Calculus Version 2

APEX
Version 2.0
Authors
Gregory Hartman, Ph.D.
Department of Applied Mathema cs
Virginia Military Ins tute
Brian Heinold, Ph.D.

Department of Mathema cs and Computer Science
Mount Saint Marys University
Troy Siemers, Ph.D.

Dimplekumar Chalishajar, Ph.D.

Editor
Jennifer Bowen, Ph.D.
Department of Mathema cs and Computer Science
The College of Wooster
Copyright 2014 Gregory Hartman

Licensed to the public under Crea ve Commons
A ribu on-Noncommercial 3.0 United States License
P
A Note on Using this Text
Thank you for reading this short preface. Allow us to share a few key points
about the text so that you may be er understand what you will nd beyond this
page.
This text comprises a threetext series on Calculus. The rst part covers material taught in many Calc 1 courses: limits, deriva ves, and the basics of integra on, found in Chapters 1 through 6.1. The second text covers material o en
taught in Calc 2: integra on and its applica ons, along with an introduc on
to sequences, series and Taylor Polynomials, found in Chapters 5 through 8. The
third text covers topics common in Calc 3 or mul variable calc: parametric
equa ons, polar coordinates, vectorvalued func ons, and func ons of more
than one variable, found in Chapters 9 through 13. All three are available separately for free at www.vmi.edu/APEX.
Prin ng the en re text as one volume makes for a large, heavy, cumbersome
book. One can certainly only print the pages they currently need, but some prefer to have a nice, bound copy of the text. Therefore this text has been split into
these three manageable parts, each of which can be purchased for about $10 at
Amazon.com.
APEX Aordable Print and Electronic teXts

APEX is a consor um of authors who collaborate to produce highquality,
lowcost textbooks. The current textbookwri ng paradigm is facing a potenal revolu on as desktop publishing and electronic formats increase in popularity. However, wri ng a good textbook is no easy task, as the me requirements
alone are substan al. It takes countless hours of work to produce text, write
examples and exercises, edit and publish. Through collabora on, however, the
cost to any individual can be lessened, allowing us to create texts that we freely
distribute electronically and sell in printed form for an incredibly low cost. Having said that, nothing is en rely free; someone always bears some cost. This
text cost the authors of this book their me, and that was not enough. APEX
Calculus would not exist had not the Virginia Military Ins tute, through a generous JacksonHope grant, given one of the authors signicant me away from
teaching so he could focus on this text.
Each text is available as a free .pdf, protected by a Crea ve Commons Attribu on - Noncommercial 3.0 copyright. That means you can give the .pdf to
anyone you like, print it in any form you like, and even edit the original content
and redistribute it. If you do the la er, you must clearly reference this work and
you cannot sell your edited work for money.
We encourage others to adapt this work to t their own needs. One might
add sec ons that are missing or remove sec ons that your students wont
need. The source les can be found at github.com/APEXCalculus.
You can learn more at www.vmi.edu/APEX.
Contents
Preface
iii
Table of Contents
1 Limits
1.1 An Introduc on To Limits . . . . .
1.2 Epsilon-Delta Deni on of a Limit
1.3 Finding Limits Analy cally . . . . .
1.4 One Sided Limits . . . . . . . . .
1.5 Con nuity . . . . . . . . . . . . .
1.6 Limits involving innity . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2 Deriva ves
2.1 Instantaneous Rates of Change: The Deriva ve
2.2 Interpreta ons of the Deriva ve . . . . . . . .
2.3 Basic Dieren a on Rules . . . . . . . . . . .
2.4 The Product and Quo ent Rules . . . . . . . .
2.5 The Chain Rule . . . . . . . . . . . . . . . . .
2.6 Implicit Dieren a on . . . . . . . . . . . . .
2.7 Deriva ves of Inverse Func ons . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
55
. 55
. 69
. 76
. 83
. 93
. 103
. 114
3 The Graphical Behavior of Func ons

3.1 Extreme Values . . . . . . . . . . .
3.2 The Mean Value Theorem . . . . . .
3.3 Increasing and Decreasing Func ons
3.4 Concavity and the Second Deriva ve
3.5 Curve Sketching . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
9
16
27
34
43
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
121
121
129
134
142
149
4 Applica ons of the Deriva ve

157
4.1 Newtons Method . . . . . . . . . . . . . . . . . . . . . . . . . 157
4.2 Related Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
4.3
4.4
Op miza on . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
Dieren als . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
5 Integra on
5.1 An deriva ves and Indenite Integra on
5.2 The Denite Integral . . . . . . . . . . .
5.3 Riemann Sums . . . . . . . . . . . . . .
5.4 The Fundamental Theorem of Calculus . .
5.5 Numerical Integra on . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
185
185
194
204
221
233
6 Techniques of An dieren a on
6.1 Subs tu on . . . . . . . . . .
6.2 Integra on by Parts . . . . . .
6.3 Trigonometric Integrals . . . .
6.4 Trigonometric Subs tu on . .
6.5 Par al Frac on Decomposi on
6.6 Hyperbolic Func ons . . . . .
6.7 LHpitals Rule . . . . . . . .
6.8 Improper Integra on . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
247
247
266
276
286
295
303
313
321
7 Applica ons of Integra on

7.1 Area Between Curves . . . . . . . . . . . . . . . . . . . .
7.2 Volume by Cross-Sec onal Area; Disk and Washer Methods
7.3 The Shell Method . . . . . . . . . . . . . . . . . . . . . .
7.4 Arc Length and Surface Area . . . . . . . . . . . . . . . .
7.5 Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.6 Fluid Forces . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
333
334
341
348
356
365
375
8 Sequences and Series

8.1 Sequences . . . . . . . . . . . . . . . . . . .
8.2 Innite Series . . . . . . . . . . . . . . . . .
8.3 Integral and Comparison Tests . . . . . . . .
8.4 Ra o and Root Tests . . . . . . . . . . . . . .
8.5 Alterna ng Series and Absolute Convergence
8.6 Power Series . . . . . . . . . . . . . . . . . .
8.7 Taylor Polynomials . . . . . . . . . . . . . .
8.8 Taylor Series . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
383
383
395
410
419
424
434
446
457
9 Curves in the Plane

9.1 Conic Sec ons . . . . . . . . . . .
9.2 Parametric Equa ons . . . . . . .
9.3 Calculus and Parametric Equa ons
9.4 Introduc on to Polar Coordinates
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
469
469
483
493
503
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
9.5
Calculus and Polar Func ons . . . . . . . . . . . . . . . . . . . 516
10 Vectors
10.1 Introduc on to Cartesian Coordinates in Space
10.2 An Introduc on to Vectors . . . . . . . . . . .
10.3 The Dot Product . . . . . . . . . . . . . . . . .
10.4 The Cross Product . . . . . . . . . . . . . . . .
10.5 Lines . . . . . . . . . . . . . . . . . . . . . . .
10.6 Planes . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
529
529
543
557
570
580
590
11 Vector Valued Func ons

11.1 VectorValued Func ons . . . . . . . . .
11.2 Calculus and VectorValued Func ons . .
11.3 The Calculus of Mo on . . . . . . . . . .
11.4 Unit Tangent and Normal Vectors . . . . .
11.5 The Arc Length Parameter and Curvature
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
599
599
605
618
631
640
12 Func
12.1
12.2
12.3
12.4
12.5
12.6
12.7
12.8
ons of Several Variables

Introduc on to Mul variable Func ons . . . . .
Limits and Con nuity of Mul variable Func ons .
Par al Deriva ves . . . . . . . . . . . . . . . . .
Dieren ability and the Total Dieren al . . . .
The Mul variable Chain Rule . . . . . . . . . . .
Direc onal Deriva ves . . . . . . . . . . . . . .
Tangent Lines, Normal Lines, and Tangent Planes
Extreme Values . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
651
651
658
668
680
689
696
705
715
13 Mul
13.1
13.2
13.3
13.4
13.5
13.6
ple Integra on
Iterated Integrals and Area . . . . . . . . .
Double Integra on and Volume . . . . . . .
Double Integra on with Polar Coordinates .
Center of Mass . . . . . . . . . . . . . . .
Surface Area . . . . . . . . . . . . . . . . .
Volume Between Surfaces and Triple Integra
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
725
725
735
746
753
765
772
A Solu ons To Selected Problems

Index
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . .
. . .
. . .
. . .
. . .
on .
A.1
A.33
1: L
Calculus means a method of calcula on or reasoning. When one computes
the sales tax on a purchase, one employs a simple calculus. When one nds the
area of a polygonal shape by breaking it up into a set of triangles, one is using
another calculus. Proving a theorem in geometry employs yet another calculus.
Despite the wonderful advances in mathema cs that had taken place into
the rst half of the 17th century, mathema cians and scien sts were keenly
aware of what they could not do. (This is true even today.) In par cular, two
important concepts eluded mastery by the great thinkers of that me: area and
rates of change.
Area seems innocuous enough; areas of circles, rectangles, parallelograms,
etc., are standard topics of study for students today just as they were then. However, the areas of arbitrary shapes could not be computed, even if the boundary
of the shape could be described exactly.
Rates of change were also important. When an object moves at a constant
rate of change, then distance = rate me. But what if the rate is not constant
can distance s ll be computed? Or, if distance is known, can we discover the
rate of change?
It turns out that these two concepts were related. Two mathema cians, Sir
Isaac Newton and Go ried Leibniz, are credited with independently formula ng
a system of compu ng that solved the above problems and showed how they
were connected. Their system of reasoning was a calculus. However, as the
power and importance of their discovery took hold, it became known to many
as the calculus. Today, we generally shorten this to discuss calculus.
The founda on of the calculus is the limit. It is a tool to describe a parcular behavior of a func on. This chapter begins our study of the limit by approxima ng its value graphically and numerically. A er a formal deni on of
the limit, proper es are established that make nding limits tractable. Once
the limit is understood, then the problems of area and rates of change can be
approached.
1.1
An Introduc on To Limits
We begin our study of limits by considering examples that demonstrate key concepts that will be explained as we progress.
Chapter 1 Limits
Consider the func on y = sinx x . When x is near the value 1, what value (if
any) is y near?
While our ques on is not precisely formed (what cons tutes near the value
1?), the answer does not seem dicult to nd. One might think rst to look at a
graph of this func on to approximate the appropriate y values. Consider Figure
1.1, where y = sinx x is graphed. For values of x near 1, it seems that y takes on
values near 0.85. In fact, when x = 1, then y = sin1 1 0.84, so it makes sense
that when x is near 1, y will be near 0.84.
Consider this again at a dierent value for x. When x is near 0, what value (if
any) is y near? By considering Figure 1.2, one can see that it seems that y takes
on values near 1. But what happens when x = 0? We have
y
1
0.8
0.6
x
1
0.5
1.5
Figure 1.1: sin(x)/x near x = 1.
0
sin 0
.
0
0
The expression 0/0 has no value; it is indeterminate. Such an expression gives

no informa on about what is going on with the func on nearby. We cannot nd
out how y behaves near x = 0 for this func on simply by le ng x = 0.
Finding a limit entails understanding how a func on behaves near a par cular value of x. Before con nuing, it will be useful to establish some nota on. Let
y = f(x); that is, let y be a func on of x for some func on f. The expression the
limit of y as x approaches 1 describes a number, o en referred to as L, that y
nears as x nears 1. We write all this as
lim y = lim f(x) = L.
x1
This is not a complete deni on (that will come in the next sec on); this is a
pseudo-deni on that will allow us to explore the idea of a limit.
Above, where f(x) = sin(x)/x, we approximated
0.9
0.8
.
1
lim
x1
Figure 1.2: sin(x)/x near x = 0.
sin x
0.84
x
and
lim
x0
sin x
1.
x
(We approximated these limits, hence used the symbol, since we are working with the pseudo-deni on of a limit, not the actual deni on.)
Once we have the true deni on of a limit, we will nd limits analy cally;
that is, exactly using a variety of mathema cal tools. For now, we will approximate limits both graphically and numerically. Graphing a func on can provide
a good approxima on, though o en not very precise. Numerical methods can
provide a more accurate approxima on. We have already approximated limits
graphically, so we now turn our a en on to numerical approxima ons.
Consider again limx1 sin(x)/x. To approximate this limit numerically, we
can create a table of x and f(x) values where x is near 1. This is done in Figure
1.3.
Notes:
x1
1.1 An Introduc on To Limits

No ce that for values of x near 1, we have sin(x)/x near 0.841. The x = 1 row
is in bold to highlight the fact that when considering limits, we are not concerned
with the value of the func on at that par cular x value; we are only concerned
with the values of the func on when x is near 1.
Now approximate limx0 sin(x)/x numerically. We already approximated
the value of this limit as 1 graphically in Figure 1.2. The table in Figure 1.4 shows
the value of sin(x)/x for values of x near 0. Ten places a er the decimal point
are shown to highlight how close to 1 the value of sin(x)/x gets as x takes on
values very near 0. We include the x = 0 row in bold again to stress that we are
not concerned with the value of our func on at x = 0, only on the behavior of
the func on near 0.
This numerical method gives condence to say that 1 is a good approximaon of limx0 sin(x)/x; that is,
x
0.9
0.99
0.999
1
1.001
1.01
1.1
Figure 1.3: Values of sin(x)/x with x near

1.
lim sin(x)/x 1.
x
-0.1
-0.01
-0.001
0
0.001
0.01
0.1
x0
Later we will be able to prove that the limit is exactly 1.

We now consider several examples that allow us explore dierent aspects
of the limit concept.
. Example 1
Approxima ng the value of a limit
Use graphical and numerical methods to approximate
x2 x 6
.
x3 6x2 19x + 3
lim
y = (x2 x 6)/(6x2 19x + 3)
0.34
on a small interval that contains 3. To numerically approximate the limit, create

a table of values where the x values are near 3. This is done in Figures 1.5 and
1.6, respec vely.
The graph shows that when x is near 3, the value of y is very near 0.3. By
considering values of x near 3, we see that y = 0.294 is a be er approxima on.
The graph and the table imply that
0.32
x x6
lim 2
0.294.
x3 6x 19x + 3
This example may bring up a few ques ons about approxima ng limits (and
the nature of limits themselves).
1. If a graph does not produce as good an approxima on as a table, why
bother with it?
Notes:
sin(x)/x
0.9983341665
0.9999833334
0.9999998333
not dened
0.9999998333
0.9999833334
0.9983341665
Figure 1.4: Values of sin(x)/x with x near

1.
To graphically approximate the limit, graph
sin(x)/x
0.870363
0.844471
0.841772
0.841471
0.84117
0.838447
0.810189
0.3
0.28
0.26
x
2.5
3.5
Figure 1.5: Graphically approxima ng a

limit in Example 1.
x
2.9
2.99
2.999
3
3.001
3.01
3.1
x2 x6
6x2 19x+3
0.29878
0.294569
0.294163
not dened
0.294073
0.293669
0.289773
Figure 1.6: Numerically approxima ng a

limit in Example 1.
Chapter 1 Limits
2. How many values of x in a table are enough? In the previous example,
could we have just used x = 3.001 and found a ne approxima on?
Graphs are useful since they give a visual understanding concerning the behavior of a func on. Some mes a func on may act erra cally near certain
x values which is hard to discern numerically but very plain graphically. Since
graphing u li es are very accessible, it makes sense to make proper use of them.
Since tables and graphs are used only to approximate the value of a limit,
there is not a rm answer to how many data points are enough. Include
enough so that a trend is clear, and use values (when possible) both less than
and greater than the value in ques on. In Example 1, we used both values less
than and greater than 3. Had we used just x = 3.001, we might have been
tempted to conclude that the limit had a value of 0.3. While this is not far o,
we could do be er. Using values on both sides of 3 helps us iden fy trends.
y
. Example 2
Approxima ng the value of a limit
Graphically and numerically approximate the limit of f(x) as x approaches 0,
where
{
x+1 x0
f(x) =
.
x2 + 1 x > 0
0.5
.1
x
0.5
0.5
Figure 1.7: Graphically approxima ng a

limit in Example 2.
x
-0.1
-0.01
-0.001
0.001
0.01
0.1
f(x)
0.9
0.99
0.999
0.999999
0.9999
0.99
Figure 1.8: Numerically approxima ng a

limit in Example 2.
Again we graph f(x) and create a table of its values near x =

S
0 to approximate the limit. Note that this is a piecewise dened func on, so it
behaves dierently on either side of 0. Figure 1.7 shows a graph of f(x), and on
either side of 0 it seems the y values approach 1.
The table shown in Figure 1.8 shows values of f(x) for values of x near 0. It
is clear that as x takes on values very near 0, f(x) takes on values very near 1.
It turns out that if we let x = 0 for either piece of f(x), 1 is returned; this is
signicant and well return to this idea later.
The graph and table allow us to say that limx0 f(x) 1; in fact, we are
probably very sure it equals 1. .
Iden fying When Limits Do Not Exist

A func on may not have a limit for all values of x. That is, we cannot say
limxc f(x) = L for some numbers L for all values of c, for there may not be a
number that f(x) is approaching. There are three ways in which a limit may fail
to exist.
1. The func on f(x) may approach dierent values on either side of c.
2. The func on may grow without upper or lower bound as x approaches c.
Notes:

3. The func on may oscillate as x approaches c.
Well explore each of these in turn.
. Example 3
Dierent Values Approached From Le and Right
Explore why lim f(x) does not exist, where
x1
{
f(x) =
y
3
x 2x + 3 x 1
.
x
x>1
2
A graph of f(x) around x = 1 and a table are given Figures

S
1.9 and 1.10, respec vely. It is clear that as x approaches 1, f(x) does not seem
to approach a single number. Instead, it seems as though f(x) approaches two
dierent numbers. When considering values of x less than 1 (approaching 1
from the le ), it seems that f(x) is approaching 2; when considering values of
x greater than 1 (approaching 1 from the right), it seems that f(x) is approaching 1. Recognizing this behavior is important; well study this in greater depth
later. Right now, it suces to say that the limit does not exist since f(x) is not
approaching one value as x approaches 1. .
. Example 4
The Func on Grows Without Bound
Explore why lim 1/(x 1)2 does not exist.
x1
A graph and table of f(x) = 1/(x 1)2 are given in Figures

1.11 and 1.12, respec vely. Both show that as x approaches 1, f(x) grows larger
and larger.
We can deduce this on our own, without the aid of the graph and table. If x
is near 1, then (x 1)2 is very small, and:
x
0.5
1.5
Figure 1.9: Observing no limit as x 1 in

Example 3.
x
0.9
0.99
0.999
1.001
1.01
1.1
f(x)
2.01
2.0001
2.000001
1.001
1.01
1.1
Figure 1.10: Values of f(x) near x = 1 in

Example 3.
1
very large number.
very small number
y
100
50
Since f(x) is not approaching a single number, we conclude that

lim
x1
1
(x 1)2
does not exist. .

. Example 5
.The Func on Oscillates
Explore why limx0 sin(1/x) does not exist.
Notes:
x
0.5
1.5
Figure 1.11: Observing no limit as x 1

in Example 4.
x
0.9
0.99
0.999
1.001
1.01
1.1
f(x)
100.
10000.
1. 106
1. 106
10000.
100.
Figure 1.12: Values of f(x) near x = 1 in

Example 4.
Chapter 1 Limits
Two graphs of f(x) = sin(1/x) are given in Figures 1.13. FigS
ure 1.13(a) shows f(x) on the interval [1, 1]; no ce how f(x) seems to oscillate
near x = 0. One might think that despite the oscilla on, as x approaches 0,
f(x) approaches 0. However, Figure 1.13(b) zooms in on sin(1/x), on the interval [0.1, 0.1]. Here the oscilla on is even more pronounced. Finally, in the
table in Figure 1.13(c), we see sin(x)/x evaluated for values of x near 0. As x
approaches 0, f(x) does not appear to approach any value.
It can be shown that in reality, as x approaches 0, sin(1/x) takes on all values
between 1 and 1 innite mes! Because of this oscilla on,
limx0 sin(1/x) does not exist...
y
0.5
0.5
x
0.5
0.5
x
0.1 5 102
0.5
0.1
0.5
5 102
.
(a)
x
0.1
0.01
0.001
0.0001
1. 105
1. 106
1. 107
(b)
sin(1/x)
0.544021
0.506366
0.82688
0.305614
0.0357488
0.349994
0.420548
(c)
Figure 1.13: Observing that f(x) = sin(1/x) has no limit as x 0 in Example 5.
Limits of Dierence Quo ents

We have approximated limits of func ons as x approached a par cular number. We will consider another important kind of limit a er explaining a few key
ideas.
Let f(x) represent the posi on func on, in feet, of some par cle that is moving in a straight line, where x is measured in seconds. Lets say that when x = 1,
the par cle is at posi on 10 ., and when x = 5, the par cle is at 20 . Another
way of expressing this is to say
20
f(1) = 10
10
x
2
f(5) = 20.
Since the par cle traveled 10 feet in 4 seconds, we can say the par cles average
velocity was 2.5 /s. We write this calcula on using a quo ent of dierences,
or, a dierence quo ent:
f(5) f(1)
10
=
= 2.5 /s.
51
4
Figure 1.14: Interpre ng a dierence

quo ent as the slope of a secant line.
Notes:
and

f
This dierence quo ent can be thought of as the familiar rise over run used
to compute the slopes of lines. In fact, that is essen ally what we are doing:
given two points on the graph of f, we are nding the slope of the secant line
through those two points. See Figure 1.14.
Now consider nding the average speed on another me interval. We again
start at x = 1, but consider the posi on of the par cle h seconds later. That is,
consider the posi ons of the par cle when x = 1 and when x = 1 + h. The
dierence quo ent is now
f(1 + h) f(1)
f(1 + h) f(1)
=
.
(1 + h) 1
h
Let f(x) = 1.5x2 + 11.5x; note that f(1) = 10 and f(5) = 20, as in our
discussion. We can compute this dierence quo ent for all values of h (even
nega ve values!) except h = 0, for then we get 0/0, the indeterminate form
introduced earlier. For all values h = 0, the dierence quo ent computes the
average velocity of the par cle over an interval of me of length h star ng at
x = 1.
For small values of h, i.e., values of h close to 0, we get average veloci es
over very short me periods and compute secant lines over small intervals. See
Figure 1.15. This leads us to wonder what the limit of the dierence quo ent is
as h approaches 0. That is,
lim
h0
10
Proper understanding of limits is key to understanding calculus. With limits,

we can accomplish seemingly impossible mathema cal things, like adding up an
innite number of numbers (and not get innity) and nding the slope of a line
between two points, where the two points are actually the same point. These
are not just mathema cal curiosi es; they allow us to link posi on, velocity and
accelera on together, connect cross-sec onal areas to volume, nd the work
done by a variable force, and much more.
In the next sec on we give the formal deni on of the limit and begin our
study of nding limits analy cally. In the following exercises, we con nue our
introduc on and approximate the value of limits.
.
(a)
20
10
.
(b)
f
f(1 + h) f(1)
= ?
h
As we do not yet have a true deni on of a limit nor an exact method for
compu ng it, we se le for approxima ng the value. While we could graph the
dierence quo ent (where the x-axis would represent h values and the y-axis
would represent values of the dierence quo ent) we se le for making a table.
See Figure 1.16. The table gives us reason to assume the value of the limit is
about 8.5.
Notes:
20
20
10
.
(c)
Figure 1.15: Secant lines of f(x) at x = 1

and x = 1 + h, for shrinking values of h
(i.e., h 0).
h
0.5
0.1
0.01
0.01
0.1
0.5
f(1+h)f(1)
h
9.25
8.65
8.515
8.485
8.35
7.75
Figure 1.16: The dierence quo ent evaluated at values of h near 0.
Exercises 1.1
Terms and Concepts
1. In your own words, what does it mean to nd the limit of
f(x) as x approaches 3?
0
0
13. lim f(x), where

x3
{ 2
x x+1
f(x) =
2x + 1
4. Describe three situa ons where lim f(x) does not exist.
14. lim f(x), where

x0
{
cos x
f(x) =
x2 + 3x + 1
5. In your own words, what is a dierence quo ent?
15.
2. An expression of the form
is called
3. T/F: The limit of f(x) as x approaches 5 is f(5).
x3
.
x>3
x0
.
x>0
xc
Problems
In Exercises 6 15, approximate the given limits both numerically and graphically.
6. lim x2 + 3x 5
x1
lim f(x), where

{
sin x x /2
f(x) =
.
cos x x > /2
x/2
In Exercises 16 24, a func on f and a value a are

given. Approximate the limit of the dierence quo ent,
f(a + h) f(a)
lim
, using h = 0.1, 0.01.
h0
h
7. lim x3 3x2 + x 5
16. f(x) = 7x + 2, a = 3
x+1
8. lim 2
x0 x + 3x
17. f(x) = 9x + 0.06, a = 1
x0
9. lim
x3
18. f(x) = x2 + 3x 7, a = 1
x2 2x 3
x2 4x + 3
19. f(x) =
x2 + 8x + 7
10. lim 2
x1 x + 6x + 5
x2 + 7x + 10
11. lim 2
x2 x 4x + 4
12. lim f(x), where
x2
{
x+2
f(x) =
3x 5
1
, a=2
x+1
20. f(x) = 4x2 + 5x 1, a = 3

21. f(x) = ln x,
x2
.
x>2
a=5
22. f(x) = sin x, a =

23. f(x) = cos x,
a=
1.2
Epsilon-Delta Deni on of a Limit
This sec on introduces the formal deni on of a limit, the epsilondelta, or

, deni on.
Before we give the actual deni on, lets consider a few informal ways of
describing a limit. Given a func on y = f(x) and an x value, call it c, we say that
the limit of the func on f is a value L:
1. if y tends to L as x tends to c.
2. if y approaches L as x approaches c.
3. if y is near L whenever x is near c.
The problem with these deni ons is that the words tends, approach,
and especially near are not exact. In what way does the variable x tend to or
approach c? How near do x and y have to be to c and L, respec vely?
The deni on we describe in this sec on comes from formalizing 3. A quick
restatement gets us closer to what we want:
3 . If x is within a certain tolerance level of c, then the corresponding value y =
f(x) is within a certain tolerance level of L.
The accepted nota on for the x-tolerance is the lowercase Greek le er delta,
or , and the y-tolerance is lowercase epsilon, or . One more rephrasing of 3
nearly gets us to the actual deni on:
3 . If x is within units of c, then the corresponding value of y is within units
of L.
Note that this means (let the represent the word implies):
c < x < c + L < y < L +
or
|x c| < |y L| <
The point is that and , being tolerances, can be any posi ve (but typically
small) values. Finally, we have the formal deni on of the limit with the nota on
seen in the previous sec on.
Notes:
Chapter 1 Limits
.
Deni on 1
The Limit of a Func on f
Let f be a func on dened on an open interval containing c. The nota on
lim f(x) = L,
xc
read as the limit of f(x), as x approaches c, is L, means that given any

> 0, there exists > 0 such that whenever |x c| < , we have
|f(x) L| < .
(Mathema cians o en enjoy wri ng ideas without using any words. Here is
the wordless deni on of the limit:
lim f(x) = L > 0, > 0 s.t. |x c| < |f(x) L| < .)
xc
There is an emphasis here that we may have passed over before. In the
deni on, the y-tolerance is given rst and then the limit will exist if we can
nd an x-tolerance that works.
It is me for an example. Note that the explana on is long, but it will take
you through all steps necessary to understand the ideas.
}
= .5
. Example 6
.Evalua ng a limit using the deni on
Show that lim x = 2.
= .5
x4
Choose > 0. Then ...
x
4
Before we use the formal deni on, lets try some numerical
S
tolerances. What if the y tolerance is 0.5, or = 0.5? How close to 4 does x
have to be so that y is within 0.5 units of 2 (or 1.5 < y < 2.5)? In this case, we
can just square these values to get 1.52 < x < 2.52 , or
2.25 < x < 6.25.
So, what is the desired x tolerance? Remember, we want to nd a symmetric

interval of x values, namely 4 < x < 4 + . The lower bound of 2.25 is 1.75
units from 4; the upper bound of 6.25 is 2.25 units from 4. We need the smaller
of these two distances; we must have = 1.75. See Figure 1.17.
= .5
}
= .5
... choose smaller

than each of these
1
z
width
= 1.75
}|
{z
width
= 2.25
}|
x
2
Now read it in the correct way: For the y tolerance = 0.5, we have found
an x tolerance, = 1.75, so that whenever x is within units of 4, then y is
within units of 2. Thats what we were trying to nd.
With = 0.5, we pick any < 1.75.
Lets try another value of .

Figure 1.17: Illustra ng the process.
Notes:
10

What if the y tolerance is 0.01, or = 0.01? How close to 4 does x have to
be in order for y to be within 0.01 units of 2 (or 1.99 < y < 2.01)? Again, we
just square these values to get 1.992 < x < 2.012 , or
3.9601 < x < 4.0401.
So, what is the desired x tolerance? In this case we must have = 0.0399. Note
that in some sense, it looks like there are two tolerances (below 4 of 0.0399
units and above 4 of 0.0401 units). However, we couldnt use the larger value
of 0.0401 for since then the interval for x would be 3.9599 < x < 4.0401
resul ng in y values of 1.98995 < y < 2.01 (which contains values NOT within
0.01 units of 2).
What we have so far: if = 0.5, then = 1.75 and if = 0.01, then
= 0.0399. A pa ern is not easy to see, so we switch to general
and and do
the calcula ons symbolically. We start by assuming y = x is within units of

2:
|y 2| <
< y 2 <
< x 2 <
2< x<2+
(2 )2 < x < (2 + )2
4 4 + 2 < x < 4 + 4 + 2
4 (4 2 ) < x < 4 + (4 + 2 )
(Deni on of absolute value)
(y = x)
(Add 2)
(Square all)
(Expand)
(Rewrite in the desired form)
Since we want this last interval to describe an x tolerance around 4, we have

that either = 4 + 2 or = 4 2 . However, as we saw in the case when
= 0.01, we want the smaller of the two values for . So, to conclude this part,
we set equal to the minimum of these two values, or = min{4+2 , 42 }.
Since > 0, the minimum will occur when = 4 2 . Thats the formula!
We can check this for our previous values. If = 0.5, the formula gives
= 4(0.5)(0.5)2 = 1.75 and when = 0.01, the formula gives = 4(0.01)
(0.01)2 = 0.399.
So given any > 0, we can set = 4 2 and the
limit deni on is sa sed. We have shown formally (and nally!) that lim x = 2. .
x4
If you are thinking this process is long,

you would be right. The previous
example is also a bit unsa sfying in that 4 = 2; why work so hard to prove
something so obvious? Many proofs are long and dicult to do. In this
Notes:
11
Chapter 1 Limits
sec on, we will focus on examples where the answer is, frankly, obvious, because the nonobvious examples are even harder. That is why theorems about
limits are so useful! A er doing a few more proofs, you will really appreciate
the analy cal short cuts found in the next sec on.
. Example 7
Show that lim x2 = 4.
x2
Lets do this example symbolically from the start. Let > 0

S
be given; we want |y 4| < , i.e., |x2 4| < . How do we nd such that
when |x 2| < , we are guaranteed that |x2 4| < ?
This is a bit trickier than the previous example, but lets start by no cing that
|x2 4| = |x 2| |x + 2|. Consider:
|x2 4| < |x 2| |x + 2| < |x 2| <
.
|x + 2|
(1.1)
?
|x + 2|
We are close to an answer, but the catch is that must be a constant value (so
it cant contain x). There is a way to work around this, but we do have to make an
assump on. Remember that is supposed to be a small number, which implies
that will also be a small value. In par cular, we can (probably) assume that
< 1. If this is true, then |x 2| < would imply that |x 2| < 1, giving
1 < x < 3.
Now, back to the frac on

. If 1 < x < 3, then 3 < x + 2 < 5. Taking
|x + 2|
1
1
1
reciprocals, we have <
< so that, in par cular,
5
|x + 2|
3
Could we not set =
<
.
5
|x + 2|
This suggests that we set =
. To see why, lets go back to the equa ons:

5
|x 2| <
|x 2| <
5
|x 2| |x + 2| < |x + 2|
5
|x2 4| < |x + 2|
5
|x2 4| < |x + 2| < |x + 2|

=
5
|x + 2|
Notes:
12
(1.2)
(Our choice of )
(Mul ply by |x + 2|)
(Combine le side)
(Using (1.2) as long as < 1)

We have arrived at |x2 4| < as desired. Note again, in order to make this
happen we needed to rst be less than 1. That is a safe assump on; we want
to be arbitrarily small, forcing to also be small.
We have also picked to be smaller than necessary. We could get by with a
slightly larger , as shown in Figure 1.18. The dashed, red lines show the boundaries dened by our choice of . The gray, dashed lines show the boundaries
dened by se ng = /5. Note how these gray lines are within the red lines.
That is perfectly ne; by choosing x within the gray lines we are guaranteed that
f(x) will be within of 4.
In summary, given > 0, set = /5. Then |x 2| < implies |x2 4| <
(i.e. |y 4| < ) as desired. We have shown that lim x2 = 4. Figure 1.18 gives
length of
= /5
. Example 8
Show that lim ex = 1.
z}|{
x2
a visualiza on of this; by restric ng x to values within = /5 of 2, we see that

f(x) is within of 4. .
length of
Figure 1.18: Choosing = /5 in Example 7.
x0
Symbolically, we want to take the equa on |ex 1| < and

unravel it to the form |x 0| < . Lets look at some calcula ons:
S
|ex 1| <
< ex 1 <
1 < ex < 1 +
ln(1 ) < x < ln(1 + )

(Add 1)
(Take natural logs)
Making the safe assump on that < 1 ensures the last inequality is valid (i.e.,
so that ln(1 ) is dened). We can then set to be the minimum of | ln(1 )|
and ln(1 + ); i.e.,
Note: Recall ln 1 = 0 and ln x < 0 when

0 < x < 1. So ln(1 ) < 0, hence we
consider its absolute value.
= min{| ln(1 )|, ln(1 + )}.

Now, we work through the actual the proof:
|x 0| <
< x <
ln(1 ) < x < ln(1 + )
1 < ex < 1 +
< ex 1 <

(By our choice of )
(Exponen ate)
(Subtract 1)
Notes:
13
Chapter 1 Limits
In summary, given > 0, let = min{| ln(1|, ln(1+)}. Then |x0| <
implies |ex 1| < as desired. We have shown that lim ex = 1. .
x0
We note that we could actually show that limxc ex = ec for any constant c.
We do this by factoring out ec from both sides, leaving us to show limxc exc =
1 instead. By using the subs tu on y = xc, this reduces to showing limy0 ey =
1 which we just did in the last example. As an added benet, this shows that in
fact the func on f(x) = ex is con nuous at all values of x, an important concept
we will dene in Sec on 1.5.
Notes:
14
Exercises 1.2
Terms and Concepts
Problems
1. What is wrong with the following deni on of a limit?
Exercises 5 11, prove the given limit using an proof.
The limit of f(x), as x approaches a, is K

means that given any > 0 there exists > 0
such that whenever |f(x) K| < , we have
|x a| < .
5. lim 3 x = 2
x5
6. lim x2 3 = 6
x3
7. lim x2 + x 5 = 15
x4
2. Which is given rst in establishing a limit, the xtolerance

or the ytolerance?
8. lim x3 1 = 7
x2
9. lim 5 = 5
x2
3. T/F: must always be posi ve.
10. lim e2x 1 = 0

x0
4. T/F: must always be posi ve.
11. lim sin x = 0 (Hint: use the fact that | sin x| |x|, with
x0
equality only when x = 0.)
Chapter 1 Limits
1.3 Finding Limits Analy cally

In Sec on 1.1 we explored the concept of the limit without a strict deni on,
meaning we could only make approxima ons. In the previous sec on we gave
the deni on of the limit and demonstrated how to use it to verify our approxima ons were correct. Thus far, our method of nding a limit is 1) make a really
good approxima on either graphically or numerically, and 2) verify our approxima on is correct using a - proof.
Recognizing that proofs are cumbersome, this sec on gives a series of
theorems which allow us to nd limits much more quickly and intui vely.
Suppose that limx2 f(x) = 2 and limx2 g(x) = 3. What is limx2 (f(x) +
g(x))? Intui on tells us that the limit should be 5, as we expect limits to behave
in a nice way. The following theorem states that already established limits do
behave nicely.
.
Theorem 1
Basic Limit Proper es
Let b, c, L and K be real numbers, let n be a posi ve integer, and let f and g be
func ons with the following limits:
lim f(x) = L and lim g(x) = K.
xc
xc
The following limits hold.

1. Constants:
2. Iden ty
3. Sums/Dierences:
4. Scalar Mul ples:
5. Products:
6. Quo ents:
7. Powers:
8. Roots:
9. Composi ons:
lim b = b
xc
lim x = c
xc
lim (f(x) g(x)) = L K
xc
lim b f(x) = bL
xc
lim f(x) g(x) = LK
xc
lim f(x)/g(x) = L/K, (K = 0)
xc
lim f(x)n = Ln
n
lim n f(x) = L
xc
xc
Adjust our previously given limit situa on to:

lim f(x) = L and lim g(x) = K.
xc
Then lim g(f(x)) = K.

xc
Notes:
16
xL
We make a note about Property #8: when n is even, L must be greater than
0. If n is odd, then the statement is true for all L.
We apply the theorem to an example.
. Example 9
Using basic limit proper es
Let
lim f(x) = 2, lim g(x) = 3 and p(x) = 3x2 5x + 7.
x2
x2
Find the following limits:

(
)
1. lim f(x) + g(x)
x2
2. lim 5f(x) + g(x)
3. lim p(x)
x2
)
2
x2
(
)
1. Using the Sum/Dierence rule, we know that lim f(x) + g(x) = 2 + 3 =
x2
5.
(
2. Using the Scalar Mul ple and Sum/Dierence rules, we nd that lim 5f(x)+
x2
)
g(x)2 = 5 2 + 32 = 19.
3. Here we combine the Power, Scalar Mul ple, Sum/Dierence and Constant Rules. We show quite a few steps, but in general these can be omitted:
lim p(x) = lim (3x2 5x + 7)
x2
x2
= lim 3x2 lim 5x + lim 7

x2
x2
x2
=32 52+7
=9
2
Part 3 of the previous example demonstrates how the limit of a quadra c

polynomial can be determined using the proper es of Theorem 1. Not only that,
recognize that
lim p(x) = 9 = p(2);
x2
i.e., the limit at 2 was found just by plugging 2 into the func on. This holds
true for all polynomials, and also for ra onal func ons (which are quo ents of
polynomials), as stated in the following theorem.
Notes:
17
Chapter 1 Limits
.
Theorem 2
Limits of Polynomial and Ra onal Func ons
Let p(x) and q(x) be polynomials and c a real number. Then:
1. lim p(x) = p(c)

xc
2. lim
xc
p(x)
p(c)
=
, where q(c) = 0.
q(x)
q(c)
. Example 10
Finding a limit of a ra onal func on
Using Theorem 2, nd
3x2 5x + 1
lim 4
.
x1 x x2 + 3
S
Using Theorem 2, we can quickly state that
3x2 5x + 1
3(1)2 5(1) + 1
=
4
2
x1 x x + 3
(1)4 (1)2 + 3
9
.
= = 3.
3
It was likely frustra ng in Sec on 1.2 to do a lot of work to prove that
lim
lim x2 = 4
x2
as it seemed fairly obvious. The previous theorems state that many func ons
behave in such an obvious fashion, as demonstrated by the ra onal func on
in Example 10.
Polynomial and ra onal func ons are not the only func ons to behave in
such a predictable way. The following theorem gives a list of func ons whose
behavior is par cularly nice in terms of limits. In the next sec on, we will give
a formal name to these func ons that behave nicely.
.
Theorem 3
Special Limits
Let c be a real number in the domain of the given func on and let n be a posi ve integer. The
following limits hold:
4. lim csc x = csc c
7. lim ax = ac (a > 0)
2. lim cos x = cos c
5. lim sec x = sec c
8. lim ln x = ln c
3. lim tan x = tan c
6. lim cot x = cot c
9. lim
xc
xc
xc
Notes:
18
1. lim sin x = sin c
xc
xc
xc
xc
xc
xc
n
x= nc

. Example 11
.Evalua ng limits analy cally
Evaluate the following limits.
1. lim cos x
4. lim eln x
x1
2. lim (sec x tan x)

2
5. lim
x3
3.
x0
lim cos x sin x
sin x
x
x/2
1. This is a straigh orward applica on of Theorem 3. lim cos x = cos =

x
1.
2. We can approach this in at least two ways. First, by directly applying Theorem 3, we have:
lim (sec2 x tan2 x) = sec2 3 tan2 3.
x3
Using the Pythagorean Theorem, this last expression is 1; therefore

lim (sec2 x tan2 x) = 1.
x3
We can also use the Pythagorean Theorem from the start.

lim (sec2 x tan2 x) = lim 1 = 1,
x3
x3
using the Constant limit rule. Either way, we nd the limit is 1.

3. Applying Theorem 3 directly gives
lim cos x sin x = cos(/2) sin(/2) = 0 1 = 0.
x/2
4. Again, we can approach this in two ways. First, we can use the exponenal/logarithmic iden ty that eln x = x and evaluate lim eln x = lim x = 1.
x1
x1
We can also use the Composi on limit rule. Using Theorem 3, we have
lim ln x = ln 1 = 0. Thus
x1
lim eln x = lim ex = e0 = 1.
x1
x0
Both approaches are valid, giving the same result.
Notes:
19
Chapter 1 Limits
5. We encountered this limit in Sec on 1.1. Applying our theorems, we attempt to nd the limit as
0
sin x
sin 0
.
x0 x
0
0
lim
This, of course, violates a condi on of Theorem 1, as the limit of the denominator is not allowed to be 0. Therefore, we are s ll unable to evaluate
this limit with tools we currently have at hand.
.
The sec on could have been tled Using Known Limits to Find Unknown
Limits. By knowing certain limits of func ons, we can nd limits involving sums,
products, powers, etc., of these func ons. We further the development of such
compara ve tools with the Squeeze Theorem, a clever and intui ve way to nd
the value of some limits.
Before sta ng this theorem formally, suppose we have func ons f, g and h
where g always takes on values between f and h; that is, for all x in an interval,
f(x) g(x) h(x).
If f and h have the same limit at c, and g is always squeezed between them,
then g must have the same limit as well. That is what the Squeeze Theorem
states.
.
Theorem 4
Squeeze Theorem
Let f, g and h be func ons on an open interval I containing c such that

for all x in I,
f(x) g(x) h(x).
If
lim f(x) = L = lim h(x),
xc
xc
then
lim g(x) = L.
xc
It can take some work to gure out appropriate func ons by which to squeeze
the given func on you are trying to evaluate a limit of. However, that is generally the only place work is necessary; the theorem makes the evalua ng the
limit part very simple.
Notes:
20

We use the Squeeze Theorem in the following example to nally prove that
sin x
lim
= 1.
x0 x
. Example 12
.Using the Squeeze Theorem
Use the Squeeze Theorem to show that
lim
x0
sin x
= 1.
x
(1, tan )
(cos , sin )
We begin by considering the unit circle. Each point on the

S
unit circle has coordinates (cos , sin ) for some angle as shown in Figure 1.19.
Using similar triangles, we can extend the line from the origin through the point
to the point (1, tan ), as shown. (Here we are assuming that 0 /2.
Later we will show that we can also consider 0.)
The area of the large triangle is 12 tan ; the area of the sector is /2; the
area of the triangle contained inside the sector is 12 sin . It is then clear from
the diagram that
(1, 0)
Figure 1.19: The unit circle and related triangles.

tan
sin
tan
2
Mul ply all terms by
sin
2
2
, giving
sin
1.
cos
sin
Taking reciprocals reverses the inequali es, giving

cos
sin
1.
(These inequali es hold for all values of near 0, even nega ve values, since
cos() = cos and sin() = sin .)
Now take limits.
Notes:
21
Chapter 1 Limits
sin
lim 1
0
0
lim cos lim
cos 0 lim
sin
1
sin
1
0
1 lim
Clearly this means that lim
sin
= 1.
Two notes about the previous example are worth men oning. First, one
might be discouraged by this applica on, thinking I would never have come up
with that on my own. This is too hard! Dont be discouraged; within this text we
will guide you in your use of the Squeeze Theorem. As one gains mathema cal
maturity, clever proofs like this are easier and easier to create.
Second, this limit tells us more than just that as x approaches 0, sin(x)/x
approaches 1. Both x and sin x are approaching 0, but the ra o of x and sin x
approaches 1, meaning that they are approaching 0 in essen ally the same way.
Another way of viewing this is: for small x, the func ons y = x and y = sin x are
essen ally indis nguishable.
We include this special limit, along with three others, in the following theorem.
.
Theorem 5
Special Limits
sin x
=1
x0 x
1. lim
cos x 1
=0
x0
x
2. lim
.3.
lim (1 + x) x = e
x0
ex 1
=1
x0
x
4. lim
A short word on how to interpret the la er three limits. We know that as

x goes to 0, cos x goes to 1. So, in the second limit, both the numerator and
denominator are approaching 0. However, since the limit is 0, we can interpret
this as saying that cos x is approaching 1 faster than x is approaching 0.
In the third limit, inside the parentheses we have an expression that is approaching 1 (though never equaling 1), and we know that 1 raised to any power
is s ll 1. At the same me, the power is growing toward innity. What happens
Notes:
22

to a number near 1 raised to a very large power? In this par cular case, the
result approaches Eulers number, e, approximately 2.718.
In the fourth limit, we see that as x 0, ex approaches 1 just as fast as
x 0, resul ng in a limit of 1.
Our nal theorem for this sec on will be mo vated by the following example.
. Example 13
Using algebra to evaluate a limit
Evaluate the following limit:
x2 1
lim
.
x1 x 1
We begin by a emp ng to apply Theorem 3 and subs tu ng
S
1 for x in the quo ent. This gives:
x2 1
12 1 0
=
=
,
x1 x 1
11
0
lim
and indeterminate form. We cannot apply the theorem.

By graphing the func on, as in Figure 1.20, we see that the func on seems
to be linear, implying that the limit should be easy to evaluate. Recognize that
the numerator of our quo ent can be factored:
x2 1
(x 1)(x + 1)
=
.
x1
x1
The func on is not dened when x = 1, but for all other x,
x
1
Figure 1.20: Graphing f in Example 13 to

understand a limit.
x2 1
(x 1)(x + 1)
(x 1)(x + 1)
=
=
= x + 1.
x1
x1
x1
Clearly lim x +1 = 2. Recall that when considering limits, we are not concerned
x1
with the value of the func on at 1, only the value the func on approaches as x
approaches 1. Since (x2 1)/(x 1) and x + 1 are the same at all points except
x = 1, they both approach the same value as x approaches 1. Therefore we can
conclude that
x2 1
lim
= 2.
.
x1 x 1
The key to the above example is that the func ons y = (x2 1)/(x 1) and
y = x + 1 are iden cal except at x = 1. Since limits describe a value the func on
is approaching, not the value the func on actually a ains, the limits of the two
func ons are always equal.
Notes:
23
Chapter 1 Limits
.
Theorem 6
Limits of Func ons Equal At All But One Point
Let g(x) = f(x) for all x in an open interval, except possibly at c, and let
lim g(x) = L for some real number L. Then
xc
lim f(x) = L.
xc
The Fundamental Theorem of Algebra tells us that when dealing with a rag(x)
onal func on of the form g(x)/f(x) and directly evalua ng the limit lim
xc f(x)
returns 0/0, then (x c) is a factor of both g(x) and f(x). One can then use
algebra to factor this term out, cancel, then apply Theorem 6. We demonstrate
this once more.
. Example 14
Evalua ng a limit using Theorem 6

x3 2x2 5x + 6
Evaluate lim 3
.
x3 2x + 3x2 32x + 15
We begin by applying Theorem 3 and subs tu ng 3 for x. This returns the familiar indeterminate form of 0/0. Since the numerator and denominator are each
polynomials, we know that (x 3) is factor of each. Using whatever method is
most comfortable to you, factor out (x3) from each (using polynomial division,
synthe c division, a computer algebra system, etc.). We nd that
x3 2x2 5x + 6
(x 3)(x2 + x 2)
=
.
2x3 + 3x2 32x + 15
(x 3)(2x2 + 9x 5)
We can cancel the (x3) terms as long as x = 3. Using Theorem 6 we conclude:
x3 2x2 5x + 6
(x 3)(x2 + x 2)
=
lim
x3 2x3 + 3x2 32x + 15
x3 (x 3)(2x2 + 9x 5)
(x2 + x 2)
= lim
x3 (2x2 + 9x 5)
10
1
=
= .
40
4
lim
.
We end this sec on by revisi ng a limit rst seen in Sec on 1.1, a limit of
a dierence quo ent. Let f(x) = 1.5x2 + 11.5x; we approximated the limit
f(1 + h) f(1)
lim
8.5. We formally evaluate this limit in the following exh0
h
ample.
Notes:
24

. Example 15
Evalua ng the limit of a dierence quo ent

f(1 + h) f(1)
Let f(x) = 1.5x + 11.5x; nd lim
.
h0
h
2
Since f is a polynomial, our rst a empt should be to em0

ploy Theorem 3 and subs tute 0 for h. However, we see that this gives us
.
0
Knowing that we have a ra onal func on hints that some algebra will help. Consider the following steps:
(
)
1.5(1 + h)2 + 11.5(1 + h) 1.5(1)2 + 11.5(1)
f(1 + h) f(1)
lim
= lim
h0
h0
h
h
1.5(1 + 2h + h2 ) + 11.5 + 11.5h 10
= lim
h0
h
1.5h2 + 8.5h
= lim
h0
h
= lim (1.5h + 8.5) (using Theorem 6, as h = 0)
S
h0
= 8.5 (using Theorem 3)

This matches our previous approxima on. .
This sec on contains several valuable tools for evalua ng limits. One of the
main results of this sec on is Theorem 3; it states that many func ons that we
use regularly behave in a very nice, predictable way. In the next sec on we give
a name to this nice behavior; we label such func ons as con nuous. Dening
that term will require us to look again at what a limit is and what causes limits
to not exist.
Notes:
25
Exercises 1.3
Terms and Concepts
1. Explain in your own words, without using formality,
why lim b = b.
xc
(
)
17. lim g 5f(x)
x1
In Exercises 18 32, evaluate the given limit.

18. lim x2 3x + 7
x3
2. Explain in your own words, without using formality,

why lim x = c.
xc
3. What does the text mean when it says that certain funcons behavior is nice in terms of limits? What, in parcular, is nice?
19. lim
20.
lim cos x sin x
21. lim ln x
x0
22. lim 4x
5. You are given the following informa on:
23.
x1
)7
x/4
4. Sketch a graph that visually demonstrates the Squeeze Theorem.
(a) lim f(x) = 0
x3
x5
8x
x3
lim csc x
x/6
24. lim ln(1 + x)
(b) lim g(x) = 0
x0
x1
(c) lim f(x)/g(x) = 2
x2 + 3x + 5
5x2 2x 3
3x + 1
26. lim
x 1 x
25. lim
x1
What can be said about the rela ve sizes of f(x) and g(x)
as x approaches 1?
Problems
27. lim
x2 4x 12
x2 13x + 42
Using:
lim f(x) = 6
28. lim
x2 + 2x
x2 2x
29. lim
x2 + 6x 16
x2 3x + 2
30. lim
x2 10x + 16
x2 x 2
x9
lim g(x) = 3
x9
x6
lim f(x) = 9
x0
x6
lim g(x) = 3
x6
evaluate the limits given in Exercises 6 13, where possible.

If it is not possible to know, state so.
x2
x2
6. lim (f(x) + g(x))

x9
7. lim (3f(x)/g(x))
x9
(
)
f(x) 2g(x)
8. lim
x9
g(x)
(
)
f(x)
9. lim
x6
3 g(x)
(
)
10. lim g f(x)
x9
(
)
11. lim f g(x)
x6
(
)
12. lim g f(f(x))
31. lim
x2 5x 14
x2 + 10x + 16
32. lim
x2 + 9x + 8
x2 6x 7
13. lim f(x)g(x) f 2 (x) + g2 (x)
35. lim f(x), where x2 f(x) 3x on [0, 3].
x2
x1
Use the Squeeze Theorem in Exercises 33 35, where appropriate, to evaluate the given limit.
( )
1
33. lim x sin
x0
x
( )
1
34. lim sin x cos 2
x0
x
x6
x6
Using:
lim f(x) = 2
x1
lim g(x) = 0
x1
lim f(x) = 1
x10
lim g(x) =
x10
evaluate the limits given in Exercises 14 17, where possible.

If it is not possible to know, state so.
14. lim f(x)g(x)
x1
(
)
15. lim cos g(x)
x10
16. lim f(x)g(x)

x1
x3
Exercises 36 39 challenge your understanding of limits but

can be evaluated using the knowledge gained in this sec on.
sin 3x
x
sin 5x
37. lim
x0 8x
36. lim
x0
ln(1 + x)
x
sin x
, where x is measured in degrees, not radians.
39. lim
x0 x
38. lim
x0
1.4 One Sided Limits
1.4
One Sided Limits
We introduced the concept of a limit gently, approxima ng their values graphically and numerically. Next came the rigorous deni on of the limit, along with
an admi edly tedious method for compu ng them. The previous sec on gave
us tools (which we call theorems) that allow us to compute limits with greater
ease. Chief among the results were the facts that polynomials and ra onal,
trigonometric, exponen al and logarithmic func ons (and their sums, products,
etc.) all behave nicely. In this sec on we rigorously dene what we mean by
nicely.
In Sec on 1.1 we explored the three ways in which limits of func ons failed
to exist:
1. The func on approached dierent values from the le and right,
2. The func on grows without bound, and
3. The func on oscillates.
In this sec on we explore in depth the concepts behind #1 by introducing
the one-sided limit. We begin with formal deni ons that are very similar to the
deni on of the limit given in Sec on 1.2, but the nota on is slightly dierent
and a short phrase has been added to the end.
Deni on 2
One Sided Limits
Le -Hand Limit
lim f(x) = L,
xc
read as the limit of f(x) as x approaches c from the le is L, or the le -hand limit of f at c is L
means that given any > 0, there exists > 0 such that |xc| < implies |f(x)L| < , for all x < c.
Right-Hand Limit

lim f(x) = L,
xc+
read as the limit of f(x) as x approaches c from the right is L, or the right-hand limit of f at c is L
means that given any > 0, there exists > 0 such that |x c| < implies |f(x) L| < , for all
x > c.
Notes:
27
Chapter 1 Limits
Prac cally speaking, when evalua ng a le -hand limit, we consider only values of x to the le of c, i.e., where x < c. The admi edly imperfect nota on
x c is used to imply that we look at values of x to the le of c. The notaon has nothing to do with posi ve or nega ve values of either x or c. A similar
statement holds for evalua ng right-hand limits; there we consider only values
of x to the right of c, i.e., x > c. We can use the theorems from previous sec ons
to help us evaluate these limits; we just restrict our view to one side of c.
We prac ce evalua ng le and right-hand limits through a series of examples.
y
2
. Example 16
Let f(x) =
1
following:
1. lim f(x)
5. lim f(x)
2. lim f(x)
6. f(0)
3. lim f(x)
7. lim f(x)
4. f(1)
8. f(2)
x1
.Evalua ng one sided limits

x
0x1
, as shown in Figure 1.21. Find each of the
3x 1<x<2
x0+
x1+
Figure 1.21: A graph of f in Example 16.
x2
x1
For these problems, the visual aid of the graph is likely more
S
eec ve in evalua ng the limits than using f itself. Therefore we will refer o en
to the graph.
1. As x goes to 1 from the le , we see that f(x) is approaching the value of 1.
Therefore lim f(x) = 1.
x1
2. As x goes to 1 from the right, we see that f(x) is approaching the value of 2.
Recall that it does not ma er that there is an open circle there; we are
evalua ng a limit, not the value of the func on. Therefore lim f(x) = 2.
x1+
3. The limit of f as x approaches 1 does not exist, as discussed in the rst

sec on. The func on does not approach one par cular value, but two
dierent values from the le and the right.
4. Using the deni on and by looking at the graph we see that f(1) = 1.
5. As x goes to 0 from the right, we see that f(x) is also approaching 0. Therefore lim f(x) = 0. Note we cannot consider a le -hand limit at 0 as f is
x0+
not dened for values of x < 0.
Notes:
28

6. Using the deni on and the graph, f(0) = 0.
7. As x goes to 2 from the le , we see that f(x) is approaching the value of
1. Therefore lim f(x) = 1.
x2
8. The graph and the deni on of the func on show that f(2) is not dened.
Note how the le and right-hand limits were dierent; this, of course, causes
the limit to not exist. The following theorem states what is fairly intui ve: the
limit exists precisely when the le and right-hand limits are equal.
Theorem 7
Limits and One Sided Limits
Let f be a func on dened on an open interval I containing c. Then
lim f(x) = L
xc
if, and only if,

lim f(x) = L
xc
and
lim f(x) = L.
xc+
The phrase if, and only if means the two statements are equivalent: they
are either both true or both false. If the limit equals L, then the le and right
hand limits both equal L. If the limit is not equal to L, then at least one of the
le and right-hand limits is not equal to L (it may not even exist).
One thing to consider in Examples 16 19 is that the value of the func on
may/may not be equal to the value(s) of its le /right-hand limits, even when
these limits agree.
y
2
. Example 17
Let f(x) =
.Evalua ng limits of a piecewisedened func on

2x
0<x<1
, as shown in Figure 1.22. Evaluate the
(x 2)2 1 < x < 2
following.
1. lim f(x)
x1
5. lim f(x)
x0+
2. lim f(x)
6. f(0)
3. lim f(x)
7. lim f(x)
4. f(1)
8. f(2)
x1+
x1
Figure 1.22: A graph of f from Example 17
x2
Notes:
29
Chapter 1 Limits
Again we will evaluate each using both the deni on of f and
its graph.
1. As x approaches 1 from the le , we see that f(x) approaches 1. Therefore
lim f(x) = 1.
x1
2. As x approaches 1 from the right, we see that again f(x) approaches 1.

Therefore lim f(x) = 1.
x1+
3. The limit of f as x approaches 1 exists and is 1, as f approaches 1 from both

the right and le . Therefore lim f(x) = 1.
x1
4. f(1) is not dened. Note that 1 is not in the domain of f as dened by the
problem, which is indicated on the graph by an open circle when x = 1.
5. As x goes to 0 from the right, f(x) approaches 2. So lim f(x) = 2.
x0+
6. f(0) is not dened as 0 is not in the domain of f.
7. As x goes to 2 from the le , f(x) approaches 0. So lim f(x) = 0.

x2
8. f(2) is not dened as 2 is not in the domain of f.
0.5
x
1
.
Figure 1.23: Graphing f in Example 18
. Example 18
Evalua ng limits of a piecewisedened func on
{
(x 1)2 0 x 2, x = 1
Let f(x) =
, as shown in Figure 1.23. Evaluate
1
x=1
the following.
1. lim f(x)
3. lim f(x)
2. lim f(x)
4. f(1)
x1
x1+
x1
It is clear by looking at the graph that both the le and rightS

hand limits of f, as x approaches 1, is 0. Thus it is also clear that the limit is 0;
i.e., lim f(x) = 0. It is also clearly stated that f(1) = 1.
x1
. Example 19
{
Let f(x) =
0.5
x
1
ing.
.
Figure 1.24: Graphing f in Example 19
Notes:
30
x2
2x
.Evalua ng limits of a piecewisedened func on

0x1
, as shown in Figure 1.24. Evaluate the follow1<x2
1. lim f(x)
3. lim f(x)
2. lim f(x)
4. f(1)
x1
x1
x1+
It is clear from the deni on of the func on and its graph

S
that all of the following are equal:
lim f(x) = lim f(x) = lim f(x) = f(1) = 1.
x1
x1
x1+
.
In Examples 16 19 we were asked to nd both lim f(x) and f(1). Consider
x1
the following table:

lim f(x)
f(1)
does not exist

1
0
1
1
not dened
1
1
x1
Example 16
Example 17
Example 18
Example 19
Only in Example 19 do both the func on and the limit exist and agree. This
seems nice; in fact, it seems normal. This is in fact an important situa on
which we explore in the next sec on, en tled Con nuity. In short, a con nuous func on is one in which when a func on approaches a value as x c (i.e.,
when lim f(x) = L), it actually a ains that value at c. Such func ons behave
xc
nicely as they are very predictable.
Notes:
31
Exercises 1.4
Terms and Concepts
y
2
1. What are the three ways in which a limit may fail to exist?
1.5
7.
2. T/F: If lim f(x) = 5, then lim f(x) = 5

x1
x1
0.5
x
.
x1
x1+
0.5
1.5
(a) lim f(x)
(d) f(1)
(b) lim f(x)
(e) lim f(x)
(c) lim f(x)
(f) lim f(x)
x1
x1
x1
x2
x1+
x0+
x1
y
2
Problems
1.5
In Exercises 5 12, evaluate each expression using the given

graph of f(x).
8.
1
0.5
x
.
2
0.5
1.5
(a) lim f(x)
(c) lim f(x)
(b) lim f(x)
(d) f(1)
x1
1.5
x1
x1+
5.
1
y
0.5
2
x
0.5
1.5
9.
(a) lim f(x)
(d) f(1)
(b) lim f(x)
(e) lim f(x)
(c) lim f(x)
(f) lim f(x)
x1
1.5
1
0.5
x0
x1+
x0+
x1
x
1
0.5
1.5
(a) lim f(x)
(c) lim f(x)
(b) lim f(x)
(d) f(1)
x1
x1
x1+
1.5
4
6.
1
2
0.5
10.
0.5
1.5
(a) lim f(x)
(d) f(1)
(b) lim f(x)
(e) lim f(x)
(c) lim f(x)
(f) lim f(x)
x1
x1+
x1
x
4 3 2 1
x2
x2+
(a) lim f(x)
(c) lim f(x)
(b) lim f(x)
(d) f(0)
x0
x0+
x0
1 cos2 x
x<a
,
sin2 x
xa
where a is a real number.
17. f(x) =
4
2
11.
x
4 3 2 1
(a)
(b)
(c) lim f(x)
(b) lim f(x)
(d) f(a)
xa
lim f(x)
(e) lim f(x)
lim f(x)
(f) lim f(x)
x2
x2
x2+
x2+
(c) lim f(x)
(g) lim f(x)
(d) f(2)
(h) f(2)
x2
x2
x+1
1
18. f(x) =
x1
(a) lim f(x)
(c) lim f(x)
(b) lim f(x)
(d) f(1)
x1
x2
x+1
19. f(x) =
x2 + 2x + 4
12.
x
4 3 2 1
(c) lim f(x)
(b) lim f(x)
(d) f(2)
x2
a(x b)2 + c
x<b
,
a(x b) + c
xb
where a, b and c are real numbers.
20. f(x) =
(a) lim f(x)
(c) lim f(x)
(b) lim f(x)
(d) f(a)
xa
xa+
(a) lim f(x)
(c) lim f(x)
x1
x1
(b) lim f(x)

x1+
{ 2
2x + 5x 1
14. f(x) =
sin x
(a) lim f(x)
x0
(e) lim f(x)
lim f(x)
(f) lim f(x)
x1
x1+
x1+
(g) lim f(x)

x1
x1
(b) lim f(x)

x +
xb
{
21. f(x) =
|x|
x
x = 0
x=0
(a) lim f(x)
(c) lim f(x)
(b) lim f(x)
(d) f(0)
x0
x0
x0+
22. Evaluate the limit: lim
x2 + 5x + 4
.
x2 3x 4
x2 16
.
x2 4x 32
x2 15x + 54
.
x2 6x
x1
x6
lim f(x)
lim f(x)
(d) f(b)
x4
x < 1
1 x 1
x>1
x1
(b) lim f(x)
Review
x<0
x0
(d) f(0)
(c) lim f(x)
(a)
(c) lim f(x)
xb
(d) f(1)
(c) lim f(x)
x0
(d) f(1)
{
cos x
16. f(x) =
sin x
(a) lim f(x)

xb+
In Exercises 13 21, evaluate the given limits of the piecewise

dened func ons f.
{
x+1
x1
13. f(x) =
x2 5
x>1
(b)
x2
x2+
Let 3 a 3 be an integer.
(a)
x<2
x=2
x>2
(a) lim f(x)
(b) lim f(x)

x0+
2
x 1
x3 + 1
15. f(x) =
2
x +1
x1
xa
x<1
x=1
x>1
x1+
xa
xa+
(a) lim f(x)
(h) f(1)
x2
x2 6x + 9
.
x2 3x
26. Approximate the limit numerically: lim
x2 4.4x + 1.6
.
x2 0.4x
x2 + 5.8x 1.2
.
x2 4.2x + 0.8
x0.4
x0.2
x<
x

(c) lim f(x)
x0.5
(d) f()
x0.1
x2 0.5x 0.5
.
x2 + 6.5x + 3
x2 + 0.9x 0.1
.
x2 + 7.9x 0.8
Chapter 1 Limits
1.5 Con nuity

As we have studied limits, we have gained the intui on that limits measure
where a func on is heading. That is, if lim f(x) = 3, then as x is close to 1,
x1
f(x) is close to 3. We have seen, though, that this is not necessarily a good indicator of what f(1) actually this. This can be problema c; func ons can tend
to one value but a ain another. This sec on focuses on func ons that do not
exhibit such behavior.
.
Deni on 3
Con nuous Func on
Let f be a func on dened on an open interval I containing c.
1. f is con nuous at c if lim f(x) = f(c).

xc
2. f is con nuous on I if f is con nuous at c for all values of c in I. If f

is con nuous on (, ), we say f is con nuous everywhere.
A useful way to establish whether or not a func on f is con nuous at c is to

verify the following three things:
1. lim f(x) exists,
xc
2. f(c) is dened, and

3. lim f(x) = f(c).
xc
. Example 20
Finding intervals of con nuity
Let f be dened as shown in Figure 1.25. Give the interval(s) on which f is connuous.
y
1.5
We proceed by examining the three criteria for con nuity.
1. The limits lim f(x) exists for all c between 0 and 3.

xc
0.5
x
1
.
Figure 1.25: A graph of f in Example 20.
2. f(c) is dened for all c between 0 and 3, except for c = 1. We know

immediately that f cannot be con nuous at x = 1.
3. The limit lim f(x) = f(c) for all c between 0 and 3, except, of course, for
xc
c = 1.
We conclude that f is con nuous at every point of (0, 3) except at x = 1.
Therefore f is con nuous on (0, 1) and (1, 3)...
Notes:
34
1.5 Con nuity

. Example 21
Finding intervals of con nuity
The oor func on, f(x) = x, returns the largest integer smaller than the input
x. (For example, f() = = 3.) The graph of f in Figure 1.26 demonstrates
why this is o en called a step func on.
Give the intervals on which f is con nuous.
We examine the three criteria for con nuity.
y
2
1. The limits limxc f(x) do not exist at the jumps from one step to the
next, which occur at all integer values of c. Therefore the limits exist for
all c except when c is an integer.
2. The func on is dened for all values of c.

3. The limit lim f(x) = f(c) for all values of c where the limit exist, since each
xc
step consists of just a line.
We conclude that f is con nuous everywhere except at integer values of c. So
the intervals on which f is con nuous are
Figure 1.26: A graph of the step func on

in Example 21.
. . . , (2, 1), (1, 0), (0, 1), (1, 2), . . . .
Our deni on of con nuity on an interval species the interval is an open

interval. We can extend the deni on of con nuity to closed intervals by considering the appropriate one-sided limits at the endpoints.
Deni on 4
Con nuity on Closed Intervals
Let f be dened on the closed interval [a, b] for some real numbers a, b.
f is con nuous on [a, b] if:
1. f is con nuous on (a, b),
2. lim f(x) = f(a) and

xa+
3. lim f(x) = f(b).

xb
We can make the appropriate adjustments to talk about con nuity on half
open intervals such as [a, b) or (a, b] if necessary.
Notes:
35
Chapter 1 Limits
. Example 22
Determining intervals on which a func on is con nuous
For each of the following func ons, give the domain of the func on and the
interval(s) on which it is con nuous.
1 x2
1. f(x) = 1/x
4. f(x) =
2. f(x) = sin x
3. f(x) = x
5. f(x) = |x|
We examine each in turn.
1. The domain of f(x) = 1/x is (, 0) (0, ). As it is a ra onal func on,

we apply Theorem 2 to recognize that f is con nuous on all of its domain.
2. The domain of f(x) = sin x is all real numbers, or (, ). Applying
Theorem 3 shows that sin x is con nuous everywhere.
3. The
domain of f(x) = x is [0, ). Applying Theorem 3 shows that f(x) =
x is con nuous on its domain of [0, ).
4. The domain of f(x) = 1 x2 is [1, 1]. Applying Theorems 1 and 3

shows that f is con nuous on all of its domain, [1, 1].
5. The domain of f(x) =
{ |x| is (, ). We can dene the absolute value
x x < 0
func on as f(x) =
. Each piece of this piece-wise dex x0
ned func on is con nuous on all of its domain, giving that f is con nuous
on (, 0) and [0, ). As we saw before, we cannot assume this implies
that f is con nuous on (, ); we need to check that lim f(x) = f(0),
x0
as x = 0 is the point where f transi ons from one piece of its deni on
to the other. It is easy to verify that this is indeed true, hence we conclude
that f(x) = |x| is con nuous everywhere.
.
Con nuity is inherently ed to the proper es of limits. Because of this, the
proper es of limits found in Theorems 1 and 2 apply to con nuity as well. Further, now knowing the deni on of con nuity we can reread Theorem 3 as
giving a list of func ons that are con nuous on their domains. The following
theorem states how con nuous func ons can be combined to form other connuous func ons, followed by a theorem which formally lists func ons that we
know are con nuous on their domains.
Notes:
36
1.5 Con nuity
Theorem 8
Proper es of Con nuous Func ons
Let f and g be con nuous func ons on an interval I, let c be a real number
and let n be a posi ve integer. The following func ons are con nuous on
I.
1. Sums/Dierences: f g
2. Constant Mul ples: c f
3. Products:
fg
4. Quo ents:
f/g
5. Powers:
fn
n
f
6. Roots:
7. Composi ons:
Theorem 9
(as longs as g = 0 on I)
(if n is even then f 0 on I; if n is odd,

then true for all values of f on I.)
Adjust the deni ons of f and g to: Let f be

con nuous on I, where the range of f on I is J,
and let g be con nuous on J. Then g f, i.e.,
g(f(x)), is con nuous on I.
Con nuous Func ons
The following func ons are con nuous on their domains.

1. f(x) = sin x
3. f(x) = tan x
5. f(x) = sec x
7. f(x) = ln x
9. f(x) = ax (a > 0)
2. f(x) = cos x
4. f(x) = cot x
6. f(x) = csc x
8. f(x) = n x,
(where n is a posi ve integer)
We apply these theorems in the following Example.
Notes:
37
Chapter 1 Limits
. Example 23
Determining intervals on which a func on is con nuous
State the interval(s) on which each of the following func ons is con nuous.
3. f(x) = tan x
1. f(x) = x 1 + 5 x
4. f(x) = ln x
2. f(x) = x sin x
We examine each in turn, applying Theorems 8 and 9 as ap-
propriate.
y
3
1. The squareroot terms are con nuous on the intervals [1, ) and (, 5],
respec vely. As f is con nuous only where each term is con nuous, f is
con nuous on [1, 5], the intersec on of these two intervals. A graph of f
is given in Figure 1.27.
2. The func ons y = x and y = sin x are each con nuous everywhere, hence
their product is, too.
x
3. Theorem 9 states that f(x) = tan x is con nuous on its domain. Its domain includes all real numbers except odd mul ples of /2. Thus f(x) =
tan x is con nuous on
(
) (
)
(
3
)
3
... ,
, ,
,
,
,...,
2
2
2 2
2 2
Figure 1.27: A graph of f in Example 23(a).
or, equivalently, on D = {x R | x = n 2 , n is an odd integer}.
4. The domain of y = x is [0, ). The range of y = ln x is (, ), but if

we restrict its domain to [1, ) its range is [0, ). So restric ng y =ln x
to the domain of [1, ) restricts its
output is [0, ), on which y = x is
dened. Thus the domain of f(x) = ln x is [1, ).
A common way of thinking of a con nuous func on is that its graph can
be sketched without li ing your pencil. That is, its graph forms a con nuous
curve, without holes, breaks or jumps. While beyond the scope of this text,
this pseudodeni on glosses over some of the ner points of con nuity. Very
strange func ons are con nuous that one would be hard pressed to actually
sketch by hand.
This intui ve no on of con nuity does help us understand another important concept as follows. Suppose f is dened on [1, 2] and f(1) = 10 and
f(2) = 5. If f is con nuous on [1, 2] (i.e., its graph can be sketched as a con nuous line from (1, 10) to (2, 5)) then we know intui vely that somewhere on
[1, 2] f must be equal to 9, and 8, and 7, 6, . . . , 0, 1/2, etc. In short, f
Notes:
38
1.5 Con nuity

takes on all intermediate values between 10 and 5. It may take on more values; f may actually equal 6 at some me, for instance, but we are guaranteed all
values between 10 and 5.
While this no on seems intui ve, it is not trivial to prove and its importance
is profound. Therefore the concept is stated in the form of a theorem.
Theorem 10
Intermediate Value Theorem
Let f be a con nuous func on on [a,

b] and, without loss of generality,
.
let f(a) < f(b). Then for every value y, where f(a) < y < f(b), there is a
value c in [a, b] such that f(c) = y.
One important applica on of the Intermediate Value Theorem is root nding. Given a func on f, we are o en interested in nding values of x where
f(x) = 0. These roots may be very dicult to nd exactly. Good approxima ons
can be found through successive applica ons of this theorem. Suppose through
direct computa on we nd that f(a) < 0 and f(b) > 0, where a < b. The Intermediate Value Theorem states that there is a c in [a, b] such that f(c) = 0. The
theorem does not give us any clue as to where that value is in the interval [a, b],
just that it exists.
There is a technique that produces a good approxima on of c. Let d be the
midpoint of the interval [a, b] and consider f(d). There are three possibili es:
1. f(d) = 0 we got lucky and stumbled on the actual value. We stop as we
found a root.
2. f(d) < 0 Then we know there is a root of f on the interval [d, b] we have
halved the size of our interval, hence are closer to a good approxima on
of the root.
3. f(d) > 0 Then we know there is a root of f on the interval [a, d] again,we
have halved the size of our interval, hence are closer to a good approxima on of the root.
Successively applying this technique is called the Bisec on Method of root
nding. We con nue un l the interval is suciently small. We demonstrate this
in the following example.
. Example 24
.Using the Bisec on Method
Approximate the root of f(x) = x cos x, accurate to three places a er the
decimal.
Consider the graph of f(x) = xcos x, shown in Figure 1.28.
S
It is clear that the graph crosses the x-axis somewhere near x = 0.8. To start the
Notes:
39
Chapter 1 Limits
Bisec on Method, pick an interval that contains 0.8. We choose [0.7, 0.9]. Note
that all we care about are signs of f(x), not their actual value, so this is all we
display.
y
0.5
x
0.5
0.5
Figure 1.28: Graphing a root of f(x) = x

cos x.
Itera on 1: f(0.7) < 0, f(0.9) > 0, and f(0.8) > 0. So replace 0.9 with 0.8 and
repeat.
Itera on 2: f(0.7) < 0, f(0.8) > 0, and at the midpoint, 0.75, we have f(0.75) >
0. So replace 0.8 with 0.75 and repeat. Note that we dont need to connue to check the endpoints, just the midpoint. Thus we put the rest of
the itera ons in Table 1.29.
No ce that in the 12th itera on we have the endpoints of the interval each
star ng with 0.739. Thus we have narrowed the zero down to an accuracy of
the rst three places a er the decimal. Using a computer, we have
f(0.7390) = 0.00014,
Itera on #
1
2
3
4
5
6
7
8
9
10
11
12
Interval
[0.7, 0.9]
[0.7, 0.8]
[0.7, 0.75]
[0.725, 0.75]
[0.7375, 0.75]
[0.7375, 0.7438]
[0.7375, 0.7407]
[0.7375, 0.7391]
[0.7383, 0.7391]
[0.7387, 0.7391]
[0.7389, 0.7391]
[0.7390, 0.7391]
Midpoint Sign
f(0.8) > 0
f(0.75) > 0
f(0.725) < 0
f(0.7375) < 0
f(0.7438) > 0
f(0.7407) > 0
f(0.7391) > 0
f(0.7383) < 0
f(0.7387) < 0
f(0.7389) < 0
f(0.7390) < 0
Figure 1.29: Itera ons of the Bisec on

Method of Root Finding
Either endpoint of the interval gives a good approxima on of where f is 0. The

Intermediate Value Theorem states that the actual zero is s ll within this interval.
While we do not know its exact value, we know it starts with 0.739.
This type of exercise is rarely done by hand. Rather, it is simple to program
a computer to run such an algorithm and stop when the endpoints dier by a
preset small amount. One of the authors did write such a program and found
the zero of f, accurate to 10 places a er the decimal, to be 0.7390851332. While
it took a few minutes to write the program, it took less than a thousandth of a
second for the program to run the necessary 35 itera ons. In less than 8 hundredths of a second, the zero was calculated to 100 decimal places (with less
than 200 itera ons). .
It is a simple ma er to extend the Bisec on Method to solve similar problems
to f(x) = 0. For instance, we can solve f(x) = 1. This may seem obvious, but
to many it is not. It actually works very well to dene a new func on g where
g(x) = f(x) 1. Then use the Bisec on Method to solve g(x) = 0.
Similarly, given two func ons f and g, we can use the Bisec on Method to
solve f(x) = g(x). Once again, create a new func on h where h(x) = f(x) g(x)
and solve h(x) = 0.
In Sec on 4.1 another equa on solving method will be introduced, called
Newtons Method. In many cases, Newtons Method is much faster. It relies on
more advanced mathema cs, though, so we will wait before introducing it.
This sec on formally dened what it means to be a con nuous func on.
Most func ons that we deal with are con nuous, so o en it feels odd to have
to formally dene this concept. Regardless, it is important, and forms the basis
of the next chapter.
In the next sec on we examine one more aspect of limits: limits that involve
innity.
Notes:
40
f(0.7391) = 0.000024.
Exercises 1.5
Terms and Concepts
y
2
1. In your own words, describe what it means for a func on

to be con nuous.
1.5
1
2. In your own words, describe what the Intermediate Value

Theorem states.
0.5
3. What is a root of a func on?
4. Given func ons f and g on an interval I, how can the Bisecon Method be used to nd a value c where f(c) = g(c)?
0.5
1.5
0.5
1.5
0.5
1.5
14. a = 0
y
2
5. T/F: If f is dened on an open interval containing c, and

lim f(x) exists, then f is con nuous at c.
1.5
xc
6. T/F: If f is con nuous at c, then lim f(x) exists.
xc
0.5
7. T/F: If f is con nuous at c, then lim f(x) = f(c).

xc+
8. T/F: If f is con nuous on [a, b], then lim f(x) = f(a).

xa
9. T/F: If f is con nuous on [0, 1) and [1, 2), then f is con nuous on [0, 2).
15. a = 1
y
2
10. T/F: The sum of con nuous func ons is also con nuous.
1.5
1
Problems
0.5
In Exercises 11 17, a graph of a func on f is given along with

a value a. Determine if f is con nuous at a; if it is not, state
why it is not.
16. a = 4
11. a = 1
2
2
1.5
x
4 3 2 1
2
0.5
0.5
1.5
(a) a = 2
17.
12. a = 1
(b) a = 0
(c) a = 2
y
1.5
2
x
0.5
13. a = 1
4 3 2 1
2
x
0.5
1.5
In Exercises 18 21, determine if f is con nuous at the indicated values. If not, explain why.
{
1
x=0
18. f(x) =
sin x
x>0
x
(a) x = 0
(b) x =
{ 3
x x
19. f(x) =
x2
x<1
x1
(a) x = 0
(b) x = 1
{ 2
x +5x+4
x2 +3x+2
20. f(x) =
x = 1
x = 1
(b) x = 10
{
2
x 64
x2 11x+24
x = 8
x=8
(a) x = 0
(b) x = 8
In Exercises 22 32, give the intervals on which the given
func on is con nuous.
22. f(x) = x2 3x + 9
23. g(x) = x2 4
24. h(k) = 1 k + k + 1
25. f(t) = 5t2 30

26. g(t) =
27. g(x) =
33. Let f be con nuous on [1, 5] where f(1) = 2 and f(5) =

10. Does a value 1 < c < 5 exist such that f(c) = 9?
Why/why not?
34. Let g be con nuous on [3, 7] where g(0) = 0 and g(2) =
25. Does a value 3 < c < 7 exist such that g(c) = 15?
Why/why not?
35. Let f be con nuous on [1, 1] where f(1) = 10 and
f(1) = 10. Does a value 1 < c < 1 exist such that
f(c) = 11? Why/why not?
36. Let h be a func on on [1, 1] where h(1) = 10 and
h(1) = 10. Does a value 1 < c < 1 exist such that
h(c) = 0? Why/why not?
In Exercises 37 40, use the Bisec on Method to approximate, accurate to two decimal places, the value of the root
of the given func on in the given interval.
(a) x = 1
21. f(x) =
32. f(x) = sin(ex + x2 )
1
1 t2
1
1 + x2
37. f(x) = x2 + 2x 4 on [1, 1.5].

38. f(x) = sin x 1/2 on [0.5, 0.55]
39. f(x) = ex 2 on [0.65, 0.7].
40. f(x) = cos x sin x on [0.7, 0.8].
Review
{
41. Let f(x) =
30. h(t) = cos t
31. f(k) = 1 ek
x<5
.
x5
(a) lim f(x)
(c) lim f(x)
(b) lim f(x)
(d) f(5)
x5
x5
x5+
42. Numerically approximate the following limits:

(a)
28. f(x) = ex
29. g(s) = ln s
x2 5
5x
(b)
lim
x2 8.2x 7.2
x2 + 5.8x + 4
lim
x2 8.2x 7.2
x2 + 5.8x + 4
x4/5+
x4/5
43. Give an example of func on f(x) for which lim f(x) does not
x0
exist.
1.6 Limits involving innity
1.6
Limits involving innity
y
100
In Deni on 1 we stated that in the equa on lim f(x) = L, both c and L were
xc
numbers. In this sec on we relax that deni on a bit by considering situa ons
when it makes sense to let c and/or L be innity.
As a mo va ng example, consider f(x) = 1/x2 , as shown in Figure 1.30.
Note how, as x approaches 0, f(x) grows very, very large. It seems appropriate,
and descrip ve, to state that
lim
x0
1
= .
x2
Also note that as x gets very large, f(x) gets very, very small. We could represent
this concept with nota on such as
lim
50
0.5
0.5
Figure 1.30: Graphing f(x) = 1/x2 for values of x near 0.
1
= 0.
x2
We explore both types of use of in turn.

Deni on 5
Limit of Innity,
We say lim f(x) = if for every M.

> 0 there exists > 0 such that if
xc
0 < |x c| < then f(x) M.
This is just like the deni on from Sec on 1.2. In that deni on, given
any (small) value , if we let x get close enough to c (within units of c) then f(x)
is guaranteed to be within of f(c). Here, given any (large) value M, if we let x
get close enough to c (within units of c), then f(x) will be at least as large as
M. In other words, if we get close enough to c, then we can make f(x) as large
as we want. We can dene limits equal to in a similar way.
It is important to note that by saying lim f(x) = we are implicitly sta ng
xc
that the limit of f(x), as x approaches c, does not exist. A limit only exists when
f(x) approaches an actual numeric value. We use the concept of limits that approach innity because they are helpful and descrip ve.
. Example 25
Find lim
x1
y
100
50
.
.Evalua ng limits involving innity
1
as shown in Figure 1.31.
(x 1)2
x
0.5
1.5
Figure 1.31: Observing innite limit as

x 1 in Example 25.
In Example 4 of Sec on 1.1, by inspec ng values of x close

S
to 1 we concluded that this limit does not exist. That is, it cannot equal any real
Notes:
43
Chapter 1 Limits
number. But the limit could be innite. And in fact, we see that the func on
does appear to be growing larger and larger, as f(.99) = 104 , f(.999) = 106 ,
f(.9999) = 108 . A similar thing happens onthe other side of 1. In general,
let a large
value M be given. Let = 1/ M. If x is within of 1, i.e., if
|x 1| < 1/ M, then:
1
|x 1| <
M
1
(x 1)2 <
M
1
> M,
(x 1)2
y
50
which is what we wanted to show. So we may say lim 1/(x 1)2 = . .

x1
x
1
0.5
0.5
It is easy to see that the func on grows without bound near

S
0, but it does so in dierent ways on dierent sides of 0. Since its behavior is not
1
consistent, we cannot say that lim = . However, we can make a statement
x0 x
1
1
about onesided limits. We can state that lim
= and lim
= .
+
x0 x
x0 x
.
50
Figure 1.32: Evalua ng lim
x0
. Example 26
Evalua ng limits involving innity
1
Find lim , as shown in Figure 1.32.
x0 x
1
.
x
Ver cal asymptotes

If the limit of f(x) as x approaches c from either the le or right (or both) is
or , we say the func on has a ver cal asymptote at c.
. Example 27
10
x
5
5
10
.
Figure 1.33: Graphing f(x) =
3x
.
x2 4
Finding ver cal asymptotes

3x
Find the ver cal asymptotes of f(x) = 2
.
x 4
Ver cal asymptotes occur where the func on grows without

S
bound; this occurs at values of c where the denominator is 0. The denominator
is small near x = c, which in turn can make the func on overall take on large
values. In the case of the given func on, the denominator is 0 at x = 2. Subs tu ng in values of x close to 2 and 2 seems to indicate that the func on tends
toward or at those points. We can graphically conrm this by looking at
Figure 1.33.
.
Notes:
44

y
When a func on has a ver cal asymptote, we can conclude that the denominator is 0 for some part of that func on. However, just because the denominator is 0 at a certain point does not mean there is a ver cal asymptote
there. For instance, f(x) = (x2 1)/(x 1) does not have a ver cal asymptote
at x = 1, as shown in Figure 1.34. While the denominator does get small near
x = 1, the numerator gets small too, matching the denominator step for step.
In fact, factoring the numerator, we get
f(x) =
(x 1)(x + 1)
.
x1
Canceling the common term, we get that f(x) = x + 1 for x = 1. So there is

clearly no asymptote, rather a hole exists in the graph at x = 1.
The above example may seem a li le contrived. Another example demonstra ng this important concept is f(x) = (sin x)/x. We have considered this
sin x
= 1;
func on several mes in the previous sec ons. We found that lim
x0 x
i.e., there is no ver cal asymptote. No simple algebraic cancella on makes this
fact obvious; we used the Squeeze Theorem in Sec on 1.3 to prove this.
.1
x
1
Figure 1.34: Graphically showing that

x2 1
f(x) =
does not have an asympx1
tote at x = 1.
If the denominator is 0 at a certain point but the numerator is not, then

there will usually be a ver cal asymptote at that point. On the other hand, if the
numerator and denominator are both zero at that point, then there may or may
not be a ver cal asymptote at that point. This case where the numerator and
denominator are both zero returns us to an important topic.
Indeterminate Forms
We have seen how the limits
sin x
x0 x
lim
and
x2 1
x1 x 1
lim
0
when we blindly plug in x = 0 and
0
x = 1, respec vely. However, 0/0 is not a valid arithme cal expression. It gives
no indica on that the respec ve limits are 1 and 2.
With a li le cleverness, one can come up 0/0 expressions which have a limit
of , 0, or any other real number. That is why this expression is called indeterminate.
A key concept to understand is that such limits do not really return 0/0.
Rather, keep in mind that we are taking limits. What is really happening is that
the numerator is shrinking to 0 while the denominator is also shrinking to 0.
each return the indeterminate form
Notes:
45
Chapter 1 Limits
The respec ve rates at which they do this are very important and determine the
actual value of the limit.
An indeterminate form indicates that one needs to do more work in order to
compute the limit. That work may be algebraic (such as factoring and canceling)
or it may require a tool such as the Squeeze Theorem. In a later sec on we will
learn a technique called lHospitals Rule that provides another way to handle
indeterminate forms.
Some other common indeterminate forms are , 0, /, 00 , 0
and 1 . Again, keep in mind that these are the blind results of evalua ng a
limit, and each, in and of itself, has no meaning. The expression does
not really mean subtract innity from innity. Rather, it means One quan ty
is subtracted from the other, but both are growing without bound. What is the
result? It is possible to get every value between and
Note that 1/0 and /0 are not indeterminate forms, though they are not
exactly valid mathema cal expressions, either. Rather they indicate that the
limit will be , , or not exist.
Limits at Innity and Horizontal Asymptotes
At the beginning of this sec on we briey considered what happens to f(x) =
1/x2 as x grew very large. Graphically, it concerns the behavior of the func on to
the far right of the graph. We make this no on more explicit in the following
deni on.
.
Deni on 6
Limits at Innity and Horizontal Asymptote
1. We say lim f(x) = L if for every > 0 there exists M > 0 such
x
that if x M, then |f(x) L| < .
2. We say lim f(x) = L if for every > 0 there exists M < 0 such
x
that if x M, then |f(x) L| < .

3. If lim f(x) = L or lim f(x) = L, we say that y = L is a horizontal
x
asymptote of f.
We can also dene limits such as lim f(x) = by combining this deni on
x
with Deni on 5.
Notes:
46

. Example 28
Approxima ng horizontal asymptotes

x2
Approximate the horizontal asymptote(s) of f(x) = 2
.
x +4
We will approximate the horizontal asymptotes by approxi-
0.5
ma ng the limits
x2
x x2 + 4
lim
x2
.
x x2 + 4
and
lim
x
20
Figure 1.36(a) shows a sketch of f, and part (b) gives values of f(x) for large magnitude values of x. It seems reasonable to conclude from both of these sources
that f has a horizontal asymptote at y = 1.
Later, we will show how to determine this analy cally...
0.5
0.5
10
10
20
0.5
Figure 1.36: Using a graph and a table

to approximate a horizontal asymptote in
Example 28.
0.5
x
20
0.9615
0.9996
0.999996
0.9615
0.9996
0.999996
(b)
x
20
10
10
20
0.5
.
(a)
20
f(x)
10
100
10000
10
100
10000
10
(a)
Horizontal asymptotes can take on a variety of forms. Figure 1.35(a) shows

that f(x) = x/(x2 + 1) has a horizontal asymptote of y = 0, where 0 is approached from both above and below.
Figure 1.35(b) shows that f(x) = x/ x2 + 1 has two horizontal asymptotes;
one at y = 1 and the other at y = 1.
Figure 1.35(c) shows that f(x) = (sin x)/x has even more interes ng behavior than at just x = 0; as x approaches , f(x) approaches 0, but oscillates as
it does this.
10
x
20
10
10
20
.
(b)
(c)
Figure 1.35: Considering dierent types of horizontal asymptotes.
We can analy cally evaluate limits at innity for ra onal func ons once we
understand lim 1/x. As x gets larger and larger, the 1/x gets smaller and smaller,
x
approaching 0. We can, in fact, make 1/x as small as we want by choosing a large
Notes:
47
Chapter 1 Limits
enough value of x. Given , we can make 1/x < by choosing x > 1/. Thus
we have limx 1/x = 0.
It is now not much of a jump to conclude the following:
1
1
lim
= 0 and
lim
=0
x xn
x xn
Now suppose we need to compute the following limit:
x3 + 2x + 1
.
x 4x3 2x2 + 9
lim
The trick to doing this is to divide through the numerator and denominator by x3
(hence dividing by 1), which is the largest power of x to appear in the func on.
Doing this, we get
x3 + 2x + 1
x3 + 2x + 1 1/x3
=
lim
x 4x3 2x2 + 9
x 4x3 2x2 + 9 1/x3
x3 /x3 + 2x/x3 + 1/x3
= lim
x 4x3 /x3 2x2 /x3 + 9/x3
1 + 2/x2 + 1/x3
= lim
.
x 4 2/x + 9/x3
lim
Then using the rules for limits (which also hold for limits at innity), as well as
the fact about limits of 1/xn , we see that the limit becomes
1+0+0
1
= .
40+0
4
This procedure works for any ra onal func on. In fact, it gives us the following theorem.
.
Theorem 11
Limits of Ra onal Func ons at Innity
Suppose we have a ra onal func on of the following form:

f(x) =
an xn + an1 xn1 + + a1 x + a0
,
bm xm + bm1 xm1 + + b1 x + b0
where any of the coecients may be 0 except for an and bm .

1. If n = m, then limx f(x) = limx f(x) =
an
bm .
2. If n < m, then limx f(x) = limx f(x) = 0.

3. If n > m, then limx f(x) and limx f(x) are both innite.
Notes:
48

We can see why this is true. If the highest power of x is the same in both
the numerator and denominator (i.e. n = m), we will be in a situa on like the
example above, where we will divide by xn and in the limit all the terms will
approach 0 except for an xn /xn and bm xn /xm . Since n = m, this will leave us with
the limit an /bm . If n < m, then a er dividing through by xm , all the terms in the
numerator will approach 0 in the limit, leaving us with 0/bm or 0. If n > m, and
we try dividing through by xn , we end up with all the terms in the denominator
tending toward 0, while the xn term in the numerator does not approach 0. This
is indica ve of some sort of innite limit.
Intui vely, as x gets very large, all the terms in the numerator are small in
comparison to an xn , and likewise all the terms in the denominator are small
compared to bn xm . If n = m, looking only at these two important terms, we
have (an xn )/(bn xm ). This reduces to an /bm . If n < m, the func on behaves
like an /(bm xmn ), which tends toward 0. If n > m, the func on behaves like
an xnm /bm , which will tend to either or depending on the values of n,
m, an , bm and whether you are looking for limx f(x) or limx f(x).
With care, we can quickly evaluate limits at innity for a large number of
func ons by considering the largest powers of x. For instance, consider again
x
lim
. When x is very large, x2 + 1 x2 . Thus
x
x2 + 1
x
x
x2 + 1 x2 = |x|, and
.
2
|x|
x +1
This expression is 1 when x is posi ve and 1 when x is nega ve. Hence we get
asymptotes of y = 1 and y = 1, respec vely.
. Example 29
.Finding a limit of a ra onal func on
Conrm analy cally that y = 1 is the horizontal asymptote of f(x) =

approximated in Example 28.
x2
x2
, as
+4
Before using Theorem 11, lets use the technique of evaluS

a ng limits at innity of ra onal func ons that led to that theorem. The largest
power of x in f is 2, so divide the numerator and denominator of f by x2 , then
take limits.
x2
x2 /x2
lim 2
= lim 2 2
x x + 4
x x /x + 4/x2
1
= lim
x 1 + 4/x2
1
=
1+0
= 1.
Notes:
49
Chapter 1 Limits
We can also use Theorem 11 directly; in this case n = m so the limit is the
ra o of the leading coecients of the numerator and denominator, i.e., 1/1 = 1. .
y
0.5
. Example 30
Finding limits of ra onal func ons
Use Theorem 11 to evaluate each of the following limits.
x
40
30
20
10
1.
x2 1
x 3 x
3. lim
x2 + 2x 1
x 1 x 3x2
0.5
x2 + 2x 1
x
x3 + 1
lim
2. lim
(a)
S
y
1. The highest power of x is in the denominator. Therefore, the limit is 0; see

Figure 1.37(a).
0.5
2. The highest power of x is x2 , which occurs in both the numerator and denominator. The limit is therefore the ra o of the coecients of x2 , which
is 1/3. See Figure 1.37(b).
x
10
0.5
20
30
40
.
(b)
3. The highest power of x is in the numerator so the limit will be or .

To see which, consider only the dominant terms from the numerator and
denominator, which are x2 and x. The expression in the limit will behave
like x2 /(x) = x for large values of x. Therefore, the limit is . See
Figure 1.37(c).
x
20
40
Chapter Summary
20
In this chapter we:

dened the limit,
40
found accessible ways to approximate their values numerically and graphically,
.
(c)
Figure 1.37: Visualizing the func ons in
Example 30.
developed a notsoeasy method of proving the value of a limit (

proofs),
explored when limits do not exist,
dened con nuity and explored proper es of con nuous func ons, and
Notes:
50

considered limits that involved innity.
Why? Mathema cs is famous for building on itself and calculus proves to be
no excep on. In the next chapter we will be interested in dividing by 0. That
is, we will want to divide a quan ty by a smaller and smaller number and see
what value the quo ent approaches. In other words, we will want to nd a limit.
These limits will enable us to, among other things, determine exactly how fast
something is moving when we are only given posi on informa on.
Later, we will want to add up an innite list of numbers. We will do so by rst
adding up a nite string of numbers, then take a limit as the number of things
we are adding approaches innity. Surprisingly, this sum o en is nite; that is,
we can add up an innite list of numbers and get, for instance, 42.
These are just two quick examples of why we are interested in limits. Many
students dislike this topic when they are rst introduced to it, but over me an
apprecia on is o en formed based on the scope of its applicability.
Notes:
51
Exercises 1.6
Terms and Concepts
10. f(x) =
1. T/F: If lim f(x) = , then we are implicitly sta ng that the

x5
limit exists.
1
.
(x 3)(x 5)2
(a) lim f(x)
(d) lim f(x)
(b) lim f(x)
(e) lim f(x)
(c) lim f(x)
(f) lim f(x)
x3
x5
x3+
2. T/F: If lim f(x) = 5, then we are implicitly sta ng that the

x
x5+
x3
limit exists.
x5
y
50
3. T/F: If lim f(x) = , then lim f(x) =

x1
x1+
4. T/F: If lim f(x) = , then f has a ver cal asymptote at

x5
x = 5.
50
5. T/F: /0 is not an indeterminate form.

11. f(x) =
6. List 5 indeterminate forms.
(a)
7. Construct a func on with a ver cal asymptote at x = 5 and

a horizontal asymptote at y = 5.
1
ex + 1
lim f(x)
(c) lim f(x)

x0
(b) lim f(x)
(d) lim f(x)
x0+
8. Let lim f(x) = . Explain how we know that f is/is not

x7
con nuous at x = 7.
1
0.5
x
10
Problems
10
10
0.5
In Exercises 9 14, evaluate the given limits using the graph

of the func on.
12. f(x) = x2 sin(x)

9. f(x) =
(a)
(b)
1
(x + 1)2
(a)
lim f(x)
lim f(x)
x1
(b) lim f(x)
lim f(x)
x1+
100
100
50
50
x
10
5
50
x
2
..
100
13. f(x) = cos(x)

(a)
In Exercises 19 24, iden fy the horizontal and ver cal

asymptotes, if any, of the given func on.
lim f(x)
19. f(x) =
2x2 2x 4
x2 + x 20
20. f(x) =
3x2 9x 6
5x2 10x 15
21. f(x) =
x2 + x 12
7x3 14x2 21x
22. f(x) =
x2 9
9x 9
23. f(x) =
x2 9
9x + 27
24. f(x) =
x2 1
x2 1
(b) lim f(x)

x
y
1
0.5
x
4 3 2
2 3
0.5
1
14. f(x) = 2x + 10
(a)
lim f(x)
(b) lim f(x)

x
25. lim
x3 + 2x2 + 1
x5
26. lim
x3 + 2x2 + 1
5x
150
100
27.
lim
x3 + 2x2 + 1
x2 5
lim
x3 + 2x2 + 1
5 x2
50
28.
.
x
5
10
In Exercises 15 18, numerically approximate the following

limits:
(a) lim f(x)
x3
(b) lim f(x)

x3+
(c) lim f(x)

x3
15. f(x) =
16. f(x) =
x2
x2 1
x6
x2 + 5x 36
3
x 5x2 + 3x + 9
x2 11x + 30
17. f(x) = 3
x 4x2 3x + 18
18. f(x) =
x2 9x + 18
x2 x 6
Review
29. Use an proof to show that
lim 5x 2 = 3.
x1
30. Let lim f(x) = 3 and lim g(x) = 1. Evaluate the following
x2
x2
limits.
(a) lim (f + g)(x)
(a) lim (f/g)(x)
(b) lim (fg)(x)
(b) lim f(x)g(x)
x2
x2
x2
x2
x2 1
x<3
.
x+5
x3
Is f con nuous everywhere?
31. Let f(x) =
32. Evaluate the limit: lim ln x.

xe
2: D
The previous chapter introduced the most fundamental of calculus topics: the
limit. This chapter introduces the second most fundamental of calculus topics:
the deriva ve. Limits describe where a func on is going; deriva ves describe
how fast the func on is going.
2.1
Instantaneous Rates of Change: The Deriva ve
A common amusement park ride li s riders to a height then allows them to

freefall a certain distance before safely stopping them. Suppose such a ride
drops riders from a height of 150 feet. Student of physics may recall that the
height (in feet) of the riders, t seconds a er freefall (and ignoring air resistance,
etc.) can be accurately modeled by f(t) = 16t2 + 150.
Using this formula, it is easy
to verify that, without interven on, the riders
will hit the ground at t = 2.5 1.5 3.06 seconds. Suppose the designers of
the ride decide to begin slow the riders fall a er 2 seconds (corresponding to a
height of 86 .). How fast will the riders be traveling at that me?
We have been given a posi on func on, but what we want to compute is a
velocity at a specic point in me, i.e., we want an instantaneous velocity. We
do not currently know how to calculate this.
However, we do know from common experience how to calculate an average
velocity. (If we travel 60 miles in 2 hours, we know we had an average velocity
of 30 mph.) We looked at this concept in Sec on 1.1 when we introduced the
dierence quo ent. We have
change in distance
rise
=
= average velocity.
change in me
run
We can approximate the instantaneous velocity at t = 2 by considering the
average velocity over some me period containing t = 2. If we make the me
interval small, we will get a good approxima on. (This fact is commonly used.
For instance, high speed cameras are used to track fast moving objects. Distances are measured over a xed number of frames to generate an accurate
approxima on of the velocity.)
Consider the interval from t = 2 to t = 3 (just before the riders hit the
Chapter 2 Deriva ves

ground). On that interval, the average velocity is
f(3) f(2)
f(3) f(3)
=
= 80 /s,
32
1
where the minus sign indicates that the riders are moving down. By narrowing
the interval we consider, we will likely get a be er approxima on of the instantaneous velocity. On [2, 2.5] we have
f(2.5) f(2)
f(2.5) f(2)
=
= 72 /s.
2.5 2
0.5
We can do this for smaller and smaller intervals of me. For instance, over
a me span of 1/10th of a second, i.e., on [2, 2.1], we have
f(2.1) f(2)
f(2.1) f(2)
=
= 65.6 /s.
2.1 2
0.1
Over a me span of 1/100th of a second, on [2, 2.01], the average velocity is
f(2.01) f(2)
f(2.01) f(2)
=
= 64.16 /s.
2.01 2
0.01
Average Velocity
/s
1
0.5
0.1
0.01
0.001
80
72
65.6
64.16
64.016
Figure 2.1: Approxima ng the instantaneous velocity with average veloci es

over a small me period h.
What we are really compu ng is the average velocity on the interval [2, 2+h]
for small values of h. That is, we are compu ng
f(2 + h) f(2)
h
where h is small.
What we really want is for h = 0, but this, of course, returns the familiar
0/0 indeterminate form. So we employ a limit, as we did in Sec on 1.1.
We can approximate the value of this limit numerically with small values of
h as seen in Figure 2.1. It looks as though the velocity is approaching 64 /s.
Compu ng the limit directly gives
f(2 + h) f(2)
16(2 + h)2 + 150 (16(2)2 + 150)
= lim
h0
h0
h
h
2
64h 16h
= lim
h0
h
= lim 64 16h
lim
h0
= 64.
Graphically, we can view the average veloci es we computed numerically as
the slopes of secant lines on the graph of f going through the points (2, f(2)) and
Notes:
56

(2+h, f(2+h)). In Figure 2.2, the secant line corresponding to h = 1 is shown in
three contexts. Figure 2.2(a) shows a zoomed out version of f with its secant
line. In (b), we zoom in around the points of intersec on between f and the
secant line. No ce how well this secant line approximates f between those two
points it is a common prac ce to approximate func ons with straight lines.
As h 0, these secant lines approach the tangent line, a line that goes
through the point (2, f(2)) with the special slope of 64. In parts (c) and (d) of
Figure 2.2, we zoom in around the point (2, 86). In (c) we see the secant line,
which approximates f well, but not as well the tangent line shown in (d).
150
100
100
50
50
x
1
.
50
(a)
(b)
.
100
50
50
x
1.5
2.5
(d)
2.5
100
(c)
x
2
x
1.5
2.5
Figure 2.2: Parts (a), (b) and (c) show the secant line to f(x) with h = 1, zoomed in
dierent amounts. Part (d) shows the tangent line to f at x = 2.
We have just introduced a number of important concepts that we will esh

out more within this sec on. First, we formally dene two of them.
Notes:
57

.
Deni on 7
Deriva ve at a Point
Let f be a con nuous func on on an open interval I and let c be in I. The

deriva ve of f at c, denoted f (c), is
lim
h0
f(c + h) f(c)
,
h
provided the limit exists. If the limit exists, we say that f is dieren able
at c; if the limit does not exist, then f is not dieren able at c. If f is
dieren able at every point in I, then f is dieren able on I.
Deni on 8
Tangent Line
Let f be con nuous on an open interval I and dieren able at c, for some
c in I. The line with equa on (x) = f (c)(x c) + f(c) is the tangent line
to the graph of f at c; that is, it is the line through (c, f(c)) whose slope
is the deriva ve of f at c.
Some examples will help us understand these deni ons.

. Example 31
.Finding deriva ves and tangent lines
Let f(x) = 3x2 + 5x 7. Find:
1. f (1)
3. f (3)
2. The equa on of the tangent line

to the graph of f at x = 1.
4. The equa on of the tangent line

to the graph f at x = 3.
1. We compute this directly using Deni on 7.

f(1 + h) f(1)
h
3(1 + h)2 + 5(1 + h) 7 (3(1)2 + 5(1) 7)
= lim
h0
h
3h2 + 11h
= lim
h0
h
= lim 3h + 11 = 11.
f (1) = lim
h0
h0
Notes:
58

2. The tangent line at x = 1 has slope f (1) and goes through the point
(1, f(1)) = (1, 1). Thus the tangent line has equa on, in point-slope form,
y = 11(x 1) + 1. In slope-intercept form we have y = 11x 10.
3. Again, using the deni on,
f(3 + h) f(3)
h
3(3 + h)2 + 5(3 + h) 7 (3(3)2 + 5(3) 7)
= lim
h0
h
3h2 + 23h
= lim
h0
h
= lim 3h + 23
f (3) = lim
h0
h0
y
60
40
20
= 23.
4. The tangent line at x = 3 has slope 23 and goes through the point (3, f(3)) =
(3, 35). Thus the tangent line has equa on y = 23(x3)+35 = 23x34.
A graph of f is given in Figure 2.3 along with the tangent lines at x = 1 and
x = 3. .
Figure 2.3: A graph of f(x) = 3x2 + 5x 7

and its tangent lines at x = 1 and x = 3.
Another important line that can be created using informa on from the derivave is the normal line. It is perpendicular to the tangent line, hence its slope is
the oppositereciprocal of the tangent lines slope.
Deni on 9
Normal Line
Let f be con nuous on an open interval I and dieren able at c, for some
c in I. The normal line to the graph of f at c is the line with equa on
n(x) =
1
(x c) + f(c),
f (c)
where f ((c) = )0. When f (c) = 0, the normal line is the ver cal line
through c, f(c) ; that is, x = c.
. Example 32
.Finding equa ons of normal lines
Let f(x) = 3x2 + 5x 7, as in Example 31. Find the equa ons of the normal lines
to the graph of f at x = 1 and x = 3.
S
In Example 31, we found that f (1) = 11. Hence at x = 1,
Notes:
59

the normal line will have slope 1/11. An equa on for the normal line is
n(x) =
x
1
Figure 2.4: A graph of f(x) = 3x2 +5x7,

along with its normal line at x = 1.
1
(x 1) + 1.
11
The normal line is plo ed with y = f(x) in Figure 2.4. Note how the line looks
perpendicular to f. (A key word here is looks. Mathema cally, we say that the
normal line is perpendicular to f at x = 1 as the slope of the normal line is the
oppositereciprocal of the slope of the tangent line. However, normal lines may
not always look perpendicular. The aspect ra o of the graph plays a big role in
this.)
We also found that f (3) = 23, so the normal line to the graph of f at x = 3
will have slope 1/23. An equa on for the normal line is
.
n(x) =
1
(x 3) + 35.
23
Linear func ons are easy to work with; many func ons that arise in the
course of solving real problems are not easy to work with. A common pracce in mathema cal problem solving is to approximate dicult func ons with
notsodicult func ons. Lines are a common choice. It turns out that at any
given point on a dieren able func on f, the best linear approxima on to f is its
tangent line. That is one reason well spend considerable me nding tangent
lines to func ons.
One type of func on that does not benet from a tangentline approximaon is a line; it is rather simple to recognize that the tangent line to a line is the
line itself. We look at this in the following example.
.
. Example 33
.Finding the Deriva ve of a Line
quad Consider f(x) = 3x + 5. Find the equa on of the tangent line to f at x = 1
and x = 7.
S
We nd the slope of the tangent line by using Deni on 7.

f(1 + h) f(1)
h0
h
3(1 + h) + 5 (3 + 5)
= lim
h0
h
3h
= lim
h0 h
= lim 3
f (1) = lim
h0
= 3.
Notes:
60

We just found that f (1) = 3. That is, we found the instantaneous rate of
change of f(x) = 3x + 5 is 3. This is not surprising; lines are characterized by
being the only func ons with a constant rate of change. That rate of change
is called the slope of the line. Since their rates of change are constant, their
instantaneous rates of change are always the same; they are all the slope.
So given a line f(x) = ax + b, the deriva ve at any point x will be a; that is,
f (x) = a.
It is now easy to see that the tangent line to the graph of f at x = 1 is just f,
with the same being true for x = 7. .
We o en desire to nd the tangent line to the graph of a func on without
knowing the actual deriva ve of the func on. In these cases, the best we may
be able to do is approximate the tangent line. We demonstrate this in the next
example.
. Example 34
Numerical Approxima on of the Tangent Line
Approximate the equa on of the tangent line to the graph of f(x) = sin x at
x = 0.
y
1
In order to nd the equa on of the tangent line, we need a

S
slope and a point. The point is given to us: (0, sin 0) = (0, 0). To compute the
slope, we need the deriva ve. This is where we will make an approxima on.
Recall that
sin(0 + h) sin 0
f (0)
h
for a small value of h. We choose (somewhat arbitrarily) to let h = 0.1. Thus
f (0)
sin(0.1) sin 0
0.9983.
0.1
0.5
x
2
0.5
Figure 2.5: f(x) = sin x graphed with an

approxima on to its tangent line at x = 0.
Thus our approxima on of the equa on of the tangent line is y = 0.9983(x

0) + 0 = 0.9983x; it is graphed in Figure 2.5. The graph seems to imply the
approxima on is rather good. .
sin x
= 1, meaning for values of x near 0,
x
sin x x. Since the slope of the line y = x is 1 at x = 0, it should seem reasonable that the slope of f(x) = sin x is near 1 at x = 0. In fact, since we
approximated the value of the slope to be 0.9983, we might guess the actual
value is 1. Well come back to this later.
Recall from Sec on 1.3 that lim
x0
Consider again Example 31. To nd the deriva ve of f at x = 1, we needed to

evaluate a limit. To nd the deriva ve of f at x = 3, we needed to again evaluate
a limit. We have this process:
Notes:
61

do something
to f and c
input specic
.
number c
return
number f (c)
This process describes a func on; given one input (the value of c), we return
exactly one output (the value of f (c)). The do something box is where the
tedious work (taking limits) of this func on occurs.
Instead of applying this func on repeatedly for dierent values of c, let us
apply it just once to the variable x. We then take a limit just once. The process
now looks like:
do something
to f and x
.
input variable
x
return
func on f (x)
The output is the deriva ve func on, f (x). The f (x) func on will take a
number c as input and return the deriva ve of f at c. This calls for a deni on.
.
Deni on 10
Deriva ve Func on
Let f be a dieren able func on on an open interval I. The func on

f(x + h) f(x)
h0
h
f (x) = lim
is the deriva ve of f.
Nota on:
Let y = f(x). The following nota on all represents the deriva ve:
f (x) = y =
dy
df
d
d
=
=
(f) =
(y).
dx
dx
dx
dx
dy
Important: The nota on
is one symbol; it is not the frac on dy/dx. The
dx
nota on, while somewhat confusing at rst, was chosen with care. A frac on
looking symbol was chosen because the deriva ve has many frac onlike proper es. Among other places, we see these proper es at work when we talk about
the units of the deriva ve, when we discuss the Chain Rule, and when we learn
about integra on (topics that appear in later sec ons and chapters).
Examples will help us understand this deni on.
. Example 35
.Finding the deriva ve of a func on
Let f(x) = 3x2 + 5x 7 as in Example 31. Find f (x).
Notes:
62

We apply Deni on 10.
f(x + h) f(x)
h
3(x + h)2 + 5(x + h) 7 (3x2 + 5x 7)
= lim
x0
h
3h2 + 6xh + 5h
= lim
x0
h
= lim 3h + 6x + 5
f (x) = lim
h0
x0
= 6x + 5
So f (x) = 6x + 5. Recall earlier we found that f (1) = 11 and f (3) = 23. Note
our new computa on of f (x) arm these facts. .
. Example 36
.Finding the deriva ve of a func on

1
Let f(x) =
. Find f (x).
x+1
S
We apply Deni on 10.

f (x) = lim
h0
= lim
h0
f(x + h) f(x)
h
1
1
x+h+1
x+1
h
Now nd common denominator then subtract; pull 1/h out front to facilitate
reading.
(
)
1
x+1
x+h+1
= lim
h0 h
(x + 1)(x + h + 1) (x + 1)(x + h + 1)
(
)
1
x + 1 (x + h + 1)
= lim
h0 h
(x + 1)(x + h + 1)
(
h
(x + 1)(x + h + 1)
1
= lim
h0 (x + 1)(x + h + 1)
1
=
(x + 1)(x + 1)
1
=
(x + 1)2
1
h0 h
= lim
Notes:
63

So f (x) =
1
. To prac ce our nota on, we could also state
(x + 1)2
(
)
d
1
1
.
=
dx x + 1
(x + 1)2
. Example 37
Finding the deriva ve of a func on
Find the deriva ve of f(x) = sin x.
Before applying Deni on 10, note that once this is found,
S
we can nd the actual tangent line to f(x) = sin x at x = 0, whereas we se led
for an approxima on in Example 34.
sin(x + h) sin x
h0
h
sin x cos h + cos x sin h sin x
= lim
h0
h
sin x(cos h 1) + cos x sin h
= lim
h0
h
(
)
sin x(cos h 1)
cos x sin h
= lim
+
h0
h
h
= sin x 0 + cos x 1
f (x) = lim
Use trig iden ty

sin(x + h) = sin x cos h + cos x sin h
(regroup)
(split into two frac ons)
(
use lim
h0
cos h 1
sin h
= 0 and lim
=1
h0 h
h
= cos x !
We have found that when f(x) = sin x, f (x) = cos x. This should be somewhat
surprising; the result of a tedious limit process and the sine func on is a nice
func on. Then again, perhaps this is not en rely surprising. The sine func on
is periodic it repeats itself on regular intervals. Therefore its rate of change
also repeats itself on the same regular intervals. We should have known the
deriva ve would be periodic; we now know exactly which periodic func on it is.
Thinking back to Example 34, we can nd the slope of the tangent line to
f(x) = sin x at x = 0 using our deriva ve. We approximated the slope as 0.9983;
we now know the slope is exactly cos 0 = 1. .
y
1
. Example 38
.Finding the deriva ve of a piecewise dened func on
Find the deriva ve of the absolute value func on,
{
x x < 0
f(x) = |x| =
.
x x0
0.5
See Figure 2.6.

x
1
0.5
0.5
Figure 2.6: The absolute value func on,

f(x) = |x|. No ce how the slope of
the lines (and hence the tangent lines)
abruptly changes at x = 0.
f(x + h) f(x)
. As f is piecewise
h
dened, we need to consider separately the limits when x < 0 and when x > 0.
S
Notes:
64
We need to evaluate lim
h0

When x < 0:
)
d(
(x + h) (x)
x = lim
h0
dx
h
h
= lim
h0 h
= lim 1
h0
= 1.
When x > 0, a similar computa on shows that
d( )
x = 1.
dx
We need to also nd the deriva ve at x = 0. By the deni on of the derivave at a point, we have
f (0) = lim
h0
f(0 + h) f(0)
.
h
Since x = 0 is the point where our func ons deni on switches from one piece
to other, we need to consider le and right-hand limits. Consider the following,
where we compute the le and right hand limits side by side.
lim
h0
f(0 + h) f(0)
=
h
h 0
lim
=
h
h0
lim 1 = 1
h0
lim
h0+
f(0 + h) f(0)
=
h
h0
lim
=
+
h
h0
lim 1 = 1
y
1
h0+
The last lines of each column tell the story: the le and right hand limits are not
equal. Therefore the limit does not exist at 0, and f is not dieren able at 0. So
we have
{
1 x < 0
f (x) =
.
1 x>0
At x = 0, f (x) does not exist; there is a jump discon nuity at 0; see Figure 2.7.
So f(x) = |x| is dieren able everywhere except at 0. .
The point of non-dieren ability came where the piecewise dened funcon switched from one piece to the other. Our next example shows that this
does not always cause trouble.
x
1
0.5
0.5
.
Figure 2.7: A graph of the deriva ve of
f(x) = |x|.
. Example 39
.Finding the deriva ve

{of a piecewise dened func on
sin x x /2
Find the deriva ve of f(x), where f(x) =
. See Figure 2.8.
1
x > /2
Notes:
65

Using Example 37, we know that when x < /2, f (x) =
cos x. It is easy to verify that when x > /2, f (x) = 0; consider:
S
lim
x0
0.5
f(x + h) f(x)
11
= lim
= lim 0 = 0.
x0
h0
h
h
So far we have
f (x) =
.
Figure 2.8: A graph of f(x) as dened in
Example 39.
cos x
0
We s ll need to nd f (/2). No ce at x = /2 that both pieces of f are 0,

meaning we can state that f (/2) = 0.
Being more rigorous, we can again evaluate the dierence quo ent limit at
x = /2, u lizing again le and righthand limits:
f(/2 + h) f(/2)
=
h
sin(/2 + h) sin(/2)
lim
=
h
h0
sin( 2 ) cos(h) + sin(h) cos( 2 ) sin( 2 )
lim
=
h
h0
1 cos(h) + sin(h) 0 1
lim
=
h
h0
0
lim
h0
0.5
.
Figure 2.9: A graph of f (x) in Example 39.
lim
h0+
f(/2 + h) f(/2)
=
h
11
lim
=
h
h0+
0
lim
=
h0+ h
0
Since both the le and right hand limits are 0 at x = /2, the limit exists and
f (/2) exists (and is 0). Therefore we can fully write f as
{
cos x x /2
f (x) =
.
0
x > /2
See Figure 2.9 for a graph of this func on. .
Recall we pseudodened a con nuous func on as one in which we could
sketch its graph without li ing our pencil. We can give a pseudodeni on for
dieren ability as well: it is a con nuous func on that does not have any sharp
corners. One such sharp corner is shown in Figure 2.6. Even though the funcon f in Example 39 is piecewisedened, the transi on is smooth hence it is
dieren able. Note how in the graph of f in Figure 2.8 it is dicult to tell when
f switches from one piece to the other; there is no corner.
This sec on dened the deriva ve; in some sense, it answers the ques on of
What is the deriva ve? The next sec on addresses the ques on What does
the deriva ve mean?
Notes:
66
x < /2
.
x > /2
Exercises 2.1
Terms and Concepts
10
,x=9
x+1
21. f(x) =
1. T/F: Let f be a posi on func on. The average rate of change

on [a, b] is the slope of the line through the points (a, f(a))
and (b, f(b)).
22. f(x) = ex , x = 2
2. T/F: The deni on of the deriva ve of a func on at a point

involves taking a limit.
24. The graph of f(x) = x2 1 is shown.
23. f(x) = cos x, x = 0
3. In your own words, explain the dierence between the average rate of change and instantaneous rate of change.
(a) Use the graph to approximate the slope of the tangent line to f at the following points: (1, 0), (0, 1)
and (2, 3).
4. In your own words, explain the dierence between Denions 7 and 10.
(b) Using the deni on, nd f (x).

(c) Find the slope of the tangent line at the points
(1, 0), (0, 1) and (2, 3).
5. Let y = f(x). Give three dierent nota ons equivalent to

f (x).
Problems
3
2
In Exercises 6 12, use the deni on of the deriva ve to compute the deriva ve of the given func on.
6. f(x) = 6
x
1
7. f(x) = 2x
8. h(t) = 4 3t
2
1
1
25. The graph of f(x) =
9. g(x) = x
1
x
(b) Using the deni on, nd f (x).
1
12. r(s) =
s2
(c) Find the slope of the tangent line at the points (0, 1)
and (1, 0.5).
In Exercises 13 19, a func on and an xvalue c are given.

(Note: these func ons are the same as those given in Exercises 6 through 12.)
y
5
(a) Find the tangent line to the graph of the func on at c.
(b) Find the normal line to the graph of the func on at c.
13. f(x) = 6, at x = 2.
14. f(x) = 2x, at x = 3.
17. f(x) = 3x2 x + 4, at x = 1.
.
1
15. h(x) = 4 3x, at x = 7.

16. g(x) = x2 , at x = 2.
In Exercises 26 29, a graph of a func on f(x) is given. Using

the graph, sketch f (x).
1
, at x = 2.
x
1
19. r(x) =
, at x = 3.
x2
18. h(x) =
3
2
In Exercises 20 23, a func on f and an xvalue a is given.

Approximate the equa on of the tangent line to the graph of
f at x = a by numerically approxima ng f (a), using h = 0.1.
20. f(x) = x2 + 2x + 1, x = 3
1
is shown.
x+1
(a) Use the graph to approximate the slope of the tangent line to f at the following points: (0, 1) and
(1, 0.5).
10. f(x) = 3x2 x + 4

11. h(x) =
1
x
2
26.
1
1
y
5
2
x
4
x
6
2
2
27. .
.
y
Review
31. Approximate lim
x5 x2
x
2
32. Use the Bisec on Method to approximate, accurate to two

decimal places, the root of g(x) = x3 + x2 + x 1 on
[0.5, 0.6].
28. .
x2 + 2x 35
.
10.5x + 27.5
33. Give intervals on which each of the following func ons are
con nuous.
y
1
1
+1
1
(b) 2
x 1
(a)
0.5
29. .
0.5
(c)
ex
(d)
5x
5 x2
34. Use the graph of f(x) provided to answer the following.

1
30. Using the graph of g(x) below, answer the following quesons.
(a)
(b)
lim f(x) =?
(c) lim f(x) =?
lim f(x) =?
(d) Where is f con nuous?
x3
x3
x3+
(a) Where is g(x) > 0?
(b) Where is g(x) < 0?
(c) Where is g(x) = 0?

1
(d) Where is g (x) < 0?
(e) Where is g (x) > 0?

(f) Where is g (x) = 0?
2
1
2.2 Interpreta ons of the Deriva ve
2.2
Interpreta ons of the Deriva ve
The previous sec on dened the deriva ve of a func on and gave examples of
how to compute it using its deni on (i.e., using limits). The sec on also started
with a brief mo va on for this deni on, that is, nding the instantaneous velocity of a falling object given its posi on func on. The next sec on will give us
more accessible tools for compu ng the deriva ve, tools that are easier to use
than repeated use of limits.
This sec on falls in between the What is the deni on of the deriva ve?
and How do I compute the deriva ve? sec ons. Here we are concerned with
What does the deriva ve mean?, or perhaps, when read with the right emphasis, What is the deriva ve? We oer two interconnected interpreta ons
of the deriva ve, hopefully explaining why we care about it and why it is worthy
of study.
Interpreta on of the Deriva ve #1: Instantaneous Rate of Change

The previous sec on started with an example of using the posi on of an
object (in this case, a falling amusementpark rider) to nd the objects velocity. This type of example is o en used when introducing the deriva ve because
we tend to readily recognize that velocity is the instantaneous rate of change
of posi on. In general, if f is a func on of x, then f (x) measures the instantaneous rate of change of f with respect to x. Put another way, the derivave answers When x changes, at what rate does f change? Thinking back to
the amusementpark ride, we asked When me changed, at what rate did the
height change? and found the answer to be By 64 feet per second.
Now imagine driving a car and looking at the speedometer, which reads 60
mph. Five minutes later, you wonder how far you have traveled. Certainly, lots
of things could have happened in those 5 minutes; you could have inten onally
sped up signicantly, you might have come to a complete stop, you might have
slowed to 20 mph as you passed through construc on. But suppose that you
know, as the driver, none of these things happened. You know you maintained
a fairly consistent speed over those 5 minutes. What is a good approxima on of
the distance traveled?
One could argue the only good approxima on, given the informa on provided, would be based on distance = rate me. In this case, we assume a
constant rate of 60 mph with a me of 5/60 hours. Hence we would approximate the distance traveled as 5 miles.
Referring back to the falling amusementpark ride, knowing that at t = 2 the
velocity was 64 /s, we could reasonably assume that 1 second later the rid-
Notes:
69

ers height would have dropped by about 64 feet. Knowing that the riders were
accelera ng as they fell would inform us that this is an underapproxima on. If
all we knew was that f(2) = 86 and f (2) = 64, wed know that wed have to
stop the riders quickly otherwise they would hit the ground!
Units of the Deriva ve
It is useful to recognize the units of the deriva ve func on. If y is a func on
of x, i.e., y = f(x) for some func on f, and y is measured in feet and x in seconds,
then the units of y = f are feet per second, commonly wri en as /s. In
general, if y is measured in units P and x is measured in units Q, then y will be
measured in units P per Q, or P/Q. Here we see the frac onlike behavior
of the deriva ve in the nota on:
the units of
dy
dx
are
units of y
.
units of x
. Example 40
The meaning of the deriva ve: World Popula on
Let P(t) represent the world popula on t minutes a er 12:00 a.m., January 1,
2012. It is fairly accurate to say that P(0) = 7, 028, 734, 178 (www.prb.org). It
is also fairly accurate to state that P (0) = 156; that is, at midnight on January 1,
2012, the popula on of the world was growing by about 156 people per minute
(note the units). Twenty days later (or, 28,800 minutes later) we could reasonably assume the popula on grew by about 28, 800 156 = 4, 492, 800 people. .
. Example 41
The meaning of the deriva ve: Manufacturing
The term widget is an economic term for a generic unit of manufacturing

output. Suppose a company produces widgets, and knows that the market supports a price of $10 per widget. Let P(n) give the prot, in dollars, earned by
manufacturing and selling n widgets. The company likely cannot make a (posi ve) prot making just one widget; the startup costs will likely exceed $10.
Mathema cally, we would write this as P(1) < 0.
What do P(1000) = 500 and P (1000) = 0.25 mean? Approximate P(1100).
The equa on P(1000) = 500 means that selling 1,000 widS
gets returns a prot of $500. We interpret P (1000) = 0.25 as meaning that the
prot per widget is increasing at rate of $0.25 per widget (the units are dollars
per widget.) Since we have no other informa on to use, our best approxima on
for P(1100) is:
P(1100) P(1000) + P (1000) 100 = $500 + 100 0.25 = $525.
We approximate that selling 1,100 widgets returns a prot of $525. .
Notes:
70

The previous examples made use of an important approxima on tool that
we rst used in our previous driving a car at 60 mph example at the beginning of this sec on. Five minutes a er looking at the speedometer, our best
approxima on for distance traveled assumed the rate of change was constant.
In Examples 40 and 41 we made similar approxima ons. We were given rate of
change informa on which we used to approximate total change. Nota onally,
we would say that
f(c + h) f(c) + f (c) h.
This approxima on is best when h is small. Small is a rela ve term; when
dealing with the world popula on, h = 22 days = 28,800 minutes is small in
comparison to years. When manufacturing widgets, 100 widgets is small when
one plans to manufacture thousands.
The Deriva ve and Mo on

One of the most fundamental applica ons of the deriva ve is the study of
mo on. Let s(t) be a posi on func on, where t is me and s(t) is distance. For
instance, s could measure the height of a projec le or the distance an object has
traveled.
Lets let s(t) measure the distance traveled, in feet, of an object a er t seconds of travel. Then s (t) has units feet per second, and s (t) measures the
instantaneous rate of distance change it measures velocity.
Now consider v(t), a velocity func on. That is, at me t, v(t) gives the velocity of an object. The deriva ve of v, v (t), gives the instantaneous rate of
velocity change accelera on. (We o en think of accelera on in terms of cars:
a car may go from 0 to 60 in 4.8 seconds. This is an average accelera on, a
measurement of how quickly the velocity changed.) If velocity is measured in
feet per second, and me is measured in seconds, then the units of accelera on
(i.e., the units of v (t)) are feet per second per second, or ( /s)/s. We o en
shorten this to feet per second squared, or /s2 , but this tends to obscure the
meaning of the units.
Perhaps the most well known accelera on is that of gravity. In this text, we
use g = 32 /s2 or g = 9.8m/s2 . What do these numbers mean?
A constant accelera on of 32( /s)/s means that the velocity changes by
32 /s each second. For instance, let v(t) measures the velocity of a ball thrown
straight up into the air, where v has units /s and t is measured in seconds. The
ball will have a posi ve velocity while traveling upwards and a nega ve velocity
while falling down. The accelera on is thus 32 /s2 . If v(1) = 20 /s2 , then
when t = 2, the velocity will have decreased by 32 /s; that is, v(t) = 12 /s.
We can con nue: v(3) = 44 /s, and we can also gure that v(0) = 42 /s.
These ideas are so important we write them out as a Key Idea.
Notes:
71

.
Key Idea 1
The Deriva ve and Mo on
1. Let s(t) be the posi on func on of an object. Then s (t) is the

velocity func on of the object.
2. Let v(t) be the velocity func on of an object. Then v (t) is the

accelera on func on of the object.
We now consider the second interpreta on of the deriva ve given in this

sec on. This interpreta on is not independent from the rst by any means;
many of the same concepts will be stressed, just from a slightly dierent perspec ve.
y
16
Interpreta on of the Deriva ve #2: The Slope of the Tangent Line
12
8
4
x
1
Figure 2.10: A graph of f(x) = x2 .
f(c + h) f(c)
Given a func on y = f(x), the dierence quo ent
gives a
h
change in y values divided by a change in x values; i.e., it is a measure of the
rise over run, or slope, of the line that goes through two points on the graph
of f. As h shrinks to 0, these two points come close together; in the limit we nd
f (c), the slope of a special line called the tangent line that intersects f only once
near x = c.
Lines have a constant rate of change, their slope. Nonlinear func ons do not
have a constant rate of change, but we can measure their instantaneous rate of
change at a given x value c by compu ng f (c). We can get an idea of how f is
behaving by looking at the slopes of its tangent lines. We explore this idea in the
following example.
. Example 42
.Understanding the deriva ve: the rate of change
Consider f(x) = x2 as shown in Figure 2.10. It is clear that at x = 3 the func on
is growing faster than at x = 1; how much faster?
16
12
8
4
..
x
1
Figure 2.11: A graph of f(x) = x2 and tangent lines.
We can answer this directly a er the following sec on, where

S
we learn to quickly compute deriva ves. For now, we will answer graphically,
by considering the slopes of the respec ve tangent lines.
With prac ce, one can fairly eec vely sketch tangent lines to a curve at a
par cular point. In Figure 2.11, we have sketched the tangent lines to f at x = 1
and x = 3, along with a grid to help us measure the slopes of these lines. At
x = 1, the slope is 2; at x = 3, the slope is 6. Thus we can say not only is f
Notes:
72

growing faster at x = 3 than at x = 1, it is growing three mes as fast. .
. Example 43
Understanding the graph of the deriva ve
Consider the graph of f(x) (in blue) and its deriva ve, f (x) (in red) in Figure 2.12.
Use these graphs to nd the slopes of the tangent lines to the graph of f at x = 1,
x = 2, and x = 3.
To nd the appropriate slopes of tangent lines to the graph
S
of f, we need to look at the corresponding values of f .
The slope of the tangent line to f at x = 1 is f (1); this looks to be about 1.
Using these slopes, the tangent lines to f are sketched in Figure 2.13. Included on the graph of f in this gure are lled circles where x = 1, x = 2 and
x = 3 to help be er visualize the y value of f at those points. .
x
1
Figure
. 2.12: Graphs of f and f in Example
43.
. Example 44
Approxima on with the deriva ve
Consider again the graph of f(x) and its deriva ve f (x) in Example 43. Use the
tangent line to f at x = 3 to approximate the value of f(3.1).
Figure 2.14 shows the graph of f along with its tangent line,
S
zoomed in at x = 3. No ce that near x = 3, the tangent line makes an excellent
approxima on of f. Since lines are easy to deal with, o en it works well to approximate a func on with its tangent line. (This is especially true when you dont
actually know much about the func on at hand, as we dont in this example.)
While the tangent line to f was drawn in Example 43, it was not explicitly
computed. Recall that the tangent line to f at x = c is y = f (c)(x c) + f(c).
While f is not explicitly given, by the graph it looks like f(3) = 4. Recalling that
f (3) = 3, we can compute the tangent line to be y = 3(x 3) + 4. It is o en
useful to leave the tangent line in pointslope form.
To use the tangent line to approximate f(3.1), we simply evaluate y at 3.1
instead of f.
y
5
x
1
Figure
. 2.13: Graphs of f and f in Example
43, along with tangent lines.
f(3.1) y(3.1) = 3(3.1 3) + 4 = .1 3 + 4 = 4.3.
We approximate f(3.1) 4.3. .

To demonstrate the accuracy of the tangent line approxima on, we now
state that in Example 44, f(x) = x3 + 7x2 12x + 4. We can evaluate f(3.1) =
4.279. Had we known f all along, certainly we could have just made this computa on. In reality, we o en only know two things:
1. What f(c) is, for some value of c, and

2
x
2.8
Notes:
3.2
Figure 2.14: Zooming in on f at x = 3 for

the func on given in Examples 43 and 44.
73

2. what f (c) is.
For instance, we can easily observe the loca on of an object and its instantaneous velocity at a par cular point in me. We do not have a func on f
for the loca on, just an observa on. This is enough to create an approxima ng
func on for f.
This last example has a direct connec on to our approxima on method explained above a er Example 41. We stated there that
f(c + h) f(c) + f (c) h.
If we know f(c) and f (c) for some value x = c, then compu ng the tangent
line at (c, f(c)) is easy: y(x) = f (c)(x c) + f(c). In Example 44, we used the
tangent line to approximate a value of f. Lets use the tangent line at x = c
to approximate a value of f near x = c; i.e., compute y(c + h) to approximate
f(c + h), assuming again that h is small. Note:
(
)
y(c + h) = f (c) (c + h) c + f(c) = f (c) h + f(c).
This is the exact same approxima on method used above! Not only does it make
intui ve sense, as explained above, it makes analy cal sense, as this approximaon method is simply using a tangent line to approximate a func ons value.
The importance of understanding the deriva ve cannot be understated. When

f is a func on of x, f (x) measures the instantaneous rate of change of f with respect to x and gives the slope of the tangent line to f at x.
Notes:
74
Exercises 2.2
Terms and Concepts
In Exercises 15 18, graphs of func ons f(x) and g(x) are

given. Iden fy which func on is the deriva ve of the other.)
1. What is the instantaneous rate of change of posi on

called?
y
g(x)
2. Given a func on y = f(x), in your own words describe how

to nd the units of f (x).
f(x)
15.
x
4
3. What func ons have a constant rate of change?
2
2
4
Problems
4. Given f(5) = 10 and f (5) = 2, approximate f(6).
5. Given P(100) = 67 and P (100) = 5, approximate

P(110).
4
g(x)
16.
x
2
2
6. Given z(25) = 187 and z (25) = 17, approximate z(20).
7. Knowing f(10) = 25 and f (10) = 5 and the methods described in this sec on, which approxima on is likely to be
most accurate: f(10.1), f(11), or f(20)? Explain your reasoning.
8. Given f(7) = 26 and f(8) = 22, approximate f (7).
y
5
f(x)
17.
. 5
9. Given H(0) = 17 and H(2) = 29, approximate H (2).
g(x)
10. Let V(x) measure the volume, in decibels, measured inside

a restaurant with x customers. What are the units of V (x)?
11. Let v(t) measure the velocity, in /s, of a car moving in a
straight line t seconds a er star ng. What are the units of
v (t)?
y
f(x)
1
18.
x
4
12. The height H, in feet, of a river is recorded t hours a er midnight, April 1. What are the units of H (t)?
13. P is the prot, in thousands of dollars, of producing and selling c cars.
(a) What are the units of P (c)?
(b) What is likely true of P(0)?
14. T is the temperature in degrees Fahrenheit, h hours a er
midnight on July 4 in Sidney, NE.
(a) What are the units of T (h)?
(b) Is T (8) likely greater than or less than 0? Why?

(c) Is T(8) likely greater than or less than 0? Why?
f(x)
g(x)
Review
In Exercises 19 20, use the deni on to compute the derivaves of the following func ons.
19. f(x) = 5x2
20. f(x) = (x 2)3
In Exercises 21 22, numerically approximate the value of
f (x) at the indicated x value.
21. f(x) = cos x at x = .
22. f(x) = x at x = 9.
2.3 Basic Dieren a on Rules

The deriva ve is a powerful tool but is admi edly awkward given its reliance on
limits. Fortunately, one thing mathema cians are good at is abstrac on. For
instance, instead of con nually nding deriva ves at a point, we abstracted and
found the deriva ve func on.
Lets prac ce abstrac on on linear func ons, y = mx + b. What is y ? Without limits, recognize that linear func on are characterized by being func ons
with a constant rate of change (the slope). The deriva ve, y , gives the instantaneous rate of change; with a linear func on, this is constant, m. Thus y = m.
Lets abstract once more. Lets nd the deriva ve of the general quadra c
func on, f(x) = ax2 + bx + c. Using the deni on of the deriva ve, we have:
a(x + h)2 + b(x + h) + c (ax2 + bx + c)
h0
h
2
ah + 2ahx + bh
= lim
h0
h
= lim ah + 2ax + b
f (x) = lim
h0
= 2ax + b.
So if y = 6x2 + 11x 13, we can immediately compute y = 12x + 11.
In this sec on (and in some sec ons to follow) we will learn some of what
mathema cians have already discovered about the deriva ves of certain funcons and how deriva ves interact with arithme c opera ons. We start with a
theorem.
.
Theorem 12
Deriva ves of Common Func ons
1. Constant Rule:
d( )
c = 0, where c is a constant.
dx
2. Power Rule:
d n
(x ) = nxn1 , where n is an integer, n > 0.
dx
d
3.
(sin x) = cos x
dx
4.
d
(cos x) = sin x
dx
d x
(e ) = ex
dx
6.
d
1
(ln x) =
dx
x
5.
This theorem starts by sta ng an intui ve fact: constant func ons have no
rate of change as they are constant. Therefore their deriva ve is 0 (they change
Notes:
76

at the rate of 0). The theorem then states some fairly amazing things. The Power
Rule states that the deriva ves of Power Func ons (of the form y = xn ) are very
straigh orward: mul ply by the power, then subtract 1 from the power. We see
something incredible about the func on y = ex : it is its own deriva ve. We also
see a new connec on between the sine and cosine func ons.
One special case of the Power Rule is when n = 1, i.e., when f(x) = x. What
is f (x)? According to the Power Rule,
f (x) =
d ( 1)
d( )
x =
x = 1 x0 = 1.
dx
dx
In words, we are asking At what rate does f change with respect to x? Since f
is x, we are asking At what rate does x change with respect to x? The answer
is: 1. They change at the same rate.
Lets prac ce using this theorem.
. Example 45
Let f(x) = x3 .
Using Theorem 12 to nd, and use, deriva ves
1. Find f (x).
y
2. Find the equa on of the line tangent to the graph of f at x = 1.
3. Use the tangent line to approximate (1.1)3 .

2
4. Sketch f, f and the found tangent line on the same axis.
x
2
1. The Power Rule states that if f(x) = x3 , then f (x) = 3x2 .

2. To nd the equa on of the line tangent to the graph of f at x = 1, we
need a point and the slope. The point is (1, f(1)) = (1, 1). The
slope is f (1) = 3. Thus the tangent line has equa on y = 3(x(1))+
(1) = 3x + 2.
Figure 2.15: A graph of f(x) = x3 , along
2
with
. its deriva ve f (x) = 3x and its tangent line at x = 1.
3. We can use the tangent line to approximate (1.1)3 as 1.1 is close to

1. We have
(1.1)3 3(1.1) + 2 = 1.3.
We can easily nd the actual answer; (1.1)3 = 1.331.
. 4. See Figure 2.15.
Notes:
77

Theorem 12 gives useful informa on, but we will need much more. For instance, using the theorem, we can easily nd the deriva ve of y = x3 , but it does
not tell how to compute the deriva ve of y = 2x3 , y = x3 +sin x nor y = x3 sin x.
The following theorem helps with the rst two of these examples (the third is
answered in the next sec on).
.
Theorem 13
Proper es of the Deriva ve
Let f and g be dieren able on an open interval I and let c be a real

number. Then:
1. Sum/Dierence Rule:
)
)
)
d(
d(
d(
f(x) g(x) =
f(x)
g(x) = f (x) g (x)
dx
dx
dx
2. Constant Mul ple Rule:
)
)
d(
d(
c f(x) = c
f(x) = c f (x).
dx
dx
Theorem 13 allows us to nd the deriva ves of a wide variety of func ons.
It can be used in conjunc on with the Power Rule to nd the deriva ves of any
polynomial. Recall in Example 35 that we found, using the limit deni on, the
deriva ve of f(x) = 3x2 +5x7. We can now nd its deriva ve without expressly
using limits:
)
d( 2
d ( 2)
d( )
d( )
3x + 5x + 7 = 3
x +5
x +
7
dx
dx
dx
dx
= 3 2x + 5 1 + 0
= 6x + 5.
We were a bit pedan c here, showing every step. Normally we would do all
)
d( 2
the arithme c and steps in our head and readily nd
3x +5x+7 = 6x+5.
dx
. Example 46
.Using the tangent line to approximate a func on value
Let f(x) = sin x + 2x + 1. Approximate f(3) using an appropriate tangent line.
This problem is inten onally ambiguous; we are to approxiS
mate using an appropriate tangent line. How good of an approxima on are we
seeking? What does appropriate mean?
In the real world, people solving problems deal with these issues all me.
One must make a judgment using whatever seems reasonable. In this example,
the actual answer is f(3) = sin 3 + 7, where the real problem spot is sin 3. What
is sin 3?
Notes:
78

Since 3 is close to , we can assume sin 3 sin = 0. Thus one guess is
f(3) 7. Can we do be er? Lets use a tangent line as instructed and examine
the results; it seems best to nd the tangent line at x = .
Using Theorem 12 we nd f (x) = cos x + 2. The slope of the tangent line is
thus f () = cos + 2 = 1. Also, f() = 2 + 1 7.28. So the tangent line to
the graph of f at x = is y = 1(x)+2+1 = x++1 x+4.14. Evaluated
at x = 3, our tangent line gives y = 3 + 4.14 = 7.14. Using the tangent line,
our nal approxima on is that f(3) 7.14.
Using a calculator, we get an answer accurate to 4 places a er the decimal:
f(3) = 7.1411. Our ini al guess was 7; our tangent line approxima on was more
accurate, at 7.14.
The point is not Heres a cool way to do some math without a calculator.
Sure, that might be handy some me, but your phone could probably give you
the answer. Rather, the point is to say that tangent lines are a good way of
approxima ng, and many scien sts, engineers and mathema cians o en face
problems too hard to solve directly. So they approximate. .
Higher Order Deriva ves

The deriva ve of a func on f is itself a func on, therefore we can take its
deriva ve. The following deni on gives a name to this concept and introduces
its nota on.
Deni on 11
Higher Order Deriva ves
Let y = f(x) be a dieren able func on on I.

1. The second deriva ve of f is:
f (x) =
( )
d( )
d dy
d2 y
f (x) =
= 2 = y .
dx
dx dx
dx
2. The third deriva ve of f is:
(
)
d ( )
d d2 y
d3 y
f (x) =
f (x) =
=
= y .
dx
dx dx2
dx3
Note: Deni on 11 comes with the

caveat Where the corresponding limits
exist. With f dieren able on I, it is possible that f is not dieren able on all of
I, and so on.
3. The nth deriva ve of f is:

f
(n)
)
(
d ( (n1) )
d d n1 y
d ny
(x) =
f
(x) =
=
= y(n) .
dx
dx dxn1
dxn
Notes:
79

In general, when nding the fourth deriva ve and on, we resort to the f (4) (x)
nota on, not f (x); a er a while, too many cks is too confusing.
Lets prac ce using this new concept.
. Example 47
Finding higher order deriva ves
Find the rst four deriva ves of the following func ons:
3. f(x) = 5ex
1. f(x) = 4x2
2. f(x) = sin x
S
1. Using the Power and Constant Mul ple Rules, we have: f (x) = 8x. Connuing on, we have
f (x) =
d( )
8x = 8;
dx
f (x) = 0;
f (4) (x) = 0.
No ce how all successive deriva ves will also be 0.

2. We employ Theorem 12 repeatedly.
f (x) = cos x;
f (x) = sin x;
f (x) = cos x;
f (4) (x) = sin x.
Note how we have come right back to f(x) again. (Can you quickly gure
what f (23) (x) is?)
3. Employing Theorem 12 and the Constant Mul ple Rule, we can see that
f (x) = f (x) = f (x) = f (4) (x) = 5ex .
.
Interpre ng Higher Order Deriva ves

What do higher order deriva ves mean? What is the prac cal interpretaon?
Our rst answer is a bit wordy, but is technically correct and benecial to
understand. That is,
The second deriva ve of a func on f is the rate of change of the rate
of change of f.
Notes:
80

One way to grasp this concept is to let f describe a posi on func on. Then, as
stated in Key Idea 1, f describes the rate of posi on change: velocity. We now
consider f , which describes the rate of velocity change. Sports car enthusiasts
talk of how fast a car can go from 0 to 60 mph; they are bragging about the
accelera on of the car.
We started this chapter with amusementpark riders freefalling with posion func on f(t) = 16t2 + 150. It is easy to compute f (t) = 32t /s and
f (t) = 32 ( /s)/s. We may recognize this la er constant; it is the acceleraon due to gravity. In keeping with the unit nota on introduced in the previous
sec on, we say the units are feet per second per second. This is usually shortened to feet per second squared, wri en as /s2 .
It can be dicult to consider the meaning of the third, and higher order,
deriva ves. The third deriva ve is the rate of change of the rate of change of
the rate of change of f. That is essen ally meaningless to the unini ated. In
the context of our posi on/velocity/accelera on example, the third deriva ve
is the rate of change of accelera on, commonly referred to as jerk.
Make no mistake: higher order deriva ves have great importance even if
their prac cal interpreta ons are hard (or impossible) to understand. The
mathema cal topic of series makes extensive use of higher order deriva ves.
Notes:
81
Exercises 2.3
Terms and Concepts
17. f(x) = 2 ln x x
1. What is the name of the rule which states that

nxn1 , where n > 0 is an integer?
)
d(
ln x ?
2. What is
dx
d ( n)
x =
dx
18. p(s) = 14 s4 + 31 s3 + 21 s2 + s + 1
19. h(t) = et sin t cos t
20. f(x) = ln(5x2 )
21. f(t) = ln(17) + e2 + sin /2
3. Give an example of a func on f(x) where f (x) = f(x).

4. Give an example of a func on f(x) where f (x) = 0.
5. The deriva ve rules introduced in this sec on explain how
to compute the deriva ve of which of the following funcons?
3
f(x) = 2
j(x) = sin x cos x
x
2
k(x) = ex
g(x) = 3x2 x + 17
m(x) = x
h(x) = 5 ln x
6. Explain in your own words how to nd the third deriva ve
of a func on f(x).
22. g(t) = (1 + 3t)2

23. g(x) = (2x 5)3
24. f(x) = (1 x)3
25. f(x) = (2 3x)2
In Exercises 26 31, compute the rst four deriva ves of the
given func on.
26. f(x) = x6
27. g(x) = 2 cos x
28. h(t) = t2 et
7. Give an example of a func on where f (x) = 0 and f (x) =

0.
29. p() = 4 3
8. Explain in your own words what the second deriva ve

means.
31. f(x) = 1, 100
9. If f(x) describes a posi on func on, then f (x) describes

what kind of func on? What kind of func on is f (x)?
10. Let f(x) be a func on measured in pounds, where x is measured in feet. What are the units of f (x)?
30. f() = sin cos
In Exercises 32 37, nd the equa ons of the tangent and

normal lines to the graph of the func on at the given point.
32. f(x) = x3 x at x = 1
33. f(t) = et + 3 at t = 0
Problems
34. g(x) = ln x at x = 1
In Exercises 11 25, compute the deriva ve of the given funcon.
36. f(x) = 2 cos x at x = /4
11. f(x) = 7x2 5x + 7

12. g(x) = 14x3 + 7x2 + 11x 29
13. m(t) = 9t5 18 t3 + 3t 8
14. f() = 9 sin + 10 cos
15. f(r) = 6er
16. g(t) = 10t cos t + 7 sin t
4
35. f(x) = 4 sin x at x = /2
37. f(x) = 2x + 3 at x = 5
Review
38. Given that e0 = 1, approximate the value of e0.1 using the
tangent line to f(x) = ex at x = 0.
39. Approximate the value of (3.01)4 using the tangent line to
f(x) = x4 at x = 3.
2.4 The Product and Quo ent Rules
2.4
The Product and Quo ent Rules
The previous sec on showed that, in some ways, deriva ves behave nicely. The
Constant Mul ple and Sum/Dierence Rules established that the deriva ve of
f(x) = 5x2 + sin x was not complicated. We neglected compu ng the deriva ve
5x2
of things like g(x) = 5x2 sin x and h(x) = sin
x on purpose; their deriva ves are
not as straigh orward. (If you had to guess what their respec ve deriva ves are,
you would probably guess wrong.) For these, we need the Product and Quo ent
Rules, respec vely, which are dened in this sec on.
We begin with the Product Rule.
Theorem 14
Product Rule
Let f and g be dieren able func ons on an open interval I. Then fg is a

dieren able func on on I, and
)
d(
f(x)g(x) = f(x)g (x) + f (x)g(x).
dx
We prac ce using this new rule in an example, followed by an example that

demonstrates why this theorem is true.
y
20
. Example 48
Using the Product Rule
Use the Product Rule to compute the deriva ve of y = 5x2 sin x. Evaluate the
deriva ve at x = /2.
15
To make our use of the Product Rule explicit, lets set f(x) =
5x and g(x) = sin x. We easily compute/recall that f (x) = 10x and g (x) =
cos x. Employing the rule, we have
10
)
d( 2
5x sin x = 5x2 cos x + 10x sin x.
dx
Figure 2.16: A graph of y = 5x2 sin x and

its tangent line at x = /2.
At x = /2, we have
y (/2) = 5
( )2
2
cos
()
2
+ 10
()
sin
= 5.
2
2
We graph y and its tangent line at x = /2, which has a slope of 5, in Figure
2.16. While this does not prove that the Produce Rule is the correct way to handle deriva ves of products, it helps validate its truth. .
Notes:
83

We now inves gate why the Product Rule is true.
. Example 49
A proof of the Product Rule
Use the deni on of the deriva ve to prove Theorem 14.
S
By the limit deni on, we have

)
d(
f(x + h)g(x + h) f(x)g(x)
f(x)g(x) = lim
.
h0
dx
h
We now do something a bit unexpected; add 0 to the numerator (so that nothing
is changed) in the form of f(x+h)g(x)+f(x+h)g(x), then do some regrouping
as shown.
)
d(
f(x + h)g(x + h) f(x)g(x)
f(x)g(x) = lim
(now add 0 to the numerator)
h0
dx
h
f(x + h)g(x + h) f(x + h)g(x) + f(x + h)g(x) f(x)g(x)
= lim
(regroup)
h0
h
(
) (
)
f(x + h)g(x + h) f(x + h)g(x) + f(x + h)g(x) f(x)g(x)
= lim
h0
h
f(x + h)g(x + h) f(x + h)g(x)
f(x + h)g(x) f(x)g(x)
= lim
+ lim
(factor)
h0
h0
h
h
g(x + h) g(x)
f(x + h) f(x)
= lim f(x + h)
+ lim
g(x) (apply limits)
h0
h0
h
h
= f(x)g (x) + f (x)g(x)
.
It is o en true that we can recognize that a theorem is true through its proof
yet somehow doubt its applicability to real problems. In the following example,
we compute the deriva ve of a product of func ons in two ways to verify that
the Product Rule is indeed right.
. Example 50
.Exploring alternate deriva ve methods
Let y = (x2 + 3x + 1)(2x2 3x + 1). Find y two ways: rst, by expanding
the given product and then taking the deriva ve, and second, by applying the
Product Rule. Verify that both methods give the same answer.
We rst expand the expression for y; a li le algebra shows
S
that y = 2x4 + 3x3 6x2 + 1. It is easy to compute y ;
y = 8x3 + 9x2 12x.
Notes:
84

Now apply the Product Rule.
y = (x2 + 3x + 1)(4x 3) + (2x + 3)(2x2 3x + 1)
(
) (
)
= 4x3 + 9x2 5x 3 + 4x3 7x + 3
= 8x3 + 9x2 12x.
The uninformed usually assume that the deriva ve of the product is the
product of the deriva ves. Thus we are tempted to say that y = (2x + 3)(4x
3) = 8x2 + 6x 9. Obviously this is not correct. .
We consider one more example before discussing another deriva ve rule.
. Example 51
Using the Product Rule
Find the deriva ves of the following func ons.
1. f(x) = x ln x
2. g(x) = x ln x x.
Recalling that the deriva ve of ln x is 1/x, we use the Product

Rule to nd our answers.
1.
)
d(
x ln x = x 1/x + 1 ln x = 1 + ln x.
dx
2. Using the result from above, we compute

)
d(
x ln x x = 1 + ln x 1 = ln x.
dx
This seems signicant; if the natural log func on ln x is an important func on (it
is), it seems worthwhile to know a func on whose deriva ve is ln x. We have
found one. (We leave it to the reader to nd another; a correct answer will be
very similar to this one.) .
We have learned how to compute the deriva ves of sums, dierences, and
products of func ons. We now learn how to nd the deriva ve of a quo ent of
func ons.
Notes:
85

.
Theorem 15
Quo ent Rule
Let f and g be func ons dened on an open interval I, where g(x) = 0

on I. Then f/g is dieren able on I, and
(
)
d f(x)
g(x)f (x) f(x)g (x)
=
.
dx g(x)
g(x)2
The Quo ent Rule is not hard to use, although it might be a bit tricky to remember. A useful mnemonic works as follows. Consider a frac ons numerator
and denominator as HI and LO, respec vely. Then
d
dx
HI
LO
)
=
LO dHI HI dLO
,
LOLO
read low dee high minus high dee low, over low low. Said fast, that phrase can
roll o the tongue, making it easy to memorize. The dee high and dee low
parts refer to the deriva ves of the numerator and denominator, respec vely.
Lets prac ce using the Quo ent Rule.
. Example 52
Using the Quo ent Rule
5x2
. Find f (x).
Let f(x) =
sin x
S
Directly applying the Quo ent Rule gives:

d
dx
5x2
sin x
sin x 10x 5x2 cos x

sin2 x
10x sin x 5x2 cos x
=
.
sin2 x
=
The Quo ent Rule allows us to ll in holes in our understanding of deriva ves
of the common trigonometric func ons. We start with nding the deriva ve of
the tangent func on.
. Example 53
.Using the Quo ent Rule to nd
Find the deriva ve of y = tan x.
S
Notes:
86
d
dx
)
tan x .
At rst, one might feel unequipped to answer this ques on.

But recall that tan x = sin x/ cos x, so we can apply the Quo ent Rule.
(
)
)
d(
d sin x
tan x =
dx
dx cos x
cos x cos x sin x( sin x)
=
cos2 x
2
cos x + sin2 x
=
cos2 x
1
=
cos2 x
= sec2 x.
This is beau ful result. To conrm its truth, we can nd the equa on of the
tangent line to y = tan x at x = /4. The slope is sec2 (/4) = 2; y = tan x,
along with its tangent line, is graphed in Figure 2.17. .
y
10
5
x
2
5
10
.
. 2.17: A graph of y = tan x along
Figure
with its tangent line at x = /4.
We include this result in the following theorem about the deriva ves of the
trigonometric func ons. Recall we found the deriva ve of y = sin x in Example
37 and stated the deriva ve of the cosine func on in Theorem 12. The derivaves of the cotangent, cosecant and secant func ons can all be computed directly using Theorem 12 and the Quo ent Rule.
Theorem 16
Deriva ves of Trigonometric Func ons
)
d(
sin x = cos x
dx
)
d(
3.
tan x = sec2 x
dx
)
d(
sec x = sec x tan x
5.
dx
1.
)
d(
cos x = sin x
dx
)
d(
4.
cot x = csc2 x
dx
)
d(
csc x = csc x cot x
6.
dx
2.
To remember the above, it may be helpful to keep in mind that the derivaves of the trigonometric func ons that start with c have a minus sign in them.
. Example 54

5x2
In Example 52 the deriva ve of f(x) =
was found using the Quo ent Rule.
sin x
2
Rewri ng f as f(x) = 5x csc x, nd f using Theorem 16 and verify the two

answers are the same.
Notes:
87

10x sin x 5x2 cos x
.
sin2 x
2
We now nd f using the Product Rule, considering f as f(x) = 5x csc x.
We found in Example 52 that the f (x) =
)
d( 2
5x csc x
dx
= 5x2 ( csc x cot x) + 10x csc x
1 cos x
10x
= 5x2
+
sin x sin x
sin x
5x2 cos x
10x
=
+
sin x
sin2 x
10x sin x 5x2 cos x
=
sin2 x
f (x) =
(now rewrite trig func ons)
(get common denominator)
Finding f using either method returned the same result. At rst, the answers
looked dierent, but some algebra veried they are the same. In general, there
is not one nal form that we seek; the immediate result from the Product Rule is
ne. Work to simplify your results into a form that is most readable and useful
to you. .
The Quo ent Rule gives other useful results, as show in the next example.
. Example 55
Using the Quo ent Rule to expand the Power Rule
1. f(x) =
1
x
2. f(x) =
1
, where n > 0 is an integer.
xn
We employ the Quo ent Rule.
1. f (x) =
x011
1
= 2.
x2
x
2. f (x) =
xn 0 1 nxn1
nxn1
n
=
= n+1 .
n
2
2n
(x )
x
x
The deriva ve of y =
Notes:
88
1
turned out to be rather nice. It gets be er. Conxn

sider:
d
dx
1
xn
d ( n )
x
dx
n
= n+1
x
= nx(n+1)
=
(apply result from Example 55)

(rewrite algebraically)
= nxn1 .
This is reminiscent of the Power Rule: mul ply by the power, then subtract 1
from the power. We now add to our previous Power Rule, which had the restric on of n > 0.
Theorem 17
Power Rule with Integer Exponents
. Then
Let f(x) = x , where n = 0 is an integer.
n
f (x) = n xn1 .
Taking the deriva ve of many func ons is rela vely straigh orward. It is
clear (with prac ce) what rules apply and in what order they should be applied.
Other func ons present mul ple paths; dierent rules may be applied depending on how the func on is treated. One of the beau ful things about calculus
is that there is not the right way; each path, when applied correctly, leads to
the same result, the deriva ve. We demonstrate this concept in an example.
. Example 56
x2 3x + 1
. Find f (x) in each of the following ways:
Let f(x) =
x
1. By applying the Quo ent Rule,
(
)
2. by viewing f as f(x) = x2 3x + 1 x1 and applying the Product and
Power Rules, and
3. by simplifying rst through division.
Verify that all three methods give the same result.
S
1. Applying the Quo ent Rule gives:

(
) (
)
x 2x 3 x2 3x + 1 1
x2 1
1
f (x) =
=
= 1 2.
2
2
x
x
x
Notes:
89

2. By rewri ng f, we can apply the Product and Power Rules as follows:
(
)
(
)
f (x) = x2 3x + 1 (1)x2 + 2x 3 x1
x2 3x + 1
x2
2
x 3x + 1
=
x2
2
x 1
=
=1
x2
=
2x 3
x
2x2 3x
+
x2
1
,
x2
+
the same result as above.
1
3. As x = 0, we can divide through by x rst, giving f(x) = x 3 + . Now
x
apply the Power Rule.
1
f (x) = 1 2 ,
x
the same result as before.
Example 56 demonstrates three methods of nding f . One is hard pressed

to argue for a best method as all three gave the same result without too much
diculty, although it is clear that using the Product Rule required more steps.
Ul mately, the important principle to take away from this is: reduce the answer
to a form that seems simple and easy to interpret. In that example, we saw
dierent expressions for f , including:
1
(
) (
)
(
)
(
)
x 2x 3 x2 3x + 1 1
1
=
= x2 3x + 1 (1)x2 + 2x 3 x1 .
2
2
x
x
They are equal; they are all correct; only the rst is clear. Work to make answers clear.
Notes:
90
Exercises 2.4
x+7
18. g(x) =
x
Terms and Concepts

)
d( 2
x sin x = 2x cos x.
dx
( 2 )
x
cos x
d
2. T/F: The Quo ent Rule states that
=
.
dx sin x
2x
t5
cos t 2t2
20. h(x) = cot x ex
1. T/F: The Product Rule states that
19. g(t) =
21. h(t) = 7t2 + 6t 2

7x 1
22. f(x) = (16x3 + 24x2 + 3x)
16x3 + 24x2 + 3x
23. f(t) = 5 t(sec t + et )

sin x
24. f(x) =
cos x + 3
(
)
25. g(x) = e2 sin(/4) 1
3. T/F: The deriva ves of the trigonometric func ons that

start with c have minus signs in them.
4. What deriva ve rule is used to extend the Power Rule to
include nega ve integer exponents?
5. T/F: Regardless of the func on, there is always exactly one
right way of compu ng its deriva ve.
26. g(t) = 4t3 et sin t cos t
6. In your own words, explain what it means to make your answers clear.
27. h(t) =
Problems
28. f(x) = x2 ex tan x

29. g(x) = 2x sin x sec x
In Exercises 7 10:
In Exercises 30 33, nd the equa ons of the tangent and

normal lines to the graph of g at the indicated point.
(a) Use the Product Rule to dieren ate the func on.
(b) Manipulate the func on algebraically and dieren ate without the Product Rule.
30. g(s) = es (s2 + 2) at (0, 2).

31. g(t) = t sin t at ( 3
, 3
)
2
2
(c) Show that the answers from (a) and (b) are equivalent.
x2
at (2, 4)
x1
cos 8
33. g() =
at (0, 5)
+1
In Exercises 34 37, nd the xvalues where the graph of the
func on has a horizontal tangent line.
32. g(x) =
7. f(x) = x(x2 + 3x)

8. g(x) = 2x2 (5x3 )
9. h(s) = (2s 1)(s + 4)
10. f(x) = (x2 + 5)(3 x3 )
34. f(x) = 6x2 18x 24
In Exercises 11 15:
35. f(x) = x sin x on [1, 1]

x
36. f(x) =
x+1
(a) Use the Quo ent Rule to dieren ate the func on.
(b) Manipulate the func on algebraically and dieren ate without the Quo ent Rule.
x2
x+1
In Exercises 38 41, nd the requested deriva ve.
37. f(x) =
(c) Show that the answers from (a) and (b) are equivalent.
11. f(x) =
x2 + 3
x
x 2x
2x2
3
13. h(s) = 3
4s
3
38. f(x) = x sin x; nd f (x).
39. f(x) = x sin x; nd f (4) (x).
12. g(x) =
14. f(t) =
15. f(x) =
40. f(x) = csc x; nd f (x).

41. f(x) = (x3 5x + 2)(x2 + x 7); nd f (8) (x).
In Exercises 42 45, use the graph of f(x) to sketch f (x).
t2 1
t+1
4
2t + 3
3t + 2
y
6
x + 2x
x+2
4
2
42.
x
3
16. f(x) = x sin x
1
2
4
1
17. f(t) = 2 (csc t 4)
t
.
.
y
6
y
10
43.
x
3
1
2
45.
6
4
2
44.
x
2 1
2
4
6
5
5
x
5
10
2.5 The Chain Rule
2.5
The Chain Rule
We have covered almost all of the deriva ve rules that deal with combina ons
of two (or more) func ons. The opera ons of addi on, subtrac on, mul plicaon (including by a constant) and division led to the Sum and Dierence rules,
the Constant Mul ple Rule, the Power Rule, the Product Rule and the Quo ent
Rule. To complete the list of dieren a on rules, we look at the last way two (or
more) func ons can be combined: the process of composi on (i.e. one func on
inside another).
Recall the nota on for composi on, (f g)(x) or f(g(x)), read as f of g of x.
In shorthand, we simply write f g or f(g) and read it as f of g. Before giving
the corresponding dieren a on rule, we note that the rule extends to mul ple
composi ons like f(g(h(x))) or f(g(h(j(x)))), etc.
To mo vate the rule, lets look at three deriva ves we can already compute.
. Example 57
Exploring similar deriva ves
Find the deriva ves of F1 (x) = (1 x)2 , F2 (x) = (1 x)3 , and F3 (x) = (1
x)4 . (Well see later why we are using subscripts for dierent func ons and an
uppercase F.)
In order to use the rules we already have, we must rst exS
pand each func on as F1 (x) = 1 2x + x2 , F2 (x) = 1 3x + 3x2 x3 and
F3 (x) = 1 4x + 6x2 4x3 + x4 .
It is not hard to see that:
F1 (x) = 2 + 2x,
F2 (x) = 3 + 6x 3x2 and
F3 (x) = 4 + 12x 12x2 + 4x3 .
An interes ng fact is that these can be rewri en as
F1 (x) = 2(x 1),
F2 (x) = 3(1 x)2 and F3 (x) = 4(1 x)3 .
A pa ern might jump out at you. Recognize that each of these func ons is a
composi on:
F1 (x) = f1 (g(x)),
where f1 (x) = x2 ,
F2 (x) = f2 (g(x)),
F3 (x) = f3 (g(x)),
where f2 (x) = x3 ,
where f3 (x) = x4 .
Well come back to this example a er giving the formal statements of the
Chain Rule; for now, we are just illustra ng a pa ern...
Notes:
93

.
Theorem 18
The Chain Rule
Let y = f(u) be a dieren able func on of u and let u = g(x) be a

dieren able func on of x. Then y = f(g(x)) is a dieren able func on
of x, and
y = f (g(x)) g (x).
To help understand the Chain Rule, we return to Example 57.

. Example 58
Using the Chain Rule
Use the Chain Rule to nd the deriva ves of the following func ons, as given in
Example 57.
Example 57 ended with the recogni on that each of the given
S
func ons was actually a composi on of func ons. To avoid confusion, we ignore
most of the subscripts here.
F1 (x) = (1 x)2 :
We found that
y = (1 x)2 = f(g(x)), where f(x) = x2 and g(x) = 1 x.
To nd y , we apply the Chain Rule. We need f (x) = 2x and g (x) = 1.
Part of the Chain Rule uses f (g(x)). This means subs tute g(x) for x in the
equa on for f (x). That is, f (x) = 2(1 x). Finishing out the Chain Rule we
have
y = f (g(x)) g (x) = 2(1 x) (1) = 2(1 x) = 2x 2.
F2 (x) = (1 x)3 :
Let y = (1 x)3 = f(g(x)), where f(x) = x3 and g(x) = (1 x). We have
f (x) = 3x2 , so f (g(x)) = 3(1 x)2 . The Chain Rule then states
y = f (g(x)) g (x) = 3(1 x)2 (1) = 3(1 x)2 .

F3 (x) = (1 x)4 :
Finally, when y = (1 x)4 , we have f(x) = x4 and g(x) = (1 x). Thus
f (x) = 4x3 and f (g(x)) = 4(1 x)3 . Thus
Notes:
94
y = f (g(x)) g (x) = 4(1 x)3 (1) = 4(1 x)3 .
2.5 The Chain Rule

Example 58 demonstrated a par cular pa ern: when f(x) = xn , then y =
n (g(x))n1 g (x). This is called the Generalized Power Rule.
Theorem 19
Generalized Power Rule
Let g(x) be a dieren able func on .

and let n = 0 be an integer. Then
)
(
)n1
d(
g(x)n = n g(x)
g (x).
dx
This allows us to quickly nd the deriva ve of func ons like y = (3x2 5x +

7 + sin x)20 . While it may look in mida ng, the Generalized Power Rule states
that
y = 20(3x2 5x + 7 + sin x)19 (6x 5 + cos x).
Treat the deriva vetaking process stepbystep. In the example just given,
rst mul ply by 20, the rewrite the inside of the parentheses, raising it all to
the 19th power. Then think about the deriva ve of the expression inside the
parentheses, and mul ply by that.
We now consider more examples that employ the Chain Rule.
. Example 59
.Using the Chain Rule
Find the deriva ves of the following func ons:
2. y = ln(4x3 2x2 )
1. y = sin 2x
3. y = ex
1. Consider y = sin 2x. Recognize that this is a composi on of func ons,

where f(x) = sin x and g(x) = 2x. Thus
y = f (g(x)) g (x) = cos(2x) 2 = 2 cos 2x.
2. Recognize that y = ln(4x3 2x2 ) is the composi on of f(x) = ln x and
g(x) = 4x3 2x2 . Also, recall that
) 1
d(
ln x = .
dx
x
This leads us to:
y =
4x3
12x2 4x
4x(3x 1)
2(3x 1)
1
(12x2 4x) = 3
=
=
.
2
2x
4x 2x2
2x(2x2 x)
2x2 x
Notes:
95

2
3. Recognize that y = ex is the composi on of f(x) = ex and g(x) = x2 .

Remembering that f (x) = ex , we have
2
y = ex (2x) = (2x)ex .
.
y
. Example 60
Using the Chain Rule to nd a tangent line
Let f(x) = cos x2 . Find the equa on of the line tangent to the graph of f at x = 1.
0.5
x
2
The tangent line goes through the point (1, f(1)) (1, 0.54)
S
with slope f (1). To nd f , we need the Chain Rule.
f (x) = sin(x2 ) (2x) = 2x sin x2 . Evaluated at x = 1, we have f (1) =
2 sin 1 1.68. Thus the equa on of the tangent line is
0.5
y = 1.68(x 1) + 0.54.
The tangent line is sketched along with f in Figure 2.18. .

2
Figure 2.18: f(x) = cos x sketched along

with its tangent line at x = 1.
The Chain Rule is used o en in taking deriva ves. Because of this, one can
become familiar with the basic process and learn pa erns that facilitate nding
deriva ves quickly. For instance,
)
d(
1
(anything)
ln(anything) =
(anything) =
.
dx
anything
anything
A concrete example of this is
) 45x14 + sin x + ex
d(
ln(3x15 cos x + ex ) = 15
.
dx
3x cos x + ex
While the deriva ve may look in mida ng at rst, look for the pa ern. The
denominator is the same as what was inside the natural log func on; the numerator is simply its deriva ve.
This pa ern recogni on process can be applied to lots of func ons. In general, instead of wri ng anything, we use u as a generic func on of x. We then
say
) u
d(
ln u = .
dx
u
The following is a short list of how the Chain Rule can be quickly applied to familiar func ons.
Notes:
96
2.5 The Chain Rule
d ( n)
u = n un1 u .
dx
d ( u)
2.
e = u eu .
dx
)
d(
3.
sin u = u cos u.
dx
)
d(
cos u = u sin u.
dx
)
d(
5.
tan u = u sec2 u.
dx
1.
4.
Of course, the Chain Rule can be applied in conjunc on with any of the other
rules we have already learned. We prac ce this next.
. Example 61
Using the Product, Quo ent and Chain Rules
1. f(x) = x5 sin 2x3
2. f(x) =
5x3
.
ex2
1. We must use the Product and Chain Rules. Do not think that you must be
able to see the whole answer immediately; rather, just proceed step
bystep.
(
)
f (x) = x5 6x2 cos 2x3 + 5x4 sin 2x3 = 6x7 cos 2x3 + 5x4 sin 2x3 .
2. We must employ the Quo ent Rule along with the Chain Rule. Again, proceed stepbystep.
)
(
)
2(
2)
2(
ex 15x2 5x3 (2x)ex
ex 30x4 + 15x2
f (x) =
=
( 2 )2
e2x2
ex
)
2(
.
= ex 30x4 + 15x2 .
A key to correctly working these problems is to break the problem down
into smaller, more manageable pieces. For instance, when using the Product
and Chain Rules together, just consider the rst part of the Product Rule at rst:
f(x)g (x). Just rewrite f(x), then nd g (x). Then move on to the f (x)g(x) part.
Dont a empt to gure out both parts at once.
Likewise, using the Quo ent Rule, approach the numerator in two steps and
handle the denominator a er comple ng that. Only simplify a erward.
We can also employ the Chain Rule itself several mes, as shown in the next
example.
Notes:
97

. Example 62
Using the Chain Rule mul ple mes
Find the deriva ve of y = tan5 (6x3 7x).
Recognize that we have the g(x) = tan(6x3 7x) func on
(
)5
inside the f(x) = x5 func on; that is, we have y = tan(6x3 7x) . We begin
using the Generalized Power Rule; in this rst step, we do not fully compute the
deriva ve. Rather, we are approaching this stepbystep.
S
(
)4
y = 5 tan(6x3 7x) g (x).
We now nd g (x). We again need the Chain Rule;
g (x) = sec2 (6x3 7x) (18x2 7).
Combine this with what we found above to give
(
)4
y = 5 tan(6x3 7x) sec2 (6x3 7x) (18x2 7)
= (90x2 35) sec2 (6x3 7x) tan4 (6x3 7x).
This func on is frankly a ridiculous func on, possessing no real prac cal
value. It is very dicult to graph, as the tangent func on has many ver cal
asymptotes and 6x3 7x grows so very fast. The important thing to learn from
this is that the deriva ve can be found. In fact, it is not hard; one must take
several simple steps and be careful to keep track of how to apply each of these
steps...
It is tradi onal mathema cal exercise to nd the deriva ves of arbitrarily
complicated func ons just to demonstrate that it can be done. Just break everything down into smaller pieces.
. Example 63
.Using the Product, Quo ent and Chain Rules

x cos(x2 ) sin2 (e4x )
Find the deriva ve of f(x) =
.
ln(x2 + 5x4 )
This func on likely has no prac cal use outside of demonS
stra ng deriva ve skills. The answer is given below without simplica on. It
employs the Quo ent Rule, the Product Rule, and the Chain Rule three mes.
f (x) =
]
[(
)
ln(x2 + 5x4 ) x ( sin(x2 )) (2x3 ) + 1 cos(x2 ) 2 sin(e4x ) cos(e4x ) (4e4x )
(
)
3
x cos(x2 ) sin2 (e4x ) 2x+20x
x2 +5x4
.
(
)
ln(x2 + 5x4 ) 2
Notes:
98
2.5 The Chain Rule

Again, in this example, there is no prac cal value to nding this deriva ve.
It just demonstrates that it can be done, no ma er how arbitrarily complicated
the func on is. .
The Chain Rule also has theore c value. That is, it can be used to nd the
deriva ves of func ons that we have not yet learned as we do in the following
example.
. Example 64
The Chain Rule and exponen al func ons
Use the Chain Rule to nd the deriva ve of f(x) = ax where a > 0, a = 1 is
constant.
We only know how to nd the deriva ve of one exponen al
S
func on: f(x) = ex ; this problem is asking us to nd the deriva ve of func ons
such as f(x) = 2x .
This can be accomplished by rewri ng ax in terms of e. Recalling that ex and
ln x are inverse func ons, we can write
a = eln a
and so
f(x) = ax = eln(a ) .
By the exponent property of logarithms, we can bring down the power to

get
f(x) = ax = ex(ln a) .
The func on is now the composi on y = f(g(x)), with f(x) = ex and g(x) =
x(ln a). Since f (x) = ex and g (x) = ln a, the Chain Rule gives
f (x) = ex(ln a) ln a.
Now one last look. Does the right hand side look at all familiar? In fact, the right
side contains the original func on itself! We have
f (x) = f(x) ln a = ax ln a.
The Chain Rule, coupled with the deriva ve rule of ex , allows us to nd the
deriva ves of all exponen al func ons. .
The previous example produced a result worthy of its own box.
Notes:
99

.
Theorem 20
Deriva ves of Exponen al Func ons
Let f(x) = a , for a > 0, a = 1. .

Then f is dieren able for all real
numbers and
f (x) = ln a ax .
x
Alternate Chain Rule Nota on

It is instruc ve to understand what the Chain Rule looks like using dy
dx
nota on instead of y nota on. Suppose that y = f(u) is a func on of u, where
u = g(x) is a func on of x (as stated in Theorem 18. Then, through the composi on f g, we can think of y as a func on of x, as y = f(g(x)). Thus the
deriva ve of y with respect to x makes sense; we can talk about dy
dx . This leads
to an interes ng progression of nota on:
y = f (g(x)) g (x)
dy
= y (u) u (x)
dx
dy
dy du
=
dx
du dx
(since y = f(u) and u = g(x))

(using frac onal nota on for the deriva ve)
Here the frac onal aspect of the deriva ve nota on stands out. On the
right hand side, it seems as though the du terms cancel out, leaving
dy
dy
= .
dx
dx
It is important to realize that we are not canceling these terms; the deriva ve
nota on of dy
dx is one symbol. It is equally important to realize that this nota on
was chosen precisely because of this behavior. It makes applying the Chain Rule
easy with mul ple variables. For instance,
dy
dy d d
=
.
dt
d d dt
where and are any variables youd like to use.
A er a while, you get be er at recognizing the pa ern and may take the
short cut of not actually wri ng down the func ons that make up the composion when you apply the Chain Rule. We simply recommend cau on and point
out thats where errors in work can (and o en do) occur.
Notes:
100
.
2.5 The Chain Rule
One of the most common ways of visualizing the Chain Rule is to consider
a set of gears, as shown in Figure 2.19. The gears have 36, 18, and 6 teeth,
respec vely. That means for every revolu on of the x gear, the u gear revolves
twice. That is: du
dx = 2. Likewise, every revolu on of u causes 3 revolu ons of
dy
y: du
= 3. How does y change with respect to x? For each revolu on of x, y
revolves 6 mes; that is,
dy
dy du
=
= 2 3 = 6.
dx
du dx
We can then extend the Chain Rule with more variables by adding more gears
to the picture.
dy
=6
dx
du
=2
dx
y
dy
=3
du
Figure 2.19: A series of gears to demon=

strate the Chain Rule. Note how dy
dx
dy
du
du
dx
Notes:
101
Exercises 2.5
Terms and Concepts
1. T/F: The Chain Rule describes how to evaluate the derivave of a composi on of func ons.
)
d(
g(x)n =
2. T/F: The Generalized Power Rule states that
dx
(
)n1
n g(x)
.
)
d(
1
ln(x2 ) = 2 .
dx
x
d ( x)
4. T/F:
3 1.1 3x .
dx
dx
dx dt
5. T/F:
=
dy
dt dy
3. T/F:
22. m(w) =
3w + 1
2w
2
23. f(x) =
3x + x
2x2
24. f(x) = x2 sin(5x)

25. g(t) = cos(t2 + 3t) sin(5t 7)
2
26. g(t) = cos( 1t )e5t
In Exercises 27 30, nd the equa ons of tangent and normal

lines to the graph of the func on at the given point. Note: the
func ons here are the same as in Exercises 6 through 9.
27. f(x) = (4x3 x)10 at x = 0
28. f(t) = (3t 2)5 at t = 1
Problems
29. g() = (sin + cos )3 at = /2

2

6. f(x) = (4x3 x)10
7. f(t) = (3t 2)5
8. g() = (sin + cos )3
2
9. h(t) = e3t +t1

(
)4
10. f(x) = x + 1x
11. f(x) = cos(3x)
12. g(x) = tan(5x)
13. h(t) = sin4 (2t)
14. p(t) = cos3 (t2 + 3t + 1)
15. f(x) = ln(cos x)
16. f(x) = ln(x2 )
17. f(x) = 2 ln(x)
18. g(r) = 4r
cos t
19. g(t) = 5
20. g(t) = 152

3w
21. m(w) = w
2
30. h(t) = e3t
+t1
at t = 1
)
d(
31. Compute
ln(kx) two ways:
dx
(a) Using the Chain Rule, and
(b) by rst using the logarithm rule ln(ab) = ln a + ln b,
then taking the deriva ve.
)
d(
32. Compute
ln(xk ) two ways:
dx
(a) Using the Chain Rule, and
(b) by rst using the logarithm rule ln(ap ) = p ln a, then
taking the deriva ve.
Review
33. The wind chill factor is a measurement of how cold it
feels during cold, windy weather. Let W(w) be the wind
chill factor, in degrees Fahrenheit, when it is 25 F outside
with a wind of w mph.
(a) What are the units of W (w)?
(b) What would you expect the sign of W (10) to be?
34. Find the deriva ves of the following func ons.
(a) f(x) = x2 ex cot x
(b) g(x) = 2x 3x 4x
2.6 Implicit Dieren a on
2.6
Implicit Dieren a on
dy
, or y , when y is
dx
given explicitly as a func on of x. That is, if we know y = f(x) for some func on
f, we can nd y . For example, given y = 3x2 7, we can easily nd y = 6x.
(Here we explicitly state how x and y are related. Knowing x, we can directly nd
y.)
Some mes the rela onship between y and x is not explicit; rather, it is implicit. For instance, we might know that x2 y = 4. This equality denes a
rela onship between x and y; if we know x, we could gure out y. Can we s ll
nd y ? In this case, sure; we solve for y to get y = x2 4 (hence we now know
y explicitly) and then dieren ate to get y = 2x.
Some mes the implicit rela onship between x and y is complicated. Suppose we are given sin(y) + y3 = 6 x3 . A graph of this implicit func on is given
in Figure 2.20. In this case there is absolutely no way to solve for y in terms of
elementary func ons. The surprising thing is, however, that we can s ll nd y
via a process known as implicit dieren a on.
Implicit dieren a on is a technique based on the Chain Rule that is used to
nd a deriva ve when the rela onship between the variables is given implicitly
rather than explicitly (solved for one variable in terms of the other).
In the previous sec ons we learned to nd the deriva ve,
x
2
2
2
.
Figure 2.20: A graph of the implicit funcon sin(y) + y3 = 6 x3 .
We begin by reviewing the Chain Rule. Let f and g be func ons of x. Then
)
d(
f(g(x)) = f (g(x)) g (x).
dx
Suppose now that y = g(x). We can rewrite the above as
)
d(
f(y)) = f (y)) y ,
dx
or
)
d(
dy
f(y)) = f (y) .
dx
dx
(2.1)
These equa ons look strange; the key concept to learn here is that we can nd
y even if we dont exactly know how y and x relate.
Lets see how it works with the equa on above.
. Example 65
.Using Implicit Dieren a on
Find y given that sin(y) + y3 = 6 x3 .
We start by taking the deriva ve of both sides (thus mainS
taining the equality.) We have :
)
)
d(
d(
sin(y) + y3 =
6 x3 .
dx
dx
Notes:
103

The right hand side is easy; it returns 3x2 .
The le hand side requires more considera on. We take the deriva ve term
byterm. Using the technique derived from Equa on 2.1 above, we can see that
)
d(
sin y = cos y y .
dx
We apply the same process to the y3 term.
d ( 3)
d ( 3)
y =
(y) = 3(y)2 y .
dx
dx
Pu ng this together with the right hand side, we have
cos(y)y + 3y2 y = 3x2 .
Now solve for y .
cos(y)y + 3y2 y = 3x2 .
(
)
cos y + 3y2 y = 3x2
y =
3x2
cos y + 3y2
This equa on for y probably seems unusual for it contains both x and y
terms. How is it to be used? Well address that next...
Implicit func ons are generally harder to deal with than explicit func ons.
With an explicit func on, given an x value, we have an explicit formula for compu ng the corresponding y value. With an implicit func on, one o en has to
nd x and y values at the same me that sa sfy the equa on. It is much easier to demonstrate that a given point sa ses the equa on than to actually nd
such a point.
For instance, we can arm easily that the point ( 3 6, 0) lies on the graph of
the implicit func on siny + y3 = 6 x3 . Plugging in 0 for y, we see the le hand
side is 0. Se ng x = 3 6, we see the right hand side is also 0; the equa on is
sa sed. The following example nds the equa on of the tangent line to this
func on at this point.
. Example 66
.Using Implicit Dieren a on to nd a tangent line
Find the equa on of the line tangent to
the curve of the implicitly dened funcon sin y + y3 = 6 x3 at the point ( 3 6, 0).
S
In Example 65 we found that

y =
Notes:
104
3x2
.
cos y + 3y2
We nd the slope of the tangent

line at the point ( 3 6, 0) by subs tu ng 3 6 for
x and 0 for y. Thus at the point ( 3 6, 0), we have the slope as
3( 3 6)2
3 3 36
y =
=
9.91.
cos 0 + 3 02
1
Therefore the equa on of thetangent line to the implicitly dened func on

sin y + y3 = 6 x3 at the point ( 3 6, 0) is
3
3
y = 3 36(x 6) + 0 9.91x + 18.
The curve and this tangent line are shown in Figure 2.21. .
This suggests a general method for implicit dieren a on. For the steps below assume y is a func on of x.
x
2
2
2
.
Figure 2.21: The func on sin y + y3 =
6 x3 and its tangent line at the point
( 3 6, 0).
1. Take the deriva ve of each term in the equa on. Treat the x terms like
normal. When taking the deriva ves of y terms, the usual rules apply
except that, because of the Chain Rule, we need to mul ply each term
by y .
2. Get all the y terms on one side of the equal sign and put the remaining
terms on the other side.
3. Factor out y ; solve for y by dividing.
Prac cal Note: When working by hand, it may be benecial to use the symbol
dy
1
dx instead of y , as the la er can be easily confused for y or y .
. Example 67
Given the implicitly dened func on y3 + x2 y4 = 1 + 2x, nd y .
We will take the implicit deriva ves term by term. The derivaS
ve of y3 is 3y2 y .
The second term, x2 y4 , is a li le tricky. It requires the Product Rule as it is the
product of two func ons of x: x2 and y4 . Its deriva ve is x2 (4y3 y ) + 2xy4 . The
rst part of this expression requires a y because we are taking the deriva ve of a
y term. The second part does not require it because we are taking the deriva ve
of x2 .
The deriva ve of the right hand side is easily found to be 2. In all, we get:
3y2 y + 4x2 y3 y + 2xy4 = 2.
Move terms around so that the le side consists only of the y terms and the
right side consists of all the other terms:
3y2 y + 4x2 y3 y = 2 2xy4 .
Notes:
105

Factor out y from the le side and solve to get
y =
x
5
10
10
.
Figure 2.22: A graph of the implicitly dened func on y3 + x2 y4 = 1 + 2x along

with its tangent line at the point (0, 1).
2 2xy4
.
3y2 + 4x2 y3
To conrm the validity of our work, lets nd the equa on of a tangent line
to this func on at a point. It is easy to conrm that the point (0, 1) lies on the
graph of this func on. At this point, y = 2/3. So the equa on of the tangent
line is y = 2/3(x0)+1. The func on and its tangent line are graphed in Figure
2.22.
No ce how our func on looks much dierent than other func ons we have
seen. For one, it fails the ver cal line test. Such func ons are important in many
areas of mathema cs, so developing tools to deal with them is also important. .
. Example 68
Given the implicitly dened func on sin(x2 y2 ) + y3 = x + y, nd y .
Dieren a ng term by term, we nd the most diculty in
S
the rst term. It requires both the Chain and Product Rules.
)
d(
d ( 2 2)
sin(x2 y2 ) = cos(x2 y2 )
x y
dx
dx
(
)
= cos(x2 y2 ) x2 (2yy ) + 2xy2
= 2(x2 yy + xy2 ) cos(x2 y2 ).
x
1
We leave the deriva ves of the other terms to the reader. A er taking the
deriva ves of both sides, we have
.
Figure 2.23: A graph of the implicitly dened func on sin(x2 y2 ) + y3 = x + y.
2(x2 yy + xy2 ) cos(x2 y2 ) + 3y2 y = 1 + y .

We now have to be careful to properly solve for y , par cularly because of
the product on the le . It is best to mul ply out the product. Doing this, we get
2x2 y cos(x2 y2 )y + 2xy2 cos(x2 y2 ) + 3y2 y = 1 + y .
From here we can safely move around terms to get the following:
2x2 y cos(x2 y2 )y + 3y2 y y = 1 2xy2 cos(x2 y2 ).
Then we can solve for y to get
y =
Notes:
106
1 2xy2 cos(x2 y2 )
.
+ 3y2 1
2x2 y cos(x2 y2 )

A graph of this implicit func on is given in Figure 2.23. It is easy to verify that
the points (0, 0), (0, 1) and (0, 1) all lie on the graph. We can nd the slopes
of the tangent lines at each of these points using our formula for y .
At (0, 0), the slope is 1.
At (0, 1), the slope is 1/2.
At (0, 1), the slope is also 1/2.
The tangent lines have been added to the graph of the func on in Figure
2.24. .
Quite a few famous curves have equa ons that are given implicitly. We can
use implicit dieren a on to nd the slope at various points on those curves.
We inves gate two such curves in the next examples.
. Example 69
Finding slopes of tangent lines to a circle
Find the slope of the tangent line to the circle x2 +y2 = 1 at the point (1/2, 3/2).
S
x
1
1
1
.
Figure 2.24: A graph of the implicitly dened func on sin(x2 y2 ) + y3 = x + y and
certain tangent lines.
Taking deriva ves, we get 2x+2yy = 0. Solving for y gives:

y =
x
.
y
y
This is a clever formula. Recall that the slope of the line through the origin and
the point (x, y) on the circle will be y/x. We have found that the slope of the
tangent line to the circle at that point is the opposite reciprocal of y/x, namely,
x/y. Hence these two
lines are always perpendicular.
At the point (1/2, 3/2), we have the tangent lines slope as
1/2
1
y =
= 0.577.
3/2
3
A graph of the circle and its tangent line at (1/2, 3/2) is given in Figure
2.25, along with a thin dashed line from the origin that is perpendicular to the
tangent line. (It turns out that all normal lines to a circle pass through the center
of the circle.) .
(1/2,
3/2)
x
1
Figure 2.25: The

unit circle with its tangent line at (1/2, 3/2).
This sec on has shown how to nd the deriva ves of implicitly dened funcons, whose graphs include a wide variety of interes ng and unusual shapes.
Implicit dieren a on can also be used to further our understanding of regular dieren a on.
One hole in our current understanding of deriva ves is this: what is the
deriva ve of the square root func on? That is,
d ( )
d ( 1/2 )
x =
x
=?
dx
dx
Notes:
107

We allude to a possible solu on, as we can write the square root func on as
a power func on with a ra onal (or, frac onal) power. We are then tempted to
apply the Power Rule and obtain
1
d ( 1/2 ) 1 1/2
x
= x
= .
dx
2
2 x
The trouble with this is that the Power Rule was ini ally dened only for
posi ve integer powers, n > 0. While we did not jus fy this at the me, generally the Power Rule is proved using something called the Binomial Theorem,
which deals only with posi ve integers. The Quo ent Rule allowed us to extend
the Power Rule to nega ve integer powers. Implicit Dieren a on allows us to
extend the Power Rule to ra onal powers, as shown below.
Let y = xm/n , where m and n are integers with no common factors (so m = 2
and n = 5 is ne, but m = 2 and n = 4 is not). We can rewrite this explicit
func on implicitly as yn = xm . Now apply implicit dieren a on.
y = xm/n
yn = xm
d ( n)
d ( m)
y =
x
dx
dx
n yn1 y = m xm1
m xm1
(now subs tute xm/n for y)
n yn1
m xm1
=
(apply lots of algebra)
n (xm/n )n1
y =
m (mn)/n
x
n
m m/n1
x
n
The above deriva on is the key to the proof extending the Power Rule to raonal powers. Using limits, we can extend this once more to include all powers,
including irra onal (even transcendental!) powers, giving the following theorem.
.
Theorem 21
Power Rule for Dieren a on
Let f(x) = x , where n = 0 is a real number. Then f is a dieren able

func on, and f (x) = n xn1 .
n
Notes:
108

y
20
This theorem allows us to say the deriva ve of x is x1 .

We now apply this nal version of the Power Rule in the next example, the
second inves ga on of a famous curve.
x
20
. Example 70
Using the Power Rule
Find the slope of x2/3 + y2/3 = 8 at the point (8, 8).
This is a par cularly interes ng curve called an astroid. It
is the shape traced out by a point on the edge of a circle that is rolling around
inside of a larger circle, as shown in Figure 2.26.
S
To nd the slope of the astroid at the point (8, 8), we take the deriva ve
implicitly.
Implicit Dieren a on and the Second Deriva ve
20
Figure 2.26: An astroid, traced out by a

point on the smaller circle as it rolls inside
the larger circle.
y
2 1/3 2 1/3
x
+ y
y =0
3
3
2
2 1/3
y
y = x1/3
3
3
x1/3
y = 1/3
y
y1/3
y = 1/3
x
Plugging in x = 8 and y = 8, we get a slope of 1. The astroid, with its
tangent line at (8, 8), is shown in Figure 2.27. .
20
20
(8, 8)
x
20
20
20
Figure 2.27: An astroid with a tangent

line.
We can use implicit dieren a on to nd higher order deriva ves. In theory,

this is simple: rst nd dy
dx , then take its deriva ve with respect to x. In prac ce,
it is not hard, but it o en requires a bit of algebra. We demonstrate this in an
example.
. Example 71
.Finding the second deriva ve

d2 y
Given x + y = 1, nd 2 = y .
dx
2
We found that y =
dy
dx
= x/y in Example 69. To nd y ,
Notes:
109

we apply implicit dieren a on to y .
d ( )
y
dx ( )
d
x
=
dx
y
y(1) x(y )
=
y2
y =
replace y with x/y:

y x(x/y)
y2
y + x2 /y
=
y2
While this is not a par cularly simple expression, it is usable. We can see that
y > 0 when y < 0 and y < 0 when y > 0. In Sec on 3.4, we will see how this
relates to the shape of the graph. .
y
Logarithmic Dieren a on
Consider the func on y = xx ; it is graphed in Figure 2.28. It is welldened

for x > 0 and we might be interested in nding equa ons of lines tangent and
normal to its graph. How do we take its deriva ve?
The func on is not a power func on: it has a power of x, not a constant.
It is not an exponen al func on: it has a base of x, not a constant.
x
1
Figure 2.28: A plot of y = xx .
A dieren a on technique known as logarithmic dieren a on becomes

useful here. The basic principle is this: take the natural log of both sides of an
equa on y = f(x), then use implicit dieren a on to nd y . We demonstrate
this in the following example.
. Example 72
.Using Logarithmic Dieren a on
Given y = xx , use logarithmic dieren a on to nd y .
S
Notes:
110
As suggested above, we start by taking the natural log of

both sides then applying implicit dieren a on.
y = xx
ln(y) = ln(xx )
y
4
(apply logarithm rule)
ln(y) = x ln x
(now use implicit dieren a on)
)
)
d(
d(
ln(y) =
x ln x
dx
dx
y
1
= ln x + x
y
x
y
= ln x + 1
y
(
)
y = y ln x + 1 (subs tute y = xx )
(
)
y = xx ln x + 1 .
x
1
Figure 2.29: A graph of y = xx and its tangent line at x = 1.5.
To test our answer, lets use it to nd the equa on of the tangent line at x =
1.5. The point on the graph our tangent line must pass through is (1.5, 1.51.5 )
(1.5, 1.837). Using the equa on for y , we nd the slope as
(
)
y = 1.51.5 ln 1.5 + 1 1.837(1.405) 2.582.
Thus the equa on of the tangent line is y = 1.6833(x 1.5) + 1.837. Figure
2.26 graphs y = xx along with this tangent line. .
Implicit dieren a on proves to be useful as it allows us to nd the instantaneous rates of change of a variety of func ons. In par cular, it extended the
Power Rule to ra onal exponents, which we then extended to all real numbers.
In the next sec on, implicit dieren a on will be used to nd the deriva ves of
inverse func ons, such as y = sin1 x.
Notes:
111
Exercises 2.6
Terms and Concepts
1. In your own words, explain the dierence between implicit
func ons and explicit func ons.
2. Implicit dieren a on is based on what other dieren aon rule?
In Exercises 23 27, nd the equa on of the tangent line to

the graph of the implicitly dened func on at the indicated
points. As a visual aid, each func on is graphed.
23. x2/5 + y2/5 = 1
(a) At (1, 0).
3. T/F: Implicitdieren a on can be used to nd the derivave of y = x.
(b) At (0.1, 0.281) (which does not exactly lie on the

curve, but is very close).
4. T/F: Implicit dieren a on can be used to nd the derivave of y = x3/4 .
y
1
Problems
0.5
(0.1, 0.281)
x
5. f(x) = 3 x
6. f(t) = 1 t2
7. g(t) = t sin t
24. x4 + y4 = 1
(a) At (1, 0).
(b) At ( 0.6, 0.8).

(c) At (0, 1).
9. x4 + y2 + y = 7
10. x2/5 + y2/5 = 1
14. x2 e2 + 2y = 5
x
1
(a) At (0, 4).
(b) At (2, 4 108).
x2 + y
= 17
x + y2
y
4
sin(x) + y
=1
cos(y) + x
2
0.5
25. (x2 + y2 4)3 = 108y2
17. (y2 + 2y x)2 = 200
19.
0.5
0.5
16. (3x2 + 2y3 )4 = 2
18.
( 0.6, 0.8)
0.5
15. x2 tan y = 50
dy
using implicit dieren a on.
dx
11. cos(x) + sin(y) = 1

x
12.
= 10
y
y
13.
= 10
x
0.5
0.5
8. h(x) = x1.5
In Exercises 9 21, nd
0.5
20. ln(x + y ) = e
2
21. ln(x + xy + y ) = 1
dy
22. Show that
is the same for each of the following implicitly
dx
dened func ons.
(a) xy = 1
(b) x2 y2 = 1
(c) sin(xy) = 1
(d) ln(xy) = 1
x
4
(2, 4 108)
26. (x2 + y2 + x)2 = x2 + y2

(a) At (0, 1).
)
(
3 3 3
(b) At ,
.
4 4
34 ,
3 3
4
In Exercises 28 31, an implicitly dened func on is given.

d2 y
Find 2 . Note: these are the same problems used in Exerdx
cises 9 through 12.
x
1
28. x4 + y2 + y = 7
29. x2/5 + y2/5 = 1
30. cos x + sin y = 1

x
31.
= 10
y
27. (x 2)2 + (y 3)2 = 9

)
7 6+3 3
,
.
2
2
(
)
4+3 3 3
(b) At
,
.
2
2
(
In Exercises 32 37, use logarithmic dieren a on to nd

dy
, then nd the equa on of the tangent line at the indicated
dx
xvalue.
(a) At
32. y = (1 + x)1/x ,
2
33. y = (2x)x ,
y
6
3.5,
6+3 3
2
34. y =
x
,
x+1
x=1
35. y = xsin(x)+2 , x = /2
( 4+33
2
x=1
x=1
, 1.5
x
2
36. y =
x+1
,
x+2
37. y =
(x + 1)(x + 2)
, x=0
(x + 3)(x + 4)
x=1
2.7 Deriva ves of Inverse Func ons
y
(1, 1.5)
(1.5, 1)
(0.5, 0.375)
x
1
1
(0.375, 0.5)
Figure 2.30: A func on f along with its inverse f1 . (Note how it does not ma er
which func on we refer to as f; the other
is f1 .)
Recall that a func on y = f(x) is said to be one to one if it passes the horizontal
line test; that is, for two dierent x values x1 and x2 , we do not have f(x1 ) = f(x2 ).
In some cases the domain of f must be restricted so that it is one to one. For
instance, consider f(x) = x2 . Clearly, f(1) = f(1), so f is not one to one on its
regular domain, but by restric ng f to (0, ), f is one to one.
Now recall that one to one func ons have inverses. That is, if f is one to one,
it has an inverse func on, denoted by f1 , such that if f(a) = b, then f1 (b) = a.
The domain of f1 is the range of f, and vice-versa. For ease of nota on, we set
g = f1 and treat g as a func on of x.
Since f(a) = b implies g(b) = a, when we compose f and g we get a nice
result:
(
)
f g(b) = f(a) = b.
(
)
(
)
In general, f g(x) = x and g f(x) = x. This gives us a convenient way to check
if two func ons are inverses of each other: compose them and if the result is x,
then they are inverses (on the appropriate domains.)
When the point (a, b) lies on the graph of f, the point (b, a) lies on the graph
of g. This leads us to discover that the graph of g is the reec on of f across the
line y = x. In Figure 2.30 we see a func on graphed along with its inverse. See
how the point (1, 1.5) lies on one graph, whereas (1.5, 1) lies on the other. Because of this rela onship, whatever we know about f can quickly be transferred
into knowledge about g.
For example, consider Figure 2.31 where the tangent line to f at the point
(a, b) is drawn. That line has slope f (a). Through reec on across y = x, we
1
can see that the tangent line to g at the point (b, a) should have slope .
f (a)
1
This then tells us that g (b) = .
f (a)
Consider:
1
(a, b)
x
1
1
(b, a)
Figure 2.31: Corresponding tangent lines

drawn to f and f1 .
Informa on about g = f1
(0.5, 0.375) lies on f
(0.375, 0.5) lies on g
Slope of tangent line to f

at x = 0.5 is 3/4
Slope of tangent line to

g at x = 0.375 is 4/3
f (0.5) = 3/4
g (0.375) = 4/3
We have discovered a rela onship between f and g in a mostly graphical

way. We can realize this rela onship analy cally as well. Let y = g(x), where
again g = f1 . We want to nd y . Since y = g(x), we know that f(y) = x. Using
the Chain Rule and Implicit Dieren a on, take the deriva ve of both sides of
Notes:
114
Informa on about f

this last equality.
)
d(
d( )
f(y) =
x
dx
dx
f (y) y = 1
1
y =
f (y)
1
y =
f (g(x))
This leads us to the following theorem.
Theorem 22
Deriva ves of Inverse Func ons
Let f be dieren able and one to one on an open interval I, where f (x) =
0 for all x in I, let J be the range of f on I, let g be the inverse func on of
f, and let f(a) = b for some a in I. Then g is a dieren able func on on
J, and in par cular,
( )
( )
1
1
1. f1 (b) = g (b) =
and 2. f1 (x) = g (x) =
f (a)
f (g(x))
The results of Theorem 22 are not trivial; the nota on may seem confusing
at rst. Careful considera on, along with examples, should earn understanding.
In the next example we apply Theorem 22 to the arcsine func on.
. Example 73
.Finding the deriva ve of an inverse trigonometric func on
Let y = arcsin x = sin1 x. Find y using Theorem 22.
Adop ng our previously dened nota on, let g(x) = arcsin x
S
and f(x) = sin x. Thus f (x) = cos x. Applying the theorem, we have
1
f (g(x))
1
=
.
cos(arcsin x)
g (x) =
This last expression is not immediately illumina ng. Drawing a gure will
help, as shown in Figure 2.33. Recall that the sine func on can be viewed as
taking in an angle and returning a ra o of sides of a right triangle, specically,
the ra o opposite over hypotenuse. This means that the arcsine func on takes
as input a ra o of sides and returns an angle. The equa on y = arcsin x can
be rewri en as y = arcsin(x/1); that is, consider a right triangle where the
Notes:
115

hypotenuse has length 1 and the side opposite
of the angle with measure y has
length x. This means the nal side has length 1 x2 , using the Pythagorean
Theorem.
1
Therefore cos(sin1 x) = cos y =
1 x2 /1 =
1 x2 , resul ng in
1 x2
)
d(
1
arcsin x = g (x) =
.
dx
1 x2
Figure 2.33: A right triangle dened by

y = sin1 (x/1) with the length of the
third leg found using the Pythagorean
Theorem.
.
Remember that the input x of the arcsine func on is a ra o of a side of a right
triangle to its hypotenuse; the absolute value of this ra o will never be greater
than 1. Therefore the inside of the square root will never be nega ve.
( 3 ,
3
2 )
x
2
4
y = sin x
In Figure 2.34 we see f(x) = sin x and f1 = sin1 x graphed

on their respec ve domains. The line tangent to sin x at the point (/3, 3/2) has slope
1
cos
/3 = 1/2. The slope of the corresponding point on sin x, the point
( 3/2, /3), is
In order to make y = sin x one to one, we restrict its domain to [/2, /2];
on this domain, the range is [1, 1]. Therefore the domain of y = arcsin x is
[1, 1] and the range is [/2, /2]. When x = 1, note how the deriva ve of
the arcsine func on is undened; this corresponds to the fact that as x 1,
the tangent lines to arcsine approach ver cal lines with undened slopes.
, 3 )
1
y = sin1 x
1
4
2
1
Figure 2.34: Graphs of sin x and sin x

along with corresponding tangent lines.
1
1
1
1
=
=
=
= 2,
1/2
1 3/4
1/4
1 ( 3/2)2
verifying yet again that at corresponding points, a func on and its inverse have
reciprocal slopes.
Using similar techniques, we can nd the deriva ves of all the inverse trigonometric func ons. In Figure 2.32 we show the restric ons of the domains of the
standard trigonometric func ons that allow them to be inver ble.
Notes:
116
Func on
Domain
Range
Inverse
Func on
Range
sin x
[/2, /2]
[1, 1]
[1, 1]
[/2, /2]
[0, ]
[1, 1]
cos1 (x)
[1, 1]
[0, ]
(, )
(, )
(/2, /2)
(, 1] [1, )
[/2, 0) (0, /2]
(x)
(, 1] [1, )
[0, /2) (/2, ]
(x)
(, )
(0, )
csc x
(/2, /2)
[/2, 0) (0, /2]
sec x
cot x
[0, /2) (/2, ]

(0, )
(, 1] [1, )
(, 1] [1, )
(, )
tan
(x)
csc
sec
cot
Domain
cos x
tan x
sin
Figure 2.32: Domains and ranges of the trigonometric and inverse trigonometric func ons.
Theorem 23
Deriva ves of Inverse Trigonometric Func ons
The inverse trigonometric func ons are dieren able on their domains
(as listed in Figure 2.32) and their deriva ves are as follows:
d ( 1 )
1
sin (x) =
dx
1 x2
d ( 1 )
1
2.
sec (x) =
dx
|x| x2 1
)
d(
1
3.
tan1 (x) =
dx
1 + x2
1.
)
d(
1
cos1 (x) =
dx
1 x2
d ( 1 )
1
5.
csc (x) =
dx
|x| x2 1
d ( 1 )
1
6.
cot (x) =
dx
1 + x2
4.
Note how the last three deriva ves are merely the opposites of the rst
three, respec vely. Because of this, the rst three are used almost exclusively
throughout this text.
)
d(
1
In Sec on 2.3, we stated without proof or explana on that
ln x = .
dx
x
We can jus fy that now using Theorem 22, as shown in the example.
. Example 74
.Finding the deriva ve of y = ln x

)
d(
Use Theorem 22 to compute
ln x .
dx
View y = ln x as the inverse of y = ex . Therefore, using our

standard nota on, let f(x) = ex and g(x) = ln x. We wish to nd g (x). Theorem
S
Notes:
117

22 gives:
1
f (g(x))
1
= ln x
e
1
= .
x
g (x) =
In this chapter we have dened the deriva ve, given rules to facilitate its
computa on, and given the deriva ves of a number of standard func ons. We
restate the most important of these in the following theorem, intended to be a
reference for further work.
.
Theorem 24
Glossary of Deriva ves of Elementary Func ons
Let u and v be dieren able func ons, and let a, c and n be a real
numbers, a > 0, n = 0.
1.
3.
5.
7.
9.
11.
13.
15.
17.
19.
Notes:
118
( )
cu = cu
(
)
d
dx u v = uv + u v
(
)
d
dx u(v) = u (v)v
( )
d
dx x = 1
( x)
d
= ex
dx e
(
) 1
d
dx ln x = x
(
)
d
dx sin x = cos x
(
)
d
dx csc x = csc x cot x
(
)
d
2
dx tan x = sec x
( 1 )
d
1
x = 1x
2
dx sin
d
dx
21.
d
dx
23.
d
dx
(
(
)
csc1 x = |x|1x2 1
)
tan1 x =
1
1+x2
2.
4.
6.
8.
10.
12.
14.
16.
18.
20.
)
u v = u v
( ) u vuv
d u
dx v =
v2
(
)
d
dx c = 0
( n)
d
= nxn1
dx x
( x)
d
= ln a ax
dx a
(
)
d
1
1
dx loga x = ln a x
(
)
d
dx cos x = sin x
(
)
d
dx sec x = sec x tan x
(
)
d
2
dx cot x = csc x
(
)
d
1
1
x = 1x
2
dx cos
d
dx
22.
d
dx
24.
d
dx
)
sec1 x =
)
1
cot1 x = 1+x
2
1
|x| x2 1
Exercises 2.7
Terms and Concepts
1. T/F: Every func on has an inverse.
2. In your own words explain what it means for a func on to
be one to one.
14. f(x) = 6e3x

Point= (0, 6)
( )
Evaluate f1 (6)
3. If (1, 10) lies on the graph of y = f(x), what can be said

about the graph of y = f1 (x)?
15. h(t) = sin1 (2t)
4. If (1, 10) lies on the graph of y = f(x) and f (1) = 5, what

can be said about y = f1 (x)?
17. g(x) = tan1 (2x)
16. f(t) = sec1 (2t)

18. f(x) = x sin1 x
Problems
19. g(t) = sin t cos1 t
In Exercises 5 8, verify that the given func ons are inverses.
20. f(t) = ln tet
5. f(x) = 2x + 6 and g(x) = 12 x 3
21. h(x) =
6. f(x) = x2 + 6x + 11, x 3 and
g(x) = x 2 3, x 2
3
, x = 5 and
x5
3 + 5x
g(x) =
, x = 0
x
x+1
8. f(x) =
, x = 1 and g(x) = f(x)
x1
7. f(x) =
In Exercises 9 14, an inver ble func on f(x) is given along

with( a point
) that lies on its graph. Using Theorem 22, evaluate f1 (x) at the indicated value.
9. f(x) = 5x + 10
Point= (2, 20)
( )
Evaluate f1 (20)
10. f(x) = x2 2x + 4, x 1
Point= (3, 7)
( )
Evaluate f1 (7)
11. f(x) = sin 2x, /4 x /4
Point= (/6, 3/2)

( )
Evaluate f1 ( 3/2)
12. f(x) = x3 6x2 + 15x 2
Point= (1, 8)
( )
Evaluate f1 (8)
1
,x0
1 + x2
Point= (1, 1/2)
( )
Evaluate f1 (1/2)
13. f(x) =
sin1 x
cos1 x
22. g(x) = tan1 ( x)

23. f(x) = sec1 (1/x)
24. f(x) = sin(sin1 x)
In Exercises 25 27, compute the deriva ve of the given funcon in two ways:
(a) By simplifying rst, then taking the deriva ve, and
(b) by using the Chain Rule rst then simplifying.
Verify that the two answers are the same.
25. f(x) = sin(sin1 x)
26. f(x) = tan1 (tan x)
27. f(x) = sin(cos1 x)
In Exercises 28 29, nd the equa on of the line tangent to
the graph of f at the indicated x value.
28. f(x) = sin1 x
at x =
29. f(x) = cos1 (2x)
2
2
at x =
3
4
Review
30. Find
dy
,
dx
where x2 y y2 x = 1.
31. Find the equa on of the line tangent to the graph of x2 +

y2 + xy = 7 at the point (1, 2).
32. Let f(x) = x3 + x.
f(x + s) f(x)
Evaluate lim
.
s0
s
3: T
F
B
y
Our study of limits led to con nuous func ons, which is a certain class of funcons that behave in a par cularly nice way. Limits then gave us an even nicer
class of func ons, func ons that are dieren able.
This chapter explores many of the ways we can take advantage of the informa on that con nuous and dieren able func ons provide.
3.1
(
.2
Extreme Values
(a)
Given any quan ty described by a func on, we are o en interested in the largest
and/or smallest values that quan ty a ains. For instance, if a func on describes
the speed of an object, it seems reasonable to want to know the fastest/slowest
the object traveled. If a func on describes the value of a stock, we might want
to know how the highest/lowest values the stock a ained over the past year.
We call such values extreme values.
Deni on 12
Extreme Values
Let f be dened on an interval I containing c.
(b)
1. f(c) is the minimum (also, absolute minimum) of f on I if f(c)

f(x) for all x in I.
2. f(c) is the maximum (also, absolute maximum) of f on I if f(c)

f(x) for all x in I.
The maximum and minimum values are the extreme values, or extrema,
of f on I.
Consider Figure 3.1. The func on displayed in (a) has a maximum, but no
minimum, as the interval over which the func on is dened is open. In (b), the
func on has a minimum, but no maximum; there is a discon nuity in the natural place for the maximum to occur. Finally, the func on shown in (c) has both a
maximum and a minimum; note that the func on is con nuous and the interval
on which it is dened is closed.
It is possible for discon nuous func ons dened on an open interval to have
both a maximum and minimum value, but we have just seen examples where
(c)
Figure 3.1: Graphs of func ons with and
without extreme values.
Note: The extreme values of a func on

are y values, values the func on a ains,
not the input values.
Chapter 3 The Graphical Behavior of Func ons

they did not. On the other hand, con nuous func ons on a closed interval always have a maximum and minimum value.
.
Theorem 25
The Extreme Value Theorem
Let f be a con nuous func on dened on a closed interval I. Then f has

both a maximum and minimum value on I.
This theorem states that f has extreme values, but it does not oer any advice about how/where to nd these values. The process can seem to be fairly
easy, as the next example illustrates. A er the example, we will draw on lessons
learned to form a more general and powerful method for nding extreme values.
y
(5, 25)
20
(0, 0)
. Example 75
Approxima ng extreme values
Consider f(x) = 2x3 9x2 on I = [1, 5], as graphed in Figure 3.2. Approximate
the extreme values of f.
(1, 11)
20
The graph is drawn in such a way to draw a en on to certain

S
points. It certainly seems that the smallest y value is 27, found when x = 3.
It also seems that the largest y value is 25, found at the endpoint of I, x = 5.
We use the word seems, for by the graph alone we cannot be sure the smallest
value is not less than 27. Since the problem asks for an approxima on, we
approximate the extreme values to be 25 and 27. .
(3, 27)
.
Figure 3.2: A graph of f(x) = 2x3 9x2 as
in Example 75.
No ce how the minimum value came at the bo om of a hill, and the maximum value came at an endpoint. Also note that while 0 is not an extreme value,
it would be if we narrowed our interval to [1, 4]. The idea that the point (0, 0)
is the loca on of an extreme value for some interval is important, leading us to
a deni on.
Note: The terms local minimum and local

maximum are o en used as synonyms for
rela ve minimum and rela ve maximum.
Deni on 13
Rela ve Minimum and Rela ve Maximum
Let f be dened on an interval I containing c.

1. If there is an open interval containing c such that f(c) is the minimum value, then
f(c) is a rela ve minimum of f. We also say that f has a rela ve minimum at (c, f(c)).
2. If there is an open interval containing c such that f(c) is the maximum value, then
f(c) is a rela ve maximum of f. We also say that f has a rela ve maximum at (c, f(c)).
The rela ve maximum and minimum values comprise the rela ve extrema of f.
Notes:
122
3.1 Extreme Values

We briey prac ce using these deni ons.
y
6
. Example 76
Approxima ng rela ve extrema
Consider f(x) = (3x4 4x3 12x2 + 5)/5, as shown in Figure 3.3. Approximate
the rela ve extrema of f. At each of these points, evaluate f .
4
2
x
1
We s ll do not have the tools to exactly nd the rela ve

extrema, but the graph does allow us to make reasonable approxima ons. It
seems f has rela ve minima at x = 1 and x = 2, with values of f(1) = 0 and
f(2) = 5.4. It also seems that f has a rela ve maximum at the point (0, 1).
S
We approximate the rela ve minima to be 0 and 5.4; we approximate the

rela ve maximum to be 1.
4
6
Figure 3.3: A graph of f(x) = (3x4 4x3

12x2 + 5)/5 as in Example 76.
It is straigh orward to evaluate f (x) = 51 (12x3 12x2 24x) at x = 0, 1

and 2. In each case, f (x) = 0. .
y
. Example 77
Approxima ng rela ve extrema
Approximate the rela ve extrema of f(x) = (x 1)2/3 + 2, shown in Figure 3.4.
At each of these points, evaluate f .
The gure implies that f does not have any rela ve maxima,
S
but has a rela ve minimum at (1, 2). In fact, the graph suggests that not only is
this point a rela ve minimum, y = f(1) = 2 the minimum value of the func on.
We compute f (x) = 32 (x 1)1/3 . When x = 1, f is undened...
What can we learn from the previous two examples? We were able to visually approximate rela ve extrema, and at each such point, the deriva ve was
either 0 or it was not dened. This observa on holds for all func ons, leading
to a deni on and a theorem.
Deni on 14
Figure 3.4: A graph of f(x) = (x1)2/3 +2

as in Example 77.
Cri cal Numbers and Cri cal Points
Let f be dened at c. The value c is a cri cal number (or cri cal value)
of f if f (c) = 0 or f (c) is not dened.
If c is a cri cal number of f, then the point (c, f(c)) is a cri cal point of f.
Notes:
123

.
Theorem 26
Rela ve Extrema and Cri cal Points
Let a func on f have a rela ve extrema at the point (c, f(c)). Then c is a
cri cal number of f.
x
1
Figure 3.5: A graph of f(x) = x3 which has

a cri cal value of x = 0, but no rela ve
extrema.
Be careful to understand that this theorem states All rela ve extrema occur
at cri cal points. It does not say All cri cal numbers produce rela ve extrema.
For instance, consider f(x) = x3 . Since f (x) = 3x2 , it is straigh orward to determine that x = 0 is a cri cal number of f. However, f has no rela ve extrema,
as illustrated in Figure 3.5.
Theorem 25 states that a con nuous func on on a closed interval will have
absolute extrema, that is, both an absolute maximum and an absolute minimum.
These extrema occur either at the endpoints or at cri cal values in the interval.
We combine these concepts to oer a strategy for nding extrema.
Key Idea 2
Finding Extrema on a Closed Interval
Let f be a con nuous func on dened on a closed interval [a, b]. To nd

the maximum and minimum values of f on [a, b]:
1. Evaluate f at the endpoints a and b of the interval.

2. Find the cri cal numbers of f in [a, b].
3. Evaluate f at each cri cal number.
4. The absolute maximum of f is the largest of these values, and the
absolute minimum of f is the least of these values.
y
We prac ce these ideas in the next examples.
40
. Example 78
.Finding extreme values
Find the extreme values of f(x) = 2x3 + 3x2 12x on [0, 3], graphed in Figure
3.6.
20
S
x
f(0) = 0 and
Figure 3.6: A graph of f(x) = 2x3 + 3x2

12x on [0, 3] as in Example 78.
f(3) = 45.
Next, we nd the cri cal values of f on [0, 3]. f (x) = 6x2 + 6x 12 = 6(x +
2)(x 1); therefore the cri cal values of f are x = 2 and x = 1. Since x = 2
Notes:
124
We follow the steps outlined in Key Idea 2. We rst evaluate
f at the endpoints:
3.1 Extreme Values

does not lie in the interval [0, 3], we ignore it. Evalua ng f at the only cri cal
number in our interval gives: f(1) = 7.
The table in Figure 3.7 gives f evaluated at the important x values in [0, 3].
We can easily see the maximum and minimum values of f: the maximum value
is 45. and the minimum value is 7.
Note that all this was done without the aid of a graph; this work followed
an analy c algorithm and did not depend on any visualiza on. Figure 3.6 shows
f and we can conrm our answer, but it is important to understand that these
answers can be found without graphical assistance.
We prac ce again.
f(x)
0
1
3
0
7
45
Figure 3.7: Finding the extreme values of

f in Example 78.
. Example 79
Finding extreme values
Find the maximum and minimum values of f on [4, 2], where
{
(x 1)2 x 0
f(x) =
.
x+1
x>0
Here f is piecewisedened, but we can s ll apply Key Idea

2. Evalua ng f at the endpoints gives:
S
f(4) = 25 and
f(2) = 3.
f(x)
4
0
2
25
1
3
Figure 3.8: Finding the extreme values of

f in Example 79.
We now nd the cri cal numbers of f. We have to dene f in a piecewise

manner; it is
{
2(x 1) x < 0
f (x) =
.
1
x>0
Note that while f is dened for all of [4, 2], f is not, as the deriva ve of f does
not exist when x = 0. (From the le , the deriva ve approaches 2; from the
right the deriva ve is 1.) Thus one cri cal number of f is x = 0.
We now set f (x) = 0. When x > 0, f (x) is never 0. When x < 0, f (x) is
also never 0. (We may be tempted to say that f (x) = 0 when x = 1. However,
this is nonsensical, for we only consider f (x) = 2(x 1) when x < 0, so we will
ignore a solu on that says x = 1.
So we have three important x values to consider: x = 4, 2 and 0. Evalua ng f at each gives, respec vely, 25, 3 and 1, shown in Figure 3.8. Thus the
absolute minimum of f is 1; the absolute maximum of f is 25. Our answer is conrmed by the graph of f in Figure 3.9. .
20
10
x
4
Figure 3.9: A graph of f(x) on [4, 2] as in

Example 79.
Notes:
125

x
f(x)
0.65
1
1
1
0.65
. Example 80
Finding extreme values
Find the extrema of f(x) = cos(x2 ) on [2, 2].
Figure 3.10: Finding the extrema of

f(x) = cos(x2 ) in Example 80.
y
1
0.5
x
2
0.5
We consider one more example.
Figure 3.11: A graph of f(x) = cos(x2 ) on

[2, 2] as in Example 80.
f(x)
1
0
1
0
1
0
Figure 3.13: Finding the extrema of the

halfcircle in Example 81.
y
1
Figure 3.12: A graph of f(x) =

on [1, 1] as in Example 81.
. Example 81
Finding extreme
values
Find the extreme values of f(x) = 1 x2 .
A closed interval is not given, so we nd the extreme values
S
of f on its domain. f is dened whenever 1 x2 0; thus the domain of f is
[1, 1]. Evalua ng f at either endpoint returns 0.
x
Using the Chain Rule, we nd f (x) =
. The cri cal points of f are
1 x2
found when f (x) = 0 or when f is undened. It is straigh orward to nd that

f (x) = 0 when x = 0, and f is undened when x = 1, the endpoints of the
interval. The table of important values is given in Figure 3.13. .
We have seen that con nuous func ons on closed intervals always have a
maximum and minimum value, and we have also developed a technique to nd
these values. In the next sec on, we further our study of the informa on we can
glean from nice func ons with the Mean Value Theorem. On a closed interval,
we can nd the average rate of change of a func on (as we did at the beginning
of Chapter 2). We will see that dieren able func ons always have a point at
which their instantaneous rate of change is same as the average rate of change.
This is surprisingly useful, as well see.
1 x2
Note: We implicitly found the deriva ve

of x2 + y2 = 1, the unit circle, in Exam= x/y. In Example 81, half
ple 69 as dy
dx
of
the
unit
circle
is given as y = f(x) =
1 x2 . We found f (x) = x 2 . Rec1x
ognize that the denominator of this fracon is y; that is, we again found f (x) =
dy
= x/y.
dx
126
We again use Key Idea 2. Evalua ng f at the endpoints of

S
the interval gives: f(2) = f(2) = cos(4) 0.6536. We now nd the cri cal
values of f.
Applying the Chain Rule, we nd f (x) = 2x sin(x2 ). Set f (x) = 0 and
solve for x to nd the cri cal values of f.
We have f (x) = 0 when x = 0 and when sin(x2 ) = 0. In general, sin t = 0
when t = . . . 2, , 0, , . . . Thus sin(x2 ) = 0 when x2 = 0, ,
2, . . .
(x2 is
2
always posi ve so we ignore , etc.) So sin(x ) = 0 when x= 0, , 2, . . ..
The only values to fall in the given interval of [2, 2] are and , approximately 1.77.
We again construct a table of important values
in Figure 3.10. In this example
we have 5 values to consider: x = 0, 2, .

From the table it is clear that the maximum value of f on [2, 2] is 1; the
minimum value is 1. The graph in Figure 3.11 conrms our results. .
Notes:
Exercises 3.1
9. f(x) = x2 6 x2
Terms and Concepts
1. Describe what an extreme value of a func on is in your

own words.
(2, 4 2)
2. Sketch the graph of a func on f on (1, 1) that has both a

maximum and minimum value.
3. Describe the dierence between and absolute and rela ve
maximum in your own words.
4. Sketch the graph of a func on f where f has a rela ve maximum at x = 1 and f (1) is undened.
x
2
(0, 0)
10. f(x) = sin x

y
(/2, 1)
5. T/F: If c is a cri cal value of a func on f, then f has either a

rela ve maximum or rela ve minimum at x = c.
x
4
Problems
1
In Exercises 6 7, iden fy each of the marked points as being

an absolute maximum or minimum, a rela ve maximum or
minimum, or none of the above.
(3/2, 1)
11. f(x) = x2 4 x
y
y
B
512
25 5
6.
( 16
10
x
4
(4, 0)
F
2
. 2
(0, 0)
{
12. f(x) =
y
C
x
4
x0
x>0
x2
x5
y
7.
x
0.5
E
2
x
1
In Exercises 8 14, evaluate f (x) at the indicated points.
0.5
0.5
{
13. f(x) =
2
8. f(x) = 2
x +1
(0, 0) 0.5
x0
x>0
x2
x
y
y
1
(0, 2)
2
0.5
1
x
1
0.5
(0, 0) 0.5
x
5
0.5
14. f(x) =
(x 2)2/3
x
19. f(x) = x +
(
6, 1 +
3
2
3
[3, 5].
21. f(x) = ex cos x
on
[0, ].
on
[0, ].
22. f(x) = e sin x

ln x
23. f(x) =
x
(2, 1)
x
5
16. f(x) = x3
17. f(x) = 3 sin x on [/4, 2/3].
18. f(x) = x2 4 x2 on [2, 2].
[0, 6].
[1, 4].
on
[0, 2].
Review
25. Find
[1, 2].
9 2
x 30x + 3 on
2
on
24. f(x) = x2/3 x
10
In Exercises 15 24, nd the extreme values of the func on

on the given interval.
15. f(x) = x2 + x + 4 on
x2
x
[1, 5].
on
20. f(x) =
on
x2
+5
y
6
3
x
dy
,
dx
where x2 y y2 x = 1.
26. Find the equa on of the line tangent to the graph of x2 +

y2 + xy = 7 at the point (1, 2).
27. Let f(x) = x3 + x.
f(x + s) f(x)
Evaluate lim
.
s0
s
3.2 The Mean Value Theorem
3.2
The Mean Value Theorem
We mo vate this sec on with the following ques on: Suppose you leave your
house and drive to your friends house in a city 100 miles away, comple ng the
trip in two hours. At any point during the trip do you necessarily have to be going
50 miles per hour?
In answering this ques on, it is clear that the average speed for the en re
trip is 50 mph (i.e. 100 miles in 2 hours), but the ques on is whether or not your
instantaneous speed is ever exactly 50 mph. More simply, does your speedometer ever read exactly 50 mph?. The answer, under some very reasonable assump ons, is yes.
Lets now see why this situa on is in a calculus text by transla ng it into
mathema cal symbols.
First assume that the func on y = f(t) gives the distance (in miles) traveled
from your home at me t (in hours) where 0 t 2. In par cular, this gives
f(0) = 0 and f(2) = 100. The slope of the secant line connec ng the star ng
and ending points (0, f(0)) and (2, f(2)) is therefore
f
f(2) f(0)
100 0
=
=
= 50 mph.
t
20
2
The slope at any point on the graph itself is given by the deriva ve f (t). So,
since the answer to the ques on above is yes, this means that at some me
during the trip, the deriva ve takes on the value of 50 mph. Symbolically,
f (c) =
f(2) f(0)
= 50
20
for some me 0 c 2.
How about more generally? Given any func on y = f(x) and a range a
x b does the value of the deriva ve at some point between a and b have to
match the slope of the secant line connec ng the points (a, f(a)) and (b, f(b))?
Or equivalently, does the equa on f (c) = f(b)f(a)
have to hold for some a <
ba
c < b?
Lets look at two func ons in an example.
. Example 82
.Comparing average and instantaneous rates of change
Consider func ons
1
f1 (x) = 2 and f2 (x) = |x|
x
with a = 1 and b = 1 as shown in Figure 3.14(a) and (b), respec vely. Both
func ons have a value of 1 at a and b. Therefore the slope of the secant line
Notes:
129

connec ng the end points is 0 in each case. But if you look at the plots of each
(below), you can see that there are no points on either graph where the tangent
lines have slope zero. Therefore we have found that there is no c in [1, 1] such
that
f(1) f(1)
= 0.
f (c) =
1 (1)
.
x
1
..
(a)
y
So what went wrong? It may not be surprising to nd that the discon nuity
of f1 and the corner of f2 play a role. If our func ons had been con nuous and
dieren able, would we have been able to nd that special value c? This is our
mo va on for the following theorem.
Theorem 27
The Mean Value Theorem of Dieren a on
Let y = f(x) be con nuous func on on the closed interval [a, b] and
dieren able on the open interval (a, b). There exists a value c, a < c <
b, such that
f(b) f(a)
.
f (c) =
ba
0.5
That is, there is a value c in (a, b) where the instantaneous rate of change
of f at c is equal to the average rate of change of f on [a, b].
(b)
Figure 3.14: A graph of f1 (x) = 1/x2 and
f2 (x) = |x| in Example 82.
Note that the reasons that the func ons in Example 82 fail are indeed that
f1 has a discon nuity on the interval [1, 1] and f2 is not dieren able at the
origin.
We will give a proof of the Mean Value Theorem below. To do so, we use a
fact, called Rolles Theorem, stated here.
Theorem 28
Rolles Theorem
Let f be con nuous on [a, b] and dieren able on (a, b), where f(a) =
f(b). There is some c in (a, b) such that f (c) = 0.
x
1 b
Figure 3.15: A graph of f(x) = x3 5x2 +

3x + 5, where f(a) = f(b). Note the existence of c, where a < c < b, where
f (c) = 0.
130
Consider Figure 3.15 where the graph of a func on f is given, where f(a) =
f(b). It should make intui ve sense that if f is dieren able (and hence, con nuous) that there would be a value c in (a, b) where f (c) = 0; that is, there would
be a rela ve maximum or minimum of f in (a, b). Rolles Theorem guarantees at
least one; there may be more.
Notes:
3.2 The Mean Value Theorem

Rolles Theorem is really just a special case of the Mean Value Theorem. If
f(a) = f(b), then the average rate of change on (a, b) is 0, and the theorem
guarantees some c where f (c) = 0. We will prove Rolles Theorem, then use it
to prove the Mean Value Theorem.
Proof of Rolles Theorem
Let f be dieren able on (a, b) where f(a) = f(b). We consider two cases.
Case 1: Consider the case when f is constant on [a, b]; that is, f(x) = f(a) = f(b)
for all x in [a, b]. Then f (x) = 0 for all x in [a, b], showing there is at least one
value c in (a, b) where f (c) = 0.
Case 2: Now assume that f is not constant on [a, b]. The Extreme Value Theorem
guarantees that f has a maximal and minimal value on [a, b], found either at the
endpoints or at a cri cal value in (a, b). Since f(a) = f(b) and f is not constant, it
is clear that the maximum and minimum cannot both be found at the endpoints.
Assume, without loss of generality, that the maximum of f is not found at the
endpoints. Therefore there is a c in (a, b) such that f(c) is the maximum value
of f. By Theorem 26, c must be a cri cal number of f; since f is dieren able, we
have that f (c) = 0, comple ng the proof of the theorem.
We can now prove the Mean Value Theorem.

Proof of the Mean Value Theorem
Dene the func on
g(x) = f(x)
f(b) f(a)
x.
ba
We know g is dieren able on (a, b) and con nuous on [a, b] since f is. We can
show g(a) = g(b) (it is actually easier to show g(b) g(a) = 0, which suces).
We can then apply Rolles theorem to guarantee the existence of c (a, b) such
that g (c) = 0. But note that
0 = g (c) = f (c)
hence
f (c) =
which is what we sought to prove.
f(b) f(a)
;
ba
f(b) f(a)
,
ba
Going back to the very beginning of the sec on, we see that the only assump on we would need about our distance func on f(t) is that it be con nuous and dieren able for t from 0 to 2 hours (both reasonable assump ons). By
the Mean Value Theorem, we are guaranteed a me during the trip where our
Notes:
131

instantaneous speed is 50 mph. This fact is used in prac ce. Some law enforcement agencies monitor trac speeds while in aircra . They do not measure
speed with radar, but rather by ming individual cars as they pass over lines
painted on the highway whose distances apart are known. The ocer is able
to measure the average speed of a car between the painted lines; if that average speed is greater than the posted speed limit, the ocer is assured that the
driver exceeded the speed limit at some me.
Note that the Mean Value Theorem is an existence theorem. It states that a
special value c exists, but it does not give any indica on about how to nd it. It
turns out that when we need the Mean Value Theorem, existence is all we need.
. Example 83
Using the Mean Value Theorem
Consider f(x) = x3 + 5x + 5 on [3, 3]. Find c in [3, 3] that sa ses the Mean
Value Theorem.
S
The average rate of change of f on [3, 3] is:

84
f(3) f(3)
=
= 14.
3 (3)
6
y
40
We want to nd c such that f (c) = 14. We nd f (x) = 3x2 + 5. We set this

equal to 14 and solve for x.
20
f (x) = 14
x
3
3x2 + 5 = 14
20
x2 = 3
x = 3 1.732
40
Figure 3.16: Demonstra ng the Mean

Value Theorem in Example 83.
We have found 2 values c in [3, 3] where the instantaneous rate of change

is equal to the average rate of change; the Mean Value Theorem guaranteed at
least one. In Figure 3.16 f is graphed with a dashed
line represen ng the average rate of change; the lines tangent to f at x = 3 are also given. Note how
these lines are parallel (i.e., have the same slope) as the dashed line. .
While the Mean Value Theorem has prac cal use (for instance, the speed
monitoring applica on men oned before), it is mostly used to advance other
theory. We will use it in the next sec on to relate the shape of a graph to its
deriva ve.
Notes:
132
Exercises 3.2
Terms and Concepts
11. f(x) = x2 + 3x 1 on [2, 2].
1. Explain in your own words what the Mean Value Theorem

states.
12. f(x) = 5x2 6x + 8 on [0, 5].
13. f(x) = 9 x2 on [0, 3].
14. f(x) = 25 x on [0, 9].
2. Explain in your own words what Rolles Theorem states.
Problems
In Exercises 3 10, a func on f(x) and interval [a, b] are given.
Check if Rolles Theorem can be applied to f on [a, b]; if so, nd
c in [a, b] such that f (c) = 0.
15. f(x) =
x2 9
on [0, 2].
x2 1
16. f(x) = ln x on [1, 5].
3. f(x) = 6 on [1, 1].
17. f(x) = tan x on [/4, /4].
4. f(x) = 6x on [1, 1].
18. f(x) = x3 2x2 + x + 1 on [2, 2].
5. f(x) = x2 + x 6 on [3, 2].

6. f(x) = x2 + x 2 on [3, 2].
19. f(x) = 2x3 5x2 + 6x + 1 on [5, 2].
7. f(x) = x2 + x on [2, 2].
20. f(x) = sin1 x on [1, 1].
8. f(x) = sin x on [/6, 5/6].

9. f(x) = cos x on [0, ].
1
10. f(x) = 2
on [0, 2].
x 2x + 1
In Exercises 11 20, a func on f(x) and interval [a, b] are
given. Check if the Mean Value Theorem can be applied to f
on [a, b]; if so, nd a value c in [a, b] guaranteed by the Mean
Value Theorem.
Review
21. Find the extreme values of f(x) = x2 3x + 9 on [2, 5].
22. Describe the cri cal points of f(x) = cos x.
23. Describe the cri cal points of f(x) = tan x.

Our study of nice func ons f in this chapter has so far focused on individual
points: points where f is maximal/minimal, points where f (x) = 0 or f does
not exist, and points c where f (c) is average rate of change of f on some interval.
In this sec on we begin to study how func ons behave between special
points; we begin studying in more detail the shape of their graphs.
We start with an intui ve concept. Given the graph in Figure 3.17, where
would you say the func on is increasing? Decreasing? Even though we have
not dened these terms mathema cally, one likely answered that f is increasing
when x > 1 and decreasing when x < 1. We formally dene these terms here.
Deni on 15
Figure 3.17: A graph of a func on f used

to illustrate the concepts of increasing
and decreasing.
Increasing and Decreasing Func ons
Let f be a func on dened on an interval I.
1. f is increasing on I if for every a < b in I, f(a) f(b).

2. f is decreasing on I if for every a < b in I, f(a) f(b).
A func on is strictly increasing when a < b in I implies f(a) < f(b), with
a similar deni on holding for strictly decreasing.
Informally, a func on is increasing if as x gets larger (i.e., looking le to right)

f(x) gets larger.
Our interest lies in nding intervals in the domain of f on which f is either
increasing or decreasing. Such informa on should seem useful. For instance, if
f describes the speed of an object, we might want to know when the speed was
increasing or decreasing (i.e., when the object was accelera ng vs. decelerating). If f describes the popula on of a city, we should be interested in when the
popula on is growing or declining.
To nd such intervals, we again consider secant lines. Let f be an increasing
func on on an interval I, such as the one shown in Figure 3.18, and let a < b be
given in I. The secant line on the graph of f from x = a to x = b is drawn; it has
a slope of (f(b) f(a))/(b a). But note:
y
2
(b, f(b))
(a, f(a))
x
a
Figure 3.18: Examining the secant line of

an increasing func on.
Average rate of
f(b) f(a)
numerator > 0
slope of the
change of f on
secant line > 0
ba
denominator > 0
[a, b] is > 0.
We have shown mathema cally what may have already been obvious: when
f is increasing, its secant lines will have a posi ve slope. Now recall the Mean
Notes:
134

Value Theorem guarantees that there is a number c, where a < c < b, such that
f (c) =
f(b) f(a)
> 0.
ba
By considering all such secant lines in I, we can conclude that f (x) > 0 on I. A
similar statement can be made for decreasing func ons.
This leads us to a method for nding when func ons are increasing and decreasing, as stated in the following theorem.
Theorem 29
Test For Increasing/Decreasing Func ons
Let f be a con nuous func on on [a, b] and dieren able on (a, b).
1. If f (c) > 0 for all c in (a, b), then f is increasing on [a, b].
Note: Theorem 29 also holds if f (c) = 0

for a nite number of values of c in I.
2. If f (c) < 0 for all c in (a, b), then f is decreasing on [a, b].
3. If f (c) = 0 for all c in (a, b), then f is constant on [a, b].
Let a and b be in I where f (a) > 0 and f (b) < 0. It follows from the
Intermediate Value Theorem that there must be some value c between a and b
where f (c) = 0. This leads us to the following method for nding intervals on
which a func on is increasing or decreasing.
Key Idea 3
Finding Intervals on Which f is Increasing or Decreasing
Let f be a dieren able func on on an interval I. To nd intervals on

which f is increasing and decreasing:
1. Find the cri cal values of f. That is, nd all c in I where f (c) = 0
or f is not dened.
2. Use the cri cal values to divide I into subintervals.

3. Pick any point p in each subinterval, and nd the sign of f (p).
(a) If f (p) > 0, then f is increasing on that subinterval.
(b) If f (p) < 0, then f is decreasing on that subinterval.
We demonstrate using this process in the following example.
Notes:
135

. Example 84
Finding intervals of increasing/decreasing
Let f(x) = x3 + x2 x + 1. Find intervals on which f is increasing or decreasing.
Using Key Idea 3, we rst nd the cri cal values of f. We
S
have f (x) = 3x2 + 2x 1 = (3x 1)(x + 1), so f (x) = 0 when x = 1 and
when x = 1/3. f is never undened.
Since an interval was not specied for us to consider, we consider the enre domain of f which is (, ). We thus break the whole real line into
three subintervals based on the two cri cal values we just found: (, 1),
(1, 1/3) and (1/3, ). This is shown in Figure 3.19.
f > 0 incr
f < 0 decr
f > 0 incr
.
1
1/3
Figure 3.19: Number line for f in Example 84.
We now pick a value p in each subinterval and nd the sign of f (p). All we
care about is the sign, so we do not actually have to fully compute f (p); pick
nice values that make this simple.
Subinterval 1, (, 1): We (arbitrarily) pick p = 2. We can compute
f (2) directly: f (2) = 3(2)2 + 2(2) 1 = 7 > 0. We conclude that f is
increasing on (, 1).
Note we can arrive at the same conclusion without computa on. For instance, we could choose p = 100. The rst term in f (100), i.e., 3(100)2 is
clearly posi ve and very large. The other terms are small in comparison, so we
know f (100) > 0. All we need is the sign.
y
10
x
2
1/3
.
Figure 3.20: A graph of f(x) in Example
84, showing where f is increasing and decreasing.
Subinterval 2, (1, 1/3): We pick p = 0 since that value seems easy to deal
with. f (0) = 1 < 0. We conclude f is decreasing on (1, 1/3).
Subinterval 3, (1/3, ): Pick an arbitrarily large value for p > 1/3 and note
that f (p) = 3p2 + 2p 1 > 0. We conclude that f is increasing on (1/3, ).
We can verify our calcula ons by considering Figure 3.20, where f is graphed.
The graph also presents f in red; note how f > 0 when f is increasing and
f < 0 when f is decreasing. .
One is jus ed in wondering why so much work is done when the graph
seems to make the intervals very clear. We give three reasons why the above
work is worthwhile.
First, the points at which f switches from increasing to decreasing are not
precisely known given a graph. The graph shows us something signicant hap-
Notes:
136

pens near x = 1 and x = 0.3, but we cannot determine exactly where from
the graph.
One could argue that just nding cri cal values is important; once we know
the signicant points are x = 1 and x = 1/3, the graph shows the increasing/decreasing traits just ne. That is true. However, the technique prescribed
here helps reinforce the rela onship between increasing/decreasing and the
sign of f . Once mastery of this concept (and several others) is obtained, one
nds that either (a) just the cri cal points are computed and the graph shows
all else that is desired, or (b) a graph is never produced, because determining
increasing/decreasing using f is straigh orward and the graph is unnecessary.
So our second reason why the above work is worthwhile is this: once mastery
of a subject is gained, one has op ons for nding needed informa on. We are
working to develop mastery.
Finally, our third reason: many problems we face in the real world are very
complex. Solu ons are tractable only through the use of computers to do many
calcula ons for us. Computers do not solve problems on their own, however;
they need to be taught (i.e., programmed) to do the right things. It would be
benecial to give a func on to a computer and have it return maximum and
minimum values, intervals on which the func on is increasing and decreasing,
the loca ons of rela ve maxima, etc. The work that we are doing here is easily
programmable. It is hard to teach a computer to look at the graph and see if it
is going up or down. It is easy to teach a computer to determine if a number
is greater than or less than 0.
In Sec on 3.1 we learned the deni on of rela ve maxima and minima and
found that they occur at cri cal points. We are now learning that func ons can
switch from increasing to decreasing (and viceversa) at cri cal points. This new
understanding of increasing and decreasing creates a great method of determining whether a cri cal point corresponds to a maximum, minimum, or neither.
Imagine a func on increasing un l a cri cal point at x = c, a er which it decreases. A quick sketch helps conrm that f(c) must be a rela ve maximum. A
similar statement can be made for rela ve minimums. We formalize this concept in a theorem.
Theorem 30
First Deriva ve Test
Let f be dieren able on I and let c be a cri cal number in I.
1. If the sign of f switches from posi ve to nega ve at c, then f(c) is a rela ve maximum of f.
2. If the sign of f switches from nega ve to posi ve at c, then f(c) is a rela ve minimum of f.
3. If the sign of f does not change at c, then f(c) is not a rela ve extrema of f.
Notes:
137

. Example 85
.Using the First Deriva ve Test
Find the intervals on which f is increasing and decreasing, and use the First
Deriva ve Test to determine the rela ve extrema of f, where
f(x) =
x2 + 3
.
x1
We start by calcula ng f using the Quo ent Rule. We nd

f (x) =
x2 2x 3
.
(x 1)2
We need to nd the cri cal values of f; we want to know when f (x) = 0 and
when f is not dened. That la er is straigh orward: when the denominator of
f is 0, f is undened. That occurs when x = 1.
f (x) = 0 when the numerator of f is 0. That occurs when x2 2x 3 =
(x 3)(x + 1) = 0; i.e., when x = 1, 3.
We have found that f has three cri cal numbers, dividing the real number
line into 4 subintervals:
(, 1),
Note: Strictly speaking, x = 1 is not a critical value of f as f is not dened at x = 1.

We therefore actually apply Key Idea 3 to
the intervals (, 1) and (1, ). We
make note of x = 1 on the number line
as we recognize that the behavior of f can
change there, as it is not dened there.
(1, 3) and
(3, ).
Pick a number p from each subinterval and test the sign of f at p to determine
whether f is increasing or decreasing on that interval. Again, we do well to avoid
complicated computa ons; no ce that the denominator of f is always posi ve
so we can ignore it during our work.
Interval 1, (, 1): Choosing a very small number (i.e., a nega ve number
with a large magnitude) p returns p2 2p 3 in the numerator of f ; that will
be posi ve. Hence f is increasing on (, 1).
Interval 2, (1, 1): Choosing 0 seems simple: f (0) = 3 < 0. We conclude
f is decreasing on (1, 1).
Interval 3, (1, 3): Choosing 2 seems simple: f (2) = 3 < 0. Again, f is
decreasing.
Interval 4, (3, ): Choosing an very large number p from this subinterval will
give a posi ve numerator and (of course) a posi ve denominator. So f is increasing on (3, ).
In summary, f is increasing on the set (, 1)(3, ) and is decreasing on
the set (1, 1) (1, 3). Since at x = 1, the sign of f switched from posi ve to
nega ve, Theorem 30 states that f(1) is a rela ve maximum of f. At x = 3, the
sign of f switched from nega ve to posi ve, meaning f(3) is a rela ve minimum.
At x = 1, f is not dened, so there is no rela ve extrema at x = 1.
This is summarized in the number line shown in Figure 3.21. Also, Figure
3.22 shows a graph of f, conrming our calcula ons. This gure also shows f in
Notes:
138
(1, 1),

red, again demonstra ng that f is increasing when f > 0 and decreasing when
f < 0.
f > 0 incr
rel.
max
f < 0 decr
f < 0 decr
rel.
min
y
20
10
f > 0 incr
.
1
10
.
One is o en tempted to think that func ons always alternate increasing,
decreasing, increasing, decreasing,. . . around cri cal values. Our previous example demonstrated that this is not always the case. While x = 1 was a cri cal
value, f was decreasing on both sides of x = 1.
20

We examine one more example.

. Example 86
.Using the First Deriva ve Test
Find the intervals on which f(x) = x8/3 4x2/3 is increasing and decreasing and
iden fy the rela ve extrema.
We again start with taking deriva ves. Since we know we
S
want to solve f (x) = 0, we will do some algebra a er taking deriva ves.
8
f(x) = x 3 4x 3
8 5
8 1
f (x) = x 3 x 3
3
3
)
8 1 ( 6
= x 3 x3 1
3
8 1
= x 3 (x2 1)
3
8 1
= x 3 (x 1)(x + 1)
3
This deriva on of f shows that f (x) = 0 when x = 1 and f is not dened when x = 0. Thus we have 3 cri cal values, breaking the number line into
4 subintervals as shown in Figure 3.23.
Interval 1, (, 1): We choose p = 2; we can easily verify that f (2) < 0.
So f is decreasing on (, 1).
Interval 2, (1, 0): Choose p = 1/2. Once more we prac ce nding the sign
of f (p) without compu ng an actual value. We have f (p) = (8/3)p1/3 (p
Notes:
139

1)(p + 1); nd the sign of each of the three terms.
f (p) =
8 1
p 3 (p 1) (p + 1) .
3 |{z} | {z } | {z }
<0
<0
>0
We have a nega ve nega ve posi ve giving a posi ve number; f is increasing on (1, 0).
Interval 3, (0, 1): We do a similar sign analysis as before, using p in (0, 1).
f (p) =
8 1
p 3 (p 1) (p + 1) .
3 |{z} | {z } | {z }
>0
<0
>0
We have 2 posi ve factors and one nega ve factor; f (p) < 0 and so f is decreasing on (0, 1).
Interval 4, (1, ): Similar work to the three intervals shows that f (x) > 0 on
(1, ), so f is increasing on this interval.
y
10
f < 0 decr
rel.
min
f > 0 incr
rel.
max
f < 0 decr
rel.
min
f > 0 incr
.
1

x
3
.
We conclude by sta ng that f is increasing on (1, 0)(1, ) and decreasing

on (, 1) (0, 1). The sign of f changes from nega ve to posi ve around
x = 1 and x = 1, meaning by Theorem 30 that f(1) and f(1) are rela ve
minima of f. As the sign of f changes from posi ve to nega ve at x = 0, we
have a rela ve maximum at f(0). Figure 3.24 shows a graph of f, conrming our
result. Once again f is graphed in red, highligh ng once more that f is increasing
when f > 0 and is decreasing when f < 0. .
Notes:
140
Exercises 3.3
Terms and Concepts
In Exercises 14 23, a func on f(x) is given.
1. In your own words describe what it means for a func on to

be increasing.
2. What does a decreasing func on look like?
3. Sketch a graph of a func on on [0, 2] that is increasing but
not strictly increasing.
4. Give an example of a func on describing a situa on where
it is bad to be increasing and good to be decreasing.
2
5. A func on f has deriva ve f (x) = (sin x + 2)ex +1 , where

f (x) > 1 for all x. Is f increasing, decreasing, or can we not
tell from the given informa on?
(a) Find the cri cal numbers of f.

(b) Create a number line to determine the intervals on
which f is increasing and decreasing.
(c) Use the First Deriva ve Test to determine whether
each cri cal point is a rela ve maximum, minimum,
or neither.
14. f(x) = x2 + 2x 3
15. f(x) = x3 + 3x2 + 3
16. f(x) = 2x3 + x2 x + 3
17. f(x) = x3 3x2 + 3x 1
Problems
18. f(x) =
19. f(x) =
(a) Compute f (x).

(b) Graph f and f on the same axes (using technology is
permi ed) and verify Theorem 29.
6. f(x) = 2x + 3
1
x2 2x + 2
x2 4
x2 1
x
20. f(x) = 2
x 2x 8
21. f(x) =
(x 2)2/3
x
7. f(x) = x2 3x + 5
22. f(x) = sin x cos x on (, ).
8. f(x) = cos x
23. f(x) = x5 5x
9. f(x) = tan x
10. f(x) = x3 5x2 + 7x 1
11. f(x) = 2x3 x2 + x 1
12. f(x) = x4 5x2 + 4
13. f(x) =
x2
1
+1
Review
24. Consider f(x) = x2 3x + 5 on [1, 2]; nd c guaranteed
by the Mean Value Theorem.
25. Consider f(x) = sin x on [/2, /2]; nd c guaranteed by
the Mean Value Theorem.

y
30
20
10
x
2
Figure 3.25: A func on f with a concave

up graph. No ce how the slopes of the
tangent lines, when looking from le to
right, are increasing.
Note: A mnemonic for remembering

what concave up/down means is: Concave up is like a cup; concave down is like
a frown. It is admi edly terrible, but it
works.
A mnemonic for remembering how
to pronounce mnemonic is to recall it begins with the same sound as
mnemotechnic.
y
30
20
10
x
2
We begin with a deni on, then explore its meaning.

Concave Up and Concave Down
Let f be dieren able on an interval.

I. The graph of f is concave up on I
if f is increasing. The graph of f is concave down on I if f is decreasing.
If f is constant then the graph of f is said to have no concavity.
The graph of a func on f is concave up when f is increasing. That means
as one looks at a concave up graph from le to right, the slopes of the tangent
lines will be increasing. Consider Figure 3.25, where a concave up graph is shown
along with some tangent lines. No ce how the tangent line on the le is steep,
downward, corresponding to a small value of f . On the right, the tangent line
is steep, upward, corresponding to a large value of f .
If a func on is decreasing and concave up, then its rate of decrease is slowing; it is leveling o. If the func on is increasing and concave up, then the rate
of increase is increasing. The func on is increasing at a faster and faster rate.
Now consider a func on which is concave down. We essen ally repeat the
above paragraphs with slight varia on.
The graph of a func on f is concave down when f is decreasing. That means
as one looks at a concave down graph from le to right, the slopes of the tangent
lines will be decreasing. Consider Figure 3.26, where a concave down graph is
shown along with some tangent lines. No ce how the tangent line on the le
is steep, upward, corresponding to a large value of f . On the right, the tangent
line is steep, downward, corresponding to a small value of f .
If a func on is increasing and concave down, then its rate of increase is slowing; it is leveling o. If the func on is decreasing and concave down, then the
rate of decrease is decreasing. The func on is decreasing at a faster and faster
Figure 3.26: A func on f with a concave

down graph. No ce how the slopes of the
tangent lines, when looking from le to
right, are increasing.
142
Concavity
Deni on 16
Note: We o en state that f is concave

up instead of the graph of f is concave
up for simplicity.
Our study of nice func ons con nues. The previous sec on showed how the
rst deriva ve of a func on, f , can relay important informa on about f. We
now apply the same technique to f itself, and learn what this tells us about f.
The key to studying f is to consider its deriva ve, namely f , which is the
second deriva ve of f. When f > 0, f is increasing. When f < 0, f is
decreasing. f has rela ve maxima and minima where f = 0 or is undened.
This sec on explores how knowing informa on about f gives informa on
about f.
Notes:

rate.
Our deni on of concave up and concave down is given in terms of when
the rst deriva ve is increasing or decreasing. We can apply the results of the
previous sec on and to nd intervals on which a graph is concave up or down.
That is, we recognize that f is increasing when f > 0, etc.
f > 0, increasing
f < 0, c. down
f < 0, decreasing
f < 0, c. down
.
Theorem 31
Test for Concavity
f < 0, decreasing
f > 0, c. up
Let f be twice dieren able on an interval I. The graph of f is concave up

if f > 0 on I, and is concave down if f < 0 on I.
If knowing where a graph is concave up/down is important, it makes sense

that the places where the graph changes from one to the other is also important.
This leads us to a deni on.
Deni on 17
f > 0, increasing
f > 0, c. up
Figure 3.27: Demonstra ng the 4 ways

that concavity interacts with increasing/decreasing, along with the rela onships with the rst and second derivaves.
Point of Inec on
A point of inec on is a point on the graph of f at which the concavity

of f changes.
Figure 3.28 shows a graph of a func on with inec on points labeled.

If the concavity of f changes at a point (c, f(c)), then f is changing from
increasing to decreasing (or, decreasing to increasing) at x = c. That means that
the sign of f is changing from posi ve to nega ve (or, nega ve to posi ve) at
x = c. This leads to the following theorem.
Note: Geometrically speaking, a func on

is concave up if its graph lies above its tangent lines. A func on is concave down if
its graph lies below its tangent lines.
Theorem 32
Points of Inec on
15
If (c, f(c)) is a point of inec on on the graph of f, then either f = 0 or

f is not dened at c.
10
f > 0
f > 0
c. up
c. up
We have iden ed the concepts of concavity and points of inec on. It is

now me to prac ce using these concepts; given a func on, we should be able
to nd its points of inec on and iden fy intervals on which it is concave up or
down. We do so in the following examples.
. Example 87
.Finding intervals of concave up/down, inec on points
Let f(x) = x3 3x + 1. Find the inec on points of f and the intervals on which
f < 0
c. down
x
1
Figure 3.28: A graph of a func on with

its inec on points marked. The intervals where concave up/down are also indicated.
Notes:
143
f < 0 c. down
it is concave up/down.
f > 0 c. up
.
0
Figure 3.29: A number line determining

the concavity of f in Example 87.
x
2
We start by nding f (x) = 3x2 3 and f (x) = 6x. To nd

the inec on points, we use Theorem 32 and nd where f (x) = 0 or where
f is undened. We nd f is always dened, and is 0 only when x = 0. So the
point (0, 1) is the only possible point of inec on.
This possible inec on point divides the real line into two intervals, (, 0)
and (0, ). We use a process similar to the one used in the previous sec on to
determine increasing/decreasing. Pick any c < 0; f (c) < 0 so f is concave
down on (, 0). Pick any c > 0; f (c) > 0 so f is concave up on (0, ). Since
the concavity changes at x = 0, the point (0, 1) is an inec on point.
The number line in Figure 3.29 illustrates the process of determining concavity; Figure 3.30 shows a graph of f and f , conrming our results. No ce how f
is concave down precisely when f (x) < 0 and concave up when f (x) > 0.
.
S
.
Figure 3.30: A graph of f(x) used in Example 87.
. Example 88
.Finding intervals of concave up/down, inec on points
Let f(x) = x/(x2 1). Find the inec on points of f and the intervals on which
it is concave up/down.
We need to nd f and f . Using the Quo ent Rule and sim-
plifying, we nd
f (x) =
(1 + x2 )
(x2 1)2
and
f (x) =
2x(x2 + 3)
.
(x2 1)3
To nd the possible points of inec on, we seek to nd where f (x) = 0 and

where f is not dened. Solving f (x) = 0 reduces to solving 2x(x2 +3) = 0; we
nd x = 0. We nd the f is not dened when x = 1, for then the denominator
of f is 0.
The possible points of inec on x = 1, x = 0 and x = 1 split the number
line into four intervals, as shown in Figure 3.31. We determine the concavity
on each. Keep in mind that all we are concerned with is the sign of f on the
interval.
Interval 1, (, 1): Select a number c in this interval with a large magnitude
(for instance, c = 100). The denominator will be posi ve. In the numerator,
the (c2 +3) will be posi ve and the 2c term will be nega ve. Thus the numerator
is nega ve and f (c) is nega ve. We conclude f is concave down on (, 1).
Interval 2, (1, 0): For any number c in this interval, ) < c2 < 1, so the denominator is nega ve. Thus the denominator will be nega ve; in the numerator, the
2c term will be nega ve whereas the (c2 + 3) is always posi ve. Thus f (c) > 0
and f is concave up on this interval.
Notes:
144

Interval 3, (0, 1): Any number c in this interval will be posi ve and small. Thus
the numerator is posi ve while the denominator is nega ve. Thus f (c) < 0
and f is concave down on this interval.
Interval 4, (1, ): Choose a large value for c. It is evident that f (c) > 0, so we
conclude that f is concave up on (1, ).
f < 0 c. down
f > 0 c. up
f < 0 c. down
y
10
5
f > 0 c. up
x
2
.
1
We conclude that f is concave up on (1, 0) (1) and concave down on

(, 1)(0, 1). There is only one point of inec on, (0, 0), as f is not dened
at x = 1. Our work is conrmed by the graph of f in Figure 3.32. Again, f is
drawn in blue and f is drawn in red. No ce how f is concave up whenever f
is posi ve, and concave down when f is nega ve. .
10
Figure 3.32: A graph of f(x) and f (x) in

Example 88.
Recall that rela ve maxima and minima of f are found at cri cal points of
f; that is, they are found when f (x) = 0 or when f is undened. Likewise,
the rela ve maxima and minima of f are found when f (x) = 0 or when f is
undened; note that these are the inec on points of f.
What does a rela ve maximum of f mean? The deriva ve measures the
rate of change of f; maximizing f means nding the where f is increasing the
most where f has the steepest tangent line. A similar statement can be made
for minimizing f ; it corresponds to where f has the steepest nega velysloped
tangent line.
We u lize this concept in the next example.
. Example 89
.Understanding inec on points
The sales of a certain product over a three-year span are modeled by S(t) =
t4 8t2 + 20, where t is the me in years, shown in Figure 3.33. Over the rst
two years, sales are decreasing. Find the point at which sales are decreasing at
their greatest rate.
We want to maximize the rate of decrease, which is to say,
we want to nd where S has a minimum. To do this, we nd where S is 0. We
nd S (t)= 4t3 16t and S (t) = 12t2 16. Se ng S (t) = 0 and solving, we
get t = 4/3 1.16 (we ignore the nega ve value of t since it does not lie in
the domain of our func on S).
This is both the inec on point and the point of maximum decrease. This
is the point at which things rst start looking up for the company. A er the
inec on point, it will s ll take some me before sales start to increase, but at
least sales are not decreasing quite as quickly as they had been.
S
S(t)
20
15
10
5
t
1
Figure 3.33: A graph of S(t) in Example 89,

modeling the sale of a product over me.
Notes:
145

A graph of S(t) and S (t) is given in Figure 3.34. When S (t) < 0, sales are
decreasing; note how at t 1.16, S (t) is minimized. That is, sales are decreasing at the fastest rate at t 1.16. On the interval of (1.16, 2), S is decreasing
but concave up, so the decline in sales is leveling o. .
S(t)
20
10
t
1
10
.
Figure 3.34: A graph of S(t) in Example 89
along with S (t) in red.
Not every cri cal point corresponds to a rela ve extrema; f(x) = x3 has a
cri cal point at (0, 0) but no rela ve maximum or minimum. Likewise, just because f (x) = 0 we cannot conclude concavity changes at that point. We were
careful before to use terminology possible point of inec on since we needed
to check to see if the concavity changed. The canonical example of f (x) = 0
without concavity changing is f(x) = x4 . At x = 0, f (x) = 0 but f is always
concave up, as shown in Figure 3.35.
The Second Deriva ve Test

y
The rst deriva ve of a func on gave us a test to nd if a cri cal value corresponded to a rela ve maximum, minimum, or neither. The second deriva ve
gives us another way to test if a cri cal point is a local maximum or minimum.
The following theorem ocially states something that is intui ve: if a cri cal
value occurs in a region where a func on f is concave up, then that cri cal value
must correspond to a rela ve minimum of f, etc. See Figure 3.36 for a visualizaon of this.
0.5
.1
0.5
0.5
x
1
Figure 3.35: A graph of f(x) = x4 . Clearly f

is always concave up, despite the fact that
f (x) = 0 when x = 0. It this example, the possible point of inec on (0, 0)
is not a point of inec on.
y
10
c. down
rel. max
5
x
1
5
c. up
10
Figure 3.36: Demonstra ng the fact that

rela ve maxima occur when the graph is
concave down and rela ve minima occur
when the graph is concave up.
146
The Second Deriva ve Test
Let c be a cri cal value of f where f (c) is dened.
1. If f (c) > 0, then f has a local minimum at (c, f(c)).

2. If f (c) < 0, then f has a local maximum at (c, f(c)).
The Second Deriva ve Test relates to the First Deriva ve Test in the following
way. If f (c) > 0, then the graph is concave up at a cri cal point c and f itself
is growing. Since f (c) = 0 and f is growing at c, then it must go from nega ve
to posi ve at c. This means the func on goes from decreasing to increasing, indica ng a local minimum at c.
rel. min
Theorem 33
. Example 90
.Using the Second Deriva ve Test
Let f(x) = 100/x + x. Find the cri cal points of f and use the Second Deriva ve
Test to label them as rela ve maxima or minima.
Notes:

We nd f (x) = 100/x2 + 1 and f (x) = 100/x3 . We set
f (x) = 0 and solve for x to nd the cri cal values (note that f is not dened at
x = 0, but neither is f so this is not a cri cal value.) We nd the cri cal values
are x = 10. Evalua ng f at x = 10 gives 0.1 > 0, so there is a local minimum
at x = 10. Evalua ng f (10) = 0.1 < 0, determining a rela ve maximum
at x = 10. These results are conrmed in Figure 3.37. .
We have been learning how the rst and second deriva ves of a func on
relate informa on about the graph of that func on. We have found intervals of
increasing and decreasing, intervals where the graph is concave up and down,
along with the loca ons of rela ve extrema and inec on points. In Chapter 1
we saw how limits explained asympto c behavior. In the next sec on we combine all of this informa on to produce accurate sketches of func ons.
40
f (10) > 0
20
x
20
10
10
20
20
f (10) < 0
40

90. The second deriva ve is evaluated
at each cri cal point. When the graph is
concave up, the cri cal point represents
a local minimum; when the graph is concave down, the cri cal point represents a
local maximum.
Notes:
147
Exercises 3.4
Terms and Concepts
1. Sketch a graph of a func on f(x) that is concave up on (0, 1)
and is concave down on (1, 2).
2. Sketch a graph of a func on f(x) that is:
(a) Increasing, concave up on (0, 1),
(b) increasing, concave down on (1, 2),
(c) decreasing, concave down on (2, 3) and
(d) increasing, concave down on (3, 4).
3. Is is possible for a func on to be increasing and concave
down on (0, ) with a horizontal asymptote of y = 1? If
so, give a sketch of such a func on.
4. Is is possible for a func on to be increasing and concave up
on (0, ) with a horizontal asymptote of y = 1? If so, give
a sketch of such a func on.
1
x2 + 1
x
24. f(x) = 2
x 1
25. f(x) = sin x + cos x on (, )
23. f(x) =
26. f(x) = x2 ex
27. f(x) = x2 ln x
28. f(x) = ex
In Exercises 29 41, a func on f(x) is given. Find the cri cal

points of f and use the Second Deriva ve Test, when possible, to determine the rela ve extrema. (Note: these are the
same func ons as in Exercises 16 28.)
29. f(x) = x2 2x + 1
30. f(x) = x2 5x + 7
31. f(x) = x3 x + 1
32. f(x) = 2x3 3x2 + 9x + 5
Problems
(a) Compute f (x).
x4
x3
+
2x + 3
4
3
34. f(x) = 3x4 + 8x3 + 6x2 24x + 2
33. f(x) =
7. f(x) = 4x2 + 3x 8
35. f(x) = x4 4x3 + 6x2 4x + 1

1
36. f(x) = 2
x +1
x
37. f(x) = 2
x 1
38. f(x) = sin x + cos x on (, )
8. f(x) = x3 3x2 + x 1
39. f(x) = x2 ex
(b) Graph f and f on the same axes (using technology is

permi ed) and verify Theorem 31.
5. f(x) = 7x + 3
6. f(x) = 4x2 + 3x 8
9. f(x) = x + x 2x + 5
10. f(x) = cos x
11. f(x) = sin x
12. f(x) = tan x
1
13. f(x) = 2
x +1
1
14. f(x) =
x
1
15. f(x) = 2
x
(a) Find the possible points of inec on of f.
(b) Create a number line to determine the intervals on
which f is concave up or concave down.
16. f(x) = x2 2x + 1
17. f(x) = x2 5x + 7
18. f(x) = x3 x + 1
19. f(x) = 2x3 3x2 + 9x + 5
x4
x3
+
2x + 3
4
3
21. f(x) = 3x4 + 8x3 + 6x2 24x + 2
20. f(x) =
22. f(x) = x4 4x3 + 6x2 4x + 1
40. f(x) = x2 ln x
41. f(x) = ex
In Exercises 42 54, a func on f(x) is given. Find the x values where f (x) has a rela ve maximum or minimum. (Note:
these are the same func ons as in Exercises 16 28.)
42. f(x) = x2 2x + 1
43. f(x) = x2 5x + 7
44. f(x) = x3 x + 1
45. f(x) = 2x3 3x2 + 9x + 5
x4
x3
+
2x + 3
4
3
47. f(x) = 3x4 + 8x3 + 6x2 24x + 2
46. f(x) =
48. f(x) = x4 4x3 + 6x2 4x + 1

1
49. f(x) = 2
x +1
x
50. f(x) = 2
x 1
51. f(x) = sin x + cos x on (, )
52. f(x) = x2 ex
53. f(x) = x2 ln x
54. f(x) = ex
3.5 Curve Sketching
3.5
Curve Sketching
We have been learning how we can understand the behavior of a func on based
on its rst and second deriva ves. While we have been trea ng the proper es
of a func on separately (increasing and decreasing, concave up and concave
down, etc.), we combine them here to produce an accurate graph of the func on
without plo ng lots of extraneous points.
Why bother? Graphing u li es are very accessible, whether on a computer,
a handheld calculator, or a smartphone. These resources are usually very fast
and accurate. We will see that our method is not par cularly fast it will require
me (but it is not hard). So again: why bother?
We a emp ng to understand the behavior of a func on f based on the informa on given by its deriva ves. While all of a func ons deriva ves relay informa on about it, it turns out that most of the behavior we care about is
explained by f and f . Understanding the interac ons between the graph of f
and f and f is important. To gain this understanding, one might argue that all
that is needed is to look at lots of graphs. This is true to a point, but is somewhat similar to sta ng that one understands how an engine works a er looking
only at pictures. It is true that the basic ideas will be conveyed, but handson
access increases understanding.
The following Key Idea summarizes what we have learned so far that is applicable to sketching func on graphs and gives a framework for pu ng that informa on together. It is followed by several examples.
Key Idea 4
Curve Sketching
To produce an accurate sketch a given func on f, consider the following

steps.
1. Find the domain of f. Generally, we assume that the domain is the
en re real line then nd restric ons, such as where a denominator
is 0 or where nega ves appear under the radical.
2. Find the cri cal values of f.
3. Find the possible points of inec on of f.

4. Find the loca on of any ver cal asymptotes of f (usually done in
conjunc on with item 1 above).
5. Consider the limits lim f(x) and lim f(x) to determine the end
x
behavior of the func on.

(con nued)
Notes:
149
Key Idea 4
Curve Sketching Con nued
6. Create a number line that includes all cri cal points, possible
points of inec on, and loca ons of ver cal asymptotes. For each
interval created, determine whether f is increasing or decreasing,
concave up or down.
7. Evaluate f at each cri cal point and possible point of inec on.
Plot these points on a set of axes. Connect these points with curves
exhibi ng the proper concavity. Sketch asymptotes and x and y
intercepts were applicable.
. Example 91
.Curve sketching
Use Key Idea 4 to sketch f(x) = 3x3 10x2 + 7x + 5.
S
We follow the steps outlined in the Key Idea.
1. The domain of f is the en re real line; there are no values x for which f(x)
is not dened.
2. Find the cri cal values of f. We compute f (x) = 9x2 20x + 7. Use the
Quadra c Formula to nd the roots of f :
)
20 (20)2 4(9)(7)
1(
x=
=
10 37 x 0.435, 1.787.
2(9)
9
3. Find the possible points of inec on of f. Compute f (x) = 18x 20. We
have
f (x) = 0 x = 10/9 1.111.
4. There are no ver cal asymptotes.
5. We determine the end behavior using limits as x approaches innity.
lim f(x) =
lim f(x) = .
We do not have any horizontal asymptotes.
6. We place the values x = (10 37)/9 and x = 10/9 on a number

line, as shown in Figure 3.38. We mark each subinterval as increasing or
Notes:
150
3.5 Curve Sketching

decreasing, concave up or down, using the techniques used in Sec ons
3.3 and 3.4.
f > 0 incr
f < 0 decr
f < 0 decr
f > 0 incr
f < 0 c. down
f < 0 c. down
f > 0 c. up
f < 0 c. up
1
9 (10
10
9
37)
0.435
1.111
1
9 (10 +
y
10
37)
1.787
x
7. We plot the appropriate points on axes as shown in Figure 3.39(a) and

connect the points with straight lines. In Figure 3.39(b) we adjust these
lines to demonstrate the proper concavity. Our curve crosses the y axis at
y = 5 and crosses the x axis near x = 0.424. In Figure 3.39(c) we show a
graph of f drawn with a computer program, verifying the accuracy of our
sketch.
.
(a)
y
10
. Example 92
.Curve sketching
x2 x 2
.
Sketch f(x) = 2
x x6
x
1
We again follow the steps outlined in Key Idea 4.
1. In determining the domain, we assume it is all real numbers and looks for
restric ons. We nd that at x = 2 and x = 3, f(x) is not dened. So the
domain of f is D = {real numbers x | x = 2, 3}.
(b)
y
2. To nd the cri cal values of f, we rst nd f (x). Using the Quo ent Rule,
we nd
8x + 4
8x + 4
f (x) = 2
=
.
(x + x 6)2
(x 3)2 (x + 2)2
f (x) = 0 when x = 1/2, and f is undened when x = 2, 3. Since f
is undened only when f is, these are not cri cal values. The only cri cal
value is x = 1/2.
3. To nd the possible points of inec on, we nd f (x), again employing
the Quo ent Rule:
f (x) =
24x2 24x + 56
.
(x 3)3 (x + 2)3
10
x
1
1
5
(c)
Figure 3.39: Sketching f in Example 91.
We nd that f (x) is never 0 (se ng the numerator equal to 0 and solving

for x, we nd the only roots to this quadra c are imaginary) and f is
Notes:
151

undened when x = 2, 3. Thus concavity will possibly only change at
x = 2 and x = 3.
y
5
4. The ver cal asymptotes of f are at x = 2 and x = 3, the places where f

is undened.
x
4
1.
(a)
f > 0 incr
y
5
f > 0 incr
> 0 c. up
< 0 c. down
2
x
4
f < 0 decr
f < 0 decr
f > 0 c. up
< 0 c. down
3
1
2
7. In Figure 3.41(a), we plot the points from the number line on a set of
axes and connect the points with straight lines to get a general idea of
what the func on looks like (these lines eec vely only convey increasing/decreasing informa on). In Figure 3.41(b), we adjust the graph with
the appropriate concavity. We also show f crossing the x axis at x = 1
and x = 2.
..
6. We place the values x = 1/2, x = 2 and x = 3 on a number line as

shown in Figure 3.40. We mark in each interval whether f is increasing or
decreasing, concave up or down. We see that f has a rela ve maximum at
x = 1/2; concavity changes only at the ver cal asymptotes.
5. There is a horizontal asymptote of y = 1, as lim f(x) = 1 and lim f(x) =
(b)
y
Figure 3.41(c) shows a computer generated graph of f, which veries the accuracy. of our sketch.
. Example 93
x
4
.Curve sketching
5(x 2)(x + 1)
Sketch f(x) =
.
x2 + 2x + 4
S
We again follow Key Idea 4.
(c)
1. We assume that the domain of f is all real numbers and consider restricons. The only restric ons come when the denominator is 0, but this
never occurs. Therefore the domain of f is all real numbers, R.
2. We nd the cri cal values of f by se ng f (x) = 0 and solving for x. We

nd
.
.
f (x) =
Notes:
152
15x(x + 4)
(x2 + 2x + 4)2
f (x) = 0 when x = 4, 0.
3.5 Curve Sketching

3. We nd the possible points of inec on by solving f (x) = 0 for x. We
nd
30x3 + 180x2 240
f (x) =
.
(x2 + 2x + 4)3
The cubic in the numerator does not factor very nicely. We instead approximate the roots at x = 5.759, x = 1.305 and x = 1.064.
4. There are no ver cal asymptotes.
x
5
5. We have a horizontal asymptote of y = 5, as lim f(x) = lim f(x) = 5.

x
6. We place the cri cal points and possible points on a number line as shown
in Figure 3.42 and mark each interval as increasing/decreasing, concave
up/down appropriately.
f > 0 incr
f > 0 incr
f < 0 decr
f < 0 decr
f > 0 incr
f > 0 decr
f > 0 c. up
f < 0 c. down
f < 0 c. down
f > 0 c. up
f > 0 c. up
f < 0 c. down
5.579
1.305
(a)
y
1.064
7. In Figure 3.43(a) we plot the signicant points from the number line as
well as the two roots of f, x = 1 and x = 2, and connect the points
with straight lines to get a general impression about the graph. In Figure
3.43(b), we add concavity. Figure 3.43(c) shows a computer generated
graph of f, arming our results.
.
In each of our examples, we found signicant points on the graph of f that
corresponding to changes in increasing/decreasing or concavity. We connected
these points with straight lines, then adjusted for concavity, and nished by
showing a very accurate, computer generated graph.
Why are computer graphics so good? It is not because computers are smarter
than we are. Rather, it is largely because computers are much faster at computing than we are. In general, computers graph func ons much like most students
do when rst learning to draw graphs: they plot equally spaced points, then connect the dots using lines. By using lots of points, the connec ng lines are short
and the graph looks smooth.
This does a ne job of graphing in most cases (in fact, this is the method
used for many graphs in this text). However, in regions where the graph is very
curvy, this can generate no ceable sharp edges on the graph unless a large
number of points are used. High quality computer algebra systems, such as
x
5
.
(b)
y
x
5
.
(c)
Notes:
153

Mathema ca, use special algorithms to plot lots of points only where the graph
is curvy.
In Figure 3.44, a graph of y = sin x is given, generated by Mathema ca.
The small points represent each of the places Mathema ca sampled the funcon. No ce how at the bends of sin x, lots of points are used; where sin x
is rela vely straight, fewer points are used. (Many points are also used at the
endpoints to ensure the end behavior is accurate.)
1.0
0.5
-0.5
-1.0
Figure 3.44: A graph of y = sin x generated by Mathema ca.
How does Mathema ca know where the graph is curvy? Calculus. When
we study curvature in a later chapter, we will see how the rst and second
deriva ves of a func on work together to provide a measurement of curviness. Mathema ca employs algorithms to determine regions of high curvature and plots extra points there.
Notes:
154
Exercises 3.5
Terms and Concepts
1. Why is sketching curves by hand benecial even though
technology is ubiquitous?
2. What does ubiquitous mean?
14. f(x) = x3 + 3x2 + 3x + 1

15. f(x) = x3 x2 x + 1
16. f(x) = (x 2) ln(x 2)
17. f(x) = (x 2)2 ln(x 2)
3. T/F: When sketching graphs of func ons, it is useful to nd

the cri cal points.
18. f(x) =
x2 4
x2

the possible points of inec on.
19. f(x) =
x2 4x + 3
x2 6x + 8

the horizontal and ver cal asymptotes.
20. f(x) =
Problems
x2 2x + 1
x2 6x + 8
21. f(x) = x x + 1
22. f(x) = x2 ex
In Exercises 6 11, prac ce using Key Idea 4 by applying the

principles to the given func ons with familiar graphs.
23. f(x) = sin x cos x on [, ]
6. f(x) = 2x + 4
24. f(x) = (x 3)2/3 + 2
7. f(x) = x2 + 1
25. f(x) =
8. f(x) = sin x
9. f(x) = ex
1
x
1
11. f(x) = 2
x
In Exercises 12 25, sketch a graph of the given func on using
Key Idea 4. Show all work; check your answer with technology.
10. f(x) =
12. f(x) = x3 2x2 + 4x + 1

13. f(x) = x3 + 5x2 3x + 2
(x 1)2/3
x
In Exercises 26 28, a func on with the parameters a and b

are given. Describe the cri cal points and possible points of
inec on of f in terms of a and b.
26. f(x) =
a
x 2 + b2
27. f(x) = sin(ax + b)

28. f(x) = (x a)(x b)
29. Given x2 + y2 = 1, use implicit dieren a on to nd
2
dy
dx
and ddx2y . Use this informa on to jus fy the sketch of the

unit circle.
4: A
D
In Chapter 3, we learned how the rst and second deriva ves of a func on inuence its graph. In this chapter we explore other applica ons of the deriva ve.
y
1
4.1
Newtons Method
Solving equa ons is one of the most important things we do in mathema cs,
yet we are surprisingly limited in what we can solve analy cally. For instance,
equa ons as simple as x5 + x + 1 = 0 or cos x = x cannot be solved by algebraic
methods in terms of familiar func ons. Fortunately, there are methods that
can give us approximate solu ons to equa ons like these. These methods can
usually give an approxima on correct to as many decimal places as we like. In
Sec on 1.5 we learned about the Bisec on Method. This sec on focuses on
another technique (which generally works faster), called Newtons Method.
Newtons Method is built around tangent lines. The main idea is that if x is
suciently close to a root of f(x), then the tangent line to the graph at (x, f(x))
will cross the x-axis at a point closer to the root than x.
We start Newtons Method with an ini al guess about roughly where the
root is. Call this x0 . (See Figure 4.1(a).) Draw the tangent line to the graph at
(x0 , f(x0 )) and see where it meets the x-axis. Call this point x1 . Then repeat the
process draw the tangent line to the graph at (x1 , f(x1 )) and see where it meets
the x-axis. (See Figure 4.1(b).) Call this point x2 . Repeat the process again to get
x3 , x4 , etc. This sequence of points will o en converge rather quickly to a root
of f.
We can use this geometric process to create an algebraic process. Lets look
at how we found x1 . We started with the tangent line to the graph at (x0 , f(x0 )).
The slope of this tangent line is f (x0 ) and the equa on of the line is
0.5
x
x0
0.5
1
.
(a)
y
1
0.5
x
x0
0 = f (x0 )(x1 x0 ) + f(x0 ).
x2
x1
0.5
1
.
(b)
y
1
0.5
x
y = f (x0 )(x x0 ) + f(x0 ).

This line crosses the x-axis when y = 0, and the xvalue where it crosses is what
we called x1 . So let y = 0 and replace x with x1 , giving the equa on:
x1
x0
x2
x3
x1
0.5
1
.
(c)
Figure 4.1: Demonstra ng the geometric
concept behind Newtons Method.
Chapter 4 Applica ons of the Deriva ve

Now solve for x1 :
x1 = x0
f(x0 )
.
f (x0 )
Since we repeat the same geometric process to nd x2 from x1 , we have

x2 = x1
f(x1 )
.
f (x1 )
In general, given an approxima on xn , we can nd the next approxima on, xn+1

as follows:
xn+1 = xn
f(xn )
.
f (xn )
We summarize this process as follows.

.
Key Idea 5
Newtons Method
Let f be a dieren able func on on an interval I with a root in I. To approximate the value of the root, accurate to d decimal places:
1. Choose a value x0 as an ini al approxima on of the root. (This is
o en done by looking at a graph of f.)
Note: Newtons Method is not infallible. The sequence of approximate values

may not converge, or it may converge so
slowly that one is tricked into thinking a
certain approxima on is be er than it actually is. These issues will be discussed at
the end of the sec on.
2. Create successive approxima ons itera vely; given an approximaon xn , compute the next approxima on xn+1 as
xn+1 = xn
f(xn )
.
f (xn )
3. Stop the itera ons when successive approxima ons do not dier
in the rst d places a er the decimal point.
Lets prac ce Newtons Method with a concrete example.

. Example 94
.Using Newtons Method
Approximate the real root of x3 x2 1 = 0, accurate to the rst 3 places a er
the decimal, using Newtons Method and an ini al approxima on of x0 = 1.
S
Notes:
158
To begin, we compute f (x) = 3x2 2x. Then we apply the
4.1 Newtons Method

Newtons Method algorithm, outlined in Key Idea 5.
f(1)
13 12 1
=
1
= 2,
f (1)
3 12 2 1
f(2)
23 22 1
=2
=2
= 1.625,
f (2)
3 22 2 2
f(1.625)
1.6253 1.6252 1
= 1.625
= 1.625
1.48579.
f (1.625)
3 1.6252 2 1.625
f(1.48579)
= 1.48579
1.46596
f (1.48579)
f(1.46596)
1.46557
= 1.46596
f (1.46596)
x1 = 1
x2
x3
x4
x5
We performed 5 itera ons of Newtons Method to nd a root accurate to the

rst 3 places a er the decimal; our nal approxima on is 1.465. The exact value
of the root, to six decimal places, is 1.465571; It turns out that our x5 is accurate
to more than just 3 decimal places.
A graph of f(x) is given in Figure 4.2. We can see from the graph that our
ini al approxima on of x0 = 1 was not par cularly accurate; a closer guess
would have been x0 = 1.5. Our choice was based on ease of ini al calcula on,
and shows that Newtons Method can be robust enough that we do not have to
make a very accurate ini al approxima on. .
y
0.5
x
0.5
1.5
0.5
1
1.5.
Figure 4.2: A graph of f(x) = x3 x2 1

in Example 94.
We can automate this process on a calculator that has an Ans key that returns the result of the previous calcula on. Start by pressing 1 and then Enter.
(We have just entered our ini al guess, x0 = 1.) Now compute
Ans
f(Ans)
f (Ans)
by entering the following and repeatedly press the Enter key:

Ans-(Ans^3-Ans^2-1)/(3*Ans^2-2*Ans)
Each me we press the Enter key, we are nding the successive approxima ons,
x1 , x2 , , and each one is ge ng closer to the root. In fact, once we get past
around x7 or so, the approxima ons dont appear to be changing. They actually
are changing, but the change is far enough to the right of the decimal point that
it doesnt show up on the calculators display. When this happens, we can be
pre y condent that we have found an accurate approxima on.
Using a calculator in this manner makes the calcula ons simple; many itera ons can be computed very quickly.
Notes:
159

. Example 95
Using Newtons Method to nd where func ons intersect
Use Newtons Method to approximate a solu on to cos x = x, accurate to 5
places a er the decimal.
y
1
0.5
x
1
0.5
0.5
0.5
Newtons Method provides a method of solving f(x) = 0; it

S
is not (directly) a method for solving equa ons like f(x) = g(x). However, this is
not a problem; we can rewrite the la er equa on as f(x) g(x) = 0 and then
use Newtons Method.
So we rewrite cos x = x as cos x x = 0. Wri en this way, we are nding
a root of f(x) = cos x x. We compute f (x) = sin x 1. Next we need a
star ng value, x0 . Consider Figure 4.3, where f(x) = cos x x is graphed. It
seems that x0 = 0.75 is pre y close to the root, so we will use that as our x0 .
(The gure also shows the graphs of y = cos x and y = x, drawn in red. Note
how they intersect at the same x value as when f(x) = 0.)
We now compute x1 , x2 , etc. The formula for x1 is
Figure 4.3: A graph of f(x) = cos x x

used to nd an ini al approxima on of its
root.
x1 = 0.75
cos(0.75) 0.75
0.7391111388.
sin(0.75) 1
Apply Newtons Method again to nd x2 :

x2 = 0.7391111388
cos(0.7391111388) 0.7391111388
0.7390851334.
sin(0.7391111388) 1
We can con nue this way, but it is really best to automate this process. On a calculator with an Ans key, we would start by pressing 0.75, then Enter, inpu ng
our ini al approxima on. We then enter:
Ans - (cos(Ans)-Ans)/(-sin(Ans)-1).
Repeatedly pressing the Enter key gives successive approxima ons. We
quickly nd:
x3 = 0.7390851332
x4 = 0.7390851332.
Our approxima ons x2 and x3 did not dier for at least the rst 5 places a er the
decimal, so we could have stopped. However, using our calculator in the manner described is easy, so nding x4 was not hard. It is interes ng to see how we
found an approxima on, accurate to as many decimal places as our calculator
displays, in just 4 itera ons. .
If you know how to program, you can translate the following pseudocode
into your favorite language to perform the computa on in this problem.
Notes:
160
4.1 Newtons Method

x = .75
while true
oldx = x
x = x - (cos(x)-x)/(-sin(x)-1)
print x
if abs(x-oldx) < .0000000001
break
This code calculates x1 , x2 , etc., storing each result in the variable x. The previous approxima on is stored in the variable oldx. We con nue looping un l
the dierence between two successive approxima ons, abs(x-oldx), is less
than some small tolerance, in this case, .0000000001.
Convergence of Newtons Method
y
0.5
x
0.5
1.5
1.5
Figure 4.4: A graph of f(x) = x3 x2 1,

showing why an ini al approxima on of
x0 = 0 with Newtons Method fails.
y
x
1
x0
.
(a)
and f (0) = 0, we see that x1 is not well dened.

This problem can also occur if, for instance, it turns out that f (x5 ) = 0.
Adjus ng the ini al approxima on x0 will likely ameliorate the problem.
It is also possible for Newtons Method to not converge while each successive
approxima on is well dened. Consider f(x) = x1/3 , as shown in Figure 4.5. It
is clear that the root is x = 0, but lets approximate this with x0 = 0.1. Figure
4.5(a) shows graphically the calcula on of x1 ; no ce how it is farther from the
root than x0 . Figures 4.5(b) and (c) show the calcula on of x2 and x3 , which are
even farther away; our successive approxima ons are ge ng worse. (It turns
out that in this par cular example, each successive approxima on is twice as far
from the true answer as the previous approxima on.)
There is no x to this problem; Newtons Method simply will not work and
another method must be used.
While Newtons Method does not always work, it does work most of the
me, and it is generally very fast. Once the approxima ons get close to the root,
Newtons Method can as much as double the number of correct decimal places
0.5
What should one use for the ini al guess, x0 ? Generally, the closer to the
actual root the ini al guess is, the be er. However, some ini al guesses should
be avoided. For instance, consider Example 94 where we sought the root to
f(x) = x3 x2 1. Choosing x0 = 0 would have been a par cularly poor choice.
Consider Figure 4.4, where f(x) is graphed along with its tangent line at x = 0.
Since f (0) = 0, the tangent line is horizontal and does not intersect the xaxis.
Graphically, we see that Newtons Method fails.
We can also see analy cally that it fails. Since
f(0)
x1 = 0
f (0)
0.5
x
1
x1 x0
.
(b)
y
x
1x3
x1 x0 x2
Notes:
.
(c)
Figure 4.5: Newtons Method fails to nd
a root of f(x) = x1/3 , regardless of the
choice of x0 .
161

with each successive approxima on. A course in Numerical Analysis will introduce the reader to more itera ve root nding methods, as well as give greater
detail about the strengths and weaknesses of Newtons Method.
Notes:
162
Exercises 4.1
Terms and Concepts
1. T/F: Given a func on f(x), Newtons Method produces an
exact solu on to f(x) = 0.
2. T/F: In order to get a solu on to f(x) = 0 accurate to d
places a er the decimal, at least d + 1 itera ons of Newtons Method must be used.
Problems
that interval. Use technology to obtain good ini al approxima ons.

8. f(x) = x3 + 5x2 x 1
9. f(x) = x4 + 2x3 7x2 x + 5
10. f(x) = x17 2x13 10x8 + 10 on (2, 2)
11. f(x) = x2 cos x + (x 1) sin x on (3, 3)
In Exercises 12 15, use Newtons Method to approximate
when the given func ons are equal, accurate to 3 places after the decimal. Use technology to obtain good ini al approxima ons.
In Exercises 3 7, the roots of f(x) are known or are easily

found. Use 5 itera ons of Newtons Method with the given
ini al approxima on to approximate the root. Compare it to
the known value of the root.
12. f(x) = x2 , g(x) = cos x
3. f(x) = cos x, x0 = 1.5
13. f(x) = x2 1, g(x) = sin x
4. f(x) = sin x, x0 = 1
5. f(x) = x + x 2, x0 = 0
2
14. f(x) = ex , g(x) = cos x
6. f(x) = x2 2, x0 = 1.5
15. f(x) = x, g(x) = tan x on [6, 6]
7. f(x) = ln x, x0 = 2
16. Why does Newtons Method fail in nding a root of f(x) =

x3 3x2 + x + 3 when x0 = 1?
In Exercises 8 11, use Newtons Method to approximate all

roots of the given func ons accurate to 3 places a er the decimal. If an interval is given, nd only the roots that lie in
17. Why does Newtons Method fail in nding a root of f(x) =

17x4 + 130x3 301x2 + 156x + 156 when x0 = 1?
4.2 Related Rates

When two quan es are related by an equa on, knowing the value of one quanty can determine the value of the other. For instance, the circumference and
radius of a circle are related by C = 2r; knowing that C = 6in determines the
radius must be 3in.
The topic of related rates takes this one step further: knowing the rate
at which one quan ty is changing can determine the rate at which the other
changes.
We demonstrate the concepts of related rates through examples.
. Example 96
Understanding related rates
The radius of a circle is growing at a rate of 5in/hr. At what rate is the circumference growing?
The circumference and radius of a circle are related by C =
S
2r. We are given informa on about how the length of r changes with respect
to me; that is, we are told dr
dt = 5in/hr. We want to know how the length of C
changes with respect to me, i.e., we want to know dC
dt .
Implicitly dieren ate both sides of C = 2r with respect to t:
C = 2r
d( )
d( )
C =
2r
dt
dt
dC
dr
= 2 .
dt
dt
Note: This sec on relies heavily on implicit dieren a on, so referring back to
Sec on 2.6 may help.
As we know
dr
dt
= 5in/hr, we know
dC
= 25 = 10 31.4in/hr.
dt
Consider another, similar example.

. Example 97
.Finding related rates
Water streams out of a faucet at a rate of 2in3 /s onto a at surface at a constant
rate, forming a circular puddle that is 1/8in deep.
1. At what rate is the area of the puddle growing?
2. At what rate is the radius of the circle growing?
Notes:
164
4.2 Related Rates

S
1. We can answer this ques on two ways: using common sense or related
rates. The common sense method states that the volume of the puddle is
growing by 2in3 /s, where
volume of puddle = area of circle depth.
Since the depth is constant at 1/8in, the area must be growing by 16in2 /s.
This approach reveals the underlying relatedrates principle. Let V and A
represent the Volume and Area of the puddle. We know V = A 18 . Take
the deriva ve of both sides with respect to t, employing implicit dierena on.
1
A
8 ( )
d( )
d 1
V =
A
dt
dt 8
dV
1 dA
=
dt
8 dt
V=
Since dV
dt = 2, we know 2 =
by 16in2 /s.
1 dA
8 dt ,
and hence
dA
dt
= 16. The area is growing
.
2. To start, we need an equa on that relates what we know to the radius.
We just learned something about the surface area of the circular puddle,
and we know A = r2 . We should be able to learn about the rate at which
the radius is growing with this informa on.
Implicitly derive both sides of A = r2 with respect to t:
A = r2
d( )
d ( 2)
A =
r
dt
dt
dA
dr
= 2r
dt
dt
Our work above told us that
dA
dt
= 16in2 /s. Solving for
dr
dt ,
we have
dr
8
= .
dt
r
Note how our answer is not a number, but rather a func on of r. In other
words, the rate at which the radius is growing depends on how big the
Notes:
165

circle already is. If the circle is very large, adding 2in3 of water will not
make the circle much bigger at all. If the circle dimesized, adding the
same amount of water will make a radical change in the radius of the circle.
In some ways, our problem was (inten onally) illposed. We need to specify a current radius in order to know a rate of change. When the puddle
has a radius of 10in, the radius is growing at a rate of
dr
8
4
=
=
0.25in/s.
dt
10
5
.
N
A = 1/2
B = 1/2
E
Figure 4.6: A sketch of a police car (at bottom) a emp ng to measure the speed of
a car (at right) in Example 98.
. Example 98
.Studying related rates
Radar guns measure the rate of distance change between the gun and the object
it is measuring. For instance, a reading of 55mph means the object is moving
away from the gun at a rate of 55 miles per hour, whereas a measurement of
25mph would mean that the object is approaching the gun at a rate of 25
miles per hour.
If the radar gun is moving (say, a ached to a police car) then radar readouts
are only immediately understandable if the gun and the object are moving along
the same line. If a police ocer is traveling 60mph and gets a readout of 15mph,
he knows that the car ahead of him is moving away at a rate of 15 miles an hour,
meaning the car is traveling 75mph. (This straightline principle is one reason
ocers park on the side of the highway and try to shoot straight back down the
road. It gives the most accurate reading.)
Suppose an ocer is driving due north at 60 mph and sees a car moving
due east, as shown in Figure 4.6. Using his radar gun, he measures a reading of
80mph. By using landmarks, he believes both he and the other car are about
1/2 mile from the intersec on of their two roads.
If the speed limit on the other road is 55mph, is the other driver speeding?
Using the diagram in Figure 4.6, lets label what we know
S
about the situa on. As both the police ocer and other driver are 1/2 mile
from the intersec on, we have A = 1/2, B = 1/2, and through the Pythagorean
Theorem, C = 1/ 2 0.707.
We know the police ocer is traveling at 60mph; that is,
dB
measurement is dC
dt = 80. We want to nd dt .
dA
dt
= 60. The radar
We need an equa on that contains relates B to A and/or C. The Pythagorean

Theorem seems like a good choice: A2 + B2 = C2 . Dieren ate both sides with
Notes:
166
4.2 Related Rates

respect to t:
2
A +B =C
)
d ( 2
d ( 2)
A + B2 =
C
dt
dt
dA
dB
dC
2A
+ 2B
= 2C
dt
dt
dt
We have values for everything except
dB
dt .
Solving for this we have
C dC A dA
dB
dt
= dt
53.12mph.
dt
B
Note: Example 98 is both interes ng and

imprac cal. It highlights the diculty in
using radar in a nonlinear fashion, and
explains why in real life the police ocer would follow the other driver to determine their speed, and not pull out pencil and paper.
The principles here are important,
though. Many automated vehicles make
judgments about other moving objects
based on perceived distances and radar
like measurements using relatedrates
ideas.
The other driver does not appear to be speeding. .

. Example 99
.Studying related rates
A camera is placed on a tripod 10 from the side of a road. The camera is to turn
to track a car that is to drive by at 100mph for a promo onal video. The videos
planners want to know what kind of motor the tripod should be equipped with
in order to properly track the car as it passes by. Figure 4.7 shows the proposed
setup.
How fast must the camera be able to turn to track the car?
We seek informa on about how fast the camera is to turn;
S
therefore, we need an equa on that will relate an angle to the posi on of the
camera and the speed and posi on of the car.
Figure 4.7 suggests we use a trigonometric equa on. Le ng x represent the
distance the car is from the point on the road directly in front of the camera, we
have
x
tan =
.
(4.1)
10
100mph
10
Figure 4.7: Tracking a speeding car (at

le ) with a rota ng camera.
As the car is moving at 100mph, we have dx

dt = 100mph. We need to convert
the measurements to common units; rewrite 100mph in terms of /s:
dx
m
m
f
1 h
= 100 = 100 5280
= 146.6 /s.
dt
h
h
m 3600 s
Now take the deriva ve of both sides of Equa on (4.1) using implicit dieren -
Notes:
167

a on:
x
10
)
d(
d ( x )
tan =
dt
dt 10
1 dx
d
=
sec2
dt
10 dt
d
cos2 dx
=
dt
10 dt
tan =
(4.2)
We want to know the fastest the camera has to turn. Common sense tells us this
is when the car is directly in front of the camera (i.e., when = 0). Our mathema cs bears this out. In Equa on (4.2) we see this is when cos2 is largest; this
is when cos = 1, or when = 0.
With dx
dt 146.67 /s, we have
1rad
d
=
146.67 /s = 14.667radians/s.
dt
10
What does this number mean? Recall that 1 circular revolu on goes through 2
radians, thus 14.667rad/s means 14.667/(2) 2.33 revolu ons per second. .
We introduced the deriva ve as a func on that gives the slopes of tangent
lines of func ons. This chapter emphasizes using the deriva ve in other ways.
Newtons Method uses the deriva ve to approximate roots of func ons; this
sec on stresses the rate of change aspect of the deriva ve to nd a rela onship between the rates of change of two related quan es.
In the next sec on we use Extreme Value concepts to op mize quan es.
Notes:
168
Exercises 4.2
Terms and Concepts
1. T/F: Implicit dieren a on is o en used when solving related rates type problems.
2. T/F: A study of related rates is part of the standard police
ocer training.
8. An F-22 aircra is ying at 500mph with an eleva on of

100 on a straightline path that will take it directly over
an an aircra gun as in Exercise 7 (note the lower elevaon here).
How fast must the gun be able to turn to accurately track
the aircra when the plane is:
(a) 1000 feet away?
Problems
(b) 100 feet away?
3. Water ows onto a at surface at a rate of 5cm3 /s forming a

circular puddle 10mm deep. How fast is the radius growing
when the radius is:
(c) Directly overhead?

9. A 24 . ladder is leaning against a house while the base is
pulled away at a constant rate of 1 /s.
(a) 1 cm?
24
(b) 10 cm?
1 /s
At what rate is the top of the ladder sliding down the side
of the house when the base is:
(c) 100 cm?

4. A circular balloon is inated with air owing at a rate of
10cm3 /s. How fast is the radius of the balloon increasing
when the radius is:
(a) 1 foot from the house?

(b) 10 feet from the house?
(a) 1 cm?
(c) 23 feet from the house?
(b) 10 cm?
(d) 24 feet from the house?
(c) 100 cm?

5. Consider the trac situa on introduced in Example 98.
How fast is the other car traveling if the ocer and the
other car are each 1/2 mile from the intersec on, the ocer is traveling 50mph, and the radar reading is 70mph?
6. Consider the trac situa on introduced in Example 98.
How fast is the other car traveling if the ocer and the
other car are each 1 mile from the intersec on, the ocer
is traveling 60mph, and the radar reading is 80mph?
10. A boat is being pulled into a dock at a constant rate of

30 /min by a winch located 10 above the deck of the
boat.
10
At what rate is the boat approaching the dock when the

boat is:
(a) 50 feet out?
7. An F-22 aircra is ying at 500mph with an eleva on of

10,000 on a straightline path that will take it directly over
an an aircra gun.
(b) 15 feet out?

(c) 1 foot from the dock?
(d) What happens when the length of rope pulling in the
boat is less than 10 feet long?
10,000
How fast must the gun be able to turn to accurately track

the aircra when the plane is:
11. An inverted cylindrical cone, 20 deep and 10 across at

the top, is being lled with water at a rate of 10 3 /min. At
what rate is the water rising in the tank when the depth of
the water is:
(a) 1 foot?
(a) 1 mile away?
(b) 10 feet?
(b) 1/5 mile away?
(c) 19 feet?
(c) Directly overhead?
How long will the tank take to ll when star ng at empty?
30
12. A rope, a ached to a weight, goes up through a pulley at

the ceiling and back down to a worker. The man holds the
rope at the same height as the connec on point between
rope and weight.
(b) How fast is the weight rising a er the man has walked
10 feet?
(c) How fast is the weight rising a er the man has walked
40 feet?
(d) How far must the man walk to raise the weight all the
way to the pulley?
2 /s
Suppose the man stands directly next to the weight (i.e., a

total rope length of 60 ) and begins to walk away at a rate
of 2 /s. How fast is the weight rising when the man has
walked:
14. A hot air balloon li s o from ground rising ver cally. From
100 feet away, a 5 woman tracks the path of the balloon.
When her sightline with the balloon makes a 45 angle with
the horizontal, she notes the angle is increasing at about
5 /min.
(a) 10 feet?
(a) What is the eleva on of the balloon?
(b) 40 feet?
(b) How fast is it rising?
How far must the man walk to raise the weight all the way
to the pulley?
13. Consider the situa on described in Exercise 12. Suppose
the man starts 40 from the weight and begins to walk
away at a rate of 2 /s.
(a) How long is the rope?
15. A company that produces landscaping materials is dumping

sand into a conical pile. The sand is being poured at a rate
of 5 3 /sec; the physical proper es of the sand, in conjuncon with gravity, ensure that the cones height is roughly
2/3 the length of the diameter of the circular base.
How fast is the cone rising when it has a height of 30 feet?
4.3 Op miza on
4.3
Op miza on
In Sec on 3.1 we learned about extreme values the largest and smallest values
a func on a ains on an interval. We mo vated our interest in such values by
discussing how it made sense to want to know the highest/lowest values of a
stock, or the fastest/slowest an object was moving. In this sec on we apply
the concepts of extreme values to solve word problems, i.e., problems stated
in terms of situa ons that require us to create the appropriate mathema cal
framework in which to solve the problem.
We start with a classic example which is followed by a discussion of the topic
of op miza on.
.
.
.
.
.
.
One can likely guess the correct answer that is great. We

S
will proceed to show how calculus can provide this answer in a context that
proves this answer is correct.
It helps to make a sketch of the situa on. Our enclosure is sketched twice in
Figure 4.8, either with green grass and nice fence boards or as a simple rectangle.
Either way, drawing a rectangle forces us to realize that we need to know the
dimensions of this rectangle so we can create an area func on a er all, we are
trying to maximize the area.
We let x and y denote the lengths of the sides of the rectangle. Clearly,
. Example 100
.Op miza on: perimeter and area
A man has 100 feet of fencing, a large yard, and a small dog. He wants to create
a rectangular enclosure for his dog with the fencing that provides the maximal
area. What dimensions provide the maximal area?
.
x
Area = xy.
We do not yet know how to handle func ons with 2 variables; we need to
reduce this down to a single variable. We know more about the situa on: the
man has 100 feet of fencing. By knowing the perimeter of the rectangle must
be 100, we can create another equa on:
Perimeter = 100 = 2x + 2y.

We now have 2 equa ons and 2 unknowns. In the la er equa on, we solve
for y:
y = 50 x.
Figure 4.8: A sketch of the enclosure in

Example 100.
Now subs tute this expression for y in the area equa on:
Area = A(x) = x(50 x).
Note we now have an equa on of one variable; we can truly call the Area a
func on of x.
Notes:
171

This func on only makes sense when 0 x 50, otherwise we get nega ve
values of area. So we nd the extreme values of A(x) on the interval [0, 50].
To nd the cri cal points, we take the deriva ve of A(x) and set it equal to
0, then solve for x.
A(x) = x(50 x)
= 50x x2
A (x) = 50 2x
We solve 50 2x = 0 to nd x = 25; this is the only cri cal point. We evaluate
A(x) at the endpoints of our interval and at this cri cal point to nd the extreme
values; in this case, all we care about is the maximum.
Clearly A(0) = 0 and A(50) = 0, whereas A(25) = 625 2 . This is the maximum. Since we earlier found y = 50 x, we nd that y is also 25. Thus the
dimensions of the rectangular enclosure with perimeter of 100 . with maximum area is a square, with sides of length 25 . .
This example is very simplis c and a bit contrived. (A er all, most people
create a design then buy fencing to meet their needs, and not buy fencing and
plan later.) But it models well the necessary process: create equa ons that describe a situa on, reduce an equa on to a single variable, then nd the needed
extreme value.
In real life the problems are much more complex. The equa ons are often not reducible to a single variable (hence mul variable calculus is needed)
and the equa ons themselves may be dicult to form. Understanding the principles here will provide a good founda on for the mathema cs you will likely
encounter later.
We outline here the basic process of solving these op miza on problems.
.
Key Idea 6
Solving Op miza on Problems
1. Understand the problem. Clearly iden fy what quan ty is to be

maximized or minimized. Make a sketch if helpful.
2. Create equa ons relevant to the context of the problem, using the
informa on given. (One of these should describe the quan ty to
be op mized. Well call this the fundamental equa on.)
3. If the fundamental equa on denes the quan ty to be op mized

as a func on of more than one variable, reduce it to a single variable func on using subs tu ons derived from the other equaons.
(con nued). . .
Notes:
172
4.3 Op miza on
Key Idea 6
Solving Op miza on Problems Con nued
4. Iden fy the domain of this func on, keeping in mind the context
of the problem.
5. Find the extreme values of this func on on the determined domain.

6. Iden fy the values of all relevant quan
es of the problem.
We will use Key Idea 6 in a variety of examples.

. Example 101
.Op miza on: perimeter and area
Here is another classic calculus problem: A woman has a 100 feet of fencing, a
small dog, and a large yard that contains a stream (that is mostly straight). She
wants to create a rectangular enclosure with maximal area that uses the stream
as one side. (Apparently her dog wont swim away.) What dimensions provide
the maximal area?
We again appeal to the perimeter; here the perimeter is

Perimeter = 100 = x + 2y.
.
.
.
.
.
This is our fundamental equa on. This denes area as a func on of two
variables, so we need another equa on to reduce it to one variable.
Area = xy.
2. We want to maximize the area; as in the example before,
1. We are maximizing area. A sketch of the region will help; Figure 4.9 gives
two sketches of the proposed enclosed area. A key feature of the sketches
is to acknowledge that one side is not fenced.
We will follow the steps outlined by Key Idea 6.
.
.
x
x
Note how this is dierent than in our previous example.

3. We now reduce the fundamental equa on to a single variable. In the
perimeter equa on, solve for y: y = 50 1/2x. We can now write Area
as
Area = A(x) = x(50 1/2x) = 50x 1/2x2 .
Area is now dened as a func on of one variable.
Figure 4.9: A sketch of the enclosure in

Example 101.
Notes:
173

4. We want the area to be nonnega ve. Since A(x) = x(501/2x), we want
x 0 and 50 1/2x 0. The la er inequality implies that x 100, so
0 x 100.
5. We now nd the extreme values. At the endpoints, the minimum is found,
giving an area of 0.
Find the cri cal points. We have A (x) = 50 x; se ng this equal to 0
and solving for x returns x = 50. This gives an area of
A(50) = 50(25) = 1250.
6. We earlier set y = 50 1/2x; thus y = 25. Thus our rectangle will have
two sides of length 25 and one side of length 50, with a total area of 1250
2
.
Keep in mind as we do these problems that we are prac cing a process; that
is, we are learning to turn a situa on into a system of equa ons. These equaons allow us to write a certain quan ty as a func on of one variable, which we
then op mize.
1000
.
.
. Example 102
.Op miza on: minimizing cost
A power line needs to be run from an power sta on located on the beach to an
oshore facility. Figure 4.10 shows the distances between the power sta on to
the facility.
It costs $50/ . to run a power line along the land, and $130/ . to run a
power line under water. How much of the power line should be run along the
land to minimize the overall cost? What is the minimal cost?
5000
Figure 4.10: Running a power line from

the power sta on to an oshore facility
with minimal cost in Example 102.
00
x2
10
1000
.
.
5000 x
We will follow the strategy of Key Idea 6 implicitly, without

S
specically numbering steps.
There are two immediate solu ons that we could consider, each of which we
will reject through common sense. First, we could minimize the distance by
directly connec ng the two loca ons with a straight line. However, this requires
that all the wire be laid underwater, the most costly op on. Second, we could
minimize the underwater length by running a wire all 5000 . along the beach,
directly across from the oshore facility. This has the undesired eect of having
the longest distance of all, probably ensuring a nonminimal cost.
The op mal solu on likely has the line being run along the ground for a
while, then underwater, as the gure implies. We need to label our unknown
distances the distance run along the ground and the distance run underwater.
Recognizing that the underwater distance can be measured as the hypotenuse
of a right triangle, we choose to label the distances as shown in Figure 4.11.
Figure 4.11: Labeling unknown distances

in Example 102.
Notes:
174
4.3 Op miza on
By choosing x as we did, we make the expression under the square root simple. We now create the cost func on.
Cost
land cost
$50 land distance
50(5000 x)
+
water cost
+ $130 water distance
+
130 x2 + 10002 .
So we have c(x) = 50(5000 x) + 130 x2 + 10002 . This func on only

makes sense on the interval [0, 5000]. While we are fairly certain the endpoints
will not give a minimal cost, we s ll evaluate c(x) at each to verify.
c(0) = 380, 000
c(5000) 662, 873.
We now nd the cri cal values of c(x). We compute c (x) as

130x
c (x) = 50 +
.
2
x + 10002
Recognize that this is never undened. Se ng c (x) = 0 and solving for x,
we have:
130x
50 +
2
x + 10002
130x
2
x + 10002
1302 x2
x2 + 10002
1302 x2
=0
= 50
= 502
= 502 (x2 + 10002 )
1302 x2 502 x2 = 502 10002

(1302 502 )x2 = 50, 0002
50, 0002
1302 502
50, 000
x=
1302 502
50, 000
2
x=
= 416
120
3
x2 =
Evalua ng c(x) at x = 416.67 gives a cost of about $370,000. The distance

the power line is laid
along land is 5000 416.67 = 4583.33 ., and the underwater distance is 416.672 + 10002 1083 . .
Notes:
175

In the exercises you will see a variety of situa ons that require you to combine problemsolving skills with calculus. Focus on the process; learn how to
form equa ons from situa ons that can be manipulated into what you need.
Eschew memorizing how to do this kind of problem as opposed to that kind
of problem. Learning a process will benet one far longer than memorizing a
specic technique.
Notes:
176
Exercises 4.3
Terms and Concepts
1. T/F: An op miza on problem is essen ally an extreme
values problem in a story problem se ng.
2. T/F: This sec on teaches one to nd the extreme values of
func on that have more than one variable.
Problems
3. Find the maximum product of two numbers (not necessarily integers) that have a sum of 100.
12. The strength S of a wooden beam is directly propor onal

to its cross sec onal width w and the square of its height h;
that is, S = kwh2 for some constant k.
w
. 12
Given a circular log with diameter of 12 inches, what sized

beam can be cut from the log with maximum strength?
5. Find the maximum sum of two numbers whose product is

500.
13. A power line is to be run to an oshore facility in the manner described in Example 102. The oshore facility is 2
miles at sea and 5 miles along the shoreline from the power
plant. It costs $50,000 per mile to lay a power line underground and $80,000 to run the line underwater.
6. Find the maximum sum of two numbers, each of which is

in [0, 300] whose product is 500.
How much of the power line should be run underground to

minimize the overall costs?
7. Find the maximal area of a right triangle with hypotenuse

of length 1.
14. A power line is to be run to an oshore facility in the manner described in Example 102. The oshore facility is 5
miles at sea and 2 miles along the shoreline from the power
plant. It costs $50,000 per mile to lay a power line underground and $80,000 to run the line underwater.
4. Find the minimum sum of two numbers whose product is

500.
8. A rancher has 1000 feet of fencing in which to construct

adjacent, equally sized rectangular pens. What dimensions
should these pens have to maximize the enclosed area?
How much of the power line should be run underground to

minimize the overall costs?
.
9. A standard soda can is roughly cylindrical and holds 355cm3
of liquid. What dimensions should the cylinder be to minimize the material needed to produce the can? Based on
your dimensions, determine whether or not the standard
can is produced to minimize the material costs.
10. Find the dimensions of a cylindrical can with a volume of
206in3 that minimizes the surface area.
The #10 canis a standard sized can used by the restaurant industry that holds about 206in3 with a diameter of 6
2/16in and height of 7in. Does it seem these dimensions
where chosen with minimiza on in mind?
11. The United States Postal Service charges more for boxes
whose combined length and girth exceeds 108 (the
length of a package is the length of its longest side; the
girth is the perimeter of the cross sec on, i.e., 2w + 2h).
What is the maximum volume of a package with a square
cross sec on (w = h) that does not exceed the 108 standard?
15. A woman throws a s ck into a lake for her dog to fetch;

the s ck is 20 feet down the shore line and 15 feet into the
water from there. The dog may jump directly into the water and swim, or run along the shore line to get closer to
the s ck before swimming. The dog runs about 22 /s and
swims about 1.5 /s.
How far along the shore should the dog run to minimize
the me it takes to get to the s ck? (Hint: the gure from
Example 102 can be useful.)
16. A woman throws a s ck into a lake for her dog to fetch;
the s ck is 15 feet down the shore line and 30 feet into the
water from there. The dog may jump directly into the water and swim, or run along the shore line to get closer to
the s ck before swimming. The dog runs about 22 /s and
swims about 1.5 /s.
How far along the shore should the dog run to minimize the
me it takes to get to the s ck? (Google calculus dog to learn
more about a dogs ability to minimize mes.)
17. What are the dimensions of the rectangle with largest area
that can be drawn inside the unit circle?
4.4 Dieren als

In Sec on 2.2 we explored the meaning and use of the deriva ve. This sec on
starts by revisi ng some of those ideas.
Recall that the deriva ve of a func on f can be used to nd the slopes of
lines tangent to the graph of f. At x = c, the tangent line to the graph of f has
equa on
y = f (c)(x c) + f(c).
The tangent line can be used to nd good approxima ons of f(x) for values of x
near c.
For instance, we can approximate sin 1.1 using the tangent
line to the graph
of f(x) = sin x at x = /3 1.05. Recall that sin(/3) = 3/2 0.866, and
cos(/3) = 1/2. Thus the tangent line to f(x) = sin x at x = /3 is:
y
1
3
2
(x) =
0.5
In Figure 4.12(a), we see a graph of f(x) = sin x graphed along with its tangent line at x = /3. The small rectangle shows the region that is displayed in
Figure 4.12(b). In this gure, we see how we are approxima ng sin 1.1 with the
tangent line, evaluated at 1.1. Together, the two gures show how close these
values are.
Using this line to approximate sin 1.1, we have:
1
(x /3) + 0.866.
2
(a)
1
(1.1 /3) + 0.866
2
1
= (0.053) + 0.866 = 0.8925.
2
(1.1) =
(1.1) sin 1.1
0.89
(We leave it to the reader to see how good of an approxima on this is.)
0.88
sin 1.1
We now generalize this concept. Given f(x) and an xvalue c, the tangent
line is (x) = f (c)(x c) + f(c). Clearly, f(c) = (c). Let x be a small number,
represen ng a small change in x value. We assert that:
0.87
f(c + x) (c + x),
1.1
(b)
Figure 4.12: Graphing f(x) = sin x and its
tangent line at x = /3 in order to es mate sin 1.1.
since the tangent line to a func on approximates well the values of that func on
near x = c.
As the x value changes from c to c + x, the y value of f changes from f(c)
to f(c + x). We call this change of y value y. That is:
y = f(c + x) f(c).
Notes:
178
4.4 Dieren als

Replacing f(c + x) with its tangent line approxima on, we have
y (c + x) f(c)
(
)
= f (c) (c + x) c + f(c) f(c)
= f (c)x
(4.3)
This nal equa on is important; well come back to it in a moment.

We introduce two new variables, dx and dy in the context of a formal denion.
Deni on 18
Dieren als of x and y.
Let y = f(x) be dieren able. The dieren al of x, denoted dx, is any

nonzero real number (usually taken to be a small number). The dieren al of y, denoted dy, is
dy = f (x)dx.
It is helpful to organize our new concepts and nota ons in one place.
Key Idea 7
Dieren al Nota on
Let y = f(x) be a dieren able func on.

1. x represents a small, nonzero change in x value.
2. dx represents a small, nonzero change in x value (i.e., x = dx).

3. y is the change in y value as x changes by x; hence
y = f(x + x) f(x).
4. dy = f (x)dx which, by Equa on (4.3), is an approxima on of the
change in y value as x changes by x; dy y.
What is the value of dieren als? Like many mathema cal concepts, dieren als provide both prac cal and theore cal benets. We explore both here.
. Example 103
.Finding and using dieren als
Consider f(x) = x2 . Knowing f(3) = 9, approximate f(3.1).
Notes:
179

The x value is changing from x = 3 to x = 3.1; therefore, we
S
see that dx = 0.1. If we know how much the y value changes from f(3) to f(3.1)
(i.e., if we know y), we will know exactly what f(3.1) is (since we already know
f(3)). We can approximate y with dy.
y dy
= f (3)dx
= 2 3 0.1 = 0.6.
We expect the y value to change by about 0.6, so we approximate f(3.1)
9.6.
We leave it to the reader to verify this, but the preceding discussion links the
dieren al to the tangent line of f(x) at x = 3. One can verify that the tangent
line, evaluated at x = 3.1, also gives y = 9.6. .
Of course, it is easy to compute the actual answer (by hand or with a calculator): 3.12 = 9.61. (Before we get too cynical and say Then why bother?, note
our approxima on is really good!)
So why bother?
In most real life situa ons, we do not know the func on that describes
a par cular behavior. Instead, we can only take measurements of how things
change measurements of the deriva ve.
Imagine water owing down a winding channel. It is easy to measure the
speed and direc on (i.e., the velocity) of water at any loca on. It is very hard
to create a func on that describes the overall ow, hence it is hard to predict
where a oa ng object placed at the beginning of the channel will end up. However, we can approximate the path of an object using dieren als. Over small
intervals, the path taken by a oa ng object is essen ally linear. Dieren als
allow us to approximate the true path by piecing together lots of short, linear
paths. This technique is called Eulers Method, studied in introductory Dierenal Equa ons courses.
We use dieren als once more to approximate the value of a func on. Even
though calculators are very accessible, it is neat to see how these techniques can
some mes be used to easily compute something that looks rather hard.
. Example 104
.Using dieren als to approximate a func on value
Approximate 4.5.
We expect 4.5 2, yet we can do be er.Let f(x) = x,

and let c = 4. Thus f(4) = 2. We can compute f (x) = 1/(2 x), so f (4) =
1/4.
We approximate the dierence between f(4.5) and f(4) using dieren als,
S
Notes:
180
4.4 Dieren als

with dx = 0.5:
f(4.5) f(4) = y dy = f (4) dx = 1/4 1/2 = 1/8 = 0.125.
The
approximate change in f from x = 4 to x = 4.5 is 0.125, so we approximate
4.5 2.125. .
Dieren als are important when we discuss integra on. When we study
that topic, we will use nota on such as
f(x) dx
quite o en. While we dont discuss here what all of that nota on means, note
the existence of the dieren al dx. Proper handling of integrals comes with
proper handling of dieren als.
In light of that, we prac ce nding dieren als in general.
. Example 105
Finding dieren als
In each of the following, nd the dieren al dy.
1. y = sin x
2. y = ex (x2 + 2)
3. y =
x2 + 3x 1
1. y = sin x:
As f(x) = sin x, f (x) = cos x. Thus

dy = cos(x)dx.
2. y = ex (x2 + 2): Let f(x) = ex (x2 + 2). We need f (x), requiring the
Product Rule.
We have f (x) = ex (x2 + 2) + 2xex , so
dy = (ex (x2 + 2) + 2xex )dx.
3. y = x2 + 3x 1: Let f(x) = x2 + 3x 1; we need f (x), requiring

the Chain Rule.
1
1
2x + 3
We have f (x) = (x2 + 3x 1) 2 (2x + 3) =
. Thus
2
2 x2 + 3x 1
(2x + 3)dx
dy =
.
2 x2 + 3x 1
Notes:
181

Finding the dieren al dy of y = f(x) is really no harder than nding the
deriva ve of f; we just mul ply f (x) by dx. It is important to remember that we
are not simply adding the symbol dx at the end.
We have seen a prac cal use of dieren als as they oer a good method of
making certain approxima ons. Another use is error propaga on. Suppose a
length is measured to be x, although the actual value is x + x (where we hope
x is small). This measurement of x may be used to compute some other value;
we can think of this as f(x) for some func on f. As the true length is x + x,
one really should have computed f(x + x). The dierence between f(x) and
f(x + x) is the propagated error.
How close are f(x) and f(x + x)? This is a dierence in y values;
f(x + x) f(x) = y dy.
We can approximate the propagated error using dieren als.
. Example 106
.Using dieren als to approximate propagated error
A steel ball bearing is to be manufactured with a diameter of 2cm. The manufacturing process has a tolerance of 0.1mm in the diameter. Given that the
density of steel is about 7.85g/cm3 , es mate the propagated error in the mass
of the ball bearing.
The mass of a ball bearing is found using the equa on mass
S
= volume density. In this situa on the mass func on is a product of the radius
of the ball bearing, hence it is m = 7.85 43 r3 . The dieren al of the mass is
dm = 31.4r2 dr.
The radius is to be 1cm; the manufacturing tolerance in the radius is 0.05mm,
or 0.005cm. The propagated error is approximately:
m dm
= 31.4(1)2 (0.005)
= 0.493g
Is this error signicant? It certainly depends on the applica on, but we can get
an idea by compu ng the rela ve error. The ra o between amount of error to
the total mass is
dm
0.493
=
m
7.85 34
0.493
=
32.88
= 0.015,
Notes:
182
4.4 Dieren als

or 1.5%.
We leave it to the reader to conrm this, but if the diameter of the ball was
supposed to be 10cm, the same manufacturing tolerance would give a propagated error in mass of 12.33g, which corresponds to a percent error of 0.188%.
While the amount of error is much greater (12.33 > 0.493), the percent error
is much lower. .
We rst learned of the deriva ve in the context of instantaneous rates of
change and slopes of tangent lines. We furthered our understanding of the
power of the deriva ve by studying how it relates to the graph of a func on
(leading to ideas of increasing/decreasing and concavity). This chapter has put
the deriva ve to yet more uses:
Equa on solving (Newtons Method)
Related Rates (furthering our use of the deriva ve to nd instantaneous
rates of change)
Op miza on (applied extreme values), and
Dieren als (useful for various approxima ons and for something called
integra on).
In the next chapters, we will consider the reverse problem to compu ng
the deriva ve: given a func on f, can we nd a func on whose deriva ve is f?
Be able to do so opens up an incredible world of mathema cs and applica ons.
Notes:
183
Exercises 4.4
Terms and Concepts
1. T/F: Given a dieren able func on y = f(x), we are generally free to choose a value for dx, which then determines
the value of dy.
2. T/F: The symbols dx and x represent the same concept.
3. T/F: The symbols dy and y represent the same concept.
4. T/F: Dieren als are important in the study of integra on.
5. How are dieren als and tangent lines related?
Problems
In Exercises 6 17, use dieren als to approximate the given
value by hand.
6. 2.052
7. 5.932
8. 5.13
9. 6.83
10. 16.5
11. 24
12. 3 63
13. 3 8.5
14. sin 3
15. cos 1.5
16. e0.1
In Exercises 17 29, compute the dieren al dy.
17. y = x2 + 3x 5
18. y = x7 x5
1
19. y = 2
4x
20. y = (2x + sin x)2
21. y = x2 e3x
4
22. y = 4
x
2x
23. y =
tan x + 1
24. y = ln(5x)
25. y = ex sin x
26. y = cos(sin x)
x+1
27. y =
x+2
28. y = 3x ln x
29. y = x ln x x
30. A set of plas c spheres are to be made with a diameter
of 1cm. If the manufacturing process is accurate to 1mm,
what is the propagated error in volume of the spheres?
31. The distance, in feet, a stone drops in t seconds is given by

d(t) = 16t2 . The depth of a hole is to be approximated by
dropping a rock and listening for it to hit the bo om. What
is the propagated error if the me measurement is accurate
to 2/10ths of a second and the measured me is:
(a) 2 seconds?
(b) 5 seconds?
32. What is the propagated error in the measurement of the
cross sec onal area of a circular log if the diameter is measured at 15 , accurate to 1/4 ?
33. A wall is to be painted that is 8 high and is measured to
be 10 , 7 long. Find the propagated error in the measurement of the walls surface area if the measurement is accurate to 1/2 .
Exercises 34 38 explore some issues related to surveying in
which distances are approximated using other measured distances and measured angles. (Hint: Convert all angles to radians before compu ng.)
34. The length l of a long wall is to be approximated. The angle
, as shown in the diagram (not to scale), is measured to be
85.2 , accurate to 1 . Assume that the triangle formed is a
right triangle.
l =?
25
(a) What is the measured length l of the wall?

(b) What is the propagated error?
(c) What is the percent error?
35. Answer the ques ons of Exercise 34, but with a measured
angle of 71.5 , accurate to 1 , measured from a point 100
from the wall.
36. The length l of a long wall is to be calculated by measuring
the angle shown in the diagram (not to scale). Assume
the formed triangle is an isosceles triangle. The measured
angle is 143 , accurate to 1 .
.50
l =?
(a) What is the measured length of the wall?

(b) What is the propagated error?
(c) What is the percent error?
37. The length of the walls in Exercises 34 36 are essen ally
the same. Which setup gives the most accurate result?
38. Consider the setup in Exercises 36. This me, assume the
angle measurement of 143 is exact but the measured 50
from the wall is accurate to 6 . What is the approximate
percent error?
5: I
We have spent considerable me considering the deriva ves of a func on and
their applica ons. In the following chapters, we are going to star ng thinking
in the other direc on. That is, given a func on f(x), we are going to consider
func ons F(x) such that F (x) = f(x).
5.1
An deriva ves and Indenite Integra on
Given a func on y = f(x), a dieren al equa on is one that incorporates y, x,

and the deriva ves of y. For instance, a simple dieren al equa on is:
y = 2x.
Solving a dieren al equa on amounts to nding a func on y that sa ses
the given equa on. Take a moment and consider that equa on; can you nd a
func on y such that y = 2x?
Can you nd another?
And yet another?
Hopefully one was able to come up with at least one solu on: y = x2 . Finding another may have seemed impossible un l one realizes that a func on like
y = x2 + 1 also has a deriva ve of 2x. Once that discovery is made, nding yet
another is not dicult; the func on y = x2 + 123, 456, 789 also has a derivave of 2x. The dieren al equa on y = 2x has many solu ons. This leads us
to some deni ons.
Deni on 19
An deriva ves and Indenite Integrals
Let a func on f(x) be given. An an deriva ve of f(x) is a func on F(x)

such that F (x) = f(x).
The set of all an deriva ves of f(x) is the indenite integral of f, denoted
by
f(x) dx.
Chapter 5 Integra on
Make a note about our deni on: we refer to an an deriva ve of f, as opposed to the an deriva ve of f, since there is always an innite number of them.
We o en use upper-case le ers to denote an deriva ves.
Knowing one an deriva ve of f allows us to nd innitely more, simply by
adding a constant. Not only does this give us more an deriva ves, it gives us all
of them.
.
Theorem 34
An deriva ve Forms
Let F(x) and G(x) be an deriva ves .

of f(x). Then there exists a constant
C such that
G(x) = F(x) + C.
Given a func on f and one of its an deriva ves F, we know all an deriva ves
of f have the form F(x) + C for some constant C. Using Deni on 19, we can say
that
f(x) dx = F(x) + C.
Lets analyze this indenite integral nota on.
Integra on
symbol
Dieren al
of x
Constant of
integra on
f(x) dx = F(x) + C
Integrand
One
an deriva ve
Figure 5.1: Understanding

the indenite integral nota on.
.
Figure 5.1shows the typical nota on of the indenite integral. The integraon symbol, , is in reality an elongated S, represen ng take the sum. We
will later see how sums and an deriva ves are related.
The func on we want to nd an an deriva ve of is called the integrand. It
contains the dieren al of the variable we are integra ng with respect to. The
symbol and the dieren al dxare not bookends with a func on sandwiched in
between; rather, the symbol means nd all an deriva ves of what follows,
and the func on f(x) and dx are mul plied together; the dx does not just sit
there.
Lets prac ce using this nota on.
Notes:
186

. Example 107
Evalua ng indenite integrals
Evaluate sin x dx.

We are asked to nd all func ons F(x) such that F (x) =
d
sin x. Some thought will lead us to one solu on: F(x) = cos x, because dx
( cos x) =
sin x.
The indenite integral of sin x is thus cos x, plus a constant of integra on.
So:
sin x dx = cos x + C.
.
S
A commonly asked ques on is What happened to the dx? The unenlightened response is Dont worry about it. It just goes away. A full understanding
includes the following.
This process of an dieren a on is really solving a dieren al ques on. The
integral
sin x dx
presents us with a dieren al, dy = sin x dx. It is asking: What is y? We found
lots of solu ons, all of the form y = cos x + C.
Le ng dy = sin x dx, rewrite
sin x dx as
dy.
This is asking: What func ons have a dieren al of the form dy? The answer
is Func ons of the form y + C, where C is a constant. What is y? We have lots
of choices, all diering by a constant; the simplest choice is y = cos x.
Understanding all of this is more important later as we try to nd an derivaves of more complicated func ons. In this sec on, we will simply explore the
rules of indenite integra on, and one can succeed for now with answering
What happened to the dx? with It went away.
Lets prac ce once more before sta ng integra on rules.
. Example 108
.Evalua ng indenite integrals
2
Evaluate (3x + 4x + 5) dx.
We seek a func on F(x) whose deriva ve is 3x2 + 4x + 5.
When taking deriva ves, we can consider func ons termbyterm, so we can
likely do that here.
What func ons have a deriva ve of 3x2 ? Some thought will lead us to a
cubic, specically x3 + C1 , where C1 is a constant.
S
Notes:
187
What func ons have a deriva ve of 4x? Here the x term is raised to the rst
power, so we likely seek a quadra c. Some thought should lead us to 2x2 + C2 ,
where C2 is a constant.
Finally, what func ons have a deriva ve of 5? Func ons of the form 5x + C3 ,
where C3 is a constant.
Our answer appears to be
(3x2 + 4x + 5) dx = x3 + C1 + 2x2 + C2 + 5x + C3 .
We do not need three separate constants of integra on; combine them as one
constant, giving the nal answer of
(3x2 + 4x + 5) dx = x3 + 2x2 + 5x + C.
It is easy to verify our answer; take the deriva ve of x3 + 2x3 + 5x + C and

see we indeed get 3x2 + 4x + 5. .
This nal step of verifying our answer is important both prac cally and
theore cally. In general, taking deriva ves is easier than nding an deriva ves
so checking our work is easy and vital as we learn.
We also see that taking the deriva ve of our answer returns the func on in
the integrand. Thus we can say that:
d
dx
)
f(x) dx = f(x).
Dieren a on undoes the work done by an dieren a on.

Theorem 24 gave a list of the deriva ves of common func ons we had learned
at that point. We restate part of that list here to stress the rela onship between
deriva ves and an deriva ves. This list will also be useful as a glossary of common an deriva ves as we learn.
Notes:
188
Theorem 35
Deriva ves and An deriva ves
Common Dieren a on Rules Common Indenite Integral Rules

(
)
d
1. dx
cf(x) = c f (x)
1. c f(x) dx = c f(x) dx
)
(
)
(
d
f(x) g(x) dx =
2. dx
f(x) g(x) =
2.
f(x) dx g(x) dx
f (x) g (x)
( )
d
3. dx
C =0
3. 0 dx = C
( )
d
4. dx
x =1
4. 1 dx = dx = x + C
( n)
d
1 n+1
5. dx
x = n xn1
5. xn dx = n+1
x
+ C (n = 1)
(
)
d
6. dx
sin x = cos x
6. cos x dx = sin x + C
(
)
d
7. dx
cos x = sin x
7. sin x dx = cos x + C
(
)
d
8. dx
tan x = sec2 x
8. sec2 x dx = tan x + C
(
)
d
9. dx
csc x = csc x cot x
9. csc x cot x dx = csc x + C
(
)
d
10. dx
sec x = sec x tan x
10. sec x tan x dx = sec x + C
(
)
d
11. dx
cot x = csc2 x
11. csc2 x dx = cot x + C
( x)
d
12. dx
e = ex
12. ex dx = ex + C
( x)
d
13. dx
a = ln a ax
13. ax dx = ln1a ax + C
(
)
d
ln x = 1x
14. dx
14. 1x dx = ln |x| + C
We highlight a few important points from Theorem 35:
Rule #1 states c f(x) dx = c f(x) dx. This is the Constant Mul ple
Rule: we can temporarily ignore constants when nding
( 2 ) an deriva ves,
d
just as we did when compu ng deriva ves (i.e., dx
3x is just as easy to
( 2)
d
compute as dx x ). An example:
5 cos x dx = 5 cos x dx = 5 (sin x + C) = 5 sin x + C.

In the last step we can consider the constant as also being mul plied by
Notes:
189
5, but 5 mes a constant is s ll a constant, so we just write C .
Rule #2 is the Sum/Dierence Rule: we can split integrals apart when the
integrand contains terms that are added/subtracted, as we did in Example
108. So:
(3x2 + 4x + 5) dx = 3x2 dx + 4x dx + 5 dx
= 3 x2 dx + 4 x dx + 5 dx
1
1
= 3 x3 + 4 x2 + 5x + C
3
2
= x3 + 2x2 + 5x + C
In prac ce we generally do not write out all these steps, but we demonstrate them here for completeness.
Rule #5 is the Power Rule of indenite integra on. There are two important things to keep in mind:
1. No ce the restric on that n = 1. This is important: 1x dx =

01 x0 + C; rather, see Rule #14.
2. We are presen ng an dieren a on as the inverse opera on of
dieren a on. Here is a useful quote to remember:
Inverse opera ons do the opposite things in the opposite
order.
When taking a deriva ve using the Power Rule, we rst mul ply by
the power, then second subtract 1 from the power. To nd the anderiva ve, do the opposite things in the opposite order: rst add
one to the power, then second divide by the power.
Note that Rule #14 incorporates the absolute value of x. The exercises will
work the reader through why this is the case; for now, know the absolute
value is important and cannot be ignored.
Ini al Value Problems

In Sec on 2.3 we saw that the deriva ve of a posi on func on gave a velocity func on, and the deriva ve of a velocity func on describes the accelera on.
We can now go the other way: the an deriva ve of an accelera on func on
gives a velocity func on, etc. While there is just one deriva ve of a given funcon, there are innite an deriva ves. Therefore we cannot ask What is the
velocity of an object whose accelera on is 32 /s2 ?, since there is more than
one answer.
Notes:
190

We can nd the answer if we provide more informa on with the ques on,
as done in the following example.
. Example 109
Solving ini al value problems
The accelera on due to gravity of a falling object is 32 /s2 . At me t = 3,
a falling object had a velocity of 10 /s. Find the equa on of the objects
velocity.
S
We want to know a velocity func on, v(t). We know two
things:
The accelera on, i.e., v (t) = 32, and
the velocity at a specic me, i.e., v(3) = 10.
Using the rst piece of informa on, we know that v(t) is an an deriva ve of
v (t) = 32. So we begin by nding the indenite integral of 32:
(32) dt = 32t + C = v(t).

Now we use the fact that v(3) = 10 to nd C:
v(t) = 32t + C
v(3) = 10
32(3) + C = 10
C = 86
Thus v(t) = 32t + 86. We can use this equa on to understand the mo on
of the object: when t = 0, the object had a velocity of v(0) = 86 /s. Since the
velocity is posi ve, the object was moving upward.
When did the object begin moving down? Immediately a er v(t) = 0:
32t + 86 = 0
t=
43
2.69s.
16
Recognize that we are able to determine quite a bit about the path of the object
knowing just its accelera on and its velocity at a single point in me. .
. Example 110
.Solving ini al value problems
Find f(t), given that f (t) = cos t, f (0) = 3 and f(0) = 5.
S
We start by nding f (t), which is an an deriva ve of f (t):
f (t) dt = cos t dt = sin t + C = f (t).
Notes:
191
So f (t) = sin t + C for the correct value of C. We are given that f (0) = 3,
so:
f (0) = 3
sin 0 + C = 3
C = 3.
Using the ini al value, we have found f (t) = sin t + 3.

We now nd f(t) by integra ng again.
f (t) dt = (sin t + 3) dt = cos t + 3t + C.

We are given that f(0) = 5, so
cos 0 + 3(0) + C = 5
1 + C = 5
C=6
Thus f(t) = cos t + 3t + 6. .
This sec on introduced an deriva ves and the indenite integral. We found
they are needed when nding a func on given informa on about its derivave(s). For instance, we found a posi on func on given a velocity func on.
In the next sec on, we will see how posi on and velocity are unexpectedly
related by the areas of certain regions on a graph of the velocity func on. Then,
in Sec on 5.4, we will see how areas and an deriva ves are closely ed together.
Notes:
192
Exercises 5.1
Terms and Concepts
1. Dene the term an deriva ve in your own words.

2. Is it more accurate to refer to the an deriva ve of f(x) or
an an deriva ve of f(x)?
3. Use your own words to dene the indenite integral of
f(x).
4. Fill in the blanks: Inverse opera ons do the
things in the
order.
func-
Problems
In Exercises 8 26, evaluate the given indenite integral.
8.
3x3 dx
x8 dx
(10x2 2) dx
10.
dt
11.
12.
1 ds
13.
14.
15.
1
dt
3t2
3
dt
t2
1
dx
x
sec2 d
16.
sin d
18.
(sec x tan x + csc x cot x) dx
5e d
19.
3t dt
20.
21.
x2 x3 dx
e dx
25.
t dx
27. This
problem inves gates why Theorem 35 states that
1
dx = ln |x| + C.
x
(a) What is the domain of y = ln x?
(
)
(b) Find dxd ln x .
(c) What is the domain of y = ln(x)?
(
)
(d) Find dxd ln(x) .
(e) You should nd that 1/x has two types of an derivaves, depending on whether x > 0 or x < 0. In
1
one expression, give a formula for
dx that takes
x
these dierent domains into account, and explain
your answer.
In Exercises 28 38, nd f(x) described by the given ini al
value problem.
28. f (x) = sin x and f(0) = 2
29. f (x) = 5ex and f(0) = 10
30. f (x) = 4x3 3x2 and f(1) = 9
31. f (x) = sec2 x and f(/4) = 5
32. f (x) = 7x and f(2) = 1
33. f (x) = 5 and f (0) = 7, f(0) = 3
34. f (x) = 7x and f (1) = 1, f(1) = 10
35. f (x) = 5ex and f (0) = 3, f(0) = 5
36. f (x) = sin and f () = 2, f() = 4
17.
24.
7. The an deriva ve of an accelera on func on is a

func on.
9.
(t2 + 3)(t3 2t) dt
23.
26.
5. What is an ini al value problem?

6. The deriva ve of a posi on func on is a
on.
(2t + 3)2 dt
22.
5t
dt
2
37. f (x) = 24x2 + 2x cos x and f (0) = 5, f(0) = 0

38. f (x) = 0 and f (1) = 3, f(1) = 1
Review
39. Use informa on gained from the rst and second deriva1
.
ves to sketch f(x) = x
e +1
40. Given y = x2 ex cos x, nd dy.
5.2 The Denite Integral
y ( /s)
5
t (s)
5
10
Figure 5.2: The area under a constant

velocity func on corresponds to distance
traveled.
y ( /s)
5
We start with an easy problem. An object travels in a straight line at a constant

velocity of 5 /s for 10 seconds. How far away from its star ng point is the
object?
We approach this problem with the familiar Distance = Rate Time equaon. In this case, Distance = 5 /s 10s = 50 feet.
It is interes ng to note that this solu on of 50 feet can be represented graphically. Consider Figure 5.2, where the constant velocity of 5 /s is graphed on the
axes. Shading the area under the line from t = 0 to t = 10 gives a rectangle
with an area of 50 square units; when one considers the units of the axes, we
can say this area represents 50 .
Now consider a slightly harder situa on (and not par cularly realis c): an
object travels in a straight line with a constant velocity of 5 /s for 10 seconds,
then instantly reverses course at a rate of 2 /s for 4 seconds. (Since the object
is traveling in the opposite direc on when reversing course, we say the velocity
is a constant 2 /s.) How far away from the star ng point is the object what
is its displacement?
Here we use Distance = Rate1 Time1 + Rate2 Time2 , which is
Distance = 5 10 + (2) 4 = 42 .
t (s)
5
10
15
Hence the object is 42 feet from its star ng loca on.

We can again depict this situa on graphically. In Figure 5.3 we have the
veloci es graphed as straight lines on [0, 10] and [10, 14], respec vely. The displacement of the object is
Area above the taxis
Figure 5.3: The total displacement is the

area above the taxis minus the area below the taxis.
Area below the taxis,
which is easy to calculate as 50 8 = 42 feet.

Now consider a more dicult problem.
. Example 111
.Finding posi on using velocity
The velocity of an object moving straight up/down under the accelera on of
gravity is given as v(t) = 32t+48, where me t is given in seconds and velocity
is in /s. When t = 0, the object had a height of 0 .
1. What was the ini al velocity of the object?
2. What was the maximum height of the object?
3. What was the height of the object at me t = 2?
It is straigh orward to nd the ini al velocity; at me t = 0,
S
v(0) = 32 0 + 48 = 48 /s.
Notes:
194

To answer ques ons about the height of the object, we need to nd the
objects posi on func on s(t). This is an ini al value problem, which we studied
in the previous sec on. We are told the ini al height is 0, i.e., s(0) = 0. We
know s (t) = v(t) = 32t + 48. To nd s, we nd the indenite integral of v(t):
v(t) dt = (32t + 48) dt = 16t2 + 48t + C = s(t).

Since s(0) = 0, we conclude that C = 0 and s(t) = 16t2 + 48t.
To nd the maximum height of the object, we need to nd the maximum of
s. Recalling our work nding extreme values, we nd the cri cal points of s by
se ng its deriva ve equal to 0 and solving for t:
s (t) = 32t + 48 = 0
t = 48/32 = 1.5s.
(No ce how we ended just nding when the velocity was 0 /s!) The rst derivave test shows this is a maximum, so the maximum height of the object is found
at
s(1.5) = 16(1.5)2 + 48(1.5) = 36 .
The height at me t = 2 is now straigh orward to compute: it is s(2) = 32 .
y ( /s)
While we have answered all three ques ons, lets look at them again graphically, using the concepts of area that we explored earlier.
Figure 5.4 shows a graph of v(t) on axes from t = 0 to t = 3. It is again
straigh orward to nd v(0). How can we use the graph to nd the maximum
height of the object?
Recall how in our previous work that the displacement of the object (in this
case, its height) was found as the area under the velocity curve, as shaded in the
gure. Moreover, the area between the curve and the taxis that is below the
taxis counted as nega ve area. That is, it represents the object coming back
toward its star ng posi on. So to nd the maximum distance from the star ng
point the maximum height we nd the area under the velocity line that is
above the taxis. This region is a triangle; its area is
Area =
1
1
Base Height = 1.5s 48 /s = 36 .
2
2
50
t (s)
1
50
.
Figure 5.4: A graph of v(t) = 32t +

48; the shaded areas help determine displacement.
Finally, nd the total signed area under the velocity func on from t = 0 to
t = 2 to nd the total displacement of the object. That is,
Displacement = Area above the taxis Area below taxis.
The regions are triangles, and we nd
Displacement =
.
1
1
(1.5s)(48 /s) (.5s)(16 /s) = 32 .
2
2
Notes:
195
The above example does not prove a rela onship between area under a velocity func on and displacement, but it does imply a rela onship exists. Sec on
5.4 will fully establish fact that the area under a velocity func on is displacement.
.
Deni on 20
The Denite Integral, Total Signed Area
Let y = f(x) be dened on a closed interval [a, b]. The total signed area
from x = a to x = b under f is:
(area under f and above xaxis on [a, b]) (area above f and under
xaxis on [a, b]).
The denite integral of f on [a, b] is the total signed area of f on [a, b],
denoted
f(x) dx,
a
where a and b are the bounds of integra on.
The previous sec on introduced the indenite integral, which related to anderiva ves. We have now dened the denite integral, which relates to areas
under a func on. The two are very much related, as well see when we learn
the Fundamental
Theorem of Calculus in Sec on 5.4. Recall that earlier we said
that the symbol was an elongated S that represented nding a sum. In

the context of the denite integral, this nota on makes a bit more sense, as we
are adding up areas under the func on f.
We prac ce using this nota on.
y
. Example 112
.Evalua ng denite integrals
Consider the func on f given in Figure 5.5.
Find:
x
1
f(x) dx
1
.
Figure 5.5: A graph of f(x) in Example 112.
f(x) dx
0
Notes:
196
0
1
5.
f(x) dx
1
3.
5f(x) dx
f(x) dx
4.
2.
1.

1.
2.
3.
4.
5.
.
f(x) dx is the area under f on the interval [0, 3]. This region is a triangle,
0
so the area is 21 (3)(1) = 1.5.
5
f(x) dx represents the area of the triangle found under the xaxis on
3
[3, 5]. The area is 21 (2)(1) = 1; since it is found under the xaxis, this is
5
nega ve area. Therefore 3 f(x) dx = 1.
5
f(x) dx is the total signed area under f on [0, 5]. This is 1.5+(1) = 0.5.
0
3
5f(x) dx is the area under 5f on [0, 3]. This is sketched in Figure 5.6.
0
Again, the region is a triangle, with height
5 mes that of the height of
3
the original triangle. Thus the area is 0 5f(x) dx = 15/2 = 7.5.
1
f(x) dx is the area under f on the interval [1, 1]. This describes a line
1
segment, not a region; it has no width. Therefore the area is 0.
y
5
x
1
5
.
Figure 5.6: A graph of 5f in Example 112.

(Yes, it looks just like the graph of f in Figure 5.5, just with a dierent y-scale.)
This example illustrates some of the proper es of the denite integral, given
here.
Theorem 36
Proper es of the Denite Integral
Let f and g be dened on a closed interval I that contains the values a, b

and c, and let k be a constant. The following hold:
a
1.
f(x) dx = 0
a
2.
a
f(x) dx
b
4.
)
f(x) g(x) dx =
5.
a
f(x) dx
f(x) dx =
.
c
f(x) dx =
b
3.
f(x) dx +
a
b
k f(x) dx = k
f(x) dx
g(x) dx
a
f(x) dx
a
We give a brief jus ca on of Theorem 36 here.

1. As demonstrated in Example 112, there is no area under the curve when
the region has no width; hence this denite integral is 0.
Notes:
197
2. This states that total area is the sum of the areas of subregions. It is easily
considered when we let a < b < c. We can break the interval [a, c] into
two subintervals, [a, b] and [b, c]. The total area over [a, c] is the area over
[a, b] plus the area over [b, c].
It is important to note that this s ll holds true even if a < b < c is not
true. We discuss this in the next point.
3. This property can be viewed a merely a conven on to make other properes work well. (Later we will see how this property has a jus ca on all its
own, not necessarily in support of other proper es.) Suppose b < a < c.
The discussion from the previous point clearly jus es
a
c
c
f(x) dx +
f(x) dx =
f(x) dx.
(5.1)
b
However, we s ll claim that, as originally stated,
f(x) dx +
a
f(x) dx =
b
f(x) dx.
(5.2)
How do Equa ons (5.1) and (5.2) relate? Start with Equa on (5.1):
a
c
c
f(x) dx +
f(x) dx =
f(x) dx
b
a
b
c
a
c
f(x) dx =
f(x) dx +
f(x) dx
a
Property (3) jus

es changing the sign and switching the bounds of intea
gra on on the
f(x) dx term; when this is done, Equa ons (5.1) and
b
(5.2) are equivalent.

The conclusion is this: by adop ng the conven on of Property (3), Property (2) holds no ma er the order of a, b and c.
4,5. Each of these may be nonintui ve. Property (5) states that when one
scales a func on by, for instance, 7, the area of the enclosed region also
is scaled by a factor of 7. Both Proper es (4) and (5) can be proved using
geometry. The details are not complicated but are not discussed here.
Notes:
198

. Example 113
Evalua ng denite integrals using Theorem 36.
Consider the graph of a func on f(x) shown in Figure 5.7.
Answer the following:
b
c
1. Which value is greater:
f(x) dx or
f(x) dx?
2. Is
f(x) dx greater or less than 0?

a
3. Which value is greater:
f(x) dx or
a
f(x) dx?
Figure 5.7: A graph of a func on in Example 113.
b
f(x) dx has a posi ve value (since the area is above the xaxis) whereas
ac
b
f(x)
dx has a nega ve value. Hence a f(x) dx is bigger.
b
c
2. a f(x) dx is the total signed area under f between x = a and x = c. Since
the region below the xaxis looks to be larger than the region above, we
conclude that the denite integral has a value less than 0.
b
3. Note how the second integral has the bounds reversed. Therefore c f(x) dx
represents a posi ve number, greater than the area described by the rst
b
denite integral. Hence c f(x) dx is greater.
1.
The area deni on of the denite integral allows us to compute the denite
integral of some simple func ons.
. Example 114
.Evalua ng denite integrals using geometry
Evaluate the following denite integrals:
5
3
1.
(2x 4) dx
2.
9 x2 dx.
2
y
10
(5, 6)
5
R2
1. It is useful to sketch the func on in the integrand, as shown in Figure 5.8.

We see we need to compute the areas of two regions, which we have
labeled R1 and R2 . Both are triangles, so the area computa on is straightforward:
1
1
R1 : (4)(8) = 16
R2 : (3)6 = 9.
2
2
x
2 R1
5
(2, 8) 10
Figure 5.8: A graph of f(x) = 2x 4 in

Example 114.
Notes:
199
Region R1 lies under the xaxis, hence it is counted as nega ve area (we
can think of the height as being 8), so
5
(2x 4) dx = 9 16 = 7.
y
5
. 3

Example 114.
9 x2 in
. Example 115
Understanding mo on given velocity
Consider the graph of a velocity func on of an object moving in a straight line,
given in Figure 5.10, where the numbers in the given regions gives the area of
that region. Assume that the denite integral of a velocity func on gives displacement. Find the maximum speed of the object and its maximum displacement from its star ng posi on.
y ( /s)
15
10
38
5
t (s)
a
5
11
2. Recognize that the integrand of this denite integral is a half circle, as

sketched in Figure 5.9, with radius 3. Thus the area is:
3
1
9
9 x2 dx = r2 = .
2
2
3
.
11
.
Figure 5.10: A graph of a velocity in Example 115.
Since the graph gives velocity, nding the maximum speed

S
is simple: it looks to be 15 /s.
At me t = 0, the displacement is 0; the object is at its star ng posi on. At
me t = a, the object has moved backward 11 feet. Between mes t = a and
t = b, the object moves forward 38 feet, bringing it into a posi on 27 feet forward of its star ng posi on. From t = b to t = c the object is moving backwards
again, hence its maximum displacement is 27 feet from its star ng posi on. .
In our examples, we have either found the areas of regions that have nice
geometric shapes (such as rectangles, triangles and circles) or the areas were
given to us. Consider Figure 5.11, where a region below y = x2 is shaded. What
is its area? The func on y = x2 is rela vely simple, yet the shape it denes has
an area that is not simple to nd geometrically.
In the next sec on we will explore how to nd the areas of such regions.
10
x
1
Figure 5.11: What is the area below y =

x2 on [0, 3]? The region is not a usual geometric shape.
200
Notes:
Exercises 5.2
Terms and Concepts
f(x) dx
1. What is total signed area?
2
3
(e)
2. What is displacement?
3
3. What is
sin x dx?
f(x) dx
5
3
(f)
2f(x) dx
4. Give a single denite integral that has the same value as

1
2
(2x + 3) dx +
(2x + 3) dx.
0
y
4
1
y = f(x)
7.
Problems
In Exercises 5 9, a graph of a func on f(x) is given. Using
the geometry of the graph, evaluate the denite integrals.
y
4
y = 2x + 4
f(x) dx
2
4
2f(x) dx
(c)
(2x + 4) dx
(2x + 4) dx
3
y=x1
(2x + 4) dx
8.
(f)
(6x + 12) dx
(a)
1
6.
x
1
f(x) dx
(d)
f(x) dx
(e)
f(x) dx
(f)
(b)
0
(c)
0
(x 1) dx
(x 1) dx
0
3
(c)
(a)
(b)
1
y = f(x)
(4x 8) dx
(2x + 4) dx
(e)
. 2
(2x 4) dx
(d)
(f)
(c)
(e)
(2x + 4) dx
4x dx
(b)
(d)
(a)
(b)
f(x) dx
(a)
5.
(d)
(x 1) dx
0
3
(x 1) dx
2
4
(x 1) dx
1
4
1
(
)
(x 1) + 1 dx
(d)
3
f(x) =
f(x) dx
0
4 (x 2)2
9.
f(x) = 3x2 3
10
1
5
x
1
12.
(a)
f(x) dx
f(x) dx
2
2
(b)
f(x) dx
(d)
f(x) dx
f(x) dx
5f(x) dx
1
1
(c)
f(x) dx
In Exercises 10 13, a graph of a func on f(x) is given; the

numbers inside the shaded regions give the area of that region. Evaluate the denite integrals using this area informaon.
1
1
(d)
f(x) dx
0
y
4
y
50
11
59
f(x) = x2
y = f(x)
10.
x
2
(a)
(c)
(b)
4
4
21
2
13.
3
1
50
1/3
.
. 100
(a)
f(x) dx
f(x) dx
f(x) dx
0
2
(d)
3f(x) dx
x
2
5x2 dx
(x2 + 3) dx
(x 1)2 dx
(
)
(x 2)2 + 5 dx
In Exercises 14 15, a graph of the velocity func on of an object moving in a straight line is given. Answer the ques ons
based on that graph.
1
4/
f(x) = sin(x/2)
y ( /s)
11.
x
1
4/
14.
(d)
(c)
(c)
(b)
(b)
(a)
7/3
1
t (s)
(a)
f(x) dx
1.
(b)
f(x) dx
(b) What is the objects maximum displacement?
(c)
f(x) dx
0
(a) What is the objects maximum velocity?

(c) What is the objects total displacement on [0, 3]?
y ( /s)
(
)
f(x) g(x) dx
(
)
3f(x) + 2g(x) dx
19.
20.
15.
2
1
t (s)
1
21. Find values for a and b such that

3
(
)
af(x) + bg(x) dx = 0
0

In Exercises 22 25, let

3
s(t) dt = 10,
(c) What is the objects total displacement on [0, 5]?
(a) What is the objects ini al velocity?

(b) When is the objects displacement 0?
(c) How long does it take for the object to return to its
ini al height?
(d) When will the object reach a height of 210 feet?
In Exercises 18 21, let
2
f(x) dx = 5,
0
3
f(x) dx = 7,
g(x) dx = 5.
Use these values to evaluate the given denite integrals.

2
(
)
18.
f(x) + g(x) dx
0
r(t) dt = 11.
0
Use these values to evaluate the given denite integrals.

3
(
)
22.
s(t) + r(t) dt
0
0
(
)
s(t) r(t) dt
23.
24.
)
s(t) 7r(t) dt
25. Find values for a and b such that

5
(
)
ar(t) + bs(t) dt = 0
0
Review
( 3
)
x 2x2 + 7x 9 dx
26.
27.
r(t) dt = 1, and
g(x) dx = 3, and
17. An object is thrown straight up with a velocity, in /s, given

by v(t) = 32t + 96, where t is in seconds, from a height
of 64 feet.
s(t) dt = 8,

(c) When does the maximum displacement occur?
16. An object is thrown straight up with a velocity, in /s, given

by v(t) = 32t + 64, where t is in seconds, from a height
of 48 feet.
(d) When will the object reach a height of 0? (Hint: nd

when the displacement is 48 .)
)
sin x cos x + sec2 x dx
(
3
)
1
+ 2t dt
2
t
)
(
1
29.
csc x cot x dx
x
28.
t+
5.3 Riemann Sums
4
3
2
1
x
1
Figure 5.12: A graph of f(x) = 4x x2 .

What is the area of the shaded region?
4
3
2
1
MPR
RHR
LHR
2
other
3
x
4
4
Figure 5.13: Approxima ng 0 (4xx2 ) dx
using rectangles. The heights of the
rectangles are determined using dierent
rules.
In the previous sec on we dened the denite integral of a func on on [a, b] to

be the signed area between the curve and the xaxis. Some areas were simple
to compute; we ended the sec on with a region whose area was not simple to
compute. In this sec on we develop a technique to nd such areas.
A fundamental calculus technique is to rst answer a given problem with an
approxima on, then rene that approxima on to make it be er, then use limits
in the rening process to nd the exact answer. That is exactly what we will do
here.
Consider the region given in Figure 5.12, which is the area under y = 4x x2
4
on [0, 4]. What is the signed area of this region i.e., what is 0 (4x x2 ) dx?
We start by approxima ng. We can surround the region with a rectangle
with height and width of 4 and nd the area is approximately 16 square units.
This is obviously an overapproxima on; we are including area in the rectangle
that is not under the parabola.
We have an approxima on of the area, using one rectangle. How can we
rene our approxima on to make it be er? The key to this sec on is this answer:
use more rectangles.
Lets use 4 rectangles of equal width of 1. This par ons the interval [0, 4]
into 4 subintervals, [0, 1], [1, 2], [2, 3] and [3, 4]. On each subinterval we will
draw a rectangle.
There are three common ways to determine the height of these rectangles:
the Le Hand Rule, the Right Hand Rule, and the Midpoint Rule. The Le Hand
Rule says to evaluate the func on at the le hand endpoint of the subinterval
and make the rectangle that height. In Figure 5.13, the rectangle drawn on the
interval [2, 3] has height determined by the Le Hand Rule; it has a height of
f(2). (The rectangle is labeled LHR.)
The Right Hand Rule says the opposite: on each subinterval, evaluate the
func on at the right endpoint and make the rectangle that height. In the gure,
the rectangle drawn on [0, 1] is drawn using f(1) as its height; this rectangle is
labeled RHR..
The Midpoint Rule says that on each subinterval, evaluate the func on at
the midpoint and make the rectangle that height. The rectangle drawn on [1, 2]
was made using the Midpoint Rule, with a height of f(1.5). That rectangle is
labeled MPR.
These are the three most common rules for determining the heights of approxima ng rectangles, but one is not forced to use one of these three methods.
The rectangle on [3, 4] has a height of approximately f(3.53), very close to the
Midpoint Rule. It was chosen so that the area of the rectangle is exactly the area
of the region under f on [3, 4]. (Later youll be able to gure how to do this, too.)
4
The following example will approximate the value of 0 (4x x2 ) dx using
Notes:
204
5.3 Riemann Sums

these rules.
. Example 116
Using the
4 Le Hand, Right Hand and Midpoint Rules
Approximate the value of 0 (4x x2 ) dx using the Le Hand Rule, the Right
Hand Rule, and the Midpoint Rule, using 4 equally spaced subintervals.
We break the interval [0, 4] into four subintervals as before.
In Figure 5.14 we see 4 rectangles drawn on f(x) = 4x x2 using the Le Hand
Rule. (The areas of the rectangles are given in each gure.)
Note how in the rst subinterval, [0, 1], the rectangle has height f(0) = 0. We
add up the areas of each rectangle (height width) for our Le Hand Rule approxima on:
4
3
2
f(0) 1 + f(1) 1 + f(2) 1 + f(3) 1 =

0 + 3 + 4 + 3 = 10.
0
.
4
2
x
4
4
using the Le Hand Rule in Example 116.
Figure 5.15 shows 4 rectangles drawn under f using the Right Hand Rule;
note how the [3, 4] subinterval has a rectangle of height 0.
In this example, these rectangle seem to be the mirror image of those found
in Figure 5.14. (This is because of the symmetry of our shaded region.) Our
approxima on gives the same answer as before, though calculated a dierent
way:
4
3
2
1
f(1) 1 + f(2) 1 + f(3) 1 + f(4) 1 =
3 + 4 + 3 + 0 = 10.
4
1
.
Figure 5.16 shows 4 rectangles
4 drawn under f using the Midpoint Rule.
This gives an approxima on of 0 (4x x2 ) dx as:
3
1
3
2
x
4
4
using the Right Hand Rule in Example 116.
f(0.5) 1 + f(1.5) 1 + f(2.5) 1 + f(3.5) 1 =

1.75 + 3.75 + 3.75 + 1.75 = 11.
Our three methods provide two approxima ons of
4
0
(4x x2 ) dx: 10 and 11. .
Summa on Nota on
3
2
It is hard to tell at this moment which is a be er approxima on: 10 or 11?

We can con nue to rene our approxima on by using more rectangles. The
nota on can become unwieldy, though, as we add up longer and longer lists of
numbers. We introduce summa on nota on to ameliorate this problem.
1.75
3.75
1
1.75
3.75
2
x
4
4
using the Midpoint Rule in Example 116.
Notes:
205
Suppose we wish to add up a list of numbers a1 , a2 , a3 , , a9 . Instead of
wri ng
a1 + a2 + a3 + a4 + a5 + a6 + a7 + a8 + a9 ,
we use summa on nota on and write
upper
bound
summand
9
ai .
i=1
i=index
.
of summa on
lower
bound
Figure 5.17: Understanding summa on nota on.
The upper case sigma represents the term sum. The index of summa on
in this example is i; any symbol can be used. By conven on, the index takes on
only the integer values between (and including) the lower and upper bounds.
Lets prac ce using this nota on.
. Example 117
.Using summa on nota on
Let the numbers {ai } be dened as ai = 2i 1 for integers i, where i 1. So
a1 = 1, a2 = 3, a3 = 5, etc. (The output is the posi ve odd integers). Evaluate
the following summa ons:
1.
ai
2.
i=1
3.
i=3
1.
(3ai 4)
(ai )2
i=1
ai = a1 + a2 + a3 + a4 + a5 + a6
i=1
= 1 + 3 + 5 + 7 + 9 + 11
= 36.
2. Note the star ng value is dierent than 1:
7
ai = (3a3 4) + (3a4 4) + (3a5 4) + (3a6 4) + (3a7 4)
i=3
= 11 + 17 + 23 + 29 + 35
= 115.
Notes:
206
5.3 Riemann Sums

3.
4
(ai )2 = (a1 )2 + (a2 )2 + (a3 )2 + (a4 )2

i=1
= 12 + 32 + 52 + 72
= 84
.
It might seem odd to stress a new, concise way of wri ng summa ons only
to write each term out as we add them up. It is. The following theorem gives
some of the proper es of summa ons that allow us to work with them without
wri ng individual terms. Examples will follow.
Theorem 37
1.
Proper es of Summa ons
c = c n, where c is a constant.
i=1
2.
n
n
n
(ai bi ) =
ai
bi
i=m
3.
i=m
c ai = c
i=1
4.
i=m
i=m
5.
ai
ai +
i=j+1
ai =
i=1
6.
i=1
i=
n(n + 1)
2
i2 =
i=1
7.
i=1
n(n + 1)(2n + 1)
6
(
i3 =
n(n + 1)
2
)2
ai
i=m
. Example 118
.Evalua ng summa ons using Theorem 37
Revisit Example 117 and, using Theorem 37, evaluate
6
i=1
ai =
(2i 1).
i=1
Notes:
207
S
6
6
6
(2i 1) =
2i
(1)
i=1
i=1
(
=
i=1
)
i
i=1
6(6 + 1)
6
2
= 42 6 = 36
=2
We obtained the same answer without wri ng out all six terms. When dealing
with small sizes of n, it may be faster to write the terms out by hand. However,
Theorem 37 is incredibly important when dealing with large sums as well soon
see. .
Riemann Sums
.
0
x1
1
x5
2
x9
3
x13
4
x17
4
Consider again 0 (4x x2 ) dx. We will approximate this denite integral
using 16 equally spaced subintervals and the Right Hand Rule in Example 119.
Before doing so, it will pay to do some careful prepara on.
Figure 5.18 shows a number line of [0, 4] divided into 16 equally spaced
subintervals. We denote 0 as x1 ; we have marked the values of x5 , x9 , x13 and
x17 . We could mark them all, but the gure would get crowded. While it is easy
to gure that x10 = 2.25, in general, we want a method of determining the value
of xi without consul ng the gure. Consider:
Figure 5.18: Dividing [0, 4] into 16 equally

spaced subintervals.
number of
subintervals
between x1 and xi
xi = x1 + (i 1)x
.
star ng
value
subinterval
size
So x10 = x1 + 9(4/16) = 2.25.

If we had par oned [0, 4] into 100 equally spaced subintervals, each subinterval would have length x = 4/100 = 0.04. We could compute x32 as
x32 = x1 + 31(4/100) = 1.24.
(That was far faster than crea ng a sketch rst.)
Notes:
208
5.3 Riemann Sums

Given any subdivision of [0, 4], the rst subinterval is [x1 , x2 ]; the second is
[x2 , x3 ]; the i th subinterval is [xi , xi+1 ].
When using the Le Hand Rule, the height of the i th rectangle will be f(xi ).
When using the Right Hand Rule, the height of the i th rectangle will be f(xi+1 ).
(
)
xi + xi+1
th
When using the Midpoint Rule, the height of the i rectangle will be f
.
2
4
Thus approxima ng 0 (4x x2 ) dx with 16 equally spaced subintervals can
be expressed as follows:
Le Hand Rule:
16
f(xi )x
i=1
Right Hand Rule:
16
f(xi+1 )x
i=1
Midpoint Rule:
)
16 (
xi + xi+1
x
f
2
i=1
We use these formulas in the next two examples. The following example lets
us prac ce using the Right Hand Rule and the summa on formulas introduced
in Theorem 37.
. Example 119
.Approxima ng denite integrals using sums
4
Approximate 0 (4x x2 ) dx using the Right Hand Rule and summa on formulas
with 16 and 1000 equally spaced intervals.
Using the formula derived before, using 16 equally spaced
S
intervals and the Right Hand Rule, we can approximate the denite integral as
16
f(xi+1 )x.
i=1
We have x = 4/16 = 0.25. Since xi = 0 + (i 1)x, we have

xi+1 = 0 + (i + 1 1)x
= ix
Notes:
209
Using the summa on formulas, consider:
(4x x ) dx
2
16
f(xi+1 )x
i=1
16
f(ix)x
i=1
16
)
4ix (ix)2 x
i=1
16
=
(4ix2 i2 x3 )
i=1
= (4x2 )
16
i x3
i=1
16
i2
(5.3)
i=1
16(17)(33)
16 17
x3
2
6
= 4 0.252 136 0.253 1496
= (4x2 )
= 10.625
We were able to sum up the areas of 16 rectangles with very li le computaon. No ce Equa on (5.3); by changing the 16s to 1,000s (and appropriately
changing the value of x), we can use that equa on to sum up 1000 rectangles!
We do so here, skipping from the original summand to the equivalent of
Equa on (5.3) to save space. Note that x = 4/1000 = 0.004.
4
3
2
1
4
with the Right Hand Rule and 16 evenly
(4x x2 ) dx
1000
f(xi+1 )x
i=1
= (4x2 )
1000
i=1
i x3
1000
i2
i=1
1000 1001
1000(1001)(2001)
= (4x2 )
x3
2
6
2
3
= 4 0.004 500500 0.004 333, 833, 500
= 10.666656
Using many, many rectangles, we have a likely good approxima on of
x2 )x. That is,
4
(4x x2 ) dx 10.666656.
.
0
Notes:
210
4
0
(4x
5.3 Riemann Sums

Before the above example, we stated what the summa ons for the Le Hand,
Right Hand and Midpoint Rules looked like. Each had the same basic structure;
the only dierence was at what values to evaluate f. All three are examples of
an even more general construc on, named a er mathema cian Georg Friedrich
Bernhard Riemann.
Deni on 21
Riemann Sum
Let f be dened on the closed interval [a, b] and let x be a par

[a, b], with
a = x1 < x2 < . . . < xn < xn+1 = b.
on of
Let xi denote the length of the i th subinterval [xi , xi+1 ] and let ci denote
any value in the i th subinterval.
The sum
n
f(ci )xi
i=1
y
is a Riemann sum of f on [a, b].

4
In this general form, the subintervals do not have be of equal length, and one
can choose a point ci inside each subinterval any way they choose (and not just
the le endpoint, or the midpoint,
4 etc.) Figure 5.20 shows the approxima ng
rectangles of a Riemann sum of 0 (4x x2 ) dx. (This par cular approxima on
is of li le use; clearly the width and heights of the rectangles were not chosen
well.)
Usually Riemann sums are calculated using one of the three methods we
have introduced. The uniformity of construc on makes computa ons easier.
Before working another example, lets summarize some of what we have learned
in a convenient way.
Key Idea 8
Consider
3
2
1
x
Figure 5.20: An example ofa general Rie4

mann sum to approximate 0 (4xx2 ) dx.
Riemann Sum Concepts

b
f(x) dx
f(ci )xi .
i=1
1. When the n subintervals have equal length, xi = x =

2. The i th term of the par
ba
.
n
on is xi = a + (i 1)x. (This makes xn+1 = b.)
(con nued . . .)
Notes:
211
Key Idea 8
Consider
Riemann Sum Concepts Con nued

b
f(x) dx
f(ci )xi .
i=1
3. The Le Hand Rule summa on is:
f(xi )x
(ci = xi ).
i=1
4. The Right Hand Rule summa on is:
f(xi+1 )x
(ci = xi+1 ).
i=1
(
)
n
xi + xx+1
f
x
5. The Midpoint Rule summa on is:
2
i=1
(ci = (xi + xi+1 )/2).
Lets do another example.

. Example 120
3 .Approxima ng denite integrals with sums
Approximate 2 (5x + 2) dx using the Midpoint Rule and 10 equally spaced
intervals.
Following Key Idea 8, we have
x =
3 (2)
= 1/2
10
and
xi = (2) + (1/2)(i 1) = i/2 5/2.
As we are using the Midpoint Rule, we will also need xi+1 and
xi = i/2 5/2, xi+1 = (i + 1)/2 5/2 = i/2 2. This gives
xi + xi+1
. Since
2
(i/2 5/2) + (i/2 2)

i 9/2
xi + xi+1
=
=
= i/2 9/4.
2
2
2
We now construct the Riemann sum and compute its value using summa on
Notes:
212
5.3 Riemann Sums

formulas.
)
10 (
xi + xi+1
x
(5x + 2) dx
f
2
2
i=1
10
f(i/2 9/4)x
i=1
10
(
)
5(i/2 9/4) + 2 x
i=1
10 [( )
]
37
= x
i
2
4
i=1
( 10
))
10 (
5
37
= x
(i)
2 i=1
4
i=1
(
)
37
1 5 10(11)
10
=
2 2
2
4
45
=
= 22.5
2
Note the graph of f(x) = 5x + 2 in Figure 5.21. The regions whose area
is computed by the denite integral are triangles, meaning we can nd the exact answer without summa on techniques. We nd that the exact answer is
indeed 22.5. One of the strengths of the Midpoint Rule is that each rectangle
includes area that should not be counted, but misses other area that should.
When the par on size is small, these two amounts are about equal and these
errors cancel each other out.
Note too that when the func on is nega ve, the rectangles have a nega ve
height. When we compute the area of the rectangle, we use f(ci )x; when f is
nega ve, the area is counted as nega ve. .
y
17
10
x
2
3
Figure 5.21: Approxima ng 2 (5x +
2) dx using the Midpoint Rule and 10
evenly spaced subintervals in Example
120.
No ce in the previous example that while we used 10 equally spaced intervals, the number 10 didnt play a big role in the calcula ons un l the very end.
Mathema cians love to abstract ideas; lets approximate another region using n
subintervals, where we do not specify a value of n un l the very end.
. Example 121
.Approxima ng denite integrals with a formula, using sums
4
Revisit 0 (4xx2 ) dx yet again. Approximate this denite integral using the Right
Hand Rule with n equally spaced subintervals.
S
Using Key Idea 8, we know x =
40
n
= 4/n. We also nd
Notes:
213
xi = 0 + x(i 1) = 4(i 1)/n. The Right Hand Rule uses xi+1 , which is
xi+1 = 4i/n.
We construct the Right Hand Rule Riemann sum as follows:
4
n
(4x x2 ) dx
f(xi+1 )x
0
i=1
( )
n
4i
x
=
f
n
i=1
[
( )2 ]
n
4i
4i
=
4
x
n
n
i=1
)
)
n (
n (
16x 2
16x
i
i
=
n
n2
i=1
i=1
(
) n
(
) n
16x
16x 2
=
i
i
n
n2
i=1
i=1
(
)
(
)
16x n(n + 1)(2n + 1)
n(n + 1)
16x
=
n
2
n2
6
32(n + 1) 32(n + 1)(2n + 1)
=
(now simplify)
3n2
(n
)
32
1
=
1 2
3
n
recall
x = 4/n
.
The result is an amazing, easy to use formula. To approximate the denite
integral with 10 equally spaced subintervals and the Right Hand Rule, set n = 10
and compute
(
)
4
32
1
(4x x2 ) dx
1 2 = 10.56.
3
10
0
Recall how earlier we approximated the denite integral with 4 subintervals;
with n = 4, the formula gives 10, our answer as before.
It is now easy to approximate the integral with 1,000,000 subintervals! Handheld calculators will round o the answer a bit prematurely giving an answer of
10.66666667. (The actual answer is 10.666666666656.)
We now take an important leap. Up to this point, our mathema cs has been
limited to geometry and algebra (nding areas and manipula ng expressions).
Now we apply calculus. For any nite n, we know that
(
)
4
32
1
(4x x2 ) dx
1 2 .
3
n
0
Notes:
214
5.3 Riemann Sums

Both common sense and highlevel mathema cs tell us that as n gets large, the
approxima on gets be er.
4 In fact, if we take the limit as n , we get the
exact area described by 0 (4x x2 ) dx. That is,
(
)
1
32
1 2
(4x x ) dx = lim
n 3
n
32
=
(1 0)
3
32
=
= 10.6
3
2
This sec on started with a fundamental calculus technique: make an approxima on, rene the approxima on to make it be er, then use limits in the
rening process to get an exact answer. That is precisely what we just did.
Lets prac ce this again.
. Example 122
.Approxima ng denite integrals with a formula, using sums
5
Find a formula that approximates 1 x3 dx using the Right Hand Rule and n
equally spaced subintervals, then take the limit as n to nd the exact
area.
Following Key Idea 8, we have x = 5(1)
= 6/n. We
n
have xi = (1) + (i 1)x; as the Right Hand Rule uses xi+1 , we have xi+1 =
(1) + ix.
The Riemann sum corresponding to the Right Hand Rule is (followed by simplica ons):
S
5
1
x3 dx
=
=
i=1
n
f(xi+1 )x
f(1 + ix)x
i=1
n
(1 + ix)3 x
i=1
n
(
)
=
(ix)3 3(ix)2 + 3ix 1 x
i=1
n
(3 4
)
i x 3i2 x3 + 3ix2 x
(now distribute x)
(now split up summa on)
i=1
Notes:
215
= x4
i3 3x3
i=1
(
4
= x
n(n + 1)
2
i2 + 3x2
i=1
)2
i=1
3x3
i=1
n(n + 1)
n(n + 1)(2n + 1)
+ 3x2
nx
6
2
(use x = 6/n)
y
=
100
1296 n2 (n + 1)2
216 n(n + 1)(2n + 1)
36 n(n + 1)
3 3
+3 2
6
n4
4
n
6
n
2
(now do a sizable amount of algebra to simplify)
= 156 +
50
x
1
5
Figure 5.22: Approxima ng 1 x3 dx using the Right Hand Rule and 10 evenly
378 216
+ 2
n
n
Once again, we have found a compact formula for approxima ng the denite
integral with n equally spaced subintervals and the Right Hand Rule. Using 10
subintervals, we have an approxima on of 195.96 (these rectangles are shown
in Figure 5.22). Using n = 100 gives an approxima on of 159.802.
Now nd the exact answer using a limit:
)
(
5
378 216
+ 2 = 156.
x3 dx = lim 156 +
n
n
n
.
1
Limits of Riemann Sums

We have used limits to evaluate exactly given denite limits. Will this always work? We will show, given notveryrestric ve condi ons, that yes, it will
always work.
The previous two examples demonstrated how an expression such as
n
f(xi+1 )x
i=1
can be rewri en as an expression explicitly involving n, such as 32/3(1 1/n2 ).

Viewed in this manner, we can think of the summa on as a func on of n.
An n value is given (where n is a posi ve integer), and the sum of areas of n
equally spaced rectangles is returned, using the Le Hand, Right Hand, or Midpoint Rules.
b
Given a denite integral a f(x) dx, let:
SL (n) =
i=1
f(xi )x, the sum of equally spaced rectangles formed using
the Le Hand Rule,
Notes:
216
5.3 Riemann Sums
SR (n) =
f(xi+1 )x, the sum of equally spaced rectangles formed us-
i=1
ing the Right Hand Rule, and

(
)
n
xi + xi+1
SM (n) =
f
x, the sum of equally spaced rectangles
2
i=1
formed using the Midpoint Rule.
Recall the deni on of a limit as n : lim SL (n) = K if, given any > 0,
n
there exists N > 0 such that
|SL (n) K| <
when
n N.
The following theorem states that we can use any of our three rules to nd
b
the exact value of a denite integral a f(x) dx. It also goes one step further.
Let x represent any par on of [a, b], and let x denote the length of the
longest subinterval of this par
n on. The theorem also states that limit of any
Riemann sum of the form i=1 f(ci )xi , as x 0, also gives the exact
value of the denite integral.
Theorem 38
Denite Integrals and the Limit of Riemann Sums
Let f be con nuous on the closed interval [a, b] and let SL (n), SR (n) and
SM (n) be dened as before. Then:
1. lim SL (n) = lim SR (n) = lim SM (n),

n
2. lim SL (n) =
n
lim
x0
f(ci )xi , where the la er sum is any Rie-
i=1
mann sum of f on [a, b], and

b
3. lim SL (n) =
f(x) dx.
n
We summarize what we have learned over the past few sec ons here.
Knowing the area under the curve can be useful. One common example
is: the area under a velocity curve is displacement.
b
We have dened the denite integral, a f(x) dx, to be the area under f
on the interval [a, b].
Notes:
217
While we can approximate a denite integral many ways, we have focused
on using rectangles whose heights can be determined using: the Le Hand
Rule, the Right Hand Rule and the Midpoint Rule.
Sums of rectangles of this type are called Riemann sums.
The exact value of the denite integral can be computed using the limit of
a Riemann sum. We generally use one of the above methods as it makes
the algebra simpler.
We rst learned of deriva ves through limits then learned rules that made
the process simpler. We know of a way to evaluate a denite integral using limits;
in the next sec on we will see how the Fundamental Theorem of Calculus makes
the process simpler. The key feature of this theorem is its connec on between
the indenite integral and the denite integral.
Notes:
218
Exercises 5.3
Terms and Concepts
18.
1. A fundamental calculus technique is to use

ne approxima ons to get an exact answer.
2. What is the upper bound in the summa on
i=1
to re14
(48i
i=7
19.
3. This sec on approximates denite integrals using what geometric shape?
Problems
10
(4i3 + 10i2 7i + 11)

i=1
201)?
4. T/F: A sum using the Right Hand Rule is an example of a

Riemann Sum.
15
(2i3 10)
20.
10
(i3 3i2 + 2i + 7)
i=1
21. 1 + 2 + 3 + . . . + 99 + 100
In Exercises 5 11, write out each term of the summa on and

compute the sum.
5.
22. 1 + 4 + 9 + . . . + 361 + 400

i2
i=2
6.
(4i 2)
Theorem 37 states
i=1
7.
sin(i/2)
ai =
i=1
i=2
ai =
i=k+1
i=1
4 (
i=1
1
i+1
)
23.
(1)i cos(i)
11.
20
i=11
i=0
In Exercises 12 15, write each sum in summa on nota on.

12. 3 + 6 + 9 + 12 + 15
24.
25
i3
i=16
13. 1 + 0 + 3 + 8 + 15 + 24 + 35 + 48 + 63
14.
1
2
3
4
+ + +
2
3
4
5
2
15. 1 e + e e + e
25.
12
i=7
In Exercises 16 22, evaluate the summa on using Theorem

37.
16.
25
i=1
17.
10
(3i2 2i)
i=1
i=1
ai , so
i=k+1
ai
ai .
i=1
Use this fact, along with other parts of Theorem 37, to evaluate the summa ons given in Exercises 23 26.
(1)i i
9.
10.
ai +
i=1
1
8.
i
i=1
26.
10
i=5
4i3
In Exercises
27 32, a denite integral
f(x) dx is given.
(a) Graph f(x) on [a, b].
rectangles.
(5 x2 ) dx, with 4 rectangles using the Midpoint Rule.

sin x dx, with 6 rectangles using the Right Hand Rule.
30.
10
37.
10
38.
(5 x) dx, using the Right Hand Rule.
(x3 x2 ) dx, using the Right Hand Rule.
2x dx, with 5 rectangles using the Le Hand Rule.

ln x dx, with 3 rectangles using the Midpoint Rule.
(2x2 3) dx, using the Le Hand Rule.
31.
(3x 1) dx, using the Midpoint Rule.
36.
29.
3
1
28.
35.
x dx, with 6 rectangles using the Le Hand Rule.
3x2 dx, using the Le Hand Rule.
27.
34.
(b) Add to the sketch rectangles using the provided rule.

b
(c) Approximate
f(x) dx by summing the areas of the
3
x3 dx, using the Right Hand Rule.
33.
1
9
1
dx, with 4 rectangles using the Right Hand Rule.
x
32.
1
Review
In Exercises 39 44, nd an an deriva ve of the given funcon.
39. f(x) = 5 sec2 x
In Exercises
33 38, a denite integral
f(x) dx is given. As demonstrated in Examples 121

a
and 122, do the following.
(a) Find a formula to approximate

subintervals and the provided rule.
f(x) dx using n
a
(b) Evaluate the formula using n = 10, 100 and 1, 000.

(c) Find the limit
of the formula, as n to nd the
b
exact value of
f(x) dx.
a
40. f(x) =
7
x
41. g(t) = 4t5 5t3 + 8

42. g(t) = 5 8t
43. g(t) = cos t + sin t
1
44. f(x) =
x
5.4 The Fundamental Theorem of Calculus
5.4
The Fundamental Theorem of Calculus
b
Let f(t) be a con nuous func on dened on [a, b]. The denite integral a f(x) dx
is the area under f on [a, b]. We can turn this into a func on by le ng the
upper (or lower) bound vary.
x
Let F(x) = a f(t) dt. It computes the area under f on [a, x] as illustrated
in Figure 5.23. We can study this func on using our knowledge of the denite
a
integral. For instance, F(a) = 0 since a f(t) dt = 0.
We can also apply calculus ideas to F(x); in par cular, we can compute its
deriva ve. While this may seem like an innocuous thing to do, it has farreaching
implica ons, as demonstrated by the fact that the result is given as an important
theorem.
Theorem 39
The Fundamental Theorem of Calculus, Part 1

x
Let f be con nuous on [a, b] and let F(x) = a f(t) dt. Then F is a dieren able func on on (a, b), and
t
b
.
Figure 5.23: The
x area of the shaded region is F(x) = a f(t) dt.
F (x) = f(x).
Ini ally this seems simple, as demonstrated in the following example.

. Example 123
Using the Fundamental Theorem of Calculus, Part 1
x
2
Let F(x) =
(t + sin t) dt. What is F (x)?
5
Using the Fundamental Theorem of Calculus, we have F (x) =
x + sin x. .
This simple example reveals something incredible: F(x) is an an deriva ve
of x2 + sin x! Therefore, F(x) = 31 x3 cos x + C for some value of C. (We can
nd C, but generally we do not care. We know that F(5) = 0, which allows us
to compute C. In this case, C = cos(5) + 125
3 .)
We have done more than found a complicated way of compu ng an anderiva ve. Consider a func on f dened on an open interval containing a, b
b
x
and c. Suppose we want to compute a f(t) dt. First, let F(x) = c f(t) dt. Using
Notes:
221
the proper es of the denite integral found in Theorem 36, we know
b
c
b
f(t) dt =
f(t) dt +
f(t) dt
a
f(t) dt +
c
f(t) dt
c
= F(a) + F(b)
= F(b) F(a).
We now see how indenite integrals and denite integrals are related: we can
evaluate a denite integral using an deriva ves! This is the second part of the
Fundamental Theorem of Calculus.
.
Theorem 40
The Fundamental Theorem of Calculus, Part 2
Let f be con nuous on [a, b] and let F be any an deriva ve of f. Then
f(x) dx = F(b) F(a).
. Example 124
Using the Fundamental Theorem of Calculus,
4 Part 2
We spent a great deal of me in the previous sec on studying 0 (4x x2 ) dx.
Using the Fundamental Theorem of Calculus, evaluate this denite integral.
We need an an deriva ve of f(x) = 4x x2 . All an derivaves of f have the form F(x) = 2x2 31 x3 + C; for simplicity, choose C = 0.
The Fundamental Theorem of Calculus states
S
(
)
1 ) (
64
(4x x2 ) dx = F(4) F(0) = 2(4)2 43 0 0 = 32
= 32/3.
3
3
This is the same answer we obtained using limits in the previous sec on, just
with much less work. .
Nota on: A special nota on is o en used in the process of evalua ng denite
integrals using the Fundamental Theorem of Calculus. Instead of explicitly writb

ing F(b) F(a), the nota on F(x) is used. Thus the solu on to Example 124
a
would be wri en as:
)4
(
4

(
)
1
1 ) (
(4x x2 ) dx = 2x2 x3 = 2(4)2 43 0 0 = 32/3.
3
3
0
0
Notes:
222

The Constant C: Any an deriva ve F(x) can be chosen when using the Fundamental Theorem of Calculus to evaluate a denite integral, meaning any value
of C can be picked. The constant always cancels out of the expression when
evalua ng F(b) F(a), so it does not ma er what value is picked. This being
the case, we might as well let C = 0.
. Example 125
Using the Fundamental Theorem of Calculus, Part 2
Evaluate the following denite integrals.
2

5
9
3
t
1.
x dx 2.
sin x dx
3.
e dt
4.
u du
5.
2
2 dx
1
2
(
) (
)
1 4
1 4
1
4
1.
x dx = x =
2
(2) = 0.
4 2
4
4
2

(
)

2.
sin x dx = cos x = cos cos 0 = 1 + 1 = 2.
(This is interes ng; it says that the area under one hump of a sine curve
is 2.)
5
5

3.
et dt = et = e5 e0 = e5 1 147.41.
0
4.
4
5.
1
u du =
9
) 2(
) 38
3
2 3
2( 3
u du = u 2 =
9 2 4 2 = 27 8 =
.
3
3
3
3
4
1
2
5

2 dx = 2x = 2(5) 2 = 2(5 1) = 8.
1
This integral is interes ng; the integrand is a constant func on, hence we
are nding the area of a rectangle with width (5 1) = 4 and height 2.
No ce how the evalua on of the denite integral led to 2(4) = 8.
b
In general, if c is a constant, then a c dx = c(b a).
Understanding Mo on with the Fundamental Theorem of Calculus

We established, star ng with Key Idea 1, that the deriva ve of a posi on
func on is a velocity func on, and the deriva ve of a velocity func on is an accelera on func on. Now consider denite integrals of velocity and accelera on
b
func ons. Specically, if v(t) is a velocity func on, what does
v(t) dt mean?
a
Notes:
223
The Fundamental Theorem of Calculus states that
b
v(t) dt = V(b) V(a),
a
where V(t) is any an deriva ve of v(t). Since v(t) is a velocity func on, V(t)
must be a posi on func on, and V(b) V(a) measures a change in posi on, or
displacement.
. Example 126
Finding displacement
A ball is thrown straight up with velocity given by v(t) = 32t + 20 /s, where
1
t is measured in seconds. Find, and interpret,
v(t) dt.
0
Using the Fundamental Theorem of Calculus, we have
v(t) dt =
0
(32t + 20) dt
0
1

= 16t2 + 20t
0
=4
Thus if a ball is thrown straight up into the air with velocity v(t) = 32t + 20,
the height of the ball, 1 second later, will be 4 feet above the ini al height. (Note
that the ball has traveled much farther. It has gone up to its peak and is falling
down, but the dierence between its height at t = 0 and t = 1 is 4 .) .
Integra ng an accelera on func on likewise gives a change in velocity. We
do not have a simple term for this analogous to displacement. If a(t) = 5miles/h2
and t is measured in hours, then
3
a(t) dt = 15
0
means the velocity has increased by 15m/h from t = 0 to t = 3.
The Fundamental Theorem of Calculus and the Chain Rule

x Part 1 of the Fundamental Theorem of Calculus (FTC) states that given F(x) =
)
d(
f(t) dt, F (x) = f(x). Using other nota on,
F(x) = f(x). While we have
dx
a
just prac ced evalua ng denite integrals, some mes nding an deriva ves is
impossible and we need to rely on other techniques to approximate the value
Notes:
224

x
of a denite integral. Func ons wri en as F(x) = a f(t) dt are useful in such
situa ons.
It may be of further use to compose such a func on with another. As an
example, we may compose F(x) with g(x) to get
(
)
F g(x) =
g(x)
f(t) dt.
a
What is the deriva ve of such a func on? The Chain Rule can be employed to
state
))
(
)
(
)
d( (
F g(x) = F g(x) g (x) = f g(x) g (x).
dx
An example will help us understand this.
. Example 127
The FTC, Part 1, and the Chain Rule

x2
Find the deriva ve of F(x) =
ln t dt.
2
x
We can view F(x) as being the func on G(x) =
ln t dt
2
(
)
composed with g(x) = x2 ; that is, F(x) = G g(x) . The Fundamental Theorem
of Calculus states that G (x) = ln x. The Chain Rule gives us
(
)
F (x) = G g(x) g (x)
S
= ln(g(x))g (x)
= ln(x2 )2x
= 2x ln x2
Normally, the steps dening G(x) and g(x) are skipped. .
Prac ce this once more.
. Example 128
The FTC, Part

1, and the Chain Rule
5
t3 dt.
Find the deriva ve of F(x) =

cos x
ve of F is straigh orward:
.
cos x
Note that F(x) =
t3 dt. Viewed this way, the deriva5
F (x) = sin x cos3 x.
Notes:
225
y
Area Between Curves

Consider con nuous func ons f(x) and g(x) dened on [a, b], where f(x)
g(x) for all x in [a, b], as demonstrated in Figure 5.24. What is the area of the
shaded region bounded by the two curves over [a, b]?
The area can be found by recognizing that this area is the area under f
the area under g. Using mathema cal nota on, the area is
f(x)
g(x)
x
a
.
a
g(x)
g(x) dx.
a
Proper es of the denite integral allow us to simplify this expression to
f(x)
f(x) dx
)
f(x) g(x) dx.
x
b
Theorem 41
.
Figure 5.24: Finding the area bounded by
two func ons on an interval; it is found
by subtrac ng the area under g from the
area under f.
Area Between Curves
Let f(x) and g(x) be con nuous func ons dened on [a, b] where f(x)
g(x) for all x in [a, b]. The area of the region bounded by the curves
y = f(x), y = g(x) and the lines x = a and x = b is
)
f(x) g(x) dx.
. Example 129
.Finding area between curves
Find the area of the region enclosed by y = x2 + x 5 and y = 3x 2.
y
It will help to sketch these two func ons, as done in Figure

S
5.25. The region whose area we seek is completely bounded by these two
func ons; they seem to intersect at x = 1 and x = 3. To check, set x2 +x5 =
3x 2 and solve for x:
y = x2 + x 5
15
10
5
x2 + x 5 = 3x 2
x
2
(x2 + x 5) (3x 2) = 0
x2 2x 3 = 0
(x 3)(x + 1) = 0
x = 1, 3
.y = 3x 2
Figure 5.25: Sketching the region enclosed by y = x2 + x 5 and y = 3x 2
in Example 129.
Notes:
226

Following Theorem 41, the area is
)
3x 2 (x2 + x 5) dx =
y
3
(x2 + 2x + 3) dx
(
)3

1 3
2
= x + x + 3x
3
1
(
)
1
1
+13
= (27) + 9 + 9
3
3
2
= 10 = 10.6
3
The Mean Value Theorem and Average Value

4 Consider the graph of a func on f in Figure 5.26 and the area dened by
f(x) dx. Three rectangles are drawn in Figure 5.27; in (a), the height of the
1
rectangle is greater than f on [1, 4], hence the area of this rectangle is is greater
4
than 0 f(x) dx.
In (b), the height of the rectangle is smaller than f on [1, 4], hence the area
4
of this rectangle is less than 1 f(x) dx.
Finally, in (c) the
4height of the rectangle is such that the area of the rectangle
is exactly that of 0 f(x) dx. Since rectangles that are too big, as in (a), and
4
rectangles that are too li le, as in (b), give areas greater/lesser than 1 f(x) dx,
it makes sense that there is a rectangle, whose top intersects f(x) somewhere
on [1, 4], whose area is exactly that of the denite integral.
We state this idea formally in a theorem.
Theorem 42
x
1
.
Figure 5.26: A graph of a func on f to introduce the Mean Value Theorem.
y
x
1
.
(a)
y
The Mean Value Theorem of Integra on
Let f be con nuous on [a, b]. There exists a value c in [a, b] such that
f(x) dx = f(c)(b a).
.
(b)
This is an existen al statement; c exists, but we do not provide a method

of nding it. Theorem 42 is directly connected to the Mean Value Theorem of
Dieren a on, given as Theorem 27; we leave it to the reader to see how.
We demonstrate the principles involved in this version of the Mean Value
Theorem in the following example.
x
1
Notes:
.
(c)
Figure 5.27: Dierently sized rectangles
4 give upper and lower bounds on
f(x) dx; the last rectangle matches the
1
area exactly.
227
. Example 130
Using the Mean Value Theorem
Consider 0 sin x dx. Find a value c guaranteed by the Mean Value Theorem.
We rst need to evaluate

done in Example 125.)
S
sin 0.69
sin x dx. (This was previously

sin x dx = cos x = 2.
0
x
c 1
Thus we seek a value c in [0, ] such that sin c = 2.
. 5.28: A graph of y = sin x on

Figure
[0, ] and the rectangle guaranteed by
the Mean Value Theorem.
sin c = 2 sin c = 2/ c = arcsin(2/) 0.69.
In Figure 5.28 sin x is sketched along with a rectangle with height sin(0.69).
The area of the rectangle is the same as the area under sin x on [0, ]. .
Let f be a func on on [a, b] with c such that f(c)(ba) =
)
b(
f(x) f(c) dx:
a
y
y = f(x)
f(c)
)
f(x) f(c) dx =
f(x)
b
a
f(x) dx. Consider
f(c) dx
a
= f(c)(b a) f(c)(b a)
= 0.
x
a
When f(x) is shi ed by f(c), the amount of area under f above the xaxis on
[a, b] is the same as the amount of area below the xaxis above f; see Figure
5.29 for an illustra on of this. In this sense, we can say that f(c) is the average
value of f on [a, b].
The value f(c) is the average value in another sense. First, recognize that the
Mean Value Theorem can be rewri en as
.
y
y = f(x) f(c)
f(c)
f(c) =
x
a
.
Figure 5.29: On top, a graph of y =
f(x) and the rectangle guaranteed by the
Mean Value Theorem. Below, y = f(x) is
shi ed down by f(c); the resul ng area
under the curve is 0.
f(x) dx,
a
for some value of c in [a, b]. Next, par on the interval [a, b] into n equally
spaced subintervals, a = x1 < x2 < . . . < xn+1 = b and choose any ci in
[xi , xi+1 ]. The average of the numbers f(c1 ), f(c2 ), , f(cn ) is:
n
) 1
1(
f(c1 ) + f(c2 ) + . . . + f(cn ) =
f(ci ).
n
n i=1
Notes:
228
1
ba

(ba)
(ba) :
Mul ply this last expression by 1 in the form of
1
1
f(ci ) =
f(ci )
n i=1
n
i=1
n
f(ci )
i=1
1 (b a)
n (b a)
1
ba
=
f(ci )
b a i=1
n
n
1
f(ci )x
b a i=1
n
Now take the limit as n :

1
lim
f(ci )x
n b a
i=1
n
1
ba
(where x = (b a)/n)
f(x) dx
f(c).
This tells us this: when we evaluate f at n (somewhat) equally spaced points in

[a, b], the average value of these samples is f(c) as n .
This leads us to a deni on.
Deni on 22
The Average Value of f on [a, b]
Let f be con nuous on [a, b]. The average value of f on [a, b] is f(c),
where c is a value in [a, b] guaranteed by the Mean Value Theorem. I.e.,
1
Average Value of f on [a, b] =
ba
f(x) dx.
a
An applica on of this deni on is given in the following example.

. Example 131
Finding the average value of a func on
An object moves back and forth along a straight line with a velocity given by
v(t) = (t 1)2 on [0, 3], where t is measured in seconds and v(t) is measured
in /s.
What is the average velocity of the object?
By our deni on, the average velocity is:
(
)3
3

)
1
1 3( 2
1 1 3
(t 1)2 dt =
t 2t + 1 dt =
t t2 + t = 1 /s.
3 . 0 0
3 0
3 3
0
S
Notes:
229
We can understand the above example through a simpler situa on. Suppose
you drove 100 miles in 2 hours. What was your average speed? The answer is
simple: displacement/ me = 100 miles/2 hours = 50 mph.
What was the displacement of the
3 object in Example 131? We calculate this
by integra ng its velocity func on: 0 (t 1)2 dt = 3 . Its nal posi on was 3
feet from its ini al posi on a er 3 seconds: its average velocity was 1 /s.
This sec on has laid the groundwork for a lot of great mathema cs to follow. The most important lesson is this: denite integrals can be evaluated using
an deriva ves. Since the previous sec on established that denite integrals are
the limit of Riemann sums, we can later create Riemann sums to approximate
values other than area under the curve, convert the sums to denite integrals,
then evaluate these using the Fundamental Theorem of Calculus. This will allow
us to compute the work done by a variable force, the volume of certain solids,
the arc length of curves, and more.
The downside is this: generally speaking, compu ng an deriva ves is much
more dicult than compu ng deriva ves. The next chapter is devoted to techniques of nding an deriva ves so that a wide variety of denite integrals can
be evaluated.
Notes:
230
Exercises 5.4
Terms and Concepts

2. What constant of integra on is most commonly used when
evalua ng denite integrals?
x
3. T/F: If f is a con nuous func on, then F(x) =
f(t) dt is
In Exercises 5 28, evaluate the denite integral.

3
5.
(3x2 2x + 1) dx
1
4
6.
0
1
23.
x3 dx
0
1
24.
x2 dx
x100 dx
0
4
25.
dx
4
5
26.
3 dx
10
2
27.
0 dx
2
/3
28.
(x 1)2 dx
0
1
7.
22.
4. The denite integral can be used to nd the area under a

curve. Give two other uses for denite integrals.
Problems
x dx
1. How are denite and indenite integrals related?
also a con nuous func on.
21.
(x3 x5 ) dx
csc x cot x dx
/6
29. Explain why:

1
(a)
xn dx = 0, when n is a posi ve, odd integer, and
8.
cos x dx
(b)
/4
9.
sec2 x dx
1
dx
x
10.
integer.
0
e
1
1
5 dx
1
1
12.
(4 2x3 ) dx
(2 cos x 2 sin x) dx
0
3
14.
9
8
t dt
1
dt
t
x dx
1
2
1
1
dx
x2
1
dx
x3
20.
1
1
dx
x
19.
0
2
x2 dx
31.
2
1
32.
ex dx
0
16
33.
x dx
In Exercises 34 39, nd the average value of the func on on

the given interval.
34. f(x) = sin x on [0, /2]
35. y = sin x on [0, ]
18.
25
17.
ex dx
16.
xn dx when n is a posi ve, even
15.
13.
In Exercises 30 33, nd a value c guaranteed by the Mean

Value Theorem.
2
30.
x2 dx
11.
xn dx = 2
/2
1
1
36. y = x on [0, 4]
37. y = x2 on [0, 4]
38. y = x3 on [0, 4]
39. g(t) = 1/t on [1, e]
In Exercises 40 44, a velocity func on of an object moving
along a straight line is given. Find the displacement of the
object over the given me interval.
40. v(t) = 32t + 20 /s on [0, 5]
41. v(t) = 32t + 200 /s on [0, 10]
42. v(t) = 2t mph on [1, 1]
50. y = x + 1, y = 3x + 6, x = 2 and x = 1.
43. v(t) = cos t /s on [0, 3/2]
44. v(t) = 4 t /s on [0, 16]
51. y = x2 2x + 5, y = 5x 5.
In Exercises 45 48, an accelera on func on of an object

moving along a straight line is given. Find the change of the
objects velocity over the given me interval.
45. a(t) = 32 /s2 on [0, 2]
52. y = 2x2 + 2x 5, y = x2 + 3x + 7.
In Exercises 53 56, nd F (x).
46. a(t) = 10 /s2 on [0, 5]

47. a(t) = t /s2 on [0, 2]
48. a(t) = cos t /s2 on [0, ]
In Exercises 49 52, sketch the given func ons and nd the
area of the enclosed region.
49. y = 2x, y = 5x, and x = 3.
x3 +x
53. F(x) =
2
0
54. F(x) =
1
dt
t
t3 dt
x3
x2
55. F(x) =
(t + 2) dt
x
ex
56. F(x) =
sin t dt
ln x
5.5 Numerical Integra on
5.5
Numerical Integra on
y
The Fundamental Theorem of Calculus gives a concrete technique for nding

the exact value of a denite integral. That technique is based on compu ng anderiva ves. Despite the power of this theorem, there are s ll situa ons where
we must approximate the value of the denite integral instead of nding its exact value. The rst situa on we explore is where we cannot compute the anderiva ve of the integrand. The second case is when we actually do not know
the integrand, but only its value when evaluated at certain points.
An elementary func on is any func on that is a combina on of polynomials, nth roots, ra onal, exponen al, logarithmic and trigonometric func ons. We
can compute the deriva ve of any elementary func on, but there are many elementary func ons that we cannot compute an an deriva ve of. For example,
the following func ons do not have an deriva ves that we can express with elementary func ons:
2
ex ,
sin(x3 )
and
0.5
x
1
0.5
y
1
sin x
.
x
y = sin(x3 )
0.5
2
The simplest way to refer to the an deriva ves of ex is to simply write

e
dx.
This sec on outlines three common methods of approxima ng the value of
denite integrals. We describe each as a systema c method of approxima ng
area under a curve. By approxima ng this area accurately, we nd an accurate
approxima on of the corresponding denite integral.
We will apply the methods we learn in this Sec on to the following denite
integrals:
y = ex
x2
ex dx,
sin(x3 ) dx,
and
0.5
sin(x)
dx,
x
x
1
1
0.5
..
y
1
y=
sin x
x
0.5
as pictured in Figure 5.30.
The Le and Right Hand Rule Methods
x
5
In Sec on 5.3 we addressed the problem of evalua ng denite integrals by

approxima ng the area under the curve using rectangles. We revisit those ideas
here before introducing other methods of approxima ng denite integrals.
We start with a review of nota on. Let f be a con nuous func on on the
b
interval [a, b]. We wish to approximate
f(x) dx. We par on [a, b] into n
a
equally spaced subintervals, each of length x =
10
15
.
Figure 5.30: Graphically represen ng
three denite integrals that cannot be
evaluated using an deriva ves.
ba
. The endpoints of these
n
Notes:
233
subintervals are labeled as
x1 = a, x2 = a + x, x3 = a + 2x, . . . , xi = a + (i 1)x, . . . , xn+1 = b.
n
Key Idea 8 states that to use the Le Hand Rule we use the summa on
n
f(xi )x and to use the Right Hand Rule we use

f(xi+1 )x. We review
i=1
the use of these rules in the context of examples.
i=1
y
y = ex
. Example 132
Approximate
.Approxima ng denite integrals with rectangles

x2
dx using the Le and Right Hand Rules with 5 equally
0.5
x
0.2
0.4
0.6
We begin by par oning the interval [0, 1] into 5 equally

S
spaced intervals. We have x = 10
5 = 1/5 = 0.2, so
0.8
x1 = 0, x2 = 0.2, x3 = 0.4, x4 = 0.6, x5 = 0.8, and x6 = 1.

Using the Le Hand Rule, we have:
y
y = ex
i=1
0.5
0.2
0.4
0.6
Figure 5.31: Approxima ng

Example 132.
0.8
1
0
(
)
f(xi )x = f(x1 ) + f(x2 ) + f(x3 ) + f(x4 ) + f(x5 ) x
(
)
= f(0) + f(0.2) + f(0.4) + f(0.6) + f(0.8) x
(
1 + 0.961 + 0.852 + 0.698 + 0.527)(0.2)
0.808.
Using the Right Hand Rule, we have:
ex dx in
i=1
(
)
f(xi+1 )x = f(x2 ) + f(x3 ) + f(x4 ) + f(x5 ) + f(x6 ) x
(
)
= f(0.2) + f(0.4) + f(0.6) + f(0.8) + f(1) x
(
0.961 + 0.852 + 0.698 + 0.527 + 0.368)(0.2)
0.681.
Figure 5.31 shows the rectangles used in each method to approximate the
denite integral. These graphs show that in this par cular case, the Le Hand
Rule is an over approxima on and the Right Hand Rule is an under approximaon. To get a be er approxima on, we could use more rectangles, as we did in
Notes:
234

Sec on 5.3. We could also average the Le and Right Hand Rule results together,
giving
0.808 + 0.681
= 0.7445.
2
The actual answer, accurate to 4 places a er the decimal, is 0.7468, showing
our average is a good approxima on. .
. Example 133
Approximate
Approxima ng denite integrals with rectangles

sin(x3 ) dx using the Le and Right Hand Rules with 10 equally
S
xi
x1
x2
x3
x4
x5
x6
x7
x8
x9
x10
x11
It is useful to write out the endpoints of the subintervals in a table; in Figure

5.32, we give the exact values of the endpoints, their decimal approxima ons,
and decimal approxima ons of sin(x3 ) evaluated at these points.
Once this table is created, it is straigh orward to approximate the denite
integral using the Le and Right Hand Rules. (Note: the table itself is easy to
create, especially with a standard spreadsheet program on a computer. The last
two columns are all that are needed.) The Le Hand Rule sums the rst 10 values
of sin(x3i ) and mul plies the sum by x; the Right Hand Rule sums the last 10
values of sin(x3i ) and mul plies by x. Therefore we have:
2
Le Hand Rule:
sin(x3 ) dx (1.91)(0.236) = 0.451.
Right Hand Rule:
sin(x3i )
0.466
0.165
0.031
0
0.004
0.061
0.246
0.601
0.971
0.690
0.670
/2 (/4)
3
ba
=
=
0.236.
n
10
40
Approx.
0.785
0.550
0.314
0.0785
0.157
0.393
0.628
0.864
1.10
1.34
1.57
Figure 5.32: Table of values used to ap

proximate 2 sin(x3 ) dx in Example 133.
We begin by nding x:
Exact
/4
7/40
/10
/40
/20
/8
/5
11/40
7/20
17/40
/2
y
1
y = sin(x3 )
0.5
x
1
1
0.5
sin(x3 ) dx (1.71)(0.236) = 0.404.

y
Average of the Le and Right Hand Rules: 0.4275.

The actual answer, accurate to 3 places a er the decimal, is 0.460. Our approxima ons were once again fairly good. The rectangles used in each approxima on are shown in Figure 5.33. It is clear from the graphs that using more
rectangles (and hence, narrower rectangles) should result in a more accurate
approxima on. .
0.5
x
1
The Trapezoidal Rule
In Example 132 we approximated the value of

0
y = sin(x3 )
1
0.5
ex dx with 5 rectangles
of equal width. Figure 5.31 shows the rectangles used in the Le and Right Hand
Figure
5.33:
Approxima ng
2
3
sin(x
)
dx
in
Example
133.
Notes:
235
y
y = ex
0.5
x
0.2
0.4
0.6
0.8
Rules. These graphs clearly show that rectangles do not match the shape of the
graph all that well, and that accurate approxima ons will only come by using
lots of rectangles.
Instead of using rectangles to approximate the area, we can instead use
2
trapezoids. In Figure 5.34, we show the region under f(x) = ex on [0, 1] approximated with 5 trapezoids of equal width; the top corners of each trapezoid lies on the graph of f(x). It is clear from this gure that these trapezoids
more accurately approximate
the area under f and hence should give a be er
1
2
approxima on of 0 ex dx. (In fact, these trapezoids seem to give a great approxima on of the area!)
The
for the area of a trapezoid is given in Figure 5.35. We approxi 1formula
2
mate 0 ex dx with these trapezoids in the following example.
. Example 134
Figure 5.34: Approxima ng 0 ex dx using 5 trapezoids of equal widths.
Approxima ng denite integrals

using trapezoids
Use 5 trapezoids of equal width to approximate
ex dx.
To compute the areas of the 5 trapezoids in Figure 5.34, it

S
will again be useful to create a table of values as shown in Figure 5.36.
The le most trapezoid has legs of length 1 and 0.961 and a height of 0.2.
Thus, by our formula, the area of the le most trapezoid is:
b
Area =
a+b
2 h
1 + 0.961
(0.2) = 0.1961.
2
.
h
Figure 5.35: The area of a trapezoid.
xi
0
0.2
0.4
0.6
0.8
1
Moving right, the next trapezoid has legs of length 0.961 and 0.852 and a height
of 0.2. Thus its area is:
0.961 + 0.852
(0.2) = 0.1813.
2
The sum of the areas of all 5 trapezoids is:
exi
1
0.961
0.852
0.698
0.527
0.368
2
Figure 5.36: A table of values of ex .
1 + 0.961
0.961 + 0.852
0.852 + 0.698
(0.2) +
(0.2) +
(0.2)+
2
2
2
0.698 + 0.527
0.527 + 0.368
(0.2) +
(0.2) = 0.7445.
2
2
1
2
We approximate
ex dx 0.7445. .
0
There are many things to observe in this example. Note how each term in
the nal summa on was mul plied by both 1/2 and by x = 0.2. We can factor
these coecients out, leaving a more concise summa on as:
[
]
1
(0.2) (1+0.961)+(0.961+0.852)+(0.852+0.698)+(0.698+0.527)+(0.527+0.368) .
2
Notes:
236

Now no ce that all numbers except for the rst and the last are added twice.
Therefore we can write the summa on even more concisely as
]
0.2 [
1 + 2(0.961 + 0.852 + 0.698 + 0.527) + 0.368 .
2
This is the heart of the Trapezoidal Rule, wherein a denite integral
f(x) dx
a
is approximated by using trapezoids of equal widths to approximate the corresponding area under f. Using n equally spaced subintervals with endpoints x1 ,
ba
x2 , . . ., xn+1 , we again have x =
. Thus:
n
b
n
f(xi ) + f(xi+1 )
x
f(x) dx
2
a
i=1
n
)
x (
=
f(xi ) + f(xi+1 )
2 i=1
x [
f(x1 ) + 2
f(xi ) + f(xn+1 ) .
2
i=2
n
. Example 135
Using the Trapezoidal Rule

2
Revisit Example 133 and approximate
sin(x3 ) dx using the Trapezoidal Rule
and 10 equally spaced subintervals.
We refer back to Figure 5.32 for the table of values of sin(x3 ).

Recall that x = 3/40 0.236. Thus we have:
S
sin(x3 ) dx
(
)
]
0.236 [
0.466 + 2 0.165 + (0.031) + . . . + 0.971 + 0.69 + (0.67)
2
= 0.4275.
No ce how quickly the Trapezoidal Rule can be implemented once the table of values is created. This is true for all the methods explored in this sec on;
the real work is crea ng a table of xi and f(xi ) values. Once this is completed, approxima ng the denite integral is not dicult. Again, using technology is wise.
Spreadsheets can make quick work of these computa ons and make using lots
of subintervals easy.
Also no ce the approxima ons the Trapezoidal Rule gives. It is the average
of the approxima ons given by the Le and Right Hand Rules! This eec vely
Notes:
237
renders the Le and Right Hand Rules obsolete. They are useful when rst learning about denite integrals, but if a real approxima on is needed, one is generally be er o using the Trapezoidal Rule instead of either the Le or Right Hand
Rule.
How can we improve on the Trapezoidal Rule, apart from using more and
more trapezoids? The answer is clear once we look back and consider what we
have really done so far. The Le Hand Rule is not really about using rectangles to
approximate area. Instead, it approximates a func on f with constant func ons
on small subintervals and then computes the denite integral of these constant
func ons. The Trapezoidal Rule is really approxima ng a func on f with a linear
func on on a small subinterval, then computes the denite integral of this linear
func on. In both of these cases the denite integrals are easy to compute in
geometric terms.
So we have a progression: we start by approxima ng f with a constant funcon and then with a linear func on. What is next? A quadra c func on. By
approxima ng the curve of a func on with lots of parabolas, we generally get
an even be er approxima on of the denite integral. We call this process Simpsons Rule, named a er Thomas Simpson (1710-1761), even though others had
used this rule as much as 100 years prior.
Simpsons Rule
3
x
1
Figure 5.37: A graph of a func on f and

a parabola that approximates it well on
[1, 3].
Given one point, we can create a constant func on that goes through that
point. Given two points, we can create a linear func on that goes through those
points. Given three points, we can create a quadra c func on that goes through
those three points (given that no two have the same xvalue).
Consider three points (x1 , y1 ), (x2 , y2 ) and (x3 , y3 ) whose xvalues are equally
spaced and x1 < x2 < x3 . Let f be the quadra c func on that goes through these
three points. It is not hard to show that
x3
)
x3 x1 (
y1 + 4y2 + y3 .
(5.4)
f(x) dx =
6
x1
Consider Figure 5.37. A func on f goes through the 3 points shown and the
parabola g that also goes through those points is graphed with a dashed line.
Using our equa on from above, we know exactly that
3
)
3 1(
g(x) dx =
3 + 4(1) + 2 = 3.
6
1
Since g is a good approxima on for f on [1, 3], we can state that
3
f(x) dx 3.
1
Notes:
238

No ce how the interval [1, 3] was split into two subintervals as we needed 3
points. Because of this, whenever we use Simpsons Rule, we need to break the
interval into an even number of subintervals.
xi
0
0.25
0.5
0.75
1
In general, to approximate
f(x) dx using Simpsons Rule, subdivide [a, b]

a
into n subintervals, where n is even and each subinterval has width x = (b

a)/n. We approximate f with n/2 parabolic curves, using Equa on (5.4) to compute the area under these parabolas. Adding up these areas gives the formula:
exi
1
0.939
0.779
0.570
0.368
(a)
]
x [
f(x1 )+4f(x2 )+2f(x3 )+4f(x4 )+. . .+2f(xn1 )+4f(xn )+f(xn+1 ) .
f(x) dx
3
y
y = ex
Note how the coecients of the terms in the summa on have the pa ern 1, 4,
2, 4, 2, 4, . . ., 2, 4, 1.
Lets demonstrate Simpsons Rule with a concrete example.
0.5
. Example 136
Using Simpsons Rule
Approximate
x2
dx using Simpsons Rule and 4 equally spaced subintervals.
0.25
We begin by making a table of values as we have in the past,

as shown in Figure 5.38(a). Simpsons Rule states that
S
ex dx
]
0.25 [
1 + 4(0.939) + 2(0.779) + 4(0.570) + 0.368 = 0.74683.
3
Recall in Example 132 we stated that the correct answer, accurate to 4 places
a er the decimal, was 0.7468. Our approxima on with Simpsons Rule, with 4
subintervals, is be er than our approxima on with the Trapezoidal Rule using
5!
2
Figure 5.38(b) shows f(x) = ex along with its approxima ng parabolas,
demonstra ng how good our approxima on is. The approxima ng curves are
nearly indis nguishable from the actual func on. .
. Example 137
Approximate
vals.
.Using Simpsons Rule

sin(x3 ) dx using Simpsons Rule and 10 equally spaced inter-
Figure 5.39 shows the table of values that we used in the

S
past for this problem, shown here again for convenience. Again, x = (/2 +
/4)/10 0.236.
0.5
0.75
(b)
Figure 5.38: A table of values to approxi1
2
mate 0 ex dx, along with a graph of the
func on.
xi
0.785
0.550
0.314
0.0785
0.157
0.393
0.628
0.864
1.10
1.34
1.57
sin(x3i )
0.466
0.165
0.031
0
0.004
0.061
0.246
0.601
0.971
0.690
0.670
Figure 5.39: Table of values used to ap

proximate 2 sin(x3 ) dx in Example 137.
4
Notes:
239
y
1
Simpsons Rule states that
y = sin(x )
0.5
= 0.4701
x
1
Recall that the actual value, accurate to 3 decimal places, is 0.460. Our approxima on is within one 1/100th of the correct value. The graph in Figure 5.40
shows how closely the parabolas match the shape of the graph. .
0.5
Figure
5.40:
Approxima ng
2
3
4 sin(x ) dx in Example 137 with
Simpsons Rule and 10 equally spaced
intervals.
0.236 [
(0.466) + 4(0.165) + 2(0.031) + . . .
3
]
. . . + 2(0.971) + 4(0.69) + (0.67)
sin(x3 ) dx
Key Idea 9
Summary and Error Analysis

We summarize the key concepts of this sec on thus far in the following Key
Idea.
Numerical Integra on
Let f be a con nuous func on on [a, b], let n be a posi ve integer, and let x =
Set x1 = a, x2 = a + x, . . ., xi = a + (i 1)x, xn+1 = b.
b
Consider
f(x) dx.
a
b
[
]
f(x) dx x f(x1 ) + f(x2 ) + . . . + f(xn ) .
Le Hand Rule:
a
Right Hand Rule:
ba
.
n
[
]
f(x) dx x f(x2 ) + f(x3 ) + . . . + f(xn+1 ) .
Trapezoidal Rule:
Simpsons Rule:
]
x [
f(x1 ) + 2f(x2 ) + 2f(x3 ) + . . . + 2f(xn ) + f(xn+1 ) .
2
a
b
]
x [
f(x1 ) + 4f(x2 ) + 2f(x3 ) + . . . + 4f(xn ) + f(xn+1 ) (n even).
f(x) dx
3
a
b
f(x) dx
In our examples, we approximated the value of a denite integral using a

given method then compared it to the right answer. This should have raised
several ques ons in the readers mind, such as:
1. How was the right answer computed?
2. If the right answer can be found, what is the point of approxima ng?
3. If there is value to approxima ng, how are we supposed to know if the
approxima on is any good?
Notes:
240

These are good ques ons, and their answers are educa onal. In the examples, the right answer was never computed. Rather, an approxima on accurate
to a certain number of places a er the decimal was given. In Example 132, we
do not know the exact answer, but we know it starts with 0.7468. These more
accurate approxima ons were computed using numerical integra on but with
more precision (i.e., more subintervals and the help of a computer).
Since the exact answer cannot be found, approxima on s ll has its place.
How are we to tell if the approxima on is any good?
Trial and error provides one way. Using technology, make an approximaon with, say, 10, 100, and 200 subintervals. This likely will not take much me
at all, and a trend should emerge. If a trend does not emerge, try using yet more
subintervals. Keep in mind that trial and error is never foolproof; you might
stumble upon a problem in which a trend will not emerge.
A second method is to use Error Analysis. While the details are beyond the
scope of this text, there are some formulas that give bounds for how good your
approxima on will be. For instance, the formula might state that the approxima on is within 0.1 of the correct answer. If the approxima on is 1.58, then
one knows that the correct answer is between 1.48 and 1.68. By using lots of
subintervals, one can get an approxima on as accurate as one likes. Theorem
43 states what these bounds are.
Theorem 43
Error Bounds in the Trapezoidal Rule and Simpsons Rule
1. Let ET be the error in approxima ng
f(x) dx using the Trapezoidal Rule.

If f has a con nuous 2nd deriva ve on [a, b] and M is any upper bound of f (x)
on [a, b], then
(b a)3
ET
M.
12n2
b
2. Let ES be the error in approxima ng
f(x) dx using Simpsons Rule.
a

If f has a con nuous 4th deriva ve on [a, b] and M is any upper bound of f (4)
on [a, b], then
(b a)5
ES
M.
180n4
a
There are some key things to note about this theorem.
Notes:
241
1. The larger the interval, the larger the error. This should make sense intui vely.
2. The error shrinks as more subintervals are used (i.e., as n gets larger).
3. The error in Simpsons Rule has a term rela ng to the 4th deriva ve of f.
Consider a cubic polynomial: its 4th deriva ve is 0. Therefore, the error in
approxima ng the denite integral of a cubic polynomial with Simpsons
Rule is 0 Simpsons Rule computes the exact answer!
x
1
0.5
We revisit Examples 134 and 136 and compute the error bounds using Theorem 43 in the following example.
. Example 138
1
x2
y=e
.Compu ng error bounds
Find the error bounds when approxima ng

2
(4x 2)
ex dx using the Trapezoidal
Rule and 5 subintervals, and using Simpsons Rule with 4 subintervals.
2
.
Figure 5.41: Graphing f (x) in Example

138 to help establish error bounds.
Trapezoidal Rule with n = 5:

2
We start by compu ng the 2nd deriva ve of f(x) = ex :
2
f (x) = ex (4x2 2).

Figure 5.41 shows a graph of f (x) on [0, 1]. It is clear that the largest value of
f , in absolute value, is 2. Thus we let M = 2 and apply the error formula from
Theorem 43.
y
ET =
10
y = ex (16x4 48x2 + 12)
(1 0)3
2 = 0.006.
12 52
Our error es ma on formula states that our approxima on of 0.7445 found

in Example 134 is within 0.0067 of the correct answer, hence we know that
x
0.5
.
Figure 5.42: Graphing f (4) (x) in Example
138 to help establish error bounds.
ex dx 0.7512 = 0.7445 + 0.0067.
0.7445 0.0067 = .7378
We had earlier computed the exact answer, correct to 4 decimal places, to be

0.7468, arming the validity of Theorem 43.
Simpsons Rule with n = 4:
2
We start by compu ng the 4th deriva ve of f(x) = ex :
2
f (4) (x) = ex (16x4 48x2 + 12).
Notes:
242

Figure 5.42 shows a graph of f (4) (x) on [0, 1]. It is clear that the largest value of
f (4) , in absolute value, is 12. Thus we let M = 12 and apply the error formula
from Theorem 43.
Es =
(1 0)5
12 = 0.00026.
180 44
Our error es ma on formula states that our approxima on of 0.74683 found

in Example 136 is within 0.00026 of the correct answer, hence we know that
0.74683 0.00026 = .74657
ex dx 0.74709 = 0.74683 + 0.00026.
Once again we arm the validity of Theorem 43. .

At the beginning of this sec on we men oned two main situa ons where
numerical integra on was desirable. We have considered the case where an
an deriva ve of the integrand cannot be computed. We now inves gate the
situa on where the integrand is not known. This is, in fact, the most widely
used applica on of Numerical Integra on methods. Most of the me we observe behavior but do not know the func on that describes it. We instead
collect data about the behavior and make approxima ons based o of this data.
We demonstrate this in an example.
. Example 139
.Approxima ng distance traveled
One of the authors drove his daughter home from school while she recorded
their speed every 30 seconds. The data is given in Figure 5.43. Approximate the
distance they traveled.
Recall that by integra ng a speed func on we get distance
S
traveled. We have informa on about v(t); we will use Simpsons Rule to approx b
imate
v(t) dt.
a
The most dicult aspect of this problem is conver ng the given data into the
form we need it to be in. The speed is measured in miles per hour, whereas the
me is measured in 30 second increments.
We need to compute x = (b a)/n. Clearly, n = 24. What are a and b?
Since we start at me t = 0, we have that a = 0. The nal recorded me came
a er 24 periods of 30 seconds, which is 12 minutes or 1/5 of an hour. Thus we
have
ba
1/5 0
1
x =
=
=
;
n
24
120
x
1
=
.
3
360
Time
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Speed
(mph)
0
25
22
19
39
0
43
59
54
51
43
35
40
43
30
0
0
28
40
42
40
39
40
23
0
Figure 5.43: Speed data collected at 30

second intervals for Example 139.
Notes:
243
Thus the distance traveled is approximately:
0.2
]
1 [
v(t) dt
f(x1 ) + 4f(x2 ) + 2f(x3 ) + + 4f(xn ) + f(xn+1 )
360
0
]
1 [
=
0 + 4 25 + 2 22 + + 2 40 + 4 23 + 0
360
6.2167 miles.
We approximate the author drove 6.2 miles. (Because we are sure the reader
wants to know, the authors odometer recorded the distance as about 6.05
miles.) .
Notes:
244
Exercises 5.5
Terms and Concepts
1. T/F: Simpsons Rule is a method of approxima ng anderiva ves.
2. What are the two basic situa ons where approxima ng the
value of a denite integral is necessary?
ln x dx
In Exercises 4 11, a denite integral is given.

(a) Approximate the denite integral with the Trapezoidal
Rule and n = 4.
(b) Approximate the denite integral with Simpsons Rule
and n = 4.
(c) Find the exact value of the integral.
1
4.
x2 dx
5x dx
sin x dx
0
7.
3
1
dx
x
( )
cos x2 dx
22.
0
5
23.
x4 dx
In Exercises 24 25, a region is given. Find the area of the

region using Simpsons Rule:
(a) where the measurements are in cen meters, taken in
1 cm increments, and
x dx
(b) where the measurements are in hundreds of yards,

taken in 100 yd increments.
0
1
x4 dx
24.
9 x2 dx
5.1
11.
4.7
cos x dx
0
6.6
10.
21.
(x3 + 2x2 5x + 7) dx
9.
8.
6.9
(b) Simpsons Rule

20.
sin x dx
4.5
(a) the Trapezoidal Rule
6.
1
dx
sin x + 2
6.3
1
dx
sin x + 2
In Exercises 20 23, nd n such that the error in approximating the given denite integral is less than 0.0001 when using:
10
19.
5.
3.6
18.
3. Why are the Le and Right Hand Rules rarely used?
Problems
17.
In Exercises 12 19, approximate the denite integral with

the Trapezoidal Rule and Simpsons Rule, with n = 6.
1
( )
12.
cos x2 dx
0
1
13.
1
x2 + 1 dx
x sin x dx
15.
0
/2
16.
0
cos x dx
25.
5.6
14.
6.6
ex dx
3.6
6: T
A
The previous chapter introduced the an deriva ve and connected it to signed
areas under a curve through the Fundamental Theorem of Calculus. The next
chapter explores more applica ons of denite integrals than just area. As evalua ng denite integrals will become important, we will want to nd an derivaves of a variety of func ons.
This chapter is devoted to exploring techniques of an dieren a on. While
not every func on has an an deriva ve in terms of elementary func ons (a
concept introduced in the sec on on Numerical Integra on), we can s ll nd
an deriva ves of many.
6.1
Subs tu on
We mo vate this sec on with an example. Let f(x) = (x2 + 3x 5)10 . We can
compute f (x) using the Chain Rule. It is:
f (x) = 10(x2 + 3x 5)9 (2x + 3) = (20x + 30)(x2 + 3x 5)9 .
Now consider this: What is (20x + 30)(x2 + 3x 5)9 dx? We have the answer
in front of us;
(20x + 30)(x2 + 3x 5)9 dx = (x2 + 3x 5)10 + C.

How would we have evaluated this indenite integral without star ng with f(x)
as we did?
This sec on explores integra on by subs tu on. It allows us to undo the
Chain Rule. Subs tu on allows us to evaluate the above integral without knowing the original func on rst.
The underlying principle is to rewrite a complicated integral of the form
f(x) dx as a notsocomplicated integral h(u) du. Well formally establish
later
how this is done. First, consider again our introductory indenite integral,
(20x + 30)(x2 + 3x 5)9 dx. Arguably the most complicated part of the
integrand is (x2 + 3x 5)9 . We wish to make this simpler; we do so through a
subs tu on. Let u = x2 + 3x 5. Thus
(x2 + 3x 5)9 = u9 .
Chapter 6 Techniques of An dieren a on

We have established u as a func on of x, so now consider the dieren al of u:
du = (2x + 3)dx.
Keep in mind that (2x+3) and dx are mul plied; the dx is not just si ng there.
Return to the original integral and do some subs tu ons through algebra:
(20x + 30)(x2 + 3x 5)9 dx = 10(2x + 3)(x2 + 3x 5)9 dx
= 10(x2 + 3x 5)9 (2x + 3) dx

| {z } | {z }
u
du
= 10u9 du
= u10 + C
(replace u with x + 3x 5)
= (x + 3x 5) + C
2
10
One might well look at this and think I (sort of) followed how that worked,
but I could never come up with that on my own, but the process is learnable.
This sec on contains numerous examples through which the reader will gain
understanding and mathema cal maturity enabling them to regard subs tu on
as a natural tool when evalua ng integrals.
We stated before that integra on by subs tu on undoes the Chain Rule.
Specically, let F(x) and g(x) be dieren able func ons and consider the derivave of their composi on:
))
d( (
F g(x) = F (g(x))g (x).
dx
Thus
F (g(x))g (x) dx = F(g(x)) + C.
Integra on by subs tu on works by recognizing the inside func on g(x) replacing it with a variable. By se ng u = g(x), we can rewrite the deriva ve
as
d ( ( ))
F u = F (u)u .
dx
Since du = g (x)dx, we can rewrite the above integral as
F (g(x))g (x) dx = F (u)du = F(u) + C = F(g(x)) + C.

This concept is important so we restate it in the context of a theorem.
Notes:
248
6.1 Subs tu on
Theorem 44
Integra on by Subs tu on
Let F and g be dieren able func ons, where the range of g is an interval
I and the domain of F is contained in I. Then
F (g(x))g (x) dx = F(g(x)) + C.
If u = g(x), then du = g (x)dx and
F (g(x))g (x) dx = F (u) du = F(u) + C = F(g(x)) + C.
The
point of subs tu on is to make the integra on step easy. Indeed, the
step F (u) du = F(u) + C looks easy, as the an deriva ve of the deriva ve of F
is just F, plus a constant. The work involved is making the proper subs tu on.
There is not a stepbystep process that one can memorize; rather, experience
will be ones guide. To gain experience, we now embark on many examples.
. Example 140
.Integra ng by subs tu on
2
Evaluate x sin(x + 5) dx.
Knowing that subs tu on is related to the Chain Rule, we
S
choose to let u be the inside func on of sin(x2 + 5). (This is not always a good
choice, but it is o en the best place to start.)
Let u = x2 + 5, hence du = 2x dx. The integrand has an x dx term, but
not a 2x dx term. (Recall that mul plica on is commuta ve, so the x does not
physically have to be next to dx for there to be an x dx term.) We can divide both
sides of the du expression by 2:
du = 2x dx
1
du = x dx.
2
We can now subs tute.
x sin(x2 + 5) dx =
sin(x2 + 5) |{z}
x dx
| {z }
1
2 du
1
sin u du
2
Notes:
249

1
= cos u + C (now replace u with x2 + 5)
2
1
= cos(x2 + 5) + C.
2
Thus x sin(x2 + 5) dx = 12 cos(x2 + 5) + C. We can check our work by evalua ng the deriva ve of the right hand side. .
. Example 141
Integra ng by subs tu on
Evaluate cos(5x) dx.

Again let u replace the inside func on. Le ng u = 5x, we
S
have du = 5dx. Since our integrand does not have a 5dx term, we can divide
the previous equa on by 5 to obtain 15 du = dx. We can now subs tute.
cos(5x) dx = cos(|{z}
5x ) |{z}
dx
1
5 du
1
cos u du
5
1
sin u + C
5
1
= sin(5x) + C.
5
=
We can again check our work through dieren a on. .

The previous example exhibited a common, and simple, type of subs tu on.
The inside func on was a linear func on (in this case, y = 5x). When the
inside func on is linear, the resul ng integra on is very predictable, outlined
here.
.
Key Idea 10
Subs tu on With A Linear Func on

Consider F (ax + b) dx, where a = 0 and b are constants. Le ng
u = ax + b gives du = a dx, leading to the result
1
F (ax + b) dx = F(ax + b) + C.
a
Thus sin(7x 4) dx = 17 cos(7x 4) + C. Our next example can use Key

Idea 10, but we will only employ it a er going through all of the steps.
Notes:
250
6.1 Subs tu on
. Example 142
Integra ng by subs tu ng a linear func on
7
Evaluate
dx.
3x + 1
View this a composi on of func ons f(g(x)), where f(x) =
S
7/x and g(x) = 3x + 1. Employing our understanding of subs tu on, we let
u = 3x + 1, the inside func on. Thus du = 3dx. The integrand lacks a 3;
hence divide the previous equa on by 3 to obtain du/3 = dx. We can now
evaluate the integral through subs tu on.
7
7 du
dx =
3x + 1
u 3
7
du
=
3
u
7
=
ln |u| + C
3
7
= ln | 3x + 1| + C.
3
Using Key Idea 10 is faster, recognizing that u is linear and a = 3. One may
want to con nue wri ng out all the steps un l they are comfortable with this
par cular shortcut. .
Not all integrals that benet from subs tu on have a clear inside func on.
Several of the following examples will demonstrate ways in which this occurs.
. Example 143
Evaluate sin x cos x dx.

There is not a composi on of func on here to exploit; rather,
S
just a product of func ons. Do not be afraid to experiment; when given an integral to evaluate, it is o en benecial to think If I let u be this, then du must be
that and see if this helps simplify the integral at all.
In this example, lets set u = sin x. Then du = cos x dx, which we have as
part of the integrand! The subs tu on becomes very straigh orward:
sin x cos x dx = u du
1 2
u +C
2
1
= sin2 x + C.
2
=
Notes:
251

One would do well to ask What would happen if we let u = cos x? The answer: the result is just as easy to nd, yet looks very dierent. The challenge to
the reader is to evaluate the integral le ng u = cos x and discovering why the
answer is the same, yet looks dierent. .
Our examples so far have required basic subs tu on. The next example
demonstrates how subs tu ons can be made that o en strike the new learner
as being nonstandard.
. Example 144
Integra ng by subs tu on
Evaluate x x + 3 dx.
Recognizing the composi on of func ons, set u = x + 3.
S
Then du = dx, giving what seems ini ally to be a simple subs tu on. But at this
stage, we have:
x x + 3 dx = x u du.
We cannot evaluate an integral that has both an x and an u in it. We need to
convert the x to an expression involving just u.
Since we set u = x+3, we can also state that u3 = x. Thus we can replace
x in the integrand with u 3. It will also be helpful, as before, to rewrite u as

1
u2 .
1
x x + 3 dx = (u 3)u 2 du
( 3
1)
=
u 2 3u 2 du
3
2 5
u 2 2u 2 + C
5
3
5
2
= (x + 3) 2 2(x + 3) 2 + C.
5
Checking your work is always a good idea. In this par cular case, some algebra
will be needed to make ones answer match the integrand in the original problem. .
. Example 145
1
Evaluate
dx.
x ln x
This is another example where there does not seem to be
S
an obvious composi on of func ons. The line of thinking used in Example 144
Notes:
252
6.1 Subs tu on
is useful here: choose something for u and consider what this implies du must
be. If u can be chosen such that du also appears in the integrand, then we have
chosen well.
Choosing u = 1/x makes du = 1/x2 dx; that does not seem helpful. However, se ng u = ln x makes du = 1/x dx, which is part of the integrand. Thus:
1
dx =
x ln x
1 1
dx
ln
x |{z}
x
|{z}
1/u
du
1
=
du
u
= ln |u| + C
= ln | ln x| + C.
The nal answer is interes ng; the natural log of the natural log. Take the derivave to conrm this answer is indeed correct. .
Integrals Involving Trigonometric Func ons

Sec on 6.3 delves deeper into integrals of a variety of trigonometric funcons; here we use subs tu on to establish a founda on that we will build upon.
The next three examples will help ll in some missing pieces of our an derivave knowledge. We know the an deriva ves of the sine and cosine func ons;
what about the other standard func ons tangent, cotangent, secant and cosecant? We discover these next.
. Example 146
.Integra on by subs tu on: an deriva ves of tan x
Evaluate tan x dx.
The previous paragraph established that we did not know

S
the an deriva ves of tangent, hence we must assume that we have learned
something in this sec on that can help us evaluate this indenite integral.
Rewrite tan x as sin x/ cos x. While the presence of a composi on of funcons may not be immediately obvious, recognize that cos x is inside the 1/x
func on. Therefore, we see if se ng u = cos x returns usable results. We have
Notes:
253

that du = sin x dx, hence du = sin x dx. We can integrate:
sin x
dx
cos x
1
=
sin
x dx}
| {z
cos
|{z}x
tan x dx =
du
1
du
u
= ln |u| + C
=
= ln | cos x| + C.
Some texts prefer to bring the 1 inside the logarithm as a power of cos x, as in:
ln | cos x| + C = ln |(cos x)1 | + C

1

+C
= ln
cos x
= ln | sec x| + C.
Thus the result they give is
equivalent. .
tan x dx = ln | sec x| + C. These two answers are
. Example 147
.Integra ng by subs tu on: an deriva ves of sec x
Evaluate sec x dx.

This example employs a wonderful trick: mul ply the inteS
grand by 1 so that we see how to integrate more clearly. In this case, we write
1 as
1=
sec x + tan x
.
sec x + tan x
This may seem like it came out of le eld, but it works beau fully. Consider:
sec x + tan x
dx
sec x + tan x
sec2 x + sec x tan x

=
dx.
sec x + tan x
sec x dx =
Notes:
254
sec x
6.1 Subs tu on
Now let u = sec x + tan x; this means du = (sec x tan x + sec2 x) dx, which is
our numerator. Thus:
du
=
u
= ln |u| + C
= ln | sec x + tan x| + C.
.
We can use similar techniques to those used in Examples 146 and 147 to nd
an deriva ves of cot x and csc x (which the reader can explore in the exercises.)
We summarize our results here.
Theorem 45
An deriva ves of Trigonometric Func ons
sin x dx = cos x + C
1.
cos x dx = sin x + C
2.
4.
.
5.
3.
csc x dx = ln | csc x + cot x| + C

sec x dx = ln | sec x + tan x| + C
tan x dx = ln | cos x|+C
6.
cot x dx = ln | sin x| + C
We explore one more common trigonometric integral.

. Example 148
.Integra on by subs tu on: powers of cos x and sin x
2
Evaluate cos x dx.
We have a composi on of func ons with cos x inside the x2
func on. However, se ng u = cos x means du = sin x dx, which we do not
have in the integral. Another technique is needed.
The process well employ is to use a Power Reducing formula for cos2 x (perhaps consult the back of this text). Note that
S
1 + cos(2x)
.
2
The right hand side of this equa on is not dicult to integrate. We have:
1 + cos(2x)
cos2 x dx =
dx
2
)
(
1 1
=
+ cos(2x) dx.
2 2
cos2 x =
Notes:
255

Now use Key Idea 10:
1
1 ( sin(2x))
x+
+C
2
2
2
sin(2x)
1
+ C.
= x
2
4
=
Well make signicant use of this powerreducing technique in future sec ons. .
Simplifying the Integrand

It is common to be reluctant to manipulate the integrand of an integral; at
rst, our grasp of integra on is tenuous and one may think that working with
the integrand will improperly change the results. Integra on by subs tu on
works using a dierent logic: as long as equality is maintained, the integrand can
be manipulated so that its form is easier to deal with. The next two examples
demonstrate common ways in which using algebra rst makes the integra on
easier to perform.
. Example 149
.Integra on by subs tu on: simplifying rst
3
x + 4x2 + 8x + 5
Evaluate
dx.
x2 + 2x + 1
One may start by se ng u equal to either the numerator or
S
denominator; in each instance, the result is not workable.
When dealing with ra onal func ons (i.e., quo ents made up of polynomial
func ons), it is an almost universal rule that everything works be er when the
degree of the numerator is less than the degree of the denominator. Hence we
use polynomial division.
We skip the specics of the steps, but note that when x2 + 2x + 1 is divided
into x3 + 4x2 + 8x + 5, it goes in x + 2 mes with a remainder of 3x + 3. Thus
3x + 3
x3 + 4x2 + 8x + 5
=x+2+ 2
.
x2 + 2x + 1
x + 2x + 1
Integra ng x + 2 is simple. The frac on can be integrated by se ng u = x2 +
2x + 1, giving du = (2x + 2) dx. This is very similar to the numerator. Note that
Notes:
256
6.1 Subs tu on
du/2 = (x + 1) dx and then consider the following:
)
3
(
x + 4x2 + 8x + 5
3x + 3
dx
=
x
+
2
+
dx
x2 + 2x + 1
x2 + 2x + 1
3(x + 1)
= (x + 2) dx +
dx
2
x + 2x + 1
1 2
3 du
= x + 2x + C1 +
2
u 2
=
1 2
3
x + 2x + C1 + ln |u| + C2
2
2
3
1 2
x + 2x + ln |x2 + 2x + 1| + C.
2
2
In some ways, we lucked out in that a er dividing, subs tu on was able to be
done. In later sec ons well develop techniques for handling ra onal func ons
where subs tu on is not directly feasible. .
=
. Example 150
.Integra on by alternate methods
2
x + 2x + 3
Evaluate
dx with, and without, subs tu on.
x
S
Rewrite
We already know how to integrate this par cular example.

1
x as x 2 and simplify the frac on:
1
1
3
x2 + 2x + 3
= x 2 + 2x 2 + 3x 2 .
1/2
x
We can now integrate using the Power Rule:
2
(
)
1
1
3
x + 2x + 3
2 + 2x 2 + 3x 2
dx
dx
=
x
x1/2
1
2 5
4 3
= x 2 + x 2 + 6x 2 + C
5
3
This is a perfectly ne approach. We demonstrate how this can also be solved

using subs tu
on as1 its implementa on is rather clever.
Let u = x = x 2 ; therefore
1 1
1
1
x 2 dx = dx 2du = dx.
2
2 x
x
2
x + 2x + 3
This gives us
dx = (x2 + 2x + 3) 2 du. What are we to do
x
1
with the other x terms? Since u = x 2 , u2 = x, etc. We can then replace x2 and
du =
Notes:
257

x with appropriate powers of u. We thus have
2
x + 2x + 3
dx = (x2 + 2x + 3) 2 du
x
= 2(u4 + 2u2 + 3) du
2 5 4 3
u + u + 6u + C
5
3
1
2 5
4 3
= x 2 + x 2 + 6x 2 + C,
5
3
=
which is obviously the same answer we obtained before. In this situa on, subs tu on is arguably more work than our other method. The fantas c thing is
that it works. It demonstrates how exible integra on is. .
Subs tu on and Inverse Trigonometric Func ons

When studying deriva ves of inverse func ons, we learned that
)
d(
1
tan1 x =
.
dx
1 + x2
Applying the Chain Rule to this is not dicult; for instance,
)
5
d(
tan1 5x =
.
dx
1 + 25x2
We now explore how Subs tu on can be used to undo certain deriva ves that
are the result of the Chain Rule and Inverse Trigonometric func ons. We begin
with an example.
. Example 151
.Integra ng by subs tu on: inverse trigonometric func ons
1
Evaluate
dx.
25 + x2
The integrand looks similar to the deriva ve of the arctanS
gent func on. Note:
1
1
=
x2
25 + x2
25(1 + 25
)
1
=
( )2
25(1 + 5x )
1
1
=
( ) .
25 1 + x 2
5
Notes:
258
6.1 Subs tu on
Thus
1
1
dx =
2
25 + x
25
1+
1
( x )2 dx.
5
This can be integrated using Subs tu on. Set u = x/5, hence du = dx/5 or
dx = 5du. Thus
1
1
1
dx =
( )2 dx
25 + x2
25
1 + 5x
1
1
=
du
5
1 + u2
1
= tan1 u + C
5
(x)
1
+C
= tan1
5
5
.
Example 151 demonstrates a general technique that can be applied to other
integrands that result in inverse trigonometric func ons. The results are summarized here.
Theorem 46
Integrals Involving Inverse Trigonomentric Func ons
Let a > 0.
(x)
1
1
1.
dx = tan1
+C
2
2
a +x
a
a
(x)
1
2.
dx = sin1
+C
a
a2 x2
( )
1
1
|x|
3.
dx = sec1
+C
a
a
x x2 a2
Lets prac ce using Theorem 46.

. Example 152
.Integra ng by subs tu on: inverse trigonometric func ons
Evaluate the given indenite integrals.
1
1
1
dx.
dx,
dx
and
9 + x2
1
5
x2
2
x x 100
Notes:
259

Each can be answered using a straigh orward applica on of
S
Theorem 46.
1
1
x
dx = tan1 + C.
9 + x2
3
3
dx = 10 sec1 10x + C.
1
2
x x 100
x
1
= sin1 + C.
5
5 x2
Most applica ons of Theorem 46 are not as straigh orward. The next examples show some common integrals that can s ll be approached with this theorem.
. Example 153
.Integra ng by subs tu on: comple ng the square

1
Evaluate 2
dx.
x 4x + 13
Ini ally, this integral seems to have nothing in common with
S
the integrals in Theorem 46. As it lacks a square root, it almost certainly is not
related to arcsine or arcsecant. It is, however, related to the arctangent func on.
We see this by comple ng the square on the denominator. We give a brief
reminder of the process here.
Start with a quadra c with a leading coecient of 1. It will have the form of
x2 +bx+c. Take 1/2 of b, square it, and add/subtract it back into the expression.
I.e.,
b2 b2
x2 + bx + c = x2 + bx +
+c
|
{z 4} 4
(x+b/2)2
(
)2
b
b2
= x+
+c
2
4
In our example, we take half of 4 and square it, ge ng 4. We add/subtract it
into the denominator as follows:
1
1
= 2
x2 4x + 13
x 4x + 4 4 + 13
| {z }
(x2)2
Notes:
260
1
(x 2)2 + 9
6.1 Subs tu on
We can now integrate this using the arctangent rule. Technically, we need to
subs tute rst with u = x 2, but we can employ Key Idea 10 instead. Thus we
have
1
1
1
x2
dx
=
dx = tan1
+ C.
x2 4x + 13
(x 2)2 + 9
3
3
.
. Example 154
Integrals require mul ple methods
4x
Evaluate
dx.
16 x2
This integral requires two dierent methods to evaluate it.
S
We get to those methods by spli ng up the integral:
4x
4
x
dx =
dx
dx.
16 x2
16 x2
16 x2
The rst integral is handled using a straigh orward applica on of Theorem 46;
the second integral is handled by subs tu on, with u = 16 x2 . We handle
each separately.
4
x
dx = 4 sin1 + C.
4
16 x2
dx: Set u = 16 x2 , so du = 2xdx and xdx = du/2. We

16 x2
have
x
du/2
dx =
2
u
16 x
1
1
du
=
2
u
= u+C
= 16 x2 + C.
Combining these together, we have
x
4x
dx = 4 sin1 + 16 x2 + C.
4
.
16 x2
Subs tu on and Denite Integra on

This sec on has focused on evalua ng indenite integrals as we are learning
a new technique for nding an deriva ves. However, much of the me integraon is used in the context of a denite integral. Denite integrals that require
subs tu on can be calculated using the following workow:
Notes:
261
1. Start with a denite integral
f(x) dx that requires subs tu on.

a
2. Ignore the bounds; use subs tu on to evaluate
f(x) dx and nd an an-
deriva ve F(x).
b

3. Evaluate F(x) at the bounds; that is, evaluate F(x) = F(b) F(a).
a
This workow works ne, but subs tu on oers an alterna ve that is powerful
and amazing (and a li le me saving).
At its heart, (using
the nota on of Theorem 44) subs tu on converts inte
grals of the form F (g(x))g (x) dx into an integral of the form F (u) du with
the subs tu on of u = g(x). The following theorem states how the bounds of
a denite integral can be changed as the subs tu on is performed.
.
Theorem 47
Subs tu on with Denite Integrals
Let f and g be dieren able func ons, where the range of g is an interval
I that contains the domain of F. Then
b
g(b)
(
)
F g(x) g (x) dx =
F (u) du.
g(a)
In eect, Theorem 47 states that once you convert to integra ng with respect to u, you do not need to switch back to integra ng with respect to x. A
few examples will help one understand.
. Example 155
.Denite integrals and subs tu on: changing the bounds
2
Evaluate
cos(3x 1) dx using Theorem 47.
0
Observing the composi on of func ons, let u = 3x 1,

S
hence du = 3dx. As 3dx does not appear in the integrand, divide the la er
equa on by 2 to get du/3 = dx.
By se ng u = 3x 1, we are implicitly sta ng that g(x) = 3x 1. Theorem
47 states that the new lower bound is g(0) = 1; the new upper bound is
Notes:
262
6.1 Subs tu on
g(2) = 5. We now evaluate the denite integral:
2
5
du
cos(3x 1) dx =
cos u
3
1
1

5
1

= sin u
3
1
)
1(
=
sin 5 sin(1)
3
0.039.
y
y = cos(3x 1)
1
0.5
x
1
0.5
No ce how once we converted the integral to be in terms of u, we never went

back to using x.
The graphs in Figure 6.1 tell more of the story. In (a) the area dened by the
original integrand is shaded, whereas in (b) the area dened by the new integrand is shaded. In this par cular situa on, the areas look very similar; the new
region is shorter but wider, giving the same area. .
(a)
y
1
y = cos(u)
0.5
u
1
. Example 156
Denite integrals and subs tu on: changing the bounds
/2
Evaluate
sin x cos x dx using Theorem 47.
0
u du
(b)
Figure 6.1: Graphing the areas dened by
the denite integrals of Example 155.
y
1
0.5
We saw the corresponding indenite integral in Example 143.

In that example we set u = sin x but stated that we could have let u = cos x.
For variety, we do the la er here.
Let u = g(x) = cos x, giving du = sin x dx. The new upper bound is
g(/2) = 0; the new lower bound is g(0) = 1. Note how the lower bound is
actually larger than the upper bound now. We have
/2
0
sin x cos x dx =
u (1)du
y = sin x cos x
0.5
(switch bounds & change sign)
1
1
u du
1 2 1
u
2 0
= 1/2.
=
In Figure 6.2 we have again graphed the two regions dened by our denite
integrals. Unlike the previous example, they bear no resemblance to each other.
However,
.
Theorem 47 guarantees that they have the same area.
. 0.5
(a)
y
y=u
0.5
u
1
Notes:
. 0.5
(b)
Figure 6.2: Graphing the areas dened by
the denite integrals of Example 156.
263
Exercises 6.1
Terms and Concepts
1. Subs tu on undoes what deriva ve rule?
tan2 (x)dx
21.
2. T/F: One can use algebra to rewrite the integrand of an integral to make it easier to evaluate.
In Exercises 22 28, use Subs tu on to evaluate the indenite integral involving exponen al func ons.
22.
e3x1 dx
Problems
23.
In Exercises 3 14, evaluate the indenite integral to develop

an understanding of Subs tu on.
(
)7
3.
3x2 x3 5 dx
4.
5.
6.
7.
8.
10.
11.
13.
14.
28.
1
dx
2x + 7
1
dx
2x + 3
x
dx
x+3
x3 x
dx
x
e x
dx
x
1
x
x4
dx
x5 + 1
+1
dx
x2
ln(x)
dx
x
cos(3 6x)dx
sec2 (4 x)dx
17.
18.
sec(2x)dx
tan2 (x) sec2 (x)dx
19.
20.
(
)5
(12x + 14) 3x2 + 7x 1 dx
In Exercises 15 21, use Subs tu on to evaluate the indenite integral involving trigonometric func ons.
15.
sin2 (x) cos(x)dx
16.
27.
( )
x cos x2 dx
ex
25.
(
)8
x x2 + 1 dx
12.
26.
ex x2 dx
24.
(
)3
(2x 5) x2 5x + 7 dx
9.
2x+1
(x 1)dx
ex + 1
dx
ex
ex ex
dx
e2x
33x dx
42x dx
In Exercises 29 32, use Subs tu on to evaluate the indenite integral involving logarithmic func ons.
ln x
29.
dx
x
ln2 (x)
30.
dx
x
( )
ln x3
31.
dx
x
1
32.
dx
x ln (x2 )
In Exercises 33 38, use Subs tu on to evaluate the indenite integral involving ra onal func ons.
2
x + 3x + 1
33.
dx
x
3
x + x2 + x + 1
34.
dx
x
3
x 1
dx
35.
x+1
2
x + 2x 5
36.
dx
x3
3x2 5x + 7
37.
dx
x+1
x2 + 2x + 1
38.
dx
x3 + 3x2 + 3x
In Exercises 39 48, use Subs tu on to evaluate the indenite integral involving inverse trigonometric func ons.
7
39.
dx
x2 + 7
40.
dx
9 x2
42.
43.
62.
dx
x x2 9
63.
45.
64.
x
dx
1 x4
65.
1
dx
x2 2x + 8
66.
47.
48.
x2
3
dx
+ 8x + 9
68.
51.
dx
1 x2
(
)
x2 csc2 x3 + 1 dx
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
sin(x) cos(x)dx
1
dx
x5
7
dx
3x + 2
3x3 + 4x2 + 2x 22
dx
x2 + 3x + 5
2x + 7
dx
x2 + 7x + 3
x3
dx
+9
x2
x3 x
dx
x2 + 4x + 9
sin(x)
dx
cos2 (x) + 1
69.
cos(x)
dx
sin2 (x) + 1
70.
cos(x)
dx
1 sin2 (x)
71.
72.
73.
3x 3
dx
2x 6
x2
x2
x3
dx
6x + 8

3
1
74.
dx
x
5
1
6
75.
x x 2dx
2
/2
76.
78.
x3 + 14x2 46x 7
dx
x2 7x + 1
79.
x
dx
x4 + 81
80.
sin2 x cos x dx
/2
1
77.
9(2x + 3)
dx
3x2 + 9x + 7
2
dx
4x2 + 1
x2 + 5x 2
dx
10x + 32
x2
67.
In Exercises 49 73, evaluate the indenite integral.
x2
dx
49.
(x3 + 3)2
( 2
)(
)8
3x + 2x 5x3 + 5x2 + 2 dx
50.
7 2x
dx
x2 + 12x + 61
2
dx
+ 6x + 7
1
dx
16 9x2
3x 2
dx
x2 2x + 10
x2
5
dx
x2 + 6x + 34
5
dx
x4 16x2
46.
dx
x 4x2 1
44.
14
dx
5 x2
41.
2x(1 x2 )4 dx
0
1
(x + 1)ex
+2x+1
2
1
1
4
2
1
dx
1 + x2
1
dx
x2 6x + 10
81.
1
1
dx
4 x2
dx
6.2 Integra on by Parts

Heres a simple integral that we cant yet do:
x cos x dx.
Its a simple ma er to take the deriva ve of the integrand using the Product
Rule, but there is no Product Rule for integrals. However, this sec on introduces
Integra on by Parts, a method of integra on that is based on the Product Rule
for deriva ves. It will enable us to evaluate this integral.
The Product Rule says that if u and v are func ons of x, then (uv) = u v+uv .
For simplicity, weve wri en u for u(x) and v for v(x). Suppose we integrate both
sides with respect to x. This gives
(uv) dx = (u v + uv ) dx.
By the Fundamental Theorem of Calculus, the le side integrates to uv. The right
side can be broken up into two integrals, and we have
uv = u v dx + uv dx.
Solving for the second integral we have
uv dx = uv u v dx.
Using dieren al nota on, we can write du = u (x)dx and dv = v (x)dx and the
expression above can be wri en as follows:
u dv = uv v du.
This is the Integra on by Parts formula. For reference purposes, we state this in
a theorem.
.
Theorem 48
Integra on by Parts
Let u and v be dieren able func ons of x on an interval I containing a

and b. Then
u dv = uv v du,
and
x=b
x=a
Notes:
266
b

u dv = uv
a
x=b
x=a
v du.

Lets try an example to understand our new technique.
. Example 157
Integra ng using Integra on by Parts
Evaluate x cos x dx.

The key to Integra on by Parts is to iden fy part of the inS
tegrand as u and part as dv. Regular prac ce will help one make good idenca ons, and later we will introduce some principles that help. For now, let
u = x and dv = cos x dx.
It is generally useful to make a small table of these values as done below.
Right now we only know u and dv as shown on the le of Figure 6.3; on the right
we ll in the rest of what we need. If u = x, then du = dx. Since dv = cos x dx,
v is an an deriva ve of cos x. We choose v = sin x.
u=x
du = ?
v=?
dv = cos x dx
u=x
du = dx
v = sin x
dv = cos x dx
Figure 6.3: Se ng up Integra on by Parts.
Now subs tute all of this into the Integra on by Parts formula, giving
x cos x = x sin x sin x dx.

We can then integrate sin x to get cos x + C and overall our answer is
x cos x dx = x sin x + cos x + C.

Note how the an deriva ve contains a product, x sin x. This product is what
makes Integra on by Parts necessary. .
The example above demonstrates how Integra on by Parts works in general.
We try to iden fy u and dv in the integral we are given, and the key is that we
usually want to choose u and dv so that du is simpler than u and v is hopefully
not too much more complicated than dv. This will mean that the integral on the
right side of the Integra on
by Parts formula, v du will be simpler to integrate
than the original integral u dv.
In the example above, we chose u = x and dv = cos x dx. Then du = dx was
simpler than u and v = sin x is no more complicated than dv. Therefore, instead
of integra ng x cos x dx, we could integrate sin x dx, which we know how to do.
A useful mnemonic for helping to determine u is LIATE, where
L = Logarithmic, I = Inverse Trig., A = Algebraic (polynomials),
T = Trigonometric, and E = Exponen al.
Notes:
267

If the integrand contains both a logarithmic and an algebraic term, in general
le ng u be the logarithmic term works best, as indicated by L coming before A
in LIATE.
We now consider another example.
. Example 158
Integra ng using Integra on by Parts
x
Evaluate xe dx.
The integrand contains an algebraic term (x) and an expoS
nen al term (ex ). Our mnemonic suggests le ng u be the algebraic term, so we
choose u = x and dv = ex dx. Then du = dx and v = ex as indicated by the
tables below.
u=x
du = ?
v=?
dv = ex dx
u=x
du = dx
v = ex
dv = ex dx
We see du is simpler than u, while there is no change in going from dv to v.

This is good. The Integra on by Parts formula gives
xex dx = xex ex dx.

The integral on the right is simple; our nal answer is
xex dx = xex ex + C.
Note again how the an deriva ves contain a product term. .
. Example 159
.Integra ng using Integra on by Parts
Evaluate x2 cos x dx.

The mnemonic suggests le ng u = x2 instead of the trigonometric func on, hence dv = cos x dx. Then du = 2x dx and v = sin x as shown
below.
S
u = x2
du = ?
v=?
dv = cos x dx
u = x2
du = 2x dx
Notes:
268
v = sin x
dv = cos x dx

The Integra on by Parts formula gives
x2 cos x dx = x2 sin x 2x sin x dx.

At this point, the integral on the right is indeed simpler than the one we started
with, but to evaluate it, we need to do Integra on by Parts again. Here we
choose u = 2x and dv = sin x and ll in the rest below.
u = 2x
du = ?
v=?
dv = sin x dx
u = 2x
du = 2 dx
v = cos x
dv = sin x dx
Figure 6.6: Se ng up Integra on by Parts (again).
(
)
x2 cos x dx = x2 sin x 2x cos x 2 cos x dx .
The integral all the way on the right is now something we can evaluate. It evaluates to 2 sin x. Then going through and simplifying, being careful to keep all
the signs straight, our answer is
x2 cos x dx = x2 sin x + 2x cos x 2 sin x + C.

.
. Example 160
.Integra ng using Integra on by Parts
x
Evaluate e cos x dx.
This is a classic problem. Our mnemonic suggests le ng u
S
be the trigonometric func on instead of the exponen al. In this par cular example, one can let u be either cos x or ex ; to demonstrate that we do not have
to follow LIATE, we choose u = ex and hence dv = cos x dx. Then du = ex dx
and v = sin x as shown below.
u = ex
du = ?
v=?
dv = cos x dx
u = ex
du = ex dx
v = sin x
dv = cos x dx
No ce that du is no simpler than u, going against our general rule (but bear
with us). The Integra on by Parts formula yields
x
x
e cos x dx = e sin x ex sin x dx.
Notes:
269

The integral on the right is not much dierent than the one we started with, so
it seems like we have go en nowhere. Lets s ck keep working and apply Integra on by Parts to the new integral, using u = ex and dv = sin x dx. This leads
us to the following:
u = ex
du = ?
v=?
dv = sin x dx
u = ex
v = cos x
dv = sin x dx
du = e dx
Figure 6.8: Se ng up Integra on by Parts (again).
The Integra on by Parts formula then gives:

(
)
x
x
x
x
e cos x dx = e sin x e cos x e cos x dx
x
x
= e sin x + e cos x ex cos x dx.
It
we are back right where we started, as the right hand side contains
seems
ex cos xdx. But this actually a good thing.
Add ex cos x dx to both sides. This gives
ex cos x dx = ex sin x + ex cos x
Now divide both sides by 2:
)
1(
ex cos x dx = ex sin x + ex cos x .
2
Simplifying a li le and adding the constant of integra on, our answer is thus
1
ex cos x dx = ex (sin x + cos x) + C.
.
2
. Example 161
.Integra ng using Integra on by Parts: an deriva ve of ln x
Evaluate ln x dx.
One may have no ced that we have rules for integra ng the
S
familiar trigonometric func ons and ex , but we have not yet given a rule for integra ng ln x. That is because ln x cant easily be done with any of the rules we
have learned up to this point. But it can be done by a clever applica on of Integra on by Parts. Set u = ln x and dv = dx. This is a good, sneaky trick to learn
Notes:
270

as it can help in other situa ons. This determines du = (1/x) dx and v = x as
shown below.
u = ln x
du = ?
v=?
dv = dx
u = ln x
du = 1/x dx
v=x
dv = dx
Pu ng this all together in the Integra on by Parts formula, things work out
very nicely:
1
ln x dx = x ln x x dx.
x
The new integral simplies to 1 dx, which is about as simple as things get. Its
integral is x + C and our answer is
ln x dx = x ln x x + C.
.
. Example 162
Integra ng using Int. by Parts: an deriva ve of arctan x
Evaluate arctan x dx.

The same sneaky trick we used above works here. Let u =
S
arctan x and dv = dx. Then du = 1/(1 + x2 ) dx and v = x. The Integra on by
Parts formula gives
x
arctan x dx = x arctan x
dx.
1 + x2
The integral on the right can be done by subs tu on. Taking u = 1 + x2 , we get
du = 2x dx. The integral then becomes
1
1
du.
arctan x dx = x arctan x
2
u
The integral on the right evaluates to ln |u| + C, which becomes ln(1 + x2 ) + C.
Therefore, the answer is
arctan x dx = x arctan x ln(1 + x2 ) + C.

.
Notes:
271
Subs tu on Before Integra on

When taking deriva ves, it was common to employ mul ple rules (such as,
using both the Quo ent and the Chain Rules). It should then come as no surprise
that some integrals are best evaluated by combining integra on techniques. In
par cular, here we illustrate making an unusual subs tu on rst before using
Integra on by Parts.
. Example 163
Integra on by Parts a er subs tu on
Evaluate cos(ln x) dx.

The integrand contains a composi on of func ons, leading
S
us to think Subs tu on would be benecial. Le ng u = ln x, we have du =
1/x dx. This seems problema c, as we do not have a 1/x in the integrand. But
consider:
1
du = dx x du = dx.
x
Since u = ln x, we can use inverse func ons and conclude that x = eu . Therefore
we have that
dx = x du
= eu du.
We can thus replace ln x with u and dx with eu du. Thus we rewrite our integral
as
cos(ln x) dx = eu cos u du.

We evaluated this integral in Example 160. Using the result there, we have:
cos(ln x) dx = eu cos u du
)
1 u(
e sin u + cos u + C
2
)
1 ln x (
= e
sin(ln x) + cos(ln x) + C
2
)
1 (
= x sin(ln x) + cos(ln x) + C
2
=
Denite Integrals and Integra on By Parts

So far we have focused only on evalua ng indenite integrals. Of course, we
can use Integra on by Parts to evaluate denite integrals as well, as Theorem
Notes:
272

48 states. We do so in the next example.
. Example 164
Denite integra on using Integra on by Parts
2
Evaluate
x2 ln x dx.
1
Once again, our mnemonic suggests we let u = ln x. (We

S
could let u = x2 and dv = ln x dx, as we now know the an deriva ves of ln x.
However, le ng u = ln x makes our next integral much simpler as it removes
the logarithm from the integral en rely.)
So we have u = ln x and dv = x2 dx. We then get du = (1/x) dx and
v = x3 /3 as shown below.
u = ln x
du = ?
v=?
dv = x2 dx
u = ln x
du = 1/x dx
v = x3 /3
dv = x2 dx
The Integra on by Parts formula then gives

2 2 3
2

x3
x 1
2
x ln x dx =
ln x
dx
3
1
1 3 x
1
2 2 2

x3
x
=
ln x
dx
3
1 3
1
2
2

x3
x3

=
ln x
3
9 1
1
) 2
( 3
x3
x
ln x
=
3
9 1
(
) (
)
8
8
1
1
=
ln 2
ln 1
3
9
3
9
7
8
= ln 2
3
9
1.07.
.
In general, Integra on by Parts is useful for integra ng certain products of
func ons, like xex dx or x3 sin x dx. It is also useful for integrals involving
logarithms and inverse trigonometric func ons.
As stated before, integra on is generally more dicult than deriva on. We
are developing tools for handling a large array of integrals, and experience will
Notes:
273

tell us when one tool is preferable/necessary over another. For instance, consider the three similarlooking integrals
3
x
x2
xe dx,
xe dx
and
xex dx.
While the rst is calculated easily with Integra on by Parts, the second is best
approached with Subs tu on. Taking things one step further, the third integral
has no answer in terms of elementary func ons, so none of the methods we
learn in calculus will get us the exact answer.
Regardless of these issues, Integra on by Parts is a very useful method, second only to subs tu on.
Notes:
274
Exercises 6.2
Terms and Concepts
25.
1. T/F: Integra on by Parts is useful in evalua ng integrands

that contain products of func ons.
26.
2. T/F: Integra on by Parts can be thought of as the opposite

of the Chain Rule.
27.
3. For what is LIATE useful?
28.
Problems
29.
4.
x sin x dx
5.
xex dx
6.
x2 sin x dx
x3 sin x dx
7.
2
8.
xex dx
9.
x3 ex dx
xe2x dx
10.
11.
ex sin x dx
12.
e2x cos x dx
13.
e2x sin(3x) dx
14.
e5x cos(5x) dx
sin x cos x dx
15.
sin1 x dx
16.
17.
tan1 (2x) dx
18.
x tan1 x dx
19.
sin1 x dx
20.
x ln x dx
21.
(x 2) ln x dx
22.
x ln(x 1) dx
23.
x ln(x2 ) dx
24.
x2 ln x dx
(ln x)2 dx
(ln(x + 1))2 dx
x sec2 x dx
x csc2 x dx
x x 2 dx
x x2 2 dx
30.
sec x tan x dx
31.
32.
x sec x tan x dx
33.
x csc x cot x dx
In Exercises 34 38, evaluate the indenite integral a er rst

making a subs tu on.
34.
sin(ln x) dx
35.
sin( x) dx
36.
ln( x) dx

37.
e x dx
38.
eln x dx
In Exercises 39 47, evaluate the denite integral. Note: the
corresponding indenite integrals appear in Exercises 4 12.

x sin x dx
39.
0
1
40.
xex dx
1
/4
41.
x2 sin x dx
/4
/2
42.
x3 sin x dx
/2
ln 2
43.
44.
2
45.
46.
xex dx
x3 ex dx
xe2x dx
ex sin x dx
0
/2
47.
/2
e2x cos x dx
6.3 Trigonometric Integrals

Func ons involving trigonometric func ons are useful as they are good at describing periodic behavior. This sec on describes several techniques for nding
an deriva ves of certain combina ons of trigonometric func ons.
Integrals of the form
sinm x cosn x dx
In learning the technique of Subs tu on, we saw the integral sin x cos x dx
in Example 143. The integra on was not dicult, and one could easily evaluate
the indenite integral by le ng u = sin x or by le ng u = cos x. This integral is
easy since the power of both sine and cosine is 1.
We generalize this integral and consider integrals of the form sinm x cosn x dx,
where m, n are nonnega ve integers. Our strategy for evalua ng these integrals is to use the iden ty cos2 x + sin2 x = 1 to convert high powers of one
trigonometric func on into the other, leaving a single sine or cosine term in the
integrand. We summarize the general technique in the following Key Idea.
.
Key Idea 11
Integrals Involving Powers of Sine and Cosine
Consider sinm x cosn x dx, where m, n are nonnega ve integers.

1. If m is odd, then m = 2k + 1 for some integer k. Rewrite
sinm x = sin2k+1 x = sin2k x sin x = (sin2 x)k sin x = (1 cos2 x)k sin x.
Then
sinm x cosn x dx =
(1 cos2 x)k sin x cosn x dx =
(1 u2 )k un du,
where u = cos x and du = sin x dx.
2. If n is odd, then using subs tu ons similar to that outlined above we have
sinm x cosn x dx =
um (1 u2 )k du,
where u = sin x and du = cos x dx.

3. If both m and n are even, use the powerreducing iden
cos2 x =
1 + cos(2x)
2
and
sin2 x =
es
1 cos(2x)
2
to reduce the degree of the integrand. Expand the result and apply the principles
of this Key Idea again.
Notes:
276

We prac ce applying Key Idea 11 in the next examples.
. Example 165
Integra ng powers of sine and cosine
5
Evaluate sin x cos8 x dx.
S
The power of the sine term is odd, so we rewrite sin5 x as
sin5 x = sin4 x sin x = (sin2 x)2 sin x = (1 cos2 x)2 sin x.
Our integral is now (1 cos2 x)2 cos8 x sin x dx. Let u = cos x, hence du =
sin x dx. Making the subs tu on and expanding the integrand gives
)
)
(
( 8
2 2
8
2 2 8
2
4 8
(1cos ) cos x sin x dx = (1u ) u du =
12u +u u du =
u 2u10 +u12 du.
This nal integral is not dicult to evaluate, giving
( 8
)
1
2
1
u 2u10 + u12 du = u9 + u11 u13 + C

9
11
13
1
2
1
= cos9 x +
cos11 x
cos13 x + C.
9
11
13
.
. Example 166
.Integra ng powers of sine and cosine
5
Evaluate sin x cos9 x dx.
The powers of both the sine and cosine terms are odd, thereS
fore we can apply the techniques of Key Idea 11 to either power. We choose to
work with the power of the cosine term since the previous example used the
sine terms power.
We rewrite cos9 x as
cos9 x = cos8 x cos x
= (cos2 x)4 cos x
= (1 sin2 x)4 cos x
= (1 4 sin2 x + 6 sin4 x 4 sin6 x + sin8 x) cos x.
We rewrite the integral as
(
)
5
9
sin x cos x dx = sin5 x 1 4 sin2 x + 6 sin4 x 4 sin6 x + sin8 x cos x dx.
Notes:
277
Now subs tute and integrate, using u = sin x and du = cos x dx.
(
)
sin5 x 1 4 sin2 x + 6 sin4 x 4 sin6 x + sin8 x cos x dx =
u5 (1 4u2 + 6u4 4u6 + u8 ) du =
( 5
)
u 4u7 + 6u9 4u11 + u13 du
1 6 1 8 3 10 1 12
1 14
u u + u u +
u +C
6
2
5
3
14
1
1
3
1
1
= sin6 x sin8 x + sin10 x sin12 x +
sin14 x + C
6
2
5
3
14
Technology Note: The work we are doing here can be a bit tedious, but the
skills it develops (problem solving, algebraic manipula on, etc.) are important.
Nowadays problems of this sort are o en solved usinga computer algebra system. The powerful program Mathema ca integrates sin5 x cos9 x dx as
y
g(x)
0.004
0.002
f(x) =
x
1
0.002
which clearly has a dierent form than our answer in Example 166, which is
f(x)
g(x) =
.
Figure 6.11: A plot of f(x) and g(x) from
Example 166 and the Technology Note.
45 cos(2x) 5 cos(4x) 19 cos(6x) cos(8x) cos(10x) cos(12x) cos(14x)
+
+
,
16384
8192
49152
4096
81920
24576
114688
1 6
1
3
1
1
sin x sin8 x + sin10 x sin12 x +
sin14 x.
6
2
5
3
14
Figure 6.11 shows a graph of f and g; they are clearly not equal. We leave it to
the reader to recognize why both answers are correct.
. Example 167
.Integra ng powers of sine and cosine
4
Evaluate cos x sin2 x dx.
The powers of sine and cosine are both even, so we employ
S
the powerreducing formulas and algebra as follows.
)2 (
)
1 + cos(2x)
1 cos(2x)
cos x sin x dx =
dx
2
2
1 + 2 cos(2x) + cos2 (2x) 1 cos(2x)

=
dx
4
2
)
1(
=
1 + cos(2x) cos2 (2x) cos3 (2x) dx
8
4
The cos(2x) term is easy to integrate, especially with Key Idea 10. The cos2 (2x)
term is another trigonometric integral with an even power, requiring the power
reducing formula again. The cos3 (2x) term is a cosine func on with an odd
power, requiring a subs tu on as done before. We integrate each in turn below.
Notes:
278
cos(2x) dx =
cos (2x) dx =
1
sin(2x) + C.
2
)
1(
1
1 + cos(4x)
dx = x + sin(4x) + C.
2
2
4
Finally, we rewrite cos3 (2x) as

(
)
cos3 (2x) = cos2 (2x) cos(2x) = 1 sin2 (2x) cos(2x).
Le ng u = sin(2x), we have du = 2 cos(2x) dx, hence
(
)
cos3 (2x) dx =
1 sin2 (2x) cos(2x) dx
1
=
(1 u2 ) du
2
1 )
1(
u u3 + C
=
2
3
)
1(
1
=
sin(2x) sin3 (2x) + C
2
3
Pu ng all the pieces together, we have
)
1(
cos4 x sin2 x dx =
1 + cos(2x) cos2 (2x) cos3 (2x) dx
8
)]
) 1(
1[
1
1(
1
1
= x + sin(2x) x + sin(4x)
sin(2x) sin3 (2x) + C
8
2
2
4
2
3
]
1[1
1
1 3
=
x sin(4x) + sin (2x) + C
8 2
8
6
.
The process above was a bit long and tedious, but being able to work a problem such as this from start to nish is important.
Integrals of the form sin(mx) sin(nx) dx,
and sin(mx) cos(nx) dx.
cos(mx) cos(nx) dx,
Func ons that contain products of sines and cosines of diering periods are
important in many applica ons including the analysis of sound waves. Integrals
of the form
sin(mx) sin(nx) dx,

cos(mx) cos(nx) dx and
sin(mx) cos(nx) dx
Notes:
279

are best approached by rst applying the Product to Sum Formulas found in the
back cover of this text, namely
(
)
(
)]
1[
cos (m n)x cos (m + n)x
2
(
)
(
)]
1[
cos(mx) cos(nx) =
cos (m n)x + cos (m + n)x
2
)
(
)]
1[ (
sin(mx) cos(nx) =
sin (m n)x + sin (m + n)x
2
sin(mx) sin(nx) =
. Example 168
Integra ng products of sin(mx) and cos(nx)
Evaluate sin(5x) cos(2x) dx.

The applica on of the formula and subsequent integra on
S
are straigh orward:
[
]
1
sin(3x) + sin(7x) dx
sin(5x) cos(2x) dx =
2
1
1
= cos(3x)
cos(7x) + C
6
14
.
Integrals of the form
tanm x secn x dx.
When evalua ng integrals of the form sinm x cosn x dx, the Pythagorean
Theorem allowed us to convert even powers of sine into even powers of cosine,
and viseversa. If, for instance, the power of sine was odd, we pulled out one
sin x and converted the remaining even power of sin x into a func on using powers of cos x, leading to an easy subs tu on.
The same basic strategy applies to integrals of the form tanm x secn x dx,
albeit a bit more nuanced. The following three facts will prove useful:
d
dx (tan x)
= sec2 x,
d
dx (sec x)
= sec x tan x , and
1 + tan2 x = sec2 x (the Pythagorean Theorem).

If the integrand can be manipulated to separate a sec2 x term with the remaining secant power even, or if a sec x tan x term can be separated with the
remaining tan x power even, the Pythagorean Theorem can be employed, leading to a simple subs tu on. This strategy is outlined in the following Key Idea.
Notes:
280
Key Idea 12
Integrals Involving Powers of Tangent and Secant
Consider tanm x secn x dx, where m, n are nonnega ve integers.

1. If n is even, then n = 2k for some integer k. Rewrite secn x as
secn x = sec2k x = sec2k2 x sec2 x = (1 + tan2 x)k1 sec2 x.
Then
tan x sec x dx =
tan x(1 + tan x)
where u = tan x and du = sec2 x dx.
k1
sec x dx =
um (1 + u2 )k1 du,
2. If m is odd, then m = 2k + 1 for some integer k. Rewrite tanm x secn x as

tanm x secn x = tan2k+1 x secn x = tan2k x secn1 x sec x tan x = (sec2 x 1)k secn1 x sec x tan x.
Then
tan x sec x dx =
(sec x 1) sec
2
n1
x sec x tan x dx =
(u2 1)k un1 du,
where u = sec x and du = sec x tan x dx.

3. If n is odd and m is even, then m = 2k for some integer k. Convert tanm x to (sec2 x 1)k . Expand
the new integrand and use Integra on By Parts, with dv = sec2 x dx.
4. If m is even and n = 0, rewrite tanm x as
tanm x = tanm2 x tan2 x = tanm2 x(sec2 x 1) = tanm2 sec2 x tanm2 x.
So
m2
tan x dx =
tan
|
sec x dx
{z
}
apply rule #1
tanm2 x dx .
|
{z
}
apply rule #4 again
The techniques described in items 1 and 2 of Key Idea 12 are rela vely straightforward, but the techniques in items 3 and 4 can be rather tedious. A few examples will help with these methods.
Notes:
281

. Example 169
Integra ng powers of tangent and secant
2
Evaluate tan x sec6 x dx.
Since the power of secant is even, we use rule #1 from Key
S
Idea 12 and pull out a sec2 x in the integrand. We convert the remaining powers
of secant into powers of tangent.
tan2 x sec6 x dx = tan2 x sec4 x sec2 x dx
(
)2
= tan2 x 1 + tan2 x sec2 x dx
Now subs tute, with u = tan x, with du = sec2 x dx.
(
)2
= u2 1 + u2 du
We leave the integra on and subsequent subs tu on to the reader. The nal
answer is
=
1
2
1
tan3 x + tan5 x + tan7 x + C.
3
5
7
. Example 170
.Integra ng powers of tangent and secant
3
Evaluate sec x dx.
We apply rule #3 from Key Idea 12 as the power of secant is
S
odd and the power of tangent is even (0 is an even number). We use Integra on
by Parts; the rule suggests le ng dv = sec2 x dx, meaning that u = sec x.
u = sec x
du = ?
v=?
2
dv = sec x dx
u = sec x
du = sec x tan x dx
Employing Integra on by Parts, we have
2
sec3 x dx = sec
| {zx dx}
|{z}x sec
u
= sec x tan x
Notes:
282
dv
sec x tan2 x dx.
v = tan x
dv = sec2 x dx

This new integral also requires applying rule #3 of Key Idea 12:
(
)
= sec x tan x sec x sec2 x 1 dx
= sec x tan x sec3 x dx + sec x dx
= sec x tan x sec3 x dx + ln | sec x + tan x|

In previous applica ons of Integra on by Parts, we have seen where
the original
integral has reappeared in our work. We resolve this by adding sec3 x dx to
both sides, giving:
2 sec3 x dx = sec x tan x + ln | sec x + tan x|
)
1(
sec x tan x + ln | sec x + tan x| + C
sec3 x dx =
.
2
We give one more example.
. Example 171
.Integra ng powers of tangent and secant
6
Evaluate tan x dx.
S
We employ rule #4 of Key Idea 12.
tan6 x dx = tan4 x tan2 x dx
(
)
= tan4 x sec2 x 1 dx
= tan4 x sec2 x dx tan4 x dx
Integrate the rst integral with subs tu on, u = tan x; integrate the second by
employing rule #4 again.
1
tan5 x tan2 x tan2 x dx
5
(
)
1
5
= tan x tan2 x sec2 x 1 dx
5
1
= tan5 x tan2 x sec2 x dx + tan2 x dx
5
=
Notes:
283

Again, use subs tu on for the rst integral and rule #4 for the second.
Notes:
284
1
= tan5 x
5
1
= tan5 x
5
( 2
)
1
3
tan x +
sec x 1 dx
3
1
tan3 x + tan x x + C
3
Exercises 6.3
Terms and Concepts
16.
sin2 x cos2 x dx cannot be evaluated using the tech-
17.
niques described in this sec on since both powers of sin x

and cos x are even.
2. T/F: sin3 x cos3 x dx cannot be evaluated using the tech-
18.
1. T/F:
niques described in this sec on since both powers of sin x

and cos x are odd.
3. T/F: This sec on addresses how to evaluate indenite integrals such as sin5 x tan3 x dx.
sin6 x cos5 x dx
In Exercises 27 33, evaluate the denite integral. Note: the

corresponding indenite integrals appear in the previous set.

27.
sin x cos4 x dx
sin2 x cos2 x dx
sin3 x cos x dx
/2
29.
sin(5x) cos(3x) dx
sin(x) cos(2x) dx
sin(3x) sin(7x) dx
/2
sin(5x) cos(3x) dx
0
/2
31.
cos(x) cos(2x) dx
sin(x) sin(2x) dx
32.
cos(x) cos(2x) dx
33.
sin2 x cos7 x dx
/2
30.
15.
tan2 x sec3 x dx
28.
14.
tan2 x sec x dx
sin2 x cos7 x dx
13.
sec5 x dx
26.
sin3 x cos3 x dx
12.
tan4 x dx
25.
11.
tan5 x sec5 x dx
sin3 x cos2 x dx
10.
tan3 x sec3 x dx
24.
9.
tan3 x sec2 x dx
21.
sin3 x cos x dx
8.
tan3 x sec4 x dx
20.
23.
7.
tan2 x sec4 x dx
4.
sin x cos4 x dx
6.
tan4 x sec2 x dx
19.
22.
( )
x cos(x) dx
2
Problems
5.
cos
/2
/4
tan4 x sec2 x dx
0
/4
/4
tan2 x sec4 x dx
6.4 Trigonometric Subs tu on

In Sec on 5.2 we dened the denite integral as the signed area under the
curve. In that sec on we had not yet learned the Fundamental Theorem of
Calculus, so we evaluated special denite integrals which described nice, geometric shapes. For instance, we were able to evaluate
3
9
9 x2 dx =
(6.1)
2
3
as we recognized that f(x) = 9 x2 described the upper half of a circle with

radius 3.
We have since learned a number of integra on techniques, including Subs tu on and Integra on by Parts, yet we are s ll unable to evaluate the above
integral without resor ng to a geometric interpreta on. This sec on introduces
Trigonometric Subs tu on, a method of integra on that lls this gap in our integra on skill. This technique works on the same principle as Subs tu on as found
in Sec on 6.1, though it can feel backward. In Sec on 6.1, we set u = f(x), for
some func on f, and replaced f(x) with u. In this sec on, we will set x = f(),
where f is a trigonometric func on, then replace x with f().
We start by demonstra ng this method in evalua ng the integral in (6.1).
A er the example, we will generalize the method and give more examples.
. Example 172
.Using Trigonometric Subs tu on
3
Evaluate
9 x2 dx.
3
We begin by no ng that 9 sin2 + 9 cos2 = 9, and hence

9 cos = 99 sin . If we let x = 3 sin , then 9x2 = 99 sin2 = 9 cos2 .
Se ng x = 3 sin gives dx = 3 cos d. We are almost ready to subs tute.
We also wish to change our bounds of integra on. The bound x = 3 corresponds to = /2 (for when = /2, x = 3 sin = 3). Likewise, the
bound of x = 3 is replaced by the bound = /2. Thus
3
/2
9 x2 dx =
9 9 sin2 (3 cos ) d
S
/2
/2
/2
3 9 cos2 cos d
/2
=
/2
3|3 cos | cos d.
On [/2, /2], cos is always posi ve, so we can drop the absolute value bars,
then employ a powerreducing formula:
Notes:
286
/2
/2
9 cos2 d
)
9(
1 + cos(2) d
/2 2
/2
)
9(
1
9
= + sin(2)
= .

2
2
2
/2
/2
This matches our answer from before. .

We now describe in detail Trigonometric
on. This method
Subs tu
excels
when dealing with integrands that contain a2 x2 , x2 a2 and x2 + a2 .
The following Key Idea outlines the procedure for each case, followed by more
examples.
Key Idea 13
Trigonometric Subs tu on
(a) For integrands containing a2 x2 :

Let x = a sin ,
Thus = sin
dx = a cos d
(x/a), for /2 /2.
On this interval, cos 0, so
a2 x2 = a cos
(b) For integrands containing
Let x = a tan ,
x2 + a2 :
dx = a sec2 d
Let x = a sec ,
x
.
x
.
dx = a sec tan d
Thus = sec (x/a). If x/a 1, then 0 < /2;

if x/a 1, then /2 < .
We restrict our work to where x a, so x/a 1, and

0 < /2. On this interval, tan 0, so
x2 a2 = a tan
a2 x2
Thus = tan1 (x/a), for /2 < < /2.

On this interval, sec > 0, so
x2 + a2 = a sec
(c) For integrands containing x2 a2 :
x2 a2
Notes:
287

. Example 173
Using Trigonometric Subs tu on
Evaluate
dx.
5 + x2
Using Key
set x =
S
Idea 213(b), we recognize a = 5 and
5
tan
.
This
makes
dx
=
5
sec
d.
We
will
use
the
fact
that
5 + x2 =
2
2
5 + 5 tan = 5 sec = 5 sec . Subs tu ng, we have:
1
1
dx =
5 sec2 d
2
2
5+x
5 + 5 tan

5 sec2
=
d
5 sec
= sec d

= ln sec + tan + C.
While the integra on steps are over, we are not yet done. The original problem
was stated in terms of x, whereas our answer is given in terms of . We must
convert back to x.
The reference triangle given in Key Idea 13(b) helps. With x = 5 tan , we
have
x
x2 + 5
tan =
and sec =
.
5
5
This gives
dx = ln sec + tan + C
2
5+x

x2 + 5
x

= ln
+ + C.

5
5
We can leave this answer as is, or we can use a logarithmic iden ty to simplify
it. Note:

x2 + 5
1 (
)
x

2

ln
+ + C = ln
x + 5 + x + C

5
5
5

1
= ln + ln x2 + 5 + x + C
5

= ln x2 + 5 + x + C,
( )
where the ln 1/ 5 term is absorbed into the constant C. (In Sec on 6.6 we
will learn another way of approaching this problem.) .
Notes:
288

. Example 174

2
Evaluate
4x 1 dx.
We start by rewri ng the integrand so that it looks like x2 a2

for some value of a:
S
(
)
1
2
4 x
4
( )2
1
= 2 x2
.
2
4x2 1 =
So we have a = 1/2, and following Key Idea 13(c), we set x = 12 sec , and hence
dx = 12 sec tan d. We now rewrite the integral with these subs tu ons:
( )2

1
2
2
4x 1 dx = 2 x
dx
2
(
)

1 1
1
2
sec
sec tan d
= 2
4
4 2
(
)
1
=
(sec2 1) sec tan d
4

(
)
1
=
tan2 sec tan d
4
1
=
tan2 sec d
2
)
1 ( 2
sec 1 sec d
=
2
)
1 ( 3
=
sec sec d.
2
We integrated sec3 in Example 170, nding its an deriva ves to be
sec3 d =
)
1(
sec tan + ln | sec + tan | + C.
2
Notes:
289

Thus

)
1 ( 3
4x2 1 dx =
sec sec d
2
(
)
)
1 1(
=
sec tan + ln | sec + tan | ln | sec + tan | + C
2 2
1
= (sec tan ln | sec + tan |) + C.
4
We are not yet done. Our original integral is given in terms of x, whereas our
nal answer, as given, is in terms of . We need to rewrite our answer in terms
of x. With a = 1/2, and x = 12 sec , the reference triangle in Key Idea 13(c)
shows that
tan = x2 1/4/(1/2) = 2 x2 1/4 and sec = 2x.

Thus

)

)
1(
1(
sec tan ln sec + tan + C =
2x 2 x2 1/4 ln 2x + 2 x2 1/4 + C
4
4

)
1( 2
4x x 1/4 ln 2x + 2 x2 1/4 + C.
=
4
The nal answer is given in the last line above, repeated here:

)
1( 2
4x2 1 dx =
4x x 1/4 ln 2x + 2 x2 1/4 + C.
4
.
. Example 175

4 x2
Evaluate
dx.
x2
S
and hence
We use Key Idea 13(a) with a = 2, x = 2 sin , dx = 2 cos
4 x2 = 2 cos . This gives

4 x2
2 cos
dx
=
(2 cos ) d
x2
4 sin2
= cot2 d
= (csc2 1) d
= cot + C.
Notes:
290

We need to rewrite our answer in terms
of x. Using the reference triangle found
in Key Idea 13(a), we have cot = 4 x2 /x and = sin1 (x/2). Thus

(x)
4 x2
4 x2
dx
=
sin1
+ C.
2
x
x
2
.
Trigonometric
Substu on can be
applied in many situa ons, even those not

of the form a2 x2 , x2 a2 or x2 + a2 . In the following example, we apply it to an integral we already know how to handle.
. Example 176
Using Trigonometric Subs tu on
1
dx.
Evaluate
x2 + 1
We know the answer already as tan1 x+C. We apply Trigonometric Subs tu on here to show that we get the same answer without inherently relying on knowledge of the deriva ve of the arctangent func on.
Using Key Idea 13(b), let x = tan , dx = sec2 d and note that x2 + 1 =
2
tan + 1 = sec2 . Thus
1
1
dx =
sec2 d
x2 + 1
sec2
= 1 d
S
= + C.
Since x = tan , = tan1 x, and we conclude that
1
dx = tan1 x + C. .
x2 + 1
The next example is similar to the previous one in that it does not involve a
squareroot. It shows how several techniques and iden es can be combined
to obtain a solu on.
. Example 177
1
Evaluate
dx.
(x2 + 6x + 10)2
We start by comple ng the square, then make the subs tuS
on u = x + 3, followed by the trigonometric subs tu on of u = tan :
1
1
dx
=
du.
2
2
2
(x + 6x + 10)
(u + 1)2
Notes:
291

Now make the subs tu on u = tan , du = sec2 d:
1
=
sec2 d
2
(tan + 1)2
1
=
sec2 d
(sec2 )2
= cos2 d.
Applying a power reducing formula, we have
)
1 1
+ cos(2) d
=
2 2
1
1
= + sin(2) + C.
2
4
(
(6.2)
We need to return to the variable x. As u = tan , = tan1 u. Using the

iden ty sin(2) = 2 sin cos and using the reference triangle found in Key
Idea 13(b), we have
1
u
1
1 u
1
sin(2) =
=
.
2
4
2 u2 + 1
2 u2 + 1
u +1
Finally, we return to x with the subs tu on u = x + 3. We start with the expression in Equa on (6.2):
1
1
1
1 u
+ sin(2) + C = tan1 u +
+C
2
4
2
2 u2 + 1
1
x+3
= tan1 (x + 3) +
+ C.
2
2(x2 + 6x + 10)
Sta ng our nal result in one line,
1
1
x+3
dx = tan1 (x + 3) +
+ C.
(x2 + 6x + 10)2
2
2(x2 + 6x + 10)
.
Our last example returns us to denite integrals, as seen in our rst example.
Given a denite integral that can be evaluated using Trigonometric Subs tu on,
we could rst evaluate the corresponding indenite integral (by changing from
an integral in terms of x to one in terms of , then conver ng back to x) and then
evaluate using the original bounds. It is much more straigh orward, though, to
change the bounds as we subs tute.
Notes:
292

. Example 178
Denite integra on and Trigonometric Subs tu on
5
x2
Evaluate
dx.
x2 + 25
0
2
Using Key Idea 13(b), we set x = 5 tan , dx = 5 sec d,
2
and note that x + 25 = 5 sec . As we subs tute, we can also change the
bounds of integra on.
The lower bound of the original integral is x = 0. As x = 5 tan , we solve for
and nd = tan1 (x/5). Thus the new lower bound is = tan1 (0) = 0. The
original upper bound is x = 5, thus the new upper bound is = tan1 (5/5) =
/4.
Thus we have
5
/4
x2
25 tan2
dx =
5 sec2 d
2
5 sec
x + 25
0
0
/4
= 25
tan2 sec d.
We encountered this indenite integral in Example 174 where we found
)
1(
tan2 sec d =
sec tan ln | sec + tan | .
2
So
/4
)
25 (
tan sec d =
sec tan ln | sec + tan |

2
0
)
25 (
2 ln( 2 + 1)
=
2
6.661.
2
25
0
/4
The following equali es are very useful when evalua ng integrals using Trigonometric Subs tu on.
Key Idea 14
Useful Equali es with Trigonometric Subs tu on
1. sin(2) = 2 sin cos
2. cos(2) = cos2 sin2 = 2 cos2 1 = 1 2 sin2

)
1(
3.
sec3 d =
sec tan + ln sec + tan + C
2
)
)
1(
1(
4.
cos2 d =
1 + cos(2) d = + sin cos + C.
2
2
Notes:
293
Exercises 6.4
Terms and Concepts
1. Trigonometric Subs tu on works on the same principles as
Integra on by Subs tu on, though it can feel
.
2. If one uses
Trigonometric Subs tu on on an integrand containing 25 x2 , then one should set x =
.
3. Consider the Pythagorean Iden ty sin2 + cos2 = 1.
(a) What iden ty is obtained when both sides are divided by cos2 ?
(b) Use the new iden ty to simplify 9 tan2 + 9.
4. Why does Key Idea 13(a) state that a2 x2 = a cos ,

and not |a cos |?
In Exercises 17 26, evaluate the indenite integrals. Some

may be evaluated without Trigonometric Subs tu on.
2
x 11
17.
dx
x
1
18.
dx
(x2 + 1)2
19.
dx
2
x 3

20.
x2 1 x2 dx
x
dx
(x2 + 9)3/2
21.
Problems
22.
In Exercises 5 16, apply Trigonometric Subs tu on to evaluate the indenite integrals.

5.
x2 + 1 dx
23.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
1
dx
(x2 + 4x + 13)2
x2 (1 x2 )3/2 dx
24.

x2 + 4 dx
25.
1 x2 dx
26.
9 x2 dx
x2 1 dx
x2 16 dx
16x2
dx
1 dx
30.
dx
7 x2
31.
5
dx
x2 8
32.
x2 16 dx
29.
dx
2
x +2
x2
dx
+3
x2
28.
1
5 x2
dx
7x2
In Exercises 27 32, evaluate the denite integrals by making the proper trigonometric subs tu on and changing the
bounds of integra on. (Note: each of the corresponding
indenite integrals has appeared previously in this Exercise
set.)
1
27.
1 x2 dx
4x2 + 1 dx
9x2
5x2
dx
10
x2
x2 + 4 dx
0
1
1
1
1
1
1
1
dx
(x2 + 1)2
9 x2 dx
x2 1 x2 dx
6.5
Par al Frac on Decomposi on
In this sec on we inves gate the an deriva ves of ra onal func ons. Recall that
ra onal func ons are func ons of the form f(x) = p(x)
q(x) , where p(x) and q(x) are
polynomials and q(x) = 0. Such func ons arise in many contexts, one of which
is the solving of certain fundamental dieren al equa ons.
We begin with an examplethat demonstrates the mo va on behind this
1
sec on. Consider the integral
dx. We do not have a simple formula
x2 1
2
for this (if the denominator were x + 1, we would recognize the an deriva ve
as being the arctangent func on). It can be solved using Trigonometric Subs tu on, but note how the integral is easy to evaluate once we realize:
x2
1
1/2
1/2
=
.
1
x1 x+1
Thus
1
dx =
x2 1
=
1/2
dx
x1
1/2
dx
x+1
1
1
ln |x 1| ln |x + 1| + C.
2
2
This sec on teaches how to decompose

1
x2 1
into
1/2
1/2
.
x1 x+1
We start with a ra onal func on f(x) = p(x)

q(x) , where p and q do not have any
common factors and the degree of p is less than the degree of q. It can be shown
that any polynomial, and hence q, can be factored into a product of linear and
irreducible quadra c terms. The following Key Idea states how to decompose a
ra onal func on into a sum of ra onal func ons whose denominators are all of
lower degree than q.
Notes:
295

.
Key Idea 15
Par al Frac on Decomposi on
p(x)
be a ra onal func on, where the degree of p is less than the
q(x)
degree of q.
Let
1. Linear Terms: Let (x a) divide q(x), where (x a)n is the highest

power of (x a) that divides q(x). Then the decomposi on of p(x)
q(x)
will contain the sum
A1
A2
An
+
+ +
.
(x a) (x a)2
(x a)n
2. Quadra c Terms: Let x2 + bx + c divide q(x), where (x2 + bx + c)n
is the highest power of x2 + bx + c that divides q(x). Then the
p(x)
decomposi on of q(x)
will contain the sum
B2 x + C 2
B1 x + C 1
Bn x + Cn
+
+ + 2
.
x2 + bx + c (x2 + bx + c)2
(x + bx + c)n
To nd the coecients Ai , Bi and Ci :
1. Mul ply all frac ons by q(x), clearing the denominators. Collect
like terms.
2. Equate the resul ng coecients of the powers of x and solve the
resul ng system of linear equa ons.
The following examples will demonstrate how to put this Key Idea into pracce. Example 179 stresses the decomposi on aspect of the Key Idea.
. Example 179
.Decomposing into par al frac ons

1
Decompose f(x) =
without solving
(x + 5)(x 2)3 (x2 + 2x + 1)(x2 + x + 7)2
for the resul ng coecients.
The denominator is already factored; we need to decomS
pose f(x) properly. Since (x + 5) is a linear term that divides the denominator,
there will be a
A
x+5
term in the decomposi on.
Notes:
296

As (x 2)3 divides the denominator, we will have the following terms in the
decomposi on:
B
C
D
,
and
.
x2
(x 2)2
(x 2)3
The x2 + 2x + 1 term in the denominator results in a
Finally, the (x2 + x + 7)2 term results in the terms
Gx + H
+x+7
x2
and
(x2
x2
Ex + F
term.
+ 2x + 1
Ix + J
.
+ x + 7)2
All together, we have

A
B
C
D
1
=
+
+
+
+
(x + 5)(x 2)3 (x2 + 2x + 1)(x2 + x + 7)2
x + 5 x 2 (x 2)2
(x 2)3
Ex + F
Gx + H
Ix + J
+ 2
+ 2
2
x + 2x + 1 x + x + 7 (x + x + 7)2
Solving for the coecients A, B . . . J would be a bit tedious but not hard. .
. Example 180
.Decomposing into par al frac ons

1
.
Perform the par al frac on decomposi on of 2
x 1
The denominator factors into two linear terms: x2 1 =
(x 1)(x + 1). Thus
1
A
B
=
+
.
x2 1
x1 x+1
S
To solve for A and B, rst mul ply through by x2 1 = (x 1)(x + 1):

A(x 1)(x + 1) B(x 1)(x+)
+
x1
x+1
= A(x + 1) + B(x 1)
= Ax + A + Bx B
1=
Now collect like terms.

= (A + B)x + (A B).
The next step is key. Note the equality we have:
1 = (A + B)x + (A B).
Notes:
297

For claritys sake, rewrite the le hand side as
0x + 1 = (A + B)x + (A B).
On the le , the coecient of the x term is 0; on the right, it is (A + B). Since
both sides are equal, we must have that 0 = A + B.
Likewise, on the le , we have a constant term of 1; on the right, the constant
term is (A B). Therefore we have 1 = A B.
We have two linear equa ons with two unknowns. This one is easy to solve
by hand, leading to
A+B=0
A = 1/2
.
AB=1
B = 1/2
Thus
x2
.
. Example 181
1/2
1/2
1
=
.
1
x1 x+1
.Integra ng using par al frac ons
Use par al frac on decomposi on to integrate
We decompose the integrand as follows, as described by Key
S
Note: Equa on 6.3 oers a direct route to
nding the values of A, B and C. Since the
equa on holds for all values of x, it holds
in par cular when x = 1. However, when
x = 1, the right hand side simplies to
A(1 + 2)2 = 9A. Since the le hand side
is s ll 1, we have 1 = 9A. Hence A = 1/9.
Likewise, the equality holds when x =
2; this leads to the equa on 1 = 3C.
Thus C = 1/3.
We can nd the value of B by expanding
the terms as shown in the example.
1
dx.
(x 1)(x + 2)2
Idea 15:
1
A
B
C
=
+
+
.
2
(x 1)(x + 2)
x 1 x + 2 (x + 2)2
To solve for A, B and C, we mul ply both sides by (x 1)(x + 2)2 and collect like
terms:
1 = A(x + 2)2 + B(x 1)(x + 2) + C(x 1)
= Ax + 4Ax + 4A + Bx + Bx 2B + Cx C
2
= (A + B)x2 + (4A + B + C)x + (4A 2B C)

We have
0x2 + 0x + 1 = (A + B)x2 + (4A + B + C)x + (4A 2B C)
leading to the equa ons
A + B = 0,
4A + B + C = 0 and
4A 2B C = 1.
These three equa ons of three unknowns lead to a unique solu on:
A = 1/9,
Notes:
298
B = 1/9 and
C = 1/3.
(6.3)

Thus
1
dx =
(x 1)(x + 2)2
1/9
dx +
x1
1/9
dx +
x+2
1/3
dx.
(x + 2)2
Each can be integrated with a simple subs tu on with u = x 1 or x = x + 2

(or by directly applying Key Idea 10 as the denominators are linear func ons).
The end result is
1
1
1
1
dx = ln |x 1| ln |x + 2| +
+ C.
2
(x
1)(x
+
2)
9
9
3(x
+ 2)
.
. Example 182
.Integra ng using par al frac ons
Use par al frac on decomposi on to integrate
x3
dx.
(x 5)(x + 3)
Key Idea 15 presumes that the degree of the numerator is

S
less than the degree of the denominator. Since this is not the case here, we
begin by using polynomial division to reduce the degree of the numerator. We
omit the steps, but encourage the reader to verify that
x3
19x + 30
=x+2+
.
(x 5)(x + 3)
(x 5)(x + 3)
Note: The values of A and B can be quickly

found using the technique described in
the margin of Example 181.
Using Key Idea 15, we can rewrite the new ra onal func on as:
19x + 30
A
B
=
+
(x 5)(x + 3)
x5 x+3
for appropriate values of A and B. Clearing denominators, we have
19x + 30 = A(x + 3) + B(x 5)
= (A + B)x + (3A 5B).
This implies that:
19 = A + B
30 = 3A 5B.
Solving this system of linear equa ons gives
125/8 = A
27/8 = B.
Notes:
299

We can now integrate.
)
(
x3
125/8
27/8
dx =
x+2+
+
dx
(x 5)(x + 3)
x5
x+3
125
27
x2
=
+ 2x +
ln |x 5| +
ln |x + 3| + C.
2
8
8
.
. Example 183
.Integra ng using par al frac

ons 2
7x + 31x + 54
Use par al frac on decomposi on to evaluate
dx.
(x + 1)(x2 + 6x + 11)
The degree of the numerator is less than the degree of the

S
denominator so we begin by applying Key Idea 15. We have:
7x2 + 31x + 54
A
Bx + C
=
+ 2
.
2
(x + 1)(x + 6x + 11)
x + 1 x + 6x + 11
Now clear the denominators.
7x2 + 31x + 54 = A(x2 + 6x + 11) + (Bx + C)(x + 1)
= (A + B)x2 + (6A + B + C)x + (11A + C).
This implies that:
7=A+B
31 = 6A + B + C
54 = 11A + C.
Solving this system of linear equa ons gives the nice result of A = 5, B = 2 and
C = 1. Thus
)
(
7x2 + 31x + 54
5
2x 1
dx =
+
dx.
(x + 1)(x2 + 6x + 11)
x + 1 x2 + 6x + 11
The rst term of this new integrand is easy to evaluate; it leads to a 5 ln |x+1|
term. The second term is not hard, but takes several steps and uses subs tu on
techniques.
2x 1
The integrand 2
has a quadra c in the denominator and a linear
x + 6x + 11
term in the numerator. This leads us to try subs tu on. Let u = x2 + 6x + 11, so
du = (2x + 6) dx. The numerator is 2x 1, not 2x + 6, but we can get a 2x + 6
Notes:
300

term in the numerator by adding 0 in the form of 7 7.
2x 1
2x 1 + 7 7
= 2
x2 + 6x + 11
x + 6x + 11
2x + 6
7
= 2
.
x + 6x + 11 x2 + 6x + 11
We can now integrate the rst term with subs tu on, leading to a ln |x2 +6x+11|
term. The nal term can be integrated using arctangent. First, complete the
square in the denominator:
7
7
=
.
x2 + 6x + 11
(x + 3)2 + 2
An an deriva ve of the la er term can be found using Theorem 46 and subs tu on:
(
)
7
7
x+3
1
dx
=
tan
+ C.
x2 + 6x + 11
2
2
Lets start at the beginning and put all of the steps together.
)
2x 1
5
+ 2
dx
x+1
x + 6x + 11
5
2x + 6
7
=
dx +
dx
dx
2
x+1
x + 6x + 11
x2 + 6x + 11
(
)
7
x+3
= 5 ln |x + 1| + ln |x2 + 6x + 11| tan1

+ C.
2
2
7x2 + 31x + 54
dx =
(x + 1)(x2 + 6x + 11)
As with many other problems in calculus, it is important to remember that one

is not expected to see the nal answer immediately a er seeing the problem.
Rather, given the ini al problem, we break it down into smaller problems that
are easier to solve. The nal answer is a combina on of the answers of the
smaller problems. .
Notes:
301
Exercises 6.5
Terms and Concepts
14.
1. Fill in the blank: Par al Frac on Decomposi on is a method

of rewri ng
func ons.
15.
2. T/F: It is some mes necessary to use polynomial division

before using Par al Frac on Decomposi on.
16.
1
without solving for the coecients, as
x2 3x
done in Example 179.
17.
3. Decompose
7x
x2 9
4. Decompose
18.
x3
5. Decompose 2
x 7
19.
2x + 5
x3 + 7x
20.
6. Decompose
21.
Problems
7x + 7
7.
dx
x2 + 3x 10
7x 2
8.
dx
x2 + x
4
9.
dx
3x2 12
x+7
10.
dx
(x + 5)2
3x 20
11.
dx
(x + 8)2
9x2 + 11x + 7
12.
dx
x(x + 1)2
12x2 x + 33
13.
dx
(x 1)(x + 3)(3 2x)
22.
23.
24.
25.
94x2 10x
dx
(7x + 3)(5x 1)(3x 1)
x2 + x + 1
dx
x2 + x 2
x3
dx
x2 x 20
2x2 4x + 6
dx
x2 2x + 3
1
dx
x3 + 2x2 + 3x
x2 + x + 5
dx
x2 + 4x + 10
12x2 + 21x + 3
dx
(x + 1)(3x2 + 5x 1)
6x2 + 8x 4
dx
(x 3)(x2 + 6x + 10)
2x2 + x + 1
dx
(x + 1)(x2 + 9)
x2 20x 69
dx
(x 7)(x2 + 2x + 17)
9x2 60x + 33
dx
(x 9)(x2 2x + 11)
6x2 + 45x + 121
dx
(x + 2)(x2 + 10x + 27)

2
8x + 21
26.
dx
(x
+
2)(x + 3)
1
5
14x + 6
dx
27.
(3x
+
2)(x + 4)
0
1
x2 + 5x 5
28.
dx
2
1 (x 10)(x + 4x + 5)
1
x
29.
dx
2
0 (x + 1)(x + 2x + 1)
6.6 Hyperbolic Func ons
6.6
Hyperbolic Func ons
y
x2 + y 2 = 1
The hyperbolic func ons are a set of func ons that have many applica ons to
mathema cs, physics, and engineering. Among many other applica ons, they
are used to describe the forma on of satellite rings around planets, to describe
the shape of a rope hanging from two points, and have applica on to the theory
of special rela vity. This sec on denes the hyperbolic func ons and describes
many of their proper es, especially their usefulness to calculus.
These func ons are some mes referred to as the hyperbolic trigonometric
func ons as there are many, many connec ons between them and the standard trigonometric func ons. Figure 6.13 demonstrates one such connec on.
Just as cosine and sine are used to dene points on the circle dened by x2 +y2 =
1, the func ons hyperbolic cosine and hyperbolic sine are used to dene points
on the hyperbola x2 y2 = 1.
We begin with their deni on.
Deni on 23
ex + ex
2
2. sinh x =
ex ex
2
3. tanh x =
sinh x
cosh x
y
2
x y =1
2
(cosh ,sinh )
4. sech x =
1
cosh x
5. csch x =
1
sinh x
6. coth x =
cosh x
sinh x
(cos ,sin )
Hyperbolic Func ons
1. cosh x =
These hyperbolic func ons are graphed in Figure 6.14. In the graphs of cosh x
and sinh x, graphs of ex /2 and ex /2 are included with dashed lines. As x gets
large, cosh x and sinh x each act like ex /2; when x is a large nega ve number,
cosh x acts like ex /2 whereas sinh x acts like ex /2.
No ce the domains of tanh x and sech x are (, ), whereas both coth x
and csch x have ver cal asymptotes at x = 0. Also note the ranges of these
func on, especially tanh x: as x , both sinh x and cosh x approach ex /2,
hence tanh x approaches 1.
The following example explores some of the proper es of these func ons
that bear remarkable resemblance to the proper es of their trigonometric counterparts.
x
2
.
Figure 6.13: Using trigonometric funcons to dene points on a circle and hyperbolic func ons to dene points on a
hyperbola. The area of the shaded regions are included in them.
Pronuncia on Note:
cosh rhymes with gosh,
sinh rhymes with pinch, and
tanh rhymes with ranch.
Notes:
303
10
10
5
x
x
2
2
5
f(x) = cosh x
10
2
f(x) = sinh x
10
.
y
y
3
f(x) = coth x
f(x) = csch x
f(x) = sech x
1
x
2
x
2
2
f(x) = tanh x
1
2
Figure 6.14: Graphs of the hyperbolic func ons.
. Example 184
.
.Exploring proper es of hyperbolic

func ons
.
Use Deni on 23 to rewrite the following expressions.

1. cosh2 x sinh2 x
2. tanh2 x + sech2 x
3. 2 cosh x sinh x
Notes:
304
)
cosh x
(
)
d
5. dx
sinh x
(
)
d
6. dx
tanh x
4.
d
dx

(
)2 ( x
)2
ex + ex
e ex
cosh x sinh x =
2
2
2x
x x
2x
e + 2e e + e
e2x 2ex ex + e2x
=
4
4
4
= = 1.
4
2
1.
So cosh2 x sinh2 x = 1.
sinh2 x
1
2 +
cosh x cosh2 x
sinh2 x + 1
Now use iden ty from #1.
=
cosh2 x
cosh2 x
=
=1
cosh2 x
tanh2 x + sech2 x =
2.
So tanh2 x + sech2 x = 1.
.
(
)( x
)
ex + ex
e ex
2
2
2x
2x
e e
=2
4
e2x e2x
=
= sinh(2x).
2
3.
2 cosh x sinh x = 2
Thus 2 cosh x sinh x = sinh(2x).

4.
So
d
dx
5.
So
d
dx
(
)
)
d ex + ex
d(
cosh x =
dx
dx
2
x
x
e e
=
2
= sinh x
)
cosh x = sinh x.
(
)
)
d(
d ex ex
sinh x =
dx
dx
2
ex + ex
=
2
= cosh x
)
sinh x = cosh x.
Notes:
305
6.
So
d
dx
(
)
)
d(
d sinh x
tanh x =
dx
dx cosh x
cosh x cosh x sinh x sinh x
=
cosh2 x
1
=
cosh2 x
= sech2 x
)
tanh x = sech2 x.
The following Key Idea summarizes many of the important iden es rela ng
to hyperbolic func ons. Each can be veried by referring back to Deni on 23.
Key Idea 16
Basic Iden
Useful Hyperbolic Func on Proper es

es
1. cosh2 x sinh2 x = 1
2. tanh2 x + sech2 x = 1
3. coth2 x csch2 x = 1
4. cosh 2x = cosh2 x + sinh2 x
5. sinh 2x = 2 sinh x cosh x
6. cosh2 x =
cosh 2x + 1
2
7. sinh2 x =
cosh 2x 1
2
Deriva ves
Integrals
(
)
d
1. dx
cosh x = sinh x
1.
cosh x dx = sinh x + C
(
)
d
2. dx sinh x = cosh x
(
)
2.
sinh x dx = cosh x + C
d
3. dx
tanh x = sech2 x
(
)
d
4. dx
sech x = sech x tanh x
3.
tanh x dx = ln(cosh x) + C
(
)
d
5. dx csch x = csch x coth x

4.
coth x dx = ln | sinh x | + C
(
)
d
6. dx
coth x = csch2 x
We prac ce using Key Idea 16.

. Example 185
.Deriva ves and integrals of hyperbolic func ons
Evaluate the following deriva ves and integrals.
)
d(
cosh 2x
dx
2.
sech2 (7t 3) dt
1.
Notes:
306
3.
ln 2
cosh x dx
0

S
1. Using the Chain Rule directly, we have
d
dx
)
cosh 2x = 2 sinh 2x.
Just to demonstrate that it works, lets also use the Basic Iden ty found in
Key Idea 16: cosh 2x = cosh2 x + sinh2 x.
)
)
d(
d(
cosh 2x =
cosh2 x + sinh2 x = 2 cosh x sinh x + 2 sinh x cosh x
dx
dx
= 4 cosh x sinh x.
Using another Basic Iden ty, we can see that 4 cosh x sinh x = 2 sinh 2x.
We get the same answer either way.
2. We employ subs tu on, with u = 7t 3 and du = 7dt. Applying Key
Ideas 10 and 16 we have:
1
sech2 (7t 3) dt = tanh(7t 3) + C.
7
3.
ln 2
0
ln 2

cosh x dx = sinh x = sinh(ln 2) sinh 0 = sinh(ln 2).
0
We can simplify this last expression as sinh x is based on exponen als:
sinh(ln 2) =
eln 2 e ln 2
2 1/2
3
=
= .
2
2
4
Inverse Hyperbolic Func ons

Just as the inverse trigonometric func ons are useful in certain integra ons,
the inverse hyperbolic func ons are useful with others. Figure 6.15 shows the
restric ons on the domains to make each func on one-to-one and the resul ng
domains and ranges of their inverse func ons. Their graphs are shown in Figure
6.16.
Because the hyperbolic func ons are dened in terms of exponen al funcons, their inverses can be expressed in terms of logarithms as(shown
in Key)Idea
17. It is o en more convenient to refer to sinh1 x than to ln x + x2 + 1 , especially when one is working on theory and does not need to compute actual
values. On the other hand, when computa ons are needed, technology is o en
helpful but many hand-held calculators lack a convenient sinh1 x bu on. (Often it can be accessed under a menu system, but not conveniently.) In such a
situa on, the logarithmic representa on is useful.
Notes:
307

Func on
cosh x
sinh x
tanh x
sech x
csch x
coth x
Domain
[0, )
(, )
(, )
[0, )
(, 0) (0, )
(, 0) (0, )
Func on
Range
[1, )
(, )
(1, 1)
(0, 1]
(, 0) (0, )
(, 1) (1, )
cosh x
sinh1 x
tanh1 x
sech1 x
csch1 x
coth1 x
Domain
Range
[1, )
[, )
(1, 1)
(0, 1]
(, 0) (0, )
(, 1) (1, )
[0, )
[, )
(, )
[0, )
(, 0) (0, )
(, 0) (0, )
Figure 6.15: Domains and ranges of the hyperbolic and inverse hyperbolic func ons.
y
y
y = cosh x
10
10
y = sinh x
5
y = cosh1 x
10
5
y = sinh1 x
5
x
10
10
10
y
3
2
y = coth1 x
1
x
2
x
2
2
2
.
.
y = csch1 x
y = tanh1 x
2
y = sech1 x
Figure 6.16: Graphs of the hyperbolic func ons and their inverses.
Key Idea 17
Logarithmic deni ons of Inverse Hyperbolic Func ons
(
)
1. cosh1 x = ln x + x2 1 ; x 1
2. tanh
1
x = ln
2
(
3. sech1 x = ln
)
1+x
; |x| < 1
1x
)
1 + 1 x2
; 0<x1
x
Notes:
308
(
)
4. sinh1 x = ln x + x2 + 1
5. coth
1
x = ln
2
(
6. csch1 x = ln
)
x+1
; |x| > 1
x1
)
1
1 + x2
+
; x = 0
x
|x|

The following Key Ideas give the deriva ves and integrals rela ng to the inverse hyperbolic func ons. In Key Idea 19, both the inverse hyperbolic and logarithmic func on representa ons of the an deriva ve are given, based on Key
Idea 17. Again, these la er func ons are o en more useful than the former.
Note how inverse hyperbolic func ons can be used to solve integrals we used
Trigonometric Subs tu on to solve in Sec on 6.4.
Key Idea 18
1.
Deriva ves Involving Inverse Hyperbolic Func ons
)
d(
1
; x>1
cosh1 x =
2
dx
x 1
)
d(
1
2.
sinh1 x =
2
dx
x +1
3.
)
d(
1
tanh1 x =
; |x| < 1
dx
1 x2
4.
)
d(
1
;0 < x < 1
sech1 x =
dx
x 1 x2
5.
)
d(
1
csch1 x =
; x = 0
dx
|x| 1 + x2
6.
)
1
d(
coth1 x =
; |x| > 1
dx
1 x2
Key Idea 19
Integrals Involving Inverse Hyperbolic Func ons
(x)
1
dx =
cosh1
+ C; 0 < a < x
1.
a
x2 a2
(x)
1
dx =
sinh1
+ C; a > 0
2.
a
x2 + a2
1
( )
x2 < a2
a tanh1 ax + C
1
3.
dx
=
( )
1
a2 x2
1 x
a2 < x2
a coth
a +C
( )
1
1
1 x
dx =
4.
sech
+ C; 0 < x < a
a
a
x a 2 x2

1
1
1 x
dx =
5.
csch
+ C; x = 0, a > 0
a
a
x x2 + a2

= ln x + x2 a2 + C

= ln x + x2 + a2 + C

1 a + x
= ln
+C
2
a x
)
(
1
x
+C
ln
a
a + a2 x2

1
x
+C
= ln
2
2
a
a+ a +x
=
We prac ce using the deriva ve and integral formulas in the following example.
Notes:
309

. Example 186
.Deriva ves and integrals involving inverse hyperbolic funcons
Evaluate the following.
[
(
)]
d
3x 2
1
1.
cosh
dx
5
1
2.
dx
x2 1
3.
1
dx
9x2 + 10
1. Applying Key Idea 18 with the Chain Rule gives:

[
(
)]
d
3x 2
1
3
1
cosh
= (
.
)
dx
5
3x2
1 5
5
1
2. Mul plying the numerator and denominator by (1) gives:
dx =
2
x 1
1
dx. The second integral can be solved with a direct applica on
1 x2
of item #3 from Key Idea 19, with a = 1. Thus
1
1
dx
=
dx
2
x 1
1 x2
x2 < 1
tanh1 (x) + C
=
coth1 (x) + C
1 < x2

1 x + 1
= ln
+C
2
x 1

1 x 1
(6.4)
= ln
+ C.
2
x + 1
We should note that this exact problem was solved at the beginning of
Sec on 6.5. In that example the answer was given as 21 ln |x1| 12 ln |x+
1| + C. Note that this is equivalent to the answer given in Equa on 6.4, as
ln(a/b) = ln a ln b.
3. This requires a subs tu on, then item #2 of Key Idea 19 can be applied.
Let u = 3x, hence du = 3dx. We have
1
1
1
dx =
du.
3
9x2 + 10
u2 + 10
Notes:
310

Note a2 = 10, hence a =
10. Now apply the integral rule.

(
)
1
3x
1
= sinh
+C
3
10

1

= ln 3x + 9x2 + 10 + C.
3
Notes:
311
Exercises 6.6
Terms and Concepts
22. f(x) = sinh x at x = 0
1. In Key Idea 16, the equa on
23. f(x) = cosh x at x = ln 2

tanh x dx = ln(cosh x)+C is
given. Why is ln | cosh x| not used i.e., why are absolute

values not necessary?
2. The hyperbolic func ons are used to dene points on the
right hand por on of the hyperbola x2 y2 = 1, as shown
in Figure 6.13. How can we use the hyperbolic func ons to
dene points on the le hand por on of the hyperbola?
24. f(x) = sech2 x at x = ln 3

25. f(x) = sinh1 x at x = 0
26. f(x) = cosh1 x at x = 2

27.
tanh(2x) dx
Problems
28.
In Exercises 3 10, verify the given iden ty using Deni on

23, as done in Example 184.
29.
3. coth2 x csch2 x = 1
4. cosh 2x = cosh2 x + sinh2 x
cosh 2x + 1
5. cosh2 x =
2
cosh
2x
1
6. sinh2 x =
2
d
7.
[sech x] = sech x tanh x
dx
d
8.
[coth x] = csch2 x
dx
9.
10.
coth x dx = ln | sinh x| + C
In Exercises 11 21, nd the deriva ve of the given func on.

11. f(x) = cosh 2x
12. f(x) = tanh(x2 )
13. f(x) = ln(sinh x)
14. f(x) = sinh x cosh x
15. f(x) = x sinh x cosh x
16. f(x) = sech1 (x2 )
17. f(x) = sinh
18. f(x) = cosh
(3x)
(2x2 )
19. f(x) = tanh1 (x + 5)
21. f(x) = cosh
(sec x)
In Exercises 22 26, nd the equa on of the line tangent to

the func on at the given x-value.
sinh x cosh x dx
30.
x cosh x dx
31.
x sinh x dx
1
dx
9 x2
32.
2x
dx
x4 4
34.
dx
1 + x3
1
35.
dx
x4 16
1
36.
dx
x2 + x
ex
dx
37.
2x
e +1
38.
sinh1 x dx
33.
tanh1 x dx
39.
40.
sech x dx
(Hint: mu ply by
cosh x
;
cosh x
set u = sinh x.)
In Exercises 41 43, evaluate the given denite integral.

1
41.
sinh x dx
20. f(x) = tanh1 (cos x)

1
cosh(3x 7) dx
1
ln 2
42.
cosh x dx
ln 2
1
43.
0
tanh1 x dx
6.7 LHpitals Rule
6.7
LHpitals Rule
While this chapter is devoted to learning techniques of integra on, this sec on
is not about integra on. Rather, it is concerned with a technique of evalua ng
certain limits that will be useful in the following sec on, where integra on is
once more discussed.
Our treatment of limits exposed us to 0/0, an indeterminate form. If lim f(x) =
xc
0 and lim g(x) = 0, we do not conclude that lim f(x)/g(x) is 0/0; rather, we use
xc
xc
0/0 as nota on to describe the fact that both the numerator and denominator
approach 0. The expression 0/0 has no numeric value; other work must be done
to evaluate the limit.
Other indeterminate forms exist; they are: /, 0 , , 00 , 1 and
0
. Just as 0/0 does not mean divide 0 by 0, the expression / does
not mean divide innity by innity. Instead, it means a quan ty is growing
without bound and is being divided by another quan ty that is growing without
bound. We cannot determine from such a statement what value, if any, results
in the limit. Likewise, 0 does not mean mul ply zero by innity. Instead,
it means one quan ty is shrinking to zero, and is being mul plied by a quan ty
that is growing without bound. We cannot determine from such a descrip on
what the result of such a limit will be.
This sec on introduces lHpitals Rule, a method of resolving limits that produce the indeterminate forms 0/0 and /. Well also show how algebraic
manipula on can be used to convert other indeterminate expressions into one
of these two form so that our new rule can be applied.
Theorem 49
LHpitals Rule, Part 1
Let lim f(x) = 0 and lim g(x) = 0, where f and g are dieren able funcxc
xc
ons on an open interval I containing c, and g (x) = 0 on I except possibly

at c. Then
f(x)
f (x)
lim
= lim .
xc g(x)
xc g (x)
We demonstrate the use of lHpitals Rule in the following examples; we

will o en use LHR as an abbrevia on of lHpitals Rule.
Notes:
313

. Example 187
Using lHpitals Rule
Evaluate the following limits, using lHpitals Rule as needed.
sin x
x0 x
x+32
2. lim
x1
1x
x2
x0 1 cos x
1. lim
3. lim
x2 + x 6
x2 x2 3x + 2
4. lim
1. We proved this limit is 1 in Example 12 using the Squeeze Theorem. Here

we use lHpitals Rule to show its power.
by LHR
sin x
x0 x
lim
x+32
lim
x1
1x
2.
lim
x0
by LHR
lim
cos x
= 1.
1
1
2 (x
x1
+ 3)1/2
1
= .
1
4
3.
by LHR
x
2x
.
= lim
x0 1 cos x
x0 sin x
This la er limit also evaluates to the 0/0 indeterminate form. To evaluate
it, we apply lHpitals Rule again.
lim
2x
x0 sin x
lim
by LHR
2
= 2.
cos x
x2
= 2.
x0 1 cos x
Thus lim
4. We already know how to evaluate this limit; rst factor the numerator and
denominator. We then have:
x2 + x 6
(x 2)(x + 3)
x+3
= lim
= lim
= 5.
x2 x2 3x + 2
x2 (x 2)(x 1)
x2 x 1
lim
We now show how to solve this using lHpitals Rule.
x2 + x 6
x2 x2 3x + 2
lim
by LHR
2x + 1
= 5.
x2 2x 3
lim
The following theorem extends our ini al version of lHpitals Rule in two
ways. It allows the technique to be applied to the indeterminate form /
and to limits where x approaches .
Notes:
314
6.7 LHpitals Rule
Theorem 50
LHpitals Rule, Part 2
1. Let lim f(x) = and lim g(x) = , where f and g are dierxa
xa
en able on an open interval I containing a. Then
f(x)
f (x)
= lim .
xa g(x)
xa g (x)
lim
2. Let f and g be dieren able func ons on the open interval (a, )
for some value a, where g (x) = 0 on (a, ) and lim f(x)/g(x)
x
returns either 0/0 or /. Then

f (x)
f(x)
= lim .
x g (x)
x g(x)
lim
A similar statement can be made for limits where x approaches

.
. Example 188
Using lHpitals Rule with limits involving
3x2 100x + 2
x 4x2 + 5x 1000
ex
.
x x3
1. lim
2. lim
1. We can evaluate this limit already using Theorem 11; the answer is 3/4.
We apply lHpitals Rule to demonstrate its applicability.
3x2 100x + 2
x 4x2 + 5x 1000
lim
by LHR
6x 100
x 8x + 5
lim
by LHR
lim
6
3
= .
8
4
ex by LHR
ex by LHR
ex by LHR
ex
=
lim
=
lim
=
lim
= .
x x3
x 3x2
x 6x
x 6
Recall that this means that the limit does not exist; as x approaches ,
the expression ex /x3 grows without bound. We can infer from this that
ex grows faster than x3 ; as x gets large, ex is far larger than x3 . (This
has important implica ons in compu ng when considering eciency of
algorithms.)
2. lim
Notes:
315
Indeterminate Forms 0 and

LHpitals Rule can only be applied to ra os of func ons. When faced with
an indeterminate form such as 0 or , we can some mes apply algebra
to rewrite the limit so that lHpitals Rule can be applied. We demonstrate the
general idea in the next example.
. Example 189
.Applying lHpitals Rule to other indeterminate forms
1. lim x e1/x
3. lim ln(x + 1) ln x
2. lim x e1/x
4. lim x2 ex
x0+
x0
1. As x 0+ , x 0 and e1/x . Thus we have the indeterminate form

e1/x
0 . We rewrite the expression x e1/x as
; now, as x 0+ , we get
1/x
the indeterminate form / to which lHpitals Rule can be applied.
lim x e1/x = lim
x0+
x0+
e1/x
1/x
by LHR
lim
x0+
(1/x2 )e1/x
= lim e1/x = .
1/x2
x0+
Interpreta on: e1/x grows faster than x shrinks to zero, meaning their
product grows without bound.
2. As x 0 , x 0 and e1/x e 0. The the limit evaluates to 0 0
which is not an indeterminate form. We conclude then that
lim x e1/x = 0.
x0
3. This limit ini ally evaluates to the indeterminate form . By applying

a logarithmic rule, we can rewrite the limit as
(
)
x+1
lim ln(x + 1) ln x = lim ln
.
x
x
x
As x , the argument of the ln term approaches /, to which we
can apply lHpitals Rule.
lim
Notes:
316
x+1
x
by LHR
1
= 1.
1
6.7 LHpitals Rule

x+1
1, it follows that
x
(
)
x+1
x implies ln
ln 1 = 0.
x
Since x implies
Thus
lim ln(x + 1) ln x = lim ln
x+1
x
)
= 0.
Interpreta on: since this limit evaluates to 0, it means that for large x,
there is essen ally no dierence between ln(x + 1) and ln x; their dierence is essen ally 0.
4. The limit lim x2 ex ini ally returns the indeterminate form . We
x
(
)
ex
x
2
2 2
can rewrite the expression by factoring out x ; x e = x 1 2 .
x
We need to evaluate how ex /x2 behaves as x :
ex
x x2
lim
by LHR
ex
x 2x
lim
by LHR
ex
= .
x 2
lim
Thus limx x2 (1 ex /x2 ) evaluates to (), which is not an indeterminate form; rather, () evaluates to . We conclude that
lim x2 ex = .
x
Interpreta on: as x gets large, the dierence between x2 and ex grows

very large.
Indeterminate Forms 00 , 1 and 0

When faced with an indeterminate form that involves a power, it o en helps
to employ the natural logarithmic func on. The following Key Idea expresses the
concept, which is followed by an example that demonstrates its use.
Key Idea 20
Evalua ng Limits Involving Indeterminate Forms

00 , 1 and 0
.
(
)
If lim ln f(x) = L, then lim f(x) = lim eln(f(x)) = e L .
xc
xc
xc
Notes:
317

. Example 190
.Using lHpitals Rule with indeterminate forms involving
exponents
)x
(
1
2. lim xx .
1. lim 1 +
x
x
x0+
S
1. This equivalent to a special limit given in Theorem 3; these limits have

important applica ons within mathema cs and nance. Note that the
exponent approaches while the base approaches 1, leading to the inx
determinate form 1 . Let f(x) = (1+1/x)
(
) ; the problem asks to evaluate
lim f(x). Lets rst evaluate lim ln f(x) .
x
(
)x
(
)
1
lim ln f(x) = lim ln 1 +
x
x
x
(
)
1
= lim x ln 1 +
x
x
(
)
1
ln 1 + x
= lim
x
1/x
This produces the indeterminate form 0/0, so we apply lHpitals Rule.
= lim
1
1+1/x
(1/x2 )
(1/x2 )
1
= lim
x 1 + 1/x
= 1.
x
(
)
Thus lim ln f(x) = 1. We return to the original limit and apply Key Idea
x
20.
(
lim
1+
1
x
)x
= lim f(x) = lim eln(f(x)) = e1 = e.
x
.
2. This limit leads to the indeterminate form 00 . Let f(x) = xx and consider
Notes:
318
6.7 LHpitals Rule

(
)
rst lim ln f(x) .
x0+
(
)
lim ln f(x) = lim ln (xx )
x0+
x0+
= lim x ln x
ln x
.
1/x
x0+
= lim
x0+
This produces the indeterminate form / so we apply lHpitals Rule.

1/x
1/x2
= lim x
= lim
x0+
f(x) = xx
x
1
Figure 6.17: A graph of f(x) = xx supporting the fact that as x 0+ , f(x) 1.
x0+
= 0.
(
)
Thus lim ln f(x) = 0. We return to the original limit and apply Key Idea
20.
x0+
lim xx = lim f(x) = lim eln(f(x)) = e0 = 1.
x0+
x0+
x0+
This result is supported by the graph of f(x) = xx given in Figure 6.17.
Our brief revisit of limits will be rewarded in the next sec on where we consider improper integra on. So far, we have only
considered denite integrals
1
where the bounds are nite numbers, such as
f(x) dx. Improper integra on

0
considers integrals where one, or both, of the bounds are innity. Such integrals have many uses and applica ons, in addi on to genera ng ideas that are
enlightening.
Notes:
319
Exercises 6.7
Terms and Concepts
24. lim
x3
1. List the dierent indeterminate forms described in this secon.

2. T/F: lHpitals Rule provides a faster method of compu ng
deriva ves.
[
]
f (x)
d f(x)
3. T/F: lHpitals Rule states that
= .
dx g(x)
g (x)
4. Explain what the indeterminate form 1 means.
f(x)
when
5. Fill in the blanks: The Quo ent Rule is applied to
g(x)
taking
; lHpitals Rule is applied when taking
certain
.
0
6. Create (but do not evaluate!) a limit that returns .
x3 5x2 + 3x + 9
x3 7x2 + 15x 9
x3 + 4x2 + 4x
x3 + 7x2 + 16x + 12
ln x
26. lim
x x
ln(x2 )
27. lim
x
x
(
)2
ln x
28. lim
x
x
29. lim x ln x
25. lim
x2
x0+
30. lim
x0+
x ln x
7. Create a func on f(x) such that lim f(x) returns 00 .
31. lim xe1/x
Problems
32. lim x3 x2
x
33. lim x ln x
x1
x0+

8.
9.
10.
11.
x2 + x 2
lim
x1
x1
2
x +x6
lim
x2 x2 7x + 10
sin x
lim
x x
sin x cos x
lim
x/4
cos(2x)
34.
lim xex
35. lim
x0+
1 1/x
e
x2
36. lim (1 + x)1/x

x0+
37. lim (2x)x

x0+
38. lim (2/x)x

x0+
sin(5x)
x0
x
sin(2x)
13. lim
x0 x + 2
sin(2x)
14. lim
x0 sin(3x)
39. lim (sin x)x
sin(ax)
15. lim
x0 sin(bx)
43. lim (ln x)1x
e 1
x2
x
e x1
17. lim
x2
x0+
x sin x
18. lim 3
2
x0+ x x
44. lim (1 + x)1/x
12. lim
16. lim
x0+
19. lim
x
ex
x
ex
ex
21. lim
x
x
x
e
22. lim x
x 2
ex
23. lim x
x 3
20. lim
x0+
Hint: use the Squeeze Theorem.
40. lim (1 x)1x

x1+
41. lim (x)1/x

x
42. lim (1/x)x

x
x1+
x
45. lim (1 + x2 )1/x

x
46.
47.
lim tan x cos x
x/2
lim tan x sin(2x)
x/2
1
1
ln x
x1
5
x
49. lim 2
x3
x3+ x 9
50. lim x tan(1/x)
48. lim
x1+
(ln x)3
x
x
2
x +x2
52. lim
x1
ln x
51. lim
6.8 Improper Integra on
6.8
Improper Integra on
We begin this sec on by considering the following denite integrals:

100
1
dx 1.5608,
1 + x2
0
1000
1
dx 1.5698,
1
+
x2
0
10,000
1
dx 1.5707.
1 + x2
0
No ce how the integrand is 1/(1 + x2 ) in each integral (which is sketched in
Figure 6.18). As the upper bound gets larger, one would expect the area under
the curve would also grow. While the denite integrals do increase in value as
the upper bound grows, they are not increasing by much. In fact, consider:
b
0
b
1
1
dx
=
tan
x
= tan1 b tan1 0 = tan1 b.
1 + x2
0
y
1
0.5
x
5
Figure 6.18: Graphing f(x) =
10
1
.
1 + x2
As b , tan1 b /2. Therefore it seems that as the upper bound b grows,

b
1
the value of the denite integral
dx approaches /2 1.5708. This
1
+
x2
0
should strike the reader as being a bit amazing: even though the curve extends
to innity, it has a nite amount of area underneath it.
b
When we dened the denite integral
f(x) dx, we made two s pula ons:
a
1. The interval over which we integrated, [a, b], was a nite interval, and
2. The func on f(x) was con nuous on [a, b] (ensuring that the range of f
was nite).
In this sec on we consider integrals where one or both of the above condions do not hold. Such integrals are called improper integrals.
Notes:
321
Improper Integrals with Innite Bounds

.
Deni on 24
Improper Integrals with Innite Bounds; Coverge, Diverge
1. Let f be a con nuous func on on [a, ). Dene
f(x) dx
to be
lim
f(x) dx.
a
2. Let f be a con nuous func on on (, b]. Dene
f(x) dx
to be
lim
f(x) dx.
a
3. Let f be a con nuous func on on (, ). Let c be any real number; dene
f(x) dx
to be
lim
f(x) dx + lim
f(x) dx.
c
An improper integral is said to converge if its corresponding limit exists; otherwise, it diverges. The improper integral in part 3 converges if and only if both of its limits exist.
. Example 191
.Evalua ng improper integrals
Evaluate the following improper integrals.
1.
1
f(x) =
1
x2
2.
1
0.5
1
dx
x2
3.
1
dx
x
4.
ex dx
1
dx
1 + x2
x
1
10
1.
1

ample 191.
1
x2
in Ex-
1
dx = lim
b
x2
1
dx
2
x
1
1 b
= lim

b x 1
1
= lim
+1
b b
= 1.
A graph of the area dened by this integral is given in Figure 6.19.

Notes:
322
2.
1
1
dx = lim
b
x
1
dx
1 x
b

= lim ln |x|
b
y
1
1
x
f(x) =
= lim ln(b)
b
0.5
= .
1
dx diverges.
x
1
Compare the graphs in Figures 6.19 and 6.20; no ce how the graph of
f(x) = 1/x is no ceably larger. This dierence is enough to cause the
improper integral to diverge.
0
0
x
3.
e dx = lim
ex dx
The limit does not exist, hence the improper integral
x
5
10
1
x

ple 191.
in Exam-
0

= lim ex
a
= lim e ea
0
= 1.
f(x) = e

4. We will need to break this into two improper integrals and choose a value
of c as in part 3 of Deni on 24. Any value of c is ne; we choose c = 0.
b
1
1
dx
+
lim
dx
2
b
1
+
x
1
+
x2
a
0
0
b

= lim tan1 x + lim tan1 x
a
b
a
0
(
)
(
)
= lim tan1 0 tan1 a + lim tan1 b tan1 0
a
b
) (
(
)
+
0 .
= 0
2
2
1
dx = lim
a
1 + x2
. 10
Figure 6.21: A graph of f(x) = ex in Example 191.
y
1
Each limit exists, hence the original integral converges and has value:
f(x) =
1
1 + x2
= .
.
. 10
Notes:
x
5

ample 191.
10
1
1+x2
in Ex-
323

The previous sec on introduced lHpitals Rule, a method of evalua ng limits that return indeterminate forms. It is not uncommon for the limits resul ng
from improper integrals to need this rule as demonstrated next.
0.4
f(x) =
ln x
x2
. Example 192
Improper integra
on and lHpitals Rule
ln x
Evaluate the improper integral
dx.
x2
1
0.2

ample 192.
10
ln x
x2
This integral will require the use of Integra on by Parts. Let

S
u = ln x and dv = 1/x2 dx. Then
in Ex-
ln x
dx = lim
b
x2
ln x
dx
x2
(1
)
b
ln x b
1
= lim
dx
+
2
b
x 1
1 x
)b
(
ln x 1
= lim
b
x
x 1
(
)
ln b 1
= lim
( ln 1 1) .
b
b
b
The 1/b and ln 1 terms go to 0, leaving lim

lim
ln b
with lHpitals Rule. We have:
b
ln b
b b
lim
by LHR
lim
ln b
+ 1. We need to evaluate
b
1/b
1
= 0.
Thus the improper integral evaluates as:
ln x
dx = 1.
x2
Improper Integrals with Innite Range

We have just considered denite integrals where the interval of integra on
was innite. We now consider another type of improper integra on, where the
range of the integrand is innite.
Notes:
324
Deni on 25
Improper Integra on with Innite Range
Let f(x) be a con nuous func on on [a, b] except at c, a c b, where

x = c is a ver cal asymptote of f. Dene
b
tc
f(x) dx = lim
f(x) dx + lim
tc+
Note: In Deni on 25, c can be one of the

endpoints (a or b). In that case, there is
only one limit to consider as part of the
deni on as the other is 0.
f(x) dx.
t
. Example 193
.Improper integra on of func ons with innite range
Evaluate the following improper integrals:
1
1
1
1
1.
dx
2.
dx.
2
x
x
0
1
y
10
1. A graph of f(x) = 1/ x is given in Figure 6.24. No ce that f has a ver cal

asymptote at x = 0; in some sense, we are trying to compute the area of
a region that has no top. Could this have a nite value?
x
1
0.5
1
dx
x
a
1
= lim 2 x
a
a0+
(
)
1 a
= lim 2
1
dx = lim
a0+
x
1
f(x) =
x

ample 193.
in Ex-
a0+
= 2.
10
It turns out that the region does have a nite area even though it has no
upper bound (strange things can occur in mathema cs when considering
the innite).
2. The func on f(x) = 1/x2 has a ver cal asymptote at x = 0, as shown
in Figure 6.25, so this integral is an improper integral. Lets eschew using
limits for a moment and proceed without recognizing the improper nature
of the integral. This leads to:
1
1 1
dx
=

x2
x 1
= 1 (1)
= 2!
f(x) =
1
x2
.
1
x
0.5
0.5

ample 193.
1
x2
in Ex-
Notes:
325

Clearly the area in ques on is above the x-axis, yet the area is supposedly
nega ve! Why does our answer not match our intui on? To answer this,
evaluate the integral using Deni on 25.
1
t
1
1
1
1
dx
=
lim
dx
+
lim
dx
2
2
2
+
t0
t0
1 x
1 x
t x
1 t
1 1
= lim + lim
x 1 t0+ x t
t0
1
1
= lim 1 + lim 1 +
t
t
t0
t0+
(
)
(
)
1 + 1+ .
Neither limit converges hence the original improper integral diverges. The
nonsensical answer we obtained by ignoring the improper nature of the
integral is just that: nonsensical.
Understanding Convergence and Divergence

O en mes we are interested in knowing simply whether or not an improper
integral converges, and not necessarily the value of a convergent integral. We
provide here several tools that help determine the convergence or divergence
of improper integrals without integra ng.
1
Our rst tool is to understand the behavior of func ons of the form p .
x
. Example 194
.Improper integra on of 1/xp
1
Determine the values of p for which
dx converges.
xp
1
y
f(x) =
1
xq
p<1<q
We begin by integra ng and then evalua ng the limit.

b
1
1
dx = lim
dx
p
p
b
x
x
1
1
b
= lim
xp dx
(assume p = 1)
b
b
1

xp+1
b p + 1
1
)
1 ( 1p
b
11p .
= lim
b 1 p
1
f(x) = p
x
= lim
x
1
Figure 6.26: Plo ng func ons of the form

1/x p in Example 194.
When does this limit converge i.e., when is this limit not ? This limit converges precisely when the power of b is less than 0: when 1 p < 0 1 < p.
Notes:
326
1
dx converges. When p < 1
xp
1
the improper integral diverges; we showed in Example 191 that when p = 1 the
integral also diverges.
Figure 6.26 graphs y = 1/x with a dashed line, along with graphs of y = 1/xp ,
p < 1, and y = 1/xq , q > 1. Somehow the dashed line forms a dividing line
between convergence and divergence. .
Our analysis shows that if p > 1, then
The result of Example 194 provides an important tool in determining the

convergence of other integrals. A similar result is proved in the exercises about
1
1
improper integrals of the form
dx. These results are summarized in the
p
0 x
following Key Idea.
Key Idea 21
Convergence of Improper Integrals

1
1. The improper integral
1
1
2. The improper integral

0
1
dx and
xp
1
0
1
dx.
xp
1
dx converges when p > 1 and diverges when p 1.
xp
1
dx converges when p < 1 and diverges when p 1.
xp
A basic technique in determining convergence of improper integrals is to

compare an integrand whose convergence is unknown to an integrand whose
convergence is known. We o en use integrands of the form 1/x p to compare
to as their convergence on certain intervals is known. This is described in the
following theorem.
Theorem 51
Note: We used the upper and lower

bound of 1 in Key Idea 21 for convenience. It can be replaced by any a where
a > 0.
Direct Comparison Test for Improper Integrals
Let f and g be con nuous on [a, ) where 0 f(x) g(x) for all x in
[a, ).

1. If
g(x) dx converges, then
f(x) dx converges.
2. If
f(x) dx diverges, then

a
g(x) dx diverges.
a
Notes:
327

. Example 195
Determining convergence of improper integrals
Determine
the
convergence
of the

following improper integrals.
2
1
1.
ex dx
2.
dx
2
x x
1
3
y
1
f(x) =
0.5
1
x2
x2
f(x) = e
1. The func on f(x) = ex does not have an an deriva ve expressible in

terms of elementary func ons, so we cannot integrate directly. It is com2
parable to g(x) = 1/x2 , and as demonstrated in Figure 6.27, ex < 1/x2
1
on [1, ). We know from Key Idea 21 that
dx converges, hence
2
x
1
x
1
3
2
Figure 6.27: Graphs of f(x) = ex and

f(x) = 1/x2 in Example 195.
ex dx also converges.
1
1
1
2. Note that for large values of x,
= . We know from Key
2
2
x
x x
x
1
Idea 21 and the subsequent note that
dx diverges, so we seek to
x
3
compare the original integrand to 1/x.
It is easy to see that when x > 0, we have x = x2 > x2 x. Taking

reciprocals reverses the inequality, giving
y
f(x) =
0.4
1
x2 x
1
1
<
.
2
x
x x
1
f(x) =
x
0.2
Using Theorem 51, we conclude that since
Figure 6.28: Graphs of f(x) = 1/ x2 x

and f(x) = 1/x in Example 195.
diverges as well. Figure 6.28 illustrates this.
1
dx diverges,
x
dx
2
x x
.
Being able to compare unknown integrals to known integrals is very useful in determining convergence. However, some of our examples were a li le
1
1
too nice. For instance, it was convenient that <
, but what if the
x
x2 x
x were replaced
with a +2x + 5? That is, what can we say about the con
1
1
1
vergence of
dx? We have >
, so we cannot
2
2
x
x + 2x + 5
x + 2x + 5
3
use Theorem 51.
In cases like this (and many more) it is useful to employ the following theorem.
Notes:
328
Theorem 52
Limit Comparison Test for Improper Integrals
Let f and g be con nuous func ons on [a, ) where f(x) > 0 and g(x) > 0
for all x. If
f(x)
lim
= L,
0 < L < ,
x g(x)
then
f(x) dx
and
g(x) dx
a
either both converge or both diverge.

. Example 196
Determining
of improper integrals
convergence
Determine the convergence of

dx.
x2 + 2x + 5
3
As x gets large, the quadra c inside the square root func on
1
1
will begin to behave much like y = x. So we compare
to with
2
x
x + 2x + 5
the Limit Comparison Test:
1/ x2 + 2x + 5
x
lim
.
= lim
x
x
1/x
x2 + 2x + 5
The immediate evalua on of this limit returns /, an indeterminate form.
Using lHpitals Rule seems appropriate, but in this situa on, it does not lead
to useful results. (We encourage the reader to employ lHpitals Rule at least
once to verify this.)
The trouble is the square root func on. To get rid of it, we employ the following fact: If lim f(x) = L, then lim f(x)2 = L2 . (This is true when either c or L
S
xc
y
f(x) =
xc
is .) So we consider now the limit
x2
.
x x2 + 2x + 5
This converges to 1, meaning the original limit also converged to 1. As x gets very
1
1
large, the func on
looks very much like . Since we know that
2
x
x + 2x + 5

1
1
dx diverges, by the Limit Comparison Test we know that

dx
2 + 2x + 5
x
x
3
3
also diverges. Figure 6.29 graphs f(x) = 1/ x2 + 2x + 5 and f(x) = 1/x, illustra ng that as x gets large, the func ons become indis nguishable. .
1
x2 + 2x + 5
0.2
lim
1
x
f(x) =
x
5
10
15
20
.
Figure 6.29: Graphing f(x) = 2 1
x +2x+5
and f(x) =
1
x
in Example 196.
Both the Direct and Limit Comparison Tests were given in terms of integrals
over an innite interval. There are versions that apply to improper integrals with
an innite range, but as they are a bit wordy and a li le more dicult to employ,
they are omi ed from this text.
Notes:
329
Exercises 6.8
Terms and Concepts
20.
1
1. The denite integral was dened with what two s pulaons?

b
f(x) dx is
2. If lim
f(x) dx exists, then the integral
b
4. For what values of p will
10
1

0
1
dx converge?
xp
1
2
23.
24.
In Exercises 7 33, evaluate the given improper integral.

7.
e52x dx
26.
x4 dx
9.
12.
( )x
1
dx
2
2
2
1
2
2
18.
1
1
19.
1
x2
x ln x dx
27.
0
28.
ln x
dx
x
1
dx
(x 1)2
1
dx
x1
1
dx
x1
1
dx
x
ln x dx
29.
0
ln x
dx
x2
ln x
dx
x
30.
1
x
dx
+4
1
dx
(x 1)2
17.
x
dx
x2 + 1
16.
1
dx
x2 + 9
15.
1
dx
ex + ex
2x dx
14.
13.
11.
xex dx
10.
1
dx
x3
8.
xex dx
25.
xex dx
Problems
1
dx
|x|
22.
1
dx converge?
xp
1
dx converge?
xp
sec2 x dx
said to
.

3. If
f(x) dx = 10, and 0 g(x) f(x) for all x, then we
1

know that
.
g(x) dx
21.
1
dx
x2
31.
1
32.
ex sin x dx
33.
0
ex cos x dx
In Exercises 34 43, use the Direct Comparison Test or the

Limit Comparison Test to determine whether the given definite integral converges or diverges. Clearly state what test
is being used and what func on the integrand is being compared to.

3
34.
dx
3x2 + 2x 5
10

4
35.
dx
7x3 x
2

x+3
36.
dx
x3 x2 + x + 1
0

37.
ex ln x dx
1
38.
39.
1
dx
x + ex
1
dx
ex x
x
dx
ex
x
dx
x2 + cos x
42.
43.
dx
1
dx
x2 + sin x
41.
+3x+1
40.
ex
7: A
We begin this chapter with a reminder of a few key concepts from Chapter 5. Let f be a con nuous
func on on [a, b] which is par oned into n subintervals as
a < x1 < x2 < < xn < xn+1 = b.
Let xi denote the length of the ith subinterval, and let ci be any x-value in that subinterval. Deni on
21 states that the sum
n
f(ci )xi
i=1
is a Riemann Sum. Riemann Sums are o en used to approximate some quan ty (area, volume, work,
pressure, etc.). The approxima on becomes exact by taking the limit
lim
||xi ||0
f(ci )xi ,
i=1
where ||xi || the length of the largest subinterval in the par

Riemann Sums to denite integrals:
lim
||xi ||0
on. Theorem 38 connects limits of
f(ci )xi =
f(x) dx.
a
i=1
Finally, the Fundamental Theorem of Calculus states how denite integrals can be evaluated using
an deriva ves.
This chapter employs the following technique to a variety of applica ons. Suppose the value Q of a
quan ty is to be calculated. We rst approximate the value of Q using a Riemann Sum, then nd the
exact value via a denite integral. We spell out this technique in the following Key Idea.
Key Idea 22
Applica on of Denite Integrals Strategy
Let a quan ty be given whose value Q is to be computed.

1. Divide the quan ty into n smaller subquan
es of value Qi .
2. Iden fy a variable x and func on f(x) such that each subquan ty can be approximated with
the product f(ci )xi , where xi represents a small change in x. Thus Qi f(ci )xi . A
sample approxima on f(ci )xi of Qi is called a dieren al element.
3. Recognize that Q =
i=1
Qi
f(ci )xi , which is a Riemann Sum.
i=1
4. Taking the appropriate limit gives Q =
f(x) dx
a
This Key Idea will make more sense a er we have had a chance to use it
several mes. We begin Area Between Curves, which we addressed briey in
Sec on 5.5.4.
Chapter 7 Applica ons of Integra on
7.1 Area Between Curves

y
f(x)
g(x)
(a)
y
f(x)
g(x)
We are o en interested in knowing the area of a region. Forget momentarily

that we addressed this already in Sec on 5.5.4 and approach it instead using
the technique described in Key Idea 22.
Let Q be the area of a region bounded by con nuous func ons f and g. If we
break the region into many subregions, we have an obvious equa on:
Total Area = sum of the areas of the subregions.
The issue to address next is how to systema cally break a region into subregions. A graph will help. Consider Figure 7.1 (a) where a region between two
curves is shaded. While there are many ways to break this into subregions, one
par cularly ecient way is to slice it ver cally, as shown in Figure 7.1 (b).
We now approximate the area of a slice. Again, we have many op ons, but
using a rectangle seems simplest. Picking any x-value ci in the i th slice, we set
the height of the rectangle to be f(ci ) g(ci ), the dierence of the corresponding y-values. The width of the rectangle is a small dierence in x-values, which
we represent with xi . Figure 7.1 (c) shows sample points ci chosen in each
subinterval and appropriate rectangles drawn. (Each of these rectangles represents
a dieren
al element.) Each slice has an area approximately equal to
(
)
f(ci ) g(ci ) xi ; hence, the total area is approximately the Riemann Sum
Q=
(
i=1
(b)
Taking the limit as ||xi || 0 gives the exact area as
Theorem 53
f(x)
)
f(x) g(x) dx.
Area Between Curves (restatement of Theorem 41)
x
a
b(
Let f(x) and g(x) be con nuous func ons dened on [a, b] where f(x)
g(x) for all x in [a, b]. The area of the region bounded by the curves
y = f(x), y = g(x) and the lines x = a and x = b is
g(x)
)
f(ci ) g(ci ) xi .
)
f(x) g(x) dx.
(c)
Figure 7.1: Subdividing a region into vercal slices and approxima ng the areas
with rectangles.
. Example 197
.Finding area enclosed by curves
Find the area of the region bounded by f(x) = sin x + 2, g(x) =
x = 0 and x = 4, as shown in Figure 7.2.
1
2
cos(2x) 1,
The graph veries that the upper boundary of the region is

S
given by f and the lower bound is given by g. Therefore the area of the region is
Notes:
334

the value of the integral
4
0
(
)
(1
))
cos(2x) 1 dx
f(x) g(x) dx =
sin x + 2
2
0
4
1

= cos x sin(2x) + 3x
4
0
= 12 37.7 units2 .
4
f(x)
. Example 198
Finding total area enclosed by curves
Find the total area of the region enclosed by the func ons f(x) = 2x + 5 and
g(x) = x3 7x2 + 12x 3 as shown in Figure 7.3.
10
g(x)
2 .
Figure 7.2: Graphing an enclosed region

in Example 197.
A quick calcula on shows

4 that f = g at x = 1, 2 and 4. One
(
)
can proceed thoughtlessly by compu ng
f(x) g(x) dx, but this ignores
S
the fact that on [1, 2], g(x) > f(x). (In fact, the thoughtless integra on returns
9/4, hardly the expected value of an area.) Thus we compute the total area by
breaking the interval [1, 4] into two subintervals, [1, 2] and [2, 4] and using the
proper integrand in each.
Total Area =
)
g(x) f(x) dx +
)
x3 7x2 + 14x 8 dx +
)
f(x) g(x) dx
2
2
)
x3 + 7x2 14x + 8 dx
= 5/12 + 8/3
= 37/12 = 3.083 units2 .
.
The previous example makes note that we are expec ng area to be posi ve.
When rst learning about the denite integral, we interpreted it as signed area
under the curve, allowing for nega ve area. That doesnt apply here; area is
to be posi ve.
The previous example also demonstrates that we o en have to break a given
region into subregions before applying Theorem 53. The following example
shows another situa on where this is applicable, along with an alternate view
of applying the Theorem.
. Example 199
.Finding area: integra ng with respect to
y
Find the area of the region enclosed by the func ons y = x + 2, y = (x
1)2 + 3 and y = 2, as shown in Figure 7.4.
x
1
2
4
.
.
Figure 7.3: Graphing a region enclosed by
two func ons in Example 198.
y
y=
y = (x 1)2 + 3
x+2
Notes:
x
1
Figure 7.4: Graphing a region for Example

199.
335

y
x=
x = (y 2)
We give two approaches to this problem. In the rst apS

proach, we no ce that the regionstop is dened by two dierent curves.
On [0, 1], the top func on is y = x + 2; on [1, 2], the top func on is y =
(x 1)2 + 3. Thus we compute the area as the sum of two integrals:
1(
2(
)
)
(
)
(
)
Total Area =
x + 2 2 dx +
(x 1)2 + 3 2 dx
3y+1
= 2/3 + 2/3
.
= 4/3
x
1
Figure 7.5: The region used in Example

199 with boundaries relabeled as funcons of y.
The second approach is clever and very useful in certain situa ons. We are
used to viewing curves as func ons of x; we input an x-value and a y-value is returned. Some curves can also be described as func ons of y: input a y-value and
an x-value is returned. We can rewrite the equa ons describing the boundary
by solving for x:
y = x + 2 x = (y 2)2
y = (x 1)2 + 3 x = 3 y + 1.
Figure 7.5 shows the region with the boundaries relabeled. A dieren al
element, a horizontal rectangle, is also pictured. The width of the rectangle is
a small change in y: y. The height of the rectangle is a dierence in x-values.
The top x-value is the largest value, i.e., the rightmost. The bo om x-value
is the smaller, i.e., the le most. Therefore the height of the rectangle is
(
)
3 y + 1 (y 2)2 .
The area is found by integra ng the above func on with respect to y with
the appropriate bounds. We determine these by considering the y-values the
region occupies. It is bounded below by y = 2, and bounded above by y = 3.
That is, both the top and bo om func ons exist on the y interval [2, 3]. Thus
3
(
)
Total Area =
3 y + 1 (y 2)2 dy
2
)3
2
1

(3 y)3/2 + y (y 2)3
3
3
2
= 4/3.
=
This calculusbased technique of nding area can be useful even with shapes
that we normally think of as easy. Example 200 computes the area of a triangle. While the formula 12 base height is well known, in arbitrary triangles
it can be nontrivial to compute the height. Calculus makes the problem simple.
Notes:
336

. Example 200
Finding the area of a triangle
Compute the area of the regions bounded by the lines
y = x + 1, y = 2x + 7 and y = 12 x + 52 , as shown in Figure 7.6.
y
3
y = 2x + 7
y=x+1
Recognize that there are two top func ons to this region,
causing us to use two denite integrals.
2
3
(
(
1
1
5 )
5 )
(x + 1) ( x + ) dx +
(2x + 7) ( x + ) dx
Total Area =
2
2
2
2
1
2
= 3/4 + 3/4
= 3/2.
S
We can also approach this by conver ng each func on into a func on of y. This
also requires 2 integrals, so there isnt really any advantage to doing so. We do
it here for demonstra on purposes.
The top func on is always x = 7y
2 while there are two bo om funcons. Being mindful of the proper integra on bounds, we have
2
3
(7 y
)
(7 y
)
Total Area =
(5 2y) dy +
(y 1) dy
2
2
1
2
= 3/4 + 3/4
y = 12 x +
5
2
x
1
Figure 7.6: Graphing a triangular region in

Example 200.
= 3/2.
Of course, the nal answer is the same. (It is interes ng to note that the area of
all 4 subregions used is 3/4. This is coincidental.) .
(a)
y
8
7
6
2.76
5.21
6.35
5
5.08
. Example 201
.Numerically approxima ng area
To approximate the area of a lake, shown in Figure 7.7 (a), the length of the
lake is measured at 200-foot increments as shown in Figure 7.7 (b), where the
lengths are given in hundreds of feet. Approximate the area of the lake.
2.25
While we have focused on producing exact answers, we are also able make
approxima ons using the principle of Theorem 53. The integrand in the theorem is a distance (top minus bo om); integra ng this distance func on gives
an area. By taking discrete measurements of distance, we can approximate an
area using Numerical Integra on techniques developed in Sec on 5.5. The following example demonstrates this.
x
1 2 3 4 5 6 7 8 9 10 11 12
The measurements of length can be viewed as measuring

S
top
minus
bo
om
of two func ons. The exact answer is found by integra ng
12
(
)
f(x) g(x) dx, but of course we dont know the func ons f and g. Our
0
(b)
Figure 7.7: (a) A sketch of a lake, and (b)
the lake with length measurements.
discrete measurements instead allow us to approximate.
Notes:
337

We have the following data points:
(0, 0), (2, 2.25), (4, 5.08), (6, 6.35), (8, 5.21), (10, 2.76), (12, 0).
We also have that x =
ba
n
= 2, so Simpsons Rule gives
)
2(
1 0 + 4 2.25 + 2 5.08 + 4 6.35 + 2 5.21 + 4 2.76 + 1 0
3
= 44.013 units2 .
Area
Since the measurements are in hundreds of feet, units2 = (100 )2 =

10, 000 2 , giving a total area of 440, 133 2 . (Since we are approxima ng, wed
likely say the area was about 440, 000 2 , which is a li le more than 10 acres.) .
Notes:
338
Exercises 7.1
Terms and Concepts
y
y = sin x + 1
1. T/F: The area between curves is always posi ve.

2. T/F: Calculus can be used to nd the area of basic geometric
shapes.
7.
1
y = sin x
x
3. In your own words, describe how to nd the total area enclosed by y = f(x) and y = g(x).
/2
.
y
y = sec2 x
Problems
In Exercises 4 10, nd the area of the shaded region in the
given graph.
8.
1
y = sin(4x)
y
y=
1
2x
.
.
+3
/8
/4
y
1
y = sin x
4.
0.5
2
y=
1
2
cos x + 1
9.
x
/4
0.5
1
y
4
10.
.
y = 4x
y = 2x
1
y = x2 + x 1
6.
y=2
y=1
/2
5/4
y = cos x
x
1
3/4
3
y = 3x3 + 3x + 2
5.
/2
x
0.5
In Exercises 11 16, nd the total area enclosed by the funcons f and g.
y = x1/3
11. f(x) = 2x2 + 5x 3, g(x) = x2 + 4x 1

12. f(x) = x2 3x + 2, g(x) = 3x + 3
13. f(x) = sin x, g(x) = 2x/
22.
0.5
14. f(x) = x 4x + x 1, g(x) = x + 2x 4
15. f(x) = x, g(x) = x

3
y=
16. f(x) = x + 5x + 2x + 1, g(x) = 3x + x + 3

3
17. The func ons f(x) = cos(2x) and g(x) = sin x intersect
innitely many mes, forming an innite number of repeated, enclosed regions. Find the areas of these regions.
In Exercises 18 22, nd the area of the enclosed region in
two ways:
1. by trea ng the boundaries as func ons of x, and
2. by trea ng the boundaries as func ons of y.
2
y=
1
4 (x
3)2 + 1
0.5
In Exercises 23 26, nd the area triangle formed by the given

three points.
23. (1, 1),
(2, 3), and
24. (1, 1),
x 1/2
(3, 3)
(2, 1)
(1, 3), and
25. (1, 1),
(3, 3), and
(3, 3)
26. (0, 0),
(2, 5), and
(5, 2)
y=x +1
18.
27. Use the Trapezoidal Rule to approximate the area of the

pictured lake whose lengths, in hundreds of feet, are measured in 100-foot increments.
y=1
y
1
y = 2x + 3
5.2
19.
4.5
0.5
7.3
4.9
y=
x
1
0.5
y = 12 x
1
y
28. Use Simpsons Rule to approximate the area of the pictured

lake whose lengths, in hundreds of feet, are measured in
200-foot increments.
4
y=x+2
y = x2
20.
6.45
y
1
x = 1/2y + 1
x
21.
1
1
2
x=
1 2
2y
4.9
7.7
x
1
6.6
4.25
7.2
Volume by Cross-Sec onal Area; Disk and Washer

Methods
The volume of a general right cylinder, as shown in Figure 7.8, is

Area of the base height.
We can use this fact as the building block in nding volumes of a variety of
shapes.
Given an arbitrary solid, we can approximate its volume by cu ng it into n
thin slices. When the slices are thin, each slice can be approximated well by a
general right cylinder. Thus the volume of each slice is approximately its crosssec onal area thickness. (These slices are the dieren al elements.)
By orien ng a solid along the x-axis, we can let A(xi ) represent the crosssec onal area of the i th slice, and let xi represent the thickness of this slice (the
thickness is a small change in x). The total volume of the solid is approximately:
Volume
V=Ah
h
.
Figure 7.8: The volume of a general right

cylinder
n [
]
Area thickness
i=1
A(xi )xi .
i=1
Recognize that this is a Riemann Sum. By taking a limit (as the thickness of
the slices goes to 0) we can nd the volume exactly.
Theorem 54
Volume By Cross-Sec onal Area
The volume V of a solid, oriented along the x-axis with cross-sec onal
area A(x) from x = a to x = b, is
V=
A(x) dx.
a
y
. Example 202
.Finding the volume of a solid
Find the volume of a pyramid with a square base of side length 10 in and a height
of 5 in.
There are many ways to orient the pyramid along the xaxis; Figure 7.9 gives one such way, with the pointed top of the pyramid at the
origin and the x-axis going through the center of the base.
Each cross sec on of the pyramid is a square; this is a sample dieren al
element. To determine its area A(x), we need to determine the side lengths of
10
2x
2x.
5
10
.
Figure 7.9: Orien ng a pyramid along the
x-axis in Example 202.
Notes:
341

the square.
When x = 5, the square has side length 10; when x = 0, the square has side
length 0. Since the edges of the pyramid are lines, it is easy to gure that each
cross-sec onal square has side length 2x, giving A(x) = (2x)2 = 4x2 . Following
Theorem 54, we have
4x2 dx
V=
0
4 5
= x3
3 0
500 3
=
in 166.67 in3 .
3
We can check our work by consul ng the general equa on for the volume of a
pyramid (see the back cover under Volume of A General Cone):
1
3 area of base height.
Certainly, using this formula from geometry is faster than our new method, but
the calculus-based method can be applied to much more than just cones. .
An important special case of Theorem 54 is when the solid is a solid of revolu on, that is, when the solid is formed by rota ng a shape around an axis.
Start with a func on y = f(x) from x = a to x = b. Revolving this curve
about a horizontal axis creates a three-dimensional solid whose cross sec ons
are disks (thin circles). Let R(x) represent the radius of the cross-sec onal disk
at x; the area of this disk is R(x)2 . Applying Theorem 54 gives the Disk Method.
.
Key Idea 23
The Disk Method
Let a solid be formed by revolving the curve y = f(x) from x = a to x = b

around a horizontal axis, and let R(x) be the radius of the cross-sec onal
disk at x. The volume of the solid is
b
V=
R(x)2 dx.
. Example 203
.Finding volume using the Disk Method
Find the volume of the solid formed by revolving the curve y = 1/x, from x = 1
to x = 2, around the x-axis.
A sketch can help us understand this problem. In Figure 7.10
S
(a) the curve y = 1/x is sketched along with the dieren al element a disk
Notes:
342

y
at x with radius R(x) = 1/x. In Figure 7.10 (b) the whole solid is pictured, along
with the dieren al element.
Using Key Idea 23, we have
2 ( )2
1
V=
dx
x
1
2
1
=
dx
2
x
1
]
[
1 2
=
x 1
]
[
1
= (1)
2
= units3 .
2
.
While Key Idea 23 is given in terms of func ons of x, the principle involved
can be applied to func ons of y when the axis of rota on is ver cal, not horizontal. We demonstrate this in the next example.
. Example 204
Finding volume using the Disk Method
Find the volume of the solid formed by revolving the curve y = 1/x, from x = 1
to x = 2, about the y-axis.
Since the axis of rota on is ver cal, we need to convert the
S
func on into a func on of y and convert the x-bounds to y-bounds. Since y =
1/x denes the curve, we rewrite it as x = 1/y. The bound x = 1 corresponds to
the y-bound y = 1, and the bound x = 2 corresponds to the y-bound y = 1/2.
Thus we are rota ng the curve x = 1/y, from y = 1/2 to y = 1 about the
y-axis to form a solid. The curve and sample dieren al element are sketched
in Figure 7.11 (a), with a full sketch of the solid in Figure 7.11 (b). We integrate
to nd the volume:
1
1
V=
dy
2
1/2 y
1
=
y 1/2
= units3 .
y = 1/x
1
0.5
R(x) = 1/x
.
(a)
y
.
(b)
Figure 7.10: Sketching a solid in Example
203.
y
x = 1/y
.
2
R(y) = 1/y
.
(a)
y
.
2
x
2
.
(b)
Figure 7.11: Sketching the solid in Example 204.
Notes:
343

y
R(x)
r(x)
.a
We can also compute the volume of solids of revolu on that have a hole in
the center. The general principle is simple: compute the volume of the solid
irrespec ve of the hole, then subtract the volume of the hole. If the outside
radius of the solid is R(x) and the inside radius (dening the hole) is r(x), then
the volume is
V=
R(x)2 dx
r(x)2 dx =
(a)
)
R(x)2 r(x)2 dx.
One can generate a solid of revolu on with a hole in the middle by revolving
a region about an axis. Consider Figure 7.12 (a), where a region is sketched along
with a dashed, horizontal axis of rota on. By rota ng the region about the axis, a
solid is formed as sketched in Figure 7.12 (b). The outside of the solid has radius
R(x), whereas the inside has radius r(x). Each cross sec on of this solid will be
a washer (a disk with a hole in the center) as sketched in Figure 7.12 (c). This
leads us to the Washer Method.
.
.
(b)
Key Idea 24
The Washer Method
R(x)
r(x)
.a
Let a region bounded by y = f(x), y = g(x), x = a and x = b be rotated about a horizontal axis that does not intersect the region, forming
a solid. Each cross sec on at x will be a washer with outside radius R(x)
and inside radius r(x). The volume of the solid is
V=
)
R(x)2 r(x)2 dx.
.
(c)
Figure 7.12:
Method.
Establishing the Washer
Even though we introduced it rst, the Disk Method is just a special case of
the Washer Method with an inside radius of r(x) = 0.
. Example 205
.Finding volume with the Washer Method
Find the volume of the solid formed by rota ng the region bounded by y =
x2 2x + 2 and y = 2x 1 about the x-axis.
.1
A sketch of the region will help, as given in Figure 7.13. RoS

ta ng about the x-axis will produce cross sec ons in the shape of washers, as
shown in Figure 7.14 (a); the complete solid is shown in part (b). The outside
radius of this washer is R(x) = 2x + 1; the inside radius is r(x) = x2 2x + 4. As
the region is bounded from x = 1 to x = 3, we integrate as follows to compute
.
Figure 7.13: A sketch of the region used
in Example 205.
344
Notes:

y
the volume.
V=
)
(2x 1)2 (x2 2x + 2)2 dx
.1
=
[
)
x4 + 4x3 4x2 + 4x 3 dx
]3
1
4

= x5 + x4 x3 + 2x2 3x
5
3
1
104
3
21.78 units .
=
15
.
(a)
y
When rota ng about a ver cal axis, the outside and inside radius func ons
must be func ons of y.
. Example 206
Finding volume with the Washer Method
Find the volume of the solid formed by rota ng the triangular region with verces at (1, 1), (2, 1) and (2, 3) about the y-axis.
The triangular region is sketched in Figure 7.15 (a); the difS
feren al element is sketched in (b) and the full solid is drawn in (c). They help us
establish the outside and inside radii. Since the axis of rota on is ver cal, each
radius is a func on of y.
The outside radius R(y) is formed by the line connec ng (2, 1) and (2, 3); it
is a constant func on, as regardless of the y-value the distance from the line to
the axis of rota on is 2. Thus R(y) = 2.
The inside radius is formed by the line connec ng (1, 1) and (2, 3). The equaon of this line is y = 2x 1, but we need to refer to it as a func on of y. Solving
for x gives r(y) = 21 (y + 1).
We integrate over the y-bounds of y = 1 to y = 3. Thus the volume is
3(
(1
)2 )
V=
22 (y + 1)
dy
2
1
3(
1
1
15 )
=
y2 y +
dy
4
2
4
1
[
1
1
15 ]3
= y3 y2 + y
12
4
4
1
10
=
10.47 units3 .
3
.
.1
.
(b)
Figure 7.14: Sketching the dieren al element and solid in Example 205.
y
r(y) =
1
2 (y
+ 1)
1
R(y) = 2
1
.
(a)
y
.
(b)
y
Notes:
.
.
(c)
Figure 7.15: Sketching the solid in Example 206.
345
Exercises 7.2
Terms and Concepts
y
1
1. T/F: A solid of revolu on is formed by revolving a shape

around an axis.
2. In your own words, explain how the Disk and Washer Methods are related.
y=
7.
3. Explain the how the units of volume are found in the in2
tegral of Theorem 54: if 3A(x) has units of in , how does
A(x) dx have units of in ?
Problems
In Exercises 4 7, a region of the Cartesian plane is shaded.
Use the Disk/Washer Method to nd the volume of the solid
of revolu on formed by revolving the region about the xaxis.
x
y=x
0.5
0.5

Use the Disk/Washer Method to nd the volume of the solid
of revolu on formed by revolving the region about the yaxis.
y
3
y = 3 x2
y
3
y = 3 x2
8.
1
4.
x
1
2
10
y
y = 5x
10
5.
y = 5x
9.
x
0.5
x
0.5
1.5
1.5
2
1
y
1
10.
6.
y = cos x
0.5
y = cos x
0.5
.
.
x
0.5
1.5
x
0.5
1.5
(Hint: Integra on By Parts will be necessary, twice. First let

u = arccos2 x, then let u = arccos x.)
In Exercises 18 21, a solid is described. Orient the solid along

the x-axis such that a cross-sec onal area func on A(x) can
be obtained, then apply Theorem 54 to nd the volume of
the solid.
1
y=
y=x
0.5
18. A right circular cone with height of 10 and base radius of 5.
0.5
In Exercises 12 17, a region of the Cartesian plane is described. Use the Disk/Washer Method to nd the volume of
the solid of revolu on formed by rota ng the region about
each of the given axes.
12. Region bounded by: y = x, y = 0 and x = 1.

Rotate about:
(a) the x-axis
10
11.
19. A skew right circular cone with height of 10 and base radius
of 5. (Hint: all cross-sec ons are circles.)
(c) the y-axis
(b) y = 1
(d) x = 1
10
13. Region bounded by: y = 4 x and y = 0.

Rotate about:
2
(a) the x-axis
(c) y = 1
(b) y = 4
(d) x = 2
14. The triangle with ver ces (1, 1), (1, 2) and (2, 1).
Rotate about:
(a) the x-axis
20. A right triangular cone with height of 10 and whose base is

a right, isosceles triangle with side length 4.
(c) the y-axis

(d) x = 1
10
(b) y = 2
15. Region bounded by y = x 2x + 2 and y = 2x 1.

Rotate about:
2
(a) the x-axis

(b) y = 1
(c) y = 5
16. Region bounded by y = 1/ x2 + 1, x = 1, x = 1 and

the x-axis.
Rotate about:
(a) the x-axis
21. A solid with length 10 with a rectangular base and triangular top, wherein one end is a square with side length 5 and
the other end is a triangle with base and height of 5.
(c) y = 1
(b) y = 1
17. Region bounded by y = 2x, y = x and x = 2.
Rotate about:
(a) the x-axis
(c) the y-axis
(b) y = 4
(d) x = 2
5
5
.
10
7.3 The Shell Method
y
1
1 + x2
y=
.
(a)
y
.
(b)
y
O en a given problem can be solved in more than one way. A par cular method
may be chosen out of convenience, personal preference, or perhaps necessity.
Ul mately, it is good to have op ons.
The previous sec on introduced the Disk and Washer Methods, which computed the volume of solids of revolu on by integra ng the cross-sec onal area
of the solid. This sec on develops another method of compu ng volume, the
Shell Method. Instead of slicing the solid perpendicular to the axis of rota on
crea ng cross-sec ons, we now slice it parallel to the axis of rota on, crea ng
shells.
Consider Figure 7.16, where the region shown in (a) rotated around the yaxis forming the solid shown in (b). A small slice of the region is drawn in (a),
parallel to the axis of rota on. When the region is rotated, this thin slice forms
a cylindrical shell, as pictured in part (c) of the gure. The previous sec on
approximated a solid with lots of thin disks (or washers); we now approximate
a solid with many thin cylindrical shells.
To compute the volume of one shell, rst consider the paper label on a soup
can with radius r and height h. What is the area of this label? A simple way of
determining this is to cut the label and lay it out at, forming a rectangle with
height h and length 2r. Thus the area is A = 2rh; see Figure 7.17 (a).
Do a similar process with a cylindrical shell, with height h, thickness x, and
approximate radius r. Cu ng the shell and laying it at forms a rectangular solid
with length 2r, height h and depth x. Thus the volume is V 2rhx; see
Figure 7.17 (b). (We say approximately since our radius was an approximaon.)
By breaking the solid into n cylindrical shells, we can approximate the volume
of the solid as
n
V=
2ri hi xi ,
i=1
.
(c)
Figure 7.16:
Method.
where ri , hi and xi are the radius, height and thickness of the i th shell, respecvely.
This is a Riemann Sum. Taking a limit as the thickness of the shells approaches 0 leads to a denite integral.
Introducing the Shell
Notes:
348
2r
(a)
V 2rhx
A = 2rh
cut here
2r
cut here
.
.
(b)
Figure 7.17: Determining the volume of a thin cylindrical shell.
Key Idea 25
The Shell Method
Let a solid be formed by revolving a region R, bounded by x = a and

x = b, around a ver cal axis. Let r(x) represent the distance from the axis
of rota on to x (i.e., the radius of a sample shell) and let h(x) represent
the height of the solid at x (i.e., the height of the shell). The volume of
the solid is
V = 2
r(x)h(x) dx.
a
Special Cases:
1. When the region R is bounded above by y = f(x) and below by y = g(x),
then h(x) = f(x) g(x).
y
2. When the axis of rota on is the y-axis (i.e., x = 0) then r(x) = x.

y=
Lets prac ce using the Shell Method.
. Example 207
.Finding volume using the Shell Method
Find the volume of the solid formed by rota ng the region bounded by y = 0,
y = 1/(1 + x2 ), x = 0 and x = 1 about the y-axis.
This is the region used to introduce the Shell Method in Figure 7.16, but is sketched again in Figure 7.18 for closer reference. A line is drawn
in the region parallel to the axis of rota on represen ng a shell that will be
S
h(x)
{z
r(x)
1
1 + x2
}x
x
1
Figure 7.18: Graphing a region in Example

207.
Notes:
349

carved out as the region is rotated about the y-axis. (This is the dieren al element.)
The distance this line is from the axis of rota on determines r(x); as the
distance from x to the y-axis is x, we have r(x) = x. The height of this line
determines h(x); the top of the line is at y = 1/(1 + x2 ), whereas the bo om
of the line is at y = 0. Thus h(x) = 1/(1 + x2 ) 0 = 1/(1 + x2 ). The region is
bounded from x = 0 to x = 1, so the volume is
V = 2
0
x
dx.
1 + x2
This requires subs tu on. Let u = 1 + x2 , so du = 2x dx. We also change the

bounds: u(0) = 1 and u(1) = 2. Thus we have:
1
du
1 u
2

= ln u
=
2x
= ln 2 2.178 units3 .
y=
Note: in order to nd this volume using the Disk Method, two integrals would
be needed to account for the regions above and below y = 1/2. .
h(x).
{z
r(x)
x
x
With the Shell Method, nothing special needs to be accounted for to compute the volume of a solid that has a hole in the middle, as demonstrated next.
(a)
. Example 208
.Finding volume using the Shell Method
Find the volume of the solid formed by rota ng the triangular region determined
by the points (0, 1), (1, 1) and (1, 3) about the line x = 3.
y
3
2
The region is sketched in Figure 7.19 (a) along with the difS
feren al element, a line within the region parallel to the axis of rota on. The
height of the dieren al element is the distance from y = 1 to y = 2x + 1, the
line that connects the points (0, 1) and (1, 3). Thus h(x) = 2x + 1 1 = 2x.
The radius of the shell formed by the dieren al element is the distance from
x to x = 3; that is, it is r(x) = 3 x. The x-bounds of the region are x = 0 to
(b)
208.
350
Notes:

x = 1, giving
V = 2
(3 x)(2x) dx
0
1
6x 2x2 ) dx
(
)
2 3 1
2
= 2 3x x
3
0
14
14.66 units3 .
=
3
= 2
.
When revolving a region around a horizontal axis, we must consider the radius and height func ons in terms of y, not x.
. Example 209
Finding volume using the Shell Method
Find the volume of the solid formed by rota ng the region given in Example 208
about the x-axis.
y
3
The region is sketched in Figure 7.20 (a) with a sample differen al element and the solid is sketched in (b). (Note that the region looks
slightly dierent than it did in the previous example as the bounds on the graph
have changed.)
The height of the dieren al element is an x-distance, between x = 12 y 21
and x = 1. Thus h(y) = 1( 12 y 12 ) = 12 y+ 32 . The radius is the distance from
y to the x-axis, so r(y) = y. The y bounds of the region are y = 1 and y = 3,
leading to the integral
S
[ (
)]
1
3
V = 2
y y+
dy
2
2
1
]
3[
1 2 3
= 2
y + y dy
2
2
[1
]
1
3
3
= 2 y3 + y2
6
4
1
[
]
9
7
= 2
4 12
10
=
10.472 units3 .
3
1
1
2
y
x=
{z
h(y)
1
r(y)
(a)
y
3
2
1
.
(b)
209.
Notes:
351

At the beginning of this sec on it was stated that it is good to have op ons.
The next example nds the volume of a solid rather easily with the Shell Method,
but using the Washer Method would be quite a chore.
. Example 210
Finding volume using the Shell Method
Find the volume of the solid formed by revolving the region bounded by y = sin x
and the x-axis from x = 0 to x = about the y-axis.
y
The region and the resul ng solid are given in Figure 7.21.
S
The radius of a sample shell is r(x) = x; the height of a sample shell is h(x) =
sin x, each from x = 0 to x = . Thus the volume of the solid is
r(x)
z }| {
.
h(x)
x sin x dx.
0
x
x
V = 2
This requires Integra on By Parts. Set u = x and dv = sin x dx; we leave it to

the reader to ll in the rest. We have:
(a)

[
]

= 2 x cos x +
cos x dx
0
] 0
[

= 2 + sin x
[
] 0
= 2 + 0
= 22 19.74 units3 .
.
(b)
210.
Note that in order to use the Washer Method, we would need to solve y =
sin x for x, requiring the use of the arcsine func on. We leave it to the reader
to verify that the outside radius func on is R(y) = arcsin y and the inside
radius func on is r(y) = arcsin y. Thus the volume can be computed as
]
( arcsin y)2 (arcsin y)2 dy.
This integral isnt terrible given that the arcsin2 y terms cancel, but it is more
onerous than the integral created by the Shell Method. .
We end this sec on with a table summarizing the usage of the Washer and
Shell Methods.
Notes:
352
Key Idea 26
Summary of the Washer and Shell Methods
Let a region R be given with x-bounds x = a and x = b and y-bounds

y = c and y = d.
Washer Method
Horizontal
Axis
Ver cal
Axis
)
R(x)2 r(x)2 dx
Shell Method
d
r(y)h(y) dy
2
c
d
c
(
)
R(y)2 r(y)2 dy
r(x)h(x) dx
a
Notes:
353
Exercises 7.3
Terms and Concepts
y
1
1. T/F: A solid of revolu on is formed by revolving a shape

around an axis.
2. T/F: The Shell Method can only be used when the Washer
Method fails.
y=
8.
x
y=x
0.5
3. T/F: The Shell Method works by integra ng crosssec onal

areas of a solid.
x
4. T/F: When nding the volume of a solid of revolu on that

was revolved around a ver cal axis, the Shell Method integrates with respect to x.
0.5

Use the Shell Method to nd the volume of the solid of revolu on formed by revolving the region about the x-axis.
Problems
Use the Shell Method to nd the volume of the solid of revolu on formed by revolving the region about the y-axis.
y
3
y
3
y = 3 x2
y = 3 x2
9.
2
5.
1
1
x
1
x
1
2
y
y
10
10
y = 5x
y = 5x
6.
10.
.
0.5
1.5
0.5
1.5
y = cos x
0.5
7.
11.
y = cos x
0.5
x
0.5
1.5
x
0.5
1.5
y
1
y=
12.
y=x
0.5
15. The triangle with ver ces (1, 1), (1, 2) and (2, 1).
Rotate about:
x
1
0.5
In Exercises 13 18, a region of the Cartesian plane is described. Use the Shell Method to nd the volume of the solid
of revolu on formed by rota ng the region about each of the
given axes.
13. Region bounded by: y = x, y = 0 and x = 1.

Rotate about:
(a) the y-axis
(c) the x-axis
(b) x = 1
(d) y = 1
14. Region bounded by: y = 4 x2 and y = 0.

Rotate about:
(a) x = 2
(b) x = 2
(c) the x-axis

(d) y = 4
(a) the y-axis
(c) the x-axis
(b) x = 1
(d) y = 2
16. Region bounded by y = x2 2x + 2 and y = 2x 1.

Rotate about:
(a) the y-axis
(c) x = 1
(b) x = 1
17. Region bounded by y = 1/ x2 + 1, x = 1 and the x and

y-axes.
Rotate about:
(a) the y-axis
(b) x = 1
18. Region bounded by y = 2x, y = x and x = 2.

Rotate about:
(a) the y-axis
(c) the x-axis
(b) x = 2
(d) y = 4
7.4 Arc Length and Surface Area
In previous sec ons we have used integra on to answer the following ques ons:
1. Given a region, what is its area?

0.5
2. Given a solid, what is its volume?

x
3
4
3
4
(a)
y
2
2
0.5
(b)
Figure 7.22: Graphing y = sin x on [0, ]
and approxima ng the curve with line
segments.
y
yi+1
In this sec on, we address a related ques on: Given a curve, what is its
length? This is o en referred to as arc length.
Consider the graph of y = sin x on [0, ] given in Figure 7.22 (a). How long is
this curve? That is, if we were to use a piece of string to exactly match the shape
of this curve, how long would the string be?
As we have done in the past, we start by approxima ng; later, we will rene
our answer using limits to get an exact solu on.
The length of straightline segments is easy to compute using the Distance
Formula. We can approximate the length of the given curve by approxima ng
the curve with straight lines and measuring their lengths.
In Figure 7.22 (b), the curve y = sin x has been approximated with 4 line
segments (the interval [0, ] has been divided into 4 equallylengthed subintervals). It is clear that these four line segments approximate y = sin x very well
on the rst and last subinterval, though not so well in the middle. Regardless,
the sum of the lengths of the line segments is 3.79, so we approximate the arc
length of y = sin x on [0, ] to be 3.79.
In general, we can approximate the arc length of y = f(x) on [a, b] in the
following manner. Let a = x1 < x2 < . . . < xn < xn+1 = b be a par on
of [a, b] into n subintervals. Let xi represent the length of the i th subinterval
[xi , xi+1 ].
Figure 7.23 zooms in on the i th subinterval where y = f(x) is approximated
by a straight line segment. The dashed lines show that we can view this line
segment as they hypotenuse of a right triangle whose sides have length xi
and yi . Using the Pythagorean Theorem, the length of this line segment is
x2i + y2i . Summing over all subintervals gives an arc length approxima on
yi
yi
x
xi
xi+1
Figure 7.23: Zooming in on the i th subinterval [xi , xi+1 ] of a par on of [a, b].
As shown here, this is not a Riemann Sum. While we could conclude that
taking a limit as the subinterval length goes to zero gives the exact arc length,
we would not be able to compute the answer with a denite integral. We need
rst to do a li le algebra.
Notes:
356
x2i + y2i .
i=1
xi

In the above expression factor out a x2i term:
(
)
n
n
y2i
2
2
xi + yi =
x2i 1 +
.
x2i
i=1
i=1
Now pull the x2i term out of the square root:
=
1+
i=1
y2i
xi .
x2i
This is nearly a Riemann Sum. Consider the y2i /x2i term. The expression
yi /xi measures the change in y/change in x, that is, the rise over run of
f on the i th subinterval. The Mean Value Theorem of Dieren a on (Theorem
27) states that there is a ci in the i th subinterval where f (ci ) = yi /xi . Thus
we can rewrite our above expression as:
=
1 + f (ci )2 xi .
i=1
This is a Riemann Sum. As long as f is con nuous, we can invoke Theorem 38

and conclude
1 + f (x)2 dx.
Key Idea 27
Arc Length
Let f be dieren able on an open interval containing [a, b], where f is

also con nuous on [a, b]. Then the arc length of f from x = a to x = b is
L=
1 + f (x)2 dx.
As the integrand contains a square root, it is o en dicult to use the formula in Key Idea 27 to nd the length exactly. When exact answers are dicult
to come by, we resort to using numerical methods of approxima ng denite integrals. The following examples will demonstrate this.
Notes:
357

. Example 211
Finding arc length
Find the arc length of f(x) = x3/2 from x = 0 to x = 4.
y
8
6
We begin by nding f (x) = 32 x1/2 . Using the formula, we

S
nd the arc length L as
4
2
L=
x
2
1+
0
4
3 1/2
x
2
)2
dx
9
1 + x dx
4
0
)1/2
4(
9
dx
=
1+ x
4
0
(
)3/2
24
9
4
=
1+ x

39
4
0
(
)
8
=
103/2 1 9.07units.
27
Figure 7.24: A graph of f(x) = x3/2 from

Example 211.
A graph of f is given in Figure 7.24. .

. Example 212
.Finding arc length

1
Find the arc length of f(x) = x2 ln x from x = 1 to x = 2.
8
This func on was chosen specically because the resul ng

S
integral can be evaluated exactly. We begin by nding f (x) = x/4 1/x. The
arc length is
L=
1+
1
358
1+
Notes:
x 1
4 x
)2
dx
x2
1
1
+
dx
16 2 x2
x2
1
1
+ + 2 dx
16
2
x
1
)2
2 (
x 1
=
+
dx
4 x
1
)
2(
x 1
+
dx
=
4 x
1
( 2
) 2
x

=
+ ln x

8
=
3
= + ln 2 1.07 units.
8
y
1
0.5
x
1
Figure 7.25: A graph of f(x) = 18 x2 ln x

from Example 212.
A graph of f is given in Figure 7.25; the por on of the curve measured in this
problem is in bold. .
The previous examples found the arc length exactly through careful choice
of the func ons. In general, exact answers are much more dicult to come by
and numerical approxima ons are necessary.
. Example 213
Approxima ng arc length numerically
Find the length of the sine curve from x = 0 to x = .
This is somewhat of a mathema cal curiosity; in Example
S
125 we found the area under one hump of the sine curve is 2 square units;
now we are measuring its arc length.
The setup is straigh orward: f(x) = sin x and f (x) = cos x. Thus

L=
1 + cos2 x dx.
0
This integral cannot be evaluated in terms of elementary func ons sowe will approximate it with Simpsons Method with n = 4. Figure 7.26 gives 1 + cos2 x
evaluated at 5 evenly spaced points in [0, ]. Simpsons Rule then states that

)
0 (
1 + cos2 x dx
2 + 4 3/2 + 2(1) + 4 3/2 + 2
43
0
= 3.82918.
Using a computer with n = 100 the approxima on is L 3.8202; our approxima on with n = 4 is quite good. .
x
0
/4
/2
3/4
1 + cos2 x
2
3/2
1
3/2
2
Figure
7.26: A table of values of y =
1 + cos2 x to evaluate a denite integral in Example 213.
Notes:
359
Surface Area of Solids of Revolu on

We have already seen how a curve y = f(x) on [a, b] can be revolved around
an axis to form a solid. Instead of compu ng its volume, we now consider its
surface area.
We begin as we have in the previous sec ons: we par on the interval [a, b]
with n subintervals, where the i th subinterval is [xi , xi+1 ]. On each subinterval,
we can approximate the curve y = f(x) with a straight line that connects f(xi )
and f(xi+1 ) as shown in Figure 7.27 (a). Revolving this line segment about the
x-axis creates part of a cone (called the frustum of a cone) as shown in Figure
7.27 (b). The surface area of a frustum of a cone is
2 length average of the two radii R and r.

.
a
xi
xi+1
.
(a)
y
}|
.a
xi
for some ci in the i th subinterval. The radii are just the func on evaluated at the
endpoints of the interval. That is,
xi+1
The length is given by L; we use the material just covered by arc length to
state that
L 1 + f (ci )xi
R = f(xi+1 )
and
r = f(xi ).
x
b
Thus the surface area of this sample frustum of the cone is approximately
2
f(xi ) + f(xi+1 )
1 + f (ci )2 xi .
2
(b)
Figure 7.27: Establishing the formula for
surface area.
Since f is a con nuous func on, the Intermediate Value Theorem states there
f(xi ) + f(xi+1 )
is some di in [xi , xi+1 ] such that f(di ) =
; we can use this to rewrite
2
the above equa on as
2f(di ) 1 + f (ci )2 xi .
Summing over all the subintervals we get the total surface area to be approximately
n
Surface Area
2f(di ) 1 + f (ci )2 xi ,
i=1
which is a Riemann Sum. Taking the limit as the subinterval lengths go to zero
gives us the exact surface area, given in the following Key Idea.
Notes:
360
Key Idea 28
Surface Area of a Solid of Revolu on
Let f be dieren able on an open interval containing [a, b] where f is

also con nuous on [a, b].
1. The surface area of the solid formed by revolving the graph of y =
f(x), where f(x) 0, about the x-axis is
Surface Area = 2
f(x)
1 + f (x)2 dx.
2. The surface area of the solid formed by revolving the graph of y =

f(x) about the y-axis, where a, b 0, is
Surface Area = 2
x 1 + f (x)2 dx.
(When revolving y = f(x) about the y-axis, the radii of the resul ng frustum
are xi and xi+1 ; their average value is simply the midpoint of the interval. In the
limit, this midpoint is just x. This gives the second part of Key Idea 28.)
. Example 214
Finding surface area of a solid of revolu on
Find the surface area of the solid formed by revolving y = sin x on [0, ] around
the x-axis, as shown in Figure 7.28.
The setup is rela vely straigh orward. Using Key Idea 28,
S
we have the surface area SA is:

SA = 2
sin x 1 + cos2 x dx
y
1
1 (

= 2
sinh1 (cos x) + cos x 1 + cos2 x
2
0
)
(
1
2 + sinh 1
= 2
14.42 units2 .
The integra on step above is nontrivial, u lizing an integra on method called

Trigonometric Subs tu on.
It is interes ng to see that the surface area of a solid, whose shape is dened
by a trigonometric func on, involves both a square root and an inverse hyperbolic trigonometric func on. .
Figure 7.28: Revolving y = sin x on [0, ]

about the x-axis.
Notes:
361
. Example 215
Finding surface area of a solid of revolu on
Find the surface area of the solid formed by revolving the curve y = x2 on [0, 1]
about:
1. the x-axis
2. the y-axis.
S
1. The integral is straigh orward to setup:

1
x2 1 + (2x)2 dx.
SA = 2
Like the integral in Example 214, this requires Trigonometric Subs tu on.
)1
(

=
2(8x3 + x) 1 + 4x2 sinh1 (2x)
32
0
)
(
=
18 5 sinh1 2
32
3.81 units2 .
0.5
.
(a)
The solid formed by revolving y = x2 around the x-axis is graphed in Figure

7.29 (a).
y
2. Since we are revolving around the y-axis, the radius of the solid is not
f(x) but rather x. Thus the integral to compute the surface area is:
1
SA = 2
x 1 + (2x)2 dx.
.
1
0.5
0.5
0.5
This integral can be solved using subs tu on. Set u = 1 + 4x2 ; the new
bounds are u = 1 to u = 5. We then have
5
=
u du
4 1
5
2 3/2
=
u
43
1
(
)

=
5 51
6
5.33 units2 .
.
(b)
Figure 7.29: The solids used in Example
215.
The solid formed by revolving y = x2 about the y-axis is graphed in Figure

. 7.29 (b).
This last example is a famous mathema cal paradox.
Notes:
362

. Example 216
The surface area and volume of Gabriels Horn
Consider the solid formed by revolving y = 1/x about the x-axis on [1, ). Find
the volume and surface area of this solid. (This shape, as graphed in Figure 7.30,
is known as Gabriels Horn since it looks like a very long horn that only a supernatural person, such as an angel, could play.)
y
To compute the volume it is natural to use the Disk Method.
We have:
1
dx
x2
1
b
1
= lim
dx
2
b
x
1
(
)b
1
= lim
b
x 1
(
)
1
= lim 1
b
b
V=
10
x
15
.
Figure 7.30: A graph of Gabriels Horn.
= units3 .
Gabriels Horn has a nite volume of cubic units. Since we have already seen
that objects with innite length can have a nite area, this is not too dicult to
accept.
We now consider its surface area. The integral is straigh orward to setup:

1
SA = 2
1 + 1/x4 dx.
x
1
Integra ng this expression is not
trivial. We can, however, compare it to other
improper integrals. Since 1 < 1 + 1/x4 on [1, ), we can state that

1
1
2
1 + 1/x4 dx.
dx < 2
x
x
1
1
By Key Idea 21, the improper integral on the le diverges. Since the integral
on the right is larger, we conclude it also diverges, meaning Gabriels Horn has
innite surface area.
Hence the paradox: we can ll Gabriels Horn with a nite amount of paint,
but since it has innite surface area, we can never paint it.
Somehow this paradox is striking when we think about it in terms of volume and area. However, we have seen a similar paradox before, as referenced
above. We know that the area under the curve y = 1/x2 on [1, ) is nite, yet
the shape has an innite perimeter. Strange things can occur when we deal with
the innite. .
Notes:
363
Exercises 7.4
Terms and Concepts
1. T/F: The integral formula for compu ng Arc Length was
found by rst approxima ng arc length with straight line
segments.
2. T/F: The integral formula for compu ng Arc Length includes
a squareroot, meaning the integra on is probably easy.
Problems
In Exercises 3 13, nd the arc length of the func on on the
given interval.
3. f(x) = x on [0, 1].
4. f(x) = 8x on [1, 1].

5. f(x) =
1 3/2
x x1/2 on [0, 1].
3
1 3 1
x + on [1, 4].
12
x
1
3/2
7. f(x) = 2x
x on [0, 9].
6
6. f(x) =
8. f(x) = cosh x on [ ln 2, ln 2].

9.
10.
11.
12.
)
1( x
f(x) =
e + ex on [0, ln 5].
2
1 5
1
f(x) =
x + 3 on [.1, 1].
12
5x
(
)
f(x) = ln sin x on [/6, /2].
(
)
f(x) = ln cos x on [0, /4].
In Exercises 13 21, set up the integral to compute the arc

length of the func on on the given interval. Do not evaluate
the integral.
2
13. f(x) = x on [0, 1].

14. f(x) = x10 on [0, 1].
15. f(x) = x on [0, 1].

16. f(x) = ln x on [1, e].
17. f(x) = 1 x2 on [1, 1]. (Note: this describes the top

half of a circle with radius 1.)
18. f(x) = 1 x2 /9 on [3, 3]. (Note: this describes the top

half of an ellipse with a major axis of length 6 and a minor
axis of length 2.)
19. f(x) =
1
on [1, 2].
x
20. f(x) = sec x on [/4, /4].

In Exercises 21 29, use Simpsons Rule, with n = 4, to approximate the arc length of the func on on the given interval.
Note: these are the same problems as in Exercises 1320.
21. f(x) = x2 on [0, 1].
22. f(x) = x10 on [0, 1].
23. f(x) = x on [0, 1]. (Note: f (x) is not dened at x = 0.)

24. f(x) = ln x on [1, e].
25. f(x) = 1 x2 on [1, 1]. (Note: f (x) is not dened at

the endpoints.)
26. f(x) = 1 x2 /9 on [3, 3]. (Note: f (x) is not dened

at the endpoints.)
27. f(x) =
1
on [1, 2].
x
28. f(x) = sec x on [/4, /4].

In Exercises 29 33, nd the Surface Area of the described
solid of revolu on.
29. The solid formed by revolving y = 2x on [0, 1] about the
x-axis.
30. The solid formed by revolving y = x2 on [0, 1] about the
y-axis.
31. The solid formed by revolving y = x3 on [0, 1] about the
x-axis.
32. The solid formed by revolving y = x on [0, 1] about the

x-axis.
33. The sphere formed by revolving y = 1 x2 on [1, 1]

about the x-axis.
7.5 Work
7.5
Work
Work is the scien c term used to describe the ac on of a force which moves
an object. When a constant force F is applied to move an object a distance d,
the amount of work performed is W = F d.
The SI unit of force is the Newton, (kgm/s2 ), and the SI unit of distance is
a meter (m). The fundamental unit of work is one Newtonmeter, or a joule
(J). That is, applying a force of one Newton for one meter performs one joule
of work. In Imperial units (as used in the United States), force is measured in
pounds (lb) and distance is measured in feet ( ), hence work is measured in
lb.
When force is constant, the measurement of work is straigh orward. For
instance, li ing a 200 lb object 5 performs 200 5 = 1000 lb of work.
What if the force applied is variable? For instance, imagine a climber pulling
a 200 rope up a ver cal face. The rope becomes lighter as more is pulled in,
requiring less force and hence the climber performs less work.
In general, let F(x) be a force func on on an interval [a, b]. We want to measure the amount of work done applying the force F from x = a to x = b. We can
approximate the amount of work being done by par oning [a, b] into subintervals a = x1 < x2 < < xn+1 = b and assuming that F is constant on each
subinterval. Let ci be a value in the i th subinterval [xi , xi+1 ]. Then the work done
on this interval is approximately Wi F(ci ) (xi+1 xi ) = F(ci )xi , a constant
force the distance over which it is applied. The total work is
W=
Wi
i=1
Note: Mass and weight are closely related, yet dierent, concepts. The mass
m of an object is a quan ta ve measure
of that objects resistance to accelera on.
The weight w of an object is a measurement of the force applied to the object by
the accelera on of gravity g.
Since the two measurements are propor onal, w = m g, they are o en
used interchangeably in everyday conversa on. When compu ng Work, one must
be careful to note which is being referred
to. When mass is given, it must be mul plied by the accelera on of gravity to reference the related force.
F(ci )xi .
i=1
This, of course, is a Riemann sum. Taking a limit as the subinterval lengths go

to zero give an exact value of work which can be evaluated through a denite
integral.
Key Idea 29
Work
Let F(x) be a con nuous func on on [a, b] describing the amount of force
being applied to an object in the direc on of travel from distance x = a
to distance x = b. The total work W done on [a, b] is
W=
F(x) dx.
a
Notes:
365

. Example 217
Compu ng work performed: applying variable force
How much work is performed pulling a 60 m climbing rope up a cli face, where
the rope has a mass of 66 g/m?
We need to create a force func on F(x) on the interval [0, 60].
S
To do so, we must rst decide what x is measuring: it is the length of the rope
s ll hanging or is it the amount of rope pulled in? As long as we are consistent,
either approach is ne. We adopt for this example the conven on that x is the
amount of rope pulled in. This seems to match intui on be er; pulling up the
rst 10 meters of rope involves x = 0 to x = 10 instead of x = 60 to x = 50.
As x is the amount of rope pulled in, the amount of rope s ll hanging is 60x.
This length of rope has a mass of 66 g/m, or 0.066 kg/m. The the mass of the
rope s ll hanging is 0.066(60 x) kg; mul plying this mass by the accelera on
of gravity, 9.8 m/s2 , gives our variable force func on
F(x) = (9.8)(0.066)(60 x) = 0.6468(60 x).
Thus the total work performed in pulling up the rope is
60
W=
0.6468(60 x) dx = 1, 164.24 J.
By comparison, consider the work done in li ing the en re rope 60 meters.

The rope weights 60 0.066 9.8 = 38.808 N, so the work applying this force
for 60 meters is 60 38.808 = 2, 328.48 J. This is exactly twice the work calculated before (and we leave it to the reader to understand why.) .
. Example 218
.Compu ng work performed: applying variable force
Consider again pulling a 60 m rope up a cli face, where the rope has a mass of
66 g/m. At what point is exactly half the work performed?
From Example 217 we know the total work performed is
S
11, 642.4 J. We want to nd a height h such that the work in pulling the rope
from a height of x = 0 to a height of x = h is 5821.2, half the total work. Thus
we want to solve the equa on
Notes:
366
6.468(60 x) dx = 5821.2
7.5 Work
for h.
6.468(60 x) dx = 5821.2
(
) h
388.08x 3.234x2 = 5821.2
Note: In Example 218, we nd that half of

the work performed in pulling up a 60 m
rope is done in the last 42.43
m. Why is it
not coincidental that 60/ 2 = 42.43?
388.08h 3.234h2 = 5821.2

3.234h2 + 388.08h 5821.2 = 0.
Apply the Quadra c Formula.
h = 17.57 and 102.43
As the rope is only 60 m long, the only sensible answer is h = 17.57. Thus about
half the work is done pulling up the rst 17.5 m; the other half of the work is
done pulling up the remaining 42.43 m. .
. Example 219
.Compu ng work performed: applying variable force
A box of 100 lb of sand is being pulled up at a uniform rate a distance of 50
over 1 minute. The sand is leaking from the box at a rate of 1 lb/s. The box itself
weighs 5 lb and is pulled by a rope weighing .2 lb/ .
1. How much work is done li ing just the rope?
2. How much work is done li ing just the box and sand?
3. What is the total amount of work performed?
1. We start by forming the force func on Fr (x) for the rope (where the subscript denotes we are considering the rope). As in the previous example,
let x denote the amount of rope, in feet, pulled in. (This is the same as
saying x denotes the height of the box.) The weight of the rope with x
feet pulled in is Fr (x) = 0.2(50 x) = 10 0.2x. (Note that we do not
have to include the accelera on of gravity here, for the weight of the rope
per foot is given, not its mass per meter as before.) The work performed
li ing the rope is
Wr =
50
(10 0.2x) dx = 250 lb.
Notes:
367

2. The sand is leaving the box at a rate of 1 lb/s. As the ver cal trip is to take
one minute, we know that 60 lb will have le when the box reaches its nal
height of 50 . Again le ng x represent the height of the box, we have
two points on the line that describes the weight of the sand: when x = 0,
the sand weight is 100 lb, producing the point (0, 100); when x = 50, the
sand in the box weighs 40 lb, producing the point (50, 40). The slope of
this line is 10040
050 = 1.2, giving the equa on of the weight of the sand
at height x as w(x) = 1.2x + 100. The box itself weighs a constant 5 lb,
so the total force func on is Fb (x) = 1.2x + 105. Integra ng from x = 0
to x = 50 gives the work performed in li ing box and sand:
50
Wb =
(1.2x + 105) dx = 3750 lb.

0
3. The total work is the sum of Wr and Wb : 250 + 3750 = 4000 lb. We
can also arrive at this via integra on:
50
W=
(Fr (x) + Fb (x)) dx
50
(10 0.2x 1.2x + 105) dx
50
(1.4x + 115) dx
=
0
= 4000 lb.
Hookes Law and Springs

Hookes Law states that the force required to compress or stretch a spring x
units from its natural length is propor onal to x; that is, this force is F(x) = kx
for some constant k. For example, if a force of 1 N stretches a given spring
2 cm, then a force of 5 N will stretch the spring 10 cm. Conver ng the distances to meters, we have that stretching a this spring 0.02 m requires a force
of F(0.02) = k(0.02) = 1 N, hence k = 1/0.02 = 50 N/m.
. Example 220
.Compu ng work performed: stretching a spring
A force of 20 lb stretches a spring from a length of 7 inches to a length of 12
inches. How much work was performed in stretching the spring to this length?
In many ways, we are not at all concerned with the actual
S
length of the spring, only with the amount of its change. Hence, we do not care
that 20 lb of force stretches the spring to a length of 12 inches, but rather that
Notes:
368
7.5 Work
a force of 20 lb stretches the spring by 5 in. This is illustrated in Figure 7.31;
we only measure the change in the springs length, not the overall length of the
spring.
F
Fluid
Concrete
Fuel Oil
Gasoline
Iodine
Methanol
Mercury
Milk
Water
lb/ 3
150
55.46
45.93
307
49.3
844
63.665.4
62.4
kg/m3
2400
890.13
737.22
4927
791.3
13546
1020 1050
1000
Figure 7.32: Weight and Mass densi es

Figure 7.31: Illustra ng the important aspects of stretching a spring in compu ng work
in Example 220.
Conver ng the units of length to feet, we have

F(5/12) = 5/12k = 20 lb.
Thus k = 48 lb/ and F(x) = 48x.
We compute the total work performed by integra ng F(x) from x = 0 to
x = 5/12:
5/12
W=
48x dx
0
5/12

= 24x2
0
= 25/6 4.1667 lb.
Pumping Fluids
Another useful example of the applica on of integra on to compute work
comes in the pumping of uids, o en illustrated in the context of emptying a
storage tank by pumping the uid out the top. This situa on is dierent than
our previous examples for the forces involved are constant. A er all, the force
required to move one cubic foot of water (about 62.4 lb) is the same regardless
of its loca on in the tank. What is variable is the distance that cubic foot of
water has to travel; water closer to the top travels less distance than water at
the bo om, producing less work.
We demonstrate how to compute the total work done in pumping a uid out
of the top of a tank in the next two examples.
Notes:
369

. Example 221
Compu ng work performed: pumping uids
A cylindrical storage tank with a radius of 10 and a height of 30 is lled with
water, which weighs approximately 62.4 lb/ 3 . Compute the amount of work
performed by pumping the water up to a point 5 feet above the top of the tank.
We will refer o en to Figure 7.33 which illustrates the salient
S
aspects of this problem.
We start as we o en do: we par on an interval into subintervals. We orient
our tank ver cally since this makes intui ve sense with the base of the tank at
y = 0. Hence the top of the water is at y = 30, meaning we are interested in
subdividing the y-interval [0, 30] into n subintervals as
y
35
10
. .
35 yi
30
0 = y1 < y2 < < yn+1 = 30.
yi+1 }
yi
yi
Figure 7.33: Illustra ng a water tank in

order to compute the work required to
empty it in Example 221.
Consider the work Wi of pumping only the water residing in the i th subinterval,
illustrated in Figure 7.33. The force required to move this water is equal to its
weight which we calculate as volume density. The volume of water in this
subinterval is Vi = 102 yi ; its density is 62.4 lb/ 3 . Thus the required force is
6240yi lb.
We approximate the distance the force is applied by using any y-value contained in the i th subinterval; for simplicity, we arbitrarily use yi for now (it will
not ma er later on). The water will be pumped to a point 5 feet above the top
of the tank, that is, to the height of y = 35 . Thus the distance the water at
height yi travels is 35 yi .
In all, the approximate work Wi peformed in moving the water in the i th
subinterval to a point 5 feet above the tank is
Wi 6240yi (35 yi ),
and the total work performed is
W
Wi =
i=1
6240yi (35 yi ).
i=1
This is a Riemann sum. Taking the limit as the subinterval length goes to 0 gives
30
W=
6240(35 y) dy
0
(
) 30
= (6240 35y 1/2y2
0
= 11, 762, 123 lb

.
Notes:
370
1.176 107 lb.
7.5 Work
y
The conical tank is sketched in Figure 7.35. We can orient

S
the tank in a variety of ways; we could let y = 0 represent the base of the tank
and y = 10 represent the top of the tank, but we choose to keep the conven on
of the wording given in the problem and let y = 0 represent ground level and
hence y = 10 represents the bo om of the tank. The actual height of the
water does not ma er; rather, we are concerned with the distance the water
travels.
The gure also sketches a dieren al element, a crosssec onal circle. The
radius of this circle is variable, depending on y. When y = 10, the circle has
radius 0; when y = 0, the circle has radius 2. These two points, (10, 0) and
(0, 2), allow us to nd the equa on of the line that gives the radius of the cross
sec onal circle, which is r(y) = 1/5y + 2. Hence the volume of water at this
height is V(y) = (1/5y + 2)2 dy, where dy represents a very small height of
the dieren al element. The force required to move the water at height y is
F(y) = 62.4 V(y).
The distance the water at height y travels is given by h(y) = 3 y. Thus the
total work done in pumping the water from the tank is
0
W=
62.4(1/5y + 2)2 (3 y) dy
10
(
)
1
17
8
= 62.4
y3 y2 y + 12 dy
25
25
5
10
220
= 62.2
14, 376 lb.
3
10
35 y
30
V(y) = 100dy
Figure 7.34: A simplied illustra on for

compu ng work.
y
3
2
. Example 222
A conical water tank has its top at ground level and its base 10 feet below ground.
The radius of the cone at ground level is 2 . It is lled with water weighing 62.4
lb/ 3 and is to be emp ed by pumping the water to a spigot 3 feet above ground
level. Find the total amount of work performed in emptying the tank.
35
3y
We can streamline the above process a bit as we may now recognize what
the important features of the problem are. Figure 7.34 shows the tank from
Example 221 without the i th subinterval iden ed. Instead, we just draw one
dieren al element. This helps establish the height a small amount of water
must travel along with the force required to move it (where the force is volume
density).
We demonstrate the concepts again in the next examples.
V(y) = ( 5y + 2)2 dy
10
Figure 7.35: A graph of the conical water

tank in Example 222.
Notes:
371
25
3 .
6 .
10 .
.
10 .
Figure 7.36: The crosssec on of a swimming pool lled with water in Example
223.
y
8
6
3
y
0
. Example 223
A rectangular swimming pool is 20 wide and has a 3 shallow end and a 6
deep end. It is to have its water pumped out to a point 2 above the current
top of the water. The crosssec onal dimensions of the water in the pool are
given in Figure 7.36; note that the dimensions are for the water, not the pool
itself. Compute the amount of work performed in draining the pool.
For the purposes of this problem we choose to set y = 0
S
to represent the bo om of the pool, meaning the top of the water is at y = 6.
Figure 7.37 shows the pool oriented with this y-axis, along with 2 dieren al
elements as the pool must be split into two dierent regions.
The top region lies in the y-interval of [3, 6], where the length of the dierenal element is 25 as shown. As the pool is 20 wide, this dieren al element
represents a this slice of water with volume V(y) = 20 25 dy. The water is
to be pumped to a height of y = 8, so the height func on is h(y) = 8 y. The
work done in pumping this top region of water is
6
Wt = 62.4
500(8 y) dy = 327, 600 lb.
3
(15, 3)
(10, 0)
x
10
15
Figure 7.37: Orien ng the pool and showing dieren al elements for Example 223.
The bo om region lies in the y-interval of [0, 3]; we need to compute the
length of the dieren al element in this interval.
One end of the dieren al element is at x = 0 and the other is along the line
segment joining the points (10, 0) and (15, 3). The equa on of this line is y =
3/5(x 10); as we will be integra ng with respect to y, we rewrite this equa on
as x = 5/3y + 10. So the length of the dieren al element is a dierence of
x-values: x = 0 and x = 5/3y + 10, giving a length of x = 5/3y + 10.
Again, as the pool is 20 wide, this dieren al element represents a thin
slice of water with volume V(y) = 20 (5/3y + 10) dy; the height func on is
the same as before at h(y) = 8 y. The work performed in emptying this part
of the pool is
3
Wb = 62.4
20(5/3y + 10)(8 y) dy = 299, 520 lb.
0
The total work in empy ng the pool is

W = Wb + Wt = 327, 600 + 299, 520 = 627, 120 lb.
No ce how the emptying of the bo om of the pool performs almost as much
work as emptying the top. The top por on travels a shorter distance but has
more water. In the end, this extra water produces more work. .
Notes:
372
Exercises 7.5
Terms and Concepts
11. A force of 1000 lb compresses a spring 3 in. How much work

is performed in compressing the spring?
1. What are the typical units of work?

2. If a man has a mass of 80 kg on Earth, will his mass on the
moon be bigger, smaller, or the same?
12. A force of 2 N stretches a spring 5 cm. How much work is

performed in stretching the spring?
3. If a woman weighs 130 lb on Earth, will her weight on the

moon be bigger, smaller, or the same?
13. A force of 50 lb compresses a spring from 18 in to 12 in.

How much work is performed in compressing the spring?
Problems
14. A force of 20 lb stretches a spring from 6 in to 8 in. How

much work is performed in stretching the spring?
4. A 100 rope, weighing 0.1 lb/ , hangs over the edge of a

tall building.
(a) How much work is done pulling the en re rope to the
top of the building?
(b) How much rope is pulled in when half of the total
work is done?
15. A force of 7 N stretches a spring from 11 cm to 21 cm. How

much work is performed in stretching the spring?
16. A force of f N stretches a spring d m. How much work is
performed in stretching the spring?
5. A 50 m rope, with a mass density of 0.2 kg/m, hangs over

the edge of a tall building.
17. A 20 lb weight is a ached to a spring. The weight rests on

the spring, compressing the spring from a natural length of
1 to 6 in.

top of the building?
How much work is done in li ing the box 1.5 (i.e, the
spring will be stretched 1 beyond its natural length)?
(b) How much work is done pulling in the rst 20 m?

6. A rope of length hangs over the edge of tall cli. (Assume the cli is taller than the length of the rope.) The
rope has a weight density of d lb/ .
top of the cli?
(b) What percentage of the total work is done pulling in
the rst half of the rope?
(c) How much rope is pulled in when half of the total
work is done?
7. A 20 m rope with mass density of 0.5 kg/m hangs over the
edge of a 10 m building. How much work is done pulling
the rope to the top?
8. A crane li s a 2,000 lb load ver cally 30
weighing 1.68 lb/ .
with a 1 cable
(a) How much work is done li ing the cable alone?
18. A 20 lb weight is a ached to a spring. The weight rests on

the spring, compressing the spring from a natural length of
1 to 6 in.
How much work is done in li ing the box 6 in (i.e, bringing
the spring back to its natural length)?
19. A 5 m tall cylindrical tank with radius of 2 m is lled with 3
m of gasoline, with a mass density of 737.22 kg/m3 . Compute the total work performed in pumping all the gasoline
to the top of the tank.
20. A 6 cylindrical tank with a radius of 3 is lled with water, which has a weight density of 62.4 lb/ 3 . The water is
to be pumped to a point 2 above the top of the tank.
(a) How much work is performed in pumping all the water from the tank?
(b) How much work is done li ing the load alone?
(b) How much work is performed in pumping 3

ter from the tank?
(c) Could one conclude that the work done li ing the cable is negligible compared to the work done li ing the
load?
(c) At what point is 1/2 of the total work done?
9. A 100 lb bag of sand is li ed uniformly 120 in one minute.

Sand leaks from the bag at a rate of 1/4 lb/s. What is the
total work done in li ing the bag?
10. A box weighing 2 lb li s 10 lb of sand ver cally 50 . A crack
in the box allows the sand to leak out such that 9 lb of sand
is in the box at the end of the trip. Assume the sand leaked
out at a uniform rate. What is the total work done in li ing
the box and sand?
of wa-
21. A gasoline tanker is lled with gasoline with a weight density of 45.93 lb/ 3 . The dispensing valve at the base is
jammed shut, forcing the operator to empty the tank via
pumping the gas to a point 1 above the top of the tank.
Assume the tank is a perfect cylinder, 20 long with a diameter of 7.5 .
How much work is performed in pumping all the gasoline
from the tank?
22. A fuel oil storage tank is 10 deep with trapezoidal sides,

5 at the top and 2 at the bo om, and is 15 wide (see
diagram below). Given that fuel oil weighs 55.46 lb/ 3 , nd
the work performed in pumping all the oil from the tank to
a point 3 above the top of the tank.
5
25. A water tank has the shape of an inverted pyramid, with dimensions given below, and is lled with water with a mass
density of 1000 kg/m3 . Find the work performed in pumping all water to a point 5 m above the top of the tank.
2m
2m
7m
10
26. A water tank has the shape of an truncated, inverted pyramid, with dimensions given below, and is lled with water with a mass density of 1000 kg/m3 . Find the work performed in pumping all water to a point 1 m above the top
of the tank.
15
5m
2
5m
23. A conical water tank is 5 m deep with a top radius of 3 m.

(This is similar to Example 222.) The tank is lled with pure
water, with a mass density of 1000 kg/m3 .
(a) Find the work performed in pumping all the water to
the top of the tank.
(b) Find the work performed in pumping the top 2.5 m
of water to the top of the tank.
(c) Find the work performed in pumping the top half of
the water, by volume, to the top of the tank.
24. A water tank has the shape of a truncated cone, with dimensions given below, and is lled with water with a weight
density of 62.4 lb/ 3 . Find the work performed in pumping
all water to a point 1 above the top of the tank.
.
2
10
5
9m
2m
2m
7.6 Fluid Forces
7.6
Fluid Forces
In the unfortunate situa on of a car driving into a body of water, the convenonal wisdom is that the water pressure on the doors will quickly be so great
that they will be eec vely unopenable. (Survival techniques suggest immediately opening the door, rolling down or breaking the window, or wai ng un l
the water lls up the interior at which point the pressure is equalized and the
door will open. See Mythbusters episode #72 to watch Adam Savage test these
op ons.)
How can this be true? How much force does it take to open the door of
a submerged car? In this sec on we will nd the answer to this ques on by
examining the forces exerted by uids.
We start with pressure, which is related to force by the following equa ons:
Force
Area
Force = Pressure Area.
Pressure =
In the context of uids, we have the following deni on.

Fluid Pressure
Let w be the weightdensity of a uid. The pressure p exerted on an

object at depth d in the uid is p = w d.
.
We use this deni on to nd the force exerted on a horizontal sheet by considering the sheets area.
. Example 224
10
Deni on 26
Figure 7.38: A cylindrical tank in Example

224.
.Compu ng uid force
1. A cylindrical storage tank has a radius of 2 and holds 10 of a uid with

a weightdensity of 50 lb/ 3 . (See Figure 7.38.) What is the force exerted
on the base of the cylinder by the uid?
10
2. A rectangular tank whose base is a 5 square has a circular hatch at the

bo om with a radius of 2 . The tank holds 10 of a uid with a weight
density of 50 lb/ 3 . (See Figure 7.39.) What is the force exerted on the
hatch by the uid?
1. Using Deni on 26, we calculate that the pressure exerted on the cylinders base is w d = 50 lb/ 3 10 = 500 lb/ 2 . The area of the base is
Figure 7.39: A rectangular tank in Example 224.
Notes:
375

22 = 4
. So the force exerted by the uid is

F = 500 4 = 6283 lb.
Note that we eec vely just computed the weight of the uid in the tank.
2. The dimensions of the tank in this problem are irrelevant. All we are concerned with are the dimensions of the hatch and the depth of the uid.
Since the dimensions of the hatch are the same as the base of the tank
in the previous part of this example, as is the depth, we see that the uid
force is the same. That is, F = 6283 lb.
The previous example demonstrates that compu ng the force exerted on a

horizontally oriented plate is rela vely easy to compute. What about a ver cally
oriented plate? For instance, suppose we have a circular porthole located on the
side of a submarine. How do we compute the uid force exerted on it?
Pascals Principle states that the pressure exerted by a uid at a depth is
equal in all direc ons. Thus the pressure on any por on of a plate that is 1
below the surface of water is the same no ma er how the plate is oriented.
(Thus a hollow cube submerged at a great depth will not simply be crushed
from above, but the sides will also crumple in. The uid will exert force on all
sides of the cube.)
So consider a ver cally oriented plate as shown in Figure 7.40 submerged in
a uid with weightdensity w. What is the total uid force exerted on this plate?
We nd this force by rst approxima ng the force on small horizontal strips.
Let the top of the plate be at depth b and let the bo om be at depth a. (For
now we assume that surface of the uid is at depth 0, so if the bo om of the
plate is 3 under the surface, we have a = 3. We will come back to this later.)
We par on the interval [a, b] into n subintervals
di
}
a = y1 < y2 < < yn+1 = b,
yi
(ci )
.
Figure 7.40: A thin, ver cally oriented
plate submerged in a uid with weight
density w.
with the i th subinterval having length yi . The force Fi exerted on the plate in
the i th subinterval is Fi = Pressure Area.
The pressure is depth w. We approximate the depth of this thin strip by
choosing any value di in [yi , yi+1 ]; the depth is approximately di . (Our convenon has di being a nega ve number, so di is posi ve.) For convenience, we let
di be an endpoint of the subinterval; we let di = yi .
The area of the thin strip is approximately length width. The width is yi .
The length is a func on of some y-value ci in the i th subinterval. We state the
Notes:
376
A key concept to understand here is that we are eec vely measuring the
weight of a 10 column of water above the hatch. The size of the tank
holding the uid does not ma er.
7.6 Fluid Forces

length is (ci ). Thus
Fi = Pressure Area
= yi w (ci ) yi .
The total force is then
F=
Fi
i=1
w yi (ci ) yi .
i=1
This is, of course, another Riemann Sum. We can nd the exact force by taking
a limit as the subinterval lengths go to 0; we evaluate this limit with a denite
integral.
Key Idea 30
Fluid Force on a Ver cally Oriented Plate
Let a ver cally oriented plate be submerged in a uid with weight

density w where the top of the plate is at y = b and the bo om is at
y = a. Let (y) be the length of the plate at y.
1. If y = 0 corresponds to the surface of the uid, then the force
exerted on the plate by the uid is
F=
w (y) (y) dy.
2. In general, let d(y) represent the distance between the surface of

the uid and the plate at y. Then the force exerted on the plate by
the uid is
F=
w d(y) (y) dy.
We approach this problem in two dierent ways to illustrate

the dierent ways Key Idea 30 can be implemented. First we will let y = 0 represent the surface of the water, then we will consider an alternate conven on.
S
. Example 225
.Finding uid force
Consider a thin plate in the shape of an isosceles triangle as shown in Figure 7.41
submerged in water with a weightdensity of 62.4 lb/ 3 . If the bo om of the
plate is 10 below the surface of the water, what is the total uid force exerted
on this plate?
.
Figure 7.41: A thin plate in the shape of
an isosceles triangle in Example 225.
Notes:
377

1. We let y = 0 represent the surface of the water; therefore the bo om of
the plate is at y = 10. We center the triangle on the y-axis as shown in
Figure 7.42. The depth of the plate at y is y as indicated by the Key Idea.
We now consider the length of the plate at y.
y
water line
x
2 1
We need to nd equa ons of the le and right edges of the plate. The
right hand side is a line that connects the points (0, 10) and (2, 6):
that line has equa on x = 1/2(y + 10). (Find the equa on in the familiar
y = mx+b format and solve for x.) Likewise, the le hand side is described
by the line x = 1/2(y + 10). The total length is the distance between
these two lines: (y) = 1/2(y + 10) (1/2(y + 10)) = y + 10.
2
d(y) = y
2
4
(2, 6)
(2, 6)
The total uid force is then:

6
62.4(y)(y + 10) dy
F=
y
8
10
10
(0, 10)
= 62.4
Figure 7.42: Sketching the triangular

plate in Example 225 with the conven on
that the water level is at y = 0.
2. Some mes it seems easier to orient the thin plate nearer the origin. For
instance, consider the conven on that the bo om of the triangular plate
is at (0, 0), as shown in Figure 7.43. The equa ons of the le and right
hand sides are easy to nd. They are y = 2x and y = 2x, respec vely,
which we rewrite as x = 1/2y and x = 1/2y. Thus the length func on
is (y) = 1/2y (1/2y) = y.
As the surface of the water is 10 above the base of the plate, we have
that the surface of the water is at y = 10. Thus the depth func on is the
distance between y = 10 and y; d(y) = 10 y. We compute the total
uid force as:
4
F=
62.4(10 y)(y) dy
y
water line
10
d(y) = 10 y
8
6
(2, 4)
(2, 4)
y
2
.
2 1
x
1
Figure 7.43: Sketching the triangular

plate in Example 225 with the conven on
that the base of the triangle is at (0, 0).
3660.8 lb.
The correct answer is, of course, independent of the placement of the plate in
the coordinate plane as long as we are consistent. .
. Example 226
.Finding uid force
Find the total uid force on a car door submerged up to the bo om of its window
in water, where the car door is a rectangle 40 long and 27 high (based on the
dimensions of a 2005 Fiat Grande Punto.)
S
length is 10/3
Notes:
378
176
3660.8 lb.
3
The car door, as a rectangle, is drawn in Figure 7.44. Its

and its height is 2.25 . We adopt the conven on that the
7.6 Fluid Forces

top of the door is at the surface of the water, both of which are at y = 0. Using
the weightdensity of water of 62.4 lb/ 3 , we have the total force as
F=
0
2.25
0
=
2.25
62.4(y)10/3 dy
(0, 0)
(3.3, 0)
208y dy
0

= 104y2
(0, 2.25)
2.25
(3.3, 2.25)
Figure 7.44: Sketching a submerged car

door in Example 226.
= 526.5 lb.
Most adults would nd it very dicult to apply over 500 lb of force to a car
door while seated inside, making the door eec vely impossible to open. This
is counterintui ve as most assume that the door would be rela vely easy to
open. The truth is that it is not, hence the survival ps men oned at the beginning of this sec on. .
. Example 227
.Finding uid force
An underwater observa on tower is being built with circular viewing portholes
enabling visitors to see underwater life. Each ver cally oriented porthole is to
have a 3 diameter whose center is to be located 50 underwater. Find the
total uid force exerted on each porthole. Also, compute the uid force on a
horizontally oriented porthole that is under 50 of water.
F = 62.4
1.5
1.5
1.5
= 62.4
1.5
1.5
= 6240
1.5
50
2
1
.
2
1
1
x
1
2
not to scale
Figure 7.45: Measuring the uid force on

an underwater porthole in Example 227.
2(50 y) 2.25 y2 dy
(
water line
d(y) = 50 y
We place the center of the porthole at the origin, meaning

S
the surface of the water is at y = 50 and the depth func on will be d(y) = 50y;
see Figure 7.45
2
2
The equa on of
a circle with a radius of 1.5 is x + y = 2.25; solving for
2
x we have x = 2.25 y , where the posi ve square root corresponds to
the right side of the circle and the nega ve square root corresponds
to the le
side of the circle. Thus the length func on at depth y is (y) = 2 2.25 y2 .
Integra ng on [1.5, 1.5] we have:
)
100 2.25 y2 2y 2.25 y2 dy
)
2.25 y2 dy 62.4
1.5
1.5
(
)
2y 2.25 y2 dy
Notes:
379

The second integral above can be evaluated using Subs tu on. Let u = 2.25y2
with du = 2y dy. The new bounds are: u(1.5) = 0 and u(1.5) = 0; the new
integral will integrate from u = 0 to u = 0, hence the integral is 0.
The rst integral above nds the area of half a circle of radius 1.5, thus the
rst integral evaluates to 6240 1.52 /2 = 22, 054. Thus the total uid force
on a ver cally oriented porthole is 22, 054 lb.
Finding the force on a horizontally oriented porthole is more straigh orward:
F = Pressure Area = 62.4 50 1.52 = 22, 054 lb.
That these two forces are equal is not coincidental; it turns out that the uid
force applied to a ver cally oriented circle whose center is at depth d is the
same as force applied to a horizontally oriented circle at depth d. .
Notes:
380
Exercises 7.6
Terms and Concepts
1. State in your own words Pascals Principle.
2. State in your own words how pressure is dierent from
force.
5
Problems
In Exercises 3 12, nd the uid force exerted on the given
plate, submerged in water with a weight density of 62.4
lb/ 3 .
8.
3.
.
2
5
9.
4.
.
1
.
5
5.
10.
6
5
1
4
6.
11.
6
.
2
.
1
5
7.
.
12.
.
2
In Exercises 13 18, the side of a container is pictured. Find

the uid force exerted on this plate when the container is full
of:
1. water, with a weight density of 62.4 lb/ 3 , and
2. concrete, with a weight density of 150 lb/ 3 .
2.
16.
y = 1 x2
y=
17.
13.
1 x2
.
2
.
6.
3
4
18.
14.
19. How deep must the center of a ver cally oriented circular
plate with a radius of 1 be submerged in water, with a
weight density of 62.4 lb/ 3 , for the uid force on the plate
to reach 1,000 lb?
y = x2
y = 4 x2
15.
.
4
y = 9 x2
20. How deep must the center of a ver cally oriented square
plate with a side length of 2 be submerged in water, with
a weight density of 62.4 lb/ 3 , for the uid force on the
plate to reach 1,000 lb?
8: S
8.1
Sequences
We commonly refer to a set of events that occur one a er the other as a sequence of events. In mathema cs, we use the word sequence to refer to an
ordered set of numbers, i.e., a set of numbers that occur one a er the other.
For instance, the numbers 2, 4, 6, 8, , form a sequence. The order is important; the rst number is 2, the second is 4, etc. It seems natural to seek a formula
that describes a given sequence, and o en this can be done. For instance, the
sequence above could be described by the func on a(n) = 2n, for the values of
n = 1, 2, . . . To nd the 10th term in the sequence, we would compute a(10).
This leads us to the following, formal deni on of a sequence.
Deni on 27
Nota on: We use N to describe the set of

natural numbers, that is, the integers 1, 2,
3,
Factorial: The expression 3! refers to the

number 3 2 1 = 6.
In general, n! = n(n1)(n2) 21,
where n is a natural number.
Sequence
A sequence is a func on a(n) whose domain is N. The range of a

sequence is the set of all dis nct values of a(n).
We dene 0! = 1. While this does not

immediately make sense, it makes many
mathema cal formulas work properly.
The terms of a sequence are the values a(1), a(2), , which are usually
denoted with subscripts as a1 , a2 , .
A sequence a(n) is o en denoted as {an }.
. Example 228
.Lis ng terms of a sequence
List the rst four terms of the following sequences.
{
}
{ n}
(1)n(n+1)/2
3
2. {an } = {4+(1)n }
3. {an } =
1. {an } =
n!
n2
S
y
5
4
3
2
an =
3
3
9
3
9
3
27
1. a1 =
= 3;
a2 =
= ;
a3 =
= ;
a4 =
=
1!
2!
2
3!
2
4!
8
We can plot the terms of a sequence with a sca er plot. The x-axis is
used for the values of n, and the values of the terms are plo ed on the
y-axis. To visualize this sequence, see Figure 8.1.
3n
n!
n
1
Figure 8.1: Plo ng a sequence from Example 228.
Chapter 8 Sequences and Series

2. a1 = 4 + (1)1 = 3;
a2 = 4 + (1)2 = 5;
a3 = 4+(1)3 = 3;
a4 = 4+(1)4 = 5. Note that the range of this
sequence is nite, consis ng of only the values 3 and 5. This sequence is
plo ed in Figure 8.2 (a).
5
4
3
(1)1(2)/2
(1)2(3)/2
1
= 1;
a2 =
=
2
1
22
4
(1)3(4)/2
1
(1)4(5)/2
1
a3 =
=
a4 =
=
;
32
9
42
16
(1)5(6)/2
1
a5 =
= .
2
5
25
We gave one extra term to begin to show the pa ern of signs is , , +,
+, , , . . ., due to the fact that the exponent of 1 is a special quadra c.
This sequence is plo ed in Figure 8.2 (b).
3. a1 =
2
n
an = 4 + (1)
n
1
(a)
y
1/2
1/4
n
1
an =
(1)n(n+1)/2
n2
. Example 229
.Determining a formula for a sequence
Find the nth term of the following sequences, i.e., nd a func on that describes
each of the given sequences.
1. 2, 5, 8, 11, 14, . . .
2. 2, 5, 10, 17, 26, 37, . . .
.
(b)
3. 1, 1, 2, 6, 24, 120, 720, . . .
Figure 8.2: Plo ng sequences in Example

228.
4.
5 5 15 5 25
, , , , , ...
2 2 8 4 32
We should rst note that there is never exactly one func on that
describes a nite set of numbers as a sequence. There are many sequences
that start with 2, then 5, as our rst example does. We are looking for a simple
formula that describes the terms given, knowing there is possibly more than one
answer.
S
1. Note how each term is 3 more than the previous one. This implies a linear
func on would be appropriate: a(n) = an = 3n + b for some appropriate
value of b. As we want a1 = 2, we set b = 1. Thus an = 3n 1.
2. First no ce how the sign changes from term to term. This is most commonly accomplished by mul plying the terms by either (1)n or (1)n+1 .
Using (1)n mul plies the odd terms by (1); using (1)n+1 mul plies
the even terms by (1). As this sequence has nega ve even terms, we
will mul ply by (1)n+1 .
Notes:
384
8.1 Sequences
A er this, we might feel a bit stuck as to how to proceed. At this point,
we are just looking for a pa ern of some sort: what do the numbers 2, 5,
10, 17, etc., have in common? There are many correct answers, but the
one that well use here is that each is one more than a perfect square.
That is, 2 = 11 + 1, 5 = 22 + 1, 10 = 32 + 1, etc. Thus our formula is
an = (1)n+1 (n2 + 1).
3. One who is familiar with the factorial func on will readily recognize these
numbers. They are 0!, 1!, 2!, 3!, etc. Since our sequences start with n = 1,
we cannot write an = n!, for this misses the 0! term. Instead, we shi by
1, and write an = (n 1)!.
4. This one may appear dicult, especially as the rst two terms are the
same, but a li le sleuthing will help. No ce how the terms in the numerator are always mul ples of 5, and the terms in the denominator are
always powers of 2. Does something as simple as an = 5n
2n work?
When n = 1, we see that we indeed get 5/2 as desired. When n = 2,

we get 10/4 = 5/2. Further checking shows that this formula indeed
matches the other terms of the sequence.
A common mathema cal endeavor is to create a new mathema cal object

(for instance, a sequence) and then apply previously known mathema cs to the
new object. We do so here. The fundamental concept of calculus is the limit, so
we will inves gate what it means to nd the limit of a sequence.
Deni on 28
Limit of a Sequence, Convergent, Divergent
Let {an } be a sequence and let L be a real number. Given any > 0, if
an m can be found such that |an L| < for all n > m, then we say the
limit of {an }, as n approaches innity, is L, denoted
lim an = L.
If lim an exists, we say the sequence converges; otherwise, the sen

quence diverges.
This deni on states, informally, that if the limit of a sequence is L, then if

you go far enough out along the sequence, all subsequent terms will be really
close to L. Of course, the terms far enough and really close are subjec ve
terms, but hopefully the intent is clear.
Notes:
385

This deni on is reminiscent of the proofs of Chapter 1. In that chapter
we developed other tools to evaluate limits apart from the formal deni on; we
do so here as well.
.
Theorem 55
Limit of a Sequence
Let {an } be a sequence and let f(x) be a func on where f(n) = an for all
n in N.
1. If lim f(x) = L, then lim an = L.

x
2. If lim f(x) does not exist, then {an } diverges.
10
5
n
20
40
60
80
100
5
10
an =
3x2 2x + 1
x2 1000
. Example 230
.Determining convergence/divergence of a sequence
Determine the convergence or divergence of the following sequences.
}
{
}
{ 2
(1)n
3n 2n + 1
2. {an } = {cos n}
3. {an } =
1. {an } =
n2 1000
n
(a)
an = cos n
When we considered limits before, the domain of the func on was an interval of real numbers. Now, as we consider limits, the domain is restricted to N,
the natural numbers. Theorem 55 states that this restric on of the domain does
not aect the outcome of the limit and whatever tools we developed in Chapter
1 to evaluate limits can be applied here as well.
0.5
n
20
40
60
80
100
3x2 2x + 1
= 3. (We could
x x2 1000
have also directly applied lHpitals Rule.) Thus the sequence {an } converges, and its limit is 3. A sca er plot of every 5 values of an is given in
Figure 8.3 (a). The values of an vary widely near n = 30, ranging from
about 73 to 125, but as n grows, the values approach 3.
1. Using Theorem 11, we can state that lim
0.5
1
.
(b)
2. The limit lim cos x does not exist, as the func on oscillates (and takes on
every value in [1, 1] innitely many mes). Thus we conclude that the
sequence {cos n} diverges. (And in this par cular case, since the domain
is restricted to N, no value of cos n is repeated!) This sequence is plo ed
in Figure 8.3 (b); because only discrete values of cosine are plo ed, it does
not bear strong resemblance to the familiar cosine wave.
y
1
0.5
n
5
10
0.5
an =
1
15
20
(1)n
n
.
(c)
Figure 8.3: Sca er plots of the sequences

in Example 230.
386
3. We cannot actually apply Theorem 55 here, as the func on f(x) = (1)x /x
Notes:
8.1 Sequences
is not well dened. (What does (1) 2 mean? In actuality, there is an answer, but it involves complex analysis, beyond the scope of this text.) So
for now we say that we cannot determine the limit. (But we will be able
to very soon.) By looking at the plot in Figure 8.3 (c), we would like to
conclude that the sequence converges to 0. That is true, but at this point
we are unable to decisively say so.
}
(1)n
converges to 0 but we
n
lack the formal tool to prove it. The following theorem gives us that tool.
It seems very clear that a sequence such as
Theorem 56
Absolute Value Theorem

.
Let {an } be a sequence. If lim |an | = 0, then lim an = 0
n
. Example 231
.Determining the convergence/divergence of a sequence
Determine the convergence or divergence of the following sequences.
{
}
{
}
(1)n
(1)n (n + 1)
1. {an } =
2. {an } =
n
n
S
1. This appeared in Example 230. We want to apply Theorem 56, so consider

the limit of {|an |}:

(1)n

lim |an | = lim
n
n
n
1
= lim
n n
= 0.
Since this limit is 0, we can apply Theorem 56 and state that lim an = 0.
n
2. Because of the alterna ng nature of this sequence (i.e., every other term
(1)x (x + 1)
is mul plied by 1), we cannot simply look at the limit lim
.
x
x
We can try to apply the techniques of Theorem 56:

(1)n (n + 1)

lim |an | = lim

n
n
n
= lim
n+1
n
= 1.
Notes:
387

We have concluded that when we ignore the alterna ng sign, the sequence
approaches 1. This means we cannot apply Theorem 56; it states the the
limit must be 0 in order to conclude anything.
y
2
an =
(1)n (n + 1)
n
In fact, since we know that the signs of the terms alternate and we know
that the limit of |an | is 1, we know that as n approaches innity, the terms
will alternate between values close to 1 and 1, meaning the sequence
diverges. A plot of this sequence is given in Figure 8.4.
1
n
5
10
15
20
1
2
Figure 8.4: A plot of a sequence in Exam.

ple 231, part 2.
We con nue our study of the limits of sequences by considering some of the
proper es of these limits.
Theorem 57
Proper es of the Limits of Sequences
Let {an } and {bn } be sequences such that lim an = L, lim bn = K, and
n
n
let c be a real number.
1. lim (an bn ) = L K
3. lim (an /bn ) = L/K, K = 0
2. lim (an bn ) = L K
4. lim c an = c L
n
n
. Example 232
.Applying proper es of limits of sequences
Let the following sequences, and their limits, be given:
{
}
n+1
{an } =
, and lim an = 0;
n
n2
{(
)n }
1
{bn } =
1+
, and lim bn = e; and
n
n
{
}
{cn } = n sin(5/n) , and lim cn = 5.
n

1. lim (an + bn )
n
2. lim (bn cn )
n
3. lim (1000 an )
n
We will use Theorem 57 to answer each of these.
1. Since lim an = 0 and lim bn = e, we conclude that lim (an + bn ) =

n
n
n
0 + e = e. So even though we are adding something to each term of the
sequence bn , we are adding something so small that the nal limit is the
same as before.
Notes:
388
8.1 Sequences
2. Since lim bn = e and lim cn = 5, we conclude that lim (bn cn ) =
n
n
n
e 5 = 5e.
3. Since lim an = 0, we have lim 1000an = 1000 0 = 0. It does not
n
n
ma er that we mul ply each term by 1000; the sequence s ll approaches
0. (It just takes longer to get close to 0.)
.
There is more to learn about sequences than just their limits. We will also
study their range and the rela onships terms have with the terms that follow.
We start with some deni ons describing proper es of the range.
Deni on 29
Bounded and Unbounded Sequences
A sequence {an } is said to be bounded if there exists real numbers m

and M such that m < an < M for all n in N.
A sequence {an } is said to be unbounded if it is not bounded.
y
1
A sequence {an } is said to be bounded above if there exists an M such

that an < M for all n in N; it is bounded below if there exists an m such
that m < an for all n in N.
an =
1
n
1/2
It follows from this deni on that an unbounded sequence may be bounded

above or bounded below; a sequence that is both bounded above and below is
simply a bounded sequence.
. Example 233
Determining boundedness of sequences
Determine the boundedness of the following sequences.
{ }
1
1. {an } =
2. {an } = {2n }
n
1/4
1/10
2. The terms of this sequence obviously grow without bound. However, it

is also true that these terms are all posi ve, meaning 0 < an . Thus we
can say the sequence is unbounded, but also bounded below. Figure 8.6
illustrates this.
.
9 10
200
n
an = 2
100
n
2
Notes:
Figure 8.5: A plot of {an } = {1/n} from

Example 233.
1. The terms of this sequence are always posi ve but are decreasing, so we
have 0 < an < 2 for all n. Thus this sequence is bounded. Figure 8.5
illustrates this.
n
1
Figure 8.6: A plot of {an } = {2n } from

Example 233.
389

The previous example produces some interes ng concepts. First, we can
recognize that the sequence {1/n} converges to 0. This says, informally, that
most of the terms of the sequence are really close to 0. This implies that
the sequence is bounded, using the following logic. First, most terms are near
0, so we could nd some sort of bound on these terms (using Deni on 28, the
bound is ). That leaves a few terms that are not near 0 (i.e., a nite number
of terms). A nite list of numbers is always bounded.
This logic implies that if a sequence converges, it must be bounded. This is
indeed true, as stated by the following theorem.
.
Note: Keep in mind what Theorem 58

does not say. It does not say that
bounded sequences must converge, nor
does it say that if a sequence does not
converge, it is not bounded.
Theorem 58
Convergent Sequences are Bounded

.
Let {an } be a convergent sequence. Then {an } is bounded.

{
n}
In Example 232 we saw the sequence {bn } = (1 + 1/n) , where it was
stated that lim bn = e. (Note that this is simply resta ng part of Theorem 5.)
n
Even though it may be dicult to intui vely grasp the behavior of this sequence,
we know immediately that it is bounded.
Another interes ng concept to come out of Example 233 again involves the
sequence {1/n}. We stated, without proof, that the terms of the sequence were
decreasing. That is, that an+1 < an for all n. (This is easy to show. Clearly
n < n + 1. Taking reciprocals ips the inequality: 1/n > 1/(n + 1). This is the
same as an > an+1 .) Sequences that either steadily increase or decrease are
important, so we give this property a name.
Deni on 30
Monotonic Sequences
1. A sequence {an } is monotonically increasing if an an+1 for all

n, i.e.,
a1 a2 a3 an an+1
Note: It is some mes useful to call

a monotonically increasing sequence
strictly increasing if an < an+1 for all
n; i.e, we remove the possibility that
subsequent terms are equal.
A similar statement holds for strictly decreasing.
2. A sequence {an } is monotonically decreasing if an an+1 for all

n, i.e.,
a1 a2 a3 an an+1
3. A sequence is monotonic if it is monotonically increasing or monotonically decreasing.
Notes:
390
8.1 Sequences
. Example 234
.Determining monotonicity
Determine the monotonicity of the following sequences.
{
1. {an } =
{
2. {an } =
n+1
n
n2 + 1
n+1
n2 9
3. {an } =
n2 10n + 26
{ 2}
n
4. {an } =
n!
In each of the following, we will examine an+1 an . If an+1

an > 0, we conclude that an < an+1 and hence the sequence is increasing. If
an+1 an < 0, we conclude that an > an+1 and the sequence is decreasing. Of
course, a sequence need not be monotonic and perhaps neither of the above
will apply.
We also give a sca er plot of each sequence. These are useful as they suggest a pa ern of monotonicity, but analy c work should be done to conrm a
graphical trend.
n+1
n
an =
n+2 n+1
n+1
n
(n + 2)(n) (n + 1)2
=
(n + 1)n
1
=
n(n + 1)
< 0 for all n.
10
(a)
y
10
an+1 an =
1.
Since an+1 an < 0 for all n, we conclude that the sequence is decreasing.
2
an =
n2 + 1
n+1
n
10
(b)
(n + 1) + 1 n + 1
n+2
n+1
(
)
2
(n + 1) + 1 (n + 1) (n2 + 1)(n + 2)
=
(n + 1)(n + 2)
2
n + 4n + 1
=
(n + 1)(n + 2)
> 0 for all n.
an+1 an =
2.
Since an+1 an > 0 for all n, we conclude the sequence is increasing. .

3. We can clearly see in Figure 8.7 (c), where the sequence is plo ed, that
it is not monotonic. However, it does seem that a er the rst 4 terms
it is decreasing. To understand why, perform the same analysis as done
before:
y
15
10
an =
n2 9
n2 10n + 26
n
5
10
(c)
Figure 8.7: Plots of sequences in Example
234.
Notes:
391

(n + 1)2 9
n2 9
2
2
(n + 1) 10(n + 1) + 26 n 10n + 26
n2 + 2n 8
n2 9
2
= 2
n 8n + 17 n 10n + 26
(n2 + 2n 8)(n2 10n + 26) (n2 9)(n2 8n + 17)
=
(n2 8n + 17)(n2 10n + 26)
2
10n + 60n 55
= 2
.
(n 8n + 17)(n2 10n + 26)
an+1 an =
We want to know when this is greater than, or less than, 0, therefore we

are only concerned with the numerator. Using the quadra c formula, we
can determine that 10n2 + 60n 55 = 0 when n 1.13, 4.87. So for
n < 1.13, the sequence is decreasing. Since we are only dealing with the
natural numbers, this means that a1 > a2 .
y
2
1.5
an =
1
Between 1.13 and 4.87, i.e., for n = 2, 3 and 4, we have that an+1 >
an and the sequence is increasing. (That is, when n = 2, 3 and 4, the
numerator 10n2 + 60n + 55 from the frac on above is > 0.)
n2
n!
When n > 4.87, i.e, for n 5, we have that 10n2 + 60n + 55 < 0,
hence an+1 an < 0, so the sequence is decreasing.
In short, the sequence is simply not monotonic. However, it is useful to
note that for n 5, the sequence is monotonically decreasing.
0.5
n
5
10
4. Again, the plot in Figure 8.8 shows that the sequence is not monotonic,
but it suggests that it is monotonically decreasing a er the rst term. We
perform the usual analysis to conrm this.
Figure 8.8: A plot of {an } = {n /n!} in

Example 234.
2
(n + 1)2
n2
(n + 1)!
n!
2
(n + 1) n2 (n + 1)
=
(n + 1)!
3
n + 2n + 1
=
(n + 1)!
an+1 an =
When n = 1, the above expression is > 0; for n 2, the above expression is < 0. Thus this sequence is not monotonic, but it is monotonically
decreasing a er the rst term.
Knowing that a sequence is monotonic can be useful. In par cular, if we

know that a sequence is bounded and monotonic, we can conclude it converges!
Consider, for example, a sequence that is monotonically decreasing and is bounded
below. We know the sequence is always ge ng smaller, but that there is a
Notes:
392
8.1 Sequences
bound to how small it can become. This is enough to prove that the sequence
will converge, as stated in the following theorem.
Theorem 59
Bounded Monotonic Sequences are Convergent
1. Let {an } be a bounded, monotonic sequence. Then {an } converges; i.e., lim an exists.
n
2. Let {an } be a monotonically increasing sequence that is bounded

above. Then {an } converges.
3. Let {an } be a monotonically decreasing sequence that is bounded
below. Then {an } converges.
Consider once again the sequence {an } = {1/n}. It is easy to show it is
monotonically decreasing and that it is always posi ve (i.e., bounded below by
0). Therefore we can conclude by Theorem 59 that the sequence converges. We
already knew this by other means, but in the following sec on this theorem will
become very useful.
Sequences are a great source of mathema cal inquiry. The On-Line Encyclopedia of Integer Sequences (http://oeis.org) contains thousands of sequences and their formulae. (As of this wri ng, there are 218,626 sequences
in the database.) Perusing this database quickly demonstrates that a single sequence can represent several dierent real life phenomena.
Interes ng as this is, our interest actually lies elsewhere. We are more interested in the sum of a sequence. That is, given a sequence {an }, we are very
interested in a1 + a2 + a3 + . Of course, one might immediately counter
with Doesnt this just add up to innity? Many mes, yes, but there are many
important cases where the answer is no. This is the topic of series, which we
begin to inves gate in the next sec on.
Notes:
393
Exercises 8.1
Terms and Concepts
20. {an } =
1. Use your own words to dene a sequence.

2. The domain of a sequence is the
numbers.
3. Use your own words to describe the range of a sequence.

4. Describe what it means for a sequence to be bounded.
Problems
In Exercises 5 8, give the rst ve terms of the given sequence.
}
{
4n
5. {an } =
(n + 1)!
{(
)n }
3
6. {bn } =
2
{
}
n+1
n
7. {cn } =
n+2
{
((
)n (
)n )}
1 5
1
1+ 5
8. {dn } =
2
2
5
In Exercises 9 12, determine the nth term of the given sequence.
9. 4, 7, 10, 13, 16, . . .
3 3
3
10. 3, , , , . . .
2 4
8
11. 10, 20, 40, 80, 160, . . .
1 1 1
1
12. 1, 1, , ,
,
, ...
2 6 24 120
In Exercises 13 16, use the following informa on to determine the limit of the given sequences.
{ n
}
2 20
{an } =
;
lim an = 1
n
2n
{(
)n }
2
{bn } =
lim bn = e2
1+
;
n
n
{cn } = {sin(3/n)};
lim cn = 0
n
{ n
}
2 20
13. {an } =
7 2n
14. {an } = {3bn an }
{
(
)n }
2
15. {an } = sin(3/n) 1 +
n
{(
)2n }
2
16. {an } =
1+
n
In Exercises 17 28, determine whether the sequence converges or diverges. If convergent, give the limit of the sequence.
}
{
n
17. {an } = (1)n
n+1
{ 2
}
4n n + 5
18. {an } =
3n2 + 1
{ n}
4
19. {an } =
5n
}
n1
n
,n2
n
n1
21. {an } = {ln(n)}

{
}
3n
22. {an } =
n2 + 1
{(
)n }
1
23. {an } =
1+
n
}
{
1
24. {an } = 5
n
}
{
(1)n+1
25. {an } =
n
{ n}
1.1
26. {an } =
n
{
}
2n
27. {an } =
n+1
{
}
n2
28. {an } = (1)n n
2 1
In Exercises 29 34, determine whether the sequence is
bounded, bounded above, bounded below, or none of the
above.
29. {an } = {sin n}
30. {an } = {tan n}
{
}
n 3n 1
31. {an } = (1)
n
{ 2
}
3n 1
32. {an } =
n
33. {an } = {n cos n}
34. {an } = {2n n!}
In Exercises 35 38, determine whether the sequence is
monotonically increasing or decreasing. If it is not, determine
if there is an m such that it is monotonic for all n m.
{
}
n
35. {an } =
n+2
{ 2
}
n 6n + 9
36. {an } =
n
{
}
n 1
37. {an } = (1) 3
n
{ 2}
n
38. {an } =
2n
39. Prove Theorem 56; that is, use the deni on of the limit of a
sequence to show that if lim |an | = 0, then lim an = 0.
n
40. Let {an } and {bn } be sequences such that lim an = L and
n
lim bn = K.
(a) Show that if an < bn for all n, then L K.

(b) Give an example where L = K.
41. Prove the Squeeze Theorem for sequences: Let {an } and
{bn } be such that lim an = L and lim bn = L, and let
n
{cn } be such that an cn bn for all n. Then lim cn = L

n
8.2 Innite Series
8.2
Innite Series
Given the sequence {an } = {1/2n } = 1/2, 1/4, 1/8, . . ., consider the following sums:
a1
a1 + a2
a1 + a2 + a3
a1 + a2 + a3 + a4
=
1/2
=
1/2 + 1/4
=
1/2 + 1/4 + 1/8
= 1/2 + 1/4 + 1/8 + 1/16
=
1/2
=
3/4
=
7/8
= 15/16
In general, we can show that

2n 1
1
= 1 n.
2n
2
Let Sn be the sum of the rst n terms of the sequence {1/2n }. From the above,
we see that S1 = 1/2, S2 = 3/4, etc. Our formula at the end shows that Sn =
1 1/2n .
(
)
Now consider the following limit: lim Sn = lim 1 1/2n = 1. This limit
n
n
can be interpreted as saying something amazing: the sum of all the terms of the
sequence {1/2n } is 1.
This example illustrates some interes ng concepts that we explore in this
sec on. We begin this explora on with some deni ons.
a1 + a2 + a3 + + an =
Deni on 31
Innite Series, nth Par al Sums, Convergence, Divergence
Let {an } be a sequence.

1. The sum
an is an innite series (or, simply series).
n=1
2. Let Sn =
ai ; the sequence {Sn } is the sequence of nth par al sums of {an }.
i=1
3. If the sequence {Sn } converges to L, we say the series

and we write
an converges to L,
n=1
an = L.
n=1
4. If the sequence {Sn } diverges, the series
an diverges.
n=1
Notes:
395
Using our new terminology, we can state that the series

and
1/2n converges,
n=1
n
1/2 = 1.
n=1
We will explore a variety of series in this sec on. We start with two series
that diverge, showing how we might discern divergence.
. Example 235
.Showing series diverge
1. Let {an } = {n2 }. Show
an diverges.
n=1
2. Let {bn } = {(1)n+1 }. Show
bn diverges.
n=1
1. Consider Sn , the nth par al sum.

Sn = a1 + a2 + a3 + + an
= 12 + 22 + 32 + n2 .
By Theorem 37, this is
=
n(n + 1)(2n + 1)
.
6
Since lim Sn = , we conclude that the series

n
instruc ve to write
n2 diverges. It is
n=1
n2 = for this tells us how the series diverges: it
n=1
grows without bound.
A sca er plot of the sequences {an } and {Sn } is given in Figure 8.9(a).
The terms of {an } are growing, so the terms of the par al sums {Sn } are
growing even faster, illustra ng that the series diverges.
Notes:
396
8.2 Innite Series

2. Consider some of the par al sums Sn of {bn }:
S1 = 1
S2 = 0
S3 = 1
S4 = 0
300
1 n is odd
. As {Sn } oscil0 n is even
lates, repea ng 1, 0, 1, 0, . . ., we conclude that lim Sn does not exist,
This pa ern repeats; we nd that Sn =
hence
(1)n+1 diverges.
100
n=1
200
n
5
A sca er plot of the sequence {bn } and the par al sums {Sn } is given in
Figure 8.9(b). When n is odd, bn = Sn so the marks for bn are drawn
oversized to show they coincide.
. an
10
Sn
(a)
While it is important to recognize when a series diverges, we are generally

more interested in the series that converge. In this sec on we will demonstrate
a few general techniques for determining convergence; later sec ons will delve
deeper into this topic.
y
1
0.5
Geometric Series
n
5
One important type of series is a geometric series.

Deni on 32
10
0.5
1
.
Geometric Series
. bn
Sn
A geometric series is a series of the form
rn = 1 + r + r2 + r3 + + rn +
n=0
Note that the index starts at n = 0, not n = 1.
(b)
Figure 8.9: Sca er plots rela ng to Example 235.
We started this sec on with a geometric series, although we dropped the

rst term of 1. One reason geometric series are important is that they have nice
convergence proper es.
Notes:
397

.
Theorem 60
Convergence of Geometric Series
Consider the geometric series
rn .
n=0
1. The nth par al sum is: Sn =
1 r n+1
.
1r
2. The series converges if, and only if, |r| < 1. When |r| < 1,
rn =
n=0
According to Theorem 60, the series
1
.
1r
1
1 1
= 1 + + + converges,
n
2
2 4
n=0
1
1
=
= 2. This concurs with our introductory example; while
n
2
1 1/2
n=0
there we got a sum of 1, we skipped the rst term of 1.
and
. Example 236
.Exploring geometric series
Check the convergence of the following series. If the series converges, nd its
sum.
1.
( )n
3
n=2
n=0
3.
3n
n=0
1. Since r = 3/4 < 1, this series converges. By Theorem 60, we have that
( )n
1
3
=
= 4.
4
1 3/4
n=0
However, note the subscript of the summa on in the given series: we are
to start with n = 2. Therefore we subtract o the rst two terms, giving:
( )n
3
3
9
=41 = .
4
4
4
n=2
n
2
. an
10
Sn
This is illustrated in Figure 8.10.
Figure 8.10: Sca er plots rela ng to the

series in Example 236.
Notes:
398
)n
(
2.
8.2 Innite Series

2. Since |r| = 1/2 < 1, this series converges, and by Theorem 60,
)n
(
1
1
2
=
= .
2
1 (1/2)
3
n=0
The par al sums of this series are plo ed in Figure 8.11(a). Note how
the par al sums are not purely increasing as some of the terms of the
sequence {(1/2)n } are nega ve.
3. Since r > 1, the series diverges. (This makes common sense; we expect
the sum
1 + 3 + 9 + 27 + 81 + 243 +
.
y
1
0.5
n
4
10
0.5
. an
Sn
1
.
(a)
to diverge.) This is illustrated in Figure 8.11(b).

y
pSeries
1,000
Another important type of series is the p-series.

Deni on 33
pSeries, General pSeries
500
1. A pseries is a series of the form
1
,
p
n
n=1
where p > 0.
n
2
. an
Sn
2. A general pseries is a series of the form
n=1
(b)
1
,
(an + b)p
where p > 0 and a, b are real numbers.

Like geometric series, one of the nice things about pseries is that they have
easy to determine convergence proper es.
Theorem 61
Convergence of General pSeries
A general pseries
n=1
.
1
will
converge if, and only if, p > 1.
(an + b)p
Note: Theorem 61 assumes that an+b =

0 for all n. If an + b = 0 for some n, then
of course the series does not converge regardless of p as not all of the terms of the
sequence are dened.
Notes:
399
. Example 237
Determining convergence of series
Determine the convergence of the following series.
1.
1
n=1
2.
1
n2
n=1
3.
4.
n
n=1
(1)n
n=1
5.
1
1
(
n
5)3
n=10 2
6.
1
2n
n=1
1. This is a pseries with p = 1. By Theorem 61, this series diverges.

This series is a famous series, called the Harmonic Series, so named because of its rela onship to harmonics in the study of music and sound.
2. This is a pseries with p = 2. By Theorem 61, it converges. Note that
the theorem does not give a formula by which we can determine what
the series converges to; we just know it converges. A famous, unexpected
2
result is that this series converges to .
6
3. This is a pseries with p = 1/2; the theorem states that it diverges.
4. This is not a pseries; the deni on does not allow for alterna ng signs.
Therefore we cannot apply Theorem 61. (Another famous result states
that this series, the Alterna ng Harmonic Series, converges to ln 2.)
5. This is a general pseries with p = 3, therefore it converges.
. 6. This is not a pseries, but a geometric series with r = 2. It converges.
Later sec ons will provide tests by which we can determine whether or not
a given series converges. This, in general, is much easier than determining what
a given series converges to. There are many cases, though, where the sum can
be determined.
. Example 238
.Telescoping series
)
(
1
1
.
Evaluate the sum
n n+1
n=1
Notes:
400
8.2 Innite Series

It will help to write down some of the rst few par al sums
of this series.
1 1
1 2 ) (
(
)
1 1
1 1
S2 =
1 2
2 3
(
) (
) (
)
1 1
1 1
1 1
S3 =
1 2
2 3
3 4
) (
) (
) (
)
(
1 1
1 1
1 1
1 1
S4 =
1 2
2 3
3 4
4 5
S1 =
1
2
1
=1
3
1
=1
4
1
=1
5
=1
Note how most of the terms in each par al sum are canceled out! In general,
1
we see that Sn = 1
. The sequence {Sn } converges, as lim Sn =
n
n
+
1
(
)
)
(
1
1
1
lim 1
= 1, and so we conclude that
= 1. Parn
n+1
n n+1
n=1
al sums of the series are plo ed in Figure 8.12. .
y
1
0.5
n
2
. an
10
Sn

series of Example 238.
The series in Example 238 is an example of a telescoping series. Informally, a

telescoping series is one in which the par al sums reduce to just a nite number
of terms. The par al sum Sn did not contain n terms, but rather just two: 1 and
1/(n + 1).
When possible, seek a way to write an explicit formula for the nth par al sum
Sn . This makes evalua ng the limit lim Sn much more approachable. We do so
n
in the next example.
. Example 239
.Evalua ng series
Evaluate each of the following innite series.
1.
n=1
2
n2 + 2n
2.
n=1
(
ln
n+1
n
1. We can decompose the frac on 2/(n2 + 2n) as

n2
2
1
1
=
.
+ 2n
n n+2
(See Sec on 6.5, Par al Frac on Decomposi on, to recall how this is done,
if necessary.)
Notes:
401

Expressing the terms of {Sn } is now more instruc ve:
1
S1 = 1
3
(
) (
1
1
S2 = 1
+
3
2
(
) (
1
1
S3 = 1
+
3
2
(
) (
1
1
S4 = 1
+
3
2
(
) (
1
1
+
S5 = 1
3
2
=1
)
1
4
) (
)
1
1
1
4
3
5
) (
) (
)
1
1
1
1
1
4
3
5
4
6
) (
) (
) (
)
1
1
1
1
1
1
1
4
3
5
4
6
5
7
=1+
=1+
=1+
=1+
1
3
1
2
1
2
1
2
1
2
1
3
1
4
1
5
1
1
4
1
5
1
6
1
y
1.5
We again have a telescoping series. In each par al sum, most of the terms
1
1
1
cancel and we obtain the formula Sn = 1 +
. Taking
2
n+1
n+2
limits allows us to determine the convergence of the series:
(
)
1
1
1
3
3
1
= , so
= .
lim Sn = lim 1 +
2
n
n
2 n+1 n+2
2
n + 2n
2
n=1
0.5
This is illustrated in Figure 8.13(a).
. an
10
2. We begin by wri ng the rst few par al sums of the series:
Sn
S1 = ln (2)
(a)
( )
3
S2 = ln (2) + ln
2
( )
( )
3
4
S3 = ln (2) + ln
+ ln
2
3
( )
( )
( )
3
4
5
S4 = ln (2) + ln
+ ln
+ ln
2
3
4
At rst, this does not seem helpful, but recall the logarithmic iden ty:
ln x + ln y = ln(xy). Applying this to S4 gives:
( )
( )
( )
(
)
3
4
5
2 3 4 5
S4 = ln (2)+ln
+ln
+ln
= ln

= ln (5) .
2
3
4
1 2 3 4
{
}
We can conclude that {Sn } = ln(n + 1) . This sequence does not con(
)
n+1
verge, as lim Sn = . Therefore
ln
= ; the series din
n
n=1
verges. Note in Figure 8.13(b) how the sequence of par al sums grows
50
. an
100
Sn
(b)
Notes:
402
8.2 Innite Series
slowly; a er 100 terms, it is not yet over 5. Graphically we may be fooled

into thinking the series converges, but our analysis above shows that it
does not.
We are learning about a new mathema cal object, the series. As done before, we apply old mathema cs to this new topic.
Theorem 62
Let
Proper es of Innite Series
an = L,
bn = K, and let c be a constant.
n=1
n=1
1. Constant Mul ple Rule:
c an = c
n=1
2. Sum/Dierence Rule:
an = c L.
n=1
(
)
an bn =
an
bn = L K.
n=1
n=1
n=1
Before using this theorem, we provide a few famous series.

Key Idea 31
Important Series
1.
1
= e.
n!
n=0
2.
2
1
=
.
n2
6
n=1
3.
(1)n+1
n2
n=1
4.
5.
2
.
12
(1)n
= .
2n
+
1
4
n=0
1
n=1
6.
(Note that the index starts with n = 0.)
diverges.
(1)n+1
n=1
.
(This is called the Harmonic Series.)
= ln 2. (This is called the Alterna ng Harmonic Series.)
Notes:
403
. Example 240
.Evalua ng series
Evaluate the given series.
(
)
(1)n+1 n2 n
1.
n3
n=1
y
2.
1000
n=1
3.
n!
1
1
1
1
+
+
+
+
16 25 36 49
0.2
1. We start by using algebra to break the series apart:

(
)
)
(1)n+1 n2 n
(1)n+1 n
(1)n+1 n2
=
n3
n3
n3
n=1
n=1
n
10
20
30
40
50
0.2
=
.
. an
Sn
(1)n+1
n=1
= ln(2)
(a)
y
(1)n+1
n=1
n2
2
0.1293.
12
This is illustrated in Figure 8.14(a).
2,000
1,500
2. This looks very similar to the series that involves e in Key Idea 31. Note,
however, that the series given in this example starts with n = 1 and not
n = 0. The rst term of the series in the Key Idea is 1/0! = 1, so we will
subtract this from our result below:
1,000
500
1000
n
5
. an
10
n=1
n!
= 1000
1
n!
n=1
= 1000 (e 1) 1718.28.
Sn
This is illustrated in Figure 8.14(b). The graph shows how this par cular
series converges very rapidly.
(b)
1
2
n
n=4
(note we start with n = 4, not n = 1). This series will converge. Using the
3. The denominators in each term are perfect squares; we are adding
Notes:
404
8.2 Innite Series

formula from Key Idea 31, we have the following:
3
1
1
1
=
+
2
2
n
n
n2
n=4
n=1
n=1
1
1
1
=
2
2
2
n
n
n
n=4
n=1
n=1
(
)
2
1 1 1
1
+ +
=
6
1 4 9
n2
n=4
2
49 1
=
6
36
n2
n=4
0.2838
.
1
2
n
n=4
It may take a while before one is comfortable with this statement, whose
truth lies at the heart of the study of innite series: it is possible that the sum of
an innite list of nonzero numbers is nite. We have seen this repeatedly in this
sec on, yet it s ll may take some ge ng used to.
As one contemplates the behavior of series, a few facts become clear.
1. In order to add an innite list of nonzero numbers and get a nite result,
most of those numbers must be very near 0.
2. If a series diverges, it means that the sum of an innite list of numbers is
not nite (it may approach or it may oscillate), and:
(a) The series will s ll diverge if the rst term is removed.
(b) The series will s ll diverge if the rst 10 terms are removed.
(c) The series will s ll diverge if the rst 1, 000, 000 terms are removed.
(d) The series will s ll diverge if any nite number of terms from anywhere in the series are removed.
These concepts are very important and lie at the heart of the next two theorems.
Notes:
405

.
Theorem 63
nth Term Test for Convergence/Divergence
Consider the series
an .
n=1
1. If
an converges, then lim an = 0.

n
n=1
2. If lim an = 0, then
n
an diverges.
n=1
Note that the two statements in Theorem 63 are really the same. In order
to converge, the limit of the terms of the sequence must approach 0; if they do
not, the series will not converge.
Looking back, we can apply this theorem to the series in Example 235. In
that example, the nth terms of both sequences do not converge to 0, therefore
we can quickly conclude that each series diverges.
Important! This theorem does not state that if lim an = 0 then

an
n
n=1
converges. The standard example of this is the Harmonic Series, as given in Key
Idea 31. The Harmonic Sequence, {1/n}, converges to 0; the Harmonic Series,
1/n, diverges.
n=1
Theorem 64
Innite Nature of Series
The convergence or divergence remains unchanged by the addi on or

subtrac on of any nite number of terms. That is:
1. A divergent series will remain divergent with the addi on or subtrac on of any nite number of terms.
2. A convergent series will remain convergent with the addi on or
subtrac on of any nite number of terms. (Of course, the sum will
likely change.)
Consider once more the Harmonic Series
1
n=1
Notes:
406
which diverges; that is, the
8.2 Innite Series

sequence of par al sums {Sn } grows (very, very slowly) without bound. One
might think that by removing the large terms of the sequence that perhaps
the series will converge. This is simply not the case. For instance, the sum of the
rst 10 million terms of the Harmonic Series is about 16.7. Removing the rst 10
million terms from the Harmonic Series changes the nth par al sums, eec vely
subtrac ng 16.7 from the sum. However, a sequence that is growing without
bound will s ll grow without bound when 16.7 is subtracted from it.
The equa ons below illustrate this. The rst line shows the innite sum of
the Harmonic Series split into the sum of the rst 10 million terms plus the sum
of everything else. The next equa on shows us subtrac ng these rst 10 million terms from both sides. The nal equa on employs a bit of psuedomath:
subtrac ng 16.7 from innity s ll leaves one with innity.
1
n=1
1
n=1
10,000,000
n=1
16.7
10,000,000
n=1
1
n
1
n
n=10,000,001
1
1
=
n
n
n=10,000,001
Notes:
407
Exercises 8.2
Terms and Concepts
16.
1. Use your own words to describe how sequences and series

are related.
n!
n
10
n=1
17.
5 n n5
5 n + n5
n=1
2. Use your own words to dene a par al sum.

3. Given a series
an , describe the two sequences related
n=1
to the series that are important.

4. Use your own words to explain what a geometric series is.
5. T/F: If {an } is convergent, then
an is also convergent.
2n + 1
2n+1
n=1
)n
(
1
19.
1+
n
n=1
18.
In Exercises 20 29, state whether the given series converges

or diverges.
20.
1
5
n
n=1
21.
1
n
5
n=0
22.
6n
5n
n=0
n=1
Problems
In Exercises 6 13, a series
an is given.
n=1
(a) Give the rst 5 par al sums of the series.

(b) Give a graph of the rst 5 terms of an and Sn on the
same axes.
6.
7.
(1)n
n
n=1
n=1
8.
24.
1
n2
cos(n)
n=1
29.
9
10
In Exercises 14 19, use Theorem 63 to show the given series

diverges.
15.
n=1
3n2
n(n + 2)
2n
n2
n=1
)n
)n
(
1
13.
10
n=1
n=1
2
(2n + 8)2
1
28.
2n
n=1
1
n
3
n=1
10
n!
n=1
(
)
1
1
26.
+
n!
n
n=1
27.
n=1
14.
25.
1
10.
n!
n=1
12.
n4
n=1
n=1
11.
n=1
n=1
9.
23.
1
2n 1
In Exercises 30 44, a series is given.

(a) Find a formula for Sn , the nth par al sum of the series.
(b) Determine whether the series converges or diverges.
If it converges, state what it converges to.
30.
1
n
4
n=0
31. 13 + 23 + 33 + 43 +
32.
(1)n n
n=1
33.
5
n
2
n=0
34.
en
44.
n=1
37.
n=1
38.
n=1
39.
n=1
40.
n=1
1
(2n 1)(2n + 1)
(
ln
n
n+1
2n + 1
n2 (n + 1)2
1
1
1
1
+
+
+
+
14
25
36
47
) (
) (
)
(
1
1
1
1
1
1
+
+
+
42. 2 +
+
+
+
2
3
4
9
8
27
n=2
)n
1
n2 1
45. Break the Harmonic Series into the sum of the odd and even
terms:
1
1
1
=
+
.
n
2n
1
2n
n=1
n=1
n=1
The goal is to show that each of the series on the right diverge.
3
n(n + 2)
41.
43.
sin 1
n=0
1
1
1
1
35. 1 +
+
+
3
9
27
81
1
36.
n(n
+ 1)
n=1
(a) Show why
n=1
1
1
>
.
2n 1
2n
n=1
(Compare each nth par al sum.)

(b) Show why
n=1
1
1
<1+
2n 1
2n
n=1
(c) Explain why (a) and (b) demonstrate that the series
of odd terms is convergent, if, and only if, the series
of even terms is also convergent. (That is, show both
converge or both diverge.)
(d) Explain why knowing the Harmonic Series is divergent determines that the even and odd series are also
divergent.
46. Show the series
n=1
n
diverges.
(2n 1)(2n + 1)
8.3 Integral and Comparison Tests

Knowing whether or not a series converges is very important, especially when
we discuss Power Series in Sec on 8.6. Theorems 60 and 61 give criteria for
when Geometric and p- series converge, and Theorem 63 gives a quick test to
determine if a series diverges. There are many important series whose convergence cannot be determined by these theorems, though, so we introduce a set
of tests that allow us to handle a broad range of series. We start with the Integral Test.
Integral Test
Note: Theorem 65 does not state that

the integral and the summa on have the
same value.
We stated in Sec on 8.1 that a sequence {an } is a func on a(n) whose domain is N, the set of natural numbers. If we can extend a(n) to R, the real numbers, and it is both posi ve and decreasing on [1, ), then the convergence of

an is the same as
a(x) dx.
1
n=1
Theorem 65
Integral Test
Let a sequence {an } be dened by an = a(n), where a(n) is con nuous,
posi ve and decreasing on [1, ). Then

an converges, if, and only if,
n=1

a(x) dx converges.
2
y = a(x)
(a)
y
We can demonstrate the truth of the Integral Test with two simple graphs.
In Figure 8.15(a), the height of each rectangle is a(n) = an for n = 1, 2, . . .,
and clearly the rectangles enclose more area than the area under y = a(x).
Therefore we can conclude that
a(x) dx <
2
y = a(x)
x
1
(8.1)
n=2
Figure 8.15: Illustra ng the truth of the

Integral Test.
with n = 2; adding a1 to both sides lets us rewrite the summa on star ng with
(b)
410
an .
n=1
In Figure 8.15(b), we draw rectangles under y = a(x) with the Right-Hand rule,
star ng with n = 2. This me, the area of the rectangles is less than the area

under y = a(x), so
an <
a(x) dx. Note how this summa on starts
Notes:

n = 1:
an < a1 +
a(x) dx.
(8.2)
n=1
Combining Equa ons (8.1) and (8.2), we have

an < a1 +
a(x) dx < a1 +
an .
1
n=1
(8.3)
n=1
From Equa on (8.3) we can make the following two statements:
an < a1 +
a(x) dx (because
an diverges, so does
1. If
1
n=1
2. If
n=1
a(x) dx
1
(because
a(x) dx <
1
a(x) dx)
n=1
an converges, so does
an .)
n=1
Therefore the series and integral either both converge or both diverge. Theorem
64 allows us to extend this theorem to series where an is posi ve and decreasing
on [b, ) for some b > 1.
. Example 241
.Using the Integral Test
ln n
. (The terms of the sequence {an } =
n2
n=1
{ln n/n2 } and the nth par al sums are given in Figure 8.16.)
ln x
Applying the Integral Test, we test the convergence of
dx.
S
x2
1
Integra ng this improper integral requires the use of Integra on by Parts, with
u = ln x and dv = 1/x2 dx.

b
ln x
ln x
dx
=
lim
dx
2
2
b
x
1
1 x
b b 1
1

= lim ln x +
dx
2
b
x
1
1 x

1
1 b
= lim ln x
b
x
x 1
1 ln b
= lim 1
. Apply LHpitals Rule:
b
b
b
= 1.

ln x
ln n
Since
dx converges, so does
.
2
x
n2
1
n=1
y
0.8
0.6
0.4
0.2
n
2
8 10 12 14 16 18 20
. an
Sn
Figure 8.16: Plo ng the sequence and

Notes:
411

Note how the sequence {an } is not strictly decreasing; it increases from
n = 1 to n = 2. However, this does not keep us from applying the Integral
Test as the sequence in posi ve and decreasing on [2, ). .
Theorem 61 was given without jus ca on, sta ng that the general p-series
1
converges if, and only if, p > 1. In the following example, we
(an
+
b)p
n=1
prove this to be true by applying the Integral Test.
. Example 242
Using the Integral Test to establish Theorem 61.
1
Use the Integral Test to prove that
converges if, and only if, p > 1.
(an
+
b)p
n=1
Consider the integral
1
dx = lim
c
(ax + b)p
1
dx; assuming p = 1,
(ax + b)p
1
dx
(ax + b)p
c
1

= lim
(ax + b)1p
c a(1 p)
1
(
)
1
1p
(ac + b)
(a + b)1p .
= lim
c a(1 p)
1
This limit converges if, and only if, p > 1. It is easy to show that the integral also
diverges in the case of p = 1. (This result is similar to the work preceding Key
Idea 21.)
1
Therefore
converges if, and only if, p > 1. .
(an + b)p
n=1
We consider two more convergence tests in this sec on, both comparison
tests. That is, we determine the convergence of one series by comparing it to
another series with known convergence.
Notes:
412
Direct Comparison Test
Theorem 66
Let {an } and {bn } be posi ve sequences where an bn for all n N,

for some N 1.
1. If
bn converges, then
n=1
2. If
n=1
an converges.
n=1
an diverges, then
Note: A sequence {an } is a posi ve

sequence if an > 0 for all n.
Because of Theorem 64, any theorem that
relies on a posi ve sequence s ll holds
true when an > 0 for all but a nite number of values of n.
bn diverges.
n=1
. Example 243
Applying the Direct Comparison Test
1
.
n
3 + n2
n=1
This series is neither a geometric or p-series, but seems reS
lated. We predict it will converge, so we look for a series with larger terms that
converges. (Note too that the Integral Test seems dicult to apply here.)
1
1
1
Since 3n < 3n + n2 , n > n
for
all
n
1.
The
series
is a
n
3
3 + n2
3
n=1
1
convergent geometric series; by Theorem 66,
converges. .
n + n2
3
n=1
. Example 244
Applying the Direct Comparison Test
1
.
n ln n
n=1
diverges, and it seems

n
that the given series is closely related to it, hence we predict it will diverge.
1
1
Since n n ln n for all n 1,
for all n 1.
n
n ln n
1
The Harmonic Series diverges, so we conclude that
diverges as
n
ln n
n=1
well. .
S
We know the Harmonic Series
n=1
Notes:
413

The concept of direct comparison is powerful and o en rela vely easy to
apply. Prac ce helps one develop the necessary intui on to quickly pick a proper
series with which to compare. However, it is easy to construct a series for which
it is dicult to apply the Direct Comparison Test.
1
. It is very similar to the divergent series given in ExConsider
n + ln n
n=1
1
1
ample 244. We suspect that it also diverges, as
for large n. Hown
n + ln n
ever, the inequality that we naturally want to use goes the wrong way: since
1
1
n n + ln n for all n 1,
for all n 1. The given series has terms
n
n + ln n
less than the terms of a divergent series, and we cannot conclude anything from
this.
Fortunately, we can apply another test to the given series to determine its
convergence.
Limit Comparison Test

.
Theorem 67
Let {an } and {bn } be posi ve sequences.
an
= L, where L is a posi ve real number, then
an and
n bn
n=1
bn either both converge or both diverge.
1. If lim
n=1
an
= 0, then if
bn converges, then so does
an .
n bn
n=1
n=1
2. If lim
an
= , then if
bn diverges, then so does
an .
n bn
n=1
n=1
3. If lim
It is helpful to remember that when using Theorem 67, the terms of the
series with known convergence go in the denominator of the frac on.
We use the Limit Comparison Test in the next example to examine the series
1
which mo vated this new test.
n + ln n
n=1
Notes:
414

. Example 245
Applying the Limit Comparison Test
1
using the Limit Comparison Test.
n + ln n
n=1
We compare the terms of
Harmonic Sequence
1
n=1
lim
n=1
1
to the terms of the
n + ln n
1/(n + ln n)
n
= lim
n n + ln n
1/n
= 1 (a er applying LHpitals Rule).
Since the Harmonic Series diverges, we conclude that
n=1
well. .
1
diverges as
n + ln n
. Example 246
1
n
3 n2
n=1
This series is similar to the one in Example 243, but now we
S
are considering 3n n2 instead of 3n + n2 . This dierence makes applying
the Direct Comparison Test dicult.
1
Instead, we use the Limit Comparison Test and compare with the series
:
3n
n=1
1/(3n n2 )
3n
=
lim
n
n 3n n2
1/3n
= 1 (a er applying LHpitals Rule twice).
lim
We know
as well. .
1
1
is
a
convergent
geometric
series,
hence
converges
n
n n2
3
3
n=1
n=1
As men oned before, prac ce helps one develop the intui on to quickly
choose a series with which to compare. A general rule of thumb is to pick a
series based on the dominant term in the expression of {an }. It is also helpful
to note that factorials dominate exponen als, which dominate algebraic funcons (e.g., polynomials), which dominate logarithms. In the previous example,
Notes:
415
1
1
n
was
3
,
so
we
compared
the
series
to
. It is
n
3n n2
3
n=1
hard to apply the Limit Comparison Test to series containing factorials, though,
as we have not learned how to apply LHpitals Rule to n!.
the dominant term of
. Example 247
x+3
.
2x+1
x
n=1
We navely a empt to apply the rule of thumb given above
S
and note that the dominant term in the expression of the series is 1/x2 . Knowing
1
that
converges, we a empt to apply the Limit Comparison Test:
2
n
n=1
( x + 3)/(x2 x + 1)
x2 ( x + 3)
=
lim
n
n x2 x + 1
1/x2
= (Apply LHpitals Rule).
lim
Theorem 67 part (3) only applies when
bn diverges; in our case, it con-
n=1
verges. Ul mately, our test has not revealed anything about the convergence
of our series.
The problem is that we chose a poor series with which to compare. Since
the numerator and denominator of the terms of the series are both algebraic
func ons, we should have compared our series to the dominant term of the
numerator divided by the dominant term of the denominator.
The dominant term of the numerator is x1/2 and the dominant term of the
denominator is x2 . Thus we should compare the terms of the given series to
x1/2 /x2 = 1/x3/2 :
( x + 3)/(x2 x + 1)
x3/2 ( x + 3)
lim
= lim
n
n x2 x + 1
1/x3/2
= 1 (Applying LHpitals Rule).
x+3
1
Since the p-series
converges, we conclude that
con2
3/2
x x+1
x
n=1
n=1
verges as well. .
Notes:
416
Exercises 8.3
Terms and Concepts
16.
1. In order to apply the Integral Test to a sequence {an }, the

func on a(n) = an must be
,
and
.
n=1
17.
2. T/F: The Integral Test can be used to determine the sum of

a convergent series.
3. What test(s) in this sec on do not work well with factorials?
an is convergent, and there are sequences

4. Suppose
n=0
18.
In Exercises 5 12, use the Integral Test to determine the convergence of the given series.
1
5.
n
2
n=1
6.
7.
1
4
n
n=1
n=1
8.
n=2
9.
n=1
10.
n=2
11.
n=1
n
n2 + 1
1
n ln n
1
n2 + 1
19.
n=1
14.
n=1
15.
n=1
21.
22.
1
n2
n=2
5n
2n
+ 10
n
n2 1
1
n2 ln n
In Exercises 23 32, use the Limit Comparison Test to determine the convergence of the given series; state what series is
used for comparison.
23.
n=1
24.
n=1
25.
26.
1
n2 3n + 5
1
4 n n2
ln n
n
3
n=4
n=1
n=1
28.
n=1
29.
1
n2 + n
n+ n
n 10
n2 + 10n + 10
(
)
sin 1/n
n=1
30.
n+5
3 5
n
n=1
31.
n+3
2 + 17
n
n=1
1
n2 + 3n 5
ln n
n
1
n2 1
n2 + n + 1
n3 5
n=1
n=2
n
2n
1
4 n + n2 n
n=1
27.
In Exercises 13 22, use the Direct Comparison Test to determine the convergence of the given series; state what series is
used for comparison.
13.
20.
1
n(ln n)2
ln n
12.
n3
n=1
n=5
n=0
Problems
1
n! + n
n=2
n=0
{bn } and {cn } such that bn an cn for all n. What
can be said about the series

bn and
cn ?
32.
n=1
1
n + 100
In Exercises 33 40, determine the convergence of the given

series. State the test used; more than one test may be appropriate.
n2
33.
2n
n=1
34.
n=1
35.
n=1
36.
n=1
37.
n=1
1
(2n + 5)3
n!
10n
ln n
n!
1
3n + n
38.
n2
10n
+5
n=1
39.
3n
3
n
n=1
40.
cos(1/n)
n
n=1
41. Given that
an converges, state which of the following
n=1
series converges, may converge, or does not converge.

(a)
(b)
an
n
n=1
an an+1
n=1
(c)
(an )2
n=1
(d)
nan
n=1
(e)
1
a
n
n=1
8.4 Ra o and Root Tests
8.4
Ra o and Root Tests
The nth Term Test of Theorem 63 states that in order for a series
an to con-
n=1
verge, lim an = 0. That is, the terms of {an } must get very small. Not only must
n
the terms approach 0, they must approach 0 fast enough: while lim 1/n = 0,
the Harmonic Series
1
n=1
diverges as the terms of {1/n} do not approach 0
fast enough.
The comparison tests of the previous sec on determine convergence by comparing terms of a series to terms of another series whose convergence is known.
This sec on introduces the Ra o and Root Tests, which determine convergence
by analyzing the terms of a series to see if they approach 0 fast enough.
Ra o Test
Theorem 68
Ra o Test
Let {an } be a posi ve sequence where lim
1. If L < 1, then
an+1
= L.
an
an converges.
n=1
2. If L > 1 or L = , then
Note: Theorem 64 allows us to apply the

Ra o Test to series where {an } is posi ve
for all but a nite number of terms.
an diverges.
n=1
3. If L = 1, the Ra o Test is inconclusive.

an+1
= L < 1, then for large n,
n an
each term of {an } is signicantly smaller than its previous term which is enough
to ensure convergence.
The principle of the Ra o Test is this: if lim
. Example 248
.Applying the Ra o Test
Use the Ra o Test to determine the convergence of the following series:
1.
2n
n=1
n!
2.
3n
n=1
n3
3.
n=1
1
.
n2 + 1
Notes:
419

S
1.
2n
n=1
n!
:
2n+1 /(n + 1)!
2n+1 n!
= lim n
n
n
n 2 (n + 1)!
2 /n!
2
= lim
n n + 1
= 0.
lim
Since the limit is 0 < 1, by the Ra o Test
2n
n=1
2.
3n
n=1
n3
n!
converges.
:
3n+1 /(n + 1)3
3n+1 n3
=
lim
n
n 3n (n + 1)3
3n /n3
3n3
= lim
n (n + 1)3
= 3.
lim
Since the limit is 3 > 1, by the Ra o Test
3n
n=1
3.
n=1
n2
n3
diverges.
1
:
+1
(
)
1/ (n + 1)2 + 1
n2 + 1
=
lim
lim
n (n + 1)2 + 1
n
1/(n2 + 1)
= 1.
Since the limit is 1, the Ra o Test is inconclusive. We can easily show this
series converges using the Direct or Limit Comparison Tests, with each
1
comparing to the series
.
n2
n=1
.
Notes:
420
8.4 Ra o and Root Tests

The Ra o Test is not eec ve when the terms of a series only contain algebraic func ons (e.g., polynomials). It is most eec ve when the terms contain some factorials or exponen als. The previous example also reinforces our
developing intui on: factorials dominate exponen als, which dominate algebraic func ons, which dominate logarithmic func ons. In Part 1 of the example,
the factorial in the denominator dominated the exponen al in the numerator,
causing the series to converge. In Part 2, the exponen al in the numerator dominated the algebraic func on in the denominator, causing the series to diverge.
While we have used factorials in previous sec ons, we have not explored
them closely and one is likely to not yet have a strong intui ve sense for how
they behave. The following example gives more prac ce with factorials.
. Example 249
Applying the Ra o Test
n!n!
.
(2n)!
n=1
Before we begin, be sure to note the dierence between
S
(2n)! and 2n!. When n = 4, the former is 8! = 8 7 . . . 2 1 = 40, 320,
whereas the la er is 2(4 3 2 1) = 48.
Applying the Ra o Test:
(
)
(n + 1)!(n + 1)!/ 2(n + 1) !
(n + 1)!(n + 1)!(2n)!
lim
= lim
n
n
n!n!/(2n)!
n!n!(2n + 2)!
No ng that (2n + 2)! = (2n + 2) (2n + 1) (2n)!, we have
= lim
(n + 1)(n + 1)
(2n + 2)(2n + 1)
= 1/4.
Since the limit is 1/4 < 1, by the Ra o Test we conclude
n!n!
converges. .
(2n)!
n=1
Root Test
The nal test we introduce is the Root Test, which works par cularly well on
series where each term is raised to a power, and does not work well with terms
containing factorials.
Notes:
421

.
Theorem 69
Root Test
Let {an } be a posi ve sequence, and let lim (an )1/n = L.

n
1. If L < 1, then
Note: Theorem 64 allows us to apply the

Root Test to series where {an } is posi ve
for all but a nite number of terms.
an converges.
n=1
2. If L > 1 or L = , then
an diverges.
n=1
3. If L = 1, the Root Test is inconclusive.

. Example 250
Applying the Root Test
Determine the convergence of the following series using the Root Test:
1.
)n
(
3n + 1
n=1
5n 2
2.
n=1
n4
(ln n)n
3.
2n
n=1
n2
((
)n )1/n
3n + 1
3n + 1
3
1. lim
= lim
= .
n
n
5n 2
5n 2
5
Since the limit is less than 1, we conclude the series converges. Note: it is
dicult to apply the Ra o Test to this series.
( 1/n )4
( 4 )1/n
n
n
2. lim
=
lim
.
n (ln n)n
n
ln n
As n grows, the numerator approaches 1 (apply LHpitals Rule) and the
denominator grows to innity. Thus
(
lim
)4
n1/n
= 0.
ln n
Since the limit is less than 1, we conclude the series converges.

( n )1/n
2
2
3. lim
= lim (
) = 2.
n n2
n n1/n 2
.
Notes:
422
Since this is greater than 2, we conclude the series diverges.
Exercises 8.4
Terms and Concepts
1. The Ra o Test is not eec ve when the terms of a sequence
only contain
func ons.
17.
2 n n2
3n
n=1
18.
1
n
n
n=1
2. The Ra o Test is most eec ve when the terms of a sequence contains

and/or
func ons.
3. What three convergence tests do not work well with terms
containing factorials?
19.
n=1
4. The Root Test works par cularly well on series where each
term is
to a
.
20.
Problems
21.

series using the Ra o Test. If the Ra o Test is inconclusive,
state so and determine convergence with another test.
2n
5.
n!
n=0
n!10n
(2n)!
n=0
8.
5 n + n4
7 n + n2
n=1
n=1
10.
1
n
n=1
1
+7
3n3
n=1
( )n
3
5
2 4 6 8 2n
13.
3
6 9 12 3n
n=1
14.
n=1
n!
5 10 15 (5n)

series using the Root Test. If the Root Test is inconclusive,
state so and determine convergence with another test.
)n
(
2n + 5
15.
3n + 11
n=1
16.
)n
(
.9n2 n 3
n=1
22.
)n
( 2
n n
n2 + n
1
2
n
n
n=1
24.
1
)n
ln n
n2 + n + 3
n2
)n
ln n
n=1
)n

series. State the test used; more than one test may be appropriate.
25.
n=1
26.
10 5n
11.
7n 3
n=1
12.
4n+7
7n
n=1
n=1
7.
9.
3n
n2 2n+1
n=1
23.
5n 3n
6.
4n
n=0
27.
n4 4 n
n!
n=1
n=1
28.
29.
n2 + 4n 2
n3 + 4n2 3n + 7
n2
+n
3n
3n
nn
n=1
n=1
n
n2 + 4n + 1
n!n!n!
30.
(3n)!
n=1
31.
32.
1
ln
n
n=1
)n
(
n+2
n+1
n=1
33.
n=1
34.
n3
)n
ln n
1
n=1
n
n+2

All of the series convergence tests we have used require that the underlying
sequence {an } be a posi ve sequence. (We can relax this with Theorem 64 and
state that there must be an N > 0 such that an > 0 for all n > N; that is, {an } is
posi ve for all but a nite number of values of n.)
In this sec on we explore series whose summa on includes nega ve terms.
We start with a very specic form of series, where the terms of the summa on
alternate between being posi ve and nega ve.
.
Deni on 34
Alterna ng Series
Let {an } be a posi ve sequence. An alterna ng series is a series of either

the form
(1)n an
or
(1)n+1 an .
n=1
n=1
Recall the terms of Harmonic Series come from the Harmonic Sequence {an } =
{1/n}. An important alterna ng series is the Alterna ng Harmonic Series:
1
1 1 1 1 1
(1)n+1 = 1 + + +
n
2 3 4 5 6
n=1
Geometric Series can also be alterna ng series when r < 0. For instance, if
r = 1/2, the geometric series is
)n
(
1
n=0
=1
1 1 1
1
1
+ +
+
2 4 8 16 32
Theorem 60 states that geometric series converge when |r| < 1 and gives
1
the sum:
rn =
. When r = 1/2 as above, we nd
1
r
n=0
)n
(
1
n=0
1
1
2
=
= .
1 (1/2)
3/2
3
A powerful convergence theorem exists for other alterna ng series that meet
a few condi ons.
Notes:
424
Theorem 70
Alterna ng Series Test
Let {an } be a posi ve, decreasing sequence where lim an = 0. Then
(1)n an
.
and
n=1
(1)n+1 an
n=1
converge.
1
The basic idea behind Theorem 70 is illustrated in Figure 8.17. A posi ve,
decreasing sequence {an } is shown along with the par al sums
Sn =
L
0.5
(1)i+1 ai = a1 a2 + a3 a4 + + (1)n an .
i=1
Because {an } is decreasing, the amount by which Sn bounces up/down decreases.

Moreover, the odd terms of Sn form a decreasing, bounded sequence, while the
even terms of Sn form an increasing, bounded sequence. Since bounded, monotonic sequences converge (see Theorem 59) and the terms of {an } approach 0,
one can show the odd and even terms of Sn converge to the same common limit
L, the sum of the series.
. an
10
Sn
Figure 8.17: Illustra ng convergence with

the Alterna ng Series Test.
. Example 251
.Applying the Alterna ng Series Test
Determine if the Alterna ng Series Test applies to each of the following series.
1.
1
(1)n+1
n
n=1
2.
ln n
(1)n
n
n=1
3.
| sin n|
(1)n+1 2
n
n=1
1. This is the Alterna ng Harmonic Series as seen previously. The underlying

sequence is {an } = {1/n}, which is posi ve, decreasing, and approaches
0 as n . Therefore we can apply the Alterna ng Series Test and
conclude this series converges.
While the test does not state what the series converges to, we will see
1
later that
(1)n+1 = ln 2.
n
n=1
2. The underlying sequence is {an } = {ln n/n}. This is posi ve and approaches 0 as n (use LHpitals Rule). However, the sequence is not
decreasing for all n. It is straigh orward to compute a1 = 0, a2 0.347,
Notes:
425

a3 0.366, and a4 0.347: the sequence is increasing for at least the
rst 3 terms.
We do not immediately conclude that we cannot apply the Alterna ng
Series Test. Rather, consider the longterm behavior of {an }. Trea ng
an = a(n) as a con nuous func on of n dened on (1, ), we can take
its deriva ve:
1 ln n
.
a (n) =
n2
The deriva ve is nega ve for all n 3 (actually, for all n > e), meaning a(n) = an is decreasing on (3, ). We can apply the Alterna ng
Series Test to the series when we start with n = 3 and conclude that
ln n
converges; adding the terms with n = 1 and n = 2 do not
(1)n
n
n=3
change the convergence (i.e., we apply Theorem 64).
The important lesson here is that as before, if a series fails to meet the
criteria of the Alterna ng Series Test on only a nite number of terms, we
can s ll apply the test.
3. The underlying sequence is {an } = | sin n|/n. This sequence is posi ve
and approaches 0 as n . However, it is not a decreasing sequence;
the value of | sin n| oscillates between 0 and 1 as n . We cannot
remove a nite number of terms to make {an } decreasing, therefore we
cannot apply the Alterna ng Series Test.
Keep in mind that this does not mean we conclude the series diverges;
in fact, it does converge. We are just unable to conclude this based on
Theorem 70.
Key Idea 31 gives the sum of some important series. Two of these are
1
2
=
1.64493
2
n
6
n=1
and
(1)n+1
n=1
n2
2
0.82247.
12
These two series converge to their sums at dierent rates. To be accurate to

two places a er the decimal, we need 202 terms of the rst series though only
13 of the second. To get 3 places of accuracy, we need 1069 terms of the rst
series though only 33 of the second. Why is it that the second series converges
so much faster than the rst?
While there are many factors involved when studying rates of convergence,
the alterna ng structure of an alterna ng series gives us a powerful tool when
approxima ng the sum of a convergent series.
Notes:
426
Theorem 71
The Alterna ng Series Approxima on Theorem
Let {an } be a sequence that sa ses the hypotheses of the Alterna ng

Series Test, and let Sn and L be the nth par al sums and sum, respec vely,
of either
(1)n an or
(1)n+1 an . Then
n=1
n=1
1. |Sn L| < an+1 , and

2. L is between Sn and Sn+1 .
Part 1 of Theorem 71 states that the nth par al sum of a convergent alternating series will be within an+1 of its total sum. Consider the alterna ng series
(1)n+1
we looked at before the statement of the theorem,
. Since a14 =
n2
n=1
1/142 0.0051, we know that S13 is within 0.0051 of the total sum. That is, we
know S13 is accurate to at least 1 place a er the decimal. (The 5 in the third
place a er the decimal could cause a carry meaning S13 isnt accurate to two
places a er the decimal; in this par cular case, that doesnt happen.)
Moreover, Part 2 of the theorem states that since S13 0.8252 and S14
0.8201, we know the sum L lies between 0.8201 and 0.8252, assuring us that
S13 is indeed accurate to two decimal places.
Some alterna ng series converge slowly. In Example 251 we determined the
ln n
series
(1)n+1
converged. With n = 1001, we nd ln n/n 0.0069,
n
n=1
meaning that S1000 0.1633 is accurate to one, maybe two, places a er the
decimal. Since S1001 0.1564, we know the sum L is 0.1564 L 0.1633.
. Example 252
.Approxima ng the sum of convergent alterna ng series
Approximate the sum of the following series, accurate to two places a er the
decimal.
1.
1
(1)n+1 3
n
n=1
2.
ln n
.
(1)n+1
n
n=1
1. To be ensure accuracy to two places a er the decimal, we need an <
Notes:
427

0.0001:
1
< 0.0001
n3
n3 > 10, 000
3
n > 10000 21.5.
With n = 22, we are assured accuracy to two places a er the decimal.
With S21 0.9015, we are condent that the sum L of the series is about
0.90.
We can arrive at this approxima on another way. Part 2 of Theorem 71
states that the sum L lies between successive par al sums. It is straightforward to compute S6 0.899782, S7 0.9027 and S8 0.9007. We
know the sum must lie between these last two par al sums; since they
agree to two places a er the decimal, we know L 0.90.
2. We again solve for n such that an < 0.0001; that is, we want n such that
ln(n)/n < 0.0001. This cannot be solved algebraically, so we approximate
the solu on using Newtons Method.
Let f(x) = ln(x)/x 0.0001. We want to nd where f(x) = 0. Assuming that x must be large, we let x(1 = 1000.
) Recall that xn+1 = xn
f(xn )/f (xn ); we compute f (x) = 1 ln(x) /x2 . Thus:
ln(1000)/1000 0.0001
)
x2 = 1000 (
1 ln(1000) /10002
= 2152.34.
Using a computer, we nd that a er 12 itera ons we nd x 116, 671.
With S116,671 0.1598 and S116,672 0.1599, we know that the sum L is
between these two values. Simply sta ng that L 0.15 is misleading, as
L is very, very close to 0.16.
One of the famous results of mathema cs is that the Harmonic Series,
n
n=1
1
diverges, yet the Alterna ng Harmonic Series,
(1)n+1 , converges. The
n
n=1
no on that alterna ng the signs of the terms in a series can make a series converge leads us to the following deni ons.
Notes:
428
Deni on 35
Absolute and Condi onal Convergence
1. A series
an converges absolutely if
n=1
2. A series
n=1
an converges condi onally if
n=1
|an | converges.
Note: In Deni on 35,
an is not nec-
n=1
essarily an alterna ng series; it just may

have some nega ve terms.
an converges but
n=1
|an | diverges.
n=1
Thus we say the Alterna ng Harmonic Series converges condi onally.

. Example 253
.Determining absolute and condi onal convergence.
Determine if the following series converges absolutely, condi onally, or diverges.
1.
(1)n
n=1
n+3
n2 + 2n + 5
n2 + 2n + 5
2.
(1)n
2n
n=1
3n 3
3.
(1)n
5n
10
n=3
1. We can show the series

(1)n

n=1
n + 3
n+3
=
n2 + 2n + 5 n=1 n2 + 2n + 5
diverges using the Limit Comparison Test, comparing with 1/n.
(1)n
n+3
converges using the Alterna ng Series
2 + 2n + 5
n
n=1
Test; we conclude it converges condi onally.
The series
2. We can show the series

n2 + 2n + 5
(1)n n + 2n + 5 =

2n
2n
n=1
n=1
converges using the Ra o Test.
n2 + 2n + 5
Therefore we conclude
converges absolutely.
(1)n
2n
n=1
Notes:
429

3. The series

3n 3
(1)n 3n 3 =

5n 10
5n 10
n=3
n=3
diverges using the nth Term Test, so it does not converge absolutely.
3n 3
(1)n
fails the condi ons of the Alterna ng Series
5n 10
n=3
Test as (3n 3)/(5n 10) does not approach 0 as n . We can state
further that this series diverges; as n , the series eec vely adds and
subtracts 3/5 over and over. This causes the sequence of par al sums to
oscillate and not converge.
The series
Therefore the series

.
3n 3
(1)n
diverges.
5n 10
n=1
Knowing that a series converges absolutely allows us to make two important statements, given in the following theorem. The rst is that absolute con
vergence is stronger than regular convergence. That is, just because

an
converges, we cannot conclude that
converges absolutely tells us that
n=1
|an | will converge, but knowing a series
n=1
an will converge.
n=1
One reason this is important is that our convergence tests all require that the
underlying sequence of terms be posi ve. By taking the absolute value of the
terms of a series where not all terms are posi ve, we are o en able to apply an
appropriate test and determine absolute convergence. This, in turn, determines
that the series we are given also converges.
The second statement relates to rearrangements of series. When dealing
with a nite set of numbers, the sum of the numbers does not depend on the
order which they are added. (So 1 + 2 + 3 = 3 + 1 +2.) One may be surprised to
nd out that when dealing with an innite set of numbers, the same statement
does not always hold true: some innite lists of numbers may be rearranged in
dierent orders to achieve dierent sums. The theorem states that the terms of
an absolutely convergent series can be rearranged in any way without aec ng
the sum.
Notes:
430
Theorem 72
Let
Absolute Convergence Theorem
an be a series that converges absolutely.
n=1
1.
an converges.
n=1
2. Let {bn } be any rearrangement of the sequence {an }. Then
n=1
bn =
an .
n=1
In Example 253, we determined the series in part 2 converges absolutely.

Theorem 72 tells us the series converges (which we could also determine using
the Alterna ng Series Test).
The theorem states that rearranging the terms of an absolutely convergent
series does not aect its sum. This implies that perhaps the sum of a condi onally convergent series can change based on the arrangement of terms. Indeed,
it can. The Riemann Rearrangement Theorem (named a er Bernhard Riemann)
states that any condi onally convergent series can have its terms rearranged so
that the sum is any desired value, including !
As an example, consider the Alterna ng Harmonic Series once more. We
have stated that
1
1 1 1 1 1 1
(1)n+1 = 1 + + + = ln 2,
n
2 3 4 5 6 7
n=1
(see Key Idea 31 or Example 251).

Consider the rearrangement where every posi ve term is followed by two
nega ve terms:
1
1 1 1 1 1 1
1
1
+ +
2 4 3 6 8 5 10 12
(Convince yourself that these are exactly the same numbers as appear in the
Alterna ng Harmonic Series, just in a dierent order.) Now group some terms
Notes:
431

and simplify:
(
)
(
)
(
)
1
1
1 1
1
1
1
1
1
+
+ =
2
4
3 6
8
5 10
12
1 1 1 1
1
1
+ +
+ =
2
4
6
8
10
12
)
(
1
1
1 1 1 1 1
1 + + + = ln 2.
2
2 3 4 5 6
2
By rearranging the terms of the series, we have arrived at a dierent sum!
(One could try to argue that the Alterna ng Harmonic Series does not actually
converge to ln 2, because rearranging the terms of the series shouldnt change
the sum. However, the Alterna ng Series Test proves this series converges to
L, for some number L, and if the rearrangement does not change the sum, then
L = L/2, implying L = 0. But the Alterna ng Series Approxima on Theorem
quickly shows that L > 0. The only conclusion is that the rearrangement did
change the sum.) This is an incredible result.
We end here our study of tests to determine convergence. The back cover
of this text contains a table summarizing the tests that one may nd useful.
While series are worthy of study in and of themselves, our ul mate goal
within calculus is the study of Power Series, which we will consider in the next
sec on. We will use power series to create func ons where the output is the
result of an innite summa on.
Notes:
432
Exercises 8.5
(
)
sin (n + 1/2)
14.
n ln n
n=1
Terms and Concepts

1. Why is
)n
(
2
15.
3
n=0
sin n not an alterna ng series?
n=1
2. A series
(1)n an converges when {an } is

n=1
and lim an =
16.
3. Give an example of a series where
n=0
an converges but
17.
n=0
|an | does not.

18.
n=0
In Exercises 5 20, an alterna ng series
(1)n
n
n=1
20.
(1000)n
n!
n=1
an is given.
n=i
(a) Determine if the series converges or diverges.
(b) Determine if
|an | converges or diverges.
n=0
(c) If
an converges, determine if the convergence is
Let Sn be the nth par al sum of a series. In Exercises 21 24, a

convergent alterna ng series is given and a value of n. Compute Sn and Sn+1 and use these values to nd bounds on the
sum of the series.
21.
(1)n
,
ln(n + 1)
n=1
22.
(1)n+1
,
n4
n=1
23.
(1)n
,
n!
n=0
24.
)n
(
,
2
n=0
n=0
condi onal or absolute.
(1)n+1
5.
n2
n=1
6.
(1)n+1
n!
n=1
7.
n+5
(1)n
3n
5
n=0
8.
2n
(1)n 2
n
n=1
9.
(1)n+1
n=0
3n + 5
n2 3n + 1
(1)n
10.
ln
n+1
n=1
n
11.
(1)n
ln
n
n=2
12.
n=1
13.
n=1
(1)n+1
1 + 3 + 5 + + (2n 1)
( )
cos n
2
(1)n 2n
19.
Problems
(1)n n2
n!
n=0
n=0
4. The sum of a
convergent series can be changed by
rearranging the order of its terms.
(e)n
n=5
n=4
n=6
n=9
In Exercises 25 28, a convergent alterna ng series is given

along with its sum and a value of . Use Theorem 71 to nd
n such that the nth par al sum of the series is within of the
sum of the series.
25.
(1)n+1
74
=
,
n4
720
n=1
26.
(1)n
1
= ,
n!
e
n=0
= 0.0001
27.
(1)n
= ,
2n
+
1
4
n=0
= 0.001
28.
(1)n
= cos 1,
(2n)!
n=0
= 0.001
= 108
8.6 Power Series

So far, our study of series has examined the ques on of Is the sum of these
innite terms nite?, i.e., Does the series converge? We now approach series
from a dierent perspec ve: as a func on. Given a value of x, we evaluate f(x)
by nding the sum of a par cular series that depends on x (assuming the series
converges). We start this new approach to series with a deni on.
.
Deni on 36
Power Series
Let {an } be a sequence, let x be a variable, and let c be a real number.

1. The power series in x is the series
an xn = a0 + a1 x + a2 x2 + a3 x3 + . . .
n=0
2. The power series in x centered at c is the series
an (x c)n = a0 + a1 (x c) + a2 (x c)2 + a3 (x c)3 + . . .
n=0
. Example 254
.Examples of power series
Write out the rst ve terms of the following power series:
1.
n=0
xn
2.
(1)n+1
n=1
(x + 1)n
n
3.
(x )2n
(1)n+1
.
(2n)!
n=0
1. One of the conven ons we adopt is that x0 = 1 regardless of the value of

x. Therefore
xn = 1 + x + x2 + x3 + x4 + . . .
n=0
This is a geometric series in x.

2. This series is centered at c = 1. Note how this series starts with n = 1.
We could rewrite this series star ng at n = 0 with the understanding that
Notes:
434
8.6 Power Series

a0 = 0, and hence the rst term is 0.
(x + 1)2 (x + 1)3 (x + 1)4 (x + 1)5

(x + 1)n
= (x+1)
+
+
...
(1)n+1
n
2
3
4
5
n=1
3. This series is centered at c = . Recall that 0! = 1.
(x )2n
(x )2 (x )4 (x )6 (x )8
(1)n+1
= 1+
...
(2n)!
2
24
6!
8!
n=0
.
We introduced power series as a type of func on, where a value of x is given
and the sum of a series is returned. Of course, not every series converges. For
instance, in part 1 of Example 254, we recognized the series

xn as a geometric
n=0
series in x. Theorem 60 states that this series converges only when |x| < 1.
This raises the ques on: For what values of x will a given power series converge?, which leads us to a theorem and deni on.
Theorem 73
Convergence of Power Series
Let a power series

true:
an (x c)n be given. Then one of the following is
n=0
1. The series converges only at x = c.

2. There is an R > 0 such that the series converges for all x in
(c R, c + R) and diverges for all x < c R and x > c + R.
3. The series converges for all x.
The value of R is important when understanding a power series, hence it is

given a name in the following deni on. Also, note that part 2 of Theorem 73
makes a statement about the interval (c R, c + R), but the not the endpoints
of that interval. A series may/may not converge at these endpoints.
Notes:
435

.
Deni on 37
Radius and Interval of Convergence
1. The number R given in Theorem 73 is the radius of convergence of

a given series. When a series converges for only x = c, we say the
radius of convergence is 0, i.e., R = 0. When a series converges
for all x, we say the series has an innite radius of convergence,
i.e., R = .
2. The interval of convergence is the set of all values of x for which

the series converges.
To nd the values of x for which a given series converges, we will use the convergence tests we studied previously (especially the Ra o Test). However, the
tests all required that the terms of a series be posi ve. The following theorem
gives us a workaround to this problem.
Theorem 74
Convergence
The series
The Radius of Convergence of a Series and Absolute
an (x c)n and
n=0
convergence R.

an (x c)n have the same radius of
n=0
Theorem 74 allows us to nd the radius of convergence R of a series by applying the Ra o Test (or any applicable test) to the absolute value of the terms
of the series. We prac ce this in the following example.
. Example 255
.Determining the radius and interval of convergence.
Find the radius and interval of convergence for each of the following series:
1.
x
n=0
Notes:
436
n!
2.
xn
(1)n+1
n
n=1
3.
n=0
2n (x 3)n
4.
n=0
n!xn
8.6 Power Series

n
x
:
1. We apply the Ra o Test to the series
n!
n=0
n+1

n+1

x /(n + 1)!
x
n!

lim
= lim n
xn /n!
n
n
x
(n + 1)!

x

= lim
n n + 1
= 0 for all x.
The Ra o Test shows us that regardless of the choice of x, the series converges. Therefore the radius of convergence is R = , and the interval of
convergence is (, ).

n

n
x

n+1 x
:

=
(1)
n
n
n=1
n=1
n+1

n+1

x /(n + 1)
x
n

lim
= lim n
xn /n
n
n
x
n + 1
n
= lim |x|
n
n+1
= |x|.
The Ra o Test states a series converges if the limit of |an+1 /an | = L < 1.
We found the limit above to be |x|; therefore, the power series converges
when |x| < 1, or when x is in (1, 1). Thus the radius of convergence is
R = 1. .
To determine the interval of convergence, we need to check the endpoints
of (1, 1). When x = 1, we have the series
(1)n
1
(1)n+1
=
n
n
n=1
n=1
= .
The series diverges when x = 1.
(1)n
, which is the Alterna ng
(1)n+1
n
n=1
Harmonic Series, which converges. Therefore the interval of convergence
is (1, 1].
When x = 1, we have the series
Notes:
437
n

2 (x 3)n :
n=0
n+1

n+1

2 (x 3)n+1
2
(x 3)n+1

lim
lim n
2n (x 3)n = n
n
2
(x 3)n

= lim 2(x 3).
n

2(x3) < 1 =
According
to
the
Ra
o
Test,
the
series
converges
when

x 3 < 1/2. The series is centered at 3, and x must be within 1/2 of 3
in order for the series to converge. Therefore the radius of convergence
is R = 1/2, and we know that the series converges absolutely for all x in
(3 1/2, 3 + 1/2) = (2.5, 3.5).
We check for convergence at the endpoints to nd the interval of convergence. When x = 2.5, we have:
2n (2.5 3)n =
n=0
2n (1/2)n
n=0
=
(1)n ,
n=0
which diverges. A similar process shows that the series also diverges at
x = 3.5. Therefore the interval of convergence is (2.5, 3.5).
4. We apply the Ra o Test to
n
n!x :
n=0

(n + 1)!xn+1

(n + 1)x

lim
=
lim
n!xn
n
n
= for all x, except x = 0.
The Ra o Test shows that the series diverges for all x except x = 0. Therefore the radius of convergence is R = 0.
We can use a power series to dene a func on:

f(x) =
a n xn
n=0
where the domain of f is a subset of the interval of convergence of the power

series. One can apply calculus techniques to such func ons; in par cular, we
can nd deriva ves and an deriva ves.
Notes:
438
8.6 Power Series
Theorem 75
Deriva ves and Indenite Integrals of Power Series
Func ons
Let f(x) =
an (x c)n be a func on dened by a power series, with
n=0
radius of convergence R.
1. f(x) is con nuous and dieren able on (c R, c + R).

2. f (x) =
an n (x c)n1 , with radius of convergence R.
n=1
3.
f(x) dx = C +
n=0
an
(x c)n+1
, with radius of convergence R.
n+1
A few notes about Theorem 75:

1. The theorem states that dieren a on and integra on do not change the
radius of convergence. It does not state anything about the interval of
convergence. They are not always the same.
2. No ce how the summa on for f (x) starts with n = 1. This is because the
constant term a0 of f(x) goes to 0.
3. Dieren a on and integra on are simply calculated termbyterm using
the Power Rules.
. Example 256
.Deriva ves and indenite integrals of power series
Let f(x) =
x . Find f (x) and F(x) =
f(x) dx, along with their respec ve
n=0
intervals of convergence.
We nd the deriva ve and indenite integral of f(x), follow-
ing Theorem 75.

1. f (x) =
nxn1 = 1 + 2x + 3x2 + 4x3 + .
n=1
In Example 254, we recognized that
xn is a geometric series in x. We
n=0
know that such a geometric series converges when |x| < 1; that is, the
interval of convergence is (1, 1).
Notes:
439

To determine the interval of convergence of f (x), we consider the endpoints of (1, 1):
f (1) = 1 2 + 3 4 + ,
f (1) = 1 + 2 + 3 + 4 + ,
which diverges.
which diverges.
Therefore, the interval of convergence of f (x) is (1, 1).
xn+1
x2
x3
2. F(x) = f(x) dx = C +
=C+x+
+
+
n+1
2
3
n=0
To nd the interval of convergence of F(x), we again consider the endpoints of (1, 1):
F(1) = C 1 + 1/2 1/3 + 1/4 +
The value of C is irrelevant; no ce that the rest of the series is an Alterna ng Series that whose terms converge to 0. By the Alterna ng Series
Test, this series converges. (In fact, we can recognize that the terms of the
series a er C are the opposite of the Alterna ng Harmonic Series. We can
thus say that F(1) = C ln 2.)
F(1) = C + 1 + 1/2 + 1/3 + 1/4 +
No ce that this summa on is C + the Harmonic Series, which diverges.
Since F converges for x = 1 and diverges for x = 1, the interval of
convergence of F(x) is [1, 1).
The previous example showed how to take the deriva ve and indenite integral of a power series without mo va on for why we care about such operaons. We may care for the sheer mathema cal enjoyment that we can, which
is mo va on enough for many. However, we would be remiss to not recognize
that we can learn a great deal from taking deriva ves and indenite integrals.
Recall that f(x) =
xn in Example 256 is a geometric series. According to
n=0
Theorem 60, this series converges to 1/(1 x) when |x| < 1. Thus we can say
f(x) =
n=0
xn =
1
,
1x
on
(1, 1).
Integra ng the power series, (as done in Example 256,) we nd
xn+1
F(x) = C1 +
,
n+1
n=0
Notes:
440
(8.4)
8.6 Power Series

while integra ng the func on f(x) = 1/(1 x) gives
F(x) = ln |1 x| + C2 .
(8.5)
Equa ng Equa ons (8.4) and (8.5), we have

F(x) = C1 +
xn+1
= ln |1 x| + C2 .
n+1
n=0
Le ng x = 0, we have F(0) = C1 = C2 . This implies that we can drop the

constants and conclude
xn+1
= ln |1 x|.
n+1
n=0
We established in Example 256 that the series on the le converges at x = 1;

subs tu ng x = 1 on both sides of the above equality gives
1 +
1 1 1 1
+ + = ln 2.
2 3 4 5
On the le we have the opposite of the Alterna ng Harmonic Series; on the

right, we have ln 2. We conclude that
1
1 1 1
+ + = ln 2.
2 3 4
Important: We stated in Key Idea 31 (in Sec on 8.2) that the Alterna ng Harmonic Series converges to ln 2, and referred to this fact again in Example 251 of
Sec on 8.5. However, we never gave an argument for why this was the case.
The work above nally shows how we conclude that the Alterna ng Harmonic
Series converges to ln 2.
We use this type of analysis in the next example.
. Example 257
.Analyzing power series func ons
Let f(x) =
. Find f (x) and f(x) dx, and use these to analyze the behavn!
n=0
ior of f(x).
We start by making two notes: rst, in Example 255, we
S
found the interval of convergence of this power series is (, ). Second,
we will nd it useful later to have a few terms of the series wri en out:
n
x
n=0
n!
=1+x+
x2
x3
x4
+
+
+
2
6
24
(8.6)
Notes:
441

We now nd the deriva ve:
f (x) =
=
xn1
n
n!
n=1
n=1
xn1
x2
=1+x+
+ .
(n 1)!
2!
Since the series starts at n = 1 and each term refers to (n 1), we can re-index
the series star ng with n = 0:
=
x
n=0
n!
= f(x).
We found the deriva ve of f(x) is f(x). The only func ons for which this is true
are of the form y = cex for some constant c. As f(0) = 1 (see Equa on (8.6)), c
must be 1. Therefore we conclude that
f(x) =
x
n=0
n!
= ex
for all x.
We can also nd f(x) dx:
f(x) dx = C +
n=0
=C+
n=0
xn+1
n!(n + 1)
xn+1
(n + 1)!
We write out a few terms of this last series:

C+
n=0
xn+1
x2
x3
x4
=C+x+
+
+
+
(n + 1)!
2
6
24
The integral of f(x) diers from f(x) only by a constant, again indica ng that
f(x) = ex . .
Example 257 and the work following Example 256 established rela onships
between a power series func on and regular func ons that we have dealt
with in the past. In general, given a power series func on, it is dicult (if not
Notes:
442
8.6 Power Series

impossible) to express the func on in terms of elementary func ons. We chose
examples where things worked out nicely.
In this sec ons last example, we show how to solve a simple dieren al
equa on with a power series.
. Example 258
.Solving a dieren al equa on with a power series.
Give the rst 4 terms of the power series solu on to y = 2y, where y(0) = 1.
The dieren al equa on y = 2y describes a func on y =
f(x) where the deriva ve of y is twice y and y(0) = 1. This is a rather simple
dieren al equa on; with a bit of thought one should realize that if y = Ce2x ,
then y = 2Ce2x , and hence y = 2y. By le ng C = 1 we sa sfy the ini al
condi on of y(0) = 1.
Lets ignore the fact that we already know the solu on and nd a power
series func on that sa ses the equa on. The solu on we seek will have the
form
f(x) =
an xn = a0 + a1 x + a2 x2 + a3 x3 +
S
n=0
for unknown coecients an . We can nd f (x) using Theorem 75:

f (x) =
an n xn1 = a1 + 2a2 x + 3a3 x2 + 4a4 x3 .
n=1
Since f (x) = 2f(x), we have
(
)
a1 + 2a2 x + 3a3 x2 + 4a4 x3 = 2 a0 + a1 x + a2 x2 + a3 x3 +
= 2a0 + 2a1 x + 2a2 x2 + 2a3 x3 +
The coecients of like powers of x must be equal, so we nd that

a1 = 2a0 ,
2a2 = 2a1 ,
3a3 = 2a2 ,
4a4 = 2a3 ,
etc.
The ini al condi on y(0) = f(0) = 1 indicates that a0 = 1; with this, we can
nd the values of the other coecients:
a0 = 1 and a1 = 2a0 a1 = 2;
a1 = 2 and 2a2 = 2a1 a2 = 4/2 = 2;
a2 = 2 and 3a3 = 2a2 a3 = 8/(2 3) = 4/3;
a3 = 4/3 and 4a4 = 2a3 a4 = 16/(2 3 4) = 2/3.
Thus the rst 5 terms of the power series solu on to the dieren al equa on
y = 2y is
4
2
f(x) = 1 + 2x + 2x2 + x3 + x4 +
3
3
Notes:
443

In Sec on 8.8, as we study Taylor Series, we will learn how to recognize this series as describing y = e2x . .
Our last example illustrates that it can be dicult to recognize an elementary
func on by its power series expansion. It is far easier to start with a known funcon, expressed in terms of elementary func ons, and represent it as a power
series func on. One may wonder why we would bother doing so, as the la er
func on probably seems more complicated. In the next two sec ons, we show
both how to do this and why such a process can be benecial.
Notes:
444
Exercises 8.6
Terms and Concepts
17.
1. We adopt the convenc on that x0 =

the value of x.
, regardless of
2. What is the dierence between the radius of convergence

and the interval of convergence?
3. If the radius of convergence of

an xn is 5, what is the radius of convergence of
n an x
n=1
4. If the radius of convergence of
an xn is 5, what is the ra-
n=0
dius of convergence of
(1)n an xn ?
n=0
In Exercises 5 8, write out the sum of the rst 5 terms of the

given power series.
5.
2n xn
6.
n=0
n=1
11.
n=0
n=1
n=0
15.
16.
n=0
(x + 2)n
n3
n=1
( x )n
10
n=0
(
)n
x+4
24.
n2
4
n=0
n!
In Exercises 25 30, a func on f(x) =
an xn is given.
xn
26.
n
n=1
27.
( )
x n
n=0
28.
nx
(1) (x 3)
n
n
(1)n (x 5)n
10n
5n (x 1)n
(2)n xn
n=0
22.
(3x)n
n=0
xn
13.
2n
n=0
14.
n=0
xn
21.
n2
n=1
(x + 4)n
12.
n!
n=0
(1)n n!(x 10)n
n=0
In Exercises 9 24, a power series is given.

(a) Find the radius of convergence.
(b) Find the interval of convergence.
(1)n+1 n
9.
x
n!
n=0
10.
20.
3n
(x 5)n
n!
n=0
n=0
(1)n 2n
x
(2n)!
n=0
19.
(a) Give a power series for f (x) and its interval of convergence.
(b) Give a power series for f(x) dx and its interval of convergence.
25.
nxn
1 n
x
n2
1 n
7.
x
n!
n=0
8.
n=0
23.
Problems
nxn
n n
18.
x
n
3
n=0
n=0
n1
29.
(1)n x2n
(2n)!
n=0
30.
(1)n xn
n!
n=0
In Exercises 31 36, give the rst 5 terms of the series that is

a solu on to the given dieren al equa on.
31. y = 3y,
y(0) = 1
32. y = 5y,
y(0) = 5
33. y = y , y(0) = 1
34. y = y + 1, y(0) = 1
35. y = y,
y(0) = 0, y (0) = 1
36. y = 2y, y(0) = 1, y (0) = 1
8.7 Taylor Polynomials
5
y = f(x)
x
4
2
y = p1 (x)
f (0) = 1
f (4) (0) = 12
f (5) (0) = 19
f(0) = 2
f (0) = 1
f (0) = 2
(
)
Consider a func on y = f(x) and a point c, f(c) . The deriva ve, f (c), gives
the instantaneous
rate of change of f at x = c. Of all lines that pass through the
(
)
point c, f(c) , the line that best approximates f at this point is the tangent line;
that is, the line whose slope (rate of change) is f (c).
In Figure 8.18, we see a func on y = f(x) graphed. The table below the
graph shows that f(0) = 2 and f (0) = 1; therefore, the tangent line to f at
x = 0 is p1 (x) = 1(x 0) + 2 = x + 2. The tangent line is also given in the gure.
Note that near x = 0, p1 (x) f(x); that is, the tangent line approximates f
well.
One shortcoming of this approxima on is that the tangent line only matches
the slope of f; it does not, for instance, match the concavity of f. We can nd a
polynomial, p2 (x), that does match the concavity without much diculty, though.
The table in Figure 8.18 gives the following informa on:
Figure 8.18: Plo ng y = f(x) and a table

of deriva ves of f evaluated at 0.
5
y = p2 (x)
x
2
y = p4 (x)
5
Figure 8.19: Plo ng f, p2 and p4 .
x
4
y = p13 (x)
.
.
f (0) = 2.
This is simply an ini alvalue problem. We can solve this using the techniques rst described in Sec on 5.1. To keep p2 (x) as simple as possible, well
assume that not only p2 (0) = 2, but that p2 (x) = 2. That is, the second derivave of p2 is constant.
If p2 (x) = 2, then p2 (x) = 2x + C for some constant C. Since we have
determined that p2 (0) = 1, we nd that C = 1 and so p2 (x) = 2x + 1. Finally,
we can compute p2 (x) = x2 + x + C. Using our ini al values, we know p2 (0) = 2
so C = 2. We conclude that p2 (x) = x2 + x + 2. This func on is plo ed with f in
Figure 8.19.
We can repeat this approxima on process by crea ng polynomials of higher
degree that match more of the deriva ves of f at x = 0. In general, a polynomial
of degree n can be created to match the rst n deriva ves of f. Figure 8.19 also
shows p4 (x) = x4 /2x3 /6+x2 +x+2, whose rst four deriva ves at 0 match
(4)
those of f. (Using the table in Figure 8.18, start with p4 (x) = 12 and solve
the related ini alvalue problem.)
As we use more and more deriva ves, our polynomial approxima on to f
gets be er and be er. In this example, the interval on which the approxima on
is good gets bigger and bigger. Figure 8.20 shows p13 (x); we can visually arm
that this polynomial approximates f very well on [2, 3]. (The polynomial p13 (x)
is not par cularly nice. It is
16901x13
13x12
1321x11
779x10
359x9
x8
139x7 11x6 19x5 x4 x3
2
+
+
+
+
+x +x+2.)
6227020800 1209600 39916800 1814400 362880 240 5040
360
120
2
6
Figure 8.20: Plo ng f and p13 .
Notes:
446
f (0) = 1
Therefore, we want our polynomial p2 (x) to have these same proper es. That
is, we need
p2 (0) = 2
p2 (0) = 1
p2 (0) = 2.
f(0) = 2
The polynomials we have created are examples of Taylor polynomials, named

a er the Bri sh mathema cian Brook Taylor who made important discoveries
about such func ons. While we created the above Taylor polynomials by solving
ini alvalue problems, it can be shown that Taylor polynomials follow a general
pa ern that make their forma on much more direct. This is described in the
following deni on.
Deni on 38
Taylor Polynomial, Maclaurin Polynomial
Let f be a func on whose rst n deriva ves exist at x = c.

1. The Taylor polynomial of degree n of f at x = c is
pn (x) = f(c)+f (c)(xc)+
f (c)
f (c)
f (n) (c)
(xc)2 +
(xc)3 + +
(xc)n .
2!
3!
n!
2. A special case of the Taylor polynomial is the Maclaurin polynomial, where c =

0. That is, the Maclaurin polynomial of degree n of f is
pn (x) = f(0) + f (0)x +
f (0) 2 f (0) 3
f (n) (0) n
x +
x + +
x .
2!
3!
n!
We will prac ce crea ng Taylor and Maclaurin polynomials in the following

examples.
. Example 259
.Finding and using Maclaurin polynomials
1. Find the nth Maclaurin polynomial for f(x) = ex .

2. Use p5 (x) to approximate the value of e.
f(x) = ex
f (x) = ex
f (x) = ex
..
.
f (n) (x) = ex
f(0) = 1
f (0) = 1
f (0) = 1
..
.
f (n) (0) = 1
Figure 8.21: The deriva ves of f(x) = ex

evaluated at x = 0.
1. We start with crea ng a table of the deriva ves of ex evaluated at x = 0.

In this par cular case, this is rela vely simple, as shown in Figure 8.21. By
Notes:
447

the deni on of the Maclaurin series, we have
f (0) 2 f (0) 3
f n (0) n
x +
x + +
x
2!
3!
n!
1
1
1
1
= 1 + x + x2 + x3 + x4 + + xn .
2
6
24
n!
10
pn (x) = f(0) + f (0)x +
2. Using our answer from part 1, we have

x
2
y = p5 (x)
1
1
1
1 5
p5 = 1 + x + x2 + x3 + x4 +
x .
2
6
24
120
Figure 8.22: A plot of f(x) = ex and its 5th

degree Maclaurin polynomial p5 (x).
To approximate the value of e, note that e = e1 = f(1) p5 (1). It is very

straigh orward to evaluate p5 (1):
p5 (1) = 1 + 1 +
A plot of f(x) = ex and p5 (x) is given in Figure 8.22.
.
f(x) = ln x
f (x) = 1/x
f (x) = 1/x2
f (x) = 2/x3
f (4) (x) = 6/x4
..
.
f (n) (x) =
(1)n+1 (n 1)!
xn
f(1) = 0
f (1) = 1
f (1) = 1
f (1) = 2
f (4) (1) = 6
..
.
f (n) (1) =
1 1
1
1
163
+ +
+
=
2.71667.
2 6 24 120
60
. Example 260
.Finding and using Taylor polynomials
1. Find the nth Taylor polynomial of y = ln x at x = 1.

2. Use p6 (x) to approximate the value of ln 1.5.
3. Use p6 (x) to approximate the value of ln 2.
(1)n+1 (n 1)!
Figure 8.23: Deriva ves of ln x evaluated

at x = 1.
1. We begin by crea ng a table of deriva ves of ln x evaluated at x = 1.

While this is not as straigh orward as it was in the previous example, a
pa ern does emerge, as shown in Figure 8.23.
Using Deni on 38, we have
f (c)
f (c)
f n (c)
(x c)2 +
(x c)3 + +
(x c)n
2!
3!
n!
(1)n+1
1
1
1
= 0 + (x 1) (x 1)2 + (x 1)3 (x 1)4 + +
(x 1)n .
2
3
4
n
pn (x) = f(c) + f (c)(x c) +
Note how the coecients of the (x 1) terms turn out to be nice.
Notes:
448

2. We can compute p6 (x) using our work above:
1
1
1
1
1
p6 (x) = (x1) (x1)2 + (x1)3 (x1)4 + (x1)5 (x1)6 .
2
3
4
5
6
y = ln x
Since p6 (x) approximates ln x well near x = 1, we approximate ln 1.5

p6 (1.5):
2
y = p6 (x)
1
1
1
p6 (1.5) = (1.5 1) (1.5 1)2 + (1.5 1)3 (1.5 1)4 +
2
3
4
1
1
+ (1.5 1)5 (1.5 1)6
5
6
259
=
640
0.404688.
..
Figure 8.24: A plot of y = ln x and its 6th

degree Taylor polynomial at x = 1.
This is a good approxima on as a calculator shows that ln 1.5 0.4055.

Figure 8.24 plots y = ln x with y = p6 (x). We can see that ln 1.5
p6 (1.5).
3. We approximate ln 2 with p6 (2):
1
1
1
p6 (2) = (2 1) (2 1)2 + (2 1)3 (2 1)4 +
2
3
4
1
1
5
6
+ (2 1) (2 1)
5
6
1 1 1 1 1
=1 + +
2 3 4 5 6
37
=
60
0.616667.
This approxima on is not terribly impressive: a hand held calculator shows
that ln 2 0.693147. The graph in Figure 8.24 shows that p6 (x) provides
less accurate approxima ons of ln x as x gets close to 0 or 2.
Surprisingly enough, even the 20th degree Taylor polynomial fails to approximate ln x for x > 2, as shown in Figure 8.25. Well soon discuss why
this is.
y
2
y = ln x
x
1
2
y = p20 (x)
4
Figure 8.25: A plot of y = ln x and its 20th

degree Taylor polynomial at x = 1.
Taylor polynomials are used to approximate func ons f(x) in mainly two situa ons:
Notes:
449

1. When f(x) is known, but perhaps hard to compute directly. For instance,
we can dene y = cos x as either the ra o of sides of a right triangle
(adjacent over hypotenuse) or with the unit circle. However, neither of
these provides a convenient way of compu ng cos 2. A Taylor polynomial
of suciently high degree can provide a reasonable method of compu ng
such values using only opera ons usually hardwired into a computer (+,
, and ).
Note: Even though Taylor polynomials

could be used in calculators and computers to calculate values of trigonometric func ons, in prac ce they generally
arent. Other more ecient and accurate
methods have been developed, such as
the CORDIC algorithm.
2. When f(x) is not known, but informa on about its deriva ves is known.
This occurs more o en than one might think, especially in the study of
dieren al equa ons.
In both situa ons, a cri cal piece of informa on to have is How good is my
approxima on? If we use a Taylor polynomial to compute cos 2, how do we
know how accurate the approxima on is?
We had the same problem when studying Numerical Integra on. Theorem
43 provided bounds on the error when using, say, Simpsons Rule to approximate
a denite integral. These bounds allowed us to determine that, for instance,
using 10 subintervals provided an approxima on within .01 of the exact value.
The following theorem gives similar bounds for Taylor (and hence Maclaurin)
polynomials.
Theorem 76
Taylors Theorem
1. Let f be a func on whose n + 1th deriva ve exists on an interval I and let c be in I.

Then, for each x in I, there exists zx between x and c such that
f(x) = f(c) + f (c)(x c) +
f (c)
f (n) (c)
(x c)2 + +
(x c)n + Rn (x),
2!
n!
f (n+1) (zx )
(x c)(n+1) .
(n + 1)!

max f (n+1) (z)

(x c)(n+1)
2. Rn (x)
(n + 1)!
where Rn (x) =
The rst part of Taylors Theorem states that f(x) = pn (x) + Rn (x), where
pn (x) is the nth order Taylor polynomial and Rn (x) is the remainder, or error, in
the Taylor approxima on. The second part gives bounds on how big that error
can be. If the (n + 1)th deriva ve is large, the error may be large; if x is far from
c, the error may also be large. However, the (n + 1)! term in the denominator
Notes:
450

tends to ensure that the error gets smaller as n increases.
The following example computes error es mates for the approxima ons of
ln 1.5 and ln 2 made in Example 260.
. Example 261
.Finding error bounds of a Taylor polynomial
Use Theorem 76 to nd error bounds when approxima ng ln 1.5 and ln 2 with
p6 (x), the Taylor polynomial of degree 6 of f(x) = ln x at x = 1, as calculated in
Example 260.
S
1. We start with the approxima on of ln 1.5 with p6 (1.5). The theorem references an open interval I that contains both x and c. The smaller the
interval we use the be er; it will give us a more accurate (and smaller!)
approxima on of the error. We let I = (0.9, 1.6), as this interval contains
both c = 1 and x = 1.5.

The theorem references max f (n+1) (z). In our situa on, this is asking
How big can the 7th deriva ve of y = ln x be on the interval (0.9, 1.6)?
The seventh deriva ve is y = 6!/x7 . The largest value it a ains on I is
about 1506. Thus we can bound the error as:
(7)

R6 (1.5) max f (z) (1.5 1)7
7!
1506 1
5040 27
0.0023.
We computed p6 (1.5) = 0.404688; using a calculator, we nd ln 1.5
0.405465, so the actual error is about 0.000778, which is less than our
bound of 0.0023. This arms Taylors Theorem; the theorem states that
our approxima on would be within about 2 thousandths of the actual
value, whereas the approxima on was actually closer.
2. We again nd an interval I that contains both c = 1 and x = 2; we choose
I = (0.9, 2.1). The maximum value of the seventh deriva ve of f on this
interval is again about 1506 (as the largest values come near x = 0.9).
Thus
(7)

R6 (2) max f (z) (2 1)7
7!
1506 7
1
5040
0.30.
Notes:
451

This bound is not as nearly as good as before. Using the degree 6 Taylor
polynomial at x = 1 will bring us within 0.3 of the correct answer. As
p6 (2) 0.61667, our error es mate guarantees that the actual value of
ln 2 is somewhere between 0.31667 and 0.91667. These bounds are not
par cularly useful.
In reality, our approxima on was only o by about 0.07. However, we

are approxima ng ostensibly because we do not know the real answer. In
order to be assured that we have a good approxima on, we would have
to resort to using a polynomial of higher degree.
We prac ce again. This me, we use Taylors theorem to nd n that guarantees our approxima on is within a certain amount.
f(x) = cos x
f (x) = sin x
f (x) = cos x
f (x) = sin x
f (4) (x) = cos x
f (5) (x) = sin x
f (6) (x) = cos x
f (7) (x) = sin x
f (8) (x) = cos x
f (9) (x) = sin x
f(0) = 1
f (0) = 0
f (0) = 1
f (0) = 0
f (4) (0) = 1
f (5) (0) = 0
f (6) (0) = 1
f (7) (0) = 0
f (8) (0) = 1
f (9) (0) = 0
. Example 262
.Finding suciently accurate Taylor polynomials
Find n such that the nth Taylor polynomial of f(x) = cos x at x = 0 approximates
cos 2 to within 0.001 of the actual answer. What is pn (2)?
Following Taylors theorem, we need bounds on the size of
S
the deriva ves of f(x) = cos x. In the case of this trigonometric func on, this is
easy. All deriva ves of cosine are sin x or cos x. In all cases, these func ons
are never greater than 1 in absolute value. We want the error to be less than
0.001. To nd the appropriate n, consider the following inequali es:

max f (n+1) (z)
(2 0)(n+1) 0.001
(n + 1)!
1
2(n+1) 0.001
(n + 1)!
Figure 8.26: A table of the deriva ves of

f(x) = cos x evaluated at x = 0.
We nd an n that sa ses this last inequality with trialanderror. When n = 8,

28+1
29+1
we have
0.0014; when n = 9, we have
0.000282 <
(8 + 1)!
(9 + 1)!
0.001. Thus we want to approximate cos 2 with p9 (2).
We now set out to compute p9 (x). We again need a table of the deriva ves
of f(x) = cos x evaluated at x = 0. A table of these values is given in Figure 8.26.
No ce how the deriva ves, evaluated at x = 0, follow a certain pa ern. All the
odd powers of x in the Taylor polynomial will disappear as their coecient is 0.
While our error bounds state that we need p9 (x), our work shows that this will
be the same as p8 (x).
Since we are forming our polynomial at x = 0, we are crea ng a Maclaurin
Notes:
452

y
polynomial, and :
f (0) 2 f (0) 3
f (8) 8
x +
x + +
x
2!
3!
8!
1
1
1
1
= 1 x2 + x4 x6 + x8
2!
4!
6!
8!
We nally approximate cos 2:
131
cos 2 p8 (2) =
0.41587.
315
Our error bound guarantee that this approxima on is within 0.001 of the correct
answer. Technology shows us that our approxima on is actually within about
0.0003 of the correct answer.
Figure 8.27 shows a graph of y = p8 (x) and y = cos x. Note how well the
two func ons agree on about (, ). .
p8 (x) = f(0) + f (0)x +
. Example 263
y = p8 (x)
x
5 4 3 2 1
.
.
f(x) = cos x
Figure 8.27: A graph of f(x) = cos x and

its degree 8 Maclaurin polynomial.
Finding and using Taylor polynomials
1. Find the degree 4 Taylor polynomial, p4 (x), for f(x) = x at x = 4.
2. Use p4 (x) to approximate 3.
3. Find bounds on the error when approxima ng 3 with p4 (3).

S
1. We begin by evalua ng the deriva ves of f at x = 4. This is done in Figure

8.28. These values allow us to form the Taylor polynomial p4 (x):
1
1/32
3/256
15/2048
p4 (x) = 2+ (x4)+
(x4)2 +
(x4)3 +
(x4)4 .
4
2!
3!
4!
2. As p4 (x) x near x = 4, we approximate 3 with p4 (3) = 1.73212.

3. To nd a bound on the error, we need an open interval that contains x = 3
and x =4. We set I = (2.9, 4.1). The largest value the h deriva ve of
f(x) = x takes on this interval is near x = 2.9, at about 0.0273. Thus

R4 (3) 0.0273 (3 4)5 0.00023.
5!
This shows our approxima on is accurate to at least the rst 2 places a er
the decimal. (It turns out that our approxima on
is actually accurate to
4 places a er the decimal.) A graph of f(x) = x and p4 (x) is given in
Figure 8.29. Note how the two func ons are nearly indis nguishable on
.
(2, 7).
f(x) =
1
f (x) =
2 x
1
f (x) = 3/2
4x
3
f (x) = 5/2
8x
15
f (4) (x) =
16x7/2
f(4) = 2
1
4
1
f (4) =
32
3
f (4) =
256
15
f (4) (4) =
2048
f (4) =
Figure 8.28:
A table of the deriva ves of
f(x) = x evaluated at x = 4.
y
3
Notes:
y= x
y = p4 (x)
x
10
Figure 8.29: A graph of f(x) = x and its

degree 4 Taylor polynomial at x = 4.
453

Our nal example gives a brief introduc on to using Taylor polynomials to
solve dieren al equa ons.
. Example 264
.Approxima ng an unknown func on
A func on y = f(x) is unknown save for the following two facts.
1. y(0) = f(0) = 1, and
2. y = y2
(This second fact says that amazingly, the deriva ve of the func on is actually
the func on squared!)
Find the degree 3 Maclaurin polynomial p3 (x) of y = f(x).
One might ini ally think that not enough informa on is given
S
to nd p3 (x). However, note how the second fact above actually lets us know
what y (0) is:
y = y2 y (0) = y2 (0).
Since y(0) = 1, we conclude that y (0) = 1.
Now we nd informa on about y . Star ng with y = y2 , take deriva ves of
both sides, with respect to x. That means we must use implicit dieren a on.
y = y2
d ( )
d ( 2)
y =
y
dx
dx
y = 2y y .
Now evaluate both sides at x = 0:
y (0) = 2y(0) y (0)
y (0) = 2
We repeat this once more to nd y (0). We again use implicit dieren a on;
this me the Product Rule is also required.
d ( )
d ( )
y =
2yy
dx
dx
y = 2y y + 2y y .
Now evaluate both sides at x = 0:
y (0) = 2y (0)2 + 2y(0)y (0)
y (0) = 2 + 4 = 6
Notes:
454

In summary, we have:
y(0) = 1
y (0) = 1
y (0) = 2
y (0) = 6.
We can now form p3 (x):
2
6
p3 (x) = 1 + x + x2 + x3
2!
3!
= 1 + x + x2 + x3 .
It turns out that the dieren al equa on we started with, y = y2 , where y(0) =
1
1, can be solved without too much diculty: y =
. Figure 8.30 shows this
1x
func on plo ed with p3 (x). Note how similar they are near x = 0. .
1
1x
y = p3 (x)
y=
x
0.5
0.5
Figure 8.30: A graph of y = 1/(x 1)

and y = p3 (x) from Example 264.
It is beyond the scope of this text to pursue error analysis when using Taylor polynomials to approximate solu ons to dieren al equa ons. This topic is
o en broached in introductory Dieren al Equa ons courses and usually covered in depth in Numerical Analysis courses. Such an analysis is very important;
one needs to know how good their approxima on is. We explored this example
simply to demonstrate the usefulness of Taylor polynomials.
Most of this chapter has been devoted to the study of innite series. This
sec on has taken a step back from this study, focusing instead on nite summaon of terms. In the next sec on, we explore Taylor Series, where we represent
a func on with an innite series.
Notes:
455
Exercises 8.7
Terms and Concepts
21. Approximate sin 0.1 with the Maclaurin polynomial of degree 3.
1. What is the dierence between a Taylor polynomial and a

Maclaurin polynomial?
2. T/F: In general, pn (x) approximates f(x) be er and be er
as n gets larger.
3. For some func on f(x), the Maclaurin polynomial of degree
4 is p4 (x) = 6 + 3x 4x2 + 5x3 7x4 . What is p2 (x)?
4. For some func on f(x), the Maclaurin polynomial of degree
4 is p4 (x) = 6 + 3x 4x2 + 5x3 7x4 . What is f (0)?
Problems
5. f(x) = ex ,
n=3
6. f(x) = sin x,
n=8
7. f(x) = x e ,
n=5
8. f(x) = tan x,
n=6
9. f(x) = e ,
24. Approximate ln 1.5 with the Taylor polynomial of degree 3

centered at x = 1.
Exercises 25 28 ask for an n to be found such that pn (x) approximates f(x) within a certain bound of accuracy.
25. Find n such that the Maclaurin polynomial of degree n of
f(x) = ex approximates e within 0.0001 of the actual value.
In Exercises 5 12, nd the Maclaurin polynomial of degree

n for the given func on.
2x
22. Approximate cos 1 with the Maclaurin polynomial of degree 4.
23. Approximate 10 with the Taylor polynomial of degree 2

centered at x = 9.

f(x) = cos x approximates cos /3 within 0.0001 of the actual value.
n=4
1
, n=4
1x
1
11. f(x) =
, n=4
1+x
1
12. f(x) =
, n=7
1+x
In Exercises 13 20, nd the Taylor polynomial of degree n,
at x = c, for the given func on.
13. f(x) = x, n = 4, c = 1
10. f(x) =
14. f(x) = ln(x + 1),
n = 4,
c=1
15. f(x) = cos x,
n = 6,
c = /4
16. f(x) = sin x,
n = 5,
c = /6
1
, n = 5, c = 2
x
1
18. f(x) = 2 , n = 8, c = 1
x
1
19. f(x) = 2
, n = 3, c = 1
x +1
17. f(x) =
20. f(x) = x2 cos x,
26. Find
degree n of f(x) =
n such that the Taylor polynomial of
x, centered at x = 4, approximates 3 within 0.0001 of
the actual value.
n = 2,
c=
In Exercises 21 24, approximate the func on value with the

indicated Taylor polynomial and give approximate bounds on
the error.

f(x) = sin x approximates cos within 0.0001 of the actual
value.
In Exercises 29 33, nd the nth term of the indicated Taylor
polynomial.
29. Find a formula for the nth term of the Maclaurin polynomial
for f(x) = ex .
for f(x) = cos x.
1
for f(x) =
.
1x
1
for f(x) =
.
1+x
33. Find a formula for the nth term of the Taylor polynomial for
f(x) = ln x.
In Exercises 34 36, approximate the solu on to the given
dieren al equa on with a degree 4 Maclaurin polynomial.
34. y = y,
y(0) = 1
35. y = 5y,
y(0) = 3
2
,
y
y(0) = 1
36. y =
8.8 Taylor Series
8.8
Taylor Series
In Sec on 8.6, we showed how certain func ons can be represented by a power
series func on. In 8.7, we showed how we can approximate func ons with polynomials, given that enough deriva ve informa on is available. In this sec on we
combine these concepts: if a func on f(x) is innitely dieren able, we show
how to represent it with a power series func on.
Deni on 39
Taylor and Maclaurin Series
Let f(x) have deriva ves of all orders at x = c.

1. The Taylor Series of f(x), centered at c is
f (n) (c)
n=0
n!
(x c)n .
2. Se ng c = 0 gives the Maclaurin Series of f(x):
f (n) (0)
n=0
n!
xn .
The dierence between a Taylor polynomial and a Taylor series is the former
is a polynomial, containing only a nite number of terms, whereas the la er is
a series, a summa on of an innite set of terms. When crea ng the Taylor polynomial of degree n for a func on f(x) at x = c, we needed to evaluate f, and the
rst n deriva ves of f, at x = c. When crea ng the Taylor series of f, it helps to
nd a pa ern that describes the nth deriva ve of f at x = c. We demonstrate
this in the next two examples.
. Example 265
.The Maclaurin series of f(x) = cos x
Find the Maclaurin series of f(x) = cos x.
In Example 262 we found the 8th degree Maclaurin polynomial of cos x. In doing so, we created the table shown in Figure 8.31. No ce how
f (n) (0) = 0 when n is odd, f (n) (0) = 1 when n is divisible by 4, and f (n) (0) = 1
when n is even but not divisible by 4. Thus the Maclaurin series of cos x is
S
x2
x4
x6
x8
+
+

2
4!
6!
8!
f(x) = cos x
f (x) = sin x
f (x) = cos x
f (x) = sin x
f (4) (x) = cos x
f (5) (x) = sin x
f (6) (x) = cos x
f (7) (x) = sin x
f (8) (x) = cos x
f (9) (x) = sin x
f(0) = 1
f (0) = 0
f (0) = 1
f (0) = 0
f (4) (0) = 1
f (5) (0) = 0
f (6) (0) = 1
f (7) (0) = 0
f (8) (0) = 1
f (9) (0) = 0
Figure 8.31: A table of the deriva ves of

f(x) = cos x evaluated at x = 0.
Notes:
457

We can go further and write this as a summa on. Since we only need the terms
where the power of x is even, we write the power series in terms of x2n :
x2n
(1)n
.
(2n)!
n=0
.
. Example 266
The Taylor series of f(x) = ln x at x = 1
Find the Taylor series of f(x) = ln x centered at x = 1.
f(x) = ln x
f (x) = 1/x
f (x) = 1/x2
f (x) = 2/x3
f (4) (x) = 6/x4
f (5) (x) = 24/x5
..
.
f (n) (x) =
(1)n+1 (n 1)!
xn
f(1) = 0
f (1) = 1
f (1) = 1
f (1) = 2
f (4) (1) = 6
f (5) (1) = 24
..
.
f (n) (1) =
(1)
n+1
(n 1)!
Figure 8.32: Deriva ves of ln x evaluated

at x = 1.
Figure 8.32 shows the nth deriva ve of ln x evaluated at x =

1 for n = 0, . . . , 5, along with an expression for the nth term:
S
f (n) (1) = (1)n+1 (n 1)!
for n 1.
Remember that this is what dis nguishes Taylor series from Taylor polynomials;
we are very interested in nding a pa ern for the nth term, not just nding a
nite set of coecients for a polynomial. Since f(1) = ln 1 = 0, we skip the
rst term and start the summa on with n = 1, giving the Taylor series for ln x,
centered at x = 1, as
1
(x 1)n
.
(1)n+1 (n 1)! (x 1)n =
(1)n+1
n!
n
n=1
n=1
It is important to note that Deni on 39 denes a Taylor series given a funcon f(x); however, we cannot yet state that f(x) is equal to its Taylor series. We
will nd that most of the me they are equal, but we need to consider the
condi ons that allow us to conclude this.
Theorem 76 states that the error between a func on f(x) and its nth degree
Taylor polynomial pn (x) is Rn (x), where
(n+1)

(z)
Rn (x) max f
(x c)(n+1) .
(n + 1)!
If Rn (x) goes to 0 for each x in an interval I as n approaches innity, we conclude that the func on is equal to its Taylor series expansion.
Notes:
458
8.8 Taylor Series
Theorem 77
Func on and Taylor Series Equality
Let f(x) have deriva ves of all orders at x = c, let Rn (x) be as stated in
Theorem 76, and let I be an interval on which the Taylor series of f(x)
converges. If lim Rn (x) = 0 for all x in I containing c, then
f(x) =
f (n) (c)
n=0
n!
(x c)n
on I.
We demonstrate the use of this theorem in an example.

. Example 267
Establishing equality of a func on and its Taylor series
Show that f(x) = cos x is equal to its Maclaurin series, as found in Example 265,
for all x.
S
bounded by
Given a value x, the magnitude of the error term Rn (x) is

max f (n+1) (z) (n+1)
Rn (x)
x
.
(n + 1)!
Since all deriva ves of cos x are sin x or cos x, whose magnitudes are bounded
by 1, we can state

(n+1)
1
Rn (x)
x
.
(n + 1)!
xn+1
= 0. Thus by the Squeeze Theorem, we conclude that
n (n + 1)!
lim Rn (x) = 0 for all x, and hence
For any x, lim

n
cos x =
x2n
(1)n
(2n)!
n=0
for all x.
It is natural to assume that a func on is equal to its Taylor series on the

series interval of convergence, but this is not the case. In order to properly
establish equality, one must use Theorem 77. This is a bit disappoin ng, as we
developed beau ful techniques for determining the interval of convergence of
a power series, and proving that Rn (x) 0 can be cumbersome as it deals with
high order deriva ves of the func on.
There is good news. A func on f(x) that is equal to its Taylor series, centered
at any point the domain of f(x), is said to be an analy c func on, and most, if
not all, func ons that we encounter within this course are analy c func ons.
Notes:
459

Generally speaking, any func on that one creates with elementary func ons
(polynomials, exponen als, trigonometric func ons, etc.) that is not piecewise
dened is probably analy c. For most func ons, we assume the func on is equal
to its Taylor series on the series interval of convergence and only use Theorem
77 when we suspect something may not work as expected.
We develop the Taylor series for one more important func on, then give a
table of the Taylor series for a number of common func ons.
. Example 268
.The Binomial Series
Find the Maclaurin series of f(x) = (1 + x)k , k = 0.
When k is a posi ve integer, the Maclaurin series is nite.
S
For instance, when k = 4, we have
f(x) = (1 + x)4 = 1 + 4x + 6x2 + 4x3 + x4 .
The coecients of x when k is a posi ve integer are known as the binomial coecients, giving the series we are developing its name.
When k = 1/2, we have f(x) = 1 + x. Knowing a

series representa on of
this func on would give a useful way of approxima ng 1.3, for instance.
To develop the Maclaurin series for f(x) = (1 + x)k for any value of k = 0,
we consider the deriva ves of f evaluated at x = 0:
f(x) = (1 + x)k
f(0) = 1
f (x) = k(1 + x)
f (0) = k
f (x) = k(k 1)(1 + x)k2
f (0) = k(k 1)
f (x) = k(k 1)(k 2)(1 + x)k3
f (0) = k(k 1)(k 2)
k1
..
.
(
)
f (n) (x) = k(k 1) k (n 1) (1 + x)kn
..
.
(
)
f (n) (0) = k(k 1) k (n 1)
Thus the Maclaurin series for f(x) = (1 + x)k is

(
)
k(k 1) k (n 1)
k(k 1) k(k 1)(k 2)
1+k+
+
+ ... +
+ ...
2!
3!
n!
It is important to determine the interval of convergence of this series. With
(
)
k(k 1) k (n 1) n
an =
x ,
n!
Notes:
460
8.8 Taylor Series

we apply the Ra o Test:
(
)

/
k(k 1) (k n) n+1
k(k
1)
(n
1)
|an+1 |

lim
= lim
x
xn

n |an |
n

(n + 1)!
n!

k n
= lim
x
n
n
= |x|.
The series converges absolutely when the limit of the Ra o Test is less than
1; therefore, we have absolute convergence when |x| < 1.
While outside the scope of this text, the interval of convergence depends
on the value of k. When k > 0, the interval of convergence is [1, 1]. When
1 < k < 0, the interval of convergence is [1, 1). If k 1, the interval of
convergence is (1, 1). .
We learned that Taylor polynomials oer a way of approxima ng a dicult
to compute func on with a polynomial. Taylor series oer a way of exactly
represen ng a func on with a series. One probably can see the use of a good
approxima on; is there any use of represen ng a func on exactly as a series?
While we should not overlook the mathema cal beauty of Taylor series (which
is reason enough to study them), there are prac cal uses as well. They provide
a valuable tool for solving a variety of problems, including problems rela ng to
integra on and dieren al equa ons.
In Key Idea 32 (on the following page) we give a table of the Taylor series
of a number of common func ons. We then give a theorem about the algebra
of power series, that is, how we can combine power series to create power
series of new func ons. This allows us to nd the Taylor series of func ons like
f(x) = ex cos x by knowing the Taylor series of ex and cos x.
Before we inves gate combining func ons, consider the Taylor series for the
arctangent func on (see Key Idea 32). Knowing that tan1 (1) = /4, we can
use this series to approximate the value of :
1 1 1 1
= tan1 (1) = 1 + +
4
3 5 7 )9
(
1 1 1 1
= 4 1 + +
3 5 7 9
Unfortunately, this par cular expansion of converges very slowly. The rst
100 terms approximate as 3.13159, which is not par cularly good.
Notes:
461
Key Idea 32
Important Taylor Series Expansions
Func on and Series

ex =
First Few Terms
x
n=0
1+x+
n!
sin x =
x2n+1
(1)n
(2n + 1)!
n=0
cos x =
x2n
(1)n
(2n)!
n=0
(x 1)n
ln x =
(1)n+1
n
n=1
x2
x3
+
+
2!
3!
Interval of
Convergence
(, )
x3
x5
x7
+
+
3!
5!
7!
(, )
x2
x4
x6
+
+
2!
4!
6!
(, )
(x 1)
(x 1)2
(x 1)3
+

2
3
(0, 2]
1
=
xn
1x
n=0
1 + x + x2 + x3 +
(1 + x)k =
(
)
k(k 1) k (n 1) n
x
n!
n=0
1 + kx +
tan1 x =
x2n+1
(1)n
2n + 1
n=0
a Convergence at x
Theorem 78
Let f(x) =
x3
x5
x7
+
+
3
5
7
(1, 1)a
[1, 1]
= 1 depends on the value of k.
Algebra of Power Series

an xn and g(x) =
n=0
bn xn converge absolutely for |x| < R, and let h(x) be con nuous.
(an bn )xn
n=0
2. f(x)g(x) =
)(
an xn
n=0
(
)
3. f h(x) =
n=0
1. f(x) g(x) =
(
)n
an h(x)
n=0
Notes:
462
k(k 1) 2
x +
2!
(1, 1)
n=0
for |x| < R.

)
bn xn
(
n=0
for |h(x)| < R.
)
a0 bn + a1 bn1 + . . . an b0 xn for |x| < R.
8.8 Taylor Series

. Example 269
Combining Taylor series
Write out the rst 3 terms of the Taylor Series for f(x) = ex cos x using Key Idea
32 and Theorem 78.
Key Idea 32 informs us that
ex = 1 + x +
x2
x3
+
+
2!
3!
and
cos x = 1
x2
x4
+
+ .
2!
4!
Applying Theorem 78, we nd that

(
)(
)
x2
x3
x2
x4
x
e cos x = 1 + x +
+
+
1
+
+ .
2!
3!
2!
4!
Distribute the right hand expression across the le :
(
)
(
)
(
)
x4
x2
x4
x2
x2
x4
x2
+
+ + x 1
+
+ +
1
+
+
=1 1
2!
4!
2!
4!
2!
2!
4!
(
)
(
)
x3
x4
x4
x4
x2
x2
+
+
+ +
+
+ +
1
1
3!
2!
4!
4!
2!
4!
Distribute again and collect like terms.
=1+x
x3
x4
x5
x7
+
+
3
6
30 630
While this process is a bit tedious, it is much faster than evalua ng all the necessary deriva ves of ex cos x and compu ng the Taylor series directly.
Because the series for ex and cos x both converge on (, ), so does the
series expansion for ex cos x. .
. Example 270
.Crea ng new Taylor series
Use Theorem 78 to create series for y = sin(x2 ) and y = ln( x).

Given that
sin x =
x2n+1
x3
x5
x7
(1)n
=x
+
+ ,
(2n + 1)!
3!
5!
7!
n=0
we simply subs tute x2 for x in the series, giving
(x2 )2n+1
x6
x10
x14
sin(x ) =
(1)n
= x2
+
.
(2n + 1)!
3!
5!
7!
n=0
2
Notes:
463

Since the Taylor series for sin x has an innite radius of convergence, so does the
Taylor series for sin(x2 ).
The Taylor expansion forln x given in Key Idea 32 is centered at x = 1, so we
will center the series for ln( x) at x = 1 as well. With
ln x =
(x 1)n
(x 1)2
(x 1)3
(1)n+1
= (x 1)
+
,
n
2
3
n=1
x for x to obtain
( x 1)2 ( x 1)3
n+1 ( x 1)
ln( x) =
(1)
= ( x 1)
+
.
n
2
3
n=1
we subs tute
While this isnot strictly a power series, it is a series that allows us to study the
func
on ln( x). Since the interval of convergence of ln x is (0, 2], and the range
of x on (0, 4] is (0, 2], the interval of convergence of this series expansion of
ln( x) is (0, 4]. .
. Example 271
.Using Taylor series to evaluate denite integrals

1
Use the Taylor series of ex to evaluate
ex dx.
0
2
We learned, when studying Numerical Integra on, that ex

does not have an an deriva ve expressible in terms of elementary func ons.
This means any denite integral of this func on must have its value approximated, and not computed exactly.
2
We can quickly write out the Taylor series for ex using the Taylor series of
ex :
S
ex =
n!
n=0
=1+x+
x2
x3
+
+
2!
3!
and so
2
ex =
=
(x2 )n
n=0
n!
(1)n
n=0
= 1 x2 +
Notes:
464
x2n
n!
x4
x6
+
2!
3!
8.8 Taylor Series

We use Theorem 75 to integrate:
2
x3
x5
x7
x2n+1
ex dx = C + x
+
+ + (1)n
+
3
5 2! 7 3!
(2n + 1)n!
2
This is the an deriva ve of ex ; while we can write it out as a series, we cannot write it out in terms of elementary func ons. We can evaluate the denite
1
integral
ex dx using this an deriva ve; subs tu ng 1 and 0 for x and sub-
trac ng gives
ex dx = 1
1
1
1
1
+
+
.
3 5 2! 7 3! 9 4!
Summing the 5 terms shown above give the approxima on of 0.74749. Since
this is an alterna ng series, we can use the Alterna ng Series Approxima on
Theorem, (Theorem 71), to determine how accurate this approxima on is. The
next term of the series is 1/(11 5!) 0.00075758. Thus we know our approxima on is within 0.00075758 of the actual value of the integral. This is arguably
much less work than using Simpsons Rule to approximate the value of the integral. .
. Example 272
Using Taylor series to solve dieren al equa ons
Solve the dieren al equa on y = 2y in terms of a power series, and use the
theory of Taylor series to recognize the solu on in terms of an elementary funcon.
We found the rst 5 terms of the power series solu on to
S
this dieren al equa on in Example 258 in Sec on 8.6. These are:
4
8
4
16
2
= 2, a3 =
= , a4 =
= .
2
23
3
234
3
We include the unsimplied expressions for the coecients found in Example
258 as we are looking for a pa ern. It can be shown that an = 2n /n!. Thus the
solu on, wri en as a power series, is
a0 = 1,
a1 = 2,
a2 =
y=
2n
n=0
n!
xn =
(2x)n
n=0
n!
Using Key Idea 32 and Theorem 78, we recognize f(x) = e2x :

ex =
.
x
n=0
n!
e2x =
(2x)n
n=0
n!
Notes:
465

Finding a pa ern in the coecients that match the series expansion of a
known func on, such as those shown in Key Idea 32, can be dicult. What if
the coecients in the previous example were given in their reduced form; how
could we s ll recover the func on y = e2x ?
Suppose that all we know is that
a0 = 1,
a1 = 2,
a2 = 2,
a3 =
4
,
3
a4 =
2
.
3
Deni on 39 states that each term of the Taylor expansion of a func on includes
an n!. This allows us to say that
a2 = 2 =
b2
,
2!
a3 =
4
b3
= ,
3
3!
and
a4 =
2
b4
=
3
4!
for some values b2 , b3 and b4 . Solving for these values, we see that b2 = 4,
b3 = 8 and b4 = 16. That is, we are recovering the pa ern we had previously
seen, allowing us to write
f(x) =
n=0
an x =
bn
n=0
n!
xn
= 1 + 2x +
4 2
8
16
x + x3 + x4 +
2!
3!
4!
From here it is easier to recognize that the series is describing an exponen al

func on.
There are simpler, more direct ways of solving the dieren al equa on y =
2y. We applied power series techniques to this equa on to demonstrate its u lity, and went on to show how some mes we are able to recover the solu on in
terms of elementary func ons using the theory of Taylor series. Most dierenal equa ons faced in real scien c and engineering situa ons are much more
complicated than this one, but power series can oer a valuable tool in nding,
or at least approxima ng, the solu on.
Notes:
466
Exercises 8.8
Terms and Concepts
1. What is the dierence between a Taylor polynomial and a
Taylor series?
2. What theorem must we use to show that a func on is equal
to its Taylor series?
Problems
Key Idea 32 gives the nth term of the Taylor series of common
func ons. In Exercises 3 6, verify the formula given in the
Key Idea by nding the rst few terms of the Taylor series of
the given func on and iden fying a pa ern.
3. f(x) = ex ;
c=0
4. f(x) = sin x;
15. f(x) = ln x
16. f(x) = 1/(1 x) (show equality only on (1, 0))
In Exercises 17 20, use the Taylor series given in Key Idea 32
to verify the given iden ty.
17. cos(x) = cos x
18. sin(x) = sin x
(
)
19. dxd sin x = cos x
(
)
20. dxd cos x = sin x
In Exercises 21 24, write out the rst 5 terms of the Binomial
series with the given k-value.
21. k = 1/2
c=0
22. k = 1/2
5. f(x) = 1/(1 x); c = 0
23. k = 1/3
6. f(x) = tan1 x; c = 0
24. k = 4
In Exercises 7 12, nd a formula for the nth term of the Taylor

series of f(x), centered at c, by nding the coecients of the
rst few powers of x and looking for a pa ern. (The formulas for several of these are found in Key Idea 32; show work
verifying these formula.)
7. f(x) = cos x;
c = /2
8. f(x) = 1/x;
c=1
9. f(x) = ex ;
c=0
10. f(x) = ln(1 + x);
c=0
11. f(x) = x/(x + 1);
c=1
12. f(x) = sin x;
c = /4
In Exercises 13 16, show that the Taylor series for f(x), as

given in Key Idea 32, is equal to f(x) by applying Theorem 77;
that is, show lim Rn (x) = 0.
n
In Exercises 25 30, use the Taylor series given in Key Idea 32

to create the Taylor series of the given func ons.
( )
25. f(x) = cos x2
26. f(x) = ex
(
)
27. f(x) = sin 2x + 3
( )
28. f(x) = tan1 x/2
29. f(x) = ex sin x
30. f(x) = (1 + x)
cos x
(only nd the rst 4 terms)
In Exercises 31 32, approximate the value of the given definite integral by using the rst 4 nonzero terms of the integrands Taylor series.

( )
sin x2 dx
31.
0
13. f(x) = ex
14. f(x) = sin x
(only nd the rst 4 terms)

1/2
2 /4
32.
cos
0
( )
x dx
9: C
9.1
Conic Sec ons
The ancient Greeks recognized that interes ng shapes can be formed by intersec ng a plane with a double napped cone (i.e., two iden cal cones placed p
to p as shown in the following gures). As these shapes are formed as sec ons
of conics, they have earned the ocial name conic sec ons.
The three most interes ng conic sec ons are given in the top row of Figure
9.1. They are the parabola, the ellipse (which includes circles) and the hyperbola. In each of these cases, the plane does not intersect the ps of the cones
(usually taken to be the origin).
Parabola
Ellipse
Point
Circle
Line
Crossed Lines
Figure 9.1: Nondegenerate Conic Sec ons
When the plane does contain the origin, three degenerate cones can be
formed as shown the bo om row of Figure 9.1: a point, a line, and crossed
lines. We focus here on the nondegenerate cases.
While the above geometric constructs dene the conics in an intui ve, visual
way, these constructs are not very helpful when trying to analyze the shapes
Hyperbola
Chapter 9 Curves in the Plane

algebraically or consider them as the graph of a func on. It can be shown that
all conics can be dened by the general seconddegree equa on
Ax2 + Bxy + Cy2 + Dx + Ey + F = 0.
While this algebraic deni on has its uses, most nd another geometric perspec ve of the conics more benecial.
Each nondegenerate conic can be dened as the locus, or set, of points that
sa sfy a certain distance property. These distance proper es can be used to
generate an algebraic formula, allowing us to study each conic as the graph of a
func on.
Parabolas
.
Deni on 40
Parabola
Axis of
Focus
Symmetry
A parabola is the locus of all points equidistant from a point (called a

focus) and a line (called the directrix) that does not contain the focus.
d
p
Vertex
Directrix
(x, y)
.}
Figure 9.2: Illustra ng the deni on of

the parabola and establishing an algebraic formula.
Figure 9.2 illustrates this deni on. The point halfway between the focus
and the directrix is the vertex. The line through the focus, perpendicular to the
directrix, is the axis of symmetry, as the por on of the parabola on one side of
this line is the mirrorimage of the por on on the opposite side.
The deni on leads us to an algebraic formula for the parabola. Let P =
(x, y) be a point on a parabola whose focus is at F = (0, p) and whose directrix
is at y = p. (Well assume for now that the focus lies on the y-axis; by placing
the focus p units above the x-axis and the directrix p units below this axis, the
vertex will be at (0, 0).)
We use the Distance Formula to nd the distance d1 between F and P:
d1 = (x 0)2 + (y p)2 .
The distance d2 from P to the directrix is more straigh orward:
d2 = y (p) = y + p.
These two distances are equal. Se ng d1 = d2 , we can solve for y in terms of x:
d1 = d2
x2 + (y p)2 = y + p
Notes:
470
9.1 Conic Sec ons

Now square both sides.
x2 + (y p)2 = (y + p)2
x2 + y2 2yp + p2 = y2 + 2yp + p2
x2 = 4yp
1 2
y=
x .
4p
The geometric deni on of the parabola has led us to the familiar quadra c
func on whose graph is a parabola with vertex at the origin. When we allow the
vertex to not be at (0, 0), we get the following standard form of the parabola.
Key Idea 33
General Equa on of a Parabola
1. Ver cal Axis of Symmetry: The equa on of the parabola with vertex at (h, k) and directrix y = k p in standard form is
y=
The focus is at (h, k + p).
1
(x h)2 + k.
4p
2. Horizontal Axis of Symmetry: The equa on of the parabola with

vertex at (h, k) and directrix x = h p in standard form is
x=
1
(y k)2 + h.
4p
The focus is at (h + p, k).

Note: p is not necessarily a posi ve number.
y
2
. Example 273
Finding the equa on of a parabola
Give the equa on of the parabola with focus at (1, 2) and directrix at y = 3.
x
2
1
1
(x 1)2 + 2.5 = (x 1)2 + 2.5.
4(0.5)
2
The parabola is sketched in Figure 9.3. .
The vertex is located halfway between the focus and directrix, so (h, k) = (1, 2.5). This gives p = 0.5. Using Key Idea 33 we have the
equa on of the parabola as
S
y=
Figure 9.3: The parabola described in Example 273.
Notes:
471

. Example 274
Finding the focus and directrix of a parabola
Find the focus and directrix of the parabola x = 18 y2 y + 1. The point (7, 12)
lies on the graph of this parabola; verify that it is equidistant from the focus and
directrix.
We need to put the equa on of the parabola in its general
S
form. This requires us to complete the square:
y
10
10
x=
10
5
=
x
10
10
Figure 9.4: The parabola described in Example 274. The distances from a point on
the parabola to the focus and directrix is
given.
=
=
1 2
y y+1
8
)
1( 2
y 8y + 8
8
)
1( 2
y 8y + 16 16 + 8
8
)
1(
(y 4)2 8
8
1
(y 4)2 1.
8
1
Hence the vertex is located at (1, 4). We have 81 = 4p
, so p = 2. We conclude
that the focus is located at (1, 4) and the directrix is x = 3. The parabola is
graphed in Figure 9.4, along with its focus and directrix.
The point (7, 12) lies on the graph and is 7 (3) = 10 units from the
directrix. The distance from (7, 12) to the focus is:
(7 1)2 + (12 4)2 =
100 = 10.
Indeed, the point on the parabola is equidistant from the focus and directrix. .
Reec ve Property
One of the fascina ng things about the nondegenerate conic sec ons is their
reec ve proper es. Parabolas have the following reec ve property:
.
Figure 9.5: Illustra ng the parabolas reec ve property.
Any ray emana ng from the focus that intersects the parabola
reects o along a line perpendicular to the directrix.
This is illustrated in Figure 9.5. The following theorem states this more rigorously.
Notes:
472
9.1 Conic Sec ons
Theorem 79
Reec ve Property of the Parabola
Let P be a point on a parabola. The tangent line to the parabola at P

makes equal angles with the following two lines:
1. The line containing P and the focus F, and

2. The line perpendicular to the directrix through P.
Because of this reec ve property, paraboloids (the 3D analogue of parabolas) make for useful ashlight reectors as the light from the bulb, ideally located
at the focus, is reected along parallel rays. Satellite dishes also have paraboloid
shapes. Signals coming from satellites eec vely approach the dish along parallel rays. The dish then focuses these rays at the focus, where the sensor is
located.
Ellipses
Deni on 41
Ellipse
An ellipse is the locus of all points whose sum of distances from two xed
points, each a focus of the ellipse, is constant.
An easy way to visualize this construc on of an ellipse is to pin both ends of

a string to a board. The pins become the foci. Holding a pencil ght against the
string places the pencil on the ellipse; the sum of distances from the pencil to
the pins is constant: the length of the string. See Figure 9.6.
We can again nd an algebraic equa on for an ellipse using this geometric
deni on. Let the foci be located along the x-axis, c units from the origin. Let
these foci be labeled as F1 = (c, 0) and F2 = (c, 0). Let P = (x, y) be a point
on the ellipse. The sum of distances from F1 to P (d1 ) and from F2 to P (d2 ) is a
constant d. That is, d1 + d2 = d. Using the Distance Formula, we have
(x + c)2 + y2 + (x c)2 + y2 = d.
Using a fair amount of algebra can produce the following equa on of an ellipse
(note that the equa on is an implicitly dened func on; it has to be, as an ellipse
fails the Ver cal Line Test):
d2
d1
.
d1 + d2 = constant
Figure 9.6: Illustra ng the construc on of

an ellipse with pins, pencil and string.
y2
x2
= 1.
( d )2 + ( d )2
2
c
2
2
Notes:
473

This is
not par cularly illumina ng, but by making the subs tu on a = d/2 and
b = a2 c2 , we can rewrite the above equa on as
Ver ces
x2
y2
+
= 1.
a2
b2
Foci
{z
} |
{z
Major axis
Minor axis
This choice of a and b is not without reason; as shown in Figure 9.7, the values
of a and b have geometric meaning in the graph of the ellipse.
In general, the two foci of an ellipse lie on the major axis of the ellipse, and
the midpoint of the segment joining the two foci is the center. The major axis
intersects the ellipse at two points, each of which is a vertex. The line segment
through the center and perpendicular to the major axis is the minor axis. The
constant sum of distances that denes the ellipse is the length of the major
axis, i.e., 2a.
Allowing for the shi ing of the ellipse gives the following standard equa ons.
Figure 9.7: Labeling the signicant fea.

tures of an ellipse.
Key Idea 34
Standard Equa on of the Ellipse
The equa on of an ellipse centered at (h, k) with major axis of length 2a

and minor axis of length 2b in standard form is:
1. Horizontal major axis:
2. Ver cal major axis:
(y k)2
(x h)2
+
= 1.
a2
b2
(x h)2
(y k)2
+
= 1.
2
b
a2
The foci lie along the major axis, c units from the center, where c2 =
a2 b2 .
y
6
. Example 275
Finding the equa on of an ellipse
Find the general equa on of the ellipse graphed in Figure 9.8.
4
2
x
6
2
2
The center is located at (3, 1). The distance from the cenS
ter to a vertex is 5 units, hence a = 5. The minor axis seems to have length 4,
so b = 2. Thus the equa on of the ellipse is
(x + 3)2
(y 1)2
+
= 1.
4
25
Figure 9.8: The ellipse used in Example

275.
.
. Example 276
.Graphing an ellipse
Graph the ellipse dened by 4x2 + 9y2 8x 36y = 4.
Notes:
474
9.1 Conic Sec ons

y
It is simple to graph an ellipse once it is in standard form. In

order to put the given equa on in standard form, we must complete the square
with both the x and y terms. We rst rewrite the equa on by regrouping:
S
4x + 9y 8x 36y = 4
2
4
3
(4x 8x) + (9y 36y) = 4.

2
2
1
Now we complete the squares.
x
2 1
1
(4x 8x) + (9y 36y) = 4

2
4(x2 2x) + 9(y2 4y) = 4

4(x2 2x + 1 1) + 9(y2 4y + 4 4) = 4
(
)
(
)
4 (x 1)2 1 + 9 (y 2)2 4 = 4
y
1
4(x 1) + 9(y 2) = 36
We see the center of the ellipse is at (1, 2). We have a = 3 and b = 2; the major axis
so the ver ces are located at (2, 2) and (4, 2). We nd
is horizontal,
c = 9 4 = 5 2.24. The foci are located along the major axis, approximately 2.24 units from the center, at (1 2.24, 2). This is all graphed in Figure
9.9 . .
Figure 9.9: Graphing the ellipse in Example 276.
(y 2)2
(x 1)2
+
= 1.
9
4
4(x 1)2 4 + 9(y 2)2 36 = 4

2
x
1
e=0
(a)
y
1
Eccentricity
x
When a = b, we have a circle. The general equa on becomes

(x h)2
(y k)2
+
=1
a2
a2
(x h)2 + (y k)2 = a2 ,
the familiar equa on of the circle centered at (h, k) with radius a. The circle has
two foci, but they lie on the same point, the center of the circle.
Consider Figure 9.10, where several ellipses are graphed with a = 1. In (a),
we have c = 0 and the ellipse is a circle. As c grows, the resul ng ellipses look
less and less circular. A measure of this noncircularness is eccentricity.
Deni on 42
Eccentricity of an Ellipse
.
c
The eccentricity e of an ellipse is e = .
a
e = 0.3
(b)
y
1
x
1
e = 0.8
(c)
y
1
x
1
Notes:
e = 0.99
(d)
Figure 9.10: Understanding the eccentricity of an ellipse.
475

The eccentricity of a circle is 0; that is, a circle has no noncircularness. As
c approaches a, e approaches 1, giving rise to a very noncircular ellipse, as seen
in Figure 9.10 (d).
It was long assumed that planets had circular orbits. This is known to be
incorrect; the orbits are ellip cal. Earth has an eccentricity of 0.0167 it has
a nearly circular orbit. Mercurys orbit is the most eccentric, with e = 0.2056.
(Plutos eccentricity is greater, at e = 0.248, the greatest of all the currently
known dwarf planets.) The planet with the most circular orbit is Venus, with
e = 0.0068. The Earths moon has an eccentricity of e = 0.0549, also very circular.
Reec ve Property
The ellipse also possesses an interes ng reec ve property. Any ray emana ng from one focus of an ellipse reects o the ellipse along a line through
the other focus, as illustrated in Figure 9.11. This property is given formally in
the following theorem.
F1
F2
Theorem 80
Reec ve Property of an Ellipse
Let P be a point on a ellipse with foci F1 and F2 . The tangent line to the
ellipse at P makes equal angles with the following two lines:
Figure 9.11: Illustra ng the reec ve

property of an ellipse.
1. The line through F1 and P, and

2. The line through F2 and P.
This reec ve property is useful in op cs and is the basis of the phenomena

experienced in whispering halls.
Hyperbolas
The deni on of a hyperbola is very similar to the deni on of an ellipse; we
essen ally just change the word sum to dierence.
.
Deni on 43
Hyperbola
A hyperbola is the locus of all points.

where the absolute value of dierence of distances from two xed points, each a focus of the hyperbola,
is constant.
Notes:
476
The two foci lie on the transverse axis of the hyperbola; the midpoint of
the line segment joining the foci is the center of the hyperbola. The transverse
axis intersects the hyperbola at two points, each a vertex of the hyperbola. The
line through the center and perpendicular to the transverse axis is the conjugate axis. This is illustrated in Figure 9.12. It is easy to show that the constant
dierence of distances used in the deni on of the hyperbola is the distance
between the ver ces, i.e., 2a.
Key Idea 35
Conjugate
We do not have a convenient way of visualizing the construc on of a hyperbola as we did for the ellipse. The geometric deni on does allow us to nd an
algebraic expression that describes it. It will be useful to dene some terms rst.
axis
9.1 Conic Sec ons
z }| { z }| {
Transverse
axis
Ver ces
Foci
Figure 9.12: Labeling the signicant features of a hyperbola.
Standard Equa on of a Hyperbola
The equa on of a hyperbola centered at (h, k) in standard form is:

1. Horizontal Transverse Axis:
2. Ver cal Transverse Axis:
(x h)2
(y k)2
= 1.
a2
b2
(y k)2
(x h)2
= 1.
a2
b2
The ver ces are located a units from the center and the foci are located
c units from the center, where c2 = a2 + b2 .
5
2
..
Graphing Hyperbolas
x2
9
y2
1
Figure 9.13: Graphing the hyperbola
1.
Consider the hyperbola = 1. Solving for y, we nd y =
As x grows
large, the 1 part of the equa on for y becomes less signicant and
y x2 /9 = x/3. That is, as x gets large, the graph of the hyperbola looks
very much like the lines y = x/3. These lines are asymptotes of the hyperbola,
x2 /9
y
1
= 1 along with its asymptotes, y =

x/3.
This is a valuable tool in sketching. Given the equa on of a hyperbola in

general form, draw a rectangle centered at (h, k) with sides of length 2a parallel
to the transverse axis and sides of length 2b parallel to the conjugate axis. (See
Figure 9.14 for an example with a horizontal transverse axis.) The diagonals of
the rectangle lie on the asymptotes.
k+b
k
These lines pass through (h, k). When the transverse axis is horizontal, the
slopes are b/a; when the transverse axis is ver cal, their slopes are a/b. This
gives equa ons:
kb
x
..
Notes:
x2
9
ha
h+a
Figure 9.14: Using the asymptotes of a hyperbola as a graphing aid.
477

Horizontal
Transverse Axis
Ver cal
Transverse Axis
b
y = (x h) + k
a
a
y = (x h) + k.
b
. Example 277
Graphing a hyperbola
(y 2)2
(x 1)2
Sketch the hyperbola given by
= 1.
25
4
y
10
x
5
5
5
.
Figure 9.15: Graphing the hyperbola in
Example 277.
The hyperbola is centered at (1, 2); a = 5 and b = 2. In

S
Figure 9.15 we draw the prescribed rectangle centered at (1, 2) along with the
asymptotes dened by its diagonals. The hyperbola has a ver cal transverse
axis, so the ver ces are located at (1, 7) and (1, 3). This is enough to make a
good sketch.
We also nd the loca on of the foci: as c2 = a2 + b2 , we have c = 29

5.4. Thus the foci are located at (1, 2 5.4) as shown in the gure. .
. Example 278
Graphing a hyperbola
Sketch the hyperbola given by 9x2 y2 + 2y = 10.
We must complete the square to put the equa on in general
S
form. (We recognize this as a hyperbola since it is a general quadra c equa on
and the x2 and y2 terms have opposite signs.)
9x2 y2 + 2y = 10
9x2 (y2 2y) = 10
9x2 (y2 2y + 1 1) = 10
(
)
9x2 (y 1)2 1 = 10
y
10
9x2 (y 1)2 = 9
x2
x
4
10
Figure 9.16: Graphing the hyperbola in

Example 278.
We see the hyperbola is centered at (0, 1), with a horizontal transverse axis,
where a = 1 and b = 3. The appropriate rectangle is sketched in Figure 9.16
along with the
asymptotes of the hyperbola. The ver ces are located at (1, 1).
We have c = 10 3.2, so the foci are located at (3.2, 1) as shown in the
gure. .
Notes:
478
(y 1)2
=1
9
9.1 Conic Sec ons
Eccentricity
y
10
Deni on 44
Eccentricity of a Hyperbola
.c
The eccentricity of a hyperbola is e = .
a
5
x
10
5
5
Note that this is the deni on of eccentricity as used for the ellipse. When
c is close in value to a (i.e., e 1), the hyperbola is very narrow (looking almost like crossed lines). Figure 9.17 shows hyperbolas centered at the origin
with a = 1. The graph in (a) has c = 1.05, giving an eccentricity of e = 1.05,
which is close to 1. As c grows larger, the hyperbola widens and begins to look
like parallel lines, as shown in part (d) of the gure.
10
e = 1.05
10
(a)
y
10
Reec ve Property
Hyperbolas share a similar reec ve property with ellipses. However, in the

case of a hyperbola, a ray emana ng from a focus that intersects the hyperbola
reects along a line containing the other focus, but moving away from that focus. This is illustrated in Figure 9.19 (on the next page). Hyperbolic mirrors are
commonly used in telescopes because of this reec ve property. It is stated
formally in the following theorem.
x
10
5
5
10
e = 1.5
10
..
(b)
y
Theorem 81
Reec ve Property of Hyperbolas
10
Let P be a point on a hyperbola with foci F1 and F2 . The tangent line to

the hyperbola at P makes equal angles with the following two lines:
1. The line through F1 and P, and
x
10
2. The line through F2 and P.
Loca on Determina on
Determining the loca on of a known event has many prac cal uses (loca ng
the epicenter of an earthquake, an airplane crash site, the posi on of the person
speaking in a large room, etc.).
To determine the loca on of an earthquakes epicenter, seismologists use
trilatera on (not to be confused with triangula on). A seismograph allows one
5
5
10
e=3
10
(c)
y
10
5
x
10
5
5
Notes:
.
10
e = 10
10
(d)
Figure 9.17: Understanding the eccentricity of a hyperbola.
479
F1
to determine how far away the epicenter was; using three separate readings,
the loca on of the epicenter can be approximated.
A key to this method is knowing distances. What if this informa on is not
available? Consider three microphones at posi ons A, B and C which all record
a noise (a persons voice, an explosion, etc.) created at unknown loca on D.
The microphone does not know when the sound was created, only when the
sound was detected. How can the loca on be determined in such a situa on?
If each loca on has a clock set to the same me, hyperbolas can be used
to determine the loca on. Suppose the microphone at posi on A records the
sound at exactly 12:00, loca on B records the me exactly 1 second later, and
loca on C records the noise exactly 2 seconds a er that. We are interested in
the dierence of mes. Since the speed of sound is approximately 340 m/s, we
can conclude quickly that the sound was created 340 meters closer to posi on A
than posi on B. If A and B are a known distance apart (as shown in Figure 9.18
(a)), then we can determine a hyperbola on which D must lie.
The dierence of distances is 340; this is also the distance between ver ces
of the hyperbola. So we know 2a = 340. Posi ons A and B lie on the foci, so
2c = 1000. From this we can nd b 470 and can sketch the hyperbola, given
in part (b) of the gure. We only care about the side closest to A. (Why?)
We can also nd the hyperbola dened by posi ons B and C. In this case,
2a = 680 as the sound traveled an extra 2 seconds to get to C. We s ll have
2c = 1000, centering this hyperbola at (500, 500). We nd b 367. This
hyperbola is sketched in part (c) of the gure. The intersec on point of the two
graphs is the loca on of the sound, at approximately (188, 222.5).
F2
.
Figure 9.19: Illustra ng the reec ve
property of a hyperbola.
1,000
1,000
500
1,000
500
A
B
500
500
1,000
1,000
500
A
500
500
500
1,000
1,000
1,000
(b)
Figure 9.18: Using hyperbolas in loca on detec on.
Notes:
..
B
1,000
500
(a)
480
1,000
500
500
500
1,000
(c)
1,000
Exercises 9.1
Terms and Concepts
1. What is the dierence between degenerate and nondegenerate conics?
18.
2. Use your own words to explain what the eccentricity of an

ellipse measures.
3. What has the largest eccentricity: an ellipse or a hyperbola?
4. Explain why the following is true: If the coecient of the x

term in the equa on of an ellipse in standard form is smaller
than the coecient of the y2 term, then the ellipse has a
horizontal major axis.
19.
x
1
5. Explain how one can quickly look at the equa on of a hyperbola in standard form and determine whether the transverse axis is horizontal or ver cal.
In Exercises 20 23, nd the equa on of the ellipse dened

by the given informa on. Sketch the elllipse.
Problems
In Exercises 6 13, nd the equa on of the parabola dened
by the given informa on. Sketch the parabola.
20. Foci: (2, 0); ver ces: (3, 0)

21. Foci: (1, 3) and (5, 3); ver ces: (3, 3) and (7, 3)
6. Focus: (3, 2); directrix: y = 1
22. Foci: (2, 2); ver ces: (2, 7)
7. Focus: (1, 4); directrix: y = 2
23. Focus: (1, 5); vertex: (1, 4); center: (1, 1)
8. Focus: (1, 5); directrix: x = 3

9. Focus: (1/4, 0); directrix: x = 1/4
10. Focus: (1, 1); vertex: (1, 2)
11. Focus: (3, 0); vertex: (0, 0)
12. Vertex: (0, 0); directrix: y = 1/16
13. Vertex: (2, 3); directrix: x = 4
In Exercises 14 15, the equa on of a parabola and a point
on its graph are given. Find the focus and directrix of the
parabola, and verify that the given point is equidistant from
the focus and directrix.
In Exercises 24 27, write the equa on of the given ellipse in

standard form.
24. x2 2x + 2y2 8y = 7
25. 5x2 + 3y2 = 15
26. 3x2 + 2y2 12y + 6 = 0
27. x2 + y2 4x 4y + 4 = 0
(x 1)2
(y 3)2
+
= 1.
4
12
(a) Verify that the foci are located at (1, 3 2 2).
(b) The points P1 = (2, 6) and P2 = (1 + 2, 3 + 6)

(2.414, 5.449) lie on the ellipse. Verify that the sum
of distances from each point to the foci is the same.
28. Consider the ellipse given by
14. y = 14 x2 , P = (2, 1)
15. x = 18 (y 2)2 + 3, P = (11, 10)
In Exercises 16 17, sketch the ellipse dened by the given
equa on. Label the center, foci and ver ces.
16.
(x 1)
(y 2)
+
=1
3
5
17.
1 2 1
x + (y + 3)2 = 1
25
9
In Exercises 29 32, nd the equa on of the hyperbola shown

in the graph.
y
In Exercises 18 19, nd the equa on of the ellipse shown in

the graph. Give the loca on of the foci and the eccentricity
of the ellipse.
29.
x
2
1
2
.
.
In Exercises 39 42, write the equa on of the hyperbola in

standard form.
30.
x
5
39. 3x2 4y2 = 12

40. 3x2 y2 + 2y = 10
41. x2 10y2 + 40y = 30
42. (4y x)(4y + x) = 4
6
4
31.
2
x
5
.
.
6
4
32.
2
x
.
In Exercises
33 34, sketch the hyperbola dened by the
given equa on. Label the center and foci.
33.
(x 1)2
(y + 2)2
=1
16
9
(x + 1)2
=1
25
In Exercises 35 38, nd the equa on of the hyperbola dened by the given informa on. Sketch the hyperbola.
34. (y 4)2
35. Foci: (3, 0); ver ces: (2, 0)

36. Foci: (0, 3); ver ces: (0, 2)
43. Johannes Kepler discovered that the planets of our solar

system have ellip cal orbits with the Sun at one focus. The
Earths ellip cal orbit is used as a standard unit of distance;
the distance from the center of Earths ellip cal orbit to one
vertex is 1 Astronomical Unit, or A.U.
The following table gives informa on about the orbits of
three planets.
Distance from
eccentricity
center to vertex
Mercury
0.387 A.U.
0.2056
Earth
1 A.U.
0.0167
Mars
1.524 A.U.
0.0934
(a) In an ellipse, knowing c2 = a2 b2 and e = c/a
allows
us to nd b in terms of a and e. Show b =
a 1 e2 .
(b) For each planet, nd equa ons of their ellip cal orbit
x2
y2
of the form 2 + 2 = 1. (This places the center at
a
b
(0, 0), but the Sun is in a dierent loca on for each
planet.)
(c) Shi the equa ons so that the Sun lies at the origin.
Plot the three ellip cal orbits.
44. A loud sound is recorded at three sta ons that lie on a line
as shown in the gure below. Sta on A recorded the sound
1 second a er Sta on B, and Sta on C recorded the sound
3 seconds a er B. Using the speed of sound as 340m/s,
determine the loca on of the sounds origina on.
.
A 1000m B
C
2000m
9.2 Parametric Equa ons
9.2
Parametric Equa ons
We are familiar with sketching shapes, such as parabolas, by following this basic
procedure:
Choose
.
x
Use a func on
(f to nd y)
y = f(x)
Plot point
(x, y)
The rectangular equa on y = f(x) works well for some shapes like a parabola
with a ver cal axis of symmetry, but in the previous sec on we encountered several shapes that could not be sketched in this manner. (To plot an ellipse using
the above procedure, we need to plot the top and bo om separately.)
In this sec on we introduce a new sketching procedure:
Use a func on
(f to nd x)
x = f(t)
Choose
.
t
Use a func on
(g to nd y)
y = g(t)
Plot point
(x, y)
Here, x and y are found separately but then plo ed together. This leads us
to a deni on.
Deni on 45
Parametric Equa ons and Curves
(Let f)and (g be con )nuous func ons on an interval I. The set of all points
x, y = f(t), g(t) in the Cartesian plane, as t varies over I, is the graph
of the parametric equa ons x = f(t) and y = g(t), where t is the parameter. A curve is a graph along with the parametric equa ons that dene
it.
This is a formal deni on of the word curve. When a curve lies in a plane
(such as the Cartesian plane), it is o en referred to as a plane curve. Examples
will help us understand the concepts introduced in the deni on.
. Example 279
.Plo ng parametric func ons
Plot the graph of the parametric equa ons x = t2 , y = t + 1 for t in [2, 2].
Notes:
483

t
2
1
0
1
2
x
4
1
0
1
4
y
1
0
1
2
3
We plot the graphs of parametric equa ons in much the

S
same manner as we plo ed graphs of func ons like y = f(x): we make a table of
values, plot points, then connect these points with a reasonable looking curve.
Figure 9.20(a) shows such a table of values; note how we have 3 columns.
The points (x, y) from the table are plo ed in Figure 9.20(b). The points have
been connected with a smooth curve. Each point has been labeled with its corresponding t-value. These values, along with the two arrows along the curve,
are used to indicate the orienta on of the graph. This informa on helps us determine the direc on in which the graph is moving. .
(a)
y
4
t=2
t=1
We o en use the le er t as the parameter as we o en regard t as representing me. Certainly there are many contexts in which the parameter is not me,
but it can be helpful to think in terms of me as one makes sense of parametric
plots and their orienta on (for instance, At me t = 0 the posi on is (1, 2) and
at me t = 3 the posi on is (5, 1).).
t=0
x
t = 1
t = 2
. 2
. Example 280
(b)
Sketch the graph of the parametric equa ons x = cos2 t, y = cos t + 1 for t
in [0, ].
Figure 9.20: A table of values of the parametric equa ons in Example 279 along
with a sketch of their graph.
t
0
/4
/2
3/4
x
1
1/2
0
1/2
1
1+
1
y
2
We again start by making a table of values in Figure 9.21(a),

S
then plot the points (x, y) on the Cartesian plane in Figure 9.21(b).
It is not dicult to show that the curves in Examples 279 and 280 are por ons
of the same parabola. While the parabola is the same, the curves are dierent.
In Example 279, if we let t vary over all real numbers, wed obtain the en re
parabola. In this example, le ng t vary over all real numbers would s ll produce
the same graph; this por on of the parabola would be traced, and retraced,
innitely. The orienta on shown in Figure 9.21 shows the orienta on on [0, ],
but this orienta on is reversed on [, 2].
These examples begin to illustrate the powerful nature of parametric equaons. Their graphs are far more diverse than the graphs of func ons produced
by y = f(x) func ons. .
2/2
2/2
(a)
y
t=0
2
t = /4
1.5
1
0.5
t = /2
t = 3/4
t=
1
0.5
Technology Note: Most graphing u li es can graph func ons given in parametric form. O en the word parametric is abbreviated as PAR or PARAM in
the op ons. The user usually needs to determine the graphing window (i.e, the
minimum and maximum x- and y-values), along with the values of t that are to
be plo ed. The user is o en prompted to give a t minimum, a t maximum, and
a t-step or t. Graphing u li es eec vely plot parametric func ons just as
weve shown here: they plots lots of points. A smaller t-step plots more points,
making for a smoother graph (but may take longer). In Figure 9.20, the t-step is
1.5
(b)
Figure 9.21: A table of values of the parametric equa ons in Example 280 along
with a sketch of their graph.
484
Plo ng parametric func ons
Notes:

y
1; in Figure 9.21, the t-step is /4.
One nice feature of parametric equa ons is that their graphs are easy to
shi . While this is not too dicult in the y = f(x) context, the resul ng funcon can look rather messy. (Plus, to shi to the right by two, we replace x with
x 2, which is counterintui ve.) The following example demonstrates this.
x = t2 + t
y = t2 t
2
x
. Example 281
Shi ing the graph of parametric func ons
Sketch the graph of the parametric equa ons x = t2 + t, y = t2 t. Find new
parametric equa ons that shi this graph to the right 3 places and down 2.
10
10
.
(a)
y
The graph of the parametric equa ons is given in Figure 9.22

(a). It is a parabola with a axis of symmetry along the line y = x; the vertex is at
(0, 0).
In order to shi the graph to the right 3 units, we need to increase the xvalue by 3 for every point. The straigh orward way to accomplish this is simply
to add 3 to the func on dening x: x = t2 + t + 3. To shi the graph down by 2
units, we wish to decrease each y-value by 2, so we subtract 2 from the func on
dening y: y = t2 t 2. Thus our parametric equa ons for the shi ed graph
are x = t2 + t + 3, y = t2 t 2. This is graphed in Figure 9.22 (b). No ce how
the vertex is now at (3, 2). .
S
Because the x- and y-values of a graph are determined independently, the

graphs of parametric func ons o en possess features not seen on y = f(x)
type graphs. The next example demonstrates how such graphs can arrive at the
same point more than once.
6
4
x = t2 + t + 3
y = t2 t 2
2
x
2
.
(b)
Figure 9.22: Illustra ng how to shi

graphs in Example 281.
. Example 282
.Graphs that cross themselves
Plot the parametric func ons x = t3 5t2 + 3t + 11 and y = t2 2t + 3 and
determine the t-values where the graph crosses itself.
15
x = t3 5t2 + 3t + 11
y = t2 2t + 3
10
Using the methods developed in this sec on, we again plot

S
points and graph the parametric equa ons as shown in Figure 9.23. It appears
that the graph crosses itself at the point (2, 6), but well need to analy cally
determine this.
. = x(t) and
We are looking for two dierent values, say, s and t, where x(s)
y(s) = y(t). That is, the x-values are the same precisely when the y-values are
the same. This gives us a system of 2 equa ons with 2 unknowns:
.5
x
5
10
15
Figure 9.23: A graph of the parametric

equa ons from Example 282.
s3 5s2 + 3s + 11 = t3 5t2 + 3t + 11
s2 2s + 3 = t2 2t + 3
Notes:
485

Solving this system is not trivial but involves only algebra. Using the quadra c
formula,
one can solve for t in the second equa on and nd that t = 1
s2 2s + 1. This can be subs tuted into the rst equa on, revealing that the
graph crosses itself at t = 1 and t = 3. We conrm our result by compu ng
x(1) = x(3) = 2 and y(1) = y(3) = 6. .
Conver ng between rectangular and parametric equa ons

It is some mes useful to rewrite equa ons in rectangular form (i.e., y = f(x))
into parametric form, and viceversa. Conver ng from rectangular to parametric can be very simple: given y = f(x), the parametric equa ons x = t, y = f(t)
produce the same graph. As an example, given y = x2 , the parametric equa ons
x = t, y = t2 produce the familiar parabola. However, other parametriza ons
can be used. The following example demonstrates one possible alterna ve.
. Example 283
Conver ng from rectangular to parametric
Consider y = x2 . Find parametric equa ons x = f(t), y = g(t) for the parabola
where t = dy
dx . That is, t = a corresponds to the point on the graph whose
tangent line has slope a.
We start by compu ng dy
dx : y = 2x. Thus we set t = 2x. We
can solve for x and nd x = t/2. Knowing that y = x2 , we have y = t2 /4. Thus
parametric equa ons for the parabola y = x2 are
x = t/2 y = t2 /4.
To nd the point where the tangent line has a slope of 2, we set t = 2. This
gives the point (1, 1). We can verify that the slope of the line tangent to the
curve at this point indeed has a slope of 2. .
We some mes chose the parameter to accurately model physical behavior.
. Example 284
.Conver ng from rectangular to parametric
An object is red from a height of 0 and lands 6 seconds later, 192 away. Assuming ideal projec le mo on, the height, in feet, of the object can be described
by h(x) = x2 /64 + 3x, where x is the distance in feet from the ini al loca on.
(Thus h(0) = h(192) = 0 .) Find parametric equa ons x = f(t), y = g(t)
for the path of the projec le where x is the horizontal distance the object has
traveled at me t (in seconds) and y is the height at me t.
Physics tells us that the horizontal mo on of the projec le
S
is linear; that is, the horizontal speed of the projec le is constant. Since the
object travels 192 in 6s, we deduce that the object is moving horizontally at
Notes:
486

a rate of 32 /s, giving the equa on x = 32t. As y = x2 /64 + 3x, we nd
y = 16t2 + 96t. We can quickly verify that y = 32 /s2 , the accelera on
due to gravity, and that the projec le reaches its maximum at t = 3, halfway
along its path.
These parametric equa ons make certain determina ons about
( the objects
)
loca
on
easy:
2
seconds
into
the
ight
the
object
is
at
the
point
x(2), y(2) =
(
)
64, 128 . That is, it has traveled horizontally 64 and is at a height of 128 , as
shown in Figure 9.24. .
It is some mes necessary to convert given parametric equa ons into rectangular form. This can be decidedly more dicult, as some simple looking
parametric equa ons can have very complicated rectangular equa ons. This
conversion is o en referred to as elimina ng the parameter, as we are looking
for a rela onship between x and y that does not involve the parameter t.
y
150
t=2
100
x = 32t
50
y = 16t2 + 96t
x
50
100
150
200
Figure 9.24: Graphing projec le mo on in

Example 284.
. Example 285
.Elimina ng the parameter
Find a rectangular equa on for the curve described by
x=
1
t2 + 1
and
y=
t2
.
t2 + 1
There is not a set way to eliminate a parameter. One method

S
is to solve for t in one equa on and then subs tute that value in the second. We
use that technique here, then show a second, simplermethod.
Star ng with x = 1/(t2 + 1), solve for t: t = 1/x 1. Subs tute this
value for t in the equa on for y:
y
2
y=1x
t2
+1
1/x 1
=
1/x 1 + 1
1/x 1
=
1/x
)
(
1
1 x
=
x
y=
1
t2 + 1
t2
y= 2
t +1
x=
t2
x
2
Figure 9.25: Graphing parametric and

rectangular equa ons for a graph in Example 285.
= 1 x.
Thus y = 1 x. One may have recognized this earlier by manipula ng the
equa on for y:
t2
1
y= 2
=1 2
= 1 x.
t +1
t +1
Notes:
487

This is a shortcut that is very specic to this problem; some mes shortcuts exist
and are worth looking for.
We should be careful to limit the domain of the func on y = 1 x. The
parametric equa ons limit x to values in (0, 1], thus to produce the same graph
we should limit the domain of y = 1 x to the same.
The graphs of these func ons is given in Figure 9.25. The por on of the graph
dened by the parametric equa ons is given in a thick line; the graph dened
by y = 1 x with unrestricted domain is given in a thin line. .
. Example 286
Elimina ng the parameter
Eliminate the parameter in x = 4 cos t + 3, y = 2 sin t + 1
y
We should not try to solve for t in this situa on as the reS

sul ng algebra/trig would be messy. Rather, we solve for cos t and sin t in each
equa on, respec vely. This gives
x
2
cos t =
x3
4
and
sin t =
y1
.
2
.
Figure 9.26: Graphing the parametric
equa ons x = 4 cos t + 3, y = 2 sin t + 1
in Example 286.
The Pythagorean Theorem gives cos2 t + sin2 t = 1, so:
cos2 t + sin2 t = 1
(
)2 (
)2
x3
y1
+
=1
4
2
(y 1)2
(x 3)2
+
=1
16
4
This nal equa on should look familiar it is the equa on of an ellipse! Figure
9.26 plots the parametric equa ons, demonstra ng that the graph is indeed of
an ellipse with a horizontal major axis with center at (3, 1). .
The Pythagorean Theorem can also be used to iden fy parametric equa ons
for hyperbolas. We give the parametric equa ons for ellipses and hyperbolas in
the following Key Ideas.
Notes:
488

y
1
Key Idea 36
Parametric Equa ons for Ellipses
The parametric equa ons

x = a cos t + h,
.y = b sin t + k
x
1
dene an ellipse with horizontal axis of length 2a and ver cal axis of
length 2b, centered at (h, k).
1
Astroid
x = cos3 t
y = sin3 t
Key Idea 37
Parametric Equa ons for Hyperbolas
The parametric equa ons

x = a tan t + h,
y = b sec t + k
dene a hyperbola with ver cal transverse axis centered at (h, k), and
x = a sec t + h,
y = b tan t + k
denes a hyperbola with horizontal transverse axis. Each has asymptotes

at y = b/a(x h) + k.
x
1
Rose Curve
x = cos(t) sin(4t)
y = sin(t) sin(4t)
Special Curves
Figure 9.27 gives a small gallery of interes ng and famous curves along
with parametric equa ons that produce them. Interested readers can begin
learning more about these curves through internet searches.
One might note a feature shared by two of these graphs: sharp corners,
or cusps. We have seen graphs with cusps before and determined that such
func ons are not dieren able at these points. This leads us to a deni on.
x
5
Deni on 46
Smooth
.
Hypotrochoid
x = 2 cos(t) + 5 cos(2t/3)
y = 2 sin(t) 5 sin(2t/3)
A curve C dened by x = f(t), y = g(t) is smooth on an interval I if f and

g are con nuous on I and not simultaneously 0 (except possibly at the
endpoints of I). A curve is piecewise smooth on I if I can be par oned
into subintervals where C is smooth on each subinterval.
y
5
x
5
Notes:
5
.
Epicycloid
x = 4 cos(t) cos(4t)
y = 4 sin(t) sin(4t)
Figure 9.27: A gallery of interes ng planar
curves.
489

Consider the astroid, given by x = cos3 t, y = sin3 t. Taking deriva ves, we
have:
x = 3 cos2 t sin t and y = 3 sin2 t cos t.
It is clear that each is 0 when t = 0, /2, , . . .. Thus the astroid is not smooth
at these points, corresponding to the cusps seen in the gure.
We demonstrate this once more.
y
. Example 287
Determine where a curve is not smooth
Let a curve C be dened by the parametric equa ons x = t3 12t + 17 and
y = t2 4t + 8. Determine the points, if any, where it is not smooth.
8
6
We begin by taking deriva ves.

x = 3t2 12,
x
5
We set each equal to 0:
10
Figure 9.28: Graphing the curve in Example 287; note it is not smooth at (1, 4).
y = 2t 4.
x = 0 3t2 12 = 0 t = 2
y = 0 2t 4 = 0 t = 2
We see at t = 2 both x and y are 0; thus C is not smooth at t = 2, corresponding to the point (1, 4). The curve is graphed in Figure 9.28, illustra ng the cusp
at (1, 4). .
If a curve is not smooth at t = t0 , it means that x (t0 ) = y (t0 ) = 0 as
dened. This, in turn, means that rate of change of x (and y) is 0; that is, at
that instant, neither x nor y is changing. If the parametric equa ons describe
the path of some object, this means the object is at rest at t0 . An object at rest
can make a sharp change in direc on, whereas moving objects tend to change
direc on in a smooth fashion.
One should be careful to note that a sharp corner does not have to occur
when a curve is not smooth. For instance, one can verify that x = t3 , y = t6
produce the familiar y = x2 parabola. However, in this parametriza on, the
curve is not smooth. A par cle traveling along the parabola according to the
given parametric equa ons comes to rest at t = 0, though no sharp point is
created.
Notes:
490
Exercises 9.2
Terms and Concepts
In Exercises 20 29, eliminate the parameter in the given

parametric equa ons.
1. T/F: When sketching the graph of parametric equa ons, the

x and y values are found separately, then plo ed together.
2. The direc on in which a graph is moving is called the
of the graph.
3. An equa on wri en as y = f(x) is wri en in
form.
4. Create parametric equa ons x = f(t), y = g(t) and sketch

their graph. Explain any interes ng features of your graph
based on the func ons f and g.
Problems
In Exercises 5 8, sketch the graph of the given parametric
equa ons by hand, making a table of points to plot. Be sure
to indicate the orienta on of the graph.
21. x = sec t,
y = 3t + 1
y = tan t
y = 3 cos t 2
22. x = 4 sin t + 1,
2
23. x = t ,
24. x =
y=t
1
,
t+1
25. x = et ,
y=
3t + 5
t+1
y = e3t 3
26. x = ln t,
y = t2 1
27. x = cot t,
y = csc t
28. x = cosh t,
29. x = cos(2t),
y = sinh t
y = sin t
In Exercises 30 33, eliminate the parameter in the given

parametric equa ons. Describe the curve dened by the
parametric equa ons based on its rectangular form.
5. x = t2 + t, y = 1 t2 , 3 t 3
/2 t /2
6. x = 1, y = 5 sin t,
20. x = 2t + 5,
y = 2, 2 t 2
30. x = at + x0 ,
8. x = t t + 3, y = t2 + 1, 2 t 2
31. x = r cos t,
In Exercises 9 17, sketch the graph of the given parametric equa ons; using a graphing u lity is advisable. Be sure to
indicate the orienta on of the graph.
32. x = a cos t + h,
y = b sin t + k
33. x = a sec t + h,
y = b tan t + k
7. x = t ,
3
9. x = t3 2t2 , y = t2 ,
10. x = 1/t, y = sin t,
0 < t 10
0 t 2
0 t 2
12. x = 3 cos t + 2, y = 5 sin t + 3,
y = r sin t
In Exercises 34 37, nd parametric equa ons for the given

dy
. Verify that
rectangular equa on using the parameter t =
dx
at t = 1, the point on the graph has a tangent line with slope
of 1.
2 t 3
11. x = 3 cos t, y = 5 sin t,
y = bt + y0
34. y = 3x2 11x + 2
13. x = cos t, y = cos(2t), 0 t
35. y = ex
14. x = cos t, y = sin(2t), 0 t 2
36. y = sin x on [0, ]
37. y = x on [0, )
15. x = 2 sec t, y = 3 tan t, /2 < t < /2

16. x = cos t + 14 cos(8t),
y = sin t + 14 sin(8t), 0 t 2
17. x = cos t + 14 sin(8t),
y = sin t + 14 cos(8t), 0 t 2
In Exercises 18 19, four sets of parametric equa ons are

given. Describe how their graphs are similar and dierent.
Be sure to discuss orienta on and ranges.
18.
< t <
(a) x = t y = t ,
y = e2t ,
< t <
(d) x = t y = t2 ,
< t <
(c) x = et
19.
(a) x = cos t
y = sin t,
2
(b) x = cos(t )
0 t 2
2
y = sin(t ),
(c) x = cos(1/t)
(d) x = cos(cos t)
0 t 2
y = sin(1/t),
y = sin(cos t),
38. x = t3 t + 3, y = t2 3
39. x = t3 4t2 + t + 7, y = t2 t
40. x = cos t, y = sin(2t) on [0, 2]
41. x = cos t cos(3t), y = sin t cos(3t) on [0, ]
< t <
(b) x = sin t y = sin2 t,
In Exercises 38 41, nd the values of t where the graph of

the parametric equa ons crosses itself.
0<t<1
0 t 2
In Exercises 42 45, nd the value(s) of t where the curve

dened by the parametric equa ons is not smooth.
42. x = t3 + t2 t,
43. x = t2 4t,
y = t2 + 2t + 3
y = t3 2t2 4t
44. x = cos t, y = 2 cos t

45. x = 2 cos t cos(2t),
y = 2 sin t sin(2t)
In Exercises 46 54, nd parametric equa ons that describe

the given situa on.
50. A circle of radius 3, centered at (1, 1), that is traced once

counterclockwise on [0, 1].
46. A projec le is red from a height of 0 , landing 16 away

in 4s.
51. An ellipse centered at (1, 3) with ver cal major axis of

length 6 and minor axis of length 2.

in 4s.
52. An ellipse with foci at (1, 0) and ver ces at (5, 0).

in 20s.
53. A hyperbola with foci at (5, 3) and (1, 3), and with
ver ces at (1, 3) and (3, 3).
49. A circle of radius 2, centered at the origin, that is traced

clockwise once on [0, 2].
54. A hyperbola with ver ces at (0, 6) and asymptotes y =

3x.
9.3
Calculus and Parametric Equa ons
The previous sec on dened curves based on parametric equa ons. In this secon well employ the techniques of calculus to study these curves.
We are s ll interested in lines tangent to points on a curve. They describe
how the y-values are changing with respect to the x-values, they are useful in
making approxima ons, and they indicate instantaneous direc on of travel.
The slope of the tangent line is s ll dy
dx , and the Chain Rule allows us to calculate this in the context of parametric equa ons. If x = f(t) and y = g(t), the
Chain Rule states that
dy
dy dx
=
.
dt
dx dt
Solving for
dy
dx ,
we get
dy / dx
g (t)
dy
=
= ,
dx
dt dt
f (t)
provided that f (t) = 0. This is important so we label it a Key Idea.

Key Idea 38
Finding
dy
dx
with Parametric Equa ons.
Let x = f(t) and y = g(t), where f and g are dieren able on some open
interval I and f (t) = 0 on I. Then
dy
g (t)
= .
dx
f (t)
We use this to dene the tangent line.

Deni on 47
Tangent and Normal Lines
Let a curve C be parametrized by x = f(t) and y = g(t), where f and g

are dieren able func ons on some interval I (containing t) = t0 . The
tangent line to C at t = t0 is the line through f(t0 ), g(t0 ) with slope
m = g (t0 )/f (t0 ), provided f (t0 ) = 0.
(
)
The normal line to C at t = t0 is the line through f(t0 ), g(t0 ) with slope
m = f (t0 )/g (t0 ), provided g (t0 ) = 0.
The deni on leaves two special cases to consider. When the tangent line is
horizontal, the normal line is undened by the above deni on as g (t0 ) = 0.
Notes:
493

Likewise, when the normal line is horizontal, the tangent line is undened. It
seems reasonable that these lines be dened (one can draw a line tangent to
the right side of a circle, for instance), so we add the following to the above
deni on.
1. If the tangent line at t = t0 has a slope of 0, the normal line to C at t = t0
is the line x = f(t0 ).
2. If the normal line at t = t0 has a slope of 0, the tangent line to C at t = t0
is the line x = f(t0 ).
. Example 288
.Tangent and Normal Lines to Curves
Let x = 5t2 6t + 4 and y = t2 + 6t 1, and let C be the curve dened by these
equa ons.
1. Find the equa ons of the tangent and normal lines to C at t = 3.
2. Find where C has ver cal and horizontal tangent lines.
S
1. We start by compu ng f (t) = 10t 6 and g (t) = 2t + 6. Thus
dy
2t + 6
=
.
dx
10t 6
40
Make note of something that might seem unusual: dy

dx is a func on of t,
not x. Just as points on the curve are found in terms of t, so are the slopes
of the tangent lines.
20
The point on C at t = 3 is (31, 26). The slope of the tangent line is m = 1/2
and the slope of the normal line is m = 2. Thus,
x
20
40
60
80
20.
1
(x 31) + 26, and
2
the equa on of the normal line is y = 2(x 31) + 26.
the equa on of the tangent line is y =
Figure 9.29: Graphing tangent and normal lines in Example 288.
This is illustrated in Figure 9.29.

2. To nd where C has a horizontal tangent line, we set dy
dx = 0 and solve
for t. In this case, this amounts to se ng g (t) = 0 and solving for t (and
making sure that f (t) = 0).
g (t) = 0
2t + 6 = 0
t = 3.
The point on C corresponding to t = 3 is (67, 10); the tangent line at

that point is horizontal (hence with equa on y = 10).
Notes:
494

To nd where C has a ver cal tangent line, we nd where it has a horizontal
normal line, and set gf (t)

(t) = 0. This amounts to se ng f (t) = 0 and
solving for t (and making sure that g (t) = 0).
f (t) = 0
10t 6 = 0
t = 0.6.
The point on C corresponding to t = 0.6 is (2.2, 2.96). The tangent line at

that point is x = 2.2.
.
The points where the tangent lines are ver cal and horizontal are indicated on the graph in Figure 9.29.
. Example 289
Tangent and Normal Lines to a Circle
1. Find where the circle, dened by x = cos t and y = sin t on [0, 2], has
ver cal and horizontal tangent lines.
2. Find the equa on of the normal line at t = t0 .
S
1. We compute the deriva ve following Key Idea 38:

g (t)
cos t
dy
=
=
.
dx
f (t)
sin t
The deriva ve is 0 when cos t = 0; that is, when t = /2, 3/2. These
are the points (0, 1) and (0, 1) on the circle.
The normal line is horizontal (and hence, the tangent line is ver cal) when
sin t = 0; that is, when t = 0, , 2, corresponding to the points (1, 0)
and (0, 1) on the circle. These results should make intui ve sense.
y
1
sin t0
= tan t0 . This normal
cos t0
line goes through the point (cos t0 , sin t0 ), giving the line
2. The slope of the normal line at t = t0 is m =
sin t0
(x cos t0 ) + sin t0
cos t0
= (tan t0 )x,
x
1
y=
as long as cos t0 = 0. It is an important fact to recognize that the normal lines to a circle pass through its center, as illustrated in Figure 9.30.
Stated in another way, any line that passes through the center of a circle
intersects the circle at right angles.
Figure 9.30: Illustra ng how a circles normal lines pass through its center.
Notes:
495

. Example 290
Tangent lines when dy

dx is not dened
Find the equa on of the tangent line to the astroid x = cos3 t, y = sin3 t at
t = 0, shown in Figure 9.31.
y
1
We start by nding x (t) and y (t):
x (t) = 3 sin t cos2 t,
x
1
Figure 9.31: A graph of an astroid.
y (t) = 3 cos t sin2 t.
Note that both of these are 0 at t = 0; the curve is not smooth at t = 0 forming
a cusp on the graph. Evalua ng dy
dx at this point returns the indeterminate form
of 0/0.
We can, however, examine the slopes of tangent lines near t = 0, and take
the limit as t 0.
y (t)
3 cos t sin2 t
= lim
t0 x (t)
t0 3 sin t cos2 t
sin t
= lim
t0
cos t
= 0.
lim
(We can cancel as t = 0.)
We have accomplished something signicant. When the deriva ve dy

dx returns an
dy
indeterminate form at t = t0 , we can dene its value by se ng it to be lim
,
tt0 dx
if that limit exists. This allows us to nd slopes of tangent lines at cusps, which
can be very benecial.
We found the slope of the tangent line at t = 0 to be 0; therefore the tangent line is y = 0, the x-axis. .
Concavity
We con nue to analyze curves in the plane by considering their concavity;
2
that is, we are interested in ddxy2 , the second deriva ve of y with respect to x.
To nd this, we need to nd the deriva ve of dy
dx with respect to x; that is,
[
]
d2 y
d dy
=
,
dx2
dx dx
but recall that dy
dx is a func on of t, not x, making this computa on not straightforward.
To make the upcoming nota on a bit simpler, let h(t) = dy
dx . We want
d
dh
[h(t)];
that
is,
we
want
.
We
again
appeal
to
the
Chain
Rule.
Note:
dx
dx
/
dh
dh dx
dh
dh
dx
=
=
.
dt
dx dt
dx
dt
dt
Notes:
496

2
In words, to nd ddxy2 , we rst take the deriva ve of

divide by x (t). We restate this as a Key Idea.
Key Idea 39
Finding
d2 y
dx2
dy
dx
with respect to t, then
with Parametric Equa ons
Let x = f(t) and y = g(t) be twice dieren able func ons on an open
interval I. Then
[ ]/
[ ]/
d2 y
d dy
dx
d dy
=
=
f (t).
dx2
dt dx
dt
dt dx
Examples will help us understand this Key Idea.

. Example 291
.Concavity of Plane Curves
Let x = 5t2 6t + 4 and y = t2 + 6t 1 as in Example 288. Determine the
t-intervals on which the graph is concave up/down.
Concavity is determined by the second deriva ve of y with
so we compute that here following Key Idea 39.
respect to x,
dy
2t + 6
In Example 288, we found
=
and f (t) = 10t 6. So:
dx
10t 6
S
d2 y
dx2 ,
20
]/
[
d2 y
d 2t + 6
=
(10t 6)
dx2
dt 10t 6
/
18
=
(10t 6)
(5t 3)2
=
t>
3/
d
ave
onc
c
;
5
n
ow
x
20
20.
9
(5t 3)3
The graph of the parametric func ons is concave up when
40
40
60
80
t < 3/5; concave up
Figure 9.32: Graphing the parametric

equa ons in Example 291 to demonstrate
concavity.
d2 y
dx2
> 0 and con-
cave down when ddxy2 < 0. We determine the intervals when the second derivave is greater/less than 0 by rst nding when it is 0 or undened.
2
9
As the numerator of
is never 0, ddxy2 = 0 for all t. It is undened
3
(5t 3)
when 5t 3 = 0; that is, when t = 3/5. Following the work established in
Sec on 3.4, we look at values of t greater/less than 3/5 on a number line:
Notes:
497

d2 y
>0
dx2
c. up
d2 y
<0
dx2
c. down
.
3/5
Reviewing Example 288, we see that when t = 3/5 = 0.6, the graph of the
parametric equa ons has a ver cal tangent line. This point is also a point of inec on for the graph, illustrated in Figure 9.32. .
. Example 292
Concavity of Plane Curves
Find the points of inec on of the graph of the parametric equa ons x = t,
y = sin t, for 0 t 16.
y
50
We need to compute
S
15
10
y = 2 cos t 4t sin t
50
and
d2 y
dx2 .
dy
y (t)
cos t
= 2 t cos t.
=
=
dx
x (t)
1/(2 t)
[ dy ]
d
d2 y
cos t/ t 2 t sin t
dt dx
=
=
= 2 cos t 4t sin t.
dx2
x (t)
1/(2 t)
t
5
dy
dx
Figure 9.33: A graph of ddx2y , showing

where it is approximately 0.
y
1
The points of inec on are found by se ng ddxy2 = 0. This is not trivial, as equaons that mix polynomials and trigonometric func ons generally do not have
nice solu ons.
In Figure 9.33 we see a plot of the second deriva ve. It shows that it has zeros
at approximately t = 0.5, 3.5, 6.5, 9.5, 12.5 and 16. These approxima ons are
not very good, made only by looking at the graph. Newtons Method provides
more accurate approxima ons. Accurate to 2 decimal places, we have:
t = 0.65, 3.29, 6.36, 9.48, 12.61 and 15.74.
0.5
x
1
0.5
1
.

equa ons in Example 292 along with the
points of inec on.
The corresponding points have been plo ed on the graph of the parametric
equa ons in Figure 9.34. Note how most occur near the x-axis, but not exactly
on the axis. .
Arc Length
We con nue our study of the features of the graphs of parametric equa ons
by compu ng their arc length.
Recall in Sec on 7.4 we found the arc length of the graph of a func on, from
x = a to x = b, to be
( )2
b
dy
L=
1+
dx.
dx
a
Notes:
498

We can use this equa on and convert it to the parametric equa on context.
Le ng x = f(t) and y = g(t), we know that dy

dx = g (t)/f (t). It will also be
useful to calculate the dieren al of x:
dx = f (t)dt
dt =
1
dx.
f (t)
Star ng with the arc length formula above, consider:
( )2
b
dy
L=
1+
dx
dx
a
b
g (t)2
=
1 + 2 dx.
f (t)
a
Factor out the f (t)2 :
f (t)2 + g (t)2
dx
| {z }
f (t)
=dt
t2
f (t)2 + g (t)2 dt.
t1
Note the new bounds (no longer x bounds, but t bounds). They are found
by nding t1 and t2 such that a = f(t1 ) and b = f(t2 ). This formula is important,
so we restate it as a theorem.
Theorem 82
Arc Length of Parametric Curves
Let x = f(t) and y = g(t) be parametric equa ons with f and g connuous on some open interval I containing t1 and t2 on which the graph
traces itself only once. The arc length of the graph, from t = t1 to t = t2 ,
is
t2
L=
f (t)2 + g (t)2 dt.
t1
As before, these integrals are o en not easy to compute. We start with a

simple example, then give another where we approximate the solu on.
Notes:
499

. Example 293
Arc Length of a Circle
Find the arc length of the circle parametrized by x = 3 cos t, y = 3 sin t on
[0, 3/2].
S
By direct applica on of Theorem 82, we have
3/2
L=
(3 sin t)2 + (3 cos t)2 dt.
Apply the Pythagorean Theorem.
3/2
3 dt
=
0
3/2

= 9/2.
= 3t
0
This should make sense; we know from geometry that the circumference of
a circle with radius 3 is 6; since we are nding the arc length of 3/4 of a circle,
the arc length is 3/4 6 = 9/2. .
. Example 294
Arc Length of a Parametric Curve
The graph of the parametric equa ons x = t(t2 1), y = t2 1 crosses itself as
shown in Figure 9.35, forming a teardrop. Find the arc length of the teardrop.
y
1
We can see by the parametriza ons of x and y that when

S
t = 1, x = 0 and y = 0. This means well integrate from t = 1 to t = 1.
Applying Theorem 82, we have
1
L=
(3t2 1)2 + (2t)2 dt
1
1
=
.

equa ons in Example 294, where the arc
length of the teardrop is calculated.
Unfortunately, the integrand does not have an an deriva ve expressible by elementary func ons. We turn to numerical integra on to approximate its value.
Using 4 subintervals, Simpsons Rule approximates the value of the integral as
2.65051. Using a computer, more subintervals are easy to employ, and n = 20
gives a value of 2.71559. Increasing n shows that this value is stable and a good
approxima on of the actual value. .
Notes:
500
9t4 2t2 + 1 dt.

Related to the formula for nding arc length is the formula for nding surface
area. We can adapt the formula found in Key Idea 28 from Sec on 7.4 in a similar
way as done to produce the formula for arc length done before.
Key Idea 40
Consider the graph of the parametric equa ons x = f(t) and y = g(t), where f and g are
con nuous on an open interval I containing t1 and t2 on which the graph does not cross itself.
1. The surface area of the solid formed by revolving the graph about the x-axis is (where
g(t) 0 on [t1 , t2 ]):
Surface Area = 2
.
t2
g(t) f (t)2 + g (t)2 dt.
t1
2. The surface area of the solid formed by revolving the graph about the y-axis is (where
f(t) 0 on [t1 , t2 ]):
t2
Surface Area = 2
f(t) f (t)2 + g (t)2 dt.
t1
. Example 295
Consider the teardrop shape formed by the parametric equa ons x = t(t2 1),
y = t2 1 as seen in Example 294. Find the surface area if this shape is rotated
about the x-axis, as shown in Figure 9.36.
The teardrop shape is formed between t = 1 and t = 1.

S
Using Key Idea 40, we see we need for g(t) 0 on [1, 1], and this is not the
case. To x this, we simplify replace g(t) with g(t), which ips the whole graph
about the x-axis (and does not change the surface area of the resul ng solid).
The surface area is:
1
Area S = 2
(1 t2 ) (3t2 1)2 + (2t)2 dt
= 2
1
1
Figure 9.36: Rota ng a teardrop shape

about the x-axis in Example 295.
(1 t2 ) 9t4 2t2 + 1 dt.
Once again we arrive at an integral that we cannot compute in terms of elementary func ons. Using Simpsons Rule with n = 20, we nd the area to be
S = 9.44. Using larger values of n shows this is accurate to 2 places a er the
decimal. .
Notes:
501
Exercises 9.3
Terms and Concepts
1. T/F: Given parametric equa ons x = f(t) and y = g(t),
dy
= f (t)/g (t), as long as g (t) = 0.
dx
In Exercises 25 32, parametric equa ons for a curve are

2
given. Find ddx2y , then determine the intervals on which the
graph of the curve is concave up/down. Note: these are the
same equa ons as in Exercises 5 12.
2. Given parametric equa ons x = f(t) and y = g(t),

the deriva ve dy
as given in Key Idea 38 is a func on of
dx
?
25. x = t, y = t2
26. x = t, y = 5t + 2
3. T/F: Given parametric equa ons (x =)f(t) and y = g(t), to

2
nd ddx2y , one simply computes dtd dy
.
dx
28. x = t2 1, y = t3 t
27. x = t2 t, y = t2 + t
29. x = sec t,
y = tan t on (/2, /2)
4. T/F: If dy
= 0 at t = t0 , then the normal line to the curve at
dx
t = t0 is a ver cal line.
30. x = cos t, y = sin(2t) on [0, 2]
Problems
32. x = et/10 cos t,
In Exercises 5 12, parametric equa ons for a curve are given.

dy
(a) Find .
dx
(b) Find the equa ons of the tangent and normal line(s)
at the point(s) given.
(c) Sketch the graph of the parametric func ons along
with the found tangent and normal lines.
2
5. x = t, y = t ; t = 1
6. x = t, y = 5t + 2; t = 4
8. x = t2 1, y = t3 t; t = 0 and t = 1
9. x = sec t, y = tan t on (/2, /2); t = /4
10. x = cos t, y = sin(2t) on [0, 2]; t = /4
11. x = cos t sin(2t), y = sin t sin(2t) on [0, 2];
12. x = e
t/10
cos t, y = e
sin t;
t = 3/4
t = /2
In Exercises 13 20, nd t-values where the curve dened by

the given parametric equa ons has a horizontal tangent line.
Note: these are the same equa ons as in Exercises 5 12.
13. x = t, y = t2
14. x = t, y = 5t + 2
15. x = t2 t, y = t2 + t
16. x = t2 1, y = t3 t
17. x = sec t, y = tan t on (/2, /2)
18. x = cos t, y = sin(2t) on [0, 2]
19. x = cos t sin(2t), y = sin t sin(2t) on [0, 2]
20. x = et/10 cos t, y = et/10 sin t
In Exercises 21 24, nd t = t0 where the graph of the given
dy
parametric equa ons is not smooth, then nd lim
.
tt0 dx
1
21. x = 2
, y = t3
t +1
22. x = t3 + 7t2 16t + 13, y = t3 5t2 + 8t 2
23. x = t3 3t2 + 3t 1, y = t2 2t + 1
24. x = cos2 t,
y = 1 sin2 t
y = et/10 sin t
In Exercises 33 36, nd the arc length of the graph of the

parametric equa ons on the given interval(s).
33. x = 3 sin(2t), y = 3 cos(2t) on [0, ]
34. x = et/10 cos t,
y = et/10 sin t on [0, 2] and [2, 4]
35. x = 5t + 2, y = 1 3t on [1, 1]
36. x = 2t3/2 , y = 3t on [0, 1]
In Exercises 37 40, numerically approximate the given arc
length.
37. Approximate the arc length of one petal of the rose curve
x = cos t cos(2t), y = sin t cos(2t) using Simpsons Rule
and n = 4.
7. x = t2 t, y = t2 + t; t = 1
t/10
31. x = cos t sin(2t), y = sin t sin(2t) on [/2, /2]
38. Approximate the arc length of the bow e curve x =

cos t, y = sin(2t) using Simpsons Rule and n = 6.
39. Approximate the arc length of the parabola x = t2 t,
y = t2 + t on [1, 1] using Simpsons Rule and n = 4.
40. A common approximate of the circumference of
an ellipse
a2 + b2
given by x = a cos t, y = b sin t is C 2
.
2
Use this formula to approximate the circumference of x =
5 cos t, y = 3 sin t and compare this to the approximaon given by Simpsons Rule and n = 6.
In Exercises 41 44, a solid of revolu on is described. Find or
approximate its surface area as specied.
41. Find the surface area of the sphere formed by rota ng the
circle x = 2 cos t, y = 2 sin t about:
(a) the x-axis and
(b) the y-axis.
42. Find the surface area of the torus (or donut) formed by
rota ng the circle x = cos t + 2, y = sin t about the yaxis.
43. Approximate the surface area of the solid formed by rotating the upper right half of the bow e curve x = cos t,
y = sin(2t) on [0, /2] about the x-axis, using Simpsons
Rule and n = 4.
44. Approximate the surface area of the solid formed by rota ng the one petal of the rose curve x = cos t cos(2t),
y = sin t cos(2t) on [0, /4] about the x-axis, using Simpsons Rule and n = 4.
9.4
Introduc on to Polar Coordinates
We are generally introduced to the idea of graphing curves by rela ng x-values

to y-values through a func on f. That is, we set y = f(x), and plot lots of point
pairs (x, y) to get a good no on of how the curve looks. This method is useful
but has limita ons, not least of which is that curves that fail the ver cal line
test cannot be graphed without using mul ple func ons.
The previous two sec ons introduced and studied a new way of plo ng
points in the x, y-plane. Using parametric equa ons, x and y values are computed independently and then plo ed together. This method allows us to graph
an extraordinary range of curves. This sec on introduces yet another way to plot
points in the plane: using polar coordinates.
P = P(r, )
Polar Coordinates
Start with a point O in the plane called the pole (we will always iden fy this
point with the origin). From the pole, draw a ray, called the ini al ray (we will
always draw this ray horizontally, iden fying it with the posi ve x-axis). A point
P in the plane is determined by the distance r that P is from O, and the angle formed between the ini al ray and the segment OP (measured counterclockwise). We record the distance and angle as an ordered pair (r, ). To avoid
confusion with rectangular coordinates, we will denote polar coordinates with
the le er P, as in P(r, ). This is illustrated in Figure 9.37
Prac ce will make this process more clear.
.
O
ini al ray
Figure 9.37:
nates.
Illustra ng polar coordi-
. Example 296
Plo ng Polar Coordinates
Plot the following polar coordinates:
A = P(1, /4) B = P(1.5, ) C = P(2, /3)
D = P(1, /4)
A
To aid in the drawing, a polar grid is provided at the bo om

of this page. To place the point A, go out 1 unit along the ini al ray (pu ng
you on the inner circle shown on the grid), then rotate counter-clockwise /4
radians (or 45 ). Alternately, one can consider the rota on rst: think about the
ray from O that forms an angle of /4 with the ini al ray, then move out 1 unit
along this ray (again placing you on the inner circle of the grid).
To plot B, go out 1.5 units along the ini al ray and rotate radians (180 ).
To plot C, go out 2 units along the ini al ray then rotate clockwise /3 radians, as the angle given is nega ve.
To plot D, move along the ini al ray 1 units in other words, back up 1
unit, then rotate counter-clockwise by /4. The results are given in Figure 9.38. .
S
.
O
D
C
Figure 9.38: Plo ng polar points in Example 296.
Notes:
.
O
503
Consider the following two points: A = P(1, ) and B = P(1, 0). To locate
A, go out 1 unit on the ini al ray then rotate radians; to locate B, go out 1
units on the ini al ray and dont rotate. One should see that A and B are located
at the same point in the plane. We can also consider C = P(1, 3), or D =
P(1, ); all four of these points share the same loca on.
This ability to iden fy a point in the plane with mul ple polar coordinates is
both a blessing and a curse. We will see that it is benecial as we can plot
beau ful func ons that intersect themselves (much like we saw with parametric
func ons). The unfortunate part of this is that it can be dicult to determine
when this happens. Well explore this more later in this sec on.
Polar to Rectangular Conversion

It is useful to recognize both the rectangular (or, Cartesian) coordinates of a
point in the plane and its polar coordinates. Figure 9.39 shows a point P in the
plane with rectangular coordinates (x, y) and polar coordinates P(r, ). Using
trigonometry, we can make the iden es given in the following Key Idea.
.
Key Idea 41
nates
Given the polar point P(r, ), the rectangular coordinates are determined
by
x = r cos
y = r sin .
P
r
.
O
Conver ng Between Rectangular and Polar Coordi-
Given the rectangular coordinates (x, y), the polar coordinates are determined by
y
r2 = x2 + y2
tan = .
x
Figure 9.39: Conver ng between rectangular and polar coordinates.
. Example 297
.Conver ng Between Polar and Rectangular Coordinates
1. Convert the polar coordinates P(2, 2/3) and P(1, 5/4) to rectangular
coordinates.
2. Convert the rectangular coordinates (1, 2) and (1, 1) to polar coordinates.
Notes:
504

1.
(a) We start with P(2, 2/3). Using Key Idea 41, we have
y = 2 sin(2/3) = 3.
So the rectangular coordinates are (1, 3) (1, 1.732).

(b) The polar point P(1, 5/4) is converted to rectangular with:
x = 1 cos(5/4) = 2/2
y = 1 sin(5/4) = 2/2.
So the rectangular coordinates are ( 2/2, 2/2) (0.707, 0.707).

x = 2 cos(2/3) = 1
P(2,
2
3 )
P(1,
These points are plo ed in Figure 9.40 (a). The rectangular coordinate
system is drawn lightly under the polar coordinate system so that the rela onship between the two can be seen.
2.
.
O
(a) To convert the rectangular point (1, 2) to polar coordinates, we use

the Key Idea to form the following two equa ons:
12 + 22 = r2
The rst equa on tells us that r =
func on, we nd
tan =
2
.
1
(1, 2)
(1)2 + 12 = r2
tan =
1
.
1
Thus r = 2. We need to be careful in compu ng : using the

inverse tangent func on, we have
tan = 1
(a)
5. Using the inverse tangent
= tan1 2 1.11 63.43 .
Thus polar coordinates of (1, 2) are P( 5, 1.11).

(b) To convert (1, 1) to polar coordinates, we form the equa ons
tan = 2
(1, 1)
1.11
3
4
(0, 0)
= tan1 (1) = /4 = 45 .
This is not the angle we desire. The range of tan1 x is (/2, /2);
that is, it returns angles that lie in the 1st and 4th quadrants. To
nd loca ons in the 2nd and 3rd quadrants, add to the result of
tan1 x. So
+ (/4) puts the angle at 3/4. Thus the polar
point is P( 2, 3/4).
An alternate method is to use the angle given by arctangent, but
change
the sign of r. Thus we could also refer to (1, 1) as
P( 2, /4).
5
4 )
(b)
Figure 9.40: Plo ng rectangular and polar points in Example 297.
These points are plo ed in Figure 9.40 (b). The polar system is drawn
lightly under the rectangular grid with rays to demonstrate the angles
used.
Notes:
505
Polar Func ons and Polar Graphs

Dening a new coordinate system allows us to create a new kind of funcon, a polar func on. Rectangular coordinates lent themselves well to crea ng
func ons that related x and y, such as y = x2 . Polar coordinates allow us to create func ons that relate r and . Normally these func ons look like r = f(),
although we can create func ons of the form = f(r). The following examples
introduce us to this concept.
. Example 298
Introduc on to Graphing Polar Func ons
Describe the graphs of the following polar func ons.
1. r = 1.5
2. = /4
S
1. The equa on r = 1.5 describes all points that are 1.5 units from the pole;
as the angle is not specied, any is allowable. All points 1.5 units from
the pole describes a circle of radius 1.5.
=
We can consider the rectangular equivalent of this equa on; using r2 =

x2 + y2 , we see that 1.52 = x2 + y2 , which we recognize as the equa on of
a circle centered at (0, 0) with radius 1.5. This is sketched in Figure 9.41.
r = 1.5
.
O
2. The equa on = /4 describes all points such that the line through them
and the pole make an angle of /4 with the ini al ray. As the radius r is not
specied, it can be any value (even nega ve). Thus = /4 describes the
line through the pole that makes an angle of /4 = 45 with the ini al
ray.
We can again consider the rectangular equivalent of this equa on. Combine tan = y/x and = /4:
Figure 9.41: Plo ng standard polar plots.
tan /4 = y/x
.
x tan /4 = y
y = x.
This graph is also plo ed in Figure 9.41.
The basic rectangular equa ons of the form x = h and y = k create ver cal
and horizontal lines, respec vely; the basic polar equa ons r = h and =
create circles and lines through the pole, respec vely. With this as a founda on,
we can create more complicated polar func ons of the form r = f(). The input
is an angle; the output is a length, how far in the direc on of the angle to go out.
Notes:
506

We sketch these func ons much like we sketch rectangular and parametric
func ons: we plot lots of points and connect the dots with curves. We demonstrate this in the following example.
. Example 299
Sketching Polar Func ons
Sketch the polar func on r = 1 + cos on [0, 2] by plo ng points.
A common ques on when sketching curves by plo ng points
is Which points should I plot? With rectangular equa ons, we o en choose
easy values integers, then added more if needed. When plo ng polar equaons, start with the common angles mul ples of /6 and /4. Figure 9.42
gives a table of just a few values of in [0, ].
Consider the point P(0, 2) determined by the rst line of the table. The angle
is 0 radians we do not rotate from the ini al ray then
we go out 2 units from
the pole. When = /6, r = 1.866 (actually, it is 1 + 3/2); so rotate by /6
radians and go out 1.866 units.
The graph shown uses more points, connected with straight lines. (The points
on the graph that correspond to points in the table are signied with larger dots.)
Such a sketch is likely good enough to give one an idea of what the graph looks
like. .
S
Technology Note: Plo ng func ons in this way can be tedious, just as it was
with rectangular func ons. To obtain very accurate graphs, technology is a great
aid. Most graphing calculators can plot polar func ons; in the menu, set the
plo ng mode to something like polar or POL, depending on ones calculator.
As with plo ng parametric func ons, the viewing window no longer determines the x-values that are plo ed, so addi onal informa on needs to be provided. O en with the window se ngs are the se ngs for the beginning and
ending values (o en called min and max ) as well as the step that is, how far
apart the values are spaced. The smaller the step value, the more accurate
the graph (which also increases plo ng me). Using technology, we graphed
the polar func on r = 1 + cos from Example 299 in Figure 9.43.
. Example 300
.Sketching Polar Func ons
Sketch the polar func on r = cos(2) on [0, 2] by plo ng points.
We start by making a table of cos(2) evaluated at common
angles , as shown in Figure 9.44. These points are then plo ed in Figure 9.45
(a). This par cular graph moves around quite a bit and one can easily forget
which points should be connected to each other. To help us with this, we numbered each point in the table and on the graph.
S
0
/6
/2
4/3
7/4
r = 1 + cos
2
1.86603
1
0.5
1.70711
.
O
Figure 9.42: Graphing a polar func on in

Example 299 by plo ng points.
.
O
Figure 9.43: Using technology to graph a

polar func on.
Notes:
507

13
14
Pt.
1
2
3
4
5
6
7
8
9
12
3
7 . 11
15
10
1
17
16
6
(a)
Pt.
10
11
12
13
14
15
16
17
cos(2)
0
1.
/6
0.5
/4
0.
/3
0.5
/2
1.
2/3
0.5
3/4
0.
5/6
0.5
1.
cos(2)
7/6
0.5
5/4
0.
4/3
0.5
3/2
1.
5/3
0.5
7/4
0.
11/6
0.5
2
1.
Figure 9.44: Tables of points for plo ng a polar curve.
Using more points (and the aid of technology) a smoother plot can be made as
shown in Figure 9.45 (b). This plot is an example of a rose curve. .
.
O
(b)
Figure 9.45: Polar plots from Example
300.
It is some mes desirable to refer to a graph via a polar equa on, and other
mes by a rectangular equa on. Therefore it is necessary to be able to convert
between polar and rectangular func ons, which we prac ce in the following example. We will make frequent use of the iden es found in Key Idea 41.
. Example 301
.Conver ng between rectangular and polar equa ons.
Convert from rectangular to polar.

1. y = x2
2. xy = 1
Convert from polar to rectangular.

3. r =
2
sin cos
4. r = 2 cos
1. Replace y with r sin and replace x with r cos , giving:

y = x2
r sin = r2 cos2
sin
=r
cos2
We have found that r = sin / cos2 = tan sec . The domain of this
polar func on is [/2, /2]; plot a few points to see how the familiar
parabola is traced out by the polar equa on.
Notes:
508

2. We again replace x and y using the standard iden
for r:
es and work to solve
y
5
xy = 1
r cos r sin = 1
1
cos sin
1
r=
cos sin
r2 =
x
5
.
This func on is valid only when the product of cos sin is posi ve. This
occurs in the rst and third quadrants, meaning the domain of this polar
func on is (0, /2) (, 3/2).
We can rewrite the original rectangular equa on xy = 1 as y = 1/x.
This is graphed in Figure 9.46; note how it only exists in the rst and third
quadrants.
Figure 9.46: Graphing xy = 1 from Example 301.
3. There is no set way to convert from polar to rectangular; in general, we

look to form the products r cos and r sin , and then replace these with
x and y, respec vely. We start in this problem by mul plying both sides
by sin cos :
2
sin cos
r(sin cos ) = 2
r sin r cos = 2.
Now replace with y and x:
r=
yx=2
y = x + 2.
The original polar equa on, r = 2/(sin cos ) does not easily reveal
that its graph is simply a line. However, our conversion shows that it is.
The upcoming gallery of polar curves gives the general equa ons of lines
in polar form.
.
4. By mul plying both sides by r, we obtain both an r2 term and an r cos
term, which we replace with x2 + y2 and x, respec vely.
r = 2 cos
2
r = 2r cos
2
x + y2 = 2x.
Notes:
509

We recognize this as a circle; by comple ng the square we can nd its
radius and center.
x2 2x + y2 = 0
(x 1)2 + y2 = 1.
The circle is centered at (1, 0) and has radius 1. The upcoming gallery
of polar curves gives the equa ons of some circles in polar form; circles
with arbitrary centers have a complicated polar equa on that we do not
consider here.
Some curves have very simple polar equa ons but rather complicated rectangular ones. For instance, the equa on r = 1 + cos describes a cardiod
(a shape important the sensi vity of microphones, among other things; one is
graphed in the gallery in the Limaon sec on). Its rectangular form is not nearly
as simple; it is the implicit equa on x4 + y4 + 2x2 y2 2xy2 2x3 y2 = 0. The
conversion is not hard, but takes several steps, and is le as a problem in the
Exercise sec on.
Gallery of Polar Curves

There are a number of basic and classic polar curves, famous for their
beauty and/or applicability to the sciences. This sec on ends with a small gallery
of some of these graphs. We encourage the reader to understand how these
graphs are formed, and to inves gate with technology other types of polar funcons.
Lines
Through the origin:
Horizontal line:
Ver cal line:
Not through origin:
r = a csc
r = a sec
r=
Notes:
510
a
z }| {
slo
pe
b
sin m cos
Spiral
Circles
Centered on x-axis:
r = a cos
Centered on y-axis:
r = a sin
Centered on origin:
r=a
Archimedean spiral
r=
.z
a
}|
Limaons
Symmetric about x-axis: r = a b cos ;

With inner loop:
a
<1
b
.z
a
}| {
Symmetric about y-axis: r = a b sin ;
Cardiod:
a
=1
b
Dimpled:
a
1< <2
b
a, b > 0
Convex:
a
>2
b
Rose Curves
Symmetric about x-axis: r = a cos(n); Symmetric about y-axis: r = a sin(n)
Curve contains 2n petals when n is even and n petals when n is odd.
r = a cos(2)
r = a sin(2)
r = a cos(3)
r = a sin(3)
Special Curves
Rose curves
Lemniscate:
r = a sin(/5)
r = a sin(2/5)
Eight Curve:
r = a cos(2)
r2 = a2 sec4 cos(2)

Earlier we discussed how each point in the plane does not have a unique
representa on in polar form. This can be a good thing, as it allows for the
beau ful and interes ng curves seen in the preceding gallery. However, it can
also be a bad thing, as it can be dicult to determine where two curves intersect.
. Example 302
.Finding points of intersec on with polar curves
Determine where the graphs of the polar equa ons r = 1+3 cos and r = cos
intersect.
/2
2
0
4
As technology is generally readily available, it is generally a

S
good idea to start with a graph. We have graphed the two func ons in Figure
9.47 (a); to be er discern the intersec on points, part (b) of the gure zooms
in around the origin. We start by se ng the two func ons equal to each other
and solving for :
1 + 3 cos = cos
2 cos = 1
1
cos =
2
2 4
=
,
.
3 3
(a)
/2
0.5
0.5
0.5
0.5
(b)
Figure 9.47: Graphs to help determine
the points of intersec on of the polar
func ons given in Example 302.
(There are, of course, innite solu ons to the equa on cos = 1/2; as the
limaon is traced out once on [0, 2], we restrict our solu ons to this interval.)
We need to analyze this solu on. When = 2/3 we obtain the point of
intersec on that lies in the 4th quadrant. When = 4/3, we get the point of
intersec on that lies in the 2nd quadrant. There is more to say about this second
intersec on point, however. The circle dened by r = cos is traced out once on
[0, ], meaning that this point of intersec on occurs while tracing out the circle
a second me. It seems strange to pass by the point once and then recognize
it as a point of intersec on only when arriving there a second me. The rst
me the circle arrives at this point is when = /3. It is key to understand that
these two points are the same: (cos /3, /3) and (cos 4/3, 4/3).
To summarize what we have done so far, we have found two points of intersec on: when = 2/3 and when = 4/3. When referencing the circle
r = cos , the la er point is be er referenced as when = /3.
There is yet another point of intersec on: the pole (or, the origin). We did
not recognize this intersec on point using our work above as each graph arrives
at the pole at a dierent value.
A graph intersects the pole when r = 0. Considering the circle r = cos ,
r = 0 when = /2 (and odd mul ples thereof, as the circle is repeatedly
Notes:
512

traced). The limaon intersects the pole when 1 + 3 cos = 0; this occurs when
cos = 1/3, or for = cos1 (1/3). This is a nonstandard angle, approximately = 1.9106 = 109.47 . The limaon intersects the pole twice in [0, 2];
the other angle at which the limaon is at the pole is the reec on of the rst
angle across the x-axis. That is, = 4.3726 = 250.53 . .
If all one is concerned with is the (x, y) coordinates at which the graphs intersect, much of the above work is extraneous. We know they intersect at (0, 0);
we might not care at what value. Likewise, using = 2/3 and = 4/3
can give us the needed rectangular coordinates. However, in the next sec on
we apply calculus concepts to polar func ons. When compu ng the area of a
region bounded by polar curves, understanding the nuances of the points of
intersec on becomes important.
Notes:
513
Exercises 9.4
Terms and Concepts
(a) A = P(3, )
1. In your own words, describe how to plot the polar point

P(r, ).
2. T/F: When plo ng a point with polar coordinate P(r, ), r
must be posi ve.
(c) C = (0, 4)
(d) D = (1, 3)
(b) B = P(1, 2/3)
In Exercises 11 29, graph the polar func on on the given

interval.
0 /2
3. T/F: Every point in the Cartesian plane can be represented

by a polar coordinate.
11. r = 2,
4. T/F: Every point in the Cartesian plane can be represented

uniquely by a polar coordinate.
13. r = 1 cos ,
[0, 2]
14. r = 2 + sin ,
[0, 2]
Problems
15. r = 2 sin ,
[0, 2]
5. Plot the points with the given polar coordinates.
16. r = 1 2 sin ,
[0, 2]
17. r = 1 + 2 sin ,
[0, 2]
(a) A = P(2, 0)
(c) C = P(2, /2)
(b) B = P(1, )
(d) D = P(1, /4)
6. Plot the points with the given polar coordinates.

(a) A = P(2, 3)
(c) C = P(1, 2)
(b) B = P(1, )
(d) D = P(1/2, 5/6)
7. For each of the given points give two sets of polar coordinates that iden fy it, where 0 2.
12. = /6, 1 r 2
18. r = cos(2),
[0, 2]
19. r = sin(3), [0, ]

20. r = cos(/3), [0, 3]
21. r = cos(2/3),
22. r = /2, [0, 4]

23. r = 3 sin(), [0, ]
24. r = cos sin ,
[0, 2]
25. r = 2 (/2)2 ,
.
O
[0, 6]
[, ]
3
26. r =
,
5 sin cos
27. r =
C
2
,
3 cos 2 sin
[0, 2]
[0, 2]
28. r = 3 sec , (/2, /2)

8. For each of the given points give two sets of polar coordinates that iden fy it, where .
29. r = 3 csc , (0, )

In Exercises 30 38, convert the polar equa on to a rectangular equa on.
30. r = 2 cos
.
O
31. r = 4 sin
32. r = cos + sin

33. r =
9. Convert each of the following polar coordinates to rectangular, and each of the following rectangular coordinates to
polar.
7
5 sin 2 cos
(a) A = P(2, /4)
(c) C = (2, 1)
3
cos
4
35. r =
sin
(b) B = P(2, /4)
(d) D = (2, 1)
36. r = tan
10. Convert each of the following polar coordinates to rectangular, and each of the following rectangular coordinates to
polar.
34. r =
37. r = 2
38. = /6
In Exercises 39 46, convert the rectangular equa on to a

polar equa on.
39. y = x
40. y = 4x + 7
41. x = 5
42. y = 5
43. x = y2
44. x2 y = 1
45. x2 + y2 = 7
46. (x + 1)2 + y2 = 1
In Exercises 47 54, nd the points of intersec on of the polar graphs.
47. r = sin(2) and r = cos on [0, ]
48. r = cos(2) and r = cos on [0, ]

49. r = 2 cos and r = 2 sin on [0, ]
50. r = sin and r = 3 + 3 sin on [0, 2]

51. r = sin(3) and r = cos(3) on [0, ]
52. r = 3 cos and r = 1 + cos on [, ]
53. r = 1 and r = 2 sin(2) on [0, 2]
54. r = 1 cos and r = 1 + sin on [0, 2]
55. Pick a integer value(for n,)where n = 2, 3, and use technolm
for three dierent integer values
ogy to plot r = sin
n
of m. Sketch these and determine a minimal interval on
which the en re graph is shown.
56. Create your own polar func on, r = f() and sketch it. Describe why the graph looks as it does.
9.5 Calculus and Polar Func ons

The previous sec on dened polar coordinates, leading to polar func ons. We
inves gated plo ng these func ons and solving a fundamental ques on about
their graphs, namely, where do two polar graphs intersect?
We now turn our a en on to answering other ques ons, whose solu ons
require the use of calculus. A basis for much of what is done in this sec on is
the ability to turn a polar func on r = f() into a set of parametric equa ons.
Using the iden es x = r cos and y = r sin , we can create the parametric
equa ons x = f() cos , y = f() sin and apply the concepts of Sec on 9.3.
Polar Func ons and
dy
dx
We are interested in the lines tangent a given graph, regardless of whether

that graph is produced by rectangular, parametric, or polar equa ons. In each
of these contexts, the slope of the tangent line is dy
dx . Given r = f(), we are
generally not concerned with r = f (); that describes how fast r changes with
respect to . Instead, we will use x = f() cos , y = f() sin to compute dy
dx .
Using Key Idea 38 we have
dy
dy / dx
=
.
dx
d d
Each of the two deriva ves on the right hand side of the equality requires the
use of the Product Rule. We state the important result as a Key Idea.
Key Idea 42
Finding
dy
dx
with Polar Func ons
Let r = f() be a polar func on. With x = f() cos and y = f() sin ,
dy
f () sin + f() cos
=
.
dx
f () cos f() sin
. Example 303
.Finding dy
dx with polar func ons.
Consider the limaon r = 1 + 2 sin on [0, 2].
1. Find the equa ons of the tangent and normal lines to the graph at =
/4.
2. Find where the graph has ver cal and horizontal tangent lines.
Notes:
516

S
1. We start by compu ng
dy
dx .
With f () = 2 cos , we have
dy
2 cos sin + cos (1 + 2 sin )
=
dx
2 cos sin (1 + 2 sin )
cos (4 sin + 1)
=
.
2(cos2 sin2 ) sin
When = /4, dy
dx = 2 2 1 (this requires a bit of simplica on).
In
coordinates, the point on the graph at = /4 is (1 +
rectangular
2/2, 1 + 2/2). Thus the rectangular equa on of the line tangent to
the limaon at = /4 is
(
)
y = (2 2 1) x (1 + 2/2) + 1 + 2/2 3.83x + 8.24.
The limaon and the tangent line are graphed in Figure 9.48.
The normal line has the oppositereciprocal slope as the tangent line, so
its equa on is
1
y
x + 1.26.
3.83
.
2. To nd the horizontal lines of tangency, we nd where
nd where the numerator of our equa on for dy
dx is 0.
cos (4 sin + 1) = 0
dy
dx
/2
3
2
.
0
1
Figure 9.48: The limaon in Example 303

with its tangent line at = /4 and
points of ver cal and horizontal tangency.
= 0; thus we
cos = 0 or 4 sin + 1 = 0.
On [0, 2], cos = 0 when = /2, 3/2.

Se ng 4 sin + 1 = 0 gives = sin1 (1/4) 0.2527 = 14.48 .
We want the results in [0, 2]; we also recognize there are two solu ons,
one in the 3rd quadrant and one in the 4th . Using reference angles, we
have our two solu ons as = 3.39 and 6.03 radians. The four points we
obtained where the limaon has a horizontal tangent line are given in Figure 9.48 with blacklled dots.
To nd the ver cal lines of tangency, we set the denominator of
dy
dx
= 0.
2(cos2 sin2 ) sin = 0.

Convert the cos2 term to 1 sin2 :
2(1 sin2 sin2 ) sin = 0
4 sin2 + sin 1 = 0.
Notes:
517

Recognize this as a quadra c in the variable sin . Using the quadra c
formula, we have
1 33
sin =
.
8
We solve sin =
1+ 33
8
and sin =
1 + 33
sin =
8
)
(
1 + 33
1
= sin
8
1 33
:
8
1 33
sin =
8
)
(
1 33
1
= sin
8
= 1.0030
= 0.6399
In each of the solu ons above, we only get one of the possible two solu ons as sin1 x only returns solu ons in [/2, /2], the 4th and 1st
quadrants. Again using reference angles, we have:
1 + 33
sin =
= 0.6399, 3.7815 radians
8
and
1 33
sin =
8
.
/2
0.5
0.5
0.5
0.5
Figure 9.49: Graphing the tangent lines at

the pole in Example 304.
= 4.1446, 5.2802 radians.
These points are also shown in Figure 9.48 with whitelled dots.
When the graph of the polar func on r = f() intersects the pole, it means
that f() = 0 for some angle . Thus the formula for dy
dx in such instances is very
simple, reducing simply to
dy
= tan .
dx
This equa on makes an interes ng point. It tells us the slope of the tangent
line at the pole is tan ; some of our previous work (see, for instance, Example
298) shows us that the line through the pole with slope tan has polar equa on
= . Thus when a polar graph touches the pole at = , the equa on of the
tangent line at the pole is = .
. Example 304
.Finding tangent lines at the pole.
Let r = 1 + 2 sin , a limaon. Find the equa ons of the lines tangent to the
graph at the pole.
Notes:
518

S
We need to know when r = 0.

1 + 2 sin = 0
sin = 1/2
7 11
,
.
=
6
6
Note: Recall that the area of a sector of a

circle with radius r subtended by an angle
is A = 12 r2 .
Thus the equa ons of the tangent lines, in polar, are = 7/6 and = 11/6.
In rectangular form, the tangent lines are y = tan(7/6)x and y = tan(11/6)x.
The full limaon can be seen in Figure 9.48; we zoom in on the tangent lines in
Figure 9.49. .
Area
/2
1
r = f()
0.5
=
0
0.5
(a)
/2
1
r = f()
When using rectangular coordinates, the equa ons x = h and y = k dened

ver cal and horzontal lines, respec vely, and combina ons of these lines create
rectangles (hence the name rectangular coordinates). It is then somewhat
natural to use rectangles to approximate area as we did when learning about
the denite integral.
When using polar coordinates, the equa ons = and r = c form lines
through the origin and circles centered at the origin, respec vely, and combina ons of these curves form sectors of circles. It is then somewhat natural to
calculate the area of regions dened by polar func ons by rst approxima ng
with sectors of circles.
Consider Figure 9.50 (a) where a region dened by r = f() on [, ] is given.
(Note how the sides of the region are the lines = and = , whereas in
rectangular coordinates the sides of regions were o en the ver cal lines x = a
and x = b.)
Par on the interval [, ] into n equally spaced subintervals as = 1 <
2 < < n+1 = . The length of each subinterval is = ( )/n,
represen ng a small change in angle. The area of the region dened by the i th
subinterval [i , i+1 ] can be approximated with a sector of a circle with radius
f(ci ), for some ci in [i , i+1 ]. The area of this sector is 12 f(ci )2 . This is shown
in part (b) of the gure, where [, ] has been divided into 4 subintervals. We
approximate the area of the whole region by summing the areas of all sectors:
0.5
0
0.5
(b)
Area
1
i=1
f(ci )2 .
Figure 9.50: Compu ng the area of a polar region.
This is a Riemann sum. By taking the limit of the sum as n , we nd the
Notes:
519

exact area of the region in the form of a denite integral.
.
Theorem 83
Note: Example
305 requires the use of
the integral cos2 d. This is handled
well by using the power reducing formula
as found in the back of this text. Due to
the nature of the area formula, integrating cos2 and sin2 is required o en.
We oer here these indenite integrals
as a mesaving measure.
1
1
cos2 d = + sin(2) + C
2
4
1
1
sin2 d = sin(2) + C
2
4
Area of a Polar Region
Let f be con nuous and non-nega ve on [, ], where 0 2.

The area A of the region bounded by the curve r = f() and the lines
= and = is
A =
1
2
f()2 d =
1
2
r 2 d
The theorem states that 0 2. This ensures that region does not
overlap itself, giving a result that does not correspond directly to the area.
. Example 305
Area of a polar region
Find the area of the circle dened by r = cos .
This is a direct applica on of Theorem 83. The circle is traced
S
out on [0, ], leading to the integral
1
Area =
cos2 d
2 0
1 1 + cos(2)
=
d
2 0
2

)
1
1(
= + sin(2)

4
2
0
1
= .
4
=
/
3
/2
Of course, we already knew the area of a circle with radius 1/2. We did this example to demonstrate that the area formula is correct. .
6
/
0
.
Figure 9.51: Finding the area of the
.
shaded
region of a cardiod in Example
306.
520
. Example 306
.Area of a polar region
Find the area of the cardiod r = 1+cos bound between = /6 and = /3,
Notes:

This is again a direct applia on of Theorem 83.
1
Area =
2
=
1
2
1
=
2
/3
(1 + cos )2 d
/6
/3
(1 + 2 cos + cos2 ) d
/6
) /3
1
1

+ 2 sin + + sin(2)

2
4
/6
)
1(
= + 4 3 4 0.7587.
8
.
/2
Area Between Curves
Area =
Key Idea 43
1
2
r22 d
1
2
r12 d =
1
2
)
r22 r12 d.
r2 = f2 ()
r1 = f1 ()
0.5
Our study of area in the context of rectangular func ons led naturally to
nding area bounded between curves. We consider the same in the context of
polar func ons.
Consider the shaded region shown in Figure 9.52. We can nd the area of
this region by compu ng the area bounded by r2 = f2 () and subtrac ng the
area bounded by r1 = f1 () on [, ]. Thus
0.5
Figure 9.52: Illustra ng area bound between two polar curves.
Area Between Polar Curves
The area A of the region bounded by r1 = f1 () and r2 = f2 (), =

and = , where f1 () f2 () on [, ], is
1
A=
2
r22
r12
/2
d.
1
. Example 307
.Area between polar curves
Find the area bounded between the curves r = 1 + cos and r = 3 cos , as
shown in Figure 9.53.
S
Notes:
We need to nd the points of intersec on between these
0
1
.
Figure 9.53: Finding the area between polar curves in Example 307.
521

two func ons. Se ng them equal to each other, we nd:
1 + cos = 3 cos
cos = 1/2
= /3
Thus we integrate
1
2
)
(3 cos )2 (1 + cos )2 on [/3, /3].
Area =
=
1
2
1
2
/3
)
(3 cos )2 (1 + cos )2 d
)
8 cos2 2 cos 1 d
/3
/3
/3
/3
)
= 2 sin(2) 2 sin + 3

(
/3
= 2.
Amazingly enough, the area between these curves has a nice value. .
/2
. Example 308
.Area dened by polar curves
Find the area bounded between the polar curves r = 1 and r = 2 cos(2), as
shown in Figure 9.54 (a).
We need to nd the point of intersec on between the two

S
curves. Se ng the two func ons equal to each other, we have
0
1
(a)
0.5
0
0.5
A1 =
(b)
Figure 9.54:
Graphing the region
bounded by the func ons in Example
308.
522
cos(2) =
1
2
2 = /3
= /6.
In part (b) of the gure, we zoom in on the region and note that it is not really
bounded between two polar curves, but rather by two polar curves, along with
= 0. The dashed line breaks the region into its component parts. Below
the dashed line, the region is dened by r = 1, = 0 and = /6. (Note:
the dashed line lies on the line = /6.) Above the dashed line the region is
bounded by r = 2 cos(2) and = /6. Since we have two separate regions,
we nd the area using two separate integrals.
Call the area below the dashed line A1 and the area above the dashed line
A2 . They are determined by the following integrals:
/2
1
2 cos(2) = 1
Notes:
1
2
/6
(1)2 d
0
A2 =
1
2
/4
/6
(
)2
2 cos(2) d.

(The upper bound of the integral compu ng A2 is /4 as r = 2 cos(2) is at the
pole when = /4.)
We omit the
integra on details and let the reader
verify that A1 = /12 and
A2 = /12 3/8; the total area is A = /6 3/8. .
Arc Length
As we have already considered the arc length of curves dened by rectangular and parametric equa ons, we now consider it in the context of polar equaons. Recall that the arc length L of the graph dened by the parametric equaons x = f(t), y = g(t) on [a, b] is
L=
2
f (t) + g (t) dt =
x (t)2 + y (t)2 dt.
(9.1)
Now consider the polar func on r = f(). We again use the iden es x =
f() cos and y = f() sin to create parametric equa ons based on the polar
func on. We compute x () and y () as done before when compu ng dy
dx , then
apply Equa on (9.1).
The expression x ()2 + y ()2 can be simplied a great deal; we leave this
as an exercise and state that
x ()2 + y ()2 = f ()2 + f()2 .
This leads us to the arc length formula.
Key Idea 44
Arc Length of Polar Curves
Let r = f() be a polar func on with f con nuous on an open interval

I containing [, ], on which the graph traces itself only once. The arc
length L of the graph on [, ] is
L=
f ()2 + f()2 d =
(r )2 + r2 d.
. Example 309
.Arc length of a limaon
Find the arc length of the limaon r = 1 + 2 sin t.
With r = 1 + 2 sin t, we have r = 2 cos t. The limaon is
S
traced out once on [0, 2], giving us our bounds of integra on. Applying Key
Notes:
523

Idea 44, we have
/2
(2 cos )2 + (1 + 2 sin )2 d
4 cos2 + 4 sin2 + 4 sin + 1 d
4 sin + 5 d
L=
=
0
.2
13.3649.
0
1
Figure 9.55: The limaon in Example 309

whose arc length is measured.
The nal integral cannot be solved in terms of elementary func ons, so we resorted to a numerical approxima on. (Simpsons Rule, with n = 4, approximates
the value with 13.0608. Using n = 22 gives the value above, which is accurate
to 4 places a er the decimal.) .
Surface Area
The formula for arc length leads us to a formula for surface area. The following Key Idea is based on Key Idea 40.
.
Key Idea 45
Consider the graph of the polar equa on r = f(), where f is con nuous
on an open interval containing [, ] on which the graph does not cross
itself.
1. The surface area of the solid formed by revolving the graph about
the ini al ray ( = 0) is:
Surface Area = 2
f() sin
f ()2 + f()2 d.
2. The surface area of the solid formed by revolving the graph about
the line = /2 is:
Surface Area = 2
f() cos
Notes:
524
f ()2 + f()2 d.

. Example 310
Surface area determined by a polar curve
Find the surface area formed by revolving one petal of the rose curve r = cos(2)
about its central axis (see Figure 9.56).
/2
1
We choose, as implied by the gure, to revolve the por on

S
of the curve that lies on [0, /4] about the ini al ray. Using Key Idea 45 and the
fact that f () = 2 sin(2), we have
/4
Surface Area = 2
(
)2 (
)2
2 sin(2) + cos(2) d
cos(2) sin()
1.36707.
(a)
The integral is another that cannot be evaluated in terms of elementary funcons. Simpsons Rule, with n = 4, approximates the value at 1.36751...
x
1
.
.
(b)
Figure 9.56: Finding the surface area of a
rosecurve petal that is revolved around
its central axis.
Notes:
525
Exercises 9.5
Terms and Concepts
1. Given polar equa on r = f(), how can one create parametric equa ons of the same curve?
2. With rectangular coordinates, it is natural to approximate
area with
; with polar coordinates, it is natural to
approximate area with
.
In Exercises 17 27, nd the area of the described region.
17. Enclosed by the circle: r = 4 sin

18. Enclosed by the circle r = 5
19. Enclosed by one petal of r = sin(3)
Problems
20. Enclosed by the cardiod r = 1 sin
In Exercises 3 10, nd:

(a)
21. Enclosed by the inner loop of the limaon r = 1 + 2 cos t
dy
dx
(b) the equa on of the tangent and normal lines to the

curve at the indicated value.
3. r = 1;
= /4
4. r = cos ;
22. Enclosed by the outer loop of the limaon r = 1 + 2 cos t

(including area enclosed by the inner loop)
23. Enclosed between the inner and outer loop of the limaon
r = 1 + 2 cos t
= /4
5. r = 1 + sin ;
= /6
24. Enclosed by r = 2 cos and r = 2 sin , as shown:

y
6. r = 1 3 cos ;
7. r = ;
= 3/4
= /2
8. r = cos(3);
= /6
x
1
9. r = sin(4);
10. r =
= /3
1
;
sin cos
In Exercises 11 14, nd the values of in the given interval where the graph of the polar func on has horizontal and
ver cal tangent lines.
25. Enclosed by r = cos(3) and r = sin(3), as shown:

y
0.5
11. r = 3; [0, 2]
x
12. r = 2 sin ;
[0, ]
.
13. r = cos(2);
14. r = 1 + cos ;
[0, 2]
[0, 2]
26. Enclosed by r = cos and r = sin(2), as shown:

y
In Exercises 15 16, nd the equa on of the lines tangent to

the graph at the pole.
15. r = sin ;
[0, ]
16. r = sin(3); [0, ]

.
x
1
27. Enclosed by r = 3 cos and r = 1 cos , as shown:

y
32. Approximate the arc length of the cardiod r = 1 + cos

with Simpsons Rule and n = 6.
x
1
31. Approximate the arc length of one petal of the rose curve
r = sin(3) with Simpsons Rule and n = 4.
In Exercises 33 37, answer the ques ons involving surface

area.
33. Use Key Idea 45 to nd the surface area of the sphere
formed by revolving the circle r = 2 about the ini al ray.
In Exercises 28 32, answer the ques ons involving arc

length.
28. Let x() = f() cos and y() = f() sin . Show, as suggested by the text, that
x ()2 + y ()2 = f ()2 + f()2 .
29. Use the arc length formula to compute the arc length of the
circle r = 2.
30. Use the arc length formula to compute the arc length of the
circle r = 4 sin .
34. Use Key Idea 45 to nd the surface area of the sphere

formed by revolving the circle r = 2 cos about the ini al
ray.
35. Find the surface area of the solid formed by revolving the
cardiod r = 1 + cos about the ini al ray.
circle r = 2 cos about the line = /2.
line r = 3 sec , /4 /4, about the line
= /2.
10: V
Up to this point in this text we have considered mathema cs in a 2dimensional
world. We have plo ed graphs on the x-y plane using rectangular and polar
coordinates and found the area of regions in the plane. We have considered
proper es of solid objects, such as volume and surface area, but only by rst
dening a curve in the plane and then rota ng it out of the plane.
While there is wonderful mathema cs to explore in 2D, we live in a 3D
world and eventually we will want to apply mathema cs involving this third dimension. In this sec on we introduce Cartesian coordinates in space and explore basic surfaces. This will lay a founda on for much of what we do in the
remainder of the text.
Each point P in space can be represented with an ordered triple, P = (a, b, c),
where a, b and c represent the rela ve posi on of P to the x-, y- and z-axes, respec vely. Each axis is perpendicular to the other two.
Visualizing points in space on paper can be problema c, as we are trying
to represent a 3-dimensional concept on a 2dimensional medium. We cannot
draw three line represen ng the three axes in which each line is perpendicular to
the other two. Despite this issue, standard conven ons exist for plo ng shapes
in space that we will discuss that are more than adequate.
One conven on is that the axes must conform to the right hand rule. This
rule states that when the index nger of the right hand is extended in the direcon of the posi ve x-axis, and the middle nger (bent inward so it is perpendicular to the palm) points along the posi ve y axis, then the extended thumb
will point in the direc on of the posi ve z-axis. (It may take some thought to
verify this, but this system is inherently dierent from the one created by using
the le hand rule.)
As long as the coordinate axes are posi oned so that they follow this rule,
it does not ma er how the axes are drawn on paper. There are two popular
methods that we briey discuss.
In Figure 10.1 we see the point P = (2, 1, 3) plo ed on a set of axes. The
basic conven on here is that the x-y plane is drawn in its standard way, with the
z-axis down to the le . The perspec ve is that the paper represents the x-y plane
and the posi ve z axis is coming up, o the page. This method is preferred by
many engineers. Because it can be hard to tell where a single point lies in rela on
y
2
1
1
.
2
1
2
x
1
1
1
3
2
Figure 10.1: Plo ng the point P

(2, 1, 3) in space.
Chapter 10 Vectors
to all the axes, dashed lines have been added to let one see how far along each
axis the point lies.
One can also consider the x-y plane as being a horizontal plane in, say, a
room, where the posi ve z-axis is poin ng up. When one steps back and looks
at this room, one might draw the axes as shown in Figure 10.2. The same point P
is drawn, again with dashed lines. This point of view is preferred by most mathema cians, and is the conven on adopted by this text.
z
2
2
2
x 2
Figure 10.2: Plo ng the point P =

(2, 1, 3) in space with a perspec ve used
in this text.
Measuring Distances
It is of cri cal importance to know how to measure distances between points
in space. The formula for doing so is based on measuring distance in the plane,
and is known (in both contexts) as the Euclidean measure of distance.
Deni on 48
Distance In Space
Let P = (x1 , y1 , z1 ) and Q = (x2 , y2 , z2 ) be points in space. The distance

D between P and Q is
D = (x2 x1 )2 + (y2 y1 )2 + (z2 z1 )2 .
We refer to the line segment that connects points P and Q in space as PQ,
and refer to the length of this segment as ||PQ||. The above distance formula
allows us to compute the length of this segment.
. Example 311
Length of a line segment
Let P = (1, 4, 1) and let Q = (2, 1, 1). Draw the line segment PQ and nd its
length.
z
2
.
x
. Figure 10.3: Plo
ng points P and Q in Ex-
The points P and Q are plo ed in Figure 10.3; no special conS

sidera on need be made to draw the line segment connec ng these two points;
simply connect them with a straight line. One cannot actually measure this line
on the page and deduce anything meaningful; its true length must be measured
analy cally. Applying Deni on 48, we have
||PQ|| = (2 1)2 + (1 4)2 + (1 (1))2 = 14 3.74.

.
ample 311.
Spheres
Just as a circle is the set of all points in the plane equidistant from a given
Notes:
530

point (its center), a sphere is the set of all points in space that are equidistant
from a given point. Deni on 48 allows us to write an equa on of the sphere.
We start with a point C = (a, b, c) which is to be the center of a sphere with
radius r. If a point P = (x, y, z) lies on the sphere, then P is r units from C; that
is,
||PC|| = (x a)2 + (y b)2 + (z c)2 = r.

Squaring both sides, we get the standard equa on of a sphere in space with
center at C = (a, b, c) with radius r, as given in the following Key Idea.
Key Idea 46
Standard Equa on of a Sphere in Space
The standard equa on of the sphere

.with radius r, centered at C =
(a, b, c), is
(x a)2 + (y b)2 + (z c)2 = r2 .
. Example 312
Equa on of a sphere
Find the center and radius of the sphere dened by x2 +2x+y2 4y+z2 6z = 2.
z
To determine the center and radius, we must put the equaS

on in standard form. This requires us to complete the square (three mes).
x2 + 2x + y2 4y + z2 6z = 2
(x2 + 2x + 1) + (y2 4y + 4) + (z2 6z + 9) 14 = 2
2
(x + 1) + (y 2) + (z 3) = 16
2
The sphere is centered at (1, 2, 3) and has a radius of 4. .
the x-y plane
The equa on of a sphere is an example of an implicit func on dening a surface in space. In the case of a sphere, the variables x, y and z are all used. We
now consider situa ons where surfaces are dened where one or two of these
variables are absent.
Introduc on to Planes in Space

The coordinate axes naturally dene three planes (shown in Figure 10.4), the
coordinate planes: the x-y plane, the y-z plane and the x-z plane. The x-y plane
is characterized as the set of all points in space where the z-value is 0. This,
in fact, gives us an equa on that describes this plane: z = 0. Likewise, the x-z
plane is all points where the y-value is 0, characterized by y = 0.
the y-z plane
.
y
the x-z plane
Notes:
Figure 10.4: The coordinate planes.
531
Chapter 10 Vectors
The equa on x = 2 describes all points in space where the x-value is 2. This
is a plane, parallel to the y-z coordinate plane, shown in Figure 10.5.
2
y
. Example 313
Regions dened by planes
Sketch the region dened by the inequali es 1 y 2.
The region is all points between the planes y = 1 and
S
y = 2. These planes are sketched in Figure 10.6, which are parallel to the x-z
plane. Thus the region extends innitely in the x and z direc ons, and is bounded
by planes in the y direc on. .
.
Figure 10.5: The plane x = 2.
Cylinders
z
.
1
.
Figure 10.6: Sketching the boundaries of
a region in Example 313.
Deni on 49
.
x
(a)
z
.
x
2
y
In this text, we consider curves C that lie in planes parallel to one of the
coordinate planes, and lines L that are perpendicular to these planes, forming
right cylinders. Thus the directrix can be dened using equa ons involving 2
variables, and the rulings will be parallel to the axis of the 3rd variable.
In the example preceding the deni on, the curve x2 + y2 = 1 in the x-y
plane is the directrix and the rulings are lines parallel to the z-axis. (Any circle
shown in Figure 10.7 can be considered a directrix; we simply choose the one
where z = 0.) Sample rulings can also be viewed in part (b) of the gure. More
examples will help us understand this deni on.
(b)
Figure 10.7: Sketching x2 + y2 = 1.
532
Cylinder
Let C be a curve in a plane and let L be

.a line not parallel to C. A cylinder
is the set of all lines parallel to L that pass through C. The curve C is the
directrix of the cylinder, and the lines are the rulings.
The equa on x = 1 obviously lacks the y and z variables, meaning it denes

points where the y and z coordinates can take on any value. Now consider the
equa on x2 + y2 = 1 in space. In the plane, this equa on describes a circle
of radius 1, centered at the origin. In space, the z coordinate is not specied,
meaning it can take on any value. In Figure 10.7 (a), we show part of the graph
of the equa on x2 +y2 = 1 by sketching 3 circles: the bo om one has a constant
z-value of 1.5, the middle one has a z-value of 0 and the top circle has a z-value
of 1. By plo ng all possible z-values, we get the surface shown in Figure 10.7
(b).
This surface looks like a tube, or a cylinder; mathema cians call this surface a cylinder for an en rely dierent reason.
Notes:
. Example 314
Graphing cylinders
Graph the cylinder following cylinders.
1. z = y2
2. x = sin z
S
1. We can view the equa on z = y2 as a parabola in the y-z plane, as illustrated in Figure 10.8 (a). As x does not appear in the equa on, the rulings
are lines through this parabola parallel to the x-axis, shown in (b). These
rulings give a general idea as to what the surface looks like, drawn in (c).
(a)
(b)
(c)
Figure 10.8: Sketching the cylinder dened by z = y2 .
2. We can view the equa on x = sin z as a sine curve that exists in the x-z
plane, as shown in Figure 10.9 (a). The rules are parallel to the y axis as
the variable y does not appear in the equa on x = sin z; some of these
are shown in part (b). The surface is shown in part (c) of the gure.
z
.
x
.
x
(a)
.
x
(b)
Figure 10.9: Sketching the cylinder dened by x = sin z.
(c)
Notes:
533
Chapter 10 Vectors
Surfaces of Revolu on
z
2
x
.
(a)
z
One of the applica ons of integra on we learned previously was to nd the

volume of solids of revolu on solids formed by revolving a curve about a horizontal or ver cal axis. We now consider how to nd the equa on of the surface
of such a solid.
Consider the surface formed by revolving y = x about the x-axis. Cross

sec ons of this surface parallel to the y-z plane are circles, as shown in Figure
2
2
2
10.10a. Each circle has equa on of the form
y + z = r for some radius r. The
radius is a func on of x; in fact, it is r(x)
= x. Thus the equa on of the surface
shown in Figure 10.10b is y2 + z2 = ( x)2 .
We generalize the above principles to give the equa ons of surfaces formed
by revolving curves about the coordinate axes.
Key Idea 47
2
x
Let r be a radius func on.
Surfaces of Revolu on, Part 1
1. The equa on of the surface formed by revolving y = r(x) or z =

r(x) about the x-axis is y2 + z2 = r(x)2 .
(b)
2. The equa on of the surface formed by revolving x = r(y) or z =

r(y) about the y-axis is x2 + z2 = r(y)2 .
Figure 10.10: Introducing surfaces of revolu on.
3. The equa on of the surface formed by revolving x = r(z) or y =

r(z) about the z-axis is x2 + y2 = r(z)2 .
. Example 315
Finding equa on of a surface of revolu on
Let y = sin z on [0, ]. Find the equa on of the surface of revolu on formed by
revolving y = sin z about the z-axis.
z
y = sin z
x = sin z
Using Key Idea 47, we nd the surface has equa on x2 +y2 =

sin z. The curve is sketched in Figure 10.11a and the surface is drawn in Figure
10.11b.
Note how the surface (and hence the resul ng equa on) is the same if we
began with the curve x = sin z, which is also drawn in Figure 10.11a. .
S
.
1
x 1
(a)
This par cular method of crea ng surfaces of revolu on is limited. For instance, in Example 210 of Sec on 7.3 we found the volume of the solid formed
by revolving y = sin x about the y-axis. Our current method of forming surfaces
can only rotate y = sin x about the x-axis. Trying to rewrite y = sin x as a funcon of y is not trivial, as simply wri ng x = sin1 y only gives part of the region
.
1
x 1
(b)
Figure 10.11: Revolving y = sin z about
the z-axis in Example 315.
534
Notes:

we desire.
What we desire is a way of wri ng the surface of revolu on formed by rota ng y = f(x) about the y-axis. We start by rst recognizing this surface is the
same as revolving z = f(x) about the z-axis. This will give us a more natural way
of viewing the surface.
A value of x is a measurement of distance from the z-axis. At the distance r,
we plot a z-height of f(r). When rota ng f(x) about the z-axis, we want all points
a distance of r from the z-axis in the x-y plane to havea z-height of f(r). All such
points sa sfy the equa on r2= x2 + y2 ; hence r = x2 + y2 . Replacing r with
x2 + y2 in f(r) gives z = f( x2 + y2 ). This is the equa on of the surface.

Key Idea 48
Surfaces of Revolu on, Part 2
.x-z plane. The surface formed by

Let z = f(x), x 0, be a curve in the
(
)
2
2
revolving this curve about the z-axis has equa on z = f
5
5
x +y .
.
(a)
. Example 316
Finding equa on of surface of revolu on
Find the equa on of the surface found by revolving z = sin x about the z-axis.
(
)
Using Key Idea 48, the surface has equa on z = sin
x2 + y2 .
S
The curve and surface are graphed in Figure 10.12. .
Quadric Surfaces
Spheres, planes and cylinders are important surfaces to understand. We

now consider one last type of surface, a quadric surface. The deni on may
look in mida ng, but we will show how to analyze these surfaces in an illumina ng way.
5
5
.
(b)
Deni on 50
Quadric Surface
A quadric surface is the graph of the general seconddegree equa on in

three variables:
Figure 10.12: Revolving z = sin x about

the z-axis in Example 316.
Ax2 + By2 + Cz2 + Dxy + Exz + Fyz + Gx + Hy + Iz + J = 0.
When the coecients D, E or F are not zero, the basic shapes of the quadric
surfaces are rotated in space. We will focus on quadric surfaces where these coeecients are 0; we will not consider rota ons. There are six basic quadric sur-
Notes:
535
Chapter 10 Vectors
faces: the ellip c paraboloid, ellip c cone, ellipsoid, hyperboloid of one sheet,
hyperboloid of two sheets, and the hyperbolic paraboloid.
We study each shape by considering traces, that is, intersec ons of each
surface with a plane parallel to a coordinate plane. For instance, consider the
ellip c paraboloid z = x2 /4 + y2 , shown in Figure 10.13. If we intersect this
shape with the plane z = d (i.e., replace z with d), we have the equa on:
In plane
y=0
In plane
x=0
d=
x2
+ y2 .
4
1=
y2
x2
+ .
4d
d
Divide both sides by d:
.
x
In plane
z=d
y
Figure 10.13: The ellip c paraboloid z =

x2 /4 + y2 .
This describes an ellipse so cross sec ons parallel to the x-y coordinate plane
are ellipses. This ellipse is drawn in the gure.
Now consider cross sec ons parallel to the x-z plane. For instance, le ng
y = 0 gives the equa on z = x2 /4, clearly a parabola. Intersec ng with the
plane x = 0 gives a cross sec on dened by z = y2 , another parabola. These
parabolas are also sketched in the gure.
Thus we see where the ellip c paraboloid gets its name: some cross sec ons
are ellipses, and others are parabolas.
Such an analysis can be made with each of the quadric surfaces. We give a
sample equa on of each, provide a sketch with representa ve traces, and describe these traces.
Notes:
536
Ellip c Paraboloid,
z=
x2
y2
+
a2
b2
Plane
x=d
y=d
z=d
Trace
Parabola
Parabola
Ellipse
In plane
y=0
In plane
x=0
.
x
In plane
z=d
y
One variable in the equa on of the ellip c paraboloid will be raised to the rst power; above,
this is the z variable. The paraboloid will open in the direc on of this variables axis. Thus
x = y2 /a2 + z2 /b2 is an ellip c paraboloid that opens along the x-axis.
Mul plying the right hand side by (1) denes an ellip c paraboloid that opens in the opposite
direc on.
z2 =
Ellip c Cone,
x2
y2
+ 2
2
a
b
Plane
x=0
y=0
Trace
Crossed Lines
Crossed Lines
x=d
y=d
z=d
Hyperbola
Hyperbola
Ellipse
in plane
y=0
x
in plane
z=d
One can rewrite the equa on as z2 x2 /a2 y2 /b2 = 0. The one variable with a posi ve
coecient corresponds to the axis that the cones open along.
x
in plane
y=d
Ellipsoid,
x2
y2
z2
+ 2 + 2 =1
2
a
b
c
Plane
x=d
y=d
z=d
Trace
Ellipse
Ellipse
Ellipse
in plane
z=0
in plane
y=0
in plane
x=0
.
If a = b = c = 0, the ellipsoid is a sphere with radius a; compare to Key Idea 46.
Hyperboloid of One Sheet,
x2
y2
z2
+ 2 2 =1
2
a
b
c
Plane
x=d
y=d
z=d
Trace
Hyperbola
Hyperbola
Ellipse
in plane
z=0
x
in plane
y=0
.
The one variable with a nega ve coecient corresponds to the axis that the hyperboloid opens
along.
y
in plane
x=0
z2
x2
y2
2 2 =1
2
c
a
b
Hyperboloid of Two Sheets,
Plane
x=d
y=d
z=d
Trace
Hyperbola
Hyperbola
Ellipse
in plane
y=0
.
The one variable with a posi ve coecient corresponds to the axis that the hyperboloid opens
along. In the case illustrated, when |d| < |c|, there is no trace.
Hyperbolic Paraboloid,
z=
x2
y2
a2
b2
Plane
x=d
y=d
z=d
Trace
Parabola
Parabola
Hyperbola
.
in plane
y=0
x
in plane
x=0
in plane
z=d
(d > 0)
.
y
in plane x
z=d
(d < 0)
.
y
The parabolic traces will open along the axis of the one variable that is raised to the rst power.
in plane
y x=0
in plane
z=d
Chapter 10 Vectors
. Example 317
.Sketching quadric surfaces
Sketch the quadric surface dened by the given equa on.
z
4
1. y =
4
x 4
y2
z2
+
= 1.
9
4
3. z = y2 x2 .
x2
z2
+ :
4
16
We rst iden fy the quadric by pa ernmatching with the equa ons given
previously. Only two surfaces have equa ons where one variable is raised
to the rst power, the ellip c paraboloid and the hyperbolic paraboloid.
In the la er case, the other variables have dierent signs, so we conclude
that this describes a hyperbolic paraboloid. As the variable with the rst
power is y, we note the parboloid opens along the y-axis.
1. y =
2. x2 +
x2
z2
+
4
16
(a)
z
To make a decent sketch by hand, we need only draw a few traces. In this
case, the traces x = 0 and z = 0 form parabolas that outline the shape.
x
x = 0: The trace is the parabola y = z2 /16
z = 0: The trace is the parabola y = x2 /4.

Graphing each trace in the respec ve plane creates a sketch as shown in
Figure 10.14 (a). This is enough to give an idea of what the paraboloid
looks like. The surface is lled in in (b).
(b)
Figure 10.14:
paraboloid.
Sketching an ellip c
y2
z2
+
=1:
9
4
This is an ellipsoid. We can get a good idea of its shape by drawing the
traces in the coordinate planes.
2. x2 +
y2
z2
x = 0: The trace is the ellipse
+
= 1. The major axis is along the
9
4
yaxis with length 6 (as b = 3, the length of the axis is 6); the minor axis
is along the z-axis with length 4.
z2
= 1. The major axis is along the
y = 0: The trace is the ellipse x2 +
4
z-axis, and the minor axis has length 2 along the x-axis.
x 3
y2
z = 0: The trace is the ellipse x2 +
= 1, with major axis along the
9
y-axis.
(a)
Graphing each trace in the respec ve plane creates a sketch as shown in

Figure 10.15 (a). Filling in the surface gives Figure 10.15 (b).
3. z = y2 x2 :
(b)
Figure 10.15: Sketching an ellipsoid.
540
Notes:

This denes a hyperbolic paraboloid, very similar to the one shown in the
gallery of quadric sec ons. Consider the traces in the yz and xz planes:
z
1
x = 0: The trace is z = y2 , a parabola opening up in the y z plane.

y = 0: The trace is z = x2 , a parabola opening down in the x z plane.
Sketching these two parabolas gives a sketch like that in Figure 10.16 (a),
and lling in the surface gives a sketch like (b).
. Example 318
Iden fying quadric surfaces
Consider the quadric surface shown in Figure 10.17. Which of the following
equa ons best ts this surface?
(a)
(b)
z2
=0
9
x2 y2 z2 = 1
x2 y2
(c)
z2 x2 y2 = 1
(d)
4x2 y2
x 1
1
(a)
z
z2
=1
9
The image clearly displays a hyperboloid of two sheets. The

2
2
2
gallery informs us that the equa on will have a form similar to cz2 ax2 by2 = 1.
We can immediately eliminate op on (a), as the constant in that equa on is
not 1.
The hyperboloid opens along the x-axis, meaning x must be the only variable with a posi ve coecient, elimina ng (c).
The hyperboloid is wider in the z-direc on than in the y-direc on, so we
need an equa on where c > b. This eliminates (b), leaving us with (d). We
2
should verify that the equa on given in (d), 4x2 y2 z9 = 1, ts.
We already established that this equa on describes a hyperboloid of two
sheets that opens in the x-direc on and is wider in the z-direc on than in the
y. Now note the coecient of the x-term. Rewri ng 4x2 in standard form, we
x2
have: 4x2 =
. Thus when y = 0 and z = 0, x must be 1/2; i.e., each
(1/2)2
hyperboloid starts at x = 1/2. This matches our gure.
z2
We conclude that 4x2 y2
= 1 best ts the graph. .
9
S
(b)
Figure 10.16:
paraboloid.
Sketching a hyperbolic
z
3
1
3
x 1
This sec on has introduced points in space and shown how equa ons can
describe surfaces. The next sec ons explore vectors, an important mathema cal
object that well use to explore curves in space.
Figure 10.17: A possible equa on of this

quadric surface is found in Example 318.
Notes:
541
Exercises 10.1
Terms and Concepts
In Exercises 23 26, a quadric surface is sketched. Determine

which of the given equa ons best ts the graph.
1. Axes drawn in space must conform to the

rule.
z
3
2. In the plane, the equa on x = 2 denes a

.
space, x = 2 denes a
; in
3. In the plane, the equa on y = x2 denes a

.
space, y = x2 denes a
; in
23.
4. Which quadric surface looks like a Pringles chip?

5. Consider the hyperbola x2 y2 = 1 in the plane. If this
hyperbola is rotated about the x-axis, what quadric surface
is formed?
.
(a)
x = y2 +
6. Consider the hyperbola x2 y2 = 1 in the plane. If this

hyperbola is rotated about the y-axis, what quadric surface
is formed?
Problems
z2
9
24.
x.
(a)
x2 y2 z2 = 0
25.
(a)
12. 0 x 3
y2
z2
x +
+
=1
3
2
2
13. x 0, y 0, z 0
In Exercises 15 18, sketch the cylinder in space.

x2
16. y = cos z
y2
z2
+
=1
9
4
26.
x2 +
(b)
14. y 3
17.
22. Revolve z = 1/x about the z-axis.
11. x2 + y2 + z2 < 1
21. Revolve z = x2 about the z-axis.
x2 y2 + z2 = 0
(b)
z
In Exercises 11 14, describe and sketch the regions in space

dened by the inequali es.
20. Revolve y = x2 about the x-axis.
1
1
9. Find the center and radius of the sphere dened by

x2 8x + y2 + 2y + z2 + 8 = 0.
x2
y2
+
=1
4
9
1
18. y =
x
In Exercises 19 22, give the equa on of the surface of revolu on described.
1
about the y-axis.
19. Revolve z =
1 + y2
z2
3
8. The points A = (1, 1, 3), B = (3, 2, 7), C = (2, 0, 8) and

D = (0, 1, 4) form a quadrilateral ABCD in space. Is this
a parallelogram?
15. z = x3
x = y2 +
(b)
z
7. The points A = (1, 4, 2), B = (2, 6, 3) and C = (4, 3, 1)

form a triangle in space. Find the distances between each
pair of points and determine if the triangle is a right triangle.
10. Find the center and radius of the sphere dened by

x2 + y2 + z2 + 4x 2y 4z + 4 = 0.
(a)
y2 x2 z2 = 1
(b)
y2 + x2 z2 = 1
In Exercises 27 32, sketch the quadric surface.

27. z y2 + x2 = 0
y2
28. z2 = x2 +
4
29. x = y2 z2
30. 16x2 16y2 16z2 = 1
z2
x2
y2 +
=1
31.
9
25
2
2
2
32. 4x + 2y + z = 4
10.2 An Introduc on to Vectors

Many quan es we think about daily can be described by a single number: temperature, speed, cost, weight and height. There are also many other concepts
we encounter daily that cannot be described with just one number. For instance,
a weather forecaster o en describes wind with its speed and its direc on (. . .
with winds from the southeast gus ng up to 30 mph . . .). When applying a
force, we are concerned with both the magnitude and direc on of that force.
In both of these examples, direc on is important. Because of this, we study
vectors, mathema cal objects that convey both magnitude and direc on informa on.
One barebones deni on of a vector is based on what we wrote above:
a vector is a mathema cal object with magnitude and direc on parameters.
This deni on leaves much to be desired, as it gives no indica on as to how
such an object is to be used. Several other deni ons exist; we choose here a
deni on rooted in a geometric visualiza on of vectors. It is very simplis c but
readily permits further inves ga on.
y
5
Deni on 51
Vector
A vector is a directed line segment.

Given points P and Q (either in the plane or in space), we denote with
#
PQ the vector from P to Q. The point P is said to be the ini al point of
the vector, and the point Q is the terminal point.
Figure 10.18: Drawing the same vector

with dierent ini al points.
#
The magnitude, or norm of PQ is the length of the line segment PQ:
#
|| PQ || = || PQ ||.
Two vectors are equal if they have the same magnitude and direc on.
y
4
Figure 10.18 shows mul ple instances of the same vector. Each directed line
segment has the same direc on and length (magnitude), hence each is the same
vector.
We use R2 (pronounced r two) to represent all the vectors in the plane,
and use R3 (pronounced r three) to represent all the vectors in space.
#
#
Consider the vectors PQ and RS as shown in Figure 10.19. The vectors look to
be equal; that is, they seem to have the same length and direc on. Indeed, they
are. Both vectors move 2 units to the right and 1 unit up from the ini al point
to reach the terminal point. One can analyze this movement to measure the
S
R
Q
P
x
2
Figure 10.19: Illustra ng how equal vectors have the same displacement.
Notes:
543
Chapter 10 Vectors
magnitude of the vector, and the movement itself gives direc on informa on
(one could also measure the slope of the line passing through P and Q or R and
S). Since they have the same length and direc on, these two vectors are equal.
This demonstrates that inherently all we care about is displacement; that is,
how far in the x, y and possibly z direc ons the terminal point is from the ini al
#
#
point. Both the vectos PQ and RS in Figure 10.19 have an x-displacement of 2
and a y-displacement of 1. This suggests a standard way of describing vectors
in the plane. A vector whose x-displacement is a and whose y-displacement is
b will have terminal point (a, b) when the ini al point is the origin, (0, 0). This
leads us to a deni on of a standard and concise way of referring to vectors.
.
Deni on 52
Component Form of a Vector
1. The component form of a vector v in R2 , whose terminal point is

(a, b) when its ini al point is (0, 0), is a, b .
2. The component form of a vector v in R3 , whose terminal point is

(a, b, c) when its ini al point is (0, 0, 0), is a, b, c .
The numbers a, b (and c, respec vely) are the components of v.
#
It follows from the deni on that the component form of the vector PQ,
where P = (x1 , y1 ) and Q = (x2 , y2 ) is
#
PQ = x2 x1 , y2 y1 ;
#
in space, where P = (x1 , y1 , z1 ) and Q = (x2 , y2 , z2 ), the component form of PQ
is
#
PQ = x2 x1 , y2 y1 , z2 z1 .
We prac ce using this nota on in the following example.
. Example 319
.Using component form nota on for vectors
1. Sketch the vector v = 2, 1 star ng at P = (3, 2) and nd its magnitude.

whose ini al point is R = (3, 2)
2. Find the component form of the vector w
and whose terminal point is S = (1, 2).
3. Sketch the vector u = 2, 1, 3 star ng at the point Q = (1, 1, 1) and
nd its magnitude.
Notes:
544

S
y
1. Using P as the ini al point, we move 2 units in the posi ve x-direc on and
1 units in the posi ve y-direc on to arrive at the terminal point P =
(5, 1), as drawn in Figure 10.20 (a).
5
S
The magnitude of v is determined directly from the component form:
|| v || = 22 + (1)2 = 5.
P
P
x
5
R
5
2. Using the note following Deni on 52, we have
(a)
#
RS = 1 (3), 2 (2) = 2, 4 .
One can readily see from Figure 10.20 (a) that the x- and y-displacement
#
of RS is 2 and 4, respec vely, as the component form suggests.
3. Using Q as the ini al point, we move 2 units in the posi ve x-direc on,
1 unit in the posi ve y-direc on, and 3 units in the posi ve z-direc on
to arrive at the terminal point Q = (3, 0, 4), illustrated in Figure 10.20
(b).
The magnitude of u is:
|| u || = 22 + 02 + 32 = 13.
.
Q
y
(b)
Figure 10.20: Graphing vectors in Example 319.
Now that we have dened vectors, and have created a nice nota on by which
to describe them, we start considering how vectors interact with each other.
That is, we dene an algebra on vectors.
Notes:
545
Chapter 10 Vectors
.
Deni on 53
Vector Algebra
1. Let u = u1 , u2 and v = v1 , v2 be vectors in R2 , and let c be a

scalar.
(a) The addi on, or sum, of the vectors u and v is the vector
u + v = u1 + v1 , u2 + v2 .
(b) The scalar product of c and v is the vector

cv = c v1 , v2 = cv1 , cv2 .
2. Let u = u1 , u2 , u3 and v = v1 , v2 , v3 be vectors in R3 , and let c
be a scalar.
(a) The addi on, or sum, of the vectors u and v is the vector
u + v = u1 + v1 , u2 + v2 , u3 + v3 .
(b) The scalar product of c and v is the vector
cv = c v1 , v2 = cv1 , cv2 , cv3 .
In short, we say addi on and scalar mul plica on are computed component
wise.
. Example 320
Adding vectors
Sketch the vectors u = 1, 3, v = 2, 1 and u + v all with ini al point at the
origin.
y
4
S
u
+
u + v = 1, 3 + 2, 1
= 3, 4 .
These are all sketched in Figure 10.21. .
Figure 10.21: Graphing the sum of vectors

in Example 320.
As vectors convey magnitude and direc on informa on, the sum of vectors
also convey length and magnitude informa on. Adding u + v suggests the following idea:
Notes:
546
We rst compute u + v.

Star ng at an ini al point, go out u, then go out v.
y
4
u
+
This idea is sketched in Figure 10.22, where the ini al point of v is the terminal point of u. This is known as the Head to Tail Rule of adding vectors. Vector
addi on is very important. For instance, if the vectors u and v represent forces
ac ng on a body, the sum u + v gives the resul ng force. Because of various
physical applica ons of vector addi on, the sum u +v is o en referred to as the
resultant vector, or just the resultant.
Analy cally, it is easy to see that u + v = v + u. Figure 10.22 also gives a
graphical representa on of this, using gray vectors. Note that the vectors u and
v, when arranged as in the gure, form a parallelogram. Because of this, the
Head to Tail Rule is also known as the Parallelogram Law: the vector u + v is
dened by forming the parallelogram dened by the vectors u and v; the ini al
point of u + v is the common ini al point of parallelogram, and the terminal
point of the sum is the common terminal point of the parallelogram.
While not illustrated here, the Head to Tail Rule and Parallelogram Law hold
for vectors in R3 as well.
It follows from the proper es of the real numbers and Deni on 53 that
v
x
Figure 10.22: Illustra ng how to add vectors using the Head to Tail Rule and Parallelogram Law.
u v = u + (1)v.
The Parallelogram Law gives us a good way to visualize this subtrac on. We
demonstrate this in the following example.
. Example 321
Vector Subtrac on
Let u = 3, 1 and v = 1, 2 . Compute and sketch u v.
The computa on of u v is straigh orward, and we show
S
all steps below. Usually the formal step of mul plying by (1) is omi ed and
we just subtract.
u v = u + (1)v
= 3, 1 + 1, 2
= 2, 1 .
. Example 322
.Scaling vectors
u
x
2
Figure 10.23 illustrates, using the Head to Tail Rule, how the subtrac on can be
viewed as the sum u + (v). The gure also illustrates how u v can be obtained by looking only at the terminal points of u andv (when their ini al points
are the same). .
.
Figure 10.23: Illustra ng how to subtract
vectors graphically.
1. Sketch the vectors v = 2, 1 and 2v with ini al point at the origin.
Notes:
547
Chapter 10 Vectors
2. Compute the magnitudes of v and 2v.
y
3
S
2
1. We compute 2v:
2v
2v = 2 2, 1
= 4, 2 .
v
x
These are sketched in Figure 10.24. Make note that 2v does not start at
the terminal point of v; rather, its ini al point is also the origin.
Figure 10.24: Graphing vectors v and 2v

in Example 322.
2. The gure suggests that 2v is twice as long as v. We compute their magnitudes to conrm this.
|| v || = 22 + 12
= 5.
|| 2v || = 42 + 22
= 20
= 4 5 = 2 5.
.
As we suspected, 2v is twice as long as v.
The zero vector is the vector whose ini al point is also its terminal point. It
is denoted by 0. Its component form, in R2 , is 0, 0; in R3 , it is 0, 0, 0. Usually
the context makes is clear whether 0 is referring to a vector in the plane or in
space.
Our examples have illustrated key principles in vector algebra: how to add
and subtract vectors and how to mul ply vectors by a scalar. The following theorem states formally the proper es of these opera ons.
Notes:
548
Theorem 84
Proper es of Vector Opera ons
The following are true for all scalars c and d, and for all vectors u, v and
, where u, v and w
are all in R2 or where u, v and w
are all in R3 :
w
1. u + v = v + u
Commuta ve Property
= u + (v + w
)
2. (u + v) + w
Associa ve Property
3. v + 0 = v
4. (cd)v = c(dv)
Addi ve Iden ty
5. c(u + v) = cu + cv
Distribu ve Property
6. (c + d)v = cv + dv
7. 0 v = 0
8. || cv || = |c| || v ||
9. || u || = 0 if, and only if, u = 0.
As stated before, each vector v conveys magnitude and direc on informaon. We have a method of extrac ng the magnitude, which we write as || v ||.
Unit vectors are a way of extrac ng just the direc on informa on from a vector.
Deni on 54
Unit Vector
. of 1; that is,
A unit vector is a vector v with a magnitude
|| v || = 1.
Consider this scenario: you are given a vectorv and are told to create a vector
of length 10 in the direc on of v. How does one do that? If we knew that u was
the unit vector in the direc on of v, the answer would be easy: 10u. So how do
we nd u ?
Property 8 of Theorem 84 holds the key. If we divide v by its magnitude, it
becomes a vector of length 1. Consider:

1

1
1

v
(we can pull out
as it is a scalar)
|| v || = || v || || v ||
|| v ||
= 1.
Notes:
549
Chapter 10 Vectors
So the vector of length 10 in the direc on of v is 10
make this more clear.
1
v. An example will
|| v ||
. Example 323
Using Unit Vectors
= 1, 2, 2.
Let v = 3, 1 and let w
1. Find the unit vector in the direc on of v.
y
3
.
2. Find the unit vector in the direc on of w
3. Find the vector in the direc on of v with magnitude 5.
5u
1. We nd || v || =
u
x
Figure 10.25: Graphing vectors in Example 323. All vectors shown have their inial point at the origin.
10. So the unit vector u in the direc on of v is
3
1
1
u = v = ,
.
10
10
10
|| = 3, so the unit vectorz in the direc on of w

is
2. We nd || w
1
1 2 2
u = w
=
, ,
.
3
3 3 3
3. To create a vector with magnitude

in the direc
on of v, we mul ply the
5
unit vector u by 5. Thus 5u = 15/ 10, 5/ 10 is the vector we seek.

This is sketched in Figure 10.25.
The basic forma on of the unit vector u in the direc on of a vector v leads
to a interes ng equa on. It is:
v = || v ||
1
v.
|| v ||
We rewrite the equa on with parentheses to make a point:

)
(
1
v .
v = || v ||
|| v ||
| {z } | {z }
magnitude
direc on
This equa on illustrates the fact that a vector has both magnitude and direc on, where we view a unit vector as supplying only direc on informa on.
Iden fying unit vectors with direc on allows us to dene parallel vectors.
Notes:
550
Deni on 55
Parallel Vectors
1. Unit vectors u1 and u2 are parallel if u1 = u2 .

2. Nonzero vectors v1 and v2 are parallel if their respec ve unit vectors are parallel.
It is equivalent to say that vectors v1 and v2 are parallel if there is a scalar

c = 0 such that v1 = cv2 (see marginal note).
If one graphed all unit vectors in R2 with the ini al point at the origin, then
the terminal points would all lie on the unit circle. Based on what we know from
trigonometry, we can then say that the component form of all unit vectors in R2
is cos , sin for some angle .
A similar construc on in R3 shows that the terminal points all lie on the unit
sphere. These vectors also have a par cular component form, but its deriva on
is not as straigh orward as the one for unit vectors in R2 . Important concepts
about unit vectors are given in the following Key Idea.
Key Idea 49
Note: 0 is direc onless; because || 0 || =

0, there is no unit vector in the direc on
of 0.
Some texts dene two vectors as being
parallel if one is a scalar mul ple of the
other. By this deni on, 0 is parallel to
all vectors as 0 = 0v for all v.
We prefer the given deni on of parallel
as it is grounded in the fact that unit vectors provide direc on informa on. One
may adopt the conven on that 0 is parallel to all vectors if they desire. (See also
the marginal note on page 573.)
Unit Vectors
1. The unit vector in the direc on of v is

1
v.
|| v ||
u =
2. A vector u in R2 is a unit vector if, and only if, its component form
is cos , sin for some angle .
3. A vector u in R3 is a unit vector if, and only if, its component form
is sin cos , sin sin , cos for some angles and .
These formulas can come in handy in a variety of situa ons, especially the
formula for unit vectors in the plane.
. Example 324
.Finding Component Forces
Consider a weight of 50lb hanging from two chains, as shown in Figure 10.26.
One chain makes an angle of 30 with the ver cal, and the other an angle of
45 . Find the force applied to each chain.
S
30
45
.
50lb
Knowing that gravity is pulling the 50lb weight straight down,

Figure 10.26: A diagram of a weight hanging from 2 chains in Example 324.
Notes:
551
Chapter 10 Vectors
we can create a vector F to represent this force.
F1
F = 50 0, 1 = 0, 50 .
F2
120
45
Figure 10.27: A diagram of the force vectors from Example 324.
We can view each chain as pulling the weight up, preven ng it from falling.
We can represent the force from each chain with a vector. Let F1 represent the
force from the chain making an angle of 30 with the ver cal, and let F2 represent the force form the other chain. Convert all angles to be measured from the
horizontal (as shown in Figure 10.27), and apply Key Idea 49. As we do not yet
know the magnitudes of these vectors, (that is the problem at hand), we use m1
and m2 to represent them.
F1 = m1 cos 120 , sin 120
F2 = m2 cos 45 , sin 45
As the weight is not moving, we know the sum of the forces is 0. This gives:
F + F1 + F2 = 0
0, 50 + m1 cos 120 , sin 120 + m2 cos 45 , sin 45 = 0
The sum of the entries in the rst component is 0, and the sum of the entries
in the second component is also 0. This leads us to the following two equa ons:
m1 cos 120 + m2 cos 45 = 0
m1 sin 120 + m2 sin 45 = 50
This is a simple 2-equa on, 2-unkown system of linear equa ons. We leave it to
the reader to verify that the solu on is
50 2
25.88.
m1 = 50( 3 1) 36.6;
m2 =
1+ 3
It might seem odd that the sum of the forces applied to the chains is more
than 50lb. We leave it to a physics class to discuss the full details, but oer this
short explana on. Our equa ons were established so that the ver cal components of each force sums to 50lb, thus suppor ng the weight. Since the chains
are at an angle, they also pull against each other, crea ng an addi onal horizontal force while holding the weight in place. .
Unit vectors were very important in the previous calcula on; they allowed
us to dene a vector in the proper direc on but with an unknown magnitude.
Our computa ons were then computed componentwise. Because such calcula ons are o en necessary, the standard unit vectors can be useful.
Notes:
552
Deni on 56
Standard Unit Vectors
1. In R2 , the standard unit vectors are

i = 1, 0
and j = 0, 1 .
2. In R3 , the standard unit vectors are

i = 1, 0, 0
. Example 325
and j = 0, 1, 0
and k = 0, 0, 1 .
Using standard unit vectors
1. Rewrite v = 2, 3 using the standard unit vectors.

= 4i 5j + 2k in component form.
2. Rewrite w
S
1.
v = 2, 3
= 2, 0 + 0, 3
= 2 1, 0 3 0, 1
= 2i 3j
2.
= 4i 5j + 2k
w
= 4, 0, 0 + 0, 5, 0 + 0, 0, 2
= 4, 5, 2
These two examples demonstrate that conver ng from component form to/from
using the standard unit vectors is rather straigh orward. Many mathema cians
prefer component form, and it is the preferred nota on in this text. Many engineers prefer using the standard unit vectors, and many engineering text use
that nota on. .
. Example 326
.Finding Component Force
A weight of 25lb is suspended from a chain of length 2 while a wind pushes the
weight to the right with constant force of 5lb as shown in Figure 10.28. What
angle will the chain make with the ver cal as a result of the winds pushing?
How much higher will the weight be?
.
Fw
25lb
Figure 10.28: A gure of a weight being

pushed by the wind in Example 326.
Notes:
553
Chapter 10 Vectors
The force of the wind is represented by the vector Fw = 5i.
The force of gravity on the weight is represented by Fg = 25j. The direc on
and magnitude of the vector represen ng the force on the chain are both unknown. We represent this force with
S
Fc = m cos , sin = m cos i + m sin j

for some magnitude m and some angle with the horizontal . (Note: is the
angle the chain makes with the ver cal; is the angle with the horizontal.)
As the weight is at equilibrium, the sum of the forces is 0:
Fc + Fw + Fg = 0
m cos i + m sin j + 5i 25j = 0
Thus the sum of the i and j components are 0, leading us to the following
system of equa ons:
5 + m cos = 0
25 + m sin = 0
(10.1)
This is enough to determine Fc already, as we know m cos = 5 and

m sin = 25. Thus Fc = 5, 25 . We can use this to nd the magnitude
m:
m = (5)2 + 252 = 5 26.

We can then use either equality from Equa on (10.1) to solve for . We choose
the rst equality as using arccosine will return an angle in the 2nd quadrant:
(
)
5
1
5 + 5 26 cos = 0 = cos
1.7682 101.31 .
5 26
Subtrac ng 90 from this angle gives us an angle of 11.31 with the ver cal.
We can now use trigonometry to nd out how high the weight is li ed.
The diagram shows that a right triangle is formed with the 2 chain as the hypotenuse with an interior angle of 11.31 . The length of the adjacent side (in
the diagram, the dashed ver cal line) is 2 cos 11.31 1.96 . Thus the weight
is li ed by about 0.04 , almost 1/2in. .
The algebra we have applied to vectors is already demonstra ng itself to be
very useful. There are two more fundamental opera ons we can perform with
vectors, the dot product and the cross product. The next two sec ons explore
each in turn.
Notes:
554
Exercises 10.2
Terms and Concepts
In Exercises 12 15, sketch u, v, u + v and u v on the same

axes.
1. Name two dierent things that cannot be described with

just one number, but rather need 2 or more numbers to
fully describe them.
y
2. What is the dierence between (1, 2) and 1, 2?

u
3. What is a unit vector?
12.
4. What does it mean for two vectors to be parallel?

5. What eect does mul plying a vector by 2 have?
Problems
In Exercises 6 9, points P and Q are given. Write the vector

#
PQ in component form and using the standard unit vectors.
13.
6. P = (2, 1), Q = (3, 5)

7. P = (3, 2),
Q = (7, 2)
8. P = (0, 3, 1), Q = (6, 2, 5)

9. P = (2, 1, 2),
Q = (4, 3, 2)
10. Let u = 1, 2 and v = 1, 1.
14.
.
u
(a) Find u + v, u v, 2u 3v.
(b) Sketch the above vectors on the same axes, along

with u and v.
(c) Find x where u + x = 2v x.
11. Let u = 1, 1, 1 and v = 2, 1, 2.

u
(a) Find u + v, u v, u
2v.
15.
(b) Sketch the above vectors on the same axes, along

with u and v.
.
(c) Find x where u + x = v + 2x.
In Exercises 16 19, nd || u ||, ||v ||, || u +v || and || u v ||.

16. u = 2, 1, v = 3, 2
17. u = 3, 2, 2, v = 1, 1, 1
18. u = 1, 2, v = 3, 6
19. u = 2, 3, 6, v = 10, 15, 30
20. Under what condi ons is || u || + || v || = || u + v ||?
In Exercises 21 24, nd the unit vector u in the direc on of
v.
In Exercises 28 31, angles and are given. Find the force

applied to each chain.
28. = 30 , = 30
29. = 60 , = 60
30. = 20 , = 15
31. = 0 , = 0
A weight of 1lb is suspended from a chain of length while
a constant force of Fw pushes the weight to the right, making
an angle of with the ver cal, as shown in the gure below.
21. v = 3, 7
22. v = 6, 8
23. v = 1, 2, 2
24. v = 2, 2, 2
25. Find the unit vector in the rst quadrant of R2 that makes
a 50 angle with the x-axis.
26. Find the unit vector in the second quadrant of R2 that

makes a 30 angle with the y-axis.
Fw
p lb
27. Verify, from Key Idea 49, thatu = sin cos , sin sin , cos
is a unit vector for all angles and .
A weight of 100lb is suspended from two chains, making angles with the ver cal of and as shown in the gure below.
In Exercises 32 35, a force Fw and length are given. Find

the angle and the height the weight is li ed as it moves to
the right.
32. Fw = 1lb,
=1 ,
p = 1lb
33. Fw = 1lb,
=1 ,
p = 10lb
100lb
34. Fw = 1lb,
= 10 ,
p = 1lb
35. Fw = 10lb,
= 10 ,
p = 1lb
10.3 The Dot Product

The previous sec on introduced vectors and described how to add them together and how to mul ply them by scalars. This sec on introduces a mul plica on on vectors called the dot product.
Deni on 57
Dot Product
1. Let u = u1 , u2 and v = v1 , v2 in R2 . The dot product of u and

v, denoted u v, is
u v = u1 v1 + u2 v2 .
2. Let u = u1 , u2 , u3 and v = v1 , v2 , v3 in R3 . The dot product of
u and v, denoted u v, is
u v = u1 v1 + u2 v2 + u3 v3 .
Note how this product of vectors returns a scalar, not another vector. We
prac ce evalua ng a dot product in the following example, then we will discuss
why this product is useful.
. Example 327
Evalua ng dot products
1. Let u = 1, 2, v = 3, 1 in R2 . Find u v.
2. Let x = 2, 2, 5 and y = 1, 0, 3 in R3 . Find x y.
S
1. Using Deni on 57, we have

u v = 1(3) + 2(1) = 1.
2. Using the deni on, we have
x y = 2(1) 2(0) + 5(3) = 13.
.
Notes:
557
Chapter 10 Vectors
The dot product, as shown by the preceding example, is very simple to evaluate. It is only the sum of products. While the deni on gives no hint as to why
we would care about this opera on, there is an amazing connec on between
the dot product and angles formed by the vectors. Before sta ng this connecon, we give a theorem sta ng some of the proper es of the dot product.
Theorem 85
Proper es of the Dot Product
be vectors in R2 or R3 and let c be a scalar.

Let u, v and w
1. u v = v u
) = u v + u w
2. u (v + w
Commuta ve Property
3. c(u v) = (cu) v = u (cv)
4. 0 v = 0
5. v v = || v ||2
(a)
The last statement of the theorem makes a handy connec on between the
magnitude of a vector and the dot product with itself. Our deni on and theorem give proper es of the dot product, but we are s ll likely wondering What
does the dot product mean? It is helpful to understand that the dot product of
a vector with itself is connected to its magnitude.
.
y
The next theorem extends this understanding by connec ng the dot product
to magnitudes and angles. Given vectorsu andv in the plane, an angle is clearly
formed whenu andv are drawn with the same ini al point as illustrated in Figure
10.29 (a). (We always take to be the angle in [0, ] as two angles are actually
created.)
(b)
Figure 10.29: Illustra ng the angle
formed by two vectors with the same
ini al point.
The same is also true of 2 vectors in space: given u andv in R3 with the same
ini al point, there is a plane that contains both u and v. (When u and v are colinear, there are innite planes that contain both vectors.) In that plane, we can
again nd an angle between them (and again, 0 ). This is illustrated
in Figure 10.29 (b).
The following theorem connects this angle to the dot product of u and v.
Notes:
558
Theorem 86
The Dot Product and Angles
Let u and v be vectors in R2 or R3 . Then
u v = || u || || v || cos ,
where , 0 , is the angle between u and v.
When is an acute angle (i.e., 0 < /2), cos is posi ve; when =
/2, cos = 0; when is an obtuse angle (/2 < ), cos is nega ve.
Thus the sign of the dot product gives a general indica on of the angle between
the vectors, illustrated in Figure 10.30.
= /2
u v > 0
u v = 0
u v < 0
Figure 10.30: Illustra ng the rela onship between the angle between vectors and the
sign of their dot product.
We can use Theorem 86 to compute the dot product, but generally this theorem is used to nd the angle between known vectors (since the dot product is
generally easy to compute). To this end, we rewrite the theorems equa on as
cos =
u v
|| u |||| v ||
= cos1
u v
|| u |||| v ||
)
.
We prac ce using this theorem in the following example.

y
. Example 328
.Using the dot product to nd angles
= 4, 3, as shown in Figure 10.31. Find the
Let u = 3, 1, v = 2, 6 and w
angles , and .
6
4
We start by compu ng the magnitude of each vector.
|| u || = 10;
|| v || = 2 10;
|| = 5.
|| w
u
x
.
Notes:
Figure 10.31: Vectors used in Example

328.
559
Chapter 10 Vectors
We now apply Theorem 86 to nd the angles.
(
u v
( 10)(2 10)
= cos1 (0) = = 90 .
2
= cos1
v w
(2 10)(5)
(
)
26
1
= cos
10 10
0.6055 34.7 .
= cos1
u w
= cos
( 10)(5)
(
)
9
= cos1
5 10
2.1763 124.7
1
.
We see from our computa on that + = , as indicated by Figure 10.31.
While we knew this should be the case, it is nice to see that this non-intui ve
formula indeed returns the results we expected.
We do a similar example next in the context of vectors in space.
. Example 329
.Using the dot product to nd angles
= 5, 1, 4, as illustrated in Figure
Let u = 1, 1, 1, v = 1, 3, 2 and w
10.32. Find the angle between each pair of vectors.
z
4
2
.
x 5 2
1. Between u and v:
1
Figure 10.32: Vectors used in Example

329.
Notes:
560
u v
= cos
|| u |||| v ||
(
)
0
1

= cos
3 14
= .
2

:
2. Between u and w
u w
||
|| u |||| w
(
)
0
= cos1
3 42
= .
2
= cos1
:
3. Between v and w
)
v w
||
|| v |||| w
(
)
0
= cos1
14 42
= .
2
= cos1
While our work shows that each angle is /2, i.e., 90 , none of these angles
looks to be a right angle in Figure 10.32. Such is the case when drawing three
dimensional objects on the page. .
All three angles between these vectors was /2, or 90 . We know from
geometry and everyday life that 90 angles are nice for a variety of reasons,
so it should seem signicant that these angles are all /2. No ce the common
feature in each calcula on (and also the calcula on of in Example 328): the
dot products of each pair of angles was 0. We use this as a basis for a deni on
of the term orthogonal, which is essen ally synonymous to perpendicular.
Deni on 58
Orthogonal
Vectors u and v are orthogonal if their dot product is 0.

. Example 330
.Finding orthogonal vectors
Let u = 3, 5 and v = 1, 2, 3.
1. Find two vectors in R2 that are orthogonal to u.
2. Find two nonparallel vectors in R3 that are orthogonal to v.
Note: The term perpendicular originally

referred to lines. As mathema cs progressed, the concept of being at right
angles to was applied to other objects,
such as vectors and planes, and the term
orthogonal was introduced. It is especially used when discussing objects that
are hard, or impossible, to visualize: two
vectors in 5-dimensional space are orthogonal if their dot product is 0. It is not
wrong to say they are perpendicular, but
common conven on gives preference to
the word orthogonal.
Notes:
561
Chapter 10 Vectors
1. Recall that a line perpendicular to a line with slope m have slope 1/m,
the opposite reciprocal slope. We can think of the slope of u as 5/3, its
rise over run. A vector orthogonal to u will have slope 3/5. There are
many such choices, though all parallel:
5, 3
or
5, 3
or
10, 6
or
15, 9 , etc.
2. There are innite direc ons in space orthogonal to any given direc on,
so there are an innite number of nonparallel vectors orthogonal to v.
Since there are so many, we have great leeway in nding some.
One way is to arbitrarily pick values for the rst two components, leaving
the third unknown. For instance, letv1 = 2, 7, z. Ifv1 is to be orthogonal
to v, then v1 v = 0, so
2 + 14 + 3z = 0
z=
16
.
3
Sov1 = 2, 7, 16/3 is orthogonal tov. We can apply a similar technique

by leaving the rst or second component unknown.
v
Another method of nding a vector orthogonal to v mirrors what we did

in part 1. Letv2 = 2, 1, 0. Here we switched the rst two components
ofv, changing the sign of one of them (similar to the opposite reciprocal
concept before). Le ng the third component be 0 eec vely ignores the
third component of v, and it is easy to see that
.
(a)
v2 v = 2, 1, 0 1, 2, 3 = 0.
.
z
.
(b)
Figure 10.33: Developing the construcon of the orthogonal projec on.
Clearly v1 and v2 are not parallel.
An important construc on is illustrated in Figure 10.33, where vectors u and

v are sketched. In part (a), a do ed line is drawn from the p of u to the line
containing v, where the do ed line is orthogonal to v. In part (b), the do ed
is formed, parallel to v. It is clear by the
line is replaced with the vectorz and w
+ z. What is important about this construc on is this: u is
diagram that u = w
decomposed as the sum of two vectors, one of which is parallel tov and one that
is perpendicular to v. It is hard to overstate the importance of this construc on
(as well see in upcoming examples).
, z and u as shown in Figure 10.33 (b) form a right triangle,
The vectors w
in terms of v and
where the angle between v and u is labeled . We can nd w
u.
Using trigonometry, we can state that
|| = || u || cos .
|| w
Notes:
562
(10.2)
is parallel to to v ; that is, the direc on of w

is the
We also know that w
is the vector in
direc on of v, described by the unit vector || 1v ||v. The vector w
the direc on || 1v ||v with magnitude || u || cos :
(
) 1
= || u || cos
v.
w
|| v ||
Replace cos using Theorem 86:
(
= || u ||
u v
|| u |||| v ||
u v
v.
|| v ||2
u v
v.
v v
1
v
|| v ||
Now apply Theorem 85.
Since this construc on is so important, it is given a special name.

Deni on 59
Orthogonal Projec on
Let u and v be given. The orthogonal projec on of u onto v, denoted

proj v u, is
u v
v.
proj v u =
v v
. Example 331
.Compu ng the orthogonal projec on
1. Let u = 2, 1 and v = 3, 1. Find proj v u, and sketch all three vectors

with ini al points at the origin.
= 2, 1, 3 and x = 1, 1, 1. Find proj x w
, and sketch all three
2. Let w
vectors with ini al points at the origin.
Notes:
563
Chapter 10 Vectors
1. Applying Deni on 59, we have
y
u v
v
v v
5
=
3, 1
10
3 1
= ,
.
2 2
proj v u =
2
u
x
2 1
projv u 1
Vectors u, v and proj v u are sketched in Figure 10.34 (a). Note how the
projec on is parallel tov; that is, it lies on the same line through the origin
asv, although it points in the opposite direc on. That is because the angle
between u and v is obtuse (i.e., greater than 90 ).
.
(a)
z
2. Apply the deni on:

x
w
x
x x
6
= 1, 1, 1
3
= 2, 2, 2 .
=
proj x w
w
x
projx
w
y
2
x
.
(b)
z
.
2
projx
w x
x
Consider Figure 10.35 where the concept of the orthogonal projec on is

again illustrated. It is clear that
u = proj v u +z.
2
z = u proj v u.
Figure 10.34: Graphing the vectors used

in Example 331.
This leads us to rewrite Equa on (10.3) in a seemingly silly way:

u = proj v u + (u proj v u).
This is not nonsense, as pointed out in the following Key Idea. (Nota on note:
the expression y means is parallel to y. We can use this nota on to state
Notes:
projv u
Figure 10.35: Illustra ng the orthogonal

projec on.
564
(10.3)
As we know what u and proj v u are, we can solve forz and state that
(c)
These vectors are sketched in Figure 10.34 (b), and again in part (c) from
a dierent perspec ve. Because of the nature of graphing these vectors,
the sketch in part (b) makes it dicult to recognize that the drawn projecon has the geometric proper es it should. The graph shown in part (c)
illustrates these proper es be er.

x y which means x is parallel to y. The expression y means is orthogonal to y, and is used similarly.)
Key Idea 50
Orthogonal Decomposi on of Vectors
Let u and v be given. Then u can be wri en as the sum of two vectors,
one of which is parallel to v, and one of which is orthogonal to v:
u = proj v u + (u proj v u).

| {z }
| {z }
v
We illustrate the use of this equality in the following example.

. Example 332
.Orthogonal decomposi on of vectors
1. Let u = 2, 1 and v = 3, 1 as in Example 331. Decompose u as the

sum of a vector parallel to v and a vector orthogonal to v.
= 2, 1, 3 and x = 1, 1, 1 as in Example 331. Decompose w
as
2. Let w
the sum of a vector parallel to x and a vector orthogonal to x.
S
1. In Example 331, we found that proj v u = 1.5, 0.5. Let

z = u proj v u = 2, 1 1.5, 0.5 = 0.5, 1.5 .
Is z orthogonal to v? (I.e, is z v ?) We check for orthogonality with the
dot product:
z v = 0.5, 1.5 3, 1 = 0.
Since the dot product is 0, we knowz v. Thus:
u = proj v u + (u proj v u)
2, 1 = 1.5, 0.5 + 0.5, 1.5 .
|
{z
}
| {z }
v
= 2, 2, 2. Applying the Key Idea,

2. We found in Example 331 that proj x w
we have:
projx w
= 2, 1, 3 2, 2, 2 = 0, 1, 1 .
z = w
Notes:
565
Chapter 10 Vectors
We check to see ifz x:
z x = 0, 1, 1 1, 1, 1 = 0.
Since the dot product is 0, we know the two vectors are orthogonal. We
as the sum of two vectors, one parallel and one orthogonal
now write w
to x:
= proj x w
+ (
)
w
w proj x w
2, 1, 3 = 2, 2, 2 + 0, 1, 1
| {z }
| {z }
We give an example of where this decomposi on is useful.

. Example 333
.Orthogonally decomposing a force vector
Consider Figure 10.36 (a), showing a box weighing 50lb on a ramp that rises 5
over a span of 20 . Find the components of force, and their magnitudes, ac ng
on the box (as sketched in part (b) of the gure):
1. in the direc on of the ramp, and
2. orthogonal to the ramp.
As the ramp rises 5 over a horizontal distance of 20 , we can
represent the direc on of the ramp with the vector r = 20, 5. Gravity pulls
down with a force of 50lb, which we represent with g = 0, 50.
S
r
5
1. To nd the force of gravity in the direc on of the ramp, we compute projr g:
g r
r
r r
250
=
20, 5
425
200 50
,
11.76, 2.94 .
=
17
17
The magnitude of projr g is || projr g || = 50/ 17 12.13lb. Though

the box weighs 50lb, a force of about 12lb is enough to keep the box from
sliding down the ramp.
projr g =
20
(a)
r
5
z
2. To nd the component z of gravity orthogonal to the ramp, we use Key

Idea 50.
20
projr g
z = g projr g
200 800
=
,
11.76, 47.06 .
17
17
(b)
Figure 10.36: Sketching the ramp and box
in Example 333. Note: The vectors are not
drawn to scale.
Notes:
566

The magnitude of this force is ||z || 48.51lb. In physics and engineering,
knowing this force is important when compu ng things like sta c fric onal
force. (For instance, we could easily compute if the sta c fric onal force
alone was enough to keep the box from sliding down the ramp.)
.
Applica on to Work
In physics, the applica on of a force F to move an object in a straight line a
distance d produces work; the amount of work W is W = Fd, (where F is in the
direc on of travel). The orthogonal projec on allows us to compute work when
the force is not in the direc on of travel.
Consider Figure 10.37, where a force F is being applied to an object moving
in the direc on of d. (The distance the object travels is the magnitude of d.) The
work done is the amount of force in the direc on of d, || proj d F ||, mes || d ||:

F d

d || d ||
|| proj d F || || d || =
d d

F d

=
|| d || || d ||
|| d ||2

F d
=
|| d ||2
|| d ||2

= F d .
F
proj d F
Figure 10.37: Finding work when the

force and direc on of travel are given as
vectors.
The expression F d will be posi ve if the angle between F and d is acute;

when the angle is obtuse (hence F d is nega ve), the force is causing mo on
in the opposite direc on of d, resul ng in nega ve work. We want to capture
this sign, so we drop the absolute value and nd that W = F d.
Deni on 60
Work
Let F be a constant force that moves an object in a straight line from point
#
P to point Q. Let d = PQ. The work W done by F along d is W = F d.
. Example 334
.Compu ng work
A man slides a box along a ramp that rises 3 over a distance of 15 by applying
50lb of force as shown in Figure 10.38. Compute the work done.
Notes:
567
Chapter 10 Vectors
The gure indicates that the force applied makes a 30 angle with the horizontal, so F = 50 cos 30 , sin 30 43.3, 25 . The ramp is
represented by d = 15, 3. The work done is simply
S
F d = 50 cos 30 , sin 30 15, 3 724.5 lb.
30
.
15
Figure 10.38: Compu ng work when sliding a box up a ramp in Example 334.
Note how we did not actually compute the distance the object traveled, nor
the magnitude of the force in the direc on of travel; this is all inherently computed by the dot product! .
The dot product is a powerful way of evalua ng computa ons that depend
on angles without actually using angles. The next sec on explores another product on vectors, the cross product. Once again, angles play an important role,
though in a much dierent way.
Notes:
568
Exercises 10.3
Terms and Concepts
21. u = 1, 2, v = 1, 3
1. The dot product of two vectors is a
, not a vector.
2. How are the concepts of the dot product and vector magnitude related?
3. How can one quickly tell if the angle between two vectors
is acute or obtuse?
4. Give a synonym for orthogonal.
Problems
In Exercises 5 11, nd the dot product of the given vectors.
22. u = 5, 5, v = 1, 3
23. u = 3, 2, v = 1, 1
24. u = 3, 2, v = 2, 3
25. u = 1, 5, 1, v = 1, 2, 3
26. u = 3, 1, 2, v = 2, 2, 1
In Exercises 27 32, vectors u and v are given. Write u as the
sum of two vectors, one of which is parallel to v and one of
which is perpendicular to v. Note: these are the same pairs
of vectors as found in Exercises 21 26.
27. u = 1, 2, v = 1, 3
5. u = 2, 4, v = 3, 7
6. u = 5, 3, v = 6, 1
28. u = 5, 5, v = 1, 3
7. u = 1, 1, 2, v = 2, 5, 3
29. u = 3, 2, v = 1, 1
8. u = 3, 5, 1, v = 4, 1, 7
30. u = 3, 2, v = 2, 3
9. u = 1, 1, v = 1, 2, 3
31. u = 1, 5, 1, v = 1, 2, 3
10. u = 1, 2, 3, v = 0, 0, 0
32. u = 3, 1, 2, v = 2, 2, 1
in R and show that

11. Create your own vectors u, v and w
u (v + w
) = u v + u w
.
2
12. Create your own vectorsu andv in R3 and scalar c and show
that c(u v) = u (cv).
In Exercises 13 16, nd the measure of the angle between
the two vectors in both radians and degrees.
13. u = 1, 1, v = 1, 2
14. u = 2, 1, v = 3, 5
15. u = 8, 1, 4, v = 2, 2, 0
16. u = 1, 7, 2, v = 4, 2, 5
In Exercises 17 20, a vectorv is given. Give two vectors that
are orthogonal to v.
17. v = 4, 7
18. v = 3, 5
19. v = 1, 1, 1
20. v = 1, 2, 3
In Exercises 21 26, vectors u and v are given. Find projv u,
the orthogonal projec on of u onto v, and sketch all three
vectors on the same axes.
33. A 10lb box sits on a ramp that rises 4 over a distance of

20 . How much force is required to keep the box from sliding down the ramp?
34. A 10lb box sits on a 15 ramp that makes a 30 angle with
the horizontal. How much force is required to keep the box
from sliding down the ramp?
35. How much work is performed in moving a box horizontally
10 with a force of 20lb applied at an angle of 45 to the
horizontal?
36. How much work is performed in moving a box horizontally
10 with a force of 20lb applied at an angle of 10 to the
horizontal?
37. How much work is performed in moving a box up the length
of a ramp that rises 2 over a distance of 10 , with a force
of 50lb applied horizontally?
of a ramp that rises 2 over a distance of 10 , with a force
of 50lb applied at an angle of 45 to the horizontal?
of a 10 ramp that makes a 5 angle with the horizontal,
with 50lb of force applied in the direc on of the ramp?
Chapter 10 Vectors
10.4 The Cross Product

Orthogonality is immensely important. A quick scan of your current environment will undoubtedly reveal numerous surfaces and edges that are perpendicular to each other (including the edges of this page). The dot product provides
a quick test for orthogonality: vectors u and v are perpendicular if, and only if,
u v = 0.
Given two, nonparallel vectors u and v in space, it is very useful to nd a
that is perpendicular to both u and v. There is a opera on, called the
vector w
cross product, that creates such a vector. This sec on denes the cross product,
then explores its proper es and applica ons.
.
Deni on 61
Cross Product
Let u = u1 , u2 , u3 and v = v1 , v2 , v3 be vectors in R3 . The cross

product of u and v, denoted u v, is the vector
u v = u2 v3 u3 v2 , (u1 v3 u3 v1 ), u1 v2 u2 v1 .
This deni on can be a bit cumbersome to remember. A er an example we

will give a convenient method for compu ng the cross product. For now, careful
examina on of the products and dierences given in the deni on should reveal
a pa ern that is not too dicult to remember. (For instance, in the rst component only 2 and 3 appear as subscripts; in the second component, only 1 and 3
appear as subscripts. Further study reveals the order in which they appear.)
Lets prac ce using this deni on by compu ng a cross product.
. Example 335
.Compu ng a cross product
Let u = 2, 1, 4 and v = 3, 2, 5. Find u v, and verify that it is orthogonal
to both u and v.
Using Deni on 61, we have
(
)
u v = (1)5 (4)2, (2)5 (4)3 , (2)2 (1)3 = 13, 2, 7 .

S
(We encourage the reader to compute this product on their own, then verify
their result.)
We test whether or not u v is orthogonal to u and v using the dot product:
(
)
u v u = 13, 2, 7 2, 1, 4 = 0,
(
)
u v v = 13, 2, 7 3, 2, 5 = 0.
Notes:
570

As each of these dot products is zero, u v is indeed orthogonal to both u andv. .
A convenient method of compu ng the cross product starts with forming a
par cular 3 3 matrix, or rectangular array. The rst row comprises the standard unit vectorsi, j, and k. The second and third rows are the vectors u and v,
respec vely. Using u and v from Example 335, we begin with:
i j k
. 4
2 1
3 2 5
Now repeat the rst two columns a er the original three:
i
2
3
j k i j
1 4. 2 1
2 5 3 2
This gives three full upper le to lower right diagonals, and three full upper right to lower le diagonals, as shown. Compute the products along each
diagonal, then add the products on the right and subtract the products on the
le :
i
2
3
3k
j
1
2
k
4.
5
i
2
3
j
1
2
5i 12j 4k
8i 10j
(
) (
)
u v = 5i + 12j + 4k 3k + 8i + 10j = 13i + 2j + 7k = 13, 2, 7 .
We prac ce using this method.
. Example 336
.Compu ng a cross product
Let u = 1, 3, 6 and v = 1, 2, 1. Compute both u v and v u.
To compute u v, we form the matrix as prescribed above,
S
complete with repeated rst columns:
i
1
1
j
3
2
k
6
1
i
1
1
j
3
2
We let the reader compute the products of the diagonals; we give the result:
(
) (
)
u v = 3i 6j + 2k 3k + 12i +j = 9, 7, 5 .
Notes:
571
Chapter 10 Vectors
To computev u, we switch the second and third rows of the above matrix,
then mul ply along diagonals and subtract:
i
1
1
j
2
3
k
1
6
i
1
1
j
2
3
Note how with the rows being switched, the products that once appeared on
the right now appear on the le , and viceversa. Thus the result is:
(
) (
)
v u = 12i +j 3k 2k + 3i 6j = 9, 7, 5 ,
which is the opposite of u v. We leave it to the reader to verify that each of
these vectors is orthogonal to u and v. .
Proper es of the Cross Product

It is not coincidence that v u = (u v) in the preceding example; one
can show using Deni on 61 that this will always be the case. The following
theorem states several useful proper es of the cross product, each of which can
be veried by referring to the deni on.
.
Theorem 87
Proper es of the Cross Product
be vectors in R3 and let c be a scalar. The following iden

Let u, v and w
1. u v = (v u)
2.
= u w
+ v w
(a) (u + v) w
) = u v + u w
(b) u (v + w
3. c(u v) = (cu) v = u (cv)

4.
(a) (u v) u = 0
es hold:
An commuta ve Property
Distribu ve Proper es
Orthogonality Proper es
(b) (u v) v = 0
5. u u = 0
6. u 0 = 0
) = (u v) w
7. u (v w
Triple Scalar Product
We introduced the cross product as a way to nd a vector orthogonal to

two given vectors, but we did not give a proof that the construc on given in
Notes:
572

Deni on 61 sa ses this property. Theorem 87 asserts this property holds; we
leave it as a problem in the Exercise sec on to verify this.
Property 5 from the theorem is also le to the reader to prove in the Exercise
sec on, but it reveals something more interes ng than the cross product of a
vector with itself is 0. Let u andv be parallel vectors; that is, let there be a scalar
c such that v = cu. Consider their cross product:
u v = u (cu)
= c(u u)
= 0.
(by Property 3 of Theorem 87)

(by Property 5 of Theorem 87)
We have just shown that the cross product of parallel vectors is 0. This hints
at something deeper. Theorem 86 related the angle between two vectors and
their dot product; there is a similar rela onship rela ng the cross product of two
vectors and the angle between them, given by the following theorem.
Theorem 88
The Cross Product and Angles
Let u and v be vectors in R3 . Then
|| u v || = || u || || v || sin ,
where , 0 , is the angle between u and v.
Note: Deni on 58 (through Theorem

86) denes u and v to be orthogonal if
u v = 0. We could use Theorem 88 to
dene u andv are parallel if u v = 0. By
such a deni on, 0 would be both orthogonal and parallel to every vector. Apparent paradoxes such as this are not uncommon in mathema cs and can be very useful. (See also the marginal note on page
551.)
Note that this theorem makes a statement about the magnitude of the cross
product. When the angle between u andv is 0 or (i.e., the vectors are parallel),
the magnitude of the cross product is 0. The only vector with a magnitude of 0
is 0 (see Property 9 of Theorem 84), hence the cross product of nonzero parallel
vectors is 0.
We demonstrate the truth of this theorem in the following example.
. Example 337
.The cross product and angles
Let u = 1, 3, 6 and v = 1, 2, 1 as in Example 336. Verify Theorem 88 by
nding , the angle between u and v, and the magnitude of u v.
Notes:
573
Chapter 10 Vectors
S
We use Theorem 86 to nd the angle between u and v.

)
(
u v
= cos1
|| u || || v ||
(
)
11
1

= cos
46 6
0.8471 = 48.54 .
Our work in Example 336 showed thatu v = 9, 7, 5, hence ||u v || =

155. Is || u v || = || u || ||v || sin ? Using numerical approxima ons, we nd:

|| u v || = 155
|| u || || v || sin = 46 6 sin 0.8471
12.45.
12.45.
Numerically, they seem equal. Using a right triangle, one can show that
(
(
))
11
155
1

sin cos
= ,
46 6
46 6
which allows us to verify the theorem exactly. .
Right Hand Rule
z
u v
v
u
y
. Figure 10.39:
Illustra ng the Right Hand

Rule of the cross product.
The an commuta ve property of the cross product demonstrates that u v

andvu dier only by a sign these vectors have the same magnitude but point
in the opposite direc on. When seeking a vector perpendicular to u and v, we
essen ally have two direc ons to choose from, one in the direc on of u v and
one in the direc on of v u. Does it ma er which we choose? How can we tell
which one we will get without graphing, etc.?
Another wonderful property of the cross product, as dened, is that it follows the right hand rule. Given u and v in R3 with the same ini al point, point
the index nger of your right hand in the direc on of u and let your middle nger
point in the direc on ofv (much as we did when establishing the right hand rule
for the 3-dimensional coordinate system). Your thumb will naturally extend in
the direc on of u v. One can prac ce this using Figure 10.39. If you switch,
and point the index nder in the direc on of v and the middle nger in the direc on of u, your thumb will now point in the opposite direc on, allowing you
to visualize the an commuta ve property of the cross product.
Applica ons of the Cross Product

There are a number of ways in which the cross product is useful in mathema cs, physics and other areas of science beyond just nding a vector perpendicular to two others. We highlight a few here.
Notes:
574

Area of a Parallelogram
It is a standard geometry fact that the area of a parallelogram is A = bh,
where b is the length of the base and h is the height of the parallelogram, as
illustrated in Figure 10.40 (a). As shown when dening the Parallelogram Law of
vector addi on, two vectors u and v dene a parallelogram when drawn from
the same ini al point, as illustrated in Figure 10.40 (b). Trigonometry tells us
that h = || u || sin , hence the area of the parallelogram is
A = || u || || v || sin = || u v ||,
.
b
(a)
(10.4)
where the second equality comes from Theorem 88. We illustrate using Equaon (10.4) in the following example.
. Example 338
Finding the area of a parallelogram
1. Find the area of the parallelogram dened by the vectors u = 2, 1 and

v = 1, 3.
2. Verify that the points A = (1, 1, 1), B = (2, 3, 2), C = (4, 5, 3) and
D = (3, 3, 2) are the ver ces of a parallelogram. Find the area of the
parallelogram.
(b)
Figure 10.40: Using the cross product to
nd the area of a parallelogram.
1. Figure 10.41 (a) sketches the parallelogram dened by the vectors u and
v. We have a slight problem in that our vectors exist in R2 , not R3 , and
the cross product is only dened on vectors in R3 . We skirt this issue by
viewingu andv as vectors in the xy plane of R3 , and rewrite them asu =
2, 1, 0 and v = 1, 3, 0. We can now compute the cross product. It is
easy to show thatu v = 0, 0, 5; therefore the area of the parallelogram
is A = || u v || = 5.
2. To show that the quadrilateral ABCD is a parallelogram (shown in Figure
10.41 (b)), we need to show that the opposite sides are parallel. We can
# #
# #
quickly show that AB = DC = 1, 2, 1 and BC = AD = 2, 2, 1. We nd
#
#
the area by compu ng the magnitude of the cross product of AB and BC:
# #
# #
AB BC = 0, 1, 2 || AB BC || = 5 2.236.
5
4
3
2
1
u
x
(a)
z
C
B
D
This applica on is perhaps more useful in nding the area of a triangle (in
short, triangles are used more o en than parallelograms). We illustrate this in
the following example.
.
x 4
Notes:
(b)
Figure 10.41: Sketching the parallelograms in Example 338.
575
Chapter 10 Vectors
. Example 339
Area of a triangle
Find the area of the triangle with ver ces A = (1, 2), B = (2, 3) and C = (3, 1),
as pictured in Figure 10.42.
y
B
x
1
Figure 10.42: Finding the area of a triangle in Example 339.
Note: The word parallelepiped is pronounced parallelehpipe-ed.
1 # #
1
1
3
|| AB AC || = || 1, 1, 0 2, 1, 0 || = || 0, 0, 3 || = .
2
2
2
2
We arrive at the same answer as before with less work. .
Volume of a Parallelepiped
The three dimensional analogue to the parallelogram is the parallelepiped.
Each face is parallel to the face opposite face, as illustrated in Figure 10.43. By
, one gets a vector whose magnitude is the area of the base.
crossing v and w
Do ng this vector with u computes the volume of parallelepiped! (Up to a sign;
take the absolute value.)
is
Thus the volume of a parallelepiped dened by vectors u, v and w
We found the area of this triangle in Example 200 to be 1.5

S
using integra on. There we discussed the fact that nding the area of a triangle
can be inconvenient using the 12 bh formula as one has to compute the height,
which generally involves nding angles, etc. Using a cross product is much more
direct.
We can choose any two sides of the triangle to use to form vectors; we
#
#
choose AB = 1, 1 and AC = 2, 1. As in the previous example, we will
rewrite these vectors with a third component of 0 so that we can apply the cross
product. The area of the triangle is
)|.
V = |u (v w
.
Figure 10.43: A parallelepiped is the three
dimensional analogue to the parallelogram.
Note how this is the Triple Scalar Product, rst seen in Theorem 87. Applying
the iden es given in the theorem shows that we can apply the Triple Scalar
Product in any order we choose to nd the volume. That is,
)| = |u (
|,
V = |u (v w
w v)| = |(u v) w
0.5
)| = | 1, 1, 0 1, 1, 1 | = 2.
|u (v w
So the volume of the parallelepiped is 2 cubic units. .

2
.
Figure 10.44: A parallelepiped in Example
340.
Notes:
576
= 1, 1, 1.
We apply Equa on (10.5). We rst nd v w
Then
.
etc.
. Example 340
Finding the volume of parallelepiped
Find the volume of the parallepiped dened by the vectors u = 1, 1, 0, v =
= 0, 1, 1.
1, 1, 0 and w
x 1
(10.5)

While this applica on of the Triple Scalar Product is interes ng, it is not used
all that o en: parallelepipeds are not a common shape in physics and engineering. The last applica on of the cross product is very applicable in engineering.
Torque
Torque is a measure of the turning force applied to an object. A classic scenario involving torque is the applica on of a wrench to a bolt. When a force is
applied to the wrench, the bolt turns. When we represent the force and wrench
with vectors F and , we see that the bolt moves (because of the threads) in a direc on orthogonal to F and . Torque is usually represented by the Greek le er
, or tau, and has units of Nm, a Newtonmeter, or lb, a footpound.
While a full understanding of torque is beyond the purposes of this book,
when a force F is applied to a lever arm , the resul ng torque is
= F.
(10.6)
. Example 341
.Compu ng torque
A lever of length 2 makes an angle with the horizontal of 45 . Find the resul ng
torque when a force of 10lb is applied to the end of the level where:
1. the force is perpendicular to the lever, and
F
F
90
60
2. the force makes an angle of 60 with the lever, as shown in Figure 10.45.
Figure 10.45: Showing a force being applied to a lever in Example 341.
1. We start by determining vectors for the force and lever arm. Since the
lever arm makes a 45 angle with the
and is 2 long, we can
horizontal
state that = 2 cos 45 , sin 45 =
2, 2 .
Since the force vector is perpendicular to the lever arm (as seen in the
le hand side of Figure 10.45), we can conclude it is making an angle of
45 with the horizontal. As it has
of 10lb, we can state
a magnitude

F = 10 cos 45 , sin 45 = 5 2, 5 2 .
Using Equa on (10.6) to nd the torque requires a cross product. We
again let the third component of each vector be 0 and compute the cross
product:
= F

=
2, 2, 0 5 2, 5 2, 0
= 0, 0, 20
Notes:
577
Chapter 10 Vectors
This clearly has a magnitude of 20 -lb.
We can view the force and lever arm vectors as lying on the page; our
computa on of shows that the torque goes into the page. This follows
the Right Hand Rule of the cross product, and it also matches well with the
example of the wrench turning the bolt. Turning a bolt clockwise moves
it in.

2. Our lever arm can s ll be represented by =
2, 2 . As our force
vector makes a 60 angle with , we can see (referencing the right hand
side of the gure) that F makes a 15 angle with the horizontal. Thus

+ 3) 5(1 + 3)
F = 10 cos 15 , sin 15 = 5(1
,
2
2
9.659, 2.588 .
We again make the third component 0 and take the cross product to nd
the torque:
= F
5(1 + 3) 5(1 + 3)
=
2, 2, 0
,
,0
2
2
= 0, 0, 10 3
0, 0, 17.321 .
As one might expect, when the force and lever arm vectors are orthogonal, the magnitude of force is greater than when the vectors are not orthogonal.
While the cross product has a variety of applica ons (as noted in this chapter), its fundamental use is nding a vector perpendicular to two others. Knowing a vector orthogonal to two others is of incredible importance, as it allows
us to nd the equa ons of lines and planes in a variety of contexts. The importance of the cross product, in some sense, relies on the importance of lines and
planes, which see widespread use throughout engineering, physics and mathema cs. We study lines and planes in the next two sec ons.
Notes:
578
Exercises 10.4
Terms and Concepts
1. The cross product of two vectors is a
scalar.
24. u = 2, 0, v = 0, 3
, not a
2. One can visualize the direc on of u v using the

.
In Exercises 25 28, nd the area of the triangle with the

given ver ces.
25. Ver ces: (0, 0, 0), (1, 3, 1) and (2, 1, 1).
26. Ver ces: (5, 2, 1), (3, 6, 2) and (1, 0, 4).
3. Give a synonym for orthogonal.
27. Ver ces: (1, 1), (1, 3) and (2, 2).
4. T/F: A fundamental principle of the cross product is that

u v is orthogonal to u and v.
28. Ver ces: (3, 1), (1, 2) and (4, 3).
5.
is a measure of the turning force applied to an

object.
In Exercises 29 30, nd the area of the quadrilateral with

the given ver ces. (Hint: break the quadrilateral into 2 triangles.)
29. Ver ces: (0, 0), (1, 2), (3, 0) and (4, 3).
Problems
30. Ver ces: (0, 0, 0), (2, 1, 1), (1, 2, 8) and (1, 1, 5).
In Exercises 6 14, vectors u and v are given. Compute u v

and show this is orthogonal to both u and v.
In Exercises 31 32, nd the volume of the parallelepiped

dened by the given vectors.
6. u = 3, 2, 2, v = 0, 1, 5
31. u = 1, 1, 1, v = 1, 2, 3,
7. u = 5, 4, 3, v = 2, 5, 1
32. u = 1, 2, 1, v = 2, 2, 1,
8. u = 4, 5, 5, v = 3, 3, 4
9. u = 4, 7, 10, v = 4, 4, 1
= 1, 0, 1
w
= 3, 1, 3
w
In Exercises 33 36, nd a unit vector orthogonal to both u

and v.
10. u = 1, 0, 1, v = 5, 0, 7
33. u = 1, 1, 1, v = 2, 0, 1
11. u = 1, 5, 4, v = 2, 10, 8
34. u = 1, 2, 1, v = 3, 2, 1
12. u = i, v = j
35. u = 5, 0, 2, v = 3, 0, 7
13. u = i, v = k
36. u = 1, 2, 1, v = 2, 4, 2
14. u = j, v = k
37. A bicycle rider applies 150lb of force, straight down,

onto a pedal that extends 7in horizontally from the
cranksha . Find the magnitude of the torque applied to
the cranksha .
in R3 and show thatu(v+

15. Pick any vectorsu,v and w
w) =
u v + u w
.
in R3 and show thatu (v
16. Pick any vectorsu,v and w
w) =
.
(u v) w
In Exercises 17 20, the magnitudes of vectors u and v in R3
are given, along with the angle between them. Use this informa on to nd the magnitude of u v.
17. || u || = 2, || v || = 5,
= 30
18. || u || = 3, || v || = 7,
= /2
19. || u || = 3, || v || = 4,
20. || u || = 2, || v || = 5,
= 5/6
In Exercises 21 24, nd the area of the parallelogram dened by the given vectors.
21. u = 1, 1, 2, v = 2, 0, 3
22. u = 2, 1, 5, v = 1, 3, 1
23. u = 1, 2, v = 2, 1
38. A bicycle rider applies 150lb of force, straight down, onto

a pedal that extends 7in from the cranksha , making a 30
angle with the horizontal. Find the magnitude of the torque
applied to the cranksha .
39. To turn a stubborn bolt, 80lb of force is applied to a 10in
wrench. What is the maximum amount of torque that can
be applied to the bolt?
40. To turn a stubborn bolt, 80lb of force is applied to a 10in
wrench in a conned space, where the direc on of applied force makes a 10 angle with the wrench. How much
torque is subsequently applied to the wrench?
41. Show, using the deni on of the Cross Product, that u (u
v) = 0; that is, that u is orthogonal to the cross product of
u and v.
42. Show, using the deni on of the Cross Product, thatuu =
0.
Chapter 10 Vectors
10.5 Lines
To nd the equa on of a line in the x y plane, we need two pieces of informaon: a point and the slope. The slope conveys direc on informa on. As ver cal
lines have an undened slope, the following statement is more accurate:
To dene a line, one needs a point on the line and the direc on of
the line.
z
d
This holds true for lines in space.
p + d
p
y
x
p 2d
Figure 10.47: Dening a line in space.
Let P be a point in space, let p be the vector with ini al point at the origin
and terminal point at P (i.e., p points to P), and let d be a vector. Consider the
points on the line through P in the direc on of d.
Clearly one point on the line is P; we can say that the vector p lies at this
point on the line. To nd another point on the line, we can start at p and move
in a direc on parallel to d. For instance, star ng at p and traveling one length of
d places one at another point on the line. Consider Figure 10.47 where certain
points along the line are indicated.
The gure illustrates how every point on the line can be obtained by star ng
with p and moving a certain distance in the direc on of d. That is, we can dene
the line as a func on of t:
(t) = p + t d.
(10.7)
In many ways, this is not a new concept. Compare Equa on (10.7) to the
familiar y = mx + b equa on of a line:
Star ng
Point
Direc on
y = b .+ m x
(t) = p + t d
How Far To
Go In That
Direc on
Figure 10.46: Understanding the vector equa on of a line.
The equa ons exhibit the same structure: they give a star ng point, dene
a direc on, and state how far in that direc on to travel.
Equa on (10.7) is an example of a vectorvalued func on; the input of the
func on is a real number and the output is a vector. We will cover vectorvalued
func ons extensively in the next chapter.
Notes:
580
10.5 Lines
There are other ways to represent a line. Let p = x0 , y0 , z0 and let d =
a, b, c. Then the equa on of the line through p in the direc on of d is:
(t) = p + td
= x0 , y0 , z0 + t a, b, c
= x0 + at, y0 + bt, z0 + ct
The last line states the the x values of the line are given by x = x0 + at, the
y values are given by y = y0 + bt, and the z values are given by z = z0 + ct.
These three equa ons, taken together, are the parametric equa ons of the line
through p in the direc on of d.
Finally, each of the equa ons for x, y and z above contain the variable t. We
can solve for t in each equa on:
x = x0 + at
y = y0 + bt
z = z0 + ct
x x0
,
a
y y0
t=
,
b
z z0
t=
,
c
t=
assuming a, b, c = 0. Since t is equal to each expression on the right, we can set

these equal to each other, forming the symmetric equa ons of the line through
p in the direc on of d:
x x0
y y0
z z0
=
=
.
a
b
c
Each representa on has its own advantages, depending on the context. We
summarize these three forms in the following deni on, then give examples of
their use.
Notes:
581
Chapter 10 Vectors
.
Deni on 62
Equa ons of Lines in Space
Consider the line in space that passes through p = x0 , y0 , z0 in the

direc on of d = a, b, c .
1. The vector equa on of the line is
(t) = p + td.
2. The parametric equa ons of the line are
x = x0 + at,
y = y0 + bt,
z = z0 + ct.
3. The symmetric equa ons of the line are

y y0
z z0
x x0
=
=
.
a
b
c
. Example 342
.Finding the equa on of a line
Give all three equa ons, as given in Deni on 62, of the line through P = (2, 3, 1)
in the direc on of d = 1, 1, 2. Does the point Q = (1, 6, 6) lie on this line?
z
5
Q
d
P
x 2
We iden fy the point P = (2, 3, 1) with the vector p =

S
2, 3, 1. Following the deni on, we have
the vector equa on of the line is (t) = 2, 3, 1 + t 1, 1, 2;
the parametric equa ons of the line are

x = 2 t,
.
Figure 10.48: Graphing a line in Example
342.
y=3+t
z = 1 + 2t; and
the symmetric equa ons of the line are

x2
y3
z1
=
=
.
1
1
2
The rst two equa ons of the line are useful when a t value is given: one
can immediately nd the corresponding point on the line. These forms are good
when calcula ng with a computer; most so ware programs easily handle equaons in these formats. (For instance, to make Figure 10.48, a certain graphics
program was given the input (2-x,3+x,1+2*x). This par cular program requires the variable always be x instead of t).
Notes:
582
10.5 Lines
Does the point Q = (1, 6, 6) lie on the line? The graph in Figure 10.48
makes it clear that it does not. We can answer this ques on without the graph
using any of the three equa on forms. Of the three, the symmetric equa ons
are probably best suited for this task. Simply plug in the values of x, y and z and
see if equality is maintained:
1 2 ? 6 3 ? 6 1
=
=
3 = 3 = 2.5.
1
1
2
We see that Q does not lie on the line as it did not sa sfy the symmetric equaons. .
. Example 343
Finding the equa on of a line through two points
Find the parametric equa ons of the line through the points P = (2, 1, 2) and
Q = (1, 3, 1).
z
Recall the statement made at the beginning of this sec on:

to nd the equa on of a line, we need a point and a direc on. We have two
points; either one will suce. The direc on of the line can be found by the
#
vector with ini al point P and terminal point Q: PQ = 1, 4, 3.
#
The parametric equa ons of the line through P in the direc on of PQ are:
S
x=2t
4
P
4
x
y = 1 + 4t z = 2 3t.
A graph of the points and line are given in Figure 10.49. Note how in the
given parametriza on of the line, t = 0 corresponds to the point P, and t = 1
corresponds to the point Q. This relates to the understanding of the vector equaon of a line described in Figure 10.46. The parametric equa ons start at the
#
point P, and t determines how far in the direc on of PQ to travel. When t = 0,
#
#
we travel 0 lengths of PQ; when t = 1, we travel one length of PQ, resul ng in
the point Q. .
2
#
2 PQ
.4
.
Figure 10.49: A graph of the line in Example 343.
Parallel, Intersec ng and Skew Lines

In the plane, two dis nct lines can either be parallel or they will intersect
at exactly one point. In space, given equa ons of two lines, it can some mes
be dicult to tell whether the lines are dis nct or not (i.e., the same line can be
represented in dierent ways). Given lines 1 (t) = p1 +td1 and 2 (t) = p2 +td2 ,
we have four possibili es: 1 and 2 are
the same line
intersec ng lines
parallel lines
skew lines
they share all points;

share only 1 point;
d1 d2 , no points in common; or
d1 d2 , no points in common.
Notes:
583
Chapter 10 Vectors
The next two examples inves gate these possibili es.
. Example 344
Comparing lines
Consider lines 1 and 2 , given in parametric equa on form:
x
1 : y
z
= 1 + 3t
= 2t
=
t
x =
2 : y =
z =
2 + 4s
3+s
5 + 2s.
Determine whether 1 and 2 are the same line, intersect, are parallel, or skew.
z
P2
d2
5
1
x
5
5
d1
P1
1 + 3t = 2 + 4s
2t =
3+s
t
= 5 + 2s.
. Figure 10.50: Sketching the lines from Example 344.
We start by looking at the direc ons of each line. Line 1

S
has the direc on given by d1 = 3, 1, 1 and line 2 has the direc on given by
d2 = 4, 1, 2. It should be clear that d1 and d2 are not parallel, hence 1 and 2
are not the same line, nor are they parallel. Figure 10.50 veries this fact (where
the points and direc ons indicated by the equa ons of each line are iden ed).
We next check to see if they intersect (if they do not, they are skew lines).
To nd if they intersect, we look for t and s values such that the respec ve x, y
and z values are the same. That is, we want s and t such that:
This is a rela vely simple system of linear equa ons. Since the last equa on is
already solved for t, subs tute that value of t into the equa on above it:
2 (5 + 2s) = 3 + s
s = 2, t = 1.
A key to remember is that we have three equa ons; we need to check if s =

2, t = 1 sa ses the rst equa on as well:
1 + 3(1) = 2 + 4(2).
It does not. Therefore, we conclude that the lines 1 and 2 are skew. .
. Example 345
.Comparing lines
Consider lines 1 and 2 , given in parametric equa on form:
x
1 : y
z
= 0.7 + 1.6t
= 4.2 + 2.72t
= 2.3 3.36t
x
2 : y
z
=
2.8 2.9s
= 10.15 4.93s
= 5.05 + 6.09s.
Determine whether 1 and 2 are the same line, intersect, are parallel, or skew.
Notes:
584
10.5 Lines
It is obviously very dicult to simply look at these equa ons
S
and discern anything. This is done inten onally. In the real world, most equaons that are used do not have nice, integer coecients. Rather, there are lots
of digits a er the decimal and the equa ons can look messy.
We again start by deciding whether or not each line has the same direc on.
The direc on of 1 is given by d1 = 1.6, 2.72, 3.36 and the direc on of 2
is given by d2 = 2.9, 4.93, 6.09. When it is not clear through observa on
whether two vectors are parallel or not, the standard way of determining this is
by comparing their respec ve unit vectors. Using a calculator, we nd:
u1 =
d1
= 0.3471, 0.5901, 0.7289
|| d1 ||
u2 =
d2
= 0.3471, 0.5901, 0.7289 .
|| d2 ||
The two vectors seem to be parallel (at least, their components are equal to
4 decimal places). In most situa ons, it would suce to conclude that the lines
are at least parallel, if not the same. One way to be sure is to rewrite d1 and d2
in terms of frac ons, not decimals. We have
d1 = 16 , 272 , 336
d2 = 29 , 493 , 609 .
10 100 100
10 100 100
One can then nd the magnitudes of each vector in terms of frac ons, then
compute the unit vectors likewise. A er a lot of manual arithme c (or a er
briey using a computer algebra system), one nds that

10 17
21
10
21
17
u1 =
u2 =
,
,
.
,
,
83
83
830
830
830
830
We can now say without equivoca on that these lines are parallel.
Are they the same line? The parametric equa ons for a line describe one
point that lies on the line, so we know that the point P1 = (0.7, 4.2, 2.3) lies
on 1 . To determine if this point also lies on 2 , plug in the x, y and z values of P1
into the symmetric equa ons for 2 :
(0.7) 2.8 ? (4.2) 10.15 ? (2.3) (5.05)
=
=
2.9
4.93
6.09
z
5
P1
d2
x 4
1.2069 = 1.2069 = 1.2069.
The point P1 lies on both lines, so we conclude they are the same line, just
parametrized dierently. Figure 10.51 graphs this line along with the points and
vectors described by the parametric equa ons. Note how d1 and d2 are parallel,
though point in opposite direc ons (as indicated by their unit vectors above). .
2
5
d1
10
P2
.
Figure 10.51: Graphing the lines in Example 345.
Notes:
585
Chapter 10 Vectors
Distances
Given a point Q and a line (t) = p + td in space, it is o en useful to know
the distance from the point to the line. (Here we use the standard deni on
of distance, i.e., the length of the shortest line segment from the point to the
line.) Iden fying p with the point P, Figure 10.52 will help establish a general
method of compu ng this distance h.
#
From trigonometry, we know h = || PQ || sin . We have a similar iden ty
#
#
involving the cross product: || PQ d || = || PQ || || d || sin . Divide both sides
of this la er equa on by || d || to obtain h:
#
PQ
h=
Figure 10.52: Establishing the distance

from a point to a line.
#
|| PQ d ||
.
|| d ||
(10.8)
It is also useful to determine the distance between lines, which we dene as

the length of the shortest line segment that connects the two lines (an argument
from geometry shows that this line segments is perpendicular to both lines). Let
lines 1 (t) = p1 + td1 and 2 (t) = p2 + td2 be given, as shown in Figure 10.53.
To nd the direc on orthogonal to both d1 and d2 , we take the cross product:
#
c = d1 d2 . The magnitude of the orthogonal projec on of P1 P2 onto c is the
distance h we seek:
P2
d2
#
h = || proj c P1 P2 ||
#

P1 P2 c

c
=
c c
#
|P1 P2 c|
|| c ||
=
|| c ||2
#
|P1 P2 c|
=
.
|| c ||
#
P1 P2
d1
P1
Figure 10.53: Establishing the distance

between lines.
A problem in the Exercise sec on is to show that this distance is 0 when the lines
#
#
intersect. Note the use of the Triple Scalar Product: P1 P2 c = P1 P2 (d1 d2 ).
The following Key Idea restates these two distance formulas.
Notes:
586
10.5 Lines
Key Idea 51
Distances to Lines
1. Let P be a point on a line that is parallel to d. The distance h from

a point Q to the line is:
#
|| PQ d ||
h=
.
|| d ||
2. Let P1 be a point on line 1 that is parallel to d1 , and let P2 be a

point on line 2 parallel to d2 , and let c = d1 d2 , where lines 1
and 2 are not parallel. The distance h between the two lines is:
#
|P1 P2 c|
.
h=
|| c ||
. Example 346
Finding the distance from a point to a line
Find the distance from the point Q = (1, 1, 3) to the line (t) = 1, 1, 1 +
t 2, 3, 1 .
The equa on of the line line gives us the point P = (1, 1, 1)
S
#
that lies on the line, hence PQ = 0, 2, 2. The equa on also gives d = 2, 3, 1.
Following Key Idea 51, we have the distance as
#
|| PQ d ||
h=
|| d ||
|| 4, 4, 4 ||
=
14
4 3
1.852.
14
The point Q is approximately 1.852 units from the line (t). .
. Example 347
.Finding the distance between lines
Find the distance between the lines
x
1 : y
z
= 1 + 3t
= 2t
=
t
x
2 : y
z
= 2 + 4s
=
3+s
= 5 + 2s.
Notes:
587
Chapter 10 Vectors
These are the sames lines as given in Example 344, where
S
we showed them to be skew. The equa ons allow us to iden fy the following
points and vectors:
P1 = (1, 2, 0)
d1 = 3, 1, 1
P2 = (2, 3, 5)
d2 = 4, 1, 2
#
P1 P2 = 3, 1, 5 .
c = d1 d2 = 3, 2, 7 .
From Key Idea 51 we have the distance h between the two lines is
#
|P1 P2 c|
|| c ||
42
= 5.334.
62
h=
The lines are approximately 5.334 units apart. .

One of the key points to understand from this sec on is this: to describe a
line, we need a point and a direc on. Whenever a problem is posed concerning a line, one needs to take whatever informa on is oered and glean point
and direc on informa on. Many ques ons can be asked (and are asked in the
Exercise sec on) whose answer immediately follows from this understanding.
Notes:
588
Exercises 10.5
Terms and Concepts
17. 1 (t) = 3, 4, 1 + t 2, 3, 4,
2 (t) = 3, 3, 3 + t 3, 2, 4.
1. To nd an equa on of a line, what two pieces of informaon are needed?
18. 1 (t) = 1, 1, 1 + t 3, 1, 3,
2 (t) = 7, 3, 7 + t 6, 2, 6.
x = 3t
x = 1 + 2t
19. 1 = y = 3 2t and 2 = y = 3 + 5t
z = 2 + 7t
z=t
x = 3.11 + 3.4t
x = 1.1 + 0.6t
20. 1 = y = 3.77 + 0.9t and 2 = y = 2 + 5.1t
z = 2.5 + 8.5t
z = 2.3 + 1.5t
x = 0.2 + 0.6t
x = 0.86 + 9.2t
21. 1 = y = 1.33 0.45t and 2 = y = 0.835 6.9t
z = 4.2 + 1.05t
z = 3.045 + 16.1t
x = 4 2.1t
x = 0.1 + 1.1t
22. 1 = y = 2.9 1.5t and 2 = y = 1.8 + 7.2t
z = 3.2 + 1.6t
z = 3.1 + 1.1t
2. Two dis nct lines in the plane can intersect or be

.
3. Two dis nct lines in space can intersect, be
.
or be
4. Use your own words to describe what it means for two lines
in space to be skew.
Problems
In Exercises 5 14, write the vector, parametric and symmetric equa ons of the lines described.
5. Passes through P = (2, 4, 1), parallel to d = 9, 2, 5.
6. Passes through P = (6, 1, 7), parallel to d = 3, 2, 5.
7. Passes through P = (2, 1, 5) and Q = (7, 2, 4).
8. Passes through P = (1, 2, 3) and Q = (5, 5, 5).
9. Passes through P = (0, 1, 2) and orthogonal to both
d1 = 2, 1, 7 and d2 = 7, 1, 3.
10. Passes through P = (5, 1, 9) and orthogonal to both
d1 = 1, 0, 1 and d2 = 2, 0, 3.
11. Passes through the point of intersec on of 1 (t) and 2 (t)
and orthogonal to both lines, where
1 (t) = 2, 1, 1 + t 5, 1, 2 and
2 (t) = 2, 1, 2 + t 3, 1, 1.
12. Passes through the point of intersec on of 1 (t) and 2 (t)
and orthogonal
to both lines, where
x = t
x = 2+t
1 = y = 2 + 2t and 2 = y = 2 t .
z=1+t
z = 3 + 2t
13. Passes through P = (1, 1), parallel to d = 2, 3.
14. Passes through P = (2, 5), parallel to d = 0, 1.
In Exercises 15 22, determine if the described lines are the
same line, parallel lines, intersec ng or skew lines. If intersec ng, give the point of intersec on.
15. 1 (t) = 1, 2, 1 + t 2, 1, 1,
2 (t) = 3, 3, 3 + t 4, 2, 2.
16. 1 (t) = 2, 1, 1 + t 5, 1, 3,
2 (t) = 14, 5, 9 + t 1, 1, 1.
In Exercises 23 26, nd the distance from the point to the

line.
23. P = (1, 1, 1),
(t) = 2, 1, 3 + t 2, 1, 2
24. P = (2, 5, 6),
(t) = 1, 1, 1 + t 1, 0, 1
25. P = (0, 3),
(t) = 2, 0 + t 1, 1
26. P = (1, 1),
(t) = 4, 5 + t 4, 3
In Exercises 27 28, nd the distance between the two lines.

27. 1 (t) = 1, 2, 1 + t 2, 1, 1,
2 (t) = 3, 3, 3 + t 4, 2, 2.
28. 1 (t) = 0, 0, 1 + t 1, 0, 0,
2 (t) = 0, 0, 3 + t 0, 1, 0.
Exercises 29 31 explore special cases of the distance formulas found in Key Idea 51.
29. Let Q be a point on the line (t). Show why the distance
formula correctly gives the distance from the point to the
line as 0.
30. Let lines 1 (t) and 2 (t) be intersec ng lines. Show why
the distance formula correctly gives the distance between
these lines as 0.
31. Let lines 1 (t) and 2 (t) be parallel. Show why the distance
formula cannot be used as stated to nd the distance be#
tween the lines, then show why le ng c = (P1 P2 d2 )d2
allows one to the use the given formula.
Chapter 10 Vectors
10.6 Planes
Figure 10.54: Illustra ng dening a plane

with a sheet of cardboard and a nail.
Any at surface, such as a wall, table top or s piece of cardboard can be

thought of as represen ng part of a plane. Consider a piece of cardboard with
a point P marked on it. One can take a nail and s ck it into the cardboard at P
such that the nail is perpendicular to the cardboard; see Figure 10.54
This nail provides a handle for the cardboard. Moving the cardboard around
moves P to dierent loca ons in space. Til ng the nail (but keeping P xed) lts
the cardboard. Both moving and l ng the cardboard denes a dierent plane
in space. In fact, we can dene a plane by: 1) the loca on of P in space, and 2)
the direc on of the nail.
The previous sec on showed that one can dene a line given a point on the
line and the direc on of the line (usually given by a vector). One can make a
similar statement about planes: we can dene a plane in space given a point on
the plane and the direc on the plane faces (using the descrip on above, the
direc on of the nail). Once again, the direc on informa on will be supplied by
a vector, called a normal vector, that is orthogonal to the plane.
What exactly does orthogonal to the plane mean? Choose any two points P
#
and Q in the plane, and consider the vector PQ. We say a vector n is orthogonal
#
to the plane if n is perpendicular to PQ for all choices of P and Q; that is, if
#
n PQ = 0 for all P and Q.
This gives us way of wri ng an equa on describing the plane. Let P =
(x0 , y0 , z0 ) be a point in the plane and let n = a, b, c be a normal vector to
the plane. A point Q = (x, y, z) lies in the plane dened by P and n if, and only
#
#
if, PQ is orthogonal to n. Knowing PQ = x x0 , y y0 , z z0 , consider:
#
PQ n = 0
x x0 , y y0 , z z0 a, b, c = 0
a(x x0 ) + b(y y0 ) + c(z z0 ) = 0
(10.9)
Equa on (10.9) denes an implicit func on describing the plane. More algebra
produces:
ax + by + cz = ax0 + by0 + cz0 .
The right hand side is just a number, so we replace it with d:
ax + by + cz = d.
(10.10)
As long as c = 0, we can solve for z:

z=
Notes:
590
1
(d ax by).
c
(10.11)
10.6 Planes
Equa on (10.11) is especially useful as many computer programs can graph funcons in this form. Equa ons (10.9) and (10.10) have specic names, given next.
Deni on 63
Equa ons of a Plane in Standard and General Forms
The plane passing through the point P = (x0 , y0 , z0 ) with normal vector
n = a, b, c can be described by an equa on with standard form
a(x x0 ) + b(y y0 ) + c(z z0 ) = 0;

the equa ons general form is
ax + by + cz = d.
A key to remember throughout this sec on is this: to nd the equa on of a

plane, we need a point and a normal vector. We will give several examples of
nding the equa on of a plane, and in each one dierent types of informa on
are given. In each case, we need to use the given informa on to nd a point on
the plane and a normal vector.
. Example 348
Finding the equa on of a plane.
Write the equa on of the plane that passes through the points P = (1, 1, 0),
Q = (1, 2, 1) and R = (0, 1, 2) in standard form.
We need a vector n that is orthogonal to the plane. Since P,
S
#
# # #
Q and R are in the plane, so are the vectors PQ and PR; PQ PR is orthogonal
#
#
to PQ and PR and hence the plane itself.
# #
It is straigh orward to compute n = PQ PR = 2, 1, 1. We can use any
point we wish in the plane (any of P, Q or R will do) and we arbitrarily choose P.
Following Deni on 63, the equa on of the plane in standard form is
2(x 1) + (y 1) + z = 0.
z
2 R
4
x 4
#
#
PQ PR
.
Figure 10.55: Sketching the plane in Example 348.
The plane is sketched in Figure 10.55. .

We have just demonstrated the fact that any three non-collinear points dene a plane. (This is why a three-legged stool does not rock; its three feet
always lie in a plane. A four-legged stool will rock unless all four feet lie in the
same plane.)
. Example 349
.Finding the equa on of a plane.
Verify that lines 1 and 2 , whose parametric equa ons are given below, inter-
Notes:
591
Chapter 10 Vectors
sect, then give the equa on of the plane that contains these two lines in general
form.
x = 5 + 2s
x = 2 + 3t
1+s
1 : y =
2 : y = 1 2t
z = 4 + 2s
z = 1+t
.
The lines clearly are not parallel. If they do not intersect,

S
they are skew, meaning there is not a plane that contains them both. If they do
intersect, there is such a plane.
To nd their point of intersec on, we set the x, y and z equa ons equal to
each other and solve for s and t:
z
2
5
x 5
5 + 2s
1+s
4 + 2s
5
y
.
Figure 10.56: Sketching the plane in Example 349.
= 2 + 3t
= 1 2t
= 1+t
s = 2,
t = 1.
When s = 2 and t = 1, the lines intersect at the point P = (1, 3, 0).

Let d1 = 2, 1, 2 and d2 = 3, 2, 1 be the direc ons of lines 1 and 2 ,
respec vely. A normal vector to the plane containing these the two lines will
also be orthogonal to d1 and d2 . Thus we nd a normal vector n by compu ng
n = d1 d2 = 5, 4 7.
We can pick any point in the plane with which to write our equa on; each
line gives us innite choices of points. We choose P, the point of intersec on.
We follow Deni on 63 to write the planes equa on in general form:
5(x + 1) + 4(y 3) 7z = 0
5x + 5 + 4y 12 7z = 0
5x + 4y 7z = 7.
The planes equa on in general form is 5x + 4y 7z = 7; it is sketched in Figure
10.56. .
. Example 350
.Finding the equa on of a plane
Give the equa on, in standard form, of the plane that passes through the point
P = (1, 0, 1) and is orthogonal to the line with vector equa on (t) = 1, 0, 1+
t 1, 2, 2.
2
P
2
x
As the plane is to be orthogonal to the line, the plane must

S
be orthogonal to the direc on of the line given by d = 1, 2, 2. We use this as
our normal vector. Thus the planes equa on, in standard form, is
Figure 10.57: The line and plane in Example 350.
(x + 1) + 2y + 2(z 1) = 0.
Notes:
592
10.6 Planes
The line and plane are sketched in Figure 10.57. .
. Example 351
Finding the intersec on of two planes
Give the parametric equa ons of the line that is the intersec on of the planes
p1 and p2 , where:
p1 : x (y 2) + (z 1) = 0
p2 : 2(x 2) + (y + 1) + (z 3) = 0
To nd an equa on of a line, we need a point on the line and
S
the direc on of the line.
We can nd a point on the line by solving each equa on of the planes for z:
p1 : z = x + y 1
p2 : z = 2x y 2
We can now set these two equa ons equal to each other (i.e., we are nding
values of x and y where the planes have the same z value:
x + y 1 = 2x y 2
2y = 3x 1
1
y = (3x 1)
2
We can choose any value for x; we choose x = 1. This determines that y = 1.
We can now use the equa ons of either plane to nd z: when x = 1 and y = 1,
z = 1 on both planes. We have found a point P on the line: P = (1, 1, 1).
We now need the direc on of the line. Since the line lies in each plane,
its direc on is orthogonal to a normal vector for each plane. Considering the
equa ons for p1 and p2 , we can quickly determine a normal vector. For p1 , n1 =
1, 1, 1 and for p2 , n2 = 2, 1, 1 . A direc on orthogonal to both of these
direc ons is their cross product: d = n1 n2 = 2, 3, 1 .
The parametric equa ons of the line through P = (1, 1, 1) in the direc on
of d = 2, 3, 1 is:
:
2
x
2 5
z
2
P
.
Figure 10.58: Graphing the planes and
their line of intersec on in Example 351.
x = 2t + 1 y = 3t + 1 z = t 1.
The planes and line are graphed in Figure 10.58. .

. Example 352
.Finding the intersec on of a plane and a line
Find the point of intersec on, if any, of the line (t) = 3, 3, 1 + t 1, 2, 1
and the plane with equa on in general form 2x + y + z = 4.
Notes:
593
Chapter 10 Vectors
z
2
(t)
x 4
4
y
The equa on of the plane shows that the vectorn = 2, 1, 1

S
is a normal vector to the plane, and the equa on of the line shows that the line
moves parallel to d = 1, 2, 1. Since these are not orthogonal, we know
there is a point of intersec on. (If there were orthogonal, it would mean that
the plane and line were parallel to each other, either never intersec ng or the
line was in the plane itself.)
To nd the point of intersec on, we need to nd a t value such that (t)
sa ses the equa on of the plane. Rewri ng the equa on of the line with parametric equa ons will help:
x = 3t
(t) = y = 3 + 2t .
z = 1 + t
Replacing x, y and z in the equa on of the plane with the expressions containing
t found in the equa on of the line allows us to determine a t value that indicates
the point of intersec on:
2x + y + z = 4
2(3 t) + (3 + 2t) + (1 + t) = 4
Figure 10.59: Illustra ng the intersec on

of a line and a plane in Example 352.
t = 2.
When t = 2, the point on the line sa ses the equa on of the plane; that point
is (2) = 1, 1, 1. Thus the point (1, 1, 1) is the point of intersec on between
the plane and the line, illustrated in Figure 10.59. .
Distances
Q
h
n
P
.
Figure 10.60: Illustra ng nding the distance from a point to a plane.
Just as it was useful to nd distances between points and lines in the previous
sec on, it is also o en necessary to nd the distance from a point to a plane.
Consider Figure 10.60, where a plane with normal vector n is sketched containing a point P and a point Q, not on the plane, is given. We measure the
#
distance from Q to the plane by measuring the length of the projec on of PQ
onto n. That is, we want:

#
#
n PQ
#

|n PQ|
|| proj n PQ || =
n
(10.12)
=

|| n ||2
|| n ||
Equa on (10.12) is important as it does more than just give the distance between
a point and a plane. We will see how it allows us to nd several other distances
as well: the distance between parallel planes and the distance from a line and a
plane. Because Equa on (10.12) is important, we restate it as a Key Idea.
Notes:
594
10.6 Planes
Key Idea 52
Distance from a Point to a Plane
Let a plane with normal vector n be given, and let Q be a point. The
distance h from Q to the plane is
#
|n PQ|
h=
,
|| n ||
where P is any point in the plane.
. Example 353
Distance between a point and a plane
Find the distance bewteen the point Q = (2, 1, 4) and the plane with equa on
2x 5y + 6z = 9.
Using the equa on of the plane, we nd the normal vector
S
n = 2, 5, 6. To nd a point on the plane, we can let x and y be anything we
choose, then let z be whatever sa ses the equa on. Le ng x and y be 0 seems
#
simple; this makes z = 1.5. Thus we let P = 0, 0, 1.5, and PQ = 2, 1, 2.5 .
The distance h from Q to the plane is given by Key Idea 52:
#
|n PQ|
|| n ||
| 2, 5, 6 2, 1, 2.5 |
=
|| 2, 5, 6 ||
| 16|
=
65
1.98.
h=
We can use Key Idea 52 to nd other distances. Given two parallel planes,
we can nd the distance between these planes by le ng P be a point on one
plane and Q a point on the other. If is a line parallel to a plane, we can use
the Key Idea to nd the distance between them as well: again, let P be a point
in the plane and let Q be any point on the line. (One can also use Key Idea 51.)
The Exercise sec on contains problems of these types.
These past two sec ons have not explored lines and planes in space as an
exercise of mathema cal curiosity. Rather, there are many, many applica ons
of these fundamental concepts. Complex shapes can be modeled (or, approximated) using planes. For instance, part of the exterior of an aircra may have
a complex, yet smooth, shape, and engineers will want to know how air ows
across this piece as well as how heat might build up due to air fric on. Many
Notes:
595
Chapter 10 Vectors
equa ons that help determine air ow and heat dissipa on are dicult to apply
to arbitrary surfaces, but simple to apply to planes. By approxima ng a surface
with millions of small planes one can more readily model the needed behavior.
Notes:
596
Exercises 10.6
Terms and Concepts
1. In order to nd the equa on of a plane, what two pieces of
informa on must one have?
2. What is the rela onship between a plane and one of its normal vectors?
17. Contains the point (5, 7, 3) and is orthogonal to the line

(t) = 4, 5, 6 + t 1, 1, 1.
18. Contains
the point (4, 1, 1) and is orthogonal to the line
x = 4 + 4t
(t) = y = 1 + 1t
z = 1 + 1t
Problems
19. Contains the point (4, 7, 2) and is parallel to the plane

3(x 2) + 8(y + 1) 10z = 0.
In Exercises 3 6, give any two points in the given plane.
20. Contains the point (1, 2, 3) and is parallel to the plane

x = 5.
3. 2x 4y + 7z = 2
4. 3(x + 2) + 5(y 9) 4z = 0
5. x = 2
In Exercises 21 22, give the equa on of the line that is the

intersec on of the given planes.
6. 4(y + 2) (z 6) = 0
21. p1 : 3(x 2) + (y 1) + 4z = 0, and

p2 : 2(x 1) 2(y + 3) + 6(z 1) = 0.
In Exercises 7 20, give the equa on of the described plane

in standard and general forms.
22. p1 : 5(x 5) + 2(y + 2) + 4(z 1) = 0, and

p2 : 3x 4(y 1) + 2(z 1) = 0.
7. Passes through (2, 3, 4) and has normal vector

n = 3, 1, 7.
In Exercises 23 26, nd the point of intersec on between

the line and the plane.
8. Passes through (1, 3, 5) and has normal vector

n = 0, 2, 4.
23. line: 5, 1, 1 + t 2, 2, 1,
plane: 5x y z = 3
9. Passes through the points (1, 2, 3), (3, 1, 4) and (1, 0, 1).
24. line: 4, 1, 0 + t 1, 0, 1,
plane: 3x + y 2z = 8
10. Passes through the points (5, 3, 8), (6, 4, 9) and (3, 3, 3).
11. Contains the intersec ng lines
1 (t) = 2, 1, 2 + t 1, 2, 3 and
2 (t) = 2, 1, 2 + t 2, 5, 4.
12. Contains the intersec ng lines
1 (t) = 5, 0, 3 + t 1, 1, 1 and
2 (t) = 1, 4, 7 + t 3, 0, 3.
13. Contains the parallel lines
1 (t) = 1, 1, 1 + t 1, 2, 3 and
2 (t) = 1, 1, 2 + t 1, 2, 3.
25. line: 1, 2, 3 + t 3, 5, 1,
plane: 3x 2y z = 4
26. line: 1, 2, 3 + t 3, 5, 1,
plane: 3x 2y z = 4
In Exercises 27 30, nd the given distances.
27. The distance from the point (1, 2, 3) to the plane
3(x 1) + (y 2) + 5(z 2) = 0.
28. The distance from the point (2, 6, 2) to the plane
2(x 1) y + 4(z + 1) = 0.
14. Contains the parallel lines

1 (t) = 1, 1, 1 + t 4, 1, 3 and
2 (t) = 2, 2, 2 + t 4, 1, 3.
29. The distance between the parallel planes

x + y + z = 0 and
(x 2) + (y 3) + (z + 4) = 0
15. Contains
the point (2, 6, 1) and the line
x = 2 + 5t
(t) = y = 2 + 2t
z = 1 + 2t
30. The distance between the parallel planes

2(x 1) + 2(y + 1) + (z 2) = 0 and
2(x 3) + 2(y 1) + (z 3) = 0
16. Contains
the point (5, 7, 3) and the line
x = t
(t) = y = t
z=t
31. Show why if the point Q lies in a plane, then the distance
formula correctly gives the distance from the point to the
plane as 0.
32. How is Exercise 30 in Sec on 10.5 easier to answer once we
have an understanding of planes?
11: V
11.1 VectorValued Func ons

We are very familiar with real valued func ons, that is, func ons whose output
is a real number. This sec on introduces vectorvalued func ons func ons
whose output is a vector.
Deni on 64
VectorValued Func ons

y
A vectorvalued func on is a func on of the form

r(t) = f(t), g(t)
or r(t) = f(t), g(t), h(t) ,
where f, g and h are real valued func ons.
3
2
r(2)
The domain ofr is the set of all values of t for whichr(t) is dened. The
range ofr is the set of all possible output vectorsr(t).
2
3
Evalua ng and Graphing VectorValued Func ons
(a)
y
Evalua ng a vectorvalued func on at a specic value of t is straigh orward;

simply evaluate
each component func on at that value of t. For instance, if
r(t) = t2 , t2 + t 1 , then r(2) = 4, 1. We can sketch this vector, as is

done in Figure 11.1 (a). Plo ng lots of vectors is cumbersome, though, so generally we do not sketch the whole vector but just the terminal point. The graph
of a vectorvalued func on is the set of all terminal points of r(t), where the
ini al point of each vector is always the origin. In Figure 11.1 (b) we sketch the
graph ofr ; we can indicate individual points on the graph with their respec ve
vector, as shown.
Vectorvalued func ons are closely( related to) parametric
equa )ons of graphs.
(
While in both methods we plot points x(t), y(t) or x(t), y(t), z(t) to produce
a graph, in the context of vectorvalued func ons each such point represents a
vector. The implica ons of this will be more fully realized in the next sec on as
we apply calculus ideas to these func ons.
3
2
r(2)
x
1
2
3
(b)
Figure 11.1: Sketching the graph of a
vectorvalued func on.
Chapter 11 Vector Valued Func ons
t t
2
1
0
1
2
6
0
0
0
6
. Example 354
Graphingvectorvalued func ons
1
3
Graphr(t) = t t, 2
, for 2 t 2. Sketchr(1) andr(2).
t +1
1
t2 + 1
1/5
1/2
1
1/2
1/5
We start by making a table of t, x and y values as shown in

S
Figure 11.2 (a). Plo ng these points gives an indica on of what the graph looks
like. In Figure 11.2 (b), we indicate these points and sketch the full graph. We
also highlightr(1) andr(2) on the graph. .
(a)
. Example 355
Graphing vectorvalued func ons.
Graphr(t) = cos t, sin t, t for 0 t 4.
y
1
r(1)
0.5
. 6
r(2)
x
(b)
We can again plot points, but careful considera on of this

S
func on is very revealing. Momentarily ignoring the third component, we see
the x and y components trace out a circle of radius 1 centered at the origin.
No cing that the z component is t, we see that as the graph winds around the
z-axis, it is also increasing at a constant rate in the posi ve z direc on, forming a
spiral. This is graphed in Figure 11.3. In the graphr(7/4) (0.707, 0.707, 5.498)
is highlighted to help us understand the graph. .
Algebra of VectorValued Func ons
Figure 11.2: Sketching the vectorvalued

func on of Example 354.
Deni on 65
Opera ons on VectorValued Func ons
Let r1 (t) = f1 (t), g1 (t) and r2 (t) = f2 (t), g2 (t) be vectorvalued

func ons in R2 and let c be a scalar. Then:
1. r1 (t) r2 (t) = f1 (t) f2 (t), g1 (t) g2 (t) .
10
r(7/4)
5
1
x
2. cr1 (t) = cf1 (t), cg1 (t) .
A similar deni on holds for vectorvalued func ons in R3 .

1
1
Figure 11.3: The graph ofr(t) in Example

355.
This deni on states that we add, subtract and scale vector-valued func ons
componentwise. Combining vectorvalued func ons in this way can be very
useful (as well as create interes ng graphs).
. Example 356
.Adding and scaling vectorvalued func ons.
Let r1 (t) = 0.2t, 0.3t , r2 (t) = cos t, sin t and r(t) = r1 (t) + r2 (t). Graph
r1 (t),r2 (t),r(t) and 5r(t) on 10 t 10.
Notes:
600

We can graphr1 andr2 easily by plo ng points (or just using
technology). Lets think about each for a moment to be er understand how
vectorvalued func ons work.
We can rewrite r1 (t) = 0.2t, 0.3t as r1 (t) = t 0.2, 0.3. That is, the
func on r1 scales the vector 0.2, 0.3 by t. This scaling of a vector produces a
line in the direc on of 0.2, 0.3.
We are familiar withr2 (t) = cos t, sin t ; it traces out a circle, centered at
the origin, of radius 1. Figure 11.5 (a) graphsr1 (t) andr2 (t).
Adding r1 (t) to r2 (t) produces r(t) = cos t + 0.2t, sin t + 0.3t , graphed
in Figure 11.5 (b). The linear movement of the line combines with the circle to
create loops that move in the direc on of 0.2, 0.3. (We encourage the reader
to experiment by changingr1 (t) to 2t, 3t, etc., and observe the eects on the
loops.)
Mul plying r(t) by 5 scales the func on by 5, producing 5r(t) = 5 cos t +
1, 5 sin t + 1.5, which is graphed in Figure 11.5 (c) along with r(t). The new
func on is 5 mes bigger than r(t). Note how the graph of 5r(t) in (c) looks
iden cal to the graph ofr(t) in (b). This is due to the fact that the x and y bounds
of the plot in (c) are exactly 5 mes larger than the bounds in (b). .
. Example 357
.Adding and scaling vectorvalued func ons.
A cycloid is a graph traced by a point p on a rolling circle, as shown in Figure
11.4. Find an equa on describing the cycloid, where the circle has radius 1.
x
4
10
20
(a)
y
4
x
4
2
2
(b)
y
20
.
10
Figure 11.4: Tracing a cycloid.
This problem is not very dicult if we approach it in a clever

way. We start by le ng p(t) describe the posi on of the point p on the circle,
where the circle is centered at the origin and only rotates clockwise (i.e., it does
not roll). This is rela vely simple given our previous experiences with parametric
equa ons; p(t) = cos t, sin t.
We now want the circle to roll. We represent this by le ng c(t) represent
the loca on of the center of the circle. It should be clear that the y component
of c(t) should be 1; the center of the circle is always going to be 1 if it rolls on a
horizontal surface.
The x component ofc(t) is a linear func on of t: f(t) = mt for some scalar m.
When t = 0, f(t) = 0 (the circle starts centered on the y-axis). When t = 2,
the circle has made one complete revolu on, traveling a distance equal to its
20
10
10
20
(c)
Figure 11.5: Graphing the func ons in Example 356.
Notes:
601

circumference, which is also 2. This gives us a point on our line f(t) = mt, the
point (2, 2). It should be clear that m = 1 and f(t) = t. So c(t) = t, 1.
We now combinep andc together to form the equa on of the cycloid: r(t) =
p(t) + c(t) = cos t + t, sin t + 1, which is graphed in Figure 11.6. .
10
Displacement
5
x
5
15
10
A vectorvalued func onr(t) is o en used to describe the posi on of a moving object at me t. At t = t0 , the object is at r(t0 ); at t = t1 , the object is at
r(t1 ). Knowing the loca onsr(t0 ) andr(t1 ) give no indica on of the path taken
between them, but o en we only care about the dierence of the loca ons,
r(t1 ) r(t0 ), the displacement.
Figure 11.6: The cycloid in Example 357.
Deni on 66
Displacement
Let r(t) be a vectorvalued func on and let t0 < t1 be values in the

domain. The displacement d ofr, from t = t0 to t = t1 , is
d = r(t1 ) r(t0 ).
When the displacement vector is drawn with ini al point atr(t0 ), its terminal
point isr(t1 ). We think of it as the vector which points from a star ng posi on
to an ending posi on.
y
1
. Example 358
Finding and graphing displacement vectors
Letr(t) = cos( 2 t), sin( 2 t) . Graphr(t) on 1 t 1, and nd the displacement ofr(t) on this interval.
d
x
1
Figure 11.7: Graphing the displacement

of a posi on func on in Example 358.
The func on r(t) traces out the unit circle, though at a difS
ferent rate than the usual cos t, sin t parametriza on. At t0 = 1, we have
r(t0 ) = 0, 1; at t1 = 1, we haver(t1 ) = 0, 1. The displacement ofr(t) on
[1, 1] is thus d = 0, 1 0, 1 = 0, 2 .
A graph ofr(t) on [1, 1] is given in Figure 11.7, along with the displacement
vector d on this interval. .
Measuring displacement makes us contemplate related, yet very dierent,
concepts. Considering the semicircular path the object in Example 358 took,
we can quickly verify that the object ended up a distance of 2 units from its ini al
loca on. That is, we can compute || d || = 2. However, measuring distance from
the star ng point is dierent from measuring distance traveled. Being a semi
Notes:
602

circle, we can measure the distance traveled by this object as 3.14 units.
Knowing distance from the star ng point allows us to compute average rate of
change.
Deni on 67
Average Rate of Change
Letr(t) be a vectorvalued func on, where each of its component funcons is con nuous on its domain, and let t0 < t1 . The average rate of
change ofr(t) on [t0 , t1 ] is
average rate of change =
r(t1 ) r(t0 )
.
t1 t0
. Example 359
Averagerate of change
Letr(t) = cos( 2 t), sin( 2 t) as in Example 358. Find the average rate of change
ofr(t) on [1, 1] and on [1, 5].
We computed in Example 358 that the displacement ofr(t)
S
on [1, 1] was d = 0, 2. Thus the average rate of change ofr(t) on [1, 1] is:
r(1) r(1)
0, 2
=
= 0, 1 .
1 (1)
2
We interpret this as follows: the object followed a semicircular path, meaning
it moved towards the right then moved back to the le , while climbing slowly,
then quickly, then slowly again. On average, however, it progressed straight up
at a constant rate of 0, 1 per unit of me.
We can quickly see that the displacement on [1, 5] is the same as on [1, 1],
so d = 0, 2. The average rate of change is dierent, though:
r(5) r(1)
0, 2
=
= 0, 1/3 .
5 (1)
6
As it took 3 mes as long to arrive at the same place, this average rate of
change on [1, 5] is 1/3 the average rate of change on [1, 1]. .
We considered average rates of change in Sec ons 1.1 and 2.1 as we studied
limits and deriva ves. The same is true here; in the following sec on we apply
calculus concepts to vectorvalued func ons as we nd limits, deriva ves, and
integrals. Understanding the average rate of change will give us an understanding of the deriva ve; displacement gives us an understanding of integra on.
Notes:
603
Exercises 11.1
Terms and Concepts
1. Vectorvalued func ons are closely related to
of graphs.
2. When sketching vectorvalued func ons, technically one
.
isnt graphing points, but rather
3. It can be useful to think of
as a vector that points
from a star ng posi on to an ending posi on.
Problems
In Exercises 4 11, sketch the vectorvalued func on on the
given interval.
4. r(t) = t2 , t2 1 , for 2 t 2.
5. r(t) = t2 , t3 , for 2 t 2.
6. r(t) = 1/t, 1/t2 , for 2 t 2.

1 2
7. r(t) = 10
t , sin t , for 2 t 2.
1 2
8. r(t) = 10
t , sin t , for 2 t 2.
9. r(t) = 3 sin(t), 2 cos(t), on [0, 2].
10. r(t) = 3 cos t, 2 sin(2t), on [0, 2].
11. r(t) = 2 sec t, tan t, on [, ].
In Exercises 12 15, sketch the vectorvalued func on on the
given interval in R3 . Technology may be useful in crea ng the
sketch.
16. r(t) = t, t2 .
17. r(t) = 5 cos t, 3 sin t.
18. r(t) = 2 cos t, 2 sin t, t.
19. r(t) = cos t, t, t2 .

In Exercises 20 27, create a vectorvalued func on whose
graph matches the given descrip on.
20. A circle of radius 2, centered at (1, 2), traced counter
clockwise once on [0, 2].
21. A circle of radius 3, centered at (5, 5), traced clockwise
once on [0, 2].
22. An ellipse, centered at (0, 0) with ver cal major axis of
length 10 and minor axis of length 3, traced once counter
clockwise on [0, 2].
23. An ellipse, centered at (3, 2) with horizontal major axis of
length 6 and minor axis of length 4, traced once clockwise
on [0, 2].
24. A line through (2, 3) with a slope of 5.
25. A line through (1, 5) with a slope of 1/2.
26. A ver cally oriented helix with radius of 2 that starts at
(2, 0, 0) and ends at (2, 0, 4) a er 1 revolu on on [0, 2].
27. A ver cally oriented helix with radius of 3 that starts at
(3, 0, 0) and ends at (3, 0, 3) a er 2 revolu ons on [0, 1].
13. r(t) = 3 cos t, sin t, t/ on [0, 2].
In Exercises 28 31, nd the average rate of change ofr(t) on

the given interval.
28. r(t) = t, t2 on [2, 2].
14. r(t) = cos t, sin t, sin t on [0, 2].
29. r(t) = t, t + sin t on [0, 2].
15. r(t) = cos t, sin t, sin(2t) on [0, 2].
30. r(t) = 3 cos t, 2 sin t, t on [0, 2].
31. r(t) = t, t2 , t3 on [1, 3].
12. r(t) = 2 cos t, t, 2 sin t, on [0, 2].
In Exercises 16 19, nd ||r(t) ||.
11.2 Calculus and VectorValued Func ons

The previous sec on introduced us to a new mathema cal object, the vector
valued func on. We now apply calculus concepts to these func ons. We start
with the limit, then work our way through deriva ves to integrals.
Limits of VectorValued Func ons

The ini al deni on of the limit of a vectorvalued func on is a bit in mida ng, as was the deni on of the limit in Deni on 1. The theorem following
the deni on shows that in prac ce, taking limits of vectorvalued func ons is
no more dicult than taking limits of realvalued func ons.
Deni on 68
Let a vectorvalued func onr(t) be given, dened on an open interval I

containing c. The limit ofr(t), as t approaches c is L, expressed as
lim r(t) = L,
tc
means that given any > 0, there exists a > 0 such that whenever
|t c| < , we have ||r(t) L || < .
Note how the measurement of distance between real numbers is the absolute value of their dierence; the measure of distance between vectors is the
vector norm, or magnitude, of their dierence.
Theorem 89
1. Let r(t) = f(t), g(t) be a vectorvalued func on in R2 dened

on an open interval I containing c. Then
lim r(t) = lim f(t) , lim g(t) .
tc
tc
tc
2. Letr(t) = f(t), g(t), h(t) be a vectorvalued func on in R3 dened on an open interval I containing c. Then
lim r(t) = lim f(t) , lim g(t) , lim h(t)

tc
tc
tc
tc
Notes:
605

Theorem 89 states that we compute limits componentwise.
. Example 360
Finding limits ofvectorvalued func ons
sin t 2
Letr(t) =
, t 3t + 3, cos t . Find lim r(t).
t0
t
We apply the theorem and compute limits componentwise.
sin t
lim r(t) = lim
, lim t2 3t + 3 , lim cos t
t0
t0 t
t0
t0
= 1, 3, 1 .
.
Con nuity
.
Deni on 69
Con nuity of VectorValued Func ons
Let r(t) be a vectorvalued func on dened on an open interval I containing c.
1. r(t) is con nuous at c if lim r(t) = r(c).

tc
2. Ifr(t) is con nuous at all c in I, thenr(t) is con nuous on I.
We again have a theorem that lets us evaluate con nuity componentwise.

.
Theorem 90
Con nuity of VectorValued Func ons
Let r(t) be a vectorvalued func on.

dened on an open interval I containing c. r(t) is con nuous at c if, and only if, each of its component
func ons is con nuous at c.
. Example 361
.Evalua ng con nuity
of vectorvalued func ons

sin t 2
Let r(t) =
, t 3t + 3, cos t . Determine whether r is con nuous at
t
t = 0 and t = 1.
While the second and third components ofr(t) are dened
S
at t = 0, the rst component, (sin t)/t, is not. Since the rst component is not
even dened at t = 0,r(t) is not dened at t = 0, and hence it is not con nuous
at t = 0.
Notes:
606

At t = 1 each of the component func ons is con nuous. Therefore r(t) is
con nuous at t = 1. .
Deriva ves
Consider a vectorvalued func onr dened on an open interval I containing
t0 and t1 . We can compute the displacement ofr on [t0 , t1 ], as shown in Figure
11.8 (a). Recall that dividing the displacement vector by t1 t0 gives the average
rate of change on [t0 , t1 ], as shown in (b).
r (t0 )
r(t1 )
r(t0 )
r(t1 ) r(t0 )
t1 t0
r(t0 )
r(t0 )
r(t1 )
r(t1 )
(a)
(b)
Figure 11.8: Illustra ng displacement, leading to an understanding of the deriva ve of vectorvalued func ons.
The deriva ve of a vectorvalued func on is a measure of the instantaneous

rate of change, measured by taking the limit as the length of [t0 , t1 ] goes to 0.
Instead of thinking of an interval as [t0 , t1 ], we think of it as [c, c + h] for some
value of h (hence the interval has length h). The average rate of change is
r(c + h) r(c)
h
for any value of h = 0. We take the limit as h 0 to measure the instantaneous
rate of change; this is the deriva ve ofr.
Deni on 70
Deriva ve of a VectorValued Func on
Letr(t) be con nuous on an open interval I containing c.

1. The deriva ve ofr at t = c is
r(c + h) r(c)
.
h0
h
r (c) = lim
Alternate nota ons for the deriva ve ofr

include:
r (t) =
)
d(
dr
r(t) = .
dt
dt
2. The deriva ve ofr is

r(t + h) r(t)
.
h0
h
r (t) = lim
Notes:
607

If a vectorvalued func on has a deriva ve for all c in an open interval I, we
say thatr(t) is dieren able on I.
Once again we might view this deni on as in mida ng, but recall that we
can evaluate limits componentwise. The following theorem veries that this
means we can compute deriva ves componentwise as well, making the task
not too dicult.
.
Theorem 91
Deriva ves of VectorValued Func ons
1. Letr(t) = f(t), g(t) . Then
r (t) = f (t), g (t) .

2. Letr(t) = f(t), g(t), h(t) . Then
r (t) = f (t), g (t), h (t) .
. Example 362
Letr(t) = t2 , t .
y
2
r(t)
1. Sketchr(t) andr (t) on the same axes.

r (t)
2. Computer (1) and sketch this vector with its ini al point at the origin and
atr(1).
x
4
1. Theorem 91 allows us to compute deriva ves componentwise, so

(a)
r (t) = 2t, 1 .
y
2
r(t) andr (t) are graphed together in Figure 11.9 (a). Note how plo ng
the two of these together, in this way, is not very illumina ng. When
dealing with realvalued func ons, plo ng f(x) with f (x) gave us useful
informa on as we were able to compare f and f at the same x-values.
When dealing with vectorvalued func ons, it is hard to tell which points
on the graph ofr correspond to which points on the graph ofr.
r (1)
r (1)
x
2. We easily computer (1) = 2, 1, which is drawn in Figure 11.9 with its

ini al point at the origin, as well as at r(1) = 1, 1 . These are sketched
in Figure 11.9 (b).
.
(b)
Figure 11.9: Graphing the deriva ve of a
vectorvalued func on in Example 362.
608
Deriva ves of vectorvalued func ons
Notes:

. Example 363
Deriva ves of vectorvalued func ons
Let r(t) = cos t, sin t, t. Compute r (t) and r (/2). Sketch r (/2) with its
ini al point at the origin and atr(/2).
6
4
We computer asr (t) = sin t, cos t, 1. At t = /2, we

haver (/2) = 1, 0, 1. Figure 11.10 shows two graphs ofr(t), from dierent perspec ves, with r (/2) plo ed with its ini al point at the origin and at
r(/2). .
S
r (/2)
r (/2)
In Examples 362 and 363, sketching a par cular deriva ve with its ini al
point at the origin did not seem to reveal anything signicant. However, when
we sketched the vector with its ini al point on the corresponding point on the
graph, we did see something signicant: the vector appeared to be tangent to
the graph. We have not yet dened what tangent means in terms of curves in
space; in fact, we use the deriva ve to dene this term.
Deni on 71
(a)
4
x
Tangent Vector, Tangent Line
Letr(t) be a dieren able vectorvalued func on on an open interval I

containing c, wherer (c) = 0.
1. A vector v is tangent to the graph ofr(t) at t = c if v is parallel to

r (c).
2
1
(b)
Figure 11.10: Viewing a vectorvalued
func on, and its deriva ve at one point,
from two dierent perspec ves.
2. The tangent line to the graph of r(t) at t = c is the line through

r(c) with direc on parallel to r (c). An equa on of the tangent
line is
(t) = r(c) + tr (c).
z
2
. Example 364
Finding tangent lines to curves in space

Letr(t) = t, t2 , t3 on [1.5, 1.5]. Find the vector equa on of the line tangent
to the graph ofr at t = 1.
To nd the equa on of a line, we need a point on the line
and the lines direc on. The point is given byr(1) = 1, 1, 1. (To be clear,
1, 1, 1 is a vector, not a point, but we use the point pointed to by this
vector.)
fromr (1). We compute, componentwise, r (t) =
The direc
on comes
2
1, 2t, 3t . Thusr (1) = 1, 2, 3.

The vector equa on of the line is (t) = 1, 1, 1 + t 1, 2, 3. This line
andr(t) are sketched, from two perspec ves, in Figure 11.11 (a) and (b). .
2
x
2
2
.
(a)
z
2
2
2
2
y 2
2
Notes:
.
(b)
Figure 11.11: Graphing a curve in space
with its tangent line.
609

. Example 365
Finding tangent lines to curves
Find the equa ons of the lines tangent tor(t) = t3 , t2 at t = 1 and t = 0.

S
r(1) = 1, 1
y
2
We nd thatr (t) = 3t2 , 2t . At t = 1, we have
and r (1) = 3, 2 ,
r(t)
so the equa on of the line tangent to the graph ofr(t) at t = 1 is

x
2
(t) = 1, 1 + t 3, 2 .
(t)
. .
Figure 11.12: Graphing r(t) and its tangent line in Example 365.
This line is graphed withr(t) in Figure 11.12.

At t = 0, we haver (0) = 0, 0 = 0! This implies that the tangent line has
no direc on. We cannot apply Deni on 71, hence cannot nd the equa on of
the tangent line. .
3 We
were unable to compute the equa on of the tangent line to r(t) =
t , t2 at t = 0 becauser (0) = 0. The graph in Figure 11.12 shows that there
is a cusp at this point. This leads us to another deni on of smooth, previously
dened by Deni on 46 in Sec on 9.2.
Deni on 72
Smooth VectorValued Func ons
Letr(t) be a dieren able vectorvalued func on on an open interval I.

r(t) is smooth on I ifr (t) = 0 on I.
Having established deriva ves of vectorvalued func ons, we now explore

the rela onships between the deriva ve and other vector opera ons. The following theorem states how the deriva ve interacts with vector addi on and the
various vector products.
Notes:
610
Theorem 92
Properies of Deriva ves of VectorValued Func ons
Letr and s be dieren able vectorvalued func ons, let f be a dierenable realvalued func on, and let c be a real number.
)
d(
r(t) s(t) = r (t) s (t)
1.
dt
)
d(
2.
cr(t) = cr (t)
dt
)
d(
3.
Product Rule
f(t)r(t) = f (t)r(t) + f(t)r (t)
dt
)
d(
4.
Product Rule
r(t) s(t) = r (t) s(t) +r(t) s (t)
dt
)
d(
5.
Product Rule
r(t) s(t) = r (t) s(t) +r(t) s (t)
dt
( )
d ( ( ))
6.
Chain Rule
r f(t) = r f(t) f (t)
dt
. Example 366
.U
sing deriva ve proper es of vectorvalued func ons

Letr(t) = t, t2 1 and let u(t) be the unit vector that points in the direc on
ofr(t).
2
r(t)
1. Graphr(t) and u(t) on the same axes, on [2, 2].
2. Find u (t) and sketch u (2), u (1) and u (0). Sketch each with ini al
point the corresponding point on the graph of u.
x
2
.
S
1. To form the unit vector that points in the direc on ofr, we need to divide
r(t) by its magnitude.
||r(t) || =
t2 + (t2 1)2
u(t) =
1
t2
(t2
1)2
u(t)
2
1
Figure 11.13: Graphingr(t) andu(t) in Example 366.
t, t2 1 .
r(t) and u(t) are graphed in Figure 11.13. Note how the graph of u(t)
forms part of a circle; this must be the case, as the length of u(t) is 1 for
all t.
Notes:
611

2. To compute u (t), we use Theorem 92, wri ng
(
)1/2
1
= t2 +(t2 1)2
.
where f(t) =
2
2
2
t + (t 1)
u(t) = f(t)r(t),
y
(We could write
u(t)
u(t) =
x
1
t2 1
,
t2 + (t2 1)2
t2 + (t2 1)2
and then take the deriva ve. It is a ma er of preference; this la er method

requires two applica ons of the Quo ent Rule where our method uses the
Product and Chain Rules.)
We nd f (t) using the Chain Rule:
)3/2 (
)
1( 2
2t + 2(t2 1)(2t)
t + (t2 1)2
2
2t(2t2 1)
= (
)3
2 t2 + (t2 1)2
Figure 11.14: Graphing some of the

deriva ves of u(t) in Example 366.
f (t) =
We now nd u (t) using part 3 of Theorem 92:

u (t) = f (t)u(t) + f(t)u (t)
2
2t(2t2 1)
1
= (
1, 2t .
)3 t, t 1 + 2
t + (t2 1)2
2 t2 + (t2 1)2
This is admi edly very messy; such is usually the case when we deal
with unit vectors. We can use this formula to compute u (2), u (1)
and u (0):
15
10
u (2) = ,
0.320, 0.213
13 13 13 13
u (1) = 0, 2
u (0) = 1, 0
Each of these is sketched in Figure 11.14. Note how the length of the
vector gives an indica on of how quickly the circle is being traced at that
point. When t = 2, the circle is being drawn rela vely slow; when t =
1, the circle is being traced much more quickly.
It is a basic geometric fact that a line tangent to a circle at a point P is perpendicular to the line passing through the center of the circle and P. This is
Notes:
612

illustrated in Figure 11.14; each tangent vector is perpendicular to the line that
passes through its ini al point and the center of the circle. Since the center of
the circle is the origin, we can state this another way: u (t) is orthogonal to u(t).
Recall that the dot product serves as a test for orthogonality: if u v = 0,
then u is orthogonal to v. Thus in the above example, u(t) u (t) = 0.
This is true of any vectorvalued func on that has a constant length, that is,
that traces out part of a circle. It has important implica ons later on, so we state
it as a theorem (and leave its formal proof as an Exercise.)
Theorem 93
VectorValued Func ons of Constant Length
Letr(t) be a dieren able vectorvalued

. func on on an open interval I of
constant length. That is, ||r(t) || = c for all t in I (equivalently,r(t)r(t) =
c2 for all t in I). Thenr(t) r (t) = 0 for all t in I.
Integra on
Indenite and denite integrals of vectorvalued func ons are also evaluated componentwise.
Theorem 94
Indenite and Denite Integrals of VectorValued

Func ons
Letr(t) = f(t), g(t) be a vectorvalued func on in R2 .
r(t) dt =
1.
f(t) dt, g(t) dt
2.
r(t) dt =
a
f(t) dt,
a
g(t) dt
a
A similar statement holds for vectorvalued func ons in R3 .
. Example 367
.Evalua ng a denite integral of a vectorvalued func on
1
2t
r(t) dt.
Letr(t) = e , sin t . Evaluate
0
Notes:
613

We follow Theorem 94.
1
1
2t
r(t) dt =
e , sin t dt
0
0
1
1
2t
=
e dt ,
sin t dt
0
0
1
1 2t 1

=
e , cos t
2
0
0
1 2
=
(e 1) , cos(1) + 1
2
3.19, 0.460 .
. Example 368
.Solving an ini al value problem
Letr (t) = 2, cos t, 12t. Findr(t) where:
r(0) = 7, 1, 2 and
r (0) = 5, 3, 0 .
Knowingr (t) = 2, cos t, 12t, we ndr (t) by evalua ng the
indenite integral.
r (t) dt =
2 dt , cos t dt , 12t dt
= 2t + C1 , sin t + C2 , 6t2 + C3
= 2t, sin t, 6t2 + C1 , C2 , C3
= 2t, sin t, 6t2 + C.

S
Note how each indenite integral creates its own constant which we collect as
one constant vector C. Knowingr (0) = 5, 3, 0 allows us to solve for C:
r (t) = 2t, sin t, 6t2 + C

r (0) = 0, 0, 0 + C
5, 3, 0 = C
Sor (t) = 2t, sin t, 6t2 + 5, 3, 0 = 2t + 5, sin t + 3, 6t2 . To nd r(t),

we integrate once more.
r (t) dt =
Notes:
614
2t + 5 dt,
sin t + 3 dt, 6t dt
= t2 + 5t, cos t + 3t, 2t3 + C

Withr(0) = 7, 1, 2, we solve for C:
r(t) = t2 + 5t, cos t + 3t, 2t3 + C

r(0) = 0, 1, 0 + C
7, 1, 2 = 0, 1, 0 + C
7, 0, 2 = C.
Sor(t) = t2 + 5t, cos t + 3t, 2t3 +7, 0, 2 = t2 + 5t 7, cos t + 3t, 2t3 + 2 .

.
What does the integra on of a vectorvalued func on mean? There are
many applica ons, but none as direct as the area under the curve that we
used in understanding the integral of a realvalued func on.
A key understanding for us comes from considering the integral of a derivave:
b
b

r (t) dt = r(t) = r(b) r(a).
a
Integra ng a rate of change func on gives displacement.

No ng that vectorvalued func ons are closely related to parametric equaons, we can describe the arc length of the graph of a vectorvalued func on
as an integral. Given parametric equa ons x = f(t), y = g(t), the arc length on
[a, b] of the graph is
Arc Length =
f (t)2 + g (t)2 dt,
as stated in Theorem 82 in Sec on 9.3. Ifr(t) = f(t), g(t), note that f (t)2 + g (t)2 =
|| r (t) ||. Therefore we can express the arc length of the graph of a vector
valued func on as an integral of the magnitude of its deriva ve.
Theorem 95
Arc Length of a VectorValued Func on
Letr(t) be a vectorvalued func on wherer (t) is con nuous on [a, b].

The arc length L of the graph ofr(t) is
L=
||r (t) || dt.
Note that we are actually integra ng a scalarfunc on here, not a vector

valued func on.
Notes:
615

The next sec on takes what we have established thus far and applies it to
objects in mo on. We will letr(t) describe the path of a mo on in the plane or
in space and will discover the informa on provided byr (t) andr (t).
Notes:
616
Exercises 11.2
Terms and Concepts
In Exercises 23 26, nd the value(s) of t for whichr(t) is not

smooth.
1. Limits, deriva ves and integrals of vectorvalued func ons

are all evaluated
wise.
2. The denite integral of a rate of change func on gives
.
3. Why is it generally not useful to graph both r(t) and r (t)

on the same axes?
4. lim 2t + 1, 3t2 1, sin t

5. lim
t3
t2 9
e,
t+3
25. r(t) = cos t sin t, sin t cos t, cos(4t)
26. r(t) = t3 3t + 2, cos(t), sin2 (t)

Exercises 27 29 ask you to verify
2 parts of Theorem 92.
3
In
each
let
f(t)
=
t
,
r(t)
=
t , t 1, 1 and s(t) =
sin t, et , t . Compute the various deriva ves as indicated.
Problems
t5
23. r(t) = cos t, sin t t
24. r(t) = t2 2t + 1, t3 + t2 5t + 3
27. Simplify f(t)r(t), then nd its deriva ve; show this is the
same as f (t)r(t) + f(t)r (t).
28. Simplifyr(t) s(t), then nd its deriva ve; show this is the
same asr (t) s(t) +r(t) s (t).
1
, (1 + t) t
t0 sin t
29. Simplifyr(t) s(t), then nd its deriva ve; show this is the
same asr (t) s(t) +r(t) s (t).
r(t + h) r(t)
, wherer(t) = t2 , t, 1 .
h0
h
In Exercises 30 33 , evaluate the given denite or indenite

integral.
30.
t , cos t, tet dt
6. lim
7. lim
In Exercises 8 9, iden fy the interval(s) on whichr(t) is connuous.
8. r(t) = t2 , 1/t
9. r(t) = cos t, et , ln t
In Exercises 10 14, nd the deriva ve of the given func on.
10. r(t) = cos t, et , ln t
1 2t 1
11. r(t) =
,
, tan t
t 3t + 1
2
12. r(t) = (t ) sin t, 2t + 5
13. r(t) = t2 + 1, t 1 sin t, 2t + 5
14. r(t) = t2 + 1, t 1, 1 sin t, 2t + 5, 1

In Exercises 15 18, ndr (t). Sketchr(t) andr (1), with the
ini al point ofr (1) atr(1).
15. r(t) = t2 + t, t2 t
16. r(t) = t2 2t + 2, t3 3t2 + 2t
17. r(t) = t2 + 1, t3 t
18. r(t) = t2 4t + 5, , t3 6t2 + 11t 6

In Exercises 19 22, give the equa on of the line tangent to
the graph ofr(t) at the given t value.
19. r(t) = t2 + t, t2 t at t = 1.
20. r(t) = 3 cos t, sin t at t = /4.
21. r(t) = 3 cos t, 3 sin t, t at t = .
22. r(t) = et , tan t, t at t = 0.

31.
1
, sec2 t
1 + t2
dt
sin t, cos t dt
32.
0
2
33.
2
2t + 1, 2t 1 dt
In Exercises 34 37 , solve the given ini al value problems.

34. Findr(t), given thatr (t) = t, sin t andr(0) = 2, 2.
35. Findr(t), given thatr (t) = 1/(t + 1), tan t and
r(0) = 1, 2.
36. Findr(t), given thatr (t) = t2 , t, 1 ,

r (0) = 1, 2, 3 andr(0) = 4, 5, 6.
37. Findr(t), given thatr (t) = cos t, sin t, et ,
r (0) = 0, 0, 0 andr(0) = 0, 0, 0.
In Exercises 38 41 , nd the arc length of r(t) on the indicated interval.
38. r(t) = 2 cos t, 2 sin t, 3t on [0, 2].
39. r(t) = 5 cos t, 3 sin t, 4 sin t on [0, 2].
40. r(t) = t3 , t2 , t3 on [0, 1].
41. r(t) = et cos t, et sin t on [0, 1].

42. Prove Theorem 93; that is, show ifr(t) has constant length
and is dieren able, then r(t)

( r (t) )= 0. (Hint: use the
Product Rule to compute dtd r(t) r(t) .)
11.3 The Calculus of Mo on

A common use of vectorvalued func ons is to describe the mo on of an object
in the plane or in space. A posi on func onr(t) gives the posi on of an object
at me t. This sec on explores how deriva ves and integrals are used to study
the mo on described by such a func on.
.
Deni on 73
Velocity, Speed and Accelera on
Letr(t) be a posi on func on in R2 or R3 .
1. Velocity, denoted v(t), is the instantaneous rate of posi on

change; that is, v(t) = r (t).
2. Speed is the magnitude of velocity, || v(t) ||.
3. Accelera on, denoted a(t), is the instantaneous rate of velocity
change; that is, a(t) = v (t) = r (t).
. Example 369
.Finding velocity and accelera on
An object is moving with posi on func onr(t) = t2 t, t2 + t , 3 t 3,

where distances are measured in feet and me is measured in seconds.
10
1. Find v(t) and a(t).

2. Sketchr(t); plot v(1), a(1), v(1) and a(1), each with their ini al point
at their corresponding point on the graph ofr(t).
3. When is the objects speed minimized?

x
5
10
(a)
1. Taking deriva ves, we nd
v(t) = r (t) = 2t 1, 2t + 1
and a(t) = r (t) = 2, 2 .
10
Note that accelera on is constant.

2. v(1) = 3, 1, a(1) = 2, 2; v(1) = 1, 3, a(1) = 2, 2.
These are plo ed withr(t) in Figure 11.15 (a).
We can think of accelera on as pulling the velocity vector in a certain

direc on. At t = 1, the velocity vector points down and to the le ; at
t = 1, the velocity vector has been pulled in the 2, 2 direc on and is
10
(b)
Figure 11.15: Graphing the posi on, velocity and accelera on of an object in Example 369.
618
Notes:

now poin ng up and to the right. In Figure 11.15 (b) we plot more velocity/accelera on vectors, making more clear the eect accelera on has on
velocity.
Sincea(t) is constant in this example, as t grows largev(t) becomes almost
parallel to a(t). For instance, when t = 10, v(10) = 19, 21, which is
nearly parallel to 2, 2.
3. The objects speed is given by
|| v(t) || = (2t 1)2 + (2t + 1)2 = 8t2 + 2.

To nd the minimal speed, we could apply calculus techniques (such as
set the deriva ve equal to 0 and solve for t, etc.) but we can nd it by
inspec on. Inside the square root we have a quadra c which is minimized
when t = 0. Thus the speed is minimized at t = 0, with a speed of 2

/s.
The graph in Figure 11.15 (b) also implies speed is minimized here. The
lled dots on the graph are located at integer values of t between 3
and 3. Dots that are far apart imply the object traveled a far distance in
1 second, indica ng high speed; dots that are close together imply the
object did not travel far in 1 second, indica ng a low speed. The dots are
closest together near t = 0, implying the speed is minimized near that
value.
. Example 370
.Analyzing Mo on
Two objects follow an iden cal path
atdierent rates on [1, 1]. The posi on
r1 (t) = t, t2 ; the posi on func on for Object 2 is
func on for Object
1
is
r2 (t) = t3 , t6 , where distances are measured in feet and me is measured

in seconds. Compare the velocity, speed and accelera on of the two objects on
the path.
y
3
We begin by compu ng the velocity and accelera on funcS

on for each object:
v1 (t) = 1, 2t
v2 (t) = 3t2 , 6t5
a1 (t) = 0, 2
a2 (t) = 6t, 30t4
We immediately see that Object 1 has constant accelera on, whereas Object 2
does not.
At t = 1, we have v1 (1) = 1, 2 and v2 (1) = 3, 6; the velocity
of Object 2 is three mes that of Object
1 and so it follows that
the speed of
Object 2 is three mes that of Object 1 (3 5 /s compared to 5 /s.)
x
2
Figure 11.16: Plo ng velocity and accelera on vectors for Object 1 in Example
370.
Notes:
619
y
1
r1 (t)
0.5
.1
x
0.5
0.5
(a)
y
1
r2 (t)
0.5
.1
x
0.5
0.5
At t = 0, the velocity of Object 1 is v(1) = 1, 0 and the velocity of Object

2 is 0! This tells us that Object 2 comes to a complete stop at t = 0.
In Figure 11.16, we see the velocity and accelera on vectors for Object 1
plo ed for t = 1, 1/2, 0, 1/2 and t = 1. Note again how the constant accelera on vector seems to pull the velocity vector from poin ng down, right to
up, right. We could plot the analogous picture for Object 2, but the velocity and
accelera on vectors are rather large (a2 (1) = 6, 30!)
Instead, we simply plot the loca ons of Object 1 and 2 on intervals of 1/10th
of a second, shown in Figure 11.17 (a) and (b). Note how the x-values of Object
1 increase at a steady rate. This is because the x-component of a(t) is 0; there is
no accelera on in the x-component. The dots are not evenly spaced; the object
is moving faster near t = 1 and t = 1 than near t = 0.
In part (b) of the Figure, we see the points plo ed for Object 2. Note the
large change in posi on from t = 1 to t = 0.9; the object starts moving very
quickly. However, it slows considerably at it approaches the origin, and comes
to a complete stop at t = 0. While it looks like there are 3 points near the origin,
there are in reality 5 points there.
Since the objects begin and end at the same loca on, the have the same displacement. Since they begin and end at the same me, with the same displacement, they have they have the same average rate of change (i.e, they have the
same average velocity). Since they follow the same path, they have the same
distance traveled. Even though these three measurements are the same, the
objects obviously travel the path in very dierent ways. .
(b)
Figure 11.17: Comparing the posi ons of
Objects 1 and 2 in Example 370.
. Example 371
.Analyzing the mo on of a whirling ball on a string
A young boy whirls a ball, a ached to a string, above his head in a counterclockwise circle. The ball follows a circular path and makes 2 revolu ons per
second. The string has length 2 .
1. Find the posi on func onr(t) that describes this situa on.
2. Find the accelera on of the ball and derive a physical interpreta on of it.
3. A tree stands 10 in front of the boy. At what t-values should the boy
release the string so that the ball hits the tree?
S
1. The ball whirls in a circle. Since the string is 2 long, the radius of the
circle is 2. The posi on func on r(t) = 2 cos t, 2 sin t describes a circle
with radius 2, centered at the origin, but makes a full revolu on every
2 seconds, not two revolu ons per second. We modify the period of the
Notes:
620

trigonometric func ons to be 1/2 by mul plying t by 4. The nal posi on
func on is thus
r(t) = 2 cos(4t), 2 sin(4t) .
(Plot this for 0 t 1/2 to verify that one revolu on is made in 1/2 a
second.)
2. To nd a(t), we deriver(t) twice.
v(t) = r (t) = 8 sin(4t), 8 cos(4t)
a(t) = r (t) = 322 cos(4t), 322 sin(4t)

= 322 cos(4t), sin(4t) .
Note how a(t) is parallel tor(t), but has a dierent magnitude and points
in the opposite direc on. Why is this?
Recall the classic phyics equa on, Force = mass accelera on. A force
ac ng on a mass induces accelera on (i.e., the mass moves); accelera on
ac ng on a mass induces a force (gravity gives our mass a weight). Thus
force and accelera on are closely related. A moving ball wants to travel
in a straight line. Why does the ball in our example move in a circle? It is
a ached to the boys hand by a string. The string applies a force to the ball,
aec ng its mo on: the string accelerates the ball. This is not acceleraon in the sense of it travels faster; rather, this accelera on is changing
the velocity of the ball. In what direc on is this force/accelera on being
applied? In the direc on of the string, towards the boys hand.
The magnitude of the accelera on is related to the speed at which the ball
is traveling. A ball whirling quickly is rapidly changing direc on/velocity.
When velocity is changing rapidly, the accelera on must be large.
There are many ways to nd this me value. We choose one that is relavely simple computa onally. As shown in Figure 11.18, the vector from
the release point to the tree is 0, 10 r(t0 ). This line segment is tangent
to the circle, which means it is also perpendicular to r(t0 ) itself, so their
dot product is 0.
r(t0 )
Let t = t0 be the me when the boy lets go of the string. The ball will be
atr(t0 ), traveling in the direc on of v(t0 ). We want to nd t0 so that this
line contains the point (0, 10) (since the tree is 10 directly in front of the
boy).
0, 10
3. When the boy releases the string, the string no longer applies a force to
the ball, meaning accelera on is 0 and the ball can now move in a straight
line in the direc on of v(t). .
Figure 11.18: Modeling the ight of a ball

in Example 371.
Notes:
621
(
)
r(t0 ) 0, 10 r(t0 ) = 0
2 cos(4t0 ), 2 sin(4t0 ) 2 cos(4t0 ), 10 2 sin(4t0 ) = 0
4 cos2 (4t0 ) + 20 sin(4t0 ) 4 sin2 (4t0 ) = 0
20 sin(4t0 ) 4 = 0
sin(4t0 ) = 1/5
4t0 = sin1 (1/5)
4t0 0.2 + 2n,
where n is an integer. Solving for t0 we have:
t0 0.016 + n/2
This is a wonderful formula. Every 1/2 second a er t = 0.016s the boy
can release the string (since the ball makes 2 revolu ons per second, he
has two chances each second to release the ball).
. Example 372
Analyzing mo on in space
An object moves in spiral with posi on func on r(t) = cos t, sin t, t, where
distances are measured in meters and me is in minutes. Describe the objects
speed and accelera on at me t.
S
Withr(t) = cos t, sin t, t, we have:

v(t) = sin t, cos t, 1 and
a(t) = cos t, sin t, 0 .
The speed of the object is ||v(t) || = ( sin t)2 + cos2 t + 1 = 2m/min;

it moves at a constant speed. Note that the object does not accelerate in the
z-direc on, but rather moves up at a constant rate of 1m/min. .
The objects in Examples 371 and 372 traveled at a constant speed. That is,
|| v(t) || = c for some constant c. Recall Theorem 93, which states that if a
vectorvalued func on r(t) has constant length, then r(t) is perpendicular to
its deriva ve: r(t) r (t) = 0. In these examples, the velocity func on has
constant length, therefore we can conclude that the velocity is perpendicular to
the accelera on: v(t) a(t) = 0. A quick check veries this.
There is an intui ve understanding of this. If accelera on is parallel to velocity, then it is only aec ng the objects speed; it does not change the direc on
of travel. (For example, consider a dropped stone. Accelera on and velocity are
Notes:
622

parallel straight down and the direc on of velocity never changes, though
speed does increase.) If accelera on is not perpendicular to velocity, then there
is some accelera on in the direc on of travel, inuencing the speed. If speed
is constant, then accelera on must be orthogonal to velocity, as it then only
aects direc on, and not speed.
Key Idea 53
Objects With Constant Speed
If an object moves with constant speed, then its velocity and accelera on
vectors are orthogonal. That is, v(t) a(t) = 0.
Projec le Mo on
An important applica on of vectorvalued posi on func ons is projec le
mo on: the mo on of objects under the inuence of gravity. We will measure
me in seconds, and distances will either be in meters or feet. We will show
that we can completely describe the path of such an object knowing its ini al
posi on and ini al velocity (i.e., where it is and where it is going.)
Suppose an object has ini al posi on r(0) = x0 , y0 and ini al velocity
v(0) = vx , vy . It is customary to rewrite v(0) in terms of its speed v0 and
direc on u, where u is a unit vector. Recall all unit vectors in R2 can be wri en
as cos , sin , where is an angle measure counterclockwise from the x-axis.
(We refer to as the angle of eleva on.) Thus v(0) = v0 cos , sin .
Since the accelera on of the object is known, namely a(t) = 0, g, where
g is the gravita onal constant, we can nd r(t) knowing our two ini al condions. We rst nd v(t):
Note: In this text we use g = 32 /s when

using Imperial units, and g = 9.8m/s
when using SI units.
v(t) =
a(t) dt
v(t) =
0, g dt
v(t) = 0, gt + C.
Knowing v(0) = v0 cos , sin , we have C = v0 cos t, sin t and so
v(t) = v0 cos , gt + v0 sin .
Notes:
623

We integrate once more to ndr(t):
r(t) = v(t) dt
r(t) =
v0 cos , gt + v0 sin dt
)
(
)
1 2 (
r(t) = v0 cos t, gt + v0 sin t + C
2
Knowingr(0) = x0 , y0 , we conclude C = x0 , y0 and
r(t) =
Key Idea 54
(
)
(
)
1
v0 cos t + x0 , gt2 + v0 sin t + y0
2
Projec le Mo on
The posi on func on of a projec le propelled from an ini al posi on of

r0 = x0 , y0 , with ini al speed v0 , with angle of eleva on and neglecting all accelera ons but gravity is
(
)
(
)
1
r(t) = v0 cos t + x0 , gt2 + v0 sin t + y0 .
2
Le ng v0 = v0 cos , sin , r(t) can be wri en as
1
r(t) = 0, gt2 + v0 t +r0 .
2
We demonstrate how to use this posi on func on in the next two examples.
. Example 373
.Projec le Mo on
Sydney shoots her Red Ryder bb gun across level ground from an eleva on of
4 , where the barrel of the gun makes a 5 angle with the horizontal. Find how
far the bb travels before landing, assuming the bb is red at the adver sed rate
of 350 /s and ignoring air resistance.
S
Notes:
624
A direct applica on of Key Idea 54 gives
r(t) = (350 cos 5 )t, 16t2 + (350 sin 5 )t + 4
346.67t, 16t2 + 30.50t + 4 ,

where we set her ini al posi on to be 0, 4. We need to nd when the bb lands,
then we can nd where. We accomplish this by se ng the y-component equal
to 0 and solving for t:
16t2 + 30.50t + 4 = 0
t=
30.50
30.502 4(16)(4)
32
t 2.03s.
(We discarded a nega ve solu on that resulted from our quadra c equa on.)
We have found that the bb lands 2.03s a er ring; with t = 2.03, we nd
the x-component of our posi on func on is 346.67(2.03) = 703.74 . The bb
lands about 704 feet away. .
. Example 374
.Projec le Mo on
Alex holds his sisters bb gun at a height of 3 and wants to shoot a target that
is 6 above the ground, 25 away. At what angle should he hold the gun to hit
his target? (We s ll assume the muzzle velocity is 350 /s.)
The posi on func on for the path of Alexs bb is
r(t) = (350 cos )t, 16t2 + (350 sin )t + 3 .
We need to nd so thatr(t) = 25, 6 for some value of t. That is, we want to

nd and t such that
(350 cos )t = 25
16t2 + (350 sin )t + 3 = 6.
and
This is not trivial (though not hard). We start by solving each equa on for cos
and sin , respec vely.
cos =
25
350t
and
sin =
3 + 16t2
.
350t
Using the Pythagorean Iden ty cos2 + sin2 = 1, we have

(
)2 (
)2
25
3 + 16t2
+
=1
350t
350t
Mul ply both sides by (350t)2 :
252 + (3 + 16t2 )2 = 3502 t2
256t4 122, 404t2 + 634 = 0
Notes:
625

This is a quadra c in t2 . That is, we can apply the quadra c formula to nd t2 ,
then solve for t itself.
122, 404 122, 4042 4(256)(634)

2
t =
512
t2 = 0.0052, 478.135
t = 0.072, 21.866
Clearly the nega ve t values do not t our context, so we have t = 0.072 and
t = 21.866. Using cos = 25/(350t), we can solve for :
(
)
(
)
25
25
= cos1
and cos1
350 0.072
350 21.866
= 7.03
and
89.8 .
Alex has two choices of angle. He can hold the rie at an angle of about 7 with
the horizontal and hit his target 0.07s a er ring, or he can hold his rie almost
straight up, with an angle of 89.8 , where hell hit his target about 22s later. The
rst op on is clearly the op on he should choose. .
Distance Traveled
Consider a driver who sets her cruisecontrol to 60mph, and travels at this
speed for an hour. We can ask:
1. How far did the driver travel?
2. How far from her star ng posi on is the driver?
The rst is easy to answer: she traveled 60 miles. The second is impossible to
answer with the given informa on. We do not know if she traveled in a straight
line, on an oval racetrack, or along a slowlywinding highway.
This highlights an important fact: to compute distance traveled, we need
only to know the speed, given by || v(t) ||.
.
Theorem 96
Distance Traveled
Letv(t) be a velocity func on for a moving object. The distance traveled

by the object on [a, b] is:
distance traveled =
|| v(t) || dt.
Note that this is just a restatement of Theorem 95: arc length is the same as distance traveled, just viewed in a dierent context.
Notes:
626

. Example 375
Distance Traveled, Displacement, and Average Speed
A par cle moves in space with posi on func onr(t) = t, t2 , sin(t) on [2, 2],
where t is measured in seconds and distances are in meters. Find:
z
1
1. The distance traveled by the par cle on [2, 2].
2. The displacement of the par cle on [2, 2].

3. The par cles average speed.
2
1
4 y
1. We use Theorem 96 to establish the integral:

2
distance traveled =
|| v(t) || dt
2
2
=
2
(a)
.
z
1
1 + (2t)2 + 2 cos2 (t) dt.
This cannot be solved in terms of elementary func ons so we turn to numerical integra on, nding the distance to be 12.88m.
2
1
2
4y
2. The displacement is the vector

r(2) r(2) = 2, 4, 0 2, 4, 0 = 4, 0, 0 .
That is, the par cle ends with an x-value increased by 4 and with y- and
z-values the same (see Figure 11.19).
(b)
Figure 11.19: The path of the par cle,

from two perspec ves, in Example 375.
3. We found above that the par cle traveled 12.88m over 4 seconds. We can
compute average speed by dividing: 12.88/4 = 3.22m/s.
We should also consider Deni on 22 of Sec on 5.4, which says that the
b
1
average value of a func on f on [a, b] is ba
f(x) dx. In our context, the
a
average value of the speed is
2
1
1
average speed =
|| v(t) || dt 12.88 = 3.22m/s.
2 (2) 2
4
.
Note how the physical context of a par cle traveling gives meaning to a
more abstract concept learned earlier.
In Deni on 22 we dened the average value of a func on f(x) on [a, b] to
be
1
ba
f(x) dx.
a
Notes:
627

Note how in Example 375 we computed the average speed as
2
distance traveled
1
=
|| v(t) || dt;
travel me
2 (2) 2
that is, we just found the average value of || v(t) || on [2, 2].
Likewise, given posi on func onr(t), the average velocity on [a, b] is
displacement
1
=
travel me
ba
r (t) dt =
r(b) r(a)
;
ba
that is, it is the average value ofr (t), or v(t), on [a, b].
The next two sec ons inves gate more proper es of the graphs of vector
valued func ons and well apply these new ideas to what we just learned about
mo on.
Notes:
628
Exercises 11.3
Terms and Concepts
1. How is velocity dierent from speed?
2. What is the dierence between displacement and distance
traveled?
3. What is the dierence between average velocity and average speed?
4. Distance traveled is the same as
viewed in a dierent context.
, just
5. Describe a scenario where an objects average speed is a

large number, but the magnitude of the average velocity is
not a large number.
6. Explain why it is not possible to have an average velocity
with a large magnitude but a small average speed.
Problems
In Exercises 7 10 , a posi on func onr(t) is given. Find v(t)
and a(t).
7. r(t) = 2t + 1, 5t 2, 7
8. r(t) = 3t2 2t + 1, t2 + t + 14
9. r(t) = cos t, sin t
10. r(t) = t/10, cos t, sin t
In Exercises 11 14 , a posi on func onr(t) is given. Sketch
r(t) on the indicated interval. Find v(t) and a(t), then add
v(t0 ) and a(t0 ) to your sketch, with their ini al points atr(t0 ),
for the given value of t0 .
11. r(t) = t, sin t on [0, /2]; t0 = /4
12. r(t) = t2 , sin t2 on [0, /2]; t0 = /4
13. r(t) = t2 + t, t2 + 2t on [2, 2]; t0 = 1
2t + 3 2
14. r(t) =
,
t
on [1, 1]; t0 = 0
t2 + 1
In Exercises 15 24 , a posi on func on r(t) of an object is
given. Find the speed of the object in terms of t, and nd
where the speed is minimized/maximized on the indicated
interval.
15. r(t) = t2 , t on [1, 1]
16. r(t) = t2 , t2 t3 on [1, 1]

17. r(t) = 5 cos t, 5 sin t on [0, 2]
18. r(t) = 2 cos t, 5 sin t on [0, 2]
19. r(t) = sec t, tan t on [0, /4]
20. r(t) = t + cos t, 1 sin t on [0, 2]
21. r(t) = 12t, 5 cos t, 5 sin t on [0, 4]
22. r(t) = t2 t, t2 + t, t on [0, 1]
23. r(t) = t, t2 , 1 t2 on [1, 1]
1
24. Projec le Mo on: r(t) = (v0 cos )t, gt2 + (v0 sin )t
2
[
]
2v0 sin
on 0,
g
In Exercises 25 28 , posi on func onsr1 (t) andr2 (s) for two
objects are given that follow the same path on the respec ve
intervals.
(a) Show that the posi ons are the same at the indicated
t0 and s0 values; i.e., showr1 (t0 ) = r2 (s0 ).
(b) Find the velocity, speed and accelera on of the two
objects at t0 and s0 , respec vely.
25. r1 (t) = t, t2 on [0, 1]; t0 = 1

r2 (s) = s2 , s4 on [0, 1]; s0 = 1
26. r1 (t) = 3 cos t, 3 sin t on [0, 2]; t0 = /2
r2 (s) = 3 cos(4s), 3 sin(4s) on [0, /2]; s0 = /8
27. r1 (t) = 3t, 2t on [0, 2]; t0 = 2
r2 (s) = 6t 6, 4t 4 on [1, 2]; s0 = 2

28. r1 (t) = t, t on [0, 1]; t0 = 1
r2 (s) = sin t, sin t on [0, /2]; s0 = /2

In Exercises 29 32 , nd the posi on func on of an object
given its accelera on and ini al velocity and posi on.
29. a(t) = 2, 3; v(0) = 1, 2, r(0) = 5, 2
30. a(t) = 2, 3; v(1) = 1, 2, r(1) = 5, 2
31. a(t) = cos t, sin t; v(0) = 0, 1, r(0) = 0, 0
32. a(t) = 0, 32; v(0) = 10, 50, r(0) = 0, 0
In Exercises 33 36 , nd the displacement, distance traveled,
average velocity and average speed of the described object
on the given interval.
33. An object with posi on func onr(t) = 2 cos t, 2 sin t, 3t,
where distances are measured in feet and me is in seconds, on [0, 2].
34. An object with posi on func on r(t) = 5 cos t, 5 sin t,
where distances are measured in feet and me is in seconds, on [0, ].
35. An object with velocity func onv(t) = cos t, sin t, where
distances are measured in feet and me is in seconds, on
[0, 2].
36. An object with velocity func on v(t) = 1, 2, 1, where
distances are measured in feet and me is in seconds, on
[0, 10].
Exercises 37 42 ask you to solve a variety of problems based
on the principles of projec le mo on.
37. A boy whirls a ball, a ached to a 3 string, above his head
in a counterclockwise circle. The ball makes 2 revolu ons
per second.
At what t-values should the boy release the string so that
the ball heads directly for a tree standing 10 in front of
him?
38. David faces Goliath with only a stone in a 3 sling, which

he whirls above his head at 4 revolu ons per second. They
stand 20 apart.
(a) At what t-values must David release the stone in his
sling in order to hit Goliath?
(b) What is the speed at which the stone is traveling
when released?
(c) Assume David releases the stone from a height of 6
and Goliaths forehead is 9 above the ground. What
angle of eleva on must David apply to the stone to hit
Goliaths head?
39. A hunter aims at a deer which is 40 yards away. Her crossbow is at a height of 5 , and she aims for a spot on the
deer 4 above the ground. The crossbow res her arrows
at 300 /s.
should she lead the deer in order to hit it in the desired loca on?
40. A baseball player hits a ball at 100mph, with an ini al height
of 3 and an angle of eleva on of 20 , at Bostons Fenway
Park. The ball ies towards the famed Green Monster, a
wall 37 high located 310 from home plate.
(a) Show that as hit, the ball hits the wall.
(b) Show that if the angle of eleva on is 21 , the ball
clears the Green Monster.
41. A Cessna ies at 1000 at 150mph and drops a box of supplies to the professor (and his wife) on an island. Ignoring
wind resistance, how far horizontally will the supplies travel
before they land?
42. A football quarterback throws a pass from a height of 6 ,
intending to hit his receiver 20yds away at a height of 5 .
(a) At what angle of eleva on should she hold the crossbow to hit her target?
(a) If the ball is thrown at a rate of 50mph, what angle of

eleva on is needed to hit his intended target?
(b) If the deer is moving perpendicularly to her line of

sight at a rate of 20mph, by approximately how much
(b) If the ball is thrown at with an angle of eleva on of

8 , what ini al ball speed is needed to hit his target?
11.4 Unit Tangent and Normal Vectors

Unit Tangent Vector
Given a smooth vectorvalued func onr(t), we dened in Deni on 71 that
any vector parallel to r (t0 ) is tangent to the graph of r(t) at t = t0 . It is o en
useful to consider just the direc on of r (t) and not its magnitude. Therefore
we are interested in the unit vector in the direc on of r (t). This leads to a
deni on.
Deni on 74
Unit Tangent Vector
Let r(t) be a smooth func on on an open interval I. The unit tangent

vector T(t) is
1
T(t) =
r (t).
||r (t) ||
. Example 376
.Compu ng the unit tangent vector
Letr(t) = 3 cos t, 3 sin t, 4t. Find T(t) and compute T(0) and T(1).
We apply Deni on 74 to nd T(t).
T(t) =
1
r (t)
||r (t) ||
= (
3 sin t, 3 cos t, 4
(
)2
+ 3 cos t + 42
3
4
3
.
= sin t, cos t,
5
5
5
3 sin t
)2
We can now easily compute T(0) and T(1):
T(0) = 0, 3 , 4 ; T(1) = 3 sin 1, 3 cos 1, 4 0.505, 0.324, 0.8 .

5 5
5
5
5
These are plo ed in Figure 11.20 with their ini al points atr(0) andr(1), respecvely. (They look rather short since they are only length 1.)
The unit tangent vector T(t) always has a magnitude of 1, though it is somemes easy to doubt that is true. We can help solidify this thought in our minds
by compu ng || T(1) ||:
|| T(1) || (0.505)2 + 0.3242 + 0.82 = 1.000001.
10
2
2
10
Figure 11.20: Plo ng unit tangent vectors in Example 376.
Notes:
631

We have rounded in our computa on of T(1), so we dont get 1 exactly. We
leave it to the reader to use the exact representa on of T(1) to verify it has
length 1. .
In many ways, the previous example was too nice. It turned out thatr (t)
was always of length 5. In the next example the length ofr (t) is variable, leaving us with a formula that is not as clean.
y
. Example 377
Compu
ng the unit tangent vector

Letr(t) = t2 t, t2 + t . Find T(t) and compute T(0) and T(1).
We ndr (t) = 2t 1, 2t + 1, and
||r (t) || = (2t 1)2 + (2t + 1)2 = 8t2 + 2.
Therefore
x
2
Figure 11.21: Plo ng unit tangent vectors in Example 377.
2t 1
2t + 1
T(t) = 1
2t 1, 2t + 1 =
,
.
8t2 + 2
8t2 + 2
8t2 + 2
When t = 0, we have T(0) = 1/ 2, 1/ 2 ; when t = 1, we have T(1) =

1/ 10, 3/ 10 . We leave it to the reader to verify each of these is a unit vector. They are plo ed in Figure 11.21 .
Unit Normal Vector

y
Figure 11.22: Given a direc on in the

plane, there are always two direc ons orthogonal to it.
Note: T(t) is a unit vector, by deni on.

This does not imply that T (t) is also a unit
vector.
Just as knowing the direc on tangent to a path is important, knowing a direcon orthogonal to a path is important. When dealing with real-valued func ons,
we dened the normal line at a point to the be the line through the point that
was perpendicular to the tangent line at that point. We can do a similar thing
with vectorvalued func ons. Givenr(t) in R2 , we have 2 direc ons perpendicular to the tangent vector, as shown in Figure 11.22. It is good to wonder Is one
of these two direc ons preferable over the other?
Given r(t) in R3 , however, there are innite vectors orthogonal to the tangent vector at a given point. Again, we might wonder Is one of these innite
choices preferable over the others? Is one of these the right choice?
The answer in both R2 and R3 is Yes, there is one vector that is not only
preferable, it is the right one to choose. Recall Theorem 93, which states that
if r(t) has constant length, then r(t) is orthogonal to r (t) for all t. We know
T(t), the unit tangent vector, has constant length. Therefore T(t) is orthogonal
to T (t).
Well see that T (t) is more than just a convenient choice of vector that is
orthogonal tor (t); rather, it is the right choice. Since all we care about is the
direc on, we dene this newly found vector to be a unit vector.
Notes:
632
Deni on 75
Unit Normal Vector
Letr(t) be a vectorvalued func on where the unit tangent vector, T(t),

is smooth on an open interval I. The unit normal vector N(t) is
N(t) =
1
T (t).
|| T (t) ||
. Example 378
Compu ng the unit normal vector
Letr(t) = 3 cos t, 3 sin t, 4t as in Example 376. Sketch both T(/2) and N(/2)
with ini al points atr(/2).
S
In Example 376, we found T(t) = (3/5) sin t, (3/5) cos t, 4/5 .
Therefore
T (t) = 3 cos t, 3 sin t, 0

5
5
10
and
3
|| T (t) || = .
5
Thus
N(t) = T (t) = cos t, sin t, 0 .

3/5
We compute T(/2) = 3/5, 0, 4/5 and N(/2) = 0, 1, 0. These are

sketched in Figure 11.23. .
2
10
2
2
Figure 11.23: Plo ng unit tangent and

normal vectors in Example 11.23.
The previous example was once again too nice. In general, the expression
for T(t) contains frac ons of squareroots, hence the expression of T (t) is very
messy. We demonstrate this in the next example.
. Example 379
.Compu
ng the unit normal vector

Letr(t) = t2 t, t2 + t as in Example 377. Find N(t) and sketchr(t) with the
unit tangent and normal vectors at t = 1, 0 and 1.
S
In Example 377, we found
T(t) =
2t 1
2t + 1
,
2
8t + 2
8t2 + 2
Finding T (t) requires two applica ons of the Quo ent Rule:
Notes:
633
(
)
8t2 + 2(2) (2t 1) 12 (8t2 + 2)1/2 (16t)
T (t) =
,
8t2 + 2
(
)
8t2 + 2(2) (2t + 1) 12 (8t2 + 2)1/2 (16t)
8t2 + 2
4(2t + 1)
4(1 2t)
=
,
3/2
3/2
(8t2 + 2)
(8t2 + 2)
This is not a unit vector; to nd N(t), we need to divide T (t) by its magnitude.
|| T (t) || =
=
=
y
6
16(2t + 1)2
16(1 2t)2
+
2
3
(8t + 2)
(8t2 + 2)3
16(8t2 + 2)
(8t2 + 2)3
8t2
4
.
+2
Finally,
1
4(2t
+
1)
4(1
2t)
N(t) =
,
4/(8t2 + 2) (8t2 + 2)3/2 (8t2 + 2)3/2
2t + 1
2t 1
=
,
.
8t2 + 2
8t2 + 2
x
2
Figure 11.24: Plo ng unit tangent and

normal vectors in Example 379.
Using this formula for N(t), we compute the unit tangent and normal vectors
for t = 1, 0 and 1 and sketch them in Figure 11.24. .
The nal result for N(t) in Example 379 is suspiciously similar to T(t). There
is a clear reason for this. If u = u1 , u2 is a unit vector in R2 , then the only unit
vectors orthogonal to u are u2 , u1 and u2 , u1 . Given T(t), we can quickly
determine N(t) if we know which term to mul ply by (1).
Consider again Figure 11.24, where we have plo ed some unit tangent and
normal vectors. Note how N(t) always points inside the curve, or to the concave side of the curve. This is not a coincidence; this is true in general. Knowing
the direc on thatr(t) turns allows us to quickly nd N(t).
Notes:
634
Theorem 97
Unit Normal Vectors in R2
Let r(t) be a vectorvalued func on in R2 where T (t) is smooth on an

open interval I. Let T(t) = t1 , t2 . Then N(t) is either
N(t) = t2 , t1
or N(t) = t2 , t1 ,
whichever is the vector that points to the concave side of the graph ofr.
Applica on to Accelera on
Let r(t) be a posi on func on. It is a fact (stated later in Theorem 98) that
accelera on, a(t), lies in the plane dened by T and N. That is, there are scalars
aT and aN such that
a(t) = aTT(t) + aN N(t).
The scalar aT measures how much accelera on is in the direc on of travel, that
is, it measures the component of accelera on that aects the speed. The scalar
aN measures how much accelera on is perpendicular to the direc on of travel,
that is, it measures the component of accelera on that aects the direc on of
travel.
We can nd aT using the orthogonal projec on of a(t) onto T(t) (review Defini on 59 in Sec on 10.3 if needed). Recalling that since T(t) is a unit vector,
T(t) T(t) = 1, so we have
proj T(t) a(t) =
(
)
a(t) T(t)
T(t) = a(t) T(t) T(t).
T(t) T(t)
{z
}
|
aT
Note: Keep in mind that both aT and

aN are func ons of t; that is, the scalar
changes depending on t. It is conven on
to drop the (t) nota on from aT (t) and
simply write aT .
Thus the amount of a(t) in the direc on of T(t) is aT = a(t) T(t). The same
logic gives aN = a(t) N(t).
While this is a ne way of compu ng aT , there are simpler ways of nding aN
(as nding N itself can be complicated). The following theorem gives alternate
formulas for aT and aN .
Notes:
635

.
Theorem 98
Accelera on in the Plane Dened by T and N
Let r(t) be a posi on func on with accelera on a(t) and unit tangent and
normal vectors T(t) and N(t). Then a(t) lies in the plane dened by T(t) and
N(t); that is, there exists scalars aT and aN such that

Moreover,
)
d(
|| v(t) ||
dt
|| a(t) v(t) ||
aN = a(t) N(t) = || a(t) ||2 a2T =
= || v(t) || || T (t) ||
|| v(t) ||
aT = a(t) T(t) =
)
d(
|| v(t) || . This measures the rate of
Note the second formula for aT :
dt
change of speed, which again is the amount of accelera on in the direc on of
travel.
. Example 380
Compu ng aT and aN
Letr(t) = 3 cos t, 3 sin t, 4t as in Examples 376 and 378. Find aT and aN .
The previous examples give a(t) = 3 cos t, 3 sin t, 0
and
T(t) =
3
3
4
sin t, cos t,
5
5
5
and N(t) = cos t, sin t, 0 .
We can nd aT and aN directly with dot products:

9
9
aT = a(t) T(t) = cos t sin t cos t sin t + 0 = 0.
5
5
aN = a(t) N(t) = 3 cos2 t + 3 sin2 t + 0 = 3.
Thus a(t) = 0T(t) + 3N(t) = 3N(t), which is clearly the case.
What is the prac cal interpreta on of these numbers? aT = 0 means the
object is moving at a constant speed, and hence all accelera on comes in the
form of direc on change. .
. Example 381
.Compu
ng aT and aN
Letr(t) = t2 t, t2 + t as in Examples 377 and 379. Find aT and aN .
Notes:
636

S
T(t) =
The previous examples give a(t) = 2, 2 and
2t 1
2t + 1
2t + 1
2t 1
,
and N(t) =
,
.
8t2 + 2
8t2 + 2
8t2 + 2
8t2 + 2
While we can compute aN using N(t), we instead demonstrate using another

formula from Theorem 98.
4t + 2
8t
4t 2
+
=
.
aT = a(t) T(t) =
8t2 + 2
8t2 + 2
8t2 + 2
(
)2
8t
4
2
2
aN = || a(t) || aT = 8
=
.
2
8t + 2
8t2 + 2
16
4
When t = 2, aT = 2.74 and aN = 0.69. We interpret this to
34
34
mean that at t = 2, the par cle is acclera ng mostly by increasing speed, not
by changing direc on. As the path near t = 2 is rela vely straight, this should
make intui ve sense. Figure 11.25 gives a graph of the path
for reference.
Contrast this with t = 0, where aT = 0 and aN = 4/ 2 2.82. Here the
par cles speed is not changing and all accelera on is in the form of direc on
change. .
y
6
t=2
r(t)
t=0
x
2
Figure 11.25: Graphing r(t) in Example

381.
. Example 382
.Analyzing projec le mo on
A ball is thrown from a height of 240 with an ini al velocity of 64 /s with an
angle of eleva on of 30 . Find the posi on func onr(t) for the ball and analyze
aT and aN .
y
Using Key Idea 54 of Sec on 11.3 we form the posi on func-
on of the ball:
r(t) =
t=1
t=2
t=0
)
(
)
64 cos 30 t, 16t2 + 64 sin 30 t + 240 ,
which we plot in Figure 11.26.

From this we ndv(t) = 64 cos 30 , 32t + 64 sin 30 anda(t) = 0, 32.
Compu ng T(t) is not dicult, and with some simplica on we nd
1t
3
T(t) =
,
.
t2 2t + 4
t2 2t + 4
With a(t) as simple as it is, nding aT is also simple:
aT = a(t) T(t) =
t=3
200
t=4
100
100
200
t=5
x
300
Figure 11.26: Plo ng the posi on of a

thrown ball, with 1s increments shown.
32t 32
.
t2 2t + 4
Notes:
637
t
0
1
2
3
4
5
aT
16
0
16
24.2
27.7
29.4
aN
27.7
32
27.7
20.9
16
12.7
Figure 11.27: A table of values of aT and

aN in Example 382.
We choose to not nd N(t) and nd aN through the formula aN = || a(t) ||2 a2T :
(
)2
32t 32
32 3
aN = 322
=
.
t2 2t + 4
t2 2t + 4
Figure 11.27 gives a table of values of aT and aN . When t = 0, we see the
balls speed is decreasing; when t = 1 the speed of the ball is unchanged. This
corresponds to the fact that at t = 1 the ball reaches its highest point.
A er t = 1 we see that aN is decreasing in value. This is because as the ball
falls, its path becomes straighter and most of the accelera on is in the form of
speeding up the ball, and not in changing its direc on. .
Our understanding of the unit tangent and normal vectors is aiding our understanding of mo on. The work in Example 382 gave quan ta ve analysis of
what we intui vely knew.
The next sec on provides two more important steps towards this analysis.
We currently describe posi on only in terms of me. In everyday life, though,
we o en describe posi on in terms of distance (The gas sta on is about 2 miles
ahead, on the le .). The arc length parameter allows us to reference a par cles
posi on in terms of distance traveled.
We also intui vely know that some paths are straighter than others and
some are curvier than others, but we lack a measurement of curviness. The
arc length parameter provides a way for us to compute curvature, a quan ta ve
measurement of how curvy a curve is.
Notes:
638
Exercises 11.4
Terms and Concepts
1. If T(t) is a unit tangent vector, what is || T(t) ||?
2. If
N(t) is a unit normal vector, what is
N(t) r (t)?
3. The accelera on vector a(t) lies in the plane dened by
what two vectors?
4. aT measures how much the accelera on is aec ng the
of an object.
Problems
In Exercises 17 20 , a posi on func on r(t) is given along

with its unit tangent vector T(t) evaluated at t = a, for some
value of a.
(a) Conrm that T(a) is as stated.
17.
18.
19.
(b) Using a graph ofr(t) and Theorem 97, nd

N(a).
5
3
r(t) = 3 cos t, 5 sin t; T(/4) = ,
.
34
34
1
1
2
r(t) = t, 2
; T(1) = , .
t +1
5
5
2
1
r(t) = (1 + 2 sin t) cos t, sin t; T(0) = , .

5
5
1
1
r(t) = cos3 t, sin3 t ; T(/4) = , .
2
2
In Exercises 5 8 , givenr(t), nd T(t) and evaluate it at the

indicated value of t.
5. r(t) = 2t2 , t2 t , t = 1
20.
6. r(t) = t, cos t, t = /4
7. r(t) = cos3 t, sin3 t , t = /4
21. r(t) = 4t, 2 sin t, 2 cos t
8. r(t) = cos t, sin t, t =

In Exercises 9 12 , nd the equa on of the line tangent to
the curve at the indicated t-value using the unit tangent vector. Note: these are the same problems as in Exercises 5
8.
9. r(t) = 2t2 , t2 t , t = 1
10. r(t) = t, cos t, t = /4
11. r(t) = cos3 t, sin3 t , t = /4

12. r(t) = cos t, sin t, t =
In Exercises 13 16 , nd
N(t) using Deni on 75. Conrm
the result using Theorem 97.
13. r(t) = 3 cos t, 3 sin t
14. r(t) = t, t2
15. r(t) = cos t, 2 sin t
16. r(t) = et , et
In Exercises 21 24 , nd
N(t).
22. r(t) = 5 cos t, 3 sin t, 4 sin t

23. r(t) = a cos t, a sin t, bt;
a>0
24. r(t) = cos(at), sin(at), t

In Exercises 25 30 , nd aT and aN givenr(t). Sketchr(t) on
the indicated interval, and comment on the rela ve sizes of
aT and aN at the indicated t values.
25. r(t) = t, t2 on [1, 1]; consider t = 0 and t = 1.

26. r(t) = t, 1/t on (0, 4]; consider t = 1 and t = 2.
27. r(t) = 2 cos t, 2 sin t on [0, 2]; consider t = 0 and
t = /2.
28. r(t) = cos(t2 ), sin(t2 ) on (0, 2]; consider t =

/2
and t = .
29. r(t) = a cos t, a sin t, bt on [0, 2], where a, b > 0; consider t = 0 and t = /2.
30. r(t) = 5 cos t, 4 sin t, 3 sin t on [0, 2]; consider t = 0
and t = /2.
y
6
t=2
t=1
r(t)
t=0
(a)
y
6
s=6
s=5
s=4
s=3
s=2
r(s)
s=1
s=0
2
x
4
(b)
In normal conversa on we describe posi on in terms of both me and distance.

For instance, imagine driving to visit a friend. If she calls and asks where you
are, you might answer I am 20 minutes from your house, or you might say I
am 10 miles from your house. Both answers provide your friend with a general
idea of where you are.
Currently, our vectorvalued func ons have dened points with a parameter
t,
which
we o en take to represent me. Consider Figure 11.28 (a), wherer(t) =
2
t t, t2 + t is graphed and the points corresponding to t = 0, 1 and 2 are
shown. Note how the arc length between t = 0 and t = 1 is smaller than the
arc length between t = 1 and t = 2; if the parameter t is me andr is posi on,
we can say that the par cle traveled faster on [1, 2] than on [0, 1].
Now consider Figure 11.28 (b), where the same graph is parametrized by a
dierent variable s. Points corresponding to s = 0 through s = 6 are plo ed.
The arc length of the graph between each adjacent pair of points is 1. We can
view this parameter s as distance; that is, the arc length of the graph from s = 0
to s = 3 is 3, the arc length from s = 2 to s = 6 is 4, etc. If one wants to nd the
point 2.5 units from an ini al loca on (i.e., s = 0), one would compute r(2.5).
This parameter s is very useful, and is called the arc length parameter.
How do we nd the arc length parameter?
Start with any parametriza on of r. We can compute the arc length of the
graph ofr on the interval [0, t] with
Figure 11.28: Introducing the arc length

parameter.
arc length =
||r (u) || du.
We can turn this into a func on: as t varies, we nd the arc length s from 0 to t.
This func on is
s(t) =
||r (u) || du.
(11.1)
This establishes a rela onship between s and t. Knowing this rela onship
explicitly, we can rewriter(t) as a func on of s: r(s). We demonstrate this in an
example.
. Example 383
.Finding the arc length parameter
Letr(t) = 3t 1, 4t + 2. Parametrizer with the arc length parameter s.
S
Using Equa on (11.1), we write
s(t) =
0
Notes:
640
||r (u) || du.

We can integrate this, explicitly nding a rela onship between s and t:
t
s(t) =
||r (u) || du
0
t
=
32 + 42 du
0
t
=
5 du
s=5
t=1
= 5t.
s=4
Since s = 5t, we can write t = s/5 and replace t inr(t) with s/5:
4
3
r(s) = 3(s/5) 1, 4(s/5) + 2 =
s 1, s + 2 .
5
5
s=3
s=2
s=1
t=0
Clearly, as shown in Figure 11.29, the graph of r is a line, where t = 0 corresponds to the point (1, 2). What point on the line is 2 units away from this
ini al point? We nd it with s(2) = 1/5, 18/5.
Is the point (1/5, 18/5) really 2 units away from (1, 2)? We use the Distance Formula to check:
(
)2 (
)2
1
18
36 64
d=
(1) +
2 =
+
= 4 = 2.
5
5
25 25
2
s=0
x
Figure 11.29: Graphing r in Example 383

with parameters t and s.
Yes, s(2) is indeed 2 units away, in the direc on of travel, from the ini al point. .
. we were able to establish
Things worked out very nicely in Example 383;
directly that s = 5t. Usually, the arc length parameter is much more dicult to
describe in terms of t, a result of integra ng a squareroot. There are a number
of things that we can learn about the arc length parameter from Equa on (11.1),
though, that are incredibly useful.
First, take the deriva ve of s with respect to t. The Fundamental Theorem
of Calculus (see Theorem 39) states that
ds
= s (t) = ||r (t) ||.
dt
(11.2)
Le ng t represent me and r(t) represent posi on, we see that the rate of
change of s with respect to t is speed; that is, the rate of change of distance
traveled is speed, which should match our intui on.
The Chain Rule states that
dr
dr ds
=
dt
ds dt
r (t) = r (s) ||r (t) ||.
Notes:
641

Solving forr (s), we have
r (s) =
r (t)
= T(t),
||r (t) ||
(11.3)
where T(t) is the unit tangent vector. Equa on 11.3 is o en misinterpreted, as

one is tempted to think it states r (t) = T(t), but there is a big dierence betweenr (s) andr (t). The key to take from it is thatr (s) is a unit vector. In fact,
the following theorem states that this characterizes the arc length parameter.
.
Theorem 99
Arc Length Parameter
Let r(s) be a vectorvalued func on. The parameter s is the arc length
parameter if, and only if, ||r (s) || = 1.
Curvature
x
1
(a)
y
Deni on 76
Curvature
Letr(s) be a vectorvalued func on where s is the arc length parameter.

The curvature of the graph ofr(s) is

dT

=
= T (s) .
ds
x
1
Consider points A and B on the curve graphed in Figure 11.30 (a). One can
readily argue that the curve curves more sharply at A than at B. It is useful to use
a number to describe how sharply the curve bends; that number is the curvature
of the curve.
We derive this number in the following way. Consider Figure 11.30 (b), where
unit tangent vectors are graphed around points A and B. No ce how the direcon of the unit tangent vector changes quite a bit near A, whereas it does not
change as much around B. This leads to an important concept: measuring the
rate of change of the unit tangent vector with respect to arc length gives us a
measurement of curvature.
(b)
Figure 11.30: Establishing the concept of
curvature.
Ifr(s) is parametrized by the arc length parameter, then

T(s) =
Notes:
642
r (s)
||r (s) ||
and N(s) =
T (s)
.
|| T (s) ||

Having dened || T (s) || = , we can rewrite the second equa on as
T (s) = N(s).
(11.4)
We already knew that T (s) is in the same direc on as N(s); that is, we can think
of T(s) as being pulled in the direc on of N(s). How hard is it being pulled?
By a factor of . When the curvature is large, T(s) is being pulled hard and the
direc on of T(s) changes rapidly. When is small, T(s) is not being pulled hard
and hence its direc on is not changing rapidly.
We use Deni on 76 to nd the curvature of the line in Example 383.
. Example 384
Finding the curvature of a line
Use Deni on 76 to nd the curvature ofr(t) = 3t 1, 4t + 2.
In Example 383, we found that the arc length parameter was
S
dened by s = 5t, so r(s) = 3t/5 1, 4t/5 + 2 parametrized r with the arc
length parameter. To nd , we need to nd T (s).
T(s) = r (s) (recall this is a unit vector)
= 3/5, 4/5 .
Therefore
T (s) = 0, 0
and

= T (s) = 0.
It probably comes as no surprise that the curvature of a line is 0. (How curvy
is a line? It is not curvy at all.) .
While the deni on of curvature is a beau ful mathema cal concept, it is
nearly impossible to use most of the me; wri ng r in terms of the arc length
parameter is generally very hard. Fortunately, there are other methods of calcula ng this value that are much easier. There is a tradeo: the deni on is
easy to understand though hard to compute, whereas these other formulas
are easy to compute though hard to understand why they work.
Notes:
643

.
Theorem 100
Formulas for Curvature
Let C be a smooth curve on an open interval I in the plane or in space.

1. If C is dened by y = f(x), then
= (
|f (x)|
.
(
)2 )3/2
1 + f (x)
2. If C is dened as a vectorvalued func on in the plane, r(t) =

x(t), y(t), then
= (
|x y x y |
)3/2 .
(x )2 + (y )2
3. If C is dened in space by a vectorvalued func onr(t), then

=
a(t) N(t)
|| T (t) ||
||r (t) r (t) ||
=
=
.
3
||r (t) ||
||r (t) ||
|| v(t) ||2
We prac ce using these formulas.

. Example 385
Finding the curvature of a circle
Find the curvature of a circle with radius r, dened byr(t) = r cos t, r sin t.
Before we start, we should expect the curvature of a circle
S
to be constant, and not dependent on t. (Why?)
We compute using the second part of Theorem 100.
=
|(r sin t)(r sin t) (r cos t)(r cos t)|

(
)3/2
(r sin t)2 + (r cos t)2
=(
r2 (sin2 t + cos2 t)
)3/2
r2 (sin2 t + cos2 t)
r2
1
= .
3
r
r
We have found that a circle with radius r has curvature = 1/r. .
Notes:
644

Example 385 gives a great result. Before this example, if we were told The
curve has a curvature of 5 at point A, we would have no idea what this really meant. Is 5 big does is correspond to a really sharp turn, or a not-sosharp turn? Now we can think of 5 in terms of a circle with radius 1/5. Knowing
the units (inches vs. miles, for instance) allows us to determine how sharply the
curve is curving.
y
Let a point P on a smooth curve C be given, and let be the curvature of the
curve at P. A circle that:
passes through P,
lies on the concave side of C,

has a common tangent line as C at P and
has radius r = 1/ (hence has curvature )

.
is the oscula ng circle, or circle of curvature, to C at P, and r is the radius of curvature. Figure 11.31 shows the graph of the curve seen earlier in Figure 11.30
and its oscula ng circles at A and B. A sharp turn corresponds to a circle with
a small radius; a gradual turn corresponds to a circle with a large radius. Being
able to think of curvature in terms of the radius of a circle is very useful. (The
word oscula ng comes from a La n word related to kissing; an oscula ng circle kisses the graph at a par cular point. Many beau ful ideas in mathema cs
have come from studying the oscula ng circles to a curve.)
x
1
Figure 11.31: Illustra ng the oscula ng

circles for the curve seen in Figure 11.30.
.
. Example 386
Finding curvature
2
Find the curvature of the parabola dened by y = x at the vertex and at x = 1.
y
10
We use the rst formula found in Theorem 100.

(x) = (
|2|
)3/2
1 + (2x)2
2
=(
)3/2 .
1 + 4x2
x
10
At the vertex (x = 0), the curvature is = 2. At x = 1, the curvature

is = 2/(5)3/2 0.179. So at x = 0, the curvature of y = x2 is that of
a circle of radius 1/2; at x = 1, the curvature is that of a circle with radius
1/0.179 5.59. This is illustrated in Figure 11.32. At x = 3, the curvature is
0.009; the graph is nearly straight as the curvature is very close to 0. .
.
Figure 11.32: Examining the curvature of
y = x2 .
Notes:
645

. Example 387
Finding curvature
Find where the curvature ofr(t) = t, t2 , 2t3 is maximized.
We use the third formula in Theorem 100 asr(t) is dened

S
in space. We leave it to the reader to verify that
r (t) = 1, 2t, 6t2 , r (t) = 0, 2, 12t , and r (t)r (t) = 12t2 , 12t, 2 .
Thus
||r (t) r (t) ||
||r (t) ||3
2
|| 12t , 12t, 2 ||
=
|| 1, 2t, 6t2 ||3
144t4 + 144t2 + 4
= (
)3
1 + 4t2 + 36t4
(t) =
(a)
z
y
x
.
.
(b)
Figure 11.33: Understanding the curvature of a curve in space.
While this is not a par cularly nice formula, it does explictly tell us what the
curvature is at a given t value. To maximize (t), we should solve (t) = 0 for
t. This is doable, but very me consuming. Instead, consider the graph of (t)
as given in Figure 11.33 (a). We see that is maximized at two t values; using a
numerical solver, we nd these values are t 0.189. In part (b) of the gure
we graphr(t) and indicate the points where curvature is maximized. .
Curvature and Mo on
Let r(t) be a posi on func on of an object, with velocity v(t) = r (t) and
accelera on a(t) = r (t). In Sec on 11.4 we established that accelera on is in
the plane formed by T(t) and N(t), and that we can nd scalars aT and aN such
that
Theorem 98 gives formulas for aT and aN :
aT =
)
d(
|| v(t) ||
and
dt
aN =
|| v(t) a(t) ||
.
|| v(t) ||
We understood that the amount of accelera on in the direc on of T relates only

to how the speed of the object is changing, and that the amount of accelera on
in the direc on of N relates to how the direc on of travel of the object is changing. (That is, if the object travels at constant speed, aT = 0; if the object travels
in a constant direc on, aN = 0.)
Notes:
646

In Equa on (11.2) at the beginning of this sec on, we found s (t) = ||v(t) ||.
We can combine this fact with the above formula for aT to write
)
d(
d( )
aT =
|| v(t) || =
s (t) = s (t).
dt
dt
Since s (t) is speed, s (t) is the rate at which speed is changing with respect to
me. We see once more that the component of accelera on in the direc on of
travel relates only to speed, not to a change in direc on.
Now compare the formula for aN above to the formula for curvature in Theorem 100:
aN =
|| v(t) a(t) ||
|| v(t) ||
and
||r (t) r (t) ||

|| v(t) a(t) ||
=
.
3
||r (t) ||
|| v(t) ||3
Thus
aN = || v(t) ||2
(
)2
= s (t)
(11.5)
This last equa on shows that the component of accelera on that changes
the objects direc on is dependent on two things: the curvature of the path and
the speed of the object.
Imagine driving a car in a clockwise circle. You will naturally feel a force pushing you towards the door (more accurately, the door is pushing you as the car
is turning and you want to travel in a straight line). If you keep the radius of
the circle constant but speed up, the door pushes harder against you (aN has
increased). If you keep your speed constant but ghten the turn, once again the
door will push harder against you.
Pu ng our new formulas for aT and aN together, we have
a(t) = s (t)T(t) + || v(t) ||2 N(t).
This is not a par cularly prac cal way of nding aT and aN , but it reveals some
great concepts about how accelera on interacts with speed and the shape of a
curve.
. Example 388
.Curvature and road design
The minimum radius of the curve in a highway cloverleaf is determined by the
opera ng speed, as given in the table in Figure 11.34. For each curve and speed,
compute aN .
Using Equa on (11.5), we can compute the accelera on
S
normal to the curve in each case. We start by conver ng each speed from miles
per hour to feet per second by mul plying by 5280/3600.
Opera ng
Speed (mph)
35
40
45
Minimum
Radius ( )
310
430
540
Figure 11.34: Opera ng speed and minimum radius in highway cloverleaf design.
Notes:
647
51.33 /s,
35mph, 310
= 1/310
aN = || v(t) ||
)2
1 (
=
51.33
310
= 8.50 /s2 .
2
58.67 /s, = 1/430

)2
1 (
aN =
58.67
430
= 8.00 /s2 .
40mph, 430
66 /s, = 1/540
1 ( )2
aN =
66
540
= 8.07 /s2 .
45mph,540
Note that each accelera on is similar; this is by design. Considering the classic
Force = mass accelera on formula, this accelera on must be kept small in
order for the res of a vehicle to keep a grip on the road. If one travels on a
turn of radius 310 at a rate of 50mph, the accelera on is double, at 17.35 /s2 .
If the accelera on is too high, the fric onal force created by the res may not be
enough to keep the car from sliding. Civil engineers rou nely compute a safe
design speed, then subtract 5-10mph to create the posted speed limit for addional safety. .
We end this chapter with a reec on on what weve covered. We started
with vectorvalued func ons, which may have seemed at the me to be just
another way of wri ng parametric equa ons. However, we have seen that the
vector perspec ve has given us great insight into the behavior of func ons and
the study of mo on. Vectorvalued posi on func ons convey displacement,
distance traveled, speed, velocity, accelera on and curvature informa on, each
of which has great importance in science and engineering.
Notes:
648
Exercises 11.5
17. C is dened byr(t) = t2 1, t3 t ; points given at t = 0

and t = 5.
Terms and Concepts

1. It is common to describe posi on in terms of both
and/or
.
2. A measure of the curviness of a curve is
3. Give two shapes with constant curvature.

4. Describe in your own words what an oscula ng circle is.
5. Complete the iden ty: T (s) =
N(s).
18. C is dened by r(t) = tan t, sec t; points given at t = 0

and t = /6.
19. C is dened byr(t) = 4t + 2, 3t 1, 2t + 5; points given
at t = 0 and t = 1.
20. C is dened byr(t) = t3 t, t3 4, t2 1 ; points given

at t = 0 and t = 1.
6. Given a posi on func on r(t), how are aT and aN aected

by the curvature?
21. C is dened by r(t) = 3 cos t, 3 sin t, 2t; points given at

t = 0 and t = /2.
Problems
22. C is dened by r(t) = 5 cos t, 13 sin t, 12 cos t; points

given at t = 0 and t = /2.
In Exercises 7 10 , a posi on func on r(t) is given, where

t = 0 corresponds to the ini al posi on. Find the arc length
parameter s, and rewriter(t) in terms of s; that is, ndr(s).
23. y =
7. r(t) = 2t, t, 2t
8. r(t) = 7 cos t, 7 sin t
9. r(t) = 3 cos t, 3 sin t, 2t
10. r(t) = 5 cos t, 13 sin t, 12 cos t
In Exercises 11 22 , a curve C is described along with 2 points
on C. Using a sketch, determine at which of these points the
curvature is greater. Find the curvature of C.
11. C is dened by y = x3 x; points given at x = 0 and
x = 1/2.
12. C is dened by y =
x = 2.
In Exercises 23 26 , nd the value of x or t where curvature

is maximized.
1
; points given at x = 0 and
x2 + 1
13. C is dened by y = cos x; points given at x = 0 and

x = /2.
14. C is dened by y = 1 x2 on (1, 1); points given at

x = 0 and x = 1/2.
15. C is dened byr(t) = cos t, sin(2t); points given at t = 0
and t = /4.
16. C is dened by r(t) = cos2 t, sin t cos t ; points given at

t = 0 and t = /3.
1 3
x
6
24. y = sin x
25. r(t) = t2 + 2t, 3t t2

26. r(t) = t, 4/t, 3/t
In Exercises 27 30 , nd the radius of curvature at the indicated value.
27. y = tan x, at x = /4
28. y = x2 + x 3, at x = /4
29. r(t) = cos t, sin(3t), at t = 0
30. r(t) = 5 cos(3t), t, at t = 0
In Exercises 31 34 , nd the equa on of the oscula ng circle
to the curve at the indicated t-value.
31. r(t) = t, t2 , at t = 0
32. r(t) = 3 cos t, sin t, at t = 0
33. r(t) = 3 cos t, sin t, at t = /2
34. r(t) = t2 t, t2 + t , at t = 0
12: F
V
A func on of the form y = f(x) is a func on of a single variable; given a value

of x, we can nd a value y. Even the vectorvalued func ons of Chapter 11 are
singlevariable func ons; the input is a single variable though the output is a
vector.
There are many situa ons where a desired quan ty is a func on of two or
more variables. For instance, wind chill is measured by knowing the temperature
and wind speed; the volume of a gas can be computed knowing the pressure
and temperature of the gas; to compute a baseball players ba ng average, one
needs to know the number of hits and the number of atbats.
This chapter studies mul variable func ons, that is, func ons with more
than one input.
12.1 Introduc on to Mul variable Func ons

Deni on 77
Func on of Two Variables
Let D be a subset of R2 . A func on f of

two variables is a rule that assigns
.
each pair (x, y) in D a value z = f(x, y) in R. D is the domain of f; the set
of all outputs of f is the range.
. Example 389
.Understanding a func on of two variables
Let z = f(x, y) = x2 y. Evaluate f(1, 2), f(2, 1), and f(2, 4); nd the domain
and range of f.
S
Using the deni on f(x, y) = x2 y, we have:

f(1, 2) = 12 2 = 1
f(2, 1) = 22 1 = 3
f(2, 4) = (2)2 4 = 0
The domain is not specied, so we take it to be all possible pairs in R2 for which
Chapter 12 Func ons of Several Variables

f is dened. In this example, f is dened for all pairs (x, y), so the domain D of f
is R2 .
The output of f can be made as large or small as possible; any real number r
can be the output. (In fact, given any real number r, f(0, r) = r.) So the range
R of f is R. .
y
5
. Example 390
x2
y2
+
=1
9
4
Let f(x, y) =
x
The domain is all pairs (x, y) allowable as input in f. Because

S
2
2
of the squareroot, we need (x, y) such that 0 1 x9 y4 :
01
Figure 12.1: Illustra ng the domain of

f(x, y) in Example 390.
0.5
.
2
(a)
z
1
0.5
.
2
x 2
y2
x2
+
1
9
4
Graphing Func ons of Two Variables

2
x 2
(
)
The graph of a func on f of two variables is the set of all points x, y, f(x, y)
where (x, y) is in the domain of f. This creates a surface in space.
One can begin sketching a graph by plo ng points, but this has limita ons.
1
Consider Figure 12.2a where 25 points have been plo ed of f(x, y) = 2
.
x + y2 + 1
More points have been plo ed than one would reasonably want to do by hand,
yet it is not clear at all what the graph of the func on looks like. Technology allows us to plot lots of points, connect adjacent points with lines and add shading
to create a graph like Figure 12.2b which does a far be er job of illustra ng the
behavior of f.
While technology is readily available to help us graph func ons of two variables, there is s ll a paperandpencil approach that is useful to understand and
master as it, combined with highquality graphics, gives one great insight into
the behavior of a func on. This technique is known as sketching level curves.
(b)
Figure 12.2: Graphing a func on of two
variables.
652
x2
y2
9
4
The above equa on describes the interior of an ellipse as shown in Figure 12.1.
We can represent the domain D graphically with the gure; in set nota on, we
2
2
can write D = {(x, y) : x9 + y4 1}.
The range is the set of all possible output values. The squareroot ensures
that all output is 0. Since the x and y terms are squared, then subtracted, inside the squareroot, the largest output value comes at x = 0, y = 0: f(0, 0) =
1. Thus the range R is the interval [0, 1]. .
Understanding a func on of two variables

y2
x2
. Find the domain and range of f.
1
9
4
Notes:
Level Curves
It may be surprising to nd that the problem of represen ng a three dimensional surface on paper is familiar to most people (they just dont realize it).
Topographical maps, like the one shown in Figure 12.3, represent the surface
of Earth by indica ng points with the same eleva on with contour lines. The
eleva ons marked are equally spaced; in this example, each thin line indicates
an eleva on change in 50 increments and each thick line indicates a change
of 200 . When lines are drawn close together, eleva on changes rapidly (as
one does not have to travel far to rise 50 ). When lines are far apart, such as
near Aspen Campground, eleva on changes more gradually as one has to walk
farther to rise 50 .
Given a func on z = f(x, y), we can draw a topographical map of f by
drawing level curves (or, contour lines). A level curve at z = c is a curve in the
x-y plane such that for all points (x, y) on the curve, f(x, y) = c.
When drawing level curves, it is important that the c values are spaced equally
apart as that gives the best insight to how quickly the eleva on is changing.
Examples will help one understand this concept.
. Example 391
Let f(x, y) =
0.8 and 1.
Figure 12.3: A topographical map displays

eleva on by drawing contour lines, along
with the eleva on is constant.
Sample taken from the public domain USGS Digital Raster Graphics,
http://topmaps.usgs.gove/drg/.
.Drawing Level Curves

x2
y2
1
. Find the level curves of f for c = 0, 0.2, 0.4, 0.6,
9
4
Consider rst
c = 0. The level curve for c = 0 is the set of
2
2
all points (x, y) such that 0 = 1 x9 y4 . Squaring both sides quickly gives
us
x2
y2
+
= 1,
9
4
S
an ellipse centered at (0, 0) with horizontal major axis of length 6 and minor axis
of length 4. Thus for any point (x, y) on this curve, f(x, y) = 0.
Now consider the level curve for c = 0.2
x2
y2
0.2 = 1
9
4
y2
x2
0.04 = 1
9
4
x2
y2
+
= 0.96
9
4
y2
x2
+
= 1.
8.64 3.84
Notes:
653
y
c = 0.6
2
1
c=1
3
1
2
(a)
ellipses that are decreasing in size as c increases. A special case is when c = 1;

there the ellipse is just the point (0, 0).
The level curves are shown in Figure 12.4(a). Note how the level curves for
c = 0 and c = 0.2 are very, very close together: this indicates that f is growing
rapidly along those curves.
In Figure 12.4(b), the curves are drawn on a graph of f in space. Note how
the eleva ons are evenly spaced. Near the level curves of c = 0 and c = 0.2 we
can see that f indeed is growing quickly. .
.
2
x
This is also an ellipse, where a = 8.64 2.94 and b = 3.84 1.96.

In general, for z = c, the level curve is:
x2
y2
c= 1
9
4
2
2
x
y
c2 = 1
9
4
y2
x2
2
+
=1c
9
4
x2
y2
+
= 1,
9(1 c2 ) 4(1 c2 )
(b)
Figure 12.4: Graphing the level curves in
Example 391.
. Example 392
.Analyzing Level Curves

x+y
Let f(x, y) = 2
. Find the level curves for z = c.
x + y2 + 1
We begin by se ng f(x, y) = c for an arbitrary c and seeing
S
if algebraic manipula on of the equa on reveals anything signicant.
x2
x+y
=c
+ y2 + 1
x + y = c(x2 + y2 + 1).
We recognize this as a circle, though the center and radius are not yet clear. By
comple ng the square, we can obtain:
(
)2 (
)2
1
1
1
x
+ y
= 2 1,
2c
2c
2c
(
)
a circle centered at 1/(2c), 1/(2c) with radius 1/(2c2 ) 1, where |c| <
1/ 2. The level curves for c = 0.2, 0.4 and 0.6 are sketched in Figure 12.5(a).
To help illustrate eleva on, we use thicker lines for c values near 0, and dashed
lines indicate where c < 0.
Notes:
654

There is one special level curve, when c = 0. The level curve in this situa on
is x + y = 0, the line y = x.
In Figure 12.5(b) we see a graph of the surface. Note how the y-axis is pointing away from the viewer to more closely resemble the orienta on of the level
curves in (a).
Seeing the level curves helps us understand the graph. For instance, the
graph does not make it clear that one can walk along the line y = x without
eleva on change, though the level curve does. .
y
c = 0.2
4
c = 0.4
2
x
5
5
2
Func ons of Three Variables

.
.
We extend our study of mul variable func ons to func ons of three variables. (One can make a func on of as many variables as one likes; we limit our
study to three variables.)
c=0
(a)
z
Deni on 78
0.7
Func on of Three Variables
Let D be a subset of R3 . A func on

f of three variables is a rule that
.
assigns each triple (x, y, z) in D a value w = f(x, y, z) in R. D is the domain
of f; the set of all outputs of f is the range.
x
5
.
Note how this deni on closely resembles that of Deni on 77.
. Example 393
Understanding a func on of three variables

x + z + 3 sin y
Let f(x, y, z) =
. Evaluate f at the point (3, 0, 2) and nd the
x + 2y z
domain and range of f.
2
y
5
0.7
(b)
Figure 12.5: Graphing the level curves in

Example 392.
32 + 2 + 3 sin 0
= 11.
3 + 2(0) 2
As the domain of f is not specied, we take it to be the set of all triples (x, y, z)
for which f(x, y, z) is dened. As we cannot divide by 0, we nd the domain D is
S
f(3, 0, 2) =
D = {(x, y, z) | x + 2y z = 0}.
We recognize that the set of all points in R3 that are not in D form a plane in
space that passes through the origin (with normal vector 1, 2, 1).
We determine the range R is R; that is, all real numbers are possible outputs
of f. There is no set way of establishing this. Rather, to get numbers near 0 we
can let y = 0 and choose z x2 . To get numbers of arbitrarily large magnitude, we can let z x + 2y. .
Notes:
655
Level Surfaces
It is very dicult to produce a meaningful graph of a func on of three variables. A func on of one variable is a curve drawn in 2 dimensions; a func on of
two variables is a surface drawn in 3 dimensions; a func on of three variables is
a hypersurface drawn in 4 dimensions.
There are a few techniques one can employ to try to picture a graph of
three variables. One is an analogue of level curves: level surfaces. Given w =
f(x, y, z), the level surface at w = c is the surface in space formed by all points
(x, y, z) where f(x, y, z) = c.
. Example 394
Finding level surfaces
If a point source S is radia ng energy, the intensity I at a given point P in space
is inversely propor onal to the square of the distance between S and P. That is,
k
for some constant k.
when S = (0, 0, 0), I(x, y, z) = 2
x + y2 + z2
Let k = 1; nd the level surfaces of I.
We can (mostly) answer this ques on using common sense.
S
If energy (say, in the form of light) is emana ng from the origin, its intensity will
be the same all a points equidistant from the origin. That is, at any point on
the surface of a sphere centered at the origin, the intensity should be the same.
Therefore, the level surfaces are spheres.
We now nd this mathema cally. The level surface at I = c is dened by
c=
c
16.
8.
4.
2.
1.
0.5
0.25
0.125
0.0625
r
0.25
0.35
0.5
0.71
1.
1.41
2.
2.83
4.
Figure 12.6: A table of c values and the

corresponding radius r of the spheres of
constant value in Example 394.
1
.
+ y2 + z2
A small amount of algebra reveals

x2 + y2 + z2 =
1
.
c
Given an intensity c, the level surface I = c is a sphere of radius 1/ c, centered

at the origin.
Figure 12.6 gives a table of the radii of the spheres for given c values. Normally one would use equally spaced c values, but these values have been chosen
purposefully. At a distance of 0.25 from the point source, the intensity is 16; to
move to a point of half that intensity, one just moves out 0.1 to 0.35 not much
at all. To again halve the intensity, one moves 0.15, a li le more than before.
Note how each me the intensity if halved, the distance required to move
away grows. We conclude that the closer one is to the source, the more rapidly
the intensity changes. .
Notes:
656
x2
Exercises 12.1
Terms and Concepts
16. f(x, y) = x2 y2 ; c = 1, 0, 1
1. Give two examples (other than those given in the text) of

real world func ons that require more than one input.
17. f(x, y) = x y2 ; c = 2, 0, 2
18. f(x, y) =
1 x2 y2
; c = 2, 0, 2
2y 2x
3. Most people are familiar with the concept of level curves in

the context of
maps.
19. f(x, y) =
2x 2y
; c = 1, 0, 1
x2 + y2 + 1
4. T/F: Along a level curve, the output of a func on does not

change.
20. f(x, y) =
2. The graph of a func on of two variables is a
5. The analogue of a level curve for func ons of three variables is a level
.
6. What does it mean when level curves are close together?
Far apart?
Problems
y x3 1
; c = 3, 1, 0, 1, 3
x
21. f(x, y) = x2 + 4y2 ; c = 1, 2, 3, 4

22. f(x, y) = x2 + 4y2 ; c = 1, 2, 3, 4
Exercises 23 26, give the domain and range of the func ons
of three variables.
x
23. f(x, y, z) =
x + 2y 4z
7. f(x, y) = x2 + y2 + 2
1
1 x2 y2 z2
25. f(x, y, z) = z x2 + y2
8. f(x, y) = x + 2y
26. f(x, y, z) = z2 sin x cos y
Exercises 7 14, give the domain and range of the mul variable func on.
9. f(x, y) = x 2y
1
10. f(x, y) =
x + 2y
1
11. f(x, y) = 2
x + y2 + 1
12. f(x, y) = sin x cos y
13. f(x, y) = 9 x2 y2
14. f(x, y) =
1
x2 + y2 9
Exercises 15 22, describe in words and sketch the level

curves for the func on and given c values.
15. f(x, y) = 3x 2y; c = 2, 0, 2
24. f(x, y, z) =
Exercises 27 30, describe the level surfaces of the given funcons of three variables.
27. f(x, y, z) = x2 + y2 + z2
28. f(x, y, z) = z x2 + y2
x2 + y2
z
z
30. f(x, y, z) =
xy
29. f(x, y, z) =
31. Compare the level curves of Exercises 21 and 22. How are
they similar, and how are they dierent? Each surface is a
quadric surface; describe how the level curves are consistent with what we know about each surface.
12.2 Limits and Con nuity of Mul variable Func ons

y
P1
P2
We con nue with the pa ern we have established in this text: a er dening a
new kind of func on, we apply calculus ideas to it. The previous sec on dened
func ons of two and three variables; this sec on inves gates what it means for
these func ons to be con nuous.
We begin with a series of deni ons. We are used to open intervals such
as (1, 3), which represents the set of all x such that 1 < x < 3, and closed
intervals such as [1, 3], which represents the set of all x such that 1 x 3.
We need analogous deni ons for open and closed sets in the x-y plane.
.
(a)
Deni on 79
Open Disk, Boundary and Interior Points, Open and
Closed Sets, Bounded Sets
An open disk B in R2
centered at (x0 , y0 ) with radius r is the set of all
points (x, y) such that (x x0 )2 + (y y0 )2 < r.
P1
Let S be a set of points in R2 . A point P in R2 is a boundary point of S

if all open disks centered at P contain both points in S and points not in S.
P2
.
(b)
A point P in S is an interior point of S if there is an open disk centered at

P that contains only points in S.
A set S is open if every point in S is an interior point.
A set S is closed if it contains all of its boundary points.
P1
A set S is bounded if there is an M > 0 such that the open disk, centered at the origin, with radius M contains S. A set that is not bounded
is unbounded.
P2
.
(c)
Figure 12.7: Illustra ng open and closed

sets in the x-y plane.
Figure 12.7 shows several sets in the x-y plane. In each set, point P1 lies on
the boundary of the set as all open disks centered there contain both points in,
and not in, the set. In contrast, point P2 is an interior point for there is an open
disk centered there that lies en rely within the set.
The set depicted in Figure 12.7(a) is a closed set as it contains all of its boundary points. The set in (b) is open, for all of its points are interior points (or, equivalently, it does not contain any of its boundary points). The set in (c) is neither
open nor closed as it contains just some of its boundary points.
Notes:
658

. Example 395
Determining open/closed, bounded/unbounded
Determine if the domain of the func on f(x, y) =

or neither, and if it is bounded.
x2
9
y2
4
is open, closed,
y
This domain of this func on was found in Example 390 to be

2
2
y2
x2
D = {(x, y) | 9 + 4 1}, the region bounded by the ellipse x9 + y4 = 1. Since
the region includes the boundary (indicated by the use of ), the set contains
all of its boundary points and hence is closed. The region is bounded as a disk
of radius 4, centered at the origin, contains D. .
S
. Example 396
Determining open/closed, bounded/unbounded
1
Determine if the domain of f(x, y) = xy
is open, closed, or neither.
As we cannot divide by 0, we nd the domain to be D =
{(x, y) | x y = 0}. In other words, the domain is the set of all points (x, y) not
on the line y = x.
The domain is sketched in Figure 12.8. Note how we can draw an open disk
around any point in the domain that lies en rely inside the domain, and also
note how the only boundary points of the domain are the points on the line
y = x. We conclude the domain is an open set. The set is unbounded. .
S
.
Figure 12.8: Sketching the domain of the
func on in Example 396.
Limits
Recall a pseudodeni on of the limit of a func on of one variable: lim f(x) =
xc
L means that if x is really close to c, then f(x) is really close to L. A similar

pseudodeni on holds for func ons of two variables. Well say that
lim
(x,y)(x0 ,y0 )
f(x, y) = L
means if the point (x, y) is really close to the point (x0 , y0 ), then f(x, y) is really
close to L. The formal deni on is given below.
Deni on 80
Limit of a Func on of Two Variables
Let f(x, y) be a func on of two variables and let (x0 , y0 ) be a point in the
domain of f. The limit of f(x, y) as (x, y) approaches (x0 , y0 ) is L, denoted
lim
(x,y)(x0 ,y0 )
.
f(x, y) = L,
if, for every > 0 there is a > 0 such that if (x, y) is in the open disk
centered at (x0 , y0 ) with radius , then |f(x, y) L| < .
Notes:
659
z
L
(x0 , y0 , L)
Theorem 101
y
The concept behind Deni on 80 is sketched in Figure 12.9. Given > 0,

nd > 0 such that if (x, y) is any point in the open disk centered at (x0 , y0 ) in
the x-y plane with radius , then f(x, y) should be within of L.
Compu ng limits using this deni on is rather cumbersome. The following
theorem allows us to evaluate limits much more easily.
Let b, x0 , y0 , L and K be real numbers, let n be a posi ve integer, and let

f and g be func ons with the following limits:
lim
(x0 , y0 , 0)
Figure 12.9: Illustra ng the deni on of

a limit. The open disk in the x-y plane has
radius . Let (x, y) be any point in this
disk; f(x, y) is within of L.
Basic Limit Proper es of Func ons of Two Variables
(x,y)(x0 ,y0 )
f(x, y) = L
and
lim
(x,y)(x0 ,y0 )
g(x, y) = K.
The following limits hold.

1. Constants:
lim
(x,y)(x0 ,y0 )
2. Iden ty
lim
(x,y)(x0 ,y0 )
3. Sums/Dierences:
4. Scalar Mul ples:
5. Products:
6. Quo ents:
7. Powers:
lim
(x,y)(x0 ,y0 )
b=b
x = x0 ;
(
lim
(x,y)(x0 ,y0 )
y = y0
)
f(x, y) g(x, y) = L K
lim
b f(x, y) = bL
lim
f(x, y) g(x, y) = LK
lim
f(x, y)/g(x, y) = L/K, (K = 0)
lim
f(x, y)n = Ln
(x,y)(x0 ,y0 )
(x,y)(x0 ,y0 )
(x,y)(x0 ,y0 )
(x,y)(x0 ,y0 )
This theorem, combined with Theorems 2 and 3 of Sec on 1.3, allows us to

evaluate many limits.
. Example 397
.Evalua ng a limit
Evaluate the following limits:
1.
Notes:
660
lim
(x,y)(1,)
y
+ cos(xy)
x
2.
lim
(x,y)(0,0)
3xy
x2 + y2

1. The aforemen oned theorems allow us to simply evaluate y/x + cos(xy)
when x = 1 and y = . If an indeterminate form is returned, we must do
more work to evaluate the limit; otherwise, the result is the limit. Therefore
y
lim
+ cos(xy) = + cos
1
(x,y)(1,) x
= 1.
2. We a empt to evaluate the limit by subs tu ng 0 in for x and y, but the
result is the indeterminate form 0/0. To evaluate this limit, we must
do more work, but we have not yet learned what kind of work to do.
.
Therefore we cannot yet evaluate this limit.
When dealing with func ons of a single variable we also considered one
sided limits and stated
lim f(x) = L
xc
if, and only if,
lim f(x) = L
xc+
and
lim f(x) = L.
xc
That is, the limit is L if and only if f(x) approaches L when x approaches c from
either direc on, the le or the right.
In the plane, there are innite direc ons from which (x, y) might approach
(x0 , y0 ). In fact, we do not have to restrict ourselves to approaching (x0 , y0 ) from
a par cular direc on, but rather we can approach that point along a path that is
not a straight line. It is possible to arrive at dierent limi ng values by approaching (x0 , y0 ) along dierent paths. If this happens, we say that
lim
f(x, y)
(x,y)(x0 ,y0 )
does not exist (this is analogous to the le and right hand limits of single variable
func ons not being equal).
Our theorems tell us that we can evaluate most limits quite simply, without
worrying about paths. When indeterminate forms arise, the limit may or may
not exist. If it does exist, it can be dicult to prove this as we need to show the
same limi ng value is obtained regardless of the path chosen. The case where
the limit does not exist is o en easier to deal with, for we can o en pick two
paths along which the limit is dierent.
. Example 398
1. Show
.Showing limits do not exist

lim
3xy
does not exist by nding the limits along the lines
x2 + y2
lim
sin(xy)
does not exist by nding the limit along the path
x+y
(x,y)(0,0)
y = mx.
2. Show
(x,y)(0,0)
y = sin x.
Notes:
661

S
3xy
along the lines y = mx means replace all ys
+ y2
with mx and evalua ng the resul ng limit:
1. Evalua ng
lim
(x,y)(0,0) x2
3x(mx)
3mx2
=
lim
x0 x2 (m2 + 1)
(x,mx)(0,0) x2 + (mx)2
3m
= lim 2
x0 m + 1
3m
= 2
.
m +1
lim
While the limit exists for each choice of m, we get a dierent limit for each
choice of m. That is, along dierent lines we get diering limi ng values,
meaning the limit does not exist.
2. Let f(x, y) =
sin(xy)
x+y .
We are to show that
lim
(x,y)(0,0)
f(x, y) does not exist
by nding the limit along the path y = sin x. First, however, consider
the limits found along the lines y = mx as done above.
(
)
sin x(mx)
sin(mx2 )
lim
= lim
x0 x(m + 1)
x + mx
(x,mx)(0,0)
1
sin(mx2 )
= lim
.
x0
x
m+1
By applying LHpitals Rule, we can show this limit is 0 except when m =
1, that is, along the line y = x. This line is not in the domain of f, so
we have found the following fact: along every line y = mx in the domain
of f,
lim f(x, y) = 0. .
(x,y)(0,0)
Now consider the limit along the path y = sin x:

(
)
(
)
sin x sin x
sin x sin x
lim
= lim
x0
x sin x
x sin x
(x, sin x)(0,0)
Now apply LHpitals Rule twice:
(
)
cos x sin x ( sin x x cos x)
( = 0/0)
x0
1 cos x
(
)
(
)
sin x sin x ( sin x x cos x)2 + cos x sin x (2 cos x + x sin x)
= lim
x0
sin x
= 2/0 the limit does not exist.
= lim
Step back and consider what we have just discovered. Along any line y =
mx in the domain of the f(x, y), the limit is 0. However, along the path
Notes:
662
y = sin x, which lies in the domain of the f(x, y) for all x = 0, the limit
does not exist. Since the limit is not the same along every path to (0, 0),
sin(xy)
we say
lim
does not exist.
(x,y)(0,0) x + y
. Example 399
Finding a limit
5x2 y2
Let f(x, y) = 2
. Find
lim f(x, y).
x + y2
(x,y)(0,0)
It is rela vely easy to show that along any line y = mx, the
S
limit is 0. This is not enough to prove that the limit exists, as demonstrated in
the previous example, but it tells us that if the limit does exist then it must be 0.
To prove the limit is 0,we apply Deni on 80. Let > 0 be given. We want
to nd > 0 such that if (x 0)2 + (y 0)2 < , then |f(x, y) 0| < .

5y2
< 5 for all (x, y) = (0, 0), and that if
Set < /5. Note that 2
x + y2
x2 + y2 < , then x2 < 2 .
Let (x 0)2 + (y 0)2 < . Consider |f(x, y) 0|:

2 2

5x y

0
|f(x, y) 0| = 2

2
x +y

5y2
= x2 2
x + y2
< 2 5
< 5
5
= .
(x 0)2 + (y 0)2 < then |f(x, y) 0| < , which is what we

5x2 y2
wanted to show. Thus
lim
= 0. .
(x,y)(0,0) x2 + y2
Thus if
Con nuity
Deni on 3 denes what it means for a func on of one variable to be connuous. In brief, it meant that the graph of the func on did not have breaks,
holes, jumps, etc. We dene con nuity for func ons of two variables in a similar
way.
Notes:
663

.
Deni on 81
Con nuous
Let a func on f(x, y) be dened on an open disk B containing the point

(x0 , y0 ).
1. f is con nuous at (x0 , y0 ) if
.lim
(x,y)(x0 ,y0 )
f(x, y) = f(x0 , y0 ).
2. f is con nuous on B if f is con nuous at all points in B. If f is con nuous at all points in R2 , we say that f is con nuous everywhere.
. Example 400
Let f(x, y) =
.Con nuity of a func on of two variables

x = 0
. Is f con nuous at (0, 0)? Is f con nuous
cos y x = 0
cos y sin x
x
everywhere?
S
lim
(x,y)(0,0)
To determine if f is con nuous at (0, 0), we need to compare

f(x, y) to f(0, 0).
Applying the deni on of f, we see that f(0, 0) = cos 0 = 1.

We now consider the limit
lim f(x, y). Subs tu ng 0 for x and y in
(x,y)(0,0)
(cos y sin x)/x returns the indeterminate form 0/0, so we need to do more
work to evaluate this limit.
sin x
Consider two related limits:
lim cos y and
lim
. The rst
(x,y)(0,0)
(x,y)(0,0) x
limit does not contain x, and since cos y is con nuous,
lim
(x,y)(0,0)
cos y = lim cos y = cos 0 = 1.

y0
The second limit does not contain y. By Theorem 5 we can say

sin x
sin x
= lim
= 1.
x0 x
(x,y)(0,0) x
lim
Finally, Theorem 101 of this sec on states that we can combine these two limits
as follows:
(
)
cos y sin x
sin x
lim
=
lim (cos y)
x
x
(x,y)(0,0)
(x,y)(0,0)
(
)(
)
sin x
=
lim cos y
lim
(x,y)(0,0)
(x,y)(0,0) x
= (1)(1)
= 1.
Notes:
664

We have found that
cos y sin x
= f(0, 0), so f is con nuous at
x
(x,y)(0,0)
lim
(0, 0).
A similar analysis shows that f is con nuous at all points in R2 . As long as
x = 0, we can evaluate the limit directly; when x = 0, a similar analysis shows
that the limit is cos y. Thus we can say that f is con nuous everywhere. A graph
of f is given in Figure 12.10. No ce how it has no breaks, jumps, etc. .
1
10
x
The following theorem is very similar to Theorem 8, giving us ways to combine con nuous func ons to create other con nuous func ons.
10
10
10 y
.
Theorem 102
Proper es of Con nuous Func ons
Let f and g be con nuous on an open disk B, let c be a real number, and
let n be a posi ve integer. The following func ons are con nuous on B.
Figure 12.10: A graph of f(x, y) in Example

400.
1. Sums/Dierences: f g
2. Constant Mul ples: c f
3. Products:
fg
4. Quo ents:
f/g
5. Powers:
fn
n
f
6. Roots:
7. Composi ons:
(as longs as g = 0 on I)
(if n is even then f 0 on I; if n is odd,

then true for all values of f on I.)
Adjust the deni ons of f and g to: Let f be

con nuous on B, where the range of f on B is
J, and let g be a single variable func on that is
con nuous on J. Then g f, i.e., g(f(x, y)), is
con nuous on B.
. Example 401
Establishing con nuity of a func on
Let f(x, y) = sin(x2 cos y). Show f is con nuous everywhere.
We will apply both Theorems 8 and 102. Let f1 (x, y) = x2 .
Since y is not actually used in the func on, and polynomials are con nuous (by
Theorem 8), we conclude f1 is con nuous everywhere. A similar statement can
be made about f2 (x, y) = cos y. Part 3 of Theorem 102 states that f3 = f1 f2
is con nuous everywhere, and Part 7 of the theorem states the composi on of
sine with f3 is con nuous: that is, sin(f3 ) = sin(x2 cos y) is con nuous everywhere. .
S
Notes:
665

The deni ons and theorems given in this sec on can be extended in a natural way to deni ons and theorems about func ons of three (or more) variables.
We cover the key concepts here; some terms from Deni ons 79 and 81 are not
redened but their analogous meanings should be clear to the reader.
.
Deni on 82
Open Balls, Limit, Con nuous
1. An open ball in R3 centered

at (x0 , y0 , z0 ) with radius r is the set of all
points (x, y, z) such that (x x0 )2 + (y y0 )2 + (z z0 )2 = r.
2. Let f(x, y, z) be a func on of three variables and let (x0 , y0 , z0 ) be a
point in the domain of f. The limit of f(x, y, z) as (x, y, z) approaches
(x0 , y0 , z0 ) is L, denoted
lim
(x,y,z)(x0 ,y0 ,z0 )
f(x, y, z) = L,
if, for every > 0 there is a > 0 such that if (x, y, z) is in the open
ball centered at (x0 , y0 , z0 ) with radius , then |f(x, y, z) L| < .
3. Let f(x, y, z) be dened on an open ball B containing (x0 , y0 , z0 ). f is
con nuous at (x0 , y0 , z0 ) if
lim
f(x, y, z) = f(x0 , y0 , z0 ).
(x,y,z)(x0 ,y0 ,z0 )
These deni ons can also be extended naturally to apply to func ons of four
or more variables. Theorem 102 also applies to func on of three or more variables, allowing us to say that the func on
2
ex +y y2 + z2 + 3
f(x, y, z) =
sin(xyz) + 5
is con nuous everywhere.
Notes:
666
Exercises 12.2
Terms and Concepts
1. Describe in your own words the dierence between boundary and interior point of a set.
2. Use your own words to describe (informally) what
lim f(x, y) = 17 means.
(x,y)(1,2)
14. f(x, y) =
Exercises 15 20, a limit is given. Evaluate the limit along the

paths given, then state why these results show why the given
limit does not exist.
15.
3. Give an example of a closed, bounded set.
lim
(x,y)(0,0)
x 2 y2
x 2 + y2
(a) Along the path y = 0.
4. Give an example of a closed, unbounded set.
(b) Along the path x = 0.
5. Give an example of a open, bounded set.

6. Give an example of a open, unbounded set.
x2 y2
x2 + y2
16.
lim
(x,y)(0,0)
x+y
xy
(a) Along the path y = mx.
Problems
17.
Exercises 7 10, give one boundary point and one interior

point, when possible, of the given set S. State whether S is an
open or a closed set.

{
}
(x 1)2
(y 3)2

7. S = (x, y)
+
1
4
9
{
}
8. S = (x, y) | y = x2
{
}
9. S = (x, y) | x2 + y2 = 1
10. S = {(x, y)|y > sin x}
(x,y)(0,0)
xy y2
y2 + x

(b) Along the path x = 0.
18.
lim
(x,y)(0,0)
sin(x2 )
y

(b) Along the path y = x2 .
19.
Exercises 11 14, give the domain of the given func on and

state whether it is an open or closed set.
x2 + y
y 2x
12. f(x, y) = y x2
lim
lim
(x,y)(1,2)
x+y3
x2 1
(a) Along the path y = 2.

(b) Along the path y = x + 1.
11. f(x, y) =
13. f(x, y) =
1
y x2
20.
lim
(x,y)(,/2)
sin x
cos y
(a) Along the path x = .

(b) Along the path y = x /2.
12.3 Par al Deriva ves

Let y be a func on of x. We have studied in great detail the deriva ve of y with
respect to x, that is, dy
dx , which measures the rate at which y changes with respect
to x. Consider now z = f(x, y). It makes sense to want to know how z changes
with respect to x and/or y. This sec on begins our inves ga on into these rates
of change.
z
20
10
4
x
2
2
2
2
4 y
(a)
The key no on to extract from this example is: by trea ng y as constant (it
does not vary) we can consider how z changes with respect to x. In a similar
fashion, we can hold x constant and consider how z changes with respect to
y. This is the underlying principle of par al deriva ves. We state the formal,
limitbased deni on rst, then show how to compute these par al deriva ves
without directly taking limits.
z
20
10
4
x
2
2
Consider the func on z = f(x, y) = x2 + 2y2 , as graphed in Figure 12.11(a).

By xing y = 2, we focus our a en on to all points on the surface where the
y-value is 2, shown in both parts (a) and (b) of the gure. These points form a
curve in space: z = f(x, 2) = x2 + 8 which is a func on of just one variable. We
can take the deriva ve of z with respect to x along this curve and nd equa ons
of tangent lines, etc.
2
2
4 y
(b)
Deni on 83
Par al Deriva ve
Let z = f(x, y) be a con nuous func on on an open set S in R2 .

1. The par al deriva ve of f with respect to x is:
Figure 12.11: By xing y = 2, the surface

f(x, y) = x2 + 2y2 is a curve in space.
f(x + h, y) f(x, y)
.
h0
h
fx (x, y) = lim
2. The par al deriva ve of f with respect to y is:

Alternate nota ons for fx (x, y) include:
f(x, y + h) f(x, y)
.
h0
h
fy (x, y) = lim
f z
f(x, y),
,
, and zx ,
x
x x
with similar nota ons for fy (x, y). For
ease of nota on, fx (x, y) is o en abbreviated fx .
. Example 402
.Compu ng par al deriva ves with the limit deni on
Let f(x, y) = x2 y + 2x + y3 . Find fx (x, y) using the limit deni on.
Notes:
668

Using Deni on 83, we have:
f(x + h, y) f(x, y)
h
(x + h)2 y + 2(x + h) + y3 (x2 y + 2x + y3 )
= lim
h0
h
(x2 y + 2xhy + h2 y + 2x + 2h + y3 (x2 y + 2x + y3 )
= lim
h0
h
2xhy + h2 y + 2h
= lim
h0
h
= lim 2xy + hy + 2
fx (x, y) = lim
h0
h0
= 2xy + 2.
We have found fx (x, y) = 2xy + 2. .
Example 402 found a par al deriva ve using the formal, limitbased deni on. Using limits is not necessary, though, as we can rely on our previous
knowledge of deriva ves to compute par al deriva ves easily. When computing fx (x, y), we hold y xed it does not vary. Therefore we can compute the
deriva ve with
to x by trea ng y as(a constant
or coecient.
( respect
)
)
d
5x2 = 10x, we compute x
x2 y = 2xy. Here we are trea ng y
Just as dx
as a coecient.
( 3)
( 3)
d
Just as dx
5 = 0, we compute x
y = 0. Here we are trea ng y as a
constant. More examples will help make this clear.
. Example 403
.Finding par al deriva ves
Find fx (x, y) and fy (x, y) in each of the following.
1. f(x, y) = x3 y2 + 5y2 x + 7
2. f(x, y) = cos(xy2 ) + sin x
2 3
3. f(x, y) = ex y x2 + 1
S
1. We have f(x, y) = x3 y2 + 5y2 x + 7.

Begin with fx (x, y). Keep y xed, trea ng it as a constant or coecient, as
appropriate:
fx (x, y) = 3x2 y2 1.
Note how the 5y2 and 7 terms go to zero.
Notes:
669

To compute fy (x, y), we hold x xed:
fy (x, y) = 2x3 y + 10y.
Note how the x and 7 terms go to zero.
2. We have f(x, y) = cos(xy2 ) + sin x.
Begin with fx (x, y). We need to apply the Chain Rule with the cosine term;
y2 is the coecient of the x-term inside the cosine func on.
fx (x, y) = sin(xy2 )(y2 ) + cos x = y2 sin(xy2 ) + cos x.
To nd fy (x, y), note that x is the coecient of the y2 term inside of the
cosine term; also note that since x is xed, sin x is also xed, and we treat
it as a constant.
fy (x, y) = sin(xy2 )(2xy) = 2xy sin(xy2 ).
2 3
3. We have f(x, y) = ex y x2 + 1.
Beginning with fx (x, y), note how we need to apply the Product Rule.
2 3
)1/2
2 3 1(
x2 + 1 + ex y
x2 + 1
2
x2 y3
e
+
.
2 x2 + 1
fx (x, y) = ex y (2xy3 )
= 2xy3 ex
2 3
Note that
when nding fy (x, y) we do not have to apply the Product Rule;
since x2 + 1 does not contain y, we treat it as xed and hence becomes
2 3
a coecient of the ex y term.
2 3
2 3
fy (x, y) = ex y (3x2 y2 ) x2 + 1 = 3x2 y2 ex y x2 + 1.
We have shown how to compute a par al deriva ve, but it may s ll not be
clear what a par al deriva ve means. Given z = f(x, y), fx (x, y) measures the
rate at which z changes as only x varies: y is held constant.
Imagine standing in a rolling meadow, then beginning to walk due east. Depending on your loca on, you might walk up, sharply down, or perhaps not
change eleva on at all. This is similar to measuring zx : you are moving only east
(in the x-direc on) and not north/south at all. Going back to your original loca on, imagine now walking due north (in the y-direc on). Perhaps walking
due north does not change your eleva on at all. This is analogous to zy = 0: z
does not change with respect to y. We can see that zx and zy do not have to be
the same, or even similar, as it is easy to imagine circumstances where walking
east means you walk downhill, though walking north makes you walk uphill.
Notes:
670

The following example helps us visualize this more.
. Example 404
Evalua ng par al deriva ves
Let z = f(x, y) = x2 12 y2 + xy + 10. Find fx (2, 1) and fy (2, 1) and interpret
their meaning.
y + x. Thus
z
10
5
We begin by compu ng fx (x, y) = 2x + y and fy (x, y) =
.
fx (2, 1) = 3
and
fy (2, 1) = 1.
(a)
z
10
It is also useful to note that f(2, 1) = 7.5. What does each of these numbers
mean?
Consider fx (2, 1) = 3, along with Figure 12.12(a). If one stands on the
surface at the point (2, 1, 7.5) and moves parallel to the x-axis (i.e., only the xvalue changes, not the y-value), then the instantaneous rate of change is 3.
Increasing the x-value will decrease the z-value; decreasing the x-value will increase the z-value.
Now consider fy (2, 1) = 1, illustrated in Figure 12.12(b). Moving along the
curve drawn on the surface, i.e., parallel to the y-axis and not changing the xvalues, increases the z-value instantaneously at a rate of 1. Increasing the yvalue by 1 would increase the z-value by approximately 1.
.
x
(b)
Figure 12.12: Illustra ng the meaning of

par al deriva ves.
Since the magnitude of fx is greater than the magnitude of fy at (2, 1), it is

steeper in the x-direc on than in the y-direc on. .
Second Par al Deriva ves

Let z = f(x, y). We have learned to nd the par al deriva ves fx (x, y) and
fy (x, y), which are each func ons of x and y. Therefore we can take par al derivaves of them, each with respect to x and y. We dene these second par als
along with the nota on, give examples, then discuss their meaning.
Notes:
671

.
Deni on 84
Second Par al Deriva ve, Mixed Par al Deriva ve
Let z = f(x, y) be con nuous on an open set S.

1. The second par al deriva ve of f with respect to x then x is
( )
( )
f
2f
= 2 = fx x = fxx
x x
x
2. The second par al deriva ve of f with respect to x then y is

( )
( )
f
2f
= fx y = fxy
=
y x
yx
Note: The terms in Deni on 84 all depend on limits, so each deni on comes
with the caveat where the limit exists.
Similar deni ons hold for
2f
2f
=
f
and
= fyx .
yy
y2
xy
The second par al deriva ves fxy and fyx are mixed par al deriva ves.
The nota on of second par al deriva ves gives some insight into the notaon of the second deriva ve of a func on of a single variable. If y = f(x), then
d2 y
f (x) = 2 . The d2 y por on means take the deriva ve of y twice, while
dx
dx2 means with respect to x both mes. When we only know of func ons of
a single variable, this la er phrase seems silly: there is only one variable to take
the deriva ve with respect to. Now that we understand func ons of mul ple
variables, we see the importance of specifying which variables we are referring
to.
. Example 405
.Second par al deriva ves
For each of the following, nd all six rst and second par al deriva ves. That is,
nd
fx , fy , fxx , fyy , fxy and fyx .
1. f(x, y) = x3 y2 + 2xy3 + cos x
2. f(x, y) =
x3
y2
3. f(x, y) = ex sin(x2 y)
In each, we give fx and fy immediately and then spend me deriving the second par al deriva ves.
Notes:
672

1. f(x, y) = x3 y2 + 2xy3 + cos x
fx (x, y) = 3x2 y2 + 2y3 sin x
fy (x, y) = 2x3 y + 6xy2
)
( )
( 2 2
fxx (x, y) =
fx =
3x y + 2y3 sin x = 6xy2 cos x
x
x
)
( )
( 3
fyy (x, y) =
fy =
2x y + 6xy2 = 2x3 + 12xy
y
y
)
( )
( 2 2
fxy (x, y) =
fx =
3x y + 2y3 sin x = 6x2 y + 6y2
y
y
(
)
)
( 3
fx =
2x y + 6xy2 = 6x2 y + 6y2
fyx (x, y) =
x
x
x3
= x3 y2
y2
3x2
fx (x, y) = 2
y
2x3
fy (x, y) = 3
y
( 3x2 ) 6x
( )
fx =
= 2
fxx (x, y) =
x
x y2
y
3)
(
)
(
2x
6x3
fyy (x, y) =
fy =
3 = 4
y
y
y
y
2)
(
)
(
3x
6x2
fx =
fxy (x, y) =
= 3
2
y
y y
y
3
( )
( 2x )
6x2
fyx (x, y) =
fx =
3 = 3
x
x
y
y
2. f(x, y) =
3. f(x, y) = ex sin(x2 y)
Because the following par al deriva ves get rather long, we omit the extra
nota on and just give the results. In several cases, mul ple applica ons
of the Product and Chain Rules will be necessary, followed by some basic
combina on of like terms.
fx (x, y) = ex sin(x2 y) + 2xyex cos(x2 y)
fy (x, y) = x2 ex cos(x2 y)
fxx (x, y) = ex sin(x2 y) + 4xyex cos(x2 y) + 2yex cos(x2 y) 4x2 y2 ex sin(x2 y)
fyy (x, y) = x4 ex sin(x2 y)
fxy (x, y) = x2 ex cos(x2 y) + 2xex cos(x2 y) 2x3 yex sin(x2 y)
.
fyx (x, y) = x2 ex cos(x2 y) + 2xex cos(x2 y) 2x3 yex sin(x2 y)
Notes:
673

No ce how in each of the three func ons in Example 405, fxy = fyx . Due to
the complexity of the examples, this likely is not a coincidence. The following
theorem states that it is not.
.
Theorem 103
Mixed Par al Deriva ves
Let f be dened such that fxy and fyx are con nuous on an open set S.
Then for each point (x, y) in S, fxy (x, y) = fyx (x, y).
Finding fxy and fyx independently and comparing the results provides a convenient way of checking our work.
Understanding Second Par al Deriva ves

Now that we know how to nd second par als, we inves gate what they tell
us.
Again we refer back to a func on y = f(x) of a single variable. The second
deriva ve of f is the deriva ve of the deriva ve, or the rate of change of the
rate of change. The second deriva ve measures how much the deriva ve is
changing. If f (x) < 0, then the deriva ve is ge ng smaller (so the graph of f is
concave down); if f (x) > 0, then the deriva ve is growing, making the graph
of f concave up.
Now consider z = f(x, y). Similar statements can be made about fxx and fyy
as could be made about f (x) above. When taking deriva ves with respect to
x twice, we measure how much fx changes with respect to x. If fxx (x, y) < 0,
it means that as x increases, fx decreases, and the graph of f will be concave
down in the x-direc on. Using the analogy of standing in the rolling meadow
used earlier in this sec on, fxx measures whether ones path is concave up/down
when walking due east.
Similarly, fyy measures the concavity in the y-direc on. If fyy (x, y) > 0, then
fy is increasing with respect to y and the graph of f will be concave up in the ydirec on. Appealing to the rolling meadow analogy again, fyy measures whether
ones path is concave up/down when walking due north.
We now consider the mixed par als fxy and fyx . The mixed par al fxy measures how much fx changes with respect to y. Once again using the rolling meadow
analogy, fx measures the slope if one walks due east. Looking east, begin walking north (sidestepping). Is the path towards the east ge ng steeper? If so,
fxy > 0. Is the path towards the east not changing in steepness? If so, then
fxy = 0. A similar thing can be said about fyx : consider the steepness of paths
heading north while sidestepping to the east.
The following example examines these ideas with concrete numbers and
Notes:
674

graphs.
. Example 406
Understanding second par al deriva ves
Let z = x2 y2 + xy. Evaluate the 6 rst and second par al deriva ves at
(1/2, 1/2) and interpret what each of these numbers mean.
We nd that:
fx (x, y) = 2x + y, fy (x, y) = 2y + x, fxx (x, y) = 2,
fxy (x, y) = fyx (x, y) = 1. Thus at (1/2, 1/2) we have
S
fx (1/2, 1/2) = 1/2,
fyy (x, y) = 2 and

z
1
fy (1/2, 1/2) = 3/2.
The slope of the tangent line at (1/2, 1/2, 1/4) in the direc on of x is 1/2:
if one moves from that point parallel to the x-axis, the instantaneous rate of
change will be 1/2. The slope of the tangent line at this point in the direc on
of y is 3/2: if one moves from this point parallel to the y-axis, the instantaneous
rate of change will be 3/2. These tangents lines are graphed in Figure 12.13(a)
and (b), respec vely, where the tangent lines are drawn in a solid line.
Now consider only Figure 12.13(a). Three directed tangent lines are drawn
(two are dashed), each in the direc on of x; that is, each has a slope determined
by fx . Note how as y increases, the slope of these lines get closer to 0. Since the
slopes are all nega ve, ge ng closer to 0 means the slopes are increasing. The
slopes given by fx are increasing as y increases, meaning fxy must be posi ve.
Since fxy = fyx , we also expect fy to increase as x increases. Consider Figure
12.13(b) where again three directed tangent lines are drawn, this me each in
the direc on of y with slopes determined by fy . As x increases, the slopes become less steep (closer to 0). Since these are nega ve slopes, this means the
slopes are increasing.
Thus far we have a visual understanding of fx , fy , and fxy = fyx . We now interpret fxx and fyy . In Figure 12.13(a), we see a curve drawn where x is held constant
at x = 1/2: only y varies. This curve is clearly concave down, corresponding
to the fact that fyy < 0. In part (b) of the gure, we see a similar curve where y
is constant and only x varies. This curve is concave up, corresponding to the fact
that fxx > 0. .
1
x
1
1
(a)
z
1
1
x
(b)
Figure 12.13: Understanding the second

par al deriva ves in Example 406.
Par al Deriva ves and Func ons of Three Variables

The concepts underlying par al deriva ves can be easily extend to more
than two variables. We give some deni ons and examples in the case of three
variables and trust the reader can extend these deni ons to more variables if
needed.
Notes:
675

.
Deni on 85
Par al Deriva ves with Three Variables
Let w = f(x, y, z) be a con nuous func on on an open set S in R3 .

The par al deriva ve of f with respect to x is:
f(x + h, y, z) f(x, y, z)
.
fx (x, y, z) = lim
h0
h
Similar deni ons hold for fy (x, y, z) and fz (x, y, z).
By taking par al deriva ves of par al deriva ves, we can nd second par al
deriva ves of f with respect to z then y, for instance, just as before.
. Example 407
Par al deriva ves of func ons of three variables
For each of the following, nd fx , fy , fz , fxz , fyz , and fzz .
1. f(x, y, z) = x2 y3 z4 + x2 y2 + x3 z3 + y4 z4
2. f(x, y, z) = x sin(yz)
S
1. fx = 2xy3 z4 + 2xy2 + 3x2 z3 ; fy = 3x2 y2 z4 + 2x2 y + 4y3 z4 ;

fz = 4x2 y3 z3 + 3x3 z2 + 4y4 z3 ; fxz = 8xy3 z3 + 9x2 z2 ;
fyz = 12x2 y2 z3 + 16y3 z3 ; fzz = 12x2 y3 z2 + 6x3 z + 12y4 z2
2. fx = sin(yz); fy = xz cos(yz); fz = xy cos(yz);
.
fxz = y cos(yz); fyz = x cos(yz) xyz sin(yz); fzz = xy2 sin(xy)
Higher Order Par al Deriva ves

We can con nue taking par al deriva ves of par al deriva ves of par al
deriva ves of ; we do not have to stop with second par al deriva ves. These
higher order par al deriva ves do not have a dy graphical interpreta on; nevertheless they are not hard to compute and worthy of some prac ce.
We do not formally dene each higher order deriva ve, but rather give just
a few examples of the nota on.
( ( ))
f
fxyx (x, y) =
and
x y x
( ( ))
f
fxyz (x, y, z) =
.
z y x
Notes:
676

. Example 408
Higher order par al deriva ves
1. Let f(x, y) = x2 y2 + sin(xy). Find fxxy and fyxx .

2. Let f(x, y, z) = x3 exy + cos(z). Find fxyz .
S
1. To nd fxxy , we rst nd fx , then fxx , then fxxy :
fx = 2xy2 + y cos(xy)
fxx = 2y2 y2 sin(xy)
fxxy = 4y 2y sin(xy) xy2 cos(xy).

To nd fyxx , we rst nd fy , then fyx , then fyxx :
fyx = 4xy + cos(xy) xy sin(xy)

)
= 4y y sin(xy) y sin(xy) + xy2 cos(xy)
fy = 2x2 y + x cos(xy)
fyxx
= 4y 2y sin(xy) xy2 cos(xy).

Note how fxxy = fyxx .
2. To nd fxyz , we nd fx , then fxy , then fxyz :
fx = 3x2 exy + x3 yexy

fxyz = 0.
fxy = 3x3 exy + x3 exy + x4 yexy = 4x3 exy + x4 yexy
In the previous example we saw that fxxy = fyxx ; this is not a coincidence.
While we do not state this as a formal theorem, as long as each par al deriva ve
is con nuous, it does not ma er the order in which the par al deriva ves are
taken. For instance, fxxy = fxyx = fyxx .
This can be useful at mes. Had we known this, the second part of Example 408 would have been much simpler to compute. Instead of compu ng fxyz
in the x, y then z orders, we could have applied the z, then x then y order (as
fxyz = fzxy ). It is easy to see that fz = sin z; then fzx and fzxy are clearly 0 as fz
does not contain an x or y.
Notes:
677

A brief review of this sec on: par al deriva ves measure the instantaneous
rate of change of a mul variable func on with respect to one variable. With
z = f(x, y), the par al deriva ves fx and fy measure the instantaneous rate of
change of z when moving parallel to the x- and y-axes, respec vely. How do we
measure the rate of change at a point when we do not move parallel to one of
these axes? What if we move in the dire on given by the vector 2, 1? Can
we measure that rate of change? The answer is, of course, yes, we can. This is
the topic of the sec on a er next. First, we need to dene what it means for a
func on of two variables to be dieren able.
Notes:
678
Exercises 12.3
Terms and Concepts
1. What is the dierence between a constant and a coecient?

2. Given a func on z = f(x, y), explain in your own words how
to compute fx .
3. In the mixed par al frac on fxy , which is computed rst, fx
or fy ?
4. In the mixed par al frac on
2f
, which is computed rst,
xy
fx or fy ?
16. f(x, y) = (x + y)3

17. f(x, y) = cos(5xy3 )
18. f(x, y) = sin(5x2 + 2y3 )
19. f(x, y) = 4xy2 + 1
20. f(x, y) = (2x + 5y) y

21. f(x, y) =
1
x2 + y2 + 1
22. f(x, y) = 5x 17y

23. f(x, y) = 3x2 + 1
Problems
24. f(x, y) = ln(x2 + y)

ln x
4y
Exercises 5 8, evaluate fx (x, y) and fy (x, y) at the indicated

point.
25. f(x, y) =
5. f(x, y) = x2 y x + 2y + 3 at (1, 2)
26. f(x, y) = 5ex sin y + 9
6. f(x, y) = x3 3x + y2 6y at (1, 3)
7. f(x, y) = sin y cos x at (/3, /3)
8. f(x, y) = ln(xy) at (2, 3)
9. f(x, y) = x2 y + 3x2 + 4y 5
2
10. f(x, y) = y + 3xy + 3x y + x
29. fx = 6xy 4y2 , fy = 3x2 8xy + 2

30. fx =
2x
2y
, fy = 2
x2 + y2
x + y2
Exercises 31 34, nd fx , fy , fz , fyz and fzy .
11. f(x, y) =
x
y
12. f(x, y) =
4
xy
13. f(x, y) = ex
27. fx = sin y + 1, fy = x cos y

28. fx = x + y, fy = x + y
Exercises 9 26, nd fx , fy , fxx , fyy , fxy and fyx .
Exercises 27 30, form a func on z = f(x, y) such that fx and

fy match those given.
31. f(x, y, z) = x2 e2y3z
32. f(x, y, z) = x3 y2 + x3 z + y2 z
+y2
14. f(x, y) = ex+2y
33. f(x, y, z) =
3x
7y2 z
34. f(x, y, z) = ln(xyz)
12.4 Dieren ability and the Total Dieren al

We studied dieren als in Sec on 4.4, where Deni on 18 states that if y = f(x)
and f is dieren able, then, dy = f (x)dx. One important use of this dieren al
is in Integra on by Subs tu on. Another important applica on is approximaon. Let x = dx represent a change in x. When dx is small, dy y, the
change in y resul ng from the change in x. Fundamental in this understanding
is this: as dx gets small, the dierence between y and dy goes to 0. Another
way of sta ng this: as dx goes to 0, the error in approxima ng y with dy goes
to 0.
We extend this idea to func ons of two variables. Let z = f(x, y), and let
x = dx and y = dy represent changes in x and y, respec vely. Let z =
f(x + dx, y + dy) f(x, y) be the change in z over the change in x and y. Recalling
that fx and fy give the instantaneous rates of z-change in the x- and y-direc ons,
respec vely, we can approximate z with dz = fx dx + fy dy; in words, the total
change in z is approximately the change caused by changing x plus the change
caused by changing y. In a moment we give an indica on of whether or not this
approxima on is any good. First we give a name to dz.
.
Deni on 86
Total Dieren al
Let z = f(x, y) be con nuous on an open set S. Let dx and dy represent

changes in x and y, respec vely. Where the par al deriva ves fx and fy
exist, the total dieren al of z is
dz = fx (x, y)dx + fy (x, y)dy.

. Example 409
Finding the total dieren al
Let z = x4 e3y . Find dz.
We compute the par al deriva ves: fx = 4x3 e3y and fy =
3x e . Following Deni on 86, we have
S
4 3y
dz = 4x3 e3y dx + 3x4 e3y dy.
We can approximate z with dz, but as with all approxima ons, there is
error involved. A good approxima on is one in which the error is small. At a
given point (x0 , y0 ), let Ex and Ey be func ons of dx and dy such that Ex dx + Ey dy
describes this error. Then
z = dz + Ex dx + Ey dy
= fx (x0 , y0 )dx + fy (x0 , y0 )dy + Ex dx + Ey dy.
Notes:
680

If the approxima on of z by dz is good, then as dx and dy get small, so does
Ex dx + Ey dy. The approxima on of z by dz is even be er if, as dx and dy go to
0, so do Ex and Ey . This leads us to our deni on of dieren ability.
Deni on 87
Mul variable Dieren ability
Let z = f(x, y) be dened on an open set S containing (x0 , y0 ) where

fx (x0 , y0 ) and fy (x0 , y0 ) exist. Let dz be the total dieren al of z at (x0 , y0 ),
let z = f(x0 + dx, y0 + dy) f(x0 , y0 ), and let Ex and Ey be func ons of
dx and dy such that
z = dz + Ex dx + Ey dy.
1. f is dieren able at (x0 , y0 ) if, given > 0, there is a > 0 such
that if || dx, dy || < , then || Ex , Ey || < . That is, as dx and dy
go to 0, so do Ex and Ey .
2. f is dieren able on S if f is dieren able at every point in S. If f is
dieren able on R2 , we say that f is dieren able everywhere.
. Example 410
.Showing a func on is dieren able
Show f(x, y) = xy + 3y2 is dieren able using Deni on 87.
We begin by nding f(x + dx, y + dy), z, fx and fy .
f(x + dx, y + dy) = (x + dx)(y + dy) + 3(y + dy)2

= xy + xdy + ydx + dxdy + 3y2 + 6ydy + 3dy2 .
z = f(x + dx, y + dy) f(x, y), so
z = xdy + ydx + dxdy + 6ydy + 3dy2 .
It is straigh orward to compute fx = y and fy = x + 6y. Consider once more z:
z = xdy + ydx + dxdy + 6ydy + 3dy2
= ydx + xdy + 6ydy + dxdy + 3dy
(now reorder)
= (y) dx + (x + 6y) dy + (dy) dx + (3dy) dy

|{z}
|{z}
| {z }
| {z }
fx
fy
Ex
Ey
= fx dx + fy dy + Ex dx + Ey dy.
With Ex = dy and Ey = 3dy, it is clear that as dx and dy go to 0, Ex and Ey also go
to 0. Since this did not depend on a specic point (x0 , y0 ), we can say that f(x, y)
Notes:
681

is dieren able for all pairs (x, y) in R2 , or, equivalently, that f is dieren able
everywhere. .
Our intui ve understanding of dieren ability of func ons y = f(x) of one
variable was that the graph of f was smooth. A similar intui ve understanding of func ons z = f(x, y) of two variables is that the surface dened by f is
also smooth, not containing cusps, edges, breaks, etc. The following theorem
states that dieren able func ons are con nuous, followed by another theorem that provides a more tangible way of determining whether a great number
of func on are dieren able or not.
.
Theorem 104
Func ons
Con nuity and Dieren ability of Mul variable
Let z = f(x, y) be dened on an open set S containing (x0 , y0 ). If f is

dieren able at (x0 , y0 ), then f is con nuous at (x0 , y0 ).
Theorem 105
Dieren ability of Mul variable Func ons
Let z = f(x, y) be dened on an open set S containing (x0 , y0 ). If fx and

fy are both con nuous on S, then f is dieren able on S.
The theorems assure us that essen ally all func ons that we see in the course
of our studies here are dieren able (and hence con nuous) on their natural domains. There is a dierence between Deni on 87 and Theorem 105, though: it
is possible for a func on f to be dieren able yet fx and/or fy is not con nuous.
Such strange behavior of func ons is a source of delight for many mathema cians.
When fx and fy exist at a point but are not con nuous at that point, we need
to use other methods to determine whether or not f is dieren able at that
point.
For instance, consider the func on
{
f(x, y) =
Notes:
682
xy
x2 +y2
(x, y) = (0, 0)
(x, y) = (0, 0)

We can nd fx (0, 0) and fy (0, 0) using Deni on 83:
f(0 + h, 0) f(0, 0)
h
0
= lim 2 = 0;
h0 h
f(0, 0 + h) f(0, 0)
fy (0, 0) = lim
h0
h
0
= lim 2 = 0.
h0 h
fx (0, 0) = lim
h0
Both fx and fy exist at (0, 0), but they are not con nuous at (0, 0), as
fx (x, y) =
y(y2 x2 )
(x2 + y2 )2
and
fy (x, y) =
x(x2 y2 )
(x2 + y2 )2
are not con nuous at (0, 0). (Take the limit of fx as (x, y) (0, 0) along the
x- and y-axes; they give dierent results.) So even though fx and fy exist at every point in the x-y plane, they are not con nuous. Therefore it is possible, by
Theorem 105, for f to not be dieren able.
Indeed, it is not. One can show that f is not con nuous at (0, 0) (see Example 398), and by Theorem 104, this means f is not dieren able at (0, 0).
Approxima ng with the Total Dieren al

By the deni on, when f is dieren able dz is a good approxima on for z
when dx and dy are small. We give some simple examples of how this is used
here.
. Example 411
.Approxima ng with the total dieren al
Let z = x sin y. Approximate f(4.1, 0.8).

Recognizing that /4 0.785 0.8, we can
S
approximate
f(4.1,
0.8)
using
f(4,
/4).
We
can
easily
compute
f(4,
/4)
=
4 sin(/4) =
)
(
2 22 = 2 1.414. Without calculus, this is the best approxima on we

could reasonably come up with. The total dieren al gives us a way of adjus ng
this ini al approxima on to hopefully get a more accurate answer.
We let z = f(4.1, 0.8)f(4, /4). The total dieren al dz is approximately
equal to z, so
f(4.1, 0.8) f(4, /4) dz
f(4.1, 0.8) dz + f(4, /4).
(12.1)
To nd dz, we need fx and fy .
Notes:
683
sin y
fx (x, y) =
2 x
fy (x, y) =
x cos y
sin /4
2 4
2/2
=
= 2/8.
4
2
fy (4, /4) = 4
2
= 2.
fx (4, /4) =
Approxima ng 4.1 with 4 gives dx = 0.1; approxima ng 0.8 with /4 gives

dy 0.015. Thus
dz(4, /4) = fx (4, /4)(0.1) + fy (4, /4)(0.015)
2
=
(0.1) + 2(0.015)
8
0.039.
Returning to Equa on (12.1), we have
f(4.1, 0.8) 0.039 + 1.414 = 1.4531.
We, of course, can compute the actual value of f(4.1, 0.8) with a calculator; the
actual value, accurate to 5 places a er the decimal, is 1.45254. Obviously our
approxima on is quite good. .
The point of the previous example was not to develop an approxima on
method for known func ons. A er all, we can very easily compute f(4.1, 0.8)
using readily available technology. Rather, it serves to illustrate how well this
method of approxima on works, and to reinforce the following concept:
New posi on = old posi on + amount of change, so
New posi on old posi on + approximate amount of change.
In the previous example, we could easily compute f(4, /4) and could approximate the amount of z-change when compu ng f(4.1, 0.8), le ng us approximate the new z-value.
It may be surprising to learn that it is not uncommon to know the values of
f, fx and fy at a par cular point without actually knowing f. The total dieren al
gives a good method of approxima ng f at nearby points.
. Example 412
.Approxima ng an unknown func on
Given that f(2, 3) = 6, fx (2, 3) = 1.3 and fy (2, 3) = 0.6, approximate
f(2.1, 3.03).
Notes:
684

The total dieren al approximates how much f changes from
S
the point (2, 3) to the point (2.1, 3.03). With dx = 0.1 and dy = 0.03, we
have
dz = fx (2, 3)dx + fy (2, 3)dy
= 1.3(0.1) + (0.6)(0.03)
= 0.148.
The change in z is approximately 0.148, so we approximate f(2.1, 3.03)
6.148. .
Error/Sensi vity Analysis

The total dieren al gives an approxima on of the change in z given small
changes in x and y. We can use this to approximate error propaga on; that is,
if the input is a li le o from what it should be, how far from correct will the
output be? We demonstrate this in an example.
. Example 413
Sensi vity analysis
A cylindrical steel storage tank is to be built that is 10 tall and 4 across in diameter. It is known that the steel will expand/contract with temperature changes;
is the overall volume of the tank more sensi ve to changes in the diameter or in
the height of the tank?
A cylindrical solid with height h and radius r has volume V =
S
r2 h. We can view V as a func on of two variables, r and h. We can compute
par al deriva ves of V:
V
= Vr (r, h) = 2rh
r
and
V
= Vh (r, h) = r2 .
h
The total dieren al is dV = (2rh)dr + (r2 )dh. When h = 10 and r = 2, we

have dV = 40dr + 4dh. Note that the coecient of dr is 40 125.7; the
coecient of dh is a tenth of that, approximately 12.57. A small change in radius
will be mul plied by 125.7, whereas a small change in height will be mul plied
by 12.57. Thus the volume of the tank is more sensi ve to changes in radius
than in height. .
The previous example showed that the volume of a par cular tank was more
sensi ve to changes in radius than in height. Keep in mind that this analysis only
applies to a tank of those dimensions. A tank with a height of 1 and radius of
5 would be more sensi ve to changes in height than in radius.
Notes:
685

One could make a chart of small changes in radius and height and nd exact
changes in volume given specic changes. While this provides exact numbers, it
does not give as much insight as the error analysis using the total dieren al.
Dieren ability of Func ons of Three Variables

The deni on of dieren ability for func ons of three variables is very similar to that of func ons of two variables. We again start with the total dieren al.
.
Deni on 88
Total Dieren al
Let w = f(x, y, z) be con nuous on an open set S. Let dx, dy and dz represent changes in x, y and z, respec vely. Where the par al deriva ves
fx , fy and fz exist, the total dieren al of w is
dz = fx (x, y, z)dx + fy (x, y, z)dy + fz (x, y, z)dz.
This dieren al can be a good approxima on of the change in w when w =

f(x, y, z) is dieren able.
.
Deni on 89
Mul variable Dieren ability
Let w = f(x, y, z) be dened on an open ball B containing (x0 , y0 , z0 )

where fx (x0 , y0 , z0 ), fy (x0 , y0 , z0 ) and fz (x0 , y0 , z0 ) exist. Let dw be the
total dieren al of w at (x0 , y0 , z0 ), let w = f(x0 + dx, y0 + dy, z0 +
dz) f(x0 , y0 , z0 ), and let Ex , Ey and Ez be func ons of dx, dy and dz such
that
w = dw + Ex dx + Ey dy + Ez dz.
1. f is dieren able at (x0 , y0 , z0 ) if, given > 0, there is a > 0

such that if || dx, dy, dz || < , then || Ex , Ey , Ez || < .
2. f is dieren able on B if f is dieren able at every point in B. If f
is dieren able on R3 , we say that f is dieren able everywhere.
Just as before, this deni on gives a rigorous statement about what it means
to be dieren able that is not very intui ve. We follow it with a theorem similar
to Theorem 105.
Notes:
686
Theorem 106
Variables
Con nuity and Dieren ability of Func ons of Three
Let w = f(x, y, z) be dened on an open ball B containing (x0 , y0 , z0 ).

1. If f is dieren able at (x0 , y0 , z0 ), then f is con nuous at (x0 , y0 , z0 ).
2. If fx , fy and fz are con nuous on B, then f is dieren able on B.
This set of deni on and theorem extends to func ons of any number of
variables. The theorem again gives us a simple way of verifying that most funcons that we enounter are dieren able on their natural domains.
Summary
This sec on has given us a formal deni on of what it means for a func ons
to be dieren able, along with a theorem that gives a more accessible understanding. The following sec ons return to no ons prompted by our study of
par al deriva ves that make use of the fact that most func ons we encounter
are dieren able.
Notes:
687
Exercises 12.4
Terms and Concepts
1. T/F: If f(x, y) is dieren able on S, the f is con nuous on S.
2. T/F: If fx and fy are con nuous on S, then f is dieren able
on S.
3. T/F: If z = f(x, y) is dieren able, then the change in z over
small changes dx and dy in x and y is approximately dz.
15. The length of a long wall is to be approximated. The angle

, as shown in the diagram (not to scale), is measured to
be 85 , and the distance x is measured to be 30. Assume
that the triangle formed is a right triangle.
Is the measurement of the length of more sensi ve to errors in the measurement of x or in ?
4. Finish the sentence: The new z-value is approximately the

old z-value plus the approximate
.
Problems
Exercises 5 8, nd the total dieren al dz.
5. z = x sin y + x2
6. z = (2x2 + 3y)2
7. z = 5x 7y
8. z = xex+y
Exercises 9 12, a func on z = f(x, y) is given. Give the indicated approxima on using the total dieren al.
2
9. f(x, y) =
x + y. Approximate f(2.95, 7.1) knowing
f(3, 7) = 4.
10. f(x, y) = sin x cos y. Approximate f(0.1, 0.1) knowing
f(0, 0) = 0.
11. f(x, y) = x2 y xy2 . Approximate f(2.04, 3.06) knowing
f(2, 3) = 6.
12. f(x, y) = ln(x y). Approximate f(5.1, 3.98) knowing
f(5, 4) = 0.
Exercises 13 16 ask a variety of ques ons dealing with approxima ng error and sensi vity analysis.
13. A cylindrical storage tank is to be 2 tall with a radius of 1 .
Is the volume of the tank more sensi ve to changes in the
radius or the height?
14. Projec le Mo on: The x-value of an object moving under the principles of projec le mo on is x(, v0 , t) =
(v0 cos )t. A par cular projec le is red with an ini al velocity of v0 = 250 /s and an angle of eleva on of = 60 .
It travels a distance of 375 in 3 seconds.
Is the projec le more sensi ve to errors in ini al speed or
angle of eleva on?
=?
16. It is common sense that it is far be er to measure a long

distance with a long measuring tape rather than a short
one. A measured distance D can be viewed as the product of the length of a measuring tape mes the number
n of mes it was used. For instance, using a 3 tape 10
mes gives a length of 30. To measure the same distance
with a 12 tape, we would use the tape 2.5 mes. (I.e.,
30 = 12 2.5.) Thus D = n.
Suppose each me a measurement is taken with the tape,
the recorded distance is within 1/16 of the actual distance.
(I.e., d = 1/16 0.005 ). Using dieren als, show
why common sense proves correct in that it is be er to use
a long tape to measure long distances.
Exercises 17 18, nd the total dieren al dw.
17. w = x2 yz3
18. w = ex sin y ln z
Exercises 19 22, use the informa on provided and the total
dieren al to make the given approxima on.
19. f(3, 1) = 7, fx (3, 1) = 9, fy (3, 1) = 2. Approximate
f(3.05, 0.9).
20. f(4, 2) = 13, fx (4, 2) = 2.6, fy (4, 2) = 5.1. Approximate f(4.12, 2.07).
21. f(2, 4, 5) = 1, fx (2, 4, 5) = 2, fy (2, 4, 5) = 3,
fz (2, 4, 5) = 3.7. Approximate f(2.5, 4.1, 4.8).
22. f(3, 3, 3) = 5, fx (3, 3, 3) = 2, fy (3, 3, 3) = 0, fz (3, 3, 3) =
2. Approximate f(3.1, 3.1, 3.1).
12.5 The Mul variable Chain Rule

The Chain Rule, as learned in Sec on 2.5, states that
If t = g(x), we can express the Chain Rule as
))
)
(
d((
f g(x) = f g(x) g (x).
dx
df
df dt
=
.
dx
dt dx
In this sec on we extend the Chain Rule to func ons of more than one variable.
Theorem 107
Mul variable Chain Rule, Part I
Let z = f(x, y), x = g(t) and y =

( h(t), where
) f, g and h are dieren able
func ons. Then z = f(x, y) = f g(t), h(t) is a func on of t, and
dz
df
dx
dy
=
= fx (x, y) + fy (x, y)
dt
dt
dt
dt
=
f dx
f dy
+
.
x dt
y dt
It is good to understand what the situa on of z = f(x, y), x = g(t) and

y = h(t) describes. We know that z = f(x, y) describes a surface; we also
recognize that x = g(t) and y = h(t) are parametric equa ons for a curve in
the x-y plane. Combining these together, we are describing a curve that lies on
the surface described
( by f. The
) parametric equa ons for this curve are x = g(t),
y = h(t) and z = f g(t), h(t) .
Consider Figure 12.14 in which a surface is drawn, along with a dashed curve
in the x-y plane. Restric ng f to just the points on this circle gives the curve
df
shown on the surface. The deriva ve dt
gives the instantaneous rate of change
of f with respect to t.
We now prac ce applying the Mul variable Chain Rule.
. Example 414
.Using the Mul variable Chain Rule

dz
Let z = x y + x, where x = sin t and y = e5t . Find
using the Chain Rule.
dt
2
Following Theorem 107, we nd

dx
fx (x, y) = 2xy + 1
fy (x, y) = x2
= cos t
dt
Applying the theorem, we have
dz
= (2xy + 1) cos t + 5x2 e5t .
dt
S
.
2
x
2
4
.
Figure 12.14: Understanding the applicaon of the Mul variable Chain Rule.
dy
= 5e5t .
dt
Notes:
689

This may look odd, as it seems that dz
dt is a func on of x, y and t. Since x and y
are func ons of t, dz
is
really
just
a
func
on of t, and we can replace x with sin t
dt
and y with e5t :
dz
2 5t
5t
5t
2
. dt = (2xy + 1) cos t + 5x e = (2 sin(t)e + 1) cos t + 5e sin t.
The previous example can make us wonder: if we subs tuted for x and y at
the end to show that dz
dt is really just a func on of t, why not subs tute before
dieren a ng, showing clearly that z is a func on of t?
That is, z = x2 y + x = (sin t)2 e5t + sin t. Applying the Chain and Product
Rules, we have
dz
= 2 sin t cos t e5t + 5 sin2 t e5t + cos t,
dt
which matches the result from the example.
This may now make one wonder Whats the point? If we could already nd
the deriva ve, why learn another way of nding it? In some cases, applying
this rule makes deriving simpler, but this is hardly the power of the Chain Rule.
Rather, in the case where z = f(x, y), x = g(t) and y = h(t), the Chain Rule is
extremely powerful when we do not know what f, g and/or h are. It may be hard
to believe, but o en in the real world we know rateofchange informa on
(i.e., informa on about deriva ves) without explicitly knowing the underlying
func ons. The Chain Rule allows us to combine several rates of change to nd
another rate of change. The Chain Rule also as theore c use, giving us insight
into the behavior of certain construc ons (as well see in the next sec on).
We apply the Chain Rule once more to solve a max/min problem.
. Example 415
.Applying the Mul variable Chain Rule
Consider the surface z = x2 + y2 xy, on which a par cle moves with x and y
coordinates given by x = cos t and y = sin t. Find dz
dt when t = 0, and nd where
the par cle reaches its maximum/minimum z-values.
1
t=0
1
.x
fx (x, y) = 2x y
1
1
fy (x, y) = 2y x
dx
= sin t
dt
Combining these according to the Chain Rule gives:

dz
= (2x y) sin t + (2y x) cos t.
dt
Figure 12.15: Plo ng the path of a par cle on a surface in Example 415.
Notes:
690
It is straigh orward to compute

dy
= cos t.
dt

dz
When t = 0, x = 1 and y = 0. Thus
= (2)(0) + (1)(1) = 1. When
dt
t = 0, the par cle is moving down, as shown in Figure 12.15.
To nd where z-value is maximized/minimized on the par cles path, we set
dz
dt = 0 and solve for t:
dz
= 0 = (2x y) sin t + (2y x) cos t
dt
0 = (2 cos t sin t) sin t + (2 sin t cos t) cos t
0 = sin2 t cos2 t
cos2 t = sin2 t
t=n
(for odd n)
4
We can use the First Deriva ve Test to nd that on [0, 2], z has reaches its
absolute maximum at t = /4 and 5/4; it reaches its absolute minimum at
t = 3/4 and 7/4, as shown in Figure 12.15. .
We can extend the Chain Rule to include the situa on where z is a func on
of more than one variable, and each of these variables is also a func on of more
than one variable. The basic case of this is where z = f(x, y), and x and y are
func ons of two variables, say s and t.
Theorem 108
Mul variable Chain Rule, Part II
1. Let z = f(x, y), x = g(s, t) and y = h(s, t), where f, g and h are
dieren able func ons. Then z is a func on of s and t, and
z
f x
f y
=
+
,
s
x s y s
z
f x
f y
=
+
.
t
x t
y t
and
2. Let z = f(x1 , x2 , . . . , xm ) be a dieren able func on of m variables,

where each of the xi is a dieren able func on of the variables
t1 , t2 , . . . , tn . Then z is a func on of the ti , and
z
f x1
f x2
f xm
=
+
+ +
.
ti
x1 ti
x2 ti
xm ti
Notes:
691

. Example 416
Using the Mul varible Chain Rule, Part II
z
Let z = x2 y + x, x = s2 + 3t and y = 2s t. Find z
s and t , and evaluate each
when s = 1 and t = 2.
Following Theorem 108, we compute the following par al
deriva ves:
f
= 2xy + 1
x
x
=3
t
x
= 2s
s
f
= x2 ,
y
y
=2
s
y
= 1.
t
Thus
z
= (2xy + 1)(2s) + (x2 )(2) = 4xys + 2s + 2x2 , and
s
z
= (2xy + 1)(3) + (x2 )(1) = 6xy x2 + 3.
t
When s = 1 and t = 2, x = 7 and y = 0, so
.
z
= 100
s
and
z
= 46.
t
. Example 417
Using the Mul varible Chain Rule, Part II
Let w = xy + z2 , where x = t2 es , y = t cos s, and z = s sin t. Find
and t = .
S
deriva ves:
w
t
when s = 0
Following Theorem 108, we compute the following par al

f
=y
x
f
=x
y
y
= cos s
t
x
= 2tes
t
f
= 2z,
z
z
= s cos t.
t
Thus
w
= y(2tes ) + x(cos s) + 2z(s cos t).
t
When s = 0 and t = , we have x = 2 , y = and z = 0. Thus
.
w
= (2) + 2 = 32 .
t
We studied nding dy
dx when y is given as an implicit func on of x in detail
in Sec on 2.6. We nd here that the Mul variable Chain Rule gives a simpler
method of nding dy
dx .
Notes:
692

For instance, consider the implicit func on x2 y xy3 = 3. We learned to use
the following steps to nd dy
dx :
)
d( 2
d( )
x y xy3 =
3
dx
dx
dy
dy
2xy + x2 y3 3xy2
=0
dx
dx
dy
2xy y3
= 2
.
dx
x 3xy2
(12.2)
Instead of using this method, consider z = x2 y xy3 . The implicit func on

above describes the level curve z = 3. Considering x and y as func ons of x, the
Mul variable Chain Rule states that
z dx z dy
dz
=
+
.
dx
x dx y dx
Since z is constant (in our example, z = 3),
Equa on (12.3) becomes
dz
dx
z dy
z
(1) +
x
y dx
dy
z / z
=
dx
x y
0=
(12.3)
= 0. We also know
dx
dx
= 1.
fx
.
fy
Note how our solu on for dy

dx in Equa on (12.2) is just the par al deriva ve
of z, with respect to x, divided by the par al deriva ve of z with respect to y.
We state the above as a theorem.
Theorem 109
Let f be a dieren able func on of x and y, where f(x, y) = c denes y

as an implicit func on of x, for some constant c. Then
dy
fx (x, y)
=
.
dx
fy (x, y)
We prac ce using Theorem 109 by applying it to a problem from Sec on 2.6.
Notes:
693

. Example 418
Given the implicitly dened func on sin(x2 y2 ) + y3 = x + y, nd y . Note: this
is the same problem as given in Example 68 of Sec on 2.6, where the solu on
took about a full page to nd.
Let f(x, y) = sin(x2 y2 ) + y3 x y; the implicitly dened
func on above is equivalent to f(x, y) = 0. We nd dy
dx by applying Theorem 109.
We nd
S
fx (x, y) = 2xy2 cos(x2 y2 ) 1

so
and
fy (x, y) = 2x2 y cos(x2 y2 ) 1,
2xy2 cos(x2 y2 ) 1
dy
= 2
,
dx
2x y cos(x2 y2 ) 1
which matches our solu on from Example 68. .
Notes:
694
Exercises 12.5
Terms and Concepts
12. z = cos x sin y,
1. Let a level curve of z = f(x, y) be described by x = g(t),

y = h(t). Explain why dz
= 0.
dt
2. Fill in the blank: The single variable Chain Rule states
))
(
)
d((
f g(x) = f g(x)
.
dx
3. Fill in the blank: The Mul variable Chain Rule states
f
dy
df
=
+
.
dt
x
dt
4. If z = f(x, y), where x = g(t) and y = h(t), we can subs tute and write z as an explicit func on of t.
T/F: Using the Mul variable Chain Rule to nd dz
is somedt
mes easier than rst subs tu ng and then taking the
deriva ve.
5. T/F: The Mul variable Chain Rule is only useful when all the
related func ons are known explicitly.
In Exercises 7 12, func ons z = f(x, y), x = g(t) and

y = h(t) are given.
dz
(a) Use the Mul variable Chain Rule to compute .
dt
dz
(b) Evaluate
at the indicated t-value.
dt
7. z = 3x + 4y,
x = t2 ,
8. z = x2 y2 ,
x = t,
y = 2t;
t=1
y = t2 1;
t=1
9. z = 5x + 2y,
x = 2 cos t + 1,
y = sin t 3;
t = /4
x
10. z = 2
,
x = cos t,
y = sin t;
t = /2
y +1
11. z = x2 + 2y2 ,
x = sin t,
y = 3 sin t;
t = /4
y = 2t + /2;
t=3
In Exercises 13 18, func ons z = f(x, y), x = g(t) and

y = h(t) are given. Find the values of t where dz
= 0. Note:
dt
these are the same surfaces/curves as found in Exercises 7
12.
13. z = 3x + 4y,
x = t2 ,
14. z = x2 y2 ,
x = t,
15. z = 5x + 2y,
x
,
16. z = 2
y +1
x = 2 cos t + 1,
x = cos t,
17. z = x2 + 2y2 ,
x = sin t,
18. z = cos x sin y,
x = t,
y = 2t
y = t2 1
y = sin t 3
y = sin t
y = 3 sin t
y = 2t + /2
In Exercises 19 22, func ons z = f(x, y), x = g(s, t) and

y = h(s, t) are given.
z
and
(a) Use the Mul variable Chain Rule to compute
s
z
.
t
z
z
(b) Evaluate
and
at the indicated s and t values.
s
t
6. The Mul variable Chain Rule allows us to compute implicit deriva ves by easily by just compu ng two
deriva ves.
Problems
x = t,
19. z = x2 y,
x = s t,
y = 2s + 4t;
s = 1, t = 0
(
)
20. z = cos x + y , x = st2 , y = s2 t; s = 1, t = 1
2
21. z = x2 + y2 ,
22. z = e(x
+y )
x = s cos t, y = s sin t;
,
x = t, y = st2 ;
s = 2, t = /4
s = 1, t = 1
dy
In Exercises 23 26, nd
using Implicit Dieren a on and
dx
Theorem 109.
23. x2 tan y = 50
24. (3x2 + 2y3 )4 = 2
25.
x2 + y
= 17
x + y2
26. ln(x2 + xy + y2 ) = 1
12.6 Direc onal Deriva ves

Par al deriva ves give us an understanding of how a surface changes when we
move in the x and y direc ons. We made the comparison to standing in a rolling
meadow and heading due east: the amount of rise/fall in doing so is comparable
to fx . Likewise, the rise/fall in moving due north is comparable to fy . The steeper
the slope, the greater in magnitude fy .
But what if we didnt move due north or east? What if we needed to move
northeast and wanted to measure the amount of rise/fall? Our par al derivaves alone cannot measure this. This sec on inves gates direc onal derivaves, which are a measure of this.
We begin with a deni on.
.
Deni on 90
Direc onal Deriva ves
Let z = f(x, y) be con nuous on an open set S and let u = u1 , u2 be a

unit vector. For all points (x, y), the direc onal deriva ve of f at (x, y) in
the direc on of u is
Du f(x, y) = lim
h0
f(x + hu1 , y + hu2 ) f(x, y)

.
h
The par al deriva ves fx and fy are dened with similar limits, but only x or
y varies with h, not both. Here both x and y vary with a weighted h, determined
by a par cular unit vector u. This may look a bit in mida ng but in reality it is
not too dicult to deal with; it o en just requires extra algebra. However, the
following theorem reduces this algebraic load.
.
Theorem 110
Direc onal Deriva ves
Let z = f(x, y) be dieren able on an open set S containing (x0 , y0 ), and

let u = u1 , u2 be a unit vector. The direc onal deriva ve of f at (x0 , y0 )
in the direc on of u is
Du f(x0 , y0 ) = fx (x0 , y0 )u1 + fy (x0 , y0 )u2 .

. Example 419
.Compu ng direc onal deriva ves
Let z = 14 x2 y2 and let P = (1, 2). Find the direc onal deriva ve of f, at P,
in the following direc ons:
1. toward the point Q = (3, 4),
2. in the direc on of 2, 1, and
Notes:
696

3. toward the origin.
The surface is plo ed in Figure 12.16, where the point P =

S
(1, 2) is indicated in the x, y-plane as well as the point (1, 2, 9) which lies on the
surface of f. We nd that fx (x, y) = 2x and fx (1, 2) = 2; fy (x, y) = 2y and
fy (1, 2) = 4.
1. Let u1 be the unit vector that points from the point (1, 2) to the point
#
Q = (3, 4), as shown in the
gure.
Thevector PQ = 2, 2; the unit vector

in this direc on is u1 = 1/ 2, 1/ 2 . Thus the direc onal deriva ve of
f at (1, 2) in the direc on of u1 is
Du1 f(1, 2) = 2(1/ 2) + (4)(1/ 2) = 6/ 2 4.24.
10
u3
P
u2
u1
Q
.
Figure 12.16: Understanding the direconal deriva ve in Example 419.
Thus the instantaneous rate of change in moving from the point (1, 2, 9)
on the surface in the direc on of u1 (which points toward the point Q) is
about 4.24. Moving in this direc on moves one steeply downward.
2. We seek the direc onal deriva ve
on
inthe direc
of 2, 1. The unit
vector in this direc on is u2 = 2/ 5, 1/ 5 . Thus the direc onal
deriva ve of f at (1, 2) in the direc on of u2 is
Du2 f(1, 2) = 2(2/ 5) + (4)(1/ 5) = 0.

Star ng on the surface of f at (1, 2) and moving in the direc on of 2, 1
(or u2 ) results in no instantaneous change in z-value. This is analogous to
standing on the side of a hill and choosing a direc on to walk that does not
change the eleva on. One neither walks up nor down, rather just along
the side of the hill.
Finding these direc ons of no eleva on change is important.
3. At P = (1, 2), the direc on towards the origin is given
by the vector
1, 2; the unit vector in this direc on is u3 = 1/ 5, 2/ 5 .

The direc onal deriva ve of f at P in the direc on of the origin is
Du3 f(1, 2) = 2(1/ 5) + (4)(2/ 5) = 10/ 5 4.47.

.
Moving towards the origin means walking uphill quite steeply, with an
ini al slope of about 4.47.
As we study direc onal deriva ves, it will help to make an important connec on between the unit vector u = u1 , u2 that describes the direc on and
the par al deriva ves fx and fy . We start with a deni on and follow this with a
Key Idea.
Notes:
697

.
Deni on 91
Gradient
Let z = f(x, y) be dieren able on an open set S that contains the point
(x0 , y0 ).
1. The gradient of f is f(x, y) = fx (x, y), fy (x, y).

2. The gradient of f at (x0 , y0 ) is f(x0 , y0 ) = fx (x0 , y0 ), fy (x0 , y0 ).
To simplify nota on, we o en express the gradient as f = fx , fy . The

gradient allows us to compute direc onal deriva ves in terms of a dot product.
Note: The symbol is named nabla,
derived from the Greek name of
. a Jewish
harp. Oddly enough, in mathema cs the
expression f is pronounced del f.
Key Idea 55
The Gradient and Direc onal Deriva ves
The direc onal deriva ve of z = f(x,.

y) in the direc on of u is
Du f = f u.
The proper es of the dot product previously studied allow us to inves gate
the proper es of the direc onal deriva ve. Given that the direc onal deriva ve
gives the instantaneous rate of change of z when moving in the direc on of u,
three ques ons naturally arise:
1. In what direc on(s) is the change in z the greatest (i.e., the steepest uphill)?
2. In what direc on(s) is the change in z the least (i.e., the steepest downhill)?
3. In what direc on(s) is there no change in z?
Using the key property of the dot product, we have
f u = || f || || u || cos = || f || cos ,
(12.4)
where is the angle between the gradient andu. (Sinceu is a unit vector, ||u || =
1.) This equa on allows us to answer the three ques ons stated previously.
1. Equa on 12.4 is maximized when cos = 1, i.e., when the gradient and u
have the same direc on; the gradient points in the direc on of greatest z
change.
Notes:
698

2. Equa on 12.4 is minimized when cos = 1, i.e., when the gradient and
u have opposite direc ons; the gradient points in the opposite direc on
of the least z change.
3. Equa on 12.4 is 0 when cos = 0, i.e., when the gradient and u are orthogonal to each other; the gradient is orthogonal to direc ons of no z
change.
This result is rather amazing. Once again imagine standing in a rolling meadow
and face the direc on that leads you steepest uphill. Then the direc on that
leads steepest downhill is directly behind you, and sidestepping either le or
right (i.e., moving perpendicularly to the direc on you face) does not change
your eleva on at all.
Recall that a level curve is dened by a path in the x-y plane along which the
z-values of a func on do not change; the direc onal deriva ve in the direc on of
a level curve is 0. This is analogous to walking along a path in the rolling meadow
along which the eleva on does not change. The gradient at a point is orthogonal
to the direc on where the z does not change; i.e., the gradient is orthogonal to
level curves.
We restate these ideas in a theorem, then use them in an example.
Theorem 111
The Gradient and Direc onal Deriva ves
Let z = f(x, y) be dieren able on an open set S with gradient f and

let u be a unit vector.
1. The maximum value of Du f is || f ||, obtained when the angle
between f and u is 0, i.e., the direc on of maximal increase is
f.
2. The minimum value of Du f is || f ||, obtained when the angle

between f and u is , i.e., the direc on of minimal increase is
f.
3. Du f = 0 when f and u are orthogonal.
. Example 420
.Finding direc ons of maximal and minimal increase
Let f(x, y) = sin x cos y and let P = (/3, /3). Find the direc ons of maximal/minimal increase, and nd a direc on where the instantaneous rate of z
change is 0.
S
We begin by nding the gradient. fx = cos x cos y and fy =
Notes:
699

z
sin x sin y, thus
f = cos x cos y, sin x sin y

x
2
4
(a)
.1
2 y
.
(b)
Figure 12.17: Graphing the surface and
important direc ons in Example 420.
0.5
1.5
2 y
1
x
and, at P, f
( ) 1 3
,
=
,
.
3 3
4 4
Thus the direc on of maximal increase is 1/4, 3/4.

In this direc on, the
instantaneous rate of z change is || 1/4, 3/4 || = 10/4 0.79.
Figure 12.17 shows the surface plo ed from two dierent perspec ves. In
each, the gradient is drawn at P with a dashed line (because of the nature of
this surface, the gradient points into the surface). Let u = u1 , u2 be the
unit vector in the direc on of f at P. Each graph of the gure also contains
the vector u1 , u2 , ||f ||. This vector has a run of 1 (because in the x-y plane
it moves 1 unit) and a rise of ||f ||, hence we can think of it as a vector with
slope of ||f || in the direc on of f, helping us visualize how steep the surface
is in its steepest direc on.
The direc on of minimal increase
is 1/4, 3/4; in this direc on the instan
taneous rate of z change is 10/4 0.79.
Any direc on orthogonal to f is a direc on of no z change. We have two
choices: the direc on of 3, 1 and the direc on of 3, 1. The unit vector
in the direc onof 3, 1 is shown in each graph of the gure as well. The level
curve at z = 3/4 is drawn: recall that along this curve the z-values do not
change. Since 3, 1 is a direc on of no z-change, this vector is tangent to the
level curve at P. .
. Example 421
Understanding when f = 0
2
Let f(x, y) = x + 2x y2 + 2y + 1. Find the direc onal deriva ve of f in any
direc on at P = (1, 1).
We nd f = 2x + 2, 2y + 2. At P, we have f(1, 1) =
S
0, 0. According to Theorem 111, this is the direc on of maximal increase.
However, 0, 0 is direc onless; it has no displacement. And regardless of the
unit vector u chosen, Du f = 0.
Figure 12.18 helps us understand what this means. We can see that P lies at
the top of a paraboloid. In all direc ons, the instantaneous rate of change is 0.
So what is the direc on of maximal increase? It is ne to give an answer of
0 = 0, 0, as this indicates that all direc onal deriva ves are 0. .
Figure 12.18: At the top of a paraboloid,

all direc onal deriva ves are 0.
The fact that the gradient of a surface always points in the direc on of steepest increase/decrease is very useful, as illustrated in the following example.
. Example 422
.The ow of water downhill
Consider the surface given by f(x, y) = 20 x2 2y2 . Water is poured on the
surface at (1, 1/4). What path does it take as it ows downhill?
Notes:
700

Let r(t) = x(t), y(t) be the vectorvalued func on deS
scribing the path of the water in the x-y plane; we seek x(t) and y(t). We know
that water will always ow downhill in the steepest direc on; therefore, at any
point on its path, it will be moving in the direc on of f. (We ignore the physical eects of momentum on the water.) Thus r (t) will be parallel to f, and
there is some constant c such that cf = r (t) = x (t), y (t).
dy
We nd f = 2x, 4y and write x (t) as dx

dt and y (t) as dt . Then
20
10
cf = x (t), y (t)
dx dy
2cx, 4cy =
,
.
dt dt
2
4
This implies
dy
dx
and 4cy = , i.e.,
dt
dt
1 dx
1 dy
c=
and c =
.
2x dt
4y dt
(a)
2cx =
As c equals both expressions, we have .
1 dy
1 dx
=
.
2x dt
4y dt
x
4
To nd an explicit rela onship between x and y, we can integrate both sides with
dx
respect to t. Recall from our study of dieren als that dt = dx. Thus:
dt
1 dx
1 dy
dt =
dt
2x dt
4y dt
1
1
dx =
dy
2x
4y
1
1
ln |x| + C = ln |y|
2
4
2 ln |x| + C = ln |y|
.
(b)
Figure 12.19: A graph of the surface described in Example 422 along with the
path in the x-y plane with the level curves.
Cx2 = y,
where we skip some algebra in the last step. As the water started at the point
(1, 1/4), we can solve for C:
C(1)2 =
1
4
C=
1
.
4
Thus the water follows the curve y = x2 /4 in the x-y plane. The surface and
the path of the water is graphed in Figure 12.19(a). In part (b) of the gure,
Notes:
701

the level curves of the surface are plo ed in the x-y plane, along with the curve
y = x2 /4. No ce how the path intersects the level curves at right angles. As the
path follows the gradient downhill, this reinforces the fact that the gradient is
orthogonal to level curves. .

The concepts of direc onal deriva ves and the gradient are easily extended
to three (and more) variables. We combine the concepts behind Deni ons 90
and 91 and Theorem 110 into one set of deni ons.
.
Deni on 92
ables
Direc onal Deriva ves and Gradient with Three Vari-
Let w = F(x, y, z) be dieren able on an open ball B and let u be a unit

vector in R3 .
1. The gradient of F is F = Fx , Fy , Fz .
2. The direc onal deriva ve of F in the direc on of u is
Du F = F u.
The same proper es of the gradient given in Theorem 111, when f is a funcon of two variables, hold for F, a func on of three variables.
.
Theorem 112
Variables
The Gradient and Direc onal Deriva ves with Three
Let w = F(x, y, z) be dieren able on an open ball B, let F be the gradient of F, and let u be a unit vector.
1. The maximum value of Du F is || F ||, obtained when the angle

between F and u is 0, i.e., the direc on of maximal increase is
F.
2. The minimum value of Du F is || F ||, obtained when the angle
between F and u is , i.e., the direc on of minimal increase is
F.
3. Du F = 0 when F and u are orthogonal.
Notes:
702

We interpret the third statement of the theorem as the gradient is orthogonal to level surfaces, the threevariable analogue to level curves.
. Example 423
Finding direc onal deriva ves with func ons of three variables
If a point source S is radia ng energy, the intensity I at a given point P in space
is inversely propor onal to the square of the distance between S and P. That is,
k
when S = (0, 0, 0), I(x, y, z) = 2
for some constant k.
x + y2 + z2
Let k = 1, let u = 2/3, 2/3, 1/3 be a unit vector, and let P = (2, 5, 3).
Measure distances in inches. Find the direc onal deriva ve of I at P in the direc on of u, and nd the direc on of greatest intensity increase at P.
We need the gradient I, meaning we need Ix , Iy and Iz . Each
S
par al deriva ve requires a simple applica on of the Quo ent Rule, giving
2x
2y
2z
I =
,
,
(x2 + y2 + z2 )2 (x2 + y2 + z2 )2 (x2 + y2 + z2 )2
4 10 6
,
,
0.003, 0.007, 0.004
I(2, 5, 3) =
1444 1444 1444
Du I = I(2, 5, 3) u
17
=
0.0078.
2166
The direc onal deriva ve tells us that moving in the direc on of u from P results in a decrease in intensity of about 0.008 units per inch. (The intensity is
decreasing as u moves one farther from the origin than P.)
The gradient gives the direc on of greatest intensity increase. No ce that
4 10 6
I(2, 5, 3) =
,
,
1444 1444 1444
2
=
2, 5, 3 .
1444
That is, the gradient at (2, 5, 3) is poin ng in the direc on of 2, 5, 3, that
is, towards the origin. That should make intui ve sense: the greatest increase
in intensity is found by moving towards to source of the energy. .
Notes:
703
Exercises 12.6
Terms and Concepts
16. f(x, y) = 4x + 3y, P = (5, 2)
1. What is the dierence between a direc onal deriva ve and

a par al deriva ve?
(b) In the direc on toward the point Q = (2, 7).

17. f(x, y) = x2 + 2y2 xy 7x, P = (4, 1)
2. For what u is Du f = fx ?
(a) In the direc on of v = 2, 5
3. For what u is Du f = fy ?
4. The gradient is
(a) In the direc on of v = 3, 1 .
to level curves.
5. The gradient points in the direc on of

increase.
6. It is generally more informa ve to view the direc onal

deriva ve not as the result of a limit, but rather as the result
product.
of a
18. f(x, y) = x2 y3 2x, P = (1, 1)

Exercises 19 24, a func on z = f(x, y) and a point P are
given.
Problems
(a) Find the direc on of maximal increase of f at P.
Exercises 7 12, a func on z = f(x, y). Find f.
(b) What is the maximal value of Du f at P?
7. f(x, y) = x2 y + xy2 + xy
(c) Find the direc on of minimal increase of f at P.
(d) Give a direc on u such that Du f = 0 at P.
9. f(x, y) =
1
x2 + y2 + 1
10. f(x, y) = 4x + 3y
11. f(x, y) = x2 + 2y2 xy 7x
12. f(x, y) = x2 y3 2x
Exercises 13 18, a func on z = f(x, y) and a point P are
given. Find the direc onal deriva ve of f in the indicated direc ons. Note: these are the same func ons as in Exercises
7 through 12.
13. f(x, y) = x y + xy + xy, P = (2, 1)
2

( )
14. f(x, y) = sin x cos y, P =
,
4 3
(a) In the direc on of v = 1, 1.
15. f(x, y) =
1
, P = (1, 1).
x2 + y2 + 1
(a) In the direc on of v = 1, 1.

Note: these are the same func ons and points as in Exercises
13 through 18.
19. f(x, y) = x2 y + xy2 + xy, P = (2, 1)
( )
20. f(x, y) = sin x cos y, P =
,
4 3
1
21. f(x, y) = 2
, P = (1, 1).
x + y2 + 1
22. f(x, y) = 4x + 3y, P = (5, 4).
23. f(x, y) = x2 + 2y2 xy 7x, P = (4, 1)
24. f(x, y) = x2 y3 2x, P = (1, 1)
Exercises 25 28, a func on w = F(x, y, z), a vector v and a
point P are given.
(a) Find F(x, y, z).
(b) Find Du F at P.
25. F(x, y, z) = 3x2 z3 + 4xy 3z2 , v = 1, 1, 1, P = (3, 2, 1)
26. F(x, y, z) = sin(x) cos(y)ez , v = 2, 2, 1, P = (0, 0, 0)
27. F(x, y, z) = x2 y2 y2 z2 , v = 1, 7, 3, P = (1, 0, 1)
28. F(x, y, z) =
2
, v = 1, 1, 2, P = (1, 1, 1)
x2 + y2 + z2
12.7 Tangent Lines, Normal Lines, and Tangent Planes

Deriva ves and tangent lines go handinhand.( Given y)= f(x), the line tangent
to the graph of f at x = x0 is the line through x0 , f(x0 ) with slope f (x0 ); that
is, the slope of the tangent line is the instantaneous rate of change of f at x0 .
When dealing with func ons of two variables, the graph is no longer a curve
but a surface. At a given point on the surface, it seems there are many lines that
t our intui on of being tangent to the surface.
In Figures 12.20 we see lines that are tangent to curves in space. Since each
curve lies on a surface, it makes sense to say that the lines are also tangent to
the surface. The next deni on formally denes what it means to be tangent
to a surface.
Deni on 93
z
10
.
x
.Figure 12.20:
Showing various lines tangent to a surface.
Direc onal Tangent Line
Let z = f(x, y) be dieren able on an open set S containing (x0 , y0 ) and let
u = u1 , u2 be a unit vector.
(
)
1. The line x through x0 , y0 , f(x0 , y0 ) parallel to 1, 0, fx (x0 , y0 ) is the
tangent line to f in the direc on of x at (x0 , y0 ).
(
)
2. The line y through x0 , y0 , f(x0 , y0 ) parallel to 0, 1, fy (x0 , y0 ) is the
tangent line to f in the direc on of y at (x0 , y0 ).
(
)
3. The line u through x0 , y0 , f(x0 , y0 ) parallel to u1 , u2 , Du f(x0 , y0 )
is the tangent line to f in the direc on of u at (x0 , y0 ).
It is instruc ve to consider each of three direc ons given in the deni on in

terms of slope. The direc on of x is 1, 0, fx (x0 , y0 ); that is, the run is one
unit in the x-direc on and the rise is fx (x0 , y0 ) units in the z-direc on. Note
how the slope is just the par al deriva ve with respect to x. A similar statement
can be made for y . The direc on of u is u1 , u2 , Du f(x0 , y0 ); the run is one
unit in the u direc on (where u is a unit vector) and the rise is the direc onal
deriva ve of z in that direc on.
Deni on 93 leads to the following parametric equa ons of direc onal tangent lines:
x = x0 + t
x = x0
x = x0 + u1 t
y = y0
y = y0 + u2 t
y = y0 + t
x (t) =
, y (t) =
.
and u (t) =
z = z0 + fx (x0 , y0 )t
z = z0 + fy (x0 , y0 )t
z = z0 + Du f(x0 , y0 )t
Notes:
705
. Example 424
Finding direc onal tangent lines
Find the lines tangent to the surface z = sin x cos y at (/2, /2) in the x and y
direc ons and also in the direc on of v = 1, 1 .
S
x
2
1
fx (/2, /2) = 0
fy (/2, /2) = 1.
At (/2, /2), the z-value is 0.

Thus the parametric equa ons of the line tangent to f at (/2, /2) in the
direc ons of x and y are:
x = /2 + t
x = /2
y
=
/2
y = /2 + t .
x (t) =
and y (t) =
z=0
z = t
(a)
fx (x, y) = cos x cos y

fy (x, y) = sin x sin y
2
4
The par al deriva ves with respect to x and y are:
2
4
(b)
Figure 12.21: A surface and direc onal

tangent lines in Example 424.
The two lines are shown with the surface in Figure 12.21(a). To nd the equaon of the tangent line inthe direc
on of v, we rst nd the unit vector in the
direc on of v: u = 1/ 2, 1/ 2 . The direc onal deriva ve at (/2, , 2) in
the direc on of u is
Du f(/2, , 2) = 0, 1 1/ 2, 1/ 2 = 1/ 2.
Thus the direc onal tangent line is
x = /2 t/2
u (t) =
y = /2
+ t/ 2 .
z = t/ 2
The curve through (/2, /2, 0) in the direc on ofv is shown in Figure 12.21(b)
along with u (t). .
. Example 425
.Finding direc onal tangent lines
Let f(x, y) = 4xy x4 y4 . Find the equa ons of all direc onal tangent lines to
f at (1, 1).
First note that f(1, 1) = 2. We need to compute direc onal
S
deriva ves, so we need f. We begin by compu ng par al deriva ves.
fx = 4y 4x3 fx (1, 1) = 0;
fy = 4x 4y3 fy (1, 1) = 0.
Thus f(1, 1) = 0, 0. Let u = u1 , u2 be any unit vector. The direc onal

deriva ve of f at (1, 1) will be Du f(1, 1) = 0, 0u1 , u2 = 0. It does not ma er
Notes:
706

what direc on we choose; the direc onal deriva ve is always 0. Therefore
x = 1 + u1 t
y = 1 + u2 t .
u (t) =
z=2
Figure 12.22 shows a graph of f and the point (1, 1, 2). Note that this point
comes at the top of a hill, and therefore every tangent line through this point
will have a slope of 0.
That is, consider any curve on the surface that goes through this point. Each
curve will have a rela ve maximum at this point, hence its tangent line will have
a slope of 0. The following sec on inves gates the points on surfaces where all
tangent lines have a slope of 0. .
z
2
1
x
1
2
2 y
.
Figure 12.22: Graphing f in Example 425.
Normal Lines
When dealing with a func on y = f(x) of one variable, we stated that a line
through (c, f(c)) was tangent to f if the line had a slope of f (c) and was normal
(or, perpendicular, orthogonal) to f if it had a slope of 1/f (c). We extend the
concept of normal, or orthogonal, to func ons of two variables.
Let z = f(x, y) be a dieren able func on of two variables. By Deni on 93,

at (x0 , y0 ), x (t) is a line parallel to the vector dx = 1, 0, fx (x0 , y0 ) and y (t) is
(a line parallel to) dy = 0, 1, fy (x0 , y0 ). Since lines in these direc ons through
x0 , y0 , f(x0 , y0 ) are tangent to the surface, a line through this point and orthogonal to these direc ons would be orthogonal, or normal, to the surface. We can
use this direc on to create a normal line.
The direc on of the normal line is orthogonal to dx and dy , hence the direcon is parallel to dn = dx dy . It turns out this cross product has a very simple
form:
dx dy = 1, 0, fx 0, 1, fy = fx , fy , 1 .
It is o en more convenient to refer to the opposite of this direc on, namely

fx , fy , 1. This leads to a deni on.
Notes:
707

.
Deni on 94
Normal Line
Let z = f(x, y) be dieren able on an open set S containing (x0 , y0 )

where
a = fx (x0 , y0 ) and b = fy (x0 , y0 )
are dened.
1. A nonzero
vector parallel
to n = a, b, 1 is orthogonal to f at
(
)
P = x0 , y0 , f(x0 , y0 ) .
2. The line n through P with direc on parallel to n is the normal line
to f at P.
(
)
Thus the parametric equa ons of the normal line to a surface f at x0 , y0 , f(x0 , y0 )
is:
x = x0 + at
y = y0 + bt
n (t) =
.
z = f(x0 , y0 ) t
. Example 426
Finding a normal line
Find the equa on of the normal line to z = x2 y2 + 2 at (0, 1).
.
.
Figure 12.23: Graphing a surface with a

normal line from Example 426.
We nd zx (x, y) = 2x and zy (x, y) = 2y; at (0, 1), we

S
have zx = 0 and zy = 2. We take the direc on of the normal line, following
Deni on 94, to be n = 0, 2, 1. The line with this direc on going through
the point (0, 1, 1) is
x=0
y = 2t + 1
n (t) =
or n (t) = 0, 2, 1 t + 0, 1, 1 .
z = t + 1
The surface z = x2 + y2 , along with the found normal line, is graphed in
Figure 12.23. .
The direc on of the normal line has many uses, one of which is the denion of the tangent plane which we dene shortly. Another use is in measuring
distances from the surface to a point. Given a point Q in space, it is general geometric concept to dene the distance from Q to the surface as being the length
of the shortest line segment PQ over all points P on the surface. This, in turn,
#
implies that PQ will be orthogonal to the surface at P. Therefore we can measure the distance from Q to the surface f by nding a point P on the surface such
Notes:
708

#
that PQ is parallel to the normal line to f at P.
. Example 427
Finding the distance from a point to a surface
Let f(x, y) = 2 x2 y2 and let Q = (2, 2, 2). Find the distance from Q to the
surface dened by f.
This surface used in Example 425, so we know that at (x, y),
S
the direc on of the normal line will be dn = 2x, 2y,1. A point P on the
#
surface will have coordinates (x, y, 2 x2 y2 ), so PQ = 2 x, 2 y, x2 + y2 .
#
#
To nd where PQ is parallel to dn , we need to nd x, y and c such that cPQ = dn .
#
cPQ = dn
c 2 x, 2 y, x2 + y2 = 2x, 2y, 1 .
This implies
c(2 x) = 2x
c(2 y) = 2y
c(x2 + y2 ) = 1
In each equa on, we can solve for c:
c=
2x
2y
1
=
= 2
.
2x
2y
x + y2
The rst two frac ons imply x = y, and so the last frac on can be rewri en as
c = 1/(2x2 ). Then
2x
1
= 2
2x
2x
2x(2x2 ) = 1(2 x)
4x3 = 2 x
4x3 + x 2 = 0.
This last equa on is a cubic, which is not dicult to solve with a numeric solver.
We nd that x = 0.689, hence P = (0.689, 0.689, 1.051). We nd the distance
from Q to the surface of f is
#
2
2
2
. || PQ || = (2 0.689) + (2 0.689) + (2 1.051) = 2.083.
We can take the concept of measuring the distance from a point to a surface
to nd a point Q a par cular distance from a surface at a given point P on the
Notes:
709

surface.
. Example 428
Finding a point
from a surface
( a set distance
)
Let f(x, y) = x y2 + 3. Let P = 2, 1, f(2, 1) = (2, 1, 4). Find points Q in space
#
that is 4 units from the surface of f at P. That is, nd Q such that || PQ || = 4 and
#
PQ is orthogonal to f at P.
We begin by nding par al deriva ves:
u =
2
2
4
fx (2, 1) = 1
fy (x, y) = 2y
fy (2, 1) = 2

n
= 1/ 6, 2/ 6, 1/ 6 0.408, 0.816, 0.408 .
|| n ||
Thus a the normal line to f at P can be wri en as

n (t) = 2, 1, 4 + t 0.408, 0.816, 0.408 .
The vector n = 1, 2, 1 is orthogonal to f at P. For reasons that will become

more clear in a moment, we nd the unit vector in the direc on of n:
fx (x, y) = 1
Figure 12.24: Graphing the surface in Example 428 along with points 4 units from
the surface.
An advantage of this parametriza on of the line is that le ng t = t0 gives a

point on the line that is |t0 | units from P. (This is because the direc on of the
line is given in terms of a unit vector.) There are thus two points in space 4 units
from P:
Q1 = n (4)
3.63, 2.27, 2.37
Q2 = n (4)
0.37, 4.27, 5.63
The surface is graphed along with points P, Q1 , Q2 and a por on of the normal
line to f at P. .
Tangent Planes
We can the direc on
( of the normal line
) to dene a plane. With a = fx (x0 , y0 ),
b = fy (x0 , y0 ) and P = x0 , y0 , f(x0 , y0 ) , the vector n = a, b, 1 is orthogonal
to f at P. The plane through P with normal vector n is therefore tangent to f at
P.
Notes:
710
Deni on 95
Tangent Plane
Let z = f(x, y) be dieren able on an open set S containing

(x0 , y(0 ), where a =) fx (x0 , y0 ), b = fy (x0 , y0 ), n = a, b, 1 and
P = x0 , y0 , f(x0 , y0 ) .
The plane through P with normal vector n is the tangent plane to f at P.

The standard form of this plane is
(
)
a(x x0 ) + b(y y0 ) z f(x0 , y0 ) = 0.
. Example 429
Finding tangent planes
Find the equa on tangent plane to z = x2 y2 + 2 at (0, 1).
Figure 12.25: Graphing a surface with tangent plane from Example 429.
Note that this is the same surface and point used in Example
S
426. There we found n = 0, 2, 1 and P = (0, 1, 1). Therefore the equa on
of the tangent plane is
2(y 1) (z 1) = 0.
The surface z = x2 + y2 and tangent plane are graphed in Figure 12.25. .
. Example 430
.Using the tangent plane to approximate func on values
The point (3, 1, 4) lies on the surface of an unknown dieren able func on f
where fx (3, 1) = 2 and fy (3, 1) = 1/2. Find the equa on of the tangent
plane to f at P, and use this to approximate the value of f(2.9, 0.8).
Knowing the par al deriva ves at (3, 1) allows us to form
S
the normal vector to the tangent plane, n = 2, 1/2, 1. Thus the equa on
of the tangent line to f at P is:
2(x3)1/2(y+1)(z4) = 0
z = 2(x3)1/2(y+1)+4. (12.5)
Just as tangent lines provide excellent approxima ons of curves near their point
of intersec on, tangent planes provide excellent approxima ons of surfaces near
their point of intersec on. So f(2.9, 0.8) z(2.9, 0.8) = 3.7.
This is not a new method of approxima on. Compare the right hand expression for z in Equa on (12.5) to the total dieren al:
dz = fx dx + fy dy
and
z = |{z}
2 (x 3) + 1/2 (y + 1) +4.
| {z } | {z } | {z }
|
fx
dx
{z
fy
dy
dz
Notes:
711

Thus the new z-value is the sum of the change in z (i.e., dz) and the old zvalue (4). As men oned when studying the total dieren al, it is not uncommon
to know par al deriva ve informa on about a unknown func on, and tangent
planes are used to give accurate approxima ons of the func on. .
The Gradient and Normal Lines, Tangent Planes

The methods developed in this sec on so far give a straigh orward method
of nding equa ons of normal lines and tangent planes for surfaces with explicit
equa ons of the form z = f(x, y). However, they do not handle implicit equaons well, such as x2 + y2 + z2 = 1. There is a technique that allows us to nd
vectors orthogonal to these surfaces based on the gradient.
.
Deni on 96
Gradient
Let w = F(x, y, z) be dieren able on an open ball B that contains the

point (x0 , y0 , z0 ).
1. The gradient of F is F(x, y, z) = fx (x, y, z), fy (x, y, z), fz (x, y, z).

2. The gradient of F at (x0 , y0 , z0 ) is
F(x0 , y0 , z0 ) = fx (x0 , y0 , z0 ), fy (x0 , y0 , z0 ), fz (x0 , y0 , z0 ) .
Recall that when z = f(x, y), the gradient f = fx , fy is orthogonal to level

curves of f. An analogous statement can be made about the gradient F, where
w = F(x, y, z). Given a point (x0 , y0 , z0 ), let c = F(x0 , y0 , z0 ). Then F(x, y, z) =
c is a level surface that contains the point (x0 , y0 , z0 ). The following theorem
states that F(x0 , y0 , z0 ) is orthogonal to this level surface.
.
Theorem 113
The Gradient and Level Surfaces
Let w = F(x, y, z) be dieren able on an open ball B containing

(x0 , y0 , z0 ) with gradient F, where F(x0 , y0 , z0 ) = c.
The vector F(x0 , y0 , z0 ) is orthogonal to the level surface F(x, y, z) = c

at (x0 , y0 , z0 ).
The gradient at a point gives a vector orthogonal to the surface at that point.
This direc on can be used to nd tangent planes and normal lines.
Notes:
712
. Example 431
Using the gradient to nd a tangent plane

x2
y2
z2
Find the equa on of the plane tangent to the ellipsoid
+
+
= 1 at
12
6
4
P = (1, 2, 1).
z
2
We consider the equa on of the ellipsoid as a level surface

2
2
x2
of a func on F of three variables, where F(x, y, z) = 12
+ y6 + z4 . The gradient
is:
S
F(x, y, z) = Fx , Fy , Fz
x y z
=
, ,
.
6 3 2
At P, the gradient is F(1, 2, 1) = 1/6, 2/3, 1/2. Thus the equa on of the
plane tangent to the ellipsoid at P is
2
x
.
2
2
2
Figure 12.26: An ellipsoid and its tangent

plane at a point.
2
1
1
(x 1) + (y 2) + (z 1) = 0.
6
3
2
The ellipsoid and tangent plane are graphed in Figure 12.26. .
Tangent lines and planes to surfaces have many uses, including the study of
instantaneous rates of changes and making approxima ons. Normal lines also
have many uses. In this sec on we focused on using them to measure distances
from a surface. Another interes ng applica on is in computer graphics, where
the eects of light on a surface are determined using normal vectors.
The next sec on inves gates another use of par al deriva ves: determining
rela ve extrema. When dealing with func ons of the form y = f(x), we found
rela ve extrema by nding x where f (x) = 0. We can start nding rela ve
extrema of z = f(x, y) by se ng fx and fy to 0, but it turns out that there is more
to consider.
Notes:
713
Exercises 12.7
Terms and Concepts
1. Explain how the vector v = 1, 0, 3 can be thought of as
having a slope of 3.
2. Explain how the vector v = 0.6, 0.8, 2 can be thought
of as having a slope of 2.
3. T/F: Let z = f(x, y) be dieren able at P. If n is a normal
vector to the tangent plane of f at P, then n is orthogonal
to fx and fy at P.
4. Explain in your own words why we do not refer to the tangent line to a surface at a point, but rather to direc onal
tangent lines to a surface at a point.
Problems
12. f(x, y) = x2 2x y2 + 4y, v = 1, 1, P = (1, 2).

Exercises 13 16, a func on z = f(x, y), a vectorv and a point
P are given. Find the two points that are 2 units from the surface f at P. Note: these are the same func ons as in Exercises
5 8.
13. f(x, y) = 2x2 y 4xy2 , v = 1, 3, P = (2, 3).
14. f(x, y) = 3 cos x sin y, v = 1, 2, P = (/3, /6).
15. f(x, y) = 3x 5y, v = 1, 1, P = (4, 2).
16. f(x, y) = x2 2x y2 + 4y, v = 1, 1, P = (1, 2).
Exercises 17 20, a func on z = f(x, y), a point P is given.
Find the equa on of the tangent plane to f at P. Note: these
are the same func ons as in Exercises 5 8.
17. f(x, y) = 2x2 y 4xy2 , v = 1, 3, P = (2, 3).
Exercises 5 8, a func on z = f(x, y), a vector v and a point

P are given. Give the parametric equa ons of the following
direc onal tangent lines to f at P:
(a) x (t)
18. f(x, y) = 3 cos x sin y, v = 1, 2, P = (/3, /6).

19. f(x, y) = 3x 5y, v = 1, 1, P = (4, 2).
20. f(x, y) = x2 2x y2 + 4y, v = 1, 1, P = (1, 2).
(b) y (t)
(c) u (t), where u is the unit vector in the direc on of v.
5. f(x, y) = 2x2 y 4xy2 , v = 1, 3, P = (2, 3).
Exercises 21 24, an implicitly dened func on of x, y and z

is given along with a point P that lies on the surface. Use the
gradient F to:
(a) nd the equa on of the normal line to the surface at
P, and
6. f(x, y) = 3 cos x sin y, v = 1, 2, P = (/3, /6).

7. f(x, y) = 3x 5y, v = 1, 1, P = (4, 2).
(b) nd the equa on of the plane tangent to the surface

at P.
8. f(x, y) = x2 2x y2 + 4y, v = 1, 1, P = (1, 2).

Exercises 9 12, a func on z = f(x, y), a vector v and a point
P are given. Find the equa on of the normal line to f at P.
Note: these are the same func ons as in Exercises 5 8.
9. f(x, y) = 2x y 4xy , v = 1, 3, P = (2, 3).
2
21.

x2
y2
z2
+
+
= 1, at P = (1, 2, 6)
8
4
16
22. z2
x2
y2
= 0, at P = (4, 3, 5)
4
9
10. f(x, y) = 3 cos x sin y, v = 1, 2, P = (/3, /6).
23. xy2 xz2 = 0, at P = (2, 1, 1)
11. f(x, y) = 3x 5y, v = 1, 1, P = (4, 2).
24. sin(xy) + cos(yz) = 0, at P = (2, /12, 4)
12.8 Extreme Values
12.8 Extreme Values

Given a func on z = f(x, y), we are o en interested in points where z takes on
the largest or smallest values. For instance, if z represents a cost func on, we
would likely want to know what (x, y) values minimize the cost. If z represents
the ra o of a volume to surface area, we would likely want to know where z is
greatest. This leads to the following deni on.
Deni on 97
Rela ve and Absolute Extrema
Let z = f(x, y) be dened on a set S containing the point P = (x0 , y0 ).

1. If there is an open disk D containing P such that f(x0 , y0 ) f(x, y)
for all (x, y) in D, then f has a rela ve maximum at P; if f(x0 , y0 )
f(x, y) for all (x, y) in D, then f has a rela ve minimum at P.
2. If f(x0 , y0 ) f(x, y) for all (x, y) in S, then f has an absolute maximum at P; if f(x0 , y0 ) f(x, y) for all (x, y) in S, then f has an
absolute minimum at P.
3. If f has a rela ve maximum or minimum at P, then f has a rela ve
extrema at P; if f has an absolute maximum or minimum at P, then
f has a absolute extrema at P.
If f has a rela ve or absolute maximum at P = (x0 , y0 ), it means every curve

on the surface of f through P will also have a rela ve or absolute maximum at P.
Recalling what we learned in Sec on 3.1, the slopes of the tangent lines to these
curves at P must be 0 or undened. Since direc onal deriva ves are computed
using fx and fy , we are led to the following deni on and theorem.
Deni on 98
Cri cal Point
Let z = f(x, y) be con nuous on an open set S. A cri cal point P =

(x0 , y0 ) of f is a point in S such that
fx (x0 , y0 ) = 0 and fy (x0 , y0 ) = 0, or

fx (x0 , y0 ) and/or fy (x0 , y0 ) is undened.
Notes:
715

.
Theorem 114
Cri cal Points and Rela ve Extrema
Let z = f(x, y) be dened on an open set S containing P = (x0 , y0 ). If f

has a rela ve extrema at P, then P is a cri cal point of f.
Therefore, to nd rela ve extrema, we nd the cri cal points of f and determine which correspond to rela ve maxima, rela ve minima, or neither. The
following examples demonstrate this process.
. Example 432
Finding cri cal points and rela ve extrema
Let f(x, y) = x2 + y2 xy x 2. Find the rela ve extrema of f.
z
5
fx (x, y) = 2x y 1
.
x
We start by compu ng the par al deriva ves of f:
Figure 12.27: The surface in Example 432

with its absolute minimum indicated.
2
x
and
x + 2y = 0.
This solu on to this system is x = 2/3, y = 1/3. (Check that at (2/3, 1/3), both
fx and fy are 0.)
The graph in Figure 12.27 shows f along with this cri cal point. It is clear from
the graph that this is a rela ve minimum; further considera on of the func on
shows that this is actually the absolute minimum. .
. Example 433
Let f(x, y) = x2 + y2 + 2. Find the rela ve extrema of f.
fy (x, y) = 2y x.
Each is never undened. A cri cal point occurs when fx and fy are simultaneously
0, leading us to solve the following system of linear equa ons:
2x y 1 = 0
and
x
fx (x, y) =
x2 + y2
2
2
We start by compu ng the par al deriva ves of f:
.
with its absolute maximum indicated.
and
fy (x, y) =
y
x2 + y2
It is clear that fx = 0 when x = 0 and that fy = 0 when y = 0. At (0, 0), both

fx and fy are not 0, but rather undened. The point (0, 0) is s ll a cri cal point,
though, because the par al deriva ves are undened.
The surface of f is graphed in Figure 12.28 along with the point (0, 0, 2). The
graph shows that this point is the absolute maximum of f. .
In each of the previous two examples, we found a cri cal point of f and then
determined whether or not it was a rela ve (or absolute) maximum or minimum
Notes:
716
12.8 Extreme Values

by graphing. It would be nice to be able to determine whether a cri cal point
corresponded to a max or a min without a graph. Before we develop such a test,
we do one more example that sheds more light on the issues our test needs to
consider.
. Example 434
Let f(x, y) = x3 3x y2 + 4y. Find the rela ve extrema of f.
z
Once again we start by nding the par al deriva ves of f:
fx (x, y) = 3x2 3
fy (x, y) = 2y + 4.
and
Each is always dened. Se ng each equal to 0 and solving for x and y, we nd

fx (x, y) = 0
fy (x, y) = 0
x = 1
y = 2.
We have two cri cal points: (1, 2) and (1, 2). To determine if they correspond
to a rela ve maximum or minimum, we consider the graph of f in Figure 12.29.
The cri cal point (1, 2) clearly corresponds to a rela ve maximum. However, the cri cal point at (1, 2) is neither a maximum nor a minimum, displaying
a dierent, interes ng characteris c.
If one walks parallel to the y-axis towards this cri cal point, then this point
becomes a rela ve maximum along this path. But if one walks towards this point
parallel to the x-axis, this point becomes a rela ve minimum along this path. A
point that seems to act as both a max and a min is a saddle point. A formal
deni on follows. .
Deni on 99
2
y
.
with both cri cal points marked.
Saddle Point
Let P = (x0 , y0 ) be in the domain of f where fx = 0 and fy = 0 at

P. P is a saddle point of f if, for every open disk D containing P, there
are points (x1 , y1 ) and (x2 , y2 ) in D such that f(x0 , y0 ) > f(x1 , y1 ) and
f(x0 , y0 ) < f(x2 , y2 ).
At a saddle point, the instantaneous rate of change in all direc ons is 0 and
there are points nearby with z-values both less than and greater than the z-value
of the saddle point.
Before Example 434 we men oned the need for a test to dieren ate between rela ve maxima and minima. We now recognize that our test also needs
Notes:
717

to account for saddle points. To do so, we consider the second par al deriva ves
of f.
Recall that with single variable func ons, such as y = f(x), if f (c) > 0, then
f is concave up at c, and if f (c) = 0, then f has a rela ve minimum at x = c. (We
called this the Second Deriva ve Test.) Note that at a saddle point, it seems the
graph is both concave up and concave down, depending on which direc on
you are considering.
It would be nice if the following were true:
fxx and fyy > 0
fxx and fyy < 0
fxx and fyy have opposite signs
rela ve minimum
rela ve maximum
saddle point.
However, this is not the case. Func ons f exist where fxx and fyy are both
posi ve but a saddle point s ll exists. In such a case, while the concavity in the
x-direc on is up (i.e., fxx > 0) and the concavity in the y-direc on is also up (i.e.,
fyy > 0), the concavity switches somewhere in between the x- and y-direc ons.
To account for this, consider D = fxx fyy fxy fyx . Since fxy and fyx are equal
when con nuous (refer back to Theorem 103), we can rewrite this as D = fxx fyy
f2xy . D can be used to test whether the concavity at a point changes depending on
direc on. If D > 0, the concavity does not switch (i.e., at that point, the graph
is concave up or down in all direc ons). If D < 0, the concavity does switch. If
D = 0, our test fails to determine whether concavity switches or not. We state
the use of D in the following theorem.
.
Theorem 115
Second Deriva ve Test
Let z = f(x, y) be dieren able on an open set containing P = (x0 , y0 ),

and let
D = fxx (x0 , y0 )fyy (x0 , y0 ) f2xy (x0 , y0 ).
1. If D > 0 and fxx (x0 , y0 ) > 0, then P is a rela ve minimum of f.

2. If D > 0 and fxx (x0 , y0 ) < 0, then P is a rela ve maximum of f.
3. If D < 0, then P is a saddle point of f.
4. If D = 0, the test is inconclusive.
We rst prac ce using this test with the func on in the previous example,
where we visually determined we had a rela ve maximum and a saddle point.
Notes:
718
12.8 Extreme Values

. Example 435
Using the Second Deriva ve Test
Let f(x, y) = x3 3xy2 +4y as in Example 434. Determine whether the func on
has a rela ve minimum, maximum, or saddle point at each cri cal point.
We determined previously that the cri cal points of f are
S
(1, 2) and (1, 2). To use the Second Deriva ve Test, we must nd the second
par al deriva ves of f:
fxx = 6x;
fyy = 2;
fxy = 0.
Thus D(x, y) = 12x.

At (1, 2): D(1, 2) = 12 > 0, and fxx (1, 2) = 6. By the Second Derivave Test, f has a rela ve maximum at (1, 2).
At (1, 2): D(1, 2) = 12 < 0. The Second Deriva ve Test states that f has a
saddle point at (1, 2).
The Second Deriva ve Test conrmed what we determined visually. .
. Example 436
.Using the Second Deriva ve Test
Find the rela ve extrema of f(x, y) = x2 y + y2 + xy.
We start by nding the rst and second par al deriva ves of
f:
fx = 2xy + y
fxx = 2y
fxy = 2x + 1
fy = x2 + 2y + x
fyy = 2
fyx = 2x + 1.
We nd the cri cal points by nding where fx and fy are simultaneously 0 (they
are both never undened). Se ng fx = 0, we have:
fx = 0
2xy + y = 0
y(2x + 1) = 0.
This implies that for fx = 0, either y = 0 or 2x + 1 = 0.

Assume y = 0 then consider fy = 0:
fy = 0
2
x + 2y + x = 0,
and since y = 0, we have
x +x=0
x(x + 1) = 0.
Thus if y = 0, we have either x = 0 or x = 1, giving two cri cal points: (1, 0)
and (0, 0).
Notes:
719

Going back to fx , now assume 2x + 1 = 0, i.e., that x = 1/2, then consider
fy = 0:
fy = 0
2
x + 2y + x = 0,
1/4 + 2y 1/2 = 0
y = 1/8.
z
0.5
0.2
0.2
0.4
.
Figure 12.30: Graphing f from Example
436 and its rela ve extrema.
and since x = 1/2, we have
Thus if x = 1/2, y = 1/8 giving the cri cal point (1/2, 1/8).
With D = 4y (2x + 1)2 , we apply the Second Deriva ve Test to each cri cal
point.
At (1, 0), D < 0, so (1, 0) is a saddle point.
At (0, 0), D < 0, so (0, 0) is also a saddle point.
At (1/2, 1/8), D > 0 and fxx > 0, so (1/2, 1/8) is a rela ve minimum.
Figure 12.30 shows a graph of f and the three cri cal points. Note how this
func on does not vary much near the cri cal points that is, visually it is dicult to determine whether a point is a saddle point or rela ve minimum (or even
a cri cal point at all!). This is one reason why the Second Deriva ve Test is so
important to have. .
Constrained Op miza on
When op mizing func ons of one variable such as y = f(x), we made use
of Theorem 25, the Extreme Value Theorem, that said that over a closed interval I, a con nuous func on has both a maximum and minimum value. To nd
these maximum and minimum values, we evaluated f at all cri cal points in the
interval, as well as at the endpoints (the boundary) of the interval.
A similar theorem and procedure applies to func ons of two variables. A
con nuous func on over a closed set also a ains a maximum and minimum
value (see the following theorem). We can nd these values by evalua ng the
func on at the cri cal values in the set and over the boundary of the set. A er
formally sta ng this extreme value theorem, we give examples.
Theorem 116
Extreme Value Theorem
Let z = f(x, y) be a con nuous func on on a closed, bounded set S. Then

f has a maximum and minimum value on S.
. Example 437
.Finding extrema on a closed set
Let f(x, y) = x2 y2 + 5 and let S be the triangle with ver ces (1, 2), (0, 1)
and (2, 2). Find the maximum and minimum values of f on S.
Notes:
720
12.8 Extreme Values

It can help to see a graph of f along with the set S. In Figure
S
12.31(a) the triangle dening S is shown in the x-y plane in a dashed line. Above
it is the surface of f; we are only concerned with the por on of f enclosed by the
triangle on its surface.
We begin by nding the cri cal points of f. With fx = 2x and fy = 2y, we
nd only one cri cal point, at (0, 0).
We now nd the maximum and minimum values that f a ains along the
boundary of S, that is, along the edges of the triangle. In Figure 12.31(b) we
see the triangle sketched in the plane with the equa ons of the lines forming its
edges labeled.
Start with the bo om edge, along the line y = 2. If y is 2, then on
the surface, we are considering points f(x, 2); that is, our func on reduces to
f(x, 2) = x2 (2)2 + 5 = x2 + 1 = f1 (x). We want to maximize/minimize
f1 (x) = x2 + 1 on the interval [1, 2]. To do so, we evaluate f1 (x) at its cri cal
points and at the endpoints. .
The cri cal points of f1 are found by se ng its deriva ve equal to 0:
f 1 (x) = 0
z
5
y
2
.
2
(a)
y
1
x = 0.
f(2, 2) = 5.
No ce how evalua ng f1 at a point is the same as evalua ng f at its corresponding point.

We need to do this process twice more, for the other two edges of the triangle.
Along the le edge, along the line y = 3x + 1, we subs tute 3x + 1 in for y
in f(x, y):
1
y=
3x +
f1 (2) = 5
f(1, 2) = 2
f(0, 2) = 1
2x
3/
f1 (1) = 2
f1 (0) = 1
y=
Evalua ng f1 at this cri cal point, and at the endpoints of [1, 1] gives:
y = 2
(b)
Figure 12.31: Plo ng the surface of f
along with the restricted domain S.
f(x, y) = f(x, 3x + 1) = x2 (3x + 1)2 + 5 = 8x2 6x + 4 = f2 (x).

We want the maximum and minimum values of f2 on the interval [1, 0], so we
evaluate f2 at its cri cal points and the endpoints of the interval. We nd the
cri cal points:
f 2 (x) = 16x 6 = 0
x = 3/8.
Evaluate f2 at its cri cal point and the endpoints of [1, 0]:
f2 (1) = 2
f2 (3/8) = 41/8 = 5.125
f2 (0) = 1
f(1, 2) = 2
f(3/8, 0.125) = 5.125
f(0, 1) = 4.
Notes:
721

Finally, we evaluate f along the right edge of the triangle, where y = 3/2x+
1.
5
f(x, y) = f(x, 3/2x + 1) = x2 (3/2x + 1)2 + 5 = x2 + 3x + 4 = f3 (x).
4
The cri cal points of f3 (x) are:
f 3 (x) = 0
z
x = 6/5 = 1.2.
We evaluate f3 at this cri cal point and at the endpoints of the interval [0, 2]:
y
2
f3 (0) = 4
f3 (1.2) = 5.8
f(0, 1) = 4
f(1.2, 0.8) = 5.8
f3 (2) = 5
f(2, 2) = 5.
x
2
.
.
Figure 12.32: The surface of f along with

important points along the boundary of S
and the interior.
One last point to test: the cri cal point of f, (0, 0). We nd f(0, 0) = 5.
We have evaluated f at a total of 7 dierent places, all shown in Figure 12.32.
We checked each vertex of the triangle twice, as each showed up as the endpoint of an interval twice. Of all the z-values found, the maximum is 5.8, found
at (1.2, 0.8); the minimum is 1, found at (0, 2). .
This por on of the text is en tled Constrained Op miza on because we
want to op mize a func on (i.e., nd its maximum and/or minimum values)
subject to a constraint some limit to what values the func on can a ain. In
the previous example, we constrained ourselves to considering a func on only
within the boundary of a triangle. This was largely arbitrary; the func on and
the boundary were chosen just as an example, with no real meaning behind
the func on or the chosen constraint.
However, solving constrained op miza on problems is a very important topic
in applied mathema cs. The techniques developed here are the basis for solving
larger problems, where more than two variables are involved.
We illustrate the technique once more with a classic problem.
. Example 438
.Constrained Op miza on
The U.S. Postal Service states that the girth+length of Standard Post Package
must not exceed 130. Given a rectangular box, the length is the longest side,
and the girth is twice the width+height.
Given a rectangular box where the width and height are equal, what are the
dimensions of the box that give the maximum volume subject to the constraint
of the size of a Standard Post Package?
Let w, h and denote the width, height and length of a rectS
angular box; we assume here that w = h. The girth is then 2(w + h) = 4w. The
Notes:
722
12.8 Extreme Values

volume of the box is V(w, ) = wh = w2 . We wish to maximize this volume
subject to the constraint 4w + 130, or 130 4w. (Common sense also
indicates that > 0, w > 0.)
We begin by nding the cri cal values of V. We nd that Vw = 2w and
V = w2 ; these are simultaneously 0 only at (0, 0). This gives a volume of 0, so
we can ignore this cri cal point.
We now consider the volume along the constraint = 130 4w. Along this
line, we have:
V(w) = V(w, 130 4w) = w2 (130 4w) = 130w2 4w3 = V1 (w).

The constraint is applicable on the w-interval [0, 32.5] as indicated in the gure.
Thus we want to maximize V1 on [0, 32.5].
Finding the cri cal values of V1 , we take the deriva ve and set it equal to 0:
V 1 (w) = 260w12w2 = 0
w(26012w) = 0
w = 0,
260
21.67.
12
We found two cri cal values: when w = 0 and when w = 21.67. We again
ignore the w = 0 solu on; the maximum volume, subject to the constraint,
comes at w = h = 21.67, = 130 4(21.6) = 43.33. This gives a volume of
V(21.67, 43.33) 19, 408in3 .
The volume func on V(w, ) is shown in Figure 12.33 along with the constraint = 130 4w. As done previously, the constraint is drawn dashed in
the x-y plane and also along the surface of the func on. The point where the
volume is maximized is indicated. .
100
.
.
50
20
Figure 12.33: Graphing the volume of a

box with girth 4w and length , subject to
a size constraint.
It is hard to overemphasize the importance of op miza on. In the real

world, we rou nely seek to make something be er. By expressing the something as a mathema cal func on, making something be er means op mize
some func on.
The techniques shown here are only the beginning of an incredibly important
eld. Many func ons that we seek to op mize are incredibly complex, making
the step of nd the gradient and set it equal to 0 highly nontrivial. Mastery
of the principles here are key to being able to tackle these more complicated
problems.
Notes:
723
Exercises 12.8
1
x2 + y2 + 1
Terms and Concepts
8. f(x, y) =
1. T/F: Theorem 114 states that if f has a cri cal point at P,

then f has a rela ve extrema at P.
9. f(x, y) = x2 + y3 3y + 1
2. T/F: A point P is a cri cal point of f if fx and fy are both 0 at

P.
3. T/F: A point P is a cri cal point of f if fx or fy are undened
at P.
4. Explain what it means to solve a constrained op miza on
problem.
1 3
1
x x + y3 4y
3
3
11. f(x, y) = x2 y2
12. f(x, y) = x4 2x2 + y3 27y 15
13. f(x, y) = 16 (x 3)2 y2
14. f(x, y) = x2 + y2
Exercises 15 18, nd the absolute maximum and minimum
of the func on subject to the given constraint.
Problems
Exercises 5 14, nd the cri cal points of the given func on.
Use the Second Deriva ve Test to determine if each cri cal
point corresponds to a rela ve maximum, minimum, or saddle point.
5. f(x, y) = 12 x2 + 2y2 8y + 4x
6. f(x, y) = x2 + 4x + y2 9y + 3xy
7. f(x, y) = x + 3y 6y + 4xy
2
10. f(x, y) =
15. f(x, y) = x2 + y2 + y + 1, constrained to the triangle with

ver ces (0, 1), (1, 1) and (1, 1).
16. f(x, y) = 5x 7y, constrained to the region bounded by
y = x2 and y = 1.
17. f(x, y) = x2 + 2x + y2 + 2y, constrained to the region
bounded by the circle x2 + y2 = 4.
18. f(x, y) = 3y 2x2 , constrained to the region bounded by
the parabola y = x2 + x 1 and the line y = x.
13: M
13.1 Iterated Integrals and Area

In Chapter 12 we found that it was useful to dieren ate func ons of several
variables with respect to one variable, while trea ng all the other variables as
constants or coecients. We can integrate func ons of several variables in a
similar way. For instance, if we are told that fx (x, y) = 2xy, we can treat y as
staying constant and integrate to obtain f(x, y):
f(x, y) = fx (x, y) dx
= 2xy dx
= x2 y + C.
Make a careful note about the constant of integra on, C. This constant is
something with a deriva ve of 0 with respect to x, so it could be any expression that contains only constants and func ons of y. For instance, if f(x, y) =
x2 y + sin y + y3 + 17, then fx (x, y) = 2xy. To signify that C is actually a func on
of y, we write:
f(x, y) = fx (x, y) dx = x2 y + C(y).

Using this process we can even evaluate denite integrals.
. Example 439
Integra
2y ng func ons of more than one variable
Evaluate the integral
2xy dx.
1
We nd the indenite integral as before, then apply the FunS

damental Theorem of Calculus to evaluate the denite integral:
2y
2y

2xy dx = x2 y
1
= (2y)2 y 2(1)y
.
= 4y3 2y.
Chapter 13 Mul ple Integra on

We can also integrate with respect to y. In general,
h2 (y)
h1 (y)
h1 (y)
and
h2 (y)
(
)
(
)

fx (x, y) dx = f(x, y)
= f h2 (y), y f h1 (y), y ,
g2 (x)
g2 (x)
(
)
(
)

fy (x, y) dy = f(x, y)
= f x, g2 (x) f x, g1 (x) .
g1 (x)
g1 (x)
Note that when integra ng with respect to x, the bounds are func ons of y
(of the form x = h1 (y) and x = h2 (y)) and the nal result is also a func on of y.
When integra ng with respect to y, the bounds are func ons of x (of the form
y = g1 (x) and y = g2 (x)) and the nal result is a func on of x. Another example
will help us understand this.
. Example 440
Integra ng func ons of more than one variable
x
( 3 3
)
Evaluate
5x y + 6y2 dy.
1
We consider x as staying constant and integrate with respect
to y:
) x

)1 (
)
(
5
5
= x3 x2 + 2x3 x3 + 2
2
2
9 3 5
= x x 2.
2
2
)
5x3 y3 + 6y2 dy =
5x3 y2
6y3
+
2
3
Note how the bounds of the integral are from y = 1 to y = x and that the nal
answer is a func on of x. .
In the previous example, we integrated a func on with respect to y and
ended up with a func on of x. We can integrate this as well. This process is
known as iterated integra on, or mul ple integra on.
. Example 441
)
2 ( x .Integra ng an integral
( 3 3
)
Evaluate
5x y + 6y2 dy dx.
1
We follow a standard order of opera ons and perform the

S
opera ons inside parentheses rst (which is the integral evaluated in Example
Notes:
726

440.)
3 3
5x y
+ 6y
)
dy
] x )
5x3 y2
6y3
dx =
+
dx

2
3
1
1
)
2(
9 3 5
=
x x 2 dx
2
2
1
(
) 2
9 4 5 2

=
x x 2x

8
4
([
89
.
=
8
Note how the bounds of x were x = 1 to x = 2 and the nal result was a number. .
The previous example showed how we could perform something called an
iterated integral; we do not yet know why we would be interested in doing so
nor what the result, such as the number 89/8, means. Before we inves gate
these ques ons, we oer some deni ons.
Deni on 100
Iterated Integra on
Iterated integra on is the process of repeatedly integra ng the results

of previous integra ons. Integra ng one integral is denoted as follows.
Let a, b, c and d be numbers and let g1 (x), g2 (x), h1 (y) and h2 (y) be
func ons of x and y, respec vely. Then:
)
d h2 (y)
d ( h2 (y)
1.
f(x, y) dx dy =
f(x, y) dx dy.
c
h1 (y)
b
g2 (x)
2.
h1 (y)
b
g2 (x)
f(x, y) dy dx =
a
g1 (x)
)
f(x, y) dy dx.
g1 (x)
Again make note of the bounds of these iterated integrals.

d h2 (y)
With
f(x, y) dx dy, x varies from h1 (y) to h2 (y), whereas y varies from
c
h1 (y)
c to d. That is, the bounds of x are curves, the curves x = h1 (y) and x = h2 (y),
whereas the bounds of y are constants, y = c and y = d. It is useful to remember
that when se ng up and evalua ng such iterated integrals, we integrate from
Notes:
727

curve to curve, then from point to point.
We now begin to inves gate why we are interested in iterated integrals and
what they mean.
Area of a plane region

y
Consider the plane region R bounded by a x b and g1 (x) y g2 (x),

shown in Figure 13.1. We learned in Sec on 7.1 that the area of R is given by
y = g2 (x)
)
g2 (x) g1 (x) dx.
y = g1 (x)
x
(
)
We can also view the expression g2 (x) g1 (x) as
Figure 13.1: Calcula ng the area of a

plane region R with an iterated integral.
(
)
g2 (x) g1 (x) =
g2 (x)
g2 (x)
1 dy =
g1 (x)
dy,
g1 (x)
meaning we can express the area of R as an iterated integral:
area of R =
y
(
)
g2 (x) g1 (x) dx =
g2 (x)
g2 (x)
dy dx =
a
g1 (x)
dy dx.
a
g1 (x)
d
x = h2 (y)
In short: a certain iterated integral can be viewed as giving the area of a

plane region.
A region R could also be dened by c y d and h1 (y) x h2 (x), as

shown in Figure 13.2. Using a process similar to that above, we have
x = h1 (y)
c
h2 (y)
the area of R =

plane region R with an iterated integral.
dx dy.
c
We state this formally in a theorem.
Notes:
728
h1 (y)
Theorem 117
Area of a plane region
1. Let R be a plane region bounded by a x b and g1 (x) y

g2 (x), where g1 and g2 are con nuous func ons on [a, b]. The area
A of R is

A=
a
g2 (x)
dy dx.
g1 (x)
2. Let R be a plane region bounded by c y d and h1 (y) x

h2 (y), where h1 and h2 are con nuous func ons on [c, d]. The area
A of R is

d
h2 (y)
dx dy.
A=
h1 (y)
The following examples should help us understand this theorem.

3
. Example 442
Area of a rectangle
Find the area A of the rectangle with corners (1, 1) and (3, 3), as shown in
Figure 13.3.
Mul ple integra on is obviously overkill in this situa on, but
S
we proceed to establish its use.
The region R is bounded by x = 1, x = 3, y = 1 and y = 3. Choosing to
integrate with respect to y rst, we have
3 3
3 ( 3 )
3
3

A=
1 dy dx =
y
dx =
2 dx = 2x = 8.
1
We could also integrate with respect to x rst, giving:

3 3
3 ( 3 )
3
3

A=
1 dx dy =
x
dy =
4 dy = 4y = 8.
1
Figure 13.3: Calcula ng the area of a rectangle with an iterated integral in Example
442.
y
5
. Example 443
.Area of a triangle
Find the area A of the triangle with ver ces at (1, 1), (3, 1) and (5, 5), as shown
in Figure 13.4.
The triangle is bounded by the lines as shown in the gure.

Choosing to integrate with respect to x rst gives that x is bounded by x = y
y=
3
2
Notes:
Clearly there are simpler ways to nd this area, but it is interes ng to note
that this method works. .
x
1
2x
y=
R
y=1
x
1
Figure 13.4: Calcula ng the area of a triangle with iterated integrals in Example
443.
729

to x = y+5
2 , while y is bounded by y = 1 to y = 5. (Recall that since x-values
increase from le to right, the le most curve, x = y, is the lower bound and the
rightmost curve, x = (y + 5)/2, is the upper bound.) The area is
A=
1
y+5
2
dx dy
( y+5 )
2
x
dy
y
1
)
5(
1
5
=
y+
dy
2
2
1
(
)
1
5 5
= y2 + y
4
2
1
= 4.
We can also nd the area by integra ng with respect to y rst. In this situaon, though, we have two func ons that act as the lower bound for the region
R, y = 1 and y = 2x 5. This requires us to use two iterated integrals. Note
how the x-bounds are dierent for each integral:
3 x
5 x
A=
1 dy dx
+
1 dy dx
1
(
)
x 1 dx
( )x
y dx
2x5
2x5
( )x
y
dx
)
x + 5 dx
=2
= 4.
As expected, we get the same answer both ways. .

. Example 444
.Area of a plane region
Find the area of the region enclosed by y = 2x and y = x2 , as shown in Figure
13.5.
y
4
3
2x
=
y
x
1
730

plane region with iterated integrals in Example 444.
1 dy dx =
Once again well nd the area of the region using both or-
ders of integra on.

Using dy dx:
2 2x
Notes:
x2
(
1 )2
4
(2x x2 ) dx = x2 x3 = .
3
3
0

Using dx dy:
4 y
1 dx dy =
y/2
( y y/2) dy =
)
2 3/2 1 2 4
4
y y = .
3
4
3
0
Changing Order of Integra on

In each of the previous examples, we have been given a region R and found
the bounds needed to nd the area of R using both orders of integra on. We
integrated using both orders of opera on to demonstrate their equality.
We now approach the skill of describing a region using both orders of integra on from a dierent perspec ve. Instead of star ng with a region and crea ng iterated integrals, we will start with an iterated integral and rewrite it in
the other integra on order. To do so, well need to understand the region over
which we are integra ng.
The simplest of all cases is when both integrals are bound by constants. The
region described by these bounds is a rectangle (see Example 442), and so:
b d
d b
1 dy dx =
1 dx dy.
a
When the inner integrals bounds are not constants, it is generally very useful
to sketch the bounds to determine what the region we are integra ng over looks
like. From the sketch we can then rewrite the integral with the other order of
integra on.
Examples will help us develop this skill.
. Example 445
Changing the order of integra on

6 x/3
Rewrite the iterated integral
1 dy dx with the order of integra on dx dy.
0
We need to use the bounds of integra on to determine the

region we are integra ng over.
The bounds tell us that y is bounded by 0 and x/3; x is bounded by 0 and 6.
We plot these four curves: y = 0, y = x/3, x = 0 and x = 6 to nd the region
described by the bounds. Figure 13.6 shows these curves, indica ng that R is a
triangle.
To change the order of integra on, we need to consider the curves that
bound the x-values. We see that the lower bound is x = 3y and the upper
bound
is x = 6. The bounds on y are 0 to 2. Thus we can rewrite the integral as
y=
3
x/
R
x
Figure 13.6: Sketching the region R described by the iterated integral in Example 445.
1 dx dy.
0
3y
Notes:
731

. Example 446
Changing the order of integra on

4 (y+4)/2
Change the order of integra on of
1 dx dy.
y2 /4
0
y
4
R
x=
(y
2 4
y/
4)
/2
x=
x
2
Figure 13.7: Drawing the region determined by the bounds of integra on in Example 446.
We sketch the region described by the bounds to help us

S
change the integra on order. x is bounded below and above (i.e., to the le and
right) by x = y2 /4 and x = (y + 4)/2 respec vely, and y is bounded between 0
and 4. Graphing the previous curves, we nd the region R to be that shown in
Figure 13.7.
To change the order of integra on, we need to establish curves that bound
y. The gure makes it clear that there are two lower bounds for y: y = 0 on
0 x 2, and y = 2x 4 on 2 x 4. Thus we need two double integrals.
The upper bound for each is y = 2 x. Thus we have
(y+4)/2
1 dx dy =
0
y2 /4
2 x
2 x
1 dy dx +
0
1 dy dx.
2
2x4
.
This sec on has introduced a new concept, the iterated integral. We developed one applica on for iterated integra on: area between curves. However,
this is not new, for we already know how to nd areas bounded by curves.
In the next sec on we apply iterated integra on to solve problems we currently do not know how to handle.
Notes:
732
Exercises 13.1
Terms and Concepts
1. When integra ng fx (x, y) with respect to x, the constant of

integra on C is really which: C(x) or C(y)? What does this
mean?
2. Integra ng an integral is called
11.
.
y
g1 (x)
Problems
(b)
6.
(a)
)
6x2 + 4xy 3y2 dy dx
(b)
7.
(
)
2x cos y + sin x dx dy
2
1
y2
(a)
9.
y2
(a)
)
x y dx dy
6
y
)
cos x sin y dx dy
In Exercises 11 16, a graph of a planar region R is given. Give

the iterated integrals, with both orders of integra on dy dx
and dx dy, that give the area of R. Evaluate one the iterated
integrals to nd the area.
10
12
y
1
x(
10.
)
1
(a)
dy
1 + x2
0
)
2 x(
1
(b)
dy dx
1 + x2
1
0
cos x sin y dx
(b)
0
14.
2 3
y/
x=
x y dx
(b)
y
1
( 2
)
x y y + 2 dy dx
13.
( 2
)
x y y + 2 dy
(b)
8.
(a)
)
2x cos y + sin x dx
/2
2
2
12.
In Exercises 5 10, evaluate the integral and subsequent iterated integral.

5
( 2
)
5. (a)
6x + 4xy 3y2 dy
4. One understanding of an iterated integral is that

b g2 (x)
of a plane region.
dy dx gives the
3. When evalua ng an iterated integral, we integrate from

to
, then from
to
.
y=
0.5
15.
R
4
y=
0.5
0.5
0.5
x
1
4x2
17.
dy dx
2
4x
2
24y2
dx dy
55x
19.
R
y=
55x2
dy dx
0
18.
16.
9x2
20.
3
In Exercises 17 22, iterated integrals are given that compute

the area of a region R in the x-y plane. Sketch the region R,
and give the iterated integral(s) that give the area of R with
the opposite order of integra on.
21.
dy dx
9x2
dx dy +
dy dx
(x1)/2
dx dy
1
(1x)/2
22.
1
y2
13.2 Double Integra on and Volume

b
The denite integral of f over [a, b], a f(x) dx, was introduced as the signed
area under the curve. We approximated the value of this area by rst subdividing [a, b] into n subintervals, where the i th subinterval has length xi , and le ng
ci be any value in the i th subinterval. We formed rectangles that approximated
part of the area under the curve with width xi , height f(ci ), and hence with
area f(ci )xi . Summing up all rectangles gave an approxima on of the denite
integral, and Theorem 38 stated that
b
f(x) dx = lim
f(ci )xi ,
x0
connec ng sums of rectangles to area under the curve.
0.5
We use a similar approach in this sec on to nd volume under a surface.

x
1
Let R be a closed, bounded region in the x-y plane and let z = f(x, y) be
a con nuous func on dened on R. We wish to nd the signed volume under
the surface of f over R. (We use the term signed volume to denote that space
above the x-y plane, under f, will have a posi ve volume; space above f and
under the x-y plane will have a nega ve volume, similar to the no on of signed
area used before.)
We start by par oning R into n rectangular subregions as shown in Figure
13.8(a). For simplicitys sake, we let all widths be x and all heights be y.
Note that the sum of the areas of the rectangles is not equal to the area of R,
but rather is a close approxima on. Arbitrarily number the rectangles 1 through
n, and pick a point (xi , yi ) in the i th subregion.
The volume of the rectangular solid whose base is the i th subregion and
whose height is f(xi , yi ) is Vi = f(xi , yi )xy. Such a solid is shown in Figure
13.8(b). Note how this rectangular solid only approximates the true volume under the surface; part of the solid is above the surface and part is below.
For each subregion Ri used to approximate R, create the rectangular solid
with base area xy and height f(xi , yi ). The sum of all rectangular solids is
n
0.5
.
(a)
z
2
.
1
1y
f(xi , yi )xy.
i=1
This approximates the signed volume under f over R. As we have done before,
to get a be er approxima on we can use more rectangles to approximate the
region R.
In general, each rectangle could have a dierent width xj and height yk ,
giving the i th rectangle an area Ai = xj yk and the i th rectangular solid a
(b)
Figure 13.8: Developing a method for
nding signed volume under a surface.
Notes:
735

volume of f(xi , yi )Ai . Let |A| denote the length of the longest diagonal of all
rectangles in the subdivision of R; |A| 0 means each rectangles width and
height are both approaching 0. If f is a con nuous func on, as A shrinks (and
n
hence n ) the summa on

f(xi , yi )Ai approximates the signed volume
i=1
be er and be er. This leads to a deni on.

.
Deni on 101
Double Integral, Signed Volume
Let z = f(x, y) be a con nuous func on dened over a closed region R

in the x-y plane. The signed volume V under f over R is denoted by the
double integral
f(x, y) dA.
V=
R
Alternate nota ons for the double integral are
f(x, y) dA =
f(x, y) dx dy =
f(x, y) dy dx.
R
The deni on above does not state how to nd the signed volume, though
the nota on oers a hint. We need the next two theorems to evaluate double
integrals to nd volume.
.
Theorem 118
Double Integrals and Signed Volume

in the x-y plane. Then the signed volume V under f over R is
V=
f(x, y) dA = lim
R
|A|0
f(xi , yi )Ai .
i=1
This theorem states that we can nd the exact signed volume using a limit
of sums. The par on of the region R is not specied, so any par oning where
the diagonal of each rectangle shrinks to 0 results in the same answer.
This does not oer a very sa sfying way of compu ng area, though. Our
experience has shown that evalua ng the limits of sums can be tedious. We
seek a more direct method.
Recall Theorem 54 in Sec on 7.2. This stated that if A(x) gives the crossb
sec onal area of a solid at x, then a A(x) dx gave the volume of that solid over
Notes:
736

[a, b].
Consider Figure 13.9, where a surface z = f(x, y) is drawn over a region R.
Fixing a par cular x value, we can consider the area under f over R where x has
that xed value. That area can be found with a denite integral, namely
z
2
g2 (x)
A(x) =
f(x, y) dy.
g1 (x)
Remember that though the integrand contains x, we are viewing x as xed.

Also note that the bounds of integra on are func ons of x: the bounds depend
on the value of x.
As A(x) is a cross-sec onal area func on, we can nd the signed volume V
under f by integra ng it:
)
(

b
V=
g2 (x)
A(x) dx =
a
g2 (x)
f(x, y) dy dx =
a
g1 (x)
f(x, y) dy dx.
a
g1 (x)
This gives a concrete method for nding signed volume under a surface. We
could do a similar procedure where we started with y xed, resul ng in a iterated
integral with the order of integra on dx dy. The following theorem states that
both methods give the same result, which is the value of the double integral. It
is such an important theorem it has a name associated with it.
Theorem 119
.
R
.
Figure 13.9: Finding volume under a surface by sweeping out a crosssec onal
area.
Fubinis Theorem
Let R be a closed, bounded region in the x-y plane and let z = f(x, y) be
a con nuous func on on R.
1. If R is bounded by a x b and g1 (x) y g2 (x), where g1
and g2 are con nuous func ons on [a, b], then
f(x, y) dA =
R
g2 (x)
f(x, y) dy dx.
g1 (x)
2. If R is bounded by c y d and h1 (y) x h2 (y), where h1

and h2 are con nuous func ons on [c, d], then
h2 (y)
f(x, y) dA =
R
f(x, y) dx dy.
c
h1 (y)
Notes:
737

Note that once again the bounds of integra on follow the curve to curve,
point to point pa ern discussed in the previous sec on. In fact, one of the
main points of the previous sec on is developing the skill of describing a region
R with the bounds of an iterated integral. Once this skill is developed, we can
use double integrals to compute many quan es, not just signed volume under
a surface.
. Example 447
Evalua ng a double integral
Let f(x, y) = xy+ey . Find the signed volume under f on the region R, which is the
rectangle with corners (3, 1) and (4, 2) pictured in Figure 13.10, using Fubinis
Theorem and both orders of integra on.
10
)
(
We wish to evaluate R xy + ey dA. As R is a rectangle,
the bounds are easily described as 3 x 4 and 1 y 2.
S
y
2
1
R
1
Using the order dy dx:
(
)
y
xy + e dA =
R
Figure 13.10: Finding the signed volume

under a surface in Example 447.
)
(
xy + ey dy dx
3
1
]2 )
4 ([

1 2
y
=
dx
xy + e
2
3
1
)
4(
3
=
x + e2 e dx
2
3
)
(
) 2
3 2 ( 2
x + e e x
=
4
4
21
=
+ e2 e 9.92.
4
Now we check the validity of Fubinis Theorem by using the order dx dy:
2 4
(
)
(
)
y
xy + e dA =
xy + ey dx dy
R
1
3
]4 )
2 ([

1 2
y
=
x y + xe dy
2
1
3
)
2(
7
y + ey dy
=
2
1
(
)2

7 2
y
=
y +e
4
1
21
+ e2 e 9.92.
=
4
Both orders of integra on return the same result, as expected. .
Notes:
738

. Example 448
Evalua ng a)double integral
(
Evaluate R 3xy x2 y2 + 6 dA, where R is the triangle bounded by x = 0,
y = 0 and x/2 + y = 1, as shown in Figure 13.11.
While it is not specied which order we are to use, we will
S
evaluate the double integral using both orders to help drive home the point that
it does not ma er which order we use.
z
8
6
4
Using the order dy dx: The bounds on y go from curve to curve, i.e., 0
y 1 x/2, and the bounds on x go from point to point, i.e., 0 x 2.
2 2x +1
)
)
(3xy x2 y2 + 6 dy dx
(3xy x2 y2 + 6 dA =
R
) 2x +1
2

1 3
3 2
2
=
xy x y y + 6y
dx
2
3
0
0
)
2(
11 3 11 2
17
=
x x x
dx
12
4
3
0
)2
(
11 4 11 3 1 2 17
=
x x x x
48
12
2
3
(
1
R
.
.

under the surface in Example 448.
17
=
= 5.6.
3
Now lets consider the order dx dy. Here x goes from curve to curve, 0
x 2 2y, and y goes from point to point, 0 y 1:
1 22y
)
)
(3xy x2 y2 + 6 dA =
(3xy x2 y2 + 6 dx dy
R
)22y

1 3
3 2
2
=
x y x xy + 6x
dy
2
3
0
0
)
(
1
28
32 3
y 22y2 + 2y +
dy
=
3
3
0
(
)1
8 4 22 3
28
=
y y + y2 + y
3
3
3
17
=
= 5.6.
3
We obtained the same result using both orders of integra on. .
Note how in these two examples that the bounds of integra on depend only
on R; the bounds of integra on have nothing to do with f(x, y). This is an important concept, so we include it as a Key Idea.
Notes:
739

.
Key Idea 56
Double Integra on Bounds
When evalua ng R f(x, y) dA using.

an iterated integral, the bounds of
integra on depend only on R. The surface f does not determine the
bounds of integra on.
Before doing another example, we give some proper es of double integrals.

Each should make sense if we view them in the context of nding signed volume
under a surface, over a region.
.
Theorem 120
Proper es of Double Integrals
Let f and g be con nuous func ons over a closed, bounded plane region
R, and let c be a constant.
1.
cf(x, y) dA = c
f(x, y) dA.
R1
R2
2.
R
f(x, y) dA
f(x, y) dA 0.
R
4. If f(x, y) g(x, y) on R, then
Figure 13.12: R is the union of two

nonoverlapping regions, R1 and R2 .
g(x, y) dA
3. If f(x, y) 0 on R, then
)
f(x, y) g(x, y) dA =
f(x, y) dA
g(x, y) dA.
5. Let R be the union of two nonoverlapping regions, R = R1

(see Figure 13.12). Then
f(x, y) dA =
f(x, y) dA +
f(x, y) dA.
R
R1
R2
R2
1z
. Example 449
.Evalua ng a double integral
Let f(x, y) = sin x cos y and R be the triangle with verces (1, 0), (1, 0) and
(0, 1) (see Figure 13.13). Evaluate the double integral R f(x, y) dA.
x
R
1 1
y

under a surface in Example 449.
740
If we a empt to integrate using an iterated integral with the

S
order dy dx, note how there are two upper bounds on R meaning well need to
use two iterated integrals. We can split the triangle into to regions along the
Notes:

y-axis, then use Theorem 120, part 5.
Instead, lets use the order dx dy. The curves bounding x are y 1 x
1 y; the bounds on y are 0 y 1. This gives us:
1y
f(x, y) dA =
R
sin x cos y dx dy
0
y1
y1
)1y

cos x cos y
dy
(
)
cos y cos(1 y) + cos(y 1) dy.
Recall that the cosine func on is an even func on; that is, cos x = cos(x).
Therefore, from the last integral above, we have cos(y 1) = cos(1 y). Thus
the integrand simplies to 0, and we have
f(x, y) dA =
R
0 dy
0
= 0.
It turns out that over R, there is just as much volume above the x-y plane as below (look again at Figure 13.13), giving a nal signed volume of 0. .
. Example 450
.Evalua ng a double integral
Evaluate R (4 y) dA, where R is the region bounded by the parabolas y2 = 4x

and x2 = 4y, graphed in Figure 13.14.
Graphing each curve can help us nd their points of intersec on; analy cally, the second equa on tells us that y = x2 /4. Subs tu ng
this value in for y in the rst equa on gives us x4 /16 = 4x. Solving for x:
S
x4
= 4x
16
x4 64x = 0
x(x3 64) = 0
x = 0, 4.
Thus weve found analy cally what was easy to approximate graphically: the
regions intersect at (0, 0) and (4, 4), as shown in Figure 13.14.
We now choose an order of integra on: dy dx or dx dy? Either order works;
since the integrand does not contain x, choosing dx dy might be simpler at
least, the rst integral is very simple.
2
x
2
4
.
Figure 13.14: Finding the volume under
the surface in Example 450.
Notes:
741

Thus we have the following curve to curve, point to point bounds: y2 /4
x 2 y, and 0 y 4.
(4 y) dA =
R
(4 y) dx dy
(
)2 y
x(4 y) 2 dy
y /4
2 y
y2 /4
4( 3
((
)
)
y
y2 )(
=
2 y
4 y) dy =
y2 2y3/2 + 8y1/2 dy
4
4
0
0
( 4
)
3
5/2
3/2 4
y
y
4y
16y

=
+

16
3
5
3
0
176
=
= 11.73.
15
4
The signed volume under the surface f is about 11.7 cubic units. .
In the previous sec on we prac ced changing the order of integra on of a
given iterated integral, where the region R was not explicitly given. Changing
the bounds of an integral is more than just an test of understanding. Rather,
there are cases where integra ng in one order is really hard, if not impossible,
whereas integra ng with the other order is feasible.
. Example 451
.Changingtheorder of integra on
3
Rewrite the iterated integral

0
y=
x
1
Figure 13.15: Determining the region R

determined by the bounds of integra on
in Example 451.
Once again we make a sketch of the region over which we

S
are integra ng to facilitate changing the order. The bounds on x are from x = y
to x = 3; the bounds on y are from y = 0 to y = 3. These curves are sketched
in Figure 13.15, enclosing the region R.
To change the bounds, note that the curves bounding y are y = 0 up to
y = x; the triangle is enclosed between x = 0 and x = 3. Thus the new
bounds ofintegra on are 0 y x and 0 x 3, giving the iterated in3
tegral
0
ex dy dx.
How easy is it to evaluate each iterated integral? Consider the order of integra ng dx dy, as given
in the original problem. The rst indenite integral we
2
need to evaluate is ex dx; we have stated before (see Sec on 5.5) that this
integral cannot be evaluated in terms of elementary func ons. We are stuck.
Changing the order of integra on makes a big dierence here. In the second
Notes:
742
on the feasibility to evaluate each integral.
ex dx dy with the order dy dx. Comment
z
x2
iterated integral, we are faced with e dy; integra ng with respect to y gives
2
us yex + C, and the rst denite integral evaluates to
x
2
2
ex dy = xex .
0
Thus
x2
e
0
dy dx =
)
2
xex dx.
This last integral is easy to evaluate with subs tu on, giving a nal answer of
1
9
2 (1 e ) 0.5. Figure 13.16 shows the surface over R.
In short, evalua ng one iterated integral is impossible; the other iterated integral is rela vely simple. .
Deni on 22 denes the average value of a singlevariable func on f(x) on
the interval [a, b] as
b
1
average value of f(x) on [a, b] =
f(x) dx;
ba a
Figure 13.16: Showing the surface f dened in Example 451 over its region R.
that is, it is the area under f over an interval divided by the length of the interval. We make an analogous statement here: the average value of z = f(x, y)
over a region R is the volume under f over R divided by the area of R.
Deni on 102
The Average Value of f on R

in the x-y plane. The average value of f on R is
f(x, y) dA
R
average value of f on R =
.
dA
. Example 452
.Finding average value of a func on over a region R
Find the average value of f(x, y) = 4 y over the region R, which is bounded by
the parabolas y2 = 4x and x2 = 4y. Note: this is the same func on and region
as used in Example 450.
S
In Example 450 we found

4 2y
176
.
f(x, y) dA =
(4 y) dx dy =
15
R
0
y2 /4
Notes:
743

We nd the area of R by compu ng
R
4
dA:
2 y
dA =
dx dy =
0
y2 /4
16
.
3
Dividing the volume under the surface by the area gives the average value:
2
average value of f on R =
While the surface, as shown in Figure 13.17, covers z-values from z = 0 to z = 4,

the average z-value on R is 2.2. .
.
2
2
x
.
Figure 13.17: Finding the average value of
f in Example 452.
The previous sec on introduced the iterated integral in the context of nding the area of plane regions. This sec on has extended our understanding of
iterated integrals; now we see they can be used to nd the signed volume under
a surface.
This new understanding allows us to revisit what
we did in the previous secon. Given a region R in the plane, we computed R 1 dA; again, our understanding at the me was that we were nding the area of R. However, we can
now view the func on
z = 1 as a surface, a at surface with constant z-value of
1. The double integral R 1 dA nds the volume, under z = 1, over R, as shown
in Figure 13.18. Basic geometry tells us that if the base of a general right cylinder
has area A, its volume is A h, where h is the height. In our case, the height is
1. We were actually compu ng the volume of a solid, though we interpreted
the number as an area.
The next sec on extends our abili es to nd volumes under surfaces. Currently, some integrals are hard to compute because either the region R we are
integra ng over is hard to dene with rectangular curves, or the integrand itself is hard to deal with. Some of these problems can be solved by conver ng
everything into polar coordinates.
.
1 y
x 2
Figure 13.18: Showing how an iterated integral used to nd area also nds a certain
volume.
Notes:
744
11
176/15
=
= 2.2.
16/3
5
Exercises 13.2
Terms and Concepts
yex dA, where R is bounded by x = 0, x = y2 and
14.
R
1. An integral can be interpreted as giving the signed area over

an interval; a double integral can be interpreted as giving
the signed
over region.
2. Explain why the following statement is false: Fu b g2 (x)
binis Theorem states that
f(x, y) dy dx =
g1 (x)
a
b g2 (y)
f(x, y) dx dy.
a
y = 1.
(
)
15.
6 3x 2y dA, where R is bounded by x = 0, y = 0
R
and 3x + 2y = 6.
16.
ey dA, where R is bounded by y = ln x and
R
g1 (y)
3. Explain why if f(x, y) > 0 over a region R, then
f(x, y) dA > 0.
R
4. If R f(x, y) dA = R g(x, y) dA, does this imply f(x, y) =

g(x, y)?
1
(x 1).
e1
y=
17.
)
x3 yx dA, where R is the half of the circle x2 +y2 = 9
in the rst and second quadrants.
(
)
18.
4 3y dA, where R is bounded by y = 0, y = x/e
R
and y = ln x.
Problems
In Exercises 5 10, evaluate the given iterated integral. Also
rewrite the integral using the other order of integra on.
)
2 1 (
x
5.
+ 3 dx dy
y
1
1
/2
(sin x cos y) dx dy
6.
/2
4
0
3
9.
20.
)
3x2 y + 2 dy dx
0
3
8.
1
(x + y + 2) dx dy
y/3
x2 y dA, where R is bounded by y =
3
x and y = x3 .
x2 y2 dA, where R is the rectangle with corners
13.
R
(1, 1), (1, 1), (1, 1) and (1, 1).
2y
dx dy
x2 + y2
x tan2 y
dy dx
1 + ln y
In Exercises 23 26, nd the average value of f over the region R. No ce how these func ons and regions are related to
the iterated integrals given in Exercises 5 8.
In Exercises 11 18, set up the iterated integrals, in both orders, that evaluate the given double integral for the described
region R. Evaluate one of the iterated integrals.
11.
x2 y dA, where R is bounded by y = x and y = x2 .
12.
1
1
)
xy2 dx dy
10.
22.
1y
( )
cos y2 dy dx
1
0
1y
9
y(
21.
( 2
)
x y xy2 dx dy
y/2
/2 /2
0
x/2+2
7.
In Exercises 19 22, state why it is dicult/impossible to integrate the iterated integral in the given order of integra on.
Change the order of integra on and evaluate the new iterated integral.
4 2
2
ex dx dy
19.
x
+ 3; R is the rectangle with opposite corners
y
(1, 1) and (1, 2).
23. f(x, y) =
24. f(x, y) = sin x cos y; R is bounded by x = 0, x = ,

y = /2 and y = /2.
25. f(x, y) = 3x2 y + 2; R is bounded by the lines y = 0,
y = 2 x/2 and x = 0.
26. f(x, y) = x2 y xy2 ;
x = 3.
R is bounded by y = x, y = 1 and
13.3 Double Integra on with Polar Coordinates

/2
1
0.5
0.5
}|r2
(a)
{z
z
r1
|
(b)
Figure 13.19: Approxima ng a region R

with por ons of sectors of circles.
We have used iterated integrals to evaluate double integrals, which give the
signed volume under a surface, z = f(x, y), over a region R of the x-y plane.
The integrand is simply f(x, y), and the bounds of the integrals are determined
by the region R.
Some regions R are easy to describe using rectangular coordinates that is,
with equa ons of the form y = f(x), x = a, etc. However, some regions are
easier to handle if we represent their boundaries with polar equa ons of the
form r = f(), = , etc.
The basic form of the double integral is R f(x, y) dA. We interpret this integral as follows: over the region R, sum up lots of products of heights (given by
f(xi , yi )) and areas (given by Ai ). That is, dA represents a li le bit of area. In
rectangular coordinates, we can describe a small rectangle as having area dx dy
or dy dx the area of a rectangle is simply lengthwidth a small change in x
mes a small change in y. Thus we replace dA in the double integral with dx dy
or dy dx.
Now consider represen ng a region R with polar coordinates. Consider Figure 13.19(a). Let R be the region in the rst quadrant bounded by the curve.
We can approximate this region using the natural shape of polar coordinates:
por ons of sectors of circles. In the gure, one such region is shaded, shown
again in part (b) of the gure.
As the area of a sector of a circle with radius r, subtended by an angle , is
A = 21 r2 , we can nd the area of the shaded region. The whole sector has area
1 2
1 2
2 r2 , whereas the smaller, unshaded sector has area 2 r1 . The area of the
shaded region is the dierence of these areas:
Ai =
)( ) r2 + r1 (
)
1 2
1
1(
r2 r21 = r22 r21 =
r2 r1 .
2
2
2
2
Note that (r2 + r1 )/2 is just the average of the two radii.
To approximate the region R, we use many such subregions; doing so shrinks
the dierence r2 r1 between radii to 0 and shrinks the change in angle also
to 0. We represent these innitesimal changes in radius and angle as dr and d,
respec vely. Finally, as dr is small, r2 r1 , and so (r2 + r1 )/2 r1 . Thus, when
dr and d are small,
Ai ri dr d.
Taking a limit, where the number of subregions goes to innity and both
r2 r1 and go to 0, we get
dA = r dr d.
So to evaluate R f(x, y) dA, replace dA with r dr d. Convert the func on

z = f(x, y) to a func on with polar coordinates with the subs tu ons x = r cos ,
Notes:
746

y = r sin . Finally, nd bounds g1 () r g2 () and that describe
R. This is the key principle of this sec on, so we restate it here as a Key Idea.
Key Idea 57
Evalua ng Double Integrals with Polar Coordinates
Let R be a plane region bounded by the polar equa ons and

g1 () r g2 (). Then
g2 ()
f(x, y) dA =
R
(
)
f r cos , r sin r dr d.
g1 ()
Examples will help us understand this Key Idea.
. Example 453
Evalua ng a double integral with polar coordinates
Find the signed volume under the plane z = 4 x 2y over the circle with
equa on x2 + y2 = 1.
The bounds of the integral are determined solely by the reS
gion R over which we are integra ng. In this case, it is a circle with equa on
x2 + y2 = 1. We need to nd polar bounds for this region. It may help to review
Sec on 9.4; the bounds for this circle are 0 r 1 and 0 2.
We replace f(x, y) with f(r cos , r sin ). That means we make the following
subs tu ons:
4 x 2y 4 r cos 2r sin .
Finally, we replace dA in the double integral with r dr d. This gives the nal
iterated integral, which we evaluate:
2 1
(
)
f(x, y) dA =
4 r cos 2r sin r dr d
R
Figure 13.20: Evalua ng a double integral

with polar coordinates in Example 453.
0
1
=
0
(
)
4r r2 (cos 2 sin ) dr d
(
)1

1
=
2r2 r3 (cos 2 sin ) d
3
0
0
)
2 (
)
1(
=
2
cos 2 sin
d
3
0
)
(
) 2
1(
= 2
sin + 2 cos
3
= 4 12.566.
The surface and region R are shown in Figure 13.20. .
Notes:
747

y
. Example 454
.Evalua ng a double integral with polar coordinates
Find the volume under the paraboloid z = 4 (x 2)2 y2 over the region
bounded by the circles (x 1)2 + y2 = 1 and (x 2)2 + y2 = 4.
x
1
1
2
.
(a)
z
5
x
At rst glance, this seems like a very hard volume to comS

pute as the region R (shown in Figure 13.21(a)) has a hole in it, cu ng out a
strange por on of the surface, as shown in part (b) of the gure. However, by
describing R in terms of polar equa ons, the volume is not very dicult to compute. It is straigh orward to show that the circle (x 1)2 + y2 = 1 has polar
equa on r = 2 cos , and that the circle (x 2)2 + y2 = 4 has polar equa on
r = 4 cos . Each of these circles is traced out on the interval 0 . The
bounds on r are 2 cos r 4 cos .
Replacing x with r cos in the integrand,along with replacing y with r sin ,
prepares us to evaluate the double integral R f(x, y) dA:
4 cos
(
)2 (
)2 )
4 r cos 2 r sin r dr d
)
r3 + 4r2 cos dr d
f(x, y) dA =
2 cos
4 cos
=
0
2 cos
)4 cos

4
1
r4 + r3 cos
d
4
3
0
2 cos
]
([
1
4
4
4
=
(256 cos ) + (64 cos )
4
3
])
[ 0
4
1
4
4
d
(16 cos ) + (8 cos )
4
3

44
=
cos4 d.
3
0
.
(b)
Figure 13.21: Showing the region R and

surface used in Example 454.
To integrate cos4 , rewrite it as cos2 cos2 and employ the power-reducing

formula twice:
cos4 = cos2 cos2
)1(
)
1(
= 1 + cos(2)
1 + cos(2)
2
2
)
1(
= 1 + 2 cos(2) + cos2 (2)
4
))
1(
1(
=
1 + 2 cos(2) + 1 + cos(4)
4
2
3 1
1
= + cos(2) + cos(4).
8 2
8
Notes:
748

Picking up from where we le o above, we have

44
=
cos4 d
3
0
(
)

44 3 1
1
=
+ cos(2) + cos(4) d
3 8 2
8
0
(
)

44 3
1
1
=
+ sin(2) +
sin(4)
3 8
4
32
0
11
=
17.279.
2
While this example was not trivial, the double integral would have been much
harder to evaluate had we used rectangular coordinates. .
z
. Example 455
Evalua ng a double integral with polar coordinates

1
Find the volume under the surface f(x, y) = 2
over the sector of the
x + y2 + 1
circle with radius a centered at the origin in the rst quadrant, as shown in Figure
13.22.
The region R we are integra ng over is a circle with radius
a, restricted to the rst quadrant. Thus, in polar, the bounds on R are 0 r a,
0 /2. The integrand is rewri en in polar as
0.5
1
1
1
2
= 2
.
2
2
2
2
2
x +y +1
r +1
r cos + r sin + 1
We nd the volume as follows:
f(x, y) dA =
R
/2
/2
4
2
.
2 a
.
a
r
dr d
2+1
r
0
)a
1(
ln |r2 + 1| d
2
0
Figure 13.22: The surface and region R

used in Example 455.
/2
1
ln(a2 + 1) d
2
0
(
)/2

1
2
=
ln(a + 1)
2
0
2
= ln(a + 1).
4
Note: Previous work has shown that

there is nite area under x2 1+1 over the
en re x-axis. However, Example 455
shows that there is innite volume under
1
over the en re x-y plane.
x2 +y2 +1
Figure 13.22 clearly shows that f shrinks to near 0 very quickly. Regardless, as a
grows, so does the volume, without bound. .
Notes:
749

. Example 456
Finding the volume of a sphere
Find the volume of a sphere with radius a.
The sphere of radius a,
centered at the origin, has equa on
S
x2 +y2 +z2 = a2 ; solving for z, we have z = a2 x2 y2 . This gives the upper
half of a sphere. We wish to nd the volume under this top half, then double it
to nd the total volume.
The region we need to integrate over is the circle of radius a, centered at the
origin. The polar bounds for this equa on are 0 r a, 0 2.
All together, the volume of a sphere with radius a is:

2 a
2
a2 x2 y2 dA = 2
a2 (r cos )2 (r sin )2 r dr d
R
=2
0
r a2 r2 dr d.
We can evaluate this inner integral with subs tu on. With u = a2 r2 , du =

2r dr. The new bounds of integra on are u(0) = a2 to u(a) = 0. Thus we
have:
2 0
(
)
=
u1/2 du d
a2
)0
2 3/2
=
u
2 d
3
0
a
(
)
2
2 3
=
a
d
3
0
(
)2
2 3
=
a
3
(a)
4
= a3 .
3
Generally, the formula for the volume of a sphere with radius r is given as 4/3r3 ;
we have jus ed this formula with our calcula on. .
. Example 457
.Finding the volume of a solid
A sculptor wants to make a solid bronze cast of the solid shown in Figure 13.23,
where the base of the solid has boundary, in polar coordinates, r = cos(3),
and the top is dened by the plane z = 1 x + 0.1y. Find the volume of the
solid.
From the outset, we should recognize that knowing how to

S
set up this problem is probably more important than knowing how to compute
(b)
.Figure 13.23:
Finding the volume of the

solid shown here from two perspec ves.
750
Notes:

the integrals. The iterated integral to come is not hard to evaluate, though it is
long, requiring lots of algebra. Once the proper iterated integral is determined,
one can use readilyavailable technology to help compute the nal answer.
The region R that we are integra ng over is bound by 0 r cos(3),
for 0 (note that this rose curve is traced out on the interval [0, ], not
[0, 2]). This gives us our bounds of integra on. The integrand is z = 1x+0.1y;
conver ng to polar, we have that the volume V is:
V=
cos(3)
f(x, y) dA =
0
(
)
1 r cos + 0.1r sin r dr d.
Distribu ng the r, the inner integral is easy to evaluate, leading to

)
(
1
1
0.1
cos2 (3) cos3 (3) cos +
cos3 (3) sin d.
2
3
3
0
This integral takes me to compute by hand; it is rather long and cumbersome.
The powers of cosine need to be reduced, and products like cos(3) cos need
to be turned to sums using the Product To Sum formulas in the back cover of
this text.
For instance, we rewrite 12 cos2 (3) as 41 (1 + cos(6)). We can also rewrite
1
3
3 cos (3) cos as:
)
1
1
1 1 + cos(6) (
cos(4)+cos(2) .
cos3 (3) cos = cos2 (3) cos(3) cos =
3
3
3
2
This last expression s ll needs simplica on, but eventually all terms can be reduced to the form a cos(m) or a sin(m) for various values of a and m.
We forgo the algebra and recommend the reader employ technology, such
as WolframAlpha, to compute the numeric answer. Such technology gives:
cos(3)
1 r cos + 0.1r sin r dr d = 0.785u3 .

4
Since the units were not specied, we leave the result as almost 0.8 cubic units
(meters, feet, etc.) Should the ar st want to scale the piece uniformly, so that
each rose petal had a length other than 1, she should keep in mind that scaling
by a factor of k scales the volume by a factor of k3 . .
We have used iterated integrals to nd areas of plane regions and volumes
under surfaces. Just as a single integral can be used to compute much more than
area under the curve, iterated integrals can be used to compute much more
than we have thus far seen. The next two sec ons show two, among many,
applica ons of iterated integrals.
Notes:
751
Exercises 13.3
Terms and Concepts
1. When evalua ng R f(x, y) dA using polar coordinates,

f(x, y) is replaced with
and dA is replaced with
.
2. Why would one be interested in evalua ng a double integral with polar coordinates?
Problems
4
4
3. f(x, y) = 3x y + 4; R is the region enclosed by the circle

x2 + y2 = 1.
4. f(x, y) = 4x + 4y; R is the region enclosed by the circle
x2 + y2 = 4.
16y2
2
8y2
13.
0
4x2
14.
)
2y x dx dy
)
x + y dx dy
2
2
0
4x2
In Exercises 3 10, a func on f(x, y) is given and

a region R of
the x-y plane is described. Set up and evaluate R f(x, y) dA
using polar coordinates.
12.
x + 5 dy dx +
1
4x2
(
)
x + 5 dy dx +
1x2
(
)
x + 5 dy dx
Hint: draw the region of each integral carefully and see how
they all connect.
In Exercises 15 16, special double integrals are presented
that are especially well suited for evalua on in polar coordinates.
2
2
15. Consider
e(x +y ) dA.
R
5. f(x, y) = 8 y; R is the region enclosed by the circles with

polar equa ons r = cos and r = 3 cos .
(a) Why is this integral dicult to evaluate in rectangular

coordinates, regardless of the region R?
6. f(x, y) = 4; R is the region enclosed by the petal of the rose

curve r = sin(2) in the rst quadrant.
(
7. f(x, y) = ln x2 + y2 ); R is the annulus enclosed by the circles x2 + y2 = 1 and x2 + y2 = 4.
(b) Let R be the region bounded by the circle of radius a

centered at the origin. Evaluate the double integral
using polar coordinates.
8. f(x, y) = 1 x2 y2 ; R is the region enclosed by the circle

x2 + y2 = 1.
(c) Take the limit of your answer from (b), as a .

What does this imply about the volume under the
2
2
surface of e(x +y ) ?
10. f(x, y) = (x y)/(x + y); R is the region enclosed by the

lines y = x, y = 0 and the circle x2 + y2 = 1 in the rst
quadrant.
16. The surface of a right circular cone with height h and

base radius
a can be described by the equa on f(x, y) =
x2
y2
hh
+ 2 , where the p of the cone lies at (0, 0, h)
a2
a
and the circular base lies in the x-y plane, centered at the
origin.
In Exercises 11 14, an iterated integral in rectangular coordinates is given. Rewrite the integral using polar coordinates.
5 25x2
11.
x2 + y2 dy dx
Conrm that the volume of a right circular cone with

1
height h and base radius a is V = a2 h by evalua ng
3
f(x, y) dA in polar coordinates.
9. f(x, y) = x2 y2 ; R is the region enclosed by the circle

x2 + y2 = 36 in the rst and fourth quadrants.
25x2
13.4 Center of Mass
13.4 Center of Mass

We have used iterated integrals to nd areas of plane regions and signed volumes under surfaces. A brief recap of these uses will be useful in this sec on as
we apply iterated integrals to compute the mass and center of mass of planar
regions.
To nd the area of a planar region, we evaluated the double integral R dA.

That is, summing up the areas
of lots of li le subregions of R gave us the total
area. Informally, we think of R dA as meaning sum up lots of li le areas over
R.
To nd the signed volume under a surface, we evaluated the double integral
f(x, y) dA. Recall that the dA is not just a bookend at the end of an inR
tegral; rather, it is mul plied by f(x, y). We regard f(x, y) as giving
a height, and
dA s ll giving an area: f(x, y) dA gives a volume. Thus, informally, R f(x, y) dA
means sum up lots of li le volumes over R.
We now extend these ideas to other contexts.
.
(a)
Mass and Weight
Consider a thin sheet of material with constant thickness and nite area.
Mathema cians (and physicists and engineers) call such a sheet a lamina. So
consider a lamina, as shown in Figure 13.24(a), with the shape of some planar
region R, as shown in part (b).
We
can write a simple double integral that represents the mass of the lamina: R dm, where dm means a li le mass. That is, the double integral
states the total mass of the lamina can be found by summing up lots of li le
masses over R.
To evaluate this double integral, par on R into n subregions as we have
done in the past. The i th subregion has area Ai . A fundamental property of
mass is that mass=densityarea. If the lamina has a constant density , then
the mass of this i th subregion is mi = Ai . That is, we can compute a small
amount of mass by mul plying a small amount of area by the density.
If density is variable, with density func on = (x, y), then we can approximate the mass of the i th subregion of R by mul plying Ai by (xi , yi ), where
(xi , yi ) is a point in that subregion. That is, for a small enough subregion of R,
the density across that region is almost constant.
The total mass M of the lamina is approximately the sum of approximate
masses of subregions:
M
i=1
mi =
i=1
(xi , yi )Ai .
y = f2 (x)
1
y=
)
f 1(x
x
1
(b)
Figure 13.24: Illustra ng the concept of a

lamina.
Note: Mass and weight are dierent

measures. Since they are scalar mul ples of each other, it is o en easy to
treat them as the same measure. In this
sec on we eec vely treat them as the
same, as our technique for nding mass is
the same as for nding weight. The density func ons used will simply have dierent units.
Notes:
753

Taking the limit as the size of the subregions shrinks to 0 gives us the actual
mass; that is, integra ng (x, y) dA over R gives the mass of the lamina.
.
Deni on 103
Mass of a Lamina with Vairable Density
Let (x, y) be a con nuous density func on of a lamina corresponding to

a plane region R. The mass M of the lamina is
mass M =
dm =
(x, y) dA.
. Example 458
Finding the mass of a lamina with constant density
Find the mass of a square lamina, with side length 1, with a density of =
3gm/cm2 .
y
We represent the lamina with a square region in the plane

S
as shown in Figure 13.25. As the density is constant, it does not ma er where
we place the square.
Following Deni on 103, the mass M of the lamina is

0.5
M=
.
0.5
3 dA =
R
3 dx dy = 3
0
dx dy = 3gm.
0
Figure 13.25: A region R represen ng a

lamina in Example 458.
This is all very straigh orward; note that all we really did was nd the area
of the lamina and mul ply it by the constant density of 3gm/cm2 . .
. Example 459
.Finding the mass of a lamina with variable density
Find the mass of a square lamina, represented by the unit square with lower
le hand corner at the origin (see Figure 13.25), with variable density (x, y) =
(x + y + 2)gm/cm2 .
The variable density , in this example, is very uniform, givS
ing a density of 3 in the center of the square and changing linearly. A graph
of (x, y) can be seen in Figure 13.26; no ce how same amount of density is
above z = 3 as below. Well comment on the signicance of this momentarily.
The mass M is found by integra ng (x, y) over R. The order of integra on
Notes:
754
13.4 Center of Mass

z
is not important; we choose dx dy arbitrarily. Thus:
M=
(x + y + 2) dA =
R
(x + y + 2) dx dy
0
)1

1 2
=
x + x(y + 2) dy
2
0
0
)
1(
5
+ y dy
=
2
0
(
)1

5
1
=
y + y2
2
2
= 3gm.
It turns out that since since the density of the lamina is so uniformly distributed
above and below z = 3 that the mass of the lamina is the same as if it had a
constant density of 3. .
z=
2
y+
z=x+
2
y
0.5
0.5
Figure 13.26: Graphing the density funcon in Example 459.
. Example 460
.Finding the weight of a lamina with variable density
Find the weight of the lamina represented by the circle with radius 2 , centered
2
at the origin, with density func on (x, y) = (x2 + y2 + 1)lb/
. Compare this
to the weight of the same lamina with density (x, y) = (2 x2 + y2 + 1)lb/ 2 .
S
A direct applica on of Deni on 103 states that the weight
of the lamina is R (x, y) dA. Since our lamina is in the shape of a circle, it
makes sense to approach the double integral using polar coordinates.
The density func on (x, y) = x2 + y2 + 1 becomes (r, ) = (r cos )2 +
(r sin )2 + 1 = r2 + 1. The circle is bounded by 0 r 2 and 0 2.
Thus the weight W is:
(r2 + 1)r dr d
W=
0
=
0
)2
1 4 1 2
r + r d
4
2
0
(6) d
=
0
= 12 37.70lb.
2
2
Now compare this with the density func on (x,
y) = 2 x + y + 1. Con2
ver ng this to polar coordinates gives (r, ) = 2 (r cos ) + (r sin )2 + 1 =
Notes:
755

2r + 1. Thus the weight W is:
W=
(2r + 1)r dr d
2
1 2
( r3 + r2 ) d
3
2
0
0
2 ( )
22
=
d
3
0
44
46.08lb.
=
3
One would expect dierent density func ons to return dierent weights, as we
have here. The density func ons were chosen, though, to be similar: each gives
a density of 1 at the origin and a density of 5 at the outside edge of the circle,
as seen in Figure 13.27.
2
.
.
(a)
.
.
(b)
Figure 13.27: Graphing the density func ons

in Example 460. In (a) is the density func on
(x, y) = x2 + y2 + 1; in (b) is (x, y) = 2 x2 + y2 + 1.
No ce how x2 + y2 + 1 2 x2 + y2 + 1 over the circle; this results in less

weight. .
Plo ng the density func ons can be useful as our understanding of mass
can be
related to our understanding of volume under a surface. We interpreted R f(x, y) dA as giving the volume under f over R; we can understand
(x, y) dA in the same way. The volume under over R is actually mass;
R
Notes:
756
13.4 Center of Mass

by compressing the volume under onto the x-y plane, we get more mass
in some areas than others i.e., areas of greater density.
Knowing the mass of a lamina is one of several important measures. Another
is the center of mass, which we discuss next.
Center of Mass
Consider a disk of radius 1 with uniform density. It is common knowledge
that the disk will balance on a point if the point is placed at the center of the
disk. What if the disk does not have a uniform density? Through trial-and-error,
we should s ll be able to nd a spot on the disk at which the disk will balance
on a point. This balance point is referred to as the center of mass, or center of
gravity. It is though all the mass is centered there. In fact, if the disk has a
mass of 3kg, the disk will behave physically as though it were a point-mass of
3kg located at its center of mass. For instance, the disk will naturally spin with
an axis through its center of mass (which is why it is important to balance the
res of your car: if they are out of balance, their center of mass will be outside
of the axle and it will shake terribly).
We nd the center of mass based on the principle of a weighted average.
Consider a college class in which your homework average is 90%, your test average is 73%, and your nal exam grade is an 85%. Experience tells us that our
nal grade is not the average of these three grades: that is, it is not:
0.9 + 0.73 + 0.85
0.837 = 83.7%.
3
That is, you are probably not pulling a B in the course. Rather, your grades are
weighted. Lets say the homework is worth 10% of the grade, tests are 60% and
the exam is 30%. Then your nal grade is:
(0.1)(0.9) + (0.6)(0.73) + (0.3)(0.85) = 0.783 = 78.3%.
Each grade is mul plied by a weight.
In general, given values x1 , x2 , . . . , xn and weights w1 , w2 , . . . , wn , the weighted
average of the n values is
/ n
n
wi .
wi xi
i=1
i=1
In the grading example above, the sum of the weights 0.1, 0.6 and 0.3 is 1,
so we dont see the division by the sum of weights in that instance.
How this relates to center of mass is given in the following theorem.
Notes:
757

.
Theorem 121
Center of Mass of Discrete Linear System
Let point masses m1 , m2 , . . . , mn be distributed along the x-axis at locaons x1 , x2 , . . . , xn , respec vely. The center of mass x of the system is
located at
/ n
n
x=
mi xi
mi .
i=1
. Example 461
i=1
Finding the center of mass of a discrete linear system
1. Point masses of 2gm are located at x = 1, x = 2 and x = 3 are connected by a thin rod of negligible weight. Find the center of mass of the
system.
2. Point masses of 10gm, 2gm and 1gm are located at x = 1, x = 2 and
x = 3, respec vely, are connected by a thin rod of negligible weight. Find
the center of mass of the system.
S
x
1
1. Following Theorem 121, we compute the center of mass as:
x
2
x=
(a)
4
2(1) + 2(2) + 2(3)
= = 1.3.
2+2+2
3
So the system would balance on a point placed at x = 4/3, as illustrated

in Figure 13.28(a).
.
x
1
2. Again following Theorem 121, we nd:
x
0
x=
(b)
Figure 13.28: Illustra ng point masses

along a thin rod and the center of mass.
10(1) + 2(2) + 1(3)

3
=
0.23.
10 + 2 + 1
13
Placing a large weight at the le hand side of the system moves the center
of mass le , as shown in Figure 13.28(b).
In a discrete system (i.e., mass is located at individual points, not along a

con nuum) we nd the center of mass by dividing the mass into a moment of
the system. In general, a moment is a weighted measure of distance from a parcular point or line. In the case described by Theorem 121, we are nding a
weighted measure of distances from the y-axis, so we refer to this as the moment about the y-axis, represented by My . Le ng M be the total mass of the
system, we have x = My /M.
Notes:
758
13.4 Center of Mass

We can extend the concept of the center of mass of discrete points along a
line to the center of mass of discrete points in the plane rather easily. To do so,
we dene some terms then give a theorem.
Deni on 104
Moments about the x- and y- Axes.
Let point masses m1 , m2 , . . . , mn be located at points (x1 , y1 ),

(x2 , y2 ) . . . , (xn , yn ), respec vely, in the x-y plane.
1. The moment about the y-axis, My , is My =
mi xi .
i=1
2. The moment about the x-axis, Mx , is Mx =
mi yi .
i=1
One can think that these deni ons are backwards as My sums up x distances. But remember, x distances are measurements of distance from the
y-axis, hence dening the moment about the y-axis.
We now dene the center of mass of discrete points in the plane.
Theorem 122
Center of Mass of Discrete Planar System
Let point masses m1 , m2 , . . . , mn be located at points (x1 , y1 ),

n
(x2 , y2 ) . . . , (xn , yn ), respec vely, in the x-y plane, and let M =

mi .
i=1
The center of mass of the system is at (x, y), where

x=
My
M
and
y=
Mx
.
M
. Example 462
.Finding the center of mass of a discrete planar system
Let point masses of 1kg, 2kg and 5kg be located at points (2, 0), (1, 1) and (3, 1),
respec vely, and are connected by thin rods of negligible weight. Find the center
of mass of the system.
We follow Theorem 122 and Deni on 104 to nd M, Mx
S
and My :
M = 1 + 2 + 5 = 8kg.
Notes:
1
(x, y)
x
1
.
Figure 13.29: Illustra ng the center of
mass of a discrete planar system in Example 462.
759
Mx =
mi yi
My =
i=1
mi xi
i=1
= 1(0) + 2(1) + 5(1)

= 7.
= 1(2) + 2(1) + 5(3)

= 19.
(
Thus the center of mass is (x, y) =
My Mx
,
M M
(
=
19 7
,
8 8
)
= (2.375, 0.875),
illustrated in Figure 13.29. .

We nally arrive at our true goal of this sec on: nding the center of mass of
a lamina with variable density. While the above measurement of center of mass
is interes ng, it does not directly answer more realis c situa ons where we need
to nd the center of mass of a con guous region. However, understanding the
discrete case allows us to approximate the center of mass of a planar lamina;
using calculus, we can rene the approxima on to an exact value.
We begin by represen ng a planar lamina with a region R in the x-y plane
with density func on (x, y). Par on R into n subdivisions, each with area
Ai . As done before, we can approximate the mass of the i th subregion with
(xi , yi )Ai , where (xi , yi ) is a point inside the i th subregion. We can approximate the moment of this subregion about the y-axis with xi (xi , yi )Ai that is,
by mul plying the approximate mass of the region by its approximate distance
from the y-axis. Similarly, we can approximate the moment about the x-axis
with yi (xi , yi )Ai . By summing over all subregions, we have:
mass: M
moment about the x-axis: Mx
moment about the y-axis: My
i=1
n
i=1
n
(xi , yi )Ai
(as seen before)
yi (xi , yi )Ai
xi (xi , yi )Ai
i=1
By taking limits, where size of each subregion shrinks to 0 in both the x and
y direc ons, we arrive at the double integrals given in the following theorem.
Notes:
760
13.4 Center of Mass
Theorem 123
Center of Mass of a Planar Lamina, Moments
Let a planar lamina be represented by a region R in the x-y plane with

density func on (x, y).
1. mass: M =
(x, y) dA
R
2. moment about the x-axis: Mx =
y(x, y) dA
R
3. moment about the y-axis: My =
x(x, y) dA
R
4. The center of mass of the lamina is

(
)
My Mx
(x, y) =
,
.
M M
We start our prac ce of nding centers of mass by revisi ng some of the

lamina used previously in this sec on when nding mass. We will mostly just
set up the integrals needed to compute M, Mx and My and leave the details of
the integra on to the reader.
y
. Example 463
.Finding the center of mass of a lamina
Find the center mass of a square lamina, with side length 1, with a density of
= 3gm/cm2 . (Note: this is the lamina from Example 458.)
We represent the lamina with a square region in the plane
as shown in Figure 13.30 as done previously.
Following Theorem 123, we nd M, Mx and My :
S
M=
3 dA =
Mx =
3 dx dy = 3gm.
0
3y dA =
My =
3y dx dy = 3/2 = 1.5.
0
1
x
0.5
Figure 13.30: A region R represen ng a

lamina in Example 458.
3x dx dy = 3/2 = 1.5.
0
(
0.5
3x dA =
R
My Mx
,
M M
)
= (1.5/3, 1.5/3) = (0.5, 0.5).
Notes:
761

This is what we should have expected: the center of mass of a square with constant density is the center of the square. .
. Example 464
Finding the center of mass of a lamina
Find the center of mass of a square lamina, represented by the unit square
with lower le hand corner at the origin (see Figure 13.30), with variable density (x, y) = (x + y + 2)gm/cm2 . (Note: this is the lamina from Example 459.)
We follow Theorem 123, to nd M, Mx and My :
(x + y + 2) dA =
M=
Mx =
(x + y + 2) dx dy = 3gm.
0
y(x + y + 2) dA =
My =
0
1
19
.
12
x(x + y + 2) dx dy =
19
.
12
x(x + y + 2) dA =
R
y(x + y + 2) dx dy =
1
) (
)
My Mx
19 19
,
=
,
(0.528, 0.528).
M M
36 36
While the mass of this lamina is the same as the lamina in the previous example,
the greater density found with greater x and y values pulls the center of mass
from the center slightly towards the upper righthand corner. .
. Example 465
.Finding the center of mass of a lamina
Find the center of mass of the lamina represented by the circle with radius 2 ,
centered at the origin, with density func on (x, y) = (x2 + y2 + 1)lb/ 2 . (Note:
this is one of the lamina used in Example 460.)
As done in Example 460, it is best to describe R using polar
S
2
coordinates. Thus when
x(x,
( we)compute My , we will(integrate
)( not
) y) = x(x +
2
2
y + 1), but rather r cos (r cos , r sin ) = r cos r + 1 . We compute
M, Mx and My :
M=
0
(r sin )(r2 + 1)r dr d = 0.
Mx =
0
762
(r cos )(r2 + 1)r dr d = 0.
My =
Notes:
(r2 + 1)r dr d = 12 37.7lb.
13.4 Center of Mass

Since R and the density of R are both symmetric about the x and y axes, it should
come as no big surprise that the moments about each axis is 0. Thus the center
of mass is (x, y) = (0, 0). .
. Example 466
Finding the center of mass of a lamina
Find the center of mass of the lamina represented by the region R shown in Figure 13.31, half an annulus with outer radius 6 and inner radius 5, with constant
density 2lb/ 2 .
Once again it will be useful to represent R in polar coorS

dinates. Using the descrip on of R and/or the illustra on, we see that R is
bounded by 1 r 2 and 0 . As the lamina is symmetric about
the y-axis, we should expect My = 0. We compute M, Mx and My :
(2)r dr d = 11lb.
0
(r sin )(2)r dr d =
0
My =
Figure 13.31: Illustra ng the region R in

Example 466.
Mx =
M=
(x, y)
364
121.33.
3
(r cos )(2)r dr d = 0.
0
( 364 )
Thus the center of mass is (x, y) = 0, 33
(0, 3.51). The center of mass is
indicated in Figure 13.31; note how it lies outside of R! .
This sec on has shown us another use for iterated integrals beyond nding
area or signed volume under the curve. While there are many uses for iterated
integrals, we give one more applica on in the following sec on: compu ng surface area.
Notes:
763
Exercises 13.4
Terms and Concepts
18. R is the annulus in the rst and second quadrants

bounded
by x2 + y2 = 9 and x2 + y2 = 36; (x, y) = x2 + y2 lb/ 2
1. Why is it easy to use mass and weight interchangeably,

even though they are dierent measures?
2. Given a point (x, y), the value of x is a measure of distance
from the
-axis.
3. We can think of R dm as meaning sum up lots of

.
4. What is a discrete planar system?
5. Why does Mx use R y(x, y) dA instead of R x(x, y) dA;

that is, why do we use y and not x?
6. Describe a situa on where the center of mass of a lamina
does not lie within the region of the lamina itself.
Problems
In Exercises 19 26, nd the center of mass of the lamina described by the region R in the plane and its density func on
(x, y).
Note: these are the same lamina as in Exercises 11 to 18.
19. R is the rectangle with corners (1, 3), (1, 2), (7, 2) and
(7, 3); (x, y) = 5gm/cm2
(7, 3); (x, y) = (x + y2 )gm/cm2
21. R is the triangle with corners (1, 0), (1, 0), and (0, 1);
(x, y) = 2lb/in2
(x, y) = (x2 + y2 + 1)lb/in2
23. R is the circle centered at the origin with radius 2; (x, y) =
(x + y + 4)kg/m2
In Exercises 7 10, point masses are given along a line or in

the plane. Find the center of mass x or (x, y), as appropriate.
(All masses are in grams and distances are in cm.)
24. R is the circle sector bounded

by x2 + y2 = 25 in the rst
2
2
quadrant; (x, y) = ( x + y + 1)kg/m2
7. m1 = 4 at x = 1;
25. R is the annulus in the rst and second quadrants bounded

by x2 + y2 = 9 and x2 + y2 = 36; (x, y) = 4lb/ 2
m2 = 3 at x = 3;
m3 = 5 at x = 10
8. m1 = 2 at x = 3; m2 = 2 at x = 1;
m3 = 3 at x = 0; m4 = 3 at x = 7
9. m1 = 2 at (2, 2);
m3 = 20 at (0, 4)
m2 = 2 at (2, 2);
10. m1 = 1 at (1, 1); m2 = 2 at (1, 1);

m3 = 2 at (1, 1); m4 = 1 at (1, 1)
In Exercises 11 18, nd the mass/weight of the lamina described by the region R in the plane and its density func on
(x, y).
(7, 3); (x, y) = 5gm/cm2
(7, 3); (x, y) = (x + y2 )gm/cm2
26. R is the annulus in the rst and second quadrants

bounded
by x2 + y2 = 9 and x2 + y2 = 36; (x, y) = x2 + y2 lb/ 2

The moment of iner a I is a measure of the tendency of a lamina to resist rota ng about an axis or con nue to rotate about
an axis. Ix is the moment of iner a about the x-axis, Ix is the
moment of iner a about the x-axis, and IO is the moment of
iner a about the origin. These are computed as follows:
Ix =
y2 dm
R
Iy =
x2 dm
R
( 2
)
IO =
x + y2 dm
R
(x, y) = 2lb/in2
In Exercises 27 30, a lamina corresponding to a planar region R is given with a mass of 16 units. For each, compute Ix ,
Iy and IO .
(x, y) = (x2 + y2 + 1)lb/in2
27. R is the 4 4 square with corners at (2, 2) and (2, 2)

with density (x, y) = 1.
15. R is the circle centered at the origin with radius 2; (x, y) =

(x + y + 4)kg/m2
28. R is the 8 2 rectangle with corners at (4, 1) and (4, 1)

16. R is the circle sector bounded

by x2 + y2 = 25 in the rst
quadrant; (x, y) = ( x2 + y2 + 1)kg/m2
29. R is the 4 2 rectangle with corners at (2, 1) and (2, 1)

17. R is the annulus in the rst and second quadrants bounded

by x2 + y2 = 9 and x2 + y2 = 36; (x, y) = 4lb/ 2
30. R is the circle with radius 2 centered at the origin with density (x, y) = 4/.
13.5 Surface Area
13.5 Surface Area

In Sec on 7.4 we used denite integrals to compute the arc length of plane
curves of the form y = f(x). We later extended these ideas to compute the
arc length of plane curves dened by parametric or polar equa ons.
The natural extension of the concept of arc length over an interval to surfaces is surface area over a region.
Consider the surface z = f(x, y) over a region R in the x-y plane, shown in
Figure 13.32(a). Because of the domed shape of the surface, the surface area will
be greater than that of the area of the region R. We can nd this area using the
same basic technique we have used over and over: well make an approxima on,
then using limits, well rene the approxima on to the exact value.
As done to nd the volume under a surface or the mass of a lamina, we
subdivide R into n subregions. Here we subdivide R into rectangles, as shown in
the gure. One such subregion is outlined in the gure, where the rectangle has
dimensions xi and yi , along with its corresponding region on the surface.
In part (b) of the gure, we zoom in on this por on of the surface. When xi
and yi are small, the func on is approximated well by the tangent plane at any
point (xi , yi ) in this subregion, which is graphed in part (b). In fact, the tangent
plane approximates the func on so well that in this gure, it is virtually indisnguishable from the surface itself! Therefore we can approximate the surface
area Si of this region of the surface with the area Ti of the corresponding por on
of the tangent plane.
This por on of the tangent plane is a parallelogram, dened by sides u and
v, as shown. One of the applica ons of the cross product from Sec on 10.4 is
that the area of this parallelogram is || u v ||. Once we can determine u and v,
we can determine the area.
u is tangent to the surface in the direc on of x, therefore, from Sec on 12.7,
u is parallel to 1, 0, fx (xi , yi ). The x-displacement of u is xi , so we know that
u = xi 1, 0, fx (xi , yi ). Similar logic shows that v = yi 0, 1, fy (xi , yi ). Thus:
z
2
.
1
1y
(a)
xi
surface area Si area of Ti

= || u v ||

= xi 1, 0, fx (xi , yi ) yi 0, 1, fy (xi , yi )
= 1 + fx (xi , yi )2 + fy (xi , yi )2 xi yi .
yi
.
(b)
Figure 13.32: Developing a method of
compu ng surface area.
Note that xi yi = Ai , the area of the i th subregion.

Summing up all n of the approxima ons to the surface area gives
n
surface area over R

1 + fx (xi , yi )2 + fy (xi , yi )2 Ai .
i=1
Notes:
765

Once again take a limit as all of the xi and yi shrink to 0; this leads to a
double integral.
.
Note:
as done before, we think of
R dS as meaning sum up lots of

li le surface areas.
The concept of surface area is dened
here, for while we already have a no on
of the area of a region in the plane, we
did not yet have a solid grasp of what the
area of a surface in space means.
Deni on 105
Surface Area
Let z = f(x, y) where fx and fy are con nuous over a closed, bounded
region R. The surface area S over R is
S=
dS
R
=
1 + fx (x, y)2 + fy (x, y)2 dA.
We test this deni on by using it to compute surface areas of known surfaces. We start with a triangle.
. Example 467
Finding the surface area of a plane over a triangle
Let f(x, y) = 4 x 2y, and let R be the region in the plane bounded by x = 0,
y = 0 and y = 2 x/2, as shown in Figure 13.33. Find the surface area of f over
R.
z
We follow Deni on 105. We start by no ng that fx (x, y) =

S
1 and fy (x, y) = 2. To dene R, we use bounds 0 y 2 x/2 and
0 x 4. Therefore
S=
dS
R
4 2x/2
Figure 13.33: Finding the area of a triangle in space in Example 467.
(
x)
6 2
dx
2
Because the surface is a triangle, we can gure out the area using geometry.
Considering the base of the
triangle to be the side in the x-y plane, we nd the
length of the base to be 20. We can nd the height using our knowledge of
vectors: let u be the side in the x-z plane
and let v be the side in the x-y plane.
The height is then || u proj v u || = 4 6/5. Geometry states that the area is
thus
1
4 6/5 20 = 4 6.
2
We arm the validity of our formula. .
Notes:
766
= 4 6.
x 4
1 + (1)2 + (2)2 dy dx
13.5 Surface Area

It is common knowledge that the surface area of a sphere of radius r is
4r2 . We conrm this in the following example, which involves using our formula with polar coordinates.
. Example 468
The surface area of a sphere.
Find the surface area of the sphere withradius a centered at the origin, whose
top hemisphere has equa on f(x, y) = a2 x2 y2 .
We start by compu ng par al deriva ves and nd
fx (x, y) =
x
a2
x2
y2
and
y
fy (x, y) =
.
2
a x2 y2
As our func on f only denes the top upper hemisphere of the sphere, we double our surface area result to get the total area:

S=2
1 + fx (x, y)2 + fy (x, y)2 dA
R

x2 + y2
=2
1+ 2
dA.
a x2 y2
R
The region R that we are integra ng over is the circle, centered at the origin,
with radius a: x2 + y2 = a2 . Because of this region, we are likely to have greater
success with our integra on by conver ng to polar coordinates. Using the subs tu ons x = r cos , y = r sin , dA = r dr d and bounds 0 2 and
0 r a, we have:
2 a
r2 cos2 + r2 sin2
r dr d
S=2
1+ 2
a r2 cos2 r2 sin2
0
0
2 a
r2
=2
r 1+ 2
dr d
a r2
0
0
2 a
a2
=2
r
dr d.
(13.1)
2
a r2
0
0
Apply subs tu on u = a2 r2 and integrate the inner integral, giving
2
2
=2
a d
0
= 4a2 .
Our work conrms our previous formula. .
Note: The inner integral in Equa on

(13.1) is an improper
integral, as the
a
a2
r
integrand of
dr d is
2
a r2
0
not dened at r = a. To properly
evaluate this integral, one must use the
techniques of Sec on 6.8.
The reason this need
arises is that the
func on f(x, y) = a2 x2 y2 fails the
requirements of Deni on 105, as fx and
fy are not con nuous on the boundary of
the circle x2 + y2 = a2 .
The computa on of the surface area is
s ll valid. The deni on makes stronger
requirements than necessary in part to
avoid the use of improper integra on, as
when fx and/or fy are not con nuous, the
resul ng improper integral may not converge. Since the improper integral does
converge in this example, the surface area
is accurately computed.
Notes:
767
. Example 469
Finding the surface area of a cone
The general formula for a right cone with height h and base radius a is
h 2
f(x, y) = h
x + y2 ,
a
shown in Figure 13.34. Find the surface area of this cone.
xh
fx (x, y) =
a x2 + y2
yh

.
a x2 + y2
This may look in mida ng at rst, but there are lots of simple simplica ons to
be done. It amazingly reduces to just
h2
1 2
1+ 2 =
a + h2 .
a
a
Our polar bounds are 0 2 and 0 r a. Thus
2 a
1
S=
r
a2 + h2 dr d
a
0
0
)a
2 (

1 2 1 2
2
=
r
a + h d
2
a
0
0
2
1
=
a a2 + h2 d
2
0
= a a2 + h2 .
Note: Note that once again fx and fy are

not con nuous on the domain of f, as
both are undened at (0, 0). (A similar
problem occurred in the previous example.) Once again the resul ng improper
integral converges and the computa on
of the surface area is valid.
This matches the formula found in the back of this text. .
. Example 470
.Finding surface area over a region
Find the area of the surface f(x, y) = x2 3y + 3 over the region R bounded by
x y x, 0 x 4, as pictured in Figure 13.35.
20
4
.
2
2
4
It is straigh orward to compute fx (x, y) = 2x and fy (x, y) =

S
3. Thus the surface area is described by the double integral

1 + (2x)2 + (3)2 dA =
10 + 4x2 dA.
R
x 4
Notes:
Figure 13.35: Graphing the surface in Example 470.
768
and
Since we are integra ng over the circle x2 + y2 = a2 , we again use polar

coordinates. Using the standard subs tu ons, our integrand becomes
(
)2 (
)2
hr cos
hr sin
1+
+
.
a r2
a r2
Figure 13.34: Finding the surface area of

a cone in Example 469.
We begin by compu ng par al deriva ves.
13.5 Surface Area

As with integrals describing arc length, double integrals describing surface area
are in general hard to evaluate directly because of the squareroot. This par cular integral can be easily evaluated, though, with judicious choice of our order
of integra on.
Integra ng with order dx dy requires us to evaluate

10 + 4x2 dx. This can
1
be done, though it involves Integra
on By Parts and sinh x. Integra ng with
order dy dxhas as its rst integral
10 + 4x2 dy, which is easy to evaluate: it
2
is simply y 10 + 4x + C. So we proceed with the order dy dx; the bounds are
already given in the statement of the problem.
4 x

10 + 4x2 dA =
10 + 4x2 dy dx
R
(
)x
y 10 + 4x2 dx
(
)
2x 10 + 4x2 dx.
Apply subs tu on with u = 10 + 4x2 :

(
)4

)
1(
2 3/2
=
10 + 4x

6
0
)
1(
= 37 74 5 10 100.825u2 .
3
So while the region R over which we integrate has an area of 16u2 , the surface
has a much greater area as its z-values change drama cally over R. .
In prac ce, technology helps greatly in the evalua on of such integrals. High
powered computer algebra systems can compute integrals that are dicult, or
at least me consuming, by hand, and can at the least produce very accurate
approxima ons with numerical methods. In general, just knowing how to set up
the proper integrals brings one very close to being able to compute the needed
value. Most of the work is actually done in just describing the region R in terms
of polar or rectangular coordinates. Once this is done, technology can usually
provide a good answer.
Notes:
769
Exercises 13.5
Terms and Concepts
8. f(x, y) =
1
;
x2 + y2 + 1
1. Surface area is analogous to what previously studied concept?
2. To approximate the area of a small por on of a surface, we

plane.
computed the area of its
3. We interpret
dS as sum up lots of li le
R is the circle x2 + y2 = 9.
4. Why is it important to know how to set up a double integral to compute surface area, even if the resul ng integral
is hard to evaluate?
5. Why do z = f(x, y) and z = g(x, y) = f(x, y) + h, for some

real number h, have the same surface area over a region
R?
x4
4 y
.
9. f(x, y) = x2 y2 ; R is the rectangle with opposite corners
(1, 1) and (1, 1).
6. Let z = f(x, y) and z = g(x, y) = 2f(x, y). Why is the surface area of g over a region R not twice the surface area of
f over R?
1
x
Problems
1
1
y
In Exercises 7 10, set up the iterated integral that computes

the surface area of the given surface over the region R.
.
7. f(x, y) = sin x cos y; R is the rectangle with bounds 0
x 2, 0 y 2.
z
1
1
; R is the rectangle bounded by
ex2 + 1
5 x 5 and 0 y 1.
10. f(x, y) =
1
1
5
1
5
x
x 5
1
y
In Exercises 11 19, nd the area of the given surface over

the region R.
11. f(x, y) = 3x 7y + 2; R is the rectangle with opposite corners (1, 0) and (1, 3).
12. f(x, y) = 2x + 2y + 2; R is the triangle with corners (0, 0),
(1, 0) and (0, 1).
13. f(x, y) = x2 + y2 + 10; R is the circle x2 + y2 = 16.
14. f(x, y) = 2x + 4y2 + 7 over R, the triangle bounded by
y = x, y = x, 0 y 1.
15. f(x, y) = x2 + y over R, the triangle bounded by y = 2x,
y = 0 and x = 2.
16. f(x, y) = 23 x3/2 + 2y3/2 over R, the rectangle with opposite
corners (0, 0) and (1, 1).
17. f(x, y) = 10 2 x2 + y2 over R, the circle x2 + y2 = 25.

(This is the cone with height 10 and base radius 5; be sure
to compare you result with the known formula.)
18. Find the surface area of the spherewith radius 5 by doubling the surface area of f(x, y) = 25 x2 y2 over R,
the circle x2 + y2 = 25. (Be sure to compare you result
with the known formula.)
19. Find the surface area of the ellipse formed by restric ng
the plane f(x, y) = cx + dy + h to the region R, the circle
x2 + y2 = 1, where c, d and h are some constants. Your
answer should be given in terms of c and d; why does the
value of h not ma er?
13.6 Volume Between Surfaces and Triple Integra on

We learned in Sec on 13.2 how
to compute the signed volume V under a surface
z = f(x, y) over a region R: V = R f(x, y) dA. It follows naturally that if f(x, y)
g(x, y) on R, then the volume between f(x, y) and g(x, y) on R is
(
)
V=
f(x, y) dA
g(x, y) dA =
f(x, y) g(x, y) dA.
R
Theorem 124
Volume Between Surfaces
Let f and g be con nuous func ons on a closed, bounded region R, where
f(x, y) g(x, y) for all (x, y) in R. The volume V between f and g over R
is
(
)
V=
f(x, y) g(x, y) dA.
z
y
. Example 471
Finding volume between surfaces
Find the volume of the space region bounded by the planes z = 3x + y 4 and
z = 8 3x 2y in the 1st octant. In Figure 13.36(a) the planes are drawn; in (b),
only the dened region is given.
.5
We need to determine the region R over which we will inteS

grate. To do so, we need to determine where the planes intersect. They have
common z-values when 3x + y 4 = 8 3x 2y. Applying a li le algebra, we
have:
3x + y 4 = 8 3x 2y
(a)
6x + 3y = 12
2x + y = 4
z
y
4
2
The planes intersect along the line 2x+y = 4. Therefore the region R is bounded
by x = 0, y = 0, and y = 4 2x; we can convert these bounds to integra on
bounds of 0 x 2, 0 y 4 2x. Thus
(
)
V=
8 3x 2y (3x + y 4) dA
=
0
(b)
)
12 6x 3y dy dx
= 16u3 .
Figure 13.36: Finding the volume between the planes given in Example 13.36.
772
R
2 42x
The volume between the surfaces is 16 cubic units. .
Notes:

In the preceding example, we found the volume by evalua ng the integral
2 42x
(
)
8 3x 2y (3x + y 4) dy dx.
0
Note how we can rewrite the integrand as an integral, much as we did in Sec on
13.1:
8 3x 2y (3x + y 4) =
83x2y
dz.
3x+y4
Thus we can rewrite the double integral that nds volume as

2 42x
2 42x ( 83x2y )
(
)
83x2y(3x+y4) dy dx =
dz dy dx.
0
3x+y4
This no longer looks like a double integral, but more like a triple integral.
Just as our rst introduc on to double integrals was in the context of nding the
area of a plane region, our introduc on into triple integrals will be in the context
of nding the volume of a space region.
To formally nd the volume of a closed, bounded region D in space, such as
the one shown in Figure 13.37(a), we start with an approxima on. Break D into
n rectangular solids; the solids near the boundary of D will either not include
por ons of D or include extra space. In Figure 13.37(b), we zoom in on a por on
of the boundary of D to show a rectangular solid that contains space not in D;
as this is an approxima on of the volume, this is acceptable and this error will
be reduced as we shrink the size of our solids.
The volume Vi of the i th solid Di is Vi = xi yi zi , where xi , yi
and zi give the dimensions of the rectangular solid in the x, y and z direc ons,
respec vely. By summing up the volumes of all n solids, we get an approxima on
of the volume V of D:
V
i=1
Vi =
2 y
x2
2
(a)
xi yi zi .
i=1
Let |D| represent the length of the longest diagonal of rectangular solids
in the subdivision of D. As |D| 0, the volume of each solid goes to 0, as do
each of xi , yi and zi , for all i. Our calculus experience tells us that taking
a limit as |D| 0 turns our approxima on of V into an exact calcula on of
V. Before we state this result in a theorem, we use a deni on to dene some
terms.
.
y
(b)
Figure 13.37: Approxima ng the volume

of a region D in space.
Notes:
773
Deni on 106
Triple Integrals, Iterated Integra on (Part I)
Let D be a closed, bounded region in space. Let a and b be real numbers, let g1 (x) and g2 (x) be
con nuous func ons of x, and let f1 (x, y) and f2 (x, y) be con nuous func ons of x and y.
1. The volume V of D is denoted by a triple integral,
V=
dV.
g2 (x)
f2 (x,y)
2. The iterated integral
dz dy dx is evaluated as
a
g2 (x)
g1 (x)
f1 (x,y)
f2 (x,y)
g2 (x)
dz dy dx =
a
g1 (x)
f2 (x,y)
dz
f1 (x,y)
g1 (x)
dy dx.
f1 (x,y)
Evalua ng the above iterated integral is triple integra on.
Our informal understanding of the nota on

dV is sum up
D
lots of li le
volumes over D, analogous to our understanding of R dA and R dm.
We now state the major theorem of this sec on.
.
Theorem 125
Triple Integra on (Part I)
Let D be a closed, bounded region in space and let D be any subdivision of D into n rectangular
solids, where the i th subregion Di has dimensions xi yi zi and volume Vi .
1. The volume V of D is
V=
dV = lim
|D|0
Vi = lim
|D|0
i=1
xi yi zi .
i=1
2. If D is dened as the region bounded by the planes x = a and x = b, the cylinders y = g( x)

and y = g2 (x), and the surfaces z = f1 (x, y) and z = f2 (x, y), where a < b, g1 (x) g2 (x)
and f1 (x, y) f2 (x, y) on D, then
g2 (x)
f2 (x,y)
dV =
D
dz dy dx.
a
g1 (x)
f1 (x,y)
3. V can be determined using iterated integra on with other orders of integra on (there are 6
total), as long as D is dened by the region enclosed by a pair of planes, a pair of cylinders,
and a pair of surfaces.
Notes:
774

We evaluated the area of a plane region R by iterated integra on, where
the bounds were from curve to curve, then from point to point. Theorem 125
allows us to nd the volume of a space region with an iterated integral with
bounds from surface to surface, then from curve to curve, then from point to
point. In the iterated integral
g2 (x)
f2 (x,y)
dz dy dx,
a
g1 (x)
f1 (x,y)
the bounds a x b and g1 (x) y g2 (x) dene a region R in the x, y plane

over which the region D exists in space. However, these bounds are also dening
surfaces in space; x = a is a plane and y = g1 (x) is a cylinder. The combina on
of these 6 surfaces enclose, and dene, D.
z
2
Examples will help us understand triple integra on, including integra ng

with various orders of integra on.
.
. Example 472
.Finding the volume of a space region with triple integra on
Find the volume of the space region in the 1 st octant bounded by the plane
z = 2 y/3 2x/3, shown in Figure 13.38(a), using the order of integra on
dz dy dx. Set up the triple integrals that give the volume in the other 5 orders of
integra on.
2
x
4
5
y
(a)
Star ng with the order of integra on dz dy dx, we need to

S
rst nd bounds on z. The region D is bounded below by the plane z = 0 (because we are restricted to the rst octant) and above by z = 2 y/3 2x/3;
0 z 2 y/3 2x/3.
To nd the bounds on y and x, we collapse the region onto the x, y plane,
giving the triangle shown in Figure 13.38(b). (We know the equa on of the line
y = 6 2x in two ways. First, by se ng z = 0, we have 0 = 2 y/3 2x/3
y = 6 2x. Secondly, we know this is going to be a straight line between the
points (3, 0) and (0, 6) in the x, y plane.)
We dene that region R, in the integra on order of dy dx, with bounds 0
.
2
x
y=6
2x
5
y
(b)
. 13.38: The region D used in ExamFigure

ple 472 in (a); in (b), the region found by
collapsing D onto the x, y plane.
Notes:
775

y 6 2x and 0 x 3. Thus the volume V of the region D is:
V=
dV
D
62x
2 13 y 32 x
=
0
62x
dz dy dz
)
2
2 13 y 3 x
=
0
dz
dy dz
2 13 y 32 x

z
dy dz
0
0
0
)
3 62x (
1
2
2 y x dy dz.
=
3
3
0
0
3
62x
From this step on, we are evalua ng a double integral as done many mes before. We skip these steps and give the nal volume,
= 6u3 .
.The order dz dx dy:
Now consider the volume using the order of integra on dz dx dy. The bounds
on z are the same as before, 0 z 2y/32x/3. Collapsing the space region
on the x, y plane as shown in Figure 13.38(b), we now describe this triangle with
the order of integra on dx dy. This gives bounds 0 x 3y/2 and 0 y 6.
Thus the volume is given by the triple integral
V=
0
3 12 y
2 31 y 23 x
dz dx dy.
The order: dx dy dz:

Following our surface to surface. . . strategy, we need to determine the
x-surfaces that bound our space region. To do so, approach the region from
behind, in the direc on of increasing x. The rst surface we hit as we enter the
region is the y, z plane, dened by x = 0. We come out of the region at the
plane z = 2 y/3 2x/3; solving for x, we have x = 3 y/2 3z/2. Thus the
bounds on x are: 0 x 3 y/2 3z/2.
Now collapse the space region onto the y, z plane, as shown in Figure 13.39(a).
(Again, we nd the equa on of the line z = 2 y/3 by se ng x = 0 in the equaon x = 3 y/2 3z/2.) We need to nd bounds on this region with the order
Notes:
776

dy dz. The curves that bound y are y = 0 and y = 6 3z; the points that bound
z are 0 and 2. Thus the triple integral giving volume is:
0 x 3 y/2 3z/2
0 y 6 3z
0z2
63z
3y/23z/2
dx dy dz.
0
The order: dx dz dy:

z=
2y/3
3y/23z/2
dx dz dy.
0
y/3
0 x 3 y/2 3z/2
0 z 2 y/3
0y6
.The x-bounds are the same as the order above. We now consider the triangle
in Figure 13.39(a) and describe it with the order dz dy: 0 z 2 y/3 and
0 y 6. Thus the volume is given by:
x
4
5
(a)
The order: dy dz dx:
We now need to determine the y-surfaces that determine our region. Approaching the space region from behind and moving in the direc on of increasing y, we rst enter the region at y = 0, and exit along the plane z =
2 y/3 2x/3. Solving for y, this plane has equa on y = 6 2x 3z. Thus y
has bounds 0 y 6 2x 3z.
Now collapse the region onto the x, z plane, as shown in Figure 13.39(b). The
curves bounding this triangle are z = 0 and z = 2 2x/3; x is bounded by the
points x = 0 to x = 3. Thus the triple integral giving volume is:
0 y 6 2x 3z
0 z 2 2x/3
0x3
22x/3
/3
2x
2
z=
.
2
62x3z
4
5
y
dy dz dx.
0
(b)
The order: dy dx dz:

The y-bounds are the same as in the order above. We now determine the
bounds of the triangle in Figure 13.39(b) using the order dy dx dz. x is bounded
by x = 0 and x = 3 2z/3; z is bounded between z = 0 and z = 2. This leads
to the triple integral:
0 y 6 2x 3z
0 x 3 2z/3
0z2
32z/3
. 13.39: The region D in Example 472

Figure
is collapsed onto the y, z plane in (a); in
(b), the region is collapsed onto the x, z
plane.
62x3z
dy dx dz.
0
Notes:
777

This problem was long, but hopefully useful, demonstra ng how to determine bounds with every order of integra on to describe the region D. In pracce, we only need 1, but being able to do them all gives us exibility to choose
the order that suits us best. .
In the previous example, we collapsed the surface into the x-y, x-z, and y-z
planes as we determined the curve to curve, point to point bounds of integraon. Since the surface was a plane, this collapsing, or projec ng, was simple:
the projec on of the boundaries of a plane onto a coordinate plane is just a line.
The following example shows us how to do this when dealing with more
complicated surfaces and curves.
z
4
. Example 473
.Finding the projec on of a curve in space onto the coordinate planes
Consider the surfaces z = 3 x2 y2 and z = 2y, as shown in Figure 13.40(a).
The curve of their intersec on is shown, along with the projec on of this curve
into the coordinate planes, shown dashed. Find the equa ons of the projec ons
into the coordinate planes.
2 y
2
x
(a)
The two surfaces are z = 3 x2 y2 and z = 2y. To nd

where they intersect, it is natural to set them equal to each other: 3 x2 y2 =
2y. This is an implicit func on of x and y that gives all points (x, y) in the x, y
plane where the z values of the two surfaces are equal.
We can rewrite this implicit func on by comple ng the square:
S
3 x2 y2 = 2y
z
4
2
.
1
2 y
3 x2 z = z2 /4
(b)
(y + 1)2 + x2 = 4.
(z + 2)2
x2
+
= 1,
16
4
and ellipse centered at (0, 2) in the x, z with a major axis of length 8 and a
minor axis of length 4.
Figure 13.40: Finding the projec ons of

the curve of intersec on in Example 473.
Notes:
778
y2 + 2y + x2 = 3
Thus in the x, y plane the projec on of the intersec on is a circle with radius 2,
centered at (0, 1).
To project onto the x, z plane, we do a similar procedure: nd the x and z
values where the y values on the surface are the same. We start by solving the
equa on of each surface for y. In this par cular case, it works well to actually
solve for y2 :
z = 3 x2 y2 y2 = 3 x2 z
z = 2y y2 = z2 /4.
Thus we have (a er again comple ng the square):

Finally, to project the curve of intersec on into the y, z plane, we solve equaon for x. Since z = 2y is a cylinder that lacks the variable x, it becomes our
equa on of the projec on in the y, z plane.
All three projec ons are shown in Figure 13.40(b). .
. Example 474
Set up the triple integrals that nd the volume of the space region D bounded
by the surfaces x2 + y2 = 1, z = 0 and z = y, as shown in Figure 13.41(a),
with the orders of integra on dz dy dx, dy dx dz and dx dz dy.
The region D is bounded below by the plane z = 0 and above by the plane
z = y. The cylinder x2 + y2 = 1 does not oer any bounds in the z-direc on,
as that surface is parallel to the z-axis. Thus 0 z y.
Collapsing the region into the x, y plane, we get part of the circle with equaon x2 + y2 = 1 as
shown in Figure 13.41(b). As a func on
of x, this half circle
has equa ony = 1 x2 . Thus y is bounded below by 1 x2 and above
by y = 0: 1 x2 y 0. The x bounds of the half circle are 1 x 1.
All together, the bounds of integra on and triple integral are as follows:
0 z y
1 x2 y 0
1 x 1
1x2
1x2
1
1
We evaluate this triple integral:

y
1 0
dz
dy
dx
=
The order dz dy dx:
1x2
z = y
1
0.5
x2 +
y2 =
1
x
(a)
dz dy dx.
0
)
y dy dx
1 )0
=
y2 2 dx
2
1x
1
1
)
1(
=
1 x2 dx
2
1
( (
))1

1
1

=
x x3

2
3
1
2 3
= u .
3
0.5
y
(b)
Figure 13.41: The region D in Example 474
is shown in (a); in (b), it is collapsed onto
the x, y plane.
With the order dy dx dz:
Notes:
779

2
2
The region
is bounded below in the y-direc on by the surface x + y =
2
1
y = 1 x and above by the surface y = z. Thus the y bounds are
1 x2 y z.
Collapsing the region onto the x, z plane gives the region shown in Figure
13.42(a); this half circle has equa on x2 + z2 = 1. (We nd this curve by solving
2
each surface for y2 , then se ng them equal to each other. We have y2 =
1 x
2
2
2
2
2
and y = z y =
z . Thus x +z = 1.) It is bounded below by x = 1 z
2
and above by x = 1 z , where z is bounded by 0 z 1. All together, we
have:
0.5
z
1 x2 y
1 z2 x 1 z2
0z1
1z2
1z2
1x2
dy dx dz.
(a)
With the order dx dz dy:

1
D is bounded below by the surface x = 1 y2 and above by 1 y2 .

We then collapse the region onto the y, z plane and get the triangle shown in
Figure 13.42(b). (The hypotenuse is the line z = y, just as the plane.) Thus z is
bounded by 0 z y and y is bounded by 1 y 0. This gives:
0.5
y
1 y2 x 1 y2
0 z y
1 y 0
(b)
Figure 13.42: The region D in Example
474 is shown collapsed onto the x, z plane
in (a); in (b), it is collapsed onto the y, z
plane.
y
0
1y2
dx dz dy.
1y2
.
The following theorem states two things that should make common sense
to us. First, using the triple integral to nd volume of a region D should always
return a posi ve number; we are compu ng volume here, not signed volume.
Secondly, to compute the volume of a complicated region, we could break
it up into subregions and compute the volumes of each subregion separately,
summing them later to nd the total volume.
Notes:
780
Theorem 126
Proper es of Triple Integrals
Let D be a closed, bounded region in

space, and let D1 and D2 be nonoverlapping regions such that D = D1 D2 .
1.
dV 0
2.
dV =
dV +
dV.
D1
D2
We use this la er property in the next example.

1
. Example 475
Find the volume of the space region D bounded by the coordinate planes, z =
1 x/2 and z = 1 y/4, as shown in Figure 13.43(a). Set up the triple integrals
that nd the volume of D in all 6 orders of integra on.
Following the boundsdetermining strategy of surface to
surface, curve to curve, and point to point, we can see that the most dicult
orders of integra on are the two in which we integrate with respect to z rst,
for there are two upper surfaces that bound D in the z-direc on. So we start
by no ng that we have
z=
1
and 0 z 1 y.
4
We now collapse the region D onto the x, y axis, as shown in Figure 13.43(b).
The boundary of D, the line from (0, 0, 1) to (2, 4, 0), is shown in part (b) of the
gure as a dashed line; it has equa on y = 2x. (We can recognize this in two
ways: one, in collapsing the line from (0, 0, 1) to (2, 4, 0) onto the x, y plane,
we simply ignore the z-values, meaning the line now goes from (0, 0) to (2, 4).
Secondly, the two surfaces meet where z = 1 x/2 is equal to z = 1 y/4:
thus 1 x/2 = 1 y/4 y = 2x.)
We use the second property of Theorem 126 to state that
dV =
dV +
dV,
D
D1
D2
where D1 and D2 are the space regions above the plane regions R1 and R2 , respec vely. Thus we can say
( 1x/2 )
( 1y/4 )
dV =
dz dA +
dz dA.
D
R1
R2
.
2
1
0z1 x
2
z=1
x 4
1
2x
(a)
.
R1
R2
x 4
.
(b)
is shown in (a); in (b), it is collapsed onto
the x, y plane.
Notes:
781

All that is le is to determine bounds of R1 and R2 , depending on whether we
are integra ng with order dx dy or dy dx. We give the nal integrals here, leaving
it to the reader to conrm these results.
dz dy dx:
0 z 1 x/2
0 y 2x
0x2
2x
dz dy dx
1y/4
dz dy dx
2x
dz dx dy:
0 z 1 x/2
y/2 x 2
0y4
2
4
2
x 4
1x/2
dV =
z
0 z 1 y/4
2x y 4
0x2
(a)
dV
D
0 z 1 y/4
0 x y/2
0y4
1x/2
dz dx dy
=
0
y/2
y/2
1y/4
dz dx dy
The remaining four orders of integra on do not require a sum of triple integrals. In Figure 13.44 we show D collapsed onto the other two coordinate
planes. Using these graphs, we give the nal orders of integra on here, again
leaving it to the reader to conrm these results.
.
1
dy dx dz:
.
2
4
0 y 4 4z
0 z 1 x/2
0x2
22z
44z
dy dx dz
0
.dy dz dx:
x 4
(b)
Figure 13.44: The region D in Example

475 is shown collapsed onto the x, z plane
in (a); in (b), it is collapsed onto the y, z
plane.
Notes:
782
0 y 4 4z
0 x 2 2z
0z1
1x/2
44z
dy dx dz
0

dx dy dz:
0 x 2 2z
0 y 4 4z
0z1
44z
22z
dx dy dz
0
dx dz dy:
0 x 2 2z
0 z 1 y/4
0y4
1y/4
22z
dx dz dy
0
We give one more example of nding the volume of a space region.

. Example 476
.Finding the volume of a space region
Set up a triple integral that gives the volume of the space region D bounded by
z = 2x2 + 2 and z = 6 2x2 y2 . These surfaces are plo ed in Figure 13.45(a)
and (b), respec vely; the region D is shown in part (c) of the gure.
.
(a)
The main point of this example is this: integra ng with reS

spect to z rst is rather straigh orward; integra ng with respect to x rst is not.
The order dz dy dx:

The bounds on z are clearly 2x2 + 2 z 6 2x2 y2 . Collapsing D onto
the x, y plane gives the ellipse shown in Figure 13.45(c). The equa on of this
ellipse is found by se ng the two surfaces equal to each other:
2x2 + 2 = 6 2x2 y2
2
x 2
4x2 + y2 = 4
.
2
y2
x2 +
= 1.
4
x 2
We can describe this ellipse with the bounds
4 4x2 y 4 4x2 and
.
(b)
1 x 1.
Thus we nd volume as
2x2 + 2 z 6 2x2 y2
4 4x2 y 4 4x2
1 x 1
44x2
44x2
5
62x2 y2
dz dy dx .
2x2 +2
The order dy dz dx:
2
x 2
.
Notes:
(c)
Figure 13.45: The region D is bounded
by the surfaces shown in (a) and (b); D is
shown in (c).
783

Integra ng with respect to y is not too dicult. Since the surface z = 2x2 +2
is a cylinder whose directrix is the y-axis, it does not create a border for y. The
paraboloid z = 6 2x2 y2 does; solving for y, we get the bounds
6 2x2 z y 6 2x2 z.
Collapsing D onto the x, z axes gives the region shown in Figure 13.46(a); the
lower curve is the from the cylinder, with equa on z = 2x2 + 2. The upper
curve is from the paraboloid; with y = 0, the curve is z = 6 2x2 . Thus bounds
on z are 2x2 + 2 z 6 2x2 ; the bounds on x are 1 x 1. Thus we have:
6 2x2 z y 6 2x2 z
1 62x2 62x2 z
dy dz dx.
2x2 + 2 z 6 2x2
1 2x2 +2
62x2 z
1 x 1
z
6
2
2
x 2
This order takes more eort as D must be split into two subregions. The
two surfaces create two sets of upper/lower bounds in terms of x; the cylinder
creates bounds
z/2 1 x z/2 1
(a)
z
6
for region D1 and the paraboloid creates bounds
3 y2 /2 z2 /2 x 3 y2 /2 z2 /2
R2
4
R1
z=4 2
y /2
z=6 2
y
The order dx dz dy:
x 2
for region D2 .
Collapsing D onto the y, z axes gives the regions shown in Figure 13.46(b).
We nd the equa on of the curve z = 4 y2 /2 by no ng that the equa on of
the ellipse seen in Figure 13.45(c) has equa on
x2 + y2 /4 = 1 x = 1 y2 /4.
(b)
Subs tute this expression for x in either surface equa on, z = 6 2x2 y2 or
z = 2x2 + 2. In both cases, we nd

is collapsed onto the x, z plane in (a); in
(b), it is collapsed onto the y, z plane.
1
z = 4 y2 .
2
.Region R1 , corresponding to D1 , has bounds
2 z 4 y2 /2,
2 y 2
and region R2 , corresponding to D2 , has bounds

4 y2 /2 z 6 y2 ,
Notes:
784
2 y 2.

Thus the volume of D is given by:
4y2 /2
z/21
dx dz dy +
z/21
6y2
3y2 /2z2 /2
4y2 /2
dx dz dy.
3y2 /2z2 /2
.
If all one wanted to do in Example 476 was nd the volume of the region D,
one would have likely stopped at the rst integra on setup (with order dz dy dx)
and computed the volume from there. However, we included the other two
methods 1) to show that it could be done, messy or not, and 2) because somemes we have to use a less desirable order of integra on in order to actually
integrate.
Triple Integra on and Func ons of Three Variables

There are uses for triple integra on beyond merely nding volume, just as
there are uses for integra on beyond area under the curve. These uses start
with understanding how to integrate func ons of three variables, which is eecvely no dierent than integra ng func ons of two variables. This leads us to a
deni on, followed by an example.
Deni on 107
Iterated Integra on, (Part II)
Let D be a closed, bounded region in space, over which g1 (x), g2 (x),

f1 (x, y), f2 (x, y) and h(x, y, z) are all con nuous, and let a and b be real
numbers.
g2 (x)
f2 (x,y)
The iterated integral
g2 (x)
h(x, y, z) dz dy dx is evaluated as
a
g1 (x)
f1 (x,y)
f2 (x,y)
g2 (x)
f2 (x,y)
h(x, y, z) dz dy dx =
a
g1 (x)
f1 (x,y)
)
h(x, y, z) dz dy dx.
g1 (x)
f1 (x,y)
. Example 477
.Evalua ng a triple integral of a func on of three variables
1 x 2x+3y
(
)
Evaluate
xy + 2xz dz dy dx.
0
x2
x2 y
We evaluate this integral according to Deni on 107.
Notes:
785
x
x2
2x+3y
x2 y
x (
)
xy + 2xz dz dy dx
2x+3y
=
x2
x(
x2 y
)
)
xy + 2xz dz dy dx
)
(
)2x+3y
2
=
xyz + xz 2
dy dx
x y
0
x2
)
(
1 x
(
)
2
2
2
2
xy(2x + 3y) + x(2x + 3y) xy(x y) + x(x y)
dy dx
=
x2
1 x(
=
0
x2
)
x5 + x3 y + 4x3 + 14x2 y + 12xy2 dy dx.
We con nue as we have in the past, showing fewer steps.

)
1(
7 7
7
=
x 8x6 x5 + 15x4 dx
2
2
0
281
=
0.836.
. 336
We now know how to evaluate a triple integral of a func on of three variables; we do not yet understand what it means. We build up this understanding
in a way very similar to how we have understood integra on and double integra on.
Let h(x, y, z) a con nuous func on of three variables, dened over some
space region D. We can par on D into n rectangularsolid subregions, each
with dimensions xi yi zi . Let (xi , yi , zi ) be some point in the i th subregion, and consider the product h(xi , yi , zi )xi yi zi . It is the product of a
func on value (thats the h(xi , yi , zi ) part) and a small volume Vi (thats the
xi yi zi part). One of the simplest understanding of this type of product is
when h describes the density of an object; then h volume = mass.
We can sum up all n products over D. Again le ng |D| represent the length
of the longest diagonal of the n rectangular solids in the par on, we can take
the limit of the sums of products as |D| 0. That is, we can nd
S = lim
|D|0
i=1
h(xi , yi , zi )Vi = lim
|D|0
h(xi , yi , zi )xi yi zi .
i=1
While this limit has lots of interpreta ons depending on the func on h, in
the case where h describes density, S is the total mass of the object described
by the region D.
Notes:
786

We now use the above limit to dene the triple integral, give a theorem that
relates triple integrals to iterated itera on, followed by the applica on of triple
integrals to nd the centers of mass of solid objects.
Deni on 108
Triple Integral
Let w = h(x, y, z) be a con nuous func on over a closed, bounded space

region D, and let D be any par on of D into n rectangular solids with
volume Vi . The triple integral of h over D is
h(x, y, z) dV = lim
|D|0
h(xi , yi , zi )Vi .
i=1
The following theorem assures us that the above limit exists for con nuous
func ons h and gives us a method of evalua ng the limit.
Theorem 127
Triple Integra on (Part II)
Let w = h(x, y, z) be a con nuous func on over a closed, bounded space

region D, and let D be any par on of D into n rectangular solids with
volume Vi .
1. The limit lim
|D|0
h(xi , yi , zi )Vi exists.
i=1
2. If D is dened as the region bounded by the planes x = a and

x = b, the cylinders y = g1 (x) and y = g2 (x), and the surfaces
z = f1 (x, y) and z = f2 (x, y), where a < b, g1 (x) g2 (x) and
f1 (x, y) f2 (x, y) on D, then
g2 (x)
f2 (x,y)
h(x, y, z) dV =
D
h(x, y, z) dz dy dx.
a
g1 (x)
f1 (x,y)
We now apply triple integra on to nd the centers of mass of solid objects.
Mass and Center of Mass

One may wish to review Sec on 13.4 for a reminder of the relevant terms
and concepts.
Notes:
787

.
Deni on 109
Mass, Center of Mass of Solids
Let a solid be represented by a region D in space with variable density

func on (x, y, z).
1. The mass of the object is M =

dm =
(x, y, z) dV.
D
2. The moment about the x,y plane is Mxy =
z(x, y, z) dV.
D
y(x, y, z) dV.
3. The moment about the x,z plane is Mxz =

D
4. The moment about the y,z plane is Myz =
x(x, y, z) dV.
D
5. The center of mass of the object is

(
)
)
(
Myz Mxz Mxy
x, y, z =
,
,
.
M M M
. Example 478
.Finding the center of mass of a solid
Find the mass, and center of mass, of the solid represented by the space region
bounded by the coordinate planes and z = 2 y/3 2x/3, shown in Figure
13.47, with constant density (x, y, z) = 3gm/cm3 . (Note: this space region was
used in Example 472.)
z
2
We apply Deni on 109. In Example 472, we found bounds

S
for the order of integra on dz dy dx to be 0 z 2y/32x/3, 0 y 62x
and 0 x 3. We nd the mass of the object:
M=
(x, y, z) dV
.
x
62x
2y/32x/3
4
5
62x
2y/32x/3
=3
dz dy dx
0
Figure 13.47: Finding the center of mass

of. this solid in Example 478.
( )
3 dz dy dx
= 3(6) = 18gm.
The evalua on of the triple integral is done in Example 472, so we skipped those
Notes:
788

steps above. Note how the mass of an object with constant density is simply
densityvolume.
We now nd the moments about the planes.
Mxy =
3z dV
D
62x
2y/32x/3
( )
3z dz dy dx
)2
1(
=
2x + y 6 dy dx
6
0
0
3
)3
4(
=
x 3 dx
9
0
= 9.
3
62x
We omit the steps of integra ng to nd the other moments.
Myz =
3x dV
D
27
=
.
2
Mxz =
3y dV
D
= 27.
The center of mass is
.
)
(
x, y, z =
27/2 27 9
, ,
18 18 18
(
)
= 0.75, 1.5, 0.5 .
. Example 479
.Finding the center of mass of a solid
Find the center of mass of the solid represented by the region bounded by the
planes z = 0 and z = y and the cylinder x2 + y2 = 1, shown in Figure 13.48,
with density func on (x, y, z) = 10 + x2 + 5y 5z. (Note: this space region
was used in Example 474.)
As we start, consider the density func on. It is symmetric
about the y, z plane, and the farther one moves from this plane, the denser the
object is. The symmetry indicates that x should be 0.
As one moves away from the origin in the y or z direc ons, the object becomes less dense, though there is more volume in these regions.
Though none of the integrals needed to compute the center of mass are
par cularly hard, they do require a number of steps. We emphasize here the
z = y
0.5
x2 +
y2 =
1
x
Figure 13.48: Finding the center of mass

of this solid in Example 479.
Notes:
789

importance of knowing how to set up the proper integrals; in complex situa ons
we can appeal to technology for a good approxima on, if not the exact answer.
We use the order of integra on dz dy dx, using the bounds found in Example
474. (As these are the same for all four triple integrals, we explicitly show the
bounds only for M.)
(
)
10 + x2 + 5y 5z dV
M=
=
1
1x2
)
10 + x2 + 5y 5z dV
64 15
=
3.855.
5 16
(
)
Myz =
x 10 + x2 + 5y 5z dV
D
= 0.
(
)
Mxz =
y 10 + x2 + 5y 5z dV
D
61
=2
1.99.
48
(
)
Mxy =
z 10 + x2 + 5y 5z dV
D
61 10
=
0.885.
96
9
Note how Myz = 0, as expected. The center of mass is
(
)
(
)
(
)
1.99 0.885
x, y, z = 0,
,
0, 0.516, 0.230 .
3.855 3.855
.
As stated before, there are many uses for triple integra on beyond nding
volume. Whenh(x,
y, z) describes a rate of change func on over some space
region D, then
h(x, y, z) dV gives the total change over D. Our one specic

D
example of this was compu ng mass; a density func on is simply a rate of mass
change per volume func on. Integra ng density gives total mass.
While knowing how to integrate is important, it is arguably much more important to know how to set up integrals. It takes skill to create a formula that describes a desired quan ty; modern technology is very useful in evalua ng these
formulas quickly and accurately.
Notes:
790
Exercises 13.6
Terms and Concepts
10. D is bounded by the planes y = 0, y = 2, x = 1, z = 0 and

z = (3 x)/2.
1. The strategy for establishing bounds for triple integrals

is
to
,
to
and
to
.
2. Give an informal interpreta on of what

dV
Evaluate the triple integral with order dx dy dz.
means.
3. Give two uses of triple integra on.

4. If an object has a constant density and a volume V, what
is its mass?
1
z=
1
2 (3
x)
Problems
In Exercises 5 8, two surfaces f1 (x, y) and f2 (x, y) and a region R in the x, y plane are given. Set up and evaluate the
double integral that nds the volume between these surfaces
over R.
1
2
x
5. f1 (x, y) = 8 x y , f2 (x, y) = 2x + y;
R is the square with corners (1, 1) and (1, 1).
2
1
2
6. f1 (x, y) = x2 + y2 , f2 (x, y) = x2 y2 ;
R is the square with corners (0, 0) and (2, 3).
7. f1 (x, y) = sin x cos y, f2 (x, y) = cos x sin y + 2;
R is the triangle with corners (0, 0), (, 0) and (, ).
8. f1 (x, y) = 2x2 + 2y2 + 3, f2 (x, y) = 6 x2 y2 ;
R is the circle x2 + y2 = 1.
In Exercises 9 16, a domain D is described by its bounding
surfaces, along with a graph. Set up the triple integrals that
give the volume of D in all 6 orders of integra on, and nd
the volume of D by evalua ng the indicated triple integral.
11. D is bounded by the planes x = 0, x = 2, z = y and by

z = y2 /2.
Evaluate the triple integral with the order dy dz dx.
9. D is bounded by the coordinate planes and

z = 2 2x/3 2y.
Evaluate the triple integral with order dz dy dx.
z
z
2
z=
z=
2
x
2
1 y
2
.
1
z = 2 23 x 2y
.
.
12. D is bounded by the planes z = 0, y = 9, x = 0 and by
z = y2 9x2 .
14. D is bounded by the plane z = 2y and by y = 4 x2 .
Evaluate the triple integral with the order dz dy dx.

Do not evaluate any triple integral.
z
z
8
6
6
4
z = y2 9x2
x2
y=
z=
2y
13. D is bounded by the planes x = 2, y = 1, z = 0 and

z = 2x + 4y 4.
15. D is bounded by the coordinate planes and by

y = 1 x2 and y = 1 z2 .
Do not evaluate any triple integral. Which order is easier to

evaluate: dz dy dx or dy dz dx? Explain why.
Evaluate the triple integral with the order dx dy dz.
z
z
1
4
y=
z = 2x + 4y 4
1
2
2
y
2
1
1
x
y = 1 x2
16. D is bounded by the coordinate planes and by

z = 1 y/3 and z = 1 x.
Evaluate the triple integral with order dx dy dz.
z
z=
1
( 2
)
x y + y2 x
z x2 +y2
dz dy dx
e
y2
y2
22. D is bounded by the planes y = 0, y = 2, x = 1, z = 0 and

z = (3 x)/2; (x, y, z) = 2gm/cm3 .
(Note: this is the same region as used in Exercise 10.)
23. D is bounded by the planes x = 2, y = 1, z = 0 and
z = 2x + 4y 4; (x, y, z) = x2 lb/in3 .
.
In Exercises 17 20, evaluate the triple integral.
/2
(
)
cos x sin y sin z dz dy dx
17.
/2
)
sin(yz) dx dy dz
21. D is bounded by the coordinate planes and

z = 2 2x/3 2y; (x, y, z) = 10gm/cm3 .
3
y/
In Exercises 21 24, nd the center of mass of the solid represented by the indicated space region D with density func on
(x, y, z).
.
z=
z
0
x3
20.
0
2
)
x + y + z dz dy dx
19.
x+y
18.
24. D is bounded by the plane z = 2y and by y = 4 x2 .

(x, y, z) = y2 lb/in3 .
A: S
T S
Chapter 1
P
This implies we can let = . Then:
|x 5| <
Sec on 1.1
< x 5 <
< x 5 <
1. Answers will vary.
< (x 3) 2 <
3. F
< (x + 3) (2) <
|3 x (2)| < ,
7. 5
which is what we wanted to prove.
9. 2
7. Let > 0 be given. We wish to nd > 0 such that when

|x 4| < , |f(x) 15| < .
Consider |f(x) 15| < , keeping in mind we want to make a
statement about |x 4|:
11. Limit does not exist.

13. 7
|f(x) 15| <

h
0.1
0.01
0.01
0.1
17.
h
0.1
0.01
0.01
0.1
19.
21.
h
0.1
0.01
0.01
0.1
23.
h
0.1
0.01
0.01
0.1
|x2 + x 5 15| <
f(a+h)f(a)
h
9
9
9
9
|x2 + x 20| <
The limit seems to be exactly 9.
|x 4| |x + 5| <
|x 4| < /|x + 5|
f(a+h)f(a)
h
0.114943
0.111483
0.110742
0.107527
The limit is approx. 0.11.
Since x is near 4, we can safely assume that, for instance,

3 < x < 5. Thus
3+5<x+5<5+5
8 < x + 5 < 10
1
1
1
<
<
8
x+5
10
<
<
8
x+5
10
f(a+h)f(a)
h
0.202027
0.2002
0.1998
0.198026
The limit is approx. 0.2.
f(a+h)f(a)
h
0.0499583
0.00499996 The limit is approx. 0.005.
0.00499996
0.0499583
Sec on 1.2
1. should be given rst, and the restric on |x a| < implies
|f(x) K| < , not the other way around.
3. T
|x 5| < , |f(x) (2)| < .
Consider |f(x) (2)| < :
|f(x) + 2| <
|(3 x) + 2| <
|5 x| <
< 5 x <
< x 5 < .
Let = 8 . Then:
|x 4| <
|x 4| <
8
|x 4| <
x+5
|x + 5|
|x 4| |x + 5| <
x+5
Assuming x is near 4, x + 4 is posi ve and we can drop the
absolute value signs on the right.
|x 4| |x + 5| <
(x + 5)
x+5
|x2 + x 20| < |(x2 + x 5) 15| < ,

which is what we wanted to prove.
|x 2| < , |f(x) 5| < . However, since f(x) = 5, a constant
func on, the la er inequality is simply |5 5| < , which is
always true. Thus we can choose any we like; we arbitrarily
choose = .

|x 0| < , |f(x) 0| < . In simpler terms, we want to show
that when |x| < , | sin x| < .
Set = . We start with assuming that |x| < . Using the hint,
we have that | sin x| < |x| < = . Hence if |x| < , we know
immediately that | sin x| < .
11.
(a) 2
(b) 2
(c) 2
(d) 0
(e) 2
Sec on 1.3
(f) 2
(g) 2

(h) Not dened

13.
5. As x is near 1, both f and g are near 0, but f is approximately twice

the size of g. (I.e., f(x) 2g(x).)
(a) 2
(b) -4
(c) Does not exist.
7. 6
(d) 2
15.
(a) 0
11. Not possible to know.
(b) 0
13. 45
(c) 0
15. 1
(d) 0
17.
(e) 2
19. 0.000000015 0
(f) 2
21. Limit does not exist
(g) 2
(h) 2
23. 2
25.
2 +3+5
5 2 23
0.6064
17.
(b) sin2 a
27. 8
(c) sin2 a
29. 10
31. 3/2
(a) 1 cos2 a = sin2 a
(d) sin2 a
19.
(a) 4
33. 0
(b) 4
35. 9
(c) 4
37. 5/8
39. /180
(d) 3
21.
(a) 1
Sec on 1.4
(b) 1
1. The func on approaches dierent values from the le and right;

the func on grows without bound; the func on oscillates.
(d) 0
(c) Does not exist
3. F
5.
(a) 2
(b) 2
(c) 2
7.
25. 1/2
27. 31/19
29. 11/81
(d) 1
Sec on 1.5
(e) As f is not dened for x < 0, this limit is not dened.
(f) 1
3. A root of a func on f is a value c such that f(c) = 0.
(a) Does not exist.
5. F
(b) Does not exist.
7. T
(c) Does not exist.

(d) Not dened.
(e) 0
(f) 0
9.
23. 2/3
(a) 2
(b) 2
9. F
11. No; lim f(x) = 2, while f(1) = 1.
x1
13. No; f(1) does not exist.

15. Yes
17.
(a) No; lim f(x) = f(2)

x2
(c) 2
(b) Yes
(d) 2
(c) No; f(2) is not dened.
19.
(a) Yes
(b) No; the le and right hand limits at 1 are not equal.
21.
(a)
x
2.9
2.99
f(x)
132.857 It seems limx3 f(x) = .
12124.4
(b)
x
3.1
3.01
f(x)
108.039 It seems limx3+ f(x) = .
11876.4
(a) Yes
(b) No. limx8 f(x) = 16/5 = f(8) = 5.
23. (, 2] [2, )
25. (, 6] [ 6, )
(c) It seems limx3 f(x) = .
27. (, )
19. Horizontal asymptote at y = 2; ver cal asymptotes at x = 5, 4.
29. (0, )
21. Horizontal asymptote at y = 0; ver cal asymptotes at

x = 1, 0, 3.
31. (, 0]
33. Yes, by the Intermediate Value Theorem.
35. We cannot say; the Intermediate Value Theorem only applies to
func on values between 10 and 10; as 11 is outside this range,
we do not know.
37. Approximate root is x = 1.23. The intervals used are:
[1, 1.5] [1, 1.25] [1.125, 1.25]
[1.1875, 1.25] [1.21875, 1.25] [1.234375, 1.25]
[1.234375, 1.2421875] [1.234375, 1.2382813]
39. Approximate root is x = 0.69. The intervals used are:
[0.65, 0.7] [0.675, 0.7] [0.6875, 0.7]
[0.6875, 0.69375] [0.690625, 0.69375]
41.
(a) 20
23. No horizontal or ver cal asymptotes.

25.
27.
29. Solu on omi ed.
31. Yes. The only ques onable place is at x = 3, but the le and
right limits agree.
Chapter 2
Sec on 2.1
1. T
(b) 25
(c) Limit does not exist

(d) 25

7. f (x) = 2
9. g (x) = 2x
Sec on 1.6
11. h (x) =
1. F
13.
(a) y = 6
3. F
(b) x = 2
5. T
(a) y = 3x + 4
15.

9.
(b) y = 1/3x + 4
(a)
(a) y = 7(x + 1) + 8
17.
(b)
11.
(b) y = 1/7(x + 1) + 8
(a) 1
(a) y = 1(x 3) + 1
19.
(b) 0
(b) y = 1(x 3) + 1
(c) 1/2
13.
1
x2
(d) 1/2
21. y = 0.99(x 9) + 1
(a) Limit does not exist
23. y = 0.05x + 1
(b) Limit does not exist
25.
(a) Approxima ons will vary; they should match (c) closely.
(b) f (x) = 1/(x + 1)2
15. Tables will vary.
(a)
x
2.9
2.99
2.999
f(x)
15.1224
It seems limx3 f(x) = .
159.12
1599.12
(b)
x
3.1
3.01
3.001
f(x)
16.8824
It seems limx3+ f(x) = .
160.88
1600.88
(c) At (0, 1), slope is 1. At (1, 0.5), slope is 1/4.

y
x
6
2
2
(c) It seems limx3 f(x) does not exist.

17. Tables will vary.
27. .
31. f (x) = f (x) = f (x) = f(4) (x) = 0
33. Tangent line: y = t + 4

Normal line: y = t + 4
0.5
35. Tangent line: y = 4

Normal line: x = /2
0.5
29.
37. Tangent line: y = 2x + 3

Normal line: y = 1/2(x 5) + 13
39. The tangent line to f(x) = x4 at x = 3 is y = 108(x 3) + 81;

thus (3.01)4 y(3.01) = 108(.01) + 81 = 82.08.
31. Approximately 24.

(a) (, )
33.
(b) (, 1) (1, 1) (1, )
Sec on 2.4
(c) (, 5]
1. F
(d) [5, 5]
3. T
Sec on 2.2
5. F
1. Velocity
7.
(b) f (x) = 3x2 + 6x
3. Linear func ons.
(c) They are equal.
5. 17
7. f(10.1) is likely most accurate, as accuracy is lost the farther from
x = 10 we go.
9.
13.
(a) h (s) = 2(s + 4) + (2s 1)(1)

(b) h (s) = 4s + 7
(c) They are equal.
9. 6
11.
(a) f (x) = (x2 + 3x) + x(2x + 3)
/s2
11.
(a) thousands of dollars per car
17. Either g(x) = f (x) or f(x) = g (x) is acceptable. The actual

answer is g(x) = f (x), but is very hard to show that f(x) = g (x)
given the level of detail given in the graph.
(a) h (s) =
4s3 (0)3(12s2 )
16s6
(b) h (s) = 9/4s4

(c) They are equal.
15.
21. f () 0.
(a) f (x) =
(x+2)(4x3 +6x2 )(x4 +2x3 )(1)

(x+2)2
(b) f(x) = x3 when x = 2, so f (x) = 3x2 .

(c) They are equal.
Sec on 2.3
17.
1. Power Rule.
f (t)
2
(csc t
t3
19. g (t) =
5. g(x) and h(x)
21. h (t) = 14t + 6
7. One possible answer is f(x) = 17x 205.

9.
f (x)
is a velocity func on, and
11.
f (x)
= 14x 5
f (x)
is accelera on.
13. m (t) = 45t4 38 t2 + 3

15. f (r) = 6er
19.
h (t)
2
x
et
1
cos t + sin t
21. f (x) = 0
4) +
1
( csc t cot t)
t2
(cos t2t2 )(5t4 )(t5 )( sin t4t)

(cos t2t2 )2
3. One answer is f(x) = 10ex .
17. f (x) =
3
x2
(c) They are equal.

13.
19. f (x) = 10x
x(2x)(x2 +3)1
x2
(b) f (x) = 1
(b) It is likely that P(0) < 0. That is, nega ve prot for not
producing any cars.
15. f(x) = g (x)
(a) f (x) =
23. f (t) = 15 x4/5 (sec t + et ) +

25. g (x) = 0
27. f (x) =
(3t +2)(ln 22t )(2t +3)(ln 3)

(3t +2)2
29. g (x) = 2 sin x sec x + 2x cos x sec x + 2x sin x sec x tan x =

2 tan x + 2x + 2x tan2 x = 2 tan x + 2x sec2 x
3
) 3
= x
2
2
3
3
)
=
x
2
2
31. Tangent line: y = (x

Normal line: y = (x
23. g (x) = 24x2 120x + 150
33. Tangent line: y = 9x 5

Normal line: y = 1/9x 5
25. f (x) = 18x 12
35. x = 0
27. g (x) = 2 sin x g (x) = 2 cos x g (x) = 2 sin x

g(4) (x) = 2 cos x
37. x = 2, 0
29. p () = 43 32 p () = 122 6 p () = 24 6
p(4) () = 24
5
t(sec t tan t + et )
39. f(4) (x) = 4 cos x + x sin x

41. f(8) = 0
13.
15.
4
2
1
2
y
x
2 sin(y) cos(y)
x
1
2y+2
19.
cos(x)(x+cos(y))+sin(x)+y
sin(y)(sin(x)+y)+x+cos(y)
21. 2x+y
2y+x
4
6
43. .
17.
x
dy
dx
23.
(a) y = 0
(b) y = 1.859(x 0.1) + 0.281
10
25.
(a) y = 4
(b) y = 0.93(x 2) +
27.
(a) y =
(b) y =
45. .5
Sec on 2.5
d y
dx2
31.
d2 y
dx2
=0
3. F
5. T
3y
5 x8/5
3(x
+
3 1
5 yx6/5
Tangent line: y = (32 /4)(x /2) + (/2)3

)
(x+1)(x+2) ( 1
1
1
1
37. y = (x+3)(x+4) x+1
+ x+2
x+3
x+4
7. f (t) = 15(3t 2)4

9. h (t) = (6t + 1)e3t
+t1
Tangent line: y = 11/72x + 1/6
11. f (x) = 3 sin(3x)

13. h (t) = 8 sin3 (2t) cos(2t)
Sec on 2.7
15. f (x) = tan x
1. F
17. f (x) = 2/x

g (t)
= ln 5
5cos t
3. The point (10, 1) lies on the graph of y = f1 (x) (assuming f is

inver ble).
sin t
5. Compose f(g(x)) and g(f(x)) to conrm that each equals x.
21. m (w) = ln(3/2)(3/2)w

23. f (x) =
2x (ln 33x x2 2x+1)(3x +x)(ln 22x 2x)

2
22x
25. g (t) = 5 cos(t2 +3t) cos(5t7)(2t+3) sin(t2 +3t) sin(5t7)

27. Tangent line: y = 0
Normal line: x = 0
7. Compose f(g(x)) and g(f(x)) to conrm that each equals x.

( )
9. f1 (20) = f 1(2) = 1/5
( )
1
11. f1 ( 3/2) = f (/6)
=1
( 1 )
13. f
(1/2) = f 1(1) = 2
29. Tangent line: y = 3( /2) + 1

Normal line: y = 1/3( /2) + 1
15. h (t) =
31. In both cases the deriva ve is the same: 1/x.
17. g (x) =
33.
)
2(
33. y = (2x)x 2x ln(2x) + x
Tangent line: y = (2 + 4 ln 2)(x 1) + 2
)
(
35. y = xsin(x)+2 cos x ln x + sin x+2
x
1. T
19.
108
72 ) + 6+32 3
4+32 3 ) + 32
1 (x
3
3/5
29.
(a)
F/mph
(b) The sign would be nega ve; when the wind is blowing at
10 mph, any increase in wind speed will make it feel colder,
i.e., a lower number on the Fahrenheit scale.
2
14t2
2
1+4x2
sin(t)
19. g (t) = cos1 (t) cos(t) 2
1t
sin
21. h (x) =
(x)+cos
(x)
1x2 cos1 (x)2
Sec on 2.6
23. f (x) = 1
25.
1x2
(a) f(x) = x, so f (x) = 1

(b) f (x) = cos(sin1 x) 1
3. T
1x2
5. f (x) = 13 x2/3 =
7.
g (t)
9.
dy
dx
11.
dy
dx
= sin(x) sec(y)
t cos t +
4x3
2y+1
3 2
x
sin
t
2 t
27.
= 1.
(a) f(x) = 1 x2 , so f (x) = x
1x2
(b)
f (x)
29. y = 4(x
cos(cos1
x)( 1 2
1x
3/4) + /6
31. y = 4/5(x 1) + 2
= x
1x2
Chapter 3
15. c.p. at c = 2, 0; increasing on (, 2) (0, ); decreasing

on (2, 0); rel. min at x = 0; rel. max at x = 2.
Sec on 3.1
17. c.p. at c = 1; increasing on (, );
19. c.p. at c = 1, 0, 1; decreasing on (, 1) (1, 0)

increasing on (0, 1) (1, ); rel. min at x = 0;
21. c.p. at c = 2, 6, 0; decreasing on (, 0) (0, 2); increasing on

(2, ); rel. min at x = 2;
5. F
7. A: abs. min B: none C: abs. max D: none E: none
9.
f (0)
11.
f (0)
=0
f (2)
=0
f (3.2)
=0
= 0 f (4) is undened
13. f (0) is not dened
23. c.p. at c = 1, 1 decreasing on (1, 1) increasing on

(, 1) (1, ); rel. min at x = 1; rel. max at x = 1
25. c = cos1 (2/)
Sec on 3.4
15. min: (0.5, 3.75)

max: (2, 10)
17. min: (/4, 3 2/2)

max: (/2, 3)

19. min: ( 3, 2 3)
max: (5, 28/5)
11. Graph and verify.
21. min: (, e )
max: (/4,
/4
2e
)
2
23. min: (1, 0)

max: (e, 1/e)
25.
dy
dx
27.
3x2
y(y2x)
x(x2y)
+1
Sec on 3.2
3. Any c in [1, 1] is valid.
5. c = 1/2
3. Yes; Answers will vary.


17. Possible points of inec on: none; concave down on (, )
19. Possible points of inec on: x = 1/2; concave down on
(, 1/2); concave up on (1/2, )
21. Possible points

of inec on: x= (1/3)(2 7); concave up on
((1/3)(2 7), (1/3)(2
+ 7)); concave
down on
(, (1/3)(2 7)) ((1/3)(2 + 7), )
23. Possible
downon
points
of inec on: x = 1/ 3; concave
(1/ 3, 1/ 3); concave up on (, 1/ 3) (1/ 3, )

25. Possible points of inec on: x = /4, 3/4; concave down on
(/4, 3/4) concave up on (, /4) (3/4, )
7. Rolles Thm. does not apply.
27. Possible points of inec on: x = 1/e3/2 ; concave down on

(0, 1/e3/2 ) concave up on (1/e3/2 , )
9. Rolles Thm. does not apply.
29. min: x = 1
11. c = 0
13. c = 3/ 2
15. The Mean Value Theorem does not apply.
17. c = sec1 (2/ )

19. c =
57 7
6
31. max: x = 1/ 3 min: x = 1/ 3

33. min: x = 1
35. min: x = 1
37. cri cal values: x = 1, 1; no max/min
39. max: x = 2; min: x = 0
41. max: x = 0
21. Max value of 19 at x = 2 and x = 5; min value of 6.75 at

x = 1.5.
43. f has no maximal or minimal value
23. They are the odd, integer valued mul ples of /2 (such as
0, /2, 3/2, 5/2, etc.)
47. f has a rela ve

max at: x = (1/3)(2 + 7) rela ve min at:
x = (1/3)(2 7)
49. f has a rela ve max at x = 1/ 3; rela ve min at x = 1/ 3
Sec on 3.3
45. f has a minimal value at x = 1/2
51. f has a rela ve min at x = 3/4; rela ve max at x = /4
53. f has a rela ve min at x = 1/ e3 = e3/2
5. Increasing
Sec on 3.5
3. T
5. T
7. A good sketch will include the x and y intercepts..
9. Use technology to verify sketch.
11.
(a) 50.92 /min

(b) 0.509 /min
(c) 0.141 /min

As the tank holds about 523.6

13.
(a) The rope is 80 long.

(c) 1.87 /sec
(d) About 34 feet.

n/2b
,
a
27. Cri cal points: x =

where n is an odd integer Points of
inec on: (n b)/a, where n is an integer.
29.
dy
dx
= x/y, so the func on is increasing in second and fourth

quadrants, decreasing in the rst and third quadrants.
d2 y
dx2
= 1/y
is posi ve when y < 0 and is
nega ve when y > 0. Hence the func on is concave down in the
rst and second quadrants and concave up in the third and fourth
quadrants.
x2 /y3 , which
Chapter 4
Sec on 4.1
1. F
3. x0 = 1.5, x1 = 1.5709148, x2 = 1.5707963, x3 = 1.5707963,
x4 = 1.5707963, x5 = 1.5707963
15. The cone is rising at a rate of 0.003 /s.
Sec on 4.3
1. T
3. 2500; the two numbers are each 50.
5. There is no maximum sum; the fundamental equa on has only 1
cri cal value that corresponds to a minimum.
7. Area = 1/4, with sides of length 1/ 2.

9. The radius should be about 3.84cm and the height should be
2r = 7.67cm. No, this is not the size of the standard can.
11. The height and width should be 18 and the length should be 36,
giving a volume of 11, 664in3 .
13. 5 10/ 39 3.4 miles should be run underground, giving a

minimum cost of $374,899.96.
15. The dog should run about 19 feet along the shore before star ng
to swim.
17. The largest area is 2 formed by a square with sides of length 2.
5. x0 = 0, x1 = 2, x2 = 1.2, x3 = 1.0117647, x4 = 1.0000458,

x5 = 1
Sec on 4.4
7. x0 = 2, x1 = 0.6137056389, x2 = 0.9133412072,
x3 = 0.9961317034, x4 = 0.9999925085, x5 = 1
1. T
9. roots are: x = 3.714, x = 0.857, x = 1 and x = 1.571
11. roots are: x = 2.165, x = 0, x = 0.525 and x = 1.813

13. x = 0.637, x = 1.410
15. x = 4.493, x = 0
17. The approxima ons alternate between x = 1, x = 2 and x = 3.
Sec on 4.2
1. T
3.
(a) 5/(2) 0.796cm/s

(b) 1/(4) 0.0796 cm/s
(c) 1/(40) 0.00796 cm/s
3. F
7. Use y = x2 ; dy = 2x dx with x = 6 and dx = 0.07. Thus
dy = 0.84; knowing 62 = 36, we have 5.932 35.16.
9. Use y = x3 ; dy = 3x2 dx with x = 7 and dx = 0.2. Thus
dy = 29.4; knowing 73 = 343, we have 6.83 313.6.
11. Use y = x; dy = 1/(2

x) dx with x =25 and dx = 1. Thus
dy = 0.1; knowing 25 = 5, we have 24 4.9.
3
13. Use y = 3 x; dy = 1/(3 x2 ) dx withx = 8 and dx = 0.5. Thus
dy
=
1/24
1/25
=
0.04;
knowing 3 8 = 2, we have
3
8.5 2.04.
15. Use y = cos x; dy = sin x dx with x = /2 1.57 and
dx 0.07. Thus dy = 0.07; knowing cos /2 = 0, we have
cos 1.5 0.07.
5. 49mph
17. dy = (2x + 3)dx
7. Due to the height of the plane, the gun does not have to rotate
very fast.
19. dy =
(a) 0.0573 rad/s

(b) 0.0725 rad/s
(c) In the limit, rate goes to 0.0733 rad/s
9.
(a) 0.04 /s
it will take about 52.36 minutes.
(b) 1.71 /sec
3,
2
dx
4x3
( 3x
)
21. dy = 2xe + 3x2 e3x dx
23. dy =
2(tan x+1)2x sec2 x

dx
(tan x+1)2
25. dy = (ex sin x + ex cos x)dx

27. dy =
1
dx
(x+2)2
(b) 0.458 /s
29. dy = (ln x)dx
(c) 3.35 /s
31.
(d) Not dened; as the distance approaches 24, the rates

approaches .
33. 48in2 , or 1/3
(a) 12.8 feet

(b) 32 feet
2
35.
(a) 298.8 feet
(f) 2
(b) 17.3
9.
(c) 5.8%
(a)
(b)
(c) 2
37. The isosceles triangle setup works the best with the smallest
percent error.
Chapter 5
(d) 10
11.
(a) 4/
(b) 4/
Sec on 5.1
(c) 0
(d) 2/
13.
(a) 40/3
(b) 26/3
(c) 8/3
7. velocity
(d) 38/3
9. 1/9x9 + C
15.
11. t + C
(a) 3 /s
(b) 9.5
13. 1/(3t) + C
15. 2 x + C
(c) 9.5
17.
(a) 96 /s
17. cos + C
(b) 6 seconds
19. 5e + C
(c) 6 seconds
21.
5t
2 ln 5
(d) Never; the maximum height is 208 .
+C
19. 5
23. t6 /6 + t4 /4 3t2 + C
21. Answers can vary; one solu on is a = 2, b = 7
25. e x + C
27.
23. 7
(a) x > 0
25. Answers can vary; one solu on is a = 11, b = 18
(b) 1/x
27. cos x sin x + tan x + C
(c) x < 0
29. ln |x| + csc x + C
(d) 1/x
Sec on 5.3
(e) ln |x| + C. Explana ons will vary.

29. 5ex + 5
1. limits
31. tan x + 4
3. Rectangles.
33. 5/2x2 + 7x + 3
5. 22 + 32 + 42 = 29
7. 0 1 + 0 + 1 + 0 = 0
35. 5ex 2x
37.
2x ln (2)+2 +x ln 2)(ln 321)+ln (2) cos(x)1ln (2)

ln2 (2)
39. No answer provided.
Sec on 5.2
9. 1 + 2 3 + 4 5 + 6 = 3
11. 1 + 1 + 1 + 1 + 1 + 1 = 6
13. Answers may vary; 8i=0 (i2 1)
15. Answers may vary; 4i=0 (1)i ei
17. 1045
3. 0
19. 8525
5.
7.
(a) 3
21. 5050
(b) 4
23. 155
(c) 3
25. 24
(d) 0
27. 19
(e) 4
29. /3 + /(2 3) 1.954
(f) 9
31. 0.388584
(a) 4
33.
(b) 2
(e) 1
(1+n)2
.
4n2
(b) 121/400, 10201/40000, 1002001/4000000
(c) 4
(d) 2
(a) Exact expressions will vary;

(c) 1/4
35.
(a) 8.
(b) 8, 8, 8
(c) 16/3 5.33
(c) 8
(a) Exact expressions will vary; 100 200/n.
37.
9.
(a) 0.2207
(b) 0.2005
(b) 80, 98, 499/5
(c) 1/5
(a) 9/2(1 + 3) 12.294
(b) 3 + 6 3 13.392
(c) 100
39. F(x) = 5 tan x + 4
11.
41. G(t) = 4/6t6 5/4t4 + 8t + 9

43. G(t) = sin t cos t 78
(c) 9/2 14.137
Sec on 5.4
13. Trapezoidal Rule: 3.0241

Simpsons Rule: 2.9315

3. T

5. 20
7. 0

9. 1
11. (5 1/5)/ ln 5
(
)
(a) n = 150 (using max f (x) = 1)
(
)
(b) n = 18 (using max f (4) (x) = 7)
( )
(a) n = 5591 (using max f (x) = 300)
(
)
(b) n = 46 (using max f (4) (x) = 24)
21.
13. 4
23.
15. 16/3
17. 45/4
(a) Area is 25.0667 cm2
25.
19. 1/2
(b) Area is 250,667 yd2
21. 1/2
Chapter 6
23. 1/4
25. 8
Sec on 6.1
27. 0
29. Explana ons will vary. A sketch will help.
31. c = 2/ 3
1. Chain Rule.
3.
1 3
(x
8
33. c = 64/9 7.1
5.
1
18
35. 2/pi
7.
1
2
9.
2
(x + 3)3/2
3
2e x + C
37. 16/3
39. 1/(e 1)
11.
41. 400
5)8 + C
( 2
)9
x +1 +C
ln |2x + 7| + C
13. 2x12
43. 1
15.
45. 64 /s
sin3 (x)
3
1
x
+C
+C
17. tan(4 x) + C
47. 2 /s
19.
49. 27/2
tan3 (x)
3
+C
21. tan(x) x + C
51. 9/2
ex
3
53. F (x) = (3x2 + 1) x3 1+x
23.
55. F (x) = 2x(x2 + 2) (x + 2)
25. x ex + C
Sec on 5.5
1. F
3. They are superseded by the Trapezoidal Rule; it takes an equal
amount of work and is generally more accurate.
5.
(a) 250
7.
27x
ln 27
29.
1
2
ln2 (x)
31.
1
6
ln2
33.
x2
2
x
3
+C
+C
( 3)
x +C
+ 3x + ln |x|) + C
x2
2
+ x 2 ln |x + 1| + C
3 2
x
2
8x + 15 ln |x + 1| + C
( )
39.
7 tan1 x + C
7
( )
41. 14 sin1 x + C
37.
2 + 3 5.15
(b) 2/3(3 + 2 + 2 3) 5.25

(a) 2 +
+C
27.
35.
(b) 250
(c) 250
6(x + 3)1/2 + C = 32 (x 6) x + 3 + C
43.
45.
5
4
sec1 (|x|/4) + C
tan
+C
(
)
1 4x
5
31
3(x +3)
41. 0
43. 1/2
47. 3 sin
49.
x1
4e54
(
)
47. 1/5 e + e
45.
+C
+C
3
4e2
Sec on 6.3
51. 1 x2 + C
1. F
53. 23 cos 2 (x) + C
3. F
7
3
ln |3x + 2| + C

57. ln x2 + 7x + 3 + C

2
59. x2 + 2 ln x2 7x + 1 + 7x + C
55.
61. tan1 (2x) + C

( )
+C
63. 13 sin1 3x
4

( x+6 )
19
1
65. 5 tan
ln x2 + 12x + 61 + C
5

2
67. x2 92 ln x2 + 9 + C
69. tan1 (cos(x)) + C
71. ln | sec x + tan x| + C (integrand simplies to sec x)
73.
x2 6x + 8 + C
75. 352/15
77. 1/5
79. /2
5.
1
4
sin4 (x) + C
7.
1
6
cos6 x
3 sin7 (x)
17.
tan5 (x)
5
6
sin3 (x)
3
tan4 (x)
4
+C
21.
sec5 (x)
5
sec3 (x)
3
+C
23.
1
3
tan3
x tan x + x + C
25.
1
2
(sec x tan x ln | sec x + tan x|) + C
27.
2
5
19.
tan (x)
6
+C
+C
33. 16/15
1. T
Sec on 6.4
3. Determining which func ons in the integrand to set equal to u

and which to set equal to dv.
1. backwards
5. ex xex + C
9. x3 ex 3x2 ex + 6xex 6ex + C
11. 1/2ex (sin x cos x) + C
13.
1/13e2x (2 sin(3x)
15.
1/2 cos2
3 cos(3x)) + C
x+C

17. x tan1 (2x) 14 ln 4x2 + 1 + C
19.
1 x2 + x sin1 x + C
5.
7.
9.
11.
13.
21. x4 + 12 x2 ln |x| + 2x 2x ln |x| + C

( )
2
23. 12 x2 ln x2 x2 + C
2
15.
25. 2x + x (ln |x|) 2x ln |x| + C
17.
27. x tan(x) + ln | cos(x)| + C
19.
2)5/2 + 43 (x 2)3/2 + C
31. sec x + C
33. x csc x ln | csc x + cot x| + C
( )
( )
35. 2 sin
x 2 x cos
x +C
x
x
37. 2 xe 2e + C
(a) tan2 + 1 = sec2
3.
7. x3 cos x + 3x2 sin x + 6x cos x 6 sin x + C
39.
3 sin5 (x)
9. 19 sin9 (x) +
+
7
5
( 1
)
1
1
11. 2 8 cos(8x) 2 cos(2x) + C
(
)
1
13. 12 14 sin(4x) 10
sin(10x) + C
(
)
15. 12 sin(x) + 13 sin(3x) + C
31. 2/3
Sec on 6.2
2
(x
5
cos4 x + C
29. 32/315
81. /6
29.
1
4
(b) 9 sec2 .
(
)
1
x x2 + 1 + ln | x2 + 1 + x| + C
2
)
(
1
sin1 x + x 1 x2 + C
2
1
x x2 1 12 ln |x + x2 1| + C
2
x x2 + 1/4 + 41 ln |2 x2 + 1/4 + 2x| + C =
1
x 4x2 + 1 + 14 ln | 4x2 + 1 + 2x| + C
2
(
)
1
4 12 x x2 1/16 32
ln |4x + 4 x2 1/16| + C =
1
x 16x2 1 18 ln |4x + 16x2 1| + C
2
( )
3 sin1 x + C (Trig. Subst. is not needed)
7
x2 11 11 sec1 (x/ 11) + C
x2 3 + C (Trig. Subst. is not needed)
21. 12
x +9
+C
23.
x+2
1
18 x2 +4x+13
25.
1
7
(Trig. Subst. is not needed)

1
54
tan1
( x+2 )
+C
2
(
)
5x2
x
sin1 (x/ 5) + C
+
27. /2
29. 2 2 + 2 ln(1 + 2)

31. 9 sin1 (1/3) + 8 Note: the new lower bound is
= sin1 (1/3) and the new upper bound is = sin1 (1/3).
The nal answer comes with recognizing that
1
sin1
and that )
( (1/3) = ) sin (1/3)
(
cos sin1 (1/3) = cos sin1 (1/3) = 8/3.
9.
tanh x dx =
Let u = cosh x; du = (sinh x)dx
1
du
u
= ln |u| + C
= ln(cosh x) + C.
1. ra onal
5.
A
x 7
11. 2 sinh 2x
B
x3
A
x
13. coth x
B
x+ 7
15. x cosh x
7. 3 ln |x 2| + 4 ln |x + 5| + C
9.
11.
1
(ln |x
3
+ 2| ln |x 2|) + C
4
x+8
3 ln |x + 8| + C
17.
19.

19. 32 ln x2 + 4x + 10 + x +
25. y = x
(
tan 1 x+2
6
27. 1/2 ln(cosh(2x)) + C

+C
29. 1/2 sinh2 x + C or 1/2 cosh2 x + C
21. 2 ln |x 3| + 2 ln |x2 + 6x + 10| 4 tan1 (x + 3) + C

(
(
))
+C
23. 12 3 ln x2 + 2x + 17 4 ln |x 7| + tan 1 x+1
4
(
)

25. 12 ln x2 + 10x + 27 + 5 ln |x + 2| 6 2 tan 1 x+5

+C
2
27. 5 ln(9/4)
1
3
ln(17/2) 3.3413
31. x cosh(x) sinh(x) + C

33. cosh1 (x2 /2) + C = ln(x2 +
35.
1
16
tan1 (x/2) +
1
32
x4 4) + C
ln |x 2| +
1
32
ln |x + 2| + C
37. tan1 (ex ) + C

39. x tanh1 x + 1/2 ln |x2 1| + C
29. 1/8
41. 0
Sec on 6.6
43. 2
1. Because cosh x is always posi ve.

( x
)2 (
)2
e + ex
2
3.
coth2 x csch2 x =
ex ex
ex ex
(e2x + 2 + e2x ) (4)
e2x 2 + e2x
e2x 2 + e2x
= 2x
e 2 + e2x
=1
( x
)2
e + ex
cosh2 x =
2
=
e2x
e2x
+2+
4
1 (e2x + e2x ) + 2
=
2
2
(
)
1 e2x + e2x
=
+1
2
2
cosh 2x + 1
=
.
2
[
]
d
d
2
[sech x] =
dx
dx ex + ex
=
7.
1
1(x+5)2
23. y = 3/4(x ln 2) + 5/4
15. x + ln |x 1| ln |x + 2| + C
17. 2x + C
3
9x2 +1
21. sec x
13. ln |2x 3| + 5 ln |x 1| + 2 ln |x + 3| + C
5.
Sec on 6.5
3.
sinh x
dx
cosh x
2(ex ex )
(ex + ex )2
2(ex ex )
= x
(e + ex )(ex + ex )
2
ex ex
= x
e + ex ex + ex
= sech x tanh x
Sec on 6.7
1. 0/0, /, 0 , , 00 , 1 , 0
3. F
5. deriva ves; limits
9. 5/3
11. 2/2
13. 0
15. a/b
17. 1/2
19. 0
21.
23. 0
25. 2
27. 0
29. 0
31.
33.
35. 0
37. 1
39. 1
41. 1
43. 1
45. 1
47. 2
49.
51. 0
Sec on 6.8
27. 219,000
Sec on 7.2
1. T
3. Recall that dx does not just sit there; it is mul plied by A(x)
and represents the thickness of a small slice of the solid.
Therefore dx has units of in, giving A(x) dx the units of in3 .
5. 175/3 units3
1. The interval of integra on is nite, and the integrand is

con nuous on that interval.
3. converges; could also state < 10.
5. p > 1
7. e5 /2
7. /6 units3
9. 35/3 units3
11. 2/15 units3
13.
(b) 256/5
9. 1/3
(c) 832/15
11. 1/ ln 2
13. diverges
(d) 128/3
15.
15. 1
(a) 104/15
(b) 64/15
17. diverges
19. diverges
(a) 512/15
(c) 32/5
17.
(a) 8
(b) 8
21. diverges
23. 1
(c) 16/3
25. 0
(d) 8/3
27. 1/4
29. 1
31. diverges
33. 1/2
35. converges; Limit Comparison Test with 1/x3/2 .
37. converges; Direct Comparison Test with xex .
19. The crosssec ons of this cone are the same as the cone in
Exercise 18. Thus they have the same volume of 250/3 units3 .
21. Orient the solid so that the x-axis is parallel to long side of the
base. All crosssec ons are trapezoids (at the far le , the
trapezoid is a square; at the far right, the trapezoid has a top
length of 0, making it a triangle). The area of the trapezoid at x is
A(x) = 1/2(1/2x + 5 + 5)(5) = 5/4x + 25. The volume is
187.5 units3 .
39. converges; Direct Comparison Test with xex .
Sec on 7.3
41. diverges; Direct Comparison Test with x/(x2 + cos x).
1. T
43. converges; Limit Comparison Test with 1/ex .
3. F
Chapter 7
5. 9/2 units3
Sec on 7.1
7. 2 2 units3
9. 48 3/5 units3
1. T
11. 2 /4 units3
13.
(a) 4/5
5. 16/3
(b) 8/15
7.
9. 2 2
(d) 5/6
11. 4.5
(c) /2
15.
13. 2 /2
(c) 4/3
15. 1/6
17. On regions such as [/6, 5/6], the

area is 3 3/2. On regions
such as [/2, /6], the area is 3 3/4.
19. 5/3
21. 9/4
23. 1
25. 4
(a) 4/3
(b) /3
17.
(d) 2/3
(a) 2( 2 1)
(b) 2(1 2 + sinh1 (1))
Sec on 7.4
1. T
3.
2
5. 4/3
7. 109/2
9. 12/5
11. ln(2 3) 1.31696

13. 01 1 + 4x2 dx

1
15. 01 1 + 4x
dx
1
x2
17. 1
1 + 1x
2 dx
19. 12 1 + x14 dx
21. 1.4790
23. Simpsons Rule fails, as it requires one to divide by 0. However,
recognize the answer should be the same as for y = x2 ; why?
Sec on 7.6
3. 499.2 lb
5. 6739.2 lb
7. 3920.7 lb
9. 2496 lb
11. 602.59 lb
13.
(a) 2340 lb
(b) 5625 lb
15.
(a) 1597.44 lb
(b) 3840 lb
17.
(a) 56.42 lb
(b) 135.62 lb
25. Simpsons Rule fails.

27. 1.4058
29. 2 01 2x 5 dx = 2 5
31. 2 01 x3 1 + 9x4 dx = /27(10 10 1)

33. 2 01 1 x2 1 + x/(1 x2 ) dx = 4
Sec on 7.5
1. In SI units, it is one joule, i.e., one Newtonmeter, or kgm/s2 m.
In Imperial Units, it is lb.
3. Smaller.
5.
(a) 2450 j
(b) 1568 j
7. 735 j
9. 11,100 lb
11. 125 lb
19. 5.1
Chapter 8
Sec on 8.1
5. 2, 38 , 83 ,
7.
32 64
,
15 45
1
, 2, 81
, 512
, 15625
3
5
3
7
9. an = 3n + 1
11. an = 10 2n1
13. 1/7
15. 0
17. diverges
19. converges to 0
13. 12.5 lb
21. diverges
15. 7/20 j
23. converges to e
17. 45 lb
25. converges to 0
19. 953, 284 j
27. converges to 2
21. 192,767 lb. Note that the tank is oriented horizontally. Let the
origin be the center of one of the circular ends of the tank. Since
the radius is 3.75 , the uid is being pumped to y = 4.75; thus
the distance the gas travels is h(y) = 4.75 y. A dieren al
element
of water is a rectangle, with length 20 and width
2 3.752 y2 . Thusthe force required to move that slab of gas is

F(y) = 40 45.93 3.752
y2 dy. Total work is
3.75
3.752 y2 dy. This can be
40
45.93
(4.75
y)
3.75
evaluated
without
actual
integra
on; split theintegral into
3.75
3.75
40 45.93 (4.75) 3.752 y2 dy + 3.75
40 45.93
3.75
2
2
(y) 3.75 y dy. The rst integral can be evaluated as
measuring half the area of a circle; the la er integral can be
shown to be 0 without much diculty. (Use subs tu on and
realize the bounds are both 0.)
29. bounded
23.
(a) approx. 577,000 j

(b) approx. 399,000 j
(c) approx 110,000 j (By volume, half of the water is between
the base of the cone and a height of 3.9685 m. If one
rounds this to 4 m, the work is approx 104,000 j.)
25. 617,400 j
31. bounded
33. neither bounded above or below
35. monotonically increasing
37. never monotonic
39. Let {an } be given such that lim |an | = 0. By the deni on of
n
the limit of a sequence, given any > 0, there is a m such that for
all n > m, | |an | 0| < . Since | |an | 0| = |an 0|, this
directly implies that for all n > m, |an 0| < , meaning that
lim an = 0.
n
41. Le to reader
Sec on 8.2
3. One sequence is the sequence of terms {a} . The other is the
sequence of nth par al sums, {Sn } = { ni=1 ai }.

5. F
7.
(a) 1, 54 ,
49 205 5269
,
,
36 144 3600
(c) If the odd series converges, the work done in (a) shows the
even series converges also. (The sequence of the nth
par al sum of the even series is bounded and
monotonically increasing.) Likewise, (b) shows that if the
even series converges, the odd series will, too. Thus if
either series converges, the other does.
Similarly, (a) and (b) can be used to show that if either
series diverges, the other does, too.
(b) Plot omi ed

9.
(a) 1, 3, 6, 10, 15
(b) Plot omi ed
11.
(a)
1 4 13 40 121
, , , ,
3 9 27 81 243
(b) Plot omi ed

13.
(d) If both the even and odd series converge, then their sum
would be a convergent series. This would imply that the
Harmonic Series, their sum, is convergent. It is not. Hence
each series diverges.
(a) 0.1, 0.11, 0.111, 0.1111, 0.11111

(b) Plot omi ed
15.
17.
19.
lim an = ; by Theorem 63 the series diverges.
Sec on 8.3
lim an = 1; by Theorem 63 the series diverges.
1. con nuous, posi ve and decreasing
lim an = e; by Theorem 63 the series diverges.
21. Converges
3. The Integral Test (we do not have a con nuous deni on of n!

yet) and the Limit Comparison Test (same as above, hence we
cannot take its deriva ve).
23. Converges
5. Converges
25. Converges
7. Diverges
9. Converges
27. Converges
11. Converges
29. Diverges
(
31.
(a) Sn =
n(n+1)
2
)2
13. Converges; compare to

all n > 1.
(b) Diverges
33.
(a) Sn = 5
11/2
1/2
15. Diverges; compare to
1
, as 1/n ln n/n for all n 2.
n
n=1
1
. Since n = n2 > n2 1,
n
n=1
(b) Converges to 10.

35.
(a) Sn =
1(1/3)n
4/3
1/n 1/ n2 1 for all n 2.
(b) Converges to 3/4.

37.
39.
(a) With par al frac ons, an = 32

(
)
1
1
n+2
.
Sn = 32 32 n+1
1
n
1
n+2
)
. Thus
(b) Converges to 9/4

(
)
(a) Sn = ln 1/(n + 1)
1
(a) an = n(n+3)
; using par al frac ons, the resul ng
telescoping
( sum reduces to
)
1
1
1
Sn = 13 1 + 12 + 13 n+1
n+2
n+3
(a) With par al frac ons, an =

(
)
1
.
Sn = 12 3/2 n1 n+1
1
2
1
n1
1
n+1
)
. Thus

45.
1
:
n
n=1
for all n 1.

43.
n2
1
n2 + n + 1
n2 + n + 1
= 3 <
,
<
3
n
n
n
n3 5
(b) Diverges (to ).

41.
1
, as 1/(n2 + 3n 5) 1/n2 for
2
n
n=1
(a) The nth par al sum of the odd series is

1
. The nth par al sum of the even
1 + 13 + 15 + + 2n1
1
series is 12 + 14 + 16 + + 2n
. Each term of the even
series is less than the corresponding term of the odd
series, giving us our result.
(b) The nth par al sum of the odd series is

1
1 + 13 + 15 + + 2n1
. The nth par al sum of 1 plus the
1
even series is 1 + 12 + 14 + + 2(n1)
. Each term of the
even series is now greater than or equal to the
corresponding term of the odd series, with equality only on
the rst term. This gives us the result.
1
. Note that
n
n=1
n
n2
1
1
= 2
> ,
n2 1
n 1 n
n
as
n2
n2 1
> 1, for all n 2.
1
.
2
n
n=1
ln n
.
n
n=1
1
.
n
n=1
1
sin n
. Just as lim
= 1,
n0 n
n
n=1
lim
sin(1/n)
= 1.
1/n
1
.
3/2
n
n=1
33. Converges; Integral Test
35. Diverges; the nth Term Test and Direct Comparison Test can be
used.
37. Converges; the Direct Comparison Test can be used with sequence
1/3n .
39. Diverges; the nth Term Test can be used, along with the Integral
Test.
41.
(a) Converges; use Direct Comparison Test as
an
n
11.
(c) n/a; diverges

13.
(c) n/a; diverges

15.
(c) absolute
17.
(a) converges
(b) converges (Ra o Test)
(e) Does not converge, using logic from (b) and nth Term Test.
Sec on 8.4
(a) converges
(b) converges (Geometric Series with r = 2/3)
(c) Converges; similar logic to part (b) so (an )2 < an .

(d) May converge; certainly nan > an but that does not mean
it does not converge.
(a) diverges (terms oscillate between 1)

(b) diverges
< n.
(b) Converges; since original series converges, we know

limn an = 0. Thus for large n, an an+1 < an .
(a) diverges (limit of terms is not 0)

(b) diverges
(c) absolute
19.
(a) converges
(b) diverges (p-Series Test with p = 1/2)
1. algebraic, or polynomial.
(c) condi onal
3. Integral Test, Limit Comparison Test, and Root Test

5. Converges
7. Converges
9. The Ra o Test is inconclusive; the p-Series Test states it diverges.
11. Converges
13. Converges; note the summa on can be rewri en as
from which the Ra o Test can be applied.
2n n!
,
3n n!
n=1
21. S5 = 1.1906; S6 = 0.6767;
(1)n
1.1906
0.6767
ln(n
+ 1)
n=1
23. S6 = 0.3681; S7 = 0.3679;
(1)n
0.3679
0.3681
n!
n=0
25. n = 5
27. Using the theorem, we nd n = 499 guarantees the sum is within
0.001 of /4. (Convergence is actually faster, as the sum is within
of /24 when n 249.)
15. Converges
17. Converges
Sec on 8.6
19. Diverges
nth -Term Test
21. Diverges. The Root Test is inconclusive, but the

shows divergence. (The terms of the sequence approach e2 , not
0, as n .)
1. 1
23. Converges
5. 1 + 2x + 4x2 + 8x3 + 16x4
25. Diverges; Limit Comparison Test
7. 1 + x +
27. Converges; Ra o Test or Limit Comparison Test with 1/3n .
9.
3. 5
29. Diverges; nth -Term Test or Limit Comparison Test with 1.

31. Diverges; Direct Comparison Test with 1/n
x3
6
x4
24
(a) R =
(b) (, )
11.
33. Converges; Root Test
Sec on 8.5
x2
2
(a) R = 1
(b) (2, 4]
13.
(a) R = 2
(b) (2, 2)
1. The signs of the terms do not alternate; in the given series, some
terms are nega ve and the others posi ve, but they do not
necessarily alternate.
15.
3. Many examples exist; one common example is an = (1)n /n.
17.
5.
(b) (4/5, 6/5)
19.
(a) diverges (limit of terms is not 0)
21.
(b) diverges
(c) n/a; diverges
9.
(c) condi onal
(a) R = 1
(b) [1, 1]
23.
(a) converges
(b) diverges (Limit Comparison Test with 1/n)
(a) R =
(b) (, )
(c) absolute
7.
(a) R = 1
(b) (1, 1)
(a) converges
(b) converges (p-Series)
(a) R = 1/5
(a) R = 0
(b) x = 0
25.
(a) f (x) =
n=1
n2 xn1 ;
(1, 1)

(b)
f(x) dx = C +
n n+1
x
;
n+1
n=0
27.
(a) f (x) =
(b)
n n1
x
;
n
2
n=1
f(x) dx = C +
3. All deriva ves of ex are ex which evaluate to 1 at x = 0.

1 3
1 4
The Taylor series starts 1 + x + 12 x2 + 3!
x + 4!
x + ;
xn
the Taylor series is
n!
n=0
(2, 2)
1
xn+1 ;
(n + 1)2n
n=0
29.
1. A Taylor polynomial is a polynomial, containing a nite number of

terms. A Taylor series is a series, the summa on of an innite
number of terms.
(1, 1)
[2, 2)
(1)n x2n1
(1)n+1 x2n+1
=
;
(2n 1)!
(2n + 1)!
n=1
n=0
(, )
(1)n x2n+1
f(x) dx = C +
(b)
; (, )
(2n + 1)!
n=0
5. The nth deriva ve of 1/(1 x) is f (n) (x) = (n)!/(1 x)n+1 ,

which evaluates to n! at x = 0.
The Taylor series starts 1 + x + x2 + x3 + ;

xn
(a) f (x) =
31. 1 + 3x + 92 x2 + 92 x3 +
n=0
7. The Taylor series starts

0 (x /2) + 0x2 + 61 (x /2)3 + 0x4
(x /2)2n+1
(1)n+1
(2n + 1)!
n=0
27 4
x
8
33. 1 + x + x2 + x3 + x4
35. 0 + x + 0x2 16 x3 + 0x4
Sec on 8.7
1. The Maclaurin polynomial is a special case of Taylor polynomials.
Taylor polynomials are centered at a specic x-value; when that
x-value is 0, it is a Maclauring polynomial.
3. p2 (x) = 6 + 3x 4x2 .
7. p8 (x) = x + x2 + 12 x3 + 16 x4 +
2x4
3
4x3
3
+ 2x2 + 2x + 1
1
5
(1+x)3 128
(1+x)4
13. p4 (x) = 1+ 12 (1+x) 81 (1+x)2 + 16
( 4 +x
)5
1
2
120 2
4 +x
)6
( 4 +x
720 2
( 4+x)2
2
( 4+x)3
6
4
+x)
( 4
24
17. p5 (x) =
1
1
1
1
x2
+ 18 (x2)2 16
(x2)3 + 32
(x2)4 64
(x2)5
2
4
19. p3 (x) =
1
2
1+x
2
9. f (n) (x) = (1)n ex ; at x = 0, f (n) (0) = 1 when n is odd and

f (n) (0) = 1 when n is even.
1 3
The Taylor series starts 1 x + 12 x2 3!
x + ;
x
(1)n .
n!
n=0
The Taylor series starts
1
1
+ 41 (x 1) 81 (x 1)2 + 16
(x 1)3 ;
2
(x 1)n
(1)n+1 n+1 .
2
n=0
1 5
x
24
11. p4 (x) = x4 x3 + x2 x + 1
15. p6 (x) =
+ 14 (1 + x)2
13. Given a value x, the magnitude of the error term Rn (x) is bounded
by

max f (n+1) (z) (n+1)
x
,
Rn (x)
(n + 1)!
where z is between 0 and x.
If x > 0, then z < x and f (n+1) (z) = ez < ex . If x < 0, then
x < z < 0 and f (n+1) (z) = ez < 1. So given a xed x value, let
M = max{ex , 1}; f (n) (z) < M. This allows us to state

Rn (x)
21. p3 (x) = x x6 ; p3 (0.1) = 0.09983. Error is bounded by

1
4!
0.14 0.000004167.
1
23. p2 (x) = 3 + 16 (9 + x) 216
(9 + x)2 ; p2 (10) = 3.16204.
The third deriva ve of f(x) = x is bounded on (8, 11) by 0.003.

13 = 0.0005.
Error is bounded by 0.003
3!
25. The nth deriva ve of f(x) = ex is bounded by 3 on intervals

3
containing 0 and 1. Thus |Rn (1)| (n+1)!
1(n+1) . When n = 7,
this is less than 0.0001.
27. The nth deriva ve of f(x) = cos x is bounded by 1 on intervals
1
containing 0 and /3. Thus |Rn (/3)| (n+1)!
(/3)(n+1) .
When n = 7, this is less than 0.0001. Since the Maclaurin
polynomial of cos x only uses even powers, we can actually just
use n = 6.
29. The nth term is
1 n
x .
n!
31. The nth term is xn .

33. The nth term is (1)n
35. 3 + 15x +
Sec on 8.8
/2)5 ;
n!
(n) (1) = (1)n+1 n!
11. f (n) (x) = (1)n+1 (x+1)
n+1 ; at x = 1, f
2n+1
5. p3 (x) = 1 x + 12 x3 16 x3
9. p4 (x) =
1
(x
120
(x1)n
.
n
375 3
1875 4
75 2
x +
x +
x
2
6
24
M (n+1)
.
x
(n + 1)!
(n+1)
x
= 0. Thus by the Squeeze
M
(n + 1)!
Theorem, we conclude that lim Rn (x) = 0 for all x, and hence
For any x, lim
ex =
n=0
xn
n!
for all x.
15. Given a value x, the magnitude of the error term Rn (x) is bounded
by

max f (n+1) (z)
Rn (x)
(x 1)(n+1) ,
(n + 1)!
where z is between 1 and x.

n!
Note that f (n+1) (x) = xn+1
.
We consider the cases when x > 1 and when x < 1 separately.
n!
If x > 1, then 1 < z < x and f (n+1) (z) = zn+1
< n!. Thus

Rn (x)

n!
(x 1)n+1
(x 1)(n+1) =
.
(n + 1)!
n+1
For a xed x,
lim
(x 1)n+1
= 0.
n+1
If 0 < x < 1, then x < z < 1 and f (n+1) (z) =

Thus
n!
zn+1
<
n!
.
xn+1
9. x = y2
1 2
11. x = 12
y
n+1
n+1

Rn (x) n!/x
(x 1)(n+1) = x
(1 x)n+1 .
(n + 1)!
n+1
13. x = 18 (y 3)2 + 2
Since 0 < x < 1, xn+1 < 1 and (1 x)n+1 < 1. We can then
extend the inequality from above to state
15. focus: (5, 2); directrix: x = 1. The point P is 10 units from each.
y
n+1

1
Rn (x) x
(1 x)n+1 <
.
n+1
n+1
x
5
As n , 1/(n + 1) 0. Thus by the Squeeze Theorem, we

conclude that lim Rn (x) = 0 for all x, and hence
ln x =
(x 1)n
(1)n+1
n
n=1
for all 0 < x 2.
x2n
(1)n
,
(2n)!
n=0
17. Given cos x =
(x)2n
x2n
cos(x) =
(1)n
=
(1)n
= cos x, as all
(2n)!
(2n)!
n=0
n=0
powers in the series are even.
19. Given sin x =

(1)n
n=0
x2n+1
,
(2n + 1)!
(
)
)
x2n+1
d(
d
(1)n
=
sin x =
dx
dx n=0
(2n + 1)!
(2n + 1)x2n
x2n
(1)n
=
(1)n
= cos x. (The
(2n
+
1)!
(2n)!
n=0
n=0
summa on s ll starts at n = 0 as there was no constant term in
the expansion of sin x).
21. 1 +
19.
(x1)
1/4
21.
n=0
(x2)2
25
(y3)2
16
=1
23.
(x+1)2
9
(y1)2
25
=1
25.
x2
3
27.
(x2)2
4
29. x2
31.
0.8877
(x1)2
9
(
x2
x6
6
x10
120
x14
5040
.
dx =
33.
35.
x2
4
37.
(x3)2
16
39.
x2
4
y2
5
=1
y2
3
(y3)2
9
x2
10
=1
(a) Solve for c in e = c/a: c = ae. Thus a2 e2 = a2 b2 , and

b2 = a2 a2 e2 . The result follows.
(b) Mercury: x2 /(0.387)2 + y2 /(0.3787)2 = 1
Earth: x2 + y2 /(0.99986)2 = 1
Mars: x2 /(1.524)2 + y2 /(1.517)2 = 1
(c) Mercury: (x 0.08)2 /(0.387)2 + y2 /(0.3787)2 = 1
Earth: (x 0.0167)2 + y2 /(0.99986)2 = 1
Mars: (x 0.1423)2 /(1.524)2 + y2 /(1.517)2 = 1
3. Hyperbola
5. With a horizontal transverse axis, the x2 term has a posi ve
coecient; with a ver cal transverse axis, the y2 term has a
posi ve coecient.
=1
=1
41. (y 2)2
=1
1. When dening the conics as the intersec ons of a plane and a

double napped cone, degenerate conics are created when the
plane intersects the ps of the cones (usually taken as the origin).
Nondegenerate conics are formed when this plane does not
contain the origin.
=1
43.
7. y =
=1
Sec on 9.1
1)2
(y2)2
4
Chapter 9
1
(x
12
y2
3
(y3)2
4
=1
y2
5
(2x + 3)2n+1
.
(2n + 1)!
x5
x3
29. x + x +
3
30

( 2)
sin x dx
31.
= 1; foci at (1, 8.75); e = 8.75/3 0.99
y
9
x2
x3
5x4
x
2
8
16
128
(1)n
x2
5x3
10x4
x
+
23. 1 +
3
9
81
243
2 )2n
x4n
(x
=
(1)n
.
25.
(1)n
(2n)!
(2n)!
n=0
n=0
27.
17. .
Sec on 9.2
1. T
3. rectangular
x
5
10
10
5.
15.
x
5
10
10
..
y
y
1
2
17.
7.
x
1
1
x
1
19.
(a)
(b)
(c)
(d)
21. x2 y2 = 1
Traces a circle of radius 1 counterclockwise once.

Traces a circle of radius 1 counterclockwise over 6 mes.
Traces a circle of radius 1 clockwise innite mes.
Traces an arc of a circle of radius 1, from an angle of -1
radians to 1 radian, twice.
23. y = x3/2
9.
25. y = x3 3
27. y2 x2 = 1
29. x = 1 2y2
31. x2 + y2 = r2 ; circle centered at (0, 0) with radius r.
10
x
5
33.
10
y
5
(xh)2
a2
(yk)2
b2 = 1; hyperbola centered at (h, k) with

horizontal transverse axis and asymptotes with slope b/a. The
parametric equa ons only give half of the hyperbola. When
a > 0, the right half; when a < 0, the le half.
35. x = ln t, y = t. At t = 1, x = 0, y = 1.
y = ex ; when x = 0, y = 1.
11.
x
5
37. x = 1/(4t2 ), y = 1/(2t). At t = 1, x = 1/4, y = 1/2.
y = 1/(2 x); when x = 1/4, y = 1.

39. t = 1, 2
41. t = /6, /2, 5/6
43. t = 2
45. t = . . . 0, 2, 4, . . .
47. x = 50t, y = 16t2 + 64t

49. x = 2 cos t, y = 2 sin t; other answers possible
51. x = cos t + 1, y = 3 sin t + 3; other answers possible
53. x = sec t + 2, y = 8 tan t 3; other answers possible
1
0.5
Sec on 9.3
13.
x
1
0.5
0.5
0.5
1. F
3. F
5.
= 2t
(a) dy
dx
(b) Tangent line: y = 2(x 1) + 1; normal line:
y = 1/2(x 1) + 1
7.
dy
dx
(a)
2t+1
2t1
(b) Tangent line: y = 3x + 2; normal line: y = 1/3x + 2

9.
dy
dx
(a)
= csc t
(b) t = /4:
Tangent
line: y =
y = 1/ 2(x 2) + 1
11.
dy
dx
(a)
2(x
cos t sin(2t)+sin t cos(2t)

sin t sin(2t)+2 cos t cos(2t)
(b) Tangent line: y = x
11.
2) + 1; normal line:
2; normal line: y = x
x
1
y
2
13. t = 0
15. t = 1/2
17. The graph does not have a horizontal tangent line.
13.
21. t0 = 0; limt0
dy
dx
= 0.
23. t0 = 1; limt1
dy
dx
= .
25.
27.
d2 y
dx2
29.
31.
= 2; always concave up
x
2
19. The solu on is non-trivial; use iden es sin(2t) = 2 sin t cos t and
cos(2t) = cos2 t sin2 t to rewrite
g (t) = 2 sin t(2 cos2 t sin2 t). On [0, 2], sin t = 0 when
2
2
t = 0, , 2,
sin
t = 0 when 1
and 2 cos t
1
1
t = tan ( 2), tan ( 2), 2 tan ( 2).
d2 y
dx2
2
1
2
y
2
15.
4
= (2t1)
3 ; concave up on (, 1/2); concave down on
(1/2, ).
2
2
d2 y
dx2
= cot3 t; concave up on (, 0); concave down on

(0, ).
d2 y
dx2
4(13+3 cos(4t))
, obtained with a computer algebra system;
(cos t+3 cos(3t))3
(
)
concave up on tan1 ( 2/2), tan1 ( 2/2) , concave down
(
) (
)
on /2, tan1 ( 2/2) tan1 ( 2/2), /2
17.
x
2
33. L = 6
35. L = 2 34
37. L 2.4416 (actual value: L = 2.42211)
y
1
39. L 4.19216 (actual value: L = 4.18308)

41. The answer is 16 for both (of course), but the integrals are
dierent.
19.
x
1
43. SA 8.50101 (actual value SA = 8.02851
Sec on 9.4
1
3. T
5.
.
1
A
2
21.
7. A = P(2.5, /4) and P(2.5, 5/4);

B = P(1, 5/6) and P(1, 11/6);
C = P(3, 4/3) and P(3, /3);
D = P(1.5, 2/3) and P(1.5, 5/3);

9. A = ( 2, 2)
B = ( 2, 2)
C = P( 5, 0.46)
D = P( 5, 2.68)
x
1
3
2
23.
1
x
(b) tangent line:

y = 53(x + 3/4) 3/4; normal line:
y = 1/5 3(x + 3/4) 3/4
11. horizontal: = /2, 3/2;

25.
ver cal: = 0, , 2
x
8
13. horizontal: = tan1 (1/ 5), /2, tan1 (1/ 5), +

tan1 (1/ 5), 3/2, 2 tan1 (1/ 5);
1
ver cal: = 0, tan1 ( 5),
tan ( 5), , +
1
1
tan ( 5), 2 tan ( 5)
.
y
4
15. In polar: = 0
= =
In rectangular: y = 0
27.
17. area = 4
x
5
5
2
19. area = /12
21. area = 3 3/2
23. area = + 3 3
/3
/6
1
1 2
1
25. area =
sin (3) d
cos2 (3) d =
+
2
2
12
24
/12
/12
y
4
2
29.
x
5
/2
27. area =
2
4
5/12
1
(1 cos )2 d +
2
0
1
(2 6 2 2) 0.105
4
5/12
31. x2 + (y + 2)2 = 4
29. 4
33. y = 2/5x + 7/5
31. L 2.2592; (actual value L = 2.22748)
35. y = 4
33. SA = 16
37. x2 + y2 = 4
35. SA = 32/5
39. = /4
37. SA = 36
41. r = 5 sec
43. r = cos / sin2
45. r = 7
47. P( 3/2, /6), P(0, /2), P( 3/2, 5/6)
49. P(0, 0) = P(0, /2), P( 2, /4)
51. P( 2/2, /12), P( 2/2, 5/12), P( 2/2, 3/4)

53. For all points, r = 1; =
/12, 5/12, 7/12, 11/12, 13/12, 17/12, 19/12, 23/12.
55. Answers will vary. If m and n do not have any common factors,
then an interval of 2n is needed to sketch the en re graph.
Sec on 9.5
(a)
dy
dx
= cot
(b) tangent line: y = (x

y=x
5.
(a)
Chapter 10
Sec on 10.1
1. right hand
3. curve (a parabola); surface (a cylinder)
5. a hyperboloid of two sheets
7. || AB || = 6; || BC || = 17; || AC || = 11. Yes, it is a right

2
2
2
triangle as || AB || + || AC || = || BC || .
9. Center at (4, 1, 0); radius = 3
1. Using x = r cos and y = r sin , we can write x = f() cos ,

y = f() sin .
3.
dy
dx
dy
dx
2/2) +
11. Interior of a sphere with radius 1 centered at the origin.

13. The rst octant of space; all points (x, y, z) where each of x, y and
z are posi ve. (Analogous to the rst quadrant in the plane.)
2/2; normal line:

z
cos (1+2 sin )

cos2 sin (1+sin )
(b) tangent line: x = 3 3/4; normal line: y = 3/4

7.
(a)
cos +sin
cos sin
15.
x 1
(b) tangent line: y = 2/x + /2; normal line:

y = /2x + /2
9.
(a)
dy
dx
1
(3 cos )2 d =
2
4 sin(t) cos(4t)+sin(4t) cos(t)

4 cos(t) cos(4t)sin(t) sin(4t)
1
y
(c) x = 1, 0, 3.
y
2
17.
2
2.
2
u + v
2
2
13.
x
v
.
19. y2 + z2 = x4
21. z = ( x2 + y2 )2 = x2 + y2
Sketch of u v shi ed for clarity.
z2
23. (a)
x = y2 +
25. (b)
y2
z2
x +
+
=1
9
4
9
z
u + v
15.
z
u v
1
x
1
27.
x 1
17. || u || =
17, || v || = 3, || u + v || = 14, || u v || = 26
19. || u || = 7, || v || = 35, || u + v || = 42, || u v || = 28

21. u = 3/ 30, 7/ 30
23. u = 1/3, 2/3, 2/3
25. u = cos 50 , sin 50 0.643, 0.766.
27.
|| u || =
1
x
1
.y
sin2 cos2 + sin2 sin2 + cos2

sin2 (cos2 + sin2 ) + cos2
sin2 + cos2
= 1.
29. The force on each chain is 100lb.

31. The force on each chain is 50lb.
33. = 5.71 ; the weight is li ed 0.005
1
5
x
Sec on 10.3
5
5
(about 1/16th of an inch).
35. = 84.29 ; the weight is li ed 9 .
31.
29.
1. Scalar
3. By considering the sign of the dot product of the two vectors. If
the dot product is posi ve, the angle is acute; if the dot product is
nega ve, the angle is obtuse.
.
Sec on 10.2
5. 22
7. 3
9. not dened
3. A vector with magnitude 1.
5. It stretches the vector by a factor of 2, and points it in the

opposite direc on.
#
7. PQ = 4, 4 = 4i + 4j
#
9. PQ = 2, 2, 0 = 2i + 2j
13. = 0.3218 18.43
11.
(a) u + v
= 3, 2,1; u v = 1,
0, 3;
u 2v = 2 2, 2, 2 2 .
15. = /4 = 45
17. Answers will vary; two possible answers are 7, 4 and 14, 8.
19. Answers will vary; two possible answers are 1, 0, 1 and
4, 5, 9.
21. projv u = 1/2, 3/2.
23. projv u = 1/2, 1/2.

25. projv u = 1, 2, 3.
27. u = 1/2, 3/2 + 3/2, 1/2.
29. u = 1/2, 1/2 + 5/2, 5/2.
31. u = 1, 2, 3 + 0, 3, 2.
33. 1.96lb
11. Answers can vary; here the direc on is given by d1 d2 : vector:

(t) = 7, 2, 1 + t 1, 1, 2
parametric: x = 7 + t, y = 2 t, z = 1 + 2t
symmetric: x 7 = 2 y = (z + 1)/2
13. vector: (t) = 1, 1 + t 2, 3
parametric: x = 1 + 2t, y = 1 + 3t
symmetric: (x 1)/2 = (y 1)/3
15. parallel
35. 141.42 lb
17. intersec ng; 1 (3) = 2 (4) = 9, 5, 13
37. 500 lb
19. skew
39. 500 lb
21. same
23.
41/3
25. 5 2/2
27. 3/ 2
Sec on 10.4
1. vector
3. Perpendicular is one answer.
5. Torque
7. u v = 11, 1, 17
9. u v = 47, 36, 44
11. u v = 0, 0, 0
13. i k = j
17. 5
#
29. Since both P and Q are on the line, PQ is parallel to d. Thus
#
PQ d = 0, giving a distance of 0.
31. The distance formula cannot be used because since d1 and d2 are
parallel,c is 0 and we cannot divide by || 0 ||.
#
Since d1 and d2 are parallel, P1 P2 lies in the plane formed by the
#
two lines. Thus P1 P2 d2 is orthogonal to this plane, and
#
c = (P1 P2 d2 ) d2 is parallel to the plane, but s ll orthogonal
#
to both d1 and d2 . We desire the length of the projec on of P1 P2
ontoc, which is what the formula provides.
Sec on 10.6
19. 0
21.
14
1. A point in the plane and a normal vector (i.e., a direc on

orthogonal to the plane).
23. 3
25. 5 2/2

27. 1
7. Standard form: 3(x 2) (y 3) + 7(z 4) = 0

general form: 3x y + 7z = 31
29. 7
31. 2
33. 1 1, 1, 2
6
35. 0, 1, 0
37. 87.5 lb
39. 200/3 66.67 lb
9. Answers may vary;

Standard form: 8(x 1) + 4(y 2) 4(z 3) = 0
general form: 8x + 4y 4z = 4
Standard form: 7(x 2) + 2(y 1) + (z 2) = 0
general form: 7x + 2y + z = 10

Standard form: 2(x 1) (y 1) = 0
u (u v) = u1 , u2 , u3 (u2 v3 u3 v2 , (u1 v3 u3 v1 ), u1 v2 u2 v1 ) general form: 2x y = 1
41. With u = u1 , u2 , u3 and v = v1 , v2 , v3 , we have
= u1 (u2 v3 u3 v2 ) u2(u1 v3 u3 v1 ) + u3(u1 v2 u2 v1 )15. Answers may vary;

Standard form: 2(x 2) (y + 6) 4(z 1) = 0
general form: 2x y 4z = 6
= 0.
Sec on 10.5
1. A point on the line and the direc on of the line.
3. parallel, skew
5. vector: (t) = 2, 4, 1 + t 9, 2, 5
parametric: x = 2 + 9t, y = 4 + 2t, z = 1 + 5t
symmetric: (x 2)/9 = (y + 4)/2 = (z 1)/5
7. Answers can vary: vector: (t) = 2, 1, 5 + t 5, 3, 1
parametric: x = 2 + 5t, y = 1 3t, z = 5 t
symmetric: (x 2)/5 = (y 1)/3 = (z 5)
9. Answers can vary; here the direc on is given by d1 d2 : vector:
(t) = 0, 1, 2 + t 10, 43, 9
parametric: x = 10t, y = 1 + 43t, z = 2 + 9t
symmetric: x/10 = (y 1)/43 = (z 2)/9

Standard form: (x 5) + (y 7) + (z 3) = 0
general form: x + y + z = 15
Standard form: 3(x + 4) + 8(y 7) 10(z 2) = 0
general form: 3x + 8y 10z = 24
21. Answers
may vary:
x = 14t
=
y = 1 10t
z = 2 8t
23. (3, 7, 5)
25. No point of intersec on; the plane and line are parallel.
27.
5/7

29. 1/ 3
#
31. If P is any point in the plane, and Q is also in the plane, then PQ
lies parallel to the plane and is orthogonal to n, the normal vector.
#
Thus n PQ = 0, giving the distance as 0.
13.
Chapter 11
2
2
Sec on 11.1
1. parametric equa ons

3. displacement
y
1
15.
x 1
5.
x
1
.y
17. ||r(t) || =
25 cos2 t + 9 sin2 t.
19. ||r(t) || = cos2 t + t2 + t4 .
21. Answers may vary; three solu ons are

r(t) = 3 sin t + 5, 3 cos t + 5,
r(t) = 3 cos t + 5, 3 sin t + 5 and
r(t) = 3 cos t + 5, 3 sin t + 5.
y
1
0.5
7.
x
4
2
0.5
25. Answers may vary, though most direct solu ons are
r(t) = t, 1/2(t 1) + 5,
r(t) = t + 1, 1/2t + 5,
r(t) = 2t + 1, t + 5 and
r(t) = 2t + 1, t + 5.
1
.
y
2
27. Answers may vary, though most direct solu on is

r(t) = 3 cos(4t), 3 sin(4t), 3t.
29. 1, 1
9.
31. 1, 2, 7
x
3
23. Answers may vary, though most direct solu ons are
r(t) = 3 cos t + 3, 2 sin t 2,
r(t) = 3 cos t + 3, 2 sin t 2 and
r(t) = 3 sin t + 3, 2 cos t 2.
Sec on 11.2
1. component
3. It is dicult to iden fy the points on the graphs ofr(t) andr (t)
that correspond to each other.
y
5
5.
3
e ,0
7. 2t, 1, 0
11.
x
10
10
9. (0, )
11. r (t) = 1/t2 , 5/(3t + 1)2 , sec2 t
13. r (t) = 2t, 1 sin t, 2t + 5 + t2 + 1, t 1 cos t, 2 =

2
(t + 1) cos t + 2t sin t + 4t + 3
6
v(/4)
1.5
4
1
15.
2
r (1)
0.5
x
a(/4)
r (t) = 2t + 1, 2t 1
0.5
1.5
13. v(t) = 2t + 1, 2t + 2, a(t) = 2, 2
r (1)
2
v(/4)
x
2
17.
2
4
a(/4)
4
4
6
r (t) = 2t, 3t2 1

19. (t) = 2, 0 + t 3, 1
21. (t) = 3, 0, + t 0, 3, 1
23. t = 0
25. r(t) is not smooth at t = 3/4 + n, where n is an integer
27. Both deriva ves return 5t4 , 4t3 3t2 , 3t2 .
15. || v(t) || = 4t2 + 1.

Min at t = 0; Max at t = 1.
17. || v(t) || = 5.
Speed is constant, so there is no dierence between min/max
19. || v(t) || = | sec t| tan2 t + sec2 t.

min: t = 0; max: t = /4
29. Both
deriva ves return
2t et 1, cos t 3t2 , (t2 + 2t)et (t 1) cos t sin t .
31. tan1 t, tan t + C
21. || v(t) || = 13.

speed is constant, so there is no dierence between min/max
23. || v(t) || = 4t2 + 1 + t2 /(1 t2 ).

min: t = 0; max: there is no max; speed approaches as
t 1
33. 4, 4
25.
35. r(t) = ln |t + 1| + 1, ln | cos t| + 2

37. r(t) = cos t + 1, t sin t, et t 1
39. 10
41.
2(1 e1 )
Sec on 11.3
(a) r1 (1) = 1, 1;r2 (1) = 1, 1
(b) v1 (1) = 1, 2; || v1 (1) || =

5; a1 (1) = 0, 2
v2 (1) = 2, 4; || v2 (1) || = 2 5; a2 (1) = 2, 12
(a) r1 (2) = 6, 4;r2 (2) = 6, 4
(b) v1 (2) = 3, 2; || v1 (2) || =

13; a1 (2) = 0, 0
v2 (2) = 6, 4; || v2 (2) || = 2 13; a2 (2) = 0, 0
29. v(t) = 2t + 1, 3t + 2,r(t) = t2 + t + 5, 3t2 /2 + 2t 2

27.
31. v(t) = sin t, cos t,r(t) = 1 cos t, sin t

1. Velocity is a vector, indica ng an objects direc on of travel and its
rate of distance change (i.e., its speed). Speed is a scalar.
33. Displacement: 0, 0, 6; distance traveled:2 13 22.65 ;

average velocity: 0, 0, 3; average speed: 13 3.61 /s
3. The average velocity is found by dividing the displacement by the

me traveled it is a vector. The average speed is found by
dividing the distance traveled by the me traveled it is a scalar.
35. Displacement: 0, 0; distance traveled: 2 6.28 ; average

velocity: 0, 0; average speed: 1 /s
5. One example is traveling at a constant speed s in a circle, ending

at the star ng posi on. Since the displacement is 0, the average
velocity is 0, hence || 0 || = 0. But traveling at constant speed s
means the average speed is also s > 0.
39.
7. v(t) = 2, 5, 0, a(t) = 0, 0, 0
9. v(t) = sin t, cos t, a(t) = cos t, sin t
11. v(t) = 1, cos t, a(t) = 0, sin t
37. At t-values of sin1 (9/30)/(4) + n/2 0.024 + n/2 seconds,

where n is an integer.
(a) Holding the crossbow at an angle of 0.013 radians,
0.745 will hit the target 0.4s later. (Another solu on
exists, with an angle of 89 , landing 18.75s later, but this is
imprac cal.)
(b) In the .4 seconds the arrow travels, a deer, traveling at
20mph or 29.33 /s, can travel 11.7 . So she needs to lead
the deer by 11.7 .
41. The posi on func on isr(t) = 220t, 16t2 + 1000 . The

y-component is 0 when t = 7.9;r(7.9) = 1739.25, 0, meaning
the box will travel about 1740 horizontally before it lands.
Sec on 11.4
5.
7. s = 3t, sor(s) = 2s/3, s/3, 2s/3
9. s = 13t, sor(s) = 3 cos(s/ 13), 3 sin(s/ 13), 2s/ 13

11. =
1. 1
(0) = 0, (1/2) =
3. T(t) and
N(t).
5. T(t) =
4t
,
20t2 4t+1
2t1
20t2 4t+1

; T(1) = 4/ 17, 1/ 17
cos t sin t
cos2 t sin2 t
cos t, sin t. (Be careful; this cannot be
simplied as just cos t, sin t as cos2

t sin2 t = cos t sin t, but
rather | cos t sin t|.) T(/4) = 2/2, 2/2

9. (t) = 2, 0 + t 4/ 17, 1/ 17 ; in parametric form,
{
x = 2 + 4t/
17
(t) =
y =
t/ 17
11. (t) =
2/4, 2/4 + t 2/2, 2/2 ; in parametric form,
{
x = 2/4 2t/2
(t) =
y =
2/4 + 2t/2
7. T(t) =
13. T(t) = sin t, cos t;

N(t) = cos t, sin t
t
2 cos t
15. T(t) = sin
,
;
4 cos2 t+sin2 t
4 cos2 t+sin2 t
N(t) = 2 cos t
, sin t
4 cos2 t+sin2 t
17.
19.
4 cos2 t+sin2 t
;
192
17 17
2.74.
| cos x|
13. =
;
(1+sin2 x)3/2
(0) = 1, (/2) = 0
15. =
|2 cos t cos(2t)+4 sin t sin(2t)|
(4 cos2 (2t)+sin2 t)3/2

(0) = 1/4, (/4) = 8
17. =
|6t2 +2|
(4t2 +(3t2 1)2 )3/2
(0) = 2, (5) =
19
1394 1394
0.0004
19. = 0;
(0) = 0, (1) = 0
3
21. = 13
;
(0) = 3/13, (/2) = 3/13
2
23. maximized at x =
4
25. maximized at t = 1/4
27. radius of curvature is 5 5/4.
(a) Be sure to show work

(b)
N(/4) = 5/ 34, 3/ 34
29. radius of curvature is 9.
(a) Be sure to show work
(b)
N(0) = 1 , 2
33. x2 + (y + 8)2 = 81, orc(t) = 9 cos t, 9 sin t 8
21. T(t) =
1
5
23. T(t) =
1+4t2
a sin t, a cos t, b;
N(t) = cos t, sin t, 0
and aN =
31. x2 + (y 1/2)2 = 1/4, orc(t) = 1/2 cos t, 1/2 sin t + 1/2
Chapter 12
2, cos t, sin t;
N(t) = 0, sin t, cos t
1
a2 +b2
25. aT = 4t
16t2
1+4t2
At t = 0, aT = 0 and
aN = 2;
At t = 1, aT = 4/ 5 and aN = 2/ 5.
At t = 0, all accelera on comes in the form of changing the
direc on of velocity and not the speed; at t = 1, more
accelera on comes in changing the speed than in changing
direc on.
27. aT = 0 and aN = 2
At t = 0, aT = 0 and aN = 2;
At t = /2, aT = 0 and aN = 2.
The object moves at constant speed, so all accelera on comes
from changing direc on, hence aT = 0. a(t) is always parallel to
N(t), but twice as long, hence aN = 2.

29. aT = 0 and aN = a
At t = 0, aT = 0 and aN = a;
At t = /2, aT = 0 and aN = a.
The object moves at constant speed, meaning that aT is always 0.
The object rises along the z-axis at a constant rate, so all
accelera on comes in the form of changing direc on circling the
z-axis. The greater the radius of this circle the greater the
accelera on, hence aN = a.
Sec on 11.5
1.
|6x|
(1+(3x2 1)2 )3/2
Sec on 12.1
3. topographical
5. surface
7. domain: R2
range: z 2
9. domain: R2
range: R
11. domain: R2
range: 0 < z 1
13. domain: {(x, y) | x2 + y2 9}, i.e., the domain is the circle and
interior of a circle centered at the origin with radius 3.
range: 0 z 3
15. Level curves are lines y = (3/2)x c/2.
y
x
2
1
2
me and/or distance
3. Answers may include lines, circles, helixes

The func on z = x2 + 4y2 can be rewri en as z2 = x2 + 4y2 ,
an ellip c cone; the func on z = x2 + 4y2 is a paraboloid, each
matching the descrip on above.
17. Level curves are parabolas x = y2 + c.

y
4
Sec on 12.2
2
c = 2
4
c=0
c=2

One possible answer: {(x, y)|x2 + y2 1}
x
4

One possible answer: {(x, y)|x2 + y2 < 1}
.
.

interior point: (1, 3)
boundary point: (3, 3)
S is a closed set
S is bounded
19. Level curves are circles, centered at (1/c, 1/c) with radius
2/c2 1. When c = 0, the level curve is the line y = x.
y

interior point: none
boundary point: (0, 1)
S is a closed set, consis ng only of boundary points
S is bounded
4
c = 1
c=0
11. D = {(x, y) | y = 2x}; D is an open set.

{
}
13. D = (x, y) | y > x2 ; D is an open set.
x
4
2
2
15.
c=1
21. Level curves are ellipses of the form

and b = c/2.
(a) Along y = 0, the limit is 1.

(b) Along x = 0, the limit is 1.
Since the above limits are not equal, the limit does not exist.
x2
c2
y2
c2 /4
= 1, i.e., a = c
17.
y
4
mx(1 m)
.
m2 x + 1
(b) Along x = 0, the limit is 1.
Since the above limits are not equal, the limit does not exist.
(a) Along y = mx, the limit is
19.
(a) Along y = 2, the limit is:
lim
(x,y)(1,2)
x
4
x+y3
x1
= lim 2
x1 x 1
x2 1
1
= lim
x1 x + 1
= 1/2.
(b) Along y = x + 1, the limit is:
lim
(x,y)(1,2)
23. domain: x + 2y 4z = 0; the set of points in R3 NOT in the

domain form a plane through the origin.
range: R
25. domain: z x2 y2 ; the set of points in R3 above (and
including) the hyperbolic paraboloid z = x2 y2 .
range: [0, )
Since the limits along the lines y = 2 and y = x + 1 dier, the

overall limit does not exist.
27. The
level surfaces are spheres, centered at the origin, with radius
c.
29. The level surfaces are paraboloids of the form z =
larger c, the wider the paraboloid.
x2
c
31. The level curves for each surface are similar; for z =
the level curves are ellipses of the form
x
c2
y
c2 /4
2(x 1)
x+y3
= lim 2
x1 x 1
x2 1
2
= lim
x1 x + 1
= 1.
y2
;
c
the
x2 + 4y2
= 1, i.e., a = c
and b = c/2; whereas for z = x2 + 4y2 the level curves are
2
y2
ellipses of the form xc + c/4
= 1, i.e., a = c and b = c/2.
The rst set of ellipses are spaced evenly apart, meaning the
func on grows at a constant rate; the second set of ellipses are
more closely spaced together as c grows, meaning the func on
grows faster and faster as c increases.
Sec on 12.3
1. A constant is a number that is added or subtracted in an
expression; a coecient is a number that is being mul plied by a
nonconstant func on.
3. fx
5. fx = 2xy 1, fy = x2 + 2
fx (1, 2) = 3, fy (1, 2) = 3
7. fx = sin x sin y, fy = cos x cos y
fx (/3, /3) = 3/4, fy (/3, /3) = 1/4
9. fx = 2xy + 6x, fy = x2 + 4
fxx = 2y + 6, fyy = 0
fxy = 2x, fyx = 2x
11. fx = 1/y, fy = x/y2

fxx = 0, fyy = 2x/y3
fxy = 1/y2 , fyx = 1/y2
2
17. dw = 2xyz3 dx + x2 z3 dy + 3x2 yz2 dz

19. dx = 0.05, dy = 0.1. dz = 9(.05) + (2)(0.1) = 0.65. So
f(3.5, 0.9) 7 + 0.65 = 7.65.
13. fx = 2xex +y , fy = 2yex +y

2
2
2
2
2
2
2
2
fxx = 2ex +y + 4x2 ex +y , fyy = 2ex +y + 4y2 ex +y
2
2
2
2
fxy = 4xyex +y , fyx = 4xyex +y
21. dx = 0.5, dy = 0.1, dz = 0.2.

dw = 2(0.5) + (3)(0.1) + 3.7(0.2) = 0.04, so
f(2.5, 4.1, 4.8) 1 0.04 = 1.04.
15. fx = cos x cos y, fy = sin x sin y

fxx = sin x cos y, fyy = sin x cos y
fxy = sin y cos x, fyx = sin y cos x
Sec on 12.5
1. Because the parametric equa ons describe a level curve, z is
constant for all t. Therefore dz
= 0.
dt
17. fx = 5y3 sin(5xy3 ), fy = 15xy2 sin(5xy3 )

fxx = 25y6 cos(5xy3 ),
fyy = 225x2 y4 cos(5xy3 ) 30xy sin(5xy3 )
fxy = 75xy5 cos(5xy3 ) 15y2 sin(5xy3 ),
fyx = 75xy5 cos(5xy3 ) 15y2 sin(5xy3 )
2
19. fx = 2y 2
4xy +1
4y4
3.
5. F
7.
, fy = 4xy2
4xy +1
16x
3 , fyy =
fxx =
3
4xy2 +1
4xy2 +1
3
fxy = 8xy
2 2
4xy2 +1
4y
, fyx
4xy2 +1
= 8xy
4xy2 +1
fxy =
8x2
(x2 +y22 +1)2 , fyy
(x2 +y2 +1)3
8xy
, f = (x2 +y8xy
2 +1)3
(x2 +y2 +1)3 yx
4y
4xy2 +1
8y2
(x2 +y2 +1)3
2
(x2 +y2 +1)2
9.
11.
19.
4xy2
dz
dt
dz
= 2x(cos t) + 4y(3 cos t).
dt
(b) At t = /4, x = 2/2, y = 3 2/2, and

(a)
3
, f = 7y6x3 z , fz
7y2 z y
6x
fyz = 7y6x
3 z2 , fzy = 7y3 z2
33. fx =
dz
dt
= 19.
= 7y3x
2 z2
21.
1. T
3. T
7. dz = 5dx 7dy
x +y
z
s
z
t
= 2xy(1) + x2 (2) = 2xy + 2x2 ;

= 2xy(1) + x2 (4) = 2xy + 4x2
(a)
z
s
= 6 and
z
s
z
t
= 2x(cos t) + 2y(sin t) = 2x cos t + 2y sin t;

= 2x(s sin t) + 2y(s cos t) = 2xs sin t + 2ys cos t
=4
(b) With s = 2, t = /4, x = 2 and y = 2. Thus z
s
and z
=
0
t
23. fx = 2x tan y, fy = x2 sec2 y;

dy
2 tan y
=
dx
x sec2 y
5. dz = (sin y + 2x)dx + (x cos y)dy

dx + 12
dz
dt
dz
= 38 cos t sin t.
dt
= 0 when t = n or n + /2, where n is any integer.
(b) With s = 1, t = 1, x = 1 and y = 2. Thus

z
=0
t
Sec on 12.4
(a)
+ 2y + C, where C is any constant.
31. fx = 2xe2y3z , fy = 2x2 e2y3z , fz = 3x2 e2y3z

fyz = 6x2 e2y3z , fzy = 6x2 e2y3z
x +y
= 14.
= 5(2 sin t) + 2(cos t) = 10 sin t + 2 cos t
(b) At t = /4, dz
= 4 2.
dt
(a)
Thus
27. f(x, y) = x sin y + x + C, where C is any constant.
9. dz = x2
dz
dt
17. We nd that
1
ln x
, f = 4y
2
4xy y
ln x
fxx = 4x12 y , fyy = 2y
3
1
1
fxy = 4xy2 , fyx = 4xy
2
= 3(2t) + 4(2) = 6t + 8.
15. t = tan1 (1/5) + n, where n is an integer
25. fx =
29. f(x, y) =
dz
dt
13. t = 4/3; this corresponds to a minimum
23. fx = 6x, fy = 0
fxx = 6, fyy = 0
fxy = 0, fyx = 0
3x2 y
(a)
(b) At t = 1,
4x
4xy2 +1
21. fx = (x2 +y2x2 +1)2 , fy = (x2 +y2y2 +1)2

fxx =
f
dx
, and y
dt
dy, with dx = 0.05 and dy = .1. At
(3, 7), dz = 3/4(0.05) + 1/8(.1) = 0.025, so

f(2.95, 7.1) 0.025 + 4 = 3.975.
11. dz = (2xy y2 )dx + (x2 2xy)dy, with dx = 0.04 and
dy = 0.06. At (2, 3), dz = 3(0.04) + (8)(0.06) = 0.36, so
f(2.04, 3.06) 0.36 6 = 6.36.
13. The total dieren al of volume is dV = 4dr + dh. The
coecient of dr is greater than the coecient of dh, so the
volume is more sensi ve to changes in the radius.
x sec2
15. Using trigonometry, = x tan , so d = tan dx +

d.
With = 85 and x = 30, we have d = 11.43dx + 3949.38d.
The measured length of the wall is much more sensi ve to errors
in than in x. While it can be dicult to compare sensi vi es
between measuring feet and measuring degrees (it is somewhat
like comparing apples to oranges), here the coecients are so
dierent that the result is clear: a small error in degree has a
much greater impact than a small error in distance.
(x + y2 )(2x) (x2 + y)(1)

,
(x + y2 )2
2
2
(x + y )(1) (x + y)(2y)
;
fy =
(x + y2 )2
2
2
dy
2x(x + y ) (x + y)
=
dx
x + y2 2y(x2 + y)
25. fx =
Sec on 12.6
1. A par al deriva ve is essen ally a special case of a direc onal
deriva ve; it is the direc onal deriva ve in the direc on of x or y,
i.e., 1, 0 or 0, 1.
3. u = 0, 1
5. maximal, or greatest
7. f = 2xy + y2 + y, x2 + 2xy + x
2y
2x
9. f = (x2 +y
2 +1)2 , (x2 +y2 +1)2
11. f = 2x y 7, 4y x
13. f = 2xy + y2 + y, x2 + 2xy + x ; f(2, 1) = 2, 2. Be

sure to change all direc ons to unit vectors.
(a) 2/5 (u = 3/5, 4/5)
(b) 2 5 (u = 1/ 5, 2 5 )
2y
2x
15. f = (x2 +y
2 +1)2 , (x2 +y2 +1)2 ; f(1, 1) = 2/9, 2/9. Be
sure to change all direc ons to unit vectors.

(a) 0 (u = 1/ 2, 1/ 2 )
(b) 2 2/9 (u = 1/ 2, 1/ 2 )
17. f = 2x y 7, 4y x; f(4, 1) = 0, 0.
(a) 0
(b) 0
19. f = 2xy + y2 + y, x2 + 2xy + x

(a) f(2, 1) = 2, 2
(b) || f(2, 1) || = || 2, 2 || =
(c) 2, 2

(d) 1/ 2, 1/ 2
2y
2x
21. f = (x2 +y
2 +1)2 , (x2 +y2 +1)2
(a) f(1, 1) = 2/9, 2/9.
(b) || f(1, 1) || = || 2/9, 2/9 || = 2 2/9

(c) 2/9, 2/9

(d) 1/ 2, 1/ 2
23. f = 2x y 7, 4y x
(a) f(4, 1) = 0, 0
x=4+t
y=2
7.
(a) x (t) =
z = 2 + 3t
x=4
y=2+t
(b) y (t) =
z = 2 5t
x = 4 + t/2
(c) u (t) =
y = 2+
t/ 2
z = 2 2t
x = 2 12t
y = 3 40t
9. n (t) =
z = 48 t
x = 4 + 3t
y = 2 5t
11. n (t) =
z=2t
13. (1.425, 1.085, 48.078), (2.575, 4.915, 47.952)
15. (5.014, 0.31, 1.662) and (2.986, 3.690, 2.338)
17. 12(x 2) 40(y 3) (z + 48) = 0
19. 3(x 4) 5(y 2) (z 2) = 0 (Note that this tangent plane
is the same as the original func on, a plane.)
21. F = x/4, y/2, z/8; at P, F = 1/4, 2/2, 6/8
x = 1+ t/4
y = 2 + 2t/2
(a) n (t) =
z = 6 + 6t/8
1
(b) 4 (x 1) + 22 (y 2) + 86 (z 6) = 0.
23. F = y2 z2 , 2xy, 2xz ; at P, F = 0, 4, 4
x=2
y = 1 + 4t
(a) n (t) =
z = 1 + 4t
(b) 4(y 1) + 4(z + 1) = 0.
(b) 0
(c) 0, 0
25.
27.
(d) All direc ons give a direc onal deriva ve of 0.
(a) F(x, y, z) = 6xz3 + 4y, 4x, 9x2 z2 6z
(b) 113/ 3
(a) F(x, y, z) = 2xy2 , 2y(x2 z2 ), 2y2 z

(b) 0
Sec on 12.7
1. Answers will vary. The displacement of the vector is one unit in
the x-direc on and 3 units in the z-direc on, with no change in y.
Thus along a line parallel to v, the change in z is 3 mes the
change in x i.e., a slope of 3. Specically, the line in the x-z
plane parallel to z has a slope of 3.
3. T
5.
x=2+t
y=3
(a) x (t) =
z = 48 12t
x=2
y=3+t
(b) y (t) =
z = 48 40t
10
x = 2 + t/
(c) u (t) =
y = 3 + 3t/ 10
z = 48 66 2/5t
Sec on 12.8
1. F; it is the other way around.
3. T
5. One cri cal point at (4, 2); fxx = 1 and D = 4, so this point
corresponds to a rela ve minimum.
7. One cri cal point at (6, 3); D = 4, so this point corresponds
to a saddle point.
9. Two cri cal points: at (0, 1); fxx = 2 and D = 12, so this point
corresponds to a saddle point;
at (0, 1), fxx = 2 and D = 12, so this corresponds to a rela ve
minimum.
11. One cri cal point at (0, 0). D = 12x2 y2 , so at (0, 0), D = 0 and
the test is inconclusive. (Some elementary thought shows that it
is the absolute minimum.)
13. One cri cal point: fx = 0 when x = 3; fy = 0 when y = 0, so one
cri cal point at (3, 0), which is a rela ve maximum, where
fxx =
y2 16
(16(x3)2 y2 )3/2
and D =
16
.
(16(x3)2 y2 )2
the circle (x 3)2
Both fx and fy are undened along

+ y2 = 16;
at any point along this curve, f(x, y) = 0, the absolute minimum
of the func on.
15. The triangle is bound by the lines y = 1, y = 2x + 1 and
y = 2x + 1.
Along y = 1, there is a cri cal point at (0, 1).
Along y = 2x + 1, there is a cri cal point at (3/5, 1/5).
Along y = 2x + 1, there is a cri cal point at (3/5, 1/5).

The func on f has one cri cal point, irrespec ve of the constraint,
at (0, 1/2).
Checking the value of f at these four points, along with the three
ver ces of the triangle, we nd the absolute maximum is at
(0, 1, 3) and the absolute minimum is at (0, 1/2, 3/4).
17. The region has no corners or ver ces, just a smooth edge.
To nd cri cal points along the circle x2 + y2 = 4,we solve for y2 :
y2 = 4 x2 . We can go further and state y = 4 x2 .
We can rewrite f as
2
2
2
f(x) = x2 + 2x + (4
x ) + 4 x = 2x + 4 + 4 x . (We
2
will return
and use
4 x later.) Solving f (x) = 0, we get

x = 2 y = 2. f (x) is also undened at x = 2, where
y = 0.
2
Using y = 4
x , we rewrite f(x, y) as
f(x) =
2x + 4 4 x2 . Solving f (x) = 0, we get
x = 2, y = 2.
The func on f itself has a cri cal point
at (1,
1).
Checking the value of f at (1, 1), ( 2, 2), ( 2, 2),

(2, 0) and (2, 0), we nd the absolute maximum is at (2, 0, 8)
and the absolute minimum is at (1, 1, 2).
area of R =
0
dx dy
y
2
R
19.
x
4
2
x2 /16 + y2 /4 = 1
2
4x2 /4
area of R =
dy dx
4x2 /4
y
4
Chapter 13
y=
21.
Sec on 13.1
4y
4y
x+
R
1. C(y), meaning that instead of being just a constant, like the
number 5, it is a func on of y, which acts like a constant when
taking deriva ves with respect to x.
(a) 18x2 + 42x 117
x+2
x2
dy dx
(b) 23/15
1. volume
(a) sin2 y
3. The double integral gives the signed volume under the surface.
Since the surface is always posi ve, it is always above the x-y
plane and hence produces only posi ve volume.
)
1 2(
x
5. 6;
+ 3 dy dx
y
1 1
2 42y
( 2
)
7. 112/3;
3x y + 2 dx dy
(b) /2
4 1
11.
dy dx and
2
9u2
1
2
dx dy.
1
area of R =
4 7x
13.
dy dx. The order dx dy needs two iterated integrals as
2
x1
x is bounded above by two dierent func ons. This gives:

5 7y
3 y+1
dx dy +
dx dy.
1
area of R =
1 x
15.
dy dx and
x4
area of R = 7/15u2
9. 16/5;
4y
y2
dx dy
0
1
0
1x2
(x + y + 2) dy dx
0
x2
1
x2 y dy dx =
y2
x2 y2 dy dx =
x2 y dx dy.
x2 y2 dx dy.
2 33/2x
(
)
15. 6 =
6 3x 2y dy dx =
0
0
3 22/3y
(
)
6 3x 2y dx dy.
17. 0 =
9x2
0
9y2
9y2
( 3
)
x y x dy dx =
( 3
)
x y x dx dy.
y = 4 x2
x
2
3
=
56
13. 0 =
17.
11.
4u2
Sec on 13.2
(a) x4 /2 x2 + 2x 3/2
9.
area of R =
(b) 108
7.
3. curve to curve, then from point to point

5.
y = x2
19. Integra ng ex with respect

to x is not possible in terms of
2
2x
elementary func ons.

0
ex dy dx = e4 1.
2y
dx gives tan1 (1/y) /4; integra ng
x2 + y2
tan1 (1/y) is hard.
1 x
2y
dy dx = ln 2.
2 + y2
x
0
0
21. Integra ng
23. average value of f = 6/2 = 3

25. average value of f =
112/3
4
= 28/3
Sec on 13.3
(
)
1. f r cos , r sin , r dr d
2 1
(
)
3r cos r sin + 4 r dr d = 4
3.
3 cos
9.
/2
/2
/2
11.
/2
5. Intui vely, adding h to f only shi s f up (i.e., parallel to the z-axis)

and does not change its shape. Therefore it will not change the
surface area over R.
Analy cally, fx = gx and fy = gy ; therefore, the surface area of
each is computed with iden cal double integrals.
2 2
7. SA =
1 + cos2 x cos2 y + sin2 x sin2 y dx dy
9. SA =
)
(
)
ln(r2 ) r dr d = 2 ln 16 3/2
(2
)
r cos2 r2 sin2 r dr d =
(2
)
r cos(2) r dr d = 0
/4
11. SA =
0
1 + 9 + 49 dx dy = 6 59 46.09
)
r cos + r sin r dr d = 16 2/3
0
2
lim (1 ea ) = . This implies that there is a nite
volume under the surface

plane.
over the en re x-y
1 + 4r2 dr d
)
(
65 65 1 273.87
6
15.
2x
SA =
1 + 1 + 4x2 dy dx
(
)
2x 2 + 4x2 dx
26
2 12.26
=
3
17. This is easier in polar:
2
2
e(x +y )
(a) This is impossible to integrate with rectangular coordinates

2
2
as e(x +y ) does not have an an deriva ve in terms of
elementary func ons.
2 a
2
2
(b)
rer dr d = (1 ea ).
(c)
1 + 4x2 + 4y2 dx dy
13. This is easier in polar:

2 4
SA =
r 1 + 4r2 cos2 t + 4r2 sin2 t dr d
( 2)
r dr d = 125/3
13.
15.
3. surface areas
/2
(
)
8 r sin r dr d = 16
1
/2
1. arc length
cos
7.
Sec on 13.5
5.
29. Ix = 16/3; Iy = 64/3; IO = 80/3
SA =
0
2
1+
0
5
4r2 cos2 t + 4r2 sin2 t

dr d
r2 sin2 t + r2 cos2 t
r 5 dr d
Sec on 13.4
1. Because they are scalar mul ples of each other.
= 25 5 175.62
3. li le masses
5. Mx measures the moment about the x-axis, meaning we need to
measure distance from the x-axis. Such measurements are
measures in the y-direc on.
19. Integra ng in polar is easiest considering R:

2 1
SA =
r 1 + c2 + d2 dr d
0
9. (x, y) = (0, 3)
13. M = 2lb
15. M = 16 50.27kg
7. x = 5.25
11. M = 150gm;
)
1 (
1 + c2 + d2 dy
2
1 + c 2 + d2 .
The value of h does not ma er as it only shi s the plane ver cally
(i.e., parallel to the z-axis). Dierent values of h do not create
dierent ellipses in the plane.
Sec on 13.6
17. M = 54 169.65lb
19. M = 150gm; My = 600; Mx = 75; (x, y) = (4, 0.5)
21. M = 2lb; My = 0; Mx = 2/3; (x, y) = (0, 1/3)
23. M = 16 50.27kg; My = 4; Mx = 4; (x, y) = (1/4, 1/4)
25. M = 54 169.65lb; My = 0; Mx = 504; (x, y) = (0, 2.97)
27. Ix = 64/3; Iy = 64/3; IO = 128/3
1. surface to surface, curve to curve and point to point

3. Answers can vary. From this sec on we used triple integra on to
nd the volume of a solid region, the mass of a solid, and the
center of mass of a solid.
1 1 (
)
2
2
5. V = 1
1 8 x y (2x + y) dx dy = 88/3
(
)
7. V = 0 0x cos x sin y + 2 sin x cos y dy dx = 2 6.728
1x/3
22x/32y
9. dz dy dx:
dz dy dx
0 1 0 33y
dz dy dx
0 3 0 22x/30
dy dz dx
0 2 0 33z/2 0 1x/3z/2
dy dx dz
dy dx dz:
0
33y3z/2
dx dz dy:
dx dz dy
1z/2
dx dy dz:
0
3 01x/3
V=
0
dz dx dy:
dy dz dx:
dx dz dy:
dy dz dx = 4/3.
2x+4y4
dz dy dx
0 1 1x/2
0
2x+4y4
dz dx dy:
dz dy dx
22y
2yz/22
dx dy dz = 4/3.
1x2
1y
dz dy dx
0 1 0 1y 0
0
1x2
1z2
0
0 1 0 1y
1y
dz dy dx
1 1 1z2
dy dz dx +
dy dz dx
0
x
0
1 1 1x2
dy dx dz +
dy dx dz
1y
dx dz dy:
dx dz dy
0
1
1z2
dx dy dz:
0
1y
dx dy dz
0
Answers will vary. Neither order is par cularly hard. The order
dz dy dx requires integra ng a square root, so powers can be
messy; the order dy dz dx requires two triple integrals, but each
uses only polynomials.
17. 8
19.
13. dz dy dx:
z/4
dx dy dz
2z
2z
dy dx dz:
dx dz dy
z/22y+2
dy dz dx:
dz dx dy
y2 /2
z
y /2 0
2
2 z
V=
0
dz dy dx
dy dx dz
z/4
dz dx dy:
2z
y 2
dx dy dz:
dx dy dz
0
V=
dy dz dx
0 0 0
dx dz dy
15. dz dy dx:
0 2 0 2 z2z
dy dx dz:
dy dx dz
dx dy dz:
dz dy dx = 1.
dy dz dx
0 4 0 1 z/22y+2
2
dx dy dz
/2
0 0 22 y y
0
2
2 2
dx dz dy:
0
22x/32y
z/4x/2+1
0 1 z/2
4y 2
33y3z/2
11. dz dy dx:
2x
dy dx dz:
1x/3z/2
dy dz dx:
0 1 0 22y
0 4 0 2 z/4x/2+1
1
0
22x/32y
dz dx dy:
dy dz dx:
21. M = 10, Myz = 15/2, Mxz = 5/2, Mxy = 5;

(x, y, z) = (3/4, 1/4, 1/2)
23. M = 16/5, Myz = 16/3, Mxz = 104/45, Mxy = 32/9;
(x, y, z) = (5/3, 13/18, 10/9) (1.67, 0.72, 1.11)
Index
!, 383
Absolute Convergence Theorem, 431
absolute maximum, 121
absolute minimum, 121
Absolute Value Theorem, 387
accelera on, 71, 618
Alterna ng Harmonic Series, 403, 428, 441
Alterna ng Series Test
for series, 425
aN , 636, 646
analy c func on, 459
angle of eleva on, 623
an deriva ve, 185
arc length, 357, 499, 523, 615, 640
arc length parameter, 640, 642
asymptote
horizontal, 46
ver cal, 44
aT , 636, 646
average rate of change, 603
average value of a func on, 743
average value of func on, 229
Binomial Series, 460
Bisec on Method, 39
boundary point, 658
bounded sequence, 389
convergence, 390
bounded set, 658
center of mass, 757759, 761, 788
Chain Rule, 94
mul variable, 689, 691
nota on, 100
circle of curvature, 645
closed, 658
closed disk, 658
concave down, 142
concave up, 142
concavity, 142, 496
inec on point, 143
test for, 143
conic sec ons, 469
degenerate, 469
ellipse, 473
hyperbola, 476
parabola, 470
Constant Mul ple Rule
of deriva ves, 78
of integra on, 189
of series, 403
constrained op miza on, 720

con nuous func on, 34, 664
proper es, 37, 665
vectorvalued, 606
contour lines, 653
convergence
absolute, 429, 431
Alterna ng Series Test, 425
condi onal, 429
Direct Comparison Test, 413
for integra on, 327
Integral Test, 410
interval of, 436
Limit Comparison Test, 414
for integra on, 329
nth term test, 406
of geometric series, 398
of improper int., 322, 327, 329
of monotonic sequences, 393
of p-series, 399
of power series, 435
of sequence, 385, 390
of series, 395
radius of, 436
Ra o Comparison Test, 419
Root Comparison Test, 422
cri cal number, 123
cri cal point, 123, 715717
cross product
and deriva ves, 611
applica ons, 574
area of parallelogram, 575
torque, 577
volume of parallelepiped, 576
deni on, 570
proper es, 572, 573
curvature, 642
and mo on, 646
equa ons for, 644
of circle, 644, 645
radius of, 645
curve
parametrically dened, 483
rectangular equa on, 483
smooth, 489
curve sketching, 149
cusp, 489
cycloid, 601
cylinder, 532
decreasing func on, 134
nding intervals, 135

strictly, 134
denite integral, 196
and subs tu on, 262
proper es, 197
deriva ve
accelera on, 72
as a func on, 62
at a point, 58
basic rules, 76
Chain Rule, 94, 100, 689, 691
Constant Mul ple Rule, 78
Constant Rule, 76
dieren al, 179
direc onal, 696, 698, 699, 702
exponen al func ons, 100
First Deriv. Test, 137
Generalized Power Rule, 95
higher order, 79
interpreta on, 80
hyperbolic funct., 306
implicit, 103, 693
interpreta on, 69
inverse func on, 114
inverse hyper., 309
inverse trig., 117
Mean Value Theorem, 130
mixed par al, 672
mo on, 72
mul variable dieren ability, 681, 686
normal line, 59
nota on, 62, 79
parametric equa ons, 493
par al, 668, 676
Power Rule, 76, 89, 108
power series, 439
Product Rule, 83
Quo ent Rule, 86
Second Deriv. Test, 146
Sum/Dierence Rule, 78
tangent line, 58
trigonometric func ons, 87
vectorvalued func ons, 607, 608, 611
velocity, 72
dieren able, 58, 681, 686
dieren al, 179
nota on, 179
for integra on, 327
for series, 413
direc onal deriva ve, 696, 698, 699, 702
directrix, 470, 532
Disk Method, 342
displacement, 223, 602, 615
distance
between lines, 587
between point and line, 587
between point and plane, 595
between points in space, 530
traveled, 626
divergence
for integra on, 327
Integral Test, 410
for integra on, 329
nth term test, 406
of geometric series, 398
of improper int., 322, 327, 329
of p-series, 399
of sequence, 385
of series, 395
dot product
and deriva ves, 611
deni on, 557
proper es, 558, 559
double integral, 736, 737
in polar, 747
proper es, 740
eccentricity, 475, 479
elementary func on, 233
ellipse
deni on, 473
eccentricity, 475
reec ve property, 476
standard equa on, 474
extrema
absolute, 121, 715
and First Deriv. Test, 137
and Second Deriv. Test, 146
nding, 124
rela ve, 122, 715, 716
Extreme Value Theorem, 122, 720
extreme values, 121
factorial, 383
First Deriva ve Test, 137
uid pressure/force, 375, 377
focus, 470, 473, 476
Fubinis Theorem, 737
func on
of three variables, 655
of two variables, 651
vectorvalued, 599
Fundamental Theorem of Calculus, 221, 222
and Chain Rule, 225
Gabriels Horn, 363
Generalized Power Rule, 95
geometric series, 397, 398
gradient, 698, 699, 702, 712
and level curves, 699
and level surfaces, 712
Harmonic Series, 403
Head To Tail Rule, 547
Hookes Law, 368

hyperbola
deni on, 476
eccentricity, 479
standard equa on, 477
hyperbolic func on
deni on, 303
deriva ves, 306
iden es, 306
integrals, 306
inverse, 307
deriva ve, 309
integra on, 309
logarithmic def., 308
implicit dieren a on, 103, 693
improper integra on, 322, 325
increasing func on, 134
nding intervals, 135
strictly, 134
indenite integral, 185
indeterminate form, 2, 45, 316, 317
inec on point, 143
ini al point, 543
ini al value problem, 190
Integral Test, 410
integra on
arc length, 357
area, 196, 728, 729
area between curves, 226, 334
average value, 229
by parts, 266
by subs tu on, 249
denite, 196
and subs tu on, 262
proper es, 197
Riemann Sums, 217
displacement, 223
distance traveled, 626
double, 736
uid force, 375, 377
Fun. Thm. of Calc., 221, 222
general applica on technique, 333
hyperbolic funct., 306
improper, 322, 325, 327, 329
indenite, 185
inverse hyper., 309
iterated, 727
Mean Value Theorem, 227
mul ple, 727
nota on, 186, 196, 222, 727
numerical, 233
Le /Right Hand Rule, 233, 240
Simpsons Rule, 238, 240, 241
Trapezoidal Rule, 236, 240, 241
of mul variable func ons, 725
of power series, 439
of trig. func ons, 255
of trig. powers, 276, 281

of vectorvalued func ons, 613
par al frac on decomp., 296
Power Rule, 190
Sum/Dierence Rule, 190
surface area, 361, 501, 524
trig. subst., 287
triple, 774, 785, 787
volume
cross-sec onal area, 341
Disk Method, 342
Shell Method, 349, 353
Washer Method, 344, 353
work, 365
interior point, 658
Intermediate Value Theorem, 39
interval of convergence, 436
iterated integra on, 727, 736, 737, 774, 785, 787
changing order, 731
proper es, 740, 781
LHpitals Rule, 313, 315
lamina, 753
Le Hand Rule, 204, 209, 212, 233
Le /Right Hand Rule, 240
level curves, 653, 699
level surface, 656, 712
limit
at innity, 46
deni on, 10
dierence quo ent, 6
does not exist, 4, 29
indeterminate form, 2, 45, 316, 317
LHpitals Rule, 313, 315
le handed, 27
of innity, 43
of mul variable func on, 659, 660, 666
of sequence, 385
of vectorvalued func ons, 605
one sided, 27
proper es, 16, 660
pseudo-deni on, 2
right handed, 27
Squeeze Theorem, 20
for integra on, 329
for series, 414
lines, 580
distances between, 587
equa ons for, 582
intersec ng, 583
parallel, 583
skew, 583
logarithmic dieren a on, 110
Maclaurin Polynomial, see Taylor Polynomial
deni on, 447
Maclaurin Series, see Taylor Series
deni on, 457
magnitude of vector, 543

mass, 753, 754, 788
center of, 757
maximum
absolute, 121, 715
and First Deriv. Test, 137
and Second Deriv. Test, 146
rela ve/local, 122, 715, 718
Mean Value Theorem
of dieren a on, 130
of integra on, 227
Midpoint Rule, 204, 209, 212
minimum
absolute, 121, 715
and First Deriv. Test, 137, 146
rela ve/local, 122, 715, 718
moment, 759, 761, 788
monotonic sequence, 390
mul ple integra on, see iterated integra on
mul variable func on, 651, 655
con nuity, 664666, 682, 687
dieren ability, 681, 682, 686, 687
domain, 651, 655
level curves, 653
level surface, 656
limit, 659, 660, 666
range, 651, 655
Newtons Method, 158
norm, 543
normal line, 59, 493, 708
normal vector, 590
nth term test, 406
numerical integra on, 233
Le /Right Hand Rule, 233, 240
Simpsons Rule, 238, 240
error bounds, 241
Trapezoidal Rule, 236, 240
error bounds, 241
open, 658
open ball, 666
open disk, 658
op miza on, 171
constrained, 720
orthogonal, 561, 708
decomposi on, 565
orthogonal decomposi on of vectors, 565
orthogonal projec on, 563
oscula ng circle, 645
p-series, 399
parabola
deni on, 470
general equa on, 471
parallel vectors, 551
Parallelogram Law, 547
parametric equa ons
arc length, 499
concavity, 496
deni on,
483
2
nding ddxy2 , 497
nding dy
dx , 493
normal line, 493
surface area, 501
tangent line, 493
par al deriva ve, 668, 676
high order, 676
meaning, 670
mixed, 672
second deriva ve, 672
total dieren al, 680, 686
perpendicular, see orthogonal
planes
coordinate plane, 531
distance between point and plane, 595
equa ons of, 591
introduc on, 531
normal vector, 590
tangent, 711
point of inec on, 143
polar
coordinates, 503
func on
arc length, 523
gallery of graphs, 510
surface area, 524
func ons, 506
area, 519
area between curves, 521
nding dy
dx , 516
graphing, 506
polar coordinates, 503
plo ng points, 503
Power Rule
dieren a on, 76, 83, 89, 108
integra on, 190
power series, 434
algebra of, 462
convergence, 435
deriva ves and integrals, 439
projec le mo on, 623, 624, 637
quadric surface
deni on, 535
ellipsoid, 538
ellip c cone, 537
ellip c paraboloid, 537
gallery, 537539
hyperbolic paraboloid, 539
hyperboloid of one sheet, 538
hyperboloid of two sheets, 539
sphere, 538
trace, 536
Quo ent Rule, 86
R, 543
radius of convergence, 436
radius of curvature, 645
Ra o Comparison Test
for series, 419

rearrangements of series, 430, 431
related rates, 164
Riemann Sum, 204, 208, 211
and denite integral, 217
Right Hand Rule, 204, 209, 212, 233
right hand rule
of Cartesian coordinates, 529
Rolles Theorem, 130
Root Comparison Test
for series, 422
saddle point, 717, 718
Second Deriva ve Test, 146, 718
sensi vity analysis, 685
sequence
posi ve, 413
sequences
boundedness, 389
convergent, 385, 390, 393
deni on, 383
divergent, 385
limit, 385
limit proper es, 388
monotonic, 390
series
absolute convergence, 429
Absolute Convergence Theorem, 431
alterna ng, 424
Approxima on Theorem, 427
Binomial, 460
condi onal convergence, 429
convergent, 395
deni on, 395
divergent, 395
geometric, 397, 398
Integral Test, 410
interval of convergence, 436
Maclaurin, 457
nth term test, 406
p-series, 399
par al sums, 395
power, 434, 435
deriva ves and integrals, 439
proper es, 403
radius of convergence, 436
rearrangements, 430, 431
Taylor, 457
telescoping, 400, 401
Shell Method, 349, 353
signed area, 196
signed volume, 736, 737
Simpsons Rule, 238, 240
error bounds, 241
smooth, 610
smooth curve, 489
speed, 618
sphere, 530
Squeeze Theorem, 20
Sum/Dierence Rule
of deriva ves, 78
of integra on, 190
of series, 403
summa on
nota on, 205
proper es, 207
surface area, 766
solid of revolu on, 361, 501, 524
surface of revolu on, 534, 535
tangent line, 58, 493, 516, 609
direc onal, 705
tangent plane, 711
Taylor Polynomial
deni on, 447
Taylors Theorem, 450
Taylor Series
common series, 462
deni on, 457
equality with genera ng func on, 459
Taylors Theorem, 450
telescoping series, 400, 401
terminal point, 543
total dieren al, 680, 686
sensi vity analysis, 685
total signed area, 196
trace, 536
Trapezoidal Rule, 236, 240
error bounds, 241
triple integral, 774, 785, 787
proper es, 781
unbounded sequence, 389
unbounded set, 658
unit normal vector
aN , 636
and accelera on, 635, 636
and curvature, 646
deni on, 633
in R2 , 635
unit tangent vector
and accelera on, 635, 636
and curvature, 642, 646
aT , 636
deni on, 631
in R2 , 635
unit vector, 549
proper es, 551
standard unit vector, 553
unit normal vector, 633
unit tangent vector, 631
vectorvalued func on
algebra of, 600
arc length, 615
average rate of change, 603

con nuity, 606
deni on, 599
deriva ves, 607, 608, 611
describing mo on, 618
displacement, 602
distance traveled, 626
graphing, 599
integra on, 613
limits, 605
of constant length, 613, 622, 623, 632
projec le mo on, 623, 624
smooth, 610
tangent line, 609
vectors, 543
algebra of, 546
algebraic proper es, 549
component form, 544
cross product, 570, 572, 573
deni on, 543
dot product, 557559
Head To Tail Rule, 547
magnitude, 543
norm, 543
normal vector, 590
orthogonal, 561
orthogonal decomposi on, 565
orthogonal projec on, 563
parallel, 551
Parallelogram Law, 547
resultant, 547
standard unit vector, 553
unit vector, 549, 551
zero vector, 547
velocity, 71, 618
volume, 736, 737, 772
Washer Method, 344, 353
work, 365, 567
Dieren a on Rules
1.
d
(cx) = c
dx
10.
d x
(a ) = ln a ax
dx
19.
d ( 1 )
1
sin x =
dx
1 x2
28.
d
(sech x) = sech x tanh x
dx
2.
d
(u v) = u v
dx
11.
d
1
(ln x) =
dx
x
20.
d ( 1 )
1
cos x =
dx
1 x2
29.
d
(csch x) = csch x coth x
dx
12.
d
1
1
(loga x) =
dx
ln a x
21.
d ( 1 )
1
csc x =
dx
|x| x2 1
30.
d
(coth x) = csch2 x
dx
13.
d
(sin x) = cos x
dx
22.
d ( 1 )
1
sec x =
dx
|x| x2 1
31.
)
d (
1
cosh1 x =
dx
x2 1
d
(u v) = uv + u v
dx
( )
d u
vu uv
4.
=
dx v
v2
3.
5.
d
(u(v)) = u (v)v
dx
14.
d
(cos x) = sin x
dx
23.
d ( 1 )
1
tan x =
dx
1 + x2
32.
)
d (
1
sinh1 x =
dx
x2 + 1
6.
d
(c) = 0
dx
15.
d
(csc x) = csc x cot x
dx
24.
d ( 1 )
1
cot x =
dx
1 + x2
33.
)
d (
1
sech1 x =
dx
x 1 x2
7.
d
(x) = 1
dx
16.
d
(sec x) = sec x tan x
dx
25.
d
(cosh x) = sinh x
dx
34.
)
1
d (
csch1 x =
dx
|x| 1 + x2
8.
d n
(x ) = nxn1
dx
17.
d
(tan x) = sec2 x
dx
26.
d
(sinh x) = cosh x
dx
35.
)
1
d (
tanh1 x =
dx
1 x2
9.
d x
(e ) = ex
dx
18.
d
(cot x) = csc2 x
dx
27.
d
(tanh x) = sech2 x
dx
36.
)
1
d (
coth1 x =
dx
1 x2
Integra on Rules
c f(x) dx = c
1.
f(x) dx
1 dx = x + C
1 n+1
5.
x
+ C, n = 1
xn dx =
n+1
n = 1
6.
ex dx = ex + C
7.
8.
9.
25.
sec2 x dx = tan x + C
26.
csc2 x dx = cot x + C
27.
sec x tan x dx = sec x + C
28.
13.
1
dx = ln |x| + C
x
19.
cos x dx = sin x + C
20.
sinh x dx = cosh x + C
coth x dx = ln | sinh x| + C
csc x cot x dx = csc x + C
cos2 x dx =
( )
1
1
x + sin 2x + C
2
4
( )
1
1
x sin 2x + C
2
4
( )
1
1
x
1
dx
=
tan
21.
+C
x2 + a2
a
a
sin2 x dx =
dx = sin1
17.
x2
16.
a2
15.
22.
14.
18.
sin x dx = cos x + C
sec x dx = ln | sec x + tan x| + C
12.
1
ax dx =
ax + C
ln a
10.
cot x dx = ln | sin x| + C
f(x) g(x) dx =
f(x) dx g(x) dx
3.
0 dx = C
2.
4.
csc x dx = ln | csc x + cot x| + C
( )
x
+C
a
( )
|x|
1
1
+C
23.
dx = sec1
a
a
x x 2 a2
cosh x dx = sinh x + C
24.
tan x dx = ln | cos x| + C
11.
29.
1
x2
a2
1
x2 + a2

dx = ln x + x2 a2 + C

dx = ln x + x2 + a2 + C

1
1 a + x
dx
=
ln
+C
a2 x2
2 a x
)
(
1
1
x
31.
dx = ln
+C
a
x a2 x2
a + a2 x2

1
1
x
+C
32.
dx = ln
a
x x 2 + a2
a + x2 + a2
30.
The Unit Circle
Deni ons of the Trigonometric Func ons
Unit Circle Deni on

(
)
1, 3
( 2) 2
22 , 22
2/3
(
)
23 , 12
3/4
120
5/6 135
(0, 1)
/2
90
60
(
23 , 2
(
4/3
)
22 , 22
(
)
12 , 23
, 22
(
3
1
2
(x, y)
sin = y
/6
30
330
11/6
315
270
3/2
(0, 1)
300
Pythagorean Iden
2
sin x + cos x = 1
2
tan x + 1 = sec x
1 + cot2 x = csc2 x
es
sec =
1
x
tan =
y
x
cot =
x
y
e
us
p
Hy
en
ot
sin =
O
H
csc =
H
O
cos =
A
H
sec =
H
A
tan =
O
A
cot =
A
O
Adjacent
Common Trigonometric Iden
1
y
Right Triangle Deni on
1
2 ,2
(
5/3
)
2
2
2 , 2
(
)
2
1
2, 2
7/4
cos = x
csc =
(1, 0)
Opposite
210
7/6
225
)
1
5/4 240
/4
45
180
/3
150
(1, 0)
1
2,
es
Cofunc on Iden es
(
)
sin
x = cos x
2
(
)
cos
x = sin x
2
(
)
tan
x = cot x
2
Sum to Product Formulas

(
)
(
)
x+y
xy
sin x + sin y = 2 sin
cos
2
2
(
)
(
)
xy
x+y
sin x sin y = 2 sin
cos
2
2
(
)
(
)
x+y
xy
cos x + cos y = 2 cos
cos
2
2
(
)
(
)
x+y
xy
cos x cos y = 2 sin
cos
2
2
Product to Sum Formulas
)
1(
sin x sin y =
cos(x y) cos(x + y)
2
)
1(
cos x cos y =
cos(x y) + cos(x + y)
2
)
1(
sin(x + y) + sin(x y)
sin x cos y =
2
x = sec x
)
sec
x = csc x
2
(
)
cot
x = tan x
2
csc
2
(
Double Angle Formulas

sin 2x = 2 sin x cos x
cos 2x = cos2 x sin2 x
= 2 cos2 x 1
= 1 2 sin2 x
2 tan x
tan 2x =
1 tan2 x
PowerReducing Formulas
1 cos 2x
sin2 x =
2
1 + cos 2x
2
cos x =
2
1
cos
2x
tan2 x =
1 + cos 2x
Even/Odd Iden
es
sin(x) = sin x
cos(x) = cos x
tan(x) = tan x
csc(x) = csc x
sec(x) = sec x
cot(x) = cot x
Angle Sum/Dierence Formulas

sin(x y) = sin x cos y cos x sin y
cos(x y) = cos x cos y sin x sin y
tan x tan y
tan(x y) =
1 tan x tan y
Areas and Volumes

Triangles
Right Circular Cone
h = a sin
Area = 12 bh
Law of Cosines:
c2 = a2 + b2 2ab cos
.
b
Parallelograms
. r
Volume = r2 h
Surface Area =
2rh + 2r2
.
b
Trapezoids
Area =
Surface Area =
r r2 + h2 + r2
Right Circular Cylinder
Area = bh
1
2 (a
Volume = 31 r2 h
Sphere
+ b)h
h
.
Volume = 34 r3
. r
Surface Area =4r2
h
.
b
Circles
General Cone
Area = r2
.
Circumference = 2r
Area of Base = A
Volume = 13 Ah
.
A
Sectors of Circles
General Right Cylinder
in radians
Area =
s = r
1 2
2 r
s
.
Area of Base = A
Volume = Ah
h
.
Algebra
Factors and Zeros of Polynomials
Let p(x) = an xn + an1 xn1 + + a1 x + a0 be a polynomial. If p(a) = 0, then a is a zero of the polynomial and a solu on of
the equa on p(x) = 0. Furthermore, (x a) is a factor of the polynomial.
Fundamental Theorem of Algebra

An nth degree polynomial has n (not necessarily dis nct) zeros. Although all of these zeros may be imaginary, a real
polynomial of odd degree must have at least one real zero.
Quadra c Formula
If p(x) = ax2 + bx + c, and 0 b2 4ac, then the real zeros of p are x = (b
b2 4ac)/2a
Special Factors
x2 a2 = (x a)(x + a)
x3 a3 = (x a)(x2 + ax + a2 )
3
3
2
2
x + a = (x + a)(x ax + a )
x4 a4 = (x2 a2 )(x2 + a2 )
n(n1)
(x + y)n = xn + nxn1 y + 2! xn2 y2 + + nxyn1 + yn
n2 2
(x y)n = xn nxn1 y + n(n1)
y nxyn1 yn
2! x
Binomial
Theorem
2
2
(x y)2 = x2 2xy + y2
(x y)3 = x3 3x2 y + 3xy2 y3
(x y)4 = x4 4x3 y + 6x2 y2 4xy3 + y4
(x + y) = x + 2xy + y2
(x + y)3 = x3 + 3x2 y + 3xy2 + y3
(x + y)4 = x4 + 4x3 y + 6x2 y2 + 4xy3 + y4
Ra onal Zero Theorem
If p(x) = an xn + an1 xn1 + + a1 x + a0 has integer coecients, then every rational zero of p is of the form x = r/s,
where r is a factor of a0 and s is a factor of an .
Factoring by Grouping
acx3 + adx2 + bcx + bd = ax2 (cs + d) + b(cx + d) = (ax2 + b)(cx + d)
Arithme c Opera ons

ab + ac = a(b + c)
(a)
( bc ) =
(a) (d)
b
a
c
ad + bc
+ =
b d
bd
(a)
ad
bc
b
c
a
bc
d
a
( )
b
ab
=
c
c
ab
ba
=
cd
dc
a+b
a b
= +
c
c
c
a
ac
( )=
b
b
c
ab + ac
=b+c
a
Exponents and Radicals

a0 = 1, a = 0
(ab)x = ax bx
( a )x
n
am = am/n
ax
= x
b
ax ay = ax+y
x
1
= x
a
a = a1/2

n
ab = n a n b
ax
= axy
ay
n
a = a1/n
x y
(a ) = a
xy
n
a
a
=
n
b
b
Addi onal Formulas

Summa
on Formulas:
n
n(n + 1)
2
i=1
)2
(
n
n(n + 1)
3
i =
2
i=1
c = cn
i=1
i2 =
i=1
n(n + 1)(2n + 1)
6
i=
Trapezoidal
Rule:
]
x [
f(x1 ) + 2f(x2 ) + 2f(x3 ) + ... + 2f(xn ) + f(xn+1 )
2
a

]
(b a)3 [
with Error
max f (x)
2
12n
b
f(x) dx
Simpsons
Rule:
]
x [
f(x1 ) + 4f(x2 ) + 2f(x3 ) + 4f(x4 ) + ... + 2f(xn1 ) + 4f(xn ) + f(xn+1 )
3
a

]
(b a)5 [
with Error
max f (4) (x)
4
180n
b
f(x) dx
Surface of Revolu on:
Arc Length:
L=
1 + f (x)2 dx
S = 2
f(x)
1 + f (x)2 dx
(where f(x) 0)
S = 2
1 + f (x)2 dx
(where a, b 0)
Work Done by a Variable Force:
Force Exerted by a Fluid:
W=
F(x) dx
F=
w d(y) (y) dy
a
Taylor Series Expansion for f(x):

pn (x) = f(c) + f (c)(x c) +
f (c)
f (c)
f (n) (c)
(x c)2 +
(x c)3 + ... +
(x c)n
2!
3!
n!
Maclaurin Series Expansion for f(x), where c = 0:

pn (x) = f(0) + f (0)x +
f (0) 2 f (0) 3
f (n) (0) n
x +
x + ... +
x
2!
3!
n!
Summary of Tests for Series:

Test
Series
nth-Term
n=1
Geometric Series
Condi on(s) of
Convergence
an
Condi on(s) of
Divergence
Comment
lim an = 0
This test cannot be used to

show convergence.
|r| 1
1
1r
)
( a
Sum =
bn L
|r| < 1
rn
n=0
Telescoping Series
p-Series
n=1
Integral Test
(bn bn+a )
n=1
lim bn = L
1
(an + b)p
an
an
n=0
Limit Comparison
an
n=0
Ra o Test
an
n=0
a(n) dn
a(n) dn
1
Direct Comparison
p1
is convergent
n=0
n=1
p>1
Sum =
is divergent
bn
n=0
converges and
0 an bn
bn
bn
n=0
diverges and
0 bn an
bn
n=0
converges and
lim an /bn 0
an = a(n) must be
con nuous
n=0
diverges and
lim an /bn > 0
an+1
lim
<1
n an
an+1
lim
>1
n an
Also diverges if
lim an /bn =
{an } must be posi ve

Also diverges if
lim an+1 /an =
Root Test
n=0
an
( )1/n
lim an
<1
( )1/n
lim an
>1
{an } must be posi ve

Also diverges if
( )1/n
lim an
=

Calculus Version 2

Uploaded by

Copyright:

Available Formats

Calculus Version 2

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Calculus Version 2

Uploaded by

Copyright:

Available Formats

APEX

Brian Heinold, Ph.D.

Troy Siemers, Ph.D.

Dimplekumar Chalishajar, Ph.D.

Copyright 2014 Gregory Hartman

APEX Aordable Print and Electronic teXts

3 The Graphical Behavior of Func ons

4 Applica ons of the Deriva ve

7 Applica ons of Integra on

8 Sequences and Series

9 Curves in the Plane

Calculus and Polar Func ons . . . . . . . . . . . . . . . . . . . 516

11 Vector Valued Func ons

ons of Several Variables

A Solu ons To Selected Problems

Figure 1.1: sin(x)/x near x = 1.

The expression 0/0 has no value; it is indeterminate. Such an expression gives

lim y = lim f(x) = L.

Figure 1.2: sin(x)/x near x = 0.

1.1 An Introduc on To Limits

Figure 1.3: Values of sin(x)/x with x near

Later we will be able to prove that the limit is exactly 1.

y = (x2 x 6)/(6x2 19x + 3)

on a small interval that contains 3. To numerically approximate the limit, create

Figure 1.4: Values of sin(x)/x with x near

To graphically approximate the limit, graph

Figure 1.5: Graphically approxima ng a

Figure 1.6: Numerically approxima ng a

Figure 1.7: Graphically approxima ng a

Figure 1.8: Numerically approxima ng a

Again we graph f(x) and create a table of its values near x =

Iden fying When Limits Do Not Exist

1.1 An Introduc on To Limits

A graph of f(x) around x = 1 and a table are given Figures

A graph and table of f(x) = 1/(x 1)2 are given in Figures

Figure 1.9: Observing no limit as x 1 in

Figure 1.10: Values of f(x) near x = 1 in

Since f(x) is not approaching a single number, we conclude that

does not exist. .

Figure 1.11: Observing no limit as x 1

Figure 1.12: Values of f(x) near x = 1 in

Figure 1.13: Observing that f(x) = sin(1/x) has no limit as x 0 in Example 5.

Limits of Dierence Quo ents

Figure 1.14: Interpre ng a dierence

1.1 An Introduc on To Limits

Proper understanding of limits is key to understanding calculus. With limits,

Figure 1.15: Secant lines of f(x) at x = 1

Figure 1.16: The dierence quo ent evaluated at values of h near 0.

13. lim f(x), where

14. lim f(x), where

5. In your own words, what is a dierence quo ent?

2. An expression of the form

3. T/F: The limit of f(x) as x approaches 5 is f(5).

lim f(x), where

In Exercises 16 24, a func on f and a value a are

17. f(x) = 9x + 0.06, a = 1

20. f(x) = 4x2 + 5x 1, a = 3

22. f(x) = sin x, a =

1.2 Epsilon-Delta Deni on of a Limit

Epsilon-Delta Deni on of a Limit