Calculo Avanzado Murray

^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 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
^
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=î,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 anÔ. 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 âiUi
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,
lovsêr
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 + Îvl +
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
Îvl
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 + î,
(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.
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, XyÔ.
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.

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

Ans. (6) Ko
all
rational
numbers
countable.
is
(6)
What
is
the cardinal
number
of the set
in (a)?
GO.
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
^.^^ô^^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 = picisî and Z2
Interpret geometrically.
prove
PjCis^j,
(a) Z122
pjp^ cis (0i
+ ^j).
^1/^2
(&)
(pi/pa) cis (0i
sâ)-
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
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.
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
aovÔ.
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.
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
avÔ,l.
Page
3.
Trigonometric functions:
sin X
smx,
cosa;, tana;
= ^^^,
^^^
cos
_ 1 _
" teE " 'ÎH:^
radians = 180). For
Îir^'
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^î^^y) = 1^ tan a; tan
2/
4.
(b)
(c)
5.
=
=
=
sin-îc,
^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-â;
sec"!
a;
cofâ;
= 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-â;,
tt)
Hyperbolic functions are defined in terms of exponential functions as follows.

(a)
sinha;
-^
e^
(d)
cscha;
(6)
coshx
,, , ,
tanha;
(c)
^^^
1
= ^^î^
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,

cothâj-l =
l-tanhâ; = sechâ;
coshâi-sinhâ; = 1
cschâ;
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-â;
in
sech-^
In
(d)
(e)
a;
Â + V^^ll\
^
+ VI
a^M
q
tanh-â;
(c)
LIMITS
of
= în(i|j,
|a;|
<
în^^Y
coth-â;
(/)
^j
|a;l
>
ô
<
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
^-*ô
/(x)
if
and only
if
lim /(x)
x-tx+
lim fix)
x-*x-
I.
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
IfBÔ
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
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^în
an
/(.)
= x, IPlim
terms of
CONTINUITY
restriction
---o-
fix)
/(.),
'
i.e.'îf
/(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"')- sifânTcr.!
\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.^cîo.1 or comvosite
This is sometimes briefly stated
u 'isas: ^ conimuous funccontinuous.
and
andift'-Tf
If i/o--/(aô), 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^
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.
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
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.
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
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.
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'
*â" ^^
"^
a;-
-^ <
16.
<
^ in(2/e-l) assuming
for
1)4
>
when
< e < 2.
>
=
^^^^
1)*
^^'^ ^"'^
2.
If
2,
0.
xî{x~
^^
<e<
if
co
^ ^^^^^^
and prove
^""
P^^^ê 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# < ê < 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
^^' ^^
ilô
We
By
on LIMITS
^^^^
^'^*^'
P'^"'^^
must show that
if
*ât
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|
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

|/(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/iî
+ i)
for
*^
83.
we have
in (i),
- xol <
- AB\
where
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
\[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.
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.
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)Ô.
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.
|^î"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.
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
<
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
<
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.
36
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]
37
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)
hÔ;
f(
5X
^^^
a;
Determine whether
^^^
^^^_ ^^^
(.)^.,ô,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
Vî^,
(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:
î^-: ^
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.
38
tan-
Prove that
(a)
46.
tan-â;
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.
^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.
Ûm
(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
Î^.-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]
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;
ôsâ^-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
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.
40
are uniformly continuous in the interval.
f(x)g(x)
m/g(x).
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-ôo
==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-ôo
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
mÛn = 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
it is
Similarly if Un+iÛn 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
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"
'
proving the existence of
6.
2
3
3. iQn
3-10'
a number
is
W>^-f
n>N,
for all
+
+
there
i.e.
e,
XT
^^
3
,.
lm
that
<
-^
<
[c/Wl
>
are constants (independent of n).

Vp
1^ <e when
Now
p>0
where -0 and
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
-ê
[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)|
Ê
|(a.
|a-A|
6)
(A
+ B)\ <
for
all
16-B|
(i)
ê
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

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-ôo
n-ôo
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
î
find
find
and Ni, we can write
iB
^^ Tiî 77^
\B\-i\B
and Problem
lim
nôo 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%î21 =
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

^^'
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 Aî
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
^
+ 'î+
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,
28. If
lim^^n
Let Un
prove that
I,
Vn
We
I.
Vl
iim
î
+ ^^+-
îî
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
<
(^^'Ûn.
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
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.
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^î^''^:'!.
<"{<?!;.}.
(*)
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-ôo
44-
If
\r\
>
prove that lim r^ does not
1,
exist.
ni->-eo
45.
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 (VîFT^
(/)
(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
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--
(^)"
iâ^--- -
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 = '-û^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
'î
^ ^ ^ '
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/(ô + ^)-/(ô)
nxo)
if this limit exists.
The derivative can

f'ixo)
also be defined in various other equivalent
Urn
f('^)-f('ô)
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
/(ô
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)-/(ô)
h->oif this limit exists.
has a derivative at this point.
if it
+ fe)-/(ô)
h (= ax)
in this case
flixo)
y-^'
it
approaches zero.
Similarly, the left
Ax
defined as
is
/(ô
fîxo)
+ 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)
/'(aô)
{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 /+(ô)
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-
îf(^)-9{x)}
^f{x)~^g{x)
3.
d^{Cm) =
4-
aîmai^)}
^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)
<â;[^(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^ÔgeU
11.
d
= aMna^
â"
12.
^
d
4^
dx
du
loga e dw
^ n =i 1
= -ITd^ a>0,a^l
d
^mU
17
du
1
^
d
J sec û
^/^^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
cschû^^
'^
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^'î
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;-aô)"
w!
This
-a)
(i
^^^^
(Û)
^^
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), Î'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)vÔ 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
/'^-''(aô)
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 /""'(^ô) >
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
Aôie;
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.
/(ô
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)
xÔ
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.
hÔ
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' + SxÂx + 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'^'Âx
(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/(^:^Î:îMl
^/(.)
,(.)
Another method:
= /(x), = g(x). Then Au = /(x + Ax) - /(x) and A,; = g(x + Ax) - g{x),
Thus
gix + Ax) = V + Av.
MAt> + ÂM + 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
AxÔ. 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-
ÂU
du
as
Am-
and
+ eAM
()
dy
0,
AmÔ
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\
,:^/'dÂM
lô\duAx ^ 'axJ
dx
dy
du
du
dx
dy
du
Am
j.
A.ôAx
Am
_j_
^^^
^ô
jj^^
aô Ax
I
(3)
CHAP.
DERIVATIVES
4]
^*- ^^''^^
d^(^"^)
(a)
^(tan:.)
secâ;,
(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^îîve, 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,
VTîlH^
a continuous function, this can be written
is
3,
^
sinî/
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
\.
--î^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-â <
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)
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
== ^î^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.
||
(êe 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),
^ ^
ô<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;
ô+ 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
â;
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-ô
'''"^^
- 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-foVsmâ;
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!^sinâ:
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*
\^
(iôiiKi;)
"
^-
/Vsinâ:/
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
xô
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
,.
xô
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.
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-ô
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-ÔO
/;(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
(^-ô)'
+ 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
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^
/â;;
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^) + fxîx),
(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
dâ;
(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: Â^*
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,
^^^^Î^^^^
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^-îa+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 xê-^'
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--"
==-ô
(/)
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
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^ - sinâ;)
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-û)(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''-û)(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.
^Â_
/'(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
xâ
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 xoâ, 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^-Ô.
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 csctiûdu
,.
-cothM
7r/4)|
22.
-'
cot^l
sechutanh udu
-sechu
f cschucoth udu
-csch^t
23.
^4.
Jsectant.d
11.
ln|tanM/2|
jsecûdu
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
_û
a ... \1 ..-i
\
1
_,,0,
77^08-'- or -sec â
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.
û
,
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)
fcschîd^*
+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,)
û)
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
limî/(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
^Âx,
/(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
â
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^tanîtf"
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)
Useînteâtion 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;
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;
cosâ;/2
sin
a;
a;
M^
fTjT^-
_ 2dM
- i+^-
2cosâ;/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
Vîni
Î^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
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'êldx
"^'^
\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' ê^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.
and
30. If fix)
are continuous in
^(a;)
for
have
A^
we
31
find C"
(fif(a;)}^(Za;
J^
real values of
all
fîmYdx
(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^îj'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
,î^^^"^'
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.
râ^^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
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
^ eîn ^= cos
|,
in
sin TTX
OSxSir
m S f(x) S M,
Q^xSl
a;
dx,
(6)
(a)êsina: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-
^Û-.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'^îM^^ 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,
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 inteâtion 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^ +,^
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.]
xê-'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.
"ô
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
= î^_>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).
û
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-
^""'^^^ ^^^*
^"
'^ *^ î^^^^"* 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
^^^ ^^^^^ ^' '^^"' ^'^^ ^PPôpriate modifications, to functions of two
ITmlre varifbles
I'?
ITERATED LIMITS
The
iterated limits
^"d
iiîS/(^'^)
they must be equal
| lim
ii/(^'^)
lim fix,y)
if
oi this last limit.
lim
is
/(x,
and
respectively]
limUim
aê
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 iÔ'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^câlled
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
fiô,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^ +
'âi^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^,
(&)
fcÔ.
{-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^^+^^-î <
'''''
- 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
- mâ;'
+ m'x'
a;"
,.
= 'mx
l^ôx'
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^XÎ =
exist.
Hence
f(x, y)
lim
and
limjlimjî =
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(îM
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
gî^^'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/=(eî'.j/)
6xy
yê'^
= a^ = 5^ (a;' + 2x3/e^) =
+ 2xy '-(e-y") + e^y'-f(2xy)
- 2xye^y^-2xy + eTO*'2a; = 4a; Ve' + 2a; e^
_
(f\
^^^
^^,
*''" '" "^"^^ '^"'""
^''
iî'^'y
/.
^
^"'^
ay
^ ^f\ -
^V
a^
3a;*
_
=
a^Wy
2xyê''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'
dÛ
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 + eÂy)
{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.
= /Âa; + /jAj/ + e,Ax + ^Ây 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-^-tÎ^'
"'
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,.(
\^
xê"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
^^
Auô^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^^dâ^ =
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)
~ î
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^^Ô.
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':î .tf

tî:^-. 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
ân^+ 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)
û
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
yû-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.
intheû^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)ê
= 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) + 3AÂ/(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
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^MOî^LÂM =
(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'
cosî
Stf,
2sin0cos0 d'V
^0
dx'
45. (a) If
<b
dp
sin0cos0 ay
^p^
Similarly,
Adding
ay
+
op
""
ginÂ 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
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) = xî^ + y^'\ ie) fix, y, z) = x^ + 4y' + 16^^
fix,
CHAP.
PARTIAL DERIVATIVES
6]
129

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.
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.,
3û
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
Gxûv^ + 2y
^'
Ans
(a)
32/
'
g + gg
differentiable. prove that
2a;M''
^'
'
+ Sy^
+y
Zxûv^
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 . =
^^'
î'^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^Mcoshô'
(b)
cos^Msinh^f,
(smûcoshî>
(c)
m sinh
cos
^)
û
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.
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.
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
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
^êetors 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 betvsêen 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 dÂ/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+â^ + '^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
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^
aîaC7^^_^
'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.
2.
Spherical Coordinates
See Fig. 7-11.
(r, d, <f).
Transformation equations:
= rsind cos<j,, y = r sin sin

z = r cos 6
where rÔ,
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^sinêd^^
r^înS
Pig. 7-11
dV =
^Û =
Laplacian:
r^de^
dr dd
r^ sin 6
r^dr\
r^
dr J
d
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.
=
=
=
(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
îzxi^
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-Bâ)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 + îk,

= 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
*ê 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 =
yzî
dM
+ yz^ -
Kzk)
^ +
du J
(Bi
\ dM
'
dM
^2^
dM
Bs"^
2,x^yi
xVz'j
3a;V^zi
x'yz^'k)
xzî,
find ^r^(</'^)
dydz
x'y'^zî
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

=
(Â)
^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-,î +
d/dx
2a;î
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 (Â).
(e)
(d)
d(a:2/^)k
dy)i
2/
Mi
dx
(">
2k,
j/ j
sin
z''
(2z cos
a;^

34. If ^ = x^yz^ and A = xzi- y^ + 2x%k,
(6)
2x'zk
2:
dA
(Â),
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'zî
^(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)
(Â)
'
(V^)
{</>Aii
A +
[CHAP.
^(V A)
+ Âsh)
'
0A2J
(Â,) + ^(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,
Â3^.
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
dÂt
dz dy
has continuous second partial derivatives so that the order of differentiation
is
CHAP.
VECTORS
7]
JACOBIANS
40.
Find ds^
(a)
Method
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
d
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
\îp.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 ^^ê.
fe
= P,
^^
Aiei
IK
<'-.'
=1
[see
(b)
div A,
Problem
19*
e.
Cs
in the results 1, 2
40(a)]
9*
-ê,
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)
iÂl)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)î
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,
rârV
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^
^îT.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
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

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
Sxzî
(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* =
= - VÂ + 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)
lîrÂi)
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'*
|^(VÎ^TÂ.)|
îF^{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]
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.
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
^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-*-^' ê^-te
integrals
INTEGRAL SIGN
..iitvi;i"rtht'?/ir:;;dttrczsr'
'-
^ "^
'-
-'- '-'"-
The
[CHAP.
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
Kiû
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.
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
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^.ô). (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
,îaF
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.
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
= îr 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

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.
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,
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)
(^
~'
+ î + '^ \
_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
'^
î
^^
= _ 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|
170
15.
sign.
Then
f(x, a) dx.
f(x, a
+ â)dx -
f(x, a)
+ Aa) da; +
+ â)dx +
f(x, a
/(a, a
X2<a) /(,
*^2(a)
dx
a)
î< + 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
+ â)dx
/(^2,
+ Aa)[M2(a + Aa) -
W2(a)]
()
ta + Aa)
X"2
-2 (a)
where
(,
is
+ Aa,
between a and a
î 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'ÎT^:^
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,
_^^
-/âlTi'
Leibnitz's rule, if
aa; da;
fit
sinaa; ["
Jo
obtain
â^
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.
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
sîj
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,
Jâ ++ '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
rlrî'^^
riy &-eosa, V
In (a -COS..) 'do;
Integrating the right side with respect to a from a to 6 yields
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.

/(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)
^{hÛ + 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)
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
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.
^^^^-
^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
^^^Âccording 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),
ôÛich
+
+
2a;
+7
:o' Jiriu!,:êits^
'^-
"'
= "''
^"'
^'
+ 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^^/.ê
.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
_^_ + î__
+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
Cose
(2),
^X^,
yields X|
1,
^X^' ^
Substituting in the
fâ-
\=
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> ^ ^î- 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.
8]
175
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-â
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.
176
[CHAP. 8
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'^.lpSinô,
(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.
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
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
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.
178
'
result
-^
sec
In (1
^^.
.>1
.. pr.v. that
- -^.
i COS x) da
and MINIMA.
MAXIMA
a;
58.
^ xyH^ ûbjec^ 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%
70.
constants and n is a Poî";^ 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.
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
î>
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 =
gîy)
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
^^
''^
^^^ ^ (^)' ^^^^^-^ - appropâte 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^""" î^'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^'LnslÎn'f"
dimension of each
"^"'k""""'
subregion
^j F{x,y,z)dV
r^rtc^lfSlfi^lif^! ^- ^-
'''
'^-^*
(7)
^-
-^-
^^
^(--)
continu-
P"'""^' ^ *^^ ^^' ^^ ^"^ ^^ Pl-"e. the

^ÎTare 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
'
^ *^^ ""^^^^^ ^^^'^^^ ô 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.

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 -
xâ -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 xâ~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
ô 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)
(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,
(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
â
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
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
-.'"û
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
vaâ
[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- "
â'
'
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 Oâ^^.
below by the cone z^ sinâ = {x^ + y') cosâ, 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
â
-,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.
(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^'
'=^)-
= â.
...
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/râ; + 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
"^^ tât 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^âxdy,
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
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

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 = -â'
a'a = ^Ma\ where
Ans. (a)
^/26.
By using
the transformation
+ y = u,y=uv, show
that
dy dx
x=
*^y~0
e'"'-^'^
-^
Ans. %â^ 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,zÔ,
^'
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. ^â'{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 = yÔ,
|(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 ^'iJê'^'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
(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
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 = yÔ, 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. îTâbc
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{{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.
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

%.
(^^)
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
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.
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)
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)
10]
Along the straight
line
from
f=
^!(
from
line
Since
at
f^_ît'-(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
(1, 2),
+ iy+t}2tdt =
points
to
(0, 0, 0)
(3a;^
(0, 0, 0)
{{3x'
J^
x^t,y = t\z = t\
JÂ-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'
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
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
):
'LiÊ
(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
+ 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.
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
^~&Â^^
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/^

{{2x)(x')
- x'} dx +
{x
Fig. 10-5
{xY}
(2x^
+ x^ + 2x') dx =
7/6
{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 =
-ô
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
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
î
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
*ô
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.
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
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
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)
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^
Â
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)
Â
P{x
P dx + Q dy = -^dx + ~dy =
follows that
{xi,yi),
= J_ f
.f>{x,y)
^ ia.hl
<p(xi,yi)
(B)
^(A)
CHAP.
10]
207
/-C3,4)
14. (a)
Prove that
{Qxy^
dx
y^)
{Qx^'y
Zxy^)
dy
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
yê-)dx
{x^sinx
2ye^)dy
around the
a^'^.
2xy sin x
Q ~
y'ê",
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.|
În 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
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.
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
*â: =
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^|fltanêC'*
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,
^ î2ik_^ 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
210
jj
^xy -2x^-^x-2y + 6
(2x2/
I
x=o *^y =
+ 6y)|^
(a;2/''-2xî/-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.
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
jî^
JJ
Aîx.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 xÔ, 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
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 =
xzî
(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
yâ
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
.^â"
- 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
^^
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
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

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
_^Â.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
Â-dr = JJ(VXA).ndS = afjn-ndS

which contradicts the hypothesis that
It follows
that
VXA=
is
Â
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
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
[CHAP. 10
A'dr

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 =

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 =
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^î,
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;
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,
Q dz
dz~a- U,
- 0;
by placing
1) to
(2, 1,
-1)
joining these
^^'
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
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.
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
AFÔ
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-ô
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.
220
[CHAP. 10
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
+ a^y + da) dx +
will the line integral
p cos 0,
P sin ^,
prove that
46;.
If
47 .
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^
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. Zâ /8
and interpret.
where C
is
the boundary of the
Ans. common value
120.
independent of the path joining
(1, 0)
and
CHAP.
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
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 theûpper 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.
= xê'' + 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.
77.

closed path
two, at least)
78.
in a region
is
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
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
î^^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
OÛn^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, Ûn also
diverges.
Example:
Since
>
2^
and
diverges,
Then
also diverges.
Quotient test for series of non-negative terms.
2.
(a)
If
ô and
i;gO and
lim
if
J^
AÔ
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)ûn, 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
(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)
(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\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
Û{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 û{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
Hoviêver, 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)\
N.
series
CO
aiUi{x)
is
THEOREMS
If
an
If
to the
a2U2{x)
uniformly convergent in
on
]â'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
Ûn{x) converges uniformly
then
S{x)dx
21
<'Ûnix)\-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 Ûn{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
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
'ÛnX"
= Xo which may
be an interior point or an endpoint of
n=
the interval of convergence, then
lim\ânX-\ =
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"
^"'^"'
ân 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
ânX"
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

^ 1.
a;
Ka;^!.
l<x<l.
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^)
Ûnix,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 PÔ
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)
^ô^^^rges
l)
Ml
The
Â2n-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
OÛn^Vn, n =
If
0.
""*"
+ {^/^^+l-^^n) = aATTI - Vl.
a necessary but not
is
diverges although Urn Un
Chapter 3
'" ^
"î2n3-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\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"
^^î 2^3-1'
'
^=
approximately
i.
'
*>
Taking
^-în^
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
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 wê-"'
1,
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.
f^.p = constant;
(a)
~=
x~''dx
j
î
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,
îf^ = Jl(''^*)ir = Jim

"'""''
'^'^
J!]|?L
-i^'îr
{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
în^ + 1
.,
11
W2
-,
_i_
required.
to each side,
<
+T--
tt
therefore proved.
ALTERNATING SERIES
15.
Given the alternating series ai a2 + a3ai+--- where 0ân+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
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,
aû)
(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,
'âM
(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
Môo
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.
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
MôoJf
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
\î:-tl\
'^n
we can prove
where
z=
L =
that
if
lim
|mi|
L<
1,
1.
lâl
\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 nê-"',
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
_-(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^
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.
^-Y
2)
The
'
odd or
1).
the series converges, and

The
lim 2"
(since
is
we have
I V\r'\
23.
depending on whether n
even.
Then
(a)
4/3.
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
nôo
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-ôo
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;
lîl >
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
|||| <
lim p^L\
M
-.
that
= -^
1,
where
^''^'^^
^'
and
a;
- -2,
i.e.
for
(a;
+ n)(x + n-ir
a;
g -i.
w
^^^^ê^'
"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
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,
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
(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
-î
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 =

ô, prove that
in the interval \x\ < |xo|,
(6) uniformly in the
interval
Since â.x"o converges,
enough,
=4>
choosing
33. If
(a)
îri'
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"
.^ ^"
nîM-"^
(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"
Î 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;,
ô ""*"
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^êlKlWs/"'^
^"
'^
(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
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
Sâ'+^-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
zîx
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 fiôf
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
în
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
îm
= 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
'û^ 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

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
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)
î^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
CONVERGENCE
53.
and
DIVERGENCE
(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

Find the
i - (1 -
fallacy:
1)
(1
^êt
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.
p>
and diverges
(a) conv.,
(6) div.,
1,
(6)
Page
^ÎlpL
diverges,
225.
'
Ans.
if
of the quotient test.
(c)
+QT1-1
('^Ît' 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.
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:îi(hiW
conv., (d) conv.,
(c)
a constant,
(a)
converges
() div.,
if
_
'
-iS^'
^^'
='
(/)
> 1 and
'=^
V^
'
'^
div.
(ft)
diverges
if
i.
2 :;^ < J-
<
^â^^TTTi'
(&)
(a) div.
^^^
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
2
=2
'"
I-
VTOO, giving the
maximum
error.
+ a/IOO
can be improved by estimating, for example, VÔ + -/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
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
(|)=

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^î,^..
~l<xS3,
^'îf-^^Ql^ ^^
(e)
all
xÔ,
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
xÔ;
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)
xÔ,
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'ê-^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'
pôve that the formal Taylor series about x
4.]
-1
2^ J" 2^T^T
xtÔ.
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
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
Use
(c)
How
[Hint:
prove that the series
<?iir>i
thot
such
that
2 TTT^
criterion,
"' '"''"f~W ""<
->'"
p/^'î'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.
Jîx),
^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"-ê-
+^^
+ ^^+ .
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;)''"
+-(2Î)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.
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]
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'îHiidw
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^Ô
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
xâ, and suppose

all
f e-dx
a;
S a,
C f{x)dx
converges,
xâ, 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
AÔ
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,
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
î?^ dx
I
does not converge (see Prob. 12).
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
pÔ
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,
"ô
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^ââ^, 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^ââ^,
[CHAP. 12
such that
X>a
M{x) dx converges,
a
uniformly and absolutely convergent in
f{x, a)
dx
Since
P^^ =
is
a^â^ 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 a
all
and
a^
f{x, a) ip{x)
^ a S a^.
dx
is
THEOREMS
Theorem
on
a^â^ a^.
UNIFORMLY CONVERGENT INTEGRALS
6.
S a and a^â^ 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^'â'â^.
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 vsîth
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 povsêr 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)

Consider
r
I
dx
-'i
yV22/ -
0<e<l,
where
- a;)
V(2
kind,
first
(b)
a proper integral.
Let
say.
x)
=1
As e->0+, we
see that consideration of the given integral
is
equivalent to con-
X/7
-
which
y^/2y
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 
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]

(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
^-|=î^^ =
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.
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^âx
The
first
Smce
Page
,
-dx
integral on the right converges [see Problem 1(e)].
hm
262, with
cos x
x"
x^'î^
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
(&)
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 yî
Since
2/
Since
integral converges by
Then the given
0.
Theorem
1,
= 2.
and p
^+T
it
\im ye~'
îlK
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
â
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
fîilL^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
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.
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-â-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
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 î^)^^^^^""'^'^^'
if
n>
and
(6)
diverges
(a)
If
wai,
the first integral in
and a
(J)
converges since the integrand
P"r'
converges
^'\" '"
â=-V*'-'^f^''^
e
_ 1, the integral
a;
x^'-ê'^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^-ê'^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)
linÔ,
If
Page
the first integral of
wÔ,
the second integral of
[Theorem
Page
264].
[Theorem
l(i),
sum
also
3(n),
Jim x-x^-'e''
converges since
(1)
Jjm x-x-'ê-'
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)
aâ^>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,
aâ^>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
for
N,
the result follows.
Method
say X
(c)
using the Weierstrass
2,
Since
lim x^
Taking
Ko.
ae'""
=0
M{x)
is
uniformly convergent to
0,
the
ai
number
&
for a
Af in the first
cannot be uniformly convergent for a
20. If
^Wj =
I
'-'a
dx
f{x, a)
>
aj
we can choose
0,
and noting that
integral
As
test.
for a
>
aj
method of
>
-^
converges,
it
for sufficiently large x,

follows that the given
0.
increases without limit, so that the integral
(6)
0.
a^â^ a^,
is
<
|ae-"l
prove that
<^(a)
is
continuous in this interval.

Let
Then
= C
<f,(a)
h)
0(a
0(a
<f>(a
f{x,a)dx
h)
f{x, a
0(a)
where R(u,a)
R(u,a),
h)
Thus
dx
R(u, a
+ h)
and so
^)
/(ic,
/(X, a
/l)
/(, a)
{/(a;,
f{x,a)dx.
a)} da;
i?(M,
dx
|iS(M, a
^)
R{u,
a)
"u
I
/l)
Since the integral

ent of a such that for
0(a)
is
a^â^ a^, we
u> N,
R{u, a
+
,
, ^
^
<
h)\
/o
e/3,
d/ d)\
R{u,
i
I
/l)
I
can, for each
^
<
+
e
>
(i)
fi(M, a)
I
0, find
/o
independ^>^
e/3
(^)
Since f{x, a)
is
continuous,
we can
\f{x,a
find 8
h)
>
fix, a)
corresponding to each
I
dx
<
e/3
for
|;i|
<
>
such that
()
CHAP.
IMPROPER INTEGRALS
12]
Using
and
(2)
Note that
a
= !,
we
(3) in {1),
0(a
+ h) -
0(a)
<
for
<
|fe|
so that ^(a) is continuous.
S,
we assume
that a and a + h are both in the interval

and right hand continuity is assumed.
in this proof
>
for example, A
see that
275
Also note the analogy of this proof with that for infinite
sg
a,
Thus
q,^.
if
series.
Other properties of uniformly convergent integrals can be proved similarly.
21. (a)
Show
that
lim
(ft)
Since
^(a)
guarantee that
22. (a)
(b)
Prove that
ae'"' dx
by Problem
Odx
lim ae~'^^\dx.
e-"""
Thus
0.
rxdx = ^^"^2
lim
a>0
for
result follows.
(see
may
96(a)
and any
Problem
19),
Prom
integration formula 34, Page 84,
M^
This follows at once from the Weierstrass
(6)
and
e'^^dx
EVALUATION
of
no
real value of
r.
aâ^b,
a^
+ r'
a cos rx) I"
^
a^
M test for integrals, by noting that
+ r^
e'"" cos rx
e""''
converges.
DEFINITE INTEGRALS
In sin
The given
is
we have
"' ^^ ^'" rx
lim ^
e-^'cosrxdx
there
not be equal to 0(0).
Prove that the integral in (a) converges uniformly and absolutely for
where
< a < & and any r.
(a)
(a).
19(ci).
not uniformly convergent for a
is
cos
(b)
Thus the required
0.
will be continuous for a
4>{a)
lim 1
ae""') dx
( '''?.
ae-'^dx
ae-^^dx
(6)
lim
ic
~ In 2.
da;
integral converges [Problem 42(/)].
In
^0
sm a;
da;
Letting
In cos
Jo
y dy
Tr/2-y,
In cos x
dx
Jd
Then
2/
Letting
2a;
J"
a;
In
sm 2a;
^{I
In cos
da;
dx
a;)
C'"\nf ^^^^^\dx
In 2 da;
-J
In sin
2a; da;
-| In
(1)
= v,
In sin 2a; da;
J^
Hence
(In sin
(1)
becomes
2/
In sin v
I)
- 1^
In 2
^
=
or
dv
/
7
i.J
(letting
- In
2
In sin v
2.
= TT-u
dv
In sin v
dv
in the last integral)
>
24.
[CHAP. 12
IMPROPER INTEGRALS
276
Prove that
Let X
g- In 2.
Then, using the results in the preceding problem,
or J
y.
= -
dx
In sin x
a;
a;
In sin X da;
~ **)
TT
TrMn 2
('^
^" s""
In sin x da;
"
a;
('^
In sin x
1" sin x
')
dx
dx
= - y In 2.
-2
+a
(b)
Show that
(d)
Prove that
^{a)
-^
r"
J^
dx
(^2
Evaluate
(c)
i)n+i
is
r'"'^
7^2^11)2
J
o
The
result follows
J^
test, since
from the Weierstrass
g i.
1-3-5..-(2m-1)
cos^^^d^
^5-47^
^^T^Tf
^^ "
TT
2"
2-4-6...(2)
1
(a)
''^
>
=
/I
*'^**
r
J^
^"^
x'
+l
converges.
(6)
0(a)
(c)
From
(6),
lim
-tan
lim
= -^^
f -^^
X +a
2va
'
tan
lim
-71
'
Differentiating both sides with respect to a,
we have
Jo
"
a^V^?+^y
Jo
the result being justified by Theorem

"
^-^ g ^^
(because
8,
Page
and
j^
(a;^
+ ar
266, since
^^
j^
^tf^
is
converges).
^'
(?ny ~
,
-3-^ = "'"
n times, we
find
J."
(-l)(-2)...(-n)J^
where
justification proceeds as in part
r'^
Jo
Substituting
26.
Prove that
J^
a;
(x^
^^r^^.
(c).
\r2A""2A"2/-\
Letting
a^l+, we
dx
l-3-5--(2w-l)
TT
2m!
tan
l)"+'
9,
f^cos'-ff d9
22 and Theorem
7,
Page
266,
we have
;2"
find
l'3-5---(2n-l)
;r
2-4-6---(2)
-^^^^^dx = 2!"^^+^ wherea,b>0.
From Problem
is
obtained.
CHAP.
IMPROPER INTEGRALS
12]
/"{/
e ""
cosrxdaldx
Ct
277
}
e-"' cos
= a l^*^x=0
''"-'""
27.
Provethat
J^^
By Problem
j;^"^
22 and Theorem
Jo
sec
a;
Page
7,
266,
Integrating again with respect to r from

1
J-ax
cosrz
;3
X
using integration by parts.
28.
Since
7(6)],
it
e'"'
cos X
dx
f
a^
f^
J^
'-'
+ r^
-dr
tan-'-
5-fln( + r)
r tan
result follows on letting r
=1
^.
cos X
for a
function of a for a
fe 0,
a;
and
f\-^^l^^dx
(Theorem
6,
Page
266).
^1? dx
is
a, 1
29.
cos z
;r-dx
fl-cosa;
J^
-î
r?ill^= T^sin^ do;

Jo
a;2
Jo
^
Provethat
Integrating by parts,
'^-
Then
limit as
e -^
0+ and
-^
>
f'l-cos^j^
ax
Jo"
sm' X
da; = 2
J^
a;"
x^
IT
f
in.
lim|tan-.i-|l(^
+
,
using Prob. 27,

1
l)|
,,
-
a;
f"
I
sin' (a:/2)
da;
-J
*'
cos X
Jo
c/O
1_Z^ _ l^j^ ^ j" sma; ax
si
shows that
a^0+,
we have
Jo
Since
letting
J.
S^'-^^^dx = (-l)(l-eos.)[ + J^"^da:

Taking the
converges [see Problem
uniformly convergent and repre-
we"have''''"*''"'"^
drXdx
to r yields
I
J^
follows by the Weirstrass test that
da
e-" cos rx
+ r^
C..r
tan-'^dr
The required
C^S^
Provethat
'^^
da
- |ln(.^ + 1), a>0.
a;
-'o
a^
we have
e-^'cosrajdx Idr
r"e--sinra^^
or
tan-'l
==
da;
lln*^ +
dx
ra;
rx dx
_
_
dx
s-
f'sin^tt,
-^dw
/o
**
on letting m
= a;/2, we
also
have
30.
[CHAP. 12
IMPROPER INTEGRALS
278
Prove that
ax
-r-
2i
'^
(2i?
1 /e3tx_g-3i;. \
e'-e-
1
-7
4V
2i
4^
"^
sin
+ 3
3a;
2i
sin
a;
Then
(
snr
âx
du
5Hid^
4\2j
i J
i Jo
i Jo
4V2
31.
Prove that
e'^'dx
By Problem
e-'^
^7/2.
e-'
Let
converges.
the integral
6,
dx
dy and
let
lim 7m
7,
Then
the required value of the integral.
^M
J,M
e-u2+v2)dxdi/
I
^M
where
12-1).
Fig. 12-1
(see Fig.
the square OACiS? of side
Since the integrand is positive, we have
%u
is
rr
e-(>.^ + ^)dxd3/
where '5^, and "5^, are the regions

M\/2 respectively.
in the first
Using polar coordinates, we have from
e-o^pdpd^
I
1\
jj e-''^ + ''''dxdy
quadrant bounded by the
circles
having radii
and
(1),
II
^0 =
e' pdpd4>
(2)
^'i-'" '^P=
^p=o
or
2^(1
32.
M^
Then taking the
limit as
Evaluate
e~^^ COB
Let
in {3),
7L
we And
S ^(1-e
lim 7^
7^
(3)
^)
7r/4
and 7
vV/2.
axdx.
e'"^ cosacc da;.
7(,
-e -"^)
Then using integration by parts and appropriate
limiting pro-
cedures,
d7
e--^cosaa;da;
differentiation under the
The
C xe-^
xe'""^
sin ax
and
dx
is
integral sign
is
justified
xe''^ dx
converges).
all
by Theorem
8,
- ^l
Page 266, and the fact that
a (since by the Weierstrass test,
xe"'' sina9;|
CHAP.
IMPROPER INTEGRALS
12]
From Problem
|e-'ôsaa;|
m=
lim 1(a)
Solving
33. (a)
31 and theûniform convergence,
and
e-^'
and thus continuity, of the given integral
converges, so that the
(since
Weierstrass test applies), we have
i^/^.
= -|/
Prove that
j^
dx
g-^'
279
/()
subject to 7(0)
^,
^\-,.-.i.,^ ^x
we
find
/()
^.
ê-<^''\
Evaluate
(6)
f^-C-^^"^)
da;.
e-<--/*)2(l-/a,2)da;.
D
^""^^^
,-.nJ?L^w?"^*'
- 0+ and that for sufficiently
-'".f?"^
large
a;
so that /'() converges uniformly for a
Now
^^ observing that the integrand remains bounded as
x,
by the Weierstrass
f e-^ dx
test, since
Jo
Jo
Jo
a;
as seen by letting alx^y in the second integral.

Thus l{a) = c, a constant.
o:H.o+
the required integral and use Problem 31 to
obtain c = y/^/2.
(b)
From
{.
fe-^--%^
Then
34.
Verify the results:
(a)
e-"e"dx
1
^{cosaa.}
^^
e-cosaxdx
(b)
^{cosax}
lim
e-<^-)M
-^^
= 1. r^-(^^-^-\. = :^,-,
Putting
=^,s>a;
^{e-}
J:{e'"}
f e-
,.
(6)
To determine
let
converges
-^^, s> 0.
e-^'-'^'dx
by Problem 22 with a
= s,
a.
Another method, using complex numbers.
From
part
(a),
Jl{e-'}
= -1.
s
JL{e''"}
.^{cos ax
1
Replace a by
s-ai
+ ai
s^ + a'
s^
.^ (cos
justified
ax}
^{cos ax}
Equating real and imaginary parts:
The above formal method can be
sin aa;}
Then
ai.
+ a" +
'
-~^,
'
i^{sinaa;}
s^
a^
.^{sin ox}
using methods of Chapter
=
17.
"
^
c,
35.
Prove that (a) ^{Y'ix)} = sUl{Yix)}

under suitable conditions on Y{x).
(a)
By
definition,
assuming that
Let
I7(a;)
=
=
^{F"(a;)}
r e-" Y(x) dx
Jo
lim e""" Y(M) =
Y(0)
s"
and so
Since
e-^'Y{x)
dxV
^{Y(x)}
F(0)
^{F(a;)}
s^{f/()}
Y"{x)
Y{x)
x,
= 0,
7(0)
Thus
U{0).
Y{0)]
Y'{0)
Y{x)}
<{Y(x)}
= ^{x},
- s F(0) -
j(L{Y"{x)}
F'(0)
we
J^{Y{x)}
j:{Y{x)}.
Y'(0)
= 2.
of both sides of the given differential equation.
Y'(0)
J^
- F'(0) = 8[s jC{Y{x)} - s Y{0) - Y'(0)
Solving for Jl{Y(x)} using the given conditions,
by methods of partial
j:{U'(x))
(o),
Solve the differential equation
j:{Y"(x)
s Y(0)
M-^00
8.<:{r(!e)}
Take the Laplace transform
e-"Y'{x)dx
M-+00
Then by part
^'(a;).
lim
lim ]e-''Y(x)
such that
s is
s'Ul{Y(x))
-m
Jo
j:{Y"{x)}
(b)
e-"'Y'(x)dx
(6)
Y{0),
.=o
JiiY'ix)}
36.
[CHAP. 12
IMPROPER INTEGRALS
280
Then by Problem
35,
1/s^
1/s^
find
fractions.
-^^ =
and
J^{x}
.^{sina;},
Hence from (1), Jl{Y(x)} = ^{x +

which is, in fact, found to be a solution.
it
follows that
;?
+ ^?:pY = ^{x +
sin x},
from which we can conclude that
call f{s) the inverse
Laplace transform of F(x) and write
smx).
Y(x)
a;
sin
Another method:
If
j:{F(x)}
By Problem
f{s),
78,
we
.^-' {/()
S'W)
-C"' {/()}
Then from
^"Mfl'(s)}-
f(s)
.C"' {F{x)}.
(i),
Inverse Laplace transforms can be read from the Table on Page 267.
IMPROPER INTEGRALS
37.
of the
FIRST KIND

(a)
,,,
fr+yd*
J
f"
(d.)
xdx
,,
f"
dx
-^^
2
sin
^^dx
r"
,,
(ff)
!^
j_^(x'
,.,
^1
a;V3a;
+2
(o) conv.,
"^^
(6) div.,
(c)
conv.,
^'"
a;
ax
'^
"'^
(d) conv.,
SC
ln^
X dx
sin'
Ans.
+ x + iy'^
(e)
conv.,
(/) div.,
(g) conv.,
(h) div.,
(t)
conv.
'
CHAP.
IMPROPER INTEGRALS
12]
f -^p-
38.
Prove that
39.
Ans.
(a) conv.,
?L_
:=.
(a)
(6) conv.,
e-^ Ir,, dx,
281
ifh^\n\
(b)
e-^r.
(1
e')
dx,
(c)
f^"^ cosh x^ rf..
(c) div.
Test for convergence, indicating absolute

or conditional convergence where
possible-
40.
f î'^
+
(a)
x'
Jo
where a,b are positive constants;
J_^~"'cos6i;da;,
^*^
r_l^rfa;;
^ Vx' + l
(c)
^=0
/
cos
>
^^^
a;
^'^^^
Prove the quotient tests
41.
IMPROPER INTEGRALS
() abs. conv.,
(6)
and
..
Ans.
-^
dx
(a) conv.,
(/)
(b) div.,
TTib^Jr^
X
Ams.
a;
^^.
v^M^
(d) div.,
(e)
abs. conv.
on Page 262.
(c)
^-^Tr^'^^
i 3in
J^
. dx
(d) conv.,
ii>
(.)
conv.,
f ^Jl^
(/)
(^)X'S
Jo
Vln(l/a;)
(c) div.,
fl^
Lv.,
ax,
(^) div.,
\kl
(/.)
<
div.,
(i)
conv.,
cos(l)..,
'"""^"' "^'"^
Prove that
(6)
TYso:^
IMPROPER INTEGRALS
(a) conv.,
cond. conv.,
sin
i - x
"' *^^ ^"*^^^^1
C^)
cos
(c)
div.
V^
(a) conv.,
Provp
tVint
^rovethat
^^
conv.
(c)
-V/^M^
a > 2, div. if
(6)
conv.
if
Jo
sb (ax)
r 5-h^^dx
,
J^
converges
if
<
^''^^
(b)
As.
X'^cos(i)...
(e)
{a)C e~'lnxdx,
(6) div.,
(g)
V^M^+1)
C"
'
-^
(3
'
dx
2 sin x)
"
Vsinh
(ax)
'
2.
|a|
<
;.
and diverges
if
|a|
&
;r.

possible:
^"^
l^'^^'
(*)
conv.
-'^ ^-e a geometric interpretation.
THIRD KIND
of the
Ans.
X |cos(A)..,
(.)
O')
diverges in the usual sense but converges

in the Cauchy principal value
*'' ''"""''''
(a) abs. conv.,
49.
cond. conv.,
(c)

possible(a)
48
abs. conv.,
(6)
'"'Iw^^'^
'ô
46.
sin
SECOND KIND
of the
^"^X^^Twf^
45.
a:
42.
44.
J"
'*"'
Hîh^
f"
((i)
'
dx-
lii^^^-
^--
()
-nd.
conv.,
(6)
abs. conv.
'
UNIFORM CONVERGENCE
50.
[CHAP. 12
IMPROPER INTEGRALS
2g2
(a)
Prove that
(6)
Prove that
5L Let
ie
^(a)
4.{a)
= jy(x,a)dx,
lim
rF(x,a)dx #
52.
Work Problem
53.
If F(x) is
51
where
(c)
all a.
F{x,c.)
Find
- <
in
F(a;).
Theorem
and
(6)
54.
Prove
55.
Prove the Weierstrass
56.
Prove that
57.
Prove that
(a)
<
a;
0(a)
is
not continuous at .
(a).
and
yF{\)d\
r"
so
e-"^
'-^
converges uniformly for a
e'"' F{x) dx
converges uniformly for a
dx
g 0,
(6)
,(.)
improper integrals of the second kind.

State the definition of uniform convergence for
59.
State and prove a theorem corresponding to
of
63.
bx
dx
In (b/a),
dx
tan-'(6/r)
j^^-L^^dx
f ?""Jd
8,
266, if a is a diflferentiable function of
Page
a,b
Justify
all
steps in each case.
>
X
'
61
Theorem
DEFINITE INTEGRALS
Establish each of the following results.
=|-tan-a,
(compare Problems 27-29).
58.
60.
S 0.
EVALUATION
= 0,
uniform convergence of integrals.
converges, then
(a)
r/2.
on Page 266.
jf^(^)^^
r"sinx^^ ^|.
Jo
Theorem
test for
if
that 0(.)
(6)
lim V{x, y)
Show
(c)
a^xe'"'.
bounded and continuous for
prove that
(a)
all a.
Ans.
lim ^(a).
<^xe-'"'.
lim F{x,a)dx.
f"
F(x,a)
if
is
f^''^
j^
continuous for
is
i>{a)
IMPROPER INTEGRALS
of
e~
^
"
X cscrx
>
a,b,r
tan-Ma/*-).
|H
/cosaa;
cos ba;\
i,Y "'
^'
X"
'
(b)
Use
(a) to
prove that
[The results of
{F(0)-F(>)}
66.
Given
C
-'''
(6)
dx
-cos
6a;
^^
,^^b
integral,
and Problem 60 are special cases of Frullanis
In^-Y
e""'"
cos ax
which
is
valid under very mild restrictions on F.]
IVîT^,
/"
I
>
P^ôê that for p
,,
a;2pg-ax2rfa;
13
=:
2*2'2'"'
1, 2, 3, ...
(2p
- 1)
2
V^
"''
2 ''" +
J^
ax
a.
'
CHAP.
IMPROPER INTEGRALS
12]
tan
Prove that
69.
'
(x/a)
tan-' (x/b)
- JiL.
~ T~r'
'^^
f"
J_^(x' + x + lY
_
-
^^
tt
283
/6\
"^^^^^
^\a)
rir- *
tt
n u,
[Hmt:
Use Problem
a>0,b>0.
38.]
jlM^rrf^ eonverges.
70.
Prove that
71.
Prove that
i
J.
x'sin^^
converges.
[Hint: Consider
n U X
, n^
A
,
^7]-
j^
X
bill
and use the fact
J;
that
J^
72.
Prove that
73.
(a)
Use
sin^
cc^
J,_^
{n;ry sin^
g
f/j^f^dx = . In +
show that
C" In sin d# = -Z in
(1
(a) to
),
Prove that
75.
Evaluate
Ans.
76.
(a)
(a) If
(a)
(6)
(6)
78.
79.
^{F(a,)}
6^
/(s),
Evaluate ^{cc cos
Prove that
>
(c)
^{e'F(a;)}
prove that
Ans.
^{x' Fix)}
^'
(b)
+ ^W} = ^-. {/(,)} +
tan-
A\
- a),
/(s
>
0.
Evaluate ^{e sin 6x}.
(6)
=
s
(-!)"/'"'(),
giving suitable restrictions on F(x).
>
^-.(^(,)},
stating any restrictions.
Solve using Laplace transforms, the

following differential equations subject
to the given conditions
(a) Y"{x) + 3Y'(x) + 2Y(x) =
Y{0) =
=
3,
(6)
Y"{x)
Y'(x)
(c)
Y"{x)
2 Y'ix)
Ans.
80.
x)/x}.
.<:{(sin
a;}.
^-{/(s)
(c)
s>\a\
prove that
o
- ay +
ax},
(6) .<;{cosh
(b)
^{Fix)}^f(s).
If
(a)
^{1/v^},
V^s,s>Q
(s
77.
fîHl^rf^ ^ E.
Ams.
0.
2.
i/o
74.
a;
j;"^^^||_ diverges.
Prove that
(6)
(a)
Yix)
2 Y(x)
6e--
= 2,
= 4
Y(0)
- 3e-^
Prove lh,t.C(F(.)) exists
It
(6)'
F(,)
= -3
K(0) = 0,
F'(0)
F'(0)
Y{x)'
^
is
-=
F'(0)
A- 2e' -
Ix'
2-^
- X,
x
stion.lly conllnuoas
(e)
[C)
in
Y(r^
nx)
-x/
i -I
e '(sm
j_
ever, fli itorv,]
a;
10
cos x)
il
.h.r. H-,
81.
[CHAP. 12
IMPROPER INTEGRALS
284
and
m=Jl{F{x)}
If
g(s)
H{x)
is
called the convolution of
Write
Hint:
F(u) G(x
where
J:{H{x)}
- m) du
and G, written F*G.
F
=
/(s) ff(s)
f(s) g(s)
J:{G{x)}, prove that
"'''^^^^^^}
{/f^'^'^^^^ '^^jlX
and then
e-(+f)jr(ti)G(v)dwd'y
let
+v
==
^0
82.
Find
(c)
Solve the integral equation
Ans.
83.
=C-{(^}.
(a)
i(sinc.-a,cos:.),
(a)
Let
/(x), i,(x)
and
g'(x)
Suppose also that h{x)
Y"{x)
Solve
(6)
Y{x)
Y(x)
-+
Y(.)
{'
f[x) g{x)
dx
f{x)
dx
is
- u) du.
- m) du,
a;
a and
(b)
integral on the
Prove that the integral on the right, and hence the
g(x),
that under the given conditions on f{x) and
test.
integral
called Abel's
is
For
For
(6), first
(a),
lim
consider
prove that
if
f(x) g(x)
\h{x)\<H
81.]
+ xV6
and suppose that g'(x)SO.
0.
S'(^)h{x)dx.
Prove that
[Hint:
lim g(x)
(a)
Use Problem
[Hint:
y(x)
(c)
finite interval
all
= y'(0) = 0.
7(0)
Yiu) sin (x
bounded for
= ~j
R^x),
i?(u) sin (x
be continuous in every
g{=c)
dx
dx after replacing Kx) by
(a constant),
then
'()
left, is
convergent.
converges and
result
sometimes
and integrating by parts.
h'{x)
Hx) dx
is
The
H{g(.a)
- g{b)};
and then
J^
lot 6
A->
-> ool
let
84.
Use Problem 83
85.
(.)
to prove that
Given that
(a)
^ ^dx
and
^"sinx^d. = j;"cos.^dx = |
evaluate
sin (x^
Explain
Ans. 7r/4
j^"
^1
sino.^ dx,
^^^
j,
^'^^^^^"'^
>
1,
converge.
'' ^""^ ''^"^^ '^''"*" ''^'
^.o
I
(6)
(6)
why
+ y^) dx dy
to evaluate the multiple integral in

the method of Problem 31 cannot be used
(a).
chapter 13
Gamma
and Beta Functions
GAMMA FUNCTION
The gamma function denoted by
which
is
w>0
convergent for
(see
recurrence formula for the
r(n) is defined
Problem
gamma
r(n
Chapter
18,
function
+ l)
by
12).
is
nT{n)
^2)
^'^'/^? ^"" '^ determined for all .>0 when the

values
Sues tor
fir Tii72''^nfr
l = 7i< 2 (or any other
""'T
mterval of unit length) are known (see table
below)
^
^'
In particular if % is a positive integer,
then
r(n
For
this reason r(w) is
Examples:
r(2)
l!
w!
r(6)
i,
5!
^(5)
r(3)
120,
mTtrirv
''"
takinf
TABLE
of
for example.
VALUES
n
and
.4!
~ 2! ~
^'^
v^
The process
GRAPH
(^)
^difference equation which has
Tin)
^^""^
7,
(5)
4) that
r(i)
See Problem
1,2,3, ...
sometimes called the factorial function.
can be shown (Problem
It
+ l) =
of the
is
^^^
(i)
as a solution.
+ ^)
(5)
called analytic continuation.
GAMMA FUNCTION
T{n)
1.00
1.0000
1.10
0.9514
1.20
0.9182
f(m)
5
1
1.30
0.8975
1.40
0.8873
1.50
0.8862
1.60
0.8935
1.70
0.9086
1.80
0.9814
Fig. 13-1
VV_-^
J
-3 -2
-1
2
3
1.90
0.9618
2.00
1.0000
A:
285
By
5.i
[CHAP. 13
2gg
ASYMPTOTIC FORMULA
If
is
parent.
for r(n)
inherent in a calculation of T{n) are aplarge, the computational difficulties

the relation
useful result in such case is supplied by
T{n
0<e<l
+ 1) = V2^n-e--e'"''-+''
which
For most practical purposes the last factor,
omitted. If n is an integer, we can write
n!
very close to 1 for large
is
(^)
"is
MISCELLANEOUS RESULTS INVOLVING
In particular
if
= i,
a;
r(i)
is
called Stirlmg's
sometimes
GAMMA FUNCTION
the
0<a;<l
= -r^^~
smfljTT
T{x)T(l-x)
n,
^/2^n''e^''
approximately equal to for large n". This

formula for n\
factorial approximation or asymptotic
where ~ means
(^)
can be
()
= V^.
22^ - 1 T{x) T{x
This
the duplication formula for
is called
(2,r-"'TM
r(x)r(.+l)r(.+|)...r(+5^) =
The
gamma
product representation for the

Euler's constant (see Problem 49, Chapter 11).
is
an
infinite
W:.
V{x
where
n(a;,
A;)
is
is
an
lim
iii^(a,
llllliiJ^
fcx
The constant
function.
limn{x,k)
is
{12)
+ i)(a; + 2)...(a; + fc)
sometimes called Gauss' ^ function.
r(a;
This
=
-
l^
+ \)
()
m = 2.
result (9) is a special case of {10) with
This
(9)
+ ^) = V^ r{2x)
the gamma function.
+ l)
V2^â;"e
f
HI
+ Y2^ +
the
called Stirlvng's asymptotic series for
20).
Problem
(see
asymptotic series
is
r'(l)
Ce-''\nxdx
-y
139
288^
51840a;3
gamma
where
1
"^
'
,j^x
"
'
'
function.
is
The
series in braces
Euler's constant.
{H)
T{x)
The
BETA FUNCTION
denoted by B(to, n)
B(m,n)
which
is
convergent for
m>0, n>0.
is
defined by
C x^-'{l-xY-^dx
See Problem 16, Chapter 12.
(ifi)
CHAP.
13]
The beta function
connected with the
is
gamma
function according to the relation
ipiM
Bim,n)
See Problem 11.

'"*^^'"^^' "^"^
^^ evaluated in terms of beta or
resuJt^Tre
Qi"Ti2m-I
valid
form>0
287
and
n>0
2n-l o
/3
Problems
[see
rr^^^ =
^/i
(10)
^
i
and
ti
(13)]
functions.
T(m)r(n)
'
2r(m + )
Two
useful
(18)
and
-^
r(p)r(l-p)
gamma
o<,,<i
(19)
See Problem 17.
DIRICHLET INTEGRALS
If V denotes the closed region
in the first octant
-j =
and the coordinate planes, then
if all
bounded by the surface
(~\ + ( A\
the constants are positive,
pqr
r(l+f + f + ?)
Integrals of this type are called Dirichlet

integrals and are often useful in evaluating
evamaung
multiple mtegrals (see Problem 21).
Solved Problems
The
1.
GAMMA FUNCTION
Prove:
(a)
T{n
+ 1) =
x-e-'dx
1, 2, 3,
r(2)
In general,
=
.
lr(l)
r(n
lim
in
1,
+ l) =
(6)
lim
n^
{-^"-" +
e-'do;
Put n
n>0;
nT{n),
1)
r{3)
2r(2)
r(n
n!
if
+ 1) =
r(n).
is
nl,
n = 1,2,3
x-e-'dx
^"-e-d.|
e-'d^
T{n
nv(n)
Um (l~e-)
ifn>0
=
Then
= 2-1 =
2!,
r(4)
a positive integer.
3r(3)
3-2!
3!
[CHAP. 13
2g8
2.
"2r{3)
F(I)
^"^
r(5.5)
_
~
6r(|)
(d)
5r(|)
30
i-ir(i)
3.
r(4)
u_
2!(1.5)(0.5)r(0.5)
_
"
r(3) r(2.5)
_
-
r(i)
2-2
2-2!
|r(|)
_
-
r(|)
,,,
(")
5^4-3j^
_5]_ _
_
~
r(6)
/.^
^'
315
(4.5)(3.5)(2.5)(1.5)(0.5) r(0.5)
6(|)(|-)r()
4
3
5r(f)
Evaluate each integral.
3.
(a).
r"xe-'d*
(6)
Let
a'e-'^dx.
r(4)
/-"/
2a;
Prove that
r(4)
5.
A/
''"''''
r(7)
6!
= -i^ =
2^
Letting x
r"e-"^dM
= u^
2(^)
this integral
= V^
becomes
using Problem 31, Chapter 12
Evaluate each integral.

(a)
^/ye-'^ dy.
r"
3-4.2^^
Letting
j/'
x,
V"
(,)
'^'
r"(e'^)'"""''dz
becomes
Let
=0'"
u.
Then x
rîZdM
Jo
^m
Letting
g-ax
ax'^y, the
".
When
a;
dx
We
(a)
r w-'/ê-dM
r(l/2)
rl
1,
(4 In 3)z'
0;
r(i/2)
when
a;
and the integral
V^
= 0, u =
The
integral
v^
^''
-Ju
where m, n, a are positive constants.
integral becomes
r{"T-Hi
Evaluate
Let
-'""'"' d.
if"
- In X =
becomes
7.
45
V^.
x-^'ê-'dx.
j/.
1(0""^
4.
3!
^^^
.;""
'"*
r(-l/2), (b) r(-5/2).
by
use the generalization to negative values defined
r(M)
i^'"""'
^^^^^^
-
CHAP.
13]
(a)
Letting:
(6)
Letting n
=
=
-|,
r(-l/2)
-3/2,
= -2V7.
îig^
r(-3/2)
_ -2\^ _ 4Vi
-3/2 ~
3
^(-1/2)
-3/2
Then
Prove that
8.
r(-5/2)
J^'
^^t^
where ^
(~^^'
fr
Page
177.
''"'"'*''"*'^'''''''"'^
become?"^
Compare with Problem
9.
particle is attracted
forTtirre'ch O.
^^
where
is
Lw
the
^^* '
"^
toward a
"''
''* *'''
by Nît^'s
'"
'"'
8,
--'"'"' dy.
fixed point
"
''
a positive integer and
is
+ l)y =
(m
u,
m>-l.
this last integral
with a force inversely proportional to
*'^ ^^^^^^^
^-^ -'
^^^^^^
^^
''"'"'' '' '*""''' '" '^' " ""'^ at

.
=.>
and
let
^'^ thTtlm:
be the origin.
Then
d^ _
k
- ^
^^dt'
mass of the
'"
Chapter
50,
"^'"^
'
= "JV^-
^=^
..".(lnx)"d..
289
*^
particle
and
>
A;
a constant of proportionality.
is
velocity of the particle.
(1)
~=~
Then
d'
- ~x
x = o, we
i^
mv^
2
at
_
=
-
find
'^^'~
'^'"^
lit"
t%".?l:it^:^jL'='ir"
taken for
the particle to go from
!'"'=^
=a
to
^-
"^^dx
upon integrating. Since
'^
cc
A In
^
"
^----^
=
rft
dt
^
dt
= -klnx +
'^
^ ^,M,.
dJ ^"^ (') becomes
:*i
(;g)
Then
a.
*i.
da;
= -A/^>/in^
-Vi>f^
-.* ---ses.
We
(^)
thus find that the time
given by
is
dx
Letting In a/x
The
10.
=u
or
a;
ae-",
this
becomes
BETA FUNCTION
Prove that
(a)
B(m, n) = B(n, m),
Using the transformation x
B(^,n)
(b)
(a)
Ix'^-Hl-xr-âx =
Using the transformation x
(fo)
~ l~y, we
sin^ $,
B(m, n)
sin^-i
cos^-'
^ dd.
"
have
j;'(l-,)->,-.rf,
= f^r-il~yr-dy =
we have
^x-Hl-xr-^dx =
(sin^*r-.(cos^.)-.2sin.cos.d.
J^
sin^^-'e cos""-' fide

n
Bin,m)
11.
[CHAP.
290
= ^^+n)
B{m,n)
Prove that
Letting z
= a:S we
Similarly,
r(w)
^''>^-
z-'e"dz =
V(m)
have
13
x'"-' e-' dx.
J^
Then
y'"'' e-y' dy.
= 4(^J%^-'e-^dx)(^J\--'e-^d2y'^
r(m)r(n)
Transforming to polar coordinates,
a;
p cos ^,
p sin
</>,
-co
p2<+n) -
J,7r/2
<f>
4(
"'
e-p2 cos'"
dp
sin'"-'
d^,
^p =
p"'+'"-'e-'''dpY
cos""-'<;6
d0
sin'""'
cos"""'
sin'"''0d0J
r{w +
w) B(to,
m)
X17-/2
= r(m + M)B(m, )
follows.
Hence the required result
using the results of Problem 10.
Chap.
The above argument can be made rigorous by using a limiting procedure as in Prob. 31,
12.
Evaluate each of the following integrals.
X'
({,)
x*(l-xydx
^"^'^
-'o
\/2-x
Letting x
4^2 j]
r(5)r(4)
'^^^^
B(5,4)
= 2v,
= 4V2 _
-f^dv
Letting
i/'
or
a'a;
that
sin^-"-^ 9 cos^"-' 6
3/
,-
v^(l-v)-^''dv
a'j\^'m-xr'^dx =
Show
/'
y*y/aF^dy.
13.
4! 3!
-^
r-
4V2r(3)r(l/2)
;(4)
aVx, the integral becomes
llMp)
= 2^1^+"^^
m.>0.
aB(5/2,3/2)
de
.-.
4A/2B(3,i)
This follows at once from Problems 10 and 11.
14.
Evaluate
(a)
Let
(a)
r'\m^9d9,
2m-l =
6,
2n-l =
Then the required
0,
(b)
i.e.
w = 7/2,
dS,
sin* 5 cos^
m=
l/2,
integral has the value
^^
in
(c)
Problem
^'^.^.
fcos^^d^.
13.
g|
64V2
1^
12.
CHAP.
13]
2m- 1 =
Letting
(6)
4,
2n
291
^(5/2) r(3)
the required integral has the value
5,
The given
(c)
integral
Thus
2m -
letting
_8_
315'
2r(ll/2)
cos* e de.
0, 2ri
-1 =
4 in Problem 13, the value
2 r(l/2) r(5/2)
is
2r(3)
15.
Prove
sm^ e de
J
'ô
=^
= ^^ j_:j;5 - (z?^2^4-6. ..p
cosêdd
-'o
tive integer,
'\':|;^-^^-
(6)
From Problem
13 with
^^
if
2m- 1 =
p,
an odd positive
is
2n- 1 =
If
= 2r+l,
2r[i(p
^^1
+ l)r(|) ^
2r(r+|)
sin^
From Problem
sin3.(l-sin^,)rf,
The method
X m^^
Letting
~o=y
J^
Evaluate
Lety* =
with p
The
_
"
-as
i^^ | = |
2.4.6-..2r
2>4-6--.2r
(2r
seen by letting
(c)
J^
în^ ^
[compare Problem
_ r
3j3,^,
1)
./2
- ^.
d^.
14()].
^.^,^
l-3'5
M'3-6-7
^
15"
_^__2i4
1*3
of Problem 14(6) can also be used.
sin e de
18.
integral equals
J-.Tr/2
^^- ^^"^"^
+ 2)]
sin3^cos^^d^,
(6)J^
15 the integral equals
cos- d..
j^
cos^cZ.,
^
The
^" ^ven posi-
^^
integer.
2r(2r-2)...2
r(r-l)...l-V^
2(r+|)(r-i)...^VF
ede
(6)
^^
the integral equals

^('
(a)
tt
8'
the integral equals
"^^"^
(6)
1)
Zir
we have
0,
X17/2
p = 2r,
If
(a)
_
~
357r
s;
2-4.6-82y
eT'
^'
= g-^, nuw
how that
mat r(p)r(l-p)
=
nP;r(l-p)=-j-!^
where 0<p<l.
''
"""r^'
the given integral becomes
y-'H-yy^dy = B(p,i-p) =
r(p)r(l-p)
and the result follows.
C J^
x.
Then the
integral becomes
result can also be obtained
C ^Lllax
+
4 J
1.
by letting
y^
tan
a;
e.
'^
_ ^V2
'^
4 sin
(,r/4)
4~
"'^
PôWem
17
.2
19.
[CHAP. 13
292
Show
that
x^
Letting x^
or
%y
IGtt
dx
9\/3'
2%/'^,
(\,/'^-i/W^)'iy-"'dy
lj^y-"'(-i--yy"dy
|b(|,|)
8 r(f ) r(|)
3
Show
= V2^^"e-
that for large n, n\
We
The function nlnx
1)
now
to
"
maximum
= n + y.
for x
Then
'
= w,
(1)
as
dx
(1)
is easily
shown by elementary
becomes
The following procedures in which we proceed formally can

is rigorous.
involved and we shall omit
suitable limiting procedures, but the proofs become
In (1
= y/n. Then
= Vnv, we
on letting y
r(K
It is of interest that
1)
from
(4)
+ x)
large a close approximation
is
=
^X
dx
use the result
(2)
When n
9V3
the analysis
be made rigorous by
them.
with x
riiW3
approximately.
X" e
has a relative
This leads us to the substitution x
Up
In
T(n
have
calculus.
FORMULA
STIRLING'S
20.
fr(i)r(t)
r(2)
^^
~"
2"
(3)
find
is
w" e""
we can
Vn
e-''2
I
dv
V^^n" e"*
also obtain the result (13) on
Page
(5)
286.
See Problem
DIRICHLET INTEGRALS
2L Evaluate
= jjJx-'-'yi'-'zy-'dxdydz
where V is the region in the first octant

bounded by the sphere x^ + y^ + z^ = l and the
to
coordinate planes.
Let
x^
= u,
y^
= v,
z^
= w. Then
rrr
~
JJ J ^
"'
du dv dw
2Vtt2\/v2Vw
1 jj
n(a/"-iv<p/2)-i ^(v/2)-icjMdvdw
(i)
the region in the uvw space bounded by

the plane u + v + w = l and the uv, vw and uw planes
as in Fig. 13-2. Thus
where
is
Fig. 13-2
= 1 ~M
74.
CHAP.
13]
Letting d
(1
- m)*, we
have
/*'"""
(1
- M)<fi+r)
r(ff/2)r(y/2
^[(^
so that {2) becomes
lW2)r(y/2 +
+
f
^!f .'^'l^y^
4rr[(^
+ y)/2 +
where we have used
22.
l)
r[(a
l]
(y/2) r(y/2)
^[^^ + ^'^'^ + ^]
+ ;8 + y)/2 + l]
^^'^''^
^>
The required mass

x^
+ y^ + ,' =
^^
In the Dirichlet integral

required result is
g
x^
Show
where
Let X*
= y. Then
Page
a'-a^-a'
2-2 -2
287,
let6
= c = a,
24.
Then
J^
l]
is
^,
-^
^^^ ^^^^.^^
the region in the
first
.^
+ 3/2 + 3/2 + 3/2)
_
^
"gil"
v^
{r(l/4)}'
r(l/4) r(3/4)
S
-yv -3.
4,ra
iî^^))'
17 with p
x^y^z\
octant bounded
'
= 1/4,
sink's; da;,
= lB(p + |,l) =
=/
1 r(l/4) r(3/2)
r(7/4)
r(l/4) r(3/4)
Prove the duplication formula

Let
+ ^ + y)/2 +
- /jfl
= r = 2^ ana
p=g
f
and a~
1
r(3/2) r(3/2) r(3/2)
r(l
r 2/-"Ml-2/)"*d2,
J ^
From Problem
>
fvT^^dx
that
8r[(a
"^nd the coordinate planes.
(20),
r(a/2)r(ff/2)r(y/2)
-
+ y2 + ^2^
23.
+ 1).
r(y/2
fff :cV^^ dx dy dz.
+ ^)/2
+ l)
+ 1]
f'
Find the mass of the region bounded by
by the sphere
293
= .^2
22''-ir(p)r(p
+ i) =
sin^" 2a; da;.
(P+i)V^
2r(p+l)
so that the required result follows.
v/^r{2p)
T>,.
Then tv,
the
[CHAP. 13
294
Letting 2x
= u, we
find
sin^"
(2 sin
dx
udu
X cos
a;)^"
u du
sin'"
2'"
sin'"
x cos'" x dx
J^
+ i)}'
r(2p + l)
22P-1 {r(p
=
Then
since I
2'"-'B(p
+f
+ i)
J,
"
2pr(p)
Show
that
2pr(2p)
follows.
"'^
25.
{r(l/4)}^
d4,
Consider
.^/.
r(i)V^
^^
-i=
cos d
iB(i,i)
-Yf^
{r(i)}'
^^
as in Problem 23.
^"^^
"
'
VW7
Jo
Vl-2sin'e/2
^0
Vcos'/2-sin'9/2
Jo
^""""^
^'"^'^^
V l-J-sin'0
the result follows.
26.
C"
Prove that
We
cos
J^
^
x"
have
"
iC
^ ,n
"<P<-^n
"^^^'^ =
= i;^
r(p)
f
Jo
2 r(p) cos
M"-'e-"dM.
(p7r/2)
.r-
'
Then
Jo
cos X ^^
'
_J^
r(P)
-ô
u^-ê-" COS X dudx
-'o
^ f^^^du
r(p)Jo
+ M"
Problem
where we have reversed the order of integration and used
17
Letting m* = -w in the last integral, we have by Problem
r"
Jo
M"
r+l?'^'*
Substitution of
27.
Evaluate
Letting
Problem
26.
x^
-1
~
(2) in (1)
r" f'"""^' J
2j ir^"''
_
~
22,
Chapter
ii-
2sin(p
+ lW2
12.
"
(2)
2cosp^/2
yields the required result.
cos x^ dx.
= y,
X ^^ ^^
2 \ 2 r(^) cos ttIa)
^ ivT/2
,
[see Problem 68(a)] are called Fresnel
This integral and the corresponding one for the sine
integrals.
CHAP.
13]
295
The
GAMMA FUNCTION
28.
Evaluate
Ans.
i^
^-^
(a)
(a) SO,
Evaluate
(b) 16/105,
(c)
e-'^dx,
(6)
Ans. (a)jr(i),
(6)^,
3t
Show
32;
Prove that
33.
Evaluate
Ans.
34.
that
(a);
(a) 24,
Evaluate
(6)
(a)
35.
Prove that
36.
Prove that
v'Je-v^da;,
J^
(c)
(c)
-3/128,
r(-7/2),
*^
is
Ans.
r(-m +
38.
{g) Find
Aws.
(a)
(a) 1/60,
(6)
is
Problem
49,
(6)B(3/2,2),
a;^(l-a;)'(far,
j^
xdx
e"- In
j^
called Euler's constant as in
BETA FUNCTION
W2,
(c)
41.
Prove that
(6)
J^
(16V^)/105,
a negative number
Chapter
Vd -
i)/a: drr,
(b)
-3r(2/3)
(-1)" 2" V7
l-3>5-(2m-l)
1)
(it is
equal to -y, where y
Ans.
(c)
(a)
1/105,
(6)4/15,
(c)
r\4 - a;T'^ da;.
(c) 3,r
Prove that
(a)
(a)
12;r,
cos'ede.
Ans.
(a)
3^/256,
cos=9sin^dff.
Ans.
(a)
16/15,
Vtan
- x'
(6)
:r
(6)
de
V3a;
(r(l/4)r
^^
sin'edtf,
45.
Ans.
*ô
sin* 9 COS* 9 de,
J-.T/2
'^^
(6)
{h)
J^
s-/V2.
^,
(6)
= 0.577215.
11).
B (1/3, 2/3).
(c)
tt''M4-)''^dt..
f"
v'hTaT^) dx.
X*
(c)
y'e-"' dy.
{a)
a positive integer,
Evaluate (a)B(3,5),
(6)-^
0-
m = 0, 1, 2,
The
24
| r(|)
r(-l/3).
(6)
>
(xlnxYdx,
(6)
(c)
'*'*'
where
r'(l)
(a)
s>0
(}^^
J^
lim T(x)
Prove that
Ans.
5v^
^"
37.
xê-^^da;.
(c)
^(4/5)
-xR,
(Inaj^da;,
if to is
r(l/2) r(3/2) r(5/2).
(c)
aj'e-'-dx,
(6)
dt
r(n)
f^-'^
f xê-dx,
(a)
EBm^
(6)
C t^M.
= _^^
(6)
5,r/8
(6)8/105
2,r/V^
46.
Prove that
f_^J^, ^^
^^^^
47.
Prove that
jje^'+iy^''
[Hint:
48.
[CHAP. 13
296
Differentiate with respect to b in
Use the method
Problem
of
31,
Chapter
--
Problem
>
' ^
"
46.]
procedure used in Problem 11.
12, to justify the
DIRICHLET INTEGRALS
in the
Find the mass of the region
49.
50.
^'"^^
yfxy-
52.
53.
Find the mass of the region bounded by the ellipsoid
Find the centroid of the region in the

Ans. x=^y = z = 21/128
54.
= 0,
if
the density
is
x-
i*
^^^ ^^"^^^^^ ^^""" ^" *^^
constant of proportionality
Ans. 4^/35
1.
octant bounded by x'"
first
+ u"'^ + z^'^ =
+ y- + z'' =
a",
1.
where
m>0,
is
given by
r(3/m)
Show that
is
1,
+ V + e%
(a^
+ y'" + z^'^ =
that the volume of the region bounded by
8 {T{l/m)y
3to2
x^'^
$+ |^+ f^
Ans.
its center.
Show
+y =
'^/24
square of the distance from

51.
xy plane bounded by
x-
the centroid of the region in the first octant bounded by
._._-_
-
given by
,,
>
a-,
where
m > 0,
3 r(2/m) r(3/m) ^
4 r(l/m) r(4/i)
'^
+ y- + z- =
55.
j'
Prove that
Let
[Hint:
56.
Evaluate
57.
Show
58.
Prove that
- a)' (6 - xY dx -
x-a =
f
(a)
^"
Let y
a!/(l
iXllSp^'^''
r,
4-
[Hint:
Let
Prove
61. D
ti,=f
that
[Hint:
62.
f 'a;"'-'(l -')""'
that
Prove *i,
J^
^^^
^.
^)m+.
x-{r+ \)y/(r +
p"
J^
Let x
Prove that
sin' 9
_
-
where p > -1, g > -1 and
Ans. (a).,
W ^^^^
where w,i>0.
in
PIT
TT
2 '^T'
B(m,TO)
^~(i
6 cos'
do
er^"
+ r)'"+''
where
to,
n and
_
"
B(m,w)
r are positive constants.
wherem,n>0.
2a'*6'
Problem 60 and choose r appropriately.]
- jT+f'a'
^
+ 1)
j/).]
sin' 'g cos'"-'

( sin' 9
a;).]
,
tan's d9
60.
1,
C ^(1 - x)ix-Z) dx.
(l>)
X"^'^
-.
- )-^'^' B(p +
__.
B{m,n)
(6
- a)j/.]
(6
imOL
that
[Hint:
(x
>
o.
CHAP.
63.
13]
m = 2, 3, 4,
Prove that for
m
[Hint:
Use
297
2""'
= (x-l)(x- a^Xx- a^)...(x- a-i),
the factored form x"-!

limit as a; -> 1.]
divide both sides
by
a;
1,
and consider the
XIT/S
[Hint:
65.
66.
67.
Square the
Take logarithms of the
In 2
using Problem
and write the limit as
result in Prob. 63
...
63.
v(^^^^^
^
x"
Discuss the cases p

sin
a;"
m -
(^-)"""""
.
(S),
Page
Problem 65 and
result in
dx
and p
dx,
-^,
2 T{p)
sm
,,.
,
{pTr/2)
m^
let
X
1
J*
70.
0<p<l.
'^
= 1.
(6)
Jo
71.
Let
x cos
Jr(x)
x" dx.
Ans.
(a)
A\/W2,
(h)
(1)"
If
>
0,
>
cot p:r,
0<p<l.
16
(a;/2)''"'"^''
^^ n\T{n + p +
l)
^ generalization of the Bessel function
'
positive integer. Prove that Jp{x) satisfies the equation
Problem 107(6) and

6
Ptt
^^.
a;
= ^
3\/3r(l/3)
-ir^ CSC
Referring to Problem 71, show that

results of
73.
dx
-r
where p may not be a

72.
3.
("-^dx
+1
Show that
and 4ac
>
(c)
of
{a) Jîx) = VZ/irx

Page 257 are valid for
sin x,
(6)
e- ""'+"" + '''' da; dj/
75.
on Page 286 from the result H) of Problem
{i5)
[Hint:
Expand
Obtain the
e"*'"*^'
resiilt (15)
Page
232, to the case
xY' + xy' +
J-in(x)
(a;^-p*)j/
^2/vx cos x,
(c)
0.
the
all p.
v/4ac
Obtain
(16),
6^ prove that
r
74.
286.]
>.]
*ô
/v.p-1
as a definite integral.]
= ^ In {2w).
In T(x) da;
J
(6)
v/2
hand side and use Problem 63 and equation
left
Prove that
(a)
da;
r() t(^^ vf^^
Prove that
[Hint:
a;
Take logarithms of the
Prove that
[Hint:
In sin
in a
on Page 286.
power
series
6*
20.
and replace the lower limit of the integral by
>.]
chapter 14
Fourier Series
PERIODIC FUNCTIONS
function f{x)
is
said to have a period
or to be periodic with period

The least value of T >
f{x + T) = f{x), where T is a positive constant.

least period or simply the period of f{x).
Example
The function
1:
...
all
sin x has periods
equal sin
"^.tt
is
The period
of sin
Example 3
The period
of tan x
Example 4
is
for
a positive integer,
is
all x,
called the
since sin {x + l-n), sin {x + Av), sin

6t
the least period or the period ai sin x.
nx or cos nx, where m
Example 2
if
is
2jr, Air,
However,
x.
(a;
+ 6jr),
is 27rln.
ir.
constant has any positive
number
Other examples of periodic functions are shown
as period.
in the
graphs of Figures 14-1 (a),
m)
/(x)
rz
(b)
and
(c)
below.
rr
(c)
(b)
Fig. 14-1
FOURIER SERIES
be defined in the interval (-L, L) and outside of this interval by f{x + 2L) - f{x),
The Fourier series or Fourier expansion corf{x) has the period 2L.
responding to f{x) is given by
Let
i.e.
f(x)
assume that
ao
where the Fourier
-^
-f,
'
coefficients an
and
\
n-rrX
& are
f{x) cos
-J
dx
w=
bn
If
/(a;)
n-irX
r"^"" L
0,1,2,
..
(^)
nirX
J-
f{x) sin
da;
has the period 2L, the coefficients a and b can be determined equivalently
from
n-nx
fix) cos
/(a;)
-^ aa;
(3)
where
c is
any
real
number.
^J^
sin
n-n-x J
-^ dx
In the special case c L,
298
(3)
becomes
(2).
CHAP.
FOURIER SERIES
14]
To determine
see that
2LJ~
co
we
Oo in (1),
r*^
is
the
(2)
or
with n =
(3)
0.
For example, from
Note that the constant term in
f{x)dx.
j^J
f^^^^^> which
use
299
mean
(i)
is
we
(2)
equal to
of f{x) over a period.
If L = Tr, the series {1) and the coefficients

function in this case has the period 27r.
or
(;8)
{3)
are particularly simple.
The
DIRICHLET CONDITIONS
Suppose that
(1)
f{x) is defined
(2)
f{x) is
(3)
f{x)
and single-valued except possibly at a
periodic outside (-L,L) with period
and
Then the
/'()
series {1) with coefficients
+ 0) and
+ e) and
f{x
lim /(x
e*0 +
number
of points in (-L, L)
2L
are sectionally continuous in {-L,L).
(a)
f{x)
,,.
/(aj
if
is
or
(;2)
{3)
converges to
a point of continuity
+ O) +/(a;-0)
^-^^
(0)
Here
finite
.^
if
a;
IS
,.
a pomt of discontmuity
- 0) are the right and left hand limits of f{x) at x and represent
lim f{x-i) respectively. For a proof see Problems 18-23.
f{x
-+0 +
The conditions
(1), (2) and (3) imposed on f{x) are sufficient but not necessary, and
are generally satisfied in practice. There are at present no known necessary and sufficient
conditions for convergence of Fourier series. It is of interest that continuity of
f{x) does
not alone insure convergence of a Fourier series.
ODD
and
EVEN FUNCTIONS
A function f{x)
are odd functions.
A function f{x)
are even functions.
is
is
called
odd
if
called even if
f{-x)
f(-x)
-f{x).
f{x).
Thus x\ x^-^x^ + 2x,

Thus x\
2x<'-4a;2
+ 5,
since,
cos
The functions portrayed graphically in Figures 14-l(a) and 14-1(6) are odd
respectively, but that of Fig. 14-l(c) is neither odd nor even.
a;,
tan
e^
3a;
+ e-^
and even
In the Fourier series corresponding to an odd function, only sine terms can be
In the Fourier series corresponding to an even function, only cosine terms (and
possibly a constant which we shall consider a cosine term) can be present.
present.
HALF RANGE FOURIER SINE

A half range Fourier sine or
or
COSINE SERIES
cosine series is a series in which only sine terms or

only cosine terms are present respectively. When a half range series corresponding
to
a given function is desired, the function is generally defined in the interval (0,L) [which
is half of the interval {-L,L), thus accounting for
the name half range] and then the
FOURIER SERIES
300
as odd or even, so that it is clearly defined in the other half of the

In such case, we have
0).
function
is specified
interval,
namely ( L,
an
[CHAP. 14
0,
^n
= Y
fi^) s^^
"T~
^^^ ^^^^ range sine series
'^^
(i)
bn
0,
an
= Y
PARSEVAL'S IDENTITY
cos-^dx
f{x)
states that
lj\f{x)rdx
L
if a
and
for half range cosine series
+ t(al+M)
bn are the Fourier coefficients corresponding to f{x)
and
(5)
if f{x) satisfies
the Dirichlet
conditions.
and INTEGRATION of FOURIER SERIES

and integration of Fourier series can be justified by using the
theorems on Pages 228 and 229 vi^hich hold for series in general. It must be emphasized,
however, that those theorems provide sufficient conditions and are not necessary. The
following theorem for integration is especially useful.
DIFFERENTIATION
Differentiation
Theorem:
The Fourier
and the resulting
COMPLEX NOTATION
Using Euler's
L g g L and
for
= V^
(see
f{x)
dx
provided that f{x)
x,
is
both a and x are in this interval.
a;
FOURIER SERIES
identities,
e'
be integrated term by term from a to
series will converge uniformly to
sectionally continuous in
where
may
series corresponding to f(x)
Problem
cos
sin
e^*^
6,
48, Chap. 11,
written as
f(x)
Page
cos 9
sin 9
251), the Fourier series for f{x)
CnB^"'"'"'
(6)
can be
^
(7)
n= so
^l
where
Cn
w^ j
/(a;)e-"'^^'^da;
(8)
In writing the equality (7), we are supposing that the Dirichlet conditions are satisfied
If f{x) is discontinuous at x, the left side of (7)
f{x) is continuous at x.
and further that
1,
,j u
A by
V.
should
be replaced
1
f(x
-^^
+ 0)+f(x-0)
-.
^
BOUNDARY-VALUE PROBLEMS
Boundary-value problems seek to determine solutions of partial differential equations
satisfying certain prescribed conditions called boundary conditions. Some of these problems
can be solved by use of Fourier series (see Problem 24).
CHAP.
FOURIER SERIES
14]
301
ORTHOGONAL FUNCTIONS
Two vectors A and B are called orthogonal (perpendicular) if A'B =
A2B2 + A3B3 = 0, where A = Aii + Aaj+Agk and B = Bii + Bzj + Bsk.
or A1B1 +
Although not
geometrically or physically evident, these ideas can be generalized to include vectors with
more than three components. In particular we can think of a function, say A{x), as being a
vector with an infinity of components^ (i.e. an infinite dimensional vector), the value of each
component being specified by substituting a particular value of x in some interval (a, h).
It is natural in such case to define two functions, A{x) and B{x), as orthogonal in {a, b) if
r A{x)B{x)dx
if
A vector A is
A'A = A2 = 1.
normalized in
(9)
called a unit vector or normalized vector if its
(a, b)
magnitude is unity, i.e.

Extending the concept, we say that the function A{x) is normal or
if
^b
{A{x)ydx
(10)
From the above it

having the properties
is
clear that
we can
consider a set of functions {^^(x)}, k
= 1,2,3,
.b
^mi^)
'l>ni^)
d^
m^W
{11)
X
In such case, each
also normalized.
is
The equations
member
{4>Jx)Ydx
m = 1,2,3,
...
of the set is orthogonal to every other
{12)
member
We
call
such a set of functions an orthonormal
{11)
and
{12)
of the set and
set.
can be summarized by writing
<kjx)'t>ni^)dx
{13)
'-'a
where
8m, called
Kronecker's symbol,
is
defined as
if
m^
and
1 if
m = n.
Just as any vector r in 3 dimensions can be expanded in a set of mutually orthogonal

unit vectors i, j,k in the form r = Cii + Caj + Csk, so we consider the possibility of expanding a function f{x) in a set of orthonormal functions, i.e.,
00
f{x)
^Cn4.Jx)
a^x^b
n=l
Such series, which are generalizations of Fourier

both from theoretical and applied viewpoints.
series, are of
(14)
great interest and utility
[CHAP. 14
FOURIER SERIES
302
Solved Problems
FOURIER SERIES
Graph each
1.
/(a) fix)
of the following functions.
0<a;<5
Period
-5 < a; <
-3
|_;
10
fix)
-- -t
3
-20
-15
-25
~*~
Per iod
--5
-10
10
IS
20
25
Fig. 14-2
Since the period is 10, that portion of the graph in -5 < a; < 5 (indicated heavy in Fig. 14-2
above) is extended periodically outside this range (indicated dashed). Note that f{x) is not defined
at a; = 0, 5, 5, 10, -10, 1 5, 15, etc. These values are the discontinuities of f(x).
sin X
Jib)
fix)
TT
O^x^TT
< < 27r
Period
2-0-
a;
/( X)
'-
-3:r
""
r^
-r
-2,r
/
/
iv
3^
2s-
Fig. 14-3
Refer to Fig. 14-3 above. Note that
0Sa;<2
g <4
4ga;<6
fO
(c)
fix)
= \l
Period
a;
x and
f{x) is defined for all
is
continuous everywhere.
fix)
Period
-12
-|
-10
-I
8
-4
-6
-8
-2
1
12
10
14
Fig. 14-4
Refer to Fig. 14-4 above. Note that f(x)
is
defined for all x and
is
discontinuous at
4, 8, 10, 14,
s>sm T^L
X.
k-TvX
2.
Prove
kirx
sin
C
J
kvx
kv.
da;
COS
kvx
Kit
-j^ ax
COS ?
ax
kirx
^ sm j^
kir
Li
-y dx
\_,
if
COS
--,
kv
-j svakir
KIT
fcTT
A;
1, 2, 3,
-; COS
( ^tt)
kiT
Li
j sm
KV
I
1
(Kir)
i\
a;
2,
CHAP.
d.
FOURIER SERIES
14]
Prove
(a)
cos
-^^ cos -j- dx
303
sm p- sm -^ dx
m=n
.L
/i,\
(o)
where
(a)
-^r
Sin
sm
Y
TIttX
7
'"^ttX
cos
^ô
^^
j^ da;
and n can assume any of the values
From
trigonometry:
cos (A
+ B)}.
Then,
cos
cos B
m-^n, by Problem
it
l{cos (A
1, 2, 3,
- B) +
+ B)},
cos (A
sin
B - A{cos(A-B)^
sin
2,
=
Similarly
irly
m =.,
m"n,
if
nirx
X.^rmrx
^sm-^sm-^do;
m = n,
If
nvx
mjra;
Note that
have
if
sin
m = m ==
cos
B =
J_^,,n
The
c,c
A +
show that
n=
(a)
- B) +
|{sin (A
/I
cos
\,
jdx
2m!ra;
h J
+ B)}.
sin (A
L,
f-,
(1
respectively.
Then by Problem
L^
gj,,!^''^
for
;^
J_^
2, if
sm^^jdx
w ?^ m,
=
and
-J_^
(6)
sm J da;
remain valid even when the limits of integration L, L are
respectively.
2/
a- cos -j--
1,2,3,
/(^)
cos-^da;
+ 2L
If the series
(m + nWx^
cos^-^jdx
these integrals are equal to 21/ and
results of parts (a)
replaced by
m = m,
If
(a)
da;
1 C
I
-;^
^sm-^cos-^rfa;
4.
nirx
X.sm j^ sm
We
we have
Xmrx
(6)
(m njirx
\ C
-J_^jcos^-^
b sin
-j-
converges uniformly to f{x) in (-L, L),
.. .,
cos"^
dx,
{b)
A +
K^^ j^^
fix)
sin"^
dx,
(c)
A = ^.
Multiplying
fix)
,
by cos
/(a;)
rriTrx
cos
da;
4-
=
Thus
om
to
Z/,
using Problem
3,
(i)
we have
cos-jdx
OmL
(âcos^ + 6sin^)
-j^ and mtegratmg from L

,
-^ j
"S!
J
<,an
r
I
(^g)
ô ^'^^ cos
cos
^"a;
-j^ dx
+
1
r''
ifm#0
f(x)
cos^^^
da;
if
m=
1, 2, 3,
cos
1
= sm rura;
-j- dx y
mjra;
[CHAP. 14
FOURIER SERIES
304
(6)
Multiplying
mtrX
sin^^
by
(-Z)
-L
LJ
lllLegXai/lllg' from
XrUIIl
ana integrating
and
Wi=
AUUlCiU 3,
XJ, using
UOmg aProblem
O, we
to
l-U L,
4lv:;
have
Ld
I
sm-jdx
fix)
Thus
(c)
sin
=
I;
(1)
from
r'
Putting
m=
The above
sm j^
cos
f(x)sin^dx
Problem
to L, using
/(x)dx
i a
(S)
-J-
dx
bn
sin
= sin -= dx >
bmL
Integration of
^ ax
= 2AL
when the
results also hold
1,2,3, ...
we
m=
gives
2,
or
in the result of part (a),
if
"^
find
C''
f^^^*^^
oLJ
f^'^^ ^"^
J^
^^^ so
A=
-L, L are replaced by c,c + 2L.
integration limits
Note that in all parts above, interchange of summation and integration is valid because the
assumed to converge uniformly to f(x) in {-L,L). Even when this assumption is not
warranted, the coefficients a and 6 as obtained above are called Fourier coefficients corresponding
series
to f(x), and the corresponding series with these values of Om and 6 is called the Fourier
corresponding to f{x). An important problem in this case is to investigate conditions under which
series is
Sufficient conditions for this convergence are the Dirichlet
this series actually converges to f(x).
conditions established below.
5.
(a)
Find the Fourier
corresponding to the function
coefficients
A)={3
(6)
Write the corresponding Fourier
(c)
How
should f(x) be defined at

series will converge to f{x) for
The graph
of f{x)
shown
is
^"^'^-^^
"0<:<5
series.
= -5, = and
5â;S5?
a;
a;
=b
in order that the Fourier
in Fig. 14-5 below.
Period
3
i
.....
1
-5
-10
-15
15
10
Fig. 14-5
(a)
Period
= 2L =
.
10 and
L=
5.
^""/(^)cos^dx
-(
sin-zo /
0,
5 \niT
lfn =
Choose the interval c to c
+ 2L
as
\fj{x)cos^dx
ifnÔ
lo
IJ cos^ dx
-5
= ^
dx
3.
to 5,
so that c
= -5. Then
CHAP.
FOURIER SERIES
14]
305
Sin
3/5
5
(b)
In COS
=l
_
=
"
Since f(x) satisfies the Dirichlet conditions,

to /(^
of continuity and
"
/(^
0)
a;
Expand
(a)
f{x)
< x < 27r
a;^,
in a
wn-a;
-
sm
sin
.
5
.
SttX
say that the series converges to f(x) at
(3
+ 0)/2 =
At
a;
-5,
and
5,
with period
f{x)
2ir is
<
a;
Period
= 2L =
""
Fourier series
shown
if
(a)
the period
is 2tt,
(6)
the period
in Fig. 14-6 below.
o
2)r
4,r
6-
Choosing c
^.
= 0, we
a;^cosna;da;
i{(.^)(^) - (2.)(:i^^) +
x'
have
=
bn
L=
f{x)cos~-dx
LJ
=
If
and
2n-
dx
2(r)
dx
a;''
sin
f(x)
x^
-T,
n-0
nx dx
'=^'^^^-(2.)p-i||^) +(2)
Then
/(x)sm
-^J^
5 +
we
10
Fig. 14-6
Period
If
a;
-2ff
-4s-
which are
x=5
5 S g 5.
r^
points
= -5
a;
/(af)
-6i-
all
as seen from the graph.
3/2
not specified.
The graph of
0<a;<5
3
3/2
then the series will converge to f{x) for
is
s-a;
^^ p^j^^^^ ^^ discontinuity.
3/2
L
cos
wtt)
z^j^^z:l^
1
+ 6 /
3s-a;
+ S^'^T
+ ^(^''^T
(.
we can
-5 <
f(x)
3(1
-^
~
rejT
3/2
points of discontinuity, the series converges to

redefine f{x) as follows.
6.
riTr)
series is
L y
Lj
cos
tlir
nirx \
6 sin -^^F^
p +
3(1
.i
n
nirx\
wti-
The corresponding Fourier

oo
(c)
= da;
? cos nx
n
4s-
Sin
COSi
(
\
J
nx
-4ir
This
<
By
a;
a;
= 2b-
the series converges to 2ir^
not specified, the Fourier series cannot be determined uniquely in general.
Ill
+ +
prove that
6,
j2
"
4
+ ^
2^ ;? =
and so
2,^^
^^
^n^
_
-
a;
TT^
4
rj-
^ + 2
the Fourier series of Problem 6 reduces to
-3-
32"^'" ^6~"
22
4^2
4:7=
and
and
the Dirichlet conditions, the series converges at
Then
ODD
At a =
27r.
Using the results of Problem

At
8.
<
valid for
is
If the period is
(6)
7.
[CHAP. 14
FOURIER SERIES
306
1(0
to
+ 4v') =
2ir^
^'
~&-
EVEN FUNCTIONS. HALF RANGE FOURIER SERIES
Classify each of the following functions according as they are even, odd, or neither
even nor odd.
mm
From
Period = 6
_ltltl
{_l
Fig. 14-7 below
it is
seen that f{-x)
so that the function is odd.
-/(),
fix)
2
Fig. 14-7
Prom
Fig. 14-8 below
it is
seen that the function
is
neither even nor odd.
-27r
Fig. 14-8
j'(c)
f{x)
^ x{10-x),
From
0<a;<10,
Period
10.
Fig. 14-9 below the function is seen to be even.
iw
y
/
\
/
\
/
\
VY
>^^ ^^s^
X.
/
/
\
\
//
\
\
\
\
^5
\\
/
/
\\
/
/
1
-10
Fig. 14-9
10
CHAP.
9,
FOURIER SERIES
14]
Show
that an even function can have no sine terms in
Method
Fourier expansion.
its
1:
No
sine terms appear if
6.
If
3O7
we make
0,
~f_J{x).in^dx
the transformation
i/(.)sin:^d.
= -m
a;
in the first integral
on the right of
(-^yu
fi-u) sin
-JJ['/Wsin^du
j^
Then
If /(x) is even, f(~x)
fî-u) sin^ du
(.)
and
in the last step that the
in particular x.
|^
f(-x)
(a,cos^ - 6.sin^).
cos^ +
6,
Thus from
(1)
sin^^
Hence
f{x).
2 6sin^
^'^d so
obtain
cos^ + K sin^) =
(a.
we
f(x)
Assume
2:
(1),
(1)
-ij''/(a.)sin^ dx
where we have used the fact that for an even function f(-u) =
f{u)
dummy variable of mtegration u can be replaced by any other symbol,
using (2), we have
Method
us write
this, let
j^fjix)sin^dx+Lfix).m^dx
hS^
To show
1, 2, 3,
0,
f{x)
i.e.
+
=
(a.
cos^ -
^+ 2
a.cos^
sin^)
and no sine terms appear.

its
In a similar manner
Fourier expansion.
10. If f{x) is even,
()
"^
--
Letting x
we can show
show that
Z Si
(a)
= u,
^fj(x)cosr^dx
an
/(^) cos
that an odd function has no cosine terms (or constant term) in
dx
^J
=
i /%(-.)
since by definition of an even function f{-u)

a.
(6)
11.
j;
f^
This follows by Method
Expand
f{x)
sin
a;,
f{x)
cos dx,
fl /(x) cos^
COS
(=^)d.
f(u).
Problem
< <
a;
tt,
+ i /'
0.
/('>')
cos^ dx
"
^f"fiu)cos^du
1 J^' /(;)
Then
cos^^ du + j-f^' fix) cos^dx

1 of
'^^
(b) 6
eos
dx
9.
in a Fourier cosine series.
Fourier series consisting of cosine terms alone is obtained only

for an even function
Hence
we extend the definition of f(x) so that it becomes even (dashed part of Fig. 14-10
below)
With this
extension /(x) is then defined in an interval of length 2^.
Taking the period as 2^, we have 21, = 2^
so that
L = v.
[CHAP. 14
FOURIER SERIES
308
f(x)
/
\
Y
2)r
-2ff
Fig.l4-H
By Problem
6n
10,
r"
H
=
{sin(a;
-2(1
and
f{x) COS
+ na;) +
%
sin
Ni
na;)}
sin
(aj
cos nv)
cos
nx dx
1 J
--!
+ l)a;
cos (w
cos (w
1
r-^
m+1
- l)a;
n-1
n^l.
if
-1)
a;
da;
n-1
+l
^^ dx
For
Ji
1,
di
sm X
cos
For
n-d.
(U,
sin
da;
a;
2sin'a;'
aa;
a;
= 2 ( cos x)
"^
1-
Then
2|(l +
IT
TT
cos.)^^^^^
n^
=2
4 /cos
2a;
-1
cos
cos
4a;
6a;
TT
12.
Expand
(o)
f{x)
Extend the
<
x,
a;
< 2,
in a half
range
sine series,
(a)
(b)
cosine series.
definition of the given function to that of the odd function of period 4

This is sometimes called the odd extension of f(x). Then 21, 4, I,
Fig. 14-11 below.
Fig. 14-11
Thus
On
6.
and
^(^\;^*=^-2-;
a;)
sin
"
Then
/(a;)
n=
_4
nV
dx
-4
nirx\
nirX
cos nir sin
n-a;
sm-g-
K-
^-^ Bin
(1)
nTT
nirX
sms
.
da;
2
2!ra;
Zrx
- 2^">-2~ + 3^'"'2~
,
-)
cos ms-
shown
= 2.
in

CHAP.
(6)
FOURIER SERIES
14]
309
Extend the
This
is
definition of f(x) to that of the even function of period

4 shown in Fig. 14-12 below
the even extension of f(x). Then 2L = 4, L = 2.
/()
.^^s
\
Fig. 14-12
Thus
0,
= 0,
ao
Then
aa;
niT-
if
1)
n 7^
X dx
f(=^)
X cos
-J
^(^)(i^^"T)-(K;^-T)
-j-i (cos
If
f(^)<ô.-^dx
lX
+ n=i
247(cos,r-l)cos^
nV
jra;
should be noted that the given function
by the two different series
- -êos- + 3,cos^ +
It
in (a)
and
377-a;
f(x)
5,73;
g,cos5|^
= x,0<x<2,
+
is
\
.)
represented equally well
(6).
PARSEVAL'S IDENTITY
13.
Assuming that the Fourier
series corresponding to f{x) converges uniformly to

f(x)
in {-L,L), prove Parseval's identity
1
where the integral

If
fix)
by term from -I,
assumed
is
^f
=i\
_i_
to
um^ dx
r'^
(which
^dx
nirx
f f_J{x) dx
^L
where we have used the
+ Mal +
bl)
nirx\
^'"TL~y'
^^^^ multiplying by f(x) and integrating term
justified since the series is
to exist.
~L~
is
uniformly convergent)
{a fjix)
cos
d^
we
obtain
f(x) sin
dx\
(al+bl)
results
J_^/(x)cos^dx =
La^,
f(x)sin^dx =
Lb,,
f''f(x)dx
La,
(2)
obtained from the Fourier coefficients.
The required
result follows on dividing both sides of (1)

than that imposed here.
less restrictive conditions
by L. Parseval's identity
is
valid under
14
1111
"
(a)
Write Parseval's identity corresponding to the Fourier series of Problem
(b)
Determine from
(a)
Here
L=
2, oo
the
(a)
sum S
= ;^2
2,
Prove that for
all
(cos n,r
rt
1),
jT+24 +
^ 0,
6,
34"'"
'^v^'^
'"
'
0.
|o
M,
positive integers
Li xJ -I,
=l
and 6n are the Fourier
S =
f'-on^^hich
l^+ S'
where
of the series
12(b).
identity becomes
Then Parseval's
15.
[CHAP. 14
FOURIER SERIES
310
corresponding to
coeificients
f(x),
and
f{x) is
assumed
sectionally continuous in {L,L).
For
Sm{x)
Let
M=
We
1,2, 3,
-j
+ bnam-j-J
facos-j^
sums
(1)
of the Fourier series corresponding to f{x).
/^t
{fix)
f{x)SM{x)dx
Multiplying both sides of
2
f(x)
(i)
limit as
and
Sl{x)dx
we
,
2
If the equality holds,
from
(5) into [3)
M^
we have
Sm{x) dx
Also, squaring (1) and integrating
(2)
SM{x)dx
{f{x)y dx
by 2/(x) and integrating from
(1)
SMix)Y dx
^l
13, gives
Substitution of
Expanding the integrand, we obtain
since the integrand is non-negative.
Taking the
2,^
this is the sequence of partial
have
Problem
(2.
2l\^ +
2^
-L
to L, using equations
{al
of
l||- + 2^K+6S)|
Problem
(4)
6?)
to L, using
(2)
-L
and dividing by
{3)
3,
we
find
(5)
obtain Bessel's inequality
n=l
J-'
^-L
Parseval's identity (Problem 13).
can think of Sm(x) as representing an approximation to f(x), while the left hand side of (2),
indicates
divided by 2L, represents the mean square error of the approximation. Parseval's identity
^ o the mean square error approaches zero, while Bessel's inequality indicates the possibility
that as
that this mean square error does not approach zero.
We
The results are connected with the idea of completeness of an orthonormal set. If, for example,
could never
were
to leave out one or more terms in a Fourier series (say cos Attx/L, for example) we
we
analogy
get the mean square error to approach zero no matter how many terms we took. For an
Problem
60.
vectors,
see
dimensional
3
with
CHAP.
FOURIER SERIES
14]
DIFFERENTIATION
16. (a)
INTEGRATION
Find a Fourier series for
12(a).
(a)
and
(b)
Use
f{x)
to
(applying the theorem of Page 300) and multiplying by 2
a;
To determine C
< a; < 2. Then
in
another way, note that
L=2
since
...)
(2)
n=i
represents the Fourier cosine series for x" in
(2)
in this case,
Then from the value of C
we have
in (a),
2=
%'
16*3
4*
3^
12
that term by term differentiation of the series in Problem

12(a)
Term by term
differentiation yields
(cos^ -
cos + cos -
Since the nth term of this series does not approach

value of X.
CONVERGENCE
18.
ttX
1
2vx
16 f
1
Stx
^{cos-^-^co.^
+ ^cos^-
n
C -
Iffi-l + l-i.-..^
C =
where
Show
find
^2
X
17.
^ ^"H"
"
12(a),
Integrating both sides from
(b)
integrating the series of Problem
w^ 1
From Problem
we
FOURIER SERIES
of
= x\ 0<x< 2, by
to evaluate the series
(a)
311
Prove that
of
(a)
not valid.
is
-..V
the series does not converge for any
0,
FOURIER SERIES
i
cos
cos 2
sin(M + i)^
cos Mt
2 sin ^t
<b)
^' ^Jo
(a)
We
have
sin(M + i)^
2smit
cos nt sin it
l)
l{sin (n
Then summing from w

sinÛcost + cos2t
""^
_
-
=1
to
in
2'
vrJ-.
sin (n
2'
- l)t}.
cos
Mt}
(sinft
(6)
,.
"^^
2sinii
M,
=
On
(M + i)t
sin
sin^t)
^{sin(M + i)t
(sin
(sinft
(M +
^)t
sinfi)
- sin(M-|)i^
sinî}
dividing by sin it and adding 1, the required result follows.
Integrate the result in (a) from -77 to

and
to ,r respectively.
smce the integrals of all the cosine terms are zero.
fix) sin
nxdx =
lim
f{x)
This gives the required results,
cosnx dx
if
f(x)
is
sec-
tionally continuous.
This follows at once from Problem 15, since
lim a = lim 6, = 0.
if
n-+oo
n-*oo
The
result is sometimes called
Riemann's theorem.
the series
^^
=i
(al
b^)
is
convergent,
20.
[CHAP. 14
FOURIER SERIES
312
Prove that
We
lim
f{x) sin
M-
(M + l)x dx =
is
if f(x)
sectionally continuous.
have
f{x) sin (Af
+ i)x
dx
{/(x) sin la;}
cosMx
da;
cos |a;} sin Ma; d
{/(a;)
J_
Then the required result follows at once by using the result of Problem 19, with f(x) replaced by
continuous if f{x) is.
f(x) sin ia; and /(a;) cos ^x respectively which are sectionally
are
a and 6 instead of ir and jr.
limits
integration
The result can also be proved when the
21.
Assuming that L =
2L = 27r, show that
Sm(x)
that the Fourier series corresponding to f{x) has period
tt,
i.e.
S (an cos
Using the formulas for the Fourier

an cos
nx
+ K
sin
nx
f^j'
r""
I
Then
SM(a;)
f{u) cos
18.
/W
cos n(M
(ft
Ji-
/(m)
dM
i r
f''
i
sin
/(m)
t,
tt,
+ x)
f{t
^^g
'^^
sin if
we have
+ (- j _
cos na;
Ma;
x)
sinnu sinna;j
/() sin
nw dwj
sin
dtt
du
cos
na;
+ i
+
(M +
.
6 sin a;)
/(w) cos
n(u
- x)
du
cosn(M-a;) ^dM
1)(m
- x)
; du
2sini(M-a;)
we have
1
sin(M +
r-^-'
A)t
Since the integrand has period 27r, we can replace the interval -jr
Thus we obtain the required result.
2jr, in particular -jr, tt.
a;, tt
a;
by any other interval
of length
22.
Prove that
From Problem
21,
1
sin(M
r"
+ l)J
Multiplying the integrals of Problem 18(6) by f(x
5^^-^
Subtracting
(2)
from
(i)
nx
iX/^"^'^'"
/()-]|
ttJ-^'^'
ux =
Letting
cos nit cos
L=
with
nu du\
/(m)
b sin nx)
coefficients
Also,
using Problem
na;
If".,
- 0)
2smlt
and f(x
+ 0)
jrj
sin(M +
l)t
respectively,
.i
sin ^t
CHAP.
FOURIER SERIES
14]
23. If f{x)
and
f'{x)
are sectionally continuous in
^^^
eontiJuouJ""'"""
^'^'
Thu<i
""^
/(<
+ a^) -
/(a:
+ 0)
'^
g sin ii
Then from Problems 20 and
JimSW.)
-{
/<-
'
'''""""''"^ ^ ^'^^^
22,
^'^^
gtg
because /{x)
1__ ^
2 sin If
Similarly,
inO<tS^
+ -)~n- + 0)
^.
'^
2 sin ij
f(^
lim
,_+
" "'""""^
tis'ts"''
fii^:0)
sectionally continuous
}lZ- ^''^l\7jr'^
'-0+
Zsm^t
''''''*''"" ^'^'^
eacf;
^'
prove that
.r),
+ o) +
fi^
lim s.[x)
(-:r,
3I3
,,^^
is
nt + .)-f(.
,"?+
-^'^^ '^-'l
sectionally
+ 0)
d--ative of
/(.)
at
,r.
-tt
<
0.
we have
+ /(-")
or
= /(^+ "> +
Jim^(.)
/(^
- 0)
BOUNDARY- VALUE PROBLEMS

24.
Find a solution U{x,
t)
of the boundary-value problem
_ 32c/
- âF
C/(0,f) = 0, C/(2,i) =
C/(x,0) =
at/
i>o,o<x<2
-Qt
i>0
0<a;<2
a;
"?^*'^ commonly employed in practice is to assume

the existence of a solution of the partial
..
differential equation having the particular
form V{x,t)
X{x) T(t) where X{x)^nA
ff^.
tions of . and * respectively, which we
shall try to deLminl.
reason the memoa
methLliis oiten
often"
called the method of separation
of variables.
mWrP
Fohht
Substitution in the differential equation yields
^(^^) = 3(^r)
(')
where we have written

Equation
(2)
and T
in place of
or
X(x) and
X^ = 3T^
dx
(^)
^^
dt
T(t).
can be written as
J_dr
3T dt
d^X
dx'
(^)
Since one side depends only on * and the

other only on x, and since x and
It is clear that each side must
be a constant c.
In Problem 47
we
see that
if c
0,
are independent
pciiueiii, variables
vdriaoies,
a solution satisfying the given boundary conditions

cannot exist.
two ori":r;'rernr'etu:ti:nr^'^''^^
f +3x^r
^"'^^^^"^ -''"'''
0,
-^
g+
-""''^
X^X
- -'' ^^- ^-
(^)
^''^^-
:=
(,)
whose solutions are respectively
solution
is
Cie-^'",
given by the product of
U{x,t)
where
and
are constants.
X =
A, cosXa;
Bi sinXa;
(5)
and T which can be written
e-''^''(AcosXa;
Bsinxa;)
ê)
[CHAP. 14
FOURIER SERIES
^.
We now
and B so that
we must have
seek to determine
the condition
t7(0, t)
0,
g-3x't(^)
(6) satisfies
JJ{x,t)
To
satisfy the condition
C/(2, t)
B=
m=
0,
1, 2,
Substitution in
(8)
i.e.
0,
avoid this choice and instead take
2X
Wjt or X
that a solution satisfying the
(10)
two boundary conditions
first
be used for different values

by B., indicating that different constants can
OT.
If
we now attempt
impossiblus^gS
jr^r-e
also
sofutLs
(called th^ principle of superposition),
U(x,
(i)
sine series.
The
0)
<
a;,
see,
^ J
X
it to be
we find
havmg the form
-J.
2,
solutions
Problem
12(a),
V,
U^)
t
= 0,
that {n) becomes
0<.<2
(^^)
function f{x)
equivalent to the problem of expanding the
solution to this is given in
<
are led to the possible solution
on placing
B^sm^
we
a.
J^Be- sin^
= x,Q<x<2, we
X
is
U{x,
However! upon recognizing the fact that sums of
the condition
This, however,
boundary condition
to satisfy the last
m.,t)
From
is
= B6^'''*-sin2n|
U{x,t)
of
we
(^)
now shows
where we have replaced
()
we must then have
0,
sin 2X
where
(^)
solution (8) identically zero,
makes the
satisfy
Be"'''* sinXa;
Be-"^'' sin 2\
Since
To
A=
or
so that (6) becomes
the given boundary conditions.
from which we
=x
<
for
see that
x <^2 into a
B =
'="-
so that [12) becomes
^
/t
actually a solution, we must show ^h^t
a /ormaJ so^^tion. To check that (U) is
^f
of
justification
consists
proof
The
conditions.
relitial differential equation and the boundary
accomplished
be
may
and
series
infinite
ttLTy ter^ Srentiation and use of limiting procedures for
Which
is
by methods of Chapter
11.
an interpretation in the theory of heat conThe boundary value problem considered here has
conduction in a thin rod or wire located
=
is the equation for heat
duction. The equation
f
=
fcg
and . = L if the surface of the wire is insulated ^^^^'

on the X axis between .
constant ^ The ^^^^^^''^'JlX
place x in the rod at time t
^/^
17(x, t) is the temperature at any
or escape
the conducting
thedensUyot
is
heat, and p
specific
the
is
s
eonduotvoity
thermal
Jwhere K is th;
^ha*
-^^-te
conditions UiO,t) = -d U L, t) =
materia!; is called the diffusivUy. The boundary
t
t>me
all
>
at zero units for
" units,
the end temperatures of the rod are kept
TîL rod i
is L - 2
2 units
the rod. In this problem the length of the
initial temperature at any point x of
while the diffusivity is fc = 3 units.
,
ORTHOGONAL FUNCTIONS
25. (a)
Show
that the set of functions

1,
sin^, cos-^,
forms an orthogonal
^^'^rtT'
^^ÎT'
set in the interval (-L,L).
'
'
CHAP.
(6)
FOURIER SERIES
14]
gj^g
Determine the corresponding normalizing constants

for the set in
orthonormal in {~L,L).
set IS
(a)
This follows at once from the results of Problems

2 and
By Problem
(b)
(a)'
^
so that the
3.
3,
dx
^^
j\lfdx =
Also,
Thus the required orthonormal
~,
2L
C ( ~k\ dx
or
set is given
dx
by
-^sin^, Xcos^, A.sin2E2
^cos^
26.
Find a Fourier series for
We
f{x)
shall take the period as
"'
= cos ax, -^^x^-., where
T:
1
- r"
l/sin(-wV
jr
ao
S'""'^
>"(*)
cos
{cos(a-)a;
2 sin
2:r.
L = .
na; da;
a-n
2L =
2. so that
cos (a
^ 0, 1, 2, 3,
Since the function

cos ax cos
is
even, 6.
..j
and
nx dx
+ w)*} da;
+ V]
+w /
sin_(a
_^
2a sin
a,r
^(a'
cos n^
- )
a,r
Then
cos ax
2a sin
4-
sinar /l
air
v;
2
COS
2q:
,
~U~^'^1^'"' +
27.
Prove that
sin
a;
a;
X' V
V
^
a;
,7-
in the
= gÎ^(J: +
a
7rC0ta:r
- i
a
is
of interest since
it
-^
2
cos
3a;
{2^)y\
{3^y^
Fourier series obtained in Problem 26.
cosa^
This result
2a:
^F^2^cos2a;
x^
-7V
Let
Wjt
-J^+
a^-1^
=
^"
2a
a=-2^
Then
2a_
^ a'-3=
_^^
a^-l^ ^ a-2^
^
.
2a
a2_32
represents an expansion of the cotangent into partial

fractions.
(i)
[CHAP. 14
FOURIER SERIES
316
S \a\ \x\<\
test, the series on the right of {!)
the Weierstrass
Thus
s rule.
L'Hospital
usmg
by
seen
is
as
and the left hand side of {1) approaches zero as a-0,
obtam
a;
to
to
from
of
(1)
we can integrate both sides
By
^'""'^
ln(
In
so that
sin.x
vx
Replacing x by xlv,
we
lim
'"^liTil^-M^-l
In
_
A_^V1_^^...r1-!^
-^s)M^)-0-s)
=
=
i-^
(^-s)(!-s
Ci_e^
...
(.>
obtain
^'^
sin^
^(^-V^K^'i^)'"
shown valid for all x. The result is of interest since

called the m/^ît. product for sin x, which can be
of a polynomial.
sin x in a manner analogous to factorization
it corresponds to a factorization of
2>2-4-4-6 -6- 8-8
TT
28.
Prove that
Let
a;
= 1/2
'
1-3
3-5
in equation {2) of
5-7
7-9
Problem
27.
Then,
(-^)o-i>)(-^)-
Taking reciprocals of both
sides,
we
^ a-i)(i-i)(i-o--
obtain the required result which
is
often called Wales' product.
^^^^-
^^^
FOURIER SERIES
3^^
FOURIER SERIES
"
^^^f^^^-{-l
(c)
*^-
trol",:ttlt^^^^^^^
/(a;)
Period 4
ltr<l
0<a;<10,
4a;,
{-
(.)/(.)=
Periodic
(d)
f{x)
^^
(0
Ans.
(a)
31.
"
32.
(a)
a;
(c)
a;
Expand
^^
^^^'^
sin^^
(6)
^ 2 ^ sin^
20
(c)
Ans.
i^
= 0, 2, 4, ...
= 0, 10, 20,
.
/(.)
(d)
-x
<
20
(d)
<
a;
Period b
(l-cos^)
,1^:,
- + 5 / 6(cos^7r-l)
â;
.^
3, 9, 15,
6 cosn^
â;l
^ ^^^^.^^ ^^^.^^ ^^
(a)
Expand
(6)
How
/(x)
<
cos x,
a;
<
should f(x) be defined at
^^^.^
^_
rL^tTvrl^^^r'Li:^'"'
a Fourier sine series.
in
:r,
a;
and
m=
a;
cosx
Ans. Answer
Expand
Ans.
(a)
f{x)
is
the
same as
in
Problem
-J8_^ 4<x<8
^2^ f
sin
Prove that f or
sin
,r,
:.
so that the series will converge to f(x)
0<x<.
result 01 i-roblem 32, explaining the

similarities
35.
S <3
-3<a;<0
,^^^^^^
A..l|{eos^ + êos2^ + i,eos^+...}
''
34.
J. JO
no discontinuities
(6)
2-^^
-4
se.es usin. properties o.
in a series of
(b)
^-(^ +
(6)
xW-x)
f(li^+^ + 5ii5._^__Nj
37.
Show
that
if
is .;
and
(6)
any.
show that
~+~-^-l + L + l__
(6) cosines.
-^cos.
i^ +
to
(a) sines,
^ % /2cosW2
xW-x)
Use the preceding problem
the period
32.
(a)
36.
if
and differences
3,r^v/2
A ^^^ ^
iorQ^x^n->
compare with the
[CHAP. 14
FOURIER SERIES
g^g
DIFFERENTIATION and INTEGRATION

38. (a) Show that for -jr < x <
FOURIER SERIES
of
tt,
By
(6)
a;
integrating the result of
(o),
By
integrating the result of

a;(T
39.
Show that
(o)
Use
(6)
By
<
\
)
show that for
-,r
/sin
12(-p
3^
s in 2a;
a;
cos 3a
25~
sin
2^"
"*"
3a;
"j
3'
ir,
3454x
^sin* + 2(j43sin2a; - a^lsinS* +
show that for
(a) to
-tt
sinx
as
40.
as
(6),
- a;)(7r + x)
cos X
a;
<
for -a-
cos2x
Vcosa;
tt''
(c)
3a:
show that for -ir^xîr,
= -^ ~ ^\^V~
*'
sin
2~"
^-j
^3
sin 2g
/sina;
tt,
cos
Problem
differentiating the result of
a;
35(6),
/ cos 2a; _
\V^
"
cos 3x
cos3x
lr\^ +
\
J
tt,
\
V
5^
3-5
"^
S
cos5x
3^
cos4x
2-4
prove that for
4 /cos X
tr
PARSEVAL'S IDENTITY
41.
35 and Parseval's identity, show that
By using Problem
2^;^ =
(a)
,n
42.
1
^^^
ot.
that*
Show *v,
43.
Show that
(a)
44.
Show that
^^.î.
2^ (2n
1)*
+
+ ^^-^ .
_L
"t"
5577?
96
'
= 5'~^
^,^
(''^
90
[Hint:
ig
^'
=i (2%
945
Use Problem
11.]
960"
-D
39
+ 2^T^^
3.
"^
F^l^^
"^
"
16
BOUNDARY-VALUE PROBLEMS
45.
(a)
^=
Solve
where
<
a;
< 4,
>
1/(0, i)
=^
0.
V{x, 0)
8 .in
2 sin
Ux.
and
solution.
Solve
0.
interpretation of the problem

(6) Give a possible physical
26-'^" sin 5,ra;.
sin,rx
Se""^"
V{x,t)
Ans. (a)
46.
0, 1/(4, t)
1/(0, t)
0,
t/(6, t)
0,
U(x,0)
= jj 5J*<6
and interpret physically.

"
Ans.
47.
t/(x,t)
.2.
Tl - cos(W3)1
cos (W3) ^_2^236
Jl
J^2|_
J'
each side of equation (5), Page 313,

satisfying the boundary-value problem.
Show that
if
'"'^*
^^;^
is
a constant c where c
S 0,
then there
is
no solution
CHAP.
14]-
FOURIER SERIES
'"'
^^^--"
. = and . = . o the . axis, its ends
transverse vibration the displacement
Y{x, t) from the x
ttr^'l^::ti:2'Xl:n'^^^^^
points. When set into small
r^^
=^S
StllS'^"^"*^^^"'"^^^^^^ ff
'^
Where
the
fi1iô:rm?='o\7:T^"'
"' ^('^'
^(^. 0)
0'
*)
gjg
sm X +
0.1
....
a^
r/,.
r = tension,
e,a.io) with a^
0.01 sin Ax, Y.(x, 0)
=4
for
Interpret physically the boundary

conditions in (a) and the solution
Ans. (a) Y{x,t) = 0.1 sin x cos 2f
+ 0.01 sin 4a; cos 8t
= mass
which
<x<
per
satisfies
>
^, t
o.
(6)
49.
Solve the boundary-value problem
(a)
Yix, 0)
Ans.
50.
(a)
- .),
0.05.(2
Y(x.t)
Y.ix, 0)
g^ =
<
where
0.
^ 5 ^^^
Solve the boundary-value problem
51.
Z7(a;, <)
f = g,
2^+5
< 2. * > 0.
1/(0,
t)
'='"' ^^^'^
4_cosm^
(6)
If
~^<x<v
and a
7^ 0,
If
< <
77
a;
jT,
t7(.,
1,
"
'' *
*)
1, 2,
r(2.
U(x, 0)
*^ '^^^^^-^-^
^-^-"
50.
Prove that
sinh
Prove that
cos
Show
a;
a;
a;
(^1
(l
that
a;
=:
(a)
a2
irV2
_J_
2 sin 2x
2'
- a^
a"
3 sin
'^
-ii^
a cos
a;
+V
Vl
a"
32
3a;
_ 2
^'
,
^x^
a;).]
11
13
3 sin
2
2a;
+ 2^
(3.rA^~(5^
15
2-2-6'6-10'10'14-14
2a;
+ 22 +
^LV,
cos
g.
^Y 1
_ i^
2 sin
;^2
J_
2a
-Â^
,
(sm
2x)/(2 sin
(6)
a;
sing
V
cos
sin
V -a'
aTT
cosh ax
2 sinh OTT
*)
2.
,,
prove that
.,
;r
(6)
r. ,
[Hint:
ajT
5sinW
2 Sinh
58.
3,
^^"^^
prove that
(a)
57.
0,
Verify that the boundary-value problem

of Problem 24 actually has the
solution (U), Page 314.
2 sin
56.
gi^^^sin(?^L^sin2(l2i^
5Sina
55.
*)
Interpret physically.
54.
sin ma;

Problem
An.. F(.,.)
53.
=:
-" ^^^^^^ cos ^^^n-lM
Eary1onS;nsL''yt^;]''^'' "'
Ans.
...
i/ 4.4.8.8-12-12-16-16... \
V3'5'7-9-ll-13'15-17..y
3a;
+ 32
_
-<!
[CHAP. 14
FOURIER SERIES
59.
Prove that
(a)
if
a' 0,1,2,
.,
then
2a
Prove that
(6)
<a<
if
_|^
2a
_ +
1,
dx
2a
(c)
60.
From
(a), (6)
and Prohlem
17,
Chapter
13,
_2a
2a
complete the proof of the fact that
Show that
Let t be any three dimensional vector.

(a)
{T'iY
(r- jf
(r)S
('
('"
J)'
inequality and Parseval's

and discuss these with reference to Bessel's
61.
w=
a minimum when
If
_,_
{0()},
1,2,3,...
^'^'^'
c,
Hx)4>Ax)dx
series.
Discuss the relevance of this result to Fourier
"^
identity.
{f(x)
is
nI
Cn^.Wrd*
is
chapter 15
Fourier Integrals
FOURIER INTEGRAL
The
Let us assume the following

conditions on
1.
2.
/(.) satisfies the Dirichlet conditions
\f{x)\dx
converges,
i.e.
(Page 299) in every
finite interval
f(x) is absolutely integrable

in (-=c,
Then Fourier's integral theorem

fix)
fix):
j^
(-L L)
oc).
states that
{A{a) COH ax
B{a) sin ax} da

(_^)
^()
Bia)
If
fix) COS ax
where
The
dx
^^^
fix) sin
ax dx
result (l) holds if is

point of continuity of fix)
'
''
wo must
the
aho.
The
Sht
replace /,)
by 0)ifc<>)
conaitions are
sucient\t not
If
i.
.
,
" "'' ' "-""""""'y.
..
^^ ^ """"
lei:
^^^^
''"'^
'"'
S^^l-SrrS-^-^---- ~^
EQUIVALENT FORMS
of
FOURIER'S INTEGRAL
THEOREM
Fourier's integral theorem

can also be written in the
forms
^^""^
"
'^
X=o
J= _
^^""^ '^"^
"(^
~ ") ^ <^
(3)
and wetaVe""'*^
fix)
^^''^
^^" be^simplified somewhat
=
=
^S^
cos ax da
'^" "^
^'^
if /(.) is
either an odd or an even

function,
aUdu
f^
fiu) cos
^^^^ '^" '^^ '^^
321
if /(^) is
even
^f /(=^) is
odd
(5)
t'^^^P- ^^
FOURIER INTEGRALS
322
FOURIER TRANSFORMS
From
follows that
{A) it
if
then
r./
n^-^
:^jy^"^
Fmjripr transform of
+V,*.
^ Tjf,;^rT^tl:.î^li^":^'^^''^'^^r
is 1/2..
product
If f{x) is
The above
is called
fix)
and
trlJfor^
is
of
sometimes written
^M
and
is
w-itten
the si/mmetrtc /orm.
an even function, equation

iF.ia)
^'^
'""''''
(5)
yields
^||J^(u)c0Sl.d
^^^
/(a;)
and we
call i^c()
If f{x) is
and we
and
and
-Fsia)
/(a,)
/(x)
G.W
and
(6)
if
F.()
^^^^
Fs()sinxda
-\(|J^
other.
iôner sine trans/orms of each
for
FOURIER INTEGRALS
of
are Fourier sine transforms
/(.)
and
==
.(x),
HD
and
(i^)
/(.)
and
.(.);
then
become respectively
{fix)Vdx
i^')
f{F4a)Vda =
j\fi^)Vdx
(^^)
relations are
Lown
as
^--fV^/,^^^^^^^^
F{a) and
Fourier transforms. Thus if
tions hold for general
prove that
0(x) respectively, we can
and
(^^)
f{x)g{x)dx
f{Fs{a)rda
of fix)
,.
,
,^
then
.(.) respectively,
J^
tralsfonns of
and Oew"are Fourier cosine
/(I)
In the special case where
The above
yields
>f|J^()sini.d^
fF.{a)Gs{a)da
Similarly
Fc{a) COS ax da
other.
Fourier cosine transforms of each
fix)
PARSEVAL'S IDENTITIES
F.W
-\||J^
an odd function, equation
call F.(a)
If
(x()
CHAP.
15]
FOURIER INTEGRALS
where the bar
The
signifies the
F{a)G{a)da
%/
00
323
f{x)]g{x)
)
dx
^g^
complex conjugate obtained by

replacing
by
-i.
See Prob. 30.
CONVOLUTION THEOREM
If Fi.)
and Gia) are^the Fourier transforms

of
/_F{)G(a)e--d
If
we
define the convolution, denoted
"
and gix)
respectively, then
j{u)gix-u)du
(^5)
by /*^, of the functions / and

^ to be
^*^
/(.)
Z^**
:^^J_.^(^)^(^-^)^^
(17)
then (16) can be written
J if* 9}
transforms.
^ran^/orms.
This
J{f}J{g}
^^g^
called the convolution
is
theorem for Fourier
Solved Problems
FOURIER INTEGRAL
The
1.
(a)
Find the Fourier transform of
(b)
Graph
(a)
The Fourier transform
/()
and
its
of f(x)
/:
For
tt
The .raphs
0,
we
of /(.)
21
o
-3
-2
Fig. 15-1
fix)
= I^
'
\0
Fourier transform for a
giaa
FOURIER TRANSFORMS
and
obtain
and
F^
<a
|;r|>a
= 3.
is
F{a)
|a;(
g-iao\
,or a
V^
=B
;2 sin
ei
V^
la
aa
a.
are shown in Figures 15-1 and

15-2 respectively
Use the
(a)
2.
[CHAP. 15
FOURIER INTEGRALS
324
F(^
= -i- f"
Then from Problem
if
then
dM
/(M) e'
The integrand
and (2), we have
^"^
"
!xl<a
1/2
lo
1 f"
r" sinaa_cosa ^^
\x\
|a;l
>a
sin
aa cos
the result of
1 in
^^
^-=0
and a
gg sin ax ^^
TT
l^l
<
^/g
||
a;
>
io
(a),
(2)
is zero.
Then from
we have
sm a
rîlL^da
sin
odd and so the integral
in the second integral of (2) is
f"
FWe-'^^da
left side of {1) is equal to
The
If X
-7=
rl
r J2sinaa^_,,^^
V2^ ^-
(b)
/(a;)
1,
_1_
(i)
sin aa~cos aX ,
f"
J_^
evaluate
1 to
Fourier's integral theorem,
From
(a)
Problem
result of
or
TT
J
d
__
'a
since the integrand is even
3.
If
an even function show that:
/(a;) is
(a)
= VIX"/(^)'=os^'^^'
F{a)
We
have
(a)
(1) is
^^^-
V2;^-^-
even and /(M)sinXtt

zero and the result can be written
f{u) cos
Xu
f(u) cos
au du
-Vl
(a),
/(m) cos
integral on the
au du
Then by using a proof exactly analogous
F(a) so that F() is an even function.

F(-a)
result follows.
required
the
(a),
in
to that
From
Then the second
odd.
is
is
-^
F(a\
vi^J-
If /(z.) is even,
right of
(6)
^W^^^'^^^'^'
'V^X
"
^"^
^^""^
(^)
can be obtained by replacing the cosine by

similar result holds for odd functions and
the sine.
4.
Let
sll
f(x) COS
/(^)
ax dx
f(,x)
cos
and choose
F(a)
^|lj'
F{a) cos ax da
of'
a)
5r
Jo
(1
axdx
cos
aa;
da
FW
=
_
|1 "
^7 /
2(1
^^^
\f
cos x)
-^
'^^
''ill'
^^ "" "^
'^ "* '^
^^-^yP-^-^'
^^^^-
FOURIER INTEGRALS
^^J
Use Problem 4
5.
to
show
f"^
that
2'
,2
As
obtained in Problem
325
4,
COS X
cos
-i
V
(
aa; da;
'^
-I
n<l
Taking the limit as a->0+, we
find
J
But
this integral
C" 2
can be written as
COS
6.
= 2m,
Show
dx
sin' (x/2)
Jo
X
a;
a;*
^^'"^ becomes
^"^
x^
~^ du
f^sinû
J^
on letting
so that the required result follows.
r^^d.
+
that
'^0
Let fix)
a^
e-' in the Fourier integral

theorem
''^*'^
"
-J^
But by Problem
X^Q.
2^-^
22,
Chapter
12,
cos
'^"^ a;
^ J
aa;
we have
da
da
awdw
e"" cos
/(m) cos
f^-cosattdM =
/o
\u du
e-
-^_
+1
da
Then
= e~'
PARSEVAL'S IDENTITY
7.
Ven^y Parseval's identity

We must show that
for Fourier integrals for

the Fourier transforms of Prob
{/(a;)}2da;
where
/(.)
This
is
Wj"
J J
,,
i.()
{F{a)r da
V^?m^.
equivalent to
/ W'dx
r^^da
=
ri?}Iâa
=
2
-'-o..
f"?!!!!^ da
Jo
r sin'
i.e.,
Jo
By
'
letting
used to find
J^
= and
r^du
using Problem
directly.
da
a'
5. it is
a^
seen that this
is correct.
The method can alsobe
PROOF
8.
[CHAP. 15
FOURIER INTEGRALS
FOURIER INTEGRAL THEOREM
of the
limitmg
Fourier's integral theorem by use of a
Present a heuristic demonstration of
of Fourier series.
form
Let
where
a.
= 1
Then by
If
oo
fix)
/W cos^
-^
and
du
11705
/^
(âncos-^ +
Ksm ^^^
^
.
(1)
jjM sin^
= ^
6.
du.
substitution (see Problem 21, Chapter 14),
we assume
/"
that
\f(u)\
du converges, the
term on the right of
first
(.)
approaches zero as
approach
L-ô, while the remaining part appears to
2
^ n=l
lim
L_
This last step

Calling
is
Aa
Li
r_
fiu) cos
y^{u-x)du
^
(3)
heuristic.
not rigorous and makes the demonstration
t/L,
{$)
can be written
f(x)
lim
Aa-*o
n=
^ ^( ^)
where we have written
But the
limit
U)
is
is
/(m) cos a(M
- x)
CF(a)da
1
-IT
Chap.
To be
COB
rigorous,
we
start with the
a(u-x)dx
-^-oo
is
considered in Problems 9-12.
12.
r sin
Let .
Cm
fda
Jo
method
and examine the convergence. This
10.
du
Fourier's integral formula.
i"*^^"^^'
(6)
- x)
f{u) COS a{u
If^daj
provide a possible result.

This demonstration serves only to
29,
(5)
du
equal to
Hx)
which
i r
f ()
-!/.
Then
at)
J_^-ir-'^"
Riemann's theorem states that
if
lim
F{x)
is
]=o J
^Hlldt,
^
= -.
^
F{x) sin
aa; da;
{a, b),
then
(see Problem
with a similar result for the cosine"
31).
Use
this to prove that
^"^^-
FOURIER INTEGRALS
^^J
where
/(x)
and fix) are assumed sectionally continuous
Using Problem
(a)
9(a), it is
{f{x
+ v)~f{x +
(0,1,)
proof of this
11. If fix) satisfies
is
since
_^lim
We
F(^)
and
exists
analogous to that in part
:=
|/(.
+ 0),
F(v)
^'^'
/(^^
+ ^) -
f(x
"^
+ 0)
sectionally
^.
IS
/(x) is sectionally continuous.
we make
{a) if
the additional condition that
i Sj(- + -)~^<^v
()
0) respectively.
0)}^^^^l^dv
This follows at once from Riemann's

theorem, because
continuous in
and {-L,
in (0, L)
seen^that a proof of the given result

amounts to proving that
lim
(6)
32^
lim
ib)
use of Problem
|/(x)|dx
fix
9(6).
converges,
prove that
+ v)'J^dv=
Ifix
- 0).
have
/(.+.) 115^
r/(.+.)^d.
r/(.+.)^
dv
(1)
dv
(2)
Subtracting,
sin
at)
(3)
J\fix + v)-fix + 0)}^dv +
Denoting the integrals
in (S)
by /,//, and
J^"/(.
+ .)^d, _
respectively,
we have
j?/
fix
= L + ll+I.
+ n\
0)
I
sin
ai)
d-y
so that
Idi;
Also
<
s
l/J
l/r^_um:
\f(x+o)\
f
f'
sin ai;
^^
Jince j;
!/(.),
d.
and
we can
f '-^
dv
both converge,
choose a so large that 17,1 < f/q

sufficiently large, so that the
required result follows.
\h]
e/3.
Also,
we can
Ti,^.,
*^
choose
//\
^'^
This result follows by reasoning exactly

analogous to that in part
12.
Prove Fourier's integral formula where
We
must prove that
Since
^fjiu) cosa(x-M)
du\
y()
COS
(.-.) du
reverse the order of integration to

obtain
"'
so large that
i.
"^""^
in
"'
\fiu)\du
~ u)
du da
./3,
''""' """^
(a)
fix) satisfies the conditions stated

f (u) cos a{x
'
|7.|
f(^
which converges,
+ 0) +
it
on Page 321.
fix
- 0)
follows by the Weir-
converges absolutely and uniformly for
all .
Thus we can
[CHAP. 15
FOURIER INTEGRALS
328
- (
da
where we have
Letting
Lett
f{u) COS a{x
/(m)
^(^)
du
cos
SinL(u-a;)
o:{x
- u) du
^^
a;
u = x+v
let
i->
-u) du
we
==,
see
f(x
^
by Problem 11 that the given integral converges to
+ 0) +
/(a;
- 0)
as
required.
13.
^=S
Solve
U{x,
t) is
bounded where x >
0, t
U{0,t)
U{x,0)
0,
Jq
y.^^,
> 0.
partial differential equation

proceed as in Problem 24, Chapter 14. A solution satisfying the
Chapter 14, the boundary
Problem
24,
Unlike
\x.
sin
Be-^"'
by
given
is
and the first boundary condition
that all values of X are possible.
conditions do not prescribe specific values for X, so we must assume
to an integration
By analogy with that problem we sum over all possible values of X, which corresponds
in this case, and are led to the possible solution
We
B{\)
e-'^
'
sin \x
d\
J'*
where B{\)
is
undetermined.
By
the second condition,
sm
AK aA
B(X)sinX.dX
ii(K}
we have
[iQ
'^lt\
s1
a;
(^)
from which we have by Fourier's integral formula

2 f"
I
B(\)
,.
f{x)
so that, at least formally, the solution
U^^^t)
is
2
-
sm \x dx
r'
_
-
sm ,\x dx
,
-cosX)
2(1
-^
(3)
given by
^j]7 ^~^""^ V-^''sinXxdc.
See Problem 26.
14.
Show
that e'"^'^
Since
e'''^'^ is
Letting
is its
own Fourier transform.
even, its Fourier transform is given
yf2u and using Problem
32,
Chapter
4.r e-^cos(aV2)dw
which proves the required
15.
by
^/2U
e'"'''"'
12, the integral
becomes
^.^-^ =
cosxa dx.
e-^"
result.
y{x)
where
g(x)
and
g{x)
y{u)r{x~u)idu
I
r(x) are given.
Suppose that the Fourier transforms of y{x), g(x) and r(x) exist, and denote them by Y(a), G(a)
and R{a) respectively. Then taking the Fourier transform of both sides o the given integral equation,
we have by the convolution theorem
Y(a}
G(a)
^f2^Y{a)R(a)
or
Y{a)
=
1
- V^-B()
CHAP.
FOURIER INTEGRALS
15]
Then
y{x)
assuming
y-
-1-
329
GM
this integral exists.
The
FOURIER INTEGRAL
16.
(a)
and
FOURIER TRANSFORMS
fix)
^^^^
1^1
'
|a;|
Determine the limit of this transform as
(6)
.
17.
18.
sin ae
(6)
Evaluate
/(x)
( "^"^^^r
^
I
f(x)
ir
fin<^
19.
V|(^-^^)
(a)
the
(5)
Show
Ans.
\/27^
(a)
| 1
" a^'
I'"!
<
~
fj
Fourier sine transform,
(a)
d.
its
(6)
Fourier cosine transform of
e'',
0.
= I e--, m >
by using the result
why
in (a).
the result in
(6)
does not hold for
+ a')]
Solve for Y{x) the integral equation
J^
Y{x) sin
xtdx
J 2
gi<
a 2
and verify the solution by direct substitution.

Ans. Y{x)
(2
2 cos
a;
4 cos 2x)/vx
PARSEVAL-S IDENTITY
21.
Evaluate
(a)
J^
[Hint:
^-^,
Use the Fourier
sine
f{x).
transform.
the viewpoint of Fourier's integral theorem
[ff/(l
^f ^.
Find the Fourier sine transform of
WÊxplain from
20.
(6)
(a)
that
result.
) cosfd..
In each case obtain the graph of f{x) and
Ans.
and discuss the
""
- 0+
(a)
If
-'
>e
(6)j^"_^!^
by use of Parseval's
and cosine transforms of e-, x
>
0.]
identity.
Ans.
(a)
W4,
(ft)
^/4
[CHAP. 15
FOURIER INTEGRALS
g3Q
22.
Use Problem 18
23.
Show that
show that
to
r"
(g cos X
^^
(a)
sin
Ze
ax
(^
^^
_
-
xf j
''3'
1'
0,
17(^.
t)
"*
^^
J^
77
15-
24.
25.
^.
U(0,
Solve
(6)
Give a physical interpretation.
f = ^.
Solve
An.
26.
^=
(a)
(a)
Show
t)
lj;"(^
t7(..*)
17.(0,
-',
= {^
^(..0)
0,
^f
28.
Establish equation
29.
Prove the result

F(a)
If
(i),
(18),
Page
Page
321,
from equation
Now make
^^f me^-du
the transformation
u+v
f(x)
(S),
where the bar
From
(a)
signifies the
> 0.
.> 0,
G(a)
'^^
and the conditions of Problem
g(x)
321.
=^
^1
| ^
\x\<l
jail
>
'
ei''^''+'
F(a)
(see
e^"^
dv.
then
f(u)g(v)dudv
x.]
GW
- (14),
Problem
^^ fjM
da
f(.x)
complex conjugate.
obtain the results (11)
Prove Riemann's theorem
Page
and
F{a) and G(a) are the Fourier transforms of
31.
0, t
323.
F(a)G(a)
(6)
>
bounded where
" 7=J
^=^
(a) satisfies
Verify the convolution theorem for the functions
(a) If
bounded where x
is
^(^.
'^lt\'
e-"^'^^
27.
30.
>
+ ^^^)e-='cosX.dX
Prove directly that the function in
[Hint:
that the solution to Problem 13 can be written

t7(;,t)
(6)
V{x, 0)
0,
Page
10).
322.
and
g{.x)
/(*)
respectively, prove that
f'*
13.
>
0.
chapter 16
Elliptic Integrals
The
INCOMPLETE ELLIPTIC INTEGRAL

u
tfLt u
^rst
first kind.
Hnd
by
Dy
F{k,4.)
^^nSf
'"^
''''
or simply K. For I^l'""'"

mUrlL^yK'^Fol
all purposes
A(/<;)
The
f ~Ji=^
the amplitude of F(k, 4,) or u, wi

""'*^" ^
'' 1^' ^^
^f .^' L.êndr.'s
The integral, ''
is also mllpri
r /
FIRST KIND
of the
^"^ ^' ^"^ ^

/orm /or f/..
'"''^'
'^ '^^
-r^^Jt
will be assumed that
of the
form for
integral
is
(n
i '''
-^-^-' written
elUptic integral of the
^"^^ ^^^^ ^"^ i denoted

is a given constant.
A;
SECOND KIND
Vl-A;2sin2^ de
also called Legendre's
0<k<i
it
defined as
is
is
defined by
0<k<l
(^)
the elliptic integral of the second kind.
called the complete
elUptic integral of the second kind

and
denottd^bf
^"*"^^^^
^"^^^
^^'^
^" *^ determination of the length
^^
o an ellipse
n
T''''
of arc of
and
supplies a reason for use of the term
elliptic integral.
is
The

-'o
also called Legendre's
THIRD KIND
do
C.
n{k,n,^)
of the
(l
+ nsin2^)v/l-A:^sin"^
form for the elliptic integral of the third

from zero since if % = 0, (3) reduces to
stant assumed different

If
<l>
= n/2
JACOBI'S
the integral
FORMS
for the
is
we
defined by
0<k<\
" <
<1
^?^
{3)
A:
kind.
Here n
is
a con
(1).
called the complete elliptic integral

of the third kind.
ELLIPTIC INTEGRALS
transformation
f^cvof above,
*K^
tegrals
is
v = sin6 is made in the Legendre forms

of the elliptic inobtain the following integrals with a; =
sin^.
dv
F,ik,x)
Ei{k,x)
C\[II^dv
331
[CHAP. 16
ELLIPTIC INTEGRALS
332
dv
ni(fc,n,a;)
/0
called Jacobi's
+ nv^W{-i--v')0kV)
ellvptic integrals of first,
forms for the
These are complete
(1
elliptic integrals
INTEGRALS REDUCIBLE
if
= l.
ELLIPTIC TYPE
to
and
x
If R{x,y) is a rational algebraic function of
nomials in X and
y,
second and third kinds respectively.
then
y,
i.e.
the quotient of two poly-
-^
,.,,
R{x,y)dx
yî
elementary functionsjalgebraic. trigonometric,

can be evaluated in terms of the usual
c,
if y = V^^+b or ?/ = Vox + & +
inverse trigonometric, exponential and logarithmic)
where a,b,c are given constants.

or
If y ^ V^[x^+bx^ + cx + d
= V^^+b^+^x^+dê where ""^f^'^J^^

mtegrals of first, second or third
of elliptic
given constants, (7) can be evaluated in terms
functions.
elementary
of
terms
in
cases
kinds or for special
polynomial of degree greater than four,
If y=. ,/P(^) where P{x) is a
(7)
may
be
functions.
integrated with the aid of hyper-elliptic
JACOBPS ELLIPTIC FUNCTIONS
,
^ =
. .v,
.
is
elliptic integral of the first kind
the
for
form
Jacobi
the
in
x
limit
The upper
it follows
am,
=
Since
=
<^
.
sin^
form by
related to the upper limits in the Legendre
= sin (am u). Thus we are led to define the elliptic functions
that
sn %
cnu
(^)
Vl-'f'sn^M
dnM
(10)
i!L!i
tn
iii)
sin(amtt)
^/l-x^
=
=
cos
^1 -
kV
(am
w)
as
to those of trigonometric functions
which have many important properties analogous
presented in the problems.
example, if
dependence
this
emphasize
To
= sn-l Note that u depends on k.
la;,
modfc.
write u^srv-^{x,k) or m = sn
elliptic functions; for
It is also possible to define inverse
= sn
then
sometimes
we
a;
LANDEN'S TRANSFORMATION
By
use of the transformation
tan
4,
'^"^^^
=
k
called Landen's transformation,
X
where
ki
|^
(see
or
we can show
i
^l-k'sm^<j>
Problem
sin
fc
<^
sin
(2.^,
<#.)
{1^)
cos 2^j
61).
that
+ kJo
^l-klsm^.i>,
This can be written
F{k,i.)
j^nK.'t>;)
(^^)
^"^^-
ELLIPTIC INTEGRALS
^^J
can prove that
hm
K=
From
this
X ^TT^n^
1.
^
.
where
333
< 1 and we
follows that
it
'^" ^' '^"'"P"*^^results after only a f^"^"^^

few applications of the
ÎZtlfT" ""f'
"
^-
^^ tan
(^j
+ -j
(,5)
I" P^^^^ it is possible"^ achieve

accurate
transformation
Solved Problems
ELLIPTIC INTEGRALS
1.
Prove that
By
0<k<l,
if

(1
-;)-'
';"l:;t:tL';'''""
'
""
""
(rl/2)(-3/2)
{-1/2X-X)
(-l/2)(-3/2)(-5/2)
,,
" "-"""
"" "'"
<'
"
,3
'"'''
b. ten.
1^1 +
using-
Problem
15,
Chapter
13.
7^1^
^ '^"'^^^ P^^""' ^^ ^^t expressing the integral as an
^'^
elliptic integral.
Let X
-Tr/2-y and
f"
^0
Let cos
1/
Then -sin ydy
cos^ u.
^y
2 c os
sin
udu
= ~2
f-dy__
^
V^^
= y/Tin
p^^
-ô
=
Substituting k
sin
2 cos
^^"COS***
Hence
cos
dy
Vcosy
m dw and
sin
uju__ ^
Vl-COS^M Vl +
du
VI +
\/2F(x/I,W2)
r^n
^0
COS'
or
COS^M
g^
V2-sin'M
2 cos
Vl +
u du
COS='M
^,
X 7f^
V2K{y/^).
the result of Problem 1 yields the

value 2.622 for the integral.
[CHAP. 16
ELLIPTIC INTEGRALS
gg^
Another method:
V2
Let
"
Vii^
Jo
sin yl2
V^^
Jo
Then V^/2 cos j//2
sin 0.
V2cos0g^
J"
Vcos^2//2-sin^^/2
-ô
dj/
cos
4-
d0.
V^ f^
^fl-2în^vl%
"!
I.
= '^''^'^^
^l
77/2
Evaluate
3.
Let cos X
/mz
v^cos^
f"
cos''
as in Problem
^'^'^^
Jo
^0
r^n
1
Jo
a;
2V2
2V2E(^/i)
7r/2
sin*
4 sin^ X
V4
Let
sin*
2 cos u du
Then
r'
4(1
C""'
i \
du
V2 -
2=
f
I
f" Vl + COS* u
""
sm* M
dM
'^'*
=====:
cos* .)
dx
V5-4cos*.d..
2 sin m
cos
in
sin
d0,
V5'Vl-tsin*,-d,
terms of incomplete
d0
cos
g cos u
_
-
dw
r*
J^
where
d0
cos
2y/\
^^^
sin*
g^
</,
COS*0d0_
C Vi-lsin*0
2 Jo
r*ziiii^:i^^^d0
-^/l
C"^
2 Jo
sin "'(2 sin ).
V5B(V475)
elliptic integrals
COS0
^
cos0.^-^===d0
Jo
_3
=
<p.
2/,
IVT^TTî^du
where
cos*" dM
- ^f^K(^)
p Vr+
V'l
Jo
da;.
f VT^TiiSHT du
Express
C"
du
Vi + cos* u
-'o
j;"V 5-4sin*^ d^
5.
Vl--i sin* udu -
integral equals
Letting
udu
V2
"^
cos'u)-l ^^
~^
2
COS* u
{l
f" cos^ u
^^'2
VI+n +
CTrli
The
2cosj^\
'=^"(vfT^^)
r^'^
/I +
Then
2.
f^"
elliptic integrals
terms of
in
rfx
sin*
tt0
Vl - i
+
sin*
-fF(i,0) + 2E(i,0)
r VT-isin*"!^ d0
-ô
S x ^ W6.
CHAP.
16]
ELLIPTIC INTEGRALS
^^==
2
COS.
(cos^./2
reducible to integrals of elliptic

type.
is
= 3-2
sin^./2)
cos^./2
a;/2
;r/2-M
in this last integral,
it
elhpt:c type.
6,
f"'
'^
from ./2
= ./2-.
t?^
\/2
v/2
-A ^f __*L__
"
becomes
seen that
it is
_2_] ^^../2
f"
./2 and ./4 respectively
f VTTÊ^
V2-sin'xda;
Find the length of arc of the
ellipse
a.
VCd^FTw
a sin
d.
The
10.
(a^-6^)/^
oVa^
j;*-^_|__
Let fcsin0
^<*
tanw.
_
_
./2.
2/
2PvrT^^d.
Jo
6 cos ^, a
2V2;(^/|)
> 6 > 0.
6'sin^,
rf,
4aJ^'^"V l-e-sin-^ d^
the square of the eccentricity

If the ellipse.
âE{e,Trlt) or 4aE{e).
terms of
in
Then k
cos ^
f" secM
d,/,
.,
elliptic integrals.
sec'
u du and
^-
sec^
f"
"Vfc-tan^,.
^"
-'o
V/bVos=M-sin^M
du
I
Now
^^^
proceed as in Problem
Upon
upper limit
retracing steps
it is
5.
Let
VF+l
sin.
A=
sin..
Then
VTlcosudu =
observed that the upper limit .

in the last integral
in the original integral
by
a;
^^'
"*"
nn-'(
Vvi +
^ ^'"
fe'sin^^/
"'
Then
is
result can be written as
Express
^,
4"VêosV
Va^-(a^--6^y^hiF^d^
e^
and
Vl-^sin^tt J
2v^j/'Vl-isinâ;rfa;
where
y=âmx, OSx^^.
f'VT+md^d.
when
/W2
9.
'^
sin^M
Vl-fsin^t*
....
"'"'^
^^^fll^
Vl - |
-i/s J,
-v/3-',r/2
a;
VaVo
Find the length of arc of the curve

Arc length
_ _2_
cos
V3^. Vl-|sin^M
8.
\/i-|cosâ;/2
V^
From Problem
since
and the given integral can be written
^J
V3-2cos=~^
Letting
335
is
, cos. d.
related to the
11.
[CHAP. 16
ELLIPTIC INTEGRALS
ggg
elliptic integrals.
Evaluate each of the following in terms of

dx
(a)
Letting x
2 sin
9,
*y
V9 -
^0
'^'^
(b)
Letting x
Vl - I
^ -ô
4 sin^ e
tan
sin= 8
e,
V(l + x^)(l + 2x^)

ds
V +
Jo
tan'' e
tan'9
a/1
l
Vr+Tto^
Vl + 2 tan^ e
Vcos^ 9
Vcos^9
J_ r"*
Letting
ff/2
this last integral
0,
1_ p'*
-'^M
V2
r'
<â;
V^^^ =
tan
9,
V2
cos* 9
In general
2)(it*
e^^
transformed into an
if some zeros are complex
12.
Evaluate
sec 9
p/
{F(Vl72, ^/3)
V2
P,{x)
is
F(Vl72,
d9
r'"'"
"^"^
VTÎ^ta^
jr/4)}
zeros,
a third degree polynomial with real
by the method
Problem 12).
integral
can be
Similar procedures are available
indicated.
sec 9 tan 9 d9
V(sec*9-l)(sec*9
Show how
,
Make
1
+
to evaluate
f
J
de
C^"
-ô
3)
13.
jfl
Let M
Jo
~
=
(see
+ 1)
^J^ V2^^î^
f -^^- where
integral becomes
Then the
becomes
this integral
sin'
F( Vl72, W4)}
^^
V(m'+
-3
.n
x^3 + uK
M or
^ Ji
{f (Vl72, W2)
V2
J^ V(x-l)(x-2)(x-3)'
Letting m
sin*
Let
2 COS* e
becomes
d,p
Vl - i
cos^ 8
de
Vl ~
-^|>
\/2
V2 -
2sin^9
2 sin^ 9
d9
p'*
2^ Vl i
= 0, 1,
>
-ô
respectively.
^T-
^ sin* 9
\f{Î2,
^12)
|k(V3/2)
sin* 9
^
^^
-^=========
x
^^
f"'
^JTTJ^ê
the fractional linear transformation
correspond to
in
terms of
elliptic integrals.
and choose a,h,o,d so that x
1,2,3
CHAP.
ELLIPTIC INTEGRALS
16]
This leads to
Using
Then
1=^,
this transformation the given integral
= u\ we
if t
In general
transformed into an
V{m^
where
Jdx
We
P4(a;)
Vf(t
,^
terms of
in
Let
V{y'
(2a
2)y
pu, where
is
/?
Choose p so that the
= 3.
where a
is
a positive constant.
10}{y^
(2a
l)y
coefficient of u' in
u = {|, du =
j^^.
we
let
impossible.
transformation
1}
- 2a + 10 =
a'
+a+7
is
equal to the constant term,
i.e.
fl*
'^
=9
du
l)(u'
+ U + 1)
dt
yj(t'
t
tan
l)(t=
+ 3)
as in Problem 11(6).
as an elliptic integral.
+ 10)(a;2_2a; + 7)
as in Problem 14, we are led to the condition a* 2a + 10 = a* 2a + 7

In this case however, completing the square in each quadratic suggests the
+a
l=y
dy
is
+ 9)(;S%^ + 3/3U + 9)
X
which
a*
Then
reducible to elliptic type by substituting
\/(a;'-2a;
is
Then
J V{u' +
Jdx
If
i.e.
a^
+ Sy + 9)
each quadratic
which
''^
I
is
method of Problem 13 is
Then the integral
dy
9){y'
p(
^~
which
can be
a constant.
Then
Let
applicable.
elliptic integrals.
zeros, so that the
+ a,
+ a'-2a +
V(/3'm'
yS
^^^
dy
or
is
a fourth degree polynomial having real zeros,
is
/ Viv' +
Let y
=^
- l)(t + 3)
Then
1.
Choose a so that the constant terms of each quadratic are equal,
or a
so that
by the method indicated. Similar procedures are available if

Problem 14; the method employed there can also be used in this
proceed as follows.
= d,c = d
and the method of Problem 12
1)(m^ + 3)
Here the polynomial under the radical has no real

mapphcable.
becomes
elliptic integral
some or all zeros are complex (see

problem).
;,
becomes
3d,
'^^
obtain
J ^/==
from which a
f|, 3=f
337
reducible to standard
form by
V(2/'
letting
+ 9){y' + 6)
\/6 tan
e.
u =
Letting
Jo
(3
du
Evaluate
16.
sece,
cos^)Vl
3cos'^
"
Show how
is
to evaluate
cos^)\Al
(3
Jo
VTTsT^?^
+ 3co?
^
+ cos^)Vr+^ôs^
(3
^-ô
^ -ô
(1 - 8 sin^ e
./l^Tf^ê
a/1
1F(V3/2,W2) - in(V3/2,-l/4,W2)
=
where the second integral
^'>
Jo
17.
[CHAP. 16
ELLIPTIC INTEGRALS
ooo
= M^
let t
iu'
C
J
^
_^=======
-
Vl - f
sin^
terms of
in
r
j
elliptic integrals.
{t
1) dt
---^^^^====
to obtain
+ i)du
r_J(3.^1+iHj
r
du
J V(M^-l)(^ +
Then proceed as
sin^ e)
third kind.
the complete elliptic integral of the

Using the same transformation as in Problem 13,
Now
r
in
Problems 12 and
Letting
Another method:
x-1 -
du
4 r
_L
+
r
3
3)
(3u^
_*L
+ l)V(^l)(' + 3)
16.
becomes
l/v, the integral
dy
J:V(l - 2/)(l - 2!/)(l - 32/)

is applicable.
for which the method of Problem 11(c)
ELLIPTIC FUNCTIONS
18.
Prove
By
(a)
^{.nu) ^ cnudnu,
definition,
if
m =
^{cnu) = -mudnu.
de
f*
)
-'o
(b)
,,
-i/l
fe
fe'
Vl -
snw = sin0 =
.,
SI
sin'
e
sin
(am
cnM =
m),
cos
(am
cos
m)
Thus
since
(a)
#(snK) = #(sin0) = cos^g = cnu^l -
i:(en)
^"' '^'
^=
dtt
^
-
= ^(eos.) = - sin
du
and so
^=
a/1
g
-
^^
k^ sin^
<P
= cnudn,
snuVT^H?^^,
sin^ = dnw.
= -sn.dn.
^"^^'
^^^
ELLIPTIC INTEGRALS
-^ (dn m) = - A;^ sn m en m,
du
Prove
19.
Prove
20.
(a)
339
Let
= - ^n .,
s (-.)
(a)
Then
Now
snM =
sin0,
sni;
e = -^ in
Thn sn
on (i,\
Thus
= - sn M.
(-U)
letting
M en M
en (-.)
(6)
= r
-A;^ sn
en .,
f"
sin (-0)
= dn u,
dn (-.)
(c)
= - sin
the second integral,
= - sn m.
it
becomes
Since
(0)
Since
cnu =
^T^-i^,
en (-M)
21.
Show
that
if
if
en (-.)
The
-r
^n M^c) dn (M
xv;
f"
N/r-A^sin^5
on the right
+ ^,
+ 2/i:) = dn M,
r'*'+'^
(^tf
J Vl
Jo
.
k^ szn^7
k'
sin^7
Vl -
X
-'^
Vl -
sn(u
+ 2K) =
cn(K
2X)
J*;
^'^
dn(K + 2X)
id)
tn (m
+ 2K)
k' sin^"7
or-
"'"
+ 2i^
in
sn(t.
cn(M
and the proof
is
complete.
+ 2ii:) =
2A^.
d^p
am(M + 2Jf) =
Vl -
+ .) = _,i^ = _,^
= eos(0 + .) = _ eos ^ = -en^.
= Vr-fe^snM.T2^ = Vl^^^îii^ =
sn it _ ^
= sn (^^ + 2g) _
cn(M + 2X-) ~ H^ - ^^n^*
A;2
sin=
+^
sin (^
Prove that snw and cnu have periods

eonal tn aj^
Replacing by .
tn (m
k^ sin^ e
'^g
(d)
vT^^^~Fî^'
the second integral on the

right becomes
X<t>+7r
22.
^^
"ô
and
dn
then
'^'^
'^
first integral
Letting
sin^
k^ sin2 ^
-'o
A;^
*" **
en u
tn?^.
Consider
en..
= VT^^î^^F^ = VT^l^î?^ =
d" (-)
ô
(a)
= Vnrî?(Z^ . ^T^T^;^ =
f'^L^^^
VI -
Vl -
'â
'^
Vrr^^r^,
dn. =
= ~
..
(6)
= - tn
tn (-.)
(.)
de
Probler21
,im
dn,.
j ^
^""^ ^^
+ 4Jf) = -sn(M + 2^) = -(-snw) = snu

+ 4/i:) = -cn( + 2^) = - (- en ,*) = cnu
""
^^^^ P""''^^ 2^"
From Problem
dn u and tn u have period 2K.
21(c) it is seen that
For
d
j- sn
(a)
(,
Jq
'o
k'
ypJT^k"' sin^
vT^^fc^
sometimes known as douhly-veriodic functions.
this reason elliptic functions are
Prove that
de
K'
23.
[CHAP. 16
ELLIPTIC INTEGRALS
340
fe)
dv
- 1;^)!! - Ic'V')
/(I
We
ia)
shall write
By Problem
18,
dx
if
snu, then
= cnwdn = Vi^:^ VT^fcV = ^f(^^^W^=l^)
A(snH)
du
du
Hence
(6)
modulus k being understood.

sn-H.,k), the dependence on the
in place of
sn- .
dx
dx
Integrating the result in
r
>'o.
(a)
V(l
from
to x,
dv
we
- a;=)(l - fcV)
=
sin
F{fe,*)
haye, since
sn-'a;
or
where ^ - am,
F(fc,
sin-
a;)
/(l-a;'')(l^^^fcV)
_
(corresponding to fc-0):
for trigonometric functions
Note the similarity with the result
r_Jv__ ^
^0
-^/l
The
gjj^-.^
trigonometric functions.
functions are generalizations of
elliptic
-u^
^_
24.
Prove that
F{y\l2,-rrl2.)
the properties of the
From
r^r(i)
2'V 2r(|)'
gamma
function,
de
sin~'" dg
Vsine
But from Problem
25.
2, this
Jo
V2F(Vl72.
integral equals
r(i)r(|)
\^r(i)
2r(J)
Tr(|)
./2)
at
and so the required result follows
of
Determine the period T of a simple pendulum
length
I.
attached
simple pendulum consists of a mass
mass and length I (see
to a rigid rod OP of negligible
suspended from a
Fig 16-1). Assuming the rod to be
equation of motion of mass
fixed point O, the differential
is
given by
rf2|0
^ _dê
m?|| = -ms-sm* or
d'e _ dp _
de _
Letting
-^ - P' dt^ ' dt
.
dp
equation becomes
^2
= 4cos9 +
I
e.
g . ,
-ysmfl
dp dl
ds dt
-fsin
9.
__
dp
^ de'
the
Integrating,
Fig. 16-1
CHAP.
.
p
16]
ELLIPTIC INTEGRALS
makes an angle
iL?de/dt - r^t^"
when e =
then
e,
at
6 = >
= - (g/l) cos 9. We
d/dt
^^'^
^S/
pendulum goes from
TU\ de/dt IS negative;

r/4),
fl
to 9
sign.
Vcos
time t =
thus have
Vcos
yligll
cos
and
released
is
from
rest,
i.e.
(which corresponds to one quarter of a period,

peôa, or
Then
^^
- V-^ r
^ "^90
hence choose the
341
-jz
=
COS
d.
Jf'"
2sr
-.//.no
fl
n= a
Thus
4^1
^^
=
_
2 ""
Letting sin 9/2
sin
26.
0,
Vsin^ 9/2
9j2
sin
sn{u
COS
+ v) =
sin''
(1
2 sin^ 9o/2)
becomes
^**
Vl -
fc
'
sin^
A;=
sin
eJ2
the approximate period for small oscillations.
is
srmcnt;dnt + cn^sn^dnM
;
a constant.
a,
'''
2 sin^ #/2
9/2
4-\p p''
Let u + v
^
Vl -
Jo
2^/
ff
this integral
it,
T = 27rVW which
Prove that
4-\/ir
^
'
T
It
^'
Vcos
'^^
*"
r
"ô
k^ sn^
Then dv/du
sn^ v
-1.
Let us define
C7
snM,
y=
snt;.
It follows
that
Û
where dots denote
/.
= {l-U')a-k'U')
Differentiating and simplifying,
U = 2k'W -
(1)
(1)
by V,
we
(2)
V'
- V'){1 - k'V)
(1
find
{l
+ k'W
(2)
V =
2fe^y=
{l
+ k')V
by U, and subtracting, we have
2k'UV(U^ ~
V')'
(5)
can be verified that (see Problem 58)
^2y2
j/y
Dividing equations
(3)
and
(5)
jj2y2
C;y
=
(1
(1
- A:'Z7'V^)(y^ i7V+ UV
t/y- UV _ _
UV-UV
= ~{UV~UV)
_ fc2[;2y2)(y2 _
jjjj
^^^
7^)
(5)
we have
J7y
But
Then
and
UV - VV =
It
dV dv
differentiation with respect to u.
U'
Multiplying
dV
and
j/y
2fc^t7y(;7y+;7y)
1
- k'WV^
(6)
-2kÛV(UV + UV) = {1- k'U'V').
d(UV- UV) ^
UV-UV
d(l
- kÛ^V]
- kÛ^V
'
so th^t (6) becomes
[CHAP. 16
ELLIPTIC INTEGRALS
342
An
integration yields
c (a constant),
i.e.,
^û^^^
sn
M cm; dm; + en m sn
1 fe^ sn^ u sn^ V
dn w
It is also clear
a solution of the differential equation.

solutions must be related as follows:
is
sn
-a;
u+v = a
that
M en f dn -u + en w sm; dn m
1
sn^ u sn^ v
__
y^^
_|_
is
a solution.
The two
^^
A:''
= 0, we see that
Then f{u + v) = sn (u 4 v)
Putting
/(m)
-y
sn u.
and the required result follows.
ELLIPTIC INTEGRALS
27.
(a)
Use the binomial theorem
show that
to
<
if lx|
1,
Vl- = l"2~'i2J4
(6)
If
|fel
<
1,
Ans.
29.
30.
31.
Show
{a) E(yf2l2,rrl2),
(a)
E(l,<p)
Evaluate
f"'
Vsin' X
(6)
Ans.
Show
y/l
^f(~, 0)
that
3 sin^
(d)
in
F(l,4>)
ln(sec.;6
+ yVA =
terms of
in
terms of
^'
sm 9
f"' VI +
ô
(c) '(0.5),
(d)
sin"'
1.
tan 0)
[Hint:
sin
f^
Let
elliptic integrals.
elliptic integrals.
(v^
kV3/2).
2.1565
where
sin^ e
2 cos^ a
^'^
Express
-'0
33.
sin 0,
-y
VI
,r/2),
1.4675,
(c)
Find the perimeter of the ellipse xVQ

Ans. 15.865
metric equations.]
-'0
32.
F(V2/2,
(6)
(6) 1.8541,
(a) 1.3506,
that
V2-4-6y'8
{ -(!)- (K)"^(l^)"
=
Evaluate
&
prove that
J''7r/2
28.
\2-a)
t)
^2^
^C'^^""W|;^
(,r/4
3 sin 0,
j/
In tan
Ans.
.a/2).
2 cos ^
^K^-^J
W^^/
V^
be the para-
^^^-
^^^
ELLIPTIC INTEGRALS
343
f ~=J^=
Evaluate
34.
^""- X y^^jTj-
C -^^
--^{KVf.O-CVi,
36.
Show
37.
Express each of the following
that
^k(
^ns
= J_x/^J_\
terms of
in
elliptic integrals.
()r^=^=
r\/3^ dx
re)
x'
'^^
(6)
Ans. (a)l^(|), (6)iiir(|),

Evaluate
38.
An.
40.
(a)
Show
39.
Show
(6)
X^(^/2,
46.
Evaluate
(a)
Prove that
^___^^^^____
Prove that
(a)
.MS.
47.
48.
(a) sn^,.
Prove that dn^ u
Find
Find
^ n(V2/2, -2/3,
1
L^V3
0.
k^
en
cn^,
cnû
u),
(6)
(a)
(cn^M-sn^M)dnM,
(a)
-(tnu),
Verify the results
en
=
=
1,
1.
(5)
-^
Show
that
f^
50.
Prove that
VI
-A;
/^J
n^
(a)
dn^w
W
,.,
dn
u =
0,
IS
(a)
sn^xdx
sn(u
(e).mO =
0.
1.
- vT^T^.
A'
tn
(d)
1.
k' sn^
rf^
f"
en
2^
snwdnw
\r+lJi^
^;r;r~-
(dn )'
(6)
sech-
(6)
(.)
k- where
= l ' en 2m
l + dn2M'
(a) sn^zt
- (sn m
(6)
- 3k^ dn'u snu cnu

(fe
j en u du
sn
Ans.
w).
(a)
= icos-Mdn^.) +
^,
c,
(5)
(5)
={u~
+ K)=^,
dn'
=
f-^L
Sli
ty
49.
./2)
Ipf^
ELLIPTIC FUNCTIONS
Show that (a)snO =
45.
./2)
^^
r_^___ ^
z\
3
Vx* + a;^ + i
V 3 'sjx' + ^^ + i = (^' + x +
l)(x'-x + l).]
42.
44.
rfa;
J^^(|)
^'^
that
[Hint:
43.
(d)
a;
r^
F(V2/2, ten- ^2/2).
that
Vs^cVi),
dx
(a)
(c)
41.
r"
it
In
en M
+ dnM
c.
E(k, amzt)}.
(6)
^"^
the complementary modulus.
cn(u
="("
m-
^' ^""^
k'
+
,,.
+
-^)--înr. (c) dn(M + i^)=.^
,
where
A:'
"
[CHAP. 16
ELLIPTIC INTEGRALS
344
51.
^
Prove
52.
Prove that
r_
,^ ,
that
dv
t^-^^^,k)
F(fc,tan-'x)
where
fc'
= ^/T^-
J_^ ^^___,^^^-^-^^^
k being understood.
briefly as tn-x, the modulus
tn-Mx,fc)
write
""
=
-
n en
2*'''
'
^^3"
Vx^
-1 Vs^
+ 1' 2
Prove that
53.
x^
sin-
dx
- 2^2E(^.f^ + ^2 f(^. f)-
Ans. 1.686 seconds
55.
Show that
any time
at
Problem 25
the angle e for the pendulum of
sin 9/2
where
56.
t is
r-r^^=e
Vsins + cose
-^
sin
cos
= V2
57.
Obtain the expansion
58.
Verify equation
59.
Prove that
60.
(a)
cn(M
i;)
(6)
dn(M +
v)
Evaluate
Ans.
sin
eJ2 sn
(a)
(4)
62.
Prove that
if
en M e n
k.
(6)
drû
\v^
+ tt/A).]
sn m sn dn
fe' sn' u sn^ v
tt
dn
dnv - k^'snusnvcnucnv
1 - fe' sn'' u sn^
./3), (6) F(0.5, ./4)
by using Landen's transformation.
0.8044
{iS),
i;
F(V2/2,
Verify the result
^^^(^"="' ^)
V"/
,.,',
61.
is vertical.
of Problem 26.
_
-
(a) 1.1424,
sin (e
given by
{yfgllt)
pendulum rod
measured from the instant where the
Showthat
[Hint:
is
Page
332, using
^.
Landen's transformation.
then lim k.
1.
Hence verify
(15).
Page
333.
We
often
chapter 17
Functions of a
Complex Variable
FUNCTIONS
variable
z,
written
= /(.)
The
^rdaL^.,
^^^"'^'''^
"" 1'
o"
operations
^^ '^' '^^^^^'^
'^""^t
with complex numbers have
already been considered in

Chapter 1
A function is single-valued if for each vahiP of ^ +>,^^^
of w; otherwise it is multiple-valued
T"^^P"^ ""^^ value
ol manv-vaZd
(^,.)
^.(.,.).
Example:
where . and . are'rerifTncSns

of
w =
z'
(x
+ iiiV
-^2
^.2
llZ
"=
/an^^^^^^^
^^^^
o-

to
thrsr.:vsr t**i '^r^z::i r^'r, "^'"' -^ -*->

S>0 such
|/(.)-Z|<. whenever T?f J <1'
given any
0,
there exists a
that
Similarly, /(.) is said to be

continuous at .0 if
such that
/(^)-/(ô)|<e
o if lim f{z) = f{z,).
whenever
>
given anv
b- J<
A
J?
Âlternatively,
^V
"'
'
I
^ ^''''*' ^ ^
tv,
.'.
/(z)
>
continuous at
is
Z-*Zq
DERIVATIVES
by
f ?/),1f <;:tal'?t'""'
'"
" -'^'"
then
/(.) is called
anaiyL
/n
'
*- " <*<'*
Tnot
^^ ^
We define
e'
6,+
4,
From
^+^+
these
...
î^,
we can show
as numerous other relations.
345
^^-al
that
^
e'
z^
n
,
z^
= e-. =
and
extension of the
"^^ '""^^^^"^
^tH TeXeTbTr T^
such complex series has already

been considered in Chapter
1:
^^^^^" ^'
?^
'T'"-'
'' -"^le-valued
L^^
necesSrty
correrp:nrr WtLr^fî^l^^^^^^^^^
/(.) exist, we can use as definitTon Thrsells
Example
/W, denoted
(1)
In orTer to be Lalv^^^^^^^^
The converse, however.
Of
AZ
As-^Ô
continuous.
"
"'
'^'^"^^^^^^^^ f
z^
^ (eos
2/
sin
2/),
as well
346
Example
^t folwhen a and b are complex numbers Since e^^ = 1,

Thus Inz
=
(pe**) = In p + <0 + 2M.
In
2
In
define
we
and
e"*""-'
lows that
manyThe various single-valued functions of which this
is a many-valued function.
branches.
called
its
are
valued function is composed
We
2:
define a^ as e-"" even

e'f
much
variable are
Rules for differentiating functions of a complex
those of real variables.
Thus ^{z^) = nz--\ ^{smz) = cosz,
CAUCHY-RIEMANN EQUATIONS
A necessary condition that w =
is
du
dx
Problem
by differentiating
{2)
real
_ _dV
(g)
dx
%, the equations are
u and
continuous,
v with respect to x and y exist and are
that
dx^
Thus the
dU
By
be analytic in %.
If the second derivatives of
find
u{x,y) + iv{x,y)
equations
dV
By'
etc.
continuous in
If the partial derivatives in (2) are
7).
sufficient conditions that f{z)
we
_
"
the same as for
be analytic in a region
f{z)
that u and v satisfy the Cauchy-Riemann
(see
[CHAP. 17
"
'
dx'^dy'
and imaginary parts satisfy Laplace's equation
tions satisfying Laplace's equation are called
harmomc
in
two dimensions.
Func-
functions.
INTEGRALS
If nz) is defined, single-valued
of f{z) along
some path
in a region
and continuous
Cm% from point
zi to
point
where
z;,
z,
%, we define the integral
x^
+ lyu
z^
- x^^ iy2,
as
complex variable can be made to

with this definition' the integral of a function of a
An alternain Chapter 10.
depend on line integrals for real functions already considered
can also
variable,
as for functions of a real
tive definition based on the limit of a sum,
the one above.
be formulated and turns out to be equivalent to
An imfor real integrals.
The rules for complex integration are similar to those
portant result
is
\(f(z)dz
where
is
an upper bound of
g
1/(^)1
f\f{z)\\dz\
on C,
i.e.
\f{z)\
M\ds
^ M, and L
=
is
ML
the length of the path C.
CAUCHY'S THEOREM
the region bounded by
be a simple closed curve. If f{z) is analytic within
that
as well as on C, then we have Cauchy's theorem
Let
j f{z)dz
f{z)dz
where the second integral emphasizes the fact that C
=
is
(5)
a simple closed curve.
CHAP.
FUNCTIONS OP A COMPLEX VARIABLE
17]
Expressed in another way,
Since
/(.)
2.
f{z)dz
has a
"^"''^ ^^^ ^^^^^ ^ corresponding results for
analytic everywhere,
is
equivalent to the statement that
(5) is
IiL^\nteiritvelpei1n"?ht^^^^^^^
Example:
347
for any simple closed curve
2zdz
i-
2zdz
we have
11
(1
+ iV -
i2iy
2i
'
^^y point interior
CAUCHY'S INTEGRAL FORMULAS

^"^^"^"^ '^''^'" ^""^ '"
"
''
to
c!\hel
/(a)
where C
'P^'
1
"^"^^^ ^"^^^
^ ^"d
fi(^d,
traversed in the positive (counterclockwise)

sense.
Also, the nth derivative of
=
f{z) at z
a is given by
is
These are called Cauchy's integral

formulas.
They are Quite rpmprV^hi. v...

^^^ closed
^^.^^/i^ C, and
fnf the
the^'^
'""^^T
various
derivatives at points within C can be
calculated
Thn if^
function of a complex variable has
a iirst derivative, it has all higher
deritativelas
""^"^'^^^'^^^
well
This of course is not necessarily true
^' ^^"
for functions of real variables.
s.-f r
'"""
n''
rrvr?thrst'i\Tow
'''
TAYLOR'S SERIES
Let f{z) be analytic inside and on a
circle having its center at
points . in the circle we have the
Taylor series representatrn of
z-n
/(.)
/(a)
+ f'ianz-a) +
T>,nr,
fr.
11
'"
Jiven by
^(.-a)^ + g^(.-a)3 +
..
(,)
See Problem 21.
SINGULAR POINTS
Example:
If /C^)
^^^,
then
is
an isolated singularity of
f{z).
POLES
^f
irntk
^(^)
= (^:r^'
<^()
^ 0.
where
:x; o:rrsv-iiL
doz.6?. pole,
etc.
'
poTe
</,(^)
is
analytic everywhere in a region including
fitt'^
i,r^^^'
''^^'" '^""^
^ ^^"^^^^
'
^-^^^^^^ ^* ^='^ -^^^^

P^^' if
n=2
it is
called a
Example
1:
/(.)
(^îfe+l)
and a pole of order

,
Examples:
,
3^~1 =
î^TJ -
f(,\ /w
or double pole at
^^^ *" singularities: a pole of order 2
Zz-"^
(^
has a hranch point at
= 0.
(see
+ 2i)(z - 2i)
^ simple poles at z
of singularities besides poles.
Problem
37).
The function
Im^
However, due to the fact that
3.
= -1.
or simple pole at z
A function can have other types

at .
[CHAP. 17
348
is finite,
2i.
For example,
/(.)
= ÎM
we
call
= V^
/(.)
has a singularity
such a singularity a
removable singularity.
LAURENT'S SERIES
a
every other pomt mside and
has a pole of order n at . = a but is analytic at
all points mside and on C
with center at a, then {z-arf{z) is analytic at
on a circle
and has a Taylor series about z = a so that
If
mC
H.\
/()
^-"
(2_a)
4-
^-" + ^
(z-a)"-i
+
^
+ -^^^ +
z-a
...
ao
,,
+ ai(z-a) + a2{z-af +
...
{9)
ij âlj^d
a^{z-a) + a.iz-af +
a Laurent series for fiz). The part ao +
the
consisting of inverse powers of z-a is called
the analytic part, while the remainder
refer to the series
t^fî^-af as a Laurent series
This
is called
principal part.
More
generally,
we
A
principal part.
where the terms with fe<0 constitute the
center
having
a region bounded by two concentric circles
panded into such a Laurent series (see Problem 92).
function which
at .
=.
analytic
is
can always be ex-
^5^
various types of singularities of a function /(.)

''' ilêroS
of
series has a fin te number
For example, when the principal part of a Laurent
series
n
order
of
are all zero, then . = a is a pole
Ss'anda-^0 while a--. a--.,
or
s^ngular^ty
. = a is called an essentâl
principal part has infinitely many terms,
It is possible to define
uZ
sometimes a pole of
Example:
infinite order.
The function
e^"
+ | + af? +
'
'
"
^^'
^'^
RESIDUES
^'"^"^'^^ singularity at z
= 0.
^,
^^
^ writing the coeffican be obtained in the customary manner by

In further developments, the
to {z-aYfiz).
cients for the Taylor series corresponding
importance.
at the pole e = a, is of considerable
The
crfficient
It
.^.
coefficients in {9)
a-u
called the residue of /(.)
can be found from the formula
calculation of the residue

the order of the pole. For simple poles the
particular simplicity since it reduces to
where n
is
of
is
a_i
RESIDUE THEOREM
lim(z-a)/(z)
(^^)
is any
except for a pole of order n at z = a and if C
If fiz) is analytic in a region
Integrating
(9),
form
the
[9).
a, then f{z) has
containing z
simple closed curve in
using the fact that
^^^^-
1'^]
dz
<r
Jc(z-a)'^
Problem
(see
13), it
"""'' '
^Lm^^S afplfe!
More
grenerally,
Theorem.
fimte
we have
*'"' "'"
call
27rm-i
(_^^)
'"'""'"^ " ''"''"
>*
"' ^>
If f{z) is analytic within
number
and on the boundary C of a region

within ^, having residue's a_, 5_, c^^.
of poles a,l>,e,...
dz
/(z)
i.e.
(12)
the integral of /(.)
Cauchy
is
27ri(a-i
2.i times the
sum
6_i
c-i
of
-5?
excent at a
respectively!
(_^_4)
of the residues of /(.) at the poles enclosed
the residue theorem.
EVALUATION
^ "-
.)
theorem and mtegral formulas are special cases of
"
the following important
J>^
by C.
ifn^l
ifn =
follows that
^^f{z)<iz
the
JO
\27ri
349
this result
which we
DEFINITE INTEGRALS
''^""' '^^^""^ integrals can often be achieved by

using the residue
^y..J^^^^^^^'''''^l
theorem
together with a suitable function
/(.) and a suitable path or corvtour C, the choice
of which may require great ingenuity.
The following types are most common in practice.
1.
F{x) dx,
F{x)
Consider
G(sin
e,
cos
Let .
^l^
See Problems
e) de,
..
The given
dz/iz.
along a contour
X ^(^)
Here we consider
Type
4.
1.
is
a rational function of sin
'-^,
and cos 6.
dz
where C
is
cose^'-+f^ and
integral is equivalent to
{sfn mx} ^^'
consisting of the line along the x axis
29, 30.
^me =
Then
with center at the origin.
^'
semi-circle above the x axis having this line

as diameter,
^"^^
tZ!
ihen 7f
let i2-*oo.
^
an even function.
F{z) dz
"^^
p"
J-2ir
is
F{z)dz
ie-d0 or de
the unit circle
^(*)
a rational function.
F(2)e-dz
where C
is
the same contour as that in
See Problem 34.
Miscellaneous integrals involving particular

contours.
[CHAP. 17
VARIABLE
FUNCTIONS OF A COMPLEX
350
Solved Problems
FUNCTIONS, LIMITS, CONTINUITY
Determine the locus represented
1.
21-3
<n\W
Method
(a)
12-21
(b)
\z-2\
1:
with center at
\x
c^
^q
1".
3|
lz
+ iy-A^\^-^ + ^y
and radius
(2, 0)
,.,,,
z-31 + +
^^^
+ 4,
12
3.
+ Oi
and 2
If this
Squaring,
Method
(&)
1:
la;
-1, a straight
line,
line.
any pou^tôn
that the distances from
Method 2: The locus is such
perpendicular
Thus the locus is the
to
,
())
(2, o)
and (-4,
^^
l^^^nX/^^Sîr^^'
T'î^^m^:''
q
10- V(^ + 3) +1/
_
x^
2/^
simplifying again yields
25
16
0)
find
are equal.
^^ ^
J^^
^ _^
^^^W^V
^^^L^: l^^'J^:^^
(c)
we
+ t2/-'i|-l'^^2/-ri|
squaring
+ ^^ - bVI^TbFT/.
of
semi-mmor axes
with semi-major and
ellipse
an
'
lengths 5 and 4 respectively.
on
distances from any point
to (3,0)
it
and
axis has length 10.
Determine the region
2.
^""^
'''
(b)
1<
interior of a circle of radius
+ 2i| g
|Z
|.
by each
represented
in the . plane
1.
See Fig. IT-l(a) below.
2.
2i\ is
^v, +
so that
the distance from . to -2^.
at -2], i.e. (0,-2); and 1^

t^;' to
represents the region exterior
J+ 2^=-
\z
of the following.
^^\
1
1
+ 2^^"
I, 4- 9,-l
bu^SIr
ow
1
1
is
center
a circle of radius 1 with
-^'^^^.^
1^^^^^^
^ên
=
1^ + 2^1
2.
<
1.
2^|
See Fig. 17-1(6)
below.
(C)
7r/3
arg
^/2.
z-pe^.
Note that arg^=*, where
g^Fig.
Hg.
these Imes. See
mcludmg ^^^^JX^
=
W2,
=
and
^/3
by the lines ^
^^^^
.^^^^.^ region bounded
below,
17-l(c)
1
J^'^-"-^
Fig.17.1 (b)
<"^
Fig. 17-1 (0)
Express each function

(a)z\
(b)
_s
()
1/il-z),
(c)
the form
e^
u{x,y)
iv{x,y),
{d)lnz.
(i.)' =
+ ,,)3 = .3 + 3x^(..) + 3x(i.)^ +
i(3x'2/ - J/')
x' - Sicj/"
=
x'-3x2/S v{x,y) = Sx^y - y^
u(x,2/)
Then
.
=
(x
-I-
where u and ^ are
x'
+ S.xV-3x.-^r
real:
CHAP.
17]
w = -J_ =
1-z
(6)
T'^.-
(c)
Inz
(d)
Inp
Prove
We
i^
sin (x
(6)
cos (x
-f-
+ iy)
use the relations
sin (
isr.Zy)
(J3^^.
tan"'
Yi
sin x cosh
e-
sin
a;
cosh y
cos ^
Similarly, COS.
e^^
sinSy
tan-'y/a;
2/
+ i cos x sinh 1/
- z sin sinh y.
a;
sin
T^
e- =
z,
^1^,
'""""
2/
cosZy,
(in this case it is infinitely

many-valued) since
""'^"^^"^ ^'^'"^ * * ''>^^"*^- ^ ^^"^
^h\t
.
called the principal branch of In z.
cos x cosh
+ %)
^""(cos ^
e^^
and
j//a;
+
+
cob (x
sin
a;)
+ iy) =
sin
z,
from which
e'-'
e-'- +
2i
e^cos
cos x sinh
= !!+^Z!
COS.
e--^""
cos ^
2i
=
=
and
+ y"),
ilnix^
=
=
sin.
vix,y)
1-x + iy
- xY + y^
(1
a multiple-valued function
is
iy)
1-^ + % _
1-x + iy "
V^?T?'+
In
'j;^^r
t-is
^ < 2^ and
(a)
Then sin.
'^^
vaL: lor
vame
foVwhTh
wnich
4.
l~x~iy
e^^(,o.Zy
u
Note that In^
= O-^^Ty^,
^3. ,3.
ln(pe'*)
-(-.-)
,3...,
} + ty)
l-{x
35^
e-"'-^">
a;
sin x)}
(sin
a;)
(j~^)
e(cosa;-isina;)}
i(cos x)
('^llHZ')
j/
e+"'
i{e-
(cos X)
(^^
= ^{e-Mcosx +
e-''^^+}
y ~
2^
^(s'n
')
(^'"2^
isinx)
'
) =
DERIVATIVES. CAUCHY-RIEMANN EQUATIONS

Prove that ^z, where 2 is the conjugate
5.
of z, does
By
definition.
which A.
= Ax + iA2/
=
lim ^
tyZl
Ay
0,
=.
j^^/(Mz^^
approaches zero.
lim ^
y-z
If
A/(,)
^^
^V
is
Ax
0,
the required limit
is
lim
lim
Ai(-o
th^t'^mZV^T^
isinx sinh 2/
not exist anywhere.
if tlî, ,jî, ^^.^^^
x-|-%-f Ax
ji
Ax-*o
If
independent of the manner in
Then
Ax - i Aj/ ^X + iAy
the required limit
cos X cosh y
(a;
- %)
tAy
Ax
^,^0 Ax
=
Ax
^^
=
lAy
iy
iAy
i
A!/
-1
approaches show that the limit depends on

the manner in which
i.e. z is non-analytic
anywhere.
that the derivative does not exist;
A.-O
so
m = j^,
U w=
(a)
6.
j^
Method
(a)
1 +
1 Hm
-^
1:
(1
Method
2:
The usual
^.
find
+ As)
+ Az)
(z
(z
(6)
+z
1-z
/(z
(1
- z - As){l - )
-^ 0.
Thus by the quotient rule for
z^l.
w =
f{x
+ Az)
the derivative does not exist;
= 1, where
is
,.
io
u{x,v)
dv
du
f(z)
equations
a region is that the Cauchy-Riemann
the region.
/(z)
_
~
rules of differentiation apply provided
Prove that a necessary condition for
Since
non-analytic.
independent of the manner in which Az
^ 1,
analytic everywhere except at z

at z 1.
non-analytic
function is
The function
is
Si
provided z
zY
Determine where
{h)
differentiation,
[CHAP. 17
352
g^
e^
+ iv{x,y) to
dv
du _
- "'d^
e^
iv{x
the
be analytic in
aatisfied in
^ satisnea
'
+ iy) = u(x,y) + iv(x,y), we have

= f[x + Ax + i(y + ^y)] = u{x + Ax.y + Ay) +
i.e.
+ Ax,y + Ay)
Then
f(g
+ i^)-fM
1^0If Aj/
= 0,
_
-
Ai
u(x
,.
the required limit
u{x,y)
^^
+ i{v{x + Ax .v + Ay)-v{x,v)}
+ i^y
is
u(x + Ax,y) -
,.
+ Ax.v + Al/) -
1^0
fv{x
u{x,y)
+ Ax,y) -
du
v(x,y)
If Aa;
= 0,
the required limit
is
u (x,y + Ay) -u(x,y)

iAy
io
,
we must have
.av
dx
dx
dv
du
- ^
and
+ Ay) -
v(x,y) \
13m
i
3t
^^
9u
dv
aj/
_^du
av
dy
dy
ia^/
^v
att
{ v(x.
two special limits must be equal,
If the derivative is to exist, these
so that
ai;
dx
dX
dy
_^
i.e.,
Bv
dy
du
g^-
derivatives of u and ^ with respect to * and 3/

Conversely, we can prove that if the first partial
equations provide sufficient conditions for /(z)
are continuous in a region, then the Cauchy-Riemann
to be analytic.
8.
(a) If
families of curves u{x,y)
by using
(a)
analytic in a region 91, prove that the one parameter

(6) Illustrate
C. and v{x,y) = C. are orthogonal families.
mûix,y) + iv(x,y)
f(z)
is
z^.
u{x,y)
Consider any two particular members of these families

at the point
Since
(aJo,j/o).
du
Also since
= u^dx +
dv
Uy
Vxdx
dy
0,
v,dy
we have
_
-
dy
^
_
-
Ux
Mj'
_Vx^
^^^
= uo,
v{x,y)
= vo
which intersect
CHAP.
17]
353
When evaluated at {xo,y<,), these

represent respectively the slopes of the
two curves at this point of intersection.
By
the Cauchy-Riemann equations,

= v^, we have the product
of the slopes at the point (xo, j/o) equal
to
Mj;
= Vy,
Uy
\ M/\ Vy) "
~^
so that any two members of the respective families are orthogonal, and
thus
the two families are orthogonal.

If
(6)
f{z)
The
a;2-j/2
Pig.
then
z",
graphs
of
Ci,
u = x^-y\v = 2xy.
members of
several
2xy
= C2
are shown in
17-2.
Fig. 17-2
9.
In aerodynamics and fluid mechanics,

the functions ^ and
f in f{z) = 4, +
where
IS analytic, are called the
velocity potential ^n^ltream functii
respecttely
x'' + ^x-y^ +
^
2y, (a) find ^ and (6) find /(z).
m=
(a)
By
U)
Method
1:
Method
where
2.
Integrating
(i),
r^,
2xy
(2),
2xy
''^'^
""'"
_ _
d^
a^-
dx~
Then
= 22/-2
Tx
+ Ay + F(x).
- 2x + G(y).
+ "'
""^'^
'''-''
-^^^^
"
- -y --1 --t-t-
(a),
Ci is
f(z)
=
-
= ^ + 4^ + ^W- Then
^
_ ~2xf+ c. Hence ^ = 2a;2/
^ + i^
(x'-y'
%-
(2),
2y
+ F-(x) =2y~2
Az
2iz
c,
result is then obtained
=^
+ iy,
= x-iy
so that
by substitution; the terms involving
z^dz
i+t
(c)
along the parabola x = t, y = t^ where 1

gig2
along the straight line joining 1-f-i and
2 + 4i,
along straight lines from l + ito2 + i and
then to 2
We
+ 4i.
have
dy)
(1,1)
''(2,4)
X(2,4)
j/=)
dx
'
2a;j/ dj/
J"
"
2a;j/ <a;
(sc^
= ^^
drop out.
INTEGRALS, CAUCHY'S THEOREM, CAUCHY'S INTEGRAL

FORMULAS
(6)
or
2a; -h c.
a pure imaginary constant.
The
2i
substituting in
-I-
= ^' + ^x - y^ + 2y + i(2xy + Ay ~2x + c)

+ 2ixy) + A(x + iy)-2i(x + iy) + ic = z^ +
This can also be accomplished by noting

that
(a)
Thus
2:
F'(x)
r
KX) i"
Z.
and i- (a)
~ -fand'îJi
From
Integrating
T=lxyV-t:M
(6)
- ^
^
dx
dy'
the Cauchy-Riemann equations,
- j,') dy
[CHAP. 17
354
Method 1:
The points
(a)
become
and
(1, 1)
(2, 4)
correspond to
=1
and
=2
Then the above
respectively.
,.
..
i
line integrals
C
t
(5)
{2(mt')dt
(1,1)
and
has the equation
(2,4)
2/
{t^-t'){2t)dt}
t
"T"^'
-t
- 1 = aZTJ
"
(^^
^^
1)
1/
3x
- 2.
Then
find
{[x^
(3a;
- 2)^]
*^j=i
From
+i
to 2
+i
+t
to 2
+ 4i
C
(|+
Adding,
Method
2)3 dx]
(x^-l)dx
dy
l,
+ (-30 -
9J)
= 2,
dx
-f
6i
-g
3i
and we have
(-i-y'Jdy
at
-^ -
and we have
2xdx
[or (2, 1) to (2, 4)],
-4ydy
3i)
[or (1, 1) to (2, 1)1,

I
2x(3x
{2x(3x-2)dx + [x^- (3x-2)^]3dx}
i\
r^
From
dx
r^
+
<c)
ij
2(t)(n2tdt}
Theline joining
we
{{t^-t')dt
30
9i
6i.
2:
integrals are independent of the path,

the methods of Chapter 10 it is seen that the line
integral
in (a), (fc) and (c) above. In such case the
obtained
values
same
for
the
accounting
thus
follows:
as
variables,
real
can be evaluated directly, as for
By
11. (a)
(b)
f{z)
dz
and on a simple closed curve C,
If /(z) is analytic inside
Prove Cauchy's theorem:
then
"
"
^ ''^
0.
^^
Under these conditions prove that
J^
V(z) dz
independent of the path joining
is
Pi and P2.
^/(^)d^
(a)
By
Green's theorem (Chapter 10),
.^-,.,
where
is
(b)
Then
jj(--it)-*.
^c-^'^"*
//(s-Sh*
the region (simply-connected) bounded by C.
Since /(.)
zero.
ûdx-rdj/ + i^^vdx + udy
^ {u + iv){dx + idy)
is
f = g- = "i^
analytic,
^ /(z) dz =
0,
assuming
/'()
^""^ ^^ *^"
(P-^^^l^"^ ^)'
^^^^ '"*'^"^'' ^'^
[and thus the partial derivatives] to be continuous.
Consider any two paths joining points Pi and P2 (see

Fig. 17-3). By Cauchy's theorem,
Pathl
r
r
Then
mdz
/(z)dz
f{z)dz
+
-
=
/(z)d2
/(z)dz
/(z)dz
Fig. 17-3
PjAPj
the integral along P.AP^ (path 1) = integral along

independent of the path joining Pi and P2.
i.e.
This explains the results of Problem 10, since
P.BP. (path
f(z)
z"
is
2),
and so the integral
analytic.
is
CHAP.
17]
355
12. If f(z) IS
analytic within and on the

boundary of
a region bounded by two
closed curves C^ and C,
(see Fig. 17-4), prove
that
As
dz
/(z)
X^
az
f(z)
in Pig. 17-4, construct

line
AB
cut)
a cross-
(called
connectmg any point on C. and

a point on
Cauchy's theorem (Problem
11),
f(z)dz
cr By
AQPABRSTBA
Fig. 17-4
Since /(.) is analytic within

the region shaded
md.
But
f(z)dz
An
^"
^(,)^^
Hence
(^)
U)
^^
gives
dz
Then
r^^^^^
BRSTB
= -jf(z)dz.
//(.)
also on the boundary.
J/(,)^,
AQPA
and
/(^)
dz
f(z)
dz
I.e.
^
need not be analytic lithin
curve C
Note that
13. (a)
f(z)
f ^L_
Jc{z-aY
Prove that
= j^î
if
if
curve bounding a region

having ^
(h)
What
(a)
Let C, be a circle of radius
is
=.
n =
n =
Jc
(^
- a)-
'
where
3 4
2,
if
having center at
that on
Fig. 17-5
e" e'
w-1,
pll-nHe
"Ode
î:
the integral equals
P'
rf
14.
Evaluate
(a)
\o)
ifn^l
.here C
is
is
not intprinv
"'^''"''^ tn
t Ul
lî
Since z
is
interior tn
mterior
to
Smce
""
2-j
For n = 0, -i, -2,

the integrand is 1, (.
-a){z~aY
Cu mcludmg z = a. Hence by
Cauchy's theorem the inte^l
.
\Zir
e-Ml^^^L
^0
(6)
a simple closed
y^ J^^r^
this last integral,

note
If
is
=a
Ihe integral equals

Jo
-1, -2, -3,
0,
and on the boundary of the

region bounded by
and C, we have by Problem
12,
To evaluate
f(z)dz
a as interior point
the value of the integral

e
/(2)rf2
J^
U
|. +
-i- i
^j
(a)
- 1.
_
^
_ 4,
1
the circle
,.,
!.
--ywhere
""
is
(,)
zero!'
the circle
,...,
'
al
the mtegral equals zero

(Problem
.,
11).
the integral equals %.i

(Problem 13).
inside
[CHAP. 17
356
closed curve C,
analytic inside and on a simple
is
15 If /(.)
figure of Problem 13,

Referring to Problem 12 and the
is
, V
integral becomes
Letting z-a^.eô, the last
continuous. Hence
and .
is
we have
f'" /(<>-+
Ha, +
J^
d6.
ee'*^).
ee
But
i)de
_
lim
any point within C.
since /(z) is analytic,
it
2)ri/{a.)
Ja
-1)
^
.
Evaluate
,
16.
(a)
Since z
^ô^^rfz
az,
/^
(a)
,.
-r
lies
Then
ir
... ^
withm C,
where C
^_-!l-.d2
J^ ^^^ ^ y)
(M
ôj
j)^ ^^r:;;
J_<^2ilcte
"*
2^:^
<C^^d. =
J^
-^
COS.
is
the circle
j^-lj-S.
by Problem 15 with
= -1
/(z)
cos
-2.i.
'c
by Problem
17.
IB, since .
2s-ie
and
Evaluate
^^^^^
Method
1:
By Cauchy's
n=
If
and
/(z)
13.
= -1
= 5Z-3Z + 2,
lOjri
2>rie-'
2,ri(l
-e
')
are both interior to C.
^^^^ ^
integral formula,
=
by Problem
^^
^"^ ^^'^ '^'"' '^""'
f"Ma)
then f'(l)
10.
2^ J^ (z - a)"-"'
Hence
'"'""'
'
''
z,
CHAP.
17]
357
SERIES and SINGULARITIES

For what values of
18.
^"^
z does each series
'^^'^'thterni
-?>;?2^-
converge?
= -^.
2"
Then
M^
lim
By
|i^!Lij
lim
the ratio test the series
|__L!^_^
converts
'verges
test fails.
\z\<9
n^ diverges
a< 2 and
if
it
(6)
series converges (absolutely)

for
>^' ^""^
2.
2
^'
(2m -1)!
3!
pl
We
5!
(-1)V"-^'
(2m + 1)
lim
Then the
series,
2^^^
3"
We
The
(2n-l)!
'
have
Mn+1
lim
lim
X2"
converges
\z\
if
Locatejn the
i\
<
3,
lim
all
values of
(z-^^)^'+'
Z
^
"""
2z'~z +
|2m(2m
'-*
and diverges
|a|
R-
,
'V
anz"
M.
sin Tnz
^
+ 2z
+ 2
,
since
circle
'""'^
i.-i,
..
z-i
3"
\z-i\
if
>
+ 1)|
- i)"
3.
^'"
''>'
"-
can write
.^
= ~^
1
~
show that
uniformly convergent
it is
.^^
is a
a ti,Îq
'S
pole
it
[^l
M.
converges,
it
follows
/J.
any. , each function and

a,e then..
^f
J
o
of order
3.
m v^ 0.
+
^^^verges since
but not on the boundary.
Since by hypothesis
.plane aU the sinûlanties,
^-mz~i)W^T+2i)
rr
z'
= 2,
and proofs for series of complex

numbers are analogous to those for
\az
M test that
thnf
finite
7T~r~7T7!)'
,^s
\z\
the ratio
and on the
is
a pole of order 2 (double

pole);
;^
and
poles of order 1 (simple poles).
'-">
=2
z.
we have
by the Weirstrass
(6)
absolutely convergent for \z\^R,
is
In this case
\z
series converges within

the circle
The definitions, theorems

real series
If
(-!)"-> z2-l
-tl^lrmllstJ'aTproIclêrras'MV!:!
for these values of
20.
2.
have
which represents sinz, converges

for
series converges if
Thus the
19- If
>
-.
^
oil points
at all
inside
i p
i-e-
{z
the
\z\
(~l)"'''g'""
lim
(c)
<
=
Izl
1^1
if
?, ^HFZ^
converges.
Thus the
k(
|-l.
5= in^2"
\z\
-j.
\z\
However, the series of absolute

values
.^
w''2"|
2.
+2
+ 22 + 2 =
. {. - (-1 + ,)h. - (-1
Since
^'
The function has the two simple
poles:
when
-,)}
z
= ~2
(,+
= ~1 + i
v/4
- 8 ^ -2 2i _
_ ,)(/^ , ^ ./
and z
= -1
-i.
2"
- 2i
are
We
-1
i,
[CHAP. 17
358
1
(d)
ljl-21i.
^
*
cos z
lim
However, since
appears to be a singularity.
0,
is
it
removable singularity.
Another method:
^=,
2!
6-'--"^
This
Then
is
1 is
an essential singularity.
finite
,.
sin.ularit.Howe= 0. We conclude that
leUin.
1^':^^^^'^^^^^:^
of
nature of a possible sn^gularity
in general, to determine the
u-0.
at
function
new
of the
and then examine the behavior
of radius
points inside and on a circle
If /(.) is analytic at all
that^
theorem
Taylor's
îs any point inside C, prove
see that z
terms.
infinite number of non-zero
the principal part has an
a Laurent series where
essential singularity at
has an
Jf
we
2U^^^~""
^"TT^^
This function has no
21
'
4!
a removable singularity.
is
(e)
U =,-+
z*
4z'
if
f
= |{l-(^l-2r+4T--6T+-V/
^^^
1
Since
/(z) at z
By Cauchy's
+ /.)
/(a)
/./'(a)
integral formula (Problem 15),
+ ^/"(a) + ^/'"(a) +
_ 1/.
.,
and
jC
f(z)
dz
(i)
substituting
"(F=^)[l-'i/(z-'^
(.) in (i)
formulas, we have
and using Cauchy's integral
on ^
C,
the length of C,
,
,
when
Now
n-
=0
IB J
z is
-. 0.
^
|j3^-rx| /(^)
Then R.
2î
and
"""
*"
...
^<f7t^r^
a)+Mz
J<, (z
\z-a\=
"
B,
?i)
so that
since
by U), Page 346, we have,
'
l;,|+iM
r,
be obtained by use of
.^ r____M^z___^
-a-
_
"
D
^"
m^
r/(^)d^
where
As
let z
division.
i-a-h
2fffl is
.e
we have
1
By
with center at
/(a
known Taylor
series.
.<
u_i
iS
r.
we can
generalize the Taylor series
CHAP.
17]
359
Find Laurent
22.
series about the indicated singularity for

each of the following functions
the singularity in each case and give the region
of convergence of each series.'
Name
e
(")
(2
_ 1)2
Let z
1-
(z-ir
-1 =
The
(6)
2 cos -
''^>
sinz
sin
z
7^
^''^
fz
Let
rfowftfe 'pole.
n-
Then
M.
+ ^)
sin (m
series converges for all values of
+m
is
0.
8z
z-0
._
is
and
^l/
_
~
it'
m=
31"^
5T~
7\
-1-
+ 2)
z.
Let
z+1 =M. Then
i+
a pole of order
1,
^^(l-M + M^-M' + M^-...)
1)
- 2m +
2
2(z
Using the binomial theorem,
16
16^
a poie of order
32'
1,
2m^
2m'
+ l) +
2(z
..
+ ir
or simple pole.
series converges for values of z such

that
The
,r
_ sin m
-7+1 +
i,
-- +
Case
1.
L.^..^
(z+l)(z
The
a removable singularity.
is
JT
+ l^rz + av
series converges for all values of z

0.
sinz
The
+ ~4!
3!
f!+fr+-
an essential singularity.
is
_
- T.
or
and
i?|i + + |7 +
^-l ^2! ^
- -i- + ^_.
^-'^
+u
= 0.
The
2,
series converges for all values of z
zcosi
z
a poZe of order
is
"^
lir
(z-ir
z
Then
M.
or simple pole.
<
|z|
<
2.
<
|z
+ 1| <
1.
[CHAP. 17
360
Case
2,
_
"
^- -^-^
~2m'
1
^
+ if
z{z
= -2
23. If
'
is
which
is
Let
+2 =
4m^
4(z
2(2+2)'
a poie of order
and the
+ 2r
where a-n
0,
(z
<
<
2|
_ a)" +
(z-a)"-!
prove that
(a)
By
we have on using Problem
integration,
f (i^'^^
27ria-i
Multiplication by
(z
- a)"
13
S'^
Taking the
- l)st
o-
call
f^{ao
a-i(z-a)"-'
derivative of both sides and letting z-*a,
lim
z-ta
f{z)
at the pole .
= a.
we
find
dr
^ ^ )" ^^^^^
;S^
dz"
result follows.
indicated
Determine the residues of each function at the
+ l)'
+ a.(.-a) + -(.-ar4-...)d.
a_. the residue of
+ a-nÂz-a) +
(w-l)!a-i
from which the required
we
gives the Taylor series
(z-arf(z)
(z-2)(z^
2-a
and on a simple closed curve C except at
27ria-i
(a)
+ 2)
32
2.
remains,
Since only the term involving a-,
24.
1^(^
RESIDUE THEOREM
/(2) dz =
(6)
3.
(a)
^m^'
16
+ 2)
8(z
analytic everywhere inside

a pole of order n so that
/(2)
-M-t<ty^(i)'<f)'-}
-2m'(1-m/2)
L_J^_^--i
16
8m
32
is
Then
u.
(u-2)v?
The
RESIDUES
-2.
These are simple
2,i,-i.
Residue at
is
lim
Residue at
=i
is
lim
Residue at
= -i
(z
Then:
4
2)
(z
poles.
poles.
(^
_ 2){z'' + 1)
5'
1
v"
- i)
(i
- 2)(2i)
r'
is
J.
(^
+ i)
i (z
- 2)(z - i){z + i)
(-i-2)(-2t)
- 2i
10
1
+ 2i
10
CHAP.
17]
^^^
z(z
'
+ 2)
'^
0,
-2.
Residue at 2
Residue at .
= -2
lim z
is
;_im^
is
= -2
a simple pole,
is
a pole of order
Then:
3.
is
361
^ ^^ |(, + ^f
^^j
_i
2A ^
8'
Note that these residues can also be obtained from the coefficients
of 1/z and l/{z
the respective Laurent series [see Problem 22(e)].
ze'*
('''
ig
_ 3)2
'
a pole of order 2 or double pole.
3,
Residueis
lim
^ |(, _
3).
_^j
Then:
^ limf(.0
=
cotz;
(d)
a pole of order
5s-,
Residueis
a,b,c,
lim (e-
Ste
(-1)(-1)
= (
lim cosi)
within and on a simple closed curve

prove that
f{z)
dz
{sum of residues of
2,71
reasoning
similar
that
to
by constructing cross cuts from

etc.), we have
(i.e.
f(z)dz
For pole
lim
except at a
number
f{z) at poles a, b, c, etc.}
f(z)dz
Problem 12
of
to Ci, Ca, Cs,
f(z)dz
...
a,
Fig. 17-6
J?/
^^
'
hence, as in Problem 23,
Similarly for pole
b.
O'
"
<h
f(z)
{z-a)"
f{z)dz
'
fit
(T^
ao
ai(z-a)
...
2rta-i.
^^ +
^
so that
...
/(^)rf^
^-^
=,
5,
6^(,_j)
...
27rib-i
Continuing in this manner, we see that
f(z)dz
-^-\-l)
Refer to Fig. 17-6.
<k
.<-)
interior to C,
J^
By
which can be shown applicable for functions of a comple

^
rule,
variable.
25. If f(z) is analytic
i^ Y
( Um
=
Where we have used L'Hospital's
e"
Then:
1.
(z-5.)-^ =
Urn
+ 2)
2rt(a-i+ 6-1+
...)
2rt (sum of residues)
of poles
362
26.
[CHAP. 17
""^" ^
r^^^TW
/c(2-l)0
Evaluate
Residue at simple pole
Residue at double pole
- -3
lim
1 is
(z
-I
^^^^" ''
^^
1)
(^T:1)(z
Since
Izl
'''
+ 3)'
''''
'''
^'^
'''
16
is
-5e-
(z-l)e'-e'
7
,.
+
^"'sdî^' ^^'lJîH^+sy
ia)
"^"^
lim
-^^
^--3
(z
r^2
1)
16
- 1,
z
3/2 encloses only the pole
TTte
the required integral
Since
(b)
\z\
10 encloses both poles z
and
the required integral
EVALUATION
27. If
constants,
are
where r
is
\16/
= -3,
2n-il
^^ J
^^g
DEFINITE INTEGRALS
of
1/(2)1
^'^'^
for 2
where fc>
Re^',
and
\\mymdz
prove that
the semi-circular arc of radius
22
=
shown
in Fig. 17-7.
By
j
the result U), Page 346,
f{z)dz
L =
that for z
Show
If
U/r
so that
M=
2,
_5^
dz
f{z)
Re'o,
|/(z)|
Re',
for example)
lim
ttB
\m\
=^'
^>'^
fe
+ R*e*'
= 4.
lim
and so
R_>
f^'''>
^^
CO
f(z)dz
|i?*e^'
inequality
Note that we have made use of the
1+^
1 ~ R*-l
\zi
+ z,\ g
*/p
<
|i
Fig. 17-7
R"-'
Then
ttB.
R-.00
ss
\f{z)\\dz\
since the length of arc
28-
we have
" R*
\z^\
if i? is
with
large enough (say
zi
B*e*'9
and
R>2,
zz
1.
dx
29.
Evaluate
^
^0
Consider
_B
to
-^
f -^
-W C
and theCtcircle
Since
z^
+l =
T,
when
is
consisting of the line from

the closed contour of Problem 27
sense.
traversed in the positive (counterclockwise)
z
these are simple poles of l/(.^

rule,
L'Hospital's
using
= e-% e--, e-- e--
C.
poles e-"* and e^"^ lie within
Then
D-
Only the
CHAP.
17]
Residue at
e'^"*
Uz -
lim
+
~
e'^*^*)
j-eiri/4
Z*
1^
Residue at
e=""*'*
-ij
lim
lim \(z^^,3itiny
lJ
ê'^""'-
e^^'*)
'z*
+ ij
Ig-^"*
-ij
lim
363
Thus
(1)
f" __^_
Taking the limit of both
*Tl
30.
Show that
The poles of
and
= 1 + 1
(^2
=i
S<Tl'
lim
is
= -1 + i
dz
X>f
r
31.
dx_
we have
required integral has the value
enclosed by the contour
- i)'
of Problem 27 are
of order 2
(z
+ i)\z - ly (z^ + 2z + 2) J
+ 1 - i)
+ DVz^ + 2z + 2^
+ ir(z+i-i)(z + i + i)
2^/^1^^+3-42!
^
,
+
^
"^
100
^''^^'^^
+ l)V + 2a; + 2)
100
5!
('
^'^^
(z^
= 9i-12
?!
(z
lim
is
J_^(a;='
^(a:^
we
28,
J (
2--l
Then
yyfi
and using the results of Problem
^'^^
dz
1.
--
Residue at z
lim
^ 1)2(^2 -(-22 + 2)
of order
Residue at z
sides of (2) as fi-*
,,,ff^
4.
~"^K~
_
-
^5~
7ff
Kn
z^dz
Xf
3-4i
_
-
7;
Jr(z^+lf(z^ + 2z +
50
2)
Taking^ the limit as fi-^. and noting that the second

integral approaches zero by Problem 27
obtam the required result.
Evaluate
Letz =
e'.
^^
3 sin
Then
sine
3sine
^1^ ^ -^
dz
X^r_i_q,'*
A-^-A
i6
where C
is
ie'o
do
izde
so that
2dz
.7:3z^ + lOZ^
.^-~,
the circle of unit radius with center at the origin, as

shown in Fig. 17-8 below.
[CHAP.
364
The
1 he
'^.
of
poles
Doles oi
^^^ *^ ^^^^^ P^
_ s3
g^2 4. 10 jz
17
+ 36
-10i V-lOO
=
Only
Fig. 17-8
1/3 lies inside C.

=
Residue at -i/3
Then
32'
,
Show that
If z
Jo
e<9,
Um
^
cos
de
+ -'
2
12"
.o
'=^3''
'
_
-
e"
e-""
(z
cos 3g
4 cos
Residue at z
is
Residue at
=|
is
+ z-')/2
d2
iz
pole of order 3 at z
+1
- l)(z - 2)
z(2z
^ i^
for ^
JJm
-j
(2
- i)
and a simple pole
^^
^.s^gg
the semi-circular
65
24"
i/2^,j21_65l
-2^{2^t)^
fc
>
and
J e'-^'Mdz
arHf
JL
as required.
12
24[
Re, where
within C.
+l
- l)(z - 2)
dz
=i
21
Hm
is
z+l
1 d'
io2!di^r *z(2z-l)(z-2)
where r
izAe.
z + l
1
" 2i X7'(22- l)(z - 2)
Then
|;(,),
d;,
the contour of Problem 31.
The integrand has a
33_ jf
_ zl+^
2
is
rule.
the required value.
^,
2i
where C
by L'Hospital's
4i
''
4 cose
'^
Then
V4V
cos 3e
5
- 3 / ~ -3 6 + lOi
3 j\__3z2 + lOiz
{^2
^ 2^/^
+ 10t2 - 3
2^^
32.
-3i, -i/3.
are constants,
prove that
the contour in Problem 27 and
constant.
Ifz =
B.',
Jê-/(2)d2
J^--''/(iê')îê*<'d9.
d9
Then
ja/O
'
Jo
g-mRsinfl |/(Be*e)|
iJdO
'0
M
"= ~
Jn
"
is
a positive
CHAP.
17]
Mow sm e a
Now
28/-n-
"^
than or equal to
for
^"'^
<
=
<
=
"
2M
/o
-/2
(see
ti
365
Problem
Chapter
77,
4).
Then the
last integral is less
f"
R'-' Jo
As
Show
34.
ie -
this
00
r"2i^
dx _
that
Consider
f^^dz
=i
lim
is
j)
(z
Then
2*
and
~ i){z + i)
=
dz
r"
,-
rfj.
are positive, and the required

result
=i H
lies
i, but only .
'^"^'^
I.e.
fe
within C.
2i7i
sin ma;
(~
\2i
r
p.m*
so
Taking the limit as
35.
we
Show
i?
and usin?
Pro>,lo,y,
obtain the required result
that
r^l^ dx
qq
^K-n
i.
* '''"^ ^'^^^ ^''^ ^t^^'-a'
around r approaches
^ ^
The method ot Problem 34 leads

us to consider the integral of e'-Zz
around the contour of
Problem 27. However, since . =
lies on this
path of mtegration and since
we cannot
we modify
grate through a singularity,
T^Z
''^;"''*!"
abde%gh1
Since 2
or
proved.
2i
dz
zero,
IS
i
the contour of Problem

27.
is
poles at z
<|(z
and
,-
where
The integrand has simple

Residue at z
approaches zero, smce
^he
path at
"'"' "^
is
outside
"'''
0,
inte-
that
as shown
^^"*-
^'
C, we have
~dz =
Fig. 17-9
dz
BOEFC
Replacing . by
-.
in the first integral
-da;
and combining with the third
-I-
^''*
or
2i
r^
dx
^ d^
dz
-f.
BDEFG
EFG
dz
BDEFG
integral,
we And
[CHAP. 17
366
first
zero.
on the right approaches
33, the second integral
Problem
Let .-^0 and R-^^. By
approaches
integral on the right
"'
de
The
^1
under the integral sign.

since the limit can be taken
Then we have
r" sin X
lim
2-i
or
VI
dx
the original triangle.

(a)
. ..o
equations. Then pomt A(2, 1)
as the transformation
B
pomts
Similarly,
Since = ^^ we have u = ^^^f" 2/
T, 7,^1 t>
figures below).
rJane
plane (see fig
into point A'(3, 4) of the
j ^^,^^^^^
in the x plane maps
-^P-îvdy. The hne
a,^,^' with equaand C map into points B' and
;^
A C B C A ii ol cu
segvents
parabolic
map respectively into
17-10(a) and (6).
tions as shown in Figures
.
-2x
"
.fl
ABC
^^
-^-
C'(T,24)
(2,1)
1/
(4,3)
Fig. 17-10
Fig. 17-10
(ft)
The
4(1
curve
slope of the tangent to the
u^
curve
slope of the tangent to the
The
Then the angle
i-^
at
2v
3-i
m,2 irii
4-
OTim.
(3, 4) is
at
^^
^^ -
is
1,
RT'
BC
is
,s
21
d^|<3_,,
*i|
^1
Û
= M =
2'
3.
^^|^^^^
by
given
and
dv\
w.
(3, 4) is
17/4
(3)(i)
^ 4 T" oT,H
i_
i.
and
between
that the angle

Therefore the ^^'^
we can show
A'B' and B'C is ./2.

responding ones of the given triangle.
function.
+ w.)
between the two curves^ at A'
tan
Similarly
(6)
(a)
AC
between
7!-/4. while the angle
./4
^^ ^^^ ^^^_
is
Jj^^/^^J^, T, a transformation where /(.)
^^^^''^f^t^.^/ing at . == .0 has the same magnitude
CHAP.
17]
VARIABLE
4 'Lli l7, f' f
tag position for the first

tiS
has returned for the second
tin,e
Let
-=1
.2
Then
e<9_
[corresponding to
w = -i/j = gie/a
A and i- in
"
367
"'"'
-t"-!
to
tart-
itrrarîn^^tlffofr^tt'ir "' "

t + a
Fi^ ITTll^d^)].
*'
*'
'^^''""^ P''^'"-
^hen
and
plane
w plane
Fig. 17-11 (a)

Fig. 17-11 (6)
must restrict 0. We can, for

'* ^ î'^^le-valued function,
example, choose
< 7< 2
we
"I
represents one branch of
the double-valued function ' - ,> ^'*"^*> "'^^"^ possibilities exist. This
are on the second branch,
" ^^ ^"^ continuing beyond this interval
e.g. 2. e
th.
t
we
is called a branchpoint.
^ê rotation is takfng
EqLalenlirwe ân fnsuTtLT.
plal
not to cross the line O.,
*'* ^^'^ "
^"^ "e single-valued by afreeing
called a hrlnoh Une.
<X
Show
S.
f "^'^^
^
that
Jo
Consider
_f^d.. Since . = is a branch point,

choose C as the contour of
Pig. 17.12 where AR r,H
^"^
actually coincident with
the
t
^""^ '^^"
separated
for visual purposes.
rn
xJi^tT^J
^
The integrand has the

Residue at z
= i =
pole z
g-^t
lying within C.
jg
Then
XT+I"^^
or,
= ~l
2e<>-'r.
omitting the integrand,
2jrie<!'-i)'H
^B
We
BDEFG
CH
HJA
thus have
wherewehavetouse. = .e- forth

in
2.
going around the circle
5Z>Sg
'
'^''
^^
"'"'^^
^*
"^
^^'
J^.
^"^^^^^
^''''^^ ^^^^
-ênt
of .
=
is
''^^^''"""
increased by
[CHAP. 17
VARIABLE
368
'""""""'
-,.
2,."
or
so that
1"^
Jo
FUNCTIONS. LIMITS, CONTINUITY
by
Describe the locus represented
case.
each
in
figure
a
Construct
39.
+ 2 - 3^| =
(a) |.
o o^
,
+ ^Y + {y-SY = 25, center (-2 3),
- 2^ + 1/' = 4, center (2,0), radius
Circle (x
= 1. where
Branch of hyperbola a.V9 1/V16
,
Ans.
(c)
Determine the region in the
Construct a figure in each case.
Ans.
41
(b)
Region in the first
(c)
Interior of ellipse
Ans.
(6)
(c)
(b)./(3
^ln((l
,,,.
(a) lim
+ x)^ +
.
_
-
Prove that
.3.
,.,
(6)
Show that
(c)
Generalize the results in
..
z''
.n.
(M
C)
,.
Zo,
42.
^^ (^
w -
(a)
T^
If
(b)
For what
Ans.
(a)
1^
- ^1 =
+ +
CO
0.^
>'
3.
3|
x+y
;^;^;"7;7
10.
1. + 31 <
^^
)_
^
the
9,
f^.^
/(2)
z
^
finite
l-l/2^
+
^
find
- 2/_^+
a;^
+ 6x +
2/^
and
values of
(ft)
z is
^^^^
^.^^ ^
^^
e'^-"' sin 2x3/
/(,)
/^^'
a;
"
+f
3x'^
+ "*-
(a)
are real.
-(^..). where . and .
= tan-rf^+2/..,
=
is
/c
continuous at .
."'y'oot o, ..= l^d>,.t .n,

1
a,
1,
0,l,2,...
P.0,. .h..
0.
(6) to
the equation z
1.
EQUATIONS
DERIVATIVES, CAUCHY-RIEMANN
44.
+ 51 "
1.
2.
""
(^. =
form .(...)
+ .), (c)e=%
=
x^-Zxt-^y, =
M
a;^ + 3x + V^
M = ^. + 6 + 1/^ + 9'
=
M = e^'-"' cos 2x1/, V
+ 2i.,
(a)
+ J/V16 -
V25
(,)
quadrant bounded by
in the
Express each function

(a).'
1^
2.
- 2) + ^ +
(x
circle
Boundary and exterior of
(a.)
(")
radius 5
-^ *
plane represented by
(5)W^3. Ogarg.Sl-
1.-2 + ^1^4,
()
(x
Circle
(a)
(6)
40.
.,,,,,_ gl.- II,

+ 2l-21. 1|,
(b)l.
5,
directly
from the
non-analytic?
definition.
...
directly
the
from the
. -e
definition.
l,-,-,--..-.
6-
'''''''* ''^
function.
Be1.nie ../...
(.)
Prove that
f(z)
Prove that
/(.)
= -1-
46.
47.
48.
z\^\
is
An.. (a).,^-.^-2,
49.
+ c.
,,iy,j, j
Prove that there
51.
Find
^^^ ^^^.^^ ^^^ .^^^^^.^^
f{z) "^""^^
whose ""^^^
real
'^
'
nart
P^^*
= 4.-3
/'(.)
iT^^l^f^t; ^^
'^^^^^'^
^^^
.-./
e
'(xcos^
;=
's
no analytic function whose

imaginary part
is
such that
/(.)
- e^^^^.SÎ
an analytic function is 2a.(l ) determine

(6)2i.-.^ + e, where . is real
Construct an analytic function

Ans. ze-'+l
50.
'"
not analytic anywhere.
i,
If the imaginary part

of
LTl^,".^':
oby
and
/(I
+ , = _3.
is
^..
^ the real part,

in^
.,
ysiny)
the function.
(6)
and for which
/(O)
.-2,.
,(.)
^.^
_ 3. + 3 - U
INTEGRALS. CAUCHY-S THEOREM.

CAUCHY'S INTEGRAL FORMULAS
Evaluate
52.
()
(6)
(22
+ 3) dz:
along the p"ath x ^ 2t

+ l, y = ^t^ _t-2
along the straight line
joining 1 - 2i and 3
T'
Evaluate
(J+^^TiJ
^^
the circle
(6)
56.
Evaluate
Ans.
I.
(a)
() -2..-
''''" 1
aL'' 17
Ans.
1 T^f,"'^'*
+ 19i
all cases
55.
+i
where
d.
-1=3..
*o 1
2^-
+i
is:
the circle
(,)
^^1+
(h\
(^
i^Jja
(a)
1)^
+i
An.()
Ans.
{a) -8ffi/3
8,7,
'^
to 3
a square with vertices at
|.|=V2:
v^.
I
(f 2!Zd-
and then
"'^" '
''
'-''
-1 -
,"
-i +
... -2s-i
\
(6)
^^'^ ^'-^
ô
'
(c)
+ ^'-'~-,
2;rt/3
-- -losing
Prove Cauchy's integral

formulas
57.
[Hint:
Use the
definition of derivative
and then apply mathematical

induction.]
SERIES and SINGULARITIES

58.
For what values of z does

each
(a)
2î^
"=>
Ans.
59.
(a) all
^
2,^
+i
(k\
'
(b) \z
f(

I.
.-I
<
i?
<
2.
<
(c)
2^ (-i)n
(^2
+ 22 + 2)^
= ~li
(c)z
^"
"
-f.
60.
series converge?
niz-i)"
(n
5
-
'S
1)
()
absolutely convergent,
(6)
|.|
i.
+ i)"
^^
(
'""^''^" "'^'*'^ -^thin any circle

of radius
such that
1.
1.
'
Locate in the
61.
[CHAP. 17
gr^Q
g~ô
Ans.
plane
finite z
the singularities,
all
y
(h\
(a) z
(c)
z^
if
cos-,
(d)
-i, pole of order 4
= 1,
simple pole;
(6)
(c)
Simple poles .
= -2,
sin
/x
e)
,,v
name them:
any, of each function and
C^)
= 0.
(e)
= ^/3,
(/)
double pole
= -1 i
z - irlZ)
,
W
'
cosz
/,!>
4- 4)2
essential singularity
removable singularity
2i, double poles
each of the following functions, naming the

Find Laurent series about the indicated singularity for
62
Indicate the region of convergence of each series.
singularity in each case.
^^^;
(a)
=^
,.
Ans.
(a)
and the
(c)
(z-ttY
16(z-l)
+
^
(,
Arm.
= 2;
=
2
0;
z
(a)
(b)
7/4,
-^
5!
4! z^
8/25,
65.
Evaluate
66.
If
- 9(il_l)
^
256
64
67.
If
less
^ (,+'1)^+8)
Use
<
double pole,
...,
|z-l|
7
'"
__^^
'
(x^
a"
'
""^'"^^
^'
^ ''^^^
i,
-,3Trt/2
Ans.
37r/2.
"^"''^'^
e'
f f(z) dz
0,
where
..
73.
r
J-
dx
(x^
"
+ 4f
^+
*'
J.
3a=
dx
""^
= JL
75.
32
= I
3
76.
= -5â-', a>0
77.
'^J
^^
(x^
+ aT
8V2
r
r
Jo
+ 4)
de
cos e
277-
v/3
4^7
+ cos ef
4 cos 9
^217
r"
l)''(x^
de
(2
sin^
Vf
"
3t
d9
(1
is
at least two
encloses all the poles of
DEFINITE INTEGRALS
_JL_
Ans. -S.i
"'"'^^ enclosing all the poles.
show that
2V2
a''
^2vr
<
{z'
74.
z#0
poles:
contour integration to verify each of the following
r" x^dx
^^J x* + l
72
all
essential singularity,
at the simple pole z
than the degree of Q(z), prove that
of
z^tt
degree of P(z)
where P(z) and Q(z) are polynomials such that the
P(z)/Q(z),
EVALUATION
all
+ iy
(c) z = 2; it^ e"
z = -t;
(d)z = f,0,
a simple closed curve enclosing
f{z)
simple pole,
..
z*
= -2; 1/4
z = -5;-8/25
Find the residue of e" tan
is
=1
RESIDUE THEOREM
64.
3)
6!
Determine the residues of each function at its

e"
,
,,
z-3
,
2z + 3
^"^^
(b) ^qrgiJ.
(a) -^fzrj'
() (j:^2?'
63.
i)f(,
(g-gf _
_^
4!
3)2^
4(z-ir
zÔ
L-+ -i^ -
+
T i
2j
,,^
("^
RESIDUES
.2
z-r
^F
--J-zr;
/fiN
(0)
zê-^'';
(6)
ef
2V2
f(z).
"^
''''''''
371
de
(ft
80
Jo
r"
81
a;''
6 cos 9)3
+4
"^
c^2^
ft>|6|
(îtfpp}?.
- -^-
83.
_lHl^_j^
^,-.
^0
82.
f"
Jo
l^}I}jrEax
(x^
i)^'*^
:r(2e-3)
__
= ^
-dx
'^'^''^
fsln^dx
=
85.
Jo
%
8
CO
J,
cosh
2 cosh (W2T
at (-iE 0)
(R
0^
^P
Consider
t^^"*^'
'
(I)
î,_ ^
W;,
-^^ *='''^
"^
rectangle with vertices
"^'''
^'
Riemanrrquatfonfire"
'"^""^'
^^'^'''^
" ^"'^
^ ^^^ Po^^r coordinates, show that the Cauchy-
d(u, v)
then at
90.
all
points where
f'(z)
Show that by the bilinear

transformation
transformed into circles of
the
91-
^-
-^ +
If /(^)
is
dx'
w = HiA
analytic inside and on

the circle
- p
U-|
\z
a\ - R,
'"^"'"
a.
+ h)
^^
J_r
<^'
^^^^ ^^-
fâ
two different ways.]
h)
= J^
-^ 0,
'' '
"
'"
''
^nd
/(^) is analytic in this

region^
/(.)d.
^r
_Mdz
^'
circles in the z plane

are
anA
closed curve in the

angular region surrounding
[Hint: Write
be
R"
'^^'*
iAec>ret that
f(a
C bemg any
<^
d,f
prove Cauchy's inequality,

namely,
I/'-' (a)!
prove Lawrewi's
"^ ^""^
''''^'^'
plane.
^"^^
C.
^^.
^^^^
,^
^^(^^
""' ^^^^"^
^^1^
93.
[CHAP. 17
372
Find a Laurent series expansion for the function

and diverges elsewhere.
-1 J. 2_
r
z
" J+1 + 2 + 2
I^Hint: Write
i)(, + 2)
(,
Ans.
Let
:j
r" e-"F(t)
dt
/(s)
than that of the denominator.
show that
pu)
J_
where
If
(see
Ans.
Chapter
(a)
is
12).
cost,
(6)
is
4+
-1
z{l
1/z)
+
^
(a)
converges for
^^''^'^
1<
l^l
-^^
+ z/2
'J
sum
_!
of degree
a given rational function with numerator
<
2)
a simple closed curve enclosing
(f e''/(z)dz
27ri t/^
this result to find F{t) if f(s) is

results in each case.
/(s) is
(,
2^4
z+
Use
[Note that
_
~
-;*- -^
/(.)
(&)
of residues of
32
+ 28 + 5'
^"^
all
^ ^
e" f{z)
8(s
the poles of
/(s),
less
we can
^ -^
^
at its poles
- If
'
^^^
(?TT?
^^
'^
the inverse Laplace transform of /()

the Laplace transform of F{t), and F{t) is
possible.]
are
/(s)
functions
other
to
Extensions
+ |te + |", (d) ^(sin t - t cos t)
le-sin2*, (c)
INDEX
Abel, integral test of, 284
Argand diagram, 6
summability, 259
theorems of, 230, 249
Absolute convergence,
Argument,
of integrals, 262, 265,

270, 271
of series, 226, 233, 239,
240
theorems on, 227, 239
Absolute maximum or minimum,
(see also
21
Maxima and minima)
Absolute value, 3
of complex numbers, 6
Acceleration, 63, 140
centripetal, 156
in cylindrical and
spherical coordinates, 159

of, 156
normal and tangential

components
Accumulation, point of, 5, 102
(see also
Limit points)
Addition, 1
associative law of,
2, 7
341, 342, 344
(see also Elliptic

functions)
6,
Binomial coefBcients, 18
series, 231, 232
theorem, 18
Bolzano- Weierstrass theorem
12
(see
Weierstrass-Bolzano theorem)
mean value theorem, 82
Boundary conditions, 300
Boundary point, 102
Boundary-value problems,
and Fourier integrals,
328
and Fourier series, 300,
313, 314
in heat conduction,
313, 314, 328
in vibration of
strings, 319
separation of variables
Bonnet
"^'"^' '''' 23' 239
65, 66
scale, 15, 19
system, 1
"' ^""' "''" Numerical methods)

by partial sums of
bvT.l7f
Fourier series, 310
by use of differentials,
115
'
method for
Bounded functions,
20, 21
sequences, 42, 47-49

sets, 6
Bounds, lower and upper,
using Newton's method,

79
using Taylor's theorem,
70, 85, 93
Arc
Bernoulli's inequality, 14
Bessel functions, 232,
250, 257 297
Bessel's differential
equation, 232 250
Binary
of elliptic integrals,
331
"tinuatio, of gamma
function, 286
Anal'îc functions, 345
Analytic
Analytic part, of a Laurent
series, 348
Anti-derivatives, 82
least square, 175

to irrational numbers,
110
relation,
Alternating series, 225,

226, 238 239
convergence test for, 225 226
"^
z,
to gamma functions,
287, 290
Bilinear transformation,
371 (see also
Fractional linear transformation)
eountability of, 12
Amplitud"e,r'"*''""^
y and
Bessel's inequality,
310, 320
of vectors, 134, 135,

143-145
Algebraic functions, 22
5,
X,
Beta functions, 273, 286,

287, 289-292
12
fundamental theorem of, 21

of complex numbers,
6, 12 13
Algebraic numbers,
Associative law, 2, 7
for vectors, 136, 143
Asymptotic series or expansions,
233 234
'
252, 253, 259
'
for gamma function,
286, 292
Axiomatic foundations,
of complex numbers, 6
of real numbers, 3
of vector analysis, 138
Axis, real, 2
Base, of logarithms, 3
Bernoulli numbers, 258
commutative law of, 2

formulas for elliptic' functions,
of complex numbers,
of vectors, 134, 144
Aerodynamics, 353
Aleph-null, 4
Algebra,
Arithmetic mean, 9
Box product, 137
5,
11
12
Branches of a function, 21
Branch line, 367
Branch point, 28, 348, 367
length, 94
element, 142, 143, 153
Area, 80, 93
expressed as a line integral
204
of an ellipse, 179
of a parallelogram,
137, 148
Briggsian system of logarithms,

3
'
373
solving.
INDEX
374
Cardinal number of the continuum, 4
Cardioid, 99
Catenary, 98
Cauchy principal value, 263, 272
Cauchy-Riemann equations, 346, 351-353
derivation of, 352
in polar form, 371
Cauchy's,
convergence criterion, 43, 50

form of remainder in Taylor's theorem,
61, 95, 231
generalized theorem of the mean, 61, 69
inequality, 371
198
Constraints, 164
Continued fraction, 52, 53,

convergence of, 52, 53
convergents of, 52, 53
55, 56
recurring, 52
Continuity, 20-40, 103, 104, 111, 112, 345,
350, 351
and differentiability, 57, 63, 64, 105, 113
an interval, 25
a region, 104
350, 351
of integrals, 89, 266

of vector functions, 139
right and left hand, 25
sectional or piecewise, 26
102
Closure law or property, 1
Cluster point, 5, 102 (see also Limit points)
Collection, 1 {see also Sets)
Commutative law, 2
theorems on, 25, 26

uniform, 26, 104
Continuous (see Continuity)
differentiability, 57, 105
Continuously differentiable functions, 57, 105
Continuum, cardinality of, 4

Contour integration, 349
for dot products, 136

for vectors, 135, 143
Convergence,
test,
for integrals, 261, 264, 268

for series, 225, 235, 236
Complementary modulus, 343

Completeness, of an orthonormal
Complex numbers, 6, 12, 13
set,
310
absolute value of, 6

amplitude of, 6
of, 6
as ordered pairs of real numbers, 6

as vectors, 18
axiomatic foundations of, 6
of, 6
equality of, 6
modulus of, 6
operations with,
6, 12,
13
6, 7,
Complex variable, 345 {see

a complex variable)
absolute (see Absolute convergence)

circle of, 232
conditional (see Conditional convergence)
criterion of Cauchy, 43, 50
domain
of,
228
interval of, 61
of continued fractions, 52, 53
of Fourier integrals
(see Fourier's integral theorem)
of Fourier series, 299, 311-313

of improper integrals
(see Improper integrals)
of infinite series (see Infinite series)
of series of constants, 234, 235
radius of, 229, 232
13
real and imaginary parts of, 6
roots of, 7, 13
Complex plane, 6
also Functions of
Components, of a vector, 136

Composite functions, 25
continuity of, 36
differentiation of, 59, 106, 116-119
Conditional convergence,
of integrals, 262, 265, 270, 271
of series, 226
Conductivity, thermal, 314
Conformal mapping or transformation, 366
Transformations)
Conjugate, complex, 6
(see also
field,
of an infinite series of functions, 228, 229, 246

of functions of a complex variable, 345,
set, 5, 11, 12,
polar form of,
Conservative
Constants, 4
in
region, 102
conjugate
102
simply-, 102, 197, 204
in
Class, 1 {see also Sets)

Closed interval, 4
argument
set,
definition of, 24, 25
integral formulas, 347, 353-356

theorem, 346, 347, 353-356
Centripetal acceleration, 156
Cesaro summability, 233, 252
Chain rules, 59, 106
for Jacobians, 108
Circle of convergence, 232
Comparison
Connected region, 197
region of, 109
uniform (see Uniform convergence)

Convergent (see also Convergence)
integrals, 260-265 (see also Improper integrals)
of continued fraction, 52, 53
sequences, 41, 227 (see also Sequences)
series, 43 (see also Infinite series)
Convolution theorem,
for Fourier transforms, 323
for Laplace transforms, 284
Coordinate curve, 141
Coordinates,
curvilinear, 109 (see also Curvilinear
coordinates)
cylindrical, 142, 153, 154
hyperbolic, 185
polar, 6
INDEX
375
Coordinates (cont.)
rectangular, 6, 101, 141
spherical, 143, 153, 154
Correspondence, 2, 10, 20,
41, 101, 141
one to one, 2, 10
Dense, everywhere, 2
Denumerable set (see Countable

Dependent variable, 20, 101
Derivatives, 57-79, 101-133 (see

also
Differentiation)
chain rules for, 59, 106
continuity and, 57, 63, 64,
Countability, 4, 10, 11
of algebraic numbers,
12
of rational numbers,
105, 113
10, 11
Countable
set, 4, 10,

directional, 163, 169
11
measure of a, 81, 87
Critical points, 63
Cross products, 137, 146-148
proof of distributive law
for 147
Curl, 140, 151, 152
in
curvilinear coordinates,
142
Curvature, radius of, 156,
159
Curve, coordinate, 141
simple closed, 102, 197,
204
space, 139
Curvilinear coordinates,
109
curl, divergence,
gradient and Laplacian in
m, i4^
i'
142
<* >
Jacobians and, 141, 142
multiple integrals in,
181, 182, 217, 218
orthogonal, 142
special, 142, 143
transformations and, 123,
124 141
vectors and, 141, 142
Cut (see Dedekind cut)

Cycloid, 99
Cylindrical coordinates,
142, 153
arc length element
divergence in, 154
gradient in, 154
Laplacian
in, 142,
multiple integrals
parabolic, 158
volume element
in, 142,
variable, 345, 351-353

ot infinite series of
functions, 229, 247
of special functions, 60, 66-68
of vector functions, 139,

150, 151
partial (see Partial
derivatives)
right and left hand,
57, 64, 65
rules for finding, 59, 66-68

table of, 60
Determinant,
for cross product, 137
for curl, 140
for scalar triple
product, 137
Jacobian (see Jacobian)
Dextral system, 136
Difference equations, 56,
285
continuous, 57
sectional or piecewise,
57
Differential equation,
Bessel's, 232, 250
Gauss', 232
solution of, by Laplace
189
in, 142,
163
Decimal representation
of real numbers,
Decimals, recurring, 1
Decreasing functions,
21, 26
monotonic, 21
transforms, 267, 280

Difterential geometry,
140, 159
Differentials, 58, 59, 65,
66, 105, 114-116
approximations by use
4, 15
Definite integrals, 80,
81, 86, 87 (see also
Integrals)
in, 83,
89-92
evaluating, 85, 92
properties of, 81, 87, 88
theorem for existence of,
81
with variable limits,
83, 163, 170, 266
Degree, of a polynomial
equation, 5
o homogeneous
functions, 106
Del (V). 140
formulas involving, 141

in curl, gradient
and divergence, 140
Deleted neighborhood,
5, 102
De Moivre's theorem,
7, 13
Denominator, 1
66
geometric interpretation of,

59
of functions of one
variable, 5
of functions of several
variables,
of vector functions,
139
total,
115
105
105
Differentiation (see also

Derivatives)
of Fourier series,
300, 311
rules for, 59, 66-68
under the integral sign,
163, 170, 266
Diffusivity, 314
mean value theorems for, 81

82 88
numerical methods for
of, 65,
exact, 106, 115, 116, 198
strictly, 21, 26
Decreasing sequences,
monotonic and strictly 42
^'
Dedekmd cuts,
change of variable
evaluation of, 63, 64

graphical interpretation of,
58
higher order, 60, 105
of elliptic functions, 338-340
of functions of a complex
Differentiability, 57, 105

and continuity, 57, 63, 64
154
153
154
in,
set)
93
Directed line segments,

134
Directional derivatives,
163, 169
Dirichlet conditions, 299
integrals, 287, 292, 293
Dirichlet's test,
for integrals, 266

for series, 228, 258
Discontinuities, 25, 104
removable, 34, 104

Distance between points,
145
Distributive law, 2
for cross products, 137
for dot products, 136
INDEX
376
Divergence, 140, 151
in curvilinear coordinates, 142
in cylindrical coordinates, 154
also
of improper integrals, 260-265 (see
Improper integrals)
of infinite series (see Infinite series)
Divergence theorem, 199, 200, 210-213
proof of, 210, 211
Divergent integrals, 260-265
sequences, 41 (see also Sequences)
series, 43 (see also Series)
Division, 1
by zero, 8
of complex numbers, 7, 12
Domain, of a function, 20, 101
of convergence, 228
Dot products, 136, 137, 145, 146

commutative law for, 136
distributive law for, 136
laws for, 136, 137
Double series, 233
Doubly-periodic functions, 340
Dummy variable, 83
Duplication formula for gamma function, 286,
293, 294
e,
proof of irrationality of, 70, 71

Electric field vector, 159
Electromagnetic theory, 159
Elementary transcendental functions, 22, 23
of a complex variable, 345, 346
Elements, of a set, 1
Ellipse, 99
area of, 179
length of arc of, 335
also
Elliptic functions, 332, 338-340 (see
Elliptic integrals)
addition formulas for, 341, 342, 344

derivatives of, 338-340
hyper-, 332
inverse, 332
Jacobi's, 332
periods of, 339, 340

Elliptic integrals, 331-344 (see also Elliptic
functions)
complementary modulus of, 343

complete and incomplete, 331
Jacobi's forms for, 331, 332
Landen's transformation for, 332, 333
Legendre's forms for, 331
modulus of, 331
of the first kind, 331-338
of the second kind, 331-338
of the third kind, 331, 332, 338
Empty
set, 1
Envelopes, 162, 163, 168, 169

Equality, of complex numbers, 6
of vectors, 134
Equations,
difference, 56, 285
differential (see Differential equation)
integral, 324, 328, 329
Equations
(cont.)
polynomial, 5, 21
Equipotential surfaces, 163
Errors, applications to, 164, 174
series,
in computing sums of alternating
226, 238, 239
mean square, 310
Essential singularity, 348, 358
Euler's, constant, 251, 286, 295
formulas or identities, 7, 251
theorem on homogeneous functions, 106
Even
functions, 299, 306-309

set, 2
Everywhere dense
Evolute, 176
Exact
differentials, 106, 115, 116, 198
(see also Differentials)
Expansion of functions,
in Fourier series (see Fourier series)
in power
Expansions
series,
231
(see Series)
Explicit functions, 107

Exponential function, 22
order, 283
Exponents, 3, 10
Factorial function (see Gamma functions)

Fibonacci sequence, 53, 55
Field, 2
conservative, 198
scalar, 138
vector, 138
Fluid mechanics, 353
Fourier coefficients, 298
expansion (see Fourier series)
Fourier integrals, 321-330 (see also Fourier
transforms)
convergence of (see Fourier's integral
theorem)
solution of boundary-value problems by, 328
Fourier series, 298-320
complex notation for, 300
convergence
of, 299,
differentiation
311-313
and integration
of, 300,
311
Dirichlet conditions for convergence of, 299

half range, 299, 300, 306-309
Parseval's identity for, 300, 309, 310
solution of boundary-value problems by, 300,
313, 314
Fourier's integral theorem, 321, 323-325
heuristic demonstration of, 326
proof of, 326-328

Fourier transforms, 322-325 (see also Fourier
integrals)
convolution theorem for, 323

inverse, 322
Parselval's identities for, 322, 323, 325
symmetric form for, 322
Fractional linear transformation, 336
(see also Bilinear transformation)
Fractions, 1
continued (see Continued fractions)
Frenet-Serret formulas, 159
Fresnel integrals, 294
INDEX
377
Frullani's integral, 282
Functions of a complex
variable (cont
Jacobians and, 371
Laplace transforms and, 372
Functional determinant
107, 120 (see also
Jacobians)
Functional notation, 20,
101
Functions, 20-40, 101, 107,
345
algebraic, 22
Bessel, 232, 250, 257, 297
beta (see Beta functions)
bounded, 20, 21
branches of, 21
composite (see Composite
contmuity of
limits of, 345, 350, 351

line integrals and, 346
multiple-valued, 345
poles of, 347, 348
real part of, 345, 352,
353
residue theorem for (see
Residue theorem)
series of, 347, 357-360
single-valued, 345
singular points of, 347
functions)
(see Continuity)
decreasing, 21, 26
Fundamental theorem,
derivatives of (see
Derivatives)
differential of (see
Differentials)
domain of, 20, 101

double-valued, 367
doubly-periodic, 340
of algebra, 21
of integral calculus,
82, 88, 89
Gamma
functions, 274, 285-297

analytic continuation of,
285
elementary transcendental,
22 23
asymptotic formulas for, 286
elliptic (see Elliptic

functions)'
even, 299, 306-309

explicit and implicit,
gamma
(see
Gamma
duplication formula for,

286, 293, 294
infinite product for, 286
recurrence formula for,
107
285, 287
formulas and asymptotic
series
for, 286, 292
table and graph of, 285
Stirling's
functions)
harmonic, 346
hyperbolic, 22, 23
hypergeometric, 232, 257

increasing, 21, 26
inverse (see Inverse
functions)
limits ot (see Limits
of functions)
maxima and minima
of (see
Maxima and
minima)
monotonic, 21
multiple-valued (see
Multiple-valued
function)
normalized, 301
odd, 299, 306-309
of a complex variable
(see
complex variable)
Functions of a
of a function (see
Composite function)
of several variables,
101. 110 in
orthogonal, 301, 314,
315
orthonormal, 301
periodic, 298
polynomial, 21
sequences and series
of, 227 228 9^9
^^'
'
242, 243
single-valued, 20,
101, 345
staircase or step, 28
transcendental, 22, 23
types of, 21, 22
value of, 20
vector (see Vector
functions)
Functions of a complex
variable, oto
345-372
ai^
analytic, 345
,
Cauchy-Riemann equations
and
(see Cauchy-Riemann
equations)
continuity of, 345,

350, 351
definition ot, 345
derivatives of, 345, 351-353
elementary, 345, 346
imaginary
part of, 345, 352, 353

integrals of, 346, 353-356
Gauss',
differential equation,
232
r function, 286
test, 227, 241
Generalized theorem of the
mean, 61, 69
treometric integral, 261
mean, 9
series, 51, 224
G.l.b (see Greatest
lower bound)
Gradient, 140, 151, 152

in curvilinear
coordinates, 142
cylindrical coordinates,
154
Graph, of a function of
one variable, 20
of a function of two
variables, 110 ^-i-i
111
.
,
Greater than, 2
Greatest limit (see Limit
superior)
Greatest lower bound, 5
of a function, 21
of a sequence, 42,
43, 49
Green's theorem in the
plane, 197, 202-205
space (see Divergence
theorem)
Grouping method, for exact
differentials, 116
Gyration, radius of,
189, 190
Half range Fourier sine

or cosine series
299, 300, 306-309
'
Harmonic functions, 346

series,
225
Heat conduction equation,

313 314
solution of, by Fourier
integrals, 328
solution of, by Fourier
series, 313, 314
Homogeneous functions, Euler's
theorem
Hyperbolic coordinates, 185
Hyperbolic functions, 22, 23
inverse, 23
Hyperboloid of one sheet.
111
Hyperelliptic functions, 332
on.
INDEX
378
series, 232,
Hypergeometric function or
Hypersphere, 101
Infinite series (cont.)
257
242, 243
of functions, 227, 228, 232,
partial sums of, 43, 224
quotient test for, 225
241
Raabe's test for, 226, 227,
241
240,
226,
for,
test
ratio
255
rearrangement of terms in, 227,
HypersTirface, 101
Hypocycloid, 207
area bounded by, 220

addition and
Identity, with respect to

multiplication, 2
Image or mapping, 108, 366
special, 224
uniform convergence
number 6
Imaginary part, of a complex
complex variable, 345,
of functions of a
M test for,
Weierstrass
352, 353
unit, 6
228, 245, 246
Infinity, 24, 42
Inflection, point of, 79

134
Initial point, of a vector,
Implicit functions, 107

and Jacobians, 119-123
260-284
Improper integrals, 85, 96, 99,
of,
convergence
conditional
absolute and
Integers, positive
Integrable, 81
262, 265, 270, 271

268
comparison test for, 261, 264,
containing a parameter, 265

definition of, 260
268-270
of the first kind, 260-262,
267, 272, 273
263-265,
260,
kind,
of the second
274
273,
266,
of the third kind, 260,
268
quotient test for, 262, 264,
266, 274, 275
uniform convergence of, 265,
274-279
266,
for,
test
Weierstrass
26
Increasing functions, 21,
21
monotonic,
and strictly, 42
Increaslng^st'quLces, monotonic
Indefinite integrals, 82 (^
^f"f,^JJ.
205-207, 215
198,
Independence of the path, 197,
101
Independent variable, 20,
71-74
Indeterminate forms, 33, 62,
L'Hospital's rules)
L^Hospital's rules for (see
72-74
Taylor's theorem and, 62,
14
Induction, mathematical, 7,
10
9,
2,
Inequalities,
Inequality,
Bernoulli's, 14
Bessel's, 310, 320
countably, 4
dimensional vectors, 301

interval, 4
258
224-25Maeeaâo
\
c
Ser.es)
226, 233, 23, ^4"

absolute convergence of,
235, 236
225,
comparison test for,
convergence
functions defined by, 232

Gauss' test for, 227, 241
236-238
integral test for, 225,
226
for,
test
Mth root
232
of complex terms,
theorem of,
Integral calculus, fundamental
82, 88, 89
329
Integral equations, 324, 328,
347 353-356
Cauchy,
of
formulas
Integral
321-344, 346,
260-297,
180-223,
Integrals, 80-100,
also
(see
360-368
347, 349, 353-356,
Integration)
also Definite integrals)
definite, 80, 81 (see
Dirichlet, 287, 292, 293
double, 180
integrals)
elliptic (see Elliptic
362-366
evaluation of, 267, 275-278, 349,
Fresnel, 294
Frullani's, 282
(see
improper
Improper integrals)
indefinite, 82
iterated, 180, 181

line {see Line integrals)
mean value theorems
for, 81, 82
multiple (see Multiple integrals)

variable, 346, 353-35b
of functions of a complex
246
series of functions, 229,
of infinite
84
of special functions, 83,
Schwarz's inequality for, 94
contour, 349
for gamma function, 286

for sin x, 315, 316
of, 2^6
conditional convergence
tests for, 225-227
series, 225, 236-238

Integral test for infinite
Integrand, 80
Integrating factor, 223
8b, ad, *
Integration, applications of,
(see also Integrals)
by parts, 84, 98
Infinite,
Infinite product, 233, 251,
and negative,
188-191
SanVJi^'fiions of, 83, 89-92, 181, 182,
266, 274, 275
265,
of,
convergence
uniform
Cauchy's, 371
Schwarz's, 9, 16, 94
inferior)
Inferior limit (see Limit
Wallis', 316
228 (see also
Infinitesimal, 65, 79
Imaginary
Infinite series, 43, 51,
of, 227,
Uniform convergence)
interchange of order of, 181

limits of, 80
311
of Fourier series, 300,
84
of special functions, 83,
95
partial fractions used in, 84, 91,
range
of,
special
80
methods
of, 84, 85,
89-92
171
under integral sign, 163, 164,
Intercepts, 110
Interior point, 102
26
Intermediate value theorem,
Intersection of sets, 11
INDEX
379
Intervals,
closed, 4
continuity
Laplacian operator
in,
(cont.)
in curvilinear
coordinates, 142
25
m
m
infinite, 4
nested, 43, 50
of convergence, 61
open, 4
Laurent's series, 348,

359, 360
theorem, 371
unbounded, 4
Invariance relations,
Least limit (see Limit

inferior)
Least square approximations,
175
Least upper bound, 5
159, 160
Invariant, scalar, 160

Inverse elliptic functions,
'""tinSrr
"^^'tVsfZtr
332
''' '-'- '-'-
'''' '''
'-'-
'-^ ^^-
'
Wlaee
elliptic integrals,
331
Leibnitz's formula for
nth derivative of
hyperbolic, 23
trig-onometric, 22
Inverse, of addition
definition
103
Jacobian determinant (see

Jacobians)
Jacobians, 107, 119-123,

141, 142, 158 154
Cham rules for, 108
curvilinear coordinates
and, 141 142
functions of a complex
variable 'and, 371
implicit functions and,
119-123
multiple integrals and,
181
of transformations,
108, 142
partial derivatives using,
107
theorems on, 108

vector interpretation
141
Jacobi's elliptic functions,
332
addrtion formulas for,
341, 342, 344
Jacobi's forms for
332
multipliers, 164, 172-174
operator)
Laplace transforms, 267,

279, 280, 284 372
convolution theorem for,
284
inverse, 280, 372
'"'^^,"372*""'*'"' ' ^ '^"'"P^^^ ^'^^^''^^
"^'e^qttirr*^^'
"'
''-' "''"
"^^'^-'^
128, 163
Hospital's rules, 62, 71-74
proofs of, 71, 72

Limit inferior, 43, 49
Limit points, 5, 11, 12
102
Weierstrass-Bolzano' theorem
on
(see Weierstrass-Bolzano
theorem)
''''' '''' 103' "1' "2,
V45,'35r3^l'
definition of, 23
iterated, 180, 181
of a complex variable,
345, 350, 351
proofs of theorems on,
31-33
right and left hand,
23
special, 24
41, 44, 45,
Lagrange's form of the

remainder, in Taylor
series, 61, 95, 231
Landen's transformation,
332 333
Laplace's equation, 113 (see
also Laplacian
267,'280''''''^ differential equations,
definition of, 41
Kronecker's symbol, 301
table of, 267
Lemniscate, 99
Length, of a vector, 134
Less than, 2
Level curves and surfaces,
theorems on, 24
Limits of integration,
80
Limits of sequences,
of,
elliptic integrals,
331,
rule for differentiating

under the integral
sign, 163, 170, 266
of, 1 (see also

Dedekind cut)
isolated singularity,
347
Iterated integrals,
180, 181
"''
product, 79
and multiplication
Irrational algebraic
functions 22
Irrationality of
V2, proof of, 8
Irrational numbers,
1, 2,' 8 9
approximations to, 8
Lagrange
of functions, 21
of sequences, 42,
43, 49
Left hand continuity,
25
derivatives, 57, 64, 65
limits, 23
Legendre's forms for
Inverse functions, 21
continuity of, 26
limits,
cylindrical coordinates,
142,' 154
spherical coordinates, 143
227
of functions, 227
theorems on, 41, 42, 45.47
Limits of vector
functions, 139
Limit superior, 43, 49
Linear dependence of
vectors, 160
Ivinear transformations,
131
fractional (see Fractional
linear
transformation)
Line integrals, 195-198,

200-202
evaluation
of,
196
independence of path
of, 197, 198,
205-207,
properties of, 196, 197

relation of, to functions
of a complex
variable, 346
vector notation for,
196
Line, normal (see
Normal
line)
tangent (see Tangent

line)
Logarithms, 3, 10, 351
as multiple-valued
functions, 351
base of 3
,
INDEX
380
Lower bound,
5, 11,
of functions, 21
of sequences, 42
Lower
Negative integers, 1
numbers, 1, 2
Neighborhoods, 5, 102
Nested intervals, 43, 50
Newton's method, 79
limit (see Limit inferior)
upper bound)
L.u.b. {see Least
Maclaurin series, 61, 231

Magnetic field vector, 159
Maenitude, of a vector, 134
function)
Mappings, 108
_.
..
(see also
Transformations)
conformal, 366
Mathematical induction,
Maxima and minima, 21,
7,
14
62, 63, 74, 75, 164,
171-174
j?
icyi
,
,
for, 164,
Lagrange's multiplier method
172-174
variable, 62, 63, 74, 75
of functions of one
variables, 164,
of functions of several
171-174
relative, 21
(see
Maxima and minima)
Maxwell's equations, 159
Mean, arithmetic, 9
Normal
line,
parametric equations for, 161
principal, 155, 159

152, 161, 165-167
to a surface, 140,
Normal plane, 162, 167, 168
wth root test, 226

Null set, 1
4
Number, cardinal,
Numbers, 1-19
numbers)
algebraic (see Algebraic
Bernoulli, 258
complex
(see
Complex numbers)
natural, 1
negative, 1, 2
operations with,
positive, 1, 2
2, 6-8, 12,
13
rational (see Rational numbers)
geometric, 9
error, 310
69
Mean, theorem or law of the, 61,
proof of, 68
Mean value theorems,
109, 124, 125

for derivatives, 61, 68-71,
94
for integrals, 81, 82, 88,
Measure
functions, 301
Normalized vectors and
numbers)
irrational (see Irrational
171, 1'^
Taylor's theorem and, 74, 75,
Mean square
vector, 135
absolute, 21
Maximum
i te
of acceleration, 156
Normal component
,
j
,
function (see Multiple-valued
M^-vaW
Napierian system of logarithms,

3
Natural base of logarithms,
1
Natural numbers,
12
real (see Real

roots of, 3
numbers)
transcendental,
5,
Numerator, 1
Numerical methods
12
.
.-
85, 92, 93
for evaluating definite integrals,
zero, 81, 87
Mechanics, 140
fluid, 353
Members, of a
Minimum
(see
Odd functions,
Open interval,
.
Modulus, complementary,
6
of a complex number,
343
331
of elliptic integrals,
210
Moebius
strip,
Moment
of inertia, 93
polar, 182, 186
Monotonic functions, 21
47-49
Monotonic sequences, 42,
42
fundamental theorem on,
180-194
Multiple integrals,
improper, 267
218
coordmates, 181, 182, ill,
in curvilinear
189
coordinates,
in cylindrical
189
spherical coordinates,
in
Jacobians and, 181

transformations of, 181, 182
21, 101, 34&
Multiple-valued functions, 20,
351
a,
as
logarithm
Multiplication, 1
associative law of, 7

involving vectors, 135
of complex numbers,
299, 306-309
region, 102, 110
set, 1
Maxima and mmima)
Operations,
IS
with complex numbers, 6, 12,
231
230,
series,
power
with
with real numbers, 2, 7, 8
6
Ordered pairs of real numbers,
138
numbers,
real
triplets of
Order, exponential, 283
of derivatives, 60
of poles, 347, 348
Orientable surface, 210
101
Origin, of a coordinate system,
(see
Orthogonal curvilinear coordinates
coordinates)
Curvilinear
Orthogonal families, 352, 353
functions, 301, 314, 315
Orthonormal functions, 301

Pappus' theorem, 194
Parabola, 27
158
Parabolic cylindrical coordmates
volume of, 137, 148, 149
Parallelepiped,
7,
12
Multiply-connected regions, 102
aUo Approximations)
(see
Parallelogram, area
law, 18, 134, 144
of, 137,
148
INDEX
381
Parametric equations, of
of normal line, 161
of space curve, 139
line,
165
Power
of, 230
Prime, relatively, 8
Principal branch,
of a function, 21
of a logarithm, 351
Principal normal, to a
space curve, 155, 159
Principal part, 59, 105
of a Laurent series,
348
Parseval's identity,
for Fourier integrals,
322, 323, 325
for Fourier series,
300, 309, 310
Partial derivatives, 101-133
applications of, 161-179
definition of, 104
evaluation of, 112-114

higher order, 105
notations for, 104
order of differentiation
using Jacobians, 107
of,
105
Partial fractions,
84, 91 95
Partial sums of infinite
;eries, 43, 224
Pendulum, period of a
simple, 340 341
Period, of a function,
298
of a simple pendulum,
340, 341
of elliptic functions,
339, 340
Piecewise continuous, 26
differentiable,
57
integrals, 261, 263
to
Point
a curve
Projection, of a vector,
196
Normal plane)
' '"'"' ^^'^ ^^"^^"* P'-)
(see
boundary, 102
branch, 28, 348, 367
cluster, 5, 102 [see
also
Limit i'""ii.b;
points)
63
interior, 102
limit (see Limit
points)
neighborhood ot, 5, 102
of accumulation,
5 (see also Limit
point,)
singular (see Singular
^
"^
points)
Point set,
one dimensional, 4
two dimensional, 101

Polar coordinates, 6
Polar form, of complex
numbers, 6, 7 13
Poles, 347, 348
defined from a Laurent
series, 348
ot infinite order,
348
residues at, 348
Polynomial equations,
5, 21
degree of, 5
Polynomial functions, 21
'
Position vector, 136

Positive definite
quadratic form, 179
Positive direction, 197
normal, 199
Positive integers, 1
1,
series, 61, 229,
248-250
Abel's theorem on, 230

expansion of functions in,
231
operations with, 230, 231
radius of convergence of,
229
special, 231,
232
theorems on, 230
series,
225
Quadratic equation, solutions

Quadratic form, 179
of,
I3
Quotient
test,
for integrals, 262,

264, 268
for series, 225,
235, 236
Raabe's test, 226, 227,

241
Radius of convergence,
229, 232
ot curvature,
156, 159
of gyration,
189, 190
of torsion, 159
Radius vector, 136

Range, of integration,
80
Rates of change, 63
Rational algebraic
functions, 22
-Kational numbers,
1, 8,
countability of, 10,

11
Ratio test, 226,

240, 241
proof of, 240
Real axis, 2
Real numbers
1 (see also
Numbers)
absolute value of, 3

axiomatic foundations
of 3
decimal representation
of 1
geometric representation
of 2
inequalities for (see
Inequality)
non-countability of,
operations with, 2,
7, 8
ordered pairs and
triplets of 6 1SR
^^
'
roots of, 3, 10
'
Potential, velocity, 353
Power
Quotient, 1
critical,
numbers,
Principal value,
of functions, 21,
22
of integrals (see
Cauchy principal value)
of inverse hyperbolic
functions, 23
of inverse trigonometric
functions, 22
of logarithms, 351
Product, 1
box, 137
cross or vector (see
Cross products)
dot or scalar (see Dot
products)
infinite (see Infinite
product)
Mth derivative of, 79
triple (see Triple
products)
Wallis', 316
Plane, complex, 6
Plane, equation of, I49
normal
series (cont.)
uniform convergence
Real part,
of a complex number,
6
* '
35? 353
"^ ^
'"'"''''''
^""^'^'' 345'
Rectangular component
vectors, 136
Rectangular coordinates,
6, 101,
141
INDEX
382
Seauences
rule for integration, 85
Recurring continued fraction, 52
asymptotic (see Asymptotic series)

binomial, 231, 232
double, 238
geometric, 51, 224
harmonic, 225
Laurent's, 348, 359, 360, 371
Maclaurin, 61, 231
357-dbU
functions of a complex variable,
of
P-,
power
sums of, 43, 224

Power series)
(see
reversion of, 231

special, 62
of, 43, 224
Taylor (see Taylor series)
telescoping, 234
terms of, 224
test for integrals, 262
sum
Riemann's theorem, 311, 326, 327

Right hand continuity, 25
derivatives, 57, 64, 65
Sets, 1
limits, 23
<.
coordinate system,
Right handed rectangular
bounded, 5
closed, 5, 11, 12
connected, 102
Countable set)
countable or denumerable (see
135, 136
Rolle's theorem, 61
proof of, 68
elements
Roots,
13
of complex numbers, 7,
of real numbers, 3, 10
Roots of equations, 21
computations of, 36
79
Newton's method for finding,
Saddle points, 172

Scalar, 134
field, 138
invariant, 160
of, 1
everywhere dense, 2
intersection of, 11
orthonormal, 301
point, 4, 101
union of, 11
204
Simple closed curves, 102, 197,
347
poles,
Simple
,. .q, ^m
197, ^04
Simply connected region, 102,
Simpson's rule, 85, 92, 93
function, 20, lOL 3^5
Single-valued
124, 260, dil,
Singular points or singularities,
357-360
34S
defined from Laurent series,
essential, 348, 358
.
product {see Dot products)

138
triple product, 137,
Scale factors, 141
Binary scale)
Scale of two (see
Schwarz's inequality,
for integrals, 94
16
for real numbers, 9,
Section (see Dedekind
Sectional continuity, 26
differentiability, 57
integration and, 81
boundary-value
Separation of variables, in
cut)
problems, 313
Sequence, Fibonacci, 53, 55
Sequences, 41-56, 227
47-49
bounded, monotonic, 42,
41, 227
divergent,
and
convergent
increasing, 42
225
partial
proof of, 361

Resultant of vectors, 134, 144
Reversion of series, 231
Riemann integrable, 81
infinite,
227
Series (see also Infinite

series)
alternating (see Alternating
singularity, 348, 358

Residues, 348, 360-362
360-362
Residue theorem, 348, 349,
349, 362-366
by,
integrals
of
evaluation
and
of,
series)
Region, 102
closed, 102
connected, 197
multiply-connected, 102
of convergence, 109
open, 102, 110
simply-connected, 102, 197, 204
Regular summability, 233, 258
Relativity, theory of, 160
Removable discontinuity, 34, 104
finite
terms of, 41
uniform convergence
decimal, 1
decreasing, 42
definition of, 41
(cont.)
Limits of sequences)
limits of, 41, 227 (see also
227
functions,
of
Rectangular neighborhood, 102
isolated, 347
removable, 348, 358
Sink, 219
Smooth function
(see Sectional differentiability)
Solenoidal vector fields, 219

Source, 219
Space curve, 139
Specific heat, 314
Spherical coordinates, 143,
arc length element

divergence in, 158
gradient in, 158
Laplacian
in,
104
153
15^5,
in, 143,
143, 158
multiple integrals
volume element
in,
189
in, 143, 153,
154
28
Staircase or step function,
Stirling's asymptotic
formula and
series, 286,
29^
INDEX
383
Stokes' theorem, 200,
213-217
proof of, 213, 214
Stream function, 353
Subset, 1
Transformations
Subtraction, 1
of complex numbers,
12
of vectors, 134
Sum,
Transforms
181, 182, 188-191
(see
Fourier transforms and

iâpiace transforms)
Transitivity, law of, 2
Trigonometric functions, 22
derivatives ot, 60
integrals of, 83, 84
inverse, 22
of series, 43, 224
of vectors, 134, 144

partial, 43, 224
Summability, 233, 252, 258
Abel, 259
Cesaro, 233, 252
Triple integrals,
181, 186-188
transformation
of, 182, 188-191

Triple products, scalar,
137 ' 133
regular, 233, 258

Superior limit (see Limit
superior)
superposition, principle
of 314
Surface, 101
vector, 137
Unbounded interval, 4
Uniform continuity, 26, 104
Uniform convergence, 227,
equipotential, 163
level, 128,
normal
(cont.)
Landen's, 332, 333

linear, 131
of integrals, 83, 89-92,
163
line to (see
Normal
orientable, 210
228, 243-245
of mtegrals, 265,
266, 274, 275
of power series, 230
of sequences, 227
of series, 227, 228
line)
tangent plane to (see

Tangent plane)
tJurface mtegrals,
198, I99, 207-210
tests for integrals,

266
tests for series, 228
Tangential component of
acceleration, 156
langent Ime, to a coordinate
curve 141
to a curve, 58, 162,
167, 168
Tangent plane,
theorems for integrals,

266
theorems for series,
228, 229, 246
161, 165-167
in curvilinear coordinates,
166, 167
Tangent vector, 139, 159
Taylor series in one variable,
61, 231 (see also

mean)
several variables, 109
of functions of a
complex
uniqueness
of,
125
approximations using, 70
for functions of one
variable, 61
for functions of several
variables, 109, 124 125
in approximate
integration, 85, 93
indeterminate forms and,
62, 72-74
proof of, 70, 125, 358
^''' "'^^ Differentials)

'
Slci''I!"''1'^''
irace, on a plane. 'f
111
Transcendental functions,
22. 23
numbers, 5, 12
Transformations, 108, 123, 124
and curvilinear coordinates,

123, 124 141
bilinear or fractional
linear, 336 ' 371
conformal, 366
Jacobians of, 108, 142
^ t-t)
infinite dimensional,
301
rectangular, 135, 136
of functions, 20, 21
of sequences, 42
257
remainder in, 61, 95, 231

Telescoping series, 234
Tensor analysis, 160
Term, of a sequence, 41
of a series, 224
Terminal point of a vector,
134
Thermal conductivity, 314
Thermodynamics, 132
Torsion, radius of, 159
247
^^^-^*--
Unit tangent vector,

139
Unit vectors, 135, 136,
301
Upper bound,
variable, 347

mean, 61, 109, 124
(see also Taylor series)
UnToHftts, u"^^*
""' ''-'
Upper
limit (see Limit

superior)
Variable,
change
Change
4,
20
of, in diflferentiation,
59,
of, in integration,
106
83, 89-92, 181
'^'""
variabS
'''" "'"' ^'^'^^^^'^"^
dependent and independent,

20 101
"a
dummy, 83
limits of integration,
83, 163, 170, 266
Vector algebra, 134,
135, 143-W5
Vector analysis (see

Vectors)
Vector field, 138
solenoidal, 219
Vector functions, 138
'"'^^ ^""^
''?5o!''m
derivatives of, 139,
Vector product (see Cross

products)
Vectors, 18, 134-160
algebra of, 134, 135,

143-145
axiomatic foundations
for, 138
complex numbers as, 18
components of, 136
curvilinear coordinates
and, 141
equality of, 134
infinite dimensional,
301
'
182
^^ '"-i^
I42
INDEX
384
(cont.)
Vectors
f 141
141
terms of,
Jacobians interpreted in
134
length or magnitude of,
normalized, 301
null, 135
position, 136
radius, 136
resultant or sum of, 134, 144
tangent, 139, 159
unit, 135, 136, 301
Vector triple product, 137

Velocity, 63, 139
of light, 159
potential, 353
319
Vibrating string, equation of,
Volume, 94
of, 142, 143, 153, 154
149
of parallelepiped, 137, 148,
element
Weierstrass-Bolzano theorem,
50, 102
proof of, 50
test,
Weierstrass
for integrals, 266, 274-279
for series, 228, 245, 246
Work, as a
line integral, 196
X axis, 101
intercept, 110
y axis, 101
intercept, 110
z axis,
101
intercept, 110
Zero, 1
division by, 8
measure, 81, 87
Wallis' product, 316
Wave equation, 319
5, 11, 12,
vector, 135
"7/^/
42, 43,

Calculo Avanzado Murray

Diunggah oleh

Informasi Dokumen

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Calculo Avanzado Murray

Diunggah oleh

Hak Cipta:

Format Tersedia

^Tf.

THEORY AXD PROBLEMS

Rensselaer Polytechnic Institute

^^Q V^* """

SCHAVM'S OUTLINE SERIES

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

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

T8910 SHSH 7B4S210

"Advanced Calculus" means differeijt things to different

essentially represents elementary calculus

with rigorous statements and proofs of theorems.

represents a variety of special

engineer and mathematician

proficient in his intended field.

review of the material of each chapter.

to stimulate further interest in the topics.

opportunity to thank the staff of the

their splendid cooperation in

Schaum Publishing Company

meeting the seemingly endless attempts

Real numbers. Decimal representation of real numbers. Geometric

FUNCTIONS, LIMITS AND CONTINUITY

Theorems on limits. Infinity. Special limits.

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

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

pretation of the derivative.

Fundamental theorem of integral calculus. Definite integrals with

Dependent and independent

derivatives using Jacobians.

Vector interpretation of Jacobians. Orthogonal curvilinear coordinates.

Special curvilinear coordinates.

APPLICATIONS OF PARTIAL DERIVATIVES

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

LINE INTEGRALS, SURFACE INTEGRALS AND

Vector notation for line integrals.

power series. Operations with power series. Expansion of functions

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

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

GAMMA AND BETA FUNCTIONS

Table of values and graph of the gamma function.

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

Equivalent forms of Fourier's integral theorem.

integral of the second kind. The

FUNCTIONS OF A COMPLEX VARIABLE

set or null set.

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

and negative integers and zero

Rational numbers or fractions

called the set of integers.

and irrational numbers

the set of real numbers.

peating cycle of digits,

numbers as points on a line called the real axis,

The geometric representation

is called the set of positive numbers; the

set to the left of

Between any two rational numbers

numbers an everywhere dense

If a, b, c belong to the set

rational (and irrational) numbers.

line there are

Commutative law of addition

the identity with

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