Calculus Concepts An Applied Approach To The Maths of Change 4ed
Uploaded by
thanhvuCalculus Concepts An Applied Approach To The Maths of Change 4ed
Uploaded by
thanhvuFOURTH EDITION
Calculus
Concepts
An Applied Approach to
the Mathematics of Change
Donald R. LaTorre
Clemson University
John W. Kenelly
Clemson University
Iris B. Reed
Clemson University
Laurel R. Carpenter
Charlotte, Michigan
Cynthia R. Harris
Reno, Nevada
Sherry Biggers
Clemson University
Houghton Mifflin Company
Boston New York
Publisher: Richard Stratton
Senior Sponsoring Editor: Molly Taylor
Marketing Manager: Jennifer Jones
Senior Development Editor: Maria Morelli
Associate Project Editor: Susan Miscio
Art and Design Manager: Gary Crespo
Cover Design Manager: Anne Katzeff
Photo Editor: Jennifer Meyer Dare
Composition Buyer: Chuck Dutton
New Title Project Manager: Susan Brooks-Pelltier
Editorial Assistant: Andrew Lipsett
Marketing Associate: Mary Legere
Editorial Assistant: Joanna Carter
Cover photograph: © Brad Rickerby, Getty Images
TI-83, TI-84 PLUS are registered trademarks of Texas Instruments Incorporated.
Excel, Microsoft and Windows are either registered trademarks or trademarks of Microsoft
Corporation in the United States and/or other countries.
Copyright © 2008 by Houghton Mifflin Company. All rights reserved.
No part of this work may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying and recording, or by any information
storage or retrieval system without the prior written permission of Houghton Mifflin
Company unless such copying is expressly permitted by federal copyright law. Address
inquiries to College Permissions, Houghton Mifflin Company, 222 Berkeley Street,
Boston MA 02116-3764.
Printed in the U.S.A.
Library of Congress Control Numbers:
Student Edition:2006935429
Brief Student Edition:2006935430
Student Edition:
ISBN 10: 0-618-78981-2
ISBN 13: 978-0-618-78981-8
Brief Student Edition:
ISBN 10: 0-618-78982-0
ISBN 13: 978-0-618-78982-5
Instructor’s Annotated Edition:
ISBN 10: 0-618-78983-9
ISBN 13: 978-0-618-78983-2
1 2 3 4 5 6 7 8 9-DOW-11 10 09 08 07
Contents
Preface xiii
1 Ingredients of Change:
Functions and Models 1
1.1 Models and Functions 2
Mathematical Models 3
Functions 4
Determining Function Output 7
Interpreting Model Output 8
Combining Functions 9
Composing Functions 15
1.1 Concept Inventory 18 • 1.1 Activities 19
1.2 Linear Functions and Models 24
Representations of a Linear Model 24
Slope and Intercept 25
Finding a Linear Model from Data 29
Aligning Data 32
Numerical Considerations: Reporting Answers 33
Numerical Considerations: Calculating Answers 34
The Four Elements of a Model 35
1.2 Concept Inventory 36 • 1.2 Activities 36
1.3 Exponential and Logarithmic Functions and Models 40
Concavity and Exponential Growth and Decay 40
Percentage Change and Exponential Models 41
Percentage Differences and Modeling from Data 43
Doubling Time and Half-Life 45
Aligning Exponential Data 46
Logarithmic Models 47
Aligning Log Data 50
Creating Inverse Functions 51
1.3 Concept Inventory 53 • 1.3 Activities 53
1.4 Logistic Functions and Models 58
Exponential Growth with Constraints 58
Finding Logistic Models 59
Limits and the Infinitely Large 63
1.4 Concept Inventory 65 • 1.4 Activities 65
1.5 Polynomial Functions and Models 68
Quadratic Modeling 69
Quadratic or Exponential? 73
Cubic Modeling 74
1.5 Concept Inventory 80 • 1.5 Activities 80
iii
iv Contents
Summary 85
Concept Check 87
Concept Review 87
Project 1.1 Compulsory School Laws 90
Project 1.2 Fundraising Campaign 92
2 Describing Change: Rates 94
2.1 Change, Percentage Change, and Average Rates of Change 95
Interpreting Descriptions of Change 96
Finding Percentage Change and Average Rate of Change Using Graphs 98
Determining Percentage Change and Average Rate of Change Using
an Equation 102
APR and APY 103
2.1 Concept Inventory 107 • 2.1 Activities 107
2.2 Instantaneous Rates of Change 112
Local Linearity and Tangent Lines 114
Secant and Tangent Lines 117
Sketching Tangent Lines 118
Where Does the Instantaneous Rate of Change Exist? 120
2.2 Concept Inventory 122 • 2.2 Activities 122
2.3 Derivative Notation and Numerical Estimates 128
Derivative Terminology and Notation 128
Interpreting Derivatives 130
Does Instantaneous Refer to Time? 133
Percentage Rate of Change 134
Finding Slopes by the Numerical Method 136
2.3 Concept Inventory 140 • 2.3 Activities 140
2.4 Algebraically Finding Slopes 145
Limits and the Infinitesimally Small 145
Continuity Revisited 148
Finding Slopes Using the Algebraic Method 149
A General Formula for Derivatives 151
2.4 Concept Inventory 155 • 2.4 Activities 155
Summary 157
Concept Check 158
Concept Review 159
Project 2.1 Fee-Refund Schedules 161
Project 2.2 Doubling Time 162
Contents v
3 Determining Change: Derivatives 163
3.1 Drawing Rate-of-Change Graphs 164
Extracting Rate-of-Change Information from a Function Graph 164
A Detailed Look at the Slope Graph 169
Points of Undefined Slope 171
3.1 Concept Inventory 173 • 3.1 Activities 173
3.2 Simple Rate-of-Change Formulas 178
The Simple Power Rule 178
The Constant Multiplier and Sum and Difference Rules 179
3.2 Concept Inventory 185 • 3.2 Activities 185
3.3 Exponential and Logarithmic Rate-of-Change Formulas 189
Exponential Rules 189
Natural Logarithm Rule 192
3.3 Concept Inventory 194 • 3.3 Activities 194
3.4 The Chain Rule 198
The First Form of the Chain Rule 198
The Second Form of the Chain Rule 200
3.4 Concept Inventory 203 • 3.4 Activities 203
3.5 The Product Rule 207
Applying the Product Rule without Equations 207
Applying the Product Rule with Equations 209
The Quotient Rule 212
3.5 Concept Inventory 212 • 3.5 Activities 212
3.6 Limiting Behavior Revisited: L’Hôpital’s Rule 217
Analyzing Limits Using Direct Substitution 218
Indeterminate Forms and L’Hôpital’s Rule 218
3.6 Concept Inventory 221 • 3.6 Activities 221
Summary 222
Concept Check 223
Concept Review 224
Project 3.1 Superhighway 225
Project 3.2 Fertility Rates 226
4 Analyzing Change:
Applications of Derivatives 227
4.1 Approximating Change 228
Using Rates of Change to Approximate Change 228
Marginal Analysis 233
4.1 Concept Inventory 236 • 4.1 Activities 236
vi Contents
4.2 Relative and Absolute Extreme Points 241
Relative Extrema 241
Conditions When Relative Extrema Might Not Exist 244
Relative Extrema on Functions That Are Not Smooth 247
Absolute Extrema 249
4.2 Concept Inventory 251 • 4.2 Activities 251
4.3 Inflection Points 256
The Second Derivative 256
Concavity and the Second Derivative 262
4.3 Concept Inventory 263 • 4.3 Activities 263
4.4 Interconnected Change: Related Rates 270
Interconnected-Change Equations and Implicit Differentiation 270
4.4 Concept Inventory 275 • 4.4 Activities 275
Summary 278
Concept Check 278
Concept Review 279
Project 4.1 Hunting License Fees 281
Project 4.2 Fundraising Campaign 282
5 Accumulating Change: Limits of
Sums and the Definite Integral 283
5.1 Results of Change and Area Approximations 284
Accumulated Change 284
Left- and Right-Rectangle Approximations 287
Midpoint-Rectangle Approximation 292
Finding a Limit of Area Estimates 294
Accumulated Change and the Definite Integral 295
5.1 Concept Inventory 301 • 5.1 Activities 301
5.2 Accumulation Functions 309
Using Limits of Sums to Sketch Accumulation Graphs 310
Using Estimated Areas to Sketch Accumulation Graphs 310
Concavity and Accumulation 316
Recovering a Function 319
5.2 Concept Inventory 320 • 5.2 Activities 320
5.3 The Fundamental Theorem 325
The Slope Graph of an Accumulation Graph 325
Antiderivative Formulas 328
Specific Antiderivatives 333
5.3 Concept Inventory 337 • 5.3 Activities 337
Contents vii
5.4 The Definite Integral 339
Antiderivatives and Definite Integrals 340
Piecewise Functions 343
Sums of Definite Integrals 343
Differences of Accumulated Changes 346
5.4 Concept Inventory 352 • 5.4 Activities 352
5.5 Average Value and Average Rate of Change 358
Average Value of a Function 360
Average Rate of Change 362
5.5 Concept Inventory 373 • 5.5 Activities 373
5.6 Integration by Substitution or Algebraic Manipulation 373
5.6 Concept Inventory 373 • 5.6 Activities 373
Summary 373
Concept Check 375
Concept Review 375
Project 5.1 Acceleration, Velocity, and Distance 377
Project 5.2 Estimating Growth 378
6 Analyzing Accumulated Change:
Integrals in Action 379
6.1 Perpetual Accumulation and Improper Integrals 380
Evaluating Improper Integrals 380
Divergence 383
6.1 Concept Inventory 384 • 6.1 Activities 384
6.2 Streams in Business and Biology 385
Determining Income Streams 385
Future Value of a Continuous Stream 387
Present Value of a Continuous Stream 389
Discrete Income Streams 390
Streams in Biology 393
6.2 Concept Inventory 395 • 6.2 Activities 395
6.3 Integrals in Economics 400
Demand Curves 400
Consumers’ Willingness and Ability to Spend 401
Consumers’ Expenditure and Surplus 404
Elasticity of Demand 407
Supply Curves 409
Producers’ Willingness and Ability to Receive 410
Producers’ Revenue and Surplus 411
Social Gain 414
6.3 Concept Inventory 417 • 6.3 Activities 417
viii Contents
6.4 Probability Distributions and Density Functions 423
Area as Probability 425
The Uniform Probability Distribution 429
Measures of Center and Variability 430
The Exponential Probability Distribution 432
The Normal Distribution 433
Cumulative Density Functions 435
6.4 Concept Inventory 440 • 6.4 Activities 440
Summary 447
Concept Check 448
Concept Review 449
Project 6.1 Arch Art 450
7 Repetitive Change: Cyclic Functions 451
7.1 Cycles and Sine Functions 452
Sine and Cosine Functions 453
Variations of the Sine Function 454
7.1 Concept Inventory 459 • 7.1 Activities 459
7.2 Sine Functions as Models 461
The Constants of a Sine Function 462
Fitting a Sine Model to Data 465
7.2 Concept Inventory 467 • 7.2 Activities 467
7.3 Rates of Change and Derivatives 473
Derivative Formulas 473
Derivative Rules 475
7.3 Concept Inventory 482 • 7.3 Activities 482
7.4 Extrema and Points of Inflection 487
Determining Extreme Points 487
Inflection Points 488
7.4 Concept Inventory 489 • 7.4 Activities 489
7.5 Accumulation in Cycles 494
Accumulated Change for Sine Models 495
Average Values and Sine Models 496
7.5 Concept Inventory 498 • 7.5 Activities 498
Summary 500
Concept Check 501
Concept Review 502
Project 7.1 Seasonal Sales 503
Project 7.2 Lake Tahoe Levels 504
Contents ix
8 Dynamics of Change: Differential Equations
and Proportionality 506
8.1 Differential Equations and Slope Fields 507
Differential Equations 507
Proportionality 510
Slope Fields 512
8.1 Concept Inventory 515 • 8.1 Activities 515
8.2 Separable Differential Equations 520
Separation of Variables 520
Differential Equations Modeling Constant Percentage Change 521
Joint Proportionality 524
Logistic Models and Their Differential Equations 526
8.2 Concept Inventory 529 • 8.2 Activities 529
8.3 Numerically Estimating by Using Differential Equations: Euler’s Method 532
8.3 Concept Inventory 536 • 8.3 Activities 536
8.4 Second-Order Differential Equations 539
8.4 Concept Inventory 542 • 8.4 Activities 542
Summary 544
Concept Check 545
Concept Review 545
Project 8.1 On-the-Job Training 547
9 Ingredients of Multivariable
Change: Models, Graphs, Rates 548
9.1 Multivariable Functions and Contour Graphs 549
Multivariable Functions 549
Sketching Contour Curves 552
Formulas for Contour Curves 557
Estimating Output and Change in Output Using Contour Graphs 558
9.1 Concept Inventory 562 • 9.1 Activities 562
9.2 Cross-Sectional Models and Rates of Change 573
Cross-Sectional Models from Data 573
Rates of Change of Cross-Sectional Models 575
9.2 Concept Inventory 577 • 9.2 Activities 577
9.3 Partial Rates of Change 583
Partial Derivatives 583
Second Partial Derivatives 587
Concept Development: Interpreting Second Partials 588
9.3 Concept Inventory 591 • 9.3 Activities 591
x Contents
9.4 Compensating for Change 597
Concept Development: Calculating Compensation for Change 598
A General Formula 600
9.4 Concept Inventory 605 • 9.4 Activities 605
Summary 608
Concept Check 609
Concept Review 610
Project 9.1 Competitive and Complementary Products 612
Project 9.2 Expert Witness 613
10 Analyzing Multivariable
Change: Optimization 614
10.1 Multivariable Critical Points 615
Critical Points 615
Locating Critical Points on Contour Graphs 616
Locating Critical Points in Tables 618
Estimating Absolute Extrema Using Tables and Contour Graphs 621
10.1 Concept Inventory 624 • 10.1 Activities 624
10.2 Multivariable Optimization 634
Finding Critical Points Algebraically 634
The Determinant Test 637
Finding Critical Points Using Matrices 638
10.2 Concept Inventory 641 • 10.2 Activities 641
10.3 Optimization Under Constraints 645
The Lagrange Multiplier 645
Interpreting 650
10.3 Concept Inventory 653 • 10.3 Activities 653
10.4 Least-Squares Optimization 657
10.4 Concept Inventory 660 • 10.4 Activities 660
Summary 662
Concept Check 663
Concept Review 663
Project 10.1 Snow Cover 665
Project 10.2 Carbonated Beverage Packaging 667
Answers to Odd Activities A-1
Subject Index A-85
Preface
Bridging Concepts
Philosophy
This book presents a fresh, intuitive approach to the concepts of calculus for students
in fields such as business, economics, liberal arts, management, and the social and life
sciences. It is appropriate for courses generally known as “brief calculus” or “applied
calculus.”
Our overall goal is to improve learning of basic calculus concepts by involving
students with new material in a way that is different from traditional practice. The de-
velopment of conceptual understanding coupled with a commitment to make calcu-
lus meaningful to the student are guiding forces. The material in this book involves
many applications of real situations through its data-driven, technology-based mod-
eling approach. It considers the ability to correctly interpret the mathematics of real-
life situations of equal importance to the understanding of the concepts of calculus in
the context of change.
Fourfold Viewpoint Complete understanding of the concepts is enhanced and
emphasized by the continual use of the fourfold viewpoint: numeric, algebraic, ver-
bal, and graphical.
Data-Driven Many everyday, real-life situations involving change are discrete in
nature and manifest themselves through data. Such situations often can be repre-
sented by continuous or piecewise continuous mathematical models so that the con-
cepts, methods, and techniques of calculus can be utilized to solve problems. Thus we
seek, when appropriate, to make real-life data a starting point for our investigations.
The use of real data and the search for appropriate models also expose students to
the reality of uncertainty. We emphasize that sometimes there can be more than one
appropriate model and that answers derived from models are only approximations.
We believe that exposure to the possibility of more than one correct approach or an-
swer is valuable.
Modeling Approach We consider modeling to be an important tool and intro-
duce it at the outset. Both linear and nonlinear models of discrete data are used to ob-
tain functional relationships between variables of interest. The functions given by the
models are the ones used by students to conduct their investigations of calculus con-
cepts. It is the connection to real-life data that most students feel shows the relevance
of the mathematics in this course to their lives and adds reality to the topics studied.
Interpretation Emphasis This book differs from traditional texts not only in its
philosophy but also in its overall focus, level of activities, development of topics, and
attention to detail. Interpretation of results is a key feature of this text that allows
students to make sense of the mathematical concepts and appreciate the usefulness of
those concepts in their future careers and in their lives.
xi
xii Preface
Informal Style Although we appreciate the formality and precision of mathemat-
ics, we also recognize that this alone can deter some students from access to mathe-
matics. Thus we have sought to make our presentations as informal as possible by
using nontechnical terminology where appropriate and a conversational style of
presentation.
Pedagogical Features
• Chapter Opener Each chapter opens with a real-life situation and several ques-
tions about the situation that relate to the key concepts in the chapter. These
applications correspond to and reference an activity in the chapter.
• Chapter Outline An outline of section titles appears on the first page of each
chapter.
• Concepts You Will Be Learning The Concepts You Will Be Learning feature
appears at the beginning of each chapter and lists the objectives of the chapter.
• Concept Inventory A Concept Inventory listed at the end of each section gives
students a brief summary of the major ideas developed in that section.
• Section Activities The Activities at the end of each section cement concepts and
allow students to explore topics using, for the most part, actual data in a variety
of real-world settings. Questions and interpretations pertinent to the data and the
concepts are always included in these activities. The activities do not mimic the
examples in the chapter discussion and thus require more independent thinking
on the part of the students. Possible answers to odd activities are given at the end
of the book.
• Chapter Summary A Chapter Summary connects the results of the chapter top-
ics and further emphasizes the importance of knowing these results.
• Concept Check A checklist is included at the end of each chapter summarizing the
main concepts and skills taught in the chapter along with sample odd activities
corresponding to each item in the list. The sample activities are to help students
assess their understanding of the chapter content and identify on which areas to
focus their study.
• Concept Review A Concept Review activity section at the end of each chapter
provides practice with techniques and concepts. Complete answers to the Con-
cept Review activitities are included in the answer key located at the back of the
text.
• Projects Projects included after each chapter are intended to be group projects
with oral or written presentations. We recognize the importance of helping stu-
dents develop the ability to work in groups, as well as hone presentation skills.
The projects also give students the opportunity to practice the kind of writing that
they will likely have to do in their future careers.
Content Changes in the Fourth Edition
This new edition contains pedagogical changes intended to improve the presentation
and flow of the concepts discussed. It contains many new examples and activities. In
addition, many data sets have been updated to include more recent data. Three
Preface xiii
important pedagogical and context changes include the streamlining of the presenta-
tion of models, restructuring of activity sets, and the addition of certain topics.
Streamlining The first two chapters of previous editions have been condensed into
one chapter on functions and modeling in the fourth edition. Thus, the instructor can
spend less time on preliminaries and start teaching calculus sooner. The important
concept of limits has been moved into later sections so it can be taught at the time it
is first needed during the development of derivatives.
Activity Sets Activity sections have been divided into subsections by type of activ-
ity. Getting Started activities give students a chance to practice basic skills. Applying
Concepts activities are the main activities of each section and are designed to apply the
concepts taught in the section to real-world situations. Discussing Concepts activities
are designed to encourage students to communicate in written form. The authors
consider Writing Across the Curriculum to be important.
Additional Topics Discussions of L’Hôpital’s Rule, Integration by Substitution,
and Elasticity of Demand are new topics in the fourth edition. Also a chapter on
proportionality, slope fields, and differential equations that in previous editions
was available only on the website is now presented in the printed fourth edition as
Chapter 7.
Bridging Technology
Technology as a Tool
Graphing Calculators and Spreadsheets Calculus has traditionally relied upon
a high level of algebraic manipulation. However, many nontechnical students are not
strong in algebraic skills, and an algebra-based approach tends to overwhelm them
and stifle their progress. Today’s easy access to technology in the forms of graphing
calculators and computers breaks down barriers to learning imposed by the tradi-
tional reliance on algebraic methods. It creates new opportunities for learning
through graphical and numerical representations. We welcome these opportunities
in this book by assuming continual and immediate access to technology.
This text requires that students use graphical representations freely, make nu-
merical calculations routinely, and find functions to fit data. Thus continual and im-
mediate access to technology is essential. Because of their low cost, portability, and
ability to personalize the mathematics, the use of graphing calculators or laptop com-
puters with software such as Excel or Maple is appropriate.
Technology Guides Because it is not the authors’ intent that class time be used to
teach technology, we provide two Technology Guides for students: a Graphing
Calculator Instruction Guide containing keystroke information adapted to materials
in the text for the TI-83/84 Plus models, and an Excel Instruction Guide
providing the same instruction for Excel spreadsheets. In the student text,
open book icons refer readers to applicable sections within the appropriate
technology guide.
xiv Preface
It is worth noting that different technologies may give different model coefficients
than those given in this book. We used a TI-83 graphing calculator to generate the
models in the text and the answer key. Other technologies may use different fit crite-
ria for some models than that used by the TI-83.
Eye on Computers and the Internet The Calculus Concepts Website (accessi-
ble through college.hmco.com/pic/latorre4e) provides an exceptional variety of
valuable resources for instructors and students alike. The instructors’ website in-
cludes worksheets, presentation slides, additional projects, data sets categorized by
type for use on tests and quizzes, and other resource materials.
The student website provides a glossary of terms, skill and drill problems, and
Excel and TI-84 Plus data for all tables presented in the text.
Building Bridges to Better Learning
Resources for Instructors In addition to the resources found at the website, the
online Instructor’s Resource Guide with Complete Solutions gives practical sugges-
tions for using the text in the manner intended by the authors. It gives suggestions for
various ways to adapt the text to your particular class situation. It contains sample syl-
labi, sample tests, ideas for in-class group work, suggestions for implementing and
grading projects, and complete activity solutions.
The Instructor’s Annotated Edition is the text with margin notes from the au-
thors to instructors. The notes contain explanations of content or approach, teaching
ideas, indications of where a topic appears in later chapters, indications of topics that
can be easily omitted or streamlined, suggestions for alternate paths through the
book, warnings of areas of likely difficulty for students based on the authors’ years of
experience teaching with Calculus Concepts, and references to topics in the Instructor’s
Resource Guide that may be helpful.
NEW! Eduspace® (powered by BlackboardTM) Eduspace is a web-based learn-
ing system that provides instructors with powerful course management tools and stu-
dents with text-specific content to support all of their online teaching and learning
needs. Eduspace makes it easy to deliver all or part of a course online. Resources such
as algorithmic automatically-graded homework exercises, tutorials, instructional
video clips, an online multimedia eBook, live online tutoring with SMARTHINK-
ING™, and additional study materials all come ready-to-use. Instructors can choose
to use the content as is, modify it, or even add their own.
NEW! Also available are Powerpoint presentations for Digital Lessons created by
one of the authors and a long-time user of Calculus Concepts.
Learning Resources for Students
1. The Student Solutions Manual contains complete solutions to the odd activities.
2. The Graphing Calculator Guide contains keystroke information adapted to ma-
terial in the text for the TI-83 and TI-86 models. Instruction on using the TI-89
graphing calculator can be found on the companion website.
Preface xv
3. An Excel Guide provides basic instruction on this spreadsheet program.
These two Technology Guides contain step-by-step solutions to exam-
ples in the text and are referenced in this book by a supplements icon.
4. The Calculus Concepts Website (accessible through college.hmco.com/pic/
latorre4e) contains extra practice problems, help with algebra, links to updated
data needed for certain activities, a glossary of terms, practice quizzes, and other
assistance.
5. The Calculus Concepts DVD Series contains chapter-by-chapter lectures by a
master teacher. The DVD series can be used by students who miss a class or by
students who think they would benefit from seeing another teacher explain a par-
ticular topic. These DVDs can also be used as training tools for graduate teach-
ing assistants.
6. A series of Lecture and Notetaking Guides written by one of the authors and a
long-time user of Calculus Concepts is available through Houghton Mifflin Cus-
tom Publishing. Students follow along with the lecture. These notes are especially
designed to help new and/or adjunct instructors cut down on preparation time.
The Notetaking Guide assists students by integrating the discussion of concepts
with a visual or graphical emphasis, providing guided solutions of examples il-
lustrating concepts in a real-world situation, and offering specific calculator in-
struction and a practical interpretation of the results of the calculations.
Acknowledgments We gratefully acknowledge the many teachers and students
who have used this book in its previous editions and who have given us feedback
and suggestions for improvement. In particular, we thank the following reviewers
whose many thoughtful comments and valuable suggestions guided the preparation
of the revision of the fourth edition.
Marsha Austin—Oklahoma City Community College
William L. Blusbaugh—University of Northern Colorado
Marcia Frobish—Northern Illinois University
Donald R. Griffin—Greenville Technical College
Karla Karstens—University of Vermont
Doreen Kelly—Mesa Community College
Robert Lewis—Linn-Benton Community College
Mehdi Razzaghi—Bloomsburg University
David Ruch—Metropolitan State College of Denver
Debra Swedburg—Casper College
We especially acknowledge the help of
Jennifer LaVare—Clemson University
Special thanks to Carrie Green for her careful work in checking the text and answer
key for accuracy. The authors express their sincere appreciation to Charlie Hartford,
who first believed in this book, and to Molly Taylor, Maria Morelli, Susan Miscio, An-
drew Lipsett, and Joanna Carter at Houghton Mifflin Company for all their work in
bringing this fourth edition into print.
Heartfelt thanks to our husbands, Sherrill Biggers and Dean Carpenter, without
whose encouragement and support this edition would not have been possible. Thanks
also to Jessica, Travis, Lydia, and Carl, whose cooperation was much appreciated.
Fourfold Viewpoint
25
1.2 Linear Functions and Models
Complete understanding of the concepts is
With words: enhanced and emphasized by the continual use
The team starts with 80 customers, and 5 new customers are added each week.
With data (see Table 1.10): of the fourfold viewpoint: numeric, algebraic,
TABLE 1.10 verbal, and graphical.
Weeks 0 1 2 3 4 5 6 7
Number of customers 80 85 90 95 100 105 110 115
With graphs (see Figure 1.21):
C
Customers Customers
150 150
100 100
50 50
w
Weeks
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Weeks
(a) (b)
FIGURE 1.21
With a model:
C(w) 5w 80 customers
where w stands for the number of weeks since the team began the subscription drive
0 w 7.
Each of these four representations is indicative of a linear model.
Two graphical representations of the linear model are shown in Figure 1.21. The
graph in Figure 1.21a is a scatter plot—a discrete representation of the data from the
table. The graph in Figure 1.21b is the graph associated with the continuous function
modeling the data. Because calculus is the study of continuous functions and their be-
havior, once we have a continuous function modeling data, we will use the function in-
stead of the data and, likewise, will use the continuous graph instead of the scatter plot.
Slope and Intercept
A linear equation is determined by two constants: a starting value and the amount of
the incremental change. All linear functions appear algebraically as
f(x) ax b
where a is the incremental change per unit input and b is the starting value.
Linear functions are graphed as lines, where a is the slope (a measure of the line’s
3
steepness) and b is the vertical axis intercept (that is, the output value at which the line
crosses the x 0 vertical axis). The slope of a graph is of primary importance in our
Determining Change:
Derivatives
Concepts Outline
3.1 Drawing Rate-of-Change
Graphs
3.2 Simple Rate-of-Change
Formulas
3.3 Exponential and Logarithmic
Rate-of-Change Formulas
3.4 The Chain Rule
3.5 The Product Rule
3.6 Limiting Behavior Revisited:
L’Hôpital’s Rule
John Henley/CORBIS
Concept Application
The aging of the American population may be one of the demographic changes that has
Real-World Motivated the greatest impact on our society over the next several decades. Given a model for the
projected number of senior Americans (65 years of age or older), the function and its
Many everyday, real-life situations involving derivative can be used to answer questions such as the following:
• What is the projected number of senior Americans in 2030?
change are discrete in nature and manifest • How rapidly will that number be changing in 2030?
themselves through data. Thus we seek, when • What is the estimated percentage rate of change in the number of senior Americans in
2030?
appropriate, to make real-life data a starting You will be able to answer these questions by using the model given in Activity 30 of
Section 3.2 and the derivative rules presented in this chapter.
point for our investigations. Real-world data
has been completely updated for this edition.
Spreadsheets containing data sets that relate to 163
exercises are also available on the companion
websites.
xvi
Modeling Approach
197
Modeling is an important tool and is introduced 3.3 Exponential and Logarithmic Rate-of-Change Formulas
at the outset. Students use real data and graph- 30. Costs Suppose the managers of a dairy company a. Find a model for the data.
have found that it costs them approximately b. Find a formula for the rate of change of the
ing technology to build their own models and c(u) 3250 75 ln u dollars to produce u units
of dairy products each week. They also know
median family income.
c. Find the rates of change and percentage rates of
interpret results. Real-world data has been com- that it costs them approximately s(u) 50u
1500 dollars to ship u units. Assume that the
change of the median family income in 1972,
1980, 1984, 1992, and 1996.
pletely updated for this edition. Spreadsheets company ships its products once each week.
a. Write the formula for the total weekly cost of d. Do you think the above rates of change and
percentage rates of change affected the reelection
containing data sets that relate to exercises are producing and shipping u units.
b. Write the formula for the rate of change of the to-
campaigns of Presidents Nixon (1972), Carter
(1980), Reagan (1984), Bush (1992), and Clinton
tal weekly cost of producing and shipping u units.
also available on the companion websites. c. How much does it cost the company to produce
(1996)?
and ship 5000 units in 1 week? 33. iPods The cumulative revenue realized by Apple
d. What is the rate of change of total production on the sales of iPods is shown in the table.
and shipping costs at 5000 units? Interpret your
Fiscal year iPod revenue
answer.
(ending September) (millions of dollars)
31. Tuition CPI The consumer price index (CPI) for
2002 53
college tuition between 1990 and 2000 is shown in
the table. 2003 174
2004 711
Year CPI Year CPI Year CPI 2005 1923
1990 175.0 1994 249.8 1998 306.5 2006* 9423
1991 192.8 1995 264.8 1999 318.7 *projected
1992 213.5 1996 279.8 2000 331.9 (Source: Based on data from “Apple Reports Fourth
Quarter Results,” 2003–2005, Apple Computer, Inc.)
1993 233.5 1997 294.1
(Source: Statistical Abstract, 2001.) a. Find an exponential model for the data.
b. Write the derivative formula for the model.
a. Align the data as the number of years since 1980, c. Determine the revenue, rate of change in
and find a log model for the CPI. revenue, and percentage rate of change in
b. Use the model to find the rate of change of the revenue in 2005. Interpret these values.
CPI in 1998.
34. VCR Homes The percentage of households with
32. Income The Bureau of the Census reports the
TVs that also have VCRs from 1990 through 2001 is
median family income since 1947 as shown in the
shown in the table.
table. (Median income means that half of American
families make more than this value and half make Households Households
less.) Year (percent) Year (percent)
1990 68.6 1996 82.2
Median family income
1992 75.0 1998 84.6
Year (constant 1997 dollars)
1994 79.0 2001 86.2
1947 20,102
1957 26,133 (Sources: Statistical Abstract, 1998, and Television Bureau of
Advertising.)
1967 35,076
5.5 Average Value and Average Rate of Change 361 1977 40,656 a. Align the input data as the number of years since
1987, and find a log model for the data.
1987 43,756
b. Write the rate-of-change formula for the model
As in the heart rate example, the estimate improves as the number of intervals 1997 44,568
in part a.
increases. Thus we obtain the exact average value by finding the limit of the estimate
as n approaches infinity:
… f(x
Average value lim [f(x1) f(x2) n1) f(xn)]x
n→
ba
which can be written as
f(x)dx
b
Average value a
ba
Thus we have
Average Value
If y f(x) is a smooth, continuous function from a to b, then the average
value of f(x) from a to b is
b
f(x)dx
Average value of
a
f(x) from a to b ba
EXAMPLE 1 Finding Average Value and Average Rate of Change
Temperature Suppose that the hourly temperatures shown in Table 5.18 were
recorded from 7 A.M. to 7 P.M. one day in September.
5.5.1
TABLE 5.18
Time Temperature (°F) Time Temperature (°F)
7 A.M. 49 2 P.M. 80
8 A.M. 54 3 P.M. 80
9 A.M. 58 4 P.M. 78
10 A.M. 66 5 P.M. 74
11 A.M. 72 6 P.M. 69
noon 76 7 P.M. 62
1 P.M. 79
Technology as a Tool
a. Find a cubic model for this set of data. Spreadsheet and graphing calculator usage is
b. Calculate the average temperature between 9 A.M. and 6 P.M.
c. Graph the equation together with the rectangle whose upper edge is determined
integrated throughout the text.
by the average value.
d. Calculate the average rate of change of temperature from 9 A.M. to 6 P.M.
The open-book icon highlights exam-
ples discussed in the Excel Guide and in
the Graphing Calculator Guide, two online
supplements.
xvii
Section Activities
4.4 Interconnected Change: Related Rates 277
Activity sections have been divided into sub-
from the building at a rate of 3 feet per second, how 28. Salt A leaking container of salt is sitting on a sections by type of activity. Getting Started ac-
quickly is the ladder sliding down the wall when the shelf in a kitchen cupboard. As salt leaks out of a
top of the ladder is 6 feet from the ground? At what hole in the side of the container, it forms a coni- tivities give students a chance to practice basic
speed is the top of the ladder moving when it hits cal pile on the counter below. As the salt falls
the ground? onto the pile, it slides down the sides of the pile
so that the pile’s radius is always equal to its
skills. Applying Concepts activities are the main
23. Height A hot-air balloon is taking off from the
end zone of a football field. An observer is sitting at
height. If the height of the pile is increasing at a
rate of 0.2 inch per day, how quickly is the salt
activities of each section and are designed to
the other end of the field 100 yards away from the leaking out of the container when the pile is 2
balloon. If the balloon is rising vertically at a rate of inches tall? How much salt has leaked out of the apply the concepts taught in the section to real-
2 feet per second, at what rate is the distance container by this time?
between the balloon and the observer changing
when the balloon is 500 yards off the ground? How 29. Yogurt Soft-serve frozen yogurt is being dispensed
world situations. Discussing Concepts activities
far is the balloon from the observer at this time? into a waffle cone at a rate of 1 tablespoon per second.
If the waffle cone has height h 15 centimeters and are designed to encourage students to commu-
24. Kite A girl flying a kite holds the string 4 feet radius r 2.5 centimeters at the top, how quickly is
above ground level and lets out string at a rate of 2 the height of the yogurt in the cone rising when the nicate in written form. The authors consider
feet per second as the kite moves horizontally at an height of the yogurt is 6 centimeters? (Hint:
altitude of 84 feet. Find the rate at which the kite is
moving horizontally when 100 feet of string has 1 cubic centimeter 0.06 tablespoon and r 6.)
h Writing Across the Curriculum to be important.
been let out. 30. Volume Boyle’s Law for gases states that when
25. Softball A softball diamond is a square with each the mass of a gas remains constant, the pressure p
side measuring 60 feet. Suppose a player is running and the volume v of the gas are related by the equa-
from second base to third base at a rate of 22 feet per tion pv c , where c is a constant whose value de-
second. At what rate is the distance between the pends on the gas. Assume that at a certain instant,
runner and home plate changing when the runner is the volume of a gas is 75 cubic inches and its pres-
halfway to third base? How far is the runner from sure is 30 pounds per square inch. Because of
home plate at this time? compression of volume, the pressure of the gas is
increasing by 2 pounds per square inch every
26. Volume Helium gas is being pumped into a minute. At what rate is the volume changing at this
spherical balloon at a rate of 5 cubic feet per minute. instant?
The pressure in the balloon remains constant.
a. What is the volume of the balloon when its
Discussing Concepts
diameter is 20 inches?
b. At what rate is the radius of the balloon changing 31. Demonstrate that the two solution methods
when the diameter is 20 inches? referred to in part d of Example 1 yield equivalent
related-rates equations for the equation given in
27. Snowball A spherical snowball is melting, and its that part of the example.
radius is decreasing at a constant rate. Its diameter
decreased from 24 centimeters to 16 centimeters in 32. In what fundamental aspect does the method of
30 minutes. related rates differ from the other rate-of-change
a. What is the volume of the snowball when its applications seen so far in this text? Explain.
radius is 10 centimeters? 33. Which step of the method of related rates do you
b. How quickly is the volume of the snowball consider to be most important? Support your
changing when its radius is 10 centimeters? answer.
Project 6.1 Arch Art
Setting 4. The artist plans to use strips of Mylar 60 inches
wide. What is the minimum number of yards
A popular historical site in Missouri is the Gateway
of Mylar that the artist will need to purchase?
Arch. Designed by Eero Saarinen, it is located on the
original riverfront town site of St. Louis and 5. Repeat Task 4 for strips 30 inches wide.
symbolizes the city’s role as gateway to the West.
6. If the 30-inch strips cost half as much as the 60-inch
The stainless steel Gateway Arch (also called the St.
strips, is there any cost benefit to using one width
Louis Arch) is 630 feet (192 meters) high and has an
instead of the other? If so, which width? Explain.
equal span.
In honor of the 200th anniversary of the Louisiana
Purchase, which made St. Louis a part of the United Reporting
States, the city has commissioned an artist to design a
work of art at the Jefferson National Expansion Write a memo telling the artist the minimum
Memorial which is a National Historic Site The artist amount of Mylar necessary. Explain how you came
plans to construct a hill beneath the Gateway Arch, to your conclusions. Include your mathematical
located at the Historic Site, and work as an attachment.
hang strips of Mylar from the arch to Height
the hill so as to completely fill the (feet)
space. (See Figure 6.45.) The artist has 630
asked for your help in determining the
570
amount of Mylar needed.
510
450
Tasks
390
1. If the hill is to be 30 feet tall at its
330
highest point, find an equation
for the height of the cross section 270
of the hill at its peak. Refer to
the figure. 210
2. Estimate the height of the arch in 150
Projects at least ten different places. Use
the estimated heights to construct 90
End-of-chapter projects help students develop a model for the height of the arch.
(You need not consider only the
30
0
Hill Span
(feet)
models presented in this text.)
the ability to work in groups, as well as hone 3. Estimate the area between the arch
0 60 120 180 240
The Gateway Arch in St. Louis
300 360 420 480 540 600
presentation skills. The projects also give stu- and the hill. Figure 6.45
dents the opportunity to practice the kind of
writing that they will likely have to do in their 450
future careers.
xviii
Preface xix
Supplements for the Instructor
Online Instructor’s Resource Guide with Complete Solutions This manual of-
fers step-by-step solutions for all text exercises, as well as section-by-section hints for
teaching reform calculus. The guide is useful for instructors who are new to the Cal-
culus Concepts approach or who want a fresh approach to a concept.
Digital Lessons Presentation slides for lectures corresponding to each of the sec-
tions in the text can be custom published in a variety of formats including PowerPoint
and transparencies. These visuals allow instructors to minimize lecture preparation
time.
HM Testing™ (Powered by Diploma®) “Testing the way you want it” HM Test-
ing offers all the tools needed to create, author, deliver, and customize multiple types
of tests—including authoring and editing algorithmic questions.
Supplements for the Students
Student Solutions Manual This manual offers step-by-step solutions for all odd-
numbered text exercises.
Student Lecture and Notetaking Guide This notebook includes prepared class
notes with blanks to be filled in by the students during class and correlates with the
Lecture Visuals. The Lecture Visuals and Student Lecture and Notetaking Guide are au-
thored by a Calculus Concepts author and available through Houghton Mifflin custom
publishing.
Includes Online Graphing Calculator Guide, Excel Guide, graphing calculator
programs, data sets, and more.
xix
xx Preface
Supplements for the Students and the Instructor
Instructional DVDs These DVDs cover selected sections of the text and pro-
vide explanations of key concepts, applications in a lecture-based format.
Eduspace® (powered by BlackboardTM)
Eduspace is a web-based learning system that provides instructors with powerful
course management tools and students with text-specific content to support all of
their online teaching and learning needs. Eduspace makes it easy to deliver all or
part of a course online. Resources such as algorithmic automatically-graded
homework exercises, tutorials, instructional video clips, an online multimedia eBook,
live online tutoring with SMARTHINKING™, and additional study materials all come
ready-to-use. Instructors can choose to use the content as is, modify it, or even add
their own. Visit www.eduspace.com for more information.
SMARTHINKING™ Live, online tutoring
SMARTHINKING provides an easy-to-use and effective on-line, text-specific tutoring
service. A dynamic Whiteboard and Graphing Calculator function enables students
and e-structors to collaborate easily. Visit smarthinking.college.hmco.com for more
information.
Online Course Content for Blackboard®, WebCT®, and eCollege®
Deliver program or text-specific Houghton Mifflin content online using your
institution’s local course management system. Houghton Mifflin offers homework,
tutorials, videos, and other resources formatted for Blackboard®, WebCT®,
eCollege®, and other course management systems. Add to an existing online
course or create a new one by selecting from a wide range of powerful learning
and instructional materials.
For more information, visit college.hmco.com/pic/latorre4e or contact your
Houghton Mifflin sales representative.
Ingredients of Change:
Functions and Models
1
Concepts Outline
1.1 Models and Functions
1.2 Linear Functions and Models
1.3 Exponential and Logarithmic
Functions and Models
1.4 Logistic Functions and
Models
1.5 Polynomial Functions and
Models
Larry Dale Gordon/TIPS IMAGES
Concept Application
On February 16, 2005, the Kyoto Protocol, a multinational attempt to slow global warm-
ing by curbing air pollution, finally came into force seven years after the initial accord
was signed. The Kyoto Protocol is a legally binding treaty requiring that ratifying, devel-
oped nations decrease their overall emissions of the six greenhouse gases: carbon
dioxide, methane, nitrous oxide, sulfur hexafluoride, hydrofluorocarbons, and perfluo-
rocarbons. Here are some questions about greenhouse-gas emissions that can be an-
swered mathematically by using functions and/or calculus:
• What was the collective amount of the six greenhouse-gases released into the atmos-
phere in 1990? in 1997? in 2002?
• At what rate was the amount of greenhouse-gas emissions increasing between 1990
and 2002?
• Based on data accumulated since 1990, what amount of greenhouse gases will be re-
leased into the atmosphere in 2012?
This chapter will provide you with some of the tools that make it possible to answer such
questions. The information needed to answer these questions is found in Activity 29 of
Section 1.2.
1
2 CHAPTER 1 Ingredients of Change: Functions and Models
Chapter Introduction
The primary goal of this book is to help you understand the two fundamental con-
cepts of calculus—the derivative and the integral—in the context of the mathematics
of change. This first chapter is therefore devoted to a study of the key ingredients of
change: functions and mathematical models. Functions provide the basis for analyz-
ing the mathematics of change because they enable us to describe relationships
between variable quantities.
This chapter introduces the process of building mathematical models. Many of
the models we use are formed from data gathered in applied situations. Although
linear functions and models are among the most frequently occurring ones in non-
science settings, nonlinear functions apply in a variety of situations. We begin with
linear, exponential, and logarithmic functions and then consider situations in which
exponential growth is constrained in some manner. Such a situation can often be
modeled by a logistic function. The final functions we consider are quadratic and cu-
bic polynomials.
Concepts You Will Be Learning
• Evaluating and interpreting functions at specified inputs or outputs (1.1)
• Using operations and function composition to construct new functions (1.1)
• Interpreting models for profit, revenue, and other business concepts (1.1)
• Finding and interpreting the rate of change (slope) of a linear model (1.2)
• Fitting a linear model to a data set (1.2)
• Using limits to interpret end behavior of a function (1.3)
• Recognizing an inverse relationship between two functions (1.3)
• Finding and interpreting doubling time and half-life of exponential functions
(1.3)
• Finding equations of the horizontal asymptotes for logistic equations (1.4)
• Visually locating and verbally interpreting inflection points on a graph (1.4)
• Using quadratic and cubic models (1.5)
• Fitting one of six models to a data set (1.2 – 1.5)
1.1 Models and Functions
Calculus is the study of change—how things change and how quickly they change.
We begin our study of calculus by considering how we describe change. Let us start
with a situation that affects nearly all of us—the price of a gallon of gas and the quan-
tity that we purchase. According to the American Petroleum Institute, on September
5, 2005, the national average retail price (including taxes) of regular-grade gasoline
reached what was at that time the record price of $3.069 per gallon. Using this price,
we can represent the cost of a varying number of gallons of gas in four ways:
With numerical data, such as Table 1.1.
With a graph, such as Figure 1.1.
1.1 Models and Functions 3
TABLE 1.1 Cost
(dollars)
Amount of gas Cost of gas
70
(gallons) (dollars)
60
0 0
50
1 3.069
40
5 15.345
30
10 30.69
20
15 46.035
10
20 61.38 Gas
5 10 15 20 (gallons)
FIGURE 1.1
With words:
C(g) is the cost (in dollars) for pumping g gallons of gasoline when the price is
$3.069 per gallon.
With a mathematical model:
C(g) 3.069 g dollars
where C(g) is the cost of pumping g gallons of gasoline, g 0.
Most of the mathematical formulas considered in this text can be viewed from each
of four perspectives: numerical, algebraic, verbal, and graphical. Each of these repre-
sentations adds a different facet to our understanding of the formula and what it
represents.
Mathematical Models
Even though all four representations enhance our understanding of the situation they
describe, only the equation and the graph enable us to apply calculus concepts to that
situation in order to study change. The process of translating a real-world problem
into a usable mathematical equation is called mathematical modeling, and the equa-
tion (with the variables described in the context) is referred to as a model.
EXAMPLE 1 Model Construction
Pressure Under Water The pressure exerted on a person who is open-water diving
can cause a painful and dangerous condition known as decompression sickness or, in
scuba diving jargon, “the bends.” In order for divers to avoid this condition, it is im-
portant for them to understand how much pressure they can expect at different
depths. Pressure is measured in atmospheres, a unit that is abbreviated atm. At sea
level, the pressure exerted by the air is 1 atm. As a diver descends, the water exerts an
additional 1 atm for every 33 feet of added depth. Table 1.2 shows the pressure for
4 CHAPTER 1 Ingredients of Change: Functions and Models
several different depths, and Figure 1.2 depicts the pressure for depths up to 132 feet
below sea level.
TABLE 1.2 Pressure
(ATM)
Depth Pressure
(feet) (atm) 5
surface 1 4
33 2
3
66 3
2
99 4
132 5 1
Depth
33 66 99 132 (feet)
FIGURE 1.2
Use the verbal description to write a model for pressure in terms of depth.
Solution
At sea level (that is, at a depth of 0 feet), the pressure is 1 atm, and the pressure in-
creases by 1 atm for each additional 33 feet of depth. In other words, the pressure in-
1
creases by 33 atm for each additional 1 foot of depth. We can write this information
as follows:
1
Pressure in atm is equal to 1 atm plus 33 atm for every foot below the surface of
the water.
Using y as the amount of pressure in atm and x as the number of feet below sea
level, we can convert that statement into the equation
1
y 1 x atm (of pressure)
33
where x is the number of feet below the surface, 0 x 132. ●
Example 1 serves as a preview to our use of mathematical models in calculus. It
uses the available data to produce a mathematical equation (model) that describes the
relationship between the variable quantities of interest (depth below sea level and
pressure). The remaining sections of this chapter deal with modeling in more detail.
However, before we do more modeling, we need to understand some of the special
properties of the models that are useful in calculus.
Functions
In this chapter, we have considered the cost of gas given the amount of gas pumped,
as well as the pressure on an object given the depth of that object under water. Each of
these is an example of the relationship between one variable, called an input, and a sec-
ond variable, called an output. In both cases, four representations (numerical data, a
1.1 Models and Functions 5
graph, words, and an equation) were used to describe the relationship, and in both
cases, a rule was defined that assigned exactly one output to each input. These rela-
tionships are examples of functions because a rule is a function if each input produces
exactly one output. If any particular input produces more than one output, then the
rule is not a function. In order to state the input/output relationship mathematically,
when f is a rule relating an input x to a specific output, we write the output as f(x).
A function is a rule that assigns exactly one output to each input.
Function notation: f(x) is the output of function f when x is the input.
To verify that the rule for the cost of gas is a function, we must ask, “Can a certain
amount of gas have different costs if the price per gallon is fixed at $3.069?” Of course
not. If the price per gallon is fixed, then the cost for a given amount of gas is also fixed.
In this case, g is the number of gallons of gasoline purchased at $3.069 per gallon (the
input). The notation for the output is C(g). The g is enclosed in parentheses to remind
us that g is the input, and the C remains outside the parentheses to remind us that C
is the rule that gives the output. Thus C(10) 30.69 means that when 10 gallons of
gasoline are purchased (the input is 10), the amount of the purchase is $30.69 (the
output is 30.69).
When a rule is a function, we can visualize the relationship using an input/out-
put diagram. Figure 1.3 shows an input/output diagram for the rule relating cost for
a purchase of gas and the amount of gas pumped.
g
Input Gallons
Rule C
Output
C(g)
FIGURE 1.3 Dollars
You probably recall from previous math courses that the standard terms for the
set of inputs and the set of outputs of a function are domain and range, respectively.
Other terms for input and output include independent variable and dependent vari-
able and controlled variable and observed variable. In this book, however, we use the
terms input and output.
In the table representation of the cost-of-gas function, the set of inputs is {0, 1, 5,
10, 15, 20}, and the set of outputs is {0, 3.069, 15.345, 30.69, 46.035, 61.38}. In the graph
representing the cost of gas, the set of inputs is all real numbers (not just integers) be-
tween 0 and 20 (0 gallons 20), and the set of outputs is all real numbers between 0
6 CHAPTER 1 Ingredients of Change: Functions and Models
and 61.38 (0 dollars 61.38). In the verbal description and equation, the set of
inputs is all nonnegative real numbers, and the set of outputs is all nonnegative real
numbers. We call the graph of this function (see Figure 1.1) continuous because it can
be drawn without lifting the writing instrument from the page. We will give a more pre-
cise definition of continuous in Chapter 2.
EXAMPLE 2 Identifying Functions, Inputs, and Outputs
Pressure Under Water Consider again the amount of pressure encountered by a
scuba diver at certain depths under water. In Example 1, the following model was de-
veloped:
1
y 1 x atm (of pressure)
33
where x is the number of feet below the surface, 0 x 132.
a. Rewrite the equation as P(x) using function notation, and draw an input/
output diagram.
b. Identify the set of inputs and the set of outputs.
c. Is P a function of x?
Solution
a. Using the letter P to represent the rule giving pressure as output when depth x
is the input, we can rewrite the model as
1
P(x) 1 x atm (of pressure)
33
where x is the number of feet below the surface, 0 x 132. The input/out-
put diagram for P is shown in Figure 1.4.
x
Input Depth
Rule P
Output
P(x)
FIGURE 1.4 Pressure
b. The input variable represents the depth of the diver in feet. In this case, the set
of inputs is all real numbers between 0 and 132. The output variable represents
the amount of pressure. The set of outputs is all real numbers between 1 and 5.
c. P is a function of x because there is only one pressure (output) corresponding
to each given depth (input). ●
1.1 Models and Functions 7
Determining Function Output
The way to find the output that corresponds to a known input depends on how the
function is represented. In a table, simply locate the desired input in the input row or
column. The output is the corresponding entry in the adjacent row or column. For
example, the output corresponding to the input 10 in Table 1.1 is 30.69, and we write
C(10) 30.69.
If a function is represented by a formula, simply substitute the value of the input
everywhere that the variable appears in the formula and calculate the result. To use the
formula C(g) 3.069g to find the cost of purchasing 4 gallons of gas at $3.069 per
gallon, substitute 4 for g in the formula: C(4) 3.069 4 12.276 12.28 dollars.
In a function represented by a graph, our convention is to place the input on the
horizontal axis. Locate the desired value of the input on the horizontal axis, move di-
rectly up (or down) along an imaginary vertical line until you reach the graph, and
then move left (or right) until you encounter the vertical axis. (You will find a see-
through ruler helpful for improved accuracy.) The value at that point on the vertical
axis is the output. For example, Figure 1.5 shows the graph of average faculty salaries
at a private liberal arts college. When the input is 2003, the output is approximately
$56,000.
Salary
(thousands of
dollars)
60
55
50
45
40
35 Year
FIGURE 1.5 1996 1998 2000 2002 2004 2006
If at any input you can draw a vertical line that crosses the graph in two or more
places, then the graph does not represent a function. The graph in Figure 1.6 shows P
as a function of t because every input produces only one output. The graph in Figure
1.7 does not describe y as a function of x because each positive x-value produces two
different y-values. This method of visual assessment is known as the Vertical Line
Test.
y
P
x
This graph is a function. This graph is not a function.
FIGURE 1.6 FIGURE 1.7
8 CHAPTER 1 Ingredients of Change: Functions and Models
Vertical Line Test
Suppose that a graph has inputs located along the horizontal axis and
outputs located along the vertical axis. If at any input you can draw a vertical
line that crosses the graph in two or more places, then the graph does not
represent a function.
One way to determine whether a formula represents a function is to graph the
equation and then apply the Vertical Line Test.
Interpreting Model Output
In learning how functions and graphs model the real world, it is important that you
understand the units of measure of the input and output of functions. In the cost-of-
gasoline example, the unit of measure of the input is gallons, and the unit of measure
of the output is dollars. The units of measure in Figure 1.5 can be read from the graph.
The unit of measure of the output is thousands of dollars, and the unit of measure of
the input is year. Note that the unit of measure is always a word or short phrase telling
how the variable is measured, not an entire description telling what the variable rep-
resents. For example, it would be incorrect to say that the unit of measure of the
output in Figure 1.5 is “average faculty salary in thousands of dollars.” This is a de-
scription of the output variable, not the unit of output.
EXAMPLE 3 Interpreting Output
This symbol indicates Land Value The value of a certain piece of property between the years 1985 and
that instructions
specific to this
2005 is given by the model
1.1.1 example for using
v(t) 3.5(1.095t ) thousand dollars
your calculator or
computer are given where t is the number of years since the end of 1985.
in a technology
supplement. A graph and some output values of this function are given in Figure 1.8 and Table
1.3, respectively.
TABLE 1.3 v(t)
(thousands of dollars)
Value
21.5
Year (thousand dollars)
1985 3.5
1990 5.5
1995 8.7
t
2000 13.7
3.5 Years since
2005 21.5 0 20
1985
FIGURE 1.8
1.1 Models and Functions 9
The situation modeled by v can be described as follows: The value of a piece of
property worth $3.5 thousand at the end of 1985 increased by 9.5% each year since
1985.
a. Describe the input variable and the output variable. What is the unit of meas-
ure for each variable?
b. What was the land value in 2000?
c. When did the land value reach $20,000?
Solution
a. The input variable t is the number of years since the end of 1985. Its unit of
measure is years. The output variable v(t) is the value of a piece of property. Its
unit of measure is thousands of dollars.
b. The input t 15 corresponds to 2000, so the value of the land in 2000 was
v(15) 3.5(1.09515) $13,655
c. In this question you know the output, and you need to find the corresponding
input. You must solve for t in the equation 20 3.5(1.095t ) . You can either
solve the equation algebraically (using logarithms) or use technology. In either
case, you should find that t 19.2. Note that because t is defined as the num-
ber of years since the end of 1985, t 19 corresponds to the end of 2004, so
t 19.2 corresponds to early in the year 2005. Thus the land reached a value of
$20,000 in 2005. ●
Many real-world applications require us to construct more complicated func-
tions from simpler functions. The basic techniques we will discuss are combining
functions using addition, multiplication, subtraction, or division, and composing
functions.
Combining Functions
We explore the processes of combining functions via function addition, multiplica-
tion, subtraction, and division by using some basic concepts from business. These fa-
miliar concepts are fixed costs (or overhead), variable costs, total cost, revenue,
profit, average cost, and break-even point. The formulas we give here will be used
throughout the text.
Total cost fixed costs variable costs
total cost
Average cost
number of units produced
Profit revenue total cost
The break-even point occurs when revenue equals total cost and thus profit
equals 0.
10 CHAPTER 1 Ingredients of Change: Functions and Models
Concept Development: Function Addition [h(x) f(x) g(x)] Consider the
total cost to a dairy company for producing milk on the xth day of last month. The
company incurred fixed costs (costs that do not change with, or depend on, the level
of production) that amounted to $20,100 for last month. The fixed cost can be rep-
resented by the function F(x) 20,100 dollars each month. The company also had
variable cost that changed according to how many gallons of milk were produced on
a given day. The model for variable cost is V(x) 0.32x2 6x 360 dollars each
day, where x is the xth day of last month. Input/output diagrams for F and V are
shown in Figure 1.9.
xth day xth day
of of
Input month Input month
Rule F Rule V
Output Output
F (x) V (x)
Dollars (each month) Dollars (each day)
FIGURE 1.9
Total cost is defined as the sum of the fixed cost and the variable cost. By adding
the output of the function for fixed costs for a specific input value to the output of the
function for variable costs for that same input value, we can find the total cost at that
input value. Before doing so, however, we need to ask ourselves two questions: “Is the
set of input values the same for each function?” and “Are the output values given in
the same unit of measure?” The answer to the first question is yes; the xth day of the
last month is the input for both function F and function V. The answer to the second
question is no. The unit of measure for the output of V is dollars each day, whereas
the unit of measure for the output of F is dollars each month. Tables 1.4 and 1.5 show
the outputs of F(x) and V(x) for selected input values as well as the outputs of F(x)
adjusted to dollars each day. Considering the month to have 30 days, we divide the
fixed costs per month by 30 to obtain the output in dollars each day instead of dollars
20,100
per month: 30 670 dollars each day.
TABLE 1.4 TABLE 1.5
Fixed cost Adjusted fixed cost Variable cost
Day (dollars each month) (dollars each day) Day (dollars each day)
7 20,100 670 7 386.32
14 20,100 670 14 381.28
21 20,100 670 21 344.88
28 20,100 670 28 277.12
1.1 Models and Functions 11
When the units of the outputs are identical, the outputs can be added, as shown
in Table 1.6.
TABLE 1.6
Adjusted fixed cost Variable cost Total cost
Day (dollars per day) (dollars per day) (dollars per day)
7 670 386.32 1056.32
14 670 381.28 1051.28
xth day 21 670 344.88 1014.88
of 28 670 277.12 947.12
Input month
Rule F
Instead of performing this addition for only the outputs displayed in a
30 table, we can construct a function C that will give the total cost per day for
any given day x. We first convert the function F into a function that will give
Output fixed costs in terms of dollars each day. An input/output diagram for this ad-
F(x) justed fixed-cost function is given in Figure 1.10.
30 The total-cost function can now be found by adding the adjusted fixed-
Dollars
cost function to the variable-cost function:
FIGURE 1.10
F(x)
C(x) V(x)
30
670 (0.32x2 6x 360)
1.1.2
670 0.32x2 6x 360
0.32x2 6x 1030
xth day
Thus C(x) 0.32x2 6x 1030 dollars is the total cost for milk produc-
of tion on the xth day of last month. Figure 1.11 shows an input/output dia-
Input month gram for the total-cost function.
Combining functions in this manner is known as function addition.
Rule Concept Development: Function Multiplication [h(x) f(x) g(x)]
F
C= +V A meaningful new function can be constructed by multiplying the outputs
30
of two existing functions as long as their inputs are the same and the prod-
Output uct of the outputs makes sense in context. The calculation of revenue is an
C (x) example of function multiplication. In business, revenue is defined as the
Dollars price per unit at which a commodity is sold times the number of units sold.
FIGURE 1.11
EXAMPLE 4 Function Multiplication
Revenue Suppose that the price of milk was M(x) 0.02x 1.90 dollars per gallon
on the xth day of last month, 0 x 30, and that S(x) 1.5 0.5(0.8x) thousand
gallons of milk were sold on the same day x, 0 x 30.
12 CHAPTER 1 Ingredients of Change: Functions and Models
a. Draw separate input/output diagrams for M and S.
b. Determine whether M and S are compatible for function multiplication and, if
so, what the resulting output unit of measure for (M S)(x) will be.
c. Draw an input/output diagram for the multiplication function.
d. Write a function for daily revenue from milk sales.
Solution
a. Input/output diagrams for M and S are shown in Figure 1.12.
xth day xth day
of of
Input month Input month
Rule M Rule S
Output Output
M (x) S (x)
Dollars per gallon Thousand gallons
FIGURE 1.12
b. The input values and their unit of measure for both functions are the same.
When the output values are multiplied, with unit of measure (dollars per gal-
lon)(thousand gallons), the result is the revenue in thousand dollars on the xth
day of last month.
c. An input/output diagram for R(x) M(x) S(x) is given in Figure 1.13.
xth day
of
Input month
Rule
R = M .S
Output
R(x)
FIGURE 1.13 Thousand dollars
d. A function for revenue is
R(x) M(x) S(x)
[0.02x 1.90] [1.5 0.5(0.8x) ] thousand dollars
on the xth day of last month, 0 x 30 ●
1.1 Models and Functions 13
Concept Development: Function Subtraction [h(x) f(x)g(x)] Function
subtraction is similar to function addition. As in function addition, it is important
that both the set of input values and the output unit of measure for each of the func-
tions to be subtracted are the same. The dairy company introduced earlier in this sec-
tion would be interested in profit, not just in revenue. Profit is the difference between
revenue and total cost—that is, profit is equal to revenue minus cost.
The functions R for revenue and C for total cost can be combined using func-
tion subtraction to create a function for profit. However, we must be careful,
because even though the sets of inputs agree, the output unit of measure for rev-
enue is thousands of dollars and the output unit of measure for cost is dollars. In
order to combine the two functions, we convert the units of one function so that
the units of R and C are in agreement. We can multiply the revenue function by
1000 so that the output unit of measure will be dollars. The function for profit can
thus be written
P(x) 1000R(x) C(x)
1000 {[0.02x 1.90] [1.5 0.5(0.8x)]} [0.32x2 6x 1030]
1000 {[0.02x 1.90] [1.5 0.5(0.8x)]} 0.32x2 6x 1030 dollars
on the xth day of last month, 0 x 30.
Note: An alternative solution would be to divide the cost function by 1000 so that
the output unit of measure will be a thousand dollars. We find the profit as follows:
C(x)
P(x) R(x)
1000
[0.32x2 6x 1030]
{[0.02x 1.90] [1.5 0.5 (0.8x)]}
1000
thousand dollars on the xth day of last month, 0 x 30.
g(x)
Concept Development: Function Division [h(x) , where f (x) 0]
f(x)
Function division is useful when we are finding averages, percentages, and other
ratios, but in order to use function division to construct a new function, we must con-
firm that the original functions have the same set of inputs and the same unit of meas-
ure for those inputs. For instance, if we know that Q(x) 1.6x2 30x 1800
gallons of milk were produced on the xth day of last month and that the total cost of
production is C(x) 0.32x2 6x 1030 dollars, where x represents the same in-
put in both functions, we can construct a function for the average cost for producing
one gallon of milk on the xth day of last month. Average cost is defined as the total cost
of production divided by the number of units produced, so
Average cost
C(x)
A(x)
Q(x)
0.32x2 6x 1030 dollars
1.6x2 30x 1800 gallons
0.32x2 6x 1030
dollars per gallon
1.6x2 30x 1800
on the xth day of last month, 0 x 30.
14 CHAPTER 1 Ingredients of Change: Functions and Models
We summarize the four function combinations as follows:
If the inputs of functions f and g are identical (corresponding values and unit
of measure), then the following new functions can be constructed and are
valid models under the given conditions:
• Function addition: h f g if the output units are identical (correspon-
ding values and unit of measure)
• Function subtraction: j f g if the output units are identical (corre-
sponding values and unit of measure)
• Function multiplication: k f g if the output units are compatible
f
• Function division: l , where g(x)
0, as long as the output units are
g
compatible
An important point in business applications is the break-even point. The break-even
point is defined as the point at which revenue equals total cost and thus profit is zero.
Example 5 uses constructed functions and graphs to determine the break-even point
for a dairy company.
EXAMPLE 5 Combining Functions
Break-even Point Suppose that a month’s total cost for the production of g gallons
of milk can be modeled as K(g) 20,100 0.2g dollars, 0 g 30. During the same
month, the average price for a gallon of milk is $2.20, so revenue can be modeled as
T(g) 2.2g dollars for g gallons of milk, g 0.
a. Graph K(g) 20,100 0.2g and T(g) 2.2g on the same set of coordinate axes.
b. Write a function giving the profit from the production and sale of g gallons of
milk. Graph the profit function.
c. How much milk needs to be produced/sold in order for the dairy company to
break even? Show this point on the graphs from parts a and b.
Solution
a. Figure 1.14 shows total cost and revenue graphed together.
1.1.3 Dollars
(thousands)
35
T
30
25
20 K
15
Break-even
10 point
5
Gallons
FIGURE 1.14 5 10 15 (thousands)
1.1 Models and Functions 15
b. Because profit is defined as revenue minus total cost, a model for profit can
be written as
P(g) 2.2g (20,100 0.2g)
2.0g 20,100 dollars
where g is the number of gallons produced and sold, g 0. See Figure 1.15.
Dollars
(thousands)
15
P
10
5
Gallons
0
5 15 (thousands)
−5
−10 Break-even
point
−15
FIGURE 1.15 −20
c. The dairy company will break even when the output of profit is zero—that is,
when P(g) 2.0g 20,100 0. Solving for g yields g 10,050 gallons. The
dairy company must produce/sell 10,050 gallons of milk during the month in
order for revenue to equal total cost. See the break-even point depicted in
Figures 1.14 and 1.15. ●
Composing Functions [h(x) (g f ) (x) g(f(x))]
Another way in which we construct functions is called function composition. This
method of constructing functions uses the output of one function as the input of
another. Consider Tables 1.7 and 1.8. Table 1.7 gives the altitude of an airplane as a
function of time, and Table 1.8 gives air temperature as a function of altitude.
TABLE 1.7 TABLE 1.8
t time into F (t ) thousand F thousand A (F ) air temperature
flight (minutes) feet above sea level feet above sea level (degrees Fahrenheit)
0 4.4 4.4 71.0
1 7.7 7.7 14.0
2 13.4 13.4 34.4
3 20.3 20.3 55.7
4 27.2 27.2 62.6
5 32.9 32.9 64.7
6 36.2 36.2 65.2
16 CHAPTER 1 Ingredients of Change: Functions and Models
It is possible to combine these two data tables into one table that shows air tem-
perature as a function of time. This process requires using each of the output values
from Table 1.7 as the corresponding input value for Table 1.8. The only restriction is
that both the numerical values and the unit of measure in the output of one function
must match those in the input of the other function. Figure 1.16 illustrates the out-
put of one function being used as the input of the next function. The resulting new
function is shown in Table 1.9.
t TABLE 1.9
Minutes
Input t = time into A(F ) = air temperature
flight (minutes) (degrees Fahrenheit)
Rule F 0 71.0
1 14.0
F(t) 2 34.4
Feet
3 55.7
4 62.6
Rule A
5 64.7
Output 6 65.2
A(F(t))
Degrees Fahrenheit
FIGURE 1.16
Because we used the output values of Table 1.7, F(t), as the input values for Table
1.8, the representation for the output of the new function is commonly written as
A(F(t)). The A is placed outside the parentheses to remind us that the output is air
temperature. The F(t) inside the parentheses reminds us that altitude from Table 1.7
is the connecting link and that time t is the input of the new function. It is customary
to refer to A as the outside function and to F as the inside function. The mathematical
symbol for the composition of an inside function F and an outside function A is A F,
so
(A F)(t) A(F(t))
The altitude data can be modeled as F(t) 0.2t3 1.8t2 1.7t 4.4 thousand
feet above sea level, where t is the time into the flight in minutes. The air temperature
can be modeled as A(F) 279 (0.85F ) 66 degrees Fahrenheit, where F is the num-
ber of feet above sea level. The composition of these two functions is
A(F) A(F(t))
279 (0.850.2t 1.8t 1.7t 4.4) 66 degrees Fahrenheit
3 2
where t is the time in minutes into the flight.
1.1 Models and Functions 17
We formalize the process of function composition as follows:
Given two functions f and g, we can form their composition if the outputs
from one of them, say f, can be used as inputs to the other function, g. In this
case, the unit of measure for the output of function f must be identical to the
unit of measure for the input of function g.
Notation: h(x) g(f(x)) (g f)(x)
In terms of inputs and outputs, we have those shown in Figure 1.17. In this case,
we can replace the portion of the diagram within the teal box by forming the com-
posite function g f, whose input/output diagram is shown in Figure 1.18.
Input x
Output
Rule f of f
Input x
becomes
input
f(x)
of g
Rule
Rule g
Output
Output
g( f(x)) g( f(x))
FIGURE 1.17 FIGURE 1.18
EXAMPLE 6 Finding a Composite Function
Lake Contamination Consider the word descriptions of the following two func-
tions and their input/output diagrams, which are shown in Figure 1.19.
C(p) parts per million of contamination in a lake when the population of the
surrounding community is p people
p(t) the population in thousands of people of the lakeside community in year t
p t
Input People Input Year
Rule C Rule p
Output Output
C(p) p(t)
Parts per million People in thousands
FIGURE 1.19
18 CHAPTER 1 Ingredients of Change: Functions and Models
a. Draw an input/output diagram for the composition function that gives the
contamination in a lake as a function of time.
b. The contamination in the lake can be modeled as C(p) p parts per million
when the community’s population is p people, p 0. The population of the
community can be modeled by p(t) 0.4t2 2.5 thousand people t years after
1980, t 0. Write the function that gives the lake contamination as a function
of time.
Solution
a. Note that the unit of measure of the output of the second function may be con-
verted to the unit of measure of the input of the first function. If we multiply
p(t) by 1000 to convert the unit of measure of the output from thousands of
people to people, we have the function P(t) 1000p(t) people in year t, t 0.
Now we can compose the two functions C and P to create the new function
C P, whose input/output diagram is shown in Figure 1.20.
t
Input Year
Rule C ° P
Output
C(P(t))
FIGURE 1.20 Parts per million
b. Use the population of the community given by P(t) 1000p(t) as input in the
function C(p). P(t) 1000(0.4t2 2.5) people t years after 1980, t 0. Thus
the contamination in the lake can be modeled as
C P(t) C(P(t)) 1000(0.4t2 2.5) parts per million
t years after 1980, t 0. ●
Calculus is the mathematics of change that occurs over continuous portions of
functions. As we embark on our study of calculus, we will consider modeling prima-
rily as a tool for developing continuous functions. Once we can model data with a
function, we can use the tools of calculus to analyze that model—especially to deter-
mine rates of change and identify maxima, minima, and inflection points.
1.1 Concept Inventory • Units of measure
• Graphs of functions
• Mathematical models • Vertical Line Test
• Function • Function addition and subtraction
• Function notation • Function multiplication and division
• Inputs and outputs • Cost, revenue, profit, break-even point
• Input/output diagram • Function composition
1.1 Models and Functions 19
7. The table gives the maximum “no-compression”
1.1 Activities dive times for open-water scuba diving with air-
filled tanks.
Getting Started
For each of the rules in Activities 1 through 4, a. specify Depth of dive Maximum dive time
the input and output descriptions, the input and output (feet) (minutes)
variables, and the input and output units of measure, b. 50 80
determine whether each rule is a function, and c. draw 60 55
an input/output diagram. 70 45
1. R(w) the first-class domestic postal rate (in 80 35
cents) of a letter weighing w ounces 90 25
2. H(a) your height (in inches) at age a years 100 22
3. A(m) the amount (in dollars) you pay for lunch 110 15
on the mth day of any week 120 12
4. C(m) the amount of credit (in dollars) that (Source: Burks Oakley II, Nitrox Scuba Diving,
Citibank Visa will allow a 20-year-old with a yearly April 30, 2000.)
income of m dollars
8.
Determine whether the tables in Activities 5 through 8
represent functions. Assume that the input is in the left Run of advertisement Total new customers
column. (weeks) (people)
5. 1 1
3 3
iPod sales
Year (millions of units) 5 8
2002 0.14 7 18
2003 0.336 10 20
2004 2.016
2005 6.451 9. Which of the following graphs represent functions?
(The input axis is horizontal.)
(Source: Compiled from “Apple
Reports Fourth Quarter Results,”
2003–2005, Apple Computer, Inc.) a. b.
6.
Military rank Basic monthly pay
(4 years of service) in 2005 (dollars)
Second Lieutenant 2948
First Lieutenant 3541
c.
Captain 3823
Major 4388
Lt. Colonel 4961
Colonel 5784
Brigadier General 7119
Major General 8459
(Source: Defense Finance and Accounting Service.)
20 CHAPTER 1 Ingredients of Change: Functions and Models
10. Which of the following graphs represent functions? sentences interpreting the following mathematical
(The input axis is horizontal.) statements. (Source: Statistical Abstract, 2001.)
a. P(Hispanic) 11.8
a. b. b. P 4.1 when r Asian and Pacific Islander
15. Sales A music store offers one free CD with the
purchase of four CDs priced at $18 each (tax
included).
a. Construct a graph showing the cost of buying x
CDs. Show input values from 0 to 10.
c. b. What is the cost of 6 CDs?
c. How many CDs could you buy if you had $36?
d. How many CDs could you buy if you had $100?
16. Sales A fraternity is selling T-shirts on the day of a
football game. The shirts sell for $8 each.
a. Complete the table. Revenue is defined as the
11. Prices P(m) is the median sale price (in thousands
number of units sold times the selling price.
of dollars) of existing one-family homes in metro-
politan area m in 2000. Write the following state-
Number of shirts sold Revenue (dollars)
ments in function notation.
(Source: Statistical Abstract, 2001.) 1
a. The median sale price in Honolulu, Hawaii, was 2
$295,000. 3
b. In Providence, Rhode Island, the median sale 4
price was $137,800. 5
c. $170,100 was the median sale price in Portland, 6
Oregon.
12. Darkness Hours H(d) is the number of hours of b. Construct a revenue graph by plotting the points
darkness in Anchorage, Alaska, on the dth day of the in the table.
year. Write the following statements in function c. How many T-shirts can be purchased with $25?
notation. d. If an 8% sales tax were added, how many T-shirts
a. On the 121st day of the year, Anchorage has 7.5 could be purchased with $25?
hours of darkness.
17. Weight A baby weighing 7 pounds at birth loses
b. The duration of darkness in Anchorage on the 7% of her weight in the 3 days after birth and then,
361st day of the year is 18.5 hours. over the next 4 days, returns to her birth weight.
c. In Anchorage there are only 4.5 hours of dark- During the next month, she steadily gains 0.5 pound
ness on the 181st day of the year. per week. Sketch a graph of the baby’s weight from
13. Exports E(t) is the value of cotton exports (in mil- birth to 4 weeks. Accurately label both axes.
lions of dollars) in year t. Write sentences interpret- 18. Medicine A patient is instructed by her doctor to take
ing the following mathematical statements. one pill containing 500 milligrams of a drug. Assume
(Source: Statistical Abstract, 1994.) that each day the patient’s body uses 20% of the re-
a. E(1988) 1975 maining amount of the drug. The amount of the drug
b. E 1999 when t 1992 remaining at the beginning of day x can be modeled as
14. Population P(r) is the percentage of U.S. resi- f(x) 500(0.8x) milligrams
dents in the year 2000 who were of origin r. Write a. Graph the model.
1.1 Models and Functions 21
b. Use the model to determine how much of the d. If you decrease the time financed from 48 to 36
drug is left in the patient’s body after 3.5 days. months, by how much will your payment
c. Use a graph of the model to estimate when the increase?
concentration of the drug will be 60 milligrams.
Monthly payment
Use the model to find the exact value.
(dollars)
19. Payment You are interested in buying a used car 1500
and in financing it for 60 months at 10% interest. As
a special promotion, the dealer is offering to finance 1200
with no down payment. The accompanying graph 900
shows the value of the car as a function of the
amount of the monthly payment. 600
300
Car value 0 Months
(dollars) 0 12 24 36 48 60 72 84
28,000 21. Social Security The accompanying graph shows
24,000 the percentage by which Social Security checks in-
20,000 creased as a result of cost-of-living adjustments
16,000 every year from 1999 through 2005.
12,000
a. What was the cost-of-living increase in 2002?
8,000
4,000 Monthly b. When was the cost-of-living increase the great-
payment est? Estimate the cost-of-living increase in that
100 200 300 400 500 600 (dollars) year.
c. When was the cost-of-living increase 2.1%?
a. Estimate the value of the car you can buy if your d. Did Social Security benefits increase or decrease
monthly payment is $200. between 2002 and 2003? Explain.
b. Estimate the monthly payment for a car that Social Security
costs $16,000. cost-of-living
increase
c. Estimate the amount by which your monthly
(percent)
payment will increase if you buy a $20,000 car
rather than a $15,000 car. 3.5
d. How would the graph change if the interest rate 3.0
were 12.5% instead of 10%?
2.5
20. Payment You have decided to purchase a car for
$18,750. You have 20% of the purchase price to use 2.0
as a down payment, and the purchase will be
financed at 10% interest. The accompanying graph 1.5
shows the monthly loan payment in terms of the 1.0 Year
number of months over which the loan is financed. 2000 2002 2004
a. What is the actual amount being financed? (Source: Based on data from Social Security Administration.)
b. Estimate your monthly payment if you finance 22. Wages The accompanying graph compares the
the purchase over 36 months. minimum wage with its value in constant 1996 dol-
c. Estimate the number of months you will have to lars. Because the minimum wage is not indexed to the
pay on the loan if you can afford to pay only $300 price level, it has been legislatively increased from
a month. time to time to make up for the loss in its value due
22 CHAPTER 1 Ingredients of Change: Functions and Models
to inflation. The last general Fair Labor Standards Act For each of the functions in Activities 31 through 34, de-
amendments were adopted in 1996, at which time the termine whether an input or an output value is given,
current federal minimum wage was set at $5.15 per and find its corresponding output or input.
hour. This wage became effective in 1997.
31. A(t) 32e0.5t ; t 15
Dollars 32. g(x) 4x2 32x 13; g(x) 247
7.50 33. f(x) 3(1.04x); f(x) 3.65
6.50 Value of the minimum wage 100
34. p(m) 0.3m ; m 2.3
(in constant 1996 dollars) 12e
5.50
35. The total cost for producing x units of a commodity
4.50 is $4.2 million, and the revenue generated by the
3.50 sale of x units is $5.3 million.
2.50 Minimum wage a. What is the profit on x units of the commodity?
1.50 Year b. Assuming T(x) is the total cost and R(x) is the
1980 1984 1988 1992 1996 2000 2004 revenue for the production/sale of x units of a
commodity, write an expression for profit.
(Source: Based on data from Bureau of Labor Statistics.)
36. The total cost for producing x units of a commodity
a. Estimate the minimum wage in 2005 and the is $400, and the revenue generated by the sale of x
value of the minimum wage in constant 1996 units is $300.
dollars in that same year. a. What is the “profit” on x units of the commodity?
b. Over what intervals did the minimum wage re- b. What does negative profit indicate?
main constant? What happened to the buying
power of the minimum wage during these 37. A company posted revenue of $35 million and a
intervals? profit of $19 million during the same quarter.
c. In what years was the 1996 value of the mini- a. What was the company’s cost during that quarter?
mum wage above $5? b. Assuming P(x) is the profit and R(x) is the rev-
d. Estimate the year in which the graphs intersect. enue posted by the company in the xth quarter,
Discuss the meaning of this point of intersec- write an expression for total cost.
tion. 38. A company posted a net loss of $3 billion last quarter.
For Activities 23 through 26, find the output of the func- a. What was its profit?
tion corresponding to each input value given.
b. If last quarter’s revenue was $5 billion, what was
23. t(s) 3s 6; s 5, s 10 its cost?
24. f(x) 7.2x 3; x 20, x 2 39. It cost a company $19.50 to produce 150 glass bottles.
25. R(w) 9.4(1.8w) ; w 3, w 0 a. What was the average cost of production of a
120 glass bottle?
26. S(t) 2t ; t 10, t 2
13e b. Assuming C(x) is the total cost for producing x
For Activities 27 through 30, find the input of the func- units, write an expression for average cost per
tion corresponding to each output value given. unit.
27. t(s) 3s 6; t(s) 18, t(s) 0 40. Natural Gas Trade The following two functions
have a common input, year t. I is the projected
28. f (x) 7.2x 3; f(x) 6.6, f(x) 3 amount of natural gas imports in quadrillion Btu, and
29. R(w) 9.4(1.8w); R(w) 9.4, R(w) 30 E is the projected natural gas exports in trillion Btu.
120 a. Using function notation, show how you could
30. S(t) 2t ; S(t) 60, S(t) 30
13e combine the two functions to create a third
1.1 Models and Functions 23
function, N, giving net trade of natural gas in and consumption of whole milk for the same period
year t. can be modeled by
b. What does it mean when net trade is negative? W(t) 0.01685t2 3.49t 188
41. Debit Cards The following two functions have a gallons per person per year
common input, year y. D is the total number of In both models, t is the number of years since 1900.
debit card transactions, and P is the number of Use the models to construct a model giving the per
point-of-sale transactions. capita consumption of milk other than whole milk,
a. Using function notation, show how you could and estimate this per capita consumption for the
combine the two functions to create a third func- year 2000.
tion, r, showing the percentage of debit card (Source: Statistical Abstract, 2001.)
transactions that were conducted at the point of
sale in year y. 45. Credit Cards The total amount of credit card debt
from 1998 and projected through 2005 can be ex-
b. What are the output units of the new function?
pressed by the function
42. Gas Prices The following two functions have a D(y) 42.4y 219.5 billion dollars
common input, year t. R is the average price (in dol-
lars) of a gallon of regular unleaded gasoline, and P y years after 1990. The number of cardholders dur-
is the purchasing power of the dollar as measured by ing that same time interval can be expressed by the
consumer prices based on 2001 dollars. function
a. Using function notation, indicate how to com- H(y) 1.7y 140.3 million cardholders
bine the two functions to create a third function
(Source: Statistical Abstract, 2000 and 2001.)
showing the price of gasoline in constant 2001
dollars. a. Construct a function that expresses the average
b. What are the output units of the new function? credit card debt per cardholder.
b. Estimate the average debt of a cardholder in 2005.
43. Earnings The salary of one of Compaq Computer
Corporation’s senior vice presidents from 1996 46. Transplants The number of kidney and liver
through 1998 can be modeled by transplants performed in the United States between
1992 and 1996 can be modeled by
S(t) 69,375t 380,208 dollars
K(x) 9.09 1.7 ln x kidney transplants
t years after 1996. His other, nonsalary compensa-
tion during the same period can be modeled by L(x) 2.42 9.2 ln x liver transplants
where x is the number of years since 1990.
C(t) 31.67t2 137.15t 233.5 thousand dollars
(Source: Statistical Abstract, 1998.)
t years after 1996. In addition, each year he received
a. Construct a model for the number of kidney and
an average bonus of $650,000.
liver transplants between 1992 and 1996.
(Source: Based on data from Compaq’s 1999 Notice of Annual
Meeting.) b. Use the model to estimate the number of kidney
and liver transplants in 1995.
a. Construct a model for this VP’s total yearly
salary package, including nonsalary compensa- 47. Births The number of births during the 1980s to
tion and bonuses. women who were 35 years of age or older can be
b. Estimate the VP’s 1997 total salary package. modeled as
n(x) 0.034x3 1.331x2 9.913x 164.447
44. Milk Consumption Per capita milk consumption thousand births
in the United States between 1980 and 1999 can be x years after 1980.
modeled by The following model gives the number of cesarean-
M(t) 0.219t 45.23 gallons per person per year section deliveries for each 1000 live births among
24 CHAPTER 1 Ingredients of Change: Functions and Models
women in the same age bracket during the same 51. R(x) is the revenue in deutsche marks from the sale
time period: of x soccer uniforms.
p(x) 0.183x2 2.891x 20.215 deliveries D(r) is the dollar value of r deutsche marks.
per 1000 live births In Activities 52 through 55, rewrite each pair of func-
x years after 1980. tions as one composite function.
Write an expression for the number of cesarean- 52. f(t) 3et ; t(p) 4p2
section deliveries performed on women 35 years of
age or older during the 1980s. 4
53. h(p) ; p(t) 1 3e0.5t
(Source: Statistical Abstract, 1992.) p
Determine whether the pairs of functions in Activities 54. g(x) 7x2 ; x(w) 4ew
48 through 51 can be combined by function composi-
tion. If so, give function notation for the new function, 55. c(x) 3x2 2x 5 ; x(t) 4 6t
and draw and label its input/output diagram.
Discussing Concepts
48. P(c) is the profit in dollars generated by the sale of c
computer chips. 56. Why is it important to understand the units of
C(t) is the number of computer chips a manufac- measure of input and output for a given function?
turer has produced after t hours of production. How can labeling units help in function construc-
tion?
49. C(t) is the number of cats in the United States at the
end of year t. 57. Describe the types of compatibility necessary be-
tween input units and output units for each of the
D(c) is the number of dogs in the United States at following types of function construction: addition,
the end of year c. multiplication, division, and composition.
50. C(t) is the average number of customers in a restau-
58. Discuss the advantages of each of the four perspec-
rant on a Saturday night t hours after 4 P.M.
tives from which a mathematical model can
P(c) is the average amount in dollars of tips gener- be viewed: numerical, algebraic, verbal, and
ated by c customers. graphical.
1.2 Linear Functions and Models
Having explored the concept of a function in the last section, we turn our attention
in the remainder of the chapter to several specific types of functions that will be help-
ful as we seek to describe real-life situations with mathematical models. Our goal is to
give you an understanding of the behavior underlying certain functions in order to
help you determine which one of those functions is most appropriate in a particular
modeling situation.
Representations of a Linear Model
We begin with the simplest of all functions: the linear function. A linear function is
one that repeatedly and at even intervals adds the same value to the output. The out-
put values form the pattern of a line when graphed; thus we use the term linear to de-
scribe the data. For example, consider a newspaper delivery team that makes weekly
deliveries of newspapers and devotes their Saturday mornings to selling new sub-
scriptions. The function describing the number of customers can be represented in
four ways.
1.2 Linear Functions and Models 25
With words:
The team starts with 80 customers, and 5 new customers are added each week.
With data (see Table 1.10):
TABLE 1.10
Weeks 0 1 2 3 4 5 6 7
Number of customers 80 85 90 95 100 105 110 115
With graphs (see Figure 1.21):
C
Customers Customers
150 150
100 100
50 50
w
Weeks
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Weeks
(a) (b)
FIGURE 1.21
With a model:
C(w) 5w 80 customers
where w stands for the number of weeks since the team began the subscription drive
0 w 7.
Each of these four representations is indicative of a linear model.
Two graphical representations of the linear model are shown in Figure 1.21. The
graph in Figure 1.21a is a scatter plot—a discrete representation of the data from the
table. The graph in Figure 1.21b is the graph associated with the continuous function
modeling the data. Because calculus is the study of continuous functions and their be-
havior, once we have a continuous function modeling data, we will use the function in-
stead of the data and, likewise, will use the continuous graph instead of the scatter plot.
Slope and Intercept
A linear equation is determined by two constants: a starting value and the amount of
the incremental change. All linear functions appear algebraically as
f(x) ax b
where a is the incremental change per unit input and b is the starting value.
Linear functions are graphed as lines, where a is the slope (a measure of the line’s
steepness) and b is the vertical axis intercept (that is, the output value at which the line
crosses the x 0 vertical axis). The slope of a graph is of primary importance in our
26 CHAPTER 1 Ingredients of Change: Functions and Models
study of calculus. As we shall see in Chapter 3, the slope of most functions is deter-
mined using calculus; however, the slope of a line can be calculated more simply.
The directed horizontal distance from one point on a graph to another is called
the run, and the corresponding directed vertical distance is called the rise. The quo-
tient of the rise divided by the run is the slope of the line connecting the two points.
Consider again the newspaper subscription graph in Figure 1.21. We have chosen the
points on the graph that correspond to 0 and 7 weeks. The rise and run correspon-
ding to those two points are shown in Figure 1.22.
C
Customers
150
100 Rise = 115 - 80 = 35 customers
Run = 7 weeks
50
w
0 7 Weeks
FIGURE 1.22
rise 35 customers
The slope is calculated as run 7 weeks 5 customers per week. (The slope
value will be the same regardless of which two points are chosen for the calculation.)
The steeper a line, the greater the magnitude of the value of the slope. Lines that fall
rather than rise have negative slope. Unlike the slope, the steepness of a line does not
depend on whether it rises or falls.
Although the slope of a particular linear model never changes, the graph of the
model may look different when the horizontal or vertical scale is changed. By using a
different vertical scale for the graph of customers (see Figure 1.23), we obtain what
appears to be a steeper line. However, a calculation of slope reveals that the difference
is only visual. Appearances can deceive, so be careful when comparing graphs with
differing scales.
C
Customers
120
110
100 Rise = 115 - 80 = 35 customers
90
80
Run = 7 weeks
w
70
0 7 Weeks
FIGURE 1.23
1.2 Linear Functions and Models 27
The slope of a graph of a function at a particular point is a measure of how
quickly the output is changing as the input changes at that point. This measure is
called the rate of change of the function at that point, and we discuss it more thor-
oughly in Chapters 2 and 3. Because linear functions are characterized by constant
incremental change, their slope (rate of change) is constant at all points. Other
types of functions have a different slope at every point; that is, their rate of change
is not constant.
EXAMPLE 1 Calculating Slopes
Sales The resale value of a used car is represented graphically in Figure 1.24.
Car value
(dollars)
12,000
Years
FIGURE 1.24 0 8
a. Locate and interpret the point at which the input is 0 and the point at which the
output is 0.
b. Calculate the slope of the graph.
c. Does the slope depend on the direction in which it was calculated?
d. Interpret the slope.
e. Write a linear model for the graph.
Solution
a. The point at which the input is 0 is where the line intercepts the vertical axis—
that is, at point (0, 12,000). The vertical axis intercept is $12,000 and corre-
sponds to the value of the car when it was purchased.
The point at which the output is 0 is where the line intercepts the horizontal
axis—that is, at point (8, 0). The horizontal axis intercept is 8 years and indi-
cates the first year in which the car had essentially no value.
b. To travel from the horizontal axis intercept 8 to the vertical axis intercept
12,000 on the graph requires that you move 12,000 units up
( $12,000 rise) and 8 units to the left (8 years run). See Figure 1.25.
$12,000
The quotient of the rise divided by the run is 8 years $1500 per year.
28 CHAPTER 1 Ingredients of Change: Functions and Models
Car value
(dollars)
Run = -8 years
12,000
Rise = $12,000
Years
FIGURE 1.25 0 8
c. If you traveled from the vertical axis intercept 12,000 to the horizontal axis in-
tercept 8, the run would be positive and the rise negative. The quotient of the
$12,000
rise divided by the run would be 8 years $1500 per year. Note that the
rise
quotient run is the same either way you compute it.
rise
d. The quotient run is equal to the slope of the linear model and is the rate of
change of the value of the car. We say that the car depreciated at a rate of $1500
per year.
e. The equation of the line that represents the value of the car has slope 1500
and starting value 12,000. If we let the variable t represent the number of years
since the car was purchased, 0 t 8, the model for the value of the car is
Value 1500t 12,000 dollars ●
It is important to be able to describe the behavior of a function or graph. We use
the words increasing, decreasing, and constant to describe the output behavior of a
graph or function. If a graph is rising as you move from left to right along the hori-
zontal axis, then it is said to be increasing. A graph is said to be decreasing if it is
falling as you move from left to right along the horizontal axis. The portion of a graph
that neither rises nor falls is called constant. The function shown in Figure 1.21 is
increasing. The function shown in Figure 1.24 is decreasing. In the case of the in-
creasing linear function the slope a is positive, and in the case of the decreasing linear
function the slope is negative.
We summarize our discussion of linear equations as follows:
Linear Model
Verbally: A linear model is one that has a constant rate of change.
Algebraically: A linear model has an equation of the form
f(x) ax b
where a and b are constants.
Graphically: The graph of a linear function is a line (see Figure 1.26). The
value a in the equation is the constant rate of change of the output and is the
slope of the line. The value b is the output at which the input is zero.
1.2 Linear Functions and Models 29
f(x) f(x)
x x
(a) f(x) = ax + b with a > 0 (b) f(x) = ax + b with a < 0
FIGURE 1.26
Finding a Linear Model from Data
Suppose you are the sole proprietor of a small business that has seen no growth in
sales for the last several years. You have noticed, however, that your federal taxes have
increased as shown in Table 1.11.
TABLE 1.11
Year 1999 2000 2001 2002 2003 2004
Tax $2532 $3073 $3614 $4155 $4696 $5237
Upon close examination of the data, you note that taxes increased by the same
amount every year.
$2532 $3073 $3614 $4155 $4696 $5237
1.2.1 $541 $541 $541 $541 $541
The changes in successive output, which are called first differences, are constant.
This constant increase is the incremental change where the increment is one year. Be-
cause the incremental change is constant, we know that the data represent a linear in-
crease in taxes. Because the data show the tax amount each year, this incremental
change is the rate of change of the tax amount and the slope of the underlying linear
model. If the data had shown taxes every other year, the first differences would still be
constant, indicating a linear pattern, but the value of the first differences would be
twice the slope. Slope values are always expressed per unit increase in input.
On the basis of the calculation of first differences, we make the following equiva-
lent statements:
Taxes increased by $541 per year.
The slope of the linear function described by the data is $541 per year.
The rate of change of the tax amount is $541 per year.
30 CHAPTER 1 Ingredients of Change: Functions and Models
If we consider 1999 to be our starting point, then the starting tax value is $2532.
We express the tax function verbally as follows: The tax amount began in 1999 at
$2532 and increased each year by $541. We express the function algebraically as
Tax 541t 2532 dollars
where t represents the number of years since 1999. Such models are important because
they often enable us to analyze the results of change. With certain assumptions, they
may even allow us to make cautious predictions about the short-term future. For ex-
ample, to predict the tax amount owed in 2007, we substitute t 8 into the tax model.
Tax 541(8) 2532 $6860
Admittedly, in this instance, an equation is not necessary to make such a predic-
tion, but there are many situations in which it is difficult to proceed without a model.
It is important to understand that when we use mathematical models to make
predictions about the short-term future, we are assuming that future events will follow
the same pattern as past events. This assumption may or may not be true. That does
not mean that such predictions are useless, only that they must be viewed with extreme
caution.
A Word of Caution
When you use a model to predict output values for input values that are
within the interval of input data used to obtain the model, you are using a
process called interpolation. Predicting output values for input values that
are outside the interval of the input data is called extrapolation. Because you
do not know what happens outside the range of given data, estimates
obtained by extrapolation must be viewed with caution and may result in
misleading predictions.
We have already noted that the rate of change of the tax amount is $541 per year,
which is how much taxes increase during a 1-year period. Thinking about rate of
change in this way helps us answer questions such as
• If you pay taxes twice a year, how much will taxes increase each time?
($541 per year)2 year $270.50
1
• How much will taxes increase each time if you pay taxes quarterly?
($541 per year)4 year $135.25
1
• How much will taxes increase during a 3-year period?
($541 per year)(3 years) $1623
EXAMPLE 2 Writing a Linear Model
Business Survival Table 1.12 shows the percent of U.S. companies that are still in
business after a given number of years in operation.
a. Find the constant rate of change in the percent of businesses surviving.
b. Find a model for the percent of companies still in business as a function of years
of operation. Graph the model and give a verbal description of that graph.
1.2 Linear Functions and Models 31
Solution
a. Calculating the first differences of the data in Table 1.12 indicates that the per-
cent of businesses surviving decreased by 3 percentage points per year. We
could also say that the rate of change of the survival percentage is 3 percent-
age points per year.
Caution: Note that this result is not the same as a rate of change of 3 percent
per year. When the output of a function is a percent, units on the rate of change
must be expressed in terms of percentage points per input unit.
b. The data reveal that the percent begins with 50% and that each year after the
first table value, the percent declines by 3 percentage points. We write this
mathematically as
TABLE 1.12 P(t)
3t 50 percent of businesses that are still in business
Companies t years after the fifth year in operation. Figure 1.27 depicts the graph of P.
Years (percent)
5 50 P(t)
(percent)
6 47
40
7 44
8 41 30
9 38
20
10 35
10
(Source: Cognetics, Cambridge,
Massachusetts, 1998.) t
0
5 6 7 8 9 10 Years
FIGURE 1.27
The function P(t) 3t 50, its graph, and the data in Table 1.12 all show
that the percent of U.S. companies that are still in business declines by 3 per-
centage points for each year after 5 years of operation and that at the end of 5
years, 50% of those companies are still in business. ●
The preceding examples illustrate methods of finding a linear model of the form
f(x) ax b for data points that fall on a line. However, real-life data values are sel-
dom perfectly linear. For instance, the tax data we considered earlier are not likely to
occur in real-life situations, because tax rates and the revenues of most businesses
change from year to year. Consider the following modification to the tax data:
Year 1999 2000 2001 2002 2003 2004
Tax $2541 $3081 $3615 $4157 $4703 $5242
$540 $534 $542 $546 $539
The first differences are not constant but “nearly constant.” A linear model may
be used if first differences are close to constant. Be sure to calculate first differences
only for data that are evenly spaced.
An examination of the scatter plot in Figure 1.28 reinforces our earlier observa-
tion from an examination of the first differences, that the data are close to being lin-
ear. How do we get a linear model in this situation? Use the linear regression routine
32 CHAPTER 1 Ingredients of Change: Functions and Models
Tax
(dollars)
5000
4000
3000
Year
FIGURE 1.28 1999 2000 2001 2002 2003 2004
that is built into your calculator or computer to find a linear equation that fits the
data. You should find the model to be
Tax1 540.371t 1,077,663.581 dollars
where t is the year. Overdrawing the line of best fit on the scatter plot confirms our
observation that a linear model is appropriate.
Aligning Data
When the input data are years, it is often desirable to modify how the years are num-
bered to reduce the number of digits you have to enter, as well as to reduce the mag-
1.2.2 nitude of some of the coefficients in the model equation. We refer to the process of
renumbering data as aligning the data. For example, if we renumber the years in the
tax data so that 1999 is year 0, 2000 is year 1, and so on, we obtain the data shown in
Table 1.13.
TABLE 1.13
Aligned year 0 1 2 3 4 5
Tax $2541 $3081 $3615 $4157 $4703 $5242
The aligned years are described as “the number of years since 1999.” Using a cal-
culator or computer to obtain a linear model for this aligned data, we have
Tax2 540.371t 2538.905 dollars
where t is the number of years since 1999. Note that this equation and the equation
for the unaligned data have the same slope (rate of change) but differ in the vertical
axis intercept. That is, the parameter a in the equation f(x) ax b is unchanged,
but the b-value differs. Aligning the data has the effect of shifting the data (usually to
the left). This does not change the slope of the line, but it does change where the line
crosses the vertical axis.
When using a model obtained from aligned data, it is important that you keep the
description of the input variable in mind at all times. If we are willing to assume that
the rate of change remains constant at about $540.37 per year, then we can estimate
1.2 Linear Functions and Models 33
the tax in 2006 by using a value of t 2006 in the Tax1 model or a value of t 7 in
the Tax2 model (the number of years that 2006 is counted from 1999).
Tax1 540.371(2006) 1,077,663.58 dollars $6322
Tax2 540.371(7) 2538.905 dollars $6322
These models give the same output values. They are equivalent. Note that the
model with the lower-numbered input (Tax2) has the smaller intercept value. It also
has the advantage that the constant term ($2538.905) is the approximate beginning
amount shown in the table.
To summarize, when the input is years, it is desirable to align data to make the
data entry faster and to reduce the magnitude of some of the coefficients. In general,
the smaller the magnitude of the input data, the smaller the magnitude of the coeffi-
cients in the function that models the data.
Numerical Considerations: Reporting Answers
When dealing with numerical results, it is important to understand the accuracy of
the data and how precise the results need to be. For instance, imagine that you have
1
a bank account with a balance of $18,532.71 paying interest at a rate of 54% com-
pounded annually. At the end of the year, the bank calculates your interest to be
(18,532.71)(0.0525) $972.967275. Would you expect your next bank statement to
record the new balance as $19,505.67728? Obviously, when reporting numerical re-
sults dealing with monetary amounts, we do not consider partial pennies. The bank
reports the balance as $19,505.67. The precision required is only two decimal places.
You should always round numerical results in a way that makes sense in the
context of the problem.
TABLE 1.14 Generally, results that represent people or objects should be rounded to the near-
est whole number. Results that represent money usually should be rounded to the
Net sales
nearest cent or, in some cases, to the nearest dollar. Consider, however, an interna-
(millions of
Year dollars)
tional company that reports net sales as shown in Table 1.14.
A linear model for these data is y 34.95x 69,945.683 million dollars,
1998 115.6 where x is the year. If we wished to estimate net sales in 2001, we would substitute
1999 80.6 x 2001 in the model to obtain y $10.73333333 million. Should we report the
2000 45.7 answer as $10,733,333.33 or $10,733,333? Neither! When we are dealing with nu-
merical results, our answer can be only as accurate as the least accurate output data.
In this case, the answer would have to be reported as $10.7 million or $10,700,000.
You should round numerical results to the same accuracy as the least
accurate of the output data given.
Keep in mind that a number by itself is likely to be absolutely worthless. For ex-
ample, it would not make sense for an international company to publish, in its annual
report, that net sales were 10.7. This could mean 10.7 dollars, 10.7 thousand euros,
34 CHAPTER 1 Ingredients of Change: Functions and Models
10.7 million yen, and so on. The label makes a big difference in our understanding of
the number.
A number is useless without a label that clearly indicates the units involved.
Numerical Considerations: Calculating Answers
Although the correct rounding and reporting of numerical results are important, it is
even more important to calculate the results correctly. Because you must sometimes
round your answers, it is tempting to think that you can round during the calculation
process. Don’t! Never round a number unless it is the final answer that you are re-
porting. Rounding during the calculation process may lead to serious errors. Your
calculator or computer is capable of working with many digits. Keep them all while
you are still working toward a final result.
When you use your calculator or computer to fit a function to data, it finds the
parameters in the equation to many digits. Although the text shows rounded coeffi-
cients and it may be acceptable to your instructor for you to round the coefficients
when reporting a model, make sure that you use all of the digits while working with
the model. This helps reduce the possibility of round-off error.
For example, suppose that your calculator or computer generates the following
equation for a set of data showing an airline’s weekly profit for a certain route as a
function of the ticket price.
Weekly profit 0.00374285714285x2 2.5528571428572x
52.71428571429 thousand dollars
where x is the ticket price in dollars. In the answer key, you would see the model re-
ported to three decimal places:
Weekly profit 0.004x2 2.553x 52.714 thousand dollars
However, if you used the rounded model to calculate weekly profit, your answers
would be incorrect because of round-off error. Table 1.15 shows the inconsistencies
between the rounded and unrounded models.
TABLE 1.15
Profit from Profit from
Ticket price rounded model unrounded model
$200 $298 thousand $308 thousand
$400 $328 thousand $370 thousand
$600 $39 thousand $132 thousand
Never use a rounded model to calculate, and never round intermediate
answers during the calculation process.
1.2 Linear Functions and Models 35
The number of decimal places shown in models in the text will vary. However,
whenever we calculate with a model, we use all the digits available at that point in the
text. If the data are given, the unrounded model will be used in calculations. When a
model found earlier is used in a later section and the data are not repeated, all calcu-
lations will be done with the rounded model given in that later section. For conven-
ience and consistency, the answer key will report all models with three decimal places
in coefficients and six decimal places in exponents.
The Four Elements of a Model
It is important to print the A model is an equation describing the relationship between an output variable and an
fewest digits possible under the input variable, together with their defining statements. The first element of a model
conditions that (a) the rounded
model visually fits the data, and
is the equation itself. As previously noted, equations obtained from data using tech-
(b) the rounded model gives nology generally have coefficients with many digits. When using the model, do not
values fairly close to answers round the coefficients, but when you are reporting the model, it may be acceptable to
obtained with the full model. In your instructor if you round the parameters to three or four decimal places as we have
particular, if there is a difference
between the results from the
done throughout the text.
full model and those from the The second element of a model is a description of what the output variable rep-
rounded model, that difference resents. Look back to the models we obtained for the small business’s tax. Tax is what
should appear only in the last is measured, and dollars is a label telling how it is measured. Always label a model
digit for which we can claim
accuracy. equation with output units to indicate how the output is measured.
The third element of a model is the description of the input variable. If all that
was written for the tax model had been Tax2 540.371t 2538.905 dollars,
you might have erroneously predicted that your federal taxes in 2006 would be
540.371(2006) 2538.905 $1,086,524. However, there is a statement with the
equation that reads “where t is the number of years since 1999.” Because 2006 is
the 7th year since 1999, you should correctly predict your federal taxes in 2006 to
be 540.371(7) 2538.905 $6322.
When both the data and a The final element of a model is a description of the valid input interval. The tax
model are present, the interval model was obtained from data between 1999 and 2004. It is important, when giving
information is not always
restated. The answer key usually
a model that will be used without the data from which it was obtained, that you indi-
restates the interval. cate the interval of input values over which the model is valid. This is the only way
that other people who use the model will know whether they are interpolating or ex-
trapolating. Here is the convention we will adopt: When a model is presented with-
out the data from which it was obtained, we will indicate the range of input data used.
For example, we would report the tax model by saying, “The tax between 1999 and
2004 can be modeled as Tax 540.371t 2538.905 dollars t years after 1999.” This
statement incorporates the four elements of a model.
There are four important elements of every model:
1. An equation.
2. A label denoting the units on the output.
3. A description (including units) of what the input variable represents.
4. An indication of the interval of input values over which the model is valid.
This information should be given whenever the model is presented or
used apart from the data from which it was obtained.
36 CHAPTER 1 Ingredients of Change: Functions and Models
As you proceed through this course, take the time to record your models com-
pletely and to report your answers accurately. If you practice proper reporting while
you work through the activities on linear models, you will begin to develop a habit
that will help you throughout your study of calculus in this text. In the next section
we will consider models other than the linear model. These reporting guidelines hold
true for them as well.
1.2 Concept Inventory 7. k(r) 3r 7
8. s(p) 3p 2
• Algebraic form of a linear function: f(x) ax b
In Activities 9 through 12, write an appropriate linear
• Rate of change (slope) of a linear model is constant
rise model for the given rate of change and initial output
• Slope as run value. Label the linear model.
• First differences
9. Cost increases by 30 cents per unit produced and
• Interpolation and extrapolation the fixed cost is 150 dollars.
• Calculation guidelines and rounding rules
10. Population, which was 49.5 thousand in 2000, has
• Units of measure on answers been increasing at a constant rate of 2.5 thousand
• Four elements of a model people per year.
11. Snow fell at a rate of 0.25 inch per hour. Three
inches had fallen by noon.
1.2 Activities
12. The speed increased at a constant rate of 2 mph per
Getting Started second from an initial speed of 30 mph.
For the linear functions given in Activities 1 through 4,
Applying Concepts
a. find the rate of change of the function, and b. find the
output of the function when the input is 0. Affix the 13. Profit The accompanying graph shows a corpora-
proper units to your answers. tion’s profit, in millions of dollars, over a period of
1. f(x) 3x 5 dollars where x is the year time.
2. k(t) 0.5t 3.2 million people of age t (in Profit
years) (millions
of dollars)
3. r(p) 2p 4.5 thousand dollars when p hundred
3
units are sold
2
4. p(r) 5r 12 hundred pounds for r inches of
1
rainfall
0 Years
For the linear functions given in Activities 5 through 8, 0 1 2 3 4 5
a. determine whether the slope is positive or negative, b.
tell whether the function is increasing or decreasing, and a. Is the profit function increasing or decreasing? Is
c. determine the vertical axis intercept. the slope of the graph positive or negative?
b. Estimate the slope of the graph, and write a sen-
5. f(x) 2x 4 tence explaining the meaning of the slope in this
6. g(t) 2t 5 context.
1.2 Linear Functions and Models 37
c. What is the rate of change of the corporation’s a. Is the rate of change positive or negative? Is the
profit during this time period? function increasing or decreasing?
d. Identify the points where input is zero and where b. What is the rate of change of the number of
output is zero, and explain their significance to Chapter 12 bankruptcy filings?
this corporation. c. Sketch a graph of the model. What is the slope of
14. Temperature The air temperature in a certain lo- the graph?
cation from 8 A.M. to 3 P.M. is shown in the accom- d. Find the point at which input is zero, and explain
panying graph. its significance in context.
a. Is the temperature function increasing or decreas- 17. Revenue The revenue for International Game
ing? Is the slope of the graph positive or negative? Technology was $2128.1 million in 2003 and
b. Estimate the slope of the graph, and explain its $2484.8 million in 2004. Assume that revenue was
meaning in the context of temperature. increasing at a constant rate.
c. How fast is the temperature rising between 8 A.M. (Source: Hoover’s, Inc.)
and 3 P.M.? a. Find the rate of change of revenue.
b. By how much did revenue increase each quarter
Air temperature
(°F) of 2004?
80 c. Assuming that the rate of increase remains con-
70 stant, complete the following table.
60
50 Revenue
40 Year (millions of dollars)
30 Time 2003
8 A.M. 10 A.M. noon 2 P.M.
2004
15. Organ Donors The number of organ donors in 2005
the United States between 1988 and 1996 can be 2006
modeled by
D(t) 382.5t 5909 donors d. Find an equation for revenue in terms of the
year.
t years after 1988.
(Source: Based on data from United Network for Organ Sharing.) 18. Car Value Suppose you bought a Honda Civic
Hybrid in 2005 for $24,000. In 2007 it was worth
a. Is the rate of change positive or negative? Is the
$18,200. Assume that the rate at which the car de-
function increasing or decreasing?
preciates is constant.
b. According to the model, what is the rate of
a. Find the rate of change of the value of the car.
change of the number of organ donors?
b. Complete the following table:
c. Sketch a graph of the model. What is the slope of
the graph?
Year Value
d. Find the point at which input is zero, and explain
its significance in context. 2005
2007
16. Bankruptcy Total Chapter 12 bankruptcy filings
between 1996 and 2000 can be modeled by 2008
2009
B(t) 83.9t 1063 filings
2010
t years after 1996.
(Source: Based on data from Administrative Office of U.S. c. Find an equation for the value in terms of the
Courts.) year.
38 CHAPTER 1 Ingredients of Change: Functions and Models
d. How much will the value of the car change dur- be able to sell the T-shirts at the premium price of
ing a 1-month period? Round your answer to the $14.50 because they will be in such great demand.
nearest dollar. a. Give the equations for monthly revenue and
19. Sales A house sells for $97,500 at the end of 2000 monthly cost as functions of the number of
and for $112,000 at the end of 2007. T-shirts sold.
a. If the market value increased linearly from 2000 b. How many shirts do you have to sell each month
through 2007, what was the rate of change of the to break even? Explain how you obtained your
market value? answer.
b. If the linear increase continues, what will the 23. Population In 1999, the U.S. Bureau of the Census
market value be in 2010? estimated that the world’s population had reached 6
c. In what year might you expect the market value billion people. A newspaper article about the popu-
to be $100,000? $150,000? lation stated that “Despite a gradual slowing of the
overall rate of growth, the world population is still
d. Find a model for the market value. What does
increasing by 78 million people a year. . . . [T]he
your model estimate for the market value in
number of humans on the planet could double again
2005? What assumption did you make when you
to 12 billion by 2050 if the current growth rate
created the model?
continues.”
20. Births Thirty-two percent of U.S. births that oc- (Source: “Population of World Ready to Hit 6 Billion,” Chicago
curred in 1995 were to unmarried women. The per- Tribune, October 21, 1999, p. A1.)
centage of U.S. births to unmarried women in 2002 a. According to the article, what is the rate of
was 34. change of the world’s population?
(Source: National Vital Statistics Reports, vol. 53, no. 9.) b. Find a linear model for the population of the
a. Find the rate of change of the percentage of world, assuming that the population was 6 bil-
births, assuming that the percentage of births to lion at the beginning of 2000.
unmarried women increased at a constant rate. c. According to the model in part b, when will the
b. Estimate the percentage of births to unmarried world’s population be 12 billion? Does this agree
women in 2003. with the estimate given in the article?
21. Credit Consumer credit in the United States was d. What assumption did you make when you made
$1719 billion in 2000 and $2040 billion in 2003. As- the prediction in part c?
sume that consumer credit increases at a constant 24. Heating Oil The following table gives the number
rate. of gallons of oil in a tank used for heating an apart-
(Source: Statistical Abstract, 2004–2005.)
ment complex t days after January 1 when the tank
a. Find the rate of increase. was filled.
b. On the basis of the rate of increase, estimate con-
sumer credit in 2006. Oil
c. Is the assumption that consumer credit increases t (gallons)
at a constant rate valid? Explain. 0 30,000
d. Assuming a constant rate of change, when will 1 29,600
consumer credit reach $3 trillion? 2 29,200
22. Break-even Point You and several of your friends 3 28,800
decide to mass-produce “I love calculus and you
4 28,400
should too!” T-shirts. Each shirt will cost you $2.50
to produce. Additional expenses include the rental
of a downtown building for a flat fee of $675 per a. What is the rate of change of the amount of oil?
month, utilities estimated at $100 each month, and b. How much oil can be expected to be used during
leased equipment costing $150 per month. You will any particular week in January?
1.2 Linear Functions and Models 39
c. Predict the amount of oil in the tank on January a. Find a linear model for the volume of cargo han-
30. What assumptions are you making when you dled by a team during an x% slowdown.
make predictions about the amount of oil? b. Use the model to estimate the volume handled if
d. Find and graph an equation for the amount of oil no slowdown occurs.
in the tank in terms of the number of days since c. For your model, what are the points at which in-
January 1. put is zero and output is zero? Discuss in context
25. Postal Rates The following table shows first-class the information given by these points.
U.S. domestic postage for mail up to 9 ounces in the d. What is the slope for your model? Interpret the
year 2006. slope in context.
27. Enrollment The accompanying table shows the
Weight not exceeding Postage enrollment for a particular university from 1965
1 oz $0.39 through 1969.
2 oz 0.63 Year 1965 1966 1967 1968 1969
3 oz 0.87 Students 5024 5540 6057 6525 7028
4 oz 1.11
a. Find a linear model for these data.
5 oz 1.39
b. Use the model to estimate the enrollment in
6 oz 1.59
1970.
7 oz 1.83
c. The actual enrollment in 1970 was 8038 stu-
8 oz 2.07 dents. How far off was your estimate? Do you
9 oz 2.31 consider the error to be significant or insignifi-
(Source: U.S. Postal Service.)
cant? Explain your reasoning.
d. Would it be wise to use the model to predict the
enrollment in the year 2000? Explain.
a. Observe a scatter plot of these data. Determine
visually whether a linear model is appropriate. 28. Cost You are an employee in the summer at a
souvenir shop. The souvenir shop owner wants to
b. Verify your observations in part a by calculating purchase 650 printed sweatshirts from a company.
first differences in the postal rates. The catalog contains a table of costs and includes di-
c. Find a formula for the postage in terms of rections to call the company for costs on orders
weight. Be specific about what the variables rep- greater than 350. The catalog costs are shown in the
resent. table. The shop owner, who has tried unsuccessfully
26. Union Pressure In order to put pressure on a for a week to contact the company, asks you to esti-
company to negotiate a contract that is more favor- mate the cost for 650 shirts.
able to employees, a workers’ union may order a
Number Total cost Number Total cost
slowdown on labor. Consider, for instance, a slow-
purchased (dollars) purchased (dollars)
down of laborers at ports along the U.S. Pacific
seaboard. The accompanying table shows the vol- 50 250 250 700
ume (in cartons) of cargo that is unloaded each day 100 375 300 825
by one team of cargo handlers during a work slow- 150 500 350 950
down when cargo handlers reduce their efficiency
200 600
by x%.
a. Find a linear model to fit the data.
Percent slowdown 10 25 50 65 80 b. Use the model to predict the cost for 650 shirts.
Note that all costs in the table are integer multi-
Cartons 108 90 60 42 24
ples of $25.
40 CHAPTER 1 Ingredients of Change: Functions and Models
c. Determine the average cost per shirt for 650 shirts. b. At what rate was the amount of greenhouse-gas
d. The shop owner is preparing a newspaper adver- emissions increasing between 1990 and 2002?
tisement to be published in a week. If the c. Assuming that the constant rate of change of the
standard markup is 700%, what should the ad- model will continue to apply, what aggregate
vertised price be? amount of greenhouse gases will be released into
e. How many of the 650 shirts will need to be sold at the atmosphere in 2012?
the price determined in part d in order to pay for
the cost of all 650 shirts (the break-even point)? Discussing Concepts
29. Kyoto Emissions The aggregate emissions by the 30. Explain the difference between using a model for in-
United States of the six greenhouse gases as defined terpolation and using it for extrapolation. Does in-
in the Kyoto Protocol can be modeled terpolation always give an accurate picture of what
as e(t) 0.107t 5.11 million gigagrams (CO2 is happening in the real world? Does extrapolation?
equivalent), where t is the number of years since Why or why not?
1990. The model is based on information for 1990
through 2002. (Source: FCCC/CP/2004/5 and FCCC/WEB/ 31. Discuss the necessity for each of the four elements of
2004/3 available at http://unfccc.int) a model.
a. What were the aggregate emissions of the six 32. Compare and contrast the terms slope, steepness,
greenhouse gases released into the atmosphere in and rate of change as they are used to describe linear
1990? in 1997? in 2002? models.
1.3 Exponential and Logarithmic Functions and Models
In Section 1.2 we studied linear functions whose output resulted from the repeated
addition of a constant at regular intervals. We now turn to a function whose output
is the result of repeated multiplication by a constant at regular intervals.
Concavity and Exponential Growth and Decay
For example, in the case of a bacterial culture that starts with 10,000 cells and doubles
every hour, the current size of the culture is determined by repeated multiplication.
The size of the culture at the end of each of the first 4 hours is shown in Table 1.16.
TABLE 1.16
P(t)
Cells Hour 0 1 2 3 4
Cells 10,000 20,000 40,000 80,000 160,000
140,000
100,000
When we examine a scatter plot of the culture size over the first 10 hours, we
60,000 see that the points certainly do not fall on a line. (See Figure 1.29.) In fact, these
points seem to lie along a specific curve. We describe curvature in terms of con-
20,000 t cavity. In this case, the graph is concave up, not because it is increasing from left
0 1 2 3 4 Hour to right, but because it appears to be a portion of an arc opening upward.
Concave up and increasing The culture grows more rapidly in the later hours. We note that the starting
FIGURE 1.29 culture size is 10,000 cells and that the size each hour is twice the prior size. The
1.3 Exponential and Logarithmic Functions and Models 41
repeated multiplication by 2 makes the hour variable t appear as an exponent on 2;
that is, the hour value counts the number of times that 2 has been used as a multiplier.
A model for the bacterial culture size is
P(t) 10,000(2t) cells
t hours after the culture was first counted. The curve underlying the scatter plot in
Figure 1.29 is the graph of the function P.
Because the variable t appears in the exponent, we call the equation an exponen-
tial equation. If the multiplier is greater than 1, the change is referred to as expo-
nential growth.
f (x) Exponential equations also arise when we see a decreasing amount of the original
Grams substance, such as in the study of radioactive material. For example, if 400 grams of a
400 radioactive substance decays by 2% per day, then the amount of the substance re-
maining each day is 98% of the previous day’s amount. The amount of the radioac-
300
tive substance after x days of decay is given by the model
200
f(x) 400(0.98x) grams
100
In this case, the multiplier is between 0 and 1. Such a situation, wherein an
x
amount diminishes by a constant multiplier, is referred to as exponential decay.
0 40 80 120 160 Days
Looking at a graph of f in Figure 1.30, we see that the graph again appears to be a por-
Concave up and decreasing tion of an arc opening upward, even though the outputs are decreasing. We say that
FIGURE 1.30 the graph of f is concave up and decreasing.
Any change in a quantity that results from repeated multiplication by a constant
f (x)
generates an exponential function. In general, if we start with an amount a and mul-
x tiply by a constant positive factor b each year, then the quantity that we have at the
end of x years is given by the exponential equation
f(x) abx
defining f as a function of x. In the cases where a is positive, the function outputs are
positive and the graphs of the function f will be similar to those in Figures 1.29 and
1.30. However, when a is negative, the function outputs are negative and the graphs
of the function f will be similar to one of the two graphs in Figure 1.31.
(a) Concave down The graph in Figure 1.31a is concave down and decreasing. The graph in Figure
and decreasing 1.31b is concave down and increasing. We do not often use negative exponential
functions as models, but they do exist.
f (x)
x Percentage Change and Exponential Models
Linear functions exhibit a constant rate of change, but exponential functions exhibit
a constant percentage change. Percentage change occurs when the amount of
growth or decay is determined by the current size. For the exponential equation
f(x) abx , we determine the constant percentage change by calculating
(b 1)100%. For the bacterial growth model P(t) 10,000(2t) cells, the constant
percentage change is (2 1)100% 100%; that is, the bacterial population in-
creases by 100% each hour. For the remaining amount of radioactive substance
(b) Concave down given by f(x) 400(0.98x) grams after x days of decay, the constant percentage
and increasing change is (0.98 1)100% (0.02)100% 2%. Thus the amount decreases by
FIGURE 1.31 2% each day.
42 CHAPTER 1 Ingredients of Change: Functions and Models
Exponential Model
Verbally: An exponential model has a constant percentage change.
Algebraically: An exponential model has an equation of the form
f(x) abx
An alternative form of the where a
0 and b 0. The percentage change is (b 1)100%, and the
exponential model f(x) abx is parameter a is the output corresponding to an input of zero.
f(x) aekx.
It is easy to convert from one Graphically: An exponential model graph with positive a has the form of one
form to the other using k ln b of the two graphs in Figure 1.32.
and b ek .
abx aekx → bx ekx → f(x) f (x)
b ek → k ln b
a a
x x
(a) Exponential growth (b) Exponential decay
f(x) = ab x with b > 1 f(x) = ab x with 0 < b < 1
FIGURE 1.32
Note that the equation in the box defines a function f with input x. From now on,
when we refer to an exponential equation or function, we will consider the equation
to be of the form f(x) abx. In the case of either exponential growth or exponential
decay (as long as a 0), a graph of the exponential function f(x) ab x is concave
up or . Its output values approach zero in one direction and increase
without bound in the opposite direction.
EXAMPLE 1 Using Percentage Change to Write an Exponential Model
iPods Apple introduced the iPod™, an MP3 digital music player, on October 23,
2001. iPod sales were 0.14 million units the first year and until 2005 increased
approximately 260% each year. Note: Apple’s fiscal year ends in September. (Source:
Based on data from “Apple Reports Fourth Quarter Results,” 2003–2005, Apple
Computer, Inc.)
a. Why is an exponential model appropriate for iPod sales?
b. Find a model for iPod sales.
c. According to the model, what were 2006 iPod sales expected to be?
d. In November of 2005, Needham & Co. analyst Charles Wolf forecasted 2006
iPod sales to be 23.5 million units. (Source: http://www.macworld.co.uk/
news/ posted on November 26, 2004.) How close is your estimate of 2006 sales
to Wolf’s forecast?
1.3 Exponential and Logarithmic Functions and Models 43
Solution
a. An exponential model is appropriate because the percentage change is con-
stant.
b. Because the percentage change is 260% per year, the constant multiplier is
1 2.6 3.6. The sales of iPods can be modeled by the equation
s(x) 0.14(3.6x) million units
where x 0 is the fiscal year ending in September 2002, 0 x 3.
c. The input for the fiscal year ending in 2006 is x 4. According to the model,
the sales of iPods in 2006 were expected to be s(4) 0.14(3.64) 23.5 million
units
d. The model gave essentially the same estimate for 2006 sales as that given by
Wolf. ●
Percentage Differences and Modeling from Data
Often we do not know what repeated multiplier we should use to write an exponen-
tial model. However, if we are given data, sometimes we can look at percentage dif-
ferences to determine the repeated multiplier. Percentage differences are calculated
from data with increasing input values by dividing each first difference by the output
value of the lesser input value and multiplying by 100.
We look at how to calculate percentage differences by considering a small town’s
dwindling population. After closing a state-subsidized steel mill near a small town,
the state has funded a six-year mandate for the town council to study the impact of
the closing. The population data (measured at the end of each year) are shown in
Table 1.17, and a scatter plot of the data is shown in Figure 1.33.
TABLE 1.17 f(x)
Year of mandate Population
0 7290
3
1 5978
x
2 4902
3 4020
4 3296
5 2703
FIGURE 1.33
6 2216
Because the inputs are evenly spaced, we examine the output data in greater detail
by calculating first differences:
7290 5978 4902 4020 3296 2703 2216
1312 1076 882 724 593 487
44 CHAPTER 1 Ingredients of Change: Functions and Models
As indicated by the shape of the scatter plot in Figure 1.33, the first differences
show us that the change in population from year to year is not constant. However, if
we calculate the percentages that these yearly changes represent, we notice a pattern.
During the first year, the population decreased by 1312 people. This represents an
18% (approximate) decrease in one year from the initial population of 7290 people.
1312 people
0.17997 18%
7290 people
During the second year, the population decreased by 1076 people. This also rep-
resents about an 18% decrease in one year from the previous year’s population.
1076 people
The percent symbol (%) means 0.17999 18%
to divide by 100. For example, 5978 people
8
8% 100 0.08. When In fact, every year the population decreased by approximately 18%.
changing a decimal number into
a percent, rewrite the number 882 people
using “divide by 100.” For
Third year: 0.17993 18%
4902 people
439
example, 4.39 100 . When you
724 people
remove “divide by 100,” insert Fourth year: 0.18010 18%
439 4020 people
“%.” That is, 100 439%.
593 people
Fifth year: 0.17992 18%
3296 people
487 people
Sixth year: 0.18017 18%
2703 people
1.3.1 We can use technology to determine an equation for the data. Enter the data
points and use your calculator or computer to fit an exponential equation. You
should obtain the model
P(x) 7290.366(0.819995x) people
where x is the number of years since the beginning of the mandate, 0 x 6.
If we assume that the population continues to decline by the same percentage (ap-
proximately 18%), we can estimate that at the end of the seventh year after the mill
closing, the town’s population was 82% of the population at the end of the study:
2216(0.82) 1817 people
This estimation can also be computed using the unrounded equation for population
with input x 7:
T(7) 7290.366(0.8199957) 1817 people
This estimate from the model agrees with the one calculated using only the data.
Although both of these predictions are valid, for consistency we adopt the
following rule of thumb:
Once an equation has been fitted to data, we will use the equation to answer
questions rather than using data points or rounded estimates.
The technology-generated equation indicates that the starting value of the popu-
lation was 7290.366 and that the population declined by 18.0005%. However, round-
1.3 Exponential and Logarithmic Functions and Models 45
ing these values to report reasonably, we would say that the starting value of the pop-
ulation was 7290 and the percentage change was 18%. Note that these values agree
with the data and with the percentage differences calculated. That is because, in this
case, the data have input values that are 1 unit apart and the percentage differences
are approximately constant. When these two conditions are true, the constant per-
centage differences equal the percentage change, (b 1)100%, and we can use the
terms interchangeably.
Doubling Time and Half-Life
One property of exponential models is that when the quantity being modeled either
doubles (during exponential growth) or halves (during exponential decay), it does so
over a constant interval. For instance, the value of an initial investment of $1000 that
increases by 8% each year can be modeled by the exponential function
I(t) 1000(1.08t) dollars
t years after the initial investment. How long will it take for the investment to be
worth $2000? We find this out by solving the equation 2000 1000(1.08t) for t by
using either technology or algebra. We find that t 9.00646. The investment will be
worth $2000 by t 9 years. In order to find out when it will double again, we solve
4000 1000(1.08t). The investment will double again by t 18 years. We say that 9
years is the doubling time of the investment. In general, the doubling time in expo-
nential growth is the amount of time it takes for the output to double (or the time it
takes for the output to increase by 100%).
Similarly, the half-life in exponential decay is the amount of time it takes for the
output to decrease by half (or the time it takes for the output to decrease by 50%).
Doubling time is the amount of time it takes for the output of an increasing
exponential function to double.
Half-life is the amount of time it takes for the output of a decreasing
exponential function to decrease by half.
Example 2 illustrates the process of building an exponential decay model given in-
formation about half-life.
EXAMPLE 2 Writing an Exponential Decay Model
Medicine Dilantin* is a drug used to control epileptic seizures. On Monday a pa-
tient takes a 300-mg Dilantin tablet at 4 P.M. Eight hours later the Dilantin reaches its
peak plasma concentration of 15 g/mL. The average plasma half-life of Dilantin is
22 hours. (Plasma half-life is the amount of time it takes for the concentration of the
drug in the plasma to reach half of its peak concentration.)
a. Write a model for the concentration of Dilantin in the patient’s plasma as a
function of the time after peak concentration is reached.
* Based on information obtained from www.parke-davis.com (accessed 6/11/00).
46 CHAPTER 1 Ingredients of Change: Functions and Models
b. What is the concentration of Dilantin in this person’s plasma at 8 A.M. on
Tuesday?
c. The minimum desired concentration of Dilantin is 10 g/mL. Will the con-
centration dip below that level before the patient takes a second tablet on Tues-
day at 4 P.M.?
Solution
a. The input of the model is the time after the peak concentration that occurs at
midnight on Monday, and the initial amount we consider is the peak concen-
tration of 15 g/mL. After 22 hours, the concentration will be half its peak, or
7.5 g/mL. Using the two data points (0, 15) and (22, 7.5), we obtain the model
C(h) 15(0.96898h) g/mL
is the concentration of Dilantin in a person’s plasma, h hours after midnight,
h 0.
b. Eight hours after midnight, the plasma concentration is
C(8) 15(0.968988) 11.7 g/mL
c. To determine when the plasma concentration reaches 10 g/mL, solve the
equation 10 15(0.96898h) using technology or algebra. You should obtain
h 12.9 hours. It takes approximately 12.9 hours after the peak concentration
at midnight for the concentration to reach 10 g/mL. This corresponds to
approximately 1 P.M. on Tuesday. Thus the concentration will be below
10 g/mL when the patient takes another tablet at 4 P.M. on Tuesday. ●
Aligning Exponential Data
As you work with exponential models, keep in mind the following principle:
When using technology to find the equation for an exponential model, align the
input data using reasonably small values to avoid numerical computation errors such
as overflow or round-off errors.
Graphically, this aligning of input values does not change the nature of the expo-
nential function. Aligning input values simply shifts the graph of the function to the
left, as illustrated in Figure 1.34.
Aligning input
values shifts the
graph horizontally
0 x
0
Original data and function in black. Aligned
data and function in teal.
FIGURE 1.34
1.3 Exponential and Logarithmic Functions and Models 47
An exponential function is characterized by increasingly rapid growth (or in-
creasingly slow decay), so its graph in one direction approaches and stays close to the
horizontal axis (namely y 0) and, in the other direction, tends to increase more
and more quickly. (See Figures 1.35a and 1.35b.) There is another class of functions
that exhibit the opposite behavior; that is, they are defined for only positive numbers
and start by hugging the vertical axis (namely x 0), and as x increases, the outputs
f(x) increase (or decrease) but more and more slowly. (See Figures 1.35c and 1.35d.)
This class of functions is known as the logarithmic functions. In the following dis-
cussion, we turn our attention to situations that lead to logarithmic models.
Output Output
Input Input
(a) Exponential (b) Exponential
Output Output
Input Input
(c) Logarithmic (d) Logarithmic
FIGURE 1.35
Logarithmic Models
Altitude and air pressure are intrinsically related. Altimeters are instruments that
determine altitude by measuring air pressure. For an altimeter, the input is air pres-
sure and the output is altitude. Table 1.18 shows altimeter data, and Figure 1.36
shows a scatter plot of the data.
Although the scatter plot has the basic declining, concave-up appearance of an ex-
ponential function, an exponential function does not fit these data well. However,
there is another function, called a logarithmic function or log function, that exhibits
similar behavior. The log equation used by most technologies is
f(x) a b ln x
This equation defines a function f with input x. The b-term in this equation de-
termines whether the function increases or decreases and how rapidly the increase or
decrease occurs. The a-term determines the vertical shift of the function. From now
48 CHAPTER 1 Ingredients of Change: Functions and Models
TABLE 1.18 Altitude
(thousands of feet)
Air pressure Altitude
(inches of mercury) (thousands of feet) 100
13.76 20
5.56 40
2.14 60
0.82 80
0.33 100 20 Air pressure
0 7 14
(inches of
mercury)
FIGURE 1.36
on, when we refer to a log equation or function, we will consider the equation to be
of the form f(x) a b ln x.
The end behavior of log functions is particularly important in determining when
they are appropriate to use to fit data. An increasing log function increases without
bound as the input increases, and a decreasing log function decreases without bound
as the input increases. Log functions are not defined for negative or zero input, and
as the input approaches zero from the right, the function either increases or decreases
without bound.
Logarithmic Model
Verbally: A logarithmic (log) function has a vertical asymptote (the line
x 0) and continues to grow or decline as x becomes large.
Algebraically: A log model has an equation of the form
f(x) a b ln x
where b
0.
Graphically: The graph of a log model has the form of one of the two graphs
shown in Figure 1.37.
f(x) f(x)
x x
(a) f(x) = a + b ln x with b > 0 (b) f(x) = a + b ln x with b < 0
FIGURE 1.37
1.3 Exponential and Logarithmic Functions and Models 49
Again consider the graphs in Figure 1.37. Note that log functions exhibit increas-
ingly smaller changes in output for constant changes in input. This slow growth or
decline characterizes log functions. The characteristic that makes log functions dif-
ferent from any other functions that we will study is that their growth or decline be-
comes increasingly slower but their output never approaches a horizontal limiting
value, as do the outputs of declining exponential functions and, as we will see later,
logistic functions.
Returning to the altitude example, we use technology to find the log model that
fits the data in Table 1.18.
A(p) 76.174 21.331 ln p thousands of feet above sea level
where p is the air pressure in inches of mercury. Figure 1.38 shows a graph of the func-
tion A overdrawn on a scatter plot of the data.
A(p)
Altitude
(thousands of feet)
100
p
20 Air pressure
(inches of
FIGURE 1.38 0 7 14 mercury)
EXAMPLE 3 Finding a Log Model
Investment An international investment fund manager models bond rates of
countries as a tool when making investment decisions. The manager uses the data in
Table 1.19 to create a yield curve for Germany, where long-term bond rates are higher
than short-term rates.
TABLE 1.19
Time to German bond Time to German bond
maturity (years) rate (percent) maturity (years) rate (percent)
1 3.60 6 4.65
2 4.10 7 4.75
3 4.25 8 4.80
4 4.40 9 4.90
5 4.50 10 4.95
(Source: Investment Digest, VALIC, vol. 12, no. 1, 1998.)
50 CHAPTER 1 Ingredients of Change: Functions and Models
a. Sketch a scatter plot of the data.
b. Find a log model for the data.
c. This investment manager estimates a 5.25% rate for 20-year bonds and a 5.50%
rate for 30-year bonds. How closely does the log model match these estimates?
Solution
a. The scatter plot in Figure 1.39 suggests the slow growth modeled by a log
function.
Bond rate
(percent)
5.0
4.5
4.0
Time
3.5
1 5 10
(years)
FIGURE 1.39
b. A log model for the data is
R(t) 3.6296 0.5696 ln t percent
is the yield of a German bond with a maturity time of t years, 1 t 10.
c. The model in part b predicts the following rates for 20- and 30-year bonds:
R(20) 5.34, and R(30) 5.57
These predictions are slightly higher than the estimates made by the fund
manager. ●
Aligning Log Data
Recall that aligning yearly data to small input values is convenient when finding a lin-
ear model and essential when finding an exponential model so that the coefficients in
the equation will not be unnecessarily large values and to avoid round-off error. In
both the linear and exponential cases, aligning input does not affect how well the
model fits the data. It simply causes a graph of the model to be shifted horizontally
closer to or farther from the origin.
However, because log models have the property that the output approaches neg-
ative or positive infinity as the inputs approach 0 from the right, differently aligned
data result in better- or worse-fitting models. For this reason, we will give a recom-
mended input alignment when we ask you to model using log functions for data that
1.3 Exponential and Logarithmic Functions and Models 51
need to be shifted. When aligning the input data for a log model, it is important to
remember that the log function is not defined for negative or 0 input. If you align the
data so that the first input value is 0, your calculator or computer will return an error
message when you attempt to find a log equation.
Creating Inverse Functions
Given a function, we can sometimes create a new function by reversing the input and
output of the original function. We can use this new function as an inverse function.
To illustrate, we again consider the German bond rates in Table 1.19. Figure 1.40
shows an input/output diagram for this function.
t
Input Years
Rule R
Output
R(t)
FIGURE 1.40 Percent
If we exchange the input values with the output values for this function, we have
Table 1.20.
TABLE 1.20 r
Input Percent
German bond Time to maturity
rate (percent) (years)
3.60 1
Rule T
4.10 2
4.25 3
4.40 4 Output
T(r)
4.50 5 Years
4.65 6 FIGURE 1.41
4.75 7
4.80 8
4.90 9
4.95 10
Each bond rate (input) corresponds to only one maturity value (output), so Table
1.20 represents a function, the inverse of the original function. Figure 1.41 shows an
input/output diagram for the inverse function.
52 CHAPTER 1 Ingredients of Change: Functions and Models
When an equation represents a function, finding an algebraic expression for an
inverse function requires that you solve for the input variable of a function in terms
of the output variable. Sometimes this is simple, sometimes it is difficult, and some-
times it is impossible. Because most of the models we use in this text are built from
data, if we need a model to approximate the inverse relationship, we simply invert the
data and model the inverted data. In most cases, this technique will not yield the ex-
act inverse function but will yield a good approximation of the inverse function.
In this case, modeling the inverted data with an exponential function yields the
time to maturity as
T(r) 0.0018(5.723r) years
where r is the bond rate (as a percentage), 3.60 r 4.95.
You may recall from previous math courses that when a function f is used as the
input for its inverse function g, the result is simply the input variable. That is, when a
function and its inverse are combined to create a composite function, the composi-
tion has the result of “undoing” the effects of the two functions.
Composition Property of Inverse Functions
If f and g are inverse functions, then
f(g(x)) (f g)(x) x and g(f(x)) (g f)(x) x
In the German bond rate example, the functions R(t) 3.6296 0.5696 ln t
and T(r) 0.0018(5.723r) are approximate inverses of each other.
As we noted earlier, the log functions and exponential functions have opposite
characteristics. If we have data whose input/output relationship can be modeled by
an exponential function, then the inverse (output/input) relationship can be modeled
by a log function, and vice versa.
Exponential functions, with their constant percentage change, and log functions,
with their slowing but not leveling end behavior, are common in health and life
sciences and in social sciences. Their importance leads us to include them in our
discussion. As we continue our exploration of calculus concepts, you will become
familiar with both log and exponential functions.
The relationship between exponential and logarithmic functions in its simplest form can be
stated as follows:
If f(x) ln x and g(x) ex, then f(g(x)) ln (ex ) x and g(f(x)) eln x x
as long as x is positive.
The inverse relationships for the model equations are more involved but still exist:
a 1
If f(x) a b ln x with b
0, then f 1(x) ABx, where A e b and B e b.
If f(x) abx, then for x 0, b 0, and b
1, f 1(x) A B ln x , where
ln a 1
A and B
ln b ln b
1.3 Exponential and Logarithmic Functions and Models 53
7. f(x) 2 ln x 8. f(x) 4 ln x
1.3 Concept Inventory f(x) 4 ln x f(x) 2 ln x
f (x) f (x)
• Exponential function: f(x) abx
• Exponential growth and decay
• Percentage change
• Doubling time and half-life x x
1 1
• Log function: f(x) a b ln x
• End behavior of log and exponential functions
• Log and exponential functions as inverses
For Activities 9 through 12, indicate whether the func-
tion describes exponential growth or decay, and give the
1.3 Activities constant percentage change.
9. f(x) 72(1.05x) 10. K(r) 33(0.92r)
Getting Started
For Activities 1 through 8, match each graph with its 11. y(x) 16.2(0.87x) 12. A(t) 0.57(1.035t)
equation. For Activities 13 and 14, find the constant percentage
1. f(x) 2(1.3x) 2. f(x) 2(1.3x) change, and interpret it in context.
f(x) 2(0.7x) f(x) 2(1.3x) 13. Bacteria After h hours, the number of bacteria in
f(x) a petri dish during a certain experiment can be
f (x) modeled as B(h) 100(0.61h) thousand bacteria.
2 14. Membership The membership of a popular club
x can be modeled by M(x) 12(2.5x) members by
2
–2
the end of the xth year after its organization.
x
Applying Concepts
3. f(x) 3(1.2x) 4. f(x) 2(0.8x) 15. Imports The U.S. Energy Information Adminis-
tration projected that imports of petroleum prod-
f(x) 3(1.4x) f(x) 2(0.6x)
ucts in 2005 would be 4.81 quadrillion Btu and
f(x) f (x) would increase by 5.47% each year through 2020.
(Source: Annual Energy Outlook, 2001.)
a. Find a model for projected petroleum product
3 2 imports between 2005 and 2020.
x x b. According to the model, when will imports ex-
ceed 10 quadrillion Btu?
5. f(x) 2 ln x 6. f(x) 3 ln x
c. Describe the end behavior of the model as time
f(x) 2 ln x f(x) ln x increases.
f (x) f (x)
16. Assets In 1994, Charles Schwab & Co. had ap-
proximately $135 billion of assets in customer ac-
3
counts, and throughout the late 1990s that value
grew by approximately 39% each year.
x x a. Find a model for Schwab’s customer account
1 1
assets in the late 1990s.
b. Use the model to estimate Schwab’s customer
account assets in 2000.
54 CHAPTER 1 Ingredients of Change: Functions and Models
17. Social Security According to the Social Security Ad- a. Find an exponential model for the data.
visory Board, the number of workers per beneficiary b. According to the model in part a, what was the
of the Social Security program was 3.3 in 1996 and is yearly percentage growth in recycled MSW from
projected to decline by 1.46% each year through 2030. 1960 through 2000?
a. Find a model for the number of workers per ben- 21. Computing Power During the last three decades,
eficiary from 1996 through 2030. Note that the computing power has grown enormously. The ac-
unit of measure is workers per beneficiary. Keep companying table gives the number of transistors
one decimal place accuracy in your answer. (in millions) in Intel processor chips.
b. What does the model predict the number of
workers per beneficiary will be in 2030? How will Processor Year Transistors (millions)
this number affect your life? 4004 1971 0.0023
18. Emissions In 1975 the EPA emissions standard 80286 1982 0.134
for cars was 3.1 grams of nitrogen oxide per mile of 386DX 1986 0.275
driving. Assume that the EPA standard decreased by 486DX 1989 1.2
9.3% each year between 1975 and 2000. Pentium® 1993 3.1
a. Find a model for the emission standard between Pentium® Pro 1995 5.5
1975 and 2000.
Pentium® II 1997 7.5
b. Estimate the EPA standard in 2000. Celeron 1999 19.0
19. Sales When a company stops advertising and pro- Pentium IV 2000 42.0
moting one of its products, sales often decrease ex- Pentium IV 2002 55.0
ponentially, provided that other market conditions
Pentium M 2003 77
remain constant. At the time that publicity was dis-
continued for a newly released popular animated Pentium M 2004 140
film, sales were 520,000 videotapes per month. One Itanium 2 2005 600
month later, videotape sales had fallen to 210,000 (Source: www.Intel.com)
tapes per month.
a. Find an exponential model for the data.
a. What was the monthly percentage decline in sales?
b. According to the model found in part a, what is the
b. Assuming that sales decreased exponentially, annual percentage increase in the number of tran-
give the equation for sales as a function of the sistors used in an Intel computer processor chip?
number of months since promotion ended.
c. On April 19, 1965, Intel co-founder Gordon
c. What will sales be 3 months after the promotion Moore predicted that the number of transistors
ends? 12 months after? on a computer chip would double approximately
20. Waste The Environmental Protection Agency re- every 2 years. This prediction is known as
ports the total amount of municipal solid waste Moore’s Law. Do the data (and does the model)
(MSW) recycled between 1960 and 2000 as shown support Moore’s Law?
in the accompanying table. 22. Farms The number of U.S. farms with milk cows
has been declining since 1980. See the accompany-
Total MSW recycling ing table.
Year (million tons)
1960 5.6 Year Farms (thousands)
1970 8.0 1980 334
1985 269
1980 14.5
1990 193
1990 33.2
1995 140
2000 68.9
2000 105
(Source: “Municipal Solid Waste in the United States:
2000 Facts and Figures,” Executive Summary.) (Source: Statistical Abstract, 1998 and 2001.)
1.3 Exponential and Logarithmic Functions and Models 55
a. Find an exponential model for these data. What c. What is the limit of the function in part b as time
is the percentage change indicated by the model? approaches infinity? Interpret this answer in the
b. Express the end behavior of the model as time in- context of the radon gas.
creases. Does this end behavior reflect what you 26. Half-Life The elimination half-life of a certain
believe will happen in the future to the number type of penicillin is 30 minutes.
of farms with milk cows? a. Write a model for the amount of this penicillin
23. Bottled Water Consumption The per capita con- left in a person’s body if the initial dose is 250 mg.
sumption of bottled water in the United States has b. If it is safe to take another dose of this penicillin
increased dramatically in the past 20 years. The ac- once the amount in the body is less than 1 mg,
companying table shows selected years and the per when should another dose be taken?
capita bottled water consumption in those years. 27. Medicine If a person takes Digoxin, a heart stim-
ulant, the concentration in the person’s blood
Bottled water consumption stream t hours after the Digoxin reaches its peak
Year (gallons per person per year) concentration is
1980 2.4
D(t) D0e0.0198t g/mL
1985 4.5
where D0 is the peak concentration in micrograms
1990 8.0
per milliliter. Find the half-life of Digoxin.
1995 12.1
28. Doubling Time The amount of an investment of
2000 17.4
P dollars with 8% interest compounded continu-
2003 22.0 ously is modeled by the equation
(Source: USDA/Economic Research Service. Last updated A(t) Pe0.08t dollars
December 21, 2004.)
t years after the initial investment. How long would
a. Find linear and exponential models for the data. it take this investment to double?
Graph the equations for these models on a scat-
29. Milk Storage According to the back of a milk car-
ter plot of the data. Which model do you think
ton sold by Model Dairy, the number of days that
better describes the per capita bottled water con-
milk will keep when stored at various temperatures
sumption?
is as shown in the table.
b. Give the rate of change of the linear model and
the percentage change of the exponential model. Temperature
(degrees Fahrenheit) Days
c. Use the two models to estimate bottled water
30 24
consumption in 2005.
38 10
d. According to each model, when will per capita
bottled water consumption exceed 25 gallons per 45 5
person per year? 50 2
24. Decay Carbon-14, 14C,has a half-life of approxi- 60 1
mately 5580 years. If a sample of an artifact contains 70 0.5
0.027 gram of 14C, how long ago did it contain 0.05
a. Find an exponential model for the data.
gram of 14C?
b. If a refrigerator is adjusted to 40°F from 37°F,
25. Decay An abandoned building is found to con- how much sooner will milk spoil when stored in
tain radioactive radon gas. Thirty hours later, 80% this refrigerator?
of the initial amount of gas is still present. c. Invert the data, and fit a log model to the in-
a. Find the half-life of this radon gas. verted data.
b. Give a model for the amount of radon gas pres- d. At what temperature should the refrigerator be
ent after t hours. set in order to keep the milk for one week?
56 CHAPTER 1 Ingredients of Change: Functions and Models
30. Weight The body weight of mice used in a drug d. Calculate R T for p 9.4, 7.5, and 7.1 and
experiment is recorded by the researcher. The data T R for y 2, 4, and 10. How do the calcula-
are given in the accompanying table. tions support R and T being approximate inverse
functions?
Age beyond 2 weeks
(weeks) Weight (grams) 32. Contaminant The American Association of Pedi-
1 11 atrics has stated that lead poisoning is the greatest
3 20
health risk to children in the United States. Because
of past use of leaded gasoline, the concentration of
5 23
lead in soil can be described in terms of how close
7 26 the soil is to a heavily traveled road. The accompa-
9 27 nying table shows some distances and the corre-
sponding lead concentrations in parts per million.
(Source: Estimated from information given in “Letters to Nature,”
Nature, vol. 381 (May 30, 1996), p. 417.)
a. Find a log model for a mouse’s weight, G, in Distance from road Lead concentration
terms of its age, w weeks. (meters) (ppm)
b. Estimate the weight of the mice when they are 4 5 90
weeks old. 10 60
c. Invert the data and find an exponential model 15 40
for a mouse’s age, A, in terms of its weight, g 20 32
grams.
(Source: Estimated from information in “Lead in the Inner City,”
d. Calculate A G for w 1, 5, and 7 and G A for American Scientist, January–February 1999, pp. 62–73.)
g 11, 23, and 27. How do the calculations sup-
port A and G being approximate inverse functions? a. Find a log model for these data.
31. Bond Rates In New Zealand, the long-term bond b. An apartment complex has a dirt play area lo-
rates are lower than short-term rates. Consider the cated 12 meters from a road. Estimate the lead
data given in the New Zealand bond rate table. concentration in the soil of the play area.
c. Find an exponential model for the data. Com-
Time to maturity New Zealand bond rate pare this model to the log model found in part a.
(years) (%) Which of the two models better displays the end
0.25 9.40 behavior suggested by the context?
2 7.90 33. Medicine The concentration of a drug in the
3 7.65 blood stream increases the longer the drug is taken
on a daily basis. The accompanying table gives esti-
4 7.50
mated concentrations (in micrograms per mil-
6 7.30 liliter) of the drug piroxicam taken in 20 mg doses
10 7.10 once a day.
a. Find a log model for the bond rate, R, in terms of Concentration Concentration
years to maturity, y. Days (g/mL) Days (g/mL)
b. The fund manager estimates 15-year rates at 1 1.5 11 6.5
7.00%. How close does your model come to this 3 3.2 13 6.9
estimate? 5 4.5 15 7.3
c. Invert the data and find an exponential model 7 5.5 17 7.5
for the time to maturity, T, in terms of the bond
9 6.2
rate, p.
1.3 Exponential and Logarithmic Functions and Models 57
a. Find a log model for the data. a. Find a log model for pH as a function of the
b. Express the end behavior of the equation using H3O concentration.
limits. b. What is the pH of orange juice with an H3O
c. Does the end behavior of the equation fit the end concentration 1.585 103?
behavior suggested by the context? c. Black coffee has a pH of 5.0. What is its concen-
d. Estimate the concentration of the drug after 2 tration of an H3O ?
days of piroxicam doses. d. A pH of 7 is neutral, a pH less than 7 indicates an
acidic solution, and a pH greater than 7 shows an
34. Peaches The following table gives the average
alkaline solution. Beer has an H3O concentra-
yearly consumption of peaches per person based on
tion of 3.162 105. Is beer acidic or alkaline?
that person’s yearly family income when the price of
peaches is $1.50 per pound. 36. Cable Subscribers On the basis of data recorded
between 1995 and 2005, the number of cable sub-
scribers in the United States as a function of the
Yearly income Consumption of peaches
(tens of thousands (pounds per person
average monthly basic cable rate can be modeled as
of dollars) per year) S(r) 1.7 20.8 ln r million subscribers
1 5.0
when the average basic rate is r dollars per month.
2 6.4 (Source: Based on data from Kagan Research LLC, Broadband
Cable Financial Databook, 2004.)
3 7.2
4 7.8
a. Use the model to estimate the number of sub-
scribers (in millions) for the following monthly
5 8.2
basic rates: $15, $30, $45.
6 8.6
b. Switch the input/output values for the three
points found in part a. Use the three inverted
a. Explain why the data are neither linear nor expo- points to find an inverse function R with input s.
nential.
Discussing Concepts
b. Find a log function to fit the data.
c. Use the log model to estimate consumption for a 37. Why does it make sense to speak of doubling time
person in a family with a yearly income of and half-life for exponential models but not for lin-
$35,000. ear models?
35. Solution pH The pH of a solution, measured on a 38. Discuss the impact of scale in the appearance of the
scale from 0 to 14, is a measure of how acidic or how graphs of exponential and log models and the con-
alkaline that solution is. The pH is a function of the ditions under which the graphs of exponential and
concentration of the hydronium ion, H3O . The log functions appear to be linear.
accompanying table shows the H3O concentration 39. Describe the end behavior of the exponential and
and associated pH for several solutions. log models. Explain how this end behavior can help
us determine which of these two functions to fit to a
H3O concentration
data set.
Solution (moles per liter) pH
Cow’s milk 3.981 107 6.4
Distilled water 1.000 107 7.0
Human blood 3.981 108 7.4
Lake Ontario water 1.259 108 7.9
Seawater 5.012 109 8.3
58 CHAPTER 1 Ingredients of Change: Functions and Models
1.4 Logistic Functions and Models
Exponential Growth with Constraints
Although exponential models are common and useful, it is sometimes unrealistic to be-
lieve that exponential growth can continue forever. In many situations, there are forces
that ultimately limit the growth. Here is a situation that may seem familiar to you.
You and a friend are shopping in a music store and find a new compact disc that
you are certain will become a hit. You each buy the CD and rush back to campus to
begin spreading the news. As word spreads, the total number of CDs sold begins to
grow exponentially, as shown in Figure 1.42.
Total sales
CD sales growing exponentially
FIGURE 1.42 Time
However, this trend cannot continue forever. Eventually, the word will spread to
people who have already bought the CD or to people who have no interest in it, and
the rate of increase in total sales will begin to decline (Figure 1.43). In fact, because
there is only a limited number of people who will ever be interested in buying the CD,
total sales ultimately must level off. The graph representing the total sales of the CD
as a function of time is a combination of rapid exponential growth followed by a
slower increase and ultimate leveling off. See Figure 1.44.
Total sales Total sales
Time Time
CD sales leveling off Total sales of a CD
FIGURE 1.43 FIGURE 1.44
S-shaped behavior such as this is common in marketing situations, the spread of
disease, the spread of information, the adoption of new technology, and the growth
of certain populations. A mathematical function that exhibits such an S-shaped curve
is called a logistic function. Its equation is of the form
L
f(x)
1 AeBx
From now on, when referring to a logistic equation or function, we will consider an
equation of this form. The number L appearing in the numerator of a logistic equa-
tion determines a horizontal asymptote y L for a graph of the function f.
1.4 Logistic Functions and Models 59
Logistic Model
Algebraically: A logistic model has an equation of the form
L
f(x)
1 AeBx
We refer to L as the limiting value of the function.
Graphically: The logistic function f increases if B is positive and decreases if B
is negative. The graph of a logistic function is bounded by the horizontal axis
and the line y L. (See Figure 1.45.)
f(x) f(x)
L L
x x
Be aware that some
technologies do not L L
(a) f(x) with B 0 (b) f(x) with B 0
1.4.1
have a built-in 1 AeBx 1 AeBx
logistic regression
routine. See the Excel FIGURE 1.45
Instruction Guide to
obtain the logistic
curve-fitting
procedure we use in The graph of a positive, increasing logistic function (such as that shown in Figure
this text. 1.45a) is trapped between the horizontal axis ( y 0) and the limiting value y L.
Finding Logistic Models
Consider a worm that has attacked the computers of an international corporation.
The worm is first detected on 100 computers. The corporation has 10,000 comput-
ers, so as time increases, the number of infected computers can approach, but never
exceed, 10,000. Consider the following representations (Table 1.21 and Figure 1.46)
of the function describing the spread of the worm.
TABLE 1.21
Total number of
Total number of infected computers
Time infected computers (thousands)
(in hours) (in thousands) 10
0 0.1 8
0.5 0.597
6
1 2.851
4
1.5 7.148
2
2 9.403
Time
2.5 9.900 0
FIGURE 1.46 0 0.5 1 1.5 2 2.5
(hours)
60 CHAPTER 1 Ingredients of Change: Functions and Models
A logistic model to fit the data in Table 1.21 is
10
I(t) thousand computers infected
1 99e3.676t
t hours after the initial attack, t 0. Note that the limiting value here is L 10
thousand computers.
A graph of the function I is shown in Figure 1.47. The dotted, horizontal line por-
trays the limiting value, L 10. Note that the curvature on the left side of this graph is
concave up , whereas the curvature on the right side is concave down .
The point on the graph at which the concavity changes is called the inflection point.
The inflection point on the logistic graph modeling the spread of the computer worm
(Figure 1.47) is marked with a black dot. In some situations, inflection points have very
important interpretations. We later use calculus to find and help interpret these spe-
cial points.
I(t)
Total number of
infected computers
(thousands)
10
2
In social science and life science
applications, the limiting value t
0
is often called the carrying 0
Hours
0.5 1 1.5 2 2.5
capacity or the saturation level.
In other applications it is FIGURE 1.47
sometimes referred to as the
leveling-off value. When you use a calculator or a computer program to find an equation and con-
struct the graph of a logistic model for a set of data, be sure to align the input data. The
same numerical computation problems that we discussed for exponential models can
occur with logistic models because of the exponential term in the denominator.
EXAMPLE 1 Finding a Logistic Model
Bacteria Table 1.22 shows the number of bacteria counted in a biology experiment.
TABLE 1.22
Day Bacteria Day Bacteria
1 4 6 619
2 15 7 733
3 52 8 771
4 165 9 782
5 391
1.4 Logistic Functions and Models 61
Find a logistic model that fits the data. What is the end behavior of the model as
time increases?
Solution
A scatter plot of the data (Figure 1.48) suggests that a logistic model is appropriate.
Bacteria
800
600
400
200
0 Day
0 1 2 3 4 5 6 7 8 9
FIGURE 1.48
A possible logistic model is
786
B(x) bacteria
1 732.9e1.318x
on day x, 1 x 9.
The limiting value of this model is L 786, or, in context, approximately 786
bacteria. The graph of B on a scatter plot appears to fit fairly well (see Figure 1.49).
The dotted line in Figure 1.49 denotes the upper limiting value. As time increases, the
number of bacteria approaches 786.
B (x)
Bacteria
800
600
400
200
x
0
0 1 2 3 4 5 6 7 8 9 (days)
FIGURE 1.49
The preceding logistic curves begin near zero and then increase toward a limiting
value L. As we have previously noted, there are also situations where the curve begins
near its limiting value and then decreases toward zero. ●
62 CHAPTER 1 Ingredients of Change: Functions and Models
EXAMPLE 2 Aligning Input and Determining End Behavior
Height Of a group of 200 college men surveyed, the number who were taller than a
given number of inches is recorded in Table 1.23.
TABLE 1.23
Number of Number of Number of
Inches men Inches men Inches men
64 200 68 139 72 23
65 194 69 105 73 12
66 184 70 70 74 6
67 166 71 42 75 3
Find an appropriate model for the data. What is the end behavior of the model as
height increases?
Solution
A scatter plot of the data is shown in Figure 1.50.
Men
200
180
160
140
120
100
80
60
40
20
Height
0
64 65 66 67 68 69 70 71 72 73 74 75 (inches)
FIGURE 1.50
We choose to align the input data by subtracting 65 from each value. A logistic
model for the aligned data is
205.51171
M(x) men out of 200 men surveyed
1 0.05822e0.70106x
who are taller than x 65 inches. The two horizontal asymptotes of a graph of this
function are y 0 and y 205.5. Because the function decreases, the end behavior
as height increases is given by the lower asymptote; that is, as the heights increase, the
number of men taller than a particular height approaches zero. ●
1.4 Logistic Functions and Models 63
L
Recall that a logistic function is of the form f(x) 1 AeBx . What happened to
the negative sign before the B in Example 2? In that case, B 0.70106. When the
formula was written, the negatives canceled.
205.51171 205.51171
Number of men
1 0.05822e1(0.70106x) 1 0.05822e0.70106x
This will always be the case for a logistic function that decreases.
It is likely that a logistic model will have a limiting value that is lower or higher
than the one indicated by the context. For instance, in Example 2 only 200 men were
surveyed, but the limiting value is L 206 men. This does not mean that the model
is invalid, but it does indicate that care should be taken when extrapolating from lo-
gistic models.
In Example 2 we chose to shift the logistic model horizontally by realigning input
data in order to avoid an extremely small value for A. It is possible to shift exponen-
tial and logistic models vertically by realigning output data. Such an alignment would
change the lower limiting values from 0 to some other number. We do this only when
there is reason to believe that the data approach a different limiting position than the
input axis or, for the case of the logistic model, when it is obvious that there is an up-
per limiting value and the shift will significantly improve end-behavior fit. In this
case, we will recommend a specific alignment.
Limits and the Infinitely Large
As we study functions in calculus, we pay attention to how the output of a function
behaves as the input becomes larger and larger in the positive direction (increases
without bound) or the magnitude of the input becomes larger and larger in the neg-
ative direction (decreases without bound). This behavior is called the end behavior
of the function. We have already discussed end behavior for a few functions and will
continue to use end-behavior analysis to choose models as well as to develop the cal-
culus topics of limits of sums and integration in later chapters.
For an example of end behavior, consider the graph of the function shown in Fig-
ure 1.51. As x becomes arbitrarily large in the positive direction (moving from x 0
to the right without bound along the graph of the function), the function outputs
u(x) increase while becoming closer and closer to 0.3. As the magnitude of x becomes
u (x) larger and larger in the negative direction (moving from x 0 to the left without
0.3 bound on the function graph), again the function outputs become closer and closer
to 0.3. We express these two facts mathematically by writing
0.2 lim u(x) 0.3
x→
and lim u(x) 0.3
x→
or with the combined statement
0.1
lim u(x) 0.3
x→
x When a function approaches a number k as the input increases or decreases with-
out bound, we call the horizontal line that has equation y k a horizontal asymp-
tote. The function depicted in Figure 1.51 has a horizontal asymptote, y 0.3. We
FIGURE 1.51 say that the output of the function has a limiting value of y 0.3.
64 CHAPTER 1 Ingredients of Change: Functions and Models
L
For an increasing logistic function f, where f(x) 1 AeBx with B 0,
lim f(x) 0
x→
lim f(x) L
and x→
L
Similarly, for a decreasing logistic function f, where f(x) 1 AeBx with
B 0 (as shown in Figure 1.45b), the end behavior can be described by the limit
statements
lim f(x) L
x→
lim f(x) 0
and x→
Thus any logistic function has two horizontal asymptotes. We refer to y 0 as the
lower asymptote and to y L as the upper asymptote or the limiting value of the
function.
Some functions appear to have no limiting value (horizontal asymptote) in at
least one direction. We say that the output of a function is increasing without bound
if the output continues to increase infinitely. If the output of a function continues to
decrease infinitely, we say that the output is decreasing without bound. This end be-
havior is seen in the linear, exponential, and log functions, as well as in polynomial
functions that we will discuss in Section 1.5. (See Figures 1.52 and 1.53.)
f(x) g(x)
x x
f ( x) decreases without bound g( x) increases without bound
as x becomes arbitrarily large as x approaches positive infin-
in the positive direction. f ( x) ity. g( x) decreases without
increases without bound as x bound as x approaches nega-
becomes arbitrarily large in tive infinity.
the negative direction. FIGURE 1.53
FIGURE 1.52
Just as we can often estimate end behavior from a graph, it is also possible to es-
timate the end behavior of a function numerically by evaluating the function at in-
creasingly large values of the input variable. This process is illustrated in Example 3.
EXAMPLE 3 Numerically Estimating End Behavior
2
3x x
An equation for the function depicted in Figure 1.54 is u(x) 2 .
1.4.2 10x 3x 2
lim u(x) andx→
Numerically estimate x→ lim u(x).
1.4 Logistic Functions and Models 65
u (x) Solution
0.3 To estimate the positive end behavior of u numerically, choose increasingly large pos-
itive values of x, as shown in Table 1.24.
0.2 TABLE 1.24
x→ 10 100 1000 10,000
0.1
u (x) 0.300388 0.300094 0.3000099 0.3000009
x
It appears that as x approaches positive infinity, the function outputs approach
0.3. Next, consider the negative end behavior of u by choosing negative values of x
FIGURE 1.54 with increasingly large magnitudes, as shown in Table 1.25.
TABLE 1.25
x → 10 100 1000 10,000
u (x) 0.298354 0.299894 0.299990 0.299999
j (x) Here also, it appears that the limiting value of the outputs is 0.3. ●
As in the example, numerical evaluation of end behavior is especially useful for
functions constructed by function division or composition. We will use a similar
technique when we look at limits at a point during our discussion of the calculus topic
of derivatives in Chapter 2.
Keep in mind that a function may oscillate and not approach any specific output
x
value as x approaches infinity. Such is the case with the function j in Figure 1.55. In
x→
lim j(x) does not exist. this case, we say that the limit of j as x approaches infinity does not exist. The sine and
FIGURE 1.55
cosine functions are good examples of such functions.
these. If you identify the scatter plot as none of these,
1.4 Concept Inventory give reasons.
1. 2.
L
• Logistic function: f(x)
1 AeBx
• Equations of horizontal asymptotes
• Concave up and concave down
• Inflection point
• Limiting value and end behavior
3. 4.
1.4 Activities
Getting Started
Identify the scatter plots in Activities 1 through 6 as
linear, exponential, logarithmic, logistic, or none of
66 CHAPTER 1 Ingredients of Change: Functions and Models
5. 6. is concave down. Label the approximate location
of the inflection point.
12. P.T.A. The accompanying table gives the total
number of states associated with the national P.T.A.
organization from 1895 through 1931.
Total number Total number
Year of states Year of states
For Activities 7 through 10, indicate whether the func-
tion is an increasing or a decreasing logistic function. 1895 1 1915 30
Also identify the upper limiting value of the function. 1899 3 1919 38
100 1903 7 1923 43
7. f(x)
1 9e0.78x 1907 15 1927 47
1925 1911 23 1931 48
8. A(t)
1 32e1.86t (Source: R. Hamblin, R. Jacobsen, and J. Miller, A Mathematical
Theory of Social Change, New York: John Wiley & Sons, 1973.)
39.2
9. h(g)
1 0.8e3g
a. Find a logistic model for the data.
16.5
10. k(x) b. What is the maximum number of states that
1 1.86e0.43x
could have joined the P.T.A. by 1931? How does
this number compare to the limiting value given
Applying Concepts
by the model in part a?
11. Postage The following table gives the number of 13. Flu In the fall of 1918, an influenza epidemic hit
European, North American, and South American the U.S. Navy. It spread to the Army, to American
countries that issued postage stamps from 1840 civilians, and ultimately to the world. It is estimated
through 1880. that 20 million people had died from the epidemic
by 1920. Of these, 550,000 were Americans—over
Year Total number of countries
10 times the number of World War I battle deaths.
1840 1 The accompanying table gives the total numbers of
1845 3 Navy, Army, and civilian deaths that resulted from
1850 9
the epidemic in 1918.
1855 16
1860 24 Total civilian
Total deaths deaths in 45
1865 30
Week ending Navy Army major cities
1870 34
August 31 2
1875 36
September 7 13 40
1880 37
September 14 56 76 68
(Source: “The Curve of Cultural Diffusion,” American September 21 292 174 517
Sociological Review, August 1936, pp. 547–556.)
September 28 1172 1146 1970
a. Find a logistic model for the data, and discuss October 5 1823 3590 6528
how well the equation fits.
October 12 2338 9760 17,914
b. Sketch the graph of the equation in part a, and
October 19 2670 15,319 37,853
mark where the curve is concave up and where it
1.4 Logistic Functions and Models 67
Total civilian
a. Find a logistic model for the data.
Total deaths deaths in 45 b. The park owners have been considering closing
Week ending Navy Army major cities the park from October 15 through March 15
October 26 2820 17,943 58,659 each year. How many visitors will they poten-
November 2 2919 19,126 73,477
tially miss by this closure?
November 9 2990 20,034 81,919 15. Chemical Reaction A chemical reaction begins
November 16 3047 20,553 86,957 when a certain mixture of chemicals reaches 95°C.
The reaction activity is measured in units (U) per
November 23 3104 20,865 90,449
100 microliters (100 L) of the mixture. Measure-
November 30 3137 21,184 93,641 ments during the first 18 minutes after the mix-
(Source: A. W. Crosby, Jr., Epidemic and Peace 1918, Westport, ture reaches 95°C are listed in the accompanying
Conn.: Greenwood Press, 1976.) table.
a. Find logistic equations to fit each set of data. In
Time Activity Time Activity
each case, graph the equation on a scatter plot of (minutes) (U/100 L) (minutes) (U/100 L)
the data.
0 0.10 10 1.40
b. Write models for each of the data sets.
2 0.10 12 1.55
c. Compare the limiting values of the models in
4 0.25 14 1.75
part b with the highest data values in the table.
Do you believe that the limiting values indicated 6 0.60 16 1.90
by the models accurately reflect the ultimate 8 1.00 18 1.95
number of deaths? Explain.
(Source: David E. Birch et al., “Simplified Hot Start PCR,” Nature,
14. Visitors The total numbers of visitors to an vol. 381 (May 30, 1996), p. 445.)
amusement park that stays open all year are given in
the accompanying table.
a. Examine a scatter plot of the data. Estimate the
limiting value. Estimate at what time the inflec-
tion point occurs.
Cumulative number of visitors
b. Find a logistic model for the data. What is the
by the end of the month
Month (thousands)
limiting value for this logistic function?
January 25 c. Use the model to estimate by how much the re-
action activity increased between 7 minutes and
February 54
11 minutes.
March 118
16. Stolen Bases San Francisco Giants legend Willie
April 250
Mays’s cumulative numbers of stolen bases between
May 500 1951 and 1963 are as shown in the table on page 68.
June 898 a. Find a logistic model for the data. Comment on
July 1440 how well the logistic equation fits the data.
August 1921 b. What is the interpretation of first differences in
September 2169 this context?
October 2339 c. Use the model in part a to estimate the number
November 2395 of bases that Mays stole in 1964. Compare this
estimate with 19, his actual number of stolen
December 2423
bases in 1964.
68 CHAPTER 1 Ingredients of Change: Functions and Models
Year Cumulative stolen bases
the model appropriate to use in predicting long-
term world population behavior?
1951 7
c. Use the model to estimate the world population
1952 11
in 1850. In 1990. Are the estimates reliable? Ex-
1953 11 plain.
1954 19
For each function given in Activities 18 through 27, a.
1955 43 consider a graph of the function and describe its con-
1956 83 cavity, b. use limit notation to describe the end behavior
1957 121
of the function (for both the positive and negative direc-
tions), and c. verbally describe the end behavior.
1958 152
1959 179
18. f(x) 5x 2
1960 204 19. g(t) 5(7 t)
1961 222 20. h(s) 5(0.7s)
1962 240
21. y(x) 3 6 ln x
1963 248
22. j(u) 3 6 ln u
17. Population A 1998 United Nations population 52
23. l(t)
study reported the world population between 1804 1 0.5e0.9t
and 1987 and projected the population through
14.3
2071. These populations are as shown in the accom- 24. m(t)
1 5e0.9t
panying table.
25. n(k) 4k2 2k 12
Population Population
Year (billions) Year (billions)
26. A(p) 3p2 4p 8
1804 1 1999 6 27. C(q) 2q3 5q2 3q 7
1927 2 2011 7
Discussing Concepts
1960 3 2025 8
1974 4 2041 9 28. Describe the graph of a logistic function using the
words concavity, inflection, and increasing/decreas-
1987 5 2071 10
ing.
a. Find a logistic model for world population. Dis- 29. Using the idea of limits, describe the end behavior
cuss how well the equation fits the data. of the logistic model, and explain how this end be-
b. According to the model, what will ultimately havior differs from that of the exponential and log
happen to world population? Do you consider models.
1.5 Polynomial Functions and Models
Polynomial functions and models have been used extensively throughout the history
of mathematics. Their successful use stems from both their presence in certain natu-
ral phenomena and their relatively simple application.
Even though higher-degree polynomials are useful in some situations, we limit
our discussion of polynomial functions and models to the linear, quadratic, and cubic
cases.
1.5 Polynomial Functions and Models 69
Quadratic Modeling
A roofing company in Miami keeps track of the number of roofing jobs it completes
each month. The data from January through June are given in Table 1.26.
TABLE 1.26
Month January February March April May June
Number of jobs 90 91 101 120 148 185
The first differences and percentage differences are not close to being constant, so
we conclude that some model other than a linear model or an exponential model is
appropriate in this case. Note, however, that the differences between the first differ-
ences are nearly constant. We call these second differences.
Number of jobs 90 91 101 120 148 185
First differences 1 10 19 28 37
Second differences 9 9 9 9
When first differences are constant, the data can be modeled by a linear equation.
When second differences are constant, the data can be modeled by the quadratic
function f(x) ax2 bx c as long as a
0. A quadratic equation of this form
defines a function f with input x. From now on, when we refer to a quadratic equa-
tion or function, we will consider an equation of the form f(x) ax2 bx c.
The graph of a quadratic function is a parabola.
Quadratic Model
Algebraically: A quadratic model has an equation of the form
f(x) ax2 bx c
where a
0.
Graphically: The graph of a quadratic function is a concave-up parabola if
a 0 and is a concave-down parabola if a 0. (See Figures 1.56a and
1.56b.)
f(x) f(x)
x x
(a) f (x) ax2 bx c with a 0 (b) f (x) ax2 bx c with a 0
lim
x → f (x) →
lim
x → f (x) →
FIGURE 1.56
70 CHAPTER 1 Ingredients of Change: Functions and Models
Using technology to obtain a quadratic equation that fits the roofing job data, you
should find the quadratic model to be
J(x) 4.5x2 12.5x 98 jobs
is the number of roofing jobs completed between January and June where x 1 in
January, x 2 in February, and so on. Note in Figure 1.57 that the function J pro-
vides an excellent fit to the data.
J(x)
Jobs
190
170
150
130
110
90
x
0 1 2 3 4 5 6
(months)
(Jan) (Feb) (Mar) (Apr) (May) (June)
Quadratic model
FIGURE 1.57
If we are willing to assume that this quadratic function continues to model the
number of roofing jobs for the next 3 months, how many jobs would we predict the
company will have in August? Substituting x 8 into the quadratic equation yields
J(8) 4.5(8)2 12.5(8) 98 286 jobs
EXAMPLE 1 Finding a Quadratic Model
Birthweight The percentage of low-birthweight babies born before 37 weeks gesta-
tion as a function of the amount of weight gained by the mother is given in Table 1.27.
TABLE 1.27
Weight gain of
18 23 28 33 38 43
mother (pounds)
Percentage of
babies born before
37 weeks weighing 48.2 42.5 38.6 36.5 35.4 35.7
less than 5 pounds
8 ounces
(Source: National Vital Statistics Report, vol. 50, no. 5, February 12, 2002.)
a. Find a quadratic model for the data.
b. Compare the minimum of the parabola with the minimum of the data.
1.5 Polynomial Functions and Models 71
Solution
a. A scatter plot of the data suggests a concave-up shape with a minimum around
38 pounds. See Figure 1.58.
Low-weight
babies
(percent)
50
45
40
Weight gain
35
0 18 23 28 33 38 43
(pounds)
FIGURE 1.58
A quadratic model for the data is
P(g) 0.0294g 2 2.286g 79.685 percent
of babies born before 37 weeks that weigh less than 5.5 pounds
when the mother’s weight gain is g pounds, 18 g 43.
b. Looking at the quadratic function graphed on the scatter plot (see Figure 1.59),
we observe that the minimum of the parabola is slightly to the right and below
the minimum data point. This means that the model slightly underestimates
the minimum percentage and estimates that the minimum occurs slightly after
it occurs in the data table.
P(g)
(percent)
50
45
40
g
35
0 18 23 28 33 38 43
(pounds)
FIGURE 1.59 ●
In the foregoing discussion and example, you have seen data sets that appear to
be quadratic. That is, they may be reasonably modeled by a quadratic equation. In
each case, the equation had a positive coefficient before the squared term. This
value is called the leading coefficient because it is usually the first one that you
write down in the equation f(x) ax2 bx c . Also, in each case, the graph of
72 CHAPTER 1 Ingredients of Change: Functions and Models
the equation appeared to be part of a parabola opening upward. Remember that we
call such curvature concave up.
Table 1.28 gives the population (in thousands) of Cleveland, Ohio, from 1900
through 1980.
TABLE 1.28
Year 1900 1910 1920 1930 1940
Population 382 561 797 900 878
Year 1950 1960 1970 1980
Population 915 876 751 574
(Source: Statistical Abstract, 1998.)
A scatter plot of the data suggests a parabola opening downward (or concave
down). The scatter plot and the graph of a quadratic equation fitted to the data are
shown in Figure 1.60.
p (t)
Population
(thousands)
900
800
700
600
500
400
300
200
100
t
Years
since
0 10 20 30 40 50 60 70 80
1900
(1900) (1940) (1980)
FIGURE 1.60
Would you expect the leading coefficient of the equation to be positive or nega-
tive? The population of Cleveland is given by
p(t) 0.280t2 24.9t 375.3 thousand people
where t is the number of years since 1900, 0 t 80. When the graph of a quadratic
function is concave up, its leading coefficient is positive; when the graph of a quad-
ratic function is concave down, its leading coefficient is negative. Curvature will be
important in later discussions that involve concepts of calculus.
1.5 Polynomial Functions and Models 73
Quadratic or Exponential?
Data sets that exhibit an obvious maximum or minimum are more easily identified as
quadratic than data sets without a maximum or minimum. Sometimes, as shown in Ex-
ample 2, all that is indicated by a scatter plot is the left side or right side of a parabola.
EXAMPLE 2 Distinguishing Between Quadratic and Exponential Models
Population Table 1.29 shows the population of the contiguous states of the United
States for selected years between 1790 and 1930.
TABLE 1.29
Population Population
Year (millions) Year (millions)
1790 3.929 1870 39.818
1810 7.240 1890 62.948
1830 12.866 1910 91.972
1850 23.192 1930 122.775
(Source: Statistical Abstract, 1998.)
Find an appropriate model for the data.
Solution An examination of the scatter plot shows an increasing, concave-up
shape (see Figure 1.61).
Population
(millions)
150
100
50
Year
1790 1810 1830 1850 1870 1890 1910 1930
FIGURE 1.61
At first, it seems logical to try an exponential model for the data. An exponential
model for the population of the contiguous United States is
E(t) 4.558(1.285t) million people
where t is the number of decades since 1790, 0 t 14. However, this equation
does not seem to fit the data very well, as Figure 1.62 shows.
74 CHAPTER 1 Ingredients of Change: Functions and Models
E(t)
Population
(millions)
150
100
50
t
Years
0 70 140 since
(1790) (1860) (1930) 1790
Exponential model
FIGURE 1.62
Q (t)
Population
(millions)
150
100
50
t
Years
0 70 140 since
(1790) (1860) (1930) 1790
Quadratic model
FIGURE 1.63
Another possibility is a quadratic model. The right half of a parabola that opens
upward could fit the scatter plot. A quadratic model for the population of the
contiguous United States is
Q(t) 0.66t2 0.77t 4.81 million people
where t is the number of decades since 1790, 0 t 14. When graphed on the scat-
ter plot, this equation appears to be a very good fit (see Figure 1.63). It is the more
appropriate model for the population of the contiguous United States on the basis of
the given data. ●
Cubic Modeling
We saw that when the first differences of a set of evenly spaced data are constant, the
data can be modeled perfectly by the linear equation y ax b. Likewise, when the
second differences of evenly spaced input data are constant, the data can be modeled
perfectly by the quadratic equation y ax2 bx c. It is also possible for the third
1.5 Polynomial Functions and Models 75
differences to be constant. In this case, the data can be modeled perfectly by a cubic
equation. A cubic equation of the form f(x) ax3 bx2 cx d is a function
with input x. From now on, when we refer to a cubic equation or function, we will
consider the equation to be of this form.
Because in the real world we are extremely unlikely to encounter data that are per-
fectly cubic, we will not look at third differences. Instead, we will examine a scatter
plot of the data to see whether a cubic model may be appropriate. Figure 1.64 shows
the graphs of some cubic equations.
y y
(a) (b)
y y
(c) (d)
FIGURE 1.64
Every cubic function f(x) ax3 bx2 cx d with a
0 has a graph that re-
sembles one of the four graphs in Figure 1.64. Figures 1.64a and 1.64c correspond to
equations in which a 0, and Figures 1.64b and 1.64d are graphs of equations in
which a 0. For a cubic equation f(x) ax3 bx2 cx d with a
0, the end
behavior shows that f(x) increases without bound in one direction and decreases
without bound in the opposite direction.
Figure 1.65 (on page 76) shows scatter plots of data sets that could be modeled
by cubic equations.
76 CHAPTER 1 Ingredients of Change: Functions and Models
y y
x
(a) (b)
y y
x
x
(c) (d)
FIGURE 1.65
You may already have noticed that in every cubic function, the curvature of the
graph changes once from concave down to concave up, or vice versa. As we noted
with the graph of a logistic curve, the point on the graph at which concavity changes
is called the inflection point. All cubic functions have one inflection point. The ap-
proximate location of the inflection point in each of the graphs in Figure 1.66 is
marked with a dot. In Chapter 4, we will see how calculus can be used to determine
the exact location of the inflection point of a cubic function.
y y
Inflection
Co
Inflection point
nc
av
point eu
p
do C
w
ave
up
on
Co
x
n
ca
Conc
ca
ve
ve
n
dow
x
n
(a) (b)
FIGURE 1.66
1.5 Polynomial Functions and Models 77
It is often the case that a portion of a cubic function appears to fit extremely well
a set of data that can be adequately modeled with a quadratic function. In an effort to
keep things as simple as possible, we adopt the following convention:
If the scatter plot of a set of data fails to exhibit an inflection point, then it is
not appropriate to fit a cubic equation to the data.
We must be extremely cautious when using cubic models to extrapolate. For the
data sets whose scatter plots are shown in Figures 1.65a and 1.65c, the functions in-
dicated by the dotted curves appear to follow the trend of the data. However, in Fig-
ure 1.65b, it would be possible for additional data to level off (as in a logistic model),
whereas the cubic function takes a downward turn. Also, additional data in Figure
1.65d might continue to get closer to the x-axis, whereas the cubic function that is fit-
ted to the available data begins to rise.
Cubic Model
Algebraically: A cubic model has an equation of the form
f(x) ax3 bx2 cx d
where a
0.
Graphically: The graph of a cubic function has one inflection point and no
limiting values. (See Figure 1.64.)
EXAMPLE 3 Finding a Cubic Model
Gas Price The average price in dollars per 1000 cubic feet of natural gas for resi-
dential use in the United States for selected years from 1980 through 2005 is given in
Table 1.30.
TABLE 1.30
Year Price (dollars) Year Price (dollars)
1980 3.68 1998 6.82
1982 5.17 2000 7.76
1985 6.12 2003 9.52
1990 5.80 2004 10.74
1995 6.06 2005 13.84
(Source: Statistical Abstract, 1992 and 2001, and Energy
Information Administration.)
a. Find an appropriate model for the data. Would it be wise to use this model to
predict future natural gas prices?
78 CHAPTER 1 Ingredients of Change: Functions and Models
b. Use the model in part a to estimate the price in 1993.
c. According to the model, when did the average price of 1000 cubic feet of natu-
ral gas first exceed $6.00?
Solution
a. An examination of a scatter plot shows that a cubic model is appropriate.
Note that the scatter plot shown in Figure 1.67 appears to be mostly concave
down between 1980 and 1990 but then is concave up between 1990 and 2005.
That is, there appears to be an inflection point (a change of concavity)
near 1990.
Price
(dollars)
13.50
11.50
9.50
7.50
5.50
3.50 Year
1980 1985 1990 1995 2000 2005
FIGURE 1.67
A cubic model for the price of natural gas is
P(x) 0.00275x3 0.0876x2 0.842x 3.72 dollars
where x is the number of years since the end of 1980, 0 x 25. A graph of
the equation over the scatter plot is shown in Figure 1.68.
Note that the graph is increasing to the right of about 1993. Natural gas prices
will probably not continue to rise indefinitely as the cubic function does, so it
is unwise to use the model to predict future prices of natural gas. Additional
data should be obtained to see the pattern past 2005.
b. In 1993 the average price of 1000 cubic feet of natural gas was P(13) $5.89.
1.5 Polynomial Functions and Models 79
P (x) P(x)
Price Price
(dollars) (dollars)
13.50 13.50
11.50 11.50
9.50 9.50
7.50 7.50
6.00
5.50 5.50
x x
Years Years
3.50 3.50
0 5 10 15 20 25 since 0 5 10 15 20 25 since
1980 1980
FIGURE 1.68 FIGURE 1.69
c. To determine when the average price first exceeded $6.00, solve the equation
P(x) 6. In Figure 1.69, a dotted line is drawn at P(x) $6.00. This line in-
tersects the graph at three places, so there are three solutions to the equation
P(x) 6. The solutions are x 4.57, 11.40, and 15.90.
We seek the smallest solution (x 4.57), which corresponds to the first time
the output is 6.00. However, the data represent yearly averages. In 1984 the av-
erage price was less than $6.00, and in 1985 the average price was slightly more
than $6.00. Therefore, the average price first exceeded $6.00 in 1985. ●
We can also model the average price of natural gas in Example 3 with a cubic
equation by renumbering the years so that x is the number of years since 1900. This
new model for the average price of natural gas will be
Price 0.00275x3 0.748x2 67.69x 2033.29 dollars
where x is the number of years since 1900, 80 x 105.
Although the two models have different inputs, they yield the same results. Note
that the coefficient on x3 is the same in both models. This will be the case for any
alignment.
In real life, data do not come with instructions attached to tell what function to
use as a possible model. It is important to consider first what underlying processes
may be influencing the relationship between input and output and what end behav-
ior may be exhibited. However, the following simple guidelines may help.
80 CHAPTER 1 Ingredients of Change: Functions and Models
Steps in Choosing a Model
1. Look at the curvature of a scatter plot of the data.
• If the points appear to lie in a straight line, try a linear model.
• If the scatter plot is curved but has no inflection point, try a quad-
ratic, an exponential, or a log model.
• If the scatter plot appears to have an inflection point, try a cubic
and/or a logistic model.
Note: If the input values are equally spaced, it might be helpful to look at
first differences versus percentage differences to decide whether a linear
or an exponential model would be more appropriate, or to compare
second differences with percentage differences to decide whether a
quadratic or an exponential model would be better.
2. Look at the fit of the possible equations. In Step 1, you should have nar-
rowed the possible models to at most two choices. Compute these equa-
tions, and graph them on a scatter plot of the data. The one that comes
closest to the most points (but does not necessarily go through the most
points) is normally the better model to choose.
3. Look at the end behavior of the scatter plot. If Step 2 does not reveal that
one model is obviously better than another, consider the end behavior of
the data, and choose the appropriate model.
4. Consider that there may be two equally good models for a particular set
of data. If that is the case, then you may choose either.
1.5 Concept Inventory 1. y 2. y
7 50
• Second differences 40
5
• Quadratic function: f(x) ax2 bx c (constant 30
second differences, no change in concavity) 3 20
• Parabola
1 10
• Cubic function: f(x) ax3 bx2 cx d (one 0 x x
0
change in concavity) 0 1 2 3 4 0 200 600 1000
• Inflection point
3. y 4. y
45 6.0
1.5 Activities 40 5.8
35 5.6
Getting Started
30 5.4
Identify the curves in Activities 1 through 6 as concave
up or concave down. In each case, indicate the portion 25 5.2
of the horizontal axis over which the part of the curve 20 x 5.0 x
that is shown is increasing or decreasing. 9 15 21 27 84 88 92 96
1.5 Polynomial Functions and Models 81
5. y 6. y giving the height of the missile, are collected via
25 15 telemetry.
20
10 Seconds after launch Feet above water
15
10 0 128
5
5 0.5 140
0 x 0 x 1 144
3 4 5 6 7 8 0 1 2 3
1.5 140
For the graphs in Activities 7 through 12, describe the 2 128
curvature by indicating the portions of the displayed 2.5 108
horizontal axis over which each curve is concave down 3 80
or concave up. Mark the approximate location of the in-
flection point on each curve. 3.5
4
7. y 8. y
25 a. Without graphing, show that the data in the table
80
20 are quadratic.
60
15 b. Without finding an equation, complete the ac-
40 10 companying table by filling in the missing values.
20 5 c. Find a quadratic model for the complete data.
0 x 0 x d. Use the model to determine when the missile hits
0 5 10 15 20 86 92 98 104 110 the water.
9. y 10. y 14. Profit The accompanying table gives the price, in
12 12 dollars, of a round-trip ticket from Denver to
10 Chicago on a certain airline and the corresponding
8 8 monthly profit, in millions of dollars, for that airline.
6
4
4
2 Ticket price (dollars) Profit (millions of dollars)
0 x
0 x 0 100 300 500 200 3.08
0 2 4 6 8
250 3.52
11. y 12. y 300 3.76
12 500 350 3.82
10 400 3.70
8 300 450 3.38
6
4
100 a. Is a quadratic model appropriate for the data?
2
0 x 0 x Explain.
0 2 4 6 8 10 0 20 60 100 140
b. Find a quadratic model for the data.
c. As the ticket price increases, the airline should
Applying Concepts collect more money. How can it be that when the
13. Missile Height During a training mission in the ticket price reaches a certain amount, profit de-
South Pacific, a Tomahawk cruise missile misfires. creases?
It goes over the side of the ship and hits the water. d. At what ticket price will the airline begin to post
Suppose the data shown in the accompanying table, a negative profit (that is, a net loss)?
82 CHAPTER 1 Ingredients of Change: Functions and Models
15. Marriage Age As listed in The 1999 World Death rate
Almanac, the median age (in years) at first marriage Age x (deaths per thousand people)
of females in the United States is shown in the
40 2.0
accompanying table.
45 3.0
50 4.4
Year 1960 1970 1980 1990
55 6.7
Age 20.3 20.8 22.0 23.9
60 10.7
65 16.5
a. Refute or defend the following statement: “The
data are perfectly quadratic.” (Source: Statistical Abstract, 2001.)
b. Without finding an equation, estimate the age
that corresponds to the year 2000. a. Find a quadratic model for the data in the table.
Discuss how well the equation fits the data.
c. Find a quadratic model for the data.
b. Use the model to complete the following table:
d. Use the model to estimate the median age in the
year 2000. Is it the same as your answer to part b? Age Model prediction Actual rate
16. Veggies The per capita consumption of commer- 51 4.7
cially produced fresh vegetables in the United States 52 5.1
from 1980 through 2000 was as shown in the ac-
53 5.6
companying table.
57 8.1
Vegetable consumption 59 9.7
Year (pounds per person) 63 14.1
1980 149.1 70 25.5
1985 156.0 75 38.0
1990 167.1 80 59.2
1995 179.1
2000 201.7 c. What can you conclude about using a model to
make predictions?
(Sources: Statistical Abstract, 2001, and
www.ers.usda.gov, accessed 9/25/02.) 18. Lead Paint Lead was banned as an ingredient in
most paints in 1978, although it is still used in some
a. Find a quadratic model for the data, and examine
specialty paints. Lead usage in paints from 1940
the equation graphed on a scatter plot of the data.
through 1980 is reported in the accompanying table.
b. Do you believe that the equation in part a is a
good fit? Explain.
Lead usage
c. The per capita consumption in 2001 had not yet Year (thousands of tons)
been tabulated when the data in the table were
1940 70
published. What does the quadratic model give
as the per capita consumption in 2001? Do you 1950 35
believe that this estimate is reliable? 1960 10
d. According to your model, in what year will con- 1970 5
sumption exceed 225 pounds per person? 1980 0.01
17. Mortality The accompanying table lists the death (Source: Estimated from information in “Lead in the
rates (number of deaths per thousand people whose Inner Cities,” American Scientist, January–February
age is x) in 1998 for the United States. 1999, pp. 62–73.)
1.5 Polynomial Functions and Models 83
a. Examine a scatter plot of the data. Find quadratic a. Examine a scatter plot of the data in the table.
and exponential models for lead usage. Com- Find a cubic model for the data, and graph the
ment on how well each equation fits the data. equation on the scatter plot. Discuss how well
b. What is the end behavior suggested by the data? the equation fits the data.
Which function in part a has the end behavior b. What trend does the model indicate beyond 2000?
suggested by the context? Do you believe that the trend is valid? Explain.
c. Which model would be more appropriate to use c. Use the model to estimate the number of live
to estimate the lead usage in 1955? Use that births in 1940. The actual number was 7558.
model to estimate the usage in 1955. How close is the model estimate?
19. Education The following table shows the amounts d. Do you believe your model would be an accurate
spent on reducing sizes of first-grade through third- predictor of the number of live births to women
grade public school classes in Nevada. 45 years of age and older for the current year?
Why or why not?
Year Amount (millions)
1990 $3 21. Gender Ratio The accompanying table shows the
number of males per 100 females in the United
1992 $31
States calculated using census data. This number is
1994 $37 referred to as the gender ratio.
1996 $42
1998 $66
Year Males per 100 females
(Source: Nevada Department of Education.) 1900 104.6
1910 106.2
a. Examine a scatter plot of the data. How does a
scatter plot indicate that a cubic model might be 1920 104.1
appropriate? 1930 102.6
b. Find a cubic model for the amount spent on class 1940 100.8
size reduction. 1950 98.7
c. Use the model to estimate the amounts in 1993 1960 97.1
and 1999. In which of these estimates can you 1970 94.8
have more confidence? 1980 94.5
d. Compare the estimates from part c with the 1990 95.1
actual amounts of $34 million spent in 1993 and
2000 96.3
$99 million spent in 1999. Does this comparison
support the statements you made in part c (Source: U.S. Bureau of the Census.)
concerning the reliability of the estimates?
a. Examine a scatter plot of the data. What behav-
20. Births The numbers of live births in the United ior of the data suggests that a cubic model is ap-
States for selected years between 1950 and 2000 to propriate?
women 45 years of age and older are as given in the
accompanying table. b. Find a cubic model for the gender ratio. What
trend does the model indicate for the years be-
Year Births Year Births yond 2000? Do you believe that this is a good
predictor of future gender ratios? Why or why
1950 5322 1980 1200
not?
1960 5182 1990 1638
22. The following table shows a manufacturer’s
1970 3146 2000 4604
total cost (in hundreds of dollars) to produce from
(Source: www.infoplease.com, accessed 9/24/02.) 1 to 33 forklifts per week.
84 CHAPTER 1 Ingredients of Change: Functions and Models
a. According to the data, for what age are the num-
Total cost
Weekly production (hundreds of dollars) bers of males and females equal?
1 18.5 b. Find cubic and logistic models for the gender ra-
tio as a function of age. Compare the fit of the
5 80
two equations. Which equation do you believe is
9 125 better for modeling these data?
13 160 c. Use the model you chose in part b to find the age
17 185 for which there are twice as many women as
21 210 men. What does this information tell you about
death rates of men and women?
25 225
29 245 24. Births The accompanying table gives the cesarean
delivery rate per 100 live births in the United States
33 280
from 1990 through 2000.
a. Examine a scatter plot of the data in the table. Cesarean delivery rate
What characteristics of the scatter plot indicate Year (per 100 live births)
that a cubic model would be appropriate? 1990 22.7
b. Find a cubic model for total manufacturing cost. 1992 22.3
c. What does the model predict as the cost to pro- 1994 21.2
duce 23 forklifts per week? 35 forklifts per week? 1996 20.7
d. Convert the cubic equation in part b for total 1998 21.2
cost to one for average cost. Then find the aver-
2000 22.9
age cost of producing 23 forklifts per week and
that of producing 35 forklifts per week. (Source: National Vital Statistics Report,
vol. 50, no. 5, February 12, 2000.)
23. Gender Ratio The gender ratio in the United
States is different for different age groups. The fol-
lowing table shows gender ratios corresponding to a. Examine a scatter plot of the data in the table,
age that were calculated from the 2000 census data. and discuss its curvature. Should the data be
modeled by a quadratic or a cubic equation?
What is the model?
Age Males per 100 females b. What does the model estimate as the Cesarean
under 1 105.0 delivery rate in the United States in 1989 and
10 105.2
1999? Compare the model estimates to the actual
rates of 22.8 births per 100 in 1989 and 22.0
20 104.7
births per 100 in 1999.
30 102.8
40 99.7 Discussing Concepts
50 96.6 25. Using the terms increasing, decreasing, and concave,
60 92.0 describe the shape of the graphs of functions of the
70 81.9 forms y ax2 bx c and y ax3 bx2
cx d.
80 63.1
90 37.6 26. Discuss how to use end-behavior analysis in deter-
mining the differences among linear and polyno-
100 and over 24.9
mial functions and among exponential, log, and
(Source: U.S. Bureau of the Census.) logistic functions.
Chapter 1 Summary 85
SUMMARY
Mathematical Modeling and Functions clearly understand that we use technology simply as a
Mathematical modeling is the process by which we tool in the service of mathematics and that no tool is a
construct a mathematical framework to represent a substitute for clear, effective thinking. Technology car-
real-life situation. In this book we often use mathemat- ries only the graphical and numerical computational
ical modeling to mean fitting a line or curve to data. The burden. You yourself must perform the mathematical
resulting equation, together with output label, input analyses, interpret the results, make the appropriate de-
description, and interval description, which we refer to cisions, and then communicate your conclusions in a
as the mathematical model, provides a representation clear and understandable manner.
of the underlying relationship between the variable
quantities of interest. A function is a description of how Function Combinations and Composition
one thing (output) changes as something else (input) There are several ways to create new functions by com-
changes. We encounter functions represented in four bining two or more other functions whose input and
ways: tables of data, graphs, word descriptions, and output units are compatible. The basic construction
equations. techniques are function addition, subtraction, multipli-
cation, division, and composition. In each of these con-
The Role of Technology structions, knowing the input and output units of the
In order to construct mathematical models from data, functions is the key to understanding how to combine
we must use appropriate tools. Normally, these tools are the functions. Table 1.31 shows the necessary input and
graphing calculators or personal computers. You should output compatibility.
TABLE 1.31
Function Input
operation compatibility Output compatibility New input units New output units
Addition Identical Same unit of measure or units Same as input Same as output unit
of measure capable of being unit of measure of measure of
combined into a larger group of original original functions
(sons daughters children) functions
Subtraction Identical Same unit of measure or units Same as input Same as output unit
of measure capable of being unit of measure of measure of
subtracted (children sons of original original functions
daughters) functions
Multiplication Identical Unit of measure of one Same as input The multiplication
function should contain “per “ unit of measure (reduced if possible)
unit of measure of the other of original of the output unit
function functions of measure of the
original functions
Division Identical Same unit of measure (or) unit Same as input The numerator
of measure of the two functions unit of measure output unit of
should make sense in a phrase of original measure “per” the
containing “. . . per . . .” functions denominator output
unit of measure
Composition Output of one function (inside function) is Same as input Same as output unit
identical to the input of the second function unit of measure of measure of
(outside function) of inside function outside function
86 CHAPTER 1 Ingredients of Change: Functions and Models
Linear Functions and Models in output toward the horizontal axis as the input values
A linear function models a constant rate of change. Its increase.
underlying equation is that of a line: y ax b, where When fitting exponential and logistic equations to
the constant a is called the slope of the line and is data, it is sometimes helpful to shift the output data.
rise This vertical shift is particularly useful when the data ap-
calculated as run. Because the slope of a line is a measure
pear to approach a value other than zero. The goal in
of its rate of increase or decrease, the slope is also known
shifting is to move the data closer to the horizontal axis.
as the rate of change for the linear model. The constant
b appearing in the linear model y ax b is simply Limits and End Behavior
the output of the model when the input is zero.
The idea of a limiting value of a function is a fundamen-
Exponential Functions and Models tal theme of calculus that can be intuitively understood
to be the behavior of the outputs of a function as the in-
Second in importance to linear functions and models
puts of the function become infinitesimally close to a
are exponential functions and models. Based on the fa-
specific value. Limits can also be used to describe the end
miliar idea of repeated multiplication by a fixed positive
behavior of a function as the magnitude of the inputs
multiplier b (the base), the basic exponential function is
becomes infinitely large.
of the form
f(x) abx Polynomial Functions and Models
Polynomial functions and models have a well-
The value a appearing in the equation is the output
established role in calculus. In this text, we consider lin-
when the input is zero.
ear functions, quadratic functions, and cubic functions.
Exponential functions model constant percentage
Quadratic equations have graphs known as parabolas.
change. In terms of the function f(x) abx, exponen-
The parabola with equation f(x) ax2 bx c opens
tial growth occurs when b is greater than 1, and expo-
upward (is concave up) if a is a positive and opens
nential decline (decay) takes place when b is between 0
downward (is concave down) if a is negative.
and 1. The constant percentage growth or decline is
Cubic equations have graphs that show a change of
given by (b 1)100%.
concavity at an inflection point, but unlike logistic mod-
Logarithmic Functions and Models els, they do not have horizontal asymptotes limiting
their end behavior. In using cubic models, we must be
The basic form of the log function that we use is especially careful when extrapolating beyond the range
f(x) a b ln x of data values from which the models are constructed.
The input of this function must be a value greater than Choosing a Model
zero. The log function is useful for situations in which Although it is not always clear which (if any) of the
the output grows or declines at an increasingly slow rate. functions we have discussed apply to a particular real-
When fitting a log equation to data, you must some- life situation, it helps to keep in mind a few general,
times align the input data to ensure that the input values common-sense guidelines: (1) Given a set of discrete
are greater than zero or to obtain a better fit. Aligning data, begin with a scatter plot. The plot will often reveal
input data has the effect of shifting the data horizontally. general characteristics that point the way to an appro-
priate model. (2) If the scatter plot does not appear to
Logistic Functions and Models be linear, consider the suggested concavity. One-way
Initial exponential growth followed by a leveling-off ap- concavity (up or down) suggests a quadratic, exponen-
proach toward a limiting value L is characteristic of lo- tial, or log model. (3) When a single change in concav-
gistic growth, which is modeled by the logistic equation ity seems apparent, think in terms of cubic or logistic
models. But remember that the graphs of logistic mod-
L
f(x) els tend to become flat on each end, whereas the graphs
1 Ae Bx
of cubic models do not. Never consider using a cubic
If the constant B is positive, the model indicates growth. or a logistic model if you cannot identify an inflection
If the constant B is negative, the model indicates decline point.
Chapter 1 Summary 87
CONCEPT CHECK
Can you To practice, try
• Identify functions, inputs, and outputs? Section 1.1 Activities 1, 5, 9
• Interpret models? Section 1.1 Activities 11, 13
• Work with functions? Section 1.1 Activities 19, 21, 23, 29, 31
• Combine or compose two functions? Section 1.1 Activities 35, 41, 43, 51, 53
• Construct a linear model given a constant rate of change? Section 1.2 Activities 9, 11, 17, 19
• Find a linear model and determine its rate of change? Section 1.2 Activities 17, 23, 26
• Construct an exponential model given a constant Section 1.3 Activities 17, 19
percentage change?
• Find an exponential model and determine its Section 1.3 Activities 21, 23
percentage change?
• Solve exponential growth and decay problems? Section 1.3 Activities 15, 27
• Find and use a log model? Section 1.3 Activities 31, 33
• Find and use a logistic model? Section 1.4 Activities 15, 17
• Determine end behavior? Section 1.4 Activities 19, 21, 23
• Find and use a quadratic model? Section 1.5 Activities 17, 19
• Find and use a cubic model? Section 1.5 Activities 23, 25
• Identify concavity and inflection points? Section 1.5 Activities 11, 13, 15
CONCEPT REVIEW
For the figures in Activities 1 through 6:
a. Describe the direction and curvature of the scatter c. Determine what type of model would best describe
plot, noting concavity. the data.
b. Describe the apparent end behavior of the scatter d. Assuming the function f(x) from part c has been fit
plot in both the negative direction and the positive to the data, use limit notation to describe the end
direction. behavior of f(x).
88 CHAPTER 1 Ingredients of Change: Functions and Models
1. y 2. y 8. Suppose S(t) is the total number of applicants to
40 40 your college or university in year t, and M(t) is the
number of male applicants to your college or uni-
30 30
versity in year t.
20 20 a. Draw input/output diagrams for S(t) and M(t).
10 10 b. What function operation must be applied to S(t)
and M(t) to construct function F, which gives the
x x
1 2 3 4 5 6 1 2 3 4 5 number of female applicants to your college or
university in year t?
3. y 4. y c. Write the new function F using appropriate
function notation.
20 20
d. Write the input and output units for the new
15 15
function.
10 10
9. Suppose R(x) is the revenue, in millions of dollars,
5 5 of a company x years after it has been in business,
and C(x) is the cost, in thousands of dollars, of a
x x
2 4 6 1 2 3 4 5 6 company x years after it has been in business.
a. Draw input/output diagrams for R(x) and C(x).
5. y 6. y b. What function can be constructed from R(x) and
40 40 C(x)?
30
c. Write the new function P using appropriate
30
function notation.
20 20
d. Write a statement of the model constructed in
10 10 part c.
x x 10. Demand Suppose the daily demand for beef can
1 2 3 4 5 6 1 2 3 4 5 6 40
be modeled by D(p) 1 0.03e0.4p million pounds,
7. At a large university, there is no student parking where p is the price of beef in dollars per pound.
available on the main campus. A transit bus is avail-
a. What function operation can be used with the
able to transport students from satellite parking to
given model to construct a function for revenue?
main campus. Suppose S(x) is the number of stu-
dents who rode the bus, where x is the number of b. Write the new function P using appropriate
hours after 7 A.M., and C(s) is the number of student function notation.
cars on the main campus that have parking tickets, c. Write a statement of the model constructed in
where s is the number of students who rode the bus? part b.
a. Draw input/output diagrams for S(x) and C(s). 11. Rearing Children According to the U.S. Census
b. What function operation must be applied to S(x) Bureau, the number of children living with their
and C(s)to construct function T, which gives the grandparents was 2.2 million in 1970 and 3.9 mil-
number of cars with parking tickets on them x lion in 1997. Assume that the number grew expo-
hours after 7 A.M. nentially between 1970 and 1997.
c. Write the new function T using appropriate a. Find an exponential model for the number of
function notation. children living with their grandparents as a func-
d. Write the input and output units for the new tion of the number of years since 1970.
function. b. What is the percentage change indicated by the
model?
Chapter 1 Concept Review 89
c. According to the model, when will the number month during 2002. At the beginning of 2002,
of children living with their grandparents reach 5 the jeep already had 68 thousand miles on it.
million? Write a model for the mileage on the jeep at the
d. What is the doubling time for the model found end of the xth month of 2002. What is the slope
in part a? Interpret the doubling time in context. of this model? What is its rate of change?
12. Demand When a certain variety of fish is mar- d. Use the models developed in parts a and c to con-
keted for human consumption, the demand for that struct a model giving the resale value of the jeep
fish can be modeled as D(p ) 6.25(0.88p ) trillion at the end of the xth month of 2002.
pounds when the fish is marketed for p dollars per 14. Epidemic In 1949 the United States experienced
pound. the second-worst polio epidemic in its history. (The
a. Is the function D an increasing or a decreasing worst was in 1952.) The accompanying table gives
exponential function? Looking only at the pa- the cumulative number of polio cases diagnosed on
rameters of the function, what indicates this in- a monthly basis.
crease or decrease?
b. Find the demand for fish when the market price
is 7 dollars; 14 dollars; 21 dollars.
Month Total number of polio cases
c. What type of model is an inverse to an exponen-
tial model? January 494
d. Using the information found in part b, find a February 759
model for market price given the amount of fish March 1016
in demand. April 1215
13. Jeep Value The following table shows the 2002 May 1619
private-party resale value of a 2000 Jeep Grand June 2964
Cherokee Laredo in excellent condition as a func- July 8489
tion of the mileage.
August 22,377
September 32,618
Mileage Resale value
(thousands) (dollars) October 38,153
20 18,520 November 41,462
40 17,120 December 42,375
60 14,670 (Source: Twelfth Annual Report, National Foundation for
Infantile Paralysis, 1949.)
80 13,295
100 12,745
120 12,270
a. Examine a scatter plot of the data. Describe the
(Source: www.kbb.com, accessed 9/30/01.) concavity, and estimate the location of the inflec-
tion point indicated by the scatter plot. What
a. Examine a scatter plot of the data, and discuss its type of model would best describe these data?
curvature. Why is a quadratic equation a good fit b. Describe the end behavior suggested by the con-
for these data? What is the model? text.
b. Use your model from part a to estimate the resale c. What type of model would be most appropriate
value for this jeep at 52 thousand miles. to describe these data and this context?
c. Suppose the owner of a 2000 Jeep Grand Chero- d. Write a model (of the type you chose in part c)
kee Laredo drives the jeep 4 thousand miles each for these data.
Project 1.1 Compulsory School Laws
Setting Tasks
In 1852, Massachusetts became the first state to enact 1. Tabulate the cumulative number of states with
a compulsory school attendance law. Sixty-six years compulsory school attendance laws for the 5-year
later, in 1918, Mississippi became the last state to periods shown in the following table.
enact a compulsory school attendance law. The
following table lists the first 48 states to enact such
1852–1856 1887–1891
laws and the year in which each of these states
enacted its first compulsory school attendance law. 1857–1861 1892–1896
1862–1866 1897–1901
1867–1871 1902–1906
State Year State Year State Year
1872–1876 1907–1911
MA 1852 SD 1883 IA 1902
1877–1881 1912–1916
NY 1853 RI 1883 MD 1902
1882–1886 1917–1921
VT 1867 ND 1883 MO 1905
MI 1871 MT 1883 TN 1905
2. Examine a scatter plot of the data in Task 1. Do
WA 1871 IL 1883 DE 1907 you believe a logistic model is appropriate?
NH 1871 MN 1885 NC 1907 Explain.
CT 1872 ID 1887 OK 1907 3. Find a logistic model for the data in Task 1.
NM 1872 NE 1887 VA 1908
4. What do most states in the third column of the
NV 1873 OR 1889 AR 1909 original data have in common? Why would these
KS 1874 CO 1889 TX 1915 states be the last to enact laws related to
CA 1874 UT 1890 FL 1915
compulsory education?
ME 1875 KY 1893 AL 1915 5. The 17 states considered to be southern states
NJ 1875 PA 1895 SC 1915
(below the Mason-Dixon Line) are AL, AR, DE,
FL, GA, KY, LA, MD, MO, MS, NC, OK, SC, TN,
WY 1876 IN 1897 LA 1916
TX, VA, and WV. Tabulate cumulative totals for
OH 1877 WV 1897 GA 1916 the southern states and the northern/western states
WI 1879 AZ 1899 MS 1918 for the dates shown in the following table.
(Source: J. Richardson, “Variation in Date of Enactment of
Compulsory School Attendance Laws,” Sociology of Education,
vol. 53 (July 1980), pp. 153–163.)
90
Project 1.1 Compulsory School Laws, continued
Northern/western states Southern states
cumulative northern state and western state totals
for the following 10-year periods:
1852–1856 1891–1895
1857–1861 1896–1900 1852–1861
1862–1866 1901–1905 1862–1871
1867–1871 1906–1910 1872–1881
1872–1876 1911–1915 1882–1891
1877–1881 1916–1920 1892–1901
1882–1886 1902–1911
1887–1891
1892–1896 10. Find a logistic model for the data in Task 9.
Comment on how well the function fits the data.
1897–1901
Compare models for the data grouped in 10-year
1902–1906 intervals and the data grouped in 5-year intervals
(Task 6). Does how the data are grouped
6. Examine scatter plots for the two data sets in Task significantly affect how well the equation fits?
5. Do you believe that logistic models are Explain.
appropriate for these data sets? Explain. 11. Find an equation that fits the data for the
7. Find logistic models for each set of data in Task southern states better than the logistic equation.
5. Comment on how well each equation fits the Write the model using the better-fitting equation.
data. Explain your reasoning.
8. It appears that the northern and western states
Reporting
were slow to follow the lead established by
Massachusetts and New York. What historical 1. Prepare a written report of your work. Include
event may have been responsible for the time lag? scatter plots, models, and graphs. Include
discussions of each of the tasks in this project.
9. One way to reduce the impact of unusual behavior
in a data set (such as that discussed in Task 8) is to 2. (Optional) Prepare a brief (15-minute)
group the data in a different way. Tabulate the presentation of your work.
91
Project 1.2 Fundraising Campaign
Setting Tasks
In order to raise funds, the mathematics department 1. Develop a slogan and a design for the T-shirt.
in your college or university is planning to sell Keep in mind that good taste is a concern. Decide
T-shirts before next year’s football game against the on a target market, and determine a strategy to
school’s biggest rival. Your team has volunteered to survey (at random) at least 100 students who
conduct the fundraiser. Because several other student represent a cross section of the target market to
groups have also volunteered to head this project, determine the demand for T-shirts (as a function
your team is to present its proposal for the fund of price) within that market. It is important that
drive, as well as predictions about its outcome, to a your sample survey group properly represent your
panel of mathematics faculty. target market. If, for example, you polled only near
campus dining facilities at lunch time, your sample
would be biased toward students who eat lunch at
such facilities.
The question you should ask is “How much
would you be willing to spend on a T-shirt
promoting the big football rivalry: $20, $18, $16,
$14, $12, $10, $8, or not interested?” Keep an
accurate tally of the number of students who
answer in each category. In your report on the
results of your poll, you should include
information such as your target market; where,
when, and how you polled within that market; and
why you believe that your polled sample is likely to
be a representative cross section of the market.
2. a. From the data you have gathered, determine
how many students from your sample survey
group would buy a T-shirt at $20, $18, $16, and
so on.
Note: This project is also used as a portion of Project 4.2 on
page 282.
92
Project 1.2 Fundraising Campaign, continued
b. Devise a marketing strategy, and determine Reporting
how many students within your target market
1. Prepare a written report summarizing your survey
you can reasonably expect to reach. Assuming
and modeling. The report should include your
that your poll is an accurate indicator of your
slogan and design, your target market, your
target population, determine the number of
marketing strategy, the results from your poll (as
students from your target market who will buy
well as the specifics of how you conducted your
a T-shirt at each of the given prices.
poll), a discussion of how and why you chose the
c. Taking into account the results of your poll and model of the demand function, a discussion of the
your projected target market, develop a model accuracy of your demand model, and graphs and
for demand as a function of price. Keep in mind equations for all of your models. Attach your
that your model must make sense for all questionnaire and data to the report as an
possible input values. appendix.
3. Use the partial price list (located on the website)
2. Prepare a 15-minute oral presentation of your
to model a function for the cost to you when
survey, modeling, and marketing strategy to be
ordering T-shirts. Use the demand function from
delivered before a panel of mathematics faculty.
Task 2c to create equations for revenue, total cost,
You will be expected to have overhead
and profit as functions of price. (Revenue, total
transparencies of all graphs and equations, as well
cost, and profit may not be among the basic
as any other information that you consider
models that were discussed in class. They are sums
appropriate as a visual aid. Remember that you are
and/or products of the demand function with
trying to sell the mathematics department on your
other functions.)
campaign idea.
93
2 Describing Change: Rates
Concepts Outline
2.1 Change, Percentage Change,
and Average Rates of Change
2.2 Instantaneous Rates of
Change
2.3 Derivative Notation and
Numerical Estimates
2.4 Algebraically Finding Slopes
Ted Horowitz/Corbis
Concept Application
The passenger airline industry has spent the first several years of the twenty-first century
undergoing extensive changes in the way that it operates. Many of the airlines that
operated during the latter part of the 1900s are no longer operating, and those that are
still operating have had to adapt quickly in order to stay financially solvent. The changes
that airline companies have had to react to include fluctuations in industry-wide demand
for seats, fluctuations in market share, newly instituted security procedures, and increases
in fuel prices. Airlines can use calculus to help answer questions such as the following:
• How quickly was the number of enplaned passengers changing at the end of 2005?
• At what point during the last decade was the price of Northwest Airline’s common
stock decreasing most rapidly?
• What was the percentage rate of change of airline fuel prices at the beginning of 2006?
This chapter will provide you with some of the tools that make it possible to answer
questions like these. To estimate answers to such questions, see Activity 23 of Section 2.3,
Activity 17 of Section 2.4, and Activity 7 of the Concept Review.
94
2.1 Change, Percentage Change, and Average Rates of Change 95
Chapter Introduction
Change is everywhere around us and affects our lives on a daily basis. In this chapter we
consider several ways to describe change. Starting from the actual change in a quantity
over an interval, it is a simple step to describe the change as a percentage change or as
an average rate of change over that interval. The notion of average rate of change, when
examined carefully in light of the underlying geometry of graphs, leads to the more sub-
tle and challenging concept of instantaneous change. Indeed, the precise description of
instantaneous change is a principal goal of calculus.
We therefore turn our attention to determining instantaneous rates of change.
Because instantaneous rates of change are slopes of tangent lines, we consider the
numerical estimation of these slopes. Then we generalize the numerical method to an
algebraic method that gives us an analytic description—a formula of sorts—for the
derivative of an arbitrary function.
Concepts You Will Be Learning
• Determining and interpreting change, percentage change, and average rate of
change (2.1)
• Finding and interpreting the APY and APR in compound interest formulas (2.1)
• Estimating and interpreting slopes of secant and tangent lines on a graph (2.1, 2.2,
2.3)
• Relating the location of a tangent line to the concavity of a graph (2.2)
• Recognizing points on a graph where a derivative does not exist (2.2)
• Using derivative notation and interpreting given derivative values in context (2.3)
• Estimating and interpreting percentage rate of change (2.3)
• Numerically estimating rates of change (2.3)
• Using limits to find rate-of-change formulas algebraically (2.4)
2.1 Change, Percentage Change, and Average Rates of Change
Calculus gives a description of change. In preparation for learning how calculus de-
scribes change, we introduce three numerical ways of reporting change. We begin
with a simple business example.
The yearly revenue for a large department store declined from $1.4 billion in 2000
to $1.1 billion in 2003. There are three common ways to express this change in revenue.
We could use a negative sign to indicate the decrease in revenue and say that the
change in revenue was $0.3 billion over 3 years. We could also express this change
by saying that the revenue declined by $0.3 billion (or $300 million) over 3 years.
It is often helpful to express the change as a percent of the 2000 revenue. This
percentage change (or percent change) is found by dividing the change by the
revenue in 2000 and multiplying by 100:
$0.3 billion
Percentage change 100
$1.4 billion
21.4%
96 CHAPTER 2 Describing Change: Rates
The company saw a 21.4% decline in revenue between 2000 and 2003. You should
already be familiar with percentage change from the discussion of exponential mod-
els. This method of calculating percentage change is the same as the way we calculated
percentage differences using data in Section 1.3. However, you should note that even
though constant percentage difference over equally spaced intervals is unique to
exponential models, the percentage change in a quantity over a specific interval may
be calculated and interpreted in almost any context.
The third way to express change involves evenly spreading the change over the in-
put interval to obtain the average rate of change. This is done by dividing the change
by the length of the interval. In the case of the department store revenue, the length
of the interval is 3 years:
$0.3 billion
Average rate of change
3 years
$0.1 billion per year
On average, the revenue declined at a rate of $0.1 billion per year (or $100 million per
year) between 2000 and 2003.
We summarize these three ways to describe change in the following box.
Change, Percentage Change, and Average Rate of Change
If a quantity changes from a value of m to a value of n over a certain interval
from a to b, then
• The change in the quantity is found by subtracting the first value from the
second.
Change n m
• The percentage change is the change divided by the first value and then
multiplied by 100.
change nm
Percentage change 100% 100%
first value m
• The average rate of change is the change divided by the length of the
interval.
change nm
Average rate of change
length of interval ba
Interpreting Descriptions of Change
Correctly calculating these descriptions of change is important, but being able to state
your result in a meaningful sentence in the context of the situation is equally
important. When interpreting a description of change, you should answer the
questions when, what, how, and by how much.
2.1 Change, Percentage Change, and Average Rates of Change 97
Interpreting Descriptions of Change
When describing change over an interval, be sure to answer the following
questions:
• When? Specify the interval.
• What? Specify the quantity that is changing.
• How? Indicate whether the change is an increase or a decrease.
• By how much? Give the numerical answer labeled with proper units:
Description Units
change output unit of measure
percentage change percent
average rate of change output unit of measure per single
input unit of measure
When stating an average rate of change, use the word average or the phrase
on average.
EXAMPLE 1 Describing Change Using a Table
Temperature Consider Table 2.1, which shows temperature values on a typical
May day in a certain midwestern city. Find the following descriptions of change, and
write a sentence giving the real-life meaning of (that is, interpret) each result.
TABLE 2.1
Time Temperature (°F) Time Temperature (°F)
7 A.M. 49 1 P.M. 80
8 A.M. 58 2 P.M. 80
9 A.M. 66 3 P.M. 78
10 A.M. 72 4 P.M. 74
11 A.M. 76 5 P.M. 69
noon 79 6 P.M. 62
a. The change in temperature from 7 A.M. to 1 P.M.
b. The percentage change in temperature between 3 P.M. and 6 P.M.
c. The average rate of change in temperature between 8 A.M. and 5 P.M.
98 CHAPTER 2 Describing Change: Rates
Solution
a. To find the change in temperature, subtract the temperature at 7 A.M. from the
temperature at 1 P.M.
Change 80°F 49°F 31°F
The interpretation statement answers the questions when (1 P.M. and 7 A.M.),
what (the temperature), how (was greater than), and by how much (31°F): The
temperature at 1 P.M. was 31°F greater than the temperature at 7 A.M.
b. To find percentage change, first find the change in temperature between 3 P.M.
and 6 P.M.
Change 62°F 78°F 16°F
Next, divide the change by the temperature at 3 P.M. (the beginning of the time
interval under consideration), and multiply by 100.
16°F
Percentage change 100% 20.5%
78°F
The temperature declined by about 20.5% between 3 P.M. and 6 P.M. In this
statement, “between 3 P.M. and 6 P.M.” answers the question when, “the
temperature” again identifies what, and “declined by about 20.5%” tells how
and by how much.
c. Begin the calculation of the average rate of change by finding the change in the
temperature from 8 A.M. to 5 P.M.
Change 69°F 58°F 11°F
Next, divide the change by the length of the time interval (9 hours).
11°F
Average rate of change 1.2°F per hour
9 hours
In interpreting this average rate of change, we must use the word average in our
sentence in addition to answering the four questions. We state the interpretation
as follows: Between 8 A.M. and 5 P.M., the temperature rose at an average rate of
1.2°F per hour. ●
Although these descriptions of change are useful, they have limitations. It appears
from the answer to part c of Example 1 that the temperature rose slowly throughout
the day. However, the average rate of change does not describe the 22°F rise in
temperature followed by the 11°F drop in temperature that occurred between 8 A.M.
and 5 P.M.
Finding Percentage Change and Average Rate of
Change Using Graphs
You may have noticed that calculating the average rate of change is the same as
calculating slope. This observation allows for the easy calculation of average rates of
change if you are given a graph. For instance, when plotted, the May daytime
temperatures in Example 1 fall in the shape of a parabola (see Figure 2.1).
2.1 Change, Percentage Change, and Average Rates of Change 99
Air temperature
(°F)
80
70
60
50
40
Time
FIGURE 2.1 8 A.M. 10 A.M. 12 P.M. 2 P.M. 4 P.M. 6 P.M.
To find the average rate of change between 9 A.M. and 4 P.M., first use a straightedge
to draw carefully a line connecting the points at 9 A.M. and 4 P.M. (see Figure 2.2). We
call this line connecting two points on a scatter plot or graph a secant line (from the
Latin secare, “to cut”). Next, approximate the slope of this secant line by estimating
the rise and the run for a portion of the line (see Figure 2.3).
Air temperature Air temperature
(°F) (°F)
80 80
70 70 Rise = 8°
Run = 7 hours
60 60
50 50
40 40
Time Time
8 A.M. 10 A.M. 12 P.M. 2 P.M. 4 P.M. 6 P.M. 8 A.M. 10 A.M. 12 P.M. 2 P.M. 4 P.M. 6 P.M.
FIGURE 2.2 FIGURE 2.3
rise change in temperature
Average rate of change
run change in time
8°F
1.1°F per hour
7 hours
Between 9 A.M. and 4 P.M., the temperature rose at an average rate of 1.1°F per hour.
Note that the rise is the change in temperature. To calculate percentage change from
the graph, divide the rise by the estimated output of the first point.
change in temperature 8°F
Percentage change 100% 12.1%
temperature at 9 A.M. 66°F
100 CHAPTER 2 Describing Change: Rates
It is important to note that this graphical method of calculating descriptions of
change is imprecise if you are given only a scatter plot or a graph. It gives only
approximations of change, percentage change, and average rate of change. The
method depends on drawing the secant line accurately and then correctly identifying
two points on the line. Slight variations in sketching are likely to result in slightly
different answers. This does not mean that the answers you obtain are incorrect. It
simply means that descriptions of change obtained from graphs are approximations.
Example 1 and the subsequent discussion of air temperature use the term between
2002 total two time values. There are other ways to describe intervals on the input axis, and we
reported take a moment now to discuss one of them.
2002 data
collected When we use data that someone else has collected, we often do not know when
the data were reported or recorded. It seems logical to assume that yearly (or monthly
or hourly and so forth) totals are reported at the end of the intervals representing
2001 2002 those periods. For instance, a 2002 total covers the period of time from the end of
FIGURE 2.4 2001 through the end of 2002 (see Figure 2.4).
Therefore, we adopt the convention that “from a through b” on the input axis
refers to the interval beginning at a and ending at b. If a and b are years and the output
represents a quantity that can be considered to have been measured at the end of the
year, then “from a to b” means the same thing as “from the end of year a through the
end of year b.” For our purposes, the phrases “between a and b” and “from a to b”
have the same meaning as “from a through b.” We use this terminology in the
remainder of the text.
EXAMPLE 2 Describing Change Using a Graph
Social Security The Social Security assets of the federal government between 2002
and 2030, as estimated by the Social Security Advisory Board, are shown in Figure 2.5.
A smooth curve connecting the points is also shown.
Social Security assets
(billions)
$4000
$3000
$2000
$1000
$0 Year
2002 2007 2012 2017 2022 2027 2032
FIGURE 2.5
(Source: www.ssab.gov accessed on 10/1/02.)
2.1 Change, Percentage Change, and Average Rates of Change 101
a. Estimate the change, percentage change, and average rate of change in Social
Security assets between 2002 and 2017. Write a sentence interpreting each
answer in context.
b. Estimate the change, percentage change, and average rate of change in Social
Security assets between 2022 and 2027. Write a sentence interpreting each
answer in context.
Solution
a. Begin by estimating from the graph the Social Security assets in 2002 and 2017.
One possible estimate is $1200 billion in 2002 and $3600 billion in 2017. To
calculate change, subtract the 2002 value from the 2017 value.
Change 3600 1200 $2400 billion
Social Security assets are expected to rise about $2400 billion between 2002 and
2017.
Convert this change to percentage change by dividing by the estimated assets in
2002.
$2400 billion
Percentage change 100% 200%
$1200 billion
Social Security assets are expected to increase about 200% between 2002 and
2017.
Convert the change to an average rate of change by dividing the change by the
length of the interval between 2002 and 2017.
$2400 billion
Average rate of change
15 years
$160 billion per year
Social Security assets are expected to increase by an average of about $160 bil-
lion per year between 2002 and 2017. The average rate of change is the slope of
the secant line through the point corresponding to 2002 and the point corre-
sponding to 2017 shown in Figure 2.6.
b. Begin by estimating the output values in 2022 and 2027. We estimate $3700 bil-
lion in 2022 and $2600 billion in 2027. Following the same procedure as in part
a, we have
Change 2600 3700 $1100 billion
$1100 billion
Percentage change 100% 29.7%
$3700 billion
$1100 billion
Average rate of change $220 billion per year
5 years
102 CHAPTER 2 Describing Change: Rates
Social Security assets
(billions)
$4000 Run = 5 years
Rise ≈
$3000 −$1100 billion
Rise ≈
$2000 $2400 billion
$1000 Run = 15 years
$0 Year
FIGURE 2.6 2002 2007 2012 2017 2022 2027 2032
Between 2022 and 2027, Social Security assets are expected to decrease by about
$1100 billion. This represents a percentage decrease of about 29.7%. On aver-
age, Social Security assets are expected to decrease by about $220 billion per
year during this time. Again, the average rate of change is the slope of the secant
line connecting the points on the graph that correspond to 2022 and 2027. (See
Figure 2.6.) ●
Determining Percentage Change and Average Rate
of Change Using an Equation
It is also possible to determine descriptions of change when we are given only an
equation. We use the equation to determine numerical points with which to work
and then calculate the change, percentage change, and average rate of change just as
we would if we had been given data. Example 3 illustrates using an equation to deter-
mine descriptions of change.
EXAMPLE 3 Descriptions of Change from an Equation
A model for the temperature data on a typical May day in a certain midwestern
city is
Temperature 0.8t 2 2t 79°F
where t is the number of hours after noon.
Calculate the percentage change and average rate of change between 11:30 A.M. and
6 P.M.
Solution
Note that at 11:30 A.M., t 0.5 and that at 6 P.M., t 6.
Substitute t 0.5 and t 6 into the equation to obtain the corresponding
temperatures.
At 11:30 A.M.: Temperature 0.8(0.5)2 2(0.5) 79 77.8°F
At 6 P.M.: Temperature 0.8(6)2 2(6) 79 62.2°F
2.1 Change, Percentage Change, and Average Rates of Change 103
We find the change in temperature from 11:30 A.M. to 6 P.M. to be
62.2°F 77.8°F 15.6°F.
To calculate percentage change, divide the change in temperature by the temperature
at 11:30 A.M. and multiply by 100.
62.2oF 77.8oF
100% 20%
77.8oF
To find the average rate of change, divide the change in temperature by the change in
time.
62.2°F 77.8°F 15.6°F
2.4°F per hour
6 (0.5) 6.5 hours
Thus, between 11:30 A.M. and 6 P.M., the temperature fell by 15.6°F. This represents a
20% decline and an average rate of decline of 2.4°F per hour. ●
Note that in Example 3, the temperature fell 15.6°F in 6.5 hours; however, when
finding an average rate of change, we state the answer (in this case) as the number of
degrees per one hour.
APR and APY
The world of finance yields a classic example of an important use of percentage
change. On an investment of P dollars invested at an interest rate of 100r%
compounded annually, the dollar amount accumulated after t years is given by the
interest formula
A P(1 r)t dollars
On an investment of P dollars invested at an interest rate of 100r% compounded n
times each year, the dollar amount accumulated after t years is given by the interest
formula
nt
r
AP 1 dollars
n
Suppose that $1000 is invested for one year in each of two accounts, one paying 6%
compounded annually and the other paying 6% compounded monthly. The nominal
rate or annual percentage rate (APR) for each investment is 6%.
When we write the formula for the amount in the account with interest paid
annually, the exponential form is A 1000(1 0.06)t 1000(1.06)t . In one year,
the amount accumulated in this account will increase by 6%. The percentage change
is 6%. This percentage change of the accumulated amount is called the effective rate
or annual percentage yield (APY).
For the investment with interest paid monthly, the annual percentage yield (APY)
is slightly more than 6%; when we write the formula for the amount accumulated, we
see the exponential form
12t
A 1000 1
0.06
12
1000(1.005)12t 1000(1.06168)t
104 CHAPTER 2 Describing Change: Rates
In one year, the amount accumulated in this account will increase by approxi-
mately 6.168%. The percentage change is 6.168%. For the investment with interest
paid monthly, the annual percentage yield (APY) is slightly more than 6%.
When an investment has a nominal rate of 100r% compounded n times during
r
the year, the constant percentage change over the compounding period is 100n %. In
0.06
the case of the investment with 6% APR compounded monthly, 12 0.005 0.5%
is the constant percentage change over each 1-month period.
The following box summarizes our discussion of the general compound interest
formula.
Compound Interest Formula
The amount accumulated in an account after t years when P dollars are
invested at an annual interest rate of 100r% compounded n times a year is
r nt
AP 1 dollars
n
The nominal rate or annual percentage rate (APR) is the percentage 100r%,
and the percentage change of the amount accumulated over one compounding
r
period is 100n%.
The effective rate or annual percentage yield (APY) is the percentage
change of the amount accumulated over one year.
We illustrate the use of the general compound interest formula in Example 4.
EXAMPLE 4 Using the Compound Interest Formula
Suppose you are 25 years old and have $10,000 to invest for retirement.
a. What APR compounded monthly is needed for your money to grow to one
million in 40 years?
b. What is the APY for this investment?
c. What are the annual and monthly percentage changes for this investment?
Solution
a. To answer the question posed, substitute 40 for t and 1,000,000 for A(t) in the
formula A(t) 10,0001 12 , and solve for r.
r 12t
r 12(40)
1,000,000 10,000 1
12
Solving for r yields r 0.11568.
An APR of approximately 11.57% compounded monthly is needed for $10,000
to grow to a million dollars in 40 years.
2.1 Change, Percentage Change, and Average Rates of Change 105
b. The APY that corresponds to the APR in part a is found by substituting the APR
into the formula for A(t) and rewriting in the form A(t) abt .
0.11568 12t
A(t) 10,000 1 10,000(1.1220t)
12
The APY is (b 1)100% 12.20%.
c. The annual percentage change in the amount of the investment is given by the
APY, 12.20%. Because the interest is compounded monthly, the monthly per-
centage change in the amount of the investment is given by the APR divided by
12, or approximately 0.964%. ●
Even though the compound interest formula is a continuous function, it has a
discrete interpretation because the amount changes only at the actual times of
compounding. For instance, we can use the monthly compounding function
A(t) 1000(1.06168t) dollars after t years to find the amount of the investment at
the end of the 3rd month of the 6th year by calculating A(6.25) $1453.63. (This
value was calculated using the unrounded value of the base.) However, it would be
incorrect to use t 6.2 to calculate the amount in the account on the 14th day of the
3rd month of the 6th year, because interest is compounded monthly, not daily.
There is one well-known interest formula that has a continuous interpretation:
the continuously compounded interest formula.
Continuously Compounded Interest Formula
The amount accumulated in an account after t years when P dollars are
invested at a nominal rate (APR) of 100r% compounded continuously is
A Pert dollars
This formula is used to model situations where interest is considered to be com-
pounded continuously. It arises as a direct result of considering what happens to the
compound interest formula as compoundings occur more and more frequently.
Given a 1-year period, the more often compounding occurs, the larger the number of
compoundings n becomes. In order to consider compoundings to occur continu-
ously, we must consider what happens as n becomes infinitely large. Numerically, it
1 n
lim 1 n e 2.71828182846 . By using algebra, it can be
can be shown that n→
1 nt
lim P1 n Pe rt . We will explore these limits numerically in the
shown that n→
activities.
EXAMPLE 5 Determining APR from APY for Continuous Compounding
An investment that has interest compounded continuously has an APY of 9.2%.
What is its APR?
106 CHAPTER 2 Describing Change: Rates
Solution
For an investment of P dollars, we consider that we have a constant yearly percentage
increase of 9.2%. We will use an exponential growth model with a P and b 1.092.
We model the amount in the investment with an initial deposit of P dollars as
A(t) P(1.092t) dollars after t years, t 0
The APR for continuously compounded interest is represented by r in the formula
A Pert. We set this expression equal to the formula for A(t) and solve for r.
P(1.092t ) Pert
Solving for r yields r 0.0880. The APR is approximately 8.8%. ●
When comparing compound interest investment opportunities, you should
always consider annual percentage yields, because the nominal rates do not reflect the
effect of the compounding periods.
EXAMPLE 6 Comparing Annual Percentage Yields
Consider three investment offers: an APR of 6.9% compounded annually, an APR of
6.8% compounded monthly, and an APR of 6.7% compounded continuously.
Determine which investment offers the best annual percentage yield.
Solution
In order to compare the three investments, we must consider the percentage change
of each over the same time interval. The easiest time interval to compare for all three
investments will be a 1-year interval. We begin by writing a formula for each of the
investments and converting those three formulas into the form f(t) abt .
Investment Offer A: APR 6.9% compounded annually
A(t) P(1 0.069)t P(1.069t)
APY is 6.9%
Investment Offer B: APR 6.8% compounded monthly
B(t) P1 12 P(1.07015t)
0.068 12t
APY is approximately 7.02%
Investment Offer C: APR 6.7% compounded continuously
C(t) Pe0.067t P(1.06930t)
APY is approximately 6.93%
The investment offer with the best annual percentage yield is Offer B, with an APR of
6.8% compounded monthly. ●
All three descriptions of change (change, percentage change, and average rate of
change) discussed in this section are valuable, but the concept of average rate of
change will be the bridge between the algebraic descriptions of change examined in
this section and the calculus description of change that we begin exploring in the next
section. The graphical interpretation of the average rate of change as a slope of a
secant line will be particularly useful.
2.1 Change, Percentage Change, and Average Rates of Change 107
specified. Write a sentence interpreting each description
2.1 Concept Inventory of change.
5. Insurance In 2005 Northwest Airlines was the
• Change
fourth-largest airline in the world. Northwest flew
• Percentage change 55.4 million enplaned passengers during 2004 and
• Average rate of change 56.5 million enplaned passengers during 2005.
• Secant line (Source: Northwest Airlines, December 2005 Financial/Traffic
Release.)
• Slope of a secant line average rate of change
6. Act Scores The national ACT college test com-
• Descriptions of change
posite average for females was 20.7 in 2002 and 20.9
• Compound interest, APR, and APY in 2005.
(Source: ACT, Inc.)
7. Population The American Indian, Eskimo, and
Aleut population in the United States was 362 thou-
2.1 Activities sand in 1930 and 2434 thousand in 2000.
(Source: U.S. Bureau of the Census.)
Getting Started
Rewrite the sentences in Activities 1 through 4 to express 8. Internet The number of Internet users in China
how rapidly, on average, the quantity changed over the grew from 12 million in 2000 to 103 million in 2005.
given interval. (Source: BDA (China), The Strategis Group and China Daily,
July 22, 2005.)
1. In five trading days, the stock price rose $2.30.
Applying Concepts
2. The nurse counted 32 heartbeats in 15 seconds.
3. The company lost $25,000 during the past 9. October Madness
3 months. a. On October 1, 1987, 193.2 million shares were
traded on the stock market. On October 30,
4. The unemployment rate has risen 4 percentage
1987, 303.4 million shares were traded. Find the
points in the past 3 years.
percentage change and average rate of change in
For Activities 5 through 8, calculate the change, percent- the number of shares traded per trading day
age change, and average rate of change over the interval between October 1 and October 30, 1987.
Shares
(millions)
600
500
400
300
200
100
Trading days
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
(Oct 1) (Oct 19) (Oct 30)
Figure for Activity 9
(Source: Phyllis S. Pierce, ed., The Dow Jones Averages 1885–1990, Homewood, IL: Business One Irwin, 1991.)
108 CHAPTER 2 Describing Change: Rates
b. The scatter plot in the figure on page 107 shows 2001 by Kelly Services, Inc., a leading global
the number of shares traded each day during Oc- provider of staffing services, is shown below.
tober of 1987. On the scatter plot, sketch a line
whose slope is the average rate of change between Sales
October 1 and October 30, 1987. (billions of dollars)
c. The behavior of the graph on October 19 and 20 5
has been referred to as “October Madness.”
Write a sentence describing how the number of 4
shares traded changed throughout the month. 3
How well does the average rate of change you
found in part a reflect what occurred throughout 2
the month?
1
10. Lake Level The graph in the accompanying figure 0 Year
shows the highest elevations above sea level attained 1991 1996 2001
by Lake Tahoe (located on the California–Nevada (Source: Based on data from Kelly Services, Inc., Annual
border) from 1982 through 1996. Reports, 1996–2001.)
a. Sketch a secant line connecting the beginning
a. Use the graph to estimate the average rate of
and ending points of the graph. Find the slope of
change in Kelly’s sales of services between 1996
this line.
and 2001. Interpret your answer.
b. Write a sentence interpreting the slope in the b. Estimate the percentage change in the service
context of Lake Tahoe levels. sales between 1996 and 2001.
c. Write a sentence summarizing how the level 12. Marriage Age The graph on page 109 shows the
of the lake changed from 1982 through 1996. median age at first marriage for men in the United
How well does your answer to part b describe the States.
change in the lake level as shown in the graph?
a. Estimate by how much and how rapidly the me-
11. Kelly Services A graph of the equation for a dian marriage age increased from 1980 through
model for the sales of services between 1991 and 1990.
Elevation
(feet above
sea level)
6230 6228.98 6229.03 6229.09
6228
6226 6224.88
6224
Natural rim (6223 feet above sea level)
No outflow below this elevation
6222
6221.87 Years
6220 since
82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 1900
Figure for Activity 10
(Source: Data from Federal Watermaster, U.S. Department of the Interior.)
2.1 Change, Percentage Change, and Average Rates of Change 109
Median age Week Total tickets sold
(years)
1 71
26.1
4 197
7 524
10 1253
23.2 13 2443
Year
1970 1980 1990 16 3660
(Source: Based on data from www.
19 4432
infoplease.com.)
22 4785
b. What is true about the average rate of change 25 4923
between any two points on a linear graph?
c. The median age at first marriage for men in
a. Find the first differences in the numbers of
2000 was 26.8 and in 2003 was 27.1. Did the
tickets sold, and convert them into average rates
median age of first marriage for men continue
of change.
to grow at the same rate during the beginning of
the twenty-first century as it did during the b. When were ticket sales growing most rapidly?
1980s? How rapidly (on average) were they growing at
that time?
13. Profit The table below gives the price, in dollars, of
c. If the travel agent made a $25 commission on
a round-trip flight from Denver to Chicago on a cer-
every ticket sold, how rapidly on average did the
tain airline and the corresponding monthly profit, in
agent’s commission revenue increase between
millions of dollars, for that airline on that route.
weeks 7 and 10?
15. Life Expectancy The life expectancies of black
Ticket price (dollars) Profit (thousands of dollars) males in the United States at various ages for 1998
200 3080 are as shown in the following table.
250 3520
300 3760
Life Life
350 3820
Age expectancy Age expectancy
400 3700 (years) (years) (years) (years)
450 3380 At birth 68.3 40 32.3
10 59.6 50 24.3
a. Find a model for the data. 20 50.0 60 17.5
b. Use the model to estimate the average rate of 30 41.1 70 11.8
change of profit when the ticket price rises from
$200 to $350. (Source: National Vital Statistics Reports, vol. 49, no. 12,
October 9, 2001.)
c. Use the model to estimate the average rate of
change of profit when the ticket price rises from
$350 to $450. a. How rapidly (on average) does the life expectancy
14. Cruise Tickets A travel agent vigorously promotes change between birth and the 70th year of life for
cruises to Alaska for several months. The number of black males in the United States?
cruise tickets sold during the first week and the total b. Compare the average rates of change of life
(cumulative) sales every 3 weeks thereafter are given expectancy for the 10-year periods between ages
in the table. 10 and 20 and ages 20 and 30.
110 CHAPTER 2 Describing Change: Rates
16. Loan The table shows the loan amount (the a. How many more cases were diagnosed in 1997
amount of money that a certain bank will lend) on than in 1995?
the basis of a monthly payment of $600 per month b. Find the percentage change and the average rate
with a 30-year term. of change in cases diagnosed between 1995 and
1997.
Monthly Loan amount (Source: Based on data from HIV/AIDS Surveillance Report,
APR (thousand dollars) vol. 13, no. 2.)
6% 100
19. Surcharges The percentage of all banks that,
8% 81.7 between 1996 and 2001, levied surcharges on the
10% 68.3 use of automated teller machines can be modeled by
12% 58.3 the equation
14% 50.6 95.98
s(t) percent
1 24,612e1.42t
a. What are the units for the change, the average t years after 1990. Find the percentage change in the
rate of change, and the percentage change in the percent of banks assessing surcharges on ATMs be-
loan amount when the interest rate changes? tween 1996 and 2001.
b. Determine the change, the average rate of change, (Source: Based on data from U.S. Public Interest Research
and the percentage change in the loan amount Group National Survey.)
when interest rates increase from 8% to 12%.
20. Refuse CPI The consumer price index (CPI)
c. Consider the inverse relationship represented by
values (1982 1984 100) for refuse collection
reversing the columns in the table. What are the
between 1990 and 2000 can be modeled by the
input units and the output units of the inverse
equation
relationship? What are the units for the change,
the average rate of change, and the percentage R(t) 133 54.5 ln t
change of the inverse relationship?
t years after 1988. Find the average rate of change in
d. Determine the change, the average rate of change, the CPI values for refuse collection between 1992
and the percentage change in the monthly APR and 2000.
when the loan amount increases from $58.3 thou- (Source: Based on data from Statistical Abstract, 1998.)
sand to $81.7 thousand.
21. Consider the linear function y 3x 4.
17. Population The population of Mexico between
a. Find the average rate of change of y over each of
1921 and 2000 is given by the model
the following intervals:
Population 7.6(1.026t) million persons
i. From x 1 to x 3
where t is the number of years since 1900. ii. From x 3 to x 6
(Source: Based on data from www.inegi.gob.mx,accessed 9/20/02.)
iii. From x 6 to x 10
a. How much did the population change from 1940
through 1955? Convert the change to percentage b. Find the percentage change in y for each of the
change. following intervals:
b. How rapidly was the population changing on i. From x 1 to x 3
average from 1983 through 1985? ii. From x 3 to x 5
18. AIDS The number of persons living with AIDS iii. From x 5 to x 7
from 1993 through 2001 can be modeled by c. On the basis of the results in part a and your
Cases diagnosed 1.51 2.15 ln x knowledge of linear functions from Chapter 1,
what generalizations can you make about per-
hundred thousand cases centage change and average rate of change for a
where x is the number of years since 1990. linear function?
2.1 Change, Percentage Change, and Average Rates of Change 111
22. Consider the exponential function y 3(0.4x). b. Use the information presented in the table to
a. Find the percentage change and average rate of find a model for the multiple-birth rate between
change of y for each of the following intervals: 1971 and 2000.
i. From x 1 to x 3 Year Multiple-birth rate (births per 100,000)
ii. From x 3 to x 5 1971 29.1
iii. From x 5 to x 7 1976 31.7
b. On the basis of the results in part a and your 1981 39.4
knowledge of exponential functions from Chap-
1986 49.9
ter 1, what generalizations can you make about
percentage change and average rate of change for 1991 80.2
an exponential function? 1996 152.6
23. Bank Account Imagine that 6 years ago you 2000 180.5
invested $1400 in an account with a fixed interest
(Sources: Greenville News, July 1, 1998, p. A1; National Vital
rate and with interest compounded continuously. Statistics Reports, vol. 50, no. 5, February 12, 2002.)
You do not remember the interest rate, but your
c. Use the equation to estimate the multiple-birth
end-of-the-year statements for the first 5 years yield
rates in 1995 and 1980. How close are those
the data shown in the table.
values to the results of part a? Are these estimates
found with interpolation or extrapolation?
End of year Amount at end of year
d. Suggest reasons why the multiple-birth rate has
1 $1489.55
been rising rapidly.
2 $1584.82
25. Finance Charge Your credit card statement indi-
3 $1686.19
cates a finance charge of 1.5% per month on the
4 $1794.04 outstanding balance.
5 $1908.80 a. What is the nominal rate (APR), assuming that
interest is compounded monthly?
a. Use the data to find the change and percentage
change in the balance from the end of year 1 b. What is the effective rate of interest (APY)?
through the end of year 5. 26. Payments In order to offset college expenses, at
b. Use the data to find the average rate of change of the beginning of your freshman year you obtained a
the balance from the end of year 1 through the nonsubsidized student loan for $15,000. Interest on
end of year 5. Interpret your answer. this loan accrues at a rate of 0.739% each month.
c. Using the data, is it possible to find the average However, you do not have to make any payments
rate of change in the balance from the middle of against either the principal or the interest until after
the fourth year through the end of the fourth you graduate.
year? Explain how this could be done or why it a. Write a function giving the total amount you will
cannot be done. owe after t years in college.
d. Find a model for the data, and use the equation b. What is the nominal rate?
to find the average rate of change in the balance c. What is the effective rate?
over the last half of the fourth year.
27. Doubling Time How long would it take an
24. Births The multiple-birth rate for births involv-
investment to double under each of the following
ing more than twins jumped 19.7% between 1995
conditions?
and 1996 and 312.4% between 1980 and 1996.
a. Interest is 6.3% compounded monthly.
a. If the birth rate for births involving three or more
babies was 152.6 per 100,000 births in 1996, find b. Interest is 8% compounded continuously.
the multiple-birth rates in 1995 and 1980. c. Interest is 6.85% compounded quarterly.
112 CHAPTER 2 Describing Change: Rates
28. Investment You have $1000 to invest, and you d. According to the table, what is the limit of the
have two options: 4.725% compounded semiannu- amount as n grows larger and larger?
ally or 4.675% compounded continuously. e. Write the limit found in part d in mathematical
a. Determine the better option by calculating the notation.
annual percentage yield for each.
30. Interest Suppose that you invest $1 for 1 year at
b. Verify your choice of option by comparing the
a rate of 100r% compounded n times during the
amount the first option would yield with the
year.
amount the second option would yield after 2, 5,
10, 15, 25, and 50 years. Does your choice of a. Use the compound interest formula to develop a
option depend on the number of years you leave function for the amount at the end of 1 year.
the money invested? (Hint: n and r will both appear as variables in the
formula.)
c. By how much would the two options differ after
10 years? b. Suppose r 0.1 and create a table that shows n
and the amount at the end of 1 year for the
29. Interest Suppose that you invest $1 for 1 year at a following compounding options: yearly, semian-
rate of 100% compounded n times during the year. nually, quarterly, monthly, weekly, daily, every
a. Use the compound interest formula to develop a hour, every minute, and every second. (Round
function for the amount at the end of 1 year the amount to four decimal places.)
given n, the number of compoundings. c. According to your table, what is the limit of the
b. Fill in the values of n on the following table. amount as the number of compoundings
becomes larger and larger?
Compounding n Amount
d. Use the information for this account and the
Yearly
continuous compounding interest formula to
Semiannually calculate the amount at the end of 1 year. How
Quarterly does your answer here compare to your estimate
Monthly
in part c?
Weekly e. Repeat parts b through d for r 0.5.
Daily
Every hour Discussing Concepts
Every minute 31. Explain how average rate of change, percent-
Every second age change, and change are related and how they
differ.
c. Use the formula from part a to fill in the Amount
column of the table. (Round values to two 32. Give a graphical interpretation of change and average
decimal places.) rate of change.
2.2 Instantaneous Rates of Change
In Section 2.1 we considered the average rate of change over an interval. Now we
consider the concept of the rate of change at a point. The most common example of
an instantaneous rate of change is as close as the nearest steering wheel. Suppose that
you begin driving north on highway I-81 at the Pennsylvania–New York border at
1:00 P.M. As you drive, you note the time at which you pass each of the indicated mile
markers (see Table 2.2).
2.2 Instantaneous Rates of Change 113
TABLE 2.2
Time Mile marker
1:00 P.M. 0
1:17 P.M. 19
1:39 P.M. 42
1:54 P.M. 56
2:03 P.M. 66
2:25 P.M. 80
2:45 P.M. 105
These data can be used to determine average rates of change. For example,
between mile 0 and mile 19, the average rate of change of distance is 67.1 mph. In this
context, the average rate of change is simply the average speed of the car. Average
speed between any of the mile markers in the table can be determined in a similar
manner. Average speed will not, however, answer the following question:
If the speed limit is 65 mph and a highway patrol officer with a radar gun
clocks your speed at mile post 17, were you exceeding the speed limit by more
than 10 mph?
The only way to answer this question is to know your speed at the instant that the
radar locked onto your car. This speed is the instantaneous rate of change of distance
with respect to time, and your car’s speedometer measures that speed in miles per
hour.
Just as an average rate of change measures the slope between two points, an
instantaneous rate of change measures the slope at a single point. Figure 2.7 shows a
continuous graph of air temperature as a function of time.
Temperature
(°F) Graph
peaks
80
B
Graph is C Graph is
rising falling
70
A
60
50
40
Time
FIGURE 2.7 8 A.M. 10 A.M. 12 P.M. 2 P.M. 4 P.M. 6 P.M.
The graph reaches its peak value at approximately 1:15 P.M. Reading from left to
right, the graph is rising until it reaches its peak at 1:15 P.M. and then is falling after
114 CHAPTER 2 Describing Change: Rates
1:15 P.M. When we read the graph from left to right, it appears that the graph is
slanted upward until we reach the peak and then slanted downward. We say that the
slope of the graph is positive at each point on the left side of the peak and is negative
at each point on the right side. The slope of the graph is zero at the top of the parabola.
Note that the graph levels off as you move from 7 A.M. to 1:15 P.M. It is not as steep at
1 P.M. as it is at 7 A.M., so the slope at 1 P.M. is smaller than the slope at
7 A.M. In fact, at each point on the graph there is a different slope, and we need to be
able to measure that slope in order to find the instantaneous rate of change at each
point.
Instantaneous Rate of Change
The instantaneous rate of change at a point on a curve is the slope of the
curve at that point.
In precalculus mathematics, the concept of slope is intrinsically linked with lines.
In terms of lines, slope is a measure of the tilt of a line. Now we wish to measure how
tilted any graph is at a point. We can consider the slope of a graph at a point as long
as the graph is continuous and not sharp at that point. Intuitively, a continuous curve
contains a sharp point when the pattern of the curve suddenly changes at that point.
We call a continuous function smooth over an interval if it has no sharp points in the
interval.
Local Linearity and Tangent Lines
For any smooth, continuous graph, we will eventually see a line as we look closer
and closer. For example, consider the temperature graph in Figure 2.7. With
the number of hours after 6 A.M. as input, close-ups of the graph 1/100 unit away
from each labeled point in both horizontal and vertical directions are as shown in
Figure 2.8.
2.2.1
65.81 80.28 77.9
A
B C
65.79 80.26 77.7
2.99 3.01 7.23 7.25 8.99 9.01
FIGURE 2.8
The graphs in Figure 2.8 illustrate the principle of local linearity. As we zoom
in on a point P on a smooth (no sharp points), continuous curve, the curve will
look more and more like a line. We call this line the tangent line, and its slope is
the instantaneous rate of change of the curve at the point P, the point of tangency.
A tangent line (from the Latin word tangere, “to touch”) at a point on a graph
touches that point and is tilted exactly the way the graph is tilted at the point of
tangency. The slope of the tangent line at a point is a measure of the slope of the
graph at that point.
2.2 Instantaneous Rates of Change 115
Local Linearity
If we look closely enough near any point on a smooth curve, the curve will
look like a line. The line is the tangent line at that point.
The tangent lines at points A, B, and C are shown in Figure 2.9. Do you see that
these are the same as the lines in Figure 2.8?
The slope of a graph at a point is the slope of the tangent line at that point.
In Figure 2.10, tangent lines are drawn on the temperature graph at 7 A.M., noon,
and 4 P.M. The tangent lines are tilted to match the tilt of the graph at each point. The
tangent lines at points D and E are tilted up, so the slope at these points is positive.
The slope at point F is negative, because the tangent line at point F is tilted down.
Even though the tangent line at point F has the least slope of these three tangents, it
is steeper than the tangent line at point E because the magnitude (absolute value) of
its slope is larger than that of the line tangent to the curve at point E. That is, the
temperature is falling more rapidly at 4 P.M. than it is rising at noon.
Temperature Temperature
(°F) (°F)
E
80 80
B F
C
70 70
A
60 60
50 50 D
40 40
Time Time
8 A.M. 10 A.M. 12 P.M. 2 P.M. 4 P.M. 6 P.M. 8 A.M. 10 A.M. 12 P.M. 2 P.M. 4 P.M. 6 P.M.
FIGURE 2.9 FIGURE 2.10
Examine Figure 2.10 carefully. The slope of the tangent line at point D is 10°F per
hour. (A method for calculating this slope will be discussed later.) Therefore, the
slope of the graph at 7 A.M. is also 10°F per hour. This is the same as saying that the
instantaneous rate of change of the temperature at 7 A.M. is 10°F per hour. In other
words, at 7 A.M., the temperature is rising 10°F per hour.
Similarly, the following statements can be made:
• The slope of the tangent line at point E is 2°F per hour.
• The slope of the graph at noon is 2°F per hour.
• The instantaneous rate of change of the temperature at noon is 2°F per hour.
• At noon, the temperature is rising 2°F per hour.
116 CHAPTER 2 Describing Change: Rates
And
• The slope of the tangent line at point F is 4.4°F per hour.
• The slope of the graph at 4 P.M. is 4.4°F per hour.
• The instantaneous rate of change of the temperature at 4 P.M. is 4.4°F per hour.
• At 4 P.M., the temperature is falling 4.4°F per hour.
We summarize the results of this discussion in the following way:
Given a function f and a point P on the graph of f, the instantaneous rate of
change at point P is the slope of the graph at P and is the slope of the line
tangent to the graph at P (provided the slope exists).
EXAMPLE 1 Comparing Slopes
NRA members NRA Membership Figure 2.11 shows the National Rifle Association membership
(millions) between 1990 and 1995. Consider the following statements:
3.6 • The slope of the line tangent to the graph at A is 0.24 million members per year.
E
3.4 • The slope of the graph is zero at point B.
3.2 D • The instantaneous rate of change of the NRA membership at point C is 340,000
3.0 members per year.
2.8 • NRA membership is increasing the most rapidly at point D. The greatest rate of
A C
2.6 Years
increase is 0.42 million members per year.
B
2.4 since • The slope of the line tangent to the graph at point E is 260,000 members per year.
0 1 2 3 4 5 1990
FIGURE 2.11 Using this information, answer the following questions:
a. At which of the indicated points is the slope of the graph (i) the greatest? (ii) the
least?
b. At which of the indicated points is the steepness of the graph (i) the greatest?
(ii) the least?
c. What is the instantaneous rate of change of the NRA membership at point A?
at point E?
d. What is the slope of the tangent line at point C? at point D?
Solution
a. The numerical values of the slopes, in million members per year, at the
indicated points are
A⬊0.24 B⬊0 C⬊0.34 D⬊0.42 E⬊0.26
The greatest value occurs at point D (the inflection point), and the least value
occurs where the only negative slope occurs, at point A.
2.2 Instantaneous Rates of Change 117
b. The steepness of the graph is a measure of how much the graph is tilted at a par-
ticular point. The direction of tilt is considered in the slope but is not consid-
ered when we are describing steepness. Thus the steepness at each of the indi-
cated points is
A⬊0.24 B⬊0 C⬊0.34 D⬊0.42 E⬊0.26
The graph is steepest at point D. The steepness is least at point B.
c. The instantaneous rate of change at A is 0.24 million, or 240,000, members
per year. At E the rate of change is 0.26 million, or 260,000, members per year.
d. The slope of the tangent line at C is 0.34 million members per year. The slope
at D is 0.42 million members per year. ●
Secant and Tangent Lines
In addition to understanding tangent lines in terms of local linearity, it is helpful to
understand the relationship between secant lines and tangent lines.
Recall that a tangent line at a point on a graph touches that point and is tilted
exactly the way the graph is tilted at that point. A secant line, on the other hand, is a
line that passes through two points on a graph. We illustrate the relationship between
secant lines and tangent lines with an example using points on a simple curve.
EXAMPLE 2 Drawing Secant and Tangent Lines
P7 P6 a. On the curve shown in Figure 2.12, draw secant lines through P1 and Q, P2 and
P8
Q
P9 Q, P3 and Q, and P4 and Q.
P4 b. Which of the four secant lines drawn in part a appears to be tilted most like the
P3
curve at Q?
P2
P1 c. Again, on the curve shown in Figure 2.12, draw secant lines through P6 and Q,
P7 and Q, P8 and Q, and P9 and Q. Which of these four secant lines appears to
match most closely the tilt of the curve at Q?
FIGURE 2.12 d. Where could you place a point P5 so that the secant line through P5 and Q
would be even closer than the secant lines in parts a and c to tilting the same
way the curve does at Q?
P7 P6
P8
P9
Q Q
P4
P3
P2
P1
FIGURE 2.13 FIGURE 2.14
118 CHAPTER 2 Describing Change: Rates
Solution
a. The four secant lines are shown in Figure 2.13.
b. The secant line through P4 and Q appears almost to hide the curve near Q. That
is, it appears to match the tilt of the curve at Q better than any of the other three
secant lines.
c. The four secant lines are shown in Figure 2.14. Again, we see that the secant line
through Q and the point closest to Q (P9 in this case) appears to match the tilt
of the curve at Q most closely.
d. The point P5 should be placed even closer to Q than P4 and P9 in order for the
tilt of the secant line to match the tilt of the curve at Q more closely. ●
We generalize the results of Example 2 by saying that the tilt of the secant line
through P and Q becomes closer and closer to the tilt of the curve at Q as P becomes
closer and closer to Q. Indeed, if you draw a secant line through the point Q and a
point P on the curve near Q, then the closer P is to Q, the more closely the secant line
approximates the tangent line at Q. You can think of the tangent line at Q as the
limiting position of the secant lines between P and Q as P gets closer and closer to Q.
Line Tangent to a Curve
The tangent line at a point Q on a smooth, continuous graph is the limiting
position of the secant lines between point Q and a point P as P approaches Q
along the graph (if the limiting position exists).
Sketching Tangent Lines
Although thinking of a tangent line as a limiting position of secant lines is vital to your
understanding of calculus (and is a subject to which we will return), it is important
for you to have an intuitive feel for tangent lines and to be able to sketch them without
first drawing secant lines. In drawing a line with the same tilt as a curve at a point, you
will find that in general, the line lies very close to the curve near the point but does
(a) If the curve is concave up
not cut through the curve.
at the point of tangency, then
the tangent line will lie below
the curve near the point of
tangency. General Rule for Tangent Lines
Lines tangent to a smooth nonlinear curve do not “cut through” the graph of
the curve at the point of tangency and lie completely on one side of the graph
near the point of tangency except at an inflection point.
It is important to note that for the graph of a linear function, the only way to draw
a line tangent to the graph at any point is to draw again the graph of the linear function.
(b) If the curve is concave Drawing a line tangent to a curve at an inflection point is dealt with after Example
down at the point of tangency, 3. For cases in which this exception does not apply, we can determine on which side
then the tangent line will lie of the curve the tangent line should lie by noting the concavity. If the curve is concave
above the curve near the point up at the point of tangency, then the tangent line will lie below the curve near the
of tangency. point of tangency. If the curve is concave down at the point of tangency, then the
FIGURE 2.15 tangent line will lie above the curve near the point of tangency. See Figure 2.15.
2.2 Instantaneous Rates of Change 119
EXAMPLE 3 Using a Tangent Line to Estimate Slope
Weight Loss A woman joins a national weight-loss program and begins to chart her
weight on a weekly basis. Figure 2.16 shows the graph of a continuous function of her
weight from the time when she began the program through 7 weeks into the program.
Weight
(pounds)
165
162
159
156
153
Time
150
0 1 2 3 4
(weeks)
5 6 7
FIGURE 2.16
a. Carefully sketch a line tangent to the curve at 5 weeks.
b. Estimate the slope of the tangent line at 5 weeks.
c. How quickly was the woman’s weight declining 5 weeks after the beginning of
the program?
d. What is the slope of the curve at 5 weeks?
Solution
a. Weight
(pounds)
165
162
159
156 Rise ≈
-5 pounds
153
Run = 4 weeks Time
150
0 1 2 3 4
(weeks)
5 6 7
FIGURE 2.17
rise 5 pounds
b. Slope run 4 weeks 1.25 pounds per week (See Figure 2.17.)
c. The woman’s weight was declining by approximately 1.25 pounds per week
after 5 weeks in the program.
d. The slope of the curve at 5 weeks is approximately 1.25 pounds per week. ●
120 CHAPTER 2 Describing Change: Rates
As we mentioned in the general rule for tangent lines, there are exceptions to the
principle that tangent lines do not cut through the graph and lie on only one side of
the graph. At a point of inflection, the graph is concave up on one side and concave
down on the other. As you might expect, the tangent line lies above the concave-down
portion of the graph and below the concave-up part. To do this, the tangent line must
cut through the graph. It does so at the point of inflection. When drawing tangent
lines at inflection points, be careful to make sure that the tangent line is tilted to
match the tilt of the graph at the point of tangency. See Figure 2.18.
FIGURE 2.18 (a) Although this line lies (b) This tangent line is
above the concave-down correctly drawn at an
portion and below the inflection point.
concave-up part of the
graph, it is not a tangent
line because it is not
tilted in the same way
that the graph is tilted at
the point.
Where Does the Instantaneous Rate of Change Exist?
Our discussion of tangent lines would not be complete without mention of piecewise
continuous functions. A piecewise continuous function is a function that is
continuous over different intervals but has a break point. It is often defined by
different equations over the different intervals. Consider Figure 2.19, which shows
the population of Indiana by official census from 1950 through 2000.
f (t)
Population
(millions)
6
4 t
Decades
5 6 7 8 9 10
since
1900
FIGURE 2.19
(Source: Based on data from World Almanac and Book
of Facts, ed. William A. McGeveran, Jr., New York:
World Almanac Education Group, Inc., 2003.)
2.2 Instantaneous Rates of Change 121
The equation of the graph in Figure 2.19 is
f(t) 0.129t2 2.2t 3.88 million people when 5 t 9
0.536t 0.72 million people when 9 t 10
where t is the number of decades after 1900.
Consider the tangent line at the point where the function f is not continuous
(1990). If we use the idea of a limiting position of secant lines, then we must conclude
that we cannot draw a line tangent to the graph at t 9 because the graph jumps at
this input.
In fact, if we zoom in close to the point at t 9, then we see something similar to
Figure 2.20. In this case, we do not have a smooth curve, and we see two lines. Again
we conclude that there is no tangent line at t 9. Does this mean that there is no in-
stantaneous rate of change in the population of Indiana in 1990? No, it simply means
that we cannot use our piecewise continuous model to calculate the rate of change in
1990. If it is necessary to determine the instantaneous rate of change at 1990 (the
break point), it would be advisable either to use a secant line estimate (an average rate
of change) or, if possible, to re-model the data with a function that would be
f (t) continuous and smooth at this point.
To help clarify the relationship between continuous and noncontinuous
5.56
functions and rates of change, consider the graphs shown in Figure 2.21. The graph
in Figure 2.21a is continuous everywhere and has a rate of change at every point. You
may think that the same is true of the graph in Figure 2.21b; however, because a
tangent line drawn at P is vertical, the slope of the tangent line is undefined. The
5.50 graph in Figure 2.21c is also continuous; however, the graph is not smooth at P be-
5.48 cause of the sharp point there. We call a point P on the graph of a continuous func-
tion a sharp point when secant lines joining P to close points on either side of it have
5.46 t different limiting positions. That is, we cannot draw a tangent line at P, because se-
8.975 9 9.025
cant lines drawn with points on the right and left do not approach the same slope.
FIGURE 2.20
P
P
(a) (b) (c) (d)
FIGURE 2.21
The graph in Figure 2.21d has a break at P and, therefore, is not continuous at
P. This graph is similar to that of the Indiana population graph, and the slope does
not exist at the break in the function. The slope does exist at all other points on the
graph.
The graph in Figure 2.21d illustrates a general rule relating continuity and rates
of change.
122 CHAPTER 2 Describing Change: Rates
Unless a function is continuous and smooth at a point, the instantaneous
rate of change does not exist at that point.
If you keep in mind the relationships among instantaneous rates of change, slopes
of tangent lines, slopes of secant lines, and local linearity, then you should have little dif-
ficulty determining the times when the instantaneous rate of change does not exist.
Recall from our discussion in Chapter 1 that calculus is applied to continuous
portions of functions. As we have just seen, instantaneous rates of change do not exist
where a function is not continuous. Modeling is a tool by which we transform data
into a continuous function. Discrete data are useful for finding change, percentage
change, and average rates of change, but a continuous or piecewise continuous
function is necessary in order to find instantaneous rates of change.
3. How are average rates of change and instantaneous
2.2 Concept Inventory rates of change measured graphically?
4. Explain in your own words how to tell visually
• Instantaneous rate of change
whether a line is tangent to a smooth graph.
• Local linearity
5. Using Table 2.2, the time/mileage table given in this
• Tangent line
section, verify that the average speed of the car from
• Slope of tangent line instantaneous rate of change mile marker 0 to mile marker 19 is 67.1 mph.
• Slope and steepness of a tangent line
6. Use Table 2.2 to answer the following questions.
• The tangent line is the limiting position of secant
a. What is the average speed (in mph) from:
lines
i. milepost 66 to milepost 80?
• Tangent lines lie beneath a concave-up graph
ii. milepost 80 to milepost 105?
• Tangent lines lie above a concave-down graph
b. What might account for the difference in speed?
• Situations in which the instantaneous rate of
change does not exist 7. Use the accompanying graph to answer the
following questions.
2.2 Activities
A
Getting Started
B D
1. In your own words, describe the difference between
a. discrete and continuous.
b. average rate of change and instantaneous rate of C
E
change.
c. secant lines and tangent lines. a. At each labeled point on the graph, determine
2. What are some advantages of using a continuous whether the instantaneous rate of change is
model instead of discrete data? What are some positive, negative, or zero.
disadvantages? b. Is the graph steeper at point A or at point B?
2.2 Instantaneous Rates of Change 123
8. At each labeled point on the graph, determine 13.
whether the slope is positive, negative, or zero.
C
E
B D
14.
A
9. Consider the accompanying graph.
15. Which of the lines drawn on the graph are not
tangent lines?
Estimate where the output is falling most rapidly. Mark
that point on the graph. A
C
10. Consider the accompanying graph.
B
16. Which of the lines drawn on the graph are not
tangent lines?
y
Estimate where the slope is greatest. Mark that point on
C
the graph.
For the figures in Activities 11 through 14, discuss the
slopes of the graphs.
11. 12. B
D
A
x
124 CHAPTER 2 Describing Change: Rates
17. On the graph, draw secant lines through P1 and Q, 20. a. Is the accompanying graph concave up, concave
P2 and Q, and P3 and Q. Repeat for the points P4 down, or neither at A, B, C, and D?
and Q, P5 and Q, and P6 and Q. Then draw the
C
tangent line at Q.
Q P P5
6 P4
P3
P2 B
D
P1
A
b. Should the tangent lines lie above or below the
curve at each of the indicated points?
18. Draw secant lines through P1 and Q, P2 and Q, and c. Carefully draw tangent lines at the labeled
P3 and Q on the graph. Repeat for the points P4 and points.
Q, P5 and Q, and P6 and Q. Then draw the tangent d. At which of the labeled points is the slope of the
line at Q. curve positive? At which of the labeled points is
the slope of the curve negative? Do any of the
P4 labeled points appear to be inflection points?
P5
Use carefully drawn tangent lines to estimate the slopes
P6 at the labeled points in Activities 21 through 24.
Q
P3
21. 5
P2
P1
4
3
19. a. Is the graph shown concave up, concave down, B
or neither (an inflection point) at A, B, C, and D? 2
1
A
B D 10 20 30 40 50 60 70
22. 10
C
8 A
6
b. Should the tangent lines lie above or below the
curve at each of the indicated points?
4
c. Carefully draw tangent lines at the labeled points
on the figure. 2
d. At which of the labeled points is the slope of the
tangent line positive? At which of the labeled
points is the slope of the tangent line negative? 1 2 3 4 5 6 7
2.2 Instantaneous Rates of Change 125
23. 600 26. Phone Bill The graph shows the average monthly
D cellular phone bill since 1987. The slope of the curve
500 at point A is 0.23.
400
Average bill
300 (dollars)
97
200
100 C
A
20 40 60 80 100 120 140
30 Year
24. 1987 1990 1993 1996 1999 2001
3.0
(Source: Based on data from the Cellular Telecommunications
2.5 and Internet Association.)
2.0 a. What should be the units on the slope at point A?
E b. How rapidly was the dollar amount growing in
1.5
1993?
1.0 c. What is the slope of the tangent line at point A?
0.5 d. What is the instantaneous rate of change of the
amount at point A?
1 2 3 4 5 6 7
27. Growth Rate The growth of a pea seedling as a
function of time can be modeled by two quadratic
Applying Concepts functions as shown. The slopes at the labeled points
25. Subscribers The graph shows the total number of are (in ascending order) 4.2, 1.3, and 5.9.
cellular phone subscribers from 1996 to 2001. The Growth rate
slope of the graph at point A is 23.1. (mm/day)
Cellular phone subscribers 54
(millions)
132.2 45
36
A
27
B C
18
44 Year 9 A
1996 1998 2000 Temperature
0
14 17 20 23 26 29 32 35 38
(°C)
(Source: Based on data from the Cellular Telecommunications
and Internet Association.)
(Source: Based on data in George L. Clarke, Elements of Ecology,
a. What are the units on the slope at point A? New York: Wiley, 1954.)
b. How rapidly was the number of subscribers a. Match the slopes with the points, A, B, and C.
growing in 2000? b. What are the units on the slopes for each of these
c. What is the slope of the tangent line at A? points?
d. What was the instantaneous rate of change of the c. How quickly is the growth rate changing with re-
number of cell phone subscribers in 2000? spect to temperature at 23°C?
126 CHAPTER 2 Describing Change: Rates
d. What is the slope of the tangent line at 32°C? f. At 30% relative humidity, the survival rates for
e. What is the instantaneous rate of change of the and are changing at
growth rate of pea seedlings at 17°C? approximately the same rate.
g. At 65% relative humidity, the survival curves for
28. Beetles The graph shows the survival rate (per-
and have approxi-
centage surviving) of three stages in the develop-
mately the same slope.
ment of a flour beetle (egg, pupa, and larva) as a
function of the relative humidity. 29. Sun Declination The figure shows a graph of the
declination of the sun (the angle of the sun from the
equator) throughout the year.
100
Larvae Declination of sun
90 Pupae Eggs
24
Degrees north
80 18
70 12
Survival rate ( % )
6
60
Days
32 60 91 121 152 181 213 244 274 305 335 365
50
Degrees south
(Jan.1) (Mar.1) (May1) (July1) (Sept.1) (Nov.1)
40 12
18
30
24
20
(Source: The Mathematics Teacher, March 1997, p. 238.)
10 a. A solstice is a time when the angle of the sun from
0 the equator is greatest. Locate the summer and
0 10 20 30 40 50 60 70 80 90 100 winter solstices on the graph. What is the slope of
Relative humidity ( % ) the graph at these points?
(Source: R.N. Chapman, Animal Ecology, New York: McGraw- b. Locate the two steepest points on the graph.
Hill, 1931.) Estimate the slopes of the tangent lines at these
two points. What is the significance of a negative
slope in this context?
In parts a through g, fill in each of the following
30. Grasshoppers The effects of temperature on
blanks with the appropriate stage (eggs, pupae, or
the percentage of grasshoppers’ eggs from West
larvae).
Australia that hatch are shown in the graph.
a. At 60% relative humidity, the instantaneous rate
of change of the survival rate of is 100
approximately zero.
80
Emergence (%)
b. An increase in relative humidity improves the
survival rate of and reduces the 60
survival rate of .
40
c. At 97% relative humidity, the survival rate of
is declining faster than that of 20
. 0
d. Any tangent lines drawn on the survival curve for 5 10 15 20 25 30
will have negative slope. Temperature (°C)
e. Any tangent lines drawn on the survival curve for (Source: Figure adapted from George L. Clarke, Elements of
will have positive slope. Ecology, New York: Wiley, 1954.)
2.2 Instantaneous Rates of Change 127
a. What is the optimum hatching temperature? Statistical Abstract for 2001, can be approximated by
b. What is the slope of the tangent line at the the model
optimum temperature? P(x) 2.56x 274.72 million people
c. Sketch tangent lines at 10°C, 17°C, and 22°C, and
where x is the number of years since 2000.
estimate the slopes at these points.
a. Compare the model in this activity with the one
d. Where does the inflection point appear to be on
in Activity 31. With the additional information
this graph?
about population available in Statistical Abstract
31. Population Predictions for the U.S. resident for 2001, were the population projections
population from 1997 through 2050, as reported by adjusted up or down?
Statistical Abstract for 1994, can be approximated by b. Was the growth rate adjusted up or down?
the model
c. Find the slope of the graph of P at t 20.
p(t) 2.37t 39.79 million people d. Describe the tangent line at t 20.
where t is the number of years since 1900. A graph e. How rapidly does the model predict the popula-
of p is shown. tion will be changing in 2020?
33. Employees The number of Houghton Mifflin
p(t) Company employees from 1993 through 2000 can
People be modeled by the accompanying graph.
(millions)
405 Employees
385 3500
365 3000
345
2500
325
305 2000 Years
285 t 1650 since
Years 4 6 8 10 1990
265
95 100 110 130 140 150
since
120 (Source: Hoover’s Online Company Capsules.)
1900
Draw tangent lines, if possible, to estimate how
quickly the number of employees was changing in
the indicated years. If it is impossible to do so,
a. Sketch a tangent line at t 120, and find its
explain why.
slope.
a. 1994
b. What is true about any line tangent to the graph
of the function p? b. 1995
c. What is the slope of any line tangent to this c. 1998
graph? 34. Employment The graph on page 128 shows em-
d. What is the slope at every point on the graph of ployment in Slovakia from 1948 through 1988.
this model? a. Estimate how rapidly employment in agriculture
e. According to the model, what is the instanta- and forestry was declining in 1958.
neous rate of change of the predicted population b. Estimate the instantaneous rate of change in
in any year from 1997 through 2050? industry employment in 1962.
32. Population Predictions for the U.S. resident c. Why is it not possible to sketch a tangent line to
population from 2001 through 2050, as reported by the industry graph at 1974?
128 CHAPTER 2 Describing Change: Rates
Industry Trade
Construction Science and research, etc.
1000 Agriculture and forestry Banking, finance, other
900 Transport and communications
800
700
600
Employment (thousands)
500
400
300
200
100
1986
1984
1959
1967
1971
1973
1977
1979
1988
1957
1969
1975
1953
1955
1961
1965
1948
1963
Figure for Activity 34
(Source: Figure from A. Smith, “From Convergence to Fragmentation,” Environment and
Planning, vol. 28, 1996. Pion Limited, London. Reprinted by permission. Data elaborated from
Statisticka rŏ centa SUSR, various dates.)
Discussing Concepts line at a break point for such a function? Discuss how
and why this might or might not happen. You may
35. Explain, using your own words, the relationship find it helpful to use these two functions as examples:
between secant lines and tangent lines.
36. Most piecewise continuous models have discontinu- f(x)
x2 8
when x 2
x3 9x 14 when x 2
ities at their break points. Consider, however, piece-
wise continuous functions that are continuous at x3 9 when x 3
g(x)
their break points. Is it possible to draw a tangent 5x2 3x when x 3
2.3 Derivative Notation and Numerical Estimates
Derivative Terminology and Notation
By now, you should be comfortable with the concepts of average rate of change and
instantaneous rate of change. Let’s summarize the differences between these two rates
of change.
2.3 Derivative Notation and Numerical Estimates 129
Average Rates of Change Instantaneous Rates of Change
• measure how rapidly (on • measure how rapidly a quan-
average) a quantity changes tity is changing at a point
over an interval
• can be obtained by calculating • can be obtained by calculating
the slope of the secant line the slope of the tangent line at
between two points a single point
• require data points or a • require a continuous, smooth
continuous curve to calculate curve to calculate
Because instantaneous rates of change are so important in calculus, we commonly
refer to them simply as rates of change. The calculus term for instantaneous rate of
change is derivative. It is important to understand that the following phrases are
equivalent.
Equivalent Terminology
All of the following phrases have the same meaning:
• instantaneous rate of change
• rate of change
• slope of the curve
• slope of the tangent line
• derivative
Even though we consider all these phrases synonymous, we must keep in mind
that the last three phrases have specific mathematical definitions and so may not exist
at a point on a function. However, the rate of change of the underlying situation does
have an interpretation at that point in context. In such cases, we have to estimate the
rate of change by using some other estimation technique.
There are also several symbolic notations that are commonly used to represent
the rate of change of a continuous function f with input t. In this book, we use three
different, but equivalent, symbolic notations:
Equivalent Notation
df This is read, “d-f-d-t,” “the rate of change of f with respect to t,” or
dt “the derivative of f with respect to t.”
(or)
f(t) This is read, “f prime of t,” or “the rate of change of f with respect to
t,” or “the derivative of f with respect to t.”
(or)
d This is read, “d-d-t of f of t,” “the rate of change of f with respect to
[f(t)]
dt t,” or “the derivative of f with respect to t.”
130 CHAPTER 2 Describing Change: Rates
Suppose that G(t) is your grade out of 100 points on the next calculus test when
you study t hours during the week before the test. The graph of G may look like that
shown in Figure 2.22.
Note how the grade changes as your studying time increases. The grade slowly
improves during the first 2 hours. The longer you study, the more rapidly the grade
improves until you have studied for approximately 7 hours. After 7 hours, the grade
improves at a slower rate. Your grade peaks after 14 hours of studying and then ac-
tually declines. What might explain the decline?
Let us compare the rates at which your grade is increasing when t 1 hour and
when t 4 hours of study. Tangent lines at t 1 and t 4 are shown in Figure
2.23. After 1 hour of studying, your grade is increasing at a rate of approximately
1.7 points per hour. (We will show you how to calculate, not estimate, this rate in
a later section.) This value is the slope of the curve when t 1 hour. After 4 hours
of studying, your grade is increasing at a rate of approximately 5.2 points per hour.
Can you see that the graph is steeper when t 4 than when t 1? The grade is im-
proving more rapidly after 4 hours of studying than it is after 1 hour. In other
words, a small amount of additional study is more beneficial when you have already
studied for 4 hours than it is when you have studied for only 1 hour.
G(t)
G(t)
Points
Points
100
100
75 75
Slope ≈ 5.2
50 50
25 Slope ≈ 1.7
t t
0
2 4 6 8 10 12 14 16 Hours 2 4 6 8 10 12 14 16 Hours
FIGURE 2.22 FIGURE 2.23 Tangent lines at t 1 and t 4
These rates can be summarized with the following notation:
dG
1.7 points per hour when t 1 hour, or
dt
G(1) 1.7 points per hour, and
dG
5.2 points per hour when t 4 hours, or
dt
G(4) 5.2 points per hour.
Interpreting Derivatives
As we have already mentioned, mathematical results are not very useful in real-world
settings unless they are stated in a form that anyone can understand. For this reason,
an interpretation of a result should be stated using a simple, nontechnical sentence.
As in the case of interpreting descriptions of change in Section 2.1, you should answer
the questions when, what, how, and by how much when interpreting a rate of change.
2.3 Derivative Notation and Numerical Estimates 131
Again, consider your score G(t) out of 100 points on the next calculus exam as a
function of the number of hours t that you have studied for the exam. Which of the
dG
following is a valid interpretation of dt 5.2 points per hour when t 4 hours?
a. The rate of change of my grade after 4 hours is 5.2 points per hour.
b. The slope of the line tangent to the grade curve at t 4 is 5.2.
c. My grade increased 5.2 points after I studied 4 more hours.
d. When I have studied for 4 hours, my grade is improving by 5.2 points per hour.
Choice a only restates the mathematical symbols in words. It does not give the
meaning of the derivative in the real-life context. Choice b is also a correct statement,
but it uses technical words that a person who has not studied calculus probably would
not understand. Also, the symbol t is used with no meaning attached to it, and units
are not included with the value 5.2. The use of the word increased in choice c refers to
an interval of time, not to change at a point in time. It is an incorrect statement.
Choice d is the only valid interpretation.
You should note that because a rate of change is measured at a point, it describes
something that is in the process of changing. Therefore, we must use the progressive
tense (verbs ending with -ing) to refer to rates of change. For example, we say that
“after 1 hour of studying, your grade is increasing by approximately 1.7 points per
hour.” It is incorrect to say that your grade increased or increases at a specific point.
These verbs refer to change over an interval rather than at a point.
EXAMPLE 1 Interpreting Derivatives
Study Time Interpret the following four mathematical statements in the context
of studying time and grades according to the function G whose graph is shown in
Figure 2.22.
dG
a. dt 6.4 points per hour when t 7 hours
b. G(12) 3.0 points per hour
c. The derivative of G with respect to t is 0 points per hour when t 14 hours.
d. The slope of the tangent line when t 15 hours is approximately 2 points
per hour.
Solution
a. The first statement says that when you have studied for 7 hours, your grade is
improving by 6.4 points per hour. As we later learn, this is the point of greatest
slope—that is, the time when a small amount of additional study will benefit
you the most.
b. The second statement says that after 12 hours of studying, your grade is
improving by 3.0 points per hour. Does this mean that at 12 hours of studying,
your grade is less than at 7 hours of studying? No! It simply means that a small
amount of additional study time beyond 12 hours may not result in as many
extra points on your test as the same amount of time produces after you have
studied for only 7 hours.
132 CHAPTER 2 Describing Change: Rates
c. The third statement says that after you have studied for 14 hours, your grade
will no longer be improving. A glance back at Figure 2.23 shows that you have
reached your best possible score; more study will not improve your grade.
d. The fourth statement tells you that after 15 hours of studying, your grade is actu-
ally declining by 2 points per hour. Additional study will only hurt your grade.
Make sure that you understand that these statements tell you nothing about what
your grade is—they tell you only how quickly your grade is changing. ●
EXAMPLE 2 Sketching Function Graphs Using Derivative Information
Medicine C(h) is the average concentration (in nanograms per milliliter, ng/mL) of
a drug in the blood stream h hours after the administration of a dose of 360 mg. On
the basis of the following information, sketch a graph of C.
C(0) 124 ng/mL C(0) 0 ng/mL per hour
C(4) 252 ng/mL C(4) 48 ng/mL per hour
The concentration of the drug is increasing most rapidly after 4 hours. The maximum
concentration, 380 ng/mL, occurs after 8 hours. Between h 8 and h 24, the con-
centration declines at a constant rate of 14 ng/mL per hour. The concentration after 24
hours is 35.9 ng/mL higher than it was when the dose was administered.
Solution
The information about C(h) at various values of h simply locates points on the graph
of C. Plot the points (0, 124), (4, 252), (8, 380), and (24, 159.9).
Because C(0) 0, the curve has a horizontal tangent at (0, 124). The point of
most rapid increase, (4, 252), is an inflection point. The graph is concave up to the
left of that point and concave down to the right. The maximum concentration occurs
after 8 hours, so the highest point on the graph of C is (8,380). Concentration
declining at a constant rate between h 8 and h 24 means that over that interval,
C is a line with slope 14.
One possible graph is shown in Figure 2.24. Compare each statement about C(h)
to the graph.
C(h)
Drug concentration
(ng/mL)
400
300
200
100
h
FIGURE 2.24 0 4 8 12 16 20 24 Hours
Note in Figure 2.24 that we cannot assign a value to C(8). However, the
maximum concentration occurs at 8 hours, and on the basis of that, we can estimate
that the rate of change of the concentration at that time is zero. ●
2.3 Derivative Notation and Numerical Estimates 133
Does Instantaneous Refer to Time?
We saw that the instantaneous rate of change at a point P on the graph of a contin-
uous function is the slope of the tangent line to the graph at P. If the function in-
puts are measured in units of time, then it is certainly natural to use the word
instantaneous when describing rates of change, because each point P on the graph
of the function corresponds to a particular instant in time. In fact, the use of the
word instantaneous in this context arose precisely from the historical need to un-
derstand how rapidly distance traveled changes as a function of time. Today, we are
accustomed to referring to the rate of change of distance traveled as a function of
time as speed.
You should be aware, however, that the use of the word instantaneous in
connection with rates of change does not necessarily mean that time units are in-
volved. For example, suppose that a graph depicts profit (in dollars) resulting from
the sale of a certain number of used cars. In this case, the slope of the tangent line
at any particular point (the instantaneous rate of change) expresses how rapidly
profit is changing per car. The unit of change is dollars per car; no time units are
involved.
You should also remember that units for instantaneous rates of change, like
average rates of change, are always expressed in output units per input unit. Without
proper units, a number that purports to describe a rate of change is meaningless.
EXAMPLE 3 Writing Derivative Notation and Slope Units
Temperature The graph in Figure 2.25* shows the temperature, T(k), of the polar
night region (in °C) as a function of k, the number of kilometers above sea level.
T(k)
(°C)
25 30 35 40 45 50 55 60 k
-20
(km)
-30
-40
-50
-60
-70
-80
FIGURE 2.25
a. Sketch the tangent line at 45 km and estimate its slope.
b. What is the derivative notation for the slope of a line tangent to the graph of T?
c. Write a sentence interpreting in context the meaning of the slope found in part a.
*“Atmospheric Exchange Processes and the Ozone Problem,” in The Ozone Layer, ed. Asit K. Biswas,
Institute for Environmental Studies, Toronto. Published for the United Nations Environment Program by
Pergamon Press, Oxford, 1979.
134 CHAPTER 2 Describing Change: Rates
T(k) Solution
(°C) a. From Figure 2.26 we calculate
25 30 35 40 45 50 55 60 k
-20 rise 23°C
-30
Run ≈10km (km) Slope 2.3°C per kilometer
Rise ≈ run 10 km
-40 23°C
-50 dT d
b. Correct derivative notations include dk , dk [T(k)], and T(k).
-60
-70 c. At 45 km above sea level, the temperature of the atmosphere is increasing 2.3°C
-80 per kilometer. In other words, the temperature rises by approximately 2.3°C
between 45 and 46 km above sea level. ●
FIGURE 2.26
Percentage Rate of Change
Recall from Section 2.1 that percentage change is found by dividing change over an
interval by the output at the beginning of the interval and multiplying by 100.
Similarly, percentage rate of change can be found by dividing the rate of change at a
point by the function value at the same point and multiplying by 100. The units of a
percentage rate of change are percent per input unit.
Percentage Rate of Change
rate of change at a point
Percentage rate of change 100%
value of the function at that point
Percentage rates of change are useful in describing the relative magnitude of a rate
of change. For example, suppose you are a city planner and estimate that the city’s
population is increasing at a rate of 50,000 people per year. Any growth in population
will affect your planning activities, but just how significant is a growth rate of 50,000
people per year? If the current population is 200,000 people, then the percentage rate
50,000 people per year
of change of the population is 200,000 people 100% 25% per year. Growth of
25% per year in population is fast growth. However, if the current population is 2
million, then the percentage rate of change of the population is
50,000 people per year
2,000,000 people 100% 2.5% per year. The steps that you, as a city planner,
must take to accommodate growth if the city is growing by 25% per year are different
from the steps you must take if the city is growing by 2.5% per year. Expressing a rate
of change as a percentage puts the rate in the context of the current size and adds
more meaning to the interpretation of the rate of change.
EXAMPLE 4 Graphically Estimating Percentage Rate of Change
Sales The graph in Figure 2.27 shows sales (in thousands of dollars) for a small
business from 1995 through 2003.
2.3 Derivative Notation and Numerical Estimates 135
Sales
(thousands
of dollars)
14
12
10
8
6
4 Year
FIGURE 2.27 1995 1997 1999 2001 2003
a. Estimate the rate of change of sales in 1999 and interpret the result.
b. Estimate and interpret the percentage rate of change of sales in 1999.
Solution
a. A tangent line is drawn at 1999, as shown in Figure 2.28.
Sales
(thousands
of dollars)
14
12
10
8
6 Rise ≈ -2.8
Run ≈ 2
4 Year
FIGURE 2.28 1995 1997 1999 2001 2003
The slope is estimated to be
$2.8 thousand
$1.4 thousand per year $1400 per year
2 years
In 1999, sales were falling at a rate of approximately $1400 per year.
b. We can express this rate of change as a percentage rate of change if we divide it
by the sales in 1999. It appears from the graph that the sales in 1999 were
approximately $7.5 thousand dollars, or $7500. Therefore, the percentage rate
of change in 1999 was approximately
$1400 per year
100% 0.187 100% per year 18.7% per year
$7500
In 1999, sales were falling by approximately 18.7% per year. Expressing the rate
of change of sales as a percentage of sales gives a much clearer picture of the
impact of the decline in sales. The business was experiencing a reduction in
sales of nearly 20% per year in 1999. ●
136 CHAPTER 2 Describing Change: Rates
Finding Slopes by the Numerical Method
By now you should have a firm graphical understanding of rates of change. Although
graphical approximations are often sufficient, there are times when we need to find a
more precise answer.
Consider the relatively simple problem of finding the slope of the graph of
f(x) 2x at x 1. Part of the graph of f(x) 2x is shown in Figure 2.29a. Take
a few moments to sketch carefully a line tangent to the graph at x 1 and estimate its
slope. You should find that the tangent line at x 1 has a slope of approximately 1. See
Figure 2.29b.
f (x) f (x)
5 5
Run ≈ 3
4 4
Rise = 3
3 3
Slope ≈ 3 = 1
3
2 2
1 1
x x
1 2 3 4 5 1 2 3 4 5
f (x) (a) (b)
5 FIGURE 2.29
4
3 Another, more precise method of estimating the slope of the graph of
(1, 2) (1.5, 2√1.5) f(x) 2x at x 1 uses a technique illustrated in Section 2.2. Recall that the
2
tangent line at a point is the limiting position of secant lines through the point of tan-
1 gency and other increasingly close points. In other words, the slope of the tangent line
is the limiting value of the slopes of secant lines drawn through the point of tangency.
x
1 2 3 4 5 To illustrate, we begin by finding the slope of the secant line on the graph of
f(x) 2x through x 1 and x 1.5. (Note that x 1.5 is an arbitrarily chosen
A secant line through x 1
and x 1.5
value that is close to x 1.) A graph of the secant line is shown in Figure 2.30. Its
FIGURE 2.30 slope is calculated as follows:
Point at x 1: (1, 21) (1, 2)
f (x) Point at x 1.5: (1.5, 21.5) (1.5, 2.449489743)
5 2.449489743 2
Slope 0.8989794856
4 1.5 1
3 This value is an approximation to the slope of the tangent line at x 1. To obtain
(1, 2) (1.1, 2√1.1) a better approximation, we must choose a point closer to x 1 than x 1.5, say
2
x 1.1. (This is also an arbitrary choice.) See Figure 2.31.
1
Point at x 1: (1, 21) (1, 2)
x
1 2 3 4 5 Point at x 1.1: (1.1, 21.1) (1.1, 2.097617696)
A secant line between x 1 2.097617696 2
Slope 0.9761769634
and x 1.1 1.1 1
FIGURE 2.31
2.3 Derivative Notation and Numerical Estimates 137
This is a better approximation to the slope of the tangent line at x 1 than the
value from the previous calculation. To get an even better approximation, we need
only choose a closer point, such as x 1.01.
Point at x 1: (1, 21) (1, 2)
Point at x 1.01: (1.01, 21.01) (1.01, 2.009975124)
2.009975124 2
Slope 0.9975124224
1.01 1
We also use x 1.001.
Point at x 1: (1, 21) (1, 2)
Point at x 1.001: (1.001, 21.001) (1.001, 2.00099975)
2.00099975 2
Slope 0.9997501248
1.001 1
As we choose points increasingly close to x 1, what do you observe about the slopes
of the secant lines shown in Figure 2.32?
5
97
0.9 0.99
51
97
e≈ ≈
f (x) 8
op lope
61 8
97 89
0.
S
9
e≈ ≈ 0.8
Sl
op pe
Sl Slo
1
e=
lop
es
lin
ent
g
Tan
x
FIGURE 2.32 1
x 1.5 Secant line slope 0.8989794856
x 1.1 Secant line slope 0.9761769634
x 1.01 Secant line slope 0.9975124224
x 1.001 Secant line slope 0.9997501248
The pattern in the slope values continues as we choose closer points: for x 1.0001,
the slope to six decimal places is 0.999975, for x 1.00001, the slope to seven decimal
places is 0.9999975, and so on. Thus the slopes of the secant lines when we use points
to the right of x 1 appear to be approaching 1. You may have noticed by now that
we are numerically estimating the limit of the slopes of secant lines. The fact that the
limit of the slopes of the secant lines that approach the point from the left and the limit
of the slopes of the secant lines that approach the point from the right is important in
138 CHAPTER 2 Describing Change: Rates
the context of finding the slope at a point. A limit exists only if the limit from the
left and the limit from the right are equal. For this reason, in order to conclude that
the slope of the tangent line at x 1 is 1, we must also consider the limit of the slopes
of secant lines using points to the left of x 1. Choosing the x-values 0.5, 0.9, 0.99,
and 0.999, we obtain the following slopes of secant lines between these x-values and
x 1:
x 0.5 Secant line slope 1.171572875
x 0.9 Secant line slope 1.026334039
x 0.99 Secant line slope 1.002512579
x 0.999 Secant line slope 1.000250125
Again, note the pattern in the slope values: for x 0.9999, the slope to six decimal
places is 1.000025. For x 0.99999, the slope to seven decimal places is 1.0000025.
Thus the slopes of secant lines using points to the left of x 1 appear to be ap-
proaching 1. Because the limit of slopes using points to the left of x 1 appears to be
the same as the limit using points to the right of x 1, we estimate that the slope of
the line tangent to the graph of f(x) 2x at x 1 is 1. We use algebraic methods
in the next section to verify that f(1) 1.
In this case, the graphical and numerical methods for estimating the slope of the
tangent line yield similar results. However, Example 5 shows that calculating the
slopes of nearby secant lines generally yields a much more precise result than
sketching a tangent line and estimating its slope.
EXAMPLE 5 Numerically Estimating a Rate of Change
Investment A multinational corporation invests $32 billion of its assets electroni-
cally in the global market, resulting in an investment with continuous compounding
at 12% APY.
a. How rapidly is the investment growing at the beginning of the fifth year?
b. At what percentage rate of change is this investment growing?
Solution
a. First, note that this is a compound interest function with accumulated amount
given by
Value
(billion dollars) f(t) 32(1.12t) billion dollars
80
after t years. A graph of this function is shown in Figure 2.33.
70
One method of approximating the slope of the tangent line is to sketch the tan-
60 gent line and estimate its slope. On the graph in Figure 2.33, draw the tangent line
at t 4, and estimate the slope. If you accurately sketched the tangent line and
50
were careful in reading two points off that line, you should have found that the
40 slope of the tangent line is some value between 5 and 6 billion dollars per year.
30 Years To obtain a more precise estimate of the slope of the tangent line, we calculate
0 2 4 6 8
slopes of nearby secant lines. We choose increasingly close points to both the right
Figure 2.33 and the left of the point where t 4: (4, 50.3526195). We calculate slopes until
2.3 Derivative Notation and Numerical Estimates 139
they remain constant to one decimal place beyond the desired accuracy for two
or three calculations. You should verify each of the computations in Table 2.3.
TABLE 2.3
Points to the left Points to the right
Point at t 3.99 :(3.99, 50.2955879) Point at t 4.001 :(4.001, 50.3583262)
50.2955879 50.3526195 50.3583262 50.3526195
Slope Slope
3.99 4 4.001 4
Slope 5.70316 Slope 5.70672
Point at t 3.999 :(3.999, 50.3469135) Point at t 4.0001 :(4.0001, 50.3531902)
50.3469135 50.3526195 50.3531902 50.3526195
Slope Slope
3.999 4 4.0001 4
Slope 5.70607 Slope 5.70642
Point at t 3.9999 :(3.9999, 50.3520489) Point at t 4.00001 :(4.00001, 50.3526766)
50.3520489 50.3526195 50.3526766 50.3526195
Slope Slope
3.9999 4 4.00001 4
Slope 5.70636 Slope 5.70640
Point at t 3.99999 :(3.99999, 50.3525625) Point at t 4.000001 :(4.000001, 50.3526252)
50.3525625 50.3526195 50.3526252 50.3526195
Slope Slope
3.99999 4 4.000001 4
Slope 5.70639 Slope 5.70640
Whether points to the left or right of t 4 are chosen, it seems clear that the
slopes are approaching approximately 5.706. That is, the limit of slopes of
secant lines using points to the left of t 4 is approximately 5.706, and the
limit of slopes of secant lines using points to the right of t 4 is approximately
5.706. Thus we conclude that the slope of the line tangent to the graph at t 4
is approximately 5.706 billion dollars per year.
Note that this numerical method of calculating slopes of nearby secant lines in
order to estimate the slope of the tangent line at t 4 gives a much more
precise answer than graphically estimating the slope of the tangent line. (Be
certain that you keep all decimal places in your calculations and enough
decimal places in your recorded slope values to be able to see the limit.) We
conclude that at the beginning of the fifth year, the accumulated amount of the
investment is growing by 5.706 billion dollars per year.
b. The percentage rate of change at the beginning of the fifth year is
5.706 billion dollars/year
100% 11.33% per year. ●
50.3526 billion dollars
Recall that for compound interest functions, the annual percentage change is constant
and is the APY. In Example 5, we found the constant percentage rate of change of a con-
tinuous compounding. By rewriting this investment function, we obtain
f(t) 32(1.12t) 32e0.1133t billion dollars is the value of the investment
140 CHAPTER 2 Describing Change: Rates
after t years. The APR from this rewritten formula is 11.33%. It is no coincidence that
the APR in this example is the constant percentage rate of change. Percentage rate of
change is a measure of the percentage change at a point. Continuous compounding
considers constant percentage change at a point.
Rate of change (or instantaneous rate of change) and percentage rate of change
are measures of change occurring at a point. Rates of change (also known as deriva-
tives) are the first concept of calculus that we will study in detail. In Chapter 4, we will
use derivatives to answer questions concerning the maxima and minima of models.
3. Typing Let W(t) be the number of words per
2.3 Concept Inventory minute that a student in a typing class can type after
t weeks in the course.
• Derivative rate of change slope of tangent line a. Is it possible for W(t) to be negative? Explain.
• Derivative notation b. What are the units on W(t)?
• Relationships between the graph of f and c. Is it possible for W(t) to be negative? Explain.
statements about f
• Percentage rate of change 4. Corn Crop Suppose that C(f ) is the number of
bushels of corn produced on a tract of farm land
• Slope of a tangent line the limiting value of
when f pounds of fertilizer are used.
slopes of secant lines
a. What are the units on C(f )?
• Numerical method of estimating the slope of a
tangent line b. Is it possible for C(f ) to be negative? Explain.
c. Is it possible for C(f ) to be negative? Explain.
5. Profit Suppose that F(p) is the weekly profit, in
thousands of dollars, that an airline makes on its
flights from Boston to Washington D.C. when the
ticket price is p dollars. Interpret the following:
2.3 Activities a. F(65) 15
b. F(65) 1.5
Getting Started dF
c. dp 2 when p 90
1. Distance Suppose that P(t) is the number of miles
from an airport that a plane has flown after t hours. 6. Sales Let T(p) be the number of tickets from
d
Boston to Washington D.C. that a certain airline
a. What are the units on dt[P(t)]? sells in 1 week when the price of each ticket is p
d
dollars. Interpret the following:
b. What common word do we use for dt[P(t)]? a. T(115) 1750
2. Mutual Fund Let B(t) be the balance, in dollars, b. T(115) 220
dT
in a mutual fund t years after the initial investment. c. dp 22 when p 125
Assume that no deposits or withdrawals are made
during the investment period. 7. On the basis of the following information, sketch a
possible graph of t with input x.
dB
a. What are the units on dt ?
• t(3) 7
dB
b. What is the financial interpretation of dt ? • t(4.4) t(8) 0
2.3 Derivative Notation and Numerical Estimates 141
dt • The average rate of change of P between 1970
• dx 0 at x 6.2
and 1980 is 0.6 percentage point per year.
• The graph of t has no concavity changes.
• Lines tangent to the graph of P lie below the
8. Using the information that follows, sketch a possible graph at all points between 1940 and 1990 and
graph of m with input t. above the graph between 1990 and 2000.
• m(0) 3 12. Enrollment E(x) is the public secondary school
d enrollment, in millions of students, in the United
• dt[m(t)] 0.34 States between 1940 and 2008 x years after 1940. Use
9. Weight Loss Suppose that W(t) is your weight t the following information to sketch a graph of E.
(Sources: Based on data appearing in Datapedia of the United
weeks after you begin a diet. Interpret the following:
States, Lanham, MD: Bernan Press, 1994; and in Statistical
a. W(0) 167 Abstract, 1998 and 2001.)
b. W(12) 142 • E(40) 13.2
dW
c. dt 2 when t 1 • The graph of E is always concave down.
dW • Between 1980 and 1990, enrollment declined at
d. dt 1 when t 9
an average rate of 0.19 million students per year.
e. W´(12) 0
• The projected enrollment for 2008 is 14,400,000
f. W´(15) 0.25 students.
g. On the basis of the information in parts a
• It is not possible to draw a line tangent to the
through f, sketch a possible graph of W.
graph of E at x 50.
10. Fuel Efficiency Suppose that G(v) equals the fuel 13. Profit Let P(x) be the profit in dollars that a fra-
efficiency, in miles per gallon, of a car going v miles ternity makes selling x number of T-shirts.
per hour. Give the practical meaning of the following
a. Is it possible for P(x) to be negative? Explain.
statements.
b. Is it possible for P´(x) to be negative? Explain.
a. G(55) 32.5
c. If P´(200) 1.5, is the fraternity losing
dG
b. dv 0.25 when v 55 money? Explain.
c. G´(45) 0.15 14. Politics Let M(t) be the number of members in a
d political organization t years after its founding.
d. dv[G(51)] 0 d
What are the units on dt[M(t)]?
11. Births P(b) is the percentage of all births to single 15. Doubling Time Let D(r) be the time, in years, that
mothers in the United States in year b from 1940 it takes for an investment to double if interest is
through 2000. Using the following information, continuously compounded at an annual rate of r%.
sketch a graph of P. (Here r is expressed as a percentage, not a decimal.)
(Source: Based on data from L. Usdansky, “Single Motherhood: dD
a. What are the units on dr ?
Stereotypes vs. Statistics,” New York Times, February 11, 1996,
Section 4, page E4, and on data from Statistical Abstract, 1998.) dD
b. Why does it make sense that dr is always negative?
• P(1940) 4 c. Give the practical interpretation of the following:
• P´(b) is never zero. i. D(9) 7.7
• P(b) 12, when b 1970 dD
ii. dr 2.77 when r 5
• P(2000) is about 21 percentage points more than dD
P(1970). iii. dr 0.48 when r 12
142 CHAPTER 2 Describing Change: Rates
16. Unemployment Let U(t) be the number of people d. Estimate the rate at which the speed is rising for
unemployed in a country t months after the election a raindrop with diameter 2 mm. Use this esti-
of a new president. mate to approximate the terminal speed of a
a. Draw and label an input/output diagram for U. raindrop with diameter 2.5 mm.
b. Is U a function? Why or why not? e. Find and interpret the percentage rate of change
of speed for a raindrop with diameter 2 mm.
c. Interpret the following facts about U(t) in state-
ments describing the unemployment situation: 18. Customers The scatter plot and graph depict the
i. U(0) 3,000,000 number of customers that a certain fast-food
restaurant serves each hour on a typical weekday.
ii. U(12) 2,800,000
iii. U´(24) 0 Customers
dU 80
iv. dt 100,000 when t 36
dU 60
v. dt 200,000 when t 48
d. On the basis of the information in part c, sketch a 40
possible graph of the number of people unem-
ployed during the first 48 months of the presi- 20
dent’s term. Label numbers and units on the axes.
0 Time
7 A.M. 9 11 1 P.M. 3 5 7 9 P.M.
Applying Concepts
17. Raindrop The accompanying graph shows the a. Estimate the average rate of change of the
terminal speed, in meters per second, of a raindrop number of customers between 7 A.M. and 11 A.M.
as a function of the size of the drop measured in Interpret your answer.
terms of its diameter. b. Estimate the instantaneous rate of change and
a. Sketch a secant line connecting the points for di- percentage rate of change of the number of
ameters of 1 mm and 5 mm, and estimate its slope. customers at 4 P.M. Interpret your answer.
What information does this secant line slope give? c. List the factors that might affect the accuracy of
b. Sketch a line tangent to the curve at a diameter of your answers to parts a and b.
4 mm. What information does the slope of this d. Use your estimate in part b to approximate the
line give? number of customers at 5 P.M.
19. Study Time Refer once more to the function G,
Terminal speed
your grade out of 100 points on the next calculus
(m/s)
10
test when you study t hours during the week before
the test. The graph of G is shown below.
8
G(t)
6
Points
4 100
2 75
Diameter
0
0 1 2 3 4 5 6 (mm) 50
(Source: R. R. Rogers and M. K. Yau, A Short Course in Cloud
Physics, White Plains, NY: Elsevier Science, 1989.) 25
c. Estimate the derivative of the speed for a diame- t
0
ter of 4 mm. Interpret your answer. 0 2 4 6 8 10 12 14 16 Hours
2.3 Derivative Notation and Numerical Estimates 143
a. Carefully draw tangent lines at 4 hours and y
11 hours. Estimate the slope of each tangent line. 16
b. Estimate the average rate of change between y = 2x
12
4 hours and 10 hours. Interpret your answer.
8
c. Estimate the percentage rate of change of the
grade after 4 hours of studying. Interpret your 4
answer.
0 x
d. Use G´(4) to estimate your grade after 4.6 hours 0 1 2 3 4
of studying. b. Use the equation y 2x to estimate numerically
20. Mortality Consider the accompanying graph of the slope of the line tangent to the graph at
rates of death from cancer among U.S. males. x 2.
80
22. a. Sketch a line tangent to the graph of
y x2 4x at the point corresponding to
x 3, and estimate its slope.
Lung
y
Rate per 100,000 male population
60 4
y = −x 2 + 4x
3
2
40
Stomach 1
0 x
Colon and Rectum 0 1 2 3 4
Prostate
b. Use the equation y x2 4x to estimate
20
numerically the slope of the line tangent to the
Pancreas graph at x 3.
23. Airport Traffic The number of passengers going
Liver
0 Year through Hartsfield-Jackson International Airport in
1930 1940 1950 1960 1970 1980 1990 Atlanta, Georgia between 2001 and 2005 yearly, can
Cancer death rates by site, males, United States, 1930–89
be modeled as
(Rates are adjusted to the 1970 U.S. census population.) P(t) 0.29t3 2.92t2 6.1t
(Source: Figure courtesy of the American Cancer Society, Inc.) 79.4 million passengers
a. Estimate how rapidly the number of deaths due to where t is the number of years since 2000. A graph
lung cancer was increasing in 1970 and in 1980. of the function is shown below.
b. Estimate the percentage rate of change of deaths P
due to liver cancer in 1980. Passengers
(millions)
c. Estimate the slope of the stomach cancer curve in 86
1960.
84
d. Describe in detail the behavior of the lung cancer
82
curve from 1930 to 1990. Explain why the lung
cancer curve differs so radically from the other 80
curves shown. 78
e. List as many factors as you can that might affect 76
a cancer death rate curve. 74 t
1 2 3 4 5 Years
21. a. Sketch a line tangent to the graph of y at 2x (Sources: Based on data from
the point corresponding to x 2, and estimate Airports Council International,
its slope. www.airports.org)
144 CHAPTER 2 Describing Change: Rates
dP a. Use the numerical method to find the rate of
a. Use the graph to estimate dt when t 4.
change of the time for a 13-year-old swimmer to
b. Use the equation to investigate P´(4) numeri- swim 100 meters freestyle.
cally. Interpret your answer. b. Determine the percentage rate of change of swim
c. Use the equation to determine the percentage time for a 13-year-old.
rate of change of P at t 4. Interpret your c. Is a 13-year-old swimmer’s time improving or
answer. getting worse as the swimmer gets older?
d. Discuss the advantages and disadvantages of
26. Sales Annual U.S. factory sales, in billions of dol-
using the two methods (in parts a and b) for
lars, of consumer electronics goods to dealers from
finding derivatives.
1990 through 2001 can be modeled by the equation
24. Bank Account The balance in a savings account is S(t) 0.0388t3 0.495t2 5.698t 43.6
shown in the graph and is given by the equation billion dollars
Balance 1500(1.0407t) dollars, where t is the
number of years since the principal was invested. where t is the number of years since 1990.
(Sources: Based on data from Statistical Abstract, 2001, and
Account balance Consumer Electronics Association.)
(dollars) a. Estimate the derivative of S when t 10.
2500
b. Interpret your answer to part a.
2400
2300 27. Profit Let P(x) 1.02x Canadian dollars be the
2200 profit from the sale of x mountain bikes. On
2100 November 25, 2002, P Canadian dollars were worth
2000 P
C(P) 1.5786 American dollars. Assume that this
1900
1800 conversion applies today.
1700 a. Write a function for profit in American dollars
1600 from the sale of x mountain bikes.
t
0 2 4 6 8 10 12 14 Years b. What is the profit in Canadian and in American
dollars from the sale of 400 mountain bikes?
a. Using only the graph, estimate how rapidly the
c. How quickly is profit (in American dollars)
balance is growing after 10 years.
changing when 400 mountain bikes are sold?
b. Use the equation to investigate numerically the
rate of change of the balance when t 10. 28. Profit Let P(x) 1.02x Canadian dollars be the
profit from the sale of x mountain bikes. On
c. Which of the two methods (part a or part b) is
November 25, 2002, P Canadian dollars were worth
more accurate? Support your answer by listing P
the different estimates that had to be made dur- C(P) 1.5786 American dollars. Assume that this
ing each method. conversion applies today.
d. Use the equation to determine the percentage a. Write a function giving average profit per
rate of change of the balance when t 10. mountain bike for the sale of x mountain bikes in
Canadian dollars.
25. Swim Time The time it takes an average athlete to
swim 100 meters freestyle at age x years can be b. Write a function for average profit in American
modeled by the equation dollars.
c. How quickly is average profit (in American
T(x) 0.181x2 8.463x 147.376 seconds
dollars) changing when 400 mountain bikes are
(Source: Based on data from Swimming World, August 1992.) sold?
2.4 Algebraically Finding Slopes 145
Discussing Concepts 31. Discuss the advantages and disadvantages of finding
rates of change graphically and numerically. Include
29. Explain how percentage change and percentage rate in your discussion a brief description of when each
of change are related and how they differ. method might be appropriate to use.
30. Describe the process of using tangent lines to 32. Explain why there may be differences between the
approximate function values. Include a discussion numerical estimate of a rate of change of a modeled
of when this technique is most accurate and when it function at a point and the actual rate of change that
is least accurate. occurred in the underlying real-world situation.
2.4 Algebraically Finding Slopes
When we have enough carefully chosen close points to observe the trend in the secant
line slopes, the numerical method is a fairly good way to find a slope to a specified
accuracy. For this reason, the process of numerically estimating slope is valuable, but
keep in mind that it gives only an estimation of the actual slope. However, we can
generalize the numerical method to develop an algebraic method that will give the
exact slope of a tangent line at a point.
We saw in Sections 2.2 and 2.3 that the slope of a tangent line is the limiting value
of the slopes of secant lines at a point. Before we proceed to evaluating slopes
algebraically, we need to revisit in a little more detail the idea of the limiting value of
a function at a point.
Limits and the Infinitesimally Small
In Section 1.4, we used the concept of limits to consider end behavior. In calculus we
are also interested in the behavior of the output of a function as the input of that
function gets closer and closer to a certain value. This type of behavior analysis is
similar to looking at the function through a microscope and increasing the power of
the magnification so as to zoom in on a very small portion of that function.
t2 16
Concept Development: Limit at a Point Consider r(t) t 4 . The graph
of r appears to be a line, but as we zoom in on the graph near t 4, a hole in the graph
becomes evident. (See Figure 2.34).
r(t)
12 9.5 8.375
8
8 8
4
0 t 6.5 t 7.625 t
4 7 3 4 5 3.75 4 4.25
FIGURE 2.34
146 CHAPTER 2 Describing Change: Rates
r(t)
t
4
FIGURE 2.35
To graphically estimate the behavior of r as t gets closer and closer to 4, place the
tips of two pencils on the graph in Figure 2.35 on either side of t 4 and, keeping the
pencils on the graph, move them toward each other. The pencil tips should move
toward the hole in the function. Indeed, as the input becomes closer and closer to 4
from either direction, the graph becomes closer and closer to an output of 8. The fact
that r(t) is never actually 8 does not affect the limit, because when determining limits,
we are interested in where the function is headed, not in whether it ever actually
arrives there.
The output r(t) is not actually 8 at t 4 because the output is not defined for
t 4. However, we can say that as the input t approaches (gets closer and closer to)
4, the output r(t) approaches 8. Mathematically, we say, “as t approaches 4, the
limiting value of r is 8.” Thus we write lim r(t) 8.
t→4
It is also possible to estimate this limit numerically. To do so, we evaluate the
function at values increasingly close to, and on either side of, t 4. Table 2.4 shows
r(t) values as t approaches 4 from the left (this is denoted t → 4 ), and Table 2.5
shows r(t) values as t approaches 4 from the right (this is denoted t → 4 ).
TABLE 2.4 TABLE 2.5
t → 4 r (t ) t → 4 r (t )
3.8 7.8 4.2 8.2
3.9 7.9 4.1 8.1
3.99 7.99 4.01 8.01
3.999 7.999 4.001 8.001
Caution: Do not confuse the symbols used to indicate approaching an input value
from the right or from the left with the statement “x → c” or “x → .” A plus or
minus sign placed in front of a number or the infinity symbol indicates whether the
number or infinity is positive or negative, whereas a plus or minus sign written after
a number (x → c or x → c ) indicates a specific direction from which that number
is to be approached.
2.4 Algebraically Finding Slopes 147
In Table 2.4 it appears that as t approaches 4 from the left, the output is becom-
ing closer and closer to 8. We symbolize this idea by writing
lim r(t) 8
t→4
This is read as “the limit of r(t) as t approaches 4 from the left is 8.” On the basis
of Table 2.5, we make a similar statement: “The limit of r(t) as t approaches 4 from
the right is 8,” which we write as
lim r(t) 8
t→4
Note that the input values near 4 in the tables were arbitrarily chosen. You should
arrive at the same conclusion if you choose other values increasingly close to t 4.
Because the limits from the left and right of 4 are the same, we conclude that
lim r 8; that is, the limit of r(t) as t approaches 4 is 8.
t→4
Another function behavior that is possible is the outputs increasing or decreasing
without bound as the inputs approach a specific value from the left or from the right.
Consider Example 1.
EXAMPLE 1 Numerically Estimating a Limit at a Point
3x
For u(x) 8x 2, numerically estimate lim u(x).
x→0.25
Solution
Note that u(x) is not defined at x 0.25 because the denominator is zero when
x 0.25. Consider values of x increasingly close to 0.25 from both the left and
the right, as shown in Table 2.6.
TABLE 2.6
2.4.1
x 0.25 u(x ) x → 0.25 u (x )
0.251 94.125 0.249 93.375
0.2501 937.875 0.2499 937.125
0.25001 9375.375 0.24999 9374.625
0.250001 93750.375 0.249999 93749.625
It appears that the output values of u are becoming increasingly large as x moves
closer and closer to 0.25 from the left. Also, the output values of u are negative, and
their magnitudes seem to become increasingly large as x moves closer and closer to
0.25 from the right. On the basis of this numerical investigation, we conclude that the
limits as x approaches 0.25 from the left and right do not exist. To indicate that u(x)
increases without bound as x approaches 0.25 from the left, we use the notation
lim u(x) → . We also write lim u(x) → to indicate that u(x)
x→0.25 x→0.25
decreases without bound as x approaches 0.25 from the right.
148 CHAPTER 2 Describing Change: Rates
Keep in mind that this numerical method gives only an estimate. Figure 2.36 shows a
graph of the function u.
u(x)
x
–1 –0.25 1
–1
FIGURE 2.36
The graph seems to support our previous estimates that lim
x→0.25
u(x) → and
that lim u(x) → . You should recognize the numerical technique illustrated
x→0.25
in Example 1 as being the same basic technique used in Section 2.3 to evaluate the slope
of a tangent line. ●
Continuity Revisited
The functions r and u seen in this section are not continuous functions. Because
calculus involves the study of continuous portions of functions, it is important to have
a complete understanding of what makes a function continuous or discontinuous.
Let us first consider some functions that are not continuous and the types of
situations that lead to a function’s being discontinuous at a point. Consider the four
functions shown in Figure 2.37. We observe that a function is not continuous at a
point where a hole in the graph occurs (see f), the output becomes infinitely large (see
g), or a break or jump occurs (see h and j).
f (x) g(x) h (x) j (x)
x x x x
c c c c
FIGURE 2.37
We say that a function is continuous on an (open) interval if the output of the
function is defined at every point on the interval and there are no breaks or jumps in
the function output.
2.4 Algebraically Finding Slopes 149
Continuity
A function is continuous on an (open) interval if the output of the function
is defined at every point on the interval and there are no breaks or jumps in
the function output.
Limits give us the power to evaluate the behavior of a continuous function at a
point. A special function that we have been working with in the chapter is the
function that gives us the slope of the secant lines of another function. This slope
function is continuous as long as the original function is continuous and smooth.
We now use the power of limits algebraically in order to determine expressions for
slope functions.
Finding Slopes Using the Algebraic Method
Consider finding the slope of f(x) 0.1x2 3x at x 2. We begin with the point
(2, 6.4). Instead of choosing close x-values such as 1.9, 2.1, 2.01, and so on, we simply
call the close value x 2 h. (Note that if h 0.1, then x 2.1; if h 0.01,
x 2.01; if h 0.01, x 1.99, and so on.) The output value that corresponds to
x 2 h is
f(2 h) 0.1(2 h)2 3(2 h)
0.1(4 4h h2) 6 3h 6.4 3.4h 0.1h2
Next we find the slope of the secant line between the point of tangency (2, 6.4) and
the close point (2 h, 6.4 3.4h 0.1h2).
(6.4 3.4h 0.1h2) 6.4 3.4h 0.1h2
Slope of secant line
(2 h) 2 h
We now have a formula for the slope of the secant line through the points at x 2
and x 2 h. We can apply the formula to obtain slopes at points increasingly close
to x 2 (see Table 2.7).
The slopes become increasingly close to approximately 3.4.
We arbitrarily chose three-
Numerically evaluating the secant line slope formula for smaller and smaller
decimal-place accuracy for the values of h gives us a good picture of where the slopes are headed, but it does not give
answer in this illustration. us the exact answer. By noting that h approaches zero as the close point approaches
TABLE 2.7
Close point Close point
3.4h 0.1h 2 3.4h 0.1h 2
to left h Slope to right h Slope
h h
1.9 0.1 3.39 2.1 0.1 3.41
1.99 0.01 3.399 2.01 0.01 3.401
1.999 0.001 3.3999 2.001 0.001 3.4001
1.9999 0.0001 3.39999 2.0001 0.0001 3.40001
1.99999 0.00001 3.399999 2.00001 0.00001 3.400001
150 CHAPTER 2 Describing Change: Rates
2, we can find the limit of the secant line slope formula as h approaches zero. That is,
we need to find
3.4h 0.1h2
lim
h→0 h
The secant line slope formula is not continuous at h 0 because it is not defined
there. However, a rule called the Cancellation Rule for limits states:
If the numerator and denominator of a rational function share a common
factor, then the new function obtained by algebraically canceling the
common factor has all limits identical to those of the original function.
If we factor h out of the numerator and cancel this common factor, then we can
find the limit by evaluating the new expression at h 0 because the new expression
(3.4 0.1h) is continuous at h 0.
h(3.4 0.1h)
lim lim (3.4 0.1h) 3.4 0.1(0) 3.4
h→0 h h→0
On the basis of this limit calculation, we state that the slope of the line tangent to
f(x) 0.1x2 3x at x 2 is exactly 3.4.
This method of finding a formula for the slope of a secant line in terms of h and then
determining the limiting value of the formula as h approaches zero is called the alge-
braic method. It is important because it always yields the exact slope of the tangent line.
EXAMPLE 2 Using the Algebraic Method to Find the Slope at a Point
Coal Production The amount of coal used quarterly for synthetic-fuel plants in the
United States between 2001 and 2004 can be modeled by
f(x) 1.6x2 15.6x 6.4 million short tons
where x is the number of years since the beginning of 2000.
df
a. Find dx at x 3.5 by writing an expression for the slope of a secant line in terms
of h and then evaluating the limit as h approaches 0.
df
b. Interpret dx at x 3.5 in the context given.
Solution
a. First, we find the output of f when x 3.5.
f(3.5) 28.6
Second, we write an expression for f(x) when x 3.5 h.
f(3.5 h) 1.6(3.5 h)2 15.6(3.5 h) 6.4
We simplify this expression as much as possible, using the fact that
(3.5 h)2 12.25 7h h2
Thus
f(3.5 h) 28.6 4.4h 1.6h2
2.4 Algebraically Finding Slopes 151
Third, we find the slope of the secant line through the two close points
(3.5, f(3.5)) and (3.5 h, f(3.5 h)) as
f(3.5 h) f(3.5)
Slope of secant line
(3.5 h) 3.5
(28.6 4.4h 1.6h2) (28.6)
(3.5 h) 3.5
Again, we simplify this as much as possible.
(28.6 4.4h 1.6h2) (28.6)
Slope of secant line (continued)
(3.5 h) 3.5
h(4.4 1.6h)
h
Finally, we find the slope of the tangent line (the derivative) at x 3.5 by
evaluating the limit of the slope of the secant line as h approaches 0.
df h(4.4 1.6h)
slope of tangent line lim
dx h→0 h
lim (4.4 1.6h)
h→0
4.4
b. In the second quarter of 2003, the amount of coal being used for synthetic-fuel
plants in the United States was increasing at a rate of 4.4 million short tons
per year. ●
A General Formula for Derivatives
The real value of the algebraic method is not in finding a slope at a particular point
but in finding general formulas for derivatives. These rate-of-change (or slope)
formulas can be used to find rates of change for many input values.
To illustrate, consider y x2. Because we desire a general equation for any x-value,
we use (x, x2) as the point of tangency. This is the same idea as the algebraic method in
Example 2 and the preceding discussion, but there we worked with a numerical value for
x. Next we choose a close point. We use x h as the x-value of the close point and find
the y-value by substituting x h into the function: y (x h)2 x2 2xh h2.
Thus the original point is (x, x2), and a close point is (x h, x2 2xh h2). Now we
find the slope of the secant line between these two points.
(x2 2xh h2) x2 2xh h2
Slope of the secant line
(x h) x h
Finally, we determine the limiting value of the secant line slope as h approaches 0.
2xh h2
lim lim (2x h) 2x 0 2x
h→0 h h→0
dy
Therefore, the slope formula for y x2 is dx 2x. Using this formula, we find
that the slope of the graph of y x2 is 6 at x 3, 12 at x 6, 0 at x 0, 1 at
x 0.5, and so on.
152 CHAPTER 2 Describing Change: Rates
This method can be generalized to obtain a formula for the derivative of an
arbitrary function.
Four-Step Method to Find f (x)
Given a function f, the equation for the derivative with respect to x can be
found as follows:
1. Begin with a typical point (x, f(x)).
2. Choose a close point (x h, f(x h)).
3. Write a formula for the slope of the secant line between the two points.
f(x h) f(x) f(x h) f(x)
Slope
(x h) x h
It is important at this step to simplify the slope formula.
4. Evaluate the limit of the slope as h approaches 0.
f(x h) f(x)
lim
h→0 h
This limiting value is the derivative formula at each input where the
limit exists.
Thus we have the following derivative formula (slope formula, rate-of-change
formula) for an arbitrary function:
Derivative Formula
dy
If y f(x), then the derivative dx is given by the formula
dy f(x h) f(x)
lim
dx h→0 h
provided that the limit exists.
Example 3 illustrates the Four-Step Method for finding a derivative formula.
EXAMPLE 3 Using the Four-Step Method to Find a Slope Formula
Coal Prices The average price paid by the synfuel industry for coal between 2002
and 2005 can be modeled by
W(t) 1.2t2 6.1t 39.5 dollars per short ton
where t is the number of years since the beginning of 2000.
a. Use the limit definition of the derivative (the Four-Step Method) to develop a
formula for the rate of change of the price of coal used by the synfuel industry.
2.4 Algebraically Finding Slopes 153
b. How quickly was the price of coal used by the synfuel industry growing in the
middle of 2003?
Solution
a. Step 1. A typical point is (t, W(t)) (t, 1.2t2 6.1t 39.5).
Step 2. A close point is (t h, W(t h))
(t h, 1.2(t h)2 6.1(t h) 39.5).
We rewrite the output of the close point before we proceed:
W(t h) 1.2(t h)2 6.1(t h) 39.5
1.2(t2 2ht h2) 6.1(t h) 39.5
1.2t2 2.4ht 1.2h2 6.1t 6.1h 39.5
Step 3. Write the slope of the secant line between the two points from Steps 1
and 2.
W(t h) W(t)
Slope
(t h) t
[(1.2t 2.4ht 1.2h2 6.1t 6.1h 39.5) (1.2t2 6.1t 39.5)]
2
(t h) t
2.4ht 1.2h 6.1h
2
h
h(2.4t 1.2h 6.1)
h
Step 4. Evaluate the limit of the secant line slope as h approaches 0.
h(2.4t 1.2h 6.1)
W(t) lim lim (2.4t 1.2h 6.1)
h→0 h h→0
2.4t 6.1
Thus the price of coal used by the synfuel industry was increasing by
W(t) 2.4t 6.1 dollars per short ton per year
t years after the beginning of 2000, 2 t 5.
b. We find that W(3.5) 2.3 dollars per short ton per year. Thus the price of
coal used by the synfuel industry was increasing by approximately 2.3 dollars
per short ton per year in the middle of 2003. ●
Using the Four-Step Method to find a derivative formula for a nonpolynomial
function may require more thoughtful algebra.
EXAMPLE 4 Algebraically Finding a Slope Formula
Consider the function f(x) 2x.
a. Use the Four-Step Method to find a formula for the slope graph of f.
b. Use the formula from part a to find the slope of the graph of f at x 1.
154 CHAPTER 2 Describing Change: Rates
Compare this answer with the one found by numerical estimation in
Section 2.3.
Solution
a. To find a slope formula, begin with a general point x, 2x and a close point
x h, 2x h . Next find the slope between these two points:
2x h 2x 2x h 2x
Secant line slope
xhx h
Now find the limit of this formula as h approaches zero:
2x h 2x
lim
h→0 h
Unlike the case for polynomial functions that always contain h as a common
factor, we cannot cancel the h here without rewriting the numerator. The key to
this cancellation is to rewrite the numerator by multiplying the numerator and
denominator by the term 2x h 2x. Observe how this multiplication
enables us to cancel the h term:
2x h 2x 2x h 2x
lim
h→0 h 2x h 2x
4(x h) 4x
lim
h→0 h 2x h 2x
4h
lim
h→0 h 2x h 2x
4
lim
h→0 2x h 2x
4 4 1
2x 0 2x 4x x
1
Thus the slope formula is f(x) .
x
1
b. Using this slope formula, we find that f(1) 1. That is, the slope of
1
the line tangent to the graph of f(x) 2x at x 1 is 1. This calculation
confirms that the numerical estimate in Section 2.3 is correct. ●
The definition of the derivative of a function gives us a formula for the slope
graph of the function, which enables us to calculate exact rates of change quickly.
Unfortunately, this method is primarily for polynomial functions and not for
exponential, logarithmic, or logistic functions. However, in Chapter 3 we will use the
algebraic method as a powerful tool to help us develop some general rules for
derivative formulas.
2.4 Algebraically Finding Slopes 155
14. The derivative of f(x) 2x3 5x 7 is
2.4 Concept Inventory df
dx 6x 5.
2
• The algebraic method for determining the slope of Applying Concepts
a graph at a given point
• The Four-Step Method for determining a rate-of- 15. Swim Time The time it takes an average athlete to
change formula swim 100 meters freestyle at age x years can be
modeled by the equation
• Limit definition of a derivative T(x) 0.181x2 8.463x 147.376
a. Find the swim time when x 13.
2.4 Activities b. Write a formula for the average swim time when
x 13 h.
Getting Started c. Write a simplified formula for the slope of the
For Activities 1 through 4, numerically determine the secant line connecting the points at x 13 and
indicated limit if it exists. x 13 h.
d. What is the limiting value of the slope formula in
x3 5x2
1. lim 2. lim part c as h approaches 0? Interpret your answer.
x→2 x 2 x→3 3x 9
2x3 7x e1x e 16. Sales Annual U.S. factory sales of consumer
3. lim 4. lim electronics goods to dealers between 1990 and 2001
x→0 x x→0 x
can be modeled by the equation
5. Consider the function f(x) 4x2. Use the algebraic
method to find f(2) by evaluating the limit of an S(x) 0.0388x3 0.495x2 5.698x 43.6
expression for the slope of a secant line. billion dollars
6. Consider the function s(t) 2.3t2. Use the alge- where x is the number of years since 1990.
braic method to find s(1.5) by evaluating the limit (Sources: Based on data from Statistical Abstract, 2001, and
of an expression for the slope of a secant line. Consumer Electronics Association.)
a. Find the sales when x 10.
7. Consider the function g(t) 6t2 7. Use the al- b. Write an expression for the sales when
dg
gebraic method to find dt at t 4 by evaluating the x 10 h.
limit of an expression for the slope of a secant line. c. Write a simplified formula for the slope of the
secant line connecting the points at x 10 and
8. Consider the function m(p) 4p p2. Use the alge- x 10 h.
dm
braic method to find dp at p 2 by evaluating the d. What is the limiting value of the slope formula
limit of an expression for the slope of a secant line. in part c as h approaches 0? Interpret your
answer.
In Activities 9 through 14, use the Four-Step Method out-
17. Airline Fuel The amount of airline fuel consumed
lined in this section to show that each statement is true.
dy by Northwest Airlines each year between 1998 and
9. The derivative of y 3x 2 is dx 3. 2004 can be modeled as
dy
10. The derivative of y 15x 32 is dx 15. f(t) 0.042t2 0.18t 1.89 billion gallons
11. The derivative of f(x) 3x2 is f(x) 6x. where t is the number of years since 1998.
(Source: Northwest Airlines Corporation, Financial and
12. The derivative of f(x) 3x2 5x is
Operating Statistics 5/27/05, ir.nwa.com)
f(x) 6x 5.
dy a. Find the amount of fuel consumed in 2001.
13. The derivative of y x3 is dx 3x2.
b. Write a formula in terms of h for the consump-
(Hint: (x h)3 x3 3x2h 3xh2 h3) tion of fuel a little after 2001.
156 CHAPTER 2 Describing Change: Rates
c. Write a simplified formula for the slope of the c. How quickly (in miles per hour) is the tractor
secant line connecting the points at 2001 and a moving?
little after 2001.
21. Drivers The number of licensed drivers between
d. What is the limiting value for the slope formula the ages of 16 and 21 in 1997 is given below.
as h approaches 0? Interpret your answer.
18. Tuition CPI The CPI for college tuition between
Age Number of drivers
1990 and 2000 can be modeled by the equation
(years) (millions)
c(t) 0.498t2 20.603t 174.458 16 0.85
where t is the number of years since 1990. 17 1.24
(Source: Based on data from Statistical Abstract, 2001.)
18 1.41
a. Find the consumer price index for college tuition 19 1.47
in 1998.
20 1.54
b. Write a formula in terms of h for the con-
21 1.51
sumer price index of college tuition a little after
1998. (Source: U.S. Department of Labor and
Transportation.)
c. Write a simplified formula for the slope of the
secant line connecting the points at 1998 and a
little after 1998. a. Find a quadratic model for the number of licensed
drivers as a function of age.
d. What is the limiting value for the slope formula
b. Use the limit definition of the derivative to
as h approaches 0? Interpret your answer.
develop a formula for the derivative of the
19. Falling Object An object is dropped off a build- rounded model.
ing. Ignoring air resistance, we know from physics c. Use the derivative formula in part b to find the
that its height above the ground t seconds after rate of change of the function in part a for an age
being dropped is given by of 20 years. Interpret your answer.
Height 16t2 100 feet d. Find the percentage rate of change in the number
of licensed drivers 20 years old. Interpret this
a. Use the Four-Step Method to find a rate-of-
result.
change equation for the height.
b. Use your answer to part a to determine how 22. Drivers The data below give the percentage of
rapidly the object is falling after 1 second. females of a certain age who were licensed drivers in
1997.
20. Distance Clinton County, Michigan, is mostly
flat farmland partitioned by straight roads (often
gravel) that run either north/south or east/west. A Age Licensed drivers
tractor driven north on Lowell Road from the (years) (percent)
Schafers’ farm is 15 0.4
d(m) 0.28m 0.6 miles 16 43.4
north of Howe Road m minutes after leaving the 17 61.9
farm’s drive. 18 72.7
a. How far is the Schafers’ drive from Howe Road? 19 73.8
b. Use the Four-Step Method to show that the trac- (Source: U.S. Department of Labor and
tor is moving at a constant speed. Transportation.)
Chapter 2 Summary 157
a. Find a quadratic model for these data. Discussing Concepts
b. Use the limit definition of the derivative to de- 23. Discuss the advantages and disadvantages of finding
velop the derivative formula for the rounded rates of change graphically, numerically, and
equation. algebraically. Include in your discussion a brief
c. Use the derivative formula in part b to find the description of when each method might be appro-
rate of change of the equation in part a when the priate to use.
input is 16 years of age. Interpret your answer.
24. Explain from a graphical viewpoint how algebraically
d. Find the percentage rate of change in the number finding a slope formula is related to numerically
of female licensed drivers 20 years old. Interpret estimating a rate of change.
this result.
SUMMARY
This chapter is devoted to describing change: the under- point of tangency. We can estimate the instantaneous rate
lying concepts, the language, and proper interpretations. of change at a point on a curve by sketching a tangent line
Change, Average Rate of Change, and Percent- at that point and approximating the tangent line’s slope.
age Change Derivatives and Percentage Rate of Change
The change in a quantity over an interval is a difference Derivative is the calculus term for (instantaneous) rate
of output values. Apart from expressing the actual of change. Accordingly, all of the following terms are
change in a quantity that occurs over an interval, change synonymous: derivative, instantaneous rate of change,
can be described as the average rate of change over an rate of change, slope of the curve, and slope of the line
interval or as a percentage change. The numerical tangent to the curve.
description of an average rate of change has an associ- Three common ways of symbolically referring to the
ated geometric interpretation—namely, the slope of the dG
secant line joining two points on a graph. derivative of a function G with respect to x are dx ,
d
Instantaneous Rates of Change dx[G(x)], and G(x). The proper units on derivatives are
output units per input unit. Rates of change also can be
Whereas average rates of change indicate how rapidly a expressed as percentages. A percentage rate of change
quantity changes (on average) over an interval, instanta- describes the relative magnitude of the rate.
neous rates of change indicate how rapidly a quantity is
Numerically and Algebraically Finding Slopes
changing at a point. The instantaneous rate of change at
a point on a graph is simply the slope of the line tangent When we have an equation y f(x) to associate with
to the graph at that point. It describes how quickly the the curve, we can improve our graphical approxima-
output is increasing or decreasing at that point. tions of the slope of the tangent line with numerical ap-
proximations of the limit of slopes of secant lines. The
Tangent Lines method of numerically estimating slopes can be general-
The principle known as local linearity guarantees that ized to provide a valuable algebraic method for finding
the graph of any continuous function looks like a line if exact slopes at points, as well as formulas for slopes at
you are close enough. A line tangent to a graph at a point any input value. We call this method the Four-Step
is the line you see when you zoom in on the graph closer Method of finding derivatives. This method yields the
and closer to that point. formal definition of a derivative: If y f(x), then
The line tangent to a graph at a point P can also be dy f(x h) f(x)
thought of as the limiting position of nearby secant f(x) lim
dx h→0 h
lines—that is, secant lines through P and nearby points
on the graph. It reflects the tilt, or slope, of the graph at the provided that the limit exists.
158 CHAPTER 2 Describing Change: Rates
CONCEPT CHECK
Can you To practice, try
• Find and interpret change, percentage change, and
average rates of change
• using data? Section 2.1 Activities 5, 7
• using graphs? Section 2.1 Activities 9, 11
• using equations? Section 2.1 Activities 17, 19
• Work with APR and APY? Section 2.1 Activities 25, 28
• Illustrate the relationship between secant lines and Section 2.2 Activities 3, 17
tangent lines?
• Accurately sketch tangent lines? Section 2.2 Activity 19
• Use tangent lines to estimate rates of change? Section 2.2 Activities 21, 25
• Use and interpret derivative notation? Section 2.3 Activities 5, 11
• Correctly interpret derivatives? Section 2.3 Activities 13, 15, 23
• Find and interpret percentage rate of change? Section 2.3 Activities 17, 25
• Graphically and numerically estimate rates of change? Section 2.3 Activity 19
• Numerically estimate the limit of a function at a point? Section 2.4 Activities 1, 3
• Use the algebraic method to find a rate of change at a point? Section 2.4 Activities 7, 15
• Use the Four-Step Method to find a rate-of-change formula? Section 2.4 Activities 11, 17
Chapter 2 Concept Review 159
CONCEPT REVIEW
1. Answer the following questions about the graph: States
50
A
40
B 30
E
20
C
D
10
a. List the labeled points at which the slope Year
appears to be (i) negative, (ii) positive, and (iii) 1900 1910 1920 1930
zero. (Source: Based on data from Hamblin, Jacobson, and Miller, A
b. If B is the inflection point, what is the relation- Mathematical Theory of Social Change, New York: Wiley, 1973.)
ship between the steepness at B and the steepness a. Use the graph to approximate how rapidly (on
at the points A, C, and D on the graph? average) the membership was growing from
c. For each of the labeled points, will a tangent line 1915 through 1931.
at the point lie above or below the graph? b. What was the percentage growth in membership
d. Sketch tangent lines at points A, B, and E. from 1915 through 1931?
2. Suppose the graph in Activity 1 represents the speed c. Approximately how rapidly (on average) was the
of a roller coaster (in feet per second) as a function membership growing from 1923 through 1927?
of the number of seconds after the roller coaster
4. Investment Suppose that to help pay for your col-
reached the bottom of the first hill.
lege education, 15 years ago your parents invested a
a. What are the units on the slopes of tangent lines? sum of money that has grown to $25,000 today.
What common word is used to describe the
a. How much did they originally invest if the invest-
quantity measured by the slope in this context?
ment earned 7.5% interest compounded monthly?
b. When, according to the graph, was the roller
b. How much will $25,000 be worth 15 years from
coaster speeding up?
now, if it is invested in an account that earns
c. When was the roller coaster’s speed the slowest? 6.5% quarterly?
d. When was the roller coaster slowing down most c. Write a formula for the accumulated amount in
rapidly? an account in which $25,000 is deposited for t
3. P.T.A. The accompanying graph models the years with interest being paid at 6.5% com-
number of states associated with the national pounded quarterly.
P.T.A. organization from 1895 through 1931. In d. Use the function in part c to determine the aver-
1915, 30 states were associated with the national age rate of change of the amount in the account
organization. The association grew to 48 states by between the fifth and tenth years. What is the
1931. percentage change for this same time period?
160 CHAPTER 2 Describing Change: Rates
5. Employees The graph of a model for the number b. What does a negative derivative indicate about a
of Dell Computer Corporation employees between swimmer’s time?
1992 and 2002 is shown below.
7. Airline Revenue Northwest Airlines revenues
E(x) from passengers for selected years between 1991
Employees and 2003 are given in the table.
40,000
Passenger revenue
30,000 Year (billion dollars)
20,000 2003 7.632
2001 8.207
10,000
1999 8.692
x 1997 8.822
0
1992 1995 1998 2001 Year
1995 7.762
(Source: Based on data in Hoover’s Online Guide Company 1993 6.620
Capsules.)
1991 5.862
a. Sketch a secant line between the points with
inputs 1993 and 1997. Describe the information (Source: Northwest Airlines
Corporation, Financial and Operating
that the slope of this line provides. Statistics 5/27/05, ir.nwa.com)
b. Sketch a tangent line at the point corresponding
to 1998. Describe the information that the slope a. Find a model for the data.
of this line provides. b. Numerically investigate the rate of change of the
c. Estimate the average rate of change in the passenger revenue in 2001. Choose at least three
number of employees between 1993 and 1997. increasingly close points. In table form, record
Interpret your answer. the close points, the slopes with four decimal
d. Estimate the rate of change and the percentage places, and the limiting value with two decimal
rate of change in the number of employees in places.
1998. Interpret your answers. c. Interpret the limiting value in part b.
6. Swim Time T(x) is the number of seconds that it d. Give the formula for the derivative of the equa-
takes an average athlete to swim 100 meters free tion in part a. Evaluate the derivative in 2001.
style at age x years. 8. In your own words, outline the Four-Step Method
a. Write sentences interpreting T(22) 49 and for calculating derivatives. Illustrate the method for
T(22) 0.5. the function f(x) 7x 3.
Project 2.1 Fee-Refund Schedules
Setting Select the nonlinear plan that you believe to be the
best choice from both the students’ and the
Some students at many colleges and universities
administration’s perspectives. Outline the reasons
enroll in courses and then later withdraw from them.
for your choice.
Such students may have part-time status upon
withdrawing. Part-time students have begun 2. Estimate the rate of change of your selected
questioning the fee-refund policy, and a public debate equation for withdrawals after 1 week, 3 weeks,
is taking place. Students who enrolled under full-time and 5 weeks. Include any other times that are
status are unable to receive a fee refund even if they indicated by your school’s schedule. Interpret the
have part-time status after dropping their classes. rates of change in this context. How might the rate
Recently, the student senate at one university passed a of change influence the administration’s view of
resolution condemning the current fee-refund the model you chose? Would the administration
schedule. Then, the associate vice president issued a consider a different model more advantageous? If
statement claiming that further erosion of the so, why? Why did you not propose it as your
university’s ability to retain student fees would reduce model of choice?
course offerings. The Higher Education Commission
has scheduled hearings on the issue. The Board of
Trustees has hired your firm as consultants to help Reporting
them prepare their presentation. 1. Prepare a written report of your results for the
Board of Trustees. Include scatter plots, models,
and graphs. Include in an appendix the reasoning
Tasks
that you used to develop each of your models.
1. Examine the current fee-refund schedule for your
2. Prepare a press release for the college or university
college or university. Present a graph and formula
to use when it announces the adoption of your
for the current fee-refund schedule. Critique the
plan. The press release should be succinct and
refund schedule.
should answer the questions who, what, when,
Create alternative fee-refund schedules that where, and why. Include the press release in your
include at least two quadratic plans (one concave report to the Board.
up and one concave down), an exponential plan, a
3. (Optional) Prepare a brief (15-minute)
logistic plan, a no-refund plan, and a complete-
presentation on your work. You will be presenting
refund plan. (Hint: Linear models have constant
it to members of the Board of Trustees of your
first differences. What is true about quadratic and
college or university.
exponential models?) For each plan, present the
refund schedule in a table, in a graph, and with an
equation. Critique each plan from the students’
viewpoint and from that of the administration.
161
Project 2.2 Doubling Time
Setting 2. Dr. Bilkins is interested in knowing how sensitive
doubling time is to changes in interest rates.
Doubling time is defined as the time it takes for an
Estimate rates of change of doubling times at 2%,
investment to double. Doubling time is calculated by
nt 8%, 14%, and 20% when interest is compounded
using the compound interest formula A P1 n
r quarterly. Interpret your answers in a way that
would be meaningful to Dr. Bilkins.
or the continuously compounded interest formula
A Pert . An approximation to doubling time can be
found by dividing 72 by 100r. This approximating Reporting
technique is known as the Rule of 72.
1. Prepare a written report for Dr. Bilkins in which
Dr. C. G. Bilkins, a nationally known financial
you discuss your results in Tasks 1 and 2. Be sure
guru, has been criticized for giving false information
to discuss whether Dr. Bilkins should continue to
about doubling time and the Rule of 72 in seminars.
present the Rule of 72 or should present other
Your team has been hired to provide mathematically
rules that depend on the number of times interest
correct information for Dr. Bilkins to use.
is compounded.
2. Prepare a summary document for Dr. Bilkins. It
Tasks should include (a) a brief summary of how to
1. Construct a table of doubling times for interest estimate doubling time using an approximation
rates from 2% to 20% (in increments of 0.25%) rule and (b) a statement about the error involved
when interest is compounded annually, in using the approximation. Also include a brief
semiannually, quarterly, monthly, weekly, and statement summarizing the sensitivity of doubling
daily. Construct a table of doubling-time time to fluctuations in interest rates. Include the
approximations for interest rates of 2% through document in your written report.
20% when using the Rule of 72. Devise similar 3. (Optional) Prepare a brief (15-minute)
rules for 71, 70, and 69. Then construct tables for presentation of your study. You will be presenting
these rules. Examine the tables and determine the it to Dr. Bilkins. Your presentation should be only
best approximating rule for interest compounded a summary, but you need to be prepared to answer
semiannually, quarterly, monthly, weekly, and any technical questions that may arise.
daily. Justify your choices.
For each interest compounding listed above,
compare percent errors when using the Rule of 72
and when using the rule you choose. Percent error
estimate true value
is true value 100%. Comment on when
the rules overestimate, when they underestimate,
and which is preferable.
162
Determining Change:
Derivatives
3
Concepts Outline
3.1 Drawing Rate-of-Change
Graphs
3.2 Simple Rate-of-Change
Formulas
3.3 Exponential and Logarithmic
Rate-of-Change Formulas
3.4 The Chain Rule
3.5 The Product Rule
3.6 Limiting Behavior Revisited:
L’Hôpital’s Rule
John Henley/CORBIS
Concept Application
The aging of the American population may be one of the demographic changes that has
the greatest impact on our society over the next several decades. Given a model for the
projected number of senior Americans (65 years of age or older), the function and its
derivative can be used to answer questions such as the following:
• What is the projected number of senior Americans in 2030?
• How rapidly will that number be changing in 2030?
• What is the estimated percentage rate of change in the number of senior Americans in
2030?
You will be able to answer these questions by using the model given in Activity 30 of
Section 3.2 and the derivative rules presented in this chapter.
163
164 CHAPTER 3 Determining Change: Derivatives
Chapter Introduction
We have described change in terms of rates: average rates, instantaneous rates, and
percentage rates. Of these three, instantaneous rates are the most important in our
study of calculus. In Chapter 2 we presented a method, using the definition of the
derivative, that enables us to find derivative models for certain functions.
In this chapter, we consider some rules for derivatives: the Simple Power Rule, the
Constant Multiplier Rule, the Sum and Difference Rules, the Chain Rule, the Product
Rule, and the Quotient Rule. These rules provide the foundation needed to work with
more complicated functions that we often encounter in the course of real-life
investigations of change.
Concepts You Will Be Learning
• Drawing rate-of-change graphs or slope graphs (3.1)
• Identifying points of undefined slope (3.1)
• Using simple derivative rules for power, exponential, and log functions (3.2, 3.3)
• Using the Chain Rule for derivatives (3.4)
• Using the Product and Quotient Rules for derivatives (3.5)
• Applying derivative rules to find rate-of-change formulas (3.2, 3.3, 3.4, 3.5)
• Determining the rate of change of a function at a point (3.2, 3.3, 3.4, 3.5)
• Using L’Hôpital’s Rule to determine limits of ratios (3.6)
3.1 Drawing Rate-of-Change Graphs
In Chapter 2 we considered the rate of change of a function at a given point. We
learned how to express the instantaneous rate of change verbally as well as estimate
it graphically, numerically, and algebraically. At the end of Section 2.4 we saw that
the algebraic method for determining the rate of change at a specific point can be
generalized to a formula. The limit definition of the derivative is a key concept of
calculus. In Section 3.2 we develop general rules about slope formulas. However, it
is important that we have a good intuitive understanding of the relationship be-
tween functions and their slope formulas before we begin using derivative rules. To
enhance this understanding, we consider the relationship between the graph of a
function and its slope graph.
Extracting Rate-of-Change Information from a
Function Graph
Every smooth, continuous curve with no vertical tangent lines has a slope associated
with each point on the curve. When these slopes are plotted, they also form a smooth,
continuous curve. We call the resulting curve a slope graph, rate-of-change graph,
or derivative graph.
3.1 Drawing Rate-of-Change Graphs 165
Consider the smooth, continuous curve shown in Figure 3.1. What do we know about
the slopes of this graph? Sketch lines tangent to the curve at the points where x A
and x C and at several other points on the curve, as shown in Figure 3.2.
y y
x x
0 A B C 0 A B C
FIGURE 3.1 FIGURE 3.2
The graph is smooth and continuous so we deduce the following facts:
• The tangent lines at the points with inputs A and C are horizontal, so the slope is
zero at those points.
• Between 0 and A, the graph is increasing, so the slopes are positive. The tangent
lines become less steep moving from 0 to A, so the slopes start off large and
become smaller as we approach A from the left.
• Between A and C, the graph is decreasing, so the slopes are negative.
• At B the graph has an inflection point. This is the point at which the graph is
decreasing most rapidly—that is, the point at which the slope is most negative.
Because the vertical-axis units of • To the right of C, the graph is again increasing, so the slopes are positive. The
the slope graph are different tangent lines become steeper as we move to the right of C, so the slopes become
from those of the function larger as the input increases beyond C.
graph, we do not draw the
slope graph on the same set of We record the above information in Figure 3.3. Then, we sketch a continuous slope
axes as the original graph. graph, as shown in Figure 3.4.
y' y'
+
+ More
Positive positive
Positive
slopes +
Less +
positive Positive
A B C
Zero slope 0 _ _ x x
0
0
Negative Negative _ Negative 0 A B C
slopes Most
negative
FIGURE 3.3 FIGURE 3.4
166 CHAPTER 3 Determining Change: Derivatives
We do not know the specifics of the slope graph—how far below the horizontal axis
it dips, where it crosses the vertical axis, how steeply it rises to the right of C, and so
on. However, we do know its basic shape.
EXAMPLE 1 Sketching a Slope Graph
A Logistic Curve The graph in Figure 3.5 is a logistic curve. Sketch its slope graph.
x
FIGURE 3.5 A
y
Less
positive Positive,
near zero
Most Solution
positive We note that the logistic curve in Figure 3.5 is always increasing. Thus its slope graph
is always positive (above the horizontal axis). Even though there is no relative maxi-
More
positive
mum or relative minimum, the logistic curve does level off at both ends. Thus its
slope graph will be near zero at both ends. (See Figure 3.6.)
x
Positive, We note that the logistic curve has its steepest slope at A because this is the loca-
A
near zero tion of the inflection point. Therefore, the slope graph is greatest (has a maximum)
FIGURE 3.6 at this point. (See Figure 3.7.) We sketch the continuous slope graph in Figure 3.8.
y' y'
+
Most positive
+ +
More Less
Slopes positive positive Slopes
Positive Positive
near zero near zero
+ +
x x
A A
●
FIGURE 3.7 FIGURE 3.8
Example 1 illustrated using information from the graph of a function to sketch the slope
graph. In Example 2, we use information concerning the slopes of a function to sketch
both the graph of the function and its slope graph.
3.1 Drawing Rate-of-Change Graphs 167
EXAMPLE 2 Relating Function and Rate-of-Change Graphs
Decreasing Functions
The graphs in parts a and b are decreasing but with different slopes.
a. Sketch a graph that is always decreasing but has slopes that are always
increasing. Also sketch its slope graph.
b. Sketch a graph that is always decreasing with slopes that are always decreasing.
Also sketch its slope graph.
Solution
y a. If a graph is decreasing, it is falling from left to right and has negative slopes. If
the negative slopes are increasing, they are becoming less negative (moving to-
ward zero) as the input increases. The graph is becoming less steep. Such a
graph must be concave up, as shown in Figure 3.9.
The slopes of this graph are always negative, so a slope graph must lie completely
below the input axis. As Figure 3.10 shows, the negative slopes are approaching
zero. They will never be positive. (See Figure 3.11.)
y
x
Negative Slopes of y
FIGURE 3.9 slope x
Less negative
slope
Even less negative
slope
x
y
FIGURE 3.10 FIGURE 3.11
b. Like the graph in part a, this graph is falling from left to right and has negative
slopes. However, the slopes are becoming more negative. They become more and
more negative. This means the graph becomes more steep as input increases.
A declining graph that becomes increasingly steep looks like the one shown
in Figure 3.12. The slopes are always negative, so the slope graph will lie
x completely below and will move farther away from the input axis as the input
FIGURE 3.12 increases. See Figures 3.13 and 3.14.
y Slopes of y
Negative x
slope
More
negative slope
Even more
negative slope
x
FIGURE 3.13 FIGURE 3.14 ●
168 CHAPTER 3 Determining Change: Derivatives
In the preceding discussion and example, the graphs we used had no labeled tick
marks on the horizontal or vertical axes. In such cases, it is not possible to estimate
the value of the slope of the graph at any given point. Instead, we sketch the general
shape of the slope graph by observing the important points and general behavior of
the original graph, such as
• Points at which a tangent line is horizontal
• The intervals over which the graph is increasing or decreasing
• Points of inflection
• Places at which the graph appears to be horizontal or leveling off
As the previous examples indicate, sketching lines tangent to a curve helps us
determine the relative magnitude of the slopes. As this process becomes more
familiar, you should be able to visualize the tangent lines. This technique is illustrated
in Example 3.
EXAMPLE 3 Using Relative Magnitudes to Sketch Slope Graphs
Growth Rate The height (in centimeters) of a plant often follows the general trend
shown in Figure 3.15. Draw a graph depicting the growth rate of the plant.
Centimeters
Cm per day
0 Days
FIGURE 3.15 0 A B C
Solution
The slopes at A, B, and C are all positive. Is the slope at A smaller or larger than that
at B? It is larger, so the slope graph at A should be higher than it is at B. The graph at
C is not as steep as it is at either A or B, so the slope graph should be lower at C than
Days
0 A B C at B. (See Figure 3.16.)
The slope is steepest near the left endpoint and least steep near the right endpoint.
FIGURE 3.16
Adding these observations to your plot yields Figure 3.17.
Now we sketch the slope graph according to the plot. (See Figure 3.18.)
Cm per day Cm per day
Days Days
0 A B C 0 A B C
FIGURE 3.17 FIGURE 3.18 ●
3.1 Drawing Rate-of-Change Graphs 169
y A Detailed Look at the Slope Graph
6 When a graph has labeled tick marks on both the horizontal and the ver-
5 tical axes or an equation for the graph is known, it is possible to estimate
4
the values of slopes at certain points on the graph. However, it would be
tedious to calculate the slope graphically or numerically for every point
3
on the graph. Instead, we calculate the slope at a few points including
2 inflection points and relative extremes.
1 Consider again a graph with a maximum, an inflection point, and
A B C
x a minimum similar to the one we saw at the beginning of this section
1 2 3 4 5 6 in Figure 3.1. Figure 3.19 shows such a graph, but this time the graph
FIGURE 3.19 has labeled tick marks on both the horizontal and the vertical axes.
We know that the slope graph crosses the horizontal axis at A and C
and that a minimum occurs on the slope graph below the horizontal axis at B. Before,
we did not know how far below the axis to draw this minimum. Now that there is a
numerical scale on the axes, we can graphically estimate the slope at the inflection
point and use that estimate to help us sketch the slope graph.
By drawing the tangent line at B and estimating its slope, we find that the
minimum of the slope graph is approximately 1.4 units below the horizontal axis.
(See Figure 3.20.)
y
8 1
7
6 1
2.7
5 1.4
4
3
1.8
2
1 1
x
1 A 3 B 4 C 6
FIGURE 3.20
If we estimate the slopes at two additional points, say at x 1 and x 6, then we can
produce a fairly accurate sketch of the slope graph. Table 3.1 shows a list of estimated
slope values. Plotting these points and sketching the slope graph give us the graph in
Figure 3.21.
dy
TABLE 3.1
dx
6 x 1 A B C 6
4 Slope 2.7 0 1.4 0 1.8
0 x If we do not have a continuous curve but have only a scatter plot, then we can
1 A 3 B C 6 sketch a rate-of-change graph after first sketching a smooth curve that fits the
–2 scatter plot. Then we can draw the slope graph of that smooth curve. We show this
FIGURE 3.21 technique in Example 4.
170 CHAPTER 3 Determining Change: Derivatives
EXAMPLE 4 Using a Curve Through Data to Sketch a Slope Graph
Population In 1797, the Lorenzo Carter family built a cabin on Lake Erie where
today the city of Cleveland, Ohio, is located. Table 3.2 gives population data for
Cleveland from 1810 through 1990.
TABLE 3.2
Year Population Year Population
1810 57 1910 560,663
1820 606 1920 796,841
1830 1076 1930 900,429
1840 6071 1940 878,336
1850 17,034 1950 914,808
1860 43,417 1960 876,050
1870 92,829 1970 750,879
1880 160,146 1980 573,822
1890 261,353 1990 505,616
1900 381,768
(Source: U.S. Department of Commerce, Bureau of the Census.)
a. Sketch a smooth curve representing population. Your curve should have no
more inflection points than the number suggested by the scatter plot.
b. Sketch a graph representing the rate of change of population.
Solution
a. Draw a scatter plot of the population data, and sketch the smooth curve. (See
Figure 3.22.)
b. The population graph is fairly level in
the early 1800s, so the slope graph will
begin near zero. The smooth sketched
Population
curve increases during the 1800s and
(thousands)
early 1900s until it peaks in the 1940s.
1000 Thus the slope graph will be positive
until the mid-1940s, at which time it
800 will cross the horizontal axis and be-
come negative. Population decreased
600 from the mid-1940s onward.
400 There appear to be two inflection
points. The point of most rapid
200 growth appears near 1910, and the
point of most rapid decline appears
Year near 1975. These are the years in which
1800 1850 1900 1950 1990 the slope graph will be at its maximum
FIGURE 3.22 and at its minimum, respectively.
3.1 Drawing Rate-of-Change Graphs 171
By drawing tangent lines at 1910 and at 1975 and estimating their slopes, we find
that population was increasing by approximately 22,500 people per year in 1910
and was decreasing by about 12,500 people per year in 1975. See Figure 3.23.
Population − 250 thousand
(thousands) Slope ≈
20 years
= −12.5 thousand
1000 people per
20 years year
800
250 thousand
450 thousand
600 Slope ≈
20 years 450 thousand
= 22.5 thousand
400 people per
year 20 years
200
Year
1800 1850 1900 1950 1990
FIGURE 3.23
Now we use all the information from this analysis to sketch the slope graph shown in
Figure 3.24.
Rate of change
of population
(thousands of
people per year)
25
20
15
10
5
Year
-5 1810 1850 1900 1950 1990
-10
FIGURE 3.24 ●
Of course, if we had a formula for the graph, we could estimate the slope numerically at
a few points instead of graphically estimating it with tangent lines. Even so, we still need
to understand curvature and horizontal-axis intercepts to sketch the rate-of-change
graph adequately.
Points of Undefined Slope
It is possible for the graph of a continuous function to have a point at which the slope
does not exist. Remember that the slope of a tangent line is the limit of slopes of ap-
proximating secant lines and that we should be able to use secant lines through points
either to the left or to the right of the point at which we are estimating the slope to find
172 CHAPTER 3 Determining Change: Derivatives
this limit. Recall that if the limits from the left and from the right are not the same, then
the derivative does not exist at that point. We depict the nonexistence of the derivative
at such points on the slope graph by drawing an open circle on each piece of the slope
graph. This is illustrated in Figure 3.25a.
Also, points at which the tangent line is vertical (that is, the slope calculation re-
sults in a zero in the denominator) are considered to have an undefined slope. The
graph of one such function is shown in Figure 3.25b.
dy
dy
dx
y dx
y
x
A
x x x
A A A
Function graph: Slope graph: Function graph: Slope graph:
No derivative at A A hole at A Vertical tangent at A Break at A
FIGURE 3.25a and no derivative at A
FIGURE 3.25b
If a function is not continuous at a point, then its slope is undefined at that point even
when the slope from the left and the slope from the right are the same.
dy
y dx
x x
A A
Function graph: Slope graph:
No derivative at A A hole at A
FIGURE 3.26
Most of the time when there is a break in the slope graph, it is because the original
function is piecewise, as shown in Figures 3.25a and 3.26. You should be careful when
drawing slope graphs of piecewise continuous functions. Figure 3.25b shows a
smooth, continuous function with a point at which the derivative is undefined. We do
not often encounter such phenomena in real-life applications, but they can happen.
3.1 Drawing Rate-of-Change Graphs 173
9. y 10. y
3.1 Concept Inventory
• Slope graph, rate-of-change graph, derivative graph
• Increasing functions have positive slopes x
• Decreasing functions have negative slopes A
x
• Maxima and minima have zero slopes B
• Inflection points are maxima or minima of slope
11. a. Sketch a graph that is increasing with increasing
graph or points of undefined slope
slopes. Also sketch its slope graph.
b. Sketch a graph that is increasing with decreasing
slopes. Also sketch its slope graph.
3.1 Activities
12. a. Sketch a graph that is decreasing with decreasing
Getting Started slopes. Also sketch its slope graph.
In Activities 1 through 10, list as many facts as you can b. Sketch a graph that is decreasing with increasing
about the slopes of the graphs. Then, on the basis of slopes. Also sketch its slope graph.
those facts, sketch the slope graph of each function.
1. y 2. y
Applying Concepts
13. Phone Bill The graph shows the average monthly
cell phone bill in the United States between 1987
and 2002.
x Average bill
A x
(dollars)
97
3. y 4. y
x x 30 Year
1987 1990 1993 1996 1999 2001
5. y 6. y (Source: Based on data from the Cellular Telecommunication
and Internet Association.)
a. Estimate and record the slopes of tangent lines
for the input values shown below.
x
x A B CD Year 1991 1993 1997 1999 2001
Slope of
7. y 8. y tangent line
b. Use the information in part a to sketch an
x accurate rate-of-change graph for the average
monthly cell phone bill. Label the axes with units
x
as well as values.
174 CHAPTER 3 Determining Change: Derivatives
14. Population The graph shows the population of 16. Fuel The graph shows the average annual fuel
Iowa between 1990 and 1999. consumption of vehicles in the United States
between 1970 and 1995.
Population
(thousands)
Average fuel consumption
(gallons per year)
2850
820
2830 790
760
2810
730
2790 700
670
Year
2770 Year 1970 1975 1980 1985 1990 1995
1990 1992 1994 1996 1998
(Source: Based on data from Statistical Abstract, 1998.)
(Source: Based on data from U.S. Bureau of the Census.)
a. Sketch tangent lines for the input values of 1992, a. Sketch tangent lines for the input values of
1994, 1996, and 1998. Also sketch a tangent line 1975, 1980, 1985, and 1990. Also sketch tangent
for a point near 1990. Estimate the slopes of these lines for a point near 1970 and a point near
five tangent lines. 1995. Estimate the slopes of these six tangent
lines.
b. Use the information in part a to sketch an accu-
rate rate-of-change graph for the population of b. Use the information in part a to sketch an
Iowa. Label the axes with units as well as values. accurate rate-of-change graph for the average
annual fuel consumption. Label the axes with
15. AIDS The graph shows the cumulative number of
units as well as values.
AIDS cases between 1985 and 2001 diagnosed in the
United States since 1984. 17. Membership The graph gives the membership in
a campus organization during its first year.
Cumulative AIDS
cases since 1985
(millions) Members
800 240
200
600
160
400
120
200 80
0 Year 40
1985 1990 1995 2000
(Source: Based on data from Centers for Disease Control.) Sep Oct Nov Dec Jan Feb Mar Apr May
a. Sketch tangent lines for the input values of 1990, a. Estimate the average rate of change in the
1995, 1997, and 2000. Also estimate the slope of membership during the academic year.
a tangent line for a point near 1985.
b. Estimate the instantaneous rates of change in
b. Use the information in part a to sketch an accu- mid-September, November, February, and April.
rate rate-of-change graph for the cumulative
number of AIDS cases diagnosed in the United c. On the basis of your answers to part b, sketch a
States. Label the axes with units as well as values. rate-of-change graph. Label the units on the axes.
3.1 Drawing Rate-of-Change Graphs 175
d. The membership of the organization was grow- j(t)
ing most rapidly in September. Not including Inmates
that month, when was the membership growing
most rapidly? What is this point on the member- 1,000
ship graph called?
900
e. Why was the result of the calculation in part a of 800
no use in part c?
700 t
18. Police Calls The scatter plot depicts the number Years
of calls placed each hour since 2 A.M. to a sheriff’s 600
0 5 10 since
department. 1990
a. Sketch a smooth curve through the scatter plot (Source: Based on data from Washoe County Jail, Reno,
Nevada.)
with no more inflection points than the number
suggested by the scatter plot. a. Sketch the slope graph of j. (Hint: Estimate j(t)
near t 0 and t 4.5 in order to sketch the
Calls slope graph accurately.)
each hour b. Label both the horizontal and the vertical axes.
80 20. Price The graph shows cattle prices (for choice
70
450-pound steer calves) from October 1994
through May 1995. Input is the number of months
60 since October 1994.
50 p(m)
Cattle price
40
(dollars per pound)
30
1.00
20
10 0.95
Hours
0.90
0 2 4 6 8 10 12 14 16 18 20 22 24
since
2 A.M. 0.85
m
(Source: Sheriff’s Office of Greenville County, South Carolina.)
0.80 Months
b. At what time(s) is the number of calls a since
0.75
minimum? a maximum? 0 1 2 3 4 5 6 7
October
1994
c. Are there any other times when the graph (Source: Based on data from the National Cattleman’s
appears to have a zero slope? If so, when? Association.)
d. Estimate the slope of your smooth curve at any a. Sketch a slope graph of p. (Hint: Estimate p(m)
inflection points. near m 0 and m 2.5 in order to sketch the
slope graph accurately.)
e. Use the information in parts a through d to
sketch a graph depicting the rate of change of b. Label the horizontal and vertical axes.
calls placed each hour. Label the units on both 21. Profit The accompanying graph depicts the
axes of the rate-of-change graph. average monthly profit for Slim’s Used Car Sales for
19. Jails The capacity of jails in a southwestern state the previous year.
has been increasing since 1990. The average daily a. Estimate the average rate of change in Slim’s
population of one jail between 1990 and 2000 is average monthly profit if the number of cars he
shown in the graph. sells increases from 40 to 70 cars.
176 CHAPTER 3 Determining Change: Derivatives
Average a. Use the graph to estimate carefully the rate of
monthly profit change in deaths of males due to lung cancer in
(dollars) 1940, 1960, and 1980.
30,000 b. Use this information to sketch an accurate rate-
of-change graph for deaths of males due to lung
20,000 cancer.
10,000 c. Label the units on both axes of the derivative
Cars graph.
10 20 30 40 50 60 70 80 90 100 110 120 sold
-10,000 In Activities 23 through 26, indicate the input values for
which the graph has no derivative. Explain why the
-20,000 derivative does not exist at those points. Sketch a
derivative graph for each of the function graphs.
b. Estimate the instantaneous rates of change at 20,
40, 60, 80, and 100 cars. 23. y
c. On the basis of your answers to part b, sketch a
rate-of-change graph. Label the units on the axes.
d. For what number of cars sold between 20 and
-2 -1
100 is average monthly profit increasing most x
rapidly? For what number of cars sold is average 1 2 3 4 5
monthly profit decreasing most rapidly? What is
the mathematical term for these points?
e. Why was the result of the calculation in part a of
no use in part c? 24. y
22. Mortality The graph shows deaths of males due to
different types of cancer.
80
Lung
Rate per 100,000 male population
60 x
0 3 6 8
25. y
40
Stomach
Colon and Rectum
20 Prostate
Pancreas
Liver
0 Year x
1930 1940 1950 1960 1970 1980 1990 0 2 3 5
3.1 Drawing Rate-of-Change Graphs 177
26. y 31. The figure below shows a graph of the function
p(m) m m.
p(m)
8
0 x
A B
0 m
0 5
a. Sketch a slope graph of p.
27. Sketch the slope graph of a function f with input t
b. Find a formula for the slope graph. Compare a
that meets these criteria: f(2) 5, the slope is
graph of this formula with the graph you sketched
positive for t 2, the slope is negative for t 2,
in part a. (Hint: Multiply the numerator and
and f(2) does not exist.
denominator of the secant line slope formula by
m h m.)
28. Sketch the slope graph of a function g with input x
that meets these criteria: g(3) does not exist, 32. The figure below shows a graph of the function
1
g(0) 4, g(x) 0 for x 3, g is concave down k(x) x x .
for x 3, g(x) 0 for x 3, g is concave up for
k (x)
x 3, lim g(x) → , and lim g(x) → .
x→3 x→3 5
29. Construct the graphs of a function h and its slope
h, with input x, such that h(1) is significantly
different from the percentage rate of change of h at
x 1. x
–5 –1 1 5
1
30. The figure shows a graph of the function q(t) t .
–5
q (t)
a. Sketch a slope graph of k.
b. Find a formula for the slope graph. [Hint:
Multiply the numerator and denominator of the
t secant line slope formula by x(x h).] Compare
a graph of this formula with the graph you
sketched in part a.
Discussing Concepts
33. What elements of a function graph are of specific
a. Sketch a slope graph of q. importance when sketching a rate-of-change graph
for that function? Explain why these elements are
b. Find a formula for the slope graph. [Hint: important.
Multiply the numerator and denominator of the
secant line slope formula by t(t h).] Compare 34. Why is it important to understand curvature and
a graph of this formula with the graph you horizontal-axis intercepts in order to adequately
sketched in part a. sketch a rate-of-change graph?
178 CHAPTER 3 Determining Change: Derivatives
3.2 Simple Rate-of-Change Formulas
By now you should have a thorough understanding of the concept of instantaneous
rate of change as the slope of a line tangent to a curve at a point, where that slope is
defined as a limiting value of slopes of secant lines. We now rely on this conceptual
understanding in order to present formulas for rapid calculation of rates of change.
You already know two rate-of-change formulas from our study of linear
functions in Chapter 1. A horizontal line has slope zero, and a nonhorizontal line of
the form y ax b has slope a. We know that the rate of change, or derivative, of a
line is its slope, so we can state the following derivative formulas:
Constant Rule for Derivatives
dy
If y b, then 0.
dx
Derivative of a Linear Function
dy
If y ax b, then a.
dx
EXAMPLE 1 Finding the Derivative of a Linear Function
Cricket’s Chirping The frequency of a cricket’s chirp is affected by air temperature
and can be modeled by
C(t) 0.212t 0.309 chirps per second
when the temperature is t°F, 50 t 85. Write a formula for the rate of change of
a cricket’s chirping speed with respect to a change in temperature.
Solution
The frequency of a cricket’s chirp is changing by
C(t) 0.212 chirps per second per degree Fahrenheit
when t is the temperature between 50°F and 85°F. ●
The Simple Power Rule
Next, consider quadratic and cubic functions. In Section 2.4, we determined that the
dy
rate-of-change formula for y x2 is dx 2x. In Activity 13 of that same section, you
dy
were asked to show that the rate-of-change formula for y x3 is dx 3x2. Note that
these are two special cases of one of the most important rules that we use for quickly
finding derivative formulas, the Simple Power Rule.
3.2 Simple Rate-of-Change Formulas 179
Simple Power Rule for Derivatives
dy
If y xn, then nxn1, where n is any nonzero real number.
dx
The Constant Multiplier and the Sum and Difference Rules
In order to find derivative formulas for any polynomial, we need two rules in addition
to the Simple Power Rule. The first derivative rule we illustrate is the Constant
Multiplier Rule. Each of the figures in Figure 3.27 shows the graph of a function and
a graph of a constant multiple of that function, and the slope graphs of the function
and the constant multiple of that function.
dy
y dx
g = 3f f g' f'
x
x
(a) (b)
dy
y dx
h
1
k= 2
h
x
h'
k'
x
(c) (d)
FIGURE 3.27
The effect of a constant multiplier is to amplify the output if the multiplier is
greater than 1 or to diminish the output if the multiplier has magnitude between 0
and 1. The behavior of the rate of change of the function will be amplified or dimin-
ished by the same factor.
180 CHAPTER 3 Determining Change: Derivatives
If the constant multiplier is negative, it has the additional effect of reflecting the
function over the horizontal axis. The rate-of-change function is likewise reflected, as
illustrated in Figures 3.28a and b.
dy
y dx
f
f'
x x
g'
g = −2 f
(a) (b)
FIGURE 3.28
The graphs in Figures 3.27 and 3.28 are basic illustrations of the following general
rule:
Constant Multiplier Rule for Derivatives
dy
If y kf(x), then kf(x).
dx
y
f +g g
This rule enables us to calculate quickly the rate-of-change formula for a function
3
such as y 5x 4: leave the 5 alone, and apply the Simple Power Rule to x 4. This
f dy
2 process gives dx 5(4x 3) 20x 3.
The final rules needed for rapid calculation of the rate-of-change formula for any
1
polynomial are the Sum and Difference Rules.
x
−2 −1 1 2
−1
Concept Development: Sum and Difference Rules We present two graphical
illustrations of the Sum and Difference Rules. Figure 3.29 illustrates the graphs of
f(x) x, g(x) x2 and the graph of (f g)(x) x x 2, the function that is the
FIGURE 3.29 sum of f and g.
3.2 Simple Rate-of-Change Formulas 181
1
The minimum of the sum function f g appears to occur at x 2, indicating
1
that the slope graph of the sum function crosses the horizontal axis at x 2 .
Moving from left to right along the horizontal axis, the outputs of f g are decreasing
1 1
before x 2 and increasing again after x 2. Thus, the sum function slope
1 1
graph is negative to the left of x 2 and positive to the right of x 2 . Further,
note that the graph of the sum function has the same basic shape as the function f, ex-
cept that it is shifted.
We now investigate the basic shape and magnitude of the slope graph by looking at
a few arbitrarily chosen inputs (x 2, x 3, and x 7). By first using the Linear
Function Rule and the Simple Power Rule to obtain the slopes of f and g, we complete
the second and third rows of Table 3.3. Numerically investigating the slope of f g at
the inputs given in the first row of Table 3.3 yields the estimates in the fourth row.
TABLE 3.3
1
x 2 2 0 3 7
df
1 1 1 1 1 1
dx
dg
2x 4 1 0 6 14
dx
d(f g)
1 2x 3 0 1 7 15
dx
dy
dx
Figure 3.30 shows the slope graphs of f, g, and the sum function f g (in teal).
(f + g)'
The graph of the sum function was obtained by plotting the outputs in the fourth row
g'
3 of the table and connecting them with a smooth curve.
Note in both Table 3.3 and Figure 3.30, that the slopes of the sum function f g
2
at a specific point can be obtained by summing the slopes of the functions f and g. A
1 similar result is obtained if we investigate the difference function f g. The slope of
f' the difference function f g at a certain input can be obtained by subtracting the
x
−2 −1 1 2 slope of g from the slope of f at that same input. See Figures 3.31a and b.
−1
dy
y dx
g g'
FIGURE 3.30 f
1 1
f'
x x
−2 −1 1 2 −2 −1 1 2
−1 −1
−2 −2
f −g (f − g)'
−3 −3
(a) (b)
FIGURE 3.31
182 CHAPTER 3 Determining Change: Derivatives
The second graphical illustration of the Sum and Difference Rules specifically
illustrates the differences of two simple power functions, k(x) x 3 and h(x) x 2.
Figure 3.32 shows graphs of these two functions and their difference function
(k h)(x) x 3 x 2.
y
k h
k −h
x
FIGURE 3.32
2
The difference function k h appears to have a minimum at x 3 and a
maximum at x 0. Therefore, the slope graph crosses the horizontal axis at these
input values. Moving from left to right along the horizontal axis, the difference
2
function is increasing before x 0 and after x 3 and is decreasing between these
two inputs. Thus the slope graph is positive to the left of x 0 and to the right of
2
x 3 and is negative between these values. Numerically evaluating the limiting value
1
of the slope of k h at x 1, x 3, and x 2 yields the estimates in the second
row of Table 3.4. Using these numerical estimates as additional points on the graph,
we sketch the slope graph of k h. See Figure 3.33.
dy
dx
(k − h)'
x
TABLE 3.4
1
x 1 3 2
d(k h)
dx 5 0.33333 8
FIGURE 3.33
Table 3.5 and Figure 3.34 illustrate the relationship of the slopes of the functions
k and h and the difference function k h. Similar to the result we found with the sum
3.2 Simple Rate-of-Change Formulas 183
function, the slopes of the difference function are the differences of the slopes of the
two individual functions.
dy
dx
k' h'
TABLE 3.5
(k − h)' = k' − h'
1 2
x 1 0 3 3 2
x
dk 1 4
dx 3x
2 3 0 12
3 3
dh 2 4
dx 2x 2 0 3 3 4
d(k h) 1
dx 5 0 3 0 8
FIGURE 3.34
We state the following general rule:
The Sum and Difference Rules for Derivatives
dj df dg
If j(x) f(x) g(x), then .
dx dx dx
dj df dg
If j(x) f(x) g(x), then .
dx dx dx
The Sum and Difference Rules also apply to sums and/or differences of more than
two functions. With the rules we now have, we can find the rate-of-change formula for
any polynomial function. For example, if
p(x) 3.22x3 0.15x2 9.98x 30
we use the Constant Multiplier and the Simple Power Rules to find the derivative for-
mula for each term, and then we combine the terms using the Sum and Difference
Rules. The rate-of-change formula for p is
p(x) 3.22(3x 2) 0.15(2x) 9.98 0 9.66x 2 0.3x 9.98
Example 2 is an application of the Sum and Difference Rules.
EXAMPLE 2 Applying Derivative Rules
Maintenance Costs Table 3.6 gives the average yearly maintenance costs per
vehicle for 15,000 miles of operation in the United States from 1993 through 2000.
TABLE 3.6
Year 1993 1994 1995 1996 1997 1998 1999 2000
Maintenance costs 2.4 2.5 2.6 2.8 2.8 3.1 3.6 3.9
(cents per mile per vehicle)
(Source: Bureau of Transportation Statistics.)
184 CHAPTER 3 Determining Change: Derivatives
Find a quadratic model for the data. Use it to approximate how rapidly maintenance
costs were increasing in 1998.
Solution
A quadratic model for the data is
g(t) 0.0304t2 0.00417t 2.446 cents per mile per vehicle
gives the maintenance cost for a vehicle where t is the number of years after 1993,
0 t 7. Applying the Sum and Difference, Power, Constant Multiplier, and
Constant Rules, we find that the derivative of g with respect to t is
dg
0.0304(2t) 0.00417(1) 0
dt
so
dg
0.0608t 0.00417 cents per mile per vehicle per year
dt
gives the rate of change in the maintenance costs, t years after 1993, 0 t 7.
Evaluating the derivative at t 5 gives 0.2998 cent per mile per vehicle per year
as the rate of change of maintenance costs in 1998. Thus we estimate that in 1998, the
average maintenance cost per vehicle was increasing at a rate of approximately 0.30
cent per mile per year. ●
In summary, here is a list of the rate-of-change formulas you should know.
Simple Derivative Rules
Rule Name Function Derivative
dy
Constant Rule yb 0
dx
dy
Linear Function Rule y ax b a
dx
dy
Power Rule y xn nxn1
dx
dy
Constant Multiplier Rule y kf(x) kf(x)
dx
dy
Sum Rule y f(x) g(x) f (x) g(x)
dx
dy
Difference Rule y f (x) g(x) f (x) g(x)
dx
3.2 Simple Rate-of-Change Formulas 185
For each function in Activities 7 through 18,
3.2 Concept Inventory a. identify which simple derivative rules apply,
• Derivative formulas b. write the formula for the derivative of the function.
• For constants a, b, and n: 7. f(x) x 5 8. f(x) x 4
If y b, then y 0. 9. f(x) 3x 3 10. f(x) 0.5x 2
If y ax b, then y a. 11. f(x) 5x 12. f(x) 7x
If y x n, then y nx n1. 13. f(x) 0.2 14. f(x) 35
If y kf(x), then y kf(x).
15. y 12x 13
If y f(x)g(x), then y f(x)g(x).
16. f(x) 7x2 9.4x 12
17. y 5x 3 3x 2 2x 5
3.2 Activities
18. y 3.2x 3 6.1x 5.3
Getting Started
For each of the functions whose graphs are given in For each function in Activities 19 through 26,
Activities 1 through 6, first sketch the slope graph and a. rewrite the expression in power notation if
then give the slope equation. necessary, and
1. y 2. y b. write the derivative formula for the function.
4 1 1
y = −x 2 + 4 19. f(x) (Hint: n xn.)
y = 2 − 7x x3 x
3
2 1 9
2 20. f(x) 21. y
x 1 x3 x2
1
x 3x 2
-3 -2 -1 1 2 3 22. f(x)
x
3x 2 1
23. j(x)
y x
3. y 4.
15 (Hint: Rewrite as two separate terms.)
4 10 4x 2 19x 6
y = x3 24. j(x)
3 5 x
x (Hint: Rewrite as three separate terms.)
2 y = x4 -3 -2 -1 1 2 3
-5 25. f(x) x (Hint: Rewrite as x x1/2.)
1
-10 26. h(x) 17 8x
x
-2 -1 1 2 -15
Applying Concepts
5. y 6. y 27. ATM Fee The average ATM transaction fee
charged by U.S. banks between 1996 and 1999 can
3
2 y = 32 y = x12
be modeled by the equation A(t) 0.1333t
0.17 dollars t years after 1990.
1 1 (Source: Based on data from the U.S. Public Interest Research
x Group.)
x -4 -2 2 4
-1
-2 -1 1 2 a. Write the derivative formula for A.
-1 b. Estimate the transaction fee in 2000.
-3
c. How quickly was the fee changing in 1999?
186 CHAPTER 3 Determining Change: Derivatives
28. Population The population of Hawaii between a. The graph is not as steep at 1:30 P.M. as it is at 7
1970 and 1990 can be modeled by A.M.
P(t) 15.48t 485.4 thousand people b. The slope of the tangent line at 7 A.M. is
10°F per hour.
t years after 1950. A graph of this model is given. c. The instantaneous rate of change of the temper-
P ature at noon is 2°F per hour.
Population d. At 4 P.M. the temperature is falling by
(thousands) 4.4°F per hour.
1250
30. Population The projected number of Americans
1125 age 65 or older for the years 1995 through 2030 can
be modeled by the equation
1000 N(x) 0.03x2 0.315x 34.23 million people
875 where x is the number of years after 2000.
(Source: Based on data from John Greenwald, “Elder Care:
t Making the Right Choice,” Time, August 30, 1999, p. 52.)
750
1970 1975 1980 1985 1990 Year a. What is the projected number of Americans
(20) (25) (30) (35) (40)
65 years of age and older in 1995? in 2029?
(Source: Based on data from George T. Kurian, Datapedia of the
in 2030?
United States, 1790–2000, Latham, MD: Bernan Press, 1994.)
a. Write the formula for P. b. What is the rate of change of the projected
number in 1996? in 2029?
b. How many people lived in Hawaii in 1970?
c. How quickly was Hawaii’s population changing c. Find the percentage rate of change in the
in 1990? projected number in 2029.
d. The Census Bureau predicts that in 2030, 20.1%
29. Temperature The graph shows the temperature of the U.S. population will be 65 years of age or
values (in °F) on a typical May day in a certain older. Use this prediction and one of the
midwestern city. unrounded answers to part a to estimate the total
Temperature U.S. population in 2030.
(°F)
31. Births The number of live births to U.S. women
80 45 years and older between 1950 and 2000 can be
modeled by the equation
60
50
B(x) 0.2685x3 15.6x2
40 t 94.684x 5378.03 births
-6
Hours
0 8
(6 A.M.) (noon) (8 P.M.) where x is the number of years since 1950.
(Source: Based on data from www.infoplease.com. Accessed
The equation of the graph is 9/24/02.)
Temperature 0.8t2 2t 79°F a. Was the number of live births rising or falling in
1970? in 1995?
where t is the number of hours since noon. Use the
derivative formula to verify each of the following b. How rapidly was the number of live births rising
statements. or falling in 1970? in 1995?
3.2 Simple Rate-of-Change Formulas 187
32. Study Time The graph represents a test grade (out a. Find a formula for a typical man’s metabolic rate.
of 100 points) as a function of hours studied. b. Write the derivative of the formula in part a.
G(t) c. What does the derivative in part b tell you about
Points a man’s metabolic rate if that man weighs 110
pounds? if he weighs 185 pounds?
100
34. Profit An artisan makes hand-crafted painted
75 benches to sell at a craft mall. Her weekly revenue and
costs (not including labor) are given in the table.
50
Number of benches Weekly revenue Weekly cost
25 sold each week (dollars) (dollars)
1 300 57
t
0
2 4 6 8 10 12 14 16 Hours 3 875 85
5 1375 107
The equation of the graph is 7 1750 121
G(t) 0.044t3 0.918t2 38.001 points 9 1975 143
11 1950 185
after t hours of study. Use the equation to verify
each of the following statements. 13 1700 213
dG
a. dt 1.704 points per hour when t 1 hour.
a. Find models for revenue, cost, and profit.
b. G(4) 5.232 points per hour. b. Write the derivative formula for profit.
c. The slope of the tangent line when t 15 hours c. Find and interpret the rates of change of profit
is approximately 2 points per hour. when the artisan sells 6, 9, and 10 benches,
33. Metabolic Rate The table shows the metabolic respectively.
rate of a typical 18- to 30-year-old male according to d. What does the information in part c tell you
his weight. about the number of benches the artisan should
produce each week?
35. Sales The accompanying table gives revenue from
Metabolic rate new-car sales and associated advertising expenditures
Weight (pounds) (kilocalories per day) for franchised new-car dealerships in the United
88 1291 States for selected years between 1980 and 2000.
110 1444
Advertising expenses Revenue
125 1551
Year (billions of dollars) (billions of dollars)
140 1658
1980 1.2 130.5
155 1750
1985 2.8 251.6
170 1857
1990 3.7 316.0
185 1964
1995 4.6 456.2
200 2071
1998 5.3 546.3
(Source: L. Smolin and M. Grosvenor, Nutrition:
2000 6.4 646.8
Science and Applications, Philadelphia, PA:
Saunders College Publishing, 1994.) (Source: Statistical Abstract, 2001.)
188 CHAPTER 3 Determining Change: Derivatives
a. Find a model that describes revenue as a function a. Write the formula for the profit made from the
of advertising expenditures. sale of one storm window when Windolux is
b. Write the formula for the derivative of the producing x windows each hour.
equation in part a. b. Write the formula for the rate of change of profit.
c. Use your equations from parts a and b to
c. What is the profit made from the sale of a win-
estimate the revenue and to find how rapidly the
dow when Windolux is producing 80 windows
revenue was changing when $5 billion was spent
each hour?
on advertising.
d. What was the percentage rate of change in the d. How rapidly is profit from the sale of a window
revenue when $5 billion was spent on advertising? changing when 80 windows are produced each
hour? Interpret your answer.
36. Costs Production costs (in dollars per hour) for a
certain company to produce between 10 and 90 38. Sales A publishing company estimates that
units per hour are given in the table. when a new book by a best-selling American au-
thor first hits the market, its sales can be predicted
Cost
by the equation n(x) 68.95x, where n(x) rep-
Units (dollars per hour)
resents the total number (in thousands) of copies
10 150 of the book sold in the United States by the end of
20 200 the xth week. The number of copies of the book
30 250 sold abroad by the end of the xth week can be
modeled by
40 400
50 750 a(x) 0.125x thousand copies of the book
60 1400 a. Write the formula for the total number of copies
70 2400 of the book sold in the United States and abroad
by the end of the xth week.
80 3850
90 5850 b. Write the formula for the rate of change of the
total number of books sold.
a. Consider the cost for producing 0 units to be $0.
c. How many books will be sold by the end of the
Include (0, 0) in the data and find a cubic model
first year (that is, after 52 weeks)?
for production costs.
b. Convert the model in part a to one for the aver- d. How rapidly are books selling at the end of the
age cost per unit produced. first year? Interpret your answer.
c. Find the slope formula for average cost.
d. How rapidly is the average cost changing when Discussing Concepts
15 units are being produced? 35 units? 85 units?
Interpret your answers. 39. Use your knowledge of the shape and end behavior
of the graph of a cubic function to explain why the
37. Profit The managers of Windolux, Inc., have mod- slope graph of a cubic function is the graph of a
eled some cost data and found that if they produce x quadratic function. Use this argument to explain
storm windows each hour, the cost (in dollars) to why the rate-of-change formula for a cubic function
produce one window is given by the function is the formula for a quadratic function.
49.6
C(x) 0.015x2 0.78x 46 ,x 0 40. Use the simple derivative rules presented in this sec-
x
tion to explain why the rate-of-change formula for a
Windolux sells its storm windows for $175 each. cubic function of the form y ax3 bx2 cx d
(You may assume that every window made will be is the formula for a quadratic function. Write your
sold.) explanation in paragraph form.
3.3 Exponential and Logarithmic Rate-of-Change Formulas 189
3.3 Exponential and Logarithmic Rate-of-Change
Formulas
In Section 3.2 we developed some general rules for rate-of-change formulas. In this
section we will continue using those rules and develop others.
Exponential Rules
Our next formulas involve derivatives of exponential functions. Because the proof of
these rules is beyond the scope of this book, we explore the functions graphically and
numerically to develop an understanding of the behavior of the derivative of the
function before stating the general derivative formula. We begin our exploration with
y
the function y e x . First, we consider the concavity and end behavior of this
function (see Figure 3.35).
4
• y e x approaches 0 as x decreases without bound, so the slope graph of y e x
3 seems to become horizontal (i.e., y → 0 as x → ).
2
• y e x increases without bound as x increases without bound (x → ) and the
graph of the function is concave up and has no vertical asymptotes. The graph of
1 y e x seems to become more and more vertical; so the slopes are increasing
without bound (i.e., y → as x → ).
x
−2 −1 1 2 Next, to obtain an idea of magnitude, we numerically investigate the slope at a few
points. Table 3.7 shows function values and slope values (rounded to three decimal
FIGURE 3.35 places) for several inputs.
TABLE 3.7
x 2 0 1 3
y ex 0.135 1.000 2.718 20.086
dy
y 0.135 1.000 2.718 20.086
dx
This function is surprising in that the rate-of-change values are precisely the same
as the function values. This function is its own derivative! In other words, if y e x,
dy
then dx y e x. The slope graph of y e x coincides with the graph of the original
function.
Derivative of e x
dy
If y ex, then ex.
dx
Does this rule apply to exponential functions that have bases different from e? In
dy
other words, if y b x, is dx b x? Consider the functions y 2x and y 3x. Graph-
ically, the descriptions of end behavior and curvature lead to the same conclusions
190 CHAPTER 3 Determining Change: Derivatives
about the shape of the derivative graphs for these exponential functions as did the analy-
sis of the shape and end behavior of y e x. (See Figures 3.36a and b.) Any differences
that occur should appear in a numerical investigation of the magnitude of the slopes.
dy
y dx
d (3 x )
y = 3x y' =
dx
4 4
y = 2x
3 3
d (2 x )
2 2 y' =
dx
1 1
x x
−2 −1 1 2 −2 −1 1 2
FIGURE 3.36 (a) (b)
TABLE 3.8
We begin by exploring the derivative of y 2x. See Table 3.8 (values are rounded
dy to five decimal places for convenience). It is obvious from the table that the derivative
x y 2x y
dx of y 2x is not y 2x . If we use the definition of derivative for the function
2 0.25 0.17329 f(x) 2x, we obtain
0 1 0.69315 2x
xh
f (x) lim 2
1 2 1.38629 h→0
h
3 8 5.54518 You may recall that 2xh 2x2h. Using this fact, we rewrite the derivative formula as
2x2h 2x 2h 1
f (x) lim
h→0 h
lim 2x
h→0 h
Because the term 2x is not affected by h approaching 0, we treat it as a constant:
2h 1
f (x) 2x lim
h→0 h
This formula indicates that the derivative of y 2x is 2x times a constant. In Table 3.9
we numerically estimate the limiting value of the multiplier lim 2h 1.
h→0 h
TABLE 3.9
2h 1 2h 1
h → 0 h → 0
h h
0.1 0.717735 0.1 0.669670
0.01 0.695555 0.01 0.690750
0.001 0.693387 0.001 0.692907
0.0001 0.693171 0.0001 0.693123
0.00001 0.693150 0.00001 0.693145
0.000001 0.693147 0.000001 0.693147
Limit 0.693147 Limit 0.693147
3.3 Exponential and Logarithmic Rate-of-Change Formulas 191
It may seem that the multiplier 0.693147 is arbitrary because it is not a familiar
number, but that is not the case. You should verify that 0.693147 ln 2. In fact, it
can be proved that the limit of the multiplier in the derivative formula for f(x) 2x
is ln 2. We state the formula for the derivative of y 2x as y (ln 2)2x A similar
exploration suggests that the formula for the derivative of y 3x is y (ln 3)3x.
d d
The two derivative formulas dx(2x) (ln 2)2x and dx(3x) (ln 3)3x are special
cases of the general derivative formula for exponential functions. The derivative of
d
y bx is y (ln b)b x if b 0. In fact, the rule dx(e x) e x is also a special case of
this formula. You will be asked to verify this fact in the activities.
Derivative of b x
dy
If y b x, where the real number b 0, then (ln b)b x.
dx
EXAMPLE 1 Using Exponential Derivative Formulas
Credit Cards If credit card purchases are not paid off by the due date on the
credit card statement, finance charges are applied to the remaining unpaid balance.
In July 2001, one major credit card company had a daily finance charge of 0.054%
on unpaid balances. Assume that the unpaid balance is $1 and that no new
purchases are made.
a. Find an exponential function for the balance owed d days after the due date.
b. How much is owed after 30 days?
c. Write the derivative formula for the function from part a.
d. How quickly is the balance changing after 30 days?
e. Repeat parts a through d, assuming that the unpaid balance is $2000.
Solution
a. Recall that the constant b in an exponential function f(x) ab x is
(1 percentage growth) and that the constant a is the value of f(0). Thus we
use the function
f(d) 1(1.00054d) 1.00054d dollars
to represent the balance due d days after the due date.
b. Thirty days after the due date, the balance is f (30) $1.02.
c. According to the derivative rules we have seen in this chapter, the derivative of
y b x is y (ln b)b x. Thus the derivative formula for our function f is
f (d) (ln 1.00054)1.00054d dollars per day
after d days.
d. We evaluate f (30). After 30 days, the balance is increasing at a rate of 0.0005 dollar
per day.
192 CHAPTER 3 Determining Change: Derivatives
e. If the unpaid balance is $2000, the balance-due function is
f(d) 2000(1.00054d) dollars
after d days. The amount due after 30 days is f(30) $2032.65.
According to the Constant Multiplier and the Exponential Rules, the derivative
of the balance function is
f (d) 2000(ln 1.00054)1.00054d dollars per day
After 30 days, the balance is increasing at a rate of f(30) $1.10 per day. ●
Natural Logarithm Rule
As with exponential functions, we motivate the derivative rule for the natural log
function graphically and numerically.
The natural logarithm function is not defined for negative input values or for an
input of zero. But as x approaches 0 from the right (x → 0 ), the outputs of the
natural log function decrease without bound. The tilt of the function appears to
become vertical. As x increases without bound, ln x increases without bound, but
more and more slowly. The slope never becomes zero. (See Figures 3.37a and b.)
TABLE 3.10 dy
y dx
Derivative of
2 3
x y ln x y = ln x
1 1 2 1
2.000 y' =
2 x
x 1
1 1.000 1 2 3 4
2 0.500 −1 x
1 2 3 4
4 0.250
10 0.100 (a) (b)
FIGURE 3.37
Numerically investigating the slope (to three decimal places) for a few input
values (to three decimal places) once again helps to establish magnitude. (See
Table 3.10.) Unlike the case for the exponential function, whose derivative at a certain
input value is dependent on the corresponding output value, these derivatives are
dependent on the input values. Note that each derivative value is the reciprocal of the
input value—that is, it is 1 divided by the input value. The derivative of y ln x at
1 1 1
x 2 is 2, the derivative at x 4 is 4, the derivative at x 10 is 10 , and so on.
Derivative of ln x
dy 1
If y ln x, then for positive x-values.
dx x
3.3 Exponential and Logarithmic Rate-of-Change Formulas 193
Our next example illustrates the use of all of the derivative rules presented in this
section and the previous section.
EXAMPLE 2 Using Simple Derivative Rules
Find the derivatives of the following functions:
a. f(x) 12.36 6.2 ln x b. g(t) 4e t 19
8 0.025 12y
c. m(r) 12r d. j(y) 17 1
r 12
Solution
a. Apply the Constant Rule to the first term and the Constant Multiplier and the
Natural Log Rules to the second term. Use the Sum Rule to add the two
derivatives together.
df 1
6.2
f (x) 0 6.2
dx x x
b. Apply the Constant Multiplier and the ex Rules to the first term and the Constant
Rule to the second term. Again, use the Sum Rule to add the two derivatives.
dg
g(t) 4e t 0 4e t
dt
c. The key to finding the derivative formula for this function is rewriting the two
terms using algebra rules for exponents. Recall that a negative exponent is used
1
to indicate that a term is in the denominator and that an exponent of 2
indicates a square root. Using these facts, rewrite m as
m(r) 8r1 12r1/2
Now apply the Constant Multiplier, the Power, and the Sum Rules to obtain
dm 1 8 6
m(r) (8)r11 (12)r1/21 8r2 6r1/2 2
dr 2 r r
d. Begin by calculating the number 1 12 in order to rewrite the formula
0.025 12
in the form abx. We have rounded this number to three decimal places, but in
using the derivative to calculate rates of change, you should keep all decimal
places stored in your calculator.
0.025 12y
j(y) 17 1 17(1.025 y)
12
Now apply the Constant Multiplier and the Exponential Rules.
dj
j(y) 17(ln 1.025)(1.025 y) 0.425(1.025 y) ●
dy
In summary, we present a list of the rate-of-change formulas you should know.
(The list includes the formulas from Section 3.2.) The formulas are best learned
through practice. Although you may need to refer to the table on page 194 for some
of the beginning activities, you should attempt to work most of them without look-
ing at this list.
194 CHAPTER 3 Determining Change: Derivatives
Simple Derivative Rules
Rule Name Function Derivative
dy
Constant Rule yb 0
dx
dy
Linear Function Rule y ax b a
dx
dy
Power Rule y xn nxn 1
dx
dy
Exponential Rule y b x, b 0 (ln b)b x
dx
dy
e x Rule y ex ex
dx
dy 1
Natural Log Rule y ln x , x 0
dx x
dy
Constant Multiplier Rule y kf(x) kf(x)
dx
dy
Sum Rule y f(x) g(x) f(x) g(x)
dx
dy
Difference Rule y f(x) g(x) f(x) g(x)
dx
3.3 Concept Inventory 3.3 Activities
• Derivative formulas Getting Started
For constants a, b, k, and n, For each of the functions whose graphs are given in
If y b, then y 0. Activities 1 through 6, first sketch the slope graph and
then give the slope equation. y
If y ax b, then y a.
1. y 2. x
If y x n, then y nx n1. -1 y = −e x
If y b x, then y (ln b)b x. 4
If y e x, then y e x. 3
1 2 y = ex
If y ln x, then y .
x 1
If y kf(x), then y kf(x). x
-2 -1 1
If y f(x) g(x), then y f(x) g(x).
3.3 Exponential and Logarithmic Rate-of-Change Formulas 195
3. y 4. y c. How quickly is the investment growing after
3 5 years?
2 y = ln x y= x
2 d. What is the percentage rate of growth after
1
1 5 years?
x
–1 1 2 3 4 5 x 22. Investment An individual has $45,000 to invest:
–1 1 2 3 4
$32,000 will be put into a low-risk mutual fund
–2 averaging 6.2% interest compounded monthly,
–3 and the remainder will be invested in a high-yield
bond fund averaging 9.7% interest compounded
5. y 6. y
continuously.
7 1 a. Find an equation for the total amount in the two
y = x2
2 investments.
5
y = –ln x + 3 1 b. Give the rate-of-change equation for the
3 x combined amount.
–3 –2 –1 1 2 3
1 c. How rapidly is the combined amount of the
0 x investments growing after 6 months? after
1 2 3 15 months?
For each function in Activities 7 through 20, give the 23. Investment The value of a $1000 investment after
derivative formula. 10 years in an account whose interest rate is 100r%
compounded continuously is
7. h(x) 3 7e x 8. y 5e x 3
A(r) 1000e10r dollars
9. g(x) 2.1x 10. y 3.5x
a. Write the rate-of-change function for the value
11. h(x) 12(1.6)x 12. y 6(0.8)x of the investment.
b. Determine the rate of change of the value of the
13. f(x) 101 4
0.05 4x
(Hint: Rewrite as abx.) investment at 7% interest. Discuss why the rate
of change is so large.
14. f(x) 241 12
0.06 12x
c. If the interest rate is input as a percentage instead
15. j(x) 4.2(0.8x) 3.5 of a decimal, the function for the value of a $1000
investment after 10 years in an account with
16. j(x) 7(1.3x) e x 17. j(x) 4 ln x interest compounded continuously is
18. j(x) ln x 19. y 12 7 ln x A(r) 1000e0.1r dollars
20. k(x) 3.7ex 2 ln x where r 1.00 when the interest rate is 1%,
r 1.25 when the interest rate is 1.25%, etc.
Write the rate-of-change function for the value
Applying Concepts
of the investment. Compare this rate-of-change
21. Investment The value of a $1000 investment in an function with that in part a.
account with 4.3% interest compounded continu- d. Using the function in part c, determine the
ously can be modeled as rate of change of the value of the investment at
7% interest. Compare this answer to that of
A e 0.043t thousand dollars after t years, t 0
part b.
a. Write the rate-of-change formula for the value of
24. Rising Dough For the first couple of hours after
the investment. (Hint: Let b e0.043, and use the
yeast dough has been kneaded, it approximately
rule for y bt .)
doubles in volume every 42 minutes. If we prepare
b. How much is the investment worth after 5 years? 1 quart of yeast dough and let it rise in a warm
196 CHAPTER 3 Determining Change: Derivatives
room, then its growth can be modeled by the b. Write a formula giving the rate of change of
function population before t 13.
V e h quarts c. Write a formula giving the rate of change of
population after t 13.
where h is the number of hours the dough has been d. How quickly was the population growing or
allowed to rise. declining in the beginning of the years 1870,
a. How many minutes will it take the dough to 1873, and 1900?
attain a volume of 2.5 quarts? 27. Weight The weight of a laboratory mouse
b. Write a formula for the rate of growth of the between 3 and 11 weeks of age can be modeled by
yeast dough. the equation
c. How quickly is the dough expanding after 24 min- w(t) 11.3 7.37 ln t grams
utes, after 42 minutes, and after 55 minutes?
where the age of the mouse is (t 2) weeks after
Report your answers in quarts per minute.
birth (thus, for a 3-week-old mouse, t 1).
25. iPod The cumulative sales of iPods can be a. What is the weight of a 9-week-old mouse, and
modeled by the equation how rapidly is its weight changing?
s(x) 0.14(4.106x) million units b. How rapidly on average does the mouse grow
between ages 6 and 11 weeks?
where x 0 is the fiscal year ending in September
c. What happens to the rate at which the mouse is
2002, 0 x 3.
growing as it gets older? Explain.
(Source: Simplified model based on data from “Apple
Reports Fourth Quarter Results,” 2003–2005, Apple 28. Web TV The projected number of homes with
Computer, Inc.)
access to the Internet via cable television between
a. How long did it take Apple to sell 2.5 million 1998 and 2005 can be modeled by the equation
iPods?
I(x) 138.27 76.29 ln x million homes
b. Write a formula for the rate at which iPods were
x years after 1990.
selling x years after the end of fiscal year 2001–
(Source: Based on data from Paul Kagen Associates, Inc., Cable
2002. Television Technology.)
c. How quickly were iPods selling at the time the a. Give the rate-of-change formula for the
2.5 millionth iPod was sold? projected number of such homes.
26. Population The population of Aurora, a Nevada b. How many homes are projected to have Internet
ghost town, can be modeled as access via a cable TV system in 2004, and how
rapidly is that number projected to be growing?
7.91t 3 121t 2 194t 29. Milk Storage The temperature at which milk
123 people when 0.7 t 13 must be stored can be modeled by the equation
p(t)
45.5(0.847t) T(d) 9.9ln(d) 60.5°F
thousand people when 13 t 55
where d is the number of days that the milk needs to
where t is the number of years since the beginning of remain fresh, d 0.
(Source: Simplified model based on data from the back of a
1860.
milk carton from Model Dairy.)
(Source: Simplified model based on data from Don Ashbaugh,
Nevada’s Turbulent Yesterday: A Study in Ghost Towns, Los a. Write a formula for the rate of change of
Angeles: Westernlore Press, 1963.) temperature.
a. The population model is defined using two b. Consider graphs of the temperature function
different functions over different time intervals. and the rate-of-change function. What do the
For what years is a cubic model used? For what graphs of the temperature function and its rate of
years is an exponential model used? change tell us as the number of days increases?
3.3 Exponential and Logarithmic Rate-of-Change Formulas 197
30. Costs Suppose the managers of a dairy company a. Find a model for the data.
have found that it costs them approximately b. Find a formula for the rate of change of the
c(u) 3250 75 ln u dollars to produce u units median family income.
of dairy products each week. They also know
c. Find the rates of change and percentage rates of
that it costs them approximately s(u) 50u
change of the median family income in 1972,
1500 dollars to ship u units. Assume that the
1980, 1984, 1992, and 1996.
company ships its products once each week.
a. Write the formula for the total weekly cost of d. Do you think the above rates of change and
producing and shipping u units. percentage rates of change affected the reelection
campaigns of Presidents Nixon (1972), Carter
b. Write the formula for the rate of change of the to-
(1980), Reagan (1984), Bush (1992), and Clinton
tal weekly cost of producing and shipping u units.
(1996)?
c. How much does it cost the company to produce
and ship 5000 units in 1 week? 33. iPods The cumulative revenue realized by Apple
d. What is the rate of change of total production on the sales of iPods is shown in the table.
and shipping costs at 5000 units? Interpret your
Fiscal year iPod revenue
answer.
(ending September) (millions of dollars)
31. Tuition CPI The consumer price index (CPI) for
2002 53
college tuition between 1990 and 2000 is shown in
the table. 2003 174
2004 711
Year CPI Year CPI Year CPI 2005 1923
1990 175.0 1994 249.8 1998 306.5 2006* 9423
1991 192.8 1995 264.8 1999 318.7 *projected
1992 213.5 1996 279.8 2000 331.9 (Source: Based on data from “Apple Reports Fourth
Quarter Results,” 2003–2005, Apple Computer, Inc.)
1993 233.5 1997 294.1
(Source: Statistical Abstract, 2001.) a. Find an exponential model for the data.
b. Write the derivative formula for the model.
a. Align the data as the number of years since 1980, c. Determine the revenue, rate of change in
and find a log model for the CPI. revenue, and percentage rate of change in
b. Use the model to find the rate of change of the revenue in 2005. Interpret these values.
CPI in 1998.
34. VCR Homes The percentage of households with
32. Income The Bureau of the Census reports the
TVs that also have VCRs from 1990 through 2001 is
median family income since 1947 as shown in the
shown in the table.
table. (Median income means that half of American
families make more than this value and half make Households Households
less.) Year (percent) Year (percent)
1990 68.6 1996 82.2
Median family income
1992 75.0 1998 84.6
Year (constant 1997 dollars)
1994 79.0 2001 86.2
1947 20,102
1957 26,133 (Sources: Statistical Abstract, 1998, and Television Bureau of
Advertising.)
1967 35,076
1977 40,656 a. Align the input data as the number of years since
1987, and find a log model for the data.
1987 43,756
b. Write the rate-of-change formula for the model
1997 44,568
in part a.
198 CHAPTER 3 Determining Change: Derivatives
c. According to the model, what was the percentage and then reconciling the result with the known
of households with TVs that also had VCRs in derivative formula for y e x.
2000? How rapidly was the percentage growing b. Use the derivative formula for y b x to find a
in that year? Interpret your answers. formula for the derivative of an exponential
function of the form y ekx, where k is some
Discussing Concepts known constant.
35. We have seen that the derivative of y b x is 36. Use your knowledge of the shape and end behavior
dy
dx (ln b)b as long as b 0. We have also seen
x of the graph of a log function of the form
dy y a b ln x, as well as your knowledge of the
that the derivative of y ex is dx ex. simple derivative rules in Sections 3.2 and 3.3, to
a. Show that the derivative formula for y e x is a describe the shape and end behavior of the graph
special case of the derivative formula for y b x and mathematical form of the rate-of-change
by applying the formula for y b x to y e x function of a log model.
3.4 The Chain Rule
We now present derivative formulas for more complicated functions than those
considered in Sections 3.2 and 3.3. In particular, we introduce the method used for
finding derivative formulas for composite functions.
The First Form of the Chain Rule
It is well known that high levels of carbon dioxide (CO2) in the atmosphere are linked
to increasing populations in highly industrialized societies. This is because large urban
environments consume enormous amounts of energy, and CO2 is a natural byproduct
of the (often incomplete) consumption of that energy. Imagine that in a certain large
city, the level of CO2 in the air is linked to the size of the population by the equation
C(p) p, where the units of C(p) are parts per million (ppm) and p is the
population. Also suppose that the population is projected to grow quadratically
between 2000 and 2015 according to the equation p(t) 400t2 2500 people, where
t is the number of years since 2000. Note that C is a function of p, and p is a function
of t. Thus, indirectly, C is also a function of t. Suppose we want to know the rate of
change of the CO2 concentration with respect to time in 2013—that is, how rapidly the
CO2 concentration is rising or falling in 2013. The mathematical notation for
dC
this rate of change is dt , and the units are ppm per year.
dC 1
The derivative of C is dp ppm per person. But this is not the rate of change
2p
dC
that we want because dp is the rate of change with respect to population, not time. The
question now becomes “How do we transform ppm per person to ppm per year?” If
we knew the rate of change of population with respect to time (people per year), then
we could multiply as indicated to get the desired units:
person year ppm
ppm people
year
3.4 The Chain Rule 199
Population is given as a function of time, so its derivative is the rate of change that we
dp
need: dt 800t people per year. This motivates
800t
dC dp dC 1 ppm people dC ppm
or
dp dt dt 2p person year dt year
dC
Because dt is a rate of change with respect to time, it is standard procedure to write the
derivative formula in terms of t. Recall that p(t) 400t2 2500 people, where t is
the number of years since 2000. Substituting 400t2 2500 for p in the equation for
dC
dt , we have
dC 1
(800t) ppm/year
dt 2400t2 2500
Now we substitute t 13 (for 2013) to obtain our desired result:
dC 1
[800(13)] 19.64 ppm/year
dt 2400(13)2 2500
In 2013, the CO2 concentration will be increasing by approximately 19.64 ppm per
dC
year. The method used to find dt in the situation above is called the Chain Rule
because it links together the derivatives of two functions to obtain the derivative of
their composite function.
The Chain Rule (Form 1)
If C is a function of p, and p is a function of t, then
dC
dt
dC dp
dp dt
EXAMPLE 1 Using the First Form of the Chain Rule
Violin Production Let A(v) denote the average cost to produce a student violin when
v violins are produced, and let v(t) represent the number (in thousands) of student
violins produced t years after 2000. Suppose that 10 thousand student violins are
produced in 2008 and that the average cost to produce a violin at that time is $142.10.
Also, suppose that in 2008 the production of violins is increasing by 100 violins per year
and the average cost of production is decreasing by 15 cents per violin.
a. Describe the meaning and give the value of each of the following in 2008:
i. v(t) ii. v(t) iii. A(v) iv. A(v)
b. Calculate the rate of change with respect to time of the average cost for student
violins in 2008.
200 CHAPTER 3 Determining Change: Derivatives
Solution
a. i. There are v(8) 10,000 violins produced in 2008.
ii. The rate of change of violin production in 2008 is v(8) 0.1 thousand
dv
violins per year. That is, dt 100 violins per year.
iii. The average cost to produce a violin is A(10) $142.10 when 10,000
violins are produced.
iv. When 10,000 violins are produced, the average cost is changing at a rate of
dA
A(10) $0.15 per violin. That is, dv $0.15 per violin.
b. The rate of change with respect to time of the average cost to produce a student
violin in 2008 is
dA dA dv
dt dv dt
($0.15 per violin)(100 violins per year) $15 per year
In 2008 the average cost to produce a violin is declining by $15 per year. ●
The Second Form of the Chain Rule
Recall the discussion at the beginning of this section concerning CO2 pollution and
population. We were given two functions—p, with input t, and C, whose input
dC
corresponds to the output of p—and then asked to find the derivative dt . You may
wonder why we did not substitute the expression for population into the CO2
equation before finding the derivative:
C(p) p with p(t) 400t2 2500 so C(p(t)) 400t2 2500
This process, which is called function composition (see Section 1.1), enables us to
dC
express C directly as a function of t. If we now take the derivative, we get dt , which is
exactly what we needed. The reason we did not do this before is that we did not know
a formula for finding the derivative of a composite function. However, we can now
use the Chain Rule to obtain a formula. First, we review some terminology from
Section 1.1.
Because p(t) was substituted into the formula for C to create the composite func-
tion C p, we call p the inside function and C the outside function. Next, recall the
Chain Rule process:
dC dC dp
dt dp dt
1
(800t)
2p
1
2400t2 2500
800t
1
The first term, , is simply the derivative of p with p replaced by
2400t 2 2500
400t2 2500. This is the derivative of the outside function with the inside function
3.4 The Chain Rule 201
substituted for p. The second term, 800t, is the derivative with respect to t of p, the inside
function. This leads us to a second form of the Chain Rule. If a function is expressed as
a result of function composition (that is, if it is a combination of an inside function and
an outside function), then its slope formula can be found as follows:
derivative of the
derivative
outside function
Slope formula of composite function of the inside
with the inside
function
function untouched
Mathematically, we state this form of the Chain Rule as follows:
The Chain Rule (Form 2)
If a function f can be expressed as the composition of two functions h and
g—that is, if
f (x) (h g)(x) h(g(x))
then its slope formula is
df
f(x) h(g(x)) g(x)
dx
In Example 2, we consider three somewhat different forms of composite
functions, identify the inside function and the outside function for each, and use the
second form of the Chain Rule to find formulas for the derivatives.
EXAMPLE 2 Using the Second Form of the Chain Rule
Write the derivatives (with respect to x) for the following three functions.
3
a. y e x b. y (x 3 2x 2 4)1/2 c. y
2
4 2x2
Solution
a. We can consider y e x as composed of an outside function y e p and an
2
inside function p x2. The derivative of the outside function is e p. (This
exponential function is its own derivative.) Form 2 of the Chain Rule instructs
us to leave the inside function untouched (that is, in its original form), so instead
2
of e p appearing in the derivative, the first expression in the slope formula is e x .
The second expression in the slope formula is the derivative, 2x, of the inside
function. The final answer is the product of these two derivatives.
dy
(e x )(2x) 2xe x
2 2
dx
b. The inside function of y (x3 2x2 4 )1/2 is p x3 2x2 4, and the
1
outside function is y p1/2. The derivative of the outside function is 2p 1/2;
202 CHAPTER 3 Determining Change: Derivatives
1
with p untouched, this becomes 2(x3 2x2 4) 1/2. The derivative of the
inside function is 3x2 4x. Thus the Chain Rule gives
dy 1
(x3 2x2 4) 1/2 (3x2 4x)
dx 2
3
c. The function y can be thought of as the composition of the outside
4 2x2
3
function y p and the inside function p 4 2x2. The derivative of the
3 3
outside function is 2 , or (4 2x2)2. The derivative of the inside function is
4x.
p
The derivative of the composite function is then
3
dy
dx
(4 2x )
2 2 (4x) 12x
(4 2x2)2
●
As illustrated in Example 3, one common use of the Chain Rule is to find the
derivative of a logistic function.
EXAMPLE 3 Using the Chain Rule to Find a Logistic Function Derivative
VCR Homes The percentage of households between 1980 and 2001 with VCRs can
be modeled* by
84.4
P(t) percent
1 33.6e
0.484t
where t is the number of years since 1900. Find the rate-of-change formula for P with
respect to t.
Solution
The function P can be rewritten as
P(t) 84.4(1 33.6e 0.484t) 1
In this form, it is easy to see that p 84.4u 1 is the outside function and that the
inside function is u 1 33.6e 0.484t . Further, we can split u into an outside and
an inside function with u 1 33.6ev as the outside function and v 0.484t as
the inside function.
Now, the derivative of P with respect to t is
P(t) (derivative of 84.4u 1)(derivative of u)
(derivative of 84.4u 1)[(derivative of 1 33.6ev)(derivative of v)]
(84.4u 2)[(33.6ev)(0.484)]
84.4(33.6ev)(0.484)
u2
1372.547ev
u2
*Based on data from Statistical Abstract, 1998.
3.4 The Chain Rule 203
Next, substitute u 1 33.6e 0.484t and v 0.484t back into the expression to
obtain the derivative in terms of t.
dP 1372.547e 0.484t
percentage points per year is the rate of change in the
dt (1 33.6e 0.484t)2
number of households with VCRs between 1980 and 2001
where t is the number of years since 1980. ●
3.4 Concept Inventory 4. Leaking Tank A gas station owner is unaware that
one of the underground gasoline tanks is leaking.
• Function composition The leaking tank currently contains 600 gallons of
• Inside and outside portions of a composite function gas and is losing 3.5 gallons per day. If the value of
the gasoline is $1.51 per gallon, how much potential
• Chain Rule: revenue is the station losing per day?
dC dC dp
(Form 1) 5. Revenue Let R(x) be the revenue in Canadian
dt dp dt
dollars from the sale of x units of a commodity, and
C(t) C(p(t)) p(t) (Form 2)
let C(R) be the U.S. dollar value of R Canadian dol-
lars. On November 25, 2002, $10,000 Canadian was
worth $6334.70 U.S., and the rate of change of the
3.4 Activities U.S. dollar value was $0.63347 U.S. per Canadian
dollar. On the same day, sales were 476 units,
Getting Started producing a revenue of $10,000 Canadian, and rev-
enue was increasing by $2.6 Canadian per unit.
1. Let x be a function of t, and let f be a function whose Identify the following values on November 25,
input corresponds to the output of x. If x(2) 6, 2002, and write a sentence interpreting each value.
f (6) 140, x(2) 1.3, and f(6) 27, give the
values of a. R(476) b. C(10,000)
df dR dC dC
a. f (x(2)) b. when x 6 c.
dx
d.
dR
e.
dx
dx
dx df
c. when t 2 d. when t 2 6. Mail Suppose that v(t) is the volume of mail (in
dt dt thousands of pieces) processed at a post office on
2. Let v be a function of x, and let g be a function whose the tth day of the current year and that E(v) is the
input corresponds to the output of v. If v(88) 17, number of employee-hours needed to process v
v(88) 1.6, g(17) 0.04, and g(17) 0.005, thousand pieces of mail. On January 1 of this year,
give the values of 150 thousand pieces of mail were processed, and
dv that number was decreasing by 200 pieces per day.
a. g(v(88)) b. when x 88 The rate of change of the number of employee-
dx
hours is a constant 12 hours per thousand pieces of
dg dg
c. when x 88 d. when x 88 mail. Identify the following quantities on January 1
dv dx of this year, and write a sentence interpreting each
3. Investment An investor has been buying gold at a value.
constant rate of 0.2 troy ounce per day. The investor dv
currently owns 400 troy ounces of gold. If gold is a. v(1) b.
dt
currently worth $323.10 per troy ounce, how
quickly is the value of the investor’s gold increasing dE dE
c. d.
per day? dv dt
204 CHAPTER 3 Determining Change: Derivatives
7. Refuse The population of a city in the Northeast 8 350
19. f(x) 20. f(x)
130
is given by p(t) 1 12e 0.02t thousand people, (x 1)3 4x 7
where t is the number of years since 2000. The 21. f(x) x2 3x 22. f (x) x2 5x
number of garbage trucks needed by the city can be
modeled by the equation g(p) 2p 0.001p3, 23. f(x) ln(35x) 24. f(x) (ln 6x)2
where p is the population in thousands. Find the 25. f(x) ln(16x2 37x) 26. f(x) e3.7x
value of the following in 2010:
27. f(x) 72e0.6x 28. f(x) e4x
2
a. p(t) b. g(p)
dp dg dg 29. f(x) 1 58e0.08x 30. f(x) 1 58e(13x)
c. d. e.
dt dp dt
f. Interpret the answers to parts a through e. 12
31. f(x) 7.3
1 18e0.6x
8. Profit Let p(x) 1.019x Canadian dollars be
the profit from the sale of x mountain bikes. On 37.5
32. f(x) 89
November 25, 2002, p Canadian dollars were worth 1 8.9e1.2x
p
C(p) 1.5786 . On the same day, sales were 476
33. f(x) x 3x
3
34. f(x) 32x
mountain bikes. Identify the following quantities
on November 25, 2002: 35. f(x) 2ln x 36. f(x) ln(2x)
a. p(x) b. C(p) L
37. f(x) AeBx 38. f(x)
dp dC dC 1 AeBx
c. d. e.
dx dp dx
Applying Concepts
f. Interpret the answers to parts a through e.
39. Advertising The marketing division of a large
Rewrite each pair of functions in Activities 9 through 16 firm has found that it can model the effectiveness of
as a single composite function, and then find the deriv- an advertising campaign by S(u) 0.75u 1.8,
ative of the composite function. where S(u) represents sales in millions of dollars
9. c(x) 3x2 2 x(t) 4 6t when the firm invests u thousand dollars in
advertising. The firm plans to invest
10. f(t) 3et t(p) 4p2 u(x) 2.3x2 53.2x 249.8 thousand dollars
4 each year x years from now.
11. h(p) p(t) 1 3e 0.5t
p a. Write the formula for predicted sales x years
12. g(x) 7 5x x(w) 4ew from now.
b. Write the formula for the rate of change of
13. k(t) 4.3t3 2t2 4t 12 t(x) ln x
predicted sales x years from now.
14. f(x) ln x x(t) 5t 11
c. What will be the rate of change of sales in 2007?
15. p(t) 7.9(1.046t) t(k) 14k3 12k2
40. Rearing Children The percentage of children
9.1
16. r(m) 3m m(f ) f 4 f 2 living with their grandparents between 1970 and
m2
2000 can be modeled by the equation
For each of the composite functions in Activities 17 p(t ) 3 0.216e0.09263t percent
through 38, identify an inside function and an outside
t years after 1970.
function, and find the derivative with respect to x of the
(Source: Based on data from the U.S. Bureau of the Census.)
composite function.
a. Write the rate-of-change formula for p.
17. f(x) (3.2x 5.7)5
b. How rapidly was the percentage of children
18. f(x) (5x2 3x 7) 1 living with their grandparents growing in 1995?
3.4 The Chain Rule 205
c. How rapidly on average did the percentage of 43. P.T.A. A model for the number of states
children living with their grandparents grow associated with the national P.T.A. organization is
between 1970 and 1990?
49
m(x) states
d. Geometrically illustrate the answers to parts b 1 36.0660e0.206743x
and c.
x years after 1895.
41. Revenue In October of 1999, iGo Corp. offered 5 (Source: Based on data from Hamblin, Jacobsen, and Miller, A
million shares of public stock at $9 per share. Mathematical Theory of Social Change, New York: Wiley, 1973.)
Revenue for the two years preceding the stock a. Write the derivative of m.
offering can be modeled by the equation
b. How many states had national P.T.A. member-
R(q) 2.9 0.0314e0.62285q million dollars ship in 1902?
q quarters after the beginning of 1998. c. How rapidly was the number of states joining the
(Source: Based on data from the Securities and Exchange P.T.A. growing in 1915? in 1927?
Commission.)
a. Write the rate-of-change formula of R. 44. Flu Civilian deaths due to the influenza epidemic
in 1918 can be modeled as
b. Complete the table below.
93,700
c(t) deaths
1 5095.9634e1.097175t
June June June
Quarter ending 1998 1999 2000
t weeks after August 31, 1918.
(Source: Based on data from A. W. Crosby, Jr., Epidemic and
Revenue Peace 1918, Westport, CT: Greenwood Press, 1976.)
Rate of change
a. How rapidly was the number of deaths growing
of revenue
on September 28, 1918?
Percentage rate
of change b. What percentage increase does the answer to
of revenue part a represent?
c. Repeat parts a and b for October 26, 1918.
42. Cable TV The percentage of households with TVs d. Why is the percentage change for parts b and c
that subscribed to cable in the years from 1970 decreasing even though the rate of change is
through 2002 can be modeled by the logistic equation increasing?
62.7 45. Tuition The tuition at a private 4-year college
P(t ) 6 percent
1 38.7e0.258t from 2000 through 2010 is projected to grow as
where t is the number of years since 1970. shown in the table.
(Source: Based on data from the Television Bureau of
Advertising.)
a. Write the rate-of-change formula for the per- Tuition Tuition
centage of households with TVs who subscribe to Year (dollars) Year (dollars)
cable. 2000 14,057 2006 16,918
b. How rapidly was the percentage growing in 2001 14,434 2007 17,561
2000? 2002 14,847 2008 18,264
c. According to the model, what will happen to the 2003 15,298 2009 19,033
percentage of cable subscribers in the long run? 2004 15,790 2010 19,873
Do you believe that the model is a correct
2005 16,329
predictor of the long-term behavior? Explain.
206 CHAPTER 3 Determining Change: Derivatives
a. Find an exponential equation of the form are made to the automated machinery to help boost
f(x) abx to fit the data. production. The company has kept a record of the
b. Convert the equation you found in part a to one average units per week that were produced in each
of the form f(x) aekx. quarter since January 2000. These data are given in
the table.
c. Find rate-of-change formulas for both
equations.
Quarter Units per week
d. Use both equations to find the rate of change in Jan–Mar 2000 2000
2008. How do your answers compare?
Apr–June 2000 2070
46. Police Calls Dispatchers at a sheriff’s office July–Sept 2000 2160
record the total number of calls received since Oct–Dec 2000 2260
5 A.M. in 3-hour intervals. Total calls for a typical
Jan–Mar 2001 2380
day are given in the table.
Apr–June 2001 2510
July–Sept 2001 2660
Time Total calls since 5 A.M.
Oct–Dec 2001 2820
8 A.M. 81
Jan–Mar 2002 3000
11 A.M. 167
Apr–June 2002 3200
2 P.M. 301
July–Sept 2002 3410
5 P.M. 495
Oct–Dec 2002 3620
8 P.M. 738
Jan–Mar 2003 3880
11 P.M. 1020
Apr–June 2003 4130
2 A.M. 1180
July–Sept 2003 4410
5 A.M. 1225
Oct–Dec 2003 4690
(Source: Greenville County, South Carolina, Sheriff’s Office.)
a. Is a cubic or a logistic model more appropriate a. Find an appropriate model for production per
for this data set? Explain. week x quarters after January 2000.
b. Find the more appropriate model for the data. b. Use the company cost model along with your
production model to write an expression model-
c. Find the rate-of-change formula for the model. ing cost per unit as a function of the number of
d. Evaluate the rate of change at noon, 10 P.M., quarters since January 2000.
midnight, and 4 A.M. Interpret the rates of c. Use your model to predict the company’s cost
change. per week for each quarter of 2004.
e. Discuss how rates of change can help a d. Carefully study a graph of the function in part b
sheriff’s office schedule dispatchers for work from January 2000 to January 2005. According
each day. to this graph, will cost ever decrease?
e. Find an expression for the rate of change of the
47. Cost A manufacturing company has found that
cost function in part b. Look at the graph of the
it can stock no more than 1 week’s worth of per-
rate-of-change function. According to this
ishable raw material for its manufacturing process.
graph, will cost ever decrease?
When purchasing this material, however, the com-
pany receives a discount based on the size of the 48. Cost A dairy company’s records reveal that it
order. Company managers have modeled the cost costs about C(u) 3250.23 74.95 ln u dollars
data as C(t) 196.25 44.45 ln t dollars to pro- per week for the company to produce u units each
duce t units per week. Each quarter, improvements week. Consumer demand has been increasing, so
3.4 The Chain Rule 207
the company has been increasing production to c. Use the company’s cost model along with your
keep up with demand. The accompanying table production model to write an expression model-
indicates the production of the company, in units ing cost per week as a function of the number of
per week, from 1990 through 2003. years since 1990.
d. Write the rate-of-change function of the cost
function you found in part c.
Production
Year (units per week) e. Use your model to estimate the company’s cost
1990 5915 per week in 2002, 2003, 2004, and 2005. Also
estimate the rates of change for those same years.
1991 5750
f. Carefully study the cost graph from 1990 to
1992 5940
2010. According to this graph, will cost ever
1993 6485 decrease? Why or why not?
1994 7385 g. Look at the slope graph of the cost function from
1995 8635 1990 to 2010. According to this graph, will cost
1996 10,245 ever decrease? Why or why not?
1997 12,210
Discussing Concepts
1998 14,530
49. When you are composing functions, why is it
1999 17,200
important to make sure that the output of the
2000 20,230 inside function agrees with the input of the outside
2001 23,610 function?
2002 27,345 50. Use your knowledge of the shape and end behavior
2003 31,440 of the graph of a shifted exponential function of the
form y aebx c, as well as your knowledge of the
simple derivative rules and the Chain Rule from
a. Describe the curvature of the scatter plot of the Sections 3.2 through 3.4, to describe the shape and
data in the table. What types of equations could end behavior of the graph and the mathematical
be used to fit these data? form of the rate-of-change function of a shifted
b. Find an appropriate model for production. exponential model.
3.5 The Product Rule
It is fairly common to construct a new function by multiplying two functions. For ex-
ample, revenue is the number of units sold multiplied by price. If both units sold and
price are given by functions, then revenue is given by the product of the two func-
tions. How to find rates of change for product functions is the topic of this section.
Applying the Product Rule Without Equations
Suppose that the enrollment in a university is given by a function E and the percent-
age (expressed as a decimal) of students who are from out of state is given by a func-
tion P. In both functions, the input is t, the year corresponding to the beginning of
the school year (that is, for the 2003–04 school year, t is 2003) because school enroll-
ment figures are stated for the beginning of the fall term. Note that the product
function N(t) E(t) P(t) gives the number of out-of-state students in year t. For
example, if in the current year, enrollment began at 17,000 students with 30% of
208 CHAPTER 3 Determining Change: Derivatives
those from out of state, then the number of out-of-state students is calculated as
(17,000)(0.30) 5100.
Suppose that, in addition to enrollment being 17,000 with 30% from out of state,
dE
the enrollment is decreasing at a rate of 1420 students per year dt 1420 ,
and the percentage of out-of-state students is increasing at a rate of 1.5 percentage
dP
points per year dt 0.015 . How rapidly is the number of out-of-state students
changing?
There are two rates to consider in determining N(t). First, we consider the rate at
which the number of out-of-state students is changing because of the overall decline in
enrollment. Of the 1420 students per year by which enrollment is declining, 30% are
from out of state. The product (1420 students per year)(0.30) 426 gives the
amount by which the number of out-of-state students is decreasing. Thus, as a conse-
quence of the decline in enrollment, the number of out-of-state students is declining by
426 students per year.
Second, we consider the rate at which the number of out-of-state students is
increasing. Of the 17,000 students enrolled, the percentage of out-of-state students is
growing by 1.5% per year. Thus the product (17,000 students) (0.015 per
year) 255 gives the amount by which the number of out-of-state students is
increasing. The increasing percentage of out-of-state students results in a rate of
increase of 255 out-of-state students per year.
To find the overall rate of change in the number of out-of-state students, we add
the rate of change due to decline in enrollment and the rate of change due to increase
in percentage.
426 students per year 255 students per year 171 students per year
We interpret this result as follows: As a consequence of the declining enrollment and
the increasing percentage of students from out of state, the number of out-of-state
students is declining by 171 per year.
The steps to obtain this rate of change can be summarized in the equation
dN dE dP
P(t) E(t)
dt dt dt
This result is known as the Product Rule and can be stated as follows: If a function is
the product of two functions, that is,
functionfunction
first second
Product function
then
Derivative derivative derivative
second first
of product of first of second
function function
function function function
3.5 The Product Rule 209
The Product Rule
If f(x) g(x) h(x), then
df
dx
dg
dx
h(x) g(x)
dh
dx
.
EXAMPLE 1 Using the Product Rule Without an Equation
Egg Production The industrialization of chicken (and egg) farming brought
improvements to the production rate of eggs. Consider a chicken farm that has 1000
laying hens, each of which lays an average of 24 eggs each month. By selling or buying
hens, the farmer can decrease or increase production. Also, by selective breeding and
genetic research, it is possible that over a period of time the farmer can increase the
average number of eggs that each hen lays.
a. How many eggs does the farm produce in a month?
b. Suppose the farmer increases the number of hens by 12 hens per month and
increases the average number of eggs laid by each hen by 1 egg per month. By
how much will the farmer’s production be increasing?
Solution
a. The monthly egg production is the product of h, the number of hens, and l, the
number of eggs each hen lays in one month. Currently, h 1000 hens and
l 24 eggs per month. The farmer’s current monthly production is
h l (1000 hens)(24 eggs per hen) 24,000 eggs
dh
b. Let t be the number of months from now. We are told that dt 12 hens per
dl
month and that dt 1 egg per hen per month. Applying the Product Rule yields
d (hl) dh dl
lh
dt dt dt
hens eggs egg/hen
12 24 (1000 hens) 1
month hen month
eggs
1288
month
The farmer’s egg production will be increasing by 1288 eggs per month. ●
Applying the Product Rule with Equations
The next example illustrates using the Product Rule for quantities that are
represented by mathematical functions.
210 CHAPTER 3 Determining Change: Derivatives
EXAMPLE 2 Using the Product Rule in a Business Setting
Sales A music store has determined from a customer survey that when the price of
each CD is x dollars, the number of CDs sold in a four-week period can be modeled
by the function
N (x) 6250 (0.92985x) CDs
Find and interpret the rates of change of revenue when CDs are priced at $10, $12,
$13.75, and $15.
Solution
Revenue is the number of units sold times the selling price. In this case, the monthly
revenue R(x) is given by
R (x) N (x) x 6250 (0.92985x) x dollars
where x dollars is the selling price. Using the Product Rule, we find that the rate-of-
change equation is
dR d d
N (x) x N (x) (x)
dx dx dx
6250 (ln 0.92985)(0.92985x) x 6250 (0.92985x)(1) dollars per dollar
where x dollars is the selling price. The output unit on the derivative, dollars per
dollar, indicates dollars of revenue per dollar of price.
dR
TABLE 3.11 Evaluating dx at the indicated values of x yields Table 3.11. At $10, revenue is
Price Rate of change increasing by $824 per $1 of CD price. In other words, increasing the price results in
(dollars) of revenue an increase in revenue. Similarly, at $12, revenue is increasing by $332 per $1 of CD
(dollars per dollar) price. At $13.75, revenue is neither increasing nor decreasing. This is the price at
which revenue has reached its peak. Finally, at $15, revenue is declining by $191 per
10.00 824
$1 of CD price.
The graph of the revenue function is shown in Figure 3.38. Review the statements
above about how the revenue is changing as they are related to the graph.
12.00 332
R(x)
13.75 0 Revenue
(dollars)
15.00 191 35,000
25,000
15,000
5,000 x
Price
5 10 15 20 (dollars)
FIGURE 3.38 ●
Often product functions are formed by multiplying a quantity function by a
function that indicates the proportion of that quantity for which a certain statement
is true. This is illustrated in Example 3.
3.5 The Product Rule 211
EXAMPLE 3 Using the Product Rule When the Product Involves a Proportion
Tourists Judging on the basis of data supplied by the Office of Travel and Tourism
Industries at the U.S. Department of Commerce, the number of overseas
international tourists who traveled to the United States between 1995 and 2000 can
be modeled by the equation
f (t) 0.148t3 3.435t2 26.673t 45.44 million tourists
where t is the number of years since 1990. Suppose that during the same time period,
the proportion (percentage expressed as a decimal) of foreign travelers to the United
States who were from Europe is given by
p(t)
0.00179t3 0.0395t2 0.275t 1.039
where t is the number of years since 1990.
a. Find a model for the number of European tourists traveling to the United
States.
b. Find the derivative of the formula in part a.
c. Find the number of European tourists to the United States in 2000, and
determine how rapidly that number was changing in that year.
Solution
a. The number N(t) of European tourists to the United States is given by the
product function N(t) f (t) p(t).
N(t) (0.148t3 3.435t2 26.673t 45.44)(0.00179t3 0.0395t2
0.275t 1.039) million European tourists
t years after 1990, 5 t 10.
b. To use the Product Rule, we need the derivatives f and p:
f (t) 0.444t 2 6.87t 26.673 million tourists per year, and
p(t) 0.00537t 2 0.079t 0.275
(Note that we did not label p(t) with units. If p(t) had been expressed as a per-
centage, then the units would have been percentage points per year. Expressed
as a decimal, p(t) is actually a proportion, which is a unitless number. It would
be possible to label p(t) as hundredths of a percentage point per year, but we
choose to state the derivative of p without a label.)
Thus, by the Product Rule,
N(t) (0.444t2 6.87t 26.673)(0.00179t3 0.0395t2 0.275t 1.039)
(0.148t3 3.435t2 26.673t 45.44)(0.00537t2 0.079t 0.275)
European tourists per year t years after 1990; 5 t 10.
c. The number of European tourists in 2000 is N(10).
N(10) (25.79)(0.449) 11.6 million tourists
212 CHAPTER 3 Determining Change: Derivatives
The rate of change of the number of European tourists in 2000 is N(10).
N(10) (2.373)(0.449) (25.79)(0.022)
1.065 0.567
0.50 million European tourists per year
In 2000, there were approximately 11.6 million tourists from Europe, and that num-
ber was growing by about 0.50 million tourists per year. ●
The Quotient Rule
Occasionally functions are constructed as the quotients of other functions—that is,
g (x)
functions of the form f (x) h (x). You may wish to rewrite f (x) as f (x) g (x)[h(x)] 1
and use the Chain and Product Rules to write the derivative formula for such functions.
Another approach is to use the Quotient Rule:
g(x) h(x) h(x) g(x)
f (x)
[h(x)]2
We present this rule without illustration because it is an algebraic consequence of the
Product and the Chain Rules.
b. If N(t) h(t) c(t), what are the input and
3.5 Concept Inventory output of N?
c. Find and interpret the values of N(2) and N(2).
• Product Rule
4. Demand Let D(x) be the demand (in units) for a
If f (x) g(x) h(x),
df dg dh new product when the price is x dollars.
then dx dx h(x) g(x) dx .
a. Interpret the following statements:
• Quotient Rule
i. D(6.25) 1000 ii. D(6.25) 50
b. Give a formula for the revenue R(x) generated
3.5 Activities from the sale of the product when the price is
x dollars. (Assume demand number sold.)
Getting Started c. Find and interpret R(x) when x 6.25.
1. Find h(2) if h(x) f(x) g(x), f(2) 6, 5. Stock Value The value of one share of a company’s
2.6
f (2) 1.5, g (2) 4, and g(2) 3. stock is given by S(x) 15 x 1 dollars x
2. Find r(100) if r(t) p(t) q(t), p(100) 4.65, weeks after it is first offered. An investor buys
p(100) 0.5, q(100) 160, and q(100) 12. some of the stock each week and owns
N(x) 100 0.25x 2 shares after x weeks. The
3. Computer Homes Let h(t) be the number of value of the investor’s stock after x weeks is given by
households in a city, and let c(t) be the proportion V(x) S(x) N(x).
(expressed as a decimal) of households in that city a. Find and interpret the following:
that own a computer. In both functions, t is the
number of years since 2005. i. S(10) and S(10) ii. N(10) and N(10)
a. Interpret the following mathematical iii. V(10) and V(10)
statements: b. Give a formula for V(x).
i. h(2) 75,000 ii. h(2)
1200 6. Education Cost The number of students in an
iii. c(2) 0.52 iv. c(2) 0.05 elementary school t years after 2002 is given by
3.5 The Product Rule 213
S(t) 100 ln (t 5) students. The yearly cost to didate A is declining by 3 percentage points per
educate one student can be modeled by week.
C(t) 1500(1.05t) dollars per student. a. If the election were held today, how many people
a. What are the input and output of the function would vote?
F(t) S(t) C(t)? b. How many of those would vote for candidate A?
b. Find and interpret the following: c. How rapidly is the number of votes that candi-
i. S(3) ii. S(3) iii. C(3) date A will receive changing?
iv. C(3) v. F(3) vi. F(3)
10. Customers Suppose a new shop has been added
c. Find a formula for F(t). in the O’Hare International Airport in Chicago.
7. Farming A wheat farmer is converting to corn At the end of the first year it is able to attract 2%
because he believes that corn is a more lucrative of the passengers passing by the store. The man-
crop. It is not feasible for him to convert all his ager estimates that at that time, the percentage of
acreage to corn at once. He is farming 500 acres of passengers it attracts is increasing by 0.05% per
corn in the current year and plans to increase that year. Airport authorities estimate that 52 thousand
number by 50 acres per year. As he becomes more passengers passed by the store each day at the end
experienced in growing corn, his output increases. of the first year and that that number was increas-
He currently harvests 130 bushels of corn per acre, ing by approximately 114 passengers per day.
but the yield is increasing by 5 bushels per acre per What is the rate of change of customers at the
year. When both the increasing acreage and the shop?
increasing yield are considered, how rapidly is
the total number of bushels of corn increasing Find derivative formulas for the functions in Activities
per year? 11 through 28.
8. Basketball A point guard for an NBA team aver- 11. f (x) (ln x) ex 12. f (x) (x 5) ex
ages 15 free-throw opportunities per game. He cur- 13. f (x) (3x2 15x 7)(32x3 49)
rently hits 72% of his free throws. As he improves,
the number of free-throw opportunities decreases 14. f (x) 2.5 (0.9x)(ln x)
by 1 free throw per game, while his percentage of 15. f (x) (12.8x2 3.7x 1.2)[29 (1.7x)]
hits increases by 0.5 percentage point per game.
When his decreasing free throws and increasing 16. f (x) (5x 29)5(15x 8)
percentage are taken into account, what is the rate
17. f (x) (5.7x2 3.5x 2.9)3(3.8x2 5.2x 7)2
of change in the number of free-throw points that
this point guard makes per game? 2x3 3 12.6 (4.8x)
18. f (x) 19. f (x)
2.7x 15 x2
9. Politics Two candidates are running for mayor
in a small town. The campaign committee for
candidate A has been conducting weekly telephone
20. f (x) (8x2 13) 1 15e
39
0.09x
polls to assess the progress of the campaign.
Currently there are 17,000 registered voters, 48% 21. f (x) (79x) 1 7.68e
198
0.85x 15
of whom are planning to vote. Of those planning
22. f (x) [ln(15.7x3)](e15.7x )
3
to vote, 57% will vote for candidate A. Candidate
B has begun some serious mud slinging, which
430(0.62x)
has resulted in increasing public interest in the 23. f (x)
election and decreasing support for candidate A. 6.42 3.3 (1.46x)
Polls show that the percentage of people who plan 24. f (x) (19 12 ln 2x)(17 3 ln 4x)
to vote is increasing by 7 percentage points per
week, while the percentage who will vote for can- 25. f (x) 4x 3x 2 93
214 CHAPTER 3 Determining Change: Derivatives
4(3x) 14x b. Use the equation to find the revenue in August
26. f (x) 27. f (x)
x 1 12.6e 0.73x and the projected revenue in September. Would
you expect the rate of change of revenue to be
1
28. f (x) (3x2 17x 4) positive or negative in August? Why?
(x 2)2
c. Give the rate-of-change formula for revenue.
d. How rapidly was revenue changing in February,
Applying Concepts
August, and September?
29. Childbirth On the basis of data from a study con- 31. Sales A music store has determined that the
ducted by the University of Colorado School of number of CDs sold monthly is approximately
Medicine at Denver, the percentage of women
receiving regional analgesia (epidural pain relief) Number 6250 (0.9286x) CDs
during childbirth at small hospitals between 1981 where x is the price in dollars.
and 1997 can be modeled by the equation a. Give an equation for revenue as a function of price.
p(x) 0.73 (1.2912x) 8 percent b. If each CD costs the store $7.50, find an equation
x years after 1980. for profit as a function of price.
(Source: Based on data from “Healthfile,” Reno Gazette c. Find formulas for the rates of change of revenue
Journal, Oct. 19, 1999, p. 4.) and profit.
Suppose that a small hospital in southern Arizona d. Complete the table below.
has seen the yearly number of women giving
Rate of change Rate of change
birth decline as described by the equation
Price of revenue of profit
b(x) 0.026x2 3.842x 538.868 women giv-
ing birth x years after 1980. $13
$14
a. Give the equation and its derivative for the num-
ber of women receiving regional analgesia while $20
giving birth at the Arizona hospital. $21
b. Was the percentage of women who received re- $22
gional analgesia while giving birth increasing or
decreasing in 1997? e. What does the table tell the store manager about
c. Was the number of women who gave birth at the the price corresponding to the highest revenue?
Arizona hospital increasing or decreasing in 1997? f. What is the price corresponding to the highest
d. Was the number of women who received re- profit?
gional analgesia during childbirth at the Arizona 32. Population The population (in millions) of the
hospital increasing or decreasing in 1997? United States as a function of the year is given by
e. If the Arizona hospital made a profit of $57 per P(d) 99.9 (1.108t) million people
woman for the use of regional analgesia, what
was the profit for the hospital from this method d decades after 1970.
of pain relief during childbirth in 1997? The percentage of people in the United States who
30. Sales During the first 8 months of last year, a live in the Midwest can be modeled by the equation
grocery store raised the price of a certain brand of tis- m(d) 6.53 (0.994d) 22 percent, where d is the
sue paper from $1.14 per package at a rate of 4 cents number of decades since 1970.
per month. Consequently, sales declined. The sales of (Source: Simplified models based on data from Statistical Ab-
tissue can be modeled as stract, 2001.)
S(m) 0.95m2 0.24m 279.91 packages a. Write an expression for the number of people
who live in the Midwest t years after 1970.
during the mth month of the year. b. Find an expression for the rate of change of the
a. Construct an equation for revenue. population of the Midwest.
3.5 The Product Rule 215
c. According to the model how rapidly was the a. Using time as the input, determine the best
population of the Midwest changing in 1990, model for each set of data.
1995, and 2000? b. Write an expression for the number of men 65
33. Costs Costs for a company to produce between 10 years old or older who are living below the
and 90 units per hour are given in the table. poverty level in the United States.
c. How rapidly was the number of male senior citi-
Cost Cost
zens living below the poverty level changing in
Units (dollars) Units (dollars) 1990 and in 2000?
10 150 60 1400 35. VCR Homes The first table gives the number of
households with TVs in the United States for
20 200 70 2400
selected years between 1970 and 2002.
30 250 80 3850
40 400 90 5850 Households Households
50 750 Year (millions) Year (millions)
1970 59 1990 92
a. Find an exponential model for production costs. 1975 69 1995 95
b. Find the slope formula for production costs. 1980 76 2000 101
c. Convert the model in part a to one for the aver- 1985 85 2002 106
age cost per unit produced.
(Sources: Statistical Abstract, 2001, and Television Bureau of
d. Find the slope formula for average cost. Advertising.)
e. How rapidly is the average cost changing when
15 units are being produced? 35 units? 85 units? a. Find a model for the data given in the table
above.
f. Examine the slope graph for average cost. Is there
an interval of production levels over which aver- b. The second table gives (for selected years be-
age cost is decreasing? tween 1978 through 2002) the percentages of
U.S. households with TVs that also have VCRs.
g. Determine the point at which average cost begins
to increase. (That is, find the point at which the
rate of change of average cost changes from nega- Year Percentage Year Percentage
tive to positive.) Explain how you found this point. 1978 0.3 1995 81.0
34. Poverty The accompanying table gives the num- 1980 1.1 2000 85.1
ber of men 65 years old or older in the United States 1985 20.9 2002 91.2
and the percentage of men age 65 or older living 1990 68.6
below the poverty level.
(Sources: Statistical Abstract, 2001, and Television Bureau of
Advertising.)
Men 65 years
or older Percentage below Align the data so that the input values corre-
Year (millions) poverty level spond with those in the model for part a (that is,
1970 8.3 20.2 if 1980 is x 10 in part a, then you want 1980 to
1980 10.3 11.1
be x 10 here also). Find a logistic model for
the data.
1985 11.0 8.7
c. Find a model for the number of U.S. households
1990 12.6 7.8 with VCRs.
1997 14.0 7.0 d. Find an equation for the rate of change of the
2000 14.4 7.5 number of U.S. households with VCRs.
(Sources: Statistical Abstract, 2001, and Current Population e. How rapidly was the number of U.S. households
Survey, March 2001.) with VCRs growing in 1990? in 1995? in 2000?
216 CHAPTER 3 Determining Change: Derivatives
36. Funding The amount of federal funds spent for 37. Dropouts The table shows the number of stu-
agricultural research and services from 1990 dents enrolled in the ninth through twelfth grades
through 2002 in the United States is given in the and the number of dropouts from those same
first table. grades in South Carolina for each school year from
1980–1981 through 1989–1990.
Amount
Year (billions of dollars) School year Enrollment Dropouts
1990 2.197 1980–81 194,072 11,651
1992 2.539 1981–82 190,372 10,599
1994 2.695 1982–83 185,248 9314
1996 2.682 1983–84 182,661 9659
1998 2.909 1984–85 181,949 8605
2000 3.189 1985–86 182,787 8048
2002 4.252 1986–87 185,131 7466
(Source: U.S. Office of Management 1987–88 183,930 7740
and Budget.) 1988–89 178,094 7466
1989–90 172,372 5768
The purchasing power of the dollar, as measured
by producer prices from 1988 through 2000 is (Source: Compiled from Rankings of the Counties and
given in the second table. (In 1982, one dollar was School Districts of South Carolina.)
worth $1.00.)
a. Find a model for enrollment and a cubic model
for the number of dropouts.
Year Purchasing power of $1
b. Use the two models that you found in part a to
1988 0.93
construct an equation for the percentage of high
1990 0.84 school students who dropped out each year.
1992 0.81 c. Find the rate-of-change formula of the percent-
1994 0.80 age of high school students who dropped out
1997 0.76 each year.
2000 0.73 d. Look at the rate of change for each school year
from 1980–1981 through 1989–1990. In which
(Source: Statistical Abstract, 1998, 2001.)
school year was the rate of change smallest?
When was it greatest?
a. Find models for both sets of data. (Remember to
e. Are the rates of change positive or negative?
align such that both models have the same input
What does this say about high school attrition in
values.)
South Carolina during the 1980s?
b. Use these models to determine a new model for
the amount, measured in constant 1982 dollars, 38. Jobs A house painter has found that the number
spent on agricultural research and services. of jobs that he has per year is decreasing in inverse
proportion to the number of years he has been in
c. Use your new model to find the rates of change
business. That is, the number of jobs he has each
and the percentage rates of change of the amount 104.25
spent on agricultural research and services in year can be modeled by j (x) x , where x is the
1992 and 2000. number of years since 1997. He has also kept a ledger
d. Why might it be of interest to consider an expen- of how much, on average, he was paid for each job.
diture problem in constant dollars? His income per job is presented in the table
3.5 The Product Rule 217
Income per job c. Write the formula for the rate of change of the
Year (dollars) painter’s income each year.
1997 430 d. What was the painter’s total income in 2003?
1998 559 e. How rapidly was the painter’s income changing
in 2003?
1999 727
2000 945 Discussing Concepts
2001 1228
39. When you are working with products of models,
2002 1597
why is it important to make sure that the input val-
2003 2075 ues of the two models correspond? (That is, why
must you align both models in the same way?)
40. We have discussed three ways to find rates of
a. Find an exponential model for his income per change: graphically, numerically, and algebraically.
job. Discuss the advantages and disadvantages of each
b. Write the formula for the painter’s total income method. Explain when it would be appropriate to
per year. use each method.
3.6 Limiting Behavior Revisited: L’Hôpital’s Rule
In Chapter 1 we looked at the end behavior of the six models defined in this text. We
noted that for a continuous, differentiable function f with input x, as x increases with-
out bound f may increase (or decrease) without bound, approach a constant value k,
or fluctuate so that the end behavior is undefined. We use the following limit nota-
tion in each case:
lim f(x)
• x→ f increases without bound
lim f(x)
• x→ f decreases without bound
lim f(x) k
• x→ f approaches the value k
• x→
lim f(x) is undefined f continues to fluctuate
In Chapter 2 we considered the behavior of a function f as its input x approaches
a specific value c from both the left and right sides. When f is continuous and differ-
entiable (except possibly at x c), then
lim f(x) k
• x→c if f approaches k as x approaches c from both the left
and the right
• x→c
lim f(x) is undefined if f increases (or decreases) without bound as x ap-
proaches c
or
if f does not approach the same value as x approaches c
from both the left and the right
218 CHAPTER 3 Determining Change: Derivatives
Analyzing Limits Using Direct Substitution
One of the concepts of calculus tells us that when a function f is continuous over an
open interval including a constant a, then as the inputs approach a, the outputs
approach f(a). This is known as the method of direct substitution: x→a lim f(x) f(a).
When the limits of f and g exist at x c, three rules for limits are
• The sum rule: lim [ f(x) g(x)] x→c
x→c
lim f(x) x→clim g(x)
• The product rule: lim [ f(x) g(x)] [lim
x→c x→c
f(x)] [lim
x→c
g(x)]
lim f(x)
g(x) lim g(x) as long as lim g(x)
0
f(x) x→c
• The quotient rule: lim
x→c x→c
x→c
These rules, as well as the method of direct substitution, are illustrated in Example 1.
EXAMPLE 1 Using Direct Substitution for Limits
Determine the following limits:
a. lim 3x 2 2 b. lim 64(2x)
x→5 x→4
542
c. lim 3x3 12x 2 4 d. lim
x→ x→ 1 3e4x
Solution
a. lim 3x 2 2 3(5)2 2 77 b. lim 64(2x) 64(24) 4
x→5 x→4
c. A polynomial function either increases or decreases without bound as its input
increases or decreases without bound. Because the cubed term is the term of
largest degree, it determines whether the function will increase or decrease
without bound. Cubing any negative number will produce another negative
number, and multiplying it by 3 will still produce a negative number. There-
fore, the function is decreasing without bound as x decreases without bound:
lim 3x3 12x 2 4
x→
d. The limit of the numerator is the constant 542. Next, we consider lim 3e4x . We
x→
know that 3e4x is a decreasing exponential that approaches 0 as x increases with-
out bound. Therefore, the denominator 1 3e4x approaches 1 as x increases
without bound, so
542 542
lim
x→ 1 3e4x
542 ●
1
Indeterminate Forms and L’Hôpital’s Rule
For our six models these basic definitions apply, but some interesting situations ap-
pear when we try to analyze the limiting behavior of functions that are constructed by
division or multiplication of other functions. For example, using the substitution rule
for limits, we find
3x2 4x 4 0
• lim
x→2 4x 8 0
3.6 Limiting Behavior Revisited: L’Hôpital’s Rule 219
ex
• lim
x→ 5x2
• lim (3x)(2 ln x) 0
x→0
0
The forms , , and 0 , are known as indeterminate forms because without further
0
analysis it is impossible to determine the actual limit. The forms and 0 are both
0
equivalent to the form , so the following rule applies to all three forms.
0
In 1696 the French mathematician Guillaume de l’Hôpital first published the
following result, which was discovered by Johann Bernoulli. This result is now known
as L’Hôpital’s Rule.
L’Hôpital’s Rule
When functions f and g with input x are continuous and smooth (except
possibly at x c ) and lim f(x) lim g(x) 0 then
x→c x→c
f(x) f (x)
lim
x→c g(x)
lim
x→c g(x)
This result says, “The ratio of the functions and the ratio of their slope functions
have the same limiting value.”
Note that L’Hôpital’s Rule uses the quotient of derivatives, not the derivative of a
quotient function as discussed in Section 3.5.
EXAMPLE 2 Using L’Hôpital’s Rule
Determine the following limits:
3x2 4x 4 ex
a. lim b. lim
x→2 4x 8 x→∞ 5x2
c. lim (3x)(2 ln x)
x→0
Solution
a. Before applying L’Hôpital’s Rule, we check whether the limit of the ratio is an
indeterminate form: lim 3x2 4x 4 3(22) 4(2) 4 0 and
x→2
lim 4x 8 4(2) 8 0 . Therefore,
x→2
3x2 4x 4 0
lim (i.e., an indeterminate form)
x→2 4x 8 0
We now apply L’Hôpital’s Rule by writing the derivatives of the functions in the
d
numerator and the denominator: (3x2 4x 4) 6x 4 and
dx
220 CHAPTER 3 Determining Change: Derivatives
d
(4x 8) 4. The limit of the ratios of these derivatives is
dx
6x 4 6(2) 4 8
lim 2
x→2 4 4 4
L’Hôpital’s Rule gives us
3x2 4x 4 6x 4
lim lim 2
x→2 4x 8 x→2 4
b. We check lim ex and lim 5x2 . Therefore,
x→∞ x→∞
ex
lim
(i.e., an indeterminate form)
x→∞ 5x2
The derivatives of the functions in the numerator and denominator of this
d x d 2
ratio are e ex and 5x 10x . Applying L’Hôpital’s Rule yields
dx dx
ex ex
lim
x→∞ 5x 2 lim
x→∞ 10x
Further analysis shows
ex
lim (i.e., an indeterminate form)
x→∞10x
d x d
We again apply L’Hôpital’s Rule: e ex and 10x 10. Thus
dx dx
ex ex
lim
x→∞ 10x
lim
x→∞ 10
Since ex increases without bound as x increases without bound
ex
lim
x→∞ 10
Therefore, applying L’Hôpital’s Rule twice we find
ex ex ex
lim 2 lim lim
x→∞ 5x x→∞ 10x x→∞ 10
c. As x approaches 0 from the right, ln x decreases without bound, so 2 ln x in-
creases without bound. At the same time 3x approaches 0. Thus,
lim (3x)(2 ln x) 0 • , (i.e., an indeterminate form)
x→0
To apply L’Hôpital’s Rule, we rewrite, (3x)(2 ln x) as a ratio of functions:
2 ln x
(3x)(2 ln x)
1
3x
1 1
Evaluating lim 2 ln x and lim yields
x→0 x→0 3x 0
2 ln x
lim
x→0 1
3x
3.6 Limiting Behavior Revisited: L’Hôpital’s Rule 221
We apply L’Hôpital’s Rule by finding the derivatives of the functions in
2
d 1
the numerator and the denominator: ( 2 ln x) 2 and
dx x x
d 1
dx 3x
d
dx
(3x) 1 1(3x) 2(3)
3
(3x)2
3
2 2.
9x
1
3x
2
x
Writing the ratio of these derivatives yields .
1
2
3x
2
3x2 6x2
We simplify this as
x
1
2
x
1
x
6x.
2
3x
2
x
Applying the limit, we have lim lim 6x 0 .
x→0 1 x→0
2
3x
Therefore, L’Hôpital’s Rule gives
2
2 ln x x
lim 3x(2 ln x) lim lim lim 6x 0
x→0 x→0 1 x→0 1 x→0
2 ●
3x 3x
Evaluating limits using L’Hôpital’s Rule can be relatively simple, as in part a of
Example 2. Multiple applications of L’Hôpital’s Rule may be necessary to evaluate a
limit, as in part b of Example 2, or it may take some extra algebraic manipulation, as
in part c of Example 2.
1 1
5. lim 6. lim
3.6 Concept Inventory x→0 ln x x→2 x2
For each limit given in Activities 7 through 10, identify
• Direct substitution for limits the indeterminate form. Use L’Hôpital’s Rule to find the
• Indeterminate forms 0
limit of indeterminate forms of type or .
• L’Hôpital’s Rule 0
ln n e2t
7. lim 8. lim
n→1 n 1 t→1 2t
3.6 Activities x 1
4 ex
9. lim 3 10. lim 2
x→1 x 1 x→ x
Getting Started For each limit given in Activities 11 through 14, rewrite
Use the method of direct substitution or end behavior 0
the indeterminate form of type 0 as either type or
analysis to determine the limits given in Activities 1 0
through 6. . Use L’Hôpital’s Rule to find the limit.
1. lim 2x3 3x 2. lim 3x2 ex
x→2 x→2 11. lim t ln t 12. lim 2n2en
t→0 n→
9
3. lim ex ln(x 1) 4. lim 13. lim x2e x 14. lim e t ln t
x→0 x→3 x x→ t→0
222 CHAPTER 3 Determining Change: Derivatives
Applying Concepts 3x2 2x 4 4x2 7
25. lim 26. lim
For Activities 15 through 28, determine the limit. If the x→ 5x2 x 1 x→ 2x3 3
limit is of an indeterminate form, indicate the form and
3x4 4x3
then use L’Hôpital’s Rule to determine the limit. 27. lim 28. lim
x→ 5x3 6 x→ 5x3 6
3x 6 x2 2x 35
15. lim 16. lim
x→2 x 2 x→7 7x x2 Discussing Concepts
x 1 2 x4 0
17. lim 18. lim 29. Explain why the forms , , and 0 are consid-
x→5 x2 25 x→4 3
x 4 2 0
ered indeterminate forms.
2x2 5x 2 3x2 2
19. lim 20. lim
x→2 5x2 7x 6 x→3 2x2 3
30. Show how the forms and 0 are equivalent to
21. lim (3x)
x→0
e2x 22. lim (4x2)(ln x)
x→0
0
the form .
0
23. lim (0.6x)(ln x)
x→
24. lim (3x)
x→
e2x
SUMMARY
Drawing Slope Graphs
Function Derivative
The smooth, continuous graphs that we use to model
dy
real-life data have slopes (derivatives) at every point on yb 0
dx
the graph except at points that have vertical tangent
lines. When these slopes (derivatives) are plotted, they dy
y ax b a
usually form a smooth, continuous graph—the slope dx
graph (rate-of-change graph, or derivative graph) of the
original graph. Slope graphs tell us a great deal about the dy
y xn nxn1
dx
change that is occurring on the original graph.
dy
Slope Formulas y ex ex
dx
The Four-Step Method of finding derivatives that was dis- dy
cussed in Chapter 2 is invaluable, but it is also cumber- y bx (ln b) bx
dx
some. For this reason, we desire formulas for derivatives dy 1
of the most common functions we encounter. Here is a y ln x
dx x
list of slope (derivative) formulas that you should know.
dy
y kf (x) kf(x)
dx
dy
y f (x) g (x) f (x) g(x)
dx
The Chain Rule
The Chain Rule tells us how to calculate rates of change
for a composite function. We present it in two forms. If
Chapter 3 Concept Check 223
C is a function of p and p is a function of t, then C can be If you need to calculate a rate of change for a general
regarded as a function of t, and f (x)
quotient function, say h (x) g (x) , simply view the
Form 1
dC
dt
dC dp
dp dt
quotient as a product
h(x) f(x)[g(x)]1
dC
Form 2 C(p(t)) p(t) and apply the Product and Chain Rules.
dt
There are other formulas for derivatives that we have
The Product Rule not given. However, we are providing the formulas that
The Product Rule tells us how to calculate rates of change are most useful for the functions encountered in every-
for a product function. If f (x) g (x) h(x), then day situations associated with business, economics, fi-
nance, management, and the social and life sciences. If
df
dx
dg
dx
h (x) g (x)
dh
dx
you ever need them, you can look up other formulas in
a calculus book that emphasizes applications in science
or engineering.
CONCEPT CHECK
Can you To practice, try
• Sketch a general rate-of-change graph? Section 3.1 Activities 3, 9
• Use tangent lines to sketch an accurate slope graph? Section 3.1 Activities 15, 17
• Apply simple derivative rules?
• Power Section 3.2 Activities 7, 13, 19
• Constant Multiplier Section 3.2 Activities 9, 11
• Sum Section 3.2 Activities 15, 17
• Exponential Section 3.3 Activities 5, 7, 11
• Natural Log Section 3.3 Activities 17, 19
• Determine simple rate-of-change formulas? Section 3.2 Activities 29, 31
Section 3.3 Activities 25, 27
• Use the Chain Rule (Form 1)? Section 3.4 Activities 5, 7
• Use the Chain Rule (Form 2)? Section 3.4 Activities 17, 23, 27
• Apply the Chain Rule? Section 3.4 Activities 39, 41
• Use the Product Rule? Section 3.5 Activities 13, 15,19
• Apply the Product Rule? Section 3.5 Activities 29, 31
• Use L’Hôpital’s Rule? Section 3.6 Activities 19, 23, 25
CONCEPT REVIEW
1. Consider the accompanying figure. 3. Mortgage The total amounts of outstanding
mortgage debt in the United States for selected
f(x)
years between 1980 and 2000 are shown in the ac-
4 companying table. Below are a model for the data
and a graph of the model.
3
A (t) 0.173t4 6.24t3 71.06t2
2
32.2t 1460.59 billion dollars
1 t years after 1980.
x A(t)
−3 −2 −1 1 2
(billions of dollars)
−1
7000
−2 6000
−3 5000
−4 4000
3000
a. Note any input values for which the slope of the t
graph is zero. 2000
0 Years since
b. For what input values is the slope of the graph 0 5 10 15 20 1980
positive? negative?
a. Use the graph to estimate how quickly the mort-
c. Are there any input values for which the slope of gage debt amount was changing in 1998.
the graph does not exist?
b. Find the derivative of the equation in 1998. In-
d. Graphically estimate the slope of f at x 2 and terpret your answer.
x 1.
4. Mortgage If N (t) is the total amount of mortgage
e. Sketch a slope graph for the function f.
debt (in billions of dollars) attributable to new
2. Turkey The average annual per capita consump- mortgages t years after 1980, we find the percentage
tion of turkey in the United States between 1980 and of outstanding mortgage debt represented by new
2002 can be modeled by the equation mortgages each year by dividing N (t) by A(t) in Ac-
tivity 3 and multiplying by 100:
8.101
D (t) 10 pounds per person
1 214.8e 0.797t P (t)
N(t)
100 percent t years after 1980
A(t)
t years after 1980. a. Discuss how you find a formula for the rate of
(Source: Based on data from Economic Research change of the percentage of total mortgage debt
Service/USDA.)
dD represented by new mortgages each year.
a. Find a formula for dt .
b. What are the input and output units of measure
b. Find the value of D(10). Interpret your answer. for P?
c. How quickly was the consumption of turkey
growing in 2001?
Year 1980 1985 1990 1992 1994 1996 1998 2000
Total outstanding debt (billions of dollars) 1465 2378 3808 4073 4380 4865 5698 6890
TABLE FOR ACTIVITY 3
(Source: Statistical Abstract, 2001.)
224
Project 3.1 Superhighway
Setting Tasks
The European Communities have decided to build a 1. Find data that give stopping distances as a function
new superhighway that will run from Berlin through of speed, and cite the source for the data. Present
Paris and Madrid and end in Lisbon. This the data in a table and as a graph. Find a model to
superhighway, like some others in Europe, will have fit the data. Justify your choice. Before using your
no posted maximum speed so that motorists may model to extrapolate, consult someone who could
drive as fast as they wish. There will be three toll be considered an authority to determine whether
stations installed, one at each border. Because these the model holds outside of the data range. Consult
stations will be so far apart and motorists may not a reliable source to determine probable speeds
anticipate their need to stop, there has been driven on such a highway. On the basis of your
widespread concern about the possibility of high- model, make a recommendation about where the
speed collisions at these stations. In response to the warning lights should be placed. Justify your
concern for safety, the Committee on Transportation recommendation.
has determined that flashing warning lights should be
2. Find rates of change of your model for at least
installed at an appropriate distance before each toll
three speeds, one of which should be the speed
station. Your firm has been contracted to study
that you believe to be most likely. Interpret these
known stopping distances and to develop a model for
rates of change in this context. Would
predicting where the warning lights should be
underestimating the most likely speed have a
installed and to suggest what other precautions could
serious adverse affect? Support your answer.
be taken to avoid accidents at the toll stations. (Bear
in mind that you are a consultant to Europeans who
wish to see all results in the metric system. However, Reporting
because you work for an American-based company,
you must also include the English equivalent.) 1. Prepare a written report of your study for the
Committee on Transportation.
2. Prepare a press release for the Committee on
Transportation to use when it announces the
implementation of your safety precautions. The
press release should be succinct and should answer
the questions who, what, when, where, and why.
Hint: Drivers’ handbooks and Department of
Transportation documents are possible sources of
data on stopping distance.
225
Project 3.2 Fertility Rates
Setting for the fertility rates for blacks in the United States.
(Note: The Statistical Abstract “Total Black” data
The Statistical Abstract of the United States (2001
for 1984 and earlier includes races other than
edition) reports fertility rates in the United States.
black.)
Data for three fertility rates are located on the
Calculus Concepts website. 4. Add any other recent data for the fertility rate of
whites. (Note: Recent editions of the Statistical
Abstract occasionally update older data points.
Tasks Therefore, you should check the given data and
1. Find a table of fertility rate data in a current change any updated values so that your data agree
edition of the Statistical Abstract, and summarize with the most recent Statistical Abstract.) Find a
in your own words the meaning of “fertility rate.” piecewise model for the updated data. Use this
Assign units to the given fertility rate data. new model to calculate the rates of change that
occurred in the years since 2000.
2. Find a piecewise model consisting of at most three
pieces that fits the given fertility rate data for
whites. Write the derivative function for this Reporting
model. Construct a table of fertility rates and the
Write a report discussing your findings and their
rate of change of fertility rates for all years from
demographic impact. Include your mathematical
1970 through 2000. Discuss any points at which
computations as an appendix.
the derivative of your model does not exist (and
why it does not exist), and explain how you
estimated the rates of change at these points.
3. Using the data in a current edition of the Statistical
Abstract, complete the same analysis as in Task 2
226
Analyzing Change:
Applications of
4
Derivatives
Concepts Outline
4.1 Approximating Change
4.2 Relative and Absolute
Extreme Points
4.3 Inflection Points
4.4 Interconnected Change:
Related Rates
James Leynse/CORBIS SABA
Concept Application
Businesses often experience growth or decline as a result of changes in the economy,
turnover in management, introduction of new products, or even whims of consumers.
A company measures its performance by measuring quantities such as revenue, profit,
productivity, and stock prices. In analyzing its performance, a company may examine
the past behavior of these quantities and seek the answers to such questions as
• When did the quantity exhibit highs and lows?
• Was there a time when the trend of the quantity changed?
• Is it possible to identify when the rate of change of the quantity was greatest?
Such analysis may be helpful as the company seeks to improve its performance. You
will find the tools in this chapter useful in answering questions such as these. One such
exploration for the Polo Ralph Lauren Corporation’s revenue is found in Activity 40 of
Section 4.3.
227
228 CHAPTER 4 Analyzing Change: Applications of Derivatives
Chapter Introduction
This chapter is devoted to exploring the ways in which rate-of-change information
can be used to help analyze change. We begin by examining how a rate of change can
be used to approximate change in a function’s output.
Next, we consider the importance of those places at which the rate of change is
zero. In many cases, these points correspond to local maximum or minimum
function values. Absolute extrema may occur at these points.
Finally, we consider situations in which we find a rate of change of a function with
respect to a variable that does not appear in the equation defining the function. The
resulting equation, which is called a related-rates equation, is useful in describing
how rates of change are interrelated.
Concepts You Will Be Learning
• Using tangent lines to approximate change (4.1)
• Using marginal analysis to estimate output (4.1)
• Using first derivatives to find relative and absolute extreme points (4.2)
• Interpreting extreme points (4.2)
• Writing second derivatives (4.3)
• Using second derivatives to find inflection points (4.3)
• Using second derivatives to determine concavity (4.3)
• Interpreting inflection points (4.3)
• Solving and interpreting related-rates equations (4.4)
4.1 Approximating Change
Recall from our discussion of linear models that the slope of a line y ax b can be
thought of as how much y changes when x changes by 1 unit. For example, the
amount spent on pollution control in the United States during the 1980s can be
described by the equation*
Amount 3.79t 252.2 billion dollars
where t is the number of years since 1900. The slope is $3.79 billion per year, so we can
say that each year during the 1980s, an additional $3.79 billion was spent on pollution
control. It follows that every 2 years spending increased by (2)($3.79) million, every
3 years spending increased by (3)($3.79) million, and so on.
Using Rates of Change to Approximate Change
Because rates of change are slopes, we can apply a similar type of reasoning to functions
other than lines. However, we must be careful in doing so, for although the slope of a
line is constant, the slopes of other functions can change at every point. Recall from the
local linearity discussion in Section 2.2 that in a small enough interval around a point
on a smooth, continuous function, the function and the line tangent to the function at
* Based on data from Statistical Abstract, 1992.
4.1 Approximating Change 229
that point appear to be the same. We call upon this similarity when we use the rate of
change (slope of the tangent) to approximate the actual change in the function.
Concept Development: Approximating Change For example, consider the
average retail price (cost to the consumer) during the 1990s of a pound of salted,
grade AA butter, which can be modeled by the equation*
p(t) 5.17t 2 28.7t 195 cents
where t is the number of years since the end of 1990. We calculate how rapidly the
average price was increasing at the end of 1998 as follows:
dp
2(5.17)t 28.7 10.34t 28.7 cents per year
dt
When t 8,
dp
10.34(8) 28.7 54 cents per year
dt
Price Thus, at the end of 1998, the price of butter was rising by approxi-
(dollars) mately 54 cents per year or $0.54 per year. On the basis of this rate of
3.55 change and the fact that the average price at the end of 1998 was
3.50 approximately p(8) 296 cents $2.96, we estimate that during
p(t) the following year (1999), the price increased by approximately $0.54
to a price of $3.50 and that during the first 6 months of 1999, the price
Tangent line 1
increased by approximately 2($0.54) $0.27 to a price of $3.23.
Figure 4.1 illustrates these approximations. Note in Figure 4.1 that the
approximation of price using the tangent line is relatively close to the
actual price. Of course, because the price function p is not linear, the
t farther we are from the point of tangency, the less accurate is our
2.96 Years tangent-line approximation.
8 8.25 8.5 8.75 9 since 1990 Compare the 6-months and 1-year approximations, as well as
FIGURE 4.1 approximations for time periods of 3 months, 9 months, and 2 years,
with the actual values given by the model. These approximations and
function values are listed in Table 4.1. In Table 4.1 we convert the price from the
equation to dollars by multiplying p(t) by 0.01 dollars per cents.
TABLE 4.1
Difference between
Time from
Approximated price Price from equation equation value and
end of 1998
approximation
3 months $2.96 0.25($0.54) $3.10 p(8.25) $3.10 $0.00
6 months $2.96 0.5($0.54) $3.23 p(8.5) $3.25 $0.02
9 months $2.96 0.75($0.54) $3.37 p(8.75) $3.40 $0.03
1 year $2.96 1($0.54) $3.50 p(9) $3.55 $0.05
2 years $2.96 2($0.54) $4.04 p(10) $4.25 $0.21
* Simplified model based on data from Statistical Abstract, 1998.
230 CHAPTER 4 Analyzing Change: Applications of Derivatives
Note that the shorter the time period, the closer the approximated price is to the
price given by the model. This is no coincidence, because the rate of change is more
likely to be nearly constant over a short time period than over a longer period of time.
Rates of change can often be used to approximate changes in a function, and they
generally give good approximations over small intervals.
To summarize, consider the following statements:
• The change in a function f from x to x h can be approximated by the change in
the line tangent to the graph of the function at x from x to x h when h is a small
number.
• The change in the tangent line from x to x h is
(Slope of the tangent line at x) h f(x) h
The mathematical notation for the statement “The change in a function from x to
x h is approximately the change of the tangent line at x from x to x h” is
f(x h) f(x) f(x) h
for small values of h. We illustrate this statement geometrically in Figure 4.2.
y y = f (x)
Slope of the f (x + h) - f (x)
tangent line = f '(x) ≈ h
f (x + h)
f (x) Rise ≈ f (x + h) - f (x)
Run = h
x
FIGURE 4.2 x x+h
Approximating Change
The approximate change in f is the rate of change of f times a small change
in x. That is,
f(x h) f(x) f(x) h
where h represents the small change in x.
It follows from this formula for approximating change that we can approximate
the function value f(x h) by adding the approximate change to f(x).
Approximating the Result of Change
When x changes by a small amount to x h, the output of f at x h is
approximately the value of f at x plus the approximate change in f .
f(x h) f(x) f(x) h
4.1 Approximating Change 231
It is the formula f(x h) f(x) f(x) h that we used to obtain the approxi-
mated price column in Table 4.1 in the butter price example.
EXAMPLE 1 Using a Tangent-Line Approximation to Estimate Outputs
Temperature The temperature for a 2-hour period during and after a thunderstorm
can be modeled by
T(h) 2.37h4 5.163h3 8.69h2 9.87h 78° Fahrenheit
where h is the number of hours since the storm began.
a. Use the rate of change of T(h) at h 0.25 to estimate by how much the
temperature changed between 15 and 20 minutes after the storm began.
b. Find the temperature and rate of change of temperature at h 1.5 hours.
c. Using only the answers to part b, estimate the temperature 1 hour and 40 min-
utes after the storm began.
d. Sketch the graph of T from h 0 to h 1.75 with lines tangent to the graph
at h 0.25 and h 1.5. On the basis of the graph, determine whether the
answers to parts a and c are overestimates or underestimates of the temperature
given by the model.
Solution
a. First we develop a rate-of-change formula for temperature:
T(h) 9.48h3 15.489h2 17.38h 9.87°F per hour
Then we evaluate that rate-of-change formula at h 0.25:
T(0.25) 9.48(0.25)3 15.489(0.25)2 17.38(0.25) 9.87
6.34°F per hour
In order to calculate the change The change in the temperature between 15 and 20 minutes after the storm
began is approximately (6.34°F/hour)12 hour 0.53°F. The temperature
in input (hour) when minutes 1
are given, we divide the change
in input by the conversion factor fell approximately half a degree.
60 minutes per hour:
b. T(1.5) 77.3°F and T(1.5) 13.3°F per hour
Interval given: “between 15 and
40 2 2
20 minutes” c. Note that 40 minutes 60 hour 3 hour, so 1 hour and 40 minutes 13
Change in input: 5 minutes hours.
5 1
60 hour 12 hour
Temperature at 1 hour temperature at approximate change
Interval given: “1.5 hours”. . .
1
and 40 minutes 12 hours in temperature
”1 hour and 40 minutes”
T 13 T(1.5) T(1.5)6
2 1
Change in input: 10 minutes
10 1
60 hour 6 hour
T 13 77.3°F (13.3°F per hour)6 hour
2 1
77.3°F 2.2°F
79.5°F
232 CHAPTER 4 Analyzing Change: Applications of Derivatives
d. Temperature
(°F)
82
81
80
79
78
77
76
75
74
h
73
0 0.25 0.5 0.75 1 1.25 1.5 1.75 Hours
FIGURE 4.3
Because the tangent line at h 0.25 hour is steeper and therefore, decreasing
more quickly than the graph between h 0.25 (15 minutes) and
h 0.33 (20 minutes), the approximate change in temperature overestimates
the actual decreases (see Figure 4.3). Thus, using the approximate change to
estimate the temperature at h 0.33 gives a temperature that is lower than the
temperature given by the model at h 0.33. The tangent line at h 1.5 is not
as steep as the graph to the right of h 1.5, so our temperature approximation
for h 1.67 is lower than the temperature given by the model. ●
Sometimes it is necessary to use a tangent line to estimate a function output value
because we do not have enough information available to develop a model. At other
times, the tangent-line approximation is used for short-term extrapolation. This is
illustrated in Example 2.
EXAMPLE 2 Using Different Extrapolation Techniques
Population (California) The population of California from 1990 through 2000
can be modeled by
P(t) 6.7t 3 89t 2 624t 29,854 thousand people
t years after July 1, 1990. Figure 4.4 shows the data and model in graphical form.
P(t)
(thousands of people)
34,000
33,000
32,000
31,000
30,000
t
29,000 Years
FIGURE 4.4 0 2 4 6 8 10 after 1990
(Source: Statistical Abstract, 2001. The data are given for July 1 of each year.)
4.1 Approximating Change 233
a. Use the fact that the population in 1999 was 33.145 million and in 2000 was
33.872 million to estimate the population on July 1, 2001.
b. Use the model to estimate the population on July 1, 2001.
c. Find the output of the model and the derivative of the model on July 1, 2000.
Use these values to estimate the population on July 1, 2001.
d. Discuss the assumptions one makes when using each estimation technique in
parts a, b, and c.
Solution
a. The average rate of change between 1999 and 2000 is
33.872 33.145 million people
0.727 million people per year
1 year
Adding this value to the population on July 1, 2000, we obtain an estimate of the
July 1, 2001, population:
33.872 0.727 34.599 million people (that is, 34,599 thousand people)
b. The model estimates the population on July 1, 2001, as
P(11) 34,867 thousand people
c. The derivative of P is
P(t) 20.1t 2 178t 624 thousand people per year
t years after July 1, 1990. On July 1, 2000, the rate of change of the population
was approximately
P(10) 854 thousand people per year
and the population according to the model was 33,894 thousand. Thus
P(11) P(10) P(10)
33,894 854
34,748 thousand people
d. To estimate using only the last two data points (part a) is to assume that the
average rate of change in the population from July 1, 2000, through July 1, 2001,
will be approximately the same as it was from July 1, 1999, through July 1, 2000.
To estimate using only the model (part b) is to assume that future growth will
continue in the manner of the cubic model. To estimate using the derivative of
the model (part c) is to assume two things: that the model may be extended be-
yond 2000 and that the rate of change on July 1, 2000 (as estimated by the model)
is a good predictor of the change in the population during the following year.
All three of these estimates are valid. If it were July 1, 2000, and someone needed
an estimate for the population a year later, that person might use any one of these or
many other techniques to make the prediction. ●
Marginal Analysis
In economics, it is customary to refer to the rates of change of cost, revenue, and profit
with respect to the number of units produced or sold as marginal cost, marginal
234 CHAPTER 4 Analyzing Change: Applications of Derivatives
revenue, and marginal profit. These rates are used to approximate actual change in
cost, revenue, or profit when the number of units produced or sold is increased by one.
The term marginal analysis is often applied to this type of approximation.
EXAMPLE 3 Understanding Marginal Cost
Cost Suppose a manufacturer of toaster ovens currently produces 250 ovens per
day with a total production cost of $12,000 and a marginal cost of $24 per oven.
a. What information does the marginal cost value give the manufacturer?
b. If C(x) is the cost to produce x toaster ovens, what is the notation for marginal
cost?
Solution
a. The marginal cost is the rate of change of cost with respect to the number of
units produced. It is the approximate increase in cost, $24, that will result if
production is increased from 250 ovens per day to 251 ovens per day.
dC
b. The marginal cost is C(x), or dx . In this example, C(250) $24 per oven. ●
We know that profit revenue cost. If p(x), r(x), and c(x) are respectively the
profit, revenue, and cost associated with x units, then we have the relationship
p(x) r(x) c(x). If we take the derivative of this expression, we have
p(x) r(x) c(x), or
Marginal profit marginal revenue marginal cost
From this equation, we see that if marginal profit is to be positive, so that
increased sales will increase profit, then marginal revenue must be greater than
marginal cost. Example 4 explores these relationships.
EXAMPLE 4 Using a Model for Marginal Analysis
TABLE 4.2
Profit A seafood restaurant has been keeping track of the price of its Monday night
all-you-can-eat buffet and the corresponding number of nightly customers. These
Number of Buffet price data are given in Table 4.2.
Customers (dollars)
a. Find a model for the data, and convert it into a model for revenue.
86 7.70
b. If the cost to the restaurant is $4.50 per person regardless of the number of
83 7.90
customers, find models for cost and profit.
80 8.20
c. Find the marginal revenue, marginal cost, and marginal profit values for 50, 91,
78 8.30
and 100 customers. Interpret the values in context.
76 8.50
73 8.70 Solution
70 9.00 a. A linear model for these buffet price data is
68 9.10
B(x) 0.0795x 14.529 dollars price of one buffet meal
4.1 Approximating Change 235
where x is the number of customers, 68 x 86. Revenue is equal to
(price)(number of customers), so it is given by the equation
R(x) 0.0795x 2 14.529x dollars revenue from x buffet meals,
where x is the number of customers, 68 x 86.
b. The cost model is c(x) 4.5x dollars for x customers. The profit model is
p(x) R(x) c(x) 0.0795x2 14.529x 4.5x
0.0795x2 10.029x dollars cost for x buffet meals
where x is the number of customers, 68 x 86.
c. The derivatives of revenue, cost, and profit are, respectively,
R(x) 0.159x 14.529 dollars revenue per customer
c(x) $4.50 cost per customer
p(x) 0.159x 10.029 dollars profit per customer
where x is the number of customers. Evaluating the derivatives at 50, 91 and 100
customers gives the marginal values shown in Table 4.3.
TABLE 4.3
Demand Marginal revenue Marginal cost Marginal profit
(number of (dollars per (dollars per (dollars per
customers) customer) customer) customer)
50 6.58 4.50 2.08
91 0.06 4.50 4.44
100 1.38 4.50 5.88
What do these marginal values tell us? If the buffet price is set on the basis of 50
customers, then revenue is increasing by $6.58 per customer. Because this value is
greater than marginal cost, we see a positive marginal profit. In other words,
increasing the number of customers to 51 (by lowering the price) will increase nightly
revenue by approximately $6.58 and nightly profit by $2.08. It would benefit the
restaurant to increase the number of customers by lowering price.
Similarly, we estimate that if the number of customers is increased from 91 to 92,
then revenue will not change significantly ($0.06 per customer) and profit will,
therefore, decline. With 91 customers, stimulating sales by lowering the price will not
benefit the restaurant.
Finally, note that when price is set so that the restaurant expects 100 customers,
the marginal revenue and profit are negative. Increasing the number of customers (by
decreasing price) to 101 will result in an approximate decrease in nightly revenue of
$1.38 and a decrease in nightly profit of $5.88. That is clearly undesirable. ●
We have seen that the change in a quantity over a small interval can be
approximated using the rate of change of that quantity over the corresponding
interval. This tangent line approximation technique can be especially useful when
there is insufficient data to calculate change directly or when it is desirable to make
short-term extrapolations. The derivative is a useful tool in marginal analysis for
business and marketing applications.
236 CHAPTER 4 Analyzing Change: Applications of Derivatives
9. Insurance A graph showing the annual premium
4.1 Concept Inventory for a one-million-dollar term life insurance policy
as a function of the age of the insured person is
• Change in function change in tangent line close given in the figure.
to the point of tangency
f(x h) f(x) f
(x) h Premium
(dollars)
• f(x h) f(x) f
(x) h for small values of h
• Marginal cost, marginal profit, marginal revenue 12,000
10,000
8000
6000
4.1 Activities 4000
2000
Getting Started Age
0
(years)
1. If the humidity is currently 32% and is falling at a 30 40 50 60 70 80
rate of 4 percentage points per hour, estimate the
Sketch a tangent line at 70 years of age, and use it to
humidity 20 minutes from now.
predict the premium for a 72-year-old person.
2. If an airplane is flying 300 mph and is accelerating at
a rate of 200 mph per hour, estimate the airplane’s 10. Life Expectancy A graph showing world life
speed in 5 minutes. expectancy as a function of the number of decades
since 1900 is given in the figure.
3. If f(3) 17 and f(3) 4.6, estimate f(3.5).
Life expectancy
4. If g(7) 4 and g(7) 12.9, estimate g(7.25). (years)
5. Interpret the following statements. 70
a. At a production level of 500 units, marginal cost 65
is $17 per unit. 60
b. When weekly sales are 150 units, marginal profit 55
is $4.75 per unit.
50
6. Sales Interpret the marginal values given in the Decades
45
since 1900
following statements. 5 6 7 8 9 10
(1950) (1990)
a. When weekly sales are 500 units, marginal
revenue is $10 per unit and marginal cost is (Source: Based on data in The True State of the Planet, ed.
$13 per unit. Ronald Bailey, New York: The Free Press for the Competitive
Enterprise Institute, 1995.)
b. When weekly sales are 10 units, marginal profit is
$3.46 per unit. Sketch the tangent line at 1990, and use it to
estimate the world life expectancy in 2000.
7. Profit A fraternity currently realizes a profit of
$400 selling T-shirts at the opening baseball game of
the season. If its marginal profit is $4 per shirt, Applying Concepts
what action should the fraternity consider taking to
11. Sales A graph of revenue from new car sales and
improve its profit?
associated advertising expenditures for franchised
8. Profit If the marginal profit is negative for the sale new-car dealerships in the United States between
of a certain number of units of a product, is the 1980 (when advertising expenditures were $1.2 bil-
company that is marketing the item losing money lion) and 2000 (when advertising expenditures were
on the sale? Explain. $6.4 billion) is shown in the figure.
4.1 Approximating Change 237
Revenue b. The model whose graph is shown in the figure is
(billions of dollars)
C(x) 6.107x 2 60.799x 315.994
600 million kilograms of CFC 11
where x is the number of years since 1988,
400 0 x 5. What estimate does the model give
for CFC-11 releases in 1993?
Advertising
c. The actual amount of CFCs released into the
200
expenditures atmosphere in 1993 was 149 million kilograms.
130
1.2 2 3 4 5 6 (billions of dollars) Which estimate is the more accurate one?
(Source: Based on data from Statistical Abstract, 2001.) 13. Population The population of South Carolina
between 1790 and 2000 can be modeled by
a. Sketch a tangent line at the point where advertis-
ing expenditures were $6 billion, and use it to P(x) 268.79(1.013087x) thousand people
estimate the revenue from new car sales when x years after 1790.
$6.5 billion was spent on advertising.
(Source: Based on data from Statistical Abstract, 2001.)
b. The model graphed in the figure is
a. Find the rate of change of the population of
R(x) 3.68x 3 47.958x 2 80.759x South Carolina in 2000.
166.98 billion dollars of revenue, b. On the basis of your answer to part a, approxi-
when x billion dollars is spent on advertising, mate by how much the population changed
1.2 x 6.5. What does the model estimate between 2000 and 2003.
as the revenue when $6.5 billion is spent on c. Write an explanation of the procedure you used to
advertising? find the approximate change in the population
c. Which estimate do you believe is the more reli- between 2000 and 2003.
able one? Why? 14. Investment The amount in an investment after t
12. Emissions In 1987, because of concern that years is given by
CFCs (chlorofluorocarbons) have a detrimental A(t) 120(1.126t) thousand dollars
effect on the stratospheric ozone layer, the Mon-
treal Protocol calling for phasing out all CFC pro- a. Give the rate-of-change formula for the amount.
duction was ratified. The accompanying graph b. Find the rate of change of the amount after 10
shows estimated releases of CFC-11 between 1988 years. Write a sentence interpreting the answer.
and 1993. c. On the basis of your answer to part b, determine
Release of
by approximately how much the investment will
CFC-11 grow during the first half of the 11th year.
(millions of kilograms) d. Find the percentage rate of change after 10 years.
Given that A is exponential, what is the
350 significance of your answer?
300
15. Population The population of Mexico between
250
1921 and 2000 can be modeled by
200
150
Years P(t) 7.567e0.026t million people
0 1 2 3 4 5 since 1988
where t is the number of years since 1900.
(Source: Based on data in The True State of the Planet, 1995.) (Source: Based on data from www.inegi.gob.mx, accessed
a. Sketch a tangent line at 1992, and use it to esti- 9/20/02.)
mate CFC-11 releases in 1993. a. How rapidly was the population growing in 1998?
238 CHAPTER 4 Analyzing Change: Applications of Derivatives
b. On the basis of your answer to part a, determine by a. Find and interpret marginal cost at production
approximately how much the population of Mex- levels of 5, 20, and 30 units.
ico should have increased between 1998 and 1999. b. Find the cost to produce the 6th unit, the 21st
16. Study Time Suppose your test grade out of 100 unit, and the 31st unit.
points as a function of the time that you spend c. Why is the cost to produce the 6th unit less than
studying can be modeled by the marginal cost at a production level of 5 units?
Why is the cost to produce the 21st unit greater
G(t) 0.044t 3 0.92t 2 38 points
than the marginal cost at a production level of 20
where t is the number of hours spent studying, units? Why is the cost to produce the 31st unit
0 t 15. greater than the marginal cost at a production
a. Confirm the following assertions: level of 30 units?
i. After 11 hours of study, the slope is 4.2 d. Find a model for average cost.
points per hour, and the grade is 90.5 points. e. Find and interpret the rate of change of average
ii. The grade after 12 hours of study is 94.2 cost at production levels of 5, 20, and 30 units.
points. 19. Bank Account Three hundred dollars is invested
b. Use the information in the first assertion of part in an account with an APR of 6.5% compounded
a to estimate the grade after 12 hours. Is this an monthly.
overestimate or an underestimate of the grade a. Find an equation for the balance in the account
given by the model? Explain. after t years.
17. Mail A model for the number of pieces of first- b. Rewrite the equation in part a to be of the form
class mail handled by the U.S. Postal Service A Pbt .
between 1980 and 2000 is c. How much is in the account after 2 years?
p(x) 17.50 26.53 ln x million pieces d. How rapidly is the value of the account growing
where x is the number of years since 1975. after 2 years?
(Source: Based on data from Statistical Abstract, 2000.) e. Use the answer in part d to approximate how
a. How rapidly was the amount of first-class mail much the value of the account changes during
growing in 1998? the first quarter of the third year.
b. On the basis of the rate of change in 1998, what 20. Bank Account Two thousand dollars is invested
approximate increase would you expect between in an account with an APR of 3.2% compounded
1998 and 1999? monthly.
c. On the basis of the equation, what was the actual a. Write an equation for the balance in the account
increase between 1998 and 1999? after t years.
d. According to the 2001 Statistical Abstract, the ac- b. Rewrite the equation in part a to be of the form
tual amounts of first-class mail handled by the U.S. A Pbt .
Postal Service in 1998 and 1999 were 100.4 million c. How much is in the account after 5 years?
pieces and 101.9 million pieces, respectively. What
was the actual 1998-through-1999 increase? d. How rapidly is the value of the account growing
after 5 years?
e. Compare the increases given by the derivative in
part b, by the model in part c, and by the data in e. Use the answer in part d to approximate how
part d. Which of the three answers is most much the value of the account changes during
accurate? Explain. the first month of the sixth year. How close is this
approximation to the actual change?
18. Costs Suppose production costs for various hourly
production levels of television sets are given by 21. Sales The owner of a concession stand finds that if
he prices hot dogs so as to sell a certain number at
c(p) 0.16p3 8.7p2 172p 69.4 dollars each sporting event, then the corresponding revenues
where p units are produced each hour. are those given in the accompanying table.
4.1 Approximating Change 239
Number of Revenue Balls produced Cost
hot dogs sold (dollars) each hour (hundreds) (dollars)
100 195 2 248
400 620 5 356
700 875 8 432
1000 1000 11 499
1200 1020 14 532
1500 975 17 567
20 625
a. Find a model for the data.
d. Convert the cost model in part a to one for aver-
b. Each hot dog costs the owner of the concession
age cost.
stand $0.50. Use this fact and the model from
part a to write models for cost and profit. e. Find and interpret the rate of change of average
Assume there are no fixed costs. cost for hourly production levels of 300 and of
c. Find marginal revenue, marginal cost, and 1700 golf balls.
marginal profit for sales levels of 200, 800, 1100, 23. CPI Rise in consumer prices is often used as a
and 1400 hot dogs. Interpret your answers. measure of inflation rate. The table shows the CPI
d. Graph revenue, profit, and cost for sales values during the 1980s for several different countries.
between 200 and 1400 hot dogs. How are the a. Find the best models for the data for the United
marginal values in part c related to the graphs? States, Canada, Peru, and Brazil.
22. Production A golf ball manufacturer knows that b. How rapidly were consumer prices rising in each
the cost associated with various hourly production of those four countries in 1987?
levels are as shown in the accompanying table. c. Considering part b, what would you expect
a. Find a model for the data. the CPI to have been in the four countries in 1988?
b. If 1000 balls are currently being produced each 24. Revenue A pizza parlor has been experimenting
hour, find and interpret the marginal cost at that with lowering the price of their large one-topping
level of production. pizza to promote sales. The average revenues from
c. Repeat part b for hourly production levels of 300 the sale of large one-topping pizzas on a Friday
golf balls and of 2100 golf balls. night at various prices are given in the table.
Year
Country 1980 1981 1982 1983 1984 1985 1986 1987
United 100 110.4 117.2 120.9 126.1 130.5 133.1 137.9
States
Canada 100 112.4 124.6 131.8 137.5 143.0 148.9 155.4
Mexico 100 127.9 203.3 410.2 679.0 1071.2 1994.9 4624.7
Japan 100 104.9 107.8 109.9 112.3 114.6 115.3 115.4
Israel 100 217 478 1174 5560 22,498 33,330 39,937
Peru 100 175.4 288.4 609 1280 3372 5999 11,150
Brazil 100 206 407 984 2924 9556 23,436 77,258
Argentina 100 204 541 2403 17,462 134,833 256,308 592,900
(Source: International Marketing Data and Statistics, 1988–89.)
Table for Activity 23
240 CHAPTER 4 Analyzing Change: Applications of Derivatives
Price
September 30 of each year between 1986 and 2000 is
9.25 10.50 11.75 13.00 14.25 shown in the accompanying table.
(dollars)
Revenue
1202.50 1228.50 1210.25 1131.00 1054.50 Year Circulation (millions)
(dollars)
1986 62.5
a. Find a model for the data. 1988 62.7
b. Find and interpret the rate of change of revenue 1990 62.3
at a price of $9.25. 1992 60.1
c. Approximate the change in revenue if the price is 1994 59.3
increased from $9.25 to $10.25. 1996 57.0
d. Find and interpret the rate of change of revenue
1998 56.2
at a price of $11.50.
2000 55.8
e. Approximate the change in revenue if the price is
increased from $11.50 to $12.50. (Source: Statistical Abstract, 1995, 2001.)
f. Explain why the approximate change is an over-
a. Find a model for the data. Why did you choose
estimate of the change in price from $9.25 to
this model?
$10.25 but an underestimate of the change in
price from $11.50 to $12.50. b. According to your model, what is the predicted
circulation of daily English-language newspapers
25. Advertising A sporting goods company keeps in 2007? Is this reasonable?
track of how much it spends on advertising each
c. Estimate how rapidly the newspaper circulation
month and of its corresponding monthly profit.
was changing in 1998.
From this information, the list of monthly advertis-
d. Use the derivative to approximate the change in
ing expenditures and the associated monthly profit
the newspaper circulation between 1990 and 1991.
shown in the accompanying table was compiled.
Discussing Concepts
Advertising Profit 27. Write a brief essay that explains why, when rates of
(thousands of dollars) (thousands of dollars)
change are used to approximate change in a func-
5 150 tion, approximations over shorter time intervals
7 200 generally give better answers than approximations
9 250
over longer time intervals. Include graphical
illustrations in your discussion.
11 325
28. Write a brief essay that explains why, when rates of
13 400
change are used to approximate the change in a con-
15 450 cave-up portion of a function, the approximation is
17 500 an underestimate and, when rates of change are used
19 525
to approximate change in a concave-down portion
of a function, the approximation is an overestimate.
Include graphical illustrations in your discussion.
a. Find a model for the data.
29. Recall that the derivative of a function f with input x
b. Find and interpret the rate of change of profit is defined as
both as a rate of change and as an approximate
change when the monthly advertising expendi- f(x h) f(x)
f(x) lim
ture is $10,000. h→0 h
c. Repeat part b for a monthly advertising expendi- provided that this limit exists. Starting with this
ture of $18,000. equation, derive the formula for the approximation
26. Newspapers The circulation of daily English- of change:
language newspapers in the United States as of f(x h) f(x) f(x) h
4.2 Relative and Absolute Extreme Points 241
4.2 Relative and Absolute Extreme Points
Extreme points are sometimes In this section, we turn our attention to finding high points (maxima) and low points
called optimal points. Relative (minima) on the graph of a function. Points at which maximum or minimum out-
extreme points are sometimes
called local extreme points. puts occur are called extreme points, and the process of optimization involves tech-
niques for finding them. Maxima and minima often can be found using derivatives,
and such points have important applications to the world in which we live.
We use the terms extreme points to refer to maxima and minima (either relative
or absolute, depending on the context) but not to inflection points.
Relative Extrema
We begin by examining a model for the population of Kentucky* from 1980 through
1993:
Population p(x) 0.395x 3 6.67x 2 30.3x 3661 thousand people
where x is the number of years since the end of 1980. It is evident from the graph
in Figure 4.5 that between 1980 and 1993, the population indicated by the model
was smallest in 1980 (3661 thousand people) and greatest in 1993 (3794 thousand
people). However, there are two other points of interest on the graph. Sometime
near 1983, the population reached a peak. We call the peak a relative maximum. It
does not represent the highest overall point, but it is a point to which the
population rises and after which it declines. Similarly, near 1988 the population
reached a relative minimum. There are lower points on the graph, but the popu-
lation decreases during a period of time immediately before this relative
minimum and then increases during a time period immediately following this
relative minimum.
p(x)
Population
(thousands)
3800
3775
3750
3725
3700
3675 x
Years
3650
0 1 2 3 4 5 6 7 8 9 10 11 12 13 since
1980
FIGURE 4.5
* Simplified model based on data from Statistical Abstract, 1994.
242 CHAPTER 4 Analyzing Change: Applications of Derivatives
p(x)
Population
(thousands)
3800
3750
Function
3700
x
Years
3650
0 1 2 3 4 5 6 7 8 9 10 11 12 13
since
1980
p'(x)
(thousands of
people per year)
40
30
Derivative 20
10 x
Years
0
1 2 3 4 5 6 7 8 9 10 11 12 13
since
-10 1980
FIGURE 4.6
It should be intuitively clear from the discussions in Chapter 2 that at a point
where the graph of a smooth, continuous function reaches a relative maximum or
minimum, the tangent line is horizontal and the slope is 0. We can consider this im-
portant link between such a function and its derivative in more detail by examining
the relationship between the Kentucky population function and its slope graph.
Horizontal tangent lines on the population function graph correspond to the points
at which the slope graph crosses the horizontal axis. Figure 4.6 shows the graphs of
the population function and its derivative.
Note that near x 3, the slope graph crosses the x-axis. This is the x-value for
which p(x) is at a local maximum. Likewise, the slope graph crosses the x-axis at the
x-value near 8 for which p(x) is at a local minimum. The connection between the
graphical and algebraic views of this situation is a key feature of optimization
techniques.
Finding the point at which the slope graph of a function crosses or touches
the input axis is the same as finding the point at which the derivative of the
function is zero. That is, the slope graph of a function f crosses or touches the
input x-axis where f (x) 0.
4.2 Relative and Absolute Extreme Points 243
An extreme point is an ordered The derivative of the population of Kentucky function is
pair of the input value and the
output value. Do not confuse dp
extreme values with the inputs. 1.185x 2 13.34x 30.3 thousand people per year
dx
The extreme value is always an
output value. where x is the number of years since 1980, 0 x 13. Setting this expression equal
to zero and solving for x results in two solutions: x 3.14 and x 8.12. This
information, together with the graph of p shown in Figure 4.6, tells us that accord-
ing to the model, the population peaked in early 1984 at approximately
p(3.14) 3703 thousand people. We also conclude that the population declined to
a relative minimum in early 1989. The population at that time was approximately
p(8.12) 3679 thousand people.
It is important to note that relative extrema do not occur at the endpoints of an
interval. A relative maximum is an output value that is larger than all other output
values in some interval around the maximum. Similarly, a relative minimum is an
output value that is smaller than all other output values in some interval around the
minimum.
EXAMPLE 1 Relating Zeros of a Derivative to Relative Extrema of the Function
Baggage Complaints The number of consumer complaints to the U.S. Depart-
ment of Transportation about baggage* on U.S. airlines between 1989 and 2000 can
be modeled by the function
B(x) 55.15x 2 524.09x 1768.65 complaints
where x is the number of years after 1989.
a. Consider the function out of its modeling context. Find any relative maxima
and minima of B on the interval 0 x 11.
b. Graph the function and its derivative. On each graph, clearly mark the input
value that corresponds to the minimum found in part a.
c. Find the year between 1989 and 2000 in which the number of baggage
complaints was at a relative minimum.
Solution
a. Because the graph of B is continuous and smooth over the interval 0 x 11,
we know that the relative extrema occur where the derivative is zero. Thus we set
the derivative equal to zero and solve for x:
B(x) 110.3x 524.09 0
x 4.8; The value of B at x 4.8 is approximately 523.5.
The equation for B is a quadratic with a positive leading coefficient; therefore,
the graph of B is a concave-up parabola. Thus, we know that (4.8, 523.5) is a
relative minimum point. (We will see later in this section that this is also an ab-
solute minimum.)
* Based on data from Statistical Abstract, 2001.
244 CHAPTER 4 Analyzing Change: Applications of Derivatives
b. B(x)
Complaints
1769
Function
Input of
relative minimum x
523
Years
0 2 4 6 8 10 since 1989
B'(x)
Complaints
per year
700
x
Derivative Years
0
2 4 6 8 10 since 1989
Input of
−525 relative minimum
FIGURE 4.7
We see in Figure 4.7 that a relative minimum on the function graph
corresponds to the point at which the derivative graph crosses the x-axis.
c. Because the model gives yearly complaint totals, it must be interpreted discretely.
Thus the minimum number of complaints occurred in either 1993 (x 4) or
1994 (x 5). Checking the value of the function in each of these years, we find
that the least number of complaints was approximately 527 in 1994. ●
Conditions When Relative Extrema Might Not Exist
We have just seen how derivatives can be used to locate relative maxima and minima.
You should use caution, however, and not automatically assume that just because the
derivative is zero at a point, there is a relative maximum or relative minimum at that
point. When a function has a point at which the derivative of the function is zero (that
is, when there is a horizontal tangent line on the graph of the function), one of the
four situations depicted in Figure 4.8 occurs.
It is evident from Figures 4.8c and 4.8d that the graph of a function may have a
horizontal tangent line at a point that is not an extreme point. It is therefore impor-
tant that you graph the function when using derivatives to locate extreme points.
Always begin the process of finding extreme points by graphing the function
to see whether there are any relative maxima or minima before proceeding to
work with derivatives.
4.2 Relative and Absolute Extreme Points 245
f (x)
Function
a a
(a) Relative maximum: (b) Relative minimum:
To the left of input a, To the left of input a,
slopes are positive, and slopes are negative, and x
to the right of a, slopes to the right of a, slopes a
are negative. are positive.
f '(x)
Derivative x
a
a a
(c) Inflection point: (d) Inflection point:
Slopes are positive to the Slopes are negative to the f (a) 0 but the graph of f has no relative max-
left and right of input a. left and right of input a. imum or minimum at a.
FIGURE 4.8 FIGURE 4.9
Let us further investigate the graph shown in Figure 4.8d. Figure 4.9 shows the graph
of the function that appears in Figure 4.8d and its slope (derivative) graph. If you
carefully examine the slope graph, you will see that it touches the x-axis but does not
cross it. Thus f does not have a relative maximum or minimum at a. You may notice
that the derivative graph reaches its maximum at a as it touches the x-axis. Do not
confuse maxima and minima on the derivative graph with maxima and minima of
the original function. In Section 4.3, we will see that maxima and minima of the de-
rivative graph have other important interpretations in terms of the original function.
EXAMPLE 2 Relating Derivative Intercepts to Relative Extrema
Revenue Acme Cable Company actively promoted sales in a town that previously
had no cable service. Once Acme saturated the market, it introduced a new 50-
channel system and raised rates. As the company began to offer its expanded system,
a different company, Bigtime Cable, began offering satellite service with more chan-
nels than Acme and at a lower price. A model for Acme’s revenue for the 26 weeks
after it began its sales campaign is
R(x) 3x 4 160x 3 3000x 2 24,000x dollars
where x is the number of weeks since Acme began sales. The graph of the model is
shown in Figure 4.10. Some points on the graph of R are given in Table 4.4.
246 CHAPTER 4 Analyzing Change: Applications of Derivatives
TABLE 4.4 R(x)
Revenue
Revenue (dollars)
Weeks (dollars)
2 37,232 80,000
6 66,672 70,000
60,000
10 70,000
50,000
14 71,792
40,000
18 78,192
30,000
22 76,912
20,000
26 37,232
10,000
x
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 Weeks
FIGURE 4.10
a. Determine the point at which Acme’s revenue peaked during this 26-week
interval.
b. At what other point is R(x) 0? Explain what happens to Acme’s revenue at
this point.
Solution
a. From the graph, we know that revenue peaks at the relative maximum. A closer
examination of the graphs of R and its derivative (see Figure 4.11) locates this
point near 20 weeks. Solving the equation
R(x) 12x 3 480x 2 6000x 24,000 0
gives two solutions, x 10 weeks and x 20 weeks. Revenue peaked at 20
weeks, with a value of R(20) $80,000. This appears to correspond to the time
immediately before Bigtime Cable’s sales began negatively affecting the Acme
company.
b. The other point at which R(x) 0 is (10, 70,000). The fact that the rate-of-
change equation is zero at two places indicates that there are two places on the
graph with horizontal tangent lines. Indeed, at x 10 weeks, the line tangent
to the curve is horizontal because the curve has leveled off. This corresponds to
the time when Acme’s revenue paused before beginning to increase again. No
local maximum occurs at this point as the derivative is positive for values on
either side of x 10. The slope graph only touches but does not cross the input
axis at x 10. Note the relationships between the rate-of-change graph and
the revenue graph shown in Figure 4.11 and how they connect to the slope
descriptions given in Figure 4.8.
The maximum occurs where the derivative graph crosses the x-axis. The leveling-off
point occurs where the derivative touches, but does not cross, the x-axis.
4.2 Relative and Absolute Extreme Points 247
R(x)
Revenue
(dollars)
80,000
70,000
60,000
50,000
Function 40,000
30,000
20,000
10,000
x
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 Weeks
R'(x)
(dollars per week)
20,000
15,000
10,000
5000
x
Derivative 0
2 4 6 8 10 12 14 16 18 20 22 24 26 Weeks
-5000
-10,000
-15,000
-20,000
FIGURE 4.11 ●
Relative Extrema on Functions That Are Not Smooth
Is it true for every continuous function that relative maxima and minima occur
only where the derivative crosses the input axis? Although this seems to be the case
for most of the functions we use, consider the piecewise continuous function de-
scribing the average concentration (in nanograms per milliliter) of a 360-mg dose
C(h) of a blood pressure drug in a patient’s blood during the 24 hours after the drug is
(ng/mL) given:
400
300
C(h) 0.51h 3 7.65h 2 125 ng/mL when 0 h 10
16.07143h 540.71430 ng/mL when 10 h 24
200
where h is the number of hours since the drug was given.
By calculating the left and right limits as h approaches 10 and comparing them
h
100
Hours with the function value at h 10, we determine that C is continuous for all input
0 10 24
values from 0 to 24. The two portions of the function join at the peak shown in
FIGURE 4.12 Figure 4.12.
248 CHAPTER 4 Analyzing Change: Applications of Derivatives
It is evident from the graph that the highest concentration of the drug occurs 10
hours after the patient receives the initial dose. Thus C(10) 380 ng/mL is the
maximum concentration. However, is C(10) 0? Does the slope graph for C cross
the horizontal axis at h 10? The slope at h 10 exists only if the graph is smooth—
that is, only if the one-sided slopes of the two portions of the graph of C are the same
at h 10. The one-sided slope of the graph on the left is zero at h 10, but the one-
sided slope of the graph on the right is 16.07143 at all points. These differing slopes
cause the sharp point on the graph of C, resulting in a graph that is continuous but
not smooth at h 10. Thus C(10) does not exist. There is no line tangent to the
graph of C at h 10. That C(10) does not exist is illustrated in the graph of C
shown in Figure 4.13. It is therefore possible for an extreme point to occur where the
derivative of the function does not exist, as long as the function has a value at that
point.
C' (h)
(ng/mL/hr)
40
30
20
10
h
0
4 8 12 16 20 24 Hours
-10
-20
FIGURE 4.13
Conditions Where Extreme Points Exist
For a function f with input x, a relative extremum can occur at x c only if
f(c) exists (is defined). Furthermore,
1. A relative extremum exists where f (c) 0 and the graph of f (x) crosses
(not just touches) the input axis at x c.
2. A relative extremum can exist where f (x) exists but f (c) 0 does not
exist. (Further investigation is needed.)
Thus in order to find relative maxima and relative minima of a function f, first deter-
mine the input values for which the derivative of f is zero or undefined, and then
examine a graph of f to determine which of these input values correspond to relative
maxima or relative minima.
4.2 Relative and Absolute Extreme Points 249
Absolute Extrema
Recall the example involving the population of Kentucky at the beginning of this
section. The model for the population of Kentucky from 1980 through 1993 is
Population p(x) 0.395x 3 6.67x 2 30.3x 3661 thousand people
where x is the number of years since the end of 1980. Figure 4.14 shows a graph of p.
p(x)
Population
(thousands)
3800
3775
3750
3725
3700
3675 x
Years
3650
0 1 2 3 4 5 6 7 8 9 10 11 12 13 since
1980
FIGURE 4.14
In the earlier discussion, it was noted that the population of Kentucky has a relative
maximum of approximately 3703 thousand people in early 1984 (x 3.14) and a rel-
ative minimum of approximately 3679 thousand people in early 1989 (x 8.12).
However, it is evident from the graph that between 1980 and 1993 there were years in
which the population was greater than 3703 thousand (the relative maximum) and
there was a time at which the population was less than 3679 thousand (the relative
minimum).
When considering maxima and minima over a closed interval, it is important to
consider not only relative extrema but also absolute extrema. A function can have
several different relative maxima (or minima) in a given closed interval. However, there
can be only one absolute maximum value and one absolute minimum value for that
interval. The absolute extremum can occur at more than one input value on the
interval. An absolute extremum can occur at a relative extremum or it can occur at an
endpoint of the interval. (In the trivial case of a linear function on a closed interval, all
points in the interval are both absolute maxima and absolute minima.)
In the case of the Kentucky population example, over the period between 1980
and 1993, the population of Kentucky reached its absolute maximum (the greatest
output) in 1993. The maximum value of the population function is approximately
3794 thousand people. On the other hand, Kentucky’s population between 1980 and
1993 was at its absolute minimum (the least output) in 1980 when its minimum
value was approximately 3661 thousand people. Note that the absolute maximum
and absolute minimum are stated in terms of the interval of years between 1980 and
1993. Obviously, there was a time before 1980 that the population was even lower,
and it has probably risen since 1993. We discuss the idea of absolute extrema over the
entire set of real numbers following Example 3.
250 CHAPTER 4 Analyzing Change: Applications of Derivatives
EXAMPLE 3 Finding Absolute Extrema on a Given Interval
Baggage Complaints Recall from Example 1 that the number of consumer
complaints to the U.S. Department of Transportation about baggage* on U.S. airlines
between 1989 and 2000 can be modeled by the function
B(x) 55.15x 2 524.09x 1768.65 complaints
where x is the number of years after 1989. Consider the function out of its modeling
context. Find the absolute maximum and the absolute minimum of B on the interval
0 x 11.
Solution
As discussed in Example 1, the graph of B is a concave-up parabola with a minimum
that occurs at x 4.8 and is approximately 523.5.
To find the absolute maximum, we must determine the value of B at the end-
points of the given interval: B(0) 1768.7 and B(11) 2676.8. From our knowledge
of the general shape of the graph of B and from these calculations, we conclude that
the absolute maximum of B is approximately 2676.8 and occurs at the right endpoint
when x 11, and that the absolute minimum of B is approximately 523.5 and occurs
at x 4.8. ●
What if we do not have a specified input interval and are asked to find the absolute
extrema? Let us again consider the function in Example 3 apart from its context and
determine the absolute extrema of the function B(x) 55.15x2 524.09x 1768.65
over all real number inputs. The limit lim B(x) → and the graph continues to
x→
increase infinitely in both directions. Thus there is no absolute maximum over all real
number inputs. We know that the absolute minimum is the relative minimum because
of the end behavior of the function. Therefore, the least output value is approximately
523.5 at x 4.8. This value is the absolute minimum of the function over all real num-
ber inputs.
In general, in order to determine whether an absolute maximum or minimum
exists for a function over all real number inputs, we must analyze the end behavior of
the function as well as consider the outputs of the function at all of the input values
for which the function is discontinuous or has relative extrema.
In order to help you be organized as you practice the concepts presented in this
section, we conclude by outlining the steps for finding relative and absolute
extrema.
* Based on data from Statistical Abstract, 2001.
4.2 Relative and Absolute Extreme Points 251
Finding Extrema
To find the relative maxima and minima of a function f,
Step 1: Determine the input values for which f 0 or f is undefined.
Step 2: Examine a graph of f to determine which input values found in
Step 1 correspond to relative maxima or relative minima.
To find the absolute maximum and minimum of a function f on an interval
from a to b,
Step 1: Find all relative extrema of f in the interval.
Step 2: Compare the relative extreme values in the interval with f(a) and
f(b), the output values at the endpoints of the interval. The largest of these
values is the absolute maximum, and the smallest of these values is the
absolute minimum.
To find the absolute maximum and minimum of a continuous function f
without a specified input interval,
Step 1: Find all relative extrema of f.
Step 2: Determine the end behavior of the function in both directions in
order to consider a complete view of the function. The absolute extrema
either do not exist or are among the relative extrema.
In Activities 3 through 8, mark the location of all relative
4.2 Concept Inventory maxima and minima with an X and the location of all
absolute maxima and minima with an O. For each
• Relative maximum extreme point that is not an endpoint, indicate whether
• Relative minimum the derivative at that point is zero or does not exist.
• An extreme point occurs at an input value 3. y 4. y
• The extreme value is an output value
• Conditions under which extreme points exist
• Absolute maximum
• Absolute minimum
x x
5. y 6. y
4.2 Activities
Getting Started
1. Which of the six basic models (linear, exponential,
logarithmic, logistic, quadratic, and cubic) could
have relative maxima or minima? x x
2. Discuss in detail all of the options you have available
for finding the relative maxima and relative minima
of a function.
252 CHAPTER 4 Analyzing Change: Applications of Derivatives
7. y 8. y a. f (x) b. f(x)
x x
2 2
x x
9. Sketch a graph of a function whose derivative is c. f(x) d. f(x)
zero at a point but that does not have a relative
maximum or minimum at that point.
10. Sketch the graph of a function with a relative
x x
minimum at a point at which the derivative does 2
2
not exist.
11. Identify for which of the graphs shown in parts a
13. Sketch the graph of a function f such that all of the
through d all of the following statements are true.
following statements are true.
For the other graphs, identify which statements are
not true. f (x) 0 for x 1
f (x) 0 for x 2 f (x) 0 for x 1
f (x) 0 for x 2 f (1) does not exist.
f (x) 0 for x 2 14. Sketch the graph of a function f such that all of the
following statements are true.
a. f (x) b. f(x) f (x) 0 for x 1
f (x) 0 for x 1
f (1) 0
x x
2 2
15. Sketch the graph of a function f such that all of the
following statements are true.
f has a relative minimum at x 3.
f has a relative maximum at x 1.
c. f (x) d. f (x) f (x) 0 for x 1 and x 3
f (x) 0 for 1 x 3
x f (1) 0 and f (3) 0
2
x 16. Sketch the graph of a function f such that all of the
2 following statements are true.
12. Identify for which of the graphs shown in parts a f (x) 0 for x 1 and x 3
through d all of the following statements are true. f (x) 0 for 1 x 3
For the other graphs, identify which statements are f (1) 0 and f (3) 0
not true.
For each function in Activities 17 through 22:
f (x) 0 for x 2
a. Write the derivative formula.
f (x) 0 for x 2
b. Locate any relative extrema and identify them as
f (x) 0 for x 2 maxima or minima.
4.2 Relative and Absolute Extreme Points 253
17. f(x) x 2 2.5x 6 9
b. Use the equation °F 5 (°C) 32 to convert
18. g(x) 3x 2 14.1x 16.2 your answer to °F.
19. h(x) x 3 8x 2 6x
26. Population The U.S. Bureau of the Census pre-
20. j(x) 0.3x 3 1.2x 2 6x 4 diction for the percentage of the population 65
21. f(t) 12(1.5t) 12(0.5t) to 74 years old from 2000 through 2050 can be
modeled by
22. j(t) 5et ln t with t 0
p(x) (1.619 105)x 4 (1.675 103)x 3
23. Consider the function
0.050x 2 0.308x 6.693 percent
g(x) 0.04x 3 0.88x 2 4.81x 12.11
where x is the number of years after 2000,
a. Find the relative maximum and relative mini- 0 x 50. Find the absolute maximum and
mum of g. the absolute minimum percentages between 2000
b. Find the absolute maximum and the absolute and 2050. Give the years and the corresponding
minimum of g on the closed interval from x 0 percentages.
through x 14.5. (Source: Based on data from Statistical Abstract, 1998.)
c. Graph the function and its derivative. Indicate
the relationship between the relative maximum 27. River Rate Suppose the flow rate (in cubic feet
and minimum of g and the corresponding points per second, cfs) of a river in the first 11 hours after
on the derivative graph. the beginning of a severe thunderstorm can be
modeled by
24. Consider the function
C(h) 0.865h 3 12.05h 2 8.95h 123.02 cfs
g(x) 5ex ln x 0.2(1.5 x)
where h is the number of hours after the storm
a. Find the relative maximum and relative began.
minimum of g.
a. What were the flow rates for h 0 and h 11?
b. Find the absolute maximum and the absolute
minimum of g on the closed interval from x 0 b. Determine the absolute maximum and minimum
through x 10. flow rates on the closed interval from h 0
through h 11.
c. Graph the function and its derivative. Indicate
the relationship between the relative maximum
28. Lake Level Lake Tahoe lies on the California/
and minimum of g and the corresponding points
Nevada border, and its level is regulated by a 17-gate
on the derivative graph.
concrete dam at the lake’s outlet. By federal court de-
Applying Concepts cree, the lake level must never be higher than 6229.1
feet above sea level. The lake level is monitored every
25. Grasshoppers The percentage of southern Aus-
midnight. The level of the lake from October 1, 1995,
tralian grasshopper eggs that hatch as a function of
through July 31, 1996, can be modeled by
temperature (for temperatures between 7°C and
25°C) can be modeled by L(d) (5.345 107)d 3 (2.543 104)d 2
P(t) 0.00645t 4 0.488t 3 12.991t 2 0.0192d 6226.192 feet above sea level
136.560t 395.154 percent d days after September 30, 1995.
where t is the temperature in °C, 7 t 25. (Source: Based on data from the Federal Watermaster, U.S. De-
partment of the Interior.)
(Source: Based on information in Elements of Ecology, George
L. Clarke, New York: Wiley, 1954, p. 170.) According to the model, did the lake remain be-
a. Find the temperature between 7°C and 25°C that low the federally mandated level from October 1,
corresponds to the greatest percentage of eggs 1995 when d 1, through July 31, 1996 when
hatching. d 304?
254 CHAPTER 4 Analyzing Change: Applications of Derivatives
29. Swim Time Swimming World (August 1992) lists b. Find and interpret the marginal cost when 40
the time in seconds that an average athlete takes to units are produced.
swim 100 meters freestyle at age x years. The data c. On the basis of the model in part a, what is the
are given in the accompanying table. equation for the average hourly cost per unit
when x units are produced each hour?
Age x Time Age x Time d. Find the production level on the interval
(years) (seconds) (years) (seconds) 7 x 61, that minimizes average hourly cost.
8 92 22 50 Give the average hourly cost and total cost at that
10 84 24 49
level.
12 70 26 51 31. Sales Consumer expenditure and revenue are
14 60 28 53
terms for the same thing from two perspectives.
Consumer expenditure is the amount of money that
16 58 30 57 consumers spend on a product, and revenue is the
18 54 32 60 amount of money that businesses take in by selling
20 51 the product. A street vendor constructs a table on
the basis of sales data.
a. Find the best model for the data.
b. Using the model, find the age at which the mini-
Price of a dozen Number of dozens
mum swim time occurs. Also find the minimum
roses (dollars) sold per week
swim time.
10 190
c. Compare the table values with the values in part b.
15 145
30. Costs A company analyzes the production costs
20 110
for one of its products and determines the hourly
operating costs when x units are produced each 25 86
hour. The results are given in the accompanying 30 65
table. 35 52
Production level, x Hourly cost
(units per hour) (dollars)
a. Find a model for quantity sold.
1 210
b. Convert the equation in part a into an equation
7 480 for consumer expenditure.
13 650 c. What price should the street vendor charge to
19 760 maximize consumer expenditure?
25 810 d. If each dozen roses costs $6, what price
31 845 should the street vendor charge to maximize
profit?
37 880
e. Why can the derivatives of the revenue and profit
43 950
equations in this activity not be used to find the
49 1070 street vendor’s marginal revenue and marginal
55 1280 profit from the sale of roses?
61 1590 32. Demand An apartment complex has an exercise
room and sauna, and tenants will be charged a
a. Find a model for hourly cost in terms of yearly fee for the use of these facilities. A survey of
production level. tenants results in these demand/price data.
4.2 Relative and Absolute Extreme Points 255
Quantity Price
34. Price Imagine that you have been hired as direc-
demanded (dollars) tor of a performing arts center for a mid-sized
community. The community orchestra gives
5 250
monthly concerts in the 400-seat auditorium. To
15 170 promote attendance, the former director lowered
25 100 the ticket price every 2 months. The ticket prices
35 50 and corresponding average attendance are given in
the table.
45 20
55 5
Price Average
(dollars) attendance
a. Find a model for price as a function of demand.
35 165
b. On the basis of the price model, give the equation
for revenue. 30 200
c. Find the maximum point on the revenue model. 25 240
What price and what demand give the highest 20 280
revenue? What is the marginal revenue at the 15 335
maximum point?
10 400
33. Refuse The yearly amount of garbage (in millions
of tons) taken to a landfill outside a city during se- a. Find quadratic and exponential models for
lected years from 1975 through 2005 is given in the the data. Which model better reflects the
table. probable attendance beyond a $35 ticket price?
Explain.
Amount
Year (millions of tons) b. On the basis of the model that you believe is
more appropriate, give the equation for revenue.
1975 81
c. Find the maximum revenue and the correspon-
1980 99
ding ticket price and average attendance.
1985 117
d. What other things besides the maximum
1990 122 revenue should you consider in setting the price?
1995 132
2000 145 Discussing Concepts
2005 180 35. If they exist, find the absolute maxima and absolute
2x 2 x 3
a. Find a model for the data. minima of y over all real number
x2 2
b. Give the slope formula for the model. inputs. If an absolute maximum or absolute
c. How rapidly was the amount of garbage taken to minimum does not exist, explain why not.
the landfill increasing in 2005? 36. If they exist, find the absolute maximum and ab-
d. Graph the derivative of your model, and solute minimum of y (2 3x x 2)(3.5 x)2
determine whether your model has a relative over all real number inputs. If an absolute
maximum and/or minimum. Explain how you maximum or absolute minimum does not exist,
reached your conclusion. explain why.
256 CHAPTER 4 Analyzing Change: Applications of Derivatives
4.3 Inflection Points
Unlike relative extrema, which In Section 4.2, we discussed extreme points on a graph. Another important point that
are output values, inflection
can occur on a graph is an inflection point.
points are coordinate points.
Recall from our earlier work with cubic and logistic functions that an inflection
point is a point where a graph changes concavity. On a smooth, continuous graph,
the inflection point can also be thought of as the point of greatest or least slope in a
region around the inflection point. In real-life applications, this point is interpreted
as the point of most rapid change or least rapid change. (See Figure 4.15.)
(a) Inflection point: point (b) Inflection point:
of least slope, point of point of greatest slope,
most rapid decrease point of most rapid
increase
(c) Inflection point: point (d) Inflection point:
of greatest slope, point of point of least slope,
least rapid decrease point of least rapid
increase
FIGURE 4.15
Relative maxima and minima on a smooth, continuous graph can be found by
locating the points at which the derivative graph crosses the horizontal axis. These
points are among those where the original graph has horizontal tangent lines. Inflection
points also can be found by examining the derivative graph and its relation to the
function graph. To find the inflection point on a smooth, continuous graph, we must
find the point where the slope (derivative) graph has a relative maximum or minimum.
That is, we apply the method for finding maxima and minima to the derivative graph.
The Second Derivative
Consider, from the discussion in the previous section, the model for the population
of Kentucky from 1980 through 1993:
Population p(x) 0.395x 3 6.67x 2 30.3x 3661 thousand people
4.3 Inflection Points 257
where x is the number of years since the end of 1980. Graphs of the function and its
derivative are the first two graphs shown in Figure 4.16.
p(x)
Population
(thousands)
3800
3750
Function
3700
x
Years
3650
since
0 1 2 3 4 5 6 7 8 9 10 11 12 13
1980
p'(x)
(thousands of
people per year)
40
30
Derivative 20
10 x
Years
0
since
-10 1 2 3 4 5 6 7 8 9 10 11 12 13
1980
p''(x)
(thousands of
people per year
per year)
15
10
5 x
Second Years
0
derivative since
-5 1 2 3 4 5 6 7 8 9 10 11 12 13
1980
-10
-15
FIGURE 4.16
We wish to determine where the inflection point occurs—that is, where the
population was declining most rapidly. It appears that p has a minimum when p has
an inflection point. In fact, this is exactly the case, so we can find the inflection point
of p by finding the minimum of p. To find the minimum of p for this smooth,
continuous function p, we must find where its derivative crosses the x-axis. The
258 CHAPTER 4 Analyzing Change: Applications of Derivatives
Other notations for the second derivative of p is called the second derivative of p, because it is the derivative of a
derivative of p with respect to x derivative. The second derivative of p is denoted p. In this case, the second derivative
d 2p d2 is given by
include 2 and [p(x)].
dx dx 2
p(x) 2.37x 13.34 thousand people per year per year
where x 0 in 1980.
Because the second derivative represents the rate of change of the first derivative,
the output units of p are
output units of p
t input units of p
Input years
An input/output diagram for this second derivative is shown in Figure
4.17.
The third graph in Figure 4.16 is a graph of the second derivative. The
graph of p crosses the x-axis where the graph of p has an inflection point.
Rule p''
Note that this identifies the minimum point on the graph of the derivative
of p where the tangent line is horizontal.
Output Setting the second derivative equal to zero and solving for x give x 5.63.
p''(t) According to the model, the population was declining most rapidly in
thousand people per mid-1986 at a rate of approximately p(5.63) 7.3 thousand people
year per year
per year. At that time, the population was approximately p(5.63) 3690
FIGURE 4.17 thousand people.
EXAMPLE 1 Using the Second Derivative to Locate an Inflection Point
Education Consider a model for the percentage* of students graduating from
high school in South Carolina from 1982 through 1990 who entered postsecondary
institutions:
f(x) 0.1057x 3 1.355x2 3.672x 50.792 percent
where x is the number of years since 1982.
a. Find the inflection point of the function.
b. Determine the year between 1982 and 1990 in which the percentage was
increasing most rapidly.
c. Determine the year between 1982 and 1990 in which the percentage was
decreasing most rapidly.
Solution
a. Consider the point(s) at which the second derivative is zero. The first derivative
formula for the percentage of students graduating from high school in South
Carolina from 1982 through 1990 is
f (x) 0.3171x 2 2.71x 3.672 percentage points per year
* Based on data in South Carolina Statistical Abstract, 1992.
4.3 Inflection Points 259
where x is the number of years since 1982. The second derivative is
f (x) 0.6342x 2.71 percentage points per year per year
where x is the number of years since 1982. The second derivative is zero when
x 4.27 years after 1982. Next, look at the graph of f shown in Figure 4.18. It
does appear that x 4.27 is the approximate input of the inflection point. The
output is f(4.27) 51.6%, and the rate of change at that point is f (4.27) 2.1
percentage points per year.
b. Although f is a continuous function, and is increasing most rapidly at x 4.27,
we need to find f (4) and f (5) to determine whether enrollment was increasing
faster in 1986 or 1987.
The rate of change of the model in 1986 is f (4) 2.09
percentage points per year. The rate of change in 1987 is
f (5) 1.95 percentage points per year. We can say that
f (x) according to the model, the percentage of South Carolina
(percent) high school graduates who enter postsecondary institutions
was increasing most rapidly in 1986. The percentage of
56 graduates going on for postsecondary education in 1986
54 was approximately f(4) 51.0%. The percentage was in-
Function 52 creasing by about f (4) 2.1 percentage points per year at
50
that time.
48 x Figure 4.18 shows the function, its derivative, and its second
Years derivative. Note again the relationship among the points at
0 2 4 6 8 since 1982 which the second derivative crosses the x-axis, at which the
f '(x) derivative has a maximum, and at which the function has an
(percentage inflection point.
points per year) c. Observe from the graph of f shown in Figure 4.18 that the
4
most rapid decrease appears to occur at one of the endpoints.
We need to assume that the model can be extended to years
2
x before 1982. Then we evaluate f (0).
Derivative 0 Years
2 4 6 8 since 1982
f (0) 3.672 percentage points per year
-2
The percentage was declining most rapidly in 1982 at a rate of
-4 approximately 3.7 percentage points per year. ●
f ''(x)
(percentage points You have just seen two examples of how the second deriva-
per year per year) tive of a function can be used to find an inflection point. It is im-
portant to use the second derivative whenever possible, because it
4
gives an exact answer. Sometimes, however, finding the second
2 derivative of a function can be tedious. In such cases, you will
x have to decide how important extreme accuracy is. If a close ap-
Second
0 Years
derivative proximation will suffice (as is often the case in real-world mod-
2 4 6 8 since 1982
-2 eling), then you may wish to find the first derivative only and use
appropriate technology to estimate where its maximum (or min-
-4
imum) occurs.
FIGURE 4.18
260 CHAPTER 4 Analyzing Change: Applications of Derivatives
EXAMPLE 2 Using Technology to Locate an Inflection Point
Epidemic Consider the following model for the number of polio cases in the
United States in 1949.
42,183.911
C(t) polio cases
1 21,484.253e1.248911t
where t 1 at the end of January, t 2 at the end of February, and so forth. Find
when the number of polio cases was increasing most rapidly, the rate of change of
polio cases at the time, and the number of cases at that time.
Solution
The graphs of C, C, and C are shown in Figure 4.19. We seek the inflection point on
the graph of C that corresponds to the maximum point on the graph of C that
corresponds to the point at which the graph of C crosses the t-axis.
C(t)
Polio cases
42,000
36,000
30,000
24,000
Function
18,000
12,000
6000
t
0 1 2 3 4 5 6 7 8 9 10 11 12 Months
C'(t)
(polio cases
per month)
12,000
9000
Derivative 6000
3000
t
0
0 1 2 3 4 5 6 7 8 9 10 11 12 Months
C''(t)
(polio cases per
month per month)
6600
4400
2200
Second t
0
derivative Months
-2200 1 2 3 4 5 6 7 8 9 10 11 12
-4400
-6600
FIGURE 4.19
4.3 Inflection Points 261
We choose to use technology to estimate the maximum point on the first derivative
graph. It occurs at t 8, where the output is approximately 13,171. It is important to
understand what these numbers represent. The t-value tells us the month in which the
greatest increase occurred: t 8 corresponds to the end of August. The output value is
a value on the first derivative graph with units cases per month. This is the slope of the
graph of C at the inflection point. We can therefore say that polio was spreading most
rapidly at the end of August 1949, at a rate of approximately 13,171 cases per month.
To find the number of polio cases at the inflection point, substitute the unrounded t-
value of the point into the original function to obtain approximately 21,092 cases. Note
that to the right of the inflection point on the graph of C, the number of polio cases con-
tinued to increase, whereas the rate at which polio cases appeared was declining. ●
In some applications, the inflection point can be regarded as the point of
diminishing returns. Consider the college student who studies for 8 hours without a
break before a major exam. The percentage of new material that the student will
retain after studying for t hours can be modeled as
45
P(t) percent
1 5.94e0.969125t
This function has an inflection point at t 1.8. That is, after approximately 1 hour
f(x) and 48 minutes, the rate at which the student is retaining new material begins to di-
minish. Studying beyond that point will improve the student’s knowledge, but not as
quickly. This is the idea behind diminishing returns: Beyond the inflection point, you
gain fewer percentage points per hour than you gain at the inflection point; that is,
a b output increases at a decreasing rate. The existence of this point of diminishing re-
Function x
turns is one factor that has led many educators and counselors to suggest studying in
2-hour increments with breaks in between.
We saw in Section 4.2 that for a smooth, continuous function, a relative
maximum or minimum occurs where the derivative graph crosses the horizontal axis,
but not where the first derivative graph touches the horizontal axis but does not cross
f '(x) it. A similar statement can be made about inflection points. If the second derivative
graph crosses the horizontal axis, then an inflection point occurs on the graph of the
function. The graphs in Figure 4.20 of a function, its derivative, and its second
derivative illustrate this issue. Note that the point at which the second derivative
a b graph touches, but does not cross, the horizontal axis actually corresponds to a
Derivative x
relative maximum on the function graph, not to an inflection point.
Two other situations that could occur are illustrated in Figure 4.21 on page 262.
Note that the graphs of both f and g have inflection points at x A because they
change concavity at that point. However, the second derivatives of f and g never cross
the horizontal axis. In fact, in each case, the second derivative does not exist at x A,
f ''(x)
because the first derivative does not exist there. The function f has a vertical tangent
line at x A, and the function g is not smooth at x A. Even though such situations
as this do not often occur in real-world applications, you should be aware that they
could happen. Keep in mind the following result:
Second a b
x
derivative
At a point of inflection on the graph of a function, the second derivative is
either zero or does not exist. If the second derivative graph is negative on one
side of an input value and positive on the other side of an input value, then
FIGURE 4.20
an inflection point of the function graph occurs at that input value.
262 CHAPTER 4 Analyzing Change: Applications of Derivatives
f (x) g(x)
Function
x x
A A
f '(x) g'(x)
Derivative x
A
x
A
f ''(x) g''(x)
Second
x x
derivative
A A
FIGURE 4.21
Concavity and the Second Derivative
Because the derivative of a function is simply the slope of the graph of that function,
we know that a positive derivative indicates that the function output is increasing and
a negative derivative indicates that the function output is decreasing. The second
derivative provides similar information about where a function graph is concave up
and where it is concave down.
In particular, if the second derivative is negative, it means that the first derivative
graph is declining, which means that the original function graph is concave down.
Similarly, a positive second derivative indicates that the first derivative is increasing,
which means that the original function graph is concave up. And, as we have already
seen, where the second derivative changes from positive to negative or from negative
to positive, the function graph has an inflection point.
Consider, for example, the following information (see Table 4.5) about the
second derivative of a function f. On the interval from x 0 through x 6, the
4.3 Inflection Points 263
second derivative f is continuous and has exactly 3 zeros. The values of the second
derivative at the input values shown in the table provide information about the
concavity of the graph of f at those points. We use “ccu” to indicate concave up and
“ccd” to indicate concave down.
TABLE 4.5
x 0 1 2 3 4 5 6
f (x ) 15 0 1 0 1 0 15
Concavity of f ccu infl. pt. ccd not infl. pt. ccd infl. pt. ccu
f (x) We can conclude from this table that a graph of the function f has two
inflection points at x 1 and x 5. Although the second derivative is zero at
x 3, the function graph does not change concavity at that value, so the
corresponding point is not an inflection point.
Next, consider some values of the first derivative of f (see Table 4.6) on the
closed interval from x 1 through x 6. Here we use the Q symbol to
indicate increasing and the R symbol to indicate decreasing.
TABLE 4.6
x x 1 0 1 2 3 4 5 6
0 1 5
f (x ) 35.6 0 5.6 4.6 4.2 3.8 2.8 8.4
FIGURE 4.22
f (x ) C min. d d d d d d
We conclude that the graph of the function f decreases to a local minimum at zero
and then increases to the right of zero. The graph is concave down between x 1 and
x 5 and concave up to the left of x 1 and to the right of x 5. A possible graph
of f, based on this information, is shown in Figure 4.22.
4.3 Concept Inventory 4.3 Activities
• Inflection point Getting Started
• Second derivative
1. Production The graph shows an estimate of the
• Point of diminishing returns ultimate crude oil production recoverable from
• Conditions under which inflection points exist Earth.
264 CHAPTER 4 Analyzing Change: Applications of Derivatives
In each of the Activities 3 through 6, identify each graph
Production (billion barrels per year)
40
as a function, its derivative, or its second derivative.
(Assume that the input is on the horizontal axis.) Give
30 reasons for your choice.
20 3. a. b.
10
0 Year A B C A B C
1900 1925 1950 1975 2000 2025 2050 2075 2100
(Source: Adapted from François Ramade, Ecology of Natural c.
Resources, New York: Wiley, 1984. Copyright 1984 by John Wiley &
Sons, Inc. Reprinted by permission of the publishers.)
A B C
a. Estimate the two inflection points on the graph.
b. Explain the meaning of the inflection points in
the context of crude oil production.
4. a. b.
2. Advertising The graph shows sales (in thousands
of dollars) for a business as a function of the amount
spent on advertising (in hundreds of dollars).
AB C D F AB D EF
Sales E C
dollars
(thousands) c.
1500
1200
AB C D EF
900
5. a. b.
600
300
Advertising
0 dollars
0 500 1000 1500 2000 2500 (hundreds) –3 –2 –1 –3 –2 –1
c.
a. Mark the approximate location of the inflection
point on the graph.
b. Explain the meaning of the inflection point in
the context of this business. –3 –2 –1
c. Explain how knowledge of the inflection point
might affect decisions made by the managers of
this business.
4.3 Inflection Points 265
6. a. 3.7
b. 21. f(x)
1 20.5e0.9x
79
22. g(t) 13
1 36e 0.2t
–3.4 –2 –3.4 –2 23. f(t) 98(1.2t) 120(0.2t)
24. j(t) 5et ln t with t 0
c. 25. Consider the function
g(x) 0.04x 3 0.88x 2 4.81x 12.11
a. Graph g, g, and g between x 0 and x 15.
Indicate the relationships among points on
–3.4 –2 the three graphs that correspond to maxima,
minima, and inflection points.
b. Find the inflection point on the graph of g. Is it
For each function in Activities 7 through 18, write the a point of most rapid decline or least rapid decline?
first and second derivatives of the function. 26. Consider the function
7. f (x) 3x 7 20
f(x)
1 19e0.5x
8. g(t) e t
a. Graph f, f, and f between x 0 and x 15.
9. c(u) 3u2 7u 5 Indicate the points on the graphs of f and f
that correspond to the inflection point on the
10. k(t) 2.1t 2 7t
graph of f.
11. p(u) 2.1u 3 3.5u 2 16 b. Find the inflection point on the graph of f. Is it a
point of most rapid or least rapid increase?
12. f(s) 32s 3 2.1s 2 7s
Applying Concepts
13. g(t) 37(1.05 t)
27. Study Time The percentage of new material that a
14. h(t) 7 3(0.02t) student will retain after studying t hours without a
15. f(x) 3.2 ln x 7.1 break can be modeled by
45
16. g(x) e 3x ln 3x P(t) percent, 0 t 8
1 5.94e0.969125t
16 a. Find the inflection point on the graph of P, and
17. L(t) interpret the answer.
1 2.1e 3.9t
b. Compare your answer with that given in the
100 discussion at the end of this section.
18. L(t)
1 99.6e0.02t
28. Population The U.S. Bureau of the Census predic-
tion for the percentage of the population that is 65
For each function in Activities 19 through 24, write the
to 74 years old from 2000 through 2050 can be
first and second derivatives of the function, and use the
modeled by
second derivative to determine inputs at which inflec-
tion points might exist. p(x) (1.619 105)x 4 (1.675 103)x 3
0.050x 2 0.308x 6.693 percent
19. f(x) x 3 6x 2 12x
where x is the number of years after 2000.
20. g(t) 0.1t 3 1.2t 2 3.6t 4.5 (Source: Based on data from Statistical Abstract, 1998.)
266 CHAPTER 4 Analyzing Change: Applications of Derivatives
a. Determine the year between 2000 and 2050 a. Sketch the graphs of p and its first and second
in which the percentage is predicted to be derivatives. Label the vertical axes appropriately.
increasing most rapidly, the percentage at that Which points on the derivative graph corre-
time, and the rate of change of the percentage at spond to the inflection points of the original
that time. function graph? Which point on the second
b. Repeat part a for the most rapid decrease. derivative graph corresponds to the inflection
point of the original function graph?
29. Grasshoppers The percentage of southern
Australian grasshopper eggs that hatch as a function b. Find the x-intercepts of the second derivative
of temperature (for temperatures between 7°C and graph, and interpret their meaning in context.
25°C) can be modeled by c. Determine when, according to the model, the
average natural gas price was declining most
P(t) 0.00645t 4
0.488t 3 12.991t 2
rapidly and when it was increasing most rapidly,
136.560t 395.154 percent between 1990 and 2001.
where t is the temperature in °C. d. Repeat part c for the interval 1995 through
(Source: Based on information in Elements of Ecology, George 1999.
L. Clarke. New York: Wiley, 1954, p. 170.)
a. Graph P, P, and P. 32. Cable TV The percentage of households with TVs
b. Find the point of most rapid decrease on the that subscribed to cable can be modeled, for the
graph of P. Interpret your answer. years from 1970 through 2002, by
30. Home Sale The median size of a new single- 62.7
P(x) 6 percent
family house built in the United States between 1 38.7e0.258x
1987 and 2001 can be modeled by the equation
where x is the number of years after 1970.
H(x) 0.359x 3 15.198x 2 221.738x (Source: Based on data from the Television Bureau of
826.514 square feet Advertising.)
where x is the number of years after 1980. a. When was the percentage of households with
(Source: Based on data from the National Association of Home TVs that subscribed to cable increasing the most
Builders Economics Division.) rapidly from 1970 through 2002?
a. Determine the time between 1987 and 2001 b. What were the percentage of households that
when the median house size was increasing least subscribed to cable and the rate of change of the
rapidly. Find the corresponding house size and percentage of households that subscribed to
rate of change in house size. cable at that time?
b. Graph H, H, and H, indicating the relation- 33. Donors The number of people who donated to an
ships between the inflection point on the graph organization supporting athletics at a certain
of H and the corresponding points on the graphs university in the southeast from 1975 through 1992
of H and H. can be modeled by
c. Determine the time between 1985 and 2001
D(t) 10.247t 3 208.114t 2
when the median house size was increasing the
most rapidly. 168.805t 9775.035 donors
31. Price The average price (per 1000 cubic feet) of t years after 1975.
(Source: Based on data from IPTAY Association at Clemson
natural gas for residential use from 1994 through
University.)
2000 is given by
a. Find any relative maxima or minima that occur
p(x) 0.03x 4 0.834x 3 8.45x 2 36.7x on a graph of the function from t 0 through
63.74 dollars t 17.
where x is the number of years since 1990. b. Find the inflection point(s) from t 0 through
(Source: Based on data from Statistical Abstract, 1998, 2001.) t 17.
4.3 Inflection Points 267
c. How do the following events in the history of b. What factors may have caused the inflection
football at that college correspond with the cur- point to occur at the time you found in part a?
vature of the graph? c. Would you expect the most rapid rise to occur at
i. In 1981, the team won the National approximately the same time each year? Explain.
Championship.
37. Labor The personnel manager for a construction
ii. In 1988, Coach F was released and Coach H company keeps track of the total number of labor
was hired. hours spent on a construction job each week during
34. Cable TV The amount spent on cable television the construction. Some of the weeks and the corre-
per person per year from 1984 through 1992 can be sponding labor hours are given in the table.
modeled by
Weeks after the Cumulative
A(x) 0.126x 3 1.596x 2
start of a project labor hours
1.802x 40.930 dollars
1 25
where x is the number of years since 1984. 4 158
(Source: Based on data from Statistical Abstract, 1994.) 7 1254
a. Find the inflection point on the graph of A and the 10 5633
corresponding points on graphs of A and A.
13 9280
b. Find the year between 1984 and 1992 in which
16 10,010
the average amount spent per person per year on
cable television was increasing most rapidly. 19 10,100
What was the rate of change of the amount spent
per person in that year? a. Find a logistic model for the data.
35. Labor A college student works for 8 hours with- b. Find the derivative of the model. What are the
out a break assembling mechanical components. units on the derivative?
The cumulative number of components she has as- c. Graph the derivative, and discuss what informa-
sembled after h hours can be modeled by tion it gives the manager.
62 d. On the interval from week 1 through week 19,
N(h) components when is the maximum number of labor hours
1 11.49e0.654h
per week needed? How many labor hours are
a. Determine when the rate at which she was needed in that week?
assembling components was greatest. e. Find the point of most rapid increase in the
b. How might her employer use the information in number of labor hours per week. How many
part a to increase the student’s productivity? weeks into the job does this occur? How rapidly
36. Lake Level The lake level of Lake Tahoe from is the number of labor hours per week increasing
October 1, 1995, through July 31, 1996, can be mod- at this point?
eled by f. Find the point of most rapid decrease in the
number of labor hours per week. How many
L(d) (5.345 107)d 3 (2.543 104)d 2 weeks into the job does this occur? How rapidly
0.0192d 6226.192 feet above sea level is the number of labor hours per week decreasing
at this point?
d days after September 30, 1995. g. Carefully explain how the exact values for the
(Source: Based on data from the Federal Watermaster, U.S.
points in parts e and f can be obtained.
Department of the Interior.)
a. Determine when the lake level was rising most
rapidly between October 1, 1995, and July 31,
1996.
268 CHAPTER 4 Analyzing Change: Applications of Derivatives
h. If the company has a second job requiring the Amount
same amount of time and the same number of Year (millions of tons)
labor hours, a good manager will schedule the
1970 81
second job to begin so that the time when the
number of labor hours per week for the first job 1975 99
is declining most rapidly corresponds to the time 1980 117
when the number of labor hours per week for the 1985 122
second job is increasing most rapidly. How many
1990 132
weeks into the first job should the second job
begin? 1995 145
38. Advertising A business owner’s sole means of 2000 180
advertising is to put fliers on cars in a nearby
shopping mall parking lot. The table shows the a. Using the table values only, determine during
number of labor hours per month spent handing which 5-year period the amount of garbage
out fliers and the corresponding profit. showed the slowest increase. What was the
average rate of change during that 5-year period?
b. Find a model for the data.
Labor hours Profit
each month (dollars) c. Give the second derivative formula for the
equation.
0 2000
d. Use the second derivative to find the point of
10 3500
slowest increase on a graph of the equation.
20 8500
e. Graph the first and second derivatives, and
30 19,000
explain how they support your answers to part d.
40 32,000
f. In what year was the rate of change of the yearly
50 43,000 amount of garbage the smallest? What were the
60 48,500 rate of increase and the amount of garbage in that
70 55,500 year?
80 56,500 40. Revenue The revenue (in millions of dollars) of
90 57,000 the Polo Ralph Lauren Corporation from 1993
through 2001 is given in the table.
a. Find a model for profit.
Revenue
b. For what number of labor hours is profit Year (millions of dollars)
increasing most rapidly? Give the number of
1993 767.3
labor hours, the profit, and the rate of change of
profit at that number. 1994 810.7
1995 846.6
c. In this context, the inflection point can be
thought of as the point of diminishing returns. 1996 1,019.9
Discuss how knowing the point of diminishing 1997 1,180.4
returns could help the business owner make 1998 1,470.9
decisions related to employee tasks.
1999 1,713.1
39. Refuse The yearly amount of garbage (in millions 2000 1,948.7
of tons) taken to a landfill outside a city during
2001 1,982.4
selected years from 1970 through 2000 is given in
the table. (Source: Hoover’s Online Guide.)
4.3 Inflection Points 269
a. Use the data to estimate the year in which a. Find a cubic model for the data.
revenue was growing most rapidly. b. Give the slope formula for the equation.
b. Find a model for the data. c. Determine when emissions were increasing most
c. Find the first and second derivatives of the model rapidly between 1940 and 1990. Give the year,
in part b. the amount of emissions, and how rapidly they
d. Determine the year in which revenue was grow- were increasing.
ing most rapidly. Find the revenue and the rate of 43. For a function f, f (x) 0 for all real number input
change of revenue in that year. values. Describe the concavity of a graph of f, and
41. Reaction A chemical reaction begins when a sketch a function for which this condition is true.
certain mixture of chemicals reaches 95°C. The 44. Draw a graph of a function g such that g (x) 0 for
reaction activity is measured in Units (U) per all real number input values.
100 microliters (100 L) of the mixture. Measure-
45. For a function f, the following statements are true:
ments at 4-minute intervals during the first
18 minutes after the mixture reaches 95°C are f (x) 0 when 0 x 2
listed in the table. f (x) 0 when x 2
f (x) 0 when 2 x 4
Time a. Describe the concavity of a graph of f between
2 6 10 14 18
(minutes)
x 0 and x 4.
Activity
0.10 0.60 1.40 1.75 1.95 b. Draw two completely different graphs that sat-
(U/100 L)
isfy these second derivative conditions.
(Source: David E. Birch et al., “Simplified Hot Start PCR,” Nature, 46. For a function h, the following statements are true:
vol. 381, May 30, 1996, p. 445.)
h(x) 0 when 0 x 2
a. According to the table, during what time interval h(x) 0 when x 2 and x 0
was the activity increasing most rapidly? h(x) 0 when x 0 and x 2
b. Find a model for the data, and use the equation a. Describe the concavity of a graph of h.
to find the inflection point. Interpret the inflec-
b. Draw a graph that satisfies all three of these
tion point in context.
second derivative conditions.
42. Emissions The table gives the total emissions in
millions of metric tons of nitrogen oxides, NOx, in Discussing Concepts
the United States from 1940 through 1990. 47. Which of the six basic models (linear, exponential,
logarithmic, logistic, quadratic, and exponential)
could have inflection points?
NOx
Year (millions of metric tons) 48. Discuss in detail all of the options that are available
1940 6.9 for finding inflection points of a function.
1950 9.4
1960 13.0
1970 18.5
1980 20.9
1990 19.6
(Source: Statistical Abstract, 1992.)
270 CHAPTER 4 Analyzing Change: Applications of Derivatives
4.4 Interconnected Change: Related Rates
We have seen that many situations in the world around us can be modeled mathemat-
ically. After a situation is modeled with a continuous function, we can use calculus to
analyze changes that have taken place and possibly predict trends in the near future. So
far we have considered how the rate of change of the output variable of a function is af-
fected by a change in the input variable. We now consider the interaction of the rates of
change of input and output variables with respect to a third variable. We shall see that
the interconnection of the input and output variables is also reflected in an interaction
between their rates of change with respect to a third variable.
Interconnected-Change Equations and Implicit Differentiation
4
An equation relating the volume V of a spherical balloon to its radius r is V 3 r 3.
When the balloon is inflated (over time), its volume increases and its radius increases.
Note that even though V and r depend on the amount of time since the balloon began
to be inflated, no time variable appears in the volume equation. In such applications,
we refer to both volume V and radius r as dependent variables because changes in
their value depend on a third variable, time t. We refer to time, the “with respect to”
variable, as the independent variable.
In order to develop an equation relating the rate of change of the balloon’s volume
with respect to time and the rate of change of its radius with respect to time, we
differentiate both sides of the equation with respect to time t. This type of differentia-
tion is known as implicit differentiation. We use the Chain Rule to differentiate
4
the right side of the volume equation, considering 3 r 3 as the outside function and r
as the inside function. The derivative of the inside function with respect to
dr
t is dt . Thus we get
dV 4 dr dr
(3r 2) 4r 2
dt 3 dt dt
This equation shows how the rates of change of the volume and the radius of a sphere
with respect to time are interconnected, and it can be used to answer questions about
those rates. Such an equation is referred to as a related-rates equation. Example 1
demonstrates how to develop related-rates equations.
EXAMPLE 1 Interconnecting Rates of Change
Using each of the given equations, find an equation relating the indicated rates.
dp dx
a. p 39x 4 ; relate dt and dt .
da dt
b. a 4 ln t ; relate dx and dx .
ds dr
c. s rr 2 h2 ; relate dx and dx , assuming that h is constant.
dr dh
d. v r 2h ; relate dt and dt , assuming that v is constant.
Solution
dp
a. Differentiating the left side of p 39x 4 with respect to t gives dt . We use
the Chain Rule to differentiate the right side of the equation with respect to t,
4.4 Interconnected Change: Related Rates 271
considering 39x 4 to be the outside function and x to be the inside
function.
dp dx
39
dt dt
b. Applying the Chain Rule to the right side of a 4 ln t yields
da 1 dt
4
dx t dx
so
da 4 dt
dx t dx
c. Differentiating the right side of s rr 2 h2 requires both the Chain Rule
and the Product Rule. Consider first the application of the Product Rule:
ds
(r)(derivative of r 2 h 2 with respect to x)
dx
(derivative of r with respect to x)(r 2 h 2)
In order to calculate the two derivatives with respect to x, remember to apply
the Chain Rule.
dx(r
ds 1 dr dr
(r) (r 2 h2)1/2(2r) 2 h2)
dx 2 dx
so
ds r2 dr
r 2 h2
dx r h2
2 dx
d. In order to develop a related-rates equation from v r 2h showing the
dr dh
interconnection between dt and dt , we can isolate either r or h on one side and
then find the derivative with respect to t, or we can find the derivative of both
dr
sides of the equation with respect to t and then solve for dt . The two methods are
equivalent, but sometimes one method has less-involved algebra than does the
other method. We choose the second method because it leads to the immediate
removal of v from the equation and thus does not require substitution later.
Differentiating the left side of the equation with respect to t gives zero because
v is constant. Applying the Product Rule and the Chain Rule to the right side of
the equation gives
0 (r 2) dhdt 2rdrdt(h)
dr
Now solve for dt :
dr dh
2rh r 2
dt dt
dr dh
2rh r2
dt dt
dr r 2 dh
dt 2rh dt
dr r dh
●
dt 2h dt
272 CHAPTER 4 Analyzing Change: Applications of Derivatives
Today’s news often brings us stories of environmental pollution in one form or
another. One serious form of pollution is groundwater contamination. If a hazardous
chemical is introduced into the ground, it can contaminate the groundwater and
make the water source unusable. The contaminant will be carried downstream via the
flow of the groundwater. This movement of contaminant as a consequence of the
flow of the groundwater is known as avection. As a result of diffusion, the chemical will
also spread out perpendicularly to the direction of flow. The region that the contam-
ination covers is known as a chemical plume.
Hydrologists who study groundwater contamination are sometimes able to
model the area of a plume as a function of the distance away from the source that the
chemical has traveled by avection. That is, they develop an equation showing the in-
terconnection between A, the area of the plume, and r, the distance from the source
that the chemical has spread because of the flow of the groundwater.
dA
If A is a function of r, then the rate of change dr describes how quickly the area of
the plume is increasing as the chemical travels farther from the source. However, of
dA
greater interest is the rate of change of A with respect to time, dt , and how it is related
dr
to the rate of change of r with respect to time, dt . These relationships are examined in
Example 2.
EXAMPLE 2 Using a Given Equation in a Related-Rates Problem
Groundwater In a certain part of Michigan, a hazardous chemical leaked from
an underground storage facility. Because of the terrain surrounding the storage
facility, the groundwater was flowing almost due south at a rate of approximately
2 feet per day. Hydrologists studying this plume drilled wells in order to sample the
groundwater in the area and determine the extent of the plume. They found that
the shape of the plume was fairly easy to predict and that the area of the plume
could be modeled as
A 0.9604r 2 1.960r 1.124 ln(0.980r 1) square feet
when the chemical had spread r feet south of the storage facility.
a. How quickly was the area of the plume growing when the chemical had traveled
3 miles south of the storage facility?
b. How much area had the plume covered when the chemical had spread 3 miles
south?
Solution
a. First, note that the question posed asks for the rate of change of area with re-
spect to time. However, the equation does not contain a variable representing
time. Second, note that we are given the rate of change of r with respect to time.
If we represent the time in days since the leak began as t, then we know that
dr dA
dt 2 feet per day and we are trying to find dt . We also have an equation that
shows the interconnection between the dependent variables A and r.
4.4 Interconnected Change: Related Rates 273
Therefore, we differentiate with respect to t both sides of the equation that
relates A and r.
0.980 drdt
dA dr dr 1
0.9604(2r) 1.960 0
dt dt dt 0.980r 1
dA 0.980 dr
1.9208r 1.960
dt 0.980r 1 dt
Next, we substitute the known values r 3 miles 15,840 feet and
dr
dt 2 feet per day and then solve for the unknown rate.
dA 0.980
1.9208(15,840) 1.960 (2)
dt 0.980(15,840) 1
60,855 square feet per day
When the chemical has spread 3 miles south, the plume is growing at a rate of
60,855 square feet per day.
b. The area of the plume is
A 0.9604(15,840)2 1.960(15,840) 1.124 ln[0.980(15,840) 1]
241,000,776 square feet
8.6 square miles
When the contamination has spread 3 miles south, the total contaminated area
is approximately 8.6 square miles. ●
Note the method we used to answer the question posed in Example 2. First, we
determined which variables were involved. Second, we identified an equation that
connected the dependent variables. Third, we determined which rates of change we
needed to relate to one another. Next, we found the derivative of each side of the
equation with respect to the independent variable. Finally, we substituted given
quantities and rates into the related-rates equation and solved for the unknown rate.
We summarize this method:
Method of Related Rates
Step 1: Read the problem carefully and determine what variables are
involved. Identify the independent variable (the “with respect to” variable)
and all dependent variables.
Step 2: Use the given equation or find an equation relating the dependent
variables. The independent variable may or may not appear in the equation.
Step 3: Differentiate both sides of the equation in Step 2 with respect to the
independent variable to produce a related-rates equation. The Chain and/or
the Product Rule(s) may be needed.
Step 4: Substitute the known quantities and rates into the related-rates
equation, and solve for the unknown rate.
Step 5: Interpret in context the solution found in Step 4.
274 CHAPTER 4 Analyzing Change: Applications of Derivatives
We illustrate this process in Example 3.
EXAMPLE 3 Using Geometric Relationships in a Related-Rates Problem
Baseball A baseball diamond is a square with each side measuring 90 feet. A baseball
team is participating in a publicity photo session, and a photographer at second base
wants to photograph runners when they are halfway to first base. Suppose that the
average speed at which a baseball player runs from home plate to first base is 20 feet per
second. The photographer needs to set the shutter speed in terms of how fast the dis-
tance between the runner and the camera is changing. At what rate is the distance
between the runner and second base changing when the runner is halfway to first base?
Solution
Step 1: The three variables involved in this problem are time, the distance
between the runner and first base, and the distance between the runner and
Second the photographer at second base. Because speed is the rate of change of dis-
base
tance with respect to time, the independent variable is time, and the two dis-
tances are the dependent variables.
Step 2: We need an equation that relates the distance between the runner
s
First and first base and the distance between the runner and the photographer at
base second base. A diagram can help us better understand the relationship be-
tween these distances. See Figure 4.23.
f
We consider the right triangle formed by the runner, first base, and sec-
et
ond base. Using the Pythagorean Theorem, we know that the relationship be-
fe
90
Home tween f and s in Figure 4.23 is
plate
f 2 902 s 2
FIGURE 4.23 Step 3: Differentiating the equation in Step 2 with respect to t gives
df ds
2f 0 2s
dt dt
ds f df
dt s dt
ds
Step 4: Note that the rate we need to find is dt . We are told that f 45 feet (half of
df
the way to first base) and dt 20 feet per second (negative because the distance
between first base and the runner is decreasing). Substitute these values into the
ds 45
related-rates equation to obtain dt s (20). Find the value of s using the equa-
tion f 2 902 s2 and the fact that f 45 feet. Thus s 10,125100.62 feet,
and we have
ds 45 feet
(20 feet per second) 8.94 feet per second
dt 10,125 feet
Step 5: When a runner is halfway to first base, the distance between the runner and
the photographer is decreasing by about 8.9 feet per second. ●
Understanding that the change in one variable is connected to the change in an-
other variable is important in a variety of applications. The independent variable is
often—but not always—time. Being able to work with related rates can help you solve
problems that occur in many real-world applications.
4.4 Interconnected Change: Related Rates 275
ground is related to the tree’s size. Suppose that a
4.4 Concept Inventory tree transpires
w 31.54 12.97 ln g gallons of water per day
• Dependent and independent variables
• Related-rates equation where g is the girth in feet of the tree trunk, meas-
ured 5 feet above the ground. A tree is currently
• Method of related rates
5 feet in girth and is gaining 2 inches of girth per
year.
a. How much water does the tree currently tran-
spire each day?
dw
b. If t is the time in years, find and interpret dt .
4.4 Activities
16. BMI The body-mass index of an individual who
Getting Started weighs w pounds and is h inches tall is given as
For each of the equations in Activities 1 through 14, 0.45w
write the indicated related-rates equation. B points
0.00064516h2
df dx
1. f 3x ; relate dt and dt . (Source: New England Journal of Medicine, September 14, 1995.)
dp ds
2. p2 5s 2; relate dx and dx . a. Write an equation showing the relationship
dk dx between the body-mass index and weight of a
3. k 6x2 7; relate dy and dy .
woman who is 5 feet 8 inches tall.
dy dx
4. y 9x 3 12x 2 4x 3 ; relate dt and dt . b. Find a related-rates equation showing the inter-
dg dx connection between the rates of change with
5. g e 3x ; relate dt and dt .
dg dx respect to time of the weight and the body-mass
6. g e15x ; relate dt and dt .
2
index.
df dx
7. f 62(1.02x) ; relate dt and dt . c. Consider a woman who weighs 160 pounds and is
dp ds dw
8. p 5 ln(7 s) ; relate dx and dx . 5 feet 8 inches tall. If dt 1 pound per month,
dh da dB
9. h 6a ln a; relate dy and dy . find and interpret dt .
dv dw d. Suppose a woman who is 5 feet 8 inches tall has a
10. v hw(x w) ; relate dt and dt , assuming that h
and x are constant. body-mass index of 24 points. If her body-mass
ds dh index is decreasing by 0.1 point per month, at
11. s rr2 h2 ; relate dt and dt , assuming that r what rate is her weight changing?
is constant.
1 dh dr 17. BMI Refer to the body-mass index equation in
12. v 3 r2h ; relate dt and dt , assuming that v is Activity 16.
constant.
dh dr
a. Write an equation showing the relationship
13. s rr2 h2 ; relate dt and dt , assuming that s between the body-mass index and the height of a
is constant. young teenager who weighs 100 pounds.
dh dw
14. v hw(x w) ; relate dt and dt , assuming that v b. Find a related-rates equation showing the
and x are constant. interconnection between the rates of change
with respect to time of the body-mass index and
Applying Concepts the height.
c. If the weight of the teenager who is 5 feet 3 inches
15. Trees Trees do a lot more than provide oxygen tall remains constant at 100 pounds while she is
and shade. They also help pump water out of the 1
ground and transpire it into the atmosphere. The growing at a rate of 2 inch per year, how quickly
amount of water an oak tree can remove from the is her body-mass index changing?
276 CHAPTER 4 Analyzing Change: Applications of Derivatives
18. Heat Index The apparent temperature A in de- which it will provide. The carrying capacity of a
grees Fahrenheit is related to the actual temperature certain wheat crop has been modeled as
t°F and the humidity 100h% by the equation
11.56P
K people per hectare
A 2.70 0.885t 78.7h D
1.20ht degrees Fahrenheit where P is the number of kilograms of wheat pro-
duced per hectare per year and D is the yearly en-
(Source: W. Bosch and L. G. Cobb, “Temperature Humidity
Indices,” UMAP Module 691, UMAP Journal, vol. 10, no. 3,
ergy requirement for one person in megajoules
Fall 1989, pp. 237–256.) per person.
(Source: R. S. Loomis and D. J. Connor, Crop Ecology:
a. If the humidity remains constant at 53% and the Productivity and Management in Agricultural Systems,
actual temperature is increasing from 80°F at a Cambridge, England: Cambridge University Press, 1982.)
rate of 2°F per hour, what is the apparent a. Write an equation showing the yearly energy
temperature and how quickly is it changing with requirement of one person as a function of the
respect to time? production of the crop.
b. If the actual temperature remains constant at b. With time t as the independent variable, write a
100°F and the relative humidity is 30% but is related-rates equation using your result from
dropping by 2 percentage points per hour, what part a.
is the apparent temperature and how quickly is it
c. If the crop currently produces 10 kilograms of
changing with respect to time?
wheat per hectare per year and the yearly energy
19. Volume The lumber industry is interested in requirement for one person is increasing by
dP
being able to calculate the volume of wood in a tree megajoules per year, find and interpret dt .
trunk. The volume of wood contained in the trunk
of a certain fir has been modeled as 21. Production A Cobb-Douglas function for the
production of mattresses is
V 0.002198d1.739925h1.133187 cubic feet
M 48.10352L0.6K 0.4 mattresses produced
where d is the diameter in feet of the tree, measured
5 feet above the ground, and h is the height of the where L is measured in thousands of worker hours
tree in feet. and K is the capital investment in thousands of
(Source: J. L. Clutter et al., Timber Management: A Quantitative dollars.
Approach, New York: Wiley, 1983.) a. Write an equation showing labor as a function of
a. If the height of a tree is 32 feet and its diameter is capital.
10 inches, how quickly is the volume of the wood b. Write the related-rates equation for the equation
changing when the tree’s height is increasing by in part a, using time as the independent variable
half a foot per year? (Assume that the tree’s and assuming that mattress production remains
diameter remains constant.) constant.
b. If the tree’s diameter is 12 inches and its height is c. If there are currently 8000 worker hours, and if the
34 feet, how quickly is the volume of the wood capital investment is $47,000 and is increasing by
changing when the tree’s diameter is increasing $500 per year, how quickly must the number of
by 2 inches per year? (Assume that the tree’s worker hours be changing in order for mattress
height remains constant.) production to remain constant?
20. Wheat Crop The carrying capacity of a crop is 22. Ladder A ladder 15 feet long leans against a tall
measured in terms of the number of people for stone wall. If the bottom of the ladder slides away
4.4 Interconnected Change: Related Rates 277
from the building at a rate of 3 feet per second, how 28. Salt A leaking container of salt is sitting on a
quickly is the ladder sliding down the wall when the shelf in a kitchen cupboard. As salt leaks out of a
top of the ladder is 6 feet from the ground? At what hole in the side of the container, it forms a coni-
speed is the top of the ladder moving when it hits cal pile on the counter below. As the salt falls
the ground? onto the pile, it slides down the sides of the pile
so that the pile’s radius is always equal to its
23. Height A hot-air balloon is taking off from the height. If the height of the pile is increasing at a
end zone of a football field. An observer is sitting at rate of 0.2 inch per day, how quickly is the salt
the other end of the field 100 yards away from the leaking out of the container when the pile is 2
balloon. If the balloon is rising vertically at a rate of inches tall? How much salt has leaked out of the
2 feet per second, at what rate is the distance container by this time?
between the balloon and the observer changing
when the balloon is 500 yards off the ground? How 29. Yogurt Soft-serve frozen yogurt is being dispensed
far is the balloon from the observer at this time? into a waffle cone at a rate of 1 tablespoon per second.
If the waffle cone has height h 15 centimeters and
24. Kite A girl flying a kite holds the string 4 feet radius r 2.5 centimeters at the top, how quickly is
above ground level and lets out string at a rate of 2 the height of the yogurt in the cone rising when the
feet per second as the kite moves horizontally at an height of the yogurt is 6 centimeters? (Hint:
altitude of 84 feet. Find the rate at which the kite is h
moving horizontally when 100 feet of string has 1 cubic centimeter 0.06 tablespoon and r 6.)
been let out. 30. Volume Boyle’s Law for gases states that when
25. Softball A softball diamond is a square with each the mass of a gas remains constant, the pressure p
side measuring 60 feet. Suppose a player is running and the volume v of the gas are related by the equa-
from second base to third base at a rate of 22 feet per tion pv c , where c is a constant whose value de-
second. At what rate is the distance between the pends on the gas. Assume that at a certain instant,
runner and home plate changing when the runner is the volume of a gas is 75 cubic inches and its pres-
halfway to third base? How far is the runner from sure is 30 pounds per square inch. Because of
home plate at this time? compression of volume, the pressure of the gas is
increasing by 2 pounds per square inch every
26. Volume Helium gas is being pumped into a minute. At what rate is the volume changing at this
spherical balloon at a rate of 5 cubic feet per minute. instant?
The pressure in the balloon remains constant.
a. What is the volume of the balloon when its
Discussing Concepts
diameter is 20 inches?
b. At what rate is the radius of the balloon changing 31. Demonstrate that the two solution methods
when the diameter is 20 inches? referred to in part d of Example 1 yield equivalent
related-rates equations for the equation given in
27. Snowball A spherical snowball is melting, and its that part of the example.
radius is decreasing at a constant rate. Its diameter
decreased from 24 centimeters to 16 centimeters in 32. In what fundamental aspect does the method of
30 minutes. related rates differ from the other rate-of-change
a. What is the volume of the snowball when its applications seen so far in this text? Explain.
radius is 10 centimeters? 33. Which step of the method of related rates do you
b. How quickly is the volume of the snowball consider to be most important? Support your
changing when its radius is 10 centimeters? answer.
278 CHAPTER 4 Analyzing Change: Applications of Derivatives
SUMMARY
This chapter is devoted to analyzing change. The princi- Inflection Points
pal topics are approximating change, optimization, Inflection points are simply points where the concavity
inflection points, and related rates. of the graph changes from concave up to concave down,
Approximating Change or vice versa. Their importance, however, is that they
identify the points of most rapid change or least rapid
One of the most useful approximations of change in a change in a region around the point.
function is to use the behavior of a tangent line to ap- Inflection points can be found where the graph of
proximate the behavior of the function. Because of the the second derivative of a function crosses the horizon-
Principle of Local Linearity, we know that tangent-line tal axis or, sometimes, where the second derivative fails
approximations are quite accurate over small intervals. to exist. In addition to locating input values of inflection
We estimate the output f(x h) as f(x) f(x) h, points, the second derivative of a function can be used to
where h represents the small change in input. determine the concavity of the function.
Optimization Related Rates
The word optimization (as we used it in Section 4.2) When the changes in one or more variables (called
refers to locating relative or absolute extreme points. A dependent variables) depend on a third variable (called
relative maximum is a point to which the graph rises and the independent variable), a related-rates equation can
after which the graph falls. Similarly, a relative mini- be developed to show how the rates of change of these
mum is a point to which the graph falls and after which variables are interconnected. The Chain Rule plays an
the graph rises. There may be several relative maxima important role in the development of a related-rates
and relative minima on a graph. The highest and lowest equation, because the independent variable (which is
points on a graph over an interval or over all possible in- often time) is not always expressed in the equation that
put values, are called the absolute maximum and relates the dependent variables.
absolute minimum points. These points may coincide
with a relative maximum or minimum, or they may oc-
cur at the endpoints of a given interval.
CONCEPT CHECK
Can you To practice, try
• Use derivatives to approximate change? Section 4.1 Activities 11, 17
• Understand marginal analysis? Section 4.1 Activities 5, 21
• Find relative and absolute extreme points? Section 4.2 Activities 17, 21, 23
• Interpret extrema? Section 4.2 Activities 25, 29
Chapter 4 Concept Review 279
• Find inflection points? Section 4.3 Activities 21, 23
• Interpret inflection points? Section 4.3 Activities 31, 35
• Understand the relationship between second derivatives Section 4.3 Activities 5, 25, 33
and concavity?
• Set up and solve related-rates equations? Section 4.4 Activities 19, 25
CONCEPT REVIEW
1. Tourists The number of tourists who visited people, and if M(2000) 4.5 thousand people per
Tahiti each year between 1988 and 1994 can be year, estimate the following:
modeled by (Source: Statistical Abstract, 2001.)
T(x) 0.4804x4 6.635x3 26.126x2 a. How much did the population of French Polyne-
26.981x 134.848 thousand tourists sia increase during the third quarter of 2001?
x years after 1988. b. What was the population at the end of 2001?
(Source: Stephen J. Page, “The Pacific Islands,” EIU International 3. Gender Ratio The table shows the number of
Reports, vol. 1, 1996, p. 91.) males per 100 females in the United States based
a. Find any relative maxima and minima of T(x) on census data. This number is referred to as the
between x 0 and x 6. Explain how you gender ratio.
found the value(s). a. Find a cubic model for the data.
b. Find any inflection points of the graph of T b. Write the second derivative of the equation you
between x 0 and x 6. Explain how you found in part a.
found the value(s). c. In what year does the output of the model exhibit
c. Graph T, T, and T . Clearly label on each graph the most rapid decline? What was the gender ra-
the points corresponding to your answers to tio in that year, and how rapidly was it changing?
parts a and b.
Year Males per 100 females
d. Between 1988 and 1994, when was the number of
tourists the greatest and when was it the least? 1900 104.6
What were the corresponding numbers of 1910 106.2
tourists in those years? 1920 104.1
e. Between 1988 and 1994, when was the number of 1930 102.6
tourists increasing the most rapidly, and when 1940 100.8
was it declining the most rapidly? Give the rates
1950 98.7
of change in each of those years.
1960 97.1
2. Population Let M(t) represent the population of 1970 94.8
French Polynesia (of which Tahiti is a part) at the 1980 94.5
middle of year t. If M(2000) 202 thousand
1990 95.1
2000 96.3
280 CHAPTER 4 Analyzing Change: Applications of Derivatives
4. The graph of the derivative of a function h is shown. b. Does the graph of h have one or more inflection
points between a and b? If so, mark and label on
h' (x) the derivative graph the input location(s) of the
inflection point(s). If not, explain why not.
5. Skid Marks The length of skid marks made on an
asphalt road when a vehicle’s brakes are applied
quickly is given by
x
a 0 b
S 0.000013wv 2 feet
where the vehicle weighs w pounds and is traveling
v mph when the brakes are applied. How quickly is
the length of the skid marks changing when the
velocity of a 4000-pound vehicle traveling at a speed
a. Does the graph of h have a relative maximum
of 60 mph is decreasing by 5 mph per second when
and/or minimum between a and b? If so, mark
the brakes are applied? (Assume that the vehicle’s
on the derivative graph the input location of each
weight remains constant.)
extreme point, and classify the type of extreme
point at that location. If not, explain why not.
Project 4.1 Hunting License Fees
Setting buy another license? How many hunters is that?
What would be the 1987 revenue?
In 1986, the state of California was trying to make a
decision about raising the fee for a deer hunting c. What would be the answers to the questions in
license. Five hundred hunters were asked how much part b if the fee were increased to $300?
they would be willing to pay in excess of the current 2. If the fee increase for 1987 were x dollars, find
fee to hunt deer. The percentage of hunters to agree formulas for the following: (a) the percentage of
to a fee increase of $x is given by the logistic model* hunters willing to pay the new fee, (b) the number
1.221 of the 75,000 hunters willing to pay the new fee,
Percentage (c) the new fee, and (d) the 1987 revenue.
1 0.221e 0.0116x
Suppose that in 1986 the license fee was $100, and 3. Use your formulas to determine the optimal
75,000 licenses were sold. Suppose that you are part license fee. Also, find the optimal fee increase, the
of the 1986 Natural Resources Team presenting a number of hunters who will buy licenses at the
proposed increase in the hunting license fee to the new fee, and the optimal revenue.
head of the California Department of Natural
Resources.
Reporting
1. Prepare a letter to the head of the California
Tasks
Department of Natural Resources. Your letter
1. Illustrate how the model can be used by answering should address the fee increase and expectations
the following questions. for revenue. You should not make it technical but
a. What was the hunting license revenue in 1986? should give some support to back up your
conclusions.
b. Suppose that in 1987 the fee increased to $150.
What percentage of the 1986 hunters would 2. Prepare a technical written report outlining your
* Based on information in Journal of Environmental Economics and findings as well as the mathematical methods you
Management, vol. 24, no. 1, January 1993. used to arrive at your conclusions.
281
Project 4.2 Fundraising Campaign
Setting If so, change your revenue, total cost, and profit
functions to reflect this adjustment and re-analyze
In order to raise funds, the mathematics department
optimal values. Show and explain the mathematics
in your college or university is planning to sell T-
that underlies your reasoning.
shirts before next year’s football game against the
school’s biggest rival. Your team has volunteered to Discuss the sensitivity of the demand function to
conduct the fundraiser. Because several other student changes in price (check rates of change for $20,
groups have also volunteered to head this project, $14, and $8). Does the demand function have an
your team is to present its proposal for the fund inflection point? If so, find it. Find the rate of
drive, as well as your predictions about its outcome, change of demand with respect to price at this
to a panel of mathematics faculty. point, and interpret its meaning and impact in this
context. How would the sensitivity of the demand
Task A curve affect your decisions about raising or
lowering your selling price?
Follow the tasks for Project 1.2 on page 92. You will
find a partial price listing for the T-shirt company on On the basis of your findings, predict the optimal
the Calculus Concepts website. selling price, the number of T-shirts you intend to
print, the costs involved, the number of T-shirts
Task B you expect to sell before realizing a profit (that is,
the break-even point), and the expected profit.
1. Review your work for Task A. If you wish to make
any changes in your marketing scheme, you
should do so now. If you decide to make any Reporting
changes, make sure that the polling that was done
1. Write a report for the mathematics department
is still applicable (for example, you will not be able
concerning your proposed campaign. They will be
to change your target market). Change (if
interested in the business interpretation as well as
necessary) any models from Task A to reflect any
in an accurate description of the mathematics
changes in your marketing scheme.
involved. Be sure to include graphical as well as
2. Use the models of demand, revenue, total cost, mathematical representations of your demand,
and profit developed in Task A to proceed with revenue, cost, and profit functions. (Include
this section. graphs of any functions and derivatives that you
Determine the selling price that generates use. Include your calculations and your survey as
maximum revenue. What is maximum revenue? Is appendices.) Do not forget to also cover Task A in
the selling price that generates maximum revenue this report.
the same as the price that generates maximum 2. Make your proposal and present your findings to a
profit? What is maximum profit? Which should panel of mathematics professors in a 15-minute
you consider (maximum revenue or maximum presentation. Your presentation should be restricted
profit) in order to get the best picture of the to the business interpretation, and you should use
effectiveness of the drive? overhead transparencies of graphs and equations of
Re-evaluate the number of shirts you may wish to all models and derivatives, along with any other
sell. Will this affect the cost you determined above? visual aids that you consider appropriate.
282
Accumulating Change:
Limits of Sums and the
5
Definite Integral
Concepts Outline
5.1 Results of Change and Area
Approximations
5.2 Accumulation Functions
5.3 The Fundamental Theorem
5.4 The Definite Integral
5.5 Average Value and Average
Rate of Change
5.6 Integration by Substitution
or Algebraic Manipulation
James Leynse/CORBIS SABA
Concept Application
If crude oil is flowing into a holding tank through a pipe, the rate at which the oil is
flowing determines how quickly the holding tank will fill. It is possible that the rate varies
with time and can be mathematically modeled. Such a model could then be used to
answer questions such as
• What is the change in the amount of oil in the tank during the first 10 minutes the oil
is flowing into the tank?
• If the initial amount of oil in the tank was 5000 cubic feet, how much oil was in the
tank after 10 minutes?
• How long can oil continue to flow into the tank before the tank is full?
Questions such as these can be answered using definite integrals. Examples of this type
of problem appear in Activities 1 and 4 of the Chapter 5 Concept Review.
283
284 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Chapter Introduction
Chapters 1 through 4 focused on the derivative, one of the fundamental concepts
in calculus. Now we begin a study of another fundamental concept in calculus, the
integral. As before, our approach is through the mathematics of change.
We start by analyzing the accumulated change in a quantity and how it is related
to areas of regions between the graph of the rate-of-change function for that quantity
and the horizontal axis. As we refine our thinking about area, we are led to consider
limits of sums, which show us how to account for the results of change in terms of
integrals. Integrals, as we shall see, are connected to derivatives by the Fundamental
Theorem of Calculus.
We conclude by considering the difference of two accumulated changes and using
integrals to calculate averages.
Concepts You Will Be Learning
• Approximating area using rectangles (5.1)
• Interpreting the area between a graph and the horizontal axis (5.1)
• Approximating area using a limiting value of sums of areas of rectangles (5.1)
• Sketching and interpreting accumulation graphs (5.2)
• Finding general antiderivatives (5.3)
• Finding and interpreting specific antiderivatives (5.3)
• Recovering a function from its rate-of-change equation (5.4)
• Evaluating and interpreting definite integrals (5.4)
• Calculating and interpreting the area between two curves (5.5)
• Calculating and interpreting the average value of a function and average rates
of change (5.5)
• Using integration by substitution to write an accumulation function (5.6)
5.1 Results of Change and Area Approximations
In our study of calculus so far, we have concentrated on finding rates of change. We
now consider the results of change.
Accumulated Change
Suppose that you have been driving on an interstate highway for 2 hours at a constant
speed of 60 mph. Because velocity, v, is the rate of change of distance traveled, s, with
respect to time, we write a function for velocity mathematically as
v(t) s(t) 60 mph, where t is the time in hours, 0 t 2. A graph of this rate
function over a 2-hour period of time appears in Figure 5.1.
5.1 Results of Change and Area Approximations 285
v(t) At a constant rate of 60 mph, the distance traveled during a time period of t hours
Velocity is (rate)(time) (60 mph)(t hours) 60t miles. Geometrically, we view this multi-
(mph) plication as giving the area of the region between the rate-of-change graph and the
horizontal axis over any time period of length t hours. Figure 5.2 illustrates this fact
for the 1-hour time period between 0.5 hour and 1.5 hours, for the 15-minute time
60 period between 1 hour 45 minutes and 2 hours, and for the first t hours of the trip.
Now, imagine that after a 2-hour drive at a constant speed of 60 mph, you accel-
erate at a constant rate to 75 mph over a 10-minute interval. You then maintain that
constant 75-mph speed during the next half hour. A graph of your speed appears in
t Figure 5.3a.
1 2 Hours We know that the distance traveled during the times when speed is constant is the
speed multiplied by the amount of time driven at that speed (represented by regions
FIGURE 5.1
v(t) v (t) v (t)
Velocity Velocity Velocity
(mph) (mph) (mph)
60 60 60
A B C
t t t
0.5 1 1.5 2 Hours 1 1.75 2 Hours t Hours
(a) Area A (60 mph) (1 hour) (b) Area B (60 mph) (0.25 hour) (c) Area C (60 mph) (t hours)
60 miles 15 miles 60t miles
Distance traveled during 1 hour Distance traveled during 15 minutes Distance traveled during t hours
FIGURE 5.2
R1 and R3 in Figure 5.3b). But how can we calculate the distance driven between
2 hours and 2 hours 10 minutes when the speed is increasing linearly? If we knew the
average speed over the 10-minute interval, then we could multiply that average speed
1
by 6 of an hour to obtain the distance traveled.
v(t) v(t)
Velocity Velocity
(mph) (mph)
75 75
R2 R3
60 60
R1
t t
40 40
1 2 2hr Hours 1 2 2hr Hours
40
2hr min 40
2hr min
10min 10min
(a) (b)
FIGURE 5.3
286 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
In this case, the average speed is simply the average of the beginning and ending
speeds during the 10-minute interval. Thus we have
Distance traveled
60 mph 75 mph
16 hour
between 2 hours average
(time)
and 2 hours speed 2
10 minutes
We represent the distance traveled during the first 2 hours while the driver
maintained a continuous speed of 60 mph as the area of a rectangle labeled R1.
Similarly, we represent the distance traveled during the next 10 minutes while the
driver is accelerating from 60 mph to 75 mph as the area of the trapezoid
Area side 1 2 side 2 (base) labeled R2 in Figure 5.3b. We represent the distance
traveled in the last 30 minutes while the driver maintained a continuous speed of
75 mph as the area of the rectangle labeled R3. Thus, the distance traveled during the
2-hour 40-minute interval is calculated as
Distance
area of region R1 area of region R2 area of region R3
traveled
60 mph 75 mph 1
(60 mph)(2 hr) 2 6
hr (75 mph)
1
2
hr
120 miles 11.25 miles 37.5 miles
168.75 miles
The distance traveled over a specific interval is given by the area of the region between
the rate-of-change graph and the horizontal axis.
Recall that the rate of change of a continuous function is negative on any interval
where the function is decreasing. In Example 1, we introduce the idea of signed area
to indicate that an area trapped below the horizontal axis represents an accumulated
decrease in the function.
EXAMPLE 1 Relating Signed Area to Accumulated Change
r(t) Draining Water
(gallons per minute) A water tank drains at a rate of r(t) 2t gallons per minute t minutes after the wa-
2 4 6 t ter begins draining. The graph of the rate-of-change function is shown in Figure 5.4.
(minutes)
−4 a. What are the units on height, width, and area of the region between the time
−8 axis and the rate graph?
−12 b. Determine the change in the volume of water in the tank during the first
4 minutes that the water was being drained.
FIGURE 5.4
Solution
a. The height corresponds to the output units, which are gallons per minute. The
width is time (in minutes). In the calculation of area, the height and width are
multiplied, giving area in gallons.
5.1 Results of Change and Area Approximations 287
r(t) b. To find the change in the volume of the water, we find the area of the region
(gallons per minute) between the time axis and the rate graph. This area is shaded in Figure 5.5.
4 t
(minutes) The region is a triangle with base of length 4. The height is determined by the
r(4) = –8
function value at t 4. Note that the function value is negative: r(4) 8.
–8 The negative sign indicates that the rate-of-change graph lies below the hori-
zontal axis. We use the signed value for height to remind us that the region lies
FIGURE 5.5 beneath the horizontal axis and represents a decrease in the amount of water in
the tank. The signed area of the region is
By signed area we mean the area
of the region with a negative 1
sign in front of the area value to (4 minutes) (8 gallons per minute) 16 gallons
2
indicate that the region lies
below the input axis.
Thus, during the first 4 minutes that the tank was being drained, the volume of
water in the tank changed by 16 gallons. That is, the amount of water in the
tank decreased by 16 gallons. ●
Results of Change
If a rate-of-change function of a quantity is continuous over a closed interval
[a, b], the accumulated change in the quantity between input values of a and
b is represented as the area or signed area of the region between the rate-of-
change function for that quantity and the horizontal axis, provided the
function does not cross the horizontal axis over the interval [a, b].
Left- and Right-Rectangle Approximations
The functions in the preceding discussion and example were carefully chosen so that
their rate-of-change graphs were easy to obtain and the areas of the desired regions
were easy to calculate. But most real-life situations are not so simple. There are two
issues that we must face:
• Obtaining the rate-of-change function for the quantity of interest
• Calculating the area of the desired region between the rate-of-change function
and the horizontal axis
In many cases, we must resort to approximating both the rate-of-change function
and the desired area. Consider the example of a store manager of a large department
store who wishes to estimate the number of customers who came to a Saturday sale
from 9 A.M. to 9 P.M. The manager stands by the entrance for 1-minute intervals at dif-
ferent times throughout the day and counts the number of people entering the store.
He uses these data as an estimate of the number of customers who enter the store each
minute. The manager’s data may look something like Table 5.1.
The number of customers who attended the sale could be calculated by summing
the number of customers who entered the store during each hour for every hour of
the 12-hour sale. We do not have enough information to determine the exact num-
ber of customers who entered the store each hour; however, we can estimate the
number. To build a model for the rate-of-change data, we choose to convert each of
288 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
TABLE 5.1
Hours since Customers (est) Cust. Hours since Customers (est) Cust.
Time 9 A.M. per minute per hour Time 9 A.M. per minute per hour
9:00 A.M. 0 1 60 2:30 P.M. 5.5 5 300
9:45 A.M. 0.75 2 120 3:10 P.M. 6.167 5 300
10:15 A.M. 1.25 3 180 4:00 P.M. 7 4 240
11:00 A.M. 2 4 240 5:15 P.M. 8.25 4 240
11:45 A.M. 2.75 4 240 6:30 P.M. 9.5 3 180
12:15 P.M. 3.25 5 300 7:30 P.M. 10.5 2 120
1:15 P.M. 4.25 5 300 8:15 P.M. 11.25 2 120
the above observation times to hours after 9:00 A.M. Thus t 0 at 9:00 A.M. and
t 11.25 at 8:15 P.M. A model for the estimated number of customers per hour
between 9 A.M. and 8:15 P.M. is
c(t) (0.6)t3 17t2 119t 53 customers per hour
where t is the number of hours after 9:00 A.M. The graph of this model is shown in
Figure 5.6.
c(t)
(customers
per hour)
300
240
180
120
60 t
Hours
0
0 3 5 7 9 11
after
9 A.M.
(9 A.M.) (12 P.M.) (2 P.M.) (4 P.M.) (6 P.M.) (8 P.M.)
FIGURE 5.6
Concept Development: Left-Rectangle Approximation To estimate the num-
ber of customers who attended the sale, we use the model to estimate the number of
customers per hour entering the store at the beginning of each hour. Summing the
estimates for each of the 12 hours results in an estimate for the total number of cus-
tomers. This process is the same as drawing a set of 12 rectangles under the graph of
c, one for each hour, and using the sum of their areas to estimate the total number of
customers who came to the sale. (See Figure 5.7.)
5.1 Results of Change and Area Approximations 289
c(t)
(customers
per hour)
300
240
180
120
60
t
0 Hours
FIGURE 5.7 0 3 7 11 after 9 A.M.
In Figure 5.7, the height of each rectangle is the function c evaluated at the left
endpoint of the base of the rectangle. It is for this reason that we call these rectangles
left rectangles. Because we are using 12 rectangles of equal width to span the 12-hour
sale, the width of each rectangle is 1 hour. We use Table 5.2 to keep track of the areas
that we are summing.
TABLE 5.2
Left endpoint
of rectangle Height of rectangle Area of rectangle*
t hours since 9 A.M. c (t ) customers per hour customers
9 A.M.: t 0 c(0) 53 53
10 A.M.: t 1 c(1) 156 156
11 A.M.: t 2 c(2) 229 229
noon: t 3 c(3) 275 275
1 P.M.: t 4 c(4) 298 298
2 P.M.: t 5 c(5) 302 302
3 P.M.: t 6 c(6) 290 290
4 P.M.: t 7 c(7) 266 266
5 P.M.: t 8 c(8) 233 233
6 P.M.: t 9 c(9) 196 196
7 P.M.: t 10 c(10) 156 156
8 P.M.: t 11 c(11) 119 119
Sum of areas 2573 customers
Change in number of customers 2573 customers
*These values were obtained using the unrounded model.
Thus, using 12 rectangles, we estimate that 2573 customers came to the Saturday sale.
Note the importance in this example of estimating the number of customers en-
tering the store per hour. If the number of customers counted per minute is not ad-
justed, then the area calculated by multiplying height (measured in customers per
290 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
minute) by width (measured in hours) is not the number of customers. Make sure the
units correspond so that the result of their multiplication gives the desired units. Also
note that our decision to use time intervals of 1 hour was arbitrary.
The previous discussion illustrates a way to approximate accumulation using left-
rectangle areas. In some situations, choosing rectangles whose heights are measured
at the right endpoint of the base of each rectangle may be preferable. Such rectangles
are called right rectangles. The use of right rectangles is illustrated in the following
example.
EXAMPLE 2 Using Right Rectangles to Approximate Change
Medicine A pharmaceutical company has tested the absorption rate of a drug that
is given in 20-milligram doses for 20 days. Researchers have modeled the rate of
change of the concentration of the drug, measured in micrograms per milliliter per
day, g/mL/day, in the bloodstream as
r(x) 1.7(0.8x) g/mL/day when 0 x 20
10.2 3 ln x g/mL/day when 20 x 30
where x is the number of days after the drug is first administered. Figure 5.8 illustrates
a graph of this rate-of-change function. Note that r(x) is defined with two equations,
with each equation applying to a specific input interval. The amount of the drug in
the bloodstream increases while the patient is taking the drug and decreases after the
patient stops taking the drug. The function r(x) is an example of a piecewise defined
function.
Rate of
change of drug
concentration
(µg/mL/day)
1.5
1
0.5
0 Day
4 8 12 16 20 24 28 30
−0.5
−1
−1.5
FIGURE 5.8
a. Use the model and right rectangles of width 2 days to estimate the change in the
drug concentration in the bloodstream while the patient is taking the drug.
b. Use the model and right rectangles of width 2 days to estimate the change in the
drug concentration in the bloodstream for the first 10 days after the patient
stops taking the drug.
c. Combine your answers to parts a and b to estimate the change in the drug
concentration in the bloodstream over the 30-day time period.
5.1 Results of Change and Area Approximations 291
Solution
a. To determine the change in the drug concentration from the beginning of day
1 (x 0) through the end of day 20 (x 20), we use the exponential portion
of the model and ten right rectangles, as shown in Figure 5.9 and Table 5.3.
r(x) TABLE 5.3
(µg/mL/day)
Right endpoint Height of the
1.5 of rectangle rectangle Area of rectangle
1 x days r (x ) (g/mL/day) g/mL
0.5 2 1.09 2.18
0 Day 4 0.70 1.39
4 8 12 16 20 24 28 30
−0.5 6 0.45 0.89
−1 8 0.29 0.57
−1.5 10 0.18 0.37
FIGURE 5.9 12 0.12 0.23
14 0.07 0.15
16 0.05 0.10
18 0.03 0.06
20 0.02 0.04
Sum of areas 5.97 g/mL
Change in concentration 5.97 g/mL
Over the 20 days that the patient took the drug, the drug concentration
increased by approximately 6.0 micrograms per milliliter.
b. To determine the change in the concentration from the beginning of day 21
through the end of day 30, we use the log portion of the model and five right
rectangles, as shown in Figure 5.10 and Table 5.4.
r(x) TABLE 5.4
(µg/mL/day)
Right endpoint Signed height of Signed area of
1.5 of rectangle the rectangle rectangle
1 x days r (x ) (g/mL/day) g/mL
0.5 22 0.93 1.85
24 28 30
0 Day 24 0.67 1.33
4 8 12 16 20 0.43 1.85
−0.5 26
−1 28 0.20 0.41
−1.5 30 0 0.01
FIGURE 5.10 Sum of signed areas 4.44 g/mL
Change in concentration 4.44 g/mL
292 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
From the beginning of day 21 through the end of day 30, the drug concentra-
tion decreased by approximately 4.4 g/mL.
c. To determine the change in concentration from the beginning of day 1 through
the end of day 30, we need only subtract the amount of decline from the
amount of increase.
5.97 g/mL 4.44 g/mL 1.54 g/mL
The drug concentration increased by approximately 1.5 g/mL from the
beginning of day 1 through the end of day 30. ●
Part c of Example 2 illustrates the results of change using signed areas. In gen-
eral, if a function f gives the rate of change of a function F and if the function f is
sometimes positive and sometimes negative between inputs a and b, then the accu-
mulated (or net) change in F is equal to the area of the region lying under the
graph of f and above the x-axis minus the area of the region lying above the graph
of f and below the x-axis. (See Figure 5.11.) In other words, the accumulated change
in a quantity is equal to the sum of the signed areas of the regions between the rate-
of-change function for that quantity and the horizontal axis over the interval from
a to b.
f(x)
R1
a
x
c b
R2
Accumulated change area of R1 area of R2
sum of signed areas
FIGURE 5.11
Midpoint-Rectangle Approximation
It is often true that if a left-rectangle approximation is an overestimate, then a right-
rectangle area will be an underestimate, and vice versa. For example, consider using
four left and four right rectangles to approximate the area of the region between the
function f(x) 4 x2 and the x-axis between x 0 and x 2. Figure 5.12 shows
the rectangles and the approximate areas.
5.1 Results of Change and Area Approximations 293
f(x) f(x)
2 2
0 x 0 x
0 0.5 1.0 1.5 2 0 0.5 1.0 1.5 2
(a) Left-rectangle (b) Right-rectangle
approximation 3.5 approximation 2.5
FIGURE 5.12
Note from Figure 5.12 that the left-rectangle approximation is an overestimate and
the right-rectangle approximation is an underestimate.
We next consider approximating area using a third type of rectangle. The mid-
point-rectangle approximation uses rectangles whose heights are calculated at the
midpoints of the subintervals. Midpoint-rectangle approximation is illustrated in Ex-
ample 3.
EXAMPLE 3 Using Midpoint Rectangles to Approximate Change
Consider again the region between the function f(x) 4 x 2 and the x-axis from
x 0 to x 2. Use four midpoint rectangles to approximate the area of this region.
(Note that we are actually seeking the area of one quarter of a circle that has a radius
of 2. Using the formula for the area of a circle, A r 2, we find that the area of the
entire circle is 4. The area of one quarter of this circle is 3.14.)
Solution
Table 5.5 shows the calculations for the areas of the midpoint rectangles shown in
Figure 5.13.
f(x)
TABLE 5.5
2
Rectangle Midpoint Height of Width of
number of interval rectangle rectangle Area
1 0.25 f(0.25) 1.98 0.5 0.99
2 0.75 f(0.75) 1.85 0.5 0.93
3 1.25 f(1.25) 1.56 0.5 0.78
0 x 4 1.75 f(1.75) 0.97 0.5 0.48
0 0.5 1 1.5 2
Total area of four midpoint rectangles 3.18
Four rectangles
FIGURE 5.13 The area of the region is approximately 3.18 (using four midpoint rectangles for
estimation). ●
294 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
f(x) Finding a Limit of Area Estimates
2 The midpoint-rectangle approximation is much closer to the actual area than are the
two other approximations. This is often the case.
We can improve the estimate of the area using more rectangles (see Figure 5.14).
In Table 5.6, we show the results when eight midpoint rectangles are used to estimate
the area of one quarter of the circle. (Remember that we know this area is .)
0 x
0 0.5 1 1.5 2
Eight rectangles
TABLE 5.6
FIGURE 5.14
Rectangle Midpoint Height of Width of
number of interval rectangle rectangle Area
1 0.125 f(0.125) 1.996 0.25 0.499
2 0.375 f(0.375) 1.965 0.25 0.491
3 0.625 f(0.625) 1.900 0.25 0.475
4 0.875 f(0.875) 1.798 0.25 0.450
5 1.125 f(1.125) 1.654 0.25 0.413
6 1.375 f(1.375) 1.452 0.25 0.363
7 1.625 f(1.625) 1.166 0.25 0.291
8 1.875 f(1.875) 0.696 0.25 0.174
Total area of eight midpoint rectangles 3.156
f(x) Using eight midpoint rectangles gives a better estimate than using four midpoint
rectangles. Sixteen midpoint rectangles appear to fit even better than eight midpoint
2 rectangles. (See Figure 5.15).
In fact, if we let the number of rectangles increase without bound, the area of the
rectangles will approach . As we use more rectangles, the shaded region accounted
for by the rectangles more closely approximates the region of interest. This fact will
help us approximate the area. If we make a table of the approximations, we may be
able to recognize a limiting value—that is, a value to which the approximations seem
to be getting closer and closer as the number of rectangles becomes larger and larger.
0 x
0 0.5 1 1.5 2
Sixteen rectangles
FIGURE 5.15
From now on, whenever we speak of approximating areas using rectangles,
unless the type of rectangle is specified, we will use midpoint rectangles
because they generally give the best approximations when compared with an
equal number of either left or right rectangles.
5.1 Results of Change and Area Approximations 295
Accumulated Change and the Definite Integral
Because the more rectangles we use to approximate area, the better we expect the ap-
proximation to be, we are led to consider area as the limiting value of the sums of ar-
eas of approximating midpoint rectangles as the number of rectangles increases with-
out bound. Let f be a function that is continuous and non-negative over the interval
from a to b. (See Figure 5.16.) Partition the interval from a to b into n subintervals of
ba
equal length x n , and on each subinterval construct a rectangle of width x
whose height is given by the value of f at the midpoint of the subinterval. Figures 5.17
through 5.20 show the rectangles when n 4, 8, 16, and 32.
f(x) f(x)
f(x)
x x x
a b a b a b
Regions under curve Four rectangles Eight rectangles
FIGURE 5.16 FIGURE 5.17 FIGURE 5.18
The heights of the rectangles are given by the values
f(x)
f(x1), f(x2), … , f(xn)
where x1, x2, … , xn are the midpoints of the subintervals. Each rectangle has width
x, so the areas of the rectangles are given by the values
f(x1)x, f(x2)x, … , f(xn)x
x
a b and the sum [f(x1) f(x2) … f(xn)]x is an approximation to the area of the
Sixteen rectangles region between the graph of f and the x-axis from a to b. As our examples have shown,
FIGURE 5.19
the approximations generally improve as n increases. In mathematical terms, the area
of the region between the graph of f and the x-axis from a to b is given by a limit of
sums as n increases:
f(x) Area lim [f(x1) f(x2) … f(xn)]x
n→
Area Beneath a Curve
x
Let f be a continuous non-negative bounded function from a to b. The area of
a b
the region between the graph of f and the x-axis from a to b is given by the limit
Thirty-two rectangles
Area lim [f(x1) f(x2) … f(xn)]x
FIGURE 5.20 n→
ba
where x1, x2, … , xn are the midpoints of n subintervals of length x n
between a and b.
296 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
EXAMPLE 4 Relating Accumulated Change to Signed Area
Wine Consumption The rate of change of the per capita consumption of wine in
the United States from 1970 through 1990 can be modeled* as
W(x) (1.243 104)x 3 0.0314x 2 2.6174x 71.977 gallons
per person per year
where x is the number of years since the end of 1900. A graph of the function is shown
in Figure 5.21.
W(x)
Rate of change of
wine consumption
(gallons per person per year)
0.14
0.08
R1
x
0.02
Years
-0.02 70 80 A
A 90 since
R2 1900
-0.08
-0.14
FIGURE 5.21
a. Find the input value of the point labeled A.
b. From 1970 through 1990, according to the model, when was wine consumption
increasing and when was it decreasing?
c. Use a limiting value of sums to estimate the areas, to two decimal places, of the
regions labeled R1 and R2. Interpret your answers.
d. According to the model, what was the change in the per capita consumption of
wine from the end of 1970 through 1990?
Solution
a. Solving W(x) 0 gives A 83.97, corresponding to the end of 1984.
b. Wine consumption was increasing where the rate-of-change graph is posi-
tive—from 1970 through 1984 (x 83.97)—and was decreasing from 1984
through 1990, where the rate-of-change graph is negative.
c. In order to find the areas of the regions R1 and R2, we must know the lower limit
and upper limit (that is, the endpoints) of each region. Because W(x) 0 when
*Based on data from Statistical Abstract, 1994.
5.1 Results of Change and Area Approximations 297
x 83.97, we use lower limit x 70 and upper limit x 83.97 for R1 and
lower limit x 83.97 and upper limit x 90 for R2.
The area of region R1 is determined by examining sums of areas of midpoint
rectangles for an increasing number of subintervals until a trend is observed, as
shown in Table 5.7. The area of region R1 is approximately 1.34 gallons per per-
son. This indicates that wine consumption increased by approximately 1.34
gallons per person from the end of 1970 through 1984 (x 83.97).
TABLE 5.7
Number of Approximation of
rectangles area of region R1
5 1.35966
10 1.34130
20 1.33671
40 1.33556
80 1.33527
160 1.33520
Limit 1.34
We calculate the signed area of region R2 in a similar way, as shown in Table 5.8.
The area of region R2 is approximately 0.45 gallon per person. This is an esti-
mate for the decrease in per capita wine consumption from 1984 (x 83.97)
through 1990.
TABLE 5.8
Number of Approximation of
rectangles signed area of region R2
5 0.44927
10 0.44870
20 0.44856
40 0.44852
Limit 0.45
d. To determine the net change in the per capita consumption of wine from the
end of 1970 through 1990, we subtract the decrease (area of R2) from the in-
crease (area of R1).
Net change 1.34 0.45 0.89 gallon per person
We see an approximate net increase of 0.89 gallon per person from the end of
1970 through 1990. ●
298 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
In the previous example, we saw that when a rate-of-change function has both
positive and negative outputs over a given input interval, the accumulated change
in the quantity function is equal to the sum of the signed areas of the regions
between the graph of the rate-of-change function and the horizontal axis. That is,
we sum the signed areas of rectangles over the entire interval to calculate accumu-
lated change. We illustrate this type of calculation using a limit of sums and the
rate-of-change function W(x) (1.243 104)x 3 0.0314x 2 2.6174x 71.977
gallons per person per year, where x is the number of years since the end of 1900,
to estimate the net change in the per capita consumption of wine from the end of
1970 through 1990. Refer to Figure 5.22. Ignoring the horizontal-axis intercept,
partition the interval from x 70 to x 90 into equal subintervals, and use the
formula for W and midpoint rectangles to calculate the sums of signed-rectangle
areas in Table 5.9.
W(x)
Rate of change of
wine consumption
(gallons per person per year)
0.14
0.08
x
0.02
Years
−0.02 70 80 A 90 since
1900
−0.08
−0.14
FIGURE 5.22
TABLE 5.9
Number of Approximate
rectangles accumulated change
5 0.92848
10 0.89712
20 0.88928
40 0.88732
80 0.88683
160 0.88671
320 0.88668
Limit 0.89 gallon per person
5.1 Results of Change and Area Approximations 299
The limit of sums lim [W(x1) W(x2) … W(xn)]x from x 70 to
n→
x 90 yields the same net change as the subtraction of nonsigned areas in part d of
Example 4.
It is tedious to write lim [W(x1) W(x2) … W(xn)]x from x a to
n→
x b every time we want to denote the limit of sums. Mathematically, we use the
f(x)dx. The sign ∫ is called an integral sign and resembles an
b
shorthand notation
a
elongated S to remind us that we are finding the limit of sums. The values a and b
identify the input interval, f is the function, and the symbol dx reminds us of the
f(x)dx is known
b
width x of each subinterval. When a and b are specific numbers,
a
as a definite integral. We now formally define the definite integral and accumulated
change.
Accumulated Change and the Definite Integral
Let f be a continuous or piecewise continuous bounded function from a to b.
The accumulated change in f from a to b is given by the limit
f(x)dx
b
lim [f(x1) f(x2) … f(xn)]x
n→
a
ba
where x1, x2, … , xn are the midpoints of n subintervals of length x n
between a and b.
f(x)dx is called the definite integral of f from a to b.
b
EXAMPLE 5 Relating Accumulated Change and the Definite Integral
Wine Consumption Recall from Example 4 the function
W(x) (1.243 104)x 3 0.0314x 2 2.6174x 71.977 gallons per person
per year
which gives the rate of change of per capita consumption of wine in gallons per per-
son per year from 1970 through 1990, where x is the number of years since 1900.
83.97 90 90
a. Find the values of W(x)dx, W(x)dx, and W(x)dx.
70 83.97 70
b. What was the per capita wine consumption in 1990?
300 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Solution
a. Refer to Figure 5.23. The area of region R1, calculated in Example 4, is the value
of the definite integral of W from 70 to 83.97.
W(x)
Rate of change of
wine consumption
(gallons per person per year)
0.14
0.08
R1
x
0.02
Years
-0.02 70 80 A
A 90 since
R2 1900
-0.08
-0.14
FIGURE 5.23
83.97
W(x)dx 1.34 gallons per person
70
The signed area of region R2, calculated in Example 4, is the value of the defi-
nite integral from 83.97 to 90.
90
W(x)dx 0.45 gallon per person
83.97
The limit of sums, calculated using Table 5.9, gives the accumulated change in
per capita wine consumption or the definite integral of W from 70 to 90.
90
W(x)dx 0.89 gallon per person
70
b. The definite integral tells us the change in a quantity over an interval, not the
value of the quantity at the endpoint of the interval. We do not know the per
capita wine consumption in 1990. ●
The first important concept in this section is that the accumulated change in a
quantity can be found by finding areas between rate-of-change graphs and the
horizontal axis. Sometimes we can calculate exact areas using geometric formulas. At
other times, we only approximate area using sums of areas of rectangles.
The other important concept in this section is the idea of finding the limiting
value of sums of areas of rectangles. We used the definite integral to indicate that we
were finding accumulated change in a quantity over an interval with specific numer-
ical endpoints. In many cases, it is possible to find accumulation functions that will
give as output the accumulated change in a rate-of-change function. We explore these
accumulation functions graphically in Section 5.2 and algebraically in Section 5.3.
5.1 Results of Change and Area Approximations 301
ii. the area of the region between the graph of A
5.1 Concept Inventory and the t-axis?
• Area or signed area of a region between a rate- 3. Braking The distance required for a car to stop is
of-change function and the horizontal axis between a function of the speed of the car when the brakes
a and b accumulated change in the amount are applied. The rate of change of the stopping
function between a and b distance could be expressed in feet per mile per
hour, where the input is the speed of the car, in
• Left-rectangle approximation miles per hour, when the brakes are applied.
• Right-rectangle approximation a. What does the area of the region between the
• Midpoint-rectangle approximation rate-of-change graph and the input axis from
• Area limiting value of sums of areas of midpoint 40 mph to 60 mph tell us about the car?
rectangles b. What are the units on
f(x)dx accumulated change in F,
b i. the heights and widths of rectangles used to
•
a estimate the area in part a?
where F f for x between a and b ii. the area in part a?
4. Emissions The atmospheric concentration of
CO2 is growing exponentially. If the growth rate in
5.1 Activities ppm per year is C(t), where t is the number of years
since 1980, what are the units on
Getting Started a. the area of the region between the graph of C and
the t-axis from t 0 to t 20?
1. Bacteria The growth rate of bacteria (in thousands
b. the heights and widths of rectangles used to
per hour) in milk at room temperature is B(t), where
estimate the area in part a?
t is the number of hours that the milk has been at
room temperature. We wish to use rectangles to esti- c. the change in the CO2 concentration from 1980
mate the area of the region between a graph of B and through 2000?
the t-axis. What are the units on 5. Population The rate of change of the population of
a. the heights of the rectangles? a country, in thousands of people per year, is mod-
b. the widths of the rectangles? eled by the function P with input t, where t is the
number of years since 1995. What are the units on
c. the areas of the rectangles?
a. the area of the region between the graph of P and
d. the area of the region between the graph of B and
the t-axis from t 0 to t 10?
the t-axis?
20
e. the accumulated change in the number of bacte- b. P(t)dt ?
ria in the milk during the first hour that the milk 10
is at room temperature? c. the change in the population from 1995 through
2000?
2. Road Test The acceleration of a car (in feet per
second per second) during a test conducted by a car 6. Lake Level During the spring thaw a mountain
manufacturer is given by A(t), where t is the num- lake rises by L(d) feet per day, where d is the num-
ber of seconds since the beginning of the test. ber of days since March 31. What are the units on
a. What does the area of the region between the a. the area of the region between the graph of L and
portion of the graph of A lying above the t-axis the d-axis from d 0 to d 15?
and the t-axis tell us about the car? 30
b. L(d)dd?
b. What are the units on 16
i. the heights and widths of rectangles used to c. the amount by which the lake rose from March
estimate area? 31 to May 31?
302 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
.7 lAgae rGow
th When warm water is released into when (f) _______ boxes are produced each day.
a river from a source such as a power plant, the in- The area between the rate-of-change-of-profit
creased temperature of the water causes some algae function and the production-level axis between
to grow and other algae to die. In particular, blue- production levels of 100 and 200 boxes each day has
green algae that can be toxic to some aquatic life units (g) _______. If p(b) represents the rate of
thrive. If A(c) is the growth rate of blue-green algae change of profit (in dollars per box) at a daily
(in organisms per °C) and c is the temperature of the 400
production level of b boxes, would p(b)db
water in °C, interpret the following in context: 300
35 be more than, less than, or the same value as
a. A(c)dc 200
25 p(b)db? (h) _______
b. The area of the region between the graph of A 100
and the c-axis from c 30°C to c 40°C 10. o
Cst The graph shows the rate of change of cost for
.8 Stock Value The value of a stock portfolio is an orchard in Florida at various production levels
growing by V(t) dollars per day, where t is the num- during grapefruit season. Fill in the blanks in the fol-
ber of days since the beginning of the year. Interpret lowing cost function discussion. If it is not possible to
the following in context: determine a value, write NA in the corresponding
blank.
a. The area of the region between the graph of V
and the t-axis from t 0 to t 120
Rate of
240
b. V(t)dt change of cost
120 (dollars per carton)
.9 Production The graph in the figure shows the rate
of change of profit at various production levels for a Weekly
pencil manufacturer. Fill in the blanks in the fol- production
level
lowing discussion of the profit. If it is not possible to 50 100 200 250
(cartons)
determine a value, write NA in the corresponding
blank.
Rate of change
of profit Cost is increasing when between (a) _______ and
(dollars per box) (b) _______ cartons of grapefruit are harvested
each week. The cost to produce 100 cartons of
grapefruit each week is (c) _______ dollars. The
cost is lower than nearby costs at a production level
of (d) _______ cartons, and it is higher than nearby
Daily costs at a production level of (e) _______ cartons of
production grapefruit each week. The cost is increasing most
100 200 500 level rapidly when (f) _______ cartons are produced
(boxes)
each week. The area between the rate-of-change-of-
cost function and the production-level axis between
Profit is increasing when between (a) _______ and production levels of 50 and 150 cartons each week
(b) _______ boxes of pencils are produced each day. has units (g) _______. If c(p) represents the rate of
The profit when 500 boxes of pencils are produced change of cost (in dollars per carton) at a weekly
each day is (c) _______ dollars. Profit is higher than production level of p cartons of grapefruit, would
nearby profits at a production level of (d) _______
250
boxes each day, and it is lower than nearby profits at c(p)dp be greater than, less than, or the same
200
a production level of (e) _______ boxes of pencils 100
each day. The profit is decreasing most rapidly value as c(p)dp? (h) ________
50
5.1 Results of Change and Area Approximations 303
11. The graph of a function g is shown. 14. Approximate the area of the region beneath the
graph of f(x) ex from x 1 to x 1 using
2
g(x)
four left rectangles, four right rectangles, and four
midpoint rectangles.
a. In each case,
i. sketch the graph of f from x 1 to x 1.
ii. label the points on the x-axis, and draw the
rectangles.
x iii. calculate the approximate areas.
0 4 8
b. Proceed as in part a to approximate the area of
the region beneath the graph of f from x 1 to
x 1 using eight left rectangles, eight right rec-
tangles, and eight midpoint rectangles.
a. Discuss how to approximate the area of the region c. The actual area, to nine decimal places, of the
region beneath the graph of f(x) ex is
2
between the graph of g and the x-axis between
x 0 and x 8 with eight right rectangles that 1.493648266. Which of the approximations
have the same width. Copy the graph, and draw found in part b is the most accurate?
the rectangles on the figure. Applying Concepts
b. Repeat part a using eight left rectangles.
15. Production The graph shows two estimates, la-
12. Stock Value On January 4, 2000, DuPont stock beled A and B, of oil production rates (in billions of
was worth $65 per share. barrels per year).
a. Write a function for the value of x shares of stock.
Graph this function. Note that this is a continuous
model representing a discrete situation.
Production (billion barrels per year)
b. Write the function for the rate of change of the
continuous model for the value of DuPont stock 40
with respect to the number of shares held. Graph
this rate-of-change function. 30
c. Find the change in the value of stock held if the 20
number of shares held is increased from 250 to
300 shares. Depict this change as the area of a 10 A
region on the rate-of-change graph. B
0 Year
13. The graph of a function f is shown. 1900 1925 1950 1975 2000 2025 2050 2075 2100
f (x)
(Source: Adapted from Ecology of Natural Resources by François
Ramade. Copyright 1984 by John Wiley & Sons, Inc. Reprinted
by permission of the publisher.)
a. Use midpoint rectangles of width 25 years to es-
timate the total amount of oil produced from
1900 through 2100 using graph A.
x
a b b. Repeat part a for graph B.
Discuss how to approximate the area of the region c. On page 31 of Ecology of Natural Resources, the to-
beneath the graph of f from x a to x b with tal oil production is estimated from graph A to be
four midpoint rectangles that have the same width. 2100 billion barrels and from graph B to be 1350
Draw the rectangles. billion barrels. How close are your estimates?
304 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
16. C
limate Scientists have long been interested in b. Find the area of the region between the graph in
studying global climatic changes and the effect of part a and the horizontal axis.
such changes on many aspects of the environment. c. What is the practical interpretation of the area
From carefully controlled experiments, two scien- found in part b?
tists constructed a model to simulate daily snow
depth in a region of the Northwest Territories in 18. Expanding aGs A certain gas expands as it is heated.
Canada. Rates of change (in equivalent centimeters The accompanying figure shows the rate of expansion
of water per day) estimated from their model are of the gas measured at several temperatures.
shown as a scatter plot in the accompanying figure. Expansion rate
Appropriate models have been sketched on the scat- (cm3 per °C)
ter plot. Note that both vertical and horizontal
scales change after June 9. 0.13
0.12
Equivalent 0.11
centimeters of 0.10
water per day 0.09
0.08
0.20
0 Temperature
0.15 200 202 204 206 208 210 (°C)
0.10
0.05
a. Use only the data and left rectangles to estimate by
(June 9) 72 73 74 75 76 how much the gas expanded as it was heated from
0 Day
0 14 28 42 56 70 200°C to 210°C. Sketch the rectangles on the scat-
(April 1) (May 12) ter plot. Do you believe your approximation is an
-4
overestimate or an underestimate? Explain.
-6
-8 b. Repeat part a using right rectangles.
-10 19. Energy Use The graph shows the energy usage in
(Source: R. G. Gallimore and J. E. Kutzbach, “Role of Orbitally megawatts for one day for a large university cam-
Induced Changes in Tundra Area in the Onset of Glaciation,” pus. The daily energy consumption for the campus
Nature, vol. 381, June 6, 1996, pp. 503–505.) is measured in megawatt-hours and is found by
a. What does the figure indicate occurred between calculating the area of the region between the graph
June 9 and June 11? and the horizontal axis.
b. Estimate the area of the region beneath the curve m(t)
from April 1 through June 9. Interpret your Energy power usage
answer. (megawatts)
c. Use four midpoint rectangles to estimate the area 40
of the region from June 11 through June 15. 35
Interpret your answer. 30
25 t
17. Robot pS eed A mechanical engineering graduate 20 Hours since
student designed a robot and is testing the ability of 0 6 12 18 24 midnight
the robot to accelerate, decelerate, and maintain
speed. The robot takes 1 minute to accelerate to 10 a. Estimate the daily energy consumption using
mph (880 feet per minute). The robot maintains that eight left rectangles.
speed for 2 minutes and then takes half a minute to b. Estimate the daily energy consumption using
come to a complete stop. Assume that this robot’s eight right rectangles.
acceleration and deceleration are constant. c. Discuss whether your estimates in parts a and b
a. Draw a graph of the robot’s speed during the are overestimates or underestimates of the actual
experiment. daily energy consumption.
5.1 Results of Change and Area Approximations 305
20. Population The rate of change of the population 22. o
Hspital Stay The rate of change of the length of
of North Dakota from 1970 through 1990 can be the average hospital stay between 1993 and 2000 can
modeled as be modeled by the equation
3.87 thousand 0.82t 0.39 days per year when 0 t 5
people per year when 0 t 15 s(t)
p(t) 0.1t days per year when 5 t 7
7.39 thousand
people per year when 15 t 20
where t is the number of years since 1993.
(Source: Based on data from Statistical Abstract, 2001.)
where t represents the number of years since 1970.
(Source: Based on data from Statistical Abstract, 1994.) a. Graph s for the years between 1993 and 2000.
a. Sketch a graph of the rate-of-change function. b. Judging on the basis of the graph, was the length
b. Find the area of the region between the graph of of the average hospital stay increasing or de-
p and the horizontal axis from 0 to 15. Interpret creasing between 1993 and 2000?
your answer. c. Find the area of the region lying above the axis
c. Find the area of the region between the graph of between the graph and the t-axis.
p and the horizontal axis from 15 to 20. Interpret d. Find the area of the region lying below the axis
your answer. between the graph and the t-axis.
d. Was the population of North Dakota in 1990 e. Judging on the basis of your answers to parts c
greater or less than the population in 1970? By and d, by how much did the average hospital stay
how much did the population change between change between 1993 and 2000? Can you deter-
1970 and 1990? mine the average stay in 2000? Why or why not?
e. What information would you need to determine 23. Temperature During a summer thunderstorm,
the population of North Dakota in 1990? the temperature drops and then rises again. The rate
21. C
ottage Ch
eese The rate of change of the per of change of the temperature during the hour and a
capita consumption of cottage cheese in the United half after the storm began is given by
States between 1980 and 1999 can be modeled by T(h) 9.5h3 15.5h2 17.4h 10.12°F per hour
the function
where h is the number of hours since the storm
0.01t 0.058 pound began.
per person per year when 0 t 13 a. Graph the function T from h 0 to h 1.5.
c(t) 0.1 pound per person Find the point A at which the graph crosses the
per year when 13 t 15 horizontal axis.
0 pound per person per year when 15 t 19 b. Consider the portion of the graph of T lying be-
low the horizontal axis. What does the area of the
where t is the number of years since 1980. region between this portion of the graph of T and
(Source: Based on data from Statistical Abstract, 2001.) the horizontal axis represent?
a. Sketch a graph of the rate-of-change function. c. What does the area of the region lying above the
b. Find the area of the region between the graph axis represent?
and the horizontal axis between t 0 and d. Consider the graph between 0 and 1.5 hours. Use
t 19. three right rectangles to approximate the area of
c. Interpret the area in part b in the context of cot- the region lying below the axis from 0 to 1.5 hours.
tage cheese consumption. e. Repeat part d for the region lying above the axis.
d. Can you determine the per capita cottage cheese
T(h) dh.
A
consumption in 1996? in 1999? Why or why not? f. Use a limit of sums to estimate
0
306 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
1.5 26. Trust Fund The rate of change of the projected
g. Use a limit of sums to estimate T(h) dh. total assets in the Social Security trust fund for the
A
years 2000 through 2033 can be modeled by the
1.5
.h Estimate T(h) dh. What information is equation
0
S(x) 0.46x 2 1.85x 108.7 billion dollars per year
needed to determine the temperature when
h 1.5? According to the model, one and a half x years after 2000.
hours after the storm began, was the temperature (Source: Based on data from the Social Security Administration.)
higher than, lower than, or the same as the tem- a. Graph S between 2000 and 2033.
perature at the beginning of the storm? If it was
higher or lower, by how many degrees? b. According to the graph of S, when will the trust
fund assets be growing and when will they be
24. Baby Boomers As the 76 million Americans born declining?
between 1946 and 1964 (the “baby boomers”) con- c. Find the point on the graph of S that corresponds
tinue to age, the United States will see an increasing to the time when the amount in the trust fund
proportion of Americans who are within one year of will be greatest.
a retirement age of 66. The model
d. Using four midpoint rectangles, estimate the
1.9 million retirees area lying above the axis and below the graph of
R(t) 0.1
1 18e0.04t year S. Interpret your answer.
gives the projected rate of change in the number of e. Using four midpoint rectangles, estimate the
people within one year of retirement, where t is the area lying below the axis and above the graph of
number of years since 1940, for years between 1940 S. Interpret your answer.
and 2050. The model is based on data from the U.S. f. By how much will the trust fund amount change
Bureau of the Census. between 2000 and 2033? What information do
we need to determine how much money is pro-
Use the equation and ten midpoint rectangles to es-
jected to be in the trust fund in 2033?
timate the change in the population within one year
of retirement or older from the end of 2005 through 27. Life Expectancy Life expectancies in the United
the end of 2010. States are always rising because of advances in
25. Birth
ewigh
t The rate of change of the percentage health care, increased education, and other factors.
of low-birthweight babies (less than 5 pounds 8 The rate of change (measured at the end of each
ounces) in 2000 can be modeled by year) of life expectancies for women in the United
States between 1970 and 2010 can be modeled by
P(w) 11.5(0.86w) percentage points per pound
E(t) 0.0004t 2 0.022t 0.36 years per year
when the mother gains w pounds during pregnancy.
The model is valid for weight gains between 18 and where t is the number of years since 1970.
(Source: Based on data from Statistical Abstract, 1998.)
43 pounds.
(Source: Based on data from National Vital Statistics Reports, a. Graph E between 1970 and 2010.
vol. 50, no. 5, February 12, 2002.) b. Find the point on the graph of E that corresponds
a. Sketch a graph of P from w 18 to w 43. to the time between 1970 and 2010 when the life
b. What does the fact that the graph of P lies below expectancy for women was growing least rapidly.
the t-axis from w 18 to w 43 tell you about c. Using four midpoint rectangles, estimate the
the percentage of low-birthweight babies in area lying between the graph of E(t) and the
2000? horizontal axis.
c. Use five midpoint rectangles to estimate the d. By how much is the life expectancy for women
area of the region between the graph of P and projected to increase between 2000 and 2010?
the w-axis from w 18 to w 43. Interpret What information do we need to determine the
your answer. projected life expectancy for women in 2010?
5.1 Results of Change and Area Approximations 307
28. eWigh
t The rate of change of the weight of a lab- 31. aSles The table records the volume of sales (in thou-
oratory mouse can be modeled by the equation sands) of a popular movie for selected months during
the first 18 months after it was released on DVD.
13.785
w(t) grams per week
t Number of DVDs
Months sold each month
where t is the age of the mouse in weeks and after release (thousands)
1 t 15.
2 565
a. Use the idea of a limit of sums to estimate the
11 4 467
value of w(t)dt . 5 321
3
7 204
b. Label units on the answer to part a. Interpret
your answer. 10 61
c. If the mouse weighed 4 grams at 3 weeks, what 11 31
was its weight at 11 weeks of age? 12 17
16 3
29. Sales The rate of change of annual U.S. factory
sales (in billions of dollars per year) of consumer 18 2
electronic goods to dealers from 1990 through 2001
can be modeled by the equation a. Find a logistic model for the data.
b. Use 5, 10, and 15 right rectangles to estimate the
s(x) 0.12x 2 x 5.7 billion dollars per year number of DVDs sold during the first 15 months
after release.
where x is the number of years since 1990.
(Sources: Based on data from Statistical Abstract, 2001, and c. Which of the following would give the most ac-
Consumer Electronics Association.) curate value of the number of DVDs sold during
a. Use the idea of a limit of sums to estimate the the first 15 months after release?
change in factory sales from 1990 through 2001. i. The answer to part b for 15 rectangles
b. Write the definite integral symbol for this limit ii. The limiting value of the sums of midpoint
of sums. rectangles using the model in part a
c. If factory sales were $43.0 billion in 1990, what iii. The sum of actual sales figures for the first 15
were they in 2001? months
30. Production On the basis of data obtained from a 32. Bank cAcount The table gives rates of change of the
preliminary report by a geological survey team, it is amount in an interest-bearing account for which in-
estimated that for the first 10 years of production, a terest is compounded continuously.
certain oil well can be expected to produce oil at the
rate of r(t) 3.94t 3.55e1.35t thousand barrels per Rate of change
year t years after production begins. At end of year (dollars per day)
a. Use the idea of a limit of sums to estimate the 1 2.06
yield from this oil field during the first 5 years of 3 2.37
production. 5 2.72
b. Use the idea of a limit of sums to estimate the 7 3.13
yield during the first 10 years of production.
9 3.60
c. Write the definite integral symbols representing
the limits of sums in parts a and b. a. Convert the input into days. Disregard leap
d. Estimate the percentage of the first 10 years’ pro- years. Why is this conversion important for a
duction that your answer to part a represents. definite integral calculation?
308 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
b. Find an exponential model for the converted data. Rate of change of
c. Use a limiting value of sums to estimate the diastolic BP
change in the balance of the account from the day Time (mm Hg per hour)
the money was invested to the last day of the 8 A.M. 3.0
ninth year after the investment was made. Again, 10 A.M. 1.8
disregard leap years.
12 P.M. 0.7
d. Give the definite integral notation for your answer 2 P.M. 0.1
to part c.
4 P.M. 0.7
33. Labor The personnel manager for a large con- 6 P.M. 1.1
struction company keeps records of the number of 8 P.M. 1.3
labor hours per week spent on typical construction
10 P.M. 1.1
jobs handled by the company. He has developed the
following model for a labor-power curve: 12 A.M. 0.7
2 A.M. 0.1
4 A.M. 0.8
6,608e0.706x 6 A.M. 1.9
m(x) million labor hours per week
(1 925e0.706x)2
a. During which time intervals was the patient’s
diastolic blood pressure rising? falling?
after the xth week of the construction job.
b. Estimate the times when diastolic blood pressure
a. Use 5, 10, and 20 right rectangles to approximate was rising and falling most rapidly.
the number of labor hours spent during the first
c. Find a model for the data.
20 weeks of a typical construction job.
d. Find the times at which the output of the model
b. If the number of labor hours spent on a is zero. Of what significance are these times in
particular job exactly coincides with the model, the context of blood pressure?
which of the following would give the most ac-
e. Use the idea of a limiting value of sums to esti-
curate value of the number of labor hours spent
mate by how much the diastolic blood pressure
during the first 20 weeks of the job?
changed from 8 A.M. to 8 P.M.
i. The 20-right-rectangle sum found in part a
f. Write the definite integral notation for your
ii. The sum of 20 midpoint rectangles answer to part e.
iii. The limiting value of the sums of midpoint Discussing Concepts
rectangles
35. Explain how area, accumulated change, and the def-
34. Blood Pressure Blood pressure (BP) varies for in- inite integral are related and how they differ.
dividuals throughout the course of a day, typically 36. Why is it important to know whether, and if so
being lowest at night and highest from late morning where, a function has horizontal-axis intercepts be-
to early afternoon. The estimated rate of change in fore using a definite integral (limit of sums) to de-
diastolic blood pressure for a patient with untreated termine the area of the region(s) bounded by the
hypertension is shown in the table. function and the horizontal axis?
5.2 Accumulation Functions 309
5.2 Accumulation Functions
In the previous section, we saw that when we have a rate-of-change function for a cer-
tain quantity, we approximate the accumulation of change in that quantity between
two values of the input variable using the area between the rate-of-change curve and
the horizontal axis. Area approximation methods are valuable, but in some situations
it would be helpful to have a formula that would answer the question “What was the
accumulated change in the quantity from a to t for any value of t?”
For instance, it is important for hydrogeologists studying a watershed to know
how much water flowed through a river since a specific starting time. Typically, data
on the rate of flow are gathered and used to create a flow rate model from which the
accumulation of flow can be calculated.
EXAMPLE 1 Estimating Accumulated Change
Rising River The flow rate past a sensor in the west fork of the Carson River in
Nevada is measured periodically. Suppose the flow rates for a 28-hour period begin-
ning at 12 A.M. Wednesday on a spring night can be modeled by the equation
f(t) 0.018t 2 0.42t 5.13 million cubic feet per hour
f (t)
(million ft3/hr) where t is the number of hours after 12 A.M. Figure 5.24 shows a graph
for the 28 hours between 12 A.M. Wednesday and 4 A.M. Thursday. A
8 heavy rain began falling around 6 A.M. on Wednesday morning. Esti-
mate to the nearest 10,000 cubic feet the amount of water that flowed
6 past the sensor from 12 A.M. Wednesday until 4 A.M. Thursday. Make the
estimate at 4-hour intervals.
4
Solution
2 The amount of water that flowed through the Carson River from 12 A.M.
t
Hours after Wednesday (t 0) and some ending time (t b) can be estimated by
12:00 A.M. the area between the graph of f and the horizontal axis from t 0 to
0 4 8 12 16 20 24 28
Wednesday t b. Table 5.10 shows the results of limits of sums for the various
ending times.
FIGURE 5.24
TABLE 5.10
Ending time Area* approximation Ending time Area* approximation
(input) (output) (input) (output)
f(t)dt 0
0 16
12 A.M. (t 0) 4 P.M. (t 16) f(t)dt 52.89
0 0
f(t)dt 17.54
4 20
4 A.M. (t 4) 8 P.M. (t 20) f(t)dt 66.6
0 0
f(t)dt 30.67
8 24
8 A.M. (t 8) 12 A.M. (t 24) f(t)dt 85.10
0 0
f(t)dt 41.69
12 28
12 P.M. (t 12) 4 A.M. (t 28) f(t)dt 110.71
0 0
*All area units are million cubic feet. ●
310 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
We see in Table 5.10 that changes in ending times lead to changes in the accumu-
lation of area. As long as the starting time stays constant at some number a, we can
f(t)dt.
x
represent the accumulated change in f from a to a variable ending time x as
a
This integral is called the accumulation function of f from a to x.
Accumulation Function
f(t)dt, gives
x
The accumulation function of a function f, denoted by A(x)
a
the accumulation of the area between the horizontal axis and the graph of f
from a to x.
Using Limits of Sums to Sketch Accumulation Graphs
When a function is continuous and bounded over a specific interval, the correspon-
ding accumulation function will also be continuous over that same interval. For the
rising river example, we can sketch an accumulation graph using the values in Table
5.10. A scatter plot of the area accumulations and a smooth curve representing river
flow is given in Figure 5.25.
x
∫0 f(t) dt
(million ft3)
100
75
50
25
x
Hours after
12:00 A.M.
0 4 8 12 16 20 24 28
Wednesday
FIGURE 5.25
Using Estimated Areas to Sketch Accumulation Graphs
In Example 1, we sketched a graph of the accumulation function based on accumu-
lated change, using the limits of sums. We were able to estimate the limits of sums be-
cause we had an equation for the flow rates past the sensor. If we have a graph of a
rate-of-change function, we can sketch an accumulation function by estimating the
signed areas trapped between the rate-of-change function and the horizontal axis. We
use a graph of f(t) sketched on a grid (see Figure 5.26).
5.2 Accumulation Functions 311
f (t)
(million ft3/hr)
8
2
t
Hours after
12:00 A.M.
0 4 8 12 16 20 24 28
Wednesday
FIGURE 5.26
We begin by finding the area of a single box of the grid. The height of one box is
1 million cubic feet per hour. The width of one box is 2 hours. Thus the area of one
box is (2 hours) 1 million cubic feet per hour = 2 million cubic feet.
f(t)
(million ft3/hr)
8
2
t
Hours after
12:00 A.M.
0 4 8 12 16 20 24 28
Wednesday
FIGURE 5.27
3
The area shaded under the curve from 0 to 2 is approximately 44 boxes or 9.5
million cubic feet. (See Figure 5.27.) The area shaded under the curve from 2 to 4 is
1
approximately 44 boxes or 8.5 million cubic feet. Adding this to the area from 0 to 2,
we find the area under the curve from 0 to 4 to be approximately 18 million cubic feet.
As we repeat this process, we approximate the area under the curve from 4 to 6
1
to be 32 boxes or 7 million cubic feet. Adding this to the area from 0 to 4, we
estimate that the accumulated area from 0 to 6 is approximately 25 million cubic
feet. There are approximately 3 boxes or 6 million cubic feet of area under the curve
312 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
from 6 to 8, so the accumulated area from 0 to 8 is approximately 31 million cu-
bic feet. See Figure 5.28a. We repeat this process until we have found the accumu-
lated area for each successive interval. The results of our accumulation are given in
Table 5.11.
f(t) f(t)
8 8
6 6
4 4
2 2
t t
0 4 8 12 16 20 24 28 0 4 8 12 16 20 24 28
(a) (b)
f (t) f(t)
8 8
6 6
4 4
2 2
t t
0 4 8 12 16 20 24 28 0 4 8 12 16 20 24 28
(c) (d)
f(t) f(t)
8 8
6 6
4 4
2 2
t t
0 4 8 12 16 20 24 28 0 4 8 12 16 20 24 28
(e) (f)
FIGURE 5.28
5.2 Accumulation Functions 313
TABLE 5.11
Interval Number of Area* Accumulated area*
a to b boxes approximation from 0 to b
0 to 2 4.75 9.5 9.5
2 to 4 4.25 8.5 18.0
4 to 6 3.5 7.0 25.0
6 to 8 3.0 6.0 31.0
8 to 10 2.75 5.5 36.5
10 to 12 2.6 5.2 41.7
12 to 14 2.6 5.2 46.9
14 to 16 2.75 5.5 52.4
16 to 18 3.25 6.5 58.9
18 to 20 3.5 7.0 65.9
20 to 22 4.25 8.5 74.4
22 to 24 5.0 10.0 84.4
24 to 26 6.0 12.0 96.4
26 to 28 7.0 14.0 110.4
*Area approximations are measured in million cubic feet.
x
∫0 f(t) dt
(million ft3)
100
75
50
25
x
Hours after
12:00 A.M.
0 4 8 12 16 20 24 28
Wednesday
FIGURE 5.29
The accumulated change found in Example 1 is 110.71 million cubic feet, which
is close to the estimated accumulation summarized in Table 5.11. Compare Figure
5.29 with Figure 5.25. Carefully counting areas using a grid is another method of
drawing an accumulation function.
When a portion of a graph is negative, the area below the horizontal axis indicates
a decrease in the accumulation. Example 2 illustrates how to sketch accumulation
graphs for graphs that go below the horizontal axis.
314 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
EXAMPLE 2 Using Estimated Grid Areas to Sketch Accumulation Functions
Consider the graph of f shown in Figure 5.30.
a. Construct a table of accumulation function values for x 3, 2.5, … , 2.5, 3.
f (t) b. Sketch a scatter plot and continuous graph of the accumulation
x
4 function A(x) f(t)dt .
3 3
2
Solution
1
t a. Begin estimating accumulated area from the far left side of the graph
-3 -2.5 -2 -1.5 -1 -0.5 0.5 1 1.5 2 2.5 3 by counting the boxes between the graph of f and the horizontal axis
-2 from 3 to x. You should obtain values similar to those in the second
-3 and six throws of Table 5.12. Note that the boxes have height 1 unit
-4 and width 0.5 unit; thus the area of each box is (1)(0.5) 0.5 unit2.
The fourth and eighth rows of Table 5.12 is the number of boxes
FIGURE 5.30
multiplied by 0.5 to obtain the accumulated area value.
For values of x greater than zero, the boxes lie below the horizontal axis. The
areas of these boxes should be subtracted from the accumulated area of boxes
above the horizontal axis in order to obtain the net number of accumulated
boxes. Because the number of boxes from 3 to 0 is the same as the number of
boxes from 0 to 3, the net result of accumulated area from 3 to 3 is zero.
TABLE 5.12
Right endpoint 3 2.5 2 1.5 1 0.5 0
Additional number of boxes 0 0.5 0.5 1 2 3 2
Additional accumulated value 0 0.25 0.25 0.5 1 1.5 1
Total accumulated value 0 0.25 0.50 1.0 2 3.5 4.5
Right endpoint 0.5 1 1.5 2 2.5 3
Additional number of boxes 2 3 2 1 0.5 0.5
Additional accumulated value 1.0 1.5 1 0.5 0.25 0.25
Total accumulated value 3.5 2.0 1.0 0.5 0.25 0
x
b. A scatter plot and continuous graph for A(x) f(t)dt are shown in Figure 5.31.
3
x
A(x) = ∫ f(t) dt
−3
4.5
x
−3 3
FIGURE 5.31 ●
5.2 Accumulation Functions 315
Sketching Accumulation Graphs
Given the graph of a function, consider accumulation beginning at the far
left, regardless of the location of the specified starting point. Sketch the
accumulation graph starting at the far left, and then shift the graph up or
down so that the output value at the specified starting point is zero.
EXAMPLE 3 Sketching Accumulation Functions with Different Starting Points
Consider again the function graph and accumulation graph from Example 2, redrawn
here as Figures 5.32 and 5.33.
f (t)
4 x
A(x) = ∫ f(t) dt
3 −3
2
1 4.5
t
-3 -2.5 -2 -1.5 -1 -0.5 0.5 1 1.5 2 2.5 3
-2
-3
x
-4 −3 3
FIGURE 5.32 FIGURE 5.33
Use the graph in Figure 5.33 to sketch a graph of the accumulation function
f(t)dt.
x
B(x)
0
Solution
In order to sketch the accumulation graph with 0 as the starting point, we vertically
shift the graph in Figure 5.33 so that the function value at the starting point is zero.
In this example, we shift the graph down 4.5 units so that the peak of the graph is the
point (0, 0). (See Figure 5.34.)
x
B(x) = ∫ f(t) dt
0
x
−3 3
−4.5
FIGURE 5.34 ●
316 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
By now, you should have a good idea of what we mean by an accumulation func-
tion. The following information gives a procedure for sketching an accumulation
function graph without making use of a grid.
Sketching Accumulation Functions
To sketch an accumulation function with starting value a given a graph of f,
Step 1: Choose as an initial starting value the input value at the left end of
the horizontal axis shown in the graph.
Step 2: Sketch the accumulation of area from the far left such that
• Regions between the graph of f and the horizontal axis that lie above the
axis contribute positive accumulation equal to the area of the region.
• Regions that lie beneath the horizontal axis contribute negative
accumulation equal to the negative of the area of the region.
• Relative extrema on the accumulation graph occur at the same input
values as the corresponding zeros on the rate-of-change graph.
• Inflection points on the accumulation graph occur at the same input
values as the relative extrema on the rate-of-change graph.
Step 3: Vertically shift the accumulation graph in Step 2 so that the graph is
zero at the actual starting value a.
In order to further refine the sketch of the graph of an accumulation function, it is
necessary to understand how rapidly the accumulated area grows or declines.
Concavity and Accumulation
slower faster
An important tool for accurately sketching accumulation function graphs is an un-
de
se
ea
cr
derstanding of how the increasing or decreasing nature of the rate-of-change func-
ea
cr
in
se
tion (as distinguished from the sign of the rate-of-change function) affects the
shape of the associated accumulation function.
The circle in Figure 5.35 may be used as a quick indicator of the general shape
de
a se
cr
that an accumulation function will take:
re
ea
c
in
se
• An accumulation function showing slower increase will be shaped a bit like the
slower faster upper-left arc of a circle. The accumulation function will be increasing and con-
cave down.
General shape of an accumulation
function • An accumulation function showing slower decrease will be shaped a bit like the
lower-left arc of a circle. The accumulation function will be decreasing and con-
FIGURE 5.35
cave up.
• An accumulation function showing faster decrease will be shaped a bit like the
Do not confuse the general
shape of the accumulation upper-right arc of a circle. The accumulation function will be decreasing and con-
function as shown by the circle cave down.
diagram as an indication of • An accumulation function showing faster increase will be shaped a bit like the
whether the accumulation
function is positive or negative lower-right arc of a circle. The accumulation function will be increasing and con-
over a given interval. cave up.
5.2 Accumulation Functions 317
Example 4 demonstrates how the graph of a rate-of-change function can be
analyzed to sketch the graph of an accumulation function.
EXAMPLE 4 Sketching a General Accumulation Function Graph
Consider the graph of y (a rate-of-change function) shown in Figure 5.36.
x
A B 0 C D E
FIGURE 5.36
a. Begin by identifying the x-intervals on which the accumulation function will be
positive and those on which the accumulation function will be negative. The
positive intervals on the rate-of-change function correspond to the intervals on
the accumulation function where the accumulation function will be increasing.
Similarly, the negative intervals on the rate-of-change function correspond to
the intervals on the accumulation function where the accumulation function
will be decreasing. Identify any relative extremes or x-intercepts on the rate-of-
change graph. Classify each of these points as a relative maximum, a relative
minimum, or an inflection point on the accumulation function.
b. Sketch the accumulation function graph with x A as the starting point.
c. Sketch the accumulation function graph with x 0 as the starting point.
Solution
a. The accumulation function will be increasing over (A, C) and decreasing over (C,
E). There is a relative maximum on the rate-of-change graph at x 0 so there is
an inflection point on the accumulation function at x 0. There is a relative min-
imum on the rate-of-change graph at x D so there is another inflection point
on the accumulation function at x D. There is a second relative minimum at
x B, so there is a third inflection point on the accumulation function at the
point x B. The slope of the accumulation graph at the point x B will be 0.
There is an x-intercept at the point x C. The accumulation function will
reach a relative maximum value at x C.
318 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Positive,
faster and Positive,
faster slower and
slower
x
A B 0 C D E
Positive,
slower and
slower Negative, Negative,
faster and slower and
faster slower
FIGURE 5.37
b. Consider the shape of the initial segment. We know
that the function is increasing because the rate-of-
change function is positive, but we see that the in-
Area accumulation crease is slowing from the point x A to x B. (See
from far left
Figure 5.37.) Begin by drawing a concave-down
shape to indicate that the rate of change of the accu-
mulation function is slowing.
Note that the rate-of-change function is smooth and
continuous. This means that the accumulation func-
tion will also be smooth and continuous. Draw your
accumulation graph so that you can keep it smooth
and continuous while incorporating inflection points
x (at points x B, x 0, and x D) and a relative max-
A B C D E imum at x C. Use the connections to the shape of
the accumulation function given in Figure 5.35.
A graph illustrating these properties is shown in Fig-
FIGURE 5.38 ure 5.38.
c. To obtain the accumulation function graph with
zero as the starting point, shift the graph down so
that it passes through (0, 0). See Figure 5.39.
Area accumulation
from zero
x
A B C D E
FIGURE 5.39 ●
5.2 Accumulation Functions 319
Recovering a Function
Recovering a function is the phrase we use for the process of beginning with a rate-
of-change function for a quantity and obtaining a function for the quantity. An im-
portant part of recovering a quantity function from its rate of change is recovering
the units of the quantity function from the units of its rate of change. We can recover
the units of the amount function by recalling that the rate of change of a function is
output units
a slope of a tangent line. The units on slope are input units , so the units of a rate-of-
change function are the output units (of its quantity function) divided by the input
units (of its quantity function). From the geometric viewpoint of area, we know that
the output units of the quantity function are
(Rate
of
change function units) (input units of quantity function)
output units of quantity function
(input units of quantity function)
input units of quantity function
dM
Suppose dt milliliters per hour is the rate-of-change function of the amount of
dM
insulin in a patient’s body t hours after an injection. In the case of dt , the units milli-
milliliters
liters per hour can be rewritten as hour . Now we can see that the output units of
M are milliliters and the input units are hours. Figure 5.40 shows input/output dia-
dM
grams for dt and M. When we recover the quantity function M, the output units will
be milliliters and the input units will be hours.
t t
Input hours Input hours
Rule dM Rule M
dt
Output Output
dM M(t)
dt milliliters
milliliters
hour
FIGURE 5.40
Occasionally, units of a rate of change are expressed in terms of a squared unit.
For example, acceleration is often expressed in feet per second squared. Suppose A(t)
is acceleration of a vehicle in feet per second squared, where t is the number of sec-
onds since the vehicle began accelerating. Again, this is obviously the rate of change
of some function. In fact, it is the rate of change of velocity. However, the input and
output units of that velocity function may not be immediately apparent because of
the use of the word squared. Rewrite feet per second squared as
feet feet feet
(second)
(second)2 (second)(second) second
(feet per second) per second
320 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
The output units of the velocity function are now identifiable as feet per second.
Input/output diagrams for these functions are shown in Figure 5.41.
t t
Input seconds Input seconds
Rule A Rule v
Output Output
A(t) v(t)
feet feet
(second)2 second
FIGURE 5.41
We have seen how to sketch an accumulated quantity function from a graph of its
rate-of-change function and how to recover the quantity units from rate-of-change
units. In the next section, we will see that the Fundamental Theorem of Calculus im-
plies that we have a powerful algebraic tool for recovering the accumulated quantity
formula from its rate-of-change formula.
5.2 Concept Inventory 5.2 Activities
• Accumulation and area Applying Concepts
• Accumulation functions 1. Velocity Refer to the velocity graph.
• Sketching accumulation graphs f(t)
• Interpreting accumulation Velocity
(ft/sec)
90
60
30
t
0
0 5 10 (seconds)
5.2 Accumulation Functions 321
a. Sketch the accumulation function with 5 seconds b. What does the area of the region between the
as the starting point. graph and the horizontal axis between days 18
b. Give the mathematical notation for the function and 47 represent?
you sketched in part a. c. Was the stock price higher or lower on day 47
c. In the context of the moving vehicle, what is the than it was on day 0? How much higher or lower?
interpretation of the output values of the func- d. Was the stock price higher or lower on day 55 than
tion in part a? it was on day 47? How much higher or lower?
2. Plant Grow
th Refer to the plant growth rate e. Using the information presented in the graph, fill
function graph. in the accumulation function values in the ac-
companying table.
g(t)
(mm/day)
x 0 8 18 35 47 55
5
r (t )dt
x
4
0
3
2
r(t)dt for values of
1 x
t f. Graph the function R(x)
Days 0
1 13 27
x between 0 and 55, labeling the vertical axis as
a. Sketch the accumulation function with day 13 as accurately as possible.
the starting point. g. If the stock price was $127 on day 0, what was the
b. Give the mathematical notation for the function price on day 55?
you sketched in part a. 4. Stock Value Consider the graph of the rate of
c. In the context of plant growth, what is the inter- change in the price of a certain technology stock
pretation of the output values of the accumula- during the first 55 trading days of 2003.
tion function in part a? a. Label each of the shaded areas as representing
3. Stock Value Consider the graph of the rate of positive or negative change in price. Also label
change in the price of a certain technology stock each region as describing faster and faster or
during the first 55 trading days of 2003. slower and slower change in stock price.
b. On the basis of your answers to part a, sketch a
a. What does the area of the region between the
graph of the accumulation function
graph and the horizontal axis between days 0 and
r(t)dt
18 represent? x
P(x)
18
r (t)
Rate of change in price c. Sketch a graph of the accumulation function
r(t)dt
(dollars per day) x
3 Q(x)
35
Area = Area = 25.4
30.4 d. What differences do you notice among the three
accumulation functions in part f of Activity 3 and
8 55 t parts b and c of this activity?
0
18 35 47 Days 5. u
S bscribers The accompanying graph shows the
Area = 8.3 Area = 11.1
rate of change of the number of subscribers to an In-
-1.5 ternet service provider during its first year of business.
Area =
7.1 a. According to the graph, did the number of sub-
Figure for Activities 3 and 4 scribers ever decline during the first year?
322 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Subscribers p(t)
per day (thousands of
70 dollars per week)
60 4
50 3
40 2
30 1
t
0
20 4 8 12 16 20 24 28 32 36 40 44 48 52 Weeks
-1
10
-2
0 Week -3
0 4 8 12 16 20 24 28 32 36 40 44 48 52
a. What does the area of each box in the grid
represent?
b. What is the significance of the peak in the rate- b. What is the interpretation, in context, of the ac-
p(t)dt?
of-change graph at 20 weeks? x
cumulation function P(x)
c. If n(x) is the number of subscribers per day 0
at the end of the xth day of the year, what does c. Count boxes to estimate accumulation function
t
the function N(t) n(x)dx values from 0 to x for the values of x given in the
140 accompanying table.
describe?
d. How many subscribers does each box in the Accumulation Accumulation
figure represent? x function value x function value
e. Use the grid and graph in the figure to estimate 0 28
the accumulation function values in the accom- 4 32
panying table. 8 36
12 40
16 44
n(x)dx n(x)dx
t t
Week t (days) Week t (days) 20 48
0 0
24 52
4 28 28 196
8 56 36 252 d. Use the data in part c to sketch an accurate graph
p(t)dt.
x
12 84 44 308 of the accumulation function P(x)
0
16 112 52 364
Label units and values on the horizontal and
20 140 vertical axes.
.7 Rainfall The graph of r(t) represents the rate of
change of rainfall in Florida during a severe thun-
f. Sketch a graph of the accumulation function
derstorm t hours after the rain began falling.
with 140 days as the starting value.
Draw a graph of the total amount of rain that fell
6. Profit The graph shows the rate of change of
during this storm, using the following facts:
profit for a new business during its first year. The in-
put is the number of weeks since the business • The rain started falling at noon and did not stop
opened, and the output units are thousands of dol- until 6 P.M.
lars per week. • Three inches of rain fell between noon and 3 P.M.
5.2 Accumulation Functions 323
f(t)dt
• The total amount of rain that fell during the x
f (t)
10. a.
storm was 5.5 inches. 0
b. f(t)dt
x
r (t)
(inches per A
hour)
t
A
f(t)dt
x
11. a. f (t)
0
b. f(t)dt
x
A t
c. f(t)dt
x A B
t B
0 3 6 Hours
f(t)dt
x
.8 Population The Brazilian government has estab- 12. f (t)
lished a program to protect a certain species of en- 0
dangered bird that lives in the Amazon rain forest.
The program is to be phased out gradually by the year t
2020. An environmental group believes that the gov-
ernment’s program is destined to fail and has pro-
jected that the rate of change in the bird population
f (t)
between 2000 and 2050 will be as shown in the figure.
f(t)dt
x
Rate of change
13.
0
(birds per year)
x f (t)
Year 14. f(t)dt
2000 2020 2050 A
t
A
Draw a graph of the bird population between 2000
and 2050, using the following:
• At the beginning of 2000 there were 1.3 million
birds in existence.
• The species will be extinct by 2050. In each of Activities 15 through 18, a graph is given.
In Activities 9 through 14, sketch the indicated accumu- Identify, from graphs a through f, the derivative graph
lation function graphs. and the accumulation graph (with 0 as the starting
point) of the given graph. Graphs a through f may be
f(t)dt
x
f (t) used more than once.
.9 a.
A
15. 16.
b. f(t)dt
x
t
B B A
A B C A B
324 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
17. 18. 20.
A B C
m p(m) Input Output Input Output
A B C D 0 0 0 0 0 0
1 8 1 12 1 3
a. b.
2 16 2 0 2 16
3 0 3 36 3 27
A B 4 64 4 96 4 0
5 200 5 180 5 125
c. d.
A B For each of the rates of change in Activities 21 through 24:
A B C a. Write the units of the rate of change as a fraction.
b. Draw an input/output diagram for the recovered
function.
c. Interpret the recovered function in a sentence.
e. f.
21. When m thousand dollars is being spent on adver-
tising, the annual revenue of a corporation is chang-
dR
ing by dm million dollars per thousand dollars spent
on advertising.
A B C D 22. The percentage of households with washing machines
dW
was changing by dt percentage points per year, where
A B C D
t is the number of years since 1950.
23. The concentration of a drug in the bloodstream of a
dc
In each of Activities 19 and 20, a table of selected values patient is changing by dh milligrams per liter per
for a function is given. Also shown are tables of values hour h hours after the drug was given.
for the derivative and the accumulation function with 0
as the starting point. Determine which table contains 24. The level of production at a tire manufacturer h
values of the derivative and which contains values of the hours after production began is increasing by P(h)
accumulation function. Justify your choice. tires per hour squared.
19.
Discussing Concepts
t f (t ) Input Output Input Output 25. Consider a rate-of-change graph that is increasing
0 4 0 0 0 0 but negative over an interval. Explain why the accu-
1 3 1 2 1 3.667
mulation graph decreases over this interval.
2 0 2 4 2 5.333 26. What behavior in a rate-of-change graph causes the
3 5 3 6 3 3 following to occur in the accumulation graph: a
minimum? a maximum? an inflection point?
4 12 4 8 4 5.333
Explain.
5.3 The Fundamental Theorem 325
5.3 The Fundamental Theorem
In Section 5.2 we graphed accumulation functions from graphs of rate-of-change
functions. We saw that we could use numerical estimates of definite integrals to help
us sketch accumulation graphs. In this section, we develop some algebraic tools to
help us write formulas for accumulation functions.
The Slope Graph of an Accumulation Graph
We began Section 5.2 with the function shown here in Figure 5.42a and drew the
accumulation function graph shown in Figure 5.42b.
x
f (t) ∫0 f(t) dt
(million ft3/hr) (million ft3)
8 100
6 75
4 50
2 25
t x
Hours after Hours after
12:00 A.M. 12:00 A.M.
0 4 8 12 16 20 24 28 0 4 8 12 16 20 24 28
Wednesday Wednesday
f (t) dt
x
(a) Function f(t) (b) Accumulation function
0
FIGURE 5.42
f (x)
(million ft3/hr)
Now we sketch the slope graph (derivative graph) of the accu-
8
mulation function in Figure 5.42b. Note that the slopes are positive
everywhere but become increasingly small as t goes from 0 to near 12.
6
There appears to be a point of least slope near t 12, after which the
slopes increase. The slope graph appears in Figure 5.43. The slope graph
4
is exactly the graph with which we began in Figure 5.42a (with the in-
put variable labeled x instead of t).
2
x In Examples 2 and 3 in Section 5.2, we began with the graph in
Hours after Figure 5.44a and sketched the accumulation function shown in Fig-
12:00 A.M. ure 5.44b.
0 4 8 12 16 20 24 28
Wednesday Again, let us sketch the slope graph of the accumulation function. To
f (t) dt
d x
Slope function the left of zero, the graph has positive slopes. The slopes are near zero to the
dx 0 far left, and the graph becomes steeper until B. Between B and 0, the
FIGURE 5.43 slopes are still positive but are approaching 0 as the accumulation function
326 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
x
f (t) A(x) = ∫ f(t) dt
0
t x
−B B −B B
f (t) dt
x
(a) Function f(t) (b) Accumulation function
0
FIGURE 5.44
A' (x) approaches its maximum. At x 0, the slope is zero. To the right of x 0,
the slopes are negative. They become more and more negative until B. To the
right of B, the slopes are still negative but are getting closer and closer to zero
as the graph levels off. The slope graph is shown in Figure 5.45. Again, this is
exactly the graph with which we began in Figure 5.44a.
x You may be noticing that if we begin with a function f with input t, graph
f(t)dt, and then
-B B x
or find a formula for the accumulation function A(x)
a
take the derivative or draw the slope graph, we get f, the function with which
we began, only in terms of x rather than t.
f (t) dt
d x
Slope function In order to explore this connection between accumulation functions and deriva-
dx 0 tives, consider the following argument:
FIGURE 5.45 Let A be the accumulation function of f from a to x. The graph in Figure 5.46a shows
the function f and the area representing the accumulation function value from a to x.
Figure 5.46b shows the region whose area is the accumulation value from a to x h.
f (t) f (t)
A(x) A(x + h)
t t
a x a x+h
(a) (b)
FIGURE 5.46
Next, consider the difference between the two areas. The small region with this
area is shown in Figure 5.47a.
5.3 The Fundamental Theorem 327
f (t) f (t)
f (x) + f (x + h)
A(x + h) − A(x) Height =
2
t t
x x+h x x+h
Width = h
(a) (b)
FIGURE 5.47
We can approximate the area of this region by using a trapezoid. Recall that the
height of a trapezoid constructed over an interval is the average of the function value
at the two endpoints. Figure 5.47b shows this trapezoid and its height and width.
The true area of the shaded region in Figure 5.47a is A(x h) A(x), and it can
be approximated by the area of the trapezoid. Thus we have
A(x h) A(x) f(x) 2f(x h)h
We now divide each side of the expression by h. The reason for this division will be
evident later.
A(x h) A(x) f(x) f(x h)
h 2
Consider what happens as h becomes smaller and smaller. In other words, what
happens when we take the limit of the above expression as h approaches zero?
lim A(x h) A(x) lim f(x) f(x h)
h→0 h→0
h 2
dA
You should recognize the term on the left as dx , the derivative of A. The term on the
right approaches f(x) as h approaches 0, because f(x h) gets closer and closer to
f(x) f(x) 2f(x)
f(x), so 2 2 f(x). Thus we have
dA
f(x)
dx
In fact, we can replace the approximation with an equality (although a discussion of
why this is the case is beyond the scope of this book):
dA
f(x)
dx
This expression says that the derivative of the accumulation function is the orig-
inal function. This is an important result because it links accumulation functions to
derivatives. The importance of this connection is reflected in the name given the
relation: the u
Fndamental Th eorem of Calculus.
328 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
The Fundamental Theorem of Calculus
For any continuous function f with input t, the derivative of an accumulation
function of f is the function f in terms of x. In symbols, we write
f(t)dt f(x)
d x
dx a
We can infer from the Fundamental Theorem that to find an accumulation for-
mula, we need only reverse the process of finding a derivative. For this reason, the ac-
f(t)dt is called an antiderivative of the function f.
x
cumulation function F(x)
a
Antiderivative
Let f be a function of x. A function F is called an antiderivative of f if
F(x) f(x); that is, the derivative of F is f.
Our motivation for developing accumulation functions (antiderivatives) is not
only to have a formula for accumulated change but also (and more important) to de-
velop a function for a quantity if we know a function for that quantity’s rate of change.
Antiderivative Formulas
Given a function, how do we find an antiderivative? As we have seen, we must reverse
the process of differentiation. Antidifferentiation starts with the known derivative and
finds the unknown function. For example, consider the constant function f(x) 3.
To find an antiderivative of f, we need to find a function of x whose derivative is 3. One
such function is F(x) 3x . Other functions whose derivatives are 3 include
F(x) 3x 7 and F(x) 3x 24.9.
In fact, having found one antiderivative F for a given function f, we can obtain in-
finitely many antiderivatives for that function by adding an arbitrary constant C to F.
Thus we call y F(x) C the general antiderivative of f. We use the notation
f(x)dx F(x) C
for the general antiderivative. The general antiderivative is a group of infinitely many
functions. (A particular accumulation function is one specific function from that
group.) Note that the integral sign has no upper and lower limits for general antideriv-
ative notation. The dx in this notation is to remind us that we are finding the general
antiderivative with respect to x, so our antiderivative formula will be in terms of x. For
example, we say that the general antiderivative for f(x) 3 is F(x) 3x C, and we
write 3dx 3x C.
An antiderivative (accumulation formula) of any constant function will be a line
because lines have constant derivatives. We can write this general rule as
kdx kx C
where k and C are any constants.
5.3 The Fundamental Theorem 329
Now consider finding the general antiderivative of f(x) 2x. We are seeking a
function whose derivative is 2x. The function is F(x) x 2 C, and we write
2x dx x 2 C
It is more difficult to reverse the derivative process for f(x) x 2. Recall that the power
d
rule for derivatives dx(x n) nx n1 instructs us to
• Multiply by the power.
• Subtract 1 from the power to get the new power.
To reverse the process for antiderivatives, we
• Add 1 to the power to get the new power.
• Divide by that new power.
This formula is known as the iSmple Pow
er Rule for antiderivatives.
Simple Power Rule for Antiderivatives
x dx nx 1 C for any n
1
n1
n
This rule requires that n
1, because otherwise, we would be dividing by
zero.
In the case of f(x) x 2, where n 2, the general antiderivative is
x dx x3 C
3
2
EXAMPLE 1 Using the Simple Power Rule to Find Antiderivatives
Find the following general antiderivatives and their appropriate units.
a. 7dx degrees per hour, where x is in hours
b. h dh parts per million per day, where h is in days
0.5
Solution
a. 7dx 7x C degrees
h dh h1.5 C parts per million
1.5
b. 0.5 ●
330 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Recall the Constant Multiplier Rule for derivatives:
If g(x) kf(x), then g(x) kf(x), where k is a constant.
A similar rule applies for antiderivatives:
Constant Multiplier Rule for Antiderivatives
kf(x)dx k f(x)dx
7 7
Thus, 12x 6dx 12 x 6dx 12 x7 C 12x7 C.
Another property of antiderivatives is the Sum Rule.
Sum Rule
[f(x) g(x)]dx f(x)dx g(x)dx
The Sum Rule lets us find an antiderivative for a sum (or difference) of functions
by operating on each function independently. For example,
(7x 3 x)dx 7x dx xdx
3
4 2
7x4 x2
C1
C2
7x 4 x2
C (Combine C1 and C2 into one constant C.)
4 2
Repeated applications of the Simple Power Rule, the Constant Multiplier Rule,
and the Sum Rule enable us to find an antiderivative of any polynomial function. We
now have the tools we need to begin with a simple polynomial rate-of-change
function for a quantity and recover an amount function for that quantity.
EXAMPLE 2 Using Given Information to Find an Antiderivative
Birth Rate An African country has an increasing population but a declining
birth rate, a situation that results in the number of babies born each year increasing but
at a slower rate. The rate of change in the number of babies born each year is given by
b(t) 87,000 1600t births per year
t years from the end of this year. Also, the number of babies born in the current year
is 1,185,800.
a. Find a function describing the number of births each year t years from now.
b. Use the function in part a to estimate the number of babies that will be born
next year.
5.3 The Fundamental Theorem 331
Solution
a. A function B describing the number of births each year is found as
B(t) b(t)dt 87,000t 1600
2
t 2 C 87,000t 800t 2 C births
t years from now.
We also know that B(0) 1,185,800, so 1,185,800 87,000 (0) 800 (02) C,
which gives C 1,185,800. Thus we have
B(t) 87,000t 800t 2 1,185,800 births each year
t years from now.
b. The number of babies born next year is estimated as B(1) 1,272,000 babies. ●
We have just presented and applied three antiderivative rules: the Simple Power
Rule, the Constant Multiplier Rule, and the Sum Rule. Now let us look at four more
rules for finding antiderivatives.
Refer to the Simple Power Rule, and note that it does not apply for n 1. The
1
case where n 1 is special. This results in the antiderivative x 1dx x dx. Recall
d 1
that dx (ln x) x. This is valid only for x 0, because ln x is not defined for x 0.
When we are not certain that x 0, we use ln x .
Natural Log Rule
1x dx ln x C
The two final antiderivative formulas that we consider are for exponential func-
tions. Recall that the derivative of f(x) e x is e x. Similarly, the general antiderivative
of f(x) e x is also ex plus a constant.
ex Rule
e dx e
x x C
The other exponential function that we have encountered is f(x) b x. Its deriv-
ative was found by multiplying bx by ln b. To find the general antiderivative, we divide
bx by ln b and add a constant:
332 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Exponential Rule
b dx lnb b C
x
x
Note that the ex Rule is a special case of this Exponential Rule with b e. If we use
the Exponential Rule to find the antiderivative of ex, we have e xdx ex
ln e
e x because
ln e 1.
Because we often encounter functions of the form f(x) e ax, it is helpful to know
this function’s derivative formula. Recall that the Chain Rule for derivatives applied
to f(x) e ax gives f(x) ae ax. Thus, to reverse the derivative process, we leave the
eax-term and divide, rather than multiply, by a.
eax Rule
e axdx
1 ax
a
e C for any a
0
In summary, we now have the following antiderivative formulas, where k, n, b,
and a are constants.
Antiderivative Formulas
Function: General antiderivative:
f f (x)dx
Constant Rule k kx C
Simple Power Rule x n, n
1 n 1 1x n1 C
1
Natural Log Rule ln x C
x
Exponential Rule b x, b 0 ln1 bb x C
ex Rule ex ex C
1 ax
eax Rule e ax, a
0 e C
a
Constant Multiplier Rule kg(x)
k g(x)dx
Sum Rule g(x) h(x) g(x)dx h(x)dx
5.3 The Fundamental Theorem 333
EXAMPLE 3 Using Antiderivative Formulas to Find General Antiderivatives
Find the following general antiderivatives.
a. 3 7e 5xdx quarts per hour, where x is measured in hours
x x
b. (4x 100e 0.46)dx mpg per mph, where x is measured in mph
0.06x
Solution
a. 3 7e 5xdx 3 dx 7 e dx 5 1x dx
x x x x
3x
7e x 5 ln x C quarts
ln 3
b. We must first rewrite x as x 1/2 and then apply the appropriate rules:
4x 100e 0.06x
0.46 dx [4x 1/2 100e 0.06x 0.46]dx
4 x 1/2dx
100 e 0.06xdx 0.46dx
x 3/2 e 0.06x
4 100 0.46x C
3/2 0.06
8 3/2 100 0.06x
x e 0.46x C mpg ●
3 0.06
Specific Antiderivatives
We have seen that any function has infinitely many antiderivatives, each differing
by a constant. When we seek an antiderivative with a particular constant, the re-
sulting function is a specific antiderivative. An example of a specific antiderivative
is the function in Example 2 for the number of births in an African country. Using
the information that in the current year the number of births is 1,185,800,
we were able to solve for the constant to obtain the specific antiderivative
B(t) 87,000t 800t 2 1,185,800 births each year t years from now.
In general, to find a specific antiderivative, you must be given additional infor-
mation about the output quantity that the antiderivative describes. After finding a
general antiderivative, you simply substitute the given input and corresponding out-
put into the general antiderivative and solve for the constant. Then you replace the
constant in the general antiderivative formula with the value you found to obtain the
specific antiderivative. Example 4 illustrates this process.
EXAMPLE 4 Finding a Specific Antiderivative
Marginal Cost Suppose that a manufacturer of small toaster ovens has collected the
data given in Table 5.13, which shows, at various production levels, the approximate
334 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
cost to produce one more oven. Recall from Section 4.1 that this is marginal cost and
can be interpreted as the rate of change of cost.
TABLE 5.13
Production level (ovens per day) 200 300 400 500 600 700
Cost to produce an additional oven $29 $20 $15 $11 $9 $7
The manufacturer also knows that the total cost to produce 250 ovens is $12,000.
a. Find a model for the marginal cost data.
b. Recover a cost model from the model you found in part a.
c. Estimate the cost of producing 500 ovens.
Solution
a. Either a quadratic or an exponential model is a good fit to the data. We choose
an exponential model:
C(x) 47.6(0.997x) dollars per oven (Note: We have rounded to simplify
calculations.)
where x is the number of ovens produced.
b. To recover a model for cost, we need an antiderivative of C satisfying the
known condition that the cost to produce 250 ovens is $12,000. The general
antiderivative is
47.6(0.997x)
C(x) K 15,843 (0.997x) K
ln 0.997
where K is a constant.
Using the fact that C(250) $12,000, we substitute 250 for x, set the
antiderivative equal to 12,000, and solve for K.
15,843(0.997250) K 12,000
7475 K 12,000
K 19,475
Thus the specific antiderivative giving the approximate cost of producing
x toaster ovens is
C(x) 15,843(0.997x) 19,475 dollars
c. You can readily determine that the cost of producing 500 toaster ovens is
estimated by C(500) $15,948. ●
It is sometimes necessary to find an antiderivative twice in order to obtain the
appropriate accumulation formula. For example, to obtain distance from accelera-
tion, you must determine the specific antiderivative of the acceleration function to
5.3 The Fundamental Theorem 335
obtain a velocity function and then determine the specific antiderivative of the veloc-
ity function to obtain a function for distance traveled.
EXAMPLE 5 Recovering Distance from Acceleration
Falling Pianos A mathematically inclined cartoonist wants to make sure his ani-
mated cartoons accurately portray the laws of physics. In a particular cartoon he is
creating, a grand piano falls from the top of a 10-story building. How many seconds
should he allow the piano to fall before it hits the ground? Assume that one story
equals 12 feet and that acceleration due to gravity is 32 feet per second squared.
Solution
We begin with the equation for acceleration, a(t) 32 feet per second squared t
seconds after the piano falls. The antiderivative of acceleration gives an equation for
velocity:
a(t)dt v(t) 32t C feet per second after t seconds of fall
To find the specific velocity equation for this example, we need information
about the velocity at a specific time. The fact that the piano falls (rather than being
pushed with an initial force) tells us that the velocity is zero when the time is zero.
Substituting zero for both t and v and solving for C give C 0. Thus the specific an-
tiderivative describing velocity is v(t) 32t feet per second after t seconds of fall.
From this velocity equation, we can derive an equation for position (the distance
the piano is above the ground) by finding the antiderivative. In this case, the general
antiderivative of v is
v(t)dt s(t) 16t 2 K feet after t seconds of fall
Again, we find the specific position equation by using information given about
the position of the piano at a certain time. In this example we know that when time
is zero, the piano is 10 stories, or 120 feet, above the ground. We substitute this
information into the position equation and solve for K, obtaining K 120 feet.
Substituting this value for K in the position equation yields the specific position
equation
s(t) 16t 2 120 feet after t seconds of fall
Now that we have equations for acceleration, velocity, and position, we can
answer the question posed: How long will it be until the piano hits the ground? Let us
phrase this question mathematically: When position is zero, what is time? To answer
this question, we set the position equation equal to zero and solve for t.
0 16t 2 120
16t 2 120
120
t2
16
t 2.7 seconds
336 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Because negative time doesn’t make sense in the context of the question, we
conclude that the cartoonist should allow approximately 2.7 seconds for the piano
to fall. ●
Let us summarize what we have learned thus far about integrals. The definite inte-
f(x)dx is a limiting value of sums of areas of rectangles and gives us the area of
b
gral
a
the region between the graph of f and the x-axis if the graph lies above the horizontal
axis from a to b. When the graph of f lies below the horizontal axis from a to b, the
definite integral is the negative of the inscribed area. If f(x) is a rate of change of some
f(x)dx is the change in the quantity from a to b. The accumulation
b
quantity, then
a
function A(x) f(t)dt is a formula in terms of x for the accumulated change in f(t)
x
from a to x. We use the integral symbol without the upper and lower limits, f(x)dx,
a
to represent the general antiderivative of f. Although these three symbols are
similar, it is important that you have a clear understanding of what each one repre-
sents. Their interpretations are summarized in Table 5.14.
TABLE 5.14
Symbol Name Interpretation
f(x)dx
b
definite integral a number that can be thought of in
a
terms of the accumulated change
in f between the input values of a
and b
f(t)dt
x
accumulation function, a formula for an accumulated
a
specific antiderivative amount between the starting input
value of a and a general input
value of x
f(x)dx general antiderivative a formula whose derivative is f
The Fundamental Theorem of Calculus tells us that accumulation functions are
specific antiderivatives. As we shall see in Section 5.4, antiderivatives enable us to find
areas algebraically by using accumulation formulas, rather than numerically as limit-
ing values of sums of areas of rectangles.
Apart from helping us find areas or accumulated change, antiderivatives are
useful in enabling us to recover functions from rates of change. We have seen several
examples of that in this section. It may seem difficult to reverse your thinking from
finding derivatives to finding antiderivatives, but with practice you will soon be
proficient at both.
5.3 The Fundamental Theorem 337
Find the general antiderivative as indicated in Activities 9
5.3 Concept Inventory through 14. Check each of your antiderivatives by taking
its derivative.
19.4(1.07 )dx
• Fundamental Theorem of Calculus
.9 x
• Antiderivative
10. 39e dx
• Recovering a function from its rate of change 3.9x
• f(x)dx general antiderivative
11. [6e 4(2 )]dx
x x
• Antiderivative formulas
12. (32.68x 3.28x 15)dx
• Specific antiderivative 3
13. (10 4x 8)dx
x
5.3 Activities
14. x dx
1 1 1 x
Getting Started
2 2x 2
In Activities 1 through 4, a and b are constants and x and
t are variables. In these activities, label each notation as For each of the rate-of-change functions in Activities 15
always representing: through 18, find the general antiderivative, and label the
i. A function of x units on the antiderivative.
ii. A function of t 15. s(m) 600(0.93m) CDs per month, m months
since the beginning of the year
iii. A number
16. p(x) 0.04x 2 0.5x 1.4 dollars per 1000 cubic
df feet per year, x years since 1989
1. a. f (t) b. c. f (3)
dx 0.8
17. c(x) x 0.38(0.01x) dollars per unit squared,
f(t)dt f(t)dt
b when x units are produced
2. a. b. f(x)dx c.
a 18. p(t) 1.7e0.03t millions of people per year, t years
3. a. f(x)dx b. f(t)dt c. f(x)dx
b x t after 1990
a a b In Activities 19 through 21, find F, the specific anti-
dx
b. f(x)dx
dx
x t a derivative of f.
d d d
4. a. f(t)dt c. f(t)dt
a dt a a 19. f(t) t 2 2t ; f(12) 700
5. Illustrate and explain the Fundamental Theorem of 2
20. f(u) u; f(1) 5
Calculus from a numerical viewpoint. u
6. Write the Fundamental Theorem of Calculus from 1
21. f(z) e z; f(2) 1
an algebraic viewpoint. z2
.7 Write the Fundamental Theorem of Calculus from
Applying Concepts
a verbal viewpoint. Do not include mathematical
symbols or graphs. 22. Bond Y ields The rate of change of the average
yield of short-term German bonds can be described
.8 Illustrate and explain the Fundamental Theorem of 0.57
Calculus from a graphical viewpoint. by the equation G(t) t percentage points per
338 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
year, for a bond with a maturity time of t years. The b. Recover the amount function, and determine the
average 10-year bond has a yield of 4.95%. Find the current value of the investment and its projected
specific antiderivative describing the average yield of value next year.
short-term German bonds. How is this specific anti-
27. D
ropped C oin The Washington Monument,
derivative related to an accumulation function of G?
located at one end of the Federal Mall in Washing-
23. W
eigh
t The rate of change of the weight of a ton, D.C., is the world’s tallest obelisk at 555 feet.
laboratory mouse can be modeled by the equation Suppose that a tourist drops a penny from the
7.37 observation deck atop the monument. Let us
W(t) t grams per week, where t is the age of the
assume that the penny falls from a height of
mouse, in weeks, beyond 2 weeks. At an age of
540 feet.
9 weeks, a mouse weighed 26 grams. Find the spe-
cific antiderivative describing the weight of the a. Recover the velocity function for the penny using
mouse. How is this specific antiderivative related to the following facts:
an accumulation function of W? i. Acceleration due to gravity near the sur-
face of the earth is 32 feet per second
24. u
Fel sUe The rate of change of the average annual
squared.
fuel consumption of passenger vehicles, buses,
and trucks from 1970 through 2000 can be modeled ii. Because the penny is dropped, velocity is 0
by the equation f(t) 0.8t 15.9 gallons per when time is 0.
vehicle per year, t years after 1970. The average an- b. Recover the distance function for the penny us-
nual fuel consumption was 712 gallons per vehicle ing the velocity function from part a and
in 1980. the fact that distance is 540 feet when time
(Source: Based on data from Bureau of Transportation is 0.
Statistics.)
c. When will the penny hit the ground?
a. Find the specific antiderivative giving the average
annual fuel consumption. 28. H
ighDive According to the Guinness Book
b. How is this specific antiderivative related to an of Records, the world’s record high dive from a
accumulation function of f ? diving board is 176 feet, 10 inches. It was made
by Olivier Favre (Switzerland) in 1987. Ignoring
25. G
ender Ratio The rate of change of the gender air resistance, approximate Favre’s impact velocity
ratio for the United States during the twentieth cen- (in miles per hour) from a height of 176 feet,
tury can be modeled as g(t) (1.67 104)t 2 10 inches.
0.02t 0.10 males per 100 females per year, t years
after 1900. In 1970 the gender ratio was 94.8 males 29. *Velocity In the 1960s, Donald McDonald claimed
per 100 females. in an article in New Scientist that plummeting cats
(Source: Based on data from U.S. Bureau of the Census.) never fall faster than 40 mph.
a. Find a specific antiderivative giving the gender a. What is the impact velocity (in feet per second and
ratio. miles per hour) of a cat that accidentally falls off a
b. How is this specific antiderivative related to an 1
building from a height of 66 feet 52 stories ?
accumulation function of g?
b. What accounts for the difference between your
26. Investment An investment worth $1 million in answer to part a and McDonald’s claim (assum-
2000 has been growing at a rate of f(t) ing McDonald’s claim is accurate)?
0.140(1.15t) million dollars per year, t years after
2000.
a. Determine how much the investment has grown * McDonald’s study was based on observations of veterinarians who
since 2000 and how much it is projected to grow treated cats that had fallen from buildings in New York City. None of
over the next year. the cats’ falls were deliberately caused by the researchers.
5.4 The Definite Integral 339
30. o
Dnors The table gives the increase or decrease in b. Find a model for the number of donors. Use the
the number of donors to a college athletics support fact that in 1990 there were 10,706 donors.
organization for selected years. c. Estimate the number of donors in 2002.
31. Employees From 1997 through 2002, an Inter-
net company was hiring new employees at a rate of
Rate of change 593
n(x) x 138 new employees per year, where x
in donors
Year (donors per year) represents the number of years since 1996. By 2001
the company had hired 896 employees.
1985 169
1988 803 a. Write the function that gives the number of em-
ployees who had been hired by the xth year after
1991 1222
1996.
1994 1087
b. For what years will the function in part a
1997 399
apply?
2000 842
c. Find the total number of employees the com-
pany had hired between 1997 and 2002. Would
this figure necessarily be the same as the number
a. Find a model for the rate of change in the num- of employees the company had at the end of
ber of donors. 2002? Explain.
5.4 The Definite Integral
So far, we have been finding general and specific antiderivatives to recover functions
from their rate-of-change functions. The Fundamental Theorem of Calculus tells us
that accumulation functions are antiderivatives and leads to an algebraic method for
x
finding an accumulation function. In general, we know that f(t)dt F(x) C,
a
where F is an antiderivative of f. We also know that when x a, the accumulation
function is zero.
f(t)dt F(a) C 0
a
This tells us that C F(a). Thus we have
f(t)dt F(x) F(a)
x
To find the value of the accumulation function from a to b, we simply substitute b
for x.
f(t)dt F(b) F(a)
b
a
340 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
We now have an efficient algebraic method for evaluating definite integrals:
Evaluating a Definite Integral
If f is a continuous function from a to b and F is any antiderivative of f, then
f(x)dx F(b) F(a)
b
Antiderivatives and Definite Integrals
f(x)dx, as the limiting
b
Recall from Section 5.1 that we define the definite integral,
a
value of sums of signed areas of rectangles. That is,
f(x)dx lim [f(x ) f(x ) … f(x )]x
b
n→ 1 2 n
a
f(x)dx F(b) F(a), gives us
b
The antiderivative definition for a definite integral,
a
a less tedious method for evaluating a definite integral for many functions. For ex-
ample, we can calculate the area of the region between the graph of f(x) x 2 2
and the x-axis from 2 to 4. (See Figure 5.48.)
f (x)
28
24
20
16
12
8
4
x
-3 -2 -1 0 1 2 3 4 5
FIGURE 5.48
To find the area of this region in Section 5.1, we would have used sums of
areas of rectangles. Now all we need to do is calculate the value of the definite integral,
4
(x 2 2)dx, by simply finding an antiderivative and subtracting the value of the
2
antiderivative at 2 from the value of the antiderivative at 4.
4
(x 2 2)dx F(4) F(2)
2
5.4 The Definite Integral 341
where F is an antiderivative of x 2 2. Here are the details of this process:
x3 2x
This notation is used
4 3 4
1. Find an antiderivative: (x 2 2)dx to indicate that we
2 2 have found an
antiderivative and
2. Evaluate at upper and lower limits and then subtract:
now must evaluate
43 (2)3 that antiderivative at
2(4) 2(2) 4 and 2 and then
3 3
find the difference of
1
293 6 3
2 the results.
36
Thus the area of the region depicted in Figure 5.48 is exactly 36.
EXAMPLE 1 Evaluating and Interpreting a Definite Integral
a. Find a formula for 70(1.07t)dt .
70(1.07 )dt.
x
b. Find a formula for the accumulation function A(x) t
1
70(1.07 )dt. Interpret your answer graphically.
4
c. Determine t
1
d. If f(t) 70(1.07t) is the rate of change of the balance in a savings account given
in dollars per year, and t is the number of years since the savings account was
opened, what does your answer in part c represent?
Solution
70(1.07 )dt 70(1.07)
t
a. t C
ln 1.07
b. The accumulation function is the specific antiderivative (in terms of x) for
which the antiderivative is zero when x 1.
70(1.071)
A(1) C0
ln 1.07
C 1107.03
Thus we have
70(1.07 )dt 70(1.07
x ) x
A(x) t 1107.03
1 ln 1.07
342 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
70(1.07 )dt A(t)
4 4
c. t
1 1
4
70(1.07t)
ln 1.07 1
70(1.074) 70(1.071)
ln 1.07 ln 1.07
249.13
The graph of f(t) 70(1.07t) is always positive. This value is the area of the
region between the graph of f(t) and the horizontal axis from 1 to 4.
d. The answer to part c represents the change in the amount in the savings account
between 1 and 4 years. The amount grew by $249.13. ●
The constant term in an antiderivative does not affect definite integral
calculations. If you are concerned only with finding change in a quantity,
finding the constant in the antiderivative is not necessary.
EXAMPLE 2 Using a Definite Integral to Find Accumulated Change
Marginal Cost In Example 4 of Section 5.3 we modeled the marginal cost for
toaster ovens using the exponential function
C(x) 47.6(0.997x) dollars per oven
where x is the number of ovens produced per day. Suppose that the current production
level is 300 ovens per day and that the manufacturer wishes to increase production to
500 ovens per day. How will this increase affect production cost?
Solution
500
The definite integral C(x)dx C(500) C(300) gives the change in cost as a
300
result of this increase. Finding the change requires two steps. First, find an antideriv-
ative of C. Then evaluate the antiderivative at the two production levels and subtract
the value at the lower limit of the integral from the value at the upper limit.
In Section 5.3 (using a rounded model), we found the general antiderivative for
C to be
C(x) C(x)dx 15,843(0.997 ) K x
Because the constant K will not affect our calculations of change, we can ignore K.
500 500
We illustrate this by finding C(x)dx C(x) and including the value of K
300 300
$19,475 that we found in Example 4 of Section 5.3.
5.4 The Definite Integral 343
The definite integral is
500 500
C(x)dx (15,843(0.997x) 19,475)
300 300
C(500) C(300)
[15,843(0.997500) 19,475][15,843(0.997300) 19,475]
$2905
Note that 19,475 will be added in and then subtracted. We ignore the constant
when calculating a definite integral.
We conclude that when production is increased from 300 to 500 ovens per day,
cost increases by approximately $2,905. ●
Piecewise Functions
Piecewise functions are defined using two or more separate continuous equations.
The values of x where the specific equation that defines the function changes are
called break points.
f(x) 18x2 135x 2,882 ft3/hr when 0 x 20
12x2 816x 19,044 ft3/hr when 20 x 27
is a piecewise function that describes flow rates of a river past a sensor in cubic feet
per hour, for 27 hours following a storm. The break point for f(x) occurs at x 20.
The graph of f(x) is shown in Figure 5.49.
Flow rate
(ft3/hr)
8,000
6,000
4,000
Time
2,000
0 12 27
(hours)
FIGURE 5.49
Calculations for definite integrals require special care. To calculate the area
trapped between f(x) and the horizontal axis for 0 x 27, we must use two defi-
nite integrals.
Sums of Definite Integrals
To estimate the amount of water that flowed through the river over the 27-hour period,
27
we calculate the value of the definite integral, f(x)dx. Note that the point of division
0
for the model occurs in the interval from 0 to 27. For this reason, we cannot calculate
344 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
the value of the definite integral simply by evaluating an antiderivative of f at 27 and
0 and subtracting.
f (x) Note that the area of the region from a to b shaded in Figure 5.50 is equal to the
sum of the area of R1 and the area of R2.
This figure illustrates the following property of integrals:
Sum Property of Integrals
R1 R2
f(x)dx f(x)dx f(x)dx
b c b
x
a c b
a a c
FIGURE 5.50 where c is between a and b.
It is this property that enables us to calculate definite integrals for piecewise con-
tinuous functions.
27
Returning to the river flow function, in order to calculate f(x)dx, we divide
0
the integral into two pieces at the point where the model changes and sum the results.
f(x)dx f(x)dx
27 20 27
f(x)dx
0 0 20
(18x 135x 2,882)dx
20
2
0
(12x 816x 19,044)dx
27
2
20
2,882x
18x 135x3 2 20
3 2 0
19,044x
12x 816x 3 2 27
3 2 20
(78,640 0) (295,488 249,680)
78,640 45,808
124,448 cubic feet
We estimate that during the first 20 hours, 78,640 thousand ft3 of water flowed past
the sensor. Between 20 and 27 hours, the volume of water was about 45,808 thousand
ft3. Summing these two values, we estimate that over the 27-hour period, 124,448
thousand ft3 of water flowed past the sensor.
In order to calculate the change from a to b in a function whose graph is some-
times above and sometimes below the horizontal axis, it is necessary only to calculate
f(x)dx; we need not calculate the change over separate intervals. This concept is
b
a
illustrated in Example 3.
5.4 The Definite Integral 345
EXAMPLE 3 Illustrating the Sum Property of Integrals
Sea Level Scientists believe that the average sea level is dropping and has been for
some 4000 years. They also believe that this was not always the case. Estimated rates
of change in the average sea level, in meters per year, during the past 7000 years are
given in Table 5.15.
TABLE 5.15
Time, t (thousands of
years before present) 7 6 5 4 3 2 1
Rate of change of
average sea level, r (t )
(meters/year) 3.8 2.6 1.0 0.1 0.6 0.9 1.0
(Source: Estimated from information in François Ramade, Ecology of Natural Resources,
New York: Wiley, 1981)
A quadratic model for the data is
r(t) 0.148t 2 0.360t 0.8 meters per yard
t thousand years from the present (past years are represented by negative numbers).
A graph of r is shown in Figure 5.51.
r (t)
(meters per year)
4
3
2
R1 1
t
Years 0
(thousands) -7 -6 -5 -4 -3 -2 R2-1
-1
FIGURE 5.51
a. Find the areas of the regions above and below the t-axis from t 7 to t 0.
Interpret the areas in the context of sea level.
0
b. Find r(t)dt , and interpret your answer.
7
Solution
a. The graph in Figure 5.51 crosses the t-axis at t 3.845 thousand years. The
area of region R1 (above the t-axis) is
3.845 3.845
0.148 t 0.8t
0.36 2
(0.148t 2 0.36t 0.8)dt t3
7 3 2 7
2.933 (2.501) 5.4 meters
346 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
The area of region R2 (below the t-axis) is
t 0.8t
0 0
3.845
(0.148t 2 0.36t 0.8)dt 0.148
3
t3
0.36 2
2 3.845
(0 2.933) 2.9 meters
From 7000 years ago through 3845 years ago, the average sea level rose by
approximately 5.4 meters. From 3845 years ago to the present, the average sea
level fell by approximately 2.9 meters.
0 0
b. r(t)dt (0.148t 2 0.36t 0.8)dt
7 7
t 0.8t
0
0.148
3
t3
0.36 2
2 7
0 (2.501) 2.5 meters
From 7000 years ago to the present, the average sea level has risen approxi-
mately 2.5 meters. This result is the same as that obtained by subtracting the
amount that the average sea level has fallen from the amount that it has risen:
5.4 meters 2.9 meters 2.5 meters ●
Differences of Accumulated Changes
Now we turn our attention to the difference of two accumulated changes. This differ-
ence can often be thought of as the area of a region between two curves. For example,
suppose the number of patients admitted to a large inner-city hospital is changing by
a(h) 0.0145h3 0.549h2 4.85h 8.00 patients per hour
h hours after 3 A.M. We find the approximate number of patients admitted to the
hospital between 7 A.M. (h 4) and 10 A.M. (h 7) as
a(h)dh (0.0145h 0.549h 4.85h 8.00)dh
7 7
3 2
4 4
8.00h
4 3 2 7
0.0145h 0.549h 4.85h
4 3 2 4
120.760 60.016 61 patients
Graphically, this value is the area of the region between the graph of a and the
horizontal axis from 4 to 7. (See Figure 5.52.)
Now suppose that the rate at which patients are discharged is modeled by the
equation
y(h)
0 patients/hour
0.028h3 0.528h2 0.056h 1.5 patients/hour when 4 h 17
0 patients/hour
when 0 h 4
when 17 h 24
5.4 The Definite Integral 347
a(h)
(patients
per hour)
20
15
10 R1
5
h
0
0 4 8 12 16 20
Hours
(3 A.M.) (7 A.M.) (11 A.M.) (3 P.M.) (7 P.M.) (11 P.M.)
The area of R1 is the number of patients admitted between 7 A.M. and
10 A.M.
FIGURE 5.52
where h is the number of hours after 3 A.M. The approximate number of patients dis-
charged between 7 A.M. and 10 A.M. is calculated as
y(h)dh (0.028h 0.528h 0.056h 1.5)dh
7 7
3 2
4 4
(0.007h4 0.176h3 0.028h2 1.5h) 74
34.433 3.92 31 patients
Graphically, this value is the area of the region between the graph of y and the
horizontal axis from 4 to 7. (See Figure 5.53.)
y(h)
(patients
per hour)
25
20
15
10
5 R2
h
0
0
Hours
4 8 12 16 20
(3 A.M.) (7 A.M.) (11 A.M.) (3 A.M.) (7 A.M.) (11 A.M.)
The area of R2 is the number of patients discharged between 7 A.M. and
10 A.M.
FIGURE 5.53
The net change in the number of patients at the hospital from 7 A.M. to 10 A.M. is
the difference between the number of patients admitted and the number discharged
between 7 A.M. and 10 A.M. That is,
348 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Change in the number of patients
a(h)dh y(h)dh
7 7
from 7 A.M. to 10 A.M.
4 4
60.744 30.513
30 patients
Geometrically, we represent this value as the area of the region below the graph of
a and above the graph of y from 4 to 7. (See Figure 5.54.)
Patients admitted
and discharged
(patients per hour)
25
20
15 R3 y(h)
10
a(h)
5
h
0 Hours
0 4 8 12 16 20 after 3 A.M.
The area of R3 is the net change in hospital patients between 7 A.M. and 10 A.M.
FIGURE 5.54
f (x) In general, when we want to find the area of a region that lies below one curve f
and above another curve g from a to b (as in Figure 5.55), we calculate it as
Area of the region between area beneath area beneath
the graphs of f and g the graph of f the graph of g
f(x)dx g(x)dx
b b
R
g(x) a a
Using the Sum Rule for antiderivatives, we obtain
[f(x) g(x)]dx
Area of the region between b
x
a b the graphs f and g a
FIGURE 5.55 Note that when f and g are obtained by fitting equations to data, the input vari-
ables of the functions must represent the same quantities measured in the same units.
If, while calculating the area of
the region between two curves,
you obtain the negative of the
Area of the Region Between Two Curves
answer you expect, then it is If the graph of f lies above the graph of g from a to b, then the area of the
likely that you have
interchanged the positions of region between the two curves from a to b is given by
[f(x)g(x)]dx
the functions in the integrand. b
a
5.4 The Definite Integral 349
EXAMPLE 4 Determining the Area of the Region Between Two Curves
Tire Manufacturers A major European tire manufacturer has seen its sales from
tires skyrocket since 1989. A model for the rate of change of sales (in U.S. dollars)
accumulated since 1989 is
s(t) 3.7(1.194t) million dollars per year
where t is the number of years since the end of 1989.
At the same time, an American tire manufacturer’s rate of change of sales
accumulated since 1989 can be modeled by
a(t) 0.04t 3 0.54t 2 2.5t 4.47 million dollars per year
where t is the number of years since the end of 1989. These models apply through the
year 2010. By how much did the amounts of accumulated sales differ for these two
companies from the end of 1999 through 2009?
Solution
First, determine whether one graph lies above the other on the entire interval in the
question. The two rate-of-change functions are graphed in Figure 5.56.
Between the years 1999 and 2010, the graphs of the two rate-of-change functions
s(t) and a(t) cross once near t 14. If we set s(t) 3.7(1.194t) equal to
a(t) 0.04t 3 0.54t 2 2.5t 4.47 and solve for t, we find that on the interval
[10, 21], the two functions intersect when t 14.28 (in 2004). (There are other
intersections outside the given interval at t 0.37, t 4.65, and t 27.97). Ac-
cumulated sales were greater for the European company than for the American com-
pany from t 4.65 to t 14.28. Between t 14.28 and t 21, the American
company saw greater accumulated sales than the European company.
Rate of change
of total sales
(million dollars
per year)
160
a (t) s (t)
140
120
100
80
60
40
20
t
0
0 6 11 16 21 Years
FIGURE 5.56 (1989) (1995) (2000) (2005) (2010)
350 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
From the beginning of 2000 (t 10) through most of the first quarter of 2004
(t 14.28), the European company accumulated approximately
14.28
[s(t) a(t)]dt 18.88 million dollars
10
more in sales than the American company. This is the area of region R1 in Figure 5.57.
Rate of change
of total sales
(million dollars
per year)
160
a (t) s (t)
140
120
R2
100
80
60
40 R1
20
t
0
0 6 11 16 21 Years
(1989) (1995) (2000) (2005) (2010)
FIGURE 5.57
From close to the end of the first quarter of 2004 (t 14.28) through 2009
(t 20), the American company accumulated approximately
20
[a(t) s(t)]dt 77.77 million dollars
14.28
more in sales than the European company. This is the area of region R2 in Figure 5.57.
In order to calculate the estimated difference in accumulated sales between the
two companies from 1999 through 2009, we subtract the portion where the Ameri-
can company’s accumulated sales were greater from the portion where the European
company’s accumulated sales were greater. That is,
Difference in 14.28 20
[s(t) a(t)]dt [a(t) s(t)]dt
accumulated sales 10 14.28
18.88 million dollars 77.77 million dollars
58.89 million dollars
The European company’s accumulated sales were nearly 59 million dollars less than
the American company’s accumulated sales over the years considered. ●
5.4 The Definite Integral 351
If we use the Constant Multiplier Rule and the Sum Rule for antiderivatives with
the functions in Example 4, we see that we did not need to split the interval from 10
to 20 into two intervals.
Difference in 14.28 20
[s(t) a(t)]dt [a(t) s(t)]dt
accumulated sales 10 14.28
[s(t) a(t)]dt
14.28 20
[a(t) s(t)]dt
10 14.28
[s(t) a(t)]dt
14.28 20
[s(t) a(t)]dt
10 14.28
20
[s(t) a(t)]dt
10
This is true in general—when you wish to find the difference between accumulated
change for two continuous rate-of-change functions, you can calculate the definite
integral of the difference of the functions, regardless of where the functions intersect.
Difference of Two Accumulated Changes
If f and g are two continuous rate-of-change functions, then the difference
between the accumulated change of f from a to b and the accumulated
change of g from a to b is
[f(x) g(x)]dx
b
[f(x) g(x)]dx may not represent the area between
b
It is important to note that
a
the graphs of f and g from a to b. In the tire sales example, the area of the regions be-
tween the two curves from 10 to 20 is 18.88 77.77 96.65, whereas the difference
in accumulated sales is 18.88 77.77 58.89.
If two rate-of-change functions intersect in the interval from a to b, then the
difference between their accumulated changes is not the same as the area of
the regions between the two curves.
Most practical applications of the area between two curves involve the difference
between two accumulated changes. However, if total area is to be calculated, remem-
ber the distinction between these quantities.
In this section, we have seen that the Fundamental Theorem of Calculus gives us
a technique for evaluating definite integrals using antiderivatives. However, there are
many antiderivative formulas not covered in this text, and there are some functions
to which no antiderivative rule applies. It is important for you to understand that if
you are ever unable to find an algebraic formula for the antiderivative of a function,
you can still estimate the value of a definite integral of that function by using the
limiting value of sums of areas of rectangles.
352 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
b. Find the area of the region between the graph of
5.4 Concept Inventory f and the x-axis from a to b. Is this area equal to
f(x)dx? Explain.
b
f(x)dx F(b) F(a), where F is an
b
a
•
c. Find f(x)dx.
a b
antiderivative of f
a
f(x)dx f(x)dx f(x)dx
b c b
• .8 f(x) 4x 2; a 1, b 4
a a c
• Area(s) of region(s) between two curves .9 f(x) 1.3x 3 0.93x 2 0.49; a 1, b 2
• Differences of accumulated changes 9.295
10. f(x) 1.472; a 5, b 10
x
11. f(x) 965.27(1.079x); a 0.5, b 3.5
5.4 Activities
Applying Concepts
Getting Started
12. iAr p
S eed The air speed of a small airplane during
For each of Activities 1 through 7, determine which of
the first 25 seconds of takeoff and flight can be
the following processes you would use when answering
modeled by
the question posed. Note that a and b are constants.
a. Finding a derivative v(t) 940,602t 2 19,269.3t 0.3 mph
b. Finding a general antiderivative (with unknown t hours after takeoff.
constant) 0.005
c. Finding a specific antiderivative (solve for the a. Find the value of v(t)dt .
0
constant)
b. Interpret your answer in context.
1. Given a rate-of-change function for population and
the population in a given year, find the population 13. Ph
one Calls The rate of change of the number
in year t. of international telephone calls billed in the
United States between 1980 and 2000 can be
2. Given a velocity function, determine the distance described by
traveled from time a to time b.
P(x) 32.432e 0.1826x million calls per year
3. Given a function, find its accumulation function
from a to x. where x is the number of years after 1980. Find and
15
4. Given a velocity function, determine acceleration at interpret the value of P(x)dx.
5
time t.
(Source: Based on data from the Federal Communications
5. Given a rate-of-change function for population, Commission.)
find the change in population from year a to 14. eWigh
t The rate of change of the weight of a lab-
year b. oratory mouse t weeks (for 1 t 15) after the
6. Given a function, find the area of the region between beginning of an experiment can be modeled by the
the function and the horizontal axis from a to b. equation
.7 Given a function, find the slope of the tangent line 13.785
w(t) grams per week
at input a. t
w(t)dt, and interpret your answer.
9
In Activities 8 through 11:
Evaluate
a. Graph the function f from a to b. 3
5.4 The Definite Integral 353
15. Revenue A corporation’s revenue flow rate can be 18. Temperature The rate of change of the tempera-
modeled by ture during the hour and a half after the beginning
of a thunderstorm is given by
r(x) 9.907x 2 40.769x 58.492
million dollars per year T(h) 9.48h3 15.49h2 17.38h 9.87 °F per hour
r(x)dx, and
5 where h is the number of hours since the storm began.
x years after the end of 1987. Evaluate
0 a. Graph the function T from h 0 to h 1.5.
interpret your answer.
1.5
16. eMdicine Consider the rate of change in the concen- b. Calculate the value of T(h)dh. Interpret your
0
tration of a drug modeled by answer.
19. Temperature The rate of change of the tempera-
1.708(0.845x)
ture in a museum during a junior high school field
g/mL/day when 0 x 20
r(x) trip can be modeled by
0.11875x 3.5854
g/mL/day when 20 x 29 T(h) 9.07h3 24.69h2 14.87h 0.03
F per hour
where x is the number of days after the drug was ad- h hours after 8:30 A.M.
ministered. Determine the values of the following def-
inite integrals, and interpret your answers. a. Find the area of the region that lies above the
axis between the graph of T and the h-axis
20 29
a. r(x)dx b. r(x)dx between 8:30 A.M. and 10:15 A.M. Interpret the
0 20 answer.
c.
29
r(x)dx b. Find the area of the region that lies below the
0 axis between the graph of T and the h-axis
17. n
S ow
Pack The rate of change of the snow pack in an between 8:30 A.M. and 10:15 A.M. Interpret the
area in the Northwest Territories in Canada can be answer.
modeled by c. There are items in the museum that should not
be exposed to temperatures greater than 73°F. If
0.00241t 0.02905 cm the temperature at 8:30 A.M. was 71°F, did the
per day when 0 t 70 temperature exceed 73°F between 8:30 A.M. and
s(t)
1.011t2 147.971t 10:15 A.M.?
5406.578 cm per day when 72 t 76
20. Road Test The acceleration of a race car
where t is the number of days since April 1. during the first 35 seconds of a road test is mod-
eled by
70
a. Evaluate s(t)dt , and interpret your answer.
0 a(t) 0.024t 2 1.72t 22.58 ft/sec2
b. Evaluate
76
s(t)dt , and interpret your answer. where t is the number of seconds since the test began.
72 a. Graph the function a from t 0 to t 35.
c. Explain why it is not possible to find the value of
b. Write the definite integral notation representing
76
s(t)dt . the amount by which the car’s speed increased
0 during the road test. Calculate the value of the
definite integral.
354 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
21. Production The estimated production rate of amount x spent on advertising. Use the fact that
marketed natural gas, in trillion cubic feet per year, revenue is approximately 877 thousand dollars
in the United States (excluding Alaska) from 1900 when $5000 is spent on advertising.
through 1960 is shown in the table. c. Find the point where returns begin to diminish
for sales revenue.
Estimated production
d. The managers of the business are considering an
rate (trillions of
Year cubic feet per year)
increase in advertising expenditures from the
current level of $8000 to $13,000. What effect
1900 0.1 could this decision have on sales revenue?
1910 0.5
23. Production The table shows the marginal cost to
1920 0.8 produce one more compact disc, given various
1930 2.0 hourly production levels.
1940 2.3
Production
1950 6.0
(CDs per hour) Cost of an additional CD
1960 12.7
100 $5
(Source: From information in Resources 150 $3.50
and Man, National Academy of Sciences,
1969, p. 165.) 200 $2.50
250 $2
a. Find a model for the data in the table.
300 $1.60
b. Use the model to estimate the total production of
natural gas from 1940 through 1960.
a. Find an appropriate model for the data.
c. Give the definite integral notation for your
answer to part b. b. Use your model from part a to derive an equa-
tion that specifies production cost C(x) as a func-
22. d
Avertising Many businesses spend money each tion of the number x of CDs produced. Use the
year on advertising in order to stimulate sales of fact that it costs approximately $750 to produce
their products. The data given show the approxi- 150 CDs in a 1-hour period.
mate increase in sales (in thousands of dollars) that
300
an additional $100 spent on advertising, at various c. Calculate the value of C(x)dx. Interpret your
levels, can be expected to generate. 200
answer.
Revenue increase due For Activities 24 and 25:
Advertising expenditures to an extra $100 advertising a. Sketch graphs of the functions f and g on the
(hundreds of dollars) (thousands of dollars) same axes.
25 5 b. Shade the region between the graphs of f and g
50 60 from a to b.
75 95 c. Calculate the area of the shaded region.
100 105 24. f(x) 10(0.85x) a2
125 104 g(x) 6(0.75x) b 10
150 79
25. f(x) x2 4x 10 a1
175 34
g(x) 2x 2 12x 14 b7
a. Find a model for these data. For Activities 26 and 27:
b. Use the model in part a to determine a model for a. Sketch graphs of the functions f and g on the
the total sales revenue R(x) as a function of the same axes.
5.4 The Definite Integral 355
b. Find the input value(s) at which the graphs of f dollars) invested in capital. The area of the shaded
and g intersect. region is 13.29.
c. Shade the region(s) between the graphs of f and g Rates of change of
from a to b. cost and revenue
(thousand dollars per
d. Calculate the difference between the area of the thousand dollars)
region between the graph of f and the horizontal
axis and the area of the region between the graph r'(x)
of g and the horizontal axis from a to b.
e. Calculate the total area of the shaded region(s).
26. f(x) 0.25x 3 a 15
c'(x)
g(x) 14(0.93x) b 50 x
Capital
27. f(x) e0.5x a 0.5 investment
1.5 5.5 (thousand
2 dollars)
g(x) b3
x
a. Interpret the area in the context of furniture
28. Revenue/oCst The figure depicts graphs of the manufacturing.
rate of change of total revenue R(x) (in billions b. Write an equation for the area of the shaded re-
of dollars per year) and the rate of change of total gion.
cost C(x) (in billions of dollars per year) of a com-
pany in year x. The area of the shaded region is 30. Epidemic The figure depicts graphs of c(t), the
126.5. rate at which people contract a virus during an epi-
demic, and r(t), the rate at which people recover
Rates of change of from the virus, where t is the number of days after
revenue and cost the epidemic begins.
(billion dollars
Rates of change of
per year)
contraction and recovery
(people per day)
R'(x)
c (t) R1
R2
r (t)
C'(x) t
Days
0 5 10 15 20 25
x a. Interpret the area of region R1 in the context of
Year the epidemic.
1997 1998 1999 2000 2001 2002
b. Interpret the area of region R2 in the context of
a. Interpret the area in context.
the epidemic.
b. Write an equation for the area of the shaded c. Explain how you could use a definite integral to
region. find the number of people who have contracted
the virus since day 0 and have not recovered by
29. Revenue/Cost The figure shows graphs of r(x),
day 20.
the rate of change of revenue, and c(x), the rate
of change of costs (both in thousands of dollars 31. Population A country is in a state of civil war. As
per thousand dollars of capital investment) associ- a consequence of deaths and people fleeing the
ated with the production of solid wood furniture country, its population is decreasing at a rate of
as functions of x, the amount (in thousands of D(x) people per month. The rate of increase of the
356 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
population as a result of births and immigration is Carerra to accelerate from 0 mph to the speeds
I(x) people per month. The variable x is the number given.
of months since the beginning of the year. Graphs of
D and I are shown in the figure. Region R1 has area Toyota Supra Porsche 911 Carerra
3690, and region R2 has area 9720. Time Speed Time Speed
a. Interpret the area of R1 in context. (seconds) (mph) (seconds) (mph)
b. Interpret the area of R2 in context. 2.2 30 1.9 30
Rates of change of 2.9 40 3.0 40
population 4.0 50 4.1 50
(people per month) 5.0 60 5.2 60
D(x) 6.5 70 6.8 70
8.0 80 8.6 80
9.9 90 10.7 90
R1 R2
11.8 100 13.3 100
I(x) (Source: Road and Track.)
x a. Find models for the speed of each car, given the
Month number of seconds after starting from 0 mph.
0 1 2 3 4 5
(Hint: Add the point (0, 0), and convert miles per
c. Find the change in population from the beginning hour into feet per second before modeling.)
of January through the end of April.
b. How much farther than a Porsche 911 Carerra
d. Explain why the answer to part c is not the sum does a Toyota Supra travel during the first 10 sec-
of the areas of the two regions. onds, assuming that both cars begin from a
32. o
Freign Trade The rate of change of the value of standing start?
goods exported from the United States between c. How much farther than a Porsche 911 does a
1990 and 2001 can be modeled as Toyota Supra travel between 5 seconds and 10
seconds of acceleration?
E(t) 1.665t 2 16.475t 7.632
billion dollars per year 34. Postal Service The table shows the approximate
rates of change of revenue for the U.S. Postal Service
t years after the end of 1990. Likewise, the rate of (USPS), Federal Express (FedEx), and United Parcel
change of the value of goods imported into the Service (UPS) from 1993 to 2001.
United States during those years can be modeled as
USPS FedEx UPS
I(t) 4.912t 40.861 billion dollars per year
(billions of (billions of (billions of
t years after 1990. dollars dollars dollars
(Source: Based on data from World Almanac and Book of Facts, Year per year) per year) per year)
ed. William A. McGeveran, Jr., New York: World Almanac 1993 3.4 0.3 1.0
Education Group, 2003.)
1994 3.3 0.6 1.2
a. Find the difference between the accumulated
value of imports and the accumulated value of 1995 3.0 1.0 1.3
exports from the end of 1990 through 2001. 1996 2.6 1.5 1.5
b. Is your answer from part a the same as the area of 1997 2.3 2.1 1.6
the region(s) between the graphs of E and I? 1998 2.0 2.3 1.8
Explain.
1999 1.8 2.1 1.9
33. Road Test The accompanying table shows the 2000 1.7 1.4 2.1
time it takes for a Toyota Supra and a Porsche 911
2001 1.8 0 2.2
(Source: Based on data from Hoover’s Online Guide.)
5.4 The Definite Integral 357
a. Using these data, find and graph equations for and that the number of pupae surviving the preda-
the rates of change of revenue for USPS and UPS. tory beetles each year during the same time period
b. Find and interpret the areas of the two regions can be modeled by the equation
bounded by the graphs in part a from 1993 p(t) 0.251 0.794 ln t hundred moths
through 2001.
per square meter of tree canopy
35. Postal Service Refer to the USPS, FedEx, and UPS
(Source: Adapted from P. J. denBoer and J. Reddingius,
data in the previous activity. Regulation and Stabilization Paradigms in Population Ecology,
a. Find and graph equations for the rates of change London: Chapman and Hall, 1996.)
of revenue for FedEx and UPS. In both equations, t is the number of years since 1960.
b. Find and interpret the areas of the three regions The area of the region below the graph of l and above
bounded by the graphs in part a between 1993 the graph of p is referred to as the accumulated
and 2001. density-dependent mortality of pupae by predatory
c. Find the definite integral of the difference of the beetles. Use an integral to estimate this value between
two equations in part a between 1993 and 2001. the years 1962 and 1965. Interpret your answer.
Interpret your answer.
37. Emissions In response to EPA regulations, a fac-
36. Mortality When modeling populations, biolo- tory that produces carbon emissions plants 22
gists consider many factors that affect mortality. hectares of forest in 1990. The trees absorb carbon
Some mortality factors (such as those that may be dioxide as they grow, thus reducing the carbon level
weather-related) are not dependent on the size of in the atmosphere. The EPA requires that the trees
the population; that is, the proportion of the popu- absorb as much carbon in 20 years as the factory
lation killed by such factors remains constant re- produces during that time. The trees absorb no car-
gardless of the size of the population. Other mortal- bon until they are 5 years old. Between 5 and 20
ity factors are dependent on the size of the years of age, the trees absorb carbon at the rates
population. In these cases, the proportion of a pop- indicated in the table.
ulation killed by a certain mortality factor increases
(or decreases) as the size of the population in- Tree age (years) 5 10 15 20
creases. Such factors are called density-dependent Carbon absorption
mortality factors. (tons per hectare 0.2 6.0 14.0 22.0
per year)
Varley and Gradwell studied the population size of
a winter moth in a wooded area between 1950 and (Source: Adapted from A. R. Ennos and S. E. R. Bailey,
1968. They found that predatory beetles represented Problem Solving in Environmental Biology, Harlow, Essex, England:
Longman House, 1995.)
the only density-dependent mortality factor in the
life cycle of the winter moth. When the population a. Find a model for the rate (in tons per year) at
of moths was small, the beetles ate few moths, which carbon is absorbed by the 22 hectares of
searching elsewhere for food, but when the popula- trees between 1990 and 2010.
tion was large, the beetles assembled in large clusters
b. The factory produced carbon at a constant rate
in the area of the moth population and laid eggs,
of 246 tons per year between 1990 and 1997. In
thus increasing the proportion of moths eaten by
1997, the factory made some equipment changes
the beetles.
that reduced the emissions to 190 tons per year.
Suppose the annual number of winter moth larvae Graph, together with the model in part a, the rate
in Varley and Gradwell’s study that survived winter of emissions produced by the factory between
kill and parasitism each year between 1961 and 1968 1990 and 2010. Find and label the time in which
can be modeled by the equation the absorption rate equals the production rate.
l(t) 0.0505 1.516 ln t hundred moths
per square meter of tree canopy
358 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
c. Label the regions of the graph in part b whose If f is the rate of change of the revenue of a small
areas correspond to the following quantities: business and g is the rate of change of the costs of the
i. The carbon emissions produced by the business x years after its establishment (both quan-
factory but not absorbed by the trees tities measured in thousands of dollars per year), in-
terpret the areas of the regions R1, R2, and R3 and the
ii. The carbon emissions produced by the fac-
[f(x g(x)]dx.
7
tory and absorbed by the trees value of the definite integral
1
iii. The carbon emissions absorbed by the trees
from sources other than the factory 39. How are the heights of rectangles (between two
d. Determine the values of the three quantities in curves) determined if one or both of the graphs lie
part c. below the horizontal axis? Consider the figure when
giving an example.
e. After 20 years, will the amount of carbon ab-
sorbed by the trees be at least as much as the (1, 3)
amount produced by the factory during that 3
f (x)
time period, as required by the EPA?
Discussing Concepts
38. Consider the regions between f and g depicted in the
x
figure. 1 4
(4, -1)
-1
f (x) g (x)
3
R3
2 -3
(4, -3)
1 R1 -4
R2 (1, -4)
x
0 1 2 3 A 4 5 6 B7
–1
–2 g(x)
–3
5.5 Average Value and Average Rate of Change
As a student, you are very familiar with grade averages and the method of calculating
averages by adding grades and dividing by the number of them. Grades are discrete
data. Let us consider a situation in which discrete averaging is not practical.
Concept Development: Averages of Continuous Functions Consider calcu-
lating a person’s average heart rate over 50 minutes of moderate activity. The actual
heart rate is calculated as the number of times the person’s heart beats during the
time period divided by the time. However, without medical monitoring devices, it is
impractical to count the number of heartbeats during a 50-minute period.
5.5 Average Value and Average Rate of Change 359
We could measure heart rate every 10 minutes and use these six data points to es-
timate the average over the entire 50-minute period. For example, by summing the
heart rates in Table 5.16 and dividing by 6, we estimate that a person’s heart rate over
the time period represented by the table is 100.8 beats per minute.
TABLE 5.16
Time into test Heart rate
(minutes) (beats per minute)
0 95
10 105
20 100
30 94
40 101
50 110
Obviously, if the heart rate were measured every 5 minutes instead of every 10
minutes, a more accurate estimate of the average heart rate could be obtained. Aver-
aging the data in Table 5.17 yields an estimate of 100.73 beats per minute.
TABLE 5.17
Time into test Heart rate Time into test Heart rate
(minutes) (beats per minute) (minutes) (beats per minute)
0 95 30 94
5 102 35 97
10 105 40 101
15 104 45 105
20 100 50 110
25 95
The preceding averages use a discrete number of heart rates, whereas heart rate is
constantly changing. In order to address this constantly changing situation, we use
the data to model heart rate with a continuous function.
H(t) (4.802 105)t 4 0.006t 3 0.229t 2
2.813t 94.371 beats per minute
where t is the number of minutes since the test began.
Integrating this function over the 50-minute interval gives a close approximation
of the total number of heart beats during that period:
50
H(t)dt 5067 beats
0
Dividing this total number of beats by 50 minutes yields the average heart rate of 101
beats per minute.
360 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Caution: Using a continuous model for a discrete situation must be done with care.
In this case, there are so many beats during the 50-minute interval (nearly 2 per sec-
ond) that it is reasonable to model the heart rate with a continuous function.
Averaging is a balancing out of extremes. Figure 5.58 depicts a graph of the heart
rate function. The average of that function’s outputs, approximately 101 beats per
minute, is shown as a dotted horizontal line on the graph.
H(t)
(beats per minute)
110
100.7
t
90
0 50 (minutes)
FIGURE 5.58
The average value of a function over an interval can be graphically interpreted as
the height (or signed height) of a rectangle whose area equals the area between the
function and the horizontal axis over the interval. (Note that we do not show the ver-
tical axis to zero in Figure 5.58; however, it is true that the area of the rectangle and
the area between the graph and the line representing the horizontal axis shown in the
figure are equal.)
Average Value of a Function
If y f(x) is a continuous function, then we can approximate the average value (or
average) of the function over an interval from x a to x b by dividing the inter-
val into n equally spaced subintervals, evaluating the function at a point in each
subinterval, summing the function values, and dividing by n.
f(x1) f(x2) … f(xn1) f(xn)
Average value
n
ba
We denote the length of each subinterval by x and calculate the length as x n .
Rewrite the average value estimate by multiplying the top and bottom terms by x.
Then make the substitution n x b a in the denominator.
[f(x1) f(x2) … f(xn1) f(xn)]x
Average value
n x
[f(x1) f(x2) … f(xn1) f(xn)]x
ba
5.5 Average Value and Average Rate of Change 361
As in the heart rate example, the estimate improves as the number of intervals
increases. Thus we obtain the exact average value by finding the limit of the estimate
as n approaches infinity:
… f(x ) f(x )]x
Average value lim [f(x1) f(x2) n1 n
n→
ba
which can be written as
b
f(x)dx
Average value a
ba
Thus we have
Average Value
If y f(x) is a smooth, continuous function from a to b, then the average
value of f(x) from a to b is
f(x)dx
b
Average value of
a
f(x) from a to b ba
EXAMPLE 1 Finding Average Value and Average Rate of Change
Temperature Suppose that the hourly temperatures shown in Table 5.18 were
recorded from 7 A.M. to 7 P.M. one day in September.
5.5.1
TABLE 5.18
Time Temperature (°F) Time Temperature (°F)
7 A.M. 49 2 P.M. 80
8 A.M. 54 3 P.M. 80
9 A.M. 58 4 P.M. 78
10 A.M. 66 5 P.M. 74
11 A.M. 72 6 P.M. 69
noon 76 7 P.M. 62
1 P.M. 79
a. Find a cubic model for this set of data.
b. Calculate the average temperature between 9 A.M. and 6 P.M.
c. Graph the equation together with the rectangle whose upper edge is determined
by the average value.
d. Calculate the average rate of change of temperature from 9 A.M. to 6 P.M.
362 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Solution
a. The temperature on this particular day can be modeled as
t(h) 0.03526h3 0.718h2 1.58h 13.69 degrees Fahrenheit
h hours after midnight. This model applies only from h 7 (7 A.M.) to
h 19 (7 P.M.).
b. The average temperature between 9 A.M. (h 9) and 6 P.M. (h 18) is
18
t(h)dh
Average temperature 9
74.4° F
18 9
c. Refer to Figure 5.59.
t (h)
(°F)
80
74.4
60
h
40 (hours after
9 18 midnight)
FIGURE 5.59
d. Recall that the average rate of change of a function on an interval is the change
in output divided by the change in input. The average rate of change of
temperature from 9 A.M. to 6 P.M. is
t(18)t(9)
Average rate of change of temperature 0.98 °F per hour ●
18 9
Average Rate of Change
The preceding example asked for an average rate of change. We know from Section
2.1 that the average rate of change of a continuous function y f(x) from x a to
f(b)f(a)
x b is calculated as b a . However, consider what happens when you have a
function that describes the rate of change of a quantity—that is, you have y f (x)—
and you need to find the average rate of change of the quantity f(x). In this case, you do
not use the average rate-of-change formula but should instead use an integral to find
the average value of the rate-of-change function. Note that we use the terms average rate
of change and average value of the rate of change interchangeably.
5.5 Average Value and Average Rate of Change 363
The Average Value of the Rate of Change
If y f (x) is a smooth, continuous rate-of-change function from a to b,
then the average value of f (x) from a to b is
Average value of the rate of
b
f (x) dx
a
change of f (x) from a to b ba
f(b)f(a)
ba
where f(x) is an antiderivative of f (x).
EXAMPLE 2 Determining Which Quantity to Average
Population Growth The growth rate of the population of South Carolina between
1790 and 2000 can be modeled* as
p(t) 0.18t 1.57 thousand people per year
where t is the number of years since 1790. The population of South Carolina in 1990
was 3486 thousand people.
a. What was the average rate of change in population from 1995 through 2000?
b. What was the average size of the population from 1995 through 2000?
Solution
a. The average rate of change in population between 1995 and 2000 is calculated
directly from the rate-of-change function as
210
p(t)dt(thousand people/year)(years)
205
35.8 thousand people per year
(210 205) years
b. In order to calculate the average population, we must have a function
for population. That is, we need an antiderivative of the rate-of-change
function:
p(t) p(t)dt
0.09t 2 1.57t C thousand people
We know that the population in 1990 was 3486 thousand people. Using this
fact, we solve for C, so the function for population is
p(t) 0.09t 2 1.57t 200 thousand people
where t is the number of years since 1790.
* Based on data from Statistical Abstract, 2001.
364 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Now we calculate the average population between 1995 and 2000 as
210
p(t)dt
205
3749 thousand people ●
210 205
Note in Example 2 that the average value of the population was found by integrat-
ing the population function, whereas the average rate of change was found by inte-
grating the population rate-of-change function. This example illustrates an important
principle:
In using integrals to find average values, integrate the function whose output
is the quantity you wish to average.
Note also that in part a of Example 2, we could have calculated the average rate of
change by using the population function and the formula
p(210)p(205) 178.9 thousand people
35.8 thousand people per year
210 205 5 years
Keeping units of measure attached to values can help you correctly calculate and label
average values.
We summarize this discussion as follows:
Average Values and Average Rates of Change
If y f(x) is a continuous or piecewise continuous function describing a
quantity from x a to x b, then the average value of the quantity from a
to b is calculated by using the quantity function and the formula
f(x)dx
b
Average value of f(x) a
ba
The average value has the same units as the output of the function f.
The average rate of change of the quantity, also called the average value
of the rate of change, can be calculated from the quantity function as
f(b)f(a)
Average rate of change
ba
or from the rate-of-change function as
f (x)dx
b
Average rate of change a
ba
The average rate of change has the same units as the rate of change of f.
5.5 Average Value and Average Rate of Change 365
EXAMPLE 3 Graphically Illustrating Average Value
Note that a(t) r(t) Carbon-14 Scientists estimate that 100 milligrams of the isotope 14C used in
carbon dating methods decays at a rate of
r(t) 0.0121(0.999879t) milligrams per year
where t is the number of years since the 100 milligrams of isotope began to decay. The
amount of the isotope that remains after t years of decay is
a(t) 100(0.999879t) milligrams
a. What is the average amount of the remaining isotope during the first 1000 years?
b. What is the average rate of decay during the first 1000 years?
c. Graphically illustrate the answers to parts a and b.
Solution
a. We calculate the average amount remaining during the first thousand years as
1000
a(t)dt
0
94.2 milligrams
1000 0
b. We find the average rate of decay during the first 1000 years as
1000
r(t)dt
0
0.0114 milligram per year
1000 0
In other words, the amount of 14C decreased by an average of 0.0114 milligram
per year during the first 1000 years. Note that this average rate of change can
also be calculated by using the amount function:
a(1000)a(0) 88.6 100 milligrams
0.0114 milligram per year
1000 0 1000 years
c. The average amount determines the top of the rectangle shown in Figure 5.60a.
The average rate of decay determines the bottom of the rectangle shown in Fig-
ure 5.60b. The average decay rate also can be graphically illustrated as the slope
of a secant line through two points on the amount function.
r (t)
a (t) (mg/year)
(mg) 0 1000 t
100 -0.0105
(years)
Average amount =
94.2 milligrams
94.2
-0.0114
Average rate of
t -0.012 change = -0.0114 mg/year
85
0 1000 (years)
(a) (b)
FIGURE 5.60 ●
366 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
a. Use a definite integral to approximate the aver-
5.5 Concept Inventory age annual value of U.S. factory sales of elec-
tronic goods to dealers from 1990 through 2001.
• Average value of a function b. Sketch the graph of sales from 1990 through
• Graphical illustration of average value 2001, and draw the horizontal line representing
• Average rate of change of a function the average value.
• Average value of a rate-of-change function 3. Phone aClls The most expensive rates (in dollars
per minute) for a 2-minute telephone call using a
long-distance carrier are listed in the accompanying
table.
5.5 Activities
Rate
Applying Concepts
Year (dollars per minute)
1. Traffic Speed The Highway Department is con- 1982 1.32
cerned about the high speed of traffic during the 1984 1.24
weekday afternoon rush hours from 4 P.M. to 7 P.M.
1985 1.14
on a newly widened stretch of interstate highway
that is just inside the city limits of a certain city. The 1986 1.01
Office of Traffic Studies has collected the data given 1987 0.83
in the table, which show typical weekday speeds 1988 0.77
during the 4 P.M. to 7 P.M. rush hours.
1989 0.65
1990 0.65
Time Speed (mph) Time Speed (mph)
1995 0.40
4:00 60 5:45 72.25
2000 0.20
4:15 61 6:00 74
4:30 62.5 6:15 74.5
a. Find a model for the data.
4:45 64 6:30 75
b. Use a definite integral to estimate the average of
5:00 66.25 6:45 74.25 the most expensive rates from 1982 through 1990.
5:15 67.5 7:00 73 c. Use a definite integral to estimate the average of
5:30 70 the most expensive rates from 1982 through 2000.
4. Ticket Price The table gives the price (in dollars)
a. Find a model for the data. of a round-trip flight from Denver to Chicago on a
b. Use the equation to approximate the average certain airline and the corresponding monthly profit
weekday rush-hour speed from 4 P.M. to 7 P.M. (in millions of dollars) for that airline for that route.
c. Use the equation to approximate the average
weekday rush-hour speed from 5 P.M. to 7 P.M. Ticket price Profit (millions
2. Electronics Sales U.S. factory sales of electronic (dollars) of dollars)
goods to dealers from 1990 through 2001 can be 200 3.08
modeled by the equation 250 3.52
Sales 0.0388x 3 0.495x 2 5.698x 43.6 300 3.76
billion dollars 350 3.82
where x is the number of years since 1990. 400 3.70
(Sources: Based on data from Statistical Abstract, 2001, and 450 3.38
Consumer Electronics Association.)
5.5 Average Value and Average Rate of Change 367
a. Find a model for the data. b. Use a definite integral to estimate the average
b. Determine the average profit for ticket prices number of accidents that occurred each year
from $325 to $450. from 1976 through 1997.
c. Determine the average rate of change of profit c. Graphically illustrate the answers to parts a and b.
when the ticket price rises from $325 to $450. .8 Temperature During a certain summer thunder-
d. Graphically illustrate the answers to parts b and c. storm, the temperature drops and then rises again.
The rate of change of the temperature during the
5. Population The population of Mexico between hour and a half after the storm began is given by
1921 and 2000 is given by the model
T(h) 9.48h3 15.49h2 17.38h 9.87 °F per hour
Population 7.567(1.02639t) million people
where h is the number of hours since the storm
where t is number of years since the end of 1900. began.
(Source: Based on data from www.inegi.gob.mx. Accessed
9/20/02.) a. Calculate the average rate of change of tempera-
ture from 0 to 1.5 hours after the storm began.
a. What was the average population of Mexico from
the beginning of 1990 through the end of 1999? b. If the temperature was 85°F at the time the storm
began, find the average temperature during the
b. In what year was the population of Mexico equal
first 1.5 hours of the storm.
to its 1990s average?
.9 Road Test The acceleration of a race car during
c. What was the average rate of change of the pop-
the first 35 seconds of a road test is modeled by
ulation of Mexico during the 1990s?
a(t) 0.024t 2 1.72t 22.58 ft/sec2
6. Veggies The per capita utilization of commer-
cially produced fresh vegetables in the United States where t is the number of seconds since the test
from 1980 through 2000 can be modeled by began. Assume that velocity and distance were both
0 at the beginning of the road test.
v(t) 0.092t 2 0.720t 149.554 pounds per person
a. Calculate the average acceleration during the
where t is the number of years after 1980. first 35 seconds of the road test.
(Sources: Based on data from Statistical Abstract, 2001, and
b. Calculate the average velocity during the first
www.ers.usda.gov. Accessed 9/25/02.)
35 seconds of the road test.
a. Use a definite integral to estimate the average per
c. Calculate the distance traveled during the first
capita utilization of commercially produced
35 seconds of the road test.
fresh vegetables in the United States between
1980 and 2000. d. If the car had been traveling at its average veloc-
ity throughout the 35 seconds, how far would the
b. Find the average rate of change in per capita
car have traveled during that 35 seconds?
utilization between 1980 and 2000.
e. Graphically illustrate the answers to parts a and
c. In which year was the per capita utilization clos-
b. Explain how the answer to part c is related to
est to the average per capita utilization between
the graphical illustration of the part b answer.
1980 and 2000?
10. O
il Production On the basis of data obtained
.7 A
ir A
ccidents The number of general-aviation from a preliminary report by a geological survey
aircraft accidents from 1975 through 1997 can be team, it is estimated that for the first 10 years of pro-
modeled by duction, a certain oil well in Texas can be expected
a(x) 100.6118x 3967.5572 accidents to produce oil at the rate of
where x is the number of years since 1975. r(t) 3.93546t 3.55e1.35135t thousand barrels per year
(Source: Based on data from Statistical Abstract, 1994 and 1998.) t years after production begins. Estimate the aver-
a. Calculate the average rate of change in the yearly age annual yield from this oil well during the first
number of accidents from 1976 through 1997. 10 years of production.
368 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
11. Velocity An article in the May 23, 1996, issue of c. Assuming that diastolic blood pressure was 95
Nature addresses the interest some physicists have mm Hg at 12 P.M., estimate the average diastolic
in studying cracks in order to answer the question blood pressure between 8 A.M. and 8 P.M.
“How fast do things break, and why?” Data given in
the table are estimated from a graph in this article
Diastolic BP
showing velocity of a crack during a 60-microsec- Time (mm Hg per hour)
ond experiment.
8 A.M. 3.0
10 A.M. 1.8
Time Velocity
(microseconds) (meters per second) noon 0.7
10 148.2 2 P.M. 0.1
20 159.3 4 P.M. 0.7
30 169.5 6 P.M. 1.1
40 180.7 8 P.M. 1.3
50 189.8 10 P.M. 1.1
60 200.0 midnight 0.7
2 A.M. 0.1
a. Find a model for the data. 4 A.M. 0.8
b. Determine the average speed at which a crack 6 A.M. 1.9
travels between 10 and 60 microseconds.
12. Newspapers The circulation (as of September 20 14. iAr p
S eed The air speed of a small airplane dur-
of each year) of daily English-language newspapers ing the first 25 seconds of takeoff and flight can be
in the United States between 1986 and 2000 can be modeled by
modeled as
v(t) 940,602t 2 19,269.3t 0.3 mph
n(x) 0.00792x 3 0.32x 2 3.457x 51.588
million newspapers t hours after takeoff.
a. Find the average air speed during the first
where x is the number of years since 1980. 25 seconds of takeoff and flight.
(Source: Based on data from Statistical Abstract, 1995 and
2001.) b. Find the average acceleration during the first
25 seconds of takeoff and flight.
a. Estimate the average newspaper circulation from
1986 through 2000. 15. Sw
im Time The rate of change of the winning
b. In what year was the newspaper circulation clos- times for the 100-meter butterfly swimming com-
est to the average circulation from 1986 through petition at selected Summer Olympic Games be-
2000? tween 1956 and 2000 can be described by w(t)
0.0106t 1.148 seconds per year, where t is the
c. Graphically illustrate the answer to part a.
number of years after 1900. Find the average rate of
13. Blood Pressure Blood pressure varies for individ- change of the winning times for the competition
uals throughout the course of a day, typically being from 1956 through 2000.
lowest at night and highest from late morning to (Source: Based on data from Statistical Abstract, 2001.)
early afternoon. The estimated rate of change in di-
16. Emissions The federal government sets stan-
astolic blood pressure for a patient with untreated
dards for toxic substances in the air. Often these
hypertension is shown in the table.
standards are stated in the form of average pollu-
a. Find a model for the data. tant levels over a period of time on the basis of the
b. Estimate the average rate of change in diastolic reasoning that exposure to high levels of toxic sub-
blood pressure from 8 A.M. to 8 P.M. stances is harmful, but prolonged exposure to
5.5 Average Value and Average Rate of Change 369
moderate levels is equally harmful. For example, 10 P.M. The city issues air quality warnings based
carbon monoxide (CO) levels may not exceed 35 on the daily average CO concentration of the
ppm (parts per million) at any time, but they also previous day between 6 A.M. and 10 P.M. The
must not exceed 9 ppm averaged over any 8-hour warnings are as follows:
period.
(Source: Douglas J. Crawford-Brown, Theoretical and
Mathematical Foundations of Human Health Risk Analysis, Average
Boston: Kluwer Academic Publishers, 1997.) concentration Warning
The concentration of carbon monoxide in the air in 0 CO 9 None
a certain metropolitan area is measured and can be 9 CO 12 Moderate pollution. People
modeled as with asthma and other
respiratory problems should
c(h) 0.004h4 0.05h3 0.27h2 2.05h 3.1 ppm remain indoors if possible.
h hours after 7 A.M. 12 CO 16 Serious pollution. Ban on all
single-passenger vehicles.
a. Did the city exceed the 35-ppm maximum in the
Everyone is encouraged to
8 hours between 7 A.M. and 3 P.M.? stay indoors.
b. Did the city exceed the 9-ppm maximum average CO 16 Severe pollution. Mandatory
between 7 A.M. and 3 P.M.? school and business closures.
17. Emissions Refer to the discussion in Activity 16.
The following table shows measured concentrations d. Judging on the basis of the data in the table and
of carbon monoxide in the air of a city on a certain your answer to part c, which warning do you
day between 6 A.M. and 10 P.M. believe should be posted?
18. Population Aurora, Nevada, was a mining boom
Time CO concentration town in the 1860s and 1870s. Its population can be
(hours since 6 A.M.) (ppm) modeled by the function
0 3
2 12
7.91t3 120.96t2 193.92t
p(t) 123.21 people when 0.7 t 13
4 22
45,544(0.8474t) people when 13 t 55
6 18
8 16 with rate-of-change function
10 20
12 28 23.73t 2 241.92t 193.92
14 16 people per year when 0.7 t 13
p(t)
16 6
7541.287(0.8474 )
t
people per year when 13 t 55
a. Consider a scatter plot of the data and deter- In both functions, t is the number of years since 1860.
mine (by examination of the data) over which 8- (Source: Based on data from Don Ashbaugh, Nevada’s
hour period the average CO concentration was Turbulent Yesterday: A Study in Ghost Towns, Los Angeles:
greatest. Westernlore Press, 1963.)
b. Model the data. Use the equation to calculate the a. What was the average population of Aurora
average CO concentration during the 8-hour between 1861 and 1871? between 1871 and 1881?
period determined in part a. b. Demonstrate two methods for calculating the
c. Use the equation in part b to estimate the average average rate of change of the population of
CO concentration in this city between 6 A.M. and Aurora between 1861 and 1871.
370 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Discussing Concepts vertical axis does not extend all the way to zero.
Instead, the vertical axis is shown above (or below)
19. We know that the area of the region between the a line y k. However, on the interval from x a
graph of a function and the horizontal axis from to x b, the area of the region between the
x a to x b is equal to the area of the rectangle graph and the line y k and the area of the rectan-
whose height is the average value of the function gle between the average value and the line y k are
from x a to x b and whose width is b a. The equal. Explain, illustrating with graphs, why this
graphs presented in this section show only a portion is true.
of the vertical axis. That is, in each graph, the
5.6 Integration by Substitution or
Algebraic Manipulation
We have thus far examined situations in which it is useful to determine accumulated
change in a quantity by finding or estimating the area of a region between the graph
of a rate-of-change function and the horizontal axis. We estimated such areas by
summing areas of rectangles until we discovered that the Fundamental Theorem of
Calculus enables us to find areas using antiderivatives. Although there are limitations
on our ability to find antiderivatives for functions, we explore two techniques that
allow us to extend our ability to find antiderivatives.
The first technique is to use algebra to change the integrand into one that can
be evaluated using the formulas from Section 5.4. We illustrate antiderivatives that
involve simple products or quotients in Example 1.
EXAMPLE 1 Finding Antiderivatives That Involve Products or
Quotients of Functions
Determine whether the following general antiderivatives involving products or quo-
tients can be found using the techniques presented in this book. If so, find the anti-
derivative.
a. xx dx b. 2xx 1 dx
2
e (e 4x 2x 2x dx
3
c. 3x 4x 4)dx d. 2
Solution
a. Rewrite xx as x(x1/2) x 3/2. Then apply the Power Rule for antiderivatives
2x 5/2
to obtain xx dx 5 C.
5.6 Integration by Substitution or Algebraic Manipulation 371
b. Unlike the expression in part a, it is not possible to rewrite 2xx2 1 in such
a way that we can easily find its antiderivative.
c. Multiply the two factors to rewrite as e7x 4e3x and apply antiderivative rules
to obtain
e (e
3x 4x 4)dx (e7x 4e3x)dx
e7x
7
4e3x
3
C
4x3 2x 4x3 2x 1
d. Rewrite 2 as 2 2 2x x . Then apply the Sum Rule, the
2x 2x 2x
Power Rule, and the Natural Log Rule for antiderivatives to obtain
4x3 2x
2
2x
dx (2x 1
x ) dx x ln x C .
2 ●
In part b of Example 1, we could not find 2xx2 1 dx using the antiderivative
rules already presented in this text. The expression 2xx2 1 involves function com-
position and function multiplication. The composite factor of the expression is
x2 1 . If we let f (x) x2 1 and g(u) u, we can write x2 1 as
f (x) g(f (x)). The factor 2x is important because f(x) 2x. Changing the order
of the factors in the expression and substituting f and g where appropriate, we get
2xx2 1 (x2 1) 2x g(f (x)) f(x).
The expression g(f (x)) f(x) should look familiar to you. Recall that for a com-
posite function G(f (x)) the derivative function is found using the Chain Rule:
d
G(f (x)) g(f (x)) f (x) where g G
dx
Thus, (x2 1) 2x is in the form of the right side of the Chain Rule. Applying
the Fundamental Theorem of Calculus to the Chain Rule (as stated here) yields
g(f (x)) f (x)dx G(f (x)) C
We shorten this notation by letting u f (x) and substituting du for f (x)dx to
obtain g(u)du G(u) C. This method is known as Integration by Substitution .
Integration by Substitution
Given g(f (x)) f (x)dx, let u f(x) and du f (x)dx.
If G is an antiderivative of g, then
g(f (x)) f (x)dx g(u)du G(u) C G(f (x)) C
372 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Applying Integration by Substitution on 2xx2 1 dx yields
2xx 1 dx (x 1) 2x dx
2 2
u du u du where u x 1 and du 2x dx
1/2 2
2 3/2
u C
3
2
(x2 1)3/2 C
3
The next example illustrates Integration by Substitution for other composite functions.
EXAMPLE 2 Using Substitution to Find Antiderivatives That Involve
Composite Functions
Use Integration by Substitution to evaluate the following integrals.
6xe dx 6xe
2
a. 3x2 7 b. 3x2 7 dx
0
c. (3x 7) dx 2
Solution
a. Rewrite the expression with the factor involving composition to the left:
6xe 3x2 7 dx (e 3x2 7) 6x dx
Let u 3x2 7 so that du 6x dx, and rewrite the integral as eu du.
Use the Exponent Rule for antiderivatives to obtain eu du eu C. Finally,
replace u with 3x2 7 so that
6xe 3x2 7 dx e3x 7 C
2
6xe
2
b. To determine 3x2 7 dx, evaluate the antiderivative from part a at 0 and 2.
0
6xe
2
dx e3x 7 e5 e 7
3x2 7 2 2
0
0
c. For the integral (3x 7)2 dx, let u 3x 7. The derivative of 3x 7
1
with respect to x is 3, so du 3dx. Multiply the integral by 3 3 to obtain the
proper form.
5.6 Integration by Substitution or Algebraic Manipulation 373
(3x 7) dx 3 (3x 7) dx (3x 7) 3dx
2 1
3
2 1
3
2
u du
1
where u 3x 7
3
2
1 1 1
3 (3 u3) C 9 u3 C
1
9 (3x 7)3 C
1
Thus (3x 7)2 dx 9 (3x 7)3 C. ●
When finding antiderivatives, it is always a good idea to take the derivative of your
answer in order to determine whether it is correct. Finding antiderivatives is one pro-
cedure for which there is a simple way to check the answer.
In Activities 9 through 20, find the exact value of the
5.6 Concept Inventory integral by using antiderivative formulas if possible. If
this is not possible, use technology to estimate the an-
• Algebraic manipulation before integration swer. In either case, state whether your answer is exact or
• Integration by substitution an approximation.
ln xdx x ln xdx
4 4
.9 10.
1 1
11. 12.
5.6 Activities ln x 5 5
5(ln x) 4
dx dx
x 2 2 x
In Activities 1 through 8, find the general antiderivative
13. 2x ln(x 1)dx 14. ln(x 1)dx
if possible. 2 2
2 2
1. 2e dx 2. 2xe dx
1 1
2x x2
15. 16.
2x 4 4 1
dx dx
x 1 x 1
3. x e dx 4. 3(ln 2)2 (1 2 ) dx
2 2
3 3
2 x2 x x 3
17. 18.
2x 6 2 x 1
4 2
dx dx
x 1
5. (1 e ) e dx 6. 1 e e dx
2 2x
1 3
x 2 x x x
19. 20. xx 1 dx
x 1 4 2 1
2
dx
.7 .8
x 2
2 x 5e x 3 0
dx dx
2 2
x e 2
x
SUMMARY
Approximating Results of Change lation is the signed area of a region between the graph of
The accumulated results of change are best understood a negative rate-of-change function and the horizontal
in geometric terms: Positive accumulation is the area of axis. We can approximate the areas of regions of interest
a region between the graph of a positive rate-of-change by summing areas of rectangular regions.
function and the horizontal axis, and negative accumu-
374 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
Limits of Sums and Accumulation Functions The Fundamental Theorem enables us to find accumu-
The area of a region between the graph of a continuous, lation function formulas by finding antiderivatives.
non-negative function f and the horizontal axis from a
to b is given by a limit of sums: The Definite Integral
The Fundamental Theorem of Calculus enabled us to
Area lim [f(x1) f(x2) … f(xn )]x
n→ show that when f is a smooth, continuous function, the
Here, the points x1, x2, … , xn are the midpoints of n
f(x)dx can be evaluated by
b
ba definite integral
rectangles of width x n between a and b. a
More generally, we consider the limit applied to an
f(x)dx F(b) F(a)
b
arbitrary continuous bounded function f over the inter-
a
val from a to b and call this limit the definite integral of
f from a to b. In symbols, we write where F is any antiderivative of f.
The Fundamental Theorem of Calculus ensures that
f(x)dx lim [f(x ) f(x ) … f(x )]x
b
each continuous bounded function does indeed have an
n→ 1 2 n
a antiderivative. Thus, to the extent that we can actually
An accumulation function is an integral of the form obtain an algebraic expression for an antiderivative, we
f(t)dt, where the upper limit x is a variable. This func-
x can easily evaluate a definite integral. In situations where
an antiderivative cannot be found, we use one of the ap-
a
proximation techniques discussed in Section 5.1 allow-
tion gives us a formula for calculating accumulated
ing technology to perform the calculations.
change in a quantity.
To compute the area between two curves, we used the
The Fundamental Theorem of Calculus fact that if the graph of f lies above the graph of g from a
[f(x) g(x)]dx is the area of the
b
The Fundamental Theorem sets forth the fundamental to b, then the integral
a
connection between the two main concepts of calculus,
region between the two graphs from a to b.
the derivative and the integral. It tells us that for any
Note that if the two functions intersect between a
continuous function f,
and b, then the difference between the accumulated
f(t)dt f(x)
d x changes of the functions is not the same as the total
dx a
area of the regions between the two rate-of-change
curves.
In other words, the derivative of an accumulation func-
tion of y f(t) is precisely y f(x). Average Values and Average Rates of Change
If we reverse the order of these two processes and
begin by differentiating first, then we obtain the starting We use definite integrals to calculate the average value of
function plus a constant. a continuous function for a quantity:
f(t)dt f(x) C Average value of
x b
f(x)dx
a a
f(x) from a to b ba
A function F is an antiderivative of f if F(x) f(x).
f(t)dt is y f(x), we When we are given a rate-of-change function y f (t),
x
Because the derivative of y the average rate of change of f(t) from t a to t b is
a
f(t)dt is an antiderivative of y f(x). Average rate of change
x b
see that y f (t)dt
a
Each continuous bounded function has infinitely many a
of f(x) from a to b ba
antiderivatives, but any two differ by only a constant.
Chapter 5 Concept Review 375
CONCEPT CHECK
Can you To practice, try
• Interpret accumulated change and area? Section 5.1 Activity 3
• Approximate areas using rectangles? Section 5.1 Activities 11, 13
• Interpret definite integrals? Section 5.1 Activities 3, 5
• Approximate area using a limiting value? Section 5.1 Activities 9, 11
• Sketch and interpret accumulation functions? Section 5.2 Activities 3, 11
• Recover the units of a quantity function? Section 5.2 Activity 21
• Find general antiderivatives? Section 5.3 Activities 11, 15, 17
• Find and interpret specific antiderivatives? Section 5.3 Activities 23, 25, 27
• Recover a function from its rate-of-change equation? Section 5.3 Activity 29
• Use the Fundamental Theorem to evaluate definite integrals? Section 5.4 Activities 13, 17
• Find and interpret areas between two curves? Section 5.4 Activity 33
• Find average value and average rate of change? Section 5.5 Activity 7
• Determine whether an approximation technique is Section 5.6 Activities 13, 15
necessary in order to estimate the value of a definite
integral?
CONCEPT REVIEW
1. iOl lFow The rate at which crude oil flows through b. Use five midpoint rectangles to estimate the
a pipe into a holding tank can be modeled by area of the region between the graph of r and
the t-axis from 0 to 25 minutes. Sketch the
r(t) 10(3.2t 2 93.3t 50.7) ft3/minute rectangles on the graph you drew in part a.
where t is the number of minutes the oil has been c. Interpret your answer to part b.
flowing into the tank.
2. p
S eed A hurricane is 300 miles off the east coast of
a. Sketch a graph of r for t between 0 and 25 minutes. Florida at 1 A.M. The speed at which the hurricane is
376 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
moving toward Florida is measured each hour. e. What does the graph in part d represent?
Speeds between 1 A.M. and 5 A.M. are recorded in the
4. iOl lFow Consider again the model for the flow
table.
rate of crude oil into a holding tank:
Time Speed (mph) Time Speed (mph) r(t) 10(3.2t 2 93.3t 50.7)ft3/minute
1 A.M. 15 4 A.M. 38 after t minutes.
2 A.M. 25 5 A.M. 40 a. If the holding tank contains 5000 ft3 of oil when
3 A.M. 35 t 0, find a model for the amount of oil in the
tank after t minutes.
a. Find a model for the data. b. Use your model in part a to find how much oil
b. Use a limiting value of sums of areas of midpoint flowed into the tank during the first 10 minutes.
rectangles to estimate how far (to the nearest tenth c. If the capacity of the tank is 150,000 ft3, according
of a mile) the hurricane traveled between 1 A.M. to the model, how long can the oil flow into the
and 5 A.M. Begin with five rectangles, doubling the tank before the tank is full?
number each time until you are confident that 5. Investment Ten thousand dollars invested in a
you know the limiting value. mutual fund is growing at a rate of
3. iDeting The accompanying graph depicts the rate a(x) 840(1.08763x) dollars per year
of change in the weight of someone who diets for 20
weeks. x years after it was invested.
2.75
a. What does the area of the shaded region beneath
a. Determine the value of a(x)dx.
the horizontal axis represent? 0
b. What does the area of the shaded region above b. Interpret your answer to part a.
the horizontal axis represent? c. What is the average rate of growth of the invest-
Rate of change in weight ment from x 0 to x 2.75?
(pounds per week)
6. Earnings Based on data provided by the U.S.
3 Bureau of the Census for the years 1980 through
Area =
1988, the full-time average annual earnings of men
15.4 and women in the United States can be modeled by
the following equations:
10 20
0 Weeks Men: m(t) 0.0625t 2 10.38t
Area = 26.7 30
466.8075 thousand dollars
-2 Women: w(t) 0.03125t 2 5.695t
234.89875 thousand dollars
c. Is this person’s weight at 30 weeks more or less
than it was at 0 weeks? How much more or less? where t is the number of years since 1900. From the
d. If w is the function shown in the figure, sketch a beginning of 1980 and through the end of 1988, by
w(t)dt. Label units and values
x how much did the 9-year earnings of a man who
graph of W(t) earned the average wage exceed those of a woman
0
on both axes of your graph. who earned the average wage?
Project 5.1 Acceleration, Velocity, and Distance
Setting Tasks
According to tests conducted by Road and Track, a 1. Convert the speed data to feet per second, and find
1993 Toyota Supra Turbo accelerates from 0 to 30 a quadratic model for velocity (in feet per second)
mph in 2.2 seconds and travels 1.4 miles (1320 feet) as a function of time (in seconds). Discuss how
in 13.5 seconds, reaching a speed of 107 mph. Road close your model comes to predicting the 107 mph
and Track reported the data given in the table. reached after 13.5 seconds.
2. Add the data point for 13.5 seconds, and find a
Time Speed reached from rest
(seconds) (mph)
quadratic model for velocity.
0 0 3. Use four rectangles and your model from Task 2 to
2.2 30
estimate the distance traveled during acceleration
from rest to a speed of 50 mph and the distance
2.9 40 traveled during acceleration from a speed of 50
4.0 50 mph to a speed of 100 mph. Repeat the estimate
5.0 60 using twice as many rectangles.
6.5 70 4. Use nine rectangles to approximate the distance
8.0 80 traveled during the first 13.5 seconds. How close is
9.0 90
your estimate to the reported value?
11.8 100 5. Find the distances traveled during
a. Acceleration from rest to a speed of 50 mph
b. Acceleration from a speed of 50 mph to a speed
of 100 mph
c. The first 13.5 seconds of acceleration
Compare these answers to your estimates in Tasks 3
and 4. Explain how estimating with areas of
rectangles is related to calculating the definite
integral.
Reporting
Prepare a written report of your work. Include scatter
plots, models, graphs, and discussions of each of the
above tasks.
377
Project 5.2 Estimating Growth
Setting Tasks
A table based on data from the Berkeley Growth 1. Use the data and right rectangles to approximate
Study is located on the Calculus Concepts website. the height of a typical 18-year-old male.
This table lists the rate of growth of a typical male
2. Sketch a smooth, continuous curve over a scatter
from birth to 18 years.
plot of the data. Find a piecewise model for the
data. Use no more than three pieces.
3. Use your piecewise model and limits of sums to
approximate the height of a typical 18-year-old
male. Convert centimeters to feet and inches, and
compare your answer to the estimate you obtained
using right rectangles. Which is likely to be the
more accurate approximation? Why?
4. Use your piecewise model and what you know
about definite integrals to find the height of a
typical 18-year-old male in feet and inches.
Compare your answer with the better of the
approximations you obtained in Task 3.
5. Randomly choose ten 18-year-old male students,
and determine their heights. (Include your data—
names are not necessary, only the heights.) Discuss
your selection process and why you feel that it is
random. Find the average height of the 18-year-
old males in your sample. Compare this average
height with your answer to Task 4. Discuss your
results.
6. Refer to your sketch of the rate-of-growth graph in
Task 2, and draw a possible graph of the height of
a typical 18-year-old male from birth to age 18.
Reporting
Prepare a report that presents your findings in Tasks
1 through 6. Explain the different methods that you
used, and discuss why these methods should all give
similar results. Attach your mathematical work as an
appendix to your report.
378
Analyzing Accumulated
Change: Integrals
6
in Action
Concepts Outline
6.1 Perpetual Accumulation and
Improper Integrals
6.2 Streams in Business and
Biology
6.3 Integrals in Economics
6.4 Probability Distributions and
Density Functions
Darama/CORBIS
Concept Application
The CEO of a large corporation must concern himself or herself with many facets of the
economy and the corporation’s relationship to it. For example, the CEO may be inter-
ested in answering questions as diverse as
• What is the 5-year future value of an income stream that the corporation invests?
• What is the market equilibrium price for a particular product produced by the corpo-
ration?
• What is the probability that an entry-level employee will continue with the company
at least 5 years?
In this chapter you will learn how to answer questions such as these.
379
380 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
Chapter Introduction
Chapter 5 established that the accumulated results of change are limiting values of ap-
proximating sums known as definite integrals. Magnitudes of accumulated change
can be expressed as areas of regions between the graph of a rate-of-change function
and the horizontal axis. The Fundamental Theorem of Calculus provides a simple
method for evaluating definite integrals using antiderivatives.
In Chapter 6 we present several applications of integration. We use integrals to
calculate perpetual accumulation, present and future values of income streams, and
future values of biological streams. We discuss how integrals can be applied to eco-
nomics topics and used to calculate economic quantities of interest to consumers and
producers. We conclude by using integrals to calculate probabilities.
Concepts You Will Be Learning
• Evaluating and interpreting improper integrals (6.1)
• Recognizing when an improper integral diverges (6.1)
• Calculating and interpreting present and future values for income streams (6.2)
• Calculating and interpreting the following economic quantities: consumers’
willingness and ability to spend, expenditure, and surplus; suppliers’ willingness
and ability to receive, revenue, and surplus; market equilibrium; total social
gain (6.3)
• Calculating and interpreting elasticity (6.3)
• Calculating and interpreting probabilities, means, and standard deviations (6.4)
• Using probability and cumulative density functions (6.4)
6.1 Perpetual Accumulation and Improper Integrals
Definite integrals have specific numbers for both the upper limit and the lower limit.
We now consider what happens to the accumulation of change when one or both of
the limits of the integral are infinite. That is, we wish to evaluate integrals of the form
b
f (x)dx , f (x)dx , or f (x)dx . We call integrals of this form improper
a
integrals. Improper integrals play a role in economics and statistics, as well as in other
fields of study.
Evaluating Improper Integrals
Consider evaluating the improper integral 2
4.3e0.06x dx. We can interpret this in-
tegral as the area of the region between the graph of y 4.3e0.06x and the x-axis from
2 to infinity. See Figure 6.1. One way to estimate this area is to consider the area be-
tween 2 and some large value. In Table 6.1 we show several area calculations for in-
creasingly large values.
6.1 Perpetual Accumulation and Improper Integrals 381
TABLE 6.1
y = 4.3e-0.06x
N
4.3e 0.06x dx
2
50 59.994558
100 63.384987
200 63.562191
400 63.562631
2 x ∞ 800 63.562631
FIGURE 6.1 1600 63.562631
Limit 63.56263
Note that the area between 2 and 400 is not very different from the area between 2 and
1600. The difference is smaller than can be shown by the technology that we used.
However, the limiting value seen in the table is still just an estimate of the
value of the integral 2
4.3e0.06x dx. You should recognize that we are numerically
N
investigating a limit in Table 6.1. We are investigating lim 4.3e0.06x dx . We can
N→ 2
calculate this limit algebraically in order to obtain the exact answer. Begin by finding
the general antiderivative of 4.3e0.06x, evaluating it at 2 and N, and subtracting the
results.
4.3e
N 4.3 0.06x N 4.3 0.06N 4.3 0.06(2)
0.06xdx e e e
2 0.06 2 0.06 0.06
4.3 0.06N 4.3 0.12
e e
0.06 0.06
An antiderivative must be found Next find the limit of this expression as N becomes infinitely large.
prior to using the limit process.
0.06
4.3 0.06N 4.3 0.12
lim e e
N→ 0.03
4.3 0.06N 4.3 0.12
lim e lim e
N→ 0.06 N→ 0.06
The first term is a decreasing exponential function, so we know that as N approaches
infinity, the value of this term approaches zero. The second term is a constant func-
tion and is not affected by the value of N.
4.3 0.06N 4.3 0.12
lim e lim e
N→ 0.06 N→ 0.06
4.3 0.12
0 e
0.06
4.3 0.12
e
0.06
We see that this answer confirms our former numerical estimate, because
4.3 0.12
e 63.56263
0.06
382 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
4.3
The answer 0.06 e0.12 is exact, whereas the answer 63.56263 is not. Although an an-
swer accurate to the fifth decimal place is sufficient for most applications, there are
situations in which greater precision is necessary.
To summarize, an improper integral f(x)dx is evaluated by replacing infinity
a
with a variable, say N, and evaluating the limit of the integral f(x)dx as N
N
approaches infinity. That is, provided the limits exist, a
f(x)dx lim f(x)dx lim F(N) F(a)
N
a N→ a N→
f(x)dx lim f(x)dx F(b) lim F(N)
b b
N→ N N→
where F is an antiderivative of f. We now have the tools we need to apply improper
integrals to some real-world problems.
EXAMPLE 1 Using a Limit to Evaluate an Improper Integral
Decay Carbon-14 dating methods are sometimes used by archeologists to deter-
mine the age of an artifact. The rate at which 100 milligrams of 14C is decaying can be
modeled by
r(t) 0.01209(0.999879t) milligrams per year
where t is the number of years since the 100 milligrams began to decay.
a. How much of the 14C will have decayed after 1000 years?
b. How much of the 14C will eventually decay?
Solution
a. The amount of 14C to decay during the first 1000 years is
1000 1000
r(t)dt 0.01209(0.999879t)dt 11.4 milligrams
0 0
Approximately 11.4 milligrams will decay during the first 1000 years. Note that
11.4 milligrams is the signed area of the shaded region in Figure 6.2.
r(t)
(milligrams
per year)
0.01
250 500 750 1000 t
Years
-0.01
FIGURE 6.2
6.1 Perpetual Accumulation and Improper Integrals 383
b. In the long run, the amount that will decay is
r(t)dt lim
N
0.01209(0.999879t)dt
0 N→ 0
0.01209(0.999879
) t N
lim
N→ ln 0.0999879 0
0.01209(0.999879
) 0.01209(0.999879 )
N 0 N
lim
N→ ln 0.0999879 ln 0.0999879 0
99.91131 lim (0.999879N ) 99.91131
N→
99.91131(0) 99.91131 100 milligrams
r(t)
Eventually all of the 14C will decay. In terms of the graph shown in Figure 6.3,
(milligrams
per year) the area of the region between the graph of the function r and the horizontal
axis gets closer and closer to 99.91131 as t gets larger and larger. Because the
0.01
parameters in the equation are rounded, the area is getting closer and closer
10,000 20,000 t to 99.91131 rather than to 100, which is the amount that must ultimately
Years decay. ●
-0.01
Divergence
FIGURE 6.3
It is possible that when we are evaluating an improper integral, the limit does not ex-
ist. (For example, as we numerically approximate the limit, the limit estimates be-
come increasingly large.) In this case, we say that the improper integral diverges.
Example 2 illustrates this situation.
EXAMPLE 2 Recognizing That an Integral Diverges
If possible, determine the value of 1xdx.
1
Solution
We begin by replacing with the variable N and finding the limit as N → .
1 N
f (x) = ln x
dx lim ln x
1 x N→ 1
lim (ln N ln 1)
N→
x lim ln N lim ln 1
N→ N→
1
To evaluate these limits, recall the shape of the graph of y ln x (see Figure 6.4). We
see from the graph that ln 1 0 and that as the input becomes increasingly large,
the output also becomes increasingly large. Thus, the limiting value does not exist.
Because lim ln N → ,
N→
1
1 x
dx → and we say that this improper integral
FIGURE 6.4 diverges. ●
384 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
which 100 milligrams of 238U is decaying can be
6.1 Concept Inventory modeled by
r(t) 1.55(0.9999999845t )106 milligrams per year
• Improper integrals
• Divergence where t is the number of years since the 100 mil-
ligrams began decaying.
a. How much of the 238U will decay during the first
100 years? during the first 1000 years?
b. How much of the 238U will eventually decay?
6.1 Activities
The following information is used in Activities 14 and 15:
Getting Started In the study of markets, economists define consumers’
For Activities 1 through 11, evaluate the indicated im- willingness and ability to spend as the maximum amount
proper integral. that consumers are willing and able to spend for a spe-
3e dt 2. 5(0.36 )dt
cific quantity of goods or services. If some consumers
1. 0.2t t
will purchase a product or service regardless of its price,
0 15
then the consumers’ willingness and ability to spend is
3. 3x dx 4. 7e dx
3
2 7x defined by
D(p)dp
10
C qp0
5. 4x dx 6.
10 1
3 p0
dx
x 2
where q is a specific quantity, p0 is the price associated
7. 9.6x 8. 1 dx
2
0.432 3 with quantity q, and D(p) is the demand for the com-
dx 2
0.36 x modity when the price is p. (Section 6.3 discusses these
9. 10. [5(0.36 ) 5]dx
2x
x concepts in greater detail.)
dx
x 1
2
2
5
14. Demand The weekly demand for a dozen roses is
11. [f(x) k] dx, where f(x)dx b and a, b, and
given by D(p) 316.765(0.949 p) dozen roses,
a a where p is the price per dozen, in dollars.
k are constants a. Find the price that corresponds to a weekly de-
mand of 80 dozen roses.
Applying Concepts b. Use the price (p0) found in part a to calculate
how much consumers are willing and able to
12. Decay The rate at which 15 grams of 14C is decay-
spend for 80 dozen roses per week.
ing can be modeled by
15. Demand The yearly demand for a certain hard-
r(t) 0.027205(0.998188t) grams per year
back science fiction novel is
where t is the number of years since the 15 grams
D(p) 499.589(0.958 p) thousand books
began decaying.
a. How much of the 14C will decay during the first where p is the price per book, in dollars.
1000 years? during the fourth 1000 years? a. Find the price that corresponds to a yearly de-
b. How much of the 14C will eventually decay? mand of 150,000 books.
b. Use the price (p0) found in part a to calculate
13. Decay An isotope of uranium, 238 U, is com-
how much consumers are willing and able to
monly used in atomic weapons and nuclear power
spend for 150,000 books each year.
generators. Because of this isotope’s radioactive
nature, the United States government is concerned 16. Work The work required to propel a 10-ton
with safe ways of storing used uranium. The rate at rocket an unlimited distance from the surface of
6.2 Streams in Business and Biology 385
Earth into space is defined in terms of force and is 17. Consider the function
given by the improper integral
0.1e 0.1x when x 1
f(x)
when x 1
160,000,000 0
W dx
4000 x2 Show that f is a probability density function.
160,000,000 18. Consider the function
The expression x2 is force in tons. The vari-
able x is the distance, measured in miles, between ⎧ 12 when x 1
the rocket and the center of Earth. g(x) ⎨x
⎩0 when x 1
a. What are the units on work in this context?
b. Calculate the work necessary to propel this Show that g is a probability density function.
rocket infinitely into space.
The following information is used in Activities 17 and 18:
A probability density function is defined as a non-
negative function f with the property that
f(x)dx 1
6.2 Streams in Business and Biology
Picture a stream flowing into a pond. You have probably just created a mental picture
of water that is flowing continuously into the pond. We can also imagine moneys that
are “flowing” continuously into an investment or new individuals that are “flowing”
continuously into an existing population.
It is not unreasonable to consider the income of large financial institutions and
major corporations as being received continuously over time in varying amounts. For
instance, consider utility companies that receive payments at varying times through-
out each month. Furthermore, with electronic transfer of funds, these payments can
be made at any time during the day or night. Such a flow of money is called a con-
tinuous income stream. When you make payments to a bank or to some other fi-
nancial institution for the purpose of investing money or repaying a loan, your pay-
ments are usually for the same fixed amount and are made at regular times that are
separated by a specified interval. Such a flow of money is called a discrete income
stream. Whether continuous or discrete, an income stream is usually described as a
rate, R(t), that varies with time t.
Determining Income Streams
Consider a business that currently posts a yearly profit of $4.3 million. The business
allocates 5% of its profits in a continuous stream among several investments. There
are several situations that could determine the flow rate into these investments. We
consider the cases where the flow is constant or either increases or decreases by a
constant or by a percentage.
If the company’s profits remain constant, then the function that describes the in-
come stream flowing into the investments is R(t) 0.05($4.3 million per year)
$0.215 million per year.
386 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
Suppose, however, that the company’s profits increase by $0.2 million each year.
In this case, the company’s profit is linear, beginning with $4.3 million dollars the first
year and increasing by a constant $0.2 million each year: 4.3 0.2t million dollars
per year after t years. The investment stream is 5% of the profit, so
R(t) 0.05(4.3 0.2t)million dollars per year t years after the company posted a
profit of $4.3 million.
It is also possible that the company’s profits could increase by a constant 7% each
year. Recall that constant percentage change is modeled by an exponential equation.
In this case, the function that describes the flow rate of the investment stream is
R(t) 0.05[4.3(1.07 t )] million dollars per year
t years after the company posted a profit of $4.3 million. Determining the rate at
which income flows into an investment is the first step in answering questions about
the present and future values of the invested income stream.
EXAMPLE 1 Writing Flow Rate Equations
Business Start-up
After you graduate from college, you start a small business that immediately becomes
successful. When you establish the business, you determine that 10% of your profits
will be invested each year. In the first year you post a profit of $579,000. Determine
the income stream flow rate for your investments over the next several years if
a. The business’s profit remains constant.
b. The profit grows by $50,000 each year.
c. The profit increases by 17% each year.
d. The profits for the first six years are as shown in Table 6.2 and are expected to
follow the trend indicated by the data.
TABLE 6.2
Year 1 2 3 4 5 6
Profit (thousands of dollars) 579 600 610 618 623 627
Solution
a. If the profit remains constant, then the flow rate of the investment stream is
constant, calculated as 10% of $579,000. Thus R(t) $57,900 per year.
b. Profit increasing by a constant amount each year indicates linear growth.
In this case, the amount of profit that is invested is described by the linear
function R(t) (0.10)(579,000 50,000t) dollars per year t years after the
first year.
c. An increase in profit of 17% each year indicates exponential growth with a con-
stant percentage change of 17%. In this case, the flow rate of the investment
stream is described by the equation R(t) 0.10[579,000(1.17 t )] dollars per
year t years after the first year of business.
6.2 Streams in Business and Biology 387
d. A scatter plot of the data in Table 6.2 indicates an increasing, concave-down
shape. (See Figure 6.5.) A quadratic model is a reasonably good fit, although the
model indicates declining profits beyond the years given in the table. A better
model for profit is the log model P(t) 580.117 26.8 ln t thousand dollars
per year after t years of business. Thus the investment flow rate is
R(t) 0.10P(t) 58.0117 2.68 ln t. Note that in parts a through c, the first
year of business corresponds to an input value of 0. In this log model, the first
year of business corresponds to an input value of 1.
Profit
(thousand
dollars)
630
615
600
585
570 Years
FIGURE 6.5 1 2 3 4 5 6 ●
Future Value of a Continuous Stream
The future value of a continuous stream is the total accumulated value of the income
stream and its earned interest. Suppose that an income stream flows continuously
into an interest-bearing account at the rate of R(t) dollars per year, where t is meas-
ured in years and the account earns interest at the annual rate of 100r% compounded
continuously. What is the future value of the account at the end of T years?
To answer this question, we begin by imagining the time interval from 0 to T years
as being divided into n subintervals, each of length t .
0 t t t T (t)(n)
We regard t as being small—so small that over a typical subinterval [t, t t],
the rate R(t) can be considered constant. Then the amount paid into the account dur-
ing this subinterval can be approximated by
Amount paid in [R(t) dollars per year](t years)
R(t)t dollars
We consider this amount as being paid in at t, the beginning of the interval, and
earning interest continuously for (T t) years. Using the continuously com-
pounded interest formula (A Pert), we see that the amount grows to
[R(t)t]er(Tt) R(t)er(Tt)t dollars
at the end of T years. Summing over the n subintervals, we have the approximation
Future value [R(t1)er(Tt1 ) R(t2)er (Tt2) … R(tn)e r (Ttn)] t dollars
where t1, t2, … , tn are the left endpoints of the n subintervals. This sum should look
familiar to you. If we simplify the expression by letting f(t) R(t)er (Tt) and rewrite
388 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
the sum as [f(t1) f(t2 ) … f (tn )]t , then you should recognize it as the type of
sum we used in Section 5.1.
Because we are considering the income as a continuous stream and interest as be-
ing compounded continuously, we let the time interval t become extremely small
(t → 0). That is, we use an infinite number of intervals (n → ). Thus
Future value lim [f(t1) f(t2) … f(tn )] t
n→
f(t)dt
T
0
R(t)e
T
r (Tt)dt dollars
0
Future Value of a Continuous Income Stream
Suppose that an income stream flows continuously into an interest-bearing
account at the rate of R(t) dollars per year, where t is measured in years and
the account earns interest at the annual rate of 100r% compounded
continuously. The future value of the account at the end of T years is
R(t)e
T
Future value r(Tt)dt dollars
0
Using the Fundamental Theorem, we can find the rate-of-change function for
future value:
Rate of change of future value
d x
dx 0
R(t)e r (Tt)dt for 0 x T
R(x)e r(Tx) dollars per year
Thus the function f(t) R(t)e r (Tt) gives the rate of change after t years of the fu-
ture value (in T years) of an income stream whose income is flowing continuously in
at a rate of R(t) dollars per year. It is the rate-of-change function f(t) R(t)e r (Tt),
not the flow rate of the income stream R(t), that we graph when depicting future
value as the area of a region beneath a rate-of-change function.
EXAMPLE 2 Finding the Future Value of a Continuous Income Stream
Airline Expansion The owners of a small airline are making big plans. They hope
to be able to buy out a larger airline 10 years from now by investing into an account
returning 9.4% APR. Assume a continuous income stream and continuous com-
pounding of interest.
a. The owners have determined that they can afford to invest $3.3 million each
year. How much will these investments be worth 10 years from now?
b. If the airline’s profits increase so that the amount the owners invest each year
increases by 8% per year, how much will their investments be worth in 10 years?
6.2 Streams in Business and Biology 389
Solution
a. The flow rate of the income stream is R(t) 3.3 million per year with
r 0.094 and T 10 years. The value of these investments in 10 years is cal-
culated as
3.3e
10
Future value 0.094(10t)dt
0
3.3e
10
0.94e0.094tdt
0
3.3e 0.94 0.094(10) 3.3e 0.94 0.094(0)
e e
0.094 0.094
35.106 89.872
$54.8 million
b. The function modeling exponential growth of 8% per year in the investment
stream is R(t) 3.3(1.08 t ) million dollars per year after t years. The future value
3.3(1.08 )e
10
is calculated (using technology)as t 0.094(10t)dt $77.7 million. ●
0
Present Value of a Continuous Stream
The present value of a continuous income stream is the amount P that would have to
be invested now in an interest-bearing account in order for the amount to grow to a
given future value. Because P dollars earning continuously compounded interest
would grow to a future value of PerT dollars in T years, we have
R(t)e R(t)e e R(t)e
T T T
PerT r(Tt)dt rT rtdt erT rtdt
0 0 0
Solving for P, we obtain
R(t)e
T
Present value P rtdt
0
Present Value of a Continuous Income Stream
Suppose that an income stream flows continuously into an interest-bearing
account at the rate of R(t) dollars per year, where t is measured in years, and
that the account earns interest at the annual rate of 100r% compounded
continuously. The present value of the account is
R(t)e
T
Present value rtdt dollars
0
It is worth noting that once you have calculated future value, it is easy to calculate the
associated present value by solving for P in the equation
Pert Future value
390 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
EXAMPLE 3 Finding Present Value from Future Value
Profit Last year, profit for the HiTech Corporation was $17.2 million. Assuming
that HiTech’s profits increase continuously for the next 5 years at a rate of $1.3 mil-
lion per year, what are the future and present values of the corporation’s 5-year prof-
its? Assume an interest rate of 12% compounded continuously.
Solution
We note that the rate of the stream is R(t) 17.2 1.3t million dollars per year
in year t. In order to calculate the future value of this stream, we evaluate
(17.2 1.3t)e
5
0.12(5t )dt . We have not developed a method for finding the anti-
0
derivative of f(t) (17.2 1.3t)e 0.12(5t), so we numerically estimate the definite
integral using a limiting value of sums or use technology to evaluate the integral.
(17.2 1.3t)e
5
Future value 0.12(5t )dt $137.9 million
0
The invested revenue will be worth approximately $138 million in 5 years. Again, nu-
merically estimate a limiting value of sums or use technology to find the present
value.
(17.2 1.3t)e
5
Present value 0.12tdt $75.7 million
0
This amount is the lump sum ($75.7 million) that would have to be invested at 12%
compounded continuously in order to earn $137.9 million (the future value) in 5
years.
We could also use the future value ($137.9 million) to calculate the present value:
Pe (0.12)(5) $137.9 million so P $75.7 million
The integral definition of the present value is useful in situations in which you do not
know the future value. ●
Discrete Income Streams
The assumptions that income is flowing continuously and that interest is com-
pounded continuously make it possible to use calculus and are often imposed by
economists. Unfortunately, they do not generally hold in the real world of business.
It is much more realistic to consider an income stream that flows monthly into an ac-
count with monthly compounding of interest or a stream that flows quarterly with
quarterly compoundings.
The process of finding the future value for discrete income streams begins in a
similar way to that for continuous streams: Determine the rate-of-flow function for
the income stream, and multiply by a term that accounts for compounding interest.
In the discrete case, we base our interest calculations on the formula A P 1 n
r nt
where A is the dollar amount accumulated after t years when P dol-
lars are invested at an annual interest rate of 100r% compounded annually n times a
year. Instead of integrating the resulting function, we sum a series of values.
Consider a small business that begins investing 7.5% of its monthly profit into an
6.2 Streams in Business and Biology 391
account that pays 10.3% annual interest compounded monthly. When the company
begins investing, the monthly profit is $12,000 and is growing by $500 each month.
We wish to determine the 2-year future value of the company’s investment (assum
ing that profit continues to grow in the manner described).
The company will make a total of 24 deposits during the 2-year period. The first
deposit is (0.075)($12,000) $900. This deposit earns interest each month for 24
months. The future value of this first deposit is $900 1 12 $1104.91. 0.103 24
The second deposit is (0.075)($12,000 $500) $937.50. This deposit earns
interest each month for 23 months. The future value of the second deposit is
$937.50 1 12
0.103 23
$1141.15.
The third deposit is (0.075)($12,000 $1000) $975. This deposit earns inter-
est each month for 22 months. The future value of the third deposit is
$975 1 12
0.103 22
$1176.70.
By now you should be able to see a pattern in the future values of these successive
monthly deposits. (See Table 6.3.)
TABLE 6.3
Time d
(months after first deposit) Future value of monthly deposit F (d )
0
F(0) $900.00 1 12
0.103 24
$1104.91
1
F(1) $937.50 1 12
0.103 23
$1141.15
2
F(2) $975.00 1 12
0.103 22
$1176.70
22 F(22) $1725.00 1 0.103 2
12
$1754.74
23
F(23) $1762.50 1 12
0.103 1
$1777.63
Future value sum of 24 values $35,204.03
The amount deposited each month is given by R(d) (0.075)($12,000 $500d),
where d is the number of deposits after the first one. The deposit associated with input
d accrues interest for 24 d months. The future value of each month’s deposit is given
by the formula
24d 24d
F(d) R(d) 1 0.103
12
(0.075)(12,000 500d) 1 0.103
12
To determine the 2-year future value, we add the future values of each month’s de-
posit beginning with 0 (month 1) and ending with 23 (month 24). Using summation
notation, we write
23 23
(0.075)(12,000 500d)1
0.103 24d
Future value F(d) d0
d0 12
392 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
23
The symbol F(d)
d0
is the notation used for the sum F(0) F(1) F(2)
. . . F(23). In this example, the future value of the deposits is $35,204.03. We
generalize the definition of future value for discrete income streams as follows:
Future Value of a Discrete Income Stream
Suppose that a deposit is made into an interest-bearing account at n equally
spaced times throughout a year. The value of the dth deposit is R(d) dollars
per period, and it earns interest at an annual percentage rate of 100r%
compounded once in each deposit period. The future value of the deposits at
the end of D deposit periods is
D1
R(d)1 n
r Dd
Future value dollars
d0
Once you have determined future value, present value can be found by solving for P
in the equation
r D
P 1
n
Future value
EXAMPLE 4 Determining the Future Value of a Discrete Income Stream
Savings
6.2.1 When you graduate from college (say, in 3 years), you would like to purchase a car.
You have a job and can put $75 into savings each month for this purchase. You
choose a money market account that offers an APR of 6.2% compounded quarterly.
a. How much money will you have deposited in 3 years?
b. What will be the value of your savings in 3 years?
c. How much money would you have to deposit now (in one lump sum) to
achieve the same future value in 3 years?
d. You are considering a second money market account that pays monthly inter-
est of 0.5%. Will this account result in a greater future value than that calcu-
lated in part b?
Solution
a. The total amount deposited is (36)($75) $2700.
b. Because interest is compounded quarterly (n 4), the monthly deposits each
quarter do not earn interest until the end of the quarter in which they are de-
posited. We therefore consider three $75 monthly deposits equivalent to one
$225 quarterly deposit, and this gives R(d) $75(3) $225 per quarter d
6.2 Streams in Business and Biology 393
quarters after the first one. We assume that the first deposit is made at the
beginning of a quarter. During the 3 years, there will be 12 quarters to consider
(D 12). Table 6.4 shows the pattern of the future values of each quarter’s
deposit.
The 3-year future value is given by
121
R(d)1
0.062 12d
Future value
d0 4
11
$2251
0.062 12d
$2988.10
d0 4
TABLE 6.4
Time (quarters after first deposit) Future value of quarterly deposit
0.062
12
0 $225 1 $270.61
4
0.062 11
1 $225 1 $266.48
4
0.062
10
2 $225 1 $262.41
4
0.062
2
10 $225 1 $232.03
4
0.062
1
11 $225 1 $228.48
4
Future value sum of 12 values $2988.10
c. Because we know the future value, we can solve P 1 4 0.062 12
2988.10 for
the present value P to obtain P $2484.48. This is the amount that you would
need to deposit now to have $2988.10 in 3 years.
d. Because the account pays interest monthly, we have the values n 12 and
D 3(12) 36. Also, because interest is compounded monthly, we consider
the deposits being made monthly, so that R(d) $75. In this case, the interest
rate is given in terms of a monthly rather than a yearly rate. In other words, we
r r
are given the term n 12 0.005. Thus the 3-year future value of $75
monthly deposits into this money market account is
35
Future value $75(1.005)36d $2964.96
d0
This is a smaller future value than for the account that pays 6.2% annual interest
compounded quarterly. ●
Streams in Biology
Biology and other fields involve situations very similar to income streams. An example
of this is the growth of populations of animals. As of 1978, there were approximately
394 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
1.5 million sperm whales* in the world’s oceans. Each year approximately 0.06 million
sperm whales are added to the population. Also each year, 4% of the sperm whale pop-
ulation either die of natural causes or are killed by hunters. Assuming that these rates
(and percentage rates) have remained constant since 1978, we estimate the sperm whale
population in 1998 using the same procedure as when determining future value of con-
tinuous income streams.
There are two aspects of the population that we must consider when estimating
the population of sperm whales in 1998. First, we must determine the number of
whales that were living in 1978 that will still be living in 1998. Because 4% of the
sperm whales die each year, we calculate the number of whales that have survived the
entire 20 years as 1.5(0.9620) 0.663 million whales.
The second aspect that we must consider is the impact on the population made
by the birth of new whales. We are told that 0.06 million whales are added to the pop-
ulation each year and that each year, 96% of those survive. Therefore, the growth rate
of the population of sperm whales associated with those that were born t years after
1978 is
f(t) 0.06(0.96)20t million whales per year
Thus the sperm whale population in 1998 is calculated as
20
Whale population 1.5(0.9620) 0.06(0.96)20tdt
0
20
0.663 0.06(0.9620)(0.96)tdt
0
20
0.663 0.06(0.9620) (0.961)tdt
0
0.06(0.9620)(0.961)t
20
0.663
ln (0.961) 0
1.48 million sperm whales
Functions that model such biological streams, in which new individuals are added
to the population and the rate of survival of the individuals is known, are referred to
as survival and renewal functions.
Future Value of a Biological Stream
The future value (in b years) of a biological stream with initial population
size P, survival rate 100s%, and renewal rate r(t), where t is the number of
years, is
r(t)s
b
Future value Ps b btdt
0
In the whale example, the initial population is P 1.5 million. The survival rate is
96% per year, so s 0.96. The renewal rate is r(t) 0.06 million whales per year.
* Delphine Haley, Marine Mammals, Seattle, WA: Pacific Search Press, 1978.
6.2 Streams in Business and Biology 395
EXAMPLE 5 Determining the Future Value of a Biological Stream
Flea Population An example of a stream in entomology is the growth of a flea pop-
ulation. In cooler areas of the country, adult fleas die before winter, but flea eggs sur-
vive and hatch the following spring when temperatures again reach 70°F. Not all the
eggs hatch at the same time, so part of the growth in the flea population is due to
the hatching of the original eggs. Another part of the growth in the flea population is
due to propagation. Suppose fleas propagate at the rate of 134% per day and that the
original set of fleas (from the dormant eggs) become reproducing adults at the rate of
600 fleas per day. What will the flea population be 10 days after the first 600 fleas be-
gin reproducing? Assume that none of the fleas die during the 10-day period and that
all fleas become reproducing adults 24 hours after hatching and propagate every day
thereafter at the rate of 134% per day.
Solution
We first note that because we begin counting when the first 600 fleas have become
mature adults, we consider the initial population to be P 600 fleas. The renewal
rate is also 600 fleas per day, so r(t) 600.
Because, in this case, the renewal rate function r does not account for renewal due
to propagation, we must incorporate the propagation rate of 134% into the survival
rate of 100%. Thus the survival/propagation rate is s 2.34.
Because the renewal rate and survival/propagation rate are given in days, we let t
be the input variable measured in days. The flea population will grow over 10 days to
10 10
Flea population Ps10 r(t)s10tdt 600(2.3410) 600(2.34)10tdt
0 0
2,953,315 3,473,166 6.4 million fleas ●
6.2 Concept Inventory nation; however, you are determined to buy a house
within 5 years of beginning your new job. Your
• Income streams starting salary is $47,000. After talking with fellow
• Flow rate of a stream employees, you consider three possibilities for what
might happen to your salary over the next 5 years:
• Future and present value of a continuous stream
• Future and present value of a discrete stream i. Your salary remains at your starting level.
• Biological stream ii. Your salary increases by $100 a month.
• Future value of a biological stream iii. Your salary increases by 0.5% each month.
a. You have decided to save 20% of your salary each
month for a down payment on a house. Give the
function describing your monthly investments
6.2 Activities for each of the three salary possibilities.
Applying Concepts b. You estimate that you will need $60,000 for a
down payment, and you are unwilling to accept
1. Savings Suppose that after you graduate, you are any investment risk. You will be investing your
hired by a company in the San Francisco Bay area. money monthly in a bank savings account that
Housing prices in that region are the highest in the pays an annual interest rate of 5% compounded
396 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
monthly. For which salary possibilities will the studies while in high school. How much should they
total amount saved in 5 years be at least $60,000? lend her so that she can invest the loaned amount
now, as one lump sum, into the account and accu-
2. Investment A company is hoping to expand its
mulate the same amount as if she had made the
facilities but needs capital to do so. In an effort to
quarterly deposits for 3 years?
position itself for expansion in 3 years, the company
will direct half of its profits into investments in a 5. Revenue The revenue of General Motors
continuous manner. The company’s profits for the Corporation (GM) in December 2001 was $177.26
past 5 years are shown in the table. billion. Assume that GM’s revenue remains con-
stant and that 3% of the revenue is invested contin-
Years ago 5 4 3 2 1 uously throughout each year, beginning at the end
Profit 860 890 930 990 1050
of December 2001, into an account that pays inter-
(thousands of dollars) est at a rate of 8.8% compounded continuously.
(Source: Hoover’s Online Guide.)
The company’s current yearly profit is $1,130,000. a. Find the value of the account in December 2008.
Find the function that describes the flow of the b. How much would GM have had to invest at the
company’s investments for each of the following end of December 2001, in one lump sum, into
profit scenarios: this account in order to build the same 7-year fu-
a. The profit for the next 3 years follows the trend ture value as the one found in part a?
shown in the table. 6. Revenue For the year ending December 31, 2001,
b. The profit increases each year for the next 3 years the General Electric Company’s revenue was
by the same percentage that it increased in the $125.68 billion. Assume that the revenue increases
current year. by 8% per year and that General Electric will (con-
c. The profit remains constant at the current year’s tinuously) invest 10% of its profits each year at an
level. APR of 8.5% compounded continuously for a pe-
riod of 9 years beginning at the end of December of
d. The profit increases each year for the next 3 years
2001. What is the present value of this 9-year in-
by the same fixed amount that it increased this
vestment?
year.
(Source: Hoover’s Online Guide.)
e. If the company’s investments can earn 16.4% an-
nual interest compounded continuously, how 7. Savings To save for the purchase of your first
much capital will it have saved after 3 years of in- home (in 6 years), suppose you begin investing $500
vesting for each of the profit scenarios given? per month in an account with a fixed rate of return
of 6.34%.
3. Revenue For the year ending June 30, 2002, the
a. Assuming a continuous stream, what will the ac-
revenue of the Sara Lee Corporation was $17.628
count be worth at the end of 6 years?
billion. Assume that Sara Lee’s revenue will increase
by 5% per year and that beginning on July 1, 2002, b. Assuming monthly activity (deposits and inter-
12.5% of the revenue was invested each year (con- est compounding), what will the account be
tinuously) at an APR of 7% compounded continu- worth at the end of 6 years?
ously. What is the future value of the investment at c. Is the answer to part a or the answer to part b
the end of the year 2006? more likely to be the actual future value of the ac-
(Source: Hoover’s Online Guide.) count? Explain.
4. Savings A high school student is trying to save 8. Investment In preparing for your retirement (in
money to help pay for her first-year college tuition. 40 years), suppose you plan to invest 14% of your
She plans to invest $300 each quarter for 3 years into salary each month in an annuity with a fixed rate of
an account that pays interest at an APR of 4% com- return of 6.2%. You currently make $2800 per
pounded quarterly. Her parents decide to lend her month and expect your income to increase by 3%
the money so that she can devote more time to her per year.
6.2 Streams in Business and Biology 397
a. Assuming a continuous stream, what will the an- a. The vending of Pepsi products on campus
nuity be worth at the end of 40 years? can be considered a continuous process. As-
b. Assuming monthly activity (deposits and inter- suming that the revenue was invested in a
est compounding), what will the annuity be continuous stream and that interest on that in-
worth at the end of 40 years? vestment was compounded continuously, how
much did Pepsi make from its 7 years of sales at
c. Is the answer to part a or the answer to part b
Alma College?
more likely to be the actual future value of the
annuity? Explain. b. Still assuming a continuous stream, find how
much Pepsi would have had to invest in 1993 to
9. Profit Ticketmaster is the world’s largest ticket re- create the same 7-year future value.
tailer. Ticketmaster’s 2002 third-quarter gross prof-
its were $82.1 million. Assume that these profits will 12. Investment Refer to Activity 8. How much would
increase by 5% per quarter and that Ticketmaster you have to invest now, in one lump sum instead of
will invest 15% of its quarterly profits in an invest- in a continuous stream, in order to build to the
ment with a quarterly return of 9%. same future (40-year) value?
(Source: Hoover’s Online Guide.) 13. Savings Refer to Activity 7.
a. Write a function for the rate at which money a. How much would you have to invest now, in one
flows into this investment each quarter. lump sum instead of in a continuous stream,
b. Write a function for the rate at which the 4-year in order to build to the same future (6-year)
future value of this investment is changing. value? Assume that interest is compounded
continuously.
c. Find the value of this investment at the end of the
year 2006. (Assume a quarterly stream beginning b. How much would you have to invest now, in one
on January 1, 2003, with the investment of lump sum instead of in a monthly stream, in or-
fourth-quarter 2002 profits.) der to build to the same future (6-year) value?
Assume monthly compounding of interest.
10. Net Income For the 2002 fiscal year, Lowe’s Com-
c. Is the answer to part a or the answer to part b
panies, Inc., reported an annual net income of
more likely to be the actual present value of the
$1,023,300,000. Assume the income can be rein-
annuity? Explain.
vested continuously at an annual rate of return of
10% compounded continuously. Also assume that 14. Revenue Between 1995 and 2001, the revenue of
Lowe’s will maintain this annual net income for the Sears Roebuck and Co. can be modeled as
next 5 years. 11.24
(Source: Hoover’s Online Guide.) R(t) 30 billion dollars per year
1 1.366e1.55t
a. What is the future value of its 5-year net income?
t years after 1995. Assume that the revenue can be
b. What is the present value of its 5-year net income?
reinvested at 9.5% compounded continuously.
11. Cola Sales In 1993, PepsiCo installed a new soccer (Source: Based on data from the Hoover’s Online Guide.)
scoreboard for Alma College in Alma, Michigan. a. How much is Sears’ revenue invested since 1995
The terms of the installation were that Pepsi would worth in 2003?
have sole vending rights at Alma College for the
b. How much was this accumulated investment
next 7 years. It is estimated that in the 3 years after
worth in 1995?
the scoreboard was installed, Pepsi sold 36.4 thou-
sand liters of Pepsi products to Alma College stu- 15. Buyout In 1956, AT&T laid its first underwater
dents, faculty, staff, and visitors. Suppose that the phone line. By 1996, AT&T Submarine Systems,
average yearly sales and associated revenue re- the division of AT&T that installs and maintains
mained constant and that the revenue from Alma undersea communication lines, had seven cable
College sales was reinvested at 4.5% APR. Also as- ships and 1000 workers. On October 5, 1996, AT&T
sume that PepsiCo makes a revenue of $0.80 per announced that it was seeking a buyer for its Sub-
liter of Pepsi. marine Systems division. The Submarine Systems
398 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
division of AT&T was posting a profit of $850 mil- 17. Buyout CSX Corporation, a railway company,
lion per year. announced in October of 1996 its intention to buy
(Source: “AT&T Seeking a Buyer for Cable-Ship Business,” Conrail Inc. for $8.1 billion. The combined com-
Wall Street Journal, October 5, 1996.) pany, CSX-Conrail, would control 29,000 miles of
a. If AT&T assumed that the Submarine Systems track and have an annual revenue of $14 billion the
division’s annual profit would remain constant first year after the merger, making it one of the
and could be reinvested at an annual return of largest railway companies in the country.
15%, what would AT&T have considered to be (Source: “Seeking Concessions from CSX-Conrail Is Seen as Most
the 20-year present value of its Submarine Sys- Likely Move by Norfolk,” Wall Street Journal, October 5, 1996.)
tems division? (Assume a continuous stream.) a. If Conrail assumed that its $2 billion annual rev-
b. If prospective bidder A considered that the an- enue would decrease by 5% each year for the next
nual profits of this division would remain con- 10 years but that the annual revenue could be
stant and could be reinvested at an annual return reinvested at an annual return of 20%, what
of 13%, what would bidder A consider to be the would Conrail consider to be its 10-year present
20-year present value of AT&T’s Submarine Sys- value at the time of CSX’s offer? Is this more or
tems? (Assume a continuous stream.) less than the amount CSX offered?
c. If prospective bidder B considered that over a 20- b. CSX Corporation forecast that its Conrail acquisi-
year period, profits of the division would grow by tion would add $1.2 billion to its annual revenue
10% per year (after which it would be obsolete) the first year and that this added annual revenue
and that profits could be reinvested at an annual would increase by 2% each year. Suppose CSX is
return of 14%, what would bidder B consider to able to reinvest that revenue at an annual return of
be the 20-year present value of AT&T’s Subma- 20%. What would CSX Corporation have consid-
rine Systems? (Assume a continuous stream.) ered to be the 10-year present value of the Conrail
acquisition in October of 1996?
16. Buyout On October 4, 1996, Tenet Healthcare
c. Why might CSX Corporation have forecast an
Corporation, the second-largest hospital company
increase in annual revenue when Conrail fore-
in the United States at that time, announced that it
cast a decrease?
would buy Ornda Healthcorp.
(Source: “Tenet to Acquire Ornda,” Wall Street Journal, 18. Buyout Company A is attempting to negotiate a
October 5, 1996.) buyout of Company B. Company B accountants
a. If Tenet Healthcare Corporation assumed that project an annual income of 2.8 million dollars per
Ornda’s annual revenue of $0.273 billion would year. Accountants for Company A project that with
increase by 10% per year and that the revenues Company B’s assets, Company A could produce an
could be continuously reinvested at an annual re- income starting at 1.4 million dollars per year and
turn of 13%, what would Tenet Healthcare Cor- growing at a rate of 5% per year. The discount rate
poration consider to be the 15-year present value (the rate at which income can be reinvested) is 8%
of Ornda Healthcorp at the time of the buyout? for both companies. Suppose that both companies
b. If Ornda Healthcorp’s forecast for its financial consider their incomes over a 10-year period. Com-
future was that its $0.273 billion annual revenue pany A’s top offer is equal to the present value of its
would remain constant and that revenues could projected income, and Company B’s bottom price is
be continuously reinvested at an annual return equal to the present value of its projected income.
of 15%, what would Ornda Healthcorp consider Will the two companies come to an agreement for
its 15-year present value to be at the time of the the buyout? Explain.
buyout? 19. Capital Value A company involved in videotape
c. Tenet Healthcare Corporation bought Ornda reproduction has just reported $1.2 million net in-
Healthcorp for $1.82 billion in stock. If the sale come during its first year of operation. Projections
price was the 15-year present value, did either of are that net income will grow over the next 5 years
the companies have to compromise on what it at the rate of 6% per year. The capital value (present
believed to be the value of Ornda Healthcorp? sales value) of the company has been set as its pres-
6.2 Streams in Business and Biology 399
ent value over the next 5 years. If the rate of return 22. Population From 1936 through 1957, a population
on reinvested income can be compounded continu- of 15,000 muskrats in Iowa bred at a rate of 468 new
ously for the next 5 years at 12% per year, what is the muskrats per year and had a survival rate of 75%.
capital value of this company? (Source: Paul L. Errington, Muskrat Population, Ames, IA:
Iowa State University Press, 1963.)
20. Population There were once more than 1 million
a. How many of the muskrats alive in 1936 were
elephants in West Africa. Now, however, the ele-
still alive in 1957?
phant population has dwindled to 19,000. Each year
17.8% of West Africa elephants die or are killed by b. Write a function for the number of muskrats that
hunters. At the same time, elephant births are de- were born t years after 1936 and were still alive in
creasing by 13% per year. 1957.
(Source: Douglas Chawick, The Fate of the Elephant, Sierra c. Estimate the muskrat population in 1957.
Club Books, 1992.)
23. Population There are approximately 200 thou-
a. How many of the current population of 19,000
sand northern fur seals. Suppose the population
elephants will still be alive 30 years from now?
is being renewed at a rate of r(t) 60 0.5t
b. Considering that 47 elephants were born in the thousand seals per year and that the survival rate
wild this year, write a function for the number of is 67%.
elephants that will be born t years from now and (Source: Delphine Haley, Marine Mammals, Seattle, WA:
will still be alive 30 years from now. Pacific Search Press, 1978.)
c. Estimate the elephant population of West Africa a. How many of the current population of 200
30 years from now. thousand seals will still be alive 50 years
from now?
21. Population In 1979 there were 12 million sooty
terns in the world. Assume that the percentage of b. Write a function for the number of seals that will
terns that survive from year to year has stayed con- be born t years from now and will still be alive 50
stant at 83% and that approximately 2.04 million years from now.
terns hatch each year. c. Estimate the northern fur seal population 50
(Source: Bryan Nelson, Seabirds: Their Biology and Ecology, years from now.
New York: Hamlyn Publishing Group, 1979.)
a. How many of the terns that were alive in 1979 are Discussing Concepts
still alive?
24. Explain, using related examples, the difference be-
b. Write a function for the number of terns that
tween a continuous income stream and a discrete
hatched t years after 1979 and are still alive.
income stream.
c. Estimate the present population of sooty terns.
400 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
6.3 Integrals in Economics
This section uses improper When you purchase an item in a store, you ordinarily have no control over the price
integrals, which were discussed that you pay. Your only choice is whether to buy or not to buy the item at the current
in Section 6.1 price. In general, consumers hold to the view that price is a variable to which they can
only respond. As the price per unit increases, consumers usually respond by pur-
chasing (demanding) less. The typical relationship between the price per unit (as in-
put) and the quantity in demand (as output) is shown in Figure 6.6a.
q
Quantity demanded
500
p
Price per unit
20
q
p
Quantity
20 Price per unit 500 demanded
(a) The mathematician’s view (b) The economist’s view
FIGURE 6.6
The traditional approach to graphing in economic theory is to put the price per
unit along the vertical axis and the quantity in demand along the horizontal axis. (See
Figure 6.6b.) In general, we choose to graph price per unit as input along the hori-
zontal axis. This will help us understand price as input and visualize the definite in-
tegrals used later in this section. However, occasionally we will present both the
mathematician’s and the economist’s graphical viewpoints.
Demand Curves
The graph relating quantity in demand q to price per unit p is called a demand curve.
In economic theory, demand is actually a function that has several input variables,
such as price per unit, consumers’ ability to buy, consumers’ need, and so on. The de-
mand curve we consider here is a simplified version. We assume that all the possible
input variables are constant except price. We denote this demand function as D with
input p.
Even though the demand function is not a rate-of-change function, there are eco-
nomic interpretations for the areas of certain regions lying beneath the demand
curve. In order to interpret the area of these regions, you must understand how to
interpret the information the demand curve represents.
6.3 Integrals in Economics 401
q For instance, suppose the graph in Figure 6.7 represents the weekly demand for
(million regular unleaded gasoline in a California city. A point on the demand curve indicates
gallons) the quantity that consumers will purchase at a given price. For instance, at $3.39 per
gallon, consumers will purchase 1 million gallons of gas. At $2.00 per gallon, con-
sumers will purchase 2 million gallons of gas.
4
D(p)
3 Consumers’ Willingness and Ability to Spend
2 Even though points on the demand curve tell us how much consumers will actually
purchase at certain prices, consumers are willing and able to pay more than this
1
p
amount for the quantity they purchase. For instance, consumers are willing and able
0 (dollars to spend approximately $3.39 million for the first million gallons of regular unleaded
0 2 3.39 per gallon) gasoline, but they are willing and able to spend only approximately $2.00 million for
FIGURE 6.7 the second million gallons. Thus, in total, consumers are willing and able to spend ap-
proximately $5.39 million for 2 million gallons of gas.
If the price of gas is $1.19 a gallon, consumers are willing and able to buy the third
million gallons. That is, consumers are willing and able to spend approximately
(1 million gallons)($3.39 per gallon) (1 million gallons)($2.00 per gallon)
(1 million gallons)($1.19 per gallon) $6.58 million
for 3 million gallons of gas, even though in actuality they spend only
(3 million gallons)($1.19 per gallon) $3.57 million
Consumers’ willingness and ability to spend can be approximated graphically as the
areas of stacked horizontal rectangles. The amount that consumers actually spend is
depicted as the area of a single vertical rectangle. (See Figures 6.8a and b.)
You should have noticed that the amount that consumers are willing and able to
spend for 3 million gallons was given as approximately $6.58 million. We can make
this approximation better by considering smaller increments for price. If we were to
approximate consumers’ willingness and ability to spend using price increments of
q q
(million (million
gallons) gallons)
3 3
2 2
1 1 D(p)
D(p)
p p
0 (dollars 0 (dollars
0 1.19 2 3.39 per gallon) 0 1.19 per gallon)
(a) Consumers are willing and able to (b) Consumers pay only $3.57 million
spend about $6.58 million for for 3 million gallons of gas.
3 million gallons of gas.
FIGURE 6.8
402 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
$0.5 per gallon, $0.25 per gallon, $0.125 per gallon, and so on, we would see that the
areas of the stacked rectangles representing these approximations would become
closer to being the true area depicted in Figure 6.9.
q
(million
gallons)
D(p)
3
p
0 (dollars
FIGURE 6.9 0 1.19 per gallon)
Thus consumers’ willingness and ability to purchase 3 million gallons of gas can
be visually represented by the sum of the area of the rectangle with width 1.19 under
the horizontal line D(p) 3 and the area under the demand curve from 1.19 to P,
where P is the price above which consumers cannot and will not purchase any gas. We
calculate the consumers’ willingness and ability to spend as
P
3(1.19) D(p) dp million dollars
1.19
Suppose the demand for gas can be modeled by
D(p) 5.43(0.607p) million gallons
where p dollars is the price per gallon. The only piece of information we still need is P,
the price above which no gas will be purchased. You should notice that the demand
function approaches 0 as p becomes large; however, it will never be exactly 0 for any p.
Hence, we let P approach . This is true for most demand functions in economics—
some people will always want the product or service, regardless of the price. In this case,
we consider the area under the demand curve as P becomes infinitely large. That is,
This limit can be viewed
graphically as the limit of a
decreasing exponential
3(1.19)
1.19
5.43(0.607p)dp
P
function.
3.57 lim 5.43(0.607p)dp
P→ 1.19
ln 0.607 1 .19
p P
3.57 lim
P→
5.43(0.607 )
P 1.19)
3.57 lim P→ 5.43(0.607
ln 0.607
)
5.43(0.607
ln 0.607
3.57 0 6.00
$9.57 million
6.3 Integrals in Economics 403
Thus consumers are willing and able to spend $9.57 million in order to purchase 3
million gallons of gas.
In general, we make the following definition:
Consumers’ Willingness and Ability to Spend
For a continuous demand function q D(p), the maximum amount that
consumers are willing and able to spend for a certain quantity q0 of goods or
services is the area of the shaded region in Figure 6.10.
q
Quantity
q0
D(p)
p
Price
p0 P per unit
FIGURE 6.10
p where p0 is the market price at which q0 units are in demand and P is the
Price price above which consumers will purchase none of the goods or services.
per unit
This area is calculated as
D(p)dp
P P
p0q0
p0
(Note that is used as the upper limit on the integral if the demand function
approaches, but does not cross, the input axis.)
D(p)
p0
Because economists graph with the price on the vertical axis, in economics books
q the market price p0 at which q0 units are in demand and the price P above which con-
q0 Quantity sumers will purchase none of the goods or services both appear on the vertical axis,
whereas q0 is on the horizontal axis. The area depicting the maximum amount that
The economist’s view consumers are willing and able to spend appears (from the economist’s viewpoint) as
FIGURE 6.11 the shaded area in Figure 6.11.
404 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
Consumers’ Expenditure and Surplus
Now that we have considered what consumers are willing and able to spend for a cer-
tain quantity of a product, let us turn our attention to calculating what consumers ac-
tually spend for that quantity. We return to the discussion of gasoline demand.
As we said before, if the market price for gas is $1.19 per gallon, consumers
will purchase 3 million gallons. The actual amount spent by consumers is
(3 million gallons)($1.19 per gallon) $3.57 million, even though they are willing
and able to spend much more. This actual amount spent is (price)(quantity), which
is the area of the rectangular region from the vertical axis to p 1.19 with height 3,
as shown in Figure 6.12. This amount is known as the consumers’ expenditure. The
amount that consumers are willing and able to spend but do not actually spend is
known as the consumers’ surplus.
q
(million
gallons)
3
Consumers’
expenditure
D(p)
Consumers’
surplus p
0 (dollars
0 1.19 per gallon)
FIGURE 6.12
Earlier we found that consumers are willing and able to spend $9.57 million to
purchase 3 million gallons of gas, so the consumers’ surplus from buying 3 million
gallons of gas at $1.19 per gallon is
$9.57 million $3.57 million $6 million
Consumers’ surplus can also be computed as the area between the demand function
and the horizontal axis as
Consumers surplus
1.19
5.43(0.607p)dp $6 million
6.3 Integrals in Economics 405
In general, we make the following definitions:
Consumers’ Expenditure and Surplus
For a continuous demand function q D(p), the amount that consumers
spend at a certain market price is called consumers’ expenditure; it is
represented by the rectangular area in Figure 6.13. Furthermore, the amount
that consumers are willing and able to spend, but do not spend, for q0 items
at a market price p0 is called consumers’ surplus; it is represented by the area
of the nonrectangular shaded region in Figure 6.13. The value P is the price
above which consumers will purchase none of the goods or services.
q
Quantity
q0
Consumers’
expenditure
D(p)
Consumers’
surplus p
Price
p0 P per unit
p
Price
FIGURE 6.13
per unit
P
These areas are calculated as follows:
Consumers’ expenditure p0q0
D(p)dp
P
Consumers’ surplus
Consumers’ p0
surplus
(Note that is used as the upper limit on the integral if the demand function
D(p)
approaches, but does not cross, the input axis.)
Consumers’
p0 expenditure
q
q0 Quantity Again it is worth noting that because economists graph with the price on the ver-
tical axis, graphs showing consumers’ expenditure and surplus appear in economics
The economist’s view books with the rectangle representing consumers’ expenditure lying below the area
FIGURE 6.14 representing consumers’ surplus, as shown in Figure 6.14.
406 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
EXAMPLE 1 Finding Areas Involving a Demand Curve
Minivans Suppose the demand for a certain model of minivan in the United States
can be described as
D(p) 14.12(0.933p) 0.25 million minivans
when the market price is p thousand dollars per minivan.
a. At what price per minivan will consumers purchase 2.5 million minivans?
b. What is the consumers’ expenditure when purchasing 2.5 million minivans?
c. Does the model indicate a possible price above which consumers will purchase
no minivans? If so, what is this price?
d. When 2.5 million minivans are purchased, what is the consumers’ surplus?
e. What is the total amount that consumers are willing and able to spend on 2.5
million minivans?
Solution
a. We solve D(p) 2.5 to find the market price at which consumers will purchase
2.5 million minivans. The equation
14.12(0.933p) 0.25 2.5
is satisfied when p 23.59033. That is, at a market price p0 of approximately
$23,600 per minivan, consumers will purchase q0 2.5 million minivans.
b. When they purchase 2.5 million minivans, consumers’ expenditure will be
p0q0 (23.59033 thousand dollars per minivan)(2.5 million minivans)
$59.0 billion
c. If the demand function approaches but does not cross the horizontal axis as
price per unit increases without bound, then there is no price above which con-
sumers will not purchase minivans. However, in this case, the demand function
crosses the horizontal axis near p 58.16701 (found by solving D(p) 0).
According to the model, the price above which consumers will purchase no
minivans is approximately p $58.2 thousand per minivan.
q
(million d. Consumers’ surplus is the area of the region shaded in Figure 6.15,
minivans) calculated as
D(p)dp
P 58.16701
D(p) [14.12(0.933p) 0.25]dp
Consumers’ P0 23.59033
expenditure
27.40482
Consumers’
2.5 surplus p To determine the appropriate units for consumers’ surplus, remember
(thousand that we are finding the area of a region whose width is measured in thou-
dollars per sand dollars per minivan and whose height is measured in million mini-
23.6 58.2 minivan)
vans. Thus the units on consumers’ surplus are (thousand dollars per
FIGURE 6.15 minivan)(million minivans), which simplify to billion dollars.
6.3 Integrals in Economics 407
Therefore, we estimate the consumers’ surplus when purchasing 2.5 million
minivans to be $27.4 billion.
e. The amount that consumers are willing and able to spend for 2.5 million mini-
vans is the combined area of the two shaded regions in Figure 6.15. This area is
approximately 59.0 27.4 $86.4 billion. ●
Elasticity of Demand
As the price per unit of a commodity increases, we expect the demand for that com-
modity to decrease. For instance, when the price per gallon of gasoline increases,
the demand for gasoline decreases slightly. On the other hand, when the price per
unit for minivans increases, the demand for minivans decreases substantially. A
measure of the responsiveness of consumers to a change in the price of a commod-
ity is elasticity of demand. Because the quantity of a commodity and the price per
unit of that commodity normally have very different units of measure, elasticity of
demand uses a ratio of percentage rates of change to compare relative changes. For
a demand function D with input p price per unit, we define elasticity of demand as
the absolute value of the percentage rate of change of the quantity demanded
divided by the percentage rate of change of the price per unit. Algebraically elastic-
ity of demand can be written as
elasticity of demand percentage
percentage rate of change of p
rate of change of D(p)
100
D(p)
D(p)
pp 100
D(p)
p 1 D(p)
Note that because
p p
d
dp
(p) 1. p1
D(p)
D(p) 1
p
p D(p)
D(p)
We will use the Greek letter eta, , to represent elasticity of demand. When is
greater than 1 at a given price per unit k, a small change in price results in a relatively
large response in the change of demand and we say that the demand is elastic at price
k. When is less than 1 at a given price per unit k, a small change in price results in a
relatively small response in the change of demand and we say that the demand is
inelastic at price k.
408 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
Elasticity of Demand
For a commodity with differentiable demand function, q D(p), where q is
the quantity demanded when the price per unit is p, the elasticity of demand is
percentage
percentage rate of change of price
rate of change of quantity
D(p)
p D(p)
Demand is elastic when 1 and inelastic when 1. Demand is at unit
elasticity when 1.
EXAMPLE 2 Finding Elasticity of Demand
Minivans Again consider the demand for a certain model of minivan in the United
States as described in Example 1:
D(p) 14.12(0.933p) 0.25 million minivans
when the market price is p thousand dollars per minivan.
a. Find the point of unit elasticity.
b. For what prices is the demand elastic? For what prices is the demand inelastic?
Solution
a. First, we find an expression for elasticity:
p D(p)
D(p)
p [14.12(ln 0.933)(0.933p)]
14.12(0.933p)0.25
Solving 1, we find that p 13.76. That is, unit elasticity occurs when mini-
vans are priced approximately 13.76 thousand dollars per minivan. At this
price, demand is approximately 5.19 million minivans.
b. We begin by checking for values of p on either side of 13.76. We choose to
check at p 10 and p 20.
When p 10, 10 [14.12(ln 0.933)(0.93310)]
14.12(0.93310)0.25 0.72. So for prices less than
13.76 thousand dollars per minivan, demand is inelastic (not highly responsive
to small increases in price.)
When p 20, 20 [14.12(ln 0.933)(0.93320)]
14.12(0.93320)0.25 1.49. So for prices greater
than 13.76 thousand dollars per minivan, demand is elastic (responsive to small
increases in price). ●
6.3 Integrals in Economics 409
A point of unit elasticity on a concave up, decreasing demand function q D(p)
for a commodity equates to a maximum on the revenue function R(p) p D(p) for
that commodity.
Supply Curves
We have seen that when prices go up, consumers usually respond by demanding less.
However, manufacturers and producers respond to higher prices by supplying more.
Thus a typical curve that relates the quantity supplied, S, to price per unit, p, is usu-
ally increasing and appears as shown in Figure 6.16.
The graph that expresses the quantity supplied in terms of the price per unit is
q
called a supply curve. You should note from Figure 6.16 that there is a price p1 below
Quantity
which producers are not willing or able to supply any quantity of the product. The
point (p1, S(p1)) is known in economics as the shutdown point. If the market price
S(p)
(and the corresponding quantity) fall below this point, producers will shut down
their production.
The supply function has an interpretation very similar to that of the demand
function. Suppose the quantity of regular unleaded gasoline that producers will sup-
p ply is modeled as
Price
p1 per
unit S(p) 0 million gallons
0.792p2 0.433p 0.314 million gallons
when p 1
when p 1
FIGURE 6.16
where the market price of gas is p dollars per gallon. This function is graphed in teal
in Figure 6.17. At $1.24 a gallon, producers will supply 1 million gallons of gas. At
$1.78 per gallon, producers will supply 2 million gallons, and if the price is $2.14 a
gallon, producers will supply 3 million gallons.
q
(million gallons)
5
4 S(p)
1
p
0 (dollars per
0 1 1.24 1.78 2.14 3 gallon)
FIGURE 6.17
410 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
Producers’ Willingness and Ability to Receive
We approximate the minimum that producers are willing and able to receive for
3 million gallons of gas by summing the amount they are willing and able to receive
for the first million gallons, for the second million gallons, and for the third million
gallons:
($1.24 per gallon)(first 1 million gallons) ($1.78 per gallon)
(second 1 million gallons) ($2.14 per gallon)(third 1 million gallons)
$1,240,000 $1,780,000 $2,140,000 $5,160,000
Thus the minimum that producers are willing and able to receive for 3 million gal-
lons of gas is approximately $5,160,000. This amount can be thought of as the sum of
the areas of the three stacked rectangles shaded in Figure 6.17.
As we use more intervals, the area of stacked rectangles comes closer to the true
area of the region above the q S(p) curve and below the q 3 line shown in Fig-
ure 6.18. Because a portion of S(p) is zero, we find the total area by dividing the re-
gion into a rectangular region and the region below q 3 and above q S(p) to the
right of the shutdown point.
q
(million gallons)
4
3 S(p)
1
p
0 (dollars
0 1 2.14 per gallon)
FIGURE 6.18
Therefore, the minimum amount that suppliers are willing and able to receive is
calculated as
2.14
3(1) [3 S(p)]dp
1
2.14
3 (0.792p2 0.433p 2.686)dp
1
$4.5 million
According to the supply model S, suppliers are willing and able to receive no less than
$4.5 million for 3 million gallons of gas.
6.3 Integrals in Economics 411
We make the following general definition:
Producers’ Willingness and Ability to Receive
For a continuous or piecewise continuous supply function q S(p), the
minimum amount that producers are willing and able to receive for a certain
quantity q0 of goods or services is the area of the shaded region in Figure 6.19.
q
Quantity
Producers’
willingness
and ability S(p)
to receive
q0
p
Price
p1 p0 per unit
FIGURE 6.19
where p0 is the market price at which q0 units are supplied and p1 is the
shutdown price. This area is calculated as
[q S(p)]dp
p0
p1q0 0
p1
If there is no shutdown price, then p1 0.
p
Price In the economist’s viewpoint (with price graphed on the
per unit vertical axis), the minimum amount that producers are willing and able
to receive for a certain quantity q0 of goods or services is the shaded re-
gion shown in Figure 6.20. The market price p0 at which q0 units are sup-
S(p) plied and the shutdown price p1 are both graphed on the vertical axis.
p0
However, because price is the input and quantity is the output, the
[q S(p)]dp.
p0
area is still calculated as p1q0 0
Producers’ p1
willingness
p1 and ability
to receive Producers’ Revenue and Surplus
q The market price that will lead to the supply of 3 million gallons
q0 Quantity of gas is p $2.14 per gallon. The producers’ revenue is price times
quantity, which is the area of the rectangle shown in Figure 6.21: ($2.14 per
The economist’s view gallon)(3 million gallons) $6.4 million. Producers will therefore receive
FIGURE 6.20 $6.4 million $4.5 million $1.9 million in excess of the minimum
412 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
q they are willing and able to receive. This excess amount is known
(million gallons) as the producers’ surplus and is the area of the shaded region in
4 Figure 6.21.
S(p) We calculate the producers’ surplus from the sale of 3 million
3
gallons of gas at the market price of approximately $2.14 directly
2 from the supply function as follows:
2.14 2.14
1 Producers’
p S(p)dp (0.792p2 0.433p 0.314)dp
surplus 1 1
0 (dollars
0 1 2.14 per gallon) $1.9 million
The area of the rectangle is producers’ revenue. In general, we find the producers’ total revenue and the
FIGURE 6.21
producers’ surplus as follows:
Producers’ Revenue and Surplus
For a continuous or piecewise continuous supply function q S(p), the amount that
producers receive at a certain market price is called the producers’ revenue and is the
area of the shaded rectangle in Figure 6.22a. Furthermore, the amount that producers
receive above the minimum amount they are willing and able to receive for q0 items at
a market price p0 is called the producers’ surplus and is the area of the region between
the supply function and the horizontal axis as shown in Figure 6.22b. The value p1 is
the price below which production shuts down. (If there is no shutdown price,
then p1 0.)
q q
Quantity Quantity
Producers’
willingness
S (p) and ability S (p)
to receive
q0 q0
Producers’
revenue Producers’
surplus
p p
Price Price
p1 p0 per unit p1 p0 per unit
(a) (b)
FIGURE 6.22
These areas are calculated as follows:
Producers’ revenue p0q0
S(p)dp
p0
Producers’ surplus
p1
6.3 Integrals in Economics 413
Figure 6.23 is the economist’s version of Figure 6.22. Because economists
graph price vertically, producers’ willingness and ability to receive and producers’ sur-
plus appear to be transposed, as shown in Figure 6.23b. However, because the supply
function S uses price per unit p as input, the calculations for producers’ revenue and
producers’ surplus are the same as given in the preceding box.
p p
Price Price
per unit per unit
Producers’
surplus
S (p) S (p)
p0 p0
Producers’
willingness
p1 p1 and ability
Producers’
to receive
revenue
q q
q0 Quantity q0 Quantity
(a) (b)
The economist’s view
FIGURE 6.23
EXAMPLE 3 Finding Areas Involving a Supply Curve
Phones Suppose the function for the average weekly supply of a certain brand of
cellular phone can be modeled by the equation
S(p) 0 phones
0.047p2 9.38p 150 phones
when p 15
when p 15
where p is the market price in dollars per phone.
a. How many phones (on average) will producers supply at a market price of
$45.95?
b. What is the least amount that producers are willing and able to receive for the
quantity of phones that corresponds to a market price of $45.95?
c. What is the producers’ revenue when the market price is $45.95?
d. What is the producers’ surplus when the market price is $45.95?
Solution
a. When the market price is $45.95, producers will supply an average of
S(45.95) 680 phones each week.
414 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
b. The minimum amount that producers are willing and able to receive is the area
of the labeled region in Figure 6.24 and is calculated as
45.95
(15)(680.247) [680.247 S(p)]dp
15
10,203.704 (0.0157p3 4.69p2 530.247p) 45.95
15
$16,300.53
q
(phones)
680 S(p)
Producers’ willingness
and ability to receive
300
Producers’
surplus
p
(dollars per
0 15 45.95 phone)
FIGURE 6.24
c. When the market price is $45.95, the producers’ revenue is
(Quantity supplied at $45.95)($45.95 per phone)
(680.247 phones)($45.95 per phone)
$31,257.35
Graphically, the producers’ revenue is the area of the rectangle in Figure 6.24.
q
d. When the market price is $45.95, the producers’ surplus (see Figure 6.24) is
Quantity
45.95
S (p) S(p)dp (0.0157p3 4.69p2 150p) 45.95
15 $14,956.82
15
Note that the producers’ surplus plus the minimum amount that the producers
are willing and able to receive is equal to the producers’ revenue. ●
q
p Social Gain
D(p)
Price
per
Consider the economic market for a particular item for which the demand and sup-
p ply curves are shown in Figure 6.25. The point (p*, q*) where the demand curve and
unit
supply curve cross is called the equilibrium point. At the equilibrium price p*, the
Market equilibrium (p*, q*) quantity demanded by consumers coincides with the quantity supplied by producers.
occurs when demand is equal This quantity is q*.
to supply.
Economists consider that society is benefited whenever consumers and/or pro-
FIGURE 6.25 ducers have surplus funds. When the market price of a product is the equilibrium
6.3 Integrals in Economics 415
price for that product, the total benefit to society is the consumers’ surplus plus the
producers’ surplus. This amount is known as the total social gain.
Market Equilibrium and Social Gain
Market equilibrium occurs when the supply of a product is equal to the
demand for that product. If q D(p) is the demand function and q S(p)
is the supply function, then the equilibrium point is the point (p*, q*),
where p* is the price that satisfies the equation D(p) S(p) and
q* D(p*) S(p*).
The total social gain for a product is the sum of the producers’
surplus and the consumers’ surplus. When q* units are produced and
sold at a market price of p*, the total social gain is the area of the entire
shaded region in Figure 6.26. The value p1 is the price below which
production shuts down, and P is the price above which consumers will
not purchase.
q
Quantity
D(p) S (p)
Producers’
surplus
q Consumers’
surplus
p
Price
p1 p P per unit
FIGURE 6.26
The combined area is calculated as
Total social gain producers’ surplus consumers’ surplus
D(p)dp
p* P
S(p)dp
p1 p
The economist’s view is shown in Figure 6.27. The visual representation does
not change the manner in which the calculations are performed in order to find the
equilibrium point and total social gain.
416 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
p
Price
per unit
D(p)
Consumers’
surplus S(p)
p
p1 Producers’
surplus
q
Quantity
q
The economist’s view
FIGURE 6.27
EXAMPLE 4 Determining Market Equilibrium and Social Gain
Gasoline The demand and supply functions for the gasoline example given near
the beginning of this section are
Demand D(p) 5.43(0.607p) million gallons
and
Supply S(p) 0 million gallons when p 1
0.792p2 0.433p 0.314 gallons when p 1
where p is the market price in dollars per gallon.
a. Find the market equilibrium point for gasoline.
b. Find the total social gain when gasoline is sold at the market equilibrium price.
Solution
a. Solving
5.43(0.607p) 0.792p2 0.433p 0.314
for p yields p* $1.83 per gallon. At this market price, q* 2.2 million gallons
of gas will be purchased. [Note: q* can be found as either D(p*) or S(p*).]
b. The total social gain at market equilibrium is the area of the shaded regions in
Figure 6.28. We must find p1, the shutdown price, and P, the price beyond
which consumers will purchase no gasoline, before we can proceed.
6.3 Integrals in Economics 417
q The shutdown price is given in the statement of the supply function
(million gallons) as p1 1. The demand function indicates that there is no price
D(p) S (p) beyond which consumers will not purchase. Thus, P → .
Now we proceed with our calculation of total social gain:
1.8311
Consumers’
surplus
Total social gain (0.792p2 0.433p 0.314)dp
q = 2.2 1.00
Producers’
surplus
p 5.43(0.607p)dp
1.8311
(dollars per
1 p = 1.83 gallon) 1.108 4.360 $5.5 million
FIGURE 6.28 At the market equilibrium price of $1.83 per gallon, the total social
gain is approximately $5.5 million. ●
c. The area of the region below the supply curve
6.3 Concept Inventory between the shutdown price and the market
price.
• Market price d. The area of the region below the demand curve
• Demand curve between the market price and the price above
• Consumers’ willingness and ability to spend which consumers will cease to purchase.
• Consumers’ expenditure 2. For each of the following amounts:
• Consumers’ surplus i. Describe the region whose area gives the
• Elasticity of demand specified amount.
• Supply curve ii. Illustrate that region by sketching an
• Shutdown point example.
• Producers’ revenue a. The maximum amount that consumers are
• Producers’ surplus willing and able to spend.
• Producers’ willingness and ability to receive b. The minimum amount that producers are
• Market equilibrium willing and able to receive.
• Total social gain c. The consumers’ expenditure.
d. The total social gain at market equilibrium.
6.3 Activities 3. Explain how to find each of the following:
Getting Started a. The price P above which consumers will
purchase none of the goods or services.
1. Give the economic name by which we call each of
b. The shutdown point.
the following:
a. The function relating the number of items the c. The point of market equilibrium.
consumer will purchase at a certain price and the 4. Demand The following two figures, drawn from
price per item. the mathematician’s and the economist’s view-
b. The function relating the quantity of items the points, depict the same demand curve for a com-
supplier of the items will sell at a certain price modity. Consider p0 to be the current market price
and the price per item. of the commodity.
418 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
q q
Quantity Quantity
(thousands) (thousands)
50
20 40 S (p)
D(p)
15 30
10 20
5
p 10
p
Price Price
(dollars (dollars
10 20 30 10 20 30 40 50
per unit) per unit)
p b. Sketch the supply curve from the economist’s
Price viewpoint. Include the shutdown price and the
(dollars per unit) point (p0, q0) on the economist’s graph.
c. Depict (by shading on each figure) the region
30 representing producers’ revenue.
20 6. Supply The following two figures, drawn from
the mathematician’s and the economist’s view-
10 D(p)
q
points, depict the same supply curve for a commod-
Quantity ity. Consider p0 $3 per unit to be the current
5 10 15 20 (thousands) market price of the commodity.
q
a. Use p0 $20 per unit and the demand curve Quantity
(millions)
to estimate q0 on the figure showing the mathe-
matician’s view. Transfer the point (p0, q0) to 20
S (p)
the figure showing the economist’s view.
15
b. Depict (by shading on each figure) the regions
representing consumers’ expenditure and con- 10
sumers’ surplus. 5
p
c. Outline on each figure the region representing Price
the amount that consumers are willing and able (dollars
2 4 6
to spend. per unit)
d. Write a formula for the calculation of consumers’
willingness and ability to spend. p
Price
5. Supply The accompanying figure depicts a sup- (dollars per unit)
ply curve, drawn from the mathematician’s view- 6 S (p)
point, for a commodity. Consider p0 $25 per
unit to be the current market price of the com- 4
modity and p1 $10 per unit to be the shutdown
2
price. q
a. Use p0 and the supply curve to estimate q0 on the Quantity
5 10 15 20 (millions)
figure showing the mathematician’s view.
6.3 Integrals in Economics 419
a. Use p0 and the supply curve to estimate q0 on the figures. Also estimate the value of, and label on
figure showing the mathematician’s view. Trans- both figures, the shutdown price p1 and the
fer the point (p0, q0) to, and label the shutdown price P above which consumers will not pur-
price on, the figure showing the economist’s view. chase the product.
b. Depict (by shading on each figure) the regions b. Depict (by shading on each figure) the regions
representing producers’ willingness and ability representing consumers’ surplus and producers’
to receive and producers’ surplus. surplus.
c. Outline on each figure the region representing c. Outline on each figure the region representing
producers’ revenue. total social gain.
d. Write a formula for producers’ willingness and d. Write a formula for the calculation of total social
ability to receive. gain.
e. Write a formula for producers’ surplus.
7. Social Gain The following two figures, drawn Applying Concepts
from the mathematician’s and the economist’s 8. Demand The demand for wooden chairs can be
viewpoints, depict the same supply and demand modeled as
curves for a product.
D(p) 0.01p 5.55 million chairs
q
Quantity where p is the price (in dollars) of a chair.
(millions) a. According to the model, at what price will con-
25 sumers no longer purchase chairs? Is this price
D(p) S (p)
20
guaranteed to be the highest price any consumer
will pay for a wooden chair? Explain.
15 b. Find the quantity of wooden chairs that con-
sumers will purchase when the market price is
10
$99.95.
5
p c. Determine the amount that consumers are will-
Price ing and able to spend to purchase 3 million
(dollars wooden chairs.
1 2 3 4 5
per unit)
d. Find the consumers’ surplus when consumers
purchase 3 million wooden chairs.
p
Price 9. Demand The demand for ceiling fans can be
(dollars per unit) modeled as
5
D(p) 25.92(0.996p) thousand ceiling fans
4 S (p)
where p is the price (in dollars) of a ceiling fan.
3 a. According to the model, is there a price above
which consumers will no longer purchase fans? If
2
so, what is it? If not, explain why not.
1 D(p) b. Find the amount that consumers are willing and
q able to spend to purchase 18 thousand ceiling
Quantity fans.
5 10 15 20 25 (millions)
c. Find the quantity of fans that consumers will
a. On both figures, mark the equilibrium point purchase when the market price is $100.
and label it (p*, q*). Estimate the values of p* d. Find the consumers’ surplus when the market
and q* and label them on the axes in both price is $100.
420 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
10. Demand The demand for a 12-ounce bottle of c. Find the consumers’ surplus when the equilib-
sparkling water is given in the accompanying table. rium price for these lanterns is $12.34 per lantern.
12. Elasticity The demand for wooden chairs can be
Price Demand modeled as
(dollars per bottle) (million bottles)
D(p) 0.01p 5.55 million chairs
2.29 25
where p is the price (in dollars) of a chair.
2.69 9
a. Find the point of unit elasticity.
3.09 3
3.49 2 b. For what prices is demand elastic? For what
prices is demand inelastic?
3.89 1
4.29 0.5 13. Elasticity The demand for ceiling fans can be
modeled as
a. Find a model for demand as a function of price. D(p) 25.92(0.996 p) thousand ceiling fans
b. Does your model indicate a price above which where p is the price (in dollars) of a ceiling fan.
consumers will purchase no bottles of water? If
a. Find the point of unit elasticity.
so, what is it? If not, explain.
b. For what prices is demand elastic? For what
c. Find the quantity of water that consumers will
prices is demand inelastic?
purchase when the market price is $2.59.
d. Find the amount that consumers are willing and 14. Elasticity In Activity 10 part a, you were asked to
able to spend to purchase the quantity you found find a model for the demand for a 12-ounce bottle
in part c. of sparkling water.
e. Find the consumers’ surplus when the market a. Use the model found in Activity 10 to find the
price is $2.59. point of unit elasticity.
11. Demand The average daily demand for a new b. For what prices is demand elastic? For what
type of kerosene lantern in a certain hardware store prices is demand inelastic?
is as shown in the table.
15. Elasticity In Activity 11 part a, you were asked to
find a model for the average daily demand for a new
Average quantity type of kerosene lantern in a certain hardware store.
Price demanded
a. Use the model found in Activity 11 to find the
(dollars per lantern) (lanterns)
point of unit elasticity.
21.52 1
b. For what prices is demand elastic? For what
17.11 3 prices is demand inelastic?
14.00 5
16. Supply The willingness of saddle producers to sup-
11.45 7
ply saddles can be modeled by the following function:
9.23 9
0 thousand saddles if p 5
7.25 11 S(p)
2.194(1.295p) thousand saddles if p 5
a. Find a model giving the average quantity de- where saddles are sold for p thousand dollars.
manded as a function of the price. a. How many saddles will producers supply when
b. How much are consumers willing and able to the market price is $4000? $8000?
spend each day for these lanterns when the market b. At what price will producers supply 10 thousand
price is $12.34 per lantern? saddles?
6.3 Integrals in Economics 421
c. Find the producers’ revenue when the market Price per print Prints supplied
price is $7500. (hundred dollars) (hundreds)
d. Find the producers’ surplus when the market 5 2
price is $7500.
6 2.2
17. Supply The willingness of answering machine 7 3
producers to supply can be modeled by the follow- 8 4.3
ing function:
9 6.3
10 8.9
0 thousand answering machines if p 20
S(p) 0.024p2 2p 60 thousand a. Find a model giving the quantity supplied as a
answering machines if p 20 function of the price per print. Note: Producers
will not supply prints when the market price falls
where answering machines are sold for p dollars. below $500.
a. How many answering machines will producers b. At what price will producers supply 5 hundred
supply when the market price is $40? $150? prints?
b. Find the producers’ revenue and the producers’ c. Find the producers’ revenue and producers’
surplus when the market price is $99.95. surplus when the market price is $630.
20. Market Equilibrium The daily demand for beef
18. Supply The table shows the number of CDs that
can be modeled by
producers will supply at the given prices.
40.007
D(p) million pounds
1 0.033e0.35382p
Price per CD CDs supplied
(dollars) (millions) where the price for beef is p dollars per pound. Like-
wise, the supply for beef can be modeled by
5.00 1
7.50 1.5
0 million pounds if p 0.5
10.00 2 S(p)
51
15.00 3 million pounds if p 0.5
1 53.98e0.3949p
20.00 4
where the price for beef is p dollars per pound.
25.00 5
a. How much beef is supplied when the price is
$1.50 per pound? Will supply exceed demand at
a. Find a model giving the quantity supplied as a this quantity?
function of the price per CD. Note: Producers b. Find the point of market equilibrium.
will not supply CDs when the market price falls
21. Social Gain The average quantity of sculptures
below $4.99.
that consumers will demand can be modeled as
b. How many CDs will producers supply when the D(p) 1.003p2 20.689p 850.375 sculptures,
market price is $15.98? and the average quantity that producers will supply
c. At what price will producers supply 2.3 million can be modeled as
CDs?
0 sculptures when p 4.5
d. Find the producers’ revenue and producers’ sur-
plus when the market price is $19.99. S(p) 0.256p2 8.132p
250.097 sculptures when p 4.5
19. Supply The table shows the average number of
prints of a famous painting that producers will sup- where the market price is p hundred dollars per
ply at the given prices. sculpture.
422 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
a. How much are consumers willing and able to f. For what price will roses be sold at the equilibrium
spend for 20 sculptures? point?
b. How many sculptures will producers supply at g. What is the total social gain from the sale of roses
$500 per sculpture? Will supply exceed demand at market equilibrium?
at this quantity?
23. Social Gain The table gives both number of
c. Determine the total social gain when sculptures copies of a hardcover science fiction novel in de-
are sold at the equilibrium price. mand and the number supplied at certain prices.
22. Social Gain A florist constructs a table on the
basis of sales data for roses.
Price Books demanded Books supplied
(dollars per book) (thousands) (thousands)
Price of 1 dozen roses Dozens sold
20 214 120
(dollars) per week
23 186 130
10 190
25 170 140
15 145
28 150 160
20 110
30 138 190
25 86
32 128 210
30 65
35 52
a. Find an exponential model for demand given the
price per book.
a. Find a model for the quantity demanded.
b. Find a model for supply given the price per book.
b. Determine how much money consumers will be Note: Producers are not willing to supply any
willing and able to spend for 80 dozen roses each books when the market price is less than $18.97.
week.
c. At what price will market equilibrium occur?
c. If the actual market price of the roses is $22 per How many books will be supplied and demanded
dozen, find the consumers’ surplus. at this price?
Suppose the suppliers of roses collect the data d. Find the total social gain from the sale of a hard-
shown in the following table. cover science fiction novel at the market equilib-
rium price.
Price of 1 dozen roses Dozens supplied 24. Social Gain The table shows both the number of a
(dollars) per week certain type of graphing calculator in demand and
20 200 the number supplied at certain prices.
18 150
14 100
Price Calculators Calculators
11 80 (dollars per demanded supplied
8 60 calculator) (millions) (millions)
5 50 60 35 10
90 31 32
d. Find an equation that models the supply data. 120 15 50
Suppliers will supply no roses for prices below 150 5 80
$5 per dozen.
180 4 100
e. What is the producers’ surplus when the market
210 3 120
price is $17 per dozen?
6.4 Probability Distributions and Density Functions 423
a. Find a model for demand given the price per d. Find the producers’ surplus at market equilib-
calculator. rium.
b. Find a model for supply given the price per cal- e. Estimate the consumers’ surplus at market
culator. Note: Producers are not willing to supply equilibrium.
any of these graphing calculators when the mar- f. Estimate the total social gain from the sale of this
ket price is less than $47.50. type of graphing calculator at the market equilib-
c. At what price will market equilibrium occur? rium price.
How many calculators will be supplied and de-
manded at this price?
6.4 Probability Distributions and Density Functions
This section uses improper We have used derivatives to measure rates of change and integrals to measure accu-
integrals, which were discussed mulation of change. This section discusses another application of integrals—to meas-
in Section 6.1.
ure the likelihood that certain events will occur in situations that involve some degree
of chance or uncertainty. When all outcomes of a particular situation are considered,
the pattern indicated by the variability in data is called the distribution of the quan-
tity being studied. Consider, for instance, the distribution of scores for 2002 college-
bound seniors on the mathematics portion of the SAT that is given in Table 6.5. This
table shows the proportion of the 1,327,831 students taking the test whose math
scores fell in each indicated interval.
TABLE 6.5
Math SAT score x Number of students Proportion of students
(points) in score interval in score interval
200 x 300 32,593 0.02455
300 x 400 159,250 0.11993
400 x 500 381,339 0.28719
500 x 600 419,310 0.31579
600 x 700 251,650 0.18952
700 x 800 83,689 0.06303
(Source: “2002 College-Bound Seniors, A Profile of SAT Program Test Takers,” College
Entrance Examination Board and Educational Testing Service.)
This distribution of the SAT math scores can be viewed with a graph composed
of rectangles called a histogram. We choose to construct the histogram so that
the area of each rectangle is the proportion of students in the corresponding score
interval. The intervals must be such that no score is in more than one interval and
all possible scores are included. The height of each rectangle is calculated by divid-
ing the area of the rectangle by the width of the score interval. (See Table 6.6)
424 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
TABLE 6.6
Height of rectangle
Math SAT score x Width of rectangle Area of rectangle (proportion of students)
(points) (points) (proportion of students per point of score)
200 x 300 100 0.02455 0.00025
300 x 400 100 0.11993 0.00120
400 x 500 100 0.28719 0.00287
500 x 600 100 0.31579 0.00316
600 x 700 100 0.18952 0.00190
700 x 800 100 0.06303 0.00063
A histogram based on the SAT score groupings in Table 6.6 is shown in Figure 6.29.
Proportion of students
per point of score
0.0040
0.0030
0.0020
0.0010
Math
0
score
200 300 400 500 600 700 800
FIGURE 6.29
EXAMPLE 1 Reading and Using a Histogram
SAT Scores Refer to Table 6.5 and the histogram of math SAT scores in Figure 6.29.
a. Describe the distribution of the scores.
b. Find the total area enclosed by the rectangles in the histogram.
c. What proportion of scores is between 300 and 600?
Solution
a. The distribution peaks near 500, and most students taking the test had math
scores that ranged between 300 and 700.
b. The total area enclosed by the rectangles in the histogram equals 1:
Area 0.02455 0.11993 0.28719 0.31579 0.18952 0.06303 1
c. From the third column in Table 6.5, we find that the proportion of scores
between 300 and 600 is 0.11993 0.28719 0.31579 0.72. ●
6.4 Probability Distributions and Density Functions 425
Area as Probability
The probability of an event is a measure of how likely it is to happen. Because there
is no uncertainty associated with a known outcome, we assume that all variables for
which probabilities are determined are random variables—that is, variables whose
numerical values are determined by the results of an experiment involving chance.
We intuitively consider the probability of an event to be the proportion of times the
event occurs when the experiment associated with the event is repeated under simi-
lar conditions a large number of times.
To illustrate this concept, we return to the distribution of SAT I math scores given
numerically in Table 6.5 and graphically in the histogram in Figure 6.29. What is the
probability that a 2002 college-bound senior made a score between 400 and 600
on the mathematics portion of the SAT I? To answer this question, we first need to
understand the terms involved and how they are related to probability.
• “Taking the math portion of the SAT I” is the experiment that was repeated, un-
der similar conditions, a large number of times (1,327,831 students took the test).
• Assuming that the 2002 college-bound senior in the question is any of those tak-
ing the test, and assuming that we do not know any student’s individual score, the
random variable is the SAT I math score.
• The event being considered is the value of the SAT I math score lying in the
interval of values between 400 and 600.
We can now compute the probability that a student’s math score is between 400 and 600
by finding the proportion of scores that are in that score interval. From the information
in Table 6.5, the proportion of students whose scores are between 400 and 600 is 0.60298,
so the probability that a student’s math score is between 400 and 600 is 0.60298.
Probabilities are proportions, so they are real numbers between 0 and 1. Suppose
we want to know the probability that a student makes a math score between 200 and
800. Because all SAT math scores lie between 200 and 800, this event is known as a
certain or sure event, so its probability should be 1. This was verified in part b of
Example 1. What is the likelihood that a student makes an SAT math score of 950?
The highest score is 800, so the probability that a student makes an SAT math score of
950 is 0. An event that has a probability of zero is known as an impossible event.
The histograms used in this section have been designed so that all rectangle
heights are non-negative values, and the total area enclosed by the rectangles in each
histogram is 1. When expressed as a proportion, each rectangle area gives the proba-
bility that the value of the random variable is in the interval that forms the base of the
rectangle. For that reason, we also call these histograms probability histograms. The
above discussions suggest the following statements:
Probability
The probability that any event occurs is a real number between 0 and 1. The
probability of a certain (or sure) event is 1, and the probability of an
impossible event is 0. Even though probabilities are real numbers between
0 and 1, they are often referred to as percentages between 0% and 100%.
When the distribution of a random variable is represented by a
probability histogram, the probability that the value of the variable lies
within certain intervals is given by the sum of the areas of the rectangles
whose bases are those intervals.
426 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
If there are a large number of values very close together, then it often simpli-
fies matters to approximate their behavior with a continuous function. Such an
approximating function, shown in Figure 6.30, is overdrawn on a more de-
tailed “SAT math scores for 2002 college-bound seniors” histogram. This con-
tinuous function is the familiar bell-shaped curve commonly called the normal
distribution.
Proportion of students
per point of score
0.0040
0.0030
0.0020
0.0010
Math
0
score
200 300 400 500 600 700 800
FIGURE 6.30
Instead of summing areas of rectangles in a histogram, we can estimate
probabilities by finding areas under curves like the one shown in Figure 6.30. Such
functions are called probability density functions. We do not discuss the inter-
pretation of output values of probability density functions. Instead, we focus
on the interpretation of areas of regions beneath these functions. Because proba-
bility density functions describe how probabilities are distributed over various
intervals associated with random variables, they are also called probability
distributions.
Probability Density Function
A probability density function is a continuous function or a piecewise
continuous function with input consisting of some interval of real numbers
and with output satisfying the following two conditions:
1. Each output value is greater than or equal to 0.
2. The area of the region between the function and the horizontal axis is 1.
Before proceeding further, we introduce a notation to simplify writing probabil-
ity statements. We write the statement “the probability that the value of x is between
a and b” using the notation P(a x b). For instance, if x is the SAT math score,
then P(500 x 800) is read “the probability that the value of the SAT math score
is between 500 and 800.” Using this notation, we restate the definition of a probabil-
6.4 Probability Distributions and Density Functions 427
ity density function and define probability for random variables having such density
functions:
Probability and Probability Density Functions
A probability density function y f(x) for a random variable x is a
continuous or piecewise continuous function such that
1. f(x) 0 for each real number x, and
2.
f(x)dx 1
The probability that a value of x lies in an interval with endpoints a and b,
where a b, is given by
f(x)dx
b
P(a x b)
a
It is possible for the interval from a through b to be the entire set of real
numbers.
Probability density functions are constructed from experimental data and/or sta-
tistical theory using techniques you will probably study if you take a course in statis-
tics. This section is not intended to discuss those statistical techniques. Rather, we are
illustrating another use of integrals to find area under a curve.
EXAMPLE 2 Finding Probabilities from a Probability Density Function
Recovery Times Suppose that the proportion of patients who recover from mild
dehydration x hours after receiving treatment is given by
f(x) 12x2 12x3 when 0 x 1
0 when x 0 or x 1
a. What is the random variable in this situation?
b. Verify that f is a probability density function for this random variable.
c. Find the probability that the recovery time is between 42 minutes and 48 minutes.
Interpret your answer.
f d. Find the probability that recovery takes at least half an hour.
2
Solution
1.5
1 a. The random variable x is the number of hours until recovery. After 1 hour all
patients will have recovered.
0.5
x b. To verify that f is a probability density function for this random variable,
Hours we must show that the two properties of a probability density function are true.
0 1
FIGURE 6.31 1. As the graph in Figure 6.31 shows, all values of f(x) are non-negative.
428 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
2. The area under the graph of f is 1 because
(12x 12x )dx 0dx
0 1
f(x)dx 0dx 2 3
0 1
1
0 (4x3 3x4) 0 0
431
c. Because 42 minutes is 0.7 hour and 48 minutes is 0.8 hour, we find the probability
that recovery time is between 42 minutes and 48 minutes by computing
0.8
P(0.7 x 0.8) (12x2 12x3)dx
0.7
0.8
(4x3 3x4) 0.7
(2.048 1.2288) (1.372 0.7203)
0.1675
There is a 16.75% chance that the recovery time for a patient will be between
42 minutes and 48 minutes. (Note that 0.1675 is the area between the graph of
f and the horizontal axis from x 0.7 to x 0.8.)
(12x 12x )dx 0dx
1
d. P(x 0.5) f(x)dx 2 3
0.5 0.5 1
(4x3 3x4) 10.5
0.6875
There is a 68.75% chance that recovery takes at least a half hour. ●
It is important that you note the difference between the meanings of f(x), the out-
put of a probability density function, and P(a x b), the probability that the
random variable is within a certain interval. To see more clearly that the output of
a probability density function is not a probability, look again at the graph of the density
function shown in Figure 6.31. There are values of f(x) that are obviously larger than 1,
and we know that probabilities can never be greater than 1. Always keep in mind that
whereas f is a function that shows how the probabilities associated with values of a ran-
dom variable x are distributed over the input interval, P(a x b) is a real number
between 0 and 1. Figure 6.32 illustrates these ideas for a probability density function f.
f f
R P(a ≤ x ≤ b)
x x
a b
The area of R is 1. The area of the shaded region is
the probability that x is between a
and b.
FIGURE 6.32
6.4 Probability Distributions and Density Functions 429
The Uniform Probability Distribution
Even though probability density functions have a variety of shapes, some naturally
occur so often that they have names. Perhaps the simplest density function is one
that assumes a constant value between two specified inputs and is zero elsewhere.
This probability distribution is called the uniform density function and has the
equation
1
b-a
u
u(x)
0
1
ba
when a x b
when x a or x b
A graph of this piecewise continuous function is shown in Figure 6.33.
The uniform density function provides a good model for random variables that
are evenly distributed over an interval. For instance, the distribution of winning
x numbers picked since the beginning of a state lottery is approximated by a uniform
a b
density function. The probability distribution that describes how tires wear (in terms
A uniform density function of the remaining tread) on a properly aligned and balanced set of wheels is a uniform
FIGURE 6.33 density function.
EXAMPLE 3 Using a Uniform Density Function
Campus Bus Buses that transport students from one location to another on a large
campus arrive at the student parking lot every 15 minutes between 7:30 A.M. and
4:30 P.M. If t is the number of minutes before the next bus arrives at the lot, then the
distribution of waiting times is modeled by the density function
1
when 0 t 15
15
u(t)
0 when t 15 or t 0
a. Explain, in the context of this situation, why it makes sense for u(t) to equal 0
when t 15 or t 0.
b. Represent on a graph of u the probability that a student arriving at the parking
u lot will have to wait more than 5 minutes for the next bus.
c. Find P(t 5). Interpret the result.
1
15
Solution
a. If buses arrive every 15 minutes at the parking lot, no one will have to wait more
than 15 minutes for a bus. Because time cannot be negative, the waiting time
t
5 10 15
distribution is zero for t 0. It therefore makes sense that u(t) 0 when t is
not in the interval with endpoints 0 and 15.
P(t 5) is the area of the
shaded region. b. As shown in Figure 6.34, P(t 5) is the area of the shaded region between the
FIGURE 6.34 graph of u and the t-axis to the right of t 5.
430 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
c. Because P(t 5) 0, P(t 5) P(t 5). This probability is represented by
1
the area of the rectangle with height 15 and width 10. (See Figure 6.34.) Thus
10
P(t 5) 0.667
15
Of course, this probability could have also been found using an integral:
u(t)dt
0dt 151 (15 5) 1015 0.667
15 1
P(t 5) dt
5 5 15 15
When a student arrives at the parking lot between 7:30 A.M. and 4:30 P.M., there
is about a 66.7% chance that he or she will have to wait at least 5 minutes for a
bus to arrive. ●
Measures of Center and Variability
One important characteristic of a density function is its central value. The measure of
central value that we consider is the mean (commonly referred to as the average). The
mean, denoted by the Greek letter , is also called the expected value and is geomet-
rically associated with the “balance point” of the region between the density function
Another common measure of
and the horizontal axis.
the center of a distribution is
the median. The median is the Distributions can have the same mean but a completely different spread, or vari-
input value such that half of the ability, about that center. One measure of how closely the values of the distribution
area under the density function cluster about its mean is the standard deviation, denoted by the Greek letter . If
lies to the left and half to the
right of it.
most of the values of the input variable are close to the mean, then the standard
deviation is small. On the other hand, if it is likely that the input values are widely
scattered about the mean, the standard deviation is large.
EXAMPLE 4 Locating the Mean and Standard Deviation on a Graph
SAT Scores Refer to the continuous function that is overdrawn on the histogram of
the 2002 college-bound seniors’ SAT math scores in Figure 6.30 on page 426. The
mean math SAT score is 516 points, and the standard deviation is 114 points.* Locate
the mean and standard deviation on the graph of this continuous function that
approximates the distribution of the SAT math scores.
Solution
Because the mean is an input value for this continuous function, is located on the
horizontal axis. (See Figure 6.35.) The standard deviation tells us how the math scores
are spread out around the mean. For instance, one standard deviation to the right of
the mean would be 516 114 630 points.
*“2002 College-Bound Seniors, A Profile of SAT Program Test takers,” College Entrance Examination
Board and Educational Testing Service.
6.4 Probability Distributions and Density Functions 431
Proportion of students
per point of score
0.0040
0.0030
0.0020
σ
0.0010
µ µ+σ
Math
0
score
200 300 400 500 600 700 800
FIGURE 6.35 ●
The mean and standard deviation of a known density function are determined using
the following definitions:
Mean and Standard Deviation
For the density function y f(x), with x defined on the interval of real
numbers,
• the mean is
xf(x)dx
• the standard deviation is (x )2f(x)dx
provided that the integrals exist.
EXAMPLE 5 Computing a Mean and Standard Deviation
Campus Bus Refer to the distribution of waiting times for campus transportation
in Example 3:
1
when 0 t 15
15
u(t)
0 when t 15 or t 0
where t is the time in minutes until the next bus arrives.
a. Find the mean waiting time.
b. Find the standard deviation of the waiting times.
432 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
Solution
t u(t)dt
0 15
a. t u(t)dt t u(t)dt t u(t)dt
0 15
0 115
t 0dt tdt t 0dt
0 15 15
1 2 15 1
0 t 0 (225 0) 7.5
30 0 30
The average time spent waiting for a bus is 7.5 minutes.
b.
(t )2u(t)dt
0
(t 7.5)2 0dt
0
15
(t 7.5)2
1
15
dt 15
(t 7.5)2 0dt
0
1
15
15
0
(t 7.5)2dt 0 1 (t 7.5)3
15
3
15
1 (15
15 3
7.5)3
(0
3
7.5)3
18.75 4.33
The standard deviation of waiting times is about 4.33 minutes. ●
As illustrated in Example 5, a definite integral involving a density function f has a
value of 0 if the height of the density function is 0. In these cases, we are finding the
area of a rectangle with height 0 and width determined by the upper and lower limits
on the integral:
Whenever f(x) 0, area (height)(width) (0)(width) 0.
Therefore, from this point on, we do not show the integrals involved in probability,
mean, or standard deviation calculations for the portions of the horizontal axis where
the value of a density function is 0.
The Exponential Probability Distribution
The amount of time it takes to learn a task, the duration of a phone call, the
E time you wait for service at a bank teller’s window, the time between arrivals at
the drive-through station of a fast-food restaurant—all are examples of random
k variables that are more likely to be small than large. With events such as these,
we are interested in the time elapsed or space between any two occurrences of
an event, rather than in the number of times the event happens. The likelihood
of encountering certain intervals of time or space between consecutive occurrences
of an event can be modeled by an exponential density function, with the general
x
0 formula
An exponential density
function
E(x) kekx when x 0
0 when x 0
FIGURE 6.36 where k is some positive constant. The graph is shown in Figure 6.36.
6.4 Probability Distributions and Density Functions 433
1
The mean of the exponential distribution is k. When, on average, an event occurs at
1
a rate of k arrivals per unit of time, the average gap between consecutive arrivals is k time
unit. For instance, suppose that you arrive at your local Burger King sometime between
11:30 A.M. and 12:30 P.M. and that the time you have to wait for service while in line at
the drive-through window is an exponential random variable with a mean of 2.5 min-
utes. This means that there are 2.5 minutes between cars coming into the line, with cars
arriving at the rate of 0.4 car per minute (that is, 4 cars every 10 minutes). When ap-
plying this distribution, remember that the values of the random variable x and the
value of k must be in the same time units.
EXAMPLE 6 Writing and Using an Exponential Density Function
ER Arrivals The distribution of the time between successive arrivals at an emergency
room of a large city hospital on Saturday nights can be approximated by an exponen-
tial density function. Two patients arrive at the emergency room every 10 minutes.
a. What is the equation for this exponential density function?
b. What is the probability that the time between successive arrivals will be more
than 1 minute? Interpret this result.
Solution
a. The mean of the distribution is the time between consecutive events. Therefore,
1 10 1
the mean is k 2 5 minutes per arrival. Thus k 5, the average
number of arrivals per minute. The equation of the exponential density
function is
1 t/5
E(t) e 0.2e0.2t when t 0
5
where t is the time, in minutes, between successive arrivals. (In the context of
time, it is assumed that E(t) 0 when t 0.)
T
b. P(t 1) 0.2e0.2t dt lim 0.2e0.2t dt
1 T→ 1
lim [e0.2T (e(0.2)(1))]
T→
0 e0.2 0.819
There is an 81.9% chance that the time between successive arrivals will be more
than 1 minute. This event is likely to occur. ●
The Normal Distribution
We now turn our attention to the single most important density function in statistics,
the normal density function, which is also called the normal distribution. The equa-
tion of the normal density function with mean and standard deviation is
(x ) 2
1
f (x) e 2 2 where x
2
434 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
The graph of a normal distribution is called a normal curve. The mean controls the lo-
cation of a normal curve, and the standard deviation controls its spread. Figure 6.37
gives us a good idea of some of the features of any normal distribution.
σ σ
x
µ-σ µ µ+σ
A normal curve
FIGURE 6.37
Note these important properties of any normal curve:
• The curve is bell-shaped with the absolute maximum occurring at the mean .
• The curve is symmetric about a vertical line through .
• The curve approaches the horizontal axis but never touches or crosses it.
• The inflection points occur at and .
EXAMPLE 7 Comparing Normal Curves
A
Consider the two normal curves shown in Figure 6.38.
a. Compare the means of the two normal distributions.
b. One of the curves has 3, and the other curve has an inflection
point at x 10. Compare the spreads of the two curves.
c. What percentage of the total area under each normal curve lies to
B
the right of x 8?
x
0 2 4 6 8 10 12 14 16 Solution
FIGURE 6.38 a. The means of the two density functions are the same, because each
curve has a maximum at x 8.
b. From Figure 6.38, we see that the rightmost inflection point on normal curve A
occurs around 10. Thus 8 10, and the standard deviation of curve A is
approximately 2. Curve B is more spread out from its center than curve A, so it
has the larger standard deviation, 3.
c. Because a normal curve is the graph of a density function, the area between
each normal curve and the horizontal axis must be 1. Each normal curve is
symmetric about its mean, so 50% of the total area is to the right of for each
normal curve. ●
The normal curve rises gradually to its absolute maximum and then decreases in
a symmetric manner. Although it is not unique in exhibiting this form, it has been
6.4 Probability Distributions and Density Functions 435
found to provide a reasonable approximation to certain other distributions that oc-
cur in many real-life situations. Also, many naturally occurring phenomena, as well
as various mental and physical characteristics of human beings, can be described by
normal distributions.
EXAMPLE 8 Using a Normal Density Function
Light Bulb A manufacturer of light bulbs advertises that the average life of these
bulbs is 900 hours with a standard deviation of 100 hours. Suppose the distribution
6.4.1
of the length of life of these light bulbs, with the life span measured in hundreds of
hours, is modeled by a normal density function.
a. Write the definite integral that represents the probability that a light bulb lasts
between 900 and 1000 hours.
b. Approximate the value of the integral in part a, and interpret the result in context.
Solution
a. Note that the input for the normal density function is measured in hundreds of
10
hours. The integral that gives P(9 x 10) is f (x)dx, where f is the normal
9
density function with 9 and 1. That is,
10 1 (x 9)
2
P(9 x 10) e 2 dx
9 2
b. There is no antiderivative formula that can be used to find the exact value of an
integral of the normal density function. However, the value of the definite
integral can be numerically approximated using technology:
10 1 0.5(x9)2
e dx 0.34
9 2
Thus the chance that any one of these light bulbs will last between 900 and 1000
hours is about 34%. ●
Cumulative Density Functions
Suppose that the length of time a student waits for a computer terminal with which
to preregister for next term’s classes is described by the uniform density function
f
1
when 0 x 25
1 25
f(x)
25
0 when x 25
x
where the value of the random variable x is measured in minutes. The probability that
0 5 15 25
Minutes
the waiting time is less than or equal to 5 minutes is found by computing the area of
FIGURE 6.39 the shaded region in Figure 6.39.
436 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
TABLE 6.7
Time interval Probability that waiting time is in time interval
Between 0 and 2 minutes P(0 x 2) 0.08
Between 0 and 5 minutes P(0 x 5) 0.20
Between 0 and 12.5 minutes P(0 x 12.5) 0.5
Between 0 and 20.38 minutes P(0 x 20.38) 0.8152
Between 0 and 25 minutes P(0 x 25) 1
Between 0 and 33 minutes P(0 x 33) P(0 x 25) P(25 x 33) 1 0 1
Between 0 and 40.2 minutes P(0 x 40.2) P(0 x 25) P(25 x 40.2) 1 0 1
Because the event that the waiting time is more than 25 minutes is impossible by the
definition of this density function, P(x 25) 0. Also, P(0 x a), when a is a
waiting time between 0 and 25 minutes, is computed by finding the area of a rectangle
1
with height 25 and width a. Table 6.7 gives the results of some of these probability
computations.
P(0 ≤ x ≤ t) P(0 ≤ x ≤ t)
P(0 ≤ x ≤ t) = 1
1 1
0.8 0.8
0.6 0.6
t
P(0 ≤ x ≤ t) =
0.4 0.4 25
0.2 0.2
0 t 0 t
0 10 20 30 40 0 10 20 30 40
(a) (b)
FIGURE 6.40
Note that when the waiting time x is between 0 and 25 minutes, the probability
that a student waits between 0 minutes and t minutes seems to be a function of the
upper endpoint of the interval. Also, the probability that a student waits between 0
minutes and t minutes, where t 25, is 1 because 25 minutes is the maximum wait-
ing time. A scatter plot of the points (t, P(0 x t)) that appear in Table 6.7 is
shown in Figure 6.40a. If all probabilities P(0 x t), where 0 t , were com-
puted and plotted, then the graph shown in Figure 6.40b would result.
A function that shows how probabilities accumulate as the value of the random
variable increases is called a cumulative density function. For instance, if y F(x) is
a cumulative density function for waiting times, where x is measured in minutes, then
the output F(5) 0.2 represents the probability that the waiting time is less than or
equal to 5 minutes. That is, F(5) P(x 5). Similarly, F(12.5) P(x 12.5), and
6.4 Probability Distributions and Density Functions 437
so forth. In general, for a random variable x with probability density function f, where
c is some constant, the cumulative density function of f at c is
c
F(c) P(x c) f (x)dx
The left end behavior of any cumulative density function will be 0, corresponding to
impossible events, and the right end behavior of any cumulative density function will
be 1, corresponding to sure events. Cumulative density functions are always nonde-
creasing. Outputs of cumulative density functions are the areas between the corre-
sponding probability density functions and the horizontal axis.
Does the link between probability density functions and cumulative density func-
tions sound familiar? It should—it is the same as the relationship between functions
and their accumulation functions, which we discussed in Chapter 5.
Cumulative Density Function
The cumulative density function for a random variable x defined on the
interval of real numbers with probability density function f is
x
F (x) f (t)dt for all real numbers x
For any value of the random variable x, say c, F(c) P(x c). The
cumulative density function F is an accumulation function of the probability
density function f.
EXAMPLE 9 Graphing a Cumulative Density Function
The graph in Figure 6.41 shows the probability density function for a random
variable x.
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
x
FIGURE 6.41 0 1 2 3 4
a. Sketch the graph of the cumulative density function for x.
b. Express P(2 x 3.4) using the probability density function f.
c. Express P(2 x 3.4) using the cumulative density function.
438 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
Solution
a. Because f (x) 0 for x 1, no area is accumulated when x 1. Thus
f (x) 0 when x 1. Note too that the accumulation of nonzero area under
the graph of f begins at x 1. Also, all accumulated area is positive because the
graph of f is above the horizontal axis. As x moves to the right of the starting
point (x 1), area is accumulating faster and faster. Thus the cumulative den-
sity function is increasing and concave up until x 4. At this point, no more
F
area is accumulated, because the value of the density function is once again 0.
1 Because f is a probability density function, the total accumulated area must be 1.
4
Thus F(4) f (x)dx 1. No area is added or subtracted past x 4, so
F(x) 1 for x 4. Combining all of this information, we can draw a possible
graph of the cumulative density function F. (See Figure 6.42.)
3.4
x b. P(2 x 3.4) f(x)dx
0 1 2 3 4 2
FIGURE 6.42 c. P(2 x 3.4) F(3.4) F(2) ●
The Fundamental Theorem guarantees the equivalence of the results of parts b
and c of Example 9. It also tells us that the derivative of a cumulative density function
is the corresponding probability density function at all points where the derivative
d
exists. That is, dx [F (x)] f (x).
EXAMPLE 10 Finding Probabilities from a Cumulative Density Function
Temperature The graph in Figure 6.43 shows the cumulative density function for
the distribution of temperatures in a southwestern city during a 24-hour period in
May. The random variable x measures the temperature recorded in degrees Fahren-
heit (beginning at midnight), and the output T(x) is the proportion of the time that
the temperature is less than or equal to x°F.
T
1
0.8
0.6
0.4
0.2
x
0
70 75 80 85 90 95 100 °F
FIGURE 6.43
6.4 Probability Distributions and Density Functions 439
a. What proportion of the time do you expect the temperature to be at most 80°F?
Interpret this result in a probability context.
b. Estimate the high and low temperatures.
c. Estimate the probability that the temperature will be above 90°F.
d. Sketch a graph of the probability density function for the distribution of May
temperatures in this location.
Solution
a. The proportion of the time that the temperature will be at most (less than or
equal to) 80°F is P(x 80) T(80) 0.41. In a southwestern city during any
24-hour period in May, the temperature will be less than or equal to 80°F about
41% of the time.
b. T(x) appears to be 0 at approximately 74°F. Because T(x) is a proportion, it can-
not be negative. Thus, the cumulative proportion to the left of 74°F must also
be 0, and the minimum temperature is 74°F. The cumulative probability has a
maximum of 1 at a temperature of about 98°F, so the maximum temperature is
approximately 98°F.
c. The temperature on any day must be either less than 90°F or greater
than or equal to 90°F. Thus the probability that the temperature is less
than 90°F added to the probability that the temperature is 90°F or greater must
equal 1.
T(90) P(x 90) 1
P(x 90) 1 T(90)
1 0.63
0.37
d. We know that the probability density function is the slope function of the
T' cumulative density function. Note that because the graph of T is always in-
creasing for temperatures between 74°F and 98°F, its slope graph is positive
over this interval. (Recall that no output of a density function can be negative.)
Also, the graph of T appears to have an inflection point (point of least slope) lo-
cated at approximately 87°F, so there is a minimum on the slope graph at this
temperature.
Using 74°F as the minimum temperature and 98°F as the maximum tempera-
ture, we know that P(x 74) and P(x 98) are both 0. Thus the value of the
x probability density function for temperatures less than 74°F and greater than
70 75 80 85 90 95 100 °F 98°F is 0.
FIGURE 6.44 A possible graph of the probability density function is shown in Figure 6.44. ●
The connection between probability density functions and cumulative density
functions lies at the very heart of calculus, for it is the relationship between a function
and its accumulation function.
440 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
a. The amount of cereal put into a box is
6.4 Concept Inventory normally distributed with 15 ounces and
1.6 ounces .
• Probability
b. The height of a certain species of plant is nor-
Histogram mally distributed with a mean of 15 inches and a
Random variable standard deviation of 5 inches.
Notation, computation, and interpretation c. You use a random number generator to choose a
• Probability density function number between 0 and 1. The random variable x
• Probability distribution is the number that is chosen.
Uniform probability distribution d. The time, in minutes, that a customer waits to
Exponential probability distribution pay for items at a department store is exponen-
Normal distribution tially distributed.
Mean and standard deviation
3. Which of the following could be probability density
• Cumulative density function functions? Explain.
6.4 Activities
a. f (x) 1.5(1 x 2) when 0 x 1
0 when x 0 or x 1
Getting Started b. h(x) 6(x x 2) when 0 x 1
0 when x 0 or x 1
1. Interpret each of the following probability state- c. r
ments in the context of the given situation.
1.2
a. Suppose the random variable x is the length, in
minutes, of a telephone call made on a computer
software technical support line. Interpret t
P(x 5) 0.46. 0 0.5 1 1.5 2
b. Suppose the random variable x is the distance in
d. s
feet between cars on a certain two-lane highway.
Interpret P(x 7) 0.25.
c. The probability that New Orleans will receive be-
tween 2 and 4 inches of rain during the month of c
March is 0.15. 0
1
2. Figures a through d show the shapes of some prob- 4. Which of the following could be probability density
ability distributions. Match each given situation to a functions? Give reasons.
possible graph of its density function. Give reasons
for each of your choices. 3x(1 x 2) when 0 x 1
a. g(x)
0 when x 0 or x 1
b. h(y) 0.625e1.6y when y 0
0 when y 0
0 15 16.6 c. g
(2, 1)
(a) (b) 1
0.75
0.50
(1, 0.50) (3, 0.25)
0.25
t
0 1 2 3
0 1 10 15
(c) (d)
6.4 Probability Distributions and Density Functions 441
d. f Suppose the waiting times follow an exponential
density function.
0.025
a. Why would customers prefer the waiting times at
the grocery checkout to follow an exponential
distribution rather than a uniform distribution
x with the same mean?
0 40 80
b. Write the equation for the exponential distribu-
tion of waiting times. Graph the equation and lo-
Applying Concepts
cate the mean waiting time on the graph. Does
5. Sales Let x represent the amount of frozen yogurt 2.5 seem reasonable for the graph?
(in hundreds of gallons) sold by the G&T restaurant c. What is the likelihood that a customer waits less
on any day during the summer. Storage limitations than 2 minutes to check out?
dictate that the maximum amount of frozen yogurt d. What is the probability of waiting between 2 and
that can be kept at G&T on any given day is 250 gal- 4 minutes to check out?
lons. Records of past sales indicate that the proba-
bility density function for x is approximated by e. What is the probability of waiting more than
y(x) 0.32x for 0 x 2.5. 5 minutes to check out?
a. What is the probability that on some summer 9. ER Arrivals Consider the exponential density
day, G&T will sell less than 100 gallons of frozen function discussed in Example 6:
yogurt? e(t) 0.2e0.2t when t 0
b. What is the mean number of gallons of frozen yo-
gurt that G&T expects to sell on a summer day? where t is the time in minutes between successive
arrivals at an emergency room in a large hospital on
c. Sketch a graph of y, and locate the mean on the Saturday nights.
graph. Also, shade the region that the answer to
part a represents. a. Find the probability that successive arrivals are
between 20 and 30 minutes apart.
6. Waiting Time Suppose that a traffic light on your b. Find the probability that 10 minutes or less
campus remains red for 30 seconds at a time. You elapses between successive arrivals.
arrive at that light and find it red. Assume that your
waiting time t (in seconds) at the light follows a uni- c. Find the probability that successive arrivals will
form density function y u(t). be more than 15 minutes apart.
a. Why does it make sense that u(t) 0 for t 0 10. a. Graph the exponential density function for
and t 30? k 0.5, 1, 1.5, 2, and 4.
b. Sketch a graph of u. b. Comment on how changing the value of k affects
c. Find your chances of waiting at least 10 seconds the shape of the graph.
at the red light. What does this value represent c. How does the mean of the exponential density
on the graph of u? function change as k increases?
d. Find the probability of waiting no more than 20 11. Learning Time The manufacturer of a new board
seconds at the red light. game believes that the time it takes a child between
e. What is the average time you would expect to the ages of 8 and 10 to learn the rules of its new
wait at the light? board game has the probability density function
7. Explain, using the definition of a density function, 3
(4t t2) when 0 t 4
why the definite integral calculation of a probability 32
P(t)
must result in a value between 0 and 1.
8. Waiting Time At a certain grocery checkout 0 when t 4
counter, the average waiting time is 2.5 minutes. where t is time measured in minutes.
442 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
a. Find the mean time that it takes a child age 8 to a. Find where the probability density function for
10 to learn the rules of the game. the number of customers who require daily
b. Find the standard deviation of the learning times. ATM service at this bank is increasing the
fastest.
c. Find P(0 t 1.5). Interpret this result.
b. Give a specific reason why it would benefit a
d. Find P(t 3). Interpret this result.
bank to know the probability distribution of its
12. Weight Suppose the weight of pieces of passen- customers who are served daily by the ATM
ger luggage for domestic airline flights follows a nor- machines.
mal distribution with 40 pounds and c. Sketch a graph of this normal distribution. In
10.63 pounds. parts i, ii, and iii, shade on the graph the area rep-
a. Find the probability that a piece of luggage resenting the probability. Remembering that the
weighs less than 45 pounds. area to the left (or right) of the mean of any nor-
b. Find the probability that the total weight of the mal distribution is 0.5, find the likelihood that,
luggage for 80 passengers on a particular flight is on a particular day,
between 1200 and 2400 pounds. (Assume each i. between 150 and 200 customers require
passenger has one piece of luggage.) service at the ATM machines.
c. Find the probability that the total weight of the ii. fewer than 220 customers require service.
luggage for 125 passengers on a particular flight iii. more than 235 customers require service.
is more than 5600 pounds. (Assume each
passenger has one piece of luggage.) 14. SAT Scores The accompanying figures are proba-
bility histograms of math SAT scores for 2002 male
d. Find where the probability density function for
and female college-bound seniors, respectively. All
the weight of passenger luggage is decreasing
scores are based on a recentered scale. (See Activity
most rapidly.
15.) Overdraw, on the histograms, continuous
13. Customers The number of customers served daily curves representing normal density functions that
by the ATM machines for a certain bank follows a approximate the distributions of scores. Discuss any
normal distribution with a mean of 167 customers similarities and/or differences between the two
and a standard deviation of 30 customers. normal distributions.
Proportion of Proportion of
male students female students
per point of score per point of score
0.004 0.004
0.003 0.003
0.002 0.002
0.001 0.001
Math Math
0 0
score score
200 300 400 500 600 700 800 200 300 400 500 600 700 800
(a) (b)
Figures for Activity 14
(Source: Based on data from “2002 College-Bound Seniors, A Profile of SAT Program Test Takers,”
College Entrance Examination Board and Educational Testing Service.)
6.4 Probability Distributions and Density Functions 443
15. SAT Scores For all test dates on or after April 1, and bell-shaped (in particular, for a normal distri-
1995, SAT Reasoning Test scores have been re- bution), the Empirical Rule states that
ported on a new, “recentered” scale. Over the years,
• Approximately 68% of the data values lie
the average score on the math portion of the SAT
between and .
moved away from 500, the midpoint of the original
200-to-800 scale. This change reestablished the • Approximately 95% of the data values lie
average score near the midpoint of the scale and between 2 and 2.
realigned the verbal and math scores so that a stu-
dent with a score of 450 on each test can conclude • Approximately 99.7% of the data values lie
that his or her math and verbal scores are equal. The between 3 and 3.
previous scales showed the average verbal score a. Verify the statements in the Empirical Rule for
to be about 425 and the average math score to be the normal probability density function with
about 475, which made comparison between the 5.3 and 8.372.
two difficult. b. Estimate P(11.444 x 13.672) using the
(Source: Used by permission from Peterson’s Guide to Four- Empirical Rule if x has a normal probability
Year Colleges, 1997, 27th edition. Princeton, NJ: Peterson’s
distribution with 5.3 and 8.372.
Guides, Inc., 1996. © 1996 by Peterson’s, Princeton, NJ.)
c. Use the normal probability density function to
a. If the interval between 200 and 800 included all
find the probability in part b.
scores within three standard deviations of the
mean score on the original scale, what was the 17. Test Scores Scores on a 100-point final exam ad-
standard deviation of the original math SAT ministered to all calculus classes at a large university
distribution? are normally distributed with a mean of 72.3 and a
b. Is the realigned mean score for verbal scores standard deviation of 28.65. What percentage of
more or less than 425? Is the realigned mean students taking the test made
score for math scores more or less than 475? a. a score between 60 and 80?
Explain.
b. a score of at least 90?
c. Most standardized test scores follow a nor-
c. a score that was more than one standard deviation
mal distribution. Using the fact that the proba-
away from the mean?
bility of a score falling in a particular interval is
the same as the percentage of students expected d. At what score was the rate of change of the prob-
to score in that interval, determine what per- ability density function for the scores a maxi-
centage of students were expected to make a mum?
math score of at least 475 under the “old” score 18. Another measure of the center of a probability dis-
scale.
tribution is the median. The median is the value m
f (x)dx f (x)dx, where f (x) 0
d. Assuming that the SAT math and verbal scores m b
follow a normal distribution, do you have enough such that
a m
information to draw a graph of the density func- for a x b.
tion for either of the recentered SAT scores?
a. Refer to the distribution of waiting times in
e. Do you think that recentering the SAT scores Example 5:
moved only the mean of the distribution or did it ⎧ 1
also change the standard deviation? Give reasons u(t) ⎨ 15 when 0 t 15
for your answer. ⎩ 0 when t 15
f. Why do the recentered “higher” scores not where t is the time until the next bus. Find the
translate into improved performance? median time to wait for the next bus.
16. As we have previously noted, a quick approxima- b. How does the median compare to the mean for a
tion is sometimes useful when an exact answer is probability distribution that is symmetric about
not required. For a distribution that is symmetric a vertical line drawn through its mean?
444 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
c. For the density function y f (x), where U.S. population
f (x) 0 for a x b, explain why the per year of age
statement (percent)
“The median is the value m such that
1.5
m 1
f (x)dx ”
a 2 1.0
is equivalent to saying 0.5
“The median is the value m such that
Age
f (x)dx f (x)dx”.
m b 0.0
0 10 30 50 70 90 110 125 (years)
a m
(Source: Based on data from the U.S. Bureau of the Census.)
d. Verify that the median of the general exponential
ln 2 a. Sketch the graph of the corresponding cumulative
density function is x k . density function.
19. Verify the following statements for the uniform b. Locate, on the graph of the cumulative density
density function function you sketched in part a and on the graph
1 of the probability density function, the probabil-
when a x b ity that a person is between 20 and 40 years old.
u(x) b a
0 when x a or x b
22. Suppose that y f(x) is the exponential density
ab function with k 2.
a. The mean is 2 .
ba a. Find F, the corresponding cumulative density
b. The standard deviation is . function.
12
c. The cumulative density function is b. Use both f and F to find the probability that
x 0.35.
0 when x 0 c. Use F to find the probability that x 0.86.
xa
F(x) ba when a x b
d. Sketch graphs of f and F.
1 when x b
23. Consider the density function
20. The graph of a cumulative density function is
2x when 0 x 1
shown. Sketch the graph of the corresponding prob- f(x)
0 when x 0 or x 1
ability density function if the input set for both
functions is all real numbers. a. Find F, the corresponding cumulative density
function.
F
b. Use both f and F to find the probability that
1 x 0.67.
c. Use F to find the probability that x 0.25.
d. Sketch graphs of f and F.
24. Population In mid-1992 the U.S. resident popu-
lation was about 255 million, with only approxi-
0 x
mately 45,000 persons aged 100 or older. The
accompanying table gives the distribution of ages of
21. Population The graph shows a probability distri- those persons less than 100 years old. F(x) is the
bution of the United States population at the time of percentage of U.S. residents less than x years of age.
the 2000 census. (Assume that no one is more than 100 years old.)
6.4 Probability Distributions and Density Functions 445
a. Fill in the value for F(0). g. Use the appropriate F(x) estimates calculated in
b. If possible, give the units of the corresponding part b and five midpoint rectangles to estimate
100
probability density function. Use symmetric dif- xF(x)dx. What does this integral represent,
ference quotients to estimate F(x) at x 10, 25, 0
30, 35, 45, 50, 70, and 90 years of age. and what are its units?
c. The accompanying scatter plot shows some ap- 25. Income The accompanying table shows income
proximate values of the continuous density func- intervals (in thousands of dollars) and the percent-
tion. Label the axes and draw a smooth curve age of Texas households with 1990 income in each
through the points so that there are no more con- interval. For instance, 9.6% of Texas households
cavity changes in your curve than those indicated had income greater than or equal to $5000 but less
by the given points. What is the demographic than $10,000 in 1990.
significance of the large bump on the graph?
Income interval Percent of
x F(x) x F(x) (thousands of dollars) households
(years) (percent) (years) (percent) 0–5 8.2
0 55 79.2 5–10 9.6
5 7.6 60 83.3 10–15 9.8
10 14.8 65 87.4 15–25 18.8
15 21.9 70 91.3 25–35 15.8
20 28.6 75 94.7 35–50 16.6
25 36.1 80 97.2 50–75 13.3
30 44.0 85 98.8 75–100 4.3
35 52.8 90 99.7
(Source: U.S. Bureau of the Census.)
40 61.1 95 99.9
45 68.4 100 100.0 a. Complete the following table, which shows the
50 74.5 cumulative percent F(x) of households with
incomes less than x thousand dollars.
(Source: Based on data from the U.S. Bureau of the Census.)
x F(x) x F(x)
1.8
5 35
1.4
10 50
1.0
15 75
0.6
0.2 25 100
0 10 20 30 40 50 60 70 80 90 100 b. Comment on the value F(100). Why is it not
d. Use the cumulative density function F to find the 100%?
percentage and number of 1992 U.S. residents c. Examine a plot of the data in part a. Does it ap-
who were age 20 or more but under age 50. pear that the continuous density function f
e. Use the density function F to write an integral would have an inflection point or limiting value?
expression for the percentage of residents who d. An approximate model for the density function is
were age 20 or more but under age 50.
f (x) 0.85403395x 0.56e0.0437980796x
f. Use three midpoint rectangles and the appropri- percentage points per thousand dollars
ate F(x) estimates from part b to estimate the
value of the integral in part e. Sketch the rectan- where x is household income in thousands
gles you use on the scatter plot in part c. of dollars. Examine a graph of this model on the
446 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
plot of density function estimates that you c. Find an exponential model for A(x).
constructed in part c. Comment on the fit. d. Use the model from part c to write a model for
e. Use the equation in part d to estimate the per- the distribution F, and then write a model for the
centage of households with 1990 annual income density function F.
xF(x)dx represent,
less than $100,000. Compare your result with
F(100) from part a. e. What does the integral
0
f. Use the equation in part d to estimate the mean and what are its units?
1990 income for Texas households with incomes
less than $100,000. 27. Dinosaurs In Michael Crichton’s novel Jurassic
26. Accidents A study of serious accidents in British Park, dinosaur clones are alive and roaming about a
coal mines focused on the time interval between suc- remote jungle island intended to be a theme park. All
cessive accidents. If, for example, successive accidents the dinosaurs have been cloned female so that the
were observed on August 2 and August 5, then the populations can be controlled in Jurassic Park. Ian
time interval between these accidents was recorded as Malcolm, a cynical mathematician who is invited to
3 days. During the study period, there were 34 serious the island, finds one of the first clues that all is not
accidents separated by 33 time intervals. The table well when he examines one of the graphs given in the
gives some of the data. A(x) is the number of time following figure. Both graphs show height distribu-
intervals that were greater than or equal to x days. tions of the “compy” (Procompsognathid). (The
park’s computer that produced the graphs con-
x A(x ) x A(x ) structed them with straight lines connecting the data
rather than using smooth curves.)
1 31 13 7
(Source: Michael Crichton, Jurassic Park, New York: Knopf,
4 19 16 4 1990.)
7 13 19 3 a. Malcolm claims that one of the graphs in the
10 9 figure is characteristic of a breeding population
and that the other graph is what would be
(Source: B. A. Maguire, E. S. Person, and A. W.
Wynn, “The Time Interval Between Industrial expected from a controlled population in which
Accidents,” Biometrika, vol. 39, 1952, 168–180.) the compys were introduced in three batches
at six-month intervals. Which distribution
a. Find A(0). Was there a day on which two acci- corresponds to which population?
dents occurred? Was there a day on which more
b. Which graph did Malcolm first see that indicated
than three accidents occurred?
something was amiss?
A(x)
b. Explain why F(x) 1 33 is the distribution c. Are the height distribution graphs in the fig-
of the time intervals observed between successive ure graphs of probability density functions?
accidents. Explain.
Animals Animals
10 10
8 8
6 6
4 4
2 2
Height Height
0 0
26 28 30 32 34 36 38 40 (cm) (cm)
25 27 29 31 33 35 37 39 41
(a) (b)
Height distribution (Procompsognathids)
Figure for Activity 27
6.4 Probability Distributions and Density Functions 447
SUMMARY
Improper Integrals calculated, the present value is determined by solving for
f (x)dx,
a P in the formula
Improper integrals of the forms f (x)dx,
a r D
P 1 future value
and f (x)dx can be evaluated by substituting a con- n
stant for each infinity symbol, finding an antiderivative Streams also have applications in biology and related
and evaluating it to determine an expression in terms of fields. The future value (in b years) of a biological stream
x and the constant(s), and then determining the limit of with initial population size P, survival rate s (in
the resulting expression as the constant approaches in- decimals), and renewal rate r(t), where t is the number
finity or negative infinity. If the limit does not exist, we of years of the stream, is
r(t)s
say the integral diverges. b
Future value Ps b bt dt
0
Streams in Business and Biology
An income stream is a flow of money into an interest-
bearing account over a period of time. If the stream Integrals in Economics
flows continuously into an account at a rate of R(t) dol- A demand curve and a supply curve for a commodity
lars per year and the account earns annual interest at the are determined by economic factors. The interaction
rate of 100r% compounded continuously, then the fu- between supply and demand usually determines the
ture value of the account at the end of T years is given by quantity of an item that is available. Areas of special in-
terest that are determined as areas associated with sup-
R(t)e
T
Future value r(Tt)dt dollars ply and demand curves are consumers’ expenditure,
0 consumers’ surplus, consumers’ willingness and ability
The present value of an income stream is the amount to spend, producers’ willingness and ability to receive,
that would have to be invested now in order for the producers’ surplus, producers’ revenue, and total social
account to grow to a given future value. The present gain.
value of a continuous income stream whose future value
is given by the previous equation is Probability Distributions and Density Functions
R(t)e
T
rtdt The probability P(a x b) is a measure of the likeli-
Present value dollars
0 hood that an outcome of an experiment involving a ran-
dom quantity x will lie between a and b. Functions that
When the income stream comes into the account dis-
describe how the probabilities associated with a random
cretely, we determine future value by summing rather
variable are distributed over various intervals of num-
than integrating:
bers are called probability density functions (or proba-
R(d)1 n
D1 r Dd bility distributions).
Future value dollars Integrals of a probability density function f have the
d0
following meanings:
where n is the number of deposits made each year, R(d) • The likelihood that x is between a and b is
is the value of the dth deposit, 100r% is the annual in-
f (x)dx
b
terest rate, and D is the total number of deposits made. P(a x b)
Once the future value of a discrete income stream is a
448 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
• A measure of the center of the distribution is the mean: sity function, the exponential density function, and the
(bell-shaped) normal distribution.
xf(x)dx A cumulative density function is an accumulation
function of a probability density function. Outputs of
• A measure of the spread of the distribution is the cumulative density functions are the areas between the
standard deviation: corresponding probability density functions and the
horizontal axis. Probabilities can be determined by us-
(x )2 f (x)dx ing either probability density functions or cumulative
density functions.
Three types of probability distributions that show
up often in real-world applications are the uniform den-
CONCEPT CHECK
Can you To practice, try
• Evaluate improper integrals? Section 6.1 Activities 1, 5
• Recognize that an improper integral diverges? Section 6.1 Activities 7, 13
• Determine income flow rate functions? Section 6.2 Activities 1, 2
• Calculate and interpret present and future values of Section 6.2 Activities 7, 13
discrete and continuous income streams?
• Find various quantities related to a demand function? Section 6.3 Activities 9, 11
• Calculate elasticity? Section 6.3 Activities 13, 15
• Find various quantities related to a supply function? Section 6.3 Activities 17, 19
• Find the market equilibrium point and total social gain? Section 6.3 Activities 21, 23
• Find and interpret probability, mean, and standard deviation? Section 6.4 Activities 11, 15
• Understand and use probability density functions? Section 6.4 Activities 3, 5
• Work with cumulative density functions? Section 6.4 Activities 23, 24
Chapter 6 Concept Review 449
CONCEPT REVIEW
1. Investment In preparing to start your own busi- 4. Social Gain The average quantity of marble
ness (in 6 years), you plan to invest 10% of your fountains that consumers will demand can be
salary each month in an account with a fixed rate of modeled as
return of 5.3%. You currently make $3000 per
D(p) 1.0p2 20.6p 900 fountains
month and expect your income to increase by $500
per year.
and the average quantity that producers will supply
a. Find a function for the yearly rate at which you can be modeled as
will invest money in the account.
0 fountains if p 2
b. If you start investing now, to what amount will S(p)
your account grow in 6 years? (Consider a con- 0.3p 2 8.1p 300 fountains if p 2
tinuous stream.) when the market price is p hundred dollars per
c. How much would you have to invest now in one fountain.
lump sum, instead of in a continuous stream, in a. How much are consumers willing to spend for 30
order to build to the same 6-year future value? fountains?
2. Investment A teacher is planning to retire in 8 b. How many fountains will producers supply at
years. To supplement her state retirement income, $1000 per fountain? Will supply exceed demand
she plans to invest 7% of her salary each month until at this quantity?
retirement in an annuity with a fixed rate of return of c. Determine the total social gain when fountains
5.2% compounded monthly. She currently makes are sold at the equilibrium price.
$3100 per month and expects her income to increase,
thanks to consulting work, by 0.4% per month. 5. The density function for a random variable x is
a. How much will be in the annuity at the end of 8
years? f (x) 0.125x when 0 x 4
0 when x 0 or x 4
b. How much would she have to invest now, in one
lump sum, to accumulate the same amount as a. Find the probability that x is less than 3.8. Inter-
the 8-year future value found in part a? pret this result.
3. Population Suppose a 1990 population of 10,000 b. Find the probability that x is between 1.3 and 5.
foxes breeds at a rate of 500 pups per year and has a Indicate, on a graph of f, what this answer
survival rate of 63%. represents.
a. Assuming that the survival and renewal rates re- c. Find the mean value of x. Locate the mean on the
main constant, determine how many of the foxes graph you drew in part b.
alive in 1990 will still be alive in 2010. d. Find and graph y F(x), the cumulative density
b. Write a function for the number of foxes that function for x.
were born t years after 1990 and will still be alive e. Use F to find the answer to part b. Show all work
in 2010. involved in your calculation.
c. Estimate the fox population in the year 2010.
Project 6.1 Arch Art
Setting 4. The artist plans to use strips of Mylar 60 inches
wide. What is the minimum number of yards
A popular historical site in Missouri is the Gateway
of Mylar that the artist will need to purchase?
Arch. Designed by Eero Saarinen, it is located on the
original riverfront town site of St. Louis and 5. Repeat Task 4 for strips 30 inches wide.
symbolizes the city’s role as gateway to the West.
6. If the 30-inch strips cost half as much as the 60-inch
The stainless steel Gateway Arch (also called the St.
strips, is there any cost benefit to using one width
Louis Arch) is 630 feet (192 meters) high and has an
instead of the other? If so, which width? Explain.
equal span.
In honor of the 200th anniversary of the Louisiana
Purchase, which made St. Louis a part of the United Reporting
States, the city has commissioned an artist to design a
work of art at the Jefferson National Expansion Write a memo telling the artist the minimum
Memorial which is a National Historic Site The artist amount of Mylar necessary. Explain how you came
plans to construct a hill beneath the Gateway Arch, to your conclusions. Include your mathematical
located at the Historic Site, and work as an attachment.
hang strips of Mylar from the arch to Height
the hill so as to completely fill the (feet)
space. (See Figure 6.45.) The artist has 630
asked for your help in determining the
570
amount of Mylar needed.
510
450
Tasks
390
1. If the hill is to be 30 feet tall at its
330
highest point, find an equation
for the height of the cross section 270
of the hill at its peak. Refer to
the figure. 210
2. Estimate the height of the arch in 150
at least ten different places. Use
the estimated heights to construct 90
a model for the height of the arch. 30 Span
(You need not consider only the 0
Hill
(feet)
models presented in this text.) 0 60 120 180 240 300 360 420 480 540 600
3. Estimate the area between the arch The Gateway Arch in St. Louis
and the hill. Figure 6.45
450
Repetitive Change:
Cyclic Functions
7
Concepts Outline
7.1 Cycles and Sine Functions
7.2 Sine Functions as Models
7.3 Rates of Change and
Derivatives
7.4 Extrema and Points
of Inflection
7.5 Accumulation in Cycles
Joseph Sohm/Chromo Sohm Media, Inc./Photo Researchers Inc.
Concept Application
The levels of carbon dioxide (CO2) in the atmosphere fluctuate in a periodic manner
throughout the year, the lowest levels occurring in the summer when plants absorb sig-
nificant amounts of CO2. The amount of CO2 in the atmosphere has been increasing for
decades, and by using a model constructed from a trigonometric function and a quad-
ratic function, we can model the carbon dioxide level over time. The model can then be
used to answer questions such as
• How quickly is the level of carbon dioxide changing in a given month?
• What are the maximum and minimum levels of CO2 in a given year, and when do
those levels occur?
• When during a year is the amount of carbon dioxide increasing most rapidly?
You will be given an opportunity to answer questions such as these in Examples 2 and 3
of Section 7.3 and Activities 10 and 18 of Section 7.4.
451
452 CHAPTER 7 Repetitive Change: Cyclic Functions
Chapter Introduction
Many things change and then return to repeat the same change. This creates a repet-
itive pattern and forms a cycle. Patterns of repeating cycles occur in nature as well as
in business applications.
Many cycles are best described with the functions from an area of mathematics
called trigonometry. This term has its roots in the Greek word for “triangle measure-
ment.” Most of early trigonometry was centered on using ratios in similar triangles to
determine indirect measurements. However, the trigonometric functions are now
widely used to describe cyclic patterns of change.
Understanding the repetitive process in cycles and the mathematical models used
to describe such cycles is the goal of the first part of this chapter. The second part of
the chapter shows how calculus concepts can be applied to cyclic functions.
Concepts You Will Be Learning
• Identifying the amplitude, reflection, period, and horizontal and vertical shifts of a
sine function (7.1, 7.2)
• Using data to construct and graph a sine model without the aid of technology
(7.2)
• Using technology to construct a sine model (7.2)
• Writing rate-of-change formulas for sine and cosine functions (7.3)
• Calculating and interpreting derivatives of sine and cosine functions (7.3)
• Locating and interpreting extrema of sine and cosine functions (7.4)
• Locating and interpreting inflection points of sine and cosine functions (7.4)
• Writing antiderivatives of sine and cosine functions (7.5)
• Calculating and interpreting definite integrals of sine and cosine functions (7.5)
• Determining the expected value of a cyclic function (7.5)
7.1 Cycles and Sine Functions
A study of functions would not be complete without an examination of functions
that repeat a pattern. If a function repeats itself and the length between the repeti-
tions is constant, then we say the function is periodic. The shortest interval over
which the function repeats itself is called the period. Furthermore, when a function
is periodic and varies continuously between two extremes, we say the function is
cyclic.
Cyclic Function
A cyclic function is a periodic, continuous function that varies between two
extremes.
7.1 Cycles and Sine Functions 453
Sine and Cosine Functions
The most common periodic functions are the trigonometric (or trig) functions.
There are six trig functions, all of them periodic. However, we restrict our study to
the two cyclic trig functions: the sine function (abbreviated sin) and the cosine func-
tion (abbreviated cos). The basic sine function is of the form y sin x; its graph is
shown in Figure 7.1.
f (x)
1 f (x) = sin x
x
− 2π _
− 3π −π − _π2 _π π _
3π 2π _
5π 3π _
7π 4π
2 2 2 2 2
−1
FIGURE 7.1
This graph extends infinitely in both horizontal directions. Note from the graph that
the sine function is periodic with a period of 2. That is, it starts at y 0 when x 0
and then rises to a maximum, falls to a minimum, and returns to y 0 again when
x 2. This pattern repeats itself every 2 input units. The sine function is also cyclic
because it varies continuously, alternating between 1 and 1.
If you studied trigonometric functions in a previous math course, you probably
It is important that you make considered the input of the sine function to be measured in degrees and may have
sure that your technology is set also related this function to angles on a unit circle. However, it is important to
to radian mode while
performing all trig
note that in this course, we use the sine function simply as a model for situations
function calculations that exhibit cyclic behavior with the input measured in real numbers (radians, not
in this text. 7.1.1 degrees).
Another trig function that we make use of is g(x) cos x. (See Figure 7.2.)
g (x)
1 g(x) = cos x
x
− 2π _
− 3π −π − _π2 _π π _
3π 2π _
5π 3π _
7π 4π
2 2 2 2 2
−1
FIGURE 7.2
The connection between the sine and cosine functions and the unit circle can be used
to show that for any input x measured in radians, cos2x sin2x 1. The cosine
function can also be considered as a horizontal shift of the sine function by the rela-
tionship cos x sin(x 2 ). We therefore restrict the trigonometric models we de-
velop to those involving only the sine function.
454 CHAPTER 7 Repetitive Change: Cyclic Functions
Variations of the Sine Function
The sine function can be modified to vary between extremes other than 1 and 1, to
shift up or down, to repeat itself more or less often than 2, and to shift right or left.
The ability to modify the sine function in these ways will be essential when we use a
sine function to model data.
Multiplying the function y sin x by a constant changes the extremes between
which the output values of the function vary. In particular, the function y a sin x
varies between a and a. The magnitude of a, a , is called the amplitude, and its ef-
fect is illustrated in Figure 7.3. A negative value of a indicates that the graph is reflected
across the horizontal axis.
y
2
1
y
_1
π 2π π 2π 2 π 2π
x x x
-_
1
2
-1
-2
1
y sin x y 2 sin x y 2 sin x
amplitude 1 amplitude 2
1
amplitude 2
FIGURE 7.3
Adding a constant to, or subtracting a constant from, the sine function shifts the
graph up or down. This constant can be thought of as the average or expected value
of one cycle because it is the midpoint between the high and low values of the func-
tion. Figure 7.4 illustrates the effect of adding or subtracting a constant to or from a
sine function.
y
6 y
y
y _3
y 4 2
5
2π
1 4 _1 2π x
2 − _1
x x 2
3 − _1 2
2
−1 2π
− 3_
2π 2 2π − _5
x x 2
1 1
y sin x y sin x 4 y sin x 2 y 2 sin x 4 y 2 sin x 2
FIGURE 7.4
7.1 Functions of Angles: Sine and Cosine 455
The graph of a sine function repeats its cycle over an interval of a specific length.
The basic sine function y sin x repeats over an interval of length 2 units. When
the input x is multiplied by a constant b (b 0) the number of times the cycle
repeats over an interval of 2 units changes. When x is multiplied by 2, the graph
cycles twice over an interval of 2 units, or we say it cycles once over an interval of
1
units. When x is multiplied by 2, the graph cycles once over an interval of 4
units. The interval over which a graph of a cyclic function completes one cycle is
called the period. For a sine function of the form y sin bx, the constant b dictates
the period. To find the period of y sin bx, we simply divide 2 by b:
2
Period
b
Similarly, if we know the period, then we can find b by using the equation
2
b
period
Figure 7.5 illustrates several functions of the form y sin bx.
y y
1 1
2π 2π
x x
−1 −1
y sin x y sin 2x
b1 b2
period 2 period
y y
1 1
2π
x x
1 2 3 4 5 6 _
3π 3π _
9π 6π
2 2
−1 −1
y sin x y sin 3 x
1
b 1
period 2 b3
period 6
FIGURE 7.5
Finally, we can shift a sine function left or right by adding a constant to the input.
h
Specifically, for y sin(bx h), the graph is shifted horizontally by b units. If h is
456 CHAPTER 7 Repetitive Change: Cyclic Functions
negative, the shift is to the right. If h is positive, the shift is to the left. One way to iden-
tify the horizontal shift of a graph is to find an input at which the function attains
its average value and is increasing. Because the function is cyclic, such a point on
the function is not unique. The function y sin x is increasing and attains its aver-
h
age value at x 0, and y sin(bx h) reaches such a point at x b . Figure 7.6
illustrates several functions with horizontal shifts.
y y
1 1
2π − 2 3 2π + 3
x x
−2
−1 −1
y sin(x 2) y sin(x 3)
shift: 2 left and period 2 shift: 3 right and period 2
y y
1 1
_3 _3
π−1 π 2+ π
x x
−1
−1 −1
y sin(2x 2) y sin( x 3)
shift: 1 left and period 3
shift: 0.955 right and period 2
FIGURE 7.6
There is one final variation of the sine function to consider. The functions
y sin(x) and y sin x are equivalent functions. That is, sin(x) sin x. The
graph of each of these functions appears to be the reflection of the graph of y sin x
across the horizontal axis. See Figure 7.7.
y y
1 1
x x
0 2 0 2
–1 –1
y sin x y sin x or y sin(x)
FIGURE 7.7
7.1 Functions of Angles: Sine and Cosine 457
We earlier stated that we assume b is positive. This restriction does not limit the func-
tions we consider, because a negative value for either a or b in y a sin(bx h) k
indicates reflection. If we encounter a function in which b is negative, we simply fac-
tor the negative 1 out of the (bx h) term and use the fact that sin(x) sin x to
rewrite the function with a positive b. For example, y 2 sin(3x 4) 6 can be
rewritten as
y 2 sin[(3x 4)] 6 2 sin(3x 4)] 6 2 sin(3x 4) 6
h
Consider the cycle of the graph y a sin(bx h) k that begins at x b . The out-
put value is always k at the point with this input. If a is positive, this cycle will first in-
crease and then decrease. If a is negative, this cycle is reflected across the line y k, so
it will first decrease and then increase.
We summarize the effects of the constants a, b, h, and k:
Instead of the equation
y a sin(bx h) k , some
technologies use
Variations of the Sine Function
y a sinb(x h) k . The The constants of the sine function
advantage in this alternative
form is that the quantities f(x) a sin(bx h) k
influenced by the parameter
have direct readouts. That is, are interpreted as follows (we assume b 0):
in the equation
y a sinb(x h) k , h is a is the amplitude, where a 0 denotes reflection across the
precisely the horizontal shift. horizontal axis.
In the first formulation, the
h
2
one used in the text, b is the b is the period.
horizontal shift.
h
b is the horizontal shift, right if h 0 and left if h 0.
k is the vertical shift, up if k 0 and down if k 0.
EXAMPLE 1 Graphing Variations of the Sine Function
For each of the following sine functions, specify the amplitude, period, horizontal
shift, and vertical shift. Graph one cycle of each function.
a. y 3 sin x 4
b. y sin(2x 6)
c. y 0.5 sin(0.2x 1) 2.5
y = 3 sin x − 4
x Solution
π 2π
−1 a. The parameters of y 3 sin x 4 are a 3, b 1, h 0, and
k 4. Thus the amplitude of this function is 3, and the graph is
shifted down 4 units. Without the shift down, the function would vary
−4 between 3 and 3. Subtracting 4 from each of these values, we see that
the graph of y 3 sinx 4 varies between 7 and 1. The period is
2. We choose to graph the cycle that begins at 0 and ends at 2. Be-
−7
cause a is positive, this cycle begins with the graph increasing. The
FIGURE 7.8 graph is shown in Figure 7.8.
458 CHAPTER 7 Repetitive Change: Cyclic Functions
y = sin(-2x – 6) b. Because b is negative in this function, we factor out the nega-
1 tive 1 and use the property that sin(x) sin x:
sin(2x 6) sin[(2x 6)] sin(2x 6)
Thus we consider the function in the form y sin(2x 6).
x For this function a 1, b 2, h 6, and k 0. The am-
-3 –3 –3
2 plitude of y sin(2x 6) is 1. There is no vertical shift, so
the function varies between 1 and 1. The constant b is 2, so
2
-1 the period is 2 . The graph is shifted horizontally
6
FIGURE 7.9 2 3 units, and the shift is to the left because 6 is positive.
Without the horizontal shift, one cycle would begin at 0 and
end at . With a horizontal shift of 3 units to the left, one cy-
cle begins at 3 and ends at 3.
Because a is negative, this cycle begins by declining rather than rising. (See
Figure 7.9.)
c. For the function y 0.5 sin(0.2x 1) 2.5, the parameters are a 0.5,
b 0.2, h 1, and k 2.5. The graph has amplitude 0.5 and is shifted
up 2.5 units. Thus the function varies between 2 and 3. The constant b of the
2
function is 0.2, so the period is 0.2 10. The graph is shifted horizontally
1
0.2 5 units, and the shift is to the right because 1 is negative. Without
the horizontal shift, one cycle begins at 0 and ends at 10. With the horizontal
shift 5 units to the right, one cycle begins at 5 and ends at 10 5. Because a
is positive, this cycle begins with the graph increasing. The graph is shown in
Figure 7.10.
y = 0.5 sin(0.2x – 1) + 2.5
2.5
0 x
5 5 + 5 10 + 5
FIGURE 7.10 ●
Now that we have seen the basic form of the sine function f(x)
a sin(bx h) k and understand how the constants a, b, h, and k are related to the
shape of a graph of the function, we can consider the usefulness of the sine function
when modeling cyclic data. In the next section, we look at modeling various sets of
cyclic data with sine functions.
7.1 Functions of Angles: Sine and Cosine 459
For each of the functions in Activities 6 through 10, state
7.1 Concept Inventory the amplitude, period, vertical shift, and horizontal
shift. Also state whether the function is reflected across
• Periodic function the horizontal axis and whether the graph is increasing
h
• Cyclic function or decreasing at x b .
• Sine and cosine functions
6. f(x) 0.1 sin(2x 4)0.5
• Sine function graph
• Sine function variations: y a sin(bx h) k 7. g(x) sin(x )
Amplitude 8. g(x) 3.62 sin(0.22x 4.81) 7.32
Horizontal shift
9. p(x) 235 sin(300x 100)65
Vertical shift
Period 10. p(x) sin(x 4) 0.7
Expected value 11. Match functions a through f with graphs i through vi.
Cycle a. y 3 sin x 8 b. y 3 sin x 3
Reflection c. y 3 sin x 2 d. y 4 sin x 8
e. y 2 sin x 3 f. y 2 sin x 2
7.1 Activities y y
5 x
Getting Started -1
1. Construct an input/output diagram for the function
s(x) sin x.
x
2. Construct an input/output diagram for the function -1 -5
c(x) cos x.
(i) (ii)
3. a. For x-values from 0 to 2, what is the largest
value that s(x) sin x achieves? Also find the
value(s) of x at which the maximum occurs. y y
b. For any real number x, what is the largest value 11
that s(x) sin x achieves? Also find the value(s) 4
of x at which the maximum occurs.
c. For x-values from 0 to 2, what is the smallest 5
value that s(x) sin x achieves? Also find the x
value(s) of x at which the minimum occurs. x
d. For any real number x, what is the smallest value (iii) (iv)
that s(x) sin x achieves? Also find the value(s)
of x at which the minimum occurs.
4. Repeat Activity 3 for the cosine function. y y
5. a. Graph c(x) cos x and s(x) sin x on the 12 x
same axes for 0 x 2.
b. Find where c(x) s(x).
c. On the basis of your answer to part b, what are 4 -6
the x-coordinates of all the points of intersection x
for the sine and cosine functions over all real
numbers? (v) (vi)
460 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
12. a. What is the distance between two consecutive re- 20. Sketch a graph of y sin x from x 0 to x 2.
peating maxima of the function f(x) sin(2x)? a. On the same axes, plot points for y sin x for
of the function g(x) sin(0.5x)? x 0, /2, , 3/2, and 2. Draw arrows from
the points on the graph of y sin x to the corre-
1 2
b. For the function h(x) sin 3 x 3 and the sponding points you plotted for y sin x.
Across which axis (vertical or horizontal) does the
1 1
function j(x) sin 3 x 3 , estimate the loca- function y sin x appear to have been reflected
tions of the points of greatest slope closest to to give the points for y sin x?
x 0.
b. On another set of axes, sketch a graph of the
c. Match the four functions given in this activity function y sin x from x 0 to x 2. On
with graphs vii through x. this new graph, plot points for y sin(x) for
x 0, /2, , 3/2, and 2. Draw arrows from
y y
the points on the graph of y sin x to the corre-
1 1 sponding points you plotted for y sin(x).
Across which axis (vertical or horizontal) does
the function y sin x appear to have been re-
x x
-10 -10 -2
flected to give the points for y sin(x)?
10 10
c. On a third set of axes, sketch a graph of
-1 -1
y sin x from x 0 to x 2. Extend the
graph to the left to x 2. In a second color
(i) (ii)
but on the same set of axes, sketch a graph of
y sin(x) from x 2 to x 0. Extend
the graph to the right to x 2 . In a third
y y
color, graph y sin x from x 2 to x 2.
1 1 How are the functions y sin x, y sin x, and
y sin(x) related?
x x 21. Sketch a graph of y cos x from x 0 to x 2.
-10 1 10 -10 10
a. On the same axes, plot points for y cos x for
-1 -1 x 0, /2, , 3/2, and 2. Draw arrows from
the points on the graph of y cos x to the cor-
(iii) (iv) responding points you plotted for y cos x.
Across which axis (vertical or horizontal) does
Without the aid of technology, graph one cycle of each the function y cos x appear to have been re-
of the functions in Activities 13 through 18. flected to give the points for y cos x?
13. f(x) sin 2x 3 b. On another set of axes, sketch a graph of the
function y cos x from x 0 to x 2. On
14. f(x) 2 sin x 3 this new graph, plot points for y cos(x) for
15. f(x) 2 sin(x 3) x 0, /2, , 3/2, and 2. Draw arrows from
the points on the graph of y cos x to the cor-
16. f(x) sin(2x 3) responding points you plotted for y cos(x).
17. f(x) 2 sin(2x 3) Across which axis (vertical or horizontal) does
the function y cos x appear to have been re-
18. f(x) 2 sin(2x 3) 3 flected to give the points for y cos(x)?
19. Compare the graph of y 2 sin x with the graph c. On a third set of axes, sketch a graph of
of y 2 sin(x ). Also compare the graphs of y cos x from x 0 to x 2. Extend the
y 0.5 sin x and y 0.5 sin(x 3). What graph to the left to x 2. In a second color
generalization can you make from your observations? but on the same set of axes, sketch a graph of
7.2 Sine Functions as Models 461
y cos(x) from x 2 to x 0. Extend Discussing Concepts
the graph to the right to x 2. In a third color,
22. Explain how to find the amplitude and vertical shift
graph y cos x from x 2 to x 2. How of a sine function given its graph and/or data.
are the functions y cos x , y cos x , and
y cos(x) related? Does cos x cos(x)? 23. Explain how to find the period and horizontal shift
of a sine function given its graph and/or data.
7.2 Sine Functions as Models
Be aware that not all The world is full of cyclic data that repeatedly return to the same maximum and min-
technologies have a built-in sine
imum. If we learn to model these data, then we can use calculus and the model to ob-
regression routine. See the Excel
Instruction Guide to obtain a sine tain special insight into the changing nature of the data. The sine and cosine functions
curve-fitting procedure. are cyclic, and they are useful in modeling data that exhibit cyclic behavior. Because
these two functions are similar (one can be obtained from the other by a horizontal
shift), we choose to use only the sine function when modeling cyclic data.
In Section 7.1 we examined the sine function of the form y a sin(bx h) k.
We saw that the constant a determines the height and reflection of the curve, the con-
2
stant b determines the period as , the constant h influences the horizontal shift, and
b
the constant k is the vertical shift. We now consider using this function to model data
in real-life situations.
The monthly amount of ultraviolet (UV) radiation (in watts per square centimeter)
received by a horizontal surface of area 1 cm2 in Auckland, New Zealand, is given in
Table 7.1, and a scatter plot of these data is shown in Figure 7.11.
TABLE 7.1
Month Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
UV radiation
55 46 32 19 10 7 8 15 27 40 52 58
(watts per cm2)
(Source: W. Rudloff, World-Climates, Stuttgart, Germany: Wissenschaftliche Verlagsgesellschaft, 1981.)
UV radiation
(watts per cm2)
50
40
30
20
10
Month
1 2 3 4 5 6 7 8 9 10 11 12
(Jan) (Mar) (May) (Jul) (Sep) (Nov)
FIGURE 7.11
462 CHAPTER 7 Repetitive Change: Cyclic Functions
We see from the data and scatter plot that the data have a minimum and a maxi-
mum. We assume that the amount of UV radiation reaches this same maximum and
minimum each year; hence the data are cyclic. Because the repetitions occur on a
yearly basis, the period is 12 months. Thus it makes sense to fit a sine function to the
data. Before using technology to fit the function f(x) a sin(bx h) k, let us in-
spect the data more carefully to see if we can estimate the constants a, b, h, and k.
The Constants of a Sine Function
We begin our inspection by looking for k. The vertical shift of the sine function is the
value of k. It tells us how high (or low) the middle value of the output range is. This
is the average (or expected) value of the cycle. In order to determine k, we must know
the maximum output and the minimum output. The maximum (max) of the data oc-
curs in December at 58 watts per cm2, and the minimum (min) occurs in June at 7
watts per cm2. To find the middle value of the output, we average the maximum and
minimum values.
max min 58 7
k 32.5 watts per cm2
2 2
In other words, if we did not take into account the cyclic fluctuations in the amount
of UV radiation, we would expect the average amount to be 32.5 watts per cm2 for any
given month. However, there are monthly variations in the amount. The amplitude
tells us by how much the amount may vary from the expected (or average) value.
We find the amplitude a of the sine model by calculating half the difference be-
tween the maximum and minimum values.
max min 58 7
a 25.5 watts per cm2
2 2
That is, UV radiation varies by 25.5 watts per cm2 away from the expected value
of 32.5 watts per cm2.
Because the data are periodic with a cycle that we assume to repeat each year, the
period is 12 months. We calculate the constant b using the equation
2
Period
b
2
Thus 12 b , or b 6 0.52.
Finally, we look for the constant h, which is an indication of the horizontal shift
of the sine function. When looking for the horizontal shift of a function of the form
f(x) a sin(bx h) k, we seek an input value x0 at which the sine function f be-
haves the same way g(x) sin x behaves at x 0. At x 0, a graph of g(x) sin x
is increasing and attains its average value, so we are looking for an input value x0 at
which the graph of f(x) a sin(bx h) k is increasing and attains its average
value.
On a function of the form f(x) a sin(bx h) k, such an input value x0
occurs exactly halfway between the x-value at which f(x) is lowest and the x-value
at which f(x) is highest. Thus for the UV radiation data in Table 7.1, we find that
a candidate for the horizontal shift x0 is halfway between June and December. If we
7.2 Sine Functions as Models 463
align so that x 1 in January and x 6 in June, then x 12 in December, and
12 6
x0 2 9 (September). This shift is to the right, indicating that h is negative.
To calculate h from x0, we use the equation
h
x0 horizontal shift
b
h
with b 6 as previously calculated. Thus 9 /6 , or h ( 6 )(9) 4.71.
Now we can write a model to approximate the amount of ultraviolet radiation re-
ceived in Auckland, New Zealand, as
r(m) 25.5 sin(0.52m 4.71) 32.5 watts per cm2
when m 1 in January, m 2 in February, and so on. Figure 7.12 shows this equation
graphed on the scatter plot of the data.
r (m)
UV radiation
(watts per cm2)
50
40
30
20
10
m
1 2 3 4 5 6 7 8 9 10 11 12 Month
(Jan) (Mar) (May) (Jul) (Sep) (Nov)
FIGURE 7.12
As we have seen, the context of a real-life situation often indicates reasonable es-
timates for the constants of a sine model. Knowing the interpretation of the con-
stants a, b, h, and k in the general sine model f(x) a sin(bx h) k enables us
to calculate an equation by hand. We again illustrate these ideas in Example 1 by
using the sine function to model the number of daylight hours at a location on the
Arctic Circle.
EXAMPLE 1 Constructing a Sine Model Without Using Technology
Daylight Hours On the Arctic Circle there are 24 hours of daylight on the summer
solstice, June 21, and 24 hours of darkness on the winter solstice, December 21. We
use this information, along with the fact that there are 12 hours of daylight and 12
hours of darkness midway between the summer and winter solstices, to construct a
model that approximates the number of hours of daylight throughout the year on the
Arctic Circle. We use as input the number of days since the beginning of the year and
show just over 2 years of data so that the pattern in the data can be easily seen. Note
that if January 1 is day 1, then December 21 of the previous year is day 10 and
464 CHAPTER 7 Repetitive Change: Cyclic Functions
December 21 of the current year is day 355. The data are given in Table 7.2, and a
scatter plot of the data is shown in Figure 7.13
TABLE 7.2
Date 12/21 6/21 12/21 6/21 12/21
Day of the year 10 81.5 173 264 355 446.5 538 629 720 811.5
Hours of daylight 0 12 24 12 0 12 24 12 0 12
Daylight
hours
24
12
Day
100 200 300 400 500 600 700 800
FIGURE 7.13
Use the data in Table 7.2 to calculate the constants a, b, h, and k of a model of the
form f (x) a sin(bx h) k.
Solution
The scatter plot suggests that a sine model is appropriate. The output values of the
data range from 0 to 24 hours; therefore, we expect the amplitude of our model to be
24
2 12 hours. However, without a vertical shift, the output of a sine model with am-
plitude 12 is between 12 and 12 hours. Because the data are between 0 and 24,
we expect a vertical shift upward of 12 hours. We also expect the model to repeat it-
self every 365 days. To find the horizontal shift, we need to locate where the number
of hours of daylight is the average value of the cycle (12 hours) and is first increasing.
This occurs at 81.5 days, so 81.5 is the horizontal shift to the right. Thus in the sine
function f(x) a sin(bx h) k, we expect
Amplitude a 12
2
Period 365, so b 0.01721
b
h
Horizontal shift to the right 81.5, so h 1.40296
b
Vertical shift k 12
A model for the number of hours of daylight on the Arctic Circle is
h(t) 12 sin(0.01721t 1.40296) 12 hours
where t is the number of days since December 31 of the previous year. Figure 7.14
shows this model graphed on the scatter plot.
7.2 Sine Functions as Models 465
h(t)
Daylight
hours
24
12
t
100 200 300 400 500 600 700 800
Day
FIGURE 7.14 ●
Fitting a Sine Model to Data
It may be the case that the data and information provided are not sufficient to predict
accurately the maximum, minimum, or period of the cycle, or you may wish to find
a model quickly. In such cases, you can use a calculator or computer to fit a sine func-
tion to the data. It is still important to understand the role of the constants a, b, h, and
k of the equation so that you can interpret the model.
At the beginning of this section we calculated a model of the data for the amount
of ultraviolet radiation in Auckland, New Zealand. However, our estimates of the
maximum and minimum of the function were not exact. The minimum of the func-
tion probably occurs somewhere between the June and the July data, and the maxi-
mum probably occurs somewhere between the December and the January data. A
In this text, all sine models
sine function fitted by technology to the data in Table 7.1 is
found using technology were
constructed on a TI-84 Plus. If A(m) 27.1 sin(0.485m 1.707) 32.9 watts per cm2
you use a different technology,
your model may vary. when m 1 in January, m 2 in February, and so on.
Let us examine this model and compare the constants to those we calculated ear-
lier. The function A gives the average value of UV radiation as approximately 32.9
watts per cm2 instead of the 32.5 watts per cm2 previously calculated from the data.
7.2.1
Ultraviolet radiation fluctuates as much as 27.1 watts per cm2 above and below this
average, according to the function A, giving a slightly larger amplitude than that calcu-
2 2
lated earlier. The period of the function A is b 0.485 13.0. Thus the equation
found using technology has a period of more than 12 months.
h 1.707
Finally, the horizontal shift of A is b 0.485 3.5 to the left. Because this
value is outside the data interval, we add one period (approximately 13) to it in order
to interpret it correctly. In this case, adding the period to the horizontal shift gives ap-
proximately 9.4. This number is within the range of our data and can be interpreted
as halfway between the September data and the October data. Thus UV radiation hits
its average value and is increasing halfway between the September data and the
October data. (The horizontal shift calculated earlier was 9, or September.) This sup-
ports our analysis that the minimum probably occurs halfway between the June and
the July data and that the maximum probably occurs halfway between the December
and the January data.
466 CHAPTER 7 Repetitive Change: Cyclic Functions
When we compare the two models, we see that the graph of the model calculated
via technology is a better fit to the data (see Figure 7.15a). However, because the pe-
riod of the technology-generated model is more than 1 year (the period we expect
from the context), the two models will begin to differ by more and more after the first
cycle (see Figure 7.15b). Thus the technology-generated model should be used only
for inputs between 0 and 12.
UV radiation UV radiation
(watts per cm2) (watts per cm2)
60 A(m)
50
Technology- r(m)
40
30 generated
function
20
10
Month Month
1 2 3 4 5 6 7 8 9 10 11 12
0 12 24 36
(Jan) (Mar) (May) (Jul) (Sep) (Nov)
(a) (b)
FIGURE 7.15
EXAMPLE 2 Using Technology to Find a Sine Model
Daylight Hours Use technology to find a sine model for the Arctic Circle hours-of-
daylight data. Compare the equation with the one calculated in Example 1.
Solution
A technology-determined model is
y 12.00011 sin(0.01721x 1.40296) 12 hours
where x is the number of days since December 31 of the previous year. We see that
the amplitude coefficient of this equation agrees to three decimal places with that of
the equation found in Example 1. The other coefficients agree to at least the number
of decimal places shown here. ●
We now have the ability to use a sine function to model data that exhibit a wave
pattern. A sine model may also be used if the data do not show more than one com-
plete cycle but the context indicates that the situation is cyclic in nature. It is also pos-
sible to use a sine model if a situation is not cyclic but the curvature of the data matches
that of a sine function. In such cases the model would be useless for extrapolation, but
it could be used for interpolation.
7.2 Sine Functions as Models 467
average number of gas thermal units per day
7.2 Concept Inventory used during the xth month of the year.
2. Minimum Wage The federal minimum wage
• The sine function: f(x) a sin(bx h) k
rates, in constant 2000 dollars, between 1950 and
• Determining sine function constants a, b, h, and k 2000 are shown in the table.
from data without the aid of technology
• Finding a sine function using technology
Mininum Minimum
wage wage
Year (dollars) Year (dollars)
1950 5.36 1980 6.48
7.2 Activities 1955 4.82 1985 5.36
1960 5.82 1990 5.01
Applying Concepts
1965 6.83 1995 4.80
1. Gas Usage The table shows the average number of 1970 7.10 2000 5.15
gas thermal units per day used by a residential gas 1975 6.72
customer in Reno, Nevada. (The data represent 15
months.) (Source: U.S. Employment Standards Administration.)
a. Renumber the input as the number of years after
Gas usage Gas usage 1950, and use the data to determine the amplitude
(therms (therms
and the vertical shift.
Month per day) Month per day)
Nov 1.2 July 0.4
b. Approximate the period and the horizontal shift
from the data.
Dec 1.7 Aug 0.3
c. Construct, without the aid of technology, a sine
Jan 3.2 Sept 0.3
model for the aligned data.
Feb 3.3 Oct 0.4
d. Graph the function in part c on a scatter plot of
Mar 3.3 Nov 1.4
the aligned data. Is the fit reasonable?
Apr 2.5 Dec 2.3
e. If the federal minimum wage data had been
May 1.5 Jan 2.9
given in current dollars rather than in constant
June 0.9 2000 dollars, would a cyclic function fit the data?
Explain.
3. Temperature The accompanying table on page
a. Plot the data for gas usage. Explain why the data 468 gives the normal daily mean temperatures,
should be cyclic. based on the 30-year period 1961 through 1990,
for Omaha, Nebraska.
b. Use the data to estimate the amplitude and the
vertical shift. a. Plot these daily mean temperatures. Explain why
the data should be cyclic.
c. Align the first January to x 1, and approxi-
mate the period and horizontal shift from the b. Use the data to determine the amplitude and the
data. vertical shift.
d. Construct, without the aid of technology and c. Approximate the period and the horizontal shift
using the alignment in part c, a sine model for the from the data.
468 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
Daily mean Daily mean
a. Renumber the input data as years since 1965. Us-
temperature temperature ing only the data, estimate the amplitude and
Month (°F) Month (°F) vertical shift.
Jan 21.1 July 76.9 b. Using only the data, estimate the period and the
Feb 26.9 Aug 74.1 horizontal shift.
Mar 38.6 Sep 65.1 c. On the basis of the values calculated in parts a
Apr 51.9 Oct 53.4
and b, give a sine model for the reduction data.
Sketch a graph of the function on a scatter plot of
May 62.4 Nov 39.0
the data.
June 72.1 Dec 25.1
d. Use the equation in part c to construct a model
(Source: U.S. National Oceanic and Atmospheric for the number of beetles not surviving in a given
Administration, Climatography of the United States, No. 81.) year. Sketch a graph of this function, and com-
pare it with the graph in part c.
d. Construct, without the aid of technology, a sine 5. Temperature The table shows the normal daily
model for mean daily temperature during the xth low temperatures, based on the 30-year period 1961
month of the year. through 1990, for Omaha, Nebraska.
e. Graph your function on the plot of the data. Is
the fit reasonable?
f. Use your function to predict the mean daily tem- Daily minimum Daily minimum
perature in Omaha in July of this year. Discuss temperature temperature
the reasonableness of the prediction. Month (°F) Month (°F)
Jan 10.9 July 65.9
4. Beetles Some populations of animals exhibit peri-
odic behavior over time. During the 1960s and Feb 16.7 Aug 62.9
1970s, two scientists carefully studied the popula- Mar 27.7 Sep 53.6
tion of the Carabid beetle in a region of the Nether- Apr 39.9 Oct 41.2
lands. The table shows data from their study. The
May 50.9 Nov 28.7
output data are log (base 10) data for the number of
beetles not surviving in a given year. These values June 60.4 Dec 15.6
are referred to as the reduction in the number of bee- (Source: U.S. National Oceanic and Atmospheric Administration,
tles. The number of beetles not surviving in a given Climatography of the United States, No. 81.)
year is found by raising 10 to the reduction value in
the table corresponding to that year.
a. How would you expect the scatter plot and
model for the normal daily low temperatures in
log (base 10) log (base 10) Omaha to be related to the scatter plot and
Year reduction Year reduction model for the normal daily mean temperatures
1965 2.7 1971 2.7
in Omaha?
1966 2.8 1972 2.9 b. Use the data for the normal daily low tempera-
tures to determine the amplitude and the vertical
1967 4.5 1973 3.7
shift. Plot the daily low temperatures, and use the
1968 4.4 1974 4.5 plot to approximate the period and horizontal
1969 3.3 1975 4.4 shift. Compare these results with those you
1970 2.8 1976 2.7 found for the daily mean temperature in Activity
3. Are the findings what you expected?
(Source: P. J. denBoer and J. Reddingius, Regulation and Sta-
bilization Paradigms in Population Ecology, London: Chapman c. On the basis of your answers to parts a and b, write
& Hall, 1996.) a model for the normal daily low temperatures.
7.2 Sine Functions as Models 469
6. Temperature The normal daily high tempera- 8. Minimum Wage Consider the data in Activity 2.
tures for the 30-year period 1961 through 1990 in a. Aligning the input as the number of years after
New Orleans, Louisiana, and Boston, Massachu- 1950, use technology to find a sine function that
setts, are given in the table. fits the aligned minimum-wage data.
b. Compare and contrast the equation found in
Activity 2 with the one found with technology
New Orleans Boston
in this activity.
daily high daily high
Month temperature (°F) temperature (°F) c. Use the model found in Activity 2 and the one
found in this activity to predict the federal min-
January 60.8 35.7
imum wage, in constant 2000 dollars, in the cur-
February 64.1 37.5 rent year. Do these predictions seem reasonable?
March 71.6 45.8
9. Temperature Use your calculator or computer
April 78.5 55.9
to find a sine function that fits the data in Activity
May 84.4 66.6 3. Compare and contrast the equation found in
June 89.2 76.3 Activity 3 with the one found with technology in
July 90.6 81.8 this activity.
August 90.2 79.8 10. Beetles Refer to the data in Activity 4.
September 86.6 72.8 a. Use technology to find a sine equation that fits
October 79.4 62.7 the data. Compare and contrast the constants
November 71.1 52.2 of the equation in part c of Activity 4 with those
of the technology-derived equation.
December 64.3 40.4
b. Compare the graph of the technology-derived
(Source: U.S. National Oceanic and Atmospheric Administration,
equation in part a with the graph in part c of
Climatography of the United States, No. 81.)
Activity 4. Judging on the basis of these graphs,
which model do you believe is the more appro-
a. Would you expect these normal daily high tem- priate one?
perature data to be cyclic? Explain.
11. Temperature Refer to the data in Activity 5.
b. Use the data to find the difference between high a. Use your calculator or computer to find a
temperatures in January and July for these two sine equation that fits the Omaha daily low-
cities. What do these differences tell you about temperature data. Graph this equation and the
how the climates of the cities compare? one found in Activity 5 on a plot of the data, and
7. Gas Usage Refer to the data for the average number examine the plot and the graphs. Discuss any dif-
of gas thermal units per day used by a residential ferences you observe.
gas customer in Reno, Nevada, given in Activity 1. b. What, if any, are the advantages of using tech-
nology to find this equation?
a. Align the data so that the first January is x 1.
Use technology to find a sine function to fit 12. Temperature Refer to the data in Activity 6.
the data. State the amplitude, period, and the
a. Discuss any differences and similarities you would
horizontal and vertical shifts of the equation.
expect in the models for the normal daily high
b. According to the function, what is the average temperature on the basis of the 30-year period
value (or expected value) of the number of 1961 through 1990 for New Orleans and Boston.
therms per day? b. Find sine models for each set of data. Compare
c. Assuming that a billing period has 30 days and the amplitude, vertical shift, horizontal shift, and
that the cost per therm is $0.5632, estimate the period of these functions. What do these values
average gas bill for this customer. indicate about the climates of the two cities?
470 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
c. Use the models to estimate the differences in a. Find the change in sales between January and
normal maximum daily temperatures for the June and between July and December.
cities between April 15 and May 1. What do b. Find a model for the ice cream data, and use it to
these values indicate about the climates of the estimate the changes in part a. Compare your
two cities during the last half of April? answers with those in part a.
13. Mass Transit In 1945, when gasoline was rationed
because of World War II, Americans took 23.4 billion 15. Temperature A model fit to the normal mean air
trips on mass transportation. Trips on mass transit temperature at Fairbanks, Alaska, is
reached a low of 6.5 billion by 1972. Now, growing f(x) 37 sin[0.0172(x 101)] 25
congestion is sending people back to public transit. degrees Fahrenheit
The table shows annual trips on U.S. mass trans-
portations systems between 1992 and 2000. where x is the number of days since the last day of
the previous year.
(Source: B. Lando and C. Lando, “Is the Curve of Temperature
Mass transit Mass transit Variation a Sine Curve?” The Mathematics Teacher, vol. 7, no. 6,
`Year trips (billions) Year trips (billions) September 1977, p. 535.)
1992 8.50 1997 8.23 a. Estimate the highest and lowest mean daily
1993 8.20 1998 8.60 temperatures.
1994 7.93 1999 9.08 b. Find the average of the highest and lowest mean
1995 7.80 2000 9.40 daily temperatures. Relate this value to the verti-
cal shift of the graph of f.
1996 7.87
(Source: American Public Transportation Association.) 16. Fever The table gives a patient’s temperature dur-
ing a 6-day illness, as recorded at noon each day.
a. Find a sine model for mass transit trips between
1992 and 2000.
b. Find a polynomial model for the data in the Day Temperature (°F)
table. 1 103.9
c. Discuss how well the functions in parts a and b 2 104.4
fit the data.
3 102.8
d. If you were to select one of these two models to
4 100.2
predict mass transit use in 2006, which would
you choose? Explain. 5 98.6
6 99.2
14. Sales The data describe the monthly sales for an
ice cream company.
a. Use the data to find the average rate of change of
Monthly sales Monthly sales
the temperature between day 2 and day 5.
(thousand (thousand
Month dollars) Month dollars) b. Find a sine model for these data.
Jan 50 July 167 c. Judging by the graph of the model, what is your
Feb 60 Aug 159 prognosis for this patient? Discuss the implica-
tions of extrapolating in this situation.
Mar 77 Sept 108
Apr 96 Oct 75 17. Production The accompanying table shows the
May 137 Nov 61 approximate numbers of large, civil-transport air-
craft produced in the United States from 1949
June 158 Dec 54
through 1963.
7.2 Sine Functions as Models 471
Year Aircraft Year Aircraft Date 1/5 2/2 3/1 3/29 4/26
1949 185 1957 230 Soup
1950 165 1958 235 (millions of 200 215 180 160 150
16-oz cans)
1951 155 1959 225
Drink
1952 160 1960 210 (millions 375 425 430 460 570
1953 160 1961 180 of pints)
1954 180 1962 160 Date 5/24 6/21 7/19 8/16 9/13
1955 215 1963 155 Soup
(millions of 120 110 100 110 130
1956 220
16-oz cans)
(Source: Data estimated from J. J. Van Duijn, The Long Drink
Wave in Economic Life, London: Allen & Unwin, 1983.)
(millions 720 740 800 740 600
of pints)
Date 10/11 11/8 12/6 1/3
Soup
a. Examine a scatter plot, and describe the curva- (millions of 155 200 210 200
ture of the graph. 16-oz cans)
b. Which of the models that we have studied Drink
could be used to model these data? (millions 500 410 375 350
of pints)
c. Find a sine model for the data.
(Source: Wall Street Journal, July 7, 1993, p. B1.)
d. What does the equation give as the number of
large, civil-transport aircraft produced in the
United States in 1964? 19. Temperature Utility needs are dynamic, and
e. Do you think the periodic nature of a sine electric companies have to plan for base and peak
model is a good description of aircraft produc- loads carefully. Forecasting is fundamental to their
tion beyond 1963? Explain. management operations. A regular pattern of sup-
ply from the coal and nuclear plants is preplanned,
18. Sales In order to smooth out seasonal work and hydroelectric units are on standby to pick up
loads, the Campbell Soup Company produces both the slack. The reason for this program is that coal
canned soup and powdered drink mix. Estimated and nuclear plants are slow to react, and their effi-
sales for 1992 are shown in the table. The accompa- ciency is lost without a regular, consistent pattern
nying table values are in millions of reconstituted of operation. Hydroelectric facilities can come on
pints or millions of 16-ounce cans sold in a 4-week line immediately, but their resources are limited
period beginning on the date indicated. and must be used sparingly. Pump storage may be
used to restore hydroelectric resources during times
a. Find a model for the soup sales. of excess.
b. Find a model for the drink sales. In the summer, high temperatures affect air con-
ditioning energy requirements; in the winter, low
c. Compare and contrast the amplitudes, periods, temperatures affect heat requirements. The weather
and the vertical and horizontal shifts of the bureau’s forecast for the high and low temperatures
equations in parts a and b. for the next 5 days will be used as the information
d. How do you suppose management would use base for a temperature model from which the plan-
the two models that you have found in their pro- ners can anticipate power load requirements. A
duction scheduling? 5-day forecast is shown on page 472.
472 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
Wednesday Thursday Friday nuclear power generators can supply enough energy
for the highs and lows predicted by the model, when
will the hydroelectric plant’s energy be needed?
21. Utilities Consider electricity production.
a. Why do electric production facilities have to
operate in sync with power demands?
b. Note your local temperature forecast, and build
52 / 37 56 / 42 60 / 42
a power demand model for your area. During
the next several days, keep a record of the daily
Saturday Sunday highs and lows, and comment on how well your
model forecasts the demand.
c. Imagine limits on your hydroelectric facilities
that would make them inadequate to meet the
residual demands. Explain how increasing the
amplitude constant on the model could reduce
the calls on the hydroelectric facilities.
56 / 34 54 / 37 22. Daylight Hours Find local information, and
build a mathematical model for the hours of day-
light in the town or city where your school is
a. Construct a table using as input the number of located.
hours since midnight on Wednesday, assuming
that the high temperatures occur at 5 P.M. and 23. Darkness Hours Use two different methods to
the low temperatures occur at 5 A.M. Also assume build a sine model for the hours of darkness in
that the midrange temperatures between the Anchorage, Alaska, using the data in the table.
highs and lows occur at 11 A.M. and 11 P.M.
b. Find a sine model for the data. What is the
Day of Hours of Day of Hours of
period of the function? the year darkness the year darkness
c. What are the high and low temperatures given 1 18 211 7
by the equation?
31 16 241 9.5
20. Temperature Activity 19 gives a 5-day forecast 61 13 271 11.5
for temperatures. The actual highs and lows are
91 10.5 301 15.5
given in the following table.
121 7.5 331 17.5
151 5 361 18.5
Weekday High (°F) Low (°F)
181 4.5
Wednesday 56 35
Thursday 59 40
24. Daylight Hours Locate appropriate data, and
Friday 71 52
build a mathematical model for the hours of day-
Saturday 66 53 light at
Sunday 56 33 a. The Tropic of Capricorn
b. The Equator
Compare the actual high and low temperatures to the 25. Neurons To study the rhythmic synchronization
highest and lowest temperatures predicted by the of neurons, a team of scientists exposed locusts to
equation in Activity 19. Assuming that the coal and odors and measured the firing patterns of pairs of
7.3 Rates of Change and Derivatives 473
neurons in response to those odors. The table sale as pets. Because reptiles hibernate during the
located on the Calculus Concepts website gives the winter months, the number of reptiles gathered is
number of times per millisecond that the pair of periodic. The table located on the Calculus Con-
neurons simultaneously fired during the experi- cepts website shows the number of lizards gathered
ment when exposed to cherry odor, apple odor, and during two consecutive years in a certain western
a combination of the two. state.
The scientists claim that the period of neuron a. From an examination of the data, estimate the
activity is not influenced by a particular odor. Find amplitude, period, and vertical shift.
models for the data for cherry odor, apple odor, and
combined odor, and report the period of each b. Use technology to find a sine function that fits
model. Do your models seem to support the scien- the 2-year data. Compare the amplitude, period,
tists’ claim? Explain. and vertical shift of the equation with those you
estimated in part a.
26. Fishing An exclusive fishing club on the Res-
tigouche River in Canada kept detailed records on c. Examine a graph of the equation on a scatter
the numbers of fish caught by its members between plot of the data. Discuss how well the graph fits
1880 and 1930. Approximate data for the number the data.
of salmon caught per rod per day are located on d. Remove the months in which no lizards were
the Calculus Concepts website. In Cycles: The Science harvested, and renumber the remaining months
of Prediction, Dewey and Dakin assert that the from 1 to 16. Find a sine model for the modified
2 data. Compare the amplitude, period, and verti-
salmon data are periodic with a period of 9 3 years.
Find a sine model for the salmon data. Does your cal shift of this equation with those of the equa-
equation agree with the assertion that the period is tion in part b. Does the graph of this equation fit
2
the data significantly better than the one in
9 3 years? part b?
27. Lizards In desert areas of the western United
States, lizards and other reptiles are harvested for
7.3 Rates of Change and Derivatives
As something moves through a cycle, we are often interested in following the change
that takes place. Derivatives are one of the tools that can be used to measure how rap-
idly change occurs.
Derivative Formulas
As we seek a function to describe the rate of change of the sine function, we note that
because the sine curve is periodic, we would expect the rate-of-change function to be
periodic as well. We know that the rate of change of a sine graph is zero at the high
and low points on the graph. The values in the sine curve repeat in a regular manner,
so the zero slopes occur repeatedly. Thus the rate-of-change function of y sin x has
3 5
a sequence of repeating x-intercepts at 2 , 2 , 2 , 2 , and so on.
474 CHAPTER 7 Repetitive Change: Cyclic Functions
Consider the slopes between the high and low points of y sin x. Figure 7.16
indicates where the slopes are positive and where they are negative. In addition, the
blue dots in the figure mark the location of the inflection points. These are points of
greatest or least slope and correspond to high or low points on the derivative graph.
y = sin x
Slope = 0 Slope = 0
Neg
e
e
slop
slop
ativ
tive
itive
x
e slo
- 3
Posi
2 5
Pos
pe
2 2 2 2
-1
Slope = 0 Slope = 0
FIGURE 7.16
Sketching the slope graph results in the graph shown in Figure 7.17a.
dy for y = sin x
__ y = cos x
dx
x x
π
−_ _π π _
3π 2π _
5π π
−_ _π π _
3π 2π _
5π
2 2 2 2 2 2 2 2
−1
(a) (b)
FIGURE 7.17
The rate-of-change graph of y sin x is periodic with amplitude 1 and period
2. A careful numerical investigation reveals that the output is 1 at x 0, 1 at
3
x , and 0 at x 2 and at x 2 . The graph of the cosine function is shown in
Figure 7.17b. Note that it has the same characteristics as those previously stated for
the rate-of-change graph of the sine function. In fact, it can be shown mathematically
that the derivative of the sine function, y sin x, is the cosine function, y cos x.
7.3 Rates of Change and Derivatives 475
Derivative of the Sine Function
When x is a real number, the rate of change of the sine function is the
cosine function.
df
If f(x) sin x, then cos x.
dx
Derivative Rules
Now that we have a derivative formula for y sin x, we can apply the Constant Mul-
tiplier Rule, the Chain Rule, and the Sum Rule given in Chapter 3 to find slope for-
mulas for functions of the form f(x) a sin(bx h) k.
To illustrate the use of these rules, we consider the temperatures in Philadelphia
on August 27, 1993, as shown in Table 7.3.
TABLE 7.3
Time 1 A.M. 2 A.M. 3 A.M. 4 A.M. 5 A.M. 6 A.M.
Temperature (°F) 78 77 77 76 76 75
Time 7 A.M. 8 A.M. 9 A.M. 10 A.M. 11 A.M. 12 P.M.
Temperature (°F) 75 77 79 83 87 90
Time 1 P.M. 2 P.M. 3 P.M. 4 P.M. 5 P.M. 6 P.M.
Temperature (°F) 91 93 94 95 93 92
Time 7 P.M. 8 P.M. 9 P.M. 10 P.M. 11 P.M. 12 A.M.
Temperature (°F) 89 86 84 83 81 79
(Source: Philadelphia Inquirer, August 28, 1993.)
The data in Table 7.3 can be modeled as
t(x) 9.6 sin(0.285x 2.93) 84.2°F
x hours after midnight on August 26, 1993. Using the Constant Multiplier Rule (with
9.6 as the constant multiplier) and the Sum Rule (with the derivative of 84.2 being
zero), we find that a rate-of-change formula for this function is
d
t(x) 9.6 [sin(0.285x 2.93)] 0
dx
Now, consider u 0.285x 2.93 to be an inside function with the sine function
being the outside function. Using the Chain Rule, we have
d d
t(x) 9.6 (sin u) (0.285x 2.93)
du dx
9.6(cos u)(0.285)
2.736 cos(0.285x 2.93)°F per hour
x hours after midnight.
476 CHAPTER 7 Repetitive Change: Cyclic Functions
According to the model and its rate-of-change formula, the temperature at
noon on August 27, 1993, was t(12) 89°F, and it was increasing by t(12) 2.4°F
per hour.
EXAMPLE 1 Writing and Interpreting a Sine Function Derivative
Police Calls The calls for service made to a county sheriff’s department in a certain
rural/suburban county can be modeled as
c(t) 2.8 sin(0.262t 2.5) 5.38 calls
during the tth hour after midnight.
a. What is the average number of calls that the department receives each hour?
b. Find a formula for c(t).
c. How quickly is the number of calls received each hour changing at noon? at
midnight?
Solution
a. The constant k of a sine model f(x) a sin(bx h) k tells us the average (or
expected) value of the cycle. Thus the department receives an average of 5.38 calls
each hour during a day.
b. A rate-of-change model for calls received each hour can be found by using the
Constant Multiplier Rule, the Chain Rule, and the Sum Rule:
d
c(t) 2.8 [sin(0.262t 2.5)] 0
dt
Now, let u 0.262t 2.5, and proceed using the Chain Rule.
d d
c(t) 2.8 (sin u) (0.262t 2.5)
du dt
2.8(cos u)(0.262)
0.7336 cos(0.262t 2.5) calls per hour
t hours after midnight.
c. At noon the number of calls received each hour is changing by c(12) 0.6 call
per hour. At midnight the number of calls received each hour is changing by
c(0) 0.6 call per hour. [Note: You could use t 24 for midnight. If the cycle
has close to a 24-hour period, then c(0) should be approximately the same as
c(24).] ●
The Sum Rule is useful when we are finding the derivative of a sine function that
is added to another function. This type of function occurs when a phenomenon cre-
ates cycles of the same period and amplitude but the expected value of each cycle
changes according to some other function. Example 2 illustrates this type of model in
the context of atmospheric carbon dioxide.
7.3 Rates of Change and Derivatives 477
EXAMPLE 2 Using the Product and Chain Rules with a Sine Function
Carbon Dioxide Figure 7.18 appeared in a Scientific American article in July of
1990.
Carbon dioxide concentration (parts per million)
355
350
345
340
335
330
325
Fluctuations reflect seasonal variation.
320
Summertime low is caused by uptake of
315 CO2 by plants. Data were collected at
the Mauna Loa Observatory in Hawaii.
310
1958 1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990
Year
FIGURE 7.18
(Source: R. M. White, “The Great Climate Debate,” Scientific American, vol. 263, no. 1, July 1990, pp.
36–43. Copyright © 1990 by Scientific American, Inc. All rights reserved. Reprinted with permission.)
This figure illustrates the rise in the amount of atmospheric carbon dioxide (CO2)
along with the seasonal fluctuations. We wish to find a model for the CO2 concen-
tration. An examination of the data from 1959 to the present (readily available on
the Web) reveals that the data are cyclic with period 1 year. Using the equation
2
b 1, we find that b 6.283. The data show that the approximate location of the
increasing inflection point of each cycle is the beginning of each year; thus there is no
horizontal shift for each cycle. On the basis of these approximations, a model for the
yearly fluctuation in the amount of atmospheric CO2 is f(x) a sin(6.283x) d
parts per million, where x is the time in years measured from the beginning of any
year, a is the amplitude, and d is the vertical shift.
The amplitude of each cycle is slightly different, but using the data to estimate
amplitudes for the years between 1959 and 1989, we observe that there is a general
rise in amplitude. Modeling the amplitudes gives the equation
a 0.010345x 2.105676 parts per million
x years after the beginning of 1900.
478 CHAPTER 7 Repetitive Change: Cyclic Functions
We also note that the average amount of carbon dioxide (the constant d) is
increasing each year. The average value of a cycle of a sine function that is not shifted
horizontally occurs at an input of zero. By fitting a quadratic model to the CO2 con-
centration in January of each year for the years between 1959 and 1989, we find that
the average amount of CO2 can be described by the equation
d(x) 0.013783x2 0.844566x 316.913080 parts per million
x years after the beginning of 1900. Adding this equation to the fluctuation cycle as
the expected value and substituting the amplitude function for the constant a give
the model for atmospheric CO2 concentration as
c(x) (0.010345x 2.105676) sin(6.283x) 0.013783x2 0.844566x
316.913080 parts per million
x years after the beginning of 1900. The model applies for years between 1959 and
1999.
a. Use the model to estimate the amount of atmospheric CO2 in July of 1989.
b. Write a formula for the rate of change of the amount of atmospheric CO2.
c. Use the formula from part b to estimate the rate of change of the amount of
atmospheric CO2 in July of 1989.
Solution
a. There were approximately c(89.5) 351.8 parts per million of CO2 in the
atmosphere in July of 1989.
b. The derivative of c is
c(x) (0.010345x 2.105676)(6.283) cos(6.283x) 0.010345 sin(6.283x)
2(0.013783)x 0.844566 parts per million per year
x years after the beginning of 1900.
c. c(89.5) 17.4 parts per million per year. The amount of atmospheric CO2 was
decreasing by approximately 17.4 parts per million per year in July of 1989. ●
Finally, let us consider finding a formula for the derivative of a rate-of-change
formula for the sine model. This is the second derivative of the sine model. Because
the derivative of y sin x is y cos x, we must know the derivative of cos x in order
to find the second derivative of y sin x.
If we were to sketch the slope graph of the cosine function, the result would be the
solid graph shown in Figure 7.19.
y
1 y = sin x
x
−π π
−_ _π π _
3π 2π
2 2 2 d (cos x)
y = __
−1 dx
FIGURE 7.19
7.3 Rates of Change and Derivatives 479
This solid graph appears to be a reflection of y sin x across the horizontal axis. That
is, the slope graph of y cos x appears to be y sin x. In fact, using more rigorous
mathematics, it can be proved that if f(x) cos x , then its derivative is
f (x) sin x.
Derivative of the Cosine Function
When x is a real number, the rate of change of the cosine function is the
negative of the sine function.
dg
If g(x) cos x, then sin x.
dx
We use this fact when writing the second derivative of a sine function. For example,
the model for Philadelphia temperatures x hours after midnight,
t(x) 9.6 sin(0.285x 2.93) 84.2°F
has the derivative
t(x) 2.736 cos(0.285x 2.93)°F per hour
and its second derivative is
t(x) 0.77976 sin(0.285x 2.93)°F per hour per hour
Note that in the first and second derivatives of t, the period and horizontal shift are
the same as for t, whereas the amplitude and vertical shift are different.
EXAMPLE 3 Writing and Interpreting a Cosine Function Derivative
Carbon Dioxide Recall from Example 2 that a model for the amount of atmos-
pheric carbon dioxide is
c(x) (0.010345x 2.105676) sin(6.283x) 0.013783x2 0.844566x
316.913080 parts per million
x years after the beginning of 1900. The model applies for years between 1959 and
1989.
a. Write a formula for the second derivative of this function.
b. Find and interpret c(80).
Solution
a. The derivative of c was found in Example 2 to be
c(x) (0.010345x 2.105676)(6.283) cos(6.283x) 0.010345 sin(6.283x)
2(0.013783)x 0.844566 parts per million per year
480 CHAPTER 7 Repetitive Change: Cyclic Functions
x years after the beginning of 1900. This derivative was found using the Product
Rule, and the Product Rule is also required for the second derivative.
c(x) (0.010345x 2.105676)(6.283)(6.283)[sin(6.283x)]
(0.010345)(6.283) cos(6.283x)
0.010345(6.283) cos(6.283x) 2(0.013783)
(0.010345x 2.105676)(6.283)(6.283)[sin(6.283x)]
2(0.010345)(6.283) cos(6.283x)
2(0.013783) parts per million per year per year
x years after the beginning of 1900.
b. c(80) 1.87 parts per million per year per year. At the beginning of 1980, the
rate of change of the amount of atmospheric carbon dioxide was increasing by
about 1.9 parts per million per year per year. ●
Even though in this section we have illustrated only some of the general derivative
rules, any previously developed rule for derivatives of functions applies to the sine
and cosine functions. Recall also that if you have a graph and not a function formula,
then you can estimate rates of change by sketching tangent lines. This is illustrated in
Example 4.
EXAMPLE 4 Estimating Rates of Change from a Graph
Temperature Figure 7.20 shows the graph of a function m giving the normal daily
mean temperature, based on the 30-year period 1961 through 1990, for Omaha,
Nebraska.
m(x)
(°F)
80
70
60
50
40
30
20
10
x
1 3 6 9 12
Month
(Jan) (Mar) (Jun) (Sep) (Dec)
FIGURE 7.20
(Source: Based on data from U.S. National Oceanic and Atmospheric
Administration, Climatography of the United States, No. 81.)
7.3 Rates of Change and Derivatives 481
a. Estimate the maximum difference in daily mean temperature. Estimate the aver-
age rate of change between the lowest daily mean temperature and the highest
daily mean temperature.
b. Estimate where the daily mean temperature is increasing most rapidly.
c. Estimate the rate of change of the daily mean temperature at the point where the
temperature is increasing most rapidly.
d. Interpret each of the following:
i. m(9)m(2) 38.1
ii. m(8) 7.3
Solution
a. The maximum difference in daily mean temperature is the difference between the
highest daily mean temperature and the lowest daily mean temperature.
According to Figure 7.20, the highest temperature is approximately 77°F (in July),
and the lowest temperature is approximately 21°F (in January). The maximum
difference is approximately 56°F. The average rate of change of daily mean
77 21
temperature between January and July is 7 1 9.3°F per month.
b. The daily mean temperature is increasing most rapidly at the inflection point of
this graph. This (according to Figure 7.20) occurs near x 4, which corresponds
to April.
c. We estimate the rate of change of daily mean temperature in April by carefully
drawing the tangent line to the graph at x 4 and approximating its slope. (See
Figure 7.21.)
m(x)
(°F)
80
70
60
50
40
Rise ≈ 29°F
30
20
Run = 2 months
10
x
1 3 6 9 12
Month
(Jan) (Mar) (Jun) (Sep) (Dec)
FIGURE 7.21
The daily mean temperature during April was increasing by approximately
29°F
2 months 14.5°F per month.
482 CHAPTER 7 Repetitive Change: Cyclic Functions
d. i. The change in the daily mean temperature between February and September
was 38.1°F.
ii. In August, the daily mean temperature was decreasing by 7.3°F per month. ●
We now have the tools we need not only to find rates of change of a sine model
but also to use rates of change to locate maxima, minima, and inflection points on
sine models. This is the topic of the next section.
7.3 Concept Inventory south of the equator at time x minutes since the
satellite was launched.
• Derivatives: If y sin x, then y cos x.
If y cos x, then y sin x. y
(kilometers)
• Second derivatives: If y sin x, then y sin x.
If y cos x, then y cos x. 3050
North of 2050
equator
7.3 Activities 1050
150 x
0
Getting Started (minutes)
18 60 78 138
In Activities 1 through 6, find formulas for the indicated South of
-1050
rates of change. equator
-2050
1. f(x) sin 3x 10.456, f(x)
-3050
dg
2. g(x) cos 0.45x 3.763,
dx a. How long does it take the satellite to complete
dt one orbit of Earth?
3. t(r) 5.2 cos(0.45r ) 80r 6.34,
dr b. What is the greatest distance south of the equa-
tor that the satellite ever reaches?
4. f(t) 3 sin(0.45t 3) 12.8t3 4.6t, f (t)
c. Carefully draw the tangent line at the time the
5. h(x) 2.08 sin(0.16x 12.3) satellite has been in orbit for 1 hour.
3.58 cos[1.35(2x 6)]4.79,
d. Use the tangent line to estimate the rate of
h(x) and h(x)
change of the satellite’s distance from the equa-
tor after it has been in orbit for 1 hour.
6. y(x) 0.023 sin(2x 7) 4 cos[3(0.5 x)],
y(x) and y(x) e. Is the answer to part d the speed of the satellite
after it has been in orbit for 1 hour? Explain.
Applying Concepts 8. Turbine Power The accompanying figure shows
the power output of a diesel engine in an irrigation
7. Satellite The graph shows a model of the distance pumping system that is supplemented by a wind
in kilometers that a weather satellite is north or turbine when the wind speeds are sufficient.
7.3 Rates of Change and Derivatives 483
Power U.S. mass
output transportation
(kW) (billion trips)
9.5
50
40 9
30
8.5
20
10
8
Wind speed
0
3 5 7 9 11 13 15 17 (m/sec)
7.5 Year
(Source: Based on data from B. A. Stout, Handbook of Energy 1992 1996 2000
for World Agriculture, London: Elsevier Applied Science, 1990.)
(Source: Based on data from the American Public Transporta-
a. Estimate the change in power output when the tion Association.)
wind speed changes from 5 m/sec to 10 m/sec a. Use tangent lines to estimate the rates of change
and from 10 m/sec to 15 m/sec. of the number of trips in 1992, 1996, and 2000.
b. Report the estimates in part a as percentage b. Use a tangent line approximation to estimate the
changes and average rates of change. number of trips in 2001.
c. Estimate the instantaneous rate of change in
10. Temperature The accompanying figure shows a
power output for wind speeds of 5, 10, and
temperature graph for a location in the south Gobi
15 m/sec. Convert your answers to percentage
Desert over a period of 2 years.
rates of change. Interpret your answers.
d. Why are the rates of change negative? What a. Estimate average rates of change in temperature
would you expect the rates of change to be for from
the power output of the wind turbine as a func- i. July through July
tion of wind speed? ii. February through February
e. The graph was generated from a sine function, iii. July through December of the first year
which is an excellent fit for the power output
b. Estimate during which month the temperature
data. Comment on the appropriateness and lim-
itations of using a sine model in this context. i. increased most rapidly
ii. decreased most rapidly
9. Mass Transit Consider the accompanying graph
showing the number of trips on U.S. mass trans- iii. increased most slowly
portation systems each year between 1992 and 2000. iv. decreased most slowly
Temperature
(°C)
30
20
10
Month
J F MAM J J A S O N D J F MAM J J A S O N D
-10
-20
(Source: Based on data from Neil E. West, ed., Ecosystems of the World: Temperate Deserts
and Semi-Deserts, Amsterdam: Elsevier Scientific Publishing Company, 1983.)
Figure for Activity 10
484 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
11. Radiation A graph of the average amount of c. Estimate the profit rate in 2003, assuming that
extraterrestrial radiation in Amarillo, Texas, for the rate of change of the profit rate remained
each month of the year is shown. The amount is constant in 2002.
measured in equivalent millimeters of water evapo- 13. Sales In order to smooth out seasonal work loads,
ration per day. Month 1 corresponds to January, the Campbell Soup Company produces both
month 12 to December. canned soup and powdered drink mix. The 1992
daily sales can be modeled by
Radiation c(x) 1.8961 sin(0.0186x 1.1801) 5.4660
(mm/day) million cans (16 ounce) of soup
17 p(x) 7.3026 cos(0.0197x 3.7526) 20.3728
15
million pints (reconstituted) of drink mix
13 where x is the day of the year. Assume that soup is
sold at an average price of $0.70 per can and drink
11
mix is sold at $0.20 per pint.
9 (Source: Based on information in Wall Street Journal, July 7,
1993, p. B1.)
7 Month
1 3 5 7 9 11 a. Find a function giving the revenue from the
(Jan) (Mar) (May) (Jul) (Sep) (Nov)
combined 1992 daily sales as a function of the
(Source: Based on data from A. A. Hanson, ed., Practical day of the year on which soup and powdered
Handbook of Agricultural Science, Boca Raton, FL: CRC
drink mix are produced.
Press, 1990.)
b. Give a function that models the rate of change of
revenue from combined 1992 sales.
a. Use the graph to estimate the change in the c. How quickly was revenue changing on February
amount of radiation from June through 15, 1992?
November. d. At what other times during 1992 was revenue
decreasing at the same rate as that found in part c?
b. Report the change in part a as a percentage
change and an average rate of change. e. Give one time during 1992 at which revenue was
increasing at the same rate at which it was
c. Use the graph to estimate the rate of change in decreasing in part c.
March.
14. Temperature A model for the normal mean air
d. Convert the rate of change in part c to a percent- temperature in Fairbanks, Alaska, is
age rate of change.
f(x) 37 sin[0.0172(x 101)] 25°F
12. Profit Rate The after-tax profit rate on invest-
ment is defined as the ratio of after-tax corporate where x is the number of days since the last day of
profits to corporate stock of plant and equipment. last year. Estimate how rapidly the normal mean
The after-tax profit rate on investment in the United temperature in Fairbanks is changing on March 15
States between 1988 and 2002 can be approximated and on December 31.
(Source: B. Lando and C. Lando, “Is the Curve of Temperature
by T(x) 1.204 sin(0.469x 2.293) 6.352 per-
Variation a Sine Curve?” The Mathematics Teacher, vol. 7,
cent, where x is the number of years after 1988. no. 6, September 1977, p. 535.)
(Source: Based on data from Bureau of Economic Analysis,
Business Week, November 4, 2002, p. 107.) 15. Temperature The normal daily high temperature
in Boston can be modeled by
a. Find a function that gives the rate of change of
the profit rate. B(m) 22.926 sin(0.510m 2.151) 58.502°F
b. Estimate the rate of change of the profit rate in at the end of the mth month of the year. Estimate
2000 and 2002. how rapidly the normal daily high temperature in
7.3 Rates of Change and Derivatives 485
Boston is changing at the end of November and in through October? Is this value an accurate repre-
the middle of May. sentation of how extraterrestrial radiation
(Source: Based on data from U.S. National Oceanic and Atmos- changed during these months? Explain.
pheric Administration, Climatography of the United States,
c. Find R(t) for t 4 and t 10. Also express
no. 81.)
your answers as percentage rates of change.
16. Sales Ice cream cones are sold at a county fair. Interpret your answers.
The total number of cones sold since the booth 19. Mortality The table shows the weekly pneumo-
opened at 8 A.M. is approximated by nia death rate (number of deaths per 100,000 peo-
c(t) 80.153 sin(0.511t 1.680) 87.564 cones ple per week) in the United States from 1923
through 1925.
where t is the number of hours since 8 A.M. Esti-
mate how rapidly the number of cones sold is
Weeks since Pneumonia Weeks since Pneumonia
changing at 12 P.M. and 8 P.M. Jan. 1, 1923 death rate Jan. 1, 1923 death rate
17. Turbine Power The power output of a diesel 0 100 80 60
engine in an irrigation pumping system that is sup- 10 210 90 45
plemented by a wind turbine can be modeled by 20 95 100 95
P(s) 20.204 sin(0.258s 0.570) 28.119 kilowatts 30 35 110 160
when the wind speed is s meters per second. 40 55 120 140
(Source: Based on data from U.S. National Oceanic and Atmos- 50 85 130 80
pheric Administration, Climatography of the United States,
no. 81.) 60 140 140 40
a. Find the change in power output when the wind 70 160 150 90
speed changes from 6 m/sec to 10 m/sec, from (Source: Estimated from E. R. Dewey and E. F. Dakin, Cycles:
10 m/sec to 14 m/sec, and from 14 m/sec to The Science of Prediction, New York: Holt, 1947.)
18 m/sec.
a. Find a sine model for the pneumonia death rate
b. Report the answers to part a as percentage
data. Why would such data be periodic?
changes and average rates of change.
b. Use the sine model to estimate the rate of change
c. Find a formula for P(s).
of the pneumonia death rate in the middle of
d. Find P(s) for s 6, s 10, s 14, and s 18. 1924 and at the end of 1924.
Interpret your answers.
20. Tides The February 1995 tide chart for the
e. Convert the answers to part d to percentage rates
Savannah River entrance notes the times of high
of change.
and low tides. A portion of the tide chart is shown
18. Radiation The average amount of extraterrestrial in the table on page 486. Assume that the tidal vari-
radiation in Amarillo, Texas, can be modeled by ation is 4 feet; that is, it varies from 2 to 2 feet
around mean sea level.
R(t) 5.059 sin(0.461t 1.287) 12.151 (Source: Sea Island Scene of Beaufort, Beaufort, SC: Sands Pub-
equivalent mm of water evaporation per day lishing Co., 1995.)
where t 1 in January and t 12 in December. a. On the basis of the data with mean sea level at
(Source: Based on data in A. A. Hanson, ed., Practical Hand- 0 feet, what would you expect the period, ampli-
book of Agricultural Science, Boca Raton, FL: CRC Press, 1990.) tude, and average water height to be if the data
a. By how much does the amount of extraterrestrial are modeled by a sine function?
radiation change between April and October? b. Use the data for March 2 through March 4 to
What is the percentage change? find a sine model for the height of the water as a
b. How quickly (on average) is the amount of function of time.
extraterrestrial radiation changing from April c. How fast is the water falling at noon on March 2?
486 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
Feb 26 Feb 27 Feb 28 Mar 1 Mar 2 Mar 3 Mar 4
Savannah River Entrance
HI 0522 LO 0003 LO 0056 LO 0145 LO 0231 LO 0313 LO 0353
LO 1146 HI 0619 HI 0708 HI 0753 HI 0835 HI 0915 HI 0955
HI 1739 LO 1238 LO 1326 LO 1411 LO 1453 LO 1532 LO 1609
Tidal Chart
HI 1833 HI 1922 HI 2006 HI 2047 HI 2128 HI 2209
Mar 5 Mar 6 Mar 7 Mar 8 Mar 9 Mar 10 Mar 11
LO 0431 LO 0509 LO 0549 HI 0024 HI 0115 HI 0208 HI 0304
HI 1036 HI 1120 HI 1206 LO 0634 LO 0728 LO 0831 LO 0934
LO 1646 LO 1724 LO 1805 HI 1255 HI 1347 HI 1442 HI 1539
HI 2251 HI 2336 LO 1853 LO 1950 LO 2053 LO 2154
Table for Activity 20
d. How fast is the water falling at noon on March 4? The functions in Activities 22 through 27 are composite
e. If a boat needs at least 1 foot of water above functions. For each function, determine the inside and
mean sea level in order to float away from being outside function, and give the formula for the derivative.
grounded at low tide, at what times is it possible 22. y 2.43 sin x 12
for this boat to float away on March 3?
23. y sin(2.4x)
21. Sun Declination The declination of the sun is the
angle of the sun as measured from the equator. The 24. y sin(2x 3)
table shows the angle of declination of the sun in 25. y 4(sin x 7)2 8(sin x 7) 13
Greenwich, England, for selected days throughout
the year. These values vary slightly from year to year. 26. y sin(3x2 19x)
27. y ln(sin x)
Day of Declination Day of Declination
the year of sun the year of sun 28. Sales The sales for a retail women’s clothing store
1 23.02° S 182 23.05° N for 5 years are given in the table located on the Cal-
culus Concepts website.
32 17.15° S 213 17.82° N
a. Why should the December data be omitted?
60 7.17° S 244 8.00° N
b. Find a model for the 5-year data set (omitting the
91 4.83° N 273 3.48° S
December data).
121 15.30° N 304 14.67° S
c. Use the model to estimate the rate of change of
152 22.09° N 334 21.92° S sales in April of both years.
(Source: Old Farmer’s Almanac, 1996.)
29. Sales The table located on the Calculus Concepts
a. Considering southern declinations to be negative website shows the monthly sales for an ice cream
and northern declinations positive, find a sine company for a 4-year time period.
model for the data. State and interpret the am- a. Find a sine model for the data. Comment on how
plitude and period of the function. well it fits the data.
b. Use the equation to find the rate of change of b. Use the data in the table to find the average of the
the declination of the sun at the two equinoxes high and low sales each year.
(when the sun is directly on the equator). c. Find a model for the average sales found in part
c. What is the relationship between the rate of b. Use as input the first month of each year in
change of the declination of the sun and the which sales would be expected to reach the aver-
summer and winter solstices? age value.
7.4 Extrema and Points of Inflection 487
d. Replace the constant d in the function in part a Discussing Concepts
with the function for average value found in part
c. Comment on how well this new function fits 30. What happens to the slope of a graph of the sine
the data compared with the function in part a. function f(x) a sin(bx c) d when f is at a
maximum? at a minimum? equal to its expected
e. Use the model in part d to estimate the rate of (average) value?
change of sales in September of the third year and
January of the fourth year. Interpret your answers.
7.4 Extrema and Points of Inflection
When describing a cycle, we must note the height of its peaks and the depth of its
valleys. The height and depth between which a cycle oscillates determine its ampli-
tude. How often the maximum values and minimum values repeat depends on how
often the cycle repeats. We might also want to find the points at which the concavity
of a cycle changes, because inflection points often have interesting interpretations in
context.
Determining Extreme Points
Recall from Section 4.2 that if a function has a relative maximum or a relative mini-
mum at an input c, then either f (c) 0 or f (c) does not exist but f(c) does exist.
We also know that if the derivative is zero and the slope graph crosses (not just
touches) the horizontal axis at a certain input, then a relative maximum or a relative
minimum exists at that input. When we are using sine or cosine functions of the
forms discussed in this chapter, the derivative exists at all input values, so in our
search for extreme points, we need consider only the inputs where the derivative of
the function crosses the input axis.
EXAMPLE 1 Locating Extreme Points Using Calculus
Sales Weekly sales of heating oil are cyclic with a winter peak and a summer lull. A
model for the weekly heating oil sales for a certain company is
f(t) 2000 sin(0.1208t 1.5707) 3000 gallons
t weeks after the beginning of the year.
a. Find the period of the heating oil sales model.
b. Sketch a graph of one cycle of f, and mark the locations of maximum and min-
imum heating oil sales.
c. Use derivatives to find the maximum and minimum of a cycle.
Solution
2
a. The period of the heating oil sales model is 0.1208 52 weeks.
488 CHAPTER 7 Repetitive Change: Cyclic Functions
b. A graph of the weekly heating oil sales for a 1-year period is shown in Figure 7.22.
f(t)
(gallons)
Maximum
(0, 5000)
5000
4000
3000
Minimum
2000 (26, 1000)
1000
t
0 10 20 30 40 50
(weeks)
FIGURE 7.22
c. The derivative of f is
f (t) 241.6 cos(0.1208t 1.5707) gallons per week
t weeks after the beginning of the year. By setting the derivative equal to zero and
solving for t, we find that f (t) 0 when t 0 and again when t 26. The graph
in Figure 7.22 shows that the second solution gives the minimum. Heating oil
sales reach a maximum of f(0) 5000 gallons at the beginning of January and
reach a minimum of f(26) 1000 gallons at the beginning of July. Sales again
reach the maximum at 52 weeks. ●
Being able to solve for the values at which the first derivative of a sine or cosine
function is zero helps us locate maxima and minima. We use a similar technique
when we are looking for inflection points.
Inflection Points
Recall from Section 4.3 that at a point of inflection on the graph of a function, the
second derivative is zero or does not exist. If the second derivative graph crosses the
horizontal axis, then an inflection point occurs on the graph of the function at that
input value. Let us again consider the model for heating oil sales.
EXAMPLE 2 Locating Inflection Points Using Calculus
Sales A model for weekly heating oil sales is
f(t) 2000 sin(0.1208t 1.5707) 3000 gallons
t weeks after the beginning of the year. A graph of f is shown in Figure 7.22 in
Example 1.
a. Use the graph in Figure 7.22 to estimate the locations of the inflection points.
b. Use derivatives to find the inflection points.
c. What would be the best course of action for the sales manager during the time
before and after each of the inflection points?
7.4 Extrema and Points of Inflection 489
Solution
a. The graph of weekly heating oil sales appears to be concave down from week 0
through week 13, concave up from week 13 through week 39, and concave down
from week 39 through week 52. Thus we estimate that inflection points occur at
the beginning of April (week 13) and at the beginning of October (week 39).
b. In order for inflection points to occur on a sine or cosine graph of the forms used
in this chapter, the second derivative of the function must be zero. The second
derivative of the heating oil function is
f (t)
29.18528 sin(0.1208t 1.5707) gallons per week per week
t weeks after the beginning of the year. By setting f (t) 0 and solving for t, we
find that inflection points occur at t 13.0 and t 39.0. In both cases, 3000 gal-
lons of heating oil are sold, so the inflection points are (13, 3000) and (39, 3000).
The point of most rapid decrease occurs at the 13th week, and the point of most
rapid increase occurs at the 39th week. (See Figure 7.22.)
c. Between January 1 and April 1, the oil manager should be concentrating on
maintaining supply, making timely deliveries, and advertising the product. From
April 1 until October 1, heating fuel demand is in the summer doldrums, and the
company should be running promotion specials to move the product. Between
October 1 and December 31, the manager should focus on ensuring that there is
enough heating oil available to meet the demand. ●
Of course, in the real world there are rarely such idealized models as that in the
previous example, but managers routinely talk about changes in the rate of increase
or rate of decrease. Through those observations, they try to determine when funda-
mental shifts in demand occur. In mathematical terms, they are looking for the
inflection points.
The connection between relative maxima and minima and inflection points of a
function and the zero values of the first and second derivatives is what makes the use
of calculus concepts so powerful in the analysis of cycles.
7.4 Concept Inventory a. Temporarily ignoring the context, find the max-
imum and minimum values of the function M
• Maxima and minima of sine and cosine models and the inputs at which those values occur for
• Inflection points of sine and cosine models one period of the function beginning at x 0.
b. Find the maximum and minimum values of the
function M and the years between 1992 and 2000
when they occur. Interpret your answers.
7.4 Activities
2. Fishing An exclusive fishing club on the Res-
tigouche River in Canada kept detailed records of
Applying Concepts
the number of fish caught by its members. A model
1. Mass Transit The number of trips on U.S. mass for the number of salmon caught per rod per day
transportation systems each year between 1992 between 1880 and 1905 is
and 2000 can be modeled by
s(y) 0.327 sin(0.675y 2.122) 1.267
M(x) 0.932 sin(0.467x 2.94) 8.737 billion trips salmon per rod per day
x years after 1992. y years after 1880.
(Source: Based on data from the American Public Transportation (Source: Based on data estimated from E. R. Dewey and E. F.
Association.) Dakin, Cycles: The Science of Prediction, New York: Holt, 1947.)
490 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
a. Find the nearest year after 1905 in which, Radiation Radiation
according to the model, the number of salmon is (mm of (mm of
expected to peak. Month H2O per day) Month H2O per day)
b. Find the nearest year after 1905 in which, Jan 8.3 July 16.8
according to the model, the number of salmon is Feb 10.2 Aug 15.6
expected to be at a minimum.
Mar 12.8 Sep 13.6
c. How might the manager of the fishing club have Apr 15.0 Oct 11.2
used the information in parts a and b to benefit
May 16.5 Nov 9.0
the club?
June 17.0 Dec 7.8
3. Temperature The table gives the normal daily
(Source: A. A. Hanson, ed., Practical Handbook of Agricultural
mean temperatures, based on the 30-year period
Science, Boca Raton, FL: CRC Press, 1990.)
1961 through 1990, for Omaha, Nebraska.
(Source: U.S. National Oceanic and Atmospheric Administra- a. Use only these data to estimate the maximum and
tion, Climatography of the United States, No. 81.)
minimum radiation levels and when those levels
occur.
b. Find a sine model for the data, and determine
Daily mean Daily mean the highest and lowest points on the graph.
temperature temperature
Interpret these points in the context of radiation.
Month (°F) Month (°F)
c. How might knowledge of high and low radiation
Jan 21.1 July 76.9
levels be useful to someone living in Amarillo?
Feb 26.9 Aug 74.1
5. Price Index The accompanying figure shows
Mar 38.6 Sept 65.1 graphs of w, the indexed price of wheat in the United
Apr 51.9 Oct 53.4 Kingdom, and a sine function, i, for the index.
May 62.4 Nov 39.0
June 72.1 Dec 25.0 75
80 w
85
90
95
a. Use only these data to estimate the maximum and 100
minimum daily mean temperatures for Omaha. 105 i
110
b. Find a sine model for the data, and determine 1770 1786 1802 1818 1834 1850 1866
115
the highest and lowest points on the graph. 120
Interpret these points in the context of Omaha’s 125
weather.
c. How do the answers obtained with the equation Eight-year cycle in Poynting’s
index numbers of the price of wheat
compare with those found using the data? What
(scale inverted)
relationship is there between the maximum and
(Source: From H. L. Moore, Generating Economic Cycles, New York:
minimum daily mean temperatures for Omaha Macmillan, 1923; repr. 1967, Augustus M. Kelley Publishers, p. 75.
found in parts a and b and the actual yearly high Reprinted by permission.)
and low temperatures in a given year?
a. How many of the maxima of the sine graph
4. Radiation The average amount of extraterres- coincide with a relative maximum in the wheat
trial radiation in Amarillo, Texas, for each month price index? Note that the vertical axis is inverted.
of the year is given in the table. The amount is b. How many of the minima of the sine graph coin-
measured in equivalent millimeters of water evap- cide with a relative minimum in the wheat price
oration per day. index?
7.4 Extrema and Points of Inflection 491
c. An equation for the sine graph is y 100.2 c. Reading carefully from the chart, estimate the
62
0.4 sin 4 t 45 , where t is the number of years
hours of daylight at the beginning of each
month, and create a scatter plot of these data.
since 1761. Use this equation to find the years in Explain why it makes sense to use a sine model
which the next maximum and minimum occur
for this data set.
after 1874.
d. Find a sine function to fit the data. Compare the
d. On the basis of the graph and your answers to period, amplitude, and vertical shift of the func-
parts a and b, how accurately do you think the tion to your estimates in part b.
years you found in part c predict when the next
maximum and minimum wheat price indices 8. Daylight Hours Refer to the chart of the hours of
occurred? Do you think the sine model should daylight in Anchorage, Alaska, given in Activity 7.
be used to predict the next maximum and mini- (A function was found in part d.)
mum price index? a. The amount of daylight contributes to one’s
moods. In particular, less daylight has been
6. Price Index A model for the Sauerbeck index of
known to cause “winter blues.” Also important
general wholesale prices in England between 1818
to mood is a person’s perception of future
and 1914 is
change. For example, when the number of hours
3.69 x 100
23 of daylight is increasing rapidly, some people
p(x) 3.1 sin
45 may develop an optimistic outlook, whereas a
pessimistic outlook may result when the number
x years after 1818. Use the model to estimate the of hours of daylight is declining rapidly. Use the
years in which the next high and low wholesale sine model to find when the number of hours of
prices occurred after 1914. daylight is increasing most rapidly and when it is
(Source: From H. L. Moore, Generating Economic Cycles,
decreasing most rapidly in Anchorage.
New York: Macmillan, 1923; repr. 1967, Augustus M. Kelley
Publishers, p. 75.) b. If you were directing a public campaign, you
would not want to introduce the campaign when
7. Daylight Hours The Anchorage Daily News in An-
the public was in a pessimistic winter mood. Use
chorage, Alaska, includes in its weather section a
your answers to part a to estimate when politi-
chart of the hours of daylight. A copy of the news-
cians in Alaska should begin a new campaign.
paper chart is shown.
9. Daylight Hours Consider the number of daylight
WEATHER hours in your hometown.
Daylight Hours a. Find a function (or find data and create a func-
20 tion) for the number of daylight hours in your
hometown.
16 June 21
Summer b. Repeat Activity 8 for your hometown.
Hours
12 Solstice
For Today: 10. Carbon Dioxide A sine model for the amount of
8 Sunrise: 6:13 A.M. carbon dioxide in the atmosphere was given in Sec-
Sunset: 9:53 P.M. Dec. 21
4
Total: 15 hrs. 40 min. Winter tion 7.3. We could also have used the following co-
Loss: 5 min. 33 sec. Solstice sine model:
0
J F M A M J J A S O N D
c(x) (0.010345x 2.105676)cos 6.283x
(Source: Used with the permission of the Anchorage Daily News.) 2
0.013783x2 0.844566x 316.913080
a. There are two graphs shown in the chart. Which parts per million
graph represents daylight hours, and what does x years after the beginning of 1900. The model
the other graph represent? applies for years between 1959 and 1999.
b. On the basis of the graph of daylight hours, esti- (Source: Based on data from R. M. White, “The Great Climate
mate the period, amplitude, and vertical shift. Debate,” Scientific American, vol. 263, no. 1, 1990, pp. 36–43.)
492 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
a. Graph the sine function given in Example 3 in The after-tax profit rate on investment in the United
Section 7.3 and the cosine function here. Are States between 1988 and 2002 can be approximated
there any differences between these two graphs? by T(x) 1.204 sin(0.469x 2.293) 6.352 per-
Why can either a sine function or a cosine func- cent, where x is the number of years after 1988.
tion be used? (Source: Based on data from the Bureau of Economic Analysis,
BusinessWeek, November 4, 2002, p. 107.)
b. Write a formula for the derivative of the cosine
function. a. Write a function for the rate of change of the
c. Use c and its derivative to find the maximum after-tax profit rate on investment.
and minimum carbon dioxide levels in 1960, b. Use calculus to find the inflection points and
1970, 1980, and 1990. In what months do these absolute extrema over one period of the function
maximum and minimum levels occur? T. Interpret your answers.
d. Why are the maxima different for each year? c. Graph the function T for x between 0 and 14. Use
the graph and the results of part b to explain the
11. Radiation A model of the data for the amount of
meaning of the phrase profit boom of the 1990s.
ultraviolet radiation in Auckland, New Zealand, is
14. Gender Ratio The number of males per 100 fe-
a(m) 27.1 sin(0.485m 1.707) 32.9
males in the United States between 1900 and 2000
watts per cm2
can be modeled by
when m 1 in January, m 2 in February, and so
m(t) 5.334 sin(0.0426t 1.265) 99.94
on.
males per 100 females
(Source: Based on data from W. Rudloff, World-Climates,
Stuttgart, Germany: Wissenschaftliche Verlagsgesellschaft, t years after 1900.
1981.)
(Source: Based on data from the U.S. Bureau of the Census.)
a. Write a function for the rate of change of the a. Find the maximum and minimum values of the
amount of ultraviolet radiation. function and the years between 1900 and 2000 in
b. Estimate the time when ultraviolet radiation is which they occur. Interpret your answers.
at its highest level and the time when it is at its
b. Determine when the graph is decreasing most
lowest level.
rapidly and how rapidly it is decreasing at that
c. How much ultraviolet radiation is received at time. Interpret your answers.
the times found in part b?
15. Star Pulses On December 15, 1995, a team of
12. Beetles Based on data for 1965 through 1976, the astronomers discovered X-ray pulses emitting from
reduction in the number of beetles in a certain geo- what they believe to be a neutron star. The speed of
graphical area in the Netherlands can be modeled the pulses (in counts per second) can be modeled by
by the equation
p(s) 40.5 sin(0.01345s 1.5708) 186.5
R(x) 100.9 sin x
3
6.5
3
3.6 beetles counts per second
after s milliseconds.
x years after 1960. (Source: Based on information in M. H. Finger et al., “Discov-
(Source: Based on data from P. J. denBoer and R. Reddingius, ery of Hard X-ray Pulsations from the Transient Source GRO
Regulation and Stabilization Paradigms in Population Ecology, J1744–28,” Nature, vol. 381, May 23, 1996, pp. 291–292.)
London: Chapman & Hall, 1996.)
a. What is the period of the pulsations?
a. What is the period of this function?
b. Determine two times when the pulse speed was
b. Use calculus to find the extrema of this function highest and two times when it was lowest. What
over one period. Interpret your answers. were the pulse speeds at those times?
13. Profit Rate The after-tax profit rate on invest- c. Find and interpret one inflection point where
ment is defined as the ratio of after-tax corporate the function is increasing and one inflection
profits to corporate stock of plant and equipment. point where the function is decreasing.
1
7.4 Extrema and Points of Inflection 493
16. Turbine Power The combined power output of a might cause the actual peak to differ from the
diesel engine and wind turbine in an irrigation one found by the equation?
pumping system can be modeled by c. Find and interpret the first inflection point lo-
P(s) 20.204 sin(0.258s 0.570) cated beyond the data given.
19.026 sin(0.247s 2.549) 44.952 kilowatts 18. Carbon Dioxide Refer to the cosine model given
when the wind speed is s meters per second. The in Activity 10 for the amount of carbon dioxide in
model should be used for wind speeds between 6 the atmosphere to answer the following questions.
and 18 meters per second, because the wind turbine a. Write a formula for c.
produces power only if wind speeds exceed 6 meters b. Determine in what months of 1960, 1970, 1980,
per second. and 1990 the carbon dioxide level was increasing
(Source: Based on information in B. A. Stout, Handbook of
most rapidly.
Energy for World Agriculture, London: Elsevier Applied
Science, 1990.) c. According to the function c, what was the carbon
a. Find the wind speeds associated with the great- dioxide level at the time of most rapid increase in
est and the least power output for this pumping 1960, 1970, 1980, and 1990? Explain why these
system. values are increasing from 1960 through 1990.
b. On the basis of your answer to part a, what do 19. Radiation Refer to the model given in Activity 11
you believe is the benefit of having a wind tur- for the amount of ultraviolet radiation in Auckland,
bine in this pumping system? New Zealand, to answer the following questions.
c. Find all inflection points of P between s 6 and a. Write a function for the second derivative of the
s 18. Interpret your answers. ultraviolet radiation model.
17. Mortality The table shows the weekly pneumonia b. Estimate the time when ultraviolet radiation is
death rate (number of deaths per 100,000 people per increasing most rapidly and the time when it is
week) in the United States from 1923 through 1925. decreasing most rapidly.
c. How much ultraviolet radiation is received at the
times found in part b?
Weeks since Pneumonia Weeks since Pneumonia
Jan 1, 1923 death rate Jan 1, 1923 death rate 20. Sun Declination The declination of the Sun in
0 100 80 60 Greenwich, England, can be described by the
10 210 90 45 equation
20 95 100 95 D(d) 23.677 sin(0.0168x 1.312) 0.292 degrees
30 35 110 160 on the xth day of the year. Negative output values
40 55 120 140 correspond to southern declinations, and positive
50 85 130 80 output values correspond to northern declinations.
(Source: Based on data from the Old Farmer’s Almanac, 1996.)
60 140 140 40
a. Use calculus to find the inflection points and ex-
70 160 150 90
trema over one period of the function. Interpret
(Source: Estimated from E. R. Dewey and E. F. Dakin, Cycles: The your answers.
Science of Prediction, New York: Holt, 1947.)
b. Relate your answers in part a to Earth’s
equinoxes and solstices.
a. Find a sine function to fit the pneumonia death
rate data. 21. Sales The Campbell Soup Company produces
both canned soup and powdered drink mix. The
b. Use the sine equation to predict the next peak in
1992 daily sales can be modeled by
the pneumonia death rate beyond the data given.
How might this information have been useful to c(x) 1.8961 sin(0.0186x 1.1801) 5.4660
health care providers in 1926? What factors million cans (16 ounce) of soup
494 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
p(x) 7.3026 cos(0.0197x 3.7526) 20.3728 c. Use the model and derivative to estimate the
million pints (reconstituted) of drink mix highest revenue and the day(s) on which this
maximum occurred.
where x is the day of the year. Assume that soup sold
at an average price of 70 cents per can and that drink d. Use the model and derivative to estimate the
mix sold at an average price of 20 cents per pint. lowest revenue and the day(s) on which this
(Source: Based on data from the Wall Street Journal, July 7, minimum occurred.
1993, p. B1.)
a. Write a function giving the revenue from the
Discussing Concepts
combined 1992 daily sales as a function of the
day of the year on which soup and powdered 22. Explain how you could determine without using de-
drink mix are produced. rivatives the maximum, minimum, and inflection
b. Write a function that models the rate of change points (both output and input) on a graph of
of the revenue from the combined 1992 sales. f(x) a sin(bx h) k.
7.5 Accumulation in Cycles
The first four sections of this chapter discussed sine and cosine functions and rates of
change of sine and cosine functions. We now consider the accumulated change in
sine and cosine functions and also investigate applications of integration using these
functions.
We begin by stating general antiderivative formulas for the sine and cosine func-
d d
tions. Because dx(cos x) 1 dx(cos x) 1(sin x) sin x, the Fundamental
Theorem of Calculus tells us that a general antiderivative of the sine function is the
d
negative of the cosine function. Also, because dx(sin x) cos x, a general antideriva-
tive of the cosine function is the sine function.
Antiderivatives of Sine and Cosine Functions
When x is a real number,
sin xdx cos x C
cos xdx sin x C
To find the general antiderivative of the sine function f(x) a sin(bx h) k,
we need only apply the antiderivative rules we used in Chapter 5. We begin by apply-
ing the Constant Multiplier Rule and the Constant Rule for antiderivatives:
[a sin(bx h) k]dx a sin(bx h)dx kdx
a sin(bx h)dx kx C 1
Now we need to find the general antiderivative of y sin(bx h). Recall that the
antiderivative process reverses the derivative process. In using the Chain Rule to find
7.5 Accumulation in Cycles 495
the derivative of y sin(bx h), we would multiply by b, the derivative of the in-
side function, to obtain y b cos(bx h). For the antiderivative, we divide by b:
[a sin(bx h) k]dx acos(bxb h) C kx C 2 1
a
cos(bx h) kx C
b
Similarly, we have
[a cos(bx h) k]dx asin(bxb h) C kx C 2 1
a
sin(bx h) kx C
b
Antiderivatives of Trigonometric Models
When x is a real number,
[a sin(bx h) k]dx ab cos(bx h) kx C
[a cos(bx h) k]dx ab sin(bx h) kx C
Accumulated Change for Sine Models
With these formulas, we can find accumulated change for sine models. For example,
tides in 1995 for the Savannah River entrance changed according to the rate-of-
change model*
t(x) 1.00934 sin(0.50467x 2.5633) feet per hour
x hours after midnight February 7, 1995. The height of the water x hours after mid-
night is the accumulated change in the rate of change from 0 to x. Thus we can find a
function for the height of the water x hours after midnight on February 7 by writing
the general antiderivative T of the rate-of-change function t:
1.00934
T(x) cos(0.50467x 2.5633) C feet above sea level
0.50467
Suppose we know that midtide occurs at 4:16 A.M. That is, T(4.26667) 0. Using this
fact, we solve for C and obtain C 0.008. Thus the height of the water x hours after
midnight on February 7 can be modeled as
T(x) 2 cos(0.50467x 2.5633) 0.008 feet above sea level
[Note: Negative values of T(x) denote tides that are lower than the average sea level.]
As illustrated in the tide example, when we have functions for rates of change, we
find the accumulation of that rate of change by using the antiderivative of the func-
tion. Example 1 again illustrates this principle.
*Based on data in Sea Island Scene of Beaufort (Beaufort, SC: Sands Publishing Co., 1995).
496 CHAPTER 7 Repetitive Change: Cyclic Functions
EXAMPLE 1 Determining Accumulated Change
Temperature Earlier in this chapter, we saw that the rate of change of the temper-
ature in Philadelphia on August 27, 1993, can be modeled as
t(x) 2.733 cos(0.285x 2.93) °F per hour
x hours after midnight. Use this rate-of-change model for temperature to find the ac-
cumulated change in the temperature between 9 A.M. and 3 P.M. on August 27, 1993.
Solution
The accumulated change in the temperature between 9 A.M. and 3 P.M. is given by the
definite integral
15 15
2.733
2.733 cos(0.285x 2.93)dx sin(0.285x 2.93)
9 0.285 9
0.285 sin[0.285(15) 2.93] 0.285 sin[0.285(9) 2.93]
2.733 2.733
9.34606 (3.42296)
12.769°F
The temperature increased by approximately 13°F between 9 A.M. and 3 P.M. ●
Average Values and Sine Models
Throughout this chapter, we have referred to the constant k of the function
f(x) a sin(bx h) k as the average (or expected) value of a cycle. This is the
value we would expect the function to take on if it were not for the fluctuations that
occur. In Chapter 5 we defined the average value of a function f from c to d as
f(x)dx
d
Average value c
dc
Example 2 illustrates that the constant k of a sine function is its average value over one
cycle, according to the definition in Chapter 5.
EXAMPLE 2 Calculating Average Value
Radiation A model for the amount of ultraviolet radiation received in Auckland,
New Zealand, is
r(m) 25.5 sin(0.52m 1.57) 32.5 watts per cm2
during the mth month of the year.
7.5 Accumulation in Cycles 497
a. Calculate the average value for one cycle of this model beginning at m 1.
b. Calculate the average value for one cycle of this model between two successive
minima. Interpret the result.
Solution
2
a. The function r has period 0.52 12.083 months. Thus the cycle beginning at
m 1 ends at m 13.083, so the average value of the cycle can be calculated
as
13.083
[25.5 sin(0.52m 1.57) 32.5]dm
1
13.083 1
25.5
1
13.083
0.52
cos(0.52m 1.57) 32.5m
12.083
400.798 (8.100)
12.083
32.5 watts per cm2
That is, discounting seasonal fluctuation, we expect the UV radiation over a pe-
riod of approximately 1 year in Auckland to be 32.5 watts per cm2. Note that
this average value is the constant k in the function r.
b. The cycle between the first two consecutive minima starts at m 0.002 and
ends at m 12.081. The average value of the cycle is calculated as
12.081
[25.5 sin(0.52m 1.57) 32.5]dm
0.002
12.081 (
0.002)
25.5
0.002
12.081
cos(0.52m 1.57) 32.5m
0.52
12.083
392.6457 (
0.0531)
12.083
32.5 watts per cm2
That is, discounting seasonal fluctuation, we expect the UV radiation in Auckland
to be 32.5 watts per cm2. This is, again, the constant k of the function r. ●
No matter which cycle you use to compute the average value of a sine function,
the average value of that cycle will always be the constant k of that sine function.
498 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
For Activities 7 through 10, find the indicated general
7.5 Concept Inventory antiderivative.
• The antiderivative of the sine function 7. (7.3 sin x 12)dx 8. 7.3 sin(x 12)dx
• The antiderivative of the cosine function
• Average value of a sine or cosine function 9. sin(7.3x 12)dx
10. [7.3 sin(7.3x 12) 12]dx
7.5 Activities 11. Find the general antiderivative of
Getting Started f(x) 4.67 sin(0.024x 3.211) 14.63
For Activities 1 and 2, use the graph of f to sketch the in- 12. Find the general antiderivative of
dicated accumulation function graphs. f(t) 49.88 cos(3.54t 4.86) 7.02
f (t)
Applying Concepts
13. Minimum Wage When the wage is expressed in
constant 2000 dollars rather than in current dollars,
t the rate of change of the federal minimum wage be-
B A C D tween 1950 and 2000 can be described by
w(t) 0.1615 cos(0.145t 1.582) dollars per year
t years after 1950.
(Source: Based on data from the U.S. Employment Standards
f(t)dt f(t)dt
x x Administration.)
1. 2. a. What are the units on the heights and widths of
A B
midpoint rectangles constructed between the
For Activities 3 through 6, consider the cyclic function graph of w and the t-axis? What are the units on
f in the figure. The area of the shaded region is 2. Using the area of those midpoint rectangles?
the symmetry indicated by the graph, find each of the b. Using 50 midpoint rectangles for each calculation,
following: approximate the area of the region between the
graph of w and the t-axis from t 7 to t 50.
y = f (x)
c. Compare the result of part b and the value of
50
1
w(t)dt . Interpret this value.
7
d. Shade on a graph of w the region represented by
45
x
-c -b -a a b c
w(t)dt .
20
45
e. Find and interpret the value of w(t)dt .
-1 20
14. Sales The rate of change of sales for a store spe-
f(x)dx f(x)dx
a c cializing in swimming pools in the summer and ski
3. 4.
0
a gear in the winter can be modeled by
5. [f(x) 1]dx 6. 3 f(x)dx
a a S(t) 23.944 cos(0.987t 1.276)
a c thousand dollars per month
7.5 Accumulation in Cycles 499
where t 1 in January, t 2 in February, and so on. 17. Lizards The number of lizards legally gathered
a. Find a model for sales in month t if sales in Jan- from a western state for selected months during a 2-
uary are $54,000. year time period can be modeled by the equation
b. Use the model to estimate when peak sales occur
L(x) 15.388 sin(0.787x 1.667) 16.346
each year. Also estimate when sales will be lowest.
thousand lizards per month
S(t)dt.
6
c. Find and interpret the value of
1 where x is a number assigned to each month. The
d. Find the area of the region between the graph of numbers 1 through 8 represent each of the eight
S and the t-axis from t 1 to t 6. consecutive months in which lizards are harvested
e. Find the average sales between January and June. in the first year, and the numbers 9 through 16 rep-
resent each of the eight consecutive harvesting
15. Profit Rate The after-tax profit rate on invest- months of the second year. In the months during
ment is defined as the ratio of after-tax corporate the 2-year period not defined by the model, there
profits to corporate stock of plant and equipment. were no lizards gathered.
The rate of change of the after-tax profit rate on in- (Source: Based on information from the Nevada Division of
vestment in the United States between 1988 and Wildlife.)
2002 can be approximated by the function
a. Use a definite integral to estimate the number of
T(x) 0.565 sin(0.469x 0.722) percent per
year, where x is the number of years after 1988. lizards harvested during the 2-year harvesting
(Source: Based on data from Bureau of Economic Analysis, period.
BusinessWeek, November 4, 2002, p. 107.) b. Because this model was formed from count data,
a. Find a model for the after-tax profit rate if the you could also estimate the number of lizards
rate in 1988 was 5.5%. harvested during the 2-year time period by sum-
b. Use the model to estimate when the after-tax ming each of the output values for the 16
profit rate was highest and when it was lowest be- months. Calculate this sum, and compare it to
tween 1988 and 2002. the estimate in part a.
c. According to the model in part a, what was the av- c. Which method (the one in part a or the one in
erage after-tax profit rate between 1988 and 2002? part b) do you prefer?
d. The average after-tax profit rate between 1960 18. Sun Declination The rate of change of the decli-
and 2000 was approximately 6.5%. Compare this nation of the Sun for Greenwich, England, can be
value with the result of part c. described by the equation
16. Temperature The rate of change of the average S(d) 0.398 sin(0.0168d 0.259) degrees per day
temperature in New York from 1873 through 1923
on the dth day of the year. Use a definite integral to
can be modeled by
determine by how many degrees the declination
T(x) 42.059 cos(0.524x 2.27) °F per month changes between January 1 and March 1.
(Source: Based on data from the Old Farmer’s Almanac, 1996.)
where x 1 in January, x 2 in February, and
19. Gender Ratio The rate of change of the ratio of
so on.
males to females in the United States from 1900
a. Find a model for the average temperature in New through 2000 can be modeled by
York. The average temperature in July is 73.5°F.
r(t) 0.2274 sin(0.0426t 2.836) males per 100
b. What does the model give as the average temper-
females per year
ature in December?
T(x)dx.
8
t years after 1900. Find and interpret the values of
c. Find and interpret the value of
2 40 100 100
r(t)dt , r(t)dt , and r(t)dt
d. Find the area of the region between the graph of 0 50 0
T and the x-axis from x 2 to x 8. (Source: Based on data from the U.S. Bureau of the Census.)
500 CHAPTER 7 Repetitive Change: Cyclic Functions
20. Daylight Hours The rate of change of the number (Source: Based on information in M. H. Finger et al., “Dis-
of hours of daylight in Honolulu, Hawaii, can be covery of Hard X-ray Pulsations from the Transient Source
GRO J1744–28,” Nature, vol. 381, May 23, 1996, pp.
modeled by 291–292.)
d(m) 1.18 cos(0.514m 1.616) hours per month a. What will the units be on the heights and widths
m months after the beginning of each year. of midpoint rectangles constructed between the
graph of p and the s-axis? What will the units be
a. Find the area beneath the graph of d that lies on the area of such midpoint rectangles?
above the m-axis between m 0 and m 12.
Interpret your answer. b. Using the fact that there are 1000 milliseconds
in 1 second, convert p(s) to counts per
b. Find the area above the graph of d that lies be- millisecond.
low the m-axis between m 0 and m 12.
Interpret your answer. c. Find and interpret the area between the graph of
p and the s-axis for one period of the function,
21. Star Pulses On December 15, 1995, a team of as- beginning at s 0.
tronomers discovered X-ray pulses being emitted
from what they believe to be a neutron star. The Discussing Concepts
speed of the pulses can be modeled by
22. Why do the inflection points of a sine function of
p(s) 40.5 sin(0.01345s 1.5708) 186.5 the form f(x) a sin(bx h) k occur at the in-
counts per second puts where the function attains its average value?
after s milliseconds.
SUMMARY
Trigonometry has been an important topic in mathe- The constant h is associated with the horizontal shift of
matics for centuries. We consider two trigonometric the function by the equation
functions, sin x and cos x. When we use sine functions to
h
model data, we consider the input as a real number with Horizontal shift
an interpretation determined by the context. b
where the shift is to the left if h 0 and to the right if
The Sine Model
h 0. Understanding these constants and their inter-
Many types of data exhibit a cyclic pattern that can be pretations in the context of the model helps us to un-
modeled by a sine function of the form derstand the model itself.
f(x) a sin(bx h) k Rates of Change
The constants of the sine function correspond to cer- As with all the other functions that we have seen in this
tain physical properties of the data. The constant k is text, we are able to use calculus to analyze the rate of
the average value of a cycle. The constant a is the am- change that is occurring in a sine or cosine model. The
plitude of the fluctuation. The constant b (considered derivative of the sine function is simply the cosine func-
positive) is associated with the period of the cycle by tion, and using derivative rules developed in Chapter 3,
the equation we have
2 d
Period [a sin(bx h) k] ab cos(bx h)
b dx
7.5 Accumulation in Cycles 501
We also saw that the derivative of the cosine function is
the negative of the sine function, which enables us to
[a sin(bx h) k] dx ab cos(bx h) kx C
find second derivatives for sine models.
Accumulation of Change
[a cos(bx h) k] dx ab sin(bx h) kx C
The antiderivative formulas for sine and cosine models We use these antiderivatives to answer questions about
are the accumulated change in cyclic models.
CONCEPT CHECK
Can you To practice, try
• Identify the amplitude, reflection, period, vertical
shift, and horizontal shift of a sine function? Section 7.1 Activity 7
• Graph a sine function? Section 7.1 Activity 15
• Construct a sine model with and without the use of Section 7.2 Activities 3, 9
technology?
• Calculate and interpret derivatives of sine and Section 7.3 Activity 17
cosine functions?
• Use derivatives to find extrema and inflection points Section 7.4 Activity 14
of sine and cosine functions?
• Find and interpret definite integrals of sine and Section 7.5 Activity 19
cosine functions?
• Find the expected (average) value of a cyclic function? Section 7.5 Activity 15
502 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
CONCEPT REVIEW
1. Sales Lawn mower sales for a certain manufac- a. Estimate the average rate of change of the price
turer are cyclic, with a low in the winter months, an index between 1860 and 1890.
increase in the spring, and a decline in the fall. Or- b. Estimate how quickly the price index was chang-
ders at this company for a 1-year period are shown ing in 1860.
in the table.
3. Price Index The equation for the graph in
Month Lawn mowers Month Lawn mowers Activity 2 is
Jan
Feb
97
450
July
Aug
1400
1050
S(t) 88.6 11.2 sin 48 t 19
10
Mar
Apr
800
1150
Sept
Oct
795
400
13.8 sin 24 t 11
36
5.3 sin 9.6 t
13
45
May 1406 Nov 190
t years after 1818.
June 1500 Dec 75
a. Use the equation to estimate the average rate of
change of the price index between 1860 and
a. Use the data to determine the amplitude, vertical 1890.
shift, period, and horizontal shift for the lawn
mower orders. b. Find and interpret S(42).
b. Use your calculator or computer to find a sine c. Find the average price index between 1860 and
function for the number of lawn mowers or- 1890.
dered in terms of the month of the year. How do 4. Sales The rate of change of the daily sales of pow-
the function constants compare to those you dered drink mix produced in 1992 by the Campbell
gave in part a? Soup Company can be modeled by
c. Use the equation to predict lawn mower orders in r(t) 0.1439 sin(0.0197t 3.7526)
April of the following year. Under what condi- million (reconstituted) pints per day
tions should this estimate be used as a prediction?
d. Use the equation to determine when lawn t days after January 1, 1992.
mower orders are greatest. (Source: Based on data in the Wall Street Journal, July 7, 1993,
e. When is the number of lawn mowers ordered in- p. B1.)
creasing most rapidly? a. Determine when the rate of change of the sales
model was greatest in 1992.
2. Price Index The graph shows the Sauerbeck Index
of general wholesale prices between 1818 and 1913. b. Find a model for the sales if the sales were 14.4
Price index million pints on January 1, 1992.
c. How is the answer to part a related to the graph
105 of the function in part b?
90
61
75 5. Sales Find and interpret the value of r(t)dt ,
32
60
where r is the rate-of-change function in Activity 4.
Year
1818 1860 1890 1913
(Source: Based on data from H. L. Moore, Generating Economic
Cycles, New York: Augustus M. Kelley, 1967.)
Project 7.1 Seasonal Sales
Setting Tasks
Some companies sell products or offer services that
1. Find a sine function for the data. Do you believe a
are seasonal in nature. Summer or winter sports
sine model is the most appropriate model for the
equipment is one example of a seasonal product, and
data you found? Give the period, amplitude, and
swimming pool cleaning is an example of a seasonal
vertical shift of the sine model. Interpret these
service. Sometimes companies specialize in two
values in the context of the company’s sales.
different products or services for which the demand
peaks in different seasons. For example, a snow ski 2. Find the maximum and minimum sales for the
shop may sell patio furniture in the summer. model. When do these sales occur? Compare the
Find a company that specializes in a seasonal extreme points of the sine function with the
product or service and that is willing to release sales highest and lowest data values and corresponding
data. Obtain data from the company. The data should times. According to the function, when are sales
span at least 1 year. If the company specializes in two increasing most rapidly? When are they decreasing
different seasonal products, you may need to consider most rapidly?
sales for the two products separately.
3. If a sine function is not a good fit for the data, find
Consider your position as being that of a
a more appropriate function. Explain your reasons
consultant. You are to make suggestions that will
for choosing a different function.
improve the company’s sales. If the company sells
only one product or offers only one service, you
should suggest another product or service to offset Reporting
the minimum sales created by the current product or
service. You should also suggest when the company 1. Write a letter addressed to the company. The letter
should begin advertising each product or service. should include your suggestions and brief reasons
If the company already sells two products or for those suggestions.
offers two services, analyze whether the peak sales of
one offset the low sales of the other. If you believe the 2. Write a report for your files, containing all the
company might benefit from offering a third product mathematics you used in performing Tasks 1
or service or from increasing advertising at particular through 3, to refer to if the sales manager for the
times during the year, offer suggestions in your letter. company calls to ask you questions.
503
Project 7.2 Lake Tahoe Levels
Setting Tasks
Lake Tahoe lies on the California-Nevada border. Its
1. Examine a scatter plot of this set of data. Discuss
level is regulated by a 17-gate concrete dam at the
why a sine function might be appropriate. If a
lake’s outlet. By federal court decree, the lake level
sine function were used to describe these data,
must never be higher than 6229.1 feet above sea level.
estimate what would be its period, amplitude,
The lake level is monitored every midnight. A table
and vertical shift. Find a sine model to fit the
showing the lake level on the first day of each month,
data. What are the function’s period, amplitude,
as reported by the United States Department of the
and vertical shift? Compare these answers to your
Interior, is located on the Calculus Concepts website.
expectations, and discuss any discrepancies.
Rewrite your model so that its output is in feet
above sea level.
2. Is there a way to construct a better-fitting model
by combining a sine function with some other
type of function using function addition,
multiplication, or composition, or by
constructing a piecewise model? If so, construct
such a model and explain why it makes sense to
do so.
3. Carefully sketch a graph of the function you
chose to model these data for the lake level. Use
the sketch to estimate the lake level in January
2003 and the rate at which the level of the lake
was changing at that time.
4. Use the model to estimate the lake level in
January 2003 and to estimate numerically how
quickly the level of the lake was changing at that
time.
5. Use the derivative of the function to find the rate
at which the level of the lake was changing in
January 2003.
6. According to the data, when was the lake at its
lowest level? When was it at its highest level? On a
sketch of the function for lake level, draw lines that
are tangent to the graph of the model at the
model’s minimum and maximum points. What is
important about the tangent lines at these points?
504
Lake Tahoe Levels, Continued
7. Use the derivative of the function to estimate the
month and day between October 1, 2002, and
September 1, 2003, when the lake was at its
lowest level. Also determine when the lake was at
its highest level between these dates. According to
the model, did the lake remain below the
federally mandated level between October 1,
2002, and September 1, 2003?
8. When (between October 1, 2002, and September
1, 2003) was the level of the lake changing most
rapidly? At that time, what was the level of the
lake and how quickly was it changing?
Reporting
1. Write a report addressed to the U.S. Department
of the Interior to communicate your findings on
Lake Tahoe levels. Keep in mind that this should
be a nonmathematical report of your
conclusions. Use graphs in your report as
appropriate. Include mathematical support for
your conclusions as an appendix. Refer to your
appendix in the body of your report as
appropriate.
2. Prepare a 10-minute presentation of your
conclusions to be given to the Department of the
Interior. You should be prepared to discuss the
mathematics if questioned, but keep in mind that
your target audience is not expecting a math talk.
Use overhead transparencies and/or other visual
aids to enhance your presentation.
3. Prepare a poster to display your conclusions. The
poster should be self-explanatory, attractive, and
easily readable from 3 feet away. Show enough
mathematics on the poster to support your
conclusions.
505
8 Dynamics of Change:
Differential Equations
and Proportionality
Concepts Outline
8.1 Differential Equations and
Slope Fields
8.2 Separable Differential
Equations
8.3 Numerically Estimating by
Using Differential Equations:
Euler’s Method
8.4 Second-Order Differential
Equations
T & L/Image Point/CORBIS
Concept Application
The rate of decay that a radioisotope undergoes dictates the amount of that isotope left
after a certain amount of time. It is this relationship that makes it possible for archeo-
logists, geologists, and environmental engineers to use radioisotopes to help establish
the age of artifacts and geological samples and to determine how much of a threat cer-
tain radioactive substances are to the environment. Activities 26 and 27 in Section 8.2
deal with developing differential equations from information about the decay of
radioisotopes.
506
8.1 Differential Equations and Slope Fields 507
Chapter Introduction
Most of the functions that we have seen in this text have been motivated by real-world
phenomena. In many cases, it is the underlying rate of change that determines
whether the equation describing a phenomenon should be linear, quadratic, logistic,
and so forth.
In this chapter we look at some applications of calculus that use information
about rates of change. We consider equations that describe the forces that drive cer-
tain rates of change. Equations stating the behavior of a rate of change (that is, equa-
tions involving derivatives) are called differential equations. We look at differential
equations graphically using slope fields, numerically using Euler’s method, and ana-
lytically using integration.
Concepts you will be learning
• Finding and graphing particular solutions to constant and linear differential
equations (8.1)
• Setting up differential equations using information about direct and inverse
proportionality (8.1)
• Using slope fields to graphically analyze a differential equation (8.1, 8.2)
• Finding and graphing particular solutions to separable differential equations (8.2)
• Setting up differential equations using information about joint proportionality
(8.2)
• Using Euler’s method to numerically analyze a differential equation (8.3)
• Solving second-order differential equations (8.4)
8.1 Differential Equations and Slope Fields
In Section 4.4 we saw how to use a known equation to develop a related-rates equa-
tion that shows us the interconnection between different rates of change. Now we
consider the opposite process: using equations that describe rates of change to de-
velop equations that describe the situation. Equations that involve rates of change
(derivatives) are called differential equations. The following equations are examples
of differential equations:
dc dy d 2s
0.16p 5.9x 3.2y 2.1s
dp dx dt2
dS k d 2y d 2y dy
3xy 4 2y
dt ln (t 1.2) dx 2 dx 2 dx
Differential Equations
Consider, for instance, the differential equation
dy
63 mph
dx
508 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
which describes the speed of a vehicle after x hours of traveling west on a straight
stretch of interstate highway in North Dakota. Suppose we want an equation for the
distance between the vehicle and the eastern border of the state after x hours. Because
dy
we are given a derivative dx and wish to find a function y, we simply write the general
antiderivative:
y(x) 63x C miles
after x hours, where C represents the distance from the border when x 0. This
equation is the solution to the differential equation given above.
It is important to note that the solution of a differential equation is a function
with derivatives that satisfy the differential equation. The solution is not a numerical
value. We define the function f to be a general solution to a differential equation
dy
dx g(x, y) if the substitution of f(x) for y gives an identity (that is, a true statement).
Differential Equation and General Solution
A differential equation involves one or more derivatives. A general solution
for a differential equation is a function that has derivatives that satisfy the
differential equation.
In the velocity example, we found y(x) 63x C miles to be a solution to
dy
dx 63 mph. To verify that this is indeed a general solution, we substitute y(x) into
the differential equation and obtain
d
(63x C) 63
dx
63 63
The statement 63 63 is called an identity. This identity confirms that we have a so-
lution. The C parameter in the solution is an arbitrary constant; it is used to designate
the family of functions that, when differentiated, gives the differential equation. In
practice, unique values for C will be determined by the initial conditions stated in the
specific problem that we are solving. In those cases, we call the solution a particular
solution.
The velocity example is a specific example of the simplest differential equation—
dy
a constant differential equation that has the form dx k, where k is a constant. To
solve this differential equation, we used our understanding of the relationships among
a rate-of-change function, the accumulation function of that rate of change, and the
quantity function. That is, we found the solution to the differential equation by deter-
mining the general antiderivative of the rate-of-change function. Another simple
dy
differential equation, called a linear differential equation, has the form dx ax b
and is given in Example 1.
8.1 Differential Equations and Slope Fields 509
EXAMPLE 1 Finding Solutions for a Linear Differential Equation
Population Judging on the basis of data compiled from Statistical Abstract, 1993,
1998, and 2001, between 1900 and 2000 the population of the United States was
growing at rate of 0.0181x 1.129 million people per year, where x is the number
of years since 1900. In 1990 the population of the United States was 249.9 million
people.
a. Write a differential equation expressing the growth rate of the population with
respect to time.
b. Find a general solution for this differential equation.
c. Find a particular solution for this differential equation. Graph the differential
equation and the particular solution.
d. Use the particular solution to estimate the population in the year 2005.
Solution
a. Let P represent the population of the United States x years after 1900. The
growth rate of the population is expressed by the differential equation
dP
0.0181x 1.129 million people per year
dx
x years after 1900.
b. The general solution is a function P describing the population. Because the dif-
dP
ferential equation dx 0.0181x 1.129 is simply the derivative of the
population function P written in terms of the input variable x, a general
dP
antiderivative of dx is a general solution for this differential equation. Thus
we have as a general solution P(x) 0.00905x2 1.129x C million people
x years after 1900.
c. We know the population in 1990, so we use the fact that P(90) 249.9 to solve
for C in the general solution. Thus we obtain the particular solution
P(x) 0.00905x2 1.129x 74.985 million people
x years after 1900. Figure 8.1 on page 510 shows graphs of the differential equa-
tion and the particular solution.
d. We estimate the population of the United States to be P(105) 293.3 million
people in the year 2005. ●
In order for us to use antiderivatives as solutions, the differential equation must
satisfy two criteria:
1. The rate of change must be in terms of the input variable only.
2. We must know an antiderivative formula for the given rate-of-change
function.
Many differential equations satisfy these criteria. Some of these differential equa-
tions are based on the ideas of proportionality.
510 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
dP
dx
Population growth P(x)
(millions of Population
people per year) (millions)
3.0 300
2.5 250
2.0 200
1.5 150
1.0 100
x 50 x
0.5
Years Years
since 0 20 40 60 80 since
0 20 40 60 80 1900
1900 (b)
(a)
FIGURE 8.1
Proportionality
dy dy
We have seen the two differential equations dx k and dx kx, where k is a constant.
The first differential equation states that the rate of change of y with respect to x is
dy
constant. The second differential equation, dx kx, states that the rate of change of
y with respect to x is proportional to the input x.
The idea of proportionality is one that is often used in setting up differential equa-
tions. We say that a variable y is directly proportional to another variable x if there
is a constant k such that y kx. We call k the constant of proportionality. We use
the terms proportional and directly proportional interchangeably.
Direct Proportionality
For input x and output y, y is directly proportional to x if there exists some
constant k such that y kx. The constant k is called the constant of
proportionality.
For example, if A(t) 23.50t dollars represents the amount it costs to purchase t
tickets to a concert, then we say that the cost is directly proportional to the number
of tickets purchased. In this case, 23.50 is the constant of proportionality.
Another type of proportionality occurs when a quantity y is related to a quantity
k
x by the equation y x , where k is a constant. In this case, we say that y is inversely
proportional to x.
Inverse Proportionality
For input x and output y, y is inversely proportional to x if there exists some
k
constant of proportionality k such that y x .
8.1 Differential Equations and Slope Fields 511
The German physiologist Gustav Fechner* said that the rate of change of the
intensity of a response R with respect to the intensity of a stimulus s is inversely
proportional to the intensity of the stimulus. That is, there is some constant k
such that
dR k
ds s
This differential equation, known as Fechner’s Law, says that if you are in a quiet en-
vironment and a small bell rings, you will perceive the sound from the bell as being
rather loud, whereas if you are in a noisy environment and the same small bell rings,
you will perceive the sound as being almost inaudible. Consider the following re-
sponse differential equation:
dR 2.94
ds s
where the stimulus (input s) is measured in decibels and the response (output R) is
measured on a scale of sound intensity where 0 represents no sound and 10 represents
unbearably loud sound. A general solution of this differential equation is simply a
general antiderivative:
R(s) 2.94 ln s C
where s is measured in decibels and is always positive.
If the smallest sound that can be detected is 10 decibels, then we can use the ini-
tial condition R(10) 0 to find a particular solution to the differential equation. In
this case, the particular solution is
R(s) 2.94 ln s 6.77
where s is measured in decibels.
EXAMPLE 2 Evaluating the Constant of Proportionality
Sales Suppose that the total sales (in billions of dollars) of a computer product are
growing in inverse proportion to ln(t 1.2), where t is the number of years since the
product was introduced. Sales totaled $53.2 billion by the end of the first year.
a. Write a differential equation representing the rate of change of sales with
respect to time.
b. At the end of the first year, total sales were growing by 8.3 billion dollars per
year. Find the constant of proportionality.
c. Can we write an explicit formula for the general solution of this differential
equation?
* D. N. Burghes and M. S. Borrie, Modelling with Differential Equations,(Chichester, England: Ellis
Horwood Limited, a division of Wiley, 1981.
512 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
Solution
a. Let S(t) represent the total sales of the computer product t years after the prod-
uct was introduced. A differential equation representing the information given is
dS k
billion dollars per year
dt ln(t 1.2)
where t is the number of years after the product was introduced.
dS
b. We are told that dt 8.3 billion dollars per year when t 1. Using this fact,
we have
k
8.3
ln(1 1.2)
Thus the constant of proportionality is k 6.544. The differential equation is
dS 6.544
billion dollars per year
dt ln(t 1.2)
where t is the number of years after the product was introduced.
c. This differential equation expresses the derivative of S as a function of the input
variable t. If we knew a formula for a general antiderivative of this function,
then we could write a general solution of the differential equation. However, we
do not know such a formula. ●
As in Example 2, even when the differential equation appears to be in a simple
form, we may not be able to find an explicit formula for its general solution. How-
ever, we still can analyze such a differential equation graphically and numerically to
develop useful insight into the nature of the function and to estimate the solution.
Slope Fields
One way to obtain a graphical representation of a solution to a differential equation
is to draw a slope field. A slope field is constructed by placing a grid on a portion of
the Cartesian plane and, at each point on the grid, drawing a short line
segment whose slope is determined by the differential equation. For instance, con-
dy
sider the differential equation dx 2x . At the point (1, 1), the slope of a solution to
this equation is 2x 2(1) 2, and at (0.5, 2), the slope is 2x 2(0.5) 1.
Using the differential equation to determine the slopes at points on a grid on a plane
where 3 x 3 and 6 y 6 and then sketching short line segments with
those slopes at the appropriate points gives the slope field shown in Figure 8.2. (This
construction is a tedious process and is usually done with computer software.)
Particular solutions for the differential equation can be sketched by following the
line segments in such a way that the solution curves are tangent to each of the segments
dy
they meet. Figure 8.3 shows the graph of a particular solution for dx 2x. This par-
ticular solution goes through the point, or initial condition, (0, 1). We use the
term initial condition to refer to a known point on the graph of a particular solu-
tion. Knowing an initial condition allows us to find a particular, rather than a
general, solution.
8.1 Differential Equations and Slope Fields 513
y y
6 6
4 4
2 2
x x
−3 −2 −1 1 2 3 −3 −2 −1 1 2 3
−2 −2
−4 −4
−6 −6
FIGURE 8.2 FIGURE 8.3
The slope field indicates that at the point (0, 1), the derivative of the solution
is zero. Thus the solution has a maximum, a minimum, or an inflection point at
(0, 1). The slopes to the left of x 0 indicate that the solution is decreasing toward
(0, 1), and the slopes to the right of x 0 indicate that the solution is increasing af-
ter it reaches (0, 1). Therefore, (0, 1) is a minimum point. We form the solution
y graph by following the general direction indicated by the slopes. You may find it help-
6
ful to consider this as a more sophisticated version of “connecting the dots.” Re-
4 member, however, that slopes in a slope field graph are plotted at only some, not all,
2 points on the plane.
dy
Figure 8.4 shows the graphs of several particular solutions for dx 2x drawn on
x
−3 −2 −1 1 2 3 the slope field. You should recognize the curves in Figure 8.4 as graphs of solutions of
−2 dy
the form y(x) x 2 C , which we know is the general solution of dx 2x. The
−4 particular solution in Figure 8.4 with initial condition (0, 1) is shown in teal. In this
−6
case, the particular solution is y(x) x 2 1. The other particular solutions shown
correspond to C 3, 2, 0, 1, 2, 3.
The particular solutions Although this example of a slope field is for a differential equation with a known
y x 2 C for solution, slope fields are particularly helpful when we are graphing solutions for a dif-
C 3,2,1, 0, 1, 2, 3 ferential equation for which we do not know solution formulas. A little practice look-
FIGURE 8.4 ing at slope fields will assist you in sketching graphs of solutions and will help you
understand how the forces that drive the rate of change of a quantity determine the
shape of the graph of the quantity function.
Each of Figures 8.5 through 8.8 shows slope fields with three particular solutions
sketched. On each figure, the particular solutions for the initial conditions (2, 2),
(1, 1), and (1, 1) are shown for the quantity function y with input x.
y y
6 3
4 2
2 1
x x
−3 −2 −1 1 2 3 −3 −2 −1 1 2 3
−2 −1
−4 −2
−6 −3
FIGURE 8.5 FIGURE 8.6
514 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
y
y
3
3
2
2
1
1
x
x
−3 −2 −1 1 2 3
−1 −3 −2 −1 1 2 3
−1
−2
−2
−3
FIGURE 8.7 FIGURE 8.8
Now that we have seen a few examples of solution graphs drawn on slope fields,
we use slope fields to help us determine a solution graph in an applied situation.
EXAMPLE 3 Sketching Graphs of Particular Solutions on a Slope Field
S(t)
Computer sales Sales In Example 2 we derived a differential equation representing the
( billions of dollars) rate of change of computer sales with respect to time:
80 dS 6.544
billion dollars per year
dt ln(t 1.2)
60
where t is the number of years after the product was introduced.
40 a. Examine the slope field of this differential equation shown in
Figure 8.9, and sketch three particular solutions on the slope field.
20 b. In Example 2 we were told that sales were $53.2 billion at the end
of the first year. Sketch the particular solution corresponding to
t this initial condition, and use it to estimate sales after 3 years.
0 1 2 3 4 5 6 Years
FIGURE 8.9 Solution
a. Figure 8.10 shows three particular solutions graphed on the slope
field. You may have sketched different solutions, but they should
all have the same general shape.
b. Figure 8.11 shows the particular solution when S(t) 53.2 at t 1.
The point on the solution graph at t 3 appears to be (3, 60). Thus
we estimate sales after 3 years as approximately $60 billion. ●
So far we have seen only a few simple differential equations in which the deriva-
tive of some function f is given in terms of only the input variable x. In these cases, a
general solution is the general antiderivative of the differential equation when one is
known. Graphing the differential equation using slope fields can be helpful in
obtaining insight into the nature of the function.
8.1 Differential Equations and Slope Fields 515
S(t) S(t)
Computer sales Computer sales
(billions of dollars) (billions of dollars)
80 80
60 60
40 40
20 20
t t
0 1 2 3 4 5 6 Years 0 1 2 3 4 5 6 Years
FIGURE 8.10 FIGURE 8.11
For each of the differential equations in Activities 5
8.1 Concept Inventory through 11:
a. Use the corresponding slope field to sketch the
• Solutions to differential equations of the form graphs of three particular solutions.
dy
dx k b. Describe how the graphs of the solutions compare
with each other.
• Solutions to differential equations of the form
dy c. Write a general solution for the differential
dx f (x) equation.
• Direct proportionality dy
5. 0 y
• Inverse proportionality dx
5
• Slope fields
4
8.1 Activities 2
Getting Started 1
For Activities 1 through 4, write an equation or differ-
x
ential equation for the given information. 0 1 2 3 4 5
1. The cost c to fill your gas tank is directly proportional dy
to the number of gallons g your tank will hold. 6. 1 y
dx
5
2. The marginal cost of producing window panes (that
is, the rate of change of cost c with respect to the 4
number of units produced) is inversely propor-
tional to the number of panes p produced. 3
3. Barometric pressure p is changing with respect to al-
2
titude a at a rate that is proportional to the altitude.
4. The rate of change of the cost c of mailing a first- 1
class letter with respect to the weight of the letter is
constant. x
0 1 2 3 4 5
516 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
dy dy
7. 1 y 11. x y
dx dx 8
5
6
4 4
2
3 x
−4 −3 −2 −1 1 2 3 4
−2
2 −4
−6
1
−8
x
0 1 2 3 4 5 Discussing Concepts
12. For the differential equations in Activities 5 through
dy 1
8. y 8, compare and contrast the particular solutions
dx 2 5 with initial condition (0, 0).
4
13. For the differential equations in Activities 9 through
11, compare and contrast the particular solutions
3 with initial condition (0, 0).
2 Applying Concepts
1
14. Consider a function y f(x) whose rate of change
with respect to x is constant.
x a. Write a differential equation describing the rate
0 1 2 3 4 5
of change of this function.
b. Write a general solution for the differential
dy 1
9. x y equation.
dx 2 8 c. Verify that the general solution you gave in part
6 b is indeed a solution by substituting it into the
4 differential equation and obtaining an identity.
2
x
15. Energy Between 1975 and 1980, energy produc-
− 4 −3 −2 −1 1 2 3 4 tion in the United States was increasing at an
−2
approximately constant rate of 0.98 quadrillion Btu
−4
per year. In 1980 the United States produced 64.8
−6
quadrillion Btu.
−8 (Source: Based on data from Statistical Abstract, 1994.)
a. Write a differential equation for the rate of change
dy of energy production.
10. 2x y
dx 8 b. Write a general solution for the differential
6 equation.
4 c. Using the initial condition, determine the partic-
2 ular solution for energy production.
x
− 4 −3 −2 −1 1 2 3 4
d. Estimate the energy production in 1975, as well
−2 as the rate at which energy production was
−4 changing at that time.
−6
e. Use the slope field to sketch the graph of the par-
−8
ticular solution indicated by the initial condition,
8.1 Differential Equations and Slope Fields 517
and use this graph to estimate energy production 17. Cropland The amount of arable and permanent
in 1975. How close is your graphical estimate to cropland worldwide increased at a slow but rela-
your estimate in part d? tively steady rate of 0.0342 million square kilome-
ters per year between 1970 and 1990). In 1980 there
Energy production
(quadrillion Btu) were 14.17 million square kilometers of cropland.
(Source: Ronald Bailey, ed., The True State of the Planet,
65
New York: The Free Press for the Competitive Enterprise
64 Institute, 1995.)
63 a. Write a differential equation representing the
62 growth of cropland.
61 b. Write a general solution for the differential
60 equation in part a.
59 c. Write the particular solution for the amount of
Years cropland.
58
since
0 1 2 3 4 5 1975 d. Use the equations to estimate the rate of change
of cropland in 1970 and in 1990. Also estimate
16. Energy Between 1975 and 1980, energy consump- the amount of cropland in those years.
tion in the United States was increasing at an 18. Air Pressure Barometric pressure p (measured in
approximately constant rate of 1.08 quadrillion Btu inches of mercury) decreases with respect to alti-
per year. In 1980 the United States consumed 76.0 tude a (measured in feet) at a rate that is directly
quadrillion Btu. proportional to the altitude. The constant of pro-
(Source: Based on data from Statistical Abstract, 1994.)
portionality is 3.7 105.
a. Write a differential equation for the rate of change a. Write a differential equation representing the
of energy consumption. rate of change of barometric pressure.
b. Write a general solution for the differential b. Write a general solution for this differential
equation. equation.
c. Determine the particular solution for energy c. Assume that barometric pressure at sea level is
consumption. 30 inches of mercury. Find a particular solution
d. Estimate the energy consumption in 1975, as for the differential equation.
well as the rate at which energy consumption was d. Use the slope field to sketch the particular solution
changing at that time. for the differential equation and initial condition
e. Use the slope field to sketch the graph of the par- given in part c.
ticular solution indicated by the initial condition,
p(a)
and use this graph to estimate energy consump- Barometric pressure
tion in 1975. How close is your graphical estimate (inches of mercury)
to your estimate in part d?
Energy consumption 30.16
(quadrillion Btu)
78 30.12
76 30.08
74 30.04
72 a
30.00
Altitude
70 0 20 40 60 80 100 (feet)
Years
68
since 19. Falling Object An object that has been dropped
0 1 2 3 4 5
1975 falls at a velocity v (in feet per second) that is
518 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
proportional to the number of seconds t after it has differential equation and simplifying to obtain
been dropped. The constant of proportionality de- an identity.
pends on the force of gravity. On Earth, the propor-
tionality constant is 32 feet per second squared. 22. Weight For the first 9 months of life, the average
weight w, in pounds, of a certain breed of dog
a. Write an equation giving velocity as a function of
increases at a rate that is inversely proportional
time.
to time t, in months. A 1-month-old puppy weighs
b. Rewrite the equation in part a as a differential 6 pounds, and a 9-month-old puppy weighs
equation giving the rate of change of distance as 80 pounds.
a function of time.
a. Write a differential equation describing the rate
c. Find a general solution to the differential of change of the weight of the puppy.
equation.
b. Give the particular solution for this differential
d. After how many seconds will the object hit the
equation on the basis of the information given.
ground if it was dropped from a height of 35 feet?
What is the terminal velocity of this object? c. Estimate the weight of the puppy at 3 months
and at 6 months.
20. Falling Object Refer to Activity 19. Use the slope
ds d. Why does this differential equation describe
field graph for dt 32t that is given to answer the
weight gain for only 8 months instead of for the
following questions.
life span of the dog?
s(t)
50 23. Height The height h, in feet, of a certain tree
increases at a rate that is inversely proportional
40
to time t, in years. The height of the tree is 4 feet
30 at the end of 2 years and reaches 30 feet at the end of
7 years.
20
a. Write a differential equation describing the rate
10 of change of the height of the tree.
0 t (seconds) b. Give a particular solution for this differential
0.5 1.0 1.5 2.0
equation.
a. Sketch a graph of the particular solution for the
c. How tall will the tree be in 15 years? What will
initial condition s 35 when t 0.
happen to the height of this tree over time?
b. Sketch a graph of the particular solution for the
initial condition s 10 when t 0. 24. Weight Refer to Activity 22. A slope field for the
c. If v represents the velocity of a falling (or thrown) differential equation in part a of Activity 22 is shown.
object in feet per second t seconds after the object
is thrown or dropped, discuss what information w(t)
is given by the graphs of the particular solutions Weight
in parts a and b. (pounds)
21. Consider a function y f(x) whose rate of change 80
with respect to x is directly proportional to the
input. 60
a. Write a differential equation describing the rate
of change of this function. 40
b. Write a general solution for the differential
20
equation.
c. Verify that the general solution you gave in part t
b is indeed a solution by substituting it into the 0 2 4 6 8 10 (months)
8.1 Differential Equations and Slope Fields 519
a. Sketch the graph of the particular solution in dy 1
a. y
part b of Activity 22. dx x 3
b. Use the graph to estimate the weight of the
2
puppy at 3 months and at 6 months.
1
25. Height Refer to Activity 23. A slope field for the
x
differential equation in part a of Activity 23 is −3 −2 −1 1 2 3
shown. −1
−2
h(t)
Height −3
(feet)
70
dy 10
60 b. y
dx x 30
50
40 20
30 10
20 x
−30 −20 −10 10 20 30
10 −10
t
0 2 4 6 8 10 12 14 16 18 20 (years) −20
−30
a. Sketch the graph of the particular solution in
part b of Activity 23.
b. Use the graph to estimate the height of the tree dy 1
c. y
after 15 years of growth. dx x 3
26. Consider a function y f(x) whose rate of change 2
with respect to x is inversely proportional to the 1
input.
x
a. Write a differential equation describing the rate −3 −2 −1 1 2 3
of change of this function. −1
b. Write a general solution for the differential −2
equation.
−3
c. Verify that the general solution you gave in part
b is indeed a solution by substituting it into the
differential equation and simplifying to obtain dy 1
d. y
an identity. dx 10x 0.3
27. For each of the following differential equations and 0.2
their slope fields:
0.1
i. Sketch the graphs of three particular solu-
tions. x
−0.3 −0.2 −0.1 0.1 0.2 0.3
ii. Describe how the graphs of the solutions −0.1
behave.
−0.2
iii. Compare and contrast the family of solu-
−0.3
tions for each of the differential equations.
520 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
In Activities 28 through 31, use the slope fields given for dy
30. cos x y
the differential equations to sketch the particular solu- dx
tion for each given initial condition. 6
a. x 0, y 0 5
dy b. x 2, y 5 4
28. 2x 1 y
dx 10 c. x 1, y 2 3
2
a. x 4, y 11 8
6 1
b. x 2, y 4 4 x
2 −3 −2 −1 − 1 1 2 3
x
−4 −3 −2 −1
−2 1 2 3 4 −2
−4
−6
−8
−10 dy
−12 31. 3x2 2x y
dx 40
dy a. x 2, y 5 30
29. sin x y
dx b. x 3, y 0 20
6
a. x 3, y 4 5
10
b. x 0, y 0 4 x
− 4 −3 −2 −1 1 2 3 4
c. x 2, y 1 3 −10
2
−20
1
x −30
−3 −2 −1 − 1 1 2 3
−2
8.2 Separable Differential Equations
y dy
In Section 8.1 we considered differential equations of the form dx f(x). In this sec-
15 dy dy
tion we consider differential equations of the forms dx f(y) and dx f(x, y). Be-
10
cause differential equations of these types give slope in terms of x and y, slope fields are
5 useful in giving graphical insight into the behavior of the underlying function. For in-
x dy 32
−4 −2 2 4 stance, a slope field for dx y (see Figure 8.12) shows that the general solution for
−5
−10
this differential equation appears to behave in a parabolic manner, with the parabola
lying on its side.
−15
FIGURE 8.12
Separation of Variables
dy dy
Because differential equations of the forms dx f(y) and dx f(x, y) are not in
terms of only the input variable x, they cannot be solved directly by writing an anti-
derivative function. Instead, we use a technique known as separation of variables to
solve such equations.
dy 32
Let us again consider the differential equation dx y . Because we do not know
32
what it means to find a general antiderivative of y with respect to the variable x, we
move all symbols containing y to one side of the equation and all symbols involving
8.2 Separable Differential Equations 521
Until now, we have considered x to the other side of the equation. This procedure is known as separating the variables.
dy
dx as a single symbol denoting In this case, we have
the rate of change of y with
respect to x. In separating ydy 32dx
variables, we consider dy and dx
to be two separate symbols, Now that the equation has the variables separated, we take antiderivatives of both
sometimes referred to as sides of the equation.
differentials.
ydy 32dx
1 2
y c1 32x c2
2
where c1 and c2 are both constants. Combining the constants and solving for y 2 yields
y 2 64x C
Finally, we write the solution equation giving y in terms of x by taking the square root
of both sides of the equation.
y 64x C
This equation does indeed yield the two sides of a horizontal parabola, as the slope
field indicates. In this case, y is not a function of x because a single value of x could
yield two different values of y.
Differential Equations Modeling Constant Percentage Change
Any time that percentage growth is constant, the situation can be described by the dif-
dy
ferential equation dx ky, where k is the constant percentage rate of change. Because
constant percentage growth characterizes exponential functions, the solution to this
differential equation is y ae kx. Example 1 illustrates the solution to differential
equations of this form.
EXAMPLE 1 Solving Differential Equations Using Separation of Variables
Bank Account Consider an account for which interest is compounded
continuously at an annual interest rate of 7%.
A(t) a. Write a differential equation expressing the rate of change of the
Amount amount in the account with respect to time.
(dollars)
b. Examine the slope field of this differential equation, shown in Figure
5000
8.13, and make a conjecture about the behavior of the function.
4000 c. Find a general solution for this differential equation.
d. If the amount after 3 years is $1000, find the particular solution.
3000
Solution
2000
a. The rate of change of A, the amount in dollars in the account, can be
1000 expressed as the differential equation
dA
0 5 10 15 20 25
t
30 Years
0.07A dollars per year
dt
FIGURE 8.13 after t years.
522 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
b. This slope field indicates a horizontal asymptote at A 0 and growth that
increases as t increases. This behavior is indicative of an exponential model.
c. To solve this equation, we use the technique of separating variables:
1
dA 0.07dt
A
Determining general antiderivatives of both sides, we have
A1 dA 0.07dt
which gives
We do not need to use A in ln A c1 0.07t c2
the antiderivative because we
know that the amount in the
account must be positive.
Combining the constants yields
ln A 0.07t C
Recalling from algebra that ln x y is equivalent to x e y, we have
A e (0.07tC)
e 0.07te C
Replacing the constant eC with the constant a gives the general solution
A ae 0.07t dollars
after t years.
d. Because the amount after 3 years is $1000, we find the particular solution by
substituting t 3 and A 1000 into the general solution and solving for a.
A 1000 ae 0.07(3)
Amount
(dollars) a 810.584
5000 The particular solution is
4000 A 810.584e0.07t dollars
3000
after t years. This solution can also be written in the form
A 810.584(1.0725t) dollars
2000
after t years. As Figure 8.14 shows, this particular solution fits the
1000
(3, 1000)
description in part b. ●
t
0 5 10 15 20 25 30 Years Example 2 illustrates the use of two different differential equations, one
FIGURE 8.14 that requires separation of variables and one that can be solved by deter-
mining an antiderivative.
EXAMPLE 2 Solving Differential Equations Using Various Techniques
GNP The gross national product (GNP) for a certain country has been increasing
continuously at a rate of approximately 5% per year. The government has decided to
8.2 Separable Differential Equations 523
keep the growth in the rate of deficit spending proportional to the GNP. Last year the
GNP was $3 billion, and the country’s national debt was $2.3 billion. The government
has mandated that this year’s national debt be held at $2.4 billion dollars.
a. Express the country’s GNP growth as a differential equation, and find the
solution.
b. Express the rate of change of the country’s national debt as a differential
equation, and find the solution.
c. Evaluate the solutions for t 2, and interpret the answers.
Solution
dG
a. Let G represent the GNP in billions of dollars. Then dt 0.05G billion dollars
per year represents the rate of change of the GNP after t years. The solution is
found by separating variables.
1
dG 0.05 dt
G
yields
ln G 0.05t C
which can be rewritten as
G ae 0.05t billion dollars
If we consider t to be the number of years since last year, then a 3 because the
GNP in year 0 (last year) was $3 billion. Thus the solution to the differential
equation is
G 3e 0.05t billion dollars
after t years.
b. Let D be the national debt in billions of dollars. The statement “the growth rate
of deficit spending is proportional to the GNP” can be translated into mathe-
matical symbols as
dD
kG billion dollars per year
dt
after t years, where k is the constant of proportionality. Note that at this point
we do not have the information we need to solve for k. However, we have
enough information to solve for k once we find a general solution. Knowing a
function for G, we substitute this into the differential equation:
dD
k(3e 0.05t)
dt
Because this differential equation gives the derivative of D in terms of only the
input t, we do not need to use separation of variables but can proceed by writ-
ing a general antiderivative:
3ke0.05t
D(t) C 60ke0.05t C billion dollars
0.05
524 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
after t years. We now have two constants, k and C, to determine, so we must
create a system of two equations that we can solve simultaneously. Because the
national debt last year (when t 0) was $2.3 billion, we substitute this infor-
mation into D:
2.3 60ke 0 C 60k C (1)
Also, the national debt is $2.4 billion when t 1, so
2.4 60ke 0.05 C (2)
Solving equation 1 for C and substituting into equation 2 give the equation
2.4 60ke 0.05 (2.3 60k)
or
0.1 (60e 0.05 60)k
0.1
k
60e 0.05 60
Thus k 0.0325. Using this value of k in equation 1 gives
C 2.3 60(0.0325) 0.3496
Thus we have the particular solution
D(t) 1.9504e0.05t 0.3496 billion dollars
after t years.
c. When t 2, G(2) $3.3 billion and D(2) $2.5 billion. In 2 years, the GNP
will be approximately $3.3 billion, and the national debt should be held to ap-
proximately $2.5 billion. ●
Joint Proportionality
Section 8.1 introduced two forms of proportionality: direct and inverse. Now we
consider a third form of proportionality. When a quantity y is proportional to the
product of two other quantities x and z—that is, when there is some constant k
such that y kxz—we say the quantity y is jointly proportional to the quantities
x and z.
Joint Proportionality
For inputs x and z and output y, y is jointly proportional to x and z if there
exists some constant of proportionality k such that y kxz.
Example 3 illustrates the concept of joint proportionality as it is used in
psychology.
8.2 Separable Differential Equations 525
EXAMPLE 3 Solving an Equation Involving Joint Proportionality
Stimulus Response The Fechner Law relates response to stimulus. A different
model that is used to describe this relationship is the Brentano-Stevens Law. This law
says that the level of response R changes according to a joint proportionality between
dR R
the level of the response and the inverse of s, the level of the stimulus. That is, ds k s
for some constant k.
Consider the following differential equation, which describes a person’s perception
of the intensity of sound:
dR 2.94R
ds s
where the sound s is measured in decibels and intensity R is measured on a scale from
0 to 10, with 0 representing inaudible sound and 10 representing painfully intense
sound.
a. Examine the slope field graph in Figure 8.15, and comment on the behavior of
the graphs of the solutions to this differential equation.
b. Write an equation giving the response R in terms of stimulus s.
R
Intensity
10 Solution
dR 2.94R
8 a. The slope field of ds s that is shown in Figure 8.15 suggests
that the solutions are exponential.
6
dR 2.94R
b. Because the differential equation ds s gives the rate of
4
change of R with respect to s in terms of both s and R, we use the
2 method of separation of variables rather than simply writing an
s antiderivative formula. Separating the variables yields
0
0.1 0.2 0.3 0.4 0.5 (decibels) 1 2.94
dR ds
FIGURE 8.15 R s
Taking antiderivatives of both sides of the equation and realizing that R 0 and
s 0 yield the equation ln R c1 2.94 ln s c2. Combining the constants c1
and c2 gives the equation ln R 2.94 ln s C. This equation is equivalent to
e ln R e 2.94 ln sC
which we simplify as follows:
eln R e2.94 ln seC
eln R (eln s)2.94eC
R s2.94eC
Replacing Ec with the constant a gives the general solution as
R as2.94
where s is measured in decibels and a is a constant. ●
526 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
Logistic Models and Their Differential Equations
We have seen differential equations that lead to linear, quadratic, logarithmic, and ex-
ponential models. What does a differential equation that yields a logistic model look
like? To help us better understand the underlying differential equation, recall the role
of the limiting value L in a logistic model:
L
y(x)
1 Ae Bx
As we saw in Chapter 1, the logistic model may be used to study the spread of a virus
in a network of computers. The virus initially spreads exponentially because the first
computer infects another, then these two computers infect two others, then the four
of them infect four others, and so on. At some point, the computers that have the
virus contact only others that are also infected instead of uninfected computers, and
the exponential sequence is broken. Eventually, the number of infected computers is
close to the limiting value L.
The rate of change of the number of infected computers depends on two things:
the number y of computers already infected and the number L y of computers not
yet infected. Thus the rate of change is jointly proportional to the number of com-
puters already infected and to the number of uninfected computers remaining. We
can describe the spread of a virus with the differential equation
dy
ky(L y)
dx
where L is the number of computers in the group and k is the constant of propor-
tionality. It is also common to refer to the limiting value L as the carrying capacity of
the system or as the saturation level.
dy
A slope graph of a differential equation of the form dx ky(L y) gives us in-
insight into the behavior of the function y. Figure 8.16 shows slope fields for differ-
ential equations of this form with different values of k and L.
In each case, the slope field of the differential equation appears to have an upper
and a lower horizontal asymptote, and a solution between these limiting values
y y
10 x
−2 2 4 6
8 −2
6 −4
4 −6
2 −8
x −10
−2 0 2 4 6
(a) k 0, L 0 (b) k 0, L 0
dy
FIGURE 8.16 Slope fields for ky (L y)
dx
8.2 Separable Differential Equations 527
y y
10 x
1 2 3 4 5
8 −2
6 −4
4 −6
2 −8
x −10
0 1 2 3 4 5
(c) k 0, L 0 (d) k 0, L 0
FIGURE 8.16 (continued)
appears to be logistic (either increasing or decreasing). In fact, it can be shown that
dy
the solution for dx ky(L y) is the logistic function
L
y
1 Ae Lkx
L
Even though the logistic equation y(x) 1 Ae Bx (where A and L are constant and
dy
B Lk) can be derived from the differential equation dx ky(L y) using separation
of variables, partial fractions, and a lot of algebra, it is not within the scope of this text
to go into this much detail. However, you will be asked in the activities to
L dy
verify that y(x) 1 Ae Bx , with B Lk, is a general solution for dx ky(L y).
EXAMPLE 4 Solving an Equation That Results in a Logistic Function
Epidemic In 1949 the United States experienced the second worst polio epidemic*
in its history. (The worst was in 1952.) In January, 494 cases of polio were diagnosed,
and by December, a total of 42,375 cases had been diagnosed. Assume that the spread
of polio followed the general principle that the rate of spread is jointly proportional to
the number of infected people and to the number of uninfected people. Also assume
that the carrying capacity for polio in the United States in 1949 was approximately
43,000 people.
a. Write a differential equation describing the spread of polio.
b. Determine a particular solution for this differential equation.
* The National Foundation for Infantile Paralysis, Twelfth Annual Report, 1949.
528 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
Solution
a. Let P(m) be the number of polio cases diagnosed by the end of the mth month
of 1949, and let k be the constant of proportionality. A differential equation
describing the spread of polio is
dP
kP(43,000 P) cases per month
dm
Because we are given information about the number of cases, not information
concerning the rate of change of the number of cases, we cannot find the con-
stant of proportionality at this point.
b. A general solution for this differential equation is the logistic equation
43,000
P(m) cases
1 Ae 43,000 km
diagnosed by the end of the mth month of 1949. This equation contains
two constants, A and k. We are given the two points (1, 494) and (12, 42,375).
Substituting these into the logistic equation, we obtain a system of two equa-
tions that we can solve simultaneously for the constant, A and k. The equations
are
43,000
494 (3)
1 Ae 43,000k
43,000
42,375 (4)
1 Ae 516,000k
Solving equation 3 for A and then substituting into equation 4 yields
43,000
42,375
1 42,506e 43,000k e 516,000k
494
Solving this equation for k yields k 1.83328271 105, or B Lk 0.788312.
Substituting k into equation 3 and then solving for A yields A 189.2704. Thus
we have the particular solution
43,000
P(m) cases
1 189.2704e 0.788312m
diagnosed by the mth month of 1949. ●
Separation of variables often yields a solution when we are considering differential
dy dy
equations of the form dx f (y) or dx f (x, y). Slope fields usually give graphical
insight into the behavior of the underlying function, and in the case where
dy
dx ky (L y), we have a specific equation that gives the solution. However, there are
many differential equations for which graphical and algebraic methods of determining
particular solutions are beyond the scope of this book. In fact, there are many differen-
tial equations for which algebraic solution methods fail. In such cases, we rely on
numerical techniques, one of which is discussed in the next section.
8.2 Separable Differential Equations 529
rate proportional to the depth S of the snowpack.
8.2 Concept Inventory Write a differential equation describing the rate
of change in the depth of the snowpack with respect
• Solutions to differential equations of the form to time.
dy
dx f(x, y)
7. Water flows into a reservoir at a rate that is in-
versely proportional to the square root of the
• Solutions to differential equations of the form depth of water in the reservoir, and water flows out
dy
dx ky
of the reservoir at a rate that is proportional to the
depth of the water in the reservoir. Write a differ-
• Solutions to differential equations of the form ential equation describing the rate of change in the
dy
dx ky(L y)
depth D of water in the reservoir with respect to
time t.
• Separation of variables
• Joint proportionality For Activities 8 through 17, sketch three particular solu-
tions for each given slope field.
8. y
6
8.2 Activities 2
x
Getting Started −3 −2 −1 1 2 3
For Activities 1 through 5, write a differential equation −2
for each of the statements. When possible, find a general −4
solution to the differential equation.
−6
1. Ice thickens with respect to time t at a rate that is in-
versely proportional to its thickness T. 9. y
6
2. The Verhulst population model assumes that a pop-
ulation P in a country will be increasing with respect 4
to time t at a rate that is jointly proportional to the 2
existing population and to the remaining amount of x
the carrying capacity C of that country. −3 −2 −1 1 2 3
−2
3. The rate of change with respect to time t of the
amount A that an investment is worth is propor- −4
tional to the amount in the investment. −6
4. The rate of change in the height h of a tree with re-
10. y
spect to its age a is inversely proportional to the
2
tree’s height.
5. In a community of N farmers, the number x of 1
farmers who own a certain tractor changes with re-
spect to time t at a rate that is jointly proportional to x
the number of farmers who own the tractor and to −2 −1 1 2
the number of farmers who do not own the tractor. −1
6. In mountainous country, snow accumulates at a
−2
rate proportional to time t and is packed down at a
530 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
11. g 15. h
6
8
6 4
4 2
2
t
x −3 −2 −1 1 2 3
−3 −2 −1 1 2 3 −2
−2
−4 −4
−6 −6
12. g 16.
6 h
3.0
4
2.5
2
t 2.0
−3 −2 −1 1 2 3
−2 1.5
−4 1.0
−6
0.5
t
13. g 0 2 4 6 8 10
3
2
17. g
1 3
t 2
−3 −2 −1 1 2 3
−1 1
−2 x
−3 −2 −1 1 2 3
−3 −1
−2
14. h −3
5
For Activities 18 through 24, identify the differential
4 equation as one that can be solved using only antideriv-
atives or as one for which separation of variables is
3 required. Then find a general solution for the differen-
tial equation.
2
dy dy
18. kx 19. ky
1 dx dx
dy k dy k
x 20. 21.
0 5 10 15 20 25 30 dx y dx x
8.2 Separable Differential Equations 531
dy kx dy ky 28. Consider a function y f(x) whose rate of change
22. 23.
dx y dx x is proportional to f.
dy a. Write a differential equation describing the rate
24. kxy of change of this function.
dx
b. Write a general solution for the differential
Applying Concepts equation.
25. Medicine The rate of change with respect to time c. Verify that the general solution you gave in part
of the quantity q of pain reliever in a person’s body b is indeed a solution by substituting it into the
t hours after the individual takes the medication is differential equation and simplifying to obtain
proportional to the quantity of medication remain- an identity.
ing. Assume that 2 hours after a person takes 200 29. Postage In 1880, 37 countries issued postage
milligrams of a pain reliever, one-half of the original stamps. The rate of change (with respect to time)
dose remains. of the number of countries issuing postage stamps
a. Write a differential equation for the rate of change between 1836 and 1880 was jointly proportional to
of the quantity of pain reliever in the body. the number of countries that had already issued
b. Find a particular solution for this differential postage stamps and to the number of countries that
equation. had not yet issued postage stamps. The constant of
c. How much pain reliever will remain after 4 hours? proportionality was approximately 0.0049. By 1855,
after 8 hours? 16 countries had issued postage stamps.
(Source: “The Curve of Cultural Diffusion,” American
26. Radioisotope Technetium-99 is a radioisotope Sociological Review, August 1936, 547–556.)
that has been used in humans to help doctors locate a. Write a differential equation describing the rate
possible malignant tumors. Radioisotopes decay of change in the number of countries issuing
(over time) at a rate that is directly proportional to postage stamps with respect to the number of
the amount of the radioisotope. Technetium-99 years since 1800.
has a half-life of 210,000 years. Assume that
0.1 milligram of technetium-99 is injected into a b. Write a general solution for the differential
person’s bloodstream. equation.
a. Write a differential equation for the rate at which c. Write the particular solution for the differential
the amount of technetium-99 decays. equation.
b. Find a particular solution for this differential d. Estimate the number of countries that were issu-
equation. ing postage stamps in 1840 and in 1860.
27. Radioisotope Radon-232 is a colorless, odorless e. Using the slope field, identify the upper and
gas that undergoes radioactive decay with a half-life lower horizontal asymptotes.
of 3.824 days. It is considered a health hazard, so N(t)
new homebuyers often have their property tested Countries
for the presence of radon-232. Because radon-232
is a radioisotope, it decays (over time) at a rate that
is directly proportional to the amount of the 30
radioisotope.
a. Write a differential equation for the rate at which 20
an amount of radon-232 decays.
b. Write a general solution for this differential 10
equation. t
c. If 1 gram of radon-232 is isolated, how much of Years
0 after
it will remain after 12 hours? after 4 days? after 40 50 60 70 80
9 days? after 30 days? 1800
532 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
f. On the slope field, sketch the particular solution N(t)
for the given initial condition. Patents
2700
30. Patents The number of patents issued for plow 2500
sulkies between 1865 and 1925 was increasing with
2000
respect to time at a rate jointly proportional to the
number of patents already obtained and to the dif- 1500
ference between the number of patents already
obtained and the carrying capacity of the system. 1000
The carrying capacity was approximately 2700
500 t
patents, and the constant of proportionality was
Years
about 7.52 105. By 1883, 980 patents had been
0 10 20 30 40 50 60
since
obtained. 1865
(Source: Hamblin, Jacobsen, and Miller, A Mathematical The-
ory of Social Change, New York: Wiley, 1973.) f. On the slope field, sketch the particular solution
a. Write a differential equation describing the rate for the given initial condition.
of change in the number of patents with respect 31. Consider a function y f(x) whose rate of change
to the number of years since 1865. is jointly proportional to f and to L f .
b. Write a general solution for the differential equa- a. Write a differential equation describing the rate
tion. of change of this function.
c. Write the particular solution for the differential b. Write a general solution for the differential equa-
equation. tion.
d. Estimate the number of patents obtained by c. Verify that the general solution you gave in part
1900. b is indeed a solution by substituting it into the
e. Using the slope field, identify the upper and differential equation and simplifying to obtain
lower horizontal asymptotes. an identity.
8.3 Numerically Estimating by Using Differential
Equations: Euler’s Method
There are differential equations that cannot be solved (or are difficult to solve) using
the methods we have discussed. In these cases, you may wish to analyze the differen-
tial equation using a numerical method. One method we use for such an analysis is
called Euler’s method after the eighteenth-century Swiss mathematician Leonhard
Euler (pronounced “oiler”).
Euler’s method relies on the use of the derivative function to approximate the
change in the quantity function. Recall that if we know a point (a, f(a)) on a function
f and we also know f (a), the slope at that point, then we can approximate the value
of the function at a close point (b, f(b)) as f(b) f(a) (b a) f(a). In using this
approximation, we are using a point on the line tangent to the graph of f at x a to
approximate a point on the function (see Figure 8.17).
We first illustrate the Euler method of approximating change for a differential
equation we can solve. Judging on the basis of data compiled from Statistical Abstract,
1993, 1998, and 2001, the growth rate of the population of the United States between
dP
1900 and 2000 can be expressed as dx 0.0181x 1.129 million people per year x
years after 1900. In 1990 the population was 249.9 million people.
8.3 Numerically Estimating by Using Differential Equations: Euler’s Method 533
The initial condition gives us a starting point, (90, 249.9), and the differential
equation gives us the slope of the line tangent to the graph of P at that point,
P(90) 2.758 million people per year. If we use this information to estimate the
population in the year 2005, then we obtain P(105) 249.9 15(2.758) 291.27
million people. Figure 8.18 illustrates the tangent-line estimate and the estimate
given by the solution P.
Population
(millions)
295 P(x)
290 estimate
Approximation 280
f (x) Tangent line
to (b, f (b))
270
estimate
(b, f (b))
260
250 x
(a, f (a))
Years
240
90 95 100 105 since
x 1900
FIGURE 8.17 FIGURE 8.18
Euler’s method says that we can obtain a better estimate by using many steps
rather than using just one step in the approximation process. In this case, instead of
approximating over 15 years all in one step, we obtain a better estimate by approxi-
mating over 1-year intervals 15 times. Using the initial condition (90, 249.9) and the
differential equation evaluated at x 90, P(90) 2.758, we approximate the
population for the year 1991 as
P(91) 249.9 (1)(2.758) 252.658 million people
Now we have an estimate of a new point, (91, 252.658), and we use the differential
equation to find the slope of the tangent line in 1991, P(91) 2.7761million people
per year. Using this new point and slope, we estimate the population in 1992 as
P(92) 252.658 (1)(2.7761) 255.4341 million people
Similarly, we estimate the population for each of the years from 1993 through 2005.
Table 8.1 records these estimates, and Figure 8.19 illustrates the estimates and the line
segments used to create them. (Note that even though the slopes found are those of
tangent lines, the actual line segments may not be lines tangent to the graph of P but,
rather, lines parallel to those tangent lines.)
Applying Euler’s method using steps of size 1 year yields the estimate that the
population in 2005 will be 293.17 million people. This estimate is slightly larger than
our original estimate of 291.27 and is closer to the estimate of 293.3 that we obtained
in Example 1 of Section 8.1 by finding a particular solution to the differential equa-
tion. (To obtain an estimate closer to that given by the particular solution, we could
use Euler’s method with 100 steps of size 0.1 to improve our results.)
534 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
The graph we obtain by plotting the Euler estimates (Figure 8.19) and drawing the
appropriate line segments is not a smooth curve. However, it does resemble a portion
of the quadratic curve that we expect to see for the solution of this differential equa-
tion and is an approximation to the graph of the population function described by the
differential equation.
TABLE 8.1 P(x)
Population
x Estimate of P (x ) Slope at x
(millions)
90 249.9000 2.7580 295
290
91 252.6580 2.7761
92 255.4341 2.7942 280
93 258.2283 2.8123 270
94 261.0406 2.8304
260
95 263.8710 2.8485
96 266.7195 2.8666 250 x
97 269.5861 2.8847 Years
240
90 95 100 105 since
98 272.4708 2.9028 1900
99 275.3736 2.9209 FIGURE 8.19
100 278.2945 2.9390
101 281.2335 2.9571
102 284.1906 2.9752 Euler’s method is a fairly straightforward way of numerically esti-
103 287.1658 2.9933 mating a solution to a differential equation. However, use caution. Be-
104 290.1591 3.0114 cause each step of Euler’s method is built on an estimate made in the
previous step, it can lead to very bad estimates when the step size is too
105 293.1705
large or when the slopes given by the differential equation change from
very large to very small over a small interval. For these reasons, there
are other methods of numerically estimating solutions that are taught in more ad-
vanced courses. However, because Euler’s method is sufficient for our uses, we will
not consider these other methods.
Example 1 illustrates the application of Euler’s method to the differential equa-
tion given in Example 2 of Section 8.1.
EXAMPLE 1 Applying Euler’s Method to an Equation in One Variable
Sales In Example 2 of Section 8.1, we saw that the rate of change of the total sales
of a computer product could be represented by the differential equation
dS 6.544
billion dollars per year
dt ln(t 1.2)
where t is the number of years after the product was introduced. Also, we were told
that at the end of the first year, sales totaled $53.2 billion. Use Euler’s method to esti-
mate the total sales at the end of the fifth year.
8.3 Numerically Estimating by Using Differential Equations: Euler’s Method 535
Solution
Our starting point (the total sales at the end of the first year) is given as (1, 53.2),
we know that the rate of change at the end of the first year is S(1)
6.544
ln(1 1.2) 8.2998, and we wish to estimate S(5), the total sales at the end
of the fifth year. We use Euler’s method to estimate over an interval of 4 years. In or-
der to keep the step sizes small, we choose to take 16 steps of size 0.25. The first step
gives an estimate for S(1.25):
S(1.25) 53.2 (0.25)(8.2998) 55.275
The estimates (to three decimal places) from each step, as well as the slopes given by
the differential equation, are recorded in Table 8.2.
TABLE 8.2 Euler’s method with step sizes of 0.25 year yields the es-
timate that the total sales for this computer product will be
Estimate of S (t ) Slope at t
$73.6 billion at the end of the fifth year. A graph (Figure
t (billion dollars) (billion dollars per year)
8.20) of the Euler estimates is an approximation of a graph
1.0 53.200 8.2998 of total sales.
1.25 55.275 7.3029
1.5 57.101 6.5885
S(t)
1.75 58.748 6.0491 Total sales
2.0 60.260 5.6261 (billions of dollars)
2.25 61.667 5.2843
75
2.5 62.988 5.0018
70
2.75 64.238 4.7637
3.0 65.429 4.5600 65
3.25 66.569 4.3834
60
3.5 67.665 4.2286
55
3.75 68.722 4.0916
t
4.0 69.745 3.9693
1 2 3 4 5
Years
4.25 70.737 3.8594
4.5 71.702 3.7599 FIGURE 8.20 ●
4.75 72.642 3.6694
5.0 73.559 Euler’s method can be used on most first-order differ-
ential equations, even when separation of variables fails.
EXAMPLE 2 Using Euler’s Method with Two Variables
dy
Consider the differential equation dx 5.9x 3.2y. Given an initial condition of
y(10) 50, estimate y(12).
Solution
Because we know both x and y at x 10, we can use the differential equation to
dy
calculate the slope at x 10 and y 50 as dx 5.9(10)3.2(50) 101.
We choose to use Euler’s method with ten steps of size 0.2. The first estimate is
y(10.2) 50 (0.2)(101) 29.8.
536 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
dy
TABLE 8.3
Because the formula for the slope dx 5.9x 3.2y relies on
x Estimate of y (x ) Slope at x knowing both x and y to estimate the slope at x 10.2, we must use our
10 50 101 estimate y(10.2) 29.8 in the slope formula. Thus, at x 10.2,
10.2 29.8 35.18 dy
5.9(10.2) 3.2(29.8) 35.18
10.4 22.764 11.485 dx
10.6 20.467 2.955 dy
We now use the estimates y(10.2) 29.8 and dx 35.18 at x 10.2
10.8 19.876 0.116 to estimate the value of y at x 10.4:
11 19.899 1.222
y(10.4) 29.8 (0.2)(35.18) 22.764
11.2 20.144 1.620 dy
To find an estimate of y(10.6), we need the slope dx at x 10.4. Again
11.4 20.468 1.763
we must estimate this slope at x 10.4 using y(10.4):
11.6 20.820 1.815
dy
11.8 21.183 1.833 5.9(10.4) 3.2(22.764) 11.4848
dx
12 21.550
Thus the value of y at x 10.6 is
y y(10.6) 22.764 (0.2)(11.4848) 20.46704
50 Proceeding in this manner, we construct Table 8.3 of Euler estimates and find that
40 y(12) 21.55.
30 Graphing the Euler estimates gives us an idea of how the function y behaves (see
Figure 8.21). There appears to be a minimum near x 10.8. The minimum value is
20
x y 19.876. ●
10 10.5 11 11.5 12
FIGURE 8.21
8.3 Concept Inventory dy
b. Given dx 2x with the initial condition (1, 4),
• Euler’s method estimate y when x 7.
• Step size versus number of steps 2. Are the answers to parts a and b of Activity 1
• Graph of Euler estimates approximate or exact? In each case, explain why.
3. Use Euler’s method and two steps to estimate the
following values.
dy 5
a. Given dx y with initial condition (1, 1),
8.3 Activities estimate y when x 5.
dy 5
b. Given dx x with initial condition (2, 2),
Getting Started
estimate y when x 8.
1. Use Euler’s method and two steps to estimate the
following values. 4. Consider the differential equations in parts a and b
dy 1 of Activity 3.
a. Given dx 2 with the initial condition (0, 0),
a. Write the particular solution for the given initial
estimate y when x 4. condition.
8.3 Numerically Estimating by Using Differential Equations: Euler’s Method 537
b. Sketch the particular solution. 7. Production It is estimated that for the first 10
c. Sketch the Euler estimate. years of production, a certain oil well can be ex-
pected to produce oil at a rate of
d. Explain why the Euler estimate deviates from the
true solution. r(t) 3.935t 3.55e1.35135
thousand barrels per year
Applying Concepts t years after production begins.
5. Weight For the first 9 months of life, the average a. Write a differential equation for the rate of
weight w, in pounds, of a certain breed of dog change of the total amount of oil produced t
increases at a rate that is inversely proportional years after production begins.
to time t, in months. A 1-month-old puppy weighs b. Use Euler’s method with ten intervals to estimate
6 pounds. The constant of proportionality is the yield from this oil well during the first 5 years
33.67885. of production.
a. Write a differential equation describing the rate c. Graph the differential equation and the Euler
of change of the weight of the puppy. estimates. Discuss how the shape of the graph of
b. Use Euler’s method with a 0.25-month step the differential equation is related to the shape of
length to estimate the weight of the puppy at 3 the graph of the Euler estimates.
months and at 6 months.
c. Use Euler’s method with a step length of 1 month 8. Labor The personnel manager for a large con-
to estimate the weight of the puppy at 3 months struction company keeps records of the worker
and at 6 months. hours per week spent on typical construction jobs
handled by the company. The manager has devel-
d. Do you expect the answer to part b or the answer oped the following model for a worker hours
to part c to be more accurate? Why? curve:
6. Postage The rate of change (with respect to time) 6,608,830e0.705989x
r(x)
of the number of countries issuing postage stamps (1 925.466e0.705989x)2
between 1836 and 1880 was jointly proportional to worker hours per week
the number of countries that had already issued
postage stamps and to the number of countries that the xth week of the construction job.
had not yet issued postage stamps. The constant of a. Use this model to write a differential equation
proportionality was approximately 0.0049. By 1855, giving the rate of change of the total number
16 countries had issued postage stamps. of worker hours used by the end of the xth
(Source: “The Curve of Cultural Diffusion,” American Socio- week.
logical Review, August 1936, 547–556.)
b. Graph this differential equation, and discuss any
a. Write a differential equation describing the rate critical points and trends that the differential
of change in the number of countries issuing equation suggests will occur.
postage stamps with respect to the number of
c. Use Euler’s method with 20 intervals to estimate
years since 1800.
the total number of worker hours used by the
b. Use Euler’s method with five steps to estimate end of the 20th week.
the number of countries issuing postage stamps
d. Graph the Euler estimates, and discuss whether
in 1840.
you believe the estimate is good. Refer to the
c. Use Euler’s method with a step length of 5 years points you discussed in part b. How could you
to estimate the number of countries issuing improve the accuracy of your estimate?
postage stamps in 1840.
d. Do you expect the answer to part b or the answer 9. Cooling Newton’s Law of Cooling says that
to part c to be more accurate? Why? the rate of change (with respect to time t) of the
538 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
temperature T of an object is proportional to b. Graph the Euler estimates in part a.
the difference between the temperature of the c. Discuss your expectations for the accuracy of
object and the temperature A of the object’s your answers.
surroundings.
a. Write a differential equation describing this law.
13. Production Refer to the information given in
b. Consider a room that has a constant temperature Activity 7 about the rate of production of oil from
of A 70°F . An object is placed in that room an oil well.
and allowed to cool. When the object is first
a. Use Euler’s method with intervals of 1 month,
placed in the room, the temperature of the object
then with intervals of 1 week, and finally with
is 98°F , and it is cooling at a rate of 1.8°F per
daily intervals to estimate the yield from the well
minute. Determine the constant of proportion-
during the first 5 years of production.
ality for the differential equation.
b. Graph the Euler estimates in part a.
c. Use Euler’s method and 15 steps to estimate the
temperature of the object after 15 minutes. c. Discuss your expectations for the accuracy of
your answers.
10. Learning A person learns a new task at a rate that
is equal to the percentage of the task not yet learned. 14. Labor Refer to the information given in Activity 8
Let p represent the percentage of the task already about the labor power on a construction job.
learned at time t (in hours). a. Use Euler’s method with 140 intervals to
a. Write a differential equation describing the rate estimate the total number of worker hours used
of change in the percentage of the task learned at by the end of the 20th week.
time t. b. Graph the Euler estimates in part a.
b. Use Euler’s method with eight steps of size 0.25 c. Discuss your expectations for the accuracy of
to estimate the percentage of the task that is your answers. Does an estimate using 140 inter-
learned in 2 hours. vals have any advantage over an estimate using
c. Graph the Euler estimates, and discuss any criti- 20 intervals? Explain.
cal points or trends.
15. Cooling Refer to the information given in Activ-
11. Explain why you would expect Euler’s method to ity 9 about Newton’s Law of Cooling. Consider
have better accuracy when more steps of smaller size particularly the specification given in part b of
are used. Illustrate with graphs. Activity 9.
a. Use Euler’s method with intervals of length 1
12. Postage Refer to the information given in Activity second to estimate the temperature of the object
6 about the rate of change of the number of coun- after 15 minutes.
tries issuing postage stamps. b. Graph the Euler estimates in part a.
a. Use Euler’s method with 40 steps and with 80 c. How does this estimate compare with that found
steps to estimate the number of countries issuing in part c of Activity 9? Discuss the accuracy of
postage stamps in 1840. this method
8.4 Second-Order Differential Equations 539
8.4 Second-Order Differential Equations
Not all differential equations involve only the first derivative of a function. Some tell
us about the second derivative. That is, they tell us about the rate at which the rate of
change is changing.
Because acceleration is the derivative of velocity v with respect to time t, and ve-
locity is the derivative of distance s with respect to time, we can write acceleration due
to gravity near the surface of the Earth as the differential equation
d2s
32 feet per second squared
dt2
after t seconds. This differential equation is of the form s(t) k for some constant
k. (We saw these equations in Chapter 5 when working with antiderivatives.) This is
a second derivative of a function, so a general solution can be found by finding the
d 2s
general antiderivative of dt 2 twice. The general antiderivative of the differential equa-
tion for acceleration is
ds
32t C feet per second
dt
Thus the general solution to the differential equation is
s(t) 16t 2 Ct D feet
after t seconds.
To confirm that this equation is a solution to the differential equation, we substi-
tute for s in the differential equation:
d2
(16t2 Ct D) 32
dt 2
Differentiating the left side of the equation twice yields the identity 32 32.
In the case where the object fell from a height of 40 feet (s 40 and v 0 at
t 0), we have the particular solution s(t) 16t2 40 feet after t seconds.
EXAMPLE 1 Solving a Second-Order Linear Differential Equation
Purchasing Power The rate of change in the purchasing power* of the U.S. dollar
was changing at a linear rate between 1988 and 2000. A differential equation describing
the rate at which the rate of change of the purchasing power was decreasing is
d 2P
0.00156x 0.0115 dollars per year squared
dx 2
where x is the number of years since 1988. The purchasing power is based on 1982
dollars, making a 1982 dollar worth $0.93 in 1988 and $0.81 in 1992.
a. Write a general solution for the purchasing power of a dollar.
b. Write a particular solution for the purchasing power of a dollar.
c. Use the particular solution to determine how much a 1982 dollar was worth in
2000.
* Based on data from Statistical Abstract, 1994 and 2001.
540 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
Solution
a. By finding the antiderivative of the differential equation twice, we obtain the
general solution
P(x) 0.00026x 3 0.00575x 2 Cx D dollars
where x is the number of years since 1988.
b. The conditions given are P(4) 0.81 and P(0) 0.93. The condition
P(0) 0.93 gives us the value for D as 0.93. Using D 0.93 and P(4) 0.81,
we solve for C and find that C 0.0488. Thus we have the particular solution
P(x) 0.00026x 3 0.00575x 2 0.0488x 0.93 dollars
where x is the number of years since 1988.
c. By 2000, a 1982 dollar was worth only P(12) $0.72. ●
Let us now consider situations in which the second derivative of an amount is
proportional to the amount function—that is, the situation in which
d 2y
ky
dx 2
We consider the case where the constant of proportionality is negative:
d 2y
ky for k 0
dx 2
d2y
Consider the differential equation dx2 y. One function we have looked at that
satisfies this equation is the function y(x) sin x. We verify this by differentiating
the function twice:
y(x) cos x
y(x) sin x y(x)
d 2y
The cosine function is another function that satisfies dx2 y.
d 2y
For the more general case dx2 ky, where k 0, it can be shown that the general
solution is
y(x) a sin(k x c)
where a and c are constants. You will be asked to verify this in the activities.
EXAMPLE 2 Solving a Second-Order Differential Equation
Fishing An exclusive fishing club on the Restigouche River in Canada kept detailed
records regarding the number of fish caught by its members.* Between 1880 and
1905, the rate of change (with respect to the year) in the number of fish per rod per
day was changing at a rate proportional to the number of fish per rod per day. The
constant of proportionality is approximately 0.455625.
* Based on information in E. R. Dewey and E. F. Dakin, Cycles: The Science of Prediction, New York:
Holt, 1947.
8.4 Second-Order Differential Equations 541
The average number of fish per rod per day for the years between 1880 and 1905
was 1.267 fish per rod per day. In 1881, the catch was 0.9425 fish per rod per day, and
in 1885, 1.5776 fish per rod per day were caught.
a. Write a differential equation expressing the rate at which the rate of change in
the number of fish per rod per day was changing with respect to the number of
years since 1880.
b. Write a particular solution to the differential equation.
c. Write a model for the average number of fish per rod per day given the num-
ber of years since 1880.
Solution
a. A differential equation expressing the rate at which the rate of change in g, the
number of fish per rod per day, was changing is
d 2g
0.455625g fish per rod per day per year squared
dy 2
where y is the number of years since 1880.
b. As previously stated, the general solution is g (y) a sin(k y c), where
k 0.455625. Thus the general solution to this differential equation is
g(y) a sin (0.675y c) fish per rod per day
where y is the number of years since 1880, and a and c are constants. In order to
determine the constants a and c for this equation, we must have two conditions.
First, we note that the general solution does not include the parameter d giving
the expected number of fish per rod per day that is normally included in a gen-
eral sine model f (x) a sin (bx c) d. Thus we must write the conditions
in terms of the difference from the expected value. That is, in 1881 the catch was
0.3245 fewer fish per rod per day than in an average year, and in 1885 the catch
was 0.3106 more fish per rod per day than in an average year. Thus the two con-
ditions are g(1) 0.3245 and g(5) 0.3106, resulting in the equations
0.3245 a sin[0.675(1) c] (1)
0.3106 a sin[0.675(5) c] (2)
Solving for a in equation 1 and then substituting in equation 2 yields
0.3245 sin(3.375 c)
0.3106
sin(0.675 c)
which gives c 1.019. Substituting this value of c into either equation 1 or
equation 2 and then solving for a yields a 0.327. Using these constants a
and c, we have the particular solution for the differential equation as
g(y) 0.327 sin(0.675y 1.019) fish per rod per day
where y is the number of years since 1880.
c. The general and particular solutions for the differential equation are both in
terms of the difference from the expected value of fish per rod per day. To write
542 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
the general sine model for this situation, we add the parameter d, the expected
value of fish per day:
g(y) 0.327 sin(0.675y 1.019) 1.267 fish per rod per day
where y is the number of years since 1880. ●
Before we leave this section, let us consider the differential equation of the form
d 2y
ky, where k 0
dx 2
In this case, the constant of proportionality is positive (rather than negative as in
Example 2), and the general solution is
y(x) aekx bekx
where a and b are constants. Note that the general solution to this differential equa-
tion appears to be very different from the general solution in the negative case. The
negative case leads to a model with which we are familiar; the positive case does not.
8.4 Concept Inventory 3. The rate of change in year y of the population P of
the United States is increasing with respect to the
• Second-order differential equations year at a constant rate.
• Solutions to differential equations of the form 4. The rate of growth of the height h of a young child
d 2y with respect to the age y of the child decreases in in-
k
dx 2 verse proportion to the age of the child.
• Solutions to differential equations of the form
d 2y Applying Concepts
f(x)
dx 2
5. Jobs The rate of change in the number of jobs for a
• Solutions to differential equations of the form Michigan roofing company is increasing by approxi-
d 2y mately 6.14 jobs per month squared. The number of
ky
dx 2 jobs in January is decreasing at the rate of 0.87 job per
month, and company records indicate that the com-
pany had 14 roofing jobs in February.
a. Write a differential equation for the rate at which
8.4 Activities the rate of change in the number of roofing jobs
for this company is changing.
Getting Started
b. Find a particular solution to the differential
For Activities 1 through 4, write a differential equation equation in part a.
expressing the information given, and when possible,
c. Use the result of part b to estimate the number
find a general solution for the differential equation.
of roofing jobs in August and the number in
1. The Rowan-Robinson model of the universe November.
assumes that the universe is expanding with respect
6. Marriage Between 1950 and 2000, the rate of
to time t at a rate that is decreasing in inverse pro-
change in the rate at which the median age of first
portion to the square of its current size S.
marriage of females in the United States was chang-
2. The rate of change with respect to time t of the de- ing was constant at 0.0042 year of age per year
mand D for a product is decreasing in proportion to squared. The median age of first marriage for these
the demand at time t. females was increasing at the rate of 0.1713 year of
8.4 Second-Order Differential Equations 543
age per year in 1991, and females were first married current first-class postage for a 1-ounce letter.
at a median age of 25.1 in 2000. Comment on the accuracy of your results. If they
(Source: Based on data from www.infoplease.com Accessed are not reasonable, give possible explanations.
9/20/02.)
9. Consider a function y f(x) whose rate of change
a. Write a differential equation for the rate at which with respect to x is changing at a constant rate.
the rate of the median age of first marriage for
United States females is changing. a. Write a differential equation describing the rate
of change in the rate of change for this function.
b. Find a particular solution to the differential equa-
tion in part a. b. Write a general solution for the differential
equation.
c. Use the result of part b to estimate the median
age of first marriage of United States females in c. Verify that the general solution you gave in part
the current year. b is indeed a solution by substituting it into the
differential equation and simplifying to obtain
7. AIDS Records of the number of AIDS cases diag- an identity.
nosed in the United States between 1988 and 1991
indicate that the rate at which the rate of change in 10. Motion Laws When a spring is stretched and then
the number of cases was changing was constant at released, it oscillates according to two laws of physics:
2099 cases per year squared. The number of AIDS Hooke’s Law and Newton’s Second Law. These two
cases diagnosed in 1988 was 33,590, and the number laws combine to form the following differential equa-
of cases was increasing at the rate of 5988.7 cases per tion in the case of free, undamped oscillation:
year in 1988. d 2x
(Source: Based on data appearing in HIV/AIDS Surveillance m kx 0
dt 2
1992 Year End Edition.)
a. Write a differential equation for the rate of change where m is the mass of an object attached to the
in the rate of change of the number of AIDS cases spring, x is the distance the spring is stretched beyond
diagnosed in year t, where t is the number of years its standard length with the object attached (its equi-
after 1988. librium point), t is time, and k is a constant associated
with the strength of the spring. Consider a spring
b. Find a particular solution to the differential with k 15 from which is hung a 30-pound weight.
equation in part a. The spring with the weight attached stretches to its
c. Estimate how rapidly the number of AIDS cases equilibrium point. The spring is then pulled 2 feet
diagnosed was changing in 1991 and the number farther than its equilibrium and released.
of AIDS cases that were diagnosed in that year.
a. Write a differential equation describing the
8. Postage Between 1919 and 1995, the rate of acceleration of the spring with respect to time
change in the rate of change of the postage required t measured in seconds. Use the fact that mass
to mail a first-class 1-ounce letter was approxi- weight
g , where g is the gravitational constant 32 feet
mately 0.022 cent per year squared. The postage was
2 cents in 1919, and it was increasing at the rate of per second per second.
approximately 0.393 cent per year in 1958. b. Write a particular solution for this differential
(Source: Based on data from the United States Postal Service.) equation. Use the fact that when the spring is
first released, its velocity is 0.
a. Write a differential equation for the rate of
change in the rate of change of the first-class c. Graph this solution over several periods, and ex-
postage for a 1-ounce letter in year t, where t is plain how to interpret the graph.
the number of years after 1900. d. How quickly is the mass moving when it passes
b. Find both a general and a particular solution to its equilibrium point?
the differential equation in part a. 11. Radiation The rate of change in the rate at which
c. Use your previous results to estimate how rapidly the average amount of extraterrestrial radiation in
the postage is changing in the current year and the Amarillo, Texas, for each month of the year is chang-
544 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
ing is proportional to the amount of extraterrestrial d. How well does your model estimate the amounts
radiation received. The constant of proportionality is of extraterrestrial radiation in March and
k 0.212531. In any given month, the expected September?
value of radiation is 12.5 mm per day. This expected
12. Consider a function y f(x) whose rate of change
value is actually obtained in March and September.
with respect to x is changing in proportion to f,
(Source: Based on data from A. A. Hanson, ed., Practical Hand-
book of Agricultural Science, Boca Raton, FL: CRC Press, 1990.) where the constant of proportionality is negative.
a. Write a differential equation for the information a. Write a differential equation describing the rate
given. of change in the rate of change for this function.
b. In June, the amount of radiation received is ap- b. Write a general solution for the differential
proximately 17.0 mm per day, and in December, equation.
the amount of radiation received is approxi- c. Verify that the general solution you gave in part
mately 7.8 mm per day. Write a particular solu- b is indeed a solution by substituting it into the
tion for this differential equation. differential equation and obtaining an identity.
c. Change the particular solution into a function
giving the average amount of extraterrestrial
radiation in Amarillo.
SUMMARY
In many situations (physical, sociological, psychologi- Slope Fields
cal, and economic), there are certain laws that govern It is often helpful to view a graphical representation of a
the dynamics of change. In these cases, differential differential equation. This can be done in the form of a
equations often can be used to describe the change that slope field. A slope field consists of line segments that
occurs. represent the slopes given by the differential equation at
Differential Equations different points on a grid placed on the coordinate plane.
Differential equations are equations involving deriva- Separable Differential Equations
tives. A solution for a differential equation is a function
that, when substituted into the differential equation, Some differential equations are given as functions not
results in an identity. A general solution contains arbi- only of the input variable but also of the output variable,
trary constants that can be determined if initial condi- so they are of the form
tions are known. When the arbitrary constants are dy
replaced by known constants, the solution is called a f(x, y)
dx
particular solution.
Some differential equations are of the simple form In this case, a general solution might be obtained by
using the method of separation of variables.
dy
f(x)
dx Euler’s Method
In this case, a general solution can be obtained by writ- Often, neither simple antidifferentiation nor separation
ing a general antiderivative for the given rate of change of variables is sufficient for determining an explicit for-
if an antiderivative is known. mula for a general solution of a differential equation. In
Chapter 8 Concept Review 545
these cases, we can numerically estimate a solution using Second-Order Differential Equations
Euler’s method. Euler’s method not only gives us a nu- Second-order differential equations involve the second
merical method for estimating solutions but also, when derivative of a function. The simple cases we considered
the Euler estimates are graphed, can help us visualize were the second-order differential equations that give
how the solution function behaves. rise to the quadratic, cubic, and sine models.
CONCEPT CHECK
Can you To practice, try
• Solve constant and linear differential equations? Section 8.1 Activities 15, 19
• Set up an equation involving direct or inverse Section 8.1 Activities 19, 23
proportionality?
• Sketch a particular solution to a differential Section 8.1 Activities 7, 9
equation on a slope field? Section 8.2 Activities 9, 13, 15
• Solve separable differential equations? Section 8.2 Activity 25
• Set up an equation involving joint proportionality? Section 8.2 Activity 29
• Estimate a solution to a differential equation Section 8.3 Activity 7
using Euler’s method?
• Set up and solve second-order differential Section 8.4 Activities 7, 11
equations using antiderivatives?
CONCEPT REVIEW
1. Accident Risk The relationship between the rela- b. Write a general solution for the differential
tive risk R (expressed as a percent) of having a car equation.
accident and the blood alcohol level of 100b% of the
c. The risk of an accident is 1% when there is no al-
driver is
cohol in the blood and 20% when the blood al-
dR cohol level is 14%. Write a particular solution for
kR
db the differential equation.
a. Write a statement interpreting this differential d. At what blood alcohol level is a crash certain to
equation. occur?
546 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
e. Use the slope field to sketch the particular solu- d. Use the particular solution from part c to esti-
tion found in part c. mate the Irish population in 1840 and 1850.
3. Famine Consider the differential equation from
R(b)
(percent) part a of Activity 2 and the initial condition that in
20
1780 there were 4.0 million people in Ireland.
18 a. Use Euler’s method and 10-year steps to estimate
16 the population of Ireland in 1840 and 1850.
14 b. Use Euler’s method and 5-year steps to estimate
12 the population of Ireland in 1840 and 1850.
10
c. How do your answers to parts a and b compare
8
to those from part d of Activity 2?
6
4 4. Population Population data for 1850 through
2 1920 show that Ireland’s population was decreasing
b according to the Verhulst model with a constant of
0 0.02 0.06 0.10 0.14
proportionality of 0.008307. Ireland reached a pop-
2. Famine In Ireland in the 1840s, potato famines ulation low of 4.4 million people in 1910 and 1920.
occurred. These famines are correlated with a To use the Verhulst model for population change,
drastic change in the dynamics of population we assume that the carrying capacity of Ireland was
growth and decline. The Verhulst model for popu- 7.154 million people above its 1910–1920 level.
lation says that the rate of change in a population is a. Write a differential equation expressing the rate
jointly proportional to the current population and of change of Ireland’s population between 1850
to the capacity remaining in the system. and 1920. Let x represent the number of years
a. From population data for 1780 through 1840, since 1800.
the carrying capacity of Ireland’s population b. Using the fact that in 1900, there were 4.5 million
appears to have been 16.396 million people. The people in Ireland (that is 0.1 million people
constant of proportionality for growth during above its 1910–1920 level), write a particular
this time was 0.001175. Write a differential equa- solution to the differential equation from part a.
tion expressing the rate of change of Ireland’s c. Adjust the model for 1850 through 1920 by
population. Let x represent the number of years adding the base population level of 4.4 million
since 1800. people, and write a piecewise model for the
b. Write a general solution for the differential equa- population of Ireland from 1780 through 1920.
tion in part a. d. Use the piecewise model from part c to estimate
c. Write a particular solution for the differential the population of Ireland in 1850. How does this
equation in part a. In 1780, there were 4.0 million differ from the estimate you found in part d of
people in Ireland. Activity 2?
Project 8.1 On-the-Job Training
Setting Tasks
You are the personnel manager for a large
1. Write a differential equation describing the rate of
manufacturing firm. The shareholders have asked you
change in the percentage of the task learned at
to estimate the cost for personnel training during the
time t.
start-up of a new plant that will employee 56 workers.
The firm has agreed to pay newly hired employees 2. Use Euler’s method at least three times with
$10 per hour during their training. The firm has also decreasing step sizes to estimate the percentage of
agreed to pay current employees from another plant the task that is learned in 2 weeks.
$45 per hour for each hour they spend training two
3. Graph the Euler estimates, and discuss any critical
new employees. It is estimated that for this job, the
points or trends.
average newly hired employee will be able to learn his
or her job at a rate that is equal to the percentage of 4. Determine at what point the newly hired
the task not yet learned, where time is measured in employees may be ready to handle the job on
40-hour weeks. their own.
5. Formulate a budget for the training of the new
personnel.
Reporting
Prepare a report that presents your findings in Tasks
1 through 5. Explain the method that you used to
arrive at your conclusions. Attach your mathematical
work as an appendix to your report.
547
9 Ingredients of
Multivariable Change:
Models, Graphs, Rates
Concepts Outline
9.1 Multivariable Functions and
Contour Graphs
9.2 Cross-Sectional Models and
Rates of Change
9.3 Partial Rates of Change
9.4 Compensating for Change
Sonda Dawes/The Image Works
Concept Application
Competition is a fundamental element of a free-market society. Although many factors
affect sales of competing products, the most obvious factor is the price of the product. If
side-by-side vending machines sell competitive products, and if sales data are collected
and modeled as a function of the two prices, then we can use the data and model to an-
swer questions such as
• If one of the competitors lowers the price, what change can the other company expect
in sales?
• How quickly are sales changing for two given prices?
• If one company lowers its price, how should the other company respond in order to
maintain its current sales level?
You will have an opportunity to answer questions such as these in Activity 28 of Section
9.3 and Activity 21 of Section 9.4.
548
9.1 Multivariable Functions and Contour Graphs 549
Chapter Introduction
The mathematics of change for functions of a single input variable was discussed in
chapters 1 through 8 of this text. The mathematics of change for multivariable func-
tions will be presented in Chapters 9 and 10. A multivariable function is a function
with two or more input variables.
Graphs of functions with two input variables are three-dimensional and represent
surfaces in space. We consider associated two-dimensional contour graphs and how
they are related to the three-dimensional surfaces. Our understanding of any func-
tion and how it changes is greatly enhanced by a visual presentation.
We investigate change in a multivariable function by considering the rate of
change with respect to any one of the input variables. We also discuss compensating
for change in the context of an important consideration in any manufacturing
process: the allocation of available resources to produce a fixed level of output.
Concepts you will be learning
• Constructing input/output diagrams for multivariable functions (9.1)
• Interpreting multivariable function inputs and outputs (9.1)
• Drawing contour curves (9.1)
• Interpreting contour graphs (9.1)
• Finding and interpreting cross-sectional models (9.2)
• Estimating partial rates of change from tables (9.2)
• Finding and interpreting first and second partial derivatives (9.3)
• Constructing a second-partials matrix (9.3)
• Calculating the slope on a contour curve at a specified point (9.3)
• Estimating the “compensation for a change” (9.4)
9.1 Multivariable Functions and Contour Graphs
In Chapters 1 through 8 we studied change using single-variable functions. Each func-
tion had one input variable and one output variable. We now consider multivariable
functions. These functions have two or more input variables and one output variable.
Multivariable Functions
Many of the functions we encounter in everyday situations are multivariable func-
tions. These are functions that depend on two or more input variables. A manufac-
turer’s profit depends on sales, variable costs, fixed costs, and perhaps other input
variables that are related to inventory costs, equipment failures, and labor strikes. In
forestry, the volume of a tree is a function of its height and diameter. Crop yield is a
function of temperature, rainfall, amount of fertilizer, and so forth.
The input of a multivariable function is denoted by (x1, x2, x3, . . . , xn). For each
input of a function f, there is only one output f(x1, x2, x3, . . . , xn).
550 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
A rule f that relates one output variable to several input variables (x1, x2, x3, . . .,
xn) is called a multivariable function if for each input (x1, x2, x3, . . . , xn) there
is exactly one output f(x1, x2, x3, . . . , xn).
A familiar example of a multivariable function is the value of an investment. The
value of an investment is modeled by
r nt
A(P, r, n, t) P 1 n dollars
where P dollars is the amount originally invested, r is the annual interest rate (writ-
ten in decimal form), n is the number of times interest is paid each year, and t years
is the time since the principal was deposited. An input/output diagram of this func-
tion is shown in Figure 9.1.
r n
P Interest Compoundings t
Dollars rate each year Years
Inputs
Rule A
Output
A(P, r, n, t)
FIGURE 9.1 Dollars
The notation A(5000, 0.06, 4, 10) represents the accumulated value of a $5000 in-
vestment after 10 years when the annual interest rate is 6% compounded quarterly.
We find this accumulated value to be
0.06 4(10)
A(5000, 0.06, 4, 10) 5000 1 $9070.09
4
and we say, “The value of a $5000 investment after 10 years with 6% interest paid
quarterly is $9070.09.”
A second example deals with the elevation of a plot of farmland. We do not have
a multivariable equation that can be used to model this situation. However, it is pos-
sible to find the elevation of the land through various surveying methods. We do have
a data table that presents the elevation at 0.1-mile intervals in two directions.
Elevation is a function whose output (distance above sea level) is determined by
knowing the location of a point in terms of a reference point. Although we could lo-
cate each point in terms of its longitude and latitude, we simplify the numbers and let
the southwest corner of the tract be our reference point. We measure distance from
this point in miles from the southern fence and miles from the western fence. The two
input variables are e and n, where e is the distance in miles east of the western fence
and n is the distance in miles north of the southern fence. The output of the elevation
9.1 Multivariable Functions and Contour Graphs 551
function is E(e, n), the elevation of the tract in feet above sea level at position (e, n).
An input/output diagram for this function is shown in Figure 9.2. Graphically, E(e, n)
can be represented in three dimensions as shown in Figure 9.3.
Numerically, E(e, n) can be represented by Table 9.1. From the table, we read the
elevation at a position 0.3 mile east and 1 mile north of the southwest corner as
E(0.3, 1) 799.7 feet above sea level. Note that e 0.3 because the position is 0.3
mile east of the western fence. Likewise, n 1 because the position is 1 mile north of
the southern fence.
e n N
Miles Miles Western W E
Inputs east north fence
E(e, n) S
(feet above
sea level)
Rule E 803
1.5
793 1.0
Output 0 n
E(e, n) Southern 0.5 0.5 (miles)
Feet above fence e 1.0
sea level (miles) 1.5
FIGURE 9.2 Three-dimensional graph of elevation
FIGURE 9.3
TABLE 9.1 Elevation E(e, n) in feet above sea level on a tract of Missouri farmland measured e miles east of the
western fence and n miles north of the southern fence
e→ 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
1.5 800.3 799.1 798.3 797.8 797.5 797.4 797.4 797.5 797.6 797.7 797.6 797.3 796.8 796.0 794.8 793.2
1.4 800.8 799.6 798.8 798.3 798.0 797.9 797.9 798.0 798.1 798.1 798.1 797.8 797.3 796.5 795.3 793.7
1.3 801.2 800.0 799.2 798.7 798.4 798.3 798.3 798.4 798.5 798.6 798.5 798.2 797.7 796.9 795.7 794.1
1.2 801.6 800.4 799.6 799.1 798.8 798.7 798.7 798.8 798.9 798.9 798.9 798.6 798.1 797.3 796.1 794.5
1.1 801.9 800.7 799.9 799.4 799.1 799.0 799.0 799.1 799.2 799.3 799.2 798.9 798.4 797.6 796.4 794.8
1.0 802.2 801.0 800.2 799.7 799.4 799.3 799.3 799.4 799.5 799.5 799.5 799.2 798.7 797.9 796.7 795.1
0.9 802.4 801.2 800.4 799.9 799.6 799.5 799.5 799.6 799.7 799.8 799.7 799.4 798.9 798.1 796.9 795.3
0.8 802.6 801.4 800.6 800.1 799.8 799.7 799.7 799.8 799.9 799.9 799.9 799.6 799.1 798.3 797.1 795.5
0.7 802.7 801.5 800.7 800.2 799.9 799.8 799.8 799.9 800.0 800.1 800.0 799.7 799.2 798.4 797.2 795.6
0.6 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.1 800.1 799.8 799.3 798.5 797.3 795.7
0.5 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.2 800.1 799.8 799.3 798.5 797.3 795.7
0.4 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.1 800.1 799.8 799.3 798.5 797.3 795.7
0.3 802.7 801.5 800.7 800.2 799.9 799.8 799.8 799.9 800.0 800.1 800.0 799.7 799.2 798.4 797.2 795.6
0.2 802.6 801.4 800.6 800.1 799.8 799.7 799.7 799.8 799.9 799.9 799.9 799.6 799.1 798.3 797.1 795.5
0.1 802.4 801.2 800.4 799.9 799.6 799.5 799.5 799.6 799.7 799.8 799.7 799.4 798.9 798.1 796.9 795.3
0 802.2 801.0 800.2 799.7 799.4 799.3 799.3 799.4 799.5 799.5 799.5 799.2 798.7 797.9 796.7 795.1
n↑
552 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
Sketching Contour Curves
A topographical map is a two-dimensional graph that shows the contour of the ter-
rain by outlining different elevations. The curves that outline the land at different but
constant elevations are known as contour curves. In general, a contour curve for a
three-dimensional function is the collection of all points (x, y) for which f(x, y) K,
where K is a constant. A contour curve is a two-dimensional outline of a three-
dimensional graph at a given output level. For a specific value of K, we sometimes call
the contour curve the K contour curve.
A contour curve for a three-dimensional function is the collection of all
points (x, y) for which f(x, y) K , where K is a constant. This curve is
referred to as the K contour curve.
Graphically, a contour curve is the two-dimensional outline of a three-
dimensional graph at a given output level.
The teal contour curves shown in Figure 9.4 outline the shape of the land at an el-
evation of 800 feet above sea level. The contour curves we will sketch and the ones you
will be asked to sketch should be smooth and should have as few concavity changes
as possible.
e 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
1.5 800.3 799.1 798.3 797.8 797.5 797.4 797.4 797.5 797.6 797.7 797.6 797.3 796.8 796.0 794.8 793.2
1.4 800.8 799.6 798.8 798.3 798.0 797.9 797.9 798.0 798.1 798.1 798.1 797.8 797.3 796.5 795.3 793.7
1.3 801.2 800.0 799.2 798.7 798.4 798.3 798.3 798.4 798.5 798.6 798.5 798.2 797.7 796.9 795.7 794.1
1.2 801.6 800.4 799.6 799.1 798.8 798.7 798.7 798.8 798.9 798.9 798.9 798.6 798.1 797.3 796.1 794.5
1.1 801.9 800.7 799.9 799.4 799.1 799.0 799.0 799.1 799.2 799.3 799.2 798.9 798.4 797.6 796.4 794.8
1.0 802.2 801.0 800.2 799.7 799.4 799.3 799.3 799.4 799.5 799.5 799.5 799.2 798.7 797.9 796.7 795.1
0.9 802.4 801.2 800.4 799.9 799.6 799.5 799.5 799.6 799.7 799.8 799.7 799.4 798.9 798.1 796.9 795.3
0.8 802.6 801.4 800.6 800.1 799.8 799.7 799.7 799.8 799.9 799.9 799.9 799.6 799.1 798.3 797.1 795.5
0.7 802.7 801.5 800.7 800.2 799.9 799.8 799.8 799.9 800.0 800.1 800.0 799.7 799.2 798.4 797.2 795.6
0.6 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.1 800.1 799.8 799.3 798.5 797.3 795.7
0.5 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.2 800.1 799.8 799.3 798.5 797.3 795.7
0.4 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.1 800.1 799.8 799.3 798.5 797.3 795.7
0.3 802.7 801.5 800.7 800.2 799.9 799.8 799.8 799.9 800.0 800.1 800.0 799.7 799.2 798.4 797.2 795.6
0.2 802.6 801.4 800.6 800.1 799.8 799.7 799.7 799.8 799.9 799.9 799.9 799.6 799.1 798.3 797.1 795.5
0.1 802.4 801.2 800.4 799.9 799.6 799.5 799.5 799.6 799.7 799.8 799.7 799.4 798.9 798.1 796.9 795.3
0 802.2 801.0 800.2 799.7 799.4 799.3 799.3 799.4 799.5 799.5 799.5 799.2 798.7 797.9 796.7 795.1
n
FIGURE 9.4
9.1 Multivariable Functions and Contour Graphs 553
N Note that there are two parts for the 800-feet contour curve. The 800-
Western W E feet contour curves separate elevations that are above 800 feet from those
fence
E(e, n) S elevations that are less than 800 feet. Elevations inside the egg-shaped
(feet above curve are above 800 feet. Figure 9.5 shows the three-dimensional repre-
sea level) sentation of the contour curve drawn in Figure 9.4.
803 Figure 9.6 gives a contour graph showing contour curves at eleva-
1.5 tions of 796, 797, 798, 799, 800, 801, and 802 feet above sea level. The
contour curves at various elevations on the tract form the topograph-
793 1.0
0 n ical map, or contour graph, that is shown in Figure 9.7a. Figure 9.7b
Southern 0.5 0.5 (miles) shows the contour curves drawn on the three-dimensional graph.
fence e 1.0 Because each contour curve represents all the points on the surface
(miles) 1.5 that correspond to a particular level of output, contour curves are
FIGURE 9.5 sometimes called level curves.
In general, a contour graph is a graph of contour curves f(x, y) K for more
than one value of K. Usually, the values of K are evenly spaced. When the values of
K are evenly spaced, we can tell something about the steepness of the underlying
three-dimensional surface by noting how close together the contour curves are. If
the contour curves (for evenly spaced K) are close together, the surface is steeper in
that region than in a portion of the graph where the contour curves are spaced far-
ther apart.
e 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
1.5 800.3 799.1 798.3 797.8 797.5 797.4 797.4 797.5 797.6 797.7 797.6 797.3 796.8 796.0 794.8 793.2
1.4 800.8 799.6 798.8 798.3 798.0 797.9 797.9 798.0 798.1 798.1 798.1 797.8 797.3 796.5 795.3 793.7
1.3 801.2 800.0 799.2 798.7 798.4 798.3 798.3 798.4 798.5 798.6 798.5 798.2 797.7 796.9 795.7 794.1
1.2 801.6 800.4 799.6 799.1 798.8 798.7 798.7 798.8 798.9 798.9 798.9 798.6 798.1 797.3 796.1 794.5
1.1 801.9 800.7 799.9 799.4 799.1 799.0 799.0 799.1 799.2 799.3 799.2 798.9 798.4 797.6 796.4 794.8
1.0 802.2 801.0 800.2 799.7 799.4 799.3 799.3 799.4 799.5 799.5 799.5 799.2 798.7 797.9 796.7 795.1
0.9 802.4 801.2 800.4 799.9 799.6 799.5 799.5 799.6 799.7 799.8 799.7 799.4 798.9 798.1 796.9 795.3
0.8 802.6 801.4 800.6 800.1 799.8 799.7 799.7 799.8 799.9 799.9 799.9 799.6 799.1 798.3 797.1 795.5
0.7 802.7 801.5 800.7 800.2 799.9 799.8 799.8 799.9 800.0 800.1 800.0 799.7 799.2 798.4 797.2 795.6
0.6 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.1 800.1 799.8 799.3 798.5 797.3 795.7
0.5 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.2 800.1 799.8 799.3 798.5 797.3 795.7
0.4 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.1 800.1 799.8 799.3 798.5 797.3 795.7
0.3 802.7 801.5 800.7 800.2 799.9 799.8 799.8 799.9 800.0 800.1 800.0 799.7 799.2 798.4 797.2 795.6
800
0.2 802.6 801.4 800.6 800.1 799.8 799.7 799.7 799.8 799.9 799.9 799.9 799.6 799.1 798.3 797.1 795.5
0.1 802.4 801.2 800.4 799.9 799.6 799.5 799.5 799.6 799.7 799.8 799.7 799.4 798.9 798.1 796.9 795.3
0 802.2 801.0 800.2 799.7 799.4 799.3 799.3 799.4 799.5 799.5 799.5 799.2 798.7 797.9 796.7 795.1
n 802 801 800 799 798 797 796
FIGURE 9.6
554 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
N N
n
W E Western W E
(miles)
S fence S
796
1.4 E(e, n)
797 (feet above
1.2
798 sea level)
Western fence
1.0 8 796
803
0.8 800 799 800 8
801 800 1.5
0.6
802 1.0
0.4 793 n
0 (miles)
0.2 Southern 0.5
0.5
e fence e
0 1.0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 (miles) (miles) 1.5
Southern fence
(a) (b)
FIGURE 9.7
EXAMPLE 1 Interpreting a Contour Curve Sketched on a Table of Data
Payments The data in the table shown in Figure 9.8 represent the monthly pay-
ments M(t, A) required to pay off a mortgage of A thousand dollars over a period of
t years when interest is paid on the loan at a rate of 7%.
t, Period of loan (years)
10 12.5 15 17.5 20 22.5 25 27.5 30
5 58.05 50.11 44.94 41.36 38.76 36.82 35.34 34.18 33.27
10 116.11 100.22 89.88 82.72 77.53 73.65 70.68 68.36 66.53
15 174.16 150.32 134.82 124.08 116.29 110.47 106.02 102.54 99.80
20 232.22 200.43 179.76 165.44 155.06 147.30 141.36 136.72 133.06
A, Amount borrowed (thousand dollars)
25 290.27 250.54 224.71 206.80 193.82 184.12 176.69 170.90 166.33
30 348.32 300.65 269.65 248.16 232.59 220.95 212.03 205.08 199.59
35 406.38 350.75 314.59 289.52 271.35 257.77 247.37 239.26 232.86
40 464.43 400.86 359.53 330.88 310.12 294.60 282.71 273.45 266.12
45 522.49 450.97 404.47 372.24 348.88 331.42 318.05 307.63 299.39
50 580.54 501.08 449.41 413.60 387.65 368.24 353.39 341.81 332.65
55 638.59 551.18 494.35 454.96 426.41 405.07 388.73 375.99 365.92
60 696.65 601.29 539.29 496.31 465.18 441.89 424.07 410.17 399.18
65 754.70 651.40 584.24 537.67 503.94 478.72 459.41 444.35 432.45
70 812.76 701.51 629.18 579.03 542.71 515.54 494.74 478.53 465.71
75 870.81 751.62 674.12 620.39 581.47 552.37 530.08 512.71 498.98
80 928.86 801.72 719.06 661.75 620.24 589.19 565.42 546.89 532.24
FIGURE 9.8 Contour curve for $520 payment
9.1 Multivariable Functions and Contour Graphs 555
a. Find and interpret M(15, 50).
b. If a prospective house buyer can afford to pay only $520 each month, what are the
options?
c. Explain how increasing the period of the loan affects the mortgage amount if the
monthly payment remains constant.
Solution
a. We find the value that is in both the column for t 15 and the row for A 50:
M(15, 50) 449.41 dollars. The monthly payment is $449.41 on a $50,000,
15-year mortgage when the rate is fixed at 7%.
b. We answer this question by sketching a contour curve at the $520 level (as
depicted in Figure 9.8) and identifying points on that curve. There are many
available options. For instance, a buyer who wants to pay off the mortgage within
12 years can apply for a loan of around $50,000. If a buyer wants a 20-year mort-
gage, then the amount borrowed can be between $65,000 and $70,000. Similarly,
if the buyer is willing to pay on the mortgage for 30 years, a mortgage slightly
larger than $75,000 can be obtained.
c. If the loan period increases and the buyer wishes to remain on the contour curve
(thus keeping the monthly payment constant), then the mortgage amount also
increases. ●
Data tables do not show every possible value for the input and output variables.
For instance, in Example 1, you cannot determine from the data table the exact loan
amount that can be obtained for a 20-year mortgage at 7% interest paid off with
monthly payments of $520. If, however, a multivariable function is given, then exact
values on a particular contour curve can be determined.
Given a function with two input variables, we draw contour graphs by plotting
points that correspond to certain levels of output. We graph the K contour curve of a
function by setting the function equal to K, picking values for one variable, solving
for values of the other variable, and plotting the points generated. This three-step
process is illustrated in the solution to part b of Example 2.
EXAMPLE 2 Obtaining a More Accurate Contour Curve
Heat Loss Siple and Passel were two explorers who accompanied Admiral Byrd to
the Antarctic and were among the first researchers to measure loss of body heat due
9.1.1 to the wind. From their data, they obtained the equation*
H(v, t) (10.45 10v v)(33 t)
which gives the body’s heat loss in kilogram-calories per square meter of body surface
area per hour, for wind speed v, in meters per second, when the air temperature is t°C.
Table 9.2 shows data generated from the model.
*W. Bosch and L. G. Cobb, “Windchill,” UMAP Module 658, The UMAP Journal, vol. 5, no. 4, (Winter
1984), 477–492.
556 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
TABLE 9.2 A body’s heat loss, in kilogram-calories per square meter of
body surface area per hour
Wind speed (meters per second)
Air temperature (°F) 0 5 10 15 20 25
20 554 1474 1700 1812 1864 1879
25 606 1613 1860 1982 2040 2056
30 658 1752 2021 2153 2216 2233
35 711 1891 2181 2324 2392 2411
40 763 2030 2341 2495 2568 2588
a. On the table, sketch a contour curve for a heat loss of 2000 kilogram-calories
per square meter of body surface area per hour.
b. Use the equation to sketch this contour curve more accurately for wind speed
values between 5 and 25 meters per second.
Solution
a. The 2000-kilogram-calories contour curve is shown on the table in Figure 9.9.
Wind speed (meters per second)
Air temperature (°F) 0 5 10 15 20 25
-20 554 1474 1700 1812 1864 1879
2000
-25 606 1613 1860 1982 2040 2056
-30 658 1752 2021 2153 2216 2233
-35 711 1891 2181 2324 2392 2411
-40 763 2030 2341 2495 2568 2588
FIGURE 9.9
b. We obtain a more accurate graph of this contour curve by setting the function
H equal to 2000-kilogram-calories, choosing values for wind speed v, and solv-
ing for values of the temperature t. This process yields a set of points that can
be used to sketch a contour curve.
Step 1. Set H(v, t) 2000.
(10.45 10v v)(33 t) 2000
The values for v are chosen so Step 2. Let v 5. Solving (10.45 105 5)(33 t) 2000 for t yields
that they are evenly spaced across t 38.9 giving the point (5, 38.9).
the input interval. We choose
enough points so that a pattern Then let v 10. Solving (10.45 1010 10)(33 t) 2000 gives ap-
can be observed in the graph
when the points are plotted.
proximately (10, 29.4).
We repeat Step 2 a few more times using 15, 20, and 25 for v. The set of points
generated is
{(5, 38.9), (10, 29.4), (15, 25.5), (20, 23.9), (25, 23.4)}
9.1 Multivariable Functions and Contour Graphs 557
Step 3. We plot the points and sketch the curve (see Figure 9.10).
t
(°C)
0 5 10 15 20 25 v
−20
(m/sec)
−25
−30
−35
−40
FIGURE 9.10
This curve has an increasing, concave-down shape similar to the one shown in
Figure 9.9 but is more accurate. We can also use the equation to sketch contour
curves for inputs other than those shown in the table. ●
Formulas for Contour Curves
Often it is helpful to be able to sketch a contour graph showing several contours. In
this case, it is beneficial to find a formula to use to graph each contour curve. Be-
cause contour curves are generated by choosing a constant value K for output, we
set the function equal to K and solve for one of the variables (whichever is easier).
This gives a general formula that we can use to graph contour curves by replacing
K with appropriate constants. In Example 3, we illustrate this approach to graph-
ing contours.
EXAMPLE 3 Using a Multivariable Function to Sketch a Contour Graph
Heat Loss Using the Siple and Passel heat-loss equation described in Example 2,
H(v, t) (10.45 10v v)(33 t)
find a general formula for the H(v, t) contour curves, and use it to sketch a contour
graph for heat-loss levels of 500, 1000, 1500, 2000, and 2500 kilogram-calories per
square meter per hour.
Solution
Replacing H(v, t) with K and then solving for t gives the general formula
K
t 33
10.45 10v v
558 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
We obtain the contour graph shown in Figure 9.11 by applying the three-step
process (illustrated in Example 1) for each of the following values of K: 500, 1000,
1500, 2000, and 2500.
t
(°C)
500
20
15
10
1000
5
v
0
5 10 15 20 25 (m/sec)
-5
1500
-10
-15
-20 2000
-25
-30
-35 2500
-40
FIGURE 9.11 ●
Estimating Output and Change in Output Using
Contour Graphs
Figures 9.12a and 9.12b show a three-dimensional representation and a contour
graph of the function z M(t, A) , which gives the amount of the monthly
payments necessary to pay off a loan of A thousand dollars over t years at a 7% in-
terest rate.
A
Mortgage amount
(thousand dollars)
M(t, A)
Monthly 80
payment C 840
75 C
(dollars) B 760 600
70
680 520
800 65
600 80
60 440
400 70 A
55 B
10
60 Mortgage
15 50 amount 360
20 50
25 (thousand t
t 30 45
Years dollars) 10 12.5 15 17.5 20 22.5 25 27.5 30 Years
(a) (b)
FIGURE 9.12
9.1 Multivariable Functions and Contour Graphs 559
Note that in Figure 9.12a, it is possible to estimate that the input values correspon-
ding to the point C are t 15 years and A 75 thousand dollars. It is difficult to es-
timate the output at that point. It is easier to use the contour graph when estimating
the input and output values. Because C is located slightly below the $680 contour curve
and is on the side closer to the $600 contour curve, we estimate that the output of C is
slightly less than $680. We estimate that the output at C is approximately $675.
As illustrated in Example 4, we can use contour graphs to help estimate the
change in outputs of a multivariable function.
EXAMPLE 4 Estimating Change Using a Contour Graph
Payments Refer to Figure 9.12, which shows three-dimensional and contour
graphs of the function z M(t, A). M(t, A) is the monthly payment (in dollars) nec-
essary to pay off a loan of A thousand dollars over t years at a 7% interest rate.
a. How does the payment on a $55,000 mortgage change when the length of the
loan is increased from 15 years to 25 years?
b. Consider a $75,000, 15-year mortgage. Does the monthly payment change
more when the amount of the mortgage is increased by $5000 or when the term
of the mortgage is increased by 10 years?
Solution
a. Refer to Figure 9.13. Locate the points M(15, 55) and M (25, 55) on the con-
tour graph. Because M(15, 55) is between the $520 and $440 contour curves but
is closer to the $520 contour, we estimate the output to be $490. Because M(25,
55) is between the $360 and $440 contours but is closer to the $360 contour, we
estimate the output to be $390.
A
Mortgage amount
(thousand dollars)
90
85
80
840
75
760 600
70
680 520
65
60 440
Increase loan
55
term
360
50
t
45
10 12.5 15 17.5 20 22.5 25 27.5 30 Years
FIGURE 9.13
560 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
When the term of a $55,000 mortgage is increased from 15 years to 25 years, the
monthly payment decreases by about $100.
b. Refer to Figure 9.14. M (15, 75) is about $675.
A
Mortgage amount
(thousand dollars)
90 840
760
85 680
80 Increase
75 amount
Increase loan term
600
70
520
65
60 440
55
360
50
t
45
10 12.5 15 17.5 20 22.5 25 27.5 30 Years
FIGURE 9.14
If the mortgage amount increases by $5000, the value of A increases by 5. M(15,
80) is approximately $720. The monthly payment increases by about $45. If the
length of the loan is increased by 10 years, we estimate M(20, 75) as about $530.
The monthly payment decreases by about $145.
The magnitude of the change (the absolute value of the change) is greater when
the length of the loan is increased from 15 years to 20 years than when the
amount borrowed is increased from $75,000 to $80,000. ●
Note that the vertical arrow in Figure 9.14 has length $5000, whereas the hori-
zontal arrow has length 10 years. We do not use the arrow lengths when determining
which change is greater, because the arrows have different units.
However, if the inputs of a function are comparable to each other, we can con-
sider the idea of steepest descent.
EXAMPLE 5 Estimating Direction of Steepest Descent
Farmland Again consider the elevation of the tract of Missouri farmland with the
contour graph shown in Figure 9.15.
a. If you are standing at (e, n) (0.4, 1.0), will you be going downhill or uphill if
you walk 0.4 mile north? south? east? west?
9.1 Multivariable Functions and Contour Graphs 561
n N
(miles) W E
S 796
1.4
797
Downhill
1.2
798 Location of
Uphill Approximately level point B
1.0
Western fence
B
800 Less than 799
Slightly
0.8 1 foot descent
uphill
801 800
0.6
802 3 foot descent
0.4
0.2
e
0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 (miles)
Southern fence
FIGURE 9.15
b. Starting at (1.0, 0.6), which direction results in the steeper descent: east 0.4 mile
or north 0.4 mile? Explain.
Solution
a. Refer to Figure 9.15. E (0.4, 1.0) is about 799.4 feet above sea level. Walking 0.4
mile north takes you to (0.4, 1.4). E(0.4, 1.4) is about 798 feet above sea level.
You have walked downhill.
There are two ways to Walking 0.4 mile south from (0.4, 1.0) takes you to (0.4, 0.6). E(0.4, 0.6) is
determine the direction of about 800 feet above sea level, so you have walked slightly uphill.
steepest descent or ascent. If
the scales on the horizontal and Walking 0.4 mile east takes you to (0.8, 1.0). E(0.8, 1.0) is slightly more than
vertical axes are the same, you
can look for the direction in
799 feet above sea level. Any difference based on the contour graph is too small
which the contours are closest to mention.
together. A more deliberate
approach is to move the same Walking west, you will walk uphill to just over 802 feet above sea level.
distance in each direction. You
can determine the direction of b. Before looking for the steeper descent, we note that both inputs are given in
steepest descent or ascent by miles. Because the input units are the same, we can compare steepness. Moving
comparing the change in output 0.4 mile east from (1.0, 0.6) to (1.4, 0.6) results in about a 3-foot descent from
in each direction. 800 feet to 797 feet. Moving north 0.4 mile results in less than a 1-foot descent.
We conclude that movement 0.4 mile to the east results in a steeper descent
than does movement 0.4 mile to the north. ●
We have seen in this section how to draw contour curves and how to estimate in-
puts and outputs at points on contour graphs and three-dimensional graphs. Being
able to visualize a multivariable function often helps us better understand the func-
tion and the relationships between the variables in the function.
562 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
4. Let N(p, s) be the number of skiers on a Saturday at
9.1 Concept Inventory a ski resort in Utah when p dollars is the price of an
all-day lift ticket and s is the number of inches of
• Multivariable function fresh snow that fell since the previous Saturday. For
• Input/output diagram parts a and b, write a sentence interpreting the
• Contour curve mathematical notation.
• Contour graph a. N(25, s)
• Three-dimensional graph b. N(p, 6)
• Sketching contour curves
c. Draw an input/output diagram for N.
• Finding equations of contour curves
• Estimating output and change in output from a 5. A multivariable function for the demand of a
contour graph commodity is D(i, p, r, c) million units, where i
thousand dollars is the average household income,
• Estimating steeper direction
p dollars is the price of the commodity, r dollars is
the price of a related commodity, and c people is the
9.1 Activities size of the consumer base. Draw an input/output
diagram for the function.
Getting Started
6. A multivariable function for the volume of a rectan-
1. Let P(c, s) be the profit in dollars from the sale of 1 gular packing container is V(l, w, h) lwh cubic
yard of fabric when c dollars is the production cost feet, where the length, l, the width, w, and the
per yard and s dollars is the selling price per yard. height, h, of the container are all measured in feet.
For parts a through c, write a sentence interpreting Draw an input/output diagram for the function.
the mathematical notation.
a. P(1.2, s) 7. A multivariable function for the principal on a
loan is A(m, r, n, t) dollars, where m dollars is paid
b. P(c, 4.5)
n times a year for t years at an interest rate of
c. P(1.2, 4.5) 3.0 100r%. Draw an input/output diagram for the
d. Draw an input/output diagram for P. function.
2. Let T(i, d) dollars be the amount of income tax owed 8. A multivariable function for the monthly cost of
by a household claiming d dependents with adjusted owning a vehicle is C(t, g, p, m, l) dollars, where t
income of i dollars. For parts a through c, write a sen- dollars is the cost of vehicle registration(s), g gallons
tence interpreting the mathematical notation. is the amount of gas used, p dollars is the insurance
a. T(36,000, d) premium, m dollars is the cost of maintenance, and
b. T(i, 4) l dollars is the loan payment. Draw an input/output
c. T(36,000, 4) 10,000 diagram for the function.
d. Draw an input/output diagram for T.
3. Let P(l, m) be the probability that a certain senator Applying Concepts
votes in favor of a tobacco-ban bill when the sena-
tor receives l letters supporting the bill and m mil- 9. Heat Index Your comfort in summer depends on
lion dollars is invested by the tobacco industry the combined effects of air temperature and humid-
lobbying against the bill. For parts a and b, write a ity. The apparent temperature table shows how hot
sentence interpreting the mathematical notation. it feels for a given air temperature and relative
humidity.
a. P(100,000, m)
The National Weather Service has established
b. P(l, 53) the following guidelines for the health threat on the
c. Draw an input/output diagram for P. basis of the apparent temperature:
9.1 Multivariable Functions and Contour Graphs 563
TABLE FOR ACTIVITY 9 Apparent temperature (°F)
Relative humidity (%)
40 45 50 55 60 65 70 75 80 85 90 95 100
110 135
108 130 137
106 124 130 137
104 119 124 130 137
102 114 119 124 130 137
Air temperature (°F)
100 109 113 118 123 129 136
98 105 108 113 117 122 128 134
96 101 104 107 111 116 121 126 132
94 97 100 103 106 110 114 119 124 129 135
92 94 96 98 101 104 108 112 116 121 126 131
90 91 92 94 97 99 102 106 109 113 117 122 126 131
88 88 89 91 93 95 97 100 103 106 109 113 117 121
86 85 86 88 89 91 93 95 97 99 102 105 108 111
84 83 84 85 86 87 89 90 92 94 96 98 100 102
82 81 82 83 83 84 85 86 87 88 90 91 93 94
80 80 80 81 81 82 82 83 83 84 85 85 86 87
(Source: National Oceanic and Atmospheric Administration.)
Apparent in Report of the Royal Society, IGY Antarctic Expe-
temperature Health risk dition to Halley Bay. (Temperatures have been
80–90°F Fatigue possible with prolonged converted to °F.)
exposure and/or physical activity a. Draw contour curves on the data from 10°F to
90–104°F Heat exhaustion or heat cramps 20°F in increments of 10°F.
possible b. Interpret these contours in the context of the
105–130°F Heat exhaustion or heat cramps mean monthly temperature.
likely c. Why are the temperatures in December and Jan-
Above 130°F Heat exhaustion or heat cramps uary warmer than the temperatures in June and
highly likely July?
a. Draw contour curves on the table for apparent d. In general, are temperatures higher at 30 mil-
temperatures of 90°F, 105°F, and 130°F. libars or at 850 millibars? Which air pressure rep-
b. Shade in the region on the table corresponding resents elevation closer to the Earth’s surface?
to apparent temperatures that are likely to cause 11. Daylight Hours The table shows the number of
heat exhaustion or heat cramps. hours of daylight for a given month at a given lati-
tude (measured in degrees away from the equator)
10. Temperature The mean monthly temperature
for the Northern and Southern Hemispheres.
table shows temperatures for Halley Bay, Antarc-
tica, at noon Greenwich Mean Time at heights a. How many hours of daylight will there be at a
where the air pressure was measured at the given location at latitude 25° north in March?
value during the particular month. These data b. How many hours of daylight will there be at a lo-
were collected during balloon ascents and reported cation at latitude 25° south in September?
564 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
TABLE FOR ACTIVITY 10 Mean monthly temperature at noon (°F)
Pressure Month
(millibars) Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
30 12.9 8.5 2.6 6.2 11.7 16.3 18.9 17.2 13.9 4.2 10.1 13.9
40 12.3 8.4 2.6 4.9 11.3 15.2 18.1 17.5 15.3 7.1 7.4 12.7
50 11.6 8.4 3.3 3.6 11.2 14.5 17.3 17.2 15.2 8.0 6.2 12.6
60 11.2 8.2 3.7 3.3 10.3 13.7 16.3 16.7 14.6 8.4 5.2 11.4
80 10.6 8.1 4.3 1.9 7.9 11.8 14.8 15.2 19.6 8.6 3.1 10.2
100 10.0 7.8 4.4 0.5 6.7 10.3 13.1 14.2 13.6 9.1 1.2 9.6
150 9.1 7.8 5.8 1.1 3.8 7.8 10.7 12.0 11.8 9.1 1.4 7.2
200 8.5 7.8 6.0 1.4 3.0 6.9 9.7 10.3 10.2 8.0 2.8 6.4
250 6.8 6.7 4.7 0.4 4.2 5.0 6.9 7.9 7.3 5.6 2.9 2.0
300 4.1 4.4 1.6 0.3 2.1 2.2 3.5 4.3 3.8 2.2 0.6 1.1
400 8.1 8.3 6.3 5.8 3.7 4.2 3.1 2.8 3.6 4.3 6.2 6.5
500 13.4 13.7 11.9 11.3 9.2 10.0 8.8 8.8 9.8 10.1 12.3 12.3
700 22.1 21.7 20.2 19.4 17.0 17.9 17.7 16.9 18.2 19.5 20.7 20.4
850 16.8 25.2 24.3 22.2 20.4 20.6 20.5 19.9 20.3 23.4 24.9 24.9
TABLE FOR ACTIVITY 11 Hours of daylight for a given month and latitude
Latitude Month
north Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
south Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun
0 12.12 12.12 12.12 12.12 12.12 12.12 12.12 12.10 12.11 12.12 12.12 12.12
5 11.87 11.96 12.08 12.22 12.35 12.41 12.38 12.28 12.16 12.02 11.90 11.83
10 11.61 11.81 12.06 12.35 12.57 12.70 12.64 12.45 12.17 11.91 11.67 11.55
15 11.34 11.66 12.04 12.47 12.82 13.00 12.92 12.62 12.22 11.81 11.44 11.25
20 11.07 11.50 12.01 12.60 13.07 13.32 13.22 12.81 12.26 11.70 11.20 10.94
25 10.78 11.33 11.97 12.74 13.34 13.66 13.53 13.02 12.31 11.58 10.94 10.62
30 10.45 11.14 11.97 12.88 13.65 14.05 13.88 13.23 12.35 11.47 10.67 10.26
35 10.09 10.95 11.95 13.06 13.98 14.47 14.27 13.47 12.42 11.33 10.36 9.86
40 9.68 10.71 11.91 13.25 14.36 14.96 14.71 13.76 12.48 11.18 10.00 9.39
45 9.19 10.45 11.87 13.48 14.82 15.55 15.25 14.09 12.55 11.01 9.60 8.85
50 8.61 10.13 11.84 13.78 15.38 16.29 15.91 14.48 12.66 10.80 9.07 8.17
55 7.83 9.73 11.79 14.10 16.14 17.28 16.78 14.99 12.76 10.55 8.45 7.28
60 6.79 9.21 11.74 14.62 17.10 18.70 18.01 15.67 12.92 10.22 7.60 6.04
(Source: A. A. Hanson, ed., Practical Handbook of Agricultural Science, Boca Raton, FL: CRC Press, 1990.)
9.1 Multivariable Functions and Contour Graphs 565
c. How many hours of daylight will there be in 14. Draw the contour curves f(L, K) 50, f(L, K)
January in the region where you attend school? 55, and f(L, K) 60 for the multivariable function
d. Draw contour curves representing 13, 14, 15, 16, f(L, K) 38.9L0.5K0.5 for 0.25 L 1.
17, and 18 hours of daylight. In a different color, 15. Draw the contour curves P(s, u) 40, P(s, u) 60,
draw contour curves representing 12, 11, 10, and and P(s, u) 100 for the multivariable function
9 hours of daylight. P(s, u) 38.6s 2su 13u 0.99u2 for 0
12. Children The Hutterite Brethren live on commu- u 5.
nal farms in parts of Canada and the United States. 16. Draw the contour curves F(w, x) 10,
They are a religious group that migrated from Europe F(w, x) 20, and F(w, x) 30 for the multivari-
to North America in the 1870s. Practically all mar- able function F(w, x) 3wx for 0 w 10
riages are within the group, and the data in the table
give the number of the 1236 married Dariusleut or 17. Capacity The function C(d, h) 0.0041d2h1.4
Lehrerleut Hutterite women born between 1879 and gives the capacity of a settled unopened silo (in tons
1936 who had s sons and d daughters. For example, of corn or grass silage) at 68% moisture content when
there were 28 women who had no sons and 1 daugh- the silo has inside diameter of d feet and the silage is
ter and 39 women with 4 sons and 4 daughters. h feet deep. Draw the 117-tons contour curve for d
between 12 and 30 feet.
a. How many women had 2 daughters and 2 sons?
5.833333A
b. How many women had 5 children? 18. Payments The function M(t, A) 1 0.932583t
gives the amount (in dollars) of the monthly pay-
c. Draw contour curves at 34, 25, 16, and 7 women.
ments necessary to pay off a loan of A thousand dol-
13. Draw the contour curves g(s, k) 200, g(s, k) lars at 7% annual interest over t years. Draw the
400, and g(s, k) 600 for the multivariable func- $400-monthly-payment contour curve for M for
tion g(s, k) 100k(1.09s) for 0 s 60. loans between 0 and 30 years.
TABLE FOR ACTIVITY 12 Number of women with s sons and d daughters
s, d, Daughters
Sons 0 1 2 3 4 5 6 7 8 9 10 11 12
0 28 8 2 3 2 2 0 0 0 0 1 0
1 21 29 21 11 5 7 6 4 1 0 0 0 1
2 11 27 22 21 21 14 10 15 5 3 1 0 0
3 6 16 27 20 35 29 18 12 10 2 2 0 0
4 9 10 20 21 39 28 30 24 10 2 5 1 0
5 3 7 22 22 40 17 18 23 16 7 2 0 0
6 2 9 15 16 27 26 26 17 10 4 1 0 0
7 1 4 7 27 19 20 16 7 2 2 0 0 0
8 0 3 12 14 12 7 10 5 3 1 0 0 0
9 0 2 4 8 11 4 5 2 0 0 0 0 0
10 0 1 1 2 3 2 2 1 1 0 0 0 0
11 0 0 1 2 0 1 0 0 0 0 0 0 0
12 0 0 1 1 0 0 0 0 0 0 0 0 0
(Source: P. Guttorp, Statistical Inference for Branching Processes, New York: Wiley, 1991, p. 194.)
566 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
19. Price Index In 1986, Cotterill developed a model models for total-stem volume for different varieties
for measuring the performance of supermarkets by of trees have been developed. In 1973, Brackett de-
considering their price index level. The price index veloped the following model for predicting the to-
level was an aggregate of 121 representative prices. tal-stem (inside bark) volume for Douglas fir trees
The lowest-price supermarket was assigned a price in British Columbia:
index of 100. According to the Cotterill study, the
V(d, h) 0.002198d1.739925 h1.133187 cubic feet
price index level of an independent supermarket
can be modeled by the function where d is the diameter of the tree at breast height
P(c, d, p, s) 109.168 0.730s 0.027s2 (4.5 feet above the ground), which is denoted by
dbh and measured in inches, and h is the height of
0.002d 0.041p 0.175c the tree in feet.
(Source: J. L. Clutter et al., Timber Management: A Quantita-
where the supermarket has s thousand square feet of tive Approach, New York: Wiley, 1983.)
sales space and is d miles from the warehouse, and the
consumer base grew by p thousand people in 10 years a. Find the stem volume of a Douglas fir with a dbh
and had a per capita income of c thousand dollars. of 1 foot and a height of 32 feet.
Assume that the distance from a supermarket to its b. Draw a contour curve representing the volume
distribution warehouse is 100 miles and that the con- you found in part a for diameters between 8 dbh
sumer base grew by 10,000 people in 10 years. and 18 dbh.
(Source: P. G. Helmberger and J. P. Chavas, The Economics of c. Explain how a change in dbh affects height if
Agricultural Prices, Upper Saddle River, NJ: Prentice-Hall, 1996.)
volume remains constant.
a. Write an equation for P(c, s) P(c, 100, 10, s).
21. BMI Your body-mass index (BMI) is a measure of
b. Draw the 110 contour curve for P(c, s) for super- how thin you are compared to your height. The body-
markets containing between 5000 and 25,000 mass index table shows BMI values for people be-
square feet of sales space. tween 5 and 6 feet tall who weigh between 90 and 200
20. Volume In the timber industry, being able to pre- pounds. For input units in inches and pounds, the
dict the volume of wood in a tree stem is important, body-mass index is calculated by the equation
especially when one is trying to determine the num- 0.4536w
B(h, w) points
ber of boards a tree will yield. Several multivariable 0.00064516h2
TABLE FOR ACTIVITY 21 Body-mass index
Weight Height (inches)
(pounds) 60 62 64 66 68 70 72
90 17.6 16.5 15.4 14.5 13.7 12.9 12.2
100 19.5 18.3 17.2 16.1 15.2 14.3 13.6
110 21.5 20.1 18.9 17.8 16.7 15.8 14.9
120 23.4 21.9 20.6 19.4 18.2 17.2 16.3
130 25.4 23.8 22.3 21.0 19.8 18.7 17.6
140 27.3 25.6 24.0 22.6 21.3 20.1 19.0
150 29.3 27.4 25.7 24.2 22.8 21.5 20.3
160 31.2 29.3 27.5 25.8 24.3 23.0 21.7
170 33.2 31.1 29.2 27.4 25.8 24.4 23.1
180 35.2 32.9 30.9 29.1 27.4 25.8 24.4
190 37.1 34.8 32.6 30.7 28.9 27.3 25.8
200 39.1 36.6 34.3 32.3 30.4 28.7 27.1
(Source: Based on information obtained from www.shapeup.org (accessed on 6/27/00).)
9.1 Multivariable Functions and Contour Graphs 567
where h is the person’s height in inches and w is the where p is the yearly average airfare (in dollars) be-
person’s weight in pounds. tween New York City and London, adjusted for
a. Estimate your BMI from the table. Calculate your inflation, and g is the U.S. yearly gross national
BMI using the equation. Use the chart that follows product (in billions of dollars), adjusted for infla-
to assess your health risk on the basis of your BMI. tion. The demand table shows selected values of this
function.
BMI Health risk based solely on BMI
a. On the table, sketch contour curves for demand
1924 Minimal values of 3000, 6000, 9000, 12,000, 15,000,
2526 Low 18,000, 21,000, and 24,000 thousand passen-
2729 Moderate gers.
3034 High b. Find a general equation for the contour curves of
3539 Very high the function D.
above 40 Extremely high
c. Use the equation in part b to sketch a contour
b. On the table, sketch contour curves for index graph for demand values of 3000, 6000, 9000,
values of 15, 20, 25, 30, and 35 points. 12,000, 15,000, 18,000, 21,000, and 24,000 thou-
sand passengers. Compare this graph with the
c. Find a general equation for the contour curves of
one sketched on the table.
the function B.
d. Use the equation in part c to sketch a contour 23. Cooking The percentage cooking loss in sausage
graph for index values of 15, 20, 25, 30, and 35 can be modeled by
points. Compare this graph with the one P(w, s) 10.65 1.13w 1.04s 5.83ws percent
sketched on the table.
when w and s represent the proportions of whey
22. Air Travel The demand for air travel between the protein and skim milk powder, respectively, used in
United States and Europe between 1965 and 1978 the sausage.
can be modeled by the function (Source: M. R. Ellekjaer, T. Naes, and P. Baardseth, “Milk
Proteins Affect Yield and Sensory Quality of Cooked
15.44g1.905
D(g, p) thousand passengers Sausages,” Journal of Food Science, vol. 61, no. 3 (1996), pp.
p1.247 660–666.)
TABLE FOR ACTIVITY 22 Demand (thousands of passengers) for air travel between
the United States and Europe from 1965 to 1978
Average yearly Yearly U.S. gross national product (billions of dollars)
airfare (dollars) 1400 1700 2000 2300 2600 2900
400 8655 12,529 17,075 22,284 28,146 34,655
500 6553 9485 12,927 16,871 21,309 26,237
600 5220 7556 10,298 13,440 16,976 20,902
700 4307 6235 8497 11,090 14,007 17,246
800 3647 5279 7194 9389 11,859 14,601
900 3148 4558 6211 8106 10,239 12,606
1000 2761 3996 5447 7108 8978 11,054
1100 2451 3549 4836 6312 7972 9816
1200 2199 3184 4339 5663 7152 8806
1300 1990 2881 3927 5125 6473 7970
1400 1815 2627 3580 4672 5902 7266
(Source: Based on information in J. M. Cigliano, “Price and Income Elasticities for Airline Travel:
The North Atlantic Market,” Business Economics, September 1980, pp. 17–21.)
568 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
a. Find a general formula for contour curves for P. a. Use the contour graph to approximate the
b. Sketch a contour graph for percentages of 10.6, production levels that will maximize revenue.
10.7, 10.8, 10.9, and 11.0. Note that because w Why did you pick the production levels you
and s are proportions, they should be graphed chose?
from 0 to 1. b. Use the contour graph and the three-dimen-
sional graph to approximate the maximum
24. Wheat Crop The carrying capacity of a particular revenue. Explain how you arrived at this conclu-
farm system is defined as the number of animals or
sion.
people that can be supported by the crop produc-
tion from a given land area. The carrying capacity of 26. Sunflowers A process to extract pectin and pigment
a wheat crop can be modeled by from sunflower heads involves washing the sunflower
heads in heated water. The three-dimensional graph
11.56P
K(P, D) people per hectare and the associated contour graph on page 569, show
D the percentage of pigment that can be removed from a
sunflower head by washing it for t minutes in r milli-
where P is the number of kilograms of wheat pro- liters of water per gram of sunflower heads when the
duced per hectare per year and D is the yearly energy water temperature is 75°C.
requirement for one person, measured in mega- (Source: X. Q. Shi et al., “Optimizing Water Washing Process
joules per person. for Sunflower Heads Before Pectin Extraction,” Journal of Food
(Source: R. S. Loomis and D. J. Connor, Crop Ecology: Produc- Science, vol. 61, no. 3 (1996), pp. 608–612.)
tivity and Management in Agricultural Systems, Cambridge,
a. Find and interpret the input and output values for
England: Cambridge University Press, 1992.)
point A.
a. Find a general formula for contour curves for K.
b. Estimate the percentage of pigment removed
b. Sketch a contour graph for carrying capacities of from 15 grams of sunflower heads if the sunflower
13, 15, 17, and 19 people per hectare. heads are washed for 10 minutes in 37.5 milliliters
25. Revenue Refer to the contour graph and the of 75°C water.
three-dimensional graph representing the revenue c. Estimate how much water would be necessary to
R(l, r), in thousands of guilders, when l thousand remove 36% of the pigment from 9 grams of sun-
tons of lower-fat (40%) cheese and r thousand tons flower heads if they were washed for 15 minutes in
of regular cheese are produced. 75°C water. Discuss why there are two answers.
r
Regular R(l, r)
cheese Revenue
(thousand tons) (million
guilders)
80
200
60 100 80
60
40 r
0 40
34 10
Regular cheese
184
20 l 20 (thousand tons)
154 20
64 Lower-fat l
124 94 30 0
0 cheese Lower-fat cheese
0 5 10 15 20 25 30 (thousand (thousand tons)
tons)
Contours of R(l, r) 3-D of R(l, r)
Figures for Activity 25
9.1 Multivariable Functions and Contour Graphs 569
Pigment t
(percent) Minutes
40
30 36
38 37
25
38
36
20
34 39
15
A t
32
Minutes 10
30 30
25 5
20
28 15 r
10 0
20 25 30 35 40 Milliliters
20 25 30 35 40
r
Milliliters
3-D of pigment function Contours of pigment function
Figures for Activity 26
H
Percent
D relative
Days to humidity
A
develop
80
14
18 75 13
12 10
70 11
14 B
65
10 60
80
55 12
6 75 P
8 70 50 Hours of
10 65 H 8 10 12 14 16
12 Percent light
60
P 14 55 relative
16
Hours of light humidity
3-D of parasitoid development Contours of parasitoid development boll weevils
Figures for Activity 27
(Source: Based on information in J. A. Morales-Ramos, S. M. Greenberg, and E. G. King, “Selection of Optimal Physical Conditions for Mass
Propagation of Catolaccus grandis,” Environmental Entomology, vol. 25, no. 1 (February 1996), pp. 165–173.)
27. Parasites Boll weevils have long presented a (in days) of C. grandis as a function of the relative
threat to cotton crops in the southern United States. humidity and the number of hours of light.
Research has been done to determine the optimal
a. Estimate and interpret the input and output val-
conditions for reproduction of Catolaccus grandis, a
ues at point B.
parasitoid that attaches to boll weevils and kills
them. The three-dimensional graph and an associ- b. Estimate and interpret the input and output val-
ated contour graph show the developmental time ues at point A.
570 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
Percentage
B moisture
21
Adhesive- A
ness 20
7
19
6
5 18
4
21
3 20 17
19 7
2
18 ure
8 17 o ist 16
10
12 16 em 6.5
14 15 tag 15
6
Days 16
18 14 c en
r 5.5
Pe 14
Days
7 8 10 12 14 16 18 19
3-D of honey adhesiveness Contours of honey adhesiveness
Figures for Activity 28
28. Honey The ability of honey to attach to a surface honey given the percentage of moisture and the
on which it is spread is known as its adhesiveness. number of days the honey is allowed to set when
The adhesiveness of honey relies on several factors: 40.9% of the honey is glucose and maltose and
the percentage of glucose and maltose (two sug- 12.5% of the honey was crystallized before setting
ars), the percentage of moisture, the percentage began.
of crystallization in the honey before it begins to (Source: J. M. Shinn and S. L. Wang, “Textural Analysis of
set, and the number of days the honey is allowed to Crystallized Honey Using Response Surface Methodology,”
set at 12°C. The three-dimensional graph and con- Canadian Institute of Food Science and Technology Journal, vol.
23, no. 4–5 (1990), pp. 178–182.)
tour graph show a measure of the adhesiveness of
y f (x, y)
3 5
10 35
2.5 15 30
30
20 25
2
25
20
1.5
15
1 20
25 10
0.5 15 5
10
0 2.5 3
0 x 1 1.5 2
0 1 2 3 4 0 1 2 3 4 0.5 y
x
Contours of f(x, y) 3-D of f(x, y)
Figures for Activity 29
(Source: M. R. Ellekjaer, T. Naes, and P. Baardseth, “Milk Proteins Affect Yield and Sensory Quality of Cooked
Sausages,” Journal of Food Science, vol. 61, no. 3 (1996), pp. 660–666.)
9.1 Multivariable Functions and Contour Graphs 571
w
20 21
15.5
10
15
4.5
10 K(s, w)
60
10
40 20
5 15.5 20
21 15
26.5
32 0 10 w
5
s 10 5
0
s 15
0
0 5 10 15 20
Contours of K(s, w) 3-D of K(s, w)
Figures for Activity 30
a. Estimate and interpret the input and output y
values at point A. 0.6
b. Estimate and interpret the input and output 0.05
0.4 .1 05
values at point B. 0.10 − 0 .05 0. .10
0.15 −0 0 15
0
0.
29. For the contour graph of f(x, y) and its associ- 0.2 0.2 0.2
ated three-dimensional graph, answer the ques-
tions by analyzing the graphs. Explain in sentences 0
x
why you answered as you did. −1 −0.8 −0.6 −0.4 −0.2 0.2 0.4 0.6 0.8 1
a. At (1.5, 2), does f(x, y) increase when y increases −0.2
or when y decreases?
−0.4
b. At (2.5, 2.5), will f(x, y) decrease more quickly as
x decreases or as y decreases?
−0.6
c. Will the change in f(x, y) be greater when (2, 2)
Contours of g(x, y)
shifts to (1, 2.5) or when (1, 0) shifts to (4, 1)?
Figure for Activity 31
30. For each of the following questions, refer to
the graphs of K(s, w) and answer the questions by
analyzing the graphs. Explain in sentences why you 31. Answer each of the following questions by analyzing
answered as you did. the contour graph of g(x, y) shown in the figure.
a. Describe in as much detail as possible the three-
a. At (10, 10), is K(s, w) increasing when w
dimensional graph that this contour graph
increases or when s decreases?
represents.
b. At (15, 2.5), will K(s, w) be decreasing more b. Is the descent greater when the position changes
rapidly when w increases or when s decreases? from (0.7, 0.1) to (0.4, 0.4) or when it changes
c. Will the change in K(s, w) be greater when the from (0.7, 0.1) to (0, 0.3)?
position is changed from (5, 5) to (6, 3) or when c. At (0, 0.1), does the function output increase as
it changes from (5, 5) to (6, 7)? x increases or as y increases?
572 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
d. Locate a point with output approximately 0.15 Surface Temperature, °F
unit greater than the output at (0.2, 0.3). 14
14
32. Answer each of the following questions by analyzing
the contour graph of h(x, y) shown in the figure. 14
23
32
y
41
10 14
50
2.0 59
1.5
1.0 5
23
0.5 00 UTC 15 Feb 90
77
32 41 50 59 68 68
x
–3 –2 –1
0 Figure for Activity 33
5
– 0. –5
.0
–1 5
– 1. .0 a. Label the regions most likely to have received
–2 snow, ice pellets, freezing rain, and rain.
–10 b. How much of a change in temperature was expe-
Contours of h(x, y) rienced by a traveler who started in southern
Kentucky and traveled to northern Missouri?
Figure for Activity 32
34. Turbine Power Wind turbines can be used to har-
ness the power of the wind and generate electricity.
a. Consider any point (x, y) on the graph. What An optimally placed wind machine is located where
happens to the function output when x remains wind power will generate the most electricity. The
constant and y increases? figure shows a contour map of average available wind
b. Consider the point (1, 5). When x is decreased power (measured in watts per square meter) over the
by 1/2, by what amount must y change in order contiguous United States.
(Source: Figure from D. M. Gates, Energy and Ecology, Sunder-
for the point to remain on the same contour
land, MA: Sinauer Associates, 1985. Copyright © 1985.
curve? Reprinted with permission of Sinauer Associates, Inc.)
c. Consider the point (2, 2). When y is in- a. Estimate four optimal wind machine placement
creased by 1.5, by what amount must x change in sites in (or near the shore of) the United States.
order for the output to remain constant?
b. What is the difference in average available
33. Temperature In meteorology, temperature con- wind power between mid-Texas and western
tours are known as isotherms. The figure shows sur- Nebraska?
face temperature isotherms14 (in °F) on February
15, 1990. The shaded regions indicate the areas of Discussing Concepts
precipitation.
(Source: Figure from R. R. Czys and R. W. Scott, “Forecasting 35. Explain how to locate a relative maximum on a con-
Techniques: A Physically Based, Nondimensional Parameter tour graph and on a table of data.
for Discriminating Between Locations of Freezing Rain and
Ice Pellets,” Weather and Forecasting, vol. II (1996) p. 595. 36. Discuss what is meant by the phrase path of steepest
Used by permission of the American Meteorological Society.) descent. Explain how you can sketch a path of steep-
est descent on a contour graph.
9.2 Cross-Sectional Models and Rates of Change 573
300 300
150 200 150
200 150 100 50 100
200
400
600
800
400 100 <100
300 300
400
300
>50
<50 150
50
150
200
100
150
Contour map of wind power
Figure for Activity 34
9.2 Cross-Sectional Models and Rates of Change
In cases where the output is based on several different input variables, it is possi-
ble to model data using a technique called multiple regression. We will not ex-
plore modeling multivariable data in this text, but we will discuss cross-sectional
modeling because it is a simple extension of the data-modeling techniques from
Chapter 1.
Cross-Sectional Models from Data
The elevation E(e, n) feet above sea level on a tract of farmland measured e miles
east of the western fence and n miles north of the southern fence is given in Table
9.3. We can model the elevation along a line parallel to the western fence at a
distance of e 0.8 mile east of that fence by looking at only that one column of
elevations.
We model part of the data in a table with two input variables by holding one in-
put variable constant and finding an equation in terms of the other input variable.
The data that we are using to find the cross-sectional model is a portion of the
larger table. We are interested in the elevation of points 0.8 mile east of the western
fence, so we take our elevation data from that column for every available point north
of the southern fence.
574 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
TABLE 9.3
e→ 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
1.5 800.3 799.1 798.3 797.8 797.5 797.4 797.4 797.5 797.6 797.7 797.6 797.3 796.8 796.0 794.8 793.2
1.4 800.8 799.6 798.8 798.3 798.0 797.9 797.9 798.0 798.1 798.1 798.1 797.8 797.3 796.5 795.3 793.7
1.3 801.2 800.0 799.2 798.7 798.4 798.3 798.3 798.4 798.5 798.6 798.5 798.2 797.7 796.9 795.7 794.1
1.2 801.6 800.4 799.6 799.1 798.8 798.7 798.7 798.8 798.9 798.9 798.9 798.6 798.1 797.3 796.1 794.5
1.1 801.9 800.7 799.9 799.4 799.1 799.0 799.0 799.1 799.2 799.3 799.2 798.9 798.4 797.6 796.4 794.8
1.0 802.2 801.0 800.2 799.7 799.4 799.3 799.3 799.4 799.5 799.5 799.5 799.2 798.7 797.9 796.7 795.1
0.9 802.4 801.2 800.4 799.9 799.6 799.5 799.5 799.6 799.7 799.8 799.7 799.4 798.9 798.1 796.9 795.3
0.8 802.6 801.4 800.6 800.1 799.8 799.7 799.7 799.8 799.9 799.9 799.9 799.6 799.1 798.3 797.1 795.5
0.7 802.7 801.5 800.7 800.2 799.9 799.8 799.8 799.9 800.0 800.1 800.0 799.7 799.2 798.4 797.2 795.6
0.6 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.1 800.1 799.8 799.3 798.5 797.3 795.7
0.5 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.2 800.1 799.8 799.3 798.5 797.3 795.7
0.4 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.1 800.1 799.8 799.3 798.5 797.3 795.7
0.3 802.7 801.5 800.7 800.2 799.9 799.8 799.8 799.9 800.0 800.1 800.0 799.7 799.2 798.4 797.2 795.6
0.2 802.6 801.4 800.6 800.1 799.8 799.7 799.7 799.8 799.9 799.9 799.9 799.6 799.1 798.3 797.1 795.5
0.1 802.4 801.2 800.4 799.9 799.6 799.5 799.5 799.6 799.7 799.8 799.7 799.4 798.9 798.1 796.9 795.3
0 802.2 801.0 800.2 799.7 799.4 799.3 799.3 799.4 799.5 799.5 799.5 799.2 798.7 797.9 796.7 795.1
n↑
n N
(miles) W E
S 796
1.4
e = 0.8 mile
797
1.2
798 N
1.0
Western fence
Western W E
fence
0.8 800 799
E(e, n) S
801 800 (feet above
0.6
802
sea level)
0.4 803
1.5
0.2
793 1.0
e 0 n
0 0.5 (miles)
(miles) Southern 0.5
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
fence e 1.0
Southern fence (miles) 1.5
(a) Contour graph with line e 0.8 (b) Three-dimensional graph with plane e 0.8
FIGURE 9.16
9.2 Cross-Sectional Models and Rates of Change 575
TABLE 9.4 We find that a model for the data set shown in Table 9.4 is
n E (0.8,n) E(0.8,n) 2.5n2 2.497n 799.490 feet above sea level
0 799.5 where the distance east of the western fence is 0.8 mile and the distance north of the
0.1 799.7 southern fence is n miles, 0 n 1.5.
0.2 799.9
A model such as E(0.8, n) is a cross-sectional model. For a function with two in-
put variables, a cross section is the curve that results when the three-dimensional
0.3 800.0 graph of the function is intersected with a plane. A cross section of a function has one
0.4 800.1 less dimension (variable) than the function itself.
0.5 800.1 The equality e 0.8 mile east of the western fence is depicted on the contour
0.6 800.1
graph as a line (Figure 9.16a) and on the three-dimensional graph as a cross-sectional
plane (Figure 9.16b). The curve resulting from the intersection of the plane e 0.8
0.7 800.0 and the elevation function E is called a cross section of E and has output denoted by
0.8 799.9 E(0.8, n).
0.9 799.7 This model is only a partial description of all of the data. It tells us nothing about
1.0 799.5
the elevations when the distance east of the western fence is something other than 0.8
mile. For this reason, we call it a cross-sectional model. Figure 9.17 shows a graph of
1.1 799.2 this cross section of E with output E(0.8,n).
1.2 798.9
1.3 798.5 E(0.8, n)
Elevation
1.4 798.1
(feet)
1.5 797.6
802
800
798
796
n
0.2 0.4 0.6 0.8 1 1.2 1.4 (miles)
FIGURE 9.17
We can also find a cross-sectional model for the elevation when the distance
north of the southern fence is held constant by modeling a specific row of data from
the table. It is also possible to find cross-sectional models when we know an equation
for a multivariable function.
Rates of Change of Cross-Sectional Models
As we have seen with the elevation example, we can use cross-sectional functions to
help us understand the behavior of a multivariable function. And we can use a cross-
sectional model to analyze the rate of change of the output with respect to one of the
input variables.
When we are given a table of data for a function with two input variables, we
approximate rates of change along one row or column by finding rates of change of
cross sections. Consider again the elevation of a tract of farmland. We use cross-
sectional models to help answer questions about the elevation and the steepness (that
is, the magnitude of the rate of change of elevation) of the tract. When we are 0.8 mile
576 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
E(0.8, n) east of the western fence and 0.6 mile north of the southern fence, how steep is the
Elevation land? In other words, how rapidly is elevation changing, or what is the rate of change
(feet) of elevation? To answer this question, we must first know in which direction to look:
from south to north or from west to east.
802
Let us first look to the north. The question is “At e 0.8 mile and n 0.6 mile,
800 what is the rate of change of elevation with respect to the distance north of the southern
fence?” This cross-sectional rate of change is the slope of the tangent line that runs in
798 the north-south direction at (0.8, 0.6). This tangent line is depicted in Figure 9.18 on
the graph of the E(0.8, n) cross section. Note that because movement is restricted to
796
one direction, this rate of change gives us only a partial answer to the overall question
n of the steepness at this point. In Section 9.3, we will consider in detail the notion of a
0.2 0.6 1 1.4 (miles) partial rate of change.
Line tangent to E(0.8, n) at The cross-sectional model for e 0.8 was found to be
n 0.6 E(0.8, n) 2.5n2 2.497n 799.490 feet above sea level
FIGURE 9.18
where n is the number of miles north of the southern fence. We find the derivative of
the cross-sectional function with output E(0.8, n) with respect to n (the distance from
the southern fence) to be
dE(0.8, n)
5n 2.497 feet per mile
dn
dE(0.8, n)
Evaluating dn at n 0.6 gives us approximately 0.5 foot per mile. This
means that at the point located 0.8 mile east of the western fence and 0.6 mile north
of the southern fence, the elevation is decreasing by 0.5 foot per mile as you go toward
the north. It also means that the elevation is increasing by 0.5 foot per mile going
toward the south.
Keep in mind that this rate of change is restricted to the north-south direction
at a particular point. It does not tell us what is happening toward the east or west. In
order to find the steepness in the east-west direction, we must look at the derivative
of a cross-sectional function with respect to distance e miles east of the western
fence.
EXAMPLE 1 Finding a Rate of Change of a Cross Section from a Table
Farmland Refer to Table 9.3 on page 574. Estimate and interpret the rate of change
of elevation with respect to distance from the western fence 0.8 mile east of the west-
ern fence and 0.6 mile north of the southern fence.
Solution
We need the cross-sectional model that has distance e miles east of the western fence as
input with the distance 0.6 mile north of the southern fence held constant. Thus, we
consider the row in Table 9.3 for n 0.6 mile to obtain the data shown in Table 9.5.
A model for E(e, 0.6), the elevation in feet above sea level of the land measured at
n 0.6 mile north of the southern fence, is
E(e, 0.6) 10.124e3 21.347e2 13.972e 802.809 feet above sea level
9.2 Cross-Sectional Models and Rates of Change 577
TABLE 9.5
e 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
E (e, 0.6) 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0
e 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
E (e, 0.6) 800.1 800.1 800.1 799.8 799.3 798.5 797.3 795.7
E(e, 0.6) e miles east of the western fence and 0.6 mile north of the southern fence,
Elevation 0 e 1.5. The rate of change of E(e, 0.6) with respect to distance e east of the
(feet) western fence is
802 dE(e, 0.6)
30.372e 2 42.694e 13.972 feet per mile
de
800
dE(e, 0.6)
798
Evaluating at e 0.8 gives us approximately 0.75 foot per mile. This
de
means that at the point located 0.8 mile east of the western fence and 0.6 mile north
796 of the southern fence, the elevation is increasing by 0.75 foot per mile as you go to-
e ward the east. It also means that elevation is decreasing by 0.75 foot per mile going to-
0.2 0.6 1 1.4
(miles) ward the west.
Line tangent to E(e, 0.6) at Figure 9.19 shows the line tangent to a graph of E(e, 0.6) at e 0.8. ●
e 0.8
FIGURE 9.19 Keep in mind that cross-sectional rates of change tell us only part of the story
about the rate of change in output, because they are restricted to change in one input
variable. They do not give us any information about what is happening with respect
to other input variables.
certain air temperatures and dew points. The dew
9.2 Concept Inventory point is the temperature to which the air needs to
be cooled in order to achieve a relative humidity
• Cross-sectional models of 100%. The higher the dew point, the greater
• Derivatives of cross-sectional models the amount of moisture in the air, and the muggier
it feels.
a. In order to model the apparent temperature as
a function of the dew point when the air tem-
perature is 95°F, which variable must be held
9.2 Activities constant?
b. Is the cross section in part a represented by a row
Getting Started
or a column of the table?
1. Apparent Temperature The table of apparent c. Find a cross-sectional model for the apparent
temperatures shows how comfortable it feels (the temperature as a function of the dew point when
apparent temperature in degrees Fahrenheit) for the air temperature is 95°F.
578 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
TABLE FOR ACTIVITIES 1 AND 2 Apparent temperature (°F)
Dew point (°F)
50 55 60 65 70 75 80 85
65 62.7 63.8 65.0 66.6
70 67.8 68.7 69.8 71.1 72.6
Air temperature (°F)
75 73.1 73.9 74.8 75.9 79.2 80.7
80 79.8 80.6 81.6 82.8 84.4 86.9 90.9
85 83.5 84.7 86.1 88.0 90.5 94.0 99.0 106.6
90 87.9 89.4 91.2 93.6 96.9 101.2 107.2 115.6
95 92.9 94.5 96.7 99.6 103.4 108.4 115.2 124.3
100 98.1 99.9 102.4 105.6 109.8 115.3 122.7 132.3
105 103.4 105.4 108.1 111.6 116.1 122.0 129.7 139.7
110 108.7 110.9 113.8 117.5 122.3 128.4 136.3 146.5
(Source: www.nws.noaa.gov, accessed 12/28/02.)
2. Apparent Temperature Consider the table of c. Find a cross-sectional model for the apparent
apparent temperatures, which shows apparent temperature as a function of the air temperature
temperature (how hot it feels) given air temperature when the dew point is 70°F.
and dew point. 3. Cloud Cover The table showing the frequency of
a. In order to model the apparent temperature as a cloud cover over Minneapolis in January, given the
function of the air temperature when the dew fraction of the sky covered by clouds and the hour of
point is 70°F, which variable must be held the day, contains output that may be interpreted as
constant? the percentage of time that a certain situation occurs.
8
b. Is the cross section in part a represented by a row a. What percentage of time is at least 10 of the sky
or a column of the table? covered by clouds at noon?
TABLE FOR ACTIVITIES 3 AND 4 Frequency of cloud cover over Minneapolis in January
Fraction of Hour of the day
sky covered Midnight 3:00 A.M. 6:00 A.M. 9:00 A.M. Noon 3:00 P.M. 6:00 P.M. 9:00 P.M.
overcast 0.45 0.48 0.47 0.45 0.41 0.39 0.41 0.38
9/10 0.48 0.52 0.49 0.57 0.55 0.53 0.46 0.46
8/10 0.50 0.55 0.52 0.62 0.60 0.57 0.49 0.43
7/10 0.52 0.56 0.54 0.64 0.63 0.59 0.52 0.47
6/10 0.54 0.59 0.57 0.66 0.65 0.61 0.54 0.49
5/10 0.55 0.59 0.59 0.69 0.67 0.64 0.55 0.51
4/10 0.56 0.59 0.62 0.72 0.68 0.66 0.58 0.56
3/10 0.58 0.62 0.64 0.74 0.71 0.69 0.60 0.60
2/10 0.60 0.67 0.66 0.78 0.74 0.71 0.62 0.62
1/10 0.66 0.71 0.72 0.89 0.86 0.84 0.79 0.66
(Source: I. I. Gringorten, “Modelling Conditional Probability,” Journal of Applied Meteorology, vol. 10, no. 4 (August 1971), pp. 646–657.)
9.2 Cross-Sectional Models and Rates of Change 579
b. In order to model the frequency of cloud cover at b. Is it possible to find a cross-sectional model for
9 A.M., which variable must be held constant? monthly payments for a loan at 9%? If so, would
What will be the input variable of the cross- this cross section be represented by a row or a
sectional model? column of the table?
c. Is the cross section in part b represented by a row c. Find an appropriate cross-sectional model, and
or a column of the table? use it to estimate the monthly payments for a 52-
d. Find a cross-sectional model for the frequency of month loan at 9%.
cloud cover at 9 A.M.
6. Payments The table of monthly payments was
4. Cloud Cover The table showing the frequency taken from a magazine advertisement and shows
of cloud cover over Minneapolis in January, given the monthly payments on a $1000 loan at different in-
fraction of the sky covered by clouds and the hour of terest rates over different loan periods.
the day, contains output that may be interpreted as a. Is it possible to find a cross-sectional model for
the percentage of time that a certain situation occurs. monthly payments for a 42-month loan? If so,
a. What percentage of time is at least half of the sky would this cross section be represented by a row
covered by clouds at 3 P.M.? or a column of the table?
b. In order to model the frequency of an overcast b. Is it possible to find a cross-sectional model for
sky given the time of day, which variable must be monthly payments for a loan at 10.5%? If so,
held constant? would this cross section be represented by a row
c. Is the cross section in part b represented by a row or a column of the table?
or a column of the table? c. Find an appropriate cross-sectional model, and
d. Find a cubic cross-sectional model for the use it to estimate the monthly payments for a 42-
frequency of an overcast sky given the time of day. month loan at 10.5%.
5. Payments The table of monthly payments was 7. Peaches The table for per capita peach consump-
taken from a magazine advertisement and shows tion shows selected values of the average yearly con-
monthly payments on a $1000 loan at different in- sumption of peaches per person, based on the price
terest rates over different loan periods. of peaches and the yearly income of the person’s
a. Is it possible to find a cross-sectional model for family.
monthly payments for a 52-month loan? If so, a. Find a cross-sectional model for a yearly income
would this cross section be represented by a row of $40,000.
or a column of the table?
TABLE FOR ACTIVITIES 5 AND 6 Monthly payments
per $1000 of loan amount TABLE FOR ACTIVITIES 7 AND 8 Per capita
Rate Months consumption of peaches (pounds per person per year)
% 24 36 42 48 60 Yearly Price per pound above $1.50
income
5 43.87 29.97 26.00 23.03 18.87 (tens of
6 44.32 30.42 26.46 23.49 19.33 thousands
7 44.77 30.88 26.91 23.95 19.80 of dollars) 0 0.10 0.20 0.30 0.40 0.50
8 45.23 31.34 27.38 24.41 20.28 1 5.0 4.9 4.8 4.7 4.7 4.6
9 45.68 31.80 27.84 24.89 20.76 2 6.4 6.3 6.2 6.1 6.1 6.0
10 46.14 32.27 28.32 25.36 21.25 3 7.2 7.1 7.0 6.9 6.9 6.8
11 46.61 32.74 28.79 25.85 21.74 4 7.8 7.7 7.6 7.5 7.4 7.4
12 47.07 33.21 29.28 26.33 22.24 5 8.2 8.1 8.0 8.0 7.9 7.8
6 8.6 8.5 8.4 8.3 8.3 8.2
(Source: Automobile, vol. 9, no. 4 (July 1994), p. 72.)
580 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
b. Use the model to estimate the consumption for a. In order to estimate sales of Coke products when
someone in a family with a yearly income of the price of Coke products is $1.40 and the price
$40,000 when the price of peaches is $1.55 per of Pepsi products is $0.50, which variable should
pound. be held constant?
b. Will the cross-section from part a be represented
8. Peaches The table for per capita peach consump-
by a row or a column of the table?
tion shows selected values of the average yearly con-
sumption of peaches per person, based on the price c. Find an appropriate cross-sectional model, and
of peaches and the yearly income of the person’s use it to estimate sales of Coke products when
family. the price of Coke products is $1.40 and the price
of Pepsi products is $0.50.
a. Find a cross-sectional model for a price of $1.80
per pound. 11. Population The table gives the population (in
b. Use the model to estimate consumption for a millions) of the United States for residents from 15
person in a family with a yearly income of to 50 years of age between 1990 and 2000, with pro-
$35,000 when the price is $1.80 per pound. jections through 2020.
a. How many people of age 20 years were living in
9. Sales Two vending machines sit side by side in a
the United States in 2000?
college dorm. One machine sells Coke products,
and the other sells Pepsi products. Daily sales of b. On the table, sketch the contour curve for a pop-
Coke products, based on the prices of the products ulation of 4 million people passing through (50,
in the two machines, are as shown in the table. 2020).
a. In order to estimate sales of Coke products when c. How rapidly will the population be increasing
the price of Coke products is $1.00 and the price with respect to age for residents who will be 35
of Pepsi products is $0.90, which variable should years old in 2005? (Hint: Begin by finding an ap-
be held constant? propriate cross-sectional model.)
b. Will the cross section from part a be represented d. How rapidly will the population be increasing
by a row or column of the table? with respect to the year for residents who will be
50 years old in 2020?
c. Find an appropriate cross-sectional model, and
use it to estimate sales of Coke products when 12. Population The table gives the population (in
the price of Coke products is $1.00 and the price millions) of the United States for residents from 15
of Pepsi products is $0.90.
10. Sales Two vending machines sit side by side in a
college dorm. One machine sells Coke products,
and the other sells Pepsi products. Daily sales of TABLE FOR ACTIVITIES 11 AND 12
Coke products, based on the prices of the products U.S. Population (in millions) by age
in the two machines, are as shown in the table.
Year
TABLE FOR ACTIVITIES 9 AND 10 1990 1995 2000 2005 2010 2015 2020
Daily sales (number of cans) of Coke products 15 3.34 3.65 3.87 4.24 4.31 4.22 4.26
from a vending machine
20 4.04 3.51 3.88 4.10 4.48 4.55 4.45
Cost of Coke products
25 4.06 3.79 3.39 3.73 3.94 4.30 4.36
$0.50 $0.75 $1.00 $1.25 $1.50
Age
Cost of Pepsi products
30 4.50 4.38 3.92 3.52 3.86 4.08 4.44
$0.50 157 143 123 98 65
35 4.27 4.59 4.47 4.00 3.61 3.95 4.17
$0.75 206 192 172 146 114
40 3.80 4.28 4.65 4.54 4.07 3.68 4.02
$1.00 255 241 221 195 163
45 2.90 3.70 4.21 4.57 4.46 4.01 3.62
$1.25 304 290 270 244 211
50 2.43 2.93 3.69 4.19 4.55 4.44 4.00
$1.50 353 339 319 293 260
(Source: U.S. Bureau of the Census.)
9.2 Cross-Sectional Models and Rates of Change 581
to 50 years of age between 1990 and 2000, with pro- a. What is the average daily weight gain/loss for a
jections through 2020. 91-kg pig when the temperature is 37.8°C?
a. How many people of age 20 years will be living in b. Find and interpret a model for W(68, t).
the United States in 2010?
c. Look at a graph of the model in part b, and
b. How rapidly will the population be increasing describe its behavior. What does the behavior
with respect to the year for residents who will be of the graph tell you about the weight gain of
20 years old in 2010? a pig?
c. Consider the specific group of people who are
age 20 in 2010. Does the answer to part b describe 14. Weight The table shows average daily weight
the change in the size of this specific group of gain/loss of a pig W(t, w) in kilograms per day as a
people as time increases? Explain. function of weight and air temperature.
a. Find a model for the average daily weight
13. Weight The table shows average daily weight
gain/loss of a 68-kg pig as a function of the air
gain/loss of a pig W(t, w) in kilograms per day as a
temperature.
function of weight and air temperature.
b. Find and interpret the rate of change of the aver-
TABLE FOR ACTIVITIES 13 AND 14 Average daily age daily weight gain of a 68-kg pig with respect
weight gain/loss for a pig (kg/day) to a change in temperature when the tempera-
w, Mean live weight (kg) ture is 26.7°C.
68 91 113 136 156 c. Find a model for the average daily weight
4.4 0.58 0.54 0.50 0.46 0.43 gain/loss of a pig as a function of the pig’s weight
t, Air temperature (°C)
10 0.67 0.71 0.76 0.80 0.85
when the air temperature is 26.7°C.
15.6 0.79 0.87 0.94 1.02 1.09 d. Find and interpret the rate of change of the aver-
age daily weight gain of a 68-kg pig with respect
21.1 0.98 1.01 0.97 0.93 0.90
to the pig’s weight when the air temperature is
26.7 0.83 0.76 0.68 0.62 0.55 26.7°C.
32.2 0.52 0.40 0.28 0.16 0.05
15. Apparent Temperature The table of apparent
37.8 0.09 0.35 0.62 0.88 1.15
temperatures shows how hot it feels, in degrees
(Source: A. A. Hanson, ed., Practical Handbook of Agricultural Fahrenheit, for certain air temperatures and dew
Science, Boca Raton, FL: CRC Press, 1990.) points.
TABLE FOR ACTIVITIES 15 AND 16 Apparent temperature (°F)
Dew point (°F)
50 55 60 65 70 75 80 85
65 62.7 63.8 65.0 66.6
70 67.8 68.7 69.8 71.1 72.6
Air temperature (°F)
75 73.1 73.9 74.8 75.9 79.2 80.7
80 79.8 80.6 81.6 82.8 84.4 86.9 90.9
85 83.5 84.7 86.1 88.0 90.5 94.0 99.0 106.6
90 87.9 89.4 91.2 93.6 96.9 101.2 107.2 115.6
95 92.9 94.5 96.7 99.6 103.4 108.4 115.2 124.3
100 98.1 99.9 102.4 105.6 109.8 115.3 122.7 132.3
105 103.4 105.4 108.1 111.6 116.1 122.0 129.7 139.7
110 108.7 110.9 113.8 117.5 122.3 128.4 136.3 146.5
(Source: www.nws.noaa.gov (accessed 12/28/02.)
582 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
a. When the air temperature is 85°F and the dew c. Repeat part a; however, instead of entering the
point is 65°F, how rapidly is the apparent tem- interest rate in whole numbers, enter r in deci-
perature increasing with respect to the dew mals so that r 0.02 at 2%, r 0.03 at 3%, and
point? so on.
dA(14,000, r)
b. What steps were necessary in determining the d. Use the model from part c to find dr
answer to part a? when the APY is 12.7%. Compare this answer to
that of part b.
16. Apparent Temperature The table of apparent
temperatures shows how hot it feels in degrees 18. Investment The table gives the value A(t, r) of an
Fahrenheit for certain air temperatures and dew initial investment of $1000 after t years in an ac-
points. count whose interest rate r% is compounded con-
a. When the air temperature is 85°F and the dew tinuously.
point is 65°F, how rapidly is the apparent temper- a. Find a cross-sectional model for the value of
ature increasing with respect to air temperature? a $1000 investment in terms of r, assuming 10
b. Consider the units on the rate of change given in years have elapsed since the initial investment.
part a. Even though the units of apparent tem- b. Use the model in part a to estimate the rate of
perature and air temperature cancel, why is it change with respect to the interest rate of a $1000
helpful to report the rate of change in degrees per investment after 10 years in an account paying 7%
degree? interest. Interpret the rate of change as an approx-
imate change. Discuss why the rate of change
17. Investment The table gives the value A(P, r) of an appears to be 100 times as large as expected.
investment of P dollars after 2 years in an account
whose annual percentage yield is r%. c. Find a cross-sectional model for the value of a
$1000 investment in terms of the interest rate ex-
a. Find a model in terms of r for the value of a pressed as a percentage, assuming 10 years have
$14,000 investment after 2 years. elapsed. Compare this equation with the one you
dA(14,000, r)
b. Find and interpret dr when the annual found in part a.
interest rate is 12.7%.
TABLE FOR ACTIVITY 17 Amount (in dollars) after 2 years
Rate Initial investment
(%) $10,000 $12,000 $14,000 $16,000 $18,000
2 10,404.00 12,484.80 14,565.60 16,646.40 18,727.20
3 10,609.00 12,730.80 14,852.60 16,974.40 19,096.20
4 10,816.00 12,979.20 15,142.40 17,305.60 19,468.80
5 11,025.00 13,230.00 15,435.00 17,640.00 19,845.00
6 11,236.00 13,483.20 15,730.40 17,977.60 20,224.80
7 11,449.00 13,738.80 16,028.60 18,318.40 20,608.20
8 11,664.00 13,996.80 16,329.60 18,662.40 20,995.20
9 11,881.00 14,257.20 16,633.40 19,009.60 21,385.80
10 12,100.00 14,520.00 16,940.00 19,360.00 21,780.00
11 12,321.00 14,785.20 17,249.40 19,713.60 22,177.80
12 12,544.00 15,052.80 17,561.60 20,070.40 22,579.20
9.3 Partial Rates of Change 583
TABLE FOR ACTIVITY 18 d. Use the model in part c to estimate the rate of
Value (in dollars) of $1000 investment change with respect to the interest rate of a $1000
Rate Time (years) investment after 10 years in an account paying
(%) 5 7 10 12
7% interest. Interpret the rate of change as an ap-
proximate change. Compare these answers with
2 1105.17 1150.27 1221.40 1271.25
those you found in part b, and reconcile any dif-
3 1161.83 1233.68 1349.86 1433.33 ferences.
4 1221.40 1323.13 1491.82 1616.07
Discussing Concepts
5 1284.03 1419.07 1648.72 1822.12
6 1349.86 1521.96 1822.12 2054.43 19. Discuss the ways in which cross-sectional functions
7 1419.07 1632.32 2013.75 2316.37 can be used to help estimate the input and output of
optimal points on a three-dimensional function.
8 1491.82 1750.67 2225.54 2611.70
9 1568.31 1877.61 2459.60 2944.68 20. Discuss the drawbacks of using cross-sectional
analysis to try to determine optimal points on a
10 1648.72 2013.75 2718.28 3320.12
three-dimensional function.
11 1733.25 2159.77 3004.17 3743.42
12 1822.12 2316.37 3320.12 4220.70
9.3 Partial Rates of Change
In Section 9.2, we found rates of change for a cross-sectional function that we mod-
eled from a given table of data. The rates of change allowed us some insight into the
behavior of a multivariable function presented as a data table. In this section, we dis-
cuss rates of change found using a multivariable equation. This approach will enable
us to choose a particular cross section and then to choose a particular point on that
cross section. In this way, one general formula can be used as the basis for finding any
cross-sectional rate of change.
Partial Derivatives
Because rates of change of a cross-sectional function give us only part of the picture,
we refer to them as partial rates of change or partial derivatives. Suppose that we have
F
a multivariable function F with input variables x and y. We use the notation x and
F dF dF
y instead of dx and dy to remind ourselves that there is more than one input
variable in the underlying function F. Another common notation for the partial de-
rivative of F with respect to x is Fx.
The revenue generated by the sale of cheeses in the Netherlands can be modeled* as
R(l, r) 0.04r2 6r 0.06lr 10l 0.4l2 million guilders
(a guilder is a Dutch monetary unit) when l thousand tons of lower-fat (40% fat)
cheese and r thousand tons of regular cheese are sold.
Suppose that we want to find the rate of change of revenue with respect to the sales
*S. Louwes, J. Boot, and S. Wage, “A Quadratic-Programming Approach to the Problem of the Optimal
Use of Milk in the Netherlands,” Journal of Farm Economics, vol. 45, no. 2 (May 1963), pp. 309–317. The
model is rounded to simplify calculation.
584 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
of regular cheese when 17 thousand tons of lower-fat cheese is sold. We can change
R(l, r) into a single-variable function by replacing l with 17. This replacement gives us
the cross-sectional model R(17, r) 0.04r2 4.98r 54.4 million guilders rev-
enue when 17 thousand tons of lower-fat cheese and r thousand tons of regular cheese
are sold (see Figure 9.20).
R(l, r)
Revenue
(million
R(17, r) guilders)
200 200
100 80
60
0 40
r
10 20 Regular
0 r 20
0 80 l 30 0 cheese
Lower-fat cheese (thousand
(thousand tons) tons)
(a) (b)
FIGURE 9.20
Taking the derivative of this cross-sectional model with respect to r, we get
dR(17, r)
0.08r 4.98 million guilders per thousand tons of regular cheese. In
dr
mathematical notation,* we say that
dR(17, r)
dr
R
r
l17
. Evaluating this derivative of
a cross-sectional model at r 40 gives the slope of the tangent shown in Figure 9.21.
The derivative
R
r l17
dR(17, r)
dr
gives us a picture of the rate of change for only
only one cross section of the function R. It would be useful to develop a function that
could be used to give us the rate of change for any cross section with respect to r. To
do this, we treat r as a variable and l as a constant. We highlight r to help us remem-
ber that it is the variable that is changing and is not held at a constant value.
R(l, r) 0.04r2 6r 0.06lr 10l 0.4l2
*When writing the partial rate of change of a function z f(x, y) with respect to the variable x, we use the
f f
notation x or fx . The partial derivative of f with respect to x and evaluated at y b is denoted by x yb
or fx yb. This partial derivative gives the slope of a line tangent to the cross-sectional function for which
f
y b. The partial derivative of f with respect to y and evaluated at x a is denoted by y xa or fy xa. This
partial derivative gives the slope of a line tangent to the cross-sectional function for which x a. We write
f f
x (x , y)(a, b) or fx (x, y)(a, b) when evaluating the partial derivative x at the point where x a and y b,
where a and b are constants.
9.3 Partial Rates of Change 585
R(l, r)
Revenue R(17, r)
(million
guilders) 200
200
100 80
60
r
0 40 r
10
Regular cheese
20 0 20 40 60 80
20 (thousand tons)
l 30 0
Lower-fat cheese
(thousand tons)
(a) (b)
FIGURE 9.21
so finding the partial derivative of R with respect to r gives
R
Rr 0.04 2r 6 1 0.06l 1 0 0
r
0.08r 6 0.06l million guilders per thousand tons of regular cheese
Note that the derivatives of the first and second terms are straightforward. In the third
term, 0.06l acts as a constant multiple of r, so the derivative follows the Constant
Multiplier Rule. Because l acts as a constant, the last two terms are both constants, and
the derivative of any constant is 0.
The partial derivative of R with respect to r denoted as either Rr or r is a func-
R
R
tion of both r and l. If we evaluate r at a particular value of l—say l 17—we obtain
a single-variable function with input r. This function gives the slope of any line tan-
R
gent to the cross section R(17, r). We use the notation r l17 to represent this slope
R
function. Evaluating r l17 , we have 0.08r 4.98 million guilders per thosand tons
of regular cheese.
The advantage of finding the partial derivative is that the formula for the partial
derivative can be used to find rates of change at points on any cross section. The first
method restricts us to points on one particular cross section.
Now we have a function that can be used to evaluate the rate of change of R with
respect to r for any l. For example, estimate the partial rate of change of revenue with
respect to regular cheese sales when 40 thousand tons of lower-fat fat cheese and 30
thousand tons of regular cheese are sold. Using l 40 and r 30, we find that
R
r (40,30)
0.08 30 6 0.06 40
1.2 million guilders per thousand tons of regular cheese
That is, when sales of lower-fat cheese remain constant at 40 thousand tons and reg-
ular cheese sales increase from 30 to 31 thousand tons, revenue increases by approx-
imately 1.2 million guilders.
586 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
What would happen if the regular cheese sales were held constant and the sales
of lower-fat cheese were varied? In this case, we could find a general formula for the
partial rate of change of revenue with respect to the quantity of lower-fat cheese
sold:
R(l, r) 0.04r2 6r 0.06lr 10l 0.4l2 million guilders
so that
R
Rl 0 0 0.06r 10 0.8l
l
0.06r 10 0.8l million guilders per thousand tons of lower-fat cheese
when sales of regular cheese are held constant.
R
We now use the partial derivative l to estimate the partial rate of change of
revenue with respect to regular cheese sales when 40 thousand tons of lower-fat
cheese and 30 thousand tons of regular cheese are sold. When l 40 and r 30,
R
(l, r)(40, 30)
0.06 30 10 0.8 40
23.8 million guilders per thousand tons of lower-fat cheese
That is, when regular cheese sales remain constant at 30 thousand tons and lower-
fat cheese sales increase from 40 to 41 thousand tons, revenue decreases by ap-
proximately 23.8 million guilders. Again, this partial rate of change is the slope of
the tangent line at l 40 on the graph of the cross section with output R(l, 30).
In general, we find the partial derivative of a multivariable function with respect
to one input variable by treating all other input variables as constants and proceeding
to take derivatives as functions of a single input variable. For a two-variable function,
a partial rate of change can be visualized graphically as the slope of a line tangent to a
cross section.
EXAMPLE 1 Writing General Partial Derivative Formulas
Investment The formula for the accumulated value of an investment of P dollars
over t years is
nt
r
A(P, r, n, t) P 1 dollars
n
If the annual interest rate is 6% compounded quarterly, then the accumulated value
formula is
4t
0.06
A(P, t) P 1 P(1.0614t) dollars
4
A
a. Write a general formula for t . What notation is used to represent this general
formula evaluated at P 1000 and t 10?
A
b. Write a general formula for P . Find and interpret AP (P, t)(7500, 10).
9.3 Partial Rates of Change 587
Solution
a. Note that for each choice of principal P,
A
At P(ln 1.0614)(1.0614t) dollars per year
t
This result should not surprise you, because it follows the Constant Multiplier
Rule for derivatives: When f(x) kg(x), then f(x) kg(x). In this case, P is
A
the constant in the Constant Multiplier Rule. The general formula for t eval-
uated at P 1000 and t 10 is denoted by
A
t
(P, t)
(1000, 10) or At (P, t)(1000, 10)
A
b. For any number of years t, AP 1.0614t dollars per dollar invested
P
AP (P, t)(7500, 10) 1.061410 $1.81 per dollar invested when $7500 is in-
vested for 10 years at 6% compounded quarterly. The accumulated value of
the investment will increase by approximately $1.81 if an additional dollar is
invested. ●
Being able to calculate partial rates of change and evaluate these partial derivatives
at various input values is important for the applications we will consider in subse-
quent sections.
Second Partial Derivatives
In Chapter 4 we considered the second derivative of a single-variable function
and used the second derivative to locate inflection points. With multivariable func-
tions, we can consider the second partial derivatives—that is, partial derivatives of par-
tial derivatives. As we shall see in Chapter 10, these second partial derivatives are use-
ful in identifying extrema and other interesting points on three-dimensional graphs.
They also indicate how the partial rates of change of the function are changing.
Consider again the Example 1 investment function A(P, t) P(1.0614t) giving
the accumulated value (in dollars) of an investment of P dollars over t years when the
interest rate is 6% compounded quarterly. The two partial derivatives are
AP 1.0614t dollars per dollar invested
At P(ln 1.0614)(1.0614t) dollars per year
What is happening to the rate of change of A with respect to t as t increases? Let us first
look at this question graphically on one cross-sectional model, say A(1000, t). Figure
9.22, on page 588, depicts A(1000, t) 1000(1.0614t) along with tangent lines drawn
at t 2, t 10, t 18, and t 26.
As t increases, the slope becomes larger. Thus the rate of change with respect to t,
At P1000, is increasing as t increases, so the second derivative must be positive. We can
verify this by finding the derivative of At P1000 with respect to t.
(A ) [1000(ln 1.0614)(1.0614t)]
t t P1000 t
1000(ln 1.0614)2(1.0614t) dollars per year per year
588 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
A(1000, t)
(dollars)
6000
t
1000
Years
2 10 18 26
FIGURE 9.22
This second partial derivative is positive for all values of t, as Figure 9.22 suggests.
Note also that because this partial derivative of the partial derivative At P1000 is
positive, the graph of the cross-sectional model with output A(1000, t) is concave up.
Before continuing with our interpretation of second partial derivatives, we find all
second partial derivatives for A.
The partial derivative At is a function with two input variables, P and t. The par-
tial derivatives of At are AtP (the partial derivative of At with respect to P) and Att (the
partial derivative of At with respect to t). Even though the partial derivative of AP is a
function of only one input variable, it is derived from a function with two input vari-
ables. For this reason, AP also has two second partial derivatives: APP (the partial de-
rivative of AP with respect to P) and APt (the partial derivative of AP with respect to t).
The second partial derivatives for A are
The second partial derivative APP 0 dollars per dollar invested per dollar invested
notation may also be written
2A 2A APt (ln 1.0614)(1.0614t) dollars per dollar invested per year
APP 2 , APt
tP ,
P
2A 2A
AtP (ln 1.0614)(1.0614t) dollars per year per dollar invested
AtP Pt , and Att 2 .
t Att P(ln 1.0614)2(1.0614t) dollars per year per year
You may have noticed that APt AtP. We refer to APt and AtP as mixed second par-
tial derivatives. The mixed second partial derivatives of a multivariable function are
equal whenever the function and its first and second partial derivatives are
continuous. This is the case for all the multivariable functions in this book.
Concept Development: Interpreting Second Partials
We now further consider the interpretation of the second partial derivatives of A(P, t).
First, APP 0 tells us that the rate of change of A with respect to P does not change
when P changes. This can be seen by looking at the cross-sectional models of A for
different values of t. Figure 9.23 shows graphs of the linear cross-sectional functions
A(P, 2), A(P, 12), A(P, 18), and A(P, 26). In APP we consider P to be changing but
9.3 Partial Rates of Change 589
consider t to be constant; that is, we look at what happens to the slope of each cross
section as P changes. The cross sections are all lines, and as P increases, the slope of
any cross section remains constant. Thus APP 0, because the slopes do not change
as P changes.
Now let us consider APt (ln 1.0614)(1.0614t). No matter what t is, APt is posi-
tive. This tells us that the rate of change (slope) of A in the P direction increases as we
move in the t direction. To see this graphically, refer again to the cross sections in Fig-
ure 9.23. Now, however, instead of considering the effect on the slope of each cross
section as P increases, we consider what happens at a particular value of P as t in-
creases—that is, as we move from one cross section to another. Figure 9.24 depicts
the cross-sectional models shown in Figure 9.23 with the slope of each cross section
labeled at P 3000. Note that as t increases, the slopes of the cross sections also
increase. Hence APt 0. There are similar interpretations for Att and AtP.
A(P, t) A(P, t)
(dollars) (dollars)
25,000 t = 26 25,000 t = 26
4.7
e¯
op
t = 18 t = 18
2.9
Sl
¯
t = 12 pe t = 12
Slo 0
t=2 e ¯ 2. t = 2
p
Slo ¯ 1.1
P Slope P
0 0
0 5000 (dollars) 0 3000 5000 (dollars)
FIGURE 9.23 FIGURE 9.24
Keeping track of all the second partial derivatives for a particular function is eas-
ier with a mathematical array called a matrix. We write the second partial derivatives
in a square matrix—that is, a table of rows and columns where the numbers of rows
and columns are equal—with one of the second partial derivatives in each position.
We call this the second partials matrix,* and we use this matrix in the next chap-
ter to help identify extreme points. A second partials matrix can be formed by labeling
the rows with the input variables and labeling the columns with the input variables
in the same order. For example, one possible setup for a second partials matrix for a
multivariable function z f(x, y) is
x y
x
y
For convenience, we choose always to use the matrix with the input variables in the
same order in which they appear in the notation used to represent the function. The
labels on the rows and columns determine which second partial derivative we write
in each position of the matrix. Using the function z f(x, y), we write fxx in the
*This matrix is also called the Hessian of the function.
590 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
position corresponding to the row and column labeled x, we write fyx in the row la-
beled y and the column labeled x, and so forth.
x y
x fxx fxy
y fyx fyy
For the function A(P, t) P(1.0614t), the second partials matrix is
P t
P 0 (ln 1.0614)(1.0614t)
t (ln 1.0614)(1.0614t) P(ln 1.0614)2(1.0614t)
Example 2 shows a function with three input variables. Because the function has three
input variables, the process of finding its partials requires that two variables be con-
sidered as constants when finding each of the three partial derivatives.
EXAMPLE 2 Working with Partial Derivatives
Cheese Spread A measure of the adhesiveness of cheese spread can be modeled* as
a function of the percentage of glycerol, salt, and lactose used in preparation:
A(g, s, l) 5600 3400s 1600s2 400l 1800g 140g2
1200sl 500s2l 200sg
where g is the percentage of glycerol, s is the percentage of salt, and l is the percentage
of lactose.
a. Write the partial derivatives of A.
b. Write the second partials matrix of A.
c. Interpret Agg and Asl.
d. Write the second partials matrix of A when g 10, s 2, and l 6.
e. Interpret Als (g, s, l)(10, 2, 6).
Solution
a. The three partial derivatives are
A
1800 280g 200s units per percentage point of glycerol
g
A
3400 3200s 1200l 1000sl 200g units per
s
percentage point of salt
A
400 1200s 500s2 units per percentage point of lactose
l
*E.Kombila-Moundounga and C. Lacroix, “Effet des combinaisons de chloure de sodium, de lactose et de
glycerol sur les caractéristiques rhéologiques et la couleur des fromages fondus à tartiner,” Canadian Insti-
tute of Food Science and Technology Journal, vol. 24, no. 5 (1991), pp. 239–251.
9.3 Partial Rates of Change 591
b. The second partials matrix is
g s l
g Agg 280 Asg 200 Alg 0
s CAgs 200 Ass 3200 1000l Als 1200 1000sS
l Agl 0 Asl 1200 1000s All 0
c. The second partial derivative Agg 280 is positive. This means that as g
increases, the rate of change (slope) of A with respect to g (in the g direction) is
increasing. When the percentage of glycerol increases by 1 percentage point, the
rate of change of adhesiveness increases by approximately 280 units per
percentage point of glycerol if salt and lactose percentages are held constant.
The second partial derivative Asl 1200 1000s is positive when s 1.2 and
negative approximately when s 1.2. Consider the percentage of glycerol as
constant. If the percentage of salt is less than 1.2%, then the rate of change of
adhesiveness with respect to salt is increasing as the percentage of lactose in-
creases. Likewise, if the percentage of salt is more than 1.2%, then the rate of
change of adhesiveness with respect to salt is decreasing as the percentage of
lactose increases.
d. The second partials matrix of A at (10, 2, 6) is
g s l
g 280 200 0
s C200
2800
800S
l 0
800 0
e. We first note that Al (g, s, l)(10, 2, 6) 800. This means that adhesiveness is in-
creasing with respect to an increase in lactose above the 6% level. Because
Als (g, s, l)(10, 2, 6)
800 is negative, as the percentage of salt increases above
2%, adhesiveness is increasing less rapidly (the rate of change Al is decreasing)
with respect to an increase in the percentage of lactose above the 6% level when
the percentage of glycerol is held constant at 10%. ●
Partial rates of change and second partial derivatives will be important to us as we
continue our study of multivariable functions.
9.3 Concept Inventory 9.3 Activities
• Partial derivatives/partial rates of change Getting Started
• Partial derivative notation In Activities 1 through 6, a. write the mathematical no-
• Second partial derivatives tation for the partial rate-of-change function needed to
answer the question posed, and b. give the units of that
• Second partials matrix rate-of-change function.
• Interpreting partial derivatives
1. W(a, h) is the weight (in pounds) of a person who is
a years old and h inches tall. What is the rate of
change of weight with respect to height?
592 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
2. P(k, c) is the price (in dollars) of a diamond that N
b. Interpret p s6.
weighs k karats and is of color code c. How quickly
is the price of a diamond changing with respect to N
Would you expect p s6 to be positive or nega-
weight?
tive? Explain.
3. T(g, t) is the mean daily temperature (in degrees
Fahrenheit) at longitude g degrees and latitude t de- In Activities 9 through 18, find the indicated partial
grees. If longitude is 23° and latitude is changing, derivatives for each of the multivariable functions.
how rapidly is temperature changing? 9. f (x, y) 3x2 5xy 2y3
4. T(i, d) is the federal income tax (in dollars) owed by f f f
a. x b. y c. x y7
a person who made i dollars and can claim d de-
pendents in a given year. When 4 dependents can be 10. g(k, m) k3m5 2km
claimed and income is changing, how quickly is the
amount of taxes owed changing? a. gk b. gm c. gm k2
5. R(b, c) is the revenue (in dollars) a farmer makes 11. f (x, y) 5x 3 3x 2y 3 9xy 14x 8
from the sale of b bulls and c cows. When the farmer f f f
a. x b. y c. x y2
sells 2 bulls and 100 cows, how rapidly is revenue
changing if the number of cows sold changes? 12. K(a, b) 5ab3 7(1.4b)
6. G(h, s) is the expected grade-point average of a typ- K K K
a. a b. b c. b a6
ical freshman college student who had a GPA of h in
high school and made a combined score of s on the
13. M(t, s) s ln t 3.75s 14.96
SAT. What is the rate of change of the expected GPA
with respect to the SAT score when the high school a. Mt b. Ms c. Ms t3
GPA is 3.5 and the SAT score is 1048? 14. g(x, y, z) 3.2x 2yz 2 2.9x y z
7. Let P(l, m) be the probability that a certain senator a. gx b. gy c. gz
votes in favor of a tobacco-ban bill when the senator
s t
receives l letters supporting the bill and m million 15. h(s, t, r) t r (st tr)2
dollars is invested by the tobacco industry lobbying
h h h h
against the bill. a. s b. t c. r d. r (s, t, r)(1, 2, 1)
P
a. Interpret m l100,000. 16. k(x, y) x ln (x y)
P k k k k
Would you expect m l100,000 to be positive or a. x b. y c. y x3 d. y (x, y)(3, 2)
negative? Explain.
P 17. f (x, y) 2xy 8x 2y 3 5e2y 10
b. Interpret l m53.
a. fx b. fy
P c. fxx fxy
Would you expect l m53 to be positive or neg-
ative? Explain. fyx fyy
8. Let N(p, s) be the number of skiers on a Saturday at d. Write the second partials matrix.
a ski resort in Utah when p dollars is the price of an
18. G(r, t) t ln r 12rt7 4(8r) tr
all-day lift ticket and s is the number of inches of
fresh snow received since the previous Saturday. G G
a. b.
N
t r
a. Interpret s p25.
c. Grr Grt
N Gtr Gtt
Would you expect s p25 to be positive or neg-
ative? Explain. d. Write the second partials matrix.
9.3 Partial Rates of Change 593
Construct a second partials matrix for each of the func- 24. Investment The value A(t, r) of an investment of
tions in Activities 19 through 22. $1000 after t years in an account for which the in-
x y terest rate is 100r% compounded continuously is
19. f (x, y) y x
given by the function A(t, r) 1000ert dollars.
20. g(x, y) 6(3x y 4)3
a. Find and interpret A(10, r).
21. h(x, y) e2x3y A
b. Find and interpret r (t , r)(10, 0.07).
22. j(x, y) y2 ln x
c. Explain how the rate of change in part b is related
Applying Concepts to a graph of the cross-sectional function in part
a. Illustrate graphically.
23. Investment The value A(P, r) of an investment of
P dollars after 2 years in an account with annual 25. Radiation Values of UV (ultraviolet) radiation
percentage yield 100r% is given by the function are sometimes useful for the medical consultant, the
A(P, r) P(1 r)2 dollars. chemist, and the technical engineer. The table below
gives the amount of a certain type of UV radiation
a. Find and interpret A(14,000, r).
A
in watts per square meter per month.
b. Find and interpret r (P, r)(14,000, 0.127). a. Find cross-sectional models for the amount of
c. Explain how the rate of change in part b is related UV radiation at a given latitude in March and for
to a graph of the cross-sectional function in part the amount of radiation for a given month at a
a. Illustrate graphically. latitude of 50° south of the equator.
TABLE FOR ACTIVITY 25 Amount of UVA radiation (watts/m2/month)
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Year
North Pole — — — 17 41 52 46 27 — — — — 183
80° — — 3 19 42 52 47 28 8 — — — 179
70° — 2 11 26 43 51 47 32 16 5 — — 233
60° 3 8 20 34 46 53 49 39 24 12 4 2 294
50° 9 16 28 40 50 55 53 44 32 20 11 7 365
40° 19 25 35 45 53 56 54 49 39 29 20 15 439
30° 26 33 41 49 54 55 54 51 44 36 28 24 495
20° 34 40 47 51 53 53 53 52 48 42 36 32 541
10 42 46 50 51 50 50 50 50 50 47 43 40 569
Equator 49 50 52 50 47 45 46 48 50 50 49 47 583
10° 53 53 50 46 41 38 39 43 48 52 53 53 569
20° 56 54 49 42 34 31 32 38 45 51 55 56 543
30° 58 53 44 35 26 22 24 31 40 49 56 58 496
40° 57 50 39 27 18 13 15 23 34 45 55 59 435
50° 55 46 32 19 10 7 8 15 27 40 52 58 369
60° 52 40 24 11 5 1 2 7 18 33 47 56 296
70° 50 35 15 4 — — — 1 10 26 43 54 238
80° 49 27 8 — — — — — 3 19 41 56 203
South Pole 48 26 — — — — — — — 16 41 56 187
(Source: W. Rudloff, World-Climates, Stuttgart, Germany: Wissenschaftliche Verlagsgesellschaft, 1981.)
594 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
TABLE FOR ACTIVITY 26 Number of cars sold, S(m, y)
Year Month m
y Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec
1960 6550 8728 12,026 14,395 14,587 13,791 9498 8251 7049 9545 9364 8456
1961 7237 9374 11,837 13,784 15,926 13,821 11,143 7975 7610 10,015 12,759 8816
1962 10,677 10,947 15,200 17,010 20,900 16,205 12,143 8997 5568 11,474 12,256 10,583
1963 10,862 10,965 14,405 20,379 20,128 17,816 12,268 8642 7962 13,932 15,936 12,628
1964 12,267 12,470 18,944 21,259 22,015 18,581 15,175 10,306 10,792 14,752 13,754 11,738
1965 12,181 12,965 19,990 23,125 23,541 21,247 15,189 14,767 10,895 17,130 17,697 16,611
1966 12,674 12,760 20,249 22,135 20,677 19,933 15,388 15,113 13,401 16,135 17,562 14,720
1967 12,225 11,608 20,985 19,692 24,081 22,114 14,220 13,434 13,598 17,187 16,119 13,713
1968 13,210 14,251 20,139 21,725 26,099 21,084 18,024 16,722 14,385 21,342 17,180 14,577
(Source: B. Abraham and J. Ledolter, Statistical Methods for Forecasting, New York: Wiley, 1983.)
b. Estimate the rate of change of UV radiation at TABLE FOR ACTIVITY 27 Per capita consumption
50° south in March with respect to latitude. of peaches (pounds per person per year)
c. Estimate the rate of change of UV radiation at Yearly income Price per pound (dollars above $1.50)
50° south in March with respect to the month. (tens of thousands
d. Graph the cross-sectional functions in part a. On of dollars) 0 0.10 0.20 0.30 0.40 0.50
the graphs, illustrate the rates of change in parts 1 5.0 4.9 4.8 4.7 4.7 4.6
b and c. 2 6.4 6.3 6.2 6.1 6.1 6.0
26. Sales Refer to the table listing the number of cars 3 7.2 7.1 7.0 6.9 6.9 6.8
sold in Québec from January 1960 through Decem- 4 7.8 7.7 7.6 7.5 7.4 7.4
ber 1968.
5 8.2 8.1 8.0 8.0 7.9 7.8
a. Find cross-sectional models for car sales S(m, y)
6 8.6 8.5 8.4 8.3 8.3 8.2
in January and June as functions of the number
of years y since 1960.
b. Approximate the partial rates of change of car
sales with respect to the year of the sales in Janu-
ary 1965 and in June 1965. b. Use an exponential cross-sectional model to esti-
S mate the rate of change of the per capita con-
c. Estimate and interpret y at m 1 (January)
sumption of peaches with respect to price when
and y 2. the yearly income is $30,000 and the price is
27. Peaches The per capita consumption table shows $1.70 per pound.
selected values of the average yearly consumption of c. The data can be modeled by the equation
peaches per person, based on the price of peaches
C(p, i) 2ln i 2.7183 p 4
and the yearly income of the person’s family.
pounds per person per year
a. Use a logarithmic cross-sectional model to esti-
mate the rate of change of the per capita where the price of peaches is $(1.50 p) per
consumption of peaches with respect to yearly pound and the person lives in a family with an-
income when the yearly income is $30,000 and nual income $10,000i. Use this model to find
the price is $1.70 per pound. the rates of change in parts a and b.
9.3 Partial Rates of Change 595
d. Compare the answers you found using the two 29. Heat Loss Siple and Passel developed the equation
methods in this activity. Which method was eas-
H(v, t) (10.45 10v v)(33 t)
ier to use? Which answers do you believe are the
more reliable? giving the body’s heat loss in kilogram-calories per
square meter of body surface area per hour for wind
28. Sales Two vending machines sit side by side in a
speed v in meters per second when the air tempera-
college dorm. One machine sells Coke products,
ture is t degrees Celsius.
and the other sells Pepsi products. Daily sales of
(Source: W. Bosch and L. G. Cobb, “Windchill,” UMAP Module
Coke products, based on the prices of the products 658, The UMAP Journal, vol. 5, no. 4 (Winter 1984), pp. 477–492.)
in the two machines, are as shown.
a. Find the function expressing the partial rate of
change of heat loss with respect to air tempera-
TABLE FOR ACTIVITY 28 Daily sales (number of cans) ture and the function expressing the partial rate
of Coke products from a vending machine of change of heat loss with respect to wind speed.
H
Cost of Cost of Coke products b. Would you expect v at (v, t) (20, 12) to be
Pepsi products $0.50 $0.75 $1.00 $1.25 $1.50 positive or negative? Give reasons.
$0.50 157 143 123 98 65 H
c. Find the value of v at (v, t) (20, 12). Inter-
$0.75 206 192 172 146 114 pret your answer.
$1.00 255 241 221 195 163 H
d. Would you expect t at (v, t) (20, 12) to be
$1.25 304 290 270 244 211 positive or negative? Give reasons.
$1.50 353 339 319 293 260 H
e. Find the value of t at (v, t) (20, 12). Inter-
pret your answer.
a. Use a quadratic cross-sectional model to esti- 30. Livestock The carrying capacity of a particular
mate the rate of change of the sale of Coke prod- farm system is defined as the number of animals or
ucts with respect to the price of Coke products people that can be supported by the crop produc-
when Coke products cost $0.75 and Pepsi prod- tion from a given land area. The carrying capacity
ucts cost $1.25. of a crop of rye grass when 70% of the crop is
b. Use a linear cross-sectional model to estimate the consumed and 80% of the amount consumed is
rate of change of the sale of Coke products with re- digested as useful nutrients is
spect to the price of Pepsi products when Coke 9.52P
products cost $0.75 and Pepsi products cost $1.25. K(P, A) animals per hectare
A
c. The data can be modeled by the equation
where P is the net crop production (in kilograms per
S(c, p) 196.42p 50.2c2 9.6c 66.4 1.04cp hectare per year) and A is the annual energy require-
cans per day ments of the animal (in megajoules per animal).
(Source: R. S. Loomis and D. J. Connor, Crop Ecology: Produc-
when Coke products cost $c and Pepsi products tivity and Management in Agricultural Systems, Cambridge,
cost $p. Use this model to find the rates of change England: Cambridge University Press, 1992.)
in parts a and b. a. Write a function for the partial rate of change of
d. Compare the answers you found using the two carrying capacity with respect to net crop pro-
methods in this activity. Which method was eas- duction.
ier to use? Which answers do you believe are the b. How quickly is carrying capacity changing with
more reliable? respect to net crop production when 15,000 kg of
e. Use the equation in part c to find and interpret rye grass is produced per hectare annually and
the crop is used to support milking cows that
S
c
(c, p)(1.30, 1.20)
and
S
p .
(c, p)(1.30, 1.20)
require approximately 64,000 megajoules of
energy per cow each year?
596 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
c. Write a function for the partial rate of change of where g is the percentage of glucose and maltose, m
the carrying capacity with respect to the animal’s is the percentage of moisture, s is the percentage of
energy requirements. seed, and h is the holding time in days.
d. How quickly is the carrying capacity changing (Source: J. M. Shinn and S. L. Wang, “Textural Analysis of
Crystallized Honey Using Response Surface Methodology,”
with respect to the animal’s energy require- Canadian Institute of Food Science and Technology Journal, vol.
ments when 15,000 kg of rye grass is produced 23, nos. 4–5 (1990), pp. 178–182.)
per hectare annually and the crop is used to
a. Write functions for each of the partial derivatives
support milking cows that require approxi-
of A.
mately 64,000 megajoules of energy per cow
each year? b. Tell which partial derivative should be used to
answer the question “How quickly is adhesive-
e. Construct a second partials matrix for K, and in-
ness changing as the percentage of glucose and
terpret each of the second partial derivatives.
maltose changes?”
31. Livestock The amount of organic matter that one c. For which input variable(s) do you need a spe-
beef cow grazing on the Northern Great Plains cific value in order to determine the actual rate at
rangeland eats each day can be modeled by which adhesiveness is changing with respect to
I(s, m) 8.62 1.24s 0.09s2 0.21m the percentage moisture?
0.036m2 0.21sm d. Find a second partials matrix for A.
kilograms per day when the cow produces m kilo- 33. Investment The value A(t, r) of an investment of
grams of milk per day and s is a number between 4 $1000 after t years in an account for which the in-
and 4 that describes the size of the cow. terest rate 100r% is compounded continuously is
(Source: E. E. Grings et al., “Efficiency of Production in Cattle of given by the function A(t, r) 1000ert dollars.
Two Growth Potentials on Northern Great Plains Rangelands A A
During Spring-Summer Grazing,” Journal of Food Science, vol. a. Write the partial derivatives t and r .
74, no. 10 (1996), pp. 2317–2326.)
b. Construct a second partials matrix for A, and in-
I I terpret each of the second partial derivatives for
a. Would you expect s and m to be positive or
negative? Explain. t 30 and r 0.047.
I I 34. Investment The value A(P, r) of an investment of
b. Write and interpret functions for s and m.
P dollars after 2 years in an account with annual
c. When a cow is of size 2 and produces 6 kg of milk percentage yield 100r% is given by the function
per day, how quickly is the amount of organic A(P, r) P(1 r)2 dollars.
matter consumed by the cow changing as its milk
a. Find the first partial derivatives of A.
production increases?
b. Construct a second partials matrix for A, and in-
d. When a cow is of size 2 and produces 6 kg of milk
terpret each of the second partial derivatives for
per day, how quickly is the amount of organic
P 10,000 and r 0.09.
matter consumed by the cow changing as its size
increases? 35. Investment The value A(r, t) of an investment of 1
million dollars with an annual yield of 100r% is given
e. Construct a second partials matrix for I. What
by the function A(r, t) (1 r)t million dollars.
does this matrix indicate about the shape of the A
surface of I? a. Write the partial derivative t . What are the
A
32. Honey A measure of the adhesiveness of honey units on t ?
that is being seeded with crystals to cause controlled A
b. Write the partial derivative r . What are the
crystallization can be modeled by A
units on r ?
A(g, m, s, h) 151.78 4.26g 5.69m
c. How quickly will the value of the investment be
0.67s 2.48h 0.05g2 0.14m2 changing with respect to time 5 years after the in-
0.03s2 0.05h2 0.07mh
9.4 Compensating for Change 597
vestment is made if the investment yields 15% when the interest rate and the monthly payment
annually? amount are fixed at 11% and $250, respectively,
d. Illustrate the answer to part c using a graph. and the period is 3 years but may vary.
d. Use graphs to illustrate the rates of change in
36. Loan Amount The following equation gives the parts b and c.
amount of a loan A (in dollars) given the interest
rate 100r%, the period n (in months), and the Discussing Concepts
monthly payments m (in dollars).
n
37. When finding a formula for the rate-of-change of
12m r a three-dimensional function f with respect to
A(r, n, m) 1 1
r 12 the variable x when the other variable y is held
constant at c, does it matter whether you substi-
a. Write an expression for the rate of change of the
loan amount with respect to the amount of the tute the known constant y c before or after
monthly payments. you algebraically find the partial derivative fx?
Explain.
b. How quickly is the loan amount changing with
respect to the amount of the monthly payments 38. What must be true about the partial derivatives of a
when $500 is paid monthly for 15 years on a loan three-dimensional function at a relative maximum?
with 12% interest? Explain from a graphical viewpoint why this is true.
c. Write the appropriate partial derivative, and cal-
culate the rate of change of the loan amount
9.4 Compensating for Change
The following question often arises in situations where output depends on two input
variables: “If output must remain fixed at some constant level and one of the input
variables changes, how must the other input variable change in order to compen-
sate?” In other words, given a function of two variables z f (x, y), if x increases (or
decreases), how must y change in order to guarantee that the output f (x, y) remains
the same?
Let us consider this question in the context of a manufacturing problem where
production depends on more than one input. The number of units produced
during a manufacturing process can often be regarded as depending on several
different input variables, two of which are the size L of the labor force and the
amount K invested in capital. Economists often use the Cobb-Douglas production
function
f (L, K) cLaK1a
as a simple model for the number of units produced. The values c and a in the
Cobb-Douglas production function are constants that depend on the particular
manufacturing process. Furthermore, the model requires that a be greater than 0
but less than 1.
The Cobb-Douglas monthly production function for a certain mattress-
manufacturing plant is
f (L, K) 48.1L0.6K0.4 mattresses
598 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
where L represents the number of worker hours (in thousands) and K represents the
amount invested in capital (in thousands of dollars). Let us suppose that currently
each month the manufacturer spends $47,000 in capital, has a labor force that works
8320 worker hours, and produces f (8.32, 47) 800 mattresses.
A competing plant offering higher wages is hiring new employees. Management
expects some of its employees to transfer to the new plant. Instead of replacing the em-
ployees, management decides to upgrade the equipment to improve efficiency. The
number of lost worker hours is not yet known, so the information that is most bene-
ficial to management is the rate of change of capital K with respect to worker hours L
dK
for the current production level. How do we find dL ?
Concept Development: Calculating Compensation
for Change
To determine how to compensate for the change in the labor force, we look at the
contour curve corresponding to 800 mattresses. This helps us understand the con-
nection between the two input variables L and K. By solving
800 48.1L0.6K0.4
K
Capital for K in terms of L, we obtain the equation of the contour curve shown in Figure 9.25.
1
expenditures
800 0.4
(thousands K
48.1L0.6
of dollars) 2.5
800L 0.6
150
800
48.1
2.5
100
800
48.1
L
1.5 thousand dollars
A tangent line is drawn on the curve in Figure 9.25 at the current amount of capital
50 (8.32, 47)
and number of worker hours.
L dK
Worker To find dL , we simply differentiate the contour curve equation with respect to L
0 to obtain
hours
0 4 8 12
(thousands) 2.5
FIGURE 9.25
dK
dL
1.5
800
48.1
L2.5 thousand dollars per thousand worker hours
Evaluating this derivative at the current number of worker hours (L 8.32) gives
dK
dL
8.475 thousand dollars per thousand worker hours. This value is the slope of
the tangent line shown in Figure 9.25.
Recall from Chapter 4 that we can use the slope of a tangent line to estimate by how
much an output quantity will change if the input changes by a small amount. That is,
dy
for a function y f(x), if the slope of the tangent line is dx when x c and the input
x changes by a small amount (x), then the corresponding change (y) in the output
y can be approximated by
dy
y x
dx
This approximation is illustrated in Figure 9.26.
9.4 Compensating for Change 599
Slope of tangent f(x)
∆y
≈ line when ∆x
∆x
is small
Actual
∆y dy ∆y = change in
≈
∆x dx f(x)
dy ∆x
∆y ≈ ∆x
dx
Approximate change
in f(x)
x
c
FIGURE 9.26
In the Cobb-Douglas example, the number of worker hours is currently L 8.32
dK
and dL 8.475 thousand dollars per thousand worker hours. That is, if the plant
decreases the monthly labor force by 1000 worker hours, the amount spent on capital
must increase by approximately $8475 if the production level is to remain constant.
Suppose that the plant loses 500 worker hours as a result of employees leaving to
work at the competing plant. In this case, the change in L is L 0.5 thousand
worker hours. Using the approximating-change equation just discussed, we estimate
the corresponding change in capital K as
dK
K L
dL
K ($8.475 thousand/thousand worker hours)
(0.5 thousand worker hours)
$4.24 thousand
Thus approximately $4240 more will need to be spent on capital improvements in or-
der to maintain the current production level and compensate for the loss of 500
worker hours. Figure 9.27 illustrates this approximation. As the figure shows, the ap-
proximation is an underestimate.
K K
55 55
80 80
0 0
50 co 50 co
nt nt
∆K ≈ $4.24 thousand ou ∆K ou
r r
∆L ≈ –0.5 thousand ∆L ≈ –0.5 thousand
45 45
worker hours worker hours
40 L 40 L
7.5 8 9 7.5 8 9
(a) Approximate change needed in K (b) Actual change needed in K
FIGURE 9.27
600 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
Similarly, if the plant loses 2500 worker hours to the competing plant, then the
change in capital is approximated as
dK
K L
dL
K ($8.475 thousand/thousand worker hours)
(2.5 thousand worker hours)
$21.2 thousand
In order not to reduce production, an additional amount of approximately $21,200 in
capital expenditures is needed to compensate for the loss of 2500 worker hours.
A General Formula
dK
The preceding method of finding dL works only if we can solve the contour curve equa-
tion for one of the variables. How do we find a derivative of one input variable with
respect to another input variable if we are not able to solve the contour curve equation
algebraically for one of the variables?
Consider a function z f (x, y) and the contour curve equation f (x, y) c. Even
if we cannot solve for y in terms of x, it is true that we can consider y to be a function*
of x. We replace y with y(x) and write the contour curve equation as f (x, y(x)) c.
Note that the function f in this form is a composite function, where f is the outside
function and y is the inside function. Recall from Chapter 3 the Chain Rule for the
derivative of a composite function f g . If we call g the inside function and f the out-
side function, then the derivative of f with respect to x is
df
(derivative of the outside function)(derivative of the inside function)
dx
df dg
dg dx
We use the Chain Rule as we develop the formula for the derivative of z f (x, y(x))
with respect to x..
To take the derivative of z f (x, y(x)) with respect to x, we consider the effect
that a change in x has on the function f as well as the effect that a change in x has on
y and the subsequent effect on the function f; that is,
Ratewith
of change f (x, y(x))
respect to x
partial rate of change
of f with respect to x
of f with respect to y with respect to x
partial rate of change rate of change of y
f f
dx
d dy
[f (x, y(x)]
dx x y
You may wonder why we use both derivative and partial derivative notation in the
preceding equation. We use partial derivative notation when considering f to be a
function of two variables, x and y, and we use derivative notation when considering
f or y to be a function of the single variable x.
*When y is not a function of x, y can be broken into parts so that each part is a function of x.
9.4 Compensating for Change 601
With this formula, we return to the contour equation, differentiate both sides with
dy
respect to x, and solve for dx .
f (x, y(x)) c
d
[ f (x, y(x))] 0
dx
f f dy
x
y dx
0
dy
xf f x
yf f
dx y
This formula gives a method of finding a rate of change of one input variable with
respect to another input variable at a point on a contour curve. It gives the slope of
the tangent line at a point on a contour curve. This method is particularly useful
when we cannot (or do not wish to) solve for one variable in terms of another.
The Slope of a Line Tangent to a Contour Curve
When a function of two variables z f (x, y) is held constant at a value c, the
slope at any point on the contour curve f (x, y) c (that is, the slope of the
line tangent to the c contour curve) is given by
dy
xf f x
yf f
dx y
whenever fy
0.
dy
The rate of change dx tells us that if the value of the variable x changes by a small
amount x, then the value of the variable y will change by a small amount that can be
dy
estimated as dx times x. Thus, as long as we remain on a constant level curve, we
can compensate for a small change in one input variable x by making a comparable
change in the other input variable determined by the approximation y dx x.
dy
We have seen how to approximate the change needed in one variable to compen-
sate for a change in another variable in order to maintain a constant function output
in two ways: using a contour curve equation solved for one of the input variables to
directly calculate a tangent line slope or using partial derivatives to calculate the
needed slope. We summarize these two methods as follows:
602 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
Compensating for Change
In order to compensate for a small change x in x to keep f (x, y) constant at
c, y must change by approximately
dx x
dy
y
which can also be calculated as
fx
y f x
y
EXAMPLE 1 Finding the Slope of a Line Tangent to a Contour Curve
BMI Your body-mass index is a measure of how thin you are compared to your
height. A person’s body-mass index* is given by
9.4.1
0.4536w
B(h, w) points
0.00064516h2
where h is your height in inches and w is your weight in pounds. A teenage boy, in the
middle of a growth spurt, has an appointment to see his doctor. He is 5 feet 7 inches
tall and weighs 129 pounds. The doctor wants to ensure that the boy’s body-mass in-
dex remains constant.
dw
a. Find dh at the point (67, 129) on the contour curve corresponding to the boy’s
current body-mass index.
dw
b. Use dh to estimate the weight change needed to compensate for growths of 0.5
inch, 1 inch, and 2 inches if the boy’s body-mass index is to remain constant.
Solution
a. We begin by finding the two partial derivatives of B:
(2)(0.4536w)
Bh points per inch
0.00064516h3
0.4536
Bw points per pound
0.00064516h2
where h is height in inches and w is weight in pounds. We evaluate these partial
derivatives at w 129 pounds and h 67 inches and obtain
Bh 0.60312 point per inch and Bw 0.15662 point per pound
Thus,
dw Bh (0.60312) point per inch
dh Bw 0.15662 point per pound
3.85 pounds per inch
*New England Journal of Medicine, September 14, 1995.
9.4 Compensating for Change 603
b. We know that the change in weight needed to compensate for an increase in
dw
height can be approximated by w dh h. Thus,
For a 0.5-inch growth, w (3.85 pounds per inch) (0.5 inch) 1.9 pounds.
For a 1-inch growth, w (3.85 pounds per inch)(1 inch) 3.85 pounds.
For a 2-inch growth, w (3.85 pounds per inch)(2 inch) 7.7 pounds. ●
It is possible to estimate compensating change by using a table of data rather than a
multivariable function. This is illustrated in Example 2.
EXAMPLE 2 Compensating for Change Using a Table
Tuition A college wishes to keep its enrollment constant at 5600 students. From ex-
perience, the trustees know that the two factors that have the greatest influence on en-
rollment are tuition and the amount of available financial aid. Some research has been
done at similar colleges, and Table 9.6 gives enrollment figures based on that research.
TABLE 9.6 College enrollment
Tuition (thousand dollars)
16 16.5 17.0 17.5 18.0 18.5 19 19.5 20.0
1.50 6358 6000 5662 5344 5043 4759 4491 4239 4000
Average financial aid per student
1.75 6722 6346 5990 5656 5339 5040 4757 4492 4240
2.00 7086 6691 6318 5967 5634 5320 5024 4745 4480
(thousand dollars)
2.25 7451 7037 6647 6279 5930 5600 5290 4997 4720
2.50 7815 7383 6975 6590 6226 5882 5557 5250 4960
2.75 8179 7729 7303 6902 6522 6163 5823 5503 5200
3.00 8543 8074 7631 7213 6817 6443 6089 5756 5440
3.25 8907 8420 7959 7525 7113 6724 6356 6009 5680
3.50 9272 8766 8288 7836 7409 7005 6622 6261 5920
3.75 9636 9111 8616 8148 7704 7285 6889 6514 6160
4.00 10,000 9457 8944 8459 8000 7566 7155 6767 6400
a. Using only the table values, estimate the change in the amount of financial aid
the college should make available if it boosts tuition from $18.5 thousand to
$19 thousand and wants to maintain an enrollment of 5600 students.
b. Use cross-sectional models to estimate the change in part a.
Solution
a. Reading in the column corresponding to a tuition of $18,500, we find that the
enrollment of 5600 students occurs when the average financial aid is $2250 per
student. If tuition is increased to $19,000, we must move one column to the
right. Because the college wishes to keep enrollment constant, we seek a value of
5600 students in the $19,000 column. This is equivalent to staying on the 5600
contour curve (shown in teal in Figure 9.28), while increasing tuition to $19,000.
604 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
Tuition (thousands of dollars)
18.0 18.5 19 19.5
student (thousands of dollars)
Average financial aid per
1.50 5043 4759 4491 4239
1.75 5339 5040 4757 4492
2.00 5634 5320 5024 4745
2.25 5930 5600 5290 4997
2.50 6226 5882 5557 5250
2.75 6522 6163 5823 5503
FIGURE 9.28 3.00 6817 6443 6089 5736
5600
A look at the table shows that there is no 5600 entry in the $19,000 column.
However, we know that 5600 occurs between 5557 (corresponding to financial
aid of $2500) and 5823 (corresponding to financial aid of $2750). Thus the
average financial aid will need to increase by a little more than $250 per student
in order for tuition to increase to $19,000 and for enrollment to remain
constant.
b. We choose to represent enrollment as E(t, a) students, where tuition is t thou-
sand dollars and average financial aid is a thousand dollars per student. Note
that if t 18.5 and E 5600, then a 2.25. We seek to estimate the change
in financial aid a needed to compensate for an increase in tuition t using the
equation
da Et
a t t
dt Ea
We begin by calculating rates of change of the appropriate cross-sectional mod-
els. To estimate Ea when t 18.5 and a 2.25, we find a cross-sectional model
for E(18.5, a). This is a linear model given by
E(18.5, a) 1122.87a 3074.55 students
when a thousand dollars of financial aid is available and the tuition is $18,500.
Now we find
E
a
t18.5
dE(18.5, a)
da
1122.87 students per thousand dollars of
financial aid
Thus if t 18.5 and a 2.25, then Ea 1122.87 students per thousand dollars.
Likewise, we find the cross-sectional model
E(t, 2.25) 46,275.637 (0.892129t) students
when tuition is t thousand dollars and financial aid is $2250. Thus
E
t
a2.25
dE(t, 2.25)
dt
5282.10(0.892129t) students per thousand
dollars of tuition
Thus if t 18.5 and a 2.25, then Et 639.3 students per thousand dollars.
9.4 Compensating for Change 605
Because the change in tuition is t 0.5 thousand dollars, we estimate the
necessary change in financial aid as
Et
a t
Ea
639.3 students per $1000 of financia1 aid
1122.87 students per $1000 of tuition
(0.5 thousand tuition dollars)
0.28 thousand dollars of financial aid
The average financial aid available per student will need to increase by approx-
imately $280 in order to compensate for the increased tuition to maintain the
current enrollment. (The estimate we made in part a using only table values, a
little more than $250, agrees with this result.) ●
In this section, we have seen examples of finding rates of change on contour
curves and using them to compensate for a change in one input variable. Compen-
sating for change is a valuable tool with many real-world applications.
9.4 Concept Inventory 6. f(s, t) s ln (2t) e 1.34t
f(s, t) 2 for 1 t 20
• Approximating change
tangent line at t 10
• Slope of a line tangent to a contour curve
7. f(a, b) 2.8a2b3 1.8a2 12b
• Compensating for change
f(a, b) 15 for 0 a 10 and 0.87 b 1.2
tangent line at b 0.9
9.4 Activities 8. f(m, n) 10n(3.67 m)2e 0.2m
f(m, n) 80 for 1 m 10 and 0 n 50
Getting Started
tangent lines at n 20
For each of the functions in Activities 1 through 4, write
a formula for the indicated rate of change. For the functions in Activities 9 through 12, determine
dx the output associated with the input given, and then
1. f(x, y) 15x2y3;
dy
approximate the change needed in one input variable
dc
2. S(c, k) c(39 ); dk
k
to compensate for the given change in the other input
dm variable.
3. g(m, n) 59.3ln m 49nm 16; dn
dr 9. f(m, n) 3m2 2mn 5n2 when m 2, n 1,
4. A(r, t) 83.2e rt ; dt
and m 0.2
For each of the functions in Activities 5 through 8, sketch
the contour curve indicated in the activity. Also sketch 10. f(h, k) (32h3 15h2 10h 47)(43k 15)
the tangent line indicated, and determine its slope. when h 4.2, k 3.7, and k 0.6
5.* g(x, y) x(1.05y) 11. f(h, s) 0.00091s[0.103(2.5h) 1] when h 3.5,
g(x, y) 100 for 0 y 10 s 1148, and h 0.5
tangent line at y 5 12. W(r, h) 2.8r2(1.08h) 59r (0.3h2 3.3h 72)
when r 10, h 60, and r 1.3
dy
*In Activity 5 it is possible to consider the slope of the tangent line as dx or
dx
as dy , depending on which input variable is placed on the horizontal axis.
A similar note applies to Activities 6 through 12.
606 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
dh
Applying Concepts c. Use dt to estimate the change in the relative
humidity needed to compensate for a 2°F in-
13. Cost The cost of having specialty T-shirts made crease in temperature if the current conditions
depends on the number of colors used in the T-shirt are those stated in part a and the apparent tem-
design and the number of T-shirts being ordered. A perature is to remain constant.
function giving A, the cost per T-shirt (the average
d. Repeat part c for a 3.5°F decrease in temperature.
cost), when c colors are used and n T-shirts are or-
dered is 15. Sunflowers A process to extract pectin and pig-
ment from sunflower heads involves washing the
A(c, n) (
0.02c2 0.35c 0.99)(0.99897n) sunflower heads in heated water. It has been shown
0.46c 2.57 dollars that the percentage of pigment that can be removed
(Source: Based on data compiled from 1993 prices at Tiger- from a sunflower head by washing for 20 minutes
town Graphics, Inc., Clemson, SC.) can be modeled as
a. Determine the average cost when 250 T-shirts p(t, r) 306.761 9.6544t 1.9836r
are printed with six colors.
0.07368t2 0.02958r2 percent
b. When 250 T-shirts are printed with six colors,
where r milliliters of t°C water is used for each gram
how quickly is the average cost changing when
of sunflower heads.
more T-shirts are printed?
(Source: X. Q. Shi et al., “Optimizing Water Washing Process
dn
c. Find a formula for dc . If average cost is to remain for Sunflower Heads Before Pectin Extraction,” Journal of Food
Science, vol. 61, no. 3 (1996), pp. 608–612.)
dn
constant, would you expect dc to be positive or a. Draw the 53% contour curve for 20 r 45
negative? Explain. and 85 t 88.
d. A fraternity is planning to buy 500 four-color b. Draw the line(s) tangent to the 53% contour
shirts. One of the members has proposed sev- curve when the temperature is 86.5°C.
eral alternative designs, some using more and c. Write a formula for the rate of change of temper-
dn
some fewer than four colors. Use dc to estimate ature with respect to a change in the amount of
the change in order size needed to compensate water used when 53% of the pigment is removed.
for an increase or decrease in the number of d. Use the formula from part c to determine the
colors if the average cost per T-shirt is to re- slope(s) of the tangent line(s) drawn in part b.
main constant.
16. Free Volume In 1965, Honer developed a model
14. Heat Index A model for the apparent tempera- for the total-stem volume of red pine trees in
ture (how hot or cold it feels) is Canada. His model is
A(h, t) 2.70 0.885t 78.7h 1.20th °F d2
V(d, h) cubic feet
0.691 363.676h 1
for an air temperature of t degrees Fahrenheit and a
where d is the diameter of the tree at breast height
relative humidity of 100h%.
(4.5 feet above the ground), which is denoted
(Source: W. Bosch and L. G. Cobb, “Temperature Humidity
Indices,” UMAP Module 691, The UMAP Journal, vol. 10, no. 3 by dbh and measured in inches, and the tree is h
(Fall 1989), pp. 237–256.) feet tall.
(Source: J. L. Clutter et al., Timber Management: A Quantita-
a. How hot does it feel when the relative humidity tive Approach, New York: Wiley, 1983.)
is 85% and the air temperature is 90°F?
a. Find the volume of a 40-foot tree with a 2-foot
dh dbh and draw a contour curve corresponding to
b. Find a formula for dt . If the apparent tempera-
dh that volume.
ture is to remain constant, would you expect dt
b. On the contour curve, draw a line whose slope
to be positive or negative? Explain. represents how quickly height is changing as dbh
9.4 Compensating for Change 607
is changing if the volume does not change for a 20. Tuition Refer to Table 9.6 on page 603, which
40-foot tree with a 2-foot dbh. shows a college’s enrollment as a function of tuition
c. Assuming a constant volume, find the rate at and financial aid. The college now has enrollment of
which height changes with respect to dbh for a 7000 and average financial aid of $3500 per student.
40-foot tree with a 2-foot dbh. a. Sketch the 7000 contour curve, and locate the
d. A logging company wishes to cut only trees that approximate point on Table 9.6 corresponding
have a volume of at least 59 cubic feet. Use the to the college’s current situation.
answer to part c to estimate the height of a tree b. The board of trustees has decided to reduce the
with dbh 25.5 inches that satisfies the volume average financial aid by $500 per student in order
requirement. Repeat for a dbh of 23 inches. to build a new dormitory. Use Table 9.6 to esti-
e. How is your answer to part c related to the line mate the decrease in tuition that will be necessary
you drew in part b? to offset the cut in financial aid and allow enroll-
ment to remain constant.
17. BMI Refer to the BMI function in Example 1.
c. Use cross-sectional models to estimate the
a. Solve for w as a function of h for the body mass change in part b.
of the teenage boy in Example 1.
dw 21. Sales The table shows the sales of Coke products
b. Use the formula in part a to find dh when
from a vending machine that is located next to a
h 67 inches and w 129 pounds.
Pepsi vending machine. The current price of Coke
c. Compare your answer in part b to the answer in products is $1.00, and Pepsi products cost $1.25.
part a of Example 1.
18. Skin The amount of skin covering a person’s body
(in square feet) depends on the person’s height and TABLE FOR ACTIVITY 21 Number of cans of
weight. One model for estimating this skin surface Coke products sold daily from a vending machine
area is given by the multivariable function Cost of Cost of Coke products
A(w, h) 0.6416w 0.425h 0.725 square feet Pepsi
products $0.50 $0.75 $1.00 $1.25 $1.50
for a person weighing w pounds who is h feet tall. If
$0.50 157 143 123 98 65
a person 5 feet 11 inches tall who weighs 130 pounds
grows 2 inches in height, by approximately how $0.75 206 192 172 146 114
much must this person’s weight change if the skin $1.00 255 241 221 195 163
surface area is to remain the same? $1.25 304 290 270 244 211
19. Payments The amount of a monthly payment on $1.50 353 339 319 293 260
a loan with 6% interest compounded monthly can
be calculated by using the function
0.005A a. Pepsi is planning to lower its price to $1.00. Use
m(A, t) dollars
1 0.9419t the table values to estimate by how much Coke
would need to change its price in order to com-
when the loan is for A dollars and is to be repaid
pensate for this change and retain current sales.
over t years.
b. Using cross-sectional models, estimate the
a. Suppose that you borrow $10,000 to buy a car
change in part a.
and will repay the loan over a period of 5 years.
What is your monthly payment? c. Judging on the basis of your estimates in parts a
and b, what action do you think Coke should
b. Suppose that you decide to finance the car over a
take when Pepsi lowers its prices?
period of 4 years instead of 5. Approximate the
amount you could borrow to buy a car without 22. Peaches The table gives the per capita consump-
increasing or decreasing the monthly payment tion of peaches as a function of annual family in-
determined in part a. come and the price of peaches. The current average
608 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
annual family income in a community is $40,000, average annual family income by $5000. Use the
and the average price of peaches is $1.99 per table to estimate the change in the price of
pound. peaches needed to compensate for the decrease
in income.
TABLE FOR ACTIVITY 22 Per capita consumption b. Use cross-sectional models to estimate the
of peaches (pounds per person per year) change in part a.
Yearly Price per pound (dollars above $1.50) c. The peach consumption data can be modeled by
income the equation
(tens of
thousands
C(p, i) 2ln i 2.718
p 4 pounds
of dollars) 0 0.10 0.20 0.30 0.40 0.50 per person per year
1 5.0 4.9 4.8 4.7 4.7 4.6 where the price of peaches is $(1.50 p) per
2 6.4 6.3 6.2 6.1 6.1 6.0 pound and the person lives in a family with an-
nual income $10,000i. Estimate the change in
3 7.2 7.1 7.0 6.9 6.9 6.8
part a using this equation.
4 7.8 7.7 7.6 7.5 7.4 7.4
d. Compare and contrast the three methods of esti-
5 8.2 8.1 8.0 8.0 7.9 7.8 mating change used in this activity in terms of
6 8.6 8.5 8.4 8.3 8.3 8.2 ease of use and accuracy.
Discussing Concepts
a. Suppose that in the next year, a company
that employs several thousand people in this 23. For a function f with inputs x and y, explain from
fx
community goes out of business, thereby a graphical viewpoint what the algebraic ratio fy
increasing unemployment and lowering the represents.
SUMMARY
Multivariable Functions and Contour Graphs When considering the rate of change of a multivari-
In applications where output depends on two or more able function, we must consider the direction of move-
input variables, multivariable functions occur. A multi- ment. Rates of change of cross-sectional models describe
variable function with two input variables is graphed us- how the function is changing when moving on the
ing a three-dimensional coordinate system; it can also be three-dimensional graph of the function in directions
graphed in two dimensions by contour curves repre- parallel to the two input axes.
senting constant values of output.
Partial Derivatives
Multivariable Functions and Rates of Change
When a function f has two input variables x and y, we
Cross-sectional models of a multivariable function with f f
two input variables can be obtained by modeling one use the symbols x and y to represent the partial rates of
column (or row) of the data given in a multivariable change and to indicate the direction of movement. Par-
table. Geometrically, cross-sectional models describe tial derivatives of multivariable functions with respect to
the curves that result when cross-sectional planes cut a a particular input variable are found by treating all the
three-dimensional graph of the function with two input other input variables as constants and computing the
variables parallel to one of the two input axes. derivative with respect to the changing variable.
CHAPTER 9 Concept Check 609
Compensating for Change variables x and y, the change is represented by the ap-
In situations with fixed output and changing input, the proximation formula
adjustment needed in one input variable to compensate
fx
for small changes in the other input variable can be esti- y x
fy
mated using the slope of a line tangent to the fixed con-
tour curve. Algebraically, for a function f with input The partial derivatives in this formula are evaluated at
the point from which the change occurs.
CONCEPT CHECK
Can you To practice, try
• Interpret multivariable function inputs and outputs? Section 9.1 Activities 3, 5
• Draw contour graphs on tables? Section 9.1 Activities 9, 11
• Draw contour graphs using equations? Section 9.1 Activities 13, 17
• Answer questions using contour graphs? Section 9.1 Activities 27, 29
• Find and interpret cross-sectional models? Section 9.2 Activities 5, 9
• Find rates of change of cross-sectional models Section 9.2 Activities 17, 19
derived from tables?
• Interpret partial rates of change? Section 9.3 Activities 5, 7
• Estimate partial rates of change using tables and Section 9.3 Activities 11, 13, 17
cross-sectional models?
• Calculate first and second partial derivatives using Section 9.3 Activities 11,13,17
an equation?
• Calculate the slope at a point on a contour curve? Section 9.4 Activities 5, 15
• Estimate compensating change? Section 9.4 Activities 19, 21
610 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
CONCEPT REVIEW
1. Apples The storage time of apples and the tem- 2. Insect Eggs A model for the number of eggs laid
perature at which they are blanched when making per female in 30 days for a certain insect is
applesauce affect the consistency of the applesauce.
E(t, h) 4191.6877 299.7038t 23.1412h
The consistometer table gives the consistency values
for applesauce as a function of the number of 5.2210t2 0.0937h2 0.4023th eggs
months the raw apples were stored and the temper- where t is the temperature in degrees Celsius and h
ature at which they were blanched. The consistome- is the percentage relative humidity.
ter value is a measure of how far (in centimeters) an (Source: J. A. Morales-Ramos, S. M. Greenberg, and E. G. King,
amount of applesauce flows down a vertical surface “Selection of Optimal Physical Conditions for Mass Propaga-
in 30 seconds. C(t, s) denotes the consistometer tion of Catolaccus grandi,” Environmental Entomology, vol. 25,
value for a storage time of s months and a blanch no. 1 (February 1996), pp. 165–173.)
temperature of t°C. a. Find and interpret E(24, 60).
E
b. Find and interpret h for a temperature of 27°C
and relative humidity 77%.
Consistometer values (cm) for Rome applesauce c. A researcher wishes to hold the egg production
constant but desires to investigate the change in
Storage time Temperature (Celsius)
humidity needed to compensate for an increase
(months) 35° 47° 59° 71° 83° in temperature. Find a formula for the derivative
0 3.0 2.8 2.6 2.6 2.8 that is most useful to this researcher.
1 3.3 3.1 2.8 2.8 3.0 d. If the researcher is currently conducting an ex-
2 3.5 3.2 3.0 2.9 3.2
periment with temperature 25°C and humidity
63%, what approximate change in humidity will
3 3.4 3.2 3.0 2.9 3.2
be needed to compensate for a decrease in tem-
4 3.2 2.9 2.7 2.7 3.0 perature of 0.5°C while maintaining the current
egg production level?
(Source: Based on information in A. M. Godfrey Usiak, M. C.
Bourne, and M. A. Rao, “Blanch Temperature/Time Effects on e. Discuss the graphical interpretation of the an-
Rheological Properties of Applesauce,” Journal of Food Science, vol. swer to part d.
60, no. 6 (1995), pp. 1289–1291.)
3. Insect Eggs The figure shows a contour graph for
the function described in Activity 2. Use the con-
tour graph to answer the questions that follow.
a. Draw an input/output diagram for C.
h
b. Find and interpret C(71, 3). (% humidity)
c. Find an appropriate cross-sectional model, and
use it to estimate the consistometer value of ap- 80
49
0
plesauce made from apples that were stored for 2
480
74
470
weeks and blanched at 35°C. 68
C
d. Interpret the notation t s4.
460
62
e. Use a cross-sectional model to estimate the value 56 A
of the partial derivative in part d for t 45°C. t
50
(°C)
Interpret your answer. 22 24 26 28 30
f. Sketch contour curves on the table for consis- a. Estimate the inputs and output associated with
tometer values of 2.8, 3.0, and 3.2 point A. Interpret your answers.
CHAPTER 9 Concept Review 611
b. At point A, which will result in a greater change d. Estimate the slope of the line tangent to the con-
in the number of eggs laid: an increase in tem- tour graph at t 24 and h 62.
perature of 3°C or an increase in humidity of 3 e. Calculate the answer to part d by using the equa-
percentage points? tion given in Activity 2. Interpret your answer.
c. At point A, will the number of eggs laid de-
crease more quickly when h decreases or when t 4. Insect Eggs Construct a second partials matrix for
decreases? the function in Activity 2.
Project 9.1 Competitive and Complementary Products
Setting Tasks
Competitive and complementary products play an
1. Let f(x, y) be the demand for a certain automobile
important role in today’s business markets. Two
sold for x dollars by Ford, and let h(x, y) be the
commodities are called competitive if an increase in
demand for a similar automobile sold for y dollars
demand for one results in a decrease in demand for
by Honda. If the total number of these two types
the other. Two different brands of the same item are
of automobiles remains constant, explain why the
competitive. For example, Pepsi and Coke products
automobiles are competitive if
in side-by-side vending machines are competitive. In
this case, when the demand for Pepsi increases, the f h
0 and 0
demand for Coke decreases, and when the demand y x
for Coke increases, the demand for Pepsi decreases. It
2. Give four examples of commodities you believe
is important to note that the driving force in the
would be competitive in today’s market.
change in the demands for competitive products is
often the increase (or decrease) in the prices of the 3. Let D(a, b) be the demand for product A that sells
products. That is, when the price of Pepsi increases, for a dollars, and let F(a, b) be the demand for
the demand for Pepsi will decrease and the demand product B that sells for b dollars. If products A and
for Coke will increase. Likewise, when the price of B are complementary, restate the definition of
Coke increases, the demand for Coke will decrease complementary products in terms of their demand
and the demand for Pepsi will increase. functions, using partial derivative notation.
Sometimes products are used together because
4. Give four examples of products you believe would
they complement each other (coffee and cream,
be complementary in today’s market.
computers and CDs, DVD players and DVDs). Two
products are called complementary when both lose
sales if one raises prices. Reporting
Prepare a written report containing detailed answers,
including explanations, for the preceding tasks.
612
Project 9.2 Expert Witness
Setting were still in the vehicle. Therefore, the investiga-
tor’s ability to predict the velocity of the vehicle ac-
Investigators at accident scenes often wish to
curately is affected by the rate of change of velocity
determine the speed at which a vehicle was traveling
with respect to weight. Find a formula for the rate
before the driver applied the brakes. Normally, all an
of change of the velocity of a vehicle with respect to
investigator has to go on is the length of the skid
its weight. How quickly will the velocity of the
marks and the weight of the vehicle that caused the
3145-pound vehicle that made skid marks 112 feet
skid marks. The investigator for an involuntary
long be changing when its weight increases?
manslaughter lawsuit has called you as an expert
witness to testify regarding the probable speed of the 3. The investigator knows that the vehicle at the time
vehicle at the time of the accident. of the accident contained one male driver 6 feet 1
inch tall and of average build, one female
passenger 5 feet 5 inches tall and of average build,
Tasks one 2-year-old child described as “chubby,” and a
child safety seat. Discuss whether the weight of the
1. Under normal driving conditions, if a person driver and passengers would significantly affect the
driving a vehicle on an asphalt road slams on the calculation of the velocity of the vehicle in the case
brakes and skids to a stop, the length of the skid under investigation. Note that the accident
marks can be modeled by S(v, w) 0.000013wv2 occurred in a 50-mph zone.
feet, where the vehicle was traveling v miles per
hour when the driver slammed on the brakes. The
vehicle weighs w pounds. Reporting
Suppose the investigator measures the skid
marks and weighs the vehicle that made them. He 1. Prepare a document, to be introduced as evidence
finds that the skid marks are 112 feet long and the in court, summarizing your analysis of the
vehicle weighs 3145 pounds. Determine the veloc- vehicle’s velocity. This report should be technical
ity at which the vehicle must have been traveling if in nature, and your conclusions should be clearly
it was empty. stated.
2. It is possible that when it was weighed, the vehicle 2. Prepare a witness statement to be given under
did not contain everything that was in it when the oath. You will be cross-examined by the defense
skid marks were created. For instance, it is ex- attorney and should be prepared to justify your
tremely unlikely that the driver and passengers assumptions and conclusions.
613
10 Analyzing Multivariable
Change: Optimization
Concepts Outline
10.1 Multivariable Critical Points
10.2 Multivariable Optimization
10.3 Optimization Under
Constraints
10.4 Least-Squares Optimization
Philip and Karen Smith/Getty Images
Concept Application
Honey is composed of sugars (fructose, glucose, and maltose) and moisture content in-
fluenced by the floral source and variations in the climate where the honey was gathered.
The percentages of sugars and moisture affect the texture of the honey. Measures of the
texture of honey include cohesiveness and adhesiveness. With the ability to model
multivariable data, food scientists are able to answer such questions as
• What is the minimum cohesiveness possible?
• What is the maximum adhesiveness possible?
• If the FDA restricts the percentage of sugars and moisture in Grade A honey, how does
this affect the smallest cohesiveness and largest adhesiveness possible if honey is to be
labeled Grade A?
You will have the opportunity to answer questions similar to these in Section 10.3.
614
10.1 Multivariable Critical Points 615
Chapter Introduction
The optimization techniques for functions with a single input variable readily gener-
alize to multivariable functions. In the same way that derivatives play an important
role in determining critical points of a function with a single input variable, partial de-
rivatives are used for locating critical points of multivariable functions. Critical points
of functions of two input variables include maxima, minima, and saddle points.
We investigate methods for unconstrained optimization where the variables are
free to change in any direction. We also consider constrained optimization, where
there are limits on the values of the variables involved. We conclude with an applica-
tion of optimization that is related to how we began: finding the equation of the line
that best fits a set of data points.
Concepts You Will Be Learning
• Estimating critical points using tables and contour graphs (10.1)
• Estimating absolute extrema using tables and contour graphs (10.1)
• Finding critical points using equations (10.2)
• Classifying critical points using the Determinant Test (10.2)
• Solving constrained optimization using Lagrange multipliers (10.3)
• Classifying optimal points subject to a constraint (10.3)
• Interpreting the Lagrange multiplier (10.3)
• Using multivariable optimization to find a best-fit linear model (10.4)
10.1 Multivariable Critical Points
In Chapter 4 we considered relative maxima and minima of single-variable functions.
Now we consider similar points on multivariable functions—maxima, minima, and
saddle points.
Critical Points
Figure 10.1 shows a three-dimensional graph with a relative maximum (an output
value greater than any of those around it). Figure 10.2 shows a three-dimensional
graph with a relative minimum (an output value smaller than any of those around it).
Relative
maximum z
z
y
x
y Relative
x minimum
FIGURE 10.1 FIGURE 10.2
616 CHAPTER 10 Analyzing Multivariable Change: Optimization
If for a given input region, there are no output values greater than a certain rela-
tive maximum, then we say that the relative maximum is also an absolute maximum
in that region. Similarly, if for a given input region, there are no output values less
than a certain relative minimum, then that relative minimum is called an absolute
minimum in that region. Figure 10.3 shows a graph that contains relative maxima at
points A and B and an absolute maximum at point B. A saddle point is a point that
corresponds to a relative maximum of a cross section in one direction and to a rela-
tive minimum of a cross section in another. Figure 10.4 shows a function with a sad-
dle point. A three-dimensional graph with a saddle point resembles a saddle in the
area around that point. Note that the function in Figure 10.3 also has a saddle point
in the valley between the points labeled A and B.
Absolute maximum
(as well as relative maximum)
z Saddle
point B
Relative A
maximum
y
x
FIGURE 10.3
Saddle
point
x
y
FIGURE 10.4
Saddle points and points corresponding to relative maxima and minima are
called critical points (or stationary points). They are identified, as we will see in the
next section, when partial rates of change are zero. Absolute extrema that occur on
edges of tables or graphs may not fit this description.
Locating Critical Points on Contour Graphs
Contour graphs that correspond to the graphs in Figures 10.1 through 10.4 are shown
in Figures 10.5 through 10.8, respectively.
10.1 Multivariable Critical Points 617
y y
0.1
0.5 9.5
0.9
7.5
1.3
5.5
1.7 3.5
x x
FIGURE 10.5 A maximum point on a contour graph FIGURE 10.6 A minimum point on a contour graph
y y
1.6
5
6 1.2
7 7 8 9
A C 10 B 0.4 0.8 0.8 0.4
1.2
1.6
x x
FIGURE 10.7 Relative and absolute maxima and saddle FIGURE 10.8 A saddle point on a contour graph
point on a contour graph
Note that on the contour graph of a continuous function, relative maximum and
minimum points lie within a simple closed contour (provided the contour graph is de-
tailed enough). A simple closed contour is a contour that completely encloses a region
but never crosses over itself. (See Figure 10.9.)
Three examples of simple closed contours Three curves that are not simple
closed contours
FIGURE 10.9
Relative maximum and minimum points lie in the center of a group of concentric
simple closed contours. (See Figures 10.5, 10.6, and 10.7.)
Saddle points are also fairly easy to locate on contour graphs. The contour curves
near the saddle point (but not through the saddle point) are all curved away from the
point. (See Figure 10.8 and the contours near C in Figure 10.7). As we move away from
the saddle point in certain directions, the contours we cross show increasing out-
618 CHAPTER 10 Analyzing Multivariable Change: Optimization
put levels; as we move away from the saddle point in other directions, the contours
show decreasing output levels.
Locating Critical Points in Tables
When given only a table of data, we can use contour curves to identify critical points
within the table. We simply sketch contour curves on the table for several equally
spaced output values and then apply the methods for locating critical points on con-
tour graphs.
Example 1 illustrates using contour curves to estimate a critical point in a table.
EXAMPLE 1 Locating Extreme Points by Sketching Contour Curves
Postal Rates To receive “bound printed matter” rates from the U.S. Postal Service,
items mailed must consist of advertising, promotional, directory, or editorial mate-
rial. The maximum weight for bound printed matter is 15 pounds. Rates are based on
weight, shape, and distance. The maximum size is 108 inches in combined length and
distance around the thickest part. (For simplicity, we consider the distance around
the thickest part of the package to be twice the width plus twice the height.) Table 10.1
shows some values of the volume of a rectangular package that contains the maxi-
mum amount of printed material according to the size requirement. All of the criti-
cal points lie within the region represented by the table.
TABLE 10.1 Volume of “Bound Printed Matter” Package
Height (inches)
3 6 9 12 15 18 21 24 27
3 864 1620 2268 2808 3240 3564 3780 3888 3888
6 1620 3024 4212 5184 5940 6480 6804 6912 6804
Width (inches)
9 2268 4212 5832 7128 8100 8748 9072 9072 8748
12 2808 5184 7128 8640 9720 10,368 10,584 10,368 9720
15 3240 5940 8100 9720 10,800 11,340 11,340 10,800 9720
18 3564 6480 8748 10,368 11,340 11,664 11,340 10,368 8748
21 3780 6804 9072 10,584 11,340 11,340 10,584 9072 6804
24 3888 6912 9072 10,368 10,800 10,368 9072 6912 3888
27 3888 6804 8748 9720 9720 8748 6804 3888 0
(Source: Based on information at www.usps.com, accessed 1/4/2003.)
Sketch contours on the table at volumes of 8000, 9000, 10,000, and 11,000 cubic
inches, and mark the location of any approximate critical points with an X. Identify
each critical point as a relative maximum, a relative minimum, or a saddle point.
10.1 Multivariable Critical Points 619
Solution
We sketch contour curves on the table as shown in Figure 10.10. From the contour
graph and the given information, we conclude that there is only one critical point.
Height (inches)
3 6 9 12 15 18 21 24 27
3 864 1620 2268 2808 3240 3564 3780 3888 3888
6 1620 3024 4212 5184 5940 6480 6804 6912 6804
8000
Width (inches)
9 2268 4212 5832 7128 8100 8748 9072 9072 8748
10,000 9000
12 2808 5184 7128 8640 9720 10,368 10,584 10,368 9720
11,000
15 3240 5940 8100 9720 10,800 11,340 11,340 10,800 9720
18 3564 6480 8748 10,368 11,340 11,664
X 11,340 10,368 8748
21 3780 6804 9072 10,584 11,340 11,340 10,584 9072 6804
24 3888 6912 9072 10,368 10,800 10,368 9072 6912 3888
27 3888 6804 8748 9720 9720 8748 6804 3888 0
FIGURE 10.10
Because this point (located near height 18 inches and width 18 inches) lies
within concentric closed contours and the values of the contour curves decrease in all
directions away from it, the point is the approximate location of a relative maximum.
Note that this method gives 11,664 cubic inches as only an approximation of the max-
imum volume because the table does not show every possible input and output of the
function. ●
Relative maxima and relative minima occur at the center of concentric closed
contour curves. However, saddle points occur where the contours appear to bend
outward away from a central point. Example 2 illustrates using contour curves to es-
timate a saddle point on a table.
EXAMPLE 2 Locating Saddle Points by Sketching Contour Curves
TV News The Nielsen Media Research ratings for television programs show the
percentage of households in a certain broadcast market tuned to a specific program.
Many television programs use the Nielsen ratings as a measure of how well they are
doing in the market. The Project for Excellence in Journalism is an independent re-
search project that has been set up to analyze the quality of local TV news broadcasts.
Preliminary findings from this project show that several factors impact a local TV
news broadcast’s Nielsen rating. In addition to documenting the influence of the
number of viewpoints presented, the amount of local news covered, the number of
stories covered, the depth of the stories covered, the number and quality of the
sources consulted, and the size of the news team, the project for Excellence in Jour-
nalism found that broadcasts with a mixture of tabloid news and serious news are not
as likely to increase their Nielsen rating as broadcasts that are dedicated to one form
of news. The preliminary findings also suggest that local news teams may be too un-
derstaffed to deliver quality reporting. Suppose that Table 10.2 shows the probability
that a local news broadcast’s Nielsen rating will be rising, given the proportion of
tabloid news on the broadcast and the amount of time (on average) that a reporter
has to research the story.
620 CHAPTER 10 Analyzing Multivariable Change: Optimization
TABLE 10.2 Probability of rise in Nielsen ratings
Research time (days)
0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2
0.1 0.15 0.22 0.28 0.34 0.38 0.41 0.42 0.43 0.42
Proportion of broadcast
that is tabloid news
0.2 0.12 0.19 0.25 0.30 0.33 0.35 0.36 0.36 0.35
0.3 0.11 0.18 0.23 0.27 0.30 0.32 0.32 0.32 0.30
0.4 0.12 0.18 0.23 0.26 0.29 0.30 0.30 0.29 0.26
0.5 0.15 0.20 0.24 0.27 0.29 0.30 0.29 0.27 0.24
0.6 0.19 0.24 0.27 0.30 0.31 0.31 0.30 0.27 0.24
0.7 0.25 0.29 0.32 0.34 0.35 0.34 0.32 0.29 0.25
0.8 0.33 0.36 0.39 0.40 0.40 0.39 0.36 0.33 0.28
0.9 0.42 0.45 0.47 0.47 0.47 0.45 0.42 0.38 0.33
Use contour curves to help estimate the critical point in Table 10.2.
Solution
We sketch contour curves corresponding to probabilities of 0.20 through 0.40 at in-
tervals of 0.05 on the table, as shown in Figure 10.11.
Research time (days)
0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2
0.20 0.25
0.40
When two or more adjacent 0.1 0.15 0.22 0.28 0.34 0.38 0.41 0.42 0.43 0.42
Proportion of broadcast
entries are equal, we estimate 0.35
0.2 0.12 0.19 0.25 0.30 0.33 0.35 0.36 0.36 0.35
that is tabloid news
the location of the critical value. 0.30
0.3 0.11 0.18 0.23 0.27 0.30 0.32 0.32 0.32 0.30
0.4 0.12 0.18 0.23 0.26 0.29 0.30 0.30 0.29 0.26
X 0.25
0.5 0.15 0.20 0.24 0.27 0.29 0.30 0.29 0.27 0.24
0.6 0.19 0.24 0.27 0.30 0.31 0.31 0.30 0.27 0.24
0.7 0.25 0.29 0.32 0.34 0.35 0.34 0.32 0.29 0.25
0.8 0.33 0.36 0.39 0.40 0.40 0.39 0.36 0.33 0.28
0.9 0.42 0.45 0.47 0.47 0.47 0.45 0.42 0.38 0.33 0.30
0.40 0.35
FIGURE 10.11
There appears to be a saddle point just above and slightly to the right of the table
entry 0.30 corresponding to 1.6 days of research and 50% tabloid news content.
(See the X in Figure 10.11.) Note that the outputs on the contour curves decrease
to the left and right of this value and increase above and below it. The value 0.30 is
a maximum in the row in which it appears but is a relative minimum in its column.
A table value that appears to be a relative maximum when approached from one di-
rection and a relative minimum when approached from another direction is iden-
tified as a saddle point. ●
10.1 Multivariable Critical Points 621
Note in Figure 10.11 that the entry 0.42 corresponding to 0.6 day research time
and 90% tabloid news content in the broadcast is the largest value in the 0.6-day
column and is smaller than nearby values in the 0.90 proportion tabloid news row.
We do not classify this point as a saddle point because it lies on the edge of the
table. In order to identify a saddle point visually in a table or on a contour graph,
we must be able to identify a point that is a maximum in one direction and a min-
imum in another direction. We cannot visually identify relative extrema of cross
sections on the edges of tables or contour graphs, because we cannot observe the
edges lying within concentric closed contours. Similarly, we cannot visually identify
saddle points on edges.
Relative extreme points and saddle points cannot be visually identified on
the edges of tables or contour graphs.
Sketching contour curves on a table is very helpful when identifying critical points.
Estimating Absolute Extrema Using Tables and Contour Graphs
Recall from Chapter 4 that in order to find absolute extrema of a continuous function
over a specified interval, we first find all relative extrema and then compare the out-
puts of the relative extrema with the function outputs at the endpoints. If we are not
given an interval over which to find the absolute extrema, then we must also consider
the end behavior of the function. This process can be extended to find absolute ex-
trema of a continuous function with two input variables (a three-dimensional func-
tion). We first locate all relative extrema (which lie within concentric closed contour
curves), and we then compare the output values of the relative extrema with any out-
put values on the edges of the function. Some functions are restricted on the basis of
the context or the nature of the function. Some functions have no edges—that is, they
extend infinitely in all directions. In this case, we must also consider the end behav-
ior of the function.
When using tables and contour graphs to estimate absolute extrema for continu-
ous, three-dimensional functions, we must keep in mind the following steps:
• Consider any restrictions on output suggested by the context.
• Sketch several contour curves on the table.
• Visually locate all relative extrema. In a table, relative extrema will be values that
are smaller or larger than all eight values around them. Relative extrema cannot
be identified on the edges of a table or contour graph, because to be visually iden-
tified, they must lie within concentric closed contour curves.
• Determine whether the edges of the table or contour graph can extend beyond
what is shown. We call any edges beyond which the table or graph cannot extend
terminal edges. Terminal edges are determined by the context of the situation that
the function describes.
622 CHAPTER 10 Analyzing Multivariable Change: Optimization
• Determine absolute extrema by comparing relative extrema and values on termi-
nal edges of the table or graph and by considering the behavior of the function
beyond the nonterminal edges. As with functions of one input variable, all absolute
extrema will either be relative extreme points or lie on terminal edges of the graph.
We illustrate these steps in Example 3.
EXAMPLE 3 Locating Absolute Extrema in a Table of Data
Cheese Spread Table 10.3 shows a measure of the consistency of cheese spread as
a function of the percent of salt and the percent of glycerol used in processing. All of
the critical points lie within the region represented by the table.
TABLE 10.3 A measure of the consistency of cheese spread
Percent glycerol
0 1 2 3 4 5 6 7 8 9 10
0 8028 6412 5078 4026 3256 2768 2562 2638 2996 3636 4558
0.3 9153 7593 6315 5319 4604 4172 4022 4154 4568 5263 6241
0.6 9953 8448 7226 6286 5627 5251 5156 5344 5814 6565 7599
1 10513 9083 7935 7069 6485 6183 6163 6425 6969 7795 8903
Percent salt
1.3 10553 9179 8087 7276 6748 6502 6538 6856 7455 8337 9501
1.6 10268 8950 7913 7159 6686 6496 6588 6961 7617 8554 9774
2 9382 8138 7176 6496 6098 5982 6148 6596 7326 8338 9632
2.3 8338 7150 6243 5619 5277 5217 5439 5942 6728 7796 9146
2.6 6968 5836 4985 4417 4131 4126 4404 4963 5805 6929 8334
3 4636 3578 2802 2308 2096 2166 2518 3152 4068 5266 6746
3.3 2507 1505 785 347 191 316 724 1414 2386 3640 5175
(Source: E. Kombila-Moundounga and C. Lacroix, “Effet des Combinaisons de Chloure de Sodium, de Lactose et de Glycerol sur les
Caractéristiques Rhéologiques et la Couleur des Fromages Fondus à Tartiner,” Canadian Institute of Food Science and Technology Journal, vol. 24,
no. 5 (1991), pp. 239–251.)
a. Sketch contour curves on Table 10.3 for consistency values of 4000, 5000, 6000,
7000, 8000, and 9000. Mark the approximate saddle point and the largest and
smallest table values.
b. Use the table and contour graph to estimate relative and absolute maximum and
minimum consistency measures.
Solution
a. In Figure 10.12, the requested contour curves are sketched, and boxes are drawn
around the largest and smallest table values. The approximate saddle point is
circled.
10.1 Multivariable Critical Points 623
Percent glycerol
800 7000 0 1 2 3 4 5 6 7 8 9 10 000
4 000
9 0 5
0 000 8028 6412 5078 4026 3256 2768 2562 2638 2996 3636 4558 0
600
0
0.3 9153 7593 6315 5319 4604 4172 4022 4154 4568 5263 6241 700
8000
0.6 9953 8448 7226 6286 5627 5251 5156 5344 5814 6565 7599 000
9
1 10513 9083 7935 7069 6485 6183 6163 6425 6969 7795 8903
Percent salt
1.3 10553 9179 8087 7276 6748 6502 6538 6856 7455 8337 9501
1.6 10268 8950 7913 7159 6686 6496 6588 6961 7617 8554 9774
2 9382 8138 7176 6496 6098 5982 6148 6596 7326 8338 9632
2.3 8338 7150 6243 5619 5277 5217 5439 5942 6728 7796 9146
2.6 6968 5836 4985 4417 4131 4126 4404 4963 5805 6929 8334
3 4636 3578 2802 2308 2096 2166 2518 3152 4068 5266 6746
3.3 2507 1505 785 347 191 316 724 1414 2386 3640 5175 60
40 500 00
00 0
FIGURE 10.12
Note that the 6502 entry corresponding to 5% glycerol and 1.3% salt is an
approximate saddle point because it is a maximum in the 5% glycerol column and
a minimum in the 1.3% salt row.
We cannot consider the values b. Because there are no closed contour curves indicated by Figure 10.12, we
4558 (upper right corner) and
2507 (lower left corner) as
conclude that the function has no relative extrema. Thus any absolute extrema
candidates for absolute must occur on terminal edges of the function. Note that the table cannot extend
extrema, even though they lie in beyond the top and left edges because negative percentages do not make sense in
the terminal edges along the this context. Thus these edges are terminal edges, and we can look in these two
top row and left column of
Figure 10.12, because these
edges for approximate absolute extrema. We must also consider the behavior of
terminal edges can extend the function beyond the lower and right edges.
down and to the right. In order The greatest value in the table is approximately 10,553, and it occurs when the
for a corner entry in a table to percentages of glycerol and salt are approximately 0 and 1.3, respectively. This
be considered in the search for
absolute extrema, both the row value is on a terminal edge; however, the contour curves suggest that the output
and the column in which the values will continue to increase for glycerol percentages greater than 10%. If the
entry lies must be terminal table were extended far enough to the right, we would probably see an output
edges. value greater than 10,553. We conclude that we do not have enough information
to identify an absolute maximum.
The smallest value in the table is 191, found in the bottom row. However, if
the table were extended downward for percentages of salt greater than 3.3%, we
would probably observe an absolute minimum of zero. The places where this
minimum occurs are difficult to estimate from the information presented in
the table. ●
Example 1 in this section discussed the volume of packages sent at the “bound printed
matter” postal rate. Neither the inputs nor the output can be negative in this context, so
0 cubic inches is the absolute minimum volume. Note that in this example, we were
given the information that all critical points lie within the region represented by Table
624 CHAPTER 10 Analyzing Multivariable Change: Optimization
10.1. We can therefore conclude that the relative maximum volume of approximately
11,664 cubic inches is the absolute maximum volume.
Three-dimensional graphs, contour graphs, and tables are helpful in estimating
critical points and absolute extrema. However, we are limited to approximations of
these points when using graphs and tables and sometimes do not have enough infor-
mation to identify absolute extrema. In the next section, we will explore a more ac-
curate method of finding critical points and absolute extrema for three-dimensional
functions.
10.1 Concept Inventory Practicing Concepts
3. Is the point at x 4.885 and y 1.382 on the con-
• Critical points: relative maxima, relative minima,
tour graph a relative maximum point, a relative
and saddle points
minimum point, or a saddle point? Explain.
• Absolute maxima, absolute minima
y
• Locating critical points and absolute extrema in ta-
4
bles and on contour and 3-D graphs
290
• Terminal edges of a table or graph 3 300
310
• Interpreting critical points and absolute extrema 2 320
1
10.1 Activities 0 x
2 3 4 5 6 7
Discussing Concepts
1. Explain how to determine whether a table of data 4. Is the point at x 4.885 and y 1.382 on the
has each of the following: contour graph a relative maximum point, a relative
a. Relative maximum minimum point, or a saddle point? Explain.
b. Relative minimum
c. Saddle point y
d. Absolute maximum 10
e. Absolute minimum 9
70
8
2. Explain, illustrating with graphs, how to determine 60
7 30
whether a contour graph has each of the following:
6 55
a. Relative maximum 5 40
b. Relative minimum 4 45
c. Saddle point 3 50
2 55
d. Absolute maximum 45
1 60
e. Absolute minimum 40
0 x
0 1 2 3 4 5 6 7 8 9 10
10.1 Multivariable Critical Points 625
5. Consider the contour graph and a three-dimensional 6. Consider the following contour graph and three-di-
graph for a function R with inputs g and h. mensional graph for a function T with inputs p and f
a. Mark on the contour plot and on the three- T(p, f)
dimensional graph the approximate locations of
all critical points. Label each point as a relative 150
maximum point, a relative minimum point, or a
saddle point. 100
b. On the basis of the graphs, approximate the in- 50
puts and output of each critical point.
0
2 8
4 6
3-D of R (g, h) 6 4
p 8 2 f
h
10 .
6
10 3-D of T (p, f)
5 25 f
4 40
85 8 60
70 40
3 55 55 100 80 80 100
40
70 6
2 60
85
25 40
1
4
10
0 g
0 2 4 6 8
2
Contours of R (g, h) 0 p
0 2 4 6 8 10
R(g, h)
Contours of T (p, f)
100
a. Mark on the contour plot and on the three-
80
dimensional graph the approximate locations of
60 all critical points. Label each point as a relative
maximum point, a relative minimum point, or a
40
saddle point.
20 b. On the basis of the graphs, approximate the
0
inputs and output of each critical point.
6
0
2 4
4 h
g 6 2
8
626 CHAPTER 10 Analyzing Multivariable Change: Optimization
Applying Concepts n N
(miles) W E
7. Consider the contour graph and associated three- S 796
1.4 794
dimensional graph of z(x, y). Is the point at which 797
x 1.8 and y 1.5 a relative maximum point, a 1.2
795
relative minimum point, or a saddle point? Explain. 798
1.0
0.8 800 799
y
801 800
3 5 0.6
10 802
2.5 15 0.4
30
20 0.2
2
25 e
0
1.5 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 (miles)
1 20
Contours of elevation
25
0.5 15
10
0 x
0 1 2 3 4
E(e, n)
Contours of z(x, y) (feet)
803
z
1.5
35
793 1.0
30 0 n
0.5 0.5 (miles)
25
e 1.0
20 (miles) 1.5
15
3-D of elevation
10
9. Lettuce The table shows the average price of ice-
5
berg lettuce, in cents per pound, in U.S. cities for se-
0 2.5 3 lected months and years.
1 1.5 2
0 1 2 3 4 0.5 y
x Price of iceberg lettuce (cents per pound)
3-D of z(x, y)
February March April May June
8. Farmland The contour graph and the three-di-
1994 54 61 53 54 52
mensional graph represent the elevation of a tract of
farmland where elevation is measured in feet above 1995 60 61 134 145 80
sea level. Identify each of the following input values 1996 59 65 65 61 67
as corresponding to a relative maximum, a relative 1997 59 61 67 60 59
minimum, a saddle point, or none of these.
1998 64 70 84 88 71
a. e 0.9 and n 0.5
1999 66 77 75 69 65
b. e 0.5 and n 0.5
(Source: Bureau of Labor Statistics.)
c. e 1.5 and n 1.5
Locate all critical points in the table. Identify each
point as a relative maximum point, a relative mini-
mum point, or a saddle point.
10.1 Multivariable Critical Points 627
10. Bananas The table shows the average price of ba- day) is shown in the table as a function of the month
nanas, in cents per pound, in U.S. cities for selected and the number of degrees of latitude.
months and years. Locate all critical points in the a. Is it possible for the table to extend in any direc-
table. Identify each point as a relative maximum tion? Explain.
point, a relative minimum point, or a saddle point.
b. Sketch contour curves on the table for 5.2, 6.2,
Price of bananas (cents per pound) 7.2, 8.2, and 9.2 kW-h/m2/day.
c. Use the contours to locate all critical points in
May June July August Sept.
the table.
1994 47 45 45 48 45
d. Locate the largest and smallest values in the table.
1995 48 49 52 51 49 Write a sentence interpreting these values.
1996 51 50 50 48 46 e. Use the table to estimate the absolute maximum
1997 48 49 49 48 46 and minimum monthly average global radiation
1998 51 51 53 49 48 levels. Are these answers the same as those you
found in part d?
1999 49 50 49 49 48
f. Under what conditions will the largest and
(Source: Bureau of Labor Statistics.) smallest values in a table also be the absolute
extrema of the function represented by the
11. Radiation The monthly average global radiation
table?
(measured in kilowatt-hours per square meter per
TABLE FOR ACTIVITY 11 Monthly and yearly means of the diurnal sums of the global radiation, in kW-h/m2 per day
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
North Pole — — — 3.1 6.9 8.9 7.9 4.9 0.80 — — —
80° — — 0.77 3.4 6.9 8.8 7.8 4.8 1.55 0.13 — —
70° — 0.51 2.0 4.5 7.1 8.5 7.8 5.5 2.9 0.95 0.06 —
60° 0.55 1.53 3.4 5.7 7.7 8.8 8.2 6.5 4.3 2.2 0.81 0.34
50° 1.66 2.8 4.7 6.7 8.4 9.1 8.8 7.4 5.5 3.4 1.97 1.35
40° 3.0 4.2 5.9 7.5 8.8 9.3 9.0 8.1 6.5 4.8 3.4 2.6
30° 4.4 5.6 6.9 8.1 9.0 9.2 9.1 8.4 7.4 6.1 4.7 4.1
20° 5.8 6.7 7.8 8.5 8.8 8.9 8.8 8.6 8.0 7.1 6.0 5.5
10° 7.1 7.7 8.3 8.5 8.4 8.3 8.3 8.4 8.3 7.9 7.2 6.8
Equator 8.1 8.5 8.6 8.3 7.8 7.5 7.6 8.0 8.4 8.4 8.2 7.9
10° 8.9 8.8 8.4 7.7 6.9 6.4 6.5 7.2 8.1 8.6 8.8 8.8
20° 9.4 9.0 8.1 6.9 5.7 5.1 5.4 6.3 7.5 8.6 9.2 9.5
30° 9.6 8.8 7.4 5.8 4.4 3.8 4.1 5.2 6.7 8.2 9.3 9.8
40° 9.6 8.3 6.5 4.6 3.1 2.5 2.7 3.9 5.6 7.5 9.1 9.9
50° 9.3 7.6 5.4 3.3 1.84 1.25 1.49 2.6 4.5 6.6 8.7 9.7
60° 8.7 6.6 4.1 2.0 0.72 0.31 0.47 1.36 3.1 5.6 8.0 9.3
70° 8.2 5.5 2.8 0.84 — — — 0.38 1.78 4.3 7.2 9.1
80° 8.2 4.7 1.42 0.099 — — — — 0.62 3.2 7.0 9.3
South Pole 8.1 4.6 0.60 — — — — — — 2.9 7.0 9.4
(Source: W. Rudolff, World-Climates, Stuttgart, Germany: Wissenschaftliche Verlagsgesellschaft, 1981.)
628 CHAPTER 10 Analyzing Multivariable Change: Optimization
12. Rainfall The table shows the probability of pre- 13. Production Expected annual yield of U.S. corn
cipitation in the U.S. corn belt being P% above or production as a percentage of the average annual
below normal and the temperature being T C yield is shown on page 629 as a function of the per-
above or below normal. Assume that if the table centage change in precipitation (above or below the
were extended in any direction, all the additional average precipitation) and the change in tempera-
entries would be zero. ture (in °C above or below the average tempera-
ture). Assume that no critical points lie outside of
Probability (expressed as a percentage) of precipitation the table.
and temperature variations
a. If a cold, wet year produces average tempera-
T P(%) tures 3°C below normal and 50% more precipi-
(°C) 40 30 20 10 0 10 20 30 40 tation than normal, what is the expected corn
yield?
4 0 0 0 0 0 0 0 0 0
b. If a hot, dry year results in 50% less precipita-
3.5 0 0 1 1 0 0 0 0 0
tion than normal and an average temperature
3 0 1 2 3 2 0 0 0 0 3°C above normal, what is the expected corn
2.5 0 1 5 7 5 1 0 0 0 yield?
2 0 2 9 15 12 4 1 0 0 c. Sketch contour curves on the table at 105%,
1.5 0 3 12 26 24 10 2 0 0 100%, 95%, and 90%.
1 0 3 15 37 40 19 4 0 0 d. Locate the absolute maximum in the table, and
write a sentence interpreting the corresponding
0.5 0 2 14 42 54 30 7 1 0
point(s).
0 0 1 11 39 59 39 11 1 0
0.5 0 1 7 30 54 42 14 2 0
14. Radiation The amount of extraterrestrial radia-
tion measured in millimeters per day equivalent
1 0 0 4 19 40 37 15 3 0
evaporation in the Northern Hemisphere is given in
1.5 0 0 2 10 24 26 12 3 0 the table as a function of the month and the number
2 0 0 1 4 12 15 9 2 0 of degrees of latitude.
2.5 0 0 0 1 5 7 5 1 0 a. According to the table, approximately how much
3 0 0 0 0 2 3 2 1 0
extraterrestrial radiation does your campus re-
ceive in September?
3.5 0 0 0 0 0 1 1 0 0
b. Locate any relative extreme points in the table.
4 0 0 0 0 0 0 0 0 0
Give the month, degrees of latitude, and
(Source: Crop Yields and Climate Change to the Year 2000, U.S. amount of radiation for the relative extreme
Department of Agriculture, 1980.) points.
c. In which directions can the table extend?
a. Sketch contour curves on the table at 10%, 20%,
30%, 40%, and 50%. d. Estimate the least and greatest amounts of extra-
terrestrial radiation in the Northern Hemi-
b. Use the contours to locate and estimate the sphere. Give the corresponding months and
absolute maximum percentage. Interpret your degrees of latitude.
answer.
10.1 Multivariable Critical Points 629
TABLE FOR ACTIVITY 13 Expected annual corn yield as a percentage of average annual yield
T P(%)
(°C) 100 90 80 70 60 50 40 30 20 10 0 10 20 30 40 50 60 70 80
6 0 7 14 18 23 27 32 40 48 56 64 67 69 72 74 75 75 75 76
5.5 0 7 15 20 25 29 34 43 51 59 67 70 73 75 78 78 78 78 78
5 0 8 16 21 26 32 37 45 54 62 71 73 76 79 81 81 81 81 81
4.5 0 8 17 22 28 34 39 48 57 65 74 77 79 82 85 85 84 84 84
4 0 9 18 24 30 36 42 51 60 68 77 80 83 86 88 88 88 87 87
3.5 0 9 19 25 32 38 45 54 62 71 80 83 86 89 92 91 91 90 89
3 0 10 20 27 33 40 47 56 65 75 84 87 90 93 96 95 94 93 92
2.5 0 10 21 28 35 42 49 59 69 78 87 91 94 96 98 97 95 94 93
2 0 11 21 29 37 44 52 63 74 82 90 95 99 99 100 98 97 96 94
1.5 0 11 22 30 38 46 54 65 76 85 93 99 101 101 102 100 99 97 95
1 0 12 23 31 40 48 56 67 78 87 95 102 103 104 104 102 100 98 96
0.5 0 12 24 33 41 50 58 69 80 90 98 104 105 106 106 104 102 100 97
0 0 12 25 34 43 51 60 71 83 92 100 107 107 108 109 106 103 101 98
0.5 0 13 26 35 43 52 61 72 84 93 101 107 108 108 107 105 102 100 97
1 0 13 27 35 44 53 61 73 85 94 103 108 109 108 106 104 101 99 96
1.5 0 14 28 36 45 53 62 73 84 93 102 107 108 107 105 103 100 98 95
2 0 14 29 37 46 54 63 73 84 93 101 105 107 106 104 102 99 97 94
2.5 0 15 29 38 46 55 63 72 82 90 98 101 103 103 103 101 98 95 93
3 0 15 30 39 47 55 64 72 80 88 96 97 99 101 102 100 97 94 92
3.5 0 15 30 38 46 54 62 70 77 85 92 94 95 97 98 96 94 91 89
4 0 15 31 38 46 53 61 68 74 81 88 90 91 93 95 92 90 88 85
4.5 0 15 31 38 45 52 59 65 72 78 85 86 88 89 91 89 87 85 82
5 0 15 31 38 44 51 58 63 69 75 81 82 84 85 87 85 83 81 79
5.5 0 16 31 37 44 50 56 61 67 72 77 79 80 82 83 81 80 78 76
6 0 16 31 37 43 49 55 59 64 69 74 75 76 78 79 78 76 75 73
(Source: Crop Yields and Climate Change to the Year 2000, U.S. Department of Agriculture, 1980.)
15. Weight The average daily weight gain of a pig b. Convert the temperatures to degrees Fahrenheit
9
(in kilograms per day) is shown in the table as a using the equation F 5( C) 32.
function of the air temperature and the pig’s c. Write a sentence interpreting the absolute ex-
weight. trema found in part a.
a. Use the table to estimate relative and absolute ex-
trema for temperatures between 4.4°C and
37.8°C and weights between 45 kilograms and
156 kilograms.
TABLE FOR ACTIVITY 14 Extraterrestrial radiation expressed in equivalent evaporation (mm/day)
Latitude Northern Hemisphere
(degrees) Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
50 3.8 6.1 9.4 12.7 15.8 17.1 16.4 14.1 10.9 7.4 4.5 3.2
48 4.3 6.6 9.8 13.0 15.9 17.2 16.5 14.3 11.2 7.8 5.0 3.7
46 4.9 7.1 10.2 13.3 16.0 17.2 16.6 14.5 11.5 8.3 5.5 4.3
44 5.3 7.6 10.6 13.7 16.1 17.2 16.6 14.7 11.9 8.7 6.0 4.7
42 5.9 8.1 11.0 14.0 16.2 17.3 16.7 15.0 12.2 9.1 6.5 5.2
40 6.4 8.6 11.4 14.3 16.4 17.3 16.7 15.2 12.5 9.6 7.0 5.7
38 6.9 9.0 11.8 14.5 16.4 17.2 16.7 15.3 12.8 10.0 7.5 6.1
36 7.4 9.4 12.1 14.7 16.4 17.2 16.7 15.4 13.1 10.6 8.0 6.6
34 7.9 9.8 12.4 14.8 16.5 17.1 16.8 15.5 13.4 10.8 8.5 7.2
32 8.3 10.2 12.8 15.0 16.5 17.0 16.8 15.6 13.6 11.2 9.0 7.8
30 8.8 10.7 13.1 15.2 16.5 17.0 16.8 15.7 13.9 11.6 9.5 8.3
28 9.3 11.1 13.4 15.3 16.5 16.8 16.7 15.7 14.1 12.0 9.9 8.8
26 9.8 11.5 13.7 15.3 16.4 16.7 16.6 15.7 14.3 12.3 10.3 9.3
24 10.2 11.9 13.9 15.4 16.4 16.6 16.5 15.8 14.5 12.6 10.7 9.7
22 10.7 12.3 14.2 15.5 16.3 16.4 16.4 15.8 14.6 13.0 11.1 10.2
20 11.2 12.7 14.4 15.6 16.3 16.4 16.3 15.9 14.8 13.3 11.6 10.7
18 11.6 13.0 14.6 15.6 16.1 16.1 16.1 15.8 14.9 13.6 12.0 11.1
16 12.0 13.3 14.7 15.6 16.0 15.9 15.9 15.7 15.0 13.9 12.4 11.6
14 12.4 13.6 14.9 15.7 15.8 15.7 15.7 15.7 15.1 14.1 12.8 12.0
12 12.8 13.9 15.1 15.7 15.7 15.5 15.5 15.6 15.2 14.4 13.3 12.5
10 13.2 14.2 15.3 15.7 15.5 15.3 15.3 15.5 15.3 14.7 13.6 12.9
8 13.6 14.5 15.3 15.6 15.3 15.0 15.1 15.4 15.3 14.8 13.9 13.3
6 13.9 14.8 15.4 15.4 15.1 14.7 14.9 15.2 15.3 15.0 14.2 13.7
4 14.3 15.0 15.5 15.5 14.9 14.4 14.6 15.1 15.3 15.1 14.5 14.1
2 14.7 15.3 15.6 15.3 14.6 14.2 14.3 14.9 15.3 15.3 14.8 14.4
0 15.0 15.5 15.7 15.3 14.4 13.9 14.1 14.8 15.3 15.4 15.1 14.8
(Source: A. A. Hanson, ed., Practical Handbook of Agricultural Science, Boca Raton, FL: CRC Press, 1990.)
TABLE FOR ACTIVITY 15 Average daily weight gain of a pig (kg/day)
Mean live Air temperature (°C)
weight (kg) 4.4 10 15.6 21.1 26.7 32.2 37.8
45 0.62 0.72 0.91 0.89 0.64 0.18
68 0.58 0.67 0.79 0.98 0.83 0.52 0.09
91 0.54 0.71 0.87 1.01 0.76 0.4 0.35
113 0.5 0.76 0.94 0.97 0.68 0.28 0.62
136 0.46 0.8 1.02 0.93 0.62 0.16 0.88
156 0.43 0.85 1.09 0.9 0.55 0.05 1.15
(Source: A. A. Hanson, ed., Practical Handbook of Agricultural Science, Boca Raton, FL: CRC
Press, 1990.)
10.1 Multivariable Critical Points 631
16. Applesauce The Food and Drug Administration 17. Applesauce The consistometer table gives the
(FDA) defines Grade A applesauce as having a consistency of applesauce as a function of the num-
consistometer value of 6.5 cm or less. It is common ber of months the raw apples were stored and the
for applesauce manufacturers to add to thick apple- temperature at which they were blanched. The con-
sauce enough water to thin it to a 6.5-cm consis- sistometer value is a measure of how far (in cen-
tometer value. The table shows the percentage of timeters) an amount of applesauce flows down a
water that can be added to applesauce while main- vertical surface in 30 seconds.
taining an FDA Grade A. This percentage of water is
a function of the number of months the apples are
stored and the blanching temperature.
Consistometer values for Rome applesauce (cm)
Storage time Temperature (°C)
(months) 35° 47° 59° 71° 83°
Percentage of water that can be added to applesauce
0 3.0 2.8 2.6 2.6 2.8
Storage time Blanching temperature (°C) 1 3.3 3.1 2.8 2.8 3.0
(months) 35° 47° 59° 71° 83° 2 3.5 3.2 3.0 2.9 3.2
0 43 48 54 53 45 3 3.4 3.2 3.0 2.9 3.2
0.5 38 44 48 47 40 4 3.2 2.9 2.7 2.7 3.0
1 35 41 45 45 38
(Source: Based on information in A. M. Godfrey Usiak, M. C.
1.5 34 40 44 44 35 Bourne, and M. A. Rao, “Blanch Temperature/Time Effects on
Rheological Properties of Applesauce,” Journal of Food Science, vol.
2 33 39 44 44 34
60, no. 6 (1995), pp. 1289–1291.)
2.5 33 38 43 43 33
3 34 40 45 45 35
3.5 35 43 47 47 38
a. Sketch contour curves on the table for consis-
4 39 45 50 50 42 tometer values of 2.7, 3.0, and 3.3.
b. Locate all critical points in the table.
c. Estimate the absolute maximum and absolute
minimum consistometer values for storage times
a. Sketch contour curves on the table for water per-
between 0 and 4 months and temperatures be-
centages of 35, 40, 45, and 50.
tween 35°C and 83°C.
b. Locate all critical points in the table.
d. If you were an applesauce manufacturer, which
c. Estimate the absolute maximum and absolute of the points found in parts b and c would be of
minimum percentage for storage times between the most interest to you? Explain.
0 and 4 months and temperatures between 35°C
and 83°C. 18. Farmland The elevation table gives elevation
d. Why would a maximum value be of interest to an E(e, n), in feet above sea level, of a tract of Missouri
applesauce manufacturer? farmland measured e miles east of the western fence
and n miles north of the southern fence.
e. If not all harvested apples can be used right away,
a. Find any relative extreme points in the table.
how long (on the basis of the table) do you think
a manufacturer should store apples before pro- b. Find the saddle point in the table.
cessing them? c. Explain how you located the saddle point.
632 CHAPTER 10 Analyzing Multivariable Change: Optimization
TABLE FOR ACTIVITY 18 Elevation (feet above sea level) e miles east of western fence and n miles
north of southern fence
e→ 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
1.5 800.3 799.1 798.3 797.8 797.5 797.4 797.5 797.6 797.6 797.7 797.6 797.3 796.8 796.0 794.8 793.2
1.4 800.8 799.6 798.8 798.3 798.0 797.9 797.9 798.0 798.1 798.1 798.1 797.8 797.3 796.5 795.3 793.7
1.3 801.2 800.0 799.2 798.7 798.4 798.3 798.4 798.5 798.5 798.6 798.5 798.2 797.7 796.9 795.7 794.1
1.2 801.6 800.4 799.6 799.1 798.8 798.7 798.7 798.8 798.9 798.9 798.9 798.6 798.1 797.3 796.1 794.5
1.1 801.9 800.7 799.9 799.4 799.1 799.0 799.1 799.2 799.2 799.3 799.2 798.9 798.4 797.6 796.4 794.8
1.0 802.2 801.0 800.2 799.7 799.4 799.3 799.3 799.4 799.5 799.5 799.5 799.2 798.7 797.9 796.7 795.1
0.9 802.4 801.2 800.4 799.9 799.6 799.5 799.6 799.7 799.7 799.8 799.7 799.4 798.9 798.1 796.9 795.3
0.8 802.6 801.4 800.6 800.1 799.8 799.7 799.7 799.8 799.9 799.9 799.9 799.6 799.1 798.3 797.1 795.5
0.7 802.7 801.5 800.7 800.2 799.9 799.8 799.9 800.0 800.0 800.1 800.0 799.7 799.2 798.4 797.2 795.6
0.6 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.1 800.1 799.8 799.3 798.5 797.3 795.7
0.5 802.8 801.6 800.8 800.3 800.0 799.9 800.0 800.1 800.1 800.2 800.1 799.8 799.3 798.5 797.3 795.7
0.4 802.8 801.6 800.8 800.3 800.0 799.9 799.9 800.0 800.1 800.1 800.1 799.8 799.3 798.5 797.3 795.7
0.3 802.7 801.5 800.7 800.2 799.9 799.8 799.9 800.0 800.0 800.1 800.0 799.7 799.2 798.4 797.2 795.6
0.2 802.6 801.4 800.6 800.1 799.8 799.7 799.7 799.8 799.9 799.9 799.9 799.6 799.1 798.3 797.1 795.5
0.1 802.4 801.2 800.4 799.9 799.6 799.5 799.6 799.7 799.7 799.8 799.7 799.4 798.9 798.1 796.9 795.3
0 802.2 801.0 800.2 799.7 799.4 799.3 799.3 799.4 799.5 799.5 799.5 799.2 798.7 797.9 796.7 795.1
n↑
Baking time
(minutes)
San N
35 Francisco
Bay
.8
103
34
Di
.2
106
abl
33
6
oR
.
108
ang
32 1
2
.0
3 e
111
31
Sa
4 5
nta
Sunnyvale
.4
113
Cr
30
uz
San José 5
Mo
29
un
tai
ns
28
27 0 mi 5
.0
111
0 km 8 5 ft = 1.5 m
26
Leavening Land subsidence in feet, 1934–1960
25
(grams) Santa Clara Valley, California
3 3.3 3.7 4 4.3 4.7 5
(Source: A. Li and C. E. Walker, “Cake Baking in Conventional, (Source: Figure from R. A. Freeze and J. A. Cherry, Groundwater,
Impingement and Hybrid Ovens,” Journal of Food Science, vol. 61, Englewood Cliffs, NJ: Prentice-Hall, 1979. Reprinted by permission
no. 1 (1996), pp. 188–192.) of Prentice-Hall, Inc. Upper Saddle River, NJ.)
Figure for Activity 20 Figure for Activity 22
10.1 Multivariable Critical Points 633
19. Therapy A technique commonly used by physical
therapists is the application of heat to muscles and MISSOURI
tissues. A contour graph showing the change in the 24-Hour Precipitation Totals
Ending 7 June 1993
temperature of a thigh model after exposure to mi- (millimeters)
crowave radiation is given. 125
50 25
B 100
75
50
A A'
25
3 2
1
Bone
Muscle
(Source: Missouri Precipitation Contour Map from S. M. Rochette and J.
Fat T. Moore, “Initiation of an Elevated Mesoscale Convective System Associ-
ated with Heavy Rainfall,” Weather and Forecasting, vol. 11 (1996), p. 444.
B'
Used by permission of the American Meteorological Society.)
Increase in temperature (°C) of a thigh model
(Source: F. J. Kottke and J. F. Lehman, Krusen’s Handbook
of Physical Medicine and Rehabilitation, Philadelphia:
Saunders, 1990.) a. Identify the Missouri and Mississippi Rivers on
the map.
a. Mark the approximate location of any relative
and absolute extrema, and estimate the output b. Mark the area of greatest rainfall. Estimate the
value at each point. amount of rain that fell on June 6 in the area you
marked.
b. Are the points you marked maxima or minima?
c. Identify any other critical points on the contour
20. Baking The contour graph shows the volume in- map, and classify them as relative maxima, rela-
dex of a cake baked at 350°F, given the baking time tive minima, or saddle points.
(in minutes) and the amount of leavening used (in
grams). An index of 100 corresponds to the volume 22. Land Subsidence Even in desert climates, the
of the batter. No critical points lie outside the con- ground contains water that in some cases can be
tour graph shown. pumped and used. When major pumping occurs, it
is possible for the land to sink. This sinking is re-
a. Would a food scientist be more interested in
ferred to as land subsidence. The figure shows the
finding a maximum or a minimum volume
land subsidence, in feet, in the Santa Clara Valley
index?
between 1934 and 1960.
b. Approximately where on the contour graph does
the optimal point occur? a. Mark with an X on the contour graph in the
figure the approximate location of any relative
c. Estimate and interpret the optimal value. extreme points.
21. Flooding In the summer of 1993, the midwestern b. Write a sentence interpreting the points in part
United States experienced catastrophic flooding that a and characterizing them as maxima or
resulted in damages estimated at over $10 billion. A minima.
particularly heavy rainfall occurred in Missouri on
June 6 of that year. A contour map of the number of
millimeters of rain that fell on June 6, 1993, is given.
634 CHAPTER 10 Analyzing Multivariable Change: Optimization
10.2 Multivariable Optimization
In Section 4.2 we saw how to locate relative extreme points (maxima or minima) on
the graph of a two-dimensional function by analyzing the function’s derivative.
Because relative extreme points on a smooth, continuous graph can occur only where
the tangent line is horizontal (that is, with slope equal to zero), we search for points
where the derivative is zero. If we find such a point, we determine whether the graph
has a local maximum, a local minimum, or neither by determining whether the
graph is concave down, is concave up, or has an inflection point at that location.
Finding Critical Points Algebraically
We now direct our attention to looking for critical points of smooth, continuous mul-
tivariable functions. As an example, we again consider the problem of maximizing the
volume of a package sent at the “bound printed matter” postal rate. The volume of a
rectangular package that contains the maximum amount of printed material according
to postal regulations size requirement* for this rate is given by the equation
V(h, w) 108hw 2h2w 2hw2 cubic inches
where h inches is the height and w inches is the width of the package. We seek the val-
ues of h and w that cause V(h, w) to be as large as possible. The three-dimensional
graph in Figure 10.13 helps us visualize what is happening.
V(h, w)
(cubic inches)
12,000
8,000
4,000
25
0 15
0 5 10 5 w
15 20 25 (inches)
h
(inches)
FIGURE 10.13
Concept Development: Relative Extrema Let us look at the maximum from
two different perspectives. First, ignore the effect of height on the volume, and think
only about the impact due to the width. In other words, look at a cross-sectional
function of V with output V(h0, w), where h is held constant at some value h0. Figure
10.14a shows graphs of cross-sectional models for h 9 inches and h 17 inches.
Figure 10.14b shows the same cross sections depicted in three dimensions.
* Based on information at www.usps.com (accessed 1/4/2003).
10.2 Multivariable Optimization 635
V(h, w)
(cubic inches)
12,000
V(h, w)
(cubic inches)
V (17, w) 8,000
12,000
4,000
8000 V (9, w) 25
0 15
4000 0 5 10 w
15 5
w 20 25 (inches)
0
h
0 5 10 15 20 25
(inches) (inches)
(a) (b)
FIGURE 10.14 Cross sections for h 9 inches and h 17 inches
Next, ignore the effect of weight on the volume, and think only about the impact
due to the height. In other words, look at a cross-sectional function of V with output
v(h, w0), where w is held constant at some value w0. Figure 10.15a shows graphs of
cross-sectional functions for w 12 inches and w 19 inches Figure 10.15b shows
the same cross sections depicted in three dimensions.
V(h, w)
(cubic inches)
12,000
V(h, w)
(cubic inches)
V (h, 19) 8,000
12,000
V (h, 12) 4,000
8000
25
0 15
4000 0 5 10 w
15 5
b 20 25 (inches)
0 h
0 5 10 15 20 25
(inches) (inches)
(a) (b)
FIGURE 10.15 Cross sections for w 12 inches and w 19 inches
In order for the volume function to have a maximum, it must have a maximum
with respect to height and a maximum with respect to width. That is, there must be a
point at which the derivative of the cross-sectional function of V with output V(h0, w)
636 CHAPTER 10 Analyzing Multivariable Change: Optimization
is zero and the derivative of the cross-sectional function of V with output V(h, w0) is
zero. In order for a maximum of V to occur, both partial derivatives must be zero:
V
108h 2h2 4hw 0 (1)
w
V
108w 4hw 2w2 0 (2)
h
We must simultaneously solve this system of equations to find a point (h0, w0) that
satisfies the condition that both partial derivatives are equal to zero. Without using
the partial derivative symbols, we rewrite one of the equations so that one of the vari-
ables is expressed in terms of the other variable. Solving equation 1 for w in terms of
10.2.1 h yields
108h 2h2
w (3)
4h
Next, we substitute this expression for w into equation 2:
108h4h 2h 4h 108h4h 2h 2 108h4h 2h
2 2 2 2
108 0 (4)
Solving equation 4 for h yields
h 18 inches
We use this value of h to determine the value of w. Substituting the unrounded value
of h in equation 3 yields
w 18 inches
The point (h0, w0) where h0 18 inches and w0 18 inches satisfies the condition
that the partial derivatives are zero. The output corresponding to these input values
is 11,664 cubic inches. (Note that this is the same result obtained from a contour
graph sketched on the table of data in Figure 10.10 on page 619 because the point
(h0, w0) (18, 18) appeared in that table.) Are we certain that h 18, w 18, and
V 11,664 is a point of maximum volume rather than a different type of critical
point? Figure 10.10 shows that a maximum is the only possibility.
This algebraic method for finding the relative maximum of the volume of a rec-
tangular package mailed at the “bound printed matter” postal rate can be used to find
any critical points of smooth, continuous three-dimensional functions because a
critical point of such a function has the characteristic that both partial derivatives are
zero. We locate critical points of any three-dimensional function as follows:
Critical Points of Three-Dimensional Functions
Let f be a three-dimensional function with input (x, y). A critical point of f
occurs at (a, b) if
1. fx(a, b) 0 and fy(a, b) 0
or
2. fx(a, b) does not exist or fy(a, b) does not exist.
10.2 Multivariable Optimization 637
The first case yields critical points that are relative extrema or saddle points. Because
the second case does not occur often in nonscientific applications, we will restrict the
remainder of our discussion of critical points to the case where both partial deriva-
tives are zero.
How do we determine which type of critical point occurs? As seen in the package
volume illustration, if we can produce either a three-dimensional graph or a contour
graph of a function, then we can visually verify whether a point at which all of the first
partial derivatives are zero is a maximum, a minimum, or a saddle point. However, if
we cannot easily produce such a graph, then we can use the second partial derivatives
to determine what type of point we have found. To do so, we begin by defining the
determinant of a second partials matrix.
The Determinant Test
If a function f has two input variables x and y, then we can calculate the determinant
D of the second partials matrix evaluated at a point for which fx fy 0 as
D ff
xx
yx
fxy
fyy fxx fyy fxy fyx
If the determinant D is 0, it tells us nothing about the nature of the point. If D 0,
then the point is either a relative maximum or a relative minimum. We can determine
which it is by looking at fxx at the point. If fxx 0, then the cross section with respect
to x is concave down, so the point is a relative maximum in the x direction. Also,
when fxx 0 and D 0, then fyy must also be less than zero. Thus there is also a
relative maximum in the y direction. Because D 0 guarantees a maximum or min-
imum, this critical point must be a relative maximum. Similarly, if fxx 0, then the
point is a relative minimum. If D 0, then the point is a saddle point regardless of
whether fxx and fyy are positive or negative.
Determinant Test
Let f be a continuous multivariable function with two input variables x and
y. Let (a, b) be a point at which the first partial derivatives of f are both 0.
The determinant of the second partials matrix evaluated at the point (a, b) is
D(a, b) ff xx
yx
fxy
fyy fxx fyy fxy fyx
If D(a, b) 0 and fxx 0 at (a, b), then f has a relative maximum at (a, b).
If D(a, b) 0 and fxx 0 at (a, b), then f has a relative minimum at (a, b).
If D(a, b) 0, then f has a saddle point at (a, b).
If D(a, b) 0, then the test does not give any information about (a, b).
In the previous package volume example, we found the point representing
h 18 inches and w 18 inches to be an extreme point. To use the Determinant
Test to verify that the point is a maximum, we first form the second partials matrix:
4w 108 4h 4w
VVhh
wh
Vhw
Vww
108 4h 4w 4h
638 CHAPTER 10 Analyzing Multivariable Change: Optimization
Next, because the second partials matrix contains variables, we substitute h 18
and w 18 in the matrix to obtain
72 36
36 72
Then we find the determinant.
D (72)(72) (36)2 3888
Because D is positive, we know the critical point is either a maximum or a minimum.
Because Vhh 72 is less than zero (and Vww 72 is less than zero), we have a
function that is concave down in both the w and h directions. Thus we know the point
is a maximum.
Finding Critical Points Using Matrices
A linear system of equations is one in which all the variables occur to the first power
and there are no terms in which two variables are multiplied or divided. It is possible
to solve a linear system of equations by forming a matrix of coefficients and using
matrix algebra (detailed mathematics not covered in this text) to obtain a solution.
Example 1 illustrates this process.
EXAMPLE 1 Algebraically Locating and Verifying a Maximum Point
Baking Consider the volume index that provides a measure of how much a cake
rises. An index of 100 corresponds to the volume of the batter. The index can be
10.2.2 modeled by
V(l, t) 3.1l2 22.4l 0.1t2 5.3t
when l grams of leavening are used and the cake is baked at 177°C for t minutes.
a. Find the maximum volume possible and the conditions needed to achieve that
volume.
b. Verify that the point you found in part a is a maximum.
Solution
a. We begin by finding the partial derivatives of V and setting them equal to zero:
Vl 6.2l 22.4 0
Vt 0.2t 5.3 0
To solve this linear system using matrices, we form a matrix of coefficients by
writing the equations with the variables in the same order in both equations and
moving the constants to the other side:
6.2l 0 t 22.4
0 l 0.2t 5.3
10.2 Multivariable Optimization 639
A matrix corresponding to these equations is
6.2 22.4
0
0 0.2 5.3
A calculator or computer software with matrix operations can be used to rewrite
this matrix as
10
0 3.6129
1 26.5
which corresponds to the equations
l 3.6
t 26.5
Thus “l 3.6 grams of leavening and t 26.5 minutes baking time” is the solu-
tion to the system of equations. These values result in a volume index of
V(3.6, 26.5) 110.7.
b. At (3.6, 26.5, 110.7), Vll 6.2 and Vtt 0.2. These second partials are both
negative, so V is concave down in both the l and t directions. Figures 10.16a and
10.16b depict the cross sections of V at l 3.6 and t 26.5, respectively.
V(3.6, t) V(l, 26.5)
110 110
t l
101 101
(minutes) (grams)
20 26.5 35 2 3.6 5
(a) (b)
FIGURE 10.16
The graphs in Figures 10.16a and 10.16b and negative Vll and Vtt suggest that the
critical point (3.6, 26.5, 110.7) is a relative maximum. We confirm this by evalu-
ating the determinant of the second partials matrix.
D 6.2
0
0
0.2
(6.2)(0.2) (0)(0) 1.24
Because D is positive and Vll is negative, the critical point is indeed a relative
maximum. ●
Note that it is not possible to use matrices to solve the system of equations that re-
sults when finding the point that maximizes the volume of a package sent at the
“bound printed matter” postal rate, because that system of partial derivative equa-
tions is not linear.
In Example 1 the determinant of the second partials matrix is positive. Example
2 illustrates the occurrence of a critical point for which the determinant of the second
partials matrix is negative.
640 CHAPTER 10 Analyzing Multivariable Change: Optimization
EXAMPLE 2 Algebraically Locating and Verifying a Saddle Point
Grazing Cattle The total daily intake of organic matter required by a beef cow
grazing the Northern Great Plains rangeland can be modeled* by
I(s, m) 8.61967 1.244s 0.0897s2 0.20988m 0.035947m2 0.214915sm
kg per day
when the cow produces m kilograms of milk per day. The variable s is a number
between 4 and 4 indicating the size of the cow. Find the critical point for I, and
determine whether it is a maximum, a minimum, or a saddle point.
Solution
A matrix corresponding to this We find the critical point by finding where the two partial derivatives of I are zero.
linear system is
0.214915 1.244 I
0.1794
0.214915 0.071894 0.20988 s
1.244 0.1794s 0.214915m 0
which can be rewritten as I
1 0 1.3316
0.20988 0.071894m 0.214915s 0
m
0 1 6.8999
Solving the linear system of equations yields m 6.90 kg of milk per day and
s 1.33. When we substitute the unrounded values of m and s in I, we obtain a
value of approximately 8.72 kg of organic matter eaten each day. At the point
(1.33, 6.90, 8.72), Iss 0.1794 and Imm 0.071894 are both positive, indicating
that the cross sections of I with outputs I (1.33, m) and I(s, 6.90) are both concave
up. We can verify this graphically. See Figure 10.17, which shows these cross
sections.
I(−1.33, m) I(s, 6.90)
(kg per day) (kg per day)
12 12
m
8 8 s
(kg)
0 12 -4 0 4
(a) (b)
FIGURE 10.17
* E. E. Grings et al., “Efficiency of Production in Cattle of Two Growth Potentials on Northern Great
Plains Rangelands During Spring-Summer Grazing,” Journal of Food Science, vol. 74, no. 10 (1996),
pp. 2317–2326.
10.2 Multivariable Optimization 641
However, to determine whether the critical point (1.33, 6.90, 8.72) is a mini-
mum or a saddle point, we apply the Determinant Test. The determinant of the sec-
ond partials matrix is
D II
ss
ms
Ism
Imm
0.1794 0.214915
0.214915 0.071894
(0.1794)(0.071894) (0.214915)2 0.033
Because this value is negative, we know that I has a saddle point rather than a mini-
mum at s 1.33 and m 6.90 kg of milk per day. Figure 10.18a shows the planes
s 1.33 and m 6.90 depicted on the graph of I. Figure 10.18b shows a contour
graph of I.
I(s, m)
(kg per day)
m
(kg)
200
14
7.5
155 12 10
8
10 9.5
8.5
8 9
6
14 8.5
10 9.5
4 9 8
m 5 7.5
(kg) 0 4 2 7
-2 0 2 6.5
-4
6.90 -1.33
s -4 -2 0 2 4
(a) (b)
FIGURE 10.18 ●
In general, when looking for the critical points of a three-dimensional function,
we solve for the point at which the partial derivatives are zero and then determine
what type of point we have found either by visually inspecting a three-dimensional
graph or a contour graph or by using the Determinant Test.
10.2 Concept Inventory 10.2 Activities
• Extreme points: maxima and minima Getting Started
• Saddle points For each of the multivariable functions in Activities 1
• Critical points through 8, find and classify any critical points.
• Solving systems of equations to find critical points 1. R(k, m) 3k2 2km 20k 3m2 4m 60
• Determinant Test 2. H(r, s) rs 2s2 r2
642 CHAPTER 10 Analyzing Multivariable Change: Optimization
3. G(t, p) pet 3p them. The number of eggs produced in 30 days by
one female C. grandis under conditions of 16 hours
4. f(a, b) a2 4a b2 2b 12
of light each day, a constant temperature of T°C,
5. h(w, z) 0.6w2 1.3z3 4.7wz and a relative humidity of H% can be modeled by
6. R(s, t) 1.1s3 2.6s2 0.9s 6 3.1t2 5.3t E(T, H) 4191.6877 299.7038T 23.1412H
7. f(x, y) 3x2 x3 12y2 8y3 60 5.2210T2 0.0937H2 0.4023TH eggs
(Source: J. A. Morales-Ramos, S. M. Greenberg, and E. G. King,
8. g(x, y) 4xy x4 y4 “Selection of Optimal Physical Conditions for Mass Propagation
of Catolaccus grandis,” Environmental Entomology, vol. 25, no. 1
Applying Concepts (1996), pp. 165–173.)
9. Revenue A restaurant mixes ground beef that a. Determine the point where the partial deriva-
costs $b per pound with pork sausage that costs tives of E are both equal to zero.
$p per pound to make a meat mixture that is used b. Write a sentence interpreting the point in part a
on the restaurant’s signature pizza. The quarterly and characterizing it as a maximum, a mini-
revenue, in thousands of dollars, from the sale of this mum, or a saddle point.
pizza is given by the equation 12. Parasite The average time for a C. grandis egg to
R(b, p) 14b 3b2 bp 2p2 12p develop into an adult can be modeled by
a. Find the prices at which the restaurant should D(H, L) 25.6691 0.838H 2.4297L 0.0084H 2
try to purchase ground beef and pork sausage in 0.0726L2 0.0181HL days
order to maximize the quarterly revenue from
where the relative humidity is held constant at H%,
the sale of the pizza.
the eggs are exposed to L hours of light each day,
b. Explain how the Determinant Test verifies that and the temperature is held constant at 30°C.
your result gives the maximum revenue. (Source: J. A. Morales-Ramos, S. M. Greenberg, and E. G. King,
c. What is the maximum quarterly revenue from “Selection of Optimal Physical Conditions for Mass Propagation
the sale of the restaurant’s signature pizza? of Catolaccus grandis,” Environmental Entomology, vol. 25, no. 1
(1996), pp. 165–173.)
10. Profit A nursery sells mulch by the truckload. a. Determine the point where the partial deriva-
Bark mulch sells for $b per load, and pine straw tives of D are both equal to zero.
sells for $p per load. The nursery’s average weekly
profit from the sale of these two types of mulch can b. Write a sentence interpreting the point in part a
be modeled by the equation and characterizing it as a maximum, a mini-
mum, or a saddle point.
P(p, b) 144p 3p2 pb 2b2 120b 35 dollars
13. Milk Fat The increased consumption of low-fat
a. Find how much the nursery should charge for foods in this country has led to a surplus of milk fat.
each type of mulch in order to maximize the Research is being done to find new uses for milk fat,
weekly profit from the sale of mulch. such as extracting flavor from the milk fat to utilize it
b. Explain in detail how the Determinant Test veri- in other products. The process of flavor extraction
fies that your result in part a gives the maximum entails hydrolysis—a chemical reaction in which fatty
profit. acids are liberated from the milk fat. It is desirable to
maximize the rate at which the fatty acids are liber-
c. What is the maximum weekly profit from the
ated. The initial liberation rate can be modeled by
sales of these two types of mulch?
R(P, T) 32.614 1.544P 1.625T
11. Parasite Boll weevils have long presented a threat
4.905P2 7.053T2 3PT micromoles
to cotton crops in the southern United States.
Research has been done to determine the optimal of free fatty acids per minute, where 5.5 1.5P is
conditions for reproduction of Catolaccus grandis, a the pH of the milk fat mixture and the temperature
parasitoid that attaches to boll weevils and kills is (60 8T)°C.
10.2 Multivariable Optimization 643
a. Find the pH and temperature that maximize the during the process. A model for the amount of pep-
initial rate. tides released is
b. Verify that the optimal point is a maximum. P(x, y) 2.14 0.26x 0.34y 0.23x2
(Source: M. T. Patel, R. Nagarajan, and A. Kilara, “Hydrolysis of
0.16y2 0.25xy milligrams
Milk Fat by Lipase in Solvent-Free Phospholipid Reverse Micel-
lar Media,” Journal of Food Science, vol. 61, no. 1 (1996), where the pH is 9 x and the temperature is
pp. 33–38.) (70 5y)°C. The contour plot shows some of the
amounts of peptides released.
14. Fatty Acids In the chemical process described in
Activity 13, it is also desirable to maximize the a. Use the contour plot to estimate the maximum
amount of fatty acids liberated in the hydrolysis amount of peptides and the pH and tempera-
process. The number of grams of fatty acids per 100 ture needed to achieve the maximum.
grams of water can be modeled by b. Use the model to determine the pH and temper-
ature that will maximize the amount of peptides
F(p, t) 25.224 0.129p 3.052t
produced.
0.446p2 3.761t2 1.249pt grams
16. Peptides The amount of peptides in Activity 15
where the pH of the milk fat is 5.5 1.5p and the can also be modeled as a function of pH and pro-
temperature is (60 8t)°C. Find the pH and tem- cessing time:
perature that maximize the amount of fatty acid.
Verify that the optimal point is a maximum. P(x, t) 2.14 0.26x 0.04t 0.23x2
0.04t2 0.07xt milligrams
15. Peptides The Louisiana crayfish industry pro-
duces over 38,600 tons of crayfish-processing where the pH is 9 x and the processing time
byproducts every year. One potential use of the is (2.5 t) hours. The contour plot shows the
byproducts is in the production of flavor extracts. amount of peptides.
In order to extract flavoring from the crayfish-
processing byproducts, peptides from the
byproducts must be drawn out. This process is a
8
complicated one with many variables. Two of these 1.5 1.0
variables are the pH and temperature maintained 7
2.0
6
2.1
Time (hours)
5
90 2.2
0 4
0.5
80 1.0 3
Temperature (°C)
1.5 2
2.0
2.1
70 1
2.3
0
60 6 7 8 9 10 11
pH
50
40 a. Use the contour plot to estimate the maximum
6 7 8 9 10 11 12 amount of peptides as well as the pH and time
pH needed to achieve the maximum.
(Source: H. H. Baek and K. R. Cadwallader, “Enzymatic Hydrolysis
of Crayfish Processing By-products,” Journal of Food Science, vol. b. Use the model to find the pH and time that will
60, no. 5 (1995), pp. 929–934.) maximize the amount of peptides.
644 CHAPTER 10 Analyzing Multivariable Change: Optimization
17. Cooking Loss Milk proteins are sometimes added a. Find the critical point of the pigment removal
to sausage to reduce shrinking due to cooking loss function.
and to improve the texture of the sausage. Research b. Identify it as a maximum, a minimum, or a saddle
has shown that when sausage is prepared with three point.
milk proteins—sodium caseinate, whey protein, and
19. Farmland A model for the elevation above sea
skim milk powder—the cooking loss (expressed as a
level of a tract of farmland previously discussed is
percentage of initial weight) can be modeled by
L(w, s) 10.65 1.13w 1.04s 5.83ws percent E(e, n)
10.124e3 21.347e2 13.972e 2.5n2
where w is the proportion of whey protein, s is the 2.497n 802.2 feet above sea level
proportion of skim milk powder, and the propor- where e is the distance in miles east of the western
tion c of sodium caseinate can be calculated as fence and n is the distance in miles north of the
c 1 w s. Determine whether the function L southern fence.
has a relative maximum, a relative minimum, or a a. Find the two critical points of E.
saddle point, and find the corresponding propor-
b. Use the contour graph for E in the figure to iden-
tions of sodium caseinate, whey protein, and skim
tify each critical point as a maximum, a mini-
milk powder at that point.
mum, or a saddle point.
(Source: M. R. Ellekjaer, T. Naes, and P. Baardseth, “Milk Pro-
teins Affect Yield and Sensory Quality of Cooked Sausages,”
c. Use the Determinant Test to verify that the
Journal of Food Science, vol. 61, no. 3 (1996), pp. 660–666.)
critical points are as you identified them in part b.
20. Revenue A model for the revenue generated by
18. Sunflowers A process to extract pectin and pig- the sale of cheeses in the Netherlands is
ment from sunflower heads involves washing the
sunflower heads in heated water. A model for the R(c, r)
52.196r2 5935.497r 59.128cr
percentage of pigment that can be removed from a 10,299.325c 384.386c2 thousand guilders
sunflower head by washing for 20 minutes in r milli-
where c thousand tons of 40% fat cheese and r
liters of water per gram of sunflower when the
thousand tons of regular cheese are sold. Find the
water temperature is T°C is
combination of cheeses whose sales result in maxi-
P(T, r) 306.761 9.6544T 1.9836r mum revenue.
0.07368T2 0.02958r2 percent (Source: S. Louwes, J. Boot, and S. Wage, “A Quadratic-Pro-
(Source: X. Q. Shi et al., “Optimizing Water Washing Process gramming Approach to the Problem of the Optimal Use of Milk
for Sunflower Heads Before Pectin Extraction,” Journal of Food in the Netherlands,” Journal of Farm Economics, vol. 45, no. 2
Science, vol. 61, no. 3 (1996), pp. 608–612.) (May 1963), pp. 309–317.)
n N
(miles) W E
S 796
1.4 794
797
795
1.2
798
1.0
E(e, n)
0.8 800 799 (feet)
801 800 803
0.6
1.5
802
0.4 793 1.0
0 n
0.2 0.5 0.5 (miles)
e e 1.0
0 (miles) 1.5
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 (miles)
Figures for Activity 19
10.3 Optimization Under Constraints 645
21. Honey A measure of the adhesiveness of honey where g is the percentage of glycerol, s is the per-
after processing can be modeled by centage of salt, and l is the percentage of lactose.
(Source: E. Kombila-Moundounga and C. Lacroix, “Effet des
A(g, m, s, h)
151.78 4.26g 5.69m 0.67s
Combinaisons de Chloure de Sodium, de Lactose et de Glycerol
2.48h 0.05g 2 0.14m2 0.03s2 sur les Caractéristiques Rhéologiques et la Couleur des Fro-
0.05h2 0.07mh mages Fondus à Tartiner,” Canadian Institute of Food Science
and Technology Journal, vol. 24, no. 5 (1991), pp. 239–251.)
where g is the percentage of glucose and maltose, m
is the percentage of moisture, s is the percentage of a. Find the point at which all three partial deriva-
crystal used to seed the honey, and h is the holding tives are zero. Find the corresponding thickness
time of the honey in days. measure at that point.
(Source: J. M. Shinn and S. L. Wang, “Textural Analysis of Crys- b. The Determinant Test applies only to functions
tallized Honey Using Response Surface Methodology,” Cana- of two variables. Use some other method to
dian Institute of Food Science and Technology Journal, vol. 23, make a conjecture about the nature of the criti-
nos. 4–5 (1990), pp. 178–182.) cal point you found. (That is, is it a maximum or
a. Find the point at which all four partial deriva- a minimum or something else?)
tives are zero. Find the corresponding adhesive-
ness measure at that point. Discussing Concepts
b. The Determinant Test applies only to functions
23. Describe in your own words how to determine
of two variables. Use some other method to
algebraically whether a three-dimensional function
make a conjecture about the nature of the criti-
has each of the following: (a) Relative maximum;
cal point you found. (That is, is it a maximum or
(b) Relative minimum; and (c) Saddle point.
a minimum or something else?)
22. Cheese Spread A measure of the thickness of 24. Using a graphical viewpoint, explain each of the
processed cheese spread can be modeled by conditional statements in the Determinant Test
given in this section.
V(g, s, l) 1000(5.647 3.436s 1.577s2
0.375l 1.757g 0.141g2
1.217sl 0.533s2l 0.186sg)
10.3 Optimization Under Constraints
y Often it is not enough to be able to find the input that leads to a maximum (or min-
imum) output. In many applications there are constraints on what input values can
be used. For example, when trying to optimize revenue made from the sale of cheese,
(x0, y0)
g the producer is constrained by the amount of cheese the company is able to produce.
In the case where a manufacturer is trying to maximize production, budget con-
straints must be considered. Often constraints are implied by a particular context.
These constraints can be written as functions with the same input variables as the
function that is being constrained.
The simplest constraint functions are linear. For other multivariable functions f,
x
the constraint function g may not be linear. The constraint curve drawn on a contour
Nonlinear constraint curve g
graph may look like the one shown in Figure 10.19.
FIGURE 10.19
The Lagrange Multiplier
The point of minimum (or maximum) output of f subject to g(x, y) c can be
determined by first finding the smallest (or largest) value M for which the constraint
curve g(x, y) c and the contour curve f (x, y) M touch and then determining the
point (x0, y0) where these two curves meet. Figure 10.19 illustrates that a nonlinear
646 CHAPTER 10 Analyzing Multivariable Change: Optimization
y constraint curve touches this lowest or highest possible contour curve at only one
point on that contour curve and does not pass through the contour curve but remains
completely on one side of it. What does this say about the lines tangent to the contour
curve and the constraint curve at this point (x0, y0)? Compare the tangent lines shown
in Figures 10.20 and 10.21. The tangent line in Figure 10.20 is the tangent line at
(x0, y0) on the constraint curve g(x, y) c, whereas the tangent line in Figure 10.21
is the tangent line at the same point but on the contour curve f(x, y) M . They
f (x, y) = M appear to be the same. In fact, they are the same.*
g(x, y) = c
x
Concept Development: Setting Up Constrained Systems We now use the
Line tangent to the constraint
curve at the optimal point fact that the tangent line at (x0, y0) on g(x, y) c is the same line as the tangent
line at the same point on f(x, y) M to develop a method for solving constrained
FIGURE 10.20
optimization problems. The tangent line in Figure 10.20 is tangent to the curve
g(x, y) c , so from Chapter 9 we know that its slope is
g
y
dy
x
gx
g
dx
y
gy
The tangent line in Figure 10.21 is a tangent to the contour curve f(x, y) M , so its
slope is
f
f (x, y) = M
dy
x
fx
g(x, y) = c
x
f
Line tangent to the contour
dx
y
fy
curve at the optimal point
FIGURE 10.21 Because these two rates of change are equal, we have
gx f
x
gy fy
and some algebra shows that
fx f
y
gx gy
For convenience, we denote this common ratio as (the Greek letter lambda).
We now use to set up a system of equations that we solve in the process of find-
ing any extreme points. First, we know that any extreme point on the constraint
curve satisfies the condition
fx f g
so
gx x x
The point also satisfies the condition
fy f g
so
gy y y
*Although we use this result to develop the optimization technique further, you should realize that we have
only graphically illustrated it in a special case. We do not give an analytic proof.
10.3 Optimization Under Constraints 647
When estimating critical Finally, the point we are trying to find must satisfy the constraint g(x, y) c. Thus we
points from contour graphs in have a system of three equations with three unknown variables, x, y, and :
constrained optimization
problems, remember that you f g
may need to adjust your initial x x
estimate to be certain that the
estimated point is actually on f g
the constraint. y y
g(x, y) c
Solving this system of equations for x and y gives a point on the constraint curve at
which an extreme value may occur. In fact, Joseph Lagrange (1736–1813), an Italian-
born French mathematician, showed that any extreme point for the function f sub-
ject to the constraint g(x, y) c occurs among the solutions to the system of equa-
tions. For this reason, is called the Lagrange multiplier.
Once a solution to the system of equations is found, a contour graph of the multi-
variable function f and the constraint g(x, y) c may be helpful in identifying the na-
ture of the corresponding extreme point. If a contour graph is not available, testing
values of f(x, y) for values of x and y that satisfy the constraint and are near the extreme
point may also help identify that point as a maximum or minimum of the function f.
EXAMPLE 1 Locating a Critical Output Subject to a Constraint
Find the optimal point of the function f (x, y) xy x 2y subject to the constraint
g(x, y) x 1.5y 8.5.
a. Write the four partial derivatives and set up the system of equations.
b. Solve the system of equations.
c. Test the solution using the close-point method to classify the solution as a maxi-
mum or a minimum.
Solution
a. fx y 1, fy x 2, gx 1, gy 1.5 are the four partial derivatives.
fx
b. so fx gx and y 1 1 (1)
gx
fy
so fy gy and x 2 1.5 (2)
gy
x 1.5y 8.5 (3)
We can solve the equations using algebra, or we can use matrices and technology
because the three equations are all linear.
We begin the solution algebraically by solving equation 3 for x.
x 8.5 1.5y
Next, we substitute for x in equation 2 and simplify.
(8.5 1.5y) 2 1.5
10.5 1.5y 1.5 (4)
648 CHAPTER 10 Analyzing Multivariable Change: Optimization
We solve for y in equation 1 and substitute into equation 4.
y 1 so 10.5 1.5( 1) 1.5
10.5 1.5 1.5 1.5
12 3 so 4
Substituting 4 into equation 1, we have y 1 4 so y 3.
Substituting 4 into equation 2, we have x 2 1.5 4 so x 4.
c. To determine the nature of our critical point f(x, y) f(4, 3) 22, we will
choose close points on the constraint equation. We choose to examine the close
points where y 2 and y 4.
If y 2, we solve the constraint equation for x and evaluate f(x, y).
x 1.5 2 8.5 so x 8.5 3 5.5 and f(5.5, 2) 20.5.
If y 4, we solve the constraint equation for x and evaluate f(x, y).
x 1.5 4 8.5 so x 8.5 6 2.5 and f(2.5, 4) 20.5.
Each of our test points had output less than that of our critical point. We conclude
that f(4, 3) 22 is a maximum value for the constrained function. ●
EXAMPLE 2 Locating a Maximum Output Subject to a Constraint
Production Consider a mattress-manufacturing process for which the Cobb-
Douglas production function is
10.3.1
f(L, K) 48.1L0.6K 0.4 mattresses
where L represents the number of worker hours (in thousands) and K represents the
amount invested in capital (in thousands of dollars). Suppose that the plant manager
has a budget of $98,000 to be invested in capital or spent on labor and that the aver-
age wage of an employee at the mattress plant is $8 per hour.
a. Write the budget constraint equation.
b. Use Lagrange multipliers to find the expenditures on labor and capital
investment that will result in the greatest number of mattresses being pro-
duced subject to the budget constraint. Also find the maximum production
level.
c. Verify that the answer to part b is a relative maximum.
Solution
a. The manager’s budget constraint equation is
g(L, K) 8L K 98 thousand dollars
where L and K are as described in the production function.
10.3 Optimization Under Constraints 649
b. The point at which the maximum production occurs must satisfy the following
system of equations:
f g
28.86 L
0.4K 0.4 (8) (5)
L L
f g
19.24 L0.6K
0.6 (1) (6)
K K
g(L, K) c 8L K 98 (7)
To solve this system, we begin by solving equation 5 for one of the variables L, K,
or . In this example, we solve for .
28.86L
0.4K 0.4
3.6075 L
0.4K 0.4
8
Substituting this expression for into equation 6, we have
19.24 L0.6K
0.6 3.6075 L
0.4K 0.4
Next, we multiply both sides of the equation by L0.4K 0.6 to remove the negative
exponents, and solve for L:
(19.24 L0.6K
0.6)L0.4K 0.6 (3.6075 L
0.4K 0.4)L0.4K 0.6
19.24 L0.60.4K
0.60.6 3.6075 L
0.40.4K 0.40.6
19.24 L 3.6075 K
L 0.1875 K
Finally, substituting L 0.1875 K into equation 7 enables us to solve for K:
8(0.1875 K) K 98
2.5 K 98
K 39.2 thousand dollars
Next we use K 39.2 to find that L 0.1875(39.2) 7.35 thousand worker
hours. Thus the labor costs are (7.35 thousand worker hours)($8 per hour)
$58,800, and the capital investment is $39,200. With those investments of time
and money, the production level is f(7.35, 39.2) 690.608 690 mattresses.
Note that we discount the fraction of a mattress and round down the production
level. A production level of 691 mattresses would (technically) result in labor
exceeding the budget constraint.
c. To verify that the values of K and L found in part b give a maximum production
level subject to the constraint, we consider nearby points on the constraint curve.
We choose values for L that are close to 7.35. In this case, we choose 7.3 and 7.4.
Substituting these values into the constraint equation g(L, K) 8L K 98,
we obtain the corresponding values of K, so the points we check are (7.3, 39.6)
and (7.4, 38.8). We find that the values f(7.3, 39.6) 690.585 and f(7.4, 38.8)
690.584 are both less than the extreme value f(7.35, 39.2) 690.608, and we
therefore have good evidence that 690 mattresses is a maximum production level.
The manager will be able to maximize the production of mattresses by investing
$39,200 in capital and having his employees put in a total of 7350 worker hours.
Note that we could also verify that our solution is a maximum by examining the
650 CHAPTER 10 Analyzing Multivariable Change: Optimization
graph in Figure 10.22. The budget constraint g(L, K) 98 intersects the
f(L, K) 290, 490, and 690 contours. Because the contour to which the con-
straint is tangent is the largest of these values, the answer to part b is the location
of a relative maximum.
100
Capital (thousand dollars)
80 690 890 1090
60 g(L, K) = 98
40 490
290
20
0 L
0 5 10 15 20
Labor (thousand worker hours)
FIGURE 10.22 ●
Interpreting
If the level of the constraint function g changes from c to c c, it will determine a
different extreme point because adding c to the constraint equation shifts the con-
straint curve (see Figure 10.23).
f (x, y) = M + ∆M
f (x, y) = M
x
g(x, y) = c
g(x, y) = c + ∆c
FIGURE 10.23 A shifted constraint curve
Note that g(x, y) c c is tangent to the contour curve f(x, y) M M ,
whereas g(x, y) c is tangent to the curve f(x, y) M . Therefore, we can consider
the extreme value M to be a function of the level c of the constraint. In order to ana-
lyze the effect that a change in the constraint level c has on the extreme value M, we
dM
look at the rate of change of M with respect to c—that is, dc . It can be shown using
more rigorous mathematics that the Lagrange multiplier is this rate of change.
10.3 Optimization Under Constraints 651
When the level of the constraint changes by a small amount c, the extreme value
M of f(x, y) changes by
M c dM
dc
c()
When f(x, y) represents the output of a production function, the Lagrange mul-
tiplier can be used to determine the approximate increase in the optimal produc-
tion level effected by an additional allocation of resources. Economists often refer to
the value of as the marginal productivity of money.
In the mattress production example, Example 2, we can use Equation 6 to find
the value of when L 7.35 and K 39.2:
19.24 (7.35)0.6 (39.2)
0.6 7.047
dM
Because can be interpreted as dc , its units are mattresses per thousand budget
dollars.
This value of can be used to approximate the maximum production level that
would occur if the budget were increased. For example, if the manager’s budget were
$99 thousand instead of $98 thousand, the production level would increase by
dM
M c ()c
dc
7.047
mattresses
thousand budget dollars
(1 thousand budget dollars)
7 mattresses
That is, the company could produce approximately 697 mattresses, instead of 690
mattresses, if the manager’s budget were $99 thousand instead of $98 thousand. ●
Interpretation of
If the constraint level c is increased by 1 unit, then the extreme value M of
the function changes by approximately units.
EXAMPLE 3 Locating a Minimum Output Subject to a Constraint
Production Consider the production of sausage, in which the amount of shrinkage
due to cooking is reduced by the addition of three milk proteins. The percentage of
10.3.2 volume lost in cooking sausage can be modeled* by the equation
s
P(c, w, s) 10.65c 11.78w 11.69s 5.83ws percent
1
where c, w, and s are the proportions of the milk proteins sodium caseinate, whey
8.3 9 protein, and skim milk powder, respectively.
.3
0.5
a. Rewrite the function P in terms of w and s, using the fact that c w s 1.
11
3.
12
.3
b. Using the constraint w s k for k 1, find the minimum value of the function
7.
3
0 w in part a and the values of s and w at which the minimum occurs. Use the contour
0 0.5 1 graph in Figure 10.24 to verify that the point is the location of a relative minimum.
FIGURE 10.24
*M. R. Ellekjaer, T. Naes, and P. Baardseth, “Milk Proteins Affect Yield and Sensory Quality of Cooked
Sausages,” Journal of Food Science, vol. 61, no. 3 (1996), pp. 660–672.
652 CHAPTER 10 Analyzing Multivariable Change: Optimization
c. To determine the effect that changing k in the constraint w s k has on the
dP
optimal cooking loss P, find the value of dk for k 1.
d. Use the answer to part b to estimate the effect on the minimum cooking loss if
sodium caseinate makes up 10% of the total milk proteins.
e. On the basis of your answer to part c, do you believe that a sausage producer
would be wise to use sodium caseinate in addition to whey and skim milk pow-
der? Explain.
Solution
a. Substituting 1 w s for c in the equation, we obtain
P(w, s) 10.65 1.13w 1.04s 5.83ws percent
b. The point at which the minimum percentage of volume lost occurs must satisfy
the system of equations Pw gw, Ps gs, and g (w, s) 1:
1.13 5.83s
1.04 5.83w
ws1
This system of equations can be solved algebraically or using matrices. A matrix
for this linear system of equations is
5.83 0 1 1.13
C 0 5.83 1 1.04S
1 1 0 1
Using technology, we find that the solution to this system is s 0.508 and
w 0.492 with
1.83. With these proportions of whey protein and skim
milk powder, the shrinkage is P(0.492, 0.508) 10.28%. To verify that this value
s
is a minimum, we graph the constraint w s 1 on the given contour graph.
1 (See Figure 10.25.) The contour curve to which the constraint curve is tangent has
the smallest value of all contour curves that intersect or touch the constraint
8.3 9
.3 curve. This confirms that we have found the minimum of P subject to the
0.5 constraint w s 1.
11
3.
12
c. From the solution to the system of equations in part b, we have 1.83.
.3
7.
dP
3
0 w Therefore, dk 1.83 percentage points per unit of k.
0 0.5 1
d. We can approximate the effect on optimal cooking loss by using the formula
FIGURE 10.25
dP
P k
dk
In this case, k is decreased by 0.1 (from 1 to 0.9), so the change in minimum
cooking loss is approximately (1.83)(0.1) 0.183 percentage point. In other
words, if sodium caseinate is added so that the proportions of whey and skim milk
proteins sum to 0.9, minimum cooking loss will increase to approximately
10.28 0.183 10.46%.
10.3 Optimization Under Constraints 653
dP
e. Because dk is negative, any decrease in k (the combined proportion of whey and
skim milk proteins) will increase P (the percentage cooking loss). Thus the addi-
tion of sodium caseinate will increase the percentage loss in cooking. Because a
sausage producer desires minimum cooking loss, it probably is not beneficial for
producers to use sodium caseinate. We do not know, however, what other benefits
sodium caseinate may confer, such as improved texture, improved taste, or
decreased processing expenses. A sausage producer would need to evaluate all
contributing factors when making a decision. ●
Because the system of equations solved in Example 3 is linear, it can be solved
using matrices. However, the system of equations solved in Example 2 is not linear
and therefore cannot be solved using the same technique.
10.3 Concept Inventory b. Classify the constrained optimal point as a max-
imum or a minimum of f(a, b).
• Constrained optimization c. Use the methods of this section to find the opti-
• Graphing a constraint on a contour graph mal point and the optimal value.
• Classifying optimal points under a constraint
2. Consider the multivariable function f(k, h) 2k2h
• Lagrange multiplier with constraint g(k, h) k2 h 2.
• Interpretation of a. Use the graph of several of the contour curves of
f and the constraint g(k, h) 2 to estimate each
optimal point and optimal value of f (k, h) sub-
10.3 Activities ject to the constraint.
h
Getting Started
3
1. Consider the multivariable function f (a, b) ab
with constraint g(a, b) a b 90.
a. Use the graph of several of the contour curves of
f and the constraint g(a, b) 90 to estimate the g(k, h) = 2
optimal point and optimal value of f (a, b) sub- 6 6
4 1 4
ject to the constraint.
2 2
b
k
90 g(a -3 -2 -1 0 1 2 3
,b
)=
75 90
b. Classify each constrained optimal point as a
60 maximum or a minimum of f (k, h).
45
c. Use the methods of this section to find the opti-
mal points and the optimal values.
2525
30 2025
1525 3. Honey The contour graph shows a measure of
15 1025 the cohesiveness of honey as a function of the per-
0 a centage of glucose and maltose and the percentage
0 15 30 45 60 75 of moisture.
654 CHAPTER 10 Analyzing Multivariable Change: Optimization
M under the constraint
(%)
g(r, p) 2r 3p 1
26
Identify the output at the point as a maximum or a
24 minimum using:
22 a. A contour graph
5 6 7 8
20 4 9 10 b. Close points
18 3
6. Find the optimal point of
16
14 f(x, y) 80x 5y2
12 under the constraint
10
g(x, y) 2x 2y 1.4
8
G Identify the output of the point as a maximum or a
(%) minimum using:
25 30 35 40 45
a. A contour graph b. Close points
a. Estimate the critical point indicated by the con-
tour graph. Interpret your answer. Applying Concepts
b. Graph the constraint g(G, M) G M 40%
7. Cooking Loss Use the model given in the solu-
on the contour graph. Estimate the optimal point
tion to part a of Example 3 and the constraint
of cohesiveness subject to the constraint. Is the
w s 0.9 to find the minimum percentage
point the location of a maximum or a minimum?
cooking loss if 10% sodium caseinate is used. Com-
4. Honey The contour graph shows a measure of pare your answer to the approximation given in
the adhesiveness of honey as a function of G, the part d of Example 3.
percentage of glucose and maltose, and M, the per- 8. Production Use the Cobb-Douglas model for
centage of moisture. mattress production given in Example 2 to find
maximum production if the manager’s budget were
M increased to $99 thousand. Compare your answer
(%) to the estimate given above the box on page 651.
22 4 9. Advertising A health club is trying to determine
5
20 6 how to allocate funds for advertising. The manager
18 7 decides to advertise on the radio and in the news-
16
paper. Previous experience with such advertising
leads the club to expect A(r, n) 0.1r 2n responses
14
G when r ads are run on the radio and n ads appear in
12 the newspaper. Each ad on the radio costs $12, and
(%)
34 36 38 40 42 44 46 48 50
each newspaper ad costs $6.
a. Estimate the critical point indicated by the con- a. Find the optimal allocation for advertising to
tour graph. Interpret your answer. produce the maximum number of responses if
b. Graph the constraint g(G, M) G M 55% the club has budgeted $504 for advertising.
on the contour graph. Estimate the optimal point b. What is the maximum number of responses
of adhesiveness subject to the constraint. Is the expected with the given advertising costs?
point the location of a maximum or a minimum? c. Suppose the manager budgeted an additional
$26 for advertising. What is the approximate
5. Find the optimal point of
change in the optimal value as a result of this
f(r, p) 2r2 rp p2 p change in the constraint level?
10.3 Optimization Under Constraints 655
10. Honey A measure of the adhesiveness of honey 12. Cost A manufacturer is designing a packaging
can be modeled by carton for shipping. The carton will be a box with
fixed volume of 8 cubic feet. The cost to construct
A(G, M) 125.48 4.26G 4.85M
each box is
0.05G2 0.14M 2
4.8 3.2
where G is the percentage of glucose and maltose C(l, w) lw dollars
w l
and M is the percentage of moisture. A contour
graph of A is shown in Activity 4. where the box is l feet long and w feet wide. If ship-
(Source: J. M. Shinn and S. L. Wang, “Textural Analysis of ping requirements are such that the length and width
Crystallized Honey Using Response Surface Methodology,” of the box are constrained by l 2w 5 feet, what
Canadian Institute of Food Science and Technology Journal, size box will cost the least to produce? What is the
vol. 23, nos. 4–5 (1990), pp. 178–182.) minimum cost?
a. Find the absolute maximum of the adhesiveness
of honey. Compare your answer with the esti- 13. Production The daily output at a plant manufac-
mate in the answer to part a of Activity 4. turing transistor radios is approximated by the
function
b. If FDA restrictions for Grade A honey require
that the combined percentages of glucose, malt- f(L, K) 10.5463L0.3 K 0.5 radios
ose, and moisture not exceed 55%, what is the where L is the size of the labor force measured in
maximum measure of adhesiveness possible? hundreds of worker hours and K is the capital
c. Verify that the point in part b corresponds to a investment in thousands of dollars.
maximum.
a. Suppose that the plant manager has a daily budget
d. Compare your answer in part b to the estimate in of $15,000 to be invested in capital or spent on la-
the answer to part b of Activity 4. bor and that the average wage of an employee at
the radio plant is $7.50 per hour. What combina-
11. Honey A measure of the cohesiveness of honey tion of worker hours and capital expenditures will
can be modeled by yield maximum daily production?
C(G, M) 106.35 3.76G 4.71M 0.04G2 b. Verify that the output value in part a is a
0.08M2 0.06GM maximum.
where G is the percentage of glucose and maltose c. Find and interpret the marginal productivity of
and M is the percentage of moisture. A contour money for this manufacturing process.
graph of C is shown in Activity 3.
(Source: J. M. Shinn and S. L. Wang, “Textural Analysis of
14. Honey Refer to the model and constraint in
Crystallized Honey Using Response Surface Methodology,” Activity 10 and its contour graph in Activity 4.
Canadian Institute of Food Science and Technology Journal,
a. If f(G, M) k is the constraint equation and A
vol. 23, nos. 4–5 (1990), pp. 178–182.) dA
is the maximum adhesiveness measure, find dk
a. Find the absolute minimum cohesiveness of
for k 55%.
honey. Compare your answer with the estimate
in the answer to part a of Activity 3. b. Use the answer to part a to estimate the change
in the maximum adhesiveness measure if the
b. If FDA restrictions for Grade A honey require
FDA approves the percentage of glucose, malt-
that the combined percentages of glucose, malt-
ose, and moisture to be 56%.
ose, and moisture not exceed 40%, what is the
minimum measure of cohesiveness possible? c. Find the maximum measure of adhesiveness if
there are no restrictions on the percentage of
c. Verify that the point in part b corresponds to a
glucose, maltose, and moisture.
minimum.
d. Compare your answer in part b to the estimate in 15. Honey Refer to the model and constraint in
the answer to part b of Activity 3. Activity 11 and its contour graph in Activity 3.
656 CHAPTER 10 Analyzing Multivariable Change: Optimization
a. If f(G, M) k is the constraint equation and C c. Maximize the equation in part a subject to the
dC constraint in part b, using the method of La-
is the minimum cohesiveness measure, find dk
for k 40%. grange multipliers.
b. Use the answer to part a to estimate the change 19. Popcorn Tin In an effort to be environmentally re-
in the minimum cohesiveness measure if the sponsible, a confectionery company is rethinking the
FDA approves the percentage of glucose, malt- dimensions of the tins in which it packages popcorn.
ose, and moisture to be 42%. Each cylindrical tin is to hold 3.5 gallons. The bottom
c. Find the minimum measure of cohesiveness if and the lid are both circular, but the lid must have an
1
there are no restrictions on the percentage of glu- additional 18 inch around it in order to form a lip.
cose, maltose, and moisture. (Consider the amount of metal needed to create a
seam on the side and to join the side to the bottom to
16. Cost Refer to Activity 12 to answer the following be negligible.) The multivariable functions for the
questions. volume of the tin and the surface area are:
a. If M is the minimum cost and f(l, w) k is the V(r, h) r2h 3.5 gallons (808.5 cubic inches),
dM
constraint equation, find dk for k 5. where r inches is the radius of the circular base of
b. Use the answer to part a to estimate the mini- the tin, and h inches is the height of the tin.
mum cost if the constraint curve equation is
l 2w 5.5
S(r, h) 2rh r2 r 8
9 2
square inches
is the surface area of the tin, where r inches is the ra-
17. Production Refer to Activity 13 to answer the fol- dius of the circular base of the tin, and h inches is the
lowing questions. height of the tin.
a. If P is the maximum daily production and the What are the dimensions of a tin that meets these
budget constraint equation is g(L, K) c, find specifications but uses the least amount of metal
dP
dc for c $15,000. possible?
b. Use the answer to part a to estimate the change a. What is the multivariable function to be mini-
in maximum production if the budget amount is mized?
increased by $1500.
b. What is the constraint function?
c. Estimate the maximum production if the budget
c. Minimize the equation in part a subject to the
amount is decreased to $14,000. constraint in part b, using the method of La-
18. Corral A rancher removed 200 feet of wire fenc- grange multipliers.
ing from a field on his ranch. He wants to reuse the
20. Revenue A travel agency offers spring-break
fencing to create a rectangular corral into which he
cruise packages. The agency advertises a cruise to
will build a 6-foot-wide wooden gate. The dimen-
Cancun, Mexico, for $1200 per person. In order to
sions of the corral with the greatest possible area are
promote the cruise among student organizations on
found using the multivariable functions for the
campus, the agency offers a discount for student
amount of fencing and for the resulting area of the
groups selling the cruise to over 50 of their mem-
corral :
bers. The price per student will be discounted by
F(l, w) 2w 2l 6 200 feet is the amount of $10 for each student in excess of 50. For example, if
fencing needed for the specified rectangular corral an organization had 55 members go on the cruise,
of width, w feet, and length, l feet. The area of the each of those 55 students would pay $1200 5($10)
specified corral is $1150.
A(l, w) lw feet squared where w feet is the width a. Write revenue as a multivariable function of the
and l feet is the length. number of students in excess of 50 and the price
a. What is the multivariable function to be maxi- per student.
mized? b. Write a constraint function in terms of the num-
b. What is the constraint function? ber of students in excess of 50 and the price.
10.4 Least-Squares Optimization 657
c. Maximize the equation in part a subject to the 22. When estimating a critical point from a contour
constraint in part b, using the method of La- graph in a constrained optimization problem, why
grange multipliers. is it important to check the input values of the esti-
mated point?
Discussing Concepts
21. Explain why any constrained extreme point of a
function f on the constraint curve g, each with inputs
fx fy
x and y, satisfies the conditions gx gy.
10.4 Least-Squares Optimization
We have used different models in this text to represent real-world situations. We
are now in a position to understand a procedure that is used to fit functions to data.
This section is devoted to the least-squares method of determining the linear model
that best fits a set of data points.
In Chapter 1, we learned how to model data using a “best-fitting” equation. Let
us review how we decide when a particular line best fits the data in the context of an
application dealing with retirement plans.
The average number of investment choices for people who invested in 401(k)
plans at their place of employment during certain years is given in Table 10.4.
TABLE 10.4
Year 1990 1991 1992 1993
Average number
3.5 4.0 4.2 4.8
of 401(k) plans
(Sources: USA Today, November 5, 1993; Foster Higgins.)
Let x represent the number of years since 1990, and let y be the average number of
401(k) plans. A scatter plot of the data and a linear model are shown in Figure 10.26.
y
Average number
of 401(k) plans
4.5
y = ax + b
4
3.5
3
x
2.5 Years
0 1 2 3 since 1990
FIGURE 10.26
658 CHAPTER 10 Analyzing Multivariable Change: Optimization
A visual indication of how well the line fits the data is given by the extent to which
the data points deviate from the line. We determine a numerical measure of this vi-
sual observation by calculating the vertical distance (the deviation) from each data
point to the line. (See Figure 10.27.)
y
Average number
of 401(k) plans
5 d4
4.5
y = ax + b
d2 d3
4
3.5 d
1
3
x
2.5 Years
0 1 2 3 since 1990
FIGURE 10.27
Each deviation measures the error between the associated data point and the line.
Deviation error ydata yline
To obtain a single numerical measure, we square the errors and sum. When there are
n data points, we have
SSE the sum of squared errors
(first error)2 (second error)2 … (last error)2
n
d12 d22 dn2
il
di2
This number, denoted by SSE (sum of squared errors), is our overall measure of how
well the line fits the data. The key to its correct use is to remember that smaller values
of SSE arise from lines where the errors are small, and larger values of SSE arise from
lines where the errors are large. The strategy for choosing the best-fitting line is to
choose the line for which the sum of squared errors (SSE) is as small as possible. Such
a line is designated the line of best fit (also called the regression line), and this proce-
dure for choosing it is called the method of least squares.
The Line of Best Fit
The method of least squares is the procedure that is used to find the best-
fitting line based on the criterion that the sum of squared errors is as small as
possible—that is, at a minimum value. The linear model obtained by this
method is called the least-squares line.
10.4 Least-Squares Optimization 659
Thus our task is to find the values of a and b in the model y ax b such that
n
Note n 4 because we have 4
parts. SSE di2 is a minimum. Because each deviation is the y-value of the data minus
i1
the y-value on the line, we proceed as follows using the data:
4
di2 [3.5 (a 0 b)]2 [4 (a 1 b)]2
i1
[4.2 (a 2 b)]2 [4.8 (a 3 b)]2
(3.5 b)2 (4 a b)2 (4.2 2a b)2 (4.8 3a b)2
This sum of squares of errors is a function of the slope, a, and the y-intercept, b, of
the least-squares line. Thus it is a multivariable function with two input variables:
f(a, b) (3.5 b)2 (4 a b)2
(4.2 2a b)2 (4.8 3a b)2
f f
Recall that to find the minimum value of f(a, b), we first find a and b :
f
0 2(4 a b)(1) 2(4.2 2a b)(2)
a
2(4.8 3a b)(3)
f
2(3.5 b)(1) 2(4 a b)(1) 2(4.2 2a b)(1)
b
2(4.8 3a b)(1)
f f
Next, we set a 0 and b 0. Some algebraic simplification results in the equa-
tions
28a 12b 53.6
12a 8b 33
An algebraic or matrix solution to this system of equations is (0.41, 3.51).
Finally, we must verify that the point (0.41, 3.51) is a minimum for the function
f by considering the determinant of the second partials matrix:
D 2812 128 224 144 80
Because D(0.41, 3.51) 0 and faa 0, we conclude that the sum of squared errors,
SSE, is at a minimum for the linear model with slope a 0.41 and y-intercept
b 3.51. The minimum value of SSE is f(0.41, 3.51) 0.027.
The best-fitting line determined by the least-squares method for the investment
data is y 0.41x 3.51 investment choices of 401(k) plans x years after 1990. You
should verify that this is the equation found by using the linear regression routine in
your calculator or computer.
The least-squares method is illustrated in this section with small data sets to sim-
plify calculations. In real-world situations, you should be cautious when using only a
few data points, because a small error in one of the points may lead to a large error in
the least-squares line.
660 CHAPTER 10 Analyzing Multivariable Change: Optimization
best-fitting line y ax b, where x is years
10.4 Concept Inventory since 1970.
b. Find the minimum value of f(a, b), and verify
• Least-squares method that it is a minimum value. Interpret the mean-
• Sum of squared errors ing of this minimum value in terms of the data
• Line of best fit and the best-fitting line.
c. Give the linear model that best fits these data.
d. Use the best-fitting line to predict the percentage
of housing units lacking complete kitchen facili-
10.4 Activities ties in the western United States this year. Is this
a reasonable prediction? Why or why not?
Getting Started
4. Singles Are young adults staying single longer?
For the data given in Activities 1 and 2:
The percentage of adults 20–24 years of age who
a. Express SSE as a multivariable function of a and have not yet been married is given in the table.
b for the best-fitting line y ax b.
f f
b. Find a, b, and the determinant of the second
partials matrix. Year 1970 1980 1990
c. Find the minimum value of f(a, b), and verify Percent
46 60 72
never married
that the value you obtain is a minimum.
d. Use the results of part c to give the linear func- (Sources: Census Questionnaire Content, 1990 CQC-6, U.S.
tion that best fits the data. Bureau of the Census, April 1993; Statistical Abstract, 2001.)
1. a. Use the method of least squares to find the mul-
x 1 6 12
tivariable function f with inputs a and b for the
y 7 11 19 best-fitting line y ax b, where x is the num-
ber of years since 1970.
2. x 2 3 6 8 b. Find the minimum value of f(a, b), and verify
y 5 7 11 15 that it is a minimum value.
c. Give the linear model that best fits these data.
Applying Concepts
d. Use your model to estimate the percentage of
3. Kitchens The U.S. Census questionnaire defines adults 20–24 years old who had never been mar-
kitchens with complete facilities as those having a ried in 2000.
sink with piped water, a range, and a refrigerator. e. The percentage of adults 20–24 years old who
Homes that lack complete kitchen facilities have had never been married in 1960 was 41. Add this
been rare in the United States for many years. The data point, and use technology to find a linear
first census for which data were tabulated on this model. Use this model to estimate the percentage
subject was in 1970. The table shows the percentage of adults 20–24 years old who had never been
of housing units lacking complete kitchen facilities married in 2000.
in the western United States. f. The percentage of adults 20–24 years old who
had never been married was 78% in 2000. How
Year 1970 1980 1990 close are your estimates from parts d and e?
Percent lacking 3 2 1 g. Use the results of parts d, e, and f, to discuss what
(Source: Census Questionnaire Content, 1990 CQC-25, effect additional data may have when one is pre-
U.S. Bureau of the Census, December 1993.) dicting by using models.
a. Use the method of least squares to find the mul- 5. Cost A factory makes 7-mm aluminum ball bear-
tivariable function f with inputs a and b for the ings. Company planners have determined how
10.4 Least-Squares Optimization 661
much it costs them to make certain numbers of 1980. The numbers for selected years are shown in
cases of ball bearings in a single run. These costs are the table. Use the method of least squares to find the
shown in the accompanying table. best-fitting linear model for the data.
Year 1970 1975 1978 1980
Cases of Cost
ball bearings (dollars) Experiments
5.5 5 4.8 4.6
(millions)
1 3.10
2 4.25 (Source: A. Rowan, Of Mice, Models, and Men, Albany: State Uni-
versity of New York Press, 1984.)
6 8.95
8. Milk Storage The table shows the number of days
9 12.29
that milk will keep as a function of the temperature.
14 18.45 Use the method of least squares to find the best-
fitting linear model for the data.
a. Construct a scatter plot of these data.
b. Use your calculator or computer to find the best- Temperature (°F) 30 38 45 50
fitting linear model, and sketch the model on the Days 24 10 5 0.5
scatter plot. Interpret the slope and vertical-axis (Source: Data taken from a milk carton produced by Model
intercept in the context of this situation. Dairy.)
c. Find the deviation of each data point from the 9. Population Before the technology was available
line, and find the sum of the squares of the devi- to fit many kinds of models to data, researchers and
ations. others were restricted to using linear models. Be-
d. Without performing any computations, discuss cause exponential data are common in many fields
the process you would use to find the multivari- of study, it has always been important to be able to
able function f with inputs a and b for the best- fit an exponential model to data. Consider the table
fitting line y ax b. How is this line related showing past and predicted world population.
to the results of parts b and c?
Population
6. Rainfall The table gives the number of inches of Year (billions)
precipitation that fell in New York City in early 1855.
1850 1.1
1930 2.0
Month Jan Feb March
1975 4.0
Precipitation
5.50 3.25 1.35
(inches) 2013 8.0
(Source: H. Helm Clayton, ed., World Weather Records, (Source: Information Please
Washington, DC: The Smithsonian Institution, 1927.) Almanac, 1994.)
a. Construct a scatter plot of the data. Comment on
a. Use the method of least squares to find the mul-
the curvature.
tivariable function f with inputs a and b for the
best-fitting line y ax b, where x is 1 in Jan- b. Change the data so that they represent the year
uary, 2 in February, and 3 in March. and the natural log of the population. Construct
a scatter plot of the new data, and comment on
b. Find the minimum value of f(a, b), and verify
the curvature.
that it is a minimum value. Interpret the mean-
ing of this minimum value in terms of the data c. Use the technique discussed in this section to
and the best-fitting line. find the best-fitting linear model for the changed
data in part b. You should keep the data in the
c. Give the linear model that best fits these data.
form ln y and should not round the values until
7. Animal Experiments The number of animal ex- the final calculation. Label input and output
periments in England declined between 1970 and units on your model.
662 CHAPTER 10 Analyzing Multivariable Change: Optimization
d. If a and b are the parameters of the linear func- Generation 1 2 3 4 5
tion y ax b found in part c, graph the func-
tion y eb(ea)x on the scatter plot of the original Infants 11 7 5 3 2
data. Label the input and output units on the ex-
ponential model.
e. Use technology and an exponential regression Discussing Concepts
routine to find the best exponential model for
11. Another possible measure of goodness of fit could
the population data. Compare it with the model
involve the sum of absolute values of the errors in-
in part d, and reconcile any differences.
stead of the sum of squared errors. Explain how this
10. Infants Repeat Activity 9 for the data shown in might change the fit of a line to data. How would the
the table, which gives the numbers of infants born use of such a measure affect the algebraic difficulty
to each of five generations in a certain family. of finding a “best-fitting” line?
SUMMARY
Multivariable Optimization To find the constrained extreme points, or to deter-
In this chapter, we consider optimization techniques for mine whether they exist, we solve the system of equations
multivariable functions. On a three-dimensional func- f g
tion, points where the partial derivatives with respect to
x x
both input variables are both zero are called critical f g
points and can be relative extremum points or saddle
y y
points. Critical points can be estimated visually on a
g(x, y) c
table or contour graph or can be located algebraically on
a smooth, continuous function f with inputs x and y by We classify the solution to this system of equations as
f f
solving the system of equations x 0 and y 0. We the location of a relative maximum or relative minimum
then classify any point found as a solution as a relative by viewing a contour graph or three-dimensional graph
maximum, a relative minimum, or a saddle point by or by checking the value of the function at nearby points
looking at a three-dimensional graph or contour plot or on the constraint curve.
by using the Determinant Test. The number is called a Lagrange multiplier. When
M is the extreme value and c is the constraint level,
Optimization Under Constraints dM
equals dc , the rate of change of the extreme value with
Many practical problems involve constraints on the op- respect to a change in the constraint.
timization process. At the point where a multivariable
function f has an extreme point subject to the constraint Least-Squares Optimization
g(x, y) c, the constant contour curve f(x, y) M and We end the chapter with a calculus explanation for the in-
the constraint g(x, y) c have the same tangent line tuitive idea of the line of best fit that was presented in
with slope Chapter 1. We define the line of best fit to be the one for
which the sum of squared errors, SSE, attains a minimum.
dy
x f
fx The optimization technique that finds the minimum SSE
dx f
yfy is called the method of least squares and is an application
of the method of unconstrained optimization.
10.1 Concept Review 663
CONCEPT CHECK
Can you To practice, try
• Identify critical points and absolute extrema on Section 10.1 Activities 5, 21
three-dimensional and contour graphs?
• Estimate critical points and absolute extrema in tables? Section 10.1 Activities 7, 17
• Find critical points using equations? Section 10.2 Activities 1, 5
• Use the Determinant Test? Section 10.2 Activity 19
• Use the method of Lagrange multipliers? Section 10.3 Activities 1, 11
• Interpret the Lagrange multiplier l? Section 10.3 Activities 13, 17
• Use multivariable optimization to find a line of best fit Section 10.4 Activities 1, 3
for a set of data?
CONCEPT REVIEW
1. Ozone Levels The contour graph shows ozone 6 7 8 9 10 11 12 1 2 3 4 5 6
levels (in thousandths of a centimeter) as a function 90° 90°
N 300 N
of the month (beginning and ending in June) and 440
the degrees of latitude from the equator. 340 340 420
(Source: From H. W. Nurenberg, ed., Pollutants and Their Eco- 360 400
60° 320 320 60°
toxicological Significance, New York: Wiley, 1985. Reproduced 380
with permission.) 300
a. On the graph, label relative maxima with an H, 280
30° 30°
relative minima with an L, and saddle points
with an S. 260
Latitude
b. Approximate the highest and lowest ozone con-
0° 0°
centrations. Tell when and where they occur.
250 260
280
2. Bananas Banana chips, a deep-fried snack food,
30° 300 30°
are popular in many countries. Research has
shown that blanching the bananas before peeling 340
360 300
increases the crispness of the chips. The accompa- 380 400 320
320
nying table shows observed measures of crispness 60° 60°
for selected blanching times and temperatures. 360 340
(The larger the measure, the crisper the banana 300 280
chip.) S 380 280 S
90° 90°
6 7 8 9 10 11 12 1 2 3 4 5 6
Month
664 CHAPTER 10 Analyzing Multivariable Change: Optimization
a. If there are no restrictions on sales, estimate the
The measure of crispness can be modeled by
maximum profit possible and the corresponding
C(x, y) 5.2x2 299.7x 0.1y2 23.1y numbers of shirts and hats sold.
0.4xy 4671 b. T-shirts cost the company $4 each, and hats cost
$1.25 each. The budget for the cost of producing
where x is the blanching time in minutes and y is the
next year’s Superbowl shirts and hats is $150
blanching temperature in degrees Celsius.
thousand. Write this constraint as a mathemati-
Measures of crispness of banana chips cal equation.
Blanching time (minutes) c. Graph the constraint function on the contour
Blanching graph. Estimate the maximum profit subject to
temperature (°C) 2 15 30
the constraint, and give the corresponding shirt
50 4.2 6.8 4.2 and hat sales.
60 4.5 4.7 7.1
4. Profit The model for the profit function in Activ-
70 8.2 11.4 8.6 ity 3 is
80 7.3 6.9 3.9
P(s, h) 52.2s2 5935.5s 59.1sh
(Source: J. C. Jackson et al., “Optimization of Blanching for 10,299.3h 384.3h2 dollars
Crispness of Banana Chips Using Response Surface Methodol-
ogy,” Journal of Food Science, vol. 61, no. 1 (1996), pp. 165–166.) when s thousand T-shirts and h thousand hats are
sold.
a. Sketch contour curves on the table for crispness
measures of 6, 8, and 10. a. Set up the system of equations used to find the
optimal profit if the constraint in part b of Activ-
b. Using only the table values, estimate the point of
ity 3 applies.
maximum crispness.
b. Solve the system of equations in part a. Find the
c. Use the equation to find the time and tempera-
profit corresponding to the solution.
ture that produce the maximum crispness. What
is the crispness measure at that point? c. Find and interpret the value of .
d. Verify that the point found in part c is a relative d. Approximate the change in the maximum profit
maximum. Is it an absolute maximum? if the cost budget in part b of Activity 3 is in-
creased to $152 thousand.
3. Profit A company sells Superbowl T-shirts and
hats each year. The company has modeled data from 5. Weight The data give the average body weight of
previous years and has obtained a model for profit as an 8-week-old pig that is kept in an environment
a function of s, the number of Superbowl T-shirts with a fixed temperature.
sold, and h, the number of Superbowl hats sold. Both
s and h are measured in thousands. The graph shows
contour curves for values of profit (in dollars). Temperature (°C) 10 15 20
h Body weight (kg) 29.9 33.4 37.5
Hats
(thousands) (Source: A. A. Hanson, ed., Practical Handbook of
Agricultural Science, Boca Raton, FL: CRC Press, 1990.)
20
15 a. Use the method of least squares to find the mul-
200,000 tivariable function f with inputs a and b for the
10
195
,00 best-fitting line y ax b, where x is the tem-
0
5 190,000 85,000
perature.
180,000 1 s
175,000
b. Find and interpret the minimum value of f(a, b),
0 T-shirts and verify that it is a minimum value.
30 40 50 60 70 (thousands)
c. Give the linear model that best fits these data.
A contour graph of profit (in dollars)
Project 10.1 Snow Cover
Setting Tasks
Often contour maps of a geographical region are
1. Carefully examine the contour graph in the figure.
not topographical maps showing land elevation.
You may wish to make several copies of the graph
Contour maps could indicate other features, such as
for use in this project. An enlarged copy may also
depth of the water table, richness of mineral deposits,
be helpful. Begin by marking on a copy of the
population density, precipitation levels, or
graph all the relative high and low percentages
atmospheric conditions. You are probably most
indicated in the graph. Use Hs and Ls to
familiar with weather maps where contours called
distinguish between high and low percentages.
isotherms indicate temperature or contours called
isobars indicate air pressure. The contour graph in the 2. Find a map or atlas of the world that indicates
figure shows the probability of snow cover in January special topographical features. Choose at least five
in the Northern Hemisphere. of the relative extreme points in the figure, and
determine the precise topographical characteristic
of the planet at each location that results in the
corresponding relative extreme point.
3. What would you expect the snow cover
probability on Mt. Everest to be in January?
Locate Mt. Everest on the contour map. Does the
graph indicate the probability you expected? If
not, why not?
Reporting
Prepare a poster to present the contour map in such a
way as to highlight the relative extrema and the
corresponding topography at each point. Give some
information about each point (such as latitude,
longitude, elevation, and topography). Include on the
poster the location of and information about Mt.
Everest. Make the poster attractive and readable from
a distance of 3 feet.
665
666
JANUARY
150°W 120°W 90°W 60°W 30°W 0° 30°E 60°E 90°E 120°E 150°E 180° 90°N
80°N
100
75 70°N
100 100
100 100
60°N
75 50 75
25 25 50
10 0 100 50°N
10 75
5 50
50 0 10
25 40°N
50
25 25 50
10 0 25 10 50 10
0 0 25 0 0 30°N
20°N
10°N
Equator
0°
10°S
20°S
30°S
40°S
50°S
Verlagsgesellschaft, 1981, p. 100. Reproduced by permission of the publisher.)
60°S
70°S
80°S
90°S
Snow cover probability in January
(Source: The figure is taken from W. Rudloff, World Climates, Stuttgart, Germany: Wissenschaftliche
Project 10.2 Carbonated Beverage Packaging
Setting Tasks
A new firm that wants to compete with Pepsi and
1. Find equations giving the volume and the surface
Coca-Cola has enlisted your services to design the
area of a cylindrical can in terms of its height and
optimal can shape for its product. The can is to hold
radius. Use these equations to determine the
12 fluid ounces of a carbonated beverage.
height and diameter of a can that will hold 12 fluid
ounces but will require the least amount of
aluminum to construct.
2. Carefully measure the diameter and height of a 12-
ounce can made by Pepsi and of a 12-ounce can
made by Coca-Cola. Compare these dimensions to
the optimal dimensions you found in Task 1. Are
there any beverage cans on the market that
conform to the optimal dimensions?
3. Repeat Tasks 1 and 2, considering that it will cost
twice as much to form the ends of the can as it
does to form the cylindrical surface.
Reporting
1. Write a letter to the company directors explaining
what dimensions you recommend that they use for
their cans. Include your findings on the optimal
dimensions as well as the dimensions of cans
produced by competitors. Attach your
mathematical work as an appendix to support
your conclusions.
2. Prepare an oral presentation to be given in front of
the company directors. You should prepare visual
aids in order to enhance your presentation. Your
goal is to persuade the directors to construct cans
according to the dimensions you suggest. Keep in
mind that the company directors are probably not
as concerned with all of the mathematical details
as they are with your analysis of your findings.
However, be prepared to answer mathematical
questions in case any of the directors asks you to
elaborate.
667
Chapter 1 Answers to Odd Activities A-1
Answers to Odd-Numbered Activities
CHAPTER 1 15. a. Cost
Section 1.1 (dollars)
144
1. a. Input: weight of a letter
126
Input variable: w 108
Input units: ounces 90
Output: first-class domestic postage 72
Output variable: R(w) 54
Output units: cents 36
b. R is a function of w 18
x
c. w CDs
Input 1 2 3 4 5 6 7 8 9 10
Ounces
b. $90
c. 2 CDs
d. 6 CDs
Rule R
17. Weight
(pounds)
Output 9
R(w)
Cents 8
3. a. Input: day of the week
Input variable: m 7
Input units: none
Output: amount spent on lunch 6 Weeks
0 1 2 3 4
Output variable: A(m)
Output units: dollars 19. a. Approximately $9000
b. A is not a function of m unless you always spend the b. Approximately $340
same amount on lunch every Monday, Tuesday, and c. Approximately $105
so on, or unless the input is the days in only 1 week. d. The graph for 12.5% would pass through (0, 0) but
c. m would lie below the graph for 10%.
Input Day 21. a. Approximately 2.6%
b. 2001; Approximately 3.5%.
c. 2004
d. Benefits increased, but they increased by a lower per-
Rule A
centage in 2002 than 2001.
23. s 5, t 21
s 10, t 36
Output
A(m) 25. w 3, R 54.8208
Dollars w 0, R 9.4
5. This is a function. 27. t(s) 18 when s 4
7. This is a function. t(s) 0 when s 2
9. Graphs b and c represent functions. 29. R(w) 9.4 when w 0
11. a. P(Honolulu, HI) 295 R(w) 30 when w 1.974
b. P(Providence, RI) 137.8 31. Input is given.
c. P(Portland, OR) 170.1 A(15) 57,857.357
13. a. In 1988 cotton exports had a value of $1,975,000,000. 33. Output is given.
b. In 1992 cotton exports had a value of $1,999,000,000. f (x) 3.65 when x 5.000
A-2 CHAPTER 1 Ingredients of Change: Functions and Models
35. a. $1.1 million 57. To combine functions using addition or subtraction, the
b. P(x) R(x) T(x) million dollars gives the profit input units must be the same, and the output units must be
from the production and sale of x units. the same or must be able to be converted to the same unit
37. a. $16 million (i.e., dollars and thousand dollars). To combine functions
b. T(x) R(x) P(x) million dollars gives the total using multiplication or division, the input units must be
cost for the company during the xth quarter. the same and the output units must be such that when
39. a. $0.13 per bottle combined, they have a practical interpretation. To com-
C(x) bine functions using composition the output units of one
b. A(x) dollars per unit gives the average cost for
x function must match the input units of the other function.
producing x units.
P(y) Section 1.2
41. a. r(y) 100
D(y)
1. a. 3 dollars per year
b. percent
b. f(0) $5
43. a. Y(t) S(t) 1000 C(t) 650,000
3. a. 2 thousand dollars per hundred units, or 20 dollars per
31,670t 2 206,525t 1,263,708 dollars
unit
gives the VP’s total yearly salary package t years after
b. r(0) $4.5 thousand
1996, 0 t 2.
5. a. negative
b. T(1) $1,438,563 b. decreasing
D(y) 42.4y 219.5 c. 4
45. a. A(y) thousand dollars gives
H(y) 1.7y 140.3 7. a. negative
the average credit card debt per cardholder y years after b. decreasing
1990, 8 y 15. c. 7
b. A(15) 5.16 thousand dollars per cardholder 9. C(x) 0.30x 150 dollars is the total cost for x units.
47. c(x) n(x)p(x) 11. S(h) 0.25h 3 inches of snow on the ground h hours
(0.034x 3 1.331x 2 9.913x 164.447) from noon, 12 h.
(0.183x 2 2.891x 20.215) 13. a. decreasing; negative
cesarean-section deliveries 0 2.5
b. Slope 0.5 million dollars per year.
performed x years after 1980 on women who were 35 years 50
of age or older. Over the 5-year period, the corporation’s profits
49. The functions cannot be combined using function decreased 0.5 million dollars each year.
composition. c. 0.5 million dollars per year
51. (D R)(x) D(R(x)) is the revenue in dollars from the d. Input is zero at (0, 2.5): At the beginning of the period
sale of x soccer uniforms. of interest, the corporation’s profits were $2.5 million
dollars.
x Output is zero at (5, 0): After 5 years, the corporation
Uniforms
Input was not making a profit.
15. a. positive; increasing
b. 382.5 donors per year
c. slope 382.5 donors per year
Rule D ° R
D(t)
Donors
Output 9000
D(R(x))
Revenue in dollars
4 5909 t
53. h(p(t)) h(1 3e0.5t) Years
1 3e0.5t 5000
55. c(x(t)) c(4 6t) 0 8 since 1988
3(4 6t)2 2(4 6t) 5 d. D(0) 5909; In 1988 there were 5909 organ donors
108t2 132t 45 in the United States.
Chapter 1 Answers to Odd Activities A-3
17. a. $356.6 million per year b. Approximately 7533 students
b. $89.175 million per quarter c. The estimate from the models is 505 less than the ac-
c. tual enrollment. Answers may vary. For a school the
Revenue
size of the one in this activity, an error of 500 students
Year (millions of dollars)
could be significant, especially in housing and faculty
2003 2128.1 loads.
2004 2484.8 d. No; This model should not be used to predict enroll-
2005 2841.5 ment in the year 2000, because the data are too far
removed from 2000 to be of any value in such a
2006 3198.2
prediction.
29. a. 1990: e(0) 5.11 million gigagrams
d. R(y) 356.7y 2128.1 million dollars y years after 1997: e(7) 5.859 million gigagrams
2003, 0 y 3. 2002: e(12) 6.394 million gigagrams
$112,000 $97,500 b. 0.107 million gigagrams per year
19. a. Rate of change
2007 2000 c. e(22) 7.464 million gigagrams
$2071 per year 31. The equation makes it possible for us to use mathematics
to answer numerical questions concerning the situation
b. $112,000 (3 years) $2071
year
$118,214 being modeled. The units of measure on the output and
the description (including units of measure) on the input
c. The value was $100,000 in early 2002 and $150,000 in make it possible for us to interpret the numerical answers
early 2026. in the context of the situation. The interval of inputs helps
d. V(t) 2071.429t 97,500 dollars gives the value of us to know when we are extrapolating.
the house t years after 2000, 0 t 7. The model as-
sumes that the rate of increase of the market value
Section 1.3
remains constant.
21. a. $107 billion per year 1. f(x) 2(1.3x) is the black graph.
b. $2361 billion f(x) 2(0.7x) is the teal graph.
c. One possible answer: No, the rate at which consumers 3. f(x) 3(1.2x) is the teal graph.
are willing to borrow money will fluctuate with f(x) 2(1.4x) is the black graph.
changes in the state of the national economy. 5. f(x) 2 ln x is the teal graph.
d. Approximately 2012 f(x) 2 ln x is the black graph.
23. a. 78 million people per year 7. f(x) 2 ln x is the teal graph.
b. P(t) 0.078t 6 billion people gives the world’s f(x) 4 ln x is the black graph.
population t years after the beginning of 2000. 9. f (x) is increasing, with a 5% change in output for every
c. Approximately 2076; No unit of input.
d. The prediction in part c assumes that the world popu- 11. y(x) is decreasing, with a 13% change in output for every
lation will grow at a constant rate of 78 million people unit of input.
per year between now and 2076. In making this pre- 13. The number of bacteria declines by 39% each hour.
diction, the Census Bureau must have assumed that 15. a. I(t) 4.81(1.0547t) quadrillion Btu is the projected
the growth rate will increase so that the 12 billion pop- amount of petroleum imports t years after 2005,
ulation will be reached sooner than our prediction 0 t 15.
based on the linear model. b. 2019
25. a. Yes c. The projected petroleum product imports increase
b. The first differences are all equal to $0.23. without bound as time increases.
c. P(w) 0.14 0.23w dollars is the first-class domes- 17. a. W(t) 3.3(0.9854t) workers per Social Security ben-
tic postage rate for weight not exceeding w ounces, eficiary t years after 1996 between 1996 and 2030.
1 w 9. b. W(34) 2 workers per beneficiary. Fewer workers
27. a. S(x) 499.3x 976,088.3 students enrolled in year per beneficiary will mean that the program will have to
x, 1965 x 1969. find other means of supplementing payments rather
A-4 CHAPTER 1 Ingredients of Change: Functions and Models
than relying solely on withholdings from workers’
c. lim (R0e(ln 0.8t)/30) 0 . Eventually, there won’t be
wages. t→
any radon present in the building.
19. a. The sales declined approximately 59.6% each month.
27. t 35 hours
b. S(t) 520,000(0.404t) videotapes sold per month t
29. a. D(t) 406.401(0.906t) is the number of days that
months after publicity was discontinued.
milk will keep when stored at a temperature of t de-
c. 3 months: S(3) 34,288 videotapes per month
grees Fahrenheit, 30 t 70.
12 months: S(12) 10 videotapes per month
b. D(40) D(37) 2.7; The milk will spoil approxi-
21. a. T(y) 0.002(1.397y) million transistors in Intel
mately 2.7 days sooner.
processor chips y years after 1970, 1 y 35.
c. t(D) 60.547 9.913 ln D degrees Fahrenheit is the
b. 39.746% each year
temperature at which milk should be stored in order
c. Yes, the data seem to support Moore’s Law. If any-
for the milk to keep for D days, 0.5 D 24.
thing the number of transistors has been doubling
d. t(7) 41.257; The refrigerator should be set at 41.3
faster than every 2 years.
degrees Fahrenheit.
23. a. L(x) 0.845x 0.790 gallons per person per year
31. a. R(y) 8.435 0.639 ln y percent gives the New
gives the per capita consumption of bottled water in
Zealand bond rate for a maturity time of t years,
the United States x years after 1980, 0 x 23.
0.25 y 10.
E(x) 2.714(1.099x) gallons per person per year gives
b. The model estimates 15-year bond rates as
the per capita consumption of bottled water in the
R(15) 6.70, which is 0.3 percentage point less than
United States x years after 1980, 0 x 23.
the fund manager’s estimate.
Consumption c. T(p) 461,733.212(0.213p) years is the time to matu-
40 rity for a New Zealand bond with a p% rate,
35
30 7.10 p 9.40.
25 d. R(T(9.4)) 9.4; R(T(7.5)) 7.5; R(T(7.1)) 7.1
20 T(R(2)) 2; T(R(4)) 4; T(R(10)) 10
15
10 x These calculations suggest that R and T are approxi-
5 Years mate inverse functions because R(T(p)) p and
0
T(R(y)) y.
0 5 10 15 20 25 since
1980 33. a. C(d) 1.182 2.216 ln d g/mL is the concentra-
tion of a drug in the blood stream after d days,
The exponential model appears to better describe the
1 d 17.
per capita bottled water consumption on the given in-
b. lim C(d) lim C(d)
terval. t→0 t→0
b. Rate of change: 0.845 gallon per person per year per c. One possible answer: The context tells us that the
year amount of concentration will continue to increase.
Percentage change: 9.927% each year The logarithmic model also indicates an increase so
c. L(25) 21.9 gallons per person per year fits the end behavior suggested by the context.
E(25) 28.7 gallons per person per year d. C(2) 2.7 g/mL
d. L(x) 25 when x 28.651 and E(x) 25 when 35. a. p(x) 9.792 10 5 0.434 ln x is the pH of a so-
x 23.521. lution, where x is the H3O concentration in moles
According to the linear model, water consumption per liter, 5.012 109 x 3.981 10 7.
will exceed 25 gallons per person in 2009. According b. p(1.585 10 3) 2.8
to the exponential model, water consumption ex- c. Approximately 1.0 105 mole per liter
ceeded 25 gallons per person in 2004. d. Beer is acidic, with a pH of approximately 4.5.
30 (ln 0.5) 37. For an exponential model the length of the input inter-
25. a. t 93.2 hours val over which the output values either double or halve
ln 0.8
will be the same no matter where the interval starts. For
b. The amount of radon gas present after t hours is given a linear model the length of the input interval over
by R(t) R0e(ln 0.8)t/30 units. which the output values double or halve is directly
affected by the starting endpoint of the interval.
Chapter 1 Answers to Odd Activities A-5
39. An exponential model in standard form ( f(x) ab x) 20,492.567
A(t) army deaths t weeks after
increases (or decreases) without bound in one direction 1 1744.15e1.212t
and approaches zero in the other direction. By contrast, a August 31, 1918, 1 t 13.
logarithmic model either increases without bound as x
approaches and decreases without bound as x ap- A(t)
Army deaths
proaches 0 from the right, or decreases without bound as x
approaches and increases without bound as x approaches 25,000
0 from the right. 20,000
Section 1.4 15,000
1. Exponential 10,000
3. Logarithmic
5,000
5. None of these; the scatter plot indicates an inflection t
point but does not indicate a limiting value. 0 Weeks after
7. Increasing; upper limiting value is 100 0 2 4 6 8 10 12 14 8/31/1918
9. Decreasing; upper limiting value is 39.2
37.195 91,317.712
11. a. C(t) countries in Europe, North C(t) civilian deaths t weeks
1 21.374e0.183t 1 1181.203e0.951t
America and South America t years after 1840, after August 31, 1918, 2 t 13.
0 t 40. The model is a good fit.
b. C(t)
C(t) Civilian deaths
Countries
100,000
38 down
ve 80,000
ca
n
Co
60,000
Inflection 40,000
p
eu
point
av
t 20,000
nc
Co Years t
0
since 0 Weeks after
0 40 0 2 4 6 8 10 12 14 8/31/1918
1840
3015.991
13. a., b. N(t) navy deaths t weeks c. No; The models given in part b have limiting values
1 119.250e1.024t
after August 31, 1918, 0 t 13. less than the number of deaths in the table for No-
vember 30. The models are not good indicators of the
N(t)
eventual number of deaths.
Navy deaths
15. a. The limiting value appears to be approximately
3,500
2 U/100L. The inflection point occurs at approxi-
3,000
mately 9 minutes. (Answers may vary.)
2,500 1.937
b. A(m) U/100L gives the reac-
2,000 1 29.064e0.421m
1,500 tion activity after m minutes, 0 m 18. The
1,000 limiting value is approximately 1.94 U/100L.
500 c. Approximately 0.74 U/100L
t
0 11.742
Weeks after 17. a. P(x) billion people gives the
0 2 4 6 8 10 12 14 8/31/1918 1 154.546e0.026x
world’s population x years after 1800, 4 x 271.
A-6 CHAPTER 1 Ingredients of Change: Functions and Models
The equation appears to be a good fit for the later equation is bounded only in one direction and is un-
(1960–2071) data but a poor fit for the early bounded in the other. A logarithmic equation must have
(1800–1960) data. positive input and increases or decreases unbounded as its
b. According to the model, the world population will input increases without bound.
level off at 11.7 billion. This is probably not an accu-
rate prediction of future world population.
Section 1.5
c. 1850: P(50) 0.266 billion people
1990: P(190) 5.320 billion people 1. Concave up, decreasing from x 0.75 to x 3, increas-
The model does a poor job of estimating the 1850 pop- ing from x 3 to x 4
ulation and a good job of estimating the 1990 popula- 3. Concave up, decreasing from x 13.5 to x 18, in-
tion. creasing from x 18 to x 22.5
19. a. g(t) is concave up. 5. Concave down, always decreasing
b. lim g(t) 0; lim g(t) 7. y
t→ t→
c. As t decreases without bound, g approaches zero. As t 80
increases without bound, g also increases without av
Conc
ed
60
bound.
ow
Inflection
point
n
21. a. y(x) is concave down. 40 C
b. lim y(x) ; lim y(x)
on
x→0 x→ 20
ca
ve
p
u
c. As x approaches zero from the right-hand side, y de-
x
creases without bound. As x increases without bound, 5 10 15 20
y also increases without bound.
9.
23. a. l(t) is concave up until t 0 is concave down. y
b. lim l(t) 0; lim l(t) 52 12
t→ t→ ave d
nc
ow
10 Co
c. As t decreases without bound, l approaches zero. As t
n
8 Inflection
increases without bound, l approaches the limiting point
6
p
eu
value of 52. 4
av
25. a. n(k) is concave up C o nc
2
b. lim n(k) x
k→
2 4 6 8
c. As k increases or decreases without bound, n(k)
increases without bound. 11. y
27. a. C(q) is concave up from to approximately q 0.8 12
and then is concave down. Inflection
e up
10
point
b. lim C(q) ; lim C(q) 8
cav
q→ q→ 6 own
on
4 ed C
c. As q decreases without bound, C(q) increases without
cav
2
Con
bound. As q increases without bound, C(q) decreases x
without bound. 2 4 6 8 10
L
29. A logistic equation of the form f (x) is 13. a. Second differences are 8, so the data are quadratic.
1 AeBx
unlike either the exponential or logarithmic equations in b. After 3.5 seconds the height is 44 feet. After 4 seconds
that it is bounded above and below so that when B 0, the height is 0 feet.
lim f (x) 0 and lim f (x) L, or when B 0, c. H(s) 16s2 32s 128 feet is the height of the
x→ x→ missile after s seconds, 0 s 4.
lim f (x) L and lim f (x) 0. An exponential d. H(s) 0 when s 4
x→ x→
Chapter 1 Answers to Odd Activities A-7
The missile hits the water after 4 seconds. b. Using “under age” as 0 and “100 and over” as 100 for
15. a. Because the data are evenly spaced and the second dif- modeling, but not prediction purposes, the model is
ferences are constant, the data are perfectly quadratic. C(a) 19.590 10 5a3 18.421 104a2
b. 26.5 years of age 0.037a 104.601 males per 100 females gives the gen-
c. A(x) 0.0035x2 0.405x 32 years is the median der ratio in the United States for individuals who are a
age at first marriage of females in the United States x years old, 0 a 100.
years after 1900, 60 x 90. 104.3
L(a) males per 100 females
d. A (100) 26.5 years of age; yes 1 9.817 10 4e0.082a
17. a. D(x) 0.025x2 2.021x 43.78 deaths per thou- gives the gender ratio in the United States for individ-
sand people gives the 1998 death rate for the United uals who are a years old, 0 a 100.
States for people aged x years, 40 x 65. The logistic equation fits the data better than the cubic
b. equation, especially for ages above 60.
Age Model prediction Actual rate c. Among 86-year-olds there are approximately twice as
51 D(51) 4.6 4.7 many women as men. This implies that men die
52 D(52) 5.2 5.1 younger than women.
25. A graph of y ax2 bx c will be concave up when a
53 D(53) 5.7 5.6
is positive. It will be decreasing to a minimum, after
57 D(57) 8.4 8.1 which it will be increasing. When a is negative, a graph of
59 D(59) 10.1 9.7 y ax2 bx c will be concave down—increasing to a
63 D(63) 14.0 14.1 maximum and then decreasing.
A graph of y ax3 bx2 cx d could take on one of
70 D(70) 22.7 25.5
four forms. If a is positive, a graph could be concave down,
75 D(75) 30.5 38.0 increasing to an inflection point and then concave up, in-
80 D(80) 39.4 59.2 creasing; or it could be concave down and increasing to a
maximum and then decreasing to an inflection point after
c. The model is more accurate when used for interpola-
which it would be concave up and decreasing to a minimum
tion than when used for extrapolation.
and then increasing. On the other hand, if a is negative, a
19. a. The scatter plot indicates an inflection point and does
graph could be concave up, decreasing to an inflection point
not indicate a limiting value.
and then concave down, decreasing; or it could be concave
b. A(t) 0.427t3 5.286t2 22.827t 3.014 million
up and decreasing to a minimum and then increasing to an
dollars spent t years after 1990, 0 t 8.
inflection point after which it would be concave down and
c. 1993: A(3) $35 million
increasing to a maximum and then decreasing.
1999: A(9) $92 million
The 1993 estimate is more likely to be accurate because
it is an interpolation rather than an extrapolation. Chapter 1 Concept Review
d. The 1993 estimate exceeded the actual amount by $1
million. The 1999 estimate is $7 million short of the 1. a. The scatter plot is concave up. It is decreasing over
actual amount. These figures confirm the statement in (0, 2) and increasing over (2, 6).
part c. b. The scatter plot appears to be increasing without
21. a. The scatter plot suggests an inflection point, a relative bound as x approaches .
maximum, and a relative minimum. c. Quadratic
b. G(x) 15.051 105x3 0.007x2 0.085x d. lim f(x)
x→
105.027 males per 100 females is the gender ratio in 2. a. The scatter plot is increasing, concave up.
the United States x years after 1900, 0 x 100. The b. End behavior to the left is not apparent from the scat-
graph of G rises beyond 2000. One possible answer: ter plot. As x increases, the scatter plot appears to be
No; The gender ratio will not continue to increase increasing without bound.
indefinitely. c. Either quadratic or exponential (shifted up 8)
23. a. The numbers of females and males are approximately d. Quadratic: lim f(x)
x→
equal for 40-year-olds.
Exponential: lim f(x) 8, and lim f(x)
x→ x→
A-8 CHAPTER 1 Ingredients of Change: Functions and Models
3. a. The scatter plot is decreasing and does not indicate 7. a.
any curvature. x
b. The scatter plot appears to be increasing without Input Hours
bound as x approaches and decreasing without
bound as x approaches .
c. Linear
Rule S
d. lim f(x) and lim f(x)
x→ x→
4. a. The scatter plot is increasing from x 0 to x 2 and
from x 4 to x 6. It in concave down over x 0 to Output
x 3 and concave up from x 3 to x 6. It appears S(x)
Students
to have an inflection point near x 3.
b. The scatter plot appears to be decreasing without s
Input Students
bound as x approaches and increasing without
bound as x approaches .
c. Cubic
d. lim f (x) , and lim f (x) Rule C
x→ x→
5. a. The scatter plot is increasing and concave up from x
0 to x 3 and concave down from x 3 to x 6. Output
b. The scatter plot appears to be increasing toward 40 as C(s)
Cars
x approaches and decreasing toward zero as x ap-
b. Composition
proaches .
c. T(x) (C S)(x) C(S(x))
c. Logistic
d. Input: hours; Output: cars
d. lim f (x) 0, and lim f (x) 40
x→ x→ 8. a.
6. a. The scatter plot is increasing, concave down. t
Input Year
b. The scatter plot appears to be decreasing without
bound as x approaches 0 from the right and increasing
without bound but more and more slowly as x
approaches . Rule S
c. Logarithmic
d. lim f(x) , and lim f (x)
t→0 x→
Output
S(t)
Total applicants
t
Input Year
Rule M
Output
M(t)
Male applicants
b. Subtraction
c. F(t) S(t) M(t)
d. Input: none; Output: applicants
Chapter 2 Answers to Odd Activities A-9
9. a. x 13. a. The scatter plot indicates a single concavity, which in-
Input Year dicates that a quadratic or exponential model could be
used. In this instance, an exponential model will not fit
the original data set well because the output data are
not approaching zero.
Rule R
J(m) 0.546m2 141.763m 21,382.5 dollars is
the 2002 private-party resale value of a 2000 Jeep
Grand Cherokee Laredo with m thousand miles on it,
Output
R(x) 20 m 120.
Million dollars b. J(52) $15,487
x
c. M(x) 4x 68 thousand miles on the 2002 Jeep by
Input Year the end of the xth month of 2002, 0 x 12.
Slope: 4 thousand miles per month
Rate of change: 4 thousand miles per month
d. J(M(x)) 0.546(4x 68)2 141.763(4x 68)
Rule C 21,382.5 dollars is the 2002 private-party resale value
of a 2000 Jeep Grand Cherokee Laredo at the end of
the xth month of 2002, 0 x 12.
Output 14. a. The scatter plot is concave up to the left of m 8 and
C(x)
Thousand dollars concave down to the right of m 8. There is an in-
b. The profit function can be constructed from the rev- flection point near (8, 21,200).
enue and cost functions. Logistic
c. P(x) 1,000,000R(x) 1000C(x) (Answers may b. As m decreases the data approach 0. As m increases the
vary.) data approach a limiting value.
d. P(x) 1,000,000R(x) 1000C(x) dollars is the com- c. Logistic
42,183.911
pany’s profit after it has been in business x years. d. P(m) polio cases by the
10. a. Multiplication 1 21,484.252e1.249m
40p mth month of 1949.
b. P(p)
1 0.03e0.4p
40p
c. P(p) million dollars when p dollars is CHAPTER 2
1 0.03e0.4p
the price of a pound of beef. Section 2.1
11. a. C(x) 2.2(1.021x) million children living with their
1. For five days, the stock price increased an average of 46
grandparents x years after 1970, 0 x 27.
cents per day.
b. Approximately 2.1% each year
3. For the past three months, the company has lost an aver-
c. C(x) 5 when x 38.717. According to the model,
age of $8333 each month.
the number of children living with their grandparents
5. Change: Between 2004 and 2005, the number of passen-
will reach 5 million in 2009.
gers flown by Northwest Airlines increased by 1.1 mil-
d. C(x) 4.4 when x 32.689 years; The number of
lion. Percentage change: Between 2004 and 2005, the
children living with their grandparents in 1970 will
number of passengers flown by Northwest Airlines in-
have doubled by 2003.
creased by approximately 2%. Average rate of change: Be-
12. a. decreasing; The number 0.88 is less than 1 so indicates
tween 2004 and 2005, the number of passengers flown
a decreasing exponential function.
by Northwest Airlines increased by 1.1 million passen-
b. $7: D(7) 2.55 trillion pounds
gers per year.
$14: D(14) 1.04 trillion pounds
7. Change: Between 1930 and 2000, the American Indian,
$21: D(21) 0.43 trillion pounds
Eskimo, and Aleut population in the United States in-
c. Logarithmic
creased by 2,072,000. Percentage change: Between 1930
d. P(d) 14.336 7.823ln d dollars per pound when d
and 2000, the American Indian, Eskimo, and Aleut
trillion pounds of fish is demanded.
A-10 CHAPTER 2 Describing Change: Rates
population in the United States increased by 572.4%.
Average rate of change: Between 1930 and 2000, the Amer-
ican Indian, Eskimo, and Aleut population in the United
States increased, on average, by 29.6 thousand people
per year.
9. a. Percent change: 57.039%
Average rate of change: 3.8 million shares per day
b. Shares The volume of shares traded on a given day
(millions) in October 1987 stayed near (or below) 200
million shares until mid-October when it
600
spiked near 600 million shares. It then de-
500 clined to near 300 million shares for the re-
400 mainder of the month. Average rate of
300 change ignores the spike during “October
200 Madness.”
100
Trading
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 days
(Oct 1) (Oct 19) (Oct 30)
4.25 3.25 17. a. Change 9.965 million people
11. a. Approximately $0.2 billion per year:
2001 1996 Percentage change 47%
Between 1996 and 2001 sales at Kelly Services, Inc. b. Average rate of change 1.69 million people per year
increased by an average of $0.2 billion each year. 19. Approximately 488.5%
4.25 3.25 21. a. i. 3; ii. 3; iii. 3
b. Approximately 100% 30.7%:
3.25 b. i. Approximately 85.7%
Between 1996 and 2001, sales at Kelly Services, Inc. ii. Approximately 69.2%
increased by 30.7%. iii. Approximately 54.5%
13. a. P(x) 0.037x2 25.529x 527.143 thousand c. The average rate of change of any linear function over
dollars monthly profit for an airline from a roundtrip any interval will be constant because the slope of a
flight from Denver to Chicago, where x dollars is the linear function is constant. The percentage change for
average cost of a roundtrip ticket and average ticket a linear function is not constant.
prices range from $200 to $450. 23. a. Change: $419.25
b. Approximately $4.943 thousand per dollar (thousand Percentage change: 28.15%
dollars profit per dollar of ticket price) b. $104.81 per year; The balance in the account in-
c. Approximately $4.414 thousand per dollar (thou- creased, on average, by $104.81 per year from the end
sand dollars profit per dollar of ticket price) of the first year until the end of the fifth year.
15. a. Approximately 0.81 year per year (year of life c. No; There are no data available for the balance in the
expectancy per year of age) account at the middle of the fourth year. There is no
b. 10 to 20: approximately 0.96 year per year (year of way to find the average rate of change in the balance
life expectancy per year of age) from the middle to the end of the fourth year.
20 to 30: approximately 0.89 year per year (year of d. The balance in the account can be modeled as
life expectancy per year of age) B(x) 1400(1.064x) dollars, where x is the number of
Life expectancy decreases with increasing age, but the years since the initial investment, 1 x 5. Average
magnitude of the rate of decrease gets smaller as a rate of change: $109.52 per year.
black male gets older.
Chapter 2 Answers to Odd Activities A-11
25. a. 18% b. An average rate of change is the change in the output
b. Approximately 19.562% values for two points evenly spread over the change in
27. a. 11.03 years the input values at the two points. An instantaneous
b. 8.66 years rate of change is the slope at a single point on a graph.
c. 10.21 years c. A secant line connects two points on a graph. A tan-
1.00 n
29. a. A(n) 11 n dollars, when $1 is invested at
gent line is drawn at a single point on a graph.
3. Average rates of change are slopes of secant lines. Instan-
100% interest compounded n times each year
taneous rates of change are slopes of tangent lines.
b., c. 19 0 miles 60 minutes
5. 17 minutes hour 67.1 mph
Compounding n Amount 7. a. The slope is positive at A, negative at B and E, and zero
at C and D.
Yearly 1 2
b. The graph is steeper at point B than at point A.
Semiannually 2 2.25 9.
Quarterly 4 2.44
Monthly 12 2.61
Weekly 52 2.69
Daily 365 2.71
Every hour 24 365 2.72
8760
x
Every minute 60 8760 2.72 1 2 3 4 5 6
525,600
11. The graph shows a decreasing linear function. The slope
Every second 60 525,600 2.72 is a constant, negative number.
31,536,000 13. The graph shows an increasing, logistic function. The
slope is positive. The magnitude of the slope increases,
d. $2.72 then decreases, but is always positive.
1 n 15. The lines drawn at points A and C are not tangent lines.
e. lim 1 n 2.72 17.
n→ Tangent
31. Change is simply a report of the difference in two output line at Q
values. Percentage change is a report of the difference in Q
P P6
two output values so that the relative magnitude of the P3 4 P
P2 5
change may be considered. Average rate of change is a
report of the difference in two output values in a P1
way that considers the associated spread of the input
values.
Section 2.2
19. a. The graph is concave down at point A; the graph is
1. a. A continuous function is defined for all possible input concave up at points C and D; and because the graph
values on an interval. A discrete function is defined for has an inflection point at point B, the graph is neither
specific input values. A continuous graph can be concave up nor concave down at point B.
drawn without lifting the pencil from the paper. A dis- b. The tangent line at point A should lie above the curve;
crete graph is a scatter plot. A continuous model or the tangent line at point B should lie below the curve
graph can be used to find either average or instanta- to the left of point B and above the curve to the right
neous rates of change. Discrete data or a scatter plot of point B; the tangent lines at points C and D should
can be used to find only average rates of change. lie below the curve.
A-12 CHAPTER 2 Describing Change: Rates
c. 31. a.
A p(t)
People
(thousands)
B D
405,000
385,000
C 365,000
345,000
325,000
305,000
d. The slope of the tangent line is positive at points A and t
285,000
D; the slope of the tangent line is negative at point C. Years
4 2.25 265,000
21. Slope at point B 0.058 (Answers may 95 100 110 120 130 140 150 since
70 40 1990
vary.)
50 0 b. Because the function is linear, any line tangent to p(t)
23. Slope at point C 3.3
25 10 coincides with the graph itself.
580 550 c., d., e. 2.37 million people per year
Slope at point D 1.5 (Answers may vary.)
100 80
25. a. Million subscribers per year 33. Employees
b. In 2000, the number of subscribers was increasing by
3500 (9, 3250)
23.1 million subscribers per year.
3000
c. The slope of the tangent line at point A is 23.1 million (8, 2875)
2500 (4, 2050)
subscribers per year. Years
2000
d. The instantaneous rate of change at point A is 23.1 1650 since
million subscribers per year. (3, 1825) 4 6 8 10 1990
27. a. The slope at point A is 1.3; the slope at point B is 5.9; 2050 1825
the slope at point C is 4.2. a. 43 225: In 1994, the number of employees
b. mm per day per degree Celsius was increasing by approximately 225 employees per
c. At a temperature of 23 oC, the growth rate is increas- year.
ing by 5.9 mm per day per degree Celsius. b. The slope of the graph cannot be found at x 5 be-
d. 4.2 mm per day per degree Celsius cause a tangent cannot be drawn at a sharp point on a
e. 1.3 mm per day per degree Celsius graph.
3250 2875
29. a. The slope at the solstices is zero. c. 98 375: In 1998, the number of employees
b. The steepest points on the graph occur at the was increasing by approximately 375 employees per
equinoxes. The estimated slope for the spring equinox year.
is 0.4 degree per day; the estimated slope at the fall 35. The line tangent to a graph at a point P is the limiting po-
equinox is 0.4 degree per day. A negative slope indi- sition of secant lines through P and nearby points on the
cates that the sun is moving away from the north, or graph.
toward the south.
Section 2.3
1. a. Miles per hour
b. Speed or velocity
3. a. The number of words per minute cannot be negative.
b. Words per minute per week
c. Yes; If the student’s typing speed were decreasing, then
W(t) would be negative.
Chapter 2 Answers to Odd Activities A-13
5. a. When the price of a ticket from Washington to Boston the graph is concave up on (1940, 1990) and concave
is $65, the airline’s weekly profit is $15,000. down on (1990, 2000).
b. When the price of a ticket from Washington to P(b)
Boston is $65, the airline’s weekly profit is increasing Births
by $1500 per dollar of ticket price. Raising the ticket (%)
price by $1 (to $66) will increase profit by approxi- 35
mately $1500.
30
c. When the price of a ticket from Washington to
Boston is $90, the profit is decreasing by $2000 per 25
dollar of ticket price. Raising the ticket price by $1 (to 20
$91) will result in a decrease in profit of approxi-
15
mately $2000.
10
7. t (x)
5
7 0 b
1940 1950 1960 1970 1980 1990 2000 Year
13. a. Yes, P(x) will be negative if the costs are more than the
revenue.
b. Yes, P(x) will be negative if the profit declines as more
x
3 4.4 6.2 8
shirts are sold. This might happen if the price of a shirt
is lowered in order to sell more shirts.
c. All that is certain if P(200) 1.5 is that the frater-
9. a. At the start of the diet, you weighed 167 pounds. nity’s profit is declining as the number of shirts that
b. After 12 weeks of dieting, your weight was 142 are sold exceeds 200. Profit may still be positive (which
pounds. means that the fraternity is making money), but the
c. After 1 week of dieting, your weight was decreasing by negative rate of change indicates that it is not making
2 pounds per week. the most profit possible.
d. After 9 weeks of dieting, your weight was decreasing 15. a. Years per percentage point
by 1 pound per week. b. As the rate of return increases, the doubling time for
e. After 12 weeks of dieting, your weight was neither in- the investment decreases.
creasing nor decreasing. c. i. When the interest rate is 9%, it takes approxi-
f. After 15 weeks of dieting, your weight was increasing mately 7.7 years for the investment to double.
by one fourth of a pound per week. ii. When the interest rate is 5%, the doubling time is
decreasing by approximately 2.8 years per per-
g. W (t) centage point.
Weight iii. When the interest rate is 12%, the doubling time is
(pounds)
decreasing by approximately one half year per per-
167 centage point.
17. Terminal speed
(m/s)
10
(4, 8.5)
142 8 (5, 9)
t (2, 7)
140
0 12 15 Weeks 6 (2, 6.5)
(.5, 4)
11. We know that the graph has the following points: (1940, 4 (1, 4)
4), (1970, 12), (2000, 33), and (1980, 18); we know that 2
the graph has no horizontal tangents; and we know that Diameter
0
0 1 2 3 4 5 6
(mm)
A-14 CHAPTER 2 Describing Change: Rates
94 21. a. Slope 3 2 3
74
a. Slope 1.25 m/s per mm of diameter. The
51
slope of the secant line gives the average rate of change y
of the terminal speed of a raindrop between 16
two diameters. y = 2x
b. The slope of the tangent line drawn at a diameter of 4 12
mm gives the instantaneous rate of change of the ter-
minal speed of a 4-mm raindrop. 8
8.5 7 (3, 7)
c. Slope 4 2 0.75 m/s per mm: The terminal
speed of a 4-mm raindrop is increasing by approxi- 4
(2, 4)
mately 0.75 m/s per mm of diameter.
6.5 4
d. Slope (at 2 mm) 2 0.5 1.7 m/s per mm: The 0 x
terminal speed of a raindrop 2.5 mm in diameter 0 1 2 3 4
would be approximately 7.35 m/s. b. The slope is approximately 4.
e. Approximately 26.2% per mm: The terminal speed of
Slope of the secant line
a 2-mm raindrop is increasing by approximately 28%
Close point between point and x 2
per mm (of increased diameter).
1.9 3.732132
19. G(t)
Points 1.99 3.972370
100 1.999 3.997228
(11, 88)
(8, 80) 1.9999 3.999723
(10, 86)
75
2.1 4.287094
(4, 50)
50 2.01 4.027822
(2, 39) 2.001 4.002774
25
2.0001 4.000277
t
0
0 2 4 6 8 10 12 14 16 Hours 23. a. Approximately 3 million passengers per year
a. 4 hours 5.5 points per hour b. In 2004, the number of passengers going through the
Atlanta International Airport each year was increasing
11 hours 2.7 points per hour
by 3.34 million passengers per year.
(86 50) points
b. Approximately (10 4) hours 6 points per hour: Slope of the secant line
As the number of hours that you study increases from Close point between point and t 4
4 to 10 hours, your expected grade on the calculus test 3.9 3.3931
increases by an average of 6 points per hour.
3.99 3.3456
6 points/hour
c. 50 points 100% 12% per hour: After studying 3.999 3.3406
for 4 hours, your expected grade is increasing by 12%
per hour. 3.9999 3.3401
d. G(4.6) G(4) 0.6 G(4) 50 0.6 5.5 53.3 4.1 3.2811
points 4.01 3.3344
4.001 3.3394
4.0001 3.3400
Chapter 2 Answers to Odd Activities A-15
c. Approximately 4% per year: In 2004, the number of Section 2.4
passengers going through the Atlanta International
1.
Airport each year was increasing by 4% per year.
d. Using numbers based on an equation is more accu-
x3 x3
rate, but there are many situations when the equation x → 2 x → 2
x2 x2
of a graph is not provided. (Answers will vary.)
1.9 68.59 2.1 92.61
25. a. Approximately 3.757 seconds per year
1.99 788.0599 2.01 812.0601
Slope of the secant line 1.999 7988.00599 2.001 8012.006001
Close point between point and x 13
1.9999 79988.0006 2.0001 80012.0006
12.9 3.7751
x3 x3
12.99 3.7588 lim → ; lim → ;
x→2 x2 x→2 x 2
12.999 3.7572
12.9999 3.7570 x3
lim does not exist
13.1 3.7389 x→2 x2
13.01 3.7552 3.
13.001 3.7568 2x 3 7x 2x 3 7x
x → 0 x → 0
13.0001 3.7570 x x
0.1 6.98 0.1 6.98
b. Approximately 5.529 percent per year 0.01 6.9998 0.01 6.9998
c. The 13-year-old swimmer is improving his time be- 0.001 6.999998 0.001 6.999998
cause the ratex of change is negative.
1.02 0.0001 6.99999998 0.0001 6.99999998
27. a. A(x) 1.5786 is the profit in American dollars from
the sale of x mountain bikes. 2x3 7x 2x3 7x
b. Canadian: C(400) $2754.66 lim 7; lim 7;
x→0 x x→0 x
American: A(400) $1745.00
c. A(400) $34.56 per mountain bike sold 2x3 7x
lim 7
29. The percentage change gives the relative amount of change x→0 x
in the output from an initial input value to a second input
value. The percentage rate of change is a relative measure of 5.
the rate of change at a particular input value in compari-
4x 2 16 4x 2 16
son to the output value at that point. x → 2 x → 2
x2 x2
31. Finding the rate of change numerically using numbers
1.9 15.6 2.1 16.4
found with an equation is both more accurate and more
time-consuming than drawing a tangent line to a graph 1.99 15.96 2.01 16.04
and estimating the slope of that tangent line. However, 1.999 15.996 2.001 16.004
there are many situations when the equation of a graph is 1.9999 15.9996 2.0001 16.0004
not provided and the rate of change must be estimated
graphically. f(x) f(2) 4x2 16
f (2) lim lim 16
x→2 x2 x→2 x 2
A-16 CHAPTER 2 Describing Change: Rates
7. 17. a. f (3) 2.052 billion gallons
b. f (3 h) 0.042(3 h)2 0.18(3 h) 1.89
(6t 2 7) (89) (6t 2 7) (89)
t → 4 t → 4 2.052 0.072h 0.042h2 billion gallons
t4 t4
0.072h 0.042h2
3.9 47.4 4.1 48.6 c. h billion gallons per year
d. 0.072 billion gallons per year; In 2001, the amount
3.99 47.94 4.01 48.06
of fuel Northwest Airlines consumed each year was
3.999 47.994 4.001 48.006 decreasing by 72 million gallons per year.
3.9999 47.9994 4.0001 48.0006 19. a. Point: (t, 16t 2 100)
Close Point: (t h, 16h2 32th 16t2 100)
g(t) g(4) (6t2 7) (89)
g(4) lim lim 48
t→4 t4 t→4 t4 16h2 32th
Slope of the Secant: h (16h 32t)
9. Point: (x, 3x 2) Limit of the Slope of the Secant:
Close Point: (x h, 3(x h) 2) dH lim (16h 32t) 32t feet per second is
dt h→0
3x 3h 2 (3x 2)
Slope of the Secant: 3 the speed of a falling object t seconds after the object
xhx
begins falling (given that the object has not reached
dy
Limit of the Slope of the Secant: lim 3 3 the ground).
dx h→0
b. After 1 second, the object is falling at a speed of 32 feet
11. Point: (x, 3x 2) per second.
Close Point: (x h, 3(x h)2) 21. a. The number of drivers of age a years in 1997 can be
modeled as D(a) 0.045a2 1.774a 16.064
3(x h)2 3x2
Slope of the Secant: million drivers.
xhx
b. D(a) 0.089a 1.774 million drivers per year (of
(3x2 6xh 3h2) 3x2
6x 3h age)
h
c. D(20) 0.012 million drivers per year (of age); In
dy 1997, the number of drivers of a certain age is decreas-
Limit of the Slope of the Secant: lim (6x 3h) 6x
dx h→0 ing by approximately 12 thousand drivers per year (of
13. Point: (x, x3) age) when the specific age under consideration is 20
Close Point: (x h, (x h)3) years.
d. Approximately 0.79% per year (of age); In 1997, the
(x h)3 x3
Slope of the Secant: number of drivers of a certain age is decreasing by ap-
xhx
proximately 0.79 percent per year (of age) when the
(x3 3x2h 3xh2 3h2) x3
3x2 3xh h2 specific age under consideration is 20 years.
h
23. Finding a slope graphically is the only approach if all that is
Limit of the Slope of the Secant:
dy available is a graph of the function without an accompany-
lim (3x2 3xh h2) 3x2 ing equation. Finding a slope graphically may be appropri-
dx h→0
ate if all that is needed is a quick approximation of the rate
15. a. T(13) 67.946 seconds of change at a point. If a more precise determination of the
b. T(13 h) 0.181(13 h)2 8.463(13 h) rate of change at one single point is needed and an equation
147.376 is available, it may be appropriate to find the slope at that
0.181h2 3.757h 67.946 one point numerically, using a table of limiting values of the
T(13 h)T(13) 0.181h2 3.757h slopes of increasingly close secant lines. If an equation is
c.
13 h 13 h available and the rate of change at several different points is
0.181h 2 3.757h needed, it might be appropriate to use an algebraic method
d. lim h lim (0.181h 3.757)
h→0 h→0 to find a formula for the slope. A final consideration when
3.757 seconds per year of age choosing between the algebraic method and the numerical
The time required for an average 13-year-old athlete method to find slope is the difficulty involved in using the
to swim 100 meters freestyle is decreasing by 3.757 sec- algebraic method. We are generally limited to using the
onds per year of age. algebraic method for linear, quadratic, or cubic functions.
Chapter 2 Answers to Odd Activities A-17
Chapter 2 Concept Review a. The slope of the secant line drawn between x 1993
and x 1997 gives the average rate of change in the
1. a. i. A, B, C ii. E iii. D
number of employees between 1993 and 1997.
b. The graph is steeper at B than it is at A, C, or D.
b. The slope of the tangent line at x 1998 gives the in-
c. Below: C, D, E; above: A; At B: above to the left of B,
stantaneous rate of change in the number of employ-
below to the right of B
ees in 1998.
d.
A c. Between 1993 and 1997, the number of employees at
Dell Computer Corp. was increasing by an average of
approximately 1250 employees each year.
B 15000 8000
d. E(1998) 3500 employees per
E 1998 1996
year; In 1998 the number of employees at Dell Com-
puter Corp. was increasing by approximately 3500
employees per year.
2. a. Feet per second squared (feet per second per second); 3500 employees/year
acceleration 15000 employees 100% 23.3 % per year; In 1998
the number of employees at Dell Computer Corp. was
b. The speed of the roller coaster increased after point D.
increasing by approximately 23.3% per year.
c. The roller coaster’s speed was slowest at point D.
6. a. An average 22-year-old athlete can swim 100 meters
d. The roller coaster’s speed was falling the fastest at
free style in 49 seconds. The time required for an aver-
point B.
age 22-year-old athlete to swim 100 meters free style is
3. a. The number of states associated with the national
decreasing by approximately one half second per year of
P.T.A. association grew by an average of 1.125 states
age.
per year between 1915 and 1931. Estimates may vary.
b. A negative derivative indicates that the swimmer’s
b. The number of states associated with the national
time to swim 100 meters free style is decreasing as he
P.T.A. association grew by 60% between 1915 and
gets older.
1931. Estimates may vary.
7. a. R(x) 0.051x2 0.884x 4.793 billion dollars
c. The number of states associated with the national
passenger revenue for Northwest Airlines between 1991
P.T.A. association grew by an average of 1 state per
and 2003, where x is the number of years since 1990.
year between 1923 and 1927. Estimates may vary.
b. Approximately 0.14 billion dollars per year
4. a. $8144.78
b. $65,761.77 Slope of the secant line
0.065 4t between point and x 10
c. A(t) 25,000 1 4 dollars in an account after
Close point
9.9 0.13357
t years when interest is compounded quarterly at 6.5%
9.99 0.13817
d. Approximately $2625.69 per year
e. Approximately 38.04% 9.999 0.13863
5. E(x) 9.9999 0.13867
Employees 10.1 0.14879
40,000
10.01 0.13919
30,000 10.001 0.13873
10.0001 0.13868
20,000
10,000 c. In 2001, passenger revenue for Northwest Airlines was
decreasing by approximately $140 million per year.
x
0 d. R(x) 0.102x 0.884 billion dollars passenger
1992 1995 1998 2001 Year
revenue per year for Northwest Airlines between 1991
and 2003, where x is the number of years since 1990;
$0.241 billion
A-18 CHAPTER 3 Determining Change: Derivatives
8. Find a point: (x, 7x 3); find a close point: (x h, 9. The slopes are negative to the left and right of x A . The
7(x h) 3); find a formula for the slope of the secant slope appears to be zero at x A .
line between the two points and simplify completely: y'
7x 7h 3 (7x 3)
Slope of the secant xhx 7; find Slopes
the limit of the slope of the secant as the point and the x
close point become closer together: lim 7 7. A
h→0
CHAPTER 3 11. a.
Section 3.1
1. The slopes are negative to the left of x A and positive to
the right of x A. The slope is zero at x A.
y'
Slopes
x Graph Slope graph
A
b.
3. The slopes are positive for all x, near zero to the left of
x 0, and increasingly positive to the right of x 0.
y'
Slopes Graph Slope graph
13. a. Table values may vary.
x Year 1991 1993 1997 1999 2001
5. The slope is zero everywhere. Slope 6.6 6.2 2.5 1.0 5.4
y' = slope = 0 b.
Rate of change
of average bill
x (dollars per year)
6
7. The slopes are negative for all x. The magnitude is large
close to x 0 and is near zero as x increases without 4
bound.
Slopes 2
x 0 Year
1987 1991 1993 1997 1999 2001
−2
−4
−6
Chapter 3 Answers to Odd Activities A-19
15. a. Table values may vary. 19.
j'(t)
Year 1985 1990 1995 1997 2000
(inmates
Slope 7.8 46.8 79.2 55.2 21.4 per year)
200
b.
Rate of change
(cases per year)
85
18.8
t
0 5 10 Years since 1990
0 Year 21. a. Profit is increasing on average by approximately $600
1985 1990 1995 1997 2000 per car.
17. a. The average rate of change during the year (found by b. Table values may vary.
estimating the slope of the secant line drawn from
Number of cars Slope
September to May) is approximately 14 members per (dollars per car)
month. (Answers may vary.)
20 0
b. By estimating the slopes of tangent lines, we obtain the
following. (Answers may vary.) 40 160
60 750
Month Slope
(members per month) 80 10
Sept 98 100 1200
Nov 9 c.
Feb 30 Average
Apr 11 monthly profit
(dollars per car)
c.
1200
Members 900
per month
600
100
300
80 Cars
60
-300
10 20 40 50 60 70 80 90 100 110 120 sold
40
-600
20
-900
Month
Oct Nov Dec Jan Feb Mar Apr May -1200
-20
d. The average monthly profit is increasing most rapidly
-40
for approximately 60 cars sold and is decreasing most
d. Mid-February. This point on the membership graph is
rapidly when approximately 100 cars are sold. The
an inflection point.
corresponding points on the graph are inflection
e. The average of change is not useful in sketching an in-
points.
stantaneous rate-of-change graph.
e. Average rates of change are not useful in graphing in-
stantaneous rates of change.
A-20 CHAPTER 3 Determining Change: Derivatives
23. The derivative does not exist at x 0, x 3, and x 4 29. Answers may vary.
because the graph is not continuous at those inputs. y
y' 60
Slopes h
-2 -1 1 2 3 4 5
x
x
0 1
h'
25. The derivative does not exist at x 2 and x 3 because
the slopes from the right and left are different at those −80
inputs.
31. a. p(m)
y'
p'(m)
Slopes
x
1 2 3 4 5
dp 2m 1 1
b. 1
dm 2m 2m
27. Answers may vary.
dp 1 dp
f ' (t) So 1 . The graph of dm is the same as
dm 2m
the one in part a.
t
0 2
Chapter 3 Answers to Odd Activities A-21
33. When sketching a rate-of-change graph, it is important to 7. The power rule applies: f (x) 5x 4
identify the following features on the graph of the original 9. The power rule and the constant multiplier rule apply:
function: 1) input values for which the derivative does not f (x) 9x 2
exist, 2) input intervals over which the function is in- 11. The linear function rule (or the power rule and the con-
creasing, 3) input intervals over which the function is de- stant multiplier rule) apply: f (x) 5
creasing, 4) input values that correspond to a relative 13. The constant rule applies: f (x) 0
maximum or minimum of the function, 5) input values 15. The linear function rule (or the power rule and the con-
for which the function appears to be increasing or de- dy
stant multiplier rule) apply: 12
creasing most rapidly, 6) input values that correspond to dx
inflection points where the function has zero slope. 17. The sum rule, the constant multiplier rule, the power rule,
dy
and the constant rule apply: 15x 2 6x 2
Section 3.2 dx
1 3
1. The graph of the function y 2 7x is a line with slope 19. f (x) 3 x3 → f (x) 3x4 4
x x
dy
7. dx 7. 9 18
21. f (x) 2 9x2 → f (x) 18x3 3
x x
dy 3x 2 1 3x 2 1
23. f (x) 3x x1 →
dx x x x
1
f (x) 3 2
x
1 1 1 1
x 25. f (x) x x 2 → f (x) x2
dy − 2 2x
= 7 27. a. A(t) 0.1333 dollars per year t years after 1990
dx
b. A(10) $1.50
c. A(9) $0.1333 per year
dy In 1999, the ATM transaction fee was increasing by
3. The slope formula is dx 4x 41 4x 3.
approximately $0.13 per year.
dy 29. a. T(5) T(1.5)
dx b. 1.6 (5) 2 10°F per hour
c. 1.6 (0) 2 2°F per hour
d. 1.6 (4) 2 4.4°F per hour
x A negative derivative indicates that the output (tem-
dy perature) is decreasing for the related input value.
= 4x 3
dx 31. a. 1970: falling
1995: rising
5. The slope of any horizontal line is 0. b. 1970: 207 births per year
1995: 322 births per year
dy
dx
dy
=0
dx
x
A-22 CHAPTER 3 Determining Change: Derivatives
33. a. m(w) 6.930w 682.188 kilocalories per day is the Section 3.3
metabolic rate for a typical 18- to 30-year-old male
1. y'
who weighs w pounds, 88 w 200.
b. m(w) 6.930 kilocalories per day per pound is the
rate of change in the metabolic rate for a typical 18- to 4
30-year-old male who weighs w pounds,
3
88 w 200.
c. The metabolic rate for any male in the 18–30-year-old 2 dy
= ex
group will increase by approximately 7 kilocalories per dx
1
day per pound of additional weight gained or will de-
crease by approximately 7 kilocalories per day per x
pound of weight lost. −2 −1 1
35. a. R(x) 3.68x 3 47.958x 2 80.759x 166.98
billion dollars is the revenue when $x billion is spent 3. dy
on advertising, 1.2 x 6.4. dx
b. R(x) 3.68(3x 2) 47.958(2x) 80.759(1) 0
11.039x 2 95.916x 80.759 billion dollars per
billion dollars (billion dollars of revenue per billion dol- dy 1
lars of advertising) is the rate of change of revenue when =
dx x
$x billion is spent on advertising, 1.2 x 6.4.
c. R(5) $122.8 billion per billion dollars (billion dol-
lars of revenue per billion dollars of advertising) x
R(5) $502.1 billion
d. Percentage rate of change 24.5% per billion dollars 5. dy
(advertising) dx
37. a. P(x) 175 0.015x 2 0.78x 46
49.6
x
x
dollars is the profit from the sale of a storm window dy –1
when x windows are produced each hour. =
dx x
49.6
b. P(x) 0.030x 0.78 2 dollars per window
x
is the rate of change of profit when x windows are pro-
duced each hour.
c. $94.78 profit 7. h(x) 7e x
d. Approximately $1.61 per window produced 9. g(x) (ln 2.1)2.1x
When 80 storm windows are produced each hour, dh
the profit from the sale of one window is decreasing by 11. 12 (ln 1.6)(1.6x)
dx
approximately $1.61 per additional window pro- 13. f (x) 0.49690(1.05095x)
duced. 15. f (x) 4.2(ln 0.8)(0.8x)
39. If a 0, the graph of ax3 bx2 cx d increases, then
4
decreases (or level off if the cubic function does not de- 17. f (x)
x
creases at all), and finally increase again. The derivative
dy 7
graph for this cubic function will be a concave up 19.
dx x
parabola. If a 0 in the cubic formula, then the slope
21. a. A(t) (ln 1.0439)(1.0439t) thousand dollars per
graph is a concave down parabola.
year, t years after the initial investment is made.
b. A(5) 1.23986 thousand dollars ($1239.86)
c. A(5) 0.0533 thousand dollars per year
d. Approximately 4.3% per year
Chapter 3 Answers to Odd Activities A-23
dA Apple from the sales of iPods for fiscal years ending
23. a. 1000 ln (e10)e10r 10,000e10r dollars per 100%
dr between September 30, 2002, and September 30, 2006,
b. A(0.07) 10,000e10(0.07) $20,137.53 per 100%. where x is the number of fiscal years since September
Working with interest can get interesting when taking 30, 2000.
derivatives. The increase of $20,137.53 represents the c. The revenue realized by Apple for the fiscal year end-
rate of change when the interest rate (currently r ing in September 30, 2003, from the sale of iPods was
0.07) increased by 1 to 1.07. This means the interest approximately $2340.44 million dollars. At that time,
rate is increased from 7% to 107%. the revenue realized by Apple from the sale of iPods
dA was increasing by approximately $2987.53 million
c. 1000 ln (e0.1)e0.1r 100e0.1r dollars per percent-
dr dollars per year. This rate of increase in the revenue re-
age point alized by Apple from the sale of iPods was approxi-
mately 127.6% per year.
d. A(7) 100e0.1(7) $201.38 per percentage point.
1 dy
This answer is 100 of the answer to part b. 35. a. y e x e1x (e1)x → (ln e1)(e1) 1e x e x
dx
25. a. Approximately 2.041 years
b. y e kx (e k)x → y (ln e k)(e k)x k(e k)x ke kx
b. s(x) 0.14(ln 4.106)(4.106x) million iPods per year
is the rate of change of iPod sales, where x is the num-
Section 3.4
ber of years after the fiscal year that ended in Septem-
ber, 2002, 0 x 3. 1. a. f (x(2)) f (6) 140
c. Approximately 3.531 million iPods per year df
b. 27
27. a. w(7) 25.64 grams dx
w(7) 1.05 grams per week dx
c. 1.3
w(9) w(4) grams 5.977 grams dt
b.
9 4 weeks 5 weeks df
d. (27)(1.3) 35.1
1.195 grams/week dt
c. The formula for the rate of change of the mouse’s 3. The value of the investor’s gold is increasing at a rate of
7.37
weight is w(x) x grams/week. As the age of the $64.62 per day.
mouse increases, the denominator will increase and 5. a. R(476) $10,000 Canadian
the growth rate will slow. On November 25, 2002, 476 units of the commodity
9.9 o were sold, producing revenue of $10,000 Canadian.
29. a. T(d) F per day is the rate of change of the
d b. C(10,000) $6334.70 U.S.
temperature at which milk must be stored in order to
On November 25, 2002, 10,000 Canadian dollars were
remain fresh, where d is the number of days that the
worth 6334.70 U.S. dollars.
milk must be stored
dR
c. $2.6 Canadian per unit
b. As the number of days that the milk must be stored in- dx
creases, the temperature at which the milk must be On November 25, 2000, revenue was increasing by 2.6
stored decreases more slowly. Consequently, the rate Canadian dollars per unit sold.
of change of the temperature becomes closer to zero. dC
d. $0.6335 U.S. per Canadian dollar
31. a. CPI(x) 351.521 227.777 ln x points is the con- dR
sumer price index for college tuition between 1990 On November 25, 2002, the exchange rate was $0.6335
and 2000, where x is the number of years since 1980 U.S. per Canadian dollar
227.777 dC
b. CPI(18) 12.7 points per year e. $1.65 U.S. per unit
18 dx
33. a. R(x) 3.960(3.584x) million dollars is the revenue On November 25, 2002, revenue was increasing at a
realized by Apple from the sales of iPods for fiscal rate of $1.65 U.S. per unit sold.
years ending between September 30, 2002, and Sep- 7. a. p(10) 8.303 thousand people
tember 30, 2006, where x is the number of fiscal years In 2010, the population of the city was approximately
since September 30, 2000. 8300 people.
b. R(x) 3.960(ln 3.584)3.584x) million dollars per b. g(p(10)) g(8.303) 16 garbage trucks
year is the rate of change of the revenue realized by In 2010, the city owned 16 garbage trucks.
A-24 CHAPTER 3 Determining Change: Derivatives
dp 31.2e0.02t 27. Inside function: u 0.6x
c. thousand people per year
dt (1 12e0.02t)2 Outside function: f 72e u
p(10) 0.155 thousand people per year df
72e 0.6x(0.6) 43.2e 0.6x
In 2010, the population was increasing at a rate of dx
approximately 220 people per year. 29. Inside function: u 0.08x
d. g(p) 2 0.003p2 trucks per thousand people Outside function: f 1 58e u
g(8.303) 1.8 trucks per thousand people df
58e 0.08x(0.08) 4.64e 0.08x
In 2010, when the population was approximately dx
8300, the number of garbage trucks needed by the city 31. Inside function:
was increasing by 1.8 trucks per thousand people.
e. When t 10,
dg
dg dp
1.24 truck per year
u 1 18e 0.6x inside: w 0.6x
outside: u 1 18e w
dt dp dt Outside function: f 12u1 7.3
In 2010, the number of trucks needed by the city was df 129.6e0.6x
decreasing at a rate of 1.24 truck per year, or approxi- 12(1 18e0.6x)2(18e0.6x)(0.6)
dx (1 18e0.6x)2
mately 5 truck every 4 years. 33. Inside function: u x 3x
9. c(x(t)) 3(4 6t)2 2 Outside function: f u3
dc
dt
6(4 6t)(6) 144 216t df
dx
3(x 3x)2 1
2x
3
4 35. Inside function: u ln x
11. h( p(t))
1 3e0.5t Outside function: f 2u
dh 4(3)(0.5)e0.5t 6e0.5t df 1
0.5t (ln 2)(2ln x)
dt (1 3e )2 (1 3e0.5t)2 dx x
13. k(t(x)) 4.3(ln x) 2(ln x) 4 ln x 12
3 2
37. Inside function: u Bx
dk 12.9(ln x)2 4 ln x 4 Outside function: f Ae u
dx x x x df
15. p(t(k)) 7.9(1.04614k 12k )
3 2
AeBx(B)
dx
dp
7.9(ln 1.046)(1.04614k 12k )(42k2 24k)
3 2
39. a. S(x) 0.752.3x 2 53.2x 249.8 1.8 mil-
dk
lion dollars is the predicted sales for a large firm,
17. Inside function: u 3.2x 5.7
where x is the number of years in the future.
Outside function: f u5
0.75 1
df
5(3.2x 5.7)4(3.2) 16(3.2x 5.7)4 b. S(x) (2.3x 2 53.2x 249.8) 2
dx 2
(4.6x 53.2) million dollars per year is the
19. Inside function: u x 1 rate of change in predicted sales for a large firm,
8
Outside function: f 3 where x is the number of years in the future.
u c. $1.26 million per year
df 41. a. R(q) 0.0314(0.62285)e 0.62285q million dollars per
8(3)(x 1)4 24(x 1)4
dx quarter q quarters after the beginning of 1998.
21. Inside function: u x 2 3x b.
Outside function: f u
df 1 1 Quarter ending June of. . . 1998 1999 2000
(x 2 3x) 2 (2x 3)
dx 2 Revenue (million dollars) 3.0 4.2 18.8
23. Inside function: u 35x Rate of change 0.07 0.82 9.92
Outside function: f ln u (million dollars per quarter)
df 1 1
(35) Percentage rate of change 2.3 19.5 52.7
dx 35x x
(percent per quarter)
25. Inside function: u 16x 2 37x
Outside function: f ln u
df 1
(32x 37)
dx 16x 2 37x
Chapter 3 Answers to Odd Activities A-25
(49)(36.0660)(0.206743)e0.206743x 49. Composite functions are formed by using the output of
43. a. m(x)
(1 36.0660e0.206743x)2 one function (the inside function) as the input of a second
365.363e0.206743x function (the outside function). The output of the inside
(1 36.0660e0.206743x)2 and the input of the outside must agree both in the quan-
states per year is the rate of change of states with na- tity measured and in the units of measurement.
tional P.T.A. membership, where x is the number of
Section 3.5
years since 1895, 0 x 36.
b. Approximately 5 states 1. h(2) f (2)g(2) f (2)g(2) 1.5(4) 6(3) 12
c. 1915: m(20) 2.4 states per year 3. a. i. In 2007, there were 75,000 households in the city.
1927: m(32) 0.4 state per year ii. In 2007, the number of households in the city was
45. a. f (x) 13865.113(1.035x) dollars is the projected tu- decreasing by 1200 per year.
ition at a private 4-year college for the years between iii. In 2007, 52% of households in that city owned a
2000 and 2010, where x is the number of years after computer.
2000. iv. In 2007, the percentage of households in that city
b. f (x) 13,865.113(e 0.0344x) with a computer was increasing by 5 percentage
c. f (x) 13,865.113(ln 1.035)(1.035x); points per year.
f (x) 13,865.113(0.0344)(e0.0344x) b. Input: the number of years since 2005
d. Because ln 1.035 0.0344, the answers will be the Output: the number of households with computers
same: f (8) $632 per year. c. N(2) 39,000 households with computers.
47. a. The data are essentially concave up, suggesting a quad- N(2) 3126 households with computers per year
ratic or exponential function. We choose a quadratic In 2007, there were 39,000 households in the city with
model: t(x) 7.763x 2 47.447x 1945.893 units is computers, and the number of households with com-
the average weekly production at a manufacturing com- puters was increasing at a rate of 3126 households per
pany during the xth quarter after January 1, 2000, for year.
the period from January 2000 through December 2003. 5. a. i. S(10) $15.24
b. C(t(x)) 196.25 44.45 ln (7.763x 2 47.447x 2.6
S(x) dollars per week
1945.893) dollars is the weekly production cost at a (x 1)2
manufacturing company during the xth quarter after S(10) $0.02 per week
January 1, 2000, for the period from January 2000 After 10 weeks, one share was worth $15.24, and the
through December 2003. value of one share was declining by $0.02 per week.
c. Extrapolating yields ii. N(10) 125 shares
Jan—Mar 2004: C(t(17)) $574.81 per week N(x) 0.5x shares per week
Apr—June 2004: C(t(18)) $577.56 per week N(10) 5 shares per week
July—Sept 2004: C(t(19)) $580.27 per week After 10 weeks, the investor owned 125 shares,
Oct—Dec 2004: C(t(20)) $582.94 per week and the number of shares was growing by 5 shares
d. The graph does not predict a decrease between Janu- per week.
ary of 2000 and January of 2005. iii. V(10) $1904.55
d dC dt V(10) $73.50 per week
e. C(t(x))
dx dt dx After 10 weeks, the investor’s stock is worth ap-
proximately $1905, and the value is increasing at a
1
44.25 (15.525x 47.447 0)
t rate of $73.50 per week.
2.6
44.25(15.525x 47.447)
7.763x 2 47.447x 1945.893
b. V(x) (x 1)2
(100 0.25x2)
dollars per quarter is the rate of change of weekly
manufacturing cost at a manufacturing company dur-
15
2.6
x1
(0.5x)
ing the xth quarter after January 1, 2000, for the period
0.65x2 260 1.3x
from January 2000 through December 2003. 7.5x
(x 1)2 x1
d
A graph of C(t(x)) never crosses the horizontal axis
dx dollars per week is the rate of change of the value of the
indicating that cost will never decrease. investor’s stock in the company after x weeks.
A-26 CHAPTER 3 Determining Change: Derivatives
7. 9000 bushels per year b. P(x) 6250[x(0.9286x) 7.50(0.9286x)] dollars is
9. a. 8160 voters the monthly profit from CD sales when the price of a
b. Approximately 4651 votes CD is x dollars.
c. Approximately 434 votes per week c. R(x) 6250[(1)(0.9286x) x(ln 0.9286)(0.9286x)]
1 x dollars of revenue per dollar of price is the rate of
11. f (x) e (ln x)(e x)
x change in the monthly revenue from CD sales when
13. f (x) (6x 15)(32x3 49) the price of a CD is x dollars.
(3x2 15x 7)(96x2) P(x) 6250[(1)(0.9286x) x(ln 0.9286)(0.9286x)
15. f (x) (25.6x 3.7) [29(1.7x)] 7.50(ln0.9286)(0.9286x)] dollars of profit per dollar
(12.8x2 3.7x 1.2) [29(ln1.7)(1.7x)] of price is the rate of change in the monthly profit
17. f (x) [3(5.7x 2 3.5x 2.9)2(11.4x 3.5)] from CD sales when the price of a CD is x dollars.
(3.8x 2 5.2x 7)2 (5.7x2 3.5x 2.9)3 d. Table entries are rounded.
[2(3.8x2 5.2x 7)3 (7.6x 5.2)]
Rate of change of Rate of change of
19. f (x) [12.6(ln 4.8)(4.8x)](x2) 12.6(4.8x)(2x3) revenue (dollars of profit (dollars of
21. f (x) 79 198
1 7.68e0.85x
15
Price
revenue per dollar
of price)
profit per dollar
of price)
198(7.68)(0.85)e 0.85x
(79x) (1 7.68e 0.85x)2
$13 88.27 1413.83
23. f (x) (430)(ln 0.62)(0.62x)[6.42 3.3(1.46x)]1 $14 82.15 1148.76
(430)(0.62x)[6.42 3.3(1.46x)]2 $20 684.05 105.17
[3.3(ln 1.46)(1.46 x)] $21 732.93 0.06
1
25. f (x) 43x 2 (6x)(3x 2) 2
$22 771.34 90.79
27. f (x) 14[1 12.6e0.73x]1
14x [(1 12.6e0.73x)2(12.6e0.73x)(0.73)] e. Because R(13) 0 and R(14) 0, the highest
0.73(1.2912x) 8
29. a. E(x) revenue will be achieved with a price between $13
100 and $14.
(0.026x 2 3.842x 538.868) f. Because P(x) 0 for 13 x 20, P(21) 0,
women received epidural pain relief during childbirth and P(21) 0, the price corresponding to the
at an Arizona hospital between 1981 and 1997, where maximum profit is approximately $21.
x is the number of years after 1980 33. a. C(x) 71.459(1.050x) dollars is the cost to produce
0.73(ln 1.2912)(1.2912x) x units in an hour, 10 x 90.
E(x)
100 b. C(x) 71.459(ln 1.050)(1.050x) dollars per unit
(0.026x 3.842x 538.868)
2 produced is the rate of change of the cost to produce
0.73(1.2912x) 8 x units in an hour, 10 x 90.
(0.052x 3.842)
100 C(x)
c. A(x) 71.459(1.050x)(x 1) dollars per
women per year is the rate of change in the number of x
unit is the average cost to produce one unit when x
women who received epidural pain relief during child-
units are produced each hour, 10 x 90.
birth at an Arizona hospital between 1981 and 1997,
d. A(x) 71.459[(ln 1.050)(1.050x)(x 1)
where x is the number of years after 1980
(1.050x)(x 2)] dollars per unit per hourly unit
b. Increasing by p(17) 14.4 percentage points per
produced is the rate of change in the average cost to
year
produce one unit when x units are produced,
c. Decreasing by approximately 5 births per year
10 x 90.
d. Increasing by E(17) 64 women per year.
e. 15 units: A(15) $0.18 per unit per hourly unit
e. Profit $57 E(17) $17,000
produced
31. a. R(x) 6250x(0.9286x) dollars is the monthly revenue
35 units: A(35) $0.23 per unit per hourly unit
from CD sales when the price of a CD is x dollars.
produced
Chapter 3 Answers to Odd Activities A-27
85 units: A(85) $1.97 per unit per hourly unit 1989–90, where x is the number of years since the
produced 1980–81 school year.
f. Average cost is decreasing over the set of inputs from d. The rate of change of the percentage of students drop-
10 to 21 units. ping out was smallest in the 1980–81 school year,
g. The graph of A crosses the x-axis near x 20.5. Prac- when it was 0.44 percentage points per year. The rate
tically speaking, the average cost is decreasing at 20 of change of the percentage of students dropping out
units and increasing at 21 units, so a production level was greatest in the 1985–86 and 1986–87 school years,
of 21 units is the one at which average cost first when it was 0.19 percentage points per year.
increases. e. Negative rates of change indicate that the number of
35. a. t(x) 0.015x 2 1.865x 59.430 million house- dropouts in South Carolina was falling during the
holds owned TVs x years after 1970, 0 x 32. 1980s. This means that more students were staying in
0.868 school.
b. v(x) (percent expressed as
1 4557.412e 0.485x 39. The inputs must correspond in order for the result of the
a decimal) of households with TVs also had VCRs, x multiplication to be meaningful.
years after 1970, 0 x 32.
c. n(x) (0.015x 2 1.865x 59.430) Section 3.6
0.868 1. 7
million households owned TVs
1 4557.412e 0.485x 3. 1
and VCRs x years after 1970. 5. 0
0.868
d. n(x) (0.03x 1.865) 0
1 4557.412e 0.485x 7. ; 1
0
(0.015x 2 1.865x 59.430) 0 4
(0.868)(4557.412)(0.485)e 0.485x 9. ;
(1 4557.412e 0.485x)2
0 3
11. 0
million households per year is the rate of change in the 13. 0
number of households with TVs and VCRs, x years 15. 0
after 1970, 0 x 32. 0 1
e. 1990: 7.4 million households per year 17. ;
0 40
1995: 1.9 million households per year 0 3
2000: 0.9 million households per year 19. ;
0 13
37. a. E(x) 151.516x3 2060.988x2 8819.062x 21. 0
195,291.201 23. 0 ; 0
students is the enrollment in ninth through twelfth
3
grades in South Carolina between 1980–81 and 25. ;
5
1989–90, where x is the number of years since the
1980–81 school year. 27. ;
D(x) 14.271x3 213.882x2 1393.655x 0 h
29. Consider the case . We know that 1) lim 0 for any
11,697.292 students is the number of students 0 h→0 c
c
dropping out from ninth through twelfth grades in non-zero real number and 2) lim is increasing (or de-
h→0 h
South Carolina between 1980–81 and 1989–90, where creasing) without bound for any non-zero real number.
x is the number of years since the 1980–81 school year. If we apply these two statements repeatedly with c
D(x)
b. P(x) 100 % is the percent of students drop- approaching 0, we end with an apparent contradiction.
E(x)
ping out from ninth through twelfth grades in South Similar arguments can be applied for and 0 .
Carolina between 1980–81 and 1989–90, where x is the
number of years since the 1980–81 school year. Chapter 3 Concept Review
c. P(x) D(x)[E(x)]1 D(x)[(E(x))2 E(x)] 1. a. x 0.8
percentage points per year is the rate of change in the b. positive slope: 0.8 x 2
percent of students dropping out from ninth through negative slope: 3 x 0, 0 x 0.8
twelfth grades in South Carolina between 1980–81 and
A-28 CHAPTER 4 Analyzing Change: Applications of Derivatives
c. x 0 b.
d. f (2) 4; f (1) 1.1 A(t)
e. f' (x) (billions of dollars)
7000 1500
4 Slope =
3
6000
3 = $500 billion
5000 per year Rise ≈ $1500 billion
2 4000 Run = 3 years
1 3000
2000 t
x Years since
−3 −2 −1 1 2 0
0 5 10 15 20 1980
−1
c. A(18) $496.4 billion per year. In 1998, the total
−2 outstanding mortgage debt in the United States was
increasing by $496.4 billion per year.
−3
4. a. Rewrite P(t) as P(t) 100N(t)[A(t)]1. So that using
−4 the Product and Chain Rules
100N(t) 100N(t)A(t)
−5 P(t) percentage points
A(t) [A(t)]2
per year is the rate of change in the percentage of out-
8.101(214.8)(0.797)e0.797t
2. a. D(t) pounds per per- standing mortgage debt that represents new mortgages,
(1 214.8e0.797t)2
t years after 1980.
son per year is the rate of change in the annual per
b. Input units: years
capita consumption of turkey in the United States
Output units: percentage points per year
between 1980 and 2002, t years after 1980.
b. D(10) 0.5 pound per person per year; In 1990 the
average annual per capita consumption of turkey in CHAPTER 4
the United States was increasing by 0.5 pound per
Section 4.1
person per year.
c. D(21) 0.0 pound per person per year; In 2001, per 1. Approximately 30.7%
capita consumption of turkey in the United States was 3. Approximately 19.3
not changing. 5. a. Increasing production from 500 to 501 units will in-
3. a. 506.3 billion per year crease cost by approximately $17.
b. If sales increase from 150 to 151 units, then profit will
increase by approximately $4.75.
7. A marginal profit of $4 per shirt means that the frater-
nity is currently losing $4 for each additional shirt sold.
The fraternity should consider selling fewer shirts or in-
creasing the sales price.
Chapter 4 Answers to Odd Activities A-29
9. Premium 13. a. In 2000 the population of South Carolina was increas-
(dollars) ing by 53.6 thousand people per year.
b. Between 2000 and 2003, the population increased by
12,000 approximately 160.8 thousand people.
10,000 c. By finding the slope of the tangent line at 2000 and
8000 multiplying by 3, we determine the change in the tan-
Rise ≈ $8600
gent line from 2000 through 2003 and use that change
6000
to estimate the change in the population function.
4000
Run ≈ 10 years 15. a. Approximately 2.5 million people per year
2000 b. Approximately 2.5 million people
Age
0 17. a. Approximately 1.15 million pieces per year
(years)
30 70 b. Approximately 1.15 million pieces
Slope of tangent line: approximately $860 per year of age c. Approximately 1.13 million pieces
Annual premium for 70-year-old: approximately $7850 d. 1.5 million pieces
Premium for 72-year-old: approximately $9570 e. The answers in parts b and c differ by less than 25
(Answers may vary slightly.) thousand pieces. They both differ from the answer in
11. a. part d by more than 300 thousand pieces. One possible
Revenue answer: the answer in part e (computed directly from
(billions of dollars)
the data) is the most accurate because the only
600 rounding introduced is in the data themselves.
Rise ≈ $97 billion sales
0.065 12t
Run = $1 billion advertising 19. a. A(t) 300 1 dollars in the account after
12
400
t years
b. A(t) 300(1.066972t) dollars in the account after t
200 Advertising
years
130 expenditures
c. $341.52
1.2 2 3 4 5 6 (billions of dollars)
d. Approximately $22.14 per year
Slope of tangent line: approximately 97 billion dollars e. Approximately $5.53
per billion dollars (sales dollars per advertising dollars)
21. a. R(x) (7.032 104)x2 1.666x 47.130 dollars
Revenue when $6 billion is spent on advertising: ap-
when x hot dogs are sold
proximately $615 billion
b. Cost: c(x) 0.5x dollars when x hot dogs are sold
Revenue when $6.5 billion is spent on advertising: ap-
Profit: p(x) (7.032 104)x2 1.166x 47.130
proximately $662.5 billion
dollars when x hot dogs are sold
(Answers may vary slightly.)
b. $658 billion
c. One possible answer: The answer from part b is more
accurate than in part a because it is calculated rather
than estimated.
A-30 CHAPTER 4 Analyzing Change: Applications of Derivatives
c. Peru: P(t) 85.112(2.01325t);
Brazil: B(t) 73.430(2.61594t)
R(x) c(x) p(x)
For all models t is the number of years since 1980.
(dollars per (dollars per (dollars per
x (dollars) hot dog) hot dog) hot dog) b., c.
200 1.38 0.50 0.88 United States Canada Peru Brazil
800 0.54 0.50 0.04 Rate of change
in 1987 (CPI 5.2 7.5 7984 59,193
1100 0.12 0.50 0.38
points per year)
1400 0.30 0.50 0.80
1988 CPI
143 163 19,134* 136,451*
If the number of hot dogs sold increases from 200 to estimate
201, the revenue increases by approximately $1.38 and
* Based on 1987 data rather than P(7) and B(7) because the models
the profit increases by approximately $0.88. If the differ significantly from the data.
number increases from 800 to 801, the revenue in-
creases by $0.54, but the profit sees almost no increase 25. a. P(A) 0.158A3 5.235A2 23.056A 154.884
($0.04). If the number increases from 1100 to 1101, thousand dollars of profit when A thousand dollars is
the increase in revenue is only approximately $0.12. spent on advertising.
Because this marginal revenue is less than the marginal b. When $10,000 is spent on advertising, profit is in-
cost at a sales level of 1100, the result of the sales in- creasing by $34.3 thousand per thousand advertising
crease from 1100 to 1101 is a decrease of $0.38 in dollars. If advertising is increased from $10,000 to
profit. If the number of hot dogs increases from 1400 $11,000, the car dealership can expect an approximate
to 1401, revenue declines by approximately $0.30 and monthly increase in profit of $34,300.
profit declines by approximately $0.80. c. When $18,000 is spent on advertising, revenue is in-
d. creasing by $12.0 thousand per thousand advertising
Dollars
dollars. If advertising is increased from $18,000 to
1200
Revenue $19,000, the sporting goods company can expect an
approximate monthly increase in profit of $12,000.
27. Essay answers will vary but should include a discussion of
how the rate of change approximation is following the
Cost path of a tangent line instead of following the curve of the
Profit function.
f(x h) f(x)
29. f (x) lim
h→0 h
x
0
Customers
200 800 1100 1400 f(x h) f(x)
f (x) for relatively small h
The marginal values in part c are the slopes of the h
graphs shown in the figure above. For example, at
x 800, the slope of the revenue graph is $0.54 per
h f (x) h
f (x h) f (x)
h
hot dog, the slope of the cost graph is $0.50 per hot
h f (x) f(x h) f(x)
dog, and the slope of the profit graph is $0.04 per hot
dog. We see from the graph that maximum profit is re- Section 4.2
alized when approximately 800 hot dogs are sold.
1. Quadratic or cubic functions could have a relative maxi-
Revenue is greatest near x 1100, so the marginal
mum and/or minimum.
revenue there is small. However, once costs are
3. y
factored in, the profit is actually declining at this sales
level. This is illustrated by the graph. X
23. a. United States: A(t) 0.109t3 1.555t2 10.927t
100.320;
Canada: C(t) 0.150t3 2.171t2
15.814t 99.650; x
The derivative is zero at the absolute maximum point.
Chapter 4 Answers to Odd Activities A-31
5. y 23. a. Using technology, the relative maximum value is ap-
proximately 19.888, which occurs at x 3.633. The
X
relative minimum value is approximately 11.779,
which occurs at x 11.034.
b. The absolute maximum and minimum are the relative
maximum and minimum found in part a.
x
c. g(x)
The derivative is zero at the absolute maximum point
marked with an X. 100
7. y
Relative
X X Relative
maximum
X
minimum
x
-5 20
x -50
The derivative is zero at both absolute maximum points. g'(x)
The derivative does not exist at the relative minimum
point. 15
9. One such graph is y x3.
11. a. All statements are true.
b. f (2) does not exist because f is not continuous at
x 2. x
c. f (x) is less than zero for x 2 because the graph to -5 20
Zero
the left of x 2 is decreasing. -5 derivative Zero derivative corresponds
d. f (2) does not exist because f is not smooth at x 2. corresponds to to relative minimum
13. One possibility is relative maximum
f (x) 25. a. The greatest percentage of eggs hatching (95.6%) oc-
curs at 9.4°C.
b. 9.4°C 49°F
27. a. C(0) 123 cfs and C(11) 331 cfs
x b. The highest flow rate is approximately 388.0 cfs; it oc-
–1 curs when h 8.9 hours. The lowest flow rate is ap-
15. One possibility is proximately 121.3 cfs; it occurs when h 0.4 hour.
29. a. S(x) 0.181x2 8.463x 147.376 seconds is the
f (x)
time it takes an x-year-old average athlete to swim 100
meters free style, 8 x 32.
b. The model gives a minimum time of 48.5 seconds oc-
curring at 23.4 years.
x c. The model reaches a minimum slightly earlier and
–1 3 lower than that reported in the data.
17. a. f (x) 2x 2.5 31. a. Q(p) 316.765(0.949p) dozen roses sold per week
b. A relative minimum of 7.5625 occurs at x 1.250. when the price per dozen is p dollars, 10 p 35
19. a. h(x) 3x2 16x 6 b. E(p) 316.765p(0.949p) dollars spent on roses each
b. A relative maximum of approximately 1.077 occurs at week when the price per dozen is p dollars,
x 0.352. A relative minimum of 108.929 occurs 10 p 35
at x 5.685. c. A price of $19.16 maximizes consumer expenditure.
21. a. f (t) 12 ln(1.5)(1.5t ) 12 ln(0.5)(0.5t ) d. A price of $25.16 maximizes profit.
b. A relative minimum of approximately 23.182 occurs e. Marginal values are with respect to the number of
at t 0.488. units sold or produced. In this activity, the input is
A-32 CHAPTER 4 Analyzing Change: Applications of Derivatives
price, so derivatives are with respect to price and are maximum point of approximately (2.732, 2.183). Thus
not marginals. the absolute maximum is approximately 2.18, and the ab-
33. a. G(t) 0.008t3 0.347t2 6.108t 79.690 million solute minimum is approximately 1.32.
tons is the yearly amount of garbage taken to a landfill
t years after 1975, 0 t 30. Section 4.3
b. G(t) 0.025t 2 0.693t 6.108 million tons per
year is the rate of change in the yearly amount of 1. a. (1982, 25) and (2018, 25); Estimates may vary.
garbage taken to a landfill t years after 1975, b. The input values of the inflection points are the years
0 t 30. in which the rate of crude oil production is estimated
c. Approximately 7.8 million tons per year. to be increasing and decreasing most rapidly. We esti-
mate that the rate of production was increasing most
d. G'(t) rapidly in 1982, when production was approximately
(millions of tons 25 billion barrels per year, and that it will be decreas-
per year) ing most rapidly in 2018, when production is esti-
8 mated to be approximately 25 billion barrels per year.
3. a. Derivative
t
b. Function
0 Years
0 30 since 1975 c. Second derivative
5. a. Second derivative
G(t) has no relative extrema because the derivative b. Derivative
graph exists for all inputs and never crosses the hori- c. Function
zontal axis. 7. f (x) 3; f (x) 0
35. The graph indicates there is an absolute maximum to the 9. c(u) 6u 7; c(u) 6
right of 3 and an absolute minimum to the left of 1. A 11. p(u) 6.3u 2 7u; p(u) 12.6u 7
view of the graph showing more of the horizontal axis 13. g(t) 37 ln(1.05)(1.05t ); g (t) 37(ln 1.05)2(1.05t)
indicates that y 2 is a horizontal asymptote for the 3.2
graph. 15. f (x) x 3.2x1; f (x) 3.2x2
y 17. L(t) 16(1 2.1e3.9t)2(0 2.1e3.9t(3.9))
131.04e3.9t(1 2.1e3.9t)2;
2 L(t) 131.04[e3.9t(2(1 2.1e 3.9t)3
(2.1e3.9t(3.9))) (1 2.1e3.9t)2(3.9e3.9t)]
19. a. f (x) 3x 2 12x 12; f (x) 6x 12
1.5 b. An inflection point may occur when x 2.
21. a. f (x) 37(1 20.5e0.9x)2(20.5e0.9x(0.9))
x 682.65e0.9x(1 20.5e0.9x)2;
−3 1
f (x) 682.65[e0.9x(2(1 20.5e0.9x)3
Use technology to find the absolute extrema, or solve the
(20.5e0.9x(0.9))) (1 20.5e0.9x)2(0.9e0.9x)]
equation
b. An inflection point may occur at x 3.356.
2x(2x2 x 3) 4x 1 23. a. f (t) 98(ln 1.2)(1.2t ) 120(ln 0.2)(0.2t );
y 2 0
(x 2 2)2 x 2
f (t) 98(ln 1.2)2(1.2t) 120(ln 0.2)2(0.2t)
In either case, you should find the absolute minimum
b. f(t) does not have an inflection point.
point of approximately (0.732, 1.317) and the absolute
Chapter 4 Answers to Odd Activities A-33
25. a. g(x) 29. a.
Relative
22 maximum P(t)
(percent)
Inflection
point 100
Point of most
Relative rapid decrease
minimum
10 x
0 3.63 7.33 11.03 15
t
g'(x) 0
7 14.2 25 (°C)
6
P'(t)
(percentage points
per °C)
15
x
7.33 15 t
-2
14.2 25 (°C)
g''(x) -15
2
P''(t)
(percentage points
per °C per °C)
x
7.33 15 10
t
-2 (°C)
14.2 25
b. The inflection point on the graph of g is approxi- -10
mately (7.333, 15.834). This is a point of most rapid
decline. b. Because the graph of P crosses the t-axis twice, there
27. a. The inflection point is approximately (1.838, 22.5). are two inflection points. These are approximately
After approximately 1.8 hours of study (1 hour and (14.2, 59.4) and (23.6, 5.8). The point of most rapid
50 minutes), the percentage of new material being decrease is (14.2, 59.4). (The other inflection point is a
retained is increasing most rapidly. At that time, point of least rapid decrease.) The most rapid decrease
approximately 22.5% of the material has been occurs at 14.2°C, when 59.4% of eggs hatch. At this
retained. temperature, the percentage of eggs hatching is declin-
b. The answer agrees with the one given in the discussion ing by 11.1 percentage points per °C. A small increase
at the end of the section. in temperature will result in a relatively large decrease
in the percentage of eggs hatching.
A-34 CHAPTER 4 Analyzing Change: Applications of Derivatives
31. a. d. The price was declining most rapidly in 1998 at a rate
p(x) of $0.19 per year and increasing most rapidly in
(dollars) 1996 at a rate of $0.55 per year.
33. a. (0.418, 9740.089) is a relative minimum point, and
7.5
(13.121, 20,242.033) is a relative maximum point on
the cubic model.
b. The inflection point is approximately (6.8, 14,991.1).
6.5 c. i. The inflection point occurs between 1981 and
1982, shortly after the team won the National
x Championship. This is when the number of donors
5.75
Years was increasing most rapidly.
4 5.78 8.12 10 since 1990
ii. The relative maximum occurred around the same
p'(x) time that a new coach was hired. After this time,
(dollars per year) the number of donors declined.
35. a. The greatest rate occurs at h 3.733, or approximately
3 hours and 44 minutes after she began working.
1.2
b. Her employer may wish to give her a break after 4
x hours to prevent a decline in her productivity.
0
Years
10,111.102
5.78 8.12 10 since 1990 37. a. H(w) is the total labor hours
1 1153.222e0.728w
−1.2 spent on a construction job after w weeks,
1 w 19.
p"(x) 10,111.102(1153.222)(0.728)e0.728w
b. H(w)
(dollars per year (1 1153.222e0.728w)2
per year) labor hours per week is the rate of change in the total
number of labor hours spent on a construction job af-
2.6
ter w weeks, 1 w 19.
c. H'(w)
(labor-hours
x
Years per week)
0
5.78 8.12 10 since 1990 2000
The maximum and minimum points of the graph of
p and the x-intercepts of the graph of p correspond
to the inflection points of the graph of p. w
0
b. The x-intercepts of p are x 5.78 and x 8.12. The 0 20 Weeks
price was declining most rapidly in January of 1999
(x 8.12) at a rate of $0.19 per year and increasing The derivative gives the manager information about
most rapidly in September of 1996 (x 5.78) at a rate the number of labor hours spent each week.
of $0.57 per year.
c. The price was declining most rapidly at the end of
1994 at a rate of $1.45 per year and increasing most
rapidly at the end of 2000 at a rate of $2.10 per year.
Chapter 4 Answers to Odd Activities A-35
d. The maximum point is approximately (9.685, f. The year with the smallest rate of change is 1984, with
1840.134). In the tenth week the most labor hours are g(14) 120 million tons of garbage, increasing at a
needed. That number is H(10) 1816 labor hours. rate of g(14) 1.4 million tons per year.
e. The point of most rapid increase on the graph of H is 41. a. The first differences are greatest between 6 and 10
(7.876, 1226.756). This occurs approximately 8 weeks minutes, indicating the most rapid increase in activity.
into the job, and the number of labor hours per week is 1.930
b. A(m) U/100L m minutes
increasing by approximately H(8) 513 labor hours 1 31.720e0.439118m
per week per week. after the mixture reaches 95°C; The inflection point is
f. The point of most rapid decrease on the graph of H is (7.872, 0.965). After approximately 7.9 minutes, the
(11.494, 1226.756). This occurs approximately 12 activity was approximately 0.97 U/100L and was in-
weeks into the job, and the number of labor hours per creasing most rapidly at a rate of approximately
week is changing by approximately H(12) 486 0.212 U/100L/min.
labor hours per week per week. Because the graph of g crosses the t-axis at 13.68, we
g. By solving the equation H(w) 0, we can find the know that input corresponds to an inflection point of
input values that correspond to a maximum or mini- the graph of g. Because g(t) is a minimum at that
mum point on the graph of H, which corresponds to same value, we know that it corresponds to a point of
inflection points on the graph of H, the weekly labor slowest increase on the graph of g.
hour curve. 43. The graph of f is always concave up. A concave-up
h. The second job should begin approximately 4 weeks parabola fits this description.
into the first job. 45. a. The graph is concave up to the left of x 2 and con-
39. a. Between 1980 and 1985, the average rate of change was cave down to the right of x 2.
smallest at 1 million tons per year. b. f (x)
1
b. g(t) 0.008t3 0.347t2 6.108t 79.690 million
tons of garbage taken to a landfill t years after 1970,
0 t 30
c. g(t) 0.051t 0.693
d. Solving g(t) 0 gives t 13.684 and g(13.684) 120 x
million tons of garbage. 0 2 4
e.
f2(x)
g' (t)
(millions of tons
per year)
10
x
0 2 4
47. Cubic and logistic models have inflection points.
t Section 4.4
1 Years
0 13.68 30 since 1970 df dx
1. 3
g'' (t) dt dt
(millions of tons dk dx
3. 12x
per year per year) dy dy
dg dx
5. 3e3x
1 dt dt
t df dx
7. 62(ln 1.02)(1.02x)
Years dt dt
13.68 30 since 1970
-1 9.
dh
dy
6 da
a dy
da 1
6 ln a 6 ln a
dy a
da
dy
A-36 CHAPTER 4 Analyzing Change: Applications of Derivatives
ds rh dh Chapter 4 Concept Review
11.
dt r 2 h2 dt
1. a. T has a relative maximum point at (0.682, 143.098)
r
13. 0
r h
2 2
dr
r h
dt
dh
dt
r2 h2
dr
dt
and a relative minimum point at (3.160, 120.687).
These points can be determined by finding the values
15. a. Approximately 52.4 gallons per day of x between 0 and 6 at which the graph of T crosses
dw the x-axis.
b. 0.4323; The amount of water transpired is
dt b. T has two inflection points: (1.762, 132.939) and
increasing by approximately 0.4323 gallon per day per (5.143, 149.067). These points can be determined by
year. In other words, in 1 year, the tree will be tran- finding the values of x between 0 and 6 at which the
spiring approximately 0.4 gallon more each day than it graph of T crosses the x-axis. These are also the points
currently is transpiring. at which T has a relative maximum and a relative
45 minimum.
17. a. B points
0.00064516h2 c.
dB 90 dh
b.
dt 0.00064516h3 dt T(x)
dB (thousand tourists)
c. 0.2789 point per year
dt
19. a. Approximately 0.0014 cubic foot per year 170
b. Approximately 0.0347 cubic foot per year 160 Inflection
Relative
point
M 5/3 M 5/3 150 maximum
21. a. L K2/3 Inflection
48.10352 K0.4 48.10352 140 point Relative
5/3 2 x
dL M dK 130 minimum
b. K5/3 Years
dt 48.10352 3 dt 120
0 0.68 1.76 3.16 5.14 6 after
c. The number of worker hours should be decreasing by
1988
approximately 57 worker hours per year. T'(x)
23. The balloon is approximately 1529.7 feet from the ob- (thousand tourists
server, and that distance is increasing by approximately per year)
1.96 feet per second.
25. The runner is approximately 67.08 feet from home plate, 50
x
and that distance is decreasing by approximately 9.84 feet Years
per second. 0
0.68 3.16 5.14 6 after
27. a. Approximately 4188.79 cubic centimeters 1988
b. By approximately 167.6 cubic centimeters per -50
minute T''(x)
29. Approximately 5.305 centimeters per second (thousand tourists
V V per year per year)
31. Begin by solving for h: h 2 r 2
r
20
Differentiate with respect to t (V is constant): x
dh V dr Years
(2r3) 0
dt dt 1.76 5.14 6 after
1988
dh r2h dr -20
Substitute r2h for V: ( 2r3)
dt dt
dh 2h dr d. The number of tourists was greatest in 1994 at 166.8
Simplify: thousand tourists. The number was least in 1991 at
dt r dt
dr r dh 120.9 thousand.
Rewrite: e. The number of tourists was increasing most rapidly in
dt 2h dt
33. One possible answer: The most important step in the 1993 at a rate of 23.1 thousand tourists per year. The
method of related rates is correctly identifying the inde-
pendent and dependent variables.
Chapter 4 Answers to Odd Activities A-37
number of tourists was decreasing most rapidly in b. i. The heights are in feet per mile per hour, and the
1990 at a rate of 13.3 thousand tourists per year. widths are in miles per hour.
1
2. a. (4.5 thousand people per year) 4 year 1.125 5.
ii. The area is in feet.
a. Thousand people
thousand people b. Thousand people
1
b. 202 2(4.5) 204.25 thousand people c. Thousand people
3. a. G(x) (5.051 105)x3 0.007x2 0.085x 7. a. This is the change in the number of organisms when
105.027 is the number of males per 100 females in the the temperature increases from 25°C to 35°C.
United States x years after 1900, 0 x 100. b. This is the magnitude of the change in the number of
b. G(x) (30.303 105)x 0.014 organisms when the temperature increases from 30°C
c. The output of the model exhibits most rapid decline to 40°C.
when x 44.598. The gender ratio was decreasing 9. a., b. Between 0 and 300 boxes and between 400 and 600
most rapidly in 1945 when the gender ratio was ap- boxes
proximately 99.8 males per 100 females. The gender c. NA
ratio was decreasing by approximately 0.216 males d 300
per 100 females per year. e. 400
4. f. Approximately 350
h'(x)
g. dollars
h. less
11. a. On the horizontal axis, mark integer values of x be-
tween 0 and 8. Construct a rectangle with width from
x 0 to x 1 and height f (1). Because the width of
Input of relative the rectangle is 1, the area will be the same as the
maximum of h height. Repeat the rectangle constructions between
x each pair of consecutive integer input values. Note
a 0 b that for the rectangles that lie below the horizontal
Inputs of inflection axis, the heights are the absolute values of the function
points of h
values. Also note that the fourth rectangle has height 0.
Sum the areas (heights) of the rectangles to obtain the
area estimate.
f(x)
ds dv
5. 0.000013(2wv)
dt dt
31.2 feet per second
The length of the skid marks is decreasing by 31.2 feet per
second.
5 6 7 8
CHAPTER 5 x
1 2 3 4
Section 5.1
1. a. Thousand bacteria per hour
b. Hours
c. Thousand bacteria
b. Repeat part a, except that the height of each rectangle
d. Thousand bacteria
is determined by the function value corresponding to
e. Thousand bacteria
the left side of the interval. In the case of the first rec-
3. a. The area would represent how much farther a car going
tangle, the height is f (0). When we use left rectangles,
60 mph would require to stop than a car going 40 mph.
the fifth rectangle has height 0.
A-38 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
f(x) b. Each rectangle has a width of 25 years.
Midpoint height
(billion barrels Rectangle area
Interval per year) (billions of barrels)
1900–1925 0 0
1925–1950 3 75
5 6 7 8 1950–1975 9 225
x
1 2 3 4 1975–2000 24 825
2000–2025 13 825
2025–2050 4 225
2050–2075 1 75
13. Divide the interval from a to b into four equal subinter-
2075–2100 0 25
vals. Determine the midpoint of each subinterval, and
substitute into the function to find the heights of the rec- Total area of rectangles 1625 billion barrels
Total oil production 1625 billion barrels
tangles. Multiply each height by the width of the subin-
tervals, and add the four resulting areas.
f (x) c. The graph A estimate is 175 billion barrels above the
journal’s estimate. The graph B estimate is 275 billion
barrels above the journal’s estimate.
17. a. Velocity
(feet per
minute)
880
0 x
a b
15. Possible solution:
Time
a. Each rectangle has a width of 25 years. 0
(minutes)
0 1 3 3.5
Midpoint height
(billion barrels Rectangle area b. 2420 feet
Interval per year) (billions of barrels) 1
c. The robot traveled 2420 feet during the 3 -minute
1900–1925 0 0 2
experiment.
1925–1950 3 75
1950–1975 9 225 19. a.
1975–2000 33 825 m(t)
(megawatts)
2000–2025 33 825
40
2025–2050 9 225
35
2050–2075 3 75 30
2075–2100 1 25 t
25
Hours
Total area of rectangles 2275 billion barrels 20
since
Total oil production 2275 billion barrels 0 6 12 18 24
midnight
Chapter 5 Answers to Odd Activities A-39
21. a.
Height Area
Interval (megawatts) (megawatt hours) Consumption rate
(pounds per person
0 to 3 24 72.0
per year)
3 to 6 23.5 70.5 t
5 10 15 20 Years
6 to 9 27.5 82.5 0 since
9 to 12 32 96.0 1980
12 to 15 37 111.0 −0.1
15 to 18 40 120.0
18 to 21 39.5 118.5
−0.2
21 to 24 34.5 103.5
b. 9.504
Total area 774 megawatt hours
c. Per capita consumption of cottage cheese decreased by
about 9.5 pounds per person between 1980 and 1996.
b. d. No; We cannot determine the per capita consumption
m(t) of cottage cheese in 1996 unless we know the value in
(megawatts) 1980 (or in some other year between 1980 and 1996).
If we knew the 1980 value, we would subtract 9.5 to
40
obtain the 1996 value.
35
23. a.
30
t T(h)
25
Hours (°F per hour)
20
0 6 12 18 24
since
midnight 15
10
5
Height Area h
0
Interval (megawatts) (megawatt hours) 0.5 1.0 1.5 Hours
−5
0 to 3 23.5 70.5
−10
3 to 6 27.5 82.5
6 to 9 32 96.0 The graph crosses the horizontal axis at h 0.907
hour.
9 to 12 37 111.0
b. The area below the axis represents the amount the
12 to 15 40 120.0 temperature declined from the time the storm began.
15 to 18 39.5 118.5 c. The area above the axis represents the amount by
18 to 21 34.5 103.5 which the temperature rose from the time it stopped
declining until 1.5 hours after the storm began.
21 to 24 23 69.0
0.907 0
Total area 771 megawatt hours d. Each rectangle has width 0.302 hour.
3
The area of the region is about 2.775°F.
1.5 0.907
c. Answers will vary. The exact area to three decimal e. Each rectangle has width 0.198 hour.
3
places is 774.426, so both estimates are very close.
The area of the region is about 4.624°F.
0.907
f. Using a limit of sums, T (h)dh
4.27 F
0
g. Using a limit of sums,
1.5
T (h)dh 3.25 F
0.907
A-40 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
1.5 d. Life expectancy for women increased approximately 5.2
h. T (h)dh
4.27 3.25
1.02. To determine years from 1970 through 2010. The life expectancy in
0.907
the temperature when h 1.5, we would have to know 1970 is needed to determine the life expectancy in 2010.
the temperature at h 0. One and one half hours after 29. a.
the storm began, the temperature was approximately Number of Approximate
1°F lower than it was at the beginning of the storm. midpoint rectangles accumulated change
25. a. 5 54.908
P(w) 10 55.307
(percentage points 20 55.407
per pound)
40 55.432
20 25 30 35 40 w
0 80 55.438
Pounds
− 0.2
From 1990 through 2001, factory sales increased by
− 0.4
approximately $55.4 billion.
− 0.6 11
− 0.8
b. s(x)dx 55.4
0
c. Approximately $98.4 billion
b. The percentage of low-birthweight babies was declin- 695.606
31. a. S(t) thousand cassettes per
ing as the mother’s weight gain increased from 18 to 1 0.081e0.495438t
43 pounds. month gives the sales of a popular movie t months
c. The signed area is approximately 4.8. When the after release, 2 t 18.
amount of weight gained during a woman’s pregnancy b.
increased from 18 to 43 pounds, the percentage of
Number of rectangles Approximation of area
low-birthweight babies decreased by approximately
4.8 percentage points. 5 3617.9
27. a. 10 3623.6
E(t) 15 3625.4
(years per year)
0.40 The estimates are about 3,618,000 cassettes, 3,624,000
cassettes, and 3,625,000 cassettes.
0.35
c. The sum of the sales figures (iii) will be most accurate
0.30
because it will give the exact total. (Note that the text
0.25 does not give all of the data needed to perform this
0.20 calculation.)
0.15 33. a. n 5: 9859 labor hours
0.10 n 10: 10,097 labor hours
t n 20: 10,100 labor hours
0.05
Years b. iii
since 35. The accumulated change of a function f with input x over
0 5 10 15 20 25 30 35 40
1970 an interval from a to b is the sum of the signed areas of re-
b. E(t) has a minimum value when t 27.5. gions bounded by the graph of f , the horizontal axis, the
c. Area 10 E(5) 10 E(15) 10 E(25) vertical lines at a and b, and any points where the graph of
10 E(35) 5.2 years. f intersects the horizontal axis. This accumulated change
f (x)dx.
b
is also defined as the definite integral
a
Chapter 5 Answers to Odd Activities A-41
Section 5.2 5. a. No
b. The peak corresponds to the time when the number of
1. a.
subscribers was increasing most rapidly.
Distance c. The number of new subscribers t days after the end of
(feet) the twentieth week
300 d. 280 subscribers
200 e.
100 Week t (days) Area Week t (days) Area
Time 4 28 350 28 196 8820
0
5 10 (seconds)
-100
8 56 924 36 252 10,430
12 84 1960 44 308 10,920
-200
16 112 3500 52 364 11,060
-300
20 140 5390
f (t) dt
x
b. D (x) f. t
5 ∫140 n (x) dx
c. The accumulation function gives the distance traveled Subscribers
between 5 seconds and x seconds. For times before 5
5670
seconds, the accumulation function is the negative of
the distance traveled, because we are looking back in
time.
3. a. The area between days 0 and 18 represents how much t
the price of the technology stock declined ($15.40 per 0
Days
share) during the first 18 trading days of 2003. 140
b. The area between days 18 and 47 represents how much
the price of the technology stock rose ($55.80 per
share) between days 18 and 47. -5390
c. $40.40 more 7. One possible answer:
d. $11.10 less
Rainfall
e.
(inches)
x x
x r (t) dt x r (t) dt 5.5
0 0
0 0 35 15.0
8 7.1 47 40.4
18 15.4 55 29.3 3
f.
x
∫ 0 r (t) dt
Price
(dollars)
Hours
40.4 0 3 6
29.3 9. a. b.
x x
∫A f (t) dt ∫B f (t) dt
15
x
x B A
8 18 35 47 55 Days
-7.1
-15.4 x
g. $156.30 A
A-42 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
11. a. b. b. h
x x Input hours
∫0 f (t) dt ∫A f (t) dt
Rule c
x
A B A B
Output
c(h)
milligrams per liter
c. x c. The concentration of a drug in the bloodstream is c(h)
∫B f (t) dt
milligrams per liter h hours after the drug is given.
25. A negative rate of change indicates a loss in the additional
accumulated amount. This loss is evidenced in a decrease
x of the accumulation graph.
A B
Section 5.3
1. a. ii b. i c. iii 3. a. iii b. i c. ii
x d 3t2 x
13. x
5.
d
dx 2
3tdt
dx 2 2
∫0 f (t) dt
d 3x2
dx 2
6 3x
7. The antiderivative of a derivative formula for a given
function returns the original function plus a constant.
x
9. 19.4(1.07x)dx
19.4(1.07x)
ln 1.07
C
11.
4(2x)
[6ex 4(2x)]dx 6ex C
ln 2
15. Derivative graph: b; accumulation graph: f
13.
10x 2 3
1
17. Derivative graph: f ; accumulation graph: e (10x 4x2 8)dx 4 x2 8x C
ln 10 3
19. Derivative: left table; accumulation function: right table
600(0.93m)
Million dollars of revenue 15. S(m) C CDs m months after the begin-
21. a. ln 0.93
Thousand advertising dollars
ning of the year
b. m 0.38(0.01x)
thousands of 17. C(x) 0.8 ln x K dollars per unit when
ln 0.01
advertising
Input
x units are produced
dollars
1
19. F(t) t3 t2 20
3
1
21. F(z)
z
ez
3
2
e2
Rule R
23. w(t) 7.37 ln t (26 7.37 ln 7) 7.37 ln t 11.66
grams is the weight of the mouse at age (t 2) weeks. The
specific antiderivative is the formula for the accumulation
Output
R(m)
function of W passing through the point (7, 26).
(1.67 10
4) 3
millions of dollars 25. a. G(t) t 0.01t2 0.10t 131.71
of revenue 3
males per 100 females gives the gender ratio for the
c. When m thousand dollars are being spent on advertis-
United States t years after 1900.
ing, the annual revenue is R(m) million dollars.
b. This specific antiderivative is the formula for the ac-
Milligrams per liter
23. a. cumulation function of g passing through the point
Hour
(70, 94.8).
Chapter 5 Answers to Odd Activities A-43
27. a., b. Velocity: v(t) 32t feet/second 11. a. 0.5 3.5
Distance: s(t) 16t2 540 feet -1000 x
where t is the number of seconds after the penny
was dropped
c. Solving for t in s(t) 0, we obtain t 5.8 seconds.
The penny will hit the ground approximately 5.8 sec-
onds after it was dropped. -1300
29. a. The impact velocity is approximately 64.99 feet/
second, or 44.31 mph. f (x)
b. Air resistance probably accounts for the difference.
b. No; Because f(x) is negative between x 0.5 and x
31. a. N(x) 593 lnx 138x 748.397 new employees at
3.5, the area from a to b equals
an Internet company x years after 1996, 1 x 6
965.27(1.079 )dx 3378.735
3.5
b. The function in part a applies from 1997 (x 1) x
through 2002 (x 6). 0.5
c. 965.27(1.079 )dx 3378.735
3.5
c. Approximately 1753; not necessarily; One possible x
answer: The number of employees hired does not take 0.5
P(x)dx 2305.357 million calls; Between 1985 and
15
into account the number of people who left the 13.
company. 5
1995, the number of international calls billed in the
Section 5.4 United States increased by about 2.3 billion.
(9.907x 40.769x 58.492)dx 195.639
5
1. c. Find a specific antiderivative and evaluate in year t 15. 2 million
3. b. Find a general antiderivative and evaluate from a to x 0
5. b. Find a general antiderivative and evaluate from a to x dollars; Between 1987 and 1992, the corporation’s rev-
7. a. Find a derivative and evaluate at a. enue increased by $195.639 million.
70
9. a. f (x) 17. a. (0.00241t 0.02905)dt 7.938 cm; In the first
0
3 70 days after April 1, the snow pack in the Northwest
Territories in Canada increased by about 7.938 cm.
x b. Between the 72nd and 76th days after April 1, the snow
-1 1 2
pack in the Northwest Territories in Canada decreased
by about 22.368 cm.
76
c. It is not possible to find the value of N(t)dt because
0
we have no information about the rate of change in the
-7 snow pack in the Northwest Territories in Canada be-
tween the 70th and 72nd days after April 1.
b. No; Because f (x) is negative between x 1.054 and x
0.8955
2, the area from a to b is approximately 19. a. T(h)dh 1.48°F; The temperature in the
0
1.054 2
f(x)dx f(x)dx 1.603 2.218 3.821 museum rose about 1.48° F in the 54 minutes follow-
1 1.054 ing 8:30 A.M.
c.
2
(1.3x3 0.93x2 0.49)dx
1
2
1.3x4 0.93x3
0.49x
4 3 1
1.74 (1.125)
0.615
A-44 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
1.75 29. a. When the amount invested in capital increases from
b. T(h)dh 1.61 F; The temperature in the
0.8955 $1500 to $5500, profit increases by approximately
museum decreased by about 1.65° F between 9:24 $1.33 thousand.
5.5
A.M. and 10:15 A.M.
b. Area [r (x) c (x)]dx 13.29
c. No, the highest temperature reached was 71 1.48 1.5
72.48°F. 31. a. The population of the country grew by 3690 people in
21. a. f(x) 0.161(1.076x) trillion cubic feet per year is the January.
rate of change in natural gas in the United States (ex- b. The population declined by 9720 people between the
cluding Alaska) between 1900 and 1960, where x is the beginning of February and the beginning of May.
number of years after 1900. c. 6030 people
b. From 1940 through 1960, the amount of natural gas d. Because the graphs intersect, the area of R1 represents
produced increased by 138.3 trillion cubic feet. an increase in population, and the area of R2 repre-
60 sents a decrease. The net change is the difference
c. f(x)dx
40 Area of R1 area of R2.
23. a. C(x) (7.714 105)x2 0.047x 8.940 dollars The total area is the sum:
per CD is the marginal cost of production of an addi- Area of R1 area of R2.
tion CD, when x CDs are produced each hour, 100 33. a. Before fitting models to the data, add the point (0,0),
x 300 and convert the data from miles per hour to feet per
5
7.714 10 3 0.047 2 second by multiplying each speed by
b. C(x) x x 8.940x
3 2 5280
1 mile 3600 seconds . The speed of the Supra after t
feet 1 hour
143.893 dollars is the production cost of CDs, when x seconds can be modeled as
CDs are produced each hour, 100 x 300 s(t) 0.702t2 20.278t 2.440 feet per second
300
c. C (x)dx 196.14 dollars; When the number of The speed of the Carrera after t seconds can be
200 modeled as
CDs produced each hour increases from 200 to 300, c(t) 0.643t2 18.963t 5.252 feet per second
the hourly production cost is increased by about $196. b. Approximately 17.96 feet
25. a., b. c. Approximately 18.04 feet
30 35. a.
f (x)
20 Rate of change of revenue
(billions of dollars per year)
10
g(x) 2.5
x UPS
2 4 6 2
c. 21.33
FedEx
27. a., c.
12 1
8 f (x) t
Years
4 0 since
0 2 4 6 8
g(x) 1993
0 x
0 1 2 3
FedEx: F(t) 0.026t3 0.198t2 0.06t 0.317
b. f(x) g(x) → x 1.134.
billion dollars per year t years after
1993
[f(x) g(x)]dx 2.812
3
d. Difference = UPS: U(t) 0.15t 1.022 billion dollars
0.5 per year t years after 1993
e. Area 4.172
Chapter 5 Answers to Odd Activities A-45
b. Area of region on left $1.28 billion Section 5.5
Area of region on middle $1.20 billion
1. a. V(t) 1.664t3 5.867t2 1.640t 60.164 mph
Area of region on right $1.59 billion
is the average speed of motorists during rush hour be-
Between the beginning of 1993 and late 1996 (t 2.8),
tween 4:00 P.M. and 7:00 P.M., where t is the number of
UPS’s accumulated revenue exceeded that of FedEx by
hours after 4 P.M.
approximately $1.28 billion. Between late 1996 and the
b. 68.99 mph
spring of 2000 (t 6.3), FedEx’s accumulated revenue
c. 72.23 mph
exceeded that of UPS by approximately $1.2 billion.
3. a. f(x) 1.822(0.899x) dollars per minute is the rate of
From the spring of 2000 until the end of 2001, UPS’s ac-
change in the most expensive charge for a 2-minute
cumulated revenue exceeded that of FedEx by about
long-distance phone call using a carrier between 1982
$1.59 billion.
and 2000, where x is the number of years since 1980.
[F(t) U(t)]dt $1.68 billion. This value is the
8
c. b. $0.99 per minute
0
c. $0.65 per minute
net amount by which UPS’s accumulated revenue ex-
5. a. 87.8 million people
ceeded that of FedEx between 1993 and 2001.
b. population 87.8 → t 94.1. This corresponds to
0 tons per year when 0 x 5 early 1995.
557.960 c. 2.287 million people per year
37. a. f(x)
1 91.202e0.318025x 7. a. 100.6 yearly accidents per year
tons per year when x 5 b. 2810.5 yearly accidents
x years after 1990 c.
b., c. Accidents
Carbon production Average rate
and absorption = Slope of line
3968 of change
(tons per year)
Absorption Average
rate 2810.5
number
Production
rate 1754
iii
246 x
190 Years
i x
ii 01 22 since
Years 1975
since
5 7 12.1 20 9. a. 2.28 feet per second squared
1990
Absorption Production b. 129.7 feet per second
=
rate rate
c. 4540.7 feet
d. i. 2054.9 tons d. 4540.7 feet
ii. 2137.1 tons
iii. 1269.0 tons
e. No, The factory produces 4192 tons, and the trees ab-
sorb 3406 tons. This does not comply with the federal
regulation.
39. The height of a rectangle is found by subtracting the
height of the bottom line from the height of the top line.
For example, the height of the left rectangle is
3 (4) 7 and the height of the right rectangle is
1 (3) 2.
A-46 CHAPTER 5 Accumulating Change: Limits of Sums and the Definite Integral
e. 19. Consider the two graphs of a function f shown below,
a(t) where A is the average value of f from a to b, and k is an
(ft/sec2) arbitrary constant.
23 f (x)
k
Average acceleration 0 x
2.28 t a b
0 35 (seconds)
We know that the areas of the two shaded regions are
equal. If we remove from each graph the rectangular
–10
region with height k and width b a, the areas of the re-
sulting regions are still equal, because we have removed
v(t) the same area from each. See the following graphs.
(ft/sec) f (x)
180
A
Average
129.7
velocity
k
cant 0 x
of se
Slope average a b
line = eration
accel t It is true for the graphs in this section with vertical axis
0 35 (seconds) shown from k rather than from zero that the area of the
region between the function and y k from a to b is the
The area of shaded region the distance traveled same as the area of the rectangle with height equal to the
the area of a rectangle with height 129.7 ft/sec and average value minus k and width equal to b a.
width 35 seconds.
11. a. V(t) 1.033t 138.413 meters per second is the rate Section 5.6
of change of the growth of a crack, t microseconds af-
ter the experiment began, 0 t 60.
1. 2e dx e 2x 2xC
13.
b. 174.58 meters per second
a. B(t) 0.030t2 0.718t 3.067 mm Hg per hour is
3. 2xe dx e C
x2 x2
the rate of change of diastolic blood pressure for a pa-
(1 e ) dx (1 3 e ) C
x 3
5. x 2
tient with untreated hypertension during one day,
where t is the number of hours after 8 A.M.
2 2 2 dx ln12 22 ln 22 dx ln12 ln(2 2) C
x x
7. x
b. 0.21 mm Hg per hour x x
c. 93.4 mm Hg
ln xdx 2.55
4
9.
15. 0.32 seconds per year 1
17. a. The average of the CO concentration is highest (ln 5)2 (ln 2)2
11. Exactly
between 10 A.M. and 6 P.M. 2 2
b. C(x) 0.248x2 4.291x 3.86 ppm is the con-
2xln(x 1)dx 3.66
2
13. 2
centration of carbon monoxide in the air in a certain 1
metropolitan area between 10 A.M. and 6 P.M., where x 15. Exactly ln17 ln10
is the number of hours since 6 A.M.
6 2x2
17. dx 8.76
1 x 1
Average concentration between 10 A.M. and 6 P.M. is 2
16.8 ppm. 13
c. Severe pollution warning 19. Exactly 12
Chapter 5 Answers to Odd Activities A-47
Chapter 5 Concept Review
4. a. R(t) 10 3.2
3
t3
2
t 50.7t 5000 cubic
93.3 2
1. a. r(t) feet is the amount of oil in the tank after t minutes.
(cubic feet b. R(10) R(0) 41,053.3 ft3
per minute)
7500 c. R(t) 150,000 → t 28 minutes
5. a. $2598.60
5750
b. At the end of the third quarter of the third year, the
$10,000 investment had increased by $2598.60 to a to-
4000
tal value of $12,598.60.
c. $944.94 per year
2250 88
6. [m(t) w(t)] dt $123 thousand
t 79
500
Minutes Between the beginning of 1980 and the end of 1988, the
0 5 10 15 20 25
wages of a man earning the average full-time wage would
b. 139,237.5 cubic feet have increased by about $123,000 more than the wages of
c. In the first 25 minutes that oil flowed into the tank, the a woman earning the average full-time wage.
amount of oil in the tank increased by approximately
139,238 cubic feet.
CHAPTER 6
2. a. S(t) 1.643t2 16.157t 0.2 miles per hour is
the speed at which a hurricane is moving toward the Section 6.1
east coast of Florida, where t is the number of hours
1. 15
since 12 midnight, 1 t 5
3. 0.3
b.
n Sum 5. 0.02
5 127.131 7. Diverges
9. Diverges
10 126.869
11. Diverges
20 126.803 13. a. Approximately 0.0002 milligram
40 126.786 Approximately 0.0015 milligram
1.55(0.9999999845t) 106
N
Limit 126.8 miles b. r(t)dt lim
0 N→ ln(0.9999999845) 0
3. a. The area beneath the horizontal axis represents the de-
crease in weight during the diet. 99.99999923 100 milligrams
b. The area above the axis represents the increase in 15. a. p $28.04
D(p)dp
weight between weeks 20 and 30. b. C qp0
c. The person’s weight was 11.3 pounds less at 30 weeks p0
150(28.04)
than it was at 0 weeks. 499.589(0.958p)dp
d. Weight
28.04
$7702 million
(pounds)
0 10 20 30 Consumers are willing and able to spend about $7.7
Weeks million for 150,000 books.
0.1e 0.1e
N
-11.3 17. 0.1xdx lim
0.1xdx
0 N→ 0
lim [
e
-26.7 N
e. The graph in part d is the overall change in the per- N→
lim e 01x
0 N→
0.1N ( e0)]
son’s weight as a function of the number of weeks af- lim (
e
0.1N) lim (e0) 0 1 1
ter the beginning of dieting. N→ N→
A-48 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
Section 6.2 b. T(t) 2.04(0.83)30t million terns born t years after
1. a. i. R(m) 0.2 $47,000
12 $783.33 per month
1979.
c. 10.91 million terns
R(m) 0.2 12 100m 783.33 20m
47,000 23. a. None
ii.
dollars a month after m months b. S(t) (60 0.5t)(0.67)50t thousand seals born t
47,000
iii. R(m) 0.2 12 (1.005m) 783.33(1.005m) years from now.
c. 90.5 thousand seals
dollars a month after m months
b. i. $53,493.40 Section 6.3
ii. $92,082.72
iii. $61,818.49 1. a. The demand function
The first option is the only one that will not result b. The supply function
in the amount needed for the down payment. c. The producers’ surplus
3. $11.2 billion d. The consumers’ surplus
5. a. $51.5 billion 3. a. To find the price P above which consumers will pur-
b. $27.8 billion chase none of the goods or services, either find the
6000e smallest positive value for which the demand function
6
7. a. 0.0634(6x)dx $43,804.70 is zero, D(p) 0, or, if D(p) is never exactly zero but
0
71 0.0634 (72m) approaches zero as p increases without bound, then let
b. 5001
m0 12
$43,896.84 P → .
c. Part b; One possible answer: Individuals do not gener- b. The supply function, S, is often a piecewise continuous
ally have access to continuous compounding. function with the first piece being the 0 function. The
9. a. r(q) 82.1(1.05q)(0.15) million dollars per quarter q value p at which S(p) is no longer 0 is the shutdown
quarters after the third quarter of 2002 price. The shutdown point is (p1, S(p1)). When S is a
b. R(q) 82.1(1.05q)(0.15)(1.09)16q million dollars
per quarter for money invested q quarters after the continuous function, the shutdown point is (0, S(0)).
third quarter of 2002 c. The market equilibrium price, p0, can be found as the
c. If the investment begins with the fourth-quarter 2002 solution to S(p) D(p). That is, it is the price at
profits, then the initial investment is based on a profit which demand is equal to supply. The equilibrium
of (82.1)(1.05) $86.205 million. Thus we calculate point is the point (p0, D(p0)) (p0, S(p0)).
15 5. a. q0 27.5 thousand units
86.205(0.15)(1.05q)(1.09)16q $629.8 million.
q0
b., c.
q
11. a. $79.87 thousand
Quantity
b. $58.29 thousand
(thousands)
13. a. $28,324.60
50
b. $28,445.37
c. Part b; One possible answer: Individuals do not gener- 40 S (p)
ally have access to continuous compounding.
15. a. $5.4 billion 30
q0
b. $6.1 billion
20
c. $11.2 billion
17. a. $7.3 billion 10
p
b. $5.6 billion Price
c. CSX was optimistic about its influence on Conrail, (dollars
10 20 30 40 50
while Conrail assumed that it would decline. p0 per unit)
19. $5.2 million
(a) Mathematician’s viewpoint
21. Answers given are based on the end of 2007.
a. 0.14 million terns
Chapter 6 Answers to Odd Activities A-49
p b., c.
Price q
(dollars Quantity
per unit) (millions)
50 25
D (p) S (p)
40 S (p) 20
Producers’
30 15 surplus
p0
20 10
Consumers’
5 surplus
10 p
q Price
Quantity
10 20 30 40 50 (thousands) 1 2 3 4 5 (dollars
q0 per unit)
(b) Economist’s viewpoint Mathematician’s viewpoint
p
7. a. Price
q
Quantity (dollars per unit)
(millions) 5
Consumers’
25 S (p)
D(p) S (p) 4 surplus
20
3
Producers’
15 surplus
2
10
1 D(p)
q (p , q ) q
5
p Quantity
Price 5 10 15 20 25 (millions)
(dollars
p0 1 2 p 3 4 P=5 Economist’s viewpoint
per unit)
D(p)dp million dollars
2.5 5
Mathematician’s viewpoint d. TSG S(p)dp
0 2.5
p 9. a. D is an exponential demand function and so does not
Price have a finite value p at which D(p) 0. Thus the
(dollars per unit) equation does not indicate a price above which con-
P=5 sumers will purchase none of the goods or services.
b. $6128.6 thousand
4 S (p)
c. 17.4 thousand fans
d. $4331.5 thousand
3
p (p , q ) 11. a. D(p) 0.025p2 1.421p 19.983 lanterns when
2 the market price is $p per lantern.
b. $109.18
1 D(p) c. $31.89
q
13. a. unit elasticity: $249.50
p0 Quantity
5 10 15 20 25 (millions) b. inelastic: p $249.50
q elastic: p $249.50
Economist’s viewpoint 15. a. unit elasticity: $9.34
p* 2.5 dollars per unit, q * 6.25 million units, b. inelastic: p $9.34
p1 0 dollars per unit, P 5 dollars per unit elastic: p $9.34
A-50 CHAPTER 6 Analyzing Accumulated Change: Integrals in Action
17. a. $40: 18.4 thousand answering machines; $50: 300 c. y
thousand answering machines
b. $9981 thousand; $3131 thousand 1
0 hundred prints when p 5 0.75
19. a. S(p) 0.300p2 3.126p
10.143 hundred prints when p 5 0.50
where p hundred dollars is the price of a print
0.25
b. $837.12 x
c. Producers revenue $148.5 thousand 0 Gallons
0 1 µ 2 2.5 (hundreds)
Producers surplus $27.3 thousand
21. a. $408.3 hundred 7. Because the f(x) values are all non-negative and
f(x)dx 1, the probability (which is the area between
b. 297 sculptures; No
c. $4542.2 hundred
23. a. D(p) 499.589(0.958086p) thousand books when the the graph of f and the input axis) must always be between
market price is $p per book 0 and 1. Another view of this explanation is that the prob-
ability of some occurrence is the proportion of time it is
0 thousand books when p 18.97 expected to happen, and all proportions are fractions be-
b. S(p) 0.532p2 20.060p tween 0 and 1.
309.025 thousand books when p 18.97
30
9. a. P(20 t 30) 0.2e
0.2tdt 0.016
where $p is the price of a book 20
c. Approximately $27.15; 156.2 thousand books b. P(t 10) 0.865
d. $4728.6 thousand c. P(t 15) 1 P(t 15) 0.050
11. a. Mean 2 minutes
Section 6.4 b. Standard deviation 0.8 0.89 minute
c. 0.316; The likelihood that any child between the ages
1. a. There is a 46% chance that any telephone call made on a of 8 and 10 learns the rules of the board game in 1.5
computer software technical support line will last 5 min- minutes or less is about 31.6%.
utes or more. d. 0.156; There is about a 15.6% chance that any child
b. The likelihood that any two cars on a certain two-lane between the ages of 8 and 10 takes between 3 and 4
road are less than 7 feet apart is approximately 25%. minutes to learn the rules of the new board game.
c. New Orleans will receive between 2 and 4 inches of rain 13. a. 137 customers
during March 15% of the time. b. One possible answer: Banks can better schedule servic-
f(x)dx f(x)dx 1.
1
3. a. Yes, f(x) 0 for all x and ing, balancing, and restocking of their ATMs.
0 c. i. P(150 x 200) 0.58
b. Yes, h(x) 0 for all x and h(x)dx h(x)dx 1.
1
0
c. No, the area between the graph of r and the horizontal
axis is 0.3 0.6
1.
d. No, some values of s(c) are negative.
y(x)dx 0.16
1
5. a. P(x 1) Customers
0 150 µ 200
b. 167 gallons ii. P(x 220) 0.50 0.46 0.96
P(x < µ) = 0.5
Customers
µ 220
Chapter 6 Answers to Odd Activities A-51
iii. P(x 235) P(x ) P( x 235) 1
[(b a)3 (b a)3]
0.50 0.49 24(b a)
0.01 2(b a)3 (b a)2
24(b a) 12
ba
Thus
12
c. The height of the density function is 0 to the left of a, so
F(x) 0 when x a. When a x b,
b 1 a dt b 1 a t bx aa. When x
Customers x x
µ 235 F(x)
a a
15. a. 6 800 200, so 100 s o the right of b, the area between the graph of u and the
b. The realigned mean score is more for each distribution, horizontal axis to the left of x is the area of a rectangle
1
because the recentering puts the mean of each at 500. with width b a and height . No more area
ba
c. 50% because 475 was the mean math score before is accumulated to the right of b
recentering. because the height of the density function is 0 at all such
d. No, the standard deviation is not given. points. Thus F(x) 1 when x b.
e. The statement “. . . and realigns the verbal and math
21. a.
scores so that a student with a score of 450 on each test
F
can conclude that his or her math and verbal scores are U.S. population
equal” indicates a change in the shape of the distribu- (percent)
tion. 100
f. The scale was recentered for interpretation purposes.
The recentering does not reflect any change in student
performance. Entrance requirements and other com- F(40)
parisons will now be made on the new scale.
17. a. P(60 x 80) 0.272, so 27.2%
F(20)
b. P(x 90) 0.50 0.232, so 26.8%
c. x
Age
P(x or x ) 1 P( ≤ x ≤ ) 0 10 30 50 70 90 110 125 (years)
1 P(43.65 x 100.95) If F equals the cumulative density function, then
1 0.683 0.317 F(40) F(20) is P(20 age 40).
Approximately 31.7% of the students are likely to make a b.
score more than one standard deviation away from the
U.S. population
mean.
per year of age
d. At 43.65
(percent)
b 1
P (20 ≤ age ≤ 40)
19. a. xdx
a b a
1.5
1
x2 b
b2 a2
b a 2 a 2(b a)
ba
2 1.0
b.
(x )3 b
1 b 1 0.5
(x ) dx
2 2
ba a ba 3 a
Age
1 0.0
[(b )3 (a )3] (years)
3(b a) 0 10 30 50 70 90 110 125
ba 3 ba 3
1
b a 0 when x 0
3(b a) 2 2
23. a. F(x) x2 when 0 x 1
ba 3 ab 3
1
3(b a) 2
2
1 when x 1
1 b. P(x 0.67) 0.4489
[(b a) (a b)3]
3
24(b a)
A-52 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
c. P(x 0.25) 0.9375 2. a. $35,143.80
d. f b. $18,277.65
3. a. Approximately 1 fox
2
b. f(t) 500(0.63)20t foxes born t years after 1990 that
will still be alive in 2010
1 c. Approximately 1083 foxes
4. a. Approximately $635.4 hundred
x b. 411 fountains. No, because D(10) 594, supply is
0 1
smaller than demand at this point.
c. Approximately $6236.0 hundred
F 5. a. P(x 3.8) 0.9025. The value of x will be smaller
2 than 3.8 about 90% of the time. This event is likely to
occur.
1 b. P(1.3 x 5) 0.8944
f
x
0 1 0.5 P(1.3 ≤ x ≤ 5)
25. a.
x 0 5 10 15 25 35 50 75 100
x
F (x) 0 8.2 17.8 27.6 46.4 62.2 78.8 92.1 96.4 0 1 2 µ3 4
c. 2.6667
b. F(100) is not 100 because the incomes of $100,000 or
more are not counted in the table.
c. Yes, there appears to be an inflection point near 0 when x 0
x 40, and the limiting value as x increases appears to d. F(x) 0.0625x2 when 0 x 4
1 when x 4
be 0.
d. The overall fit is very good.
F
100
e. f(x)dx 96.41% F(100)
0 1
100
xf (x)
f. dx $31,000
0 100
27. a. The distribution in graph a, the bell-shaped curve, is
characteristic of a normal breeding population. x
0 1 2 3 4
Malcolm argues that the one in graph b, with three
peaks, is what would be expected from three separate e. P(1.3 t 5) 0.8944
(controlled) populations.
b. The dinosaurs were supposed to be all female and CHAPTER 7
therefore not reproduce, so Ian first saw the normal Section 7.1
curve in graph a.
c. Neither graph is that of a probability density function. 1.
Input x
Even though all the outputs are non-negative, the area
under each curve is more than 1.
Chapter 6 Concept Review Rule s
1. a. R(t) (0.1)(3000 12 500t) 3600 50t
dollars per year after t years
Output
b. $29,064 s(x)
c. $17,664 sine of x
Chapter 7 Answers to Odd Activities A-53
15.
3. a. The maximum is 1 and occurs at x . f(x)
2
2
b. The absolute maximum is 1 and occurs at the values
x 2k, where k is an integer.
2
3 x
c. The minimum is 1 and occurs at x . -3 2 – 3
2
d. The absolute minimum is 1 and occurs at the values
3
x 2k, where k is an integer. -2
2
5. a. 1.5
17. f(x)
1 s(x) = sin x 2
0.5 c(x) = cos x
0 x
π
_ π 3π
_ 2π
-0.5 2 2 0 x
3 3
+
-1 2 2
-1.5
b. c(x) s(x) 0.707 when x 0.785, and -2
c(x) s(x) 0.707 when x 3.927.
c. x 0.785 2k, where k is an integer, and 19. The members of each pair of graphs are identical. Gen-
x 3.927 2k, where k is an integer. eralization: Begin with a sine graph that is not vertically
7. Amplitude: 1 shifted, and reflect it across the horizontal axis. Begin
Period: 2 with the same sine graph, and shift it right or left by an
Vertical shift: none integer multiple of half of the period of the original
Horizontal shift: graph. This new graph will be identical to the reflected
The function is reflected across the horizontal axis, so the graph.
cycle starting at x begins by decreasing.
21. a. Horizontal axis
9. Amplitude: 235
1 y = cos x
Period:
150
Vertical shift: 65 down
1
Horizontal shift: 0 x
3 π
_ π 3π
_
The function is reflected across the horizontal axis, so the 2π
2 2
1
cycle starting at x begins by decreasing.
3 -1
11. i. c ii. e iii. a
iv. f v. d vi. b
b. Vertical axis
13. f(x)
4 1 y = cos x
2 0 x
π
_ π 3π
_ 2π
2 2
x -1
0
A-54 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
c. y = -cos x y = cos (-x) b. Amplitude: approximately 1.5 therms per day. Vertical
1
shift: 1.8 therms per day
c. Period: approximately 12 months
Horizontal shift : approximately 1.5 months to the left
1.5
-2π -3π π 0
_ -π - _ _π π _
3π 2π
x d. G(x) 1.5 sin
6
6
1.8 therms per day of
2 2 2 2
gas used daily in month x
-1 3. a. The data are cyclic because the average daily tempera-
ture repeats itself over one year.
(i) Daily mean
temperature
y = -cos x y = cos x y = cos (-x) (°F)
1 90
x
-2π -3π π 0
_ -π - _ _π π _
3π 2π
2 2 2 2
-1
(ii)
10 Month
1 12
It appears from the graphs that the functions
(January) (December)
y cos(x) and y cos x are the same. The graph
b. Amplitude: approximately 27.9°F
of y cos x is the reflection, across the x-axis, of the
Vertical shift: approximately 49°F
graph of y cos x; that is, the outputs of y cos x
c. Period: approximately 12 months
are the negatives of the outputs of y cos x.
Horizontal shift: approximately 4 months to the right
23. In order to estimate amplitude and vertical shift, locate
the maximum and minimum output values. Subtract d. m(x) 27.9 sin x
6
2
3
49°F during the xth
the minimum value from the maximum value and di- month of the year
vide by 2 to calculate amplitude. Add the maximum and e.
minimum values and divide by 2 to calculate vertical
m(x)
shift.
(°F)
90
Section 7.2
1. a. Gas usage
(therms per day)
x
1 10
1 12 Month
0 Month f. The model gives an estimate of 76.9°F as the average
Nov Apr Sep Jan July temperature. (Note that this value agrees with the
data.) Although the function is an excellent descrip-
The data are cyclic because the residential gas usage is
tion of the normal mean daily temperature for the
determined by the daily temperature, which usually
years 1961–1990, it cannot predict with any certainty
repeats itself over a year’s time.
the mean daily temperatures in Omaha this July.
Chapter 7 Answers to Odd Activities A-55
5. a. The low temperatures should have the same period g(x)
and approximately the same amplitude and horizontal (°F)
shift as the average temperatures, but they should have 70
a smaller vertical shift.
b. Amplitude: approximately 27.5°F
Vertical shift: approximately 38.4°F
Period: approximately 12 months 35
Horizontal shift: approximately 4 months to the right
As expected, the period and horizontal shift are the x
10
same. The amplitude is slightly lower than for the 1 12 Month
mean temperature data, and the vertical shift is less. Both models appear to fit the data well.
2 13. a. M(x) 0.932 sin(0.467x 2.940) 8.737 billion
c. We calculate b using the equation 12 , so b .
b 6 trips x years after 1992
h b. T(x) 0.0052x 3 0.127x 2 0.573x 8.56 bil-
We calculate h using the equation 4 and the in-
b lion trips x years after 1992
formation that the horizontal shift is to the right, so h is
c. The cubic function fits the data fairly well, but the sine
2
negative: h 4b . Using a 27.5 and function better follows the curvature of the data and
3
provides a very good fit.
k 38.4, we have f (x) 27.5 sin x
6
2
3
d. M(14) 8.32 billion trips and T(14) 11.24 billion
38.4°F during the xth month of the year. trips. Reasons will vary.
7. a. g(x) 1.610 sin(0.534x 0.322) 1.591 therms 15. a. Maximum is 62°F; minimum is 12°F.
per day x months after November 2000 b. The average is 25°F, which equals k, the vertical shift.
Amplitude: 1.610 therms per day 17. a. The data appear concave up between 1949 and 1955,
Period: approximately 11.8 months concave down between 1956 and 1960, and concave
Horizontal shift: approximately 0.603 month to the left up between 1961 and 1963.
Vertical shift: up 1.591 therms per day b. A cubic model is not a good fit and is not appropriate
b. Approximately 1.6 therms per day because more than one inflection point are indicated
c. Approximately $26.88 by the data. Possibly a piecewise continuous model
9. g(x) 28.533 sin(0.479x 1.801) 47.985°F during that is quadratic between 1949 and 1955 and cubic be-
the xth month of the year; Essay answers will vary. tween 1956 and 1963 would fit the data. However, the
11. g(x) 27.357 sin(0.489x 1.878) 37.641°F during sine model seems most appropriate.
the xth month of the year. c. f (x) 41.547 sin(0.528x 2.913) 194.493 air-
f (x) craft x years since 1949
(°F) d. Approximately 155 aircraft
e. Answers will vary.
70
19. a. Hours °F Hours °F
5 37 65 60
35 11 44.5 71 47
17 52 77 34
x 23 47 83 45
10
1 12 Month
29 42 89 56
35 49 95 46.5
41 56 101 37
47 49 107 45.5
53 42 113 54
59 51
A-56 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
b. y 8.709 sin (0.261x 2.840) 47.090°F x hours c. Approximately 2.4 mm per day per month; Extrater-
after midnight on Wednesday: The period is approxi- restrial radiation in Amarillo is increasing by 2.4 mil-
mately 24 hours. limeters of equivalent water evaporation per day per
c. High: approximately 56°F month in March.
Low: approximately 38°F d. 18.4% per month
21. Essay answers will vary. 13. a. R(x) 1.32727 sin (0.0186x 1.1801) 1.46052 cos
23. Without technology: D(t) 7 sin 2
365
t
542
365
11.5 (0.0197x 3.7526) 7.90076 million dollars where
x is the day of the year
hours of daylight, where t is the number of days since De-
b. R(x) 0.0247 cos (0.0186x 1.1801)0.0288 sin
cember 31 of the previous year.
(0.0197x 3.7526) million dollars per day where x is
With technology: d(t) 6.677 sin (0.016t 1.908)
the day of the year
11.730 hours of daylight, where t is the number of days
c. R(46) 0.003 million dollars per day. Combined
since December 31 of the previous year.
daily sales were decreasing by approximately $3000
per day on February 15, 1992.
Section 7.3
d. Approximately 220 days and 293 days after the begin-
1. f (x) 3 cos 3x ning of 1992. (These values correspond to August 7
3. t(r) 2.34 sin (0.45r ) 80 and October 19. Note that 1992 was a leap year.)
5. h(x) 0.3328 cos (0.16x 12.3) e. x 90 days (March 30) or 166 days (June 14) or 339
9.666 sin [1.35(2x 6)] days (December 4).
h(x) 0.053248 sin (0.16x 12.3) 15. B(11) 11.1°F per month
26.0982 cos [1.35(2x 6)] B(4.5) 11.6°F per month
7. a. 2 hours 17. a., b.
b. 3050 km at 48 minutes after launching
Wind speed interval
c. 6–10 10–14 14–18
(m/sec)
d(x)
Change (kilowatts) 17.4 17.3 0.3
(kilometers)
Percentage change 38.4 61.7 2.7
3050
Average rate of change
4.4 4.3 0.07
2050 (kilowatts/m/sec)
North of
equator c. P(s) 5.2126 cos (0.258s 0.570) kilowatts per
1050
meter per second when the wind speed is s meters per
x
0 second
(minutes)
18 60 78 138 d., e.
-1050
South of Wind speed (m/sec) 6 10 14 18
equator
-2050
P(s) (kilowatts/m/sec) 2.7 5.2 2.6 2.5
-3050 Percentage rate of
6.0 18.6 24.7 24.1
change (%/m/sec)
d. Approximately 93.9 kilometers per minute The power output of the engine is decreasing at the
e. No; speed is the rate of change of the total distance rate of 2.7 kilowatts per meter per second at a wind
traveled with respect to the traveling time. speed of 6 meters per second. When the wind speed is
9. a. 1992: approximately –0.4 billion trips per year 10 meters per second and increases to 11 meters per
1996: approximately 0.2 billion trips per year second, the power output of the engine decreases by
2000: approximately 0.3 billion trips per year approximately 5.2 kilowatts. When the wind speed is
b. Approximately 9.7 billion trips 14 meters per second and increases to 15 meters per
11. a. Approximately 8 mm per day second, the power output of the engine decreases by
b. 47.1% decline; 1.6 mm per day per month approximately 2.6 kilowatts. At a wind speed of 18 me-
ters per second, the power output of the engine is
Chapter 7 Answers to Odd Activities A-57
increasing at the rate of 2.5 kilowatts per meter per between 1992 and 2002 was approximately 7.81 billion
second. (in 1995).
19. a. d(t) 60.7407 sin (0.1206t 0.3134) 99.9194 3. a. The maximum daily mean temperature is approxi-
deaths per 100,000 people per week, where t is the mately 76.9°F (in July). The minimum daily mean
number of weeks since January 1, 1923. Answers will temperature is approximately 21.1°F (in January).
vary. b. t(x) 28.533 sin (0.479x 1.801) 47.985°F in the
b. xth month of the year. The highest point on the model
Date t d(t)
is (7.041, 76.517), and the lowest point is (0.481,
Middle of 1924 78 7.0111* 19.452). The maximum normal daily mean tempera-
End of 1924 104 7.0252* ture in Omaha between 1961 and 1990 was 76.5°F and
occurred at the beginning of August. The normal daily
*deaths per one hundred thousand people per
week per week minimum temperature was at a minimum of 19.5°F
21. a. D(d) 23.677 sin (0.017d 1.312)0.292 degrees near the middle of January.
on the dth day of the year; the amplitude of the model c. The maximum and minimum temperatures obtained
is 23.7 degrees. This is the greatest angle of declination with the model are slightly lower than those found us-
that the Sun reaches. The period is approximately 374 ing the data. Because the data are based on a monthly
days (calculated with the unrounded b-value). We ex- average over a 30-year period, actual high and low
pect the Sun’s declination to complete a cycle every temperatures in any given year are likely to vary from
year. The period in the model is slightly long. those found in parts a and b.
b. The equinoxes occur at d 78.7 days and d 264.4 5. a. Ten maxima of the sine function i are near a relative
days. D(78.7) 0.4 degree per day, and maximum of the actual index w. Note that the maxima
D(264.4) 0.4 degree per day. At the equinoxes, appear as minima because of the inverted scale. (An-
the declination of the Sun is changing by approxi- swers may vary.)
mately 0.4 degree per day. b. Ten minima of the sine function i are near a relative
c. The summer and winter solstices occur when the minimum of the actual index w. (Answers may vary.)
declination is greatest in both north and south direc- c. A relative maximum of approximately 104.6 occurs at
tions. On the graph, the summer solstice corre- t 120.25 years after 1761, and a relative minimum
sponds to the maximum, and the winter solstice of 95.8 occurs at t 116.25 years after 1761. Thus the
corresponds to the minimum. At these points, the next maximum and minimum of i(t) after 1874 are
rate of change is zero. predicted to have occurred in 1878 and in 1882, re-
23. Inside: u(x) 2.4x spectively.
Outside: y(u) sin u d. Answers will vary.
y (ln 2.4)(2.4x)[cos (2.4x)] 7. a. The teal graph represents daylight hours.
25. Inside: u(x) sin x 7 The black graph represents nighttime hours.
Outside: y(u) 4u2 8u 13 b. Period: approximately 12 months
y 8 cos x(sin x 6) Amplitude: approximately 7.5 hours
27. Inside: u(x) sin x Vertical shift: approximately 12 hours up
Outside: y(u) ln u c. Values in table may vary.
cos x
y
sin x Month Hours Month Hours
Jan 5 July 19
Section 7.4
Feb 8 Aug 17
1. a. Between x 0 and x 13.45, the maximum value of
Mar 10.5 Sept 14.5
9.67 occurs at x 9.66. The minimum value of 7.81
Apr 13.5 Oct 11.5
occurs at x 2.93.
b. The greatest number of yearly mass transit trips be- May 16.5 Nov 8.5
tween 1992 and 2002 was approximately 9.66 billion June 19 Dec 6
(in 2000). The least number of yearly mass transit trips
A-58 CHAPTER 7 Repetitive Change: Cycles and Trigonometry
d. H(m) 7.021 sin (0.480m 1.619) 11.786 hours 15. a. Approximately 467.2 milliseconds
at the beginning of the mth month. b. Two possible answers are: The pulse speed was at its
Period: approximately 13 months highest value, 227 counts per second, after approxi-
Amplitude: approximately 7 hours mately 233.6 milliseconds and 700.7 milliseconds. The
Vertical shift: approximately 11.8 hours pulse speed was at its lowest value of 146 counts per
9. a., b. Answers will vary depending on the latitude of the second after approximately 4.2 milliseconds and 467.2
student’s hometown. milliseconds.
11. a. a(m) 27.1(0.485) cos (0.485x 1.707) watts per c. One possible answer is: The speed of the pulses emit-
centimeter squared per month when m 1 in Janu- ted from the star was increasing the fastest after ap-
ary, m 2 in February, and so on proximately 116.8 milliseconds. The speed at that
b. Radiation is at its highest level in July (m 6.7853) time was approximately 186.5 counts per second.
and at its lowest level in January (m 0.2808). The speed of the pulses emitted from the star was
c. The highest level of radiation is 60 watts per centime- decreasing the fastest after approximately 350.4 mil-
ter squared. The lowest level of radiation is 5.8 watts liseconds when the speed was also 186.5 counts per
per centimeter squared. second.
13. a. T(x) 0.565 cos (0.469x 2.293) percent per year x 17. a. d(t) 60.741 sin (0.121t 0.313) 99.919 deaths
years after 1988 per 100,000 people per week, where t is the number of
b. For one period of T, between x 0 and x 13.4, weeks since January 1, 1923
solve the equation T(x) 0 to find that x 1.54 (in- b. The period is approximately 52 weeks, so the death
put for relative minimum) and x 8.24 (input for rate should peak again at t 166.8 weeks since Janu-
relative maximum). Compare the outputs at these in- ary 1, 1923.
puts to T(0) and T(13.4) to find the absolute extrema. c. t 153.7; d(t) 100; The first time after 150 weeks
The smallest after-tax profit rate on investment, ap- since January 1, 1923, that the weekly pneumonia death
proximately 5.18%, occurred in 1990, and the largest rate was increasing the fastest was near 154 weeks. The
rate, approximately 7.5%, occurred in 1996. weekly death rate at that time was approximately 100
Solve T (x) 0 to find that the inputs at the inflec- deaths per 100,000 people per week.
tion points are x 4.89 and x 11.59. The most 19. a. a(m) 6.3745975 sin (0.485m 1.707) watts per
rapid increase in the after-tax profit rate was 0.56% cm2 per month, where m 1 in January, m 2 in
per year (in 1993), and the most rapid decrease in the February, and so on
rate was 0.55% per year (in 2000). b. The point of most rapid increase occurs in mid-April
c. Explanations will vary. (m 3.5), and the point of most rapid decrease oc-
curs at the beginning of October (m 10).
T (x)
(percent) c. 32.9 watts per cm2
21. a. R(x) 1.327 sin (0.019x 1.180) 1.461 cos
7.5 (0.020x 3.753) 7.901 million dollars where x is
the day of the year
7 b. R(x) 0.025 cos (0.019x 1.180)0.029 sin
(0.020x 3.753) million dollars per day
6.5
where x is the day of the year
6 c. $8.03 million, x 193 (July 12)
d. The absolute minimum value is $7.66 million, occur-
5.5 x ring at x 66 (March 7). A local minimum value of
Years $7.68 million occurs at x 318 (November 14).
5 after
0 2 4 6 8 10 12 14
1988
Chapter 7 Answers to Odd Activities A-59
Section 7.5 15. a. T(x) 1.205 cos (0.469x 0.722) 6.404 per-
cent x years after 1988
1.
b. The after-tax profit rate was lowest in 1990 (x 1.54)
x
at approximately 5.2% and was highest in 1996
∫A f(t) dt
(x 8.24) near 7.6%.
c. Approximately 6.4%
d. The values are close, but not identical, because a
model (and not the actual data) was used to determine
the average value in part c.
16
x 17. a. L(x)dx 259 thousand lizards
B A C D 1
b. 261 thousand lizards. This value is 2 thousand more
3. 1 than the answer to part a.
5. 2 2a c. Answers will vary.
7. 7.3 cos x 12x C
40
19. r(t)dt 4.2 males per 100 females,
1
9. cos(7.3x 12) C 0
7.3
100
r(t)dt 2.33 males per 100 females, and
11. F(x) 194.583 cos (0.024x 3.211) 14.63x C 50
13. a. Height units are dollars per year, and width units are 100
r(t)dt 8.8 males per 100 females. From 1900
years. Area units are dollars. 0
b. The t-intercepts of w between t 7 and t 50 are to 1940, the number of males per 100 females decreased by
t 21.74 and t 43.41. The area between t 7 and approximately 4.2 males per 100 females. From 1950 to
t 21.74 is $1.71, the area between t 21.74 and 2000, the number of males per 100 females decreased by
t 43.41 is $2.23, and the area between t 43.41 and approximately 2.3 males per 100 females. Between 1900
t 50 is $0.47. The total area is approximately $4.41. and 2000, the number of males per 100 females decreased
50
by approximately 8.8 males per 100 females.
c. w(t)dt $0.04, which is not the same as the re-
7 21. a. Height in counts per second, width in milliseconds,
sult of part b because w crosses the t-axis. The federal area in (counts per second)(milliseconds)
minimum wage, expressed in constant 2000 dollars, b. p(s) 0.0405 sin (0.0135s 1.571) 0.187 counts
decreased by approximately 4 cents between 1957 and per millisecond after s milliseconds
2000. 467.151324
c. p(s)ds 87 counts
d. w(t) 0
(dollars per year) Approximately 87 pulses are emitted by the star over
one period (approximately 0.5 second).
0.15
Chapter 7 Content Review
0.1
1. a. Amplitude: 712.5
0.05 t
Years Vertical shift: 787.5
0 after Period: 12 months
10 20 30 40 50
−0.05 1950 Horizontal shift: 3 months
b. l(x) 713.251 sin (0.525x 1.557) 777.883 lawn
−0.1 mowers ordered x months since December of the pre-
−0.15
vious year. The amplitude is very close, whereas the
vertical shift of the model is approximately 10 lawn
e. The federal minimum wage, expressed in constant mowers less than that found using the data. The pe-
2000 dollars, decreased by approximately $2.16 be- riod of the model is approximately 12 months. The
tween 1970 and 1995. horizontal shift of the model is approximately 3. Both
the period and the horizontal shift are very close to
those found using the data.
A-60 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
c. l(16) 1157 lawn mowers; For this estimate to be b. All particular solutions are horizontal lines, each pass-
valid, the cyclic pattern shown by the data should con- ing through the chosen initial condition.
tinue for the following year. c. y(x) C, where C is a constant
d. Lawn mower orders were highest in June (x 6). 7. a. The displayed particular solutions go through the
e. Solving l(x) 0, finding where l(x) is greatest, or points (0, 0.5), (2, 2), and (4, 5). Answers will vary.
c y
using in the sine model, we find that orders were
b 5
increasing most rapidly in March (x 3).
2. a. Approximately 1.6 units per year 4
b. Approximately 0 unit per year
3. a. Approximately 1.32 units per year 3
b. S(42) 0.37 unit per year. According to the model,
the Sauerbeck index of general wholesale prices was 2
increasing at the rate of 0.37 unit per year in 1860.
1 72 1
c. S(t)dt 90.8
72 42 42
4. a. The rate of change was greatest when t 111 days, x
0 1 2 3 4 5
which corresponds to March 20 (1992 was a leap year).
b. R(t) 7.305 cos (0.0197t 3.753) 20.383 million b. All particular solutions are parallel lines with negative
pints t days after January 1, 1992. slopes.
c. The answer to part a is the input corresponding to the c. y(x) x C, where C is a constant
inflection point for the model in part b. 9. a. The displayed particular solutions go through the
5. Approximately 1.24 million pints. At the end of February points (0, 1), (1, 3), and (2, 6.5). Answers will vary.
1992, the Campbell Soup Company was selling approxi- y
mately 1.24 million more pints than it was selling at the 8
beginning of February. 6
4
2
CHAPTER 8
x
Section 8.1 − 4 −3 −2 −1 1 2 3 4
−2
−4
1. c kg
dp −6
3. ka −8
da
5. a. The displayed particular solutions go through the b. All particular solutions are concave-up parabolas with
points (0, 0.5), (2, 2), and (4, 5). Answers will vary. minimum points on the vertical axis.
y c. y(x) 0.25x 2 C, where C is a constant
5 11. a. The displayed particular solutions go through the
points (0, 1), (2, 3), and (2, 6.5). Answers will vary.
4 y
8
3 6
4
2 2
x
1 − 4 −3 −2 −1 1 2 3 4
−2
0.5
x −4
0 1 2 3 4 5 −6
−8
Chapter 8 Answers to Odd Activities A-61
b. All particular solutions are concave-down parabolas d. Approximately 1.5 seconds after the object is dropped,
with maximum points on the vertical axis. it will hit the ground at a velocity of about 47.3 feet
c. y(x) 0.5x 2 C, where C is a constant per second. (The negative sign on the velocity indi-
13. The particular solutions with initial condition (0, 0) all pass cates downward motion.)
through the point (0, 0). Each particular solution has a con- df
21. a. kx
stant term of zero, but the nonconstant terms all differ. dx
dp k
15. a. 0.98 quadrillion BTU per year t years after 1975 b. f (x) x 2 C
dt 2
b. p(t) 0.98t C quadrillion BTU t years after 1975
c. p(t) 0.98t 59.9 quadrillion BTU t years after
c.
d k 2
dx 2
x C kx leads to the identity kx kx.
1975 dh k
23. a. feet per year after t years
d. In 1975, production was 59.9 quadrillion BTU and dt t
was increasing by 0.98 quadrillion BTU per year. b. h(t) 20.75 ln t 10.39 feet after t years
e. p(t) c. h(15) 45.8 feet. Over time, the tree will continue to
(quadrillion Btu) grow, but the rate of increase will be smaller and
65 smaller.
64 25. a. h(t)
63 (feet)
70
62
61 60
60 50
59 t 40
58 Years 30
since
0 1 2 3 4 5 20
1975
10
p(0) 59.9 quadrillion BTU; They are the same. t
dc (years)
17. a. 0.0342 million square kilometers per year t years 0 2 4 6 8 10 12 14 16 18 20
dt
after 1970 b. h(15) 46 feet
b. c(t) 0.0342t C million square kilometers t years 27. a. i. The particular solutions that are sketched go
after 1970 through (1, 1.5), (2, 1), and (1, 0). Answers will
c. c(t) 0.0342t 13.828 million square kilometers t vary.
years after 1970 y
3
d. Cropland was increasing by 0.0342 million square
kilometers per year in both 1970 and 1990. In 1970 2
there were 13.828 million square kilometers of crop-
1
land, and in 1990 there were 14.512 million square
kilometers. x
−3 −2 −1 1 2 3
19. a. v(t) 32t feet per second t seconds after the object −1
is dropped
−2
ds
b. 32t feet per second t seconds after the object is
dt −3
dropped
ii. When x 0, the graph of the particular solution
c. s(t) 16t 2 C feet t seconds after the object is rises as x gets larger. When x 0, the particular-
dropped solution graph rises as x gets smaller. The particu-
lar-solution graphs are concave down.
A-62 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
iii. The family of solutions appears to increase rapidly as x moves away from the origin (in both direc-
as x moves away from the origin (in both direc- tions), and then the decrease levels off. Again, the
tions), and then the increase slows down. The line line x 0 appears to be a vertical asymptote for
x 0 (lying along the y-axis) appears to be a ver- the family.
tical asymptote for the family. d. i. The particular solutions that are sketched go
b. i. The particular solutions that are sketched go through (0.1, 0), (0.1, 0), and (0.05, 0.05). Answers
through (10, 0), (10, 0), and (5, 5). Answers will will vary.
vary. y
y 0.3
30
0.2
20
0.1
10 x
x − 0.3 − 0.2 − 0.1 0.1 0.2 0.3
−30 −20 −10 10 20 30 − 0.1
−10
− 0.2
− 20
− 0.3
−30
ii. When x 0, the graph of the particular solution
ii. When x 0, the graph of the particular solution rises as x gets larger. When x 0, the particular-
rises as x gets larger. When x 0, the particular- solution graph rises as x gets smaller. The particu-
solution graph rises as x gets smaller. The particu- lar-solution graphs are concave down.
lar-solution graphs are concave down. iii. The family of solutions appears to behave the
iii. The family of solutions appears to behave the same same as that in part a, but the slope at each point
1
as that in part a, but the slope at each point on a on a particular-solution graph is 10 times the slope
particular-solution graph is 10 times the slope at at the corresponding point on a particular-solu-
the corresponding point on a particular-solution tion graph in part a. Again, the line x 0 appears
graph in part a. Again, the line x 0 appears to be to be a vertical asymptote for the family.
a vertical asymptote for the family. 29. y
c. i. The particular solutions that are sketched go
6
through (1, 1.5), (2, 1), and (1, 0). Answers will vary.
5
y
(a) 4
3
3
2 2
1
1
x
x −3 −2 −1 − 1 1 2 3
(c)
−3 −2 −1 1 2 3 (b) −2
−1
−2 31. y
40
−3
30
ii. When x 0, the graph of the particular solution
20
falls as x gets larger. When x 0, the particular-
solution graph falls as x gets smaller. The particu- 10
lar-solution graphs are concave up. x
− 4 −3 −2 −1 1 2 3 4
iii. The slope at each point on a particular-solution −10
graph is the negative of the slope at a correspon-
−20
ding point on a particular-solution graph in part a.
−30
The family of solutions appears to decrease rapidly (b) (a)
Chapter 8 Answers to Odd Activities A-63
Section 8.2 15. The particular solutions that are sketched go through (0,
0), (1, 4), and (1, 3). Answers will vary.
dT k
1. , so T 2kt C h
dt T 6
(We use the positive square root because thickness can’t
4
be negative.)
dA 2
3. kA, so A(t) ae kt
dt
t
dx N −3 −2 −1 1 2 3
5. kx (N x), so x(t) −2
dt 1 AeNkt
dD k −4
7. hD, where k and h are both constants
dt D
−6
9. The particular solutions that are sketched go through
(2, 2), (2, 1), and (1, 1). Answers will vary. 17. The particular solutions that are sketched go through (0,
3), (1, 0), and (2, 1). Answers will vary.
y g
6 3
4 2
2 1
x x
−3 −2 −1 1 2 3 −3 −2 −1 1 2 3
−2 −1
−4 −2
−6 −3
11. The particular solutions that are sketched go through (0, 19. Solve by separation of variables; y(x) ae kx
2), (2, 4), and (1, 6). Answers will vary. 21. Solve by antidifferentiation; y(x) k ln x C
g 23. Solve by separation of variables; y(x) ax k
8 dq
25. a. kq milligrams per hour
6 dt
4 b. q(t) 200e0.346574t milligrams after t hours
2
c. Approximately 50 milligrams after 4 hours and 12.5
milligrams after 8 hours.
x
−3 −2 −1 1 2 3 da
−2 27. a. ka units per day
dt
−4 b. a(t) ce0.181262t units after t days
−6 c. Approximately 0.9 gram after 12 hours, 0.5 gram after 4
13. The particular solutions that are sketched go through days, 0.2 gram after 9 days, and 0.004 gram after 30 days
(2, 1), (2, 1), and (1, 1). Answers will vary. dN
29. a. 0.0049N(37 N) countries per year
g dt
3 37
b. N(t) countries t years after 1800
2 1 Ae0.1813t
37
1 c. N(t) countries t years after
1 28,097.439e0.1813t
t 1800
−3 −2 −1 1 2 3 d. 2 countries in 1840 and 24 countries in 1860
−1
e. The upper asymptote is N(t) 37, and the lower as-
−2 ymptote is N(t) 0.
−3
A-64 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
f. N (t) b. y(8) 12.5
Countries
x Estimate of y(x) Slope at x
37
2 2 2.5
30
5 9.5 1
8 12.5
20
dw 33.67885
5. a. pounds per month after t months
dt t
10
t b. w(3) 46.0 pounds; w(6) 70.0 pounds
Years
0 after
40 50 60 70 80 Estimate of w (t ) Slope at t
1800 t (months) (pounds) (pounds per month)
df 1 6 33.679
31. a. kf (L f )
dx 1.25 14.420 26.943
L
b. f (x) 1.50 21.155 22.453
1 AeLkx
1.75 26.769 19.245
c.
d
L
dx 1 AeLkx
k L
1 AeLkx
L
L
1 AeLkx
2 31.580 16.839
Each side of this equation can be simplified to 2.25 35.790 14.968
L2AkeLkx
2.5 39.532 13.472
(1 AeLkx)2
2.75 42.900 12.247
Section 8.3 3 45.961 11.226
1. a. y(4) 2 3.25 48.768 10.363
3.5 51.359 9.623
x Estimate of y(x) Slope at x 3.75 53.764 8.981
0 0 0.5 4 56.010 8.420
2 1 0.5 4.25 58.115 7.924
4 2 4.5 60.100 7.484
4.75 61.967 7.090
b. y(7) 34
5 63.739 6.736
5.5 67.027 6.123
x Estimate of y(x) Slope at x
5.75 68.558 5.857
1 4 2
6 70.022
4 10 8
c. w(3) 56.5 pounds; w(6) 82.9 pounds
7 34
Estimate of w (t ) Slope at t
3. a. y(5) 11.91 t (months) (pounds) (pounds per month)
1 6 33.679
x Estimate of y(x) Slope at x
2 39.679 16.839
1 1 5
3 56.518 11.226
3 11 0.45
4 67.745 8.420
5 11.91
5 76.164 6.736
6 82.90
Chapter 8 Answers to Odd Activities A-65
d. The answer to part b should give a more accurate ap- dT
9. a. k(T A)°F per minute after t minutes
proximation of the true weight at each input because dt
it uses a smaller step size. b. k 0.064
dp c.
7. a. 3.935t3.55e1.35135t thousand barrels per year t T Estimate of T (t ) T
(t )
dt
years after production begins, where p(t) is the total 0 98 1.8
amount of oil produced after t years 1 96.2 1.684
b. See table for this activity. During the first 5 years, the 2 94.516 1.576
oil well will produce approximately 10.3 thousand
3 92.940 1.475
barrels.
4 91.465 1.380
t Estimate of p (t ) p
(t ) 5 90.085 1.291
0 0 0 6 88.794 1.208
0.5 0 0.17094 7 87.586 1.131
1 0.0855 1.0187 8 86.455 1.058
1.5 0.5948 2.1865 9 85.397 0.990
2 1.6881 3.0891 10 84.408 0.926
2.5 3.2326 3.4707 11 83.481 0.867
3 4.9680 3.3734 12 82.615 0.811
3.5 6.6547 2.9668 13 81.804 0.759
4 8.1381 2.4251 14 81.044 0.710
4.5 9.3507 1.8744 15 80.355
5 10.2879 The temperature of the object is approximately 80.3°F
after 15 minutes.
c. p (t) 11. Essay answers will vary but should include a discussion of
dp (thousand barrels) the accuracy of tangent line approximations for function
dt values.
(thousand barrels 10
per year) 8 Section 8.4
4.0 6 d 2S k
3.0 1. 2
4 dt 2 S
2.0 d 2P k
1.0 2 3. k, so P(y) y 2 Cy D
dy 2 2
t t
(years) (years) d 2R
1 2 3 4 5 1 2 3 4 5 5. a. 6.14
dt 2
The graph of the differential equation is the slope graph b. R(t) 3.07t 2 7.01t 15.94 jobs in the tth month
for the graph of the Euler estimates. Similarly, the graph of the year
of the Euler estimates is an approximation to the accu- c. 156 jobs in August and 310 jobs in November
mulation graph of the differential equation graph. d 2A
7. a. 2099 cases per year per year
dt 2
b. A(t) 1049.5t2 5988.7t 33,590 cases t years
after 1988
c. 308.3 cases per year and 42,111 cases
d 2f
9. a. k
dx 2
k
b. f (x) x 2 Cx D
2
A-66 CHAPTER 8 Dynamics of Change: Differential Equations and Proportionality
d2 k 2
dP
c. x Cx D k leads to the identity k k. 2. a. 0.001175P(16.396 P) million people per
dx 2 2 dx
dE2 year, where x is the number of years since 1800.
11. a. 0.212531E
dt 2 16.396
b. P(x) million people, where x is the
b. E(t) 4.713 sin (0.461011t 2.248008) mm per 1 Ae0.019265x
day, where t is the month of the year. number of years since 1800.
c. E(t) 4.713 sin (0.461011t 2.248008) 12.5 mm 16.396
c. P(x) million people, where x is
per day, where t is the month of the year. 1 2.108e0.019265x
d. The model overestimates by 2.2 mm per day in March the number of years since 1800.
and 12.0 mm per day in September. d. 8.3 million people in 1840 and 9.1 million people in
1850
Chapter 8 Concept Review 3. a. 1840 8.18 million people; 1850 8.97 million
people
1. a. The relative risk of having a car accident is changing
b. 1840 8.24 million people; 1850 9.03 million
with respect to the blood alcohol level at a rate that is
people
proportional to the risk of having a car accident at a
c. These estimates are slightly smaller than those found
certain blood alcohol level.
in part d of Activity 2.
b. R(b) ae kb percent, where b is the proportion of al-
dQ
cohol in the blood stream. 4. a. 0.008307Q(7.154 Q) million people per
dx
c. R(b) e 21.398b percent, where b is the proportion of year, where the population is P(x) Q(x) 4.4 mil-
alcohol in the blood stream. lion people and x represents the number of years since
d. A crash is certain to occur (that is, it has a 100% prob- 1800
ability of occurring) when the blood alcohol level is
21.5%.
e. R (b)
(percent)
20
18
16
14
12
10
8
6
4
2
b
0 0.05 0.10 0.15
Chapter 9 Answers to Odd Activities A-67
7.154 vor of the bill received by the senator when the to-
b. Q(x) million people, where x is
1 0.185e0.059428x bacco industry spends $53 million lobbying against
the number of years since 1800. the bill.
c. m
16.396
million people when x 40 l Million
1 2.108e0.019265x Letters dollars
c. P(x) Inputs
7.154
4.4 million people when x 50
1 0.185e 0.059428x
where x is the number of years since 1800.
d. 6.0 million people in 1850. This answer is significantly Rule P
smaller than the one found in part d of Activity 2.
Output
P(l, m)
CHAPTER 9 Probability
Section 9.1 5.
1. a. P(1.2, s) is the profit in dollars from the sale of a yard i
of fabric as a function of s, the selling price per yard, Thousand p r c
Inputs dollars Dollars Dollars People
when production cost is $1.20 per yard.
b. P(c, 4.5) is the profit in dollars from the sale of a
yard of fabric as a function of c, the production cost
per yard, when the selling price is $4.50 per yard. Rule D
c. When the production cost is $1.20 per yard and the
selling price is $4.50 per yard, the profit is $3.00 for
each yard sold. Output
d. D(i, p, r, c)
c s
Million units
Inputs Dollars Dollars
7.
n
Rule P r Compoundings
m Interest each year t
Inputs Dollars rate Years
Output
P(c, s)
Dollars
Rule A
3. a. P(100,000, m) is the probability of the senator voting
in favor of the bill as a function of the amount m mil-
lion dollars invested by the tobacco industry lobbying Output
against the bill when the senator receives 100,000 let- A(m, r, n, t)
ters supporting the bill. Dollars
b. P(l, 53) is the probability of the senator voting in favor
of the bill as a function of the number l of letters in fa-
A-68 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
9. a., b.
Relative humidity (%)
40 45 50 55 60 65 70 75 80 85 90 95 100
110 135
108 130 137
106 124 130 137
104 119 124 130 137
102 114 119 124 130 137
100 109 113 118 123 129 136
Air temperature (°F)
98 105 108 113 117 122 128 134
96 101 104 107 111 116 121 126 132
94 97 100 103 106 110 114 119 124 129 135
92 94 96 98 101 104 108 112 116 121 126 131
90 91 92 94 97 99 102 106 109 113 117 122 126 131
130
88 88 89 91 93 95 97 100 103 106 109 113 117 121
86 85 86 88 89 91 93 95 97 99 102 105 108 111
84 83 84 85 86 87 89 90 92 94 96 98 100 102 105
82 81 82 83 83 84 85 86 87 88 90 91 93 94
90
80 80 80 81 81 82 82 83 83 84 85 85 86 87
11. a. 11.97 hours
b. 11.97 hours
c. Answers will vary. For example, if your college is at the
45th parallel north, it will receive 9.19 hours of day-
light each day during January.
d.
Latitude Month
North Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
South Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun
0 12 12.12 12.12 12.12 12.12 12.12 12.12 12.12 12.10 12.11 12.12 12.12 12.12 12
5 11.87 11.96 12.08 12.22 12.35 12.41 12.38 12.28 12.16 12.02 11.90 11.83
10 11.61 11.81 12.06 12.35 12.57 12.70 12.64 12.45 12.17 11.91 11.67 11.55
15 11.34 11.66 12.04 12.47 12.82 13.00 12.92 12.62 12.22 11.81 11.44 11.25 11
20 11 11.07 11.50 12.01 12.60 13.07 13.32 13.22 12.81 12.26 11.70 11.20 10.94
25 10.78 11.33 11.97 12.74 13.34 13.66 13.53 13.02 12.31 11.58 10.94 10.62
30 10.45 11.14 11.97 12.88 13.65 14.05 13.88 13.23 12.35 11.47 10.67 10.26
10
35 10 10.09 10.95 11.95 13.06 13.98 14.47 14.27 13.47 12.42 11.33 10.36 9.86
40 9.68 10.71 11.91 13.25 14.36 14.96 14.71 13.76 12.48 11.18 10.00 9.39
9
45 9 9.19 10.45 11.87 13.48 14.82 15.55 15.25 14.09 12.55 11.01 9.60 8.85
50 8.61 10.13 11.84 13.78 15.38 16.29 15.91 14.48 12.66 10.80 9.07 8.17
55 7.83 9.73 11.79 14.10 16.14 17.28 16.78 14.99 12.76 10.55 8.45 7.28
60 6.79 9.21 11.74 14.62 17.10 18.70 18.01 15.67 12.92 10.22 7.60 6.04
18 17 16 15 14 13
Chapter 9 Answers to Odd Activities A-69
13. k 17. Height
2 (feet)
45
1.5 35
200
400
25
1 600
15
0.5 5 Diameter
12 14 16 18 20 22 24 26 28 30 (feet)
0
s 19. a. P(c, s) 108.958 0.730s 0.027s2 0.175c for a
20 40 60 supermarket with s thousand square feet of sales space
15. s and a customer base with a per capita income of $c
thousand
3 b. c
(thousand
100
dollars)
60 35
40
0 u
5
-2 s
13 (thousand
5 10 15 20 25 square feet)
21. a. Answers will vary. For example, the BMI for a person
who is 5 feet 8 inches tall and weighs 150 pounds is
22.8 points.
b.
Weight Height (inches)
(pounds) 60 62 64 66 68 70 72
90 17.6 16.5 15.4 14.5 13.7 12.9 12.2
100 19.5 18.3 17.2 16.1 15.2 14.3 13.6
110 21.5 20.1 18.9 17.8 16.7 15.8 14.9
15
120 23.4 21.9 20.6 19.4 18.2 17.2 16.3
130 25.4 23.8 22.3 21.0 19.8 18.7 17.6
140 27.3 25.6 24.0 22.6 21.3 20.1 19.0
150 29.3 27.4 25.7 24.2 22.8 21.5 20.3
20
160 31.2 29.3 27.5 25.8 24.3 23.0 21.7
170 33.2 31.1 29.2 27.4 25.8 24.4 23.1
180 35.2 32.9 30.9 29.1 27.4 25.8 24.4
190 37.1 34.8 32.6 30.7 28.9 27.3 25.8 25
200 39.1 36.6 34.3 32.3 30.4 28.7 27.1
35 30
A-70 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
K(0.00064516)h2 revenue is greater than 184 million guilders but less
c. w pounds, where h is the height
0.4536 than 214 million guilders (or else we would see the 214
in inches and K is the BMI. contour curve). From the three-dimensional graph, it
d. appears that the maximum revenue is near 200 million
w guilders. One possible approximation is 190 million
(pounds) guilders.
200 35 30 27. a. Input: P 10 hours, H 70%; Output: approxi-
mately 12.5 days
25 When C. grandis is exposed to 70% relative humidity
and 10 hours of light each day, it will take approxi-
mately 12.5 days to develop.
b. Input: P 10 hours, H 60%; Output: approxi-
20 mately 11.5 days
When C. grandis is exposed to 60% relative humidity
and 10 hours of light each day, it will take approxi-
mately 11.5 days to develop.
15
29. a. When y increases, the contour curves have smaller
numbers, so f (x, y) decreases. Therefore, the function
90
h
must increase when y decreases.
60 72 (inches)
b. Because the contour curves are more closely spaced to
This graph is the upside-down mirror reflection of the the left of (2.5, 2.5) than they are directly below that
graph in part b. It is also more accurately drawn. point, the function decreases more quickly as x de-
K 10.65 1.13w K 10.65 1.04s creases than it does as y decreases.
23. a. s or w
1.04 5.83w 1.13 5.83s c. The change is greater when (2, 2) shifts to (1, 2.5),
b. s causing the contour values to change from about 21 to
1 about 14, than it is when (1, 0) shifts to (4, 1), causing
virtually no change in contour values.
31. a. The three-dimensional graph has two peaks approxi-
mately the same height—one around (0.7, 0, 0.25)
and the other near (0.7, 0, 0.25).
b. The descent is greater from (0.7, 0.1) to (0, 0.3).
0
11.
1 c. The output increases as x increases.
10 0.6
10 .7
.9
.8 d. There are many correct answers. One possible answer
10
is (0.4, 0.2).
10 .9
.8
10 10 .0
.7 11
0 w
0 1
25. a. The center of the 184 thousand contour curve corre-
sponds to approximately 10 thousand tons of lower-
fat cheese and 55 thousand tons of regular cheese.
b. From the contour graph, we see that the maximum
Chapter 9 Answers to Odd Activities A-71
33. a. to spot this number if you first quickly sketch a couple of
contour curves. Once you have located a number that ap-
pears to be a maximum value, you describe its location by
following its column and row to the edges of the table to
determine the input values that produce that maximum.
Keep in mind that for a contour graph as well as a table,
there may exist more than one relative maximum.
w
Sno Section 9.2
Ice Pellets
n
g Rai 1. a. Air temperature
ezin
Fre b. Row
c. A(p, 95) 0.0238p2 2.3455p 151.31°F when
in
Ra the dew point is p°F.
3. a. 60% of the time
b. Constant: hour of the day, Input: fraction of sky
covered
c. column
d. C(9, f ) 1.651f 3 2.686f 2 1.597f 1.019,
where f is the fraction (expressed as a decimal) of the
b. The temperature dropped 36°F. sky covered by clouds.
35. To locate a relative maximum on a contour graph, look 5. a. No
for contour curves that form a simple, closed curve (or, if b. Yes, row
completed, appear to form a closed curve). Estimate the c. p(m, 9) 0.017m2 2.091m 85.977 dollars when
point in the center of this closed curve. Moving away m months is the length of the loan (m between 24 and
from the “center,” check that contour levels are decreas- 60); p(52, 9) 22.65 dollars
ing. If the levels are decreasing by a constant amount d, 7. a. P(x) 0.857x 7.781 pounds/person/year, where
the maximum value at the point can be estimated be- $(1.50 x) per pound is the price of peaches
tween the largest indicated contour value k and k d. b. P(0.05) 7.7 pounds/person/year
On a table of data, locate a number whose eight surround- 9. a. Cost of Coke products
ing numbers are all less than the number itself. It may help b. Column
c. Approximately 201 cans
11. a. 3.88 million people
b.
Year
Age 1990 1995 2000 2005 2010 2015 2020
15 3.34 3.65 3.87 4.24 4.31 4.22 4.26
20 4.04 3.51 3.88 4.10 4.48 4.55 4.45
25 4.06 3.79 3.39 3.73 3.94 4.30 4.36
30 4.50 4.38 3.92 3.52 3.86 4.08 4.44
35 4.27 4.59 4.47 4.00 3.61 3.95 4.17
40 3.80 4.28 4.65 4.54 4.07 3.68 4.02
45 2.90 3.70 4.21 4.57 4.46 4.01 3.62
50 2.43 2.93 3.69 4.19 4.55 4.44 4.00
A-72 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
c. Approximately 100 thousand people per year P
d. Approximately 16 thousand people per year 7. a. m l100,000 is the rate of change of the probability that
13. a. 0.35 kg/day the senator will vote for the bill with respect to the
b. W(68, t) 0.00009t3 0.003t2 0.013t 0.572 amount spent by the tobacco industry on lobbying
kg/day, where t°C is the temperature. A pig weighing when the senator receives 100,000 letters in opposition
68 g will (on average) gain/lose W(68, t) kilograms to the bill. We expect this rate of change to be negative,
during a day when the temperature is t°C. because if the number of letters is constant but lobby-
c. The graph of W(68, t) is concave up, increasing until ing funding against the bill increases, the probability
approximately t 10. After t 10 the graph is con- that the senator votes for the bill is likely to decline.
cave down. It reaches a maximum just after t 21.1.
A pig weighing 68 kilograms will gain the most weight
P
b. l
m53 is the rate of change of the probability that the
when the temperature is near 21.1°C. For tempera- senator votes for the bill with respect to the number of
tures in excess of 21.1°C, the weight gain diminishes letters received when $53 million is spent on lobbying
more and more rapidly until weight is actually lost at efforts. We expect this rate of change to be positive, be-
temperatures in excess of approximately 37.5°C. cause if the number of letters increases (while lobbying
15. a. Approximately 0.52°F per °F funding remains constant), the probability of the sen-
b. First, determine a cross-sectional model: A(p, 85) ator voting in favor of the bill is likely to increase.
0.0194p2 2.0052p 135.832°F when the dew point f
9. a. 6x 5y
is p°F. Second, determine the derivative of this model: x
d(p, 85) f
0.0388p 2.0052 Third, evaluate the de- b. 5x 6y
dp y
rivative for p 65°F .
17. a. A(14,000, r) 1.4r2 280r 14,000 dollars, where
c.
f
x y7 6x 35
r is the interest rate expressed as a percentage. f
dA(14,000, r) 11. a. 15x 2 6xy 3 9y 14
b. When r 0.127, dr $315.56 per percent- x
age point. f
b. 9x 2y 2 9x
c. A(14,000, r) 14,000r2 28,000r 14,000 dollars , y
f
where r is the interest rate expressed as a decimal.
dA(14,000, r) c. 15x 2 48x 32
d. When r 12.7%, $31,556 per 100 x y2
dr
percentage points. The derivative tells us approxi- 13. a. Mt s t t 1 s
mately how much the output will change when the in-
put increases by one unit. If the input is a percentage b. Ms ln t 3.75
expressed as a decimal, then an increase of one unit c. Ms t3 ln 3 3.75
corresponds to 100 percentage points. For example, if h 1
15. a. s t 2t(st tr)
r 0.127 and is increased by 1 to r 1.127, the cor-
h s 1
responding percentages are 12.7% and 112.7%. b. t t2 r 2(st tr)(s r)
19. Derivatives of cross-sectional models of a three-dimen- h t
c. r r 2 2t(st tr)
sional function can be used to pinpoint optimal points on
h
those cross sections. These optimal points in turn may be d. r (s, t, r)(1, 2, 1) 14
used to estimate the point at which a critical point may
17. a. fx 2y 16xy3
appear on the three-dimensional function.
b. fy 2x 24x2y2 10e2y
Section 9.3 fxx 16y3
W c. fxy 2 48xy2
1. h pounds per inch fyx 2 48xy2
T fyy 48x2y 20e2y
3. t g23°F per degree of latitude
2 48xy2
16y3
R d.
5. c b2 dollars per cow for c 100 2 48xy 2 48x2 20e2y
Chapter 9 Answers to Odd Activities A-73
⎥⎧ 0.389m
19. x y 3 3.321m2 when
⎥ 3.425m 61.857 0m7
x 2y 1 1
2 watts per square
x3 y2 x ⎥ meter per month
y 1 1
2
2x R(m, 50 ) ⎨ 0.565m3 16.008m2 when
y2 x y3 ⎥⎥ 137.689m 381.183 7 m 12
21. x y ⎥ watts per square
⎩ meter per month
x 4e 2x3y
y 6e 2x3y
6e 2x3y
9e 2x3y b. Quadratic model: For l 50°, Rl mMarch 0.70
23. a. A(14,000, r) 14,000(1 r)2
dollars. This function watt per square meter per month per degree of
gives the value of an investment after 2 years when the latitude.
earned interest is 100r%. Sine model: For l 50°, Rl mMarch 0.75 watt per
A
b. r (P, r)(14,000, 0.127) $31,556 per 100 percentage square meter per month per degree of latitude.
points c. Sine model: For m 3, Rm l50° 13.13 watts
c. The slope of the line tangent to A(14,000, r) at per square meter per month per month.
r 0.127 is $31,566 per 100 percentage points. Piecewise model: For m 3, Rm l50° 12.85
watts per square meter per month per month.
d. R (March, l)
A(14,000, r) (watts/m2/month)
(dollars) 60
20,000
Slope = Slope ≈
31,556 0.7
l°
–90 –50 90
R (m, –50)
14,000 r (watts/m2/month)
0 0.127 0.18 60
Slope ≈ –13
25. a. R(March, l) 0.007l 2 0.035l 49.749 watts
per square meter per month, where l is the number
of degrees north of the equator (negative values of l
correspond to locations south of the equator). An
alternative model is
R(March, l) 35.943 sin(0.023l 1.633) 15.513
m
watts per square meter per month 0
1 3 12 (months)
R(m, 50°) 27.105 sin(0.485m 1.707) 32.911
watts per square meter per month, where m 1 cor- 27. a. Approximately 0.67 pound/person/year per ten thou-
responds to January, m 2 corresponds to February, sand dollars of income
and so on. An alternative model is b. Approximately 0.77 pound/person/year per dollar
per pound
A-74 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
A A
c.
c
i
3
0.67 pound/person/year per ten thousand 33. a. t 1000re rt dollars per year, r 1000te rt dollars
per 100 percentage points
b. t r
C
p 0.2
0.82 pound/person/year per dollar per t
1000r 2e rt 1000ert 1000rte rt
pound r 1000tre rt 1000e rt 1000t 2e rt
d. The estimates for rate of change with respect to in- For t 30 and r 0.047, the second partials matrix is
come are close. The estimates for rate of change with
9.05 9871.25
respect to price differ by approximately 0.044. One .
9871.25 3,686,359.86
possible answer: The method in part c is easier, but the At the time a $1000 investment has been earning
estimate from part b is more reliable. 4.7% compounded continuously for 30 years.
29. a. Ht 10.45 10v v kilogram
calories per
(1) The rate at which the amount is growing with re-
square meter per hour per degree Celsius
spect to time is increasing with respect to time by
Hv (33 t)
v
5
1 kilogram-calories per square $9.05 per year per year.
meter per hour per meter per second (2) The rate at which the amount is growing with
H respect to time is increasing with respect to the in-
b. v should be positive because an increase in wind
speed (when temperature is constant) should increase terest rate by $9871.25 per year per 100 percent-
heat loss. age points.
H
c. When v 20 and t 12, v 2.48 kilogram
(3) The rate at which the amount is growing with re-
calories per square meter per hour per meter per spect to the interest rate is increasing with respect
second. to time by $9871.25 per 100 percentage points per
H year.
d. t should be negative because an increase in temper-
(4) The rate at which the amount is growing with re-
ature (when wind speed is constant) should decrease
spect to the interest rate is increasing with respect
heat loss.
H to the rate by $3,686,359.86 per 100 percentage
e. When v 20 and t 12, t 35.17 kilogram
points per 100 percentage points.
calories per square meter per hour per degree Celsius.
A
31. a. We expect food intake to increase as either milk pro- 35. a. t (1 r)t ln(1 r) million dollars per year
duction or size increases. Therefore, we expect both A
b. r t(1 r)t1 million dollars per 100 percentage
partial derivatives to be positive. points
I
b. s 1.24 0.18s 0.21m kilograms per day per A
c. When t 5 and r 0.15, t 0.28 million dollars
unit of size index. This equation is the rate of change per year.
of the amount of organic matter eaten with respect to d.
the size of the cow (when the amount of milk pro-
A (0.15, t)
duced is constant).
I
(million dollars)
m 0.21 0.072m 0.21s kilograms per day
per kilogram of milk per day. This equation is the rate
of change of the amount of organic matter eaten with
Slope ≈ 0.28
respect to the amount of milk produced (when the size
of the cow is constant).
I
c. m 0.64 kilogram per day per kilogram of milk per
day 1
I
d. s 0.38 kilogram per day per unit of size index
t
e. s m
5 (years)
s 0.18 0.21
m 0.21 0.072
37. No, since the y is being treated as a constant during dif-
Because all second partials are positive, we know that
ferentiation, it does not matter if the substitution of the
the rates of change in the s and m directions increase
specific constant is made before or after the derivative is
as both s and m increase. This indicates that the surface
determined.
is concave up in the s and m directions.
Chapter 9 Answers to Odd Activities A-75
Section 9.4 9. 21; 0.2
The value of n should decrease by approximately 0.2 in
dx f 45x 2y 2
3x
1. y order to compensate for an increase of 0.2 in m.
dy fx 30xy 3 2y
11. Approximately 3.703; 377.57
dm gn 49m
3. 13. a. A(6, 250) $7.16
dn gm 1
59.3 m
49n A
5. b. (0.02c 2 0.35c 0.99)(ln 0.99897)(0.99897n)
x y n
100
10 $0.002 per shirt
dn Ac
c.
dc An
77.94 5
[(0.04c 0.35)(0.99897n) 0.46]
(0.02c 2 0.35c 0.99)(ln 0.99897)(0.99897n)
60.75 y x
0 shirts per color
0 5 10
60.75 77.94 100
dn
We expect dc to be positive because if the number of
dx dy
3.82 0.26 colors increases, the order size would also need to in-
dy dx
crease to keep average cost constant.
7. b dn
1.2 d. dc 450 shirts per color. For each additional color,
the order size would need to increase by approxi-
mately 450 shirts. Similarly, if the number of colors
decreases by 1, the order size could decrease by ap-
proximately 450 shirts and the average cost would re-
main constant.
15. a., b.
t
(°C)
88
0.87 a
0 3.9 10
87
db
0.016
da 86
a r
10 85
20 45
(mL)
dt pr (1.9836 0.05916r)
c. °C per milliliter
dr pt 9.6544 0.14736t
dt
d. At t 86.5°C and r 23.125 ml, dr 0.199°C per
0.9 milliliter.
dt
0 b At t 86.5°C and r 43.9 ml, dr 0.199°C per
0.87 1.2
milliliter.
da (Answers for the alternative view with t on the hori-
64.85 zontal axis are given in the Student Solutions Guide.)
db
A-76 CHAPTER 9 Ingredients of Multivariable Change: Models, Graphs, Rates
17. a. w
0.00064516Kh2
0.4536 pounds, where K 20.204 c. 0.05s2 0.35s 3 cm when 0 s 2
0.05s2 0.15s 3.4 cm when 2 s 4
dw 2(0.00064516)Kh where s is the storage time in months.
b. dh 0.4536 pounds per inch. When h 67,
Assuming that 1 month 4.3 weeks, 2 weeks corre-
dw
dh 3.85 pounds per inch.
2
sponds to s 4.3 0.465 month. Thus we have
c. The answer to part b agrees with the answer given in
C(35, 0.465) 3.2 cm.
Example 1.
19. a. Approximately $193.32
b. Approximately $8232
d.
C
t s4
is the rate of change of the consistometer value
with respect to the blanching temperature when the
21. a. From the table, we estimate that Coke would need to
storage time is 4 months.
lower its prices by more than $0.50 per can.
e. Using C(t, 4) (6.944 104)t 2 0.087t 5.412 cm,
b. Using the equations S(1.00, P) 196P 25 cans
when the apples are blanched at t°C, we have
of Coke products and S(C, 1.25) 50.286C 2
C
7.771C 312.6 cans of Coke products, where $P is the (1.389 103)t 0.087 cm per °C. When
t s4
price of Pepsi products and $C is the price of Coke C
products, we estimate the change in Coke prices as t 45°C, 0.0244 cm per°C. When the
t s4
Sp storage time is a constant 4 months and the blanching
C
Sc
196
P 92.8 (0.25) $0.53. Coke temperature is 45°C, the consistometer value is de-
creasing by approximately 0.024 cm per°C. That is, if
would need to lower its price from $1.00 a can to
the blanching temperature is increased to 46°C, the
about $0.47 a can.
consistometer value should decrease by approximately
c. Rather than lower its prices so drastically, Coke should
0.024 cm.
probably consider such alternatives as more advertis-
f.
ing on campus.
f
23. For a function f with inputs x and y, the ratio x represents
fy Storage time Temperature (Celsius)
how much more tilted a tangent plane at point (x, y, f ) is (months) 35° 47° 59° 71° 83°
f
fy
in the x direction than in the y direction. If x 1, then 0
3.0
3.2
3.0 2.8 2.6 2.6 2.8
2.8
3.0
the tangent plane is tilted more in the x direction. If 1 3.3 3.1 2.8 2.8 3.0
3.2
fx
fy
1, then the tangent plane is tilted more in the y 2
3
3.5
3.4
3.2
3.2
3.0
3.0
2.9
2.9
3.2
3.2
f
fx
direction. When x 1, the tilt in the x direction is the 4 3.2 2.9 2.7 2.7
2.8
3.0
same as that in the y direction.
2. a. E(24, 60) 466 eggs. A female insect kept at 24°C and
60% relative humidity will lay approximately 466 eggs
Chapter 9 Content Review in 30 days.
E
b. For t 27°C and h 77%, h 2.2 eggs per per-
1. a. t s
centage point. When the temperature is held constant
°C Months
Inputs at 27°C and the relative humidity is 77%, the number
of eggs is decreasing by 2.2 eggs per percentage point
of relative humidity. That is, if the relative humidity
were increased to 78%, the number of eggs would de-
Rule C
crease by approximately 2.
dh E (299.7038 10.442t 0.4023h)
c. dt Eh t 23.1412 0.1874h 0.4023t percent-
Output age points per°C
C(t, s) dh
Centimeters d. When t 25°C and h 63%, dt 10.4 per-
b. 2.9 cm. When apples are stored for 3 months and dh
centage points per°C. Thus h dt t (10.4)
blanched at 71°C, the applesauce flows 2.9 cm down a
(0.5) 5.2 percentage points.
vertical surface in 30 seconds.
Chapter 10 Answers to Odd Activities A-77
e. On the contour curve corresponding to the egg pro- a table when determining critical points and absolute
duction for t 25°C and h 63% (the 490.1601 egg extrema.
contour curve), the slope of the tangent line at 3. The point is a relative maximum point because the values
dh
t 25°C and h 63% is dt 10.4 percentage of the contour curves decrease in all directions away from
points per°C. The h-value on the tangent line when the point.
t 24.5°C is h 63 5.2 68.2%. This is an ap- 5. a. Relative
R(g, h)
proximation to the value of h that corresponds to 100 maximum
t 24.5°C on the 490.1601 egg contour curve.
3. a. Inputs: t 26°C, h 56%; Output: approximately 80
485 eggs; When temperatures remain near 26°C and
60
humidity levels remain near 56% over a 30-day period,
an average female Catolaccus grandi will lay 485 eggs. 40
b. An increase in temperature of 3°C
c. The number of eggs laid will decrease more rapidly 20
when the temperature decreases than when the hu- 0 6
midity decreases. 0
2 4
4 h
d. By sketching a tangent line on the figure and calculat- g 6 2
ing its slope, you should obtain an estimate of approx- 8
Saddle
imately 13 percentage points per °C.
point
dh E 24.1532
e. dt Eh t 1.8672 12.9 percentage points
per °C
When the temperature is 24°C and the humidity is h
62%, the change in humidity needed to compensate 6
for a small change in temperature (so that the number 10
of eggs remains constant) can be estimated as 5 25
h (12.9)t percentage points. Relative
4. t h 4 maximum Saddle
point
t 10.442 0.4023 85
40
70
h 0.4023 0.1874 3 55 55
40
70
CHAPTER 10 2
85
Section 10.1 25
1
1. a. A relative maximum occurs when a table value is 10
greater than all the values surrounding it. 0 g
0 2 4 6 8
b. A relative minimum occurs when a table value is less
b. Relative maximum point: g 2, h 3, R 95
than all the values surrounding it.
Saddle point: g 6, h 3, R 30
c. If a value appears to be a maximum in one direction
7. The point is a saddle point because it is a maximum con-
but a minimum in another direction, then the value
tour level along a cross section extending from the x 4, y
corresponds to a saddle point.
0 corner diagonally back through the point and is a min-
d., e. If all the edges of a table are terminal edges, then the
imum contour level along a cross section extending from
absolute maximum and minimum are simply the
the x 0, y 0 corner diagonally through the point.
largest and smallest values in the table. If all the edges
9. Relative maximum point: (May, 1995, $1.45 per pound);
are not terminal edges, then you must know whether
Relative maximum point: (May, 1998, 88 cents per pound)
any critical points exist outside the table in order to
determine whether absolute extrema exist. If no criti- 11. a. Yes; The table gives yearly averages, so it doesn’t make
cal points exist outside the table, then in determining sense to extend the columns. However, the choice of
absolute extrema, you must consider relative ex- January as the first column is not mandatory. The best
trema, output values on terminal edges, and the be- way to visualize this table is to wrap it around a cylin-
havior of the function beyond the edges of the table. der so that the January and December columns are ad-
jacent columns and there are no left and right edges of
It is often helpful to sketch several contour curves on
the table. The top and bottom rows are terminal edges.
A-78 CHAPTER 10 Analyzing Multivariable Change: Optimization
b.
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
North Pole 3.1 6.9 8.9 7.9 4.9 0.80
80° 0.77 3.4 6.9 8.8 7.8 4.8 1.55 0.13
70° 0.51 2.0 4.5 7.1 8.5 7.8 5.5 2.9 0.95 0.06
60° 0.55 1.53 3.4 5.7 7.7 8.8 8.2 6.5 4.3 2.2 0.81 0.34
50° 1.66 2.8 4.7 6.7 8.4 9.1 8.8 7.4 5.5 3.4 1.97 1.35
40° 3.0 4.2 5.9 7.5 8.8 9.3 9.0 8.1 6.5 4.8 3.4 2.6
2
9.
.2
30° 4.4 5 5.6 6.9 8.1 9.0 9.2 9.1 8.4 7.4 6.1 4.7 4.1
.2 .2
20° 5.8 6 6.7 7 7.8 8.5 8.8 8.9 8.8 8.6 8.0 7.1 6.0 5.5 5.2
10° 7.1 7.7 8.2 8.3 8.5 8.4 8.3 8.3 8.4 8.3 7.9 7.2 6.8 6.2
Equator 8.1 8.5 8.6 8.3 7.8 7.5 7.6 8.0 8.4 8.4 8.2 7.9 7.2
10° 8.9 9.2 8.8 8.4 7.7 6.9 6.4 6.5 7.2 8.1 8.6 8.8 8.8 8.2
20° 9.4 9.0 8.1 6.9 5.7 5.1 5.4 6.3 7.5 8.6 9.2 9.5 9.2
30° 9.6 8.8 7.4 5.8 4.4 3.8 4.1 5.2 6.7 8.2 9.3 9.8
40° 9.6 8.3 6.5 4.6 3.1 2.5 2.7 3.9 5.6 7.5 9.1 9.9
50° 9.3 7.6 5.4 3.3 1.84 1.25 1.49 2.6 4.5 6.6 8.7 9.7
60° 8.7 6.6 4.1 2.0 0.72 0.31 0.47 1.36 3.1 5.6 8.0 9.3
70° 8.2 5.5 2.8 0.84 0.38 1.78 4.3 7.2 9.1
80° 8.2 4.7 1.42 0.099 0.62 3.2 7.0 9.3
South Pole 8.1 4.6 0.60 2.9 7.0 9.4
5.2
6.2
7.2
8.2
c. Relative maximum points: (June, 40° N, 9.3 kW-h/m2), d. The greatest radiation level shown in the table is 9.9
(June, North Pole, 8.9 kW-h/m2), (December, 40° S, 9.9 kW-h/m2, which occurs in December at 40°S. The
kW-h/m2) smallest radiation level shown is 0.06 kW-h/m2, which
Relative minima: It is difficult to estimate these accu- occurs in November at 70°N. If we consider the dashes
rately because of the dashes in the table, which we can to be zeros, then the smallest radiation level is zero and
interpret to mean radiation levels of essentially zero. occurs many times in the table.
Thus we conclude that the regions of the underlying e. The absolute maximum value is 9.9 kW-h/m2, and the
function represented by the dashes in the table are those absolute minimum is approximately zero. Because the
in which the minimum radiation level occurs. There are table cannot be extended in any direction, these an-
two such regions: one near the North Pole between Oc- swers do correspond to those in part d.
tober and March and one near the South Pole between f. The largest and smallest values in a table will be the ab-
April and September. If there are two specific relative solute maximum and minimum, respectively, if the
minima of the underlying function, then we estimate table cannot be extended in any direction. That is, ei-
that they occur at the end of December at the North Pole ther the edges are terminal edges or the table “wraps
and in the middle of June at the South Pole. around,” as in this case.
Saddle points: (April, 10°N, 8.5 kW-h/m2), (June, 70°N, 13. a. The expected corn yield is 100% of the annual average
8.5 kW-h/m2), (August, 10°N, 8.4 kW-h/m2), (December, yield. That is, there is no expected increase or decrease
70°S, 9.1 kW-h/m2) in yield from the average.
b. 40%
Chapter 10 Answers to Odd Activities A-79
c. See table below.
∆P
(%)
-100 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80
6 0 7 14 18 23 27 32 40 48 56 64 67 69 72 74 75 75 75 76
0 7 15 20 25 29 34 43 51 59 67 70 73 75 78 78 78 78 78
5 0 8 16 21 26 32 37 45 54 62 71 73 76 79 81 81 81 81 81
0 8 17 22 28 34 39 48 57 65 74 77 79 82 85 85 84 84 84
4 0 9 18 24 30 36 42 51 60 68 77 80 83 86 88 88 88 87 87
0 9 19 25 32 38 45 54 62 71 80 83 86 89 92 91 91 90 89
3 0 10 20 27 33 40 47 56 65 75 84 87 90 93 96 95 94 93 92
90
0 10 21 28 35 42 49 59 69 78 87 91 94 96 98 97 95 94 93
2 0 11 21 29 37 44 52 63 74 82 90 95 99 99 100 98 97 96 94
0 11 22 30 38 46 54 65 76 85 93 99 101 101 102 100 99
10 97 95
95
1 0 12 23 31 40 48 56 67 78 87 95 102 103 104 104 102 100 0 98 96
∆T 0 12 24 33 41 50 58 69 80 90 98 104 105 106 106 104 102 100 97
105
0 0 12 25 34 43 51 60 71 83 92 100 107 107 108 109 106 103 101 98
(°C) 0 13 26 35 43 52 61 72 84 93 101 107 108 108 107 105 102 100 97
-1 0 13 27 35 44 53 61 73 85 94 103 108 109 108 106 104 101 99 96
0 14 28 36 45 53 62 73 84 93 102 107 108 107 105 103 100 98 95
-2 0 14 29 37 46 54 63 73 84 93 101 105 107 106 104 102 99 97 94
0 15 29 38 46 55 63 72 82 90 98 101 103 103 103 101 98 95 93
-3 0 15 30 39 47 55 64 72 80 88 96 97 99 101 102 100 97 94 92
0 15 30 38 46 54 62 70 77 85 92 94 95 98 98 96 94 91 89
-4 0 15 31 38 46 53 61 68 74 81 88 90 91 93 95 92 90 88 85
0 15 31 38 45 52 59 65 72 78 85 86 88 89 91 89 87 85 82
-5 0 15 31 38 44 51 58 63 69 75 81 82 84 85 87 85 83 81 79
0 16 31 37 44 50 56 61 67 72 77 79 80 82 83 81 80 78 76
-6 0 16 31 37 43 49 55 59 64 69 74 75 76 78 79 78 76 75 73
d. The maximum percentage yield is 109%. This maxi- 17. a.
mum occurs twice, at the points (40%, 0°C, 109%)
Temperature (°C)
and (20%, 1°C, 109%). A yield of 109% above nor- Storage time
mal can be expected when temperatures are average (months) 35° 47° 59° 71° 83°
and there is 40% more precipitation than normal or
2.7
0 3.0 2.8 2.6 2.6 2.8
when temperatures are 1°C below normal and precip- 3.0
itation is 20% above normal. 1 3.3 3.1 2.8 2.8 3.0
15. a. absolute minimum: 156 kg, 37.8°C, 1.15 kg/day) 2 3.5 3.2 3.0 2.9 3.2
relative minimum: none
3 3.4 3.2 3.0 2.9 3.2
absolute maximum: (156 kg, 15.6° C, 1.09 kg/day)
relative maximum: (91 kg, 21.1° C, 1.01 kg/day) 4 3.3 3.2 2.9 2.7 2.7 3.0
b. 3.0 2.7
°C 4.4 10 15.6 21.1 26.7 32.2 37.8
b. Saddle point: (63°C, 2.25 months, 2.9 cm)
°F 40 50 60 70 80 90 100 c. Absolute maximum: 3.5 cm
Absolute minimum: 2.6 cm
c. On average, the daily weight gain will be least for a pig
d. One possible answer: The saddle point is important
weighing 156 kg when the temperature is 100° F, and
because it shows the optimal storage time and temper-
it will be greatest for a pig weighing 156 kg when the
ature to keep applesance from getting too thick or too
temperature is 60° F.
thin.
A-80 CHAPTER 10 Analyzing Multivariable Change: Optimization
19. a. Section 10.2
B
1. A relative minimum of 16 is located at k 4 and m 2.
3. A saddle point is located at (ln 3, 0, 0).
A X A' 5. A relative minimum of approximately 68.35 is located
at z 4.72 and w 18.49. (0, 0, 0) is the location of a
3 2
X 1 saddle point.
Bone 7. Relative minimum point: (0, 0, 60)
Relative maximum point: (2, 1, 68)
Saddle point: (0, 1, 64)
Muscle Saddle point: (2, 0, 64)
9. a. The manager should try to buy ground beef at $1.91 a
pound and sausage at $2.52 a pound.
b. Because the determinant is positive and Rbb is negative,
Fat
the critical point corresponds to a relative maximum.
B' c. $28.5 thousand
11. a. The critical point is (26.1°C, 67.4%, 500 eggs).
The absolute maximum nearest the fat layer has a b. When exposed to 26.1°C and 67.4% relative humid-
value between 12 and 13, and the relative maximum at ity, a C. grandis female will lay approximately 500
the bone has a value between 3 and 4. eggs in 30 days. This is the maximum number of eggs
b. Both points are maxima. possible.
21. a., b.
MISSOURI
24-Hour Precipitation Totals
M
M issis
i ss
Ending 7 June 1993
ou
ri
(millimeters)
si p
Rive
pi
R
r
iv
125
er
50 25
100
Area of 75
50
greatest
rainfall
>125 mm
c. There is a relative maximum in the simple, closed, 50-mm contour curve near the eastern side of Missouri.
Chapter 10 Answers to Odd Activities A-81
13. a. A pH of approximately 5.3 and a temperature of ap- Section 10.3
proximately 59.3°C will maximize the initial rate.
1. a. One possible estimate is (45, 45, 2025).
b. We verify that this is a maximum by calculating the
b. The point is a constrained maximum.
determinant of the second partials matrix:
c. Solving the system of equations
D 9.81 3
3 14.106 129.38 b
Because D is greater than zero and Rpp is less than zero, a
the critical point must be a maximum. a b 90
15. a. The maximum appears to be approximately 2.35 mg gives a 45 and b 45 , corresponding to
when pH 9 and temperature 65°C. f(45, 45) 2025.
b. A pH of approximately 9.02 and a temperature of ap- 3. a. Approximate relative minimum point: (35%, 17%,
proximately 64.6°C will maximize the amount of pep- 2.5). The minimum measure of cohesiveness is about
tides produced. 2.5, which occurs when the percent of glucose and
17. The model has a saddle point for s 0.19, w 0.18, and maltose is 35 and the percent of moisture is 17.
c 0.63. b. M
19. a. The critical points are A (0.5185, 0.4994, 799.91) (%)
and B (0.8872, 0.4994, 800.16). 26
b. Point A is a saddle point, and point B corresponds to a 24
relative maximum. 22
c. The determinant at A is D 55.98, which indicates 5 6 7
20 4 8 9
that A is a saddle point. Because the determinant at B 3 10
18
is D 55.98 and Enn is negative, point B must corre-
16
spond to a relative maximum.
14
21. a. The critical point is g 42.6%, m 17.1%,
s 11.2%, h 12.8 days, and A 7.29. 12
b. By evaluating the function at many points close to the 10
critical point, it is possible to conjecture that the point 8
G
is a relative maximum. There are also other methods
25 30 35 40 45 (%)
of arriving at this conjecture.
23. In order to locate a critical point of a three-dimensional g (G, M ) = 40%
function, determine both first partial derivatives, set each
of them equal to zero, and solve. Any input (a, b) for The constrained minimum is approximately 7, which
which both first partial derivatives are zero will yield a occurs when G 26% and M 15% .
critical point.
To determine the type of critical point found, write the 5. a. The constrained optimal point is 16 , 8, 32 . 1 3 7
four second partial derivatives of the function. Next eval- p
uate the product of the non-mixed second partials minus 0.8
0.1
the product of the mixed partials at (a, b)—this is known
as D(a, b), the determinant of the second partials matrix. 0.6
g(r, 0.2 0.4
0.5
If D(a, b) is negative, then (a, b) yields a saddle point. If p) = 0.3
1
D(a, b) is positive, then evaluate one of the non-mixed 0.4
0.3
second partials at (a, b). If that second partial is negative, 0.4
0.5 0.2
then (a, b) yields a maximum. If it is positive, then (a, b) 0.2
yields a minimum. Finally, if D(a, b) is exactly zero, then 0.1
r
the determinant test fails and estimation graphically or -0.4 -0.2 0 0.2 0.4
numerically may help.
-0.2
A-82 CHAPTER 10 Analyzing Multivariable Change: Optimization
b. i. The contour graph confirms that the point is a con- c. The relative minimum when there are no constraints
strained minimum, because the contour to which is approximately 2.5, which is obtained when the per-
the constraint line is tangent is the smallest-valued centage of glucose and maltose is approximately 34.7
contour that the constraint line touches. and the percentage of moisture is approximately 16.4.
1 dP
ii. Evaluating f (r, p) for values of r near 16 and p near 17. a. 3.15 radios per thousand dollars
dc
3 7
8 gives values greater than 32 , which suggests that b. P (3.15)(1.5) 4.7 5 radios
the point is a constrained minimum. c. Approximately 56 radios
7. 10.45%; The approximation is 0.01% less than that in
9 2
Example 2 for the constraint w s 1 was 10.46%. 19. a. S(r, h) 2rh r2 r 8 square inches
9. a. The health club should allocate $336 for 28 radio ads
and $168 for 28 newspaper ads. when the radius is r inches and the height is h inches.
b. A(28, 28) 2195 responses b. V(r, h) r2h 808.5 cubic inches
c. The Lagrange multiplier is 13.07 responses per c. Solving the equations
dollar. The change in the number of responses can be
approximated as 2h 2r 2 r 9
8
2rh
A (13.07 responses per dollar) ($26) 2r r2
340 additional responses r2h 808.5
11. a. The point of absolute minimum cohesiveness is gives r 4.87 inches, h 10.86 inches, and
(34.7%, 16.4%, 2.5). This value is close to the estimate S(r. h) 519.5 square inches.
in Activity 3. f f
b. The minimum cohesiveness is approximately 7.2, 21. The condition x y is equivalent to guaranteeing that
gx gy
which corresponds to approximately 25.4% glucose the slope of the extreme-contour curve is the same as the
and maltose and 14.6% moisture. slope of the constraint curve at their point of intersection.
c. To verify that the value in part b is a minimum, eval-
uate the cohesiveness function at close points on the
constraint curve, or examine the constraint curve Section 10.4
graphed on a contour graph of the cohesiveness 1. a. f(a, b) (7 a b)2 (11 6a b)2
function. (19 12a b)2
d. The answers are comparable.
f
13. a. 750 worker hours and $9375 in capital expenditures b. a 2(7 a b) 12(11 6a b)
will yield maximum daily production. 24(19 12a b)
b. To verify that the value in part a is a maximum, evalu-
f
ate the production function at close points on the con- b 2(7 a b) 2(11 6a b)
straint curve, or examine the constraint curve graphed 2(19 12a b)
on a contour graph of the production function.
c. The marginal productivity of money is
D
362 38
38 6
728
3.15 radios per thousand dollars. An increase in
c. The minimum value of f (a, b) is approximately 1.407.
the budget of $1000 will result in an increase in output
We verify that this is a minimum by noting that D 0
of approximately 3 radios.
and faa 0.
dC
15. a. 0.85 unit per percentage point d. y 1.099x 5.374
dk
b. C (0.85 unit per percentage point) 3. a. f(a, b) (3 b)2 (2 10a b)2
(2 percentage points) 1.7 units. The minimum (1 20a b)2
cohesiveness measure should decline by approxi- b. The minimum value of f (a, b) is zero. We verify that
mately 1.7. this is a minimum by calculating D 2400 and
faa 1000 and noting that both are greater than zero.
Chapter 10 Answers to Odd Activities A-83
Because the minimum SSE is zero, we know that the 9. a.
line of best fit is a perfect fit—that is, all of the data Population
points lie on the line. (billions)
c. y 0.1x 3 percent, where x is the number of
8
years since 1970, 0 x 20.
7
d. No, the extrapolation is too far from the data used to
find a trend. 6
5. a., b. 5
y 4
Cost 3
(dollars) 2
1
18
Years since
0
0 80 125 163
1850
15
12 The data points appear to be concave up.
b.
9
Natural log
6 of population
x
3 2
0 1 2 6 9 14
Cases
1
y 1.176x 1.880 dollars to make x cases of ball
bearings. The variable costs to make a case of ball bear-
ings is $1.176. The fixed cost for making ball bearings Years since
0
is $1.880. 1850
0 80 125 163
c. The deviations are approximately 0.044, 0.018, 0.013,
0.176, and 0.102; SSE 0.044.
The data are also concave up, but less so than in part a.
d. To find the best-fitting line, first construct the func-
tion f with inputs a and b, which represents the sum of c. y 0.012x 0.037, gives the natural log of the pop-
the squared errors of the data points from the line ulation in billions x years after 1850, 0 x 163.
y ax b. Find the partial derivatives of f with re- d.
spect to a and b. Simplify the partials, and find the y
point (a, b) where the partials are simultaneously zero. Population
These are the coefficients of the model given in part b. (billions)
The function f evaluated at (a, b) gives the value of SSE
shown in part c. 8
7. y 0.089x 5.484 million experiments x years af- 7
6
ter 1970, 0 x 10.
5
4
3
2
1 x
0 Years since
0 80 125 163 1850
y e0.037(e0.012)x billion people x years after 1850
A-84 CHAPTER 10 Analyzing Multivariable Change: Optimization
e. y 0.964(1.012x) billion people x years after 1850. b. The maximum crispness appears to be about 11.4,
The model in part d simplifies to this model. when the blanching temperature is 70°C and the
11. A single large outlier will not have as profound an influ- blanching time is 15 minutes.
ence on the overall fit of the line if absolute errors are c. A crispness index of approximately 10 occurs at
used as it would if squared errors are used. However, x 26.4 minutes and y 62.7°C.
algebraically simplifying (and solving) the sum of ab- d. Because the determinant of the second partials ma-
solute error expressions is much more complicated than trix is greater than zero (D 1.92) and Cxx is nega-
simplifying the sum of squared algebraic expressions. tive (Cxx 10.4), the point found in part a is a
maximum. Yes, There is only one critical point, and
the table of values indicates that the function de-
Chapter 10 Concept Review
creases in all directions away from that critical point.
1. a. These two observations lead to the conclusion that
the relative maximum is also an absolute maximum.
6 7 8 9 10 11 12 1 2 3 4 5 6
90° 90° 3. a. The maximum profit is approximately $202,500 for
N 300 N about 52,000 shirts and 9000 hats. (Answers may vary.)
440
340 340 420 b. 4s 1.25h 150
360 S 400
60° 320 320 60° c.
380
h
300
Hats
30° 280 30° (thousands)
260
20
Latitude
L
0° 0° 15
S
260 260 200,000
280 10
195
,00
30° 300 30° 0
5 190,000 000
360 180,00
0 185, s
340 300
380 400 320 0 175,000 T-shirts
H 320
60° 60° 30 40 50 60 70 (thousands)
360 340
300 L The constrained maximum profit is slightly more than
S H 280 S $185,000 for about 34,000 shirts and 10,000 hats.
90° 90°
6 7 8 9 10 11 12 1 2 3 4 5 6 4. a. 104.4s 5935.5 59.1h 4
59.1s 10,299.3 768.6h 1.25
Month
4s 1.25h 150
b. Highest ozone level 450 thousandths of a centime- b. h 10.04 thousand hats, s 34.36 thousand shirts,
ter at 90° N (the North Pole) in mid-March. Lowest P $186,599
ozone level 250 thousandths of a centimeter at or c. 439 dollars of profit per thousand dollars spent;
just north of the equator between October and For each additional thousand dollars budgeted, profit
March. will increase by approximately $439.
2. a. See table below. d. Profit would increase by approximately $877.
5. a. f(a, b) (29.9 10a b)2 (33.4 15a b)2
Blanching time (minutes) (37.5 20a b)2
Blanching temperature
(°C) 2 15 30 b. The minimum value of f is 0.06. We verify that this is
a minimum by calculating the determinant of the sec-
50 4.2 6.8 4.2
6 6 ond partials matrix (D 600) and noting that it is
60 4.5 4.7 7.1 8 positive and that faa 1450 is also positive. The min-
70 8.2 11.4 10 8.6 imum value is the sum of the squares of the deviations
6
of the data from the best-fitting line.
80 7.3 6.9 3.9
c. y 0.76x 22.2 kg, where the temperature is x°C
Subject Index
of probability density functions, Power Rule for, 329, 332
A 437–439 of products of functions, 370–373
Absolute extrema starting value for, 310 of quotients of functions, 370–373
definition of, 249, 278, 616 units of measure of, 319 of sine functions, 494–495, 501
end behavior and, 250, 621 Addition, function specific (see Specific antiderivatives)
finding of, 251, 621 for accumulated change, 295, Sum Rule for, 330, 332
graphs of, 616–617, 621–623 298–299 from tabular data, 333–334
of linear functions, 249 definition of, 10–11 Approximating change
in multivariable functions, 616–617, derivatives and, 180–184, 194, 222 concavity and, 240
621–623 input/output diagrams for, 10, 11 in cubic functions, 232–233
in normal distribution, 434 input variables in, 10, 85 formula for, 230
of quadratic functions, 250 notation for, 391–392 from graphs, 373
from tabular data, 621–623 output variable in, 10–11, 14, 85 in multivariable functions, 598–602,
on terminal edges, 621–623 rules for (see Sum Rules) 609
Acceleration Sum Rules for (see Sum Rules) in quadratic functions, 229
definition of, 539 units of measure of, 14, 85 in quartic functions, 231–232
due to gravity, 335 Aligning data, 32–33 with tangent lines, 228–232,
input/output diagrams for, 320 Amplitude of trig functions, 454, 457, 598–599
units of measure of, 319–320 462 APR. See Annual percentage rate (APR)
velocity and, 334–335 Annual percentage rate (APR) APY, 103–106, 138–140
Accumulated change. See also Definite average change from, 103–106 Area
integrals continuous income stream and, in accumulated change, 285–298, 370
area and, 285–298, 370 388–389 of accumulation functions, 326–327
in cubic functions, 288–290, 296–300 differential equations for, 521–522 of circles, 293
definition of, 287 discrete income stream and, between curves, 348–351, 374
difference of, 346–351 392–393 definite integrals and, 336, 340–351
from equations, 290–294, 296–300, instantaneous change from, 138–140 of demand functions, 400–407
309 multivariable formula for, 550 in histograms, 423–425
in exponential functions, 290–292, Annual percentage yield (APY) as limit of sums, 295, 374
303 average change from, 103–106 median, 430
from graphs, 285, 287, 289, 291, instantaneous change from, of normal density functions, 434
293–298 138–140 of probability density functions,
graphs of, 292 Antiderivatives. See also Accumulation 426, 428
integrals and, 299–300, 339–340 function; Integrals signed, 287
limits of, 294–299 of composite functions, 371–372 of supply functions, 410–414
in linear functions, 284–287 of constant functions, 328, 329, 332 surface, of cans, 656
in logarithmic functions, 290–292 Constant Multiplier Rule for, 330, 332 of total social gain, 415–417
in piecewise functions, 290–292 constant term in, 328, 342 of trapezoids, 286, 327
in quadratic functions, 309 of cosine functions, 494–495, 501 of uniform density functions, 429–430
in sine functions, 495–496 of cubic functions, 330 Avection, 272
from tabular data, 287–290 definition of, 328, 374 Average. See also Mean
units of measure of, 286, 289–290 Difference Rules for, 330, 332 costs, 9, 13
from words, 284–290 of differential equations, 508–512 definition of, 430
Accumulated density-dependent from equations, 330–331 rate of change (see Average rate of
mortality, 357 of exponential functions, 331–334, change)
Accumulated value of investments, 550 370–372 value (see Average value)
Accumulation function. See also Exponential Rule for, 332 Average costs, 9, 13
Antiderivatives ex Rule for, 331, 332 Average rate of change
algebraic method for, 339–340 general (see General antiderivatives) average value and, 364
area of, 326–327 from integrals, 329, 340–341 from continuously compounded
definition of, 310, 336, 374 of linear functions, 329, 330–331 interest, 105–106
derivative of, 325–328, 374 logarithmic functions from, definition of, 96, 364
of exponential functions, 341 331–334, 341–342, 371 from equations, 102–106, 363–365
extrema of, 316–318 Natural Logarithm Rule for, 331, 332 of exponential functions, 365
graphs of, 310–318, 325–327 notation for, 328, 336, 341 formula for, 96, 362, 364, 374
input/output diagrams for, 319 of power functions, 329–331, 370, from graphs, 98–102
notation for, 336 372–373 graphs of, 365
A-85
A-86 Subject Index
vs. instantaneous rate of change, units of measure of, 97 Constant of proportionality,
129, 157 Circles, area of, 293 510–512
integrals for, 362–363 Cobb-Douglas production function, Constant percentage growth, 86
in linear functions, 285–286 597–600, 648–650 Constant Rule, 178, 184, 194
in revenue, 96 Competitive commodities, 612 Constraint curves
rise/run calculation of, 99 Complementary products, 612 contour curves and, 645–646
secant lines and, 98–100, 157, 365 Composite functions extrema on, 646–647, 650, 652
slope and, 98–100 antiderivatives of, 371–372 shifting of, 650
from tabular data, 97–98, 112–113, construction of, 15–18, 85 tangent lines to, 646, 662
359–362 definition of, 17 Constraint functions
units of measure of, 97, 363–364 derivatives of, 198–203 definition of, 645
Average value input/output diagrams for, 16, 17 graphs of (see Constraint curves)
average rate of change and, 364 inverse functions and, 52 multivariable functions and,
from equations, 363–365 multivariable, 600–601 645–653
of exponential functions, 365 slope of, 201 Consumers’ expenditure and surplus,
formula for, 361, 364, 374 Composition of functions 254, 404–407, 414–417
graphs of, 365 inside/outside functions in, 85, 200 Consumers’ willingness/ability to
horizontal shift and, 456, 462 units of measure in, 16, 85 spend, 384, 401–404
limits of, 361 variables in, 15–17 Continuity
over intervals, 360–361, 496 Compound interest definition of, 148–149
over piecewise functions, 364 average change from, 103–106 discrete interpretation of, 105
of rate of change, 362–364 continuous (see Continuously in graphs, 6, 121
of sine functions, 454, 456, 462, compounded interest) rates of change and, 129
496–497 for discrete income stream, 390–393 sharp points and, 114, 121
from tabular data, 361–362 exponential functions and, 105 slope and, 164–165, 222
units of measure of, 364 formulas for, 103–104, 162, 550 Continuous functions
vertical shift and, 462, 496–497 instantaneous change from, 138–140 instantaneous rate of change in,
limits of, 105 113–121
Concavity piecewise (see Piecewise functions)
on accumulation graphs, 316 Continuous income stream, 385–390,
B approximating rate of change and, 447
Base of exponential function, 86 240 Continuously compounded interest
Bell-shaped curves. See Normal of cubic functions, 76, 86 average change from, 105–106
distribution definition of, 40 for continuous income stream,
Biological streams, 393–395, 447 of exponential functions, 40–41, 387–390
Boyle’s Law, 277 86, 189 formula for, 105, 162
Break-even point, 9, 14–15 extrema and, 634 instantaneous rate of change from,
Break points, 121–122, 343 inflection point and, 60, 120, 256, 138–140
Brentano-Stevens Law, 525 261 limits of, 105
instantaneous rate of change and, 120 Contour curves
leading coefficient and, 72 constraint curves and, 645–646
of logarithmic functions, 47, 86 definition of, 552
C of logistic functions, 60, 86 from equations, 555–558
Calculus, definition of, 2 at point of tangency, 118, 120 parts of, 553
Cancellation Rule for limits, 150 of quadratic functions, 69, 72, 86 slope of, 601
Cans, surface area of, 656 second derivative and, 262–263, 278 spacing of, 553, 561
Capital value, 398 tangent lines and, 118, 120 steepness on, 553, 560–561
Carrying capacity, 60, 526 Constant behavior, 28 on tabular data, 552–556, 618–623
Certain event, 425 Constant differential equations, 508, tangent lines to, 576–577, 584,
Chain Rule, 198–203, 222–223 510 601–602, 646
Change Constant functions on topographical maps, 552–554,
accumulated (see Accumulated antiderivatives of, 328, 329, 332 560–561
change) derivatives of, 222 on weather maps, 572
approximating (see Approximating in differential equations, 508 Contour graphs
change) slope fields for, 515 critical points on, 616–624
definition of, 157 uniform density, 429–430, 444 in cross-sectional modeling,
incremental (see Incremental change) Constant Multiplier Rule 574–575
percentage (see Percentage change) for antiderivatives, 330, 332 curves on (see Contour curves)
rate of (see Rate of change) for derivatives, 179–180, 184, 194 definition of, 553
Subject Index A-87
Counter graphs (continued) integration of, 299–300, 346–351 interpretation of, 401
output estimates using, 558–561 limits of, 77, 218 Demand functions
rate of change from, 559–560, vs. quadratic functions, 77 area of, 400–407
576–577 second derivative of, 256–259 consumers’ expenditure and surplus,
Controlled variable, 5. See also Input slope fields for, 520 254, 400–407, 414–417
variables slope graphs of, 242, 257 consumers’ willingness/ability to
Cosine functions from tabular data, 77–79 spend, 384, 401–404
antiderivatives of, 494–495, 501 third differences and, 74–75 examples of, 406–407
derivative of, 478–480 Cumulative density functions, graphs of (see Demand curves)
derivatives of, 478–480 435–439, 444 in market equilibrium, 414–415
differential equations for, 540 Cyclic functions Density-dependent mortality factors, 357
end behavior of, 65 cosine (see Cosine functions) Density functions
extrema of, 487–488 definition of, 452 cumulative, 435–439, 444
formula for, 453 sine (see Sine functions) exponential, 432–433, 444
graphs of, 453, 474 normal (see Normal distribution)
inflection points of, 488–489 probability (see Probability density
integration of, 494–495, 501 functions)
limits of, 65 D uniform, 429–430, 444
sine function, relationship to, 453, Decay Dependent variable. See also Output
474–475, 494 constant percentage, 86 variable
slope fields for, 520 exponential, 41, 42, 45–46, 86 definition of, 5, 270
slope graph of, 474, 478 half-life, 45 in related-rates problems, 273, 278
slope graphs for, 474, 478 logarithmic, 49, 86 Derivatives. See also Instantaneous rate
Costs logistic, 86 of change
average, 9, 13 Decreasing behavior, 28, 64 of accumulation functions, 325–328,
at break-even point, 9 Definite integrals. See also Specific 374
fixed, 9–10 antiderivatives addition of functions and, 180–184,
marginal (see Marginal costs) accumulated change and, 299–300, 194, 222
profit and, 9, 234–235 339–340 Chain Rule for, 198–203, 222–223
total, 9–11, 334 antiderivatives and, 340–341 of composite functions, 198–203
variable, 9–10 area and, 336, 340–351 of constant functions, 222
Cotterill price index, 566 of cubic functions, 299–300, Constant Multiplier Rule for,
Critical points 346–351 179–180, 184, 194
on contour graphs, 616–624 definition of, 299, 336, 340–341, Constant Rule for, 178, 184, 194
definition of, 662 374 of cosine functions, 478–480
extrema (see Extrema) difference between, 346–351 of cubic functions, 155, 183,
of multivariable functions, 615–616, from equations, 299–300, 340–351 201–202, 256–259
634–641 of exponential functions, 341–343, of cumulative density functions, 438
saddle points (see Saddle points) 349–351, 382 definition of, 129, 157
in tables, 618–624 improper, 380–383, 447 derivatives of (see Second derivative)
Cross-sectional models, 573–577, 608 of piecewise functions, 343–344, Difference Rules for, 180–184, 194
Cubic functions 346–348 in differential equations (see
accumulated change in, 288–290, of power functions, 340–341 Differential equations)
296–300 of quadratic functions, 343–346 division of functions and, 212
aligning data for, 79 Sum Property of, 344 from equations, 193
antiderivatives of, 330 units of measure of, 348 of exponential functions, 189–192,
approximating change in, 232–233 Demand 194, 201, 222
behavior of, 75, 86 elasticity of, 407–409 Exponential Rule for, 191, 194
concavity of, 76, 86 graphs of (see Demand curves) ex Rule for, 189, 194
derivatives of, 155, 183, 201–202, percentage rate of change of, extrema and, 242, 244, 256–258,
256–259 407–408 634–636
differential equations for, 539–540 price and, 400 formula for, 151–152
end behavior of, 75, 86 Demand curves Four-Step Method for, 152–154
extrema of, 241–243, 256–259 for consumers’ expenditure and from graphs, 133–134, 222
formula for, 75, 77 surplus, 404–406 graphs of, 132, 263 (see also Slope
graphs of, 75–77, 86, 241, 257, 288 for consumers’ willingness/ability to graphs)
horizontal asymptote of, 86 spend, 401–403 inflection point and, 256–263, 278
inflection point of, 76–77, 80, definition of, 400 of inside functions, 200–203
256–259 equilibrium point on, 414 interpretation of, 130–132
A-88 Subject Index
Linear Function Rule for, 178, 184, for logistic functions, 526–528 logarithmic functions and, 52
194 particular solutions to, 508, 512, 544 slope fields for, 522
of linear functions, 155, 178, 184, for percentage rate of change, Effective rate, 103–104. See also Annual
222 521–522 percentage yield (APY)
of logarithmic functions, 192, 222, for power functions, 513 Elasticity of demand, 407–409
331 proportionality in (see Empirical Rule, 443
of logistic functions, 202–203 Proportionality) End behavior. See also Limits
multiplication of functions and, for quadratic functions, 509–510 of cosine functions, 65
207–212, 222 response, 511, 525 of cubic functions, 75, 86
Natural Logarithm Rule for, 192, 194 second derivative in, 539–542 of cumulative density functions, 437
notation for, 129, 157 second-order, 539–542 definition of, 28, 63
of outside function, 200–203 separation of variables in, 520–522 of exponential functions, 64, 189
partial (see Partial derivatives) for sine functions, 540–542 extrema and, 250, 621
of piecewise functions, 171–172, 248, from words, 511–512, 521–525, of linear functions, 28, 64
261–262 527–528 of logarithmic functions, 64
of power functions, 155, 179, 184, 222 Differentials, definition of, 521 of logistic functions, 64, 86
Power Rule for, 179, 184, 194 Differentiation, implicit, 270–274 numerical estimate method for,
Product Rule for, 207–212, 222 Direct proportionality, 510 64–65
of quadratic functions, 151–153, Direct substitution method, 218–219 of sine functions, 65
155, 184 Discount rate, 398 . See Elasticity of demand
Quotient Rule for, 212 Discrete income stream, 385, 390–393, Euler’s method, 532–536
secant lines and, 169–171 447 Expected value. See Average
of sine functions, 473–479, 500 Distance, velocity and, 335 Exponential density function,
slope graphs of (see Slope graphs) Distributions, probability 432–433, 444
subtraction of functions and, definition of, 423, 426 Exponential distribution, 433
180–184, 194, 222 exponential, 433 Exponential functions
Sum Rule for, 180–184, 194 normal (see Normal distribution) accumulated change in, 290–292, 303
from tabular data, 183–184 spread in, 430, 434 accumulation function of, 341
units of measure of, 133, 157, uniform, 429–430 aligning data for, 46, 50, 63, 86
198–199 Divergence of improper integrals, 383 antiderivatives of, 331–334,
Determinant Test, 637–641, 645 Division, function 370–372
Deviation, best fit and, 658 antiderivatives of, 370–373 average rate of change of, 365
Dew point, 577 definition of, 13 average value of, 365
Diagrams, input/output. See derivatives and, 212 base of, 86
Input/output diagrams input variables in, 13, 85 behavior of, 64
Difference Rules integration and, 370–373 compound interest and, 105
for antiderivatives, 330, 332 limits and, 218–219 concavity of, 40–41, 86, 189
for derivatives, 180–184, 194 output variable in, 14, 85 for decay, 41, 42, 45–46, 86
Differential equations Quotient Rule for, 212, 218 derivatives of, 189–192, 194, 201,
antiderivatives of, 508–512 units of measure of, 14, 85 222
for APR, 521–522 Domain differential equations for, 521–525,
Brentano-Stevens Law, 525 definition of, 5 542
constant, 508, 510 in graphs and tables, 7 end behavior of, 64, 189
constant functions in, 508 units of measure of, 8 formula for, 41, 42, 86
for cosine functions, 540 Doubling time, 45, 162 fractions as, 193
criteria for, 509 graphs of, 40–42, 47, 86, 189
for cubic functions, 539–540 for growth, 41, 42, 45, 58, 86
definition of, 507 horizontal asymptote of, 64, 522
in Euler’s method, 532–536 E improper integrals of, 380–383
for exponential functions, 521–525, e inflection point of, 80
542 accumulated change in, 303 integration of, 341–343, 349–351,
Fechner’s Law, 511, 525 antiderivatives of, 331–333, 370–372 380–383
general solution to, 508, 512, 544 compound interest and, 105 limits of, 42, 49, 64, 189, 220
graphs of solution to (see Slope derivatives of, 189, 194, 201, 222 logarithmic functions and, 52
fields) differential equations for, 521–525, percentage change in, 41–43, 86, 96
linear, 508–510, 539–540 542 percentage differences and, 43–44
for linear functions, 507–508 in exponential functions, 42 vs. quadratic functions, 73–74
for logarithmic functions, 511–512, integration of, 380–382 rate of change in, 41–43
514, 534–535 limits of, 220 slope fields for, 521–522
Subject Index A-89
Exponential functions (continued) percentage differences and, 43–44, 80 subtraction of (see Subtraction,
square roots as, 193 second differences and, 69 function)
from tabular data, 43–45 slope and, 29 supply, 409–417
from words, 42–43, 45–46 in tabular data, 29 survival and renewal, 394–395
Exponential Rule Fixed costs, 9–10 trigonometric (see Trigonometric
for antiderivatives, 332 Flow rate functions)
for derivatives, 191, 194 continuous, 385–389 Vertical Line Test for, 7–8
Extrapolation of data, 30, 77 discrete, 390 Fundamental Theorem of Calculus,
Extrema Four-Step Method for derivatives, 327–328, 374
absolute (see Absolute extrema) 152–154 Future value
on accumulation graphs, 316–318 Functions of biological stream, 394–395, 447
concavity and, 634 accumulation (see Accumulation of continuous income stream,
conditions for, 248 function) 387–390, 447
on constraint curves, 646–647, 650, addition of (see Addition, function) of discrete income stream, 390–393,
652 Cobb-Douglas production, 597–600, 447
constraints and, 645–653 648–650
on contour graphs, 616–623 combining of, 9–15, 85
of cosine functions, 487–488 composite (see Composite functions) G
of cubic functions, 241–243, 256–259 composition of (see Composition of General antiderivatives. See also
definition of, 241 functions) Indefinite integrals
derivatives and, 242, 244, 256–258, constant (see Constant functions) of composite functions, 371–372
634–636 constraint (see Constraint functions) of constant functions, 328, 329, 332
determinant test for, 637–638 cosine (see Cosine functions) Constant Multiplier Rule for, 330,
end behavior and, 250, 621 cubic (see Cubic functions) 332
endpoints and, 243 cumulative density, 435–439, 444 constant term in, 328
finding of, 251, 621 cyclic (see Cyclic functions) of cosine functions, 494–495, 501
graphs of, 241–242, 245, 615–617, definition of, 5, 85 of cubic functions, 330
621–623, 634–635 demand (see Demand functions) definition of, 328–329, 336
inflection point and, 256–258, 634 division of (see Division, function) Difference Rules for, 330, 332
initial conditions and, 513 examples of, 6 of differential equations, 508–512
Lagrange multiplier and, 646–647 exponential (see Exponential from equations, 330–331
least-squares method for, 659 functions) of exponential functions, 331–333,
of linear functions, 249 exponential density, 432–433, 444 370–372
local (see Local extrema) as general solutions, 508 Exponential Rule for, 332
of logistic functions, 166, 260–261 Hessian of the (see Second partials ex Rule for, 331, 332
of multivariable functions, 615–617, matrix) of linear functions, 329, 330–331
621–623, 634–641 identification of, 6 Natural Logarithm Rule for, 331, 332
in normal distribution, 434 inside (see Inside functions) notation for, 328
of piecewise functions, 247–248 inverse, 51–52 of power functions, 329–331, 370,
of quadratic functions, 243–244, 250 linear (see Linear functions) 372–373
of quartic functions, 245–247 logarithmic (see Logarithmic Power Rule for, 329, 332
relative (see Relative extrema) functions) of products of functions, 370–373
second derivative and, 261 logistic (see Logistic functions) of quotients of functions, 370–373
sharp points and, 247–248 multiplication of (see Multiplication, of sine functions, 494–495, 501
of sine functions, 462, 487–488 function) Sum Rule for, 330, 332
slope at, 242, 634 multivariable (see Multivariable Graphs. See also Input/output
slope graphs and, 242, 245, 261 functions) diagrams; Maps
steps for finding single variable, 251 normal (see Normal distribution) of accumulated change, 292
from tabular data, 621–623 notation for basic, 5, 16 accumulated change from, 285, 287,
tangent lines at, 242, 634 outside (see Outside function) 289, 291, 293–298
on terminal edges, 621–623 periodic (see Periodic functions) of accumulation functions, 310–318,
piecewise (see Piecewise functions) 325–327
power (see Power functions) approximating change from, 373
F quadratic (see Quadratic functions) of average rate of change, 365
Fechner’s Law, 511, 525 quartic, 231–232, 245–247 average rate of change from, 98–102
First differences recovering of, 319 of average value, 365
definition of, 29 role of, 2 axis, scale of, 26
input variables and, 29, 31 sine (see Sine functions) of break-even point, 14–15
linear functions and, 29, 31, 74 smoothness of, 114 change on, 98–102
A-90 Subject Index
of constraint curves functions (see scatter plots, 25, 169 Indefinite integrals, 341. See also
Constraint curves) of secant lines, 99, 102, 117–118, General antiderivatives
of consumers’ willingness/ability to 136–138 Independent variable. See also Input
spend, 401–402 of second derivative, 257 variables
continuity in, 6, 121 shutdown point on, 409 definition of, 5, 270
contour (see Contour graphs) of sine functions, 453–458 in related-rates problems, 273, 278
of cosine functions, 453, 474 slope (see Slope graphs) Indeterminate forms, 219
in cross-sectional modeling, 574–575 slope fields (see Slope fields) Inelastic demand, 408
of cubic functions, 75–77, 86, 241, slope from, 119, 136–138, 169–171 Inflection point
257, 288 smoothness of, 114 on accumulation graphs, 316–318
of demand functions (see Demand standard deviation on, 430 concavity and, 60, 120, 256, 261
curves) steepness of, 25–26, 115, 117 of cosine functions, 488–489
of derivatives, 132, 263 (see also of supply functions, 409–414 of cubic functions, 76–77, 80,
Slope graphs) of tangent lines, 114–118, 120, 256–259
derivatives from, 133–134, 222 136–138 definition of, 60, 165, 256, 278
of differential equation solution (see tangent lines from, 119, 121, 165 derivatives and, 256–263, 278
Slope fields) of total social gain, 415–417 of exponential functions, 80
domain on, 7 of uniform density functions, 429 extrema and, 256–258, 634
equilibrium point on, 414 variables in, 7 initial conditions and, 513
of exponential functions, 40–42, 47, vertical axis intercept on, 25 instantaneous rate of change and,
86, 189 Vertical Line Test for, 7–8 116, 120
of extrema, 241–242, 245, 316–318, Growth of logarithmic functions, 80
615–623, 634–635 constant percentage, 86 on logistic curves, 166
fit in, 80 doubling time, 45, 162 of logistic functions, 60, 80, 260–261
histograms, 423–425 exponential, 41, 42, 45, 58, 86 in normal distribution, 434
horizontal axis intercept in, 27 logarithmic, 49, 86 of piecewise functions, 261–262
inflection point in (see Inflection logistic, 58, 86 as point of diminishing returns, 261
point) point of tangency and, 118, 120
input/output variables on, 7 of quadratic functions, 80
limits from, 145–146 second derivatives and, 256–263,
of linear functions, 24–28, 118,
H 278
285–287 Half-life, 45 of sine functions, 474, 488–489
linear functions from, 27–28 Hessian of the function. See Second slope and, 165, 256
local linearity in, 114–115 partials matrix slope graphs and, 256–257
of logarithmic functions, 47–49, 86 Histograms, 423–425 tangent lines and, 118, 120
of logistic functions, 58–60, 166, 260 Horizontal asymptote Initial condition on slope fields,
of market equilibrium, 414–417 of cubic functions, 86 512–513
mean on, 430 definition of, 63 Input/output diagrams
of multivariable functions, 549, 551, of exponential functions, 64, 522 for acceleration, 320
615–616 (see also Contour of linear functions, 64 for accumulation function, 319
graphs; Cross-sectional models) of logarithmic functions, 64 for addition of functions, 10, 11
of normal distribution, 426, 434 of logistic functions, 58, 64, 86, 526 basic, 5
percentage change from, 98–102 of polynomials, 64 for composite functions, 16, 17
percentage rate of change from, Horizontal axis intercept, 27 for multiplication of functions, 12
134–135 Horizontal shift of sine functions, for multivariable functions, 550
of piecewise functions, 121, 148, 172, 455–457, 462–463 for second derivative, 258
247 Horner tree stem model, 606 for velocity, 320
point of tangency on (see Point of Input variables
tangency) in addition of functions, 10, 85
of probability density functions, 427,
439
I in composition of functions, 15–17
definition of, 4–5
of proportions, 568 Identity, definition of, 508 description of, 35
of quadratic functions, 69–72, 86, Implicit differentiation, 270–274 in division of functions, 13, 85
244 (see also Parabola) Improper integrals, 380–383, 447 first differences and, 29, 31
of quartic functions, 246 Income streams, 385–393, 447 on graphs, 7
range on, 7 Increasing behavior, 28, 64 interval, 35
rate of change from, 98–102, 480–482 Incremental change in multiplication of functions,
rise/run on, 26 first differences (see First differences) 11, 85
of saddle points, 616 in linear functions, 25, 27, 29 in multivariable functions, 549
Subject Index A-91
Input variables (continued) Inverse functions, 51–52 overview of, 86, 217
in second partial matrix, 589 Inverse proportionality, 510 of percentage change, 105
in subtraction of functions, 13, 85 Investments, accumulated value of, 550 of polynomials, 64, 69, 218
in tables, 7 Isobars, 665 of power functions, 218
units of measure of, 8, 35, 85 Isotherms, 572, 665 of price, 402–403, 405–406
Inside functions Product Rule for, 218
in composition of functions, 85, 200 of quadratic functions, 69
definition of, 16 J Quotient Rule for, 218
derivatives of, 200–203 Joint proportionality, 524 saturation level, 60, 526
in multivariable functions, 600 of secant line slope, 118, 121,
Instantaneous rate of change. See also 136–139, 145, 150–154, 157
Derivatives K of sine functions, 65
vs. average rate of change, 129, 157 Sum Rule for, 218
at break points, 121–122 K contour curve. See Contour curves Linear differential equations, 508–510,
concavity and, 120 539–540
in continuous functions, 113–121 Linear Function Rule, 178, 184, 194
from continuously compounded L Linear functions
interest, 138–140 Lagrange multiplier, 646–653, 662 accumulated change in, 284–287
definition of, 114, 157, 178 . See Lagrange multiplier aligning data for, 32–33, 50
examples of, 116–117 Land subsidence, 633 antiderivatives of, 329, 330–331
inflection point and, 116, 120 Leading coefficient, 71, 72 average rate of change in, 285–286
notation for, 129 Least-squares line, 658–659 behavior of, 28, 64
in piecewise functions, 120–122 Least-squares method, 657–659 definition of, 24, 86
point of tangency and, 114 Left rectangles, 289, 292–293 derivatives of, 155, 178, 184, 222
in revenue, 210–211 Level curves. See Contour curves differential equations for, 507–508
secant lines and, 117–118, 121 Leveling-off value, 60, 246 end behavior of, 28, 64
sharp points and, 114, 121–122 L’Hôpital’s Rule, 219–221 equations for, 25, 28, 86
slope and, 113–117, 119, 121 Limit of sums extrema in, 249
tangent lines and, 114–118, 120, for accumulated change, 298 first differences and, 29, 31, 74
121, 157 area as, 295, 374 from graphs, 27–28
terminology for, 129 notation for, 299 graphs of, 24–28, 118, 285–287
units of measure of, 133 Limits. See also End behavior horizontal asymptote of, 64
from words, 113 of accumulated change, 294–299 horizontal axis intercept in, 27
Integrals. See also Antiderivatives of average value, 361 implicit differentiation of, 270–271
for average value of the rate of Cancellation Rule for, 150 incremental change in, 25, 27, 29
change, 362–363 carrying capacity, 60, 526 least-squares method for, 657–659
definite (see Definite integrals) of compound interest, 105 limits of, 64
improper, 380–383, 447 of consumers’ willingness/ability to percentage change in, 31
indefinite, 341 spend, 402 point of tangency in, 118
notation for, 299, 336 of cosine functions, 65 rate of change in, 27–29, 31, 41, 86,
Integration by substitution, 371–372 of cubic functions, 77, 218 178, 284–287
Intercepts of Determinant Test, 645 slope fields for, 515–516
aligning data and, 32–33 direct substitution for, 218–219 slope of, 25–29, 86, 114, 178, 228
horizontal axis, 27 division of functions and, 218–219 starting value in, 25
in linear functions, 25, 27 of exponential functions, 42, 49, 64, steepness in, 25–26
vertical axis, 25, 27, 32–33 189, 220 from tabular data, 25, 29–35
Interest, compound. See Compound from graphs, 145–146 vertical axis intercept in, 25, 27
interest of improper integrals, 380–383 Linear system of equations, 638
Interpolation of data, 30 interpretation of, 147 Line of best fit, 658–659
Intervals L’Hôpital’s Rule for, 219–221 Local extrema, 634. See also Relative
average rate of change in (see of linear functions, 64 extrema
Average rate of change) of logarithmic functions, 48–49, 64, Local linearity, 114–115, 157
average value over, 360–361, 496 220–221 Logarithmic functions
extrema in, 243, 249–250 of logistic functions, 49, 59, 64, 218 accumulated change in, 290–292
in integrals, 299 multiplication of functions and, aligning data for, 50–51, 86
percentage change in (see Percentage 218–219 from antiderivatives, 331–334,
change) notation for, 63, 146, 217 341–342, 371
smoothness over, 114 numerical estimate method for, behavior of, 64
terminology for, 100 64–65, 146–148 concavity of, 47, 86
A-92 Subject Index
derivatives of, 192, 222, 331 Matrix notation for, 550
differential equations for, 511–512, definition of, 589 optimization of, 634–641, 645–653,
514, 534–535 for linear system of equations, 638–639 657–659
end behavior of, 64 for second partial derivatives, output variable in, 549
exponential functions and, 52 588–590, 637 outside function in, 600
formula for, 47–48, 86 Mean. See also Average rate of change in, 559–560, 575–577,
graphs of, 47–49, 86 definition of, 430, 448 597–605 (see also Partial
for growth, 49, 86 examples of, 431–432 derivatives)
horizontal asymptote of, 64 formula for, 431 from tabular data, 550–551
implicit differentiation of, 270–271 on graphs, 430
from improper integrals, 383 in normal distribution, 430–431, 434
inflection point of, 80 notation for, 430
limits of, 48–49, 64, 220–221 of uniform density functions, 444
N
slope fields for, 514–515, 519 Median Natural Logarithm Rule
from tabular data, 49–50 definition of, 430 for antiderivatives, 331, 332
vertical asymptote in, 48 formulas for, 443–444 for derivatives, 192, 194
Logistic curves, 166 Midpoint rectangles, 293–294 Newton’s Law of Cooling, 537–538
Logistic functions Mixed second partial derivatives, 588 Nominal rate, 103–105. See also Annual
aligning data for, 60, 62–63, 86 Models percentage rate (APR)
behavior of, 64, 86 accuracy of, 33–35 Normal distribution, 426, 430–431,
concavity of, 60, 86 choice of, 80, 86 433–435, 443
derivatives of, 202–203 components of, 85
differential equations for, 526–528 construction of, 3–4
end behavior of, 64, 86 cross-sectional, 573–577, 608
extrema of, 166, 260–261 definition of, 3, 35
O
formula for, 58, 59, 86 elements of, 35 Observed variable, 5. See also Output
graphs of, 58–60, 166, 260 examples of, 3–4 variable
for growth, 58, 86 extrapolation vs. interpolation of Optimal points. See Extrema
horizontal asymptote of, 58, 64, 86, 526 data in, 30 Optimization, definition of, 241
inflection point of, 60, 80, 260–261 interpretation of, 8–9 Oscillation equation, 543
limits of, 49, 59, 64, 218 role of, 85 Output variable
second derivative of, 260–261 Moore’s Law, 54 in addition of functions, 10–11, 14, 85
slope fields for, 526–527 . See Mean behavior of, 28, 64
slope graphs of, 166, 260 Multiplication, function in composition of functions, 15–17
slope of, 166 definition of, 11–12 definition of, 4–5
from tabular data, 60–63 derivatives and, 207–212, 222 description of, 35
from words, 59–60 input/output diagrams for, 12 in division of functions, 14, 85
input variables in, 11, 85 on graphs, 7
integration and, 370–373 limits of (see Limits)
limits and, 218–219 in multiplication of functions, 11,
M output variable in, 11, 14, 85 14, 85
Maps units of measure of, 14, 85, 211 in multivariable functions, 549
contour curves on (see Contour Multivariable functions proportionality of, 510, 524
curves) approximating change in, 598–602, 609 in subtraction of functions, 13, 14, 85
topographical, 552–554, 560–561 constraints on, 645–653 in tables, 7
weather, 572 critical points of, 615–616, 634–641 units of measure of, 8, 35, 85
Marginal costs cross-sectional modeling of, 573–575 Outside function
definition of, 233–234 definition of, 549–550 in composition of functions, 85, 200
examples of, 234–235, 333–334, from equations, 550 definition of, 16
342–343, 515 extrema of, 615–617, 621–623, derivatives of, 200–203
Marginal productivity of money, 651 634–641 in multivariable functions, 600
Marginal profit, 233–234 from graphs, 558–561
Marginal revenue, 233–235 graphs of, 549, 551, 615–616 (see also
Market equilibrium, 415–417
Market price, 409, 411–416
Contour graphs)
input/output diagram for, 550
P
Mass equation, 543 input variables in, 549 Parabola
Mathematical modeling inside functions in, 600 concavity in, 86
definition of, 3, 85 Lagrange multiplier and, 646–653, 662 definition of, 69
models for (see Models) least-squares method for, 657–659 minimum of, 71
Subject Index A-93
Partial derivatives graphs of, 121, 148, 172, 247 cumulative density functions and,
critical points and, 636 inflection points of, 261–262 437–439
definition of, 583–584 instantaneous rate of change in, definition of, 385, 426, 427, 447
from equations, 583–587, 590–591, 120–122 examples of, 427–428
608 integration of, 343–344, 346–348 exponential density, 432–433, 444
notation for, 583 rate of change in, 120–122 graphs of, 427, 439
second (see Second partial second derivative of, 261–262 normal (see Normal distribution)
derivatives) slope graphs of, 171–172, 248, notation for, 426–427
Percent 261–262 uniform, 429–430, 444
definition of, 44 Point of diminishing returns, 261 Producers’ revenue/surplus, 411–417
error formula, 162 Point of tangency, 114, 118, 120, Producers’ willingness/ability to
Percentage change 136–138 receive, 410–414
definition of, 96 Polynomials Product Rule
from equations, 102–106 behavior of, 64, 75 for derivatives, 207–212, 222
examples of, 95–96 cubic (see Cubic functions) for limits, 218
in exponential functions, 41–43, limits of, 64, 69, 218 Profit
86, 96 quadratic (see Quadratic functions) after-tax rate, 484
formula for, 96 quartic (see Quartic functions) at break-even point, 9
from graphs, 98–102 Position, velocity and, 335 continuous income stream and,
of investments, 103–106 Power functions 385–390
limits of, 105 antiderivatives of, 329–331, 370, cost and, 9, 234–235
in linear functions, 31 372–373 definition of, 13
percentage differences and, 45 derivatives of, 155, 179, 184, 222 discrete income stream and,
vs. percentage rate of change, 134 differential equations for, 513 390–391
rise and, 99 integration of, 340–341 examples of, 13–15, 234–235
from tabular data, 97–98 limits of, 218 formula for, 9
units of measure of, 97 slope fields for, 513, 516 marginal, 233–234
Percentage differences Power Rule, Simple revenue and, 9, 234–235
calculation of, 43, 96 for antiderivatives, 329, 332 Proportionality
examples of, 43–45 for derivatives, 179, 184, 194 constant of, 510–512
exponential functions and, 43–44 Present value, 389–390, 392–393, 447 direct, 510
first differences and, 43–44, 80 Price inverse, 510
percentage change and, 45 Cotterill price index, 566 joint, 524
second differences and, 69, 80 demand and, 400 of output variable, 510, 524
Percentage rate of change elasticity of demand and, 407–408 Proportions
definition of, 134, 157 graphs of (see Demand curves; graphs of, 568
of demand, 407–408 Supply curves) from histograms, 424
differential equations for, 521–522 limits of, 402–403, 405–406 from tabular data, 423, 425
from graphs, 134–135 market, 409, 411–416 units of measure of, 211
of investments, 138–140 percentage rate of change of,
vs. percentage change, 134 407–408
of price, 407–408 quantity demanded and, 400,
units of measure of, 134 406–407
Q
Period revenue and, 234–235, 411 Quadratic functions
annual, 462 shutdown, 409, 411–412, 417 accumulated change in, 309
definition of, 452, 455 Probability approximating change in, 229
of sine functions, 453, 455, 457 definition of, 425, 427, 447 concavity of, 69, 72, 86
Periodic functions distributions (see Distributions, vs. cubic functions, 77
cosine (see Cosine functions) probability) derivatives of, 151–153, 155, 184
definition of, 452 examples of, 427–428 differential equations for, 509–510
sine (see Sine functions) histograms, 423–425 vs. exponential functions, 73–74
Piecewise functions of an impossible event, 425 extrema of, 243–244, 250
accumulated change in, 290–292 interpretation of, 425 formula for, 69
average value over, 364 notation for, 426 graphs of, 69–72, 86, 244 (see also
break points in, 343 range of, 425 Parabola)
definition of, 120 of a sure event, 425 inflection point of, 80
derivatives of, 171–172, 248, Probability density functions integration of, 343–346
261–262 accumulation function of, 437–439 leading coefficient in, 71
extrema in, 247–248 area of, 426, 428 limits of, 69
A-94 Subject Index
second differences and, 69, 74 partial (see Partial derivatives) profit and, 9, 234–235
slope fields for, 516, 520 percentage (see Percentage rate of units of measure of, 210
slope graphs of, 244 change) Right rectangles, 290, 292–293
from tabular data, 69–72 in piecewise functions, 120–122 Rise
Quartic functions from tabular data, 97–98, 603–604 change and, 99
approximating change in, 231–232 terminology for, 129 definition of, 26
extrema in, 245–247 units of measure of, 97, 133, 319 Round-off error, 34
graphs of, 246 Ratio, gender, 83 Rule of 72, 162
slope graphs of, 247 Recovering a function, 319 Run
Quotient Rule Rectangles change and, 99
for derivatives, 212 left, 289, 292–293 definition of, 26
for limits, 218 midpoint, 293–294
right, 290, 292–293
Reduction value, 468
Reflection, 454, 456–457
S
R Regression line, 658–659 Saddle points
Random variables, 425 Related rates, 270–274 on contour graphs, 617–621, 623
Range Relative extrema definition of, 616
definition of, 5 on accumulation graphs, 316–318 determinant test for, 637, 641
in graphs and tables, 7 conditions for, 248 graphs of, 616
of probability, 425 on contour graphs, 616–623 in multivariable functions, 616
units of measure of, 8 of cubic functions, 241–243, Saturation level, 60, 526
Rate of change. See also Derivatives 256–259 Scatter plots
accumulated (see Accumulated definition of, 241, 278, 615 definition of, 25
change) derivatives and, 242, 244, 256–258, slope graphs of, 169
aligning data and, 32 634–636 Secant lines
approximating (see Approximating determinant test for, 637–638 average rate of change and, 98–100,
change) endpoints and, 243 157, 365
average (see Average rate of change) graphs of, 241–242, 245, 615–616, definition of, 99
average value of, 362–364 634–635 derivatives and, 169–171
compensating for, 597–605 inflection point and, 256–258 graphs of, 99, 102, 117–118,
continuity and, 129 least-squares method for, 659 136–138
from contour graph, 559–560 in logistic curves, 166 instantaneous rate of change and,
in cross-sectional models, 575–577, in multivariable functions, 615–616, 117–118, 121
608 (see also Partial derivatives) 634–641 limiting position of, 118, 121,
definition of, 27, 96 in piecewise functions, 247–248 136–139, 145, 150–154, 157
description of, 96–98 of quadratic functions, 243–244 point of tangency and, 136–138
in differential equations (see of quartic functions, 245–247 sharp points and, 121
Differential equations) slope at, 242, 634 slope of, 99, 136–139, 145, 149–154,
from equations, 102–106 slope graphs and, 242, 245, 261 157, 365
Euler’s method for, 532–536 steps for finding single variable, tangent lines and, 117–118, 121,
in exponential functions, 41–43 251 136–139, 157, 169–171
formula for, 96 tangent lines at, 242, 634 Second derivative
of future value, 388 Response differential equations, concavity and, 262–263, 278
from graphs, 98–102, 480–482 (see 511, 525 of cubic functions, 256–259
also Slope graphs) Revenue definition of, 257–258
implicit differentiation, 270–274 average rate of change in, 96 in differential equations, 539–542
instantaneous (see Instantaneous at break-even point, 9 extrema and, 261
rate of change) consumer expenditure and, 254 graphs of, 257
interconnecting, 270–273 definition of, 11, 207, 254 inflection point and, 256–263, 278
interpretation of, 96–98, 130–132, elasticity of demand and, 409 input/output diagram for, 258
230 instantaneous rate of change in, of logistic functions, 260–261
in linear functions, 27–29, 31, 41, 86, 210–211 notation for, 258
178, 284–287 marginal, 233–235 of piecewise functions, 261–262
in multivariable functions, 559–560, multiplication of functions for, units of measure of, 258
575–577, 597–605 11–12, 210–211 Second differences, 69, 74, 80
notation for, 129 percentage change in, 95–96 Second-order differential equations,
numerical estimate method for, price and, 234–235, 411 539–542
136–140, 145 producers’, 411–417 Second partial derivatives, 587–591, 637
Subject Index A-95
Second partials matrix magnitude of, 26, 115 Speed, 113, 133, 273
definition of, 589 numerical estimate method for, Square root, in exponential notation, 193
determinant of, 637 136–139, 145, 157 SSE, 658–659
labeling of, 589–590 at points, 114–117 Standard deviation
Sharp points, 114, 121–122, 247–248 rise/run calculation of, 26 definition of, 430, 448
Shutdown point, 409 of secant lines, 99, 136–139, 145, Empirical Rule for, 443
Shutdown price, 409, 411–412, 417 149–154, 157, 365 examples of, 431–432
. See Standard deviation of sine functions, 473–474 formula for, 431
Signed area, 287 SSE and, 659 on graphs, 430
Simple closed contours, 617 steepness and, 115, 117 in normal distribution, 430–431,
Simple Power Rule of tangent lines, 114–121, 129, 434, 443
for antiderivatives, 329, 332 136–139, 145, 157, 169–172, notation for, 430
for derivatives, 179, 184, 194 598–599, 646 of uniform density functions, 444
Sine functions tilt and, 115, 117 Starting value
accumulated change in, 495–496 units of measure of, 319 for accumulation function, 310
amplitude of, 454, 457, 462 from words, 116–117 in linear functions, 25
antiderivatives of, 494–495, 501 Slope fields Steepness
average value of, 454, 456, 462, for constant functions, 515 on contour curves, 553, 560–561
496–497 construction of, 512 definition of, 117
cosine functions, relationship to, for cosine functions, 520 in linear functions, 25–26
453, 474–475, 494 for cubic functions, 520 slope and, 115, 117
derivatives of, 473–479, 500 for exponential functions, 521–522 of tangent lines, 115, 117
differential equations for, 540–542 initial condition on, 512–513 Subtraction, function
end behavior of, 65 for linear functions, 515–516 definition of, 13
expected value of, 454 for logarithmic functions, 514–515, derivatives and, 180–184, 194, 222
extrema of, 462, 487–488 519 difference rules for (see Difference
formula for, 457, 500 for logistic functions, 526–527 Rules)
graphs of, 453–458 for power functions, 513, 516 input variables in, 13, 85
inflection points of, 474, 488–489 for quadratic functions, 516, 520 output variable in, 13, 14, 85
integral of, 494–495, 501 for response differential equations, units of measure of, 14, 85
limits of, 65 525 Sum of squared errors (SSE), 658–659
maximum/minimum of, 487–488 for sine functions, 520 Sum Property of integrals, 344
period of, 453, 455, 457 solution graphs drawn on, 513–515 Sum Rules
reflection of, 456–457 Slope graphs for antiderivatives, 330, 332
shifting of, 454–457, 462–463 of accumulation graphs, 325–326 for derivatives, 180–184, 194
slope fields for, 520 for cosine functions, 474, 478 for limits, 218
slope graphs for, 474 of cubic functions, 242, 257 Supply curves, 409–414
slope of, 473–474 definition of, 164 Supply functions, 409–417
from tabular data, 461–463 drawing of, 165–169 Sure event, 425
from words, 463–465 extrema and, 242, 245, 261 Surveys, bias in, 92
Slope function graphs from, 167–168 Survival and renewal functions, 394–395
algebraic method for, 149–154, 157 inflection point and, 256–257
aligning data and, 32 of logistic functions, 166, 260
average rate of change and, 98–100 overview of, 222
comparison of, 116–117 piecewise, 171–172, 248, 261–262
T
of composite functions, 201 of quadratic functions, 244 Tables
continuity and, 164–165, 222 of quartic functions, 247 accuracy of, 33–35
on contour curves, 601 for sine functions, 474 aligning data in, 32–33
definition of, 25 from tabular data, 170–171 extrapolation vs. interpolation of
at extrema, 242, 634 tangent lines and, 169 data in, 30
first differences and, 29 Specific antiderivatives. See also first differences in, 29
Four-Step Method for, 152–154, 222 Definite integrals Tangent lines
from graphs, 119, 136–138, 169–171 of composite functions, 372 approximating change with,
inflection point and, 165, 256 constant term in, 342 228–232, 598–599
instantaneous rate of change and, definite integrals and, 340–341 at break points, 121
113–117, 119, 121 definition of, 333, 336 concavity and, 118, 120
of linear functions, 25–29, 86, 114, of exponential functions, 334, 372 to constraint curves, 646, 662
178, 228 notation for, 336, 341 to contour curves, 576–577, 584,
of logistic functions, 166 from tabular data, 333–334 601–602, 646
A-96 Subject Index
definition of, 114 in addition of functions, 14, 85 dependent, 5, 270, 273, 278
at extrema, 242, 634 of average rate of change, 97, independent, 5, 270, 273, 278
from graphs, 119, 121, 165 363–364 input (see Input variables)
graphs of, 114–118, 120, 136–138 of average value, 364 Lagrange multiplier, 646–653,
inflection point and, 118, 120 of change, 97 662
instantaneous rate of change and, in combining of functions, 14, 85 in linear system of equations,
114–118, 120, 121, 157 in composition of functions, 16, 85 638
in optimization, 646 for consumers’ expenditure and observed, 5
secant lines and, 117–118, 121, surplus, 406 output (see Output variable)
136–139, 157, 169–171 of definite integrals, 348 random, 425
sharp points and, 114, 121 definition of, 8 Velocity
slope graphs and, 169 of derivatives, 133, 157, 198–199 acceleration and, 334–335
slope of, 114–121, 129, 136–139, 145, in division of functions, 14, 85 definition of, 284
157, 169–172, 598–599, 646 of domain, 8 distance and, 335
steepness of, 115, 117 of input variables, 8, 35, 85 input/output diagram for, 320
Temperature conversion equation, 253 of instantaneous rate of change, 133 position and, 335
Terminal edges, 621–623 in multiplication of functions, 14, units of measure of, 319–320
Third differences, 74–75 85, 211 Vertical asymptote, 48
Topographical maps, 552–554, 560–561 of output variable, 8, 35, 85 Vertical axis intercept
Total cost, 9–11, 334 of percentage change, 97 aligning data and, 32–33
Total social gain, 414–417 of percentage rate of change, 134 in linear functions, 25, 27
Trapezoids, 286, 327 for pressure, 3 Vertical Line Test, 7–8
Trigonometric functions of proportions, 211 Vertical shift
cosine (see Cosine functions) of range, 8 average value and, 462, 496–497
mode for, 453 of rate of change, 97, 133, 319 of sine functions, 454, 457, 462
sine (see Sine functions) recovering a function and, 319 Volume
reporting of, 33–34 of balloons, 270
of revenue, 210 of boxes, 562
of second derivatives, 258 Boyle’s Law for gas, 277
U of slope, 319 Brackett tree equation, 566
Uniform density functions, 429–430, 444 in subtraction of functions, 14, 85 of tin cans, 656
Uniform probability distribution,
429–430
Unit elasticity, 408–409
Units of measure V W
of accumulated change, 286, Variable costs, 9–10 Weather maps, 572
289–290 Variables
of accumulation functions, 319 controlled, 5
Index of Applications
Business and Economics Applications Consumer credit, 38
Accumulated value of an investment, 159, 195, 338, 529 Consumer expenditure and revenue, 254
Advertising effectiveness, 19, 54, 187-188, 204, 236-237, Consumer Price Index (CPI) for college tuition, 157, 197
240, 264, 324, 354, 654 Consumer Price Index (CPI) for different countries, 239
Advertising with fliers, 268 Consumer Price Index (CPI) for refuse collection, 110
After-tax profit rate on investment, 484, 492, 499 Continuously compounded interest, 105-106, 111, 112,
Airline company expansion, 388-389 138, 195, 307, 390-392, 521-522, 582
Airline profit and ticket price, 34, 81, 109, 140, 366-367 Corn production, 140, 213, 628-629
Airport shop customers, 213 Corporation profit, 36-37, 138, 379, 385-386, 390
APR (annual percentage rate), 103, 106 Cotterill model for supermarket performance, 566
APY (annual percentage yield), 103, 106, 138-139 Cotton exports value, 20
ATM customers served daily, 442 Credit card debt, 23
ATM transaction fees, 110, 185 Credit card finance charge, 111, 191
Baggage complaints against airlines, 243-244, 250 Crude oil production, 263-264
Balance in an account, 33, 103-104, 111, 144, 238, Cruise ticket sales/revenue, 109, 656-657
307-308, 521-522, 582-583 Customer assets at Charles Schwab & Co., 53
Balance in mutual fund, 140 Customer waiting times at department store, 440
Ball bearing costs, 660-661 Dairy company costs, 10-15, 197, 206-207
Banana prices, 627 Debit card transactions, 23
Bankruptcy filings, 37 Dell Computer Corporation employees, 160
Beef supply and demand, 88, 421 Demand for air travel between U.S. and Europe, 567
Bottled water consumption, 55 Demand for ceiling fans, 419, 420
Butter price increase, 229 Demand for exercise facilities, 254-255
Buyout of a company, 397-398 Demand for kerosene lanterns, 420
Cable subscribers, 57, 205, 266 Demand for minivans, 406-407, 408
Cable subscribers with Internet access, 196 Demand for sparkling water, 420
Cable television expenditure per person, 267 Demand for wooden chairs, 419, 420
Cable television revenue, 245-247 Demand curves/function, 384, 400-401, 417, 542, 562
Campbell Soup Company sales, 471, 484, 493-494, 502 Department store customer count, 287-290
Capital value of a company, 398-399 Department store revenue, 95
Car dealership advertising, 236-237 Diamond pricing, 592
Car sales in Quebec, 594 DuPont stock value, 303
Car value depreciation, 37 Earnings of men and women in U.S., 376
Carbonated beverage packaging cost, 667 Egg production, 209
Cargo unloaded during slowdown, 39 Exports from/imports into United States, 356
Cattle prices, 175 Fabric sales profit, 562
CD production cost, 354 Factory sales of electronics in U.S., 144, 155, 307, 366
CD sales, 20, 58, 210, 214 Fast food restaurant customers, 142
Cellular phone average monthly bill, 125, 173 Federal Express revenue, 356-357
Cellular phone subscribers, 125 Federal income tax and dependents, 542
Cellular phone supply, 413-414 Fee refund schedules, 161
Cheese sales revenue, 568, 583-586, 644 Fish production, 89
Civil-transport aircraft production, 470-471 Flow rate of profits into investments, 385, 386-387
Cobb-Douglas production function, 276, 597, 648, 654 Fork lift production costs, 83-84
Cola sales, 397, 548, 580, 595, 607 Frozen yogurt sales, 441
Compaq Computer Corp. executive compensation, 23 Future value of an investment, 388-389
Competitive and complementary products, 612 Future value of corporation profits, 387, 391-392
Compound interest, 103-104, 112, 195, 307, 391-393, Gasoline fill-up cost, 515
550, 586-590, 593, 596-597, 607 Gasoline prices, 2-3, 23
Computer product sales growth, 511-512, 514-515, Gasoline supply and demand, 401-404, 409-410, 411-413,
534-535 416-417
Construction company worker hours, 267-268, 308, German bond rates, 49-50, 51-52, 337-338
537, 538 Gold investment, 203
Golf ball production costs, 239 New Zealand bond rates, 56
Grapefruit production costs, 302 Newspaper circulation, 24-26, 240, 368
Graphing calculator supply and demand, 422-423 Nominal interest rate, 103, 111
Gross national product (GNP), 522-524 Northwest Airlines revenue, 160
Handcrafted bench profit, 187 Number of skiers and lift ticket price, 592
Health club advertising, 654 Oil production rates, 303, 307, 367, 537, 538
Heating oil for apartment complex, 38-39 Pencil production profit, 302
Heating oil sales, 487-489 Personnel training for new plant, 547
Hot dog sales and price, 238-239 Petroleum product imports, 53
Houghton Mifflin employees, 127 Pizza sales revenue, 239-240, 642
House painting income, 216-217 Pollution control spending, 228
House’s market value, 38 Postal rates, 39, 515, 543, 618-619, 634-636, 637-638
Ice cream sales, 470, 485 Production costs, 188, 206, 215, 238, 254
iGo Corp. stock offering, 205 Purchasing power of U.S. dollar, 216, 539-540
Income tax and dependents, 562 Resale value of Jeep Grand Cherokee, 89
Indexed price of wheat in United Kingdom, 490-491 Revenue flow rate, 353
International Game Technology revenue, 37 Revenue from bull and cow sales, 592
Internet access through cable TV, 196 Revenue/profit in Canadian dollars, 203, 204
Internet company employees, 339 Revenue investment, 396
Internet service subscribers, 321-322 Revenue of General Electric Company, 396
Investing for retirement, 104-105, 396-397, 449 Revenue of General Motors Corporation, 396
Investment doubling time, 45, 55, 111, 141, 162 Revenue of Polo Ralph Lauren Corporation, 227, 268-269
Investment growth, 104-106, 112, 138-139, 195, 237, Revenue of Sara Lee Corporation, 396
307-308, 338, 376, 392-393, 449, 550, 582-583, Revenue of Sears, Roebuck, and Company, 397
586-590, 593, 596-597 Revenue optimization, 254, 255, 642
Investment of company’s profits, 390-392, 396 Revenue rates of change, 355, 356-357, 358
Investment percentage yields, 112, 582 Roofing company jobs, 69-70, 542
Investments with 401 (k) plans, 657-659 Rose market supply and/or demand, 254, 384, 422
iPod revenue, 197 Sauerbeck index of wholesale prices, 491, 502
iPod sales, 19, 42-43, 196 Saving for college tuition, 396
Land value, 8-9 Saving to buy a car, 392-393
Lawn mower sales, 502 Saving to buy a house, 395-396, 397
Lettuce prices, 626 Science fiction novel supply and/or demand, 384, 422
Life insurance premiums, 236 Sculptures supply and demand, 421-422
Lift ticket sales and snow depth, 562 Seafood restaurant buffet prices, 234-235
Loan amount, 110, 554-555, 558-560, 562, 597 Seasonal sales, 498-499, 503
Long distance telephone rates, 366 Shipping carton of minimal cost, 655, 656
Marble fountain supply and demand, 449 Small business taxes/sales, 29-30, 31-33, 134-135
Marginal analysis, 233-235, 236, 238-239, 333-334, Social gain, 414-416, 419, 421-423
342-343, 515 Startup business investments, 386-387
Mattress production, 276, 597-600, 648-650, 651, 654 Stock market “October Madness,” 107-108
Mechanical component assembly time, 267 Stock value, 212-213, 302, 321
Median new-house size, 266 Student loan payment, 111
Median sale price of one-family homes, 20 Superbowl T-shirt and hat sales, 664
Minimum wage, 21-22, 467, 469, 498 Supply curve for a commodity, 409, 417-419
Monthly payment on a loan, 110, 111, 554-555, 558-560, Supply of answering machines, 421
562, 565, 579, 597, 607 Supply of artist’s prints, 421
Mortgage debt in U.S., 224 Supply of CDs, 421
Mountain bike sales profit, 144, 204 Supply of saddles, 420-421
Movie video sales, 307 Survival of U.S. business companies, 30-31
Mulch sales profit, 642 Sweatshirt costs and break-even point, 39-40
Natural gas imports and exports, 22-23 Swimming pool and ski gear sales, 498-499
Natural gas prices, 77-79, 266 Ticketmaster profits, 397
Natural gas production rate, 354 Tire production, 324
Net income of Lowe’s Companies, Inc., 397 Tire sales, 349-351
Net sales of international company, 33 Tissue paper sales, 214
New business profit, 322 Toaster oven production costs, 234, 333-334, 342-343
New car dealership revenue, 187-188 Transistor radio manufacturing, 655, 656
T-shirt production costs, 606 Luggage weight on passenger flights, 442
T-shirt sales/profit, 20, 38, 92-93, 141, 236, 282, 606 Mail processing at post office, 203, 238
U.S. Postal Service revenue, 356-357 Milk consumption, 23
United Parcel Service revenue, 356-357 Milk cow farms’ decline, 54-55
Used car financing, 21 National Rifle Association membership, 116-117
Used car sales, 27-28, 175-176 Northwest Airline’s fuel consumption, 155-156
Vehicle maintenance costs, 183-184, 562 Northwest Airline passengers, 107
Violin production, 199-200 Overseas international tourists, 211-212
Window pane production, Passengers at Atlanta airport, 143
Women’s clothing sales, 486-487 Peach consumption, 57, 579-580, 594-595, 607-608
Pectin and pigment extraction from sunflowers, 568-569,
General Applications 606, 644
Accidents in British coal mines, 446 Peptides released in crayfish processing, 643
Adhesiveness of cheese spread, 590-591 Popcorn tin construction, 656
Adhesiveness of honey, 570-571, 596, 614, 645, 654, Salmon catch cycles on Restigouche River, 473, 489-490,
655 540-542
Agriculture research spending, 216 Salt leakage, 277
Aircraft accidents, 367 Sausage cooking loss, 567-568, 644, 651-653, 654
Amusement park visitors, 67 Scuba diving times, 19
Applesauce consistency, 610, 631 Snowball size, 277
Athletic organization donors, 266-267, 339 Student’s grade-point average and SAT score, 592
Banana chip crispness, 663-664 Student’s test grade and study time, 130, 131-132, 142-143,
Baseball photography, 274 187, 238, 261, 265
Basketball free-throw points, 213 Studying point of diminishing returns, 131-132, 261
Board game learning time, 441-442 Superhighway in Europe, 225
Calculus class exam scores, 443 Swimmer’s times, 144, 155, 160, 254, 368
Campus bus waiting time, 429-430, 431-432, 443-444 Technical support call, 440
Cereal in a box, 440 Ticket price and attendance, 255
Cheese spread consistency, 622-624, 645 Tourists visiting Tahiti, 279
Coal price for synthetic-fuel plants, 150-151, 152-153 Traffic light waiting time, 441
Cohesiveness of honey, 614, 653-654, 655, 656 Tree trunk wood volume, 276
Computer terminal waiting time, 435-436 Turkey consumption, 224
Computer worm in network, 59-60, 526-527 Typing speed, 140
Corral constructed from fencing, 656 Vegetable consumption/utilization, 82, 367
Cottage cheese consumption in U.S., 305 Volume index of a cake, 632-633, 638-639
Countries issuing postage stamps, 66, 531-532, 537, 538 Volume of rectangular packing container, 562
Cropland increase worldwide, 517 Wheat crop carrying capacity, 276, 568
Distance between cars, 440 Willie Mays’ stolen bases, 67-68
Fish caught per rod per day, 473, 489-490, 540-542 Wine consumption in U.S., 296-300
Foreign travelers to U.S. from Europe, 211-212 Yeast dough expansion, 196
Frozen yogurt cone, 277
Garbage taken to landfill, 255, 268 Life Science Applications
Gateway Arch in St. Louis, 450 Absorption rate of drug, 290-292
Greenhouse gas emissions, 1, 40 AIDS cases, 110, 174, 543
Grocery checkout waiting time, 441 Algae growth, 302
Homes lacking complete kitchen facilities, 660 Animal experiments in England, 661
Households with cable TV, 205 Apparent temperature, 276, 562-563, 581-582, 606
Households with computers, 212 Atmospheric carbon dioxide, 198-199, 200, 301, 451, 477-478,
Households with TVs, 215 479-480, 491-492, 493
Households with TVs and VCRs, 197-198 Baby weight from birth to 4 weeks, 20
Households with VCRs, 202-203, 215 Bacterial growth/decline, 40-41, 53, 60-61
Households with washing machines, 324 Bacterial growth in milk, 301
Hunting license fees, 281 Beef cow intake of organic matter, 596, 640-641
Intel processor computing power, 54 Beetle population in Netherlands, 468, 469, 492
International telephone calls, 352 Berkley Growth Study, 378
Internet users in China, 107 Blood-alcohol level and accident risk, 545-546
Licensed drivers and age, 156-157 Blood pressure and time of day, 308-368
Light bulb lifespan, 435 Body heat loss due to wind, 555-558, 595
Continued on back inside cover
Life Science Applications (continued) Plant height distribution, 440
Body temperature after taking Tylenol, Pneumonia death rate in U.S., 485, 493
Body-mass index (BMI), 275, 566-567, 602-603, 607 Polio epidemic of 1949, 89, 260-261, 527-528
Boll weevil parasitoids, 569-570, 642 Puppy weight gain, 518-519, 537
Cancer death rates, 143, 176 Radon gas health hazard, 531
Carbon emissions absorbed by trees, 357-358 Regional analgesia during childbirth, 214
Carbon monoxide emissions, 368-369 Silo capacity, 565
Cesarean delivery rate, 84 Siple and Passel heat loss equation, 555-558, 595
Crickets chirping, 178 Sooty tern population, 399
Death rates in United States, 82, 143 Sperm whale population, 393-394
Density-dependent mortality, 357 Stem volume for Douglas firs in British Columbia, 566
Dinosaur height distributions, 446 Stem volume for red pine trees in Canada, 606-607
Drug concentration, 56-57, 132-133, 247-248, 290-292, Survival rate of flour beetle, 126
319, 324, 353 Technetium-99 in bloodstream, 531
Drug elimination from body, 18-19, 45-46, 55 Tree absorption of atmospheric carbon, 357-358
Elephant population in West Africa, 399 Tree girth and water transpiration, 275
Emergency room arrivals, 433, 441 Tree growth, 518, 519, 529
Endangered bird protection, 323 Weight and height of a person, 591
Epidemic of viral infection, 355 Weight loss and dieting, 119, 141, 375
Fatty acid extraction from milk fat, 642-643 Weight of laboratory mouse, 56, 196, 307, 338, 352
Flea population growth, 395
Fox population, 449 Physical Science Applications
Grasshopper eggs hatching temperature, 126-127, 253, 266 Acceleration due to gravity, 338, 539
Heart beats/rate, 358-360 Acceleration of a vehicle, 301, 320, 356, 377
Height and age of a young child, 542 Acceleration of a race car, 353, 367
Heights of college men, 62-63 Air pressure and altitude, 47-49, 515, 517
Influenza epidemic of 1918, 66-67, 205 Air temperature, 37, 113-114, 115-116
Insect egg production, 610-611, 642 Airplane altitude and air temperature, 15-16
Kidney transplants in the U.S., 23 Airplane distance, 140
Lake contamination, 17-18 Airplane acceleration/speed, 236, 352, 368
Lead concentration in soil, 56 Apparent temperature, 562-563, 577-578, 581-582
Lead paint usage, 82 Average temperature in New York, 499
Life expectancy for black males, 109 Balloon volume and radius, 270, 277
Life expectancy for women, 306 Boyle’s Law, 277
Life expectancy in world, 236 Carbon-14 dating, 55, 365-366, 382-383, 384
Liver transplants in the U.S., 23 Chemical plume in groundwater, 272-273
Livestock supported by rye grass, 595-596 Chemical reaction activity, 67, 269
Lizards gathered for pets, 473, 499 Chlorofluorocarbons (CFCs) in atmosphere, 237
Low birth weight babies, 70-71, 306 Cloud cover frequency over Minneapolis, 578-579
Male growth rate, 378 Crude oil flow rates, 283, 375
Metabolic rate of male, 187 Daily high temperatures in Boston, 469-470, 484-485
Milk storage and temperature, 55, 196, 661 Daily high temperatures in New Orleans, 469-470
Moth population, 357 Daily low temperatures in Omaha, Nebraska, 468, 469
Multiple birth rates, 111 Daily mean temperature by longitude and latitude, 592
Muskrat population in Iowa, 399 Daily mean temperature in Fairbanks, Alaska, 470, 484
Neuron firing patterns in response to odors, 472-473 Daily mean temperatures in Omaha, Nebraska, 467-468, 469,
Nitrogen oxide emissions, 54, 269 480-482, 490
Northern fur seal population, 399 Darkness hours in Anchorage, Alaska, 20, 472
Organ donors in the U.S., 37 Daylight hours at the Equator, 472
Pain reliever remaining in body, 531 Daylight hours at Tropic of Capricorn, 472
Patients admitted to and discharged from a hospital, 346-348 Daylight hours for Northern and Southern Hemispheres,
Patients recovering from mild dehydration, 427-428 563-565
Patient’s temperature during illness, 470 Daylight hours in a town or city, 472-491
Pea seedling growth rate, 125-126 Daylight hours in Anchorage, Alaska, 491
Physical therapy with heat, 633 Daylight hours in Honolulu, Hawaii, 500
Piglet’s average weight at fixed temperature, 664 Daylight hours on Arctic Circle, 463-465, 466
Pig’s average daily weight gain/loss, 581, 629-630 Depth of water in a reservoir, 529
Plant growth rate, 168, 321 Dew point, 577-578, 581-582
Distance of tractor from a farm, 156 Temperature during and after a thunderstorm, 231-232,
Diver’s velocity, 338 305-306, 353, 367
Draining water tank, 286-287 Temperature in a museum, 353
Electricity production and temperature, 471-472 Temperature in the south Gobi desert, 483
Elevation of Missouri farmland, 550-553, 560-561, 573-577, Temperature of polar night region, 133-134
626, 631-632, 644 Temperature over 24 hours in a southwestern city, 438-439
Expanding gas, 304 Temperatures in a Midwestern city, 97-99, 102-103, 186
Extraterrestrial radiation, 484, 485, 490, 543-544, 628, 630 Temperatures in Philadelphia, 475-476, 479, 496
Falling object, 156, 335-336, 338, 517-518 Temperatures during a September day, 361-362
Flooding in Missouri, 633 Tides in Savannah River entrance, 485-486, 495
Flow rate of a river, 253, 309, 310-313, 325-326, 343-344 Ultraviolet radiation by latitude and month, 593-594
Fuel consumption by vehicles in U.S., 174, 338 Ultraviolet radiation in Auckland, New Zealand, 461-463,
Fuel efficiency, 141 465-466, 492, 493, 496-497
Gas pressure and volume, 277 Uranium decay rate, 384
Global radiation by month and latitude, 627 Velocity and distance traveled, 284-286
Heat Index, 276, 562-563, 606 Velocity of a crack, 368
Hooke’s Law, 543 Velocity of a vehicle, 320-321, 508
Hot-air balloon height and distance, 277 Water pressure, 3-4, 6
Humidity level, 236, 562-563 Weather satellite orbit, 482
Hurricane speed, 375 Wind turbine use, 482-483, 485, 493, 572-573
Ice rate of thickening, 529 Work required to propel a rocket, 384-385
Isotherms and precipitation, 572
Kite’s rate of movement, 277 Social and Behavioral Sciences Applications
Ladder speed while sliding down wall, 277-278 ACT college test average for females, 107
Lake Tahoe levels, 108, 253, 267, 301, 504-505 Adults 20-24 years of age not yet married, 660
Land subsidence in Santa Clara Valley, 633 Birth rate in African country, 330-331, 333
Leaking gasoline tank, 203 Births to unmarried women, 38, 141
Mean monthly temperature and pressure at Halley Bay, 611 Births to women 35 years and older, 23-24
Missile misfiring, 81 Births to women 45 years and older, 83, 146
Newton’s Law of Cooling, 537-538 Campus organization membership, 174-175
Newton’s Second Law, 543 Children living with grandparents, 88-89, 204-205
Ozone levels, 663 Club membership, 53
Power output of diesel engine with wind turbine, 482-483, Coal mine accidents, 446
485, 493 College enrollment based on tuition and financial aid,
Precipitation in New York City in early 1855, 661 603-605, 607
Precipitation in U.S. corn belt, 628 College tuition, 205-206
Pulses from neutron star, 492, 500 Compulsory school laws, 90-91
Radioactive decay, 506, 531 Elementary education funding/cost, 83, 213
Radioactive radon gas decay, 55, 531 Employment in Slovakia, 127-128
Raindrop speed and size, 142 Energy consumption by university, 304
Rainfall in Florida during thunderstorm, 322-323 Energy production and consumption in U.S., 516-517
Rainfall in New Orleans, 440 Farmers who own a tractor, 529
Road test of Toyota and/or Porsche, 356, 377 Fechner’s Law of stimulus and response, 511, 525
Robot speed, 304 Fertility rates, 226
Rowan-Robinson universe expansion model, 542 Garbage trucks needed for city’s population, 204
Sea level changes, 345-346 Gender ratio, 83, 84, 279, 338, 492, 499-500
Skid marks, 280, 613 High school enrollment and dropouts in S.C., 216
Snowball size, 277 High school graduates in South Carolina, 258-259
Snow cover in January in Northern Hemisphere, 665-666 Hospital stay length, 305
Snow pack depth in mountains, 529 Hutterite Brethren children, 565
Snow pack in Northwest Territories, 304, 353 Infants born to five generations of a family, 662
Solution pH, 57 Irish potato famine, 546
Speed of a roller coaster, 159 Jail capacity in a southwestern state, 175
Speed of a vehicle, 113, 122, 284-286, 356, 366, 377, Learning rate for new task, 538
507-508, 613 Marriage age of men, 109
Stopping distance of a car, 301 Marriage age of women, 82, 108-109, 542-543
Sun declination, 126, 486, 493, 499 Mass transit trips, 470, 483, 489
Temperature affecting utility company, 471-472 Mayoral election, 213
Population of South Carolina, 237, 363-364
Military rank and pay, 19 Population of United States, 73-74, 127, 214-215, 509-510,
Municipal solid waste recycling, 54 532-534, 542
Nielsen Media Research ratings, 619-620 Population of U.S. 15 to 50 years old, 580-581
Patents for plow sulkies, 532 Population of U.S. 65 to 74 years old, 253, 265-266
Political organization members, 141 Population of U.S. 65 years or older, 163, 186
Pollution control spending, 228 Population of U.S. distributed by age, 444-445
Population change due to a civil war, 355-356 Poverty in the United States, 215
Population decline of small town, 43-44 PTA-affiliated states, 66, 159, 205
Population growth of a city, 134 Public secondary school enrollment, 141
Population of a country, 301 Residential gas use in Reno, Nevada, 467, 469
Population of American Indians, Eskimos, and Aleuts, 107 SAT score realignment, 443
Population of Aurora, Nevada, 196, 369 SAT scores, 423-424, 425-426, 430-431, 442
Population of baby boomers, 306 SAT scores and grade-point average, 592
Population of California, 232-233 Sheriff ’s office calls received, 175, 206, 476
Population of Cleveland, Ohio, 72, 170-171 Social Security assets, 100-102
Population of a city in the Northeast, 204 Social Security cost-of-living adjustments, 21
Population of French Polynesia, 279 Social Security trust fund, 306
Population of Hawaii, 186 Social Security workers per beneficiary, 54
Population of Indiana, 120-121 Texas household income distribution, 445-446
Population of Iowa, 174 Tobacco-ban bill in Senate, 562, 592
Population of Ireland, 546 Unemployment rates, 142
Population of Kentucky, 241-243, 249, 256-258 University enrollment, 39, 207-208
Population of Mexico, 110, 237-238, 367 Verhulst population model, 529
Population of North Dakota, 305 World population growth, 38, 68, 661-662
You might also like
- Introduction - Calculus and Vectors PortfolioNo ratings yetIntroduction - Calculus and Vectors Portfolio55 pages
- (Ebook PDF) Functions and Change: A Modeling Approach To College Algebra 6th Edition Instant Download100% (2)(Ebook PDF) Functions and Change: A Modeling Approach To College Algebra 6th Edition Instant Download55 pages
- (Textbooks in Mathematics) Marius Ghergu - Differential Calculus in Several Variables (Textbooks in Mathematics) - Chapman and Hall - CRC (2024)100% (1)(Textbooks in Mathematics) Marius Ghergu - Differential Calculus in Several Variables (Textbooks in Mathematics) - Chapman and Hall - CRC (2024)325 pages
- (Ebook PDF) Functions Modeling Change: A Preparation For Calculus, 5th Edition Instant Download100% (2)(Ebook PDF) Functions Modeling Change: A Preparation For Calculus, 5th Edition Instant Download54 pages
- Transition To Advanced Mathematics Textbooks in Mathematics100% (7)Transition To Advanced Mathematics Textbooks in Mathematics552 pages
- Pre-Calculus, Calculus, and Beyond: Hung-Hsi Wu100% (5)Pre-Calculus, Calculus, and Beyond: Hung-Hsi Wu459 pages
- (Ebook PDF) Functions and Change: A Modeling Approach To College Algebra 6th Edition Download100% (1)(Ebook PDF) Functions and Change: A Modeling Approach To College Algebra 6th Edition Download46 pages
- (Ebook PDF) Functions and Change: A Modeling Approach To College Algebra 6th Editionpdf Download100% (3)(Ebook PDF) Functions and Change: A Modeling Approach To College Algebra 6th Editionpdf Download54 pages
- Linear Algebra, 4th Edition (2009) Lipschutz-Lipson-1 PDF100% (4)Linear Algebra, 4th Edition (2009) Lipschutz-Lipson-1 PDF432 pages
- Brief Calculus With Integrated Precalculus by Joseph CollisonNo ratings yetBrief Calculus With Integrated Precalculus by Joseph Collison352 pages
- Linear Algebra, 4th Edition (2009) Lipschutz-Lipson100% (2)Linear Algebra, 4th Edition (2009) Lipschutz-Lipson432 pages
- The Six Pillars of Calculus Lorenzo Sadun Business EditionNo ratings yetThe Six Pillars of Calculus Lorenzo Sadun Business Edition379 pages
- Michael Akritas Probability Statistics With R PDF100% (1)Michael Akritas Probability Statistics With R PDF526 pages
- Calculus 2nd Edition - Single Variable Solutions94% (18)Calculus 2nd Edition - Single Variable Solutions816 pages
- Mehdi Rahmani-Andebili - Precalculus - Practice Problems, Methods, and Solutions (2021)100% (1)Mehdi Rahmani-Andebili - Precalculus - Practice Problems, Methods, and Solutions (2021)119 pages
- (Ebook PDF) College Algebra and Trigonometry 4th Edition by J. S. Ratti Instant Download100% (2)(Ebook PDF) College Algebra and Trigonometry 4th Edition by J. S. Ratti Instant Download53 pages
- (Ebook PDF) Functions and Change: A Modeling Approach To College Algebra 6th Edition PDF DownloadNo ratings yet(Ebook PDF) Functions and Change: A Modeling Approach To College Algebra 6th Edition PDF Download51 pages
- Calculus For The Life Sciences - A Modeling Approach-Libro2011No ratings yetCalculus For The Life Sciences - A Modeling Approach-Libro20111,105 pages
- Modern Mathematical Statistics With Applications (2nd Edition)13% (32)Modern Mathematical Statistics With Applications (2nd Edition)13 pages
- Calculus Concepts and Applications Foerster100% (8)Calculus Concepts and Applications Foerster731 pages
- 2024 Term 2 Content Workshop Manual (Calculus) - 073228No ratings yet2024 Term 2 Content Workshop Manual (Calculus) - 0732288 pages
- 2024 Term 2 Enrichment Classes Manual (Calculus) - 044423No ratings yet2024 Term 2 Enrichment Classes Manual (Calculus) - 0444238 pages
- Calculus - Concepts and Applications - Foerster83% (58)Calculus - Concepts and Applications - Foerster731 pages
- A Course in Mathematical Statistics 0125993153100% (10)A Course in Mathematical Statistics 0125993153593 pages
- O Collection Problem/: Yas - Bu G R O L/ S.M. Nikolsky100% (1)O Collection Problem/: Yas - Bu G R O L/ S.M. Nikolsky203 pages
- (Undergraduate Texts in Mathematics) Murray H. Protter, Charles B. Jr. Morrey - Intermediate Calculus (1986, Springer)75% (4)(Undergraduate Texts in Mathematics) Murray H. Protter, Charles B. Jr. Morrey - Intermediate Calculus (1986, Springer)664 pages
- A Primer of Real Functions Carus Mathematical Monographs100% (1)A Primer of Real Functions Carus Mathematical Monographs322 pages
- Advanced Functions and Introductory Calculus100% (20)Advanced Functions and Introductory Calculus498 pages
- Modelling Change and Rates of Change 1: 38 Chapter Review 39 115 Chapter Review 116No ratings yetModelling Change and Rates of Change 1: 38 Chapter Review 39 115 Chapter Review 1163 pages
- Marek Capinski, Tomasz Jerzy Zastawniak - Probability Through Problems (2000, Springer) PDFNo ratings yetMarek Capinski, Tomasz Jerzy Zastawniak - Probability Through Problems (2000, Springer) PDF133 pages
- The AP Calculus Problem Book: Chuck Garner, PH.DNo ratings yetThe AP Calculus Problem Book: Chuck Garner, PH.D216 pages
- D'Angelo, West - Mathematical Thinking Problem Solving and Proof (2nd Ed, 2000)100% (2)D'Angelo, West - Mathematical Thinking Problem Solving and Proof (2nd Ed, 2000)433 pages
- Elementary Mathematical Models Order Aplenty and A Glimpse of Chaos (Dan Kalman)No ratings yetElementary Mathematical Models Order Aplenty and A Glimpse of Chaos (Dan Kalman)362 pages
- M1-R5: Information Technology Tools and Network Basics SET - 01No ratings yetM1-R5: Information Technology Tools and Network Basics SET - 0124 pages
- Bslbatt 15kw Slimline Victron Setup Guide v1.3No ratings yetBslbatt 15kw Slimline Victron Setup Guide v1.323 pages
- Constraints of Effective Electronic Records Managment System in The OrganizaitonsNo ratings yetConstraints of Effective Electronic Records Managment System in The Organizaitons47 pages
- Lecture 1: Introduction To Scilab: Kebaier@No ratings yetLecture 1: Introduction To Scilab: Kebaier@55 pages
- Office Building, Nagar Parishad Pachore: First FloorNo ratings yetOffice Building, Nagar Parishad Pachore: First Floor1 page
- Honeywell Sensing Micropressure Board Mount Pressure MPR Series Datasheet 32332628 G en PDFNo ratings yetHoneywell Sensing Micropressure Board Mount Pressure MPR Series Datasheet 32332628 G en PDF25 pages
- Categorical Reparameterization With Gumbel SoftmaxNo ratings yetCategorical Reparameterization With Gumbel Softmax13 pages
- Cisco 7600 Series Ethernet Services Plus 20 and 40-Gbps Line Cards Data SheetNo ratings yetCisco 7600 Series Ethernet Services Plus 20 and 40-Gbps Line Cards Data Sheet10 pages
