Elementary Calculus

Elementary
Calculus
v02
2g
0 v02
g
Michael Corral
Elementary Calculus
Michael Corral
Schoolcraft College
About the author:
Michael Corral is an Adjunct Faculty member of the Department of Mathematics at School-
craft College. He received a B.A. in Mathematics from the University of California, Berkeley,
and received an M.A. in Mathematics and an M.S. in Industrial & Operations Engineering
from the University of Michigan.
This text was typeset in LATEX with the KOMA-Script bundle, using the GNU Emacs
text editor on a Fedora Linux system. The graphics were created using TikZ and Gnuplot.
Copyright © 2020 Michael Corral.

Permission is granted to copy, distribute and/or modify this document under the terms of the
GNU Free Documentation License, Version 1.3 or any later version published by the Free
Software Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover
Texts.
Preface
This book covers calculus of a single variable. It is suitable for a year-long (or two-semester)
course, normally known as Calculus I and II in the United States. The prerequisites are high
school or college algebra, geometry and trigonometry. The book is designed for students in
engineering, physics, mathematics, chemistry and other sciences.
One reason for writing this text was because I had already written its sequel, Vector Cal-
culus. More importantly, I was dissatisfied with the current crop of calculus textbooks, which
I feel are bloated and keep moving further away from the subject’s roots in physics. In addi-
tion, many of the intuitive approaches and techniques from the early days of calculus—which
I think often yield more insights for students—seem to have been lost.
I agree with the views of the late Russian mathematician V.I. Arnold on teaching mathe-
matics, in particular the idea that “Mathematics is the part of physics where experiments are
cheap.”1 The ties to physics are especially important in calculus, so this book tries to introduce
new concepts with physical motivations (what other motivations can there be?). The book con-
tains exercises and examples that I hope will adequately prepare students who continue on in
physics and engineering.2
Perhaps controversially, the book uses infinitesimals, making it a bit of a “throwback” or
“retro” calculus text. My justification for this heretical act was purely pedagogical: infinitesi-
mals make learning calculus easier, and their use aligns more with the way students will see
calculus in their physics, chemistry and other science classes and textbooks (where infinitesi-
mals are employed liberally). This might ruffle some feathers among mathematical “purists,”
but they are not the main audience for this book. That said, the book is still compatible with the
usual limit-based approach, so an instructor could simply ignore the parts involving infinites-
imals and teach the material as he or she normally would. I did not want to be dogmatic, so I
used infinitesimals where I thought it made sense, and used limits where appropriate (e.g. in
discussing continuity, series). Again, pedagogy was my priority.
The exercises at the end of each section are divided into three categories: A, B and C. The
A exercises are mostly of a routine computational nature, the B exercises are slightly more
involved, and the C exercises usually require some effort or insight to solve. A crude way of
1 A RNOLD, V.I., “On Teaching Mathematics”, Russian Math. Surveys 53 (1998), No. 1, 229-236. An HTML version
is at https://www.uni-muenster.de/Physik.TP/~munsteg/arnold.html
2 The book covers some of the types of problems and techniques for solving them that such students will likely
encounter. Facility with using named constants (e.g. c, h, T ) is also emphasized.
iii
iv P REFACE
describing A, B and C would be “Easy”, “Moderate” and “Challenging”, respectively. However,

many of the B exercises are easy and not all the C exercises are difficult. Appendix A provides
answers and hints to many of the odd-numbered and some of the even-numbered exercises.
A few exercises require the student to write a computer program to solve numerical approx-
imation problems (e.g. numerical methods for approximating definite integrals). Algorithms
are presented in pseudocode, with code implementations in various languages (primarily Java,
but also Python, Octave, Sage). I hope the code comments will help the reader figure out what
is being done, regardless of familiarity with those languages. Students are free to implement
solutions using the language of their choice. There are no dedicated “calculator exercises,” as
those have been rendered pointless by modern computing (with which students need to become
acquainted).
Stylistically I made a conscious effort to break from an unfortunate but all too common
mode of writing in mathematics texts, lamented in the preface of a physics book: “Nothing
is more repellent to normal human beings than the clinical succession of definitions, axioms,
and theorems generated by the labours of pure mathematicians.”3 I have been guilty of that
sin myself, but I have changed my ways and banished all traces of that sort of thing from this
book. So you won’t find Definition 1.2, Theorem 3.3, Corollary 4.6, Lemma 5.7, Axiom 1B,
etc. Instead, I tried to borrow the best of the styles from the physics and foreign languages
textbooks I enjoyed so much in college. I also deliberately avoided what the author Gore Vidal
called the “we-ness” that prevails in academic writing. There is no good reason for the “royal
we” in a textbook, and it comes off as a bit pompous, so we won’t use it.
This book is released under the GNU Free Documentation License (GFDL), which allows
others to not only copy and distribute the book but also to modify it. For more details, see
the included copy of the GFDL. So that there is no ambiguity on this matter, anyone can
make as many copies of this book as desired and distribute it as desired, without needing my
permission. The PDF version will always be freely available to the public at no cost (go to
http://www.mecmath.net/calculus). Feel free to contact me at [email protected] for
any questions on this or any other matter regarding the book. I welcome your feedback.
Schoolcraft College M ICHAEL C ORRAL

December 2020
3 Z IMAN, J.M., Elements of Advanced Quantum Theory, Cambridge, U.K.: Cambridge University Press, 1969.
Contents
Preface iii
1 The Derivative 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 The Derivative: Limit Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 The Derivative: Infinitesimal Approach . . . . . . . . . . . . . . . . . . . . . . . 15
1.4 Derivatives of Sums, Products and Quotients . . . . . . . . . . . . . . . . . . . . 21
1.5 The Chain Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.6 Higher Order Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2 Derivatives of Common Functions 37

2.1 Inverse Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.2 Trigonometric Functions and Their Inverses . . . . . . . . . . . . . . . . . . . . 41
2.3 The Exponential and Natural Logarithm Functions . . . . . . . . . . . . . . . . 45
2.4 General Exponential and Logarithmic Functions . . . . . . . . . . . . . . . . . . 53
3 Topics in Differential Calculus 56

3.1 Tangent Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
3.2 Limits: Formal Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.3 Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.4 Implicit Differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.5 Related Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.6 Differentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4 Applications of Derivatives 90
4.1 Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.2 Curve Sketching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
4.3 Numerical Approximation of Roots of Functions . . . . . . . . . . . . . . . . . . 109
4.4 The Mean Value Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5 The Integral 124

5.1 The Indefinite Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
v
vi C ONTENTS
5.2 The Definite Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132

5.3 The Fundamental Theorem of Calculus . . . . . . . . . . . . . . . . . . . . . . . 140
5.4 Integration by Substitution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
5.5 Improper Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
6 Methods of Integration 159

6.1 Integration by Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
6.2 Trigonometric Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
6.3 Trigonometric Substitutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
6.4 Partial Fractions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
6.5 Miscellaneous Integration Methods . . . . . . . . . . . . . . . . . . . . . . . . . . 185
6.6 Numerical Integration Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
7 Analytic Geometry and Plane Curves 202

7.1 Ellipses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
7.2 Parabolas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
7.3 Hyperbolas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
7.4 Translations and Rotations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
7.5 Hyperbolic Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
7.6 Parametric Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
7.7 Polar Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
8 Applications of Integrals 252

8.1 Area Between Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
8.2 Average Value of a Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
8.3 Arc Length and Curvature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
8.4 Surfaces and Solids of Revolution . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
8.5 Applications in Physics and Statistics . . . . . . . . . . . . . . . . . . . . . . . . 277
9 Infinite Sequences and Series 286

9.1 Sequences and Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
9.2 Tests for Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
9.3 Alternating Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
9.4 Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
9.5 Taylor’s Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308
Appendix A Answers and Hints to Selected Exercises 315
GNU Free Documentation License 321
History 330
Index 332
The Greek Alphabet
Letters Name Letters Name Letters Name

A α alpha I ι iota P ρ rho
B β beta K κ kappa Σ σ sigma
Γ γ gamma Λ λ lambda T τ tau
∆ δ delta M µ mu Υ υ upsilon
E ǫ epsilon N ν nu Φ φ phi
Z ζ zeta Ξ ξ xi X χ chi
H η eta O o omicron Ψ ψ psi
Θ θ theta Π π pi Ω ω omega
Mathematical Notation
Symbol Meaning Example

⇒ if...then; implies | x | > 1 ⇒ x2 > 1
⇔ if and only if; two-way implication | x | > 1 ⇔ x2 > 1
iff if and only if; two-way implication | x | > 1 iff x2 > 1
; does not imply | x| > 1 ; x > 1
∃ there exists ∃ a number c > 0
Ø there does not exist Ø x such that x2 < 0
∃! there exists a unique ∃! x such that 2x − 1 = 3
p
∀ for every ∀ x ≥ 0, x is a real number
≡ is identically equal to f ≡ 0 ⇒ f (x) = 0 for all x
∝ is proportional to y ∝ x2 ⇒ y = kx2 for some k
⊆ is a subset of {0, 1} ⊆ {0, 1, 2}
∈ is an element of 1 ∈ {1, 2, 3}
∉ is not an element of 1 ∉ {2, 3}
∪ union of sets {0, 1} ∪ {2, 3} = {0, 1, 2, 3}
∩ intersection of sets {0, 1} ∩ {1, 2} = {1}
∅ empty set {0, 1} ∩ {2, 3} = ∅
∴ therefore ∴ n must exist
CHAPTER 1
The Derivative
1.1 Introduction
Calculus can be thought of as the analysis of curved shapes.1 Its development grew out of
attempts to solve physical problems. For example, suppose that an object at rest 100 ft above
the ground is dropped. Ignoring air resistance and wind, the object will fall straight down until
it hits the ground (see Figure 1.1.1(a)). As will be proved later, t seconds after being dropped
the object will be s = s(t) = −16t2 + 100 ft above the ground. The object will thus hit the ground
after 2.5 seconds (when s = 0). While the object’s path is a straight line, the graph of its position
s above the ground as a function of time t is curved, part of a parabola (see Figure 1.1.1(b)).
t = 0 sec 100
s = −16 t2 + 100
100 ft object falling
t
t = 2.5 sec 0 2.5
(a) Path of the object (b) Position s as a function of time t
Figure 1.1.1 An object dropped from 100 ft above the ground
How fast is the object moving before it hits the ground? This is where calculus comes in. The
solution, presented now, will motivate much of this chapter.
1 It is more than that, of course, but that definition puts us in good company: the first European textbook on
calculus, written by the French mathematician Guillaume de l’Hôpital in 1696, was titled Analyse des Infiniment
Petits pour l’Intelligence des Lignes Courbes (which translates as Analysis of the Infinitely Small for Understanding
Curved Lines). That book (in French) can be obtained freely in electronic form at https://archive.org
1
2 Chapter 1 • The Derivative §1.1
First, the object travels 100 ft in 2.5 seconds, so its average speed in that time is
distance traveled 100 ft

= = 40 ft/s,
time elapsed 2.5 seconds
and its average velocity in that time is
change in position final position − initial position 0 ft − 100 ft

= = = −40 ft/s.
change in time end time − start time 2.5 sec − 0 sec
Unlike speed, velocity takes direction into account. Thus, the object’s downward motion means
it has negative velocity. Positive velocity implies upward motion.
Using the idea of average velocity over an interval of time, there is a natural way to define
the object’s instantaneous velocity at a particular instant of time t:
1. Find the average velocity over an interval of time.
2. Let the interval become smaller and smaller indefinitely, shrinking to a point t. If the
average velocity over that smaller and smaller interval approaches some value, call that
value the instantaneous velocity at time t.
Figure 1.1.2 below shows how to choose the interval: for any time t between 0 and 2.5, use
the interval [t, t + ∆ t], where ∆ t (pronounced “delta t”) is a small positive number. So ∆ t is the
change in time over the interval; denote by ∆ s the change in the position s over that interval.
100
s( t)
∆s s( t + ∆ t) − s( t)
average velocity = =
∆s ∆t ∆t
s( t + ∆ t)
∆t
t
0 t t + ∆t 2.5
∆s
Figure 1.1.2 Average velocity ∆t over the interval [ t, t + ∆ t]
The average velocity of the object over the interval [t, t + ∆ t] is ∆s

∆t , so since s(t) = −16t2 + 100:
Introduction • Section 1.1 3
∆s s(t + ∆ t) − s(t)
=
∆t ∆t
−16(t + ∆ t)2 + 100 − (−16t2 + 100)

=
∆t
−16t2 − 32t∆ t − 16(∆ t)2 + 100 + 16t2 − 100

=
∆t
−32t∆ t − 16(∆ t)2 ∆✚t (−32t − 16∆ t)

✚
= =
∆t ∆✚t
✚
= −32t − 16∆ t ,
Now let the interval [t, t + ∆ t] get smaller and smaller indefinitely—that is, let ∆ t get closer and
s
closer to 0. Then the average velocity ∆ ∆ t = −32t − 16∆ t gets closer and closer to −32t − 0 = −32t.
Thus, the object has instantaneous velocity −32t at time t. This calculation can be interpreted
s
as taking the limit of ∆∆ t as ∆ t approaches 0, written as follows:
instantaneous velocity at t = limit of average velocity over [t, t + ∆ t] as ∆ t approaches to 0
∆s
= lim
∆ t→0 ∆t
= lim (−32t − 16∆ t)

∆ t→0
= −32t − 16(0)
= −32t
s
Notice that ∆ t is not replaced by 0 in the ratio ∆∆ t until after doing as much cancellation as
possible. Notice also that the instantaneous velocity of the object varies with t, as it should
(why?). In particular, at the instant when the object hits the ground at time t = 2.5 sec, the
instantaneous velocity is −32(2.5) = −80 ft/s.
If this makes sense so far, then you understand the crux of the idea of what a limit is and
how to calculate a limit. The instantaneous velocity v(t) = −32t is called the derivative of the
position function s(t) = −16t2 + 100. Calculating derivatives, analyzing their properties, and
using them to solve various problems are part of differential calculus.
What does this have to do with curved shapes? Instantaneous velocity is a special case of an
instantaneous rate of change of a function; in this case the instantaneous rate of change
of the position (height above the ground) of the object. Similar to how the rate of change of a
line is its slope, the instantaneous rate of change of a general curve represents the slope of
the curve. For example, the parabola s(t) = −16t2 + 100 has slope −32t for all t. Note that the
slope of this curve varies (as a function of t), unlike the slope of a straight line.
Finding the area inside curved regions is another type of problem that calculus can solve.
The basic idea is to use simpler regions—rectangles—whose areas are known, then use those
to approximate the area inside the curved region. One such method is to draw more and more
rectangles of diminishing widths inside the curved region,2 so that the sums of their areas
approach the area of the curved region. Figure 1.1.3 shows an example with four rectangles to
approximate the area under a curve y = f (x) over an interval [a, b] on which f (x) ≥ 0.
y = f ( x)
x
a b
Figure 1.1.3 The area of a curved region
The limit of these sums of rectangular areas is called an integral. The study and application
of integrals are part of integral calculus. Perhaps the most remarkable result in calculus is
that there is a connection between derivatives and integrals—the Fundamental Theorem of
Calculus, discovered in the 17th century, independently, by the two men who invented calculus
as we know it: English physicist, astronomer and mathematician Isaac Newton (1642-1727)
and German mathematician and philosopher Gottfried Wilhelm von Leibniz (1646-1716).
Calculus makes extensive use of infinite sequences and series. An infinite series is just a
sum of an infinite number of terms. For example, it will be shown later in the text that
π 1 1 1 1
= 1 − + − + − ··· , (1.1)
4 3 5 7 9
where the sum on the right involves an infinite number of terms. A power series is a partic-
ular type of infinite series applied to functions; it can be thought of as a polynomial of infinite
degree. For example, the trigonometric function sin x does not appear to be a polynomial. But
it turns out that sin x has a power series representation as
x3 x5 x7 x9
sin x = x − + − + − ··· , (1.2)
3! 5! 7! 9!
where again the sum continues infinitely, and the formula holds for all x (in radians).
The idea of replacing a function by its power series played an important role throughout the
development of calculus, and is a powerful technique in many applications.
All the functions in this text will be functions of a single real variable—that is, the values
that the variable can take are real numbers. Below is some standard notation for commonly-
used sets of numbers:
2 It will be shown later (in Chapter 5) that the rectangles do not have to be completely inside the region.
N = the set of all natural numbers, i.e. the set of nonnegative integers: 0, 1, 2, 3, 4, . . .
Z = the set of all integers: 0, ±1, ±2, ±3, ±4, . . .

m
Q = the set of all rational numbers , where m and n are integers, with n 6= 0
n
R = the set of all real numbers
Note that N ⊂ Z ⊂ Q ⊂ R.
The set of real numbers consists of the rational numbers
p together with numbers that are
not rational, called irrational numbers. For example, 2 is irrational. That is, 2 is not the
square of a rational number. In fact, if the square of a rational number q were an integer,
then q itself would have to be an integer: write q as m/n, where m and n are positive integers
with no common positive integer divisors other than 1. Since q2 = m2 /n2 simply duplicates the
integer divisors of m and n, then q2 can be an integer only if n = 1, i.e. q is an integer. Clearly
2 is not the square of an integer,
p p and p thus
p itp cannot
p be the square of a rational number. This
argument also shows that 3, 5, 6, 7, 8, 10, and so on, are irrational.3
It turns out that there are far more irrational numbers–and hence real numbers—than ra-
tional numbers. In fact, whereas the rational numbers can be listed in a sequence (i.e. first,
second, third, etc.), the set of real numbers cannot.4 For example, in the closed interval [0, 1]
there is no “next” real number after the number 0. Thus, some infinite sets are larger than
others—R is larger than Q. Intervals such as [0, 1] or R itself are examples of a continuum
of objects, i.e. no gaps exist.5 A famous unsolved problem in mathematics—the Continuum
Hypothesis—is whether an infinite set exists that is larger in size than Q but smaller than R.
Infinity is an important notion in calculus. Whether it is the idea of infinitely large or in-
finitesimally small, calculus attempts to give the idea some mathematical meaning (typically
by way of limits).6 The mathematical use of infinity has been a subject of philosophical debate.7
Though several centuries old, calculus was the beginning of modern mathematics. Classical
mathematics (e.g. algebra, geometry, trigonometry)—whose origins date back to the ancient
Babylonians, Egyptians, and Greeks—was concerned mostly with the study of static quantities.
Calculus produced a way to analyze dynamic (i.e. changing) quantities. The period from the
17th through the 19th century also saw revolutionary advances in physics, chemistry, biology
and other sciences. The birth of calculus was one part of that qualitative leap.
3 This argument is due to the British philosopher Bertrand Russell (1872-1970). For an alternative proof that
p
2
is irrational, see pp. 97-98 in G ELFAND, I.M. AND A. S HEN, Algebra, Boston: Birkhäuser, 1993.
4 For a proof see Ch.1 in K AMKE , E., Theory of Sets, New York: Dover Publications, Inc., 1950.
5 For a study of the structure of the real number system, see B URRILL , C.W., Foundations of Real Numbers, New
York: McGraw-Hill Book Company, 1967.

6 Not everyone agrees that calculus does this satisfactorily. For example, for an alternative development of ba-
sically the same material in “standard” calculus but without the use of limits—called infinitesimal analysis—see
K EISLER , H.J., Elementary Calculus: An Infinitesimal Approach, Boston: Prindle, Weber & Schmidt, 1976.
7 For example, see the essays by L. E. J. Brouwer, Hermann Weyl and David Hilbert in H EIJENOORT, J. VAN, From
Frege to Gödel: A Source Book in Mathematical Logic, 1879-1931, Cambridge, MA: Harvard University Press, 1967.
Exercises
A
For Exercises 1-4, suppose that an object moves in a straight line such that its position s after time t
is the given function s = s( t). Find the instantaneous velocity of the object at a general time t ≥ 0. You
should mimic the earlier example for the instantaneous velocity when s = −16 t2 + 100.
1. s = t2 2. s = 9.8 t2 3. s = −16 t2 + 2 t 4. s = t3
¡ ¢
5. By equation (1.1), π = 4 1 − 13 + 1
5 − 1
7 + 19 − · · · , where the n th term in the sum inside the
(−1)n+1 8
parentheses is 2 n−1 (starting at n = 1). So the first approximation of π using this formula is
¡ ¢
π ≈ 4 (1) = 4.0, and the second approximation is π ≈ 4 1 − 31 = 8/3 ≈ 2.66667. Continue like this until
two consecutive approximations have 3 as the first digit before the decimal point. How many terms
in the sum did this require? Be careful with rounding off in the approximations.
B
6. In elementary geometry you learned that the area inside a circle of radius r > 0 is πr2 (that formula
will be proved later in the text). So in particular, let C be a circle of radius 1. Then the area inside
C is π. That area can be approximated by Eudoxus’ method of exhaustion.9 The idea is to inscribe
regular polygons inside the circle, i.e. the vertexes of the polygons touch C . Recall from geometry
that a polygon is regular if its sides are of equal length. By increasing the number of sides of the
polygons, the areas inside the polygons will approach the area (π) of C . This was an early attempt
at using what is now called a limit.10
C C
1 1
Figure 1.1.4 Inscribed square Figure 1.1.5 Inscribed regular hexagon
(a) Inscribe a square inside C , as in Figure 1.1.4. Show that the area inside the square is 2. This is
a poor approximation of π = 3.14159265..., obviously.
(b) Inscribe a regular
p
hexagon (6-sided) inside C , as in Figure 1.1.5. Show that the area inside the
3 3
hexagon is 2 ≈ 2.59807621. This is a slightly better—though still poor—approximation of π.
(c) Inscribe a regular dodecagon (12-sided) inside C . Show that the area inside the dodecagon is 3.
It thus takes 12 sides for the approximation to get the first digit of π correct.
(d) Inscribe a regular 100-sided polygon inside C . Show that the area inside this polygon is approx-
imately 3.13952598. This is getting closer to π.
(e) Show that
µ the ¶general formula for the area inside a regular n-sided polygon inscribed inside C
n 360◦
is sin . (Hint: The double-angle identity sin 2θ = 2 sin θ cos θ might help.)
2 n
8 This, by the way, is a terrible formula for calculating π; getting just the 3.14 part requires 119 terms in the sum!
9 Originally due to another ancient Greek mathematician, Antiphon (ca. 430 B . C .)
10 The great ancient Greek mathematician, physicist and astronomer Archimedes (ca. 287-212 B . C .) used this
method, together with circumscribed regular polygons, to calculate π.

C
7. What is the flaw in the following “proof ” that π = 4?:
Step 1: Draw a square around a circle of diameter d = 1. The circumference
of the circle is thus π d = π, and the perimeter of the square is 4.
d=1
Step 2: Remove corners from the square as shown in the picture on the right,
so that four new corners touch the circle. Notice that the perimeter of the
resulting polygon is still 4, since the lengths of the removed corner pieces are
duplicated in the new polygon, so that the lengths of all the vertical sides add d=1
up to 2 while the lengths of all the horizontal sides add up to 2.
Step 3: Remove corners from the polygon in Step 2, as shown in the picture
on the right, so that eight new corners touch the circle. The perimeter of the
resulting polygon is again still 4.
d=1
Step 4: Continue this procedure indefinitely, with each successive polygon

still having a perimeter of 4 and becoming increasingly indistinguishable from
the circle. Since the perimeters of the polygons always equal 4 and approach
the circle’s circumference (π), then π must equal 4. d=1
8. An infinite set is countable if its members can be put into a one-to-one correspondence with the
members of N, the set of natural numbers (0, 1, 2, 3, 4,... ). Clearly N is itself countable. The set Z of
all integers is also countable, by means of the following one-to-one correspondence with N:
N 0 1 2 3 4 5 6 7 8 ...
Z 0 1 -1 2 -2 3 -3 4 -4 ...
Show that Q (the set of all rational numbers) is countable. (Hint: The above correspondence for Z is
an infinite list in one dimension (the horizontal direction). For Q think two-dimensionally.)
1.2 The Derivative: Limit Approach

The following definition generalizes the example from the previous section (concerning instan-
taneous velocity) to a general function f (x):
The derivative of a real-valued function f (x), denoted by f ′ (x), is
∆f f (x + ∆ x) − f (x)
f ′ (x) = lim = lim (1.3)
∆ x→0 ∆x ∆ x→0 ∆x
for x in the domain of f , provided that the limit exists.11
For a general function f (x), the derivative f ′ (x) represents the instantaneous rate of
change of f at x, i.e. the rate at which f changes at the “instant” x. For the limit part of
the definition only the intuitive idea of how to take a limit—as in the previous section—is
needed for now. Notice that the above definition makes the derivative f ′ itself a function of the
variable x. The function f ′ can be evaluated at specific values of x, or you can write its general
formula f ′ (x).
The (instantaneous) velocity of an object as the derivative of the object’s position as a func-
tion of time is only one physical application of derivatives. There are many other examples:
Field Function Derivative Field Function Derivative
Physics position velocity Physics angular momentum torque
velocity acceleration Engineering electric charge electric current
momentum force magnetic flux induced voltage
work power Economics profit marginal profit
The limit definition can be used for finding the derivatives of simple functions.
Example 1.1
Find the derivative of the function f ( x) = 1.
Solution: By definition, f ( x) = 1 for all x, so:
f ( x + ∆ x) − f ( x)
f ′ ( x) = lim
∆ x→0 ∆x
1 − 1
= lim
∆ x→0 ∆x
0
= lim
∆ x→0 ∆x
= lim 0
∆ x→0
f ′ ( x) = 0
11 Recall that the domain of f is the set of all numbers x such that f ( x) is defined.
The Derivative: Limit Approach • Section 1.2 9
Notice in the above example that replacing ∆ x by 0 was unnecessary when taking the limit,
f ( x +∆ x ) − f ( x )
since the ratio ∆x simplified to 0 before taking the limit, and the limit of 0 is 0 regard-
less of what ∆ x approaches. In fact, the answer—namely, f ′ (x) = 0 for all x—should have been
obvious without any calculations: the function f (x) = 1 is a constant function, so its value (1)
never changes , and thus its rate of change is always 0. Hence, its derivative is 0 everywhere.
Replacing the constant 1 by any constant yields the following important result:
The derivative of any constant function is 0.
The above discussion shows that the calculation in Example 1.1 was unnecessary. Consider
another example where no calculation is required to find the derivative: the function f (x) = x.
The graph of this function is just the line y = x in the x y-plane, and the rate of change of a line
is a constant, called its slope. The line y = x has a slope of 1, so the derivative of f (x) = x is
f ′ (x) = 1 for all x. The formal calculation of the derivative, though unnecessary, verifies this:
f (x + ∆ x) − f (x) (x + ∆ x) − x ∆x
f ′ (x) = lim = lim = lim = lim 1 = 1
∆ x→0 ∆x ∆ x→0 ∆x ∆ x→0 ∆ x ∆ x→0
Recall that a function whose graph is a line is called a linear function. For a general
linear function f (x) = mx + b, where m is the slope of the line and b is its y-intercept, the same
argument as above for f (x) = x yields the following result:
The derivative of any linear function is the slope of the line itself:
If f (x) = mx + b then f ′ (x) = m for all x.
The function f (x) = 1 from Example 1.1 is the special case where m = 0 and b = 1; its graph
is a horizontal line, so its slope (and hence its derivative) is 0 for all x. Likewise, the function
f (x) = 2x − 1 represents a line of slope m = 2, so its derivative is 2 for all x. Figure 1.2.1 shows
these and other linear functions y = f (x).
y
y y f ( x) = 2 x − 1
2 f ( x) = − x + 2
f ( x) = 1 slope = 0 slope = 2
slope = −1
1 f ′ ( x) = 0 f ′ ( x) = 2
f ′ ( x ) = −1
x 0 x 0 x
0 1 2
2
−1
Figure 1.2.1 Slopes and derivatives of lines

Linear functions have a constant derivative—the constant being the slope of the line. The
converse turns out to be true: a function with a constant derivative must be a linear function.12
What types of functions do not have constant derivatives? The previous section discussed
such a function: the parabola s(t) = −16t2 + 100, whose derivative s′ (t) = −32t is clearly not a
constant function. In general, functions that represent curves (i.e. not straight lines) do not
change at a constant rate—that is precisely what makes them curved. So such functions do not
have a constant derivative.
Example 1.2
1
Find the derivative of the function f ( x) = . Also, find the instantaneous rate of change of f at x = 2.
x
Solution: For all x 6= 0, the derivative f ′ ( x) is:
f ( x + ∆ x) − f ( x)
f ′ ( x) = lim
∆ x→0 ∆x
1 1
−
x + ∆x x 0
= lim → , so simplify the ratio before plugging in ∆ x = 0,
∆ x→0 ∆x 0
x − ( x + ∆ x)
( x + ∆ x) x
= lim (after getting a common denominator)
∆ x→0 ∆x
−✟∆✟x
= lim
∆ x→0 ∆✟
✟x( x + ∆ x) x
−1 −1
= lim =
∆ x→0 ( x + ∆ x) x ( x + 0) x
1
f ′ ( x) = −
x2
The instantaneous rate of change of f at x = 2 is just the derivative f ′ ( x) evaluated at x = 2, that is,
f ′ (2) = − 212 = − 41 .
Notice that the instantaneous rate of change f ′ (2) = − 41 in the y

above example is a negative number. This should make sense, since 1
f ( x) =
the function f (x) = 1x is changing in the negative direction at x = 2; x
that is, f (x) is decreasing in value at x = 2. This is plain to see from

0
the graph of f (x) = 1x shown on the right. In fact, for all x 6= 0 the x
2
function f (x) = 1x is decreasing as x grows.13 This is reflected in the
derivative f ′ (x) = − x12 being negative for all x 6= 0. In general, a nega-
tive derivative means that the function is decreasing, while a positive
derivative means that it is increasing.
12 This will be proved in Chapter 5.
13 In this text, the rate of change of f ( x) is always taken in the direction of increasing x, i.e. in the positive x
direction.
The problem with using the limit definition to find the derivative of a curved function is that
the calculations require more work, as the above example shows. As the functions become more
complicated those calculations can become difficult or even impossible. And though limits have
not yet been defined formally, for now the intuitively obvious idea of limits suffices, namely:
For a real number a and a real-valued function f (x), say that the limit of f (x) as x ap-
proaches a equals the number L, written as
lim f (x) = L ,
x →a
if f (x) approaches L as x approaches a.

Equivalently, this means that f (x) can be made as close as you want to L by choosing x
close enough to a. Note that x can approach a from any direction.
Below are some simple rules for limits, which will be proved later:
Rules for Limits: Suppose that a is a real number and that f (x) and g(x) are real-valued
functions such that limx³→a f (x) and
´ lim
³ x→a g(x)´ both exist. Then:
(a) lim ( f (x) + g(x)) = lim f (x) + lim g(x)
x →a x →a x →a
³ ´ ³ ´
(b) lim ( f (x) − g(x)) = lim f (x) − lim g(x)
³ ´
(c) lim (k · f (x)) = k · lim f (x) for any constant k
x →a x →a
³ ´ ³ ´
(d) lim ( f (x) · g(x)) = lim f (x) · lim g(x)
f (x) lim f (x)

x →a
(e) lim = , if lim g(x) 6= 0
x →a g(x) lim g(x) x →a
x →a
The above rules say that the limit of sums, differences, constant multiples, products, and
quotients is the sum, difference, constant multiple, product, and quotient, respectively, of the
limits. This seems intuitively obvious.
These rules can be used for finding other expressions for the derivative. The quantity ∆ x rep-
resents a small number—positive or negative—that approaches 0, but it is common in mathe-
matics texts to use the letter h instead:14
f (x + h) − f (x)
f ′ (x) = lim (1.4)
h→0 h
14 Physics texts typically prefer the delta notation, since ∆ x represents a small change in some physical quantity x.
Another formulation is to set h = w − x in formula (1.4), which yields

f (x + h) − f (x) f (x + (w − x)) − f (x)
f ′ (x) = lim = lim ,
h→0 h w− x→0 w − x
so that
f (w) − f (x)
f ′ (x) = lim (1.5)
w→ x w − x
since w − x approaches 0 if and only if w approaches x. Another formulation replaces h by − h:
f (x + h) − f (x) f (x + − h) − f (x) − ( f (x) − f (x − h))
f ′ (x) = lim = lim = lim ,
h→0 h − h→0 −h − h→0 −h
and thus
f (x) − f (x − h)
f ′ (x) = lim (1.6)
h→0 h
since − h approaches 0 if and only if h approaches 0. The above formulations did not use the
Limit Rules, but the following result does:
Suppose that f ′ (x) exists. Then
f (x + h) − f (x − h)
f ′ (x) = lim . (1.7)
h→0 2h
f (x + h) − f (x) f (x) − f (x − h)
Proof: Since f ′ (x) = lim = lim by formulas (1.4) and (1.6),
h→0 h h→0 h
then Limit Rule (c) shows that
1 ′ f (x + h) − f (x) f (x) − f (x − h)
f (x) = lim = lim .
2 h→0 2h h→0 2h
Now use the idea that a − b = (a − c) + (c − b) for all a, b, and c to write:
f (x + h) − f (x − h) ( f (x + h) − f (x)) + ( f (x) − f (x − h))

lim = lim
h→0 2h h→0 2h
f (x + h) − f (x) f (x) − f (x − h)
= lim + lim (by Limit Rule (a))
h→0 2h h →0 2h
1 ′ 1
= · f (x) + · f ′ (x)
2 2
= f ′ (x)
QED
As an example of using these different formulations, recall that a function f is even if

f (− x) = f (x) for all x in the domain of f , and f is odd if f (− x) = − f (x) for all x in its domain.
For example, x2 , x4 , and cos x are even functions; x, x3 , and sin x are odd functions. The
following result is often useful:
The derivative of an even function is an odd function.

The derivative of an odd function is an even function.
To prove the first statement—the second is an exercise—suppose that f is an even function

and that f ′ (x) exists for all x in its domain. Then
f (− x + h) − f (− x)
f ′ (− x) = lim by formula (1.4) with x replaced by − x
h→0 h
f (−(x − h)) − f (− x)
= lim
h→0 h
f (x − h) − f (x)
= lim since f is even
h→0 h
− ( f (x) − f (x − h))
= lim
h→0 h
f (x) − f (x − h)
= − lim by Limit Rule (c), so
h→0 h
f ′ (− x) = − f ′ (x) by formula (1.6),
which shows that f ′ is an odd function.

Derivatives do not always exist, as the following example shows.
Example 1.3
Let f ( x) = | x |. Show that f ′ (0) does not exist. y
Solution: Recall that the absolute value function f ( x) = | x | is defined as
(
x if x ≥ 0 y = −x y=x
f ( x) = | x | =
−x if x < 0 x
0
The graph consists of two lines meeting at the origin. For x ≥ 0 the graph f ( x) = | x |
is the line y = x, which has slope 1. For x ≤ 0 the graph is the line y = − x,
which has slope -1. These lines agree in value ( y = 0) at x = 0, but their
slopes do not agree in value at x = 0. Therefore the derivative of f does not exist at x = 0, since the
derivative of a curve is just its slope. A more “formal” proof (which amounts to the same argument) is
outlined in the exercises.
If the derivative f ′ (x) exists then f is differentiable at x. A differentiable function is

one that is differentiable at every point in its domain. For example, f (x) = x is a differentiable
function, but f (x) = | x | is not differentiable at x = 0. The act of calculating a derivative is called
differentiation. For example, differentiating the function f (x) = x yields f ′ (x) = 1.
Exercises
Note: For all exercises, you can use anything discussed so far (including previous exercises).
A
For Exercises 1-11, find the derivative of the given function f ( x) for all x (unless indicated otherwise).
1. f ( x) = 0 2. f ( x) = 1 − 3 x 3. f ( x) = ( x + 1)2 4. f ( x) = 2 x2 − 3 x + 1
1 −1 1
5. f ( x) = x +1 , for all x 6= −1 6. f ( x) = x +1 , for all x 6= −1 7. f ( x) = x2
, for all x 6= 0
p
8. f ( x) = x, for all x > 0 (Hint: Rationalize the numerator in the definition of the derivative.)
p p p
9. f ( x) = x + 1, for all x > −1 10. f ( x) = x2 + 1 11. f ( x) = x2 + 3 x + 4
12. In Exercise 8 the point x = 0 was excluded when calculating f ′ ( x), even though x = 0 is in the
p
domain of f ( x) = x. Can you explain why x = 0 was excluded?
B
f ( x ) − f ( w)
13. Show that for all functions f such that f ′ ( x) exists, f ′ ( x) = lim .
w→ x x − w
14. True or false: If f and g are differentiable functions on an interval (a, b) and f ( x) < g( x) for all x in
(a, b), then f ′ ( x) < g′ ( x) for all x in (a, b). If true, prove it; if false, give a counterexample. Would your
answer change if the restriction of x to (a, b) were removed and all real x were used instead?
15. Show that the derivative of an odd function is an even function.
C
For Exercises 16-21, assuming that f ′ ( x) exists, prove the given formula.
f ( x + 2h) − f ( x − 2h) f ( x + 3h) − f ( x − 3h)
16. f ′ ( x) = lim 17. f ′ ( x) = lim
h→0 4h h→0 6h
f ( x + 2h) − f ( x − 3h) f ( x + ah) − f ( x − bh)

18. f ′ ( x) = lim 19. f ′ ( x) = lim (a, b > 0)
h→0 5h h→0 ( a + b) h
w f ( x ) − x f ( w) w2 f ( x ) − x 2 f ( w)
20. lim = f ( x) − x f ′ ( x) 21. lim = 2 x f ( x) − x2 f ′ ( x)
w→ x w − x w→ x w − x
22. Show that f ( x) = | x | is not differentiable at x = 0, using formula (1.4) for the derivative. Here you
will have to use a part of the definition which has not been used yet: as h approaches 0, h can be
either positive or negative. Consider those two cases in showing that the limit is not defined at x = 0.
23. Suppose that f (a + b) = f (a) f (b) for all a and b, and f ′ (0) exists. Show that f ′ ( x) exists for all x.
The Derivative: Infinitesimal Approach • Section 1.3 15
1.3 The Derivative: Infinitesimal Approach

Traditionally a function f of a variable x is written as y = f (x). The dependent variable y is
considered a function of the independent variable x. This allows taking the derivative of y
dy
with respect to x, i.e. the derivative of y as a function of x, denoted by dx . This is simply a
different way of writing f ′ (x), and is just one of many ways of denoting the derivative:
Notation for the derivative of y = f (x): The following are all equivalent:
dy d df
, f ′ (x) , ( f (x)) , y′ , ẏ , f˙(x) , , D f (x)
dx dx dx
dy
The notation dx appears to denote a fraction: a quantity dy divided by a quantity dx. It turns
out that the derivative really can be thought of in that way, as a ratio of infinitesimals. In fact,
this was the way in which derivatives were used by the founders of calculus—Newton and, in
particular, Leibniz.15 Even today, this is often the way in which derivatives are thought of and
used in fields outside of mathematics, such as physics, engineering, and chemistry, perhaps
due to its more intuitive nature.
The concept of infinitesimals used here is based on the nilsquare infinitesimal approach
developed by J.L. Bell16 , namely:
A number δ is an infinitesimal if the conditions (a)-(d) hold:

(a) δ 6= 0
(b) if δ > 0 then δ is smaller than any positive real number
(c) if δ < 0 then δ is larger than any negative real number
(d) δ2 = 0 (and hence all higher powers of δ, such as δ3 and δ4 , are also 0)
Note: Any infinitesimal multiplied by a nonzero real number is also an infinitesimal, while
0 times an infinitesimal is 0.
The above definition says that infinitesimals are numbers which are closer to 0 than any
positive or negative number without being zero themselves, and raising them to powers greater
than or equal to 2 makes them 0. So infinitesimals are not real numbers.17 This is not a
problem, since calculus deals with other numbers, such as infinity, which are not real. An
infinitesimal can be thought of as an infinitely small number arbitrarily close to 0 but not 0.
15 It was Leibniz who created the notation dy . For this reason dy is called the Leibniz notation for the derivative.
dx dx
Newton used the dot notation ẏ, which has fallen out of favor with mathematicians but is still used by many
physicists, especially when the independent variable represents time. Newton called derivatives fluxions. The
prime notation f ′ is due to the French mathematician and physicist Joseph Louis Lagrange (1736-1813).
16 B ELL , J.L., A Primer of Infinitesimal Analysis, Cambridge, U.K.: Cambridge University Press, 1998.
17 In an equivalent treatment, infinitesimals are part of the hyperreal number system. See K EISLER , H.J., Elemen-
tary Calculus: An Infinitesimal Approach, Boston: Prindle, Weber & Schmidt, 1976.
This might seem like a strange notion, but it really is not all that different from the limit
notion where, say, you let ∆ x approach 0 but not necessarily let it equal 0.18
As for the square of a nonzero infinitesimal being 0, think of how a calculator handles the
squares of small numbers. For example, most calculators can display 10−8 as 0.00000001, and
will even let you add 1 to that to get 1.00000001. But when you square 10−8 and add 1 to it,
most calculators will display the sum as simply 1. The calculator treats the square of 10−8 ,
namely 10−16 , as a number so small compared to 1 that it is effectively zero.19
Notice a major difference between 0 and an infinitesimal δ: 2 · 0 and 0 are the same, but 2 δ
and δ are distinct. This holds for any nonzero constant multiple, not just the number 2.
dy
The derivative dx of a function y = f (x) can now be defined in terms of infinitesimals:
Let dx be an infinitesimal such that f (x + dx) is defined. Then dy = f (x + dx) − f (x) is also
an infinitesimal, and the derivative of y = f (x) at x is the ratio of dy to dx:
dy f (x + dx) − f (x)
= (1.8)
dx dx
The basic idea is that dx is an infinitesimally small change in the variable x, producing an
infinitesimally small change dy in the value of y = f (x).
Example 1.4
Show that the derivative of y = f ( x) = x2 is
dy
dx = 2 x.
Solution: For any real number x,
dy f ( x + dx) − f ( x)
=
dx dx
( x + dx)2 − x2
=
dx
x2 + 2 x dx + (dx)2 − x2
=
dx
2 x dx + 0
= since dx is an infinitesimal ⇒ (dx)2 = 0
dx
✚
2 x✚
dx
=
✚
✚
dx
= 2x
18 The infinitesimal approach was first developed in an axiomatic manner in the landmark book R OBINSON, A.,
Non-Standard Analysis, Amsterdam: North-Holland, 1966. Robinson showed that for all practical purposes calcu-
lus can be developed without resorting to limits, with equivalent results.
19 Calculators do this for display reasons—most can show only 10-12 digits. Try this experiment on your calculator:
¡ ¢ ¡ ¡ ¢¢ ¡ ¡ ¢ ¢
Add 1030 , − 1030 , and 1 in two different ways: 1030 + − 1030 + 1, and 1030 + − 1030 + 1 . The first way will
give you the correct answer 1, but the second way yields 0. So addition is not always associative on calculators!
You might have noticed that the above example did not involve limits, and that the derivative
2x represents a real number (i.e. no infinitesimals appear in the final answer); this will always
be the case. Infinitesimals possess another useful property:
Microstraightness Property: For the graph of a differentiable function, any part of the
curve with infinitesimal length is a straight line segment.
In other words, at the infinitesimal level differentiable curves are straight. The idea behind
this is simple. At various points on a nonstraight differentiable curve y = f (x) the distances
along the curve between the points are not quite the same as the lengths of the line segments
joining the points. For example, in Figure 1.3.1 the distance s measured along the curve from
the point A to the point B is not the same as the length of the line segment AB joining A to B.
y
y = f ( x) B
B
magnify AB s
AB
s
A
A
x
Figure 1.3.1 Curved real-valued distance s 6= length of AB, but infinitesimal s = length of AB
However, as the points A and B get closer to each other, the difference between that part of
the curve joining A to B and the line segment AB becomes less noticeable. That is, the curve
is almost linear when A and B are close. The Microstraightness Property simply goes one step
further and says that the curve actually is linear when the distance s between the points is
infinitesimal (so that s equals the length of AB at the infinitesimal level).
At first this might seem nonsensical. After all, how could any nonstraight part of a curve
be straight? You have to remember that an infinitesimal is an abstraction—it does not exist
physically. A curve y = f (x) is also an abstraction, which exists in a purely mathematical sense,
so its geometric properties at the “normal” scale do not have to match those at the infinitesimal
scale (which can be defined in any way, provided the properties at that scale are consistent).
This abstraction finally reveals what an instantaneous rate of change y
is: the average rate of change over an infinitesimal interval. Moving y = f ( x)
an infinitesimal amount dx away from a value x produces an infinites- dy
imal change dy in a differentiable function y = f (x). The average rate dx
dy x
of change of y = f (x) over the infinitesimal interval [x, x+ dx] is thus dx , x x + dx
i.e. the slope—rise over run—of the straight line segment represented
by the curve y = f (x) over that interval, as in the figure on the right.20
20 Notice that the figure implies that the Pythagorean Theorem does not apply to infinitesimal triangles. This will
be discussed in Chapter 8.
The Microstraightness Property can be extended to smooth curves—that is, curves without
sharp edges or cusps. For example, circles and ellipses are smooth, but polygons are not.
The properties of infinitesimals can be applied to determine the derivatives of the sine and
cosine functions. Consider a circle of radius 1 with center O and points A and B on the circle
such that the line segment AB is a diameter. Let C be a point on the circle such that the angle
∠ BAC has an infinitesimal measure dx (in radians) as in Figure 1.3.2(a).
C C
= 2dx
s = BC = 2dx
s = BC
dx dx 2dx
A B A B
1 O 1 1 O 1
(a) (b)
Figure 1.3.2 Circle O: BC = 2 sin dx, ∠ BOC = 2 ∠ B AC
By Thales’ Theorem from elementary geometry, the angle ∠ ACB is a right angle. Thus:
BC BC
sin dx = = ⇒ BC = 2 sin dx
AB 2
Figure 1.3.2(b) shows that ∠ OAC + ∠ OC A + ∠ AOC = π. Thus, 1 = OC = OA ⇒ ∠ OC A =
∠ OAC = dx ⇒ ∠ AOC = π − dx − dx = π − 2dx ⇒ ∠ BOC = 2dx. By the arc length formula from
along the circle from B to C is the radius times the
trigonometry, the length s of the arc BC
= BC, and thus:
central angle ∠ BOC: s = BC = 1 · 2dx = 2dx. But by Microstraightness, BC
= 2dx
2 sin dx = BC = BC ⇒ sin dx = dx
Since dx is an infinitesimal, (dx)2 = 0. So since sin2 dx + cos2 dx = 1, then:
cos2 dx = 1 − sin2 dx = 1 − (dx)2 = 1 − 0 = 1 ⇒ cos dx = 1
The derivative of y = sin x is then:

d dy sin(x + dx) − sin x
(sin x) = =
dx dx dx
(sin x cos dx + sin dx cos x) − sin x
= by the sine addition formula
dx
✘
✘✘
(sin
✘ x) ✘
(1) + dx cos x − ✘ ✘
sin✘x ✚ cos x
✚
dx
= = , and thus:
dx ✚✚
dx
d
(sin x) = cos x
dx
A similar argument (left as an exercise) using the cosine addition formula shows:
d
(cos x) = − sin x
dx
One of the intermediate results proved here bears closer examination. Namely, sin dx = dx
for an infinitesimal angle dx measured in radians. At first, it might seem that this cannot be
true. After all, an infinitesimal dx is thought of as being infinitely close to 0, and sin 0 = 0, so
you might expect that sin dx = 0. But this is not the case. The formula sin dx = dx says that in
an infinitesimal interval around 0, the function y = sin x is identical to the line y = x (not the
line y = 0). This, in turn, suggests that for real-valued x close to 0, sin x ≈ x.
This indeed turns out to be the case. The free graphing software Gnuplot21 can display the
graphs of y = sin x and y = x. Figure 1.3.3(a) below shows how those graphs compare over the
interval [−π, π]. Outside the interval [−1, 1] there is a noticeable difference.
4 0.3
sin( x) sin( x)
3
x 0.2 x
2
0.1
1
0 0
y
-1
-0.1
-2
-0.2
-3
-4 -0.3
-3 -2 -1 0 1 2 3 -0.3 -0.2 -0.1 0 0.1 0.2 0.3
x x
(a) (b)
Figure 1.3.3 sin dx = dx: Comparing y = sin x and y = x near x = 0
Figure 1.3.3(b) shows that there is virtually no difference in the graphs even over the non-
infinitesimal interval [−0.3, 0.3]. So sin x ≈ x is a good approximation when x is close to 0, that
is, when | x | ≪ 1 (the symbol ≪ means “much less than”). This approximation is used in many
applications in engineering and physics when the angle x is assumed to be small.
Notice something else suggested by the relation sin dx = dx: there is a fundamental differ-
ence at the infinitesimal level between a line of slope 1 (y = x) and a line of slope 0 (y = 0). In a
real interval (−a, a) around x = 0 the difference between the two lines can be made as small as
desired by choosing a > 0 small enough. But in an infinitesimal interval (−δ, δ) around x = 0
there is unbridgeable gulf between the two lines. This is the crucial difference in sin dx being
equal to dx rather than 0.
21 Available at http://www.gnuplot.info.
Notice also that the value of a function at an infinitesimal may itself be an infinitesimal (e.g.
sin dx = dx) or a real number (e.g. cos dx = 1).
df
For a differentiable function f (x), dx = f ′ (x) and so multiplying both sides by dx yields the
important relation:
df = f ′ (x) dx (1.9)
Note that both sides of the above equation are infinitesimals for each value of x in the domain
of f ′ , since f ′ (x) would then be a real number.
The notion of an infinitesimal was fairly radical at the time (and still is). Some mathemati-
cians embraced it, e.g. the outstanding Swiss mathematician Leonhard Euler (1707-1783),
who produced a large amount of work using infinitesimals. But it was too radical for many
mathematicians (and philosophers22), enough so that by the 19th century some mathemati-
cians (notably Augustin Cauchy and Karl Weierstrass) felt the need to put calculus on what
they considered a more “rigorous” footing, based on limits.23 Yet it was precisely the notion of
an infinitesimal which lent calculus its modern character, by showing the power and usefulness
of such an abstraction (especially one that did not obey the rules of classical mathematics).
Exercises
A
For Exercises 1-9, let dx be an infinitesimal and prove the given formula.
1. (dx + 1)2 = 2dx + 1 2. (dx + 1)3 = 3dx + 1 3. (dx + 1)−1 = 1 − dx
4. tan dx = dx 5. sin 2dx = 2dx 6. cos 2dx = 1
7. sin 3dx = 3dx 8. cos 3dx = 1 9. sin 4dx = 4dx
10. Is cot dx defined for an infinitesimal dx? If so, then find its value. If not, then explain why.
11. In the proof of the derivative formulas for sin x and cos x, the equation cos2 dx = 1 was solved to
give cos dx = 1. Why was the other possible solution cos dx = −1 ignored?
B
12. Show that d
dx (cos x) = − sin x.
d
13. Show that dx (cos 2 x) = −2 sin 2 x. (Hint: Use Exercises 5 and 6.)
C
14. Show that d
dx (tan x) = sec2 x. (Hint: Use Exercise 4.)
22 The English philosopher George Berkeley (1685-1753) famously derided infinitesimals as “the ghosts of departed
quantities” in his book The Analyst (1734), which had the disquieting subtitle “A Discourse Addressed to an Infidel
Mathematician” (directed at Newton).
23 However, the limit approach turns out, ultimately, to be equivalent to the infinitesimal approach. In essence,
only the terminology is different.

Derivatives of Sums, Products and Quotients • Section 1.4 21
1.4 Derivatives of Sums, Products and Quotients

So far the derivatives of only a few simple functions have been calculated. The following rules
will make it easier to calculate derivatives of more functions:
Rules for Derivatives: Suppose that f and g are differentiable functions of x. Then:
d df dg
Sum Rule: ( f + g) = +
dx dx dx
d df dg
Difference Rule: ( f − g) = −
dx dx dx
d df
Constant Multiple Rule: (c f ) = c · for any constant c
dx dx
d dg df
Product Rule: ( f · g) = f · + g·
dx dx dx
µ ¶ df dg
d f g · dx − f · dx
Quotient Rule: =
dx g g2
The above rules can be written using the prime notation for derivatives:
Sum Rule: ( f + g)′ (x) = f ′ (x) + g′ (x)
Difference Rule: ( f − g)′ (x) = f ′ (x) − g′ (x)
Constant Multiple Rule: (c f )′ (x) = c · f ′ (x) for any constant c
Product Rule: ( f · g) (x) = f (x) · g (x) + g(x) · f ′ (x)

′ ′
µ ¶′
f g(x) · f ′ (x) − f (x) · g′ (x)
Quotient Rule: (x) =
g (g(x))2
df dg
The proof of the Sum Rule is straightforward. Since dx and dx both exist then:
d ( f + g)(x + dx) − ( f + g)(x) f (x + dx) + g(x + dx) − ( f (x) + g(x))

( f + g) = =
dx dx dx
f (x + dx) − f (x) + g(x + dx) − g(x) f (x + dx) − f (x) g(x + dx) − g(x)
= = +
dx dx dx
df dg
= + X
dx dx
The proofs of the Difference and Constant Multiple Rules are similar and are left as exercises.
Note that by the Product Rule, in general the derivative of a product is not the product of
d ( f · g) df dg
the derivatives. That is, dx 6= dx · dx . This should be obvious from some earlier examples. For
d ( f · g) df dg
instance, if f (x) = x and g(x) = 1 then ( f · g)(x) = x · 1 = x so that dx = 1, but dx · dx = 1 · 0 = 0.
There is a proof of the Product Rule similar to the proof of the Sum Rule (see Exercise 20),
but there is a more geometric way of seeing why the formula holds, described below.
Construct a rectangle whose perpendicular sides g( x + dx)
have lengths f (x) and g(x) for some x, as in the draw- df
ing on the right. Change x by some infinitesimal
amount dx, which produces infinitesimal changes df
f ( x + dx)
and dg in f (x) and g(x), respectively. Assume those f ( x)
changes are positive and extend the original rectan-
gle by those amounts, creating a larger rectangle with
perpendicular sides f (x + dx) and g(x + dx). Then
g( x ) dg
d( f · g) = ( f · g)(x + dx) − ( f · g)(x)
= f (x + dx) · g(x + dx) − f (x) · g(x)
= (area of outer rectangle) − (area of original rectangle)
= sum of the areas of the three shaded inner rectangles
= f (x) · dg + g(x) · df + df · dg
= f (x) · dg + g(x) · df + ( f ′ (x) dx) · (g′ (x) dx)
= f (x) · dg + g(x) · df + ( f ′ (x)g′ (x)) · (dx)2
= f (x) · dg + g(x) · df + ( f ′ (x)g′ (x)) · 0
d( f · g) = f (x) · dg + g(x) · df , so dividing both sides by dx yields
d( f · g) dg df
= f (x) · + g(x) · X
dx dx dx
f
To prove the Quotient Rule, let y = g , so f = g · y. If y were a differentiable function of x,
then the Product Rule would give
df d(g · y) dy dg dy f dg dy df f dg
= = g· + y· = g· + · ⇒ g· = − ·
dx dx dx dx dx g dx dx dx g dx
and so dividing both sides by g and getting a common denominator gives

df dg
dy 1 df f dg g · dx − f · dx
= · − 2· = X
dx g dx g dx g2
HI f
A simple mnemonic device for remembering the Quotient Rule is: write g as HO —so that HI
represents the “high” (numerator) part of the quotient and HO represents the “low” (denomi-
nator)
³ ´ part—and think of dHI and dHO as the derivatives of HI and HO, respectively. Then
d f HO· d HI − HI· d HO
dx g = 2 , pronounced as “ho-dee-hi minus hi-dee-ho over ho-ho.”
HO
Example 1.5
Use the Quotient Rule to show that d
dx (tan x) = sec2 x.
sin x
Solution: Since tan x = cos x then:
µ ¶ d d
d d sin x (cos x) · dx (sin x) − (sin x) · dx (cos x) (cos x) · (cos x) − (sin x) · (− sin x)
(tan x) = = 2
=
dx dx cos x cos x cos2 x
cos2 x + sin2 x 1
= 2
= = sec2 x
cos x cos2 x
Example 1.6
Use the Quotient Rule to show that d
dx (sec x) = sec x tan x.
1
Solution: Since sec x = cos x then:
µ ¶ d d
d d 1 (cos x) · dx (1) − 1 · dx (cos x) (cos x) · 0 − 1 · (− sin x)
(sec x) = = =
dx dx cos x cos2 x cos2 x
sin x 1 sin x
= 2
= · = sec x tan x
cos x cos x cos x
Similar to the above examples, the derivatives of cot x and csc x can be found using the
Quotient Rule (left as exercises). The derivatives of all six trigonometric functions are:
d d
(cos x) = − sin x (sec x) = sec x tan x
dx dx
d d
(sin x) = cos x (csc x) = − csc x cot x
dx dx
d d
(tan x) = sec2 x (cot x) = − csc2 x
dx dx
Note that the Sum and Difference Rules can be applied to sums and differences, respectively,
of not just two functions but any finite (integer) number of functions. For example, for three
differentiable functions f 1 , f 2 , and f 3 ,
d df 1 d
( f1 + f2 + f3) = + ( f 2 + f 3) by the Sum Rule
dx dx dx
df 1 df 2 df 3
= + + by the Sum Rule again.
dx dx dx
Continuing like this for four functions, then five functions, and so forth, the Sum and Differ-
ence Rules combined with the Constant Multiple Rule yield the following formula:
For n ≥ 1 differentiable functions f 1 , . . ., f n and constants c 1 , . . ., c n :
d df 1 df n
(c 1 f 1 + · · · + c n f n ) = c 1 + · · · + cn (1.10)
dx dx dx
Note that the above formula includes differences, by using negative constants. The formula
also shows that differentiation is a linear operation, which makes dx d
a linear operator. The
d
idea is that dx “operates” on differentiable functions by taking their derivatives with respect to
the variable x. The sum c 1 f 1 + · · · + c n f n is called a linear combination of functions, and the
derivative of that linear combination can be taken term by term, with the constant multiples
taken outside the derivatives.
A special case of the above formula is replacing the functions f 1 , . . ., f n by nonnegative
powers of x, making the sum a polynomial in x. In previous sections the derivatives of a few
polynomials—such as x and x2 —were calculated. For the derivative of a general polynomial,
the following rule is needed:
d ¡ n¢
Power Rule: x = n x n −1 for any integer n
dx
There are several ways to prove this formula; one such way being a proof by induction,
which in general means using the following principle:24
Principle of Mathematical Induction

A statement P(n) about integers n ≥ k is true for all n ≥ k if:
1. P(k) is true.
2. If P(n) is true for some integer n ≥ k then P(n + 1) is true.
The idea behind mathematical induction is simple: if a statement is true about some initial
integer k (Step 1 above) and if the statement being true for some integer implies it is true for
the next integer (Step 2 above), then the statement being true for k implies it is true for k + 1,
which in turn implies it is true for k + 2, which implies it is true for k + 3, and so forth, making
it true for all integers n ≥ k.
Typically the initial integer k will be 0 or 1. To prove the Power Rule for all integers, first
use induction to prove the rule for all nonnegative integers n ≥ 0, using k = 0 for the initial
integer. For the proof by induction, let P(n) be the statement: dx d
(x n ) = n x n−1 .
1. Show that P(0) is true. ¡ ¢
That means showing that the Power Rule holds for n = 0, i.e. d
dx x0 = 0 x0−1 = 0. But
x0 = 1 is a constant, so its derivative is 0. X
24 As Bertrand Russell noted, the name is really a misnomer: it is actually a definition of the natural numbers
rather than a principle, and induction technically has a different meaning.

2. Assuming P(n) is true for some n ≥ 0, show that P(n + 1) is true.

¡ n +1 ¢
d
Assuming that dx (x n ) = n x n−1 , show that dx
d
x = (n + 1) x(n+1)−1 = (n + 1) x n . It was
d
shown in Section 1.2 that dx (x) = 1, so:
d ¡ n +1 ¢ d ¡ ¢
x = x · xn
dx dx
d ¡ n¢ d
= x· x + xn · (x) (by the Product Rule)
dx dx
= x · n x n −1 + x n · 1 (by the assumption that P(n) is true)
= n x n + x n = (n + 1) x n X
Thus, by induction, the Power Rule is true for all nonnegative integers n ≥ 0.
To show that the Power Rule is true for all negative integers n < 0, write n = − m, where m
is positive (namely, m = | n |). Then:
µ ¶
d ¡ n¢ d ¡ −m ¢ d 1
d
x m · dx (1) − 1 · dx
d
(x m )
x = x = = (by the Quotient Rule)
dx dx dx x m (x m )2
x m · 0 − 1 · m x m −1
= (by the Power Rule for positive integers)
x2 m
= − m x m −1−2 m − m x − m −1 = n x n −1 X
Thus, the Power Rule is true for all integers, which completes the proof. QED
Example 1.7
Find the derivative of f ( x) = x4 − 4 x3 + 6 x2 − 4 x + 1.
Solution: Differentiate the polynomial term by term and use the Power Rule:
df d ¡ 4 ¢
= x − 4 x3 + 6 x2 − 4 x + 1
dx dx
d ¡ 4¢ d ¡ 3¢ d ¡ 2¢ d d
= x − 4· x + 6· x − 4· ( x) + (1)
dx dx dx dx dx
= 4 x 4 −1 − 4 · 3 x 3 −1 + 6 · 2 x 2 −1 − 4 · 1 + 0
= 4 x3 − 12 x2 + 12 x − 4
In general, the derivative of a polynomial of degree n ≥ 0 is given by:
For any constants a 0 , . . ., a n with n ≥ 0:
d ¡ ¢
a n x n + a n−1 x n−1 · · · + a 2 x2 + a 1 x + a 0 = na n x n−1 + (n−1) a n−1 x n−2 + · · · + 2a 2 x + a 1
dx
A way to remember the Power Rule is: bring the exponent down in front of the variable then
reduce the variable’s original exponent by 1. This works even for negative exponents.
Example 1.8
2
Find the derivative of f ( t) = 3 t100 − t100
.
Solution: Differentiate term by term:
µ ¶
df d 100 2 d ¡ 100 ¢ ¡ ¢ 200
= 3t − 100 = 3t − 2 t−100 = 3 · 100 t99 − 2 · −100 t−101 = 300 t99 + 101
dt dt t dt t
Exercises
A
For Exercises 1-14, use the rules from this section to find the derivative of the given function.
1. f ( x) = x2 − x − 1 2. f ( x) = x8 + 2 x4 + 1
2 x6 3 sin x + cos x
3. f ( x) = − 4. f ( x) =
3 2 x6 4
5. f ( x) = x sin x 6. f ( x) = x2 tan x
sin x sin x
7. f ( x) = 8. f ( x) =
x x2
2t 1 − t2
9. f ( t) = 10. g( t) =
1 + t2 1 + t2
ax + b Gm 1 m 2
11. f ( x) = (a, b, c, d are constants) 12. F ( r) = − (G , m 1 , m 2 are constants)
cx + d r2
4
13. A (r) = π r2 14. V ( r) = π r3
3
15. Show that d
dx (cot x) = − csc2 x. 16. Show that dx d
(csc x) = − csc x cot x.
B
17. Prove the Difference Rule. 18. Prove the Constant Multiple Rule.
19. Use the Product Rule to show that for three differentiable functions f , g, and h, the derivative of
their product is ( f gh)′ = f ′ gh + f g′ h + f gh′ .
C
20. Provide an alternative proof of the Product Rule for two differentiable functions f and g of x:
(a) Show that (df )(dg) = 0.
(b) By definition, the derivative of the product f · g is
d f ( x + dx) · g( x + dx) − f ( x) · g( x)
( f · g) = .
dx dx
dg df
Use that formula along with part (a) to show that d
dx ( f · g) = f · dx + g· dx .
(Hint: Recall that df = f ( x + dx) − f ( x).)
The Chain Rule • Section 1.5 27
1.5 The Chain Rule

From what has been discussed so far it might be tempting to think that the derivative of a
function like sin 2x is simply cos 2x, since the derivative of sin x is cos x. It turns out that is
not correct:
d d
(sin 2x) = (2 sin x cos x) (by the double-angle formula for sine)
dx dx
d
= 2 (sin x cos x) (by the Constant Multiple Rule)
dx
µ ¶
d d
= 2 sin x · (cos x) + cos x · (sin x) (by the Product Rule)
dx dx
= 2 (sin x · (− sin x) + cos x · cos x)
¡ ¢
= 2 cos2 x − sin2 x
= 2 cos 2x (by the double-angle formula for cosine)
So the derivative of sin 2x is 2 cos 2x, not cos 2x.

In other words, you cannot simply replace x by 2x in the derivative formula for sin x. Instead,
regard sin 2x as a composition of two functions: the sine function and the 2x function. That is,
let f (u) = sin u, where the variable u itself represents a function of another variable x, namely
u(x) = 2x. So since f is a function of u, and u is a function of x, then f is a function of x, namely:
f (x) = sin 2x. Since f is a differentiable function of u, and u is a differentiable function of x,
df df
then du and du du
dx both exist (with du = cos u and dx = 2), and multiplying the derivatives shows
that f is a differentiable function of x:
df ✚ ✚
du df
· = since the infinitesimals du cancel, so
✚✚
du dx dx
df df
(cos u) · 2 = ⇒ = 2 cos u = 2 cos 2x
dx dx
The above argument holds in general, and is known as the Chain Rule:
Chain Rule: If f is a differentiable function of u, and u is a differentiable function of x,

then f is a differentiable function of x, and its derivative with respect to x is:
df df du
= ·
dx du dx
Notice how simple the proof is—the infinitesimals du cancel.25

25 Some textbooks give dire warnings to not think that du is an actual quantity that can be canceled. However, you
can safely ignore those warnings, because du is just an infinitesimal and hence can be canceled!
df
The Chain Rule should make sense intuitively. For example, if du = 4 then that means f is
increasing 4 times as fast as u, and if du
dx = 3 then u is increasing 3 times as fast as x, so overall
f should be increasing 12 = 4 · 3 times as fast as x, exactly as the Chain Rule says.
Example 1.9
Find the derivative of f ( x) = sin ( x2 + x + 1).
Solution: The idea is to make a substitution u = x2 + x + 1 so that f ( x) = sin u. By the Chain Rule,
df df du
= ·
dx du dx
d d 2
= (sin u) · ( x + x + 1)
du dx
= (cos u) · (2 x + 1)
= (2 x + 1) cos ( x2 + x + 1)
after replacing u by its definition as a function of x in the last step; the final answer for the derivative
should be in terms of x, not u.
In the Chain Rule you can think of the function in question as the composition of an “outer”
function f and an “inner” function u; first take the derivative of the “outer” function then
multiply by the derivative of the “inner” function. Think of the “inner” function as a box into
which you can put any function of x, and the “outer” function being a function of that empty
box.
For instance, for the function f (x) = sin(x2 + x + 1) in the previous example, think of the
“outer” function as sin ✷, where ✷ = x2 + x + 1 is the “inner” function, so that
f (x) = sin(x2 + x + 1)
= sin ✷
df d
= (cos ✷) · ✷
dx dx
³ ´ d
= cos x2 + x + 1 · x2 + x + 1
dx
= (2x + 1) cos (x2 + x + 1)
Example 1.10
Find the derivative of f ( x) = (2 x4 − 3 cos x)10 .
Solution: Here the “outer” function is f (✷) = ✷10 and the “inner” function is ✷ = u = 2 x4 − 3 cos x:
df df du d
= · = 10 ✷9 · (✷) = 10 (2 x4 − 3 cos x)9 (8 x3 + 3 sin x)
dx du dx dx
Recall that the composition f ◦ g of two functions f and g is defined as ( f ◦ g)(x) = f (g(x)).
Using prime notation the Chain Rule can be written as:
Chain Rule: If g is a differentiable function of x, and f is a differentiable function on the

range of g, then f ◦ g is a differentiable function of x, and its derivative with respect to x is:
( f ◦ g)′ (x) = f ′ (g(x)) · g′ (x)
Using the Chain Rule, the Power Rule can be extended to include exponents that are rational
numbers:26
d ¡ r¢
x = r x r −1 for any rational number r
dx
To prove this, let r = m/n, where m and n are integers with n 6= 0. Then y = x r = x m/n = (x m )1/n ,
so that yn = x m . Taking the derivative with respect to x of both sides of this equation gives
d ¡ n¢ d ¡ m¢
y = x , so evaluating the left side by the Chain Rule gives
dx dx
dy
n y n −1 · = mx m−1
dx
³ ´ n−1 dy
n x m/ n · = mx m−1
dx
dy mx m−1 m m−1−(m−(m/n)) m (m/n)−1
= m − ( m / n )
= x = x = r x r −1 X
dx nx n n
Example 1.11
p
Find the derivative of f ( x) = x.
p
Solution: Since x = x1/2 then by the Power Rule:
df d ³ 1/2 ´ 1 1 1
= x = x1/2−1 = x−1/2 = p
dx dx 2 2 2 x
Example 1.12
2
Find the derivative of f ( x) = p
3 x
.
Solution: µ ¶
df d 2 −1/2 2 −1 −3/2 1
= x = · x = − 3/2
dx dx 3 3 2 3x
26 It will be shown in Chapter 2 how to define any real number as an exponent. The Power Rule extends to that
case as well.
Exercises
A
For Exercises 1-18, find the derivative of the given function.
1. f ( x) = (1 − 5 x)4 2. f ( x) = 5 ( x3 + x − 1)4
p 3
3. f ( x) = 1 − 2x 4. f ( x) = (1 − x2 ) 2
p p
x x + 1
5. f ( x) = 6. f ( x) = p
x + 1 x − 1
µ ¶ µ 2 ¶6
1 − t 4 x + 1
7. f ( t) = 8. f ( x) =
1 + t x − 1
¡p ¢
9. f ( x) = sin2 x 10. f ( x) = cos x
11. f ( x) = 3 tan (5 x) 12. f ( x) = A cos (ω x + φ) ( A , ω, φ are constants)

µ ¶ µ ¶
1 1
13. f ( x) = sec( x2 ) 14. f ( x) = sin2 + cos2
1−x 1−x
µ µ ¶ ¶−1
1 1 x−l 2
15. L(β) = p 16. f ( x) = 1 + (s, l are constants)
1 − β2 πs s
p p
17. f ( x) = cos (cos x) 18. f ( x) = 1 + x
B
19. In a certain type of electronic circuit27 the overall gain A v is given by
Ao
Av =
1 − T
where the loop gain T is a function of the open-loop gain A o .
(a) Show that
dA v 1 Ao d(1 − T )
= − .
dA o 1 − T (1 − T )2 dA o
(b) In the case where T is directly proportional to A o , use part (a) to show that
dA v 1
= .
dA o (1 − T )2
(Hint: First show that A o · d(1 −T )

dA o = −T .)
20. Show that the Chain Rule can be extended to 3 functions: if u is a differentiable function of x, v is
a differentiable function of u, and f is a differentiable function of v, then
df df dv du
= · ·
dx dv du dx
so that f is a differentiable function of x. Notice that the 3 derivatives are linked together as in a
chain (hence the name of the rule). The Chain Rule can be extended to any finite number of functions
by the above technique.
27 This is an example of a current-differencing negative feedback amplifier. See pp.473-479 in S CHILLING, D.L.
AND C. B ELOVE, Electronic Circuits: Discrete and Integrated, 2nd ed., New York: McGraw-Hill, Inc., 1979.
21. In an internal combustion engine, as a piston moves downward the connecting rod rotates the crank
in the clockwise direction, as shown in Figure 1.5.1 below.28
piston
conn
l
e
cting
s rod
B
θ
k
an
cr
a
O
Figure 1.5.1
The point A can only move vertically, causing the point B to move around a circle of radius a centered
at the point O, which is directly below the point A and does not move. As the crank rotates it makes
an angle θ with the line O A . Let l = AB and s = O A as in the picture. Assume that all lengths are
measured in centimeters, and let the time variable t be measured in minutes.
¡ ¢1/2
(a) Show that s = a cos θ + l 2 − a2 sin2 θ for 0 ≤ θ ≤ π.
(b) The mean piston speed is S̄ p = 2LN , where L = 2a is the piston stroke, and N is the rotational
velocity of the crank, measured in revolutions per minute (rpm). The instantaneous piston
dt . Let R = l /a. Show that for 0 ≤ θ ≤ π,
velocity is S p = ds
¯ ¯ " #
¯ Sp ¯
¯ ¯ = π sin θ 1 + cos θ
¯ ¯ 2 ¡ ¢1/2 .
S̄ p R 2 − sin2 θ
28 See pp.43-45 in H EYWOOD, J.B., Internal Combustion Engine Fundamentals, New York: McGraw-Hill Inc., 1988.
1.6 Higher Order Derivatives

The derivative f ′ (x) of a differentiable function f (x) can be thought of as a function in its
own right, and if it is differentiable then its derivative—denoted by f ′′ (x)—is the second
derivative of f (x) (the first derivative being f ′ (x)). Likewise, the derivative of f ′′ (x) would
be the third derivative of f (x), written as f ′′′(x). Continuing like this yields the fourth
derivative, fifth derivative, and so on. In general the n-th derivative of f (x) is obtained
by differentiating f (x) a total of n times. Derivatives beyond the first are called higher order
derivatives.
Example 1.13
For f ( x) = 3 x4 find f ′′ ( x) and f ′′′ ( x).
Solution: Since f ′ ( x) = 12 x3 then the second derivative f ′′ ( x) is the derivative of 12 x3 , namely:
f ′′ ( x) = 36 x2
The third derivative f ′′′ ( x) is then the derivative of 36 x2 , namely:
f ′′′ ( x) = 72 x
Since the prime notation for higher order derivatives can be cumbersome (e.g. writing 50
prime marks for the fiftieth derivative), other notations have been created:
Notation for the second derivative of y = f (x): The following are all equivalent:
d2 y d2 d2 f
f ′′ (x) , f (2) (x) , , ( f (x)) , y′′ , y(2) , ÿ , f¨(x) , , D 2 f (x)
dx2 dx2 dx2
Notation for the n-th derivative of y = f (x): The following are all equivalent:
dn y dn dn f
f (n) (x) , , ( f (x)) , y(n) , , D n f (x)
dx n dx n dx n
Notice that the parentheses around n in the notation f (n) (x) indicate that n is not an exponent—
dn y
it is the number of derivatives to take. The n in the Leibniz notation dxn indicates the same
thing, and in general makes working with higher order derivatives easier:
µ ¶
d2 y d dy
=
dx2 dx dx
µ ¶ µ ¶
d3 y d d2 y d 2 dy
= =
dx3 dx dx2 dx2 dx
..
.
µ ¶ µ ¶
n
d y d d n −1 y d n−1 dy
= =
dx n dx dx n−1 dx n−1 dx
Higher Order Derivatives • Section 1.6 33
A natural question to ask is: what do higher order derivatives represent? Recall that the
first derivative f ′ (x) represents the instantaneous rate of change of a function f (x) at the value
x. So the second derivative f ′′ (x) represents the instantaneous rate of change of the function
f ′ (x) at the value x. In other words, the second derivative is a rate of change of a rate of change.
The most famous example of this is for motion in a straight line: s<0 s>0
let s(t) be the position of an object at time t as the object moves along s=0
the line. The motion can take two directions, e.g. forward/backward
or up/down. Take one direction to represent positive position and the other to represent neg-
ative direction, as in the drawing on the right. The (instantaneous) velocity v(t) of the object
at time t is s′ (t), i.e. the first derivative of s(t). The acceleration a(t) of the object at time
t is defined as v′ (t), the instantaneous rate of change of the velocity. Thus, a(t) = s′′ (t), i.e.
acceleration is the second derivative of position. To summarize:
s(t) = position at time t
v(t) = velocity at time t

ds
= = s′ (t) = ṡ(t)
dt
a(t) = acceleration at time t
dv
= = v′ (t) = v̇(t)
dt
µ ¶
d ds d2s
= = = s′′ (t) = s̈(t)
dt dt dt2
Example 1.14
Ignoring wind and air resistance, the position s of a ball thrown straight up with an initial velocity of
34 m/s from a starting point 2 m off the ground is given by s( t) = −4.9 t2 + 34 t + 2 at time t (measured in
seconds) with s measured in meters. Find the velocity and acceleration of the ball at any time t ≥ 0.
Solution: The ball moves in a straight vertical line, first straight up then straight down until it hits the
ground. Its velocity v( t) is
ds
v( t) = = −9.8 t + 34 m/s
dt
while its acceleration a( t) is
µ ¶
d2 s d ds d
a( t) = = = (−9.8 t + 34) = −9.8 m/s2 ,
dt2 dt dt dt
which is the acceleration due to the force of gravity on Earth.
Note that time t = 0 is the time at which the ball was thrown, so that v(0) is the initial velocity of the
ball. Indeed, v(0) = −9.8(0) + 34 = 34 m/s, as expected.
Notice in Example 1.14 that the acceleration of the ball is not t = 3.47
v=0
only constant but also negative. To see why this makes sense,
first consider the case where the ball is moving upward. The
ball has an initial upward velocity of 34 m/s then slows down v>0
v<0
to 0 m/s at the instant it reaches its maximum height above the a <0
a<0
ground. So the velocity is decreasing, i.e. its rate of change—the s=2
acceleration—is negative. t=0
The ball reaches its maximum height above the ground when its
s=0
velocity is zero, that is, when v(t) = −9.8t + 34 = 0, i.e. at time t =
34/9.8 = 3.47 seconds after being thrown (see the picture above). The ball then starts moving
downward and its velocity is negative (e.g. at time t = 4 s the velocity is v(4) = −9.8(4) + 34 =
−5.2 m/s). Recall that negative velocity indicates downward motion, while positive velocity
means the motion is upward (away from the Earth’s center). So in the case where the ball
begins moving downward it goes from 0 m/s to a negative velocity, with the ball moving faster
towards the ground, which it hits with a velocity of −33.43 m/s (why?). So again the velocity is
decreasing, which again means that the acceleration is negative.
Common terminology involving motion might cause some confusion with the above discus-
sion. For example, even though the ball’s acceleration is negative as it falls to the ground, it
is common to say that the ball is accelerating in that situation, not decelerating (as the ball is
doing while moving upward). In general, acceleration is understood to mean that the mag-
nitude (i.e. the absolute value) of the velocity is increasing. That magnitude is called the
speed of the object. Deceleration means the speed is decreasing.
The first and second derivatives of an object’s position with respect to time represent the
object’s velocity and acceleration, respectively. Do the third, fourth, and other higher order
derivatives have any physical meanings? It turns out they do. The third derivative of posi-
tion is called the jerk of the object. It represents the rate of change of acceleration, and is
often used in fields such as vehicle dynamics (e.g. minimizing jerk to provide smoother brak-
ing). The fourth, fifth, and sixth derivatives of position are called snap, crackle, and pop,
respectively.29
The zero-th derivative f (0) (x) of a function f (x) is defined to be the function f (x) itself:
f (0) (x) = f (x). There is a way to define fractional derivatives, e.g. the one-half derivative
f (1/2) (x), which will be discussed in Chapter 6.
An immediate consequence of the definition of higher order derivatives is:
µ ¶
d m+ n dm dn
( f (x)) = ( f (x)) for all integers m ≥ 0 and n ≥ 0.
dx m+n dx m dx n
Recall that the factorial n! of an integer n > 0 is the product of the integers from 1 to n:
n! = 1 · 2 · 3 · · · · · n
29 Yes, those really are their names, obviously inspired by a certain breakfast cereal. Snap has found some uses in
flight dynamics, e.g. minimizing snap to optimize flight paths of drones.

Higher Order Derivatives • Section 1.6 35
For example:
1! = 1 3! = 1 · 2 · 3 = 6
2! = 1 · 2 = 2 4! = 1 · 2 · 3 · 4 = 24
By convention 0! is defined to be 1. The following statement can be proved using induction:
dn
(x n ) = n! for all integers n ≥ 0
dx n
Thus, µ ¶
d n +1 ¡ n ¢ d d n ¡ n¢ d
n + 1
x = n
x = (n!) = 0
dx dx dx dx
for all integers n ≥ 0, since n! is a constant. Polynomials are linear combinations of nonnegative
powers of a variable (e.g. x), so the above result combined with the Sum Rule and the Constant
Multiple rule—which also hold for higher-order derivatives—yields this important fact:
The (n + 1)-st derivative (“n plus first derivative”) of a polynomial of degree n is 0:

d n +1
For any polynomial p(x) = a 0 + a 1 x + a 2 x2 + · · · + a n x n of degree n, (p(x)) = 0.
dx n+1
This is the basis for the commonly-used statement that “any polynomial can be differentiated
to 0” by taking a sufficient number of derivatives. For example, differentiating the polynomial
p(x) = 100x100 + 50x99 101 times would yield 0 (as would differentiating more than 101 times).
Exercises
A
For Exercises 1-6 find the second derivative of the given function.
1. f ( x) = x3 + x2 + x + 1 2. f ( x) = x2 sin x 3. f ( x) = cos3 x
sin x 1 Gm 1 m 2
4. f ( x) = 5. f ( x) = 6. F (r) =
x x r2
7. Find the first five derivatives of f ( x) = sin x. Use those to find f (100) ( x) and f (2014)( x).
8. Find the first five derivatives of f ( x) = cos x. Use those to find f (100) ( x) and f (2014)( x).
9. If an object moves along a straight line such that its position s( t) at time t is directly proportional to
t for all t (written as s ∝ t), then show that the object’s acceleration is always 0.
B
dn
10. Use induction to show that dxn ( x n ) = n! for all integers n ≥ 1.
dm n!
11. Show that for all integers n ≥ m ≥ 1, dxm (xn) = ( n− m)! x n− m .
1
12. Find the general expression for the n-th derivative of f ( x) = ax+ b for all constants a and b (a 6= 0).
13. Show that the function y = A cos (ω t + φ) + B sin (ω t + φ) satisfies the differential equation
d2 y
+ ω2 y = 0
dt2
for all constants A , B, ω, and φ.
14. If s( t) represents the position at time t of an object moving along a straight line, then show that:
s′ and s′′ have the same sign ⇒ the object is accelerating
′ ′′
s and s have opposite signs ⇒ the object is decelerating
15. For all twice-differentiable functions f and g, show that ( f · g)′′ = f ′′ · g + 2 f ′ · g′ + f · g′′ .
C
d
16. Recall that taking a derivative is a way of operating on a function. That is, think of dx as the
differentiation operator on the collection of differentiable functions, taking a function f ( x) to its
df
derivative function dx :
d
dx
df
f ( x) dx
d2
Likewise, dx2
is an operator on twice-differentiable functions, taking a function f ( x) to its second
d 2f
derivative function dx2
:
d2
dx2 d 2f
f ( x)
dx2
In general, an eigenfunction of an operator A is a function φ( x) such that A (φ( x)) = λ · φ( x), that is,
A
φ( x ) λ · φ( x )
for all x in the domain of φ, for some constant λ called the eigenvalue of the eigenfunction.
d2
(a) Show for all constants k that φ( x) = cos kx is an eigenfunction of the dx2
operator, and find its
d2
eigenvalue. That is, show that dx2
(φ( x)) = λ · φ( x) for some constant λ (the eigenvalue).
(b) The wave function ψ for a particle of mass m moving in a one-dimensional box of length L, given
by r
2 πx
ψ( x) = sin for 0 ≤ x ≤ L,
L L
is a solution (assuming zero potential energy) of the time-independent Schrödinger equation
h2 d 2ψ
− = E ψ( x)
8π2 m dx2
where h is Planck’s constant and E is a constant that represents the total energy of the wave
function. Find an expression for the constant E in terms of the other constants. Notice that this
d2
makes ψ( x) an eigenfunction of the dx 2 operator.
CHAPTER 2
Derivatives of Common Functions
2.1 Inverse Functions

The derivatives calculated in the previous chapter were mostly for polynomials and a few
trigonometric functions. This chapter will show how to find the derivatives of other types of
functions, beginning in this section with inverse functions. The idea here is that if a function
is differentiable and has an inverse then that inverse function is also differentiable.
Recall that a function is a rule that assigns a single object Domain Range
y from one set (the range) to each object x from another set f
(the domain). That rule can be written as y = f (x), where f is x y
y = f ( x)
the function (see Figure 2.1.1). There is a simple vertical rule for
determining whether a rule y = f (x) is a function: f is a function Figure 2.1.1
if and only if every vertical line intersects the graph of y = f (x)
in the x y-coordinate plane at most once (see Figure 2.1.2).
y y
y = f ( x) y = f ( x)
x x
(a) f is a function (b) f is not a function
Figure 2.1.2 Vertical rule for functions
Recall that a function f is one-to-one (often written as 1 − 1) if it assigns distinct values

of y to distinct values of x. In other words, if x1 6= x2 then f (x1 ) 6= f (x2 ). Equivalently, f is
one-to-one if f (x1 ) = f (x2 ) implies x1 = x2 . There is a simple horizontal rule for determining
whether a function y = f (x) is one-to-one: f is one-to-one if and only if every horizontal line
intersects the graph of y = f (x) in the x y-coordinate plane at most once (see Figure 2.1.3).
37
38 Chapter 2 • Derivatives of Common Functions §2.1
y y
y = f ( x) y = f ( x)
x x
(a) f is one-to-one (b) f is not one-to-one
Figure 2.1.3 Horizontal rule for one-to-one functions
If a function f is one-to-one on its domain, then f has an inverse function, denoted by f −1,
such that y = f (x) if and only if f −1 (y) = x. The domain of f −1 is the range of f .
The basic idea is that f −1 “undoes” what f does, and vice versa. In other words,
f −1 ( f (x)) = x for all x in the domain of f , and
−1
f (f (y)) = y for all y in the range of f .
Intuitively it is clear that a function is one-to-one (and hence invertible) when it is either
strictly increasing or strictly decreasing; if the function, say, increases and then decreases (as
in Figure 2.1.3(b)) then the horizontal rule would be violated around that “turning point.” For
differentiable functions a positive derivative means the function is increasing, while a negative
derivative means the function is decreasing (this will be proved in Chapter 3).
However, a function can still be one-to-one even if its derivative is zero only at isolated points
(i.e. not identically zero over an entire interval of points) and either positive everywhere else
or negative everywhere else. For example, the function f (x) = x3 has derivative f ′ (x) = 3x2 ,
which is zero only at the isolated point x = 0 and positive for all other values of x. Clearly
f is one-to-one over the set of all real numbers (why?) and hence it has an inverse function
p
x = f −1 (y) = 3 y defined for all real numbers y (i.e. the range of f ).
Thus, having a derivative that is either always positive or always negative is sufficient for a
function to be one-to-one but not necessary. Having a nonzero derivative is necessary, though,
for the inverse function to be differentiable.
In algebra you learned that ab = 1b for all real numbers a 6= 0 and b 6= 0 (and hence ab 6= 0).
a
The same holds true for the infinitesimals dy and dx (nonzero by definition) since they can
be treated like numbers, which immediately yields a formula for the derivative of an inverse
function:
Derivative of an Inverse Function: If y = f (x) is differentiable and has an inverse

function x = f −1 (y), then f −1 is differentiable and its derivative is
dx 1 dy
= dy if 6= 0.
dy dx
dx
dy
The inverse of a function would still exist at a point where dx = 0 but it would not be differen-
tiable there, since its derivative would be the undefined quantity 10 .
Inverse Functions • Section 2.1 39
dy 1
Since y is a function of x, dx will be in terms of x and hence dy will be in terms of x. However,
dx
dx
since (by invertibility) x is a function of y, dy would normally be in terms of y, not x, so that
dx 1
the two sides of the equation dy = dy are not in the same terms! One way to handle this
dx
discrepancy is to use the formula y = f (x) to solve for x in terms of y then substitute that
dy dx 1
expression into dx , so that dy = dy is now in terms of y. That might not always be possible,
dx
however (e.g. try solving for x in the formula y = x sin x).
Example 2.1
Find the inverse f −1 of the function f ( x) = x3 then find the derivative of f −1 .
Solution: The function y = f ( x) = x3 is one-to-one over the set of all real numbers (why?) so it has an
p
inverse function x = f −1 ( y) defined for all real numbers, namely x = f −1 ( y) = 3 y.
The derivative of f −1 is
dx 1 1
= dy = , which is in terms of x, so putting it in terms of y yields
dy 3 x2
dx
1 1
= ¡ p ¢2 = 2/3
3 y3 3 y
p
which agrees with the derivative obtained by differentiating
p x = 3 y directly. Note that this derivative
3
is defined for all y except y = 0, which occurs when x = 0 = 0, i.e. at the point ( x, y) = (0, 0).
Functions are often expressed in terms of x, so it is common to see an inverse function also
p p
expressed in terms of x: writing the inverse of f (x) = x3 as f −1 (x) = 3 x (not as f −1 (y) = 3 y)),
as confusing as that might be. In that case, the idea is to switch the roles of x and y in the
original function y = f (x), making it x = f (y), and then write y = f −1 (x) and use
dy 1
= dx
dx
dy
−1
to put the derivative of f in terms of x, following the same procedure mentioned earlier.
Example 2.2
Find the inverse f −1 of the function f ( x) = x3 then find the derivative of f −1 .
p
Solution: Rewrite y = f ( x) = x3 as x = f ( y) = y3 , so that its inverse function y = f −1 ( x) = 3
x has deriva-
tive
dy 1 1
= dx = , which is in terms of y, so putting it in terms of x yields
dx 3 y2
dy
1 1
= ¡p ¢ 2
=
3 3x 3 x2/3
p
which agrees with the derivative obtained by differentiating y = 3
x directly.
To obtain a formula in prime notation for the derivative of an inverse function, notice that
for all x in the domain of an invertible differentiable function f ,
d ¡ −1 ¢ d ¡ ¢′
f −1 ( f (x)) = x ⇒ f ( f (x)) = (x) ⇒ f −1 ( f (x)) · f ′ (x) = 1
dx dx
by the Chain Rule, and hence:
¡ ¢′ 1
f −1 ( f (x)) = if f ′ (x) 6= 0
f ′ (x)
Two equivalent ways to write this are:

¡ ¢′ 1
f −1 (c) = where c = f (a) and f ′ (a) 6= 0
f ′ (a)
and
¡ ¢′ 1
f −1 (x) = if f ′ ( f −1(x)) 6= 0
f ′ ( f −1 (x))
Exercises
A
For Exercises 1-8, show that the given function y = f ( x) is one-to-one over the given interval, then find
the formulas for the inverse function f −1 and its derivative. Use Example 2.2 as a guide, including
putting f −1 and its derivative in terms of x.
1. f ( x) = x, for all x 2. f ( x) = 3 x, for all x

p
3. f ( x) = x2 , for all x ≥ 0 4. f ( x) = x, for all x ≥ 0
5. f ( x) = 1x , for all x > 0 6. f ( x) = 1x , for all x < 0

1
7. f ( x) = x2
, for all x > 0 8. f ( x) = x5 , for all x
p
9. The unit circle x2 + y2 = 1 does not define y as a single function of x, since y = ± 1 − x2 defines two
separate functions. But p the part of the unit circle in the first quadrant, i.e. for 0 ≤ x ≤ 1 and 0 ≤ y ≤ 1,
does define y = f ( x) = 1 − x2 as a single function of x that is one-to-one on the interval [0, 1]. Find
the formulas for its inverse function f −1 and its derivative.
B
10. Show that if f is differentiable and invertible, and if f −1 is twice-differentiable, then
¡ ¢′′ f ′′ ( f −1 ( x))
f −1 ( x) = − ¡ ¢3 .
f ′ ( f −1 ( x))
Trigonometric Functions and Their Inverses • Section 2.2 41
2.2 Trigonometric Functions and Their Inverses

The graphs of the six trigonometric functions are shown in Figure 2.2.1:
y y y
1 1
0 x 0 x x
π π 3π 2π π π 3π 2π − π2 0 π
2 2 2 2 2
−1 −1
(a) y = sin x (b) y = cos x (c) y = tan x
y y y
1 1
0 x 0 x 0 x
π π 3π 2π π π 3π 2π π π
2 2 2 2 2
−1 −1
(d) y = csc x (e) y = sec x (f) y = cot x
Figure 2.2.1 Graphs of the six trigonometric functions
Recall that sin x, cos x, csc x, and sec x have a period of 2π (i.e. repeat the same values every
2π radians), while tan x and cot x have a period of π.
The derivatives of the six trigonometric functions—given in Section 1.4—are:
d d
(sin x) = cos x (csc x) = − csc x cot x
dx dx
d d
(cos x) = − sin x (sec x) = sec x tan x
dx dx
d d
(tan x) = sec2 x (cot x) = − csc2 x
dx dx
The six trigonometric functions are not one-to-one over their entire domains, but recall from
trigonometry that they are one-to-one when restricted to smaller domains, and hence have
inverse functions, called the inverse trigonometric functions.
£ ¤
For example, y = sin x is one-to-one over the interval − π2 , π2 , as shown in Figure 2.2.2 below:
y
y = sin x
1
x
−π − π2 0 π π
2
−1
£ ¤
Figure 2.2.2 y = sin x is one-to-one with x restricted to − π2 , π2
Similarly, recall that cos x is one-to-one over [0, π], tan x is one-to-one over (−π/2, π/2), csc x
is one-to-one over (−π/2, 0) ∪ (0, π/2), sec x is one-to-one over (0, π/2) ∪ (π/2, π), and cot x is one-
to-one over (0, π). Hence, the inverse trigonometric functions sin−1 x, cos−1 x, tan−1 x, csc−1 x,
sec−1 x and cot−1 x are defined,1 with the following domains and ranges:
function sin−1 x cos−1 x tan−1 x csc−1 x sec−1 x cot−1 x
domain [−1, 1] [−1, 1] (−∞, ∞) | x| ≥ 1 | x| ≥ 1 (−∞, ∞)

¡ π π¢ ¡ π ¢ ¡ π¢ ¡ π¢ ¡π ¢
range [− π2 , π2 ] [0, π] −2, 2 − 2 , 0 ∪ 0, 2 0, 2 ∪ 2 , π (0, π)
The graphs of all six inverse trigonometric functions are shown in Figures 2.2.3 and 2.2.4
below:
y y y
π
π π 2
2
x x
0 0
−1 1
− π2 x − π2
−1 0 1
(a) y = sin−1 x (b) y = cos−1 x (c) y = tan−1 x
Figure 2.2.3 Graphs of sin−1 x, cos−1 x, tan−1 x
1 The arc notation arcsin x, arccos x, arctan x, arccsc x, arcsec x, arccot x is often used in place of sin−1 x, cos−1 x,
tan−1 x, csc−1 x, sec−1 x, cot−1 x, respectively.

Trigonometric Functions and Their Inverses • Section 2.2 43
y y y
π π
2 π
0 π
x 2 π
−1 1 2
− π2 x x
−1 0 1 0
(a) y = csc−1 x (b) y = sec−1 x (c) y = cot−1 x
Figure 2.2.4 Graphs of csc−1 x, sec−1 x, cot−1 x
The derivatives of the six inverse trigonometric functions are:
d 1 d 1
(sin−1 x) = p (for | x | < 1) (csc−1 x) = − p (for | x | > 1)
dx 1 − x2 dx | x | x2 − 1
d 1 d 1
(cos−1 x) = − p (for | x | < 1) (sec−1 x) = p (for | x | > 1)
dx 1 − x2 dx | x | x2 − 1
d 1 d 1
(tan−1 x) = (cot−1 x) = −
dx 1 + x2 dx 1 + x2
For the derivative of cos−1 x, recall that y = cos−1 x is an angle between 0 and π radians, defined
dx
for −1 ≤ x ≤ 1. Since cos y = x by the definition of y, then dy = − sin y and
p p
sin2 y = 1 − cos2 y = 1 − x2 ⇒ sin y = ± 1 − x2 = 1 − x2
since 0 ≤ y ≤ π (which means sin y must be nonnegative). Thus:

d dy 1 1 1
(cos−1 x) = = dx = = −p X
dx dx − sin y 1 − x2
dy
For the derivative of sec−1 x, since y = sec−1 x is defined for | x | ≥ 1, then 0 ≤ y < π/2 for x ≥ 1
and π/2 < y ≤ π for x ≤ −1. Recall also that sec y and tan y are both positive when 0 < y < π/2
and are both negative when π/2 < y < π. So in both cases the product sec y tan y is nonnegative,
i.e. sec y tan y = | sec y tan y |. Thus, since sec y = x and
q p
1 + tan2 y = sec2 y ⇒ tan2 y = sec2 y − 1 ⇒ tan y = ± sec2 y − 1 = ± x2 − 1
then for | x | > 1:

d dy 1 1 1 1 1
(sec−1 x) = = dx = = = ¯ p ¯ = p X
dx dx sec y tan y | sec y tan y | ¯ x x2 − 1 ¯ | x | x2 − 1
dy
The proofs of the derivative formulas for the remaining inverse trigonometric functions are
similar, and are left as exercises.
Example 2.3
Find the derivative of the function y = 3 tan (π − 2 x).
Solution: By the Chain Rule with u = π − 2 x, the derivative of y = 3 tan (π − 2 x) = 3 tan u is:
dy dy du ¡ ¢
= · = 3 sec2 u (−2) = −6 sec2 (π − 2 x)
dx du dx
Example 2.4
Find the derivative of the function y = sin−1 ( x/4).
Solution: By the Chain Rule with u = x/4, the derivative of y = sin−1 ( x/4) = sin−1 u is:
dy dy du 1 1 1 1
= · = p · = p = p
dx du dx 1−u 2 4 2
4 1 − ( x /16) 16 − x2
Exercises
A
For Exercises 1-16, find the derivative of the given function y = f ( x).
1. y = sec2 3 x 2. y = csc( x2 + 1) 3. y = cot 3 x 4. y = cos (tan x)
5. y = tan−1 ( x/3) 6. y = sec−1 ( x2 + 1) 7. y = cot−1 3 x 8. y = cos−1 (sin x)

p ¡ ¢2
9. y = cot−1 (1/ x) 10. y = tan−1 x 11. y = sin−1 3 x 12. y = tan−1 1x + tan−1 x
x −1
13. y = tan−1 x +1 14. y = x sin−1 (2 x + 1) 15. y = x cot−1 x 16. y = tan−1 1x + cot−1 x
17. Find the derivative of y = sin−1 x + cos−1 x . Explain why no derivative formulas were needed.
B
For Exercises 18-21 prove the given derivative formula.
d 1 d 1
18. (sin−1 x) = p 19. (tan−1 x) =
dx 1 − x2 dx 1 + x2
d 1 d 1
20. (cot−1 x) = − 21. (csc−1 x) = − p
dx 1 + x2 dx | x | x2 − 1
22. The Chebyshev polynomials T n ( x) = cos (n cos−1 x) are defined for all | x | ≤ 1 and n = 0, 1, 2,... .
(a) Show that the Chebyshev polynomials T n ( x) satisfy the differential equation
(1 − x2 ) T n′′ ( x) − x T n′ ( x) + n2 T n ( x) = 0
(b) Find polynomial expressions for T0 ( x), T1 ( x) and T2 ( x).

(c) Show that T n+1 ( x) + T n−1 ( x) = 2 x T n ( x) for all n ≥ 1. (Hint: Write θ = cos−1 x so that cos θ = x)
The Exponential and Natural Logarithm Functions • Section 2.3 45
2.3 The Exponential and Natural Logarithm Functions

Functions of the form a x , where the exponent x varies, are called exponential functions. Unless
otherwise noted, assume that a > 0 (0 x is just 0, and (−1)1/2 is not a real number). You already
know how a x is defined when the exponent x is a rational number
p (i.e. x = m/n
p where m and n
are integers, n 6= 0).pBut what if x were irrational, such as 2? What would 3 2 mean?
The idea is that 2 = 1.414213562 . . . can be approximated by rational numbers 14/10 = 1.4,
141/100 = 1.41, 1414/1000 = 1.414, 14142/10000 = 1.4142, and so on, taking larger and larger
numeratorsp and denominators
p in the rational approximations to get more and more decimal
2
places of 2. Then 3 would be the number that 3 raised to those rational approximations
approaches: p
3 2 = limp 3m/n
m/n→ 2
m/n rational
Each quantity 3m/n is p defined inside the above limit, and as the rational numbers m/n get
closer to the value of 2 it can be shown the limit of the values of 3m/n will exist:2
14
31.4 = 3 10 = 4.65553672174608
141
31.41 = 3 100 = 4.70696500171657
1414
31.414 = 3 1000 = 4.72769503526854
14142
31.4142 = 3 10000 = 4.72873393017119
141421
31.41421 = 3 100000 = 4.72878588090861
1414213
31.414213 = 3 1000000 = 4.72880146624114
.. ..
. .
p
31.414213562... = 3 2
= 4.72880438783742
Of course you would never do all this by hand—you would simply use a computer or calculator,
which use much more efficient algorithms for calculating powers in general.3
All the usual rules of exponents that you learned in algebra apply to a x when defined in
the manner described above, with a > 0 and x varying over all real numbers. Of all the pos-
sible values for the base a, the one that appears the most in mathematics, the sciences and
engineering is the base e, defined as:
µ ¶
1 x
e = lim 1 + (2.1)
x→∞ x
The approximate value of is e = 2.71828182845905 . . . (often called the Euler number).
2 For the general case see pp.61-63 in F RANKLIN, P., A Treatise on Advanced Calculus, New York: Dover Publica-
tions, Inc., 1964.

3 For example, to see how square roots and cube roots are calculated see Chapter 2 in F IKE , C.T., Computer
Evaluation of Mathematical Functions, Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1968.
The limit in this definition means that as x becomes larger—approaching infinity (∞)—the
¡ ¢x
values of 1 + 1x approach a number, denoted by e. More decimal places for e can be obtained
by making x sufficiently large.4 For example, when x = 5 × 106 the value is 2.718281555200129.
For extremely large values of x, that is, when x ≫ 1 (the symbol ≫ means “much larger than”),
µ ¶x µµ ¶ x ¶1/ x ³ ´ µ ¶
1 1/ x 1 1 1/ x 1
e ≈ 1 + ⇒ e ≈ 1 + = 1 + ⇒ e − 1 x ≈ x = 1,
x x x x
so letting h = 1/x, and noting that h = 1/x → 0 if and only if x → ∞, yields the useful limit:5
eh − 1
lim = 1 (2.2)
h→0 h
Using the above limit, the derivative of y = e x can be found:
d ¡ x¢
e = ex
dx
Proof: Using the limit definition of the derivative for f (x) = e x ,
d ¡ x¢ f (x + h) − f (x) e x+ h − e x
e = lim = lim
dx h→0 h h→0 h
¡ ¢
ex eh − 1 eh − 1
= lim = e x · lim (since e x does not depend on h)
h→0 h h→0 h
= ex · 1 = ex
QED
In general, for a differentiable function u = u(x) as the exponent the Chain Rule yields:
d ¡ u¢ du
e = eu ·
dx dx
Example 2.5
2
Find the derivative of y = 4 e− x .
2 2 2
(− x2 ) = 4 e− x · (−2 x) = −8 xe− x
dy
Solution: dx = 4 e− x · d
dx
4 It will be shown in Chapter 9 that this limit does in fact exist.

5 This is admittedly a “hand waving” argument. In Chapter 3 a more exact method will be discussed for proving
limits like this one.

The function e x is often referred to simply as the exponential function, even though there
are obviously many exponential functions. What makes the base e so special? Take y = A e kt to
represent the amount of some physical quantity at time t, for some constants A and k. Then
dy d ³ kt´
= Ae = k · A e kt = k y ,
dt dt
which says that the instantaneous rate of change of the quantity is directly proportional to
the amount present at that instant. It turns out that many physical quantities exhibit that
behavior, some of which will be discussed shortly. Conversely, in Chapter 5 it will be shown
dy
that any solution to the differential equation dt = k y must be of the form y = A e kt for some
constant A. This is what gives the exponential function its special significance.
Let f (x) = e x be the exponential function. Then f (x) > 0 for all x and f ′ (x) = f (x) = e x > 0 for
all x, and so f (x) is strictly increasing. The graph is shown in Figure 2.3.1.
y y
y = ln x
y = ex
1
x
0 1
x
0
Figure 2.3.1 The exponential function Figure 2.3.2 Natural logarithm function
Thus, the exponential function is one-to-one over the set of all real numbers and hence has
an inverse function, called the natural logarithm function, denoted (as a function of x) as
f −1 (x) = ln x. The graph is shown in Figure 2.3.2. Below is a summary of the relationship
between e x and ln x:
domain of ln x = all x > 0 = range of e x

range of ln x = all x = domain of e x
y = ex if and only if x = ln y (2.3)

ln x
e = x for all x > 0 (2.4)
¡ x¢
ln e = x for all x (2.5)
The reader should be aware that many—if not most—fields outside of mathematics use the
notation log x instead of ln x for the natural logarithm function.6
6 This text almost used log x as well, prevented only by the desire for compatibility with other mathematics texts.
From algebra you should be familiar with the following properties of the natural logarithm,
along with their equivalent properties in terms of the exponential function:7
ln(ab) = ln a + ln b e a · e b = e a+ b
³a´ ea
ln = ln a − ln b = e a− b
b eb
¡ a ¢b
ln a b = b ln a e = e ab
ln 1 = 0 e0 = 1
To find the derivative of y = ln x, use x = e y :
dy 1 1 1 1
= dx = d
= y
=
dx
dy dy (e y ) e x
Hence:
d 1
(ln x) =
dx x
In general, for a differentiable function u = u(x), the Chain Rule yields:
d 1 du u′
(ln u) = · =
dx u dx u
Example 2.6
¡ ¢
Find the derivative of y = ln x2 + 3 x − 1 .
dy 1 d 2 2x + 3
Solution: = 2 · ( x + 3 x − 1) = 2
dx x + 3 x − 1 dx x + 3x − 1
Recall that | x | = − x for x < 0, in which case ln(− x) is defined and
d d 1 1
(ln | x |) = (ln(− x)) = · (−1) = .
dx dx −x x
³ ´
7 Note that when using the formula ln a = ln a − ln b in numerical computations—especially on hand-held
b ³ ´
calculators—it is preferable to use the left side of the equation, i.e. ln ab , since the right side ln a − ln b is vulner-
able to the problem of subtractive cancellation, which can give an incorrect answer of 0 if a and b are nearly equal.
For a discussion of subtractive cancellation see § 1.3 in H ENRICI , P., Essentials of Numerical Analysis, with Pocket
Calculator Demonstrations, New York: John Wiley & Sons, Inc., 1982.
Combine that result with the derivative d

dx (ln x) = 1x for x > 0 to get:
d 1
(ln | x |) =
dx x
Logarithmic Differentiation
For some functions it is easier to differentiate the natural logarithm of the function
first and then solve for the derivative of the original function. This technique is called
logarithmic differentiation, demonstrated in the following two examples.
Example 2.7
Find the derivative of y = x x .
Solution: For this example assume x > 0 (since x is both the base and the exponent). Note that you
cannot use the Power Rule for this function since the exponent x is a variable, not a fixed number.
Instead, take the natural logarithm of both sides of the equation y = x x and then take the derivative of
both sides and solve for y′ :
¡ ¢
ln y = ln x x = x · ln x
d d
(ln y) = ( x · ln x)
dx dx
y′ 1
= 1 · ln x + x ·
y x
y′ = y (ln x + 1) = x x (ln x + 1)
Example 2.8
(2 x+1)7 (3 x3 −7 x+6)4
Find the derivative of y = (1+sin x)5
.
Solution: Use logarithmic differentiation by taking the natural logarithm of y and then use properties
of logarithms to simplify the differentiation before solving for y′ :
µ ¶
(2 x + 1)7 (3 x3 − 7 x + 6)4 ¡ ¢ ¡ ¢
ln y = ln = ln (2 x + 1)7 (3 x3 − 7 x + 6)4 − ln (1 + sin x)5
(1 + sin x)5
¡ ¢ ¡ ¢ ¡ ¢
= ln (2 x + 1)7 + ln (3 x3 − 7 x + 6)4 − ln (1 + sin x)5
d d ¡ ¢
(ln y) = 7 ln (2 x + 1) + 4 ln (3 x3 − 7 x + 6) − 5 ln (1 + sin x)
dx dx
y′ 2 9 x2 − 7 cos x
= 7· + 4· 3 − 5·
y 2x + 1 3x − 7x + 6 1 + sin x
µ ¶
14 36 x2 − 28 5 cos x
y′ = y · + −
2x + 1 3 x3 − 7 x + 6 1 + sin x
µ ¶
(2 x + 1)7 (3 x3 − 7 x + 6)4 14 36 x2 − 28 5 cos x
= · + −
(1 + sin x)5 2x + 1 3 x3 − 7 x + 6 1 + sin x
Radioactive decay
dy
A classic example of the differential equation dt = k y is the case of exponential decay
of a radioactive substance, often referred to simply as radioactive decay. In this case
the general solution y = Ae kt represents the amount of the substance at time t ≥ 0,
dy
and the decay constant k is negative: dt < 0 since the substance is decaying (i.e. the
dy
amount of substance is decreasing) while y > 0 , so dt
= k y implies that k < 0.
The constant A is the initial amount of the y
substance, i.e. the amount at time t = 0: y(0) =
Ae0 t = Ae0 = A . For this reason A is some-
A0
times denoted by A 0 . The constant k turns
out to be related to the half-life of the sub- y = A 0 e kt (k < 0)
stance, defined as the time t H required for 1
A
half the current amount of substance to de- 2 0
cay (see Figure 2.3.3).
You might be tempted to think that the
t
half-life is not a constant, that it might change 0 tH
depending on the amount of substance present.
Figure 2.3.3
For example, perhaps it would take longer
for 100 g of the substance to decay to 50 g than it would for 10 g to decay to 5 g. How-
ever, this is not so. To see why, pick any t ≥ 0 as the current time, so that y( t) = A 0 e kt
is the current amount of the substance. By definition, that amount should be halved
when the time t H has passed, that is, y( t + t H ) = 21 y( t). Then t H does turn out to be
independent of the initial amount A 0 and depends only on k, since
1 1 1
y( t + t H ) = y( t) ⇒ A 0 e k ( t+ t H ) = A 0 e kt ⇒ A✚
✚ e✚
0✚
kt
· e kt H = ✚ A✚e✚
0✚
kt
2 2 2
µ ¶
1 1
⇒ e kt H = ⇒ kt H = ln = − ln 2
2 2
and so:
ln 2 ln 2
tH = − and k = − (2.6)
k tH
Example 2.9
Suppose that 5 mg of a radioactive substance decays to 3 g in 6 hours. Find the half-life of the substance.
Solution: Consider A 0 = 5 mg as the initial amount, so that y( t) = 5 e kt is the amount at time t ≥ 0 hours.
Use the given information that y(6) = 3 mg to find k, the decay constant of the substance:
µ ¶
3 1
3 = y(6) = 5 e k6 ⇒ 6k = ln ⇒ k = ln 0.6
5 6
Then the half-life t H is:

ln 2 ln 2
tH = − = −1 = 8.14 hours
6 ln 0.6
k
Note in the above example that the given time t = 6 was used for finding the constant
k and then the half-life t H . For the converse problem—given the half-life find the time
required for a certain amount to decay—you would do the opposite: use the given t H
to find k and then solve for the required time t from the equation y( t) = A 0 e kt .
Example 2.10
dy
Another example of the differential equation dt = k y is exponential growth of cell bacteria, in which
case k > 0 since the number of cells y( t) at time t is increasing.
Example 2.11
Another example is for the current I in a simple series electric R
circuit with a constant direct current (DC) source of voltage V , a
capacitor with capacitance C , a resistor with resistance R , and a
switch, as in Figure 2.3.4. If the capacitor is initially uncharged V I ( t) C
when the switch is open, and if the switch is closed at time t = 0,
then the current I ( t) through the circuit at time t ≥ 0 satisfies (by
Kirchoff ’s Second Law) the differential equation
Figure 2.3.4
dI dI I
RC + I = 0 ⇒ = −
dt dt RC
so that I ( t) = I 0 e− t/RC where I 0 is the initial current at t = 0. Ohm’s Law says that V = I 0 R , so
V − t/RC
I ( t) = e
R
is the current at time t ≥ 0, which decreases exponentially.
Example 2.12
In the previous examples the quantities that decayed or grew exponentially did so as functions of
time. There are other possible variables besides time, though. For example, the atmospheric pres-
sure p measured as a function of height h above the surface of the Earth satisfies—assuming constant
temperature—the differential equation
dp w0
= − p
dh p0
where p0 is the pressure at height h = 0 (i.e. ground level) and w0 is the weight of a cubic foot of air at
pressure p0 (with air pressure measured in lbs per square foot and height measured in feet). Thus,
w
− p0 h
p(h) = p0 e 0 .
So the atmospheric pressure decreases exponentially as the height above the ground increases.
Exercises
A
2
1. y = e2 x 2. y = xe x 3. y = e− x − e x
1 + e x 1
4. y = esin x 5. y = 6. y =
1 − ex 1 + e −2 x
8. y = e2 ln x
x
7. y = e e 9. y = ln (3 x)
¡ ¢3
10. y = ln ( x2 + 2 x + 1)4 11. y = ln(tan x2 ) 12. y = ln ( e x + e2 x )
d 1
13. Show that (ln (kx)) = for all constants k > 0.
dx x
d n
14. Show that (ln ( x n )) = for all integers n ≥ 1.
dx x
dy
For Exercises 15-18, use logarithmic differentiation to find dx .
2 ( x + 2)8 (3 x − 1)7
15. y = x x 16. y = xln x 17. y = xsin x 18. y =
(1 − 5 x)4
B
19. Suppose it takes 8 hours for 30% of a radioactive substance to decay. Find the half-life of the
substance.
20. The radioactive isotope radium-223 has a half-life of 11.43 days. How long would it take for 3 kg of
radium-223 to decay to 1 kg?
21. If a certain cell population grows exponentially—i.e. is of the form A 0 e kt with k > 0—and if the
population doubles in 6 hours, how long would it take for the population to quadruple?
For Exercises 22-25, use induction to prove the given formula for all n ≥ 0.
d n ¡ kx ¢ dn
22. e = k n e kx (any constant k 6= 0) 23. (x e x ) = ( x + n) e x
dx n dx n
dn d n +1 n!
24. ( x e− x ) = (−1)n ( x − n) e− x 25. 1
( x n ln x) =
dx n dx n + x
1
26. Show that f ′ ( x) = f ( x) (1 − f ( x)) for the sigmoid neuron function f ( x) = . This derivative
1 + e− x
relation is used in neural network learning algorithms.
³p ´
27. If y = Ce−κ t cos n2 − κ2 t + γ then show that
d2 y dy
2
+ 2κ + n2 y = 0
dt dt
for all constants C , n, κ, γ, with 0 ≤ κ ≤ n.
C
dy
28. Suppose that e y + e x = e y+ x . Show that dx = − e y− x .
d
29. For an infinitesimal dx show that e dx = 1 + dx. (Hint: Use dx ( e x ) = e x .)
30. For an infinitesimal dx show that ln (1 + dx) = dx.
General Exponential and Logarithmic Functions • Section 2.4 53
2.4 General Exponential and Logarithmic Functions

For a general exponential function y = a x , with a > 0, use logarithmic differentiation
to find its derivative:
¡ ¢
ln y = ln a x = x ln a
d d
(ln y) = ( x ln a) = ln a
dx dx
y′
= ln a ⇒ y′ = y · ln a
y
Thus, the derivative of y = a x is:
d ¡ x¢
a = (ln a) a x
dx
In general, for an exponent of the form u = u( x):
d ¡ u¢ du
a = (ln a) a u ·
dx dx
Example 2.13
Find the derivative of y = 2cos x .
Solution: This is the case where a = 2, so:
dy d
= (ln 2) 2cos x · (cos x) = −(ln 2) (sin x) 2cos x
dx dx
Note that any exponential function y = a x can be expressed in terms of the exponen-
tial function e x . Since
a x > 0 ⇒ eln (a ) = a x ,
x
and since ln (a x ) = x ln a, then:
a x = e x ln a
Computers and calculators often use the above formula to calculate a x .

The function y = a x has an inverse for any a > 0, except for a = 1 (in that case
y = 1 x = 1 is just a constant function). To see this, notice that since a x > 0 for all x, and
dy
ln a < 0 for 0 < a < 1, while ln a > 0 for a > 1, then dx = (ln a) a x is always negative if
0 < a < 1 and always positive if a > 1. Thus, y = a x is a strictly decreasing function if
0 < a < 1, and it is a strictly increasing function if a > 1. The graphs in each case are
shown in Figure 2.4.1.
y y
a>1
a<1
a>1 x
0 1
1 a<1
x
0
Figure 2.4.1 y = ax Figure 2.4.2 y = loga x
Hence, for any a > 0 with a 6= 1 the function f ( x) = a x is one-to-one, so it has an

inverse function, called the base a logarithm and denoted by f −1 ( x) = loga x. It is
often spoken as “log base a of x”. The graphs for a < 1 and a > 1 are shown in Figure
2.4.2. Note that the natural logarithm is just the base a logarithm in the special case
with a = e, i.e. ln x = log e x. The base a logarithm has properties similar to those of the
natural logarithm (and the corresponding properties of a x ):
loga ( bc) = loga b + loga c a b · a c = a b+ c

µ ¶
b ab
loga = loga b − loga c c
= a b− c
c a
³ ´c
loga b c = c loga b a b
= a bc
loga 1 = 0 a0 = 1
Note that loga x can be put in terms of the natural logarithm, since
³ ´
x = aloga x ⇒ ln x = ln aloga x = (loga x) · (ln a)
so dividing the last expression by ln a gives:
ln x
loga x =
ln a
The above formula is useful on calculators that do not have a loga x key or function.
Taking the derivative of both sides yields:
General Exponential and Logarithmic Functions • Section 2.4 55
d ¡ ¢ 1
loga x =
dx x ln a
In general, when taking the logarithm of a function u = u( x):
d ¡ ¢ 1 du u′
loga u = · =
dx u ln a dx u ln a
Example 2.14
Find the derivative of y = log2 (cos 4 x).
Solution: This is the case where a = 2, so:
dy 1 d 4 sin 4 x
= · (cos 4 x) = −
dx (cos 4 x) (ln 2) dx (ln 2) (cos 4 x)
The number a is the base of both the logarithm function loga x and the exponential
function a x . Base 2 and base 10 are the most commonly used bases other than base
e. Base 10 is how numbers are normally expressed, as combinations of powers of 10
(e.g. 2014 = 2 · 103 + 0 · 102 + 1 · 101 + 4 · 100 ). Base 2 is especially useful in computer
science, since computers represent all numbers in binary format, i.e. as a sequence
of zeros and ones, indicating how many successive powers of two to take and then
sum up.8 For example, the number 6 is represented in binary format as 110, since
1 · 22 + 1 · 21 + 0 · 20 = 4 + 2 + 0 = 6.
Exercises
A
3 x + 3− x
2. y = 2ln 3 x 3. y = 22
x
1. y =
2
−1 x
4. y = tan π 5. y = log2 ( x2 + 1) 6. y = log10 e x
¡ ¢
7. y = sin log2 π x 8. y = log2 42 x 9. y = 8log2 x
B
kx dy
10. Show that for all constants k the function y = Aa ln a satisfies the differential equation dx = k y.
Does this contradict the statement made in Section 2.3 that the only solution to that differential
equation is of the form y = Ae kx ? Explain your answer.
8 Binary notation leads to the joke “There are 10 kinds of people in the world: those who understand binary and
those who do not.”

CHAPTER 3
Topics in Differential Calculus
3.1 Tangent Lines

Everyone knows that the Earth is not flat, but locally, e.g. in your immediate vicinity,
isn’t the Earth effectively flat? In other words, “flat” is a fairly good approximation of
the Earth’s surface “near” you, and it simplifies matters enough for you to do some
useful things.
This idea of approximating curved shapes by straight shapes is a frequent theme
in calculus. Recall from Chapter 1 that by the Microstraightness Property a differen-
tiable curve y = f ( x) actually is a straight line over an infinitesimal interval, having
dy
slope dx . The extension of that line to all values of x is called the tangent line:
For a curve y = f ( x) that is differentiable at x = a, the tangent line to the curve at

the point P = (a, f (a)) is the unique line through P with slope m = f ′ (a). P is called
the point of tangency. The equation of the tangent line is thus given by:
y − f ( a) = f ′ ( a) · ( x − a) (3.1)
Figure 3.1.1 on the right shows the tangent line y

to a curve y = f ( x) at a point P . If you were to look slope = f ′ (a)
at the curve near P with a microscope, it would look
almost identical to its tangent line through P . Why tangent
line
is this line—of all possible lines through P —such a P
y = f ( x)
good approximation of the curve near P ? It is be-
cause at the point P the tangent line and the curve
x
both have the same rate of change, namely, f ′ (a). a
So the curve’s values and the line’s values change
Figure 3.1.1 Tangent line
by roughly the same amount slightly away from P
(where the line and curve have the same value), making their values nearly equal.
56
Tangent Lines • Section 3.1 57
Example 3.1
Find the tangent line to the curve y = x2 at x = 1. y
Solution: By formula (3.1), the equation of the tangent line is y = x2
y − f (a) = f ′ (a) · ( x − a)
with a = 1 and f ( x) = x2 . So f (a) = f (1) = 12 = 1. Both the curve (1, 1) y = 2x − 1

y = x2 and the tangent line pass through the point (1, f (1)) = (1, 1). The 1
derivative of f ( x) = x2 is f ′ ( x) = 2 x, so f ′ (a) = f ′ (1) = 2(1) = 2, which
is the slope of the tangent line at (1, 1). Hence, the equation of the x
tangent line is y − 1 = 2( x − 1), or (in slope-intercept form) y = 2 x − 1. 0 1
The curve and tangent line are shown in Figure 3.1.2. Near the point −1
(1, 1) the curve and tangent line are close together, but the separation
grows farther from that point, especially in the negative x direction. Figure 3.1.2
In trigonometry you probably learned about tangent lines to circles,

where a tangent line is defined as the unique line that touches the cir-
cle at only one point, as in the figure on the right. In this case the tangent
line is always on one side of the circle, namely, the exterior of the circle;
it does not cut through the interior of the circle. In fact, that definition is a special
case of the calculus definition. In general, though, the tangent line to any other type
of curve will not necessarily be on only one side of the curve, as it was in Example 3.1.
Example 3.2
Find the tangent line to the curve y = x3 at x = 0. y
y = x3
3
Solution: Use formula (3.1) with a = 0 and f ( x) = x . Then f (a) = f (0) =
03 = 0. The derivative of f ( x) = x3 is f ′ ( x) = 3 x2 , so f ′ (a) = f ′ (0) = 3(0)2 =
0. Hence, the equation of the tangent line is y − 0 = 0( x − 0), which is y=0
x
y = 0. In other words, the tangent line is the x-axis itself. (0, 0)
As shown in Figure 3.1.3, the tangent line cuts through the curve. In
general it is possible for a tangent line to intersect the curve at more
than one point, depending on the function. Figure 3.1.3
Example 3.3
Find the tangent line to the curve y = sin x at x = 0. y
y = x y = sin x
Solution: Use formula (3.1) with a = 0 and f ( x) = sin x. Then f (a) =
f (0) = sin 0 = 0. The derivative of f ( x) = sin x is f ′ ( x) = cos x, so
f ′ (a) = f ′ (0) = cos 0 = 1. Hence, the equation of the tangent line is x
y − 0 = 1( x − 0), which is y = x, as in Figure 3.1.4. Near x = 0, the (0, 0)
tangent line y = x is close to the line y = sin x, which was shown in
Section 1.3 (namely, sin dx = dx, so that sin x ≈ x for x ≪ 1).
Figure 3.1.4
58 Chapter 3 • Topics in Differential Calculus §3.1
There are several important things to note about tangent lines:

• The slope of a curve’s tangent line is the slope of the curve.
Since the slope of a tangent line equals the derivative of the curve at the point of
tangency, the slope of a curve at a particular point can be defined as the slope of
its tangent line at that point. So curves can have varying slopes, depending on the
point, unlike straight lines, which have a constant slope. An easy way to remember
all this is to think “slope = derivative.”
• The tangent line to a straight line is the straight line itself.

This follows easily from the definition of a tangent line, but is also easy to see with
the “slope = derivative” idea: a straight line’s slope (i.e. derivative) never changes,
so its tangent line—having the same slope—will be parallel and hence must coincide
with the straight line (since they have the points of tangency in common). For
example, the tangent line to the straight line y = −3 x + 2 is y = −3 x + 2 at every
point on the straight line.
• The tangent line can be thought of as a limit of secant lines.

A secant line to a curve is a line that passes through two points on the curve.
Figure 3.1.5 shows a secant line L PQ passing through the points P = ( x0 , f ( x0 )) and
Q = (w, f (w)) on the curve y = f ( x),
y
y = f ( x)
L PQ
f ( w)
Q
TP
P
f ( x0 )
x
x0 w
Figure 3.1.5 Secant line L PQ approaching the tangent line T P as Q → P
As the point Q moves along the curve toward P , the line L PQ approaches the tan-
gent line TP at the point P , provided the curve is smooth at P (i.e. f ′ ( x0 ) exists).
This is because the slope of L PQ is ( f (w) − f ( x0 ))/(w − x0 ), and so
¡ ¢ f (w) − f ( x0 )
lim slope of L PQ = lim = f ′ ( x0 ) = slope of TP
Q →P w → x0 w − x0
which means that as Q approaches P the “limit” of the secant line L PQ has the same
slope and goes through the same point P as the tangent line TP .
• Smooth curves have tangent lines, nonsmooth curves do not.

For example, think of the absolute value function f ( x) = | x |. Its graph has a sharp
edge at the point (0, 0), making it nonsmooth there, as shown in Figure 3.1.6(a)
below. There is no real way to define a tangent line at (0, 0), because as mentioned
in Section 1.2, the derivative of f ( x) does not exist at x = 0. The same holds true for
curves with cusps, as in Figure 3.1.6(b).
y
y = |x|
y = f ( x)
x
0
(a) f ( x) = | x | (b) Curve with a cusp
Figure 3.1.6 Nonsmooth curves: no tangent line at the nonsmooth points
Many lines go through the point of nonsmoothness, some of which are indicated by
the dashed lines in the above figures, but none of them can be the tangent line.
Sharp edges and cusps have to be “smoothed out” to have a tangent line.
As a point moves along a smooth curve, the corresponding tangent lines to the curve
make varying angles with the positive x-axis—the angle is thus a function of x. Let
φ = φ( x) be the smallest angle that the tangent line L to a curve y = f ( x) makes with
the positive x-axis, so that −90◦ < φ( x) < 90◦ for all x (see Figure 3.1.7).
y y = f ( x)
φ( x 3 )
x
x1 x2 L x3
φ( x 1 )
φ ( x 2 ) = 0◦ L
Figure 3.1.7 The angle φ( x) between the tangent line and positive x-axis
As Figure 3.1.7 shows, −90◦ < φ( x) < 0◦ when the tangent line L L
has negative slope, 0◦ < φ( x) < 90◦ when L has positive slope, and rise
φ( x) = 0◦ when L is horizontal (i.e. has zero slope). The slope of a line φ( x )
run
is usually defined as the rise divided by the run in a right triangle,
as shown in the figure on the right. The figure shows as well that by definition of the
tangent of an angle, tan φ( x) also equals the rise (opposite) over run (adjacent). Thus,
since the slope of L is f ′( x), this means that tan φ( x) = f ′ ( x). In other words:
The tangent line to a curve y = f ( x) makes an angle φ( x) with the positive x-axis,
given by
φ( x) = tan−1 f ′ ( x) . (3.2)
Example 3.4
Find the angle φ that the tangent line to the curve y = e2 x at x = − 21 makes y
y = e2 x
with the positive x-axis, such that −90◦ < φ < 90◦ .
Solution: The angle is φ = φ(−1/2) = tan−1 f ‘(−1/2), where f ( x) = e2 x . Since
L
f ′ ( x) = 2 e2 x , then ¡ 1 1¢
−2, e
φ = φ(−1/2) = tan−1 f ‘(−1/2) = tan−1 2 e−1 = tan−1 0.7358 = 36.3◦ . φ
x
The figure on the right shows the tangent line L to the curve at x = − 21 and the angle φ.
You learned about perpendicular lines in elementary y N tangent

geometry. Figure 3.1.8 shows the natural way to define
normal line L
how a line N can be perpendicular to a curve y = f ( x)
line
at a point P on the curve: the line is perpendicular to
y = f ( x)
the tangent line of the curve at P . Call this line N P
the normal line to the curve at P . Since N and L are
x
perpendicular, their slopes are negative reciprocals of a
each other (provided neither slope is 0). The equation
Figure 3.1.8 Normal line N
of the normal line follows easily:
The equation of the normal line to a curve y = f ( x) at a point P = (a, f (a)) is
1
y − f ( a) = − · ( x − a) if f ′ (a) 6= 0. (3.3)
f ′ ( a)
If f ′ (a) = 0, then the normal line is vertical and is given by x = a.
Example 3.5
Find the normal line to the curve y = x2 at x = 1. (Note: This is the curve from Example 3.1.)
Solution: The equation of the normal line is
1
y − f (a) = − · ( x − a)
f ′ (a)
with a = 1, f ( x) = x2 , and f ′ ( x) = 2 x. So f (a) = 1 and f ′ (a) = 2. Hence, the equation of the normal line is
y − 1 = − 21 ( x − 1), or (in slope-intercept form) y = − 21 x + 32 .
Exercises
A
For Exercises 1-12, find the equation of the tangent line to the curve y = f ( x) at x = a.
1. f ( x) = x2 + 1; at x = 2 2. f ( x) = x2 − 1; at x = 2 3. f ( x) = − x2 + 1; at x = 3
4. f ( x) = 1; at x = −1 5. f ( x) = 4 x; at x = 1 6. f ( x) = e x ; at x = 0
x + 1
7. f ( x) = x2 − 3 x + 7; at x = 2 8. f ( x) = ; at x = 0 9. f ( x) = ( x3 + 2 x − 1)3 ; at x = −1
x − 1
p p
10. f ( x) = tan x; at x = 0 11. f ( x) = sin 2 x; at x = 0 12. f ( x) = 1 − x2 ; at x = 1/ 2
13. Find the equations of the tangent lines to the curve y = x3 − 2 x2 + 4 x + 1 which are parallel to the
line y = 3 x − 5.
14. Draw an example of a curve having a tangent line that intersects the curve at more than one point.
For Exercises 15-17, find the angle φ that the tangent line to the curve y = f ( x) at x = a makes with the
positive x-axis, such that −90◦ < φ < 90◦ .
15. f ( x) = x2 ; at x = 2 16. f ( x) = cos 2 x; at x = π/6 17. f ( x) = x2 + 2 x − 3; at x = −1
18. Show that if φ( x) is the angle that the tangent line to a curve y = f ( x) makes with the positive
x-axis such that 0◦ ≤ φ( x) < 180◦ , then
 Ã !


 −1 1

 cos p when f ′ ( x) ≥ 0
 1 + ( f ′ ( x))2
φ( x ) = Ã !

 − 1

 1
 cos when f ′ ( x) < 0.
−
 p
1 + ( f ( x))
′ 2
(Hint: Draw a right triangle.)

For Exercises 19-21, find the angle φ that the tangent line to the curve y = f ( x) at x = a makes with the
positive x-axis, such that 0◦ ≤ φ < 180◦ .
19. f ( x) = x2 ; at x = −1 20. f ( x) = e− x ; at x = 1 21. f ( x) = ln 2 x; at x = 10
For Exercises 22-24, find the equation of the normal line to the curve y = f ( x) at x = a.
p
22. f ( x) = x; at x = 4 23. f ( x) = x2 + 1; at x = 2 24. f ( x) = x2 − 7 x + 4; at x = 3
25. Find the equations of the normal lines to the curve y = x3 − 2 x2 − 11 x + 3 which have a slope of − 14 .
B
26. Show that the area of the triangle formed by the x-axis, the y-axis, and the tangent line to the
curve y = 1/ x at any point P is constant (i.e. the area is the same for all P ).
27. For a constant a > 0, let P be a point on the curve y = ax2 , and let Q be the point where the tangent
line to the curve at P intersects the y-axis. Show that the x-axis bisects the line segment PQ .
28. Let P be a point on the curve y = 1/ x in the first quadrant, and let Q be the point where the tangent
line to the curve at P intersects the x-axis. Show that the triangle ∆ POQ is isosceles, where O is
the origin.
3.2 Limits: Formal Definition

So far only the intuitive notion of a limit has been used, namely:
A real number L is the limit of f ( x) as x approaches a if the values of f ( x) can be

made arbitrarily close to L by picking values of x sufficiently close to a.
That notion can be put in terms of a formal definition as follows:
Let L and a be real numbers. Then L is the limit of a function f ( x) as x approaches

a, written as
lim f ( x) = L ,
x→a
if for any given number ǫ > 0, there exists a number δ > 0, such that
| f ( x) − L | < ǫ whenever 0 < | x − a| < δ .
A visual way of thinking of this definition is shown in Figure 3.2.1 below:

y
y = f ( x)
L+ǫ
| f ( x) − L | < ǫ L
L−ǫ
( ) x
a−δ a a+δ
0 < | x − a| < δ
Figure 3.2.1 lim f ( x) = L
x→ a
Figure 3.2.1 says that for any interval around L on the y-axis, you will be able to
find at least one small interval around x = a (but excluding a) on the x-axis that the
function y = f ( x) maps completely inside that interval on the y-axis. Choosing smaller
intervals around L on the y-axis could force you to find smaller intervals around a on
the x-axis.
In Figure 3.2.1, f ( x) is made arbitrarily close to L (within any distance ǫ > 0) by
picking x sufficiently close to a (within some distance δ > 0). Since 0 < | x − a | < δ
means that x = a itself is excluded, the solid dot at (a, L) could even be a hollow dot.
That is, f (a) does not have to equal L, or even be defined; f ( x) just needs to approach L
as x approaches a. Thus–perhaps counter-intuitively—the existence of the limit does
not actually depend on what happens at x = a itself.
Limits: Formal Definition • Section 3.2 63
Example 3.6
Show that lim x = a for any real number a.
x→ a
Solution: Though the limit is obvious, the following “epsilon-delta” proof shows how to use the formal
definition. The idea is to let ǫ > 0 be given, then “work backward” from the inequality | f ( x) − L | < ǫ to
get an inequality of the form | x − a | < δ, where δ > 0 usually depends on ǫ. In this case the limit is L = a
and the function is f ( x) = x, so since
| f ( x) − a | < ǫ ⇔ | x − a| < ǫ ,
then choosing δ = ǫ means that
0 < | x − a| < δ ⇒ | x − a| < ǫ ⇒ | f ( x) − a | < ǫ ,
which by definition means that lim x = a.

x→ a
Calculating limits in this way might seem silly since—as in Example 3.6—it requires
extra effort for a result that is obvious. The formal definition is used most often in
proofs of general results and theorems. For example, the rules for limits—listed in
Section 1.2—can be proved by using the formal definition.
Example 3.7
Suppose that lim f ( x) and lim g( x) both exist. Show that
x→ a x→ a
³ ´ ³ ´
lim ( f ( x) + g( x)) = lim f ( x) + lim g( x)
x→ a x→ a x→ a
Solution: Let lim f ( x) = L 1 and lim g( x) = L 2 . The goal is to show that lim ( f ( x) + g( x)) = L 1 + L 2 . So let
x→ a x→ a x→ a
ǫ > 0. Then ǫ/2 > 0, and so by definition there exist numbers δ1 > 0 and δ2 > 0 such that
0 < | x − a | < δ1 ⇒ | f ( x) − L 1 | < ǫ/2 , and

0 < | x − a | < δ2 ⇒ | g( x) − L 2 | < ǫ/2 .
Now let δ = min(δ1 , δ2 ). Then δ > 0 and
0 < | x − a| < δ ⇒ 0 < | x − a | < δ1 and 0 < | x − a | < δ2

⇒ | f ( x) − L 1 | < ǫ/2 and | g( x) − L 2 | < ǫ/2
Since | A + B | ≤ | A | + | B | for all real numbers A and B, then
| f ( x ) + g( x ) − ( L 1 + L 2 ) | = | ( f ( x ) − L 1 ) + ( g( x ) − L 2 ) | ≤ | f ( x ) − L 1 | + | g( x ) − L 2 |
and thus
0 < | x − a| < δ ⇒ | f ( x) + g( x) − (L 1 + L 2 ) | < ǫ/2 + ǫ/2 = ǫ
which by definition means that lim ( f ( x) + g( x)) = L 1 + L 2 . X
x→ a
The proofs of the other limit rules are similar.1 In general, using the formal defini-
tion will not be necessary for evaluating limits of specific functions—in many cases a
simple analysis of the function is all that is needed, often from its graph.
Example 3.8
Evaluate lim f ( x) for the following function: y
x→1
 y = f ( x)
x
 if x > 1
2
f ( x) = 2 if x = 1


1 if x < 1
1
Solution: From the graph of f ( x) in Figure 3.2.2, it is clear that as x
approaches 1 from the right (i.e. for x > 1) f ( x) approaches 1 along the x
line y = x, whereas as x approaches 1 from the left (i.e. for x < 1) f ( x) 0 1
approaches 1 along the horizontal line y = 1. Thus, lim f ( x) = 1 . Figure 3.2.2
x→1
Note that the limit did not depend on the value of f ( x) at x = 1.
As Example 3.8 shows, what matters for a limit is what happens to the value of f ( x)
as x gets near a, not at x = a itself. Figure 3.2.3(a) below shows how as x approaches
a, f ( x) approaches a number different from f (a). Figure 3.2.3(b) shows that x = a does
not even need to be in the domain of f ( x), i.e. f (a) does not have to be defined. So it
will not always be the case that lim f ( x) = f (a).
x→a
y y
(a, f (a)) y = f ( x) y = f ( x)
f (a)
L L
x x
a a
(a) lim f ( x) = L 6= f (a) (b) lim f ( x) = L, f ( x) not defined at x = a
x→ a x→ a
Figure 3.2.3 Excluding x = a from lim f ( x)

x→ a
In Example 3.8 the direction in which x approached the number 1 did not affect
the limit. But what if f ( x) had approached different values depending on how x ap-
proached 1? In that case the limit would not exist. The following definitions and
notation for one-sided limits will make situations like that simpler to state.
1 For example, see Section 2.2 in P ROTTER , M.H. AND C.B. M ORREY , A First Course in Real Analysis, New York:
Springer-Verlag, 1977.
Call L the right limit of a function f ( x) as x approaches a, written as
lim f ( x) = L ,
x→a+
if f ( x) approaches L as x approaches a for values of x larger than a.

Call L the left limit of a function f ( x) as x approaches a, written as
lim f ( x) = L ,
x→a−
if f ( x) approaches L as x approaches a for values of x smaller than a.
The following statement follows immediately from the above definitions:
The limit of a function exists if and only if both its right limit and left limit exist
and are equal:
lim f ( x) = L ⇔ lim f ( x) = L = lim f ( x)

x→a x→a− x→a+
Example 3.9
Evaluate lim f ( x), lim f ( x), and lim f ( x) for the following function: y
x→0− x→0+ x→0
(
x2 if x < 0 y = f ( x)
f ( x) = 2
2−x if x ≥ 0
Solution: From the graph of f ( x) in Figure 3.2.4, it is clear that as x 1

approaches 0 from the left (i.e. for x < 0) f ( x) approaches 0 along the
parabola y = x2 , whereas as x approaches 0 from the right (i.e. for
x > 0) f ( x) approaches 2 along the line y = 2 − x. Hence, lim f ( x) = 0 x
x→0− −1 0 1
and lim f ( x) = 2 . Thus, lim f ( x) does not exist since the left and
x→0+ x→0 Figure 3.2.4
right limits do not agree at x = 0.
Example 3.10
µ ¶
1 y
Evaluate lim sin .
x→0+ x y = sin(1/ x)
1
Solution: For x > 0 the function f ( x) = sin (1/ x) is defined, and its graph
is shown in Figure 3.2.5. As x approaches 0 from the right, sin (1/ x) will 0 x
be 1 for the numbers x = 2/π, 2/5π, 2/9π, 2/13π, ... (which approach 0),
and sin (1/ x) will be −1 for the numbers x = 2/3π, 2/7π, 2/11π, 2/15π, ... −1
(which also approach 0). So as x approaches 0µ from ¶ the right, sin (1/ x)
1 Figure 3.2.5
will oscillate between 1 and −1. Thus, lim sin does not exist .
x→0+ x
So far only finite limits have been considered, that is, L = lim f ( x) where L is a real
x→a
(i.e. finite) number. Define an infinite limit, with L = ∞ or −∞, as follows:
For a real number a, the limit of a function f ( x) equals infinity as x approaches

a, written as
lim f ( x) = ∞ ,
x→a
if f ( x) grows without bound as x approaches a, i.e. f ( x) can be made larger than

any positive number by picking x sufficiently close to a:
For any given number M > 0, there exists a number δ > 0, such that
f ( x) > M whenever 0 < | x − a | < δ .
For a real number a, the limit of a function f ( x) equals negative infinity as x

approaches a, written as
lim f ( x) = −∞ ,
x→a
if f ( x) grows negatively without bound as x approaches a, i.e. f ( x) can be made

smaller than any negative number by picking x sufficiently close to a:
For any given number M < 0, there exists a number δ > 0, such that
f ( x) < M whenever 0 < | x − a | < δ .
The above definitions can be modified accordingly for one-sided limits. If lim f ( x) =
x→a
∞ or lim f ( x) = −∞, then the line x = a is a vertical asymptote of f ( x), and f ( x)
x→a
approaches the line x = a asymptotically. The formal definitions are rarely needed.
Example 3.11
1 y
Evaluate lim .
x→0 x
Solution: For x 6= 0 the function f ( x) = 1x is defined, and its graph 1
is shown in Figure 3.2.6. As x approaches 0 from the right, 1/ x y= x
approaches ∞, that is,
1
lim = ∞. x
x→0+ x
0
As x approaches 0 from the left, 1/ x approaches −∞, that is,
1
lim = −∞ .
x→0− x
1 Figure 3.2.6
Since the right limit and the left limit are not equal, then lim does not exist .
x→0x
Note that the y-axis (i.e. the line x = 0) is a vertical asymptote for f ( x) = 1x .
Example 3.12
1 y
Evaluate lim .
x→0 x2
Solution: For x 6= 0 the function f ( x) = x12 is defined, and its graph 1
is shown in Figure 3.2.7. As x approaches 0 from either the right y= x2
or the left, 1/ x2 approaches ∞, that is,
1 1 x
lim = ∞ = lim 2 . 0
x→0+ x2 x→0− x
Figure 3.2.7
1
Since the right limit and the left limit both equal ∞, then lim = ∞.
x→0 x2
Note that the y-axis (i.e. the line x = 0) is a vertical asymptote for f ( x) = x12 .
In the limit lim f ( x) so far only real values of a have been considered. However, a
x→a
could be either ∞ or −∞:
For a real number L, the limit of a function f ( x) equals L as x approaches ∞,

written as
lim f ( x) = L ,
x→∞
if f ( x) can be made arbitrarily close to L for x sufficiently large and positive:

For any given number ǫ > 0, there exists a number N > 0, such that
| f ( x) − L | < ǫ whenever x>N .
For a real number L, the limit of a function f ( x) equals L as x approaches −∞,

written as
lim f ( x) = L ,
x→−∞
if f ( x) can be made arbitrarily close to L for x sufficiently small and negative:

For any given number ǫ > 0, there exists a number N < 0, such that
| f ( x) − L | < ǫ whenever x<N .
The above definitions can be modified accordingly for L replaced by either ∞ or −∞.
One way to interpret the statement lim f ( x) = L is: the long-term behavior of f ( x) is
x→∞
to approach a steady-state at L. If lim f ( x) = L or lim f ( x) = L, then the line y = L
x→∞ x→−∞
is a horizontal asymptote of f ( x), and f ( x) approaches the line y = L asymptotically.
Again, for most limits of specific functions, only the intuitive notions are needed.
Example 3.13
From Figures 3.2.6 and 3.2.7, it is clear that
1 1 1 1
lim = 0 = lim and lim = 0 = lim
x→∞ x x→−∞ x x→∞ x2 x→−∞ x2
1 1
Note that the x-axis (i.e. the line y = 0) is a horizontal asymptote for f ( x) = x and f ( x) = x2
.
Some limits are obvious—you can use them for calculating other limits:
( (
∞ for any real n > 0 ∞ for n = 2, 4, 6, 8, . . .
lim xn = lim xn =
x→∞ 0 for any real n < 0 x→−∞ −∞ for n = 1, 3, 5, 7, . . .
lim e x = ∞ lim e x = 0 lim e−x = 0 lim e−x = ∞

x→∞ x→−∞ x→∞ x→−∞
lim ln x = ∞ lim ln x = −∞
x→∞ x →0 +
A related notion is that of Big O notation (that is the capital letter O, not a zero):
Say that
f ( x) = O( g( x)) as x → ∞,
spoken as “ f is big O of g”, if there exist positive numbers M and x0 such that
| f ( x) | ≤ M | g( x) | for all x ≥ x0 .
For example, obviously 2 x3 = O( x3 ), by picking M = 2, with x0 any positive number.

In general, f ( x) = O( g( x)) means that f exhibits the same long-term behavior as g, up
to a constant multiple. You can think of g as the more basic “type” of function that
describes f , as far as long-term behavior.
Example 3.14
Show that 5 x4 − 2 = O( x4 ).
Solution: First, recall from algebra that | a + b | ≤ | a | + | b | for all real numbers a and b. Thus,
¯ 4 ¯ ¯ ¯ ¯ ¯
¯ 5 x − 2 ¯ ≤ ¯ 5 x4 ¯ + |−2 | = 5 ¯ x4 ¯ + 2
¯ ¯
for all x. So since ¯ x4 ¯ = x4 ≥ 1 for all x ≥ 1, then
¯ 4 ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯
¯ 5 x − 2 ¯ ≤ 5 ¯ x4 ¯ + 2 ≤ 5 ¯ x4 ¯ + 2 ¯ x4 ¯ ⇒ ¯ 5 x4 − 2 ¯ ≤ 7 ¯ x4 ¯ for all x ≥ 1,
which shows that 5 x4 − 2 = O( x4 ), with M = 7 and x0 = 1.

Some limits need algebraic manipulation before they can be evaluated.

Example 3.15
³p p ´
Evaluate lim x+1 −x .
x→∞
p p
Solution: Note that both x + 1 and x approach ∞ as x goes to ∞, resulting in a limit of the form
∞ − ∞. This is an example of an indeterminate form, which can equal anything (as will be discussed
shortly); it does not
p have to equal 0 (i.e. the ∞’s do not necessarily “cancel out”). The trick here is to use
p
the conjugate of x + 1 − x, so that
³p p p
p ´ ³p p ´ x+1 + x ( x + 1) − x 1
lim x + 1 − x = lim x+1 − x · p p = xlim p p = xlim p p = 0
x→∞ x→∞ x+1 + x →∞ x+1 + x →∞ x+1 + x
1
since the numerator is 1 and both terms in the sum in the denominator approach ∞ (i.e. ∞ = 0).
Some other indeterminate forms are ∞/∞, 0/0 and ∞· 0. How would you handle such
limits? One way is to use L’Hôpital’s Rule2 ; a simplified form is stated below:
L’Hôpital’s Rule: If f and g are differentiable functions and
f ( x) ±∞ 0
lim = or
x→a g ( x) ±∞ 0
then
f ( x) f ′ ( x)
lim = lim ′ .
x→a g ( x) x → a g ( x)
The number a can be real, ∞, or −∞.
Example 3.16
2x − 1
Evaluate lim .
x→∞ ex
Solution: This limit is of the form ∞/∞:
2x − 1 ∞
lim x
→
x→∞ e ∞
2
= lim by L’Hôpital’s Rule
x→∞ ex
= 0
since the numerator is 2 and e → ∞ as x → ∞.
x
Note that one way of interpreting the limit being 0 is that e x grows much faster than 2 x − 1. In fact,
using L’Hôpital’s Rule it can be shown that e x grows much faster than any polynomial, i.e. exponential
growth outstrips polynomial growth.
2 For a proof, see pp.89-91 in P ROTTER , M.H. AND C.B. M ORREY , A First Course in Real Analysis, New York:
Example 3.17
x
Evaluate lim .
x→∞ ln x
x ∞
lim →
x→∞ ln x ∞
1
x→∞ 1
x
= lim x = ∞
x→∞
Note that one way of interpreting the limit being ∞ is that x grows much faster than ln x. In fact,
using L’Hôpital’s Rule it can be shown that any polynomial grows much faster than ln x, i.e. polynomial
growth outstrips logarithmic growth.
Example 3.18
Evaluate lim xe−2 x .
x→∞
Solution: This limit is of the form ∞ · 0, which can be converted to ∞/∞:
x ∞
lim xe−2 x = lim →
x→∞ x→∞ e2 x ∞
1
x→∞ 2 e2 x
= 0
Note that the limit is another consequence of exponential growth outstripping polynomial growth.
Example 3.19
2 x2 − 7 x − 5
Evaluate lim .
x→∞ 3 x2 + 2 x − 1
2 x2 − 7 x − 5 ∞
lim →
x→∞ 3 x2 + 2 x − 1 ∞
4x − 7
x→∞ 6x + 2
∞
→ , so use L’Hôpital’s Rule again
∞
4 2
= =
6 3
Note that the limit ended up being the ratio of the leading coefficients of the polynomials in the numer-
ator and denominator of the original limit. Note also that the lower-order terms (degree less than 2)
ended up not mattering. In general you can always discard the lower-order terms when taking the limit
of a ratio of polynomials (i.e. a rational function).
Example 3.20
1 − cos x
Evaluate lim .
x→0 x
Solution: This limit is of the form 0/0:
1 − cos x 0
lim →
x→0 x 0
sin x
x→0 1
sin 0
= = 0
1
f ( x)
There is an intuitive justification for L’Hôpital’s Rule: since the limit lim x→a g( x)
uses a ratio to compare how f changes relative to g as x approaches a, then it is
really the rates of change of f and g—namely f ′ and g′ , respectively—that are being
compared in that ratio, which is what L’Hôpital’s Rule says.
The following result provides another way to calculate certain limits:
Squeeze Theorem: Suppose that for some functions f , g and h there is a number
x0 ≥ 0 such that
g( x) ≤ f ( x) ≤ h( x) for all x > x0
and that lim g( x) = lim h( x) = L. Then lim f ( x) = L.
x→∞ x→∞ x→∞
Similarly, if g( x) ≤ f ( x) ≤ h( x) for all x 6= a in some interval I containing a, and if

lim g( x) = lim h( x) = L, then lim f ( x) = L.
x→a x→a x→a
Intuitively, the Squeeze Theorem says that if one function is “squeezed” between
two functions approaching the same limit, then the function in the middle must also
approach that limit. The theorem also applies to one-sided limits ( x → a+ or x → a−).
Example 3.21
sin x
Evaluate lim .
x→∞ x
Solution: Since −1 ≤ sin x ≤ 1 for all x, then dividing all parts of those inequalities by x > 0 yields
1 sin x 1 sin x
− ≤ ≤ for all x > 0 ⇒ lim = 0
x x x x→∞ x
−1 1
by the Squeeze Theorem, since lim = 0 = lim .
x→∞ x x →∞ x
Exercises
A
For Exercises 1-18 evaluate the given limit.
x2 + 3 x − 10 x2 + 3 x − 10 x2 + 3 x − 10 x3 + 3 x − 10
1. lim 2. lim 3. lim 4. lim
x→2 x2 − x − 2 x→∞ 2 x2 − x − 2 x→∞ 2 x3 − x − 2 x→∞ 2 x2 − x − 2
cos x x2 ln x
5. lim 6. lim 7. lim x2 e x 8. lim 1/ x
x→π/2 x − π/2 x→∞ e x x →−∞ x → 0 + e µ ¶
tan x − x sin 3 x cos x 1
9. lim 10. lim 11. lim 12. lim x sin
x→0 x − sin x x→0 sin 4 x x →∞ x x → 0 x
2 µ ¶
ln (1 − x) − sin x 1 1
13. lim 14. lim − 15. lim (sec x − tan x)
x→0 1 − cos2 x x→1 ln x x−1 x→π/2
e −1/ x ³p ´ cot x
16. lim 17. lim x2 + 4 − x 18. lim
x→0+ x x→∞ x→0 csc x
B
19. The famous “twin paradox,” a result of Einstein’s special theory of relativity, says that if one of a
pair of twins leaves the earth in a rocket traveling at a high speed, then he will be younger than his
twin upon returning to earth.3 This is due to the phenomenon of time dilation, which says that a
clock moving with a speed v relative to a clock at rest in some inertial reference frame counts time
slower relative to the clock at rest, by a factor of
1
γ= p ,
1 − β2
called the Lorentz factor, where β = vc is the fraction of the speed of light c at which the clock is
moving ( c ≈ 2.998 × 108 m/sec). Notice that 0 ≤ β < 1 (why?). For example, a clock moving at half
the speed of light, so that β = 0.5, would have γ = 1.1547, meaning that the clock runs about 15.47%
slower than the clock on earth.
(a) Evaluate lim γ . What is the physical interpretation of this limit?
β→1−
(b) Suppose an astronaut and his twin just turned 30 years old when the astronaut leaves earth on
a high-speed journey through space. Upon returning to earth the astronaut is 35 and his twin
is 70. At roughly what fraction of the speed of light must the astronaut have been traveling?
p( x )
20. Show that lim = 0 for all polynomials p( x) of degree n ≥ 1 with a positive leading coefficient.
x→∞ ex
p( x )
21. Show that lim = ∞ for all polynomials p( x) of degree n ≥ 1 with a positive leading coefficient.
x→∞ ln x
2 x2 + 1
22. Show that 5 x3 + 6 x2 − 4 x + 3 = O( x3 ). 23. Show that = O( x). (Hint: Consider x ≥ 1)
x+1
24. Call h( x) an infinitesimal function as x → a if lim h( x) = 0. That is, an infinitesimal function
x→ a
approaches zero near some point. Prove the following result, where a and L are real numbers:
lim f ( x) = L ⇔ f ( x) = L + h( x) for all x, where h( x) is an infinitesimal function as x → a

x→ a
3 See p.154-159 in F RENCH , A.P., Special Relativity, Surrey, U.K.: Thomas Nelson & Sons Ltd., 1968.
Continuity • Section 3.3 73
3.3 Continuity
Recall from the previous section that a limit lim x→a f ( x) can exist without being equal
to f (a), or with f (a) not even being defined. Many functions encountered in applica-
tions, however, will meet those conditions, and they have a special name:
A function f is continuous at x = a if
lim f ( x) = f (a) . (3.4)

x→a
A function is continuous on an interval I if it is continuous at every point in the

interval. For a closed interval I = [a, b], a function f is continuous on I if it is
continuous on the open interval (a, b) and if limx→a+ f ( x) = f (a) (i.e. f is right
continuous at x = a) and lim x→b− f ( x) = f ( b) (i.e. f is left continuous at x = b).
A function is discontinuous at a point if it is not continuous there. A continuous
function is one that is continuous over its entire domain.
Equation (3.4) in the above definition implies that f (a) is defined, i.e. x = a is in the
domain of f . Figure 3.3.1 below shows some examples of continuity and discontinuity:
y y = f ( x)
x
x1 x2 x3 x4
Figure 3.3.1 Continuous at x1 , discontinuous at x2 , x3 and x4
In the above figure, f is not continuous at x = x2 because lim x→ x2 f ( x) 6= f ( x2 ); f is not

continuous at x = x3 because lim x→ x3 f ( x) does not exist (the right and left limits do
not agree— f is said to have a jump discontinuity at x = x3 ); and f is not continuous
at x = x4 because f ( x4 ) is not defined. However, f is continuous at x = x1 .
A function is continuous if its graph is one unbroken piece over its entire domain.
Polynomials, rational functions, trigonometric functions, exponential functions, and
logarithmic functions are all continuous on their domains. For example, tan x is con-
tinuous over its domain, which is broken into disjoint intervals (−π/2, π/2), (π/2, 3π/2),
(3π/2, 5π/2), and so forth; the graph is unbroken on each of those intervals. However,
tan x is not continuous over all of R, since the function is not defined at all points in R.
In the language of infinitesimals, a function f is continuous at x = a if f (a + dx) − f (a)
is an infinitesimal for any infinitesimal dx. This definition is rarely used.
Physical examples of continuous functions are position, speed, velocity, acceleration,
temperature, and pressure. Some discontinuous functions do arise in applications.
Example 3.22
The floor function ⌊ x⌋ is defined as
⌊ x⌋ = the largest integer less than or equal to x .
In other words, ⌊ x⌋ rounds a non-integer down to the previous integer, and integers stay the same. For
example, ⌊0.1⌋ = 0, ⌊0.9⌋ = 0, ⌊0⌋ = 0, and ⌊−1.3⌋ = −2. The graph of ⌊ x⌋ is shown in Figure 3.3.2(a).
y y
2 2
1 1
x x
−2 −1 0 1 2 3 −3 −2 −1 0 1 2
−1 −1
y = ⌊ x⌋ y = ⌈ x⌉
−2 −2
(a) Floor function ⌊ x⌋ (b) Ceiling function ⌈ x⌉
Figure 3.3.2 Floor and ceiling functions
Similarly, the ceiling function ⌈ x⌉ is defined as
⌈ x⌉ = the smallest integer greater than or equal to x .
In other words, ⌈ x⌉ rounds a non-integer up to the next integer, and integers stay the same. For example,
⌈0.1⌉ = 1, ⌈0.9⌉ = 1, ⌈1⌉ = 1, and ⌈−1.3⌉ = −1. The graph of ⌈ x⌉ is shown in Figure 3.3.2(b).
Clearly both ⌊ x⌋ and ⌈ x⌉ have jump discontinuities at the integers, but both are continuous at all
non-integer values of x. Both functions are also examples of step functions, due to the staircase
appearance of their graphs. Step functions are useful in situations when you want to model a quantity
that takes only a discrete set of values. For example, in a car with a 4-gear transmission, f ( x) could
be the gear the transmission has shifted to while the car travels at speed x. Up to a certain speed the
car remains in first gear ( f = 1) and then shifts to second gear ( f = 2) after attaining that speed, then
it remains in second gear until reaching another speed, upon which the car then shifts to third gear
( f = 3), and so on. In general, discrete changes in state are often modeled with step functions.
Example 3.23
For an extreme case of discontinuity, consider the function
(
0 if x is rational
f ( x) =
1 if x is irrational
This function is discontinuous at every value of x in R, since within any positive distance δ of a real
number x—no matter how small δ is—there will be an infinite number of both rational and irrational
numbers. This is a property of R. So the value of f will keep jumping between 0 and 1 no matter how
close you get to x. In other words, for any number a in R, f (a) exists but it will never equal lim x→a f ( x)
because that limit will not exist.
By the various rules for limits, it is straightforward to show that sums, differences,
constant multiples, products and quotients of continuous functions are continuous.
Likewise a continuous function of a continuous function (i.e. a composition of continu-
ous functions) is also continuous. In addition, a continuous function of a finite limit of
a function can be passed inside the limit:
If f is a continuous function and lim g( x) exists and is finite, then:

x→a
³ ´
f lim g( x) = lim f ( g( x)) (3.5)
x→a x→a
The same relation holds for one-sided limits.
The above result is useful in evaluating the indeterminate forms 00 , ∞0 , and 1∞ . The
idea is to take the natural logarithm of the limit by passing the continuous function
ln x inside the limit and evaluate the resulting limit.
Example 3.24
Evaluate lim x x .
x→0+
Solution: This limit is of the form 00 , so let y = lim x x and then take the natural logarithm of y:
x→0+
µ ¶
ln y = ln lim x x
x→0+
= lim ln x x (pass the natural logarithm function inside the limit)

x→0+
= lim x ln x → 0 · (−∞)
x→0+
ln x −∞
= lim →
x→0+ 1/ x ∞
1/ x
x→0+ −1/ x2
ln y = lim (− x) = 0
x→0+
Thus, lim x x = y = e0 = 1.
x→0+
There is an important relationship between differentiability and continuity:
Every differentiable function is continuous.
f ( x)− f (a)
Proof: If a function f is differentiable at x = a then f ′ (a) = lim x −a exists, so
x→a
x−a f ( x) − f ( a)
lim ( f ( x) − f (a)) = lim ( f ( x) − f (a)) ·
= lim · lim ( x − a) = f ′ (a) · 0 = 0
x→a x→a x−a x→a x−a x→a
which means that lim f ( x) = f (a) , i.e. f is continuous at x = a . X

x→a
Note that the converse is not true. For example, the absolute y
value function f ( x) = | x | is continuous everywhere—its graph
is unbroken, as shown in the picture on the right—but recall
from Example 1.3 in Section 1.2 that it is not differentiable at y = |x|
x = 0. Continuous curves can have sharp edges and cusps, but 0
x
differentiable curves cannot.
Two other important theorems4 about continuous functions are:
Extreme Value Theorem: If f is a continuous function on a closed interval [a, b]

then f attains both a maximum value and a minimum value on that interval.
Intermediate Value Theorem: If f is a continuous function on a closed interval

[a, b] then f attains every value between f (a) and f ( b).
y y
2
y = f ( x) k
1 y = f ( x)
x x
a b c d a x0 b c d
Figure 3.3.3 Extreme Value Theorem Figure 3.3.4 Intermediate Value Theorem
Figure 3.3.3 shows why a closed interval is required for the Extreme Value Theorem,
as f attains neither a maximum nor minimum on the open interval ( c, d ). The Inter-
mediate Value Theorem says that continuous functions cannot “skip over” intermedi-
ate values between two other function values. In Figure 3.3.4 the function f skips the
value k between f ( c) = 1 and f ( d ) = 2 because f is not continuous over all of [ c, d ]. On
[a, b] the value k is attained by f at x = x0 , i.e. f ( x0 ) = k, since f is continuous on [a, b].
4 The full proofs require some advanced results. See pp.97-98 and p.558 in T AYLOR , A.E. AND W.R. M ANN,
Advanced Calculus, 2nd ed., New York: John Wiley & Sons, Inc., 1972.
Example 3.25
Show that there is a solution to the equation cos x = x.
Solution: Let f ( x) = cos x − x. Since f is continuous for all x, in particular it is continuous on [0, 1]. So
since f (0) = 1 > 0 and f (1) = −0.459698 < 0, then by the Intermediate Value Theorem there is a number
c in the open interval (0, 1) such that f ( c) = 0, since 0 is between the values f (0) and f (1). Hence,
cos c − c = 0, which means that cos c = c. That is, x = c is a solution of cos x = x.
Note in the above example that the Intermediate Value Theorem does not tell you
how to find the solution, just that the solution exists. To find the solution you can use
the bisection method: divide the interval [0, 1] in half and apply the Intermediate
Value Theorem to each half-interval to determine which one contains the solution;
repeat this procedure on that half-interval, resulting in a smaller interval containing
the solution, then repeat the procedure over and over, until you eventually obtain
an interval so small that the midpoint of that interval can be taken as the solution.
Listing 3.1 below shows one way of implementing the bisection method for Example
3.25 to find the root of f ( x) = cos x − x, using the Python programming language.
Listing 3.1 Bisection method in Python
1 import math
2
3 def f(x):
4 return math.cos(x) - x
5
6 def bisect(a, b):
7 midpt = (a+b)/2.0
8 tol = 1e-15
9 if b - a > tol:
10 val = f(midpt)
11 if val*f(a) < 0:
12 bisect(a, midpt)
13 elif val*f(b) < 0:
14 bisect(midpt, b)
15 else:
16 print("Root = %.13f" % (midpt))
17 else:
18 print("Root = %.13f" % (midpt))
19
20 bisect(0, 1)
Line 8 sets the tolerance to 10−15 : the program terminates upon reaching an interval
whose length is smaller than that. The output is shown below:
Root = 0.7390851332152
This is the number obtained by taking the cosine of a number (in radians) repeatedly.
Exercises
A
For Exercises 1-18, indicate whether the given function f ( x) is continuous or discontinuous at the given
value x = a by comparing f (a) with lim x→a f ( x).
1. f ( x) = | x |; at x = 0 2. f ( x) = | x − 1 |; at x = 0 3. f ( x) = ⌊ x⌋; at x = 0
4. f ( x) = ⌊ x⌋; at x = 0.3 5. f ( x) = ⌈ x⌉; at x = 0 6. f ( x) = ⌈ x⌉; at x = 0.5
7. f ( x) = x − ⌊ x⌋; at x = 0 8. f ( x) = x − ⌊ x⌋; at x = 1.1 9. f ( x) = x − | x |; at x = 0

( ( (
0 if x ≤ 0, 0 if x ≤ 0, x2 if x ≤ 0,
10. f ( x) = 11. f ( x) = 12. f ( x) =
1 if x > 0; 1 if x > 0; 1 if x > 0;
at x = 0 at x = 1 at x = 0
( ( (
x + 1 if x ≤ 0, sin ( x2 ) if x 6= 0, sin (1/ x) if x 6= 0,
13. f ( x) = 14. f ( x) = 15. f ( x) =
1 if x > 0; 0 if x = 0; 0 if x = 0;
at x = 1 at x = 0 at x = 0
( ( (
0 if x is rational, 0 if x is rational, 0 if x is rational,
16. f ( x) = 17. f ( x) = 18. f ( x) =
1 if x is irrational; x if x is irrational; x if x is irrational;
p
at x = 3 at x = 0 at x = 1
2
19. Evaluate lim x x . 20. Evaluate lim x1/ x . 21. Evaluate lim (1 − x)1/ x .
x→0+ x→∞ x→0
2
x +x−2
22. If f ( x) = for x 6= 1, how should f (1) be defined so that f ( x) is continuous at x = 1 ?
x−1
23. If f ( x) = 1/ x for x 6= 0, is there a way to define f (0) so that f ( x) is continuous for all x ?
B
24. Can a function that is not continuous over a closed interval attain a maximum value and a mini-
mum value in that interval? If so, then give an example; if not then explain why.
25. Show that there is a number x such that x5 − x = 3.
26. Prove that f ( x) = x8 + 3 x4 − 1 has at least two distinct real roots.
27. Suppose that a function f is continuous on the interval [0, 3], f has no roots in [0, 3], and f (1) = 1.
Prove that f ( x) > 0 for all x in [0, 3].
28. Show that an object whose average speed is vavg over the time interval a ≤ t ≤ b will move with
speed vavg at some time t in [a, b].
29. Let f ( x) = 1/( x − 1). Then f (0) = −1 < 0 and f (2) = 1 > 0. Can you conclude by the Intermediate
Value Theorem that f ( x) must be 0 for some x in [0, 2] ? Explain.
30. Show that if f ′ and f ′′ exist and are continuous at x then
f ( x) − 2 f ( x − h) + f ( x − 2h)
f ′′ ( x) = lim .
h→0 h2
31. Show that if f ′ , f ′′ and f ′′′ exist and are continuous at x then
f ( x) − 3 f ( x − h) + 3 f ( x − 2h) − f ( x − 3h)
f ′′′ ( x) = lim .
h→0 h3
Implicit Differentiation • Section 3.4 79
3.4 Implicit Differentiation

A function y = f ( x) is usually given by an explicit formula, such as y = x2 . It is then
dy
straightforward to find dx using the differentiation rules you have learned so far. But
suppose instead that you were given merely an equation involving x and y, such as
x3 y2 esin ( x y) = x2 + x y + y3 .
The set of points ( x, y) satisfying this equation describes some sort of curve in the x y-
plane, but it might not be possible to solve for y in terms of x—that is, there might not
be an explicit formula for y as a function of the variable x. So in this case does the
dy
derivative dx even have any meaning, and if so then how would you find it?
dy
It turns out that dx does make sense in such a case, because an equation involving
x and y such as the one above implicitly defines y in terms of x in the following sense:
as x varies so does y. Hence it should be possible to find the rate of change of y with
dy d
respect to the variable x (i.e. dx ). To do so, take dx of both sides of the equation, then
assume that y really is a function of x so that you can use the Chain Rule to solve for
dy
dx
. The example below illustrates this procedure, called implicit differentiation.
Example 3.26
given the equation x3 + 3 x + 2 = y2 .
dy
Find dx
10
Solution: The above equation implicitly defines an elliptic curve, and its
5
graph is shown on the right. This curve is not a function y = f ( x), since it
dy d
violates the vertical line test, but y still varies with x. To find dx take dx of
dy 0 y
both sides of the equation then solve for dx :
−5
d 3 d 2
( x + 3 x + 2) = (y )
dx dx −10
dy −4 −2 0 2 4 6
3 x2 + 3 = 2 y · by the Chain Rule, so x
dx
dy 3 x2 + 3
=
dx 2y
dy
At first this might seem unsatisfying—or confusing—since dx is given in terms of both x and y.
However, the derivative can still be evaluated at specific points ( x, y) on the p curve, i.e. any ( x, y) satis-
fying the original equation. For example, p
it is easy to check that ( x, y ) = (1 , 6) satisfies the equation
3 2 dy p 3(1)p2 +3 6 dy
x + 3 x + 2 = y , so dx (1, 6) = = 2 . Note that dx is not defined when y = 0.
2 6
Notice that taking the psquare root of both sides of the original equation does not result in an explicit
formula for y, since y = ± x3 + 3 x + 2 defines two functions, not just one. The beauty of implicit differen-
2
tiation is that the derivative dx = 3 x2 y+3 calculated above gives you a single expression for the derivative
dy
of both those functions.

An algebraic curve is defined as the set of all points ( x, y) satisfying a polynomial

equation in the variables x and y, such as x2 − 3 x y4 + 1 = x5 − y2 . An elliptic curve
is a special case of an algebraic curve, where the polynomial has the specific form
x3 + ax + b = y2 , such as the equation x3 + 3 x + 2 = y2 from Example 3.26. Elliptic curves
have certain properties that have found applications in cryptography.5
Example 3.27
given the equation x + y = x3 + y3 .
dy
Find dx
3
Solution: The above equation implicitly defines an algebraic curve and its 2
dy
graph is shown on the right. To find dx take dx
d
of both sides of the equation 1
dy
then solve for dx : 0 y
d d 3 −1
( x + y) = ( x + y3 )
dx dx −2
dy dy −3
1 + = 3 x 2 + 3 y2 · by the Chain Rule, so
dx dx −3 −2 −1 0 1 2 3
x
dy 3 x2 − 1
=
dx 1 − 3 y2
Notice that the curve consists of an oval shape (an ellipse, actually) with a line through it. In fact,
that line is y = − x, as can be verified by replacing each instance of y in the equation x + y = x3 + y3 by
dy
− x (resulting in the equation 0 = 0). You might be wondering how dx is defined at the points where that
line intersects the ellipse: is it the slope of the line y = − x (i.e. −1), or is it the slope of the tangent line
to the ellipse at those points (which would not equal −1)? This is discussed in the exercises.
The graph was created with the free open-source graphing program Gnuplot6 using the following
Gnuplot commands (which give an idea of how to plot implicit functions in general):
set size square

set view 0,0
set isosamples 500,500
set contour base
set cntrparam levels discrete 0
unset surface
set grid
unset key
unset ztics
set xlabel ’x’
set ylabel ’y’
f(x,y) = x + y - x**3 - y**3
splot [-3:3][-3:3] f(x,y) lw 3
5 For example, see Section 12.2 in B UCHMANN, J.A., Introduction to Cryptography, New York: Springer-Verlag,
2001.
6 See the documentation at http://www.gnuplot.info/documentation.html
Implicit Differentiation • Section 3.4 81
Example 3.28
Find the tangent line to the curve x2 + y2 = 1 at the point (4/5, 3/5). y
Solution: This curve is the unit circle, shown in the picture on the right. x 2 + y2 = 1
dy
First use implicit differentiation to find dx :
(4/5, 3/5)
d 2 d dy
( x + y2 ) = (1) ⇒ 2x + 2 y · = 0 by the Chain Rule 0 1
x
dx dx dx
dy x
⇒ = −
dx y
dy 4/5
The slope m of the tangent line to the curve at (4/5, 3/5) is then m = dx (4/5, 3/5) = − 3/5 = −4/3. Thus, the
¡ ¢
equation of the tangent line is y − 53 = − 43 x − 54 .
Exercises
A
dy
For Exercises 1-9, use implicit differentiation to find dx .
1. x3 y − 4 x y2 = y + x2 2. x y = ( x + y)3 3. ( x + y)3 = ( x − y + 1)2

x+ y
4. x2/3 + y2/3 = a2/3 5. ( x2 − y2 )2 = 2 x2 + y2 6. = x 2 + y2
x− y
7. cos ( x y) = sin ( x2 y2 ) 8. x3 − x = y2 9. x3 y2 esin ( x y) = x2 + x y + y3
10. In Example 3.28 is it possible to solve the equation x2 + y2 = 1 explicitly for y in terms of x? Explain.
11. In Example 3.28 what happens to the tangent line at the point (1, 0)? Why does this make sense
geometrically?
12. Find the equation of the tangent line to the curve x3 + 3 x2 y + y3 = 8 at the point (2, 0).
B
d2 y
13. Find dx2
for the curve x2 + y2 = 1. You may use the results from Example 3.28.
14. Show that at every point ( x0 , y0 ) on the curve y2 = 4ax, the equation of the tangent line to the curve
is yy0 = 2a( x + x0 ).
x 2 y2
15. Show that at every point ( x0 , y0 ) on the ellipse 2 + 2 = 1, the equation of the tangent line to the
a b
xx0 yy0
ellipse is 2 + 2 = 1.
a b
x 2 y2
16. Show that at every point ( x0 , y0 ) on the hyperbola 2 − 2 = 1, the equation of the tangent line to
a b
xx0 yy0
the hyperbola is 2 − 2 = 1.
a b
dy
17. Show that dx is not defined at the points of intersection of the line and ellipse described by the
curve x + y = x3 + y3 from Example 3.27. (Hint: Factor the equation x + y = x3 + y3 .)
18. Show that the points P = (2, 4) and Q = (−31/64, −337/512) are on the elliptic curve x3 + 3 x + 2 = y2
from Example 3.26, and that the tangent line to the curve at P also goes through Q .
3.5 Related Rates

If several quantities are related by an equation, then differentiating both sides of that
equation with respect to a variable (usually t, representing time) produces a relation
between the rates of change of those quantities. The known rates of change are then
used in that relation to determine an unknown related rate.
Example 3.29
Suppose that water is being pumped into a rectangular pool at a
rate of 60,000 cubic feet per minute. If the pool is 300 ft long, 100
10
ft wide, and 10 ft deep, how fast is the height of the water inside
the pool changing?
Solution: Let V be the volume of the water in the pool. Since the
volume of a rectangular solid is the product of the length, width, h 300
and height of the solid, then
100
V = (300)(100)h = 30000h ft3
where h is the height of the water, as in the picture on the right. Both V and h are functions of time t
3
(measured in minutes), and dV dh
dt = 60000 ft /min was given. The goal is to find dt . Since
dV d dh
= (30000h) = 30000
dt dt dt
then
dh 1 dV 1
= = · 60000 = 2 ft/min.
dt 30000 dt 30000
Example 3.30
Suppose that the angle of inclination from the top of a 100 ft pole to ✸
the sun is decreasing at a rate of 0.05 radians per minute. How fast
is the length of the pole’s shadow on the ground increasing when the
angle of inclination is π/6 radians? You may assume that the pole is θ
perpendicular to the ground.
Solution: Let θ be the angle of inclination and let x be the length of the 100
shadow, as in the picture on the right. Both θ and x are functions of θ
time t (measured in minutes), and ddtθ = −0.05 rad/min was given (the x
derivative is negative since θ is decreasing). The goal is to find dx
dt when
¯
¯
dx ¯
θ = π/6, denoted by dt ¯ (the vertical bar means “evaluated at” the value of the subscript to the right
θ =π/6
of the bar). Since
dx dθ
x = 100 cot θ ⇒ = −100 csc2 θ · = −100 csc2 θ · (−0.05) = 5 csc2 θ
dt dt
then ¯
dx ¯¯
= 5 csc2 (π/6) = 5 (2)2 = 20 ft/min.
dt ¯θ =π/6
Related Rates • Section 3.5 83
Example 3.31
The radius of a right circular cylinder is decreasing at the rate of 3 cm/min, while the height is increas-
ing at the rate of 2 cm/min. Find the rate of change of the volume of the cylinder when the radius is 8
cm and the height is 6 cm.
Solution: Let r, h, and V be the radius, height, and volume, respectively, of the cylinder. Then V = π r2 h.
Since dr dh
dt = −3 cm/min and dt = 2 cm/min, then by the Product Rule:
µ ¶ ¯
dV d dr dh dV ¯¯ cm3
= (π r2 h) = 2π r · h + π r2 · ⇒ = 2π (8) (−3) (6) + π (82 ) (2) = −160π
dt dt dt dt dt ¯ r = 8 min
h=6
Exercises
A
1. A stone is dropped into still water. If the radius of the circular outer ripple increases at the rate of 4
ft/s, how fast is the area of the circle of disturbed water increasing when the radius is 10 ft?
2. The radius of a sphere decreases at a rate of 3 mm/hr. Determine how fast the volume and surface
area of the sphere are changing when the radius is 5 mm.
3. A kite 80 ft above level ground moves horizontally at a rate of 4 ft/s away from the person flying it.
How fast is the string being released at the instant when 100 ft of string have been released?
4. A 10-ft ladder is leaning against a wall on level ground. If the bottom of the ladder is dragged away
from the wall at the rate of 5 ft/s, how fast will the top of the ladder descend at the instant when it
is 8 ft from the ground?
5. A person 6 ft tall is walking at a rate of 6 ft/s away from a light which is 15 ft above the ground. At
what rate is the end of the person’s shadow moving along the ground away from the light?
6. An object moves along the curve y = x3 in the x y-plane. At what points on the curve are the x and y
coordinates of the object changing at the same rate?
7. The radius of a right circular cone is decreasing at the rate of 4 cm/min, while the height is increasing
at the rate of 3 cm/min. Find the rate of change of the volume of the cone when the radius is 6 cm
and the height is 7 cm.
8. Two boats leave the same dock at the same time, one goes north at 25 mph and the other goes east
at 30 mph. How fast is the distance between the boats changing when they are 100 miles apart?
9. Repeat Exercise 8 with the angle between the boats being 110◦ .
B
10. An angle θ changes with time. For what values of θ do sin θ and tan θ change at the same rate?
11. Repeat Example 3.30 but with the ground making a 100◦ angle with the pole to the left of the pole.
C
12. An upright cylindrical tank full of water is tipped over at a constant angular speed. Assume that
the height of the tank is at least twice its radius. Show that at the instant the tank has been tipped
45◦ , water is leaving the tank twice as fast as it did at the instant the tank was first tipped. (Hint:
Think of how the water looks inside the tank as it is being tipped.)
3.6 Differentials
An ideal gas satisfies the equation PV = RT , where R is a constant and P , V , and T
are the pressure, volume per mole, and temperature, respectively, of the gas. It will be
proved that
dP dV dT
+ = . (3.6)
P V T
Recall that dP , dV , and dT represent infinitesimal changes in the quantities P , V ,
and T , respectively. Notice that none of the quotients in Equation (3.6) have an in-
finitesimal in the denominator. For example, dP is divided not by dx or dt, as it would
be in a derivative such as dPdx or dt . Instead it is divided by P , which is not an in-
dP
finitesimal. So Equation (3.6) is an equation that relates infinitesimals themselves,

i.e. infinitesimal changes, not infinitesimal rates of change. This is, in fact, how many
physical laws are stated, for reasons that will be discussed shortly.
Though infinitesimals have been used throughout this text, many calculus text-
books7 do not even mention them, instead preferring to call them differentials.8 For
compatibility, the definition is given here:
For a differentiable function f ( x), the differential of f ( x) is
df = f ′ ( x) dx (3.7)
where dx is an infinitesimal change in x.
Note that this is identical to Equation (1.9) in Section 1.3.

Example 3.32
Find the differential df of f ( x) = x3 .
Solution: By definition,
df = f ′ ( x) dx = 3 x2 dx
Equivalently, this can be written as
d ( x3 ) = 3 x2 dx ,
which is often the way it would appear in textbooks in the sciences.
All the rules for derivatives (e.g. sum rule, product rule) apply to differentials, and
can be proved simply by multiplying the corresponding derivative rule by dx on both
sides of the equation:
7 More accurately, many current calculus textbooks never mention them. Calculus texts up through the 1930s or
so not only mentioned infinitesimals but used them extensively, even to the point of the texts themselves having
titles such as Introduction to Infinitesimal Calculus.
8 Though often in an unclear and sometimes confusing and misleading manner, as will be seen later in this section.
Differentials • Section 3.6 85
Let f and g be differentiable functions, and let c be a constant. Then:

(a) d ( c) = 0
(b) d ( c f ) = c df (Constant Multiple Rule)
(c) d ( f + g) = df + dg (Sum Rule)
(d) d ( f − g) = df − dg (Difference Rule)
(e) d ( f g) = f dg + g df (Product Rule)

µ ¶
f g df − f dg
(f) d = (Quotient Rule)
g g2
(g) d ( f n ) = n f n−1 df (Power Rule)

df
(h) d ( f ( g)) = dg (Chain Rule)
dg
For example, to prove (e), multiply both sides of the usual Product Rule by dx so that
µ ¶
d ( f g) dg df dg df
= f + g ⇒ d ( f g) = ✚ ✚ f
dx + g
dx dx dx ✚✚
dx ✚✚
dx
⇒ d ( f g) = f dg + g df X
since the dx terms all cancel. The proofs of the other rules are similar.
The differential version of the ideal gas law in Equation (3.6)
dP dV dT
+ =
P V T
can now be proved by taking the differential of both sides of the equation PV = RT :
d (PV ) = d (RT ) = R · dT by the Constant Multiple Rule

PV PV
V dP + P dV = dT by the Product Rule and since R =
T T
V dP P dV dT
+ = after dividing both sides by PV
PV PV T
dP dV dT
+ = X
P V T
Notice that dP
P , V and T represent the relative infinitesimal changes in P , V , and T ,
dV dT
respectively. The differential formulation is useful for finding one relative infinitesi-
mal change when the other two are known.
Example 3.33
Suppose that M is the total mass of a rocket and its unburnt fuel at any time t (so M is a function of
t). Over an infinitesimal time dt a mass dm of fuel is burnt and the gas byproducts are expelled out the
rear of the rocket at a velocity vE relative to the rocket. Using the law of conservation of momentum
over the interval dt, show that
vE dm = M dv
where m and v are the mass of burnt fuel and the velocity of the rocket, respectively, at the beginning
of the time dt.
Solution: Momentum is defined as mass times velocity. The momentum of the rocket at the beginning
of the time dt is thus Mv. At the end of the time dt, the momentum of the rocket consists of two parts,
namely the momentum of the rocket and its remaining unburnt fuel, which is
((mass before dt) − (increase in burnt fuel)) × ((velocity before dt) + (increase in velocity)) (3.8)
( M − dm)(v + dv) (3.9)
and the momentum of the fuel that was burnt and expelled out the rear, which is
(v − vE ) dm .
So by conservation of momentum,
Mv = ( M − dm)(v + dv) + (v − vE ) dm
Mv = Mv − v dm + M dv − (dm)(dv) + v dm − vE dm, so
vE dm = M dv − (dm)(dv) = M dv
since (dm)(dv) = (m′ ( t) dt)(v′ ( t) dt) = m′ ( t)v′ ( t)(dt)2 = m′ ( t)v′ ( t) · 0 = 0.

Dividing both sides of vE dm = M dv by dt yields the equation
M v̇ = ṁ vE
using the dot notation—mentioned in Section 1.3—for the derivative with respect to the time variable
t, which is still popular with physicists. Since v̇ is just acceleration a, this formulation is the classic
equation for the acceleration of a rocket.9
Letting f be the natural logarithm function and letting g = u in the differential

version of the Chain Rule yields the following useful result:
du
d (ln u) =
u
This is often used in a differential version of the technique of logarithmic differentia-

tion discussed in Section 2.3.
9 For other formulations see Chapter 1 in R OSSER , J.B., R.R. N EWTON AND G.L. G ROSS , Mathematical Theory
of Rocket Flight, New York: McGraw-Hill Book Company, Inc., 1947.

Example 3.34
dP
Prove the relation P + dV
V =
dT
T using logarithmic differentiation.
Solution: Take the natural logarithm and then the differential of both sides of the equation PV = RT :
ln (PV ) = ln (RT ) ⇒ ln P + ln V = ln R + ln T
⇒ d (ln P + ln V ) = d (ln R + ln T )
dP dV dT dT
⇒ + = 0 + = (since ln R is a constant)
P V T T
Example 3.35
The derivative of the area πr2 of a circle of radius r, as a function of r, equals its circumference 2πr. Use
the notion of a differential as an infinitesimal change to explain why this makes sense geometrically.
Solution: Let A = πr2 be the area of a circle of varying radius r. Then A ′ (r) = 2πr, which is equivalent
to saying dA = 2πr dr. To see why this makes sense geometrically, imagine increasing the radius by
dr, as in the picture below on the left. This increases the area A of the circle to A + dA , with dA the
infinitesimal area of the shaded ring in the picture.
dr slice and 2π r
roll flat dr area = dA
r 2π(r + dr)
dA
dr area = 0
π dr
Slice that ring along the dashed line then roll it flat, yielding a trapezoid with height dr, top length
2πr (from the circumference of the inner circle of the ring), and bottom length 2π(r + dr) (from the
circumference of the outer circle of the ring), as shown in the picture above on the right. The triangular
edges of the trapezoid contribute nothing to the area of the trapezoid, since (by the Microstraightness
Property) the hypotenuse of each is indeed a straight line, so each is a right triangle with height dr and
(by symmetry) base πdr, thus having area 21 π(dr)2 = 0. Hence the entire area dA of the trapezoid comes
from the rectangular portion of height dr and base 2πr, which means dA = 2πr dr, as expected.
The above example answers the question of whether it is a happy coincidence that
the derivative of a circle’s area turns out to be the circle’s circumference—no, it is not!
Some other such cases (e.g. the derivative of a sphere’s volume is its surface area) are
left to the exercises. Note that a similar “coincidence” does not occur for a square: if
x is the length of each side then the area is x2 , but the derivative of x2 is 2 x, which is
not the perimeter of the square (i.e. 4 x). Why does this not follow the same pattern as
the circle? Think about a key difference in the shape of a square in comparison to a
circle, keeping differentiability in mind.
There are many benefits to using differentials—i.e. infinitesimals—in calculus.10

For example, recall Example 3.31 in Section 3.5 on related rates, where the volume V
of a right circular cylinder with radius r and height h changes with time t as
µ ¶
dV dr dh
= 2π r · h + π r2 · .
dt dt dt
The above equation forces you to consider only the derivative with respect to the time
variable t. What if you wanted to see the rates of change with respect to another
variable, such as r , h, or some other quantity? In that case using the differential
version of the above equation, namely
dV = 2π rh dr + π r 2 dh
provides more flexibility—you are free to divide both sides by any differential, not just
by dt. Many related rates problems would likely benefit from this approach.
Present-day calculus textbooks confuse the notion of a differential (infinitesimal) dx
with the idea of a small but real value ∆ x. The two are not the same. An infinitesimal
is not a real number and cannot be assigned a real value, no matter how small; ∆ x
can be assigned real values. Using dx and ∆ x interchangeably is a source of much
confusion for students (likewise for dy and ∆ y). This confusion rears its head in exer-
cises involving the linear approximation of a curve by its tangent p line near a point x0 ,
namely f ( x) ≈ f ( x0 ) + f ( x0 )( x − x0 ) when x − x0 is “small” (e.g. 63 ≈ 7.9375, by using
′
p
f ( x) = x, x = 63, x0 = 64, and x − xo = ∆ x = −1). Such exercises have nothing to do
with differentials, not to mention having dubious value nowadays. They are remnants
of a bygone era, before the advent of modern computing obviated the need for such
(generally) poor approximations.
Exercises
A
1. Find the differential df of f ( x) = x2 − 2 x + 5. 2. Find the differential df of f ( x) = sin2 ( x2 ).
¡ ¢ x dy − y dx
3. Show that d tan−1 ( y/ x) = 4. Given y2 − x y + 2 x2 = 3, find dy.
x 2 + y2
x dy d (ln y)
5. The elasticity of a function y = f ( x) is E ( y) = · . Show that E ( y) = .
y dx d (ln x)
6. Prove the differential version of the Quotient Rule:
µ ¶
f g df − f dg
d =
g g2
10 For an excellent overview on this subject, see D RAY, T. AND C.A. M ANOGUE , Putting Differentials Back into
Calculus, College Math. J. 41 (2010), 90-100. Some of the material in this section is indebted to that paper, which
is available at http://www.math.oregonstate.edu/bridge/papers/differentials.pdf
7. Let y = c u n , where c and n are constants. Show that

dy du
= n .
y u
8. Obviously the derivative of the constant π2 is not 2π. But is d (π2 ) = 2π d (π) true? Explain.
B
9. The continuity relation for an ideal gas is
PM
p = constant
T
where P and T are the pressure and temperature, respectively, of the gas, and M is the Mach
number. Show that
dP dM dT
+ = .
P M 2T
10. For an ideal gas, satisfying the equation PV = RT as before, the Gibbs energy G is defined as
G = H − T S , where H and S are the enthalpy and entropy, respectively, of the gas.
(a) Show that µ ¶
G 1 G
d = dG − dT .
RT RT RT 2
(b) One of the fundamental property relations for an ideal gas (which you do not need to prove) is
dG = V dP − S dT .
Use this and part (a) to show that
µ ¶
G V H
d = dP − dT .
RT RT RT 2
11. The derivative of the volume πr2 h of a right circular cylinder of radius r and height h, as a function
of r, equals its lateral surface area 2πrh. Use the notion of a differential as an infinitesimal change
to explain why this makes sense geometrically.
12. The derivative of the volume 43π r3 of a sphere of radius r, as a function of r, equals its surface area
4πr2 . Use the notion of a differential as an infinitesimal change to explain why this makes sense
geometrically.
13. In quantum calculus the q-differential of a function f ( x) is
d q f ( x) = f ( qx) − f ( x) ,
and the q-derivative of f ( x) is
d q f ( x) f ( qx) − f ( x) f ( qx) − f ( x)
D q f ( x) = = = .
dq x qx − x ( q − 1) x
(a) Show that for all positive integers n,
¡ ¢
D q x n = [n] x n−1 ,
where [n] = 1 + q + q2 + · · · + q n−1 .
(b) Use part (a) to show that for all positive integers n,
¡ ¢ d ¡ n¢
lim D q x n = x .
q→1 dx
CHAPTER 4
Applications of Derivatives
4.1 Optimization
Many physical problems involve optimization: finding either a maximum or mini-
mum value of some quantity. Optimization problems often have a constraint involv-
ing two variables which allows you to rewrite the objective function—the function to
optimize—as a function of a single variable: use the constraint to solve for one variable
in terms of another, then substitute that expression into the objective function.
First, the intuitive notions of maximum and minimum need clarifying.
A function f has a global maximum at x = c if f ( c) ≥ f ( x) for all x in the domain of

f . Similarly, f has a global minimum at x = c if f ( c) ≤ f ( x) for all x in the domain
of f . Say that f has a local maximum at x = c if f ( c) ≥ f ( x) for all x “near” c,
i.e. for all x such that | x − c | < δ for some number δ > 0. Likewise, f has a local
minimum at x = c if f ( c) ≤ f ( x) for all x such that | x − c | < δ for some number δ > 0.
In other words, a global maximum is the largest value everywhere (“globally”), whereas
a local maximum is only the largest value “locally.” Likewise for a global vs local min-
imum. The picture below illustrates the differences.
y
y = f ( x)
x
a c1 c2 b
In the picture, on the interval [a, b] the function f has a global minimum at x = a, a
global maximum at x = c 1 , a local minimum at x = c 2 , and a local maximum at x = b.
90
Optimization • Section 4.1 91
Every global maximum [minimum] is a local maximum [minimum], but not vice
versa. In physical applications global maxima or minima1 are the primary interest.
The Extreme Value Theorem in Section 3.3 guarantees the existence of at least one
global maximum and at least one global minimum for continuous functions defined on
closed intervals (i.e. intervals of the form [a, b]). All the functions under consideration
here will be differentiable, and hence continuous. So the only issues will be how to
find the global maxima or minima, and how to handle intervals that are not closed.
Consider again the picture from the previous page, this time looking at how the
derivative f ′ changes over [a, b]. Intuitively it is obvious that near an internal max-
imum (i.e. in the open interval (a, b)) such as at x = c 1 , the function should increase
before that point and then decrease after that point. That means that f ′ ( x) > 0 before
x = c 1 and f ′ ( x) < 0 after the “turning point” x = c 1 , as shown below.
y
f′=0 y = f ( x)
f′<0
f′>0
f′>0
f′=0
x
a c1 c2 b
Assuming that f ′ is continuous (which will be the case for all the functions in this
section), then this means that f ′ = 0 at x = c 1 , that is, f ′ ( c 1 ) = 0. Similarly, near
the internal minimum at x = c 2 , f ′( x) < 0 before x = c 2 and f ′ ( x) > 0 after x = c 2 , so
that f ′ ( c 2 ) = 0. Points at which the derivative is zero are called critical points (or
stationary points) of the function. So x = c 1 and x = c 2 are critical points of f .
Note in the picture that f ′ goes from positive to zero to negative around x = c 1 , so
that f ′ is decreasing around x = c 1 , i.e. f ′′ = ( f ′ )′ < 0. Similarly, f ′ is increasing around
x = c 2 , i.e. f ′′ > 0. This leads to the following test for local maxima and minima:2
Second Derivative Test: Let x = c be a critical point of f (i.e f ′ ( c) = 0). Then:

(a) If f ′′ ( c) > 0 then f has a local minimum at x = c.
(b) If f ′′ ( c) < 0 then f has a local maximum at x = c.
(c) If f ′′ ( c) = 0 then the test fails.
To see why the test fails when f ′′ ( c) = 0, consider f ( x) = x3 : f ′ (0) = 0 and f ′′ (0) = 0, yet
x = 0 is neither a local minimum nor maximum in any open interval containing x = 0.
Section 4.2 will present an alternative for when the Second Derivative Test fails.
1 The words “maxima” and “minima” are the traditional plural forms of maximum and minimum, respectively.
2 A formal proof requires the Mean Value Theorem, which will be presented in Section 4.4.
92 Chapter 4 • Applications of Derivatives §4.1
There is a simple visual mnemonic device for remembering the Second Derivative
Test, due to a generic minimum or maximum resembling a smile or frown, respec-
tively:
+ + − − o o
f ′′ > 0 f ′′ < 0 f ′′ = 0
local min. local max. test fails
The “eyes” in the faces represent the sign of f ′′ at a critical point, while the “mouths”
indicate the nature of that point (when f ′′ = 0 nothing is known). The procedure for
finding a global maximum or minimum can now be stated:
How to find a global maximum or minimum

Suppose that f is defined on an interval I . There are two cases:
1. The interval I is closed: The global maximum of f will occur either at an
interior local maximum or at one of the endpoints of I whichever of these points
provides the largest value of f will be where the global maximum occurs.
Similarly, the global minimum of f will occur either at an interior local min-
imum or at one of the endpoints of I ; whichever of these points provides the
smallest value of f will be where the global minimum occurs.
2. The interval I is not closed and has only one critical point: If the only
critical point is a local maximum then it is a global maximum. If the only critical
point is a local minimum then it is a global minimum.
In each case of the above procedure try to use the Second Derivative Test to verify that
a critical point is a local minimum or maximum, unless it is obvious from the nature
of the problem that there can be only a minimum or only a maximum.
Example 4.1
Show that the rectangle with the largest area for a fixed perimeter is a
square. y
Solution: Let L be the perimeter of a rectangle with sides x and y. The y
idea is that L is a fixed constant, but x and y can vary. Figure 4.1.1
x x
shows that there are many possible shapes for the rectangle, but in all
cases L = 2 x + 2 y. Let A be the area of such a rectangle. Then A = x y, Figure 4.1.1
which is a function of two variables. But
L
L = 2x + 2 y ⇒ y = − x,
2
and hence µ ¶
L Lx
A = x − x = − x2
2 2
is now a function of x alone, on the open interval (0, L/2) (since the length x is positive). Now find the
critical points of A :
L
A ′ ( x) = 0 ⇒ − 2x = 0
2
L
⇒ x = is the only critical point
4
This problem is thus the case of a function defined on an open interval having only one critical point.
Use the Second Derivative Test to verify that the sole critical point x = L/4 is a local maximum for A :
A ′′ ( x) = −2 ⇒ A ′′ (L/4) = −2 < 0 ⇒ A has a local maximum at x = L/4
Thus, A has a global maximum at x = L/4. Also, y = L/2 − x = L/2 − L/4 = L/4, which means that x = y,
i.e. the rectangle is a square.
Note: The constraint in this example was L = 2 x + 2 y and the objective function was A = x y.
Example 4.2
Suppose a right circular cylindrical can with top and bottom lids will be assembled r
to have a fixed volume. Find the radius and height of the can that minimizes the
total surface area of the can.
Solution: Let V be the fixed volume of the can with radius r and height h, as in
Figure 4.1.2. The volume V is a constant, with V = πr2 h. Let S be the total surface h
area of the can, including the lids. Then
S = 2πr2 + 2πrh
where the first term in the sum on the right side of the equation is the combined
area of the two circular lids and the second term is the lateral surface area of the Figure 4.1.2
can. So S is a function of r and h, but h can be eliminated since
V
V = π r2 h ⇒ h =
π r2
and so
V 2V
S = 2π r 2 + 2π r · = 2π r 2 +
π r2 r
making S a function of r alone. Now find the critical points of S (i.e. solve S ′ (r) = 0):
2V
S ′ ( r) = 0 ⇒ 4π r − = 0
r2
V
⇒ r3 =
2π
s
3 V
⇒ r = is the only critical point
2π
Since both r and h are lengths and have to be positive, then 0 < r < ∞. So this is another case of a
function defined on an open interval
q having only one critical point. Use the Second Derivative Test to
3 V
verify that this critical point r = 2π is a local minimum for S :
Ãs ! s
′′ 4V ′′ 3 V 4V 3 V
S ( r ) = 4π + 3 ⇒ S = 4π + = 12π > 0 ⇒ S has a local minimum at r =
r 2π V 2π
2π
q
3 V
Thus, S has a global minimum at r = 2π , and
V V
r · r2 = r3 = ⇒ 2r = = h.
2π π r2
q q
3 V 3 V
Hence, r = 2π and h = 2 2π will minimize the total surface area, i.e. the height should equal the
diameter.
Note that this result can be applied to soda cans, where the volume is V = 12 fluid ounces ≈ 21.6 cubic
inches: both a diameter and height of about 3.8 inches will minimize the amount (and hence the cost)
of the aluminum used for the can. Yet soda cans are not that wide and short—they are usually thinner
and taller. So why is a non-optimal size used in practice? Other factors—e.g. packing requirements, the
need for small children to hold the can in one hand—might override the desire to minimize the cost of
the aluminum. The lesson is that an optimal solution for one factor (material cost) might not always be
truly optimal when all factors are considered; compromise is often necessary.
Example 4.3
Suppose that a projectile is launched from the ground with a fixed
initial velocity v0 at an angle θ with the ground. What value of θ
v0
would maximize the horizontal distance traveled by the projectile, θ

assuming the ground is flat and not sloped (i.e. horizontal)? L
Solution: Let x and y represent the horizontal position and vertical
position, respectively, of the projectile at time t ≥ 0. From the trian-
gle at the bottom of Figure 4.1.3, the horizontal and vertical com-
ponents of the initial velocity are v0 cos θ and v0 sin θ, respectively. v0 sin θ
v0
Since distance is the product of velocity and time, then the horizon- θ
tal and vertical distances traveled by the projectile by time t due to v0 cos θ
the initial velocity are (v0 cos θ) t and (v0 sin θ) t, respectively. Ignor- Figure 4.1.3
ing wind and air resistance, the only other force on the projectile
will be the downward force g due to gravity, so that the equations of motion for the projectile are:
x = (v0 cos θ) t
1
y = − gt2 + (v0 sin θ) t
2
The goal is to find θ that maximizes the length L shown in Figure 4.1.3. First write y as a function of x:
µ ¶2
x 1 x x
x = (v0 cos θ) t ⇒ t = ⇒ y = − g + (v0 sin θ) ·
v0 cos θ 2 v0 cos θ v0 cos θ
gx2
⇒ y = − + x tan θ
2v02 cos2 θ
Then L is the value of x > 0 that makes y = 0:
gL2 2v02 sin θ cos θ v02 sin 2θ

0 = − + L tan θ ⇒ L = =
2v02 cos2 θ g g
So L is now a function of θ, with 0 < θ < π/2 (why?). So if there is a single local maximum then it must
be the global maximum. Now get the critical points of L:
2v02 cos 2θ
L ′ (θ ) = 0 ⇒ = 0
g
⇒ cos 2θ = 0
π π
⇒ 2θ = ⇒ θ = is the only critical point
2 4
Use the Second Derivative Test to verify that L has a local maximum at θ = π/4:
4v02 sin 2θ 4v02
L′′ (θ) = − ⇒ L′′ (π/4) = − < 0
g g
π
⇒ L has a local maximum at θ =
4
Thus, L has a global maximum at θ = π
4,
i.e. the projectile travels the farthest horizontally when
¡ ¢ v2
launched at a 45 angle with the ground (with L π4 = g0 being the maximum horizontal distance).
◦
v02 sin 2θ
Note that once the formula for L as a function of θ was found to be L = g , calculus was not actually
needed to solve this problem. Why? Since and g are positive constants (recall g = 9.8m/s2 ), L would
v02
have its largest value when sin 2θ has its largest value 1, which occurs when θ = π/4.
Example 4.4
Fermat’s Principle states that light always travels along the path that takes the least amount of time.
So suppose that a ray of light is shone from a point A onto a flat horizontal reflective surface at an angle
θ1 with the surface and then reflects off the surface at an angle θ2 to a point B . Show that Fermat’s
Principle implies that θ1 = θ2 .
Solution: Let L be the horizontal distance between A and B, let d1 be the distance the light travels
from A to the point of contact C with the surface a horizontal distance x from A , let d2 be the distance
from C to B, and let y1 and y2 be the vertical distances from A and B, respectively, to the surface, as in
the picture below.
A
d2
y2
d1
y1
C
θ1 θ2
x L−x
L
Since time is distance divided by speed, and since the speed of light is constant, then minimizing the
total time elapsed is equivalent to minimizing the total distance traveled, namely D = d1 + d2 . The basic
idea here is that Fermat’s Principle implies that for the light to go from A to B in the shortest time, the
unknown point C —and hence the unknown distance x—will have to be at a point that makes θ1 = θ2 .
The distances L, y1 and y2 are constants, so the goal is to write the total distance D as a function of x,
find the x that minimizes D , then show that that value of x makes θ1 = θ2 .
First, note that C has to be between A and B as in the picture, otherwise the total distance D would
be larger than if C were directly below either A or B. This ensures that θ1 and θ2 are between 0 and
π/2, and that 0 ≤ x ≤ L.
Next, by the Pythagorean Theorem and the above picture,
q q
d1 = x2 + y12 and d2 = (L − x)2 + y22
and so the total distance D = d1 + d2 traveled by the light is a function of x:

q q
D ( x) = x2 + y12 + (L − x)2 + y22
To find the critical points of D , solve the equation D ′ ( x) = 0:

x L−x x L−x
D ′ ( x) = q − q = 0 ⇒ = ⇒ sin θ1 = sin θ2 ⇒ θ1 = θ2
x2 + y12 (L − x)2 + y22 d1 d2
since the sine function is one-to-one over the interval [0, π2 ].

This seems to prove the result, except for one remaining issue to resolve: verifying that the minimum
for D really does occur at the x between 0 and L where D ′ ( x) = 0, not at the endpoints x = 0 or x = L
of the closed interval [0, L]. Note that using the Second Derivative Test in this case does not matter,
since you would have to check the value of D at the endpoints anyway and compare those values to the
values of D at the critical points. To find expressions for the critical points, note that
x L−x x2 ( L − x )2
D ′ ( x) = 0 ⇒ q = q ⇒ =
x2 + y12 (L − x)2 + y22 x2 + y12 (L − x)2 + y22
✘ ✘ ✘ ✘
⇒ ✘ −✘
(L✘ x)2 x2 + x2 y22 = ✘ −✘
(L✘ x)2 x2 + (L − x)2 y12
L y1
⇒ x y2 = (L − x) y1 ⇒ x = is the only critical point,
y1 + y2
and x is between 0 and L. Now compare the values of D 2 ( x) at x = 0, x = L, and x =

L y1
y1 + y2 :
q
D 2 (0) = L2 + y12 + y22 + 2 y1 L2 + y22
q
D 2 (L) = L2 + y12 + y22 + 2 y2 L2 + y12
³ ´
D 2 y +1y = L2 + y12 + y22 + 2 y1 y2
Ly
1 2
q q ³ ´
L2 + y22 and y1 < L2 + y12 , then D 2
L y1
Since y2 < y1 + y2 is the smallest of the three values above, so that
2 L y1
D ( x) has its minimum value at x = y1 + y2 , which means D ( x) has its minimum value there. X
Example 4.5
A man is in a boat 4 miles off a straight coast. He wants to reach a point 10 miles down the coast in the
minimum possible time. If he can row 4 mi/hr and run 5 mi/hr, where should he land the boat?
Solution: Let T be the total time traveled. The goal is to minimize T . From the
Y p
picture on the right, since time is distance divided by speed, break the total time x2 + 16
into two parts: the time rowing in the water and the time running on the coast, 4
so that
p x 10 − x
distrow distrun x2 + 16 10 − x
T = timerow + timerun = + = + , 10
speedrow speedrun 4 5
where 0 ≤ x ≤ 10 is the distance along the coast where the boat lands. Then
x 1 p 16
T ′ ( x) = p − = 0 ⇒ 5 x = 4 x2 + 16 ⇒ 25 x2 = 16 ( x2 + 16) ⇒ x = ,
4 x2 + 16 5 3
16 16
so x = 3 is the only critical point. Thus, the (global) minimum of T will occur at x = 3 , x = 0, or x = 10.
Since p µ ¶
29 16 13
T (0) = 3 , T (10) = ≈ 2.693 , T = = 2.6
2 3 5
then T ( 16
3 ) < T (10) < T (0). Hence, the global minimum occurs when landing the boat x =
16
3 ≈ 5.33 miles
down the coast.
Note that this example shows the importance of checking the endpoints. It was quite close between
landing about 5.33 miles down the coast (2.6 hours) or simply rowing all the way to the destination
(about 2.693 hours)—the difference is only about 5.6 minutes. With just a slight change in a few of the
numbers, the minimum could have occurred at an endpoint. Moral: always check the endpoints!3
Example 4.6
p
Find the point ( x, y) on the graph of the curve y = x that is closest to the point (1, 0).
p p
Solution: Let ( x, y) be a point on the curve y = x. Then ( x, y) = ( x, x ), y
so by the distance formula the distance D between ( x, y) and (1, 0) is p
y= x
given by
p ( x, y)
D 2 = ( x − 1)2 + ( y − 0)2 = ( x − 1)2 + ( x )2 = ( x − 1)2 + x ,
which is a function of x ≥ 0 alone. Note that minimizing D is equivalent D

to minimizing D 2 . Since x
0 (1, 0)
d(D 2 ) 1
= 2 ( x − 1) + 1 = 2 x − 1 = 0 ⇒ x = 2 is the only critical point,
dx
2
(D ) 2
and since d dx 2 = 2 > 0 for all x, then by the Second Derivative Test x = 1/2 is a local minimum. Hence,
the global minimum for D 2 must occur at the endpoint x = 0 or at x = 1/2. But D 2p (0) = 1 > D 2 (1/2) = 3/4,
so the global minimum occurs at x = 1/2. Hence, the closest point is ( x, y) = (1/2, 1/2).
3 Another possible lesson is that optimal in the mathematical sense might, again, not mean optimal in a practical
sense. After all, presumably after the man is finished with whatever he had to do at the destination 10 miles down
the coast, he then has the inconvenience of going back about 4.67 miles to retrieve his boat. At his running speed
of 10 mph this would take 28 minutes, wiping out the 5.6 minutes he gained with his “optimal” landing spot!
Example 4.7
Find the width and height of the rectangle with the largest possible perimeter inscribed in a semicircle
of radius r.
Solution: Let w be the width of the rectangle and let h be the height, as
in the picture. Then the perimeter is P = 2w + 2h. By symmetry and the
Pythagorean Theorem,
r h
³ w ´2 1p 2
2 2
h = r − ⇒ h = 4 r − w2
2 2 w
2
p w
and so P = 2w + 4r2 − w2 for 0 < w < 2r. Find the critical points of P :
w p
P ′ ( w) = 2 − p = 0 ⇒ w = 2 4 r 2 − w2
4 r 2 − w2
⇒ w2 = 16r2 − 4w2
4r
⇒ w = p is the only critical point,
5
and since
4r2 ³ ´ 53/2
4r
P ′′ (w) = − ⇒ P ′′ p = − < 0
(4r2 − w2 )3/2 5 2r
4r
then P has a local maximum at w = p , by the Second Derivative Test. Since P (w) is defined for w in
5
4r
the open interval (0, 2r), the local maximum is a global maximum. For the width w = p the height is
5
h= pr , which gives the dimensions for the maximum perimeter.
5
Note: If w were extended to include the cases of “degenerate” rectangles of zero³ width
´ or height, i.e.
w = 0 or w = 2r, then the maximum perimeter would still occur at w = p , since P p = 10
4 r 4 r p r ≈ 4.472 r is
5 5 5
larger than P (0) = 2r and P (2r) = 4r.
Exercises
A
1. Find the point on the curve y = x2 that is closest to the point (4, −1/2).
2. Prove that for 0 ≤ p ≤ 1, p (1 − p) ≤ 41 .
3. A farmer wishes to fence a field bordering a straight stream with 1000 yd of fencing material. It is
not necessary to fence the side bordering the stream. What is the maximum area of a rectangular
field that can be fenced in this way?
4. The power output P of a battery is given by P = V I − R I 2 , where I , V , and R are the current,
voltage, and resistance, respectively, of the battery. If V and R are constant, find the current I that
maximizes P .
5. An impulse turbine consists of a high speed jet of water striking circularly mounted blades. The
power P generated by the turbine is P = VU (V − U ), where V is the speed of the jet and U is the
speed of the turbine. If the jet speed V is constant, find the turbine speed U that maximizes P .
6. A man is in a boat 5 miles off a straight coast. He wants to reach a point 15 miles down the coast in
the minimum possible time. If he can row 6 mi/hr and run 10 mi/hr, where should he land the boat?
7. The current I in a voltaic cell is

E
I = ,
R+r
where E is the electromotive force and R and r are the external and internal resistance, respectively.
Both E and r are internal characteristics of the cell, and hence can be treated as constants. The
power P developed in the cell is P = R I 2 . For which value of R is the power P maximized?
x2 y2
8. Find the point(s) on the ellipse 25 + 16 = 1 closest to (−1, 0).
2 y2
9. Find the maximum area of a rectangle that can be inscribed in an ellipse ax 2 + b2
= 1, where a > 0
and b > 0 are arbitrary constants. Your answer should be in terms of a and b.
10. Find the radius and angle of the circular sector with the maximum area and a fixed perimeter P .
11. In Example 4.3 show that the maximum height reached by the projectile when launched with an
v02 sin2 θ
initial velocity v0 at an angle 0 < θ < π
2 to the ground is 2g .
12. The phase velocity v of a capillary wave with surface tension T and water density p is
s
2π T λ g
v = +
λp 2π
where λ is the wavelength. Find the value of λ that minimizes v.
13. For an inventory model with a constant order quantity Q > 0 and a constant linear inventory de-
pletion rate D , the total unit cost TC to maintain an average inventory of Q /2 units is
P (I + W ) Q
TC = C + +
Q 2D
where C is the capital investment cost, P is the cost per order, I is the per unit interest charge per
unit time, and W is the overall inventory holding cost. Find the value of Q that minimizes TC .
14. The opening of a rain gutter—shown in the figure on the right—has

a bottom and two sides each with length a. The sides make an angle a a
θ with the bottom. Find the value of θ that maximizes the amount of
rain the gutter can hold. θ a θ
B
r0 x0
15. In an electric circuit with a supplied voltage (emf) E , a resistor
with resistance r 0 , and an inductor with reactance x0 , suppose
you want to add a second resistor. If r represents the resistance
of this second resistor then the power P delivered to that resistor E r
is given by
E2 r
P =
( r + r 0 )2 + x 0 2
with E , r 0 , and x0 treated as constants. For which value of r is the power P maximized?
16. The stress τ in the x y-plane along a varying angle φ is given by

σx − σ y
τ = τ(φ) = sin 2φ + τ x y cos 2φ ,
2
where σ x , σ y , and τ x y are stress components that can be treated as constants. Show that the maxi-
mum stress is q¡ ¢2
σ x − σ y + 4τ2x y
τ = .
2
(Hint: Draw a right triangle with angle 2φ after finding the critical point(s).)
17. A certain jogger can run 0.16 km/min, and walks at half that speed. If he runs along a circular trail
with circumference 50 km and then—before completing one full circle—walks back straight across
to his starting point, what is the maximum time he can spend on the run/walk?
18. A rectangular poster is to contain 50 square inches of printed material, with 4-inch top and bottom
margins and 2-inch side margins. What dimensions for the poster would use the least paper?
19. Find the maximum volume of a right circular cylinder that can be inscribed in a sphere of radius 3.
20. A figure consists of a rectangle whose top side coincides with the diameter of a semicircle atop it.
If the perimeter of the figure is 20 m, find the radius and height of the semicircle and rectangle,
respectively, that maximizes the area inside the figure.
21. A thin steel pipe 25 ft long is carried down a narrow corridor 5.4 ft wide. At the end of the corridor
is a right-angle turn into a wider corridor. How wide must this corridor be in order to get the pipe
around the corner? You may assume that the width of the pipe can be ignored.
22. A rectangle is inscribed in a right triangle, with one corner of the rectangle at the
right angle of the triangle. Show that the maximum area of the rectangle occurs
when a corner of the rectangle is at the midpoint of the hypotenuse of the triangle.
23. Find the relation between the radius and height of a cylindrical can with an open top that maxi-
mizes the volume of the can, given that the surface area of the can is always the same fixed amount.
24. An isosceles triangle is circumscribed about a circle of radius r. Find the height of the triangle that
minimizes the perimeter of the triangle.
25. Suppose N voltaic cells are arranged in N / x rows in parallel, with each row consisting of x cells in
series, creating a current I through an external resistance R . Each cell has internal resistance r and
EMF (voltage) e. Find the x that maximizes the current I , which—due to Ohm’s Law—is given by
xe
I = .
( x2 r/ N ) + R
26. A single-degree-of-freedom harmonically forced vibration system with damping factor ζ has mag-
nification factor MF = ((1 − r2 )2 + (2ζ r)2 )−1/2 , where r is the frequency ratio. Find the value of r that
maximizes MF .
27. At a distance x ≥ 0 from the center of a uniform ring with charge q and radius a, the magnitude E
of the electric field for points on the axis of the ring is
qx
E =
4πǫ0 (a2 + x2 )3/2
where ǫ0 is the permittivity of free space. Find the distance x that maximizes E .
2 y2
28. Find the equation of the tangent line to the ellipse ax 2 + b2 = 1 in the first quadrant that forms with
the coordinate axes the right triangle with minimal area.
29. A “cold” star that has exhausted its nuclear fuel—called a white dwarf —has total energy E , given
by
µ ¶−2/3
ħ2 (3π2 N q)5/3 4 3 3GM 2 N 2
E = 2
π R −
10π m 3 5R
where ħ is the reduced Planck constant, N is the number of nucleons—protons and neutrons—in the
star, q is the charge of an electron, m is the mass of an electron, M is the mass of a nucleon, G is the
gravitational constant, and R is the radius of the star. Show that the radius R that minimizes E is
µ ¶2/3
9π ħ2 q5/3
R = .
4 GmM 2 N 1/3
C
30. An object of mass m has orbital angular momentum l around a black hole with Schwarzchild radius
r S and mass M . The effective potential Φ of the object is
GM l2 rS l2
Φ = − + −
r 2 m2 r 2 2 m2 r 3
where G is the gravitational constant and r is the object’s distance from the black hole. Show that Φ
has a local maximum and minimum at r = r 1 and r = r 2 , respectively, where
 s   s 
l2 1 − 1 − 6 GMm 2r
S  l 2
1 + 1 − 6GMm 2r
S 
r1 = and r 2 = .
2GMm2 l2 2GMm2 l2
31. Recall Fermat’s Principle from Example 4.4, which states that light travels A
along the path that takes the least amount of time. The speed of light in a
vacuum is approximately c = 2.998 × 108 m/s, but in some other medium (e.g. θ1
water) light is slower. Suppose that a ray of light goes from a point A in one
medium where it moves at a speed v1 and ends up at a point B in another θ2
medium where it moves at a speed v2 . Use Fermat’s Principle to prove Snell’s
Law, which says that the light is refracted through the boundary between B
the two media such that
sin θ1 v1
=
sin θ2 v2
where θ1 and θ2 are the angles that the light makes with the normal line perpendicular to the
boundary of the media in the first and second medium, respectively, as in the picture above.
32. A sphere of radius a is inscribed in a right circular cone, with the sphere touching the base of the
cone. Find the radius and height of the cone if its volume is a minimum.
33. Find the length of the shortest line segment from the positive x-axis to the positive y-axis going
through a point (a, b) in the first quadrant.
34. Find the radius r of a circle c whose center is on a fixed circle C of radius R such that the arc length
of the part of c within C is a maximum.
4.2 Curve Sketching
A function can increase between two points in different ways, as shown in Figure 4.2.1.
y y y
y = f ( x)
y = f ( x)
y = f ( x)
x x x
(a) f = 0: straight
′′
(b) f > 0: concave up
′′
(c) f < 0: concave down
′′
Figure 4.2.1 Increasing function f : f ′ > 0, different signs for f ′′
In each case in the above figure the function is increasing, so that f ′ ( x) > 0, but the
manner in which the function increases is determined by its concavity, that is, by
the sign of the second derivative f ′′ ( x). The function in the graph on the far left is
linear, i.e. of the form f ( x) = ax + b for some constants a and b, so that f ′′ ( x) = 0 for all
x. But the functions in the other two graphs are nonlinear. In the middle graph the
derivative f ′ is increasing, so that f ′′ > 0; in this case the function is called concave
up. In the graph on the far right the derivative f ′ is decreasing, so that f ′′ < 0; in
this case the function is called concave down. The same definitions would hold if the
function were decreasing, as shown in Figure 4.2.2 below:
y y y
y = f ( x)
y = f ( x)
y = f ( x)
x x x
(a) f = 0 ′′
(b) f > 0: concave up
′′
(c) f < 0: concave down
′′
Figure 4.2.2 Decreasing function f : f ′ < 0, different signs for f ′′
In Figures 4.2.1(b) and 4.2.2(b) the function is below the line joining the points at
each end, while in Figure 4.2.1(c) and 4.2.2(c) the function is above that line. This
turns out to be true in general, as a result of the following theorem:
Curve Sketching • Section 4.2 103
Concavity Theorem: Suppose that f is a twice-differentiable function on [a, b].

Then:
(a) If f ′′ ( x) > 0 on (a, b) then f ( x) is below the line l ( x) joining the points (a, f (a))
and ( b, f ( b)) for all x in (a, b).
(b) If f ′′ ( x) < 0 on (a, b) then f ( x) is above the line l ( x) joining the points (a, f (a))
and ( b, f ( b)) for all x in (a, b).
Proof: Only part (a) will be proved; the proof of part y

(b) is similar and left as an exercise. So assume that
f ( b)
f ′′ ( x) > 0 on (a, b), and l ( x) be the line joining (a, f (a)) l ( x)
and ( b, f ( b)), as in the drawing on the right. The draw-
f (a)
ing suggests that f ( x) < l ( x) over (a, b), but this is
what has to be proved. y = f ( x)
The goal is to show that g( x) = f ( x) − l ( x) < 0 on
x
(a, b), since this will show that f ( x) < l ( x) on (a, b). a b
Since f and l are both continuous on [a, b] then so is
g. Hence g has a global maximum somewhere in [a, b], by the Extreme Value Theo-
rem. Suppose the global maximum occurs at an interior point x = c, i.e. for some c
in the open interval (a, b). Then g′ ( c) = 0 and g′′ ( c) = f ′′ ( c) − l ′′ ( c) = f ′′ ( c) > 0, since
l ( x) is a line and hence has a second derivative of 0 for all x. Then by the Sec-
ond Derivative Test g has a local minimum at x = c, which contradicts g having a
global maximum at x = c. Thus, the global maximum of g cannot occur at an inte-
rior point, so it must occur at one of the end points x = a or x = b. In other words,
either g( x) < g(a) or g( x) < g( b) for all x in (a, b). But f (a) = l (a) and f ( b) = l ( b), so
g(a) = 0 = g( b). Hence, g( x) < 0 for all x in (a, b), i.e. f ( x) < l ( x) for all x in (a, b). X
Points where the concavity of a function changes have a special name:
A function f has an inflection point at x = c if the concavity of f changes around

x = c. That is, the function goes from concave up to concave down, or vice versa.
Note that to be an inflection point it does not suffice for the second derivative to
be 0 at that point; the second derivative must change sign around that point, either
from positive to negative or from negative to positive. For example, f ( x) = x3 has an
inflection point at x = 0, since f ′′ ( x) = 6 x < 0 for x < 0 and f ′′ ( x) = 6 x > 0 for x > 0, i.e.
f ′′ ( x) changes sign around x = 0 (and of course f ′′ (0) = 0). But for f ( x) = x4 , x = 0 is not
an inflection point even though f ′′ (0) = 0, since f ′′ ( x) = 12 x2 ≥ 0 is always nonnegative.
That is, f ( x) = x4 is always concave up. Figure 4.2.3 below shows the difference:
y
y
x
0
x
0
(a) f ( x) = x3 : inflection point at x = 0 (b) f ( x) = x4 : non-inflection point at x = 0
Figure 4.2.3 Inflection vs non-inflection point at x = 0 with f ′′ (0) = 0
Figure 4.2.3(b) shows that a point where the second derivative is 0 is a possible inflec-
tion point, but you still must check that the second derivative changes sign around that
point. Using local minima and maxima, concavity and inflection points, and where a
function increases or decreases, you can sketch the graph of a function.
Example 4.8
Sketch the graph of f ( x) = x3 − 6 x2 + 9 x + 1. Find all local maxima and minima, inflection points,
where the function is increasing or decreasing, and where the function is concave up or concave down.
Solution: Since f ′ ( x) = 3 x2 − 12 x + 9 = 3 ( x − 1) ( x − 3) then x = 1 and x = 3 are the only critical points.
And since f ′′ ( x) = 6 x − 12 then f ′′ (1) = −6 < 0 and f ′′ (3) = 6 > 0. So by the Second Derivative Test, f
has a local maximum at x = 1 and a local minimum at x = 3. Since f ′′ ( x) = 6 x − 12 < 0 for x < 2 and
f ′′ ( x) = 6 x − 12 > 0 for x > 2, then x = 2 is an inflection point, and f is concave down for x < 2 and concave
up for x > 2. The table below shows where f is increasing and decreasing, based on the sign of f ′ :
x values 3 ( x − 1) ( x − 3) f ′ ( x) direction
x<1 − − + f is increasing
1<x<3 + − − f is decreasing
x>3 + + + f is increasing
The graph is shown below:

6
3
y
0
0 0.5 1 1.5 2 2.5 3 3.5 4
x
Example 4.9
Sketch the graph of f ( x) = 1 − x
+ x2
. Find all local maxima and minima, inflection points, where the func-
tion is increasing or decreasing, and where the function is concave up or concave down. Also indicate
any asymptotes.
x 2 −1 2 x (3− x2)
Solution: Since f ′ ( x) = (1+ x2 )2
then x = 1 and x = −1 are the only critical points. And since f ′′ ( x) = (1+ x2 )3
then f ′′ (1) = 21 > 0 and f (−1) = − 12 < 0. So by the Second Derivative Test, f has a local minimum at
′′
p p
x = 1 and
p a local maximum at x =p
−1. Since f ′′ ( x) >p0 for x < − 3, f ′′ ( x) < 0 for − 3 < x < 0, f ′′ ( x) > 0 p
for
0 < x < 3, and p f ′′ ( x) < 0 for x > 3, then x = 0, ±p 3 are inflection points, p f is concave up for x < − 3
and for 0 < x < 3, and f is concave down for − 3 < x < 0 and for x > 3. Since f ′ ( x) > 0 for x < −1
and x > 1 then f is increasing for | x | > 1. And f ′ ( x) < 0 for −1 < x < 1 means f is decreasing for | x | < 1.
Finally, since lim f ( x) = 0 and lim f ( x) = 0 then the x-axis ( y = 0) is a horizontal asymptote. There are
x→∞ x→−∞
no vertical asymptotes (why?).
The graph is shown below:
0.5
0
y
−0.5
−1
−4 −3 −2 −1 0 1 2 3 4
x
If the Second Derivative Test fails then one alternative is the following test:
First Derivative Test: For a continuous function f on an interval I , let x = c be

a number in I such that f ( c) is defined, and either f ′ ( c) = 0 or f ′( c) does not exist.
Then:
(a) If f ′( x) changes from negative to positive around x = c then f has a local mini-
mum at x = c.
(b) If f ′ ( x) changes from positive to negative around x = c then f has a local max-
imum at x = c.
This test merely states the obvious: a function decreases then increases around a
minimum, and it increases then decreases around a maximum.
Example 4.10
Sketch the graph of f ( x) = x2/3 .
1.4
Solution: Clearly f ( x) is continuous for all x, including x = 1.2
0 (since f (0) = 0), but f ′ ( x) = 2p
3 is not defined at x = 0. 1
3 x 0.8
y
Since f ′ ( x) changes from negative to positive around x = 0 0.6
( f ′ ( x) < 0 when x < 0 and f ′ ( x) > 0 when x > 0), then by the 0.4
0.2
First Derivative Test f has a local minimum at x = 0. Since 0
f ′′ ( x) = − 9 x24/3 < 0 for all x 6= 0, then f is always concave −2−1.5−1−0.5 0 0.5 1 1.5 2
down. There are no vertical or horizontal asymptotes. The
x
graph is shown on the right.
Note that the Second Derivative Test could not be used for this function, since f ′ ( x) 6= 0 for all x (notice
also that f ′′ ( x) is not defined at x = 0).
A more complete alternative to the Second Derivative Test is the following:4
Nth Derivative Test: A non-constant function f with continuous derivatives of

all orders up to and including n > 1 at x = c has either a local minimum, local
maximum or inflection point at x = c if and only if
f (k) ( c) = 0 for k = 1, 2, . . ., n − 1 and f (n) ( c) 6= 0
(i.e. the nth derivative is the first nonzero derivative at x = c). If so, then:
(a) If n > 1 is even and f (n) ( c) > 0 then f has a local minimum at x = c.
(b) If n > 1 is even and f (n) ( c) < 0 then f has a local maximum at x = c.
(c) If n > 1 is odd then f has an inflection point at x = c.
Note that the Second Derivative Test is the special case where n = 2 in the Nth Deriva-
tive Test. Though this test gives necessary and sufficient conditions for a local max-
imum, local minimum, and inflection point, calculating the first n derivatives can be
complicated if n is large and the given function is not simple.
Example 4.11
The Second Derivative Test fails for f ( x) = x4 at the critical point x = 0, since f ′′ (0) = 0. But the first
4 derivatives of f ( x) = x4 are f ′ ( x) = 4 x3 , f ′′ ( x) = 12 x2 , f (3) ( x) = 24 x, and f (4) ( x) = 24, which are all
continuous and
f ( k) (0) = 0 for k = 1, 2, 3 and f (4) (0) = 24 6= 0 .
So by the Nth Derivative Test, since n = 4 is even and f (4) (0) = 24 > 0 then f ( x) = x4 has a local minimum
at x = 0. Note that f ( x) ≥ 0 = f (0) for all x, so x = 0 is actually a global minimum for f .
4 For a proof, see pp.10-11 in K OO, D., Elements of Optimization, New York: Springer-Verlag, 1977.
A common practice in many fields of science and engineering is to combine multiple

named constants (e.g. π) or variables in a function into one variable and then sketch a
graph of that function. The example below illustrates the technique.
Example 4.12
A hydrogen atom has one electron, and the probability of finding the electron in the ground state of the
hydrogen atom between radii r and r + dr is D (r) dr, where dr is an infinitesimal change in the radius
r (the distance from the electron to the nucleus), D (r) is the radial probability density function
4
D ( r) = r 2 e −2 r / a 0
a30
and a0 ≈ 5.291772 × 10−11 m is the Bohr radius. It is useful to analyze this function in terms of r ≥ 0 in
relation to the Bohr radius a0 (e.g. r = 0.5a0 , a0 , 2a0 , 3a0 ). To do this, let x = ar0 , so that
µ ¶2 ³ ´
4 r −2 ar
D ( r) = e 0 ⇒ a 0 D ( x ) = 4 x 2 e −2 x
a0 a0
and then sketch the graph of a0 D ( x), which is shown below:
0.6
0.5
0.4
³ ´
0
a 0 D ar
0.3
0.2
0.1
0
0 1 2 3 4 5 6
r
a0
From the graph it looks like x = 1 (i.e. r = a0 ) is a local (and global) maximum, so that the electron is
most likely to be found near r = a0 , and the probability drops off dramatically past a distance r = 3a0 .
In the exercises
³ you will
´ be asked to show that r = a0 is indeed a local maximum and that the inflection
1
points are r = 1 ± p a0 .
2
Note that the right side of the formula a0 D ( x) = 4 x2 e−2 x does not involve a0 , which was multiplied
over to the left side. In general that is the strategy when dealing with these sorts of functions where
variables and constants are combined. In this case the stray constant a0 can be multiplied with D since
that will not affect the location of critical and inflection points, nor fundamentally alter the general
shape of the graph.
Example 4.13
For a single particle with two states—energy 0 and energy ǫ—in thermal contact with a reservoir at
temperature τ, the average energy U and heat capacity CV are given by
e−ǫ/τ ³ ǫ ´2 eǫ/τ
U = ǫ and C V = k B ¡ ¢2
1 + e−ǫ/τ τ 1 + eǫ/τ
where k B ≈ 1.38065 × 10−23 J/K is the Boltzmann constant. The graph below shows both quantities as
functions of τ/ǫ (not ǫ/τ, as you might expect). See Exercise 9.
Average Energy U /ǫ vs Heat Capacity CV /k B

0.5
0.45
0.4
0.35 e−ǫ/τ
U /ǫ =
0.3 1 + e−ǫ/τ
0.25
0.2 ³ ǫ ´2 eǫ/τ
0.15 C V /k B = ¡ ¢2
τ 1 + eǫ/τ
0.1
0.05
0
0 1 2 3 4 5
τ/ǫ
Exercises
A
For Exercises 1-8 sketch the graph of the given function. Find all local maxima and minima, inflection
points, where the function is increasing or decreasing, where the function is concave up or concave
down, and indicate any asymptotes.
2
1. f ( x) = x3 − 3 x 2. f ( x) = x3 − 3 x2 + 1 3. f ( x) = xe− x 4. f ( x) = x2 e− x
1 x2 e − x − e −2 x
5. f ( x) = 6. f ( x) = 7. f ( x) = 8. f ( x) = e− x sin x
1 + x2 ( x − 1)2 2
9. Write U /ǫ and CV /k B from Example 4.13 as functions of x = τ/ǫ. You do not need to sketch the graphs.
10. Show that the function D (r) = 43 r2 e−2 r/a 0 from Example 4.12 has a local maximum at r = a0 and
³ ´ a0
1
inflection points at r = 1 ± p a0 .
2
³ 2
´
11. Sketch the graph of Kratzer’s molecular potential V (r) = −2D ar − 12 ar2 as a function of x = ar , with
a > 0 and D > 0 as constants.
p − K
2p
N K e kT
12. Sketch the graph of f (K ) = π ( kT )3/2
as a function of x = kT ,
K
with N , k and T as positive constants.
13. Prove part (b) of the Concavity Theorem.
Numerical Approximation of Roots of Functions • Section 4.3 109
4.3 Numerical Approximation of Roots of Functions

When finding critical points of a function f , you encounter the problem of solving the
equation f ′ ( x) = 0. The examples and exercises so far were set up carefully so that
solutions to that equation could be found in a simple closed form. But in practice this
will not always be the case—in fact it is almost never the case. For example, finding
2
the critical points of f ( x) = sin x − x2 entails solving the equation f ′ ( x) = cos x − x = 0,
for which there is no solution in a closed-form expression.
What should you do in such a situation?5 One possibility is to use the bisection
method mentioned in Section 3.3. In fact, in Example 3.25 the solution to the equation
cos x = x (i.e. cos x − x = 0) was shown to exist in the interval [0, 1], and then a
demonstration of the bisection method was given to find that solution.
The bisection method is one of many numerical methods for finding roots of a func-
tion (i.e. where the function is zero). Finding the critical points of a function means
finding the roots of its derivative. Though the bisection method could be used for that
purpose, it is not efficient—convergence to the root is slow. A far more efficient method
is Newton’s method6 , whose geometric interpretation is shown in Figure 4.3.1 below.
y
( x0 , f ( x0 ))
y = f ( x)
( x1 , f ( x1 ))
x
0 x̄ x2 x1 x0
Figure 4.3.1 Newton’s method for finding a root x̄ of f ( x)
The idea behind Newton’s method is simple: to find a root x̄ of a function f , choose an
initial guess x0 and then go up—or down—to the curve y = f ( x) and draw the tangent
line to the curve at the point ( x0 , f ( x0 )). Let x1 be where that tangent line intersects
the x-axis, as shown above; repeat this procedure on x1 to get the next number x2 ,
repeat on x2 to get x3 , and so on. The resulting sequence of numbers x0 , x1 , x2 , x3 , . . .,
will approach the root x̄. Convergence under certain conditions can be proved.7
5 Note: To “just give up”—as suggested semi-seriously by some students I have had—is not an option.
6 Sometimes called the Newton-Raphson method.
7 See pp.58-62 in S AATY, T.L. AND J. B RAM, Nonlinear Mathematics, New York: McGraw-Hill, Inc., 1964.
The general formula for the number xn obtained after n ≥ 1 iterations in Newton’s
method can be determined by considering the formula for x1 . First, the tangent line to
y = f ( x) at the point ( x0 , f ( x0 )) has slope f ′( x0 ), so the equation of the line is
y − f ( x0 ) = f ′( x0 ) ( x − x0 ) . (4.1)
The point ( x1 , 0) is (by design) also on that line, so that
f ( x0 )
0 − f ( x0 ) = f ′ ( x0 ) ( x1 − x0 ) ⇒ x1 = x0 −
f ′ ( x0 )
provided that f ′ ( x0 ) 6= 0. The general formula for xn is given by the following algorithm:
Newton’s method: For an initial guess x0 , the numbers xn for n ≥ 1 are computed
iteratively as:
f ( xn−1 )
xn = xn−1 − ′ for n = 1, 2, 3, . . .
f ( xn−1 )
That is, each “next” number xn depends on the previous number xn−1 . The algo-
rithm terminates whenever f ′ ( xn ) = 0, or when the desired accuracy is reached. If
f ′ ( xn ) = 0 for some n ≥ 0, then you could start over with a different initial guess x0 .
To implement this algorithm in a programming language (for which Newton’s method

is well-suited), the following language-independent pseudocode can be used as a guide:
Algorithm pseudocode for Newton’s method
N EWTON ’ S METHOD
1 N ← N UMBER - OF - ITERATIONS ✄ User supplies this value
2 x ← INITIAL - GUESS ✄ User supplies this value
3 for n ← 1 to N
4 do
5 if f ′ ( x) 6= 0
6 then
f ( x)
7 x←x − ′
f ( x)
8 print x
9 else
10 error “division by zero”
Example 4.14
Use Newton’s method to find the root of f ( x) = cos x − x.
Solution: Since the root is already known to be in the interval [0, 1], choose x0 = 1 as the initial guess.
The numbers x n for n ≥ 1 can be computed with a hand-held scientific calculator, but the process is
tedious and error-prone. Using a computer is far more efficient and allows more flexibility.
For example, the algorithm is easily implemented in the Java programming language. Save this code
in a plain text file as newton.java:
Listing 4.1 Newton’s method in Java (newton.java)

1 public class newton {
2 public static void main(String[] args) {
3 int N = Integer.parseInt(args[0]); //Number of iterations
4 double x = 1.0; //initial guess
5 System.out.println("n=0: " + x);
6 for (int i = 1; i <= N; i++) {
7 x = x - f(x)/derivf(x);
8 System.out.println("n=" + i + ": " + x);
9 }
10 }
11
12 //Define the function f(x)
13 public static double f(double x) {
14 return Math.cos(x) - x;
15 }
16
17 //Define the derivative f’(x)
18 public static double derivf(double x) {
19 return -Math.sin(x) - 1.0;
20 }
21 }
Though knowledge of Java would help, it should not be that difficult to figure out what the above code
is doing. The number of iterations N is passed as a command-line parameter to the program, and x n
is computed and printed for n = 0, 1, 2, ... , N . Note that the derivative of f ( x) is “hard-coded” into the
program.8 There is also no error checking for the derivative being zero at any x n . The program would
simply halt on a division by zero error.
Compile the code, then run the program with 10 iterations:
javac newton.java
java newton 10
The output is shown below:
8 There are some programming language libraries for calculating derivatives of functions “on the fly,” i.e. dynami-
cally. For example, the GNU libmatheval C/Fortran library can perform such symbolic operations. It is available
at http://www.gnu.org/software/libmatheval/
n=0: 1.0
n=1: 0.7503638678402439
n=2: 0.7391128909113617
n=3: 0.739085133385284
n=4: 0.7390851332151607
n=5: 0.7390851332151607
n=6: 0.7390851332151607
n=7: 0.7390851332151607
n=8: 0.7390851332151607
n=9: 0.7390851332151607
n=10: 0.7390851332151607
Note that the solution x̄ = 0.7390851332151607 was found after only 4 iterations; the numbers x n repeat
for n ≥ 5. This is much faster than the bisection method.
Another root-finding numerical method similar to Newton’s method is the secant

method, whose geometric interpretation is shown in Figure 4.3.2 below:
y
( x0 , f ( x0 ))
y = f ( x)
( x1 , f ( x1 ))
x
0 x̄ x2 x1 x0
Figure 4.3.2 Secant method for finding a root x̄ of f ( x)
The idea behind the secant method is simple: to find a root x̄ of a function f , choose
two initial guesses x0 and x1 , then go up—or down—to the curve y = f ( x) and draw the
secant line through the points ( x0 , f ( x0 )) and ( x1 , f ( x1 )) on the curve. Let x2 be where
that secant line intersects the x-axis, as shown above; repeat this procedure on x1 and
x2 to get the next number x3 , and keep repeating in this way. The resulting sequence
of numbers x0 , x1 , x2 , x3 , . . ., will approach the root x̄, under the right conditions.9
9 See pp.227-229 in D AHLQUIST, G. AND Å. B JÖRCK, Numerical Methods, Englewood Cliffs, NJ: Prentice-Hall,
Inc., 1974.
f ( x1 )− f ( x0 )
Since the secant line through ( x0 , f ( x0 )) and ( x1 , f ( x1 )) has slope x1 − x0 , the equa-
tion of that secant line is:
f ( x1 ) − f ( x0 )
y − f ( x1 ) = ( x − x1 )
x1 − x0
The point ( x2 , 0) is on that line, so that
f ( x1 ) − f ( x0 ) ( x1 − x0 ) · f ( x1 )
0 − f ( x1 ) = ( x2 − x1 ) ⇒ x2 = x1 −
x1 − x0 f ( x1 ) − f ( x0 )
provided that x1 6= x0 . The general formula for xn is given by the following algorithm:
Secant method: For two initial guesses x0 and x1 , the numbers xn for n ≥ 2 are
computed iteratively as:
( xn−1 − xn−2 ) · f ( xn−1 )

xn = xn−1 − for n = 2, 3, 4, . . . (4.2)
f ( xn−1 ) − f ( xn−2 )
That is, each “next” number xn depends on the previous two numbers xn−1 and
xn−2 . The algorithm terminates whenever xn = xn−1 (i.e. the numbers start repeat-
ing) or when the desired accuracy is reached.
Algorithm pseudocode for the secant method
S ECANT METHOD
1 N ← N UMBER - OF - ITERATIONS ✄ User supplies this value
2 x0 ← FIRST- INITIAL - GUESS ✄ User supplies this value
3 x1 ← SECOND - INITIAL - GUESS ✄ User supplies this value
4 f 0 ← f ( x0 )
5 for n ← 1 to N
6 do
7 f 1 ← f ( x1 )
8 if f 0 6= f 1
9 then
( x1 − x0 ) · f 1
10 x ← x1 −
f1 − f0
11 print x
12 x0 ← x1
13 f0 ← f1 ✄ Re-use f 1 as f 0 in the next iteration
14 x1 ← x
15 else
16 error “division by zero”
One difference you might have noticed between the secant method and Newton’s
method is that the secant method does not use derivatives. The secant method replaces
the derivative in Newton’s method with the slope of a secant line which approximates
the derivative (recall how the tangent line is the limit of slopes of secant lines). This
might seem like a drawback, perhaps giving a “less accurate” slope than the tangent
line, but in practice it is not really a problem. In fact, in many cases the secant method
is preferable, since computing derivatives can often be quite complicated.
Example 4.15
Use the secant method to find the root of f ( x) = cos x − x.
Solution: Since the root is already known to be in the interval [0, 1], choose x0 = 0 and x1 = 1 as the
two initial guesses. The algorithm is easily implemented in the Java programming language. Save this
code in a plain text file as secant.java:
Listing 4.2 Secant method in Java (secant.java)

1 import java.math.*;
2 public class secant {
3 public static void main(String[] args) {
4 int N = Integer.parseInt(args[0]); //Number of iterations
5 double x0 = 0.0; //first initial guess
6 double x1 = 1.0; //second initial guess
7 double f0 = f(x0);
8 double f1;
9 double x = 0.0;
10 for (int i = 2; i <= N; i++) {
11 f1 = f(x1);
12 x = x1 - (x1 - x0)*f1/(f1 - f0);
13 x0 = x1;
14 f0 = f1; //Re-use f1 as f0 in the next iteration
15 x1 = x;
16 System.out.println("n=" + i + ": " + x);
17 }
18 }
19
20 //Define the function f(x)
21 public static double f(double x) {
22 return Math.cos(x) - x;
23 }
24 }
Compile the code, then run the program:
javac secant.java
java secant 10
The output is shown below:

n=2: 0.6850733573260451
n=3: 0.736298997613654
n=4: 0.7391193619116293
n=5: 0.7390851121274639
n=6: 0.7390851332150012
n=7: 0.7390851332151607
n=8: 0.7390851332151607
n=9: NaN
n=10: NaN
Notice that the root was found after 6 iterations (n = 7). The undefined number NaN (which stands
for “Not a Number”) was returned starting with the eighth iteration (n = 9) because x7 = x8 , so that
f ( x8 ) − f ( x7 ) = 0, causing a division by zero error in the term
( x8 − x7 ) · f ( x8 )
x9 = x8 − .
f ( x8 ) − f ( x7 )
For the function f ( x) = cos x − x, the table below summarizes the results of 10 itera-
tions of the bisection method, Newton’s method and the secant method:
Term Bisection Newton Secant

x0 0.5 1.0 0.0
x1 0.75 0.7503638678402439 1.0
x2 0.625 0.7391128909113617 0.6850733573260451
x3 0.6875 0.739085133385284 0.736298997613654
x4 0.71875 0.7390851332151607 0.7391193619116293
x5 0.734375 0.7390851332151607 0.7390851121274639
x6 0.7421875 0.7390851332151607 0.7390851332150012
x7 0.73828125 0.7390851332151607 0.7390851332151607
x8 0.740234375 0.7390851332151607 0.7390851332151607
x9 0.7392578125 0.7390851332151607 undefined
x10 0.73876953125 0.7390851332151607 undefined
Newton’s method found the root after 4 iterations, while the secant method needed
6 iterations. After 10 iterations the bisection method had yet to find the root to the
same level of precision as the other methods—it would take 52 iterations (that is, x52 )
to achieve similar accuracy to 16 decimal places.
In general Newton’s method requires fewer iterations to find a root than the secant
method does, but this does not necessarily mean that it will always be faster. De-
pending on the complexity of the function and its derivative, Newton’s method could
involve more “expensive” operations (i.e. computing values, as opposed to assigning
values) than the secant method, so that the few more iterations possibly required by
the secant method are made up for by fewer total computations.
To see this, notice that Newton’s method always requires that both f ( xn−1 ) and
f ( xn−1 ) be computed for the nth term xn in the sequence. The secant method needs
′
f ( xn−1 ) and f ( xn−2 ) for the nth term, but a good programmer would save the value of
f ( xn−1 ) so that it could be re-used (and hence not re-computed) as f ( xn−2 ) in the next
iteration, resulting in potentially fewer total computations for the secant method.
There are occasional pitfalls in using Newton’s method. For example, if f ′ ( xn ) = 0
for some n ≥ 1 then Newton’s method fails, due to division by 0 in the algorithm. The
geometric reason is clear: the tangent line to the curve at that point would be parallel
to the x-axis and hence would not intersect it (assuming f ( xn ) 6= 0). There would be no
“next number” xn+1 in the iteration! See Figure 4.3.3(a) below.
y y
y = f ( x) y = f ( x)
f ( xn ) = 0
′
x x
x̄ xn x n −1 x̄ x0 x1 x2 ...
(a) f ′ ( x n ) = 0 (b) Moving away from a root
Figure 4.3.3 Newton’s method: Potential pitfalls
Another possible problem is that Newton’s method might move you away from the
root, i.e. not get closer, typically by a poor choice of x0 . See Figure 4.3.3(b) above. In
some extreme cases, it is possible that Newton’s method simply loops back and forth
endlessly between the same two numbers, as in Figure 4.3.4:
y y
y = f ( x)
y = f ( x)
x
x1 x̄ x0
x
x̄ x0 x1
Figure 4.3.4 Newton’s method: Infinite loop
In most cases a different choice for the initial guess x0 will fix such problems. Most
textbooks on the subject of numerical analysis discuss these issues.10
10 For example, R ALSTON, A. AND P. R ABINOWITZ , A First Course in Numerical Analysis, 2nd ed., New York:
McGraw-Hill, Inc., 1978.

There are conditions under which Newton’s method is guaranteed to work, and
convergence is fast. Newton’s method has a quadratic rate of convergence, mean-
ing roughly that the error terms—the differences between approximate roots and the
actual root—are being squared in the long term. More precisely, if the numbers xn for
n ≥ 0 converge to a root x̄, then the error terms ǫn = xn − x̄ for n ≥ 0 satisfy the limit
| ǫn+1 |
lim = C
n→∞ | ǫn |2
for some constant C . Squared error terms might sound like a bad thing, but the xn
terms are converging to the root, making the error terms closer to 0 for large n. Squar-
ing a number ǫn when | ǫn | < 1 results in a smaller number, not a larger one.
The numerical methods that you have learned will make it possible to sketch the
graphs of many more functions, since finding local minima and maxima involves find-
ing roots of f ′ , and finding inflection points involves finding roots of f ′′ . You now know
some methods for finding those roots.
Finally, despite being slower, the bisection method has the nice advantage of al-
ways working. The speed of modern computers makes the difference in algorithmic
efficiency negligible in many cases.
Exercises
A
1. Use Newton’s method to find the root of f ( x) = cos x − 2 x.
2. Use Newton’s method to find the positive root of f ( x) = sin x − x/2.
3. Use Newton’s method to find the solution of the equation e− x = x.
4. Use Newton’s method to find the solution of the equation e− x = x2 .
p
5. Use Newton’s method and f ( x) = x2 − 2 to approximate 2 accurate to six decimal places.
p
6. Use Newton’s method to approximate 3 accurate to six decimal places.
7. Repeat Exercise 1 with the secant method. 8. Repeat Exercise 3 with the secant method.
9. Repeat Exercise 5 with the secant method. 10. Repeat Exercise 6 with the secant method.
3
11. Cosmic microwave background radiation is described by a function similar to f ( x) = −1x+ e x for x ≥ 0.
Use Newton’s method to find the global maximum of f accurate to four decimal places.
12. Would a different choice for x0 in either graph in Figure 4.3.4 eliminate the infinite loop? Explain.
13. Draw a graph without any symmetry that has the infinite loop problem for Newton’s method.
14. The time t( p) required for a p-way merge of a file on a single disk drive into memory is
pa + m
t( p) = ,
ln p
where p > 1, a is the disk access time, and m is the time to read in one segment of the size of memory.
Find the integer closest to the value of p that minimizes t( p) when a = 24.3ms and m = 3500.3ms.
4.4 The Mean Value Theorem

The difference between instantaneous and average rates of change has been discussed
in earlier sections. Recall that there is no difference between the two for linear func-
tions. For nonlinear functions the average rate of change over an interval [a, b] of
positive length (i.e. b − a > 0) will not be the same as the instantaneous rate of change
at every point in the interval. However, the following theorem guarantees that they
will be the same at some point in the interval:
Mean Value Theorem: Let a and b be real numbers such that a < b, and suppose
that f is a function such that
(a) f is continuous on [a, b], and
(b) f is differentiable on (a, b).
Then there is at least one number c in the interval (a, b) such that
f ( b ) − f ( a)
f ′ ( c) = . (4.3)
b − a
Figure 4.4.1 below shows the geometric interpretation of the theorem:

y
slope = f ′ ( c) (b, f (b))

f ( b)
y = f ( x)
f (b) − f (a)
slope =
f (a) b−a
(a, f (a))
x
a c b
Figure 4.4.1 Mean Value Theorem: parallel tangent line and secant line
The idea is that there is at least one point on the curve y = f ( x) where the tangent
line will be parallel to the secant line joining the points (a, f (a)) and ( b, f ( b)). For each
f (b)− f (a)
c in (a, b) the tangent line has slope f ′ ( c), while the secant line has slope b−a . The
Mean Value Theorem says that these two slopes will be equal somewhere in (a, b).
To prove the Mean Value Theorem (sometimes called Lagrange’s Theorem), the fol-
lowing intermediate result is needed, and is important in its own right:
The Mean Value Theorem • Section 4.4 119
Rolle’s Theorem: Let a and b be real numbers such that a < b, and suppose that
f is a function such that
(a) f is continuous on [a, b],
(b) f is differentiable on (a, b), and
(c) f (a) = f ( b) = 0.
Then there is at least one number c in the interval (a, b) such that f ′ ( c) = 0.
Figure 4.4.2 on the right shows the geometric in- y

terpretation of the theorem. To prove the theorem, slope = f ′ ( c)
assume that f is not the constant function f ( x) = 0
for all x in [a, b] (if it were then Rolle’s Theorem y = f ( x)
would hold trivially). Then there must be at least
x
one x0 in (a, b) such that either f ( x0 ) > 0 or f ( x0 ) < 0. a c b
If f ( x0 ) > 0 then by the Extreme Value Theorem f
Figure 4.4.2
attains a global maximum at some x = c in the open
interval (a, b), since f is zero at the endpoints x = a and x = b of the closed interval
[a, b]. Then f ′ ( c) = 0 since f has a maximum at x = c. Likewise if f ( x0 ) < 0 then f
attains a global minimum at some x = c in (a, b), and thus again f ′ ( c) = 0 . X
The Mean Value Theorem can now be proved by applying Rolle’s Theorem to the
function
f ( b ) − f ( a)
F ( x) = f ( x) − f ( a) − ( x − a)
b−a
where f satisfies the conditions of the Mean Value Theorem. Basically, F “tilts” the
graph of f from Figure 4.4.1 to look like the graph in Figure 4.4.2. It is trivial to check
that F (a) = F ( b) = 0, and F is continuous on [a, b] and differentiable on (a, b) since f
is. Thus, by Rolle’s Theorem, F ′ ( c) = 0 for some c in (a, b). However,
f ( b ) − f ( a)
F ′ ( x) = f ′ ( x) −
b−a
and so
f ( b ) − f ( a) f ( b ) − f ( a)
F ′ ( c) = f ′ ( c) − = 0 ⇒ f ′ ( c) = X
b−a b−a
Note that both the Mean Value Theorem and Rolle’s Theorem are purely existence
theorems—they tell you only that a certain number exists. The task of finding the
numbers is left to you. For the Mean Value Theorem that task involves solving the
f (b)− f (a)
equation f ′ ( x) = b−a (or f ′ ( x) = 0 for Rolle’s Theorem). The numerical root-finding
methods from Section 4.3 could come in handy, since obtaining closed-form solutions
might be impossible. For that reason, the Mean Value Theorem is more useful for
theoretical purposes. One such application is the following important result:
If f is a differentiable function on an interval I such that f ′ ( x) = 0 for all x in I ,

then f is a constant function on I .
Note that I can be any interval, even the entire real line (−∞, ∞). It is already known
that f = constant ⇒ f ′ = 0; the above result says that the converse is true. The proof
is by contradiction: assume that f is not a constant function and show this contradicts
the Mean Value Theorem. If f is not constant then there exist numbers a < b in I such
that f (a) 6= f ( b). However, by the Mean Value Theorem there must exist a number c
in the interval (a, b) such that
f ( b ) − f ( a)
f ′ ( c) = .
b−a
Since the derivative of f is 0 everywhere in I , then f ′ ( c) = 0 and so
f ( b ) − f ( a)
= 0 ⇒ f ( b ) − f ( a) = 0 ⇒ f ( a) = f ( b ) ,
b−a
a contradiction of f (a) 6= f ( b). Thus, f must be a constant function. X
Another theoretical result can be proved with the Mean Value Theorem:
Let f be a differentiable function on an interval I . Then:

(a) If f ′ > 0 on I then f is increasing on I .
(b) If f ′ < 0 on I then f is decreasing on I .
To prove part (a), assume that f ′ ( x) > 0 for all x in I , and choose arbitrary numbers
a and b in I with a < b. To prove that f is increasing on I it suffices to show that
f (a) < f ( b). By the Mean Value Theorem there is a number c in (a, b) (and hence in I )
such that
f ( b ) − f ( a)
= f ′ ( c) , and so
b−a
f ( b ) − f ( a) = ( b − a) f ′ ( c ) > 0
since b − a > 0 and f ′ ( c) > 0. Thus, f ( b) > f (a), and so f is increasing on I. X
The proof of part (b) is similar and is left as an exercise. You might wonder why such
a proof is necessary. After all, an intuitive explanation was provided in Section 1.2 for
why positive or negative derivatives imply that a function is increasing or decreasing,
respectively. That knowledge has been assumed and used in the subsequent sections.
Intuitive so-called “hand-waving” explanations, in fact, often yield more insight than
a “formal” proof, such as the one above. However, it is good to know that such intuition
has a solid basis and can be proved, if needed.
The Mean Value Theorem can help in proving inequalities, often used in the sciences
for establishing upper or lower bounds on a quantity (e.g. worst-case scenario).
Example 4.16
Show that sin x ≤ x for all x ≥ 0.
Solution: The inequality holds trivially for x = 0, since sin 0 = 0 ≤ 0. So assume that x > 0. Then by the
Mean Value Theorem there is a number c in (0, x) such that for f ( x) = sin x,
f ( x) − f (0) sin x − sin 0
= f ′ ( c) ⇒ = cos c
x−0 x−0
⇒ sin x = x cos c
⇒ sin x ≤ x
since cos c ≤ 1 and x > 0. Note that sin x ≤ x is a sharper inequality than sin x ≤ 1 when 0 < x < 1.
There is a useful alternative form of the Mean Value Theorem. If a 0, so that a + h = b. Then any number c in (a, b) can be written as c = a + θ h
for some number θ in (0, 1). To see this, let c be in (a, b). Then 0 < c − a 0 be real numbers, and
suppose that f is a function such that
(a) f is continuous on [a, a + h], and
(b) f is differentiable on (a, a + h).
Then there is a number θ in the interval (0, 1) such that
f ( a + h ) − f ( a) = h f ′ ( a + θ h ) . (4.4)
The Mean Value Theorem is the special case of g( x) = x in the following generalization:
Extended Mean Value Theorem: Let a and b be real numbers such that a < b,
and suppose that f and g are functions such that
(a) f and g are continuous on [a, b],
(b) f and g are differentiable on (a, b), and
(c) g′ ( x) 6= 0 for all x in (a, b).

Then there is at least one number c in the interval (a, b) such that
f ′ ( c) f ( b ) − f ( a)
= . (4.5)
g′ ( c) g ( b ) − g ( a)
The Mean Value Theorem says that the derivative of a differentiable function will
always attain one particular value on a closed interval: the function’s average rate
of change over the interval. It turns out that the derivative will take on every value
between its values at the endpoints, similar to how the Intermediate Value Theorem
applies to continuous functions:11
Darboux’s Theorem: If f is a differentiable function on a closed interval [a, b]

then its derivative f ′ attains every value between f ′ (a) and f ′ ( b).
In other words, if f ′ (a) < γ < f ′ ( b) (or f ′ ( b) < γ < f ′ (a)) then there is a number c
in (a, b) such that f ′ ( c) = γ. If f ′ were continuous on [a, b] then the result would fol-
low trivially by the Intermediate Value Theorem for continuous functions. What is
perhaps surprising is that Darboux’s Theorem holds even for derivatives that are not
continuous. This means that a discontinuous derivative cannot have the type of sim-
ple jump discontinuities that would allow it to “skip” over intermediate values—the
points of discontinuity must be of a more complicated type. One rough interpretation
of Darboux’s Theorem is that even if a derivative is not a continuous function, it will
behave sort of as if it were.
Exercises
A
1. Does Rolle’s Theorem apply to the function f ( x) = 1 −| x | on the interval [−1, 1]? If so, find the number
in (−1, 1) that Rolle’s Theorem guarantees to exist. If not, explain why not.
2. Suppose that two horses run a race starting together and ending in a tie. Show that, at some time
during the race, they must have had the same speed.
¯ ¯ ¯ ¯
3. Use the ¯ ¯ ¯ ¯
¯ Mean Value¯Theorem
¯ ¯ show that sin A − sin B ≤ A − B for all A and B (in radians).
to
¯ ¯ ¯ ¯
Does sin A + sin B ≤ A + B for all A and B? Explain.
¯ ¯ ¯ ¯ ¯ ¯ ¯ ¯
4. Show that ¯ cos A − cos B ¯ ≤ ¯ A − B ¯ for all A and B (in radians). Does ¯ cos A + cos B ¯ ≤ ¯ A + B ¯
for all A and B? Explain.
5. Show that tan x ≥ x for all 0 ≤ x < π2 .
¯ ¯ ¯ ¯ ¡ ¢
6. Show that ¯ tan A − tan B ¯ ≥ ¯ A − B ¯ for all A and B (in radians) in − π2 , π2 . Can the inequality be
extended to all A and B? Explain your answer.
B
7. Use Rolle’s Theorem to show that for all constants a and b with a > 0, f ( x) = x3 − ax + b can not have
three positive roots. Also, show that it can not have three negative roots.
11 For a proof see pp.16-17 in O STROWSKI , A.M., Solution of Equations and Systems of Equations, 2nd ed., New
York: Academic Press Inc., 1966.

8. Use the Mean Value Theorem to show that if f ′ < 0 on an interval I then f is decreasing on I .
9. Suppose that f and g are continuous on [a, b] and differentiable on (a, b), and that f ′ ( x) > g′ ( x) for
all a < x < b. Show that f (b) − g(b) > f (a) − g(a).
10. Prove the Extended Mean Value Theorem, by applying Rolle’s Theorem to the function
f (b) − f (a)
F ( x) = f ( x) − f (a) − ( g( x) − g(a)) .
g(b) − g(a)
11. Show that e x ≥ 1 + x for all x. (Hint: Consider f ( x) = e x − x.)
12. Show that ln (1 + x) < x for all x > 0. 13. Show that tan−1 x < x for all x.
14. Show that for 0 < α ≤ β < π2 ,
β−α β−α
≤ tan β − tan α ≤ .
cos2 α cos2 β
15. Show that for 0 < a ≤ b,

b−a b b−a
≤ ln ≤ .
b a a
16. Show that for n > 1 and a > b,
nb n−1 (a − b) < an − b n < nan−1 (a − b) .
C
p b
17. Show that a2 + b < a +for all positive numbers a and b.
2a
¡ ¢ ¡ ¢
18. Show that f ( x) = cos2 x + cos2 π3 + x − cos x cos π3 + x is a constant function. What is its value?
19. Suppose that f ( x) is a differentiable function and that f (0) = 0 and f (1) = 1. Show that f ′ ( x0 ) = 2 x0
for some x0 in the interval (0, 1).
20. Prove the inequality ¯ ¯
¯ x1 + x2 ¯
¯ ¯ < 1 for −1 < x1 , x2 < 1
¯ 1+x x ¯
1 2
as follows:
(a) First prove the special case where x1 = x2 .
(b) For the case x1 < x2 define
x+a
f ( x) =
1 + ax
for −1 ≤ x ≤ 1, where −1 < a < 1. Show that f is increasing on [−1, 1], then use a = x2 and x = x1 .
Note that proving the case x2 < x1 is unnecessary (why?).
This inequality is a generalization of the same inequality for 0 ≤ x1 , x2 < 1 in the relativistic velocity
addition law from the theory of special relativity: if object 1 has velocity v1 relative to a frame of
reference F , and if object 2 has a velocity v2 relative to object 1, so that x1 = v1 / c and x2 = v2 / c
represent the fractions of the speed of light c at which the objects are moving, then the fraction of
the speed of light at which object 2 is moving with respect to F is x = ( x1 + x2 )/(1 + x1 x2 ). So it should
be true that 0 ≤ x < 1, since nothing can move faster than the speed of light.
CHAPTER 5
The Integral
5.1 The Indefinite Integral

Derivatives appear in many physical phenomena, such as the motion of objects. Recall,
for example, that given the position function s( t) of an object moving along a straight
line at time t, you could find the velocity v( t) = s′ ( t) and the acceleration a( t) = v′ ( t) of
the object at time t by taking derivatives. Suppose the situation were reversed: given
the velocity function how would you find the position function, or given the accelera-
tion function how would you find the velocity function?
d d
dt dt
s(t) v(t) a(t)
? ?
Figure 5.1.1 Differentiation vs antidifferentiation for motion functions
In this case calculating a derivative would not help, since the reverse process is
needed: instead of differentiation you need a way of performing antidifferentiation,
i.e. you would calculate an antiderivative.
An antiderivative F ( x) of a function f ( x) is a function whose derivative is f ( x).

In other words, F ′ ( x) = f ( x).
Differentiation is relatively straightforward. You have learned the derivatives of

many classes of functions (e.g. polynomials, trigonometric functions, exponential and
logarithmic functions), and with the various rules for differentiation you can calculate
derivatives of complicated expressions involving those functions (e.g. sums, powers,
products, quotients). Antidifferentiation, however, is a different story.
124
The Indefinite Integral • Section 5.1 125
To see some of the issues involved, consider a simple function like f ( x) = 2 x. Of

course you know that dx d
( x2 ) = 2 x, so it seems that F ( x) = x2 is the antiderivative of
f ( x) = 2 x. But is it the only antiderivative of f ( x)? No. For example, if F ( x) = x2 +1 then
F ′ ( x) = 2 x = f ( x), and so F ( x) = x2 + 1 is another antiderivative of f ( x) = 2 x. Likewise,
so is F ( x) = x2 + 2. In fact, any function of the form F ( x) = x2 + C , where C is some
constant, is an antiderivative of f ( x) = 2 x.
Another potential issue is that functions of the form F ( x) = x2 + C are just the most
obvious antiderivatives of f ( x) = 2 x. Could there be some other completely different
function—one that cannot be simplified into the form x2 + C —whose derivative also
turns out to be f ( x) = 2 x? The answer, luckily, is no:
Suppose that F ( x) and G ( x) are antiderivatives of a function f ( x). Then F ( x) and

G ( x) differ only by a constant. That is, F ( x) = G ( x) + C for some constant C .
To prove this, consider the function H ( x) = F ( x)−G ( x), defined for all x in the common
domain I of F and G . Since F ′ ( x) = G ′ ( x) = f ( x), then
H ′ ( x) = F ′ ( x) − G ′ ( x) = f ( x) − f ( x) = 0
for all x in I , so H ( x) is a constant function on I , as was shown in Section 4.4 on the
Mean Value Theorem. Thus, there is a constant C such that
H ( x) = C ⇒ F ( x) − G ( x) = C ⇒ F ( x) = G ( x) + C
for all x in I. X
The practical consequence of the above result can be stated as follows:
To find all antiderivatives of a function, it is necessary only to find one antideriva-

tive and then add a generic constant to it.
So for the function f ( x) = 2 x, since F ( x) = x2 is one antiderivative then all antideriva-

tives of f ( x) are of the form F ( x) = x2 +C , where C is a generic constant. Thus, functions
do not have just one antiderivative but a whole family of antiderivatives, all differing
only by a constant. The following notation makes all this easier to express:
The indefinite integral of a function f ( x) is denoted by

Z
f ( x) dx
and represents the entire family of antiderivatives of f ( x).
The large S-shaped symbol before f ( x) is called an integral sign.

126 Chapter 5 • The Integral §5.1
R
Though the indefinite integral f ( x) dx represents all antiderivatives of f ( x), the in-
tegral can be thought of as a single object or function in its own right, whose derivative
is f ′ ( x):
µZ ¶
d
f ( x) dx = f ( x)
dx
You might be wondering what the integral sign in the indefinite integral represents,
and why an infinitesimal dx is included. It has to do with what an infinitesimal rep-
resents: an infinitesimal “piece” of a quantity. For an antiderivative F ( x) of a function
f ( x), the infinitesimal (or differential) dF is given by dF = F ′ ( x) dx = f ( x) dx, and so
Z Z
F ( x) = f ( x) dx = dF .
The integral sign thus acts as a summation symbol: it sums up the infinitesimal
“pieces” dF of the function F ( x) at each x so that they add up to the entire function
F ( x). Think of it R
as similar to the usual summation symbol Σ used for discrete sums;
the integral sign takes the sum of a continuum of infinitesimal quantities instead.
Finding (or evaluating) the indefinite integral of a function is called integrating
the function, and integration is antidifferentiation.
Example 5.1
Z
Evaluate 0 dx.
R
Solution: Since the derivative of any constant function is 0, then 0 dx = C , where C is a generic
constant.
Note: From now on C will simply be assumed to represent a generic constant, without having to explic-
itly say so every time.
Example 5.2
Z
Evaluate 1 dx.
R
Solution: Since the derivative of F ( x) = x is F ′ ( x) = 1, then 1 dx = x + C .
Example 5.3
Z
Evaluate x dx.
x2
R x2
Solution: Since the derivative of F ( x) = 2 is F ′ ( x) = x, then x dx = 2 + C.
³ ´
d x n+1 1
Since dx n+1 = x for any number n 6= −1, and
n d
dx (ln | x |) = x = x−1 , then any power
of x can be integrated:
 n+1
Z x

if n 6= −1
Power Formula: x dx = n + 1
n


ln | x | if n = −1
The following rules for indefinite integrals are immediate consequences of the rules
for derivatives:
Let f and g be functions and let k be a constant. Then:

Z Z
1. k f ( x) dx = k f ( x) dx
Z Z Z
2. ( f ( x) + g( x)) dx = f ( x) dx + g( x) dx
Z Z Z
3. ( f ( x) − g( x)) dx = f ( x) dx − g( x) dx
The above rules are easily proved. For example, the first
R rule is a simple consequence
of the Constant Multiple Rule for derivatives: if F ( x) = f ( x) dx, then
Z Z
d d
( k F ( x)) = k (F ( x)) = k f ( x) ⇒ k f ( x) dx = k F ( x) = k f ( x) dx . X
dx dx
The other rules are proved similarly and are left as exercises. Repeated use of the
above rules along with the Power Formula shows that any polynomial can be inte-
grated term by term—in fact any finite sum of functions can be integrated in that
manner:
For any functions f 1 , . . ., f n and constants k 1 , . . ., k n ,

Z Z Z
( k 1 f 1 ( x) + · · · + k n f n ( x)) dx = k 1 f 1 ( x) dx + · · · + k n f n ( x) dx . (5.1)
Example 5.4
Z
Evaluate ( x7 − 3 x4 ) dx.
Solution: Integrate term by term, pulling constant multiple outside the integral:
Z Z Z
x8 3 x5
( x7 − 3 x4 ) dx = x7 dx − 3 x4 dx = − + C
8 5
Example 5.5
Z
p
Evaluate x dx.
Solution: Use the Power Formula:
Z Z
p x3/2 2 x3/2
x dx = x1/2 dx = + C = + C
3/2 3
Example 5.6
Zµ ¶
1 1
Evaluate + dx.
x2 x
Solution: Use the Power Formula and integrate term by term:
Zµ ¶ Zµ ¶
1 1 −2 1 x −1 1
+ dx = x + dx = + ln | x | + C = − + ln | x | + C
x2 x x −1 x
The following indefinite integrals are just re-statements of the corresponding deriva-
tive formulas for the six basic trigonometric functions:
Z
cos x dx = sin x + C
Z
sin x dx = − cos x + C
Z
sec2 x dx = tan x + C
Z
sec x tan x dx = sec x + C
Z
csc x cot x dx = − csc x + C
Z
csc2 x dx = − cot x + C
Since dx ( e ) = e ,
d x x
then:
Z
e x dx = e x + C
Example 5.7
Z
Evaluate (3 sin x + 4 cos x − 5 e x ) dx.
Solution: Integrate term by term:
Z Z Z Z
(3 sin x + 4 cos x − 5 e x ) dx = 3 sin x dx + 4 cos x dx − 5 e x dx
= −3 cos x + 4 sin x − 5 e x + C
Example 5.8
Recall from Section 1.1 the example of an object dropped from a height of 100 ft. Show that the height
s( t) of the object t seconds after being dropped is s( t) = −16 t2 + 100, measured in feet.
Solution: When the object is dropped at time t = 0 the only force acting on it is t = 0 sec
gravity, causing the object to accelerate downward at the known constant rate of
32 ft/s2 . The object’s acceleration a( t) at time t is thus a( t) = −32. If v( t) is the
t>0
object’s velocity at time t, then v′ ( t) = a( t), which means that 100 ft
Z Z s( t)
v( t) = a( t) dt = −32 dt = −32 t + C
for some constant C . The constant C here is not generic—it has a specific
value determined by the initial condition on the velocity: the object was at rest at time t = 0. That is,
v(0) = 0, which means
0 = v(0) = −32(0) + C = C ⇒ v( t) = −32 t
for all t ≥ 0. Likewise, since s′ ( t) = v( t) then
Z Z
s( t) = v( t) dt = −32 t dt = −16 t2 + C
for some constant C , determined by the initial condition that the object was 100 ft above the ground at
time t = 0. That is, s(0) = 100, which means
100 = s(0) = −16(0)2 + C = C ⇒ s( t) = −16 t2 + 100
for all t ≥ 0. X
The formula for s( t) in Example 5.8 can be generalized as follows: denote the ob-
ject’s initial position at time t = 0 by s 0 , let v0 be the object’s initial velocity (positive
if thrown upward, negative if thrown downward), and let g represent the (positive)
constant acceleration due to gravity. By Newton’s First Law of motion the only accel-
eration imparted to the object after throwing it is due to gravity:
Z Z
a( t ) = − g ⇒ v( t ) = a( t) dt = − g dt = − gt + C
for some constant C : v0 = v(0) = − g(0) + C = C . Thus, v( t) = − gt + v0 for all t ≥ 0, and so

Z Z
s( t ) = v( t) dt = (− gt + v0 ) dt = − 12 gt2 + v0 t + C
for some constant C : s 0 = s(0) = − 21 g(0)2 + v0 (0) + C = C . To summarize:
Free fall motion: At time t ≥ 0:
acceleration: a( t ) = − g
velocity: v( t) = − gt + v0
position: s( t) = − 12 gt2 + v0 t + s 0
initial conditions: s 0 = s(0) , v0 = v(0)
Note that the units are not specified—they just need to be consistent. In metric units,
g = 9.8 m/s2 , while g = 32 ft/s2 in English units.
Thinking of an indefinite integral as the sum of all the infinitesimal “pieces” of a
function—for the purpose of retrieving that function—provides a handy way of inte-
grating a differential equation to obtain the solution. The key idea is to transform the
differential equation into an equation of differentials, which has the effect of treating
functions as variables. Some examples will illustrate the technique.
Example 5.9
dy
For any constant k, show that every solution of the differential equation dt = k y is of the form y = Ae kt
for some constant A . You can assume that y( t) > 0 for all t.
Solution: Put the y terms on the left and the t terms on the right, i.e. separate the variables:
dy
= k dt
y
Now integrate both sides (notice how the function y is treated as a variable):
Z Z
dy
= k dt
y
ln y + C1 = kt + C2 (C1 and C2 are constants)

ln y = kt + C (combine C1 and C2 into the constant C )
y = e kt+C = e kt · e C = Ae kt
where A = e C is a constant. Note that this is the formula for radioactive decay from Section 2.3.
Example 5.10
Recall from Section 3.6 the equation of differentials
dP dV dT
+ =
P V T
relating the pressure P , volume V and temperature T of an ideal gas. Integrate that equation to obtain
the original ideal gas law PV = RT , where R is a constant. .
Solution: Integrating both sides of the equation yields

Z Z Z
dP dV dT
+ =
P V T
ln P + ln V = ln T + C (C is a constant)
ln (PV ) = ln T + C
PV = eln T +C = eln T · e C = T e C = RT
where R = e C is a constant. X
The integration formulas in this section depended on already knowing the deriva-
tives of certain functions and then “working backward” from their derivatives to obtain
the original functions. Without that prior knowledge you would be reduced to guess-
ing, or perhaps recognizing a pattern from some derivative you have encountered. A
number of integration techniques will be presented shortly, but there are many indef-
R 2 R
inite integrals for which no simple closed form exists (e.g. e x dx and sin( x2 ) dx).
Exercises
A
For Exercises 1-15, evaluate the given indefinite integral.
Z Z Z
¡ 2 ¢
1. x + 5 x − 3 dx 2. 3 cos x dx 3. 4 e x dx
Z Z Z
¡ 5 ¢ 3ex
4. x − 8 x4 − 3 x3 + 1 dx 5. 5 sin x dx 6. dx
Z Z Z 5
6 4 ¡ p ¢
7. dx 8. dx 9. −2 x dx
Zx 3
Z ³x Z
1 ´ 1
10. p dx 11. x + x4/3 dx 12. p dx
Z3 x 3 3
Z Z x
13. 3 sec x tan x dx 14. 5 sec2 x dx 15. 7 csc2 x dx
R R R
16. Prove the sum and difference rules for indefinite integrals: ( f ( x) ± g( x)) dx = f ( x) dx ± g( x) dx
17. Integrate both sides of the equation
dP dM dT
+ =
P M 2T
PM
to obtain the ideal gas continuity relation: p = constant.
T
18. Use the free fall motion equation for position to show that the maximum height reached by an
v02
object launched straight up from the ground with an initial velocity v0 is 2g .
5.2 The Definite Integral

Recall from the last section that the integral sign in the indefinite integral
Z
f ( x) dx
represents a summation of the infinitesimals f ( x) dx = dF for an antiderivative F ( x) of

f ( x). Why is the term “indefinite” used? Because the summation is indefinite: the x
in f ( x) dx is defined generically, meaning “ x in general,” that is, not for x in a specific
range of values. The same summation over a specific, definite range of values of x, say,
over an interval [a, b], is a different type of integral:
The definite integral of a function f ( x) over an interval [a, b] is denoted by

Zb
f ( x) dx
a
and represents the sum of the infinitesimals f ( x) dx for all x in [a, b].
An indefinite integral yields a generic function, whereas a definite integral yields

either a number or a specific function. There are many ways to calculate the specific
summation in a definite integral, one of which is motivated by a geometric interpreta-
tion of the infinitesimal f ( x) dx as the area of a rectangle, as in Figure 5.2.1 below:
y
y = f ( x)
f ( x + dx) − f ( x)
f ( x)
x
a x x + dx b
dx
Figure 5.2.1 The infinitesimal f ( x) dx as the area of a rectangle
The shaded rectangle in the above picture has height f ( x) and width dx, and so its
area is f ( x) dx. In fact, it appears that that area is just a little bit smaller than the
area under the curve y = f ( x) and above the x-axis between x and x + dx; there is a
small gap between the curve and the top of the rectangle, accounting for the difference
in the area. However, the area of that gap turns out to be zero, as shown below:
The Definite Integral • Section 5.2 133
B
y = f ( x)
df = f ( x + dx) − f ( x)
A
C
f ( x)
x x + dx
dx
Figure 5.2.2 Area under the curve y = f ( x) over [ x, x + dx]
By the Microstraightness Property, the curve y = f ( x) shown in Figure 5.2.1 is a

straight line over the infinitesimal interval [ x, x + dx], as shown in Figure 5.2.2.1 Thus,
the part of the area between the curve and the x-axis over the interval [ x, x + dx] con-
sists of two parts: the area f ( x) dx of the shaded rectangle and the area of the right
triangle △ ABC , both of which are shown in Figure 5.2.2. However, the area of △ ABC
is zero:
1 1 1 1
Area of △ ABC = (base) × (height) = ( dx)( df ) = ( dx)( f ′ ( x) dx) = f ′ ( x)( dx)2 = 0
2 2 2 2
The function f shown in Figure 5.2.2 is increasing at x, but a similar argument could
be made if f were decreasing at x. Hence, the area between the curve y = f ( x) and the
x-axis comes solely from the rectangles with area f ( x) dx, as x varies from a to b. The
sum of all those rectangular areas, though, equals the definite integral of f ( x) over
[a, b]. The definite integral can thus be interpreted as an area:
For a function f ( x) ≥ 0 over [a, b], the area under the curve y = f ( x) between
x = a and x = b, denoted by A , is given by
Zb
A = f ( x) dx
a
and represents the area of the region R bounded above by y = f ( x), bounded below
by the x-axis, and bounded on the sides by x = a and x = b (with a < b).
1 The function f is assumed to be differentiable at x, in this case. If not then the points where f is not differentiable
can be excluded without affecting the integral.

y = f ( x)
x
a b
Rb
Figure 5.2.3 The area A of the region R equals a f ( x) dx
In Figure 5.2.3 the area under the curve y = f ( x) between x = a and x = b is the
Rb
area A of the shaded region R , namely A = a f ( x) dx. To calculate that area for a
specific function, rectangles can again be used, but this time with widths that are
small positive numbers instead of infinitesimals. The procedure is as follows:
1. Create a partition P = { x0 < x1 < · · · < xn−1 < xn } of the interval [a, b] into n ≥ 1
subintervals [ x0 , x1 ], [ x1 , x2 ], . . ., [ xn−1 , xn ], with x0 = a and xn = b.
2. In each subinterval [ x i−1 , x i ] of P pick a number x∗i , so that x i−1 ≤ x∗i ≤ x i for i = 1
to n.
3. For i = 1 to n, form a rectangle whose base is the subinterval [ x i−1 , x i ] of length

∆ x i = x i − x i−1 > 0 and whose height is f ( x∗i ).
4. Take the sum f ( x1∗ )∆ x1 + f ( x2∗ )∆ x2 + · · · + f ( x∗n )∆ xn of the areas of these rectangles,
called a Riemann sum.
y
y = f ( x)
f ( x1∗ ) f ( x∗ ) ... f ( x∗n )

2
x
a = x0 x1∗ x1 x2∗ x2 ... x n −1 x∗n x n = b
∆ x1 ∆ x2 ∆xn
Figure 5.2.4 A Riemann sum of f ( x) over a partition of [a, b]
5. Take the limit of the Riemann sums as n → ∞, so that the subinterval lengths
approach 0. If the limit exists then that limit is the area A of the region R :
Zb Xn
Area A = f ( x) dx = lim f ( x∗i ) ∆ x i (5.2)
a n→∞
i=1
The limit in Formula (5.2) should be taken over all partitions whose norm—the
length of the largest subinterval—approaches 0. In practice, however, the partitions
are usually chosen so that the subintervals are of equal length, and then simply make
those equal lengths smaller and smaller by dividing the interval [a, b] into more and
more such subintervals. Note that the points x∗i in each subinterval can be anywhere
in the subinterval—often the midpoint of the subinterval is chosen, but the left and
right endpoints are also typical choices.
In the above procedure the gaps between the rectangles and the curve will have ar-
eas approaching 0 as the number n of subintervals grows and the subinterval lengths
approach 0. This is true if the function f is differentiable, and in fact even if f is merely
continuous.2 Thus, the area under the curve can be defined by the above procedure.
To calculate the area under a curve in this manner, the reader should have some
familiarity with the summation notation in Formula (5.2).
For real numbers a 1 , a 2 , . . ., a n and an integer n ≥ 1,

X
n
a k = a1 + a2 + · · · + a n
k=1
is the sum of a 1 , . . ., a n . The symbol Σ is called the summation sign, which is the
Greek capital letter Sigma.
The following rules for this “Sigma notation” are intuitively obvious:
Let a 1 , a 2 , . . ., a n , and b 1 , b 2 , . . ., b n be real numbers, and let c be a constant. Then:

Xn Xn Xn
(1) (a k + b k ) = ak + bk
k=1 k=1 k=1
X
n X
n X
n
(2) (a k − b k ) = ak − bk
k=1 k=1 k=1
X
n X
n
(3) ca k = c ak
k=1 k=1
X
n X
n
(4) ak = ai (i.e. the sum is independent of the summation index letter)
k=1 i=1
2 For a proof and fuller discussion of all this, see Ch.1-2 in K NOPP, M.I., Theory of Area, Chicago: Markham
Publishing Co., 1969. The book attempts to define precisely what an “area” actually means, including that of a
rectangle (showing agreement with the intuitive notion of width times height).
The following summation formulas can be helpful when calculating Riemann sums:
Let n ≥ 1 be a positive integer. Then:

Xn
(1) 1 = n
k=1
n
X n ( n + 1)
(2) k = 1 + 2 + ··· + n =
k=1 2
n
X n ( n + 1) (2 n + 1)
(3) k 2 = 12 + 22 + · · · + n 2 =
k=1 6
X
n n2 ( n + 1)2
(4) k 3 = 13 + 23 + · · · + n 3 =
k=1 4
X
n n ( n + 1) (6 n3 + 9 n2 + n − 1)
(5) k 4 = 14 + 24 + · · · + n 4 =
k=1 30
Formula (1) is obvious: add the number 1 a total of n times and the sum is n.
Formula (2) can be proved by induction:
X
n n ( n + 1)
1. Show that k = for n = 1:
k=1 2
1
X 1 (1 + 1)
k = 1 = X
k=1 2
X
nn ( n + 1)
2. Assume that k = for some integer n ≥ 1. Show that the formula holds
k=1 2
for n replaced by n + 1, that is:
+1
nX ( n + 1) (( n + 1) + 1) ( n + 1) ( n + 2)
k = =
k=1 2 2
To show this, note that
+1
nX X
n
k = 1 + 2 + · · · + n + ( n + 1) = k + ( n + 1)
k=1 k=1
n ( n + 1) n ( n + 1) + 2( n + 1) ( n + 1) ( n + 2)
= + ( n + 1) = = X
2 2 2
3. By induction, this proves the formula for all integers n ≥ 1. QED
Formulas (3)-(5) can be proved similarly by induction (see the exercises). The example
below shows how Formulas (2) and (3) are used in finding the limit of a Riemann sum.
Example 5.11
Z2
Use Riemann sums to calculate x2 dx.
1
Solution: The definite integral is the area under the curve y = f ( x) = x2 between x = 1 and x = 2, as
shown in Figure 5.2.5(a):
y y
...
x x
0 1 2 0 1 = x0 x1 x2 ... x n−1 x n = 2
(a) Area under y = x2 over [1, 2] (b) Riemann sums using left endpoints: x∗i = x i−1
R2
Figure 5.2.5 Calculating 1 x2 dx
Divide the interval [1, 2] into n subintervals of equal length ∆ x i = (2 − 1)/n = 1/n for i = 1 to n, so that
the partition P is { x0 < x1 < ... x n } where x i = 1 + ni for i = 0, 1, ..., n (and hence x0 = 1 and x n = 2). In
each subinterval [ x i−1 , x i ] pick the point x∗i to be the left endpoint x i−1 , so that the rectangles appear as
in Figure 5.2.5(b). Then
Z2 X n X n 1 X n 1
x2 dx = lim f ( x∗i ) ∆ x i = lim f ( x i−1 ) = lim x2i−1
1 n →∞
i =1
n →∞
i =1 n n →∞
i =1 n
µ ¶ n µ1 ¶
n
X i −1 2 1 X 2 1
= lim 1+ = lim + 2 ( i − 1) + 3 ( i − 1)2
n→∞
i =1 n n n→∞
i =1 n n n
Ã ! Ã !
X n 1 2 Xn 1 Xn −1
2 nX −1
1 nX
2 2
= lim + 2 ( i − 1) + 3 ( i − 1) = lim 1 + 2 i + 3 i
i =1 n n i =1 n i =1 n i =1 n i =1
n→∞ n→∞
µ ¶
2 (n − 1)n 1 (n − 1)n(2n − 1)
= lim 1+ 2· + 3· (replace n by n − 1 in Formulas (2) and (3))
n→∞ n 2 n 6
³ ´ µ ¶ µ ¶
n−1 2n2 − 3n + 1
= lim 1 + lim + lim
n→∞ n→∞ n n→∞ 6n2
1 2 7
= 1 + + =
1 6 3
It it often simpler to use a computer to calculate approximations of a definite inte-

gral, by taking the Riemann sum of a sufficiently large number of rectangles in order
to achieve the desired accuracy. Choosing subintervals of equal length, as in Example
5.11, makes it easier to use an algorithm to calculate the integral.
For example, the table below summarizes the calculations of Riemann sums for the
function in Example 5.11—namely f ( x) = x2 over [1, 2]—using different values for the
points x∗i in the subintervals (left endpoints, midpoints, and right endpoints):
# of rectangles Left endpoint Midpoint Right endpoint

1 1 2.25 4
2 1.625 2.3125 3.125
3 1.851851851852 2.324074074074 2.851851851852
4 1.96875 2.328125 2.71875
5 2.04 2.33 2.64
10 2.185 2.3325 2.485
100 2.31835 2.333325 2.34835
1000 2.3318335 2.33333325 2.3348335
10000 2.333183335 2.3333333325 2.333483335
100000 2.33331833335 2.333333333325 2.33334833335
1000000 2.333331833333 2.333333333333 2.333334833333
Due to the concavity of the curve y = x2 , using the left endpoints underestimates
the actual area, whereas using the right endpoints yields an overestimate. Using the
midpoints usually gives better results (i.e. more accuracy in fewer iterations).
So far only definite integrals of nonnegative functions have been considered—that
is, functions f ( x) ≥ 0 over an interval [a, b]. If f ( x) is either negative or changes sign
over [a, b], then the definite integral can be defined as follows:
Let R be the region bounded by y = f ( x) and the x-axis between x = a and x = b. If

f ( x) ≤ 0 over [a, b], then
Zb
f ( x) dx = the negative of the area of R
a
If f ( x) changes sign over [a, b], then

Zb
f ( x) dx = the net area of R ,
a
where the parts of R above the x-axis count as positive area and the parts below
count as negative area.
Zb
Note: In the definite integral f ( x) dx the numbers a and b are called the limits
a
of integration, with a being the lower limit of integration and b the upper limit
of integration. The function f ( x) being integrated is called the integrand, in both
definite and indefinite integrals.
Exercises
A
Zb
1. Explain why c dx = c(b − a) for any constant c.
a
R2
2. Would using left endpoints in the Riemann sums underestimate or overestimate 1 ln x dx? Explain.
B
Z1 Z1
3. Use Riemann sums to calculate x dx. 4. Use Riemann sums to calculate x2 dx.
0 0
Z1 Z1
2
5. Use Riemann sums to calculate 3 x dx. 6. Use Riemann sums to calculate x3 dx.
0 0
n
X 2 n (n + 1) (2n + 1)
7. Prove the formula k = by induction on n ≥ 1.
k =1 6
X n (n + 1) (2n + 1)
n
8. Prove the formula k2 = as follows:
k =1 6
n ¡
X ¢
(a) Show that (k + 1)3 − k3 = (n + 1)3 − 1 .
k =1
(b) Show that (k + 1)3 − k3 = 3k2 + 3k + 1 .
X n n (n + 1) X n n (n + 1) (2n + 1)
(c) Use the formula k = and parts (a) and (b) to show that k2 = .
k =1 2 k =1 6
n
X n2 (n + 1)2
9. Prove the formula k3 = by induction on n ≥ 1.
k =1 4
10. The famous quicksort algorithm in computer science is a popular method for placing objects in
some order (e.g. numerical, alphabetical). On average the algorithm needs O(n log n) comparisons
to sort n objects (here log n means the natural logarithm of n). The proof of that average complexity
depends on the inequality
mX−1 Zm
k ln k ≤ x ln x dx
k =2 2
for all integers m > 2. Explain why that inequality is true.
C
n
X n (n + 1) (6n3 + 9n2 + n − 1)
11. Prove the formula k4 = 14 + 24 + · · · + n4 = by induction on n ≥ 1.
k =1 30
12. Calculate the following sum:
1 + (1 + 2) + (1 + 2 + 3) + (1 + 2 + 3 + 4) + · · · + (1 + 2 + 3 + 4 + · · · + 50)
5.3 The Fundamental Theorem of Calculus

Using Riemann sums to calculate definite integrals can be tedious, as was seen in the
previous section. In fact the technique shown in that section depended on the function
being a low-degree polynomial, which obviously will not always be the case. Luckily
there is a better way, involving antiderivatives, given by the following theorem:
Fundamental Theorem of Calculus: Suppose that a function f is differentiable

on [a, b]. Then:
(I) The function A ( x) defined on [a, b] by
Zx
A ( x) = f ( t) dt
a
is differentiable on [a, b], and
A ′ ( x) = f ( x)
for all x in [a, b].
(II) If F is an antiderivative of f on [a, b], i.e. F ′ ( x) = f ( x) for all x in [a, b], then
Zb
f ( x) dx = F ( b) − F (a) .
a
The function A ( x) in Part I of the theorem is sometimes called the area function
because it represents the area under the curve y = f ( x) over the interval [a, x], as
shown in Figure 5.3.1 below.
y y
y = f ( x) y = f ( x)
A ( x) dA
x x
a x b a x x + dx b
Rx
Figure 5.3.1 The area function A ( x) = a f ( t) dt Figure 5.3.2 dA = A ( x + dx) − A ( x)
To prove Part I, assume that f ( x) ≥ 0 on [a, b] as in Figure 5.3.1 (the proofs for f ( x)
either negative or switching sign over [a, b] are similar). The goal is to show that for
any x in [a, b] the differential dA exists and equals f ( x) dx. First, dA = A ( x + dx) − A ( x)
is the area under the curve y = f ( x) over the interval [ x, x + dx], as shown in Figure
5.3.2 above.
The Fundamental Theorem of Calculus • Section 5.3 141
By the Microstraightness Property f is a straight line over the infinitesimal interval

[ x, x + dx], so f must be either increasing, constant, or decreasing over that interval.
The three possibilities are shown in Figure 5.3.3:
B A
y = f ( x)
y = f ( x)
df −df
A y = f ( x) B
C C
f ( x)
f ( x) f ( x) f ( x) f ( x + dx)
x x + dx x x + dx x x + dx
dx dx dx
(a) f is increasing (b) f is constant (c) f is decreasing
Figure 5.3.3 The three possibilities for dA
In the case where f is increasing over [ x, x + dx], the infinitesimal area dA is the sum
of the area of the rectangle of height f ( x) and width dx and the area of the right trian-
gle △ ABC shown in Figure 5.3.3(a). The area of △ ABC is 12 ( df )( dx) = 12 f ′ ( x)( dx)2 = 0,
so dA = f ( x) dx.
In the case where f is constant over [ x, x + dx], the infinitesimal area dA is the area
of the rectangle of height f ( x) and width dx, as shown in Figure 5.3.3(b). So again,
dA = f ( x) dx.
In the case where f is decreasing over [ x, x + dx], the infinitesimal area dA is the
sum of the area of the rectangle of height f ( x + dx) and width dx and the area of the
right triangle △ ABC shown in Figure 5.3.3(c). Note that df < 0 since f is decreasing,
and so the area of △ ABC is 21 (− df )( dx) = − 21 f ′ ( x)( dx)2 = 0. Thus,
dA = f ( x + dx) dx = ( f ( x) + df ) dx = f ( x) dx + f ′ ( x)( dx)2 = f ( x) dx + 0 = f ( x) dx .
dx = f ( x), which shows that A ( x) is

So in all three cases, dA = f ( x) dx, and so A ′ ( x) = dA
differentiable and has derivative f ( x). This proves Part I of the Fundamental Theorem
of Calculus. X
To prove
Rx Part II of the theorem, let F ( x) be an antiderivative of f ( x) over [a, b]. Since
A ( x) = a f ( x) dx is also an antiderivative of f ( x) over [a, b] by Part I of the theorem,
then A ( x) and F ( x) differ by a constant C over [a, b]. In other words:
A ( x) = F ( x) + C for all x in [a, b]

By definition A (a) = 0, since it is the area under the curve over the interval [a, a] of
zero length. Thus,
0 = A ( a) = F ( a) + C ⇒ C = − F ( a) ⇒ A ( x) = F ( x) − F ( a) for all x in [a, b]
and so Zb
f ( x) dx = A ( b) = F ( b) − F (a)
a
which proves Part II of the theorem.3 X
Note: In some textbooks Part I is called the First Fundamental Theorem of Calculus
and Part II is called the Second Fundamental Theorem of Calculus. The following
notation provides a shorthand way of writing F ( b) − F (a):
¯b
¯
F ( x) ¯¯ = F ( b) − F (a)
a
Example 5.12
Z2
Calculate x2 dx.
1
Solution: Recall from Example 5.11 in the previous section that the integral equals 7/3. In that example
Riemann sums were used, but Part II of the Fundamental Theorem of Calculus makes the integral much
3
easier to calculate. Since F ( x) = x3 is an antiderivative of f ( x) = x2 , then
Z2 ¯2
2 x3 ¯¯ 23 13 7
x dx = ¯ = − = .
1 3 1 3 3 3
Note in the above example that any antiderivative of f ( x) = x2 could have been used,
3
e.g. F ( x) = x3 + 5. Notice that the constant 5 would have been canceled out when evalu-
ating F (2) − F (1). So you do not need to add a generic constant C to the antiderivative
of f ( x) in a definite integral, as you would in an indefinite integral.
Example 5.13
Zπ
Calculate sin x dx.
0
Solution: Since F ( x) = − cos x is an antiderivative of f ( x) = sin x, then
Zπ ¯π
¯
sin x dx = − cos x ¯¯ = − cos π − (− cos 0) = −(−1) − (−1) = 2 .
0 0
3 The theorem can be proved for the weaker condition that f is merely continuous on [a, b]. See p.173-175 in
PARZYNSKI , W.R. AND P.W. Z IPSE, Introduction to Mathematical Analysis, New York: McGraw-Hill, Inc., 1982.
Example 5.14
Z1
Calculate x3 dx.
−1
x4
Solution: Since F ( x) = 4 is an antiderivative of f ( x) = x3 , then
Z1 ¯1
x4 ¯¯ 14 (−1)4 1 1
x3 dx = ¯ = − = − = 0.
−1 4 −1 4 4 4 4
Example 5.14 is a special case of the following result for odd functions:
If f is an odd function, i.e. f (− x) = − f ( x) for all x, then

Za
f ( x) dx = 0
−a
for all a > 0 such that f is continuous on [−a, a].
The idea is that since an odd function is symmetric around the origin, then the area
between the curve and the x-axis over [0, a] will cancel out the area between the curve
and the x-axis over [−a, 0]. Both areas are the same but one gets counted as positive
and the other negative, as shown in Figure 5.3.4 below:
y y
y = f ( x) y = f ( x)
−a +
x
a
—
x
−a a
Figure 5.3.4 Odd function f over [−a, a] Figure 5.3.5 Even function f over [−a, a]
By symmetry around the y-axis, a similar result holds for even functions (see Figure
5.3.5):
If f is an even function, i.e. f (− x) = f ( x) for all x, then

Za Za
f ( x) dx = 2 f ( x) dx
−a 0
for all a > 0 such that f is continuous on [−a, a].

The following rules for definite integrals are a consequence of the corresponding
rules for indefinite integrals:
Let f and g be continuous functions on [a, b] and let k be a constant. Then:

Zb Zb
1. k f ( x) dx = k f ( x) dx
a a
Zb Zb Zb
2. ( f ( x) + g( x)) dx = f ( x) dx + g( x) dx
a a a
Zb Zb Zb
3. ( f ( x) − g( x)) dx = f ( x) dx − g( x) dx
a a a
The following results for definite integrals are a consequence of the Fundamental
Theorem of Calculus:
Let f be a continuous function on [a, b] and suppose that a < c < b. Then:
Za
(1) f ( x) dx = 0
a
Za Zb
(2) f ( x) dx = − f ( x) dx
b a
Zb Zc Zb
(3) f ( x) dx = f ( x) dx + f ( x) dx
a a c
For example, if F ( x) is an antiderivative of f ( x) on [a, b], then

Zc Zb Zb
f ( x) dx + f ( x) dx = (F ( c) − F (a)) + (F ( b) − F ( c)) = F ( b) − F (a) = f ( x) dx
a c a
which proves rule (3).
The following result is a consequence of Part I of the Fundamental Theorem of Cal-
culus along with the Chain Rule:
Chain Rule for integrals: Let f be a continuous function on an interval I con-

taining x = a, and let g( x) be a differentiable function on I . If
Z g ( x)
F ( x) = f ( t) dt for all x in I
a
then F ′ ( x) = f ( g( x)) · g′( x) for all x in I .

Example 5.15
Z x2
2
Let F ( x) = e− t dt for all x > 0. Find F ′ ( x).
0
2
Solution: By the Chain Rule for integrals, with f ( t) = e t and g( x) = x2 :
2 )2 4
F ′ ( x) = f ( g( x)) · g′ ( x) = e−( x · (2 x) = 2 x e− x
Exercises
A
For Exercises 1-12, evaluate the given definite integral.
Z1 Z1 Z1 Z1
1. x2 dx 2. x2 dx 3. x3 dx 4. ( x2 + 3 x − 4) dx
0 −1 0 −1
Z2 Z3 Zπ/2 Z1
1 1
5. 2
dx 6. 3
dx 7. cos x dx 8. e x dx
1 x 2 x 0 0
Z1 Zπ Z4 Z2 3 x 2
x p x e
9. 2 e dx 10. sin x dx 11. x dx 12. dx
−1 −π 0 −2 cos 2x
Zx Z Z
1 a b
13. Show that ln x = dt for all x > 0. 14. Show that f ( x) dx = − f ( x) dx.
1 t b a
Zx
t
15. Given f ( x) = p dt , find f ′ (3) and f ′ (−2).
1 4
t +1
B
16. Prove the Chain Rule for integrals.
¯ Zb ¯ Zb
¯ ¯
¯
17. Explain why for any continuous function f on [a, b], ¯ f ( x) dx ¯¯ ≤ | f ( x) | dx .
a a
Zb Zb
18. Explain why if f ( x) ≤ g( x) on [a, b] then f ( x) dx ≤ g( x) dx .
a a
C
19. Show that if f ( x) is continuous on [a, b] then there is a number c in (a, b) such that
Zb
f ( x) dx = f ( c) · (b − a) .
a
20. Let f ( t) be a continuous function for all t ≥ 0, and for each x ≥ 0 define a function g( x) by
Zx
g( x ) = ( x − t) f ( t) dt .
0
Zx
Show that g′ ( x) = f ( t) dt for all x ≥ 0.
0
21. Show that for all x > 0,
Zx Z1/ x
dt dt
+
0 1 + t2 0 1 + t2
is independent of x.
5.4 Integration by Substitution

The integrals encountered so far—whether indefinite or definite—have been the sim-
plest
R kind, since the antiderivatives were givenRby known formulas. For example,
cos x dx = sin x + C . What if the integral were cos 2 x dx instead? No formula has
been discussed yet for this integral, and the answer is not sin 2 x + C , since the deriva-
tive of sin 2 x is 2 cos 2 x, not cos 2 x. But
R dividing sin 2 x by 2 first and then taking the
derivative would yield cos 2 x, so that cos 2 x dx = 21 sin 2 x + C .
Evaluating an integral in such a manner is often done when the function is not
too complicated, as the one above. Usually it will not be quite that simple, and so
a general technique called substitution is needed. The idea behind substitution is
to replace part of the function being integrated by a new variable—typically u—so
that a complicated function of x is now a simpler function of u that you know how to
integrate.
Example 5.16
Z
Evaluate cos 2 x dx by substitution.
Solution: The 2 x in the cosine function is what makes this integral unknown, so replace it by u: let
u = 2 x. The integral is now Z
cos u dx
which is a problem because the point of doing substitution is to eliminate all references to the variable
x, including in the infinitesimal dx. The entire integral needs to be in terms of u and du, but the dx is
still there. So put dx in terms of du:
1
u = 2x ⇒ du = 2dx ⇒ dx = du
2
The integral now becomes
Z µ ¶ Z
1 1 1
cos u du = cos u du = sin u + C
2 2 2
by the formula already known, just with the letter u as the variable instead of x. The original integral
was in terms of x, so the final answer—for an indefinite integral—should also be in terms of x. Thus,
the final step is to substitute back into the answer what u equals in terms of x, namely 2 x:
Z
1 1
cos 2 x dx = sin u + C = sin 2 x + C
2 2
If the procedure in the above example seems similar to making a substitution when
using the Chain Rule to take a derivative, that is because it is similar: you are basi-
cally doing the same thing only in reverse. Just as for differentiation, it is not always
obvious what part of the function is the best candidate for substitution when perform-
ing integration. There is one obvious rule: never make the substitution u = x, because
that changes nothing.
Integration by Substitution • Section 5.4 147
Example 5.17
Z
Evaluate e−3 x dx.
Solution: The −3 x in the exponential function is what makes this integral unknown, so make the
substitution u = −3 x, which means that du = −3dx, and so dx = − 13 du. Thus:
Z Z µ ¶ Z
1 1 1 1
e−3 x dx = e u − du = − e u du = − e u + C = − e−3 x + C
3 3 3 3
The above example can be generalized as follows:

Z
1 ax
e ax dx = e + C for any constant a 6= 0
a
Example 5.17 was the special case with a = −3.

Example 5.18
Z
Evaluate (1 + 4 x)5 dx.
Solution: You might be tempted to make the substitution u = 4 x, but that would then require finding
the integral of (1 + u)5 , for which there is not yet any formula. But there is a formula for the integral of
u5 . Hence, let u = 1 + 4 x, so that du = 4dx ⇒ dx = 41 du. Thus:
Z Z
5 1 1 u6 1
(1 + 4 x) dx = u5 du = + C = (1 + 4 x)6 + C
4 4 6 24
Example 5.19
Z
2
Evaluate 2 x e x dx.
Solution: It might be unclear whether you should make the substitution u = 2 x or u = x2 , but the hint
here is that the derivative of the x2 inside the exponential function is 2 x, which appears outside the
exponential function. Indeed, you could check that letting u = 2 x would result in an integral no simpler
R 2
than the current one (namely, 12 u e u /4 du). So let u = x2 , which means du = 2 x dx. Thus:
Z Z
2 2
2 x e x dx = e u du = e u + C = e x + C
Example 5.20
Z p
Evaluate x 1 + 3 x2 dx.
Solution: Note that the derivative of the 1 + 3 x2 term inside the square root function is 6 x, which is
almost the function x outside the square root—all that is missing is the constant multiple 6. That is a
hint to let u = 1 + 3 x2 . Notice also that du = 6 xdx ⇒ x dx = 61 du, and x dx is the remaining part of the
integral outside the square root. Thus:
Z p Z
1 p 1 u3/2 1
x 1 + 3 x2 dx = u du = + C = (1 + 3 x2 )3/2 + C
6 6 3/2 9
Example 5.21
Z 2
x dx
Evaluate p .
x3 + 9
Solution: Let u = x3 + 9, so that du = 3 x2 dx ⇒ x2 dx = 13 du. Thus:
Z Z 1 Z
x2 dx 3du 1 1 u1/2 2p 3
p = p = u−1/2 du = + C = x +9 + C
x3 + 9 u 3 3 1/2 3
Example 5.22
Z
2 x dx
Evaluate .
x2 − 1
Solution: Notice that the numerator 2 x in the function is exactly the derivative of the denominator
x2 − 1. That is a hint to substitute on the denominator so that the integral is the natural logarithm
function. Let u = x2 − 1, so that du = 2 x dx. Thus:
Z Z
2 x dx du
2
= = ln | u | + C = ln | x2 − 1 | + C
x −1 u
Example 5.23
Z
Evaluate tan x dx.
sin x
Solution: Notice that tan x = cos x and that the numerator sin x is almost the derivative of the denom-
inator cos x; all that is missing is a negative sign. That is a hint to substitute on the denominator:
u = cos x, so that du = − sin x dx ⇒ sin x dx = −du. Thus:
Z Z
sin x
tan x dx = dx
cos x
Z
−du
= = − ln | u | + C = − ln | cos x | + C = ln | cos x |−1 + C = ln | sec x | + C
u
Example 5.24
Z
Evaluate sec x dx.
Solution: Notice that
Z Z Z
sec x (sec x + tan x) sec x tan x + sec2 x
sec x dx = dx = dx
sec x + tan x sec x + tan x
and that the numerator in the last integral is the derivative of the denominator: let u = sec x + tan x,
so that du = (sec x tan x + sec2 x) dx. Thus:
Z Z
du
sec x dx = = ln | u | + C = ln | sec x + tan x | + C
u
The following formulas are straightforward consequences of substitution and the

derivatives of inverse trigonometric functions discussed in Section 2.2:
For any constant a > 0:

Z ³x´
dx −1
p = sin + C (if | x | < a) (5.3)
a2 − x2 a
Z ³ ´
dx 1 −1 x
= tan + C (5.4)
a2 + x2 a a
Z ³ x´
dx 1
p = sec−1 + C (if | x | > a) (5.5)
| x | x2 − a2 a a
¡ ¢
For example, to prove the second formula, recall that dxd
tan−1 x = 1+1x2 . Make the
substitution u = x/a, so that x = au and dx = a du. Thus:
Z Z Z ³ ´
dx a du 1 du 1 −1 1 −1 x
= = = tan u + C = tan + C X
a2 + x2 a2 + a2 u 2 a 1 + u2 a a a
Example 5.25
Z
dx
Evaluate p .
4 − 9 x2
Solution: The 4 − 9 x2 inside the square root is almost of the form a2 − x2 , except for the 9. The goal is to
have u2 = 9 x2 , so let u = 3 x, which means that dx = 31 du and u2 = 9 x2 . Thus,
Z Z ³u´ µ ¶
dx 1 du 1 1 3x
p = p = sin−1 + C = sin−1 + C
4 − 9x 2 3 4−u 2 3 2 3 2
by Formula (5.2) with a = 2.
To use substitution with definite integrals, follow the same procedure as with indef-
inite integrals but add one extra step: replace the limits of integration x = a and x = b
Rb
in the original integral a f ( x) dx by u = g(a) and u = g( b), respectively, in the new
integral involving u, where u = g( x) is your substitution.
Example 5.26
Z2
Evaluate (2 x + 1)3 dx.
1
Solution: Let u = g( x) = 2 x + 1, which means that dx = 12 du. The upper limit of integration x = 2 becomes
u = g(2) = 2(2) + 1 = 5 in the new u-based integral, while the lower limit of integration x = 1 becomes
u = g(1) = 2(1) + 1 = 3. Thus:
Z2 Z ¯5
1 5 3 1 ¯ 1
(2 x + 1)3 dx = u du = u4 ¯¯ = (54 − 34 ) = 68
1 2 3 8 3 8
Note that you could have put everything back in terms of x at the end, but there was no need to since
you would get the same numerical answer.
The following property of definite integrals comes in handy for evaluating certain
types of definite integrals:
For any constant a, Za Za

f ( x) dx = f (a − x) dx . (5.6)
0 0
This is simple to prove, using the substitution u = a − x, so x = a − u and dx = − du,

while x = 0 becomes u = a and x = a becomes u = 0 in the limits of integration:
Za Z0 Za Za
f ( x) dx = − f (a − u) du = f (a − u) du = f (a − x) dx X
0 a 0 0
Example 5.27
Zπ
x sin x
Evaluate dx.
0 1 + cos2 x
Zπ
x sin x
Solution: Let I = 2
dx. Then by the above property (with a = π):
0 1 + cos x
Zπ Zπ Zπ Zπ
(π − x) sin (π − x) (π − x) sin x sin x x sin x
I = dx = dx = π dx − dx
0 1 + cos2 (π − x) 0 1 + cos2 x 2
0 1 + cos x
2
0 1 + cos x
Zπ
sin x
I = π dx − I
0 1 + cos2 x
Zπ ¯π ³ ´ 2
sin x −1
¯
¯ = −π − π − π = π
2I = π 2
dx = − π tan (cos x ) ¯
1 + cos x
0 0 4 4 2
2
π
I =
4
Exercises
A
For Exercises 1-24 evaluate the given integral.
Z Z 2x Z³ ´
e + e −2 x 2
1. (3 cos 5 x + 4 sin 5 x) dx 2. dx 3. xe− x + x2 cos x3 dx
Z 2
Z ex Z
x − 2 1
4. dx 5. dx 6. dx
2 1 + ex 1
Z xp − 4 x + 9 Z Z + ex
7. x x + 4 dx 8. cos2 x dx 9. tan2 x dx
Z Z Z
3 3
10. p dx 11. dx 12. sin2 x cos3 x dx
4 − 25 x2 4 + 25 x2
Z1 Z1 Z8 p
13. (2 x + 1)3 dx 14. (2 x − 1)3 dx 15. x 1 + x dx
0 0 0
Zπ/2 Zπ/4 Zpπ
16. 4 sin ( x/2) dx 17. 4 sin x cos x dx 18. 5 x cos ( x2 ) dx
0 0 0
Z−1 Zln 3 Z3
x ex dx
19. dx 20. dx 21. p
−2 ( x + 2)3
2
− ln 3 ex + 4 1 x ( x + 1)
Z1 Z2 Z3
x2 dx dx x5 dx
22. p 23. 2
24. 2
−1 x3 + 9 1 x − 6x + 9 −3 ex
25. Evaluate the indefinite integral Z
sin x cos x dx
three different ways:

(a) Use the substitution u = sin x.
(b) Use the substitution u = cos x.
(c) Use the trigonometric identity 2 sin x cos x = sin 2 x.
(d) Are the three answers from parts (a)-(c) actually different? Explain.
26. For all positive constants L, show the following:
ZL ³
x ´2 L
(a) 1 − dx =
0 L 3
ZL/2 µ ¶
1 x 2 L
(b) − dx =
−L/2 2 L 3
ZL ³
x ´3 ³ x ´2 L
(c) 1 − dx =
0 L L 60
B
27. Recall from trigonometry that sin2 x = 1
(1 − cos 2 x) for all x.
2
Zπ
(a) Use the Fundamental Theorem of Calculus to evaluate sin2 x dx .
0
(b) Approximate the integral from part (a) by dividing the interval [0, π] into n = 2 subintervals of
equal length, [0, π/2] and [π/2, π], and finding the exact value of the sum of the areas of the
rectangles whose heights are determined at the right endpoints of the subintervals.
(c) Repeat part (b) with n = 3.
(d) Repeat part (b) with n = 4.
(e) Repeat part (b) with n = 6.
R
28. Show that csc x dx = − ln | csc x + cot x | + C . (Hint: See Example 5.24.)
C
Ra Ra
29. Use the property 0 f ( x) dx = 0 f (a − x) dx to show that
Zπ/2
sin2 x 1 ³p ´
dx = p ln 2 + 1 .
0 sin x + cos x 2
(Hint: Use Exercise 28 and the sine addition formula.)
5.5 Improper Integrals

Definite integrals so far have been defined only for continuous functions over finite
closed intervals. There are times when you will need to perform integration despite
those conditions not being met. For example, in quantum mechanics the Dirac delta
function4 δ is defined on R by four properties:
∞
(1): δ( x) = 0 for all x 6= 0
y
(2): δ(0) = ∞
Z∞
(3): δ( x) dx = 1
−∞
x
(4): For any continuous function f on R, 0
Z∞
Figure 5.5.1 y = δ( x)
f ( x) δ( x) dx = f (0).
−∞
Properties (3) and (4) provide examples of one type of improper integral: an inte-
gral over an infinite interval (in this case the entire real line R = (−∞, ∞)). Define this
type of improper integral as follows:
For a continuous function f and a real number a, define the improper integral
of f over [a, ∞) by
Z∞ Zb
f ( x) dx = lim f ( x) dx ,
a b→∞ a
define the improper integral of f over (−∞, a] by
Za Za
f ( x) dx = lim f ( x) dx ,
−∞ b→−∞ b
and define the improper integral of f over (−∞, ∞) by

Z∞ Zc Z∞
f ( x) dx = f ( x) dx + f ( x) dx ,
−∞ −∞ c
for any real number c (typically c = 0). If the given limit exists (i.e. is a real
number) then the improper integral is convergent; otherwise it is divergent.
4 Created by the physicist P.A.M. Dirac (1902-1984), who won a Nobel Prize in physics in 1933. The function is
neither real-valued nor continuous at x = 0. The “graph” in Figure 5.5.1 is perhaps misleading, as ∞ is not an
actual point on the y-axis. One interpretation is that δ is an abstraction of an instantaneous pulse or burst of
something, preceded and followed by nothing. To learn more about this fascinating and useful function, see §15 in
D IRAC, P.A.M., The Principles of Quantum Mechanics, 4th ed., Oxford, UK: Oxford University Press, 1958.
Improper Integrals • Section 5.5 153
The limits in the above definitions are always taken after evaluating the integral
inside the limit. Just as for “proper” definite integrals, improper integrals can be
interpreted as representing the area under a curve.
Example 5.28
Z∞
dx
Evaluate .
1 x
Solution: For all real numbers b > 1, y
Z∞ Zb Ã ¯b ! 1
dx dx ¯ y= x
= lim = lim ln x ¯¯
1 x b→∞ 1 x b→∞ 1
= lim (ln b − ln 1) = lim b = ∞

b→∞ b→∞ x
0 1
and so the integral is divergent. This means that the area under the
curve y = 1/ x over the interval [1, ∞)—as shown in the graph above—is infinite.
Example 5.29
Z∞
dx
Evaluate .
1 x2
Solution: For all real numbers b > 1, y
Z∞ Zb Ã ¯b ! 1
dx dx 1 ¯ y= x2
= lim = lim − ¯
x 2 2 x ¯
1 b →∞ 1 x b →∞ 1
µ µ ¶¶ µ ¶
1 1 1
= lim − − − = lim 1 − = 1 − 0 = 1. x
b→∞ b 1 b→∞ b
0 1
This means that the area under the curve y = 1/ x2 over the interval [1, ∞)—as shown in the graph
above—equals 1. Thus, an infinite region can have a finite area. Length and area are different and
not necessarily related concepts, as this example illustrates. Notice that y = 1/ x2 approaches the x-axis
asymptote much faster than y = 1/ x does—fast enough to make the integral convergent.
Example 5.30
Z0
Evaluate e x dx .
−∞
Solution: For all real numbers b < 0, y

Z0 Z0 Ã ¯0 !
¯
x ¯
e x dx = lim e x dx = lim e ¯
−∞ b→−∞ b b→−∞ b y = ex
= lim (1 − e b ) = 1 − 0 = 1 .
b→−∞ x
0
This means that the area under the curve y = e x over the interval
(−∞, 0]—as shown in the graph above—equals 1.
Example 5.31
Z∞
Evaluate sin x dx .
0
Solution: Since y
Z∞ Zb
sin x dx = lim sin x dx y = sin x
0 b→∞ 0
Ã 1
¯b !
¯
= lim − cos x ¯¯ = lim (− cos b + 1)
b→∞ 0 b→∞ x
0 2π
then the integral is divergent, since lim b→∞ cos b does not exist
−1
(cos b oscillates between 1 and -1). This means that the net area
over [0, ∞)—counted as positive above the x-axis and negative below—is indeterminate.
Example 5.32
Z∞
dx
Evaluate 2
.
−∞ 1 + x
Solution: Split the integral at x = 0: y

Z∞ Z0 Z∞
dx dx dx
= + 1
2 2 1 + x2 y=
−∞ 1 + x −∞ 1 + x 0 1+ x 2
µ Z0 ¶ µ Zb ¶
dx dx
= lim + lim
b→−∞ b 1 + x2 b→∞ 0 1 + x2 x
Ã ¯0 ! Ã ¯b ! 0
−1 ¯
¯ −1 ¯
¯
= lim tan x ¯ + lim tan x ¯
b→−∞ b b→∞ 0
= lim (tan−1 0 − tan−1 b) + lim (tan−1 b − tan−1 0)

b→−∞ b→∞
= (0 − (−π/2)) + (π/2 − 0) = π
This means that the area under the curve y = 1+1x2 over the entire real line (−∞, ∞)—as shown in the
graph above—equals π. Note that if the integral were split at any number c then the answer would
be the same. Another way to evaluate the integral would have been to use the symmetry around the
y-axis—as f ( x) = 1+1x2 is an even function—so that
Z∞ Z∞
dx dx
2
= 2 = · · · = 2(π/2 − 0) = π .
−∞ 1 + x 0 1 + x2
Since the integrand is continuous over R, a common way of evaluating the integral—especially among
students—is to simply use ±∞ as actual limits of integration, thus avoiding the need to take a limit:
Z∞ ¯∞
dx −1 ¯
¯ π −π
2
= tan x ¯ = tan−1 (∞) − tan−1 (−∞) = − = π
−∞ 1 + x −∞ 2 2
This type of shortcut is fine as long as you are aware of what plugging x = ±∞ into tan−1 x actually
means, and that there are no numbers for which the integrand is undefined (which would yield an
improper integral of a different type, to be discussed shortly).
The second type of improper integral is of a function not continuous or not bounded
over its interval of integration. For example, the integral in property (3) of the Dirac
delta function is of that type, since δ is discontinuous at x = 0. Define this type of
improper integral as follows:
For a function f that is continuous on [a, b) but has either a discontinuity or vertical
asymptote at x = b, define the improper integral of f over [a, b) by
Zb Zc
f ( x) dx = lim f ( x) dx .
a c→ b− a
Likewise, if f is continuous on (a, b] but has either a discontinuity or vertical

asymptote at x = a, then define the improper integral of f over (a, b] by
Zb Zb
f ( x) dx = lim f ( x) dx .
a c→a+ c
If f is continuous on [a, b] but has either a discontinuity or vertical asymptote at

x = c for a < c < b, then define the improper integral of f over [a, b] by
Zb Zc Zb
f ( x) dx = f ( x) dx + f ( x) dx ,
a a c
where the integrals on the right are evaluated as in the first two definitions. If the
given limit exists (i.e. is a real number) then the improper integral is convergent;
otherwise it is divergent.
Adjust these definitions accordingly for infinite intervals—e.g. [a, ∞), (−∞, b], or
(−∞, ∞)—to be consistent with the definitions of improper integrals of that type.
Example 5.33
Z1
dx
Evaluate .
0 x
Solution: Since x = 0 is a vertical asymptote for y = 1x , y
Z1 Z1 Ã ¯1 !
dx dx ¯
= lim = lim ln x ¯ 1
x c→0+ x c→0+ ¯ y=
0 c c x
= lim (ln 1 − ln c) = 0 − (−∞) = ∞

c→0+
x
and so the integral is divergent. This means that the area under the curve y = 1/ x 0 1
over the interval (0, 1]—as shown in the graph above—is infinite. The region is infinite in the y direction.
Example 5.34
Z1
dx
Evaluate p .
0 x
Solution: Since x = 0 is a vertical asymptote for y = p1 , y
x
Z1 Z1 Ã ¯1 !
dx dx p ¯
p = lim p = lim 2 x ¯
c→0+ c→0+ ¯ y= p1
0 x c x c x
p
= lim (2 − 2 c ) = 2 − 0 = 2 .
c→0+
p x
This means that the area under the curve y = 1/ x over the interval (0, 1]—as shown
0 1
in the graph above—equals 2. The region is infinite in the y direction.
Example 5.35
Z3
Evaluate ⌊ x⌋ dx .
1
Solution: Recall from Example 3.22 in Section 3.3 that the floor function y
y = ⌊ x⌋ has jump discontinuities at each Rinteger value of x, as shown
3 2
in the graph on the right. The integral 1 ⌊ x⌋ dx is thus an improper y = ⌊ x⌋
integral over the interval [1, 3), which needs to be split at the point of
discontinuity x = 2 within that interval: 1
Z3 Z2 Z3
⌊ x⌋ dx = ⌊ x⌋ dx + ⌊ x⌋ dx
1 1 2 x
µ Zb ¶ µ Zc ¶ 0 1 2 3
= lim ⌊ x⌋ dx + lim ⌊ x⌋ dx
b→2− 1 c→3− 2
µ Zb ¶ µ Zc ¶
= lim 1 dx + lim 2 dx
b→2− 1 c→3− 2
Ã ¯b ! µ ¯c ¶
¯ ¯
= lim x ¯¯ + lim 2 x ¯¯
b→2− c→3− 1 2
= lim (b − 1) + lim (2 c − 4) = (2 − 1) + (6 − 4) = 3
b→2− c→3−
Similar to some of the above examples, the following result is easy to prove (see the
exercises):
For any real number a > 0, the improper integral

Z∞
dx
a xp
is convergent if p > 1, and divergent if 0 < p ≤ 1.
The following test for convergence or divergence is sometimes helpful:
Comparison Test for Improper Integrals:

R∞ R∞
(a) If | f ( x) | ≤ g( x) for all x in [a, ∞), and if a g( x) dx is convergent, then a f ( x) dx
is convergent.
R∞ R∞
(b) If f ( x) ≥ g( x) ≥ 0 for all x in [a, ∞), and if a g( x) dx is divergent, then a f ( x) dx
is divergent.
The idea behind part (a) is that if − g( x) ≤ f ( x) ≤ g( x) over [a, ∞), then—thinking
of improper integrals as areas—the integral of f is “squeezed” between the two finite
integrals for ± g. There are, however, some subtle issues to prove about the limit in
the integral of f —finite bounds might not necessarily mean the limit exists.5
Example 5.36
Z∞
sin x
Show that dx is convergent.
1 x2
R∞ 1
Solution: By Example 5.29, the integral 1 x2
dx is convergent. So y
1
since | sin x | ≤ 1 for all x, then y= x2 sin x
x2
¯ ¯
¯ sin x ¯ 1
¯ ¯ 0
¯ x2 ¯ ≤ x2 x
1
R∞ sin x
for all x in [1, ∞). Thus, by the Comparison Test, 1 x2
dx is y = − x12
sin x
convergent. The graph on the right shows how the curve y = x2
is bounded between the curves y = ± x12 .
The rules and properties from Section 5.3 concerning definite integrals still apply
to improper integrals, provided the improper integrals are convergent. For example,
suppose a function f has a discontinuity or vertical asymptote at x = c. If both im-
Rc Rb
proper integrals a f ( x) dx and c f ( x) dx are convergent, then the improper integral
Rb
a f ( x) dx is convergent and
Zb Zc Zb
f ( x) dx = f ( x) dx + f ( x) dx .
a a c
Rc R∞ R∞
Likewise, if a f ( x) dx and c f ( x) dx are convergent, then so is a f ( x) dx, with
Z∞ Zc Z∞
f ( x) dx = f ( x) dx + f ( x) dx .
a a c
5 See pp.140-141 in B UCK , R.C., Advanced Calculus, 2nd ed., New York: McGraw-Hill Book Co., 1965.
Exercises
A
For Exercises 1-15, evaluate the given improper integral.
Z∞ Z1 Z∞ Z∞ Z1
dx dx dx
1. 2. p 3. e− x dx 4. e−2 x dx 5.
1 x3 0
3
x 0 0 −1 x
Z∞ Z0 Zπ/2 Z1 Z1
2 ln x dx
6. xe− x dx 7. 2 x dx 8. tan x dx 9. dx 10. p
0 −∞ 0 0 x −1 1 − x2
Z3 Z∞ Z1 Z∞ Z∞
dx dx dx dx
11. ⌈ x⌉ dx 12. 2
13. 14. 15. p
0 −∞ x + 4 0 ( x − 1)3 2 x ln x 1 x x2 − 1
16. In a standby system of two non-identical components, the normal operating component A has a
failure rate of λ A > 0 failures per unit time, while the standby component B—which takes over when
A fails—has a failure rate λB > 0 (with λ A 6= λB ).
(a) Find the standby system’s reliability R ( t) beyond time t ≥ 0, where
Z∞
λ A λ B ³ −λB x ´
R ( t) = e − e−λ A x dx .
t λ A − λB
R∞ 1
(b) Show that the system’s mean time to failure (MTTF) m, where m = 0 R ( t) dt , is m = λA + λ1 .
B
Z∞
dx
17. Show that for all a > 0, is convergent if p > 1, and divergent if 0 0, p
is convergent if 0 < p < 1, and divergent if p ≥ 1.
Z∞ 0 x Z∞
dx dx
19. Is 4
convergent? Explain. 20. Is p convergent? Explain.
1 x + x 2 x − x
B
R∞
21. Example 5.31 showed that 0 sin x dx is divergent. What is the flaw in the argument that the
integral must be 0 since each “hump” of sin x above the x-axis is canceled by one below the x-axis?
R∞ R∞ R∞
22. This exercise concerns the subtraction rule a ( f ( x) − g( x)) dx = a f ( x) dx − a g( x) dx.
1 1 1
(a) Show that x( x+1) = x − x +1 for all x except 0 and -1
R∞
(b) Show that 1 x( x+1) is convergent.
dx
R∞ R∞ dx
(c) Show that both 1 dx x and 1 x+1 are divergent.
(d) Does part (c) contradict parts(a)-(b) and the subtraction rule? Explain.
R∞ y
23. The improper integral −∞ δ( x) dx = 1 is one of the notable
“improprieties” of the Dirac delta function δ. One way to
n
think of that integral is by approximating δ by triangular
“pulse” functions D n (for n ≥ 1), as in the picture on the right. y = D n ( x)
(a) Write a formula for each D n ( x) over all of R.
R∞
(b) Show that −∞ D n ( x) dx = 1 for all integers n ≥ 1.
x
(c) Show that lim n→∞ D n (0) = ∞ = δ(0). 0
− n1 1
n
(d) Do the D n functions begin to resemble δ as n → ∞?
CHAPTER 6
Methods of Integration
6.1 Integration by Parts

In physics and engineering the Gamma function1 Γ ( t), defined by
Z∞
Γ ( t) = x t−1 e−x dx for all t > 0,
0
has found many uses. Evaluating Γ (2) entails integrating the function f ( x) = x e−x . No
formula or substitution you have learned so far would be of help. Differentiating that
function, on the other hand, is easy. By the Product Rule for differentials,
d ( x e − x ) = x d ( e − x ) + d ( x) e − x
= − x e−x dx + e−x dx
d ( x e−x ) = − x e−x dx − d ( e−x )
x e−x dx = − d ( x e−x ) − d ( e−x ), so integrate both sides to get
Z Z Z
x e dx = − d ( x e ) − d ( e−x ) = − x e−x − e−x + C
−x −x
R
since dF = F + C . Generalizing this process for functions u and v,
d ( uv) = u dv + v du
u dv = d ( uv) − v du
so that integrating both sides yields the integration by parts formula:
For differentiable functions u and v:

Z Z
u dv = uv − v du (6.1)
1 Created by the Swiss mathematician, physicist and astronomer Leonhard Euler (1707-1783). The use of the
Greek capital letter Γ for this function is due to the French mathematician Adrien-Marie Legendre (1752-1833).
159
160 Chapter 6 • Methods of Integration §6.1
Integration by parts is just

R the Product Rule for derivatives in integral form,
R typically
used when the integral v du would be simpler than the original integral u dv.
Example 6.1
Z
Use integration by parts to evaluate x e− x dx . Use the answer to evaluate Γ (2).
R
Solution: The original integral is always of the form u dv, so you must decide which parts of xe− x dx
will represent u and dv. Typically you would choose dv to be a differential that you could integrate
easily (since you will need to integrate dv to get v) and choose u to be a function whose derivative is
simpler than u (since that derivative will appear in v du, which R you Rhope to be a simpler integral). In
this case, pick u = x and dv = e− x dx. Then du = dx and v = dv R = e− x dx = − e− x (you can omit the
generic constant C for now—include it when you have evaluated v du). Thus,
Z Z
u dv = uv − v du
Z Z
x e− x dx = x (− e− x ) − − e− x dx
u dv u v v du
Z
x e− x dx = − x e− x − e− x + C
R R
which agrees with the example at the beginning of this section. Note that v du = − e− x dx, which
indeed is simpler than the original integral. The Gamma function value Γ (2) can now be evaluated:
Z∞
Γ (2) = x e− x dx
0
¯∞
−x
¯
−x ¯
= −x e − e ¯
0
= lim (− x e −x
− e − x ) − ( −0 e 0 − e 0 )
x→∞
³ x´ ³ ´
= − lim x − lim e− x − (0 − 1)
x→∞ e x→∞
= 0 − 0 + 1 = 1
Example 6.2
What would happen in Example 6.1 if you let u = e− x and dv = x dx?
R R
Solution: In this case du = − e− x dx and v = dv = x dx = 21 x2 , so that
Z Z
x e− x dx = e− x x dx
u dv
Z
1 2 1 2
= e− x x − x (− e− x ) dx
2 2
u v v du
Z
1 2 −x 1
x e + = x2 e− x dx
2 2
which leads you in the wrong direction: a more difficult integral than the original.
Integration by Parts • Section 6.1 161
Example 6.2 showed the importance of an appropriate choice for u and dv. There
are some rough guidelines for that choice—as in Example 6.1—but no rules that are
guaranteed to always work. It might not be clear when you should even attempt
integration by parts.
Example 6.3
Z
Evaluate ln x dx .
Solution: Integration by parts ostensibly requires two functions in the integral, whereas here ln x ap-
pears to be the only one. However, the choice for dv is a differential,
R R and one exists here: dx. Choosing
dv = dx obliges you to let u = ln x. Then du = 1x dx and v = dv = dx = x. Now integrate by parts:
Z Z
u dv = uv − v du
Z Z
1
ln x dx = (ln x) ( x) − x· dx
x
Z
= x ln x − 1 dx
= x ln x − x + C
Note that choosing dv = ln x dx would be pointless, as integrating dv to get v is the original problem!
Example 6.4
Z
2
Evaluate x3 e x dx .
Solution: One frequently useful guideline for integration by parts is to eliminate the most complicated
function in the integral by integrating it—as dv—into something simpler (which becomes v). In this
2 2
integral, e x is somewhat complicated but has no closed form antiderivative. However, x e x appears in
2
the integral and can be integrated easily (using a substitution as in Section 5.4). So choose dv = x e x dx,
R R 2 2
which means u = x2 . Then du = 2 x dx and v = dv = x e x dx = 21 e x . Now integrate by parts:
Z Z
u dv = uv − v du
Z Z
2 1 x2 1 x2
x3 e x dx = x2 · e − e · 2 x dx
2 2
Z
x 2 x2 2
= e − x e x dx
2
x 2 x2 1 2
= e − ex + C
2 2
( x2 − 1) x2
= e + C
2
Sometimes multiple rounds of integration by parts are needed, as in the following

example.
Example 6.5
Z
Evaluate x2 e− x dx .
Solution: This integral appears

R −x similar to the one in Example 6.1, so choose dv = e− x dx and u = x2 .
Then du = 2 x dx and v = e dv = − e− x . Now integrate by parts:
Z Z
u dv = uv − v du
Z Z
x2 e− x dx = x2 · (− e− x ) − − e− x · 2 x dx
Z
= − x2 e− x + 2 x e− x dx (integrate by parts again)
Z
x2 e− x dx = − x2 e− x + 2 (− x e− x − e− x ) + C (by Example 6.1)
= − x2 e− x − 2 x e− x − 2 e− x + C
In the above example, notice that the u = 2 x in the second integral came from the
derivative of the u = x2 in the first integral. Likewise, the dv = − e−x dx in the second
integral came from integrating the dv = e−x dx from the first integral. In general, if
n rounds of integration by parts were needed, with u i and v i representing the u and
v, respectively, for round i = 1, 2, . . ., n, then the repeated integration by parts would
look like this:
Z Z
u 1 dv1 = u 1 v1 − v1 du 1
Z
= u 1 v1 − u 2 dv2
µ Z ¶ Z
= u 1 v1 − u 2 v2 − v2 du 2 = u 1 v1 − u 2 v2 + u 3 dv3
µ Z ¶
= u 1 v1 − u 2 v2 + u 3 v3 − u 4 dv4
µ Z ¶
= u 1 v1 − u 2 v2 + u 3 v3 − u 4 v4 − u 5 dv5
= ···
Z
= u 1 v1 − u 2 v2 + u 3 v3 − u 4 v4 + u 5 v5 − · · · u n dvn
R
The last integral u n dvn is one you could presumably integrate easily.
The above procedure is called the tabular method for integration by parts, since it
can be shown in a table (the arrows indicate multiplication):
u dv
u1 dv1
u2 v1 (+) + u 1 v1
u3 v2 (–) − u 2 v2
u3 v3 (+) + u 3 v3
u4 v4 (–) − u 4 v4
.. .. ..
. . .
The idea is to differentiate down the u column and integrate down the dv column. If
the u in the original integral is a polynomial of degree n, then you know from Section
1.6 that its ( n +1)-st derivative will be 0, at which point the tabular method terminates.
The integral is then the sum of the indicated products with alternating signs.
For example, the tabular method on the integral from Example 6.5 looks like this:
u dv
x2 e−x dx
2x − e−x (+) + ( x 2 ) (− e − x )
2 e− x (–) − (2 x) ( e−x )
STOP 0 − e−x (+) + (2) (− e−x )
The integral is the sum of the products, and agrees with the result in Example 6.5:
Z
x2 e−x dx = + ( x2 ) (− e−x ) − (2 x) ( e−x ) + (2) (− e−x ) + C = − x2 e−x − 2 x e−x − 2 e−x + C
Example 6.6
Z
Evaluate x3 e− x dx .
Solution: Use the tabular method with u = x3 and dv = e− x dx:

u dv
x3 e− x dx
3x 2 − e− x (+) + ( x 3 ) (− e − x )
6x e− x (–) − (3 x2 ) ( e− x )
6 −e −x (+) + (6 x) (− e− x )
STOP 0 e −x (–) − (6) ( e− x )
Z
x3 e− x dx = − x3 e− x − 3 x2 e− x − 6 x e− x − 6 e− x + C
Integration by parts can sometimes result in the original integral reappearing, al-
lowing it to be combined with the original integral.
Example 6.7
Z
Evaluate sec3 x dx .
R R
Solution: Let u = sec x and dv = sec2 x dx, so that du = sec x tan x dx and v = dv = sec2 x dx = tan x.
Then
Z Z
u dv = uv − v du
Z Z
sec3 x dx = sec x tan x − sec x tan2 x dx
Z Z
sec3 x dx = sec x tan x − sec x (sec2 x − 1) dx
Z Z Z
sec3 x dx = sec x tan x + sec x dx − sec3 x dx
Z
2 sec3 x dx = sec x tan x + ln | sec x + tan x | + C
Z
1
sec3 x dx = (sec x tan x + ln | sec x + tan x |) + C
2
Example 6.8
Z
Evaluate e x sin x dx .
R R
Solution: Let u = e x and dv = sin x dx, so that du = e x dx and v = dv = sin x dx = − cos x. Then
Z Z
u dv = uv − v du
Z Z
e x sin x dx = − e x cos x + e x cos x dx
and so integration by parts

R is R needed again, for the integral on the right: let u = e x and dv = cos x dx, so
that du = e dx and v = dv = cos x dx = sin x. Then
x
Z µ Z ¶
e x sin x dx = − e x cos x + uv − v du
Z µ Z ¶
e x sin x dx = − e x cos x + e x sin x − e x sin x dx
Z
2 e x sin x dx = − e x cos x + e x sin x
Z
ex
e x sin x dx = (sin x − cos x) + C
2
In Example 6.1 integration by parts was used in evaluating an improper integral.

In general, in definite or improper integrals where a and b are real numbers or ± ∞,
Zb ¯b Zb
¯
u dv = uv ¯¯ − v du .
a a a
Example 6.9
Z1 p
Evaluate x3 1 − x2 dx .
0
p p p
Solution: Since xp3 1 − x2 = x2 · x 1 − x2 , and x 1 R
− x2 isReasy
p to integrate (via a substitution), let
u = x and dv = x 1 − x dx. Then du = 2 x dx and v = dv = x 1 − x2 dx = − 13 (1 − x2 )3/2 , and so:
2 2
Zb ¯b Zb
¯
u dv = uv ¯¯ − v du
a a a
Z1 ¯1 Z1
p x2 ¯ 2x
x3 1 − x2 dx = − (1 − x2 )3/2 ¯¯ + (1 − x2 )3/2 dx
0 3 0 0 3
Ã ¯1 !
−2 ¯
2 5/2 ¯ 2 2
= (0 − 0) + (1 − x ) ¯ = 0 + =
15 0 15 15
Exercises
A
For Exercises 1-25, evaluate the given integral.
Z Z Z Z Z
2 x
1. x ln x dx 2. x e dx 3. x cos x dx 4. x
x 3 dx 5. x2 a x dx (a > 0)
Z Z Z Z Z
6. ln 4 x dx 7. ln x2 dx 8. x2 sin x dx 9. x cos2 x dx 10. sin x cos 2 x dx
Z Z Z Z Z
11. sin−1 x dx 12. cos−1 2 x dx 13. tan−1 3 x dx 14. x sec2 x dx 15. sin x sin 3 x dx
Z Z Z Z2 3 Z1 p
ln x 3 2 x dx
16. dx 17. x ln x dx 18. x5 e x dx 19. p 20. x3 1 + x2 dx
x3 0 4 − x2 0
Z Z Z Z Z p
−1 p
21. sin (ln x) dx 22. ln (1 + x2 ) dx 23. x tan−1 x dx 24. cot x dx 25. e x dx
R x
26. Evaluate the integral
R e sin x dx fromRExample 6.8 by using two rounds of the tabular method
and the formula u1 dv1 = u1 v1 − u2 v2 + v2 du2 from p.162.
B
27. Show that for all constants a and b 6= 0,
Z Z
e ax (a cos bx + b sin bx) e ax (a sin bx − b cos bx)
e ax cos bx dx = and e ax sin bx dx = .
a2 + b2 a2 + b2
28. For the Gamma function Γ ( t) show the following:

(a) Γ ( t + 1) = t Γ ( t) for all t > 0. (Hint: Use integration by parts.)
(b) Γ (n) = (n − 1) ! for all positive integers n. (Hint: Use part (a) and induction.)
Note that by part (b) the Gamma function can be thought of as an extension of the factorial operation
to all positive real numbers. In fact, the Gamma function was created for that purpose.
29. Use Exercise 28 to prove for all integers n ≥ 1:
Z∞ Z∞
r n e−r ln r dr = (n − 1) ! + n r n−1 e−r ln r dr
0 0
30. By the Maxwell speed distribution for gas molecules, the average speed 〈 ν 〉 of molecules of mass m
in a gas at temperature T is
³ m ´3/2 Z∞ 2
〈 ν 〉 = 4π ν3 e−mν /2 kT dν ,
2πkT 0
where k ≈ 1.38056 × 10−23 J/K is the Boltzmann constant. Show that

s
8kT
〈ν〉 = .
πm
31. Some physics texts write integrals in a form like this energy integral from statistical mechanics,
Z∞ ³ ´
2
ln 1 − α e− x d ( x3 )
0
which uses the differential of a function—in this case d ( x3 )—instead of a variable (e.g. not just plain
dx).RThis often signals that integration by parts is on the way, with the added benefit of having the
v = dv calculation done for you—in the above integral d ( x3 ) means that v = x3 , with dv = d ( x3 ) not
really being needed for anything else. With that understanding, show that for 0 < α < 1,
Z∞ ³ ´ Z∞ 4 − x 2
2 x e dx
ln 1 − α e− x d ( x3 ) = −2α 2
.
0 0 1 − αe x
−
32. Find the flaw in the following “proof ” that 0 = 1:

Z
dx 1 dx
Evaluating the integral using integration by parts with u = and dv = dx, so that du = − 2
x x x
and v = x, shows that
Z Z
u dv = uv − v du
Z µ ¶ Z µ ¶
dx 1 dx
= ·x − x· − 2
x x x
Z Z
dx dx
= 1 +
x x
0 = 1 X
Trigonometric Integrals • Section 6.2 167
6.2 Trigonometric Integrals

In engineering applications you sometimes encounter integrals of the form
Z
cos (α t + φ1 ) cos (β t + φ2 ) dt
where α t+φ1 and β t+φ2 are different angles (e.g. when the voltage and current are out
of phase in an AC circuit). In general, integrals involving products of sines and cosines
with “mixed” angles can be simplified with the useful product-to-sum formulas:2
1
sin A cos B = 2
(sin ( A + B) + sin ( A − B)) (6.2)
1
cos A sin B = 2
(sin ( A + B) − sin ( A − B)) (6.3)
1
cos A cos B = 2
(cos ( A + B) + cos ( A − B)) (6.4)
sin A sin B = − 12 (cos ( A + B) − cos ( A − B)) (6.5)
Example 6.10
Z
Evaluate 0.5 sin x sin 12 x dx .
Solution: Using the product-to-sum formula (6.5) with A = x and B = 12 x,
sin A sin B = − 21 (cos ( A + B) − cos ( A − B))

sin x sin 12 x = − 21 (cos ( x + 12 x) − cos ( x − 12 x))
sin x sin 12 x = − 21 (cos 13 x − cos 11 x)
since cos(−11 x) = cos 11 x. Then

Z Z
1
0.5 sin x sin 12 x dx = − (cos 13 x − cos 11 x) dx
4
1 1
= − sin 13 x + sin 11 x + C
52 44
Notice how the product-to-sum formula turned an integral of products of sines into integrals of individ-
ual cosines, which are easily integrated. The integrand is an example of a modulated wave, commonly
used in electronic communications (e.g. radio broadcasting). The graph is shown below:
y
y = 0.5 sin x sin 12 x
0.5
0 x
π 2π
−0.5
The curves y = ±0.5 sin x (shown in dashed lines) form an amplitude envelope for the modulated wave.
2 See Section 3.4 in C ORRAL , M., Trigonometry, http://mecmath.net/trig/, 2009.

On occasion you might need to integrate trigonometric functions raised to powers

higher than two. For the sine function raised to odd powers of the form 2 n + 1 (for
n ≥ 1), the trick is to replace sin2 x by 1 − cos2 x, so that
Z Z
2 n+1
sin x dx = (sin2 x)n sin x dx
Z
= (1 − cos2 x)n sin x dx
Z
= p( u) du
where p( u) is a polynomial in the variable u = cos x, and the single sin x is now part of
du = − sin x dx. You can then use the Power Formula to integrate that polynomial.
Example 6.11
Z
Evaluate sin3 x dx .
Solution: Let u = cos x so that du = − sin x dx:
Z Z
sin3 x dx = (sin2 x) sin x dx
Z
= (1 − cos2 x) sin x dx
Z
= (1 − u2 ) (−du)
Z
= (u2 − 1) du
1 3
= u − u + C
3
1
= cos3 x − cos x + C
3
R
In general sin2n+1 x dx will be a polynomial of degree 2 n + 1 in terms of cos x. Sim-
ilarly, use cos2 x = 1 − sin2 x to integrate odd powers of cos x, with the substitution
u = sin x:
Z Z
2 n+1
cos x dx = (cos2 x)n cos x dx
Z
= (1 − sin2 x)n cos x dx
p ( u) du
R
Integrals of the form sinm x cos n x dx, where either m or n is odd, can be evaluated
using the above trick for the function having the odd power.
Example 6.12
Z
Evaluate sin2 x cos3 x dx .
Solution: Replace cos2 x by 1 − sin2 x, then let u = sin x so that du = cos x dx:
Z Z
sin2 x cos3 x dx = sin2 x (cos2 x) cos x dx
Z
= sin2 x (1 − sin2 x) cos x dx
p( u) du
Z
= (u2 − u4 ) du
1 3 1
= u − u5 + C
3 5
1 1
= sin3 x − sin5 x + C
3 5
For even powers of sin x or cos x. You would replace sin2 x or cos2 x with either
1 − cos 2 x 1 + cos 2 x
sin2 x = or cos2 x = ,
2 2
respectively, as often as necessary, then proceed as before if odd powers occur.
Example 6.13
Z
Evaluate sin4 x dx .
1−cos 2 x
Solution: Replace sin2 x by 2 :
Z Zµ ¶
1 − cos 2 x 2
sin4 x dx = dx
2
Z
1
= (1 − 2 cos 2 x + cos2 2 x) dx
4
Zµ ¶
1 1 + cos 4 x
= 1 − 2 cos 2 x + dx
4 2
Z
1
= (3 − 4 cos 2 x + cos 4 x) dx
8
3x 1 1
= − sin 2 x + sin 4 x + C
8 4 32
R
Similar methods can be used for integrals of the form sec m x tann x dx when either
m is even or n is odd. For an even power m = 2 k + 2, use sec2 x = 1 + tan2 x for all but
two of the m powers of sec x, then use the substitution u = tan x, so that du = sec2 x dx.
This results in an integral of a polynomial p( u) in terms of u = tan x:
Z Z
2 k+2
sec x tan x dx = (sec2 x)k sec2 x tann x dx
n
Z
= (1 + tan2 x)k tann x sec2 x dx
p ( u) du
Likewise for an odd power n = 2 k + 1, use tan2 x = sec2 x − 1 for all but one of the n
powers of tan x, then use the substitution u = sec x, so that du = sec x tan x dx. This
results in an integral of a polynomial p( u) in terms of u = sec x:
Z Z
2 k+1
m
sec x tan x dx = secm−1 x sec x (tan2 x)k tan x dx
Z
= secm−1 x (sec2 x − 1)k sec x tan xdx
p ( u) du
R
Mimic the above procedure for integrals of the form cscm x cotn x dx when either m
is even or n is odd, using the identity csc2 x = 1 + cot2 x in a similar manner.
Example 6.14
Z
Evaluate sec4 x tan x dx .
Solution: Use sec2 x = 1 + tan2 x for one sec2 x term, then substitute u = tan x, so that du = sec2 x dx:
Z Z
sec4 x tan x dx = sec2 x sec2 x tan x dx
Z
= (1 + tan2 x) tan x sec2 x dx
Z
= (1 + u2 ) u du
Z
= (u + u3 ) du
1 2 1
= u + u4 + C
2 4
1 1
= tan2 x + tan4 x + C
2 4
For some trigonometric integrals try putting everything in terms of sines and cosines.
Example 6.15
Z
cot4 x
Evaluate dx .
csc5 x
Solution: Put cot x and csc x in terms of sin x and cos x:
Z Z
cot4 x cos4 x sin5 x
dx = dx
csc5 x sin4 x
Z
= cos4 x sin x dx (now let u = cos x, du = − sin x dx)
Z
1
= − u4 dx = − cos5 x + C
5
Exercises
A
Z Z Z Z
1. sin 2 x cos 5 x dx 2. cos 2 x cos 5 x dx 3. sin 2 x sin 5 x dx 4. cos 2π x sin 3π x dx
Z Z Z Z
5. sin3 x cos3/2 x dx 6. cos4 x dx 7. sin6 x dx 8. sin x sin 2 x sin 3 x dx
Z Z Z Z
tan3 x dx
9. sec4 x dx 10. sec2 x tan3 x dx 11. dx 12.
sec4 x csc2 x cot x
B
R
13. Evaluate sin3 x cos3 x dx in two different ways:
3
(a) Use sin x = (1 − cos2 x) sin x and the substitution u = cos x.
(b) Use cos3 x = (1 − sin2 x) cos x and the substitution u = sin x.
Are the answers from parts(a) and (b) equivalent? Explain.
R
14. Evaluate sec4 x tan x dx by using sec4 x tan x = sec3 x (sec x tan x) and the substitution u = sec x.
Is your answer equivalent to the answer in Example 6.14? Explain.
Z
4 tan x (1 − tan2 x) 1
15. Show that dx = − cos 4 x + C .
(1 + tan2 x)2 4
16. The autocorrelation function R x (τ) of the periodic function x( t) = A cos (ω t + θ) is given by
Z2π/ω
ω
R x (τ) = x( t) x( t − τ) dt
2π 0
where A , ω and θ are constants. Show that
A2
R x (τ) = cos ωτ .
2
6.3 Trigonometric Substitutions

One of the fundamental formulas in geometry is for the area A of a circle of radius r:
A = πr 2 . The calculus-based proof of that formula uses a definite integral evaluated
by means of a trigonometric substitution, as will now be demonstrated.
Use the circle of radius r > 0 centered at the origin (0, 0) in the x y-plane, whose
equation is x2 + y2 = r 2 (see Figure 6.3.1(a) below).
y
x 2 + y2 = r 2
( x, y) = (r cos θ, r sin θ)
y
r
p
θ y= r2 − x2
x
0 r
θ=π θ=0
x
−r 0 r
(a) Full circle (b) Upper hemisphere
Figure 6.3.1 Circle of radius r
By symmetry about the x-axis, the area A of the circle is twice the area of its
p upper
hemisphere (see Figure 6.3.1(b) above), which is the area under the curve y = r 2 − x2 :
Zr p
A = 2 r 2 − x2 dx
−r
To evaluate this integral, recall from trigonometry that any point ( x, y) on the circle
can be written as ( x, y) = ( r cos θ , r sin θ ), where 0 ≤ θ < 2π (in radians) is the angle
shown in Figure 6.3.1(a). Figure 6.3.1(b) shows that as x goes from x = − r to x = r , the
angle θ goes from θ = π to θ = 0. Now substitute x = r cos θ and dx = − r sin θ dθ into
the integral and change the limits of integration from x = − r and x = r to θ = π and
θ = 0, respectively:
Z0 p Zπ p
A = 2 2 2 2
r − r cos θ (− r sin θ ) dθ = −2 r 2 (1 − cos2 θ ) (− r sin θ ) dθ
π 0
Zπ p Zπ Zπ
1 − cos 2θ
= 2 r sin2 θ r sin θ dθ = 2 r 2 2r2
sin2 θ dθ = ✓ dθ
0 0 0 2
✓
µ ¶ ¯π µ µ ¶¶
2 1 ¯
¯ 2 1 1
= r θ − sin 2θ ¯ = r π − sin 2π − 0 − sin 0
2 0 2 2
= πr2 X
Trigonometric Substitutions • Section 6.3 173
Rp
For an indefinite integral of the general form a2 − u2 du, the same calculation as
above with the substitutions u = a cos θ and du = −a sin θ dθ yields
Zp Zp Z
2
2 2
a − u du = 2 2 2
a − a cos θ (−a sin θ ) dθ = −a sin2 θ dθ
Z
1 − cos 2θ a2 a2 sin 2θ
= − a2 dθ = − θ + + C,
2 2 4
which is still in terms of θ . To put this back in terms of u, use θ = cos−1 ( ua ), the double-
p p
angle formula sin 2θ = 2 sin θ cos θ , and a2 − u2 = a2 sin2 θ = a sin θ . Then
Zp ³ ´
2 2
a2 −1 u 2a2 sin θ cos θ
a − u du = − cos +
2 a 4
a2 ³u´ (a cos θ ) (a sin θ )
= − cos−1 + + C
2 a 2
which results in the following formula:
Zp ³u´
a2 1 p
a2 − u2 du = − cos−1 + u a2 − u 2 + C (6.6)
2 a 2
It is left as an exercise to show that the substitution u = a sin θ gives:
Zp ³u´
a2 1 p
a2 − u2 du = sin−1 + u a2 − u 2 + C (6.7)
2 a 2
That these two seemingly different antiderivatives are equivalent follows immediately
from the identity sin−1 x + cos−1 x = π2 for all −1 ≤ x ≤ 1, which shows that the an-
πa2
tiderivatives differ by the constant 4
(absorbed in the generic constant C ):
a2 ³u´ 1 p a2 ³ π ³ u ´´ 1 p
sin−1 + u a2 − u 2 + C = − cos−1 + u a2 − u 2 + C
2 a 2 2 2 a 2
a2 ³u´ 1 p π a2
= − cos−1 + u a2 − u 2 + C +
2 a 2 4
C
Thus, either
R p substitution— u = a cos θ or u = a sin θ —can be used when evaluating the
integral a2 − u2 du. The latter choice is sometimes preferred, to avoid the negative
sign in du and the resulting formula.
Example 6.16
Zp
Evaluate 9 − 4 x2 dx .
p
Solution: The integrand is of the form a2 − u2 with a = 3 and u = 2 x, so that du = 2dx. Then dx = 12 du
and so:
Zp Z
1 p 2
9 − 4 x2 dx = a − u2 du
2
µ ³ ´ ¶
1 a2 −1 u 1 p 2 2
= sin + u a −u + C
2 2 a 2
µ ¶
9 −1 2 x 1 p
= sin + x 9 − 4 x2 + C
4 3 2
In general, when other methods fail, use the table below as a guide for certain types
of integrals, making use of the specified substitution and trigonometric identity:
Integral contains Substitution Identity

p
a2 − u2 u = a sin θ 1 − sin2 θ = cos2 θ
p
a2 + u2 u = a tan θ 1 + tan2 θ = sec2 θ
p
u2 − a2 u = a sec θ sec2 θ − 1 = tan2 θ
For example, the substitution u = a tan θ leads to the following formula:
Zp
1 p 2 a2 ¯ p ¯
a2 + u 2 2
du = u a + u + ln ¯ u + a2 + u2 ¯ + C (6.8)
2 2
Similarly, the substitution u = a sec θ yields this formula:
Zp
1 p 2 a2 ¯ p ¯
u2 − a2 du = u u − a2 − ln ¯ u + u2 − a2 ¯ + C (6.9)
2 2
The proof of each formula requires this result from Example 6.7 in Section 6.1:
Z
1
sec3 θ dθ = (sec θ tan θ + ln | sec θ + tan θ |) + C (6.10)
2
The above substitutions can be used even if no square roots are present.
Example 6.17
Z
dx
Evaluate .
(1 + x2 )2
Solution: Notice that this integral cannot be evaluated by using the Power Formula with the substi-
tution u = 1 + x2 (why?). Integration by parts does not look promising, either. So try a trigonometric
substitution. The integrand contains a term of the form a2 + u2 (with a = 1 and u = x), so use the
substitution x = tan θ. Then dx = sec2 θ dθ and so
Z Z
dx sec2 θ dθ
=
(1 + x2 )2 (1 + tan2 θ)2
Z
sec2 θ dθ
=
(sec2 θ)2
Z
dθ
=
sec2 θ2
Z
= cos2 θ dθ
Z
1 + cos 2θ
= dθ
2
θ 1
= + sin 2θ + C
2 4
θ 1
= + sin θ cos θ + C
2 2
by the trigonometric double-angle identity sin 2θ = 2 sin θ cos θ.
The simplest way to get expressions for sin θ and cos θ in terms of x is to draw 2
a right triangle with an angle θ such that tan θ = x = 1x , as in the drawing on
p p 1+x
x
the right. The hypotenuse must then be 1 + x2 (by the Pythagorean Theorem),
which makes it easy to read off the values of sin θ and cos θ: θ
1
x 1
sin θ = p and cos θ = p
1 + x2 1 + x2
Since θ = tan−1 x, putting the integral back in terms of x yields:
Z
dx 1 1 x 1
2 2
= tan−1 x + p p + C
(1 + x ) 2 2 1 + x2 1 + x2
1 x
= tan−1 x + + C
2 2 (1 + x2 )
Note: An alternative method for getting sin θ and cos θ in terms of x would be to put tan θ = x in the
identity sec2 θ = 1 + tan2 θ to solve for cos θ, then use the identity sin2 θ = 1 − cos2 θ to solve for sin θ.
By completing the square, quadratic expressions in x can be put in one of the forms
a2 ± u2 or u2 − a2 , enabling the use of the corresponding trigonometric substitution.
Example 6.18
Z
dx
Evaluate .
(4 x + 8 x − 5)3/2
2
Solution: This integral cannot be evaluated by using the Power Formula, so try a trigonometric substi-
tution. Complete the square on the expression 4 x2 + 8 x − 5:
4 x2 + 8 x − 5 = 4 ( x2 + 2 x) − 5 = 4 ( x2 + 2 x + 1) − 5 − 4 = 4 ( x + 1)2 − 9
This expression is now of the form u2 − a2 for u = 2 ( x + 1) and a = 3. Use the substitution u = a sec θ,
which means 2 ( x + 1) = 3 sec θ. Then 2 dx = 3 sec θ tan θ dθ and so:
Z Z
dx dx
=
(4 x2 + 8 x − 5)3/2 (4 ( x + 1)2 − 9)3/2
Z Z
3 sec θ tan θ dθ 3 sec θ tan θ dθ
= =
2 (9 sec2 θ − 9)3/2 2 (9 (sec2 θ − 1))3/2
Z Z Z
1 sec θ tan θ dθ 1 sec θ dθ 1 cos θ dθ
= = =
18 tan3 θ 18 tan2 θ 18 sin2 θ
Z
1 1
= csc θ cot θ dθ = − csc θ + C
18 18
To get an expression for csc θ in terms of x, draw a right triangle with )

an angle θ such that sec θ = 2( x3+1) , as in the drawing on the right. +1 p
p 2( x 4 x2 + 8 x − 5
The side opposite θ must then be 4 x2 + 8 x − 5 (by the Pythagorean
θ
Theorem), and hence:
2( x + 1) 3
csc θ = p
4 x2 + 8 x − 5
Putting the integral back in terms of x yields:
Z
dx 1 2( x + 1)
= − p + C
(4 x2 + 8 x − 5)3/2 18 4 x2 + 8 x − 5
x+1
= − p + C
9 4 x2 + 8 x − 5
Note: Trigonometric identities could have been used to obtain csc θ by knowing sec θ.
The following integrals from Section 5.4 might be helpful for the exercises:
Z
tan u du = ln | sec u | + C (6.11)
Z
sec u du = ln | sec u + tan u | + C (6.12)
Z
csc u du = − ln | csc u + cot u | + C (6.13)
Exercises
A
Zp Zp Zp Zp
1. 9 + 4 x2 dx 2. 2 − 3 x2 dx 3. 4 x2 − 9 dx 4. x2 + 2 x + 10 dx
Zp Z 2 Z Z
1 − x2 x dx dx dx
5. dx 6. p 7. p 8. p (a > 0)
x2 x2 − 9 x 1+x 2 x a2 + x2
2
Z 3 Z Z Z
x dx dx dx x2 dx
9. p 10. 2 − 9)3/2
11. 2 2
12. p (a > 0)
2
x +4 (4 x (9 + 4 x ) a2 − x2
Z 3 Zp Z Z
x dx 4 − x2 ( x − 4) dx dx
13. p 14. dx 15. p 16. 2 + 16 x + 15)3/2
9 − x2 x 2
−9 x + 36 x − 32 (4 x
17. Prove formula (6.7) directly by using the substitution u = a sin θ.
18. Prove formula (6.8). 19. Prove formula (6.9).

Rp
20. Show that using the substitution u = a cot θ to evaluate the integral a2 + u2 dx leads to an
antiderivative equivalent to the one in formula (6.8).
Rp
21. Show that using the substitution u = a csc θ to evaluate the integral u2 − a2 dx leads to an
antiderivative equivalent to the one in formula (6.9).
B
Z
dx
22. The integrals p can be evaluated without the use of trigonometric substitutions, by using
x2 ± a2
differentials:
(a) For u2 = x2 ± a2 , show that
dx d ( x + u)
= .
u x+u
(b) Integrate both sides of the result from part (a).
p
Note: In general, many integrals involving x2 ± a2 can be handled with a similar manipulation of
differentials, with varying complexity.
23. According to Newtonian physics the path of a photon grazing the surface of the Sun should be
deflected by the Sun’s gravitational field by an angle θ, given approximately by
¯ Z ¯
¯ 2 GMR 0 dy ¯
θ = ¯¯ 2 2 2 3/2
¯
¯
c ∞ (R + y )
where c = 2.998×108 m/s is the speed of light, G = 6.67×10−11 N/m2 /kg2 is the gravitational constant,
M = 1.99 × 1030 kg is the mass of the Sun, and R = 6.96 × 108 m is the radius of the Sun. Show that
2 GM
θ = = 4.24 × 10−6 radians = 2.43 × 10−4 degrees ≈ 0.875 seconds of arc,
c2 R
where 1 second of arc = 1/3600 of 1 degree.3
3 Albert Einstein published this result in 1911, then showed in 1915 that the true angle should be double that
amount, due to the curvature of space. Experiments verified Einstein’s prediction. See pp.69-71 in S ERWAY, R.A.,
C.J. M OSES AND C.A. M OYER, Modern Physics, Orlando, FL: Harcourt Brace Jovanovich Publishers, 1989.
6.4 Partial Fractions

In the last two sections integrals involving trigonometric functions were simplified by
using various trigonometric identities. For integrals of rational functions—quotients
of polynomials—some algebraic identities (e.g. x2 − a2 = ( x − a)( x + a)) will be useful in
the method of partial fractions. The idea behind this method is simple: replace a
complicated rational function with simpler ones that are easy to integrate.
For example, there is no formula for evaluating the integral
Z
dx
2
x +x
but notice that you can write
1 1 1 1
= = − ,
x2 + x x ( x + 1) x x+1
1
called the partial fraction decomposition of x2 + x
, so that
Z Zµ ¶
dx 1 1
2
= − dx
x +x x x+1
= ln | x | − ln | x + 1 | + C .
There is a systematic way to find this decomposition. First, assume that

1 A B
= +
x ( x + 1) x x+1
for some constants A and B. Get a common denominator on the right side:
1 A ( x + 1) + Bx
=
x ( x + 1) x ( x + 1)
0x + 1 ( A + B) x + A
=
x ( x + 1) x ( x + 1)
Now equate coefficients in the numerators of both sides to solve for A and B:
constant term : A = 1
coefficient of x : A + B = 0 ⇒ B = − A = −1
Thus,
1 A B 1 −1 1 1
= + = + = −
x ( x + 1) x x+1 x x+1 x x+1
as before.
Partial Fractions • Section 6.4 179
The partial fraction method can be discussed in general, and its assumptions proved4 ,
but only the simplest cases—linear and quadratic factors— will be considered here. In
all cases it will be assumed that the degree of the polynomial in the numerator of the
rational function is less than the degree of the polynomial in the denominator.
Start with the most basic case, similar to the example above:
p( x)
Case 1 - Distinct linear factors: A rational function q( x) such that degree ( p( x)) <
degree ( q( x)), where q( x) is a product of n > 1 distinct linear factors
q ( x) = ( a 1 x + b 1 ) ( a 2 x + b 2 ) · · · ( a n x + b n ) ,
can be written as a sum of partial fractions
p ( x) A1 A2 An
= + + ··· +
q ( x) a1 x + b1 a2 x + b2 an x + bn
for some constants A 1 , A 2 , . . . , A n . Those constants can be solved for by getting

a common denominator on the right side of the equation and then equating the
coefficients of the numerators of both sides.
Example 6.19
Z
dx
Evaluate .
x2 − 7 x + 10
Solution: Since x2 − 7 x + 10 = ( x − 2) ( x − 5), then
1 1 A B
= = +
x2 − 7 x + 10 ( x − 2) ( x − 5) x−2 x−5
( A + B ) x + ( −5 A − 2 B )
=
( x − 2) ( x − 5)
so that
coefficient of x : A + B = 0 ⇒ B = −A
1 1
constant term : −5 A − 2B = 1 ⇒ −5 A + 2 A = 1 ⇒ A = − and B =
3 3
Thus,
Z ZÃ !
dx − 13 1
3
2
= + dx
x − 7 x + 10 x−2 x−5
1 1
= − ln | x − 2 | + ln | x − 5 | + C
3 3
4 For example, see Section 5.10 in H ILLMAN, A.P., AND G.L. A LEXANDERSON, A First Undergraduate Course in
Abstract Algebra, 3rd ed., Belmont, CA: Wadsworth Publishing Co., 1983.
If one linear factor in the denominator is repeated more than once, and all other
factors are distinct, then use the following decomposition:
Case 2 - One repeated linear factor + distinct linear factors: A rational

p( x)
function q( x) such that degree ( p( x)) < degree ( q( x)), where q( x) is a product of n > 1
distinct linear factors and one linear factor repeated m > 1 times
q( x) = (ax + b)m (a 1 x + b 1 ) (a 2 x + b 2 ) · · · (a n x + b n ) ,
can be written as as a sum of partial fractions
p ( x) A1 A2 Am B1 Bn
= + + · · · + + + · · · +
q ( x) ax + b (ax + b)2 (ax + b)m a1 x + b1 an x + bn
for some constants A 1 , A 2 , . . . , A m and B1 , . . . , B n . Those constants can be solved

for by the same method as in Case 1.
Example 6.20
Z 2
x +x−1
Evaluate dx .
x3 + x2
Solution: Since x3 + x2 = x2 ( x + 1), then
x2 + x − 1 x2 + x − 1 A B C
= = + 2 +
x3 + x2 x2 ( x + 1) x x x+1
Ax ( x + 1) + B ( x + 1) + Cx2
=
x2 ( x + 1)
( A + C ) x2 + ( A + B ) x + B
=
x2 ( x + 1)
so that
constant term : B = −1
coefficient of x : A + B = 1 ⇒ A = 1 − B = 2
2
coefficient of x : A + C = 1 ⇒ C = 1 − A = −1
Thus,
Z Zµ ¶
x2 + x − 1 2 −1 −1
dx = + 2 + dx
x3 + x2 x x x+1
1
= 2 ln | x | + − ln | x + 1 | + C0 (C0 = generic constant)
x
Case 2 can be extended to more than one repeated factor—the partial fraction de-
composition would then have more terms similar to those for the first repeated factor.
Example 6.21
Z
dx
Evaluate .
x ( x + 1)2
2
Solution: Expanding Case 2 to two repeated factors,

1 A B C D
= + 2 + +
x2 ( x + 1)2 x x x+1 ( x + 1)2
Ax ( x + 1)2 + B ( x + 1)2 + Cx2 ( x + 1) + Dx2
=
x2 ( x + 1)2
( A + C ) x3 + (2 A + B + C + D ) x2 + ( A + 2B) x + B
=
x2 ( x + 1)
so that
constant term : B = 1
coefficient of x : A + 2B = 0 ⇒ A = −2 B = −2
3
coefficient of x : A + C = 0 ⇒ C = −A = 2
2
coefficient of x : 2A + B + C + D = 0 ⇒ D = −2 A − B − C = 1
Thus,
Z Zµ ¶
dx −2 1 2 1
= + 2 + + dx
x2 ( x + 1)2 x x x+1 ( x + 1)2
1 1
= −2 ln | x | − + 2 ln | x + 1 | − + C0 (C0 = generic constant)
x x+1
The partial fraction decompositions for quadratic factors are similar to those for
linear factors, except the numerators in each partial fraction can now contain linear
terms. A factor of the form ax2 + bx + c is considered quadratic only if it cannot be
factored into a product of linear terms (i.e. has no real roots) and a 6= 0.
p( x)
Case 3 - Distinct quadratic factors: A rational function q( x) such that
degree ( p( x)) < degree ( q( x)), and q( x) is a product of n > 1 distinct quadratic factors
q ( x) = ( a 1 x2 + b 1 x + c 1 ) ( a 2 x2 + b 2 x + c 2 ) · · · ( a n x2 + b n x + c n ) ,
p ( x) A 1 x + B1 A 2 x + B2 A n x + Bn
= 2
+ 2
+ ··· +
q ( x) a1 x + b1 x + c1 a2 x + b2 x + c2 a n x2 + b n x + c n
for some constants A 1 , A 2 , . . . , A n and B1 , B2 , . . . , B n . Those constants can be

solved for by the same method as in Case 1.
Example 6.22
Z
dx
Evaluate .
( x + 1) ( x2 + 4)
2
Solution: Neither x2 + 1 nor x2 + 4 has real roots, so by Case 3,

1 Ax + B Cx + D
= 2 + 2
( x2 + 1) ( x2 + 4) x +1 x +4
( Ax + B) ( x2 + 4) + (Cx + D ) ( x2 + 1)
=
( x2 + 1) ( x2 + 4)
( A + C ) x3 + (B + D ) x2 + (4 A + C ) x + (4B + D )
=
( x2 + 1) ( x2 + 4)
so that
coefficient of x3 : A + C = 0 ⇒ C = −A
2
coefficient of x : B + D = 0 ⇒ D = −B
coefficient of x : 4A + C = 0 ⇒ 4A − A = 0 ⇒ A = 0
1 1
constant term : 4B + D = 1 ⇒ 4B − B = 1 ⇒ B = and D = −
3 3
Thus,
Z ZÃ 1
!
dx 3 − 31
= + dx
( x + 1) ( x2 + 4)
2 x2 + 1 x2 + 4
1 1 ³ x´
= tan−1 x − tan−1 + C0
3 6 2
by formula (5.4) in Section 5.4.
A repeated quadratic factor is handled in the same way as a repeated linear factor:
Case 4 - One repeated quadratic factor + distinct quadratic factors: A ra-

p( x)
tional function q( x) such that degree ( p( x)) < degree ( q( x)), where q( x) is a product
of n > 1 distinct quadratic factors and one quadratic factor repeated m > 1 times
q( x) = (ax2 + bx + c)m (a 1 x2 + b 1 x + c 1 ) (a 2 x2 + b 2 x + c 2 ) · · · (a n x2 + b n x + c n ) ,
p ( x) A 1 x + B1 A m x + Bm C1 x + D1 Cn x + Dn
= 2
+ ··· + 2
+ 2
+ ··· +
q ( x) ax + bx + c (ax + bx + c) m a1 x + b1 x + c1 a n x2 + b n x + c n
for some constants A 1 , . . . , A m , B1 , . . . , B m , C 1 , . . . , C n , and D 1 , . . . , D n . Those

constants can be solved for by the same method as in Case 1.
Example 6.23
Z
dx
Evaluate .
( x + 1)2 ( x2 + 4)
2
Solution: Neither x2 + 1 nor x2 + 4 has real roots, and x2 + 1 is repeated, so by Case 4,

1 Ax + B Cx + D Ex + F
= + 2 + 2
( x2 + 1)2 ( x2 + 4) 2
x +1 ( x + 1) 2 x +4
( Ax + B) ( x2 + 1) ( x2 + 4) + (Cx + D ) ( x2 + 4) + (Ex + F ) ( x2 + 1)2
=
( x2 + 1) ( x2 + 4)
with the right side of the equation expanded as
( A + E ) x5 + (B + F ) x4 + (5 A + C + 2E ) x3 + (5B + D + 2F ) x2 + (4 A + 4C + E ) x + (4B + 4D + F )
( x2 + 1)2 ( x2 + 4)
so that equating coefficients of both sides gives
coefficient of x5 : A+E =0 ⇒ E = −A
4
coefficient of x : B+F =0 ⇒ F = −B
3
coefficient of x : 5 A + C + 2E = 0 ⇒ 5A + C − 2A = 0 ⇒ C = −3 A
2
coefficient of x : 5B + D + 2 F = 0 ⇒ 5B + D − 2 F = 0 ⇒ D = −3 B
coefficient of x : 4 A + 4C + E = 0 ⇒ 4 A − 12 A − A = 0 ⇒ A = 0 ⇒ C = 0 and E = 0
1 1 1
constant term : 4B + 4 D + F = 1 ⇒ 4B − 12B − B = 1 ⇒ B = − ⇒ D = and F =
9 3 9
Thus,
Z ZÃ !
dx − 19 1
3
1
9
= + + dx
( x + 1)2 ( x2 + 4)
2 x2 + 1 ( x2 + 1)2 x2 + 4
µ ¶ ³ ´
1 −1 1 1 −1 x 1 −1 x
= − tan x + tan x + + tan + C0
9 3 2 2 ( x2 + 1) 18 2
1 x 1 ³x´
= tan−1 x + 2
+ tan−1 + C0
18 6 ( x + 1) 18 2
where the middle integral on the right is from Example 6.17 in Section 6.3.
When a rational function has a numerator with degree larger than its denominator,
dividing the numerator by the denominator leaves the sum of a polynomial and a new
rational function perhaps satisfying the conditions for Cases 1-4. When the numerator
and denominator have the same degree, a trick like this might be easier.
x2 + 2 ( x2 + 3 x + 2) − 3 x x2 + 3 x + 2 3x 3x
= = − = 1 −
( x + 1) ( x + 2) ( x + 1) ( x + 2) ( x + 1) ( x + 2) ( x + 1) ( x + 2) ( x + 1) ( x + 2)
The last rational function on the right can be integrated using partial fractions.
Exercises
A
Z Z Z Z
dx x+1 dx dx
1. 2. dx 3. 4.
2
x −x x2 − x 2 x2 + 3 x − 2 x2 + x − 6
Z Z Z Z
dx x x−1 x2
5. 6. dx 7. dx 8. dx
x4 − x2 ( x − 2)3 x2 ( x + 1) ( x − 1)2
Z Z Z Z 2
x−2 dx dx ( x − 1)
9. 2 2
dx 10. 11. 12. dx
x ( x − 1) x − x2
4 x + 5 x2 + 4
4 ( x2 + 1)2
B
13. For all numbers a 6= b show that
Z ¯ ¯
dx 1 ¯ x−a¯
= ln ¯¯ ¯ + C.
( x − a) ( x − b) a−b x−b ¯
14. Let q( x) = ( x − a1 ) ( x − a2 ), where a1 6= a2 . Show that
q′ ( x ) 1 1
= + .
q( x ) x − a1 x − a2
15. For q( x) as in Exercise 14, show that
1 1 1
= ′ + ′ .
q( x ) q ( a1 ) ( x − a1 ) q ( a2 ) ( x − a2 )
16. Extend Exercise 15 to three distinct linear factors: if q( x) = ( x − a1 ) ( x − a2 ) ( x − a3 ) then
1 1 1 1
= ′ + ′ + ′ .
q( x ) q ( a1 ) ( x − a1 ) q ( a2 ) ( x − a2 ) q ( a3 ) ( x − a3 )
This result can be extended to any n ≥ 2 distinct factors, though you do not need to prove that.
µ ¶
d 100 1 d 2020 ¡ 2 ¢
17. Find 100 2
. 18. Find ( x − 1)−1 .
dx x − 9 x + 20 dx2020
19. It is possible to use a form of partial fractions to evaluate integrals that are not rational functions.
For example, evaluate the integral Z
dx
x + x4/3
by finding constants A , B and C such that
1 1 A B x−2/3 + C
= = + .
x + x4/3 x (1 + x1/3 ) x 1 + x1/3
Notice how in the second partial fraction the highest power of x in the numerator is one less than in
the denominator, similar to a partial fraction for a quadratic factor in a rational function.
C
20. Find a different way to evaluate the integral in Exercise 19.
Miscellaneous Integration Methods • Section 6.5 185
6.5 Miscellaneous Integration Methods

The integration methods presented so far are considered “standard,” meaning every
calculus student should know them. This section will discuss a few additional meth-
ods, some more common than others. One such method is the Leibniz integral rule
for “differentiation under the integral sign.”5 This powerful and useful method is best
explained with a simple example.
Recall from Section 5.4 that
Z
eα x dx = α1 eα x + C (6.14)
for any constant α 6= 0. This antiderivative involves the variable x and the constant α.
The idea behind the Leibniz rule is to reverse those roles: view α as the variable and x
as a constant. The antiderivative is then seen as a function of α, and so its derivative
can be taken with respect to α. The big leap that this method makes is to move the
differentiation operation inside the integral:6
Z Z Z
d αx d αx
e dx = ( e ) dx = x eα x dx
dα dα
However, differentiating the right side of the formula (6.14) shows that
Z
d d ¡ 1 αx ¢ α ( x eα x ) − 1 · eα x
eα x dx = α
e + C = 2
= α1 x eα x − 1
α2
eα x
dα dα α
Thus, Z
1 1
x eα x dx = α
x eα x − α2
eα x + C
which can be verified via integration by parts with the tabular method:
u dv
x eα x dx
1 αx
1 α
e (+) + ( x) ( α1 eα x )
1 αx
STOP 0 α2
e (–) − (1) ( α12 eα x )
Z
1 1
x eα x dx = α
x eα x − α2
eα x + C X
5 The renowned physicist Richard Feynman (1918-1988) famously lamented that the technique was no longer
being taught. See p.72 in F EYNMAN, R.P., Surely You’re Joking, Mr. Feynman!, New York: Bantam Books, 1986.
6 It can be proved that this is valid when the derivative of the integrand is a continuous function of α, which will
always be the case in this book. See pp.121-122 in S OKOLNIKOFF , I.S., Advanced Calculus, New York: McGraw-
Hill Book Company, Inc., 1939.
What was actually done in the above example? A known integral,

Z
eα x dx = α1 eα x + C ,
was differentiated with respect to α via the Leibniz rule to produce a new integral,
Z
x eα x dx = α1 x eα x − α12 eα x + C ,
with the constant α treated temporarily—only during the differentiation—as a vari-

able. In general, that is how the Leibniz rule is used. Typically this means if you want
to evaluate a certain integral with the Leibniz rule, then you “work backwards” to fig-
ure out which integral you need to differentiate with respect to some constant (e.g. α)
in the integrand.
Example 6.24
Z
dx
Use the Leibniz rule to evaluate .
(1 + x2 )2
Solution: By formula (5.4) in Section 5.4,
Z
dx ¡x¢
= 1
tan−1 + C
a2 + x2 a a
for any constant a > 0. So differentiate both sides with respect to a:

Z
d dx d ¡1 ¡ ¢ ¢
= tan−1 ax + C
da a2 + x2 da a
Z µ ¶
d 1 ¡ ¢ 1
2 2
dx = − a12 tan−1 ax + 1a · x
¡ x ¢2 · − a2
da a + x 1+ a
Z
2a ¡x¢ x
− dx = − a12 tan−1 −
(a2 + x2 )2 a a (a2 + x2 )
Z
dx ¡x¢ x
= 1
2a3
tan−1 + + C
(a2 + x2 )2 a
2a2 (a2 + x2 )
That general formula is useful in itself. In particular, for a = 1,

Z
dx x
= 12 tan−1 x + + C,
(1 + x2 )2 2 (1 + x2 )
which agrees with the result from Example 6.17 in Section 6.3.
Notice that there was no generic constant (e.g. a or α) in the statement of the problem. When that
happens, you will need to figure out where the constant should be in order to use the Leibniz rule.
You can also use differentiation under the integral sign to evaluate definite integrals.
Example 6.25
Z∞
2
1p
Show that e− x dx = 2 π .
0
R∞ 2
Solution: Let I = 0 e− x dx. The integral is convergent, since by Exercise 11 in Section 4.4, for all x
2 2 1
e x ≥ 1 + x2 ⇒ 0 ≤ e− x ≤
1 + x2
R∞ 1 1
implies I is convergent by the Comparison Test, since 0 1+ x 2
dx is convergent (and equals 2 π) by
Example 5.32 in Section 5.5. For α ≥ 0, define
Z∞ 2 x2
α e −α
φ(α) = dx .
0 1 + x2
Then clearly φ(0) = 0, and differentiating under the integral sign shows
Z∞ 2 2 2 x2 Z∞
−2α2 e−α x + e−α dx 1
φ′ (α) = dx ⇒ φ′ (0) = = 2π .
0 1 + x2 0 1 + x2
The substitution y = α x, so that dy = α dx, shows φ(α) can be written as
Z∞ 2
e− y
φ(α) = ¡ y ¢2 dy ⇒ 0 ≤ lim φ(α) ≤ I < ∞ .
α→∞
0 1+ α
Also, for α > 0,
µ ¶ Z∞ −α2 (1+ x2 ) Z∞ 2 2
d 1 −α 2 d e −2α (1 + x2 ) e−α (1+ x )
e φ(α) = dx = dx
dα α dα 0 1 + x2 0 1 + x2
Z∞
2 2 2
= −2 α e − α e−α x dx , now substitute u = α x and du = αdx to get
0
Z∞
2 1 2 2
= −2 α e − α e−u du = −2 e−α I , and so integrating both sides yields
α 0
Z∞ µ ¶ Z∞
d 1 −α 2 2
e φ(α) dα = −2 I e−α dα = −2 I 2 .
0 dα α 0
However, by the Fundamental Theorem of Calculus.

Z∞ µ ¶ ¯∞ µ ¶ µ ¶
d 1 −α 2 1 −α 2 ¯
¯ φ(α) φ(α)
e φ(α) dα = e φ(α) ¯ = lim − lim
0 dα α α α→∞ α e α2 α→0 α e α2
0
µ ¶
φ(α) 0
= 0 − lim → , so by L’Hôpital’s Rule
α→0 α e α2 0
φ′ (α) φ′ (0)
= − lim 2 2
= − = − 12 π .
α→0 eα + 2α2 eα 1+0
Thus,
1p
−2 I 2 = − 12 π ⇒ I = 2 π
which is the desired result.
One immediate consequence of Example 6.25 is that

Z∞
2 p
e−x dx = π
−∞
2
since e−x is an even function. The following example shows another consequence, as
well as how useful substitutions can be in writing integrals in a different form.
Example 6.26
Show that the Gamma function Γ ( t) can be written as
Z∞
2
Γ ( t) = 2 y2 t−1 e− y dy for all t > 0,
0
¡1¢ p
and that Γ 2 = π.
2
Solution: Let x = y , so that dx = 2 y dy. Then x = 0 ⇒ y = 0 and x = ∞ ⇒ y = ∞, so
Z∞ Z∞ Z∞
2 2
Γ ( t) = x t−1 e− x dx = ( y2 ) t−1 e− y 2 y dy = 2 y2 t−1 e− y dy .
0 0 0
¡ ¢
In this form, with the help of Example 6.25 it is now easy to evaluate Γ 21 :
Z∞ Z∞
¡ ¢ 2 2 ¡ p ¢ p
Γ 21 = 2 y1−1 e− y dy = 2 e− y dy = 2 21 π = π
0 0
A function closely related to the Gamma function is the Beta function B( x, y), de-
fined by: Z1
B( x, y) = t x−1 (1 − t) y−1 dt for all x > 0 and y > 0 (6.15)
0
7
It can be shown that
Γ ( x) Γ ( y)
B( x, y) = for all x > 0 and y > 0. (6.16)
Γ ( x + y)
Example 6.27
Show that the Beta function B( x, y) can be written as
Z∞
u x −1
B( x, y) = du .
0 (1 + u) x+ y
1
Solution: Let u = 1−t t , so that t = u
1+ u , 1− t = 1+ u , and dt = du
(1+ u)2
. Then t = 0 ⇒ u = 0 and t = 1 ⇒ u = ∞,
so Z Z∞ ³ µ ¶ Z∞
1 u ´ x −1 1 y−1 du u x −1
B( x, y) = t x−1 (1 − t) y−1 dt = = du .
0 0 1+u 1+u (1 + u)2 0 (1 + u)
x+ y
7 See p.18-19 in R AINVILLE , E.D., Special Functions, New York: Chelsea Publishing Company, 1971.
Another application of substitutions in integrals is in the evaluation of fractional

derivatives. Recall from Section 1.6 that the zero-th derivative of a function is just
the function itself, and that derivatives of order n are well-defined for integer values
n ≥ 1. It turns out that derivatives of fractional orders—e.g. 1/2—can be defined, with
the Riemann-Louiville definition being the most common:
For all 0 < α < 1, the fractional derivative of order α of a function f ( x) is

Zx
dα 1 d f ( t)
f ( x) = dt . (6.17)
dx α Γ (1 − α) dx 0 ( x − t)α
Example 6.28
d 1/2
Calculate ( x) .
dx1/2
Solution: Here α = 21 and f ( x) = x, so that
Zx Zx
d 1/2 1 d t 1 d t dt
( x) = dt = p p
dx1/2 Γ (1 − 1/2) dx
0 ( x − t)
1/2 π dx 0 x−t
¡1¢ p p
since Γ 2 = π by Example 6.26. Use the substitution u = x − t, so that t = x − u2 and dt = −2u du.
p
Then t = 0 ⇒ u = x and t = x ⇒ u = 0, so
Z0 Zp x
d 1/2 1 d ( x − u2 ) (−2u du) 2 d
( x ) = p p = p ( x − u2 ) du
dx1/2 π dx x u π dx 0
¯ u =p x
2 d ¡ ¢ ¯
3 2 d ³ 3/2 ´
= p xu − 1
3u
¯
¯ = p x − 1
3 x3/2
π dx u =0 π dx
2 d ³ 2 3/2 ´ 2 p
= p 3 x = p x
π dx π
Part of the motivation for creating fractional derivatives was to find if it were possible
to take two “half” derivatives to form a “whole” derivative:
µ ¶
d 1/2 d 1/2 d
f ( x ) = f ( x)
dx1/2 dx1/2 dx
It is left as an exercise to show that the above relation does hold for the function
f ( x) = x. Derivatives with fractional order 0 < α < 1 and integer order n ≥ 1 can be
combined by taking the derivative of integer order first:8
µ ¶
d n+α dα d n
f ( x) = f ( x)
dxn+α dxα dxn
8 For more details about fractional derivatives, as well as examples of their applications in physics and engineering,
see O LDHAM , K.B. AND J. S PANIER, The Fractional Calculus, New York: Academic Press, 1974.
Recall from Section 6.3 that the trigonometric substitution x = r cos θ —or its sister
substitution x = r sin θ —was motivated by trying to find the area of a circle of radius
r . To simplify matters, let r = 1 so that points on the unit circle can be identified with
the angle θ via that substitution, with θ as shown in Figure 6.5.1(a) below.
y y
slope = 1
slope = 2
( x, y) = (cos θ, sin θ) slope = 1
2
1
slope = 0
θ
x x
0 1 −1 0 1
slope = − 21
slope = −2
slope = -1
(a) Identify points by angle (b) Identify points by slope
Figure 6.5.1 Points on the unit circle x2 + y2 = 1
Figure 6.5.1(b) shows a different identification of points on the unit circle—by slope.
This will be the basis for a half-angle substitution for evaluating certain integrals.
Let A be the point (−1, 0), then for any other point P on the unit circle draw a line
from A through P until it intersects the line x = 1, as shown in Figure 6.5.2 below:
y y
x=1
P x=1
t2
2
p 1+t
1+
p
t t
θ/2 θ θ/2 θ
x x
A 1 0 1 A 1 0 1
(a) 0 ≤ θ ≤ π
2 (b) π
2 <θ<π
¡θ¢
Figure 6.5.2 Half-angle substitution: t = tan 2 = slope of AP
From geometry you know that the inscribed angle that the line AP makes with the
x-axis is half the measure of the central angle θ . So the slope of AP is the tangent of
that angle: tan 21 θ = 1t = t, which is measured along the y-axis and can take any real
value. Each point on the unit circle—except A —can be identified with that slope t.
Figure 6.5.2 shows only positive slopes—reflect the picture about the x-axis for neg-
ative slopes. The figure shows that
t 1
sin 12 θ = p and cos 21 θ = p
1 + t2 1 + t2
so that by the double-angle identities for sine and cosine,
t 1 2t
sin θ = 2 sin 21 θ cos 12 θ = 2 p p =
1 + t2 1 + t2 1 + t2
and
1 t2 1 − t2
cos θ = cos2 21 θ − sin2 21 θ = − = .
1 + t2 1 + t2 1 + t2
Since θ = 2 tan−1 t, then
¡ ¢ 2 dt
dθ = d 2 tan−1 t = .
1 + t2
Below is a summary of the substitution:
Half-angle substitution: The substitution t = tan 21 θ yields:
2t 1 − t2 2 dt
sin θ = cos θ = dθ =
1 + t2 1 + t2 1 + t2
The half-angle substitution thus turns rational functions of sin θ and cos θ into ratio-
nal functions of t, which can be integrated using partial fractions or another method.
Example 6.29
Z
dθ
Evaluate .
1 + sin θ + cos θ
Solution: Using t = tan 12 θ, the denominator of the integrand is
1 + t2 2t 1 − t2 2t + 2
1 + sin θ + cos θ = + + =
1 + t2 1 + t2 1 + t2 1 + t2
so that
Z
Z 2 dt Z
dθ 1+ t 2 dt
= 2 t +2
=
1 + sin θ + cos θ t+1
1+ t 2
= ln | t + 1 | + C
¯ ¯
= ln ¯ tan 21 θ + 1 ¯ + C
Example 6.30
Z
dθ
Evaluate .
3 sin θ + 4 cos θ
Solution: Using t = tan 21 θ, the integral becomes
Z
Z 2 dt Z
dθ 1+ t 2 −1
= = dt
3 sin θ + 4 cos θ 3 2t
+4 1− t 2 2 t2 − 3 t − 2
1+ t 2 1+ t 2
Z Zµ ¶
−1 A B
= dt = + dt
(2 t + 1) ( t − 2) 2t + 1 t−2
where
coefficient of t : A + 2B = 0 ⇒ A = −2 B
1 2
constant term : −2 A + B = −1 ⇒ 4 B + B = −1 ⇒ B = − and A =
5 5
Thus,
Z ZÃ 2
!
dθ 5 − 15 1 1
= + dt = ln | 2 t + 1 | − ln | t − 2 | + C
3 sin θ + 4 cos θ 2t + 1 t−2 5 5
1 ¯¯ ¯ 1 ¯ ¯
= ln 2 tan 21 θ + 1 ¯ − ln ¯ tan 21 θ − 2 ¯ + C
5 5
By the half-angle substitution t = tan 12 θ ,

2t 2t
sin θ 1 + t2 1 + t2 = t
= =
1 + cos θ 1+t 2
1−t 2 2
+
1+t 2 1+t 2 1 + t2
which yields the useful half-angle identities:9
sin θ 1 − cos θ
tan 21 θ = = (6.18)
1 + cos θ sin θ
Example 6.31
Z
sin θ
Evaluate dθ .
1 + cos θ
Solution: Though you could use the half-angle substitution t = tan 12 θ, it is easier to use the half-angle
identity (6.18) directly, since
Z Z
sin θ ¯ ¯
dθ = tan 21 θ dθ = 2 ln ¯ sec 12 θ ¯ + C
1 + cos θ
by formula (6.11) in Section 6.3.
9 For a different derivation, see pp.79-80 in C ORRAL , M., Trigonometry, http://mecmath.net/trig/, 2009.
Exercises
A
Z Z Z Z
1 − 2 cos θ dθ dθ dθ
1. dθ 2. 3. 4.
sin θ 3 − 5 sin θ 2 − sin θ 4 + sin θ
Z Z Z Z
sin θ dθ dθ dθ
5. dθ 6. 7. 8.
2 − sin θ 5 − 3 cos θ 1
Z∞ + sin θ − cos θ
Z1 − sin θ + cos θ
Z Z
cot θ 1 − cos θ − x2 /2
∞ 6
9. dθ 10. dθ 11. e dx 12. x2 e− x dx
1 + sin θ 3 sin θ −∞ −∞
Z
sin θ
13. Consider the integral dθ from Example 6.31.
1 + cos θ
(a) Evaluate the integral using the substitution u = 1 + cos θ.
(b) Evaluate the integral using the half-angle substitution t = tan 21 θ.
(c) Show that the answers from parts (a) and (b) are equivalent to the result from Example 6.31.
B Z
dθ
14. Evaluate the integral from Example 6.30 by noting that
3 sin θ + 4 cos θ
5
Z Z 4
dθ dθ
= ¡3 ¢
3 sin θ + 4 cos θ 5 5 sin θ + 54 cos θ φ
Z 3
dθ
= ¡ ¢
5 cos φ sin θ + sin φ cos θ
Z Z
dθ 1
= = csc (θ + φ) dθ
5 sin (θ + φ) 5
by the sine addition formula, where φ is the angle in the right triangle shown above. Complete the
integration and show that your answer is equivalent to the result from Example 6.30.
15. Show directly from the definition of the Beta function that B( x, y) = B( y, x) for all x > 0 and y > 0.
16. Show that the Beta function B( x, y) can be written as
Zπ/2
B( x, y) = 2 sin2 x−1 (θ) cos2 y−1 (θ) dθ for all x > 0 and y > 0.
0
17. Use Exercise 16 and formula (6.16) to show that

µ ¶ µ ¶
m+1 n+1
Zπ/2 Γ Γ
2 2
sin m θ cosn θ dθ = ³m+n ´ for all m > −1 and n > −1.
0 2Γ +1
2
18. Use Exercise 28 from Section 6.1, as well as Exercise 17 above, to show that for m = 1, 2, 3, ...,
Zπ/2 p ¡ ¢ Zπ/2 p
2m
π Γ m + 12 2 m +1 π ( m!)
sin θ dθ = and sin θ dθ = ¡ ¢ .
0 2 (m!) 0 2 Γ m + 23
Z∞ Z∞
ln x xa Γ (a + 1)
19. Show that dx = 0. 20. Show that dx = for a > 1.
0 1 + x2 0 a x (ln a)a+1
21. Use the result from Example 6.28 to show that
µ ¶
d 1/2 d 1/2 d
( x) = 1 = ( x) .
dx1/2 dx1/2 dx
d 1/2 d 1/3
22. Calculate ( c) for all constants c. 23. Calculate ( x) .
dx1/2 dx1/3
Z1
1 ¡ ¢
24. Show that p dx = n1 B n1 , 12 for n ≥ 1.
0 1−x n
25. Show that the Gamma function Γ ( t) can be written as

Z∞
Γ ( t) = p t u t−1 e− pu du for all t > 0 and p > 0.
0
26. Show that the Gamma function Γ ( t) can be written as

Z1 µ µ ¶¶ t−1
1
Γ ( t) = ln du for all t > 0.
0 u
27. Using the result from Exercise 27 in Section 6.1 that

Z
e ax (a cos bx + b sin bx)
e ax cos bx dx =
a2 + b2
for all constants a and b 6= 0, differentiate under the integral sign to show that for all α > 0
Z∞
2α
x e− x sin α x dx = .
0 (1 + α2 )2
C
28. Use the Leibniz rule and formula (6.8) from Section 6.3 to show that for all a > 0,
Z p
dx ¯ ¯
p = ln ¯ x + a2 + x2 ¯ + C .
2
a +x 2
29. Use Example 6.27 to show that the Beta function satisfies the relation
Z1
t− x + t x−1
B( x, 1 − x) = dt for all 0 < x < 1.
0 1+t
Z∞ Z∞
u x −1 t− x
(Hint: First use a substitution to show that du = dt.)
0 1+u 0 1+t
30. Show that for all a > −1,
Zπ/2
dθ π
= p .
0 1 + a sin2 θ 2 1+a
Numerical Integration Methods • Section 6.6 195
6.6 Numerical Integration Methods

Section 5.2 showed how to obtain exact values for definite integrals of some simple
functions (low-degree polynomials) by using areas of rectangles. For functions with no
closed-form antiderivative, the rectangle method typically produces an approximate
value of the definite integral—the more rectangles, the better the approximation.
For example, suppose you wanted to evaluate y
y = sin ( x2 )
Zpπ 1
sin ( x2 ) dx
0
with the rectangle method. This means finding the

area of the shaded region in the figure on the right. p x
0 π
Computers have eliminated the need to do these sorts
of calculations by hand. Though the rectangle method is simple to implement in a
traditional programming language (e.g. via a looping construct), there are easier ways
in a domain-specific language (DSL) geared toward scientific computing. One such
DSL is MATLAB® , or its free open-source clone Octave.10
Implementing the rectangle method from scratch in Octave is a one-liner. For exam-
p
ple, suppose you divide the interval [0, π] into 100,000 (105) subintervals of equal
length, producing 100,001 (1 + 1e5 in scientific notation) equally spaced points in
p p
[0, π] (including 0 and π ). First use the left endpoints of the 105 subintervals:
octave> sum(sin(linspace(0,sqrt(pi),1+1e5)(1:end-1).^2)*sqrt(pi)/1e5)
ans = 0.8948314693913354
Now use the right endpoints:
octave> sum(sin(linspace(0,sqrt(pi),1+1e5)(2:end).^2)*sqrt(pi)/1e5)
ans = 0.8948314693913354
Finally, use the midpoints of the subintervals:
octave> sum(sin((linspace(0,sqrt(pi),1+1e5)(1:end-1)+sqrt(pi)/2e5).^2)*
sqrt(pi)/1e5)
ans = 0.8948314695305527
The true value of the integral up to 15 decimal places is 0.894831469484145, so all

three approximations are accurate to 9 decimal places.
10 Octave is freely available at https://www.gnu.org/software/octave/
The syntax in the above commands can be explained with some examples. The fol-
lowing command creates 4 equally spaced points in the interval [1, 7] (including x = 1
and x = 7), thus dividing [1, 7] into 3 subintervals each of length (7 − 1)/3 = 2:
octave> linspace(1,7,4)
ans =
1 3 5 7
Get all but the last number in the above list:11
octave> linspace(1,7,4)(1:end-1)
ans =
1 3 5
Now square each number in that list of numbers (the dot before the exponentiation
operator ^ applies the squaring operation ^2 element-wise in the list):
octave> linspace(1,7,4)(1:end-1).^2
ans =
1 9 25
Now take the sine of each of those squared numbers (measured in radians):
octave> sin(linspace(1,7,4)(1:end-1).^2)
ans =
0.8414709848078965 0.4121184852417566 -0.132351750097773
Now multiply each of those numbers (the heights of the rectangles) by the width (2) of
each rectangle, then add up those areas:
octave> sum(sin(linspace(1,7,4)(1:end-1).^2)*2)
ans = 2.24247543990376
11 MATLAB would require you to use two steps:
MATLAB>> x = linspace(1,7,4);
MATLAB>> x(1:end-1)
Before the advent of modern computing, the rectangle method was considered ineffi-
cient, and so alternative methods were created. Two such methods are the trapezoid
rule and Simpson’s rule. The idea behind both methods is to take advantage of a
nonlinear function’s changing slope by using nonrectangular regions. For the trape-
zoid rule those regions are trapezoids, while Simpson’s rule uses quasi-rectangular
regions whose top edges are parabolas, as shown in Figure 6.6.1:
y y y
y = f ( x) y = f ( x) y = f ( x)
x x x
a x i x i +1 b a x i x i +1 b a x i x i +1 x i +2 b
(a) rectangle method (b) trapezoid rule (c) Simpson’s rule
Rb
Figure 6.6.1 Comparison of numerical integration methods for a f ( x) dx
For a partition P = {a = x0 < x1 < · · · < xn−1 < xn = b} of an interval [a, b] into n ≥ 1
subintervals of equal width h = ( b − a)/ n, let yi = f ( x i ) for i = 0, 1, . . ., n. The trapezoid
rule adds up the areas of trapezoids on each subinterval [ x i , x i+1 ], with the top edge
being the line segment joining the points ( x i , yi ) and ( x i+1 , yi+1 ). The approximation
formula is straightforward to derive, based on areas of trapezoids:
Trapezoid rule:
Zb
h
f ( x) dx ≈ ( y0 + 2 y1 + 2 y2 + · · · + 2 yn−1 + yn )
a 2
Simpson’s rule depends on pairs of neighboring subintervals: [ x0 , x1 ] and [ x1 , x2 ],

[ x2 , x3 ] and [ x3 , x4 ], . . . , [ xn−2 , xn−1 ] and [ xn−1 , xn ]. Thus, n ≥ 2 must be even. The top
edge of the region over each pair [ x i , x i+1 ] and [ x i+1 , x i+2 ] is the unique parabola joining
the 3 points ( x i , yi ), ( x i+1 , yi+1 ), and ( x i+2 , yi+2 ). The approximation formula is then:12
Simpson’s rule:
Zb
h
f ( x) dx ≈ ( y0 + 4 y1 + 2 y2 + 4 y3 + 2 y4 + · · · + + 2 yn−2 + 4 yn−1 + yn )
a 3
12 For a full derivation of both formulas, see pp.144-149 in H ORNBECK , R.W., Numerical Methods, New York:
Quantum Publishers, Inc., 1975.

Example 6.32
Zpπ
Approximate the value of sin ( x2 ) dx by using the trapezoid rule and Simpson’s rule with n = 105
0
subintervals.
p
Solution: Since x0 = 0 and x n = π, then y0 = sin ( x02 ) = sin 0 = 0 and yn = sin ( x2n ) = sin π = 0. Thus, y0
and yn contribute nothing to the summation formulas for both rules. In particular the trapezoid rule
approximation becomes
Zpπ Ã !
h nX−1
2
sin ( x ) dx ≈ (0 + 2 y1 + 2 y2 + · · · + 2 yn−1 + 0) = h · yk
0 2 k =1
which is simple to implement in Octave:
octave> x = linspace(0,sqrt(pi),1+1e5);
octave> h = sqrt(pi)/1e5;
octave> h*sum(sin(x(2:end-1).^2))
ans = 0.8948314693913405
Likewise, the Simpson’s rule approximation becomes

Zpπ
h
sin ( x2 ) dx ≈ (4 y1 + 2 y2 + 4 y3 + 2 y4 + · · · + + 2 yn−2 + 4 yn−1 )
0 3
which can be implemented easily by using Octave’s powerful indexing features:
octave> (h/3)*(4*sum(sin(x(2:2:end-1).^2)) + 2*sum(sin(x(3:2:end-1).^2)))

ans = 0.8948314694841457
In the above command, the statement x(3:2:end-1) allows you to skip every other element in the list
x after position 3, by moving up the list in increments of 2 positions all the way to the next-to-last
position in the list (end-1). Similarly for x(2:2:end-1), which starts at position 2 and then moves up in
increments of 2.
Note that Simpson’s rule gives essentially the true value in this case, and the value from the trapezoid
rule is virtually the same as the value produced by the built-in trapz function in Octave/MATLAB:
octave> trapz(x,sin(x.^2))
ans = 0.8948314693913402
In general you are better off using these sorts of built-in functions instead of implementing your own.
Typically Simpson’s rule is slightly more efficient than the trapezoid rule, which is
slightly more efficient than the rectangle method. However, in the above examples all
the approximations were accurate to at least 9 decimal places (equivalent to getting
the distance between Detroit and Chicago correct within the thickness of a toothpick).
The running time of each calculation was only a few thousandths of a second. Modern
computing has generally made the efficiency differences negligible.
Notice that the approximations in the rectangle method, the trapezoid rule and
Simpson’s rule can all be written as linear combinations of function values f (a i ) mul-
tiplied by “weights” w i :
Zb X
n
f ( x) dx ≈ w i f (a i )
a i=0
For example, the weights in Simpson’s rule are w i = h3 , 23h , or 43h , depending on the
points a i in the interval [a, b]. The method of Gaussian quadrature transforms an
integral over any interval [a, b] into an integral over the specific interval [−1, 1] and
then uses a standard set of points in [−1, 1] and known weights for those points:13
Rb
Gaussian quadrature: Transform the integral a f ( x) dx into an integral over
[−1, 1] by means of the substitution u = b−1 a (2 x − a − b), so that x = b−2 a u + a+2 b and
dx = b−2 a du. Then
Zb Z1
b−a b−a X
n
f ( x) dx = g( u) du ≈ w i g(a i )
a 2 −1 2 i=1
¡ ¢
where g( u) = f b−2 a u + a+2 b , with the points a 1 , . . ., a n and weights w1 , . . ., wn
given in Table 6.1 below for any choice of 2 ≤ n ≤ 10 points in [−1, 1].
Table 6.1 Table of Gaussian quadrature points and weights
n a1 , ..., a n w1 , ..., wn n a1 , ..., a n w1 , ..., wn

2 ±0.577350 1 8 ±0.183435 0.362684
3 0 8/9 ±0.525532 0.313707
±0.774597 5/9 ±0.796666 0.222381
4 ±0.339981 0.652145 ±0.960290 0.101229
±0.861136 0.347855 9 0 0.330239
5 0 0.568889 ±0.324253 0.312347
±0.538469 0.478629 ±0.613371 0.260611
±0.906180 0.236927 ±0.836031 0.180648
6 ±0.238619 0.467914 ±0.968160 0.081274
±0.661209 0.360762 10 ±0.148874 0.295524
±0.932470 0.171324 ±0.433395 0.269267
7 0 0.417959 ±0.679410 0.219086
±0.405845 0.381830 ±0.865063 0.149451
±0.741531 0.279705 ±0.973907 0.066671
±0.949108 0.129485
13 The details are beyond the scope of this book. See Chapter 4 in R ALSTON, A. AND P. R ABINOWITZ , A First Course
in Numerical Analysis, 2nd ed., New York: McGraw-Hill, Inc., 1978. See also Table 1 in S TROUD, A.H. AND D.
S ECREST, Gaussian Quadrature Formulas, Englewood Cliffs, NJ: Prentice-Hall, Inc., 1966.
Example 6.33
Z2
dx
Approximate the value of 3
by using Gaussian quadrature with n = 4 points.
0 1+x
Solution: For a = 0 and b = 2, use the substitution u = b−1 a (2 x − a − b) = x − 1, so that x = u + 1 and dx = du.
Thus, g(u) = f (u + 1) = 1+( u1+1)3 . Using n = 4 in Table 6.1, the points a i and weights w i are
a1 = −0.339981 w1 = 0.652145
a2 = 0.339981 w2 = 0.652145
a3 = −0.861136 w3 = 0.347855
a4 = 0.861136 w4 = 0.347855
and so
Z2 Z Z1 4 4
dx 2−0 1 du X X 1
3
= g ( u ) du = 3
≈ w i g( a i ) = wi ·
0 1+x 2 −1 −1 1 + ( u + 1) i =1 i =1 1 + (a i + 1)3
0.652145 0.652145 0.347855 0.347855

≈ + + +
1 + (−0.339981 + 1)3 1 + (0.339981 + 1)3 1 + (−0.861136 + 1)3 1 + (0.861136 + 1)3
≈ 1.091621
The true value of the integral to six decimal places is 1.090002.

Using more points (e.g. n = 7) is easy to implement in Octave, using element-wise operations on arrays:
octave> a = [ 0 -0.405845 0.405845 -0.741531 0.741531 -0.949108 0.949108 ];

octave> w = [ 0.417959 0.381830 0.381830 0.279705 0.279705 0.129485 0.129485 ];
octave> sum(w./(1 + (a+1).^3))
ans = 1.090016688064804
Gaussian quadrature can be applied to improper integrals. For example,

Z∞ Xn
−x
f ( x) e dx ≈ w i f (a i )
0 i=1
using the points a i and weights w i in Table 6.214 for n = 3, 4, or 5 points in [0, ∞):
R∞
Table 6.2 Table of Gaussian quadrature points and weights for 0 f ( x) e− x dx
n a1 , ..., a n w1 , ..., wn n a1 , ..., a n w1 , ..., wn n a1 , ..., a n w1 , ..., wn

3 0.415775 0.711093 4 0.322548 0.603154 5 0.263560 0.521756
2.294280 0.278518 1.745761 0.357419 1.413403 0.398667
6.289945 0.010389 4.536620 0.038888 3.596426 0.075942
9.395071 0.000539 7.085810 0.003612
12.640801 0.000023
14 See Table 6 in S TROUD, A.H. AND D. S ECREST , Gaussian Quadrature Formulas.

Example 6.34
Z∞
Approximate the value of x5 e− x dx by using Gaussian quadrature with n = 3 points in Table 6.2.
0
Solution: For n = 3, Table 6.2 gives a1 = 0.415775, a2 = 2.294280, a3 = 6.289945, and w1 = 0.711093,
w2 = 0.278518, w3 = 0.010389. Then for f ( x) = x5 ,
Z∞ n
X n
X
x5 e− x dx ≈ wi f (a i ) = w i a5i
0 i =1 i =1
≈ 0.711093 (0.415775)5 + 0.278518 (2.294280)5 + 0.010389 (6.289945)5

≈ 119.9974709727211
The true value is Γ (6) = 5! = 120.

Note: The points a i in Table 6.2 are the roots of the Laguerre polynomials of degree n.
Exercises
A
1. A simple pendulum of length l swings through an angle of 90◦ on either side of the vertical with
period P , given by s Z
l π/2 dθ
P = 4 p
g 0 1 − 0.5 sin2 θ
where g = 9.8 m/s2 is the acceleration due to gravity. Use the rectangle method
p (with left endpoints),
the trapezoid rule, and Simpson’s rule to write P as a constant multiple of l / g . Preferably, use a
computer and n = 105 subintervals of equal width (or n = 10 subintervals if calculating by hand).
2. Repeat Exercise 1 using Gaussian quadrature with n = 5 points.
Zpπ
3. Approximate the value of sin ( x2 ) dx by using Gaussian quadrature with n = 7 points.
0
4. Repeat Exercise 3 with n = 9 points. 5. Repeat Exercise 3 with n = 10 points.
6. The points a i in Table 6.1 for Gaussian quadrature are the roots of the Legendre polynomials P n ( x),
with P0 ( x) = 1, P1 ( x) = x, and P n ( x) defined for integers n ≥ 2 by the recursion formula
n P n ( x) = (2n − 1) x P n−1 ( x) − (n − 1) P n−2 ( x)
(a) Write out P n ( x) explicitly in standard polynomial form for n = 2, 3, 4, 5.

(b) Verify that the roots of P n ( x) match the n points a1 , ..., a n in Table 6.1 for n = 2, 3, 4, 5.
R1 R1 R1
(c) With no calculations, explain why −1 P2 ( x) P3 ( x) dx = −1 P3 ( x) P4 ( x) dx = −1 P4 ( x) P5 ( x) dx = 0.
(d) For n = 0, 1, 2, 3, verify that
Z1
2
P n2 ( x) dx = .
−1 2n + 1
7. Repeat Example 6.34 with n = 4 points.
8. Repeat Example 6.34 with n = 5 points.
Z∞
9. Use Table 6.2 to approximate the value of ln (1 + x) e− x dx with n = 5 points.
0
CHAPTER 7
Analytic Geometry and Plane Curves
7.1 Ellipses
If you were to ask a random person “What is a circle?” a typical response would be to
kick the can down the road: “Something that’s round.” There is a simple definition:
A circle is the set of all points in a plane that are a fixed distance from a fixed point
in that plane. The fixed point is the center of the circle, while the fixed distance
from the center is the circle’s radius.
Similarly, the question “What is an ellipse?” would likely be answered with “an oval,”
“something egg-shaped,” or “a squished circle.” A precise definition would be:
An ellipse is the set of all points in a plane such that the sum of the distances from
each of those points to two fixed points in the plane is the same constant. The two
fixed points are the foci—plural of focus—of the ellipse.
Figure 7.1.1 illustrates the above definitions, with a point P moving along each curve.
P P
r d1 d2
C F1 F2
(a) Circle: center C , radius r = constant (b) Ellipse: foci F1 and F2 , d1 + d2 = constant
Figure 7.1.1 The circle and its “squished” sibling the ellipse
Along the ellipse the sum d 1 + d 2 of the distances from P to the foci remains constant.
202
Ellipses • Section 7.1 203
The circle’s definition makes it easy to imagine its shape, especially for anyone who
has drawn a circle with a compass. The definition of the ellipse, on the other hand,
might not immediately suggest an “oval” shape. Its shape becomes apparent when
physically constructing an ellipse by hand, using only the definition. Stick two pins in
a board and tie the ends of a piece of string to the pins, with the string long enough so
that there is some slack (see Figure 7.1.2(a)). The pins will be the foci of the ellipse.
P
(a) Two pins and string on a board (b) Pull the string taut with a pencil
Figure 7.1.2 Construction of an ellipse
Pull the string taut with a pencil whose point is touching the board, then move the
pencil around as far as possible on all sides of the pins. The drawn figure will be an
ellipse, as in Figure 7.1.2(b). The length of the string is the constant sum d 1 + d 2 of
distances from points on the ellipse to the foci. The symmetry of the ellipse is obvious.
There is some terminology connected with ellipses. The principal axis is the line
containing the foci, and the center is halfway between the foci, as in Figure 7.1.3:
major axis principal

axis
V1 F1 minor C (center) F2 V2
(vertex) (focus) axis (focus) (vertex)
Figure 7.1.3 The parts of an ellipse
The vertexes are the points where the ellipse intersects the principal axis. The
major axis is the chord joining the vertexes, and the minor axis is the chord through
the center that is perpendicular to the major axis. The two semi-major axes are the
halves of the major axis joining the center to the vertexes (CV1 and CV2 in Figure
7.1.3). Likewise the semi-minor axes are the two halves of the minor axis. A chord
through the center is a diameter. Notice that a circle is the special case of an ellipse
with identical foci (i.e. the foci and center are all the same point).
204 Chapter 7 • Analytic Geometry and Plane Curves §7.1
Ellipses appear in nature (e.g. the orbits of planets around the Sun) and in many
applications. The ancient Greeks were able to derive many properties of the ellipse
from its purely geometric definition.1 Nowadays those properties are typically derived
using methods from analytic geometry—the study of geometric objects in the context of
coordinate systems.2 You have already seen the equation of an ellipse in the x y-plane
2 y2
centered at the origin: ax 2 + b2 = 1, where a > b > 0, with the x-axis as the principal
axis. The equation is straightforward to derive from the definition of the ellipse.
In the x y-plane, let the foci of an ellipse be the y
points (± c, 0) for some c > 0, so that the center is the ( x, y)
origin (0, 0) and the x-axis is the principal axis, as in d1 d2
the figure on the right. Denote by 2a the constant
sum d 1 + d 2 of the distances from points ( x, y) on the x
(− c, 0) ( c, 0)
ellipse to the foci, with a > 0. Notice that a > c, since
the distance 2 c between the foci must be less than
d 1 + d 2 = 2a. Then by the distance formula,
d 1 + d 2 = 2a
q q
( x + c)2 + y2 + ( x − c)2 + y2 = 2a
µq ¶2 µ q ¶2
2
( x − c) + y 2 2
= 2 a − ( x + c) + y 2
q
( x − c )2 + ✓y✓2 = 4a2 − 4a ( x + c)2 + y2 + ( x + c)2 + ✓
y✓2
q
4a ( x + c)2 + y2 = 4a2 + ( x + c)2 − ( x − c)2
q
4a ( x + c)2 + y2 = ✓
✓ 4 a2 + ✓
4 xc
q
( x + c)2 + y2 = a + ac x
✟ + c2 + y2 = a2 + 2 cx✟ c2 2
x2 + ✟2 cx ✟ + a2
x
³ 2
´
1 − ac 2 x2 + y2 = a2 − c2
x2 y2
+ = 1
a2 a2 − c 2
x2 y2
2
+ 2 = 1 where b2 = a2 − c2 (and so a > b > 0) X
a b
1 The word ellipse is in fact due to the Greek astronomer and geometer Apollonius of Perga (ca. 262-190 B.C.),
which seems an improvement over the name “thyreos” that Euclid (ca. 360-300 B.C.) had given the shape.
2 Pioneered by the French mathematician and philosopher René Descartes (1596-1650), for whom the Cartesian
coordinate system is named. The proposition “I think, therefore I am” (Cogito, ergo sum) is due to Descartes.
2 y2
The graph of the resulting ellipse ax2 + b2 = 1 y
with a > b > 0 and foci at (± c, 0) is shown in Fig-
x2 y2
ure 7.1.4. Since the the x-axis is principal axis b
a2
+ b2 = 1
then the vertexes are found by setting y = 0: x =
±a. The vertexes are thus (±a, 0), so the major
x
axis goes from (−a, 0) to (a, 0) and has length 2a −a −c 0 c a
(i.e. the semi-major axis has length a). Simi-
larly, setting x = 0 shows the minor axis goes
−b
from (0, − b) to (0, b), (i.e. the semi-minor axis has
length b).pSince a > b > 0 and b2 = a2 − c2 , Figure 7.1.4
then c = a2 − b2 . Thus, for any ellipse of the
2 y2
p
form ax2 + b2 = 1 with a > b > 0, the foci will be at (± c, 0), where c = a2 − b2 . The foci
can be used to define an important geometric property of the ellipse:
The eccentricity of an ellipse is the ratio of the distance between the foci to the
x2 y2
length of the major axis. In the case of an ellipse a2
+ b2 = 1 with a > b > 0, the
eccentricity e is given by p
c a2 − b 2
e = = .
a a
The eccentricity e is a measure of how “oval” an ellipse is, with 0 < e < 1. The
boundary case e = 0 is a circle, while e = 1 is a line segment; an ellipse is somewhere
in between—the closer e gets to 1, the more “squished” the ellipse. See Figure 7.1.5.
C F1 F2 F1 F2 F1 F2
(a) Circle: e = 0 (b) Ellipse: e = 0.5 (c) Ellipse: e = 0.75 (d) Segment: e =
1
Figure 7.1.5 Eccentricity e
Earth’s elliptical orbit around the Sun is almost circular: the eccentricity is 0.017.
Only the orbits of Venus and Neptune (both at 0.007) have a lower eccentricity among
the nine planets in the solar system, while Pluto’s (0.252) has the highest.
x2 y2
It is left as an exercise to show that the ellipse a2
+ b2 = 1 with a > b > 0 can be
written in terms of the eccentricity e:
y2 = (1 − e2 ) (a2 − x2 ) (7.1)
Example 7.1
x2 y2
Find the area inside the ellipse a2
+ b2 = 1.
Solution: By symmetry the area will be four times the area in the first quadrant. Solving for y in the
equation of the ellipse gives
s
b2 x 2 x2 bp 2
y2 = b 2 − ⇒ y = b 1− = a − x2
a2 a2 a
for the upper hemisphere. Thus,

Za Z
4b a p 2
Area = 4 y dx = a − x2 dx
0 a 0
µ ³x´ ¯a ¶
4b a2 1 p ¯
= sin−1 + x a2 − x2 ¯
¯ (by formula (6.7) in Section 6.3)
a 2 a 2 0
µ ¶
4b a2 π
= + 0 − (0 + 0)
a 2 2
= π ab
A remarkable feature of the ellipse is the reflection prop- P

erty: light shone from one focus to any point on the ellipse
will reflect to the other focus. Figure 7.1.6 shows light em-
anating from the focus F1 and reflecting off a point P on
F1 F2
the ellipse to the other focus F2 . Fermat’s Principle from
Example 4.4 in Section 4.1 showed that the incoming angle
θ1 (angle of incidence) of light from point A will equal the Figure 7.1.6
outgoing angle θ2 (angle of reflection) to point B for light re-
flecting off a flat reflective surface at point P , as in Figure 7.1.7(a). Fermat’s Principle
also applies to curved surfaces—e.g. an ellipse—with the angles measured relative to
the tangent line to the curve at the point of reflection, as in Figure 7.1.7(b).
A A
B B
tangent
θ1 θ2 line
θ1 θ2 P
P
(a) Flat surface (b) Curved surface
Figure 7.1.7 Fermat’s Principle for reflection: θ1 = θ2

Notice that Fermat’s Principle is equivalent to saying that the angles α1 and α2 that
the light’s path makes with the normal line through the point of reflection are equal,
since each angle would equal 90◦ − θ , as in Figure 7.1.8(a):
y n = normal line
tangent line
normal line P ( x0 , y0 )
A B b
α1
α1 α2 tangent α2
θ θ line F1 F2
x
P −a −c 0 N c a
(a) α1 = α2 (b) Ellipse: show that α1 = α2
Figure 7.1.8 Fermat’s Principle with normal line
Thus, to prove the reflection property, it suffices to prove that the normal line n to
the ellipse at P bisects the angle ∠F1 PF2 in Figure 7.1.8(b)—this would make α1 = α2 ,
so that the indicated path from F1 to P to F2 satisfies Fermat’s Principle. First, let
2 y2
P = ( x0 , y0 ) be a point on the ellipse ax2 + b2 = 1, with a > b > 0. Assume that P is not a
vertex (i.e. y0 6= 0), otherwise the reflection property holds trivially. By Exercise 15 in
Section 3.4, the equation of the tangent line to the ellipse at P = ( x0 , y0 ) is
xx0 yy0
2
+ 2 = 1, (7.2)
a b
2 a2 y
so that its slope is − ab2 xy0 . Hence, the negative reciprocal b2 x0 is the slope of the normal
0 0
line n, whose equation—valid even when x0 = 0 (i.e. when y0 = ± b)— is then
b2 x0 ( y − y0 ) = a2 y0 ( x − x0 ) . (7.3)
Setting y = 0 and solving for x shows the x-intercept of n is at
(a2 − b2 ) x0 y0 c2
x = 2
= 2
x0 = e2 x0
a y0 a
Let N = ( e2 x0 , 0) be that x-intercept, as in Figure 7.1.8(b). The distance F1 N from the
focus F1 = (− c, 0) = (− ea, 0) to N is then
F1 N = e2 x0 − (− ea) = e (a + ex0 )
while the distance F2 N from the focus F2 = ( c, 0) = ( ea, 0) to N is
F2 N = ea − e2 x0 = e (a − ex0 ) .
Therefore,
F1 N e (a + ex0 ) a + ex0
= = .
F2 N e (a − ex0 ) a − ex0
By the distance formula, the distance F1 P from F1 = (− ea, 0) to P = ( x0 , y0 ) is given by
(F1 P )2 = ( x0 + ea)2 + y02

= x02 + 2 eax0 + e2 a2 + (1 − e2 ) (a2 − x02 ) (by formula (7.1))
(F1 P )2 = a2 + 2 eax0 + e2 x02 = (a + ex0 )2
F1 P = a + ex0 .
Similarly, the distance F2 P from F2 = ( ea, 0) to P = ( x0 , y0 ) is given by
(F2 P )2 = ( x0 − ea)2 + y02 = x02 − 2 eax0 + e2 a2 + (1 − e2 ) (a2 − x02 )

(F2 P )2 = a2 − 2 eax0 + e2 x02 = (a − ex0 )2
F2 P = a − ex0 .
Thus, P
F1 P a + ex0 F1 N
= = , α1 α2
F2 P a − ex0 F2 N
which means that α1 = α2 :3 by the Law of Sines, and with 180◦ − θ θ
θ = ∠F2 NP as in the figure on the right, F1 N F2
sin α2 sin θ sin (180◦ − θ ) sin (180◦ − θ ) F1 P sin α1 F1 N sin α1

= = = · = · =
F2 N F2 P F2 P F1 P F2 P F1 N F2 N F2 N
and thus sin α2 = sin α1 , so that α2 = α1 (since 0◦ < α1 , α2 < 90◦). X
An ellipse of the form y

x2 y2
2 2 a
b2
+ a2 = 1
x y
2
+ 2 = 1
b a
c
with a > b > 0 simply switches the roles of x and y in the
previous examples: the principalp axis is now the y-axis, the
foci are at (0, ± c), where c = a2 − b2 , and the vertexes are x
−b 0 b
at (0, ±a), as in Figure 7.1.9.
Thus, the largest denominator on the left side of an equa- −c
x2 y2
tion of the form + = 1 tells you which axis is the prin-
ä2 ä2
cipal axis. For example, the principal axis of the ellipse
−a
x2 y2
25
+ 16 = 1 is the x-axis (since 25 > 16), while the ellipse
Figure 7.1.9
x2 y2
4
+ 9
= 1 has the y-axis as its principal axis (since 9 > 4).
3 This follows directly from Proposition 3 in Book VI of Euclid’s Elements. See the purely geometric proof on
pp.125-126 in E UCLID, Elements, (Thomas L. Heath translation), Santa Fe, NM: Green Lion Press, 2002.
Exercises
A
1. Construct an ellipse using the procedure shown in Figure 7.1.2. Place the two pins 7in apart and
use a 10in piece of string.
For Exercises 2-6, sketch the graph of the given ellipse, indicate the major and minor axes and exact
locations of the foci and vertexes, and find the eccentricity e.
x 2 y2 x 2 y2 4 x 2 y2
2. + =1 3. + =1 4. + =1 5. x2 + 4 y2 = 1 6. 25 x2 + 9 y2 = 225
25 16 4 9 25 4
x2 y2
7. Show that for a > b > 0 the ellipse = 1 with eccentricity e can be written as y2 = (1− e2 ) (a2 − x2 ).
+
a2 b2
2 y2
p
8. Use Example 7.1 to show the area inside the ellipse ax 2 + b2 = 1 with eccentricity e is πa2 1 − e2 .
x2 y2 x2 y2
9. For all a > b > 0, find the points of intersection of the ellipses a2
+ b2 = 1 and b2
+ a2 = 1.
10. Show that the vertexes are the closest and farthest points on an ellipse to either focus.
B
2 y2
11. Show that any line of slope m that is tangent to the ellipse ax2 + b2 = 1 must be of the form
p
y = mx ± a2 m2 + b2 .
12. A 10ft ladder with a mark 3ft from the top rests against a wall. If the top of the ladder slides down
the wall, with the foot of the ladder sliding away from the wall on the ground, as in Figure 7.1.10,
then show the mark moves along part of an ellipse.
y
P T1
G
3
F
7
T2
x D
Figure 7.1.10 Exercise 12 Figure 7.1.11 Exercise 13 Figure 7.1.12 Exercise 14
13. Another definition of an ellipse is the set of points P for which the ratio of the distance from P to a
fixed point F (a focus) to the distance from P to a fixed line D (the directrix) is a constant e < 1 (the
PF
eccentricity): PG = e, as in Figure 7.1.11. Use this definition to show that the equation of an ellipse
x2 y2
with focus ( c, 0) can be written as a2
+ b2 = 1 for some a > b > 0. Find the equation of the directrix.
14. Show that the set of intersection points of all perpendicular tangent lines to an ellipse form a circle,
as in Figure 7.1.12 (showing two such tangent lines T1 ⊥ T2 ).
15. A chord of an ellipse that passes through a focus and is perpendicular to the major axis is a latus
x2 y2 2 b2
rectum. Show that for the ellipse a2
+ b2 = 1 with a > b > 0 the length of each latus rectum is a .
16. Suppose that the normal line at one end of a latus rectum of an ellipse passes through an end of
the minor axis. Show that the eccentricity e is a root of the equation e4 + e2 − 1 = 0, then find e.
17. Show that the set of all midpoints of a family of parallel chords in an ellipse lie on a diameter.
(Hint: Use symmetry with chords of slope m 6= 0.)
7.2 Parabolas
Like ellipses, you have seen parabolas (e.g. y = x2 ) and some of their applications (e.g.
projectile trajectories), but perhaps without knowing their purely geometric definition.
The alternative definition of an ellipse described in Exercise 13 in Section 7.1 is, in
fact, similar to the definition of the parabola:
A parabola is the set of all points in a plane that are equidistant from a fixed point
(the focus) and a fixed line (the directrix).
Figure 7.2.1 illustrates the above definition, with a

P
point P moving along the parabola so that the distance F
from P to the focus F equals the distance from P to
the directrix D . Note that the point halfway between D
the focus and directrix must be on the parabola—that G
point is the vertex, which is the point on the parabola Figure 7.2.1 Parabola: PF = PG
closest to the directrix. The axis of the parabola is the
line that passes through the focus and is perpendicular to the directrix. Notice that the
ratio PG
PF
equals 1, whereas that ratio for an ellipse—by the alternative definition—was
the eccentricity e < 1. The eccentricity of the parabola, therefore, is always 1.4
To construct a parabola from the definition, cut a piece of string to have the same
length AB as one side of a drafting triangle, as in Figure 7.2.2.
A
P
F
P
D
C B
Figure 7.2.2 Construction of a parabola
Fasten one end of the string to the vertex A of the triangle and the other end to a pin
somewhere between A and B—the pin will be the focus F of the parabola. Hold the
string taut against the edge AB of the triangle at a point P on either side of the pin,
then move the edge BC of the triangle along the directrix D . The drawn figure will
be a parabola, as the lengths PF and PB will be equal (since the length of the string
being AB = AP + PF means PF = PB).
4 The eccentricity e of the parabola being 1 means there is no second vertex, unlike the ellipse (where e < 1 forced
the existence of two vertexes in the alternative definition).

Parabolas • Section 7.2 211
To derive the equation of a parabola in the x y- y

plane, start with the simple case of the focus on the
y-axis at (0, p), with p > 0, and the line y = − p as d1 (0, p)
the directrix, as in the figure on the right. The ver- ( x, y)
tex is then at the origin (0, 0). Pick a point ( x, y) 0
x
whose distances d 1 and d 2 from the focus (0, p) and d2
y = −p
directrix y = − p, respectively, are equal. Then −p
( x, − p)
d 12 = d 22
( x − 0)2 + ( y − p)2 = ( x − x)2 + ( y + p)2
y✓2 − 2 p y + p2 = ✓
x2 + ✓ y✓2 + 2 p y + p2
x2 = 4 p y
1 2
In other words, y = 4p x , which is the more familiar form of a parabola. Thus, any
2
curve of the form y = ax , with a 6= 0, is a parabola whose ¡ afocus
¢ and directrix can be
a
found by dividing a by 4: p = 4 , so that the focus is at 0, 4 and the directrix is the
¡ ¢
line y = − a4 . For example, the parabola y = x2 has its focus at 0, 41 and its directrix is
the line y = − 41 .
When p > 0 the parabola 4 p y = x2 extends upward; for p < 0 it extends downward,
as in Figure 7.2.3(a) below:
y y
y
x = −p x = −p
−p y = −p
x x
−p 0 ( p, 0) ( p, 0) 0 −p
x
0
(0, p)
(a) 4 py = x2 : p < 0 (b) 4 px = y2 : p > 0 (c) 4 px = y2 : p < 0
Figure 7.2.3 Parabolas with vertex at the origin
Switching the roles of x and y yields the parabola 4 px = y2 , with focus at ( p, 0) and
directrix x = − p. For p > 0 this parabola extends rightward, while for p < 0 it extends
leftward. See Figure 7.2.3(b) and (c).
It is left as an exercise to show that in general a curve of the form y = ax2 + bx + c
is a parabola. Just like not every oval shape is an ellipse, not every “cupped” or “U”
shape is a parabola (e.g. y = x4 ).
The slope of the parabola 4 p y = x2 is dx = 42px = 2xp , so that the equation of the tangent
dy
line to the parabola at a point ( x0 , y0 ) is:

x0
y − y0 = ( x − x0 )
2p
2 p ( y − y0 ) = x0 x − x02
2 p y − 2 p y0 = x0 x − 4 p y0
2 p ( y + y0 ) = x0 x (7.4)
Likewise, switching the roles of x and y, the tangent line to the parabola 4 px = y2 at a
point ( x0 , y0 ) is:
2 p ( x + x0 ) = y0 y (7.5)
Formula (7.5) simplifies the proof of the reflection prop- y
erty for parabolas: light shone from the focus to any point
on the parabola will reflect in a path parallel to the axis ( x0 , y0 ) β
of the parabola. Figure 7.2.4 shows light emanating from P
the focus F = ( p, 0) and reflecting off a point P = ( x0 , y0 ) on Q β
x
the parabola 4 px = y2 . If that line of reflection is parallel (− x0 , 0) 0 F = ( p, 0)
to the x-axis—the axis of the parabola—then the tangent
line to the parabola at ( x0 , y0 ) should make the same an-
gle β with the line of reflection as it does with the x-axis.
Figure 7.2.4 4 px = y2
So extend the tangent line to intersect the x-axis and use
formula (7.5) to find the x-intercept:
2 p ( x + x0 ) = y0 y = y0 · 0 = 0 ⇒ x = − x0
Let Q = (− x0 , 0), so that the distance FQ equals p + x0 . The goal is to show that the
angle of incidence ∠FPQ equals the angle of reflection β. The focal radius FP has
length
q q q
FP = ( p − x0 )2 + (0 − y0 )2 = p2 − 2 px0 + x02 + 4 px0 = p2 + 2 px0 + x02 = p + x0 .
Thus, FQ = FP in the triangle △FPQ , so that ∠FPQ = ∠FQP = β, i.e. the light’s path
does indeed satisfy Fermat’s Principle for curved surfaces. X
The parabola’s reflection property shows up in some engineering applications, typi-
cally by revolving part of a parabola around its axis, producing a parabolic surface in
three dimensions called a paraboloid. For example, it used to be common for vehi-
cle headlights to use paraboloids for their inner reflective surface, with a bulb at the
focus, so that—by the reflection property—the light would shine straight ahead in a
solid beam. Many flashlights still operate on that principle. The reflection property
also works in the opposite direction, which is why satellite dishes and radio telescopes
are often wide paraboloids with a signal receiver at the focus, to maximize reception
of incoming reflected signals.
Example 7.2
Suppose that an object is launched from the ground with an initial velocity v0 and at varying angles
with the ground. Show that the family of all the possible trajectories—which are parabolic—form a
region whose boundary (called the envelope of the trajectories) is itself a parabola.
Solution: Recall from Example 4.3 in Section v02
4.1 that if the object is launched at an angle 2g
0 < θ < π2 with the ground, then the height y
attained by the object as a function of the hori-
zontal distance x that it travels is given by
gx2
y = − + x tan θ .
2v02 cos2 θ
The curve is a parabola, with the figure on the
right showing these parabolic trajectories for 500
values of the angle θ. Clearly each parabola in-
tersects at least one other.The maximum hori-
v2 0 v02
zontal distance g0 occurs only for θ = π4 , as was g
shown in Example 4.3. The maximum vertical
v2
height 2 0g is attained when the object is launched straight up (i.e. θ = π2 ), as was shown in Exercise 18
in Section 5.1. By symmetry only the angles 0 < θ ≤ π2 in the same vertical plane need be considered. So
in the above figure, imagine if the trajectories for all possible angles were included, filling up a region
that does appear to have a parabolic boundary. This will now be shown to be true.
v02
First, it turns out that all the parabolas for 0 < θ < π
2 have the same directrix y = 2 g . To see why,
v2 sin2 θ
recall from Exercise 11 in Section 4.1 that the maximum height reached by the object is 0 2 g , which
is thus the y-coordinate of the vertex of the parabola. That vertex is midway between the focus and
the directrix. The parabola is of the form 4 py = x2 + bx, where b is a constant that does not affect the
distance between the vertex and directrix5 , and 4 p is a constant with p < 0 such that the directrix is
− p units above the vertex (since p < 0), just as in the case 4 py = x2 . The equation of the parabola then
shows that
1 g v2 cos2 θ
= − 2 ⇒ p = − 0
4p 2v0 cos2 θ 2g
so that the directrix is at
y = y-coordinate of the vertex + (− p)
Ã 2 !
v02 sin2 θ v0 cos2 θ v2
= + − − = 0 (sin2 θ + cos2 θ)
2g 2g 2g
v02
y =
2g
v2
It is perhaps surprising that all the parabolic trajectories share the same directrix y = 2 0g , which is
µ 2 2 ¶ µ 2 ¶
v sin θ v
independent of the angle θ. Note that the heights of each vertex 0 2 g and focus 2 0g (sin2 θ − cos2 θ)
do depend on θ. The common directrix is the key to the remainder of the proof.
5 This will be discussed further in Section 7.4.
v02
Now let P be a point in the first quadrant of the x y-plane below the common directrix y = 2g , denoted
by D . Then P can be either inside, outside, or on the envelope, as in Figure 7.2.5:
y y y
v02 v02 v02
2g
D 2g
D 2g
D
P
P
P
x x x
O v02 / g O v02 / g O v02 / g
(a) Inside the envelope (b) Outside the envelope (c) On the envelope
Figure 7.2.5 Trajectory envelope and a point P
The origin O = (0, 0) is on each trajectory, so by definition of a parabola the foci for all the trajectories
v02
must be a distance 2g from O, i.e. the distance from O to D . That is, the foci of all the trajectories must
v2
lie on the circle C0 of radius 2 0g centered at O. If P is any other point inside the envelope, so that it lies
on at least one trajectory, then it must be a distance r > 0 below the line D . By definition of a parabola,
P must be the same distance from the foci of any trajectories it belongs to. That is, the foci must be on
a circle C of radius r centered at P and touching the directrix D , as in Figure 7.2.6:
y y y
v02 v02 v02
2g
D 2g
D 2g
D
C0 r C0 r C0 r
C C
F1 C P P
P F
x x x
O F2 v02 / g O v02 / g O v02 / g
(a) Two intersections F1 , F2 (b) No intersections (c) One intersection F
Figure 7.2.6 Circles C and C0 intersect at foci of trajectories
In Figure 7.2.6(a) C and C0 intersect at two points F1 and F2 , so P y

belongs to two trajectories; P must then be inside the envelope. In v02
L
Figure 7.2.6(b) C and C0 do not intersect, so P must be outside the g
envelope (since it is not on a parabola with a focus on C0 ). If C and v02
C0 intersect at only one point F , as in Figure 7.2.6(c), then P must 2g
v2 v02
be on the envelope. In that case, P is a distance r + 2 0g from O, which 2g
D
r
v02
is also the distance from P to the line y = (denoted by L). Thus, by
g P
definition of a parabola,
µ P
¶ is on a parabola with focus O and directrix
v02
L. The vertex is at 0, 2 g . Therefore, the envelope is a parabola: the x
O v02 / g
boundary of the shaded region in the figure on the right. X
In Example 7.2 all the trajectories were in the x y-plane only. Removing that restric-
tion, so that trajectories in all vertical planes through the y-axis are possible, would
result in a solid paraboloid consisting of all possible trajectories from the origin.
Parabolas also appear in suspension bridges: the suspension ca-
bles supporting a horizontal bridge (via vertical suspenders, as in
the figure on the right) have to be parabolas if the weight of the
bridge is uniformly distributed.6
Exercises
A
1. Construct a parabola using the procedure shown in Figure 7.2.2.
For Exercises 2-6, sketch the graph of the given parabola and indicate the exact locations of the focus,
vertex, and directrix.
2. 8 y = x2 3. y = 8 x2 4. x = y2 5. x = −3 y2 6. −1000 y = x2
7. Find the points of intersection of the parabolas 4 py = x2 and 4 px = y2 when p > 0. What is the
equation of the line through those points?
8. A vehicle headlight in the shape of a paraboloid is 3in deep and has an open edge with diameter 8in.
Where should the center of the bulb be placed in order to be at the focus, measured in inches relative
to the vertex?
9. The latus rectum of a parabola is the chord that passes through the focus and is parallel to the
directrix. Find the length of the latus rectum for the parabola 4 py = x2 .
10. Show that the circle whose diameter is the latus rectum of a parabola touches the parabola’s direc-
trix at one point.
11. Find the points on the parabola 4 px = y2 such that the focal radii to those points have the same
length as the latus rectum.
12. From each end of the latus rectum of a parabola draw a line to the point where the directrix and
axis intersect. Show that the two drawn lines are perpendicular.
B
13. Show that any point not on a parabola is on either zero or two tangent lines to the parabola.
14. Show that y = mx − 2mp − m3 p is the normal line of slope m to the parabola 4 px = y2 .
15. From a point P on a parabola with vertex V let PQ be the line segment perpendicular to the axis
at a point Q . Show that PQ 2 equals the product of QV and the length of the latus rectum.
16. Show that the curve y = ax2 + bx + c is a parabola for a 6= 0, using only the definition of a parabola.
Find the focus, vertex and directrix.
17. Show that the set of all midpoints of a family of parallel chords in a parabola lie on a line parallel
to the parabola’s axis.
6 See pp.159-161 in S MITH , C.E., Applied Mechanics: Statics, New York: John Wiley & Sons, Inc., 1976.
7.3 Hyperbolas
In the previous two sections you have seen curves with eccentricity e = 0 (circles),
0 < e < 1 (ellipses) and e = 1 (parabolas). The remaining case is e > 1: the hyperbola,
whose definition is similar to the second definition of the ellipse.
A hyperbola is the set of all points in a plane such that the ratio of the distance
from a fixed point (a focus) to the distance from a fixed line (a directrix) is a
constant e > 1, called the eccentricity of the hyperbola.
It will be shown in Section 7.4 that the curve y = 1x is a hy- y

perbola, which has two branches (see Figure 7.3.1). In general
1
a hyperbola resembles a “wider” or less “cupped” parabola, and y= x
it has two symmetric branches (and hence two foci and two di-
rectrices) as well as two asymptotes. x
0
The ratio of distances referred to in the definition of the hyper-
bola also appears in the second definition of the ellipse (where
the ratio is smaller than 1) and in the definition of the parabola
(where the ratio equals 1). In all three cases that ratio is the Figure 7.3.1
eccentricity. See Figure 7.3.2(a) for the comparisons.
e>1 v > vE
e=1 v = vE
D D
F e<1 Sun v < vE
(a) Eccentricity e (b) Orbit velocity v and escape velocity vE
Figure 7.3.2 Hyperbola, parabola and ellipse with focus F and directrix D
There is an analogue for Figure 7.3.2(a) in terms of orbits. For example, an object
approaching the Sun must meet or exceed the escape velocity to overcome the Sun’s
gravitational pull and avoid returning to orbit. Figure 7.3.2(b) shows the three possible
trajectories—hyperbola, parabola and ellipse—in terms of the object’s velocity v and
the escape velocity vE . Note the apparent correlation between the eccentricity of the
object’s path and its speed as a fraction of the escape velocity (i.e. vvE ).
Hyperbolas • Section 7.3 217
Figure 7.3.3 illustrates the definition of a hyper- F

P
bola, consisting of points P whose distance PF from
the focus F exceeds the distance PG to the directrix D
G
D in a way so that the ratio PG PF
is always the same
constant e > 1 (the eccentricity). Figure 7.3.3 Hyperbola: PF
PG = e>1
To derive the equation of a hyperbola with eccentricity y

e > 1, assume the focus is on the x-axis at ( ea, 0), with d2
( x, y)
a > 0, and the line x = ae is the directrix, as in Figure
7.3.4. Pick a point ( x, y) whose distances d 1 and d 2 from d1
the focus ( ea, 0) and directrix x = ae , respectively, satisfy
d1 x
the condition for a hyperbola: d2 = e > 1. Then 0 ( ea, 0)
a
x= e
d 12 = e2 d 22
µ³ ¶
2 2a ´2 2
( x − ea) + ( y − 0) = e x− + ( y − y)2
e Figure 7.3.4
2 ✘✘ 2 2 2
2 eax + e a + y = e x − ✘
x −✘ 2 2 ✘✘ + a2
2 eax
( e2 − 1) x2 − y2 = ( e2 − 1) a2
x2 y2
− = 1
a2 b2
p
where b2 = ( e2 − 1) a2 > 0. The hyperbola thus has two branches ( x = ± ab y2 + b2 ), as
in Figure 7.3.5. Let c = ea, so that c > a and b2 = c2 − a2 . By symmetry the hyperbola
2
has two foci (± c, 0) and two directrices x = ± ac , and the lines y = ± ab x are asymptotes.
The vertexes are the points on the hyperbola closest to the directrices.
y
directrix directrix
asymptote asymptote
y = − ab x y = ab x
transverse
axis vertex vertex
x
(− c, 0)
−a
− a2 0 a2 a
( c, 0)
c c
focus focus
conjugate
axis
x2 y2
Figure 7.3.5 Parts of the hyperbola a2
− b2 = 1, with b2 = c2 − a2
The center is the point midway between the foci. The transverse axis and conju-
gate axis are the perpendicular lines through the foci and center, respectively.
In Figure 7.3.5 the vertexes are (±a, 0), the x-axis is the transverse axis, the center
is the origin (0, 0), and the conjugate axis is the y-axis. Note that the existence of two
foci and directrices—when the definition of the hyperbola mentioned only a focus and
a directrix—is simply a consequence of the symmetry about both axes imposed by the
x2 y2
equation − b2 = 1. A parabola, in comparison, is symmetric about only one axis.
a2 p
To see why the lines y = ± ab x are asymptotes, consider the upper half y = ab x2 − a2
of both branches of the hyperbola. The difference between the line y = ab x and the
upper right branch approaches zero as x approaches infinity:
µ ¶ p
b bp 2 b ³ p ´ x + x2 − a2
lim x − x −a = 2 2
lim x − x − a · 2 p
x→∞ a a a x→∞ x + x2 − a2
b x2 − ( x2 − a2 )
= lim p
a x→∞ x + x2 − a2
ab
= lim p = 0
x→∞ x + x2 − a2
p
Thus the line y = ab x is an (oblique) asymptote for the upper half y = ab x2 − a2 of the
right branch in the first quadrant of the x y-plane. So by symmetry the lines y = ± ab x
are asymptotes for both branches, i.e. for the entire hyperbola.
x2 y2
Conversely, given a hyperbola in the form a2
− b2 = 1, let c2 = a2 + b2 to get the foci
2
(± c, 0) and the directrices x = ± ac , while the eccentricity is e = ac .
Example 7.3
For the hyperbola x2 − y2 = 1 find the vertexes, foci, directrices, asymptotes and eccentricity.
p
Solution: Here a = b = 1, so that c2p= a2 + b2 = 2, i.e. c = 2. The vertexes are thus (±1, 0),pthe
asymptotes are y = ± x, the foci are (± 2, 0), the directrices are x = ± p1 , and the eccentricity is e = 2.
2
Switching the roles of x and y yields the hyperbola y

y2 x2
a2
− = 1, shown in Figure 7.3.6. The transverse
b2
c y = ab x
a
axis is the y-axis, the conjugate axis is the x-axis,
a2
the vertexes are at (0, ±a), and the foci are at (0, ± c), y= c
2
where c2 = a2 + b2 . The directrices are y = ± ac , the x
asymptotes are y = ± ab x, and the eccentricity is e = ac . a2
y=−
2 2 c
For example,
p the hyperbola y − x = 1 has a = b = 1, −a
so that c = 2. The vertexes are p (0, ±1), the asymp- −c y = − ab x
totes are y = ± x, the foci are (0, ± 2), the
p directrices
1 y2 2
are y = ± , and the eccentricity is e = 2. The hy-
p Figure 7.3.6 a2
− bx 2 = 1
2
perbola y2 − x2 = 1 is just the hyperbola x2 − y2 = 1
rotated 90◦ .
There is another way to define a hyperbola, in terms of two foci:
A hyperbola is the set of all points in a plane such that the absolute value of the
difference of the distances from two fixed points (the foci) is a positive constant.
Figure 7.3.7 illustrates the above definition with foci F1 and F2 . The difference
d 1 − d 2 of the distances d 1 and d 2 can be positive or negative depending on which
branch the point ( x, y) is on, which is why the absolute value | d 1 − d 2 | is used.
y
d1 ( x, y)
d2 L
P A
P
x
F1 0 F2 F1 F2
Figure 7.3.7 Hyperbola: | d1 − d2 | = constant > 0 Figure 7.3.8 Construction of a hyperbola
It is left as an exercise to show that this second definition yields the same equation
of the hyperbola as from the first definition. The second definition is often used as
the primary definition in many textbooks, perhaps because it provides a simple way to
construct a hyperbola by hand. Figure 7.3.8 shows the procedure for foci F1 and F2 : at
the focus F1 fasten one end of a ruler of length L, and at the other end A of the ruler
fasten one end of a string of length L − d for some number 0 < d < F1 F2 . Fasten the
other end of the string to the focus F2 and hold the string taut with a pencil against
the ruler at a point P , while rotating the ruler about F1 . The drawn figure will be
one branch of a hyperbola, since the difference PF1 − PF2 will always be the positive
constant d :
PF1 − PF2 = (L − AP ) − PF2 = L − ( AP + PF2 ) = L − (L − d ) = d
Reverse the roles of F1 and F2 to draw the other branch of the hyperbola.
x2 y2
By Exercise 16 in Section 3.4, the tangent line to the hyperbola a2
− b2 = 1 at a point
( x0 , y0 ) is
xx0 yy0
2
− 2 =1 , (7.6)
a b
2
so that its slope is ab2 xy0 when y0 6= 0. Note that by the above equation, when y0 = 0 (so
0
that x0 = ±a) the two tangent lines are the vertical lines x = ±a.
That slope will be used in proving the reflection property for the hyperbola: Light
shone from one focus will reflect off the hyperbola in the opposite direction from the
other focus. Figure 7.3.9 shows the light’s path from focus F2 as it reflects at the point
P along the line through P and the other focus F1 .
y
L
P
( x0 , y0 ) θ2
P
θ2 θ1
F1 F2
α1 θ α2
x
F1 F2 (− c, 0) A ( c, 0)
x2 y2
Figure 7.3.9 Reflection property Figure 7.3.10 Hyperbola a2
− b2 = 1 with foci (± c, 0)
By Fermat’s Principle for curved surfaces, the reflection property is equivalent to

saying the tangent line L to the hyperbola at the point P = ( x0 , y0 ) bisects the angle
∠F1 PF2 , i.e. θ1 = θ2 as in Figure 7.3.10. The reflection property holds trivially when
y0 = 0 (the light reflects straight back along the x-axis), so to show that θ1 = θ2 assume
y0 6= 0. By symmetry only x0 > 0 need be considered. For the case x0 6= c, let A be the
x-intercept of the tangent line L. Then by Figure 7.3.10, since the sum of the angles
in the triangle △F1 P A equals 180◦,
α1 + θ2 + (180◦ − θ ) = 180◦ ⇒ θ2 = θ − α 1 ⇒ tan θ2 = tan(θ − α1 ) .

b 2 x0 y0
Thus, since tan θ is the slope of the tangent line L (i.e. a2 y0
), and tan α1 = x0 + c , then
by the subtraction formula for the tangent function
b x0 2 y b2 x02 − a2 y02 + b2 x0 c
− x +0 c ✭
✭
tan θ − tan α1 a2 y0 0 ✭a2 ✭ x0✭
y0 (✭ + c)
tan θ2 = = 2x
=
1 + tan θ tan α1 b y (a2 + b2 ) x0 y0 + a2 y0 c
1 + a2 y · x0 +0
0
x0✭
y0 (✭
a2 ✭
✭
✭
0 c ✭ + c)
a2 b2 + b2 x0 c x02 y2
= (since c2 = a2 + b2 , and a2
− b02 = 1 ⇒ b2 x02 − a2 y02 = a2 b2 )
c2 x0 y0 + a2 y0 c
2 ✘✘ ✘✘
b2 ✘
(a✘ + x0 c) b2
=
2 ✘✘✘
✘ =
(a✘
c y0 ✘ + x0 c) c y0
Similarly, since the sum of the angles in the triangle △F2 P A equals 180◦,
α2 + θ1 + θ = 180◦ ⇒ θ1 = 180◦ − (θ + α2 ) ⇒ tan θ1 = − tan(θ + α2 ) .

y
Thus, since tan α2 = − tan (180◦ − α2 ) = − x−0c ,
b 2 x0 − y0 b2 x02 − a2 y02 − b2 x0 c
✭
✭
x0✭
tan θ + tan α2 a2 y0
+ y0 (✭
a2 ✭
x0 − c ✭ − c)
tan θ1 = − = − = −
1 − tan θ tan α2 b 2 x0 − y0 (a2 + b2 ) x0 y0 − a2 y0 c
1 − a2 y0
·
x0✭
y0 (✭
a2 ✭
✭
✭
x0 − c ✭ − c)
2 ✘✘ ✘✘
a2 b2 − b2 x0 c b2 ✘(a✘ − x0 c) b2
= − 2 = ✘ = = tan θ2 ,
c x0 y0 − a2 y0 c 2 ✘✘✘
(a✘
c y0 ✘ − x0 c) c y0
i.e. θ1 = θ2 (since 0◦ < θ1 , θ2 < 90◦). X Note: The case x0 = c is left as an exercise.
Ellipses, parabolas and hyperbolas are sometimes called conic sections, due to
being formed by intersections of planes with a double circular cone of unlimited extent:
(a) Ellipse (b) Parabola (c) Hyperbola
Figure 7.3.11 Conic sections
Each double cone in Figure 7.3.11 has two nappes—a cone extending upward and
one extending downward. When a plane intersects only one nappe in a closed noncir-
cular curve, as in Figure 7.3.11(a), that curve is an ellipse. A plane that is parallel to
a line on one nappe, as in Figure 7.3.11(b), intersects only that nappe in a parabola.
The intersection of a plane with both nappes, as in Figure 7.3.11(c), is a hyperbola.
To prove that the ellipse, parabola and hyperbola really are P1
represented by the indicated conic sections, first a minor result is
needed from three-dimensional geometry: tangent line segments
Q O
to a sphere from the same point have equal lengths, as in Figure
7.3.12. Since the right triangles △QOP1 and △QOP2 share the
same hypotenuse QO and have legs OP1 and OP2 of equal length P2
(the radius of the sphere), the result QP1 = QP2 follows by the Figure 7.3.12
Pythagorean Theorem.
For the case of a right circular double

cone (i.e. the base of each nappe is a cir-
cle in a plane perpendicular to the axis C
of the cone7 ), let β be the complement Pc
P F
of the angle that the cone makes with
its axis, as in Figure 7.3.13. So β is the D
angle the cone makes with any circular Q β A α G
P0
base of the cone. This constant angle β,
with 0◦ < β < 90◦ , is an intrinsic property
of the cone. Let P c be a plane that in-
β
tersects the lower nappe of the cone in a
curve C , such that P c makes an angle α Figure 7.3.13
with any base circle of the cone. By sym-
metry, only the angles 0◦ < α ≤ 90◦ need be considered.8 Inscribe a sphere in the cone
so that it touches P c at a point F , as in Figure 7.3.13 (P c is the tangent plane to the
sphere at F ). Let P0 be the plane through the circle where the inscribed sphere touches
the cone, and let D be the line of intersection of the planes P c and P0 . It will be shown
that F and D are the focus and directrix, respectively, of the curve C .
Let P be any point on the curve C , then let Q be the point on the plane P0 that lies
on a line through P and the cone’s vertex. Drop a perpendicular line segment from P
to the point A in the plane P0 . From A draw a perpendicular line segment to the point
G on the line D . Then as Figure 7.3.13 shows, △Q AP and △P AG are right triangles,
with
PA PA
sin α = and sin β = .
PG PQ
However, since PF and PQ are both tangent line segments to the inscribed sphere
from the same point P , the result proved earlier shows that PQ = PF . Thus,
PF sin α
PG sin α = P A = PQ sin β = PF sin β ⇒ = .
PG sin β
Let e = PG
PF
. Then e is the same constant sin sin β for any point P on the curve C . Thus, by
α
definition, e is the eccentricity of the curve C with focus F and directrix D . If 0◦ < α < β
then 0◦ < sin α < sin β so that 0 < sin
sin β < 1, which means that 0 < e < 1, and hence C is
α
an ellipse (by the second definition of an ellipse). Likewise, if α = β then e = 1, so that

C is a parabola. Finally, if β < α ≤ 90◦ then e > 1, so that C is a hyperbola (and will
intersect both nappes of the cone). Thus, the ellipse, parabola and hyperbola truly are
conic sections. X
7 The proof can be extended to oblique double cones. See §364 in S ALMON, G.S., A Treatise on Conic Sections,
London: Longmans, Green and Co., 1929.

8 The case where α = 0◦ results in a circle, which is typically not considered a conic section.
Exercises
A
1. Construct a hyperbola using the procedure shown in Figure 7.3.8. Place the two focus pins 7in apart
and use a 9in piece of string attached to a 12in ruler.
For Exercises 2-6, sketch the graph of the given hyperbola, indicate the exact locations of the foci and
vertexes, indicate each directrix and asymptote, and find the eccentricity e.
x 2 y2 x 2 y2 4 x 2 y2
2. − =1 3. − =1 4. − =1 5. x2 − 4 y2 = 1 6. 25 y2 − 9 x2 = 225
16 9 8 15 25 4
7. Find the equation of the hyperbola with foci (±5, 0) and vertexes (±3, 0).
B
8. In the second definition of the hyperbola on p.219, let 2a be the constant absolute value of the
difference of distances from points on the hyperbola to the foci, for some a > 0. Let the foci be at
(± c, 0) for some c > 0. Show that this second definition yields a hyperbola having an equation of the
x2 y2
form a2
− b2
= 1, with b > 0.
9. Prove the reflection property for the hyperbola (see pp.220-221) when x0 = c for the point P = ( x0 , y0 )
on the hyperbola with foci at (± c, 0).
10. Show that a straight line parallel to an asymptote of a hyperbola intersects the hyperbola at exactly
one point.
11. A latus rectum (plural: latus recta) of a hyperbola is a chord through either focus perpendicular to
x2 y2 2 b2
the transverse axis. Show that the latus recta of the hyperbola a2
− b2 = 1 have length a .
12. Show that the segment of an asymptote of a hyperbola between the two directrices has the same
length as the line segment between the vertexes.
13. Show that the segment of a tangent line to a hyperbola between the hyperbola’s asymptotes has its
midpoint at the point of tangency.
14. The focal radii of a hyperbola are the line segments from the foci to points on the hyperbola. Show
x2 y2
that the lengths of the focal radii to points ( x, y) on the hyperbola a2
− b2
= 1 are a + ex and ex − a,
where e is the eccentricity.
15. Show that a tangent line to a hyperbola together with the hyperbola’s asymptotes bounds a triangle
of constant area (i.e. the area is independent of the point of tangency on the hyperbola).
x2 y2
16. Show that the product of the perpendicular distances from any point on the hyperbola a2
− b2
=1
to the asymptotes y = ± ab x is a constant.
17. A person at a point P = ( x, y) hears the crack of a rifle located at the point F1 = (−1000, 0) and the
sound of the fired bullet hitting its target located at the point F2 = (1000, 0) at the same time. The
bullet’s speed is 2000 ft/sec and the speed of sound is 1100 ft/sec. Find an equation relating x and y.
18. Show that the set of all midpoints of a family of parallel chords either in one branch or between the
two branches of a hyperbola lie on a line through the center of the hyperbola.
19. Prove a second reflection property of hyperbolas: a light shone between the two branches and
directed toward one focus will reflect toward the other focus.
7.4 Translations and Rotations

For convenience the ellipses, parabolas and hyperbolas in the previous sections were
centered at the origin and had their foci on one of the coordinate axes. In general
the center and foci of those curves can be moved anywhere by means of coordinate
transformations.
You have probably seen in earlier courses how the y′
graph of a function y = f ( x) can be shifted horizontally
by an amount h and vertically by an amount k by re- y
P = ( x, y)
placing x and y by x − h and y − k, respectively: P ′ = ( x ′ , y′ )
x−h
y − k = f ( x − h)
y−k
This coordinate transformation is called translation,
and can be applied to any curve in the x y-plane. The k x′
O′
origin O = (0, 0) is shifted to the point O′ = ( h, k), which
serves as the origin of the x′ y′ -plane,9 as in Figure O
x
h
7.4.1. Let P = ( x, y) be a point in the x y-plane. Con-
sider P as a point P ′ = ( x′ , y′ ) in the x′ y′ -plane, so that Figure 7.4.1 Translation
relative to the origin O , the x and y coordinates of P are:
′ ′ ′ ′
x′ = x − h
y′ = y − k
Using these translation equations (i.e. the substitutions x 7→ x − h and y 7→ y − k), the
graphs of the conic sections can be translated to any center ( h, k):
Ellipse: For a > b > 0, an equation of the form
( x − h )2 ( y − k )2
+ = 1
a2 b2
describes an ellipse with center ( h, k), vertexes ( h ± a, k), and foci ( h ± c, k), where
c2 = a2 − b2 . The eccentricity is e = ac , and the principal axis is the line y = k.
Likewise, an equation of the form
( y − k )2 ( x − h )2
+ = 1
a2 b2
describes an ellipse with center ( h, k), vertexes ( h, k ± a), and foci ( h, k ± c), where
c2 = a2 − b2 . The eccentricity is e = ac , and the principal axis is the line x = h.
9 The prime symbol (′ ) does not indicate differentiation—it acts merely to distinguish the new axes.
Translations and Rotations • Section 7.4 225
Parabola: For p 6= 0, an equation of the form
( x − h )2 = 4 p ( y − k )
describes a parabola with vertex ( h, k) and focus ( h, k + p). The directrix is the line
y = k − p and the axis is the line x = h.
( y − k )2 = 4 p ( x − h )
describes a parabola with vertex ( h, k) and focus ( h + p, k). The directrix is the line
x = h − p and the axis is the line y = k
Hyperbola: For a 6= 0 and b 6= 0, an equation of the form
( x − h )2 ( y − k )2
− = 1
a2 b2
describes a hyperbola with center ( h, k), vertexes ( h ± a, k), and foci ( h ± c, k), where
2
c2 = a2 + b2 . The eccentricity is e = ac , the directrices are the lines x = h ± ac , the
asymptotes are the lines y − k = ± ab ( x − h), and the transverse axis is the line y = k.
( y − k )2 ( x − h )2
− = 1
a2 b2
describes a hyperbola with center ( h, k), vertexes ( h, k ± a), and foci ( h, k ± c), where
2
c2 = a2 + b2 . The eccentricity is e = ac , the directrices are the lines y = k ± ac , the
asymptotes are the lines y − k = ± ab ( x − h), and the transverse axis is the line x = h.
Example 7.4
( x−2)2
Find the vertexes and foci of the ellipse 4 + ( y − 1)2 = 1. y x=2
Solution: The translation coordinates are (h, k) = (2, 1). Also, 2
a = 2 and b = 1. Thus, the vertexes are (h ± a, kp) = (2 ± 2, 1) = p p
(0, 1) and (4, 1). Since 2 2 2 (2 − 3, 1) (2 + 3, 1)
p c = a − b = 3 then c = 3, so the foci y=1
are (h ± c, k) = (2 ± 3, 1), as shown in Figure 7.4.2.
2
Note that the ellipse ( x−42) + ( y − 1)2 = 1 in the x y-plane is
′2 x
the ellipse x4 + y′2 = 1 in the x′ y′ -plane, for the translation 0 4
equations x′ = x − 2 and y′ = y − 1 (i.e. the x′ -axis is the line
( x−2)2
y = 1 and the y′ -axis is the line x = 2). Figure 7.4.2 Ellipse 4 + ( y − 1)2 = 1
Example 7.5
For a 6= 0 and constants b and c, find the vertex, focus and directrix of the parabola y = ax2 + bx + c.
Solution: The idea here is to write y = ax2 + bx + c in the form ( x − h)2 = 4 p( y − k) for some h, k, and p,
by completing the square:
ax2 + bx + c = y
µ ¶
b
a x2 + x = y − c
a
µ ¶
b b2 b2
a x2 + x + = y − c + a
a 4a2 4a2
µ ¶ µ ¶
b 2 1 b2 − 4ac
x + = y +
2a a 4a
³ ´
b2 b 4ac− b2
So h = − 2ba , k = 4ac4− 1 1
a , and 4 p = a means p = 4a . Thus, the vertex is ( h, k) = − 2a , 4a , the focus is
³ ´
b 4ac− b2 +1 4ac− b2 −1
(h, k + p) = − 2a , 4a , and the directrix is the line y = k − p = 4a .
Rotation is another common coordinate transforma- y

′
y
tion. Consider the case of rotating the x y-plane about
P = ( x, y)
the origin by an angle θ , as in Figure 7.4.3. The origin
P ′ = ( x ′ , y′ )
of the resulting x′ y′ -plane is unchanged from the origin
O = (0, 0) of the x y-plane. To find the rotation equations α ′
x
for x′ and y′ , let P = ( x, y) be a point in the x y-plane
r
θ
away from the p origin. From trigonometry you know α−
θ x
that for r = x2 + y2 6= 0 and the angle 0◦ ≤ α < 360◦ O
measured from the positive x-axis to OP ,
Figure 7.4.3 Translation
x = r cos α and y = r sin α .
Considering P as a point P ′ = ( x′ , y′ ) in the x′ y′ -plane, OP ′ makes an angle α − θ with
the positive x′ -axis, so that by the sine and cosine subtraction identities,
x′ = r cos (α − θ ) = r cos α cos θ + r sin α sin θ = x cos θ + y sin θ (7.7)
and
y′ = r sin (α − θ ) = r sin α cos θ − r cos α sin θ = y cos θ − x sin θ . (7.8)
Similar to the translation substitutions, the above rotation equations allow any curve
in the x y-plane to be rotated:
To rotate a curve in the x y-plane about the origin by an angle θ , make the following
substitutions:
x 7→ x cos θ + y sin θ and y 7→ − x sin θ + y cos θ (7.9)
Example 7.6
x2
Find the equation of the ellipse 4 + y2 = 1 when rotated 45◦ counterclockwise about the origin. Simplify
the equation.
Solution: For θ = 45◦ the substitutions are: y

x+ y
x 7→ x cos θ + y sin θ = x cos 45◦ + y sin 45◦ = p 2
x′
2
y′
◦ ◦ −x + y
y 7→ − x sin θ + y cos θ = − x sin 45 + y cos 45 = p
2
The equation of the rotated ellipse (shown in Figure 7.4.4) is then: 45◦
x
µ ¶2 µ ¶2 −2 0 2
1 x+ y −x + y
p + p = 1
4 2 2
x2 + 2 x y + y2 + 4( x2 − 2 x y + y2 ) = 8
−2
5 x 2 − 6 x y + 5 y2 − 8 = 0
Figure 7.4.4
In the above example note the presence of the 6 x y term in the equation of the rotated
ellipse. In general if a conic section has a second-degree equation of the form
Ax2 + Bx y + C y2 + Dx + E y + F = 0 (7.10)
then B 6= 0 indicates rotation, and either D 6= 0 or E 6= 0 indicates translation.

Example 7.7
x2 y2
Find the value of a such that rotating the hyperbola a2
− a2 = 1 by 45◦ counterclockwise about the origin
results in the curve x y = 1.
Solution: Since θ = 45◦ then as in Example 7.6 the substitutions are again:
x+ y −x + y
x 7→ p and y 7→ p
2 2
The equation of the rotated hyperbola is then:
µ ¶2 µ ¶
1 x+ y 1 −x + y 2
p − p = 1
a2 2 a2 2
x 2 + 2 x y + y2 − ( x 2 − 2 x y + y2 )
= 1
2a2
2x y
= 1
a2
p 1
Thus, when a = 2 the rotated hyperbola has the equation x y = 1, which shows that the curve y = x is
a hyperbola. In general any curve of the form Bx y = 1 is a hyperbola for B 6= 0.
The following result can be used for determining the type of conic section described
by a second-degree equation:10
The graph of Ax2 + Bx y + C y2 + Dx + E y + F = 0 (with A , B, C not all zero) describes

a curve whose type is based on the sign of B2 − 4 AC :
(a) B2 − 4 AC < 0: an ellipse (or a circle, point, or no curve)
(b) B2 − 4 AC = 0: a parabola (or a line, two parallel lines, or no curve)
(c) B2 − 4 AC > 0: a hyperbola (or two intersecting lines)
If B 6= 0 and the curve is a conic section, then the rotation angle θ is given by:
(a) θ = 45◦ if A = C
B
(b) tan 2θ = A −C if A 6= C , with 0◦ < θ < 90◦
Recall that the rotation equations (7.7) and (7.8) for x′ and y′ in terms of x and y
provided substitutions that allowed the second-degree equation of the rotated conic
section to be found. Conversely, to transform a second-degree equation in x and y into
a “standard” conic section equation in terms of x′ and y′ (to simplify sketching the
graph), the “reverse” rotation equations for x and y in terms of x′ and y′ are needed:
x = x′ cos θ − y sin θ (7.11)

′ ′
y = x sin θ + y cos θ (7.12)
Example 7.8
Determine the type of curve whose equation is 5 x2 + 4 x y + 8 y2 − 36 = 0, and sketch its graph.
Solution: Since A = 5, B = 4, and C = 8, then B2 − 4 AC = −144 < 0, so the curve is an ellipse if it is a
4
conic section. Since B 6= 0 and A 6= C , the rotation angle θ would be given by tan 2θ = AB
−C = −3 , with
4 −3
0 < θ < 90 . Then 2θ is in the second quadrant, so that sin 2θ = 5 and cos 2θ = 5 . Use a half-angle
◦ ◦
identity to find θ:
4
sin 2θ 5
tan θ = = = 2
1 + cos 2θ 1 + −3 5
Hence, sin θ = p2 and cos θ = p1 . Now use equations (7.11) and (7.12) to get expressions for x and y in
5 5
terms of x′ and y : ′
x ′ − 2 y′
x = x′ cos θ − y sin θ = p
5
2 x ′ + y′
y = x′ sin θ + y′ cos θ = p
5
10 For a proof see Section 6.8 in P ROTTER , M.H. AND C.B. M ORREY , Analytic Geometry, 2nd ed., Reading, MA:
Addison-Wesley Publishing Company, Inc., 1975.

Substitute those expressions into 5 x2 + 4 x y + 8 y2 − 36 = 0: y

µ ¶2 µ ¶µ ¶ µ ′ ¶2
x ′ − 2 y′ x ′ − 2 y′ 2 x ′ + y′ 2 x + y′ 3
x′
5 p +4 p p +8 p − 36 = 0
5 5 5 5
45 x′2 + 20 y′2 = 180
y′
x′2 y′2
+ = 1 64.3◦
4 9 x
′2
−3 0 3
x ′2 y
So the curve’s equation in the x′ y′ -plane is 4 + 9 = 1. In
2
x2 y
other words, the curve is just the ellipse + 4 9 = 1 rotated by
θ = tan−1 2 ≈ 63.4◦ , as shown in the figure on the right.
−3
Exercises
A
For Exercises 1-3, sketch the graph of the given ellipse with the exact locations of the foci and vertexes.
( x − 3)2 ( y − 2)2 ( x + 1)2 ( y + 3)2
1. + =1 2. + =1 3. 4 x2 + y2 + 24 x − 2 y + 21 = 0
25 16 9 4
For Exercises 4-7, sketch the graph of the given parabola with the exact locations of the focus, vertex
and directrix.
4. 4( x − 1)2 = 3( y + 2) 5. y = 4 x2 + 24 x + 21 6. 3( y + 2)2 = 5( x − 2) 7. 4 x2 − 4 x + 4 y − 5 = 0
For Exercises 8-10, sketch the graph of the given hyperbola with the exact locations of the foci, vertexes,
directrices, and asymptotes.
( x − 3)2 ( y − 2)2 ( x + 1)2 ( y + 3)2
8. − =1 9. − =1 10. 4 x2 − y2 + 24 x − 2 y + 21 = 0
25 16 9 4
1
11. Find the foci, vertexes, directrices and asymptotes for the hyperbola y = x in Example 7.7.
12. Find the equation of the parabola 4 y = x2 when rotated 60◦ counterclockwise about the origin.
13. Prove equations (7.11) and (7.12). (Hint: Use equations (7.7) and (7.8).)
B
14. Determine the type of curve whose equation is 8 x2 − 12 x y + 17 y2 − 20 = 0, and sketch its graph.
15. Determine the type of curve whose equation is 7 x2 + 6 x y − y2 − 32 = 0, and sketch its graph.
16. Let A ′ x′2 + B′ x′ y′ + C ′ y′2 + D ′ x′ + E ′ y′ + F ′ = 0 be the equation obtained by substituting the “reverse”
translation equations x = x′ + h and y = y′ + k into equation (7.10). Show that A = A ′ , B = B′ and
C = C ′ (i.e. A , B and C are invariant under translation).
17. Similar to Exercise 16, substitute the “reverse” rotation equations (7.11) and (7.12) into equation
(7.10) to show that A + C and B2 − 4 AC are invariant under rotation.
18. Sketch the graph of the conic section whose equation is x2 − 2 x y + y2 + 4 x − 4 y + 10 = 0 = 0. (Hint:
Handle the rotation first (as in Example 7.8), then the translation (by completing the square).)
7.5 Hyperbolic Functions

In some textbooks you might see the sine and cosine functions called circular functions,
since any point on the unit circle x2 + y2 = 1 can be defined in terms of those functions
(see Figure 7.5.1). Those definitions motivate a similar idea for the unit hyperbola
x2 − y2 = 1, whose points can be defined in terms of hyperbolic functions.
y y
a
2 = sector area = 21 r2 θ = θ
2
P = ( x, y)
P = ( x, y)
x 2 + y2 = 1 x = cos θ = cos a a
= shaded area
2
θ y = sin θ = sin a
x x
O 1 O 1 x
x 2 − y2 = 1
Figure 7.5.1 Circular Figure 7.5.2 Hyperbolic
For a point P = ( x, y) on the unit hyperbola x2 − y2 = 1, the hyperbolic angle a is

twice the area of the shaded hyperbolic sector in Figure 7.5.2).11 The area a2 thus
equals the area of the right triangle with hypotenuse OP and legs of length x and y (so
that the triangle’s area is 21 x y) minus the area under the hyperbola over the interval
p
[1, x]. So since the upper half of the hyperbola x2 − y2 = 1 is the function y = x2 − 1,
a
= (area of triangle) − (area under the hyperbola from 1 to x)
2 Zx p
1
= xy − u2 − 1 du
2 1
µ p ´¶
a 1 p 2 1 2
1 ³ p
2
= x x −1 − x x − 1 − ln x + x − 1 (by formula (6.9))
2 2 2 2
³ p ´
a = ln x + x2 − 1
p
e a = x + x2 − 1
³p ´2
( e a − x )2 = x2 − 1
e2a − 2 xe a + x2 = x2 − 1
e 2a + 1 e a + e −a
x = = = cosh a
2 ea 2
where cosh a is the hyperbolic cosine of a.
11 The reason for using twice the area is merely to obtain a “cleaner” final result involving a instead of 2a.
Hyperbolic Functions • Section 7.5 231
The y-coordinate of P can then be found:

s s
p µ a ¶2
e +e − a e2a + 2 + e−2a 4
y = x2 − 1 = −1 = −
2 4 4
s s
µ a ¶
e2a − 2 + e−2a e − e −a 2
= = , so since a ≥ 0
4 2
e a − e −a
y = = sinh a
2
where sinh a is the hyperbolic sine of a. All six hyperbolic functions can now be
defined in general, analogous to the trigonometric (circular) functions:
The hyperbolic sine, hyperbolic cosine, hyperbolic tangent, hyperbolic

cotangent, hyperbolic secant and hyperbolic cosecant, denoted by sinh, cosh,
tanh, coth, sech and csch, respectively, are:
e x − e− x e x + e− x
sinh x = for all x cosh x = for all x
2 2
sinh x 1
tanh x = for all x coth x = for all x 6= 0
cosh x tanh x
1 1
sech x = for all x csch x = for all x 6= 0
cosh x sinh x
The graphs of the hyperbolic functions are shown below:

y y y
y = cosh x y = csch x
y = coth x
1 1 y = sech x 1
x x x
0 y = tanh x 0 0
−1
y = sinh x
(a) sinh x and cosh x (b) tanh x and coth x (c) sech x and csch x
Figure 7.5.3 Graphs of the six hyperbolic functions
The graph of y = cosh x in Figure 7.5.3(a) might look familiar: a catenary—a uniform
cable hanging from two fixed points—has the shape of a hyperbolic cosine function.
The hyperbolic functions satisfy the following identities:
cosh2 x − sinh2 x = 1 tanh2 x + sech2 x = 1 coth2 x − csch2 x = 1
sinh (− x) = − sinh x cosh (− x) = cosh x tanh (− x) = − tanh x

sech (− x) = sech x csch (− x) = − csch x coth (− x) = − coth x
sinh ( u ± v) = sinh u cosh v ± cosh u sinh v sinh 2 x = 2 sinh x cosh x

cosh ( u ± v) = cosh u cosh v ± sinh u sinh v cosh 2 x = cosh2 x + sinh2 x
tanh u ± tanh v 2 tanh x
tanh ( u ± v) = tanh 2 x =
1 ± tanh u tanh v 1 + tanh2 x
The identity cosh2 x − sinh2 x = 1 was proved when deriving the coordinates of points
on the unit hyperbola x2 − y2 = 1 in terms of the hyperbolic angle (since such a point
( x, y) = (cosh a, sinh a) must satisfy x2 − y2 = 1). The addition identities can be proved
similarly using hyperbolic angles (i.e. areas).12 However, it is simpler to use the
definitions of sinh and cosh in terms of exponential functions. For example:
e u − e − u e v + e −v e u + e − u e v − e −v
sinh u cosh v + cosh u sinh v = · + ·
2 2 2 2
e u+v + e u−v − e − u+v − e − u−v + e u+v − e u−v + e − u+v − e − u−v
=
4
2e u+v
− 2e − u−v
=
4
e u+v − e−(u+v)
=
2
= sinh ( u + v) X
The identity for sinh 2 x is then easy to prove by letting u = v = x in the above identity:
sinh 2 x = sinh ( x + x) = sinh x cosh x + cosh x sinh x = 2 sinh x cosh x X
Note that the identities cosh (− x) = cosh x and sinh (− x) = − sinh x mean that cosh is
an even function and sinh is an odd function. Those two functions thus (sort of) serve
as even and odd versions of the exponential function (which is neither even nor odd).
Both cosh x and sinh x grow exponentially (the e−x term for both functions becomes
negligible as x → ∞), while sinh x decreases exponentially to −∞ as x → −∞.
12 See pp.25-29 in S HERVATOV, V.G., Hyperbolic Functions, Boston: D.C. Heath and Company, 1963.
The derivatives of the hyperbolic functions and their integral equivalents are:
d d
(sinh x) = cosh x (csch x) = − csch x coth x
dx dx
d d
(cosh x) = sinh x (sech x) = − sech x tanh x
dx dx
d d
(tanh x) = sech2 x (coth x) = − csch2 x
dx dx
Z Z
cosh x dx = sinh x + C csch x coth x dx = − csch x + C
Z Z
sinh x dx = cosh x + C sech x tanh x dx = − sech x + C
Z Z
2
sech x dx = tanh x + C csch2 x dx = − coth x + C
For example, by definition of cosh x:

µ ¶
d d e x + e− x e x − e− x
(cosh x) = = = sinh x X
dx dx 2 2
Example 7.9
Find the derivative of y = sinh x3 .
dy
Solution: By the Chain Rule, = 3 x2 cosh x3 .
dx
Example 7.10
Z
Evaluate tanh x dx.
Solution: Use the definition of tanh x and the substitutions u = cosh x, du = sinh x dx:
Z Z Z
sinh x du
tanh x dx = dx = = ln | u | + C = ln (cosh x) + C
cosh x u
For any constant a > 0, both y = cosh at and y = sinh at satisfy the differential equation
y′′ ( t) = a2 y( t) ,
which models rectilinear motion of a particle under a repulsive force proportional to

the displacement. This is just one of the many reasons why hyperbolic functions ap-
pear in so many physical applications.
Example 7.11
In the classical theory of paramagnetism, the total number n of molecules in a gas subject to a magnetic
field of strength H is Z π µH
cos θ
n = 2π N e kT sin θ dθ ,
0
where N is the number of molecules per unit solid angle having zero potential energy, µ is the magnetic
moment, k is Boltzmann’s constant, and T is the temperature of the gas. Show that
µ ¶
4π NkT µH
n = sinh .
µH kT
µH
Solution: Let a = kT , and let u = cos θ so that du = − sin θ dθ:
Z−1 Z1 ¯1 µ ¶
2π N au ¯¯ 2π N e a − e −a
n = −2 π N e au du = 2π N e au du = e ¯ = 2·
1 −1 a −1 a 2
µ ¶
4π NkT µH
= sinh
µH kT
x −x
d
Since dx (sinh x) = cosh x = e +2e > 0 for all x, then y = sinh x is an increasing func-
tion and thus its inverse function x = sinh−1 y is defined. The remaining inverse hy-
perbolic functions can be defined similarly, with the following domains and ranges
(switching the roles of x and y, as usual) plus their graphs:
function sinh−1 x cosh−1 x tanh−1 x csch−1 x sech−1 x coth−1 x

domain all x x≥1 | x| < 1 all x 6= 0 0<x≤1 | x| > 1
range all y y≥0 all y all y 6= 0 y≥0 all y 6= 0
y y = sinh−1 x y y
y = tanh−1 x
y = csch−1 x
y = coth−1 x
y = cosh−1 x y = sech−1 x
x x x
1 −1 1 1
(a) sinh−1 x and cosh−1 x (b) tanh−1 x and coth−1 (c) sech−1 x and csch−1 x
Figure 7.5.4 Graphs of the six inverse hyperbolic functions

The inverse hyperbolic functions can be expressed in terms of the natural logarithm:
p p
sinh−1 x = ln ( x + x2 + 1 ) cosh−1 x = ln ( x + x2 − 1 ) for x ≥ 1
1 1+x 1 x+1
tanh−1 x = ln for | x | < 1 coth−1 x = ln for | x | > 1
2 1−x 2 x−1
p  s 
−1 1 + 1 − x2 1 1
sech x = ln for 0 < x ≤ 1 csch−1 x = ln  + + 1  for x 6= 0
x x x2
Notice that the above formula for cosh−1 x was actually proved at the beginning of this
section. The formula for sinh−1 x follows from the definition of an inverse function:
e y − e− y
y = sinh−1 x ⇒ x = sinh y =
2
⇒ e y − 2 x − e− y = 0
⇒ e2 y − 2 xe y − 1 = 0
⇒ u2 − 2 xu − 1 = 0 for u = e y
p
2 x ± 4 x2 − 4(1)(−1) p
⇒ u = = x ± x2 + 1
2
p
⇒ e y = u = x + x2 + 1 since e y > 0
p
⇒ y = ln ( x + x2 + 1 ) X
The remaining formulas can be proved similarly.
Example 7.12
Show that for the hyperbolic angle a of a point P = ( x, y) on the y x 2 − y2 = 1
unit hyperbola x2 − y2 = 1, the area a2 of the hyperbolic sector
O AP (the shaded region in the figure on the right) is P
( x, y) = (cosh a, sinh a)
a 1
= cosh−1 x .
2 2
Solution: It was shown earlier that the area of O AP is x
a 1 p O A = (1, 0)
= ln ( x + x2 − 1 )
2 2 a
p 2 = shaded area of O AP
−1 2
so by the formula cosh x = ln ( x + x − 1 ) the result follows.
This makes sense, since x = cosh a and so cosh−1 x = cosh−1 (cosh a) = a, by definition of an inverse.
You can use either the general formula for the derivative of an inverse function or the
above formulas to find the derivatives of the inverse hyperbolic functions:
d 1 d 1
(sinh−1 x) = p (cosh−1 x) = p for x ≥ 1
dx x2 + 1 dx x2 − 1
d 1 d 1
(tanh−1 x) = for | x | < 1 (coth−1 x) = for | x | > 1
dx 1 − x2 dx 1 − x2
d −1 d −1
(sech−1 x) = p for 0 < x ≤ 1 (csch−1 x) = p for x 6= 0
dx x 1 − x2 dx | x | 1 + x2
For example, here is one way to find the derivative of tanh−1 x:

µ ¶
d −1 d 1 1+x 1 d
(tanh x) = ln = (ln (1 + x) − ln (1 − x))
dx dx 2 1−x 2 dx
µ ¶
1 1 −1 1 − x + (1 + x) 1
= − = = = X
2 1+x 1−x 2(1 + x)(1 − x) 1 − x2
Exercises
A
1. Prove the identities for sinh (− x), cosh (− x), tanh (− x), coth (− x), sech (− x), and csch (− x) on p.232.
2. Prove the identities for cosh (u ± v), tanh (u ± v), and tanh 2 x on p.232.
For Exercises 3-15 prove the given identity. Your proofs can use other identities.
3. (cosh x + sinh x)r = cosh rx + sinh rx for all r 4. sinh 3 x = 3 sinh x + 4 sinh3 x
1
5. sinh A cosh B = 2 (sinh ( A + B) + sinh ( A − B)) 6. tanh2 x + sech2 x = 1
1
7. sinh A sinh B = 2 (cosh ( A + B) − cosh ( A − B)) 8. coth2 x − csch2 x = 1
1
¡ ¢
9. cosh A cosh B = 2 (cosh ( A + B ) + cosh ( A − B)) 10. coth−1 1x = tanh−1 x
cosh x − 1
11. cosh 2 x = cosh2 x + sinh2 x = 2 cosh2 x − 1 = 1 + 2 sinh2 x 12. sinh2 2x =
2
cosh x + 1 cosh x − 1 cosh x − 1 sinh x
13. cosh2 2x = 14. tanh2 2x = 15. tanh x
2 = =
2 cosh x + 1 sinh x cosh x + 1
16. Prove the identities for tanh−1 x, coth−1 x, sech−1 x, and csch−1 x on p.235.
17. Prove the derivative formulas for sinh x, tanh x, coth x, sech x, and csch x on p.233.
18. Prove the derivative formulas for sinh−1 x, cosh−1 x, coth−1 x, sech−1 x, and csch−1 x on p.236.
d
19. Show that cosh x = O( e x ) and sinh x = O( e x ). 20. Show that (tan−1 (sinh x)) = sech x .
dx
21. Verify that the curve y = tanh x has asymptotes y = ±1.
1
22. Show that sinh dx = dx and cosh dx = 1 for any infinitesimal dx. (Hint: Use 1+ dx = 1 − dx from
Exercise 3 in Section 1.3 along with e dx = 1 + dx from Exercise 29 in Section 2.3.)
d
23. Use Exercise 22 and the addition formula for sinh x to show that dx (sinh x) = cosh x.
−2 x
24. Sketch the graph of f ( x) = e sinh x. Find all local maxima and minima, inflection points, and
vertical or horizontal asymptotes.
25. Denoting the speed of light by c, the Lorentz transformations for two inertial frames S and S ′ are
x − vt t − vx/ c2
x′ = p and t′ = p
1 − ( v / c )2 1 − ( v / c )2
where S ′ moves with speed v > 0 parallel to the x-axis of S . Let v/ c = tanh ξ. Show that
x′ = − ct sinh ξ + x cosh ξ and t′ = t cosh ξ − ( x/ c) sinh ξ .

R3
26. Evaluate the integral I = dx
2 1− x 2 in two different ways:
−1 1
(a) Use d
dx (coth x) = 1− x 2
for | x | > 1 to show that I = coth−1 3 − coth−1 2.
1
(b) Use the substitution u = x in the integral, then use d
dx (tanh−1 x) = 1
1− x 2
for | x | < 1 to show that
I = tanh−1 13 − tanh−1 12 . Is this answer equivalent to the answer from part (a)? Explain.
B
27. The general solution of the differential equation y′′ = a2 y is y( t) = y1 ( t) = c1 e at + c2 e−at , where a is
a positive constant, and c1 and c2 are arbitrary constants.
(a) Verify that y( t) = y2 ( t) = k1 cosh at + k2 sinh at is also a solution of y′′ = a2 y.
(b) Show that for any c1 and c2 , y1 ( t) = c1 e at + c2 e−at can be written as y1 ( t) = k1 cosh at + k2 sinh at
for some constants k1 and k2 in terms of c1 and c2 .
28. Verify that for positive constants β, p, l and c the function
βx β c sinh ( px/ c)
η( x) = −
p p2 cosh ( pl / c)
is a solution of the differential equation (related to water displacement in a canal of length 2l )
d2 η p2 βx p
2
− 2
η = − 2 .
dx c c
29. For any constant a and for s > | a | the Laplace transform L (s) of the function f ( t) = sinh at is
Z∞
L (s) = e−st sinh at dt .
0
a
Show that L (s) = s2 −a 2
.
C
30. In quantum mechanics the scaling factor eπ/s0 of the three-particle Efimov trimer is the solution
s = s0 > 0 of the equation ¡ ¢
¡ πs ¢ 8 sinh π6s
s cosh 2 = p ,
3
Use a numerical method to approximate s0 , then calculate eπ/s0 .
31. Continuing Example 7.11, the total magnetic moment M is defined as

Zπ
µH
M = 2π N µ e kT cos θ sin θ cos θ dθ .
0
1
(a) Use the value of n from Example 7.11 to show that M
nµ = L(a), where L(a) = coth a − a is the
µH
Langevin function and a = kT .
µH µ H
(b) Show that for a = kT , µmax > 0, and x = max
kT ,
Zµmax µ ¶
1 sinh x
L(a) dµ = ln .
0 x x
32. The age t0 of the universe (in years) is given by

Z1
dx
t0 = τ0 q ,
2 q0
0 x + 1 − 2 q0
where τ0 = 2 × 1010 years is the Hubble time and q0 ≥ 0 is the deceleration parameter. For the
2q
cosmological model with 0 < q0 < 21 , use the substitution x = 1−20q sinh θ to show that
0
Ã Ã !!
1 2 q0 −1 1
t0 = τ0 − cosh p .
1 − 2 q0 (1 − 2 q0 )3/2 2 q0
What fraction of τ0 is t0 (i.e. what is t0

τ0
) when q0 = 14 ? (Hint: Use Exercise 11 or 12.)
33. Circular rotations preserve the area of circular sectors (see Figure 7.5.5(a)). For any constant c > 0
the hyperbolic rotation φ : ( x, y) 7→ ( cx, y/ c) moves points along the hyperbola x y = k (for k > 0), as
shown in Figure 7.5.5(b) for c > 1. This hyperbolic rotation preserves the area of hyperbolic sectors.
y y y
rotate OBQ by α P′
B′ to OB′ Q ′
Q′ ( x, y)
Q P = A′
α ( cx, y/ c) (cosh a, sinh a)
θ B
θ x
x x
−1 O 1 0 O A = (1, 0)
(a) Unit circle x2 + y2 = 1 (b) Hyperbola x y = k (c) Unit hyperbola x2 − y2 = 1
Figure 7.5.5 Circular and hyperbolic rotations
As an example of why this is true, let a > 0 and consider the unit hyperbola in Figure 7.5.5(c). Then:
(a) Let c > 0. When c 6= 1 the mapping φ : ( x, y) 7→ ( cx, y/ c) does not move points along the unit
hyperbola. Find the formula for φ on the unit hyperbola as follows: use the rotation equations
from Section 7.4 to rotate the unit hyperbola 45◦ to a hyperbola of the form x y = k, apply φ to a
generic point on that hyperbola, then rotate that hyperbola by −45◦ back to the unit hyperbola.
(b) Use part (a) to find the value of c such that φ maps A = (1, 0) to P = (cosh a, sinh a).
(c) Let P ′ be the point that φ maps P to, and let A ′ = P . Use part (b) to find the coordinates of P ′ ,
then show that the hyperbolic sectors O AP and O A ′ P ′ have the same area a/2.
Parametric Equations • Section 7.6 239
7.6 Parametric Equations

Recall that Section 6.5 presented two different ways to “identify” or represent points
on the unit circle—by angle and by slope, as in Figure 7.6.1:
y y
slope = t
( x, y) = (cos θ, sin θ)
( x, y)
1
θ θ
x x
0 1 −1 0 1
(a) Identify points by angle θ (b) Identify points by slope t
Figure 7.6.1 Points on the unit circle x2 + y2 = 1
When identifying by the angle θ , all points ( x, y) on the unit circle can be written as
x = cos θ and y = sin θ
for any angle θ . When identifying by the slope t of lines through the point (−1, 0), recall
2t t2
from the derivation of the half-angle substitution that sin θ = 1+ t2
and cos θ = 11−
+ t2
. So
all points ( x, y) on the unit circle except (−1, 0) can be written as
1 − t2 2t
x =and y =
1 + t2 1 + t2
for any slope t. These two distinct “identifications” are called parametrizations of
the unit circle, with parameter θ in the first case and parameter t in the second.
In general, one way to describe a plane curve C (i.e. a curve in the x y-plane) is to
write its x and y-coordinates as functions of a variable t:
x = x( t) and y = y( t)
These are parametric equations of C , which consists of all points ( x, y) such that
x = x( t) and y = y( t) for the parameter t in some interval I . The shorthand for this is:
C : x = x( t), y = y( t), t in I
Notice the flexibility that parametric equations provide, since plane curves can take
any shape, not limited to the graph of a single function y = f ( x). In fact, a curve
y = f ( x) is the special case where the parametric equations are x = t and y = f ( t). In
physical settings the parameter t often denotes time, but it can represent anything
and any symbol can be used in its place. A curve can have many parametrizations.
Example 7.13
Show that for any constants ω 6= 0 and r > 0, and for t measured in radians,
x = h + r cos ω t and y = k + r sin ω t for −∞ < t < ∞
is a parametrization of the circle ( x − h)2 + ( y − k)2 = r2 with center (h, k) and radius r.
Solution: Since ω t is similar to the angle θ in Figure 7.6.1(a), it suffices to show that ( x − h)2 +( y− k)2 = r2 :
( x − h)2 + ( y − k)2 = r2 cos2 ω t + r2 sin2 ω t = r2 X
The constant ω determines how fast and in which direction the circle is traced as the parameter t varies.
For example, for ω = 2 the circle C is traced counterclockwise at twice the speed of the parametrization
C : x = h + r cos t, y = k + r sin t. In all cases the circle is re-traced every 2π/ω radians. For that reason
the interval for t is often restricted to the interval [0, 2π/ω], so that the circle is traced only once.
Example 7.14
Show that for a > 0 and b > 0 the parametric equations y
x = a cos t and y = b sin t b ( x, y) = (a cos t, b sin t)
for 0 ≤ t ≤ 2π describe an ellipse. x

−a a
Solution: Since
−b
x2 y2 a2 cos2 t b2 sin2 t
2
+ 2 = 2
+ = cos2 t + sin2 t = 1
a b a b2
2 y2
for all t, then the points ( x, y) = (a cos t, b sin t) lie on the ellipse ax2 + b2 = 1. It should be obvious that
the entire ellipse is traced, but it is left as an exercise to show what the parameter t represents.
Example 7.15
y
Show that the parametric equations
( x, y) = (cosh t, sinh t)
x = cosh t and y = sinh t
t
for −∞ < t < ∞ describe one branch of a hyperbola. 2 = shaded area
Solution: Since x
O 1
x2 − y2 = cosh2 t − sinh2 t = 1
x 2 − y2 = 1
for all t, then the points ( x, y) = (cosh t, sinh t) lie on the unit
hyperbola x2 − y2 = 1. This was in fact shown in Section 7.5, where t is half the area of the shaded
region in the above figure for t > 0. For t < 0 the shaded region is reflected below the x-axis. Since
cosh t ≥ 1 and sinh t can take any value, then the entire right branch of the hyperbola is traced as t
varies. Likewise the left branch has parametric equations x = − cosh t and y = sinh t. The hyperbola
is thus parametrized by area (or negative area for t < 0). In general, for a > 0 and b > 0 the hyperbola
x2 y2
a2
− b2 = 1 has parametric equations x = ±a cosh t and y = b sinh t for all t.
Example 7.16
Bézier curves13 are used in Computer Aided Design (CAD) to join the ends of an open polygonal path
of noncollinear control points with a smooth curve that models the “shape” of the path. The curve is
created via repeated linear interpolation, illustrated in Figure 7.6.2 and described below for n = 3 points:
B1 B1
40%
P A1
A0 B
40%
40%
B0 B2 B0 B2
(a) t = 0.4 = 40% (b) Many values of t in [0, 1]
Figure 7.6.2 Bézier curve B with 3 control points B0 , B1 , B2
For three control points B0 , B1 , B2 , pick 0 ≤ t ≤ 1. Say t = 0.4, which you can think of as a percentage:
0.4 = 40%. Let A 0 and A 1 be the points 40% of the way from B0 to B1 and from B1 to B2 , respectively,
as in Figure 7.6.2(a). Then the point P that is 40% of the way from A 0 to A 1 is on the Bézier curve B
joining B0 and B2 . Do this for every t in [0, 1] to fill out the curve B, as in Figure 7.6.2(b).
It can be shown via de Casteljau’s algorithm that the Bézier curve B for any three control points
B0 = ( x0 , y0 ), B1 = ( x1 , y1 ) and B2 = ( x2 , y2 ) in the x y-plane has parametric equations
x = (1 − t)2 x0 + 2 t (1 − t) x1 + t2 x2 and y = (1 − t)2 y0 + 2 t (1 − t) y1 + t2 y2 (7.13)
for 0 ≤ t ≤ 1. Write out and simplify the explicit parametric equations for the Bézier curve B with control
points B0 = (1, 2), B1 = (2, 4) and B2 = (4, 1).
Solution: The parametric equations for B0 = ( x0 , y0 ) = (1, 2), B1 = ( x1 , y1 ) = (2, 4), B2 = ( x2 , y2 ) = (4, 1) are:
x = (1 − t)2 (1) + 2 t (1 − t) (2) + t2 (4) = 1 − 2 t + t2 + 4 t − 4 t2 + 4 t2 = t2 + 2 t + 1

y = (1 − t)2 (2) + 2 t (1 − t) (4) + t2 (1) = 2 − 4 t + 2 t2 + 8 t − 8 t2 + t2 = −5 t2 + 4 t + 2
The Bézier curve B : x = t2 + 2 t + 1, y = −5 t2 + 4 t + 2, 0 ≤ t ≤ 1 for y

B0 , B1 , B2 is shown in the figure on the right. It is left as an exercise B1
to show that this curve is part of a parabola. In general Bézier curves
3
can be created for n ≥ 3 control points in the plane, with the parametric
equations being polynomials of degree n − 1 in the parameter t. In the B
2 B0
exercises you will be guided in how to derive the parametric equations
in the cases n = 3 and n = 4. Bézier curves can also be constructed 1 B2
for control points in three-dimensional space. A similar construct—a
Bézier surface—is used in three dimensions to model the boundary of x
0 1 2 3 4
a polyhedron (i.e. a solid whose faces are polygons).
13 Developed and popularized in the 1960s by two engineers, Pierre Bézier and Paul de Casteljau, for vehicle body
modeling at the French automotive manufacturers Renault and Citroën, respectively.

A curve with parametric equations x = x( t) and y = y( t) might not be the graph

dy
of a single function y = f ( x), but the derivative dx can still be found by using the
differentials of x and y as functions of t: dy = y ( t) dt and dx = x′ ( t) dt, so that
′
dy
dy y′ ( t) dt y′ ( t) dt
= ′ = ′ = (7.14)
dx x ( t) dt x ( t) dx
dt
d2 y
when dx
dt 6= 0. The second derivative can then be found via the Chain Rule:
dx2
µ ¶
d dy
µ ¶ µ µ ¶¶
d dy d dy dx d 2 y dx d2 y dt dx
= · = 2
· ⇒ 2
= (7.15)
dt dx dx dx dt dx dt dx dx
dt
R x2
For t in [a, b] with x1 = x(a) and x2 = x( b), the integral x1 y dx is given by:
Z x2 Zb
y dx = y( t) x′ ( t) dt (7.16)
x1 a
Example 7.17
A cycloid is the path of a point P on a circle rolling along a straight line. Figure 7.6.3 shows the cycloid
C traced by a circle of radius a rolling along the x-axis so that P touches the origin during the roll:
y
C
P
y
t
a
θ
x
0 x aθ 2π a
Figure 7.6.3 Cycloid C for a circle of radius a
dy
Find parametric equations for C and find dx .
Solution: For the angles t and θ—measured in radians—shown in Figure 7.6.3, t + θ + π/2 = 2π, so
t = 3π/2 − θ. The point P touches the origin as the circle rolls, so the horizontal distance from the
circle’s center to the y-axis is the length of the circular arc with central angle θ, namely aθ. So by the
parametrization of the circle as in Example 7.13, but with center (h, k) = (aθ, a), radius r = a, and ω = 1,
¡ ¡ ¢¢
x = a θ + a cos t = a θ + cos 32π − θ = a (θ − sin θ)
¡ ¡ ¢¢
y = a + a sin t = a 1 + sin 32π − θ = a (1 − cos θ)
Thus, C : x = a (θ − sin θ), y = a (1 − cos θ), −∞ < θ < ∞ is a parametrization of the cycloid C .
dy
As in formula (7.14) the derivative dx is given by:
dy y ′ (θ ) a sin θ sin θ
= ′ = = = cot 21 θ
dx x (θ ) a (1 − cos θ) 1 − cos θ
dy
Thus, dx is undefined when cos θ = 1, namely, when θ = 2πk for all integers k, i.e. when x = a (θ − sin θ) =
a (2πk − sin 2πk) = 2πka. Notice from Figure 7.6.3 that the cycloid has cusps at those values of x.
A cycloid appears in the solution of the famous brachistochrone problem:14 A
find the plane curve joining two points A and B—where B is at a lower height
than A but not directly under it—along which an object slides frictionless
under the force of gravity alone from A to B in the shortest time. It turns out
that the optimal path is not a straight line, but part of an inverted (upside- B
down) cycloid with a cusp at A , as in the figure on the right.15
Exercises
A
2 y2
1. For a > b > 0, in the ellipse ax2 + b2 = 1 inscribe a circle of radius b centered at the origin, called the
minor auxiliary circle. This circle is parametrized by x = b cos t and y = b sin t for 0 ≤ t ≤ 2π, with the
angle t shown in Figure 7.6.4. From each point on that circle draw a horizontal line segment to the
point P on the ellipse in the same quadrant, as shown. Show that P = (a cos t, b sin t).
Note: The angle t is called the eccentric angle of the ellipse, and it is the parameter for the parametriza-
tion of the ellipse in Example 7.14.
y
y (a cos t, a sin t)
b (b cos t, b sin t) b
P = ( x, y)
P = ( x, y)
t x t x
−a a −a a
−b −b
Figure 7.6.4 Exercise 1 Figure 7.6.5 Exercise 2
2. Similar to Exercise 1, circumscribe a circle of radius a (called the major auxiliary circle) around the
2 y2
ellipse ax2 + b2 = 1. This circle is parametrized by x = a cos t and y = a sin t for 0 ≤ t ≤ 2π, with the
eccentric angle t shown in Figure 7.6.5. From each point on that circle draw a vertical line segment
to the point P on the ellipse in the same quadrant, as shown. Show again that P = (a cos t, b sin t).
3. Show that the cycloid in Example 7.17 has global maxima at x = (2k + 1)πa for all integers k, and
that the cycloid is always concave down.
4. Show that the area under the cycloid in Example 7.17 over the interval [0, 2πa] is 3πa2 .
14 First solved in 1696 by the Swiss physicist and mathematician Johann Bernoulli (1667-1748).
15 See pp.60-62 in C LEGG, J.C., Calculus of Variations, Edinburgh: Oliver & Boyd, Ltd., 1968. For Bernoulli’s proof
see pp.644-655 in S MITH , D.E., A Source Book in Mathematics, New York: Dover Publications, Inc., 1959.
2
5. The parametrization C : x = 11+−t
t2
2t
, y = 1+ t2
, −∞ < t < ∞ of the unit circle C shown earlier makes the
unit circle a rational curve, since x and y are rational functions of the parameter t. Is the ellipse
x2 y2
a2
+ b2 = 1 a rational curve? Justify your answer.
6. Let P0 = ( x0 , y0 ) and P1 = ( x1 , y1 ) be distinct points in the x y-plane. For t in [0, 1] define
x( t) = (1 − t) x0 + t x1 and y( t) = (1 − t) y0 + t y1
and let P ( t) = ( x( t), y( t)).
(a) Show that C : x = x( t), y = y( t), 0 ≤ t ≤ 1 is a parametrization of the line segment from P0 to P1 .
P0 P ( t )
(b) Show that the parameter t is the proportion of the length P0 P ( t) to the length P0 P1 : P0 P1 = t.
7. In an ellipsograph a rod of length a has a peg at one end and another peg a distance b from the other
end, so that the rod can slide along two thin perpendicular rails, with the pegs in the rails as in
Figure 7.6.6. Show that a marker at the endpoint P traces an ellipse as the rod moves. (Hint: Treat
the rails as the x and y-axes. Find parametric equations, with a different angle than in Exercise 1.)
y
C
B
P
a
b
P = ( x, y)
θ x
O A
b
a−
B
8. The end P of a thread is held taut as it is unwound from a circle of radius a starting from a point A ,
as in Figure 7.6.7. Show that the path C that the end traces—called the involute of the circle—has
parametric equations x = a(cos θ + θ sin θ), y = a(sin θ − θ cos θ).
d2 y
9. Recall that a parabola of the form y = ax2 + bx + c has a constant second derivative dx2
= 2a. Consider
the Bézier curve B in Example 7.16.
d2 y
(a) Find dx2
for B. Is it a constant?
(b) Show that B is part of a parabola. Does this contradict part (a)? Explain. (Hint: Show the curve
has a second-degree equation of the form (7.10).)
(c) Find the point where the curve B has a global maximum.
10. Use Exercise 6 to derive the formulas (7.13) for the parametric equations of the general Bézier
curve for n = 3 control points.
11. To form the Bézier curve for n = 4 control points B0 = ( x0 , y0 ), B1 = ( x1 , y1 ), B2 = ( x2 , y2 ), B3 = ( x3 , y3 ),
for 0 ≤ t ≤ 1 use the three points that are 100 t% of the way along the line segments B0 B1 , B1 B2 , B2 B3
as the control points in the n = 3 case. Show that the resulting parametrization is:
x = (1 − t)3 x0 + 3 t(1 − t)2 x1 + 3 t2 (1 − t) x2 + t3 x3 and y = (1 − t)3 y0 + 3 t(1 − t)2 y1 + 3 t2 (1 − t) y2 + t3 y3
12. Each point on a plane curve lies on some line through the origin. Use that fact to show that
the equation y2 = x2 + x3 defines a rational curve (see Exercise 5). (Hint: For all real t, find the
intersections of the lines y = tx with the curve. Consider also the special case of the line x = 0.)
13. Sketch the graph of the curve C : x = 2 t − 4 t3 , y = t3 − 3 t4 , −∞ < t < ∞.
Polar Coordinates • Section 7.7 245
7.7 Polar Coordinates

Suppose that you wanted to write the equation of a spiral, like the one in Figure 7.7.1.
The curve is clearly not the graph of a function y = f ( x) in Cartesian coordinates, as
it violates the vertical line test. However, this spiral is simple to express using polar
coordinates.16 Recall that any point P distinct from the origin (denoted by O) in the
−−→
x y-plane is a distance r > 0 from the origin, and the ray OP makes an angle θ with the
−−→
positive x-axis, as in Figure 7.7.2. Imagine OP swings around the “pole” at the origin.
y
y
y
P = (r, θ)
r
x θ
0 1 2 3 θ x
x O
O −r
P = (−r, θ)
Figure 7.7.1 Spiral Figure 7.7.2 Polar coordinates (r, θ) Figure 7.7.3 Negative r: (−r, θ)
The pair ( r, θ ) contains the polar coordinates of P , and the positive x-axis is called
the polar axis of this coordinate system. For the angle θ measured in radians, ( r, θ ) =
( r, θ + 2πk) for all integers k. Thus, the polar coordinates of a point are not unique. By
−−→
convention r can be negative, by defining (− r, θ ) = ( r, θ + π) for any angle θ : the ray OP
is drawn in the opposite direction from the angle θ , as in Figure 7.7.3. When r = 0, the
point ( r, θ ) = (0, θ ) is the origin O, regardless of the value of θ .
Example 7.18
Express the spiral in Figure 7.7.1 in polar coordinates, such that the distance between any two points
separated by 2π radians is always 1.
Solution: The goal is to find some equation involving r and θ (measured in radians) that describes the
spiral. The distance between any two points separated by 2π radians is always 1, for example:
θ = 0 ⇒ r = 1
θ = 2π ⇒ r = 2
θ = 4π ⇒ r = 3
...
θ = 2π k ⇒ r = 1+k
for all integers k ≥ 0. In fact r = 1 + k when θ = 2πk for all real k ≥ 0, by the assumption about the
distance. So solving for k in terms of θ yields the polar equation r = 1 + 2θπ for all θ ≥ 0.
16 Created by the Flemish mathematician Grégoire de Saint-Vincent (1584-1667) and Italian mathematician
Bonaventura Cavalieri (1598-1647) in the 17th century, later used by Newton in his Method of Fluxions (1671).
You might be familiar with graphing paper, for plotting points or functions given in
Cartesian coordinates. Such paper consists of a rectangular grid, where the horizontal
and vertical lines represent where x and y, respectively, are constants, at regular
intervals. Similar graphing paper exists for polar coordinates, as in Figure 7.7.4.
90◦
105◦ 75◦
120◦ 60◦
135◦ 45◦
150◦ 30◦
165◦ 15◦
180◦ O 0◦
195◦ 345◦
210◦ 330◦
225◦ 315◦
240◦ 300◦
255◦ 285◦
270◦
Figure 7.7.4 Polar coordinate graphing paper
This polar grid is radial, not rectangular. The concentric circles around the origin O
are where r is constant (e.g. r = 1, r = 2), while the lines through the origin are where θ
is constant, at regular intervals for each. The angle θ shown here is in degrees, though
radians are often preferred for their “unitless” nature.
In general, polar coordinates are useful in describing plane curves that exhibit sym-
metry about the origin (though there are other situations), which arise in many phys-
ical applications.
Figure 7.7.5 shows how to convert between polar coordinates y
( x, y) (r, θ)
and Cartesian coordinates. Namely, for a point with polar coordi-
nates ( r, θ ) and Cartesian coordinates ( x, y): r
y
Polar to Cartesian:
θ
x
x = r cos θ y = r sin θ (7.17) O
x
Cartesian to Polar: Figure 7.7.5

q
y
r = ± x2 + y2 tan θ = if x 6= 0 (7.18)
x
In formula (7.18), if x = 0 then θ = π/2 or θ = 3π/2. If x 6= 0 and y 6= 0 then the

y
two possible solutions for θ in the equation tan θ = x are in opposite quadrants (for
p
same quadrant as the point ( x, y), then r = x2 + y2
0 ≤ θ < 2π). If the angle θ is in thep
(i.e. r is positive); otherwise r = − x2 + y2 (i.e. r is negative).
Example 7.19
Write the equation of the unit circle x2 + y2 = 1 in polar coordinates.
Solution: By formula (7.18), r2 = x2 + y2 = 1, so in polar coordinates the equation is simply r = 1. In
general the circle x2 + y2 = a2 of radius a > 0 has the simpler expression r = a in polar coordinates.
Example 7.20
Write the equation x2 + ( y − 4)2 = 16 in polar coordinates.
Solution: This is the equation of a circle of radius 4 centered at the point (0, 4). Expand the equation:
x2 + ( y − 4)2 = 16
x2 + y2 − 8 y + 16 = 16
x 2 + y2 = 8 y
r2 = 8 r sin θ (by formulas (7.17) and (7.18))
r = 8 sin θ
Is it valid to cancel r from both sides in the last step? Yes. The point (0, 0) is on the circle, so canceling
r does not eliminate r = 0 as a potential solution of the equation (e.g. θ = 0 would make r = 8 sin θ = 0).
Thus, the polar equation is r = 8 sin θ.
Notice that this polar equation is actually less intuitive than its Cartesian equivalent. From the equa-
tion x2 + ( y − 4)2 = 16 it is easy to identify the curve as a circle and read off its radius and center; these
properties are not so obvious from the polar equation. Since the circle’s center is not the origin, there is
no symmetry about the origin, which is when polar coordinates are often better suited.
Derivatives in Polar Coordinates
Suppose that the polar coordinates ( r, θ ) for a plane curve are related by a function:
r = r (θ ). Then by formula (7.17), x = r (θ ) cos θ and y = r (θ ) sin θ are now parametric
equations for the curve in the parameter θ . Thus, by the Product Rule and formulas
(7.14) and (7.15) from Section 7.6, with dx = x′ (θ ) dθ and dy = y′ (θ ) dθ :
For a plane curve with polar equation r = r (θ ),

µ ¶
d dy
dy r ′ (θ ) sin θ + r (θ ) cos θ d2 y dθ dx
= ′ and 2
= ′ . (7.19)
dx r (θ ) cos θ − r (θ ) sin θ dx r (θ ) cos θ − r (θ ) sin θ
Example 7.21
Sketch the graph of r = 1 + cos θ.
Solution: First sketch the graph treating (r, θ) as Cartesian coordinates, for 0 ≤ θ ≤ 2π as in Figure
7.7.6(a). Then use that graph to trace out a rough graph in polar coordinates, as in Figure 7.7.6(b).17
y ¡3 ¢
2, 3
π
r
1
¡1 2π
¢
2 2, 3
x
¡1 4π
¢ 0 2
2, 3
−1
¡3 5π
¢
2, 3
θ
0 π 2π
(a) θ r-plane (b) Polar coordinates (r, θ)
Figure 7.7.6 Graph of r = 1 + cos θ

dy d2 y
To find the maxima, minima, and inflection points it is still necessary to find dx and dx2
. It is left as an
exercise to use formula (7.19) and double-angle identities to show that
dy cos θ + cos 2θ d2 y 3 (1 + cos θ)
= − and 2
= −
dx sin θ + sin 2θ dx (sin θ + sin 2θ)3
and that for θ in [0, 2π], dx = 0 only when θ = π3 and 53π , while dx is undefined for θ = 0, 23π , π, 43π , and
dy dy
¯ ¯
d2 y ¯ d2 y ¯
2π. Since dx2 ¯ π < 0 and dx2 ¯ 5π > 0 then the curve has a local maximum when θ = π3 and a local
θ= 3 θ= 3
5π d2 y
minimum when θ = It can also be shown that dx2 changes sign around θ = 0, 23π , π, 43π , so that the
3 .
inflection points occur at those values of θ, as shown (with the correct concavity) in Figure 7.7.6(b).
17 There are far too many interesting plane curves to cover here. For an extensive collection, see L AWRENCE , J.D.,
A Catalog of Special Plane Curves, New York: Dover Publications, Inc., 1972. See also S EGGERN, D.H. VON , CRC
Handbook of Mathematical Curves and Surfaces, Boca Raton, FL: CRC Press, Inc., 1990.
Integration in Polar Coordinates
In some cases polar coordinates can simplify evaluation of a definite integral or find-
ing an area. To determine the polar form of a definite integral, suppose that r is a
function of θ : r = f (θ ). A polar region swept out by r = f (θ ) between θ = α and θ = β
would look like the shaded region in Figure 7.7.7 with area A :
y
θ=β y
(r + dr, θ + dθ) B
r = f (θ )
dr
ds
c a
r+
θ=α
(r, θ)
x dθ r
O x A b C
Figure 7.7.7 Area A Figure 7.7.8 Area dA Figure 7.7.9 Area = 12 bc sin A
A typical infinitesimal wedge of that region—shown in Figure 7.7.8—is produced by

an infinitesimal change dθ in the angle θ , which results in an infinitesimal change
dr in r . By the Microstraightness Property, the curve r = f (θ ) is a straight line with
infinitesimal length ds over that infinitesimal angle dθ . The area d A of the wedge thus
equals the area of the triangle with sides of lengths r and r + dr with included angle
dθ and a side of length ds. Since sin dθ = dθ and dr = f ′ (θ ) dθ , then by the formula for
the area of a triangle (as shown in Figure 7.7.9),18
1
dA = 2
( r ) ( r + dr ) sin dθ
1 2 1
= 2
r dθ + 2
r ( dr ) dθ
1 2 1
= 2
r dθ + 2
r ( f ′(θ ) dθ ) dθ
1 2 1 2
= 2 r dθ + 0 = 2 r dθ
since ( dθ )2 = 0. The area A of the region is the sum of these infinitesimal areas dA :
For a plane curve with polar equation r = f (θ ), the area A of the region swept out
between θ = α and θ = β is:
Zθ=β Zβ Zβ
1 2
A = dA = 2
r dθ = 1
2
( f (θ ))2 dθ (7.20)
θ=α α α
If f is periodic then choose the angle interval [α, β] so that the area is swept out
only once.
18 The formula 1 bc sin A for the area of a triangle △ ABC is derived in most trigonometry texts. For example, see
2
p.54 in C ORRAL , M., Trigonometry, http://mecmath.net/trig/, 2009.
Example 7.22
Use polar coordinates to show that the area of a circle of radius R is πR 2 .
Solution: Let the origin be the center of the circle. Then r = R is the polar equation of the circle, with
0 ≤ θ ≤ 2π sweeping out exactly one full circle. The area A inside the circle is then
Z 2π Z2π ¯2π
1 2
¯
A = 2 r dθ = 1
2 R 2 dθ = 1
2 R 2 θ ¯¯ = 1
2 R 2 (2π − 0) = πR 2 . X
0 0 0
Note the simplicity of this integral compared to the trigonometric substitution required when using
Cartesian coordinates, as in Section 6.3. Notice also that using a larger interval—say [0, 4π]—for θ
would result in an incorrect area (4πR 2 ) even though the region inside the curve is the same.
Example 7.23
Find the area inside the curve r = 1 + cos θ.
Solution: Choose 0 ≤ θ ≤ 2π as in Example 7.21, so that the area A is:
Z 2π Z 2π Z2π µ ¶
1 2 1 2 1 cos2 θ
A = 2 r dθ = 2 (1 + cos θ) dθ = 2
+ cos θ +
2
dθ
0 0 0
Z 2π µ ¶ Z2π µ ¶
1 1 + cos 2θ 3 cos 2θ
= + cos θ + dθ = + cos θ + dθ
0 2 4 0 4 4
¯2π
3 1 ¯ 3π
= θ + sin θ + sin 2θ ¯¯ =
4 8 0 2
Exercises
A
For Exercises 1-8 write the given equation in polar coordinates.
1. ( x − 3)2 + y2 = 9 2. y = x 3. x2 − y2 = 1 4. 3 x2 + 4 y2 − 6 x = 9
5. y = − x 6. y = x + 1 7. y = x2 8. y = x3
9. Write the polar equation r2 = 4 cos 2θ in Cartesian coordinates.

¡ ¢ ¡ ¢
10. Find the tangent line to r = cos 2θ at 21 , π6 . 11. Find the tangent line to r = 8 sin2 θ at 2, 56π .
12. Recall the curve r = 1 + cos θ from Example 7.21.

(a) Verify that for this curve,
dy cos θ + cos 2θ d2 y 3 (1 + cos θ)

= − and = − .
dx sin θ + sin 2θ dx2 (sin θ + sin 2θ)3
dy 5π 2π 4π
(b) Verify that for θ in [0, 2π], dx = 0 only when θ = π
3 and 3 , and is undefined for θ = 0, 3 , π, 3 ,
and 2π.
¯ ¯
d 2 y ¯¯ d 2 y ¯¯
(c) Verify that dx2 ¯ < 0 and dx2 ¯ > 0.
π 5π
θ= 3 θ= 3
d2 y 2π 4π
(d) Verify that dx2
changes sign around θ = 0, 3 , π, and 3 .
For Exercises 13-15, sketch the graph of the given curve and indicate all local maxima and minima.
13. r = 1 + sin θ 14. r = 1 − cos θ 15. r = sin 2θ
16. Find the area inside r = 1 + sin θ. 17. Find the area inside r = sin 2θ.
B
18. Sketch a rough graph of the meridian voltage component E θ for a half-wavelength linear antenna:
cos ( π2 cos θ)
E θ = r (θ ) =
sin θ
19. Show that the distance d between two points (r 1 , θ1 ) and (r 2 , θ2 ) in polar coordinates is
q
d = r21 + r22 − 2r 1 r 2 cos (θ1 − θ2 ) .
20. For a point P = (r, θ) on the curve r = r(θ), let α be the angle that the tangent line through P makes
with the positive x-axis. Let ψ = α − θ be the angle between the tangent line and the line through P
and the origin. Show that tan ψ = rr′((θθ)) . (Hint: Use formulas (7.19) and (3.2).)
21. For the parabola with focus at (0, 0), vertex at (0, − p), and directrix y = −2 p (with p > 0), show that
the polar equation is
2p
r = .
1 − sin θ
c
22. For the ellipse with foci (0, 0) and (2 c, 0), vertexes ( c ± a, 0), and eccentricity e = a (with 0 < c < a),
show that the polar equation is
a (1 − e2 )
r = .
1 − e cos θ
23. For the hyperbola with foci (0, 0) and (2 c, 0), vertexes ( c ± a, 0), and eccentricity e = c
a (with 0 < a < c),
show that the polar equation is
a ( e2 − 1)
r = .
1 + e cos θ
y
24. The bipolar coordinates19 (ξ, η) of a point P = ( x, y) relative to
two poles F1 = (−a, 0) and F2 = (a, 0) are given by ξ = ln dd1 and P = ( x, y)
2
η = ∠F1 PF2 , where d1 = F1 P and d2 = F2 P , as in the figure on η
the right . Then −∞ < ξ < ∞ except at F1 (which corresponds d1 d2
to ξ = −∞) and F2 (ξ = ∞), while 0 ≤ η ≤ π for points on or
F1 F2
above the x-axis, and π < η < 2π below the x-axis. x
−a 0 a
(a) Show that
a sinh ξ a sin η
x = and y = .
cosh ξ − cos η cosh ξ − cos η
(Hint: Use the distance formula, the Law of Cosines, and the Law of Sines.)
(b) Show that for constants τ 6= 0 and σ 6= 0 or π, the curves ξ = τ and η = σ are the respective circles
a2 a2
( x − a coth τ)2 + y2 = and x2 + ( y − a cot σ)2 = .
sinh2 τ sin2 σ
19 Inspired by the lines of force and equipotential lines for an electric dipole. See pp.55-56 in S TRATTON, J.A.,
Electromagnetic Theory, New York: McGraw-Hill Book Company, Inc., 1941.

CHAPTER 8
Applications of Integrals
8.1 Area Between Curves

The “area under a curve” was defined in Chapter 5 as the area below some curve
y = f ( x) and above the x-axis over some interval. That was a special case of the area
between curves, where in general one curve y = f 1 ( x) is not necessarily always above
another curve y = f 2 ( x) over the entire interval, as in Figure 8.1.1 for an interval [a, b].
y
dx
y = f 1 ( x)
dA
h( x) = | f 1 ( x) − f 2 ( x) |
y = f 2 ( x)
x
0 a x x + dx b
Figure 8.1.1 The area A between two curves y = f 1 ( x) and y = f 2 ( x) over [a, b]
The area A of the region between the curves in Figure 8.1.1 cannot be negative.
Thus, a typical infinitesimal area element d A of the region is of the form h( x) dx, where
the height function h( x) is the ¯ nonnegative ¯ difference in the y-coordinates of the
curves at each x in [a, b]: h( x) = f 1 ( x) − f 2 ( x) ¯. Hence:
¯
The area A between two curves y = f 1 ( x) and y = f 2 ( x) over an interval [a, b] is:
Zb Zb
¯ ¯
A = dA = ¯ f 1 ( x) − f 2 ( x) ¯ dx (8.1)
a a
The interval [a, b] can be replaced by any interval—finite or infinite—over which

the integral is defined. Neither curve is required to be above the x-axis.
252
Area Between Curves • Section 8.1 253
Example 8.1
Find the area between y = e x and y = e− x over [0, 2]. y
y = ex
Solution: Since e ≥ e for x in [0, 2], the¯ height function
x −x
¯ h( x) for the region
between the curves over [0, 2] is h( x) = ¯ e x − e− x ¯ = e x − e− x . The area A of h( x)
the region is thus
Z2 y = e− x
A = ( e x − e− x ) dx x
0 0 2
¯2
¯
x
= e +e −x ¯ = e2 + e−2 − (1 + 1)
¯
0
= 2 (cosh 2 − 1) .
Example 8.2
Find the area of the region bounded by y = x2 and y = x. y
Solution: A “bounded” region will always mean a region of finite area, as op-
posed to unbounded regions. The curves y = x2 and y = x intersect at x = 0 and
x
y=
x = 1, so the region the curves bound is the shaded region shown in the figure y = x2
on the right. Since x ≥ x2 for 0 ≤ x ≤ 1, then the height function for the region x
0 1
is h( x) = | x2 − x | = x − x2 . The region’s area A is then
Z1
A = ( x − x2 ) dx
0
¯1
1 1 ¯ 1 1 1
= x − x2 ¯¯ = − = .
2 3 0 2 3 6
Example 8.3
Find the area of the region bounded by y = sin x and y = cos x over [0, π/3]. y
Solution: As shown in the figure on the right, the curves intersect at x = π
4,
and cos x ≥ sin x for 0 ≤ x ≤ π4 , while sin x ≥ cos x for π4 ≤ x ≤ π3 . The area A 1 y = sin x
of the region thus needs to be split into two integrals:
Zπ/3 y = cos x
A = | sin x − cos x | dx
0 x
Zπ/4 Zπ/3 0 π π
4 3
= (cos x − sin x) dx + (sin x − cos x) dx
0 π/4
Ã ¯π/4 ! Ã ¯π/3 !
¯ ¯
= sin x + cos x ¯¯ + − cos x − sin x ¯¯
0 π/4
µ ¶ p ! µÃ ¶
1 1 1 3 1 1
= p +p − (0 − 1) + − − − −p − p
2 2 2 2 2 2
p p
4 2−3− 3
=
2
254 Chapter 8 • Applications of Integrals §8.1
Formula (8.1) can be extended to find the area between any number of curves, by
splitting the integral over subintervals with different height functions.
Example 8.4
Find the area of the region bounded by y = 6 − x2 , y = x and y = −5 x above the x-axis.
Solution: As shown in the figure on the right, above the x-axis the curve y
y = 6 − x2 intersects the line y = x at x = 2 and intersects the line y = −5 x
y = −5 x
at x = −1. Since 6 − x2 ≥ −5 x over [−1, 0] and 6 − x2 ≥ x over [0, 2], the area
A of the shaded region needs to be split into two integrals: y = 6 − x2
Z0 Z2
¯ ¯ ¯ ¯
A = ¯ 6 − x2 − (−5 x) ¯ dx + ¯ 6 − x2 − x ¯ dx
−1 0
y=x
Z0 Z2
= (6 − x2 + 5 x) dx + (6 − x2 − x) dx
−1 0
Ã ¯0 ! Ã ¯2 ! x
1 3 5 2 ¯¯ 1 1 ¯ −1 0 2
= 6x − x + x ¯ + 6 x − x3 − x2 ¯
¯
3 2 −1 3 2 0
µ ¶ µ ¶
1 5 8 21
= 0 − −6 + + + 12 − − 2 − 0 =
3 2 3 2
For some areas between curves it might be easier to switch the roles of x and y, so that
instead of a vertical height function you would use a horizontal width function.
Example 8.5
Find the area of the region bounded by x = y2 − 2 and y = x.
Solution: As shown in the figure on the right, the parabola x = y2 − 2 y
intersects the line y = x at x = −1 and x = 2. The region has differ-
ent height functions h( x) for −2 ≤ x ≤ −1 and −1 ≤ x ≤ 2, so that two x = y2 − 2
integrals would be required for the area A . However, notice that the y=x
w ( y)
width function w( y) has one definition over¯ the entire region
¯ between
x
the curves x = y2 − 2 and x = y: w( y) = ¯ y − ( y2 − 2) ¯ = y − ( y2 − 2). −2 −1 0 2
Thus, instead of integrating the vertical strips dA = h( x) dx along the
x-axis, integrate the horizontal strips dA = w( y) dy along the y-axis,
from y = −1 to y = 2:
Z2 Z2
¯ ¯
A = w( y) dy = ¯ y − ( y2 − 2) ¯ dy
−1 −1
Z2
= ( y − ( y2 − 2)) dy
−1
¯2
1 2 1 3 ¯
= y − y + 2 y ¯¯
2 3 −1
µ ¶ µ ¶
8 1 1 9
= 2− +4 − + −2 =
3 2 3 2
The area between curves given by polar equations can be y

found similarly. For example, consider curves r = r 1 (θ ) and θ=β
r = r 2 (θ ) with r 1 (θ ) ≥ r 2 (θ ) when α ≤ θ ≤ β as in Figure 8.1.2. r1
The area A of the region between the curves and those angles θ=α
r2
is simply the difference between the “outer” and “inner” areas,
each given by formula (7.20):
x
Zβ Zβ Zβ O
1 2 1 2 1 2
A = 2 1
r dθ − r dθ =
2 2
( r − r 22 ) dθ
2 1 Figure 8.1.2
α α α
In general, to include cases where the “outer” and “inner” curves switch positions, take
the absolute value of the difference:
The area A between two polar curves r = r 1 (θ ) and r = r 2 (θ ) for α ≤ θ ≤ β is:

Zβ
¯
1¯ 2
¯
A = 2
r 1 − r 22 ¯ dθ (8.2)
α
Example 8.6
Find the area between r = 1 + cos θ and r = 1 − cos θ for π
6 ≤ θ ≤ π3 . y r = 1 + cos θ
Solution: Let r 1 (θ) = 1 + cos θ and r 2 (θ) = 1 − cos θ. Since r 1 (θ) > r 2 (θ) for
π
3
π
6 ≤ θ ≤ 3 , the area A of the region (shown in the figure on the right) is
π
θ=
Zπ/3 Zπ/3
¯ 2 ¯
1 ¯ r − r2 ¯ dθ = 1 2 2 π
A = 2 1 2 2 ((1 + cos θ) − (1 − cos θ) ) dθ
θ=
6
π/6 π/6
Zπ/3 ¯π/3 Ãp !
¯ 3 1 p r = 1 − cos θ
= 2 cos θ dθ = 2 sin θ ¯¯ = 2 − = 3−1 x
π/6 π/6 2 2 O
Monte Carlo integration is a technique for approxi- y

mating the area of a region by taking a large number of
random points in a rectangle that encloses the region (see
Figure 8.1.3). The idea is simple:
# of points in the region area of the region
≈
# of points in the rectangle area of the rectangle
x
For example, if 20% of the random points in the rectangle
fall inside the region, then—by randomness—you would Figure 8.1.3
expect the area of the region to be about 20% of the area of
the rectangle. The more random points you take, the better the approximation. Since
the area of the rectangle is known, as well as the number of random points inside the
region and the rectangle, the area of the region is easy to approximate.
Example 8.7
Use Monte Carlo integration to approximate the area of the region y 2
2 y = ex
in the first quadrant above the curves y = e x and y = 2 cos x2 , and
inside the circle x2 + y2 = 9. 3
Solution: The region is the shaded area shown in the figure on the
right, enclosed in a rectangle of width 2 and height 3. The area of the 2
x 2 + y2 = 9
rectangle is 6, and a point ( x, y) in the rectangle is inside the region
if only if the following conditions are met:
1
x2 2 2 2
y = 2 cos x2
y > e and y > 2 cos x and x +y < 9
x
Notice that these conditions are all in the form of inequalities. The 0 1 2 3
Monte Carlo integration is then simple to perform in Octave, using 10 million random points:
octave> N = 1e7;
octave> x = 2*rand(1,N);
octave> y = 3*rand(1,N);
octave> 6*(sum(y > exp(x.^2) & y > 2*cos(x.^2) & x.^2 + y.^2 < 9))/N
ans = 0.94612
The actual value—accurate to 5 decimal places—is 0.94606.

The rand(1,N) command returns an array of N random numbers between 0 and 1. So the statement
x = 2*rand(1,N) stores N random numbers between 0 and 2 in an array for the x-coordinates, and the
statement y = 3*rand(1,N) stores N random numbers between 0 and 3 in an array for the y-coordinates.
2
The statement y > exp(x.^2) returns a value of 1 if the condition y > e x is met, 0 otherwise. Similarly
for the statements y > 2*cos(x.^2) and x.^2 + y.^2 < 9. Joining those three statements with & sym-
bols returns a value of 1 if all three conditions are met, 0 otherwise. The sum command thus counts how
many of the N points are inside the region. Dividing that count by N gives the ratio of points inside the
region, then multiplying by 6 (the area of the rectangle) gives the approximate area of the region.
Note that the size of the rectangle can affect the approximation—generally the larger the rectangle
the more points must be used. Note also in this example that finding the area by using definite integrals
2
would require numerical integration methods, since f ( x) = e x and f ( x) = 2 cos x2 cannot be integrated
in a closed form. In fact, even finding the points of intersection of the three curves—in order to split the
integrals—would require a numerical root-finding method (e.g. Newton’s method).
Exercises
A
For Exercises 1-6, find the area of the region bounded by the given curves.
1. y = x2 and y = 2 x + 3 2. x = − y2 + 2 y and x = 0 3. y = x2 − 1 and y = x3 − 1
4. y = x4 and y = x 5. x = y2 and x = y + 2 6. y = 4 − 4 x2 and y = 1 − x2
7. Find the area between y = 4 x − x2 and y = x over [0, 4].

8. Find the area between y = cosh x and y = sinh x over [0, ∞).
9. Find the area of the region defined by the inequalities 0 ≤ x ≤ y − x ≤ 1 − y ≤ 1.
10. Find the area between r = 1 + cos θ and r = 2 + 2 cos θ.

11. Find the area between r = 1 + cos θ and r = 2 + cos θ.
B
12. Find the area of the region in the first quadrant between the unit y
x 2 − y2 = 1
hyperbola x2 − y2 = 1 and the circle x2 + y2 = 4 (i.e. the shaded region
shown in the figure on the right) in two ways: 2
(a) Integration using formula (8.1). P

(b) Draw a line segment from the origin O to the point of intersection x 2 + y2 = 4
P on the hyperbola, then use the areas of the resulting circular
sector and hyperbolic sector, without resorting to integration. x
O 1 2
Does your answer agree with part (a)? Explain.
13. Find the area common to the four circles of radius 5 shown in Figure 8.1.4. (Hint: Use symmetry.)
C
b y = mx + b
5 m x+
y=
T
5 5 B y = ax2
A 2
y = ax
5 D
(a) Area A (b) Area T
14. Let A be the area of the region bounded by the parabola y = ax2 and the line y = mx + b, where a,
m, and b are positive constants (see Figure 8.1.5(a)). Let T be the area of the triangle △ BCD , where
B and C are where the line intersects the parabola, and the point D on the parabola has the same
x-coordinate as the midpoint of the line segment BC (see Figure 8.1.5(b)). Show that A = 34 T .
15. In Example 8.7 the region was enclosed in the rectangle R = { ( x, y) : 0 ≤ x ≤ 2 , 0 ≤ y ≤ 3 }. Use
Monte Carlo integration to approximate the area of the region again, using a different enclosing
rectangle R :
(a) R = { ( x, y) : 0 ≤ x ≤ 2 , 1 ≤ y ≤ 3 }
(b) R = { ( x, y) : 0 ≤ x ≤ 3 , 0 ≤ y ≤ 3 }
Are the results significantly different than before?
16. Approximate the area of the region bounded by y = x2 and y = cos x in two different ways:
(a) Use Monte Carlo integration with 10 million points.
(b) Use a numerical root-finding method to find the points of intersection of the curves, then use
those points in a numerical integration method to find the area.
C
17. A dog is chained to a fixed point at the circular base of a cylindrical silo. The silo’s radius is 50 ft
and the chain can wrap exactly halfway around the silo. How much total area can the dog roam, not
counting the area inside the silo?
8.2 Average Value of a Function
According to Kepler’s laws of planetary motion, a planet planet

revolving around the Sun follows an elliptical orbit, with
the Sun at one focus of the ellipse, as in Figure 8.2.1. The d
✸
distance d between the planet and the Sun varies over the Sun
ellipse, reaching a minimum distance and a maximum dis-
tance (by the Extreme Value Theorem). How would you find
Figure 8.2.1
the average distance between the planet and the Sun over
one complete orbit? The idea is to generalize the usual notion of an average of num-
bers.
Recall that for n numbers x1 , x2 , . . . , xn the average, denoted by x̄, is simply the sum
of the numbers divided by how many numbers there are, namely
x1 + x2 + · · · + xn
x̄ = .
n
In statistics x̄ is called the mean of x1 , x2 , . . . , xn . This definition makes sense for a
finite set of numbers, but in the case of a planet revolving around the Sun, there are an
uncountably infinite number of distances between the planet and the Sun, making the
above definition impossible to use. A way of taking a sum over an infinite continuum
of values is needed instead. Such a method has already been encountered: the definite
integral, which is merely a sum of a continuum of infinitesimal quantities.
To motivate the definition of the average value y f ( x n −1 )
of a function f over a closed interval [a, b], de- f ( x2 )
f ( xn )
noted by 〈 f 〉, consider a partition f ( x1 )
P = {a = x0 < x1 < x2 < · · · < xn−1 < xn = b} f ( x0 )

...
that divides [a, b] into n subintervals [ x i−1 , x i ]
of equal length ∆ x i = x i − x i−1 = ( b − a)/ n, as in x
a = x0 x1 x2 ... x n−1 x n = b
Figure 8.2.2. The n function values f ( x1 ), f ( x2 ),
. . . , f ( xn ) constitute only a finite subset of all Figure 8.2.2 ∆ x i = x i − x i−1 = (b − a)/n
the function values f ( x) over [a, b], so their av-
erage would be an approximation of the true function average 〈 f 〉, namely:
f ( x1 ) + f ( x2 ) + · · · + f ( xn ) Xn f (x )
i
〈f 〉 ≈ =
n i=1 n
By properties of summations, divide the entire sum by the constant b − a and multiply
each term in the sum by b − a to get:
Average Value of a Function • Section 8.2 259
1 X n b−a 1 X n
〈f 〉 ≈ f (xi ) · = f (xi ) ∆xi
b − a i=1 n b − a i=1
Note that the last summation on the right is just a Riemann sum for the definite in-
Rb
tegral a f ( x) dx, with the points x∗i chosen to be the right endpoints of the intervals
[ x i−1 , x i ] for i = 1 to n. Thus, taking the limit of that sum as n → ∞ (which means in-
cluding more and more function values in the average) yields the following definition:
The average value 〈 f 〉 of a function f over a closed interval [a, b] is:1

Zb
1
〈f 〉 = f ( x) dx (8.3)
b−a a
Example 8.8
Find the average value of f ( x) = x2 over [0, 1].
Solution: By definition, with a = 0 and b = 1,
Z1 Z1 ¯1
1 2 x3 ¯¯ 13 03 1
〈f 〉 = f ( x) dx = x dx = ¯ = − =
1−0 0 0 3 0 3 3 3
Note that this says that if you took all the numbers between 0 and 1 and squared them, then the average
of those squares would be 1/3.
Example 8.9
Find the average value of f ( x) = x2 over [−1, 1].
Solution: By definition, with a = −1 and b = 1,
Z1 Z1 ¯1
1 1 x3 ¯¯ 13 (−1)3 1
〈f 〉 = f ( x) dx = x2 dx = ¯ = − =
1 − (−1) −1 2 −1 6 −1 6 6 3
Note that this is the same as the average over [0, 1], as shown in the previous example. This should
make sense, since the function f ( x) = x2 is symmetric about the y-axis, so the values of f ( x) from [−1, 0]
are the same as those from [0, 1]. The values from [−1, 1] just duplicate the values from [0, 1] and hence
do not change the average.
Example 8.10
Find the average value of f ( x) = sin x over [0, π].
Solution: By definition, with a = 0 and b = π,
Zπ Z ¯π
1 1 π 1 ¯ 1 1 2
〈f 〉 = f ( x) dx = sin x dx = − cos x ¯¯ = − (cos π − cos 0) = − (−1 − 1) =
π−0 0 π 0 π 0 π π π
1 In some statistics or mathematics texts you might see the notation f¯ for the average value.
Example 8.11
2
x2 y
Find the average distance from the ellipse 25 + 9 = 1 to the point y
(4, 0).
( x, y) 3
Solution: Let d represent the distance from any point ( x, y) on
the ellipse to the point (4, 0), as in Figure 8.2.3. If ( x, y) is on d
x2 y2 2 x2 9 2
the ellipse + = 1 then y = 9(1 −
25 9 25 ) = 25 (25 − x ). So by the x
−5 0 (4, 0) 5
distance formula, d is given by
d 2 = ( x − 4)2 + ( y − 0)2 = ( x − 4)2 + y2 −3

9 x2 y2
= ( x − 4)2 + (25 − x2 ) Figure 8.2.3 25 + 9 =1
25
25( x − 4)2 + 9(25 − x)2 25 x2 − 200 x + 400 + 225 − 9 x2
= =
25 25
16 x2 − 200 x + 625
=
25
(4 x − 25)2
d2 = , and so taking square roots gives
25
4 x − 25 4 x − 25 25 − 4 x
d = ± = − =
5 5 5
for −5 ≤ x ≤ 5, since d = (4 x − 25)/5 < 0 on [−5, 5] and the distance cannot be negative. Note that by
symmetry of the ellipse about the x-axis, only the upper half of the ellipse is needed for the average
distance, since the lower half just duplicates the distances. Hence, the average distance is
Z5 ¯5
1 25 − 4 x 1 ¯ 1
〈d〉 = dx = (25 x − 2 x2 ) ¯¯ = (125 − 50 − (−125 − 50)) = 5 .
5 − (−5) −5 5 50 −5 50
x2 y2
Notice that the point (4, 0) is a focus of the ellipse 25 + 9 = 1 (why?), which as it turns out makes the
calculation of the average distance fairly simple.
What if you wanted the average value of a function f that is not easily integrable?
One alternative to numerical integration techniques is the Monte Carlo method.
The idea behind it is simple: go back to the usual definition of an average, by taking
a large number N of random numbers x1 , x2 , . . . , xN in [a, b] and then using the
approximation
f ( x1 ) + f ( x2 ) + · · · + f ( xN )
〈f 〉 ≈ .
N
This might seem like taking a step backward from calculus, and it is, but it is sur-
prisingly useful, as well as simple to implement with a computer. In addition, it can
be shown that as the number of random points in [a, b] increases, the approximations
converge to the actual average.
Average Value of a Function • Section 8.2 261
Example 8.12
The Monte Carlo method is easy to implement in Octave/MATLAB. Typically only a “one-liner” is
needed, owing to Octave’s vectorization—i.e. the ability to perform mathematical operations on entire
arrays of objects all at once.
For example, recall from Example 8.8 that the average value of f ( x) = x2 over [0, 1] is 1/3 = 0.33333 ... .
Approximate the average value using an array of 100 million (108 ) random numbers in [0, 1]:
octave> mean(rand(1,1e8).^2)
ans = 0.3333292094741531
The dot in the rand(1,1e8).^2 command applies the squaring operation (^2) to each of the 108 random
numbers in the array returned by the rand(1,1e8) command. The aggregate mean function then calcu-
lates the array’s mean. Trigonometric, exponential and other functions can be applied to arrays, with
the function evaluating each array element individually. In general the command (b − a).*rand(1,N)+a
will return an array of N random numbers in the interval (a, b).
For example, the function f ( x) = sin ( x2 ) cannot be integrated in a closed form, but its average value
over [π, 2π] can be approximated easily in Octave (actual average = -0.04154374531416104):
octave> mean(sin((pi.*rand(1,1e8)+pi).^2))
ans = -0.04153426177596753
Exercises
A
For Exercises 1-9, find the average value of the function f ( x) over the given interval.
1. f ( x) = 1, over [0, 3] 2. f ( x) = x, over [0, 1] 3. f ( x) = x2 , over [0, 2]
4. f ( x) = x3 , over [0, 2] 5. f ( x) = sin 2 x, over [0, π/2] 6. f ( x) = e x , over [−1, 4]

1
7. f ( x) = x3 , over [−1, 1] 8. f ( x) = sin x, over [−π/2, π/2] 9. f ( x) = , over [1, 3]
x
10. Electrical signals are commonly represented by a periodic waveform x( t), which is a function of
time t and has period T (i.e. T is the smallest positive number such that x( t + T ) = x( t) for all t). The
average power of the waveform is defined as the average value of its square over a single period:
ZT
® 1
x2 ( t) = x2 ( t) dt .
T 0
(a) Find the average power of the waveform x( t) = A cos(ω t + φ), where A > 0 and ω > 0 and φ are all
constants.
(b) The root mean square of a waveform, abbreviated as rms, is the square root of the average
power. Calculate the rms of the waveform from part (a). Write your answer in decimal form as
a percentage of the amplitude A .
11. An electric circuit with a supplied voltage (electromotive force) s C

E , a capacitor with capacitance C , and a resistor with resistance
R , is shown in the picture on the right. When a switch s in the
E
circuit is opened at time t = 0 the current I through the circuit R
begins to decrease exponentially as a function of time t (mea-
sured in seconds after the switch is opened), given by
E − t/RC
I = e
R
for t ≥ 0.
(a) Sketch a rough graph of I as a function of t.
E
(b) Note that at time t = 0 the current is I = R (measured in amperes), which is the familiar formula
from Ohm’s Law. That is the peak value of I . What is the current I at time t = 5RC ? Write your
answer in decimal form as a percentage of the peak current R E
(e.g. 0.42 R
E
, which would be 42%
of the peak current).
(c) Find the average current in the circuit over the time interval [0, 5RC ]. Write your answer in
decimal form as a percentage of the peak current.
12. A spring with spring constant k and damping constant ν connects two point particles with mass m
in a gravitational wave detector. A gravitational wave passes through the detector at time t = 0 and
induces oscillation in the spring, with a period of 2π/Ω and energy E at time t ≥ 0 given by
1 ¡ ¢
E ( t) = mR 2 Ω2 sin2 (Ω t + φ) + ω20 cos2 (Ω t + φ) ,
4
where ω20 = 2k/m, φ = tan−1 (2νΩ/(m(ω20 − Ω2 )), and R is a constant.

(a) Show that the average energy 〈 E 〉 over one period [0, 2π/Ω] of oscillation is
1
〈E〉 = mR 2 (ω20 + Ω2 ) .
8
(b) Suppose a large number of identical detectors of this type are uniformly distributed in a planar
array at a density of σ detectors per unit area. The energy E σ ( t) imparted to each detector at
time t ≥ 0 by a gravitational wave is
E σ ( t) = νΩ2 R 2 sin2 (Ω t + φ) .
Show that the average energy 〈 E σ 〉 over one period [0, 2π/Ω] of oscillation is
1
〈 Eσ 〉 = νΩ2 R 2 .
2
B
2 y2
p
13. For the ellipse ax 2 + b2 = 1, with a > b > 0, the foci are the points ( c, 0) and (− c, 0), where c = a2 − b2 .
Find the average distance from the ellipse to either of its foci in terms of the constants a, b, and c.
14. Write a computer program to use the Monte Carlo method with 1 million random points to approx-
x 2 y2
imate the average distance from the ellipse 25 + 9 = 1 to the point (0, 0). Use symmetry to choose
the smallest interval for the points. Could you have used formula (8.3) instead? Explain.
Arc Length and Curvature • Section 8.3 263
8.3 Arc Length and Curvature

Just like the area of a plane region can be found using calculus, so too can the length
of a plane curve. Along the way the mystery mentioned in a footnote in Chapter 1
will finally be solved: what is the length of the hypotenuse of a right triangle with
infinitesimal sides?
For a function y = f ( x) denote by s the length of the piece of that curve over an
interval [a, b], as in Figure 8.3.1 (a). Call s the curve’s arc length over [a, b].
y
y = f ( x)
ds
ds dx dy
dy
s dx
x
a b dx 1
(a) Arc length s (b) Infinitesimal triangle (c) Noninfinitesimal triangle
Figure 8.3.1 Arc length
By the Microstraightness Property, for a ≤ x 0 is the
infinitesimal change in s over that interval. Notice that you p cannot simplypapply the
Pythagorean Theorem here, since that would make ds = ( dx)2 + ( dy)2 = 0 + 0 = 0,
which is false. The trick is to divide all sides of this infinitesimal right triangle by
dx, which yields the similar—and noninfinitesimal—right triangle shown in Figure
8.3.1(c). The Pythagorean Theorem can then be applied to that triangle:
s s
µ ¶2 µ ¶2
ds 2
dy dy
= 1 + ⇒ ds = 1 + dx
dx dx dx
Summing up those infinitesimal lengths ds then yields the arc length s:
The arc length s of a curve y = f ( x) over [a, b] is:

s
Zb Zb µ ¶2
dy
s = ds = 1+ dx (8.4)
a a dx
That such a formula exists is of course good news, but as you have probably guessed,
the integral cannot be evaluated in a closed form except for a few functions.2 In most
cases numerical integration methods will be required.
2 This will be the last “good news/bad news” scenario in this book. That’s the good news.
Example 8.13
Find the arc length of the curve y = cosh x over [0, 1]. y
Solution: Since
dy
= sinh x, then the arc length s is: y = cosh x
dx
s
Z1 µ ¶2 Z1 p
dy s
s = 1+ dx = 1 + sinh2 x dx 1
0 dx 0
Z1 ¯1 x
¯
= cosh x dx = sinh x ¯¯ = sinh 1 − sinh 0 = sinh 1 ≈ 1.1752 0 1
0 0
Example 8.14
A catenary—a hanging uniform cable whose ends are fastened at the same y
height h a distance L apart—has its lowest point—the apex—a distance
a > 0 above the ground. It can be shown3 that with the apex at (0, a), the h
equation of the catenary is y = a cosh ax . Find the arc length of the catenary. s
Solution: The figure on the right shows the catenary. By symmetry the a
y = a cosh ax
total arc length s is twice the arc length over [0, L/2]:
s x
ZL/2 µ ¶2 ZL/2 q −L/2 0 L/2
dy
s = 2 1+ dx = 2 1 + sinh2 ax dx
0 dx 0
ZL/2 ¯L/2
¯
= 2 cosh ax dx = 2a sinh ax ¯¯ = 2a sinh 2La
0 0
Elliptic Integrals
Suppose you tried to find the circumference s of the ellipse y

x2 y2 x2 y2
+= 1, with a > b > 0, which has eccentricity e = c
, b a2
+ b2 = 1
a2 b2
p a
where c = a2 − b2 . By symmetry, s q
is quadruple the arc
2
x
length of the upper hemisphere y = b 1 − ax2 over [0, a]: −a a
s s −b
Za µ ¶2 Za µ ¶2
dy bx
s = 4 1+ dx = 4 1+ − p dx
0 dx 0 a a2 − x2
s
Za
a2 ( a2 − x2 ) + b 2 x2
= 4 dx
0 a2 ( a2 − x2 )
s
Z
4 a a4 − ( a2 − b 2 ) x2
= dx
a 0 a2 − x2
Now try the trigonometric substitution x = a sin θ , dx = a cos θ dθ :
3 See pp.162-163 in S MITH , C.E., Applied Mechanics: Statics, New York: John Wiley & Sons, Inc., 1976.
s
Zπ/2
4 a4 − (a2 − b2 ) a2 sin2 θ
s = a cos θ dθ
a 0 a2 − a2 sin2 θ
s
Zπ/2
a2 − (a2 − b2 ) sin2 θ ✘✘
= 4 cos
✘
2✘
θ dθ
✘✘
✘sin
1−
0 ✘ θ
s
Zπ/2
a2 − b 2
= 4a 1− sin2 θ dθ
0 a2
Zπ/2 p
c 2 a2 − b 2
s = 4a 1 − e2 sin2 θ dθ (since e2 = = )
0 a2 a2
The last integral is a special case of the elliptic integral of the second kind
Zφ p
E ( k, φ) = 1 − k2 sin2 θ dθ ,
0
with k = e and φ = 2.
This special case is denoted by E ( k) = E ( k, π2 ). Thus, the circum-
π
ference s of the ellipse is:

s = 4a E ( e) = 4a E ( e, π2 )
The integral E ( k) for 0 < k < 1 cannot be evaluated in a closed form. There are ta-
bles4 for certain values of k between 0 and 1, but a number of scientific computing
applications have built-in functions to evaluate elliptic integrals.
x 2 y2
For example, suppose you want to find the circumference s of the ellipse 25 + 9 = 1.
p
Then a = 5, b = 3, c = a2 − b2 = 4, and e = a = 0.8, so s = 4a E ( e) = 20 E (0.8). In the
c
Python-based open-source mathematical software system Sage5 , the elliptic integral

E ( k, φ) is provided by the function elliptic_e(φ, k2 ). Use k = e = 0.8 and φ = π2 :
In [1]: 20*elliptic_e(pi/2,0.8^2)
Out[1]: 25.5269988633981
The circumference is thus approximately 25.5269988633981. In Octave/MATLAB the

function ellipke( e2 ) evaluates the elliptic integral E ( e), with one extra step:
MATLAB>> [K,E] = ellipke(0.8^2);

MATLAB>> 20*E
ans = 25.526998863398131
4 See p.609 in A BRAMOWITZ , M. AND I.A. S TEGUN, Handbook of Mathematical Functions, New York: Dover
Publications, Inc., 1965.

5 Available at https://www.sagemath.org
The parametric formula for arc length can be derived by dividing all sides of the
infinitesimal right triangle in Figure 8.3.1(b) by dt, then applying the Pythagorean
Theorem to the resulting noninfinitesimal right triangle:
s s
µ ¶2 µ ¶2 µ ¶2 µ ¶2
ds dx dy dx dy
= + ⇒ ds = + dt
dt dt dt dt dt
Sum up those infinitesimal lengths ds to obtain the arc length s:
The arc length s of a parametric curve x = x( t), y = y( t), a ≤ t ≤ b is:

s
Zb Z b µ ¶2 µ ¶2
dx dy
s = ds = + dt (8.5)
a a dt dt
Example 8.15
Find the arc length of the parametric curve x = cos3 t, y = sin3 t, 0 ≤ t ≤ π/2. y
dx 2 dy 2
Solution: Since = −3 cos t sin t and
dt = 3 sin t cos t, then the arc length s is:
dt
s 1 x = cos3 t
Zπ/2 µ ¶2 µ ¶2 Zπ/2 q
dx dy y = sin3 t
s = + dt = (−3 cos2 t sin t)2 + (3 sin2 t cos t)2 dt
0 dt dt 0
Zπ/2 p Zπ/2 q x
= 9 cos4 t sin2 t + 9 sin4 t cos2 t dt = 3 sin2 t cos2 t (cos2 t + sin2 t) dt 0 1
0 0
Zπ/2
= 3 sin t cos t dt (since cos t ≥ 0 and sin t ≥ 0 for 0 ≤ t ≤ π/2)
0
¯π/2
3 ¯ 3
= sin2 t ¯¯ =
2 0 2
The polar formula for arc length can be considered a special case of the parametric
formula. A polar curve r = r (θ ) for α ≤ θ ≤ β has Cartesian coordinates x = r (θ ) cos θ
and y = r (θ ) sin θ , so that
dx dr dy dr
= cos θ − r sin θ and = sin θ + r cos θ .
dθ dθ dθ dθ
It is left as an exercise to show that putting these derivatives into formula (8.5)—using
the parameter θ instead of t—yields the polar arc length formula:
The arc length s of a polar curve r = r (θ ) for α ≤ θ ≤ β is:

s
Zβ Zβ µ ¶2
2
dr
s = ds = r + dθ (8.6)
α α dθ
Example 8.16
Prove that the circumference of a circle of radius R is 2πR .
dr
Solution: Use the polar curve r = R for 0 ≤ θ ≤ 2π. Then dθ = 0, so:
s ¯ 2π
Z2π µ ¶2 Z 2π p Z2π
dr ¯
s = r2 + dθ = R 2 + 02 dθ = R dθ = R θ ¯¯ = 2πR X
0 dθ 0 0 0
Curvature
In Chapter 4 you saw one simple measure of curvature: y

the second derivative. From Figure 8.3.2 it is clear that the
parabola y = x2 is less curved at the point (1, 1) than at the
d2 y 1
origin, yet = 2 at each point. So the second derivative—
dx2 y = x2
dy
the rate of change of the slopes dx of the tangent lines—does x
not fully capture the curvature of a curve; more information 0 1
is needed. It turns out that the rate of change of the angles of Figure 8.3.2
the tangent lines is the key to curvature.
First consider a curve with arc length s between two points A and B on the curve.
Let α be the angle between the tangent lines to the curve at A and B, as in Figure
8.3.3(a).
B B
s
α s
α
A A
(a) Small curvature (b) Larger curvature and α
Figure 8.3.3 Curvature between A and B: same s, different α
For the same arc length s but larger angle α as in Figure 8.3.3(b), the curvature ap-
pears greater. This suggests that curvature should measure α relative to s:
The average curvature κ̄ AB of a curve between two points A and B on the curve
is
α
κ̄ AB = (8.7)
s
where s is the arc length of the curve between A and B, and α is the angle between
the tangent lines to the curve at A and B, called the angle of contingence.
Similar to how the instantaneous rate of change of a function at a point is the aver-
age rate of change over an infinitesimal interval, the curvature of a curve at a point
can be defined as the average curvature over an infinitesimal length of the curve:
The curvature κ of a curve at a point A on the curve is
dφ
κ = (8.8)
ds
where s is the arc length function of the curve and φ is the angle that the tangent
line to the curve at A makes with the positive x-axis.
The idea is that moving an infinitesimal length ds along the curve induces an infinites-
imal difference dφ in the angles of the tangent lines. Roughly, as B moves toward A :
dφ α
lim κ̄ AB = lim = κ =
B→ A B→ A s ds
Curvature can be viewed as the instantaneous rate of change of direction of a curve,
but with respect to arc length (i.e. distance traveled) instead of time.
For a curve y = f ( x), recall from formula (3.2) in Section 3.1 that φ = φ( x) = tan−1 f ′ ( x)
at each point ( x, f ( x)) on the curve. Thus, by the Chain Rule:
f ′′ ( x) dx
dφ d (tan−1 f ′ ( x)) 1 + ( f ′ ( x))2
κ = = =
ds ds ds
p
So since ds = 1 + ( f ′ ( x))2 dx by formula (8.4), then:
The curvature κ of a curve y = f ( x) is:
f ′′ ( x)
κ = (8.9)
(1 + ( f ′( x))2 )3/2
The above formula makes κ a function of x. Note also that κ = 0 for a straight line, and
that the curve y = f ( x) is concave up if κ > 0 and concave down if κ < 0.
Example 8.17
Find the curvature of the curve y = x2 for x = 0 and x = 1.
Solution: For f ( x) = x2 , f ′ ( x) = 2 x and f ′′ ( x) = 2, so for any x the curvature κ = κ( x) is:
f ′′ ( x) 2
κ = =
(1 + ( f ′ ( x))2 )3/2 (1 + 4 x2 )3/2
2
In particular, κ(0) = 2 and κ(1) = 53/2
≈ 0.1789. So y = x2 has more curvature at the origin than at (1, 1).
For a parametric curve x = x( t), y = y( t), the curvature κ will be a function of the
dy y′ ( t )
parameter t. Since dx = x′ ( t) by formula (7.14) in Section 7.6, then by formula (7.15):
µ ¶ µ ¶
d dy d y′ ( t)
d2 y dt dx dt x′ ( t) x′ ( t) y′′ ( t) − y′ ( t) x′′ ( t)
= = =
dx2 dx x′ ( t ) ( x′ ( t))3
dt
So by formula (8.9):
d2 y x′ ( t) y′′ ( t) − y′ ( t) x′′ ( t)
dx2 ( x′ ( t))3 x′ ( t) y′′ ( t) − y′ ( t) x′′ ( t)
κ = µ = Ã =
µ ¶2 ¶3/2 µ ′ ¶2 !3/2 Ã µ ′ ¶2 !3/2
dy y ( t) ¡ ¢3/2 y ( t)
1+ 1+ ′ ( x ( t))
′ 2 1+ ′
dx x ( t) x ( t)
Simplify the denominator to obtain the parametric curvature formula:
The curvature κ of a parametric curve x = x( t), y = y( t) is:
x′ ( t) y′′ ( t) − y′ ( t) x′′ ( t)
κ = ¡ ¢3/2 (8.10)
( x′ ( t))2 + ( y′ ( t))2
The derivation of the curvature formula in polar coordinates is left as an exercise:
The curvature κ of a polar curve r = r (θ ) is:
( r (θ ))2 + 2 ( r ′(θ ))2 − r (θ ) r ′′ (θ )

κ = ¡ ¢3/2 (8.11)
( r (θ ))2 + ( r ′ (θ ))2
Example 8.18
Find the curvature of a circle of radius R .
Solution: Use the polar curve r = r(θ) = R , so that r′ (θ) = 0 = r′′ (θ):
(r(θ))2 + 2 (r′ (θ))2 − r(θ) r′′ (θ) R 2 + 2 · 02 − R · 0 R2 1

κ = ¡ ¢3/2 = 2 2 3/2
= 3
=
2
(r(θ)) + (r (θ))
′ 2 (R + 0 ) R R
A circle thus has constant curvature, as you would expect by symmetry. It turns out that any planar
curve with constant curvature is either a line or part of a circle.6
6 See pp.62-63 in O’N EILL , B., Elementary Differential Geometry, New York: Academic Press, Inc., 1966.
Exercises
A
For Exercises 1-10, find the arc length of the given curve over the given interval.
1. y = x3/2 ; 1 ≤ x ≤ 4 2. y = x2 ; 0 ≤ x ≤ 1 3. y = x2/3 ; 1 ≤ x ≤ 8
x2 ln x x4 1 ex + 1
4. y = − ; 1≤x≤2 5. y = + ; 1≤x≤2 6. y = ln ; 1≤x≤2
4 2 4 8 x2 ex − 1
7. x = e t cos t, y = e t sin t ; 0 ≤ θ ≤ π 8. x = cos t + t sin t, y = sin t − t cos t ; 0 ≤ t ≤ π
9. polar curve r = 1 + cos θ ; 0 ≤ θ ≤ 2π 10. polar curve r = eθ ; 0 ≤ θ ≤ 2
11. Find the arc length of the curve in Example 8.15 for 0 ≤ t ≤ π.
12. Use formula (8.5) to find the circumference of the unit circle using two different parametrizations:
1 − t2 2t
(a) x = cos t, y = sin t, 0 ≤ t ≤ 2π (b) x = 2
, y= , −∞ < t < ∞
1+t 1 + t2
For Exercises 13-18 find the curvature of the given curve at the indicated points.
x2 y2
13. y = sin x at x = 0 and x = π
2 14. y = ln x at x = 1 15. a2
+ b2 = 1 at (a, 0) and (0, b)
16. y = e x at x = 0 17. x2 − y2 = 1 at (1, 0) 18. r = 1 + cos θ at θ = 0
B
19. Prove formula (8.6). 20. Prove formula (8.11).
x
21. Let α and β be nonzero constants. Show that the arc length s of y = β sin α
over the interval [0, x0 ]
can be put in terms of the elliptic integral E (k, φ):
q Ãs !
2 2
β2 x0
s = α +β · E ,
α2 + β2 α
22. For −1 < k < 1 and −1 ≤ x ≤ 1, define u( x) and K by

Zx Z1
dt dt
u( x ) = p p and K = u(1) = p p
0 1−t 2 2
1−k t 2 0 1−t2 1 − k2 t2
so that all the square roots are defined and positive.
(a) Show that u is an increasing function of x and thus has an inverse function, call it x = sn u, with
domain [−K, K ] and range [−1, 1].
p p
(b) Define cn u = 1 − sn2 u and dn u = 1 − k2 sn2 u. Show that:
d d d
(sn u) = cn u dn u , (cn u) = − sn u dn u , and (dn u) = −k2 sn u cn u
du du du
The functions sn u, cn u, and dn u are called the Jacobian elliptic functions.
Zu
(c) Suppose that sin φ = sn u. Show that E (k, φ) = dn2 v dv.
0
23. The ends of a 50 ft long catenary are fastened 40 ft apart. Use a numerical method to find how
much the apex dips below the ends. (Hint: Solve for a in Example 8.14, then use symmetry.)
Surfaces and Solids of Revolution • Section 8.4 271
8.4 Surfaces and Solids of Revolution

Long before calculus was invented the ancient Greeks (e.g. Archimedes) discovered
the formulas for the volume and surface area of familiar three-dimensional objects
such as the sphere.7 Volumes and surface areas of arbitrary solids and surfaces can
be found using multivariable calculus. However, single-variable calculus can be used
in the special case of the objects possessing symmetry about an axis, via methods that
involve revolving a curve or region in the x y-plane around an axis.
For example, revolve a curve y = f ( x) ≥ 0 around the x-axis, for a ≤ x ≤ b. This
produces a surface of revolution in three dimensions, as in Figure 8.4.1(a).
y y y
y = f ( x) y = f ( x) ds
f ( x) f ( x + dx)
x x x
a b a x b dx
(a) Revolve y = f ( x) around x-axis (b) dS over [ x, x + dx] (c) Side view of frus-
trum
Figure 8.4.1 Surface of revolution over [a, b] with lateral surface area S
To find the total lateral surface area S , pick x in [a, b) then find r1
the infinitesimal surface area dS swept out over the infinitesimal
interval [ x, x + dx], as in Figure 8.4.1(b). By the Microstraight- l
ness Property, the curve y = f ( x) is a straight line segment of r2
length ds over that interval, so that the infinitesimal surface is a
frustrum—a right circular cone with the vertex chopped off by a Figure 8.4.2
plane parallel to the base circle. From geometry8 you might re-
call the formula for the lateral surface area of the frustrum in Figure 8.4.2: p π (r 1 + r 2 ) l .
Use that formula with r 1 = f ( x), r 2 = f ( x + dx) = f ( x) + dy, and l = ds = 1 + ( f ′( x))2 dx
(by formula (8.4) in Section 8.3) as in Figure 8.4.1(c), so that dS is
q
dS = π ( f ( x) + ( f ( x) + dy)) 1 + ( f ′( x))2 dx
q q
= 2 π f ( x) 1 + ( f ( x)) dx + π 1 + ( f ′ ( x))2 dy dx
′ 2
q
= 2 π f ( x) 1 + ( f ′ ( x))2 dx + 0
7 See Propositions 33 and 34 in On the Sphere and Cylinder, Book I, appearing in H EATH , T.L., The Works of
Archimedes, Mineola, NY: Dover Publications, Inc., 2002. This work is also available at https://archive.org
8 See pp.136-137 in W ELCHONS A.M. AND W.R. K RICKENBERGER , Solid Geometry, Boston: Ginn & Co., 1936.
since dy dx = f ′ ( x) ( dx)2 = 0. The surface area S is then the sum of all the areas dS :
The surface area S of the surface of revolution obtained by revolving the curve
y = f ( x) ≥ 0 around the x-axis for a ≤ x ≤ b is:
Zb Zb q
S = dS = 2 π f ( x) 1 + ( f ′( x))2 dx (8.12)
a a
For a general curve y = f ( x), possibly negative in [a, b], the surface area S is:
Zb Zb Zb
¯ ¯q
S = dS = 2 π | y | ds = 2 π f ( x) ¯ 1 + ( f ′ ( x))2 dx
¯ (8.13)
a a a
Note that formula (8.13) holds by symmetry

¯ ¯and formula (8.12).
y
¯ ¯
A curve y = f ( x) < 0 and the curve y = f ( x) = − f ( x) are sym- ¯ ¯
y = ¯ f ( x) ¯
metric with respect to the x-axis, as in the figure on the right.
Thus, both curves sweep out the same surface of revolution
x
when revolved around the x-axis. This means formula (8.13) a b
also holds if y = f ( x) changes sign in [a, b]: similar to the
area between two curves, you would split the integral over y = f ( x)
different subintervals depending on the sign.
Example 8.19
Show that the surface area of a sphere of radius r is 4πr2 .
Solution: Use thepcircle x2 + y2 = r2 . The upper half of that circle is the y
curve y = f ( x) = r2 − x2 over the interval [−r, r], as in the figure on p
the right. Revolving that curve around the x-axis produces a sphere y= r2 − x2
of radius r, whose surface area S is:
Zr q x
S = 2 π f ( x) 1 + ( f ′ ( x))2 dx −r r
−r
s µ ¶2
Zr p
2 2
−x
= 2π r − x 1+ p dx
−r r2 − x2
s
Zr p
✘ ✘ r2
= 2 π✘✘ r2 − x2 ✘ ✘ dx
−r ✘
r − x2
2
¯r
¯
= 2 π rx ¯¯ = 4 πr2 X
−r
A similar derivation using a frustrum yields the surface area S of the surface of
revolution obtained by revolving a curve y = f ( x) around the y-axis, for 0 ≤ a ≤ x ≤ b:
Zb Zb Zb q
S = dS = 2 π | x | ds = 2 π x 1 + ( f ′( x))2 dx (8.14)
a a a
Now suppose you revolve the region between a curve y = f ( x) ≥ 0 and the x-axis
around the x-axis, for a ≤ x ≤ b (see Figure 8.4.3(a)). This produces a solid of revo-
lution in three dimensions, as in Figure 8.4.3(b). Notice that this solid consists of the
surface of revolution as before along with its interior.
y y y
y = f ( x) y = f ( x)
ds
dy
x x f ( x) f ( x + dx)
a b a x b
x
dx
(a) Revolve region around x-axis (b) Solid (c) Infinitesimal strip
Figure 8.4.3 Solid of revolution over [a, b] with volume area V
The goal is to find the volume V of this solid. The idea is to divide the solid into slices,
like a loaf of bread. First, the infinitesimal volume dV of the frustrum swept out by a
strip of infinitesimal width dx at x in [a, b)—shown in Figure 8.4.3(c)—is needed. By
the Microstraightness Property the curve y = f ( x) is a straight line of length ds over
the interval [ x, x + dx]. There is thus a right triangle at the top of the strip—unshaded
in Figure 8.4.3(c)—whose area A is zero: A = 21 ( dy)( dx) = 12 f ′ ( x)( dx)2 = 0.
That triangle thus contributes no volume when revolved around the dx
x-axis: the volume dV swept out by that strip all comes from the
shaded rectangle of height f ( x) and width dx. That rectangle sweeps f ( x)
out a right circular cylinder of radius f ( x) and height dx (see Figure
8.4.4). The volume of a right circular cylinder of radius r and height h
is defined as the area of the base circle times the height: πr 2 h. Hence,
Figure 8.4.4
dV = π ( f ( x))2 dx .
The total volume V of the solid is then the sum of all those infinitesimal volumes dV :
The volume V of the solid of revolution obtained by revolving the region between
the curve y = f ( x) and the x-axis around the x-axis for a ≤ x ≤ b is:
Zb Zb
V = dV = π ( f ( x))2 dx (8.15)
a a
This method for finding the volume is called the disc method, since the cylinder of
volume dV resembles a disc. Think of the discs as being similar to infinitesimally thin
slices of a loaf of bread. Notice that absolute values are not needed in formula (8.15)
since f ( x) is squared, so the formula holds even when f ( x) is negative.
Example 8.20
Show that the volume of a sphere of radius r is 34 πr3 .
Solution: Use the circle x2 + y2 = rp2 . Revolve the region between the y
upper half of the circle y = f ( x) = r2 − x2 and the x-axis around the p
x-axis over the interval [−r, r], as in the figure on the right. The solid y= r2 − x2
of revolution swept out is a sphere of radius r, whose volume V is:
Zr Zr ¯r x
2 2 2 2 1 3 ¯¯ −r r
V = π ( f ( x)) dx = π ( r − x ) dx = π r x − π x ¯
−r −r 3 −r
µ ¶ µ ¶
3 1 3 3 1 3 4 3
= π r − π r − −π r + π r = π r X
3 3 3
Instead of memorizing formula (8.15), try to remember the more generic approach
of revolving an infinitesimal rectangular strip around an axis, which might not be the
x-axis. The idea is to find the radius r and height h—typically dx or dy—of the disc
swept out by that strip, so that the disc’s volume is dV = πr 2 h. Then integrate dV over
the appropriate interval to find the volume V of the entire solid.
Example 8.21
Suppose the region bounded by the curve y = x2 and the x-axis for 0 ≤ x ≤ 1 is revolved around the line
x = 1. Find the volume of the resulting solid of revolution.
Solution: The region is shaded in the figure on the right. Since the y
x=1
region is revolved around a vertical axis, the disc method will use
discs with height dy, not dx. At a point x in [0, 1] go up to the curve 1
y = x2
y = x2 and draw a horizontal rectangular strip to the line x = 1, as
shown in the figure. Let h = dy and revolve that strip around the r
line x = 1, producing a disc of radius r − 1 − x and height h = dy. x2 = y h
p x
Since y = x2 implies x = y, the volume dV of that disc is 0 x 1
p p
dV = π r2 h = π (1 − x)2 dy = π (1 − y)2 dy = π (1 − 2 y + y) dy .
The volume V of the entire solid is then the sum of those volumes dV along the y-axis for 0 ≤ y ≤ 1:
Z1 Z1 µ ¶ ¯1 µ ¶
p 4 3/2 1 2 ¯¯ 4 1 π
V = dV = π (1 − 2 y + y) dy = π y − y + y ¯ = π 1 − + =
0 0 3 2 0 3 2 6
The shell method can be used for finding the vol- y

ume of a solid with a “hole” in the middle, as in the
solid of revolution produced by revolving the shaded y = f ( x)
region in the figure on the right around the y-axis. The
hole in the solid between x = −a and x = a is a result of x
−b −a 0 a b
the gap between the y-axis and the region.
To find the volume V of that solid, at a point x in [a, b) form an infinitesimal strip of
width dx from the x-axis up to the curve y = f ( x), as in Figure 8.4.5(a).
y
y
ds
dy x dx
y = f ( x)
f ( x)
f ( x) f ( x + dx)
x x
0 a x b dx
(a) Revolve region around y-axis (b) Infinitesimal strip (c) Cylindrical shell
Figure 8.4.5 Shell method
Just like the strip in the disc method, the right triangle at the top of this strip—as
in Figure 8.4.5(b)—has zero area and thus does not contribute to the volume dV of the
right circular cylindrical shell swept out by the strip, shown in Figure 8.4.5(c). The
volume of that shell is just the volume of the “outer” cylinder of radius x + dx minus
the volume of the “inner” cylinder of radius x, both with height f ( x):
dV = π ( x + dx)2 f ( x) − π x2 f ( x)
✘✘ ✯0
✟ ✘ ✘
=✘π x✘2✘
f ( x) + 2 π x f ( x) dx + π✟
( dx π x✘2✘
✟)2 f ( x) − ✘ f ( x)
= 2 π x f ( x) dx
The volume V of the entire solid is then the sum of those volumes dV , using an abso-
lute value to handle any sign for f ( x):
The volume V of the solid of revolution obtained by revolving the region between
the curve y = f ( x) and the x-axis around the y-axis for 0 ≤ a ≤ x ≤ b is:
Zb Zb
¯ ¯
V = dV = 2 π x ¯ f ( x) ¯ dx (8.16)
a a
Example 8.22
Suppose the region bounded by the curve y = x2 and the x-axis for 0 ≤ x ≤ 1 is revolved around the
y-axis. Find the volume of the resulting solid of revolution.
Solution: The region is shaded in the figure on the¯ right. ¯ The vertical strip at y
x in [0, 1) with infinitesimal width dx and height ¯ f ( x) ¯ = f ( x) is shown in the
figure. That strip produces the shell with volume dV in formula (8.16), so by the 1
y = x2
shell method the volume V of the solid of revolution is:
Z1 Z1 Z1 ¯1
¯ ¯ 2 π 4 ¯¯ π
V = dV = ¯ ¯
2 π x f ( x) dx = 2 π x · x dx = x ¯ =
0 0 0 2 2 x
0
0 x 1
The volume dV in formula (8.16) can be generalized to dV = 2πrhw, where r is the

distance from the axis of revolution to a generic vertical strip of infinitesimal width w
in the region, and h is the height of the strip.
Example 8.23
Suppose the region bounded by the curves y = x2 and y = x is revolved around the y-axis. Find the
volume of the resulting solid of revolution.
Solution: The region is shaded in the figure on the right, along with a vertical y
strip with infinitesimal width w = dx at the distance r = x from the y-axis in
the region and height h = x − x2 . That strip produces the shell with volume 1 y=x
dV = 2πrhw = 2π x( x − x2 ) dx, so that the volume V of the solid of revolution is:
Z1 Z1 ¯1 y = x2
2π 3 π ¯ π
V = dV = 2 π x ( x − x2 ) dx = x − x4 ¯¯ = x
0 0 3 2 0 6 0 x 1
Exercises
A
For Exercises 1-3, find the surface area of the surface of revolution produced by revolving the given
curve around the x-axis for the given interval.
p 3
1. y = 4 − x2 for 1 ≤ x ≤ 2 2. y = cosh x for 0 ≤ x ≤ 1 3. y = x6 + 21x for 1 ≤ x ≤ 3
For Exercises 4-6, find the volume of the solid of revolution produced by revolving the region between
the given curve and the x-axis around the x-axis for the given interval.
p
4. y = x3 for 0 ≤ x ≤ 1 5. y = sin x for 0 ≤ x ≤ π 6. y = x for 0 ≤ x ≤ 1
For Exercises 7-9, find the volume of the solid of revolution produced by revolving the region between
the given curve and the x-axis around the y-axis for the given interval.
p
7. y = sin ( x2 ) for 0 ≤ x ≤ π 8. y = sin x for 0 ≤ x ≤ π 9. y = x2 − x3 for 0 ≤ x ≤ 1
10. Revolve the region in Example 8.23 around the line x = 1 and find the volume of the resulting solid.
x2 y2
11. Revolving the ellipse a2
+ = 1 around the x-axis produces an ellipsoid, for a > b > 0. Show that
b2 ¡ ¢
the surface area of the ellipsoid is 2 π b2 1 + eb
a
sin−1 e , where e is the eccentricity of the ellipse.
4
12. Show that the volume inside the ellipsoid from Exercise 11 is 3 π a b2 .
13. Find the surface area and volume of a right circular cone of radius r and height h.
14. Formulas (8.13), (8.15) and (8.16) can be extended to include regions over infinite intervals—the
integrals in those formulas simply become improper integrals. Consider the region between the
curve y = 1x and the x-axis over the interval [1, ∞). Revolve that region around the x-axis.
(a) Show that the surface area of the resulting surface of revolution is infinite.
(b) Show that the volume of the resulting solid of revolution is π.
15. For 0 < a < b, revolving the region inside the circle ( x − b)2 + y2 = a2 around the y-axis produces a
donut-shaped solid of revolution called a torus. Show that the volume of the torus is 2π2 a2 b.
16. Use formula (8.14) and symmetry to show that the torus from Exercise 15 has surface area 4π2 ab.
Applications in Physics and Statistics • Section 8.5 277
8.5 Applications in Physics and Statistics

This chapter concludes with a few applications showing how some familiar discrete
sums can be replaced by integrals, which are essentially continuous sums.
Center of Gravity
Suppose a thin uniform rod has n > 1 masses m1 , . . ., m n attached, with m1 and m n
at the ends. The center of gravity of the masses is the point where—due to the
Earth’s gravity—the rod would be balanced if a fulcrum were placed there (see Figure
8.5.1(a)). Imagine the rod as part of the x-axis and the weights as point masses—with
each mass m k at xk —and let the center of gravity be at x̄, as in Figure 8.5.1(b).
m1 m2 ··· m n −1 mn
m1 m2 ··· m n −1 mn
x
x1 x2 x̄ x n −1 xn
(a) Rod balanced on fulcrum (b) Masses and x̄ on x-axis
Figure 8.5.1 Center of gravity x̄ for masses m 1 ,... , m n
The rod is balanced if the masses do not rotate the rod, i.e. the total torque is zero.
Torque is defined here as force times position relative to x̄. Each mass m k applies
a force m k g to the rod—where g is the (downward) acceleration due to the Earth’s
gravity—at position ( xx − x̄) relative to x̄. The total torque is thus zero if
( m1 g) ( x1 − x̄) + ( m2 g) ( x2 − x̄) + · · · + ( m n g) ( xn − x̄) = 0
so that solving for x̄ yields:

Pn
m1 gx1 + · · · + m n gxn m1 x1 + · · · + m n xn k=1
m k xk
x̄ = = = Pn (8.17)
m1 g + · · · + m n g m1 + · · · + m n k=1
mk
Each quantity m k xk is called the moment of the mass m k . Thus, x̄ is the sum of
the moments divided by the total mass. This idea can be extended to regions in the
x y-plane, using an integral of a continuum of moments instead of a finite sum. The
center of gravity of a planar region is defined as the point such that any force along
a line through that point produces no rotation of the region about that line.9 There
should thus be zero torque in both the x and y directions, so the idea is to apply formula
(8.17) in both directions to obtain the region’s center of gravity ( x̄, ȳ).
9 For a proof that such a point exists, see p.206 in B ROWN, F.L., Engineering Mechanics, 2nd ed., New York: John
Wiley & Sons, Inc., 1942. Some texts use the terms “center of mass” or “centroid” instead of “center of gravity,” and
there are differences in the meanings. However, for the situation presented here, where the gravitational field is
assumed to have constant magnitude and direction throughout the region, they all mean the same thing.
A region can be thought of as a lamina—a thin plate with uniform density. Take the
area of the region as its mass, which makes sense given the uniform density. For the
region between two curves y = f 1 ( x) and y = f 2 ( x) over [a, b], with f 1 ( x) ≥ f 2 ( x), take
a vertical slice of width dx at some x, as in Figure 8.5.2(a). By the same arguments
used in Section 8.4, all the area from that strip comes from the rectangle of height
f 1 ( x) − f 2 ( x) and width dx (see the shaded rectangle in Figure 8.5.2(b)).
y y
y = f 1 ( x)
ds
dy
f 1 ( x) − f 2 ( x) ( x + 21 dx, 12 ( f 1 ( x) + f 2 ( x)))
y = f 2 ( x)
x x
0 a x b dx
(a) Region between y = f 1 ( x) and y = f 2 ( x) (b) Infinitesimal strip over [ x, x + dx]
Figure 8.5.2 Center of gravity of a region
By the assumption of uniform density, the center ¡ of1 gravity of that rectangle
¢ is
1
clearly its geometric center, whose coordinates are x + 2 dx, 2 ( f 1 ( x) + f 2 ( x)) . The en-
tire mass of the strip can be treated as if it is concentrated at that point. The moment
m x of the strip about the x-axis is its mass times the position of its center of gravity
relative to the x-axis (i.e. its y coordinate):
2 2
m x = ( f 1 ( x) − f 2 ( x)) dx · ( 21 ( f 1 ( x) + f 2 ( x))) = 1
2 (( f 1 ( x)) − ( f 2 ( x)) ) dx
Similarly the moment m y of the strip about the y-axis is its mass times the x coordi-
nate of its center of gravity:
2
m y = ( f 1 ( x) − f 2 ( x)) dx · ( x + 21 dx) = x ( f 1 ( x) − f 2 ( x)) dx + 1
2 ( f 1 ( x) − f 2 ( x)) ( dx)
= x ( f 1 ( x) − f 2 ( x)) dx
The moments M x and M y of the entire region about the x-axis and y-axis, respectively,
are defined as the sum of the respective moments m x and m y of all strips over [a, b]:
Zb Zb Zb Zb
1 2 2
Mx = mx = 2 (( f 1 ( x)) − ( f 2 ( x)) ) dx and M y = my = x ( f 1 ( x) − f 2 ( x)) dx
a a a a
Note in formula (8.17) that the denominator is the sum of all the masses in the system.
For the region that total mass would simply be its area M :
Zb
M = ( f 1 ( x) − f 2 ( x)) dx
a
Dividing the moments M x adn M y by M yields the formula for the center of gravity:
The center of gravity ( x̄, ȳ) of the region between the curves y = f 1 ( x) and y = f 2 ( x)
over [a, b], with f 1 ( x) ≥ f 2 ( x), is given by:
Zb Zb
x ( f 1 ( x) − f 2 ( x)) dx 1
2
(( f 1 ( x))2 − ( f 2 ( x))2 ) dx
My a Mx a
x̄ = = Zb and ȳ = = Zb
M M
( f 1 ( x) − f 2 ( x)) dx ( f 1 ( x) − f 2 ( x)) dx
a a
(8.18)
Example 8.24
Find the center of gravity of the region bounded by the curve y = x2 and the x-axis for 0 ≤ x ≤ 1.
Solution: The region is shaded in the figure on the right. Using y = f 1 ( x) = x2 y
and y = f 2 ( x) = 0 in formula (8.18) yields
1
Z1 Z1 ¯1
¯ y = x2
1 2 1 4 1 5 1
Mx = 2 ( f 1 ( x)) dx = 2 x dx = 10 x ¯¯ = 10
0 0 0 ȳ
Z1 ¯1
Z1
3 1 4 ¯
¯ x
My = x f 1 ( x) dx = x dx = 4 x ¯ = 41 0 x̄ 1
0 0 0
Z1 Z1 ¯1
¯
M = f 1 ( x) dx = x2 dx = 13 x3 ¯¯ = 31
0 0 0
so that the center of gravity ( x̄, ȳ) is:
My 1/4 3 Mx 1/10 3
x̄ = = = and ȳ = = =
M 1/3 4 M 1/3 10
Example 8.25
Find the center of gravity of the region bounded by the curves y = x and y = x2 . y
Solution: The region is shaded in the figure on the right. Using y = f 1 ( x) = x and
1
y = f 2 ( x) = x2 in formula (8.18) yields y=x
Z1 Z1 ¯1 ȳ y = x2
1 2 2 1 2 4 1 3 1 5
¯ 1
Mx = 2 (( f 1 ( x)) − ( f 2 ( x)) ) dx = 2 ( x − x ) dx = 6 x − 10 x ¯¯ = 15
0 0 0
x
Z1 Z1 ¯1 0 x̄ 1
¯
My = x ( f 1 ( x) − f 2 ( x)) dx = ( x2 − x3 ) dx = 13 x3 − 14 x4 ¯¯ = 1
12
0 0 0
Z1 Z1 ¯1
¯
M = ( f 1 ( x) − f 2 ( x)) dx = ( x − x2 ) dx = 12 x2 − 13 x3 ¯¯ = 61
0 0 0
so that the center of gravity ( x̄, ȳ) is:
My 1/12 1 Mx 1/15 2
x̄ = = = and ȳ = = =
M 1/6 2 M 1/6 5
Work
Suppose that a constant force displaces an object along a line in the same direction
in which the force is applied. The work done by the force is defined as the force times
the displacement. For example, if the constant force F moves an object from position
x = a to x = b on the x-axis, as in Figure 8.5.3(a), then the work W done by the force is:
W = force × displacement = force × (final position − initial position) = F · ( b − a)
F ( x + dx)
F
x F ( x)
a b dx
(a) Constant force F (b) Variable force F over [ x, x + dx]
Figure 8.5.3 Work W as the effect of a force F displacing an object from x = a to x = b
Suppose now that the force F is a function of position x over [a, b]: F = F ( x). By the
Microstraightness Property, over an infinitesimal interval [ x, x + dx] the curve y = F ( x)
is a straight line, as in Figure 8.5.3(b). How should the work dW performed by F over
this infinitesimal interval be defined? After all, F is not constant over [ x, x + dx]—it
takes every value between F ( x) and F ( x + dx). It is left as an exercise to show that any
value in that range can be used—they all result in the same amount F ( x) dx for the
work performed.10
For example, suppose you use the value halfway between F ( x) and F ( x + dx) as the
value of F : 21 (F ( x) + F ( x + dx)). Then the work dW as force times displacement is:
1 1
dW = (F ( x) + F ( x + dx)) dx = (F ( x) + F ( x) + F ′ ( x) dx) dx
2 2
1 ✯0
✟
= F ( x) dx + F ′ ( x)✟ ✟)2
( dx
2
= F ( x) dx
Define the total work W over [a, b] as the sum of all the dW :
The work performed by a force F ( x) in displacing an object along the x-axis from
x = a to x = b is: Zb Zb
W = dW = F ( x) dx (8.19)
a a
10 Note how this is different than claiming that F is “essentially constant over small intervals,” as most textbooks
do. Instead, the additional infinitesimal force beyond F ( x) contributes zero work over [ x, x + dx].
Before continuing, some possible confusion needs to be cleared up. First, force is
always a vector—it has both a magnitude and a direction. For the forces considered
here, which act in a single dimension (e.g. along the x-axis), by convention the direc-
tion of the force is indicated by its sign: positive in the direction toward +∞, negative
in the direction toward −∞. So a force of 3 N acts in the opposite direction as a force
of −3 N, but they have the same magnitude | 3 | = 3.
Second, work is not a vector—it is a scalar, meaning it has a magnitude but no
direction. That magnitude can have any sign, though. Work is positive if the object is
displaced in the same direction as the force, but is negative if the displacement is in
the opposite direction of the force. For example, if you lift an object straight up from
the ground, then you did positive work—the object moved in the same direction as the
force you used. However, the force of gravity did negative work on the object as you
lifted, since gravity works downward yet the object moved upward.
Last, zero work is done by a force if no displacement in its direc- N
tion occurs. In particular, forces acting perpendicular to the line of Fµ
F
displacement perform no work. For example, consider an object of
mass m on a flat horizontal table top as in the figure on the right. If
you push that object to the right with a force F (performing positive −mg
work), then both the downward force of gravity − mg and the upward normal force N
exerted by the table perform zero work on the object. The force of friction Fµ from the
table surface does negative work, as it opposes the force F . As an another example, no
work is performed by holding a 100 lb object still and above the ground.
Example 8.26
Hooke’s law states that a coiled spring has an elastic restoring force compress stretch
F = −kx, where x is the displacement of the end of the spring from
its equilibrium position as the spring is stretched or compressed,
and k > 0 is the spring constant—or stiffness coefficient—specific to
the spring. This force always tries to restore the spring to its equi- x
librium position, and the law holds only for a limited range of x.
0
For a spring laid horizontally imagine it lies on the x-axis with the
equilibrium position at x = 0, as in the figure on the right.
(a) Find the spring constant k if a force of 2 N stretches the spring by 4 cm.
(b) Use part (a) to find the work performed by compressing the spring 3 cm.
Solution: (a) The force required to stretch the spring by an amount x is F = kx, since that force must
2N
counter the restoring force. Thus, k = Fx = 4cm = 0.2N
04m = 50 N/m.
(b) By part (a) the force required to compress the string to position x is F ( x) = kx = 50 x, since again it
must counter the restoring force. Thus, since 3 cm is 0.03 m, the work W performed is:
Z−0.03 Z−0.03 ¯−0.03
¯
W = F ( x) dx = 50 x dx = 25 x2 ¯¯ = 25 (−0.03)2 − 0 = 0.0225 Nm
0 0 0
Probability
Suppose you flip two evenly balanced pennies and let X be the number of heads
in the result. Then X is a discrete random variable—discrete because it can take
only a discrete set of values (0, 1 and 2); random because its value is left to chance.
The probability of a penny being flipped heads is 50% = 12 , i.e. that is its theoretical
likelihood since heads and tails are equally likely. The sample space S of all possible
outcomes is the set S = {TT, T H, HT, HH }, where H is heads and T is tails (e.g. HT
means the first penny came up heads and the second came up tails). Figure 8.5.4(a)
shows a bar chart of the probabilities—as numbers between 0 and 1—with P ( X = x)
denoting the probability of the event that X equals the number x. Notice that the
sum of the probabilities is 1, and P ( X = x) = 0 if x is not 0, 1, or 2.
P ( X = x) f ( x)
probability
1
2
1
4
x x
0 1 2 a b
(a) Discrete: P ( X = x) (b) Continuous: P (a < X < b)
Figure 8.5.4 Probability: discrete vs continuous random variables
The idea behind a continuous random variable X is to fill in those gaps between
the bars in Figure 8.5.4(a), so that X would represent a continuous quantity, e.g. time,
distance, temperature. Rather than finding P ( X = x) you would find the probability
that X is in a continuum such as an interval, e.g. P (a < X < b) (see Figure 8.5.4(b)).
For a continuous random variable X define P ( X = x) = 0 for all real x, and define
the probability density function for X as a function f ( x) ≥ 0 such that:
Z∞
(a) f ( x) dx = 1
−∞
Zb
(b) P (a < X < b) = f ( x) dx for all a 
and ≥). In the remainder of this section it will be assumed that all random variables
are continuous, for which the sample space is typically all of R or some interval, finite
or infinite (e.g. (0, ∞)).
Example 8.27
Let X be the lifetime—i.e. the time to failure—of an electronic component. If the average lifetime of the
component is 700 days, then the probability density function f ( x) for the random variable X is
(
λ e−λ x if x ≥ 0,
f ( x) = (8.20)
0 if x < 0
1
where λ = 700 and x is the number of days. In this case X is said to have the exponential distribution
with parameter λ. Find the probability that the lifetime of the component is:
(a) between 600 and 800 days
(b) greater than 700 days
Solution: (a) The probability is:
Z800 Z800 ¯800
x x ¯ 800 600
1
P (600 < X < 800) = f ( x) dx = 700 e
− 700
dx = − e− 700 ¯¯ = − e− 700 + e− 700 ≈ 0.1055
600 600 600
Thus, there is about a 10.55% chance that the component’s lifetime will be between 600 and 800 days.
(b) The probability is:
Z∞ Z∞ ¯∞
x x ¯
P ( X > 700) = f ( x) dx = 1
700 e − 700
dx = − e − 700 ¯
¯ = 0 + e−1 ≈ 0.3679
700 700 700
Exercises
A
For Exercises 1-3, find the center of gravity of the region bounded by the given curves over the given
interval.
1. y = x3 and y = 0 ; 0 ≤ x ≤ 1 2. y = − x + 1 and y = 0 ; 0 ≤ x ≤ 1 3. y = x2 and y = x3 ; 0 ≤ x ≤ 1
4. Find the center of gravity of the region inside the circle x2 + y2 = r2 and above the x-axis.
5. Find the center of gravity of the region inside the circle x2 + y2 = r2 in the first quadrant.
x2 y2
6. Find the center of gravity of the region inside the ellipse a2
+ b2 = 1 and above the x-axis.
x2
7. Find the center of gravity of the region between the circle x2 + y2 = 4 and the ellipse 4 + y2 = 1 above
the x-axis.
8. Would formula (8.18) for the center of gravity change if the mass of a region were proportional—but
not equal—to its area, say, by a constant positive proportion δ 6= 1? Explain.
9. If a spring requires 3 N of force to be compressed 5 cm, how much work would be performed in
stretching the spring 8 cm?
10. The gravitational force F ( x) exerted by the Earth on an object of mass m at a distance x from the
center of the Earth is
mgr2
F ( x) = − 2 e
x
where r e is the radius of the Earth. If the object is released from rest at a distance r o from the center
of the Earth, find the work performed by gravity in bringing the object to the Earth’s surface.
11. Recall that the ideal gas law states that PV = RT , where R is a constant, P is the pressure, V is
the volume, and T is the temperature. It can be shown that the work W done by an ideal gas in
expanding the volume from Va to Vb is
ZVb
W = P dV .
Va
Calculate W .
Z∞
12. Verify that f ( x) dx = 1 for the function f ( x) in formula (8.20) in Example 8.27 for all λ > 0.
−∞
13. Find P ( X < 300) in Example 8.27.
14. The distribution function F ( x) for a random variable X is defined as F ( x) = P ( X ≤ x) for all x. Show
that F ′ ( x) = f ( x), where f ( x) is the probability density function for X .
B
15. Formula (8.18) can be extended to regions over an infinite interval, provided the area is finite. Use
that fact to find the center of gravity of the region between y = e− x and the x-axis for 0 ≤ x < ∞.
16. The expected value (or mean) E [ X ] of a random variable X with probability density function f ( x) is
Z∞
E[ X ] = x f ( x) dx .
−∞
1
Show that E [ X ] = if X has the exponential distribution with parameter λ > 0.
λ
Note: The expected value can be thought of as the weighted average of all possible values of X , with
weights determined by probability. It is analogous to the idea of a center of gravity.
17. A random variable X is said to have a normal distribution if its probability density function f ( x) is
2
1 ( x−µ)
f ( x) = p e 2σ2 for all x
σ 2π
where σ > 0 and µ are constants. This is the famous “bell curve” in statistics.
Z∞
(a) Verify that f ( x) dx = 1. (Hint: Use Example 6.25 and a substitution.)
−∞
(b) Show that E [ X ] = µ.
(c) Use numerical integration to show that P (−1 < X < 1) ≈ 0.6827 when µ = 0 and σ = 1.
18. A random variable X has the beta distribution if its probability density function f ( x) is
(
1 a −1
B(a,b) x (1 − x)b−1 if 0 ≤ x ≤ 1
f ( x) =
0 elsewhere
a
for positive constants a and b, where B(a, b) is the Beta function. Show that E [ X ] = a+ b .
19. Show that any value between F ( x) and F ( x + dx) for the force over [ x, x + dx] gives same the formula
dW = F ( x) dx for the work performed over that interval. (Hint: Consider F ( x + α dx) for 0 ≤ α ≤ 1.)
20. A drop of water of mass M is released from rest at a height sufficient for the drop to evaporate
completely, losing mass m each second (i.e. at a constant rate). Ignoring air resistance, show that
g2 M2
the work performed by gravity on the drop up to complete evaporation is 6 m2
.
21. This exercise is related to Einstein’s famous law E = mc2 . The relativistic momentum p of a particle
of mass m moving at a speed v along a straight line (say, the x-axis) is
mv
p = q ,
2
1 − vc2
where c is the speed of light. The relativistic force on the particle along that line is
dp
F = ,
dt
which is the same formula as Newton’s Second Law of motion in classical mechanics. Assume that
the particle starts at rest at position x1 and ends at position x2 along the x-axis. The work done by
the force F on the particle is: Z x2 Z x2
dp
W = F dx = dx
x1 x1 dt
(a) Show that
dp m
= ³ ´ .
dv 2 3/2
1 − vc2
(b) Use the Chain Rule formula

dp dp dv dx
=
dt dv dx dt
to show that
dp
F dx = v dv .
dv
(c) Use parts (a) and (b) to show that
Zv Zv
dp mv
W = v dv = ³ ´3/2 dv .
0 dv 0 v2
1 − c2
(d) Use part (c) to show that

mc2
W = q − mc2 .
v2
1 − c2
(e) Define the relativistic kinetic energy K of the particle to be K = W , and define the total energy E
to be
mc2
E= q .
2
1 − vc2
So by part (d), K = E − mc2 . Show that
E 2 = p2 c2 + (mc2 )2 .
(Hint: Expand the right side of that equation.)

(f) What is E when the particle is at rest?
22. A median of a triangle is a line segment from a vertex to the midpoint of the opposite side, and the
three medians intersect at a common point. Show that this point is a triangle’s center of gravity.
CHAPTER 9
Infinite Sequences and Series
9.1 Sequences and Series

In the 5th century B.C. the ancient Greek philosopher Zeno of Elea devised several
paradoxes, the most famous of which—The Dichotomy—asserts that if space is in-
finitely divisible then motion is impossible. The argument goes like this: imagine a
line segment of finite length, say 1 m, with a person at one end as in Figure 9.1.1.
1 1 1
0 ... 8 4 2 1
Figure 9.1.1 Zeno’s motion paradox
Before traversing the entire distance the person would first need to travel one half
the distance. Before doing that, though, he would need to travel one fourth the dis-
tance, and before that one eighth the distance, and so on. There is thus no “first”
distance for him to traverse, meaning his motion cannot even begin!
Obviously motion is possible, or you would not be reading this. Does that mean
Zeno’s reasoning is flawed? More on that later. In the meantime, notice a few things
in Figure 9.1.1. First, the distance markers 21 , 14 , 81 , . . . form an infinite sequence of
numbers approaching 0. Second, the sum of the distances between successive markers
is an infinite series which should equal the total segment length 1:
1 1 1
+ + + ··· = 1
2 4 8
It will be shown shortly that the sum is indeed 1, which turns out to have no bearing
on Zeno’s paradox. First some definitions are needed.
286
Sequences and Series • Section 9.1 287
A sequence is an ordered list of objects, which in this book will always be real
numbers. Sequences can be finite or infinite: finite if there is a last number in the list,
infinite if every number in the list is followed by another number (i.e. a “successor”).
Sequences should not be confused with sets—order matters in a sequence but not in a
set, and numbers may repeat in a sequence but not in a set.
For example, the sequences 〈1, 2, 3〉 and 〈1, 3, 2〉 are different, since order matters.
However, the numbers in those sequences comprise the same set {1, 2, 3}.
The simplest example of an infinite sequence is N, the set of natural numbers:
0, 1, 2, 3, . . .. In fact, the numbers a 0 , a 1 , a 2 , . . . in any infinite sequence of real num-
bers can be written as the range of a function f mapping N into R:
f ( n) = a n (9.1)
Typically notation such as { a n }∞n=0 is used for representing an infinite sequence, or

simply { a n } when the initial value of the index n is understood (and n is always an in-
teger). The intuitive notion of the limit of an infinite sequence can be stated formally:
A real number L is the limit of an infinite sequence { a n }, written as
lim a n = L or simply an → L ,
n→∞
if for any given number ǫ > 0 there exists an integer N such that
¯ ¯
¯ a n − L ¯ < ǫ for all n > N . (9.2)
In this case the sequence { a n } is said to converge to L and is called a convergent

sequence. A sequence that is not convergent is divergent.
In other words, a sequence { a n } converges to L if the terms a n can be made arbi-

trarily close to L for n sufficiently large. In most cases the formal definition will not
be needed, since by formula (9.1) the same rules and formulas from Chapters 1 and 3
for the limit of a function f ( x) as x approaches ∞ apply to sequences (e.g. sums and
products of limits, L’Hôpital’s Rule). All you have to do is replace x by n.
Example 9.1
1
For integers n ≥ 1 define a n = 2n . Find lim a n if it exists.
n→∞
1
Solution: Since lim x = 0 then replacing x by n shows that
x→∞ 2
1
lim a n = lim = 0.
n→∞ n→∞ 2n
288 Chapter 9 • Infinite Sequences and Series §9.1
Example 9.2
2 n +1
For integers n ≥ 0 define a n = 3 n +2 . Is { a n } a convergent sequence? If so then find its limit.
Solution: By L’Hôpital’s Rule, treating an integer n ≥ 0 as a real-valued variable x,
2n + 1 2 2
lim a n = lim = lim = .
n→∞ n→∞ 3n + 2 n→∞ 3 3
Thus the sequence is convergent and its limit is 23 .
Example 9.3
en
For integers n ≥ 0 define a n = 3 n +2 . Is { a n } a convergent sequence? If so then find its limit.
Solution: By L’Hôpital’s Rule, treating an integer n ≥ 0 as a real-valued variable x,
en en
lim a n = lim = lim = ∞
n→∞ n→∞ 3n + 2 n→∞ 3
Thus the sequence is divergent.
Example 9.4
The famous Fibonacci sequence1 { F n } starts with the numbers 0 and 1, then each successive term is
the sum of the previous two terms: F0 = 0, F1 = 1,
F n = F n−1 + F n−2 for integers n ≥ 2 (9.3)
Equation (9.3) is a recurrence relation. The first ten Fibonacci numbers are 0, 1, 1, 2, 3, 5, 8, 13, 21, 34.
Clearly { F n } is a divergent sequence, since F n → ∞. For n ≥ 2 define a n = F n /F n−1 . The first few values
are:
F2 1 F3 2 F4 3 F5 5
a2 = = = 1 , a3 = = = 2 , a4 = = = 1.5 , a5 = = ≈ 1.667
F1 1 F2 1 F3 2 F4 3
Show that { a n } is convergent. In other words, in the Fibonacci sequence the ratios of each term to the
previous term converge to some number.
Solution: There are many ways to prove this, perhaps the simplest being to assume the sequence is
convergent and then find the limit (which would be impossible if the sequence were divergent). So
assume that a n → a for some real number a. Then divide both sides of formula (9.3) by F n−1 , so that
Fn F n −2 1
= 1 + ⇒ an = 1 + for integers n ≥ 2
F n −1 F n −1 a n −1
Now take the limit of both sides of the last equation as n → ∞:
µ ¶ p
1 1 1± 5
lim a n = lim 1 + ⇒ a = 1 + ⇒ a2 − a − 1 = 0 ⇒ a =
n→∞ n→∞ a n −1 a 2
p p
Since a must be positive then a = 1+2 5 . Thus, the sequence is convergent and converges to 1+2 5 ≈ 1.618.
Note: This number is the famous golden ratio, the subject of many claims regarding its appearance in
nature and supposed aesthetic appeal as a ratio of sides of a rectangle.2
1 Due to the Italian mathematician Leonardo Fibonacci (ca. 1170-1250).

2 For example, see H UNTLEY, H.E., The Divine Proportion, New York: Dover Publications, Inc., 1970.
An infinite series is the sum of an infinite sequence. If the infinite sequence is

n=0 then the series can be written as
{ a n }∞
∞
X
a n = a0 + a1 + a2 + · · · + a n + · · ·
n=0
P
or simply as a n when the initial value of the index n is understood. There is a
natural way to define the sum of such a series:
∞
X
The sum of an infinite series a n is
n=0
∞
X
a n = lim s n (9.4)
n→∞
n=0
n=0 is the sequence of partial sums of the series:

where { s n }∞
X
n
sn = a k = a0 + a1 + a2 + · · · + a n for n ≥ 0
k=0
If the partial sums { s n } converge to a real number s then the series is convergent,
and converges to s; if { s n } diverges than the series is divergent.
One important convergent series is a geometric progression:

a + ar + ar 2 + ar 3 + · · · + ar n + · · · (9.5)
with a 6= 0 and | r | < 1. Multiply the n-th partial sum s n by r :
rs n = r (a + ar + ar 2 + ar 3 + · · · + ar n−1 + ar n ) = ar + ar 2 + ar 3 + ar 4 + · · · + ar n + ar n+1
Now subtract rs n from s n :
s n − rs n = a + ar + ar 2 + ar 3 + · · · + ar n − (ar + ar 2 + ar 3 + ar 4 + · · · + ar n + ar n+1 )
s n − rs n = a − ar n+1 so that
a (1 − r n+1 ) a
lim s n = lim =
n→∞ n→∞ 1−r 1−r
n+1
since r → 0 as n → ∞ when | r | < 1. Thus:
∞
X
For a 6= 0 and | r | < 1 the geometric progression ar n is convergent:
n=0
∞
X a
ar n = a + ar + ar 2 + ar 3 + · · · + ar n + · · · = (9.6)
n=0 1−r
Example 9.5
X∞ 1
Show that = 1.
n =1 2
n
1
Solution: This is a geometric progression with a = 2 and r = 12 . So by formula (9.6) the sum is:
X∞ 1 X∞ µ 1 ¶ µ 1 ¶n a
1
2
= · = = = 1 X
n =1 2
n
n =0 2 2 1−r 1 − 21
Example 9.6
Write the repeating decimal 0.17 = 0.17171717 ... as a rational number.
17 1
Solution: This is a geometric progression with a = 0.17 = 100 and r = 0.01 = 100 :
0.171717 ... = 0.17 + 0.0017 + 0.000017 + · · · = 0.17 (0.01)0 + 0.17 (0.01)1 + 0.17 (0.01)2 + · · ·
X∞ a 0.17 0.17 17
= 0.17 (0.01)n = = = =
n =0 1−r 1 − 0.01 0.99 99
Back to Zeno’s motion paradox, by Example 9.5 the sum of the infinite number of
distances between successive markers in Figure 9.1.1 is 1, as expected. This fact is
often mistaken as proof that Zeno was wrong, yet it does not actually address Zeno’s
argument, since the person is attempting to begin motion at the tail end—the “infinite
end”—of the geometric progression, not at the beginning (i.e. at n = 0). Zeno’s point
remains that there is no “first step” at that tail end.
In fact, even if you were to reverse the person’s position to start at the other end, so
that he would first move a distance 21 , then a distance 14 , and so on, as in Figure 9.1.2,
then a new problem is introduced: the person keeps moving no matter how close he is
to the end point. There is now no “last step” and motion is thus still impossible.
1 1 1
8 4 2
···
1 1 1
0 ... 8 4 2 1
Figure 9.1.2 Zeno’s motion paradox in reverse
The convergence of the geometric progression has misled many people to argue that
Zeno is wrong, by claiming it shows an infinite number of movements can be completed
in a finite amount of time. But this line of argument fails on (at least) two counts. First,
Zeno never argued about time—it is irrelevant to his paradox. The more fundamental
flaw is that the introduction of time brings along the concept of speed, typically taken
to be constant (though it need not be). Speed is distance over time, but it is precisely
the distance part of that ratio that Zeno rejects—that distance can never be traveled.
In other words, it is circular—and hence faulty—reasoning to “prove” that motion is
possible by assuming it is possible.
The geometric progression also doesn’t help when considering only the distances and
ignoring time. Even assuming each of those distances could be traveled, the partial
sums approach 1 but never actually reach it. The limit of a sequence is defined in terms
of an inequality—you can get arbitrarily close to the limit, and that is all. The equality
in formula (9.6) is merely a shorthand way of saying that. It is an abstraction based
on properties of the real number system, not necessarily based on physical reality.
Zeno’s paradox is not purely mathematical—it is about space and hence is physical,
with a tinge of philosophy. Physicists have recognized this—and noted the flaw in
the purely mathematical line of attack—and devised new arguments against Zeno,
some more sophisticated than others. However, they all invariably end up in some
sort of circular reasoning. All this calls into question the original assumption: the
infinite divisibility of space. If space had some smallest unit that could not be divided
any further then there is no paradox—motion over a finite distance could always be
decomposed into a large but finite number of irreducible steps.3
Exercises
A
For Exercises 1-8, determine if the given sequence is convergent. If so then find its limit.
½ ¾∞ ½ ¾∞ ½ ¾∞
−n ∞ n2 n2 n3
1. { n e }n=0 2. 3. 4.
3n2 + 7n − 2 n=1 3n3 + 7n − 2 n=1 3n2 + 7n − 2 n=1
n n o∞ ½ ¾∞ n ³ ´ o
n! nπ ∞
5. 6. 7. sin 8. { (−1)n cos nπ }∞
n =0
ln n n=2 (n + 2) ! n=0 2 n =0
For Exercises 9-12 determine whether the given series is convergent. If so then find its sum.
X∞ µ ¶n X∞ X
∞ X∞
2 3 n + 5 n +1
9. 2 10. 7− n 11. 12. n
n =0
3 n =1 n =0
6n n =0
For Exercises 13-16 use a geometric progression to write the given repeating decimal as a rational
number.
13. 0.113 14. 0.9 15. 0.249 16. 0.017
B
p
5 −1
17. In Example 9.4 define b n = F n /F n+1 for n ≥ 0 and show that { b n }∞
n=0 converges to 2 .
3 As of this writing there is not yet a definitive answer as to whether space is continuous or discrete (quantized).
A “smallest unit” would have to be below the Planck level—around 10−33 cm, well below current measurement
capabilities. Some recent advances in the field of loop quantum gravity suggest the possibility of quantized space.
See C HAMSEDDINE , A.H., C ONNES, A. & M UKHANOV, V., “Geometry and the quantum: basics.” J. High Energy
Phys. 2014, 98 (2014). https://doi.org/10.1007/JHEP12(2014)098
18. Show that formula (4.1) for Newton’s method with f ( x) = x2 − 2 and x0 = 1 yields the sequence
n=0 with
{ x n }∞
1 1
x n = x n −1 + for n ≥ 1 ,
2 x n −1
p
then assuming { x n } is convergent show that it must converge to 2. (Hint: See Example 9.4.)
19. In this exercise you will prove a formula for the general number F n in the Fibonacci
p
sequence.
p
Denote the positive and negative solutions to the equation x2 − x − 1 = 0 by φ+ = 1+2 5 and φ− = 1−2 5 ,
respectively. Note that φ+ is the golden ratio mentioned in Example 9.4.
(a) Use the equation x2 − x − 1 = 0 to show that
n +1 n n −1 n +1 n n −1
φ+ = φ+ + φ+ and φ− = φ− + φ− .
(b) Use part (a) and induction to show that

n n
φ+ − φ−
Fn = p for n ≥ 0.
5
20. A ball is dropped from a height of 4 ft above the ground, and upon each bounce off the ground the
ball bounces straight up to a height equal to 65% of its previous height. Find the theoretical total
distance the ball could travel if it could bounce indefinitely. Why is this physically unrealistic?
21. A continued fraction is a type of infinite sum involving a fraction with a denominator that continues
indefinitely:
1
a = a0 +
1
a1 +
1
a2 +
a3 + · · ·
p
1+ 5
(a) Show that the golden ratio φ+ = 2 (a solution of x2 − x − 1 = 0) can be written as
1
φ+ = 1 + .
1
1+
1
1+
1 + ···
(Hint: Look for a recurrence relation in the fraction.)
(b) Show that
p 1
2 = 1 + .
1
2+
1
2+
2 + ···
p q p
1+ 5
p
22. Show that the golden ratio φ+ = 2 can be written as φ+ = 1 + 1 + 1 + · · ·.
23. Write a computer program to approximate the result in Exercise 22 using 100 terms. After how
many iterations do the approximate values start repeating?
24. Would the existence of infinitesimals as a measure of space resolve Zeno’s paradox? Explain.
Tests for Convergence • Section 9.2 293
9.2 Tests for Convergence

There are many ways to determine if a sequence converges—two are listed below. In
all cases changing or removing a finite number of terms in a sequence does not affect
its convergence or divergence:
1. Monotone Bounded Test: A sequence that is bounded and monotone—i.e.

either always increasing or always decreasing—is convergent.
2. Comparison Test: If { b n } diverges to ∞ and if { a n } is a sequence such that

a n ≥ b n for all n > N for some N , then { a n } also diverges to ∞.
Likewise, if { c n } diverges to −∞ and a n ≤ c n for all n > N for some N , then { a n }
also diverges to −∞.
The Comparison Test makes sense intuitively, since some- x

thing larger than a quantity going to infinity must also go a1 a2 a3 a4 · · · → M
to infinity. The Monotone Bounded Test can be understood Figure 9.2.1
by thinking of a bound on a sequence as a wall that the se-
quence can never pass, as in Figure 9.2.1. The increasing sequence { a n } in the figure
moves toward M but can never pass it. The sequence thus cannot diverge to ∞, and
it cannot fluctuate back and forth since it always increases. Thus it must converge
somewhere before or at M .4 Notice that the Monotone Bounded Test tells you only
that the sequence converges, not what it converges to.
Example 9.7
Show that the sequence { a n }∞
n=1 defined for n ≥ 1 by
1 · 3 · 5 · · · (2n − 1)
an =
2 · 4 · 6 · · · (2n)
is convergent.
Solution: Notice that { a n } is always decreasing, since
1 · 3 · 5 · · · (2n − 1) · (2n + 1) 2n + 1
a n +1 = = an · < a n · (1) = a n
2 · 4 · 6 · · · (2n) · (2n + 2) 2n + 2
for n ≥ 1. The sequence is also bounded, since 0 < a n and
1 3 5 2n − 1
an = · · ··· < 1 for n ≥ 1
2 4 6 2n
since each fraction in the above product is less than 1. Thus, by the Monotone Bounded Test the
sequence is convergent.
Note that for a decreasing sequence only the lower bound is needed for the Monotone Bounded Test, not
the upper bound. Similarly, for an increasing sequence only the upper bound matters.
4 For a proof see pp.48-49 in B UCK , R.C., Advanced Calculus, 2nd ed., New York: McGraw-Hill Book Co., 1965.
Some tests for convergence of a series are listed below:
P
1. n-th Term Test: If a n converges then lim a n = 0.
n→∞
P
2. Ratio Test: For a series a n of positive terms let
a n+1
R = lim .
n→∞ an
Then
(a) if R < 1 then the series converges,
(b) if R > 1 (including R = ∞) then the series diverges,
(c) if R = 1 then the test fails.
∞
X
3. Integral Test: For a series a n of positive terms let f ( x) be a decreasing
n=1
function on [1, ∞) such that f ( n) = a n for all integers n ≥ 1. Then
∞
X Z∞
a n and f ( x) dx
n=1 1
either both converge or both diverge.

X∞ 1
4. p-series Test: The series p
converges for p > 1, and diverges for p ≤ 1.
n=1 n
P
5. Comparison Test: If 0 ≤ a n ≤ b n for n > N for some N , and if b n is convergent
P
then a n is convergent. Similarly, if 0 ≤ b n ≤ a n for n > N for some N , and if
P P
b n is divergent then a n is divergent.
P P
6. Limit Comparison Test: For two series a n and b n of positive terms let
an
L = lim .
n→∞ bn
Then
P P
(a) if 0 < L < ∞ then a n and b n either both converge or both diverge,
P P
(b) if L = 0 and b n converges then a n converges,
P P
(c) if L = ∞ and b n diverges then a n diverges.
P P
7. Telescoping Series Test: Suppose a n = ( b n − b n+1 ) for some sequence { b n }.
P P
Then a n converges if and only if b n → L, in which case a n = b 1 − L.
Most of the above tests have fairly short proofs or at least intuitive explanations.
For example, the n-th Term Test follows from the definition of convergence of a series:
P
if a n converges to a number L then since each term a n = s n − s n−1 is the difference
of successive partial sums, taking the limit yields
lim a n = lim ( s n − s n−1 ) = L − L = 0

n→∞ n→∞
by definition of the convergence of a series. X

Since the n-th Term Test can never be used to prove convergence of a series, it is often
stated in the following logically equivalent manner:
P
n-th Term Test: If lim a n 6= 0 then a n diverges.
n→∞
Example 9.8
∞
X n 1 2 3
Show that = + + + · · · is divergent.
n =1 2n + 1 3 5 7
Solution: Since
n 1
lim = 6= 0
2n + 1
n→∞ 2
then by the n-th Term Test the series diverges.
The Ratio Test takes a bit more effort to prove.5 When the ratio R in the Ratio Test
is larger than 1 then that means the terms in the series do not approach 0, and thus
the series diverges by the n-th Term Test. When R = 1 the test fails, meaning it is
inconclusive—another test would need to be used. When the test shows convergence
it does not tell you what the series converges to, merely that it converges.
Example 9.9
X∞ n
Determine if is convergent.
n =1 2
n
Solution: For the series general term a n = 2n ,

n
n+1
a n +1 n +1 n+1 1
R = lim = lim 2 n = lim = < 1,
n→∞ a n n→∞ n→∞ 2 n 2
2n
so by the Ratio Test the series converges.
5 See pp.612-613 in T AYLOR , A.E. AND W.R. M ANN, Advanced Calculus, 2nd ed., New York: John Wiley & Sons,
Inc., 1972.
Figure 9.2.2 shows why the Integral Test works.

y y
y = f ( x) y = f ( x)
a1
a1
a2 a2
a3 a4 a5 a3 a4
x x
0 1 2 3 4 5 0 1 2 3 4 5
Z∞ ∞
X Z∞ ∞
X
(a) f ( x) dx > an (b) f ( x) dx < an
1 n =2 1 n =1
Figure 9.2.2 Integral Test

R∞
In Figure 9.2.2(a) the area 1 f ( x) dx is greater than the total area S of all the
rectangles under the curve. Since each rectangle has height a n and width 1, then
P
S= ∞ 2 a n . Thus, since removing the single term a 1 from the series does not affect the
convergence or divergence of the series, the series converges if the improper integral
converges, and converselyR the integral diverges if the series diverges. Similarly, in
∞ P
Figure 9.2.2(b) the area 1 f ( x) dx is less than the total area S = ∞ 1 a n of all the
rectangles, so the integral converges if the series converges, and the series diverges if
the integral diverges. Notice how in both graphs the rectangles are either all below
the curve or all protrude above the curve due to f ( x) being a decreasing function.
Example 9.10
X∞ 1
Show that the p-series p
converges for p > 1 and diverges for p = 1.
n =1 n
Solution: For p ≥ 1 let f ( x) = x1p on [1, ∞). Then f ( x) is decreasing, and f (n) = a n = n1p for all integers
n ≥ 1. For p > 1,
Z∞ Z∞ ¯∞ µ ¶
dx −1 ¯
¯ −1 1
f ( x) dx = = p −1 ¯
= 0 − p −1
= < ∞
1 1 x p ( p − 1) x 1 ( p − 1) (1) p − 1
so the integral converges. Thus, by the Integral Test, the series converges for p > 1. For p = 1,
Z∞ Z∞ ¯∞
dx ¯
f ( x) dx = = ln x ¯¯ = ∞
1 1 x 1
so the integral diverges. So by the Integral Test, the series diverges for p = 1. The harmonic series
X∞ 1 1 1 1
= 1 + + + + ···
n =1 n 2 3 4
1
thus diverges even though a n = n → 0 (which is hence not a sufficient condition for a series to converge).
Note that this example partly proves the p-series Test. The remaining case ( p < 1) is left as an exercise.
The divergence part of the Comparison Test is clear enough to understand, but for
the convergence part with 0 ≤ a n ≤ b n for all n larger than some N , ignore the (finite)
P
number of terms before a N and b N . Since b n converges then its partial sums must
P
be bounded. The partial sums for a n then must also be bounded, since 0 ≤ a n ≤ b n
P
for n > N . So since a n ≥ 0 means that the partial sums for a n are increasing, by
P P
the Monotone Bounded Test the partial sums for a n must converge, i.e. a n is
convergent.
Example 9.11
X∞ 1
Determine if n
is convergent.
n =1 n
Solution: Since nn ≥ n2 > 0 for n > 2, then
1 1
0 < n
≤ 2
n n
P 1
for n > 2. Thus, since ∞n=1 n2 converges (by the p-series Test with p = 2 > 1, as in Example 9.10), the
P∞ 1
series n=1 nn converges by the Comparison Test.
For the Limit Comparison Test with ab nn → L < ∞ and L > 0, by definition of the limit
of a sequence, ab nn can be made arbitrarily close to L. In particular there is an integer
N such that
L an 3L
< <
2 bn 2
for all n > N . Then
3L X X
0 < an < b n and b n converges ⇒ a n converges
2
by the Comparison test. Likewise,
L X X
0 < bn < an and b n diverges ⇒ a n diverges
2
by the Comparison Test again. The cases L = 0 and L = ∞ are handled similarly.
Example 9.12
X∞ n+3
n =1 n · 2
n
P 1
Solution: Since ∞ n=1 2n is convergent (as part of a geometric progression) and
(n + 3)/(n · 2n ) n+3
lim = lim = 1
n→∞ 1/2n n→∞ n
P n +3
then by the Limit Comparison Test ∞n=1 n · 2n is convergent..
P
A series a n is telescoping if a n = b n − b n+1 for some sequence { b n }. Assume the
P P
series a n and sequence { b n } both start at n = 1. Then the partial sum s n for a n is
s n = a 1 + a 2 + · · · + a n = ( b 1 − b 2 ) + ( b 2 − b 3 ) + · · · + ( b n − b n+1 ) = b 1 − b n+1
for n ≥ 1. Thus, since b 1 is a fixed number, limn→∞ s n exists if and only if limn→∞ b n+1
P
exists, i.e. a n converges if and only if { b n } converges. So if b n → L then s n → b − L,
P
i.e. a n converges to L, which proves the Telescoping Series Test. Note that the
number b 1 , as the first number in the sequence { b n }, could be replaced by whatever
the first number is, in case the index n starts at a number different from 1.
Example 9.13
∞
X 1
Determine if is convergent. If it converges then can you find its sum?
n =1 n (n + 1)
1
Solution: For the sequence { b n } with b n = n for n ≥ 1, each term in the series can be written as
1 1 1
= − = b n − b n +1
n (n + 1) n n+1
Thus, since { b n } converges to 0, by the Telescoping Series Test the series also converges, to b1 − 0 = 1.
Convergent series have the following properties (based on similar properties of limits):
P P
Let a n and b n be convergent series, and let c be a number. Then:
P P P P
(a) (a n ± b n ) is convergent, with (a n ± b n ) = a n ± b n
P P P
(b) ca n is convergent, with ca n = c · a n
Exercises
A
For Exercises 1-5 show that the given sequence { a n }∞
n=1 is convergent.
2 · 4 · 6 · · · (2n) 2n 2 · 4 · 6 · · · (2n) 1
1. a n = 2. a n = 1 − 3. a n = ·
3 · 5 · 7 · · · (2n + 1) n! 1 · 3 · 5 · · · (2n − 1) 2n + 2
µ ¶
1 2 · 4 · 6 · · · (2n) 2 1 3 3 5 5 7 2n − 1 2n + 1
4. a n = 5. a n = · · · · · · · · ·
n 1 · 3 · 5 · · · (2n − 1) 2 2 4 4 6 6 2n 2n
For Exercises 6-17 determine whether the given series is convergent.
X
∞ ³ nπ ´ X
∞
1 X
∞
1 X
∞
n
6. sin 7. 8. 9.
n =0 2 n =1 n (n + 2) n =1 2n n =1 (n + 1) 2n
X
∞
1 X
∞
n! X
∞
n X
∞
1
10. p 11. 12. 13.
n =2 n ln n n =1
(2n) ! n =1
en n =1 cosh2 n
X∞
n! X∞
1 X∞
1 X∞
ln (n + 1)
14. 15. p 16. 17.
n =1
2n n =1 n n =1
n (2n − 1) n =1 n2
For Exercises 18-21 determine whether the given series is convergent. If convergent then find its sum.
X
∞
1 X∞
1 X∞
2 X∞
1
18. 19. 20. 21. 2 −1
n =1 (2n + 1) (2n + 3) n =1 (2 n + 3) (2 n + 5) n =1 (3 n + 1) (3 n + 4) n =1 4 n
B
22. Continue Example 9.10 with a proof of the p-series Test for p < 1.
23. Show that { a n }∞
n=1 is convergent, where
1 1 1 1 1
an = + + + + ··· +
1! 2! 3! 4! n!
1
for n ≥ 1. (Hint: Use the Monotone Bounded test by using a bound on n! for n > 2.)
X
∞
1 1 1 1
24. Consider the series = 1 + + + + ··· .
n =1 2n − 1 3 5 7
(a) Show that the series is divergent.
(b) The textbook Applied Mathematics for Physical Chemistry (3rd ed.), J. Barrante, provides the
following argument that the above series converges: Since
1 1 1 1 1 1 1 1 1
1 + + + + ··· < 1 + + + + ··· < 1 + + + + ···
4 9 16 3 5 7 2 3 4
where the series on the left converges (by the p-series Test with p = 2) and the series on the right
diverges (by the p-series Test with p = 1), and since each term in the middle series is between
its corresponding terms in the left series and right series, then there must be a p-series for some
value 1 < p < 2 such that each term in the middle series is less than the corresponding term in
that p-series. That is,
1 1 1 1 1 1 1 1 1
1 + + + + ··· < 1 + + + + ··· < 1 + + p + p + ···
4 9 16 3 5 7 2p 3 4
for that value of p between 1 and 2. But p > 1 for that p-series on the right, so it converges,
which means that the middle series converges! Find and explain the flaw in this argument.
25. Wallis’ formula6 for π is given by the infinite product
π 2 2 4 4 6 6 2n 2n
= · · · · · ··· · · ··· .
2 1 3 3 5 5 7 2n − 1 2n + 1
Notice that this is the limit of the reciprocal of the sequence in Exercise 5. Write a computer program
to approximate the limit using 1 million iterations. How close is your approximation to π2 ?
6 Due to the English mathematician and theologian John Wallis (1616-1703). For a proof of the formula see pp.738-
739 in T AYLOR , A.E. AND W.R. M ANN, Advanced Calculus, 2nd ed., New York: John Wiley & Sons, Inc., 1972.
9.3 Alternating Series

In the last section the harmonic series
X∞ 1 1 1 1 1
= 1 + + + + + ···
n=1 n 2 3 4 5
was shown to diverge. If you were to alternate the signs of successive terms, as in
X∞ (−1) n−1 1 1 1 1
= 1 − + − + − ··· (9.7)
n=1 n 2 3 4 5
then it turns out that this new series—called an alternating series—converges, due
to the following test:
P
Alternating Series Test: If a n is an alternating series—i.e. the signs of the
terms a n alternate between positive and negative—such that the absolute values
of the terms are decreasing to 0, then the series converges.
The condition for the test means that | a n+1 | ≤ | a n | for all n and | a n | → 0 as n → ∞. To
see why the test works, consider the alternating series given above by formula (9.7),
n−1
with a n = −1n . The odd-numbered partial sums s 1 , s 3 , s 5 , . . ., can be written as
µ ¶ µ ¶ µ ¶
1 1 1 1 1 1
s1 = 1 , s3 = 1 − − , s5 = 1 − − − − , ...
2 3 2 3 4 5
| {z } | {z } | {z }
>0 >0 >0
while the even-numbered partial sums s 2 , s 4 , s 6 , . . ., can be written as

µ ¶ µ ¶ µ ¶
1 1 1 1 1 1 1 1 1
s2 = 1 − , s4 = 1 − + − , s6 = 1 − + − + − , ...
2 2 3 4 2 3 4 5 6
| {z } | {z } | {z }
>0 >0 >0
Thus the odd-numbered partial sums { s 2n−1 } are decreasing from s 1 = 1, and the even-
numbered ones { s 2n } are increasing from s 2 = 1 − 21 = 21 , with 21 < s n < 1 for all n, i.e. the
partial sums are bounded. So by the Monotone Bounded Test, both sequences { s 2n−1 }
1
and { s 2n } must converge. Since s 2n − s 2n−1 = a 2n = −
2 n for all n ≥ 1, then
−1
lim ( s 2n − s 2n−1 ) = lim = 0 ⇒ lim s 2n = lim s 2n+1
n→∞ n→∞ 2n n→∞ n→∞
Thus, the partial sums s n have a common limit, so the series converges. Notice that
the key to the convergence was having the terms decreasing in absolute value to zero.
Alternating Series • Section 9.3 301
Example 9.14
X∞ (−1) n
n=2 ln n
Solution: For the general term a n = (− 1) n

ln n , since ln ( n + 1) > ln n for n ≥ 2 and ln n → ∞ as n → ∞, then
| a n | decreases to 0 as n → ∞. Thus, by the Alternating Series Test the series converges.
P∞ (−1)n−1
The series n=1 converges by the Alternating Series Test, though the series
n
P∞ 1 P (−1)n−1
n=1 n diverges. This makes ∞n=1 n an example of a conditionally convergent
series:
P P P
A series a n is conditionally convergent if a n converges but | a n | diverges.
P P
If | a n | converges then a n is absolutely convergent.
P∞ (−1)n−1 P∞ 1
For example, n=1 n is not absolutely convergent, since n=1 n diverges.
Example 9.15
X∞ (−1) n−1
Is conditionally convergent or absolutely convergent?
n =1 n2
P 1 P∞ (−1)n−1
Solution: Since ∞n=1 n2 converges (by the p-series Test) then n=1 n2
is absolutely convergent.
It turns out that absolute convergence implies ordinary convergence:
P P
Absolute Convergence Test: If | a n | converges then a n converges.
The test is obvious if the terms a n are

© all positive,
ª so assume the series has both
© pos-
ª
itive terms (denoted by the sequence a pos ) and negative terms (denoted by a neg ).
Then the series can be decomposed into the difference of two series:
X X X
an = a pos − | a neg |
Since each of the sums on the right side of the equation is part of the convergent
P
series | a n |, then each sum itself converges (being part of a finite sum). Thus their
P P
difference, namely a n , is finite, i.e. a n converges.
The test can be stated in the following logically equivalent manner:
P P
Absolute Convergence Test: If a n diverges then | a n | diverges.
One unusual feature of a conditionally convergent series is that its terms can be re-
arranged to converge to any number, a result known as Riemann’s Rearrangement
Theorem. For example, the alternating harmonic series
1 1 1 1
1− + − + − ···
2 3 4 5
consists of one divergent series of positive terms subtracted from another series of
positive terms, namely:
µ ¶ µ ¶
1 1 1 1 1 1 1 1 1 1
1 − + − + − ··· = 1 + + + + ··· − + + + ···
2 3 4 5 3 5 7 2 4 6
The idea is that since the first series on the right diverges then that series has some
partial sum that could be made just larger than any positive number A . Likewise,
since the second series on the right that is being subtracted also diverges, it has some
partial sum that when subtracted from the first partial sum results in a number just
less than A . Continue like this indefinitely, first adding a partial sum to get a number
just bigger than A then subtracting another partial sum to get just less than A . Since
the terms in each series approach zero, the overall series can be made to converge to A !
It also turns out that an absolutely convergent series does not have this feature—any
rearrangement of terms results in the same sum.7
Exercises
A
For Exercises 1-5 determine whether the given alternating series is convergent. If convergent, then
determine if it is conditionally convergent or absolutely convergent.
X
∞
(−1)n−1 X
∞
(−1)n−1 n ! X
∞
(−1)n−1 X
∞
(−1)n X
∞
(−1)n−1
1. p 2. 3. 4. 5.
n =1 n n =1 2n n =1 2n − 1 n =2 n ln n n =1 n!
6. Rearrangements of a divergent alternating series can make it appear to converge to different num-
bers. For example, find different rearrangements of the terms in the divergent series
∞
X
(−1)n = 1 − 1 + 1 − 1 + · · ·
n =0
so that the series appears to converge to 0, 1, -1, and 2.

7. A guest arrives at the Aleph Null Hotel,8 which has an infinite number of rooms, numbered Room 0,
Room 1, Room 2, and so on. The hotel manager says all the rooms are taken, but he can still give the
guest his own room. How is that possible? Would it still be possible if an infinite number of guests
showed up, each wanting his own room? Explain your answers.
7 For formal proofs of these statements, see pp.442-444 in K LAMBAUER , G., Aspects of Calculus, New York:
8 The symbol ℵ is called aleph null and represents the cardinality of N, i.e. its size. ℵ is the smallest infinity.
0 0
Power Series • Section 9.4 303
9.4 Power Series

A power series is an infinite series whose terms involve constants a n and powers of
P
x − c, where x is a variable and c is a constant: a n ( x − c)n . In many cases c will be 0.
For example, the geometric progression
∞
X 1
r n = 1 + r + r2 + r3 + · · · =
n=0 1−r
converges when | r | < 1, i.e. for −1 < r < 1, as shown in Section 9.1. Replacing the
constant r by a variable x yields the power series
∞
X 1
x n = 1 + x + x2 + x3 + · · · = (9.8)
n=0 1−x
that converges to 1−1 x when −1 < x < 1. Note that the series diverges for | x | ≥ 1, by the
n-th Term Test.
P
In general a power series of the form f n ( x), where f n ( x) = a n ( x − c)n is a sequence
of functions, has an interval of convergence defined as the set of all x such that the
series converges. The interval can be any combination of open or closed, as well as
the extreme cases of a single point or all real numbers. On its interval of convergence
the power series is thus a function of x. The radius of convergence R of a power
series is defined as half the length of the interval of convergence. In the case where
the interval of convergence is all of R you would say R = ∞.
P
For example, for the above power series ∞ n=0 f n ( x), where f n ( x) = x for n ≥ 0, the
n
interval of convergence is −1 < x < 1, so the radius of convergence is R = 1. Notice that

∞
X
f ( x) = xn for −1 < x < 1
n=0
is thus a well-defined function on the interval (−1, 1), where it happens to equal 1−1 x .
This power series can be thought of as a polynomial of infinite degree.
P
To find the interval of convergence of a power series f n ( x), you typically would
use the Ratio Test on the absolute values of the terms (since the Ratio Test requires
positive terms): ¯ ¯
¯ f n+1 ( x) ¯
¯
r ( x) = lim ¯ ¯ (9.9)
n→∞ f ( x) ¯n
Note that the limit r ( x) in this case is a function of x. When taking the limit, though,
treat x as fixed. By the Ratio Test the power series will then converge for all x such
that r ( x) < 1, and diverge when r ( x) > 1. When r ( x) = 1 the test is inconclusive, so you
would have to check those cases individually to see if those values of x should be added
to the interval of convergence (along with the points where r ( x) < 1).
Example 9.16
X∞ xn
Find the interval of convergence of the power series .
n =0 n !
xn
Solution: For f n ( x) = n! ,
¯ ¯ ¯ n +1 ¯ ¯ ¯
¯ f n +1 ( x ) ¯ ¯x /(n + 1) ! ¯¯ ¯ 1 ¯
¯
r( x) = lim ¯ ¯ ¯
= lim ¯ ¯ ¯ = |x| · 0 = 0
n→∞ f n ( x) ¯ n→∞ x n /n ! ¯ = | x | · nlim
→∞ ¯ n + 1 ¯
for any fixed x. Thus, r( x) = 0 < 1 for all x, so the interval of convergence is all of R, i.e. −∞ < x < ∞.
Example 9.17
X∞ xn
Find the interval of convergence of the power series .
n =1 n
n
Solution: For f n ( x) = xn ,
¯ ¯ ¯ n +1 ¯ ¯ ¯ ¯ ¯
¯ f n +1 ( x ) ¯ ¯ /(n + 1) ¯¯ ¯ nx ¯ ¯ ¯
r( x) = lim ¯¯ ¯ = lim ¯ x = lim ¯ ¯ = | x | · lim ¯ n ¯ = | x | · 1 = | x |
n→∞ f n ( x) ¯ n→∞ ¯ x /n
n ¯ ¯
n→∞ n + 1 ¯ n→∞ n + 1 ¯
¯
for any fixed x. Thus, the series converges when r( x) = | x | < 1 and diverges when r( x) = | x | > 1.
P 1
The cases r( x) = | x | = 1 need to be checked individually. For x = 1 the series is ∞n=1 n , which diverges.
P∞ (−1)n−1
For x = −1 the series is n=1 n , which converges. Thus, the interval of convergence is −1 ≤ x < 1.
Example 9.18
∞
X
Find the interval of convergence of the power series n ! xn .
n =0
Solution: For f n ( x) = n ! x n ,
¯ ¯ ¯ ¯ (
¯ f n +1 ( x ) ¯ ¯ (n + 1) ! x n+1 ¯ 0 if x = 0,
r( x) = lim ¯ ¯ ¯ = lim ¯¯ ¯ = | x | · lim | n + 1 | =
n→∞ f n ( x) ¯ n →∞ n! x n ¯ n →∞ ∞ if x 6= 0.
Thus, r( x) = ∞ > 1 for all x 6= 0, so the series diverges for x 6= 0. So since r( x) = 0 < 1 only for x = 0, the
interval of convergence is the single point x = 0.
It turns out that power series can be both differentiated and integrated term by term:9
∞
X
For a power series f ( x) = a n ( x − c)n that converges for | x − c | < R , both
n=0
∞
X Z ∞
X
n−1 an
′
f ( x) = n a n ( x − c) and f ( x) dx = C + ( x − c)n+1 (9.10)
n=1 n=0 n+1
converge for | x − c | < R .
9 The proofs require uniform convergence, a stronger condition than ordinary convergence. See pp.129-134 in
B ROMWICH , T.J., An Introduction to the Theory of Infinite Series, 2nd ed., London: Macmillan & Co. Ltd., 1955.
R Notice that the above statement says nothing about the convergence of f ( x) or
′
f ( x) dx at the endpoints of the interval | x − c | < R . In each case convergence at the

endpoints can be checked individually.
Example 9.19
∞
X
Write the power series form of the derivative of f ( x) = x n and find its interval of convergence. Can
n =0
f ′ ( x) be written in a non-series form?.
Solution: Differentiate f ( x) term by term:
Ã !
d X ∞ d X∞
′
f ( x) = x n
= (1 + x + x2 + x3 + · · · ) = 0 + 1 + 2 x + 3 x2 + · · · = nx n−1
dx n=0 dx n =1
Since f ( x) converges for −1 < x < 1 then so does f ′ ( x). Checking the endpoints, at x = 1 and x = −1 the
P P∞ n −1
series for f ′ ( x) are ∞n=1 n and n=1 (−1) n, respectively, both of which diverge by the n-th Term Test.
Thus, the interval of convergence for f ′ ( x) is−1 < x < 1.
1 1
Since f ( x) = 1− x for −1 < x < 1 then f ′ ( x) = (1− x)2
for −1 < x < 1. Thus,
∞
X 1
nx n−1 = 1 + 2 x + 3 x2 + · · · = for −1 < x < 1.
n =1 (1 − x)2
Bessel Functions
Many physical applications—especially those involving oscillations and mechanical

vibrations—involve solving differential equations of the form
d2 y 1 dy
2
+ + y = 0 , (9.11)
dx x dx
known as Bessel’s equation. This equation has a solution J0 ( x), known as Bessel’s
function of order zero,10 defined in terms of a power series:
X
∞
x2 n x2 x4 x6
J0 ( x) = (−1)n = 1 − + − + ··· (9.12)
n=0 ( n !)2 · 22n 22 22 · 42 22 · 42 · 62
The Ratio Test shows that J0 ( x) converges for all x, since for any fixed x,
¯ 2 n+2
¯
¯ x ¯
¯ (−1)n+1 ¯ ¯ ¯
¯ ¯ 1
¯ (( n + 1) !)2 · 22n+2 ¯ 2
¯
¯
¯
¯ = 0 < 1.
r ( x) = lim ¯ ¯ = x · lim
n→∞ ¯ x 2n ¯ n→∞ 4 ( n + 1)2 ¯
¯
¯ (−1)n ¯
¯ ¯
( n !)2 · 22n
10 Due to the German astronomer Friedrich Wilhelm Bessel (1784-1846), from a study of elliptic planetary motion.
The general Bessel equation of order m,

µ ¶
d2 y 1 dy m2
+ + 1− 2 y = 0 (9.13)
dx2 x dx x
for m = 0, 1, 2, . . ., has a solution Jm ( x), called Bessel’s function of order m:
X
∞
1 ³ x ´2n+m
Jm ( x ) = (−1)n (9.14)
n=0
n ! · ( n + m) ! 2
For example, the Bessel function J1 ( x) of order 1 is

X
∞
1 ³ x ´2n+1 x x3 x5 x7
n
J1 ( x) = (−1) = − 2 + 2 2 − 2 2 2 + ···
n=0
n ! · ( n + 1) ! 2 2 2 ·4 2 ·4 ·6 2 ·4 ·6 ·8
Term by term differentiation shows that J0′ ( x) = − J1 ( x):

µ ¶
d d x2 x4 x6 x8
( J0 ( x)) = 1 − 2 + 2 2 − 2 2 2 + 2 2 2 2 − ···
dx dx 2 2 ·4 2 ·4 ·6 2 ·4 ·6 ·8
3 5 7
x x x x
= − + 2 − 2 2 + 2 2 2 − · · · = − J1 ( x)
2 2 ·4 2 ·4 ·6 2 ·4 ·6 ·8
Graphs of J0 ( x) and J1 ( x) are shown in Figure 9.4.1 below. As you can see, J0 ( x) and
J1 ( x) behave as sort of “poor man’s” cosine and sine functions, respectively.
J (x)
0
J (x)
1
0.5
0
y
-0.5
-1
0 5 10 15 20 25
x
Figure 9.4.1 Bessel functions J0 ( x) and J1 ( x)

Exercises
A
For Exercises 1-8 find the interval of convergence of the given power series.
X
∞
nx n X
∞
nx n X
∞ X
∞
( x + 4)n
1. 2. 3. n2 ( x − 2)n 4.
n =1 (n + 1)2 n =1
2n n =1 n =0
2n
X∞
( x + 1)n X∞ X∞ X∞
nx n
5. 6. nn x n 7. (−1)n x n 8.
n =1
nn n =1 n =0 n =1
n+1
P∞
9. Note that power series of the form n=0 a n x n have an issue at x = 0 when n = 0: 00 is an indeter-
minate form—it can equal anything (or nothing). What value has it implicitly been assigned so far?
P
What would be the technically correct way to write the series ∞n=0 a n x so that this issue goes away?
n
10. Show that

∞
X x
n xn = for −1 < x < 1.
n =1 (1 − x)2
11. Write the following infinite series as a rational number:
1 2 3 4
+ + + + ···
10 100 1000 10000
B
12. Differentiating term by term, verify that the Bessel function J0 ( x) satisfies Bessel’s equation (see
equation (9.11)).
13. Show that for all m ≥ 1 the Bessel functions Jm ( x) converge for all x.
14. For all m ≥ 1 verify that the Bessel functions Jm ( x) satisfy the general Bessel equation of order m
(see equation(9.13)).
15. For the Bessel functions J0 ( x) and J1 ( x) show that:
d
(a) ( x J1 ( x)) = x J0 ( x)
dx
R
(b) J0 ( x) J1 ( x) dx = − 21 J02 ( x)
R R
(c) x J0 ( x) J1 ( x) dx = − 21 x J02 ( x) + 21 J02 ( x) dx
(d) For all integers n ≥ 2,
Z Z
n −1 2
n n
x J0 ( x) dx = x J1 ( x) + (n − 1) x J0 ( x) − (n − 1) x n−2 J0 ( x) dx .
(Hint: Use part (a) and integration by parts twice.)

16. For all integers m ≥ 2 show that the Bessel functions Jm ( x) satisfy the relations:
(a) m Jm ( x) + x Jm
′
( x) = x Jm−1 ( x)
(b) Jm−1 ( x) − Jm+1 ( x) = 2 Jm
′
( x)
1
17. Use long division to obtain the first three terms of , then integrate term by term to show that
xJ02 ( x)
Z
dx x2 3 x4
J0 ( x) = J0 ( x) · ln x + − + ··· .
xJ02 ( x) 4 128
This function is a Bessel function of the second kind and is another solution of Bessel’s equation.
9.5 Taylor’s Series

In the previous section a few functions, e.g. f ( x) = 1−1 x , turned out to be the sum of a
power series. This section will discuss a general method for representing a function as
a power series, called a Taylor’s series.11 Suppose that a function f ( x) can be written
as
∞
X
f ( x) = a n ( x − c)n
n=0
either for all x or for | x − c | < R , for some R > 0. Then f ( c) = a 0 , and differentiating
term by term yields
∞
X
f ′ ( x) = n a n ( x − c)n−1 ⇒ f ′ ( c) = 1 · a 1
n=1
X∞
f ′′ ( x) = n ( n − 1) a n ( x − c)n−2 ⇒ f ′′ ( c) = 2 · 1 · a 2
n=2
X∞
f ′′′ ( x) = n ( n − 1) ( n − 2) a n ( x − c)n−3 ⇒ f ′′′ ( c) = 3 · 2 · 1 · a 3
n=3
··· ···
∞
X
f ( k ) ( x) = n ( n − 1) ( n − 2) · · · ( n − k + 1) a n ( x − c)n−k ⇒ f ( k) ( c ) = k ! a k
n= k
so that in general (since f (0) ( x) = f ( x) and 0! = 1):
f ( n) ( c )
an = for n ≥ 0 (9.15)
n!
These { a n } are the Taylor’s series coefficients of f ( x) at x = c. The full power series
representation of f ( x) can now be stated:
Taylor’s formula: If f ( x) has a power series representation in powers of x − c,

where x = c is inside the interval of convergence, then that representation is unique
in that interval and is given by
X
∞
f ( n) ( c )
f ( x) = ( x − c)n (9.16)
n=0
n!
for all x in the interval. This is the Taylor’s series for f ( x) about x = c.
11 Named after English mathematician Brook Taylor (1685-1731), though such series were known to others (e.g.
James Gregory, Johann Bernoulli) before Taylor.

Taylor’s Series • Section 9.5 309
Example 9.20
Find the Taylor’s series for f ( x) = e x about x = 0.
d
Solution: Since dx ( e x ) = e x , then for all n ≥ 0,
f ( n) ( x) = e x ⇒ f ( n) (0) = e0 = 1 .
Thus, by Taylor’s formula with c = 0:12

X
∞
f ( n) (0) X
∞
xn
ex = xn =
n =0
n! n =0
n!
2 3
x x x4 x5
= 1 + x + + + + + ···
2! 3! 4! 5!
For what x is this Taylor’s series valid? Recall that Example 9.16 showed the interval of convergence is
all of R. Thus the above Taylor’s series holds for all x.
Before continuing, you might be wondering why you should even bother with finding
the Taylor’s series—after all, in the above example why replace a simple function like
e x by a far more complicated expression? One reason is that it often helps simplify
some computations, especially in integrals. The idea is to use only a few terms in the
series, i.e. a polynomial, as an approximation, since polynomials are generally easier
to work with. Perhaps surprisingly, in many practical applications no more than two
terms are needed, and often only one.
For example, using only the first two terms of the Taylor’s series for e x in Example
9.20, e x ≈ 1 + x is a good approximation when x is close to 0 (i.e. | x | ≪ 1). Using more
terms does not necessarily help—for | x | ≪ 1 and n > 1, xn will be effectively 0. So the
added complexity would not make the approximation significantly better.
Example 9.21
The energy density E of electromagnetic radiation at wavelength λ from a black-body at temperature
T degrees Kelvin is given by Planck’s Law of black-body radiation,
8πhc
E (λ) =
λ5 ( e hc/λ kT − 1)
where h is Planck’s constant, c is the speed of light, and k is Boltzmann’s constant. Show that for λ ≫ 1:
8πkT
E (λ) ≈
λ4
Solution: Since e x ≈ 1 + x for | x | ≪ 1 by the Taylor series for e x , let x = hc/λkT . Then x ≪ 1 and so
8πhc 8πhc 8πhc 8πkT
E (λ) = ≈ ³³ ´ ´ ≈ ≈
λ5 ( e hc/λ kT − 1) λ5 1+ hc
−1 λ5 λhc λ4
λ kT kT
12 The special case of c = 0 in Taylor’s formula yields what is sometimes called the Maclaurin’s series for f ( x),
though that terminology is typically not used in fields of study outside mathematics.
Example 9.22
Find the Taylor’s series for f ( x) = sin x about x = 0.
Solution: The derivatives of f ( x) = sin x repeat every four derivatives:
f ( x) = sin x , f ′ ( x) = cos x , f ′′ ( x) = − sin x , f ′′′ ( x) = − cos x , f (4) ( x) = sin x
So at x = 0:
f (0) = 0 , f ′ ( x) = 1 , f ′′ ( x) = 0 , f ′′′ ( x) = −1 , f (4) ( x) = 0
So for n ≥ 0,

0
 if n is even,
f ( n) (0) = 1 if n = 1, 5, 9,... ,


−1 if n = 3, 7, 11,... .
Thus, by Taylor’s formula with c = 0
X
∞
f ( n) (0)
sin x = xn
n =0
n!
x3 x5 x7 x9
= x − + − + − ···
3! 5! 7! 9!
X
∞
x 2 n +1
= (−1)n
n =0 (2n + 1)!
By the Ratio Test this series converges for all x, since for any fixed x,
¯ ¯
2 n +3 ¯
¯
¯ (−1)n+1 x ¯ ¯ ¯
¯ ¯ ¯ 1 ¯
¯ (2n + 3)! ¯ 2 ¯ ¯
r( x) = lim ¯ 2 n +1
¯ = x · lim ¯ (2n + 3) (2n + 2) ¯ = 0 < 1 .
n→∞ ¯ x ¯ n →∞
¯ (−1)n ¯
¯ ¯
(2n + 1)!
Example 9.23
Find the Taylor’s series for f ( x) = cos x about x = 0.
Solution: The Taylor’s series can be found using the same procedure as in Example 9.22, but it is
simpler to just differentiate the Taylor’s series for sin x term by term for all x:
µ ¶
d d x3 x5 x7 x9
cos x = (sin x) = x − + − + − ···
dx dx 3! 5! 7! 9!
x2 x4 x6 x8
= 1 − + − + − ···
2! 4! 6! 8!
X
∞
x2 n
= (−1)n
n =0 (2n)!
Since the Taylor’s series for sin x converges for all x then so does its derivative. Thus the Taylor’s series
for cos x converges for all x.
Notice that the Taylor’s series for cos x has only even powers of x, while the series for sin x has only odd
powers of x. This makes sense since cos x and sin x are even and odd functions, respectively.
Example 9.24
The function ln x is not defined at x = 0 and hence has no Taylor’s series about x = 0. Instead, find the
Taylor’s series for f ( x) = ln (1 + x) about x = 0.
Solution: Take successive derivatives:
1 1 1·2 1·2·3
f ( x) = ln (1 + x) , f ′ ( x) = , f ′′ ( x) = − , f ′′′ ( x) = , f (4) ( x) = −
1+x (1 + x)2 (1 + x)3 (1 + x)4
So f (0) = 0 and for n ≥ 1:
(n − 1)!
f ( n) ( x) = (−1)n−1 ⇒ f ( n) (0) = (−1)n−1 (n − 1)!
(1 + x)n
Thus, by Taylor’s formula,
X
∞
f ( n) (0) X
∞
(n − 1)! x n
ln (1 + x) = xn = (−1)n−1
n =0 n! n =1 n!
X
∞
xn
= (−1)n−1
n =1
n
x2 x3 x4 x5
= x − + − + − ···
2 3 4 5
Use the Ratio Test to find the interval of convergence:
¯ ¯
¯ x n +1 ¯
¯ (−1)n ¯ ¯ ¯
¯ n + 1 ¯ ¯ n ¯
r( x) = lim ¯¯ ¯ ¯
= | x | · lim ¯ ¯ = |x| · 1 = |x|
n→∞ ¯
1
x n ¯¯ n→∞ n + 1 ¯
¯ (−1) n −
¯
n
So the series converges when | x | < 1. Check the cases r( x) = | x | = 1 individually. For x = 1 the series
P n−1 P
is the alternating harmonic series n=1 (−1)n , which converges. For x = −1 the series is − n=1 n1 , the
negative of the harmonic series, which diverges. Thus, the series converges for −1 < x ≤ 1.
Example 9.25
2
Find the Taylor’s series for f ( x) = e x about x = 0.
Solution: The Taylor’s series can be found using the same procedure as in Example 9.20, but it is simpler
to just replace each occurrence of x in the Taylor’s series for e x by x2 (since the series for e x converges
for all x). In other words, make the substitution u = x2 in the Taylor’s series for e u about u = 0:
X
∞
un
eu =
n =0
n!
2
X
∞
( x2 ) n X
∞
x2 n
ex = =
n =0
n! n =0
n!
x4 x6 x8 x10
= 1 + x2 + + + + + ···
2! 3! 4! 5!
Define the n-th degree Taylor polynomial P n ( x) for a function f ( x) about x = c by
X
n
f ( k) ( c )
P n ( x) = ( x − c)k
k=0
k!
f ′ ( c) f ′′ ( c) f ( n) ( c )
= f ( c) + ( x − c) + ( x − c )2 + · · · + ( x − c)n
1! 2! n!
for x in the interval of convergence for the full Taylor’s series. In other words, P n ( x) is
the n-th partial sum of the Taylor’s series. Since some of the coefficients could be zero,
P n ( x) is a polynomial of degree at most n. Thus, P n ( x) = O( xn ). For that reason P n ( x)
is sometimes called the O( x n ) approximation to f ( x).
Figure 9.5.1 shows a comparison of sin x with a few of its approximations:
O(x7) approx.
1 O(x11) approx.
O(x15) approx.
sin(x)
0.5
0
y
−0.5
−1
−8 −6 −4 −2 0 2 4 6 8
x
Figure 9.5.1 sin x and Taylor’s series approximations
As you can see, the Taylor polynomials of degree 7, 11 and 15 are all good approxi-
mations over the interval [−2, 2], with the O( x15 ) approximation still being fairly good
over [−6, 6]. Clearly those approximations all become poor quite quickly for | x | > 6;
they approach ±∞, unlike sin x.
The following theorem shows how to measure the accuracy of the approximations:
Remainder Theorem:13 If P n ( x) is the n-th degree Taylor polynomial about x = c

for a function f ( x) in some interval containing x = c, then for all x in that interval,
f ( x) = P n ( x) + R n ( x) , (9.17)
where
f (n+1) ( c + θ ( x − c))
R n ( x) = ( x − c)n+1 (9.18)
( n + 1)!
for some number θ between 0 and 1. Alternatively,
Z
1 x
R n ( x) = ( x − t)n f (n+1) ( t) dt . (9.19)
n! c
Since the number θ is unknown in equation (9.18), usually only an upper bound on
the remainder R n ( x) can be found by that formula. For practical purposes formula
(9.19) for R n ( x) might be easier to use (via numerical integration).
A common misconception is that hand-held calculators use Taylor’s series to com-
pute values of functions like sin x, cos x, e x , etc. However, that is typically well be-
yond their capability, especially for large values of x—far too many terms would be
required. Instead, many calculators use an algorithm called CORDIC14 (Coordinate
Rotation Digital Computer), and—perhaps surprisingly—lookup tables. CORDIC uses
the computationally inexpensive operation of bit-shifting to translate large input val-
ues into a smaller range, then uses tables stored in memory for values in that range,
along with interpolation for numbers between those in the tables.15
Exercises
A
For Exercises 1-9 write out the first three nonzero terms in the Taylor’s series for the given function
f ( x) about the given value c. You may use any method you like.
1. f ( x) = sin x ; c = π/2 2. f ( x) = sinh x ; c = 0 3. f ( x) = cosh x ; c = 0
4. f ( x) = tan x ; c = 0 5. f ( x) = tanh x ; c = 0 6. f ( x) = sec x ; c = 0

1 1 p
7. f ( x) = ; c=0 8. f ( x) = ; c=1 9. f ( x) = 1 + x2 ; c = 0
1 + x2 1 + x2
10. Use Example 9.19 from Section 9.4 to write out the first three nonzero terms in the Taylor’s series
for f ( x) = (1−1x)3 about x = 0.
13 For a proof see pp.171-172 in K LAMBAUER , G., Aspects of Calculus, New York: Springer-Verlag, 1986.
14 See Ch.7 in S CHMID, H., Decimal Computation, New York: John Wiley & Sons, Inc., 1974.
15 To see how inaccurate calculators can be, compute tan(355/226) in radian mode on a calculator. The true value
to 3 decimal places is −7497258.185, but few calculators produce an answer close to that.
p
11. Use the derivative of the Taylor’s series for 1 + x2 from Exercise 9 to write out the first three
2
nonzero terms in the Taylor’s series for f ( x) = px about x = 0.
1+ x2
For Exercises 12-15R replace the function f ( x) by its Taylor’s series about x = 0 to evaluate the given
indefinite integral f ( x) dx (up to the first three nonzero terms in the series).
Z Z Z Zp
sin x 2
12. dx 13. cos ( x2 ) dx 14. e− x dx 15. 1 + x6 dx
x
d
16. Use dx (tan−1 x) = 1+1x2 along with Exercise 7 to find the Taylor’s series for f ( x) = tan−1 x about x = 0,
along with its interval of convergence.
17. Use Exercise 16 to show that
µ ¶
1 1 1
π = 4 1 − + − − ··· .
3 5 7
2 /2
18. Use the first three nonzero terms in the Taylor’s series about x = 0 for e− x to evaluate the definite
integral
Z1
1 2
p e− x /2 dx .
−1 2π
Note: The actual value (rounded to 4 decimal places) is 0.6826.
19. Recall that the surface area S of the solid obtained by revolving the curve y = x2 around the x-axis
between x = 0 and x = 2 is given by the integral
Z2 p
S = 2π x2 1 + 4 x2 dx .
0
The exact value of the integral rounded
p to 3 decimal places is S = 53.226. Use the first two nonzero
terms in the Taylor’s series for 1 + 4 x2 about x = 0 to approximate the integral. How close is this
approximation to the actual value? Does the approximation become better if you use the first three
nonzero terms in the Taylor’s series? Justify your answer.
20. The tangential component of a space shuttle’s velocity during reentry is approximately
µ µ ¶¶
g −1 v0
v( t) = v c tanh t + tanh
vc vc
³ ´
where v0 is the velocity at time t = 0 and v c is the terminal velocity. If tanh−1 vv0c = 21 then show
that v( t) ≈ gt + 21 v c .
³ ´
2π h
21. The velocity of a water wave of length L in water of depth h satisfies the equation v2 =
gL
2π tanh L .
p
Show that v ≈ gh.
22. A disk of radius a has a charge of constant density σ. A point P lies at a distance r directly above
p 2
the disk. The electrical potential V at point P is given by V = 2πσ( r2 + a2 − r). Show that V ≈ πar σ
for large r (i.e. r ≫ 1).
23. The fifth-degree Padé approximation uses rational functions to approximate tanh x:
x5 + 105 x3 + 945 x
tanh x ≈
15 x4 + 420 x2 + 945
Compare the values of the Padé approximation and the fifth-degree Taylor’s series approximation
from Exercise 5, evaluated at x = 1. Which is better? The actual value of tanh(1) is 0.7615941559558.
How do the two approximations compare at x = 2?
APPENDIX A
Answers and Hints to Selected Exercises
Chapter 1 17. sin(cos x) sin x 21. Hint for part(b):

Use part(a) and the Chain Rule to find S p ,
Section 1.1 (p. 6) then recall how to convert from radians
per second to revolutions per minute.
1. 2 t 2. 19.6 t 3. −32 t + 2 4. 3 t2
Section 1.2 (p. 14) Section 1.6 (p. 35)

1. 0 3. 2 x + 2 5. − ( x+11)2 7. − x23 2
1. 6 x + 2 3. −9 cos3 x 5. x3
9. p1 11. p 2 x+3
2 x+1 2 x2 +3 x+4
Section 1.3 (p. 20) Chapter 2

5. Hint: Use the sine double-angle for-
mula. 7. Hint: Use Exercise 5 and the Section 2.1 (p. 40)
sine addition formula. ¡ ¢′
1. f −1 ( x) = x, f −1 ( x) = 1
p ¡ ¢′ 1
Section 1.4 (p. 26) 3. f −1 ( x) = x, f −1 ( x) = 2p x
1 1
¡ 1
¢ ′ 1
5. x cos x + sin x 5. f ( x) = x , f¡ ( x) = − x 2
− −
1. 2 x − 1 3. 4 x5 + x97 ¢′
1
7. x cos x−sin x 9. 2−2 t
2
11. ad−bc
−1
7. f ( x) = px , f − 1
( x) = − 21 x−3/2
x2 (1+ t2 )2 ( cx+d )2
13. 2πr

1 p 1− x
1. −20(1 − 5 x)3 3. − p 5. 2 x( x+1)2 1. 6 sec2 3 x tan 3 x 3. −3 csc2 3 x 5. 3
1−2 x 9+ x 2
3 6psin−1 3 x
7. − 8(1 − t)
(1+ t)5
9. 2 sin x cos x 11. 15 sec (5 x) 7. − 1+39 x2
2 9. 1
1+ x 2
11.
1−9 x2
1
13. 2 x sec( x2 ) tan( x2 ) 15. β(1 − β2 )−3/2 13. 1+ x 2
15. cot −1
x− x
1+ x 2
315
316 Appendix A: Answers and Hints to Selected Exercises

2ex −3 x2 y+4 y2 +2 x 2( x− y+1)−3( x+ y)2
1. 2 e2 x
x
3. − e−x − e x 5. (1− e x )2
7. e e e x 1. 3. 2( x− y+1)+3( x+ y)2
x3 −8 x y−1
2
1 6 x(ln(tan x2 )) sec2 x2 2 x(1−( x2 − y2 )) y
9. 11. 2
5. y(2( y2 −x2 )−1) 7. − x
x tan x
x2 −2 x− y+3 x2 y2 e sin(x y) + x3 y3 e sin(x y) cos( x y)
15. x ( x¡+ 2 x ln x) ¢ 9. − x4 y2 esin(x y) cos( x y)+2 x3 yesin(x y) −3 y2 −x
17. xsin x cos x ln x + sinx x 19. 15.5 hours x 2 + y2
21. 12 hours 13. − y3
Section 3.5 (p. 83)

Section 2.4 (p. 55)
ln3 (3 x −3−x )
1. 80π ft/s 3. 2.4 ft/s 5. 10 ft/s
3. (ln 2)2 22 2 x
x
1. 2 7. −76π cm3 /min 9. 45.14 mph
2x cos(log2 π x)
5. (ln 2)( x2 +1)
7. x ln 2 9. 3 x2 11. 155.8 ft/min
Section 3.6 (p. 88)

Chapter 3
1. (2 x − 2) dx 2. 4 x sin( x2 ) cos( x2 ) dx
Section 3.1 (p. 61) 11. Hint: Mimic Example 3.35.
1. y = 4 x − 3 3. y = −6 x + 10 5. y = 4 x
7. y = x + 3 9. y = 240 x + 176 11. y = 2 x
Chapter 4
31
13. y = 3 x + 27 , y = 3 x + 1 15. 75.96◦
Section 4.1 (p. 98)
17. 0 ◦
19. 116.6◦ 21. 5.71◦
23. y = − 41 x + 11
2 1. (1, 1) 3. 125, 000 sq yd 5. U = V2
1 81 1 1159 q
25. y = − 4 x − 4 , y = − 4 x + 108
7. R = r 9. 2ab 13. Q = 2I +DP W
q
2 2
15. r = r 0 + x0 17. 380.62 minutes
Section 3.2 (p. 72) p
19. 12π 3 21. 12.8 ft 22. Hint: Place
1. 37 3. 0 5. −1 7. 0 9. 2 11. 0 the right angle of the triangle q at the origin
1
14. 2 15. 0 17. 0 in the x y-plane. 25. x = RN
r
27. x = pa 33. (a2/3 + b ) 2/3 3/2
2
Section 3.3 (p. 78) 34. Hint: You can leave your answer in
terms of R and an angle satisfying a cer-
1. continuous 3. discontinuous tain equation.
5. discontinuous 7. discontinuous
9. continuous 11. continuous
Section 4.2 (p. 108)
13. continuous 15. discontinuous
17. continuous 19. 1 21. e−1 2. local maximum at x = 0, local mini-
25. Hint: Use the Intermediate Value mum at x = 2, inflection pt at x = 1, in-
Theorem. creasing for x < 0 and x > 2, decreasing
Appendix A: Answers and Hints to Selected Exercises 317
for 0 < x < 2, concave up for x > 1, con- Section 5.3 (p. 145)
cave down for x < 1 3. local maximum at
x = 1, inflection pt at x = 2, increasing for 1. 31 3. 1
4
5. 1
2
7. 1 9. 2 e − 2 e−1
x < 1, decreasing for x > 1, concave up for 11. 16
3
x > 2, concave down for x < 2, horizontal
asymptote: y = 0 5. local maximum at
x = 0, inflection pts at x = ± p1 , increasing
3
for x < 0, decreasing for x > 0, concave up 1. 3 sin 5 x − 4 cos5 x
+C
5
for x < − p1 and x > p1 , concave down for 3. 1 − x2
−2 e 1
+ 3 sin x3 + C
3 3
− p1 < x < p1 , horizontal asymptote: y = 0 5. ln(1 + e ) + Cx
3 3
7. local maximum at x = ln 2, inflection pt 7. 52 ( x + 4)5/2 − 38 ( x + 4)3/2 + C
3
¡ ¢
at x = ln 4, increasing for x < ln 2, decreas- 9. tan x − x + C 11. 10 tan−1 52x + C
ing for x > ln 2, concave up for x > ln 4, con- 13. 10 15. 1192 17. 1 19. − 48 1
15
cave down for x < ln 4, horizontal asymp- 21. π6 23. 12
tote: y = 0

1. x = 0.450184 3. x = 0.567143 1. 12 3. 1 5. divergent 7. ln12

5. x = 1.414213 11. global maximum at 9. divergent 11. 6 13. divergent
x = 2.8214 14. 50 15. 2 19. Yes 20. No
π
Section 4.4 (p. 122) Chapter 6

1. No 3. Yes 6. No 18. Hint: Calcu-
late f ′ ( x) and use the cosine addition for- Section 6.1 (p. 165)
mula. 2 2
1. x 2ln x − x4 + C 2. ( x2 − 2 x + 2) e x + C
3. x sin x + cos x + C
2 x
5. xlnaa − 2 xa 2a x
x
Chapter 5 2 + 3 +C
ln a ln a
7. x ln x2 − 2 x + C
Section 5.1 (p. 131) 9. Hint: Use apdouble-angle identity.
3 2 11. x sin−1 x + 1 − x2 + C
1. x3 + 52x − 3 x + C 3. 4 e x + C 13. x tan−1 3 x − 61 ln(1 + 9 x2 ) + C
5. −5 cos x + C 7. 6 ln| x | + C 15. − 38 sin x cos 3 x + 18 cos x sin 3 x + C
2
9. − 43 x3/2 + C 11. x2 + 37 x7/3 + C
17. 41 x4 ln2 x − 81 x4 ln x + 32
1 4
x + C 19. 16
3
13. 3 sec x + C 15. −7 cot x + C p
20. 2 2+2
15 21. x sin(ln
2
x)
− x cos(ln
2
x)
+C
2 −1 −1
x tan x
Section 5.2 (p. 139) 23. 2 − 2x + tan2 x + C
−1 p p −1 p
24. x cot x + x + cot x+C
1 1 1 p
3. 2 4. 3 5. 1 6. 4 25. Hint: Try the substitution t = x.

1 p
1. − 14 cos 7 x + 16 cos 3 x + C 1. The true value is P ≈ 7.4163 l / g (i.e.
1
3. − 14 sin 7 x + 61 sin 3 x + C the integral is ≈ 1.8541)
p
5. − 52 cos5/2 x + 92 cos9/2 x + C 2. 7.416331870724302 l / g
7. − 41 sin 2 x + 48
1 5
sin2 2 x + 16 3
x + 64 sin 4 x + C 3. 0.8948311310564181
1 3 1 4 7. 119.9785845899309
9. 3 tan x + tan x + C 11. 4 sin x + C
9. 0.5967390281992041 (The true value is
0.5963473623231939)
p p
1. 21 x 9 + 4 x2 + 49 ln | 2 x + 9 + 4 x2 | + C
p p Chapter 7
3. 12 x 4 x2 − 9 − 49 ln | 2 x + 4 x2 − 9 | + C
p
1− x 2
5. − sin−1 x − x +C Section 7.1 (p. 209)
p
7. ln| x | − ln | 1 + p1 + x2 | + C
9. 13 ( x2 + 4)3/2 − 4 x2 + 4 + C 2. Foci: (±3, 0), vertexes: (±5, 0), e = 53
1
¡ ¢ p p
11. 108 tan−1 23x + 18(9+x 4 x2) + C 3. Foci: (0 , ± 5 ), vertexes: (0, ±3), e = 35
p ³ p ´ p
13. −9 9 − x2 + 31 (9 − x2 )3/2 + C 5. Foci: ± 3
, 0 , vertexes: ( ± 1 , 0), e = 3
p ¡ 3 x−6 ¢ ³ 2
´ 2
1 2 1
15. − 9 −9 x2 + 36 x − 32 − 3 sin −
+C 2 2 2 2
2 9. ± aa2 +bb2 , ± aa2 +bb2 13. Hint: Use the
two points you know for certain are on the
Section 6.4 (p. 184) ellipse to find the location of the directrix.
1. − ln| x | + ln| x − 1 | + C 16. Hint: Use Exercise 15 and formula
3. 1 ln| 2 x − 1 | − 1 ln| x + 2 | + C (7.3).
5 5
5. 1x + 12 ln| x − 1 | − 21 ln| x + 1 | + C
7. 2 ln| x | + 1x − 2 ln| x + 1 | + C Section 7.2 (p. 215)
9. −3 ln| x | + 2x + 3 ln| x − 1 | + x−1 1 + C
¡ ¢ 2. Focus: (0, 2), vertex: (0, 0), directrix:
11. 31 tan−1 x − 16 tan−1 2x + C
y = −2 ¡ 1¢
3. Focus: 0, 32 , vertex: (0, 0), directrix:
Section 6.5 (p. 193) y = − 321
¡ ¢
1. tan 21 θ − ln | sin θ | + C 4. Focus: 12 , 0 , vertex: (0, 0), directrix:
µ 1 ¶ x = − 21
2 1 2 tan 2 θ−1 ¡ ¢
3. p tan − p +C 5. Focus: −121 , 0 , vertex: (0, 0), directrix:
3 3
µ 1 ¶ 1
4 −1 2 tanp2 θ−1
x = 12
5. p tan −θ+C 7. (0, 0) andp(4 p, 4 p); y = x 9. | 4 p |
3 3
¯ ¯ ¯ ¯
7. ln ¯ tan 21 θ ¯ − ln ¯ tan 12 θ + 1 ¯ + C 11. (3 p, ±2 3 p) ³ ´
¯ ¯ ¯ ¯ −b 4ac−b2 +1
9. ln ¯ tan 21 θ ¯ − 2 ln ¯ tan 21 θ + 1 ¯ + C 16. Focus: 2a
, 4a
, vertex:
p ³ ´
11. 2π 23. 3¡ 2 ¢ x2/3 −b 4ac−b 2 2
2Γ 2a
, 4a , directrix: y = 4ac−4ab −1
3
Appendix A: Answers and Hints to Selected Exercises 319

12. Hint: Use Exercise 11. p
2. Foci: (±5, 0), vertexes: (±4, 0), directri- 24. local maximum at x = ln 3, inflection
ces: x = ± 16 3
x, e = 45
5p, asymptotes: y = ± 4 p pt at x = ln 3, horizontal asymptote: y = 0
3. Foci: (± 23, 0), vertexes: (±2 2, 0), di- 26. (b) Hint: See Exercise 10.
rectrices: x = ± p8 , 27. (b) k 1 = c 1 + c 2 , k 2 = c 1 − c 2
23p p
15
asymptotes: y = ± p x, e = p 23 30. s 0 ≈ 1.006237835313385,
p 2 2 2 2 eπ/s0 = 22.69438187638412
41 5
4. Foci: (± 2 , 0), vertexes: (± 2 , 0), direc- 33. (a) x = cx+ c y − y−x , y = cx+ c y + y−x
2 2c 2 2c
25
trices: x = ± p , asymptotes: y = ± 54 x, (b) c = cosh a + sinh a
p 2 41
(c) Hint: Use Example 7.12.
e = 541
p
5
5. Foci: (± 2 , 0), vertexes:
(±1, 0), directri- Section 7.6 (p. 243)
p
ces: x = ± p2 , asymptotes:y = ± 21 x, e = 25
5 p 5. Yes 6. (a) Hint: Solve for t in terms
6. Foci: (0, ± 34), vertexes: (0, ±3), di- of x then substitute into y.
rectrices: y = ± p9 , asymptotes: y = ± 53 x, (b) Hint: Use the distance formula.
p 34
e= 3 34 8. Hint: Does BP = AB? 9. (a) − (2 t+71)3
¡ 36 14 ¢
2 y2 x2 y2 (c) 25 , 5 12. x = t2 − 1, y = t( t2 − 1)
7. x9 − 16 = 1 17. 302500 − 697500 = 1
1. r = 6 cos θ 3. r 2 = sec 2θ
Section 7.4 (p. 229) 5. θ = 34π 7. r = sec θ tan θ
9. x4 + 2 x2 y2 +p y4 = 4 x2 − 4 y2
p
1. Foci: (0, 2) and (6, 2), vertexes: (−2, 2) 11. y − 1 = − 3 5 3 ( x + 3 )
¡ ¢ ¡ ¢
and (8, 2) p 13. local maxima at 2, π2 and 0, 32π ,
¡ 1 7π ¢ ¡ 1 11π ¢
3. Foci: (−3, 1 ± 2 3), vertexes: (−3, −3) local minima at 2 , 6 and 2 , 6
³ p ´
and (−3, 5) 15. local maxima at 2 3 2 , α and
5. Focus: (−3, − 239
16
), vertex: (−3, −15), di- ³ p ´
rectrix: y = − 241 − 2 3 2 , 2π − α , local minima at
¡ 16 ¢ ¡ ¢ ³ p ´ ³ p ´
7. Focus: 12 , − 74 , vertex: 12 , − 32 , direc- − 2 3 2 , π − α and 2 3 2 , π + α ,
trix: y = − 45 p
p where α = tan−1 2 17. π2
9. Foci: (−1 ± 13), vertexes: (−4, −3) and
(2, −3), directrices: x = −1 ± p9 , asymp-
13 Chapter 8
totes: y = ± 32 ( x + 1) − 3
p p p p
11. Foci: ( 2, 2) and (− 2, − 2), ver- Section 8.1 (p. 256)
texes:p (1, 1) and −1, −1), directices: y =
− x ± 2, asymptotes: x = 0 and y = 0 1. 32
3 2. 43 3. 12
1
5. 29 7. 19
3 9. 9
140
¡π p ¢
15. hyperbola 11. 3π 13. 25 3 + 1 − 3
6250π3 113
17. sq ft 7. Divergent 9. 6 11. 32 13. 999
3
14. 1 15. 41 20. 132
7 ft 24. No
4 2 1
1. 1 3. 3
5. π
7. 0 9. 2
ln 3
6. Divergent 7. Convergent
8 3/2 1 3/2 12. Convergent 13. Convergent
1. 27 (10 ) − 27 (13 ) ≈ 7.634
p
5 1 −1 14. Divergent 15. Divergent
2. 2 + 4 sinh 2 ≈ 1.479 16. Convergent 17. Convergent
8 3/2 1 3/2
3. 27 (10 p ) − 27 (13 ) ≈ 7.634 18. 61 19. 10
1
20. 16 21. 12
3
5. 4 + ln 2 ≈ 1.097
p
7. 2( e − 1) ≈ 31.312
π
9. 8 11. 3 13. κ(0) = 0, κ( π2 ) = −1
15. κ(0) = ba2 at (a, 0), κ( π2 ) = ab2 at (0, b) 1. Conditionally convergent
17. −1 23. 13.27 ft 3. Conditionally convergent
5. Absolutely convergent
7. Answer to second question: Yes
1. 4π 2. 2 (2 + sinh 2)
π
3. 208
9
π
π2
5. 2 7. 2π 9. 10 π
10. π6 1. −1 ≤ x < 1 2. −2 < x < 2 3. 1 < x < 3
p
13. S = πr r 2 + h2 , V = 31 πr 2 h 4. −6 < x < −2 5. −∞ < x < ∞ 6. x = 0
7. −1 < x < 1 10. Hint: Use Example
Section 8.5 (p. 283) 9.19
¡ ¢ ¡ 12 ¢ ¡ ¢
1. 45 , 72 3. 53 , 35 5. 34πr , 34πr Section 9.5 (p. 313)
¡ ¢
7. 0, 411π 9. 0.192 Nm
µ ¶ π 2 π 4
1 1
¡ 1 ¢ 1. 1 − ( x− 2 ) + ( x− 2 ) − · · ·
11. RT V 2 − V 2 13. 0.3486 15. 1, 4 3
2!
5
4!
2 4
a b 2. x + x3! + x5! + · · · 3. 1 + x2! + x4! + · · ·
18. Hint: Use Exercise 28 in Section 6.1. x3 2 x5 x3 2 x5
20. Hint: Use the equation from Section 4. x + 32 + 154 + · · · 5. x − 3 + 15 − · · ·
5.1 for free fall motion to write time as a 6. 1 + x2 + 524 x
+ · · · 7. 1 − x2 + x4 − · · ·
2
function of height. 8. 1 − ( x−1) + ( x−1) − · · ·
2 2 4
x2 x4 4 6
9. 1 + 2 − 8 + · · · 11. x2 − x2 + 38x − · · ·
Chapter 9 x5 x9 x x 7 13
13. x − 10 + 216 −··· 15. x + 14 − 104 +···
18. 0.68485 19. 97.18; −132.605
1
1. Converges to 0 2. Converges to 3
3. Converges to 0 5. Divergent
GNU Free Documentation License
Version 1.3, 3 November 2008

Copyright © 2000, 2001, 2002, 2007, 2008 Free Software Foundation, Inc.
<http://fsf.org/>
Everyone is permitted to copy and distribute verbatim copies of this license

document, but changing it is not allowed.
Preamble
The purpose of this License is to make a manual, textbook, or other functional and
useful document “free” in the sense of freedom: to assure everyone the effective free-
dom to copy and redistribute it, with or without modifying it, either commercially or
noncommercially. Secondarily, this License preserves for the author and publisher a
way to get credit for their work, while not being considered responsible for modifica-
tions made by others.
This License is a kind of “copyleft”, which means that derivative works of the doc-
ument must themselves be free in the same sense. It complements the GNU General
Public License, which is a copyleft license designed for free software.
We have designed this License in order to use it for manuals for free software, be-
cause free software needs free documentation: a free program should come with manu-
als providing the same freedoms that the software does. But this License is not limited
to software manuals; it can be used for any textual work, regardless of subject matter
or whether it is published as a printed book. We recommend this License principally
for works whose purpose is instruction or reference.
1. APPLICABILITY AND DEFINITIONS

This License applies to any manual or other work, in any medium, that contains
a notice placed by the copyright holder saying it can be distributed under the terms
of this License. Such a notice grants a world-wide, royalty-free license, unlimited
in duration, to use that work under the conditions stated herein. The “Document”,
321
322 GNU Free Documentation License
below, refers to any such manual or work. Any member of the public is a licensee, and
is addressed as “you”. You accept the license if you copy, modify or distribute the work
in a way requiring permission under copyright law.
A “Modified Version” of the Document means any work containing the Document
or a portion of it, either copied verbatim, or with modifications and/or translated into
another language.
A “Secondary Section” is a named appendix or a front-matter section of the Doc-
ument that deals exclusively with the relationship of the publishers or authors of the
Document to the Document’s overall subject (or to related matters) and contains noth-
ing that could fall directly within that overall subject. (Thus, if the Document is in
part a textbook of mathematics, a Secondary Section may not explain any mathemat-
ics.) The relationship could be a matter of historical connection with the subject or
with related matters, or of legal, commercial, philosophical, ethical or political posi-
tion regarding them.
The “Invariant Sections” are certain Secondary Sections whose titles are desig-
nated, as being those of Invariant Sections, in the notice that says that the Document
is released under this License. If a section does not fit the above definition of Secondary
then it is not allowed to be designated as Invariant. The Document may contain zero
Invariant Sections. If the Document does not identify any Invariant Sections then
there are none.
The “Cover Texts” are certain short passages of text that are listed, as Front-Cover
Texts or Back-Cover Texts, in the notice that says that the Document is released under
this License. A Front-Cover Text may be at most 5 words, and a Back-Cover Text may
be at most 25 words.
A “Transparent” copy of the Document means a machine-readable copy, repre-
sented in a format whose specification is available to the general public, that is suit-
able for revising the document straightforwardly with generic text editors or (for im-
ages composed of pixels) generic paint programs or (for drawings) some widely avail-
able drawing editor, and that is suitable for input to text formatters or for automatic
translation to a variety of formats suitable for input to text formatters. A copy made in
an otherwise Transparent file format whose markup, or absence of markup, has been
arranged to thwart or discourage subsequent modification by readers is not Transpar-
ent. An image format is not Transparent if used for any substantial amount of text. A
copy that is not “Transparent” is called “Opaque”.
Examples of suitable formats for Transparent copies include plain ASCII without
markup, Texinfo input format, LaTeX input format, SGML or XML using a publicly
available DTD, and standard-conforming simple HTML, PostScript or PDF designed
for human modification. Examples of transparent image formats include PNG, XCF
and JPG. Opaque formats include proprietary formats that can be read and edited only
by proprietary word processors, SGML or XML for which the DTD and/or processing
tools are not generally available, and the machine-generated HTML, PostScript or
PDF produced by some word processors for output purposes only.

The “Title Page” means, for a printed book, the title page itself, plus such following
pages as are needed to hold, legibly, the material this License requires to appear in the
title page. For works in formats which do not have any title page as such, “Title Page”
means the text near the most prominent appearance of the work’s title, preceding the
beginning of the body of the text.
The “publisher” means any person or entity that distributes copies of the Document
to the public.
A section “Entitled XYZ” means a named subunit of the Document whose title ei-
ther is precisely XYZ or contains XYZ in parentheses following text that translates
XYZ in another language. (Here XYZ stands for a specific section name mentioned be-
low, such as “Acknowledgments”, “Dedications”, “Endorsements”, or “History”.)
To “Preserve the Title” of such a section when you modify the Document means that
it remains a section “Entitled XYZ” according to this definition.
The Document may include Warranty Disclaimers next to the notice which states
that this License applies to the Document. These Warranty Disclaimers are considered
to be included by reference in this License, but only as regards disclaiming warranties:
any other implication that these Warranty Disclaimers may have is void and has no
effect on the meaning of this License.
2. VERBATIM COPYING
You may copy and distribute the Document in any medium, either commercially or
noncommercially, provided that this License, the copyright notices, and the license
notice saying this License applies to the Document are reproduced in all copies, and
that you add no other conditions whatsoever to those of this License. You may not use
technical measures to obstruct or control the reading or further copying of the copies
you make or distribute. However, you may accept compensation in exchange for copies.
If you distribute a large enough number of copies you must also follow the conditions
in section 3.
You may also lend copies, under the same conditions stated above, and you may
publicly display copies.
3. COPYING IN QUANTITY
If you publish printed copies (or copies in media that commonly have printed cov-
ers) of the Document, numbering more than 100, and the Document’s license notice
requires Cover Texts, you must enclose the copies in covers that carry, clearly and
legibly, all these Cover Texts: Front-Cover Texts on the front cover, and Back-Cover
Texts on the back cover. Both covers must also clearly and legibly identify you as the
publisher of these copies. The front cover must present the full title with all words of
the title equally prominent and visible. You may add other material on the covers in
addition. Copying with changes limited to the covers, as long as they preserve the title
of the Document and satisfy these conditions, can be treated as verbatim copying in
other respects.
If the required texts for either cover are too voluminous to fit legibly, you should put
the first ones listed (as many as fit reasonably) on the actual cover, and continue the
rest onto adjacent pages.
If you publish or distribute Opaque copies of the Document numbering more than
100, you must either include a machine-readable Transparent copy along with each
Opaque copy, or state in or with each Opaque copy a computer-network location from
which the general network-using public has access to download using public-standard
network protocols a complete Transparent copy of the Document, free of added mate-
rial. If you use the latter option, you must take reasonably prudent steps, when you
begin distribution of Opaque copies in quantity, to ensure that this Transparent copy
will remain thus accessible at the stated location until at least one year after the last
time you distribute an Opaque copy (directly or through your agents or retailers) of
that edition to the public.
It is requested, but not required, that you contact the authors of the Document well
before redistributing any large number of copies, to give them a chance to provide you
with an updated version of the Document.
4. MODIFICATIONS
You may copy and distribute a Modified Version of the Document under the condi-
tions of sections 2 and 3 above, provided that you release the Modified Version under
precisely this License, with the Modified Version filling the role of the Document, thus
licensing distribution and modification of the Modified Version to whoever possesses a
copy of it. In addition, you must do these things in the Modified Version:
A. Use in the Title Page (and on the covers, if any) a title distinct from that of the
Document, and from those of previous versions (which should, if there were any,
be listed in the History section of the Document). You may use the same title as a
previous version if the original publisher of that version gives permission.
B. List on the Title Page, as authors, one or more persons or entities responsible for
authorship of the modifications in the Modified Version, together with at least five
of the principal authors of the Document (all of its principal authors, if it has fewer
than five), unless they release you from this requirement.
C. State on the Title page the name of the publisher of the Modified Version, as the
publisher.
D. Preserve all the copyright notices of the Document.

E. Add an appropriate copyright notice for your modifications adjacent to the other
copyright notices.
F. Include, immediately after the copyright notices, a license notice giving the public
permission to use the Modified Version under the terms of this License, in the form
shown in the Addendum below.
G. Preserve in that license notice the full lists of Invariant Sections and required
Cover Texts given in the Document’s license notice.
H. Include an unaltered copy of this License.
I. Preserve the section Entitled “History”, Preserve its Title, and add to it an item
stating at least the title, year, new authors, and publisher of the Modified Version
as given on the Title Page. If there is no section Entitled “History” in the Document,
create one stating the title, year, authors, and publisher of the Document as given
on its Title Page, then add an item describing the Modified Version as stated in the
previous sentence.
J. Preserve the network location, if any, given in the Document for public access to
a Transparent copy of the Document, and likewise the network locations given in
the Document for previous versions it was based on. These may be placed in the
“History” section. You may omit a network location for a work that was published
at least four years before the Document itself, or if the original publisher of the
version it refers to gives permission.
K. For any section Entitled “Acknowledgments” or “Dedications”, Preserve the Title

of the section, and preserve in the section all the substance and tone of each of the
contributor acknowledgments and/or dedications given therein.
L. Preserve all the Invariant Sections of the Document, unaltered in their text and in
their titles. Section numbers or the equivalent are not considered part of the section
titles.
M. Delete any section Entitled “Endorsements”. Such a section may not be included
in the Modified Version.
N. Do not retitle any existing section to be Entitled “Endorsements” or to conflict in

title with any Invariant Section.
O. Preserve any Warranty Disclaimers.
If the Modified Version includes new front-matter sections or appendices that qualify
as Secondary Sections and contain no material copied from the Document, you may at
your option designate some or all of these sections as invariant. To do this, add their
titles to the list of Invariant Sections in the Modified Version’s license notice. These
titles must be distinct from any other section titles.
You may add a section Entitled “Endorsements”, provided it contains nothing but
endorsements of your Modified Version by various parties—for example, statements of
peer review or that the text has been approved by an organization as the authoritative
definition of a standard.
You may add a passage of up to five words as a Front-Cover Text, and a passage of up
to 25 words as a Back-Cover Text, to the end of the list of Cover Texts in the Modified
Version. Only one passage of Front-Cover Text and one of Back-Cover Text may be
added by (or through arrangements made by) any one entity. If the Document already
includes a cover text for the same cover, previously added by you or by arrangement
made by the same entity you are acting on behalf of, you may not add another; but
you may replace the old one, on explicit permission from the previous publisher that
added the old one.
The author(s) and publisher(s) of the Document do not by this License give per-
mission to use their names for publicity for or to assert or imply endorsement of any
Modified Version.
5. COMBINING DOCUMENTS
You may combine the Document with other documents released under this License,
under the terms defined in section 4 above for modified versions, provided that you
include in the combination all of the Invariant Sections of all of the original documents,
unmodified, and list them all as Invariant Sections of your combined work in its license
notice, and that you preserve all their Warranty Disclaimers.
The combined work need only contain one copy of this License, and multiple iden-
tical Invariant Sections may be replaced with a single copy. If there are multiple
Invariant Sections with the same name but different contents, make the title of each
such section unique by adding at the end of it, in parentheses, the name of the original
author or publisher of that section if known, or else a unique number. Make the same
adjustment to the section titles in the list of Invariant Sections in the license notice of
the combined work.
In the combination, you must combine any sections Entitled “History” in the various
original documents, forming one section Entitled “History”; likewise combine any sec-
tions Entitled “Acknowledgments”, and any sections Entitled “Dedications”. You must
delete all sections Entitled “Endorsements”.
6. COLLECTIONS OF DOCUMENTS
You may make a collection consisting of the Document and other documents released
under this License, and replace the individual copies of this License in the various
documents with a single copy that is included in the collection, provided that you
follow the rules of this License for verbatim copying of each of the documents in all
other respects.
You may extract a single document from such a collection, and distribute it individ-
ually under this License, provided you insert a copy of this License into the extracted
document, and follow this License in all other respects regarding verbatim copying of
that document.
7. AGGREGATION WITH INDEPENDENT WORKS

A compilation of the Document or its derivatives with other separate and indepen-
dent documents or works, in or on a volume of a storage or distribution medium, is
called an “aggregate” if the copyright resulting from the compilation is not used to
limit the legal rights of the compilation’s users beyond what the individual works per-
mit. When the Document is included in an aggregate, this License does not apply to
the other works in the aggregate which are not themselves derivative works of the
Document.
If the Cover Text requirement of section 3 is applicable to these copies of the Doc-
ument, then if the Document is less than one half of the entire aggregate, the Doc-
ument’s Cover Texts may be placed on covers that bracket the Document within the
aggregate, or the electronic equivalent of covers if the Document is in electronic form.
Otherwise they must appear on printed covers that bracket the whole aggregate.
8. TRANSLATION
Translation is considered a kind of modification, so you may distribute translations
of the Document under the terms of section 4. Replacing Invariant Sections with
translations requires special permission from their copyright holders, but you may in-
clude translations of some or all Invariant Sections in addition to the original versions
of these Invariant Sections. You may include a translation of this License, and all the
license notices in the Document, and any Warranty Disclaimers, provided that you
also include the original English version of this License and the original versions of
those notices and disclaimers. In case of a disagreement between the translation and
the original version of this License or a notice or disclaimer, the original version will
prevail.
If a section in the Document is Entitled “Acknowledgments”, “Dedications”, or “His-
tory”, the requirement (section 4) to Preserve its Title (section 1) will typically require
changing the actual title.
9. TERMINATION
You may not copy, modify, sublicense, or distribute the Document except as expressly
provided under this License. Any attempt otherwise to copy, modify, sublicense, or
distribute it is void, and will automatically terminate your rights under this License.
However, if you cease all violation of this License, then your license from a particular
copyright holder is reinstated (a) provisionally, unless and until the copyright holder
explicitly and finally terminates your license, and (b) permanently, if the copyright
holder fails to notify you of the violation by some reasonable means prior to 60 days
after the cessation.
Moreover, your license from a particular copyright holder is reinstated permanently
if the copyright holder notifies you of the violation by some reasonable means, this is
the first time you have received notice of violation of this License (for any work) from
that copyright holder, and you cure the violation prior to 30 days after your receipt of
the notice.
Termination of your rights under this section does not terminate the licenses of par-
ties who have received copies or rights from you under this License. If your rights
have been terminated and not permanently reinstated, receipt of a copy of some or all
of the same material does not give you any rights to use it.
10. FUTURE REVISIONS OF THIS LICENSE

The Free Software Foundation may publish new, revised versions of the GNU Free
Documentation License from time to time. Such new versions will be similar in spirit
to the present version, but may differ in detail to address new problems or concerns.
See http://www.gnu.org/copyleft/.
Each version of the License is given a distinguishing version number. If the Docu-
ment specifies that a particular numbered version of this License “or any later version”
applies to it, you have the option of following the terms and conditions either of that
specified version or of any later version that has been published (not as a draft) by
the Free Software Foundation. If the Document does not specify a version number of
this License, you may choose any version ever published (not as a draft) by the Free
Software Foundation. If the Document specifies that a proxy can decide which future
versions of this License can be used, that proxy’s public statement of acceptance of a
version permanently authorizes you to choose that version for the Document.
11. RELICENSING
“Massive Multiauthor Collaboration Site” (or “MMC Site”) means any World Wide
Web server that publishes copyrightable works and also provides prominent facilities
for anybody to edit those works. A public wiki that anybody can edit is an example
of such a server. A “Massive Multiauthor Collaboration” (or “MMC”) contained in the
site means any set of copyrightable works thus published on the MMC site.
“CC-BY-SA” means the Creative Commons Attribution-Share Alike 3.0 license pub-
lished by Creative Commons Corporation, a not-for-profit corporation with a principal
place of business in San Francisco, California, as well as future copyleft versions of
that license published by that same organization.
“Incorporate” means to publish or republish a Document, in whole or in part, as part

of another Document.
An MMC is “eligible for relicensing” if it is licensed under this License, and if all
works that were first published under this License somewhere other than this MMC,
and subsequently incorporated in whole or in part into the MMC, (1) had no cover
texts or invariant sections, and (2) were thus incorporated prior to November 1, 2008.
The operator of an MMC Site may republish an MMC contained in the site under
CC-BY-SA on the same site at any time before August 1, 2009, provided the MMC is
eligible for relicensing.
ADDENDUM: How to use this License for your documents

To use this License in a document you have written, include a copy of the License
in the document and put the following copyright and license notices just after the title
page:
Copyright © YEAR YOUR NAME. Permission is granted to copy, distribute and/or

modify this document under the terms of the GNU Free Documentation License,
Version 1.3 or any later version published by the Free Software Foundation; with
no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts. A copy
of the license is included in the section entitled “GNU Free Documentation Li-
cense”.
If you have Invariant Sections, Front-Cover Texts and Back-Cover Texts, replace the
“with . . . Texts.” line with this:
with the Invariant Sections being LIST THEIR TITLES, with the Front-Cover
Texts being LIST, and with the Back-Cover Texts being LIST.
If you have Invariant Sections without Cover Texts, or some other combination of
the three, merge those two alternatives to suit the situation.
If your document contains nontrivial examples of program code, we recommend re-
leasing these examples in parallel under your choice of free software license, such as
the GNU General Public License, to permit their use in free software.
History
This section contains the revision history of the book. For persons making modifica-
tions to the book, please record the pertinent information here, following the format
in the first item below.
1. VERSION: 0.1
Date: 2016-01-24
Author(s): Michael Corral
Title: Elementary Calculus
Modification(s): Initial version
2. VERSION: 0.5
Date: 2020-05-27
Modification(s): Numerous corrections of typos and errors, as well as enhance-
ments of some sections and more exercises.
3. VERSION: 0.5a
Date: 2020-06-03
Modification(s): Replaced the old Section 5.5 (Average Value of a Function) with
the new Section 5.5 (Improper Integrals). Average Value of a Function will be
moved to Chapter 8 in the second half of the book.
4. VERSION: 1.0
Date: 2020-12-31
330
History 331

Modification(s): Full version.
Index
Symbols polar. . . . . . . . . . . . . . . . . . . . . . . . . . . .266

∆x . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 area between curves. . . . . . . . . . . . . . . .252
〈 f 〉 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .258 area function . . . . . . . . . . . . . . . . . . . . . . . 140
f˙ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 area under a curve . . . . . . . . . . . . . . . . . 133
dx . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 asymptote . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
d2 y
dx2
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 horizontal . . . . . . . . . . . . . . . . . . . . . . . 67
dn
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 vertical . . . . . . . . . . . . . . . . . . . . . . . . . . 66
dx n
≫ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 average value . . . . . . . . . . . . . . . . . . . . . . 258
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238 axes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
≪ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 of an ellipse . . . . . . . . . . . . . . . . . . . . 203
I.......................................4
N......................................4 B
Q......................................4 Bézier curve . . . . . . . . . . . . . . . . . . . . . . . . 241
R ......................................4 base a logarithm . . . . . . . . . . . . . . . . . . . . .54
f ′ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8 Bessel functions . . . . . . . . . . . . . . . . . . . . 305
f (n) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32 order m . . . . . . . . . . . . . . . . . . . . . . . . .306
n-th Term Test . . . . . . . . . . . . . . . . . . . . . 294 order zero . . . . . . . . . . . . . . . . . . . . . . 305
Bessel’s equation . . . . . . . . . . . . . . . . . . . 305
A order m . . . . . . . . . . . . . . . . . . . . . . . . .306
absolute convergence . . . . . . . . . . . . . . . 301 Beta function . . . . . . . . . . . . . . . . . . . . . . . 188
absolute value . . . . . . . . . . . . . . . . . . . . . . . 13 Big O notation . . . . . . . . . . . . . . . . . . . . . . . 68
accelerating. . . . . . . . . . . . . . . . . . . . . . . . . .34 bipolar coordinates . . . . . . . . . . . . . . . . . 251
acceleration . . . . . . . . . . . . . . . . . . . . . . . . . . 33 bisection method . . . . . . . . . . . . . . . . . . . 109
algebraic curve . . . . . . . . . . . . . . . . . . . . . . 80 Bohr radius . . . . . . . . . . . . . . . . . . . . . . . . 107
alternating series . . . . . . . . . . . . . . . . . . 300 brachistochrone . . . . . . . . . . . . . . . . . . . . 243
angle of contingence . . . . . . . . . . . . . . . . 267
angle of tangent line . . . . . . . . . . . . . . . . . 59 C
antiderivative . . . . . . . . . . . . . . . . . . . . . . 124 catenary . . . . . . . . . . . . . . . . . . . . . . . 231, 264
arc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 ceiling function . . . . . . . . . . . . . . . . . . . . . . 74
arc length . . . . . . . . . . . . . . . . . . . . . . . . . . 263 center of gravity . . . . . . . . . . . . . . . . . . . . 277
parametric . . . . . . . . . . . . . . . . . . . . . 266 Chain Rule . . . . . . . . . . . . . . . . . . . . . . . . . . 27
332
Index 333
Chebyshev polynomials. . . . . . . . . . . . . .44 fractional . . . . . . . . . . . . . . . . . . . . . . . 189

circle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 higher order . . . . . . . . . . . . . . . . . . . . . 32
definition . . . . . . . . . . . . . . . . . . . . . . . 202 natural logarithm function . . . . . . 48
involute . . . . . . . . . . . . . . . . . . . . . . . . 244 polar coordinates . . . . . . . . . . . . . . . 248
circular functions. . . . . . . . . . . . . . . . . . .230 Power Rule . . . . . . . . . . . . . . . . . . . . . . 24
closed interval . . . . . . . . . . . . . . . . . . . . . . . 91 Product Rule . . . . . . . . . . . . . . . . . . . . 21
Comparison Test . . . . . . . . . . . . . . . . . . . 293 Quotient Rule . . . . . . . . . . . . . . . . . . . 21
concave down . . . . . . . . . . . . . . . . . . . . . . . 102 second . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
concave up. . . . . . . . . . . . . . . . . . . . . . . . . .102 Sum Rule . . . . . . . . . . . . . . . . . . . . . . . . 21
concavity . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 trigonometric functions . . . . . . . . . .23
conditional convergence . . . . . . . . . . . . 301 with respect to x . . . . . . . . . . . . . . . . . 15
conic sections . . . . . . . . . . . . . . . . . . . . . . . 221 zero-th . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
conjugate axis . . . . . . . . . . . . . . . . . . . . . . 217 Difference Rule . . . . . . . . . . . . . . . . . . . . . . 21
constant function . . . . . . . . . . . . . . . . . . . . . 9 differentiable . . . . . . . . . . . . . . . . . . . . . . . . 14
Constant Multiple Rule. . . . . . . . . . . . . .21 differential. . . . . . . . . . . . . . . . . . . . . . . . . . .84
continued fraction . . . . . . . . . . . . . . . . . . 292 differential equation . . . . . . . . . . . . . 36, 47
continuity. . . . . . . . . . . . . . . . . . . . . . . . . . . .73 differentiation . . . . . . . . . . . . . . . . . . . . . . . 14
continuous function. . . . . . . . . . . . . . . . . .73 implicit . . . . . . . . . . . . . . . . . . . . . . . . . . 79
convergent integral . . . . . . . . . . . 152, 155 under the integral sign . . . . . . . . . 185
coordinate transformations . . . . . . . . 224 Dirac delta function . . . . . . . . . . . . . . . . 152
CORDIC . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313 directrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
countable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 disc method . . . . . . . . . . . . . . . . . . . . . . . . 273
crackle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 discontinuous . . . . . . . . . . . . . . . . . . . . . . . . 73
critical point . . . . . . . . . . . . . . . . . . . . . . . . . 91 distribution function . . . . . . . . . . . . . . . 284
curvature . . . . . . . . . . . . . . . . . . . . . . 267, 268 divergent integral . . . . . . . . . . . . . 152, 155
average . . . . . . . . . . . . . . . . . . . . . . . . . 267 domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8, 37
parametric . . . . . . . . . . . . . . . . . . . . . 269 dot notation . . . . . . . . . . . . . . . . . . . . . . . . . .15
polar. . . . . . . . . . . . . . . . . . . . . . . . . . . .269
cycloid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242 E
e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
D eccentric angle . . . . . . . . . . . . . . . . . . . . . 243
Darboux’s Theorem. . . . . . . . . . . . . . . . .122 Efimov trimer . . . . . . . . . . . . . . . . . . . . . . 237
decelerating . . . . . . . . . . . . . . . . . . . . . . . . . 34 eigenfunction . . . . . . . . . . . . . . . . . . . . . . . . 36
definite integral . . . . . . . . . . . . . . . . . . . . 132 eigenvalue . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
derivative . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 8 elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
n-th . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 ellipse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
Chain Rule . . . . . . . . . . . . . . . . . . . . . . 27 area . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Constant Multiple Rule . . . . . . . . . 21 center . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Difference Rule . . . . . . . . . . . . . . . . . . 21 construction . . . . . . . . . . . . . . . . . . . . 203
exponential function . . . . . . . . . . . . . 46 definition . . . . . . . . . . . . . . . . . . . . . . . 202
first . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 diameter . . . . . . . . . . . . . . . . . . . . . . . 203
334 Index
directrix . . . . . . . . . . . . . . . . . . . . . . . . 209 free fall formulas . . . . . . . . . . . . . . 129, 130

eccentric angle . . . . . . . . . . . . . . . . . 243 frustrum. . . . . . . . . . . . . . . . . . . . . . . . . . . .271
foci . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37
latus rectum . . . . . . . . . . . . . . . . . . . .209 absolute value . . . . . . . . . . . . . . . . . . . 13
major auxiliary circle. . . . . . . . . . .243 area . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
major axis . . . . . . . . . . . . . . . . . . . . . . 203 average value . . . . . . . . . . . . . . . . . . 258
minor auxiliary circle . . . . . . . . . . 243 ceiling . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
minor axis . . . . . . . . . . . . . . . . . . . . . . 203 circular . . . . . . . . . . . . . . . . . . . . . . . . . 230
principal axis . . . . . . . . . . . . . . . . . . . 203 composition . . . . . . . . . . . . . . . . . . . . . . 27
reflection property. . . . . . . . . . . . . .206 constant . . . . . . . . . . . . . . . . . . . . . . . . . . 9
second definition . . . . . . . . . . . . . . . 209 continuous . . . . . . . . . . . . . . . . . . . . . . . 73
semi-major axis . . . . . . . . . . . . . . . . 203 differentiable . . . . . . . . . . . . . . . . . . . . 14
semi-minor axis . . . . . . . . . . . . . . . . 203 elliptic . . . . . . . . . . . . . . . . . . . . . . . . . . 270
vertex . . . . . . . . . . . . . . . . . . . . . . . . . . 203 even . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
ellipsograph . . . . . . . . . . . . . . . . . . . . . . . . 244 exponential . . . . . . . . . . . . . . . . . . . . . . 45
ellipsoid . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276 floor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .74
elliptic curve . . . . . . . . . . . . . . . . . . . . . . . . . 79 hyperbolic . . . . . . . . . . . . . . . . . . . . . . 230
elliptic function . . . . . . . . . . . . . . . . . . . . .270 infinitesimal . . . . . . . . . . . . . . . . . . . . . 72
elliptic integral . . . . . . . . . . . . . . . . . . . . . 264 inverse. . . . . . . . . . . . . . . . . . . . . . . . . . .38
enthalpy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 inverse hyperbolic . . . . . . . . . . . . . . 234
entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 inverse trigonometric . . . . . . . . . . . . 41
escape velocity . . . . . . . . . . . . . . . . . . . . . 216 Langevin . . . . . . . . . . . . . . . . . . . . . . . 238
Eudoxus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 linear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Euler number. . . . . . . . . . . . . . . . . . . . . . . .45 natural logarithm . . . . . . . . . . . . . . . 47
even function . . . . . . . . . . . . . . . . . . . . . . . . 13 objective . . . . . . . . . . . . . . . . . . . . . . . . . 90
event . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 odd . . . . . . . . . . . . . . . . . . . . . . . . . 13, 143
expected value. . . . . . . . . . . . . . . . . . . . . .284 rational . . . . . . . . . . . . . . . . . . . . . 70, 178
exponential decay . . . . . . . . . . . . . . . . . . . 50 root . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
exponential distribution . . . . . . . . . . . . 283 sigmoid neuron . . . . . . . . . . . . . . . . . . 52
exponential functions . . . . . . . . . . . . . . . . 45 step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
exponential growth . . . . . . . . . . . . . . . . . . 51 trigonometric . . . . . . . . . . . . . . . . . . . . 41
Extended Mean Value Theorem . . . . 121 Fundamental Theorem of Calculus 140
G
F Gamma function . . . . . . . . . . . . . . 159, 188
factorial. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34
Gaussian quadrature . . . . . . . . . . . . . . .199
Fermat’s Principle . . . . . . . . . . . . . . 95, 101
Gibbs energy . . . . . . . . . . . . . . . . . . . . . . . . 89
Fibonacci sequence . . . . . . . . . . . . 288, 292
golden ratio . . . . . . . . . . . . . . . . . . . 288, 292
First Derivative Test . . . . . . . . . . . . . . . 105
floor function . . . . . . . . . . . . . . . . . . . . . . . . 74 H
focal radius . . . . . . . . . . . . . . . . . . . . 212, 223 half-angle substitution . . . . . . . . . . . . . 190
fractional derivative . . . . . . . . . . . . . . . . 189 half-life . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Index 335
harmonic series . . . . . . . . . . . . . . . . . . . . 296 inflection point . . . . . . . . . . . . . . . . . . . . . 103

height function . . . . . . . . . . . . . . . . . . . . . 252 instantaneous rate of change . . . . . . .3, 8
Hooke’s Law . . . . . . . . . . . . . . . . . . . . . . . . 281 integers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
horizontal asymptote . . . . . . . . . . . . . . . . 67 integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
hyperbola . . . . . . . . . . . . . . . . . . . . . . . . . . .216 convergent . . . . . . . . . . . . . . . . 152, 155
asymptotes . . . . . . . . . . . . . . . . . . . . . 217 definite . . . . . . . . . . . . . . . . . . . . . . . . . 132
center . . . . . . . . . . . . . . . . . . . . . . . . . . 217 divergent . . . . . . . . . . . . . . . . . . 152, 155
conjugate axis . . . . . . . . . . . . . . . . . . 217 exponential function . . . . . . . . . . . 128
construction . . . . . . . . . . . . . . . . . . . . 219 improper . . . . . . . . . . . . . . . . . . . . . . . 152
definition . . . . . . . . . . . . . . . . . . . . . . . 216 indefinite . . . . . . . . . . . . . . . . . . . . . . . 125
directrix . . . . . . . . . . . . . . . . . . . . . . . . 216 trigonometric . . . . . . . . . . . . . .128, 167
eccentricity . . . . . . . . . . . . . . . . . . . . . 216 integral sign . . . . . . . . . . . . . . . . . . . . . . . . 125
focal radius . . . . . . . . . . . . . . . . . . . . . 223 Integral Test . . . . . . . . . . . . . . . . . . . . . . . 294
focus . . . . . . . . . . . . . . . . . . . . . . . . . . . .216 integrand . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
latus rectum . . . . . . . . . . . . . . . . . . . .223 integration . . . . . . . . . . . . . . . . . . . . . . . . . 126
reflection property. . . . . . . . . . . . . .220 by parts . . . . . . . . . . . . . . . . . . . . . . . . 159
sector . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 Monte Carlo . . . . . . . . . . . . . . . . . . . . 255
transverse axis . . . . . . . . . . . . . . . . . 217 numerical . . . . . . . . . . . . . . . . . . . . . . 195
vertex . . . . . . . . . . . . . . . . . . . . . . . . . . 217 tabular method . . . . . . . . . . . . . . . . . 163
hyperbolic angle . . . . . . . . . . . . . . . . . . . . 230 interval of convergence . . . . . . . . . . . . . 303
hyperbolic cosine . . . . . . . . . . . . . . . . . . . 230 inverse function . . . . . . . . . . . . . . . . . . . . . 38
hyperbolic functions . . . . . . . . . . . 230, 231 inverse hyperbolic functions . . . . . . . 234
inverses . . . . . . . . . . . . . . . . . . . . . . . . 234 inverse trigonometric functions . . . . . 41
hyperbolic rotation . . . . . . . . . . . . . . . . . 238 involute of circle . . . . . . . . . . . . . . . . . . . . 244
hyperbolic sector . . . . . . . . . . . . . . . . . . . 230 irrational numbers. . . . . . . . . . . . . . . . . . . .5
hyperbolic sine . . . . . . . . . . . . . . . . . . . . . 231 J
Jacobian elliptic functions . . . . . . . . . .270
I jerk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
ideal gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
jump discontinuity . . . . . . . . . . . . . . . . . . 73
implicit differentiation . . . . . . . . . . . . . . 79
improper integral . . . . . . . . . . . . . . . . . . 152 L
indefinite integral . . . . . . . . . . . . . . . . . . 125 L’Hôpital’s Rule . . . . . . . . . . . . . . . . . . . . . . 69
indeterminate form . . . . . . . . . . . . . . 69, 75 Laguerre polynomials . . . . . . . . . . . . . . 201
induction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 lamina . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
infinite series . . . . . . . . . . . . . . . . . . . . 4, 289 Langevin function . . . . . . . . . . . . . . . . . . 238
convergent . . . . . . . . . . . . . . . . . . . . . 289 Laplace transform . . . . . . . . . . . . . . . . . . 237
divergent . . . . . . . . . . . . . . . . . . . . . . . 289 latus rectum . . . . . . . . . . . . . . . . . . .209, 223
partial sum . . . . . . . . . . . . . . . . . . . . . 289 of ellipse . . . . . . . . . . . . . . . . . . . . . . . . 209
sum. . . . . . . . . . . . . . . . . . . . . . . . . . . . .289 of parabola . . . . . . . . . . . . . . . . . . . . . 215
infinitesimal . . . . . . . . . . . . . . . . . . . . . . . . . 15 left continuous . . . . . . . . . . . . . . . . . . . . . . . 73
infinitesimal function . . . . . . . . . . . . . . . .72 Legendre polynomials . . . . . . . . . . . . . . 201
336 Index
Leibniz integral rule . . . . . . . . . . . . . . . 185 natural logarithm . . . . . . . . . . . . . . . . . . . 47

Leibniz notation . . . . . . . . . . . . . . . . . . . . . 15 natural numbers . . . . . . . . . . . . . . . . . . . . . . 4
limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3, 62 Newton’s method . . . . . . . . . . . . . . . . . . . 109
infinite . . . . . . . . . . . . . . . . . . . . . . . . . . .66 norm of a partition . . . . . . . . . . . . . . . . . 135
left . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 normal distribution . . . . . . . . . . . . . . . . 284
one-sided . . . . . . . . . . . . . . . . . . . . . . . . 64 normal line . . . . . . . . . . . . . . . . . . . . . . . . . . 60
right . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Nth Derivative Test . . . . . . . . . . . . . . . . .106
Limit Comparison Test . . . . . . . . . . . . . 294 numerical integration . . . . . . . . . . . . . . 195
limits of integration . . . . . . . . . . . . . . . . 139
line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 O
normal . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 O notation . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
tangent . . . . . . . . . . . . . . . . . . . . . . . . . . 56 objective function . . . . . . . . . . . . . . . . . . . . 90
linear combination. . . . . . . . . . . . . . . . . . .24 odd function . . . . . . . . . . . . . . . . . . . . 13, 143
linear function . . . . . . . . . . . . . . . . . . . . . . . . 9 one-to-one . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
linear operator. . . . . . . . . . . . . . . . . . . . . . .24 open interval . . . . . . . . . . . . . . . . . . . . . . . . 91
logarithmic differentiation . . . . . . .49, 86 optimization . . . . . . . . . . . . . . . . . . . . . . . . . 90
Lorentz factor . . . . . . . . . . . . . . . . . . . . . . . 72 P
Lorentz transformation . . . . . . . . . . . . 237 p-series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
M p-series Test . . . . . . . . . . . . . . . . . . . . . . . . 294
Mach number . . . . . . . . . . . . . . . . . . . . . . . . 89 Padé approximation . . . . . . . . . . . . . . . . 314
Maclaurin’s series . . . . . . . . . . . . . . . . . . 309 parabola . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
magnitude . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 axis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
major auxiliary circle . . . . . . . . . . . . . . .243 construction . . . . . . . . . . . . . . . . . . . . 210
maximum . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 definition . . . . . . . . . . . . . . . . . . . . . . . 210
Maxwell speed distribution . . . . . . . . 166 directrix . . . . . . . . . . . . . . . . . . . . . . . . 210
Mean Value Theorem. . . . . . . . . . . . . . .118 eccentricity . . . . . . . . . . . . . . . . . . . . . 210
alternative form . . . . . . . . . . . . . . . . 121 focal radius . . . . . . . . . . . . . . . . . . . . . 212
Extended . . . . . . . . . . . . . . . . . . . . . . . 121 focus . . . . . . . . . . . . . . . . . . . . . . . . . . . .210
median of triangle . . . . . . . . . . . . . . . . . . 285 latus rectum . . . . . . . . . . . . . . . . . . . .215
method of exhaustion . . . . . . . . . . . . . . . . . 6 reflection property. . . . . . . . . . . . . .212
Microstraightness Property . . . . . . . . . 17 vertex . . . . . . . . . . . . . . . . . . . . . . . . . . 210
minimum . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 paraboloid . . . . . . . . . . . . . . . . . . . . . . . . . . 212
minor auxiliary circle . . . . . . . . . . . . . . 243 parameter . . . . . . . . . . . . . . . . . . . . . . . . . . 239
modulated wave . . . . . . . . . . . . . . . . . . . . 167 parametric equations. . . . . . . . . . . . . . .239
moment . . . . . . . . . . . . . . . . . . . . . . . 277, 278 parametrization of a curve . . . . . . . . . 239
Monotone Bounded Test . . . . . . . . . . . . 293 partial fractions . . . . . . . . . . . . . . . . . . . . 178
Monte Carlo integration . . . . . . . . . . . .255 partition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Monte Carlo method . . . . . . . . . . . . . . . 260 norm . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
periodic waveform . . . . . . . . . . . . . . . . . . 261
N Planck’s constant . . . . . . . . . . . . . . . . . . . . 36
nappe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .221 plane curve . . . . . . . . . . . . . . . . . . . . . . . . . 239
Index 337
point of tangency . . . . . . . . . . . . . . . . . . . . 56 hyperbola. . . . . . . . . . . . . . . . . . . . . . .220

polar axis . . . . . . . . . . . . . . . . . . . . . . . . . . . 245 parabola . . . . . . . . . . . . . . . . . . . . . . . . 212
polar coordinates . . . . . . . . . . . . . . . . . . . 245 regular polygon . . . . . . . . . . . . . . . . . . . . . . . 6
derivatives . . . . . . . . . . . . . . . . . . . . . 248 related rates . . . . . . . . . . . . . . . . . . . . . . . . . 82
pop. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34 relativistic momentum . . . . . . . . . . . . . 285
Power Formula . . . . . . . . . . . . . . . . . . . . . 127 relativistic velocity addition law . . . 123
Power Rule . . . . . . . . . . . . . . . . . . . . . . . . . . 24 reliability . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
power series . . . . . . . . . . . . . . . . . . . . . 4, 303 Riemann sum . . . . . . . . . . . . . . . . . . . . . . 134
differentiation . . . . . . . . . . . . . . . . . . 304 Riemann’s Rearrangement Theorem
integration . . . . . . . . . . . . . . . . . . . . . 304 302
interval of convergence . . . . . . . . . 303 right continuous . . . . . . . . . . . . . . . . . . . . . 73
radius of convergence . . . . . . . . . . 303 Rolle’s Theorem . . . . . . . . . . . . . . . . . . . . 119
principal axis . . . . . . . . . . . . . . . . . . . . . . . 203 root mean square . . . . . . . . . . . . . . . . . . . 261
of an ellipse . . . . . . . . . . . . . . . . . . . . 203 root of function . . . . . . . . . . . . . . . . . . . . . 109
probability. . . . . . . . . . . . . . . . . . . . . . . . . .282 rotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . .226
probability density function . . . . . . . . 282 rotational velocity . . . . . . . . . . . . . . . . . . . 31
Product Rule. . . . . . . . . . . . . . . . . . . . . . . . .21
product-to-sum formulas . . . . . . . . . . . 167 S
proof by induction . . . . . . . . . . . . . . . . . . . 24 scalar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
Schrödinger equation . . . . . . . . . . . . . . . . 36
Q Schwarzchild radius . . . . . . . . . . . . . . . .101
q-derivative. . . . . . . . . . . . . . . . . . . . . . . . . .89
secant line . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
q-differential . . . . . . . . . . . . . . . . . . . . . . . . 89
secant method . . . . . . . . . . . . . . . . . . . . . . 112
quantum calculus. . . . . . . . . . . . . . . . . . . .89
Second Derivative Test . . . . . . . . . . . . . . 91
quicksort algorithm . . . . . . . . . . . . . . . . 139
second-degree equation . . . . . . . . . . . . 227
Quotient Rule . . . . . . . . . . . . . . . . . . . . . . . 21
sequence . . . . . . . . . . . . . . . . . . . . 4, 286, 287
R Comparison Test . . . . . . . . . . . . . . . 293
radioactive decay . . . . . . . . . . . . . . . . . . . . 50 convergent . . . . . . . . . . . . . . . . . . . . . 287
radius of convergence . . . . . . . . . . . . . . 303 divergent . . . . . . . . . . . . . . . . . . . . . . . 287
random variable . . . . . . . . . . . . . . . . . . . . 282 Fibonacci . . . . . . . . . . . . . . . . . . . . . . . 288
continuous . . . . . . . . . . . . . . . . . . . . . 282 limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
discrete . . . . . . . . . . . . . . . . . . . . . . . . . 282 Monotone Bounded Test . . . . . . . .293
range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 series . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4, 286
Ratio Test . . . . . . . . . . . . . . . . . . . . . . . . . . 294 absolutely convergent . . . . . . . . . . 301
rational curve . . . . . . . . . . . . . . . . . . . . . . 244 alternating . . . . . . . . . . . . . . . . . . . . . 300
rational function . . . . . . . . . . . . . . . .70, 178 conditional convergent . . . . . . . . . 301
rational numbers . . . . . . . . . . . . . . . . . . . . . 4 harmonic . . . . . . . . . . . . . . . . . . 296, 300
real numbers. . . . . . . . . . . . . . . . . . . . . . . . . .4 infinite . . . . . . . . . . . . . . . . . . . . . . . . . 289
recurrence relation . . . . . . . . . . . . . . . . . 288 Integral Test . . . . . . . . . . . . . . . . . . . 294
reflection property . . . . . . . . . . . . . . . . . .206 Limit Comparison Test . . . . . . . . . 294
ellipse . . . . . . . . . . . . . . . . . . . . . . . . . . 206 n-th Term Test . . . . . . . . . . . . . . . . . 294
338 Index
p-series Test . . . . . . . . . . . . . . . . . . . . 294 torus. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .276

Ratio Test . . . . . . . . . . . . . . . . . . . . . . 294 trajectory envelope . . . . . . . . . . . . . . . . . 213
rearrangement . . . . . . . . . . . . . . . . . 302 translation . . . . . . . . . . . . . . . . . . . . . . . . . 224
telescoping . . . . . . . . . . . . . . . . . . . . . 298 transverse axis . . . . . . . . . . . . . . . . . . . . . 217
Telescoping Series Test. . . . . . . . .294 trapezoid rule . . . . . . . . . . . . . . . . . . . . . . 197
shell method. . . . . . . . . . . . . . . . . . . . . . . .274 trigonometric functions . . . . . . . . . . 23, 41
sigma notation . . . . . . . . . . . . . . . . . . . . . 135 inverses. . . . . . . . . . . . . . . . . . . . . . . . . .41
sigmoid neuron function . . . . . . . . . . . . . 52 trigonometric integrals . . . . . . . . 128, 167
Simpson’s rule . . . . . . . . . . . . . . . . . . . . . . 197 trigonometric substitution . . . . . . . . . 172
slope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
slope of a curve . . . . . . . . . . . . . . . . . . . . . . 58 U
smooth curve . . . . . . . . . . . . . . . . . . . . . . . . 18 unit hyperbola . . . . . . . . . . . . . . . . . . . . . .230
snap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 V
Snell’s Law . . . . . . . . . . . . . . . . . . . . . . . . . 101 vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .281
solid of revolution . . . . . . . . . . . . . . . . . . 273 velocity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2
disc method . . . . . . . . . . . . . . . . . . . . 273 average . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2, 34 instantaneous. . . . . . . . . . . . . . . . . . . . .2
Squeeze Theorem . . . . . . . . . . . . . . . . . . . . 71 vertical asymptote . . . . . . . . . . . . . . . . . . . 66
stationary point . . . . . . . . . . . . . . . . . . . . . 91
step function . . . . . . . . . . . . . . . . . . . . . . . . . 74 W
straight line motion . . . . . . . . . . 1, 33, 129 Wallis’ formula . . . . . . . . . . . . . . . . . . . . . 299
substitution . . . . . . . . . . . . . . . . . . . . . . . . 146 work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
half-angle . . . . . . . . . . . . . . . . . . . . . . 190
trigonometric . . . . . . . . . . . . . . . . . . . 172 Z
substitution method . . . . . . . . . . . . . . . . 146 Zeno’s motion paradox . . . . . . . . 286, 290
Sum Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
surface of revolution . . . . . . . . . . . . . . . .271
T
tabular integration . . . . . . . . . . . . . . . . . 163
tangent line. . . . . . . . . . . . . . . . . . . . . . . . . .56
angle of . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Taylor polynomial . . . . . . . . . . . . . . . . . . 312
Taylor’s formula . . . . . . . . . . . . . . . . . . . . 308
Taylor’s series . . . . . . . . . . . . . . . . . . . . . . 308
Remainder Theorem . . . . . . . . . . . 313
telescoping series . . . . . . . . . . . . . . . . . . .298
Telescoping Series Test . . . . . . . . . . . . .294
Thales’ Theorem . . . . . . . . . . . . . . . . . . . . . 18
time dilation . . . . . . . . . . . . . . . . . . . . . . . . . 72
torque . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277

Elementary Calculus

Uploaded by

Copyright:

Available Formats

Elementary Calculus

Uploaded by

Document Information

Original Title

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

Elementary Calculus

Uploaded by

Copyright:

Available Formats

Elementary

Copyright © 2020 Michael Corral.

encounter. Facility with using named constants (e.g. c, h, T ) is also emphasized.

describing A, B and C would be “Easy”, “Moderate” and “Challenging”, respectively. However,

Schoolcraft College M ICHAEL C ORRAL

2 Derivatives of Common Functions 37

3 Topics in Differential Calculus 56

5 The Integral 124

5.2 The Definite Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132

6 Methods of Integration 159

7 Analytic Geometry and Plane Curves 202

8 Applications of Integrals 252

9 Infinite Sequences and Series 286

Appendix A Answers and Hints to Selected Exercises 315

GNU Free Documentation License 321

Letters Name Letters Name Letters Name

Symbol Meaning Example

Figure 1.1.1 An object dropped from 100 ft above the ground

distance traveled 100 ft

and its average velocity in that time is

change in position final position − initial position 0 ft − 100 ft

1. Find the average velocity over an interval of time.

The average velocity of the object over the interval [t, t + ∆ t] is ∆s

−16(t + ∆ t)2 + 100 − (−16t2 + 100)

−16t2 − 32t∆ t − 16(∆ t)2 + 100 + 16t2 − 100

−32t∆ t − 16(∆ t)2 ∆✚t (−32t − 16∆ t)

instantaneous velocity at t = limit of average velocity over [t, t + ∆ t] as ∆ t approaches to 0

= lim (−32t − 16∆ t)

Z = the set of all integers: 0, ±1, ±2, ±3, ±4, . . .

York: McGraw-Hill Book Company, 1967.

Figure 1.1.4 Inscribed square Figure 1.1.5 Inscribed regular hexagon

method, together with circumscribed regular polygons, to calculate π.

Step 4: Continue this procedure indefinitely, with each successive polygon

1.2 The Derivative: Limit Approach

The derivative of a real-valued function f (x), denoted by f ′ (x), is

for x in the domain of f , provided that the limit exists.11

The derivative of any constant function is 0.

Figure 1.2.1 Slopes and derivatives of lines

Notice that the instantaneous rate of change f ′ (2) = − 41 in the y

that is, f (x) is decreasing in value at x = 2. This is plain to see from

if f (x) approaches L as x approaches a.

f (x) lim f (x)

Another formulation is to set h = w − x in formula (1.4), which yields

Suppose that f ′ (x) exists. Then

Now use the idea that a − b = (a − c) + (c − b) for all a, b, and c to write:

f (x + h) − f (x − h) ( f (x + h) − f (x)) + ( f (x) − f (x − h))

As an example of using these different formulations, recall that a function f is even if

The derivative of an even function is an odd function.

To prove the first statement—the second is an exercise—suppose that f is an even function

f ′ (− x) = − f ′ (x) by formula (1.6),

which shows that f ′ is an odd function.

If the derivative f ′ (x) exists then f is differentiable at x. A differentiable function is

f ( x + 2h) − f ( x − 3h) f ( x + ah) − f ( x − bh)

1.3 The Derivative: Infinitesimal Approach

A number δ is an infinitesimal if the conditions (a)-(d) hold:

(b) if δ > 0 then δ is smaller than any positive real number

(c) if δ < 0 then δ is larger than any negative real number

Figure 1.3.2 Circle O: BC = 2 sin dx, ∠ BOC = 2 ∠ B AC

Since dx is an infinitesimal, (dx)2 = 0. So since sin2 dx + cos2 dx = 1, then: