Vous êtes sur la page 1sur 394

^Tf.

't.-^^fr-

15317
SCHAVM'S OVTLIPiE OF

THEORY AXD PROBLEMS


of

ADVABf CED
C A L C IJ LIT S
;P!972

BY

MURRAY

R.

SPIEGEL, Ph.D.

Professor of Mathematics

Rensselaer Polytechnic Institute

^^Q V^* """

SCHAVM'S OUTLINE SERIES


McGRAW-HILL BOOK COMPANY
New

York, St. Louis, San Francisco, Toronto, Sydney

Copyright 1963 by McGraw-Hill, Inc. All Rights Reserved. Printed in the


United States of America. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means;
electronic, mechanical, photocopying, recording, or otherwise, without the
prior written permission of the publisher.
60229

T8910 SHSH 7B4S210

Preface
The
people.

commonly

subject

To some

it

called

"Advanced Calculus" means differeijt things to different


from an advanced viewpoint, i.e.

essentially represents elementary calculus

with rigorous statements and proofs of theorems.

To

others

it

represents a variety of special

advanced topics which are considered important but which cannot be covered

in

an elementary

course.

In this book an effort has been made to adopt a reasonable compromise between these
extreme approaches which, it is believed, will serve a variety of individuals. The early chapters
of the book serve in general to review and extend fundamental concepts already presented
in
elementary calculus. This should be valuable to those who have forgotten some of the calculus
studied previously and who need "a bit of refreshing". It may also serve to provide
a common
background for students who have been given different types of courses in elementary calculus.
Later chapters serve to present special advanced topics which are fundamental to the
scientist,

engineer and mathematician

if

he

is

to

become

proficient in his intended field.

This book has been designed for use either as a supplement to all current standard textbooks or as a textbook for a formal course in advanced calculus. It should also prove useful
to students taking courses in physics, engineering or any of the
numerous other fields in which
advanced mathematical methods are employed.

Each chapter begins with a clear statement of pertinent definitions, principles and theorems
together with illustrative and other descriptive material. This is followed by graded
sets of solved
and supplementary problems. The solved problems serve to illustrate and amplify the theory,
bring into sharp focus those fine points without which the student continually feels
himself on
unsafe ground, and provide the repetition of basic principles so vital to effective
learning.

Numerous proofs of theorems and derivations of basic results are included among the solved
The large number of supplementary problems with answers serve as a complete

problems.

review of the material of each chapter.

Topics covered include the differential and integral calculus of functions of one or more
and their applications. Vector methods, which lend themselves so readily to concise
notation and to geometric and physical interpretations, are introduced early and used
whenever
they can contribute to motivation and understanding. Special topics include
line and surface
variables

and integral theorems, infinite series, improper integrals, gamma and beta functions,
and Fourier series. Added features are the chapters on Fourier integrals, elliptic integrals and
functions of a complex variable which should prove extremely useful in the
study of advanced
engineering, physics and mathematics.
integrals

Considerably more material has been included here than can be covered in most courses.
This has been done to make the book more flexible, to provide a more useful book
of reference

and

to stimulate further interest in the topics.


I

wish to take

this

opportunity to thank the staff of the

their splendid cooperation in

Schaum Publishing Company

meeting the seemingly endless attempts

at perfection

M. R. Spiegel
Rensselaer Polytechnic Institute

December, 1962

for

by the author.

CONTENTS
Page
Chapter

NUMBERS

Real numbers. Decimal representation of real numbers. Geometric


representation of real numbers. Operations with real numbers. Inequalities. Absolute value of real numbers. Exponents and roots. Logarithms.
Axiomatic foundations of the real number system. Point sets. Intervals.
Countability. Neighborhoods. Limit points. Bounds. Weierstrass-Bolzano
theorem. Algebraic and transcendental numbers. The complex number
system. Polar form of complex numbers. Mathematical induction.
Sets.

Chapter

FUNCTIONS, LIMITS AND CONTINUITY

20

Functions. Graph of a function. Bounded functions. Monotonic functions. Inverse functions. Principal values. Maxima and minima. Types
of functions.
Special transcendental functions.
Limits of functions.

Right and

left

Continuity.

hand

left

Theorems on continuity.

Chapter

Theorems on limits. Infinity. Special limits.


hand continuity. Continuity in an interval.
Sectional continuity. Uniform continuity.

limits.

Right and

SEQUENCES

41

Definition of a sequence. Limit of a sequence. Theorems on limits of


sequences. Infinity. Bounded, monotonic sequences. Least upper bound

and greatest lower bound of a sequence. Limit superior. Limit inferior.


Nested intervals. Cauchy's convergence criterion. Infinite series.

Chapter

DERIVATIVES

57

Definition of a derivative.
tiability

in

an

interval.

Right and
Sectional

left

hand

derivatives.

differentiability.

Differen-

Graphical

inter-

pretation of the derivative.


Differentials.
Rules for differentiation.
Derivatives of special functions. Higher order derivatives. Mean value
theorems. Rolle's theorem. The theorem of the mean. Cauchy's generalized theorem of the mean. Taylor's theorem of the mean. Special
expansions. L'Hospital's rules. Applications.

Chapter

INTEGRALS
Definition of a definite integral.
integrals.
Mean value theorems

80

Measure
for

zero.

integrals.

Properties of definite
Indefinite

integrals.

Fundamental theorem of integral calculus. Definite integrals with


variable limits of integration.
Change of variable of integration.
Integrals of special functions. Special methods of integration. ImpVoper
integrals.
Numerical methods for evaluating definite integrals. Applications.

CONTENTS
Page
Chapter

PARTIAL DERIVATIVES

101

Dependent and independent


Three dimensional rectangular coordinate systems. Neighborhoods. Regions. Limits. Iterated limits. Continuity. Uniform continuity. Partial derivatives. Higher order partial
Theorems on differentials. Differentiation
Differentials.
derivatives.
Euler's theorem on homogeneous functions.
of composite functions.
Functions

Implicit

of

two

Domain

variables.

or

more

functions.

Jacobians.

Theorems on Jacobians.

Mean

Chapter

variables.

of a function.

Partial

derivatives using Jacobians.


Curvilinear coordinates.

Transformations.

value theorems.

VECTORS

134

Laws

Unit
Vectors and scalars.
Components of a vector. Dot or
Rectangular unit vectors.
vectors.
scalar product. Cross or vector product. Triple products. Axiomatic
approach to vector analysis. Vector functions. Limits, continuity and
derivatives of vector functions. Geometric interpretation of a vector
Formulas involving V.
Gradient, divergence and curl.
derivative.
Vector algebra.

of vector

algebra.

Vector interpretation of Jacobians. Orthogonal curvilinear coordinates.


Gradient, divergence, curl and Laplacian in orthogonal curvilinear coordinates.

Chapter

Special curvilinear coordinates.

APPLICATIONS OF PARTIAL DERIVATIVES

161

Applications to geometry. Tangent plane to a surface. Normal line to


a surface. Tangent line to a curve. Normal plane to a curve. Envelopes.
Directional derivatives. Differentiation under the integral sign. Maxima
and minima. Method of Lagrange multipliers for maxima and minima.
Applications to errors.

Chapter

MULTIPLE INTEGRALS
Double integrals. Iterated integrals.

180
Triple integrals.

Transformations

of multiple integrals.

Chapter

10

LINE INTEGRALS, SURFACE INTEGRALS AND

INTEGRAL THEOREMS

195
Evaluation of line

Vector notation for line integrals.


Properties of line integrals. Simple closed curves. Simply
and multiply-connected regions. Green's theorem in the plane. Conditions for a line integral to be independent of the path. Surface integrals.
The divergence theorem. Stokes' theorem.
Line integrals.

integrals.

Chapter

11

INFINITE SERIES

224

Convergence and divergence of infinite series. Fundamental facts concerning infinite series. Special series. Geometric series. The p series.
Tests for convergence and divergence of series of constants. Comparison
Quotient test. Integral test. Alternating series test. Absolute
test.
and conditional convergence. Ratio test. The nth root test. Raabe's
Gauss' test. Theorems on absolutely convergent series. Infinite
test.
sequences and series of functions. Uniform convergence. Special tests
test. Dirichlet's test.
for uniform convergence of series. Weierstrass
Theorems on uniformly convergent series. Power series. Theorems on

CONTENTS
Page

power series. Operations with power series. Expansion of functions


in power series. Some important power series. Special topics.
Functions defined by series. Bessel and hypergeometric functions. Infinite
series of complex terms. Infinite series of functions of two
(or more)
variables.

Double

series.

Infinite products.

Summability.

Asymptotic

series.

Chapter

12

IMPROPER INTEGRALS

260

Definition of an improper integral. Improper integrals of the first kind.


Special improper integrals of the first kind. Geomewic or exponential
integral. The p integral of the first kind. Convergence tests
for

im-

proper integrals of the first kind. Comparison test. Quotient test.


Series test. Absolute and conditional convergence. Improper integrals
of the second kind. Cauchy principal value. Special improper integrals
of the second kind. Convergence tests for improper integrals of the
second kind. Improper integrals of the third kind. Improper integrals
containing a parameter. Uniform convergence. Special tests for uniform
convergence of integrals. Weierstrass
test. Dirichlet's test. Theorems
on uniformly convergent integrals. Evaluation of definite integrals.
Laplace transforms. Improper multiple integrals.

Chapter

13

GAMMA AND BETA FUNCTIONS


Gamma

285

Table of values and graph of the gamma function.


Asymptotic formula for r{n). Miscellaneous results involving the gamma
function. Beta function. Dirichlet integrals.

Chapter

14

function.

FOURIER SERIES

298

Periodic functions. Fourier series. Dirichlet conditions. Odd and even


functions. Half range Fourier sine or cosine series. Parseval's identity.
Differentiation and integration of Fourier series. Complex notation for

Fourier

Chapter

15

series.

Boundary-value problems.

FOURIER INTEGRALS
The Fourier

integral.

Fourier transforms.
convolution theorem.

Chapter

16

Orthogonal functions.

321

Equivalent forms of Fourier's integral theorem.


Parseval's identities for Fourier integrals. The

ELLIPTIC INTEGRALS
The incomplete

331
the first kind. The incomplete elliptic
incomplete elliptic integral of the third

elliptic integral of

integral of the second kind. The


Jacobi's forms for the elliptic integrals. Integrals reducible to
elliptic type. Jacobi's elliptic functions. Landen's transformation.

kind.

Chapter

17

FUNCTIONS OF A COMPLEX VARIABLE

345

Limits and continuity. Derivatives. Cauchy-Riemann equations. Integrals. Cauchy's theorem. Cauchy's integral formulas. Taylor's series. Singular points. Poles. Laurent's series. Residues. Residue
theorem. Evaluation of definite integrals.
Functions.

INDEX

373

chapter

Numbers
SETS
Fundamental in mathematics is the concept of a set, class or collection of objects
having specified characteristics. For example we speak of the set of all university
professors, the set of all letters A,B,C,D, .. .,Z of the English alphabet,
etc.
The individual
objects of the set are called members or elements. Any part of a set is called
a subset of
the given set, e.g. A, B,C is a subset of A,B,C,D,..., Z. The set consisting
of no elements
is

called the

empty

set or null set.

REAL NUMBERS
The following types of numbers are already familiar
1.

Natural numbers

to the student.

1,2, 3, 4,
., also called positive integers, are used in counting
of a set.
The symbols varied with the times, e.g. the Romans used
III, IV, ....
The sum a + b and product a-b or ab of any two natural num.

members
I, II,

bers a and b is also a natural number. This is often expressed by saying that
the set of natural numbers is closed under the operations of addition and
multiplication, or satisfies the closure property with respect to these operations.
2.

Negative integers and zero denoted by -1, -2, -3,


and
respectively, arose
to permit solutions of equations such as x + b - a where a and b are
any natural
numbers. This leads to the operation of subtraction, or inverse of addition, and
we write x = a b.
.

The
3.

set of positive

and negative integers and zero

Rational numbers or fractions


equations such as bx = a for all
operation of division, or inverse
where a is the numerator and b

is

called the set of integers.

such as f -f,
arose to permit solutions of
integers a and b where b v^ 0. This leads to the
of multiplication, and we write x-a/b or a-^b
the denominator.
,

The set of integers is a subset of the rational numbers, since integers correspond to rational numbers where & = 1.
4.

numbers such as \/2 and tt are numbers which are not rational, i.e.
cannot be expressed as % (called the quotient of a and b) where a and b are integers

Irrational

and b^Q.

The

set of rational

and irrational numbers

DECIMAL REPRESENTATION

of

is called

the set of real numbers.

REAL NUMBERS

Any real number can be expressed in decimal form, e.g. 17/10 = 1.7, 9/100 = 0.09,
1/6 = 0.16666 .... In the case of a rational number the decimal expansion either terminates
or, if it does not terminate, one or a group of digits in the
expansion will ultimately
repeat as, for example, in i = 0.142857 142857 142
In the case of an irrational number such as V2 = 1.41423. .. or ,7 = 3.14159... no such repetition can occur.
We can
always consider a decimal expansion as unending, e.g. 1.375 is the same as 1.37500000.
or 1.3749999
To indicate recurring decimals we sometimes place dots over the ree.g. \ = 0.142857, ^ = 3.16.
The decimal system uses the ten digits 0, 1,2,
It is possible to design number
., 9.
systems with fewer or more digits, e.g. the binary system uses only two digits
and 1

peating cycle of digits,

(see

Problems 32 and

33).

NUMBERS

GEOMETRIC REPRESENTATION

of

[CHAP.

REAL NUMBERS

numbers as points on a line called the real axis,


For each real number there
as in the figure below, is also well known to the student.
corresponds one and only one point on the line and conversely, i.e. there is a one to one
(1-1) correspondence between the set of real numbers and the set of points on the line.
Because of this we often use point and number interchangeably.

The geometric representation

of real

-4-3-2-10

-5

in

Fig. 1-1

The

is called the set of positive numbers; the


to the right of
itself is neither positive nor
the set of negative numbers, while

numbers

set of real

set to the left of

is

negative.

Between any two rational numbers


infinitely

many

(or irrational)

numbers an everywhere dense

OPERATIONS

with

REAL NUMBERS
R of real
belong to R

If a, b, c belong to the set


1.

2.
3.
4.
5.
6.

7.

(or irrational

rational (and irrational) numbers.

is

call

line there are


the set of rational

set.

numbers, then:

+ b and ab
a+b = b+a
a + {b + c) = {a + b) + c
ab = ba
a(bc) (ab)c
a{b + c) ab + ac
= + a = a, l'a = a*l =
a +
a

numbers) on the

This leads us to

Closure law

Commutative law of addition


Associative law of addition
Commutative law of multiplication
Associative law of multiplication
Distributive law

a
to addition, 1
with
respect
the
identity
called

the identity with

is called

respect to multiplication.
8.

X
9.

set,

a number x in R such that x + a 0.


called the inverse of a with respect to addition and

For any a there


is

is

is

denoted by a.

is a number x in ^ such that ax 1.


X is called the inverse of a with respect to multiplication and
c~^ or 1/a.

For any ut-O there

These enable us to operate according to the usual rules of algebra.


such as R, whose members satisfy the above is called a field.

is

denoted by

In general any

INEQUALITIES
If a 6 is a positive number we

say that a is greater than & or b is less than a and


write respectively a > & or & < a. If the possibility that a = b also exists, we write a S b
or b g a. Geometrically, a > b if the point on the real axis corresponding to a lies to the
right of the point corresponding to b.
Examples: 3 < 5 or 5 > 3; 2 <
be 3 or less than 3.
If a, b
1.

2.
3.
4.
5.

and

or

1 >

2;

a;

means that

a;

is

a real number which may-

are any given real numbers, then:

Either a> b, a = b or a<b


If a > b and b > c, then a> c
If a > b, then a + c > b + c
If a > b and c > 0, then ac > be
If a > b and c < 0, then ac < be

Law of trichotomy
Law of transitivity

CHAP.

NUMBERS

1]

ABSOLUTE VALUE

of

REAL NUMBERS

The absolute value of a

a<0, and

if

Examples:
1.

\ah\

2.

|a

+ 5|

\a-b\

3.

real

number

|-||

a,

|+2|

5,

\a\

+
-

|,

|-V2|

\ahc...m\

or

l&l

=
|a + 5 + c +

or

l&l
|a|

2,

\h

^2,

|a] |&| |c|

defined as a

is

|a|,

|o|

+m| S

if

a>

0,

-a

if

o.

|w|
|a|

+ |6| +

|c|

+ |m|

\h\

The distance between any two points


\a-h\

denoted by

= 0.
|-5i

|a|

numbers) a and h on the real axis

(real

a\.

is

EXPONENTS

and ROOTS
The product a-a. .a of a
called the exponent and a is
.

is

1.

These and extensions

to

a"

any

real

-a''

real

cluded.

number a by

called the base.

In particular by using
definitions a" = 1, a~ = l/a".

p times

itself

The following

0"+'

3.

(a")''

is

denoted by a" where p

rules hold.

a"''

numbers are possible so long as division by zero is ex2, with p=^q and p =
respectively, we are led to the

= N, where p is a positive integer, we call a a pth root of N, written \^. There


be more than one real pth root of A^. For example since 2^ = 4 and (-2)^ =
4, there
are two real square roots of 4, namely 2 and -2. It is customary to denote the
positive
square root by \/4 = 2 and the negative one by -^=-2.
If a"

may

If

p and q are

we

positive integers,

define

a''^"

= \^.

LOGARITHMS
If

a''

= N,p

is

called the logarithm of A^ to the base a, written p = logaN. If a and


1, there is only one real value for p.
The following rules hold.

are positive and a ^^


1.

lOgaMA^

= logaM + logaA^
3.

logaAf

2.

loga

= logaM-logaiV

= rlogaM

In practice two bases are used, the Briggsian system uses base a
tem uses the natural base a = e = 2.71828 ....

AXIOMATIC FOUNDATIONS
The number system can be

of the

= 10,

the Napierian sys-

REAL NUMBER SYSTEM

up

logically, starting from a basic set of axioms or


from experience, such as statements 1-9, Page 2.
we assume as given the natural numbers and the operations of addition and multi-

built

"self evident" truths, usually taken


If

plication (although it is possible to start even further back with the concept
of sets),
find that statements 1-6, Page 2, with
as the set of natural numbers,

we

hold while 7-9

do not hold.

and

Taking 7 and 8 as additional requirements, we introduce the numbers -1, -2, -3,
Then by taking 9 we introduce the rational numbers.

0.

NUMBERS

[CHAP.

Operations with these newly obtained numbers can be defined by adopting axioms 1-6,
R is now the set of integers. These lead to proofs of statements such as (-2) (-3) = 6,
-(-4) = 4, (0)(5) = 0, etc., which are usually taken for granted in elementary mathematics.

where

can also introduce the concept of order or inequality for integers, and from these
For example if a, b, c, d are positive integers we define
eld if and only if ad > be, with similar extensions to negative integers.

We

inequalities for rational numbers.

a/b

>

Once we have the set of rational numbers and the rules of inequality concerning
them, we can order them geometrically as points on the real axis, as already indicated.
We can then show that there are points on the line which do not represent rational numbers (such as v^,7r,etc.). These irrational numbers can be defined in various ways one
From this we can show that
of which uses the idea of Dedekind cuts (see Problem 34).
the usual rules of algebra apply to irrational numbers and that no further real numbers
are possible.

POINT SETS, INTERVALS

set of points (real

numbers) located on the real axis

is

called a one-dimensional

point set.

The set of points x such that a^x^b is called a closed interval and is denoted by
The set a<x<b is called an open interval, denoted by {a,b). The sets a<x^b
[a,b].
and a g X < b, denoted by {a, b] and [a, b) respectively, are called half open or half closed
intervals.

The symbol x, which can represent any number or point


The given numbers a or & are called constants.
Example:

The

set of all x such that

|x|

<

4,

i.e.

-4 <

a:

< 4,

of a set,

is

is

called a variable.

represented by (-4,4), an open

interval.

The set x>a can also be represented by a<x<cc. Such a set is called an
or unbounded interval. Similarly - < x < represents all real numbers x.

infinite

COUNTABILITY

denumerable
respondence with the natural numbers.
set is called countable or

Example:

The even natural numbers

2, 4, 6, 8,

if

its

is

elements can be placed in 1-1 cor-

a countable set because of the 1-1 correspond-

ence shown.

Given set

Natural numbers

set is infinite if it

infinite set

The
or

which

is

8...

i;

...

can be placed in 1-1 correspondence with a subset of


is called countably infinite.

itself.

An

countable

numbers is countably infinite while the set of irrational numbers


non-countably infinite (see Problems 17-20).

set of rational

all real

numbers

is

set which is
of elements in a set is called its cardinal number.
countably infinite is assigned the cardinal number K^, (the Hebrew letter aleph-null). The
with this set)
set of real numbers (or any sets which can be placed into 1-1 correspondence
the continuum.
is given the cardinal number C, called the cardinality of

The number

CHAP.

NUMBERS

1]

NEIGHBORHOODS
The

set of all points x such that

of the point
is called

The

a.

a deleted

\x-a\<h where 8 > 0,


< - a| < 8

set of all points x such that

neighborhood of

|a;

neighborhood

is

called a

in

which

=a

a;

is

excluded,

a.

LIMIT POINTS

limit point, point of accumulation or cluster point of a set of numbers is a number I


8 neighborhood of I contains members of the set.
In other words

such that every deleted

8 > 0, however small, we can always find a member x of the set which is not equal
but which is such that \x-l\ < 8. By considering smaller and smaller values of 8
see that there must be infinitely many such values of x.

for any
to

we

A
point.

cannot have a limit point.

finite set

Thus the natural numbers have no

has infinitely

many

An

may

infinite set

or

may

not have a limit

limit point while the set of rational

numbers

limit points.

A set containing all its limit points is called a closed set.


not a closed set since, for example, the limit point \/2
(Problem 5). However, the set
x g 1 is a closed set.

The

is

is

set of rational

not a

member

numbers

of the set

BOUNDS
If for all

above and

numbers
is

a;

of a set there

M such

a number
Similarly if

is

an upper bound.

called

that

x^m,

x^M,

the set

is

the set is bounded


bounded below and

m^x^M,

x we have
the set is called bounded.
a number such that no member of the set is greater than
but there is at
least one member which exceeds
for every e > 0, then
is called the least upper
bound (l.u.b.) of the set. Similarly if no member of the set is smaller than
but at least
one member is smaller than
+ e for every >0, then
is called the greatest lower
bound (g.l.b.) of the set.
is

called a lower bound.

If

If for all

is

M-

WEIERSTRASS-BOLZANO THEOREM
The Weierstrass-Bolzano theorem states that every bounded infinite set has at least
one limit point. A proof of this is given in Problem 23, Chapter 3.

ALGEBRAIC and TRANSCENDENTAL NUMBERS


A number x which is a solution to the polynomial
+

aox"

where

ttix"-'

a^x"-^

...

equation

an-ix

{1)

a are integers and n is a positive integer, called the degree of


an algebraic number. A number which cannot be expressed as a
solution of any polynomial equation with integer coefficients is called a transcendental
number.
ao = 0,

the equation,

ai, aa,

is

Examples

called

| and ^2 which are solutions of 3x


numbers.

2 =

and

x^

2 =

respectively, are algebraic

The numbers
and e can be shown to be transcendental numbers.
determine whether some numbers such as e-n- or e + tt are algebraic or not.
-n-

The

numbers is a countably infinite


numbers is non-countably infinite.

set of algebraic

of transcendental

set (see

Problem

We

still

cannot

23) but the set

NUMBERS
The

[CHAP.

COMPLEX NUMBER SYSTEM

Since there is no real number x which satisfies the polynomial equation


or similar equations, the set of complex numbers is introduced.

x^

+1 =

We

can consider a complex number as having the form a + bi where a and b are
called the real and imaginary parts, and i \/ 1 is called the imaginary
unit. Two complex numbers a + bi and c + di are equal if and only if a = c and b = d. We
can consider real numbers as a subset of the set of complex numbers with b 0. The
complex number + Oi corresponds to the real number 0.
real

numbers

The absolute value or modulus


conjugate of a
is

bi is defined as a

of a

bi is defined as |a

bi\

The complex conjugate

bi.

y/aF+~b^.

of the

The com,plex

complex number

often indicated by z or z*.

The set of complex numbers obeys rules 1-9 of Page 2, and thus constitutes a field.
In performing operations with complex numbers we can operate as in the algebra of real
numbers, replacing t^ by 1 when it occurs. Inequalities for complex numbers are not
defined.

From

the point of view of an axiomatic foundation of complex numbers, it is desirable


complex number as an ordered pair {a, b) of real numbers a and b subject to
certain operational rules which turn out to be equivalent to those above. For example, we
define {a,b) + {c,d) = {a + c, b + d), {a,b){c,d) = {ac bd, ad + bc), m{a,b) = {ma,mb),
etc.
We then find that (a, b) a(l, 0) + &(0, 1) and we associate this with a + bi,
where i is the symbol for (0,1).
to treat a

POLAR FORM

of

COMPLEX NUMBERS

chosen on two mutually perpendicular axes X'OX and Y'OY (the x


and y axes) as in Fig. 1-2 below, we can locate any point in the plane determined by these
lines by the ordered pair of numbers {x, y) called rectangular coordinates of the point.
Examples of the location of such points are indicated by P, Q, R, S and T in Fig. 1-2.
If real scales are

Y
*P(3,4)

Q(-3,3)

Pio'.y)

T(2.5, 0)
-t-

R(-2.5, -1.5)
*S(2,

-2)
Y'

Y'

Fig. 1-3

Fig. 1-2

Since a complex number x + iy can be considered as an ordered pair (x, y), we can
represent such numbers by points in an xy plane called the complex plane or Argand
diagram. Referring to Fig. 1-3 above we see that x p cos
y = p sin ^ where
p = \/x^ + y^ = \x+iy\ and ^, called the amplitude or argument, is the angle which line
OP makes with the positive x axis OX. It follows that
<l>,

called the polar


It is

ly

p(cos

(j>

form of the complex number, where

sometimes convenient to write

cis

+
p

sin ^)

and

^ instead of cos

<^

<j>

(2)

are called polar coordinates.


sin 0.

CHAP.

NUMBERS

1]

If

Zi

we can show

Xi

iyi

pj(cos

+ i sin

<j>^

and

<j>^)

Z2

X2

+ iy^ =

Pii'ios

<i>^

sin ^2)

that
Z1Z2

P1P2 { cos (^1

^{cos(</.i-,^2)

3"

{p(cos<^

+ ^2) +

*'

sin

(</>!

+ .^2)}

(S)

*sin(,^j-<^2)}

isin^)}"

p" {cos

kV)
714,

sin

n<j,)

(5)

where n

is any real number.


Equation (5) is sometimes called De Moivre's theorem. We
can use this to determine roots of complex numbers. For example if % is a positive integer,
2*^"

(p(cos^

p-|cos(i^)+^sin(i^)}

isin^)}*''"

{6)

A.

= 0,l,2,3,...,n-1

from which it follows that there are in general n different values for z^"^.
will show that e** = cos
+ i sin ^ where e = 2,71828
This

we

<j!>

Later (Chap. 11)


is

called

Euler's

formula.

MATHEMATICAL INDUCTION
The

principle of mathematical induction

is an important property of the positive


proving statements involving all positive integers when
it is known for example that the statements are valid for w = 1, 2, 3 but it is suspected or
conjectured that they hold for all positive integers. The method of proof consists of the

integers.

It is especially useful in

following steps.
1.

Prove the statement for n =

2.

Assume

3.

the assumption in 2 prove that the statement must be true for n = k + 1.


the part of the proof establishing the induction and may be difficult or
impossible.

integer).

positive integer.

From

This

4.

some other positive


n = k where A; is any

l (or

the statement true for

is

Since the statement is true for n = l [from step 1] it must [from step 3] be true
for w = 1 + 1 = 2 and from this for = 2 + 1 = 3, etc., and so must be true for
all

positive integers.

Solved Problems

OPERATIONS
1.

NUMBERS

with

Ifx = 4,y = 15,z- -3, p=^i,q


(c)

p(gr),

(d) (pq)r,

+ (2/ + z) =
+ y) + z =

(a)

a;

(6)

(x

{c)

p{qr)

(d)

(pq)r

The

(4

=
=

x(v

= -J, and

(6)

and

(d)

= |,

evaluate

(a)

{y

+ z),

(&) {x

+ y) +

z,

+ 12 =

16

19 - 3 = 16
are equal illustrates the associative law of addition,

|{(-i)(|)} = (|)(-^) =
{(f)(-i)}(|) = (-fs)(f) =
(c)

+ q).

+ [IB + (-3)] =
+ 15) + (-3) =

fact that (a) and

The fact that


(e)

(e)

(f)(-i)
(-l)(|)

= -^ = -3L
= -I = -3L

are equal illustrates the associative law of multiplication.

+ g) = 4(1-1) = 4(1-1) = 4(i) = f = 2


Another method: x{p + q) = xp + xq = (4)(f) +

a;(p

distributive law.

(4)(-i)

= |-| =

|-|=f=2

using the

NUMBERS

2.

why we do

Explain
If

we

Ox

= 0.

(a)

However, this

we

represent,
(6)

As

not consider

define a/b as that

in (a), if

is

consider

we

number
it

(a)

(b)

[CHAP.

as numbers.

such that bx a, then 0/0 is that number x such that


numbers. Since there is no unique number which 0/0 can

(if it exists)

true for

all

undefined.

number x

define 1/0 as that

(if it exists)

such that

Oa;

= 1, we

conclude that there

no such number.

is

Because of these facts we must look upon division by zero as meaningless.

3.

Simplify
~2
a;

# 3.

For

RATIONAL
4.

ZTq

=3

a;

and

Qw XT\

XT

the given fraction

provided that the cancelled factor

4rw*

3)

is

not zero,

i.e.

IRRATIONAL NUMBERS

+ 4m =

2(2?n^

+ Im),

is 1

more than the even

the result follows.

no rational number whose square is 2.


Let p/g be a rational number whose square is 2, where we assume that p/g is in lowest terms, i.e.
p and q have no common integer factors except 1 (we sometimes call such integers relatively prime).
Then (p/q)' = 2, p* = 2g* and p^ is even. From Problem 4, p is even since if p were odd, p' would
Prove that there

is

= 2m.
Substituting p = 2m

be odd.

Thus p

in p"

= 2q^

yields q^

Thus p and q have the common factor

common

By

factors other than 1.

whose square
6.

(a;

undefined.

is

Prove that the square of any odd integer is odd.


Any odd integer has the form 2m + 1. Since (2m + 1)^ = 4m* + 4m +
integer

5.

^,_^^_3

= 2m*,

so that q^ is even

and q

even.

is

contradicting the original assumption that they had no


virtue of this contradiction there can be no rational number
2,

is 2.

Show how to find rational numbers whose squares can be made arbitrarily close to 2.
We restrict ourselves to positive rational numbers. Since (1)* = 1 and (2)* = 4, we are led to choose
rational

numbers between 1 and


(1.4)* = 1.96 and (1.5)*

Since

2, e.g. 1.1, 1.2, 1.8, ..., 1.9.

= 2.25, we

consider rational numbers between 1.4 and

1.5, e.g.

1.41,

1.42, ...,1.49.

Continuing
(1.414213562)*

7.

in

is less

manner we can obtain closer and closer rational


than 2 while (1.414213563)* is greater than 2.

Given the equation aox" + aia;""^ + ... + a


and ao and an^O. Show that if the equation

must divide a and q must divide


Since p/q

is

approximations,

this

a root

we

OOP"

where

is to

...,a are integers

ao, ai,

have a rational root p/q, then p

ao exactly.

have, on substituting in the given equation and multiplying by

aip"-'q

a2p''-*q*

or dividing by p,
OOP"-'

e.g.

dip-'-^g

a,-ipg-'

ttn-ig"-'

ag

q",

the result

(1)

-^^

Since the left side of (2) is an integer the right side must also be an integer.
p does not divide g" exactly and so must divide a.

(z)

Then

since

p and g are

relatively prime,

In a similar manner, by transposing the

must
8.

divide

first

term of

{!)

and dividing by

g,

we can show

that g

ao.

Prove that \/2 + a/3 cannot be a rational number.


X = V2 + \/3 then a;* = 5 + 2^6, a;* - 5 = 2V6 and
If

squaring, * - 10a;* +1 = 0. The


only possible rational roots of this equation are 1 by Problem 7, and these do not satisfy the equation.
It follows that Vz + \/S, which satisfies the equation, cannot be a rational number.

CHAP.

9.

NUMBERS

1]

Prove that between any two rational numbers there


If

a and

assume a <

this

5^^ <

<

<

<

6.

a = and

a rational number, let

is

= -

where

A = 1( + I^ = \(^ + TL\ = 5+91


2
2\gs
qsj
2qs
2\g s/

Then

< a+h and a < "^

to both sides, 2a

26 and

6.
.

6.

^r

To prove that

97^0,8^0.

a rational number between a and

is

Then by adding a

6.

Similarly adding 6 to both sides, a

Thus

numbers, then

6 are rational

To prove

another rational number.

is

^^

p, g, r, s are integers

and

rational number.

INEQUALITIES
10.

For what values of


+ 3(2 - x) a 4 a:

11.

is

ic
a;

For what values of

a;

cc

3(2

x) S

a;

6-43

+ 6-3a;g4-a;, 6-2a;g4-a;,

when

a;

is

x^

- Bx -

<

10

- 2x

2aa;,

-a;,

2a;

i.e.

a;

g 2.

The required inequality holds when

a;''-3x-2-10 +

<

2a;

0,

a;'

- a;

12

<

or

(a;

- 4)(a; + 3) <

This last inequality holds only in the following cases.

Case 1
a; - 4 >
and a; + 3 < 0, i.e.
both greater than 4 and less than 3.
:

Case

12. If

>

a;

and x

<

-3.

This

impossible since x cannot be

is

and a; + 3 > 0, i.e. a; < 4 and x > -3. This is possible when -3 <
2: a; - 4 <
Thus the inequality holds for the set of all x such that 3 < a; < 4.

ago
A

and

&

prove that ^{a

0,

method of proof

+ b) g

a;

<

4.

t/oE.

often arrived at by assuming the required result to be true and performing


valid operations until a result is obtained which is known to be true. By reversing the steps (assuming
this possible) the proof follows.
is

In this problem we start with the required result to obtain successively a + 6 S 2^/ab, {a + by g
or a'-2ab + b'' S 0, i.e. (a-bf g 0, which is known to be true. Retracing the steps, the

4ab

result follows.

Another method:

Since

we have a 2^/ab +

(Va-y/bY S

This result can be generalized to

o or

g ^aiUi

A(a+6) ^ -/^,

negative. The left and right sides are called respectively the arithmetic
of the numbers ai,
On.
.

13. If

Ui, a2,

For

all real

6i, bi,

a2b2

numbers

X,

are non-

mean and geometric mean

are any real numbers, prove Schwarz's inequality

.,b

anb^f

(a?

a|

al)(bl

6|

+ &')

we have

(aiX+6i)'

Expanding and

ai,

and

(aibi

where

(aiX

biV

(a\

6)^

collecting terms yields

A'\'

+ 2C\ +

B"

(1)

where

A^

Now

{1)

a?

+ oi +

+ a,

B'

= bl+bl+

+bl,

C =

oi6i

+ 0262 +

+ o6

{S)

can be written
X^

gx

or

(x

+^y

But this last inequality is true for all real X if and only it
gives the required inequality upon using (2).

+ 1; -

-^ - -^ ^
A
A

(S)

or

C^S A^B'

which

NUMBERS

10

14.

111
+ +

Prove that

Let

15.

'

'^

"

'

2r^ -

^^

^+...+^+^

Subtracting,

1S

= i -

-^-

>

1.

+^

^+

IS.

positive integers

^^^

= i + 1 +

Then

EXPONENTS, ROOTS

...

Thus S

g^

==

for

all n.

LOGARITHMS

and

Evaluate each of the following.

=-

(a)

aF^^P^

(c)

log2/3(f)

(log 6)(logb a)

Since

>

ilf

AT

0,

Let log

i^

>

MM
77

X, log

""

^.^^.^ ^

^ ^Z,^^,^ ^
or

(f)"'

a;

-3.

assuming a,b >

2/

5. 10-= or 0.00005

and

a, 6 7^ 1.

b, b"

a>0

and

= f - ()' =
loga b = x, logb a =

= a and m = xy.
(a^) = a^ = 6 = a we have

Then a'=

16. If

(f^

Let

u.

q-2

^W.lO^

Then

X.

Q4 + 8-14

(^)

(d)

N = y.
-

""

_"

= a'

a^l,

but

Then a'-M,

a"

a""

or

=1

prove that

a'^N

the required value.

loga

loga

M - loga

A^.

and so

a;^/

lo&]v

= logaM -

loga AT

COUNTABILITY
17.

"'"

Sn

[CHAP.

Prove that the


Write
i' 4. |.

set of all rational

numbers between

and

1 inclusive is countable.

fractions with denominator 2, then 3,


considering equivalent fractions such as
no more than once. Then the 1-1 correspondence with the natural numbers can be accomall

plished as follows.

Rational numbers

Oliifilif

Natural numbers

123456789...

Thus the set of all rational numbers between


number Xo (see Page 4).

18. If

and

or

are two countable sets, prove that the set consisting of

is

countable, there

Suppose elements of
or

is

or

elements of

are

countable since

a 1-1 correspondence between elements of


elements by ai, aa, as, ...

is

we can denote these


Similarly we can denote the

Case

1 inclusive is countable

all

and has cardinal

elements from

(or both) is also countable.

Since A
bers so that

from

and

...

by

and the natural num-

61, 62, 63,

from elements of B. Then the set consisting of elements


establish the following 1-1 correspondence.

all distinct

we can

Natural numbers

ai

61

^
^

A
V

as

62

az

63

**

^
^

"^

...

...

Case 2: If some elements of A and B are the same, we count them only once as in Problem 17.
the set of elements belonging to A or B (or both) is countable.

Then

CHAP.

NUMBERS

1]

The

and B, denoted by

The set consisting of


and B, denoted by A n
The

A UB

set consisting

Prove that the

or

(or both) is often called the

union of

A+B.

or

and B is called the intersection of


A n B.
of all elements in A but not in B is written A B. If we let B be the set of
in B, we can also write A B = AB. If A and B are countable, so is A B.

elements which are not

19.

which belong to

set consisting of all elements

11

all

elements which are contained in both


AB. If A and B are countable, so

or

set of all positive rational

Consider all rational numbers x >


only one rational number 1/x in (0, 1),

numbers

is

countable.

is

rational number we can associate one and


there is a one to one correspondence between all rational
numbers > 1 and all rational numbers in (0, 1). Since these last are countable by Problem 17, it follows
that the set of all rational numbers > 1 are also countable.

With each such

1.

i.e.

From Problem 18 it then follows that the set consisting of all positive rational numbers is
countable, since this is composed of the two countable sets of rationals between
and 1 and those
greater than or equal to 1.
From

20.

this

we can show that

Prove that the


Every
the digits

real

set of all real

number

0, 1,2,

the set of

in

rational

all

numbers

in

numbers

non-countable.

is

[0, 1]

has a decimal expansion

[0, 1]

countable (see Problem 59).

is

mi aiUs

where

oi, as,

are any of

.,9.

We

assume that numbers whose decimal expansions terminate such as 0.7324 are written
and that this is the same as 0.73239999
If all real numbers in [0, 1] are countable we can place them in 1-1 correspondence with the
natural numbers as in the following list.
0.73240000...

We now

<^

O.aii

an

ais

an

->

O.Uii

a-zz

023 a2i

(r->

O.asi

ttsa

ass 034 ..

form a number
0.6i 62 63 64

where

61

# an,

62

# 022,

63 f^ ass, 64 v^ 044,

and where

all

b's

beyond some position are not

This number, which is in [0, 1], is different from all numbers in the above
the list, contradicting the assumption that all numbers in [0, 1] were included.

list

and

is

all 9's.

thus not in

Because of this contradiction it follows that the real numbers in [0, 1] cannot be placed in 1-1
correspondence with the natural numbers, i.e. the set of real numbers in [0, 1] is non-countable.

LIMIT POINTS, BOUNDS, WEIERSTRASS-BOLZANO


21. (a)

Prove that the


upper bound

infinite set of

least
is

a limit point of the

(d) Is

set.

trate the Weierstrass-Bolzano


(a)

Since
2

is

We
(6)

all

members

numbers

and greatest

(l.u.b.)

is

(namely

member
1)

of the set

which exceeds

member

is

bounded,

(g.l.b.)

the set a closed set?

(b)

of the set.
(e)

How

Determine the
(c) Prove that

does this set

illus-

and greater than 1

(for example), the set is

bounded;

a lower bound.

can find smaller upper bounds

Since no

bound

theorem?

of the set are less than 2

an upper bound, 1

1, i, i, i,

lovs^er

THEOREM

is

(e.g. |

and larger lower bounds

(e.g.

^).

greater than 1 and since there is at least one member of the set
for every positive number e, we see that 1 is the l.u.b. of the set.

of the set is less than


and since there is at least one member of the set
which is less than -I- e for every positive c (we can always choose for this purpose the number
l/w where n is a positive integer greater than 1/e), we see that
is the g.l.b. of the set.

Since no

NUMBERS

12

Let X be any member of the

(c)

for any positive number S


integer greater than 1/S),

we

we

small

take

we

see that

is

a positive

set.
To put this another way,
always includes members of the set, no matter how

a limit point of the

> 0.

The

(e)

Since the set is bounded and infinite it must, by the Weierstrass-Bolzano theorem, have at least
one limit point. We have found this to be the case, so that the theorem is illustrated.

does not belong to the given

set is not a closed set since the limit point

set.

TRANSCENDENTAL NUMBERS

and

Prove that \/2 + \/S is an algebraic number.


Let X = yf2 + \/3. Then x i/S = t/2. Cubing both sides and simplifying, we find ^ + Oa; 2
= 3v'3(^ + 1). Then squaring both sides and simplifying we find a" 9a;'' ix^ + 27x^ + 36a; 23 = 0.
Since this
solution, is

23.

<\x\<

set.

(e.g.

neighborhood of

(d)

ALGEBRAIC
22.

any deleted

see that

Since we can always And a number x such that


we can always pick x to be the number 1/n where n is

[CHAP.

a polynomial equation with integral coeflScients

is

Prove that the set of


n

follows that \/2

it

numbers

algebraic

all

a countable

is

which

is

set.

Algebraic numbers are solutions to polynomial equations of the form


where ao, ai,
a are integers.
.

Let

+ '\/3,

an algebraic number.

\ao\

oo a;"

!""'

...

|ai|

|a|

For any given value of

n.

number

of possible polynomial equations and thus only a finite

there are only a finite

number

of possible algebraic numbers.

Write all algebraic numbers corresponding to P = 1,2, 3, 4,


avoiding repetitions. Thus all
algebraic numbers can be placed into 1-1 correspondence with the natural numbers and so are countable.
.

COMPLEX NUMBERS
24.

Perform the indicated operations.


(a)
(4 - 2i) + (- 6 + 5i) = 4 - 2i - 6 +
(6)
(c)

(d)

(- 2

+ 5)i = -2 + 3i

- 4i) = - 7 + 3i - 2 + 4i = - 9 + 7i
(3 - 2i)(l + 3i) = 3(1 + Si) - 2i{l + 3i) = 3 + 9i - 2i - 6P = 3 + 9i - 2i +
-5 + 5i _
-5 + 5i 4 + 3i _ (-5 + 5z)(4 + 3i) _ -20 - 15i + 20i + 15t^
16 + 9
4-3i
4 - 3i
4 + 3i
16 - 9i'
-35 + 5t
5(-7 + i)
-7,1.
-5- + 5'
25
25
(- 7

(e)

+ 3i) -

5i

(2

+ i' + i^ + i^ + i^
1 + i

1
1

(/)

- 4il

|4

1
iff)

25. If zi

+ 3i\

3i

3t

and

Let
\ziZ2\

Zi

Z2

+ 3i

-6i

2^

25

= V(or+(-j%r = f

are two complex numbers, prove that

\ziZ2\

+ iyz.
+ iyi){x2 + iy2) = xiXi yiy2 + i{xiy2 + Xzyi)
= ^/{XiXz yiyiY + (xiy^ + XtVif = ^/x\xl + y\yl + ^Ivl +
Xi

(5)(5)

10

di"

1,1.

i+1

VW+iFWVW+W

iyi,

Z2

\/ (.x\

+ y\){x\ +

y\)

\zi\ \Z2\.

Then

X2

(xi

_ i-1-i+l + i

(V)H

1-

3i

+ (V-f +
+i
i i^

(i')(j)

\fx\

y^ \/x\

y\

\xi

iyi\\xi

^Ivl

iy2\

|l| \Z2\.

7i

CHAP.

26.

NUMBERS

1]

Solve

- 2x -

a;3

13

0.

The possible rational roots using Problem 7 are 1, 2, 4. By trial we find x = 2 is a root.
Then the given equation can be written (x 2)(x^ + 2x + 2) = 0. The solutions to the quadratic equation

-2

ax"

bx

'^

^4-8

= ~^ ~

"
cb

The

of

2i

1 +

- 4(ic

V6^
2a

= -1

set of solutions is 2,

POLAR FORM
27.

o^"
are

(^

-2 ^/-i _ -2

a=l,

For

= 2,

=2

this

gives

i.

i,

COMPLEX NUMBERS

Express in polar form

(a)

3i,

- 1 + ^i,

(6)

- 1,

(c)

-2 -

(d)

2v/3 i.

>y
180

45

-2^3

(a)

(c)

Fig. 1-4

(a)

Amplitude
3

(&)

3i

Amplitude

=
=

45

120

ff/4

p(cos

-1 +
(c)

Amplitude

sin 0)

tt

Amplitude

240

-2 -

28.

Evaluate
(a)

(a)

(-l

(6)

(-1

By Problem

-1 +

47r/3

2\/3

=
=
=

l(cos

[2(cos2r/3

1024 [cos

(2,7/3

1024(-l

= V2 (cos 135 +

J-/3

6,7)

\/(-l)^

Then

= 1. Then
= e"

V(-l)^+(0)'

cis

77

V(-2)*

sin 4,7/3)

+ (V^)'' = yT = 2.
= 26^'^""

2 cis 2,7/3

+ (-2Vsy = 4.
cis 4,7/3 = 4e*''*"

Then

i^'K

= 0, 1, 2

2"'(cos 20,7/3

(2W3 +

^/2 [cos (135

[eos

6,r)]

+
=

sin 20,7/3)

1024(cos 2,7/3

sin 2,7/3)

fc '

360)

sin (135

k 360)]

(i5^1+_fc^3601^
135
i

sin

fe-360

0]

are

V^ (cos 45 + t sin 45),


V^ (cos 165 + i sin 165),
V^ (cos 285 + i sin 285)
The results for = 8, 4, 5, 6, 7, ... give
fc

sin

= -512 + 512^3 i

1)

sin 135)

+ ^)V3 = (V2)-

fe

sin w)

sin 2,7/3)]"

results for

(-1

+
The

sin 27r/3)
p

3\/2e"'*

and De Moivre's theorem,

Then
(-l

J7-

4(cos 4j7/3

(&)

Modulus

radians.

+ \/3 zy,

27(6)

VSi)^"

Modulus

radians.

= ^/WT = SVZ. Then


+ i sin jr/4) = 3V2cis7r/4 =

jr/4

Modulus p

2(cos 27r/3

1 =
(d)

3\/2(cos

2ir/3 radians.

s/Zi

180

Modulus

radians.

repetitions of these.

These complex roots are represented geometrically


plane by points Pi, P2, Pa on the circle of Fig. 1-5.

in the

complex
Fig. 1-5

a;

NUMBERS

14

[CHAP.

MATHEMATICAL INDUCTION
29.

Prove that

1^

The statement

2^
is

3^

4^

%=

true for

Assume the statement true


1^

Adding

(k

+ 1)^

P + 2^+ 3^+

1, it

+ n^ =

1 since

n=

for

2^

+ l){2n + 1).

^(1)(1

1)(2

1)

1.

Then

k.

3^

1"

in{n

A;^

:^k{k

l){2k

1)

to both sides,

+ l){2k+l) + (k + iy = {k + l)[^k{2k + 1) + k + 1]
+ l)(2fc^ + 7fc + 6) = ^(fe + l)(fc + 2)(2fe + 3)
statement is true for n = k + 1 if it is true for n k. But since it is true for
it is true for n 1 + 1 = 2 and for m = 2 + 1 = 3,
., i.e. it is true for all

+fc^

which shows that the

n=

follows that

(fc

+ l)^ =
=

^k{k
|(fe

positive integers n.

30.

Prove that

a;"

2/"

has x

as a factor for
x^ y^ =

n=l

since
The statement is true for
Assume the statement true for n = k,

x y.

assume that x^

i.e.

positive integers n.

all

a;''(a;

i/) +

i;(a;'^

y^

has

Consider

as a factor.

y^)
y&s,

The first term on the right has a; j/ as a factor, and the second term on the right also has x
a factor because of the above assumption.
Thus
Then
x
31.

0!''+'

has

2/''+'

since x^

y^

a;

as a
y&s

factor if x" -~'y^ does.

2/

factor, it follows that x^

has x

Prove Bernoulli's inequality

(1

has x

(1

Thus the statement

But
true for

+ x)""-^ >

is

true for

as a factor, x^

y^

has

is

+ nx

for

positive since x

>

1).

if

x>

-1, Xy^O.

Then we have

+ x){l + kx) - 1 + (fc + l)a; + fcx* >


n k + 1 if it is true for n = k.

= 1.

not true for %

is true for n
2, it
integers greater than or equal to 2.

Note that the result


TO

(1

since the statement


all

n = 2,3,
1 + 2a; + a;^ > 1 + 2a;.
(l + x)^
i.e., {1 + xY > 1 + kx.

+ a;)" >

The statement is true for n = 2 since


Assume the statement true for n = fe,
Multiply both sides by 1 + a; (which is

for

y^

as a factor, etc.

must be true for n

+
2

However, the modified result

+ l)x

(fc

+1 =
(1

S,

+ a;)" S

...

and

thus

is

nx

true

is

1,2, 3,

MISCELLANEOUS PROBLEMS
32.

Prove that every positive integer P can be expressed uniquely in the form
+ ai2"-i + 022""^ + + a where the a's are O's or I's.
Dividing P by 2, we have F/2 = (10 2"-' + ai2"-^ + + -i + a/2.
Then

a is the remainder,

or

1,

P =

when P is divided by 2 and is unique.


Then Pi = ao2"-i + ai2-^ + + a-i.
or 1, obtained when Pi is divided by
is the remainder,

ao2"

obtained

Let Pi be the integer part of P/2.


Dividing Pi by 2
is

By continuing
33.

we

see that a-i

unique.
in this

manner,

Express the number 23

in the

The determination of the

all

the a's can be determined as O's or I's and are unique.

form of Problem

coefficients

32.
can be arranged as follows.

2^23
2 ^11

2)5
12
2 )3
2

Remainder
Remainder
Remainder
Remainder
Remainder

1
1
1
1

and

CHAP.

NUMBERS

1]

The

coefficients are

The number 10111

34.

10 111.
is

Check: 23

2"

2'

2^

1.

said to represent 23 in the scale of two or binary scale.

Dedekind defined a

cut, section

tion of all rational

numbers

15

or partition in the rational number system as a separatwo classes or sets called L (the left hand class) and
(the right hand class) having the following properties:
into

The classes are non-empty (i.e. at least one number belongs


Every rational number is in one class or the other.
Every number in L is less than every number in R.

I.

II.

III.

to each class).

Prove each of the following statements:


(a)
(b)

There cannot be a largest number

for L to have a largest number and for R to have no smallest number.


type of number does the cut define in this case?

for L to have no largest number and for R to have a smallest number.


type of number does the cut define in this case?
for L to have no largest number and for R to have no smallest number.
type of number does the cut define in this case?

It is possible

What
(a)

a smallest number in R.

It is possible

What
(d)

L and

It is possible

What
(c)

in

Let a be the largest rational number in L, and b the smallest rational number in R.
a b or a<b.

We
number

cannot have a
R.

=b

since

by

definition of the cut every

number

in

is less

Then

either

than every

in

We cannot have a<b since by Problem 9, ^{a+b) is a rational number which would be
greater than a (and so would have to be in R) but less than b (and so would have to be in L), and
by definition a rational number cannot belong to both L and R.
(b)

As an

indication of the possibility let


contain the number f and all rational numbers less than
contains all rational numbers greater than |. In this case the cut defines the rational
number |.
similar argument replacing | by any other rational number shows that in such case
the cut defines a rational number.
|, while

(c)

As an

(d)

indication of the possibility let

numbers greater than |.


argument shows that this cut always

contain all rational numbers less than | while


This cut also defines the rational number |.
defines a rational number.

As an

all

rational

indication of the possibility let

consist of all negative rational numbers


than 2, while
consists of all positive

R contains
A similar

and all positive


numbers whose
squares are greater than 2. We can show that if a is any number of the L class there is always
a larger number of the L class, while if b is any number of the R class there is always a smaller
number of the R class (see Problem 106). A cut of this type defines an irrational number.
From (6), (c), (d) it follows that every cut in the rational number system, called a Dedekind
cut, defines either a rational or an irrational number. By use of Dedekind cuts we can define
operations (such as addition, multiplication, etc.) with irrational numbers.
rational

numbers whose squares are

less

'

NUMBERS

16

[CHAP.

Supplementary Problems
OPERATIONS with NUMBERS
35. Given x = 3, y 2, z = 5,
(a)

yK3y +

{2x

Ams.

(a)

2200,

Find the

36.

^"^^

(a) 2,

RATIONAL

3(2a;

(6) 6,

g^fe

_i

(6)

1)}

= 1,

and

-51/41,

(c)

a;-2

evaluate:

'

2a^6^

+1

^^^^

'

-4,

(aj/

fexf

which the following equations are true. Justify

2(2a;

1)

+ 2) -

12(a;

(c)

Va;''-2a;

-1,

-1

(d)

1,

+ (x-y)(y-z)
+
.^_,X_

{z-x)(x-y)

62/)^

f, _
.
,,,_
(2/-z)(-a!)

'

'

all steps in

each case.

\/SM^8^+T - V2^T2 =

(<^)

(c)

(a;

(d) 1

,_J,_,

Prove that

37.

(6) 32,

- 2) +

8-a;

Ans.

2z),

set of values of x for

4{(x

(a)

z)(5x

=|

+5

a;

giving restrictions

any.

if

IRRATIONAL NUMBERS

and

38.

Find decimal expansions for

39.

Show that a fraction with denominator 17 and with numerator 1,2, 3, ...,16 has 16 digits in the
repeating portion of its decimal expansion. Is there any relation between the orders of the digits in
these expansions?

40.

Prove that

(a)

V^,

41.

Prove that

(a)

y/E

42.

Determine a positive rational number whose square

43.

Prove that every rational number can be expressed as a repeating decimal.

44.

Find the values of x for which


(a)

2*'

Ans.
45.

46.

If

(a)

is

^/^ + y/h

Sa;"

(6)

0,

8/3,

(6)

Ams.

(6) \/5.

(h) yf2

yfi,

18

-2, 3/2

3,

0.428571, (6)2.2360679...

(a)

are irrational numbers.

(6) ^f2

- 9a; +

5a;''

(a)

-2

are irrational numbers.

+ 4a;' - 35a; + 8 =
(c)

"v/s

|(5

from 7 by

differs

+ h^fm =
+ V3 + Vs ^ 12^5 - 2^15 + 14^3 - 7
^^
l-\/3 + V5

not a perfect square, prove that a

(c)

0,

a;*

l(-5

^/ri),

-21a;'

than .000001.

+4 =

0.

y/vf)

+ d-\fm

less

if

and only

if

and

= d.

Prove that

INEQUALITIES
47.

Find the
^"^

X
Ans.

+
(a)

- ^' (''> ^(*' + 2) =


< s 1, (ft) -6 s s
a;

48.

Prove

49.

Prove that for

50.

If

51.

If

52.

Prove that for

53.

Show
.

54.

a'

(a)

|a;

+ 6' =

a;

j/|

all

and

all

|a;|

\y\,

real x, y,

a;>0, prove that

.,n

x for which each of the following inequalities holds.

set of values of

c''

d'

a;"+'

a'0,

z,

24,

x^
1,

(c)

4,

(6)

|x

la;

+ 2| <

<

3/2,

+ y + ] S

+ y^ + z' S
prove that

+ ^^ >
\a

(c)

+ l/a\ a

x"

-{

xy

(d)

ja;]

(d)

la;-5|,

a;

>

\y\

>

^^

-1< < -1,

3,

^
a;

\z\,

(c)

|a;-2/|

or

a;

< -2

|a;|

|2/|.

+ yz + zx.

ac+hd S
where n

1.

is

any

positive integer.

2.

that in Schwarz's inequality (Problem 13) the equality holds


where k is any constant.

If ai, ai, 03 are positive, prove that A(ai

+ 02 + as) s

y/aiaiaz.

if

and only

if

Op

= kbp,

1,2, 3,

CHAP.

NUMBERS

1]

EXPONENTS, ROOTS
55.

Evaluate

Ans.

(o) 64,

56.

Prove

57.

Prove

(a)

6'''b<'

(6)

(6) 7/4,

loga

LOGARITHMS

and

4'-.

(a)

I7

log.. s (if,),

50,000,

(c)

MAT =

(d) 1/25,

M + Iog

Iog

a giving restrictions

iV,

if

-^/iM^|||M

(c)

()

(_l)./3

(_27)-/'.

-7/144

(e)

(6)

(d)3--3^

log

r loga

indicating restrictions if any.

any.

COUNTABILITY
58.

(a)

Prove that there

-5 ^

Ans.
59.

(a)

a one to one correspondence between the points of the interval


is the cardinal number of the sets in (a.)?
C, the cardinal number of the continuum.

-3.

(b)

(6)

is

and

What

Prove that the set of


Ans. (6) Ko

all

rational

numbers

countable.

is

(6)

What

is

the cardinal

number

of the set

in (a)?
GO.

Prove that the set of

61.

The

intersection of

belonging to both

numbers,

(a) all real

(6) all

irrational

numbers

two sets A and B, denoted by A n 5 or AB,


and B. Prove that if A and B are countable,

is

is

non-countable.

the set consisting of

all

elements

so is their intersection.

62.

Prove that a countable set of countable sets

63.

Prove that the cardinal number of the set of points inside a square is equal to the cardinal number
(a) one side, (6) all four sides, (c) What is the cardinal number in this case?
(d) Does a corresponding result hold for a cube?
Ans. (c) C

is

countable.

of the set of points on

LIMIT POINTS.

BOUNDS. WEIERSTRASS-BOLZANO THEOREM

64.

Given the set of numbers 1, 1.1, .9, 1.01, .99, 1.001, .999, .... (a) Is the set bounded?
(6) Does the set
have a l.u.b. and g.l.b.? If so, determine them, (c) Does the set have any limit points? If so, determine
them, (d) Is the set a closed set?
Ans. (a) Yes
(6) l.u.b. = 1.1, g.l.b. = .9
(c) 1
(d) Yes

65.

Given the set -.9, .9, -.99, .99, -.999, .999 answer the questions of Problem
Ans. (a) Yes
(6) l.u.b. = 1, g.l.b. == -1
(c) 1, -1
(d) No

66.

Give an example of a set which has

67.

(a)
(6)

68.

Let

no limit points.

Prove that every point of the interval


< ic < 1 is a limit point.
limit points which do not belong to the set in (a) ? Justify your answer.

be the set of

all

numbers

rational
(6) Is

having denominator

in (0,1)

and

n =

1,2,3,

(a)

Does S

is

not bounded.

(6)

Does this contradict

TRANSCENDENTAL NUMBERS
%/3

Prove that

2",

closed?

(a) Give an example of a set which has limit points but which
the Weierstrass-Bolzano theorem? Explain.

ALGEBRAIC
70.

(6)

Are there any

have any limit points?


69.

(a) 3 limit points,

64.

(a)-

-1/2

(6) -^2

V3 + V2

+ y^ + -/B

are algebraic numbers.

71.

Prove that the set of transcendental numbers

72.

Prove that every rational number

is

in (0, 1) is not countable.

algebraic but every irrational

number

is

not necessarily algebraic.

COMPLEX NUMBERS. POLAR FORM


73.

Perform each of the indicated operations:

+ IT3?'

<^Hm^

Ans.(a)l-4i,

{b)

-9 -

2-4i
(^)

46i,

(1
(f.

+ 7i
{c)^-U,
^^'

- Si) - 3(-2 + i) +
+ i){2 + 3i)(4 - 2i)
{l + 2iril-i)

(a) 2(5

(d)

-1,

(e)

if,

5{i-

(f)f-U

3),

(6)

(3

2^)^

(c)

-41

NUMBERS

18

74.

and

If zi

are complex numbers, prove

za

= M,

11

(a)

\Z2\

75.

Prove

76.

Find

77.

|zi|

solutions of

all

(c)

l^sl,

3,

|zi

- 2| ^

|zi|

|z2|.

-li

i,

and

zi

zi

is

two dimensions.

in
78.

Interpret geometrically the inequalities of Problem 75.

79.

Express
Ans.

form

in polar

6 cis

(a)

Evaluate
Ans.

81.

giving any restrictions.

\z^\'

be represented by points Pi and Pa in the Argand diagram. Construct lines OPi and OP2,
the origin. Show that zi + 22 can be represented by the point Pa, where OPs is the diagonal
of a parallelogram having sides OPi and 0P%. This is called the parallelogram law of addition of
complex numbers. Because of this and other properties, complex numbers can be considered as vectors
Let

where

80.

\z\\

(6)

\Zl\

+ Izz], (6) \zi + Zi + zj[ S [zij + l^^l +


Ans.
2x* - Sa;' - Ta;^ - 8a; + 6 = 0.

+ Z2I ^

(a) |zi

[CHAP.

-5V2 +

(a)

Determine
Ans.

(c)

(c)

2 cis 57r/3

sin25)][5(cosll0

l-^fli,

(d) 5,

id) 5 cis

sin 110)],

-5i.

(e)

5 els 37r/2

(e)

12 cis 16
^.^^^o^^g cis 62)

(6)

^g

'

-2i

(6)

5-v/2i,

the indicated roots and represent them graphically:

all

(4V2 + 4V2i)'',

(a)

[2(cos25

(a)

-2-2i,

(6)

2^/2 cis 57r/4

(6)

7r/6

+ 3t,

3\/3

{a)

(e)(-Jz-iY'\

(6) (-l)''^

=-1,

180

cis 36, cis 108, cis

(c)

v^ cis 110, -^ cis 230, V^ cis 350


292.5

(d) cis 22.5, cis 112.5, cis 202.5, cis

83.

If

+ Vsi

Prove that 1

324

cis 252, cis

(6)

82.

i"\

(d)

2 cis 135, 2 cis 255

(a) 2 cis 15,

an algebraic number.

is

zi = picis^i and Z2
Interpret geometrically.

prove

PjCis^j,

(a) Z122

pjp^ cis (0i

+ ^j).

^1/^2

(&)

(pi/pa) cis (0i

s^a)-

MATHEMATICAL INDUCTION
Prove each of the following.

(2m

- 1)

84.

85.

7^^
+
+ -p^ +
l-33-55'7

86.

(a

2'

2(5)^

90.

1(5)

91.

x'"-^

92.

(cos0

93.

A +

sin

a;

2/^"-'

cos

a;

+ [a+ {n- l)d]


^

^n[2a

n{n + l){n +

^lljZ-i
*

by

cosw^

cos 2x

2i+l

+ (n- l)d]
4(n

2)

+ l)(n + 2)

^^1

= In^n + ly

3(5)

+ l)

(2w-l)(2w

ar"--^

is divisible

sin 2a;

3'4'5

sin0)"

+ 2d) +

ar"

3'

-^

94.

{a

2-3'4

ar

1^

+ d) +

89.

l'2-3
88.

sin

a;

(5)-^

+
i

cos

nx

j/

for w

sinn^.

nx

=
=

^g

1, 2, 3,

Can

this be proved if

r^
+^x
cos Ax cos

2 sm

is

a rational number?

i)x

sin {n

Z sin
.

(ti

a;

7^ 0, 2ir, 4n-,

+ A)a;
,

-.

ia;

a;

where
is

C,

defined as

= a" + Cia''-'6 + ,.C%a''-^h' + +


= n(n- !)(.- 2)-(^-r + 1) ^
^ c.-,.
^!
r\(n r)\
r\

. ,

nC

1.

This

is

called the binomial theorem.

are called the binomial coefficients.

The

# 0, 2jr, 4jr,

(a+6)

95.

.Cn-iaft""'

Here

p!

Co = 1,

coefficients

Cr is also written

6"

=p(p-l)-l
Ci = n,

nCa

and
,

nin

g-j-

0!
,.

1)

CHAP.

NUMBERS

1]

19

MISCELLANEOUS PROBLEMS
96.

Express each of the following integers (scale of 10) in the scale of notation indicated: (a) 87
64 (three), (c) 1736 (nine). Check each answer.
Ans. (a) 1010111, (6) 2101, (c) 2338

(two),

(6)

97.

If a

Ans.
98.

number

is 144 in the scale of


110001, (6) 61

(a)

5,

what

is

number

the

Prove that every rational number p/q between

and

can be expressed

(a) 2,

(6)

in the

8?

form

^2"

2"

in the scale of

where the a's can be determined uniquely as O's or I's and where the process may or may not terminate.
The representation O.ai a^
a
is then called the binary form of the rational number.
[Hint:
Multiply both sides successively by 2 and consider remainders.]
.

99.

Express f in the scale of


An^. (a) 0.1010101..., (6)

100.

101.

In

number
what

106.

(c)

(6) 3,

0.2 or 0.2000.

(d)

8,
.,

(c)

What

10.

0.5252.

is

the

number

Show

.,

(d)

number

0.6666...

in the scale of 10.

Ans. 3.28125

two additional symbols t and e must be used to designate the "digits" ten and
Using these symbols, represent the integer 5110 (scale of ten) in the scale of 12.

in the scale of 10 consists of six digits.

the

Ans. 5

12?

Find a rational number whose decimal expansion

first digit,

105.

(a) 2,

scale of notation is

eleven respectively.
Ans. 2 e 5 <

104.

in the scale of 2 is 11.01001.

102. In the scale of 12,

103.

new number

is

one-third as large.

that the rational numbers form a

is

1.636363

Ans. 18/11

If the last digit is removed and placed before the


Find the original number.
Ans. 428571

field.

Using as axioms the relations 1-9 on Page 2, prove that


= 0, (6) (-2)(+3)=-6, (c) (-2)(-3) = 6.

(a) (-3)(0)

number whose square is less than 2, show that x + {2- x")/10 is a larger such
a rational number whose square is greater than 2, find in terms of x a smaller
rational number whose square is greater than 2.

107. (a) If X is a rational

number.

108. Illustrate

(6)

If

is

how you would use Dedekind

(a)V5 + V3, (b)^/^-^/2,

(c)

(V3)(V2),

cuts to defi.ne
(d)

^2/^3.

chapter 2
and Continuity

Functions, Limits
FUNCTIONS

A function is a rule which establishes a correspondence between two sets. For our
present purposes we consider sets of real numbers. If to each value which a variable x
can assume there corresponds one or more values of a variable y, we call y a function of x
symbolizing the function while
and write y - f{x), y = G{x), ... the letters f,G,
=
a.
function
at
x
value
the
G{a),
denote
the
of
f{a),
.

The set of values which x can assume is called the domain of definition or simply
domain of the function; x is called the independent variable and y the dependent variable.
If only one value of y corresponds to each value of x in the domain of definition, the
function is called single-valued. If more than one value of y corresponds to some values
Since a multiple-valued function can be
considered as a collection of single-valued functions, we shall assume functions to be
single-valued unless otherwise indicated.
of X, the function

Examples:

1.

is

called multiple-valued.

If to each number in 1 S a;
correspondence between x and

^
ic"

we

associate a

number y given by

defines a function /

which

is

x^,

The domain of / is l^x^X. The value of / at a; is given by 3/ =


For feample, /(I) = (1)^ = 1 is the value of the function at x = 1.
2.

With each time

after the year 1800

we can

then the

single-valued.
f(x)

x^.

associate a value P for the population


P and t defines a single-valued

of the United States. The correspondence between


function, say F, and we can write P F{t).
3.

y' = x where x>


then to each x there correspond two values of y. Hence y is
a double-valued function of x. We can consider this as two single-valued functions
/ and g where f{x) = ^/x and g{x) = ^/x.

If

Note that although a function is often defined by means of a formula as in Examples 1 and 3, it does not have to be, as seen in Example 2.
For convenience we shall often speak of the function f{x) rather than the function /
whose value at x is f{x). The distinction should however be kept in mind.

GRAPH

of a

FUNCTION

of a function defined hy y = f{x) is a pictorial representation of the funcand can be obtained by locating on a rectangular coordinate system the points defined

The graph

tion

by the number pairs

{x,y) or

[x,f(x)].

BOUNDED FUNCTIONS
If there is a constant

numbers),

we say

M such

that f(x) is
upper bound of the function.

^M

for all x in an interval (or other set of


an
hounded above in the interval (or the set) and call
that f{x)

20

CHAP.

FUNCTIONS, LIMITS AND CONTINUITY

2]

21

If a constant m exists such that f{x) g m for all x in an interval, we say that
f{x)
bounded below in the interval and call m a lotver bound.
If mS/(a;)gM in an interval, we call f{x) bounded.
Frequently, when we wish to
indicate that a function is bounded we shall write |/(a;)| < P.
is

Examples:

= 3 + a;
greater than

f(x)

1.

is
4).

bounded

-1

in

lower bound

a;

1.

An upper bound

2 (or any

is

number

less

4 (or

is

than

any number

2).

1/x is not bounded in


< a; < 4 since by choosing x sufficiently close to zero,
can be made as large as we wish, so that there is no upper bound. However,
a lower bound is given by
J (or any number less than i).
f(x)

2.

f(x)

it

If f{x) has an upper bound it has a least upper bound (l.u.b.); if it has a lower
bound
has a greatest lower bound (g.l.b.). (See Chapter 1 for these definitions.)

MONOTONIC FUNCTIONS
A function is called monotonic
X2 in the interval such that
strictly increasing.

Similarly
f{xi)

> f{x2),

if f{xi)

increasing in an interval

Xi<X2, f(xi)^f{x2).

^ f{x2) whenever

it is strictly

Xi

< X2, then

If

if

for any two points Xi and


the function is called

fixi)<f{x2)

f{x) is

monotonic decreasing; while

if

decreasing.

INVERSE FUNCTIONS. PRINCIPAL VALUES


a function of

denoted by f{x), then a; is a function of y, denoted by a; = f-\y),


Interchange of x and y leads to consideration of y = f-\x).
If f{x) is single-valued, f~\x) may be multiple-valued in which case it can
be considered as a collection of single-valued functions each of which is called a branch.
It is
often convenient to choose one of these branches, called the principal branch, and denote
it by f-\x).
In such case the value of the inverse function is called the principal value.
If

2/

is

x,

called the inverse function.

The function

2/

since for each

MAXIMA

and

leads to consideration of y = sin'^ x which is multiple-valued,


a; S 1
there are many values of y. By restricting sin"' a; to be
such that -jr/2 ^ sin-'a; S jr/2, for example, the function becomes single-valued.
In such case the principal value of sin~'( i) = ir/6.

Example:

a;

sin

in

a;

-1 S

MINIMA

If xo is a point of an interval such that f{x)Sf{xo) [or


f{x) s f{xo)] for all x in the
interval, then f{x) is said to have an absolute maximum [or absolute
minimum] in the
interval at x = cco of magnitude f{xo). If this is true only for x in some

deleted

borhood of
a relative

TYPES
1.

xo

where

maximum

of

>

[i.e.

for

(or relative

all

x such that

minimum)

0<\x-Xo\<

8],

then f{x)

is

neigh-

said to have

at Xo.

FUNCTIONS

Polynomial functions have the form


fix)

where

ao,

polynomial

if

aox"

a are constants

aix"-!

and w

is

...

Un-ix

a positive integer called

(i)
tjie

degree of the

aov^O.

The fundamental theorem

of algebra states that every polynomial equation


has at least one root. From this we can show that if the degree is n the
equation has exactly n roots (counting a repeated root of multiplicity r as r roots).
f{x)

[CHAP.

FUNCTIONS, LIMITS AND CONTINUITY

22

2.

Algebraic functions are functions y


Po(iC)r

where

po{x),

Pl(a;)r~'

=
+

satisfying an equation of the

fix)

Pn-lix)y

form

Pn{x)

(2)

are polynomials in x.

.,pn{x)

can be expressed as the quotient of two polynomials, i.e.


P{x)/Q{x) where P(x) and Q(x) are polynomials, it is called a rational algebraic
function; otherwise it is an irrational algebraic function.
If the function

3.

Transcendental functions are functions which are not algebraic,


equations of the form

i.e.

do not satisfy

(2).

to integers, rational

Note the analogy with real numbers, polynomials corresponding


functions to rational numbers, etc.

SPECIAL TRANSCENDENTAL FUNCTIONS


called elementary transcendental functions.

The following are sometimes


1.

Exponential function:

2.

Logarithmic function:

= a^

/(ic)

3.

This and the exponential function


f{x)-logaX, a ^0,1.
called the natural base of logarithms,
are inverse functions. If a = e = 2.71828
we write f{x) = \ogeX - In a;, called the natural logarithm of x. For properties,
.

see
3.

For properties, see Page

av^O,l.

Page

3.

Trigonometric functions:
sin X

smx,

cosa;, tana;

= ^^^,

^^^

cos
_ 1 _
" teE " '^IH:^
radians = 180). For

^Iir^'

a;

^^ ^
"^
,

^^^=^-^3^'

The variable x is generally expressed in radians (tt


values of x, sin a; and cos a; lie between -1 and 1 inclusive.
The following are some properties of these functions.
1 + tan^ x = sec^ x
x + cos^ a; = 1
sin {xi:y) = sin x coay cos x sin y
cos {xy) = cos X cosy ^ sin x sin y
tan xta.ny
i^^i^^y) = 1^ tan a; tan
2/
4.

(b)

(c)

5.

=
=
=

sin-^ic,

^y^
g g

{-it/2

cos-i X, (0

1/

tan-la:, {-kI2

7:/2)

tt)

<y

<itI2)

(d)

(e)

(/)

i/

=
=
=

cot^

sin {-x)

cos {-x)

tan (-)

The following
Inverse trigonometric functions.
nometric functions and their principal values.
(a)

sin^

csc-^a;

sec"!

a;

cof^a;

= csc^ x
= - sin
- cos x
= - tan

a;

a;

of the inverse trigo-

list

=
=
=

sin-il/a;,

is

cos"'
7r/2

real

^ y ^ 7r/2)
(0 g ?/ g
(0 < 2/ <

{-7t/2

tt)

1/a;,

tan-^a;,

tt)

Hyperbolic functions are defined in terms of exponential functions as follows.


(a)

sinha;

-^
e^

(d)

cscha;

(6)

coshx

,, , ,
tanha;

(c)

^^^
1

= ^^^i^
e^-e-"

sinhx
= ^^^j^ = ^.^^-x

e'

P
^^r+i^

(e)

secha;

= ^^^^ =

,.,

x
e^ + e'
cosh
__
_
- -^^^^ - g._g-x

(/)

^
cotha:
,,

a;

The following are some properties of these functions,


coth^aj-l =
l-tanh^a; = sech^a;
cosh^ai-sinh^a; = 1

csch^a;

CHAP.

FUNCTIONS, LIMITS

2]

sinh

cosh

{xy) =
{xy) =

cosh y

sinh x cosh y

cosh

a;

AND CONTINUITY

cosh x sinh y
sinh a; sinh y

sinh {-x)

cosh

Unh ixy) = tanhxtanhj/


1

6.

23

(-a;)

tanh(-a;)

= - sinh
= cosh x
= -tanh

a;

a;

tanh X tanh ?/

Inverse hyperbolic functions.


If x = sinhi/ then y = sinh-'a; is the inverse
hyperbolic sine of x. The following list gives the principal values of the inverse
hyperbolic functions in terms of natural logarithms and the domains for which
they are real.
sinh-ia;

(a)

cosh-

(b)

a;

In

(a;

+ V' + 1),

alia;

In

(a;

a;

\/^^^),

csch-^a;

in

sech-^

In

(d)

(e)

a;

^A + V^^ll\
^

+ VI

a^M
q

tanh-^a;

(c)

LIMITS

of

= ^in(i|j,

|a;|

<

^in^^Y

coth-^a;

(/)

^j

|a;l

>

^o
<

3.

FUNCTIONS

Let f{x) be defined and single-valued for all values of x near a; =


a;o with the possible
exception of a; = a;o itself (i.e. in a deleted 8 neighborhood of
a;o).
We say that the number I
IS the limit of f{x) as x approaches Xo and write
lim f{x) = I if for any positive number e
x-fxa

we can

(however small)

<

find

some

number

positive

(usually depending on

.)

such that

whenever 0<\x-Xo\< 8. In such case we also say that f{x) approaches I


as X approaches xo and write f{x)^l as x-^ Xo.
In words this means essentially that we can make the absolute value
of the difference
between f{x) and I as small as we wish by choosing x sufficiently close to
xo, i.e. by choosing
thq difference in absolute value between x and Xo sufficiently
small (but not zero, i.e. we
\f(x)-l\

exclttde x

= Xo).

Example:

Let

/()

if

We

closer to 4.

a;

= 2-

'^^^^ ^^

^*

thus suspect that lim f(x)

the above definition of limit (with

Note that lim


s-^2

f(x) = /(2),

value of fix) at x = 2 since /(2)


were not defined at a; = 2.

1-4)
i.e.

'='^''

* ^

To prove

4.

is satisfied.

For

definition.

The

this

we must

this proof see

the limit of f(x) as

by

x approaches

(i.e.

a;

limit

is

would

2), f(x)

gets

see whether

Problem

10.

not the same as the


in fact be 4 even if

f(x)

When

RIGHT

the limit of a function exists

it is

unique,

i.e. it is

the only one (see Prob. 17).

LEFT HAND LIMITS

and

In the definition of limit no restriction was made as to how x should


approach Xo.
It is sometimes found convenient to restrict this approach.
Considering x and Xo as
points on the real axis where Xo is fixed and x is moving, then x can
approach xo from the
right or from the left. We indicate these respective approaches
by writing x ^ xo+ and
a;-* Xo~.
lim f{x)

If
a;

-+ xq

and

left

hand

The

e,S definitions of limit

lim f{x)

x -+ Xo

h,

we

call

U and h respectively the right and

and denote them by /(a;o+) or /(xo + 0) and f{Xo-) or /(xo-0).


of f{x) as a;->a;+ or x-^xo- are the same as those for x-> xo
except for the fact that values of x are restricted to x > Xo or x <
xo respectively.

We

limits of f{x) at Xo

have lim
^-*^o

/(x)

if

and only

if

lim /(x)
x-tx+

lim fix)
x-*x-

I.

FUNCTIONS, LIMITS AND CONTINUITY

24

THEOREMS

LIMITS
A and lim

[CHAP.

on

If lim f{x)

g{x)

B, then

X-*X,j

X-fXf,

+ g{x)) =

1.

lim (f{x)

2.

\im(f{x)-g{x)) = lim

3.

limf/(x)g(a;)^

4.

lim^
x^x

lira f{x)

- r"
lim

fl'(a;)

lim g{x)

= A+B

lim

f lim /(a;)Ylim

lim /()

f(-.

/(a;)

flrC*)

=5B4

3^(0;)

flr(a;))

A-B
AS

IfB^O

X^*Xq

Similar results hold for right and left hand limits.

INFINITY
sometimes happens that as a; ^ Xo, f{x) increases or decreases without bound. In
such case it is customary to write lim f{x) = +00 or lim f{x) = - respectively. The
It

X-+-XO

X-*Xfl

symbols +=

and - are read pliis infinity (or infinity) and minus


must be emphasized that they are not numbers.

written

(also

respectively, but

it

In precise language,

=0)

we say

can find a positive number

0<\x~Xo\<B.

lim f{x)

that

00

if

for each positive

number

we

whenever
in general) such that f{x) >
(depending on
we
that lim f(x) = -00 if for each positive number

we say

Similarly

infinity

<

can find a positive number 8 such that /()


remarks apply in case x -^ Xo+ or x -^ xo.

-M

whenever

<

|a;

- Xo] <

8.

Analogous

Frequently we wish to examine the behavior of a function as x increases or decreases


without bound. In such cases it is customary to write x-^ +> (or <) or -> =0 respectively.

We

lim f{x)

say that

I,

or f{x)->

as

x^

+<x>,

if

for any positive

number

we

X-* +00

can find a positive number

x>N. A

(depending on e in general) such that


similar definition can be formulated for lim /().
X-*

i/(a;)

l\<e. whenever

SPECIAL LIMITS
sin
hm^
X

a;

,.

1.

,.

hm

1,

lim(l+-)
Xj

X-fM \

3.

hm
X-+0

e,

lim

cosa;

X-+0

X-+-0

2.

[1

xf"'

X-+0 +

1,

lim
x-*i

mX

CONTINUITY
Let
a;

lim f(x)

be defined and single-valued for all values of x near x = Xo as well as at


The function f{x) is called continuous at x = xo if
implies
three conditions which must be met in order that
that
this
Note
f{xo).

f{x)

== a;o (i.e.

in a 6 neighborhood of xo).

X-*Xo

f{x)

be continuous at x

xo.

^^-2J

FUNCTIONS, LIMITS AND CONTINUITY

3.

exist,

exist.

f{x) is defined at xa.

i.e.

/(.-To)

Equivalently

must

must

2. f{xo)

Jim f{x)

lim f{x)

1.

25

if fix)

continuous at

is

we can write

^,

/(lim X).

Examples:

1.

If

/(x)

= {/ ^^g

Hence lim
2.

m=

If

/(ck)

o;^

/(2)

^^'^'^

fails to

^he example on Page 23, lim

and the function

for all x, then lim

/(:)

this in the suggestive

/(2)

is

not continuous aTa;

and

f{x) is

/(a.)

4.

But

form

/(2)

= 0.

= 2.

continuous at

be continuous are called discontinuities of

.iH to
f"^"*^/*^^^/(^)
said
be discontinuous at these points.

/(a;)

c.

- 2.

and

^^

fix) is

In constructing a graph of a continuous


function the pencil need never leave the

' ' ^^^--teristic property and not a

c:Sufty ^r dTsco^ruitr"
^^''''

continuity,

"^
a:-/!fT"nv*Vn'71'^"^"^'''^'"**^^*
t^."^"""
"^

Not^tWt^-

t?at

" '^

.?L

RIGHT

"^

'"^^

^'^"''"" "' ''^'

Similarly, /(x)

Definitions

is

THEOREMS
Theorem

= .o

^"d r^^^l f the

continuous ion the left) ^tx


e and S can be given.

definition does not apply.

if

^Hm

in

on
1.

ft^tSui^^in

an

/(.)

= x, IPlim

terms of

CONTINUITY

restriction

---o-

fix)

/(.),

'

i.e.'^if

/(xo),
''
'^

i.e.

In such case

f^xJ^^fTl
fixo-)
'^^^
>

fix.)
'^^'>'-

INTERVAL

CONTINUITY
If fix)

and gix) are continuous at x^Xo, so

llnttir

polynom-r"')- sif^anTcr.!

\ZV

is

^"^

IS continuous at x = x..
tion of a continuous function is

il'

*^^ ^^^^

ST^/a^
continuous at

-^^

^^

^"'""^"^

also

^(-)--

are the functions

'"-^^ ->ts

^" ^^^^^ '"^^^ ^"^^^^^^

hold

^^^

^"

= Xo and

^ = ^(2/) is continuous ^t y =
y,
a /^ncizo^ 0/ a /t.^c^io.1 or comvosite
This is sometimes briefly stated
u 'isas: ^ conimuous funccontinuous.

and
andift'-Tf
If i/o--/(a^o), then
th^ the function
function,

define fix) as continuous at


|/(^)-/(a:o)|<. whenever b-r.|<8.

"^**^ ^-^(^-"^

x^Xo, the above

continuous ion the Hght) at .

call /(.)

we can

LEFT HAND CONTINUITY

and

If fix) is defined only for

we

definition of

=^

..

/(x)], called

AconZuZfut^

FUNCTIONS, LIMITS AND CONTINUITY

26

Theorem
Theorem

4.

5.

interval about x

If f{x) is continuous in a closed interval, it is


If f{x) is continuous at

= Xo

in vi^hich f{x)

>

a;

= a;o

[or f{x)

and f{Xo)>0

[CHAP. 2

bounded

[or /(a;o)

in the interval.

<0], there exists an

< 0].

continuous in an interval and either strictly infunction f~^{x) is single-valued, continuous


inverse
creasing or strictly decreasing, the
decreasing.
and either strictly increasing or strictly

Theorem

6.

If a function f{x) is

[a, b] and if f{a) = A and f{b) = B, then correbetween A and B there exists at least one number c in [a, b]
is sometimes called the intermediate value theorem.
Theorem 8. If f{x) is continuous in [a, b] and if /(a) and f{b) have opposite signs,
where a<c<b. This is related to
there is at least one number c for which /(c) =

Theorem

7.

If f{x) is continuous in

sponding to any number


such that /(c) = C. This

Theorem 7.
Theorem

9.
If f{x) is continuous in a closed interval, then f{x) has a maximum
for at least
for at least one value of x in the interval and a minimum value
value
and
one value of x in the interval. Furthermore, f{x) assumes all values between
for one or more values of x in the interval.
are respecand
Theorem 10. If f{x) is continuous in a closed interval and if
there
exists
of
bound
(g.l.b.)
lower
f(x),
greatest
and
(l.u.b.)
upper
bound
least
tively the
related
to
=
This
is
m.
or
which
for
f(x)
interval
f{x)
the
of
x
in
value
at least one

-M

Theorem

9.

SECTIONAL CONTINUITY

function is called sectionally continuous or piecewise continuous in an interval


if the interval can be subdivided into a finite number of intervals in each of
which the function is continuous and has finite right and left hand limits. Such a function has only a finite number of discontinuities. An example of a function which is secThis function
is shown graphically in Fig. 2-1 below.
tionally continuous in

a^x^b

a^x^b

has discontinuities at Xi, X2, Xa

and

X4.

Fig. 2-1

UNIFORM CONTINUITY
be continuous in an interval. Then by definition at each point Xo of the
interval and for any >0, we can find S>0 (which will in general depend on both e and
the particular point Xo) such that \f{x) - f{xo)\ < e whenever \x - xo\ < 8. If we can find
not on
8 for each e which holds for all points of the interval (i.e. if 8 depends only on e and
interval.
the
in
continuous
is
uniformly
that
say
f{x)
Xo), we
we can find
Alternatively, f{x) is uniformly continuous in an interval if for any >

Let

8>0

f(x)

such that

\f{xi)~fix2)\

<

whenever

|a;i-a;2|

<

where

Xi

and

X2 are

any two

points in the interval.

Theorem.
the interval.

If fix) is continuous in a closed interval, it is

uniformly continuous in

CHAP.

FUNCTIONS, LIMITS AND CONTINUITY

21

27

Solved Problems

FUNCTIONS
Let f{x)

1.

(a)

for 2

a;

8.

Find

(a)

- 2)(8 - 6) = 4 2 = 8
not defined since f{x) is defined only for 2
The set of all x such that 2 g a; s 8.
/(6)

/(-I)
(6)

= {x- 2)(8 ~x)

/(6)

a;

- 2t) = {(1 - 2t) - 2}{8 - (1 - 2f)}


(1 + 2t)(7 + 2t)
-7/2 S t g -1/2.
= (3 - 2)(8 - 3) = 5, /[/(3)] = /(5) = (5 - 2)(8 - 5) =
/(5) = 9 so that /[/(5)] = /(9) is not defined.

(e)

What

th^ dn

is

8.

where

t is

i.e.

(d)

(b)

(6

is

/(I

(c)

and /(-I)

The following

such that

2t S

8,

9.

table shows f{x) for various values of


x.
8-

8
8

f{^)

2.5

7.5

2.75

2.75

^''^^' ^''^^' ^^''^' ^^'^^' ^'^'^^' ^^'^'

(2.5jit),''aT2.?5)''^'

These points are only a few of the


infinitely many
on the required graph shown in the
adjoining
2-2. This set of points defines
a curve which is part

points

* ig.
of a parabola.

for

''

of2(ldt7nf?r.*^
and that of f{x) m
o

fl^(a;)

swer
swelTart"s
parts llVand"
(6) and

2<:.<8.

Problem

1.

(a)

(6)

Discuss the difference between the graph


is the l.u.b. and g 1 b of of^)'
(c)^n^^

What

^"^ ^^^"^ ^^ ^ ^ *^^ domain


M '/?.
for the T
function f{x) of Problem

'of

de/nitn?

(c)

^'^
i'T
^

1.

(a)

(6)

The

l.u.b.

(c)

The

l.u.b.

of g{x)

is 9.

The

g.l.b.

of g{z)

is 0.

of gix) is attained for the value


a; = 5
The e-lh of
no value of x in the domain of definition
such that ^(iLo

nM
'^^

j
,=***
"* """""''
"

'"^'^^

^^'"^

^^

(d)

^sed interval attams its l.u.b.


andg]r:^tL:S:::?^;h:tt::v2^^'^:^?;-:^^
"'"' " "' '="*^"""^
in a closed interval need not
attain its Lu.bTnd g.S"te pTo^^^^^^
ff
'

Let

if =^ is a rational number
1 0, if X is an irrational number
Construct a graph of /(a;) and explain
why it
= 1 since f is a rational number
/(|)
/(-5)
- 1 since -5 is a rational number
/(1^1423) = 1
since 1.41423 is a rational number
=
/(V2)
since \/2 is an irrational number
ffa;^
^

' ^

(6)
()

(6)

= / 1'

^"^

"

is

^'"^

^(*)'

misleading by

^-5), /(1.41423), /(V2),


itself.

fM

The graph is shown in the adjoining Fig.


2-3
From its
appearance it would seem that there are
two functional
values
and 1 corresponding to each value of x, i.e.
that f<x)
is multiple-valued, whereas
it is actually single-valued
Fig. 2-3

FUNCTIONS, LIMITS AND CONTINUITY

28

4.

[CHAP.

construct the graph of f-Hx), (b) find an expression for


f~^{x) and show that f~^{x) is not single-valued.
y = /-(x)
8- A
The graph of j/ = /() or a; = f-^(y) is shown in Fig. 2-2
(a)
of Problem 1(6). To obtain the graph of y-f-\x), we
6have only to interchange the x and y axes. We obtain
)P
the graph shown in the adjoining Fig. 2-4 after
4-

Referring to Problem

1,

(a)

^,

orienting the axes in the usual manner.

(6)

+ lQ + y =

have y = (x- 2)(8 - x) or x^-lQx


Using the quadratic formula,

We

B
1

Vl00-4(16 + y) ^

10
f-'{y)

2h

0.

^ ^/g^_

Fig. 2-4

Then, y

/-'()

V9 -

for
In the graph, AP represents i/ = 5 -I- Vs-^, BP represents 2/ = 5 - V9 - a;. Thus
S a; < 9, /"'(a;) is double-valued. This is seen graphically from the fact
each value of a; in
that every vertical line to the left of P and the right of AB meets the graph in two points.

The functions 5 + yj'i-x and 5 - yj^-x represent the two branches oi i'\x). The point
where the two branches meet (or have the same value) is sometimes called a branch point, in this
case at a; = 9, y 5.

(a)

Prove that g{x)

g a; g 9. (b) Is
decreasing in
Does g{x) have a single-valued inverse?

+ ^9-x

is strictly

tonic decreasing in this interval? (c)


(a)
^(a;) is strictly decreasin g if g {xi) > g(X i)

y/9-Xi > V9 - X2,


(6)

(c)

+ V9-X1 >

If

(see

y
a;

<

x-l.

If

- xi >

xi

<

is

strictly decreasing.

X2

then

- xi,

- 2)(8 - y)
then 2/ - 5 = V'9 - a; or squaring, x = -16 + lOy-y^
5 +
a single-valued function of y, i.e. the inverse function is smgle-valued.
single-valued inverse
In general, any strictly decreasing (or increasing) function has a
=

V9^^

is

Theorem
The

6,

Page

26).

the figure of
results of this problem can be interpreted graphically using

Construct graphs for the functions


greatest integer
(a)

xi

V9-a;2 showing that g{x)

mono-

since if g{xi) > g(xi) it is also


Yes, any strictly decreasing function is also monotonic decreasing,
it is not necessarily strictly
decreasing,
monotonic
g(x)
is
if
However
^
g(x2).
true that g(xi)
decreasing.

and

6.

wh enever

it

(a)

f{x)

^^^^

'

a;

'

^^^

^^^^

Problem

'-*'

4.

x.

the graph is included


The required graph is shown in Fig. 2-5 below. Since la; sin l/a;] S |a;l,
=
m,r, m - 1, 2, 3, 4, ^. .,
1/a;
=
or
=
1/a;
sin
=
when
that
=
f(x)
-x. Note
x and y
between y
The curve oscillates infinitely often between x = l/ir and a; - 0.
i.e. where x = I/tt, l/27r, l/3ir, ....

Fig. 2-5

(6)

shown in Fig. 2-6 above. If 1 ^ a; < 2, then [a;] = 1. Thus [1.8] - 1, [V2] - 1,
However, [2] = 2. Similarly for 2 g a; < 3, [a;] = 2, etc. Thus there are jumps at
The function is sometimes called the staircase function or step function.

The required graph


[1.99999]

1.

the integers.

Fig. 2-6

is

CHAP.

FUNCTIONS, LIMITS AND CONTINUITY

2]

29

(a) Construct the graph of /(x)


= tan a;. (6) Construct the graph of tan- :.. (c) Show
graphically why tan->a; is a multiple-valued
function,
(d) Indicate possible principal
Prmcipai
values for tan"' a;, (e) Using your choice,
evaluate tan-i(-l).
(a)
The graph of f{x) = tan x appears in Fig. 2-7
below.

7.

= /W =

Pig. 2-7

If

(6)

/"'()

tan-

tan I

Fig. 2-8

2/-/(a;)-taniB, then

a;

a^CrTr^r2-rabovr

= /-%) =

^ ^^^^

Then the graph of /" '(a:)


2/.
^^^^^ ^ '^' ^'^ -'*' ^^^^

tan"'

^" ^'^^

tan"'

a;

is

obtained

oriente/rurat

(c)

To define tan- a; as a single-valued function, it


is clear from the graph that
we can only do so bv
restricting its value to any of the following:
< tan- ic < 1/2 ^/2 <
/o
.
''^^'' '*'
shall agree to take the first as defining^he

(rf)

-W2

We

"

valued'^nverse.*

8.

*"'

'""^ "'

*""" "

''""'^'''

'^

tan -(-!) = -^/4 is the only value lying


between
according to our choice in (d).

(e)

Show
Tf

Z^-^S^\

prinSpaTvalue

= -^1,
- V^+1
*^" (x

that fix)
^

1m^

~1,

is

" ^*"^*'^ increasing function with a single-

-W2

+ l)2/-l = V^

or squaring,

it is
Srv21trnolThrouottnT;r"T

9.

If fix) = coshaj
inverse function,

H^=
yVy^-i.
Since

Thus
i/

= Ue^ + e-% prove that we


cosh-ix = In (a; -FV^^),

i(^+e-^),
a;

Choosingthe

(a;

Va;

e^'-ZyC+l =
In (i/ Vl/' - 1).

- Vy^~l =
a;

1).

{y

0.

(x

"
is

^-

+ 1)V -

it

2(a;

'^^'^^ ^(*>

the principal value

is

l)j/

a;

-l^^'^-- f--on.

an irrational algebraic function.

can choose as the principal value of the


x SI.

Then using the quadratic formula,

e'

2j/

V4y' -

yfy

ln(2/

i.e.

an irrational algebraic function.

^^^
not the quotient of two polynomials,
r^^^^T^^^^
so that it

flowever,

and W2,

we can
V2/'-l)

or

cosh-^/

= in

(j/

also write

y^^^Ti)

sign as defining the principal value


and replacing y by x,
a; 6 1 is made so that
the inverse function is real.

The choice

we have cosh-

a.

LIMITS
10. If

^"^

ia)fix)

= x\

(&)/(^)={o;

^;|,

Z^-^^V:^ZV<\T-T<\':'

"'

prove that lim


'^^^

''^^'^'

/(a:)

^^'^'-^"^^

4.

- ^

----)

-h

that

FUNCTIONS, LIMITS AND CONTINUITY

30

< - 2| < 1 or 1<


S 1 so that
= - 2| 1 + 21 <
=
2)(a;
+
|(a;
2)|
|a;^-4|

Choose

Then

|a;

|a;

as 1 or e/5, whichever
and the required result is proved.

Take

that
a;'

if

11. Prove

a;

make

to

4| <

\x^

we can

6,

choose

=1

|a;-l|

<

>

we can

find S

Smce a;#l, we can write

S.

>0.

4a;''

and

2x*

.
that
such .^
,

6a;'

<

whenever

- 2| <

jx

this will he satisfied.


in (a), since in both cases

3a; 3

a;^

2a;^

6a;

+ a:' + 3

on cancelling the

^~

^^^^

+3 _
""

(2a:'

_.|

'

'I

- ix' ~3x- 3)(x - 1) _


"
--l

'^T^

common factor a; 1 7^
Then we must show that for any e > 0, we can find 8 >
< a; < 2, a; 7^ 1.
< - 1| < 8. Choosing 8 S 1, we have

2x^

- ^\ <

merely coincidental.

is

must show that for any

<

56.

\x^

2x*-6x^ + X^ + S
a
= -o.
7^^

lim
"''^

when

<

= 2.

that

We

Then we have

# 2.

no difference between the proof for this case and the proof

is

exclude

.,

=2

a;

we wish

If

There

a;

2i

some numerical values. If for example we wish to make


choose 8 = e/5 = .05/5 = .01. To see that this is actually the case, note
< |iB - 2| < .01 then 1.99 < a; < 2.01 (x ^ 2) and so 3.9601 < a;^ < 4.0401, -.0399 <
.0401 and certainly |a;'' - 4| < .05 (a;'' = 4). The fact that these inequalities also happen

-4 <

we

3,

we can

.05,

to hold at

(6)

smaller.

<

S|a;

of interest to consider

It is

- 4| <

|x^

is

a;

[CHAP.

0.

such that

|2a;'

4a;''

- 3a; + 5| <

when

|a;

Now |2x' (g_l_4 + 5)3 = i7g_

4a;''

C \x-S\

12.

Let

=\x-3

f{x)

For

For

12a;''

- 2a; - 5| <

3,

a;

3,

Graph the

function.

Find lim

(5)

-3) _ -1

a;

>

3;

= 1,

j/

a;

(c)

f{x).

Find

right, f(x) -^ 1,

lim f{x)

i.e.

= 1,

(3,0)

l/(a!)

- 1] <

To prove

this

we

we can

>
- 1 < 8.

-1

11

find 8

<

whenever
f{x) = 1 and

whenever

the proof

so

1 <

a;

F>S-2-9

As x^

Since

lim f(x)

from the
=

left,

f{x)^ -1,

lim

<
<

Making

lim_ f(x)

= -l. A

proof can be formulated as in

(fe).

0.

must show that given any

|a;-0l

i.e.

lim f{x) does not exist.

f(x),

Prove that lim x sin \lx

If

<

<3

0<*-l<8.

We

5)

in the adjoining Fig.

= l,

Now since a; > 1,


consists in the triviality that

<

(3, 0).

from the

such that

13.

l2a;I

/()

a;-3

as seems clear from the graph.


must show that given any e > 0,

(d)

-3 ~

-(a;

\x--31

<

a;

X--3

and the point

(c)

6(l2a;''l

a;-3

la;--31

>

a;

a;

<

f{x).

Then the graph, shown

As

- 2a; - 51

2-9, consists of the lines

(b)

|2x''

x-*3

x-*3(a)

- 11

(a)

Find lim

(d)

fix),

|a;

as the smaller of 1 and e/17, the required result follows.

^^

'

(0,

lim

- 3a; + 5|

Taking

>

0,

we can

find

>

such that

la;

sin 1/a;

0|

<

when

6.
la;|

<

8,

then

the choice

|a;

sin

= e, we

l/a;l

|a;l

see that

S
sin l/a;| <

|sin l/a;|

la;

la;l

<

when

since

<

|sin l/a;|
la;l

<

8,

for all

a;

7^ 0.

completing the proof.

CHAP.

FUNCTIONS, LIMITS AND CONTINUITY

2]

Evaluate

14.

lim

To prove

X-H.O+

e-i/^'

we must show

this conjecture

2
1

Now

-^ > i
^

- 1)' ^

In (2/e

any value of

15.

>

>

0,

will

work

Explain exactly what

when

<

^'^

e-'''

(t" V'

*^a" ^^

"^

a;-

-^ <

16.

<

^ in(2/e-l) assuming

for

1)4

>

when

< e < 2.

>

=
^^^^
1)*

^^'^ ^"'^

2.

If

2,

0.

x^i{x~
^^

<e<

if

co

^ ^^^^^^

and prove
^""
P^^^^e the

-->-

(^^^P-d-

0<|a;-l|<s
<

when

all

lim

^"^^^^^ ""'^^ ^'

^^^ > M

this note that

such that

conjecture has been proved

meant by the statement

is

("

Choosmg

>

<

a;

e'/'+l

^'"'^

*^* our

since in such case

ol^t grerlirs^chltt'"^ ^^^'


To prove

find S

2-2-2e-'''

validity of this statement.


ing

we can

^ ^

^'

Choosing

<

>

g"^

+ e-^"

that, given

+ ->/' -2I =

when

31

(x

- 1)* <

<

or

|a:

- 1| < J-

1/y/M, the required result follows.

Present a geometric proof that

lim

^^

e-o

Construct a circle with center at O and


radius OA ^
t 5 on OA extended and point
C on OD so that lines BD

OD =
and

AC

It IS geometrically evident that

Area of triangle

OAC <

Area of
f smi?cos# < ^e < Atanij

Dividing by 1 sin

OAO <

sector

as in Fie- 2 in hlw
ru
'""^
are perpendicular S, Oi.

Area

of triangle

OBD

e,

<

cos 9

^H*

<

cos

<

sin e

COS*
tan

<

cos

As

0,

COS e

and

it

follows that lim

^^ =

Fig. 2-10

THEOREMS
^^' ^^

il^o

We
By

on LIMITS

^^^^

^'^*^'

P'^"'^^

must show that

if

*^at

Jim

hypothesis, given any

/(x)

>

it

must be unique.

h and

we can

\f{i>)~li\

\f(x)~h\

<
<

e/2
e/2

lim /(x)

find

>

when
when

k,

then

Z,

i,.

such that

<
<

<
\x
xo\ <
|a;

a;o|

FUNCTIONS, LIMITS AND CONTINUITY

32
Then by inequality

\h-h\
i.e.

18. If

\li-h\

lim

= B

lim g(x)

Since

\g{x)\

>

we can

find 8

g(x)

i.e.

19.

<

\B\

\\B\

Given lim

= A and

f{x)

(5)

lim /(i,(.)

(a)

We

= AB,

Using inequality

By

>

{A

>

\[f{x)

where
(6)

We

(A

have

lim fix)

Since

A - 1<

f(x)

a;ol

<
<

(c)

We

la;

ja;

a;o|

8.

\g(x)\

[f{x)

(^)

lim

+ g{x)] = A + B,

if

5^0.

such that
Q

<\x-Xo\ <

&

>

and

e/2

82

<
<

when
when

e/2

lg{x)-B]\

>

\g{.x)-B\

{1)

such that

<
<

|a;-a;o|

|a;

\f(x)-A\

a;o|

8^

(2)

8^

<

i5)|

81

=
^
S

AB\

e/2

and

- B] +
\f(x)\\g{x)-B\ +
\f{x)\ \g(x) -B\ +

f(x) [g{x)

8,

such that

A, given

>

<

when

|a5

aJo]

<

8^.

e/2

i.e.

we can

B[f{x)

- A]

lB||/(x)-A|

(\B\

1) \f{x)

l/(a;)-A|

<

~A\

<

for

< P where P is
such that
8^ >

|/(a;)l

find

>

we can

find S3

(i)
\

|a;-a;ol

>

<

\g(x)-B\

\f(x)-A\

such that

<

Sj,

i.e.

a positive constant.

8^.

lim /(x)

<

ia.

(S),

<

e/2P

g/i^i

+ i)

for

*^

83.

we have

in (i),

- xol <

- AB\

where

must show that for any

g(x)

P-^

m+

the smaller of

>

b\

we can

find 6

lf(f)~-^l

<

1)

81, 8^, 8,

>

2(|g|

"=

>

\g{x)

we can
B\

find 81

< \B\

and the proof

is

complete.

such that

when

>

such that

when

1)

0<la;-a;l<S

\B\\g(x)\
e

hypothesis, given

<

is

J:__i| =
By

find 8i

A, we can find

\f(x) g{x)

<

B, given

Using these

for

<

for

lim

when

<\x-xo\ <

(a)

>

find 8

lim g(x)

<

^\B\

\\B\

5^0,

if

so that f{x) is bounded,

Since

- B\ <

<

B, prove

< A + 1,

Since
la;

g(x)

Zi

:^\B\.

\[f(x)-A]

chosen as the smaller of

8 is

|/(a;)

<

and

(1), (2)

we can

|/(a;)-A|

g(x)]

Thus

we have

\g(x)-B\

Then from

\g{x)\

we can

3,

B)\

hypothesis, given

<

|a;-a;ol

\g{x)

\g{x)\

- (A+B)| <

Page

2,

+
>

\B-g{x)\

\[f{x)+g(x)]

+ g{x)\

such that

iL^ = i

must show that for any

\[f(x)

>

e/2

zero.

such that

<

for

lim g{x)

(c)

>

e/2

must be

(however small) and so

i|B|

from which

\g(x)\,

<

\f{x)-h\

we have

g(x),

IB]

\h-f{x)\

prove that there exists

B,

B = B-

Writing

+ f{x)-h\

\h-f{x)

- 0,

3,

than any positive number

is less

^(a;)

Page

2,

[CHAP.

<

[a;

a;o|

<

Sj

(5)

CHAP.

FUNCTIONS, LIMITS AND CONTINUITY

2]

By Problem

18, since

lim g{x)

>

\9{<e)\

Then

if S is

Prom

parts

and

8j,

^ |gW-g| <
\B\ \g{x)

and the required result


(d)

S^

and

(6)

we

0,

can find

<

when

l\B\

the smaller of

= B

we can

|a;

>

6,

a;o|

<

33
such that

8,

write

<

whenever

il^l

<

a;o|

proved.

is

(c),

lim

/(a;)

ff(a;)

= i^(=)
^"''
=
j t;^
g(x)

^'i = I

x'.i;x '

i'Ji;"|j

This can also be proved directly (see Problem


69).
The above results can also be proved in the cases
x^Xo+, x-^ x,-,
Note: In the proof oi (a) we have used the
results

A\

\f(x)

la;

'"* * ^"

iS-r .T ^T!

foVtt^ror/sS^rcVind^x'

"^'^^^ '"""^^^^

^^

^"^

and

,/2

+ g(x)-(A + B)| < J Of

l/(*)

"'^"^

<

a;^ , a;-^-=o.

\g(x)

- b'\ <

e/2

so that

course the proof would be

^^^"^^

"'

^-

^ ^^"-

f^t

^"^^^

Evaluate each of the following, using theorems on


limits.

20.

(a)

lim (x^

- 6a; + 4) =

lim

a;^

(2)(2)

lim (-6a;)

(-6)(2)

lim 4

= -4

In practice the intermediate steps are omitted,


lim
(6)

(c)

lim

i--i

a;''

-oo

+ 3a;

6a;''

+ 3)

lim

- Sa;" + 1
+ a;' 3a;

2a;*

lim

(a;

(^|H2Z_1)

CO

(a;='

lim

\/4

fe

lim 2

lim 6

lim

sill u^

Note that

-I +

lim

- +
X

2_

lim

lim

-|

i-n

a;"

lim

lim

V4 + fe +

A/m+2

^
"-">

lim

(-3)

-4

a;'

lim

x-oo

V4 + ft -

=
1.

19.

()

6+^-4

-K

(d)

X*

a;

by Problem

- 1)

(2a;

+ 3a;-2)

a;*

lim
*-+

lim

4 + A-4
A(V4+1 + 2)

sin

a;
'

v^ =

(d) and (e) if we use the theorems on


limits indiscriminately we obtain the
so c^WeA mdeternv^nate forms =c/co and 0/0. To
avoid such predicaments, note that in each case the
form of the limit is suitably modified. For other methods
of evaluating limits, see Chapter 4
(c),

CONTINUITY
21.

Prove that
Method

f{x)

ia;

x" is continuous at

a;

= 2.

1:

By Problem
Method

lim

10,

/(a;)

/(2)

and so

f(x)

is

continuous at

a;

= 2.

2:

We must show that given


- 4| < e when
- 2| < S.
[a;

any e > 0, we can find S > (depending on e) such that


The proof patterns that given in Problem 10.

|/(a;)
v
i

'

-/(2)|
m ;i

[CHAP.

FUNCTIONS, LIMITS AND CONTINUITY

34

22. (a)

Prove that

define /(O) so that f{x)


(a)

From Problem

continuous at

is

lim f(x)

13,

.;

= 0.

= 5,

so that f(x)

discon-

is

continuous. Because the function can


redefining f{x) so that /(O) = 0, the function becomes
at the point, we call the pomt a
function
the
redefining
be made continuous at a point simply by
removable discontinuity.

23. Is the

function

lim f(x}

f{x)

= -8

Problem

(see

Prove that

if

and g{x) are continuous at

fix)

(c)|||

{b)f{x)g(x),

if

= l. By redefining f(x) so that /(I) =


= 1, i.e. = 1 is a removable discontmuity.

not continuous at x

is

becomes continuous at x

11), it

= !?

continuous at

^^zii

not exist, so that f{x)

/(I) does

a;

a;

= Xo,

so also

are

<\x-xo\<

and rewriting

Prove that

We

/()

when \x - xn\ <

Prove that

f{x)

Since X

is

cb,

finally 2x'

By

8.

is

a;

<

as |aj-Xo|

8,

including x

i.e.

continuous at any point x = a;o.


any e>0, we can find 8>0 such that

choosing

2x^

+ g{x),

+x

is

=e

A = f{xo)

and

B = g(xo)

x<,.

that, given

must show

f{x)

(a)

/(xo)^O.

taking
These results follow at once from the proofs given in Problem 19 by

26.

re-

By

(b)

25.

Can one

(6)

= 0?

this limit is not equal to /(O)

But

0.

a;

not continuous at

is

^J

0.

tinuous at x

24.

|^^i"l/^'

f{x)

]/()

-/{o)|

|a;

<

a;o|

the result follows at once.

continuous at any point

a;

= Xa.

x-x = x\x^-x = x\ 2x' and


continuous at any point x = a; (Problem 25) so also is
products
of continuous functions are
and
sums
that
(Problem
24)
using the theorem

continuous.

27.

Prove that
If X, is

lim

= ^x^^

f{x)

if

any point such that

V^"^^ =

and

/(5)

for

<

a;o

<

9,

then

28.

xa.

An

e,

8 proof, directly

For what values of x

a;

lim

from the

in the

9,

then f{x)

lim f{x)

Hm yf^^ =

'"'^Here we have used the resuir that


at

/(9)

yfW) =

definition,

is

lim

continuo us in this interval.

y/x-h = Vo-5 =

Thus the

Hm

/(^)

fi.x,).

Also,

result follows.

= V7(^

if

/(^)

is

continuous

can also be employed.


definition is each of the following functions

domain of

continuous.

(a)

f(x)

fM ^
rW

(c)

fix)

/6^

i^
l

+
+

<="'^

sin

^ns.

all

x except

Ans.

all

Ans.

all

> -10

Ans.

all

a;

#3

Ans.

all ^,

x-l

(where the denominator

is zero)

a;

J
= Tr-
VlO +
l0-i''<^=^~3'^
a;

(d)

f(x)

(e)

/W =

(/)

/(a:)

If

10-''^""''"' *

J|

(see

since

Problem

55)

/(3)

Umf{x)

^r
a;

>

Then

0,

/(a;)

f{x) is

= ^^^ =

0.

continuous for

If
all

a;

< 0,

/(a;)

x except

a;

=
= 0.

^^ =2.

At

a;

= 0,

/(a;)

is

undefined.

CHAP.

FUNCTIONS, LIMITS AND CONTINUITY

2]

fi.x)

(ff)

""

a;==0

2,

As

in

f(x) is

(/),

continuous for

^^M

lim
follows that f{x)

it

Thus

is

= xc&cx = -^.

/()

a^csc,

/(O)

continuous for

all

continuous for

f{x) is

a;

< 0. Then

+ =

li^

continuous (from the

f(x)

{h)

35

all

Ans.

a;

all

Since

1.

S 0,

lim

esc

a.

a;

li^^

[compare

/(O)

domain of

its

0, ,r, 2;r, 3:r,

a.

= 0.

everywhere in

i.e.

x except

x except ^, 2w, 3,r,

at

left)

lim 2

definition.

....

^,

^(o),

,,, that /(.)

(h)].

UNIFORM CONTINUITY
29.

Prove that
Method

1,

f{x)

x^

is

uniformly continuous in

<

a;

<

1.

using definition.

We
where

must show that given any e > we can find S >


such that
e and not on Xo where
< xo < 1.
X and a;o are any points in
< a; < 1, then

If

lx'-xl\

la;

a;| (a;

<

a;o|

+ 1(

11

la;

Choosing

i^'-*l<2^r.''"a;f<'rwwAV"T*,"*
"^ " ' ^""^ ""* "
'
0<a;< J
The above can also be used to prove that f{x) = x'
"^

^''-

is

- xl\ <

\x'

depends only on

Prove that
Method

/(x)

1/a;

is

- a;|<

a;oI

|a;

'

Let

a;

and

Xo

However,

0</<r

^" ^'^^

" ''^^

= ./2, we see that |.^-.S(<. when


/(') = ^ i uniforinly continuous in
g^s

uniformly continuous in

<

a;

= ~-^.

Then
1

J^

in the given interval.

'' ^"'^^ **^^*

|a;-a;o|

i.

Hence bv the theorem on

<

it

^(^>l

"

<

1.

Then for any

-'^^

'^

= T >

Xi,

^ contradiction and

'^^^^

" -I <
<

(since

follows that

<

m^l/x

<

e>0 we
^^^ " -

"

should

S.

1).

cannot be uniformly continuous in

2:

Let Xo and

a;o

be any two points in


fi<o)

f{xo

(0, 1).

Then

Xo

a;o

+S

'

xo\

H^'^^^

not uniformly continuous in

""**ly continuous

Ju^^Ta^.^^ ^^""l''*"""
i'
L tt interval. ''"

'

1:

bp

Method

\x
'

Method 2:
The function fix) = x' is continuous in the closed
interval OSa;^!.
Page 26 it is uniformly continuous in S a; g 1
and thus in < a; < 1.

30.

when

xo{xo

+ S)

z': r t!s;tntrouT*^^^ ""^"^^ '^ ^'--^^'^^ ^^ ^-^^-^'^^^ ^^^-^ *

--

*^

^-et-

[CHAP.

FUNCTIONS, LIMITS AND CONTINUITY

36

MISCELLANEOUS PROBLEMS
3L

If

y^f{x)

continuous at x = Xo, and z^g{y)


is continuous at x=^Xo.

is

prove that

continuous at y

= yo where

yo

= f{xo),

and yo respectively,
Since by hypothesis f{x) and g(y) are continuous at xo

= g{f{x)}.

Let h{x)
we have

Then

which proves that h(x)

lim f(x)

/(lim x)

f{xo)

lim g{v)

g(l\m y)

gM

lim hix)

32,

is

z = g{f{x)}

lim g{f(x)}

g{f(x)}

is

flr{

lim /(x) I

continuous at x

ffifM)

g{f(xo)}

/^-(sco)

= xo.

8, Page 26.
must be an interval (a a+h),
Suppose that /(a) < and f{b) > 0. Since f(x) is continuous there
and so has a least upper
bound
upper
an
has
a+h)
{a,
points
of
h>0 for which fix) < 0. The set
if /(c)
because
bound which we call c. Then /(c) S 0. Now we cannot have /(c) < 0,
T^^f/^^^*"^^ ^^^
c) for which f{x) < 0; but smce
than
greater
values
(including
c
about
would be able to find an interval
and so we must have /(c) - as required.
c is the least upper bound this is impossible,

Prove Theorem

>

If /(a)

33. (a)

Given

and

f(x)

< 0,

2a;^

a similar argument can be used.

- 3rr^ + 7x - 10,

evaluate /(I) and

number x such that l<a;<2.

for some real

X in

f(b)

(c)

Show how

Prove that

(b)

/(2).

/(a;)

to calculate the value of

(b).

(a)

/(I)

(b)

If

2(1)'

- 3(1)^ + 7(1) - 10 =

-4,

continuous in tt^^Sb and


of X between a and 6 such that f(x) =
is

/(2)
if /(a)

2(2)

and

- 3(2)^ + 7(2) - 10 =

/(b)

have opposite

8.

signs, then there is a value

(Problem 32).
continuous in
realize that the given polynomial is
only
need
we
theorem
To apply this
exists a
there
Thus
0.
and
>
/(I)
<
/(2)
that
(a)
iaa;g2, since we have already shown in
0.

/(c)
that
such
2
and
1
between
number c

f(l 5^

2(1 5)

again, we see that


0.5. Then applying the theorem of (b)
since
and 1.5 and is "most likely" closer to 1.5 than to 1,
is
but
conclusion
valid
=
a
always
-4 (this is not
than /(I)

- 3(1.5)^ + 7(1.5) - 10 =

lies between 1
has a value closer to
worth pursuing in practice).
= 2(1.4) - 3(1.4)^ + 7(1.4) - 10 = -0.592 we conThus we consider x = 1.4. Since /(1.4)
1.4.
is most likely closer to 1.5 than to
clude that there is a root between 1.4 and 1.5 which
to 2 decimal places.
Continuing in this manner, we find that the root is 1.46

the required root

= 0.5

/(1.5)

34.

Prove Theorem

10,

Given any >0,


Tv><.

of

Theorem

hypothesis,
f{x)

=M

i
4,

Page

Page

we can

> i
26.

26.

find

x such that

so that

However

we must have

.,
if

^,,

M - f(x)
is

we suppose

<

by

definition of the l.u.b.

M.

not bounded and hence cannot be continuous in view

also continuous.

M - f{x)

is

continuous, by

In view of this contradiction,

we must have

that f{x)=M, then since

J^^J,^)

for at least one value of x in the interval.

Similarly

we can show

that
that there exists an x in the interval such

f{x)=m (Problem

93).

^^^^2]

FUNCTIONS, LIMITS AND CONTINUITY

37

Supplementary Problems
FUNCTIONS
Give the domain of definition for which each of the
following functions

35.

Ans.

If

36.

(6)

all

{a)

-2Sa;g3,

x>3,~2<x<2.
find:

If

fix)

iQg

/A
(/)

^^0;

<

s 2,

find

2^/(0)

h^O;

f(

5X

^^^

a;

Determine whether

^^^

^^^_ ^^^

(.)^.,^o,2

|,.^0,2

(d)

(a)

the

and (6) the g.l.b. of


and g.l.b.
Ans. (a)

l.u.b.

f{x) attains its l.u.b.

f(x)
(6)

8,

Construct a graph for each of the following functions.

m=

(a)

(b)

-3 S . ^ 3

1x1.

(/)

= 2-i^, -2Sa;S2

fix)

/(x)

(fif)

CO,

where

(e)

/(x)

a;2

sin

[.]

greatest integer

coshx

x<0
sin

39.

^^^_^^^.^

/{/(a:)}.

(/)

|^,,^04,2

and single-valued:

(d)
log. (a,' _ 3a, - 4a: + 12).
s (2m + 1W3, m = 0, 1 2 ',

'-^-5,2

-^+5-.

= 2a;^

5 /(-I)

2m;./3

(c)

is real

V^i^,

(c)

2,

M^m

(,)

_^

7^

a;

(.)

(e)

^-Wi

38.

(x

(d)

/W + /(4/.;),

- 2)/(x^ - 4),

(6)

/(.)=f^,.^2,

(d)

37.

+ 4),

V(3-:)(2x

(a)

1/a;,

a;

.^
f(^x)

(^^

a;

^'"'^

Construct graphs for (a) ^Va= + z/V6^ =


1,
(6) xVa^-j,V6^ = 1,
(c) y^ = 2px,
and (d) = 2aa:-a:^
a, b, p are given constants.
If y = fix) in each of these cases, I
fix) single-valued?

where
40.

(a)

cos

From
r

,3

the choice

the graph of y ^ cos a: construct the graph


oi y = cos-'x.
(6) Show graphically
."
, multiple-jalued function. Indicate possible choices of a principal
value oLos" .

(6),

find

cos-

(1/2)

- cos- (-1/2).

Does the value of this depend on the choice?

41.

Work

42.

Given the graph for y = fix), show how to obtain the


graph for y = fiax
given constants. Illustrate the procedure by obtaining
the graphs of
(a) y - cos 3a;,
(6) y = sin (5a; + W3),
(c) 2/ = tan (,7/6 - 2a;).

43.

44.

parts

(a)

and

(6)

Construct graphs for

of Problem 40 for

(a)

2/

6"^,

(6)

2/

(a)

In

|a;],

seC' x,

(c)

sin- (-V3/2)

(6)

(D- tan- (-1)


cot- (l/x/s) - cot- (-l/v/3)

(c)

(/)

tan-

(d)

cosh-y2

(e)

e-e..- <.v.,

(^)
(fe)

(i)
(^.)

Ans.

45.

2/

Using the conventional principal values on Pages 22 and


(a)

(c)

-^/3

t/2

id)

ln(l

(a) cos {^

sinh (In 2)},

(a)

-:./3

(&)

Evaluate

+ V2)
(6)

whv

U^^^

(6)

2/

e'l-l sin

cot"' .

b),

where a and 6 are

a;.

23, evaluate:

^i^-: ^

sinsin-

-1 g g 1
S S ^/2
W2 g g 3;./2

^^^-,

(cos 2a;),
(cos 2x),

a;,

a;

a;

tanh(csch-3x), a;#0
eos(2tan-a;^)

- 2a;

(e)

(^)

^/2

(/)

^/2

ih)

2a;-3W2

cosh"' {coth

ExplZ

(In 3)}.

Ans.

i_^4

\x\
^'^

(a)

W^^+t

-^2/2,

(6)

In 2

^^^

1+^

'

[CHAP.

FUNCTIONS, LIMITS AND CONTINUITY

38
tan-

Prove that

(a)

46.

tan-^a;

Is

47.

If

f{x)

48.

Prove that tan"' a

4<J

Prove the

tanh ^x

(sinha;)/(l

cot'' b

coshx),

= 1.

/(O)

(sin^x)/a;S

a;^/

= 1.

cot"' a.

3 sin x

(e)

In esc
|

4 sin^ x,

a;

Discuss the cases when /(O)

(a)

of:

4 cos'

a=

3 cos x,

In |tan ^x\.

undefined

is

cos 3x

(c)

cota;|

Find the relative and absolute maxima and minima

50.

discussing the case

(^^^,

sin 3x

(b)

sech^ x.

Ry)

identities:

if

also?

f(.x)

tan"' 6

are taken.
the conventional principal values on Page 22
Explain.

= W2

:.

'^/2

t^n-^x, prove that

ST- tanh^=^
(d)

cot-

tan-'(l/a;)

(6)

:.

fix)

or

m^l;

.(sinx)/a;

defined but

/(O) is

f(x)

(6)

1.

LIMITS
by using the

first

limits,

Evaluate the following

51

(a)lim(x=-3a.
(/)

li^.

Let

f{x)

Evaluate

- 1,

0,

lim
'"'

(6)

.,

+ 5,

a;

Ans.

/(fe)

- /(0+)

{c)

lim^^zT^'

4,

(d)

-1,

(0

(a)

Construct a graph of

(d)

/W,

id) 5,

(e)

_d
ana

/feN
(O)

54.

(a) 2,

(a) If

Mm

fix),

-1,

if)

Prove that

5e.

Let

we

if

/(.)=|lp,

Find

lim 1^,

(a)

_o+

(6)

a;

59.

Explain exactly what

^f^ =

60.

Prove that

61.

Explain

62

If

If

[.]

64

If

lim fix)

What

m=

.#0,/(0)

(f)

Urn

fix).

Justifying your answer

is

the function of Prob. 52.

y.(^)

j^

justifying your answer.

is

lim !^

Evaluate

(a)

-1,

(c)

(a) |,

(6)

Illustrate

lim sin

(a)

Hm

(6)

lim 10 -'

(&) 1/6,

(c)

2,

a;

(1

= A

and

(a)

u2{fix)-9ix))

Km /W.

^Um

(c)

/(.),

lim

justi-

/(.),

does not exist.

lim/(la;|) exist?

-,

:=

lim

(a)

(e)

x.

^ljm^^ =

/(.).

Ans.

(a) 1,

(6)

-1

lim gix)

(6)

ib)

(a)

=T-B,

ib)

B,

lim 3^:r2

lim e-sina;

does exist.

Umjix). (0 Jn^

Urn {.-[.]},

Jim [fix)}^

Explain.

/(*),

id)

Um

fix),

(e)

Ans.

(a)

Hm

/(.).

does not exist

evaluate
(a)

(c)

= A\

true?
generalizations of these do youluspect are

lim fix)

Explain.

your answers graphically.

does not exist,

A. prove that

If

= 2?

(a) still

writes:

- e-)

(b)

0,

(d) 1/6,

/(0)

Does your answer to

(6)

a;

meant when one

-,

:.#0,

cc^cosl/*,

= i.

Ans.

i-o-

largest integer

63

65

(a)

why

(a) 2,

16

using the definition.

/(,)=.3^+W,"Evaluate

Arts,

I'

'

limits.

h)*

the function defined in Problem 56, does

/(a;) is

If

lim

all cases.

58.

(a)

consider

lim 10-i/c'-3>'

fying answers in
57

evaluate

x^ cosl/x,

remain the same


55.

^'^

does not exist

f ih)-fiO-

li^
^nm_

(2

,.

/(a;).

fix),

lirn

ie)

(/)

(6)

fix)

32.

(e)

jj^

Atis.

\/S-2
^I^.-J^T

^'^

-1,

-10,

(c)

(6) 9,

lim

(a)

(6)

(0 MmJ(x),

/(.),

....

.
^
Evaluate

<
=
>

a;

in each case.
53.

(a) 2,

Arts,
r SiB

52.

WUm^^^,

+ 2),

and then using theorems on

definition

a,i_4

(6)

(6)

Jhn

Hm

VM

{.

[a.]}.

0,

^/A-

Can you prove them?

prove that

lim {afix)

b gix)}

= aA + bB where

a,b

any constants.

(6)

<^^^P-2]

FUNCTIONS, LIMITS AND CONTINUITY

If the limits of f(x), g{x) and h{x) are


(a)
{f{x)
g(x)
h{x)}

66.

hm

(a)

(3i;-l)(2a;

(b)
^^'

lim
i

^'^

i!l?}^(^

(5a;-3)(4a;

If

2 -3a;

Given

lim

Ans.

(a) 3,

X-

io

If

lim

()

72.

Generalize these results.

limits.

0,

fe

(">

-8/21

(*)

3/10

1/32

Ans. 1/12

x').

prove directly that

lim

-x g(x)

evaluate:

('^^

- 3)

(=

i,

(c) 1/2,

(6) 0,

-+0

gX

= B ^

^1^^^=:^^
^"'iro-^^

^^^

Let 8

(Hint:

lim g(x)
"*'

1,

()il"3^
-.

^''^

and

^"*-

3^)

-^^

= A

/(a;)

"

= ABC.

+ 3)
+ 6)

""".^

k-fO

71.

respectively, prove that-

lm,J(x) g(x) h{x)

\(3x + 2)(5x-3) ~ x'-5x +


ZJ

Evaluate lim ^^^

68.

g^g'

lirr,

'

70.

and
(b)

Evaluate each of the following using the theorems


on

67.

69.

A,

A+B + C,

39

^">1

CSC

(d) -1/^,,

4?^-^^
+ 3 sin

(e) lira

^*

(e)

(/)

,,,
(^)

40;

^os^a^-cosfcg

lim

(/)

2/7,

2a;

^^^

i(6-a),

(p)

-i,

- 2cos. +

i!
:-+o

cos2x

a;^

^.^ 3 sin ^x

sin

3:ra;

(h) 4^

Prove that

prove that:

1,

W l^ =

= *-;

/()

if

and only

lim

if

Inf
f{x)

>

a,6

0;

lim

(c)

lim /(x)

^^

a.

I.

CONTINUITY
73.

Prove that

/()

a;^

74.

Prove that

/(a;)

1/a;

75.

Investigate the continuity of each of the following


functions at the indicated points:

- 3a; + 2
is

(a) /(.)

^;

(6) f(x)

a;-la;|;a; =

4n.

,^0,

continuous at x

continuous

/(0)

= 0;

(a)

at

=,

= 4.
2,

in 1

(6)

(o)

f(x)

a;

(a)

discontinuous,

(6)

'"*'^^''

Prove that

78.

If

f{x)

= x'

m/g(x) and

is

80.

Prove that a polynomial


'^"^

liiiio"^

"''

(tan"'

'"

(c)

continuous,

(<)

= 3;

a;

=2

^,
*"

l<a;<2'

a;

discontinuous

''^^^^t'^^t^ *> continuity of f(x)

g(x) are continuous at

Prove that

''

continuous,

/(2)

[x]

in the interval (a)

1< < 2,
a;

continuous in every finite interval.

79.

f(x)

x^2,

r-

^llZ H*'''*

77.

3.

g^;
/In

'"
(6)

is

a;)/a;,

is

/(O)

=1

a;

=
is

a;.,

prove that fix) must be continuous at

continuous at

a;

a;

a;..

= 0.

continuous in every finite interval.

polynomials, prove that

/(a;)/^(a;)

is

continuous at each point

a;

= a;.

for which

82.

functions.
Give the points of discontinuity of each of the following

(6) f(x)

An.

(a)

a;

# 0, /(O) =

x' sin 1/x,

a;

none,

2, 4,

(b)

UNIFORM CONTINUITY
Prove that /(sc) = z' is
83.
= x^

(c)

(d) /(a^)

none,

(d)

7,r/6

uniformly continuous in

a;

<

not uniformly continuous in

<

a;

<

a constant, prove that f{x) = l/x' is (a) continuous in


tinuous in a < a; < " if a > 0, (c) not uniformly continuous in
If

is

(c)

any

finite interval.

a< <
a:

<

a;

<

"

if

aS 0,

(b)

uniformly con-

1.

and
(a) f{x) g{x)
interval, prove that
State and prove an analogous theorem for

and g{x) are uniformly continuous in the same

If f(x)
(6)

0, 1, 2,

>.

85.

is

a;

{b)0^x^ 2,

2,

Prove that

f{x)

'

+ 2 sin
2m,r, m =

2m-, II77/6

<

(a)

84.

86

[CHAP.

FUNCTIONS, LIMITS AND CONTINUITY

40

are uniformly continuous in the interval.

f(x)g(x)

m/g(x).

MISCELLANEOUS PROBLEMS
87.

Give an

88.

(a)

"e,

B" proof of the theorem of Problem 31.

Prove that the equation tan

<

3,r/2

a;

<

5ir/2,

5n-/2

<

a;

<

a;

a;

has a real positive root in each of the intervals

77r/2,

the result in (a) graphically by constructing the graphs of y


locating their points of intersection.
Ans.
tan x = x.
(c) Determine the value of the smallest positive root of
(6) Illustrate

89.

Prove that the only real solution of sin

90.

(a)
(6)

a;

is

a;

tan x

(c)

and

^
y^x

and

4.49 approximately

= 0.

=
has infinitely many real roots.
a; + 1
Prove that for large values of x the roots approximate those of cos

Prove

92.

Suppose

.,

that

h),

,.

lim
j:_o

f(x) is

a;''

sin

(l/a;)

^-^

smx

a;

0.

0.

Prove that there exists an interval (a: - h,


a; = a;o and assume f(x) > 0.
which f(x) > 0. (See Theorem 5, Page 26.) [Hint: Show that we can make
Then show that f(x) S /(a;o) - |/(a;) - /(a;o)l > ^fixo) > 0.]

continuous at

where h>0,

|/(a;)-/(a;o)|

93.

a;

Prove that cos x cosh

91.

xo

r/2<x< 3ir/2,

in

< ^ f(x<,).

bound
(a) Prove Theorem 10, Page 26, for the greatest lower
Page 26, and explain its relationship to Theorem 10.

(see

Problem

34).

(6)

Prove Theorem

9,

chapter 3
Sequences
DEFINITION

of a

SEQUENCE

function of a positive integral variable, designated by f{n) or u, where n = l,2, 3,


is called a sequence.
Thus a sequence is a set of numbers ui, 112, Us,
in a definite
order of arrangement (i.e. a correspondence with the natural numbers) and formed according to a definite rule. Each number in the sequence is called a term; u is called the
nth. term.
The sequence is called finite or infinite according as there are or are not a
finite number of terms.
The sequence Ui,U2,U3,
is also designated briefly by {Un}.
.

Examples:

1.

The
Un

2.

The
Un

f(n)

numbers 2,7,12,17,
.,32 is a finite sequence; the nth term
= 2 + 5(w - 1) = 5n - 3, n = 1, 2,
7.

set

of

set of

numbers

l/(2n-l), w

Unless otherwise specified,

LIMIT

we

1, 1/3, 1/5,

1/7, ...

is

is

given by

an

sequence with

infinite

Jith

term

1,2, 3,

shall consider infinite sequences only.

SEQUENCE

of a

number I is called the limit of an infinite sequence Ui, U2, Us,


number we can find a positive number N depending on e such that
integers n>N. In such case we write lim m =
.

if

for any positive


- Z| < e for all

\Un

I.

n~*oo

Example:

m = 3 + l/w
lim u = 3.

If

(3n

the sequence

l)/w,

is

4,7/2,10/3,... and

we can show that

If the limit of a sequence exists, the sequence is called convergent; otherwise it is

A sequence can converge to only one limit,


See Problem 8.

called divergent.

unique.

i.e.

if

a limit exists

it

is

A more intuitive but unrigorous way of expressing this concept of limit is to say
that a sequence Ui, Ui, Us,
has a limit I if the successive terms get "closer and closer"
to I. This is often used to provide a "guess" as to the value of the limit, after which the
definition is applied to see if the guess is really correct.
.

One should observe the similarities and differences between limits of functions and
sequences. In defining lim f{x) = I, the limit I is attained for all possible approaches to
infinity.

In defining lim f(n)

I,

the limit

need exist only along a certain approach to

n-+oo

infinity, namely along the positive integers.


Other possibilities present themselves. For
example, in some cases it may be important to consider the limit of f{x) as x approaches <
(or in fact any number Xo) along a sequence of rational numbers.

THEOREMS

SEQUENCES

If

on LIMITS of
lim a = A and lim

1.

lim (a

lim a

2.

lim

lim a

+ b) =
(a &)
=

B,

lim b

then

lim 6

= A+B
= AB
41

SEQUENCES

42

3.

lim (&)

4.

lim y-

(lim

lim a
ttn

,.

n~*

=B

v^

A v^ 0,

lim ^^ does not exist.

and

A = 0,

lim

^-^

may

or

may

not exist.

A,

for p

= any

real

number

if

A"

p^,

for p

= any

real

number

if

p^ exists.

exists.

n-^oo

==2)"^""

lim p""

6.

6n

and

(lim an)"
^

n-+oo

i^

B=

If J5

lim a^

CO

If

5.

iTzrir

= AB

(lim b)

ff)

[CHAP.

INFINITY

We
ber

write

lim an

(depending on

<x>

M) such

number

for each positive

if

that

a>M

for

all

we can find a positive numSimilarly we write lim a = -^

n> N.

n-^oo

for each positive number


we can find a positive number
such that a <
for all
n>N. It should be emphasized that =o and = are not numbers and the sequences are
not convergent. The terminology employed merely indicates that the sequences diverge
in a certain manner.
if

BOUNDED, MONOTONIC SEQUENCES

M
M

If Un^M for n l,2,S,..., where


the sequence {u} is bounded above and

quence
If

is

bounded below and

m^Un = M
Un-n

= Un

is

is

is called bounded, often indicated by \Un\


bounded, but the converse is not necessarily true.

the sequence

convergent sequence
If

a constant (independent of n), we say that


called an upper bound.
If Un'^m, the secalled a lower bound.
is

is

the sequence

is

called

monotonic increasing;

if

u+i>Un

S P.

Every

it

is

called

if

u+i<Un

strictly increasing.

it is

Similarly if Un+i^Un the sequence


strictly decreasing.
Examples:

1.

is called

The sequence 1,1.1,1.11,1.111,...

monotonic decreasing, while

is

bounded and monotonic increasing.

It is also

strictly increasing.
2.

The sequence

1,

1,

1,

1,

1,

Is

bounded but not monotonic increasing or decreas-

ing.
3.

The sequence 1,
However,

it is

1.5, 2, 2.5, 3, ...

is

monotonic decreasing and not bounded.

bounded above.

The following theorem is fundamental and is related to the Weierstrass-Bolzano


theorem (Chapter 1, Page 5) which is proved in Problem 23.
Theorem. Every bounded monotonic (increasing or decreasing) sequence has a limit.

LEAST UPPER BOUND

number

and

GREATEST LOWER BOUND

of a

SEQUENCE

upper bound (l.u.b.) of the sequence {u}


e for any > 0.
ti = 1, 2, 3,
while at least one term is greater than
A number m is called the greatest lower bound (g.l.b.) of the sequence {m}
while at least one term is less than m + c for any > 0.
% = 1, 2, 3,
.

is

called the least

if

Un

= M,

Un

= m,

if

CHAP.

SEQUENCES

3]

Compare with the


Page

definition of I.u.b.

and

43

g.l.b.

numbers

for sets of

in general (see

5).

LIMIT SUPERIOR, LIMIT INFERIOR

number I is called the liTfiit superior, greatest limit or upper limit (lim sup or lini)
of the sequence {m} if infinitely many terms of the sequence are greater than I - 1 while
only a finite number of terms are greater than l + t, where is any positive number.

number

called the limit inferior, least limit or lower limit (lim inf or lim)

is

of the sequence (m}

only a finite

if

number

many terms of the sequence


I e, where is any

infinitely

are less than

of terms are less than

positive

+t

while

number.

These correspond to least and greatest limiting points of general sets of numbers.
If infinitely many terms of {m} exceed any positive number M, we define lim sup {^t}
CO.
If infinitely many terms are less than ~M, where
is any positive number, we
If

If

lim Un

lim Un

= =, we

n-+

= <.

lim inf {m}

define

we

lim sup {Un}

define

lim sup {Un}

define

lim inf {m}


lim inf {Un}

oo.

= <.

oo

Although every bounded sequence


lim sup and lim inf.

not necessarily convergent,

is

it

always has a

finite

sequence [Un] converges

if

and only

if

lim sup

lim inf m

is finite.

NESTED INTERVALS
Consider a set of intervals [a, b], n = l,2,3, ..., where each interval is contained
and lim (a - bn) = 0. Such intervals are called nested intervals.

in the preceding one

We can prove that to every set of nested intervals there corresponds one and only
one real number. This can be used to establish the Weierstrass-Bolzano theorem of Chap. 1.
(See Problems 22 and 23.)
CAUCHY'S CONVERGENCE CRITERION
Cauchy's convergence criterion states that a sequence {} converges if and only if
for each >
we can find a number
such that \Up -Ug\ < e for all p,q> N. This criterion has the advantage that one need not know the limit I in order to demonstrate
convergence.

INFINITE SERIES
Let
Si

where

ui, U2, Us,

= Ui,

S2

S, called the

The sequence

Ui

be a given sequence.

+ U-z,

Ss

= Ui+U^ + Ua,

nth partial sum,


Si, S2, S3,

Form a new

is

is

the

sum

.,

sequence

Si, S2,

Ui+Uz + Ua-i

of the first

n terms

-S3,

where

\-Un,

of the sequence {u}.

symbolized by
00

Ul

which

is called

an

infinite series.

+ U2 +
If

Us

lim Sn

= S

^ Un
n=l
exists, the series is called

and S is its sum, otherwise the series is called divergent.


Further discussion of infinite series and other topics related
in Chapter 11.

convergent

to sequences is given

SEQUENCES

44

[CHAP. 3

Solved Problems

SEQUENCES
1.

Write the

first five

'*'**

^'\sn-{-2j
fl

( D"!
,

(&) i

terms of each of the following sequences.


5' 8' 11' 14' 17

0, gi, 0, gs

Ans. ^,
2' 2'4' 2'46' 2'4'6'8' 2

K-[
2>4-6--2nJ

(c) i

o.A. I

(^){f + 1 + 1 ++!;}

^^'

^'^^

(-!)"-' a;"-' 1

(2m-1)!

Ans. jj,

g
'^'*'

i+

i'

-jg"

1!'

-'

a;"

3!

5!'

'

4+i+

i'

7!

We

define

i+i+4+

4'

10

i + i + i + i^ +

i^'

'
9!

'

= l'2'3-4..-%. Thus
= 1.

Note that n!
etc.

= 1,

1!

3!

= 1'2'3 =

6,

5!

= l'2-3-4'5 =

120,

Two

students were asked to write an nth term for the sequence 1, 16, 81, 256,
and to write the 5th term of the sequence. One student gave the wth term as Un n*.
The other student, who did not recognize this simple law of formation, wrote Un =
lOn^ S5n^ + 50n 24. Which student gave the correct 5th term?
which agrees with the first
If u = n*, then
mi = 1* = 1, mj = 2* = 16, M3 = 3* = 81, M4 = 4* = 256
.

four terms of the sequence. Hence the


If
first

ttn

lOn' 35%^ + 50n 24,

student gave the 5th term as Ms

first

ui = 1,

then

U2 = 16,

Us = 81,

U4 256

Hence the second student gave the 5th term as Ms

four terms given.

= 5* = 625.

which also agrees with the

= 601.

Both students were correct. Merely giving a finite number of terms of a sequence does not define
a unique wth term. In fact an infinite number of mth terms is possible.

LIMIT
3.

SEQUENCE

of a

sequence has

its

nth term given by Un

3^

17?\

-i

(a)

Write the

1st, 5th, 10th,

100th,

1000th, 10,000th and 100,000th terms of the sequence in decimal form. Make a guess
as to the limit of this sequence as w-* >. {b) Using the definition of limit verify that
the guess in (a) is actually correct.
(a)

% =

.22222

M = 10
.64444

.56000

good guess is that the limit is .75000 ... =


of n that a possible limit may become apparent.

We
on

c)

must show that for any given


such that

Now

|m

3n 1
r-3
4m + 5

-;
I

-||

<

for

all

>

|4(4w

= J(19/4e
complete.

Choosing iV
is

Note that

if e

= .001

when

4.5>g,

5), we

see that

(for example),

of the sequence beyond the 1186th

+ 5)| <

i^>l,
and the proof

10,000
.74988

it is

n=

100,000
.74998

only for large enough values

small) there

is

a number

(depending

n> N.

-19

3
-r

how

(no matter

Note that

n =

1000

.74881

(6)

n=

n-lOQ
.73827

term

|m

19
4(4m +

5)

>l@-5)

fl <e

for

N = 1(19000/4 - 5) =
differ

from f

all

n>N,

so that

lim m
''"*

=f

1186J. This means that all terms


value by less than .001.

in absolute

CHAP.

4.

SEQUENCES

3]

lim =

Prove that

We

on

e),

we

^
Prove

see that

,,

We
n>N.

(5

+2-10"

10")

-3

>

1/c,

10"

(a)

is

If for each positive

an> M

for

all

>

>

10"

7/3e

a number

<

3-10")

n>N.

(depending

0.

+
+

such that

- 5,

when

>

10"

is

2'10''

S-IO"

<

for

all

>

we can

iq")

n >

or

0,

''^^^'^^'i

''

''

log.o {i(7/3e

< e < 7/15.

i.e.

- 5)} =

N,

If

a 7/15, we

see

we can

we

write

a.

(1

- 2n) =

->.

(depending on

M) such

that

00,

(&)

lim

number

M,

n > if^^^^
+ l)
' \ log 3
/

00

00

- 1)

(2n

lim S^"-!

(a)

find a positive

lim

write

when

a<M

n> 0.

all

we

then

result.

7/3e - 5 >

real only if

4-

3(5

- 5)

l(7/3e

meant by the statements

3'^""'

log 3

>

log

find a positive
lim = (.
n^

In this case, 1

for all

iV^^Mj

Choosing

and thus establishing the required

each positive number


for all n> N, then

If for

n>nfj

or

is

(6)

0| <

le/n"

-7

number

n> N,

In this case,

such that

proving that lim (c/W)

there

3(5

Note that the above value of


1 + 2-10"
2
that
< e for
5 + 3-10"

Explain exactly what

'

proving the existence of

6.

2
3

3. iQn

3-10'

a number

is

W>^-f

n>N,

for all

+
+

there

i.e.

e,

must show that for any

XT
^^
3

,.

lm

that

<

-^

<

[c/Wl

>

are constants (independent of n).


Vp

1^ <e when

Now

p>0

where -0 and

must show that for any

45

i.e.

number

N.

M) such

(depending on

that

00

- 2n < M when 2n-l >

or

n > ^{M +

1)

N.

should be emphasized that the use of the notations and > for limits does not in any
way imply convergence of the given sequences, since > and -=o are not numbers. Instead, these
are notations used to describe that the sequences diverge in specific ways.
It

Prove that lim x"

if

<

\x\

1.

n-+oo

Method

1:

We

can restrict ourselves to a; v^


exists AT such that

show that there

Dividing by logic

Method

which

is

<

a;

for n

negative, yields

n >

the result

> AT. Now


2p^'u|

is
|a;"|

clearly true.

|a;i"

<

>

0,

we must

nlogio

]a;|

<

Given

when

^' Proving the required

result.

2:

Let

\x\

|a;"|

THEOREMS
8.

|a;|,

since if
|a;"|

on

Prove that

We must

if

1/(1

+ p), where

|a;|

1/(1

LIMITS

of

if

>

+ j,)" <

0.

By

1/(1

Bernoulli's inequality (Prob. 31, Chap.

+ np) <

for

all

n> N.

SEQUENCES

lim Un exists,

show that

it

lim Un

must be unique.

- h and

lim Un

= h,

then h

- h.

Thus

1),

lim a"

we have

0.

logioe.

SEQUENCES

46

By

any

hypothesis, given

|m

>

<

li\

we can

find jV such that

n>N,

when

-^e

[CHAP.

\u

hi <

n> N

when

-Je

Then
Ih
i.e.

9.

If

[li

h\ is less

h] =

than any positive

= A and

lim an

We must
n> N. From

lim 6

hypothesis, given

>

prove that lim

> 0, we can
we have

3,

B)\

(1), (2)

and

we can

10.

for

lim an

Thus

li

= h.

= A + B.

b)

such that

(6-B)|

^E

|(a.

|a-A|

6)

(A

+ B)\ <

for

all

16-B|

(i)

^e

foralln>iV,
foralln>iVi

(2)

Ae
2

forallw>Ar2

(3)

- (A+B)l <

Prove that a convergent sequence


Given

Ni and Ni such that

find

le

le

f or all

ti

>

A^

result follows.

bounded.

is

that there exists a positive number

we must show

A,

Thus the required

chosen as the larger of Ni and Ni.

iV is

-|e

(5),

!(an+6)

where

<

\Unh\

{an

N>0

find

|(a-A)

an-A\
a-Al <
6-B| <
Then from

U\ +

\li

(however small) and so must be zero.

B,

show that for any


inequality 2, Page

\{an+bn)-(A +

By

+ Un h\ =

\li~Un

such that

< P

\a\

Now

all n.

la|

But by hypothesis we can

It follows

that

\an\

<P

for

find

all

|a-A + A|

such that

|a|

< e+
we

if

A| <

|a

for

\A\

choose

|a-A|
for

k>

all

\A\

all

i>Ar,

i.e.,

AT

as the largest one of the numbers

ai, ai,

.,

on,

e+\A\.

11. If

lim bn
Since

=B

-0,

prove there exists a number

B = B-6+6, we

Now we

have:

can choose AT so that

K]

\B\

{!)

\B -

such that

|B-fc|

\bn-B\

<

^\B\

>

|61

^\B\

for

all

n>N.

|6|.

for

all

n> N,

since

lim

K-B

by

hypothesis.

Hence from

12. If

lim On

(i),

= A and

i\B\

lim6

B,

>

|6|

l|B|

for

prove that lima6

w>

all

AT.

= AB.

n-+oo

since

lim o

= A

\an(bn-B)

and

\bn-B\ < 2p for

ATa.

or

\bn\

have, using Problem 10,

la6-AB|

But

<

n-^oo

n-^oo

We

\B\

Hence from (1), |a6-AB|


Thus the result is proved.

lim b.
all

<

+ B{an-A)\

B, given any

n>Ni
le

^ |a| |6,.- J51 + IBj |a- A|


S P\b-B\ + (|B| + l)|a,-A|

+ |e =

\an-A\ <
e

for

all

>

we can

find

for

^^|g[^

n>N,

(1)

Ni and Ni such that

all

n> N.

^^

where

AT is the larger of

Ni and

CHAP.

13. If

SEQUENCES

3]

lima

We

(a)

= A and

= Sv^0,

lim6

must show that for any given

prove

>

find

11
1
_ IB - 6|
\k~B\ - \B\ \K\ <

By

any e>0, we can

hypothesis, given

Problem

we can

0,

and the proof


(6)

Prom

part

Ni such that \hn~ B\ < ^Bh for all n>Ni.


N^ such that \K\>t\B\
for all n>N,^ fsee
^i

find

find

and Ni, we can write

iB
^^ Ti^i 77^
\B\-i\B

and Problem

lim

n^oo On

lim
-+

T-

Ony

lim a
n-

(a)

(6)

lim

~ ^^

n^5n' + 2n-6
lim } ^(^

3-5/n

lim

n"t,5

+ ^) -

'^^

+ 2/n-6M

n^

,;^ J

_
5

(c)

lim(V^H^-V^) =

(rf)

iim%^i21 =

lim

lim

= A 4 =
B

_
~

,.

3
5
l

0)

+ l/n + 2/n^

(1

w>A/',

we can

~2n

rn
^-^^

^^>

Urv,

2n' - 4n'
3' + n'-10

,.

+
+

2'10
3.10"

number

by choosing

16
81

= (I)'
3

22

+
+

S-lO-'

respectively, the limit does

larger than any positive

'

2m

2/w' - 4M
+ lM*-10/n' =

10""

,.

= ir.

made

'

lim

(.>r.l^)*
_ ,.
- i

^^

"^^^

~2

write, if desired,

.-(StI)'

+ 4/-

n-tco

">

0)

1
(V^r+l-V^)^^^5iV^ = H^
^
Vw + 1 + a/m
"-^VJT+T+Vn

Since the limits of the numerator and denominator are 3 and


not exist.

2nl ^

4
B

limits.

(1

^^^<i^

n> N

all

41).

3 +
+ +

+ n' + 2n

lim

n-*ooOn

Evaluate each of the following, using theorems on


^^'

for

'

we have

12,

This can also be proved directly (see Problem

14.

as

complete.

is

(a)

\B\\K\

(1)

\K-B\

= -

(i)

AT is the larger of A^i

if

lim ?^

(b)

forallw>iV

'

""*"

11).

Then

=B^

lim 6

= 4,

such that

Also, since

lim

(a)

we can

0,

47

(Compare with Prob.

5.)

BOUNDED MONOTONIC SEQUENCES


15.

Prove that the sequence with nth term


(&) is
(a)

bounded above,

{m}

is

(c)

is

monotonic increasing

Un

bounded below,
if

m+i

m,

= g^~^

(d) is

m = 1, 2, 3,

(a)

bounded,
....

Now

is

monotonic increasing,

(e)

has a

limit.

[CHAP.

SEQUENCES

48

+ l) +

2(w+l)

which

11,

"

if

"'

llw

10

-7

3n + 2
+5
S en" - llw - 35, i.e. -10
3

Reversal of steps proves that 2

true.

is

Since this particular sequence


Any
Mn s -1, M = 1,2, 3,
.

(d)

&n^

2m

2nonly

-35,

wliich

true.

-4n <
(c)

"
" and

If

Thus by reversal of steps in the inequalities, we see that {%} is monotonic increasing.
Actually, since 10 > 35, the sequence is strictly increasing.
By writing some terms of the sequence, we may guess that an upper bound is 2 (for example).
To prove this we must show that m s 2. If (2to - 7)/(3ti + 2) ^ 2 then 2n - 7 ^ 6m + 4 or
is

(6)

- 5)(3n + 2)

(2n

or

2n-

- 3n + 2
a (2m - 7)(3m + 5),

3{n

is

an upper bound.

monotonic increasing, the first term -1


number less than -1 is also a lower bound.

a lower bound,

is

is

i.e.

Since the sequence has an upper and lower bound, it is bounded. Thus for example we can write
^ 2 for all n.
Since every bounded monotonic (increasing or decreasing) sequence has a limit, the given sequence

1m|
(e)

has a

16.

A
(a)

2wlim
X 4 3w + 2

In fact,

limit.

sequence {%}

defined

is

Prove that liniM

2-7/w

lim

+ 2ln

by the recursion formula Wn+i =

exists.

Find the limit

(6)

Mi-1.

^JZUn,

in (a).

n-+oo
(a)

of the sequence are

The terms
The

term

TCth

induction (Chapter

By Problem

m..

x
.

lim w + i

Since

excluded since m s
1/2+1/4+ +1/2""'
lim 3

\/3m2

3"^ + "*,

as can be proved by mathematical

monotone increasing.
i.e. m is bounded above.

Hence m,

is

bounded

bounded and monotonic increasing.

lim VSm^,

= ,._
hm

we have

V^

and

= 3.

(The other

1~1'2"
3

_
_

_
_

lim (1- 1/2")

3n^<

01

_^

Verify the validity of the entries in the following table.

2, 1.9, 1.8, 1.7,

.,

1,-1,1,-1

\,-\,l,-h
.6, .66,

By

(n-l)/10

1(1-1/10"),

lim

.,

Letting x

a;)"

+-

mx

if

+
,

...

...

(-l)"w,

the binomial theorem,


(1

...,{~\Y-'l{n^^),

-1, +2, -3, +4, -5,

Prove that

(-l)-',

.666, ...,

Increasing

Monotonic
Decreasing

Limit
Exists

No

No

Yes

No

Yes

No

No

No

Yes

No

No

Yes

(0)

Yes

Yes

No

Yes

(|)

No

No

No

Monotonic

Bounded

Sequence

18.

3"^_ Ma
+1/2

i).

,.

Another method:

17.

3,

1,

0, is
,

is zero).

required limit.

possibility,

is

3'

limit exists, since the sequence is

Thus a
Let X

Then the sequence

Chapter

14,

(since a lower bound

(6)

V'Sitl

31/2+1/4+
3

M2

1,

1).

m+i

Clearly,

given by

is

ux

e.

is

a positive integer (see Problem 95, Chapter

n{n-V)
^
'

^,

x"

+
,

n{n
-^

- l)(w - 2)
g,

..3

x"

w(w

1/w,

04)"=
=

No

n(w
^

+ '^i+

1)

2!

J^

n(n

1),

1)---(m

-,

1)---(w

n+

n\

if-

-'+if(-0-M'-i)(-l
n!

w/V

- w+

1)

1)

1^

CHAP.

SEQUENCES

3]

49

Since each term beyond the first two terms in the last expression
follows that the sequence m is a monotonic increasing
sequence.
It is also clear that

nj
by Problem

Thus
value of

19.

Chapter

14,

= 2.71828.

Prove that

^3!

..

2^

2"-i

lim(^l

+-y =

largest integer

where

e,

then w

x,

n + iy

.-\

a;-* .

a;

in

any manner whatsoever

Problem

lim
lim

The

gm+

and

( 1

n+X )

.^\^

+iY =

( 1

(i.e.

not neces-

18).

(^1

+ -V^

(^1

+ i Y*'

y^'-^n^x)

lim(l+iY''= lim(l+lYCl +

and

follows that

e.

Since

it

-^

-^

n, it

1.

sarily along the positive integers, as in


If

^^

n\

bounded and monotonic increasing, and so has a limit which we


denote by

is

w.

2!

an increasing function of

is

i')

e.

LEAST UPPER BOUND, GREATEST LOWER BOUND, LIMIT SUPERIOR, LIMIT

INFERIOR
20.

Find the
2)
{a)

2,
1

(a) l.u.b.,

1,

1,

u.b.

= 2,
e

than 2
(6)

g.l.b.

is less
(c)

(namely
e

>

lim inf or
e

(InH),

and

lim inf (lim)

(d)

>

while at least one term (the 1st)

1,

since infinitely
the sequence)

is

greater

while at least one term (the 2nd)

of the sequence are greater than 1 - e for any


while only a finite number of terms are greater than 1 + e

many terms

-1, since infinitely many terms of the sequence are less than -1
+ e for any
-I's in the sequence) while only a finite number of terms
are less than -1 - .
(namely the 2nd term).
all

results are

(6) g.l.b.,

shown

(c)

2,1.9,1.8,1.7,

l.u.b.

...,2-(-l)/10

1,-1,1,-1, ...,(-i)-i,

.,

lim inf (lim) for the sequences

g.l.b.

lim sup
or lim

...

1(1-1/10"),

-1, +2, -3, +4, -5,

(d)

-1

-i

...

(-l)m,

lim inf
or lim

..

,(-l)"-V(n + l), ...

.66, .666, ...,

lim sup (lim), and

in the following table.

Sequence

h-h\'-h

for the sequence

Hm =

Find the (a) l.u.b.,


Problem 17.

.6,

lim sup

all I's in

The

(c)

(namely the 1st term).

(namely

for any

21.

(&) g.l.b.,

1, ....

-2, since all terms are greater than or equal to -2


than -2 + e for any e > 0.

for any

c>0

1,

since all terms are less than or equal to 2


for any e > 0.

lim sup or lim

e>0
(d)

1,

1,

none

+00

SEQUENCES

50

[CHAP.

NESTED INTERVALS
22.

Prove that to every set of nested intervals


one and only one real number.
By

a + i

definition of nested intervals,

[a, &],

n=

l,2,3,

n=

l,2,3, ...

6+i

a,

6,

there corresponds

.,

and lim
nf

(a

6)

0.

CO

Then ai ^ a ^ 6 g 6i, and the sequences {a} and {6} are bounded and respectively monotonic
increasing and decreasing sequences and so converge to a and 6.

To

shovsr

a=b

that

and thus prove the required

6-a =

(6-6)

|6-6|

+
+

such that for

all

\b-a\

Now

given any

so that

23.

from

(2),

>

\b

find

|6-6]

<

we can

0,

a\ <

Since

e.

any

]6-a|

>

<

+
+

{an

-a)

{1)

|a

- a|

(2)

AT

<

la-al

e/3,

Page

(see

note that

e/3

{3)

number, we must have

positive

Prove the Weierstrass-Bolzano theorem

we

(6-a)

16.-al

e/3,
c is

result,

a =

or

a=6.

5).

Suppose the given bounded infinite set is contained in the finite interval [a, b]. Divide this interval
into two equal intervals. Then at least one of these, denoted by [ai, bi] contains infinitely many points.
Dividing [ai, 6i] into two equal intervals we obtain another interval, say [02,62], containing infinitely
many points. Continuing this process we obtain a set of intervals [a, 6], w = 1, 2, 3, ., each interval
contained in the preceding one and such that
.

hi-ai

from which we see

- a)/2, 62-02 = (61 - a,)/2 =


that lim (6 a) = 0.
(6

n-+

(6

- a)/2^

- a =

(6

- a)/2"

00

This set of nested intervals, by Problem 22, corresponds to a real number which represents a
and so proves the theorem.

limit point

CAUCHY'S CONVERGENCE CRITERION


24.

Prove Cauchy's convergence criterion as stated on Page


Suppose the sequence {} converges to

Necessity.

43.

Then given any

I.

>

0,

we can

find

such that
\up

-l\

p> N

Then for both

\u-U,\

<

and

e/2

q>

for all

p>N

and

jit,

<

e/2

\l-U,\

<

l\

for

all

q>N

N, we have

\(u,-l)

(l-U,)\

\u^-l\

e/2

e/2

and any e > 0. Then all the numbers uh,


Suppose \Uf - m,] < e for all p,q>
a finite interval, i.e. the set is bounded and infinite. Hence by the Weierstrass-Bolzano
theorem there is at least one limit point, say a.
Sufficiency.

un + 1,

...

If

lie in

is

the only limit point,

we have

the desired proof and

lim m

a.

Suppose there are two distinct limit points, say a and 6, and
suppose 6 > a (see Fig. 3-1). By definition of limit points, we have
\up-a\
lla

Then

since

|6-a|

<
<

a)/3
(6 a)/3
(6

for infinitely
for infinitely

many
many

values of

(1)

values of q

(2)

- it,) + (m, - Up) + {ut.-o), we


= b-a ^ \b - Uq\ + \up - Uq\ + \Uf - a\

b-a =

(6

Using (1) and (2) in (3), we see that |Mp thus contradicting the hypothesis that \Up - u,\
one limit point and the theorem is proved.

>

m,1

<

(6

have

'*~3~*'

'*~s~*'
1-

^.

g_j

{3)

- a)/3 for infinitely many values of p and q,


> N and any e > 0. Hence there is only

for p,q

CHAP.

SEQUENCES

3]

51

INFINITE SERIES
25.

Prove that the

(sometimes called the geometric series)

infinite series

ar

ar^

^J ar"-^

n=l
(a)

- r)

converges to a/(l

< 1,

(b)

Sn

rSn

=
=

if

Let

Then
Subtract,

\r\

(l-r)S,

diverges

If

\r\

<

1,

(6)

If

\r\

>

1,

lim S

lim "^|

1.

ar^

ar"''^

ar

ar'

ar"'^

ar''

r")
1-r

a{l

f^

\r\

ar

Sn
(a)

if

by problem

7.

lim Sn does not exist (see Problem 44).


n-+
00

26.

Prove that
Since

a series converges,

if

Sn

Ui

+ u,+

...

+ u^,

its

Sn-i

nth term must necessarily approach zero.


U1 + U2+ .-. +Un-t
we have Un = S^- Sn-u

If the series converges to S, then

lim Un

27.

Prove that the


Method

Sn-i)

lim S

1-1 + 1-1 + 1-1+..

series

S-S

lim Sn-i

S (-l)"""^

diverges.

"=^
1:

^~'^^"
it

lim (S

iifi
diverges.

^^ ^^'^^

^'

Method 2:
The sequence

of partial

Then by Problem 26 the

doesn't exist.

**

gums

is

1,

- 1,

- 1 + 1,

- 1 + 1 - 1,

series cannot converge,

i.e.

Since this sequence has no limit, the series diverges.

i.e.

1, 0, 1, 0, 1, 0, 1,

MISCELLANEOUS PROBLEMS
28. If

lim^^n

Let Un

prove that

I,

Vn

We

I.

Vl

iim

^i

+ ^^+-

^i^i

n
lim ^'

must show that

+ V^+ ..+Vn

Vl

+ -^^ +

'

"

+ Vz+ .+VP

^^

Vp + i

^
+

t,p +

if

lim

t;

...

Now

so that
\

vi

+ V2+

Since Jim Vn

0,

P we

|-t>P

it;i

.i;,+

choose

+ z|

can choose

...+^p|

we can

\Vp+l\

After choosing

+^>J

so that

|i;,|

|v|

e/2

so that ior
\Vl

<
+

for

e/2

e/2

{S)

and

(3), (i)
1^1
I

[i;p+2|

'

thus proving the required result.

+ Vn

|i;|

(1)

w > P. Then
+ e/2 _ ( - P)e/2

...

+Vp\

<

(*)

becomes

+ Va +

n> N> P,

+ V2+
n

Then using

kp+i|

<

K^

75"

for

n> N

SEQUENCES

52

29.

Prove that lim


Let

(1

+ n + n^Y'" =

(l

1+% + M^
Then

Prove that
The

lim

Un

l],

= -^.

lim m

for

we can prove

-^^ =

If

2|a|

lim

2'

(1

+ m) =

1.

-Un

gP

+
,

"

Mn

6(n" + w)
^
< -^^^^^^-^^

+ w + w")"" =

lim

Problem

(see

1^

(1

We

39).

n>2\a\, and

can assume a
if

we

7^ 0.

call

<

(^^'^Un.

lim m

follows that

it

7),

or

i^)"'"

Un
Un-i

'"''

2'

0.

n*- CO

71-+ 00

31.

l){n 2)

then

1,

^ <

(using Problem

-^^

large enough, say

is

Un + 2
Un + 1

^
<

n{n

a.

lim j-

that

""^

'ul

-^-gj

Thus

0.

Multiplying these inequalities yields


lim (i)"""

constants

all

Un + 1
Un

Since

WM

the binomial theorem,

n(n 1)

Un-i
the greatest integer

i.e.

Now by

0.

Then

n\

+^w(n-l)(%-2)
3!~^'*"

and

a"

M)"

>

i.

where m

+ n + n^
,

+ m

(1

result follows if

Let
[2|a|

,^

Hence lim m'

30.

+ n + n^)!/" =

[CHAP. 3

The expression

ai

indicated briefly by

a2

ai

^-q: ^-q:

where

+
,

!, a2,

are positive integers,

an example of a continued fraction.

is

defined as the limit of the sequence

ai

ai,

+ ,

ai

-\

0,2

Its value is

when

this limit

-\

as
exists,

the continued fraction is said to converge to this limit.

and

The successive

terms of the sequence are called the successive convergents of the continued fraction.
repeat after some point, the continued fraction is
In case the constants ai, a2,
Given the recurring continued fraction
called recurring.
.

(a)

Find the

first

A_LJ.

2+ 2+ 2+

ten convergents and guess

at a possible limit.

(6)

Assuming that

the limit exists, find its value.


(a)

The first convergent


The second convergent

=
=

2
2

1/2

The third convergent

The fourth convergent

^2/5

The

29/12

fifth

convergent

Similarly,

if =

2..0....

From
places

we

is

S=..4U2...,
it is

5/2

,^

find for the sixth

the results

2.4142.

=
1

2.5
1

29
12
70
29

'^

~
^

5/2

12
~5

^'^

2.4166...

2.4137...

through tenth convergents respectively the values

S = ""^-- ^ = ^-^-.

iS=^-^-

reasonable to guess that the required limit accurate to four decimal

CHAP.

SEQUENCES

3]

It is of interest to note that if PJQ and


respectively, then the (n + 2)nd convergent is

P. + 2

Pn+JQn+i are the nth and

any continued

in the case of

+
+

2P+I
2Q.+1

Q.+2

For the general result

53

(w

+ l)st

convergents

P
Q
Problem

fraction, see

76(a).

Assume

the limit to be given by x. Then clearly we must have x = 2 + 1/x. Thus =' 2a; - 1 =
X = ldz^/2. Since the limit cannot be negative, it must be 1 + \/2
This agrees with the
guess in (a), since ^/2 = 1.4142 approximately.

(6)

or

Note that this continued fraction can be defined by the recursion formula
Un+i

and

lim u

if

exists, this yields

a;

1/m,

l/x

Ml

as above.

Supplementary Problems
SEQUENCES
32.

Write the

first

^^

(n\

cos

<{^.

<-'ts^}
Avs

four terms of each of the following sequences:

VS_

2x
3=

1^

1!'
33.

^'

8'

5=

'^"'>

4!

LIMITS

of

+ y/5),

a;^

cos

a;

1^
'

a;^

is

|(1

^'

2'^

'

a;^

3^^

cos
'

x^

4a;

+ 4^

T~' 1T3' 1.3-5'l-3-5-7


first 5

terms are indicated and find the 6th term:

(c)

|,0,|,0,f,

...

- lu+i + m and ui = 1,U2 = 1. (a) Find the


that the nth term is given by m = (a" - 6")/\/5 where

the sequence {m} where u+2

- VS).

(6)

Show
Ans.

(a)

1, 1, 2, 3, 5,

definition of limit, prove that:

tS =

cos 3*

2a;

SEQUENCES

Using the
('^^

cos
^"^

7=

(6)1,0,1,0,1,...

terms of the sequence.

1(1

8a;=
'

wa;

''

__L

iT'TI'Tf'

The Fibonacci sequence


first 6

35.

4x'
'

Find a possible nth term for the sequences whose


('^^

34.

-^
.
2!'3!'

^'^{.r^^i^''^:'!.

<"{<?!;.}.

(*)

1 2-V.=

1,

(c)

lim

(d)

lim

5i^

36.

Find the least positive integer

= .001,

(6) 6

Using the

38.

Prove that

39.

Prove that

40

If

Ans.

.0001.

K| =

lim

if

502,

(a)

then lim m

prove that

I,

lim

(a)

lim

Give a direct proof that

cm,,

Prove that

lim 3"

(a)

If

>

1,

(6)

5002,

<

for

w > AT

all

if

(a)

(c)

lim ul

.01,

cannot be ^.

V^,

= A

lim r"

and

lim {%)""

(6)

<,

lim (f )

(c)

1,

=B #

lim 6
n-+

n+oo

1,

lim ul

(6)

l^,

=V

0.

lim a,

if

any constant,

c is

0.

00

0.

n-*oo

-foo

prove that

50,002

(c)

- l)/(3w + 4)

where

cl

= A/B

a/fe

n-*oo

43.

+ 2)/(n - 1) -

Is the converse true?

0.

n-^- 00

42.

(3n

lim (2w

lim Vwii

(d)

a positive integer,

is

does not exist.

lim ( l)"n

lim M

where p

(c) 6

such that

definition of limit, prove that

37.

41.

[CHAP.

SEQUENCES

54

carefully explaining the significance of this statement.

w-^oo

44-

If

\r\

>

prove that lim r^ does not

1,

exist.

ni->-eo

45.

Evaluate each of the following, using theorems on

Wlim -^-^rr'"'
(6)

lim

Ans.

(o)lim

^ (3-\A^)(V^ +
-3/2,

(a)

2)

-1/2,

(6)

(d) lin,

(c)

(d)

V3/2,

limits.

^^":-^" + ^

.^^Zll'}^!-.
-15,

(e)

1/2,

(/)

ie)

lim (V^iFT^

(/)

(2"

n)

+ 3")"

BOUNDED MONOTONIC SEQUENCES


46.

Prove that the sequence with wth term u - Vn/(n


(d) has a limit.
(c) is bounded above,

+ 1)

(a) is

monotonic decreasing,

(6) is

bounded

below,

n2 + w3 + n

47
*'

Tf

"

^^

48.

If

M+i

= Vwn +

49.

If

M+i

= Uun + p/iin)

used to determine
50.

Mi

1,

l,

prove that

p>0

where

+ n'
1

lim Un

and mi

>

prove that lim m


^
,_

=
0,

^(1

exists

and

lim u

prove that

Show how

Vp-

monotonic increasing (or monotonic decreasing) prove that SJn, where S


also monotonic increasing (or monotonic decreasing).

Find the
(a)
(6)

1,

|,-J,f,-|,

Ails.

52.

l.u.b., g.l.b.,

-1, J, -i.

(a)

this can. be

\f2.

mi

+ Ma +

LEAST UPPER BOUND, GREATEST LOWER BOUND, LIMIT SUPERIOR, LIMIT INFERIOR
51.

1.

+ \/5).

If M is
is

and

between

lies

. .

lim sup

(-l)V(2n -

...,(-l)''+'(n

^,-1,0,0

(Iim),

(6)

1),

lim inf (lim) for each sequence:


(c)

+ l)/(w + 2),

1,-1,1,-1

Prove that a bounded sequence {m}

(c)

is

...

id)

none, none,

convergent

if

1,

-3,

5,

-7,

...

1,4,1,16,1,36,

+,-"
and only

(d)

if

(-1)-' (2n

...,m' +

none,l,+,l

lim m

- 1),

'-"",...

lim m.

+ ttn,

CHAP.

SEQUENCES

3]

55

INFINITE SERIES
00

53.

2 (I)".

Find the sum of the series

n=l

54.

2 (-1)"-V6".

Evaluate

Ans. 2

''

Ans. 1
o

n=l

ji^ + -J + -J + _1_ +

_.,.,

55.

Prove that

56.

Prove that multiplication of each term of an infinite series


by a constant (not zero) does not affect the
convergence or divergence.

57.

Prove that the series

1+| + |+. ..+! + ...

diverges.

Then prove that

\S2n

- S| >

1, giving

MISCELLANEOUS PROBLEMS
58. If 0, g M g 6 for all n>N,
59.

If

jim a

result true

60.

Jim

when

= id +

Let u.

61.

Prove that

62.

If

63.

If

(a)

lim \un+i/un\

<

\a\

1,

lim b

independent of

e is

prove that lim m

I,

prove that lim (a cosne

n,

(-!)},

Jim

>/-

<

1,

\a\

1. 2, 3,

1,

(6)

If

Jim (a +

prove that

prove that lim n'a"

S,

m.

n)^/"

lim m

+ ,+

...+

.,

where the constant p

>

66.

If {m} is the Fibonacci sequence (Problem


34), prove that

lim m+,/m

Prove that the sequence

is

If

69.

If

70.

Prove that

a.

|m|

6 for all

|i;|

71.

Prove that
fining the

72.

73.

Show

lim

> AT and

[a,6],

lim

SJn =

(1

+ l/n)"-, r. =
S 1.]

1,2,3, ...

J-(l

+ Vs).

a monotonic decreasing sequence whose

lim a^

A, lim 6,

B,

prove that lim m

0,

where a

...

prove that

A^B.

0.

+ 1^ =
w/

(1

1/re)"

and

(l

+ l/re)-,

is

a set of nested intervals de-

e.

Prove that every bounded monotonic (increasing or decreasing)


sequence has a

limit.

Verify the values of each of the following continued fractions.

()3

+ 2T3T + 2i--

(^)"

i^a^--- -

X.

m/m-i

i(l + 1 + ^ +
\
2
3

number

Is the

1.

that

and lim v

0.

0.

sin 1/n

re

0.

Prove that lim n


* CO

68.

6. sinne)

0.

65.

[Hint:

I.

where a and p are constants.

Prove that lim

IS e.

prove that

64.

limit

!.

depends on w?

m-*oo

67.

a contradiction with Cauchy's convergence criterion.]

and lim a

and

0,

= 1+1+1+... +

Let S

[Hint:

+ Vr5)

4(3

i(

+ V^M^)

,e)a

(c^)-l

^...

+
1

...

f + VfTf
1

74.

[CHAP.

SEQUENCES

56

Express
Ans.

174/251,

(a)

Vs,

(b)

(a)

1+ 2+ 3+ 1+ 5+ 1+

(&)

[Hint:

In

+ 1+ 2+ 1+
(6)

(c)

and

\/6,

^"^

2+-

+ (V3-i)

(V3

+ i)/2

than ^/^ (namely

less

^^ 25+ 7+

^g_^

:^_^

1) to

obtain

+ iT^vTiy = '-^u^T^
(^f3 + l)/2 (namely 1) to obtain
=

+ (V3-i)/2

Then add and subtract the greatest integer in

V5 +

rf

in

Then add and subtract the greatest integer

^ 2+ 4+ 2+ 4+

add and subtract the greatest integer

V3

as continued fractions.

3.14159

(d)

+ (V3-l)

\/3

^;^

(namely

^7^

2)

'"-vtl

to obtain

'"-^^^2

after which repetition occurs.]

75.

Given the continued fraction

'^i

^ ^ ^ '

o^"

>

''*'<'^^

0'

'^^^^

convergent

is

PJQn, prove

each of the following and illustrate by means of examples.


(a)
(6)
(c)

(d)

(e)

76.

(a)

OnPr^-l

Q.

(-1)""'

UnQn-l

Qn-2

PnQu-1
fraction.
The successive convergents are alternately less than and greater than the continued
the conThe convergents of odd order are less than the continued fraction but are increasing;
decreasing.
but
are
fraction
continued
vergents of even order are greater than the

The continued

fraction always converges.

Prove that

PJQn and

Problem

75,

if

then

P+i
Qn+l

P+i/Q+i are two successive convergents to the continued fraction in

P.
Qn

curate to two decimal places.


77.

Pn-%,

P-iQ

an+iQl

Ans.

(6)

g
77?
~ Q

(6)

Find the

first

convergent to ^f^ which

is

ac-

26/15

constants. This is called


Let {m} be a sequence such that m+2 = awn+i + hun where a and 6 are
of the form m - r where r is a
a second order difference equation for m. (a) Assuming a solution
- ar - 6 = 0. (6) Use (a) to show that a solution
constant, prove that r must satisfy the equation r^
where A and B are arbitrary
of the difference equation (called a general solution) is m = An" -f- Br?,
different, (c) In case ri =
assumed
r'-ar-h
of
solutions
two
the
constants and r, and r^ are
in (6), show that a (general) solution is m = (A + JS)r.

78.

m+2
Solve the following difference equations subject to the given conditions: (a)
= 1 (compare Prob. 34); (h) m+2 = 2m+i + 3m, Mi = 3, Ha = 5; (c) m+2 = 4m+i
(c) it. = n 2
(6) m = 2(3)""' + (-l)""'
Ans. (a) Same as in Prob. 34,
M2

79.

(a)

Prove that the nth convergent to the continued fraction

[Hint:

(6)

80.

By

Use Prob.

73(a)

(1

yfhr

jTjl

jq:

^^

(i-\/6)'+'
(1

- VB)"

34.]

taking the limits as n -^ in

Work Problems

(i+\/5)"+^

(d)

by

first

(a),

find the value of the continued fraction.

finding the nth convergent.

= m+i+^, mi = 1,
- 4m, mi = 2, m^ - 8.

chapter 4
Derivatives
DEFINITION
Let

of a

DERIVATIVE

be defined at any point Xo in

f{x)

The

(a,b).

derivative of f(x) at

.,-,.,,

fined as

a;

= a;o

is

de-

= iim/(^o + ^)-/(^o)

nxo)
if this limit exists.

The derivative can


f'ixo)

also be defined in various other equivalent

Urn

f('^)-f('^o)

A
I.e.

function

x^x

is

RIGHT

LEFT HAND DERIVATIVES

not necessarily true (see

f{x) at

if this

as

it

limit exists.

Note that

a;

= xo

- a;o

lira

hand derivative of

fix) at

lim

/(^o

a;

In this case h

DIFFERENTIABILITY

is

is

If a function

in

an

a;

= a;o

However

= a;o

restricted only to positive values


is

defined as
,

v*/

restricted to negative values as

function f{x) has a derivative at

there.

+ fe)-/(^o)

h->oif this limit exists.

has a derivative at this point.

if it

+ fe)-/(^o)

h (= ax)

in this case

flixo)

y-^'

it

approaches zero.
Similarly, the left

Ax

defined as

is

/(^o

f^ixo)

+ Ax)-f(xo)

must be continuous
Problems 3 and 4).

the converse

The right hand derivative of

f{xo

li^
Ax.o

called differentiable at a point a;


If f{x) is differentiable at a; = iCo

is

if f'{xo) exists.

and

xo

ways; for example,

if

and only

if

f^{xo)

it

approaches zero.
fi{xo).

INTERVAL

has a derivative at

points of an interval, it is said to be differentiable


defined in the closed interval
ie [a 61
differentiable in the interval if and only if f'{x,) exists for
each x, such that

the interval.

In particular

all

a^x^b

if f{x) is

then f(x) IS
a < a;o < 6 and
If

if /;(a) and fL{b) both exist.


a function has a continuous derivative,

it is

sometimes called continuously

differ-

entiable.

SECTIONAL DIFFERENTIABILITY

A
smooth

function
in

is

called sectionally or piecewise differentiable or sectionally


or piecewise
if f'(x) is sectionally continuous.
An example of a sec-

an interval

aSxSb

tionally continuous function is

shown graphically on Page


57

26.

[CHAP. 4

DERIVATIVES

58

GRAPHICAL INTERPRETATION
Let the graph of
below.

The

2/

of the

DERIVATIVE

APQB

be represented by the curve

f{x)

in Fig. 4-1

difference quotient

QR ^
PR

fjxo

+ Ax) -

fjxo)

AX

^ ^^^

(5)

and Q of the curve.


the slope of the secant line joining points P
at the point P.
curve
the
to
PS
line
line approaches the tangent
is

A^

Ax-0

is

shown

'

PR

As Ax-

0, this

secant

Then
{6)

^^

the point P.
the slope of the tangent line to the curve at

Fig. 4-2

Fig. 4-1

An

equation for the tangent line to the curve y

given by

f(xa)

/'(a^o)

{X

f{x)

at the point

where x=^x,

- xo)

is

(7)

point and yet not be differentiable


fact that a function can be continuous at a
In this case there are ^^^
"^
there is shown graphically in Fig. 4-2.
^f
,*f"f .^^f,! reof these tangent lines are L(x,) and /+(^o)
slopes
The
PN.
and
represented by

The

PM

spectively.

DIFFERENTIALS
Let Ax

= dx

be an increment given to

Ay
IS called

the increment in y

f{x).

f(x

x.

Then

+ Ax)-f{x)

If f{x) is

()

continuous and has a continuous

first de-

rivative in an interval, then

Ay
where -^0 as Ax-0.

f'{x)AX

+ cAX

The expression

f'(x)dX

edx

(9)

CHAP.

DERIVATIVES

4]

dy

gg

f'{x)dx
(JO)
called the differential of y or f(x) or the
principal part of Ay. Note that Ay^dv in
genera
However if Ax = dx is small, then dy is a close approximation
of Ay (see Problem 11). The quantity dx, called the differential
of x, and dy need not be small.
Because of the definitions {8) and {10), we often
write
is

^
^-'-'''
-^^-0
Ax
^^_o A
emphasized that da; and dy are noi the limits of Ax
and Aj/ as Ax-^0, since these
limits are zero whereas dx and d?/ are not
necessarily zero.
Instead, given dx we determine dy from {10), i.e. rf^/ is a dependent variable
determined from the independent
variable dx for a given x.
It IS

Geometrically, dy

by the

RULES
If

g and h are differentiable functions, the following differentiation rules are

1-

~{m+g{x))

^f{:c)+^g{x)

2-

^if(^)-9{x)}

^f{x)~^g{x)

3.

d^{Cm) =

4-

a^imai^)}

^mX

^(-)^/(-)-/(-)^^(-_)

,.

li

The

cr{x)

where C

+ g{x)^f{x)

f^^)j^9{x)

if

f{u)

results {12)

where u =
dy _
dx ~
and

then

f{x),

If

f{t)

and y

dy
dx

.du

f^d^

g{x)}

d{f{x)g{x)}

df{x)

iffl'(a;)^0

and v =
dy du dv
du' dv' dx

g{v)

h{x),

(12)

then
(^^)

chain rules for differentiation of

__

dx/dy

i^^)

then

rfj//df

dx/dt

dg{x)

f{x) dg{x)

/(.),.(.)

= f'iama'ix)

^
~

f'{t)

f'{x)dx

g{x) df{x)

(iS)

^'(i)

Similar rules can be formulated for differentials.


d{f{x)

g{x)f'{x)

and dy/dx and dx/dy are related by

f-\y);

g{t),

f{x)g'{x)

.(.)n.) -

_
-

{13) are often called

dy
dx
8.

any constant

is

^^^^

where u = g{x), then


dy _ dy du
d^ - m'lhc =

valid.

f'{x)-g'{x)

composite functions.
7.

= xo

f'{x)+g'{x)

[5,(^)]2

f{u)

Similarly

C-^f{x)

<^a;[^(a;)/
6.

DIFFERENTIATION

for

/,

represented in Fig. 4-1 above, for the particular


value
is represented by QR.

is

segment SR, whereas Ay

line

For example,
g'{x)dx
{f{x)g'{x)

=
+

{f'{x)

g'{x)}dx

g{x)f'{x))dx

[CHAP. 4

DERIVATIVES

60

DERIVATIVES

SPECIAL FUNCTIONS

of

function of x; if u^x, du/dx In the following we assume that u is a differentiable


Chapter
principal values given
The inverse functions are defined according to the

1-

4(^)

2.

^z."

'

du
n"-^

<?

du

dw

du
secittantt^
,

'^-

d^^^'^''

-CSCitCOtM^

8.

j-cscu

9-

d^l^"^

10-

d^^OgeU

11.

d
= aMna^
^a"

12.
^

d
4^
dx

du

loga e dw
^ n =i 1
= -ITd^ a>0,a^l

d
^mU

17

du
1
^
d
J sec ^u
^/^^3ida;
dx

18.

d
dx

19.

d
T smh u
dx

du
^^da;

20.

-y-

cosh u

smh M -J

21.

tanh u
Tdx

22.

f cothtf
dx

23.

-V-

d
sech u
dx

- sech w ^tanh u

24.

d
V,
u
csch
T
da;

- csch u coth w

+ u^dx

dM

-^^

dx

du

da;

'^^^"""dx

csch^u^^

'^

sinh-^M

cosh- lit

27.

e" -T-

J- tanh" ^M

dit

28.

5 coth-l^t

14.

f^

J- COS
dx
d

du

1
'

-jz

dw

y/l-u^
du
1

'^^

du

1-u^dx'

1-z.^da;'

^
H>^

dtt

-,

dM

u^/u^-1'^''
1

30.

. -,

T-sech^'^i
da;

"'^

l^l<^

1
29.

^^

du

da;

da;

..

Vm^ 1

du
dx

e"

f
da;

^l+u'dx

dx
26

du

2E

f-ifM>l
\ + irtt<-l

dM

1
fi^

+ ifM>l
l-itu<
f

^Vw2-1^^
.

<^^

d
^ 1
j-cot"*
dx
.

r,

du

16.

1.

2.

:j-csch-iii
da;

M^M^ +

dtt

dx

HIGHER ORDER DERIVATIVES


or dy/dx,
interval, its derivative is given by /'(a;),_ y'
by
denoted
is
derivative
its
interval,
If fix) is also differentiable in the
denoted
the tith derivative of /(a;), if it exists, is

an

If fix) is differentiable in

where y = f(x).
,j"
f.(r\
\Jt'),
y

or
oi

by

2/'"^

the

/'"H^)'
first,

i-?^Vf^.
dx^

dx\dxj
or

0,

second, third,

Similarly

where
.

is called

the order of the derivative.

orders are given by

f'{x), f"{x), f"'(x),

Thus derivatives
....

repeated application of the


Computation of higher order derivatives follows by
ferentiation rules given above.

of

dif-

^^-^P-

DERIVATIVES

^]

gj

MEAN VALUE THEOREMS


^'

^^ ^^""^ '^ continuous in [a, b] and differentiable in


then there exists a point | in (a,h) such that
/'(|) = 0.

5!!*!'1?^''*"'-

/W-/{o)-0,

"^ *^* "*^"^^ ^(^) ^^ continuous in


exists a point ^ in (a, 6) such that

^" /^''^^

b-a
Rolle's

theorem

The result
X and xo are in

is

(16)

f{xo)

(16)

with

f'ii){x-Xo)

of the

mean

a+

differentiable in (a,

b),

in

=x

between

which case

If f{x)

and

g{x).

where Rn,

called the

and

a+

Lagrange's form:

n^l^H^
R^

Cauchy's form:

i?

...

+ f">(aKb-a)" ^ ^^

/'"-"(I) (ft-a)"^^

a^f<;o

+ !)!

/'"^"() (&-^)"(6

In both forms the values of

+ nx,)(x-xo) +

/^"^(^ o)(a;-a^o)"

w!

This

-a)

(i

^^^^

(^U)

^^

x and xo
between xo

if

/<"^"(g)(x-a;o)"^

often called a Taylor series for f(x) with a


remainder and
f(x) by a polynomial, in which case Rn is the error
term.

hm Rn =

Note

are different in general.

(%

IS

If

(18)

in either of the following forms:

Xf

f(xo)

where

and differentiable

(i7)

are continuous in
such that

'^^^
^^^^ '^^^ '^^ written in various alternative forms; for example,
^f "xl*
are in (a,b),
then using Lagrange's form of the remainder
R, we have for
fix)

^/i,

are not simultaneously zero.

/'(a;), flr'(a;)

SLXlCx

a;

a<i<b

WW)

remamder, can be written

See Problems 26, 81-84.

a;o

yields (16).

+ r(a)(b-a) +

/(a)

in (19), the infinite series obtained is


called a

is

+ !)!

V'<''<'J

used to approximate

Taylor series for /(ic)


the series is called a MacZawrm series. Such
series, called vower
series, generally converge for all values
of x in some interval, called the interval
of convergence, and diverge for all x outside this interval
(see Chapter 11 for further discussion).
In referring to Taylor's theorem of the mean,
the Lagrange form of the remainder
will be assumed unless otherwise stated.

about x

= x,.

If

a;o

= 0,

if

(a, &)

Taylor's theorem of the mean. If ^^^(x) is


continuous in [a,b]
in (a, b), then there exists a point
i in (a, b) such that
fib)

'

mean.

then there exists a point | in

where we assume gia)^g{b) and


that the special case g{x)

also called the law of the

is

9ib)-g(a)

4.

h,

Cauchy's generalized theorem of the mean.

and

(16)

can be written in various alternative forms; for


examnle
then

can also write

The theorem

[a, b\

in

(a, b),

< ^ < 1.

3.

a<i<b

f(^^

if

a special case of this where f(a)=f{b)-0.

/(a:)

We

and

and differentiable

[a. b]

(a, 0), ^I'r'"*!!!


then there

(a, h)

[CHAP. 4

DERIVATIVES

g2

SPECIAL EXPANSIONS
The following are some important expansions.
be obtained by using either {20) or {21).
1.

e^

a;

^-3T
1

~
-

a;

oT
2!

qT
3!

^^
+ ;^
n\ +

5T~7T'^'"^'

2.

sin

3.

cosx

+ x)

tan-' X

^-Y~5~^

In (1

4.

5.

In 1-3, lim Rn

-l^x^l.

for

all x.

4T

2T

Y
In

4,

'S

^+

lim

~i

fi

^"

/Y*2n2

(2n-2)!

67
X*

x^

(2n-l)!
in

in each case can

The remainder Rn

(-!)"-' a;"
y->
-

...

Further discussion of such expansions

for
is

+ R^

'

'

2^2^^^!

-1< x g 1.

In

5,

lim Rn

given in Chapter 11.

L'HOSPITAL'S RULES
If

lim /(x)

lim^

finite,

= A and
is

lim g{x)

for

are either both zero or both

= B where A and B

co/oo

form 0/0 or

often called an indeterminate of the

m-

respectively,

usually nothing indealthoug1uch terminology is somewhat misleading since there is


facilitate evaluarules,
L'Hospital's
called
terminate involved. The following theorems,
tion of such limits.
1.

and g{x) are differentiable in the interval (a,


this interval, and if sr'(x)v^O for Xv^xo, then

If f{x)

xo in

limM

b)

except possibly at a point

lim/;M

{23)

x-,x g'{x)

x-xo g{x)

and g'{x) satisfy the


whenever the limit on the right can be found. In case /'(x)
be repeated.
can
process
same conditions as f{x) and ^(x) given above, the
2.

If

lim /(x)

00

and

lim g{x)

the result {23)

>,

is

also vahd.

X-^X

X-*Xq

-, and to cases where x = a or xo= b


These can be extended to cases where x-^ or
a+ or x^ b-, are involved.
in which only one sided limits, such as x-^
, ooO, 0, 1" and - co
forms
indeterminate
Limits represented by the so-called
limits for which the above rules are
can be evaluated on replacing them by equivalent
applicable (see Problems 33-36).
Taylor's theorem of
Sometimes evaluation of such limits is facilitated by using

the mean, as in Problems 32 and 36.

APPLICATIONS
1

Maxima and minima. Suppose


= /"(Xo) = /'(xo)

that at x

/'^-''(a^o)

f{x)

has a relative

(b) f{x)

has a relative

maximum
minimum

at
at

f{x) satisfies the conditions

= l). Then
x = xo if f'-^''^{xo) <
x = xo if /""'(^^o) >

for some positive integer p (usually P


(a)

= Xo,

and

f ^'''(Xo) ^

{2J,)

'^^^^-

4]

DERIVATIVES

gg

See Problem 39. In practice, to find the


relative maxima and minima of /(x)
we
so ve the equation /'(x) = to obtain
the critical points xo and then use
Graphi(2i).
cally the necessary condition
/'(:.) =
follows, since at a relative maximum

mmimum
2.

x^xo

point

We

Rates of change.

can interpret dv/dx

:tr:^^, It "
3.

the tangent line to y

^"- *^" '

Velocity and acceleration.

J^TJ-^T
s/df^ IS its
d

= fix) must

= f'{x)

as the rate of change of v


-'
^'^")<

''-

'=

--^

If s is the instantaneous

"'^^

^'"' ^*

or
be parallel to the x axis.

*^"" ^"/^^ ^
J'\^
instantaneous
acceleration at time t.
^

"

= f(x)

'S

displacement of a particle
instantaneous velocity and

^*

Solved Problems
DERIVATIVES
1.

Let

3 + a;
= g-^,

fix)

By

A^oie;

Evaluate

3.

points
at all

^^

<.

a;

from the

using rules of elementary calculus,

(3-,,)A(3 + ^) _

nx)

/'(2)

(3

+ ^)

we

Let

/(cc)

/'(5)

(3_^)

Evaluate

/'(5)

= ii^/(5 + fe)-/(5) ^
ft

I/(. +

ft)

-/(.)

since /'(ko) exists by hypothesis.


lim/(a;o
/() is

6
(3

i,

definition.

ft

= xo, prove

- a;)'

V9 + 2ft-3

^=^

If fix) has a derivative at x

showing that

from the

v^-r^..-a,
+ 2ft-3 V9
2ft + 3
V + 2fc
i;^ V9
"
V^T2ft + 3

..

Then

+ ^)(-l^ _
~

(3-r

=2 we find r(2\-e, AH),n<.i,


indL^Lnatefy t"e PrfbS^^S

"-

3.

(3-a.)(l)-(3

where the derivative exists. Putting a;


must be careful not to apply thel

= ^2^;-!.

find

rf^

often useful, one

2.

definition.

/(^o
ji^^^

lim
""

+ 2ft -9
ft(V9T2ft + 3)
9

o
= lim
~ ^ V9T2ft + 3
tL

that /(^) must be continuous at x

+ ft)

-/(.,)

^^^^^^

^,^^^^^^

Thus

+ ft)-/(aj) =

continuous at x-x^.

or

lim

/(a;

+ ft) =

/()

_ 11
" ^

= xo.

'"

[CHAP. 4

DERIVATIVES

_.

64

m=

Let

4.

^^0

|q^

Is fix) continuous at

(a)
(a)

By Problem

(6)

/'(O)

22(6) of

= 0?

Chapter

m + h)-m
j^-^^

Does

(&)

i:..
lijw

f{x)

have a derivative at

fW-m)

a;

= 0?

x= 0.

continuous at

2, f(x) is

_
-

limi^^

which does not

fesinl/fc-0

^ limsinl
'-

'^

'^

exist.

not have
function is continuous at a point it need
This example shows that even though a
is not necessarily true.
Problem
in
3
theorem
the
of
converse
a derivative at 4e point, i.e. the
but
which is continuous at every point of an mterval
a function

It is possible to construct

has a derivative nowhere.

s,

J.

Let

5.

(a)

3.

|q^

Is fix) diflferentiable at

by Problem
(6)

x^O

faj^sinl/a;,

fix)

13,

Chapter

exist)

= 0?

Then

2.

f(x)

Is fix) continuous at

(&)

has a derivative

^^(cosi)(-i) +

=
lim

a;

calculus diflferentiation rules,

From elementary

Since

= lim(-cosi +
=
conthiuous at

/'(,)

fix) clnnot be

if

sin

= 0?

at

(is diflferentiable)

x=

and

its

value

is 0.

x=0,

i)

-cosi + 2.sinl

(sini)(2.)

2a:

a;

does not exist (because

lim cos

l/a=

does not

in spite of the fact that /'(O) exists.

a;

/'(x) and puttmg


this case by simply calculating
This shows that we cannot calculate /'(O) in
derivative of a
the
when
only
It
is
^-0 as is frTquently supposed in elementary calculus.
answer. This happens to be
right
the
gives
procedure
this
that
action irco^rrufat appoint
calculus.
true for most functions arising in elementary

6.

Present an
fix)

"e,

8" definition of the derivative of fix) at

has a derivative

/'(x)

fixo

at x

= x

+ h)-

if,

fjxo)

given any

^,^^^J

>

<

0,

we

a;

= Xo.

can find

<

^jjen

1^1

<

>

such that

RIGHT and LEFT HAND DERIVATIVES


7

Let

fix)

1^1.

the left haid


id) Illustrate

at
= 0. (6) Calculate
Calculate the right hand derivative of fix)
^
at x = 0'!
derivative
a
have
(c) Does fix)
derivative of /(.) at x = 0.
(a)

the conclusions in

(0.

U^^>

(a)

since

since

\h\

|/i|

=h

= -^

for

for

(a), (b)

and

_
=
Hm ^W-^(^
- h+
h

^>0.

h<Q.

(c)

lim
1.-0+

from a graph.

M^
h

1 Ih =

"-*"+

CHAP.

DERIVATIVES

4]

The derivative at
does not exist if
the right and left hand derivatives are
unequal.

(c)

No.

(d)

The required graph is shown in the adjoining Pig. 4-3. Note that the slopes of the
lines y x and y x are 1 and 1 respectively, representing the right and left

= Q.

hand derivatives at x
derivative at

8.

65

Prove that

a;

f{x)

However, the

does not exist.

x^

Fig. 4-3

S g 1.

differentiable in

is

Let xa be any value such that

<

<

a;o

a;

Then

1.

h^O

At

the end point

At the end

Then
IS

9.

point

a;

= 0,

a;

= 1,

differentiable in

is

customary to write

/; (0)

Find an equation for the tangent


Prom Problem

(a)

8,

y-f(x,)
(6)

As

f'(xo)

in part (o),

2x.

f{l)

fL (1)

line to

f'(l) {x

-1)

or

= 2a;

/'(x)

for any x in this interval.

It

in this case.

at the point

=
2/

where

(a)

= 1/3,

Then the equation of the tangent

2/3.

j/-i

or

write

/'(I)

= a;^

2/

so that /'(1/3)

f'{x)(x-xo)

We may

1.

and

/'(O)

|(a;--i),

2(a;

= 1.

line is

i.e.

1),

i.e.

2/

2a;

- 1.

6Aa;

a;'

-f-

6a;

fa;

(&)

DIFFERENTIALS
10. If
(a)

(b)

1/

/(a;)

A^

dy

/(

a;3

- 6a;,

+ Aa;) -

find

/(a;)

principal part of

(a) At/,

dy,

(6)

(a)

and

Ay

(6),

Note that

e-0

- #.

+ Axf - 6(a; + Aa;)} - {a;= - 6a;}


x' + Sx^Ax + SxiAxr + {Axy - 6x (3a;2-6)Aa;
3a;(Aa;)^ + (Ax)'

Ay

{(a;

-I-

(3a;^-6)Aa;

and dy

(Sx^

i ., that
.w'i^^*
^.'^f
sized
dy and
dx are not necessarily small.

From

A?/

=
=
=

=Bx^~6

(c)

(c)

- dy =

3a;(Aa;)^

as Aa;-0,

i.e.

(Aa;)'

- 6)da;,

(3a;^

- 6)dx,
eAa;,

^^-0

i.e.

since

by

dy/dx

where

as Aa;-0.

definition Aa;

= da;.

3a;^-6. It must be empha-

3a;Aa;

Hence Ay

(Aa;)^

~ dy

is

an infinitesimal

of higher order than Aa; (see Problem


92).

In case Aa;

11.

is

small, dy

and Ay are approximately equal.

Evaluate \/25 approximately by use of differentials.


If Aa; is small.
Ay = fjx + Ax ) - f{x) = /'(a;)Aa; approximately.
Let

fix)

= ^. Then ^/W+A^ -

^^

^x'^'^Ax

(where

denotes approximately equal

to).

[CHAP. 4

DERIVATIVES

-,

DO

If

= 27

a;

and

Aa;

= -2, we

have

^/27^-V^
\/^ - 3-2/27

Then

or

It is interesting to

^(21)-'" (-2),

V^ - 3

i.e.

2.926.

good.
so that the approximation is fairly

= 25.05,

observe that (2.926)'

DIFFERENTIATION RULES. DIFFERENTIATION


12.

SPECIAL FUNCTIONS

of

/(-)^^(-) +

{/(.). W)
|:
dx

Prove the formula

- -2/27

^(^)

assuming

^^(^)'

and .

are differentiable.

By

definition,

sf,

,x

Mrr.

/(a;

+ Ax) g(x + Ao:) -

f(x

+ Ax)

.,

/(x)

^.

{g{x

+ Ax -

^,(,) +

g{x)}

r g(x

f{x)g(x)

g{x) {fjx

+ Aa;)-g(x) ]

+ Ax) -

f(x)}

,.^_(^J/(^:^^I:^iMl

^/(.)

,(.)

Another method:

= /(x), = g(x). Then Au = /(x + Ax) - /(x) and A,; = g(x + Ax) - g{x),
Thus
gix + Ax) = V + Av.
MAt> + ^AM + AMAi;
_
(u + Au)iv + Av)-uv
d
- "
lim
T-M-y =
Ax
iiX
A*-*"

Let u

u + Aw,

dx

13. If

it is

/(x

_
+ Ax) ,

A:t->0

/ Av

where

,,

i.e.

r,

Ax-

as

noted that At;-

= fiu) Where . = .(.),

Am

0,

since v

is

Am

*?i

*L

supposed differentiable and thus continuous.

%=%-%

prove that

4-

assuming that

and g are

differentiable.

Let X be given an increment

Ay

u and y take on increments Am and

as a consequence

Ax^O. Then

where

respectively,

A2/

/(m

+ Am) -

AM

f(u),

ff(x

+ Ax) -

g{x)

Note that as Ax->0, Aj/-0 and Am-0.


If

AM #0,

let

us write

f^

^y
"
If

we

AM =

define

for values of Ax, then

Ax

e-

^AU
du

as

Am-

and

+ eAM

()

dy

0,

Am^O

For such

shows that Ay=^0 for these values of Ax.

= 0.

It follows that in both cases,

limit as

(!)

so that

or Am

= 0,

(2)

Dividing

holds.

(2)

by

Ax#0

cases,

j.x.
and takmg the

we have

_
AoA^ ~
,.

"^

dbo

_
~

Ay

dydM.^dM
dudx

Au\

,:^/'d^AM

l^o\duAx ^ 'axJ
dx

dy

du

du
dx

dy
du

Am
j.
A.^oAx

Am
_j_

^^^

^^o

jj^^

a^o Ax
I

(3)

CHAP.

DERIVATIVES

4]

^*- ^^''^^

d^(^"^)

(a)

^(tan:.)

sec^a;,

(a)

fit.n.)

A/^li2^>

and ^(cos^)

^=08^

--(-.)-

(cos a;)(cos X)

-sin^, derive the formulas

^.

~:{sm~^x)

(b)

(sin

.)(- sin

It

sin-'., then .

sin

^
dx

or

"'^

(COS.)

_
"

cos".

^_
cos

VI -

J/

We

have supposed here that the principal value -^/2


-J^^i^ive, thus accounting for our writing cosy =

''

^^'"'"'^

?unctron':f'r"''
Consider y

f(u)

lim

=
Since the logarithm

^'^

By

log.w.

19,

Chapter

Thenby Problem

16.

Calculate dy/dx

if

Vl-.'

g sin-. S

log(w

^(.2/^)

where

iS.sr--(^)=

.;i.o..(i +

f)""

= m/Am.
;^(logaW) =
^

^.o..

with .

13,

dx^

xy'

(a)

]M1

'

dx

^ xy +

Zx^

y'

=(^2/) +(5)

^(3.")

= dy/dx.

Solving,

4^^")+i(^l-) =

2/

Let

(b)

h,

e^^

+ ylnx ^

cos

cosh,.,

l()
(sinhM)(2)

_
~

=
+

or

(.)(32/y)

(2/3)(i)_6^

Then

(^)(^,)

e-(.2,'

+ 2/)+^+(ln.)2,' = -2

find

(^)(j)^(,

2. sin 2.
."e^

+
+

.j/e^

sin 2..

. In .

#/dx, (6) dVda;^


- 3. + 1. Then dy/du = sinhu, du/dx =
2.-3,
dy du _
^^
d^'d^ - (sinhM)(2.-3) = (2. - 3) sinh (a;" - 3. + 1)

l),

where

dy
dx
^'^

2a;.

- 3/' + 2/)/(3.2/" - .).

(6.

(<=2.),
'

cosh(a;2-3a;
2/

2/'

Solving,

(a)

^ differentiable

+ AM)-logM

..)

17. If

Differentiate with respect to ., considering


y as a function of .. (We sometimes say that y
^mplu^^t funcUon of x, since we cannot
solve explicitly for y in terms of

(a)

^''

is chosen so that
rather than cosj, =

^/2,

VT^ilH^

a continuous function, this can be written

is

3,

^
sin^i/

definition,

+ AM)-f(M) _

/(w

x,

^">'' ^^^)' "^^ -

^-{i'.(-fy'")
by Problem

^^'^''^

Taking the derivative with respect to

2/.

cosj/^

sin.

_
"~

a;)

cos".
(b)

67

(a)

a."

(^nh.g) =
(coshu)(2.-3)"

and

sinhg+cosh.(f^y

2 sinh (."

- 3. +

1)

(2^

- 3)"

cosh

(."

- 3. + 1)

is

an

[CHAP.

DERIVATIVES

68

18. If

x'^y

y^

2,

find

{b)

(a) y',

(a)

y" at the point


x^y'

Differentiating with respect to x,


y'

x^

(6)

d
dx

y"

i-(\x^ +
dx

x-1,

Substituting

= -i

2/

=1

and

and
at (1,1)

32/='

-2xy

(y')
^^ '

(1, 1).

+ 2xy + Zy^y' =
-Ixy

(x^

+ W){2xy' + iy) -

{2xy)(2x

+ 6yy')

-1,
;

we And y" = -|.

MEAN VALUE THEOREMS


19.

Prove Rolle's theorem.


Case

1:

fix)

Case

Then

in [a,b].

f'(x)

for all x in

(a, 6).

2:

attains
Since f{x) is continuous there are points at which f{x)
Chapter
2).
Problem
(see
34,
respectively
and
minimum values, denoted by
Since f(x) # 0, at least one of the values M,m is not
and that /(|) =
zero. Suppose, for example,
fix)

in

its

[a, b].

(see Fig. 4-4).

For

, >0,

then

maximum and

+ h) S

this case, /(|

i^^-0

/(|).

and

If

h<0,

then

f(i

+ h)

and
Fig. 4-4

(2)

But by hypothesis

to the left

must be equal
which case

A
20. (a)

f'(i)

similar

hand derivative (1)


has a derivative at all points in (a, 6). Then the right
if they are both equal to zero,
only
happen
can
This
derivative
(2).
hand

fix)

as required.
in case

argument can be used

Prove the theorem of the mean,

theorem.
(a)

M=

Define

Fix)

Then Fia)
Also,

fix)

if fix)

/(a)

(a;

and

Give a geometric interpretation of this

(b)

- a).fib)-fia)

and Fib)

m # 0.

0.

specified in Rolle's
the conditions on continuity and differentiability

satisfies

theorem, then Fix) satisfies them also.

Then applying
F'io

(6)

no

Rolle's

theorem to the function Fix), we obtain

"
f^
^
'

fib)

0,

a<i<b

Let curve ACB in Pig. 4-5 represent the graph of


that there is a point |
fix). Geometrically it appears
between a and 6 where the tangent line to this
curve at point C is parallel to the chord AB.

Slope of tangent line


Slope of chord

Then

is

AB

such that

fH).

^ZT^
fib)-

/'(?)

fia)

It is interesting to note that the function Fix)


of part (a) represents the difference in ordinates of
at any point x in (a, b).
and line
curve

ACB

AB

fib)
f'ii)

-fia)
a

a<i<b

CHAP.

21.

DERIVATIVES

4]

Verify the theorem of the

\.
--^i^lr
4)/(5 - 2)

/o.
(25

22. If f'{x)

^''^^

"

or

^^'

'^'^^^

mean for f{x) =


= 4| - 7. Then the

Since 2

3.5.

<|<

5,

69

theorem of the mean

the theorem

at all points of the interval

a = 2, &

+ l0,

2a;2-7a;

states

that

4f

-7 =

is verified.

prove that

(a, h),

= 5.

f{x)

must be a constant

the interval.
Let

xi

<

any two

X, be

different points in (a,

f^^'^-fi^^)
X2

Thus

fix,)

at all

pomts of

23. If f'{x)

fix.)

>

<

X2 be

24. (a)

Prove that

(6)

Showthat
Let fix)

(a)

f'ii)

any two

different points in (a,

for X2

> Xi, and

^-^
J

= tan-x.

<

prove that f{x)

(a, b),

By

6).

^^'^'^

>

a,

1/(1

tan"' h

Since fix)

+ e) <

1/(1

+ a').
1

>

f'ii)

a;i

^^

- tan-^a <

1/(1

+ a;^)

and

tan~'a

/'(^)

Since

<

h,

1/(1

tan"' &

if

= 4/3

and a

I
25.

=1

<

1/(1

5.

+ e) >

in the result of part (a).

< tan-i- tan-1 < |

1/(1

have by the theorem of

Then

6^).

1
1

+^

- a.

Then

since tan"' 1

f+

or

+ 1^), we

tan~'a

b~a

+ 6^

and the required result follows on multiplying by


Let 6

Xi<i< x,,

< tan-i| < + 1.


J

ib)

strictly increasing.

so fix) is strictly increasing.

tan~'6

is

the theorem of the mean, for

the mean,

Since

x^<i< x^,

From this it follows that if two functions have the same derivative
the functions can only differ by a constant.

a;2

> fixi)

the theorem of the mean, for

constant.

(a, 6),

^<^')

fix^)

Xl

at all points of the interval

Let Xi

Then

By

h).

in

we have

,r/4,

< tan-| <

J+

Prove Cauchy's generalized theorem of the mean.


Consider
Gix) = fix) - fia) - a{gix)- gia)},
where a is a constant.
Then G(x) satisfies
the conditions of Rolle's theorem, provided
fix) and gix) satisfy the continuity and differentiability
conditions of Rolle's theorem and if Gia) = Gib) = 0. Both
latter conditions are satisfied if the constant

== ^^i^l/M
gib)

Applying

- gia)
Rolle's theorem,

f'(i)-ag'ii)
as required.

G'H)

for

or

<f<

6,

<^)

we have

SMn.

ff(b)-gia)

a<i<b

[CHAP.

DERIVATIVES

rjQ

TAYLOR'S THEOREM
26.

MEAN

of the

for the case n =


Prove Taylor's theorem of the mean with the Lagrange remainder

We

must show that

/(6)

m+
= m -

H(x)

in (1)

H(b)

(2),

To obtain H(a)

0.

= 0, we must

_
^4 "
Assuming
then H{x) also
that H'ii)

From
result

(6

- xYA

fib)

(2)

attained by replacing a by

is

H'(x)

= -f"(x){b-x) +

# 6).

a;

choose

f{a)

f'(a)(b-a)

($)

(b-aY
differentiability conditions of Rolle's theorem,
is a value i between a and b such

0.

(2)

2(6

- x)A

Substituting this value of

A -

for

/(6) yields the required

and solving for

into (3)

2(&

- |)A =
.

= -f"{i)(b~0 +

and H'(i)

{1).

>

follow by similar reasoning.

Prove that

sinx

where

| is

Let

fix)

Let

/"(x)

cos x,

51

= 5W180
(a),

x,

since

(x-a) +

/'(a)

the required result

and

radians

sin 45

sin 51

45

Hence the sum

28. (a)

Prove that

e^

(6)

Prove that

e is

| is

between

'^^

x,

so

that

/(a)

sin a,

2, i.e.,

= 45W180

'-

radians.

Then

a;

-a =

ff/30

radians.

\/2/2,
(cos|)(7r/30)

(V2/2)(7r/30)''

3!

2!

|-(co sg)(7r/30)^

namely

more terms

g^

JL/'-HlV

<-

31^^30^

<

0002

0.777, is accurate to 3 decimal places.

in Taylor's

formula should be taken.

+ xV2! + x^Bl +

= e\ Then all derivatives


= e" = 1 while /<"+"(?) = e^,
e'

where

-cos

an irrational number.

fix)

/c)(o)

obtained.

^+T~V30y

desired,

is

/' "(x)

x,

i-^,

V2/^\

V^

of the first three terms,

more accuracy

is

cos 45

The absolute value of the error term

Let

3!

= - cos J.

f"'iO

fia)

between a and

Thus from part

If

-sin

substituting into Taylor's formula with w


fix)

J is

/'(x)

= - sin a,

cos a, /"(a)

where

gl

x.

Then

sin X.

Then

(cosg)(x-a)3

a)(.-a)^

evaluate sin 51 and estimate the error made.

(a) to

(sin

+ (cosa)(x-a)

sina

between a and

Use part

f'{a)

(a)

- X) -

the conditions of Rolle's theorem; hence there

Generalizations to n

(6)

(6

and fix) satisfy the continuity and

f(x)

satisfies

(since

f"(i)/2\

{a)

fix)

is

From

(b)

f(x)

l.

<?<&

an undetermined constant. Motivation for forming H(x)


and transposing all terms to the right.

where

27. (a)

+^(b-ar

f'(a)(b-a)

Consider the function

and

=
x.

of fix) are equal to e\ If a = 0, then


so that Taylor's formula becomes
x^

x'

x"

x'''^'e^

+ X+2T + 3T+---+ n\'^{n + l)l

/(O)

/'(O)

=
/<\

^'

CHAP.

DERIVATIVES

4]

^^* ^"

"

Ux<0,

^^

(^Tfiyr"'-

all x,

i.e.

-<
e"

'''

^>0'

< ^^^^^ and

|i?|

Thus for

i.e.

the series converges for

<

<

%> g

< (^ri)T''

lim

|i?|

lim

|i?|

a;

^'^'^

l^-l

= 0,

If ^

also.

||

(^ee Prob. 30,

and 11m

Chap.

\R,\

3).

0.

and we may write

xV2\

xVZ\

...

(^^

all x,

^3^^t^'zi::r^'
Choose

l-^-l

,
1

rj-^

''^* -

"^^^^--^^ '^"-^^^^

'-

and multiply both sides of

(i)

by n!

'^^

^--^ -^- - -^^

'-^^*

Then

Now

e(/in + l) is a number between


and 1, while every other term in U) is a positive integer.
Thus assummg that e is rational, we arrive at the contradiction
that an integer is equal to a
non-mteger. Hence e must be irrational.

L'HOSPITAL'S RULE
29.

Prove L'Hospital's rule for the case of the "indeterminate


forms"
(a)

Weshallsupposethat/(a;)andsr(iK)aredifrerentiableina<a;<6 and

By

Cauchy's generalized theorem of the mean (Problem

m.

f(^)-f{xo)
9{x)-g{xo)

9{x)

^^^^
'*'o^0(x)

__

g'ii)

x"+flr'a)

0/0,

^0,

g(xo)

=o/oo.

(6)

= 0,

where

25),

^ ^
^o<i<x

f'ii)

"

f{xo)

(a)

.i+ff'w

-^

since as x -> xo+, i -> X(,+.

Modification of the above procedure can be used to


establish the result

(6)

Wesupposethat/(a;)andflr(a;)aredifferentiablein

a<a;<6, and

where a<a;o<6.

Assume

x^ is

such that

a<x,<x<x,<b. By
^^^^-^^^^

Hence

f(x)-f(xi)

~
f(x)

=,

9(^)

Let us now suppose that

^W

W'({)

lim

^,

f(x,)/f(x)

l-g(xi)/g{x)

x,~

Um

a;

g(x)

i_;+*^^^

->

x.

/^
g'd)

- ff(xi)/g(x)

f (t)

ff'ii)

l-f{xi)/f{x)

= L

oo,

->

'^"

from which we see that

/M

/W =

Cauchy's generalized theorem of the mean,

^'(f)

g(x)

ff(!)-ff{xi)

lim
*-o+

if

and write

"-Al-ZM/ZW; +

(i)

(^)

as

l-Kx^)lf{x) )

(^)

[CHAP.

DERIVATIVES

-72

We

Xi so close to Xo

can choose

that

\f'{i)/s'(i)

<

L\

Keeping

e.

xi fixed,

f(xi)/f{x)

-*+

'"-'''o

limit as a;-> a;o+ on both sides of

Then taking the

(2),

= L =

44

lim

we

see that, as required,

fM

lim

Appropriate modifications of the above procedure establish the result

Evaluate

e2x_i
^-,

lim

(a)

see that

lim

30.

we

(&)

l + cosTra;
^2
2a; + 1'

,,

->

a;o-,

-> ko,

In cos 3a;

,.

("^

if

In cos 2a;-

.!i?l

All of these have the "indeterminate form" 0/0.


(a)

lim

,,,

,.

l +
hm ,

(6)

cosira;

lim

.,

ttx
_
5-sin5
-

,.

TT

lim

,.

nm

g'

_
-

cosra;
s

;^

again yields the


ATote; Here L'Hospital's rule is applied twice, since the first application
"indeterminate form" 0/0 and the conditions for L'Hospital's rule are satisfied once more.

^"^

In cos

3a;

In cos

2a;

_
~

,._.

-3

sin 3a;)/(cos

(- 2 sin

3a;)

2a;)/(cos 2a;)

3 sin

3a;

cos

2a;

^o+ 2 sin

2a;

cos

3a;

jj^^

3 COS 2a; \ _
sin3a;\ / ,.
,.
.llT+^hT^/Uli'JV 2l^^3^; -

3cos3a; \
/ ,.
(^.i, 2 COS 2a;;

/3
\i

^ !

(I)

Note that in the fourth step we have taken advantage of a theorem on limits to simplify
further calculations.

31.

Evaluate

0^2 _ a; 4lim 5^2 6^33+

(a)

In tan 2a;

(^^Hm^^^-^

(^)

^T,

hTtiiTS^-

/.
All of these have or can be arranged to have the "indeterminate form"
,

3x

,.

''

- +5
+ 6a;-3 "

(''^

(6)

lim a;'e-

5a;^

a;

lim

_
"

lntan2x
(''^

tan 3x

.11?+ In

6a;

,.

T^

o.

Evaluate

lim

2a;

-^ =

6_

,.

.H?+
,.

2 sec'

2a;

10

lim

(3 sec^ 3a;)/(tan 3x)

= \}}^^

32.

_
"

(2 sec^ 2a;)/(tan 2a;)

,.

10^^+6

Vt

_
"

=
2 sec' 2x tan 3x

_
"

!i?JV

3 sec' 3x tan 2x

tan3x\ _

3 sec'SxjV-'?-^ t^^^2^J

^a;
a; tan
^21^(1 + ^)

/2Y

3 sec'

3a;

" VVV^2sec'2^

sin

1-

\xxa

Although L'Hospital's rule is applicable here, its successive use becomes involved. Using Taylor's
Page 62)
theorem of the mean, however, the limit is obtained quickly and easily. We use the results (see
3in^

^_|! +

^,

tan-a;

^+

Qx=.

In (1

+ x) =

f+

fix^

CHAP.

DERIVATIVES

4]

Then the required

limit equals
(X

li^

33.

Evaluate

lim

73

- xVZ + Px^) -(X- xVZ + Qx^)


x'-xV2 + Rx'

l + (P-Q)x^

x^lnx.

a:-0

_
~

iol-a;/2d-i?a;^

xHnx =

lim

Hm

i2_

_1/*

,i

lim

The given

limit has the "indeterminate form"


.
In the second step the form
mdeterminate form oo/oo and L'Hospital's rule is then applied.

to give the

34.

Find

a;

jjj^lncos^

Then

In F(a;)

the limit takes the "indeterminate form" !.

(In cos a;)/a;^

-sin a;

^^^^

which L'Hospital's rule can be applied.

to

lim

^_o2a;cosa;

2a!

- -i. But

lnF(a!)

oc,

;i^^(-sina;)/(cosa;)

'-

=*

Thus lim

and lim l/x^

F{x)^(cosxY"\

Let
have

'-*"

since the logarithm

is

x-^o

'''"^^

- 2a!

sin

= -i

F{x)

(e^-

find

bxf'-,

or

Let

G(x)

\nF(x)

= JLIL^^zM

in (a)

and

and

^hen

(b)

Evaluate

lim'
i-foVsm^a;

34,

lim

a00

- -1
2 cosa;

5xy" =

a;)i/^'

is

1:

applicable.

The required

<

applies.

hm

^7^^^

e""^

We

have

e^

By writing

the limit as lim

'^'"^'"''^
a!^sin^a:

it is

seen that

'

Two methods

of procedure

limit can be written

limr ^'-r'^Y-^")

_
_

and lim G(x) assume the indeterminate

However, this process proves laborious.

.-0 \
a!"

^'-

lim F(x).

are possible.

Method

InClimFfa!)!

are " and 1".

-*o

rule

x^

This has the indeterminate form

L Hospital's

{e^'

lim (cos

(6)

lim G(x)

forms /oo and 0/0 respectively, and L'Hospital's rule

Then, as in Problem

limF(a;)

lim F{x)

(a)

The respective indeterminate forms

since

'-*

ln(limF(a;))

36.

a;

a continuous function, lim lnF(a!)

Then

35. If

altered so as

lim (cos xY""'.

Since lim cos

We

is

a!

X*

\^

(i^oiiKi;)

"

^-

/Vsin^a:/

Now

lim

'^'-^'"''^
a;*

by successive applications of L'Hospital's

rule,

[CHAP.

DERIVATIVES

74

x"

s.va?

Iim

= ,.hm

=
Method

2x

2 sin

TT

x_ko

cos a

a;

lo

2 cos

2a;

4 sin

,.

lim

sm 2a;
r~?

4a;'

2a;

fT-j

,.

lim

24a;

x_fO

12a;^

2a;

lim

x^o

4a;''

x-to

8 cos
SI
24

2a;

Using Taylor's theorem, we have

2:

x^

sin"

sin"

a;"

,.

_
~

a;"

a;

(a;

"(x
x^/3

,.

x^o

a;"

_
~

1/3

xUo

- gV6 + Px'-y
xV6 + Pa;^

+ terms involving x" and higher


+ terms involving x^ and higher

+
+

terms involving x" and higher


terms involving x" and higher

1.

MISCELLANEOUS PROBLEMS
37. If

a;

(a)

= f{t) and

= sr(i)

(a) di//d,

are twice diflferentiable, find

Letting primes denote derivatives with respect to

d^ -

d^Tdt

t,

y
1

38.

Let

(a)

iffl'W^"

g'(t)

d/r(

If

= 1/m,

ifi

lim

Prove that

^^Q.

f(h)-m)
^^

^>

lim

r'(0)

Similarly, replacing
/: (0) = 0, and so

lim

h^0+
/'(O)

f'(h)-f'(Q)
- ^ '

Similarly,

lim u/e'

by

,'

/i' (0)

In general, /<*"(0)

Let

fix)

/'"(ico)

(6) /"(O)

0.

e-""'
r

hm

A,

^ 0-

byM-^-, we

and m-^o

find

/I (0)

_
- hm

2u*

/"^'(a;o)

e-i"''-27i-'

,.

hm

n=

for

<

0.

Thus

0.

and so /"(O)

lim l/2Me"
ti-foO

p
2e-"'''

,.

lim

,.

or

Taylor's theorem of the mean,

0.

1,2,3,

>

if J is

_
-

rule.

..

(see

Problem

89).

be such that /"^'(x) exists in a^x^b, and suppose that /'(xo)


where a<a;o<b. Prove that f{x) has a relative maximum or

at xo according as

By

f'(t)g"(t)

0,

,.

by successive applications of L'Hospital's

39.

(a) /'(O)

W-+00

[^'(*)]'

using L'Hospital's rule this limit equals


lim we""^

(6)

g'it)

a'(t)f"(t)

e-i"''-0

,.

'

U-^OO

/;(0)

_
"

dtVg'(t).

dxidt

WW

1^"'^''

dtW(t)J _

g'fflrW -f(t)g"W l

?wl

m=
f'{0)

dff'(t)\
dx\g'(t)J

d^y/dx^

we have

d ff'(t)\

(6)

respectively.

between xo and

x,

then

= /"(a;o) =
minimum

CHAP.

DERIVATIVES

4]

fix)

f{xo)

- xo) +

f'(xo) {X

/"(")

75

(^-^o)'

+ f"'M{x-x,Y ^ r^'HOix-xoY
3!

2!

=
If

0, then for all a; in a S neighborhood of xo we have /(x) g


/(a;), so that fix) has a
value atx^xo. Similarly if /"v.(^) < q, then for all a;
in a S neighborhood of xo
so that /(a;) has a relative maximum value at a; = xo.
/(;),

minimum

we have

40.

^""(^)(f--)^
4!

>

/"^'(a;o)

relative

fix.)

4;

f{x)

Find the length of the longest ladder which can be


carried around the corner of a
whose dimensions are indicated in the figure below, if it is
assumed that

corridor,

the ladder

carried parallel to the floor.

is

The length of the longest ladder is the same as the shortest


straight line segment
[Fig. 4-6] which touches both outer
walls and the corner formed by the inner walls.

AB

As

seen from Fig. 4-6, the length of the ladder

L
i

is

a sec

i.e.,

a sin'

a sec

Then

tan 9

b cos' e

Va^'^

so that

a sec

Although

is

esc e

for

=
fbh,

J 1/3

b cse

geometrically evident that this

is

it

we can prove

length,

tan

b"

6 esc e cot 9

or

,1/3

d'^LIde'^

AB

6 esc e

minimum when

dL/de

minimum

4-

= tan-'v/^

gives

the

by showing that
Problem 88).

this analytically

is positive (see

Fig. 4-G

Supplementary Problems
DERIVATIVES
41.

Use the

definition to

((.)(3a;-4)/(2a;

Ans.
42

T Bt

(a)

f(rv\

'^^'

17/25,

at
41

a;

Tot

-Liet

= 0,

(c)

/c-v^
/^a;;

45.

46.

(h) 2,

(c)

1,

*^ si" I/*''

a;

7^

0,

a;

f{x)

is

{xe~^"''^,

^''''

a;#0
^

Ans.

^^

(a)

a;

*^*

(") /(*) is

= e', show

Use the

results

that

/'(xo)

= e'o

lim (sin /i)^

a;

0,

(6) f{x)

_,

(6)

/(x)

(a)

3,

Page

depends on the result lim


1,

has a derivative

is

continuous at x

= 0,

(6)

has a deriva-

Yes,

Give an alternative proof of the theorem in Problem


If /(x)

continuous at

= 0.

Determine whether

Yes;

point:

(d)

continuous at

| ^^

= 0.

tive at X

44.

+ 3),

compute the derivatives of each of the following functions at the


indicated
a; = l;
(6) x'- 3x^ + 2x - 5, x = 2;
{c)^[i,x^^^ (d) ^^6^^ a; = 2.

lim (l-cos/i)M

63,

(e"

using

- \)lh -

"e, 8

definitions".

1.

to prove that if /(x)

sinx,

/'(x.)

cosxo.

RIGHT and LEFT HAND DERIVATIVES


Let /() = x\x\. (a) Calculate the right hand
47.
derivative of /(x) at

and

(a), (6)

48-

Discuss the

Let

g
50.

51.

= 0.

Does

(c)

(a)

and

continuity

(a) 0;

(6) 0;

(6) differentiability

What happens

in case

= 0.

derivative of f{x) at a;
= 0?

(c)

Yes,

of

/(*)

Calculate the left hand


the conclusions in

(6)

(d) Illustrate

have a derivative at x

f(x)

Ans.

- l^f^^' 2~*5^.
Ix^ Z, 2<a;S4

fix)
a;

a;

from a graph.

(c)

positive number.

49

[CHAP.

DERIVATIVES

76

" sin llx, /(O)

where p

0,

is

any

any real number?

is

Discuss the

and

continuity

(a)

differentiability of f{x)

(6)

in

4.

Prove that the derivative of

f(x) at

a;

and only

exists if

aJo

if

/+ (xo)

fL (x).

Prove that f{x) = x" ^ x" + 5x - 6 is differentiable in a S x ^ ft, where o and 6 are any constants.
Find equations for the tangent lines to the curve j/ = x^ - x^ + 5x 6 at x = and x = 1. Illus(c) Determine the point of intersection of the tangent lines in (6).
trate by means of a graph,
(d) Find /'(x), f"{x), f"'{x), r"'(x), ....
Ans. (6) 2/ = 5x - 6, 3/ = 6x - 7; (c) (1, -1); (d) Zx" - 2x + 5, 6x - 2, 6, 0, 0, 0, ...
(a)
(6)

52.

Explain clearly the difference between

53.

If

/(x)

x^'lxl,

(a) /; (xo)

and

/'(xo+),

(6) /I

(jko)

and

discuss the existence of successive derivatives of f{x) at x

/'(xo-).

= 0.

DIFFERENTIALS

54.

If

55.

If fix)

2/

Ans.
56.

57.

/(x)

+ Sx,

x'^

(a) .0501,

Using

\/36.

If

3/

(c) 5.01,

(6) .05,

Ans.

(a) A^/,

%,

(6)

find ia) Ay, (b) dy,

differentials,

(c)

And

1/x,

Ay/Ax,

(c)

(d) 5,

(c) Aj/

- dy,

{d) (Aj/

dy/dx and

id)

- dy)/Ax,

(e)

(Ay

- dy)/Ax,

(6) 0.12,

sin X, evaluate

(a)

Ay,

(6) dj/.

if

a;

=1

and Ax

.01.

(e) .01

compute approximate values for each of the following:


(a) 0.515,

dy/dx.

(e)

(c)

2.0125

(c)

Prove that (Ay

- d2/)/Ax

->

as

sin 31,

(a)

Ax

(6)

In (1.12),

0.

DIFFERENTIATION RULES and SPECIAL FUNCTIONS


58.

Prove:

^"^

59.

Evaluate

'

dx

tanhii

Compute

f^n^) + fx^ix),

(6)

^{/(x)

p(x)}

J^fi^^)

'

^s(x),

'^"'^'^"'

[g(x)r

j^ {x"

3 In 4, (6)

g(x)}

In (x^

- 2x + 5)}

at x

= 1,

(6)

~ {sin^ (3x + n/6)}

at x

= 0.

|V^

Derive the formulas:

(c)

61.

(a)

(a)

dx\gix)j

Ans.

60.

f^{fix)

ia)

(a)

a"

sech^M-^ where m
dx

(a)-^tan-'x,
dx

use of principal values.

= a"lna ^, a>0, a#l;


is

a differentiable function of

(b)^csc-'x,
dx

(c)

-^
dx

sinh"' x,

(6)

^cscm =

^;

x.

(d)^coth-'x,
dx

-esc m. cot m

paying attention to the

CHAP.

62.

63.

''

65.

Uy-x'.comvniedy/dx.
If

2/

67.

Calculate

If

If

2/

69.

dy/dx

(a)

and

(a) 2^7(1 -a;2^),

tan

sec

a;

Aws.
68

+ 2)}''-'<^'+=>,

{ln(3..

Prove

(a)

2/

(6)

-1,

dx"

Establish formulas

y'"

tan
(c)

_
"

t^'i/

^^

tJiQf

and

V2,

= '^^^' ^^^

differentiating].

Ans.

^^^^

Ans. x^X

V
(j^ + ^
a-g- -Ling
'" '" ^

\nx).

2)^/

/.

. and . are

a;2/^l

+ 2/2)(l + ^y^),

2(1

evaluate

t,

:= 0.

d'y/dx^ if xy -\ny = 1.
- 2a;2/V(l - a;?/)^ provided

(6)

prove that

a;,

dy/dx at *

find

^"' ^

(Sj/'

(j)

77

Take logarithms before

[Hint:

= ^^^^
LLlSle!''^^^ "

Atis.

66.

DERIVATIVES

4]

(a)

d2//dx,

d'y/dx\

(6)

(c)

d2//da;',

at

= :r/4

3V2

d^a;

(a) 7,

dx\

/(

\d^J

~dyy

and

18

(6)

stating precise conditions under which

'

on Page

27,

(c)

it

holds.

60.

MEAN VALUE THEOREMS


^"'

71.

^^'

~^- 1"^

aLn.V
fapplicable ?
to TJ-^'
this ;
function,
"^

i'nof
IS
not

Verify Rolle's theorem for f(x)

m?

i.e.

x'{l

no value

is

- x)\

^'^^ ""^^^ ''*' *

S g
a;

"''*''"'

^*r,?,*''^
[Hint: ^A^*
Apply
Rolle's theorem to the function
73.

0<a<6,

(a)

If

(b)

Use the

prove that

result of (a) to

74.

Prove that

75.

Prove the statement

76.

77.

78.

(:./6

If f(x)

(6)

(a)

(6)

Prove that

at all points of

(sin

(a)

By

placing a

Prove that

81.

and

of the

In (1

<

(a, 6),

=
a;

= ^ *h^^
- sin

In 1.2

(./6

'

-1.

< (b/a-l)
< 15'
by using the theorem of the mean.

1/8)

prove that fix)

is

20(6).

monotonic decreasing in

(a, 6).

strictly decreasing in (a, 6)?

cot|,

in (a),

where

show that

W2).
is

(6)

Prove that

between a and

= a;/2.

sin x

2a;/;r

f or

a;

S ^/2.

6.

Does the result hold

if

a;

<

0?

MEAN

a.)
'

cc

^2

and estimate the accuracy.

Evaluate (a) cos 64, (6) tan- 0.2,


Ans. (a) 0.438, (6) 0.197, (c) 1.543,

^3 ~ ^
+
4

Ans.

(6)

(c)

cosh

(d)

0.741

1,

="'

5(1

+ j)5

'

0<,<.
" < ? < *

/m
(o) tt
Use

(o)^ .
to
.

evaluate

0.09531 with an error less than 2-10-.

(d)

e~'-

to 3 decimalplaces.

Prove Taylor's theorem of the mean with the Lagrange


remainder for (a)t
positive integer.

^t l^t ne real root of e^cosa:

a;.]

In 6/a

*o <

strictly decreasing in (0,

^^^^^I^^^^

In (1.1)

80.

a;)/a; is

.6

e''

1.

paragraph of Problem

is f{x)

Prove that

TAYLOR'S THEOREM
79.

in the last

Under what conditions

- a/6) <

< sin-

V3/15)

(a)

(a)

(1

show that

the graph of /(=.). (5) Explain why Rolle's theorem


j for which /'(c) = 0,
< { < 2.

C'^^t'^'=t

there

= 2,
>

(6)/
\

=3

(c) w

vj

anv
any

82.

where

m-

f(x)

chosen so that H(a)

is

- ^^^_

f'{x)(b-x)

-,

^)^

(6

= 0.]

theorem can be written in


Prove that the Lagrange and Cauchy forms of the remainder in Taylor's
the forms
h^+'-r^-^ia+eh)

where

respectively,

84.

Derive the result (21), Page 61, for the Cauchy form of the remainder.
[Hint: Apply RoUe's theorem to

H{x)

83.

[CHAP.

DERIVATIVES

rjg

By

writing

/i

<<

and

fc"+'(l

-)"/'"'"'(

+ gfc)

1.

obtain Taylor's
in place of the last term (6 - x)A in the hint of Problem 82,
special values
the Lagrange remainder, (6) the Cauchy remainder, using suitable

- xYA

(6

theorem with

(a)

for p.

RULE

L'HOSPITAL'S
85.

Evaluate each of the following

s'"

,s

(6)

lim

__../
cos 3a; - 2 cos 2x

'^

lim

(d)

lim x^e-^'

Ams.

(o;^

(a) ^,

[n)

9 4-

_Z + cos
-,

lim

-1,

(6)

e-"\

(o)

e\

(c)

a;

-4/,r,

(3^

(/)
"
-o

- 1) tan ^x/2

(c)

lim x^ In x

(e)

2*

limits.

lim {llx

(i)

2')/x

(j)

Im

/a +
/

(g)

lim

(1

- S/x^

(k)

lim (1/x^

lim

(1

+ 2.)-

(I)

Urn'-^^^lE^Z^LlR

(d) 0,

(e)

0,

<

(n)

x--"

==-^o

(/)

In 3/2,

(g)

e-\

cot^ x)

(h) 1,

(t)

0,

(j)

,.

lim

^^

(m) lim x ln(

esc x)

/^sina;\''^

(^-^
x
-o\

+ e'+

(o)

lim (x

(,)

Jhn^(six)-^
(fc)

1,

(0 i,

|,

e^'Y'^

(m)

6,

[p) e

MISCELLANEOUS PROBLEMS
86.

Prove that

87.

If

A/(x)

(6)

89.

90.

91.

f{x

+ ^x)-f{x),

(a)

if

if this

0<x<l.
A{A/(x)}

prove that

derive an expression for A''f(x) where n

/'"* (x)

88.

^/^ < ^|^

is

any positive

^.'f{x)

integer,

+ Ax) +

^^

f(x),

A" f(x)

"

(^x)"

40.

Find the relative maxima and minima of


Ans. f(x) has a relative minimum when x

f(x)

x",

x>0.

= e~^.

particle travels with constant velocities Vi and Vi in mediums


II respectively (see adjoining Fig. 4-7). Show that in order
to go from point P to point Q in the least time, it must follow

and

PAQ

2 f(x

show that

(c)

for
MO) =
If fix) is the function of Problem 38, prove that
Taylor's series with a remainder for this function and prove that f{x)
cannot approach zero as w^ and discuss the consequences of this.
(a)

path

2Ax)

limit exists.

Complete the analytic proof mentioned at the end of Problem

f{x

where

is

such that

(sin 9i)/(sin e^)

vihi

ti

1,2,3, ...
(6) Write
(c) Explam why K
Rn.

P ""v

N^,

Medium

velocity i

CHAP.

92.

93.

94.

DERIVATIVES

4]

79

A variable a is called an infinitesimal if it has zero as a limit. Given two infinitesimals a and we
^8,
say that a is an mfinitesimal of higher order (or the same order)
if lim a/ji =
(or lim alp = l= 0)
Prove that as a;-*0, {a) sin^2a; and (1 - cos 3a;) are infinitesimals
of the same order, (6) (x^ - sin^a;)
IS an mfinitesimal of higher order than
{a; - In (1 + a;) - 1 + cos x).
Why

can we not use L'Hospital's rule to prove that

Can we use L'Hospital's

lim^^-^^J^ =

(see Prob. 91,

rule to evaluate the limit of the sequence

Un

m'e-"',

Explain.
95.

[Hint:

Assume

96.

the actual root

hf'{a)

isa+h,

If

a;

4a;

that a better approximation, in general,

f(a+h) =

so that

to 3 decimal places.

95, obtain the positive root of

Ans.

denotes the operator d/dx so that

Dy =

(a) 3.268,

(6)

98.

= {D-u)v + Ci(D''-'u)(Dv) + X,(D-^u)(D'v) +


Xr = (") are the binomial coefiicients (see Problem 95,

99.

If

f'(x.)

borhood of
100.

-^ (x^ sin x) =

Prove that

{x^

n{n

- 1)}

= ... = r-^xo) =
- x. The point xo in such

f"(x,)
a;

Let fix) be twice differentiable in


at least one point

>

'

given by

fact that for small h,

i in (a, b)

but
case

{a, b)

such that

sin (x

+ w^/2) -

a;'

- 2a;^ - 2a; - 7 =

0,

...

d'-y/dx".

prove Leibnitz's formula

Xr(D''-^u)(D'v)

Chapter

...

uD^v

1).

2wa; cos {x

+ nW2).

^ 0, discuss the behavior of f(x) in the neighoften called a point of inflection.

/<^"+"(a;)
is

and suppose that

|/"(c)|

(a)

1.131

dy/dx while D'-y

D-(uv)

where

Then use the

0.

is

'

approximately.]

Using successive applications of Problem


(6) 5 sin

97.

f{a)

= 0, show

2) ?

2 3

1
'

If a is an approximate root of /(x)


-/()//() {Newton's method).

f(a+h)

w=

Chap.

involving velocity and acceleration of a particle.

^^A_

/'(a)

{/(ft)

f'(b)

_ /(^)j.

0.

(jj^^

Prove that there exists


^ pj^y^j^^ interpretation

chapter 5
Integrals

DEFINITION

of a

DEFINITE INTEGRAL

the area
The concept of a definite integral is often motivated by consideration of
a; = b
x^a
and
at
erected
ordinates
=
the
bounded by the curve y f{x), the x axis and
(see Fig. 5-1).

However, the

i,

xi

definition

fj

can be given without appealing to geometry.

Xi

Fig. 5-1

a^x^h

into n sub-intervals by means of the points


Subdivide the interval
.,
In each of the new intervals {a,Xi),{xi,X2),
chosen arbitrarily.
Xi,X2, ...,a;-i
Form the sum
.,! arbitrarily.
{x-i,b) choose points i^,i^,
.

m;)(xi-a) + f{Q{x2-Xi) + ms){x2-X2) +

By

writing xo^a, Xn

= b and Xk-Xk-i =

AXk,

|;/(4)(a;fc-a;.-i)

this

/(l)(b-a;n-i)

(1)

can be written

^/djAx.

(2)

sum represents the total area of all rectangles in the above figure.
We now let the number of subdivisions n increase in such a way that each Ax^-^O.
not depend on the mode
as a result the sum (1) or {2) approaches a limit which does
subdivision, we denote this limit by

Geometrically this

If

of

f{x)

dx

(3)

In this symbol
called the definite integral of f{x) between a and b.
integration.
often called the integrand, and [a, b] is called the range of
and b the limits of integration, a being the lower limit of integration

which
call

is

We

upper

limit.

80

f{x) dx is
sometimes
and b the

^^^^-

INTEGRALS

^]

gj^

The limit (3) exists whenever f{x) is continuous (or sectionally


continuous) in a^x^b
Problem 35). When this limit exists we say that
is
Riemann integrable
f(z)
or
y
sim-ply mtegrable in [a, 6].
(see

<^

Geometrically the value of this definite integral


represents the area bounded by the
curve y - t(x), the x axis and the ordinates at x =
a and x = b only if f{x) SO
If fix)
IS sometimes positive and sometimes
negative, the definite integral represents the algebraic sum of the areas above and below the
x axis, treating areas above the x axis as
positive and areas below the x axis as negative.

MEASURE ZERO
A set of points

on the x axis

of intervals enclosing
positive number .).

all

We

is

said to have

can show

measure zero

if the sum of the lengths


arbitrary small (less than any given
(see Problem 6) that any countable set
of points on
In particular, the set of rational numbers
which is

the points can be

made

the rea axis has measure zero.


countable (see Problems 17 and 59, Chapter

1), has measure zero.


important theorem in the theory of Riemann integration
is the following:
Theorem. If f{x) is bounded in [a,b], then a necessary
and sufficient

An

condition for

the existence of

PROPERTIES
If f{x)
1.

f{x)dx

fa

DEFINITE INTEGRALS

of

and

that the set of discontinuities of f{x) have


measure zero

is

g{x) are integrable in

{f{x)g{x)}dx

Af{x)dx

A C

then

[a, b]

C g{x)dx

f{x)dx

f{x)dx

where

is

any constant

3.

f{x)dx

provided f{x)

is

f{x)dx

dx

f{x)

j^ f{x)dx +

J'f{x)dx

integrable in [a,c] and [c,b].

- r

f{x)dx

6.

If in

SXg

6,

m g f(x) S M
m{b-a)

7.

If in a

a;

6, f{x)

MEAN VALUE THEOREMS


^'

sue? that "

J*

f{x)dx

and

are constants,

then

M{b-a)

g{x) then

J
f{x)dx

where

g{x)dx

INTEGRALS

"^^'"^ *^'"'*'"; ^^

iia<b

\f{x)\dx

for

f{x)dx

^^""^ '"

^(^)^^

continuous in

(&-)/(!)

[a, b],

there

is

a point

^ in (a, b)

(4)

[CHAP.

INTEGRALS

g2

in [a, h], and


Generalized first mean value theorem. If f{x) and g{x) are continuous
h) such that
(a,
in
point
|
g{x) does not change sign in the interval, then there is a

r
This reduces to

3.

(^)

f{x)g{x)dx

if sr(x)

(5)

Sf(x)dx

/(^)

= 1.

continuous in [a,b] and


Bonnet's second mean value theorem. If fix) and g{x) are
then there is a point | in (a, b)
if ?(x) is a positive monotonic decreasing function,
/-b
such that
^/ X 7
//?^
(6)
= g{a) jr^ f{x)dx
f{x)g{x)dx
,

monotonic increasing function, then there

If g{x) is a positive

in

such that

[a, b)

f{x)g{x)dx

a point

is

sr(b)

(?)

/(a;) da;

and sr(x) are continuous in [a,b]


and is not necessarily
decreasing
and if g{x) is monotonic increasing or monotonic
that
such
b)
always positive as in 3, there is a point | in (a,

4 Generalized second

C"

mean value theorem.

fix) g{x)

This result holds also

dx

we

if

gia)

If f{x)

fix)dx

gib)

by

replace continuity

()

fix)dx

integrability.

INDEFINITE INTEGRALS
If fix) is given,

Clearly

integral or anti-derivative of /(x).


also is

Fix)

= /(x) is called an indefinite


an indefinite integral of fix) so

then any function F(x) such that F'(x)

+ c where

definite integrals differ

c is

Fix)

if

[F(x)

any constant since

+ c]' =

F'(x)

Thus

/(x).

all

in-

often use the symbol //(x) dx to denote any

We

by a constant.

is

indefinite integral of fix).

Example:

If F'(x)

tive of

= x^

F{x)

then

dx

i x'

= xVS +

is

an indefinite integral or anti-deriva-

x'.

FUNDAMENTAL THEOREM

of

INTEGRAL CALCULUS

is any function such that F'(x)


If fix) is continuous in [a, b]
of fix)], then
antiderivative
is an indefinite integral or

and F(x)

fix)dx

Fib)

fix)

[i.e.

F(x)

(9)

Fia)

integrals without direct use of


This important theorem enables us to calculate definite

the definitio

-.

whenever an

Example:

To

calculate

indefinite integral is

f xUlx,
C x'dx

we note

Since c disappears anyway,

F(2)
it is

that

known.
F'{x)

F(l)

= x\

F(x)

= xys + c,

f-)-a"-

convenient to write more simply

and we have

= i

<^HAP.

INTEGRALS

5]

83

DEFINITE INTEGRALS with VARIABLE LIMITS of INTEGRATION


An indefinite integral can be expressed as a definite integral with
limit

by writing

^
J

It follows that

^ f{x)dx

f{x)dx

f ^(^)^^

d^J

(^10)

/(^)

{11)

Since a definite integral depends only on the limits of


integration,
variable as the symbol of integration. For example, f
dx =

For

etc.

this reason the variable is often called


for instance, as
.
x

The

^^'^'^'

d^X

result can be generalized to the case

limits are variable.

f iiElj^j
dx Ji

variable.

f{t)

dt

We

use any

f' f{u) du,

can write

^^^^

(11)
(^^)

where the lower and upper integration

sin x" d(x^)

x^

of

f{x)

dummy

we can

Thus we have

Example:

CHANGE

variable upper

VARIABLE

of

If a determination of

_ ^n^d(x)

dx

2 sin x'

dx

sin x

INTEGRATION
fix)

dx

not immediately

obvious in terms of elementary


J
may be obtained by changing the variable form x to t according
= g{t). The fundamental theorem enabling us to do this is sumis

functions, useful results


to the transformation x
marized in the statement

X f{x)dx
where after obtaining the
terms of

x, i.e.

//{fir(i)}sr'(t)cJi

indefinite integral

{u)

on the right we replace t by its value in


This result is analogous to the chain

g-\x), assumed single-valued.

rule for differentiation (see

Page 59).
The corresponding theorem for definite

y
where
fix) IS

f{x)dx

integrals is

J^

f{g{t)}g'{t)dt

{15)

= a and g{/3)^b, i.e. a^g~\a), I3 = g-i{b). This result is certainly valid if


continuous in [a, b] and if g{t) is continuous and has a continuous derivative
in
g{a)

INTEGRALS

of

SPECIAL FUNCTIONS

The following
an

identity.

be added.

^"^^
^-

3.

results can be demonstrated by differentiating both sides to


produce
In each case an arbitrary constant c (which has been omitted
here) should

J IT ~
^
smudu
J

^^pi

^^-1

^"1^1

4.

^-

-coszi

6.

J
j

cosMdM

sinM

tanwdw

=
=
=

fcotttdM

In |secM|
In |cosm|

InisinM|

[CHAP.

INTEGRALS

84

X'
("

7.

=
=

sec udu

9.

10.

tanw

csc^Mdw

-cotu

jsecii'udu

taiihu

C cscti^udu
,.

-cothM

7r/4)|

22.

-'

cot^l

sechutanh udu

-sechu

f cschucoth udu

-csch^t

23.

^4.

Jsectant.d

11.

ln|tanM/2|

jsec^udu

In |csc%

21.

tanMl

In tan {u/2

CSC udu

ln|sec%

^^

25.

sin-i- or -cos-^-

/^

^6.

Inlii

^/vFa^
r
du ^
C
^
\
=
^-j
2
I

2_o,2

sec

+ V^?^^

12.

JI

cscucotudu

-cscM
on
27.

"
u
1
1 Xtnn-i

Trtan-i

+ a^

14.

e"
I

15

-'
^

C^tanh M dw
^

17

30.

coshudu

du
^=

29.

13

16

u^

d.u
dw
An
,

sinhM

-In

_^u

a ... \1 ..-i
\
1
_,,0,
77^08-'- or -sec ^a
a
M
a

= %y/u'd
^

^2

,
r-j
ri +
f yoj^-u^du
= u2^Va2_^2

Inlsinhtt]
'

'

32.

\/i?T-^

g-ln |w

Jcothwdtt

a.

Jfv^^^^^dM
"

31.

^u

,
In cosh

i
^

Ma
M+a

1^0^^^^

-^

**
1
^
+-1,u

cot
--cot

or

*^

,-!1
_,sin

19

sechwdM

tan-i(sinh%)

*^

20

33.

hu du

gau cos

5^

e^/^cosbt^
\

of

(^^

bu

sin

e"" (a

b cos bu)

WV^

=^

+ bsinbM)

WV^

INTEGRATION

Integration by parts.

^ udv

where u =

uv-^vdu

f{x)

and v

Partial fractions.

is

= g{x). The

Any

2:

corresponding result for definite integrals over


and g{x) are continuous and have con-

P(x)

less

where P{x) and Q{x) are polynomials,

than that of Q{x),

form
in

^^^.^

can be written as the

Ax + B
:^^2 + {,x + cY

^'^^^^ r

_
-

sum

o Q
1,^,6,

of

terms of elementary functions.

A
-3

3a;-2
_
~
(4* - 3)(2x + 5)'

+2
5a:' + 2a; + 4)Ma; - 1)

"^
{2a;

4a;

_
""

a;

(a;''

^^

rational function

which can always be integrated

Example

j f'{x)g{x)dx

See Problems 18-20.

rational functions having the

,
1:

f{x)g{x)

certainly valid if /(x)

with the degree of P{x)

,
Example

^ i{x)g'{x)dx

or

the interval
tinuous derivatives in [a,b].
[a, b]

2.

34

SPECIAL METHODS
1.

e"" sin

-coth-i(coshu)

fcsch^id^*

+g
+ 2a; + 4)'

Aa;
(a;'

g
+ 5^

C
I

(2x

+ Z)
+ 2a; + 4

Ca;
a;'

+ 5F

D
2a;

+5

g
x-1

The constants, A, B, C, etc., can be found by clearing of fractions and equating coefficients
methods (see Problem 21).
of like powers of x on both sides of the equation or by using special

CHAP.

INTEGRALS

5]

85

3.

Rational functions of sin j: and cosx can always be


integrated in terms of elementary functions by the substitution tana;/2 = M (see
Problem

4.

Special devices depending on the particular form of the


integrand are often employed (see Problems 23 and 24).

22).

IMPROPER INTEGRALS
If the range of integration [a, b] is not finite
or if f{x) is not defined or not bounded
at one or more points of [a,b], then the integral of
f{x) over this range is called an improper integral. By use of appropriate limiting operations,
we may define the integrals

such cases.
Example
Example

f^

1:

f^=

2:

Examples:

f'
Jo ^

Um

f j^

lim

f'^

li,

->0+

f'
J^ X

limtan-x

=
=

lim

Hm+

2Vi
In.

=
=

-..0

Since this limit does not exist

we

lim tan~'Af

7r/2

lim (2-2V^)
lim ( In

e).

say that the integral diverges

(i.e.

converge).

For further examples,


tegrals, see

Chapter

see Problems 33, 78-80.

does not

For further discussion of improper

in-

12.

NUMERICAL METHODS

for

EVALUATING DEFINITE INTEGRALS

Numerical methods for evaluating definite integrals are


available in case the integrals cannot be evaluated exactly.
The following special numerical methods are based
on subdividmg the interval [a,b] into n equal parts
of length Ax = <b~a)/n.
For simplicity we denote fia + kAx) =
f{xu) by y^,
where k = 0,1,2. .. .,n. The symbol ^
means approximately equal". In general, the approximation
improves as n increases.
1. Rectangular rule.

yf{x)dx

Aa;{2/o

The geometric
2.

2/i

+ 2/2+...+i/-i}

interpretation

is

or

Ax

{y,

+ y, + y,+

..

+y,)

^u)

evident from the figure on Page 80.

Trapezoidal rule.

f{x)dx

^{2/0 + 22/1 + 22/2+... +22/-i+ ^4

(i7)

This is obtained by taking the mean of the


approximations in (16)
Geometrically this replaces the curve y^f{x) by a set
of approximating line segments.
3.

Simpson's rule.

yf{x)dx
This
(I.e.

is

^^''

+ ^^ + 21/2 + 42/3 + 22/4 + 42/5 +...+22/n-2 + 42/-i + 2/}

{18)

obtained by dividing [a,b] into an even number of


equal intervals
even) and approximating f{x) by a quadratic through
3 successive points
is

corresponding to
the curve y
4.

x,, xi, x^; x,, x^, Xs;

= f{x) by

...; a;-2, x-i, x.


Geometrically this replaces
a set of approximating parabolic arcs.

Taylor's theorem of the

mean can sometimes be

used, as in

Problem

26.

[CHAP. 6

INTEGRALS

gg

APPLICATIONS

solve many physical or


of the integral as a limit of a sum enables us to
lengths, moments of
arc
volumes,
areas,
geometrical problems such as determination of
inertia, centroids, etc.

The use

Solved Problems
DEFINITION
1.

DEFINITE INTEGRAL

of a

prove that

If f{x) is continuous in [a,b]

of the mode of subdivision (see Problem 35).


Since f(x) is continuous, the limit exists independent
length ^x = (h - a)ln (see Fig. 5-1, Page 80).
equal
of
parts
equal
into
n
[a.,
of
subdivision
6]
the
Choose
Let i^ = a + fc(b-a)/, fe = l,2,...,ri. Then

2.

Express
Let a

limi
= 0,

^
^ Jk/(

in

lim^i/(a + ^^)

limi/(|.)A..

(a)

Problem

f x^dx

Express

definite integral.
(a)

If

fix)

= x\

(6)

1.

- i

/f-")

then f(kln)

(fe/)''

fcVw^

Thus by Problem

i^

n(n +l)(2w

+ l)

6^?

_
~

2,

^da;

1,

(1

,.

+ l/w)(2 +

1/n)

/*

the required limit.

virhieh is

Note:

By using

1V3

0V3

the fundamental theorem of the calculus,

limj^

The required

we

observe that

1/3.

The area bounded by the curve y

Evaluate

dx

/(*)

Interpret the result geometrically.

4.

given
as a limit of a sum, and use the result to evaluate the

This can be w^ritten, using Problem 29 of Chapter

(6)

^^

Then

lim

(a;V3)|J

/(-)

as a definite integral.

lim

3.

^j^pg

limit can be

= x\

"
"

"

the x axis and the line

a;

n + n\-

vifritten

Jo
using Problem 2 and the fundamental theorem of the calculus.

a;

=1

is

equal to i.

x^dx
J^

^
=

CHAP.

INTEGRALS

5]

Prove that

5.

Iim;^jsin|
n-foo 71

Let

a-0,

= t,

71

f(x)

+ sin +
71

sin

71

cost

Then

1.

"2 sin

Mw

^'""^

inv^,

+ sin^^?-^l =

...

Problem

in

87

sin

a;

da;

l-cos(

and so
lim
n-oo

using the fact that

^^^ =

lim

i
n

=l

COS

0.

MEASURE ZERO
6.

Prove that a countable point


'"*

.M* Js^ /1^*


"^T^^f
^P<=t'^'y
V'J^'
hu^ ofJ.i--.
of
the lengths
the intervals is
e/2

25(a) of

Chapter

PROPERTIES
'

3),

of
''

set

has measure zero.

and suppose that intervals of lengths less than


where e is any positive number Then the sum
less than e/2 + e/4 + e/8 +
= e Qet a Jt>
ifi
and r
~ ./2 anH
- ^ m Problem
showing that the set has measure zero
''^

=^^,<c..x,,x., ...

^'^

enclose the points,

iZ

DEFINITE INTEGRALS

''"'^""""^ ^"

f"^'

proJe'thal

^^

m(6-a)

mS

^"^

/(a.)

/(a;)da;

g M where

and

g M(&-a)

(6)

Interpret the result of

(a)

We

(a)

geometrically.

have

mAx
Summing from k =

to

^^Ax,

/(U^**

MAX,,

l,2,...,n

w and using the fact that

(x,-a)

(x.-x,)

...

(6-a;-0

= 6-a

follows that

it

<b - o)
Taking the limit as
the required result.

(6)

Assume
shown

f(x)

/(4)

w-^ and each Axu-^0

and continuous

in the adjoining Fig. 5-2.

in [a, h]

Aa;;c

S M(b - a)

yields

with graph

It is geometrically

evident that

Area

ABCD g

Area under y

i.e.,

w(6-a)

A
fix) IS

/()da;

M(b-a)

similar interpretation can be

made if the
removed. The result also holds
sectionally continuous in [a, b].

restriction
II

= f{x) Area ABEF

f(x)^0

is

Fig. 5-2

are constants,

8.

[CHAP.

INTEGRALS

gg

Prove that
By

f{x)dx\

inequality

Page

2,

\f{x)\dx

Ip'^sinnx,

Then

aj^

P^^^^d^l

liml

If

(a)

Interpret the result of


Since

/(a;) is

by Problem

I*

Jo

prove that there

[a,h],

/(a;) da;

[a, h\,

we can

;. ^(^)

^^

and

such that

w g /()

^ M. Then

and

(see

Chapter

2,

Problems

34, 93).

/() da;

result follows on multiplying

by

a.

with graph as shown in the figure of Problem

such that

in (a,b)

(b-a)/(|)

Since /(x) is continuous it takes on all values between


In particular there must be a value | such that

If fix)

a point

is

find constants

m s

(b)

2w

geometrically.

7,

The required

dx

INTEGRALS

for

(a)

continuous in

n^

<

+*^

and so the required result follows.

r
(ft)

the required result.

p'^l sinwa:

Jo

0,

continuous in

/(a;) is

we have

-^ 0,

w"

MEAN VALUE THEOREMS


10. (a)

'"'

I==l

and each Axt

|Jo

a<b.

if

3,

1=1

Taking the limit as n

7(6),

we can

interpret

j^

f(x)

dx

as

= f(x). Geometrically this area should equal that of a recthe shaded area under the curve y
for some value | between a and 6.
height
6
and
a
base
f(0
with
tangle

FUNDAMENTAL THEOREM
IL

If

F{x)

C f{t)dt

Fix

by the

first

Then

+ h)-

where
Fix)

mean value theorem

if

is

of

any point

INTEGRAL CALCULUS
f{x) is
C

-\\

/(^)

f{t)dt
^

[a,b],

prove that
-.

mdtV
"I

F'{x)

f{x).

/-' + "

f(t)dt
dt
fit)

-^J

between x and x-\-h

for integrals (Problem 10).

interior to

F'ix)

continuous in
x+h
^

^
li_>0

[a, 6],

F(^

+ /^)-^W
ft

Hm/(|)

/(.)

''-*''

since / is continuous.
If

=a

cases as well.

or

= b, we

the result holds in these


use right or left hand limits respectively and

^^^^-

INTEGRALS

5]

gg

Prove the fundamental theorem of the integral calculus.


By Problem 11, if F(x) is any function whose derivative is

12.

where

any constant

c is

= c,

Since F{a)

13. If fix) is

it

follows that

continuous in

is

any point

and F(x)

is

continuous.

li

(see last line of

If

at X

a;
a,

= a and
and x

a;

= 6, we

F{x

dt

/(t)

prove that

[a, b],

lira

Chapter

22,

i?'(x)

we can

or

j' /(t) dt =

4).

F(a)

write

^a

Problem

F(h)

\a,h\,

interior to

r mdt +

F(x)

f(x),

f{t)dt

is

F{b)

F{a).

continuous in

[a,b].

then as in Problem 11,

+ h) -

= limhm) =
"-*"

F(x)

use right and left hand limits respectively to show that


F(x)

is

continuous

o.

Another method:
contifuou^s'"*'^^'"

CHANGE
14.

"

^""''^'"

^'"'^

VARIABLES

of

and

Prove the result {U), Page


Let

F(x)

dF =

Then

f{^)dx

/(a;)

dx,

^^^^^^"^ ^'

^'

SPECIAL METHODS

83, for

and

dG =

of

INTEGRATION

changing the variable of integration.

G(t)

i{g{t)}g\t)dt,

where x

J^

p(t).

f{g(t)} g\t)dt.

'**"^^t^^*

FW

*^"'^' ^^""^ ^'^"^ "^'"* ^"'^ ^ -^(^^ '""^t be

^'

/(^^'^^

="G(t?+I'''^*^'^*'

i^()

Fi.)lZ^'l:^xJb=:.Z^^r..''^^^^
r
f

C
f

/(a;) da;
f(x)dx

dFW =

/{i/()}^'Wdt so that

()

O,

dG(*),

so that

from which

= .

Hence

f{g(t)}g'(t)dt

as required.

15.

Evaluate:

r(a;

(a)

(6)

+ 2)sin(a;2 + 4a;-6)da;

^^

f'

(c)

J*

^2*^)^^

^'^

r2-nanh2i-dx

(^^

X -"
sin
- ^C 7!^^^^
^ a;^

,o\

J-.V(x + 2)(3-x)

Jo

*^
(a)

Method
Let

;^

J
(6)

(a;

+ 4a;-6=M. Then
J

sinwdit

(2a,

+ 4) da; =

dzt.

--cosm +

+ +i
a;

+ 2) da; = J dz. and the integral


= - 1 cos + 4a; - 6) + c

(a;

becomes

(a;'>

2:

+ 2)sin(a;^ + 4a;-6)dx =

Let In

Va;'

1:

Method

^^^

(/)

a;

M.

Then

(da;)/a;

cotMdtt

sin

(a;^

+ 4x - 6) d(a;^ + 4a; - 6) =

-1

du and the integral becomes

Inlsinwl

In

sin (In

a;)

cos (x^

+ 4a; - 6) +

[CHAP.

INTEGRALS

Method

(c)

1:

x-i-=u,

Letting

f'

by Problem

J 1^=

Method

as in

dx

= sin"'^ +

= sm-'[^^)+c.

'-(*^)|-, = -'-(l)-"'"-(-|

sin-'. 2

sin-'

2:

a;-i = M

dx

becomes

this

Then

Let

dx

dx

Method

Now when

1.

.6

-1, m

= -|;

_
_
- 1, u-^.
i

and when

a;

rpi, lo

Ihus

14,
1/2

r'

dx
r'
J-. V25/4 - (a;

da;

J-iV(x
iV(x + 2)(3-a;)

=
Let 2'-^

id)

Let

(.)

sin-'.^

du

Then

sm

' X''

r
j

xdx

V x' + x + l
=

lJ(.2 + , +

^/S^

1,

M = tan-'
ri-/

0;

\/3 sec"

2/,

(dx)/x

_
""

dy.

2,

In

l/^/3

u du
[Ssec'uf"

When

= e,
y'

j/

1")"

irin-'
4^sm

^^

-^

144

V^^* + l

l)_|J_^=

- 1 = Vstan.,
Then the

W6.

i
3

= 1; when

becomes

r^g

Let x

\/3 sec'

dy

r_^+j_ ^, _i
-^^
V^T^^Tl

+ .+

Jo

j;,

5/2 1.
"'"'-^'^

and the integral becomes

i(sin-'a;T

+ i + \/(^T|r+l

u = tan-'

C""^

sin

so that the integral

2.^

l)-.2,(,2

+1 - ^

when x =

u du
IsT^tan^itf"

Let In X

^(^^^r^^^r^

Write the integral as


X

2^

^^,

^%

/,-! ^2\2
X )
i(sm

+ i-i rf,
2j Vx^ + x + 1
1

= -

and 2-'d.

iu'

_,
dx

VisTI^^

'^-^-a/^

^==2^^^

^ C udu
X

- If

sin-'.

- 2'-^(ln 2) dx = du

Then

..

sin-'. 2

^1
-2K

COS

It

dx

V3sec^,.d..

When

integral becomes

dtt

:^

sin

a;

e^
.

j/

= 2.

Then the integral becomes

^^^^-

18.

5]

INTEGRALS

jx" In x dx

Find

if

(a)

-1,

(6)

Use^inte^ation by parts, letting u

(.)

91

n = -1.

ln.,

d.

.d.,

n+

^
a;" +

(6)

19.

a;-Mna;dx

In

x d(ln

V2^Tl =

2a;

2/,

Integrate by parts, letting u


S"

Find

a;

Let

ydy

J"
M

In

(a;

udv

+ 3)

In (x

+ 3),

\{\nxY

= y,dv =

+ l).

da;

c.

S> dy;

and the integral becomes

then d.

= y3'

dy,

In 3

Then

dw
"

'^^

=
a;

= y^"(- + 3)-|/Srl
|-ln(a;

^^1^
^
Jf (a;-3)(2a;
fa; -3 (2a; +
5)

Use the method of partial

3)

3V(ln

ln3

(hTir

"

a;ln(a;

+ 3'

- T*='

tt
^^"*= on integrating by parts,

= fln(. + 3)-i/(.-3 +
9 1n(x

+ 3)| +

+ 3)rfa= = | -

4 In 4

+|l3

+Z

]dx

6?a;.

fractions.

6- 3)(2a; + 6)
a;

Let

a;

^ "^
^
- 3 + 2ST5"

1:

To determine the constants

6-. =
Since this

is

an

and B, multiply both sides by

(a.

- 3)(2. + 5)

to obtain

A(2.

identity,

+ 5) + B(.-3)
or
6-. = 54 - 35 + (2A + 5).
5A~BB = &, 2A + B = -1 and 4=3/11.5 = -17/11
Then
jjln|a;-3|

Method

SS'-ydy

and we have

3),

InS'*^

- l|^-3a; +

Method

= y dy

/ du

x dx.

Then

Determine

Then dx

j/^

wy

dv

||-ln|2a!

+ 5| +

2:

Substitute suitable values for x in the identitTr li\


^ ^^^
find at once A = 3/11, 5 =
-17/11!

W, we

22.

...V(n

(n+V)

a;)

Jw +

+I

^^r^ +
J..

(d.)/.,

da;.

/.ln(. + 3)..

21.

a;

Find J*3v2^+^da;.
Let

20.

In
+ l'"""

du

so that

Evaluate

From

^^
i? example, lettmg
""^
i

da;

Fig. 5-3

^'^

3 COS*

we

sin a;/2

"^^"^ *^^ substitution tan

see that

u
cos

Vl + M'

a;/2

=
Vl + M"

a;/2

u.

a;

=3

and

a;

-5/2

in

[CHAP.

INTEGRALS

92
Then

cos

du

Also

Thus the

a;

cos^a;/2

sin

a;

a;

M^

fTjT^-

_ 2dM
- i+^-

2cos^a;/2dM

r <^w
J ^^T+J

IT,
integral becomes
.

dx

or

^sec''x/2da;

sin*a;/2

1 tan
m/^
^ +an-'M/2

-r

c
i-

i
i tan"'

(A tan x/2)
v^

c.

J"""

Let

r''_rf(cos^_j

J
i.e.

= irV^-I

Then

TT-y.

cos'

or

-;.tan-Mcosj/r

\/sina

Letting a

>rV2

7rV4.

_
-

aa;

^^

J^

Vsmic + vcosa;
= 7r/2 - y, we have

J'"'^

J/

P&=dx

Then
,

from which 2/ =

7r/2

Jo

V^ini
^I^l
dx
Vsin^ + Vcos*

Jo

\/sin

r"

^
.

and 7

-v/cos

"

jj-/4.

The same method can be used

to prove that for all real values of

ir/2

_
^

sin" X
J
+ cos" X

sin" X

1
-'

m,

!L

(see

Problem

first

some definite integrals can be evaluated without


Note: This problem and Problem 23 show that
finding the corresponding indefinite integrals.

94).

NUMERICAL METHODS
25.

da;

Evaluate
rule,

r^

where the

for

EVALUATING DEFINITE INTEGRALS

approximately, using

interval [0,1]

is

divided into

(a)

the trapezoidal rule,

n = 4 equal

Page 85, we
Let fix) = 1/(1 + a:^). Using the notation on
= 1.0000, y.
=
Then keeping 4 decimal places, we have: y. /(O)

(b)

Simpson's

parts.

Ax = {b - a)/n = d y 0)M = 0.25.


= /(0.25) = 0.9412, y. = /(0.50) = 0.8000,

find

j/3

(a)

= /(0.75) = 0.6400,
The trapezoidal

2/4

= /(I) = 0.5000.

rule gives

^{y. + 2y. + 2y. + 2y. + y.} =

f {1.0000 + 2(0.9412)

+ 2(0.8000) + 2(0.6400) + 0.500} =

0.7828.

CHAP.

(b)

INTEGRALS

5]

Simpson's rule gives

+ %. + 2y, + 42/3 + 2,4 = 0|5

{yo

The true value

26. (a)
(b)

93

Evaluate

As

in

Problem

jr/4

e^^dx

J^

estimate the

is

4,

we

^^^-

Now

"*"

lA

'

4!"

0<l<

+ ~5r

++ +_+ __

a'

+ fy + fT+

^^

+ 5"^ + 7^^ +

IC'^eldx
"^'^

\E\
'

3T

0<i<x'

to 1,

X('

x^,

by using Taylor's theorem of the mean


and

find

2!

Integrating from

0.7854.

error.

Chapter

Then replacing x by

+ 4(0.9412) + 2(0.8000) + 4(0.6400) + 0.5000} =

0.7854.

approximately

maximum
28,

{j.ooOO

5!

<
-

+ ^

fih

9T4T

+ ^

r'h'.^
15!-^''^^

<=

(where the error

1-4618

r'a'"

= 'X

-51'^^

E =C' ^e^dx)

+ E
e

= 1IT57 <

0.0021.

APPLICATIONS

(a)

Subdivide the region into rectangles


as
on Page 80. A typical rectangle
is

adjoming Fig.

in

the

Then

5-4.

Required area

in the figure

shown

2=

lim
"-oo

f{Q Aa;*

fc

lim

=
I

(6)

(4-iDAa;.

(4 -a;") da;

" =/()

^2

Assuming unit

density, the moment of inertia


about
the y axis of the typical rectangle
shown above is

ilfiQAx^.

Then

Required moment of inertia

= Um i ^JiQ Ax, =
=

a;^(4

-x'')dx

lim

= iM
15

2 g (4 -

jj) Aa;.

= 4-{;

28 Find the length of arc of the parabola y

Required arc length

29.

[CHAP.

INTEGRALS

g^

j^ Vi

from x =

x^

(dy/dxY dx

y/l

to x

ij Vl + u'

r Vl +

l{iMVT^H?+ iln(M + VT + ^)}|^

4a:'

dx

l.

(2xr dx

du
l\/5

+ iln(2 +

V5)

of Problem 27 about the x


Find the volume generated by revolving the region

Required volume

lim

^yl^x. = ^ f (i-x^dx =

axis.

512,r/15.

MISCELLANEOUS PROBLEMS
and

30. If fix)

are continuous in

^(a;)

for

have

A^

we

31

find C"

(fif(a;)}^(Za;

J^

real values of

all

f^imYdx

(f f{x)g{x)d^
We

prove Schwarz's inequality for integrals:

[a, b],

Hence by Problem 13

X.

B'

(\g(x)Ydx,

t,

of Chapter

using

1,

with^

C =

{f(^)rdx,

(1)

f(x)g{x)dx

A''B^ which gives the required result.

Page 82, under the assumption


Prove the second mean value theorem of equation [8),
to the other assumptions.
addition
in
[a,h]
of the existence and continuity of g'{x) in
f{t) dt.

Then on integrating by

parts,

a.

f{x)g(x)dx

r g{x)F'(x)dx

g{x)F(x)\l

g'{x)F(x)dx

\
*y a

=
Case

1:

g(x) is monotonic increasing,

Then by the generalized

first

i is

in (a, b)

2:

is

F{i)

g'(x)F(x)dx

0.

for integrals (see

^^g'{x)dx

Page

82)

F(i)[g{h)-g{a)]

so that

g(x) is monotonic decreasing,

The proof

mean value theorem

C" f(x)g(x)dx

Case

g'{x)

i.e.

(\'(x)F(x)dx
where

g{b)F{b)

g(b)F(b)

F(i)[g(b)-g(a)]

g{a)F(0

g{b)[F(b)-F(i)]

g(a)

i.e.

similar to that in Case

g'{x)
1.

r
g

0.

f(x)dx

gib)

f{x)dx

we

have,

*^H-^P-

INTEGRALS

5]

32. (a) If /"+^(a;)

continuous in

is

95

prove that for x in

[a,b],

[a,b],

dt
(6) Use
theorem
(a)

We

to obtain the

(a)

(see

Page

use mathematical induction (see Chapter

Assume that

^^dt
we

Lagrange and Cauchy forms of the remainder

it

f(a)

f mat

holds for n

dv,

The

1).

f(a)

= k. Then

/-"(*) = u

We

have by the generalized

f
Letting

giving the

F(t)

LafirmTiflre

i:

/(a)

-^jf^.

/'' + ( a)(a;-a)i'+i

f(x)

since

f(a)

(i)

du

/<--(*)d*

first

fc

+ 1. Thus

mean value theorem

F{t) G(t) dt

F{i)

/<->((), G(t)

= ^^~^,

have

Accordmg

:^

to the

x^

holds for all integers

Page

n S

0.

82),

between a and x

(20),

^,

Page

61],

with

6 replaced

by

x.

,_

(21),

Page

61],

with h replaced by

x.

4^^

/'"^^'Wdf

obtain

form of the remainder, [equation

C -^

lim
M-

we

1,

it

('-)'="'

for integrals (see

G(t) dt

giving the Cawc/ij/ /orm of the remainder,


[equation

Prove that

^*

+ (fe+TyrX

(Ffl)!

F(e)=/^!:!lMi^, ^(, =

Letting

We

/(f)

w=

integrating by parts, using

so that

which shows that the result holds for

33.

result holds for

find

_
-

(6)

in Taylor's
j^^ij'iui

61).

4x^

(^^

+ Zf -

^^xV

(x^

+ 2 + 2x)(x^ + 2 - 2a:).

method of partial fractions, assume

= _ Ax + B ^ Cx + D
+4
- 2a; + 2
+ 2a; + 2
{A
C)x^+
(B~2A
1
2C
D)x'+
+
=
+
{2A-2B + 2C + 2D)x + 2B + 2D
+
A + C-0, B-2A + 2C + C = 0, 2A - 2fi + 2C + 2i? =
0, 2B + 2i? = 1
simultaneously, A = ^, B = 1, C = -i, D =
Thus
^.

x*

Then
so that

Solving

x^ +
Jf_^_

4,

1
i

f x+2
J x^ + 2x + 2

8j
=

(a;

,^ln(x^

i)

i''*

a;''

r
J

+ sj

a;2

a;-2

^^
;' - 2a; + "
x2-2a;-'
2

(a;

+ 2a; + 2) +|tan-(a; +

1)^

l)

" gj

i (^.

(^rriy^TT ^"

da;

(^:rijF+3

_ 3^ + 2) +itan-(x-l) + C

'^*'";.

We

dx

r"

ij^ rj_i

denote this limit by

The

dt
Si sin t^
-j

lim

Evaluate

called

an improper

See also Problem

(M +

+ j tan-' (M - 1)

1)

mt^flrrai o/ tfee /?rst fcind.

Let

if f{x) is

continuous in

,^i^^^"^'

sin.3

[a, b]

then

m(b-a) S

l.u.b.

and

g.l.b.

= im.A^. g

are the g.l.b. and l.u.b. of


and
lower and upper sums respectively.

is

lim?^^

Since f(x)

is

continuous

of /(.) in the interval [x.-u^.],

M.Aa;.

we can
i.e.

= S ^ M(6 - a)

The sums

fix) in [a,b].

exists.

/(x) da;

J^

using the notation of Page 80.

f(Q^x.,

numbers M. and m. representing the


mk S f(x) S Mk. We then have

Such integrals are

78.

that the required limit


conditions of L'Hospital's rule are satisfied, so

Prove that

where

1 ^^_,

ifr^''''

35.

/M!2M + 2 \

X^m'

considered further in Chapter 12.

34.

[CHAP.

INTEGRALS

96

and S are sometimes

find

such that

W
called the

consider the corresponding lower and upper


choose a second mode of subdivision of [a, b] and
sums denoted by s' and S' respectively. We must have

Now

s'

this

S'

(^)

we

and T respectively.

and

using the division points of both the


choose a third mode of subdivision obtained by
lower and upper sums, denoted by
corresponding
the
consider
and second modes of subdivision and

To prove
first

By Problem
s

which proves

89,

S T S

we have
S'

and

s'

S T S S

()

(f).

are
number of subdivisions is increased, the upper sums
to (1) these sums
accordmg
Since
increasing.
monotonic
are
sums
monotonk decreasing and the lower
shall call s and S respectively.
r^a^^rboundTit follows that they have limiting values which we must show that s - S.
we
exists,
integral
the
that
prove
order
to
By Problem 90, s ^ S. In
any
[a,b], it is uniformly continuous. Then given

From

Since

>

0,

it is

(2)

also clear that as the

we can

interval
is continuous in the closed
take each Ax. so small that M^-m^

S-s =

<

(M,.-m4Aa;fc

e/(b

<

- a).

It follows that

^37^2^^^" "

U)

positive
and it follows that each term in parentheses is
it must be zero, i.e. S-s.
number
definite
is
a
5
S
since
and so is less than e by (i).' In particular,
and the proof is complete.
Thus the limits of the upper and lower sums are equal

Now

S-s = (S-S) +

(S-s) + {s-s)

'

C^^P-

INTEGRALS

5]

Qrj

Supplementary Problems
DEFINITION
36.

(a)

DEFINITE INTEGRAL

of a

Express

x^dx

j^
integral,

37.

38.

39.

definition, evaluate

Prove that

Jim J

Prove that

^^

/l"

2-

(a)

^ +

3-

(Sx

42.

43.

Work Problem
Prove that

lim J

Prove that

lim

PROPERTIES

i -^L^

45.

If f(x) is integrable in (ac)

47.

48.

If

/(a;)

Property

Prove that

2,

(6)

cos

51.

and

'^

x'

In 2

a;

for

Prove that there

>

1.

/,

/s.

,^0.

is

for

a value

/(a:)

(a)

J"

/() da:

^(a;),

( f(x)dx

prove that

f
s

f(x)dx.

f^Wdx.

INTEGRALS

{^

and

J in

^ e^in ^= cos

|,

in

sin TTX

OSxSir

m S f(x) S M,
Q^xSl

a;

dx,

(6)

(a)^esina:3

^^

S J^-^'V\'^)
f(x)ff(x) <
m 9(h
g(x)
JNow mNow
in
-M

such that

57

C e-

such that

of

mg(x)

then

cos

a;

da:

gin

INTEGRATION

f'^^^^^dt,

C'

(c)

-^1

^'^s.

127-

^^U-.16^-

J/(a;)da;

s-/2.

VARIABLES and SPECIAL METHODS

Evaluate:

jf

'''*

Prove that there exist values

of

all n.

Prove the result (5). Page 82. [Hint: If


tegrate and apply Property 7, Page
82.]

CHANGE
53.

sm X

+l

and

in [a,h]

X
52.

if

(6)9

e".

prove that

(c, 6),

for

ajVs-

Jo

Prove that

a;

MEAN VALUE THEOREMS


50.

...

(a) 8.

Property 3 on Page 81.

f'^^iM^^ s
x+1

49.

te^

and 9(x) are integrable

Prove that

Ans.

DEFINITE INTEGRALS

of

Prove

46.

44.

(a)

e''

using Problem 94 of Chapter

5 directly,

^_l

(6)

C\x--4x)dx.

(6)

result of (a) to evaluate the given definite

Ans.

l)dx,

n-

e'dx
41.

Use the

(5)

Interpret the result geometrically.

(c)

Using the

as a limit of a sum.

(6),rV32,

(c)W3,

(d)

-~:=,
V4a!

-2

(d)

coth y/H

-^

a;'

c.

csch'V^

V^

(e)iln3.

^
**'

INTEGRALS

98

54.

55.

Show

(a)

J^

Prove that

(6)

56.

that

fVa''

Find

^2-'

-ia"

yfof+2xTh -

Ans.

sin"' w/a

In

c,

a;

>

0.

58.

Evaluate

59.

Show that

(a)

cos

a;

(a)

3a; da;,

tan"'

a;^

(6)

da;

a;

a;

e"^^

M = /(a;) and

= sf(a;)

t;

evaluate

Show

that

[Hint:

wo'-'^^dx

x* sin

J^

Use

Mi;<-

(x

a;

ir(x^

(c)

Prove that

dx

-,

Evaluate

r-^r

i "r

i/q

'^

for

-2/9,

(a)

-le-2-{4a;'

(6)

In 2

>

+ m'V"-" -

- (-1)"

M<'i'da;
I

simplifications occur if m'"'

,.,
l)Ma;^

+
.

^,

-,

+ iy

Prove the

66.

Prove Simpson's

(a)

value,

In 2

.,90

(a)

the trapezoidal rule,

"^

m~
+

x^

^^^

^^"^

Simpson's rule, taking m=4.

(6)

0.6931.

and compare with the exact value

rectangular rule,

(a) U.3ZZ,

Use

1.

(6)

trapezoidal rule,

i.e.

(16)

and

by obtaining the values of

ir/4.

Page

(17) of

85.

rule.

dx

Ans.

(c)

Discuss,

EVALUATING DEFINITE INTEGRALS

approximately, using

0,10,

65.

0?

+ -rrr
x+ 1

r^rr^

(x

1)

XTT/2 sin^ X dx
a;

a;

Compare with the exact

sin^x at

+ 6a;^ + 6a; + 3)+

12)7^ + 48

assume

i.e.

(a;

NUMERICAL METHODS
63.

jr*

l)

partial fractions,

What

(6)

Ans.

da;.

mV"-^'

A B C D]
62.

c.

have continuous wth derivatives, prove that

called generalized integration by parts.


(a) to

Ans.

da;.

- ^ + ^

_3L^

61.

+ 2a; + 5 +

Va:"

+ 2a; + 5

Va;'

Establish the validity of the method of inte^ation by parts.

(a) If

''

57.

60.

"^^ +

J ^^y^?^ "

^'^

= 1m\A?"^ JaMn w + ^f>F^\

\u-\Ja^-u'

^^^

f
^

du

+ 2^_^y.

f v^?^dM

(o)

- m'

(3

[CHAP.

[O)

X^ 3-v+x
_

i.iuo

J-,

(6)

r' cosh
I
j

x^ dx.

APPLICATIONS
68.

Find the (a) area and (6) moment of inertia about the y axis of the region in the xy plane bounded
Ans. (a) 2, (6) ir^ 4
by 2/ = sin X, S x^tt and the x axis, assuming unit density.

69.

Find the moment of inertia about the x axis of the region bounded by y
Ans. ^M, where
is proportional to the distance from the x axis.

70.

Show

that the arc length of the catenary

cosh x

from x

to

= x' and y = x, if the


mass of the region.

In 2

is

density

CHAP.

71.

72.

73.

INTEGRALS

5]

Show

that the length of one arch of the cycloid

Prove that the area bounded by the

xVa'

ellipse

a(tf

sin

y^/b''

'''''''"'''

W3*X'

a^t'To!

(a) If

curve and the lines ^

lemnwcate

p"

''"""'^"^ ^'^

''''''''

cos

e),

^ 9 ^ 2^)

(0

= 0,

and ^ = 0,
Ans. (b)

a' cos 20.

| f*'

is

a^

p^d^.

is 8a.

is B-a6.

d^xS^,

sinx,

" '^^'^^' - =

"^

a{l

'^

about the x axis.

^'^d the

x axis

is

the equation of a curve in polar coordinates,


show that the area bounded

is

/(0)

' *^'

e),

Find the volume of the region obtained by revolving the


curve y

^*'

75.

a;

99

located

by

this

Find the area bounded by one loop of the

(6)

'^<bi

(a) Prove that the arc length of the


curve in Problem 75(a)
length of arc of the cardioid p = o(l - cos^).
Ans.

f Vo^ +

is

(b)

(do/daY

da.

^"

^*:

(h\

^^

T?f^ +,^

MISCELLANEOUS PROBLEMS
77.

Establish the theorem of the

[Hmt: Let
78.

f(x)

Prove that

F'(x)

l^

(a)

mean

f'

for derivatives

Page

in (4),

4,

and give a geometric interpretation of the

lim

(6)

first

mean value theorem

for integrals.

f'^

lim

(c)

6,

f"'

^^

. K

results.

j^^^,

[These limits, denoted usually by

from the

81.]

and

f^ 4^

X'7=

respectively, are called

^'""^ ^'"" Problem" SSrsince the integrands are not


bounded in the
t^Z7"^^T'"'f
'^f'
range
of mtegration.
For T"""^
further discussion of improper integrals, see Chapter
12.]

x^e-'dx
80.

Evaluate

fj^.
1+x^'

(a)

(6)
'^

Jo
Ams.

(a)

-^

(6)

(c)

4!

"'"^'^

24

lim

(6)

= -

'^^

f
^^

Xf"' (sin.)-'^*'.
d^;

Xf" ^TT^f^-

(rf
()

does not exist

3V3
81.

82.

Evaluate

Prove:

^hm^

^ f^yt^ +

83.

Prove the result

84.

Prove that

85.

^-^^^

(13)

+ 1) dt =

on Page

= 0. But

3.

+ .' - 2x3 + 3.- 2.,

(6)

A J^^^^

,.

dx

4,

87.

Evaluate

lim
""

2. cos .^

cos

.^

- _

^'^

l).

f
^
the .transformation
- -A usmg 4.^
x-l/y.
J_il + a;= ~
J_^ 1+2^2
= tan- (1) - tan- (-1) ="1/4 -(-./4) = ./2. Thus ./2 = 0.
I

Vl + a;'

JV^^^ + ^f^^ +
1

^ V21n(V2 +

f^' ^r^^^f^^

(.)

'

1/2

J,

^,

83.

VlTl^x

(a)

Explain the fallacy:

Hence /

^,,. ,/,^

;^57^

2
-

+ V^^T^ l
h

2,0 /^
Ans. |(2V2

,.
1)

88.

Prove that
[Hint:

division

89.

[CHAP.

INTEGRALS

100

f{x)

jj

j^

^^

SlonaT^

"* Eiemann integrable in

[0, 1].

then irrational points of sub(2), Page 80, let |k, fc = 1,2,3, .. .,w be first rational and
and examine the lower and upper sums of Problem 35.]

In

Prove the result

(3)

of Problem 35.

First consider the effect of only one additional point of

[Hint:

subdivision.]

Assume

[Hint:

90.

In Problem 35, prove that s

91.

If /() is sectionally continuous in [a, 6],

S.

the contrary and obtain a contradiction.]

prove that

f{x)dx

exists.

[Hint:

Enclose each point of

discontinuity in an interval, noting that the sum of the lengths of such intervals can be
Then consider the difference between the upper and lower sums.]

made

arbi-

trarily small.

< <1
=1
[6a;-ll<a;<2
(2x

92.

93.

If

f(x)

Evaluate

{a;

a;

-h |.}

[a;]

/"17-/2

(a)'
^

Prove that

sm" X

Prove that

::

,0.5

Show that

j-

[x]

-\-

dx

rr dx
cos" X

ir

-r

for

all

exists.

f cos^ X

Ans. 9

denotes the greatest integer less than or equal to

Ans. 3

0.4872 approximately.

X dx
j

Interpret the result graphically.

tan* x

dx

tan~' X

J'

97.

Jo

f(x) dx.

dx

J'zu'

95.

"^o

sin" X

-:z

Jo

find

where

dx

pret the result graphically.


94.

^2

a;

^ 3

_ w^
" 2V2

'

real values of

m.

x.

Inter-

chapter 6
Partial Derivatives

FUNCTIONS of TWO or MORE VARIABLES


A variable z is said to be a function of two

variables x and y if for each given pair


or more values of z. This definition is in keeping
with the
general definition of function as a correspondence between
two sets (see Page 20)
Here
the two sets are (1) a set of number pairs {x,y)
[represented geometrically by a two
dimensional point set in the xy plane] and
(2) a set of real numbers represented
by
c
r
j the
variable z.

we can determine one

{x,v)

i-

We

use the notation f(x,y), F{x,y),

to denote the value of the function at (x


y)
^^so sometimes use the notation z = z(x
y)
that in this case z is used in two senses, namely
as a
etc.,

^^'''^^' ^ = ^^'^'^^' ^*^W^ ^^"


f"
It should be understood

^".u "^""l
although
function and as a variable.

Example:

If

The concept

f(x,y)

x^

2y\ then

easily extended.

is

/(3,

Thus

-1)

(3)^

2(-l)'

w = F{x,y,z)

at (x,y,z) [a point in 3 dimensional space], etc.

DEPENDENT

7.

denotes the value of a function

INDEPENDENT VARIABLES. DOMAIN

and

of a FUNCTION
a dependent variable and x and 2/ the independent
variables
m,
Ihe l^
function is called single-valued if only one value of z
corresponds to each pair (x y)
for which the function is defined. If there is
more than one value of z, the function is
multiple-valued and can be considered as a collection
of single-valued functions.
Hence
^

we

= ^i^_>y)' we

call z

shall restrict ourselves to single-valued functions,


unless otherwise indicated.

^'^^l^^^

* ^^^"^^^ (points), {x,y) for which a function


domain of the function.

.
of definition or simply

Example:

If

= Vl-(x^ + j/^), the


+ y-^ X,

(x,y) such that x'

having center at

(0, 0)

is

defined

is

called the

domain for which

z is real consists of the set of points


the set of points inside and on a circle in the xy
plane
and radius 1.
i.e.

THREE DIMENSIONAL RECTANGULAR COORDINATE SYSTEMS


A three dimensional rectangular coordinate system obtained by

mutually perpendicular axes (the

domain

constructing 3

y and z axes) intersecting in point O (the origin)


forms a natural extension of the usual xy plane for
representing functions of two variables
graphically A point in 3 dimensions is represented
by the triplet {x, y, z) called coordinates
of the point. In this coordinate system z =
f{x,y) [or Fix, y,z) = 0] represents a surface.
Example:

The

x,

set of points (x,y,z) such that

hemisphere of radius

and center at

= Vl-(x^ + 2/^)

comprises the surface of a

(0, 0, 0).

^f^^""'^tions of more than two variables such geometric interpretation


fails, althe terminology is still employed. For example,
{x,y,z,w) is a point in 4 dimensional space, and w = f{x,y,z) [or F{x,y,z,w) =
represents a hypersurface in 4 di0]
mensions; thus x^ + y2 + ^2 + ^2 = ^2 represents a
hypersphere in 4 dimensions with
radius a>0 and center at (0,0,0,0).
^u
though

101

[CHAP.

PARTIAL DERIVATIVES

102

NEIGHBORHOODS
where 8 > 0, is called
\x - Xo\ <8,\y- yo\ < 8
set of all points {x, y) such that
< |a: - a;o| < 8. < \y-yo\< 8 which
a rectangular 8 neighborhood of (xo.yo); the set
neighborhood of ixo,yo). Similar remarks
excludes (xo,yo) is called a rectangular deleted S
circular 8 neighe.g. {x-x,f + {y-yof < 8^ is a

The

can be made for other neighborhoods,


borhood of {xo,yo).

or cluster point of a point


point {x,,yo) is called a limU point, accumulation point
the case of one
of
S. As
points
contains
set S if every deleted 8 neighborhood of {xo, y,)
(the Weirpomt
limit
one
least
dimensional point sets, every bounded infinite set has at
points is
limit
its
all
containing
set
strass-Bolzano theorem, see Pages 5 and 50).

called a closed

set.

REGIONS
point of S if there exists
point P belonging to a point set S is called an interior
point P not belonging
S.
to
belong
a deleted 8 neighborhood of P all of whose points
of P all of
neighborhood
deleted
8
a
to S is called an exterior point of S if there exists
point of S
boundary
a
called
is
S
to
whose points do not belong to S. A point P belonging
not
points
also
and
S
to
belonging
contains points
if every deleted 8 neighborhood of P

belonging to S.
joined by a path consisting of a finite number of
then S is called a connected set.
broken line segments all of whose points belong to S,
points or interior and boundary
interior
of
A region is a connected set which consists
An open region
points.
boundary
its
all
containing
points. A closed region is a region
If

any two points of a

set

S can be

consists only of interior points.


in Figures 6-l(a), (b) and (c) below.
represents the set of points
boundary,
the
including
The rectangular region of Fig. 6-l(a),
for one
interval
closed
the
of
extension
which is a natural
excluded.
being
boundary
the
to
corresponds
dimension. The set a<x<b,

Examples

of

some regions are shown graphically

a^x^b

a^x^b c^y^d

c<y<d

simple closed curve (one which does


In the regions of Figures 6-l(a) and 6-1(6), any
can be shrunk to a point which
not intersect itself anywhere) lying inside the region
6-l(c)
Such regions are called simply-connected regions. In Fig.
also lies in the region.
canregion
of the "holes" in the
however, a simple closed curve ABCD surrounding one
multiplySuch regions are called
not be shrunk to a point without leaving the region.

connected regions.
(&)

(a)

(")

c:>
X

<

Fig. 6-1

LIMITS

may not be
Let f(x,y) be defined in a deleted 8 neighborhood of (xo.yo) [i.e. f{x,y)
approaches
and
x,
approaches
x
as
y
of
f{x, y)
defined at {xo, 3/0)]. We say that I is the Umit
if for
f{x,y)-l\
lim
[or
^l
Mm
f{x,y)
approaches (a;o, 2/0)] and write
2/0 [or (X, 3/)
(a:,i^)--(xo,i/o)
X'^Xq

CHAP.

PARTIAL DERIVATIVES

6]

any positive number e we can find some positive


general] such that \f{x,y)~l\ < , whenever
If desired

we can

103

number

[depending on

0<\x~xo\<8

use the deleted circular neighborhood

Let

={^-

/(...)

As .

^^ = Sg-

If

gets closer to 3(1)(2)

-. 1

and .

and we suspect that Jim f{x,y)

show that the above definition of limit with 1 = 6


supplied by a method similar to that of Problem
4.
Note that }im f(x,y)

/(1, 2)

and

since

/(1, 2)

is

2 [or

6.

satisfied.

0.

The

{xo,yo), in

0<\v~l\<8

< (x-XoY +

mstead of the deleted rectangular neighborhood.


Example:

and

(v-vo)''
;
vj'

(., .) -^ (1. 2)

To prove

this

<

S^

],/(, ,)

we must

Such a proof can be

limit

would in fact be 6

3/->2

even

ii

fix,y)

(X, y) -* (a;o, yo)

were not defined at (1,2). Thus the existence of the limit


of f{x,y) as
is m no way dependent on the
existence of a value of f(x, y) at (xo, yo).

Note that in order for

Jim^ ^^Jix,y)

^^

to exist,

"

it

must have the same value

re-

^""'^^^ ^^^*

^"
'^ *^ ^i^^^^"* approaches
give
Sve^dSerL^v
T^'^'tt
f'^^
different values,
the limit
cannot exist (see Problem 7). This implies, as in the
case
of functions of one variable, that if a limit
exists it is unique.
The concept of one-sided limits for functions of one
variable is easily extended to
functions of more than one variable.

Example

^lim tan"' (y/x)

1:

^'^'"^''^-

lim tan"' (y/x)

^/2,

a:-+0-

= ~W2

y-ti

Example

lim

2:

tan-

(2//a:)

does not exist, as

proaches of Example

is

clear

from the fact that the two

different ap-

give different results.

^"^""^^ *^^ theorems on limits, concepts of infinity, etc., for functions


of one
^^^ ^^^^^ ^' '^^"' ^'^^ ^PP^opriate modifications, to functions of two
ITmlre varifbles

I'?

ITERATED LIMITS
The

iterated limits

^"d

ii^iS/(^'^)
they must be equal

| lim

ii/(^'^)
lim fix,y)

if

oi this last limit.

lim

is

/(x,

and

respectively]

limUim
a^e

fix,y)\,

[also

\U "^necessarily

equal.

denoted by

Although

to exist, their equality does not guarantee


the existence

j/-*;/"

lnn(

1)

-1.

Thus the

iterated limits are not equal and so lim /(x,

2/)

cannot

exist.

CONTINUITY
(X.
(xo

as well as near it]. 'V


yTJtn^^.\t^r"tf
We
2/0)

znai\r(x,y)
conditions

t(xo,yo)\

must be

<

'

"l^h^jhd

say that f(x,y)

whenever

satisfied in

f i^O'yo)
is

[i-e.

f{x,y)

continuous at (xo,yo)

must be defined at
if for any positive

x-xo < 8 and \y-yo\ < 8. Note that thrPP


order that fix,y) be continuous It
(tyo).

[CHAP.

PARTIAL DERIVATIVES

]^04

lim

1.

f{x,y)

i.e.

I,

the limit exists as {x,y)

{xo,yo)

-^

(.r,y)->(xo,o^

2.

f{xo, yo)

3.

must

exist,

i.e.

f{x, y) is defined at {xo, yo)

fixo,yo)

write this in the suggestive form

we can

If desired

Example:

If
/(a;,

= [f'

fi.,y)

t:y)

*-

= a:2y

lim

f(x, y)

.Ji^..J^^''^ =

'

/(lim x, lim y).

continuous at (1, 2). If we redefine the function so that f(x, y)


(1,2), then the function is continuous at (1,2).

y) is not

(x,y)

^^"'^^

f^'''^'

6 for

discontinuous at
a function is not continuous at a point {xo,yo), it is said to be
If, as in the above example, it is
{xo,yo) which is then called a point of discontinuity.
discontinuity that the new
possible so to redefine the value of a function at a point of
of the old
function is continuous, we say that the point is a removable discontinuity
it is conif
plane
of the xy
function. A function is said to be continuous in a region
If

tinuous at every point of %.


can, with
of the theorems on continuity for functions of a single variable
variables.
more
or
two
of
functions
to
extended
suitable modification, be

Many

UNIFORM CONTINUITY
and also (xo.yo) in
In the definition of continuity of f{x,y) at (xo,yo), 8 depends on c
not on any parbut
on
only

we can find a 8 which depends


general. If in a region
then
f{x,y) is said
in
points
%],
[i.e. the same 8 will work for all
ticular point (xo,2/o) in
it can be
variable,
one
of
functions
As in the case of
to be uniformly continuous in %.
uniformly
is
region
bounded
proved that a function which is continuous in a closed and

continuous in the region.

PARTIAL DERIVATIVES
of a function of several variables with respect to one of
constant, is called
the independent variables, keeping all other independent variables
Partial derivatives
the partial derivative of the function with respect to the variable.

The ordinary derivative

and y are denoted by

of f{x, y) with respect to x

f.{x,y),

or

fx,

when

respectively, the latter notations being used

and

-^

fx{x,y),

-i-

dy

or

/,

needed to emphasize

it is

which variables are held constant.

By

definition,

df

ei

when

fix + Ax,y)- f{x,y)

,.

llZ

these limits exist.

often indicated by

Example:

If

must be continuous
alone is not enough

fix,y)

_
"

df

'

+ 3xy\
+ 3(2)^ =

2x'

6(1)^

f^xo.yo)
fx{xo,yo)

then
18,

|^

and
/.

/(1,2)

df/dx

in

,,

At/

^^

fyixo,Vo)

respectively.

Gx"

+ Sy" and

6(1)(2)

has continuous partial derivatives


to

f{x,y+Ay)-f{x,y)

at the particular point {xo,yo)

The derivatives evaluated

/.(1, 2)

If a function /

df/By

6xy.

are

Also,

12.

df/dx, df/dy

in a region, then /

However, the existence of these partial derivatives


the region.
guarantee the continuity of / (see Problem 9).

CHAP.

PARTIAL DERIVATIVES

6]

105

HIGHER ORDER PARTIAL DERIVATIVES


If f{x,y) has partial derivatives at each point {x,y)
in a region, then df/dx and df/dy
are themselves functions of x and y which may also
have partial derivatives. These
second derivatives are denoted by

B_(sl\d^^
dx\dxj
If /x.

and

othervsrise

fd\_3^_

d^f

6^ =

8/3f\

,
f^^'

a^l^toj

8^f

If

f(x,y)

+ 3xy' (see
Ua.2) = 12,

2x'

In such case

In a similar

manner higher order

preceding example), then f..= 12x, fyy


/,(!, 2) = 6, /^(1,2) = ^1,2) = 12.

derivatives are defined.

,
= ^-

fl^to

fy, are continuous, then /., = /,, and the order of differentiation
they may not be equal (see Problems 13 and 43).

Example:

is

dfsf\_

ey^dyj-sy^-f'"' ^[b^ '


J

^-'

dx-

immaterial-

is

= 6x,

For example
^

'

(^)

Ly

= 6y =z

^'^

dx^dy

f
""'''

the derivative of / taken once with respect to


y and twice with respect to x.

DIFFERENTIALS
Let ^x

= dx and Ay = dy
=

Az
is called

the increment in z

be increments given to x and y respectively.

f{a:+Ax,y+Ay)

f{x,y).

If f[x,y)

f{x,y)

^^

where

= to^^ +
e^

and

e^

"^

V^^

'i^^^

'2^^

^'

(5)

first partial

^d^ + ^dy +

approach zero as Ax and Ay approach zero

pression

a/

has continuous

a region, then

Then

e^dx

(see

derivatives in

e,dy

Problem

14)

a/

The

(4)

ex-

ft

or

aJ^^+a|^^

df

%dx +

'ldy

(5)

the total differential or simply differential of z


or /, or the principal part of
^*^ *^^* ^^ ^ '^^ '" general. However, if Ax = dx and Ay = dy are
"small",
fxf
""^^^^ approximation of Az (see Problem 15).
''^'.^
The
quantities
dx
and
dy,
called
^JlT"
f
differentials of a; and y respectively, need not be small.
is called

"'"/

If / is such that

zero as

A/

(or az) can be expressed in the form


(4) where
zero, we call / differentiable at (x.y). The

e, and ., approach
mere existence of
however, continuity of U and /

AX and Ay approach

and /, does not


itself guarantee differentiability;
does (although this condition happens to be slightly stronger
than necessary).
and / are continuous in a region %, we shall say that is
/
continuously
/x

In case f,

differentiable in

THEOREMS

on

%.

DIFFERENTIALS

In the following we shall assume that all functions


have continuous first partial
derivatives in a region %, i.e. the functions are continuously
differentiable in %.
1.

If

f{Xi,X2,

..,Xn),

then

regardless of whether the variables xi,X2, ...,Xn are


independent or dependent
on other variables (see Problem 20). This is a generalization
of the result (5).

In

(6)

we

often use z in place of

/.

'

[CHAP.

PARTIAL DERIVATIVES

106

2.

Xi,X2,

3.

f{xi, Xi, ..., x)

If

.,Xn cannot all

of f{x,y) if and only


exact differential.

=
dx

dz

P{x,y,z)dx

The expression

Qdy + Rdz

^^ =

if

tial

4.

+ Q{x,y)dy

P{x,y)dx

The expression

is

a constant, then df
be independent variables.

c,

Q(x,y,z)dy

if

-\-

Q dy

R dz

-X-

is

the differen-

if

an

P dx +

dQ

dR

5^' g^

an exact

is called

is called

dQ
aP
=
^

.^

and only

case

this

or briefly

R{x,y,z)dz
,

in

Pdx + Qdy

I" ^"^^ ^^^^


"a^-

P dx

Pdx + Qdy

or briefly

the differential of f{x,y,z)

In such case

Note that

0.

5^'

differential,

Proofs of Theorems 3 and 4 are best supplied by methods of later chapters (see
Chapter 10, Problems 13 and 30).

COMPOSITE FUNCTIONS
where x = g{r, s), y = h{r, s) so that

DIFFERENTIATION
Let z

f{x, y)

of

_
~

dz_

dr

In general,

^^^"

if

du

w =

dzdy

dz_

dy dr

ds

F{xi,

xi, X2,

.,Xn

ds

Xi

/i(n,

fc

dxdrk

depend on only one variable


du dxi

du dXi

dxi ds

aa;2

,^.

.,

Xn

l,2,

fn{ri,

/g\

from one

rules.

HOMOGENEOUS FUNCTIONS

on

function F{xi, X2, ..,Xn) is called homogeneous of degree p


the parameter A and some constant p, we have the identity
.

F{XXi, \X2,
Fix.y)

(8)

dXn ds

'

Higher derivatives are obtained by repeated application of the chain

Example:

.,rp),

then

s,

results, often called chain rules, are useful in transforming derivatives


to' another.
variables
set of

,p

These

EULER'S THEOREM

du dxn

"

ds

Then

s.

dy ds

.,rp),

and

dzdy

dzdx^
dx ds

dud^

dud_x^
dXidrk

_
~

du

where

.,x)

dudjc^
dxidrk

drk

If in particular

dzdx
dx dr

z is a function of r

x*

2xy^

F(\x,\y)

.,

by*

(\x)*

XXn)

is

A" F{Xi, X2,

2(\x)(\yr

5{\yr

\*{x'

(see

dF

all

values of

(10)

4,

since

+ 2xy' - 5y*) =

if

F{xu X2,

.,

\*F{x,y)

Xn)

is

homo-

Problem 25)

dF

+
xi^ + X23
3a;2
-

aici

for

Xn)

.,

homogeneous of degree

Euler's theorem on homogeneous functions states that

geneous of degree p then

if,

dF
+ - =
.

dXn

pF

,^^.
(ll)

CHAP.

PARTIAL DERIVATIVES

6]

107

IMPLICIT FUNCTIONS
In general, an equation such as F{x,y,z) =
defines one variable, say z, as a function of the other two variables x and y. Then z is
sometimes called an implicit function
ot X and y, as distinguished from a so-called explicit
function f, where z = f(x y) which
^
is such that F[x,y,f{x,y)] = 0.
Differentiation of implicit functions is not
independent variables are kept clearly in mind.

difficult

provided the dependent and

JACOBIANS
and G{u,v) are
F and
determinant defined by
If F{u,v)

briefly the Jacobian, of

differentiable in a region, the Jacobian determinant


or
with respect to u and v is the second order functional

dF_

dF_

B{F,G)

dU

dv

Fu

F,

d(u, v)

BG

BG

Gu

Gv

dU

dv

(7)

Similarly, the third order determinant

djF.G.H)
d{u, V, w)
is^c^alled

the Jacobian of F,

G and

Fu

Fv

F^

Gu

Gv

Gw

Mu

Mil

Hu

with respect to

v and w.

u,

Extensions are easily

PARTIAL DERIVATIVES USING JACOBIANS


Jacobians often prove useful in obtaining partial
derivatives of implicit functions,

inus tor example, given the simultaneous e([uations


F{x,y,u,v)

Problem

3"

P"^""^^'

^""^^^^^

^ ^^d

djF.G)

d{F, G)

'

Sy

The ideas are

d(F, G)

F{u,v,w,x,y)

0,

Thus

du
dx

'

if

we

djF.G.H)
d{x, V, w)

d{u, y)

d{F, G)

d{F,G,H)
S{U' '> W)

'

dy

d{F,G)
d{u, v)

consider the simultaneous equations

0,

H{u,v,w,x,y)

as functions of x and y.

In this case.

d{F,G,H)
'

we have
d{F,G)

d{u, v)

G{u,v,w,x,y)

for example, consider u, v and

In this case,

d{F,G)
dv
9x

d{u, v)

easily extended.

v as functions of x and y.

^{y,v)

d(u, v)

we may,

G{x,y,u,v)

0,

d{F,G)

dU
dx

dw

d{u, V, y)

'dy
dy

djF.G.H)
d(u, V, w)

with similar results for the remaining partial


derivatives (see Problem

33).

(see

[CHAP.

PARTIAL DERIVATIVES

108

THEOREMS

on

JACOBIANS

In the following
1.

we assume

necessary and

that

sufficient

functions are continuously diflferentiable.

all

Similar results are valid for


If X
(see

equations in

F{u,v,x,y,z)

condition that the equations

can be solved for u and v (for


G{u,v,x,y,z) =
in a region %.
identically
zero
not

2.

,..,,,
example) is that
variables,

where

0,

d(F, G)

-q^jj^

is

m<n.

and ij are functions of u and v while u and v are functions of r and


Problem 45)
^ 5(a;,^) d{u,v)

s,

^^^

then
^^^

d{u,v) d{r,s)

d{r,s)

an example of a chain rule for Jacobians. These ideas are capable of


generalization (see Problems 114 and 116, for example).
This

3.

is

u =

f{x, y)

and v

g{x, y),

then a necessary and sufficient condition that a

J
j-i,
between u and v is that
Similar results hold for n functions of n variables.

functional relation of the

be identically zero.

form

^{u, v)

Further discussion of Jacobians appears

exists

J-

5(m,v)

^^^^

in

Chapter 7 where vector interpretations

=
^

F(u,v)

are employed.

TRANSFORMATIONS
The

X
| ^
{

set of equations

(,f)\
(^0)

^J^^^j

a transformation or mapping which establishes a correspondence


between points in the uv and xy planes. If to each point in the uv plane there corone
responds one and only one point in the xy plane, and conversely, we speak of a one to
differentiable
continuously
are
and
G
transformation or mapping. This will be so if F
with Jacobian not identically zero in a region. In such case (which we shall assume
differentiable
unless otherwise stated) equations {10) are said to define a continuously

defines,

in

general,

transformation or mapping.

Under the transformation


mapped into a closed region %'

{10)

a closed region

of the

tively the areas of these regions,

uv

plane.

we can show

Then

xy plane is, in general,


AA^y and aA denote respec-

of the
if

that
d{x, y)

lim

(li)

d{U, V)

limit as AA^y (or aAuv) approaches zero. The Jacobian on the right
of {11) is often called the Jacobian of the transformation {10).

where lim denotes the

solve {10) for u and v in terms of x and y, we obtain the transformation


often called the inverse transformation corresponding to {10).
f{x,y), V = g{x,y)

If

u =

we

The Jacobians
(see

Problem

^
a\x, y)

45).

and

Hence

5^
o\u, V)
if

of these transformations are reciprocals of each other

one Jacobian

is

different

from zero

in a region, so also is

the other.
ideas can be extended to transformations in three or higher dimensions.
simplicity
shall deal further with these topics in Chapter 7, where use is made of the
of vector notation and interpretation.

The above

We

CHAP.

PARTIAL DERIVATIVES

6]

109

CURVILINEAR COORDINATES
If {x, y) are the rectangular coordinates of a point in the xy
plane, we can think of
(u,v) as also specifying coordinates of the same point, since by
knowing {u,v) we can
determine {x,y) from {10). The coordinates {u,v) are called curvilinear coordinates
of
the point.

The polar coordinates (p, 0) of a point correspond to the case m


the transformation equations {10) are a; = p cos
0, 2/ = p sin 0.

Example:

For curvilinear coordinates

in higher dimensional spaces, see

= p,

-y

= 0.

Chapter

in this case

7.

MEAN VALUE THEOREMS


First mean value theorem. If f{x,y) is continuous in a
closed region and if the first
partial derivatives exist in the open region (i.e. excluding
boundary points), then

1.

f{xo +h,yo+k}

This

is

f(xo,

2/0)

= h f:,(xo + eh, yo + ek) +

sometimes written in a form in which h

k fy{xo + 6h,Vo+ek)

= ^x = x-Xo and

<

<

(12)

= Ay = y-yo.

Taylor's theorem of the mean. If all the nth partial derivatives of


f{x, y) are continuous
a closed region and if the {n + l)st partial derivatives exist in the open
region, then

2.

3""

'
3
is\^,
1 A a
fi^o,yo)+[^h^ + kyxo,yo)+j,(^hl
+ k^jfixo,yo) +

f{xo + h,yo + k)

+ ^(^^to +
where Rn, the remainder after n terms,

^5^j^(^'2/)

+ ^"

...

(13)

given by

is

'"*'
a

^-

(^rnyrV^s^ +

'^ai^y

o<e<i

f{xo+9h,y,+ek)

(u)

and where we use the operator notation

{"^Tx

^(^h^^ +

'^,^J/(-o,2/o)

=
etc.,

where we expand

Equation {13)
y - t/o. Note that
In case

limi

is

(h^

h^fxx{X(,,yo)

k~\

k^^yixo,yo)

2hkf^y{xo,yo)

a special case of {13) where


for

all {x,y)

k^fyy{xo,yo)

{15)

formally by the binomial theorem.

= ax = x-xo and
=
w 0.

sometimes written in a form where h

{12) is
/?

2hk^ +

= Ay

in a region, the result can be used to obtain

=^

an
expansion of f{x, y) in powers of a; - Xo and y-yo convergent in this region,
called the region of convergence.
This series is called a Taylor series in 2 variables.
Extensions to 3 or more variables can be made.
infinite series

[CHAP.

PARTIAL DERIVATIVES

110

Solved Problems

FUNCTIONS
1.

If

GRAPHS

and

= x^-2xy + ^v\

f{x,y)

(a)

find:

/(-2,3);

fix,y + k)-f(x,y)

-.

fi^,

(&)

fc^O.

{-2f

2(-2)(3)

(a)

/{-2, 3)

(6)

K^'f) " (^) "

(c)

f(x,y + k)- f(x,y)


k

= -8 +

3(3f

vXf)^

-i-{[a;

\y)

2a;(j/

12

27

31

i i

3(2/

+ k)'] -

[x'

+ fe) +

^^^

- 2xy + 3y^]}

= V (*'' ~ 2a;j/ - 2kx + Sj/^ + eky + 3k^ - x^ + 2xy - 3y')


=

= ^(-2kx + 6ky + 3k')


2.

-2x + 6y +

3k.

Give the domain of definition for which each of the following functions are defined

and

and indicate

real,

f{x,y)

{a)

this

domain graphically.

\n{{16-x'-y'){x^

The function

is

defined

+ y'-m

and real for

points (x,y) such that

all

(16 - x^ - y^)(x' + V' -

'i)

>

0,

i.e.

<

x^

+ y' <

16

the required domain of definition. This point set consists of all points interior to the circle
origin,
of radius 4 with center at the origin and exterior to the circle of radius 2 with center at the
as in the figure. The corresponding region, shown shaded in Fig. 6-2 below, is an open region.

which

is

Fig. 6-3

Fig. 6-2

- V6 -

f{x,y)

(b)

{2x

+ Sy)

and real for all points {x, y) such that 2x + Zy S


domain of definition.
The corresponding (unbounded) region of the xy plane is shown shaded

The function

3.

is

defined

6,

which

is

in Fig. 6-3 above.

Sketch and name the surface in 3 dimensional space


represented by each of the following.
(a)

2x

+ 'iy + Sz =

12.
(z

= 0)

is

the straight line

Trace on yz plane
12, x = Q.

= 0)

is

the straight line

Trace on xz plane
12, y = Q.

{y

= 0)

is

the straight line

Trace on xy plane

+ 2y -

42/

+ 3Z =

2a!

+ 3 =

6,

= 0.

These are represented by AB,

The surface
points A(6, 0,

0C = 4

0),

is

BC

and

AC

A(P.n.o)

in Fig. 6-4.

a plane intersecting the x, y and_z axes_in the


3, 0), C(0, 0, 4). The lengths OA =6,0B = 3,

B(0,

are called the x, y and z intercepts respectively.

the required

Fig. 6-4

CHAP.

PARTIAL DERIVATIVES

6]

-J.2

(^)^ +

y2

^2

T?

Trace on xy plane

Trace on yz plane

IpI

= 0)

(z

(2/

Trace on any plane

As

(x

Trace on xz plane

Ill

^+ ^ =

the ellipse

is

i^

= 0.

= 0)

is

the hyperbola

^-^ =

1,

a;

= 0.

= 0)

is

the hyperbola

-^ =

=n

z=p

parallel to the

plane

a;?/

is

the ellipse

Pig. 6-5

increases from zero, the elliptic


cross-section increases in size.

The surface

a hyperboloid of one sheet (see Fig.

is

LIMITS and CONTINUITY


4.
Prove that lim {x^ + 2y) =

6-5).

5.

M-^2

Method

1,

using definition of

limit.

exclul;

;JlViJ

Thus

1-28 + 8^

^"'

'

"'

^''"'

<r-r< f7<T-t<

"'''

" ''"" <

<^^<i + 2s + 8'

Now if 6^1, it certainly follows


< x-ll < s
< \v~2\ <

and

Method

2,

+ 2a.

+ 2 - 5 <

a;^

r.

v,

<

<

c.'

Prove that

f{x,y)

By Problem

4,

Jim

{x"

x^

+ 2y)

+ 2y

f(x, y)

lim

a;^

1 -f

continuous at (1,2).

is

Also, /(1, 2)

5.

lim 2m

1^

+ 2(2) =

5.

6.

lim fix, y)

Determine whether

f(l,2)

v)
f(x.
'^'^^

and the function

= / ^^ "^ ^^'
to,

(a)

has a limit as

a;

and

2/

2,

(6)

is

is

continuous at (1,2).

(^' ^) '^ (1' 2)


{x,y) = {l,2)

continuous at

(1, 2).

2-a <

+ 2^, - 5 <

'

S-.2

Then

and

-^'

<

+ a,

Adding.

ie

58

i^^+^^-^i <

'''''

- 4S + 5^ <

or

Th<>

2^/

^""^^

lrr=>
'

using theorems on limits.


lim

5.

- 26 <

-56 <

that
x

'^

*^- l- <

^'

5-4S + S^<^^ + 22/<5 + 4S + S^


ever

'"'

a"'

46

S^

2
+
^
+^3/-5|
<
p;I

^s
5S

i,
when-

[CHAP.

PARTIAL DERIVATIVES

112

By Problem

(a)

Since \im f(x,v)

is

since the limit has nothing to do with the

B,

-2

value at (1,2).

(6)

Umf{x,v)

follows that

it

4,

and

discontinuous at

/(1, 2)

\imf{x,y)

follows that

it

0,

/(1, 2).

'

>

Hence the function

(1, 2).

y^
= \ x^ + y""
[0
x^

7.

Investigate the continuity of

Let

a;

and

3/

in

f{x,y)

such a

x^-y''
_
lTo^^+ "

way

that y

- m^a;'
+ m'x'

a;"

,.

= 'mx

l^^ox'

i^'^'^^

{x,y)

{0,0)

xy plane).

(a line in the

_
"

x\l

,.

i'i^x^d

at (0,0).

- m")
+ m^)

Then along

this line,

1-w^
1

Since the limit of the function depends on the manner of approach to


the line), the function cannot be continuous at (0, 0).

+ m^
(0, 0)

(i.e.

the slope

of

Another method:
Since

cannot

limjlim^X^I =

exist.

Hence

f(x, y)

lim

and

limjlimj^i =

cannot be continuous at

PARTIAL DERIVATIVES
If f{x,y) = 2x^-xy + y^,
8.

(a) df/dx,

find

-1 are not

equal, lim f(x.y)

(0, 0).

and

directly

dfldy at {xo,y^)

(6)

from the

definition.
,

(a)

a/

_
-

V-

/^

,,

J '{> y<>)

f2(a;o

lim

i,-

f{xo+h,yo)-

+ hy -

(xo

h)ya

^.^4te + 2fe'-%,

/.(*o,5/o)

lim

[2a;?

/,

lim(4^

+ yl]

ii-*"

lim/(il!l4lli(^iM

a;o(j/o

/b)

(2/0

fe)']

+ 2fc-j/o)

ft

[2a;S

Xayo

4a;-2/

^^ -kxo + 2kyo + kk_frO

[ixl- Xoyo

h_0

(*)

+ yl\ -

-i

h-fO

f(xo,yo)

""J

/C

lim (-X0

+ 21,0 +

fc)

yl]

- x +

2^.

fc-vO

Since the limits exist for all points (xo,yo), we can write
X + 2j/ which are themselves functions of x and 1/.

fx{x,y)

fx

4a; -.2/,

fy{x,y)

respect to x,
Note that formally fAxo.yo) is obtained from f(x,y) by differentiating with
diiferentiatmg
=
Similarly
l/o) is obtained by
fAxo,
x-=xo,y
putting
J/o.
and
then
constant
keeping j/
while often lucrative in practice, need
/ with respect to y, keeping x constant. This procedure,
partial derivatives are
not always yield correct results (see Problem 9). It will work if the
continuous.

CHAP.

PARTIAL DERIVATIVES

6]

exist but that


()

/40,o)

ii^

/,(0.0)

H^

Let

(6)

a;

and

-^

+m

2/

discontinuous at (0,0).

/(x,2/) is

(ft)

/(fe.O)-/(0.0)

iio

/(0.fe)-/(0,0 )

j.^0

along the line y

-^

213

= mx

xy plane. Then

in the

lim f(x, y)

lim

so that the limit depends on the approach and


therefore does not exist.

not continuous at

Hence f{x,y)

is

(0, 0).

Note that unlike the situation for functions of one


variable, the existence of the first partial
derivatives at a point does not imply continuity
at the point.

Note

also

that

if

(...)^(0.0),

l7Zl2\ZcZJT
10. If

4>{x,y)

+ e-\

x^v

find

(a) ^^,

(6) ^^,

"'""

^ li ^ ^{x'y + e^')

3x'y

^^^

" l| =

x'

M
^^

^"'^

^.

_ 3V _
~ asc^ -

e'"'")

/a^N

a^VW

-0

g^i^^'y

(c) ^^^.

e^y^-y'

+ v'e^y')

^^

and .

= 0.

(d) ^^^,

(.)

3x^y

e^y'-2xy

f^

x'

6a;j/

+
+

,,,

,,(0,0).

/.(0,0)

See remark at the end of

^^,

(/)

^^^.

+ y'e^
2xye'^'

j/=(e^i'.j/)

6xy

y^e'^

= a^ = 5^ (a;' + 2x3/e^) =
+ 2xy '-(e-y") + e^y'-f(2xy)
- 2xye^y^-2xy + eTO*'2a; = 4a; Ve' + 2a; e^
_

(f\
^^^

^^,

*''" '" "^"^^ '^"'""

^''

i^i'^'y

/.

^
^"'^

ay

^ ^f\ -

^V
a^

3a;*

_
=

a^Wy

2xy^e''y^

a /ac6\

3a;*

_a

"

2ye^^
,

^(a;'

2a;3/'e*

+ 2a;2/e^*)

Zx^

2xy

e='y^ -

y^

e-y^

2y

22/ew2

Note that 0^ = 0^^ in this case. This is because the


second partial derivatives exist and
are continuous for all (x,y) in a region
%. When this is not true we may have d> ^ 4.
^"^
Problem 43, for example).
^
^" (see

11.

Show

that

We

V{x,y,z)

{x^

+ y2 + ^2yu2

assume here that (x,y,z)

(0,0,0).

^^^^^^^^ Laplace's partial differential equation

Then

ITT

-^ =
"a^

-i(a;*

+ 3/= + 2*)-'.2a; =

^[-^(='

+ 2/' + 2')-''T

3a!*
(a;

+ 2/^ + zT'*

-a;(a;*

+ y' + zT'"
_ 2a;* - - z*
- (x* + + ^2]

{-=c)[-Ux'

+ j/* + g*)
(a;2 + 3/2 + 22)5
(a;*

+ j/* + 2)-s/2

1/*

j/^

5/2

2x]

(a;*

+ y^ + z^"'

(-1)

'

[CHAP.

PARTIAL DERIVATIVES

114

dW _ 2y^ -x^-z^
-^ - (^r+^qrj^'

Similarly,

-J^+^+

Adding,

= :nan-|,

12. If

^
-

dz_

d^

The

'

result can be written

z^yil, 1)

"51,

prove that

fxy

fyr

G =

Then

f(xo

4>(x,y)

(i)

G =

(3)

region

^^(^^^^_

1.

the fact that

Zx is

continuous at

(see

(1, 1)

remark at the

and

if

/ and

and are continuous at a point

fy. exist

at this point.

Let (xo,y) be the point of %.

Define

9).

13. If f{x,y) is defined in a

of

^^.-^
x' + y^
X
x-' + y^
\ ^ fa' + i/-)(3x-)-(a^-)(2a;) ^

we are using

Note: In this calculation

end of Problem

at (1,1).

X'

-x^-y^
+ y^ + z')"^

2z'
(x'

sz^

+ {y/xYdy\xJ

3rdz\

a^z

find

1^1^^^

,
"^

ez'

d^U

Consider

+ h,yo + k) -

f(x

h,y)

-pixo,

yo

+ k) -

f{xo, yo

f(x,y)

+ k) -

<p{x,y)

(2)

,p(x,yo)

f{xo

G =

(i)

h,

/(o, yo)

+ k) -

f{x,

^{x^

j/o)

h, yo)

f{x,y)

^l^{xo,yo)

(see Page 61) to (S) and


Applying the theorem of the mean for functions of one variable
we have
- fy(xo,yo + e,k)} Q<e,<l
G ^ k<t,y{xo,yo+e,k) = k{fy{x,^h,y^+e,k)
(5)

(6)

=^

H,{xo

+ e^h,y,) =

h{f.{x<,

Applying the theorem of the mean again


(7)
(8)

From

and

(7)

(8)

G =
G =

fe

-^

and

e,h,yo

+ k) -

and

to (5)

(6),

f.(xo

< 1,
< 1,

tf^

ff^

<
<

9,

<

<1
94 < 1

ffj

fy4xo

fe

-*

+ e^h,yo+e^k) =

in (9)

we

f^y(xo

have, since

/x,

fyx(Xa,yo)

and

e^h,ya

/,x

+ e^k)

are assumed continuous at

Let f{x,y) have continuous

this fails to hold, see

first partial

(a;o, J/o),

fxy{Xo,Vo)

Problem

43.

DIFFERENTIALS
14.

<

e,h,yo))

we have

<
<

+ e^h,yo + e^k)
hkf:^{xo + e^h,yo + 9^k)

hkfy.{xo

For an example where

as required.

derivatives in a region

of the xy plane.

Prove that
A/

where a and

=
e^

f{x

+ Ax, y + Ay) -

f{x,y)

U^x +

fyAy

c.ax

..Ay

approach zero as Ax and Ay approach zero.

Applying the theorem of the mean for functions of one variable


{1)

A/

(i),

we have
(9)

Letting

=
=

(see

Page

61),

we have

+ Aa;, + A2/) - f(x, y + Ay)} + {fix, y + Ay)- fix, y)}


0<S^<1, <
Ax fx{x + fl,Ax, y+Ay) + Ay fyix, y + e^Ay)
{fix

2/

9,

<

^^'^^-

PARTIAL DERIVATIVES

6]

Since,

by hypothesis,

and

/,

are continuous,

U{x + e,^x.y + ^y)


where

e^ -* 0, e^ -^

Thus

Ao; -

as

follows that

it

= Mx,y) +

and

II5

fy{x,y

15.

e,

A/

call

U^z =

fy(x,y)

A2/ -> 0.

= /^Aa; + /jAj/ + e,Ax + ^^Ay as required.


= dx, Ay = dy. we have A/ = f.dx + f,dy +
df = /xdx + /d2/ the differential of
/ (or z) or

Defining Ax

We

+ e^^y)

= x^jz-S^/,
-0.01, A2/ = 0.02.

f{x,y)

y~Z, Az =

find

(a)

(d)

How

e,dx

e,dy.

the principal part of A/ (or Aa).

(b) d^.
(c) Determine Az and dz if
x=4
might you determine /(5.12, 6.85) without direct

A^,

computation ?
Solution:
(a)

Az

= fix + Ax,y + Ay) - f(x, y)


= {(x + AxY{y + Ay) - 3{y + Ay)} - {x'y - 3y}
= 2xyAx + (x'-3)Ay + (Axfy + 2x Ax Ay +

{Axf Ay

'
'

(A)

The sum (A)

is

the principal part of Az and

^*^

(d)

= 2xyAx +

'^^

Another method:
(c)

Jb)

Az

=
=

/(x

+ Ax,

d^

and

AT^d^ltZL:^Zr'

that

/(4

{(4)=(3)

= 2xydx + (x^-3)dy =

2x2/

dx

(a;^

+ Ax =

a:

- 3) dj/
3)

+ Ay =

nx + Ax,y + Ay) = f(x,y) + Az is approximately equal to


Now
z = f{x,y) = /(5,7) = (5)^(7)
-3(7) = 154
dz

2xy dx

(x'

direct?omUtio7fll9.oSl^^

- 3) dy =
154

2(5)(7)(0.12)

+ 5.1 =

0.02

e^-1- <i- to the fact that A.

and

5.12

(5^

We

6.85.

f(x,y)

dz,

- 3)(-0.15) =

approximately.

159.1

Thus

z, i.e. dz.

(x^-z)dy

(4^-3)(0.02)

^" ^PP-^-t^I^'

"'

2xy dx

- 0.01, 3 + 0.02) - /(4


= 0.018702
3(3)}

2(4)(3)(-0.01)

+ Ax,2, + A2/) when

/(r.

(x^-3)A2/

/(x,j/)

3(3.02)}

dz

Wemustfind

the differential of

= grf^ + gdj/ =

+ A2/) -

2/

{(3.99)^(3.02)

is

= d.

can accomplish

i.e.

+ dz.

5.1.

The value obtained by

Let C7 = a;V/^ Find dU. (b) Show that


iSx^y-2y^)dx + (x^ - 4xy + 6y^) dy can
be written as an exact differential of a
function ^{x,y) and find this function.

16. (a)

(a)

Method

1:

S
^''^"

Method

Method

If*^"^

^-"

^dy

'"(;
+

(2xey"-yey")dx

xe^'^

dy

2:

=
=

x^die"")

ey'-d{x')

^dy~ydx

^.^y,.(

\^

x^e"d{y/x)

2xe''da;

2x6^'-

(2xey''

dx

ye^") dx

1:

Suppose that

Then

'^^

dU

(6)

= '-"(-)

(^x^y

- 2y') dx +
(,)

(x^

- ixy + &y^) dy =

= 3.V -

2y^

(.)

fj

= ^dx + ^di,"'
dx
dy

dd,

,3

4,^

6^.

a;e'-<

PARTIAL DERIVATIVES

116

From

=
is

x^

F'(y)

x'

Hence the required function

is

4xy

2xy'

F{y)

Substituting this into

the "constant" of integration.

4:xy

x'y

from which

Gy'

F'(y)

2xy^ + 2y^ + c,

x^y

we have

integrating with respect to x keeping y constant,

(1),

where F{y)

[CHAP.

where

yields

(2)

6y\

F(y)

i.e.

2y^

an arbitrary constant.

c is

Note that by Theorem 3, Page 106, the existence of such a function is guaranteed, since if
- 22/' and Q = x' - Axy + 6y\ then dP/dy = Bx^ - iy = SQ/dx identically. If
dP/dy = bQIbx this function would not exist and the given expression would not be an exact

P =

Zx^y

differential.

Method

2:

{Zx^y

- 23/") dx +

('

Then the required function

- 4*2/ +

dx

(Sx^'y

x' dy)

2xy^ + 2y' + c.

x'y

is

- (2y^ dx + Axy dy) + 62/^ dy


= d{x^y - 2xy' + 2y')
d{x'y)
d{2xy^) + d{2y')
2y'
d{x^y
+ c)
2xy'' +

=
=
=

dy

62/'')

This method, called the grouping method, is based on one's ability to recognize exact differential
combinations and is less direct than Method 1. Naturally, before attempting to apply any method,
one should determine whether the given expression is an exact differential by using Theorem 3,
Page 106. See last paragraph of Method 1.

DIFFERENTIATION
17.

Let

of

and x

f{x, y)

COMPOSITE FUNCTIONS
= 4>{t), y = f{t) where /,

Using the results of Problem


dt

smce as At

= e^

dt

_
"

dz_

18. If

_
"

dz
dt

A-o

At

we have Ax

COS

t,

By At

-^ 0,

Ay

sin

(v'e=^y^){- t sin t

= f{x, y) where x =
.,93 _ dzdx_dzdy
'dxdu
^"''du ~
dydu'
(a)

</>(tt,

From Problem
dz

du
(b)

20.

The

14,

Az_

(h\

a"-*oAm

Prove that

dz

dz

3a;

dy

t)

dx

even

if

Ay

dx

= xdu +

Xvdv

+ Xwdw

dz

dy

dy dt

-jr

Tr/2.

(0)(1)

and then

x and

1/

^1

'Am

'

sin

t).

-,rV8.
differentiate.

(2)

Au j

letting Ai;

9z dx

dz dy

dxdu

dydu

-^ 0.

are dependent variables.

Suppose X and y depend on three variables u,v,w, for example.


(jf)

dx

dy

{2xye^y'){t cos

by replacing Au by Av and

-dx+^dy

5z

dx dt

53 a^
dydv'
/, 0, ^t, we have

32 to

dxdv

dy Au

At J

prove that

^{u, v),

^^

^'

-^ -tt

'''

"'"

e*

differentiability of

^^
Au^o^dxAu
j

dz

COS

^'dv

result is proved as in (a)

A.X

(,rV4)(-,r/2)

and y =

v)

assuming the

= 0,y^7r/2. Then ^\

^
0,

"

^Jl\.

At

compute dzldt at

t,

Another method. Substitute x and y to obtain

19. If

*'

e^^

e^ -^ 0,

dz\

Att = TT/2,

^+

i^^_|_^4i
[flccAf

=
|5^
+ |5^
ay at
"X dt

dzd^
dy dt

At-.o

-^ 0,

dzdx
dx dt

we have

14,

are assumed differentiable.

./r

<^,

^^^^

dt/

Then

= y^du +

yvdv

+ ywdw

Prove

CH-^P-

PARTIAL DERIVATIVES

^]

Thus

z.dx

=
=

zydy

+ 2y) du + (z.a;. +
+ Zvdv + Zy,dw =

(z.x.

Zudu

using obvious generalizations of Prob.

r = x^-xy + y\

21. If

3T

may

]L

(a)

psm<j>,

l^J^^d^a^ =

dT

dT

dx_

aa;

30

also be

dr

dp

^JL

^ dU

dx dr

dT dy

find

dT/dp,

(a)

iSx^-y){cos^)

_
~

30

32/

+ 2s,

3r^

dy_

ZyyJ)

dw

(b)

dT/d<t>.

dT dy

dy dr

(^.a;^

dz

^^^^

~ 2/)(- P

1/

{Sy--x){sm^)

sin 0)

worked by direct substitution of x and y

^-^^^^y'^ where

^^'

pcosi>,

dTdx

d^

z,j/.)

19.

_
~

30
This

II7

(Sj/^

- x){p

cos 0)

in T.

- 4r-2s3,

= 2r2-3s^

find

(a)

flCZ/flr,

dz^

dz dr

"f)(- J)} + {(-f)(j)}< + (sinf>


X

"'

(6)

dx 3s

dUdxdUdydUdz

ds

32/

z cos

23. If

ds

dz 3s

f)(-?)}'^'+{( f)}<-)+(='nf)M..

22/?

= pco^^,y =

6s'z

2/

show that

p sin ^,

(^^ + (^^X

(^^X+

=.

(^^^

Using the subscript notation for partial derivatives,


we have

=
=

Vp

V^
Dividing both sides of

(2)

Then from

we have

by

p,

V.Xp

V.x^

+
+

'^p

24.

Show

that z

Let

x^'y

and

(5),

= u. Then

a:

ycos0 + y, sin0

Vyy^

y. (-p sin

0)

yj(p cos 0)

~ ^i

is

Vy sin

0)2

sin

dz du

2*;^,

(2)

Vy cos

(- y^ sin

differentiable,

(3)

satisfies

y,,

cos 0)^

xidz/dx)

= f(u). Thus

/'(k)

(Vx cos ^

where

fix^y),

dz

Then

-^Vl

we have
~^<l>

(1)

y,2/p

...

22/

1^
"

dz

2;.^

and

dz

so

dii

a;|i
3a;

22/^.
32/

V^

2y{dz/dy).

V^

[CHAP.

PARTIAL DERIVATIVES

118

Another method:

We

have

dz

f'{x'y)d{x'y)

Also,

dz

= ^dx +

Then

dx

-r-

dy

x" dy).

dy.

2xyf'{x'y),

Elimination of f'{x'y) yields

f'{x'y)(2xydx

x^f'ix'y).

dz
a;r

values of the parameter A and for some constant p, F{Xx,\y) = X^F{x,y)


identically, where F is assumed differentiable, prove that x{dF/dx) + y{dF/dy) = pF.

25. If

for

all

Let Xx

= u,

\y

= v. Then
= XTix.y)

F(u,v)

The derivative with respect

to X of the left side of (1) is

dF

SF du
8F

d\

du dX

The derivative with respect

dF

26. If

F(x,y)

in (2), so that

x'^y^sin-^y/x,

F(Xx,Xy)

Since

from Problem 25 with p

27.

du

-ri

dv

Then

to X of the right side of (1) is pX^'^F.

dv

=1

SF

_
"

dv

dv dx

.^+yf
du
Letting X

(1)

u=

x,

= y, we

show that

pX^-^F

have

x{dF/dx)

x{dF/dx)

(2)

y(dF/dy)

y{dF/dy)

pF.

6F.

Xy/Xx = xVy^sm'^y/x = X^F(x,y), the result follows


can of course also be shown by direct differentiation.

(\x)*(Xy)^ sin''

= 6.

It

Prove that Y = /( + at) + g{x - at) satisfies d^Y/dt^ = a\dW/dx^), where / and g
are assumed to be at least twice differentiable and a is any constant.
Let u

dY
By
d^Y
^

'

Si^

x-\- at,

dY du

dY

at

Y =

so that

dv

,,,

f{u)
,,

dYt^

dt

_
~

dYt dv

aFt du
du dt

dv dt

-^

f"{u)

{af'(u)

aV"W
2 f",
+

^
d^Y

(^)

dYr

dYx du

dYx dv

,,,,

Then from

(i)

and

(2),

S'^F/at'

a^d-'YIdx').

d^f/du'^,

ag'(v)} (a)

f'(u)

d/Zdw,

g"(v)

,,

p"(^)

\^

-^ {af'(u)

mi

fl''(-D)

,,,

d^g/dv^,

a^9"{v)

dx^

= /"W +

if

dY du ,dYdv

dY

further differentiation, using the notation

_
~

+ g{v). Then

//

\i

dgfdv,
,,

we have

ag'iv)} (-a)

CHAP.

6]

28. If

PARTIAL DERIVATIVES

= 2r-s and

Solving

Then

dr/dx

2r

- s,

+ 2s,

as/aa;

2/5,

find
2s

8U

_a_

^2

ar;

(2^+
^
25 V
U

t,

2acr_iac7
5 dr

5 as ar

(1)

and

dr ds

Solving

(5)

dx

and

dt

dU

Then
n

5 ds

J dy

a''t/Y2

5 ds'

2t

Zy'

5x*

2x

32/

J\5

dUdy

t.

dy dt

t\

(implicit) functions of

2x(dx/dt)

(4)

dt

-2t

(Zx'y}
(

15x*y'

- 2x

- defines z as an implicit function of x


plane, prove that (a) az/ajc = -FJF. and (b) dzjdy =
is

a function of x and

Since x and y are independent,

aF
(')

from which the required

F{x,y,u,v)

aF

5x*

2x

2t

5x*

2x

3y'

(2)

and y

in a region

of the xy

-Fy/F where i^^^O.

sf
aF a^
a^ + "ai a"^

If desired, equations (1)

0,

find

(a)

du/dx,

and

(2)

(6) flM/ai/,

can be written directly.

(c) Sv/Sa;,

'

of x and y,
Ma:dx

+ Uydy

(i)

dv

v^dx

(d)

S-y/fli/.

(implicit) functions of the

dF = F,dx + Fydy + Fudu + Fdv =


dG = Gsdx + Gydy + Gudu + Gvdv =

- 2x

\l5xV - 2x

u and v as
Using the subscript notation, we have

u and v are functions


du

ISxV'

yiOxH - 2x_
(x'

in general define the dependent variables

(1)

2f

^^^

"

results are obtained.

y.

Zy'(dy/dt)

dw.
+ T
dy "

az

a^+a;a^ -

independent variables x and

(S)

- dx
dx

differentiating with respect

we have

and G(x,y,u,v)

The two equations

dx

i/,

Then

diz/dt,

Zy'--2t
15a;V -2x

ac/d X
ax d t

dy/dt

+ y^ =

x^

30. If F(x, y,z)

Also, since

dr

ds'

simultaneously for dx/dt and

(^)

dt

and y as

(2) define x

5x\dx/dt)

(3)

31. If

_ lacA as

a_/' 2 at/

ds\5

J dy

^2 dHJ_ _

5 as"

has continuous second partial derivatives.

Equations
we have

Since z

Hence we have

2/5.

i3f7\ ar
5 ds

IMPLICIT FUNCTIONS and JACOBIANS


29. If Z7 = x^y, find dU/dt if (1) x' + y = t, (2)
to

+ y)/5,

s.

= {2y - x)/5.

-.IE.- o!E

dr'

assuming

_
~

ds

2dHl_l i^C^yi^ +
_^
t;

(2x

as/S^/

ds dx

dr\hdr

dy

s:

1/5,

du

dr^

dr dx

d_fdU
\dx

dy dx

dr/dy

du

ax

_a^

and

f or r

-1/5,

terms of derivatives with respect to r and

in

dydx

119

v^dy.

[CHAP.

PARTIAL DERIVATIVES

120

Substituting

and

(5)

=
=

dF
dG

(5)
(6)

Since x and

2/

and

(i) in (1)

(2)

yields

+ F.u. + F.v,)dx+(Fy + F.Uy +


+ GMx + Gi;.)da; + (G + G.u, +

(F.
(G.

are independent, the coefficients of dx and

Solving

and

(r)

(8)

= -Fx
=
G

|GM. + G.v,

Ux

(c)

ax

-Fy
-Gy

ly

F.

with respect to u and

d{y,v)

du

1-12V

dx

8uv

aw

We

_ -2-4-u
1 _ g^t^

mnemonic

(see also

Problem

du

dv

dx

dx

dv
dx

dy

33,

,.

If

F{u, V, w, X, y)

,,.
^

'

dy

dx

9{u, v)

-Fy
-Gy

a(F,G)

F,

S(F.G)

G.

d{u, v)

F
Gu
J,

or

^\J

,s

32/

4m

find (a) du/dx,

dv/dx,

(6)

to x, considering

3m

Sf

dv

(c)

du/dy,

{d) dvfdy.

u and v as functions

of

du

- Sm'U

Suv

F =

G{u, v, w, x, y)

dw

the Jacobian of

u,v

u'

F.
G.

Fu
G.

= 0.

- v -Sx- y =

G = u-2v^ - x + 2y =

0,

-1

-3
-1

4-u

F.

2m

-1

F.
G.

41)

obtained.
Similarly the other partial derivatives are

0,

i
is

S(u,y)

derivarules for writing at once the required partial

d(u, v)
0.

d(F,G)

G.

d(F, G)

- 8mi; ^

d(u, x)

we have

d(x, v)

dx

a{F,G)

33).

d{F, G)

<?s

dv
'

Method 2: The given equations are


Problem 31,

provided

Fs

2m-3
SM-y'

have, of course, assumed that

du

dy

2m^
dy

-^ -

3(F, G)
Fu Fv
/
denoted by -.
d{u, v)
Gu Gv
and is supposed = 0.

ij

'

iS^
^^'

Hence we

-F
-G

F.
G.

B(u, v)

Differentiating with respect to y,

Solvmg,

=
=

F
G.

HF, G)

to devise

'

Vy

(d)

Method 1: Differentiate the given equations with respect


X and y. Then

Solving,

are zero.

(S)

F.
G.

dV
dx

ejF.G)

u'-v = Sx + y and u-2v' = x-2y,

Uj/

d{u, v)

F.
G.

Note that it is possible


tives in terms of Jacobians

32.

Us

Go

dJF.G)

The functional determinant


and

^'

(6)

F.
G.

Gu

Gu Ms

d(F, G)

F.
G.

Fu
G.

du

li,

and

gives

-F. F.
-G. G.
(a)

G.Vy)dy

Fu Ms
(S)

=
-

F.Vy)dy

in (5)

obtain
^'^^

0,

H(u, v, w,

x, y)

0,

find

1-12V
1

8mi;

0.

Then by

^^^^-

PARTIAL DERIVATIVES

^1

j21
'^^"'- ' ^--bles in terms of

theremr4T*i:3\Llat:sl^^^
S.S., .e

know
that

wo.a .no.

^t:z^::^^'^;tzx^ :sst^:s^::

the remaining independent variable.

we

However, the particular notation pL\

are to obtain Bv/Sy keeping . constant,

i.e.

serves to indicate

the other independent variable.

is

Differentiating the given equations


with respect to y, keeping x constant,
gives

(a)

Solving simultaneously for

Vy,

we have
F

dv

T
^y

Equations

{!), (2)

and

31.

H^ H, H

d{F,G,H)
S{u, y, w)

F.
G

F
G.

F
G

diF,G,H)

/?

H.

H^

d(u, V,

w)

can also be obtained by using


differentials as in Problem

(S)

The Jacobian method


and Problem

F
G

F
Gy

(?

31.

very suggestive for writing results


immediately, as seen in this problem
Thus observe that in calculating
g| the result is the negative of the quotient of
is

s^sriaS: r-^:r=:tj:^^

^^^^

d(F,G,H)

d(F,G ,H)
du

a(F,G,H)

d{F,G,H)

^(''y'^)

34. If

z^~xz-y =

-^
dy

prove that

0,

d{w,x,v)

= _

dx

J^+^
{3z^

xf

v,.s~^r/o:rsSnr; vts/;:rrwr,r "''-'"'


^'

~ ^to ~

"

and

U)

Differentiating with respect to


y, keeping x constant,

^^^^-1
Differentiating
^^^

a.a.

The

'''

(2)

_
-

and

with respect to x and using

-1

/-.

az

(3?^(6z--iJ

result can also be obtained

(1),

we

^
dx

""'^

Sz"

(.)

= "

-X

_^

we have

~j^^^
1

'''

find

6zrz/(3z^-a;)l

by differentiating U) with respect

""^'^^

the

=
to

_l

-0^^

y and using

(2).

continuously

^
differentiate in some
' '''
regiorJ^'pL:eV.t^'^'''^'
necessary and sufficient condition
* ^
that there exists a
!
hI
i\u
honal relation between . and of the
.
form ,(., .) =
is the vanishin/of
Jacobiat

'^-

di^

Te

Identically.

[CHAP.

PARTIAL DERIVATIVES

122

We

Necessity.
^(^''")

have to prove that

To do

identically.

if

d{x, y)

= ^^du +
= (<p,u. +

d^

Then

0Mx

(1)

Now

<f.dv

(<f>Uy

contrary to hypothesis. Hence

it

(2)

from

follows

Uydy)

4>,i^,)dy

-P.v.

and ,. cannot be identically zero since

,pAu.dx

<f,,v,)dx

_
-

d(u,v)

is

<t>.vy

^ aKji ^

that

(2)

.- .,
T
I.have to prove that if the Jacobian ^^^-^
0.
v)
0(m,
tional relation between u and v, i.e.

Vydy)

functional relation.
they were, there would be no

if

and

(1)

+ <f.Av.dx +
=

f>.y-y

We

Sufficiency.

then the Jacobian

exists,

note that

we

this,

,(u,v)

the functional relation

Uy

identically.

d(x, y)

Vy

a func-

then there exists


iripn+icallv tnen
identically,

f.

= and Uy = 0. In this case the Jacobian


Let us first suppose that both u.
relation = c. is obtained.
functional
trivial
the
that
so
a constant Cu

identically zero

is

and

both u. = and . = 0; for "iefi-;^


Let us now assume that we do not have
F(u,y},
T^FiZ^,
to obtam x
solve for x in the equation ,. - f{x,y)
108,
Page
may then, according to Theorem 1,
from which it follows that
V = g{F{u,y),
(2)
u = f{F(u,y),y}
(1)

From

these

{S)

du

(-4)

dv

From

we have

respectively,

= u.dx + uydy = u4F.du + Fydy) + uydy = u.F.du +


= v.dx + vydy = vAFudu + Fydy) + Vydy = v.F.du +

(3),

u.Fu

and u,Fy

+ m, =

or

= v.F.du + {vA-uJu.) +

dv

(e)

d(u, v)

But by hypothesis

Ux

Uy

Vx

Vy

^(^^

(5)

UxVy

UyVx

(v,Fy

+
+

Uy)

dy

Vy)

dy

U) becomes

this,

UxVy

= v.Fudu +

Vy}dy

Using

-Uylux.

(u.Fy

UyVx

dy.

Ux
identically,

so that

>r: ;tsi S':^i .tf


t^i:^-. r..=.-sr
= exists.
functional relation

36. {a) It
(b)

u=^

4>(u,

^^

Are u and

t;

d{u, v)

and v

= tan^'x +

functionally related?

Ux

Uy

Vx

Vy

(1

r/.i'

find

v)

-^^^y

If so, find the relationship.


l

+ x'

{1-xyY

if

xy^l.

d(x,y)

By ProWn.

other.

9(2^,

tan-^y,

- xyY
+ x'

+ y'

S5. ..no. the J.c.bi.n i. iden.ic.lly

Thus, for example, from v


x

1
1

becomes

i))

1+j/'
(a)

(S)

Then substituting

tan (i;-

tan-, 2/),

this in

^ =

(x

= tan- x +
_
-

tan
i

^;

+ y)ni-xy)

'<f

tan

tan

;! ^l' ^ :

'j/

we

find

tan (tan- j/)


tan (tan -^2/)

tan

t;

and simplifying, we

tan

^an^+ ytanv
j/

find

u = tan^.

i"'*Tl
2/

an

CHAP.

PARTIAL DERIVATIVES

6]

123

>.2_
X = u-v + '^'
w, 2'
y =
u^
= '-^^-^^
and z = u^
u' + v,
^/^' ^' ^^
v. evaluate the Jacobian
Jac
V.,:~
and
(b) explain the significance of the
d{u,v,w)
e non-vanishinff
non-vanishing of this .Tamhinn
Jacobian

37. (a) If
^

s(x, y, z)

(a)

d{u, V,

w)

^u

fl?i>

Xu

yu

yv

y-K

2m

-1
2v

Zu

Zx,

Zv,

3m^

6tyM^

+ 2u + 6uH + 2w

The given equations can be solved simultaneously for


u,v,w in terms of x,y,z in a region
s" >
the Jacobian is not zero in 9^.

(6)

%^

if
ji

TRANSFORMATIONS, CURVILINEAR COORDINATES


38.

region g^ in the xy plane is bounded by a; +


3/ =
termine the region %' in the uv plane into which

tion

y^u-v.
%

= u + v,

the ratio of the areas of


(a)

The region

shown shaded

9?

"

respectively^"

(b)

Compute

and %'.

6,
is

a{x, y)

^g^.
<'{u,v)

(c)

a;

-^ =

and

mapped under
Compare the

2/

0.

(a)

De-

the transforma-

result of (6)
v
) with

in Fig. 6-6(a)
*'"'

below is a triangle bounded by the lines x + y = 6


distinguishing purposes are shown dotted, dashed
and heavy

/
.^y

/
/

y=

{a)

xy plane
(6)

uv plane

Fig. 6-6

I.e.

Under the given transformation the line x + y = 6 \s


transformed into {u + v) +
^M - 6 or M - 3, which is a line (shown dotted) in
the uv plane of Fig. 6-6(6) above.

inthe^u^rilw
m
the uv plane.

^ *"'""'''
Tl^
In
manner, y =

(u

like

2l7L''^Te-eir
a(a;,

(6)

(c)

y)
d{u, v)

dx
du

dx
dv

Sy
du

dv

dy^

+ v)-(u~v) = 2
becomes r* =
1;

^''''''''' ""'''''' '^ ''''''''''''

''''

(u-v)^e

= 1, which is a line (shown dashed)


zt = ^, which is a line shown
heavy
.^ = 3,
. = 1 and . = .. andTsshowT

or ^
or

(u + v) ~{u + v)

iu-v) -(u~v)

-1

The area of triangular region % is 4, whereas the


area of triangular region %' is 2. Hence the
'^'1'"^ "'''''^^ "^'"^ * ''^^ '^'''''"^^ '" (^)- '"<= the Jacobian is constant in
thfcll
"^ " ''' '^ ^'^"^ ^''^ '"''^^ *^ ^^^^^ ''^ corresponding mapped

r
Sons VtTi:
'

p'l^L!'

39.

[CHAP.

PARTIAL DERIVATIVES

124

region

and 1/-0,
xy plane is bounded by x^ + y' = a?, x^ + y^ = h\ x^Q
under the
mapped
is
which
%
(a) Determine the region %' into
what
Discuss
(b)
0g^<2,r.
x = pcos</., y = psin^, where p>0,

in the

0<a<&.

where

transformation

happens when a =

Compute

(c)

0.

-g^.

Compute

(d)

^^^.

%'

= a}
*

O'

''

(6)

Fig. 6-7

(a)

The region
x'^

+ y^ =

'^l

[shaded in Pig. 6-7(a) above]

x^

a" (dashed),

y''

Under the given transformation,


or

and p =
becomes

a,

a^ x^b

= 0,

^p^

The required region %'

is

x'

Also,

respectively.

is

bounded by x=

(dotted),

(dotted

and dashed),

6' (heavy).

+ y' =
x^Q,

a?

and

a;^

a^ySb

+ j/^ =

6^

becomes

become

- jr/2,

p^

= a^

and

=p^

p^

= 6^

y-U,

b;

6.

shown shaded

in Fig. 6-7(&) above.

the origin O of the xy plane and


Another method: Using the fact that p is the distance from
given by
is clear that the required region is
it
axis,
x
positive
the
from
is the angle measured

(6)

S W2

6,

as indicated in Fig. 6-7(6).

b (bounded by 3 sides)
= is mapped into
x
=
point
the
0,y
that
is
this
for
reason
The
while %' remains a rectangle.
which is somepoint
this
at
one
one
to
= an indeterminate and the transformation is not
p = 0,

If

the region

becomes one-fourth of a circular region of radius

times called a singular point.

d{x, y)

dp

(p

cos 0)

4- (P c^s

~
dp

(p

sin 0)

i00

(c)

d(p, 0)

=
{d)

p(cosV

From Problem

45(&)

+
we

*)

cos

sin 0)

sin

(p

sinV)

3(p, 0) 3(a;,

p cos

have, letting

d{x,y) d{p, 0)

u-p,

<p,

a(p,0)
^

so that, using

(c).

3/)

This can also be obtained by direct differentiation.


Note that from the Jacobians of these transformations
is

sin

00

d{x, y)

it is

clear

1
p

why

(i.e.

a;

= 0,

2/

= 0)

a singular point.

MEAN VALUE THEOREMS, TAYLOR'S THEOREM


40.

Prove the

mean value theorem for functions of two variables.


= f(xo + ht, yo + kt). By the mean value theorem for functions of one
< 9 < 1
F(l) - F(0) = F'(e)

first

Let F(t)

variable,

^^^^-

PARTIAL DERIVATIVES

^]

If

x,

+ ht,y =

+ kt,

y,

= Md./dt)+Mdym)

F'(t)

where

<

<

where

<

<

Thus

1,

fixo

(1)

==

then F(t)
hf.

and F'ie)

kf,

by Problem

17,

^ hM.. + eh, y.+

ok)

so that

f(x,y),

125

+ kfAxo + eh,y. + ek)

becomes

+ h,vo + k)-

fix,, y)

hf^ (^

+ eh,y, + ek) +

+ eh,

h /, (a;

y,

Prove Taylor's theorem of the mean for


functions of two variables.
^^'" "" '^*' "'''''^
'' '" ^^"^"^ '" ^^ '^'^^'^'^ *^ "^ the mean
of ontlar^JSle:

F(t)

andifi =

(2)

-.a.

more^;::^Sit^^tS^,::.iil---U^

41.

+ ek)

as required.

F(0)

+ ElMJl +
2!
^

P'(0)t

l,

Prom Problem

...

^'"'(0)
,.
*

ti!

F<'^"{e) ,_,

+ \n +
i

40,

Uj^

+ kMx.,y,) =
where we have used the symbolic operator
notation. Similarly,
i^'(O)

hf4x.,y,)

l)\

o.

be..

for functions

0<e<t

(1)

+ i).

fcf>(.o,^)
^

from which
^"(0)

'^V..(x,2/)

2;ifc/.,(x,2/)

(h^

fcV.(c.o,2/o)

Since the second partial derivatives


of / are supposed continuous.
In like manner we can verify (by
mathematical induction)

that for

where

42.

< < 1.
<P

Expand

Substituting these in

.'y

4-

3. - 2,

/.

. 2.y,

/.

All higher derivatives are zero.


/(I,

-2)

..

+ 3,

= 2y,

/.

= 2x,

- y

/. . 0,

where
/. = 0.

= 1, y,^ -2 Then
/. . 2, /,, = , /,, = o

x,

Thus

= -xo f4l,-2)^-4, /(i,-2) . 4, /,.(!, -2) = -4,


/.(!, -2) - 0, /(!, -2) =
2, A(l,-2) = 0,

/.,(l,-2)
/,(!,

-2)

= 2,
=

/,(!,

-2)

Taylor's theorem,

fi^,y)

where R^

is

/(I,

+ kfAl,-2) + i,{.V..(l,-2) + 2../.,(l,-2) + .V.(l,-2)}


+ 3,{AV...a,-2) + 3A^A/(l,-2) + 3MV.(l,-2) + k'f,il,-2)}
+ R,
-2)

./.(I, -2)

the remainder and

is

zero in this case.

Substituting the values of the derivatives


obtained above,
.^y

positive integers n,

x^y

By

y,)

the required result follows.

(2).

+ ^y~2 in powers of x-1 and 2/ + 2.


Use Taylor's theorem of the mean with h =
x - x k = y

/(., .)

all

+ k^"^ fix,

+ By-2 = -^0~4i.-l) +

,iy

we

+ 2)-2(x-ir +

as can be verified directly in this case


by algebraic processes.

find

2i.-l)iy

+ 2) + i,-iriy + 2)

[CHAP.

PARTIAL DERIVATIVES

126

MISCELLANEOUS PROBLEMS
43.

Let

f{x,y)

\^ +2/ /

Compute

If

(a)/x(0,0),

y)

(a;,

(a;,!/)

.0

(0,0)

(c)/.40,0),

(6)/.(0,0),

/(0,0),

(d)

(/)/.x(0,0).

(e) /x(0,0),

(0, 0),

.y_^^M!_^ +

af...A^-.=_M

,(5^

Then

li^MO^i^L^AM =

(e)

/.(0,0)

(/)

/.^(O.O)

= lim/.(M)-MO,0)

Note that

44.

Show

ew

5-

/.

/.

,.^^^1

We

dW

^
becomes

-r-s-

IdV ^
-

-1

See Problem 13.

at (0,0).

that under the transformation

li^njf

"

^r

cos ^,

= psm^

the equation

^^^

dW _

1
2

aT2

"

have
(-'^

Differentiate
tions of X

a^

'

p cos 0,

"^
a;^

1/

a0 ax

flx

with respect to

p sin

dp ay

sy

'

x,

B4,

remembering that

dy
p

and

are func-

and y

-psin0g+

cos0|^,

PCOS0U+

sin^ll

Solving simultaneously,
dp
-^

Similarly, differentiate with respect to y.

-psin0g+

_
~ _

30
V"
3x

^
cos d>,

ax

sin0

ig)

Then

cos^l^,

pcos0|j+ sin^l^

Solving simultaneously,
02/

a?/

Then from

(1)

dv

and

(2),

3V

sin

ay

,..

ay

-^

ay

cos0 dv

CHAP.

PARTIAL DERIVATIVES

6]

127

Hence

dx\axj

3x'

^ /lo ^

"^

dp\ax/di

^^

d4,\ax/di

SV\dp

sin

sirn6

aF\a0

d<j>JSx

ap

sin

which simplifies to
^'^

Pn^^g!Z

dx'

and

(7)

p''

^-

sinv

2 si"

^p'

we

()

cos

3(a;,

a;

= /(%,?;) and

2/

-a^y
-

2 sin

_|_

d<f,

= g{u,v), where m =

sm0

fXsmVg ay

d'v
apa0

cos
o
p

^"^

find, as required,

dV

ap a0

ay

cos
p'

cos^i

Stf,

2sin0cos0 d'V

^0

dx'

45. (a) If

<b

dp

sin0cos0 ay

^p^

Similarly,

Adding

ay
+
op

""

gin^A gn^

ap

cos^0

ay

ap

"^

d^v

lay

dy

ap^

p dp

p^

+
1

,Kr,s),

a^^

cos^ a^y
O V
p^

p2 30^

prove that ^^^'^^

d{r, s)

y) d(u, v)

d(u,v) d{r,s)
(b)
(0)

(a)

di^>y)diu,v)

ProvP that
Frove

^^^ ^^^-^

_
_

XvVr

XuUs

+ yvVr

y,cUs

+ yvVs

XuUr+

d{x, y)
S{r, s)

y,Ur

y-

V'

Xu

d(x,y)

^
^^
+

provided

Xr,

Mr

Ms

J/

Vr

Vs

XvV,

a(a:,

2/

and interpret geometrically.

0,

y) d(u, v)

d{u,v) d(r,s)
I

using a theorem on multiplication of determinants


(see Problem 115).
o course, the existence of the partial derivatives
involved.
(6)

= x,s = y

Place r

in the result of (o).

Then

^^^' ^^ ^^^' '")


a(M, v) d{x, y)

^(^>

i/)

We
=

have assumed here

a(x, y)

The equations x = f{u,v), y = g{u,v) defines a transformation


between points (x,y)
and pomts (m,^) in the uv plane. The inverse transformation
is given by m =
^(x,y)
The result obtained states that the Jacobians of these

a;j/_plane

V-

46.

Show

Let u
<^)

that F{xy,z~2x)

y{dz/dy)

What

2x.

= xy,

satisfies

under suitable conditions the equation x(dz/dx)-

are these conditions?

= z- 2x. Then

F{u, v)

dF = F.du + F.dv

and

= F.(xdy +

ydx)

+ F.(dz~2dx) =

dependent variable and x and y as independent variables,

TV.
Then'^^^l^'^J!:^
substituting

in the
4,(x y)

transformations are reciprocals


^

of each other.

(i),

we

we have

find

(yF^

Fr,z^

2)

dx

(xF.

F^zy) dy

Hence we have, since x and y are independent,


(2)

yF,

+ F.z.-2 =

(3)

xFu

F,Zy

dz

a^^

a^y

"*"

and v

(?)

""

a^y

ct>ir,s)

sin

z.dx
^ ^

z,dy
-r zyu.y.

[CHAP.

PARTIAL DERIVATIVES

128

Solve for F in (3) and substitute in (2). Then


dividing by Fv (supposed not equal to zero).
will certainly be valid if

The result
that Fr - 0.

xz.

obtain the required result

we

that F{u,v)

we assume

is

- yzy =

2x

upon

continuously differentiable and

Supplementary Problems
FUNCTIONS
47.

If

48

If

= x'-yz + 3xy,

g(x,y,z)

-1,

(a)

-3),

2x

W
,^.

g^g^Tpg).

(6)

-i,

(a) /(I,

find

11

ii.\

{a)

Ans.
49.

2^,

/(^,^)

Ans.

GRAPHS

and

^^'
(6)

^^''''

'i~

/(^

+ 2/,x2/).

+ 2y + xy

t-x'y-xy^

find

(a) f?(l,-2,2),

x'-x-2- yz' + z" + Zxy + Sy,

{b)

+ ^'

g{x + l,y-l,z'),

(6)
(c)

g(xy,xz,x + y).

(c)

xV - x'z - xyz + Sx'yz

are defined and


Give the domain of definition for which each of the following functions
indicate this

Ans.

x^
x^

(a)

and

domain graphically.

() /(^.J/)

real,

+ y^-l

+ y"-

\,

(^)

'

(b)

/(^'^)

+y >

=
0,

ln{x
(c)

+ y),

(c)

2x-y ^
x+

= sin->(^^^.

f(x,y)
,

't

50.

(a)
(6)

the domain of definition for which


Indicate this domain graphically.

What

Ans.
51.

(a)

is

+z ^

1,

x'

+ y^ + z^ <

and x

f{x,y,z)

+y+z ^

^ J^^.^^^2 _

1,

x^

+ y' + z^ >

defined and real?

1.

the following,
Sketch and name the surface in 3 dimensional space represented by each of
(a)
(6)
(c)

+ 2z = 12,
42 = x^ + y\
z = x'- Ay\

(d)

3x

(e)
if)

+ z' = y\
x' + y' + z' = 16,
x' - 4y' - 4.z' = 36,
x'

iff)

+ y' = 2y,
= x + y,
y' = 4:z,
+ y^ + z^-'ix +
x^

(h) z
(i)

(i)

x''

hyperbolic paraboloid,
Ans. (a) plane, (6) paraboloid of revolution,
(g) right circular cylinder,
(e) sphere,
(/) hyperboloid of two sheets,
cylinder, (j) sphere, center at (2, -3, -1) and radius 4.
(c)

52.

Construct a graph of the region bounded by x'

53.

Describe graphically the set of points


(a)

54.

is

x^

y^

z''

1,

x^

y"-

z^;

(b)

(x, y, z)

x^

+ y' =

a'

and

x'

+ z'

=^

Gy

+ 2z-2 =

(d)

{h)

0.

right circular cone,


plane, (t) parabolic

a\ where a

is

a constant.

such that:

+ y" <

<

+ y.

The level curves for a function z = f{x, y) are curves in the xy plane defined by
They provide a way of representing the function graphically.
is any constant.

f{x, y)

c,

where

Similarly, the level


system defined by
coordinate
(xyz)
rectangular
in
a
=
surfaces
are the
surfaces of w
f{x, y, z)
graph
the level curves and surfaces for each
and
Describe
=
constant.
any
where
c
is
c,
z)
fix y
= 4xy, (c) f(x,y) = tan-' y/(x + l),
of the following functions: (a) f(x,y) = \n(x^ + y^-l), (b) f(x,y)
- y).
(/) sin ix + z)/il
(d)
y) = x^i^ + y^'\ ie) fix, y, z) = x^ + 4y' + 16^^
fix,

CHAP.

PARTIAL DERIVATIVES

6]

129

LIMITS and CONTINUITY


Prove that

55.

lim (3^-2y)

(.)

and

14

(6)

Jim^^^
^^
(x,yy^
(2, 1)

y-+ -1

56.

If

lim/OK,j/)

(a)

hm{fix,y)

Evaluate each of the following limits where they


^''^

lim^^

(6)

(^)

rd)

lim ^-5ilL(2il!l

+ x~2y

il"? 4

-*

As.
59.

^' S'"

does not exist,

(6)

Formulate a

/A

1-

2a;

-J/

sin"' toy

-2tan

''

(c)

8\/2,

(d) 0,

(e)

0,

(/)

(a) 3,

(6)

does not exist,

Investigate the continuity of each of the following


functions at the indicated points:

62.

() -^

+ y^;

Ans.

(a)

(xo, yo).

continuous,

Using the

definition,

^^'

oiT^tts^'g

^^

(6)

(6)

^) -. (0, 0, 0)

(O, 0).

exist?

(c) (x^

discontinuous,

prove that f(x,y)

?"'''"" ^ ^'"'"' ^^

Justify your answer.

sin

if

(X, y)

if (.,

(0, 0),

?/)

(0, 0); (0, 0).

continuous

(c)

y^)

6)

n variables.

61.

(a;, j/,

-2^

'(3a;j/

(h) 1/3

(g) 0,

Does

as

^ + ^"1

j;^

(^)

60.

lim

prove that:
P

lim e-'^<.->^

()

definition of limit for functions of

2^-^^

{x,y) -^ {xo,yo),
>!">

definition.

exist.

s_.o

(a) 4,

as

by using the

the limit of the quotient of two functions equal


to the quotient of their imii>-o.
limits'

is

Prove your answer.


58.

- 3. + 4) =

A and lim gix,y) = B, where lim denotes limit


+ g{x,y)} = A+B, (6) lim {f{x,y) ff{x,y)} = AB.

Under what conditions

57.

(xy

xy

+ 6x

is

continuous at

uniformly continuous

*'

(a) (1,2),

in

(6)

(x,,y,).

the square region defined by

PARTIAL DERIVATIVES
64.

If fix,y)

1^,

by differentiation

65.

If

Ans.

rules.

(^'

f{x,y)

find (a) df/3x

- 2/)/(a' +

2/)

|0
Aws.
66.

(a)

1,

Jim^^^ f4x,y)

,^

-2,

for {x,y)
for (x, 2/)

3f/Sy at (2,-1)
(6)

from the

definition

and verify your answer

-4

(0,0)

(0,0)'

..,,,,,
(a) /, (0,0),

,,,
(6)

find

/, (0,0).

for the function in the preceding problem


and explain

why

this limit

exists) is or is not equal to /x(0,


0).

Ans. (a)l(.

V3),

^^'

2z

(a)

(6)

(6)

Investigate
(if it

and

ff''rr?L'^/n
m*
If (x,y)
(0,0).

(6)

i(2.

3^3).

(c)|(.^-2),

(d)

|(,V3

3),

(e)

1(2.^3

1),

(/)

^^fff'-^""^""" that . = xy t^n(y/x) satisfies the equation x{ez/Sx)


(6) Discuss part (a) for all other points (x,y)
assuming ^ = at

1)

yOz/8y)

1(2.^/3

(0,0)

69.

Verify that /

mdicatmg
70.

Show

that

= /,. for the functions (a) (2x - 2/)/(x + 2/), (6)


possible exceptional points and investigate
these points.
.

ln{(a5-a)^

(2,-6)^}

satisfies

aVfla;^

aVSj/^

^t^^xy

and

except at

(c)

cosh
osn

(a, 6).

(2/
^2/

cos x),
a;)

71.

[CHAP.

PARTIAL DERIVATIVES

130

Show
a;

that z

x cos

(ylx)

tan {ylx)

x^z^x

satisfies

2a;j/2x

J/^Zss

except at points for which

= 0.
(a;
a;

+ z"N"
+ 2/
J/
2/

Indicate possible exceptional points.

DIFFERENTIALS
73.

= x^-xy\- 3y\

If z

why Az and
Ans.
74.

-11.658,

(a)

Compute

compute

and

Az

(a)

dz are approximately equal,


(6)

^^(3.8)'

-12.3,

Az

(c)

(c)

= -66,

= 5, = 4, Aa; = -0.2, Ay - 0.1.


x = 5, = 4, Aa; = -2, Aj/ = 1.

a;

Explain

3/

if

j/

= -123.

dz

using differentials.

approximately,

2(2.1)'

(6) dz where
Find Az and dz

Ans. 2.01
75.

76.

Find dF and dG
xy' In

(j//a;).

Arts,

(a)

(d)

77.

- 'iy^) dx +

{Sx'y

V-

(6)

(81/

(c)

{y^ In

Prove that

F(x,y)

(o)

if

(dy)/(l

(6xy

y^)

dx

DIFFERENTIATION

of

U{x,y,z)

(a) If
(b)

H(a;,2/)

If

Ans.

>

(a)

If

F(a;,2/)

(d) d^F/dv^,

80.

If

Z7

81.

If

a;

12t

where m

that

if

a;

V sin a and y

p cos ^,

dy

bx

Use Problem 82

to

2/

Bu

8v

IT-

dy

if so, find

the function.

= *'-* + !.
= t' + 3t,
,

+
4.,

3^u

aa;"

32/"

Ans.

?)

cos

(dVldyf

36"',

dC//dt at

find

= 0.

find dH/dt.

find

a,

aF/aw,

(o)

-14,

(a) 7, (b)

(c)

x(dU/ax)

(b)

aF/av,

14, (d) 112, (e)

+ y(dU/dy) =

where a

is

a constant, show that

{dV/dur

{BV/dvy

(c)

-49
217.

the equations

dv
^rdx

= u + 2v,

aw

become

show that under the transformation

av

+ T-i

3a;2/

Jcos{Sx-y)

sin a

P sin

42/'

suitable restrictions on F,

show that under

du
^

<;-2/^

2t',

= 2u-Zv,
= 2, v = 1.

(dV/dxr

Show

2/

a;z'

+ 9a;"-6tM-6xH + 18\

+ y)/{y -2x),

(c)

e^vTl

d^F/du dv,

cos a

not exact,

+ 21/ =

a;'

x^F(y/x),

= M

F(a;,2/)

dz

a;

sin(3x-2/),

(2x

(e)

3a;z^

COMPOSITE FUNCTIONS
= 2 sin t,
2a;" - 2/z + a;z^

/36t=2/
^

/l^

(b)

24,

(c)

(2xe

83.

- Sx^yz,

= UdV+VdU, (b) d{U/V) = {V dU - U dV)/V\ (c) d(ln JT) = (dU)/U,


+ V^) where U and V are diflferentiable functions of two or more variables.

- x^) dy
3a;) dy +
31/) dx + (122/^
(e) (z'
Ans. (a) x'y'^ + y sin 3a; + c, (b)

82.

Sxy^z"

(16a;3/z'

79.

G(x,y,z)

(b)

Determine whether each of the following are exact differentials of a function and
(a) (2x2/" + 3y cos 3a;) da; + (2x^y + sin 3a;) dy
(b)

78.

- ^xy^ + Sy\

- 8xy + 24y') dy
- 3a;2) di/ + (2'^xyV - ^x^y) dz
+
- 1/^} da; + {2xy In (^//a;) + xy} dy

d(UV)

(a)

d(tan-' 7)

x^y

(x^

6a;2/z) da;

(2//a;)

becomes

-r-

dp
a;

a"M

p cos ^,

+
xT
^P
,

av

1 a-y

- JT'
p d,p

aw

" Z

P P

2/

"^

T~
dp

P sin 0,

1 a^w
Ta 5:12"

_
"

du

XT
d<j>

the equation
.

"

d'FlSu',

^H-^P-

PARTIAL DERIVATIVES

6]

131

IMPLICIT FUNCTIONS and JACOBIANS

Find

(a)

Ans.

(a) (y

85.

86.

87.

F(x, v)

If

84.

xu^

If

If

dy/dx and

y\ 2yu

- ~ FJFy.

dy/dx

d^y/dx^

(6)

~ xy(y^ - x),

+v =

prove that

0,

x^

if

-2xyl{y^

(h)

- xv'' =

f(x,y),v^gix,y) are

4x,

+ y^-Zxy =
- xY

find

(o)

^
^

(6)

0.

- ^'''uV + 4
Gx^uv^ + 2y

^'

Ans

(a)

32/

'

g + gg

differentiable. prove that

2a;M''

^'

'

+ Sy^
+y

Zx^uv^

Explain clearly which

1.

variables are considered independent in each


partial derivative.
88.

If

f(x,y,r,s)

What

independent.

89.

If F(x, y)

0,

,,,ix,y,r,s)

gg + g|

prove that

,,

show that

^x.Fg

2F.,F.Fy
p3

dx'

90.

9L
^^'

^^'

94.

^^^

Evaluate

U F=

= Zu^-uv,

F(u,v)

2x^yz,

" =,.''^""^ + i'^"2' /'Id V


i!i
relationship between u and i;, and

(a) If

=.
"^

5(m,i;,w)
If

+ ^^ + ^'

fix, y,z)

and

G(u,v)

,;.^
^''^

"

if

so find

and

^'^^

^"1

y, z)

0,

Ans.

(a)

+ ixyz

at (1,-1,0).

determine whether there

+ .,

determine whether there


Ans. H''~G-2F =

2/

prove that

^^,1;,^),

V,

(a) in

+x -

dy

dz

d{f,g)

d(f,g)

d{z, x)

d(x, y)

__

is

Ana. 10
a functional

^('''^'^) d{u,v,w)
S{u,v,w) d{x.y,w)

is

a functional

^ ^Provmed
_-^;hh

terms of transformations.

valid.

is

w, tmd

{a)

~,

(b)

z as twice differentiable functions of


u,
(6)

- Zv^

show that

S{y, z)

'^

l&uv^

it.

mterpretation of the result of

giving conditions under which the result

y-rz

^gj^

^(f'9)

y and

explaining which variables are

Ans. 24u^v

evaluate

g{u,v,w),

^.

+ v\

2uv^

= 2z^-xy,

=/'

and g{x,

0,

+ FyyFl

= xVT^^+j/VT^^,

dx

define x,

+ ?'' + ^^ and
= 0^ +
connecting F, G, and H. and if so find it.

x=f(u,v,w),y =

^(=^.2/.^)

95.

if

+ Zy^-~.^G =

=^^.2'
I!l
relationship

notation could you use to indicate the independent


variables considered?

- iyz - 32a;'z'
(1 + i^y^Y

16a;'j/

(c)

^^^

assuming that the equations

v and w.

ISj/'z
<

~,

(")

- &xz - Z2x^y^
+ Sxyzr

(1

TRANSFORMATIONS. CURVILINEAR COORDINATES


transformation x = 2u + v, y = u-Zv.
of the .. plane bounded by . =

^^'

^i'^Ti^^
which
the region

^"''"*'

iTTtr-T^'^
99.

SS-

^''^

(a)

0,

^""^P""-" *^ "^^^"'^

Sketch the region

of the uv plane into

= 1, ^ . 0, ^ = l^s mapped undTr the trans!


^'t*> *e ratios of the areas of
*
% and ^.

(''^

Prove that under a linear transformation x = a.u+


a,v, y ^ 6,,.+ 6.t, (a,6.-a.6> ^0) lines
circles m the
ane are mapped respectively into lines and circles in
p
the uv plane. (6) compute
Compute
the Jacobian J of ^^
the transformation and discuss the significance
of J=
C'^)

and

[CHAP.

PARTIAL DERIVATIVES

^32

\<)(x,y)\

\d(u,v)\

Ans.

sin^Mcosh^o'

(b)

cos^Msinh^f,

(sm^ucosh^i>

(c)

m sinh

cos

^)

^u

region
- - 2 +
o _L o - .
TO
z = 2m - 2v + w. (a) Sketch the
w, .,
+w
y -- u Zv
Given the transformation x = 2u + Sv
^
x
x
by
8,
0,
j/
bounded
0,1/
4,
of the a.?/, space
%' of the uvw space into which the region
with the ratios of the volumes
(c) Compare the result of (6)
,

101.

l'o4^^0,

=6

of

102.

is

mapped.

and

Compute

(6)

Ares.

"iR'.

^J^.

(6)

x = rsincos.,
Given the spherical coordinate transformation
surfaces (a) r
coordinate
the
rgO OSfiS^ OS0<27r. Describe

- b, and
c
^-T
- a, rt
(6) 9 -^kndTcIT^

Ares,

a,b,c are any constants.


103. (a)
(b)

spheres,

(a)

{&) cones,

wllere
wnere

c,

(c)

planes

(c)

transformation of Problem 102,


Verify that for the spherical coordinate
0.
=
J
where
case
the
Discuss

_
-

9(a!, j/, z)

Qi^r,e,4>)

_
"

^ ^"^

"

MEAN VALUE THEOREMS


104.

Prove that

105.

Expand
Ans.

106.

In

/(x, .)

.Vx^

l-3(x-l)-2(2/ +
10[1

Prove the

108. Generalize

first

e{y

powers of x

in

l)

1)Y

+ 1, up

mean value theorem

"^

'

"

and including second degree terms and

to

+ 6(x-l)^ + 6(x-l)(l/ + l) + (l/ + ir


+ 12[1 - e{y + 1)] [1 + ejx - 1)] + 3[1 +
[1 + e{x - 1)1

e{x

- 1)Y

^^^^^^

F{P,V,T)^0, prove

(a)

gy|^ Jyl^

= ~ wl'

^^^

P,V,T

thermodynamics, where

in

P\

^^

^^

that

^\

dTW dVi

^J^
dPlr

1.

= -1

correspond to pressure, volume and tem-

perature of a physical system.


that F{x/y,z/y)

110.

Show

111.

Show that

satisfies

f{u,v)

113. If

f{u,v), y

and y

^ Au,v,w),

g{u,v),

g(u,v), z

= Hu,v,^)
d{x,y)

W^

_
~

x(dz/dx)

nx + y-z,x^ + y') =

112. If

<e<

for functions of 3 variables.


and prove Taylor's theorem of the mean

These results are useful

114. If

for functions of 3 variables.

MISCELLANEOUS PROBLEMS
109. If

and .

"^ "'

up to and including second degree terms.

/(x,.)

write the remainder.

107.

"'^^^ ^

= sin x in powers of x - 1 and y-^.,


- i^'ix - ly - iAx - 1)(2/ - i'^) - y - ^'

Expand
Ans.

'^'^''

^TTTli^W^y

satisfies

prove that

dv

x(Bz/dy)

- -j

and F{x,y,z)

d(u,v)

"^

and u

d(u,v)

/(r,s),

8ix,y)S(u,v)

3(^)3(r,s)

z.

0,

y.

S(x, v)

eiu.v)'

prove that
3(u,v)

"

T
^nere J

g^

y(Sz/ax)

1 dy

__

h(u,v)

y(dz/dy)

g(r,s).

w=

d(x^ djj^^
a{v,w)

a(,r,s)

h(r,s).

prove that

^M ^^

d(w,u) a{r,s)

^^^^-

PARTIAL DKRIVATIVES

^J

Prove that

115. (a)

116. If ..

ae

ce

. and . are functions of ., . and

+ bg
+ dg

d(.r, s, t)

"'
c^'ondL*;:

(6)

Illustrate

r:sp;ct to
120.

homogeneous of degree

by using the

X.]

B(u, V,

special case

2,

prove that

F{x,y)

{x,D.,

Let x and 2/ be determined from u and


transformation the equation

5^ +
(a)

,.

w)

a;^

establishing

are functions of

true

., .

the

and

*,

rule

for

the

prove that

'

d(r, s,

,-

t)

1,2,...,..

^+

2xw

Prove that under suitable

^^

i!^

9p

x^ ln{y/x).

Tf

pr-,-

'^'

\
'

'

"'''^ ''

Problem

(.),

25, twice'

with

homogeneous of degree

p,

then

+ X2D.,+ ...+xD.J'-F = p(p-l)...(p-r+l)F

121. (a)

the result in

thus

+ dh

D:fferent.ate both sides of equation

Generalize the result of Problem 119 as follow.?


for any positive integer r, if D,, = a/a^

(6) Is

cf

^.(-,...,^......,.,)=0 where

oT^"^

119. (a) If F(a;,2/) is

af+bh

while ., . and

,.

^33

^=
if

+ iy =

is

1;

according
to
x + iy
e,
^^ ^-riy

transformed into

F(u +

iv)l

(u
{u

iv)'
+ iv)

Prove your stItemenL.

th., under
.h
bioProve that
this

Chapter 7
Vectors
VECTORS

and

SCALARS

There are quantities in physics characterized by both


magnitude and direction, such as displacement, velocity,
To describe such quantities, we
force and acceleration.
introduce the concept of a vector as a directed line segment

PQ from
Q called

one point P called the


the terminal point.

another point
denote vectors by bold

initial point to

We

faced letters or letters with an arrow over them. Thus PQ


is denoted by A or A as in Fig. 7-1.
The magnitude or
length of the vector

is

then denoted by \PQ\, PQ,

|A|

or \A\.

Fig. 7-1

Other quantities in physics are characterized by magnitude only, such as mass,


length and temperature.
Such quantities are often called scalars to distinguish them
from vectors, but it must be emphasized that apart from units such as feet, degrees, etc.,
they are nothing more than real numbers. We can thus denote them by ordinary letters
as usual.

VECTOR ALGEBRA
The operations

of addition, subtraction

tion familiar in the algebra of

numbers

definition, capable of extension to

The following
1.

Two

and multiplicawith suitable

are,

an algebra of vectors.

definitions are fundamental.

vectors

and

are eqtial

if

they have the

same magnitude and direction regardless of


initial points.
Thus A = B in Fig. 7-1 above.
2.

A
A

their

vector having direction opposite to that of vector


but with the same magnitude is denoted by A

Fig. 7-2

[see Fig. 7-2].

The sum or resultant of vectors A and B of Fig. 7-3(a) below is a vector C formed
by placing the initial point of B on the terminal point of A and joining the
initial point of A to the terminal point of B [see Fig. 7-3(6) below].
The sum C
is written C = A + B.
The definition here is equivalent to the parallelogram law
for vector addition as indicated in Fig. 7-3(c) below.

C =

A+B

Fig. 7-3

134

CHAP.

VECTORS

7]

135

Extensions to sums of more than two vectors


are immediate. For example
^^^ ^w to obtain the sum or resultant E of the
vectors A,

J'tj'^^''^

ID

T^Tri+^+^
Fig. 7-4

^"^ ^' represented by A-B, is that vector C which


Equivalently,
- B may be defined as + (-B) If
defined as the null or zero vector and is
represented by the symbol o!
TV^ has a magnitude of zero but its
This
direction is not defined.
''''*'"'

!?! ff'T"'
added
to B gives A.

^"

5.

Multiplication of a vector A by a scalar


produces a vector
with magnitude |m| times the magnitude of A and direction
the same as or opposite to that
of A according as
is positive or negative.
= 0,
= 0, the null vector
If

mA

LAWS

of

If A,
1.

2.

mA

VECTOR ALGEBRA
B

and

are vectors, and

A + B=:B + A
A + (B + C) = (A + B) + C

4.

m(nA) = (mn)A = n{mA)


(m + n)\ = mA + nA

5.

m(A + B) =

3.

I=b

wA + mB

m and n are scalars,

then

Commutative Law for Addition

Law for
Law for
Distributive Law
Distributive Law
Associative

Addition

Associative

Multiplication

'"
^^^^ "^y multiplication of a vector
^''V^tJ
defined.
On Pages ^^tT
136 and 137 we define products of vectors.

by one or more scalars

is

UNIT VECTORS
thPv,
then

^^^1

AM

^^eetors are vectors


is

having unit length. If


is any vector with length
a unit vector, denoted by a, having the
same direction as A. Then A

RECTANGULAR UNIT VECTORS


The rectangular unit vectors i, j and k are unit
vectors having the direction of the
positive x, y and z axes
ot a rectangular coordinate system
[see Fig. 7-5].
use

We

right-handed rectangular coordinate systems


unless otherwise specified. Such systems derive their
name from the
fact that a right threaded screw rotated
through 90 from
Ox to Oy will advance in the positive z direction. In
general,

/O

Fig. 7-5

A>0
= Aa!

[CHAP.

VECTORS

136

coplanar are said


three vectors A, B and C which have coincident initial points and are not
screw rotated through
to form a right-handed system or dextral system if a right threaded
an angle less than 180 from A to B will advance in the direction C [see Fig. 7-6 below].

(AijA^.A,)

Fig. 7-7

Fig. 7-6

COMPONENTS

of a

VECTOR

in 3 dimensions can be represented with initial point at the origin O


vector
rectangular
of a rectangular coordinate system [see Fig. 7-7 above]. Let (Ai, A2, A3) be the
vectors
Aii,
The
at
point
0.
initial
with
vector
A
point
of
coordinates of the terminal
vectors,
component
simply
or
vectors,
component
rectangular
A2J and Ask are called the
Ai, A2 and As are called the rectangular
in the x, y and z directions respectively.
of
the
x, y and z directions respectively.
in
of
components,
components, or simply

Any

The sum or resultant

of

Ad, A2J and Ask

A =
The magnitude

of

Aii

is

A23

the vector A, so that

+ Ask

{!)

lAl

V^fTAlTAf

{2)

In particular, the position vector or radius vector r from


written
^ a;i + 2/3 + 2k
^

and has magnitude r

or

|r|

\/x''

to the point {x,y,z) is


(5)

+ j/^ + z^.

SCALAR PRODUCT

dot B)
and B, denoted by A'B (read
scalar product of two vectors
angle
the
of
cosine
the
and
B
and
defined as the product of the magnitudes of

The dot or

is

write

is

A =

DOT

we can

In symbols,

between them.

A-B =
Note that

A B

is

3.
4.

i-i

j-j

Commutative Law for Dot Products


Distributive

(^)

valid:

A'B = BA (B + C) = A B + A C
m(A B) = (mA) B = A

2.

O^^Stt

a scalar and not a vector.

The following laws are


1.

ABcos6,

= k-k =

1,

Law

(mB) = (A B)m,
i-j = j-k ^ k-i =

where

is

a scalar.

CHAP.

VECTORS

7]

If

5.

A=

Aii

+ A2J + Ask and B =

A-B =
A- A
If

6.

CROSS

or

A B =

and

and

B2J

+ Bgk,

then

AiBi + A2B2 + A3B3


A2 = Af + Al + Al
B^ = 52 + g2 + 52

=
=

B-B

Bii

I37

are not null vectors, then

and

are perpendicular.

VECTOR PRODUCT

The cross or vector product of A and B is a vector C = A X B (read A cross B).


The magnitude of A x B is defined as the product of the magnitudes of A and B and the
sine of the angle betvs^een them. The direction of the vector C = A X B is perpendicular
to the plane of A and B and such that A, B and C form a right-handed system.
In
symbols,

where u

A X B = AS sin
is

to B, then

a unit vector indicating the direction of


sin

6*

and we

The following laws are

AxB

1.

6>

define

AXB =

A X B.

tt

If

(5)

A=B

3.
4.

if

is

parallel

valid:

-BxA

(Commutative

Law

for Cross Products Pails)

Ax(B + C) = AxB + AxC


Distributive Law
m(A X B) = (mA) x B = A x (mB) = (A x B)m,
where m
ixi = jxj = kxk = 0,
ixj = k, jxk =
kxi = j
If A = Aii + A2J + Ask and B = BA + Bij+Bik, then

2.

or

0.

is

a scalar.

i,

5.

AXB

Ai

A2

k
A3

Bi

B2

B3

= the area of a parallelogram with sides A and B.


A X B = and A and B are not null vectors, then A and B

6.

|A X B]

7.

If

are parallel.

TRIPLE PRODUCTS
Dot and cross multiplication of three vectors A, B and C may produce meaningful
products of the form (A-B)C, A- (B x C) and A x (B x C). The following laws are valid:
1.

(A -6)0

2.

A(B-C)

in general

(B X C) = B (C X A) = C (A X B) = volume of a parallelepiped having A, B,


and C as edges, or the negative of this volume according as A, B and C do or
do not form a right-handed system. If
= Aii + Aaj + Ask, B = Bii + B2 j + Bsk
and C = Cii + C2J + C3k, then

A
3.

4.

AX

(B X C)

Ai

A2

Bi
Ci

B2

As
Bs

C2

Ca

^ (A X B) X C
(Associative Law
(BXC) = (A-C)B - (A-B)C
(A X B) X C = (A- C)B - (B- C)A
X (B X C)

(S)

for Cross Products Fails)

The product A (B x C) is sometimes called the scalar triple product or box product
and may be denoted by [ABC]. The product A x (B x C) is called the vector triple

product.

VECTORS

138

[CHAP. 7

In A (B X C) parentheses are sometimes omitted and we write A B X C. However,


parentheses must be used in A x (B x C) (see Problem 29).
Note that A (B x C) =
(A X B) C. This is often expressed by stating that in a scalar triple product the dot and
the cross can be interchanged without affecting the result (see Problem 26).

AXIOMATIC APPROACH

to

From the above remarks


when its 3 components {x,y,z)
ing an axiomatic approach

Definition.

With
Thus

VECTOR ANALYSIS
seen that a vector r = xi + y] + zk is determined
some coordinate system are known. In adoptthus quite natural for us to make the following

it

is

relative to

it is

3 dimensional vector

this as starting point

we can

is

an ordered

2.
3.

A-B

4.

and subtraction,

etc.

= (Ai-Bi, A2-B2, A3-B3)

6.
7.

Length or magnitude of

A =

= VAvA =

|A|

From these we obtain other properties of


(A + B) + C, A-(B + C) = A-B + A-C,
i

we can then show

(1,0.0),

vectors, such as
etc.

By

(0,1,0),

+ A| + A|
A + B = B + A,

\/^f

A+

(B

+ C) =

defining the unit vectors

k =

(0,0,1)

(7)

that

A In like

(Ai, A2, A3).

(0,0,0)

toA = m(Ai,A2,A3) = {mAi,mA2,mAa)


A-B = AiBi + A2B2 + A3B3

5.

numbers

define equality, vector addition

A = (Ai, A2, A3) and B = (Bi, B2, Ba), we define


A = B if and only if Ai = Bi, A2 Bi, A3 = Bs
A + B = (Ai + Bi, A2 + B2, A3 + B3)

if
1.

manner we can

define

AxB

All

A2J

+ A3k

(S)

= {A2B3-ASB2, A3B1-A1B3, A1B2-A2B1).

After this axiomatic approach has been developed we can interpret the results geoFor example, we can show that A B = AB cos 6, |A X B] =

metrically or physically.
sin 0, etc.

AB

triplet of real

In the above we have considered three dimensional vectors. It is easy to extend the
idea of a vector to higher dimensions. For example, a four dimensional vector is defined
as an ordered quadruple (Ai, A2, A3, A4).

VECTOR FUNCTIONS
If corresponding to

each value of a scalar u we associate a vector A, then


In three dimensions we can write A{u)

a function of u denoted by A{u).


A2(m)j

is

Ai{u)i

called

+ A3{u)k.

The function concept is easily extended. Thus if to each point (x,y,z) there corresponds a vector A, then A is a function of {x,y,z), indicated by A{x,y,z) Ai{x,y,z)\ +
A2 {x, y, 2:) j

We

+ A3

(a;,

y, z)k.

sometimes say that a vector function A.{x,y,z) defines a vector field since it
associates a vector with each point of a region. Similarly 4>{x,y,z) defines a scalar field
since it associates a scalar with each point of a region.

CHAP.

VECTORS

7]

139

LIMITS, CONTINUITY and DERIVATIVES

of VECTOR FUNCTIONS
Limits, continuity and derivatives of vector functions
follow rules similar to those
for scalar functions already considered.
The following statements show the analogy

which

exists.

The vector function A{u) is said to be continuous at Uo if given any


number , we can find some positive number 8 such that |A(m)-A(mo)| <

1.

ever

The

2.

\u-ua\

<

This

S.

derivative of A(tt)

is

is

equivalent to the statement

lim A{u)

positive
e

when-

A(mo).

defined as

dA
^
du

provided this limit exists.

+ A^<)~A(^)

A(^<

lijn
Au-0

In case

A{u)

{u)i

+ Aa {u)j + As {u)k;

then

dAs,

du
du^
Higher derivatives such as d^A/du^,
3. If

is

A(a;, y^z)

du^
etc.,

'^

'W^

can be similarly defined.

= Ai {x, y, z)i + A2 (x, y, z)j + A3 {x, y, z)k,

then

the differential of A.

Derivatives of products obey rules similar to those for scalar functions.


However,
when cross products are involved the order may be important. Some examples

4.

are:

|^(M)

^(A-B)

(c)

|(AXB) =

(a)

GEOMETRIC INTERPRETATION

'^

of a

du

du

dy
,

^ aB

'

dy
,

aA

VECTOR DERIVATIVE

If r is the vector joining the origin

of a
coordinate system and the point {x,y,z), then
specification of the vector function r{u) defines
X, y and z as functions of u. As u changes, the
terminal point of r describes a space curve
(see

Fig.
x{u),

7-8)

hailing

y{u), z

parametric equations

If the parameter u
the arc length s measured from some fixed
point on the curve, then
z{u).

At

is

ds

(9)

is a unit vector in the direction of the tangent


to the curve and is called the unit tangent
vector. If u is the time t, then

dr

is

^
(io)
di =
the velocity with which the terminal point

Fig. 7-8

r(M

+ Am)

r(M)

[CHAP.

VECTORS

140

We

of r describes the curve.

from which we

have

_
~

_
~

dr
di

dtds^
ds dt

magnitude of v

see that the

ill)

dt

dt^

is

Similarly,

ds/dt.

'

These con-

point of r describes the curve.


is the acceleration with which the terminal
and differential geometry.
mechanics
cepts have important applications in

GRADIENT, DIVERGENCE and CURL

Consider the vector operator

V
Then

if

.^(ic,

y. z)

which

dition

1.

is

many

z)

Gradient. The gradient of

grad ^

(13)

'd^ + 's^ + ^d^

have continuous first partial derivatives in a region (a concases stronger than necessary), we can define the following.

and A{x, y,
in

by

defined

(del)

'^^^dy +

'te+^a^ + '^S^f "

V*^

defined by

is

<j>

dz

dy

dx

{U)

^dz

of a
interesting interpretation is that if ^{x, y,z) = c is the equation
Problem
36).
(see
surface, then V^ is a normal to this surface

An

2.

Divergence. The divergence of

VA

divA =

[^Tx

is

Ml
dx
3.

Curl.

The

curl

curl of

A =

is

defined

A =

'^

defined by

-^

(All

^Jz

'S^J

dAi

dAa

dy

dz

A2J

+ Ask)

{15)

by

i^ + ^4 +
i

dy

{16)

^^'^^

A
A
dx dy
Ai

*^^)"^^^^ + ^^^ +

1dz

Ai As
d

dz

dx

_5_

dz

dx

dy

+ k

A, A2

A, As

Ms

dA

dy

dz

dAi
i+{^-^n
dz

Ai A2

Ma

Ml

5a;

dy

Note that in the expansion of the determinant, the operators d/dx,d/dy,d/dz must
precede Ai, A2, A3.


CHAP.

VECTORS

7]

141

FORMULAS INVOLVING V
If the partial derivatives of A, B,
1.

2.
3.

4.
5.
6.
7.

8.
9.

11.
12.

and

are assumed to exist, then

+ F) = vf/ + VF
or
grad(f/+F) = grad 2* + grad F
V(A + B) =
+
or
div(A + B) = div A + div B
VX(A + B) = VXA + VXB
or
curl(A + B) = curl A + curl B
V{UA) = (VC/)-A + f/{vA)
V X (C/A) = (Vf/)XA + Z7(VXA)
V(AXB) = B-(VXA) - A-(vxB)
VX(AXB) = (B-V)A - B(VA) - (A-V)B + A(VB)
V(A-B) ^ (B-V)A + (A-V)B + BX (VXA) + Ax (VXB)
V(C/

VA

V.(vf7)

V
V

VB

v^t/

V2

and
10.

^ + ^,
32

52

is

called the Laplacian of

52
.

a^ ~^ 5^2

5a;2

called the Laplacian operator.

is

X (vt/)

= 0. The curl of the gradient of V is zero.


= 0. The divergence of the curl of A is zero.
Vx(vxA) = V(VA) - V'A

(V X A)

VECTOR INTERPRETATION of JACOBIANS.


ORTHOGONAL CURVILINEAR COORDINATES.
The transformation equations

f{ui,U2,U3),

g{ui,U2,U3),

h{Ui,Ui,us)

(17)

[where we assume that f,g,h are continuous, have


continuous partial derivatives and
have a smgle-valued inverse] establish a one to one
correspondence between points in an
xyz and ^1^*2^3 rectangular coordinate system. In vector
notation the transformation (17)
can be written
r

xi

y]

zk

point P in Fig. 7-9 can then


by rectangular coordinates {x,
nates iui,U2,U3) as well.
We
curvilinear coordinates of the

f{uuU2,uz)i

g{ui,U2,Us)j

h{Mi,U2,U3)k

(18)

be defined not only


y, z) but by coordicall

{ui.m.Ua) the

point.

and U3 are constant, then as Ui varies,


which we call the Ui coordinate
curve. Similarly we define the U2 and Uz coordinate curves through P.
If U2

r describes a curve

From
dr

{18),

we have

- dui
aUi

dU2

dU2

Fig. 7-9

(19)

dUa
dt/du^ is tangent to the u^ coordinate curve
at P.
direction, we can write dr/du,^h^^, where
=

The vector

this

write
tively.

dT/dU2

Then

and dT/du^^hzez,
can be written

h2e2

(19)

dv

The

hiduiei

where

h2

If ei is a unit vector at

h^

\dTlduA.

\3r/dU2\

and

h,

Similarly

\dr/du,\
\

hzduzez

hsduaca

quantities hi,h2,hz are sometimes called scale factors.

we can
respecy

(20)

VECTORS

142

If

[CHAP.

are mutually perpendicular at any point P, the curvilinear coordinates are


In such case the element of arc length ds is given by

ei, 62, es

called orthogonal.

and corresponds

dx-dx

h\du\

h^dul

h^dul

{21)

to the square of the length of the diagonal in the

above parallelepiped.

ds^

Also, in the case of orthogonal coordinates the

dV

{hi

dui

ei)

{h2 du2 62)

volume of the parallelepiped

{ha dua es)

is

given by

hi /12 ha dui du2 dua

{22)

which can be written as

dV
where

dx

dx

dx

dUi

dUi

dUa

d{x,y,z)/d{ui,U2,Ua)

d{x,y,z)

dUi dUi dUa

dui du2 dua

{23)

d{Ui, U2, Ua)

the Jacobian of the transformation.

is

It is clear that when the Jacobian vanishes there is no parallelepiped and explains
geometrically the significance of the vanishing of a Jacobian as treated in Chapter 6.

GRADIENT, DIVERGENCE, CURL and LAPLACIAN

in

ORTHOGONAL

CURVILINEAR COORDINATES
If * is a scalar function and
curvilinear coordinates Ui,U2,Ua,
1.

V*
V* A

div

A =

curl

ei
Jhi dUi

A =

+ A2e2 + Aaea a vector function of orthogonal


the following results.

Aiei

grad $

A =

we have

{hihaAi)

dUi

'

dU2

hiCi

^262

haea

dU2

dUa

hih2ha

dUi

Laplacian of *

V'*

hih2ha

{hahiAi)

{hihiAa)

h2A2 haAa

hiAi
4.

ha dUa

A =

es
'

hi dUz

T
hihiha

62

J-

dUi
a

a$

/ /12/13
\

hi

/hahi a*_

dU2 [

dUi

hi

dU2

f hih2 a*

dUa\^

ha

dUa

These reduce to the usual expressions in rectangular coordinates


by {x, y, z), in which case ei, 02 and es are replaced by i, j and k and

{ui, 2, Ua)

if

hi

we replace
= h2 = ha = 1.

SPECIAL CURVILINEAR COORDINATES


1.

Cylindrical Coordinates

{p, <^, z).

See Fig. 7-10.

P(x,v,z)

Transformation equations:
X

where

p cos

0,

Scale factors:

^ <

27r,

hi

1,

Element of arc length:


r

Jacobian:

d{x, y, z)
- f

-tt

d{p,

<j>,

<^,

z)

Element of volume:

psm<j>,

< .
h2 = p, /13 = 1
ds^ = dp^ + p^ d<j>^+
=0

<

2;

'^

dV =

pdpd<f>dz

Laplacian:

^ ^ -

~pdp[p'd^J

dz^

7a^

a^iaC7^^_^
'W

dp^

P dp

"^

p2

a^2

9^2

CHAP.

VECTORS

7]

143

Note that corresponding- results can be obtained for polar


coordinates in the
plane by omitting z dependence.
In such case for example, ds" = dp" + p^ d<j>^
while the element of volume is replaced by the element
of area, dA =
pdpd<i>.

2.

Spherical Coordinates

See Fig. 7-11.

(r, d, <f).

Transformation equations:

= rsind cos<j,, y = r sin sin


z = r cos 6
where r^O,
g g
g ^ < 2,r.
Scale factors: hi - 1, h^-r, ha = r sin e
X

<^,

ir,S,<!>)

,r,

Element of arc length:


ds^

dr^

'^^

JacoUan:

Element of volume:

r^sin^ed^^

r^^inS
Pig. 7-11

dV =

^^U =

Laplacian:

r^de^

dr dd

r^ sin 6

r^dr\

r^

dr J

d<i>

J-_ ^
r^smOdd

Other types of coordinate systems are possible

Solved Problems

VECTOR ALGEBRA
1.

Show that
and
Then

addition of vectors

is

commutative,

OP + PQ = OQ
OK + RQ = OQ

A+B = B+

i.e.

or
or

A + B = B + A.

See Fig. 7-12 below.

A + B = C,
B +A = C

A.

Fig. 7-12

2.

Show that the addition of vectors is associative,


Fig. 7-13 above.
OP + PQ = OQ ^
^"'^^

we have

(A

B)

OP + PR = OR =
OQ + QR = OR =

A+

(B

C)

vec JsTiiTaii!;'

(A + B)

'"""^

and
D,

D,

i.e.

-I-

(B

+ C) = (A + B)

PQ + QR = PR
i.e.
A + (B + C) = D
i.e.
(A + B) + C = D

(B

-f-

C.

See

+ C)

C.

"' ^^^'^^"^ ' ^"'^ ' ^''"^ ^'^*

'"^^ '-'^^'-

"' ^'^'''^^

'^^

-y -->-

of

VECTORS

144

[CHAP.

An

automobile travels 3 miles due north, then 5 miles northeast as shown in Fig. 7-14.
Represent these displacements graphically and determine the resultant displacement
(a)

graphically,
Vector

(&) analytically.

OP

or

PQ

or

represents displacement of 3 mi due

north.

Vector

represents displacement of 5 mi north-

east.

Vector OQ or C represents the resultant displacement


or sum of vectors A and B, i.e. C = A + B. This is the
triangle law of vector addition.
The resultant vector OQ can also be obtained by constructing the diagonal of the parallelogram OPQR having
vectors OP = A and OK (equal to vector PQ or B) as sides.
This is the parallelogram law of vector addition.
(a)

(6)

Graphical Determination of Resultant. Lay oflf the


1 mile unit on vector OQ to find the magnitude 7.4 mi
(approximately). Angle EOQ = 61.5, using a protractor. Then vector OQ has magnitude 7.4 mi and
direction 61.5 north of east.
Analytical Determination of Resultant. From triangle
OPQ, denoting the magnitudes of A, B, C by A, B, C, we
have by the law of cosines

C'

and

A'

C=
By

B'

- 2AB

the law of sines,

Thus vector

Prove that

5.

If

same

OQ

2(3)(5) cos 135

5^

sin

sin

A
lOQP

lOPQ =

+ yb

+ yih =

X2a

+ y2h,

34

C
sin Z

55.21

3(0.707)
.

Then

OPQ

ookk
0.2855

implies

v^^here a

xa

10QP =

a
and

(45

1635')

+ yh =

ABCD

Wz =

0,

2/1

Problem

6135' north of east.

implies

0.

or a = -(j//x)b, i.e. a and b must be parallel


Thus x = 0; then yh = 0, from whicA v = 0.

and b are

+ y2h can be written


xia + yih (Xia + yih) =
4,

1635'

xa = -j/b

non-collinear, then

3/2 =

or

{xi

or xi

X2)a +

xz,

2/1

{yi

a;i

y2)h =

2/2.

49).

Prove that the diagonals of a parallelogram bisect each other.


be the given parallelogram with diagonals

P as shown in Fig. 7-15.


Since BD + a = b, BD = b - a. Then BP =
Since AC = a + b, AP = y{a + b).
AB = AP + PB = AP - BP,
But
a = y(a + b) a;(b a) = (x + y)a + {y x)b.

intersecting at

i.e.

mile

I5V2

-I-

X2a

Extensions are possible (see

Let

Fig. 7-14

has magnitude 7.43 mi and direction

Then xa

Hence by Problem

6.

3"

line (collinear), contrary to hypothesis.

Xia+2/ib
Xia

lOPQ =

a and b are non-collinear, then

if

Suppose x=0.
to the

cos

7.43 (approximately).

A
,
sm lOQP =

4.

Unit

x(b

a).

Since a and b are non-collinear we have by Problem 5,


x + y = 1 and y x = 0, i.e. x y = ^ and P is the
midpoint of both diagonals.

= X2 and

1/1

= 1/2.

CHAP.

VECTORS

7]

Prove that the line joining the midpoints of


third side and has half its length.

7.

From
and CB.

sides of a triangle

AC + CB = AB or b + a = c.
be the line joining the midpoints of sides

parallel to the

is

Fig. 7-16,

DE = d

Let

two

145

Then

= DC + CE = lb +

Thus d

la

l(b

and has half

parallel to c

is

AC

length.

its

Fisr.

Prove that the magnitude

8.

AA + Aij+Ask
By
where

A of the vector A =
A = A/Af+Af+Af. See Fig.

is

_
OP

{OQY + (QPy

A,]i

= (ORf + (RQ)\
(OPy = (OBY + {RQy + (QPy

(OQ)'

AI

AI

A\,

or

A = VAH^4|TAf.

i.e.

Determine the vector having initial point P(xu


Vx, 2i)
and terminal point Q{x2,y2,Z2) and find its magni-

9.

tude.

P
Q

position vector of

position vector of

+ PQ =
PQ = r2 ri

=
=

xii
iC2i

+ yij + zik.
+ 2/2J + z^k.
Q(.Jt-s,

ri

= fei + 2/2J + Z2k) - (xii + j/ij + zjk)


= {X2 - X,)i + (2/2 - 2/i)j + (z, _ 2,)k_
PQ = PQ

- xi)2 +

V(a;2

this is the distance

or

is ra

or

=
Note that

is ri

r2

Magnitude of

DOT

i'{xi,yi,z,)

See Fig. 7-18.

The
The

The

<{

denotes the magnitude of vector OP, etc

A-

iOPr

Then

- y,)" + (z, - ztY.


P and Q.

(2/2

between points

Fig. 7-18

SCALAR PRODUCT
A on B

Prove that the projection of


vs^here

is

is equal to A-b,
a unit vector in the direction of B.

Through the
perpendicular to
Fig. 7-19; then
Projection of

11.

P(A...A,,A,)

the Pythagorean theorem,

Similarly,

10.

7-16

Prove

initial

at

and terminal points of

G and

B =

on

GH = EF = A

A-(B-fC) = A-B

-h

-h

(B

C) on

A =

= A-b

cos 9

Fig. 7-19

A-C.

Let a be a unit vector in the direction of


A;
Projection of (B

pass planes

respectively as in the adjacent

projection of B on
+ projection of

then

A
C on

+ C)-a = B-a + C-a

Multiplying by A,
(B-f-C)

and

(B

C)

Aa
-

= B-Aa

A = B A

-f-

-I-

C-Aa
C

Then by the commutative law for dot products,

A-(B + C) = A-B4-A-C
and the distributive law

is

valid.

G A
Fig. 7-20

Vt, 2)

[CHAP.

VECTORS

146

12.Provethat (A + B)-(C + D) = A- C + A-D + B- C + B-D.


By Problem 11, (A + B) (C + D) = A (C + D) + B (C + D) = A C + A D + B C + B

D.

0-0-0 =

of algebra are valid for dot products.

The ordinary laws

13.

Evaluate each of the following.

=
=
=

(a)

i-i

(b)

i-k

(c)

k-j

id)

i(2i-3j +

(e)

(2i-j)-(3i + k)

14. If

A =

|i|

= (1)(1)(1) = 1
= (1)(1)(0) =
= (1)(1)(0) =
cos 90
= 2j-i - 3j-j + j'k = 0-3 +
= 2i-(3i + k) - i-(3i + k) = 6i-i +

cosO

|i|

|i||k|cos90

HlJl

Aii

k)

+ Azj + Ask and B =

Bii

= -3
-3ji-jk =

2i-k

+ B23 + B3k,

prove that

A-B =

A1B1 + A2B2 +

A3B3.

A-B =

+ A2J + A3k) (Bii + Baj + Bsk)


Aii (Bii + B2 j + Bak) + A2 j (Bii + ^2 j + Bak) + Ask (BJ + B2 j + Bsk)
AiBii i + AiBA j + AiJSai k + A2B1 j i + A2B2 j 3 + A2B3J k
+ AsBik'i + AzBtk'i + AaBsk-k
= AiBi + A2B2 + A3B3
= jj=k*k = l and all other dot products are zero.
(Aii

=
=

since

15. If

i'i

Aii + Aaj + Ask, show that A = \/A-A =


A-A = (A)(A) cos 0 = A^ Then A = VA A.
Also, A'A = (A,i + A2J+A3k) (A,i + A2J+A3k)

A =

\/A\

+ A| + A

+ {A3){A,) = A\ + A\ + Al
B = A.
Then A = VAvA = VAf+Af+Af is the magnitude of A.
=

by Problem

Prove

(A2)(A2)

14, taking

The CROSS or
16.

{A,)(A,)

Sometimes

A A

is

written

A^

VECTOR PRODUCT

AXB

-BxA.

IS

(6)

(a)

Fig. 7-21

AXB = C
system

has magnitude

AB

sin e

and direction such that A,

and C form a right-handed

[Fig. 7-21 (a) above].

B X A = D has magnitude BA sin e and direction such that B, A and D form a right-handed
system [Fig. 7-21(6) above].
Then D has the same magnitude as C but is opposite in direction, i.e. C = D or A X B = B X A.
The commutative law for

cross products is not valid.

CHAP.

17.

VECTORS

7]

147

AxB AxC

Ax(B

Prove that
+ C) =^
+
for the case where
is perpendicular to

B and

also to C.

Since A is perpendicular to B, A X B is a
vector perpendicular to the plane of A and B
and having magnitude AB sin 90 = AB or
magnitude of AB. This is equivalent to multiplying vector B by A and rotating the resultant
vector through 90 to the position shown in
Fig. 7-22.
Similarly, A X C is the vector obtained by
multiplying C by A and rotating the resultant
vector through 90 to the position shown.

In like manner, A X (B + C) is the vector


obtained by multiplying B + C by A and rotating the resultant vector through 90 to the position shown.

Since A X (B + C) is the diagonal of the


parallelogram with A X B and A X C as sides,
we have
+ C) =

AX(B

18.

AXB-FAXC.

Prove that Ax(B-f-C)

AXB

+ AxC
B and C are

in the general case vs^here A,

non-coplanar.

Fig. 7-22

See Fig. 7-23.

Resolve B into two component vectors, one


perpendicular to A and the other parallel to A,
and denote them by Bi and Bn respectively. Then
B = Bx-HB||.

angle between A and B, then


Thus the magnitude of A X Bi is
AB sin e, the same as the magnitude of A X B.
Also, the direction of A X Bj. is the same as the
direction of A X B. Hence A X Bx = A X B.
Similarly if C is resolved into two component
vectors Cm and C^, parallel and perpendicular
respectively to A, then A X Cx = A X C.
If

B^

= B

is

sin

the

e.

Also, since

Now

B+ C =

B,

+ C, + d, = (B, + C,) + (B,, + C,,)


A X (B. + C) = A X (B + C)

B,,

Bx and Ci are vectors perpendicular to

AX

Fig. 7-23

and

so

by Problem

it

follows that

17,

+ Cx) = A X Bx + A X Cx
AX(B + C) = AXB + AXC

Then

(Bx

and the distributive law holds. Multiplying by -1, using Problem 16, this becomes (B -1- C) X A
A.
Note that the order of factors in cross products is important. ~'
The usual laws of

BXA + CX

"

algebra apply only

if

proper order

is

maintained.

19. If

A=

Aii-I-Aaj

AXB

=
=
=

"

+ Aak and B =

Bii

+ Bzj+Bah, provethat

AXB

+ Aak) X (Ba + B2J + B3k)


{BA + B,i + Bak) + A2 j X {BA + B2J + Bsk) + Aak X (J5ii + B2J + Bak)
AiBii X i + AiBzi X j + AiBA X k + A2B1J X i + A2B2J X j + A^Baj X k
+ AsBik X i + AaBak X j + AaBak X k
i
k
j
{A2B,-A,B2)i + (AzBr-AiBs)j + (A1B2 - Ai,B,)k =
Ai A2 A3
(Aii-I-Azi

AAX

Si B2 Bs

Ai A2 As
Hi B2 B3

20. If

A =

3i

- j + 2k and B =
i

2i

3-12

AXB

- k,

Sj

-1

5i

7j

Ax B.

find

-1

3-1

21.

[CHAP.

VECTORS

148

-1

+ k

-1

+ Ilk

Prove that the area of a parallelogram with


sides A and B is lAxBj.
Area

=
=
=

of parallelogram

|B|

|A|sin9|B|

lAXEJ.

Note that the area of the triangle with sides

and

B = llAXBl.
22.

Fig. 7-24

Q(4,2,-l),
Find the area of the triangle with vertices at P(2,3,5),
PQ = (4 - 2)i + (2 - 3)j + (-1 - 5)k = 2i - j - 6k
PR = (3 - 2)i + (6 - 3)i + (4 - 5)k = i + 3j - k
= 1 |(2i- j - 6k) X (i + 3j -k)
= 1 PQ X PR
Area of triangle

/2(3,6,4).

k
-6

-1

13-1

iV(W+T^WTW

19i

- 4j + 7k

iV^

TRIPLE PRODUCTS
23.

(B X C) is in absolute value
that
of a parallelepiped
volume
equal to the

Show

with sides A,

B and

^izxi^

C.

Let n be a unit normal to parallelogram 7,


having the direction of B X C, and let h be the
height of the terminal point of A above the

parallelogram

Fig. 7-25

/.

Volume

of parallelepiped

(height fe)(area of parallelogram

(A'n)(|BXCl)

= A
If A,

24. If

B and C

{|B

do not form a right-handed system,

C| n}

= A

/)

(B X C)

and the volume

A n <

A = Aa + A.j + A3k, B = Bd + B.i + B,k, C = Cd + C,2 + Csk


A

(B X C)

Ai

A2

Bi

Bi
Ci

C\
i

(B

C)

= A

Bi B%
C\

Ci

= A

(B

A3
Bs
Ca

k
Cl

{B,C^-B.C,)}

+ {B.C^- B,Ci)k]
Ai A2 As

= A,{B^C,-B,C^)+A^{B^C,-B,C,) + A.{B,C.~B.Cd

C)

show that

B:

= {AA + A.j+Ask) [{B,Cz-B^a)i +

Bi

B. B,

C,

C2

Cs

CHAP.

25.

VECTORS

7]

Find the volume of a parallelepiped with sides


By Problems 23 and

149

3i

B =

j,

C =

2k,

5j

4k.

3-10

volume of parallelepiped

24,

(B

12
15

C)
i

-201

26.

Prove that

(B X C)

= (A x

B)

C,

i.e.

24:

(B

C)

the dot and cross can be interchanged

At Aa

By Problem

Ci

(AXB)-C =

Bi B2 B:
Ci

20.

C2

C-(AXB)

Cs

C2 03

Ai A2 A3
Bi Bz Bs

Since the two determinants are equal, the required result


follows.

27.

Let

ri

and

rs

= xii + i/ij + ^ik,


= Xsi + ^3 j + 2:3k

ti = Xii + ij^j + Z2k


be the position vec-

tors of points Pi{xi,yi,zi), P2{x2,y2,Z2)

and
Find an equation for the plane
passing through Pi, P2 and Ps. See Fig. 7-26.
P3(a;3, ya, zs).

We assume that Pi, P2 and P3 do not lie


same straight line; hence they determine a
Let r
tor of

=:

xi

in the

plane.

+ yi + zk

denote the position vecin the plane. Consider


r^-r,, P1P3 = rs - n and P,P =

any point P(x,y,z)

vectors P1P2 =
r - ri which all lie in the plane.

PiP-PiP2 X PiPa
or

(r

- ri)

(r2

- r,) x

(ra

/'

Then

=
- ri)

Fig. 7-26

In terms of rectangular coordinates this becomes


[(X

or,

28.

- X^)i + (y- 2/OJ + (Z~ Zi)k] [(X2 - X,)i + (y, y,)j +


X [(X3 ~ Xi)i + (j/3 - 2/i)j + (Z3 - ^j)k] =
X Xi
Z~ Zi
-yi
using Problem 24,
X2 Xi
-2/1
Z%~Zi
Xi Xi
-2/1
Zs ~ Zi

(g,

_ g^)^

Find an equation for the plane passing through the


points
The

position vectors of

ri

''^^

.oK
equation

IS

Si

P^P^.Pa and any point P{x,y,z) on the plane are

+ j - 2k,

rs

-i

2j

4k,

ra

" ^ ~/
^'^' 7 '' ~
^'^' " "' ~ "''
T'
(r - n)
(r^ - n) X (ra - n) = 0,
{(X - 3)i +{y- l)j + (z + 2)k}
{-4i +
""''

""''

2i

- j + k,

^^^ ^^^ '"

xi

-2), P2(-l,2,4),

1,

respectively

yj

zk

*^e required plane and so the required

i.e.,

- 3)i + (y- l)j + (z + 2)k} {15i +


15(x-3) + 6{y-l) + 9{z + 2) =0

{{x

Another method:

Pi(3,

By Problem

6j

-h

6k}

9k}
or

X {-i -

2j -h

3k}

=
5x

2y

Sz

11

27, the required equation is

x~3 y-1 z+2


-1-3 2-1 4+2
2-3 -1-1 1 + 2

5x

2y

3z

11

VECTORS

150

A = +

29. If

B = 2i-3j+k, C = 4j-3k,

j,

AXB

(a)

BXC

(b)

[CHAP. 7

i - i - 5k.

-3

(A X B) X C =

Then

--1

k
-5
-3

-3

-3

5i

6j

AX

Then

8k.

(B

(A X B) X

AX

(B

23i

+ 3j + 4k.

110

C))

It follows that, in general,

(&)Ax(BxC).

(A x B) X C,

(a)

find

= 8i-8j+k.

C).

DERIVATIVES
30. If
t

(a)

(i3

+ 2i)i-3e-2j + 2sin5k,

and give a possible physical

^(t'

At
(6)

From

dt'

2)i

tion

and magnitude of the acceleration at

|dVd'|

(3J^

+ 2)i +

6e-j

10 cos5

/lio

6e-"j

at

2\/35

10 cos 5f k}

0.

6ti

12e-"3

50 sin 5t

-12j.

From

(c),

at

df

10k

(d)

If

31.

{(3i^

dt

= 0, dVdi'

6]

(b)

significance.

VW+~W+JW

dt\dtj

At

2i

fdr

_
~

d^r
(c)

dridt

\dtldt\

(a),

^,

(a)

+ 2t)i + ^(-3e-")j + ^(2sin5)k =

= 0,

dh

dr

find

12

at

0.

represents time, these represent respectively the velocity, magnitude of the velocity, accelerat =
of a particle moving along the space curve x t^ + 2t,

-3e-^', z

2 sin 5t.

Prove that

-p-

(A B)

dB

dA

\-

-^

-j^

where

B,

and

are

diff erentiable

func-

tions of u.

Method

1:

;v-(A-B)
du

+ AA)

(A

lim

(B

A'AB + AA'B + AA'AB

lim

Am

A-o

hm

A-o

Method

2:

A =

Let
d_

du

Aii

x^yz and

(x^yz){Zx'^yi

(Sx^y^zi

dz

'

Ask,

B =

Bit

B^j

_|_

dM

dA^
dM

Then

Bsk.

du

_
~

0A =

^ dB

A. (A1B1+ A2B2 + A3B3)

(A-B)

'

\^

A2J

(a
""

32. If ^{x,y,z)

AA
AA ^
AB
^^
A
+ -r-'^
+ 7
Am
Am
Am

/.

,.

+ AB) - A-B

Am

A->o

;^

^+ ^+A
A.

dM

dB ^ dA
du

dM

A =
yz^i

dM

+ yz^ -

Kzk)

^ +

du J

(Bi
\ dM

'

dM

^2^

dM

Bs"^

2,x^yi

xVz'j

3a;V^zi

x'yz^'k)

xz^i,

find ^r^(</'^)
dydz

x'y'^z^i

3x*yH

^^ ^^^ point (1,-2,-1).

x^yz^k

Sa^j/^gSj

2x^yzk

CHAP.

VECTORS

7]

dy dz

If

33. If

(3a;Vi

dy

= l, y~-2, z- -1,

A =

x2 sini/i

Method 1:
3A _
;r

(0A)

z^ cosi/j

sm yi .

2a;

,.
dA

Method

aA

=
=
=

aA

y' k,

gy

(2x
(2x

xy^k,

a;''

cos

find

dA.

2^ sin

3/ i

2/

=
=

sin

dix''
(2a;

sin

+
+

2/)i

y dx

d(z^ cos y)i

x^ cos

^^

2xyz''i

^^^

+ xVi +

(i+^^

aa;

(^^)

22 cos yj

j/ j

y dz

y dz

z^

+
-

2xyk) dy
sin

d2/)j

j/

(2z cos
(^/^

j)

da;

dz

2a;j/

dy)k

(2z cos

find

z^

sin

v^,

(a)

dj/)j

j/

(6)

div (0A)

d/dy

B/dz

xz

y^

2x^y

curl (5SA)

9x

A,

da;

(c)

|-

2xy dy)k

V X A,

(a;^2/2)j

(d)

2x^yk)

^(2^'2/)

(a;

- i^-^'^y + (^(-) 4a;2/)j


=

(^A)

^x^yz*

(x''yz*i

V X (0A)

VX

(x'yzH

d/dx

d/dy

d/dz

x^yz*

-xVz^

2a;V2'

(4a;Vz'

+ ^(-a;yz) +
Zx'y'z' + 6x''yV

^{x'yz')

^ {x^yz^Yi

"

+|k).(a;zi-,^i +

d/dx

2a;^i

Mj+Mk
By
dz

(j/"

22/

+ h^ + ^k)x(a;zi-2/^j + 2x^2/k)

(^*

Zx^yz^k

^(-2/')

(4<^^^^)

(e)

dA

2a;2/k,

curl (^A).

(e)

(d)

d(a:2/^)k

dy)i

2/

Mi
dx

(">

2k,

j/ j

sin

z''

(2z cos

a;^

GRADIENT, DIVERGENCE and CURL


34. If ^ = x^yz^ and A = xzi- y^ + 2x%k,

(6)

2x'zk

2:

dA

(^A),

12j

Bx^yz'j

2/

2/ i

Gx'yi

-12i

3A
flA
+ ^-dj/
+ -T-dz
dy
dz
sin i - y^k) dx + {x^ cos
sin
da; +
cos y dy)i +
,

-^-dx
ox

2x'yzk)

becomes

this

9a;

Sx^y^z^

151

3a;Vz')i

x^y'z^i

^(2.^^))i

^ i-y') + (&

2x*y'z^)

^(2a;2/V)

xVz^J

+ (4a;V -

2x''y'z'^)

8a;Vz')J

(2a;2/'2'

a;V)k

^(<^'')y

div

(a;2/z)k

VECTORS

152

35.

Prove

(0A)

(^A)

'

(V^)

{</>Aii

A +

[CHAP.

^(V A)
+ ^Ash)
'

0A2J

(^A,) + ^(0A.) + ^(0As)


dz
^\

dz

dy

aa;

dy

dx

dz

(V0)-A + 0(V-A)

36.

V^

Prove that

a vector perpendicular to the surface

is

<j,{x,y,z)

c,

where

c is a

constant.
Let

Then dx

i.e.

37.

V0

dr

+ y} + zk be the position vector to any point P{x, y, z) on the surface.


= da; i + dj/ j + dz k lies in the plane tangent to the surface at P. But

xi

V^

so that

is

perpendicular to dr and therefore to the surface.

Find a unit normal to the surface 2x^ + ^yz By Problem 36, a vector normal to the surface is
V(2x^

Then a unit normal

2x^y

(6)

=
=

VV

to the surface at

is

dy

and

V^

(a)

=
=

'^ct>

at the point P(3,-l,2).

at (3,-1,2)
= 12i + 8j - 24k
12i + 8j - 24k
_ 31 + 2j - 6k
(-24)"
(8)='
+
V(12)^ +
31 + 2j - 6k

10)k

V^^-

(&)

(ixy

dz

'

Laplacian of

Ay

(4^/

is

dx

4zj

find

xz^,

4x1

to the surface at

Another unit normal

38. If

+ 4:yz-hz^)

= -10

5z^

z^)'\ +
=

2*")

3x2^k

l(4xy-z^) +^(2.^) +|(-3a..^)

6xz

Another method:

_
-

5V
dx-

Ay

39.

Prove

dy-

^{2x-y-xz')
dz-

A =

0.

(V X A)

\.

>
d_

dx

assuming that
immaterial.

+ ^{2x-y-xz') + ^{2x-y-xz'>)

Gxz

div curl

div curl

/ a A _ 5A^\
dz J

\dy

d-As

d-Ai

dxdy

dx dz

a/aa;

d/dy

d/dz

Ai

A2

A3

fdA,

^A3^.

dy dz

dy dx

aAj

dxj'
aAsN
d /sAi
\
A/'
Ml _ 3A3
dy\dz
dx J
3M3
aUa
d-Ai
(,a.

dz dx

_ 5A

3a;

9_fdA2

dz\ 3x

0^]

31/

dA
dy

d^At
dz dy

has continuous second partial derivatives so that the order of differentiation

is

CHAP.

VECTORS

7]

JACOBIANS
40.

Find ds^
(a)

Method

CURVILINEAR COORDINATES

and

in

(a)

and

cylindrical

(6)

a;

Then

= -p
=

ds^

Method

= h^^ 1,
The

2:

sin

c?0

dx^

h,

position vector

dx

dp

(cos

dr-dr

a;

Then

= -r

cos

(p

<b

cos

p cos

p sin

<p

dp,

dz

dz

dp)^

<t>d<i>

sin

sin

rfp)^

(dg)^

^,^^^^,

Then

zk.

Sz

+ (-psin0i + pcos0J)d0 + kdz


+ (sin dp + p cos d0)j + kdz
dp
d0)^ + (sin
p sin
d,P + P cos
d0)^ +
+ p'id^y + (dzr

sin0j)rfp

=
=
r sin

IS

fl0

(cos0i

(6)

P cos

(- p sin ^d,p

dz""

=z
d,f +

p sin 0,

p^(d^Y

dp

ds^

2/

dy

dp,

+ (d^). = f^.^^^y ^ f^^^^^y ^


= h^=p, h,^h,^l are the scale factors.

(dpY

=
=
Thus

dy''

P cos 0,

cos

^
h,

spherical coordinates and determine the


scale factors.

1:

dx

and

153

p sin
(cos

(dp)'

ff

d0)i

<f,

cos 0,

sin sin0 d0

j/

r sin # sin 0,

r cosff cos0 de

(dz)2

r cos 9

sin 9 cos

dr

dy

and

41.

- r sin cos d0 + r cos e sin ds + sin e sin dr


rfz
= r sin d9 + cos e dr
(dsr = (dccf + (dyr + (dzr
= (c^rf + r^(d9f + r^ sin^ 9(^0)^
The scale factors are h, = K = 1, h. = h, =
r, hs = k^ = r sin e.

Find the volume element


The volume element

dV
(a)

dV

in (a) cylindrical

and

(6)

in orthogonal curvilinear
coordinates

In cylindrical coordinates,

h.h.h.du,du.du,

Ui= p,
p,
dV =

U2i

d,
-P,

u!,
us

(l)(p)(l)

is

\^ip.l^\du,du.du.

= ^,
z ni
h,-

dp d0 dx

spherical coordinates and sketch.


u,,m,u,

\
1,

h,~
/i2-p,

,
h.3

-1

r
[see

r.
u,
Problem

./ .,
40(o)].

Then

pdpd^dz

This can also be observed directly from Pig.


7-27(a) below.

/'

/(' =

J=

(r

sin

r^ sin

(9

(o)

Volume element

in cylindrical coordinates.

(6)

Fig. 7-27

Volume element

rf0)

(r rf6') (i/r)

i9 rfr rfi9 cl(p

sin6'

rf<^

in spherical coordinates.

[CHAP.

VECTORS

j^54

(6)

In spherical coordinates, Wi

= r,

dV

= e,

m.

Us

= <t>,h, =

(l)(r)(r sin 0)

dr de

1,

d<p

h^^r,hz =

r sin e

dr de

r" sin e

[see

Problem

40(6)].

Then

d<p

This can also be observed directly from Fig. 7-27(&) above.

42.

Express in cylindrical coordinates:


Let M,

Page

(a)

= p,

M2

142.

Then

grad*

where

ei, ez, Cs

= 0,

W3

= 2,

(c)

A =

V^*

=
"

j ^^^e.

fe

= P,

^^

Aiei

IK

<'-.'

=1

[see

(b)

div A,

Problem

19*

e.

Cs

in the results 1, 2

40(a)]

9*
-^e,

V^*-

(c)

+
p,

S*

and 4 on

5*
+ -ea
,

e.

0, 2 respectively.

^]

+ ^262 + Aaea.
riA^Ki)^*^

(l)(p)(l)

*,

/is

15*

grad

are the unit vectors in the directions of increasing

=
where

= 1,

a*

V*

fei

(a)

dpV

Lap\
dpj

dp

(1)

"^

p'a0^

50\

J
"^

a /(i)(p)

i^Al)a)a*\

90/

(p)

92\

9*\"|

9z/J

(1)

92"

IMISCELLANEOUS PROBLEMS
where
43. Prove that grad /(r) = ^r,

yx^+^M^

and

/'(r)

d//dr

is

assumed

to exist.
grad/(r)

V /(r)

/'(r)^i

Another method:

^f{r)i + ^/(r)j + ^/(r) k

f'(r)^j

/'(r)^k

^(xi + yj + zk)

we have

In orthogonal curvilinear coordinates ui,ui,U3,

a*

hi dui

a*

h-t

du%

9*

/,N

hi 3u3

particular we use spherical coordinates, we have tti = r, U2 = e, us = 0.


function of r alone, the last two terms on the right of (1) are zero. Hence
observing that ei = r/r and hi = 1, the result
If in

* = f(r), a

'^

44.

(ft)

Find the Laplacian of

equation

V^^ =

0.

<^

'

/(r).

(b)

dr

we

letting

have, on

Prove that

Then

<i,

= Vr

is

a solution of Laplace's

*^HAP.

(a)

VECTORS

7]

By Problem

43,

V0
By Problem

35,

Laplacian of ^

Another method:

assuming that

(6)

U = f(r),

From

Sry

"^

(V96)

the result in part

showing that 1/r

is

we have

V-J^^rl

we have
ao/ +

r^sinffSeV'"

the last two terms on the right are zero and

we

r= sin^ e

~d^

find

we have

(a),

v{^)
\rj

45.

^^r
r

has continuous second partial derivatives,

f(r)

V/(r)

In spherical coordinates,

r^arV
ii

I55

^(l\

_
~

dfl\

dr^\rj

rdr\rj

2
r"

_
-

a solution of Laplace's equation.

particle moves along a space curve r = r{t),


where t is the time measured from
some mitial time. If i; = \dr/dt\ = ds/dt is the magnitude of the
velocity of the particle
(s IS the arc length along the space
curve measured from the initial position), prove
that the acceleration a of the particle is given by

dv

where T and

v^

are unit tangent and normal vectors to the space


curve and

rf^r-

fa^yV

r/d.^Y
,

ds^

The

velocity of the particle is given

dv
dt

Since

T has

_
- dt^ +

dt^"^^

unit magnitude,

^^^*

^'lF

,''

1.

_
-

dvdT ds
di^ + ^'Tsdi =

Then

2X.^ =.
-^T-^

u,
0,

/d^z^''~"'

+1^

Then the acceleration

vt.

dT

we have T T =

+ -- =
rfj-r^-x
''S
tfTrTll'V^
^^iT.u^^^
dT/ds, and called
01
the principal

by v

di^

A/T^

^\d^)

ds^J

"^

dv

d^T +

dT
T-^

or

dF

given by

dT

^^f

differentiating with respect to

P^'-P'^di"l^r to T. Denoting by
to the space curve, we have

normal

is

s,

the unit vector in the direction

(2)

[CHAP.

VECTORS

156

where
dT/ds

k is the

Now

magnitude of dT/ds.

f/

=
Defining p

= 1//C,

(2)

becomes dT/ds

N/p.

dr/ds

\2

,,

{(S)

ISI

T =

since

= dVds^ Hence

i;Vp in the direction of

The components dv/dt and

/j2

(S)

- (i?)

(i)

we

Page

(9),

\0

/^2 \^

Thus from

^T
dt

equation

[see

we have

139],

*"

have, as required,

^N
p

T and N

are called the tangential and normal

The

acceleration.
components of the acceleration, the latter being sometimes called the centripetal
of the space curve.
curvature
and
curvature
radius
the
of
respectively
are
and
k
quantities p

Supplementary Problems
VECTOR ALGEBRA
46.

Given any two vectors

47

A man

48.

If

50

and B,

illustrate geometrically the equality

4A +

3(B

- A) = A +

3B.

south. By using an appropriate


travels 25 miles northeast, 15 miles due east and 10 miles due
is from his startmg position.
he
direction
scale determine graphically (a) how far and (b) in what
13.2 north of east
miles,
Ans.
33.6
analytically?
answer
the
determine
Is it possible to

is

49.

prove that
are any two non-zero vectors which do not have the same direction,
and
B.
by
determined
plane
the
in
lying
a vector

and

B and C are non-coplanar vectors


a;iA + i/iB + 2iC = X2A + 3/2B + 22C, prove
If

mA + nB

A,

(vectors which do not all lie in the same


that necessarily xi = X2, yi^yz, Z1-Z2.

and

plane)

ABCD be any quadrilateral and points P,Q,R and S the midpoints of successive sides. Prove
PQRS is a parallelogram and (6) that the perimeter of PQRS is equal to the sum of the
that
(a)
lengths of the diagonals of ABCD.
Let

51.

Prove that the medians of a triangle intersect at a point which

52.

Find a unit vector

C = 3i-2j +

in

the direction of the resultant of vectors


(6i - 2j + 7k)/\/89

4k.

SCALAR PRODUCT
(A-B)|
|(A + B)

DOT

53.

Evaluate

54.

Prove the law of cosines for a triangle.


use

a trisection point of each median.

A =

2i-i + k,

B =

+ ] + 2k,

Ans.

The

or

is

if

A =

21

- 3j +
[Hint:

5k

and

B =

Take the

3i

+ j-2k.

sides as A, B,

Ans. 24

C where C =

A - B.

Then

C-C = (A-B)'(A-B).]

55.

Find a so that

56.

If

A =

2i

21

- 3] + 5k

and

31

+ i + k, B = i-2j + 2k

direction of B.

Ans. 17/3

aj

- 2k

and

are perpendicular.

C = 3i-4j +

2k,

find the

Ans. a

-4/3

projection of

A+C

in

the

^^^^-

"

7]

VECTORS

-j^g,^

^''''

tro^Tl'^TelT^!t)1''-'y^~y^^^^^

()

the

58.

Prove that the diagonals of a rhombus


are perpendicular to each other.

59.

Prove that the vector (AB

The CROSS or
60.

If

A =

"'

In'

+ BA)/(A + B)

of the

represents the bisector of the angle between

median drawn

and B.

VECTOR PRODUCT

2i-j + k and B =

tS+W^'

AXB = AX

2j

- 3k,

find

+ B) X

|(2A

(A-2B)|.

perpendicular to the plane of the vectors

62.

If

63.

Find the area of the triangle with


vertices

U. Find^the

len^h

C,

B=C

does

shortest distance

A =

31

Ans. 25^3

-2i + 4k and B =

+ i - 2k.

necessarily?

from the point

(2.-3,1), (1,-1,2), (-1,2,3).

Ans.

(3,2,1) to the plane determined

by

^^3

(1,1,0),

(3,-1,1),

TRIPLE PRODUCTS

66.

Prove that

(a)

(6)

AX

= B (C X A) = C (A X B)
(BXC) = B(A-C) - C(A-B).
X

(B

C)

''

2' T4T-t:=l ''^ ^^^"^ ^^

68.

Find the volume of the tetrahedron


with vertices at

69.

Prove that

(A X B)

(C

X D) +

(B

C)

^'''"^^

(^'-^'-2)' (-l'2.-3). (4,1,0).

(A X D)

(2,1,1), (l,-i,2), (0,1,-1), (1,-2,1).

(C X A)

Ans. |3

(B

X D)

are differentiable functions of u.

0.

DERIVATIVES

71.

72.

73.

Prove that

AX^
du

:^(AXB) =

du'

^vn

+
^ du^^

Find a unit vector tangent to the


space curve
Ans. a + 2} + 3k)/Vu
If

a cos .

b sin

scalar, prove that

yz]

()

.t,

u
^here
a;

= f.

,y
2/

and

- f^
^

/s
t'

at +i,
the
4-

pomt. where
,

where a and b are any constant noncollinear


vectors and .

+ xyzk,

(a

find

b),

1^

(6)

_|

.^r

0.

at the point (2,1,-2).

Ans. leVE

= 1.

is

a constant

[CHAP.

VECTORS

158

GRADIENT, DIVERGENCE and CURL


76.

If

77.

If

4,

continuous partial derivatives prove that:

B have

U, V, A,

VU+VV,

(a)V(U + V) =

V -(A + B)

(6)

+ yz + zx and A =

xy

at the point

-1,

(3,

(a) 25,

where

78.

Show

79.

Prove:

80.

Prove that

81.

Find a unit normal to the surface


Ans. (3i + 4] - 6k)/V6i

82.

If

83.

(a)

that VX(r''r)

VX

(a)

A =

(C7A)

yzj

V X

Prove that
Problem 82.

(Vt7)

U = 0,

grad

curl

Sxz^i

(x

+ y'zj + z'xk,

x'yi

Ans.

2).

V-A + V-B,

:=

(6)

xi

A +

find

A),

(a)

and r
(6)

V X

V0,

(A

+ B) = V X A + V X B.
V A

(6)

and

(c)

(V<p)

- 30] + 47k

56i

(c)

+ z'k

yj

t/(V

2/

(c)

|r|.

(A X B)

= B

(V X A)

- A

(V X

B).

stating appropriate conditions on U.


x^y

+ 2z)k,

- 2xz + 2y^z* =

= - V^A + V(V

(V X A)

Ans. - 6a;i

curl curl A.

find

A).

at the point

10

(b)

(2, 1,

-1).

- l)k

(62

Verify the result in

(a) if

is

given as in

JACOBIANS and CURVILINEAR COORDINATES


84.

85.

Express

(a)

l^ir^Ai)
1

^r_

dr

8U2

dU3

V^*

in spherical coordinates.

3*

1
;_

o<p
sm aa^^^

(smeA2)

/ ,a*\

(0)

dr

div A,

(6)

,19*
+ -^-^2
+
r ad

(b)

S'i'\

^-^

AiCi

+ AiCi + Ase^

5^

f
+ ^s-ihT^^l^sms
^^y + ^s^^2gQ^2
,

v^^\J"a^J

to parabolic cylindrical coordinates is defined by the equations


that the system is orthogonal. (6) Find ds^ and the scale
Prove
uv, z
y
Find the Jacobian of the transformation and the volume element.

X{v} - r^),

factors,

Ans.

Write
Ans.

(c)

(6)

ds^

(c)

u^

(a)

=
v",

(u"

{u"

V"* and

(a)

V^*

(6)

div

+ t)^) du" + (m' + v^) dv^ +


+ v^) du dv dz
(6)

div
1

dz\

hi

- y/vFTV,

h2

h^

in parabolic cylindrical coordinates.

/s^*

^7+^(^5^ +

(a)

z.

a^<f>\

^j

^
a'*

|^(V^I^T^A.)|
^iF^{air(^^^^'+^^^) +

a As

dz

Prove that for orthogonal curvilinear coordinates,

_
"

ei

a<!>_

hi dUi

62

a*_

e3

d<t>

ha r)U3

h2 dU2

[Hint: Let V* = aiCi + a2e2 + asCs and use the fact that
rectangular and the curvilinear coordinates.]

89.

A =

where

The transformation from rectangular


X

88.

grad *,

a*
-ei

(c)

87.

flr

dUi

(a)

Ans.

86.

d(x, y, z)
d{ui, U%, Ma)

Prove that

d*

= V*

Give a vector interpretation to the theorem in Problem 35 of Chapter

dr

6.

must be the same

in both

CHAP.

VECTORS

7]

MISCELLANEOUS PROBLEMS
If A is a differentiable function
90.

u and

of

91.

Prove formulas

92.

If p and
are polar coordinates
satisfies Laplace's equation.

93.

If

94.

Find the most general function of


the spherical coordinate e which
Ans. (a) A + B Inp,
(6) A + B/r,

y =

2 cos

6,

and 8 on Page

sin='

prove that dA/du

1,

is

perpendicular to A.

141.

U =

and A,B,n are any constants, prove that

e cos

^^

^^y

^^^^

6 sin g cos

the cylindrical coordinate

(a)

(c)

95.

|A(m)|

I59

-5

(4

+B

sin n<p)

sin'e)

the spherical coordinate

(6)

p,

p'{A cos n,p

r,

satisfies Laplace's equation.

A + B

(c)

In (esc

where

cote)

and

are any constants.

Let T and N denote respectively the unit tangent vector and unit principal normal vector
to a space
curve r = r(M), where r(M) is assumed differentiable. Define a vector B = T X N called
the unit
hinormal vector to the space curve. Prove that

dT

_
-

-d^

dB
'^'

dN
^

^,
= --N,

dF

= .B-T

These are called the Frenet-Serret formulas and are of fundamental importance
in differential
geometry^ In these formulas k is called the curvature, t is called the torsion; and
the reciprocals of
these, p = 1//C and a - l/r, are called the radius of curvature and radius
of torsion respectively.
96.

Prove that the radius of curvature at any point of the plane curve

(a)

differentiable, is given

by

...

(1

Find the radius of curvature at the point


Ans. (h) 2\/2
(6)

97.

spherical coordinates

(r

r'e^

sinx, z

0.

where

f{x) is

of the curve

(57/2, 1, 0)

is

given respectively in

(a) cylindrical,

by

r^^ sin^ e)e^

(re

(p

2r|

pi>'')^p

where dots denote time derivatives and


creasing

Prove that the acceleration of a particle along a space curve


(h)

98.

+ y'T'
y"

= f(x),

,,,,,

(p0

2p0)e^

r0^ sin e cos e)eg


Cp, e^, e., Cr, e^,

e^

26^

(2r0 sin

2r#0 cos

r0 sin e)e^

are unit vectors in the directions of in-

respectively.

p, 0, z, r, e,

Let E and H be two vectors assumed to have continuous partial derivatives (of second
order at least)
with respect to position and time. Suppose further that E and H satisfy the equations

V-E =
prove that

E and

0,

V-H =

VXE
V ^

I/,
0,

Cl

VXH =

1 5H
-i|a,

=

.-r

. -

aE

^fl

(i)

satisfy the equation

[The vectors E and


are called electric and magnetic field vectors in electromagnetic theory. Equations (1) are a special case of Maxwell's equations. The result
(2) led Maxwell to the conclusion that
light was an electromagnetic phenomena. The constant c is the
velocity of light.]
99.

Use the

relations in

Problem 98 to show that

j^{l(E'
100.

m}

+ cV.(EXH)

Let At,Ai,A3 be the components of vector A in an xyz rectangular coordinate


system with unit vectors
(the usual i, j,k vectors), and A'l.A'^.A'^ the components of
A in an x'y'z' rectangular coordinate
system which has the same origin as the xyz system but is rotated with respect to it
and has the unit
vectors i'l, i'2, i'3. Prove that the following relations (often called
invariance relations) must hold:
ii, i2, ia

An
where i

L,..

hnA[

+ hA2 +

hnAs

M=

1, 2,

[CHAP.

VECTORS

160

101. If

is

the vector of Problem 100, prove that the divergence of A,


prove that
i.e.

i.e.

A,

is

an invariant

(often

called a scalar invariant),

Mi

Mi

4-

dy'

dx'

Mi

Mi + Mi + Mi
dz
dy
dx

dz'

The resvdts of this and the preceding problem express an obvious requirement that physical quantities
must not depend on coordinate systems in which they are observed. Such ideas when generalized lead
to

102.

an important subject called tensor analysis which

Prove that

(a)

103. If ui,U2,U3 are

B,

(6)

A X B,

V XA

basic to the theory of relativity.

are invariant under the transformation of Problem 100.

orthogonal curvilinear coordinates, prove that

()%fff?

Vk..V.XV3

and give the significance of these


104.

Use the axiomatic approach

105.

in

(6)

(^

)(V..

V. X V3)

terms of Jacobians.

to vectors to

prove relation

(S)

on Page 138.

Ai,A2, ...,A is called linearly dependent if there exists a set of scalars


identically, otherwise the set is
cAi + C2A2 + + cA =
ci,C2, ...,Cn not all zero such that
Ai = 21 3j + 5k, Aa = i + j - 2k, A3 =
(a) Prove that the vectors
called linearly independent.
(b) Prove that any four 3 dimensional vectors are linearly
are linearly dependent.
3i - 7j + 12k
Ai = aii+ bij + Cik,
(c) Prove that a necessary and sufficient condition that the vectors
dependent.
As = oai + feaj + csk. As = aai+feaj + csk be linearly independent is that Ai'AzXAa ^ 0. Give
set of

vectors

a geometrical interpretation of
106.

(c)

is

this.

complex number can be defined as an ordered pair (a, b) of real numbers a and 6 subject to certain
rules
rules of operation for addition and multiplication, (a) What are these rules? (6) How can the
considered
be
can
numbers
complex
why
Explain
(c)
in {a) be used to define subtraction and division?
as two-dimensional vectors, (d) Describe similarities and differences between various operations involving complex numbers and the vectors considered in this chapter.

chapter 8
Applications of Partial Derivatives
APPLICATIONS
1.

to

Tangent Plane

GEOMETRY

to a Surface.

Let F{x,y,z) =
be the equation of a
surface S such as shown in Fig. 8-1. We
shall assume that F is continuously differentiable unless otherwise indicated.
Suppose
we wish to find the equation of a tangent
plane to S at the point P{xo,yo,Zo). A vector

K.

s-

VFl,

this point is No = VF\p, the


indicating that the gradient is
to be evaluated at the point P{xo,yo,Zo).

normal to S at
subscript

If ro

spectively

and r are the vectors drawn refrom


to P{xo, yo, zo) and Q{x, y, z)

on the plane, the equation of the plane


(r-ro) -No
since

r-ro

(r-ro)

is

V^|p =

(1)
Fig. 8-1

perpendicular to No.
In rectangular form this is
is

dF
dx

{x~Xo)

dF
dy

(y-yo)

dF
dz

~ Zo)

{z

(2)

In case the equation of the surface is given in orthogonal curvilinear


coordinates in
the form F{ui, Uz, Ws) = 0, the equation of the tangent plane
can be obtained using the
result on Page 142 for the gradient in these coordinates.
See Problem 4.

Normal Line to a Surface.


Suppose we require equations for the normal line to the surface S at
P{xo,yo,Zo).
If
we now let r be the vector drawn from O in Fig. 8-1 to any point {x,y,z) on the normal
No, we see that r-ro is collinear with No and so the required
condition is
2.

(r

In rectangular

form

- ro) X

No

(r-

ro)

XVF\p

Xa

dF
dx

yo

{3)

Zo

dF

dF

dy

dz

Setting each of these ratios equal to a parameter (such as


yields the -parametric equations of the normal line.

The equations for the normal


is

this is

surface

H)
,

or u) and solving for

line can also be written


expressed in orthogonal curvilinear coordinates.

161

x,

when the equation

y and

of the

[CHAP.

APPLICATIONS OF PARTIAL DERIVATIVES

162

Tangent Line to a Curve.

3.

Let the parametric equations of curve C


of Fig. 8-2 be a; = f{u), y = g{u), z = h{u)
where we shall suppose, unless otherwise indicated, that /, 9 and h are continuously
differentiable. We wish to find equations for
the tangent line to C at the point P{xo, y, Zo)

where u uo.

R=

If

gent to
If ro

f{u)i

+ g(u)i + h(u)k,

at the point

a vector tan-

P is given by

To

^j^.

and r denote the vectors drawn respecto P{xo, Vo, zo) and Q{x, y, z) on
from

tively

the tangent line, then since r


with To we have

- ro

collinear

is

(r-ro)xTo = (r-ro)X-^^ =
In rectangular form this becomes

The parametric form


If the

and G{x, y, z) =

xo

Va

Zo
(6)

obtained by setting each ratio equal to u.

F{x,y,
given as the intersection of two surfaces with equations
are
line
the corresponding equations of the tangent

curve

is

Fig. 8-2

(5)

is

0,

F.
G.

Fy
Gy

F.
G,

z-Z(,_

y-yo

x-xa

z)

n
G.

Fy
Gy

F,
G.

be found when the


Note that the determinants in (7) are Jacobians. A similar result can
coordinates.
surfaces are given in terms of orthogonal curvilinear

4.

Normal Plane to a Curve.


Suppose we wish to find an equation for the normal plane

of Fig. 8-2

(i.e.

the plane perpendicular to the tangent


to any point {x,y,z) on this plane,
Then the required equation is

line to
it

be the vector from


dicular to To.

(r

- ro)

dn

To

to curve C at Pixo,yo,zo)
at this point). Letting r

follows that

r-ro

is

perpen-

(8)

("^-^"^-d^

In rectangular form this becomes


f'{uo){x-xo)

when

5.

g'{uo){y-yo)

the curve has parametric equations x

F.
G.

Fy
Gy

when

the curve

is

{x

xo) +

defined

F.
G.

by F{x, y,

z)

F.
Gx

0,

= f(u),

h'{uo){z-Zo)

(y-yo)
G{x, y,

z)

= g{u),
F.
Gx

+
=

(9)

- h{u), and

F
Gy

[z

-zo)

[10)

0.

Envelopes.
If 4,{x, y,a)

a curve

which

is

be
a one parameter family of curves in the xy plane, there may
that
such
and
family
the
of
member
tangent at each point to some
is

^^^^-

and

APPLICATIONS OF PARTIAL DERIVATIVES

^J

" is called

163

the envelope of the family.

* ^^^""i"^ *he envelope of a one parameter


family of
surfaces' <f>(x,y,z,a). ^"i!?^*^"^^^
suriaces
This envelope can be found from

JS?/T

Extensions to two (or more) parameter


families can be made.

DIRECTIONAL DERIVATIVES
Suppose Fix,y,z)

is

defined at a point

4f =

lim ZCa^

lim
.
4s-o As

'l^lTfLnT' '"'

+ Ay,z + Az) ~

dF ^

dFdx

ds

dx ds

"^

^' *'^ P^"* (^'^'^)

dFdy
"^

F(x, y, z)

As^

'^"'"'^''^ ^'

m vector form this can be written

The maximum value

+ AX,

4,^0

'""''""'

on a given soace eurvP

ix,y,z)

{13)

^1-^

t*>e

curve

sFdz
^ dzd^

dy ds

T of

(U)

of

the directional derivative Is


given by |vf
These maxima occur in directions
normal to the surfaces Fl-r ^\
any constant) which are sometimes
called
|.

...^.0.^1^^ oi^^/ ;J/lt.^"

DIFFERENTIATION UNDER
^^*

where

Ut

the

^W =
and

tia

may depend

INTEGRAL SIGN

f{x.a)dx

on the parameter

'^

derivatives for

In case
(see

and u^ are constants, the

Proll'emfifs^g)""''

INTEGRATION UNDER

if

the

^^

/,.x

^ ^"^ ^ ^" - -^-^ ^ ^^^


are continuous and have continuous

'''*'

^:;

z.,

last

"^"'"''''^ '"^^'

gag 6

Then

a.

r^"'
Xe Sinf i!<:^<':f fi'^T
rf
and
= - ^"^
aZ^
'

' ''

z.,

two terms of

^"^"

"^^^"^

(17) are zero

-^1-*-^' ^e^-te

integrals

INTEGRAL SIGN

..iitvi;i"rtht'?/ir:;;dttrczsr'

'-

^ "^
'-

-'- '-'"-

The

[CHAP.

APPLICATIONS OF PARTIAL DERIVATIVES

164

result is

known

under the
as interchange of the order of integration or integration

integral sign.

MAXIMA and MINIMA


A point {xo,y(,) is called

a relative maximum point or relative minimum point of f{x,v)


h,yo + k) S f{xo,yo) for
respectively according as f{xo + h,yo + k) g fixo,yo) or f{xo +
< |/i| < S, < Ifcj < 8 where 8 is a sufficiently small positive number.
all h and k such that
minimum is
necessary condition that f{x,y) have a relative maximum or

dx
If ixo,yo) is

a point

(called a critical point) satisfying equations {19)

by

fined

dy

'

1.

maximum

2.

(a;o,

^0) is

a relative

minimum

point

point

A>

if

and
and

A>

if

METHOD
A

of

is

obtained

A=

if

relative

^^^

d'f

^
^

>

maximum
<l>{x,y,z) =

F{x,y,z) subject to a constraint condition


auxiliary function
^ F{x,y,z)
G{x,y,z)

or
0,

\
j

or

<
^^^ ^^^

>

(a:o,!/o>

point

MAXIMA

for

de-

32/

[ov

<
^^^

if

A<0.

If

such case further investigation

(in

LAGRANGE MULTIPLIERS

method for obtaining the

^
^

minimum

a relative maximum or
a saddle point.
called
sometimes

{xo,yo) is neither

no information

is

{20)

32/

a relative

4.

"^^

(xo, 1/0) is

is

if

_ jfs2f\(s^\ _ fj^
Kdxdy
dx^jKdy^J

then

3.

and

""

'

and

A<0,

is

(aJo.l/o)

necessary).

MINIMA

values of a function
the formation of the
of
consists

minimum

{21)

X^{x,y,z)

subject to the conditions

"

dx

"'

dy

'

dz

The parameter

minimum.
which are necessary conditions for a relative maximum or
multiplier.
Lagrange
a
called
is
x,y,z,
of
which is independent
relative
The method can be generalized. If we wish to find the

minimum
^^{xu

values of a function

.,xn)

0,

4>M' ..,^n)

F{xX2,Xs,
...,

0,

G{x,,X2,...,Xu)

maximum

X,

or

subject to the constravnt conditions


.,Xn)
.,x) = 0, we form the auxiliary function

Ki^u
F + X,<j., +
.

X,4>,

A,c/>,

{23)

subject to the (necessary) conditions


dXi

where

A,, A,,

dX2

'

'

dXn

Lagrange multipliers.
...,X which are independent of x^x^, ...,Xn, are the

APPLICATIONS
etc.,

'

to

ERRORS

function of x,y,z,
The theory of diflferentials can be applied to obtain errors in a
28.
when the errors in x,y,z, etc., are known. See Problem

CHAP.

APPLICATIONS OF PARTIAL DERIVATIVES

8]

165

Solved Problems

TANGENT PLANE
1.

NORMAL LINE

and

Find equations for the

tangent plane and

(a)

SURFACE

to a

normal

(&)

line to the surface x^yz

+ 3y^ =

2xz^ Sz
(a)

at the point (1,2,-1).


The equation of the surface is
(1,2,-1)

F =

x^yz

+ Sy^ 2x2^ + 8z =

0.

normal

surface at

to the

is

No

=
=

VF|_,,_

Referring to Fig.

Page

8-1,

{2xyz-2z')i

-6i

llj

+
+

(x'z

+ 6y)j +

(x'y

- 4xz + S)kl^_^^_,^

14k

161:

The vector from O

to

The vector from O

to the point (1,2,-1) on the tangent plane is

The vector

ro =

any point

(x

1)1 +

on the tangent plane

{x, y, z)

2)j +

(j/

(z

+ l)k

is

ro

+ zk.
= i-|-2j k.
a;i

j/j

the tangent plane and

lies in

is

thus

perpendicular to No.

Then the required equation

- ro)

(r

No

-6(a;

(6)

Let r

a;i

=
- 1) +

is

+ j/j-|-zk

l)k

is

normal

collinear with No.

- l)i + (y - 2)j + (z +
or
+ 1) =
O

to

- ro) X

is

{-6i

llj/

+ llj + 14k} =
14z + 2 =

any point (x,y,z) of the normal


i + 23 k.
The vector r ro

The vector from O

No.

(a

l)i +

(2/

2)j +

Then

No

x-1 y-2

i.e.

-6
which

6a;

ro

is

(r

l)k}

14(z

be the vector from

to the point (1,2,-1) on the


(2

{(x

i.e.

- 2) +

11(2/

11

+1
14

equivalent to the equations

ll(a;-l)

-6(2/

-2),

-2)

_
~

11

14(2/

14(a;-l)

11(^+1),

-6(z

+ 1)

These can be written as


X

-6
parameter

2 _
~

form for the equations of a

often called the standard


to the

14

By

line.

setting each of these ratios equal

we have

(,

= l-6f,

a;

2/

lit,

14t

called the parametric equations for the line.

2.

In what point does the normal line of Problem 1(6) meet the plane x
Substituting the parametric equations of Problem 1(6), we have

Then

3,

Show

a;

- 6t =

7,

+
2

that the surface x^

of surfaces

x^

{2

3(2

llt

lit)

2(14t-l)

-9, z

2yz + y^ =
4a)y^ + az^

14t

10

or

- 1 = -15

+ 3y 2z =

= -1

and the required point

is (7,

x'

y^

2a;i

(32/'-2z)j

2yz

and

G =

a;^

(2

- 4a)2/^ -

az"

Then

VF =

-9, -15).

is perpendicular to any member of the family


at the point of intersection (1,-1,2).

Let the equations of the two surfaces be written in the form

F =

10?

22/k,

VG =

2a;i

2(2

- 4a)2/j -

2a2k

'

'

Thus the normals

to the

two surfaces at

Ni

all a,

Ni Na

Since

4.

[CHAP. 8

APPLICATIONS OF PARTIAL DERIVATIVES

IQQ

(2)(2)

2i

(1,

-1, 2) are given by

N2

2k,

- 2(2 - 4a) - (2)(4a) =

0,

2i

- 4a)j -

2(2

4ak

follows that Ni and Nz are perpendicular for

it

and so the required result follows.

where
is given in spherical coordinates by F{r, 6, <l>) = 0,
the
tangent
for
equation
an
(a)
Find
differentiable.
continuously
we
plane to the surface at the point {r^, e^.^o). (6) Find an equation for the tangent plane
(c) Find a set of equations
at the point (2^2, 7r/4, SttM).
to the surface r = 4 cos

The equation

of a surface

suppose that

is

for the normal line to the surface in


(a)

The gradient of *

in orthogonal curvilinear coordinates is

13*

,13*

dui

hi dU2

All

Pages 141,

3r

hs dUs

142).

m = r,

zk

dUs
1

es

hi dUi

hi 3ui

In spherical coordinates
r sin e cos <pi + r sin e sin 0]

fls

62

ei

,13*

3r

3r

where
(see

at the indicated point.

(b)

+r

U2

= e,

u%

0,

fei

fe

1,

= r,

As

and

r sin e

xi

+ j/j +

cos ek.

Then

= sin e cos
+ sin 9
= cos e cos i + cos 9
= sin i + cos j

1-61
J

<t>\

62

t 63

sin

sin

cos
sin 9

k
k

(1)

and

,^iaF

3F

As on Page

Now

161, the required equation

substituting

VF|p

0.

we have

(i) in (2),

dF

(r-ro)

is

BF

_sin9cos0

+
.

dr

dF

cos

^^g^

f aF\

So

sin 00
r

-~

cos 00

ro sin e

dF\

3F

3<f,

cos 00

dF
TT

dF

-^f

Ai + Bj + Ck,
Denoting the expressions in braces by A, B, C respectively so that VF|p =
This can be written
x) + B{y - 1/0) + C(z - Zo) = 0.
see that the required equation is A{x
for x, y and z in these coordmates.
in spherical coordinates by using the transformation equations

we

(6)

We

have

F -

4 cos

= 7r/4,

0.

Then

= 3!r/4, we

dF/dr

1,

dF/de

4 sin

VF|p

9,

Ai

dF/d-f,

Bj

0.

+ Ck =

-i

Since

u = 2\/2,

From

the transformation equations the given point has rectangular coordinates (- V2, V2,

and so

r-n

{x

+ V2)i

+ {y ~ ^/2)i +

have from part

(a),

j.

2),

(z- 2)k.

The required equation of the plane is thus


In spherical coordinates this becomes r sin 9 sin

- + Vi) + - \f2) =
r sin 9 cos = 2V2.
= 4 cos 9 becomes
+
(a;

(3/

or

2V2.

= 4 and
a;^
j/^ + (2 - 2)'
In rectangular coordinates the equation r
cases, however, it may
the tangent plane can be determined from this as in Problem 1. In other
method of part (a)
not be so easy to obtain the equation in rectangular form, and in such cases the
is simpler to use.

CHAP.

APPLICATIONS OF PARTIAL DERIVATIVES

8]

The equations

(c)

of the normal line can be represented by

-110

+ y/2.

X + V2

5.

The vector from

(3

sin 2t)j

To

III/

lt

(1

sinf)i

The vector from O

to the point
to

equation

- ro =

normal plane
where t - ^tt.

- ro) X

To

- l)k

(z

(z

3j

k.

xi

R = (t
= ^ir is

21

2j

cost,

cos

()i

3k

+ yj + zk.

collinear with To, so that the required

is

ai

2-2
2/

-^ = ^ = ^1
^

or in parametric form x

2t

+ ir
2

+ j/j + zk

be the vector from O to any point (a;, y, z) of the normal plane. The vector
where t = \w\s, ro = \tt\ + 3j + k. The vector r - ro = {x~ \Tr)\ + (?/ - 3)3 +
the normal plane and hence is perpendicular to To. Then the required equation is

to the point

l)k lies in

(r-r)-To

6.

line is

^n-i

^Jr

0,

and the required equations are


= 3-2t, z ^ St+1.

ro

is

= t-

to the curve x

3sin3fk|,=i4:,

on the tangent

{x, y, z)

(y - 3)j +

- lv)i +

(x

+ 2cos2tj -

where t=lir

any point

Let r

0,

(b)

from O

the

is

(r

(6)

2/

z-2. Thus

CURVE

to a

at the point

The vector from O


Then

Page 162) to any point of curve C is


Then a vector tangent to C at the point where t

= 1^7j-

a;

the plane

lies in

(see Fig. 8-2,

cos St)k.

-n-

or

tangent line and

cos St

origin

(1

dR

(a)

that the line

'

NORMAL PLANE

and

Find equations for the


y = 3 +sin2t, z - 1
(a)

y-yfi

ZZx

y-V2 ^ z-2

member being

the significance of the right hand


required line is given by

TANGENT LINE

167

or

2(a;

-i-,r)

2(1/

- 3) +

3(z

- 1) =

0.

Find equations for the (a) tangent line and (&) normal plane to the curve
= -2, 2xz - x^y ^ 3 at the point (1, -1, 1).

Zx^y

y^z

The equations

(a)

of the surfaces intersecting in the curve are

F =
The normals

N2

As

To

in

Problem

The
Page

162.

3{x

5(6) the

and

G ^

0,

2xz

x^y

+ 2yz)} + y'k = -6i + j +


- x^j + 2a;k = 41 - j + 2k

(3x^

P is
(-61 + j + k) X

tangent line

z
= 2

{(x

or

(6)

1)1

or

normal plane

- 1) + U{y + 1) +

results in (a)

(2z~2xy)i

or

(r-ro)-To =
i.e.,

6xyi

5(a), the

xl_y+l
- ^Q"3

^^'

(6)

- r) X

2/^z

to the curve at

Nx X N2

Thus, as in Problem
(r

=
=

VF\p
VG|p

Then a tangent vector


To

to each surface at the point F(l, 1, 1) are respectively

=
=

Ni

Bx^y

is

+
a;

is

(41

- j + 2k)

31

16j

2k

given by
(y

l)j

+
3i,

(z

l)k}

16t

{31

1,

16j
z

+ 2k} =
=

2i

given by

{(x

2(z

- 1)

1)1

+ (j/ + l)i + (z- l)k} {31 + 16j + 2k} =


=
or
Sx + 16y + 2z = -11

can also be obtained by using equations

(7)

and

(10)

respectively on

APPLICATIONS OP PARTIAL DERIVATIVES

168

7.

[CHAP. 8

Establish equation (10), Page 162.


Suppose the curve
G{x, y,z)

is

defined

by the intersection of two surfaces whose equations are F{x, y,


and G continuously differentiable.

z)

0,

where we assume

The normals to each surface at point P are given respectively by Ni = VF\p and N2 = VG|p.
Then a tangent vector to the curve at P is To = Ni X N2 = Vif'|p X VG|p. Thus the equation of the
normal plane is (r ro) To = 0. Now

To

juation

V/^|p

Fyj

Fzk)

(G.i

Gyj

Fz F,

Gz G,

Gzk)}

|p

F,

Gx Gy

is

Fy F,

Fy Fz
Gy G.

Fy Fz

G. Gy G.

- ro)

{{FA

F:z

(r

VF|p X VG|p

{x

""'

ajo)

Gy Gz

Fz Fx

(y-yo)

Gz Gx

Fx Fy

(z

Gx Gy

-zo)

ENVELOPES
8.

Prove that the envelope of the family


solving simultaneously the equations

Assume parametric equations

<^

y, a)

<j){x,

and

of the envelope to be

<^^

f(a),

9(a).

and so upon differentiating with respect to a [assuming that


derivatives], we have
,
,,
identically,

The

slope of

0x

any member of the family

The
slope
""" "^"^^

<t>{x,

{x, y)
of the
envelope ^
at v^,
is
""^ ci.,civ,i/c
"^
/ io

= Q at {x, y) is
_ dy/da _ g'(a)
_

y,a)

dy
^^

dxida
envelope and a member of the family are tangent, we must have
(t>y'

Comparing

(a)

(2)

with

(1)

we

0x/'()

/'(a)

0,

see that 00

56,

Then <p(f(a), g{a), a) =


/ and g have continuous

dy
dx

can be obtained by

if it exists,

0,

= 0.

Find the envelope of the family

given by 0^ dx

Then

at

<i>

dy

or

any point where the

f'(a)'

0 -'()

(2)

and the required result follows

x sin a

+ y cos a =

1.

(6)

Illustrate the results

geometrically.
(o)

By Problem 8 the envelope, if it exists, is obtained by


solving simultaneously the equations 0(a;, y,a) = x sin a +
and 0 (x, y,a) = x cos a y cos a = 0.
y sin a 1 =
From these equations we find x = sin a, y = cos a or
x^

(6)

1.

The given family is a family of straight lines, some members


of which are indicated in Fig. 8-3. The envelope is the circle
x^

10.

+ y' =

+ y'^ =

1.

Fig. 8-3

Find the envelope of the family of surfaces z = 2ax- o?y.


By a generalization of Problem 8 the required envelope, if it exists,

is

simultaneously the equations


(1)

From

(2)

x/y. Then

2ax

a^y

substitution in

2
(1)

and

yields x^

(2)

2x

2ay

yz, the required envelope.

obtained by solving

CHAP.

11.

APPLICATIONS OF PARTIAL DERIVATIVES

8]

Find the envelope of the two parameter family of surfaces


The envelope

of the family

F = 0,

between the equations

F ^
Then p

x, a

ax

= y and we

i^
I3y

have

F(x, y,

= 0,

F^

z, a, /3)

+ aP =
z = xy.

ax

+ py ap.
/?

Now

43).

F = -x +

0,

obtained by eliminating a and

if it exists, is

0,

Problem

(see

169

/3

= -y +

F^

0,

DIRECTIONAL DERIVATIVES
12.

Find the directional derivative of F = x^yz^ along the curve x


z = u cos u at the point P where u 0.
The point

corresponding to m

VF =
A

is (1, 1,

2xyzH

tangent vector to the curve

du^^"'^

-e""i

"*"

and the unit tangent vector

Sx^yz^s.

^*""

^^

cosMJ

"*"

^^^

1,

3k at

-i

2j

(1

in this direction is

To

cosM)k}

sinM)k

~'

k atP

Directional derivative

13.

-2i

("

Then

Since this

2 sinw

Then

1).

x^z^j

e~",

IS

du

is

positive,

is

VF'To

(-21

- j + 3k)

(^

~'

+ ^i + '^ \

_A

i^

increasing in this direction.

Prove that the greatest rate of change of F, i.e. the maximum directional derivative,
takes place in the direction of, and has the magnitude of, the vector V^-

~ ^^*ds

'ds

when

'^

projection of

VF

and di/ds have the same


direction of VF, and the magnitude

VF

This projection

Then the maximum value

direction.
is

~.

in the direction

is

maximum

of dF/ds takes place in the

\VF\.

Find the directional derivative of U = 2x^y-3y^z at P(l,2,-1) in a direction


toward Q(3, -1, 5). (6) In what direction from P is the directional derivative a maximum? (c) What is the magnitude of the maximum directional derivative?

14. (a)

(a)

VC7

ex^yi

{2x^

- 6yz)i -

The vector from

to

The unit vector from

+ 14] = (3 - 1)1 +

= T =

Sy'k

to

121

i.e.

is

P =

(121

(-1
21

- 2)j +

- 3j +

[5

- (-l)]k =

6k

VW+TW+W

Then
Directional derivative at

12k at P.

14j

- 12k)

21

21

3j

6k.

- 3j + 6k
'^

^i

^^

= _ 90

decreasing in this direction.

(6)

From Problem

(")

F rom Problem

13, the directional derivative is

1 3,

Vl44 + 196 + 144 =

the
22.

value of the

maximum

maximum

in the direction 121

directional derivative is

|121

14j

- 12k.

+ 14j 12k|

APPLICATIONS OF PARTIAL DERIVATIVES

170

DIFFERENTIATION UNDER
15.

sign.

Then

f(x, a) dx.

f(x, a

+ ^a)dx -

f(x, a)

+ Aa) da; +

+ ^a)dx +

f(x, a

/(a, a

X2<a) /(,

*^2(a)

dx

a)

^i< + Aa)

^2<a + Aa)

X2(a>

[f(x,

+ Aa) -

/(, a)] dsc

the

f(x, a

jCa)

By

+ ha)dx -

f{x, a

+ Aa) d

-^niCa)

^2<a)

mean value theorems

+ Aa) diB

-^"ifa)

i<a+Aa)

dx

-',()

-i(a + Aa)

INTEGRAL SIGN

the

Prove Leibnitz's rule for differentiating under the integral

/(a;,

[CHAP.

we have

for integrals,

^2<'

,2<a'

+ Aa) -

[/(a;,

/(C, a)]

dx

Aa

/ {x,

|) da;

(1)

^"i (a)
(a + Aa)

Xwj

f{x, a

+ Aa) dir

/(^j,

+ Aa)[Wi (a + Aa) -

-Ui (a)]

(2)

f(x,a

+ ^a)dx

/(^2,

+ Aa)[M2(a + Aa) -

W2(a)]

()

ta + Aa)

X"2
-2 (a)
where

(,

is

+ Aa,

between a and a

^i is

between ui{a) and Mi(a

+ Aa)

and

C2 is

between

M2(a)

and M2(a + Aa).

Then

^
Taking the

limit as

derivatives,

we

Aa

^(a)

By

^) ''^

/ ('''

/(-

^)

^-

^(^- "

^yA<^,a)dx + f[u.{a),a]^-

= f'^^^^dx,

find

f[u,(a),a]^

where ^0.

i>'{a)

Leibnitz's rule,
/

_=..

2\

J
2,

;
<
sin (g

cos
^

r_^
-

By

cos

=
a;

^(a)

X'^IT^:^

2 sin a'

>1
=

a \) J
d

da

()

sin a^

sin a'

find

= (^^^T)^

ft

3 sin a

,
cos a;)^
Jo (2- tl^,2

f'^^ ^
COS
W/

2 sin a"

i/o

'T'^-

2 sin a

L,

_^^
-/^alTi'

Leibnitz's rule, if

aa; da;

fit

sinaa; ["

Jo

obtain

^a^

17. If

^>

assumed to have continuous

of the fact that the functions are

making use

-* 0,

g
16. If

'"'

J(a^-l)-'/^

from which

(See

then

f(2-z^

Problem

62,

Chapter

5.)

CHAP.

APPLICATIONS OF PARTIAL DERIVATIVES

8]

INTEGRATION UNDER
18.

Prove the result


Consider

By

163, for integration under the integral sign.

1)1

tp(a)

(1)

/(^>a)

da fdx

Leibnitz's rule,

'P'M

Then by

<f>{a)

from

(1),

Putting a

= 6,

we have

Jj

,p{a)

(2)

f{x,a)daldx

s^ij

integration,

Since

19.

INTEGRAL SIGN

the

Page

{18),

171

0(a)

da

f{x,a)dxidx

<p(a)

Thus from

in (2).

f(x,a)dx

(1)

and

(2)

with c

'/(,) d*

= 0, we

find

the required result follows.

f U'-^^^)dx
\a-cosxJ

Prove that

Jo

From Problem

62,

Chapter

'

^^

5,

J^a ++ 'fzl
]
V^^F^J

cos X

'^

-J a'

>

a,b>

ii

1.

1.

Integrating the left side with respect to a from a to 6 yields

rlr^i'^^

riy &-eosa, V

In (a -COS..) 'do;

Integrating the right side with respect to a from a to 6 yields

and the required result

MAXIMA
20.

TT

In (a

1)

'b

TT

ln(

+ Vb^
)

follows.

MINIMA

and

Prove that a necessary condition for

minimum)

+ Va -

at {xo,yo)

that

is

f{x, y) to

fx{xo,yo)

0,

have a relative extremum (maximum or

fy{xo,yo)

0.

be an extreme value for f(x, y), then it must be an extreme value for both f(x, y)
But a necessary condition that these have extreme values at a; = ajo and 2/ = j/o respectively

If f(xo, yo) is to

and
is

21.

Let
in

f{xt y).

fx{xo,yo)

f{x, y)

0, f,{x:,,yo)

(using results for functions of one variable).

be continuous and have continuous partial derivatives, of order two at least,


% including the point {xo,yo). Prove that a sufficient condition that
a relative maximum is that A = fxx{xo,yo)fyy{xo,yo) - fxy{xo,yo) >
and

some region

fixo.yo) is
fxx{xo,yo)

By

<

0.

Taylor's theorem of the


/(ko

mean

(see

+ h,yo + k) -

Page

109),

f{xo, yo)

using fAxo,yo)

f(xo

+ h,yo + k) -

(1)

we

/(xo, yo)

0, /,(a;o,3/o)

^{h^U + 2hkf^ +

where the second derivatives on the right are evaluated at


pleting the square on the right of

Xo

+ eh,

yo

0,

we have

k'fyy)

+ ek

where

(1)

< 9 < 1. On

com-

find

luUh + ^fc Y + ("hhfJkyA

(2)

APPLICATIONS OF PARTIAL DERIVATIVES

172

Now
in braces

small h and

all sufficiently

we can

Similarly

22.

+ h,y(, + k) s

But

k.

2/o)

establish sufficient conditions for a relative

Sx^ S

f(xo,

this states that f(xo, yo) is a relative

Find the relative maxima and minima of


f^

by hypothesis there is a neighborhood of (xo, 2/o) such that /xx < 0. Also the sum of the terms
by hypothesis. Thus it follows that
must be positive, since f^^Jyy fly >
/(xo

for

[CHAP.

when x

1,

fy

f{x, y)

Sy" 12 =

x^

maximum.

minimum.

+ y^-Sx- 12y + 20.

when y = 2.

Then

critical points are

P(l,2)' Q(-l,2), B(l,-2), S(-l,-2).

/x.

At

6a;,

A>0

P{1, 2),

and

At Q( 1, 2), A <
At

R(l, 2),

occurring at

23.

fly

is

a relative

hence

0;

is

neither a relative

is

neither a relative

and

<

/xx (or/jj)

so

maximum
maximum
is

36xy.

at

point.

or

minimum

or

minimum

point.

maximum

point.

a relative

minimum value of f(x, y) occurring


Points Q and R are saddle points.

minimum

is 2,

point.

maximum

while the relative

rectangular box, open at the top, is to have a volume of 32 cubic feet.


be the dimensions so that the total surface is a minimum?
and

are the edges (see Fig.

(1)

Volume

(2)

Surface area

or, since z

32/ xy

from

^ = y-^
Dividing equations
a;

then

= V = xyz = 32
of box S xy +

2yz

'.^-

2xz

x^+Mx

= = 4, A =
2/

lows that the dimensions

64

"

when

(S)

J'-

^
^

(1),

y^

of box

For

8-4),

value

What must

If X, y

and

(i),

S..Syy

4 ft

Fig.8.4

(3)

we
Sly

x'y
find

= M,
y

s,o

that x^

(^Y^V^

x-^
= 64
>

X 4 ft X 2 ft give the minimum

LAGRANGE MULTIPLIERS
24.

ffyy

A>0

is 38.

>

/xx {or fyy)

and

relative

A =

Then

0.

and Q

A<

At S{l,2),
Thus the

/^

6j/,

for

MAXIMA

and

or x

when

(.4)

= y A and

^^^ Sxx

Xi/^

= 64

= 2.

^>

0.

Hence

it fol-

surface.

MINIMA

Prove that a
Consider F{x,y,z) subject to the constraint condition G{x,y,z)-(i.
FyGx = 0.
that
Gy
extreme
value
is
Fx
F{x,
z)
have
an
necessary condition that
y,
G{x,y,z) 0, we can consider z as a function of x and y, say z f{x,y). A necessary
F[x,y, f{x,y)] have an extreme value is that the partial derivatives with respect to
X and y be zero. This gives
Since

condition that

{1)

Since G(x, y,

z)

0,

we

also
(3)

From

(1)

and

Then from

(3)

(5)

The above

we have
(6) we

and

(5)

find

Px

F=Z.

(2)

Fy

-(-

F^Zy

G.Zx

Gy

G.Zy

(3)

and

have
Gx

F^G^-F^G^ = 0, and from


F^Gy-FyGx = 0.

results hold only if F^ ?^ 0, G.

^ 0.

(4)

we have

(6)

FyG^-F^Gy =

0.

^^^^-

APPLICATIONS OF PARTIAL DERIVATIVES

^J

yields"

where^^/.

26.

"

/:G,-irfG^=^;."

= o!'?i'o'

"

'^'^'^'

+ ^^" =

^='

Elimination of X between these equations

"

^'^^^^^^^ multiplier. If desired

"' '*""

173

we can

consider equivalently

= xF + G

Find the shortest distance from the origin


to the hyperbola x^ + 8xy + ly- =
225, z = 0.
We must find the minimum value of x^ + y" (the sauarp of tlio Hic+or, f^
'' *^' "^'" * "'^^
point in the xy plane) subject to the
constraint .^ + 8.7+ 7,^1 225
^^^^According to the method of Lagrange multipliers,
we consider ^ = x^ + 80=2/ + ly^ - 225 + X(x^

^"

i.x.

+ y^).

=
-

*'
^,

Prom

{!)

and

(2),

8x

since

82/

142/

(a;, 2/)

+1

+
+

=
=

2X0;
2Xj/

or

{1)

(x

or

{2)

4a;

+ l)x + % =
+ (X + 7)j/ =

we must have

(0,0),

^o^Uich

+
+

2a;

+7

:o' Jiriu!,:^eits^

'^-

"'

= "''

^"'

^'

+ 8X-9 =

^"'^^^^^*^''" ^"

^^

or

+ 8.. + 7.^ =

1,-9

225

yieMs -5.^

= 225,

^^+8.2. + 72, = 225 yields 45.^ = 225.


''^:L\^%:l^J::Zti ^nit
^ l^f/-%r^?:Th^"
^ ^ - 2^- Thus the required shortest distance is ^25 =
'^

5.

Find the maximum and minimum


values of x^
conditions a;V4 + .,75 + ^725 = 1
^ -

27. (a)

tion of the result in


()

We^mu^t

5+25

find the

1-0

and

and

- ^T+T
+ ^.

(a).

F =

extrema of

^,

+ j/-^ =

.^
o.

+ ^^ + ,.

+ v^ + t?

<?nhiPf.+ tr.
+i,.
t^^/.^e

.object to the constraint conditions

In this case

we use two Lagrange

,,

multipliers X, X,

consider the function

Taking the partial derivatives of


G.

2.+^4-X.

Solving these equations for

0,

x, y, z,

G,

we

find

with respect to .,y,. and setting


them equal to zero,
22,

= ^2X2_
Xi + 4

2^+x.

= O.a =

-5X2

2Xi

10

'

Prom

the second constraint condition,


x + v-z - O
different from zero (this is Justified
since otLerwIse we would
satisfy the first constraint conHit.ioriV ti,. ^.=,,1.
condition), the result

_^_ + ^i__
+4
2X1
+

Xi

Multiplying both sides by

2(X,

17X?

from which

X^

= -10

+ 4)(X, +
245X,

or -75/17.

5)(X.

750

+ 25)

_25_

10 "^2X1

50"

_
-

2.

(X,

%-x.

we

J
Ld

find

(.)

25X2
2X:

+ 50

(^)

,
havet
a'-O, 2/-O,
T" ;j-0 which
wh" I"would
'lr1
not
j-

^^+.

S0

'^

and simplifying yields


or

-.
^
.constraint

(M
Pi,"
(&) Give a geometric mterpreta-

10)(17X,

75)

174

APPLICATIONS OF PARTIAL DERIVATIVES

Cose

(2),

^X^,

yields X|

1,

^X^' ^

Substituting in the

f^a-

\=

or

180/19

6\/57l9.

The value

2/V5

constraint condition,

two

a;^/4

+ y'/5 +

critical points

(-2\/57l9, -3-/6719, -5\/57l9)

corresponding to these critical points

+ 45 + 126)/19 =

(20

is

10.

\ = -75/17.

2:

From

+ y^ + z^

of x'

first

Tliis gives the

(2\/57l9, 3\/57l9, 5V^7l9),

Case

\ = -10.

1:

From
2725

[CHAP.

a;=^X2>

(2),

z^/25

~^2> ^ ^^i- Substituting


= 140/(17\/646) which gives

2/

1,

of

x''

+ y^ + z^

corresponding to these

Thus the required maximum value

(1600

is

+ 1225 + 25)/646 =

minimum

10 and the

is

the first constraint condition,

a;V4

(-40/V646, 35/V646, -5/V646)

(40/V646, -35/V646, 5/V646),

The value

in

the critical points

yields Xj

value

is

75/17.

75/17.

+ y^ + z^ represents the square of the distance of (x, y, z) from the origin (0, 0, 0), the
equivalent to determining the largest and smallest distances from the origin to the curve
of intersection of the ellipsoid a;V4 + 2/^/5 + z^/25 = 1 and the plane z x + y. Since this curve is
Since

(6)

problem

x''

is

an ellipse, we have the interpretation that


semi-minor axes of this ellipse.

and y/lB/ll are the lengths of the semi-major and

-v/lO

The fact that the maximum and minimum values happen to be given by Xj in both Case 1 and
It follows, in fact, on multiplying equations (!) by x, y and z in
is more than a coincidence.

Case 2

succession and adding, for

2x'

^+
+

X'

i.e.

Then using the constraint


For a generalization

APPLICATIONS
28.

The period T
(a)

if
(o)

error and

y^

\,x

obtain

^^ +

22/^

+ K\j
^\i +

z'

conditions,

we

X,2/

2z'

+ y^

-\-

Problem

X,

K(x + y-z)

5 +25^*
^+i^J+
x'

find

of this problem, see

= Xj.

z^

76.

ERRORS

to

pendulum of length I is given hy T = 2ir^/T/g. Find the


made in computing T by using J = 2 ft and g = 32 ft/sec^,
= 1.95 ft and g = 32.2 ft/sec^?

of a simple
(b)

percent error

the true values are

T =

we then

Then

2-n-l"'9-'''\

= (2^g-mU-'"dl) +

dT

= -j^

{2^l'"){-U-"'dg)

dl

tt

Error

The error
have from

Error

in

in

is

in

= Ag = dg =

+0.2; error in

actually AT, which

is in this

r ^ dT = -=^=^

The value of T for

{-0.05)

ir-x

= Al = dl = 0.05

case approximately equal to dT.

= 2,

sr

-D
* error (or
/
Percent
relative error)

Another method:

Since

dT _
T ~
Note that

In

1 d;

2
(2)

L^(+0.2) =

= 32

Then the adjusted value for T

as before.

(1)

Thus we

(1),

V(2)(32)

ih\
(o)

dg
^fl
9^

ylg

T=

is

1.571

is

mi
.

rp

dT
-=r
1

2jr

aI^ "= | =

- 0.025 =

2Y

-0.025 sec (approx.)

^^'^^ ^'=

(approx.)

1.546 or 1.55 sec.

_
-^
0.025
rm l-567o.
1.0/1

-ir/128

ts
vli-

T = In 2s- + ^ In J 1
Idg _ l^ -0.05 \

--^
'-'^^

(*^^)

Inff,

l/+0.2\

2\-r-)-2\-w)

_
-

^
-^^^^''

can be written

Percent error in

T =

|-

Percent error in

Percent error in g

CHAP.

APPLICATIONS OF PARTIAL DERIVATIVES

8]

175

MISCELLANEOUS PROBLEMS
29.

Evaluate

x^
r ^x~l
dx.

In

In order to evaluate this integral,

Then by

we

resort to the following device.

Leibnitz's rule

'<'

X'i(e)-

Integrating with respect to

a,

4>{a)

x'^-

In (a

1)

In 2.

-~

X'

But

c.

+ 1).

In (a

Define

Then the value of the required integral

is <p{l)

x" dx

since 0(0)

= 0,

30.

aj

<

applicability of Leibnitz's rule can be justified here, since if we define F{x, a)


1, F(0, a) = 0, F{1, a) = a, then F(x, a) is continuous in both x and a for

finite

>

4,(0)

= (x" l)/ln x,
s ^ 1 and all
a;

0.

Find constants a and

b for

which

F{a,b)
IS

and so

The

<

{sin

{ax^

+ bx)y dx

aF/db

a minimum.
The necessary conditions for a minimum are dF/da

tiations,

we

3F

From

aa

56

these

0,

we

C^
)

a^<-sin:

(a;^

Hj^sina;

(ax^+6a;)}2da;

+ MFrfa; =

-2

-2

Performing these

0.

diflferen-

C'^
I

x""

{%yn.x

{ax"

hx))

dx

a;

{sin

(aa;^

6a;)}

dx

a;

find

a;"

da;

a;^

da:

-5IT-^a

Solving for a and

6,

we

a;^

da;

a;^

da;

.^0

a;^

a;

sin x

sin

a;

dx

da;

J(,

- -'-4

77'6

find

20

320
-0.40065,

~-^77-*

We can show that for these values, F{a,


on Page 164.
The

obtain

b) is

indeed a

1.24798

Tr"

minimum

using the sufficiency conditions

The polynomial ax^ + bx is said to be a least square approximation of sin x over the
interval (0, jr).
ideas involved here are of importance in many branches of
mathematics and their applications.

APPLICATIONS OF PARTIAL DERIVATIVES

176

[CHAP. 8

Supplementary Problems
TANGENT PLANE

NORMAL LINE

and

3L Find the equations


Ans.

(2,-4,5).

32.

(a)

x-%y-z =

(a)

SURFACE

to a

tangent plane and


5,

(6)

^^

normal

(6)

line to the surface

x^

+ y^

20

/|p(!-a;o)

and

/!,|p(?/-2/o)

at

Az

^.

If a = /(, y), prove that the equations for the tangent plane and normal line at point
are given respectively by
(a)

33.

of the

(6)

P(a;o,

^^

= I-^ =

.5--^

/1p

/Ip

~J-

Prove that the acute angle y between the 2 axis and the normal to the surface F{x, y,z)
is given by sec y = VFf+Ff+FI/\F;,\.

j/o,

at

zo)

any

point
34.

The equation

of a surface

is

P(P( 0o> ^)

A(x
where

xo

cos

A =
35-

= 2,

Ans.

^,

2/0

B(y

yo) +

C(z

zo) =

and

(a)

^^

P'^.lpSin^o,

(a)

^^

+y+z =
=

(6)

^^

y<t _
B

2o

B =

Fp|pSin 0o

-^.xlpcos

C =

<6o.

irz

p<j>

F,\p

at the point

where

check your answer work the problem using rectangular coordinates.

^1^

of the (a) tangent line

Ans.

and

p sin ^

NORMAL PLANE

0, where F is continuously
the normal line at the point

(6)

to find the equation of the tangent plane to the surface

and

The surfaces x

z)

xo _

to a

CURVE

Find the equations of the (a) tangent line and (6) normal plane
y = 4 cos 3t, z = 2 sin 5t at the point where t = v/4.
A.S.

37.

xd) +

= 57/2, z = 1. To
2x try + 2irz =

TANGENT LINE
36.

<t>,

i''p|pcos0o

Use Problem 34
p

given in cylindrical coordinates by F{p,


(a) the tangent plane and

Prove that the equations of


^^^ given respectively by

differentiable.

(6)

3.-6.-52 =

to

the space curve

and x" y' + 2z^ = 2 intersect in a space curve.


normal plane to this space curve at the point (1, 1, 1).

3a!-:^-2z

(6)

6 sin

t,

26V2.

2^,

a;

Find the equations

ENVELOPES
38.

Find the envelope of each of the following families of curves in the xy plane. In each case construct
a graph,
Ans.

39.

40.

(a)

(a) x'

= 4y;

ax
(b)

a",
x

(6)

+y =

1,

-~

a = 1.
x y = 1

Find the envelope of a family of lines having the property that the length intercepted between the
X and y axes is a constant a.
Ans. x"^ + y^'" = a^'^
Find the envelope of the family of circles having centers on the parabola y
vertex. [Hint: Let {a, a^) be any point on the parabola.]
Ans. x^ =

its

41.

Find the envelope of the normals (called an evolute) to the parabola y


Ans. %(.y-\Y = 21x^

= x' and passing


j/V(22/ + 1)

~ -J**

through

and construct a graph.

CHAP.

42.

APPLICATIONS OF PARTIAL DERIVATIVES

8]

177

Find the envelope of the following families of surfaces:

iz

(a)

= (x-y)\

-y) - aH =
= z^^2xz

a(x

(a)

Ans.

y^

(6)

- af +

(6) {x

1,

2/^

2az

43.

Prove that the envelope of the two parameter family of surfaces F(x,
y,
obtained by eliminating a and /? in the equations F = 0, F = 0, F^ = 0.

44.

Find the envelope of the two parameter families (a) z = ax + /iy z cos y = a where cosV + cos'/3 + cos'y = 1 and a is
a constant.
Ans. (a) 4z = ^ + J/^ (6) x^ + y' + z" = a'

a'

z, a,

- p'

p)

and

(6)

0,

a;

if it exists, is

cos a

2/

cos

j8

DIRECTIONAL DERIVATIVES
45.

(a)
(6)

Ans.
46.

U =

Find the directional derivative of


In

what

direction

(a) 10/3,

-2i

(6)

- z'

2xy

+ 4j - 2k,

at (2,-1,1) in a direction toward (3,1,-1).


(c) What is the value of this maximum?

maximum?

the directional derivative a

is

2V6

(c)

The temperature at any point {x,y) in the xy plane is given hy T


lOOxy/^x" + y'). (a) Find the
directional derivative at the point (2, 1) in a direction making an angle of 60
with the positive x axis.
(6) In what direction from (2, 1) would the derivative be a maximum?
(c) What is the value of this

maximum?
Ans.

12\/3-6;

(a)

47.

Prove that
point

is

4,,

z)

12v5

DIFFERENTIATION UNDER

the

cos ax^ dx,


/s
49.

50.

(a)

If

F{a)

51.

52.

53.

Given

^ -I-^,p>_i.

x" dx

Prove that

Prove that

Show

that

-^

by Leibnitz's

rule.

In (1

cos a)

In (1

2 cos

Cj^-i^
(5-3 cos

jlj^

a;

r.

a;)'

cos
4

(a'

x')

77-

x^lnx^dx =

~"

"^

ln(^ ^

a^) da;

in

(a)

1"

7'

l"i

<

H<1

'^

^^^~t, m =

1-

Discuss the case

lal

1.

2048

dxXda

sin x)

J |J
dx

{a

(a^

- a;^) da Ida;.

2ira,

prove that for

xY} dx

all

constants a and

1277

{(6-

by direct

1,2,3.....

^.

J-2V {a -

any

cos a*

INTEGRAL SIGN

the

at

2v^

- 1 In (a^ + 1)

Prove that

da;

Check the result

(6)

J^

INTEGRATION UNDER
Verify that

x^ sin ax'' da;

-'vzF

^^

find

2a tan"' a

-\

Ans.

-f""

Ans.

54.

find

integration.

J^

INTEGRAL SIGN

^
tan-'|da;,
(a)

directional derivative of

L(^\ + /'MLV

0?

maximum

continuously differentiable, the

is

(^\' +

-J

given by

a direction making an angle of 7r-tan-2 with the positive x axis, or in the

(c)

F(p,

if

(6) in

+ 2j;

direction

sin

xY

sin

2j7(6^

- a^)

6,

D.. a,,

56.

[CHAP.

APPLICATIONS OF PARTIAL DERIVATIVES

178

'

result

-^

sec

In (1

^^.

.>1

.. pr.v. that

- -^.

i COS x) da

LAGRANGE MULTIPLIERS

and MINIMA.

MAXIMA

a;

58.

^ xyH^ ^ubjec^ to the conditions


Find the maxima and minima of F(x, y, z)
1, y 2, z 3
=
x
at
108
value
maximum
Ans.
z> 0.

59.

What
a;V9

60.

a;

2/

maximum and minimum

61.

Solve Problem 23 using Lagrange multipliers.

62.

Prove that

63

(a)

is

ABC

there is a point
the intersection of the medians.
in

any triangle

Prove that the

Find the extreme values of


Ans.

maximum =

5,

such that

values of f(x,y)
(b) Can the result of

y equal

on the surface

2x'

3x^

PA' + PB^ + PC'

maximum and minimum

i xTl,0^ y Glares and respectively,


derivatives of f{x, y) with respect to x and
64.

6,

>

0,

2/

>

0,

which can be inscribed in the ellipsoid


the volume of the largest rectangular parallelepiped
Ans. 64\/3
J/V16 + 2V36 = 1?

Find the

that

is

values of x^ + y' subject to the condition


the results in (a).
of
interpretation
geometrical
(6) Give a
Ans. maximum value = 70, minimum value = 20
(a)

=
(a)

+ 4xy + Gy^ =

is

140.

a minimum and

+ xy + y' in the unit square


be obtained by settmg the partial

x'

Explam.

to zero.

+ 3y' + z' - 12xy + 4xz =

35.

minimum = 5

65.

values
case where we wish to find the extreme
Establish the method of Lagrange multipliers in the
0.
H(x,y,z)
G{x,y,z)
0,
conditions
of F{x,y,z) subject to the two constraint

66.

Prove that the shortest distance fro m the origin


and y^hx where a > 0, 6 > 0, is 3Va(&' + l)/2&-

67.

Find the volume of the

APPLICATIONS
68

69.

to

ellipsoid

\\x'

+ 9y^ +

surfaces
to the curve of intersection of the

15z^

- Axy +

lOj/z

- 20x2 =

80.

xyz

=a

Ans. 64,rA/2/3

ERRORS

0.03 inches, while its height


The diameter of a right circular cylinder is measured as 6.0
and (6) percent error made
error
possible
()
largest
as 4.0 0.02 inches. What is the
1.5%
in^
1.70
(a)
Ans.
(6)
volume?
the

is

^^^^^^^
computmg

"

^he

15.0 feet, and the included ^"eJ%^0-0sides of a triangle are measured to be 12.0 and
while the angle can be measured to with n 2%
accuracy
1%
within
to
measured
be
can
lengths
side
in determining the {a) area and (6) opposite
find the maximum error and percent error

The

accuracy,

of the triangle.

Ans.

(a)

2.501 ft^ 3.21%;

(6)

0.287

ft,

2.08%

MISCELLANEOUS PROBLEMS
70.

constants and n is a Po^i";^ i"*^^^;;


are cylindrical coordinates, a and 6 are any positive
mutually perpendicular along their
=
are
6
=
p'cosn^
and
a
prove that the surfaces psin^0
curves of intersection.
If p

and

4>

CHAP.

71.

APPLICATIONS OF PARTIAL DERIVATIVES

8]

Find an equation for the (a) tangent plane and (6) normal line
where r-1, e~ ttIA, = n-/2, (r, e,
being spherical coordinates.

179

to the surface Sre^

jr'

at the point

</,)

Ans.
72.

Ax

(a)

(77-^

477)1/

(47r

-7r'V2,

(6)

Prove that the shortest distance from the point

(a)

-4

73.

The

potential

where p

due to a charge distribution

is

Prove that

~X-dx

m> -1, M>


(a)

If

<

4ac

and

2n-/V4ac - 6^ [Hint:
ax^ + bxy + cy^ 1.]
76.

the plane

Z--V/272
77^-477

Ax + By + Cz + D =

+D
2x~iy + Qz =

cos

(r,

tf,

.p)

(b)

by

any point

is

4 cos^ e

m > 0,

if
'

Ans.

20.

directional derivative at
g

is

n>Q. Can you

extend the result to the case

-1?

fc'

'<m)

In

maximum

Prove that the

a constant.

+ 477

given in spherical coordinates

is

pVsin"

75.

Cc

Find the shortest distance from (1,2,-3) to the plane

T,

74.

77^

(a, b, c) to

Aa + Bb +
(6)

_
~

3/-\/2"/2

- 77^)2 =

>

0,

Find the

c>

prove that the area of the ellipse ax'^ + bxy + cy^ = 1 is


values of x^ + y" subject to the constraint

0,

maximum and minimum

Prove that the maximum and minimum distances from the origin to the curve of
intersection defined
by xVa' + yVb' + z'/c' = 1 and Ax + By + Cz =
can be obtained by solving for d the equation

BV

C'c'

77.

Prove that the last equation in the preceding problem always has two real solutions
df and di for any
real non-zero constants a,b,c and any real constants A,B,C (not
all zero).
Discuss the geometrical
significance of this.

78.

(a)

dx

Prove that

(6)

(C)

Find

lim Im.
"-*

79.

80.

81.

(a)

82.

Use

(a)

Find

(b)

Examine

[Hint:
(i.e.

is

+ y^ which

f(x, y)

(x^

cos X dx

is closest to

the point

(3,

-6,

prove that

2x + Ay^ SyY.

2(a^

cos'
(2

cos x

a;

1)

da
sin

a^
877

xY

8 In 2

50

a relative maximum or minimum of w = f(x, y, z).


+ y' + z' - 6xy + 8xz - lOyz for maxima and minima.
the fact that the quadratic form Aa' + B/3' + Cy' + 2Z)a/? + 2Eay +
ZFfiy >

sufficient conditions for

For

w =
(a)

x''

use

4)

In a

a cos X ^ - sin X
'""^

(a) to

of

x^

+ ay

25

p^^

Prove that

Jo
(6)

(x'

maxima and minima

Ans. minimum value

da
z

+ aj

dx

A. lim

(x'

Find the point on the paraboloid


Ans. (1,-2,5)
Investigate the

(x^

da

M_oo

This can be denoted by

f JCj~^^
+

lim

Is

J_tan-^+

+ a'^Y

(x^

positive definite) if

A >

0,

A D
D B

>

0,

A D F
D B E
F E C

>

chapter 9
Multiple Integrals

DOUBLE INTEGRALS

of
Let F{x,y) be defined in a closed region
subn
into
Subdivide
9-1).
Fig.
the xy plane (see
w. Let (|^, r,^)
regions A%^ of area aA^, fc = 1, 2,
sum
the
Form
A%^.
be some point of

F(^^,r,^)AAu
k=l

Consider
n-*oo k

limit is taken so that the number % of subdivisions increases without limit and such that the

where the

largest linear dimension of each A%^ approaches


If this limit exists it is denoted by
zero.

Jj"
and

is

F{x,y)dA

Fig. 9-1

{3)

called the double integral of F{x,y) over the region

%.

F{x,y)

is

can be proved that the limit does exist


tinuous) in %.
It

if

continuous (or sectionally con-

ITERATED INTEGRALS
of % in at most
such that any lines parallel to the y axis meet the boundary
curves ACB
the
of
equations
the
two points (as is true in Fig. 9-1), then we can write
and
where
respectively,
Mx)
=
/2(^) are
Mx)
and ADB bounding % as y = fi{x) and y
double
the
evaluate
can
we
case
this
In
single-valued and continuous in a^x^b.
grid
ot
a
constructmg
formed
by
rectangles
integral (3) by choosing the regions A%,, as
can
be
Then
areas.
corresponding
{3)
the
lines parallel to the x and y axes and aA^ as
If

is

written

^^

ff

^i>

F{x,y)

dxdy

^h<.x^

F{x,y)dydx
j

' y

(*)

= fi'.x'l

F{x,y)dyydx

i
I

constant) and finally


in braces is to be evaluated first (keeping x
a double integral
how
indicates
integrating with respect to x from a to b. The result {A)
integrals.
iterated
called
integrals
single
can be evaluated by expressing it in terms of two

where the integral

180

^^^^-

MULTIPLE INTEGRALS

9]

If

such that any lines parallel to the x axis


meet the boundary of

is

x~gx{v) and x =

g^iy)

respectively and

J J F{x,y)dxdy

we

If the double integral exists,

t Tsed^Ve
into regions

{J,)

and

in at

F{x,y)dxdy

(5) yield

g?

most

find similarly

other fornT

jgj^

(5\

F{x,y)dx\dy

the same value.

(See,

however Problem 17

^^

''^
^^^ ^ (^)' ^^^^^-^ - approp^ate miy
calloVel^^^^^^^^" ^^^r.;.a^. o/
^^^ order of integration with respect to
the

"w' 1

shown
'V ''whl5f^'
which

<2^j,q^^,

by taking the sum

in the above figure, it can generally


be subdivided
are of this type. Then the double integral
over <3?
^ is found
of the double integrals over
%^, %^, ....

TRIPLE INTEGRALS
The above

results are easily generalized to


closed regions in three dimensions
For^""" ^i^'V'-) defined in a closed three
dTmLsra"
region
Subdivide 'T'^^'.^
the region into n subregions of
volume aF., A; = -^.^,...,i.
1,2
LettS/rf
Letting
be some point in each subregion,
(4,^^, ?{

rj-;

we form

approaches infinity in such a way that the


largest
approaches zero. If this limit exists we denote
it by

linear
Hnt'r^'Lnsl^In'f"
dimension of each
"^"'k""""'
subregion

^j F{x,y,z)dV

r^rtc^lfSlfi^lif^! ^- ^-

'''

'^-^*

(7)

^-

-^-

^^

^(--)

continu-

P"'""^' ^ *^^ ^^' ^^ ^"^ ^^ Pl-"e. the


^^ITare rectangular
which
parallelepipeds
In such
integral over ^ given by
(7) as an\-.e.a*:^ S';, J"of the

'
region
is subdivided into '"f*'"^
region'risTubd?vMed^[nt
subregions

case^we can express the triple

(where the innermost integral is to be


evaluated first) 0/ the Urn' of such integrals
integration can also be performed in
any other ordei to give an equSerifresult
Extensions to higher dimensions are also
possible.

TRANSFORMATIONS

of

MULTIPLE INTEGRALS

The

[CHAP.

MULTIPLE INTEGRALS

182

there will be a set of


be curvilinear coordinates of points in a plane,
of the ^y plane
(a;, i/)
points
=
mapping
=
g{u,v)
f{u,v), y
transformation equations x
plane is mapped
xy
the
of
region
the
case
such
In
into points {u,v) of the uv plane.
then have
into a region %' of the uv plane. We
If

we

let {u,v)

Cj" F{x,tj)dxdy

G{u,v)

where

JJ

G{u,v)

d{x, y)

dudv

{9)

d{u, v)

and

F{f{u,v), g{u,v)}

du

dx
dv

dy_

dy_

du

dv

dx
djx, y)

d(u, v)

{10)

v (see Chapter 6).


the Jacobian of x and y with respect to u and
in three dimensions, there will be
Similarly if {u,v,w) are curvilinear coordinates
z^h{u,v,w) and we can
g{u,v,w),
=
= fiu,v,w), y
a set of transformation equations x
write
djx, y, z)
rnr
CCC ,
is

du dv dw

((( F(x,y,z)dxdydz = jJJ ^(^'^'^^l^^


^^^

{11)

3?.'

where

G{u,v,w) = F{f{u,v,w), g{u,v,w), h{u,v,w)}


dx
du
djx, y,z)

dy^

du

d{u,v,w)

is

dx
dv
dy
dv

and
dx

dW
dy

{12)

dw

dz_

dz

dz

du

dv

dw

and w.
the Jacobian of x, y and z with respect to u, v
and triple
and {11) correspond to change of variables for double
results

The

{9)

integrals.

made.
Generalizations to higher dimensions are easily

Solved Problems

DOUBLE INTEGRALS
1.

in the

xy plane bounded by

(a)

Sketch the region

{b)

Give a physical interpretation to

(c)

Evaluate the double integral in

jj

{x^

y^x^

= 2,

y=^l.

+ y"") dx dy.

'R.

(a)
(6)

The required region

is

(b).

shown shaded

in Fig. 9-2 below.

point (x,y) to (0,0) we can consider the


Since x^ + y^ is the square of the distance from any
(i.e. moment of inertia with respect
mertm
moment
o/
pojar
the
representing
double integral as
(assuming unit density).
to the origin) of the region
assuming
double integral as representing the mass ol the region
the
We can also consider
a density varying as x^ + y^.

CHAP.

MULTIPLE INTEGRALS

9]

183

!/

X
x

=l

a;

=2

Fig. 9-2

Method

(c)

1:

The double

Fig. 9-3

integral can be expressed as the iterated


integral
X2

dx
j=i

1006
105

The integration with respect

to y (keeping x constant) from


v = l to y = x' corresBonds
"^'*"^'
"'"^ ^^'^ ^''- '-'^- '^'^^ -bseque'nt integration wTh reTpec
corresponds to addition of contributions
from all such vertical columns

ttT? cm V-Tr^-Y
betLen"=7and ^^ri
Method

The double integral can

2:

{x'

+ y')dxdy

also be expressed as the iterated


integral

f V?

(x'
(^

+ y')
y^}dz\
dx [dy
'"'

colum^ as n

riTto

respect to

pL

S'l^*-

T^'V'-

lb

''^""

xy-

rfj/

=v

""

1006
105

"^ '" ^'^- '"' "^'^^^ '^ ^^^'^^^^ ^'^ ^ horizontal

''f''"'^

/="l

an7,-V""'

'"

for

'

^ *^^ ""^^^^^ ^^^'^^^ ^o the intersecting cylinders

Required volume

r ^3

-' = !

integration with respect to x (keeping


"-2 corrltlrV
y constant) from
>n this horizontal column. Subsequent
^^^ ^"-""-^
integration with
TfromT It
'''^'^''"
contributions

co3ns LLeen

^'

9 S

1^1

II

all

x^

such horizontal

+ y^ =

a-

and

nnt

8 times volume of region shown in Fig. 9-4

zdydx

'

+ j'

^Vaz-xz

-y

(ffl='-a;^)(;a;

16a^

Iw rarSfshadedfn
-

'^'' '^^''^ corresponds to the volume of a column


" "''^"* ^"^^ integrating with respect to y
''^'"^
fromr-0 tr
Vc.
a^^
corresponds to adding the volumes of all
such columns in a slab parallel
7
tlZe.'.luJf^
to
the yz plane, thus givmg the volume of
this slab. Finally, integrating with
respect to Tfroml""""'^"'^ '' ^'''"" '""^ ^'^ "^ ^" --^^ ^^^^^
"'^^ 4ion,'hus%tin^,
theTquiJ

sucht

volu'mT

,F^

tb

//""""l'

[CHAP.

MULTIPLE INTEGRALS

184

Find the volume of the region bounded by


z = x + y, z-Q, a; = 0, 2/ = 0, =
Required volume

volume of region shown


{6

+ y)} dydx

(x

^9 =

x=

in Fig. 9-5

- x)y -

(6

dx

hy'
j,

|(6

- xy dx

36

Fig. 9-5

{6
darkly shaded) corresponds to
In this case the volume of a typical column (shown
of. the
over th^ region
integrating
by
obtained
then
are
integration
of
limits
(x + y)} dydx. The
(obtamed
to y to y from y =
figure. Keeping x constant and integrating with respect
parallel to the yz plane
slab
in
a
columns
all
summing
=
to
=
corresponds
x + y)
from z 6 and z
= to ; = 6 corresponds to adding the volumes of all
Finally, integrating with respect to x from a;
volume.
required
such slabs and gives the

G-x ^

TRANSFORMATION
4.

DOUBLE INTEGRALS

of

Justify equation {9), Page 182, for changing


variables in a double integral.
In rectangular coordinates, the double integral
(shaded in Fig. 9-6) is

of F{x,y) over the region

jj F{x,y)dxdy.

We

= cops*^

can also evaluate this double

integral by considering a grid formed by a family of


constructed on
It and V curvilinear coordinate curves
as shown in the figure.
the region

be any point with coordinates {x,y) or


where x = f(u, v) and y = g{u, v). Then the
vector r from O to P is given hy r - xi + yj = f(u,v)i +
g(u, v)}. The tangent vectors to the coordinate curves
u = ci and v = e%, where ci and cj are constants, are
dridv and dr/Bu respectively. Then the area of region ^%

Let

(m, v),

of Fig. 9-6

is

given approximately by

But

du

\^-^

dx dy
du du

Sv

dx
dv
ler

SO that
is

k
-

is

3{x,y)

3(m, v)

d{x,y)

Am Av

\s{u, v)

sum
g{u, v)}

d(x,y)

AuAv

d(u,v)

investigation reveals that this limit

If
where

dy_

dx dy
dv dv

2 F{f{u, v),

An

dx

Su du

-r-lilUikV
dv\

the limit of the

taken over the entire region %.

Am.Ai>.

\du

The double integral

dy
dv

Fig. 9-6

is

P{i(n,v),g{u,v)}\^^\dudv

the region in the nv plane into which the region

is

mapped under

the transformation

y = g{u,v).
^ ,.
^
i
of variables makes use of Ime integrals
Another method of justifying the above method of change
Problem 32).
and Green's theorem in the plane (see Chapter 10,

f{u,v),

CHAP.

li

5.

MULTIPLE INTEGRALS

9]

u~

x^-y^ and v = 2xy,

From

d(u, v)

__

9{x, y)

the identity

{x^

(a;2

Then by Problem

^ yy =

Chapter

{:.2

v}

y)
d{u, v)

y,

- 2/^)^ +

^-

i^2xyY

2a;

-22/

22/

2a;

terms of u and

v.

we have

and

v^

in

a;^

= V^?T^

2/^

6,

^(a;,

Anothe^r .ethod:

d{x,y)/3(u,v)

Ux Uy

+ 2/T =

45,

find

185

d{u, v)/d(x, y)

4(a;2

+ y^)

4Vm'

i;^

Solve the .iven equations for


. and y in terms of . and .
and find the Jacobian

xy-i, xy =

1.

assuming unit density.

Fig. 9-7

Under the transformation x'-y''

R,l-l polar

= u

..,., i.

2a;i/

JJ
%P
4/

The .,..,. ,,.,

7.

Evaluate
a;2

+ 2/2 =

(see

+ y^ =

%'

4v';7T^

,,|&M|,,,

OKU, v)

/-

dw

di;

4j^^J^^^

5.

i.

,iV"wr :r<.'r """ '

the region in the . piane


bounded by

^^^^^'^^'

t\t ra:2rt:tirtf
*'^^"'*^*'" ***^ ^^'

[Fig. 9-8(5) below]

fr,.. +
I"/

9.

pilCrctt:^;^"Sr

into the region

,ere

+ ,,,,,,

. ,. _^.^,^ r."j je"

J/v'J^..,,,
4 and x^

,..

__dudv_
^^ V^JTV'

=
where we have used the results
of Problem

t>, ,..

r'^-^
[Pi^- 9-8()

os..
below]

2.

is

. sin,

mapped

[CHAP.

MULTIPLE INTEGRALS

186

Itt-

ir~

:m:
1

!I

(6)

Fig. 9-8

S{x,y)

Since

follows that

it

d{p,4.)

S{x,y)\
I

y/x"

dpd</>

+ y^ dx dy

\j p-p dp

d,p

"K

19

d(^

38ff

immediately on observing the region %, smce


can also write the integration limits for
integration
=
3 within the sector shown dashed in Fig. 9-8(a). An
for fixed 0, p varies from p = 2 to p
Geometrically
all sectors.
from
contribution
the
=
then
gives
=
to
2r
from
with respect to

We

p dp

8.

d0 represents the area dA as shown

in Fig. 9-8(a).

Find the area of the region in the xy plane


bounded by the lemniscate p^ = o? cos 2^.
given directly in polar
various values to
assigning
coordinates (p,0).
and finding corresponding values of p we obtain
the graph shown in Fig. 9-9. The required area

Here the curve

is

By

(making use of symmetry)

Fig. 9-9

is
O'
laVcos

,77/4
,77/4

pdpd<i,

2c()

d0
2

a?-

cos 20 d0

a^ sin

20

f/V

- dtdy

X77/4

TRIPLE INTEGRALS
9.

(a)

Sketch

the

bounded by x
2/

(5)

= 0,

-I-

dimensional region %.
a; = 0,
1/ + 2 = a (a > 0),

= 0.

Give a physical interpretation to


0.

\\\

[x^

J-

1/

+ if + z^) dx dy dz

(c)

Evaluate the triple integral in

(a)

The required region

(6)

{x, y,
Since ^ + j/^ + z^ is the square of the distance from any point
inertia
the triple integral as representing the polar moment of

is

shown

(b).

in Fig. 9-10

Fig. 9-10
z)

to (0, 0, 0),

(i.e.

moment

we can

consider
of inertia with

CHAP.

MULTIPLE INTEGRALS

9]

respect to the origin) of the region

We

The

(assuming unit density).

can also consider the triple integral as representing the


mass of the region

density varies as x^

(c)

187

triple integral

+ y" + z^.

if

the

can be expressed as the iterated integral

(x''

y"-

z^)

dz dy dx

x^z

y^z

~x)-

x%a

dy dx

-?^+

rLv-^)2 -

x^a -x)y

+ (a-

x-y

x)y^

i<^~^)y'

+ (^_JL^yY

y^^ ^^

_yl_ (a-x-y)
ia-xY

(-x)^

g!(g^:M!

y^

(a~xV

f f x^a~xr

(a-xa
6

rdx
J

dx

20

The integration with respect to z (keeping x and


y constant) from 2 =
to z =
corresponds to summing the polar moments of
inertia (or masses) corresponding to each
cube
n a vertical column. The subsequent integration
with respect toy f rom^/ =
to y = a-x

a-x-v

' contributions from all vertical columns contained


mn'aTab
a slab narilerf'r""'^''
parallel to the yz plane. 'V^f'^
Finally, integration with respect to x from
x = Otox
^o x = a adds
up contributions from
slabs parallel to the yz plane.

all

IS

10.

Although the above integration has been accomplished


clearly possible and the final answer
should be the same.

Find the

volume and

(a)

'

.7''

centroid of the region

(b)

= 0,y = 0,y^6,z =

sta"nt

The region

(a)

9^ is

shown

Required volume

in the

order z,y,x, any other order

bounded by the parabolic cylinder


assuming the density to be a con-

in Fig. 9-11.

fffda; dy dz

(4

dzdydx

a;2) dy dy.

*^ X = (i *^y =

(4

(24-6x2)

dx

x^)y

da;

32
Fig. 9-11

oJ,.X=

"d^'^ydx

32. by part

(a),

since

<r

is

constant.

24(7

Total mass

Total mass

32(7

Then

[CHAP.

MULTIPLE INTEGRALS

188

moment about xz

Total

Total

_
-

_
"

32<j

azdzdydx

32a

Total mass

Total mass

Thus the centroid has coordinates

960-

Total mass

Total mass

moment about xy
^""plane

ay dz dy dx

(3/4,3,8/5).

because of symmetry.
Note that the value for y could have been predicted

TRANSFORMATION
11.

of

Justify equation (11),

TRIPLE INTEGRALS
Page

182, for

changing variables

in a triple integral.

Fig. 9-12

surfaces which subconstruct a grid of curvilinear coordinate


9-1^).
Fig.
(see
is
which
L'=K
of
one
into subregions, a typical

By analogy with Problem


divide the region

The vector

"R.

we

4,

from the origin O


r

==

xi

3/j

to point

zk

is

/(u,i;,'M;)i

assuming that the transformation equations are

fl'(w,-i',w)J

x^f{u,v,w), y

'l(^t,'y,')k

9(u,v,w)

and

h(u,v,w).

the intersection of pairs of coordmate

to
Tangent vectors to the coordinate curves corresponding
of the region
volume
the
Then
,rle^.
Brl^v,
ar/a,
by
surfals are gWen
approximately by
S{x,y,z)
Au At) Am

A^

of Fig. 9-12 is given

\du

The

dv

d{u,v,w)

dw\

region is the limit of the


triple integral of F(x, y, z) over the
d(x,y,z)

^F{f[u,v,w), g(u,v,w), h{u,v,w)} d{u,v,w)

An

investigation reveals that this limit

sum
Am. A-y

Aw

is

du dv dm

where '^

is

the region in the

uvw

mapped under the transformation.


variables in triple integrals makes use of

space into which the region

Another method for justifying the above change of


Stokes' theorem (see Problem 84, Chapter 10).

is

CHAP.

12.

MULTIPLE INTEGRALS

9]

Express

F{x, y,

J JJ

dx dy dz

z)

The transformation equations

(a)

As

in

Problem

and

d{x,y,z)/d{p,4,,z)

6,

spherical coordinates.

(6)

in cylindrical coordinates are

Chapter

39,

in (a) cylindrical

189

.x

p cos 0,

Then by Problem

p.

becomes

III

p sin 0, z

= z.

11 the triple integral

G{p,ip,z) pdpd(f,dz

%'

where %'
p

sm 0,

is

the region in the

p,

The transformation equations

(6)

0, z

space corresponding to

r cos

spherical

in

III

13.

sm

is the region in
cos 0, r sin 9 sin 0, r cos

H{r,

e,

%'
the r,e,<p

where %'
(r

and where

G(p, 0, )

coordinates

are

r sin

cos 0,

F(p cos

y-r sin e

0) r^ sin

Then by Problem 11 the

r^ sin .

dr de

0,

sin

d,

The volume

triple

d<p

space corresponding to

%, and where

H(r,,0)

9).

Find the volume of the region above the xy plane bounded


by the paraboloid
and the cylinder x^ + y- = a^.
cylindrical

9.

By Problem 103, Chapter 6, dix,y,z)/d(r,e,,f,)


mtegral becomes
//'/

z).

x^

+ ifl

most easily found by using

is

coordinates.

In these coordinates the


equations for the paraboloid and cylinder are re-

= p' and
Required volume

spectively z

4 times
4 X1T/2

p-a.

Then

volume shown
^a
I

in Fig. 9-13

pdzdpdtp

J-.ir/2

p^

dp

d<t,

0, l' -

or

.=0 4

fr^m

It"

11

2
Fig. 9-13

The integration with respect

oZZ

;/'

/)

to z (keeping p and
constant) from ^
p^ corresponds to
to z
by dV) in a vertical column extending from the xy plane to
""t
subsequent
mtegration with respect to p (keeping
constant) from p =
to p
a
"*"' (indicated

-.'"^u
The

ff

corresponds to addition of volumes of all columns


in the wedge shaped region.
with respect to
corresponds to adding volumes of all such wedge

Finally, integration
shaped regions.

The integration can

also be

performed

in other orders to yield the

^^ determining the region %'

"^*'''

manJ/bv
"T^^^ transformation.
mapped
by V''r'^
the cylindrical coordinate

^^'

same

result.

in p,0,z space into

moment

?^
1?
thatT'^^
the density
(a)

is

The moment of

which

is

of inertia about the z axis of the region in Problem


13, assuming
the constant <r. (6) Find the radius of gyration.

inertia about the z axis

is

^7r/2

p^

4o

p^dpd^

apdz dp

d,p

4<r

d0

va^a

[CHAP.

MULTIPLE INTEGRALS

190
The

result can be expressed in terms of the

Note that

respect to the z axis and

(b)

so that

up the integral for

in setting

of the cubical volume element,

The

-a'a

volume X density

p^

p'

JJJ

ap dz dp

of the region, since

- -3- "

^a'

'

by Problem

as being the mass


of inertia of this mass with

can think of op dz dp d^

as the

Problem

such that

13,

3^'*'^

moment

opdzdpd^, as the total moment of inertia about the

the value

is

^-

L we

d,f>,

limits of integration are determined as in

The radius of gyration

mass

z axis.

13.

MK' = |Ma%

i.e.

K'

K = a^/2/3.

or

|a^

were concentrated in a thin


is that if all the mass
The physical significance of
this shell about the axis of the
cylindrical shell of radius K, then the moment of inertia of

cylinder would be h.

z^ = a^ and
Find the volume of the region bounded above by the sphere x^ + y^ +
such that O^a^^.
below by the cone z^ sin^a = {x^ + y') cos^a, where is a constant
sphere of radius a.
(b) From the result in (a), find the volume of a

15. (a)

= a.

This

y-r sm e

sm 0,

= a and that of the cone is


In spherical coordinates the equation of the sphere is r
a; = r sm 9 cos 0,
equations
transformation
the
can be seen directly or by using
becomes, on using these equations.
z' siV a = (x' + v') cos^
2 = r cos 9. For example,

from which tan

(a)

(r^ sin^ e

r^ cos" e sin' a

I.e.,

say

r^ cos^ e sin'

= tan

and

so e

=a

or

cos^

=
e =^

r' sin^ e sin^

<p)

cos' a

r' sin' e cos' a

ir

- a.

It is sufficient to consider

one of these,

= a.

Required volume

volume (shaded)

4 times

in Fig. 9-14

^a

-,0!

J,TT/2

r' sin e

dr de

d,j>

"O'

=:

i_

^0

"

- r" sin rfr cj

d^

d4>

sine ded^

de

sin 9

-s-

^ = 0^0 =

V_ dl'

^6 =

4o' X7r/2
3
) = n

cos

e\

dtji

18=0

-(1 -

Fig. 9-14

cos a)

to r- a corconstant) from
(keeping e and
in a column
dV)
indicated
by
(such
as
elements
cubical
responds to summing the volumes of all
constant)
respect to e (keeping
with
integration
=
=
subsequent
The
a.
r
to
r
from
extending
shaped
wedge
the
in
columns
all
of
volumes
the
summing
from 6 = to e = ,7/4 corresponds to
to adding volumes of all such wedge
corresponds
to
respect
with
integration
Finally,
region.
shaped regions.

The integration with respect

(6)

Letting

a-w,

r-0

to r

the volume of the sphere thus obtained


27ra'
(1

cos

tt)

is

-^va

CHAP.

Find the centroid of the region

16. (a)
(b)
{a)

MULTIPLE INTEGRALS

9]

Use the

Problem

in

191

15.

result in (a) to find the centroid of a hemisphere.

The centroid

{x, y, z) is,

due to symmetry, given hy x

Total

= y=^0

moment about xy plane

and

fSS

zadV
/J"/ dV

Total mass

Since z

= rcosB

and a

is

constant the numerator

'''^^"*'^^'^^*

J^=<,J,=J,_/'=^*

is

^''j

sin 9 cos 9 de d<p

ad'

The denominator, obtained by multiplying the

Then

= b-/2,

P.ove that

(e,

Toa* sin^ a

la.o

by

<r,

is

|r<7a(l

- cos a).

cos a)

"

8"^^

,,,

J^' jj^'

'=^)-

= ^a.

MISCELLANEOUS PROBLEMS
...

fTraa^d

Letting a

sin^g

r
.

result of Prob. 15(a)

^Toa* sin^ a
\Traa*
a:

_
(6)

Sincos9d6id0

f^[f^

.,},.

-'o

V(a;

Jo

(x

1
,

dx\dy = -^.

dx

+ 2/r^a; + 2/y

(6)

lf

This follows at once on formally interchanging


x and y in
and then multiplying both sides by -1.

results^''
results.

(a) to

obtain f'-f f'


-'o

[-'

^ -jydxVdy = I
(x

"^^ t^at interchange in order of integration may not always


produce equal
under which the order may be interchanged is
that

A Z^m^-^
fcondition
A
sufficient

the

integral over the corresponding region


exists.

region
origin.

18.

1,

O^ySl

The mtegral

Provethat
Let

a;

I(x)

is

In this case

ff^^axdy,

fails to exist

j^ |j^

is

the

because of the discontinuity of the integrand


the
^* ^^^
^^"'^ at
actually an improper double integral
(see Chapter 12)

\fyiu)du\dt = f^x - u) Fiu) du.


=

where

doubL

F{u)dujdt,

I'(x)

J{x)

F(u)du,

(x-u)F(u)du.
J'(x)

Then

F(u)du

[CHAP.

MULTIPLE INTEGRALS

j^g2

The

The

form

result is sometimes written in the

Jr

F(x)dx'-

('

...

54)

(;r^/(^-)""^Wdr*

CF{x)dx'

(x-u)F(u)du

Problem

result can be generalized to give (see

Supplementary Problems
DOUBLE INTEGRALS
y 19.

xy plane bounded by y^ = 2x and y


constant density <t.
the polar moment of inertia of % assuming
the mass of 1i.
is
=
where
72M/35,
(c) 48<r/35
Arts. (6) f
(a,)

Sketch the region

in the

v4o.

problem.
Find the centroid of the region in the preceding

/21.

Given

j^{x + y)dxdy.

the doubl"integl-al.

Ans.

22.

(6)

Find the area of %.

(6)

(c)

Fmdj
to-

= x.

(6)

(x

(a)

Ans. x

^,

1
j

interpretation of
Sketch the region and give a possible physical

Interchange the order of integration,

+ y)dydx,

(c)

Evaluate the double in^gral.

241/60

(c)

Showthat

+ X=J.=V.""^ '
bounded by x/a + y/b + z/c =

""'

X^J^^^-'^g'^^'^^
23.

24.

25.

Find the volume^ of the'^tetrahedron


Ans. abc/Q

Find the volume of the region bounded by


Ans. 8aV3

x'

+ y\

= 0,

= -a,

Evaluate

If

is

of

= a,

= -a,

DOUBLE INTEGRALS

(T V^^T^dxdy,

where

is

the region x'

JJ e-'''+>'>

the region of Prob. 26, evaluate

+ y' S

dxdy.

a\
Ans.

<^

28.

TRANSFORMATION

27.

and the coordinate planes.

the centroid of the region


Find (a) the moment of inertia about the z axis and (6)
assuming a constant density a.
= mass; (6) * = ^ = 0, 2 = -^a'
a'a = ^Ma\ where
Ans. (a)

^/26.

By using

the transformation

+ y = u,y=uv, show

that

dy dx

x=

*^y~0

e'"'-^'^

-^

Ans. %^a^ v^

,r(l

- e""')

= a.
in

Problem 24

CHAP.

29.

30.

MULTIPLE INTEGRALS

9]

Find the area of the region bounded by xy

= 4,xy:= 8,

31.

Show thatjhe volume generated by revolving the region in


~x,y -8x,x' = y, x" = 8y about the x axis is 2797r/2.

32.

Find the area of the region

Let

[Hint:

in the first

= 5,

xy'

1,

= Q,

xy"

= 15.

= u,

Let xy

[Hint:

xy'

= v.]

the first quadrant bounded by the parabolas


[Hint: Let j/" = ux, x' = vy.\

quadrant bounded by

be the region bounded by x + y


Let x-y = u, x + y = v.]

I93

2/

= a;S

= 0. Show

J/

= 4a;S

that

a;

= j/,

a;

= 4j/.

Ans. 4

=
(( co/^^^^\dxdy
\x
"
+ yJ

-'-'

5HlI
2

'

TRIPLE INTEGRALS
^33.

Evaluate

(a)

34.

35.

J J ^^^ xyzdzdydx.

J^
^

Find the
where a,

(a)
h, c

Find the
Ans.

(a)

(a)

volume and

(6)

are positive.

moment

Mia'

centroid of the region in the first octant bounded


Ans. (a) abc/6; (b) x = a/4, y = 6/4, z = c/4

of inert ia and

+ b'^j/lO,

(b)

\/(a'

(6)

39.

40.

+ y' + z'^4, x^0.yiO,z^O,

^'

Find the volume of the region bounded by z

TRANSFORMATION

x'

+ y'

and

z^4-x'-y'

Find the centroid of the region


Ans. x = y = 0, z-^

38,

Evaluate

cone

^^^

y/x'

mtegration

= 2x.

iWe.***^*
42.

43.

Vx'

in

Problem

+ y' + z' dxdy dz,

and the xy plane.

assuming constant density


"

where

is

+ y\ (b) Give a physical interpretation of the integral in


cylindrical coordinates in the order p,z,<^.]
Ans.

JJJ

^-^-^^^___^ ^here

Tni latttiTb)'^
''

45.

is

"'"' a>b>0.ib)

Ans. 8^

= ^^^^+^

- l)/i

^'^'l **>

paraboloid z

6,

about

=^ x'

^^^^^
a

,r/2.

- = 2. cos., and
Ans. ^^a'{l- cos^ a)

+ y^ + g^ =

beW

Find the centroid of a hemispherical shell having


outer radius a and inner radius 6
constant, (6) vanes as the square of the
distance from the base.

(a) IS

Discuss the case

^' ""'^ ' symmetry:

ia)x = y^O,

|(a^- 6^)/(3_

j,).

+ y'

axis if the

Give a physical interpretation of the integral


in

'"1''
!rL?ltro'r;//f
a where 0<a<^/2. %ftfDiscuss
n"
the case
(6)

f ^'iJ^e'^'fevt^^ft'ef^

its

x'

and the

Perform the

[Hint-

(a).

the region bounded by the spheres

is

=3

the region bounded by the plane .

^^^ '''' * *^' ''^'" '"""''^"^ ^^ *^' """^


'

Evaluate

the density

if

a.

Find the moment of inertia of a right circular


cylinder of radius a and height
density is proportional to the distance from
the axis.
Ans. filfa^
(a)

Ans. ^/2

27.(2v/2

*'

+ y/h + z/c ^

TRIPLE INTEGRALS

of

Find the volume of the region bounded by

(a)

by xia

radius of gyration about the z axis of the region in


Prob 34

^Xlfox'^r'''Ans'l^'''''''''^"^''^'^^'

38.

(a).

+ b^lO

''

37.

Give a physical interpretation to the integral in

(6)

'

if

a'

(a).

by the cone

the densitv

a=b
(j)

= ^ = o.

[CHAP.

MULTIPLE INTEGRALS

194

MISCELLANEOUS PROBLEMS
46.

47.

48.

49.

Find the mass of a right circular cylinder of radius a and height


of the distance from a point on the circumference of the base.
Ans. ^Tra'bk{9a^ + 26^), where k = constant of proportionality.

b, if

the density varies as the square

Find the (a) volume and (6) centroid of the region bounded above by the sphere
and below by the plane z = b where a > 6 > 0, assuming constant density.
Ans. (a) l^(2a' - Sa'b + b'); (b) x = y^O, z = |(a + b)V(2a + 6)

x'

+ y' + z' =

a^

A sphere of radius a has a cylindrical hole of radius 6 bored from it, the axis of the cylinder coinciding
6^)''*].
with a diameter of the sphere. Show that the volume of the sphere which remains is j7r[a^ - (a" simple closed curve in a plane is revolved about an axis in the plane which does not intersect the
Prove that the volume generated is equal to the area bounded by the curve multiplied by the
distance traveled by the centroid of the area (Pappus' theorem).

curve.

50.

Use Problem 49
about the x

to find the

volume generated by revolving the

circle

51.

Find the volume of the region bounded by the hyperbolic cylinders


Ans. 64
yz = 25, yz = 49. [Hint: Let xy = u, xz = v, yz = w.]

52.

Evaluate
x^/a"

f)

x^

+ (y-bY =

a^,

>a>

Ans. 2n-V6

axis.

VT^^TxV^T^WTzV?)

+ y^/b' + zVc^ =

1.

[Hint:

Let x

dxdydz,

= au,

where

bv, z

'5^

is

= cw. Then

xy

= l,

xy=:9, xz

4:,

xz

= 36,

the region interior to the ellipsoid

use spherical coordinates.]

Ans. ^iT^abc

53.

If

'ff

is

[Hint:

the region

Let x

Prove that

+ xy + y^ ^

= w cos a
Then

in the integrand.

54.

x^

-u

let

1,

sin a,

u = ap

C' F{x)dx'^

li

prove that

= m sin a +

cos

<p,

^^w

e-<"^^+^+^'

da;% = -p(e-l).

V3

%
i>

and choose a so as to eliminate the xy term


where a and 6 are appropriately chosen.]

cos a

bp sin

(x

- u)''-^ F{u) du

for

w=

1,2,3,

..

(see Prob. 18).

chapter 10
Line Integrals, Surface Integrals

and
LINE INTEGRALS
Let C be a curve
connects

xy plane which
and B{a2, hi), (see
Let P{x,y) and Q{x,y) be single-

points

Fig. 10-1).

Theorems

Integral

in the

A{ai, bi)

valued functions defined at all points of C.


Subdivide C into n parts by choosing (n 1)
points on it given by {xi,yx), (0:2,1/2),
., (a;-i,
.

2/n-i).

Call AXfc

1,2,

= Xh- Xk-\ and

.,n where

(ai, 61)

\yk

{xo, yo), (a2, ^2)

(4,

chosen so that they are situated on


points (Xk-i, Vk-i) and

Form

{Xk, yk).

(Xi,y,)

= yk- yk-i,

and suppose that points

{xn, yn)

(k+i.ii.+i)

77^,)

are

Ol

at

C between
the

sum

Fig. 10-1

{P{ik'Vk)^^<^

(i)

Q{k'nk)^Vk)

The limit of this sum as tc ^ =0 in such a way that all the quantities A^^, Ayk
zero, if such limit exists, is called a line integral along C
and is denoted by
P{x,y)dx

+ Q{x,y)dy

or

approach

Pdx + Qdy

(2)

The limit does exist if P and Q are continuous (or sectionally continuous) at all points of C.
The value of the integral depends in general on P, Q, the particular curve C, and on the
limits (ai, 61) and (a2, 62).
In an exactly analogous
three dimensional space as

lim

{A, (4,

manner one may

Q AXk

r,,,

where Ai, A2 and A3 are functions of

define a line integral along a curve

^2(4, rj^,

A?/,

A3(4,

,;

Q AZk}

in

(3)

Aidx + Azdy + Asdz


x,

y and

z.

Other types of
example,
{xk,yk)

if

and

line integrals, depending on particular curves, can be defined.


For
ASk denotes the arc length along curve C in the above figure between points
(xk+uyk+i), then

lim
is

called the line integral of

U{ik' Vk) ASic

U{x,y) along curve C.

dimensions are possible.

195

U(x, y) ds

U)

Extensions to three (or higher)

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

196

VECTOR NOTATION

for

[CHAP. 10

LINE INTEGRALS

convenient to express a line integral in vector form as an aid in physical


For example, we can
or geometric understanding as well as for brevity of notation.
form
the
express the line integral (3) in
It is often

C Aidx + Aidy + Azdz

J^

(AJ + A2J + Aak)

fA
special
If
field is

(dxi

dy\

dzk)

(5)

dr

+ Aaj+Ask and dr = dxi + dy] + dz^k. The line integral {2)


this
with ^ = 0.
of
case
(i.e. if a
at each point {x, y, z) we associate a force F acting on an object
defined), then
^
\
F-dv

A =

where

Aii

represents physically the total

EVALUATION

of

work done

in

moving the object along the curve

is

force
(^)

C.

LINE INTEGRALS

is given as y = f{x), the line integral


curve C in the plane 2 =
the integrand to obtain the definite
=
in
dy^f'{x)dx
y
placing
f{x),
evaluated by

If the equation of a
{2) is

integral

^"2

P{x,f{x)}dx

,^^

Q{x,f{x)}f'{x)dx

(7)

which

is

then evaluated in the usual manner.

Similarly

if

is

given as

a;

g{y),

then dx

g'(y)

dy and the

line integral

becomes

^' P{g{v),v)g'{v)dy + Q{g{y),y}dy


If

is

given in parametric form

C
where h and

ta

^(),

the line integral becomes

^(t),

+ Qm),m}rit)dt

p{<i>{t),m)4>'ip)dt

denote the values of

corresponding to points

Combinations of the above methods

(8)

may

()

and

respectively.

be used in the evaluation.

along space curves.


Similar methods are used for evaluating line integrals

PROPERTIES

of

LINE INTEGRALS

of ordinary integrals.
Line integrals have properties which are analogous to those
For example:
1.

2.

P{x,ij)dx

+ Q{x,y)dy

Pdx + Qdy

J(oi.bi)

Thus reversal

=
-

j" P{x,y)dx

J'^Q{x,y)dy

Pdx + Qdy

^'(aj.bj)

line integral.
of the path of integration changes the sign of the

CHAP.

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

10]

^(03,63)
J''^2.f'2)

where

Pdx + Qdy
(as, 63) is

^(02,62)

Pdx + Qdy +

197

Pdx + Qdy

another point on C.

Similar properties hold for line integrals in space.

SIMPLE CLOSED CURVES. SIMPLY

MULTIPLY-CONNECTED REGIONS

and

simple closed curve is a closed curve which does not intersect


itself anywhere.
Mathematically, a curve in the xy plane is defined by the parametric
equations x = </,(<),
y-<P{t) where 4, and ^ are single-valued and continuous in an interval
ti^t^ti. If
^{ti) = <f,(t2) and ^(ii) = f{t2), the curve is said to be closed.
If ^(2*) = <j,(v) and f (m) = f{v}
only when m = v (except in the special case where u = ti and v^ti), the
curve is closed and
does not intersect itself and so is a simple closed curve. We shall
also assume, unless
otherwise stated, that
and ^ are piecewise differentiable in ti^tStz.
<l>

If a plane region

has the property that any closed curve in it can be continuously


shrunk to a point without leaving the region, then the region is called
simply-connected,
otherwise it is called multiply-connected (see Page 102 of Chapter
6).
As the parameter t varies from ti to ^2, the plane curve is described in a certain
sense
or direction. For curves in the xy plane, we arbitrarily describe
this direction as positive

or negative according as a person traversing the curve in this


direction with his head
pointing in the positive z direction has the region enclosed by the curve
always toward his
left or right respectively. If we look down upon a simple
closed curve in the xy plane, this
amounts to saying that traversal of the curve in the counterclockwise
direction is taken
as positive while traversal in the clockwise direction is taken
as negative.

GREEN'S THEOREM

in the

PLANE

Let P, Q, dP/dy, dQ/dx be single-valued and continuous in a


simply-connected region

bounded by a simple closed curve C.

Pdx
where

j)^

is

Then

^ Qdy

used to emphasize that

Xr(g-|V,,

closed

is

and that

it

(,,)

described in the positive

is

direction.

This theorem is also true for regions bounded by two or


multiply-connected regions). See Problem 10.

CONDITIONS
Theorem
path

for a

LINE INTEGRAL

to be

INDEPENDENT

closed curves

of the

(i

PATH

1.

necessary and sufficient condition for

joining any two given points in a region


dP^

dy

where

more

it is

J^Pdx + Qdy

is

that in

to be independent of the

dQ^

dx

supposed that these partial derivatives are continuous in


%.

(^^)

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

198

i.e.

Pdx + Qdy is an
Pdx + Qdy =

exact differential,
In such case if
d,j>.
is given by
integral
the value of the line

(11) is also the condition that


that there exists a function 4>{x,y) such that

The condition

the end points of curve

holds and

if {11)

and

{x^, 2/2),

(xi, yi)

Pdx + Qdy

In particular

are

is closed,

[CHAP. 10

d4>

we have

<i>ix2,y2)

a;i

= ir2,

1/1

(12)

<l>{xuyi)

= 1/2 and

^Pdx + Qdy For proofs and related theorems, see Problems

The

results in

Theorem

Theorem

necessary and sufficient condition for

^ Aidx
+ Azdy + Asdz

to be inde-

joining any two given points in a region

Thus we have

Ml

M2

dAs

Ml

dy

dX

dX

dz

'

M2

M3

dz

dy

is

that in

%
{U)

supposed that these partial derivatives are continuous in %.

it is

The

11-13.

can be extended to line integrals in space.

2.

pendent of the path

where

(i5)

terms of vectors.

results can be expressed concisely in

the line integral can be written

A-dr

and condition

{1I^)

If

A =

is

equivalent to the con-

Aii

A^.^

Ask,

represents a force field F which acts on an object, the result


in moving the object from one point
is equivalent to the statement that the work done
X
- 0.
if and only if
to another is independent of the path joining the two points
Such a force field is often called conservative.
dition

A =

If

0.

V A

condition
The condition {lA) [or the equivalent condition V X A = 0] is also the
i.e. that there exists a
diiferential,
exact
an
is
A-dr]
[or
Asdz
that Aidx + Aidy +
case if the endpomts
function <i,{x,y,z) such that A,dx + A^dy + A^dz = d^. In such
integral is given by
of curve C are (x^yuZi) and {x2,y2,Z2), the value of the line

In particular

if

A-dr

is

closed

d<j>

<l>{x2,y2,Z2)

^{xuyuZi)

and

A=

0,

i A'dv

we have

{16)

SURFACE INTEGRALS
Let S be a two-sided surface having
on the xy plane as in the
projection
adjoining Fig. 10-2. Assume that an
equation for S is 2 = f{x, y), where / is
single-valued and continuous for all x

and y

in

%.

Divide

of area AAp, p

% into n subregions

1,2,

{15)

.,n,

and

erect

a vertical column on each of these subregions to intersect S in an area ASp.

36/-

Fig. 10-2

CHAP.

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

10]

Let

(j>{x,y,z)

be single-valued and continuous at

all

points of S.

Form

the

199

sum

p=l

where {^^, rj^, CJ is some point of aS^. If the limit of this sum as % -> in such a way
that each ASp -^
exists, the resulting limit is called the surface integral of ^{x, y, z)
over S and is designated by

JX <j){x,y,z)dS

= [sec y^l AAp approximately, where y^ is the angle between the normal
the positive z axis, the limit of the sum (17) can be written

ASp

Since
line to

(18)

S and

JJ
The quantity

]sec y| is

>j>{x,y,z)\secy\dA

(ig)

given by

z = f{x, y)
has continuous (or sectionally continuous) derivatives
can be written in rectangular form as

Then assuming that


in

%,

{19)

SUi^.y,^)^ i+(|J+(|Jdxd,
In case the equation for

is

given as

F{x,y,z)

ran

The

0,

(21)

{21)

can also be written

dx dy

(22)

results (21) or (22) can be used to evaluate (18).

In the above we have assumed that S is such that any line parallel to the z axis
S in only one point. In case S is not of this type, we can usually subdivide S
into surfaces Si, &,
which are of this type. Then the surface integral over S is defined as the sum of the surface integrals over Si, Si,
intersects

The results stated hold when S is projected on to a region g^ of the xy plane. In


some cases it is better to project S on to the yz or xz planes. For such cases {18) can
be evaluated by appropriately modifying {21) and {22).

The

DIVERGENCE THEOREM
Let

be a closed surface bounding a region of volume Y.


Choose the outward
to the surface as the positive normal and assume that a,
p, y are the angles
which this normal makes with the positive x, y and z axes respectively. Then if Ay, A2
and A3 are continuous and have continuous partial derivatives in the region

drawn normal

also be written

jjj

i^+^ + lfy^

^j Ardydz

+ A2dzdx + A.dxdy

{2A)

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

200

A =

In vector form with


be simply written as

Aii

A2J

+ Ask and n =

jrjvAdF

cosi

cos^j

[CHAP. 10

these can

cos yk,

Jj An dS

(25)

Green's theorem in space, states


In words this theorem, called the divergence theorem or
taken over a closed
vector
that the surface integral of the normal component of a
the volume enclosed
over
taken
surface is equal to the integral of the divergence of
by the surface.

STOKES'

THEOREM

non-intersecting curve C
Let S be an open, two-sided surface bounded by a closed
if it is on one side
positive
as
to
S
normal
line
(simple closed curve). Consider a directed
side is positive is
which
of
choice
The
of
S.
side
of S, and negative if it is on the other
sense of C posior
direction
the
Call
advance.
in
arbitrary but should be decided upon
the direction
pointing
head
his
with
of
S
boundary
tive if an observer, walking on the
single-valued,
are
A,,A2,A,
if
Then
left.
his
on
surface
the
of the positive normal, has
t>,
have continuous first partial derivatives in a region of space including

continuous, and

we have

C Aidx + Aidy + Aadz = JJ

+
In vector form with
simply expressed as

A = Ad +

A23

dA

By

dz

aAi

dAi

dz

dx

(26)

cos a

aA2
dAi
^
|cos^+i-9^--g^ COSy dS
,

+ Ask and n =
=

fA-dr

dAs

cos a

cos

/8 j

this

cos y k,

(VXA)-ndS

JJ

is

(27)

the line integral of the


In words this theorem, called Stokes' theorem, states that
curve C is equal to the
closed
tangential component of a vector A taken around a simple
over any surface -S
taken
of
surface integral of the normal component of the curl
in (27), we obtain
=
X
case
V
having C as a boundary. Note that if, as a special

the result (16).

Solved Problems
LINE INTEGRALS
,(1,2)

I.

Evaluate

-y)dx +

(x^

(y^

+ x) dy

along

a straight line from

(a)

(0, 1)

to

(1, 2),

^(0,1)

(6)

straight lines

= t,

(a)

t^

from

(0,1) to (1,1)

and then from

(1,1) to (1,2),

(c)

the parabola

+ 1.

An

equation for the line joining


the line integral equals

(0, 1)

and

(1, 2)

in the

xy plane

+ 1. Then

/"*

is

{x^

-(x + 1)} dx +

{(X

+ If +

x}dx

(2x'

2x) dx

5/3

dy

= dx

and

CHAP.

(6)

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

10]

Along the straight

line

from

f=

^!(

from

line

Since

at

f^_^it'-(t'

2.

If

+ l)dy =

= l,

t=l

+ l)}dt +

at

{(t'

-2/3

10/3

(a)

x=

(b)

the straight lines from

(c)

the straight line joining

= t^

J^

dr

+ l)dy =

line integral equals

10/3

the line integral equals

(1, 2),

+ iy+t}2tdt =

points

to

(0, 0, 0)

(3a;^

(0, 0, 0)

{{3x'

J^

x^t,y = t\z = t\
J^A-dr

and the

j'\2t' + U' +

2t'

+ 2t~l)dt =

)A-dr

evaluate

f rom

(0, 0, 0)

"

to

= t^.

=
If

-2/3

8/3.

{3x^-6yz)i + {2y + Sxz)j + {l-4xyz')k,


along the following paths C:

t,

dx

{y^

A =

(1, 1, 1)

(a)

(1,1) to (1,2),
{y'

line integral equals

r (x'-l)dx =

+ a;)(0)

(l

and the

Jt

and

(0, 1)

(l-2/)(0)

= l,dy =

(1, 1),

Then the required value


(c)

ix'-l)dx

Along the straight

to

(0, 1)

201

and

- eyz)i +

and

and then

(0, 1, 1),

to

(1, 1, 1).

(1, 1, 1).

+ 3xz)j +

{2y

- 6yz) dx +

(0, 0, 0)

then to

(0, 0, 1),

+ Zxz) dy +

(2y

- 6(n{t^)} dt +

(Se

- 6t=) dt +

(4t

{2t'

{1

+ smt')} d{t') +
dt

6t=)

{St'

{dxi

+ dy} + dzk)

~ 4xyz') dz

correspond to

(1, 1, 1)

iSt'

- 4xyz^)k}

(1

and
{1

=1

respectively.

Then

- 4{t){t^)(tY} d{t')
=

12t") dt

Another method:
C, A = (3t' - 6t')i +
+ 2j + 3tk) dt. Then

Along

dr

(i

J A
(6)

dr

(St"

(2t'

+ 3t*)i +

- 6f=) dt +

(1

(4t'

- 4t')k

+ 6t') dt +

Along the straight line from (0, 0, 0) to (0, 0, 1), x = 0,y


to 1.
Then the integral over this part of the path is
{3(0)^

J
from

{2(0)

+ 3(0)(^)}0 +

{1

- 6(2/)(l)}0 +

{22/

+ 3(0)(1)} dy +

{1

dx

- 6(1)(1)} da; +

{2(1)

+ 3a;(l)}0 +

{1

A'dr

1-5

0, z

-3.

+ yj + zk =

= 0,dy =

= 1,

da;

(i

f
= 0,

ti+ tH

+ t'k,

while z varies from

dz

while y varies

is

- 4a;(l)(l)}0 =

xi

- 12") dt

- 4(0)(a/)(l)no =

'='

Adding,

- MO)(0)(z')} dz ^

Along the straight line from (0, 1, 1) to (1, 1, 1), y = l, z


to 1.
Then the integral over this part of the path is
{3a;^

(Bt'

= 0,

Along the straight line from (0, 0, 1) to (0, 1, 1), x =


to 1.
Then the integral over this part of the path
{3(0)^

from

- 6(0)(z)}0 +

and

= 1,

dy

f'
Jx=o

(
= 0,

2y dy

(3a;^

dz

while x varies

- 6) da; = -5

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

202

The straight
Then

(c)

given in parametric form by

is

(1, 1, 1)

= t,

t,

t.

XA
3.

and

line joining (0, 0, 0)

[CHAP. 10

dr

{Bf

- 6t^) dt +

+ 31")

{2t

dt

(1

- U*) dt =

6/5

Find the work done in moving a particle once around an ellipse C in the xy plane,
the ellipse has center at the origin with semi-major and semi-minor axes 4 and 3
respectively, as indicated in Fig. 10-3, and if the force field is given by
if

F
=

In the plane z

dr

dxi

(fr-dr

dyj

(3a;

0,

F =

so that the

- 4i/ + 2z)i +
(3x

- 4y)i +

work done

{{3x-'iy)i+ i4x

- 4y) dx +

(B (3a;

+ 2y)i -

[Ax

'i

sin

where

from

varies

to

- Ay'^ + z^)ii.

and

4y'k}

{3(4 cos

x:

*)

a;

4 cos

cos

t)

t}

dt

dt

{4(4 cos

(48t

t)

15 sin* t)\f

OBtt

have chosen the counterclockwise direction indicated in Fig. 10-3. We call


C has been traversed in the positive sense. If C were traversed
the clockwise (negative) direction the value of the integral would be 96jr.

Evaluate
Since

I
-'c

ds

i/ds

along the curve

y/dx^

yds

+ dy^ = Vl +

2\/x

Vl +

(y'V

given by
dx

= yl +
=

l/ dx

i/

2\/ic

GREEN'S THEOREM

in the

^/x+l dx

PLANE

from

=3

cc

to x

= ^

(x

Let the equations of the curves AEB and AFB


10-4) be y = Yi(x) and y = Y2{x)
respectively. If 9^ is the region bounded by C, we

):
'Li^E

(see adjoining Fig.

have
Fig. 10-4

dx dy

- r

P{x,v)\llyl,^,dx

P(x,Yi)dx

156

^^^^'"^^*\

straight line parallel to the coordinate axes


cuts C in at most two points.

dy

1)""

Prove Green's theorem in the plane if C is a


closed curve which has the property that any

XX

= 24.

we have

l/xdx,

5.

t j

2(3 sin f)}{3 cos t} dt

C we

In traversing

4.

3 sin

Fig. 10-3

this the positive direction, or say that


in

t,

Then the

(see Fig. 10-3).

2tt

30 sin

t i

2y) dy

4(3 sint)}{- 4 sin

(48

4 cos

+ dyj)

(dxi

line integral equals

+ yi

xi

Choose the parametric equations of the ellipse as

4y^k

{2xz

is

+ 2y)j-

(Ax

+ 2y-Bz^)j +

{ix

[P(x, y^)

P(x,Yi)dx

P(X, F,)] dx

-JPd

CHAP.

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

10]

Then

^ p dx

{1)

= -

EAF

Similarly let the equations of curves

JJ ^ dx dy

%
EBF

and

203

be a

and

Xi{y)

Xi{y) respectively.

Then

r rP"

j/S'^''''^

Q{Xuy)dy +

J"

Then

Adding

(1)

and

^ P dx

(2),

J"

^Qdy =

(2)

+ Q

^~&^A^^

dy

( [Q(X.,y)-Q(Xuy)]dy

^ Q dy

Q(X^,y)dy

CC^dxdy
(((y'^-

dx dy.

6.

Verify Green's theorem in the plane for

{2xy-x^)dx +

{x

+ y^)dy

where C is the closed curve


bounded by y = x^ and y^ x.
(0, 0)

is

of the region

The plane curves y x^ and y^ x intersect at


and (1, 1). The positive direction in traversing
as shown in Fig. 10-5.
Along y

x^,
r

Along

2/^

the line integral equals


{{2x)(x')

- x'} dx +

{x

Fig. 10-5

{xY}

(2x^

+ x^ + 2x') dx =

7/6

the line integral equals

{2{y'){y)

Then the required

- {y^} d(y') +

line integral

7/6

{y'

+ y^}dy =

17/15

{y

- 2xy)[f^^^

dx

^1 =

Hence Green's theorem

(4y'

- 2y' + 2y') dy =

-17/15

1/30.

\\ {X-2x)dxdy

7.

d(x')

f
=

(a;'

(1

- 2a;) dy dx

- 2a;="2 - a;^ + 2a;') da =

-^o

is verified.

Extend the proof of Green's theorem in the


plane given in Problem 5 to the curves C for
virhich lines parallel to the coordinate axes
may cut C in more than two points.
Consider a closed curve C such as shown in the
adjoining Fig. 10-6, in which lines parallel to the axes

may meet C in more than two points. By constructing


line ST the region is divided into two regions -^j and
%^ which are of the type considered in Problem 5 and
for which Green's theorem applies, i.e..

Fig. 10-6

1/30

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

204

STVS

Adding the

hand

left

STVS

ST

and

sides of {!)

SVTS

using the fact that

dx

have, omitting the integrand

SVT

TVS

+Q

dy in each

TVSYT

SVT

TVS

TS

dx

TS

Adding the right hand sides of (1) and


where "3^ consists of regions ^j and %2j

we

(2),

ST

Then

%2

SVTS

%i

[CHAP. 10

+ Q dy

omitting the integrand,

(2),

"*"

%,

l)(-j^

J-

and the theorem

)'^^ '^V

II

'^f^

is

proved.

TtJSrT

A region "5^ such as considered here and in Problem 5, for which any closed curve lying in
can be continuously shrunk to a point without leaving %, is called a simply-connected region. A
region which is not simply-connected is called trndtiply-connected. We have shown here that Green's
theorem in the plane applies to simply-connected regions bounded by closed curves. In Problem 10
the theorem is extended to multiply-connected regions.
For more complicated simply-connected regions
as ST, to establish the theorem.

8.

Show

may

it

be necessary to construct more

that the area bounded by a simple closed curve

\ j xdy

given by

is

such

-ydx.

P = y, Q = x. Then

In Green's theorem, put

i)

lines,

xdy ydx

T* (^'^

""

a"

jdxdy

(~2/)

dx dy

= 2A

%
where

9.

is

the required area.

Find the area of the


Area

ii

Thus

ellipse

x dy

A =

y dx

X dy

^i

a cos

6,

-|

sin^ e)

de

y dx.

b sin

6.

(a cos e){b cos e) de

(6 sin

e)( a sin

e)

de

^277

J2Tr

o-6(cos^ e

abde

vah

*^o

10.

Show

that Green's theorem in the plane

valid for a multiply-connected region

as

shown

in Fig.

is also

such

10-7.

The shaded region %, shown in the figure, is


multiply-connected since not every closed curve lying
in
can be shrunk to a point without leaving %, as is
observed by considering a curve surrounding DEFGD
for example. The boundary of "5^, which consists of the
exterior boundary AHJKLA and the interior boundary
DEFGD, is to be traversed in the positive direction, so
that a person traveling in this direction always has the
region on his left. It is seen that the positive directions
are those indicated in the adjoining figure.

O
Fig. 10-7

In order to establish the theorem, construct a line, such as


The region bounded by
so Green's theorem is valid. Then
exterior and interior boundaries.

AD,

called a cross-cut, connecting the

ADEFGDALKJHA

is

simply-connected, and

CHAP.

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

10]

ADEFGDALKJHA

But the integral on the

^'""^

of

if

OEFGD

Thus

J AD ^ "J^;

^^

9^

leaving out the integrand,

left,

205

is

the curve

DA

Ci

equal to

is

ALKJHA

ALB:JHA,

Ci, is

the curve

DEFGD

ALKJHA

DSFGD

consisting of C, and C2 (traversed in the positive directions),


then

and C

f + f

is

the boundary

and so

Jt

INDEPENDENCE
11.

of the PATH
Let P{x, y) and Q{x, y) be continuous and have continuous first
partial derivatives at
each point of a simply connected region %. Prove that
a necessary and sufficient

pdx

condition that

dP/dy

dQ/dx

where

is

identically in

Suppose 3P/dy

Sufficiency.

+ Qdy

around every closed path

in

is

dQ/dx

that

%.

dQ/dx.

Then by Green's theorem,

'^

the region bounded by C.

Necessity.

^^Pdx + Qdy =

Suppose

around every closed path

some point of %. In particular suppose

By

tammg
Green

in

"3^

and that

dP/dy

at

dQ/dx >
at the point (xo,yo).
hypothesis dP/dy and dQ/dx are continuous in
^, so that there must be some region r con{x,yo) as an interior point for which dP/dy - dQ/dx >
0.
If r is the boundary of r, then by
'
dP/dy

theorem

contradicting the hypothesis that

cannot be positive.

we can show

Similarly

identically zero,

12.

Let

and

dition that
is

that

i.e.

dQ/dx

If 3P/dy

for

aZi closed

curves in

dQ/dx -dP/dy cannot be negative, and


dQ/dx identically in %.

yPdx + Qdy
=

that

be defined as in Problem 11.

dP/dy

Sufficiency.

dP/dy

P dx + Q dy
^

Thus dQ/dx
-

follows that

Prove that a necessary and

be independent of the path in

identically in

it

??.
^

dQ/dx, then by Problem 11,

Pdx + Qdy

A-

ADBEA
(see Pig.

From

10-8).

Pdx + Q dy, we

this,

omitting for brevity the integrand

have

/^/"'
"">'
i.e.

the integral

is

"'^

ABB

independent of the path.

= ~ f " f
BBA

AEB

and so

r=r^

^"l

"^

must be

sufficient con-

joining points

%.

it

- dP/dv
/"

and

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

206

[CHAP. 10

Necessity.

and C2

If the integral is independent of the path, then for all paths Ci

X
From

X'

this it follows that the line

Problem 11 that dP/dy

13.

and

(a)

Prove that a necessary and

be as in Problem

differential of a function

(o)

is

zero,

ADBEA

any closed path

in

and hence by

11.

Pdx + Qdy

sufficient condition

(j>{x,

dP/dy

y) is that

Pdx + Qdy

we have

AEB

that
dQ/dx.

d^

^A
I

and

dQ/dx.

integral around

Let

and

in

,j>{B)

be an exact

where

^(A)

are any two points.

Necessity.

Pdx + Qdy = d^ =

If

an exact

and

(2)

^dy,
dy

dif>/dx

(1)

(x,y).

P,

*{x

+ Ax,

y)

Thus by differentiating (!)


Q.
with respect to y and x respectively,

B,/,/dy

(2)

then

differential,

^dx
+
dx

= 8Q/dx since we are assuming


tinuity of the partial derivatives.
dP/dy

con-

Sufficiency.

By
\

Prob.

12,

P dx + Q dy

is

joining two points.


points be (a, 6) and

dP/dy

if

dQ/dx,

then
(a.h)*

independent of the path


In particular, let the two
(x, y)

and define

Fig. 10-9

P dx + Qdy
(a, b)

Then

Pdx + Qdy
(a, b)

Pdx + Q(

Pdx + Qdy

Since the last integral is independent of the path joining (x,y) and (x + &.x,y), we can choose the
path to be a straight line joining these points (see Fig. 10-9) so that dy = 0. Then by the mean
value theorem for integrals,
.f.(x

+ Ax,y) -

Taking the limit as Ax


Similarly

Thus

(6)

Let

it

^f^ + Ajj/)

-* 0,

we have

we can show

B=

that

P dx ^

d<(>/dx

Q.

d(f,/dy

(xa, 2/2).

From part

Then omitting the integrand

^ (.x^,y{>

P dx

-\r

0<ff

<1

P.

Pdx

-f

dif,.

Qdy

Q dy, we have
<p{xi,y2)

'^ (.a.bl

eAx,y)

(a).

<f>{x,y)

^A

P{x

P dx + Q dy = -^dx + ~dy =

follows that

{xi,yi),

= J_ f

.f>{x,y)

^ ia.hl

<p(xi,yi)

<I>(B)

^(A)

CHAP.

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

10]

207

/-C3,4)

14. (a)

Prove that

{Qxy^

dx

y^)

{Qx^'y

Zxy^)

dy

independent of the path

is

(1.2)

joining (1,2) and (3,4).


(a)

P = 6xy^-y\ Q =
line integral is

(6)

Method

Evaluate the integral in

(6)

- 3xy\

Qx'y

Then

dP/dy

12xy

(a).

- 3y^ =

dQ/dx

and by Problem 12 the

independent of the path.

Since the line integral is independent of the path, choose any path joining
(1,2) and (3,4),
for example that consisting of lines from (1,
2) to (3, 2) [along which y = 2, dj/ = 0] and then
(3,2) to (3,4) [along which x = 3, dx = Q]. Then the required integral equals

-S)dx +

(24a;
J

Method

(54^/

- 9y^) dy =

(6xy^-y')dx

(6x^y

=
f(y)

- Sxy"-) dy =

236

Note that

We

- y^) dx +

{x^ycosx

x^^^

cos X

+
+

y^'^

(6a;V

clear that

it is

^
x^'y

way

in

Hence

- a;^/' + c)

(i(3a;V

+ c|";^J =

can be omitted.

236

See also Prob.

16,

Page

115.

could also have noted by inspection that

from which

hypocycloid

only

g{x)

3a;V-x2/'

in this evaluation the arbitrary constant c

{6xy'

Evaluate

- xy' + g{x). The


= c, a constant.

3x'y'

P =

156

2:

From {1), 4, = 3x'y^- xy' + f(y). From (2),


which these two expressions for
are equal is if
3x'y^-xy^ + c. Then by Problem 13,

15.

80

+
-

{Gxy" dx

rf(3y)

6x'y dy)
d(xy')

(y^

dx

d(3x^y^

3xy^ dy)

xy")

= 3xV xy^ + c.

<f>

+ 2xy

=
=

3xy^) dy

sinx

y^e-)dx

{x^sinx

2ye^)dy

around the

a^'^.

2xy sin x

Q ~

y'^e",

a;^

Then
bPlby = X' cosx + 2x sin a; - 2ye''
around any closed path, in particular x^'^ + y'"^

sin

=
=

a;

2ye''.

dQ/Bx,
a"^,

so that

by Problem 11 the

line integral

zero.

is

SURFACE INTEGRALS
16. If y is

the angle between the normal line to any point {x,y,z)


of a surface

positive z axis, prove that

according as the equation for


If the equation of

is

VF-k =
from which

|secy|

VTTWT4

|sec,|

F(x,y,z)

is

^"^

~^"''

Fy

'

equation

Zy,

F^

=l

is

or F{x, y,

f{x, y)

a normal to

|VF||k|cosy

F.

or

at {x,y,z)

= V^l +

z)

F|

^^

0.

VF = FA + Fyj+F.y..

is

the

Then

Fi cos y

as required.

|F.|

^In casejhe

0,

= y^'+Jf +

S and

and we

f{x,y),
find

we can write

[see y|

\/l

zl

F(x ,y,z)

z}.

z-f(x,y)

0,

from which

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

208

17.

ji

Evaluate

U{x, y,

dS where S

z)

above the xy plane and U(x, y, z)


interpretation in each case.
The required

integral

11 U{x, y, z)
where

(6) x^

(a) 1,

+ y^,

= 2-{x^ + y^)
Give a physical

(c) 3z.

equal to

is

\/l

the surface of the paraboloid z

is

equal to

is

[CHAP. 10

zl

zl

dx dy

(1)

%
S on

the projection of

is

given by x^

Since Zx

\(

y^

0.
z, = 2y,

2,

2x,

U(x, y,

z)

the xy plane

Vl +

(i)

4.T=

can be written

+ 4^ dx dy

(2)

%
(a)

If ll(x, y, z)

1, (2)

becomes

+ 4x^ + 4y^ dx

To evaluate this, transform to polar coordinates (p, 0). Then the integral becomes
/2ir

V2
I

^(l + 4pV

VTTVpdpd0 =

X2T-

Physically this could represent the surface area of S, or the mass of

(6)

U(x,y,z)

If

x^

+ y^

(2)

becomes

jj

+ y^Wl +

4a;''

p'VT+Vdpd0 =

where the integration with respect

{x'

to p is accomplished

4)/^

S assuming

unit density.

or in polar coordinates

dx dy

149)7

30

by the substitution

Physically this could represent the moment of inertia of


mass of S assuming a density = x^ + y'.

S about

Vl + 4p' -

u.

the z axis assuming unit

density, or the

(c)

U{x, y,

If

z)

f(

3z, (2)

3Vl

becomes
4x^

4y' dx dy

Jj

3{2

(x'

+ y')} Vl +

4x^

4y' dx dy

density

32, or

or in polar coordinates.
3p(2

Physically this could represent the mass of


of S about the xy plane.

111,7
(

10

S assuming a

three times the

moment

first

18,

- p^) Vl + 4p' dp

Find the surface area of a hemisphere of


radius a cut off by a cylinder having this
radius as diameter.
Equations for the hemisphere and cylinder (see
by x^ + y^ + z' = a'
(or 2 = Va" - x^- y^) and (x - a/2y + y^ = aV4 (or
Fig. 10-11 ) are given respectively

a;^ -I- j/^

ax).

Since
Zx

=
Va^

we have

x^ y^

and

2s

=
V^^^

x''

1/''

Fig. 10-11

CHAP.

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

10]

Required surface area

Two methods
Method

^/l

JJ

x''

7r/2

(f>

The

2:

is

a cos

p dp dtp

dx dy

the integral becomes

4>,

v^^

integral

tan'' e,

d(f>

1^

p=

2a Xrr/2

(1

sin

(tt

<p) d<t>

- 2)a^

equal to

is

^^9 =

Va'

=r:r dy

a;'

da;

2a

x:
^=o

2/^

this integral

becomes

tan

de

costt>

2a

(9

sec^ e

=
a;

)
^p

*^a: =

Letting

dx dy

z^

/%acos<l

=
Method

zl

of evaluation are possible.

Using polar coordinates.


+ y^ = ax in polar coordinates

Since

2a

4an J^

2a^|fltan^eC'*

2a^{:7/4

-/aa;

sm"

sin

da;

os'

Va +

da;
a;

tan^ $[

(tan 6

(sec^

- e)C^

1)

(^

de I

Note that the above integrals are actually improper and should be treated
limiting procedures (see Problem 78, Chapter 5, and also Chapter
12).

19.

Find the centroid of the surface

dS

{^'^^

^Jl/M.

SS

zVTT4^+4y^ dx dy

J J A-ndS,

Evaluate
plane 2x

A
and

+ 2y + z =

normal

to

is

V(2a;

where

A =

xyi

included in the

2y

results of Prob. 17(c) and 17(a) respec-

^ All
130

137r/3

20.

+ z-6) =

2i

x^j

first octant,

(x

+ z)k, S

and n

is

is

that portion of the

a unit normal to S.

+ 2j + k,

^ ^i2ik_^ 2i + 2i + k
3
V2' + 2^ + 1^
A-n = {xyi - a;'j + (a; + z)k} ^ Ji + ^j + k
_ 2xy - 2a;' + (x + z)
so

2xy

2x'

(a;

+ 6 - 2a; - 2y)

2xy

2x^-x-2y + Q
3

The required surface integral

is

by appropriate

_ %
" SSVl + ix^ + iy'dxdy

The numerator and denominator can be obtained from the


and we thus have

2)a^

Problem 17

in

Sf z

By symmetry, x = y~0 and

tively,

209

therefore
X'

Fig. 10-12

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

210

jj

^xy -2x^-^x-2y + 6

(2x2/
I
x=o *^y =

+ 6y)|^

(a;2/''-2x^i/-X2/-2/''

two-sided.

^^

"'dx

27/4

with surface integrals we have restricted ourselves to surfaces which are


Give an example of a surface which is not two-sided.

Take a strip of paper such as ABCD as


shown in the adjoining Pig. 10-13. Twist the
strip so that points A and B fall on D and C
respectively, as in the adjoining figure. If n is
the positive normal at point P of the surface,
we find that as n moves around the surface it

f.

AD

reverses its original direction when it reaches P


again. If we tried to color only one side of the
surface we would find the whole thing colored.
This surface, called a Moebius strip, is an example of a one-sided surface. This is sometimes
called a non-orientable surface.
A two-sided
surface is orientable.

The
22.

^^^

- 2x* - X - 2j/ + 6) % dx

J.1 =
21. In dealing

^,
^,
^,
^^ ^ ^

[CHAP. 10

Fig. 10-13

DIVERGENCE THEOREM

Prove the divergence theorem.

'>

11)

Fig. 10-14

Let S be a closed surface which is such that any line parallel to the coordinate axes cuts S in at
most two points. Assume the equations of the lower and upper portions. Si and S2, to be z = fi(x,y)
and z = f2 (x, y) respectively. Denote the projection of the surface on the xy plane by "5^. Consider
8 As

dz

dy dx

dz

%
JJ

A3{x,y,z)

dy dx

jj

[A3(x. y,fi)

As(x,y,fi)]

dydx

51

For the upper portion


angle y^ with k.

S2,

dy dx

cos y^ dSi

k na dS% since the normal

nz to S2

makes an acute

CHAP.

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

10]

For the lower portion


an obtuse angle y^ with k.

Then

- -cosy^dSi = -k-nidSi

dy dx

Si,

Ij A3(x,y,f2)dydx

A3(x,y,fi)dydx

since the normal ni to Si

makes

nz dS2

S2

'

JJ

A3 k

211

1) A3 k

ni dSi

and

A4x,y,f2)dydx

j^i^

JJ

A^ix.y,

dy dx

ft)

A3 k

JS2J

n2 dSz

+ j

A3 k

dS,

'sj

jj Ask-ndS

so that
(^)

Similarly, by projecting

Adding

and

(1), (2)

XfA3k.dS

fff't'^^

on the other coordinate planes,

{3),

ffK^-'W^ty
or

V A dy

JjV J

//(Aii + A2i+A3k)..S

=
'

I'l'

n dS

The theorem can be extended to surfaces which are such that lines parallel to the coordinate axes
meet them in more than two points. To establish this extension, subdivide the region bounded by
S
into subregions whose surfaces do satisfy this condition. The
procedure is analogous to that used in
Green's theorem for the plane.

23.

Verify the divergence theorem for A


bounded by x^O, x^l, y ^0, y^l,

We

first

evaluate

surface of the cube

Face DEFG:

JJ

A-n

dS

i,

in

xr^- dS

=
z

(2a; - z)i + x^yj= 0, z^l.

where S

is

the

dj/

xz^k

taken over the region

Fig. 10-15.

Then

x=^l.

- 2)i +

= J J
"

{(2

{2-z)dydz

z^k}

. i

'^

DEFG

Face ABCO: n

JJ

-i, x

n dS

=
=

0.

3/2

Then

j'0 j
^0
^0

^0

i-zi)

'

zdydz

(-i)

dy dz

1/2
Fig. 10-15

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

212

Face

ABEF:

= j,

= 1. Then

2/

llA-ndS=
Face OGDC:

n=-i,

{{2x

- z)i +

A-ndS =

\\

= k,

x'-j

{(2a;

dx dz

dx dz

x''

1/3

{(2a;

z)i

X2^k}

(-j)

dxdz

- l)i +

a;Vj

- xk}-kdxdy =
J

a;

da; d^/

-1/2

Then

0.

A-ndS

l\

Then

1.

jJA-ndS = J J
n = -k,

xz'Vi)

"

BCDE

Face AFGO:

Then

0.

OGDC

Face BCDE:

[CHAP. 10

- x^yj}

{2x\

(-k)

da; dj/

AFGO

Adding,

Ci"

A-ndS = f+^ +

O-l +

^.

Since

\\\ V A dy

the divergence theorem

24.

Evaluate

{2

+ x^- 2xz) dx dy dz

verified in this case.

is

dS,

where S

is

a closed surface,

By

the divergence theorem,

jjr.ndS

jjf^-^dV
V

where

25.

is

the volume enclosed by S.

Evaluate

xz^

dy dz

(x^y

z^) dz dx +

{2xy

+ y^z) dx dy

where S

is

the entire

surface of the hemispherical region bounded by z = \/a^ x^


divergence theorem (Green's theorem in space), (&) directly.
dy dz

Since

(a)

dS

cos

a,

{xz^ cos a

dz dx
{x^y

= dS

cos

/?,

z^) cos f3 +

da;

(2xy

dy

= dS

dS

where

A =

xz^i

(x'^y

z^)i +

(2xy

y^zyk

and

cos a

cos

/3 j

is

(a)

by the

n dS

cos y k,

Then by the divergence theorem the integral equals

where

drawn unit normal.

the integral can be written

cos y,

y^z) cos 7}

y^ and

the region bounded by the hemisphere and the xy plane.

the outward

CHAP.

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL


THEOREMS

10]

By

use of spherical coordinates, as in Problem


15, Chapter

XTT/2

^Tr/2

9, this

integral

213

equal to

is

y^a

r'

r^ sin

dr de

d<f,

5
If Si is the

(6)

convex surface of the hemispherical region and S^

{X'y

JJ

ffxz^dydz

- Z>) dzdx

f^'^zWa^-y'-z'dzdy (

'"''

{X'y/d'-x^-z^

+ j/^^)da;rf2/

(2a:2/

xz'dydz

JJ

jj(2a.2/

f"'

+ 2,^.)da;d2/

jj

*2

(a;'2/

- =) dz da; =

{2a;2/

+ 2,^(0)} dx d,/

(T

0,

{20=2/

addition of the above,

3/

"^

we

j/Va^

^3}

a,^

{z

f''^
d^

= 0),

then

.^^a"

- y^ ~ z' dzdy

rf^

a;^

- 2^ -

dz

da;

- j,^} ^^ ^^

0,

C^"^^

"^ x=-a

Sa

By

{-arVa'

=-a ^2 =

JJ

the base

is

2xy dy dx

^3=-Vo2-i2

obtain

z^

dz dx

.Va2-i2

-x'~y' dy dx
Since by symmetry

these integrals are equal, the result

2/Va^-a;^-y^d2/da;

12

f" C

p'

is,

on using polar coordinates,

^W

4,

V^F^ p dp

d<i>

^^

THEOREM

STOKES'
26.

all

Prove Stokes' theorem.


S

Let

be a surface which is such that its proand xz planes are regions


bounded by simple closed curves, as indicated in
jections on the xy, yz
Fig.

Assume S

10-16.

to have representation
or y-h{x,z), where f,g,h
are single-valued, continuous and differentiable
functions. We must show that
z

= f(x,y)

JJ

or x

= g{y,z)

(VX A)'ndS
=

Jj
s
j

[V X (Aii

+ A^j+Aak)] -ndS

A'dr
dx dy

where

is

the boundary of S.

Consider

first
^^

[V X

(A,i)]

n dS.

Fig. 10-16

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

214

VX

Since

Ml

A i_
A
dx dy

(Aii)

[CHAP. 10

aAi,
ay

dz

dz

Ai

If

is

f(x, y)

yj+zk =

xi

+ yj+

f{x,y)k

dy

so that

= ^'^^^ ^ ^^%^'

|^

^"*

is

a vector tangent to

is

n-j

T n-k

= dz
r n-k
dy
,

n-j

3y

to obtain

'3Ai

~ n-k ]dS

dy

dz

Now

a; i

to n, so that

(1)

(i)

taken as the equation of S, then the position vector to any point of

and thus perpendicular

Substitute in

-Mtn-k^s

= (^Mln.j

[VX(A.i)].ndS

_
^

[VX(Aii)]-ndS

on S, A,(x,y,z)

[V X

A^lx,y,f(x,y)]

-n dS

(Aii)]

Ml n'k) dS

BAi dz
^
dz By

/Ml

dy

Ml

dz^\

()

\Sy

F(x,y);

n-kdS

= -

^+^^ =

hence

|^ and

(2)

becomes

= - - dx dy

Then

rr

[V X

(Aii)]

n dS

where

is

'?i

the projection of

^ F dx

equals

where r

is

-j-dxdy

\\

S on

By

the xy plane.

the boundary of %.

Green's theorem for the plane the last integral

Since at each point

same as the value of Ai at each point (x,y,z) of C, and since dx


must have
^
^
F dx = 3>Ai.dx

is

is

the

the same for both curves,

we

{x, y)

of

r the value of

or

[V X

jj

(All)]

-ndS =

j)

Aidx

Similarly, by projections on the other coordinate planes,

rr

[V X (Ao)]

n dS

A^dy,

Thus by

[V X (Ask)]

J* J

n dS

A, dz

addition,

rr

(V X A) n dS

9 A

dr

The theorem is also valid for surfaces S which may not satisfy the restrictions imposed above.
.,Ck which
Sk with boundaries Ci, d,
For assume that S can be subdivided into surfaces Si, S2,
do satisfy the restrictions. Then Stokes' theorem holds for each such surface. Adding these surface
integrals, the total surface integral over S is obtained. Adding the corresponding line integrals over
.

Ci, Cz,

Ck, the line integral over

is

obtained.

. ,

CHAP.

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL


THEOREMS

10]

Verify Stokes' theorem for A = 3yi ~ xzj + yz'k,


where
paraboloid 2z = x^ + y^ bounded by 2 = 2 and C is its
boundary.

27.

is

215

the surface of the

The boundary C of S is a circle with equations x^ + y^ =


4,
= 2 and parametric equations x 2 cost, y = 2 smt z = 2
where s t < 2i-. Then
z

dr

J)

Zydx - xzdy +

3(2 sin t)(- 2 sin

yz' dz

t)

dt

(2

cos t)(2)(2 cos

dt

t)

'277

VXA =

Also,

+ 8cosH)dt =

(12sin^

_fl_

_a_

_a_

dx

dy

dz

207r

{z'

x)i

(z

+ 3)k
Fig. 10-17

3y xz yz^

and

xi + yj k
\/^+ y' + l

+ y'-2z)
+ 2/^-2z)|

V{x'
I

V(a;^

Then

JJ(VXA).ndS =
=

JJ(VXA).^

ffH^y

+tVl

x'

fjixz^ +

x^

+ z + S)dxdy

+ sUxdy

In polar coordinates this becomes


^27r

28.

Up

Prove that a necessary and


curve

is

Sufficiency.

that

Suppose

A=

VXA =

cos 0)(pV2)

p2 cos^

+ p72 +

sufficient condition that

j>

<

d0

2077

A-dr =

for every closed

*^c

identically.

0.

3} p dp

Then by Stokes' theorem

A'dr

(VXA)-ndS =
JJ
s

Necessity.

Suppose

_^^A.dr

around every closed path C, and assume

VXA #
at some point P.
Then assuming: VX A is continuous there will be
a region with P as an interior point, where
V X A ^ 0. Let S be a surface contained in this region whose
normal n at each point has the same
direction as V X A i.e. V X A = an where
is a positive constant. Let C be the boundary of S.
Ihen by Stokes' theorem

^A-dr = JJ(VXA).ndS = afjn-ndS


which contradicts the hypothesis that
It follows

that

VXA=

is

^A

dr

and shows that

also a necessary

and

VXA=

sufficient

dr to be independent of 4he jath joining


points Pi and Pz.

>

0.

condition

for

a line integral

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

216

29.

Sufficiency.

A=

If

VXA=VxV0 = O

V^, then

Necessity.

is

by Prob.

80,

A =

Chap.

7,

is

Page

that

A=

independent of the path joining two points which

we

VXA =

then by Prob. 28,

0,

(B

v^-

158.

dr

If

sufficient condition that

Prove that a necessary and

[CHAP. 10

A'dr

around every closed path


take as

and

(o, b, c)

(x, y, z).

and

dt

Let us define

Aidx + Aidy + A^dz

Then

+ ^x,y,z) -

<t,(x

<t>{x,y,z)

Aidx + Aidy + At

dz

Since the last integral is independent of the path joining (x,y,z) and (a: + Aa;, j/,z),
the path to be a straight line joining these points so that dy and dz are zero. Then
(a:

+ Aa!.

3/.

Taking the limit of both

we can show

Similarly

Thus

that

d(p/dy

A = Aa + A.j + A.k =

Prove that a necessary and


an exact differential, is that

30. (a)

(&)

Show

Necessity.

integrals.

d<f>/dx

B<l>ldz

Ai.

As.

gi+gj+lfk =

V^.

Aidx + Aidy + A^dz =

sufficient condition that

yxA

A =

where

If

Aidy + Aidz =

-<r

IS = ^^

Then by

differentiating

we

dAi
By
is

Aii

+ Aaj + Ask.

from which

Sufficiency.

If

_
~

aA%
dx

(^)

d0

eAi_
'

VXA =

0,

'^^'^

^^

^{x^.y^.z^)

"*"

af''^

(^)

VXA =

dAi

dAz
dy

dz

~^^''

<i,{xi,yuZi)

*^"

^'
fz =

'

_ Ms
dx

dz

0.

Aidx + A^dy + Aidz

VXA=VxV0
If

have, assuming continuity of the partial derivatives,

precisely the condition

Another method:

d<^

A^dx + A^dy + A^dz


(^)

which

Ai,

< < 1

+ eAx,y,z)

Ai{x

that in such case,

Aidx

(a)

Aidx

Aa;^0 gives

sides as

d^,

then

O.

then by Problem 29,

A=

V<f>

and

Aidx + Aidy + A,dz = A'dr = V0 dr = ff <^ +


J.(x,!(,)
(a, b. c)

choose

J,

where we have applied the law of the mean for

we can

^^^2/

A,dx + Aidy + Asdz.

+ ^d =

d0

d^,

CHAP.

LINE INTEGRALS, SURFACE INTEGRALS. INTEGRAL


THEOREMS

10]

Then omitting the integrand Aidx

31. (a)

F =

Prove that

field.

Evaluate

(6)

(2xz^

+ 6y)i +

(Sx

F- dr where

J^

+ A^dy +

A,dz, we have

~ 2yz)i +

(Bx^z^

- y^^i,

any path from

is

217

(1,

i,

-1,

a conservative force

1) to (2, 1,

-1).

(c)

Give

a physical interpretation of the results.


(a)

force field

any two

is

conservative

points.

VXF =

Since here

dy

dz

Method

From

+ 6y

6a;

independent of the path

be conservative

- 2j/z

we

x^z^

Gxy

1,

/. (y, z)

F-dr

we may

Alternatively

F-dr
from which

0,

x'z^

"

is

conservative.

3a;V-j/^

6x2/

?/^.

is

=
=

+ h (x, z)
- r^^ Az\ ~

t (^

'

..\

^^^"'^^

y^z

f, (x, y)

"

'''''"'

^^''^'^

^"^

x'z'

6xy

y'z

c\';-'\ll

15

notice by inspection that

(2xz'dx
d(x^z')

+ 3xVdz) + (&vdx +
+ d(6x2/) - d(y^z) =

6xdy)
d(a;2z3

- (2yzdv + y^dz)
+ g^^ _ ^2^ ^ ^^

determined.

2:

*^
r,rt,VnJ"'^

'"*T*^

'^

independent of the path,

J^^^(2x-6)dx +

J^^^^

(12-22/)d2/

we can

choose any path to evaluate if in

\l2z^-l)dz

15

where the first integral is obtained from


the line integral by placing y = -l z = i^,
l ay
dy
^
"
the second integral bv niacins'
'

^^ ~ n j
Y
^ " 2, .r-i
1, dx - 0, d. = 0; and the third integral
=
=
x
2,y X. dx =
=

x~2

0,

Physically

J^F-dr

dy

represents the work done in moving


an object from

In a conservative force field the

work done

is

(1,

-1,

independent of the path

Q dz
dz~a- U,
- 0;
by placing

1) to

(2, 1,

-1)

joining these

MISCELLANEOUS PROBLEMS
^^'

0.

-O

X(2,

along C.

joining

obtain respectively

These are consistent if fi{y,z) = ~y^z


+ c h(x
f = x'z^ + 6xy-y^z + c. Thus by Problem sj

(c)

VXF =

that

is

1:

these

Method

is

dx
2a;z

(6)

jj>dr

the line integral

if

necessary and sufficient condition that

"" "
^." ^^"'""^ "^^^"^^ ^ transformation which maps a region g? of
?i
i^"^'"^'
the
^
xy if
plane mto
a region %' of the z^t; plane, prove
that

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

218

J/|^ dudv

JJdxdy =
(b)

Interpret geometrically the result in


If

(a)

(assumed to be a simple closed curve)


I

(a).
is

the boundary of %, then by Problem

-^xdy

dxdy

[CHAP. 10

8,

ydx

%
Under the given transformation the integral on the right of

liO

S*)-'(*+S-)

is the mapping of
simple closed curve also).

where

By

Green's theorem

if

is

<>

IX.(^-'S>"-(l^-)*

uv plane (we suppose the mapping

in the

%'

becomes

(1)

the region in the uv plane bounded by

to be such that

C, the right

is

side of (^)

equals

dx

(x^
dv\ du

ff

Idx

dx dy\ J

dy

=SJ

CClH^jyX du dv
J J \d{u,v)

to ensure that the result is non-negative as

where we have inserted absolute value signs so as


is

dxdy.

%
we can show

In general,

Problem

(see

jjdxdy

and

%'

represent the area of region %,

du dv

J/ ||g^

(S)

%.

(6)

|M dudv

jj F{f(u, v), g{u, v)}

jj F(x, y) dx dy

83) that

the

first

expressed in rec-

tangular coordinates, the second in curvilinear coordinates

33,

Let

F =

T'^V
+

yxF.

Calculate

(a)

i F-dr

Evaluate

(b)

around any closed

-'

path and explain the results.


k

(o)

_a_

VXF

dy

dx

in

any region excluding

(0,0).

dz

-y
x^

(6)

C^F'dr =

+ y^

x'

+ y^

ydx + xdy^

.^^^

pcos4>,

j/

p sin 0,

where

(p,<p)

are polar coordinates.

Then
dx

p sin

d0

y dx +
and so

x'

For a closed curve

= 27r

ABCDA

dp cos 0,
X(

+ y'

dy

p cos

d(.arc tan

[see Fig. 10-18(o) below]

after a complete circuit back to A.

(f>

d<f,

dp sin

</>

surrounding the origin,

In this case the line integral equals

=
J

at
d<f>

A
=

and
Ztt.

CHAP.

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

10]

219

(a)

(b)

Fig. 10-18

For a closed curve


and

- 00

Q =

F = Pi +

VXF =

Qj,

to contradict those of

Problem

is

not surrounding the origin,

In this case the line integral equals

equivalent to

dP/dy

However, no contradiction

11.

x'

+ y^

exists since

P =

^ "* ^^^^ continuous derivatives throughout any region including


in

Problem

,i

= ^

<

at

0.

and the results would

BQ/dx

was assumed

34. If div

[see Fig.l0-18(6) above]

after a complete circuit back to P.

Since

seem

PQRSP

-~
+

and

and

this

x^

(0, 0),

y'

11.

denotes the divergence of a vector

field

at a point P,

show that

SSA'tidS

A =

div

lim

AF

AV-tO

aV is the volume
aF to the point P.

vi^here

ing

By

enclosed by the surface

the divergence theorem,

div

AS and

A dV =

AV

By

the limit

is

obtained by shrink-

n dS
i

AS

the mean-value theorem for integrals, the left side can be written

A iijdV =

div

where div
Then

is

div

A AV

some value intermediate between the maximum and minimum of div


/;
div

Taking the limit as

AF^O

such that

is

A =

always interior
SS

A =

throughout AV.

AV
to

AV, dlTA approaches the value div

point P; hence
div

n dS

lim
Av-^o

at

A-ndS

AV

This result can be taken as a starting point for defining the divergence
of A, and from it all the
properties may be derived including proof of the divergence
theorem. We can also use this to extend
the concept of divergence to coordinate systems other than
rectangular (see Page 142).
Physically,

ijjj A-ndSj/AV
AS

represents the flux or net outflow per unit volume of the vector

A from the surface AS. If div A is positive in the neighborhood of a point P it means that the
outflow
from P IS positive and we call P a source. Similarly, if div A
is negative in the neighborhood of P
the outflow IS really an inflow and P is called a sink. If
in a region there are no sources or sinks
then div A =
and we call A a solenoidal vector field.

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

220

[CHAP. 10

Supplementary Problems
LINE INTEGRALS
35.

Evaluate

from

lines

Ans.
36.

f'"

(x

to

(1, 1)

+ y)dx + (y- x) dy

along

and then to

(d)

(1, 2)

(6) 11,

(a) 34/3,

(c)

^ (2x - y +

Evaluate

4)

dx

the parabola y^

the curve

2t^

+ 3a; - 6) dy

(5y

Evaluate the line integral in the preceding problem around a


Ans. QAtt

38.

(a)

y =
Ans.
39.

F =

(x'

- y')i +

x'-x from

2xyi,

the point

evaluate

(1, 0)

a straight

(b)

1,

t'

+ 1-

to

J"

the xy

in

F =

(2x

where C

y) ds,

is

the curve in the xy plane given by

x'

+ y' ^

41.

If

(a)

the

(a)

23/15,

s is

the

Ans. 15

is a path
+ 2z)i - x^k. evaluate J^F-dr from (0,0,0) to (1,1,1), where C
points, (c)_the straight
curve x = t,y = t\z = t\ (h)\ straight line joining these

(b) 5/3, (c) 0,

and then

to (1,1,1),

(d)

x^z,z^y.

the curve

13/30

{d)

the unit tangent vector to a curve

is

and

25

(y

(0,0,0) to (0,1,0), then to (0,1,1)

from

Ans.

(3x-22/)i

consisting of
lines

plane given by

124/15

(a)

Evaluate

If

(0,0).

Interpret physically the result obtained.

(6)

(2, 2).

with center at

circle of radius 4

along the curve

dr

xy plane with vertices at

in the

along the shortest path.


arc length parameter, from the point (3,4) to (4,3)

40.

line, (c) straight

Ans. 12

37.

If

+t+

around a triangle

traversed in a counterclockwise direction.

(0,0), (3,0), (3,2)

= x,

32/3

(d)

14,

(4, 2),

(a)

that under appropriate conditions

J^

(plane or space curve) and

dr

F -T

where

ds

is

a given force

field,

prove

the arc length parameter.

is

J^

Interpret the result physically and geometrically.

GREEN'S THEOREM
42.

in the

PLANE. INDEPENDENCE

43.

Evaluate the

44.

(a)

Let

line integrals of

Problem 36 and

/.

d) (o-ix

Under what conditions

+ a^y + da) dx +

will the line integral

p cos 0,

P sin ^,

prove that

46;.

If

47 .

Verify Green's theorem in the plane for

J/

region enclosed by the circles

48.

,,

(a)

Prove that

(2,1).

{y'

{h)

x^

y^

(a).

h^y

+
,

63)

{x^

j
dy

+ y''' - f"= a sm' t,

x^

2/^

- 4xy^) dy
Ans.

(h)

=
is

=
C

Stg

2/

x dy - y dx

and

is

a square with

around any path

y-x^y-)dx +

{2xy -y^ + Z)dx +


f"'"
(1.0)

Evaluate the integral in

where C

-2xy) dy

Problem 37 by Green's theorem.

(6)

(b^x

Find the area bounded by the hypocycloid x'''


= a cos' t,
[Hint: Parametric equations are a;
X

- xy^) dx +

Prove that

having area A.
be any simple closed curve bounding a region

are constants,

(6)

(a)

PATH

Ans. common value

vertices at (0,0), (2,0), (2,2), (0,2).

45.

^^{x^

Verify Green's theorem in the plane for

of the

:L

xy^ dy,

16.

ti.

(61

if a., a., (H, b,, 6., b,

\a
- <h)A

Ans.

be zero?

(b) a,

b.

,,

2^.]

,^

A^

Ans. Z^a /8

and interpret.

where C

is

the boundary of the

Ans. common value

120.

independent of the path joining

(1, 0)

and

CHAP.

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

10]

Evaluate

49.

to

(0, 0)

dx

(1

= Ty'

from

problem around a parallelogram with vertices at

(0,0),

2y sin x

+ SxY) dy

along the parabola

2x

J-V4

line integral in the preceding

Ans.

(3,0), (5,2), (2,2).

51.

y' cos x)

Atis.

(ff/2, 1).

Evaluate the

50.

(2xy'

J^

221

(a) Prove that


G = (2a;" + xy - 2y') dx + (3a;" + 2xy) dy is not an exact differential. (6) Prove that
e"G/x is an exact differential of
and find 0.
(c) Find a solution of the differential equation
(2a;" + xy- 2y^) dx + (3a;" + 2xy) dy = 0.
Ans. (b) 4, = e"{x^ + 2xy) + c, (c) x' + 2xy + ce-" =

SURFACE INTEGRALS
52.

Evaluate

(a)

{x'

^j

+ y^)dS,

where S

the surface of the cone

is

z"

3(a;"

+ j/")

bounded by z

and

Interpret physically the result in

(b)

3.

Ans.

(a).

53.

Determine the surface area of the plane 2x + y + 2z = 16


(b) x = 0,y =
and a;" + 1/" = 64.
Atis. (a) 9, (6) 24j-

54.

Find the surface area of the paraboloid


Ans. |!r(5V5 - 1)

55.

56.

57.

Find the area of the surface of the cone


Ans. 6jr

a)

An^.

<

J/",

< :r/2.

(a) 2jra"(l

(a)

Find

a),

t he

<

Ans.

outside the cone

a;"

cut out by the paraboloid

+ 2/" + z" =

a;"

2/"

a"

< 7r/2.

ia(l

(a)

(6)

From

cos

a),

a;"

the result in
(b) a/2

Va;"

+ 2/" + z" -

a"

= 3,

J/".

x''

and

x'

+ z"

contained within the cone

a"

= 2,

2/".

z tan a
(c)

o".

Explain

point on the surface.

contained within the cone

obtain the centroid of the surface of a

(a)

DIVERGENCE THEOREM

A =
= 0.

Verify the divergence theorem for


2x + 2y + z = e, x = 0, y = Q, z

Evaluate

Js J

bounded by
62.

+ j/")

is

result in (a) to find the surface area of a hemisphere,


jr in the result of (a) yields the total surface area of a sphere.
(6) 2jra" (consider the limit as a -* v/2)

by
61.

3(a;"

which

2/"

cent roid of the surface of the sphere

= \/^M^,

hemisphere.

60.

cos

Determine the moment of inertia of the surface of a sphere of radius a about a


Assume a constant density a.
Ans. 2Mei", where mass M = 47ra"a

z tan a

The

a;"

x-0, y = 0,

(a)

Use the

(6)

formally placing a

cut off by

to the intersecting cylinders

Obta in the surface area of the sphere

why

59.

z"

Find the surface area of the region common


Ans. 16a"

Va;"

58.

2z

(a) djr

Evaluate

F-ndS,

2/",

jj A-ndS,

where
a;

= 0,

a;

F =

A =

where

(2xy

(z'

+ z)i +

2/"j

- x)i -

xyj

+ 32/)k taken
= 27

+ 3z)i -

{xz

and S

Szk

and the xy plane.


(2a;

(x

over the region bounded

Ans. common value

is

the

surface of the

region

Ans. 16

+ y)i +

(2/"

+ 2z)k

and S

is

the surface of the

sphere having center at


63.

(3,

-1,

Determine the value of

2)

and radius

3.

Ans.

xdy dz + y dzdx + zdx dy,

lOSn-

where

is

the

surface of the region

bounded by the cylinder


theorem,

(6) directly.

a;"

+ 2/" =

Ans. Slv

and the planes 2

and

= 3,

(a)

by using the divergence

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

222

64.

ff 4xz dy dz -

Evaluate
^ = 0, z =
Ans. 3/2

= 1,

0,*a:

65.

Prove that

66.

Prove that

67.

If

2/

= 1,

y^ dz

dx

= 1,

where S

yz dx dy,

(a) directly,

fj (V X A) n dS

the surface of the cube bounded by

is

by Green's theorem

(fe)

[CHAP. 10

a;

= 0,

in space (divergence theorem).

for any closed surface S.

ndS =

where n

0,

drawn normal

the outward

is

to

any closed surface S.

drawn normal

the unit outward

is

any closed surface S bounding the region V, prove that

to

jJJdivndF = S
V

STOKES' THEOREM
68.

a;2

69.

70.

A =

Verify Stokes' theorem for

+ z"

j^2

and C

2yi

^ Zx] - z'Xa,

where S is the^upper half surface of the sphere


Ans. common value - 9r

boundary.

is its

of the region in the


Verify Stokes' theorem for X = (y + z)i- xzj + y^k, where S is the surface
xy plane, (6) plane
(a)
in
the
=
included
=
not
which
is
2
S and y
first octant bounded by 2x + z
boundary.
corresponding
=
is
the
and
C
6
=
2x
+
z
plane
(c)
y
2,
Ans. The common value is (a) 6, (6) 9, <c) 18

Evaluate

JJ

VX

A)

\ =

n dS, where

(x

- z)i +

(x'

+ yz)j -

and S

Sxy^k

is

the surface of the

s'

cone
71.

If

= 2-

is

y/x^

+ y^

Ans.

above the xy plane.

12it

S and B = V X

a region bounded by a closed surface

A,

jj B'n dS =

prove that

0.

72.

Prove that

(a)

F = V0.
Ans.
73.

x'

74.

Let

- 42/ +

3x

Show

b\

- 4z)j C

4j/k

a conservative force

any path from (3,-1,2)

is

(c)

is

if

f F- dr

F=
if

5r'r,

F=

(h)

field,

to

(2, 1,

a-

to

Find

4>

such that

-1).

where r

/(r)r,

x'

where

+ y' + z^ =

a;i

+ 2/j + zk,

/(r) is

then

assumed

any point on the sphere

J^

dr

ft'^- o=.

to be continuous.

r f(r) dr

a function

is

<p

such that

- y)i + (x'y + z^)j + (3a;z^ - xy)k.


If so,
Zxe-'i + (cosz - aj'e"")] - y sinzk.
= x^e'' + y cosz +
does not exist. (6)
(a)

F =
F =

Ans.

(x^

+ constant,

that

Determine whether there


(b)

F-dr, where

In Problem 73 evaluate

(a)

76.

x^j/

3)1

be any path joining any point on the sphere

+ y' + z^ =

Ans.
75.

(2xy

Evaluate

(c)

(6)

F =

F = V^,

where:

(xz

(z' - Axy) dx
Solve the differential equation
Ans. xz' - 2x^y + 3y' + z = constant

(6y

find

it.

constant

- 2x') dy +

(Sxz'

+ l)dz =

0.

MISCELLANEOUS PROBLEMS
77.

Prove that a necessary and


closed path

two, at least)

78.

in a region
is

that

sufficient condition that

(where

-^ + -^

is

^^^dy-j^dx

be zero around every simple

order
continuous and has continuous partial derivatives of

0.

containing two "holes" (see Problem


Verify Green's theorem for a multiply-connected region

10).

CHAP.

79.

If

LINE INTEGRALS, SURFACE INTEGRALS, INTEGRAL THEOREMS

10]

Pdx + Qdy

223

not an exact differential but p.(P dx + Q dy) is an exact


differential where ^ is
y, then ,, is called an integrating factor,
(a) Prove that if F and G are
functions of x alone, then (Fy + G) dx + dy has an
integrating factor ^ which is a function of x
alone and find /.. What must be assumed about F
and G? (6) Use (a) to find solutions of the
differential equation xy' = 2x + Zy.
is

some function of x and

Ans.
80.

81.

(a)

p.

e'^^"'^'

(6)

= cx^-x,

Find the surface area of the sphere

where

x^

y'

c is

any constant

+ {z- aY =

If /(r) is a continuously differentiable function


of

82.

Prove that
position

83.

VX

Js J

and n

is

(0n)

contained within the paraboloid z

yjx^

+ y' + z'

x^

+ y\

prove that

dS =

a unit outward

a'

where ^

drawn normal

any continuously

is

differentiable scalar function

of

to a closed surface S.

Establish equation {3), Problem 32, by using Green's theorem in the


plane.
[Hint: Let the closed region "3^ in the xy plane have boundary
C and suppose that under the transformation X - f{u, v), y = g(u, v) these are transformed into %' and
in the uv plane respectively.

First prove that

from sign

^^ F(x,y)dxdy

^ Q(x,y)dy

this last integral is equal to

Q^^fiu.v),

J^
theorem to transform this into

84.

Flfiu.v), g(u,v)]

\\
JJ

Then show that apart

F(x,y).

g{u,v)f^du + ^dv'].

Finally use Green's

^i^^ldudv.
B(u,v)\

a; = f{u, v, w),
y = g(u, v, w), z = h{u, v, w) defines a transformation which maps a region
xyz space into a region %' of uvw space, prove using Stokes' theorem
that

valid.

F(x,y,z)dxdydz

where G{u,
IS

v, w) = F[f{u,
See Problem 83.

G{u,v,w)

jj

d{x, y, z)

d(u, V,

v, w), g{u, v, w), h{u, v, w)].

w)

of

du dv dw

State sufficient conditions under which the result

(a) Show that in general the equation


r = i{u, v) geometrically represents a surface.
(6) Discuss
the geometric significance of u = cu v ^ c. where ci and e, are
constants, (c) Prove that the element
of arc length on this surface is given by

= Edu' + 2Fdudv + G dv^

ds^

E =

where

86.

dQ/dy

If

JJJ

85.

y,here

^
OW

'^
bu

F=^-^
G=^.^
du

dv

'

dv'

dv

(a)

Referring to Problem 85, show that the element of surface area

(b)

Deduce from

(a)

that the area of a surface r

t{u,v)

is

is

given by dS

= y/EG-F'dudv.

fC y/EG-F^dudv.
s

[Hint:

(A X B)
87.

88.

Use the fact that

(C

||

X D) = (A C)(B D)

(a) Prove that


r
sphere of radius a.

=
(b)

0| =

V(^^1>(S^

(A D)(B

^""^

'""^^

"^

*^^

*''-*-

C).

a sin m cos i; i + a sin m sin v j + a cos m,


S m S j-, S ^ < 2^ represents a
Use Problem 86 to show that the surface area of this sphere is Ava'.

Use the result of Problem 34


Fage 142.

to obtain

divA

in

(a)

cylindrical

and

(6)

spherical coordinates.

See

chapter

11

Infinite Series

CONVERGENCE
Consider the

and

DIVERGENCE

INFINITE SERIES

of

infinite series

'^ U

Ui

U2

Us

n= l

Let the sequence of partial sums of the series be


Sl

= Uu

S2

Ml

Sz

M2,

If this sequence is convergent,

series (1) is called convergent


is called

=^

Ui

U2

Si, S2, Ss,

+ Us,

and S

called its

is

where

.
.

= U1+U2+

Sn

...,

there exists a

i.e. if

+U

limS

number S such that

S,

(2)

the

If lim Sn does not exist, the series

sum.

(Compare Chap. 3, Page 43). We shall sometimes abbreviate the


by Swn and shall refer to Un as the nth term of the series.

divergent.

infinite series {1)

Example

Example

1:

2|r =

f+i + i+--

Chap.

Then

2:

3).

(-1)"''

lim

since

Here S = sum

Tl -

n-+oo \

=
^^
^ /

= 1-1 + 1-1+--.

Hence lim Sn does not

exist

1,

the series

Here S

and the

of first

n terms
is

- 1;

(see Prob. 25,

convergent and has sum

or 1 according as n

is

S=

1.

even or odd.

series is divergent.

n-+oo

FUNDAMENTAL FACTS CONCERNING INFINITE SERIES


1.

If

limM

2 converges, then

ever

is

not necessarily true,

It follows that if the

i.e.

(see Frob. 26,


if

lim Un

0,

Chap.

2m

3).

may

or

The converse how-

may

not converge.

nth term of a series does not approach zero the series

is

divergent.
2.

3.

Multiplication of each term of a series by a constant different


not affect the convergence or divergence.

Removal

(or addition) of a finite

number

of terms

from

from zero does

(or to) a series does not

affect the convergence or divergence.

SPECIAL SERIES
1.

Geometric series

J ^""'

a + ar + ar^+...,

where a and r are constants,

n=l

converges to

terms

is

S=

= '^^^^

if

\r\

(see

< 1 and

diverges

Prob. 25, Chap.

224

if

3).

\r\

^ 1. The sum

of the first

CHAP.

11]

INFINITE SERIES
*

The p

2.

>

?>

TESTS

series

J^

The

^here p

series with

DIVERGENCE

and

Let

O^Un^Vn

t;

for

go

for

a constant, converges for

is

=1

is called

SERIES

of

of

the harmonic series.

CONSTANTS

Example:

n>N

N = l.

Often,

^^ g 1

Since

n > iV and suppose that 2v converges. Then


2m also converges. Note that
means

all

n>N,

all

from some term onward.

(6)

test for series of non-negative terms.

Convergence.
if

and diverges for p^l.

Comparison
(a)

+ __+_+...

CONVERGENCE

for

1.

225

2 |r converges, ^~r^

and

also converges.

Divergence. Let t;gO for all n.>N and suppose


that 2i; diverges.
if Un^Vn for all n>N, ^Un also
diverges.
Example:

Since

>

2^

and

diverges,

Then

also diverges.

Quotient test for series of non-negative terms.

2.

(a)

If

^o and

i;gO and

lim

if

J^

A^O

or

=,

then 2i. and 2^ either

both converge or both diverge.


(&)

If

(c)

If

to

Theorem

it.

1.

A=6
A = 00

in (a)

This test is related to the comparison test and is often


a very useful alternative
In particular, taking Vn = l/n", we have from known facts
about the p series the

Let

lim%''M

(i)

lun converges

(w)

2m diverges
Examples:

3.

and 2vn converges, then 2m converges.


and 2!; diverges, then ^lu diverges.

in (a)

if

if

1.

>

2:^

pg

-^

2i

and

^ converges since

Inn
mn
.

A is finite
A v^ (A may

and

.^

"S,

Then

A.

,.

VV+T

infinite).

lim n'

-^

In

diverges since

lim n}
--

~.

(n+l)'

Integral test for series of non-negative terms.


If f{x) is positive, continuous

such that

f{n)^un, n

and monotonic decreasing for x^N and is


then 2m converges or diverges ac.,

= N,N+l,N+2,

fix)

dx

^m

we may have N^l,

as

is

often true in practice.

Example:

4.

be

-j converges since

fix)

Hm

dx

converges or diverges.

^=

Alternating series test. An alternating series


alternately positive and negative.

Hm
is

-^

In particular

exists.

one whose successive terms are

[CHAP.

INFINITE SERIES

226

An
(see

alternating series converges

Problem

the following two conditions are satisfied

if

15).

For the

Example:

11

\u+i\

(b)

lim u

ior

\Un\

n^l

(or lim

l-i + i-i + i--' =

series

=-,

\un\

(a)

\un+i\

=^.

Then for a

2^

\un+i\

1,

\un\

O)

^
^

|m,|.

we have
Also

lim

Mn
|m|

=0. Hence

the series converges.

Theorem 2.
The numerical error made

stopping at any particular term of a convergent


alternating series which satisfies conditions (a) and (b) is less than the absolute value of
the next term.
Example:

we

If

stop at the 4th term of the series

than 1

5,

in

.the error made

l-i + i-j + i

is less

= 0.2.

Absolute and conditional convergence. The series 2m is called absolutely convergent if 2 \un\ converges. If Swn converges but 2 \un\ diverges, then 2m is
called conditionally convergent.

Theorem

3.

is

convergent (see Prob.


Example

Example

1:

T?

2:

In words, an absolutely convergent series

Sm converges.

If 2|Mn| converges, then


17).

+ 4-|?-J5-+^ + ^--is

absolutely convergent and thus convergent, since

the series of absolute values ]T+2i^

+ 3?+45"+

^-^ + ^-^+

is

converges, but 1

converges.

+ 4+

diverges.

Thus

2"

~
3

conditionally convergent.

Any

of the tests used for series with non-negative terms can be used to test

for absolute convergence.

6.

Ratio

Un+l

lim

converges (absolutely)

(6)

diverges

L=1

if

test.

Let

converges (absolutely)

(b)

diverges

if

if

series

2m

L<1

fails.

(a)

L=

Then the

L>1.

the test

The nth root

If

8.

Let

(a)

If

7.

test.

lim y/juj^
if

Then the

L.

series

2m

L<1

L > 1.

1 the test fails.

Raabe's

test.

Let

limw(

'^

L>

(a)

converges (absolutely)

(b)

diverges or converges conditionally

if

^-

Then the

1
if

L<

1.

series

2m

CHAP.

INFINITE SERIES

11]

L=1

If

the test

This test

Gauss'

9.

(6)

If

fails.

often used

when

the ratio test fails.

L c
'^~n'^^'

Un + l
vtn

2m

series
(a)

test.

is

"^^^^^

converges (absolutely) if L>1


diverges or converges conditionally

This test

THEOREMS

on

is

227

if

\Cn\<P

for

all

n>N,

then the

LSI.

often used vs^hen Raabe's test fails.

ABSOLUTELY CONVERGENT SERIES

Theorem 4.
The terms

of an absolutely convergent series can be rearranged in


any order, and
such rearranged series will converge to the same sum.
However, if the terms of a
conditionally convergent series are suitably rearranged, the
resulting series may diverge
or converge to amj desired sum (see Problem
76).
all

Theorem 5.
The sum, difference and product of two
convergent. The operations can be performed

INFINITE SEQUENCES and SERIES

UNIFORM CONVERGENCE

of

absolutely convergent series


as for finite series.

absolutely

is

FUNCTIONS.

Let {Un{x)}, n = 1,2,3, .. be a sequence of functions defined in


[a, b].
The sequence
said to converge to F(x), or to have the limit F{x) in [a,b],
if for each c>0
.

IS

and each x
\Un{x)-F{x)\ < . for all n>N. In such case we
write hmun{x) = F{x).
The number
may depend on x as well as e. If it depends
only on c and not on x, the sequence is said to converge
to F{x) uniformly in la, b] or to
be uniformly convergent in [a,b].

we can

[a,b]

find

iV>0

such that

The

infinite series of functions


PC

^U{x)

n1 = l1

is

Ui{x)

U2{x)

Uaix)

...

said to be convergent in [a,b] if the sequence of partial sums


Ui{x) + U2{x) + ... +Un{x), is convergent in [a,b].
S{x)
=
and call S{x) the sum of the series.
"(*^

where S{x) =

(3)/
V

{S(a;)},

It follows that l,Un(x)

we can

find

and not on

converges to S{x) in [a,b] if for each


such that \Sn(x)~S{x)\ < e for all n>N.
the series is called uniformly convergent in [a,b].

iV>0

x,

Since S{x)-Sn{x) = Rn{x), the remainder after n terms,


that ^u{x) is uniformly convergent in [a,b] if for each e>0

but not on x such that \Rnix)\

<

for

all

n>N and all x in

1,2,3,

In such case

we

write

>0 and

UN

each x in \a b]
depends only on .

we can equivalents say


we can find A^ depending on

[a, b].

[CHAP.

INFINITE SERIES

228

These definitions can be modified to include other intervals besides

a<x<b,

a^x^b,

11

such as

etc.

of convergence (absolute or uniform) of a series is the set of values of x


the series of functions converges (absolutely or uniformly).

The domain
for which

SPECIAL TESTS

UNIFORM CONVERGENCE

for

SERIES

of

test. If a sequence of positive constants Mi, M2, Mi,


Weierstrass
that
in some interval
found such

1.

then 1un{x)

is

Example:

\Un{x)\

(6)

2Af converges

can be

1,2,3, ...

uniformly and absolutely convergent in the interval.


is

5-^

n=

g Mn

(a)

uniformly and absolutely convergent in

[0,277-]

since


-j

-r and

converges.

This test supplies a sufficient but not a necessary condition for uniform
convergence, i.e. a series may be uniformly convergent even vi^hen the test cannot
be made to apply.
One may be led because of this test to believe that uniformly convergent
Hovi^ever, the two properseries must be absolutely convergent, and conversely.
without being
convergent
uniformly
be
can
series
i.e.
a
independent,
ties are
See Problems 30, 123.

absolutely convergent, and conversely.

Dirichlet's test. Suppose that

the sequence {a}

(a)

is

a monotonic decreasing sequence of positive constants

having limit zero,


there exists a constant

(b)

Then the

Ui {x)

+ Ui{x) +

such that for

...+Un{x)\

< P

a^x^h
for

all

n>N.

series
CO

aiUi{x)
is

THEOREMS
If

an

If

to the

a2U2{x)

uniformly convergent in

on

]^a'u(x)

a^x^b.

UNIFORMLY CONVERGENT SERIES

infinite series of functions is

ties possessed

Theorem

by sums

uniformly convergent,

it

has

many

of the proper-

of finite series of functions, as indicated in the following theorems.

6.

{Un{x)],

sum

TO

S{x) in

are continuous in [a,b] and


..
then S{x) is continuous in [a, b].

1,2,3,

[a, b],

if

2un{x) converges uniformly

Briefly, this states that a uniformly convergent series of continuous functions is a


continuous function. This result is often used to demonstrate that a given series is not
uniformly convergent by showing that the sum function S(x) is discontinuous at some

point (see Problem 30).

CHAP.

INFINITE SERIES

11]

In particular

if Xo is in [a, b],

then the theorem states that

\imX'^n{x)

x-*x n = l

where we use right or

Theorem
If

to the

left

229

lim

M(a;)

y,Un{xo)

=i

n=lx-*XQ

hand

limits in case Xo is

..

are continuous in [a,b] and

an endpoint of

[a, b].

7.

n-

{Unix)},

sum

S{x) in

1,2,3,

[a, b],

.,

if

^Un{x) converges uniformly

then

S{x)dx

21

<'^Unix)\-dx

Un(x)dx

(4)

"=l-'a

"

or

Unix)dx

(5)

a uniformly convergent series of continuous functions can be integrated

Briefly,

term by term.

Theorem
If

and

8.

n- 1,2,3, .. ., are continuous and have continuous derivatives in [a,b]


converges to S{x) while ^Un{x) is uniformly convergent in [a, b], then in [a, b]

{Unix)},

if 2w(a;)

S'{x)

X<i^)

(6)

n=l

or

rx{tM^)} = tj-xMx)

(7)

This shows conditions under which a series can be differentiated term by term.

Theorems similar
= 1,2,3, ..

{Un{x)},

to the above can be formulated for sequences.


is

uniformly convergent in

Un{x)dx

which

is

POWER

the analog of Theorem

SERIES.

series

[a, b],

For example,

if

then

Mm. Un{x)dx

(8)

7.

having the form


CO

ao

aix

a^x^

...

^= <^n^"

(5)

where ao, ai, a%,


are constants, is called a
to abbreviate the series (9) as 'lanX".
.

power

series in x.

It is often

convenient

In general a power series converges for \x\ < R and diverges for |x| > R, where the
constant R is called the radius of convergence of the series. For |a;|=.B, the series may
or may not converge.

-R<x<R,

The interval \x\<R or


with possible inclusion of endpoints, is called
the interval of convergence of the series. Although the ratio test is often successful in
obtaining this interval, it may fail and in such cases other tests may be used (see Prob. 22).
The two

special cases

R=

and

R = ^

-><a:<>

(see

Problem

25).

can

arise.
In the first case the series conconverges for all x, sometimes written
speak of a convergent power series we shall

verges only for x=^0; in the second case

it

When we
R>0.

assume, unless otherwise indicated, that

Similar remarks hold for a power series of the form

(9),

where x

is

replaced by (x

- a).

[CHAP.

INFINITE SERIES

230

POWER SERIES

THEOREMS

on

Theorem 9.
A power

series converges uniformly

within

11

and absolutely

any interval which

in

lies entirely

of convergence.

its interval

Theorem 10.
A power

series can be differentiated or integrated term by term over any interval


convergent power
lying entirely within the interval of convergence. Also, the sum of a
convergence.
of
interval
its
within
entirely
lying
interval
any
series is continuous in

This follows at once from Theorem 9 and the theorems on uniformly convergent
The results can be extended to include endpoints of the
series on Pages 228 and 229.
interval of convergence by the following theorems.

Theorem

Abel's theorem.

11.

of
a power series converges up to and including an endpoint of its interval
this
include
to
far
as
so
extends
also
convergence
uniform
convergence, the interval of
See Problem 42.
endpoint.

When

Theorem
If

Abel's limit theorem.

12.

converges at x

'^UnX"

= Xo which may

be an interior point or an endpoint of

n=

the interval of convergence, then

lim\^anX-\ =
n=

x-*x

If xo is
left

|;llimaa;4
n=

2 aa;?

(10)

n=

\x-yx
J

an endpoint we must use x-^xo+

or

in {10) according as Xo is a

x^xo-

or right hand endpoint.

of
This follows at once from Theorem 11 and Theorem 6 on the continuity of sums
uniformly convergent series.

OPERATIONS

with

POWER SERIES

In the following theorems

we assume

that

all

power

series are convergent in

some

interval.

Theorem

13.

added or subtracted term by term for each value of x comconvergence.


to their intervals of

Two power
mon

Theorem
"

series can be

14.

Two power

series, for

^= "^"

example

^^^

2j CnX"

^"'^"'

^an be multiplied to obtain

n=

11

where
Cn

aobn

CLlbn-l

the result being valid for each x within the

Theorem

Cl2bn-2

common

dnbo

(H)

interval of convergence.

15.

If the

power

series

^anX"

is

divided by the power series 26a;" where &o=^0, the

of x.
quotient can be written as a power series which converges for sufficiently small values

CHAP.

INFINITE SERIES

11]

Theorem

231

16.

""> then by substituting x =


we can obtain the coefficients
&n2/"
n=0
=0
bn in terms of a. This process is often called reversion of series.
If

EXPANSION

of

FUNCTIONS

Suppose that

f{x)
in the closed interval

and

Then

in

as

f{x)

f{a)

i,

which

lies

f'(x),f"{x),
.,/<">(x) exist and are continuous
and that f" +(a;) exists in the open interval a<x<b.
previous chapters (see Pages 61 and 95)

+ na){x -a) +

where Rn, the remainder,

where

derivatives

its

is

^"

^n

t^{x -af+...+ ^^("^ - )"

(^

(a^-)"^'

i)j

^^(a;-|)(a;-a)

between a and

/(a)

-K"

(^^)

given in either of the forms

/'(a)(a:

- a) +

(Lagrange's form)

(13)

(Cauchy's form)

(U)

general different in the two forms.

x, is in

As n changes, ^ also changes in general.


then (J^) can be written
fix)

a^x^h

have seen

-v/e

POWER SERIES

in

If for all x

and

$ in [a, b]

we have

lim

i2

"-*"

/^

(a;

- a)^ +

^^(x -a)^+

...

0,

(15)

which is called the Taylor series or expansion of f{x). In case a = 0, it is often called
the Maclaurin series or expansion of f{x). For problems involving such
expansions see
also Chapter 6.

One might be tempted to believe that if all derivatives of f{x) exist at x = a, the
expansion (15) would be valid. This however is not necessarily the case, for although
one can then formally obtain the series on the right of (15), the resulting series may not
converge to f(x). For an example of this see Problem 104.
Precise conditions under which the series converges to f{x) are best obtained
by
of the theory of functions of a complex variable.
See Chapter 17.

means

SOME IMPORTANT POWER SERIES


The following series, convergent to the given function in the indicated intervals, are
frequently employed in practice.
x^

x'^

smx

^-31+5!-

2.

cos^

3.

e^

4.

ln|l

5.

iln

1 + x
1-a;

-,

1.

a;|

-_+___+

-f + ^- J+

^+3-

cc-^+^-^+

X^

O'^

6.

tan-x

7.

(l

+ x)"

:.

X*

x^

X^

'7*5

X^"~^
...(_i)-x^^__
+

x^

2T+3T+

a;2-i

x''

71+ (-1)""^(2^+

_<^<,

...

+(^+
a;""'

-^<x<^

...(-l)n-.J+

...

-Kx^l

-l<x<l

T+--- +2^:ri

/v2n

/y7

1+PX+ ^^P^x^
^l

-~<^<~

-l^^^l

.(-1)-|^3T +
...

J>iv--L)---iP-n
ni

l)^

[CHAP.

INFINITE SERIES

232

This

is

the binomial series.

a positive integer or zero, the series terminates.

(a)

If

(6)

If

25

(c)

If

l<p<0,

(d)

If

For

all

11

is

>

but

= 1,

is

not an integer, the series converges (absolutely) for


the series converges for

^ 1.

a;

Ka;^!.

l<x<l.

the series converges for

p the series certainly converges

-1 g

1< < 1.

if

a;

SPECIAL TOPICS
1.

Functions defined by series are often useful in applications and frequently arise
as solutions of differential equations. For example, the function defined by
x''

Jv{x)

x^

2^|-^ ~

2(2p

x'^

+ 2) ^2-4(2p + 2)(2p + 4) ~

^"'^^

^ (-l)"(a:/2)''+^"
n\{n + p)\

n=o

a solution of Bessel's differential equation x^y" + xy' + (x^-p^)y


See Problems 46, 106-109.
thus called a Bessel function of order p.
is

and

is

Similarly, the hyper geometric function


,

F{a,h;c;x)
is

a(a

a-

+ 17^ ^ +

a solution of Gauss' differential equation

aby

+ 1)6(& + 1)
,..,,

^^^^
+ l) ^ +
x{l - x)y" + (c - (a + 6 + \)x}y' ,

i.2.c(c

0.

These functions have

many important

properties.
00

2.

Infinite series of

complex terms,

in particular

power

series of the

form

^=

""'^">

where

+ iy and

may

be complex, can be handled in a manner similar

to real series.

Such power series converge for \z\ < R, i.e. interior to a circle of convergence
+ y^ = R2^ where R is the radius of convergence (if the series converges only
for 2 = 0, we say that the radius of convergence R is zero; if it converges for all z,
we say that the radius of convergence is infinite). On the boundary of this circle,
= R, the series may or may not converge, depending on the particular z.
i.e.
Note that for y = the circle of convergence reduces to the interval of conx^

I^l

vergence for real power series. Greater insight into the behavior of power series
is obtained by use of the theory of functions of a complex variable (see Chapter 17).
PC

3.

Infinite series of functions of

two (or more) variables, such as

be treated in a manner analogous to series in one variable.


can discuss power series in x and y having the form
ttoo

(aioa;

amy)

{a2oX^

+ anxy +

ao2y^)

^Unix,y) can
n=l

In particular,

we
(18)

using double subscripts for the constants. As for one variable, we can expand
suitable functions of x and y in such power series using results of Chapter 6,
Page 109 and showing that the remainder Rn-^0 as n^ =. In general, such power
series converge inside a rectangular region |x| < A, \y\ < B and possibly on the

boundary.

CHAP.

4.

INFINITE SERIES

11]

Double

Series.

Consider the array of numbers (or functions)

Un

Let

Smn

Ui2 Uis

U21 U22 U2S

U31 U32 M33

X 2 ^1

sum

be the

columns of this array.

^ ^ Mp,

converges to the

limi

P,

rows and

lim Smn

S; otherwise

it

first

S,

we

diverges.

+ u,){l + ,)(! +us)...{l + u^) denoted by


ft (1 + u,)
u^^-l, k = 1,2,3,.... If there exists a numbeV P^O

Let P

where we suppose that


such that

sum

series are very similar to those for series

already considered.
Infinite Products.

in the first

a number S such that

and theorems for double

Definitions

numbers

of the

If there exists

say that the double series

5.

233

(1

we say

that the infinite product

(1

+ ui){l + U2){1 + U3)... =

00

(1

+ Uk),

or briefly n(l

+ Uk),

converges to P; otherwise

it

diverges.

If Il{l + \Uk\) converges, we call the infinite product n{l +


Uk) absolutely
convergent. It can be shown that an absolutely convergent infinite product
converges and that factors can in such cases be rearranged without affecting
the
result.

to

Theorems about infinite products can (by taking logarithms), often be made
depend on theorems for infinite series. Thus, for example, we have the

Theorem.
is

6.

necessary and sufficient condition that n(l


that luk converge absolutely.

Summability.
If the

Let

sequence

^-g"^,

in the

converge absolutely

be the partial sums of a divergent series 2m.

n terms of

of the first

summable

Si,

Si, S2, S3,

+ Uk)

-,

(formed by taking arithmetic means

Si, S2, S3,


.) converges to S, we say that the series
Cesaro sense, or C-1 summable to S (see Problem 51).
.

52t is

If 2m converges to S, the Cesaro method also yields the result S.


For this
reason the Cesaro method is said to be a regular method of summability.

In case the Cesaro limit does not exist,


the sequence

'^

Si,

exists and equals S, we say that


can be continued indefinitely.
7.

Asymptotic

series.

1,

'S.Uk

....

we can apply
If the

converges to

the same technique to

C-1 limit for this sequence

in the C-2 sense.

The process

Consider the series

sw

0.

- +

...

+; +

...

,)

and suppose that


S(a.)

5+|| +

are the partial sums of this series.

...

+^

^ = 0,1,2...

i20)

INFINITE SERIES

234

Rn (x) =

If

Sn (x), where

f{x)

then S{x)

is

an asymptotic expansion of

called

11

such that for every n

f{x) is given, is

lim x'^Rnix)

fix)

[CHAP.

f{x)

and we denote

this

by writing

S(x).

However, by taking the sum of suc-

In practice, the series (19) diverges.

cessive terms of the series, stopping ju^t before the terms

we may

begin to increase,

obtain a useful approximation for f{x).

Various operations with asymptotic series are permissible. For example,


asymptotic series may be multiplied or integrated term by term to yield another
asymptotic series.

Solved Problems

CONVERGENCE
1.

111

Prove that pg- + ^^^^

(a)
its

DIVERGENCE

and

+ ^-^ +

of

Sn

- l)(2n
(2n-l)(2n

"

"^

+ l)
1)

(2to

~ I^T+Iy-

2 V2^r=^l

Since

lim S

series is

last, cancel

{1

lim

(&) find

Prove that

(l)^

Subtract:

iSn

lim S

Another method:

Let a

lim

(|)3

^2n-l

^\
+ iy

2M

the series converges and

17,

= f,

j^

+
1.

(f

=f

)"}

2,

converges and

(f)-

= I + (|)^ + (!)'+
=
(|)^ + (|) +
= | - (|)"+i

lim 2{1

Prove that the series i


lim M

sum

its

i_\

^ 1/
2

2n

\^

+ iy

is A.

sometimes called a telescoping series since the terms of Sn, other than the

Since

first

and

out in pairs.

|S

3.

and

'^**^"

iA_i+i_i+i_...+^^
2^1

(a)

^^o^^^rges
l)

Ml

The

^A2n-l){2n +

CONSTANTS

of

sum
"*"

2,

SERIES

+
+

(|)

)"+>

(f

or

2{1

the series converges and

+ #+ =

Hence by Problem

2^
26,

sum.

(!)

in Prob. 25 of Chap. 3; then the

find its

(b)

:rT

Chapter

its

sum

(f)"}

sum

is

is 2.

a/{l

r) =

f/(l

diverges.
3,

the series

is

divergent.

f) =

2.

CHAP.

INFINITE SERIES

11]

Show that the series whose wth term


The fact that

lim m

follows

235

\/n

+l-

from Problem

14(c),

is

Now

Sn = Ut + U2+---+Un = {V2-^fl) + (^/S-^/2)+...


Then S increases without bound and the series diverges.

This problem shows that Jim m


vergence of 2m. See also Problem

COMPARISON TEST
5.

1, 2, 3,

Let

Sn

Ul

+ Ui+

Then

Sn

Thus Sn

is

2m converges.

We

Ui

+ 112 +

Thus

to

+ lu ^

lnn<n

Then the given

iin

Also, since

+ Vn s T

t)

SO, r g

T.

OSSnST.

or

^^

prove that

test

(i

*)

(i

made

+i+i+

if

>

Chap.

(see

3)

diverges.

""'

= 1

_L(8terms)

and Vn be

s 1+

...

we can show

that

^+1 +

...

number by choosing enough terms, the

-^, where p

is

a constant, diverges

This can also be shown in other ways [see Problem 13(a)].

1.

^^ S^,

and

i)

larger than any positive

Test for convergence or divergence

Let

+V.

+ vi +

vi

analogous to that used here,

and converges

Since

+ V2+

any desired number of terms,

By methods
1

V1

ps

r exists and equals T, say.

Since the right hand side can be


given series diverges.

8.

Ivn converges, prove that 2 also converges

if

1 + ^ + ^+... =
1
s

+
^
i
i
i + i = i
i
s i+i+ i+i = J
i + i + i +
+ l^+--- + T^ S Jg + J5 + j.+ ... +

have

7.

and

a bounded monotonic increasing sequence and must have a limit

Using the comparison

if

+Un, Tn

hm

Since 2v converges,

etc.

establish the comparison test for convergence).

(i.e.

6.

sufficient condition for the con-

6.

QUOTIENT TEST

and

O^Un^Vn, n =

If

0.

""*"

+ {^/^^+l-^^n) = aATTI - Vl.

a necessary but not

is

diverges although Urn Un

Chapter 3

'" ^
"^i2n3-r

we have

series converges, since

positive.

lim -^

If

^
=

^ = J,

converges.

constant

A # 0,

prove that

2tt

converges

or diverges according as 2v converges or diverges.


for

By hypothesis, given
n = iV + 1, iV + 2,

>

choose an integer

such that

AT-I- 1 to

\~-a\<
I

-^ <

Summing from

we can

<=o

^- ^

<

'

(A- )Vn <

or

(more precisely from iV+

(A~e)

%Vn ^

1 to

Un

^n

+ e)

for

all

< {A+ )Vn

Af and then letting

(A

"2 Vn

n>N

Then

(1)
ilf -> =0),

(2)

[CHAP. 11

INFINITE SERIES

236

the right hand inequality of (2),


There is no loss in generality in assuming: A - e > 0. Then from
2m diverges when Ivn does.
In, converges when 2i; does. Prom the left hand inequality of (2),
62.
Problem
For the cases A = or A = , see

Test for convergence:

9.

(a)

For large

we have

n.

+ 2n

n"

^^^i 2^3-1'

'

^=

approximately

i.

'

*>

Taking

^-^in^

t^n

+ S'

^$=^

and

v.=^,

1.

= 42

2i;

2.^

is

^^J^|^

lim ^

Since

(a)

2m

diverges,

1/w

also diverges

by Problem

8.

obtain an appropriate
Note that the purpose of considering the behavior of Un for large n is^to
1/m.
v
taken
=
well
have
comparison series v. In the above we could just as

lim

Another method:

(6)

For large

Un

w,

+ yfn

2n^-i

lim

Another method:

lim

Since

nf ^'Z^^^ ^ =
.

^^

Page

1,

225, the series diverges.

_w_l
- -^ - -^

approximately Vn

1v, ^ ^"2^

and

Then by Theorem

4.

converges (p series with p==2), the given series

converges.

(e)

lim yv^'f-^^^^^')

Then by Theorem

^(|^)
lim

with p

n"'(^^ =
= 3/2,

2 l/n

10.

Examine
(a)

Page

225, the series converges.

lim

(by L'Hospital's rule or otherwise).

^^ < ^ =

6(a) yields

i'

^"^^ ^''^^^''^ "^"^

2 ^"

(o)

>

n=l

2 sin^(X"/

(^)

n=l

Then by Theorem

(by L'Hospital's rule or otherwise).

with p

converges.

(6)

For large

n, sin (l/n) is

approximately l/n.

This leads to consideration of

/A

r
lim

hm n'smM -) =
,.

from which we deduce, by Theorem

with p

= 3,

/sinOMV
<
f
rr-

1
1

that the given series converges.

INTEGRAL TEST
11.

Establish the integral test (see Page 225).

We

perform the proof taking

From

^^ ^^^"^"^

diverges.

for convergence:

lim w^e-"'

1,

the series converges.

Note that the method of Problem


since

Then by Theorem
I"-

N = 1.

the monotonicity of f(x),


Mn+i

Integrating from

a;

= nto

= n+

l,

made

if iV

/(a;)

/(n)

using Property

7,

Page

1,2,3,...

81,

n+l

M+i

> 1.

we have

= /(n+1)
a5

Modifications are easily

f{x)dx

1,2,3,...

= 2,

the series

CHAP,

INFINITE SERIES

Summing from w =

M2

1,

Af

1 to

M3

Mj,

f{x)

Ma

+ M3 +
^^

Thus the proof

Problem

is

'

M2

M_,

{1)

can be omitted.

we

equal to S,

is

we

unbounded,

see

from the

hand inequality

left

in {1) that

see

from the right hand inequality

2m diverges.

in {1) that

complete.

geometrically

12. Illustrate

and

Ml

monotonic increasing and bounded above by S, so that Szt, converges.

is

^^^^ '^^

ii^ J

exists

^^'^^ ^'^

J
+ M

dx

If /() is strictly decreasing, the equality signs in


(1)
^*

237

the

proof

in

11.

Geometrically, Mz

+ Ma +

+ m

the total

is

area of the rectangles shown shaded in Fig. 11-1,


while M1 + M2+
+Mj,_, is the total area of
the rectangles which are shaded and non-shaded.

The area under the curve y = f(x) from


x = X to x =
is intermediate in value between
the two areas given above, thus illustrating the

result

13.

(1)

of

Problem

11.

Test for convergence:

f^.p = constant;

(a)

~=

x~''dx
j

^i

P<^, j^

^^

P>'^>

j^

M'-'-l
p^ \

j_

^=

",

"_

Af'-"-!

i-p

'

and thus the

^ ^^^^ *^^ integral

-# = InM and

|;^;

(c)

{d)

| ne'

where p=l.

1-p

so that the integral

t~^;

(6)

lim InAf

series diverges.

and thus the

",

series converges.

SO that the integral

and thus the

series diverges.

Thus the

^^^

^^

JirX

series converges if

" Jim i In

(05^

p>

and diverges

if

S 1.

+ 1)1 r = Jim {lln(M+l) - Jln2} =

=0

and the

series

diverges.

XM
.

^'^^

Jif. J,

^if^ = Jl(''^*)ir = Jim


"'""''

'^'^

J!]|?L

-i^'^ir

{In (In ilf)- In (In 2)}

^4""'

*""*">

= 4"

and the

series diverges.

^"id the series converges.

Note that when the series converges, the value of the corresponding integral is not
(in
general) the same as the sum of the series. However, the
approximate sum of a series can often
be obtained quite accurately by using integrals. See Problem 70.

ce

14.

[CHAP.

INFINITE SERIES

238

Prove that

<

-r

2)

From Problem

11

it

.,

follows that

lim

M-foo = 2
00

i-e.

<

Since

+
I

<

lim

The required

result

is

-,

from which

lim

M-.00 n = l
1

2 ~TTT

<

-r

obtain, on adding

<

TT

< ^, we

:;

<

T^

J^ X'

M_>oo

00-1

-|

TT

< 77+
t2
4

^in^ + 1
.,

11

W2

-,

_i_

required.

to each side,

<

+T--

tt

therefore proved.

ALTERNATING SERIES
15.

Given the alternating series ai a2 + a3ai+--- where 0^an+i = an and where


lim a = 0. Prove that (a) the series converges, (6) the error made in stopping at

any term
(a)

not greater than the absolute value of the next term.

is

The sum

2M terms is
= (di - c(2) + {as-at) +
+
ai (a2 as) (a4 as)

of the series to

SjM

Therefore {S^m}
Also,

S2M + 1

it

lim a

0),

Thus the

(6)

is

('*2M-|-l

and

is

SiM

partial

+ a^M +

sums

2M + 2)

~ <*2M + +
2

16. (a)

Sam

lim Szm+i
of the series

2M

is

ttjM-i)

"-aM

at

lim Sum

and

made

{"'iM

lim

ct^M

+ 4)

O'lM

non-positive and greater than

^ o

Prove that the series

1"

2M +

~ ~{'''2M + 2~
O2M+2.

'"

by hypothesis,

(for,

S.

is

in stopping after

+ 3~"''^2M + 4)

lim ajM + i

approach the limit S and the series converges.

terms

"" ''2M

(^BM + S

= S

thus non-negative and less than or equal to aiM + i, the

Similarly, the error

which

lim Sim

Since

i-

in stopping after

we have

S2^S4^Se^S,^

0,

a^u)

(a2_2

a bounded monotonic increasing sequence and thus has limit S.

follows that

The error made

Since the quantities in parentheses are non-negative,

S2M

(ffljM-i

converges,

'^aM

(b)

first

terms

+ s)

term which

is

('''2M

+2

'''2M

omitted.

is

+ 3) ""

('^BM

Find the

+4

"2M + 5)

maximum

"
'

'

error

made

in

(c) How
first 8 terms and the first 9 terms of the series,
of the series are needed in order to obtain an error which does not

approximating the sum by the

many terms

exceed .001 in absolute value?


(a)

The

series is

'"
'

'

'

2m

+ 1'

^+

|^

Since

that the series converges.

-^

-prr

2n

+l

2n-l^

-^

If

and since

p,

^r

lim

then

n^2n-lr

0,

it

\u\

= - y,

a+i

follows by Problem 15(a)

CHAP.

INFINITE SERIES

11]

Use the

(6)

is

positive

Similarly, the first 9 terms are 1 negative and greater than or equal to -^,

i+ J - ^+ i-

fj-

"

tV

t^

+ iV

and the error

IS

the error does not exceed j^ in absolute value.

i.e.

The absolute value of the error made in stopping


the desired accuracy we must have 1/(2M+1)

(c)

1-^ + |.-1 + 1-jl.+ i-ju

Problem 15(6). Then the first 8 terms give


and does not exceed jV-

results of

and the error

239

M terms

after

than l/(2Af+l).

is less

M a 499.5.

from which

.001,

Thus at

To obtain
least 500

terms are needed.

ABSOLUTE
17.

CONDITIONAL CONVERGENCE

and

Prove that an absolutely convergent series


Given that 2
Let Sm

+m +

Ml

we must show

converges,

\un\

+ Mj,

=
^

Sm + Tm

Since

|m|

and Tm
(Ml

m+

2|m2|

|m|

bounded monotonic increasing sequence, and so


Also, since

lim

Tu

lim
M^oo

must

18.

also exist

convergent.

that 5m converges.

iMil)

2|mi|

converges and since

\ui\

is

\u2\

{U2

+
+

+
|M2|)

...

lim (Sm

Sm

+Tm Tm) =

lim (Sm
(Sm-

M-<

23

337r

is in

>

(Mj,

1,2,3,

.. .,

is

|m|)

follows that

it

Sm+Tm

is

Tu) exists.

lim (Sm

by hypothesis),

Tm)

Tm

lim
M-o<>

proved.

^^^^ - ^^^ +
^

jL

...

M-oa

Investigate the convergence of the series

]172^

exists (since the series is absolutely convergent

and the result

Since each term

Then

2|Mj|

for

0,

\uj.

^^^
O

....

absolute value less than or equal to the corresponding term of the series

which converges,

follows that the given series is absolutely convergent and

it

hence convergent by Problem 17.

19.

Examine for convergence and


(\

^ {-'^y-'n

(a)

The

absolute convergence:

^ (-1)"-'
2

series of absolute values is

^(-l)"-i2

.^

,^. which

is

divergent by Problem 13(6).

Hence the given

series is not absolutely convergent.

However,
all i

1,

if

and also

|m.|

lim a

^^

lim

-j^

Since the series converges but

(6)

The

series of absolute values

is

and

n=2

is

cin+i

0.

|m+i|

= ^-^1-^,

Hence by Problem 15 the

not absolutely convergent,

it

is

then a+iSa for

series converges.

conditionally convergent.

n Wn
-

By the integral test, this series converges or diverges according as


or does not exist.

lim f _^^_
"- ./,
* In^ x

exists

[CHAP.

INFINITE SERIES

240

C
Hence

lim

M^ooJf

dx

a;

f^,
In^

du

lim

= t\
r^ - ^r^
In 2
In My

and the integral

M_,\^ln2

ic

11

exists.

Thus the

p^^-

i^^'^^' **^"*

series converges.

Then

=2

In

converges absolutely and thus converges.

Another method:

S'"-

(.+i)inMn+i) =

;nk

^^^

To examine

the given alternating series converges.


as above.

where

lim Un=

Since

(c)

^^

lim m ?^

that

""^

we must

absolute convergence

its

proceed

To show

On

show that

to

it suffices

0,

*"-'

^*

'

the given series cannot be convergent.

n-+oo

;nk

by L'Hospital's rule or other methods

lim

|ttn|

lim

-j-

This can be accomplished

0.

[see Prob. 21(6)].

RATIO TEST
20.

Establish the ratio test for convergence.


Consider
if

lim H-tL
n^ M Un

By

first

the series

= L<

hypothesis,

L<r<l.

U1

+ U2 + U3+

where each term

we can

choose an integer AT so large that for

< ru^
< r-M^ + <
< -M + 2 <

Mk + 2

<

u{r

and so the given series converges by the comparison

above proof and Problem

17, it follows that if

we

\^i:-tl\
'^n

we can prove

where

z=

L =

that

if

lim

|mi|

L<

1,

1.

l^al

\us\

Then by the

then 2m, converges (absolutely).

Hiili

i,

>

the

series

2m diverges, while

the ratio test fails [see Prob. 21(c)].

21. Investigate the


(a)

<

<r<

consider

lim fetil
^

n-+oo

Similarly

+ r^ + r^+

test, since

In case the series has terms with mixed signs,

(un+i/un)

r'u^

addition,
tt

lim

n^N,

all

r"M

Mn + 3

n* w

must prove that

Then
M^ + s

By

We

non-negative.

then necessarily 2m converges.

1,

Mn + 1

etc.

is

Here M

convergence of

(a)

T n^e-"',

lim

{n

+ ly
n

n-+oo

1,

e"

lim ('^-lYe-^-'

<

'^-

11=1

'-^

(c)

Jl=l

^2.1
'( -r i

Then

n''e~"''.

lim

Since

{b)

n=l

the series converges.

_-(m2 + 2n + l)

liml2L+l)!^
n 6""

n-oij

lim (^^?^t^ ^)

n^x

lim e"^"-'
n-+oo

I'O

if

CHAP.

INFINITE SERIES

11]

(-l)"-'2"

Here u

(6)

Then

Mn+l

lim

(-1)2"+'

lim

Here

(c)

>

1,

the series diverges.

(-l)"-'w

u.

and the ratio

(-l)"-'2"

Compare Prob.

(n

+ 1)^

19(c).

Then

.^' + l

n^| (-l)'(n + l)
lim
^|(n+ 1)^ + 1

lim|*Li|

2n'

+ lf

(n

Since 2

241

By using

test fails.

(-l)"-in

(^+l)(n-+l)

H^

other tests [see Prob. 19(a)], the series

is

seen to be convergent.

MISCELLANEOUS TESTS
22.

Test for convergence


(c)

l+2r +r^ + 2r^ + T^ + 2r^+

where

...

Here the

ratio test

Un+l

inapplicable, since

is

2|r|

or i|r|

However, using the wth root

test,

^^

2/3,

(6)

-2/3

V2\r\

if

if

Mis even

lim

Thus

y/lul]

if

\r\

Hence the

=
<

\r\

y/2\F\

^ /
=

series converges for cases (a)

ratio test fails since

li,

lim

]^)

n(l-

\r\

if

and

>

|>-|

is

odd

+ 3-6-9
ihn

li^

the series diverges.

and diverges

(6),

1-4-7

=.

|^|

=
-

...

in case

(c).

+ l-4-7-..(3n-2)
3-6-9...(3n)

(||lj =

1.

However, by Raabe's

....

test,

=4

Ji _ fSniy\

and so the series converges.

24.

Test for convergence

^-Y

2)
The

'

odd or

1).

the series converges, and

Test for convergence


The

lim 2"

(since

is

we have

I V\r'\

23.

depending on whether n

even.

Then

(a)

4/3.

ratio test fails since

n-.

However, using long


|Mn+i|

Mn

+ f'^-^Y +
V2-4y

P^^\ =

lim

\^

Un

'

_\

('^'^'^

f l-S-5...{2n-l) \

V2-4-6; +

lim' ^^
n-V 2m

\J

-^\^

+^
+2
[

1.

2-4-6...(2n)

'

Also, Raabe's test fails since

\2n + 2y

division,

/'2re+l

V2n + 2y

1
"

1
n

so that the series diverges by Gauss' test.

_^
'

5-4/w
4i='

+ 8i + 4

where

|c|

<P

INFINITE SERIES

242

SERIES
25.

of

(a)

/yin

11

FUNCTIONS

For what values of x do the following


o3

[CHAP,

00

;-^

\n 1 /y.2n

Assuming x^'Q

oo

(if

a;

(n

00

^<
\x\

series converges if

a;

If

a;

=3

= 3

the series becomes

n=i

Then the interval


Note that the

\x\

(-1)"
on

\x\

If

1.

y^ =

1,

i.e.

3, the test

fails.

which diverges.

(-1)"-^

which converges.

3=i

3 S

of convergence is

+ l)'

"^

n=l3W

il!^ 3(w

y^ >

if

1
= ^2~
3n=lW

-I

2 tt"

the series becomes

a;"-'

and diverges

1,

CO

If

+ l)'3"+'

\n

we have

the series converges),

lim
n-*

m(^

00

a;"

lim

Then the

series converge?

< 3. The

a;

series diverges outside this interval.

3 < < 3. At

series converges absolutely for

a;

a;

= 3

the series converges

conditionally.

(6)

Proceed as in part

lim
n-^

Then the

M < <
a;

(c)

Un

n\{x

'*"

w(a;

a)",

- 1)"
- 1)

2(3n

lim

(2n - 1)
n 1 ^2n
(-1)"-'*'

x^'

+ l)!

(2n

(2w

Then

1)

-,o(2n+l)(2w)(2TO-l)!

lim

'*""''"

(w+l)(a;-l)''+'
2+'(3n + 2)
(tt

=
Un

lim
n-+ 00
n^oo

series converges for

The

test fails for

For

a;

=3

Then the

x'a.

is infinite if

'

lim

"^

|a;

1|

[a;

the series becomes

i.e.

n=i

zero.

For X = 1 the
approach zero.
,

Then the

series

becomes

(2

l)!"'

x^

i.e.

the interval of (absolute) convergence

lim

r;(2n+l)(2n)

'

lim (n

1)

|a;

is

a],

= a.

series converges only for x

Then

x-1

l)(3w-l)(a:-l)
2n(3n + 2)

1| <
2,

x^

+ l)!(x-a)" +
n\(x a)"

(TO

lim

n-*oo

Thus the

_<.,

series converges (absolutely) for all x,


.

This limit

lim

.'

(-1)"

00

n-^oo

iA\

with m

(a)

and diverges for

a;-l

= 2

or

a!

=3

\x-\\

1| > 2.
and x = -1.

which diverges since the nth term does not approach


n6n r
1

n=i

series converges only for

which
^3w -^r
1
la;

1|

<

2,

also diverges since the wth

i.e.

2 <

a;

1 <

or

term does not

1 <

a;

<

3.

CHAP.

26.

INFINITE SERIES

11]

For What va.ue. Of. does

(,

243

2__i___

_|^(^|J,

^
ce

Then the
if

If

a;

If

or

If

a;

The

<

J-

diverges

1,

+ 21
= b-TT
la;

l^il >
1

if

^^1.-2.
^^

.^

If

and the

1,

test fails

a;

= 1 the series diverges.


= 2 the series converges.
= -^

Thus the

(6)

r^^

series converges if

+2

]x

+2

^,,,^,

the series

2^ 2^^1\ which converges.

is

|||| <

series converges for

lim p^L\
M

ratio test fails since

-.

that

= -^

1,

where
^''^'^^

^'

and

a;

- -2,

i.e.

for

(a;

+ n)(x + n-ir

a;

g -i.

w
^^^^^e^'

"O^'^S^

1
(a;

we

see that if

Sn

a;

+ n){x + n-l)

- 1, -2,

?^ 0,

lim Sn

provided

1/a;,

n-*eo

Then the

+n

~n,

and

(l~-A\+(^

+ .+ ...+ =

MX

+ n-1

x
0,

L^+.

+n
-2 -3

-1,

'

series converges for all x except

ar

0,-1, -2, -3,

and

.,

its

sum

is

1/a;.

UNIFORM CONVERGENCE
27.

Find the domain of convergence of (l-x)


Method

x{l

-x) + x%l -x) +

1:

Sum

of first

M terms

=
=

S,(x)

(l-x)

.T

-l-x'
=

- a;") =

If kl

<

1,

lim Sn{x)

If kl

>

1,

lim Sn(x) does not exist.

If

a;

1,

If

a;

-1, Sn{x)

S,(a;)

lim

and

(1

lim S(x)
(-1)" and

+
+

- a;) + x'{l-x) + ... + a;-' (l-x)


- a;^ + a;^ - a;' + ... + a;"-' - x"

x(l
a;

1.

0.

lim Sn(x) does not exist.


w-t-oo

Thus the

Method

using the ratio

2,

The

series converges for

converges;

if

a!

= l.

series converges if

a;

= -1

a;

= 1,

i.e.

for

-1< S 1.
a;

test.

series converges if

Thus the

\x\<l and

\x\

If

<

a;^l and

1,

the series diverges.

diverges

a;-^ (1

a;),

then

lim

|a;| > 1.
The test fails if \x\
series converges for -1 < i

if

Then the

f^l =

= 1.
g 1.

If

a;

lim

\x\.

= rthe

series

[CHAP.

INFINITE SERIES

244

the uniform convergence of the series of Problem 27

28. Investigate

(a)-~h<x<h {b)-i^x^h {c)-.99^xS.9d,


(a)

By Problem

^X-x",

SAx)

27,

We

in this interval.

S(x)

have

Remainder after M terms

Rn(x)

--

e,

n>

by In

since division

But

if

\x\

of X, the series

In this case

<

(which

In

i,

\x\

(d)

<

la;|

<

negative since

is

|,

In

|a;|

and

In (i)

^ S S
a;

<

number by choosing

|a;|

Discuss the continuity of the

find

N dependent

>

or

j^

inequality.
^) reverses the sense of the

Thus

since

is

independent

j^ = N,

> j^?^ S

so that the series

is

also

sufficiently close to 1.

1 < <
a;

sum

all

^^

Thus no

can be made larger than any

and

exists

it

follows that the

1.

points in this interval,

function

series is uniformly con-

shows that the

.99,

S(x)

cannot converge uniformly

it

limSn(x)

Problem 27 for the

of

a;

Oga;<l,

Thus
points in

x'

^ ^ 1.

interval

a;

<\ne

in this case, since

Since the series does not even converge at


in the interval.

If

we can

>

-J.

The arguments used above break down

If

- a;")

(1

'^

series is not uniformly convergent in

29.

any

1^ > 1^ = ^-

and >

In (^)

n\T:i\x\

\x\

Now

Reasoning similar to the above, with | replaced by


vergent in -.99 xS .99.

positive

(e)

when

thus the series cotiwrfl-es

uniformly convergent in the interval.

is

\x\

|a;|

la;"l

uniformly convergent in

(c)

Sn(x)

the interval

in

0^x<2.

(e)

-^<'^<h

i*

series is uniformly convergent in the interval if given


N.
but not on x, such that \Rn(x)\ < e for all
\Rn(x)\

(6)

S{x)

The

on

{d)

ln ()

"""

-l<x<l,

11

Sn(a;)

l,

S(x)

a;

<

S{x)

P
I

Urn Sn(x)

and S(x)
^*
if a;

= ^1

"^

=
"'

lim

(1

- a;") ^

1.

0.

and S(x)

is

discontinuous at

a;

but continuous at

all

other

1.

function
In Problem 34 it is shown that if a series is uniformly convergent in an interval, the sum
in an
continuous
not
function
is
sum
if
the
that
follows
It
interval.
in
the
continuous
S(x) must be
This fact is often used to demonstrate the
interval, the series cannot be uniformly convergent.
non-uniform convergence of a series (or sequence).

30. Investigate

the uniform convergence of

Suppose xv^Q.
Prob. 25, Chap. 3).

Then the

x^

+ y+x^ '^ (T+xY

series is a geometric series with ratio

^(-^

l-l/(l +

x-)

^^

"^

"

"^

"
'

1/(1

+ x^)

(1

+ x^Y

'^

whose sum

'"
is

'

(see

CHAP.

11]

INFINITE SERIES

U x~0

sum

Since

the

lim S{x)

n terms

of the first

S(x)

S(0),

S(0)

is

hence

0;

discontinuous at

is

245

;.

S(0)

= 0.

lim S(0)

0.

The^by Problem

34, the series

cannot he uniformly convergent in any interval


which includes x = 0, although it
convergent
any interval. However, it is uniformly convergent
in any interval which

is

(absolutely)

Lludes

"

= o!

This can also be shown directly (see Problem


89).

WEIERSTRASS

TEST

Prove the Weierstrass

31.

'''''^

test,

^^^^

i.e.

^^^

\un{x)\^Mn, n = 1,2,3. ...,


where ilf are
*^^" ^^'(^) i ""iformly (and absolutely)

if

'=""''^''^^''

?onfergent"'^^"*''

The remainder of the

l(^)i

\u,,{x)

series 2m(x) after

u,,{x)+

n terms

R^x) = u.,4x)+uMx) +

is

Now

+ 1-^,,(^)| +
g M + + M+. + ...
But M+, + M,,+ ... can be made less than e
by choosing n>N, since 2M, converges. Since AT is
clearly mdependent of. we have
\RAx)\ < e for n > N, and the series is uniformly
convergent The
absolute convergence follows at once from
the comparison test.
...\

==

|Mn+i(:r)|

...

32.

Test for uniform convergence:

cosnx
y2?i^

(a)

|5J**

(a)

w*

^-

Then

By

all

converges.

sin

nx

\n^

for

n-

2M

^sinnx
v^sinnx
n

-^i

1
^
+ x-\=
I

J^
n-'

= Mt g

-1 g

_1_.

+ x^

a;

i,

1),

the series

j.g.

|a;|

is

uniformly

1.

it/=^, we
-1 g s i by the M
a;

does not converge.

The

see that

2M,

test.

test cannot be used

conclude anything about the uniform


convergence
cxgcuue by
uy mis
this test

^ converges.

Then by the

test the given series converges uniformly

a;.

a power series 2ax" converges for x =


^o, prove that
in the interval \x\ < |xo|,
(6) uniformly in the
interval

Since ^a.x"o converges,

enough,

=4>

choosing

33. If

(a)

^iri'

M=^,

^n'*

test.

series converges uniformly for

we cannot

however. Problem 121).

all

||1|

.^
,
^""f

'

converges (p series with p

x by the

However, 5M, where

case and

this

(see,

all

in this interval,

Thus the given

(d)

'

\~^^\ =
in

,
^

'

the ratio test, the series converges in the


interval

For

^"^

since

(and absolutely) convergent for

(6)

X"
.^ ^"
n^iM-"^

(h\

i.e.

)a|

<

Jim ax^
for

n>N.

\a.x'\

\x\

and so we can make

it

converges

(a)

where

|x.|

|xi|,

\a.x2\

<

absolutely

< |xo|.

by choosing n large

Then

|alki"

<

5t^

[CHAP.

INFINITE SERIES

246

11

Since the last series in {1) converges for |a;| < |xo], it follows by the comparison test that
the first series converges, i.e. the given series is absolutely convergent.

(b)

M,=r^-

Let

Then 2M converges

since

<

la;i|

As

|a;o|.

part

in

(a),

|ana;"|

< M

for

Ixol

jxj

so that by the Weierstrass

jXiL

It follows that a power series


of convergence.

THEOREMS
34.

uniformly convergent in any interval within

is

S{x)

where we choose h

continuous in

is

+ Rr.{x),

S{x)

S{x

S{x

so that

+ h)-S{x)

so that both x

h)

Sn(x

+ h) -

S^(x

and x

h)

Since the series, by hypothesis,

is

<

\Rn(x)\

Then from

{1), (2)

\S(x

35.

\Sn{x

\h\

<

S,

interval

[a, b].

h)

Sn(x)

lie in [a, b]

(if

+ Rn{x +
+

Rn(x

= b,

and thus

h)

h)

and

+ h)-

e/3

S{x)

<

whenever

c/3

and

\Rn{x

\h\

we can

uniformly convergent,

+ h)\ <

R,(x)

(1)

for example, this will require h

Since Sn{x) is a sum of a finite number of continuous functions,


Then given e > 0, we can find S so that

for

its

Page 228.

6,

must show that S{x)

Now

uniformly convergent.

is

UNIFORM CONVERGENCE

on

Prove Theorem

We

2 a,. a;"

test,

<

also be continuous.

(2)

choose

for

e/3

must

it

< 0).

so that

n> N

(3)

(3),

\Sn(X+h)-Sn{.x)\

=S

S{X)\

and so the continuity

is

\Rn{x

h)\

<

\Rn{x)\

established.

Prove Theorem

7,

Page 229.

If a function
continuous,

is

continuous in [a,b],

its integral

Then

exists.

S{x)

since S(x), S(x)

and Rn{x) are

S{x)dx

Rn{x)dx

To prove the theorem we must show that


\

r"
I

S{x)dx

r'

Sn(x)dx

r'
I

Rn{x)dx

can be made arbitrarily small by choosing n large enough. This, however, follows at once, since by
independent of x
the uniform convergence of the series we can make |i?(x)| < e/(5 ct) for n>
in [a, b], and so

This

is

r"

Jj a

Rn{x)dx\
I

r"

^I a

\R(x)\dx

r"

<

.>'a

r^^^
^
^

"

equivalent to the statements

S(x)dx

lim

S{x)dx

or

lim

Sn{x) dx

lim Sn{x)} dx

CHAP. H]

36.

INFINITE SERIES

Prove Theorem
Let

g(x)

8,

247

Page 229.

2^m:(x).

Since,

by hypothesis, this

series converges

uniformly in

[a,b],

we can

integrate term by term (by Problem 35) to obtain

g{x)dx

u:{x)dx

u{x)

{u.(x)~uJa)}

Un{a)

S{a)

*^- ^'^-

^^^^

S{x)

00

because, by hypothesis,

prove?frt"hiri

37.

Let Sn{x)

2^ m^x)

converges to S{x) in

'"^^ ^^'" '

nxe-^^, n =

(a)

Determine whether

[b)

Explain the result in

(a)

j^

SWdx =

1, 2, 3,

a;

'

''^'

-(^)

'(^).

which

1.

f &(x)daj =

lim

f' lim Sn(x)dx.

(a).

_ a;e--^dx
lim

S(aj)

''^'

X^^^^^'^ =

[a, 6].

lim S(a;)

lim

= 1(1-6-).

-ie--^|;

S(a;)da;

nxe"^ =

lim

0,

s(ic
(cc)

1.(1

- g-")

whether x

dx

Then

= Oor 0<a;Si.

Then

^0

It follows that

rs.(x)da;^

lim

"

the integral sign.

The reason for the

(6)

f limS(*)da;,
^o ""*"

i.e.

the limit cannot be taken under

result in (a) is that although the sequence


S(^) converges to 0, it does not
To show this, observe that the function nxe-n^^ has a
0.
maximum at

convevgeumforr>ily to

j^

*^^ ""'"^^ ''"^^' * elementary calculus), the value of this


maximum being
./f Hence
""/I^
y/^ne
asn^<^, S{x) cannot be made arbitrarily small for all x and
so cannot converge uniformly to 0.
.

38.

Let

f{x)

We
all

a;,

= J?H.!^.

have

^Hl^^

in particular

^.

Then by the Weierstrass

Q-Sx^^, and can be

f{x)dx

Prove that

C0SW;7

1_
2^
-fi(27Z-l)

test the series is

integrated term by term.

uniformly convergent for

Thus

<F+^+F+-)

^|(2^

1)^

[CHAP.

INFINITE SERIES

248

11

POWER SERIES
00

39.

2= "^"

Prove that both the power series

series of derivatives

and the corresponding

Tl

OS

'^nanX"'^ have the same radius of convergence.


n=

Let

we can

i?

>

choose

so that

|a|

<

< R.

Then, as in Problem 33,

\xor

Thus the terms of the

series

corresponding terms of the series

InunX"'^

2w

however,

|a;|>i2,

lim "

= 2

can for

re |al |a;|"-'

n> N

which converges, by the ratio

'

,
,

Hence 2waa;"-' converges absolutely for


If,

\xo\

n> N.

for

rr

<

Let

be the radius of convergence of 2aa;".

all

points xo (no matter

and thus

lim oa;""'

how

be made less than

close

|a;o|

so that

0,

for

test,

|a;|

<

|a!o|

< B.

R).

is to

2notna;"~' does not

converge.

Thus

the radius of convergence of 2 nanX"'^.

is

Note that the

40. Illustrate

Problem 39 by using the


lim

Un + 1

lim

-+00

so that the series converges for


convergence is 3 = x = 3.

The

may

series of derivatives

\x\

may

or

3.

o
^2 'o
=in

At = 3

m^

,.

lim
n-*

00

3(w

By Problem

n^

3"

\x\

% 3"

3 ^

have the same radius of convergence,

a;

i.e.

<

3.

R = S,

although they do not have the

Note that the result of Problem 39 can also be proved by the ratio

The proof given

there, however, applies even

when the

test if this test is applicable.

test is not applicable, as in the series of

22.

Prove that

in

any interval within

its

interval of convergence a

(a)

represents a continuous function, say f{x),

(b)

can be integrated term by term to yield the integral of

(c)

can be differentiated term by term

We

(c)

'

^
.=1

The two
same interval of convergence.

(b)

the series also converges, so that the interval of

25(a) this has the interval of convergence


series

(a)

+ lf'

series of derivatives is

fi

41.

= i2.

a;"

2nx

Problem

|a;|

X"

series

(n+l)''3""

<

not converge for values of x such that

power

series

f{x),

to yield the derivative of f{x).

consider the power series 2aa;", although analogous results hold for 2a(a;

This follows from Problems 33, 34 and the fact that each term " of the series

This follows from Problems 33, 35 and the fact that each term anX" of the series
and thus integrable.

is

is

a)".

continuous.

continuous

39, the series of derivatives of a power series always converges within the interval
of convergence of the original power series and therefore is uniformly convergent within this
interval. Thus the required result follows from Problems 33 and 36.

From Problem

If a power series converges at one (or both) end points of the interval of convergence, it is
possible to establish (a) and (6) to include the end point (or end points). See Problem 42.

^^^^- "]

INFINITE SERIES

^^'

if

249

a P^r series converges at an end point


of its interval
of uniform convergence includes this
end point

rrZ,?'^''
^^^.r'",?^*
of convergence,
then the interval
For

simplicity in the proof,

we assume

power

the

series to be

.'

interval of convergence at a; =
1, so that the series surely converges
^"'"^^''^^^ ^'^'^
for
show that the series converges uniformly in
this interval
its

with the end point of

<
<
=r=
a;

TV,o we
..
Then
must

1
1.

Let

Now

Hence for

a;

<

{i2

R^,)xn

fin*"

'

(;''

...

i)

i,

lR,{x)l

smce (l-a;)a

+ (^__^^ _ i?, + .)a;+' + (i?.,, - i?,3)a;+^ +


+ ie+. + - ) + ie.+,(a;+2 - a; + + ...
!"{ - a-x){R^, + R^,x + Rn^.x' + ...)}

=
=
=

Rn(x)

|fil

+ + a;2 + a;'+...) = 1
= l, |fi (a;)| = |5.j < ,
a;

(if

Also, for x

re.uire'd^^elKlWs/"'^

^"

'^

(1

-^)(|fi,,|

for

^' "'^'^

''

<

a;

>

1^,,,,^

,^^,,^.

...J

^^^

1)

^^ ^"'^^^^"^^'^^ >*

*^

-'-

of . in

1.

and the

Extensions to other power series are


easily made.
43.

Prove Abel's limit theorem (see Page


230).
As in Problem 42, assume the power series

to be

Then we must show that

lim

a^x"

a.x\ convergent for

Sx g 1.

*.

Extensions to other power series are


easily made.
44, (a)

Prove that

tan-i x

vergent in -1

(6)

Provethat

(a)

By Problem

;r

x-~^ + ^-^-u
^
7 +

1.

r,
where
the

series is uniformly con-

~= i-i + i-.! +
25 of Chapter

3,

Y+^
Integrating from

to

with r

= 1~

where

a;,

= -x'
x'

-K x <

and a

= 1, we

+ x'-x'+
1,

have

...

-i<a;<i

^j^

yields

.-^ ^-^+...
7^

= ^--'- =
+ 5
./o +
(:2)
3
using Problems 33 and 35.
Since the series on the right of te)
- +1 it 4-^11
converses ^''
for r '
.,
t,
..,
"^
series is uniformly convergent
-}' ^* '"""^^ by Problem 42 that the
in
y
1 = x s 1 and represents tan-'a; In this
interval.

flfx^

m -l-^x^TlZ

(6)

By Problem
^lim

43 and part

tan-x

(a),

^lim

we have

(x

-^+ f - ^+

..

.^

:^

357^
,

45.

[CHAP.

INFINITE SERIES

250

We

have

to 3 decimal place accuracy.

1-^ dx

Evaluate

e"

+ 7T+"^+'
+ -5T
4!
5!
3!

-57

2!

ThenifM = -x^
Thus

we

zi
X'

= l-=^+f!--f!+ir-i!-+---'

e-^

=
-

X^

x"

X*

^~2T3!
1

^^

_
-

_
~

=
Note that the error made
fifth term,

i.e.

less

than 0.001

1- +

adding the
Problem

(see

_^__ _

I'

'"L

0.03333

9'5!

first

0.00595

0.00092

0.862

four terms of the alternating series

is less

than the

15).

(i 6),

x^y"

S a; S 1,

1_ + ^^

7'4!

5'3!

0.16666

MISCELLANEOUS PROBLEMS
46. Prove that y ^ Mx) defined by
The

<

*=

converges uniformly for

9-5!

7-4!

5-3!

3-2!

in

in particular,

so,

_x^ _ _x^

3-2!

<

and

_^^

-=

4!"^ 5!

Since the series converges for all x


can integrate term by term to obtain

X'X-e-2
-2

11

232, satisfies Bessel's differential equation

Page

xy'

(x^

- p^)y =

be
x [see Problem 106(a)]. Since a power series can
x,
for
all
interval of convergence, we have

series for Jp(x) converges for all

differentiated

term by term within

its

.ro 2''+''n!(M

^'

,?o

+ p)!
+ 2n)(p + 2w--l)x^+^

2''+^"m!(TO

" (-1)- (p

y"

+ p)!

n=

Then,

{x^-p^)y

_
*^'

a;Y'

i^gp+^'MKn + p)!

+ 2w) x'^'"
+ p)!
" (-1)" (p + 2w)(p + 2w - 1) x''+"'
2^
2''+'"i!(n + p)!
2''+'"'w!(M

00

X2/'

(x'-p')y

J\njj.p + 2n + 2

S^a'+^-wKw +
"*"

P)!

(-1)" [-p'

"^

(p

+ 2w) +

(-i)"a;''+''+^

?o 2-+^n!(M

.?! 2"+'"-" (n

"

2"+='"

+ p)!

(_l)-i

+ p)!

" (-1)" (p

.=0

Adding,
xj/"

.r2''"M!(n

- 1)

"*"

(-1)" [4ji(m

(n

+ 2ffl)(p + 2n - D]

x'-^"

?)!

+ p)] x"'^*'
+ p)l

2^*'"-n]{n

a;p+2''
!

(p

n !(. +

- 1 + p)

(-l)Mx''+^'

~.?i 2'>+^(-l)!( + P-l)!

"
-r,

.|

( 1)" 4x''+'"
2"+=''

(n-

1)! (m

+ p- 1)!

(-l)Mx'-^"

.=1 2''+"(n-l)!(w

+ P-l)!

CHAP.

INFINITE SERIES

11]

251
-n

so

47.

Since

lim

lim
(n

w
-^<'^,

i.e.

<

\z\

For

- 3,

|z|

1)'

Thus the

Assuming the power

z^ix

e''

cos X

Similarly,

49.

Prove that

lim(

n--oo \

Letting f{x)

==

2,

i sin

e"'^

cos

Consider

S..,

a;

sin

respect to x from w to m
'

M
I.e.

Sn+i

Sn
/"

*
n fi^of
to
0.577215.

1^1

Z
-g-,

2A

so that the series

|z[

absolutely

complex numbers, show that

sin x

we have
,

..

results are called Euler's identities.

+1 -

inw

ft

we

11,

by

exists.

find

n,

I = l+i + | + i+-+^-ln ^ 1
+1 + 1 + 1+ ... +l_in jg bounded by

1,

in

(H1\

By

and

1.

integrating the inequality

-^ g 1 g 1

with

we have

^in
so that S.

is

monotonic decreasing.

infinite

product Yl{l

+ ui,), where Uk>0,

is

converges

''"'

equation

is

= 3.

^""^^^ ^""^ monotonic decreasing, it has a limit. This limit, denoted


by
and is called Euler's constant. It is not yet known whether

Prove that the


By

the series converges

'^

.
.

50.

+1
S 0,

^%3

' + ^'
/'i
\ ^
Y x' x' \
(^l---...j
+ ,^^__+^_...-)

The

a;.

Problem

= -1^ -

_)_

(i),

'

3,

from which we have on replacing

circle

cos X

^ + ~+...,

+|^ + J-O + i+
^

1/x in

Thus the sequence

2 jSJ^

is

and on the

+ ^++ -

ix+
+ tx+
+

g/

3.

series for e^ holds for

e'^l +

in

^im

= 3.

\z\

e"
Letting

>

\z\

the series of absolute values

series converges within

3"

and diverges for

3,

convergent and thus convergent for

48.

z
5 -^r-^

Test for convergence the complex power series

for

(J)

of Problem 28, Chapter

if

+x S

e'

tor

x>0,

equal

'^u^ converges.

k=
4,

y, is

rational or not.

so that

Pn

- n(l + Wfc)

(1

+ Wi)(l + M2)

(1+ M)

e"i

e"2 ..- e"

e"i+"z +

"-+%

fc=i

Since M1 + M2+ ... converges, it follows that


so has a limit, thus proving the required
result.

is

a bounded monotonic increasing sequence and

51.

[CHAP.

INFINITE SERIES

252

1-1 + 1-1 + 1-1+--- isC-1 summable

Prove that the series


The sequence

Then

Si

sums

of partial
Si + S,
2

+ _
- ^

1,

we

Continuing in this manner,

52. (a)

1 1/2
^ %/(2i

- 1)

Prove that

if

1, 0, 1, 0, 1, 0, ...

is

if

is

if

n.

is

even
odd

>

'

obtain the sequence

^^^^

and p >

= 1^

jj^^

p{p

+ l)

to 1/2.

+ S2 + S, _
~
3

S1

+ +

_
~

3
1, |,

2
'

'

'

|, f. |, i-

'

'

the nth term being

and the required result

follows.

0,

g
J-^t"

(b)

Use

i.e.

(a)

(a) to

Jl

f?

.. p(f>

+ l)---(p+w-l) l

prove that

the series on the right

an asymptotic expansion of the function on

the, left.

we have

Integrating by parts,
/.

is

^du

lim

Hm

e-'u-'-du

i(-e-")(Jt-'')''

(-6-")(-pM-''-')dMl
J"

dw
^

p/p+i

a:"
a;

Similarly,

e
~
- ^ri

/p+i

(P

+ 1) ^"^^

so that

a;

By continuing

(6)

Let

S(a;)

Then

in this

e-|- "

Rn(x)

\Bn{x)\

p(p

manner the required

f(x)

^p+.

Sn(x)

"

result follows.

("l)

^i^J"

(-l)+p(p

+ l)...{p + n) r -If^dw

+ 1) (? + n)

e-'dw
J

e-'dzt

1.

Thus

z^r^du.

p(p

+ l)...(p + w) f

p(p

+ l)---(p + w)
_p+ll+l

since

Now

^^^dM

CHAP.

INFINITE SERIES

11]

lim \x^BAx)\

and

follows that

it

B. (a;)

lirn^

Mn
|Mn+l|

,.

lltTl

_
~

^^^ *^

+ ^]---iP + ^)

Hence the required

0.

diverges for

'"*^

result is proved.

p{p + l)---(p + w)/a:P+"+'


P(p + l)---(p + i-l)/a;''+"

^^^

Note that since

PiP

lim

253

we have

for

all fixed x,

x by the ratio test

all

Supplementary Problems
CONVERGENCE
53.

and

DIVERGENCE

Prove that the series

(a)

(6) find its

sum.

of

i:

SERIES

3-7 ^ 7-11

Ans.

of

CONSTANTS

11-15

...

i-

{4n-l){4n

.-f.

+ 3)

<=<'^''^^Ses

and

1/12

(6)

54.

Prove that the convergence or divergence of a series is


not affected by (o) multiplying each term by
the same non-zero constant, (6) removing (or
adding) a finite number of terms.

55.

If

56.

57.

2m and

Prove that the series


Find the
i - (1 -

fallacy:
1)

(1

^^et

and

respectively, prove that

(fP

- 1 + 1 - 1 + 1 - 1 + ...
=0. Hence 1 = 0.

f- 1) +

(1

(f)'

5(|)"

5(w

+ i;)

converges to

A +B.

diverges.

Then

5 =

(1

- 1) - (1 - i) -

...

and

QUOTIENT TEST

and

Test for convergence;

Ans.
59.

- 1) +

COMPARISON TEST
58.

converge to

2t;

(a) conv.,

(6) div.,

(c) div.,

Investigate the convergence of

(d) conv.,

(a)
^

^w^

(e) div.,

5n

= I w(m2

(/)

-2

1)^'2

lo;

'

conv.

60.

Establish the comparison test for divergence (see Page


225).

61.

Use the comparison


^"'^

.?.

converges

test to
if

62.

Establish the results

63.

Test for convergence:

p>

and diverges

(a) conv.,

(6) div.,

1,

(6)

Page

^^IlpL

diverges,

225.

'

Ans.

if

of the quotient test.

(c)

+QT1-1
('^^It' wiV.tan-MW),
l/-

In~n"

lon'

'^^-

conv.,

converges.

(6)

prove that

and

(6)

\ 2 +

..

,-fJ

/I

(c) div.,

/S\

/\
(0)

(d) div.

11^;^,
n(l +
fi

(d)

l.sinMlM).

(c)

div.

2m converges, prove that lim nun

64.

If

65.

(a)

0.

Test for convergence

2 -^-

Does your answer to

INTEGRAL TEST

>

^"'""'.

67

Prove that

68.

Prove that

contradict the statement about the

(a)

Ans.

1?

(a)

Ams.

(a) div.,

where p

./

=2 %(ln n)"

(6) conv.,

is

J:^ii(hiW

conv., (d) conv.,

(c)

a constant,

(a)

converges

() div.,

if

_
'

-iS^'

^^'

='

(/)

> 1 and

'=^

V^

'

'^

div.

(ft)

diverges

if

i.

2 :;^ < J-

<

^^a^^TTTi'

(&)

(a) div.

Test for convergence:

^^^

69.

11

n-*co

p series made on Page 225 that 2 l/n" converges for p

66.

[CHAP.

INFINITE SERIES

254

Investigate the convergence of


conv.

Arts.

70.

(a)
(b)

Provethat

Use

(a) to

+ V^ =
+ i ^ VT + V2 + Vs +
estimate the value ofVl + V2 + V3+--+

Show how

(c)

(b)

671.5

Test for convergence:

2
=2

'"

I-

VTOO, giving the

maximum

error.

+ a/IOO
can be improved by estimating, for example, V^O + -/ll +
accuracy.
of
degree
desired

some
to
computed
/2
+
+
+ \/9

(~1)"V^.
ln%

(a)

Aws.

%^^
(a) conv.,

,,-,,

(6) conv.,

(c) div.,

1 ^3^^,

(^)

(d) conv.,

(e)

(d)

What is the largest absolute error made in approximating the sum of the
Atis. 1/192
the sum of the first 5 terms?
which must be taken in order that 3
(6) What is the least number of terms

Prove that

(b)

How many

accuracy?

74.

and

S -

^+

23

+ 33+""

""

i^

2'

3'

1)

^^

/
in order to calculate

to six decimal place

CONDITIONAL CONVERGENCE

(0

(a) abs. conv.,

Prove that

decimal place accuracy

Test for absolute or conditional convergence:

Ams"

(1)"

2^ 2"(n

'

terms of the series on the right are needed


Ans. (6) at least 100 terms

<a)^^^

75

series

Ans. 8 terms

(a)

ABSOLUTE

("D" sin--,

div.

(a,)

will result?

73.

"

r_1^+

^2;^,^^^.

72.

(6)

4.5

ALTERNATING SERIES
71.

the accuracy in

and adding on the value of Vl


Ans.

"'

|n='^

=i

(6)

TT'"?
a;^ + n^

2^

cond. conv.,

(c)

(e)2^sinA,

cond. conv., (d) div.,

converges absolutely for

all

real

a;

(e)

and

abs. conv.,

a.

(/)

abs. conv.

CHAP.

+ 1^ 1 +

If 1

76.

INFINITE SERIES

11]

converges to S, prove that the rearranged series 1


Explain.

!+ ^=|S.

Take 1/2

[Hint:

term

255

of the first series

to the first series.

and write it as
Note that S = In 2, as shown

A +

in

0-J +

Problem

+ +a 4 + i + l-J- + la + JU
lY
5

+ J+.--;

then add term by

96.]

Prove that the terms of an absolutely convergent series can always be rearranged
without altering
the sum.

77.

RATIO TEST
Test for convergence:

78.

Ans.

conv. (abs.), (6) conv.,

(a)

79.

Show that

80.

Prove that

(c) div.,

conv. (abs.),

(d)

(e) div.

the ratio test cannot be used to establish the conditional convergence


of a series.
(a)

2 ~ converges

and

lim

(6)

n-+oo

0.

'W.

MISCELLANEOUS TESTS
81.

Establish the validity of the nth root test on Page 226.

82.

Apply the wth root


Prove that

83.

|+

test to

(|)=

Test for convergence:


Ans. (a) div., (6) conv.

84.

If a, b

85.

work Problems

(i )^

(a)

(|)4

86.

of

if

-a >

d,

and diverges

(e).

converges.

2-5
,,,2^2-5
2
r^^

2-5-8

9-12 '9- 12 -15

> a, prove that


+
a(a + d){a + 2rf)
"^
b{b + d) '^ 6(6 + d)(6 + 2d)
b(b
if 6 - a g d

rf)

""

FUNCTIONS

<->.l,5. < .l,i-.igfef

87.

'^

and

a{a
a(g

Find the domain of convergence of the

Aws.

(|)

(d)

(c),

-^'^'^

'^''^

"^

and d are positive numbers and

converges

1-4
1-4 -J?
1
^
+ ^~^
+ illll
+

a^
a

SERIES

(ly

78(a),

()

-l^si,

Prove that

(6)

series:

i^^i,^..

~l<xS3,

^'^if-^^Ql^ ^^

(e)

all

x^O,

converges for

.|,.-(fS)". <.)i;.^=n
(d)

x>0,

-1 g

<

(e)

a;

S 0.

1.

UNIFORM CONVERGENCE
88.

By

use of the definition, investigate the uniform


convergence of the series
n=i [1

[Hint:
1

(n-l)x]{l

+ nx\

Resolve the nth term into partial fractions and


show that the nth partial

sum

is

S.{,x)

+ nx''

^'^'

''"''^'"^^"*

^thJ^nU^ll

"

^"y

^"^^"^^^

^^^"^ mcWAes

a;

= 0;

uniformly convergent in any

[CHAP.

INFINITE SERIES

256
30 directly by

obtaining Sn(x).

89.

Work Problem

90.

Investigate by any method the convergence and uniform convergence of the series:

conv. for \x\ <


not unif. conv. for

Ans.

91.

If

F(x)

(a)

Fix)

is

^^,

first

unif conv. for |a;i S r < 3.


but unif. conv. for a; S r

3;

(a.)

>

all x.

(c)

conv. for

x^O;

0.

prove that:
all x,

(6)

lim F(x)

92.

Provethat

Q^ ^ S^ + ^-f^

93.

Prove that

Fix)

94.

Examine the sequence

95.

Provethat

has derivatives of

all

-,

= j^f^. =

u^x)

1.2,3,

F'(x)

(c)

0,

<^x

^"!^"'
n=i Sinn fiTT

dx
'^^
r'
J^ j^^^j^.

lim

unif conv. for

(6)

x^O,

continuous for

11

^^

continuous everywhere.

is

0.

orders for any real x.

for uniform convergence.

.. .,

6-'

POWER SERIES
(a)

Prove that

ln(l

(6)

Prove that

In 2

[Hint:

Prove that

98.

Evaluate
Ans.

+ a5) = ^~Y'^Y~'4"^
= 1- ^ + i-i+

^-^ = 1 - x + x^ - +

Use the fact that

97.

sm"' x

(a) 0.461,

(6)

a.

f'^e-^dx,

(a)

X^

X^

96.

a''

a.

s"

j^'

(6)

3 ^'

2^

4.

~5

xi
7

^
,

Evaluate (a) sin 40, (6) cos 65, (c) tan 12


Ans. (a) 0.643, (6) 0.423, (c) 0.213

100.

Verify the expansions

101.

By

102.

Showthat

103.

Obtain the expansions

Let
exists

/(x)

4,

[Hint: See

x'

tanh-;

(6)

ln(x

but that

j^"'""
it

_i

'

''

-^

<
-

J.

<
-

correct to 3 decimal places.

31x

4x*

r^-r^v
V^ + l)

a;#0

eT"^"/'

2 sin

a;

cos x

38,

Chapter

sin 2x.

^ ^^<r ,
-<*'<-

-1< x < 1

^+^ + f+--

-2

1 x

3"+

x''

1-3-5

x'

p^ove that the formal Taylor series about x

4.]

-1

2^ J" 2^T^T

xt^O.
does not converge to the given function for any

Problem

1
^

and 6 on Page 231.

= (^1-2T + T!

(a)

'

to 3 decimal places, justifying all steps.

multiplying the series for sin x and cos x, verify that

e-"

integrate.]

0.486

99.

104.

1-3-5
2-4'6

Lll|^^dx

and

<
=

a,
a;

1
i

corresponding to /(x)

CHAP.

105.

INFINITE SERIES

11]

257

Prove that

^^

()

+ .)r

{ln(l

(6)

-(l

+ (l+| + |)=c' -

x^-(l+i)^+(l+| + |)?|!-... for-K.gl

+!)''

for-l<a.<l

MISCELLANEOUS PROBLEMS
Prove that the series for

106.

Mx)

converges

for all x,

(a)

absolutely and uniformly in any finite

(h)

interval.

107.

Prove that

{a)-^{Mx)} = -J,(x),

= ~Jp(x) -

(c)Jp+i(x)

108.

Assuming that the

109.

Prove that

eV4'-'/"

Write the

[Hint:

\z\

111. (a) If

2^

Suppose that
If I,

113.

114.

r^r^n;-?'-is
expansion

=1

lEi

J-^(x)

0,-1,-2,

.,

w=

(-D'Jnix),

prove that
1,2,3, ...

expand and use Problem

108.]

absolutely and uniformly convergent at all points


within and on the

for

all

=.' ^' 2'


unique.

v'R =

(c)

common

in the

Use

to

(a)

interval of convergence

show that

if

\x\

<R

the Taylor expansion of a function

Prove that 2m converges or diverges according


as

L.

L<1

or

to find the radius of convergence of


the series in

Prove that a necessary and

Use

(c)

How

[Hint:

prove that the series

<?iir>i
thot
such
that

2 TTT^
criterion,

"' '"''"f~W ""<

->'"

p/^'^i'i"'^
F(-p,l;l.,-x)

(1

l<j

|&p

Show

that S'ix)

<:

limits,

converge

i,
whenever

is that,
"lai,,

p>N
-.

e ^
> u,
> iV, where

given any

and g

converges.

2-

Page

(fw 282)

_ S,| ^
<

>1

22.

diverges'

43.]

() is

.b.l.t,ly convergent fcr

!.|

<

1,

(6) i,

dlvg,l

hypergeometric function defined by the series on


Page 232, prove that
(,) xF(l,l;2;-x) = ln(l + .), (.) f(i,|;|;.^) =
(sfn-.)}..

+ .)..

Find the sum of the series


[Hint:

f
^

Use the Cauchy convergence

the

"^Jf^
(a)

on
"

could you use (a) to prove that the series

>'l'

Problem

sufficient condition that the series 2e.

deDendinff
"''^"'""'"^

j_

a.

(a) to

""'

f^lrf

N>Q

find

<?-+

(h)

i^

the test fails.

Use Problem 113

we can

B>0,

where

Prove that the radius of convergence of the series


2a can be determined by the following
when they exist, and give examples: (a) liml-^^l,
^
1
u) ij^
(6) lim

115. (a)

118.

x^ J,.,{x),

t^

e"'' e-'',

^^

6a;"
**

exists, the
the

112.

J^ix),

^p(a;)

left side as

aa;

2^^:p2J^

1.

Problem 107(c) holds for p

JMx) =

(6)

no. Prove that


circle

J^-i{x).

result of

= -Mx),

J-,(x)

(a)

j-{x^ J,(x)}

{h)

S(x)
l

xS{x)

-^ +
1*3

and

'^^

j-

l'3-5 ^

solve.]

'

Ans.

..
e'^'^

Jo

e-'^'^dx

[CHAP.

INFINITE SERIES

258

11

Prove that
at

119.

^'"^'^

Prove that

Use the

[Hint:

-4!"^

uniformly convergent in any interval which does not include

is

n=i

2*

on Page 228.

120. Establish the Dirichlet test

121.

~ ^V^~^'^2^-2!'5"2-3!-7"^

lT3""'"l'35"^ l-3-5-7'^

Page

Dirichlet test,

and Problem

228,

Chapter

94,

0, ff, 2?r,

1.]

on Page 232 concerning the binomial series.


Examine the Lagrange and Cauchy forms of the remainder

122. Establish the results

[Hint:

123.

Prove that

124.

Prove that

125. If

a;

converges
2
x^
n+r-r1 -^ + ^- ^+
4
7
10
^^

= ye",

^1

prove that

1/

11=1

126.

Prove that the equation

uniformly for

3y^

("1)" '^"
ni

Let

^
e''

has only one real root and show that

Show that

(a)

numbers, satisfy the recursion formula {B + 1)" - B"


expanding. (6) Using (a) or otherwise, determine Bi,
Ans.

(6)

Bi

= -^,

- ^

128 (a) Prove that


if

129.

ft

= 1,

^2

= 0,

Bz

Bi

= -^,

= -(coth|-2^
^

^:"i

1,2,3,...

it is

given by

" (-l)'-'n"-^e-

+^^
+ ^^+ .
3
2

= 1+ Bix
1

<x^ lie.

-1/e

for

x"

_
127.

but not absolutely.

all x,

^ + |-ln2
3

= X-1

e"'^

in Taylor's theorem.]

1).

= 0,

Bs

Be

=
.

the numbers B, called the Bernoulli

where B"
.,

formally replaced by Bh after

is

Be.

=h

Use Problem 127 and part

(6)

(a)

to

show that B2h+i

Derive the series expansions:

,x

x'

B2n

(a)

cothos

^+ 3-^5+

(&)

coto;

^-3-45

(-1)

[Hint:

For

for

use esc

(d)

130.

Prove that

131.

Use the

132.

Prove that

(a)

fl

use Problem 128; for

a;

f1

cota;

+ -j

(2a;)''"

+-(2^I)Tir+

-(2;oiir+

(6)

replace x by ix in

for

(a);

(c)

use

tana;

cot a;

2 cot 2a;;

tana;/2.]

converges.

definition to prove that

(l

+-)

diverges.

00

n=

133. (a)

Prove that

(1

- *").

where

< m <

1,

converges

if

and only

if

2m converges.

11

"^

converges to i.

(6)

Evaluate the

infinite

product in

(a)

places and compare with the true value.

+ 0-1 + 1 + 0-1 + 1 + 0-1+--

134.

Prove that the series

135.

Prove that the C6saro method of summability

is

regular.

is

C-1

[Hint:

summable

to zero.

See Prob. 28, Chap.

3.]

to 2 decimal

CHAP.

136.

137.

INFINITE SERIES

11]

Prove that the series

series

is

+ 2x + 3x' + 'ix^+

...

Abel summable to S

called

+ nx"-' +
S =

if

(-l)"(n

^
2,

(-!)"(

n=

th\
(6)

1)

~+^-^

l){n

is

+ 2)

'-

IS

Prove that the double series


as p

>

Abel, summable to

1 or

140.

Use

(a) to

On*"

for

|a;|

< 1.

Prove that

exists.

J^

CO

2 (^.

1/8.

.|

+ ^y

where p

is

a constant, converges or diverges according

1 respectively.

= l-J- + 2!_3!

f!Zd^
(6)

- x^

Abel summable to 1/4 and

00

138.

lim

converges to 1/(1

X-+1- n =

00

(a.)

259

prove that

C ?^du
M

J.

(-ir-'(n-l)!

r^-".,

i--l+21_31 ^
x
a;=

a;

a;*

Prove that

f'^iHiidw
M

^SlULdu
X

141. If /(a;)

9l^{l_2}_,4l^_
a;'
x*
^
\

2= ^^^
% 1

~ 21^( 1-3!, 5|_ \ _


X \x
x^^x^
"J

has an asymptotic expansion given by

expansion

?\

sinx/l_3! 5!
X \x
x^'^x'

'")

sing /
X V;

4!

x^"*"^

'")

-, prove that

2!

f{x)dx

has the asymptotic

chapter 12
Improper Integrals
DEFINITION

of

IMPROPER INTEGRAL

an

dx

f{x)

is called

an improper integral

if

(1)
(2)

a=-a>

or &

or both,

<

one or both integration limits

i.e.

one or more points of

fix) is unbounded at
larities of f{x).

a^x^b.

...

is

infinite,

Such points are

called singu-

integrals of the first and


Integrals corresponding to (1) and (2) are called improper
and
conditions
(2) are called improper
both
with
(1)
Integrals
second kinds respectively.
integrals of the third kind.
sin x^

Example

2-

-^

Vx

J,
4:

is

is

an improper integral of the

first kind.

an improper integral of the second kind.

00

Example

dx

dx

is

^^^dx

an improper integral of the third kind.

is

a proper integral since ^Hm


I_>0+

sm

convergence or divergence of improper mtheorems bear close analogy to


It will be found that such tests and proofs of
tegrals.
for infinite series (see
theorems
corresponding
convergence and divergence tests and
Chapter 11).
In this chapter

we formulate

IMPROPER INTEGRALS
Let

f{x)

of the

tests for

FIRST KIND

be bounded and integrable in every

f f{x)dx
The

integral on the left

series

ztn

where Un

infinite series.

Similarly,

We
we

Note that
while

f{n),

often write

lim

f{x)dx

a^x^h. Then we

define

convergent or divergent according as the limit on the

is called

right does or does not exist.

finite interval

bears close analogy to the infinite

f{x)dx

f{x)dx corresponds to the partial sums of such

in place of b in {1).

define

f{x)dx

lim

260

f{x)dx

(2)

CHAP.

IMPROPER INTEGRALS

12]

261

and call the integral on the left convergent or divergent according as the
limit on the
right does or does not exist.
Examplel:

Example

2:

cos

exist,

= J

lim

cos X dx

we

In like manner,

da;

a;

Hm

f"^

cos

"

(^

da;

a;

l)

=
lim

so that

(sin

?^

converges to

1.

Since this limit does not

sin a).

divergent.

is

define
/^^

y^^O

fix)dx

-^CO

j_^f(x)dx +

f{x)dx

(3)

where

Xo is a real number, and call the integral convergent or


divergent according as the
integrals on the right converge or not as in definitions
(1) and (2).

SPECIAL IMPROPER INTEGRALS


1.

Geometric or exponential integral


if

t>0 and

diverges

if i

S 0.

FIRST KIND

of the

e'^da;,

where

a constant, converges

is

Note the analogy with the geometric

series if r = e"'

so that e~'^==r^.

~,
verges

> 1 and

if

diverges

CONVERGENCE TESTS

for

if

g 1.

where p

is

Compare with the v

IMPROPER INTEGRALS

a>0,

a constant and

of the

con-

series.

FIRST KIND

The following tests are given for cases where an integration limit
is oo.
Similar
tests exist where an integration limit is -=o (a change
of variable x = -i/ then makes
integration limit

Unless otherwise specified we shall assume that


and thus integrable in every finite interval aSccSfe.
1.

Comparison
(a)

verges.

(6)

Let g{x)^0 for

Then

if

Since

Divergence.

f{x) is

for integrals with non-negative integrands.

test

Convergence.

Example:

the
continuous

=o).

all

0^f{x)^g{x) for

^ ^ = ^^

and

Let g{x)^0 for

all

x^a, and suppose


all

f e-dx

a;

S a,

C f{x)dx

converges,

x^a, and suppose


f{x)dx

that

con-

also converges.

f^
that

j'" g{x)dx

also converges.

C g{x)dx

diverges.

also diverges.

Example:

Since

3^

diverges.

>i

for

a;

2 and

f^

"^2

diverges (p integral with p

f^
da;

1),

J,

In:

also

[CHAP. 12

IMPROPER INTEGRALS

262

2.

Quotient test for integrals with non-negative integrands.


If fix)

(a)

g{x)

and

0, and

g(x)

if

= A ^

4^

lim

or ,

then

dx and

f{x)

either both converge or both diverge.

dx

(b)

If

A=

(c)

If

A = 00

Theorem
j

(ii)

dx converges, then

then

dx

g{x)

diverges,

a;" /(a;)

dx converges
diverges

fix)dx

if

if

A.

>

f{x)

dx converges.

dx

diverges.

Then
and

pgl

dx

x'

X 4^4

and

is finite

A^O

x^

infinite).

+ 25

'

"-^"

+l

be

X
Ax*

lim x

diverges since

Vx* +

may

(A

lim X

converges since

+ 25

Va:*

Similar tests can be devised using gix)

3.

f{x)

is

Let lim

1.

fix)

(i)

and

in (a)

9{x)

related to the comparison test and is often a very useful alternative to


In particular, taking g{x) = l/ic", we have from known facts about the v integral, the

This test

it.

and

in (a)

a;''

1.

+l

e~*^.

Series test for integrals with non-negative integrands.

fix)

J^

dx converges or

diverges according as 2m, where Un-fin), converges or diverges.

4.

Absolute and conditional convergence.


converges.

C\fix)\dx

if

fix)dx

If

fix)

dx

is called

converges

fix)dx

but

absolutely convergent

\fix)\dx

diverges,

called conditionally convergent.

is

Theorem

2.

If

^^

\fix)\dx

converges,

then

fix)dx converges.

In words, an abso-

lutely convergent integral converges.

Example

1:

f -^^^dx
J

^^1

is

^"'^

absolutely convergent and thus convergent since

Jo

-^n

^HL^ dx converges
X

5^2^ dx

is

^^

M^*

erges.

(see Prob. 11),

but

^0

^i?^ dx
I

does not converge (see Prob. 12).

conditionally convergent.

Any

of the tests used for integrals with non-negative integrands can be used to

test for absolute convergence.

CHAP.

IMPROPER INTEGRALS

12]

IMPROPER INTEGRALS
we

of the

define

on the right of
divergent.

If the limit
it is

Similarly

a^x^b,

if

=a

aSx^b,

of the interval

f{x)dx

(4) exists,

=
we

lim
call

f<x)dx

then
(A)

the integral on the left convergent; other-

becomes unbounded only at the end point

f{x)

we

then

a;

(,

wise

SECOND KIND

becomes unbounded only at the end point

If f{x)

263

a;

=&

of the

interval

define

f{x)dx

lim

f{x)dx

-.0+ tJa

(5)

In such case the integral on the left of (5) is called convergent or divergent according
as
the limit on the right exists or does not exist.
If f{x)

we

then

becomes unbounded only at an interior point x

f{x)dx

The

= Xo

of the interval

aSxSb,

define

lim

fix)dx

lim

f(x)dx

(6)

integral on the left of (6) converges or diverges according as the limits on the right

exist or do not exist.

Extensions of these definitions can be


or more points of the interval aSx^b.

made

in case f{x)

becomes unbounded at two

CAUCHY PRINCIPAL VALUE


It

may happen

zero independently.

that the limits on the right of {6) do not exist when e^ and
In such case it is possible that by choosing t^ = e^ = e in (6),

f{x)dx

^limjj'"

f{x)dx

+ C

e^

approach

i.e.

f{x)dxi

writing
(7)

the limit does exist. If the limit on the right of (7) does exist, we call this limiting
value
the Cauchy principal value of the integral on the left. See Problem 14.

SPECIAL IMPROPER INTEGRALS


-,

C"

J
o

dx
(x

(b

- x)"

SECOND KIND

it

p<l

and diverges

if

^'o^verges if

p<l

and diverges

if

p^l.

<*''"

verges

dx

r'

a)"

of the

g 1.

These can be called p integrals of the second kind.

Note that when

p^O

the in-

tegrals are proper.

CONVERGENCE TESTS
The following
the interval

IMPROPER INTEGRALS

tests are given for the case

a^x^b.

where a<Xo<b.

for

where

Similar tests are available

of the

if f{x) is

SECOND KIND

unbounded only at x = a in
unbounded at a; = 6 or at a; = a;o

f(x) is

[CHAP. 12

IMPROPER INTEGRALS

264

(a)

with non-negative integrands.

test for integrals

Comparison

1.

Let g{x)

Convergence.

Then

verges.

SO

a<x^b,

for

for

(i^f{x)^g{x)

if

and suppose that


r"

a<a;gb,

~=

dx

con-

also converges.

/(a;) da;

g{x)

J,

(&)

= -!),

Divergence.

converges (p integral with a

1,

also converges.

Let

g(a;)

SO

for a

<x^

and suppose that

b,

g{x)

dx diverges.

also diverges.

f{x)dx
a

Example:

-^^ > ^^^


P

= 4),

"^^ dx

(^

^3

for

a;

> 3. Then

since

j -^^^

diverges (p integral with a

= 3,

also diverges.

"^l

Quotient test for integrals with non-negative integrands

2.

(a)

If

a < a; g 6, and
SO for a<a;g6,
/(x)SO
and gix)
gix)^0
fix) S
fix)

dx and

(b)

If

A=

(c)

If

A = 00

in

gix)

and

(a),

and

in (a),

^
fix)

= A^

lim
lim^

if

or

>,

then

dx either both converge or both diverge.

gix)

dx converges, then

fl'(a;)

dx diverges, then

fix)

fix)

dx converges.

dx diverges.

This test is related to the comparison test and is a very useful alternative to it. In
particular taking gix) = l/(x - a)" we have from known facts about the p integral the

Theorems.
(i)

Let

lim (x-a)''/(a;)

A.

Then

x~*a +

fi^)

d^ converges

fix)

dx diverges

if

< 1 and A

is finite

S 1 and A ^

(A

if

may

be

infinite).

If fix)

becomes unbounded only at the upper limit these conditions are replaced by

those in

Theorem
(i)

(ii)

fix)

lim ih -xYf{x)

Let

4.

dx converges

/(a;)da;

diverges

if

if

"^

Example

1:

Example

2:

?)<

pSl

B.

and

and

is finite

B^Q

converges since
^

Then

iB

may

^lirn^ (x

diverges since

be

- !)"'

lim

(3

infinite).

,4

- x)

_ ^y,^ ^^^ \

a;*

1 ~

";==

"2"

CHAP.

IMPROPER INTEGRALS

12]

fix)
if

Theorem
If

converges.

\fix)\dx

J^
then

dx

is called

absolutely convergent

"

"

265

If

f(x)dx

converges but

C"

"

^b
f(x)

dx

\f{x)\dx

diverges,

"^o

called conditionally convergent.

is

5.

\f{x)\dx

converges, then

f{x)dx converges.

J^

In words, an absolutely con-

vergent integral converges.


Example:

sin X

Since

it

\
'

^
-=

W^^\

follows that

lutely),

Xdx
X _
/

and

y/^r^

converges (p integral with

3.

^/

k^do,
IVa;-ii-|

J
'^n-

Arr
"^

'

converges and thus

f
^^

^^do.

a-v, p = J),

converges (abso-

\lx-Tr

Any of the tests used for integrals with non-negative integrands can
be used to test
lor absolute convergence.

IMPROPER INTEGRALS
o
IS

of the

THIRD KIND

Improper integrals of the third kind can be expressed in terms


of improper integrals
the first and second kinds, and hence the question of
their convergence or divergence
answered by using results already established.

IMPROPER INTEGRALS CONTAINING


CONVERGENCE

PARAMETER. UNIFORM

Let
^()

f{x,a)dx

(5)

This integral is analogous to an infinite series of functions.


In seeking conditions under
which we may differentiate or integrate ,^(a) with respect to
, it is convenient to introduce the concept of uniform convergence for integrals by analogy
with infinite series.

We

shall suppose that the integral (8) converges for

a^^a^a^, or

briefly [a^,a,].

Definition.

The integral (8) is said to be uniformly convergent


number N depending on e but not on a such that

in

[a^,

aj

if

for each

find a

4y{a)

J^

fix, ot)dx

This can be restated by noting that


analogous

an

The above

<

.^()

for

infinite series to the absolute

all

M>

iV

and

C fix,a)dx

all

a in

[a^,

>

we can

aj

C fix,a)dx

value of the remainder after

which

is

terms.

and the properties of uniform convergence to be developed are


formulated in terms of improper integrals of the first kind. However,
analogous results
can be given for improper integrals of the second and third kinds.
definition

IMPROPER INTEGRALS

266

SPECIAL TESTS
1.

for

UNIFORM CONVERGENCE

Weierstrass

M test.

(a)

g M{x)

\f{x,a)\

we can

If

of

INTEGRALS

M{x)

find a function

a^^a^a^,

[CHAP. 12

such that

X>a

M{x) dx converges,
a

uniformly and absolutely convergent in

f{x, a)

dx

Since

P^^ =

is

a^^a^ a^.

Example:

is

As

^"*^

"iTTT

converges,

uniformly and absolutely convergent for

follows that

it

values of

all real

da;

a.

possible for integrals to be uniformly


convergent without being absolutely convergent, and conversely.

2.

in the case of infinite series,

Dirichlet's test.
(a)

4i(x)

it

is

Suppose that

a positive monotonic decreasing function which approaches zero as

is

a;-^ 00

(b)

\f{x,a)dx < P

for

u>a

all

and

a^

f{x, a) ip{x)

^ a S a^.

dx

uniformly convergent for

is

THEOREMS
Theorem

on

a^^a^ a^.

UNIFORMLY CONVERGENT INTEGRALS

6.

S a and a^^a^ a^, and

If f{x, a) is continuous for x

f{x,a)dx

is

if

f{x, a)

dx

continuous in

is

uniformly

a^'^a'^a^.

In

particular, if a^

is

any point of

aj

^ a g a^, we
f(x,

If

ttj

is

Theorem 7.
Under
a

to a

one of the end points,

we

a)dx

a)

dx

(9)

we can

6,

integrate ^(a) with respect to a

from

to obtain
I

4,{a)da

-^1

/(X,

a)dx^da

which corresponds

Theorem

to a

1)1

f{x,a)daydx

(10)

change of the order of integration.

8.

If f{x, a) is continuous

and has a continuous partial derivative with respect

dx
not oepena on

a,

converges uniformly in

depends on a this result

a^

g a g a^,

then

to a for
if

a does

"

f
da
If a

lim f{x,

use right or left hand limits.

Theorem

the conditions of

can write

is easily

ur^^dx
a oa

modified (see Leibnitz's rule, Page 163).

(11)

CHAP.

IMPROPER INTEGRALS

12]

EVALUATION

of

267

DEFINITE INTEGRALS

Evaluation of definite integrals which are improper can be achieved by a variety


of
One useful device consists of introducing- an appropriately placed parameter
in the integral and then differentiating or integrating vs^ith
respect to the parameter,
employing the above properties of uniform convergence.
techniques.

LAPLACE TRANSFORMS

F(x)

JL{F(x)}

The Laplace transform of a function


F{x)

is

f{s)

J:{F{x)}

e-'-F{x)dx

One
is

sin

n =

useful application of Laplace transto the solution of differential equa-

'-^'^

ax
s^

+ a^

s'

+ a'

^^

sJ:{Y(x)}

Y"{x)

^ n

s>0

1,2,3,...

Y'ix)

Problems 34-36).

tions (see

cos ax

x"

*>o

e"

{12)

and is analogous to povs^er series as seen by


replacing e'^ by t so that e""^ = i^. Many
properties of power series also apply to
Laplace transforms. The adjacent short table of Laplace transforms is useful. In each
case a is a real constant.
forms

defined as

s'j:{Y{x)}

Y{0)

8Y{0)

Y'{0)

IMPROPER MULTIPLE INTEGRALS


The

definitions

and results for improper single integrals can be extended

to

improper

multiple integrals.

Solved Problems

IMPROPER INTEGRALS
1.

Classify according to the type of improper integral.

dx
\/x{x

_xdx

dx

x^dx

x^ + 1

(a)

Second kind (integrand

(6)

Third kind (integration limit

(c)

This

(d)
(e)

is

unbounded at
is infinite

a;

and x

= 1).

and integrand

is

3SxS 10).

First kind (integration limits are infinite but integrand

This

is

a proper integral (since

''
lim

~
'

cos X
a;

unbounded where tan x

a proper integral (integrand becomes unbounded at

integration

dx

X''

+ tanx

is

COSX

+ 1)

is

a;

= 2,

but this

is

-1).

outside the range of

bounded).

_L

= ^1 by

applying L'Hospital's

rule).

IMPROPER INTEGRALS

268

[CHAP. 12

dx
J' y/x{2 -

into

(a)

(a)

an improper integral of the


Consider

r
I

dx

-'i

yV22/ -

0<e<l,

where

- a;)

V(2

kind,

first

(b)

a proper integral.
Let

say.

x)

=1

Then the integral becomes

As e->0+, we

see that consideration of the given integral

is

equivalent to con-

X/7
-

which

y^/2y

an improper integral of the

is

first kind.

1
,

We

J.i
sideration of

r
I

VeVv^ +

dv

^0

tJv^

which

are thus led to con-

a proper integral.

is

+2

From the above we see that an improper integral of the first kind m,ay be transformed into
an improper integral of the second kind, and conversely (actually this can always be done).

We also see that an improper integral


only sometimes be done).

IMPROPER INTEGRALS
3.

may

be transformed into a proper integral (this can

FIRST KIND

of the

Prove the comparison test (Page 261) for convergence of improper integrals of the
first kind.

Since

OS/(a;)

= g(x)

for

S a, we

a;

S
But by hypothesis the

Prove the quotient


By

hypothesis,

f(x)

fix)

^-^ = A >

44

= ^ +

and hence

(A

or

loss of generality in choosing

g(x) dx

>

we can

0,

gix) dx

- e) gix) S

find AT such that

r"
I

fix)
f
M^
\g(x)
I

A <
I

fix)

For the cases where

A=

A c >

and

A=

=
>,

a>

e)

gix)

(A

e)

and

p(x) da;

(1)

0.

dx exists, and so

fix) rfx

(A

/(x)dx

r"
I

fix)

so

(i),

dx converges

diverges, then by the inequality on the left of

lim

f(x)

converges, then by the inequality on the right of


lim

dx converges

r(x)dx

no

If

f(x)

Then given
any

0.

(A-e) r

If

dx

Then

is

g(x)

Thus for x'^N, we have

A -

There

g(x) dx

81,

Thus

exists,

i_fr(x)

when x'^N.

Page

7,

on Page 262.

test (a)

lim

dx

dx

last integral exists.

lim

4.

f(x)

have using Property

fix)

(i),

dx diverges

see Problem 41.

As seen in this and the preceding problem, there


proofs for infinite series and improper integrals.

is

in general a

marked

similarity between

CHAP.

5.

IMPROPER INTEGRALS

12]

Test for convergence:


(a)

Method

(a)

For large

^'"*^

x, the

""'"^

f^ ^^^tx^ +
integrand

+ 5a;2 + 1 " 3^'

3a;*

Sx^

2:

verges,

Let /(.)

^"'^

+l

f(x)

3^,^^^,

sT^

converges.

(6)

Method

Method^:

Let

j-/

integrand

^ ^^-

converges (p integral with p

integrand for large x

"the

Since

= 3),

it

follows

we have

U"?

,(.)=!.

and

obtain a suitable

/%(.) d.

con-

test.

discarded the factor I.

approximately x'/^/^

''""

to

is

/g =1,

lim

Hence by Theorem

l/3a;^

^^ converges.

ff{x),

|.

is

'''''''''

Then

1,

Page

1/x.

how-

262, the required integral

f.^r/'^
1

since

It could,

1,

also diverges.

\(x) dx

and

dx also diverges.

]im

Since

3:

/(.)=-|^.

f(x)

x, the

Ml

^''''''^''

Method

>

For large

diverges,

g^.^
+ "-

3^4

J" Va;" + 16 dx.


'2

dx also converges by the quotient

jirn x'

3:

^^^

'

.(.)

Note that in the comparison function


ever, just as well have been included.

Method

approximately x/3x*

is

Note that the purpose of examining


comparison integral.

Method

269

^-|=^i^^ =

a;

1,

the required

diverges by

integral

Theorem

1,

Note that Method 1 may (and often does) require one to obtain a suitable
inequality factor
(m this case i, or any positive constant less than 1) before the comparison test
can be applied.
Methods 2 and 3, however, do not require this.

6.

Prove that

e-^^

Jim xU--'

dx converges.
(by L'Hospital's rule or otherwise).

the given integral converges.

7.

Compare Problem

Then by Theorem

Chapter

10(a),

1,

with

A=

0,

= 2,

11.

Examine for convergence:


^"^

J 'xTa ^^' ^^^^^

^"^

IT^'^'ITT^ =

<^

is

a positive constant;

Theorem

"

H'^'=

l^'-'^^dx + J"L

*>y

1.

(&)

Page

-*-

262, with

""

A=

dx.
,

= l,

the given integral

diverges.

(,)

fjl^^ax
The

first

Smce
Page

,
-dx

integral on the right converges [see Problem 1(e)].

hm

262, with

cos x
x"

x^'^i^

A-0

Thus the given

and p

j =

0,

= 3/2.

integral converges.

the second integral on the right converges by

Theorem

1,

[CHAP. 12

IMPROPER INTEGRALS

270

da;,

(a)

= 2/. Then

Let X

(&)

the integral becomes

J_ ^sTi'^^r^V-

i/j

Method

1:

y^l. Then

e"" for

since

e-" dy

converges,

^dy

Ji

Ji

converges; hence

the given integral converges.

Method

Page

y^(]=

\\m

2:

262, with

A=

- f ^^f^dy.
+1
^d

becomes

^t^^

lim '(

'^'*'

9-~

Letting x

1^4^ "^^^

lim y^i

Since

2/

Since

integral converges by

Then the given

0.

Theorem

1,

= 2.

and p

^+T
it

\im ye~'

^ilK
2/

J^

"

^^''^

1'

= -yin

the

first integral,

integral converges.

the second integral converges.

1,

Thus the given integral converges.

ABSOLUTE and CONDITIONAL CONVERGENCE


INTEGRALS of the FIRST KIND
converges

f{x)dx

if

We

have

is

|/(a;)|

ing

|/(a;)|

i.e.

an absolutely con-

convergent.

fix)

|/(a;)|,

g
|/(a;)|da;

converges,

\f{x)\dx

IMPROPER

^a

vergent integral

for

converges,

i.e.

[f{x)

f(x)

we

dx, which converges,

\f(x)\

see that

[f(x)

f{x)

^ 2\m\.

^ aj^

\f{x)\]dx

follows that

it

Then

\f(x)\dx

\fix)\]

dx converges.

Hence by subtract-

dx converges.

10.

Prove that
Method
I

52i^

I
I

^ dx

converges.

1:

cos X
x-'

Ji

xM

Method

X
1
1

for

a;

Then by the comparison

1.

test, since

x'

(a;

converges,

i.e.

^^^

Ji

dx

converges,

j-

it

follows that

dx converges absolutely, and so converges by Problem

9.

x''

2:

Since

lim a;'H^^
X

a:-t-oo

= Hm
\

P^^ldx

a;-*-Qo

F^
(

-*

0,

it

follows from

Theorem

1,

Page

262, with

converges, and hence

^^^dx

converges (absolutely).

A=

and

CHAP.

IMPROPER INTEGRALS

12]

-^^dx
Since

we need

~dx

converges.

-^^^r +

SI"

f"cosx,
as

(1)

_
-

^rf^
a'

M^

j^

Since the integral on the right of

and

lim

^HL

= A

r"cosx,

cosl1

cos X

^^^ =

0,

-^dx

+ j^
1

lim

(2)

converges by Problem 10, the required result follows.

(2)

The technique of integration by parts

to establish convergence is often useful in practice.

2:

a;

!;

+ n?7,

sm X

if

-da;

the summation becomes

J,
is

cosM

and using the fact that

>

This

S1

or on taking the limit on both sides of

Letting

a;

Integration by parts yields

f^iilL^da;

Method

<

continuous in

is

^^^^dx converges.

/j

Method

(because 5HL

converges

only show that

271

an alternating

+ nw

^j^^ s

Since

series.

J_sinjy_
v
Also,

J,

lim
n-*

+ n^

^ ^^^

^^^

^
=

.
'"

r^^d*
+ wt

^0

Jo

1^

lim

i'

and sin^

t;

+ 2;r

in [0,,], it follows that

sin v

r'^

Jo

Jo

iT

(%

"^^

1)^

r"*^
njT

-i.oo ./o

Thus each term of the alternating series is in absolute value less than or equal
to the preceding
term, and the reth term approaches zero as n^ =o. Hence by the
alternating series test (Page 226) the
series and thus the integral converges.

12.

Prove that

dx

converges conditionally.

Since by Problem 11 the given integral converges,


vergent,

As

i.e.

in

\~-^ dx

Problem

2,

\^\dx

N"^

Vhi

= (d-lV
'

*
fr

n?o

(^TflV

-I

*;

2 f

?^\dx

=oJ,r

g ^ S ..

f ^^+^''"
sin'i;

'^'^^'^^^'

is

not absolutely con-

r -^r^d.
v + nv
~=oJo

^^>

Hence

^
-

OT+TV

r"
Jo

"""

dv

^^^ series on the right of

dx diverges and the required result follows.


I

it

we have

Jo
^'""^

that

diverges.

Method

11,

we must show

(n

(1)

l)n-

diverges by the comparison

(2)

test.

IMPROPER INTEGRALS
13. (a)

Prove that

The integrand

converges and

unbounded at

is

^^

li^

SECOND KIND. CAUCHY PRINCIPAL VALUE

of the

(b) find its value.

v^ + 1

^-1

a;

= -1. Then we

Determine whether

define the integral as

lim(^+i^'

This shows that the integral converges to

14.

[CHAP. 12

IMPROPER INTEGRALS

272

6,

converges

,^ _^-^y

limfe-fe-)

=^

(a)

(b) in

in the usual sense,

the Cauchy

principal value sense.

(a)

By

definition,

dx

C'

..

dx

f'-^'

dx

r'

,.

32

_,

and since the limits do not exist, the integral does not converge in the usual sense.

(6)

Since

the integral exists in the Cauchy principal value sense.

15.

The principal value

is 3/32.

Investigate the convergence of:

,,

r'

r\
(a)

Um
^hm

dx

/-r
(X

2W3.
I)

verges by Theorem

(6)

lim x'

i_>0+

(c)

^^^^

1
^,(^3

_ 8)2/3

3(i),

dx

c'

.,

Page

lim

i(

5:

J^^^ x\x^-.tx^ A J

Write the integral as

Since

lim
X-*.

Since

(a;

lim

V(5-)(a;-l)

Hence the

3(m) on

Page

264.

1)"^

V(5
(^

- x)(x - 1)
"*"

a^)'^'

- = -

V(5

- a;)(a; - 1)

di
I

-'s

^,

V(5-a;)(a;-l)
the

first

- a;)

^r^

ir.

the second integral converges.

integral converges.

ir/2

integral converges.

'^

2'~'^
(1

264.

Thus the given

lim

x^

^1

(d)

g^

Hence the integral diverges by Theorem

1.

dx

,.

Hence the integral diverges.

integral

con-

CHAP.

IMPROPER INTEGRALS

12]

273

Another method:
2'
2""

'

2'""

r'

1-x ~ 1-x'
(e)

16. If

lim
(r/2
-%7T-

simultaneously and

mgl

For

Se a; Ss 1.

(6)

<m< 1

using Theorem

and

<

Page

3(i),

Page

4(i),

x-^-^{l-xY-^ dx

If

m g 0, hm

<

1,

x'-^

l-m

p =

-w

^(1 - x)- =

00.

and

Hence the

Letting X

= 1/^,

"

n>0

and a

'^

of the

f ^dy

a;"-' (1

{1- xy-" x''-'{l-

- sc)"-" =

x)'-'

using

simultaneously.

regardless of the

= 0.

^^^

""''''''^'^ ^' *

*^*" ^'^^^^-^

^* -'

-< ^^^

^-*^--'

^^''^

and the required result follows from Prob.

12.

THIRD KIND

prove that J["x"-'e-Mx

^^'

converges conditionally.

the integral becomes

IMPROPER INTEGRALS

lim -

first integral in (1) diverges,

=1

continuous in

""*'

1.

m>0

''^'^""^ '^^^^^"

lim

is

(i)

'""''

coSLSTctXlf^^ "' ^'^ "^^^" ^"*^^'^'' '^^"^ *^


-sin-dx

m>0

if

- xr-^ dx

anda = 0.

and 6

integral converges if

requirfrS'fS.Lr'''

{1

the first integral converges since

264, with

value of n, by Theorem 3(n), Page 264, with p

are

converges

(a)

diverges otherwise.

x-^a-xr-'dx +

264, with

Thus the given


(b)

'

Similarly, the second integral converges


since

Theorem

^
*i,
Hence
the integral converges.

and n^l simultaneously, the integral converges


since the integrand
Write the integral as

j
If

numbers, prove that

real

Hence the given integral diverges.

diverges.

lim ( '^dlzL^Y" _
~
-i-\ cosx J

dx

T^^

(cos^)""

'

and n are

and n >
(a)

- xy"

^^^

(a)

converges

{J ^i^)^^^^^""'^'^^'

if

n>

and

(6)

diverges

Write the integral as

(a)

If

wai,

the first integral in

and a

(J)

converges since the integrand

P"r'
converges

^'\" '"

^a=-V*'-'^f^''^
e
_ 1, the integral
a;

x^'-^e'^dx

(ij

is

continuous in

'"*'^"^' *

by Theorem

3(i),

*^

S x ^ 1.

^^'^"'^'^ ''i"'^

Page

264, with

^* -

= o-

1-rj

= 0.
Thus the

fii-st

jim
lim a;"."-"-^
a;
a;
e

Page

a;"

"-''.

Since "lir
smce
^nm x

x^-^e'^dx
'e-'^dx

262, with
**"'

for. Jo"'

integral converges for

?i

> 0.

1'^
^ ^^^P^"^
-%Tvt'H''Tr'"
(by L Hospital's rule or"otherwise),
this

'"^^^''^l

''^

t^

^^^t kind.

integral converges

Since

by Theorem

l(i).

= 2.

'""''

'"*''''^'

^'' '""^""^'^ *""

>0, and

so the given integral converges

[CHAP. 12

IMPROPER INTEGRALS

274

(6)

lin^O,
If

Page

the first integral of

w^O,

the second integral of

[Theorem

Page

264].

[Theorem

l(i),

sum

also

3(n),

Jim x-x^-'e''

converges since

(1)

Jjm x-x-'^e-'

diverges since

(1)

262].

Since the first integral in (1) diverges while the second integral converges, their
diverges, i.e. the given integral diverges if n = 0.
interesting properties of the given integral, called the

Some
in

Chapter

UNIFORM CONVERGENCE
19. (a)

Evaluate

= f

<^(a)

(c)

Explain

(o)

<f>(a)

a^a^>0.

converges uniformly to 1 for

in (a)

the integral does not converge uniformly to 1 for a

why
lim

ae-"''dx

e -ax

lim

Thus the integral converges

Method

a>0.

for

Jl

Prove that the integral

(6)

function, are considered

IMPROPER INTEGRALS

of

ae-'"'dx

(&)

gamma

13.

>

to 1 for all a

lim 1

e"""

>

0.

a>0.

if

0.

using definition.

1,

a^a^>0

integral converges uniformly to 1 in

The

pending on

but not on

such that

a,

ae-<"dx\

r'

(1

<

ae "'dx

- e-"")

for each

if

for

e-

we can

find

N, de-

u> N.

all

<

e-"!"

>

u>-\n-1<