Sam Melkonian, MATH1005 - Differential Equations and Infinite Series

Ordinary Differential Equations
and
Infinite Series
Second Edition
Sam Melkonian
School of Mathematics and Statistics
Carleton University
Ottawa, Canada
NELSON
NELSON
COPYRIGHT@ 2016 by No part of this work covered ISBN-13: 978-0-17-671614-1

Sam Melkonian. by the copyright herein may ISBN-10: 0-17-671614-9
ALL RIGHTS RESERVED. be reproduced, transcribed,
Published by or used in any form or by any Consists of Original Works:
Nelson Education Ltd. means-graphic, electronic, Ordinary Differential Equations
or mechanical, including and Infinite Series
Printed and bound in Canada photocopying, recording, taping,
6 7 8 9 19 18 17 16 Sam Melkonian
Web distribution, or information
For more information contact storage and retrieval systems
without the written permission Cover Credit:
Nelson Education Ltd.,
1120 Birchmount Road, Toronto, of the publisher.
EtiAmmos/Shutterstock
Ontario, M1 K 5G4. Or you can visit For permission to use material
our Internet site at nelson.com from this text or product, submit
all requests online at
cengage.com/permissions.
Further questions about
permissions can be emailed to
[email protected]
Every effort has been made to
trace ownership of all copyrighted
material ·and to secure permission
from copyright holders. In the
event of any question arising as
to the use of any material, we will
be pleased to make the necessary
corrections in future printings.
This textbook is a Nelson custom
publication. Because your
instructor has chosen to produce a
custom publication, you pay only
for material that you will use in
your course.
111
Preface
The prerequisites for this book are elementary differential and integral calculus and
linear algebra. The exercises at the end of each section follow the progression of the
topics in that section, and are approximately in increasing order of difficulty. The
exercises at the end of certain chapters cover the material of the entire chapter, may
be more difficult, may involve material from earlier chapters, or may be of a more
theoretical nature. The solutions of all of the exercises are provided in Appendix C.
Proofs of theorems which are mere computations are furnished in order to provide
the reader with a better understanding of the subject. More abstract or otherwise
difficult proofs are omitted from the main text. The proofs of selected theorems are
given in Appendix B.
Appendix A contains formulas and techniques from trigonometry, calculus and

linear algebra which may prove useful to the reader.
The second edition is identical to the first, with the exception of minor changes
and corrections.
Sam Melkonian
April 2015
Contents
Preface 111
I Ordinary Differential Equations 1
1 Introduction 3
1.1 Basic Concepts 3
Exercises 1.1 .. 7
2 First-Order Equations 9
.
2 1 Separable Equations ...... 9
2.1.1 Orthogonal Trajectories 1 1
Exercises . 2 1 ....... 1 4
.2 2 Homogeneous Equations 1 5
Exercises 2.2 ....... 8 1
. 2 3 Linear Equations .... 81
.
2 . 3 1 Bernoulli Equations . 2 1
Exercises . 2 3 ........ 2 3
2.4 Functions of Two Variables 2 4
2.4.1 Partial Derivatives 2 4
2.4.2 T he Chain Rule . 26
Exercises 2.4 ........ 27
. 2 5 Exact Equations ..... 82
2.5.1 Integrating Factors 3 3
Exercises 2.5 .. 36
Chapter 2 Exercises ..... 38
3 Second-Order Equations 43
.
3 1 Basic Definitions ... 4 3
Exercises 3.1 ........... 64
.3 2 Linear Homogeneous Equations 7 4
3.2.1 Equations with Constant Coefficients 8 4
.
3 .
2 2 Cauchy-Euler Equations ...... 5 2
V
Vl CONTENTS
Exercises 3.2 . . . . . . . . . . . . . . . . . . . . . 55
3.3 Linear Nonhomogeneous Equations . . . . . . . . 57
3.3.1 The Method of Undetermined Coefficients 57
3.3.2 Variation of Parameters . . . . . . . . 67
Exercises 3.3 . . . . . . . . . . . . . . . . . . . 74
3.4 Equations Reducible to First-Order Equations 76
Exercises 3.4 . . 83
Chapter 3 Exercises . . . . . . . . . . 84
4 Higher-Order Linear Equations 87

4.1 Homogeneous Equations . . . . . . . . . . . 88
4.1.1 Equations with Constant Coefficients 88
4.1.2 Cauchy-Euler Equations 98
Exercises 4.1 . . . . . . . . . . . 101
4.2 Nonhomogencous Equations . . 103
4.2.1 Variation of Parameters 103
4.2.2 Undetermined Coefficients Annihilators 111
Chapter 4 Exercises . 119
5 Linear Systems 121

5.1 Homogeneous Systems . . . . . . . . . . . 123
5.1.1 General Theory . . . . . . . . . . . 123
5.1.2 Systems with Constant Coefficients 126
Complex Eigenvalues . . . 135
Generalized Eigenvectors . 146
Exercises 5.1 . . . . . . . . 168
5.2 Nonhomogeneous Systems 171
Chapter 5 Exercises . . . . . . . 186
II Infinite Series 189

6 Sequences and Series 191
6.1 Sequences . . 191
Exercises 6.1 . . . . 205
6.2 Series . . . . . . . . 208
6.2.1 The Integral Test 215
Approximations of Series . 219
6.2.2 The Comparison Tests . . 223
6.2.3 Alternating Series . . . . . 229
Approximations of Alternating Series 231
CONTENTS Vll
6.2.4 Absolute and Conditional Convergence 233

Chapter 6 Exercises . 243
7 Taylor Series 245

7.1 Power Series . 245
Exercises 7.1 . 251
7.2 Representations of Functions by Power Series 252
The Binomial Series . . . . . . . . . . . 263
Taylor Polynomials and Approximations 269
Exercises 7.2 . 279
8 Fourier Series 283

8.1 Fourier Series of Periodic Functions 283
Exercises 8.1 . . . . . . . . . . . . . 297
8.2 Fourier Series of Functions on F inite Intervals 299
Chapter 8 Exercises . . . . . . . . . . . . . . . . 313
9 Series Solutions of Differential Equations 317

9.1 Solutions About Ordinary Points 317
Exercises 9 .1 . . . . 326
9.2 Periodic Solutions . 328
Exercises 9.2 . . . . 332
Appendices
A Formulas and Techniques 335

A.I Trigonometric Identities 335
A.2 Integration Techniques 335
A.3 Matrix Inversion . . . . 340
B Proofs of Selec ted Theorems 341

B.1 The Comparison Test . . . . 341
B.2 The Limit Comparison Test 342
B.3 The Alternating Series Test 343
B.4 The Ratio Test . . . . . 343
B.5 The Root Test . . . . . . 345
B.6 The Binomial Theorem . 346
viii CONTENTS
C Solutions 347
C.1 Chapter 1 Solutions . 347
Index 445
Part I
Ordinary Differential Equations
1
...
Chapter 1
Introduction
1.1 Basic Concepts

Definition 1. 1. An ordinary differential_ equation of order n � l is a relation which
contains the derivative of order n of an unknown function y, called the dependent
variable, and may also contain lower-order derivatives of y as well as given functions
of an independent variable x or t.
Example 1. 1. y' + y = x2 + ex is a first-order ordinary differential equation.
Example 1.2. y111 = 2y2 + y' + 1 is a third-order ordinary differential equation.
The term ordinary refers to differential equations in which the unknown function
y is a function of a single variable. In contrast, differential equations which contain
unknown functions of more than one variable are called partial differential equations.
This book is concerned only with ordinary differential equations, to which we shall
refer simply as differential equations, or equations for short.
Definition 1.2. A solution of a differential equation is any differentiable function f

such that, if y is replaced by J(x), y' by f'(x), y" by J"(x), etc., then the equation is
satisfied for all x in an interval.
Example 1.3. The function f (x) = e 2x is a solution of the equation y' - 2y = 0

because f' (x) = 2e 2x and, hence, f' ( x) - 2f ( x) = 2e 2x - 2e 2x = 0 for all x, the relevant
....
interval here being the entire real line IR = (-oo, oo).
3
4
CHAPTER 1. INTRODUCTION
Exampl� 1.4. The first-order equation y' +x = 0 can be solved simply by integration
and reqmres no special techniques:
Y +x=0 ⇒ y x2
= -x ⇒ y
J
=
I /
-xdx = -- + c
2
where c is an arbitrary constant of integration. Thus, f(x) = --

+ c is a solution
2
x
for every value of· c. In particular, 2
= - x , J(x) = - x + 1 and f(x) = -2 - 2

2 2
x2
f(x)
2 2
are solutions, corresponding to c = 0, c =1 and c = -2, respectively. Their graphs
are displayed in Figure 1.1.
Figure 1.1: Particular solutions of y' + x = 0 and directions.
Definition 1.3. The solution of a first-order equation which contains one arbitrary
constant is called the general solution. It is a one-parameter family of solutions, i.e.,
a set of solutions in which every solution corresponds to a particular value of the
parameter c. A solution in which c is assigned a particular value is called a particular
solution.
In Example 1.4, y + c is the general solution, and

2
= -�
y y=- +1 and y
2 -2
2
x2 x2
= -x , =-
2 2
1.1. BASIC CONCEPTS 5
are particular solutions. At every point (x, y) on any one of these curves, the slope of
the tangent line is :� = -x, i.e., y' + x = 0, which is the differential equation. Thus,
the differential equation defines a direction at every point (x, y) in the plane (where
it is defined), and the graph of every solution is a curve which follows these directions
at every point on the curve. A few directions are indicated in Figure 1.1.
Definition 1.4. The set of directions defined by a first-order differential equation

of the form y' = g(x, y) is called a direction field, and the graph of a solution is called
a solution curve.
In Definition 1.4, g(.rc, y) denotes an expression which may include both x and y.
For example, g( x, y) = x 2 + y2 or g( x, y) = '!!_. Such functions of two variables will be
X
discussed at greater length in Section 2.4.
Definition 1.5. An initial-value problem for a first-order equation consists of a first

order equation, together with an initial condition, which specifies the value of y at a
single point x and, thereby, determines a particular solution.
For the equation y' + x = 0 in Example 1.4, the initial condition y(0) = 1 forces
c = 1. Hence, the solution of the initial-value problem
x2
+ x = 0, y(0) = 1, is y =-
y'
2 + 1.
The initial condition y(4) = -1 forces c = 7. Hence, the solution of the initial-value
problem
x2
+ X = 0, = -1, =-
y' y(4) is y
2 + 7.
Example 1.5. Let y(t) > 0 denote the amount of a substance present at time t, and
suppose that the rate at which y changes with time is proportional to the amount
present at time t. This means that
dy
= ky(t),
dt
which is a first-order ordinary differential equation, where k is the proportionality
constant. This equation determines, for example, the (approximate) number y(t) of
bacteria present in a culture at time t. Since y is an increasing function oft, y' > 0
and, hence, k > 0. This equation also determines the amount y(t) of a radioactive
substance present at time t. Since the substance decays with time, y' < 0 and, hence,
k < 0. If a quantity y(t) is constant in time, then y' = 0 and, hence, k = 0.
6 CHAPTER 1. INTRODUCTION
If we attempt to solve the equation as in Example 1.4, we obtain
y' = ky * y= J kydt,
and the integral on the right cannot be evaluated because y is not known as a function
of t. It is therefore necessary to perform certain algebraic manipulations before an
integration can be performed.
Express the equation in the equivalent form � y ' =k and integrate both sides with
. y
respect to t to obtam
y' = ky * ty' =k * J ty' dt = J kdt.
By the substitution rule,
J
1 , .
-:;/
dt = Jy 1 dy
dt
dt = Jy1
dy,
and the integral on the right can be evaluated without knowledge of y as a function
of t. Thus, we obtain
j tdy = j kdt.
Evaluating both integrals, we obtain
where c1 and c2 are arbitrary constants of integration and the combination c2 - c1

may be denoted by the single arbitrary constant c3. Taking the exponential of both
sides and employing the fact that e10 (a) = a for any a > 0, we obtain
where c = ±ec3 is an arbitrary constant since c3 is arbitrary. At t = 0, we obtain

y(0) = c. Thus, the solution may be expressed as y(t) = y(0)e kt _ The quantity y
grows exponentially if k > 0, decays exponentially if k < 0, and has the constant
value y(0) if k = 0.
dy
Example 1.6. The equation = 2y has the general solution y(t) = y(0)e 2t by
dt
Example 1.5 with k = 2.
Example 1. 7. The initial-value problem y' = -3y, y(0) 2, has the solution
y(t) = 2e- 3t by Example 1.5 with k = -3 and y(0) = 2.
1.1. BASIC CONCEPTS 7
Example 1.8. Find the general solution of y' - xy = 0 and solve the initial-value
problem given the initial condition y(0) = -2.
If y(x) = 0 (i.e., y = 0 for all x), then y'(x) = 0, and the equation is satisfied. If
tv' j tv' jt
y(x) ¢. 0, then, proceeding as in Example 1.5,
y' - xy = 0 ⇒ y' = xy ⇒ =x ⇒ dx = j x dx ⇒ dy = j x dx
2
x
⇒ ln lvl = -
2
+ C ⇒ lvl = e e
c x2 /2
⇒ y = ±ecex2 /2 = Kex2 /2
is the general solution, where K = 0 corresponds to the solution y 0. y(0) = -2 ⇒ =
K = -2 ⇒ y = -2ex / 2 is the solution of the initial-value problem.
2
The method employed in solving the first-order equations in the present chapter
ca,11 be applied to all equations of a particular type, and will be discussed in greater
generality in Chapter 2, as well as the methods required to solve other types of
first-order equation. Second-order equations will be discussed in Chapter 3, where
the concepts general solution, particular solution and initial-value problem will be
suitably generalized.
Exercises 1.1
1. Find the general solution of the equation y' + 2x = xe x .
1
2. Find the general solution of the equation -y' = 4cos(2x).
X
3. Find the general solution of the equation !� = 2x cos(x2 ).
2
dy
4. Find the general solution of the equation ( = 4x2e 2x . 2
dx)
1
5. Find the general solution of the equation -- - y' = 2 sin(x).
COS (X )
dy
6. Solve the initial-value problem = ln(t), l > 0, y(l) = 2.
dt
7. Solve the initial-value problem y' = 4cos 2(x), y (�) = i
. . . x+3
8. Fm d the genera1 so1ut10n of the equation y' =
x _x_6
2
8 CHAPTER 1. INTRODUCTION
y
9. Find the general solution of the equation d = 12e 2x sin(3x).
x
!� =
d
10. Solve the initial-value problem 3y, y(O) = 2.
y
11. Solve the initial-value problem d = -2y, y(O) = 3.
dt
12. Solve the initial-value problem y' + cos(x)y = 0, y(1r) = -3.
13. Solve the initial-value problem 3y2 y' = 1, y(O) = 2.

Chapter 2
First-Order Equations
2.1 Separable Equations

Definition 2.1. A first-order differential equation is separable if it can be placed in
the form
dy
g(y) = (x)
dx J ,
possibly after some algebraic manipulation.
The essential characteristic of a separable equation is that the quantity g(y) which
dy
multiplies is independent of x, whereas the right-hand side f(x) of the equation is
dx
independent of y. Either or both of these functions may, of course, be constant. The
reader may readily verify that the equations in Examples 1.4, 1.5 and 1.8 in Chapter
1 are separable. The method employed in their solution is applicable to any separable
equation. Thus, employing the substitution rule, we obtain
g(y) :� = f(x) ⇒ J g(y) :� dx = J J(x) dx ⇒ J g(y) dy = J J(x) dx,
and evaluation of the last two integrals gives a relation between x and y containing
one arbitrary constant. (As seen in the examples in Chapter 1, a difference c2 - c 1
of two arbitrary constants may be replaced by a single one.) It may or may not
be possible to solve the relation for y explicitly as a function of x. Nevertheless, it
constitutes, at least in principle, the general solution.
3x 2
Example 2.1. The equation 2d is separable because it can be placed in the
dy
X
= y
dy
form 2y- = 3x 2
. Then
dx
J 2y �� dx = j 3x 2 dx ⇒ j 2y dy = j 3x 2 dx ⇒ y2 = x3 + c ⇒ y = ±Jx3 + c
9
10 CHAPTER 2. FIRST-ORDER EQUATIONS
is the general solution. If the initial condition y(O) = 2 is imposed, then it requires
that 2 = ±Jc, which gives c = 4 and we must take the plus sign. The solution of
the initial-value problem is then y = ✓x3 + 4. If, instead, the initial condition is
y(O) = -2, then,again, c = 4, but we must take the minus sign, and y = - ✓x3 + 4.
Example 2.2. The equation (x2 + 4)y' = .!__ is separable because it can be placed
2y
in the form 2yy' = -!--.
X+4
Then
j 2yy' dx = j :
x2 4
dx => j 2y dy = j x2
:
4
dx => y2 = � ln(x2 + 4) + c,
du
employing the substitution rule with u = x2 + 4, - = 2x,to obtain
dx
J x2
x
+4
dx =
1
2 J 2x
� dx = 2
1
J 1 du
; dx dx =
1
2 J 1
; du =
1
2 ln(u) + c.
Hence, y = ± ✓� ln(x2 + 4) + c.
Example 2.3. Consider the initial-value problem

2
2xyy' = e- Y ln(x), x > 0, y(l) = 2.
2 , ln(x)
The equation can be placed in the form 2yeY y = --,and is separable. Thus,
X
j 2yeY y' dx =
2
j ln x)
� dx => j 2ye Y dy =
2
j ln x) .
� dx
du
Employing the substitution rule with u = y2 , - = 2y, in the integral on the left,
dy
dv
and v = ln(x), = ! ,in the integral on the right, we obtain
dx x
=> y2 = ln { �[ln(x)] 2 + c} => y = ± In { �[ln(x)] 2 + c}
is the general solution, and
y(l) = 2 => ±� = 2 => ln(c) = 4 => c = e4 => y = In { �[ln(x)] 2 + e4 }
is the solution of the initial-value problem.
. \ .
2.1. SEPARABLE EQUATIONS 11
2.1.1 Orthogonal Trajectories

A circle of radius r > 0 centred at the origin can be described by the equation
x2 + y2 = r 2 . The set ( or family or collection) of all such circles constitutes a one
parameter family of circles, one circle for each value of the parameter r. If we let
J(x, y) = x 2 + y 2 , then the family may be described by the equation f(x, y) = r2 or,
alternatively, g(x, y, r) = x 2 + y 2 - r2 = 0.
A straight line through the origin with slope m can be described by the equation
y = mx. The set of all such lines constitutes a one-parameter family of lines, one line
for each value of the parameter m. With f(x) = x, the family may be described by
y = mf(x) or, alternatively, g(x, y, m) = mx - y = 0.
A few circles centred at the origin and a few lines through the origin are displayed
in Figure 2.1.
Figure 2.1: Circles centred at the origin and lines through the origin.
More generally, a relation between x and y which includes a parameter (constant)

defines a set of curves in the plane, one curve (or more) corresponding to each value
of the parameter.
Definition 2.2. A one-parameter family of curves is a set of curves defined by an

equation of the form g(x, y, c) = 0, where c is the parameter.
The relation g(x, y, c) = 0 my take more particular forms, such as y = cf(x),

y = J(x, c), or J(x, y) = c. In the first case, g(x, y, c) = y - cf(x) = 0, in the second,
g(x, y, c) = y - f(x, c) = 0, and, in the third, g(x, y, c) = f(x, y) - c = 0.
Notice that, in Figure 2.1, every line intersects every circle orthogonally, i.e., at
right angles. More precisely, at every point of intersection, the line tangent to a circle
at that point is orthogonal (perpendicular) to the line through the origin.
Definition 2.3. A curve is an orthogonal trajectory of a one-parameter family of

curves if it intersects every member of the family at right angles.
Thus, every straight line through the origin is an orthogonal trajectory of the
family of circles centred at the origin, and every circle centred at the origin is an
orthogonal trajectory of the family of lines through the origin.
Given a one-parameter family of curves, it is often necessary to determine its

orthogonal trajectories. To this end, we note that if a line L has slope m, then a
1
line perpendicular to L must have slope --. This fact enables us to determine the
m
orthogonal trajectories of a given family of curves with the aid of differential equations.
X
Example 2.4. The general solution of the first-order equation y' = -- is a one
y
parameter family of curves. Find the general solution and its orthogonal trajectories.
The equation is separable, and

y X2 2
YY
I
= -x ⇒ 2 = -2 + c1 ⇒ y2 = -x 2 + 2c1 ⇒ x2 + y2 = 2c1 = c,
which defines a one-parameter family of circles centred at the origin for c > 0. Its
orthogonal trajectories must satisfy the equation y' = '}!_, which is also separable. If
X
y ¢. 0, then
�y'
y
=�
X
⇒ ln IYI = ln lxl + C ⇒ lvl = e ixl
c
⇒ y = ±ecx = K x,
=
which defines a one-parameter family of lines through the origin, in agreement with
the result inferred from Figure 2.1. The constant function y 0 (the x-axis) satisfies
the equation and corresponds to K = 0. The vertical line x = 0 (the y-axis) is an
orthogonal trajectory, but is excluded from the solution y = K x. It can be included
by regarding x as a function of y, expressing the equation y' = '}!_ as
X
dx
x, =-=-=-=-
1 1 x
dy dy y' y'
dx
and noting that x = 0 is a solution.

2.1. SEPARABLE EQUATIONS 13
In Example 2.4, the one-parameter family of curves was obtained by solving a

first-order equation. If no equation is given, then an equation must be determined for
which the given family is the general solution. Thus, given a one-parameter family
of curves, g(x, y, c) = 0, express the relation as f(x, y) = c and differentiate with
respect to x, regarding y as a function of x, to eliminate c. Or, eliminate c from the
two equations g(x, y, c) = 0 and its derivative with respect to x, regarding y as a
function of x.
Example 2.5. Find the orthogonal trajectories of the one-parameter family of

parabolas y = x 2 + c.
Differentiation with respect to x gives y' = 2x. The orthogonal trajectories must
1 1
therefore satisfy y' = --, the general solution of which is y = -- ln lxl + k, and
2x 2
defines the orthogonal trajectories (Figure 2.2).
y )
Figure 2.2: The parabolas y = x2 + c and their orthogonal trajectories.
Example 2.6. Find the orthogonal trajectories of the one-parameter family of

parabolas y = kx2 .
. as -
Y = x2 y' - 2xy
We may express the equat10n
X2
k and then differentiate to obtain ---- = O'
x4
2y
.1.e., xyI - 2 y = Q, or yI = -. Alternatively,
X
2
y = kx2 ⇒ :2 = k and y' = 2kx ⇒ y' = 2x ( :2 ) = :,
as before. The orthogonal trajectories must therefore satisfy y' = _ _.:._, which is
2y
separable. Thus,
y' = -;
y
==> 2yy' = -x ==> y2
2
= -� + c ==> y = ±g, or �
2
+ y2 = c,
which defines a one-parameter family of ellipses for c > 0. A few members of each
family are displayed in Figure 2.3
Figure 2.3: The parabolas y = kx 2 and their orthogonal trajectories.
Exercises 2 .1
1
1. Solve the initial-value problem y' = 3x 2y2, y(0) =
_
°
2
• •
2. Fmd the general solution of the equation y =
• I Jx+I .
y3
dy 2x + 1
3. Solve the initial-value problem (x 2 + 1) = y(0) = 3.
dx 3y2 ,
. . y' X
4. Find the general solution of the equation
4x2 + 3x- 2 - 5y 4
- 4y3 + 1-
I
. .. , ln(x)
5. Solve the m1tial-value problem xy - -- = 0, x > 0, y(l) = -3.
y
6. Find the general solution of the equation xy' = (x + l)e Y .
---- ---
2.2. HOMOGENEOUS EQUATIONS 15
' xy + x
7. Solve the initial-value problem y = , y(2) = -2.
xy-y
8. Solve the initial-value problem cos(x) ln(y) 1� = ysin(x), Y > 0, y(O) = e 2.

x2 . .
9. Find the general solution of the equation y' = - and its orthogonal traJectones.
y
10. Find the general solution of the equation y' = 2x(y2 + 1) and its orthogonal
trajectories.
11. Find the orthogonal trajectories of the one-parameter family of curves y = x3 +c.
12. Find the orthogonal trajectories of the one-parameter family of curves y = kx 4 .
13. Find the orthogonal trajectories of the one-parameter family of curves y = ecx.
2.2 Homogeneous Equations

Definition 2.4. A first-order equation is homogeneous if it can be placed in the form
y' = f (�),possibly after some algebraic manipulation.
For example, the equations
y
y' = '!!_, y' = X
+ = l + '!!._
X X X
and
x2 + y 2 x y l y
y' =---=-+-=-+-
xy y x y/x x
/ X2 + Y . not, smce
Y 1s . x and y may appear only
are homogeneous, whereas y = -- =x+-
X X
as the combination '!!._.
X
A homogeneous equation is solved by making the substitution (also called a change

of variable) u = '!!_, where u is a new function of x. The homogeneous equation for the
X
unknown function y is then transformed into a separable equation for the unknown
function u: Employing the product rule, we obtain
u= � ⇒ y = xu ⇒ y' = u + xu' ⇒ u + xu' = y' =f (;-) = f (u)

1 1
⇒ xu' = J (u) - u ⇒ f(u) - U
u' -
X
,
which has the form g(u)u' = h(x), and is separable. After u is found, y is obtained
by y = XU.
,
'
I
Example 2.7. The equation

2 + 2
/ y 1
y=--=-+-
X Y
xy y/x x ---
shown above. The substitution u = '!!... gives
,-L--
is homogeneous, as
X
1 y l
+ XU,=y,= - + - = - + U.
: yjx X
U
U
1
Then xu' = - ,which is separable. Thus,
u
1 1 1
xu' ⇒ uu = -
I
⇒ -u2=ln x
l l + c1 ⇒ u2=21n lxl + 2c1 = 21n lxl + c
U X 2
⇒ u = ±J2 ln lxl + c ⇒ y=xu = ±xJ2 ln lxl + c ..
xy' = yn
l (y) -yln(x), x > 0, y > 0,
can be placed in the form
y'='!!... [ln(y) - ln(x)]= ln

X
IX (I),
X
and is homogeneous. Then u = '}!_ gives u + xu' = u ln( u)

X
⇒ xu' = u ln( u) - u ⇒
1 , 1
u =- ⇒ J l
] du=
l- dx=ln lxl + c1.
unu-l
[l ( ) ] x
[
unu-l
l ( ) J x
In order to evaluate the integral on the left, make the substitution v= ln(u) - 1.
dv l
Then - =-,and we obtain
du u
J
unu
l
[l ( ) - l
] du=
J l dv
- -du=
v du J l
- dv=ln v
v
l l = ln I ln(u) - 11-
\
x2 y' = x2 + xy + y2 , y(l) = l.
= I + ;; + (Y)
y 2
The equation can be placed in the form y I ;; , . h omogeneous.
and 1s
y 1 / 1
'/1, = - =} y = XU =} U + XU
I
= yI = l + U + U2 =} XU
I
= 1+ U 2 =} --- U
1 2+u
f- J
X X
1
⇒ - du= _!-_ dx ⇒ tan-1(u) = ln lxl +c ⇒ u = tan(ln lxl + c).
1 + u2 X
7f
Hence, y = x tan(ln lxl + c) is the general_ solution. y(l) = 1 ⇒ tan(c) = 1⇒ c = ,
4
and the solution of the initial-value problem is y = x tan (in lxl + �).
y e- y/x
can be placed in the form y' = - + --, and is homogeneous.
X y1 X
e-U
=} U + XUI = y I = U +-
J J�
u
⇒ ueu du = dx = ln lxl + c.
Integration by parts gives J ueu du= ueu - eu . Hence, ueu - eu = ln lxl + c, i.e.,
It is not possible to solve this relation for y explicitly in terms of x. Nevertheless, ,

this is the general solution, with y defined implicitly by the above relation.
Definition 2.5. A solution of a first-order equation defined by a relation between

x and y is called an implicit solution. If it contains one arbitrary constant, then it is
the general solution and has the form g(x, y, c) = 0.
In Example 2.10, the solution may be placed in the form
g(x, y, c) = '!/__ ey/x - ey/x - ln lxl - c = 0.

X
(
Exercises 2. 2
x+
1. Solve the initial-value problem y' = Y, y(l) = 2.
X
2. Solve the initial-value problem xyy' = y2 - x2 , y(-1) = 2.
3. Solve the initial-value problem y' = V:, + '!!_, x > 0, y > 0, y(l) = 9.
vY X
4. Find the general solution of the equation y' = '}!_ +� x

X x2 y + ,/xY + X
3
5. Solve the initial-value problem y +x sec(�) -xy' = 0, y(l) = i·

2x + 3Y
6. Find the general solution of the equation y' =
3x - 2y
x3 +x2 y +xy2 + y3
7. Find the general solution of the equation y' =
X 3 + X 2 y +xy2
8. Solve the initial-value problem yy' = 6x + y, y(l) = 4.
2.3 Linear Equations

Definition 2.6. A first-order equation is linear if it has the form
a(x)y' + b(x)y = c(x),
where a(x) ¢ 0, b(x) and c(x) may be any functions of x whatsoever (or constants),
but are independent of y and y'. An equation which is not linear is called nonlinear.
For example, the equations ex y' + ln(x)y = x3 + 1, y' = 6 and y' + x2 y = 0 are
linear, but yy' = l, y' + sin(y) = 0 and y' + y2 = x are nonlinear.
)
b(x
In order to solve a linear equation, divide the equation by a(x) and let P(x) = )
. ax(
c(x)
and Q(x) = to place the equation in the standard form
ax( )
J y' + P(x)y = Q(x).
Multiply the standard form of the equatiun by a function I(x), to be determined, to

obtain
I(x)y' + I(x)P(x)y = I(x)Q(x).
2.3. LINEAR EQUATIONS 19
The function J(x) is chosen so that the left-hand side I(x)y'+ I(x)P(x)ybecomes
[J(x)y]'. Employing the product rule, this requirement becomes
I (x)y'+ I(x)P(x)y= [J(x)y]'= I(x)y'+ I'(x)y,
'(x)
which gives l(x)P(x)y= I '(x)y, or I(x)P(x) = I' (x) . Then 1!· = P(x) and,
(x)
employing the substitution rule, we obtain
j} :�dx =JP(x)dx ⇒ J }dI= P(x)dx j ⇒ lnlIx()I = P(x)dx , j

from which we obtain
J(x) = ef P(x )dx ,
calle<.l an integrating factor. Arbitrary constants and ± signs are not necessary here,
.since we seek only one function I ( x), and not all such functions. The equation then
becomes [i(x)y]=
' J(x)Q(x) , and one integration gives I(x)y= J I(x)Q(x)dx, from
which we obtain
y= ! �
()
J J(x)Q(x)dx.
Example 2.11. The equation y' + 2y = 1 is linear, with the integrating factor
I(x) = ef
2d
x =e 2x _
Tli,:: eq L1i'Ltion then becomes
·;ri1egrat10n
· of w h"1ch gives
· e
2x
y= J e dx
2x
= 1 e 2x + c and, hence, y =
2
1
2 + ce-2 x_
Exct.mple 2.12. Consider the initial-value problem I
/ 1
y --y=lnx)
( , x>O, y)=-3.
l(
X
The equation is linear, with the integrating factor J(x) = d

ef -� x = e- ln( ) =
x
!.
X
The
equation then becomes
! y' _ 2_2 y=ln(x) , 1.e., l- y '

( )
ln()
x
X x X x X
du ]:_
integration of which , employing the substitution rule with u = ln(x) and = '
.
gives
dx x
l
-y =
X
X
J ln(x)
--dx
X
=
J du
u- dx
dX
=
J l
udu = -u2
2
+c=
X
1
-[ln(x)] 2
2
+ c'
and y = -[ln(x)] 2 + ex. Then y(l) = -3 ==> c = - ==> y = -[ln(x)] 2 - 3x.
3
2 2
Example 2.13. The equation xexy' + ex y = 4xe-x is linear. In standard form, it is

1
y' + - y = 4e-2x, with the integrating factor
X
I(x) = ef ½dx = l l = lxl

e' n x = ±x,
and I(x) = x will do . The equation then becomes
· , (xy)' = 4xe-2x
xyI + y = 4xe-2x, 1.e.
J J
Integration by parts then gives
xy = 4x e-2x dx = -2xe-2x + 2e-2x dx = -2xe-2x - e-2x + c,
e-2x C
and y = -2e-2x - -- + -.
X X

cos(x)y' + sin(x)y = 2sin(x)cos3(x), y(1r) = - 2.
The equation is linear, with the standard form y' + tan(x)y = 2sin(x)cos2 (x). The
integrating factor is
1
J(x) = eftan( x )dx = eln lsec(x)I = x I = ±--,
I sec()
cos(x)
and I ( x) = __J_ will do. The equation then becomes

COS()X
1 , sin(x) .
- - y + 2 ()
y = 2sm(x)cos(x),
COS(X -) COS X
l.C.'
[cos\x)
v]'
= 2sin(x)cos(x) = sin(2x),
J
integration of which gives
l l 1
- - - y = sin(2x) dx = -- cos(2x) + c ==> y = --cos(x
2
)cos(2x) + ccos(x).
COS(X ) 2
1 5 1 5
Then y(1r) = - 2==> -2 = - c ==> c = ==> y = - cos(x)cos(2x) + cos(x).
2 2 2 2
2.3.1 Bernoulli Equations

Definition 2. 7. A first-order equation is a Bernoulli equation if it has the form
a(x)y' + b(x)y = c(x)ya ,
wherea(x) ¢. 0, b(x) and c(x) are functions of x (or constants), but are independent
of y and y', and a is a real number.
If a= 0, then the equation isa(x)y' + b(x)y = c(x), which is linear, and if a= 1,

then the equation isa(x)y' + b(x)y = c(x)y, i.e.,a(x)y' + [b(x) - c(x)]y = 0, which
is also linear. For a -/= 0, 1, a Bernoulli equation can be transformed into a linear
equation by means of the substitution (change of variable) u = y 1 -a. Then
and, empioying the chain rule,
1 ---1 I
1 1
YI = --U 1 0 U = _g_ I
--U1-0U
1-a 1-a
and the equation becomes
1
+ b(x)u� = c(x)uT=o.
et I I a
--a(x)u 1 -au
1-a
-Ct
Multiplication by u� then gives
� a(x)u' + b(x)u = c(x),

1 a
which is linear. Once the linear equation is solved for u(x), the solution y(x) of the
Bernoulli equation is obtained by y = u �.
Example 2.15. The equation xy' + 2y = y2 is a Bernoulli equation with a = 2.

With 1 - a= -1, u = y- 1 ⇒ y = u- 1 ⇒ y' = -u- 2 u1 , y2 = u- 2, and the equation
becomes
2 1
-xu' + 2u = 1, or u' - -u = --
x x'
which is linear, with the integrating factor
Then
l , 2 1 1 '
x 2u - u= -
x x3
⇒ ( 2 u)
1
⇒ I_u=j-I_dx= --+c
1
3
x x3 x2 x3 2x2
⇒ u=- 1 +cx2 ⇒ y =u
-1
=
1
1 2
2
2
+ cx2 - 1 + kx2
is the general solution.
Example 2.16. The equation 2y' - y = exy3 is a Bernoulli equation with a = 3.

With 1 - a = -2, u = y -2 ⇒y= u- 1/2 ⇒ y' = -tu - 31 2
u', y
3 = u- 3 1 2 , and the
equation becomes
i.e.' -uI -u= e,

X
or u' +u= -ex ,
which is linear, with the integrating factor I(x) = ef 1 dx = ex . Then
is the general solution. Note thatu112 = ±Ju. In fact, it is evident from the Bernoulli
equation that if y is a solution, then so is -y, because
2x2
I
y' + 2x2 y= vY, y > 0, y (O) =4.
1 3
The equation is a Bernoulli equation with a= -
2. With 1 - a=
2,
and the equation becomes
i.e.'
which is linear, with the integrating factor J(x) = ef3x 2

dx = ex3 . Then
3
3
e x u' + 3x2 ex3 u = 3x2 ex3 ⇒ ( ex u)
'
= 3x2 ex3 ⇒ ex u =
3 J 2
3x ex dx = ex
3 3
+c
⇒ u = 1 + ce-x ⇒ y = u
3 2 /3 = (1 + ce-x3 ) 2 /3
is the general solution, and y(0) = 4 ⇒ (1 + c)2 1 3 = 4 ⇒ c 43/2 - 1 7 ⇒
y = (1 + 7e- x )
3
213
Exercises 2. 3 \
1. Solve the initial-value problem y' + y = e- x , y(0) = -2.

2. Find the general solution of the equation y' - 3y = 1.
3. Solve the initial-value problem ex y' + 2ex y -1 = 0, y(0) = -3.
4. Find the general solution of the equation y' + 2xy = 2x.
1
5. Solve the initial-value problem xy' + y = -, x > 0, y(l) = 2.
X
6. Find the general solution of the equation x 3 y' - 2y + 1 = 0.

1
7. Find the general solution of the equation x2 y' + xy = , x > 0.
ln(x)
+ sin(x)y = cos4 (x), y(1r) = .

7f
8. Solve the initial-value problem cos(x)y'
2
9. Find the general solution of the equation xy' + 2y = e- x
.
x3
lO. Find the general solution of the equation xy' - y = _ __
x+l
e-x
11. Find the general solution of the equation y' + y =
X 2 + 2x + 2
12. Find the general solution of the equation x ln(x)y' + y = - 22x , x > 0.
X -1
1
13. Solve the initial-value problem 3y' - y = 2, y(0) = 2.
y
14. Find the general solution of the equation y' + 4y = 2e- x y'Y, y > 0.
15. Solve the initial-value problem 2x2 y' - 2xy = -y3 sin(x), y (�) = 1r.
16. Find the general solution of the equation xyy' + y 2 - xex = 0.
17. Find the general solution of the equation x ln(x )y' + y = 2yy/y, x > l, y > 0.
2 .4 Functions of Two Variables

The study of a type of first-order equation called exact, to be defined in Section 2.5,
requires that certain basic properties of functions of two variables be established first.
Definition 2.8. A function f of two variables assigns to any pair (x, y) in its domain
a real number denoted byf ( x, y), where the domain of f is either specified or taken
to be all points (x, y) in the plane where f ( x, y) is defined.
Several such functions have already been encountered in the preceding sections.
For example,
f(x, y) = x2 + xy - y3, g(x, y) =-+1

X
and h(x, y) = Jx - y
y
are functions of two variables. In the.above, the domain off is the entire plane IR2,
that of g is the set of all points (x, y) with y i= 0, and that of his the set of all points
(x, y) with x - y 2'. 0, i.e., x 2'. y.
It is important to note that, in the definition of a function of two variables, y is

not a function of .1:, and that x and y may be assigned values independently. In other
words, x and y are independent variables. For example,
J(x, y) = y - x2 + 1 ⇒ J(l, 2) = 2, J(2, 1) = -2, and f(0, 0) = 1.
It is only a relation, such as f(x, y) = 0 or f(x, y) = c, which defines y as a function

of x. For example, f(x, y) = y - x 2 + 1 = 0 ⇒ y = x 2 - 1.
2.4.1 Partial Derivatives

The (first-order) derivative of a functionf of a single variable x is the function f' or
df
defined by
dx
f J(x + h) - J(x).
J'(x) = d (x) = lim
dx 1i-o h
If z =f(x), then dz = J'(x) is the rate of change of z with respect to x.

dx
In contrast to functions of one variable, there are different ways in which functions
of two variables can be differentiated.
2.4. FUNCTIONS OF TWO VARIABLES 25
Definition 2.9. Let f be a function of two variables x and y. The partial derivative
off with respect to x is the function j� or �� of two variables defined by
_of x, y -l· f(x+h,y)-J(x,y)

X ( x, y) -
! ( ) - Im .
!:}
uX h->O h
The partial derivative off with respect to y is the function J� or �� of two variables
defined by
fy (x,y ) _
- a f(x,y) _- 1.Im J(x,y + h)-J(x,y)
.
!:}
uy h->O h
z
If z = J(x,y), then o = fx (x,y) is the rate of change of z with respect to x with
ox
y held constant, and �: = j�(x, y) is the rate of change of z with respect toy with
x held constant.
Example 2.18. Let f(x,y) = x4y2 + 2x- 3y+ e x - ln(y) - sin(y) + cos(x) + 2. Then
fx (x,y) = 4x3y2 + 2 + e x - sin(x) and fy (x,y) = 2x4y - 3 - � - cos(y).

y
In the computation of fx , y is regarded as a constant, and all expressions which are

'independent of x, such as y, ln(y) and sin(y) have derivative with respect to x equal to
0. In the computation of fy , xis regarded as a constant, and all expressions which are
independent of y, such as x, e x and cos(x) have derivative with respect toy equal to 0.
A function J of two variables has four second-order partial derivatives, namely,

the first-order partial derivatives of the two functions fx and fy • The four possibilities
are as follows:
a Jx = O af 3 f. h
2
fxx = . 1 denvat1ve
1s t e partia . . of fx wit . h respect to x,
ox ox ox ox2
a
� fx = �� =
aa
f a2 f
� .1s tl1e partia . 1 denvative
.
.
. of 1·x with respect toy,
uy uy ux uyux
a fy
aa
f o2 f . .
. i denvative
1s the partia
.
. of fy with respect to x,
ox· = ox oy = oxoy
a1
a af
= �� =
a2 f . .
uy ·y
� !:}
2
1s the partia
. i denvative
. . of fy with respect toy.
uy uy uy
Example 2.19. Let f(x,y) = x3 y4 + e 2x ln(y) + x2 - y3 - 3. Then
fx (x,y) 3x2 y4 + 2e2x ln(y) + 2x,

e2x
fy (x,y) 4x3 y3 + - - 3y2 ,
y
fxx (x,y) 6xy + 4e2x ln(y) + 2,
4
2e 2x
fxy (x,y) 12x2 y3 + --,
y
?e 2x
12x2 y3 + -�-,
y
e 2x
12x3 y 2 - - - 6y.
y2
Note that, in this example, fxy = fyx · Although this is not true in general, the
conditions under which it is true are given by the following:
Theorem 2.1. If fxy and fyx are continuous in an open region R of the (x,y)-plane
(i.e., R excludes its boundary), then f�y (x,y) = fyx (x,y) for all (x,y) in R.
Continuity of a function of two variables is defined in a manner analogous to the

definition for a function of a single variable:
Definition 2.10. A function f of two variables is continuous at a point (a, b) if
lim f(x,y) = f(a, b).

(x,y)-+(a,b)
f is continuous in a region R if f is continuous at every point (a, b) in R.
2.4.2 The Chain Rule

The chain rule for functions of a single variable states that, if z = f ( x) and x = g ( t),
then the composition h(t) = f(g(t)) has derivative
h'(t) = ! h(t) = J'(g(t))g'(i).
. . . � ��
In other words, 1f z 1s a funct10n of x and xis a function oft, then - = --.
dt dx dt
Several versions of the chain rule exist for functions of two (or more) variables.
The one of interest in the present study is the following:
2.4. FUNCTIONS OF TWO VARIABLES 27
Theorem 2.2. (The Chain Rule) If f is a function of two variables and x and
y are functions of a single variable t, then the composition h(t) = f(x(t), y(l)) has
derivative
d � �
h'(l) = J(x(t), y(t)) = fx(x(t), y(t)) dl + fy (x(l), y(l)) dt . (2.1)
dl
In other words,if z = f (x, y) and x and y are functions of t, then
dz 8 f dx 8 f dy
-=--+--.
dt 8x d t 8y dl
Note that, in the special case where z = f(x) is independent of y, �� = 0 and

8f df dz df dx d z dx . .
- =-,and the above formula reduces to - = -- = -- , 1.e.,the cham rule
8x dx dl d x dl dx dl
for functions of one variable.
Example 2.20. Let f(x, y) = x2 + y3, x(t) = t3 - 6l + 1 and y(t) =et +3t. Then
h(t) = J(x(t), y(l)) = (t 6t +1)2 +(e t +3t)3

3
-
is a function of t, and
8 ! dx 8 ! dy
h' (t)
8X dt
+
8y d t
= 2x(t)(3t2 - 6) +3[y(t)] 2(e l +3)
2(t3 - 6l +1)(3t2 - 6) +3(e +3t) (i +3).
t 2
dx
In Theorem 2.2, consider the special case where x(t) = l. In that case, =1,
dt
and Equation (2.1) reduces to
or, employing the fact that x =t,
d dy
J(x, y(x)) = fx(x, y(x)) + fy (x, y(x)) dx , (2.2)
dx
which is the version of the chain rule required in the study of exact equations.
Exercises 2.4
1. Let f(x, y) = x3 - y2 + x2 y2 + 1.
(a) Determine J(l, 0), f(3, 1), f(l, 3) and f(2, -1).
(b) Find fx (x,y) and fy (x,y), and the values of these functions at the points
(1, 0), (3, 1), (1, 3) and (2, -1).
(c) Find fxx (x,y), fxy (x,y), fyx (x,y) and fyy (x,y), and the values of these
functions at the points (1, 0), (3, 1), (1, 3) and (2, -1).
2. Let J(x,y) = x2 ln(y) - y3 ex + x3 - y2 + 1. Find fx , fy , fxx , fxy , fyx and fyy ·

3. Let f(x,y) = x4 + 3x2 y2 + y4, x(t) = t2 + 2t and y(t) = t3 - 2t. Determine
d
f ( X (t) ' y ( t)).
dt
4. Let f(x,y) = x2 - cos(y) + x sin(y), x(t) = 2t 2 + 3t - 1 and y(t) =et + 2t.
Determine : f(x(t),y(t)).
t
5. Let f(x,y) = sin(xy) + y!XY, x(t) el + l and y(t) ln(t). Determine
d
f ( X (t)' y (t)).
dt
6. Let f(x,y) = x4 + 3x2 y2 + y4 and y(x) = x3 + 6x2 - x. Determine J(x,y(x)).

..!!:.._
dx
7. Let f(x,y) = eY - x2 + y3 ln(x) and y(x) = sin(x). Determine :J(x,y(x)).

d
l d
8. Let f(x,y) = + xe- y and y(x) = sin(2x). Determine xf(x,y(x)).
X2 +y 2
d
9. Let f(x,y)= x4 - y4 - l. The relation J(x,y) = 0 defines y as a function of x.
dy
Determine - m. terms of x and y.
dx
10. Let f(x,y) = x3 + y2 - eY .
The relation f(x,y) = 0 defines y as a function of
dy
x. D eterminc - m. terms of x and y.
dx
2.5 Exact Equations

Consider a first-order equation expressed in the form
dy
P(x,y) + Q(x,y) = 0, (2.3)
dx
where P and Q are functions of two variables.
Suppose that there exists a function J of two variables such that fx = P and
fy = Q. Then Equation (2.3) takes the form
dy
fx + fy =
dx 0
2.5. EXACT EQUATIONS 29
which, by Equation (2.2), is !I:_

dx
J(x,y(x)) = 0. Integration with respect to x then
yields f(x,y(x)) = c, an arbitrary constant. Thus, the general solution of Equation
(2.3) is J(x,y) = c. The problem, then, is the determination of the function f, called
a potential function.
Definition 2.11. An equation of the form P(x,y) + Q(x,y) ;� = 0, with P and Q

continuous in an open region R of the (x, y)-plane, is exact in R if there exists a
potential function J defined for all (x, y) in R.
Example 2.21. Consider the equation y + xy' = 0. Evidently, J(x,y) = xy is a

potential function because fx = y = P and fy = x = Q. Hence, the equation is exact
in IR 2 , with the general solution f(x,y) = c, i.e., xy = c.
Theurern 2.3. Suppose that P and Q have continuous first-order partial derivatives
in an open region R of the plane. If the equation P(x,y) + Q(x,y)y' = 0 is exact in
R, then Py (x,y) = Qx (x,y) for all (x,y) in R.
Proof. Since the equation is exact, there exists a potential function f. Then fx = P
and fy = Q * fxy = Py and fyx = Qx , Since Py and Qx are continuous in R, fxy and
fyx are continuous in R, hence equal by Theorem 2.1. Thus, Py = Qx in R.
Theorem 2.3 is equivalent to the following:
Theorem 2.4. If Py =I= Qx in R, then the equation P(x,y) + Q(x,y)y' = 0 is not

exact in R.
Proof. Suppose that Py =I= Qx in R. If the equation is exact, then Py = Qx by

Theorem 2.3, which contradicts Py =I= Qx . Hence, the equation is not exact.
dy
Example 2.22. Consider the equation x 2 + y2 + xy = 0.
dx
P(x,y) = x 2 + y2 , Q(x,y) = xy, Py = 2y, Qx = y, Py =I= Qx, hence, the equation is
not exact, by Theorem 2.4.
It is not true, in general, that if Py = Qx, then the equation P(x, y)+Q(x,y)y' = 0
is exact. In order to state a converse of Theorem 2.3, i.e., a theorem which gives the
conditions under which exactness follows from Py = Qx , the following definition is
required.
Definition 2.12. A region R in the (x, y)-plane is connected if it consists of a single

"piece." It is simply connected if it is connected and has no "holes" in it.
Definition 2.12 is, of necessity, merely informal, since the mathematically rigorous
definitions of connected and simply connected sets are beyond the scope of this book.
For example, the region R enclosed by a non-self-intersecting closed curve (e.g.,

a circle) is simply connected. The entire plane is simply connected. However, the
plane with the origin excluded is not simply connected due to the "hole" at the origin.
Theorem 2.5. Suppose that P and Q have continuous first-order partial derivatives
in an open, simply connected region R of the plane. If Py = Qx in R, then the
equation P(x, y) + Q(x, y)y' = 0 is exact in R.
Note that, if the region R is not simply connected, then the equation may not be
exact, even if Py = Qx -
Example 2.23. Consider the equation x 2 + y2 + 2xyy' = 0.
P(x, y) = x + y 2 , Q(x, y) = 2xy, Py = 2y, Qx = 2y, Px = 2x and Qy = 2x

2
are continuous in the plane, which is simply connected, and Py = Qx . Hence, the
equation is exact by Theorem 2.5. A potential function f exists, and satisfies the two
conditions fx = P and fy = Q.
fx = P ⇒ f(x, y) = J P(x, y) dx = J x2
3
+ y2 dx = � + xy2 + g(y),
where g(y) is an arbitrary function of y. Note that, in the above integration with
respect to x, y is regarded as a constant, and g(y) has the role of the arbitrary
d
"constant" of integration, being the most general expression such that g(y) = .
dx 0
Then fy = Q ⇒
x3
2xy + g'(y) = 2xy ⇒ g'(y) = 0 ⇒ g(y) = c1 =? f ( X, Y) =
3 + xy2 + C1,
and the general solution of the equation is
x3
f ( x, y) = or
3 + xy + c1 = c2,
2
Example 2.24. Consider the equation 4x3 - 2x + 3y + (3x + 2y)y' = 0.
P(x,y) = 4x3 - 2x + 3y, Q(x,y) = 3x + 2y, all first-order partial derivatives of both
P and Q are continuous in the plane, which is simply connected, and Py = 3 = Qx .
Hence, the equation is exact by Theorem 2.5. A potential function f exists, and
satisfies the two conditions fx = P and fy = Q.
fx = P => f(x,y) = J P(x,y)dx =

f 4x3 -2x+3ydx=x4 -x +3xy+g(y),
2
where g(y) is an arbitrary function of y. Then fy = Q ⇒

3x + g'(y) = 3x + 2y ⇒ g'(y) = 2y ⇒ g(y) = y2 + c1
⇒ J(x,y) = x4 - x2 + 3xy + y2 + c1,
and the general solution of the equation is
I
'f(x,y) = x4 -x2 +3xy+y2 +c 1 = c2, or x4 -x2 +3xy+y2 = k.
Note that, if we begin with the condition fy = Q instead of fx = P, then the result
will be the same. Thus,
fy = Q => J(x,y) =
J Q(x,y)dy= f 2
3x+2ydy= 3xy+y +g(x),
where g(x) is an arbitrary function of x, being the most general expression such that
d
g(x) = 0. In the above integration with respect toy, x is regarded as a constant.
dy
Then fx = P =>
3
3y+g'(x)=4x -2x+3y ⇒ g'(x)=4x3 -2x ⇒ g(x)=x4 -x2 +c 1
=> f(x,y) = 3xy + y2 + x4 - x 2 + c 1,

as before.
Example 2.25. Consider the equation x2 + y2 + xyy' = 0.
As shown in Example 2.22, this equation is not exact because Py =I= Qx. Thus, any
attempt to determine a function f such that fx = P and fy = Q must fail. Proceeding
as in the last two examples,
fx = P => f(x,y) = J P(x,y)dx = J x 2 + y2 dx

3
= � + xy2 + g(y),
and fy =Q⇒
2xy + g'(y)= xy ⇒ g'(y)= -xy,
which is a contradiction because the left-hand side is independent of x, whereas the

right-hand side is not, making their equality impossible. It follows that a function
f(x,y) with fx = P and fy = Q does not exist.
ye + sin(y) +[e x +xcos(y) + 1]:� = 0, y(0)= -3.

x
P(x,y) = ye x + sin(y), Q(x,y)=ex +xcos(y) + 1, all first-order partial derivatives

of both P and Q are continuous in the plane, and Py = e x +cos(y) = Qx - Hence, the
equation is exact by Theorem 2.5.
fx = P ⇒ f(x,y) = J P(x,y)dx = J ye x + sin(y) dx = ye + x sin(y) + g(y),

x
and fy = Q ⇒
ex +xcos(y) + g'(y)=e x +xcos(y) + 1 ⇒ g'(y) = 1 ⇒ g(y) = y + c1
⇒ f(x,y)=ye x +xsin(y)+y +c1,
and the general solution is ye x + x sin(y) + y = k. Setting x = 0 and y = -3 then
gives k = -6, and the solution of the initial-value problem is ye x + x sin(y) + y = -6.
yln(x)+2x-[x- xln(x)+3y ]y'=0, x>0, y(1)=2.

2
P(x,y) = y ln(x) +2x, Q(x,y) = -x+x ln(x)-3y2 , all first-order partial derivatives of
both P and Qare continuous in the right half-plane x > 0, which is simply connected,
and Py = ln(x) = Qx - Hence, the equation is exact by Theorem 2.5.
fy = Q ⇒ J(x,y) J Q(x,y)dy = f-x +xln(x)- 3y dy

2
-xy + xy ln(x) - y3 + g(x),
and fx = P ⇒
-y + y ln(x) + y + g'(x) = y ln(x) + 2x ⇒ g'(x) = 2x ⇒ g(x) = x2 + c 1
⇒ J(x,y) = -xy + xy ln(x) - y + x + c 1,

3 2
and the general solution is xy ln(x) - xy + x2 - y3 = k. Setting x = 1 and y = 2 then

gives k = -9, and the solution of the initial-value problem is
xyln(x)- xy +x 2 - y3 = -9.
2.5.1 Integrating Factors

If the equation P(x, y) + Q(x, y)y' = 0 is not exact because Py =/- Qx , then it can be
made exact by multiplication by an integrating factor I(x, y), chosen such that the
resulting, equivalent, equation
T(x, y)P(x, y) + I(x, y)Q(x, y)y' = 0 (2.4)
is exact. The exactness of Equation (2.4) requires that (IP)y = (IQ)x - Employing
the product rule, this condition becomes
This partial differential equation for the unknown function I ( x, y) is more difficult to
solve than the given ordinary differential equation, and simplifying assumptions must
be made, which may or may not succeed.
Seek an integrating factor J(x) which is independent of y. Then Iy = 0, Ix = I'(x),

and the condition (2.5) becomes
I'(x)
(2.6)
I(x)
If the right-hand side of Equation (2.6) is independent of y, then I(x) is determined
by Equation (2.6). Otherwise, Equation (2.6) is a contradiction because the left-hand
side is independent of y. In that case, I ( x) does not exist.
If I(x) does not exist, then seek an integrating factor I(y) which is independent
of x. Then Ix = 0, Iy = J'(y), and the condition (2.5) becomes
I'(y)
(2.7)
I(y) (
If the right-hand side of Equation (2.7) is independent of x, then I(y) is determined
by Equation (2.7). Otherwise, Equation (2.7) is a contradiction because the left-hand
side is independent of x. In that case, I (y) does not exist.
Example 2.28. Consider the equation x 2 + y 2 + xyy' = 0.

As shown in Example 2.22, this equation is not exact because Py =/- Qx - Since
Py - Qx 2y -y 1
Q xy X
is independent of y, I ( x) exists and is determined by
J'(x) 1
I(x) X
⇒ ln III= ln /xi ⇒ I(x) = ±x,

where arbitrary constants of integration are unnecessary and have been suppressed,
because we seek only one solution I(x), and not all possible solutions. Since both
±x are integrating factors, we select the simpler one, I(x) = x. The equation then
becomes
x
3
+ xy2 + x2 yy' = 0,
and is exact. Then
l 1
fx = P ⇒ f(x,y) = x
3
+ xy2 dx = 4 x4 + 2 x2 y2 + g(y),
J
and fy = Q ⇒
1 1
f(x,y)= x + x y +c1,
4 2 2
2
x y+g'(y)=x y
2
⇒ I
g (y)=0 ⇒ g(y)=c 1 ⇒ 4 2
. . 1 1 2 2 k y = 4k = k1-
and the general solution 1s x 4
4 + 2 x y = �, or x + 2x
4 2 2
Example 2.29. Consider the equation
jx + y
3
+ 2xy2 y' = 0, X > 0.
P(x,y) = ..jx + y , Q(x,y) = 2xy , Py (x,y) = 3y , Qx (x,y) = 2y , Py =/= Qx, hence,
3 2 2 2
the equation is not exact. Since

2 -
Py - Qx - 3y 2y2 = �
Q 2xy2 2x
is independent of y, I ( x) exists and is determined by
I'(x)
= � ⇒ ln II(x)I = ! ln(x) = ln (x 1 12
) ⇒ I(x) = ±x 112 = ±Jx.
I(x) 2x 2
With I(x) = x 112, the equation becomes
x + x 1f2y3 + 2x3f2y2 y' = o,

and is exact. Then
fx = p ⇒ J(x,y) = J x + x y dx = 1x + �x3f2y3 + g(y),

1f 2 3 2
and fy = Q ⇒
1 2 3/2 3
2x3 f 2 y2 + g'(y) = 2x3f2y2 ⇒ g(y) = c1 ⇒ J(x,y) =
2
2
x +
3x y + c1,
. . 1 2 2
and the general solut10n 1s x + 3/2 3
y = k�, or 3x2 + 4x3/ 2 Y 3 _
- k 1-
2 3x
x2 )
+y+ ( + X + XY y1 = 0.
X
2
Py= 1, Q x = x + 1 + y, Py # Q x , hence, the equation is not exact. Since
-x-y
x2 + X + xy
is not independent of y, I(x) does not exist. Since
Qx Py x+
-
= y = l
p x+y
is independent of x, J(y) exists and is determined by

ThP equation then becomes
��? = 1, which gives J(y) = eY .
2
(x+y)e Y + (� +x+xy)eY y'=O,
and is exact. Then
and fu = Q ==>
(�x2 +xy+x)e Y +g'(y)= (�x2 +xy+x

1
)e Y ==> g(y)=c1
==> f (x, y) = ( � x2 + xy)eY + c1,
and the general solution is (�x2 + xy)e Y = k, or ( x2 + 2xy) e Y = k 1.
4xy + (x 2 + v'Y) y' = 0, y > 0, y(l) = 4.

Py = 4x, Q x = 2x, Py -I Qx , hence, the equation is not exact. An integrating factor
I (y) exists and is determined by
J'(y) -2x 1 1
==> ln II(y)I=-- ln(y) ==> I(y) =y- 112.
J(y) 4xy 2y 2
The equation then becomes
and is exact. Then
fx = 4xy112 ⇒ f(x, y) = J 4xy112 dx = 2x2y 112 + g(y),
and fv =Q⇒
x 2 y-l/ 2 + g'(y) = x 2y -l/2 + 1 ⇒ g'(y) = 1 ⇒ g(y) = y + C1
⇒ J(x, y) = 2x2y 112 + y + c1,
and the general solution is 2x 2 y 112 + y = k. Setting x = 1 and y = 4 gives k = 8, and
the solution of the initial-value problem is 2x 2 y112 + y = 8.
Exercises 2. 5
1. Find the general solution of 3x2 + y + 1 - (3y2 - x + l)y' = 0.
2. Find the general solution of xy + x + 1 + ( �x 2 + y + 1) !� = 0.
3. Find the general solution of 3x2y2 - 2x + (2x3 y + 3y2)y' = 0.

4. Solve the initial-value problem y
3
- 3x2y2 + 4x + (3xy2 - 2x
3 3
y - 4y3 ) y' = 0,
y(0) = 1.
3x 2y2 - y - 2xy3
5. Solve the initial-value problem y' = , y(2) = 1.
3x 2y2 - 2x 3 y+x
6. Solve the initial-value problem 21 + y + ( � + x)

2
�� = 0, x > 0, y > 0,
y(l) = 4.
7. Find the general solution of (6xy + eY)y' + xex + 3y2 = 0.

d
+ 4x3 + ( 2yex2+Y2 - 2y dy = 0.
. . 2+ Y2 )
8. Fmd the general solution of 2xe x
x
9. Find the general solution of ex+y + x + y + cos(x) + (ex+y + x - y) y' = 0.
x y y
10. Solve the initial-value problem 2 - 2x + ( 2 + 2y) dd = 0,
x+y+l2 x+y+l2 x
y(0) = 0.
11. Find the general solution of
2x + 2
-------- 2y - 3 dy = .
+ 1 + -------- O
x + y + 2x - 3y + 3
2 2 x + y + 2x - 3y + 3 dx
2 2
12. Consider the equation ✓x x+ y

2 2
+1+ ( ✓x y+ y
2 2
+ 1)
dy
dx
= 0.
(a) Show that Py = Qx in the region R = {(x,y): (x,y) =/= (0,0)}.

(b) Solve the equation.
(c) Is the equation exact in R?
y X
13. Consider the equation yI = O.
X2 +y 2 X2 +y 2
(a) Show that Py = Qx in the region R = {(x, y): (x, y) =/= (0, 0)}.
(b) Solve the equation.
(c) Is the equation exact in R?
14. Solve the initial-value problem x + 6y2 + 4xyy' = 0, y(2) = 0, and express y in
terms of x.
15. Find the general solution of x3 + 2y2 + xyy' = 0.
16. Find the general solution of e- x - cos(y) + sin(y )y' = 0.

17. Find the general solution of xex + xy2 + yy' = 0.
2
18. Find the general solution of ft+ y2 + 4xyy' = 0, x > 0.
19. Find the general solution of x + y2 + 3xyy' = 0.
20. Solve the initial-value problem 4xy + (6x 2 + y)y' = 0, y(l) = 1.
21. Find the general solution of sin(x) + [e-Y - cos(x)] y' = 0.
22. Find the general solution of y + (1 + 2x + xy)y' = 0. \

dy (y)
23. Find the general solution of = ;r� , y > 0.
dx x n y + 2x
Chapter 2 Exercises
1. Consider the equation xy + y2 - x2 y' = O.
(a) Solve it as a homogeneous equation.
(b) Solve it as a Bernoulli equation.
y2
2. Consider the equation - y' = 0.
X2
(a) Solve it as a separable equation.

(b) Solve it as a homogeneous equation.
(c) Solve it as a Bernoulli equation.
(d) Find an integrating factor which makes the equation exact and solve it.
(e) Find the orthogonal trajectories of the one-parameter family of curves
defined by the general solution.
3. Consider the equation x + y - xy' = 0.
(a) Solve it as a homogeneous equation.

(b) Solve it as a linear equation.
(c) Find an integrating factor which makes the equation exact and solve it.
( d) Find the orthogonal trajectories of the one-parameter family of curves
4. Consider the equation 3xy' + 2y = 0.

(a) Solve it as a separable equation.
(c) Solve it as a linear equation.
(d) Find an integrating factor which makes the equation exact and solve it.
(e) Find the orthogonal trajectories of the one-parameter family of curves
5. Find the orthogonal trajectories of the one-parameter family of curves defined

by the equation 2x - 2e-x - y2 = k.
6. Find the orthogonal trajectories of the one-parameter family of curves defined

by the equation y = 2x ln Jxl + ex.
7. The relation f(x, y) = 0 defines y as a function of x, and x as a function of y.

dy dx dy dx
Determine - and - in terms of Jx and fy, and express - in terms of -.
dX dy dX dy
CHAPTER 2 EXERCISES 39
8. Consider the equation 3x - 2y + (y - 2x)y' = 0.
(a) Solve it as an exact equation and express yin terms of x.

9. Consider the equation xex - x ln(x) + y + xy' = 0, x > 0.
(a) Solve it as an exact equation and express yin terms of x.

(b) Solve it as a linear equation.
10. (a) Show that any separable equation can be expressed as an exact equation.
(b) Solve the equation y + x ln(x) ln(y)y' = 0, x > 0, y > 0, as a separable
equation.
(c) Solve the separable equation in part (b) as an exact equation.
11. Consider an equation of the form y' = J(ax +by+ c), b-=/- 0.
(a) Show that u( x) = ax + by + c transforms the equation into one which is

separable.
(b) Solve y' = (x + y + 1)2.
1
(c) Solve y' = , x + y > 2.
Jx + y- 2
(d) Solve y' = (x + y) ln(x + y) - 1, x + y > 0.
12. Consider an equation of the form a(y) + [b(y)x - c(y)]y' = 0.
(a) Show that an integrating factor I (y) exists.

(b) Regard y as the independent variable and x as the unknown function of
y. Show that the resulting equation for x(y) is linear, employing the fact
1
that y' = - by the chain rule (see Exercise 7).
x'
13. Consider the equation y + [(2y2 + l)x - 2y]y' = 0.
(a) Find an integrating factor which makes the equation exact and solve it.
(b) Proceed as in Exercise 12(b) and solve the linear equation.
14. Consider the equation y2 + (3xy + 2)y' = 0.

(a) Find an integrating factor which makes the equation exact and solve it.
(b) Proceed as in Exercise 12(b) and solve the linear equation.
15. Solve the equation 1 + (xy - x- 3 y3 )y' = 0 by transforming it into an equation

for x as a function of y (see Exercise 12(b)).
16. Consider the equation y' = 2-jy=1".

(a) Find the general solution.
(b) By inspection of the equation, find a solution which is not a particular
solution, i.e., it does not correspond to any particular value of the arbitrary
constant c. Such a solution is called a singular solution, and may occur
only as the solution of a nonlinear equation.
(c) The envelope of a one-parameter family of curves is a curve C with the
property that, at every point of C, it is tangent to the graph of a member
of the family. The envelope of the one-parameter family of curves defined
by the general solution, if it exists, is a singular solution. However, a
singular solution is not necessarily an envelope. Confirm that the graph of
the singular solution found in part (b) is the envelope of the one-parameter
family of curves defined by the general solution found in part (a).
17. Given the general solution J(x, y, c) = 0 of a nonlinear equation, where f is

a (single-valued) function with continuous first-order partial derivatives, if a
singular solution exists, then it is obtained by the elimination of c from the two
equations f(x, y, c) = 0 and of
Be (x, y, c) = 0.
1 y2
(a) Find the general solution of the equation (y') 2 = �
y
(b) By inspection of the equation, find two singular solutions.
(c) Express the general solution as f (x, y, c) = 0 and determine the singular
solutions by elimination of c from J(x, y, c) = 0 and fc (x, y, c) = 0.
(d) Confirm that each singular solution is an envelope of the general solution.
(y') 2 + 1
18. Consider the equation = 1.
(y-xy')2
(a) Show that y = mx ± ✓m2 + 1 is the general solution, where m is an

arbitrary constant.
(b) Express the general solution in part (a) as f(x, y, m) = 0 and determine x2
as a function of m 2 from the equations f(x, y, m) = 0 and fm (x, y, m) = 0.
(c) Employ the result of part (b) to determine y2 as a function of m2.
(d) Employ the results of parts (b) and (c) to eliminate m, thereby obtaining
a relation g(x, y) = 0 between x and y, which defines a singular solution.
(e) The general solution y = mx± ✓m 2 + l of part (a) defines a one-parameter
family of straight lines of slope m and y-intercept ± ✓m2 + 1. Identify
the graph of the singular solution obtained in part (d) as a well-known
geometrical object, and confirm that it is the envelope of the family of
straight lines.
19. Peano 's existence theorem states that, if f is continuous on a rectangular region
R with sides parallel to the axes, and if (x0 , y0) is an interior point of R, then
the initial-value problem y' = f(x, y), y(x0) = y0, has a solution.
(a) Find the general solution of the equation y' = '#._ and express y as a function
X
of x.
(b) Show that the initial-value problem with the initial condition y(0) = 1 has
no solution, and give the reason.
20. Picard 's existence and uniqueness theorem states that, if f and fy are continuous
on a rectangular region R with sides parallel to the axes, and if (x0, y0) is an
interior point of R, then the initial-value problem y' = J(x, y), y(x0) = y0, has
a unique solution.
(a) Find the general solution of the equation y' = 3xy 1 1 3.

(b) Solve the initial-value problem with the initial condition y(0) = 0.
(c) Find a singular solution (which is not an envelope).
(d) Show that the solution of the initial-value problem is not unique, and give
the reason.
L
Chapter 3
Second-Order Equations
3.1 Basic Definitions

Definition 3.1. The general solution of a second-order equation is the solution which
contains two arbitrary constants which cannot be combined. It is a two-parameter
family of solutions, i.e., a set of solutions such that every solution corresponds to
particular values of the arbitrary constants. A solution in which the constants are
assigned particular values is called a particular solution.
Example 3.1. The second-order equation y" = x can be solved by two integrations:
yII = X ⇒ yI = 1 x 2 + C1
2
is the general solution since it contains two arbitrary constants c1 and c2 which cannot
be replaced by a single arbitrary constant. Particular solutions are
1 3 1 1 3
y= x , y = x3 + 2x + 1, and y = x - 4x - 3,
6 6 6
corresponding to c1 = c2 = 0, c1 = 2, c2 = 1, and c1 = -4, c2 = -3, respectively.
Definition 3.2. An initial-value problem for a second-order equation consists of the

equation, together with two initial conditions, which specify the values of y and y' at
a single point x and, thereby, determine a particular solution.
In Example 3.1, if the initial conditions y(0) = 2 and y'(0) = 3 are imposed, then
y = 6x
1 3 + C1X + C2 and y 0 = 2 ⇒ = 2 ⇒ y = 61x3 + C1X + 2 ⇒ yI = 2x
12 + C1,
() C2
1
and y' (0) = 3 ⇒ c 1 = 3 ⇒ y = -x 3 + 3x + 2
6
43
44 CHAPTER 3. SECOND-ORDER EQUATIONS
f In contrast to first-order equations, only a few types of second-order equation can

be solved in terms of elementary functions.
Definition 3.3. A second-order equation is linear if it has the form

a(x)y" + b(x)y' + c(x)y = g(x),
!�:; ,
where a(x) :/= 0, b(x), c(x) and g(x) are functions of x (or constants). Otherwise,
the equation is nonlinear. Dividing the equation by a(x) and letting p(x) =
c(x) g(x)
q(x) = and J(x) = , the equation may be placed in the standard form
a(x) a(x)
y" + p(x)y' + q(x)y = f(x).
=
If f(x) 0, then the equation y" +p(x)y' +q(x)y = 0 (or a(x)y" +b(x)y' +c(x)y = 0)
is called homogeneous. Otherwise, it is nonhomogeneous.
For example, the equation
is linear and homogeneous, with the standard form

y" + e-xy' - 2e-xy = 0.
The equation
2y" + 5xy' + y = sin(x)
is linear and nonhomogeneous, with the standard form
5x , l 1 .
y11 + -y + -y = - Slll (X ) .
2 2 2
The equations
yy" + y = x2 , y" + yy' = 0 and y" + y' + eY = x
are nonlinear.
Definition 3.4. Two functions y1 and y2 are linearly dependent on an interval I if

there exist constants c1 and c2, not both zero, such that C1Y1 + C2Y2 = 0 on I (i.e.,
c1 y1 ( x) + c2y2 ( x) = 0 for all x in I). Two functions y1 and Y2 are linearly indepen
dent on I if they are not linearly dependent on I, i.e., if c1 y1 + C2Y2 = 0 on I, then
C1 = C2 = 0.
3.1. BASIC DEFINITIONS 45
It follows by Definition 3.4 that two functions are linearly dependent on I if and
only if one of the functions is a constant multiple of the other on I: Suppose that
C1Y1 + C2 Y2 = 0.
-C2 -C1
If Ci-/=- 0, then Yi = -y2 , and if C2 -/=- 0, then Y2 = -yi.
� �
Conversely, if Yi = cy2 , then 1 · y1 - cy2 = 0 (with ci 1 -/=- 0 and c2 = -c), and if
=
y2 = cy1, then cy1 - 1 · Y2 = 0 (with c1 = c and c2 = -1 -/=- 0). Two functions are
linearly independent on J if and only if neither one is a constant multiple of the other
on J.
Example 3.2. y 1 = x and y2 = ex are linearly independent on IR because neither

one is a constant multiple of the other. If c1x + c2 ex = 0 for all x, then x = 0 ⇒
c2 = 0 ⇒ cix = 0 for all x, and x = 1 ⇒ c1 = 0.
Example 3.3. y1 = ex and y2 = e 2x are linearly independent on IR because neither

one is a constant multiple of the other. If ciex + c2 e 2x = 0 for all x, then x = 0 ⇒
Ci+ c2 = 0 =} c2 = -c1 ⇒ c1(ex - e 2x) = 0 for all x, and x = 1 ⇒ ci(e - e 2 ) = 0 =}
Ci = 0 =} C2 = 0.
More generally, Yi = e
rix
and y2 = e
rz x
are linearly independent on IR if and only if
ri -/=- r2.
Example 3.4. Yi = e 3x and Y2 = 2e 3x are linearly dependent on IR because y2 = 2yi

for all x. Equivalently, 2yi + (-l)y2 = 0 for all x, with Ci = 2 and c2 = -1.
Example 3.5. Yi = cos(x) and y2 = sin(x) are linearly independent on IR because

. is a constant multiple of the other. If c1 cos(x)
neither one
�
+ c2 sin(x) = 0 for all x,
then x = 0 ⇒ ci = 0 ⇒ c2 sm(x) = 0 for all x, and x = ⇒ C2 = 0.
.
2
The terms "dependent" and "independent" are often employed to mean "linearly
dependent" and "linearly independent," respectively. If an interval J is not specified,
then it is taken to be any interval on which y 1 and y2 are defined.
The definitions of linear dependence and independence can be extended to any

number of functions:
Definition 3.5. The functions Yi, Y2 , · · ·, Yn are linearly dependent on an interval T

if there exist constants ci, C2 , · · · , en, not all zero, such that ciyi + c2y2 + · · · + CnYn = 0
on I. The functions Yi, Y2, · · ·, Yn are linearly independent on I if they are not linearly
dependent on I, i.e., if CiYi + C2 Y2 + · · ·+ CnYn = 0 on I, then C1 = C2 = · · · = Cn = 0.
Exercises 3 .1
1. Find the general solution of the equation y" + 6x2 + sin(x) = 2.
2. Find the general solution of the equation xy" = ln(x), x > 0.
3. Solve the initial-value problem y" - 4 cos2 (x) = 2, y(0) = 0, y'(0) = 1.
4. Solve the initial-value problem y" = xex , y(0) = l, y'(0) = 1.
5. Determine whether the given equation is linear and homogeneous, linear and
nonhomogeneous, or nonlinear.
(a) xy" - y' + 2y =0

(b) 2y" + ln(x)y = sin(x)
(c) y"+y'+(x2 +y2)=0
(d) = y3
y" + xy'
(e) x2 y" + xy' = 2y
(f) y" + y' + y + X = 0
(g) y" + yy' + y = 0
6. Show that the functions y1 = er i x and y2 = er2 x are linearly dependent on JR if
and only if r1 = r2.
7. Show that the functions y 1 = x0 and y2 = xf3 are linearly dependent on I if and
only if a = /3, where I is any interval where both functions are defined.
8. Show that the functions y 1 = xr and y2 = esx are linearly dependent on I if

and only if r = 0 and s = 0, where I is any interval where both functions are
defined.
9. Show that the functions y 1, Y2, · ··, Yn are linearly dependent if and only if (at
least) one of the functions is a linear combination of the others.
10. Are the functions Y1 = x, Y2 = ex and y3 = e- x linearly independent?

11. Are the functions y1 = x, y2 = ex and y3 = 2x linearly independent?
12. Under what condition(s) is the single function y linearly independent?
3.2. LINEAR HOMOGENEOUS EQUATIONS 47
3.2 Linear Homogeneous Equations

Before methods of solution of linear, second-order equations can be discussed, certain
fundamental results must be established.
Theorem 3.1. If y1 and y2 arc any two solutions of a linear, homogeneous equation
y" + p(x )y' + q(x)y = 0, then any linear combination y = C1Y1 + C2Y2 is a solution.
Proof.
y" + p(x)y' + q(x)y (ciY1 + C2Y2) 11 + p(x)(ciyi + c2y2)' + q(x)(ciYt + c2y2)

ci[y� + p(x)y� + q(x)yi] + c2[y� + p(x)y; + q(x)y2]
0.
Theorem 3.2. The equation y" + p(x)y' + q(x)y = 0 has precisely two linearly
independent solutions y 1 and Y2, and the general solution is y = CiY1 + c2y2, where ci
and c2 are arbitrary constants.
Example 3.6. Let y1 = cos(x) and Y2 = sin(x). Then y� = -y1 ⇒ y� + Yi = 0,

and YI = -y2 ⇒ YI + Y2 = 0. Hence, Y1 and Y2 are solutions of y" + y = 0.
Since y1 and y2 are linearly independent on JR, the general solution of y" + y = 0 is
y = c1 cos(x) + c2 sin(x).
Note that, if we take Yi = cos(x) and Y2 = 2 cos(x), then, although both Y1 and Y2
are solutions of y" + y = 0, they are not linearly independent since y2 = 2yi. Then
y = C1Yi + C2Y2 = ci cos(x) + 2c2 cos(x) = (c1 + 2c2) cos(x) = kcos(x) is not the
general solution. Thus, it is essential that Yi and y2 be linearly independent solutions
in order for C1Y1 + C2 Y2 to be the general solution.
It is often possible to obtain one solution Yi of an equation easily. The method of

reduction of order, discussed below, is a method of determining a second, independent
solution y2 (i.e., a solution y2 such that Yi and y2 are independent).
Reduction of Order
Suppose that Yi is a solution of
y" + p(x)y' + q(x)y = 0,

i.e., y� +p(x)y� +q(x)y 1 = 0, and let y2 = u(x)yi, where u is to be determined. Then,
by the product rule,
Y2 = uyi ==>- Y2
I
= u Yi + UYi,
I I
Y2
fl
=u
ff
Y1 + 2' '
u Yi + UY1 ,
ff
and Y2 is a solution if and only if y� + p(x)y� + q(x)y2 = 0, i.e.,
[u"y1 + 2u'y� + uy�] + p(x)[u'y1 + uy�] + q(x)[uy1] = b.

Rearranging terms, we obtain
u[y� + p(x)y� + q(x)y1] + u"y1 + u'[2y� + p(x)y1] = 0. (3.1)

Since yJ + p(x)y� + q(x)y1 = 0, Equation (3.1) reduces to
u"y1 + u'[2y� + p(x)y1] = 0,

which is a first-order, separable equation for the unknown function v = u'. Thus,
v' 2 y�
v'y1 + v[2y� + p(x)y1] = 0 =} - = -- - p(x)
V Yi
1 e- f p(x)dx
-
=} lnlvl = -2 ln lY1I- J p(x)dx =} V =
Yi
(3.2)
where arbitrary constants of integration and± signs have been omitted, since we seek
1
only one function u(x), and not all such functions. Since 2ef p(x) dx =/= 0 (provided
Y1
that p is continuous), u(x) is not a constant and, hence, y1 and Y2 = uy1 are linearly
independent.
3.2.1 Equations with Constant Coefficients

Definition 3.6. A linear, homogeneous equation with constant coefficients has the
general form
ay" + by' + cy = 0,
where a, b and c are constants.
Seek solutions in the form y = e rx , where r is a constant to be determined. Then
and y = e rx is a solution if and only if ar2 e rx + bre rx + ce rx = 0, i.e.,
ar2 + br + c = 0, (3.3)
called the indicial equation. Employing the quadratic formula, the solutions of the
indicial equation (3.3) are given by
-b ± Jb - 4ac
r=------
2
(3.4)
2a
There are three possibilities, namely, b2 - 4ac > 0, b2 - 4ac = 0, and b2 - 4ac < 0,
and each must be analysed separately.
Case (i) If b2 - 4ac > 0, then
-b + Jb2 - 4ac -b - Jb 2 - 4ac

r1 = and r2 = ------
2a 2a
are two distinct, real roots, and y 1 = e i and y2 = e are two independent solutions.
r x r2 x
The general solution is then Y = C1er1 x + C2er2 x _
Example 3. 7. Consider the constant-coefficient equation

3y" + 7y' + 2y = 0.
Setting y = erx gives the indicial equation 3r 2 + 7r + 2 = 0, with the solutions
-7 ± J49 - 24 . 1
r= , 1.e., r1 = - and r2 = -2. Since the roots of the quadratic are
6 3
real and distinct, y 1 = r½x and y2 = e-2x are two independent solutions, and the
general solution is y = c1e-½ x + c2 e-2x _

y" - y' - 6y = 0, y(0) = 0, y'(0) = 10.
Setting y = erx gives the indicial equation r2 - r - 6 = (r - 3)(r + 2) = 0, with the
solutions r 1 = 3 and r2 = -2. Since the roots of the quadratic are real and distinct,
y 1 = e3x and Y2 = e- 2x are two independent solutions, and the general solution is
y = c1e3x + c2 e-2x . Then y(O) = 0 =} C1 + C2 = 0 =}
y = c1(e3x - e-2x ) =} y' = c1(3e3x + 2e-2x ),
and y'(0) 10 =} c1 = 2 =} y = 2(e3x - e-2x ) is the solution of the initial-value
problem.
-b
Case (ii) If b2 - 4ac = 0, then r = - is the only root, and one solution is y1 = erx _
2a
In order to determine a second, independent solution, employ reduction of order. The
standard form of the equation ay" +by'+cy = 0 is y" + �y' + <::._y = 0, with p(x) = �.
a a a
By Equation (3.2),
v = u' = �e- f p(x)dx = -1-e-!x = 1 =} u(x) = x

Yi 2rx
e
Example 3.9. Consider the constant-coefficient equation
3y" + 4v16 y' + 8y = 0.

-4\1'6 ± J96 - 96 2\1'6 .
y =e ⇒rx
3r2 + 4\1'6 r + 8 = 0 ⇒ r = ------- ⇒ r = --- 1s the only
6 3
root. Thus, two independent solutions are y1 = 2v'6
e--3-
x
and y2 = 2v'6
xe--3-x , and the
general solution is y = e- f x (c1 + c2 x).
2
4y" - 12y' + 9y = 0, y(0) = 0, y'(0) = 1.

12 ± ✓144 - 144 3
y = e
rx
⇒ 4r 2 - 12r +9 = 0 ⇒ r=- ----- ⇒ r = -
8 2
is the only root.
Note that 4r 2 - 12r + 9 = (2r - 3)2 = 0 also gives r = � as the only root. Thus,
two independent solutions are y1 e� x and y2
=;= = xe � x , and the general solution is
y = e2 x (c1 + c2 x). Then y(0) = 0
3
⇒ c1 = 0 ⇒
and y'(0) = 1 ⇒ c2 = 1 ⇒ y = xe� x is the solution of the initial-value problem.
Case (iii) If b2 - 4ac < 0, then the roots of the quadratic are complex,
-b ± Jb2 4ac -b ± iJ4ac- b2

- = ------ = a± i/3 '
-
r = ---- -
2a 2a
-b J4ac- b2
where a= - and /3 = ----. Two complex solutions are then given by
2a 2a
where z1 is the complex conjuga te of z1.
In order to extract two real, independent solutions from the complex ones, employ
the fact that any linear combination of solutions is a solution (by Theorem 3.1), and
Euler's identity, which states that, for any real 0,
e
i0
= cos(0) + isin(0).
Thus, express the complex solutions as
<o+if3) x = eaxeif3x = eax [cos(f3x) + isin(/3x)] and
z1 e
z2 e<a-if3)
x
= eax e-if3x = e ax [
cos(/3x) - isin(/3x)],
and let
+ Z2 + Zt
Zl
= Zl
= 1n
:J1. ( z 1 ) = e
QX
r:i ) and
cos( /Jx
2 2
= Zi = :::.S ({JX
°'( Z1 ) = e
Z1 - Z2 Z1 - o
)'
x ·
Sln
2i 2i
where �(z1) is the real part of z 1 and <s(z 1) its imaginary part. Then Y1 and Y2 are
two real, independent solutions, and the general solution is
Example 3.11. Consider the constant-coefficient equation
4y" + 6y' + 3y = 0.
y= e
rx
-6 ± J36 - 48
⇒ 4r2 + 6r + 3 = 0 ⇒ r = --- J3 i ⇒
--- = - -3 ± -
8 4 4
Yi = e-!, cos ( � x) and y2 = e-l• sin ( � x)

are two independent solutions, and the general solution is
y" - 6y' + 13y = 0, y(0) = 0, y'(0) = -6.

± J36 - 52
y =e ⇒ rx
r2 - 6r + 13 = 0 ⇒r=6 2
= 3 ± 2i ⇒
y = e3x [c1 cos(2x) + c2 sin(2x)]

is the general solution. Then y(0) =0 ⇒ c1 = 0 ⇒
y = c2e3x sin(2x) ·⇒ y' = c2[3e3x sin(2x) + 2e3x cos(2x)],
and y'(0) = -6 ⇒ c2 = -3 ⇒ y = -3e3x sin(2x) is the solution of the initial-value
problem.
3.2.2 Cauchy-Euler Equations

Definition 3. 7. An equation of the form
ax2y11 + bxy I + cy = 0 (3.5)
where a, b and care constants, is called a Cauchy-Euler or Euler equation.
The standard form of an Euler equation is y " + ..!!_ y , + __!:_2 y = 0. Since p(x) = ..!!_
ax ax ax
c
and q(x) = -2 are undefined at x = 0, the solution may be undefined at x = 0. Such
ax
a point is called a singular point or singularity of the equation.
For x > 0, seek solutions of the Euler equation (3.5) in the form y = xr , where r
is a constant to be determined. Then
and y = xr is a solution if and only if ar(r - l)xr + brxr + cxr = 0, i.e.,

ar(r - 1) + br + c= 0, or ar2 + (b - a)r + c= 0, (3.6) .,,.-
called the indicial equation.
The three cases where the quadratic ar2 + (b - a)r + c in Equation (3.6) has two
real, distinct roots, a repeated real root, and two complex conjugate roots must be
analysed separately.
Case (i) If the quadratic has two real, distinct roots r 1 and r2 , then two independent
solutions of the Euler equation are y1 = xr 1 and y2 = xr2, and the general solution is
y = C1Xr1 + C2Xr2 .
Case (ii) If the quadratic has a repeated real root r = a - b, which occurs if and
2a
only if (b-a)2 -4ac = 0, then one solution is y1 = xr . A second, independent solution
y2 = u(x)y 1 is obtained by reduction of order. By Equation (3.2) on page 48,
2 -
V = u' = �e-fp(x)dx = _l_e-f abx dx _e-� ln(x) = _l_e(r2 -l)ln()x = x r l
= _l2r
1
YI x2r
x 2rx 2 xr X
⇒ u = ln(x) ⇒ Y2 = xr ln(x),
and the general solution is y = c 1 xr + c2 xr ln(x) = xr [c1 + c2 ln(x)].
Case (iii) If the quadratic has two complex conjugate roots r = a± i/3, then two
complex solutions are z 1 = xa +i/3 and z2 = xa -i/3 = z 1. Since x7 = e 10(x ) = e 7 1 n(x) for
-Y
any real number 'Y, we define the quantity xi0 by x = e ( ) . Employing Euler's
iO iOln x
identity, the two complex solutions may be expressed as
z1 xa+i/3= x°' xi/3= x°' eif3ln (x) =x°' {cos[,6ln(x)]+isin[,6ln(x)]}, and

xa -i/3= x°' x-i/3 =x°' e-if3 ln(x ) =x°' {cos[,6ln(x)] -isin[,Bln(x)]}.
z2
By Theorem 3.1, since the Euler equation is linear and homogeneous, any linear
combination of solutions is a solution. Thus, two real, independent solutions are
�+� �+�
= =�(z1 )=x°' cos[,Bln(x)l and
2 2
Z1 - Z2 Z1 - Z1
= =S'(z1)=x°' sm[,Bln(x)],
2i 2i
and the general solution is y=x°' { c1 cos[,Bln(x)]+ c2 sin[,Bln(x)]}.
If x < 0, let t =-x > 0 and y(x)= z(l). Then, by the chain rule,
dy _ dz dl _ dz d2y _ d dz dt _ d2 z _ d2 z
- - ( ) ( l)
dx dl dx dl' 2
dx - dt dt dx - dt 2 - dl 2'
and the equation ax2 y" + bxy'+ cy=0 transforms into at2 z" + blz' + cz = 0, i.e., an
Euler equation for z(t) with l > 0. Since y(x)=z(l)=z(-x), the solutions y(x) for
x < 0 are obtained from the solutions z(t) by replacing l by -x. Employing the fact
that
x, x > 0
lxl= { }
-X, X <0
the solutions of an Euler equation for any x f O can be obtained from the solutions
for x > 0 by replacing x by lxl.
Example 3.13. Consider the Euler equation
2x 2 y" + xy'-3y = 0.
For x > 0, y = xr ⇒
2r(r-l)+r-3=0 ⇒ 2r2 -r-3=0 ⇒ r=

1± ✓ 1 + 24
4
3
⇒ r1 = Since r1 and r2 are real and distinct, the general solution for
2 and r2 = -1.
3 2 1
any xi- 0 is y=c1lxl / + c2 ixl- .
1
Since lxl= ±x, the term c2ixl- may be replaced
by c3x- 1, where C3= ±c2 .
X
2 /I
y + xyI -y = 0 y(l) = 0, y'(l) = -4.
For x > 0, y = xr ⇒
r ( r - l) + r - l = r2 - 1 = (r - l) (r + l) = 0 ..
⇒ r1 = 1 and r2 = -1. The general solution for any x =I= 0 is
Then y(l) = 0 ⇒ c3 + C4 = 0 ⇒ y = c3(x - x-1 ), y'(x) = c3 (1 + x-2), and y'(l) = -4

⇒ c3 = -2 ⇒ y = -2(x - x- 1) is the solution of the initial-value problem.
4x2 y" + l6xy' + 9y = 0.

For x > 0, y = xr ⇒
4r(r - 1) + 16r + 9 = 4r2 + 12r + 9 = (2r + 3) 2 = 0
2
is the only root. The general solution, for any x =I= 0, is y = lxl-3/ (c1 + c2 ln lxl).
Note that lxi- 3/2 =I= ±x- 3 1 2, and x -3/2 is not real for x < 0. Hence, the absolute
values must be retained.
x2y" - 5xy' + 9y = 0, y(l) = 0, y'(l) = 2.

For x > 0, y = xr ⇒
r(r - 1) - 5r + 9 = r 2 - 6r + 9 = (r - 3) 2 = 0 ⇒ r = 3
is the only root. The general solution for any x =I= 0 is
Then y(l) = 0 ⇒ C3 = 0 ⇒ y = c4x3 ln lxl, y' = c4 (3x2 ln lxl + x2), and y'(l) =2⇒
c4 = 2 ⇒ y = 2x3 ln lxl is the solution of the initial-value problem.
3x2 y" - 2xy' + 4y = 0.

For x > 0, y ⇒ 3r(r - 1) - 2r + 4 = 3 r2 - 5r + 4 = 0
= xr
5 ± J25 - 48
⇒ r=-----=-±-i 5 J23 .
6 6 6
⇒ y - lxl'i' [c, cos ( � ln lxl) + c, sin ( � ln lxl)]
is the general solution for any x =/= 0.
x2 y" - 5xy' + 13 y = 0, y(l) = 0, y'(l) = 6.

For x > 0, y = xr ⇒ r(r - 1) - 5r + 13 = r2 - 6r + 13 =0 ⇒
6± 3 - 52
r = j 6
2
= 3 ± 2i ⇒
y = lxl3 [c1 cos(2ln lxl) + c2 sin(2ln lxl)] = x3 [c3 cos(2ln lxl) + C4 sin(2ln lxl)]
is the general solution for any x =/= 0. Then y(l) = 0 ⇒ c3 = 0 ⇒
and y'(l) =6⇒ c4 = 3. Hence, the solution of the initial-value problem is
y = 3x3 sin(2ln lxl)-
Exercises 3.2
1 . Find the general solution of the equation y" - y' - 2y = 0.
2. Solve the initial-value problem y" + y' - 6y = 0, y(0) = 0 , y'(0) = 10, and
determine y(l) and y'(l).
3. Solve the initial-value problem y" + 5y' + 4y = 0, y(O) = 0, y'(O) = -6.
4. Find the general solution of the equation 4y" + y' - 3y = 0.

5. Find two independent solutions of the equation y" - 3y' + y = 0.
6 . Solve the initial-value problem y" - y = 0, y(0) = 1, y'(0) = 2, and determine

y(l) and y(-1).
7. Find two independent solutions of the equation 9y" - 24y' + l6y = 0.
8. Solve the initial-value problem y" - lOy' + 25y = 0, y(0) = 1, y'(0) = 1.
9. Find the general solution of the equation 4y" - 4y' + y = 0.
10. Employ two different methods to solve the equation y" = 0.
11. Find the general solution of the equation y" + y' + y = 0.
12. Solve the initial-value problem y" - 2y' + 2y = 0, y(0) = 0, y'(0) = 1.

13. Find the general solution of the equation 2y" - 4y' + 3y = 0.
14. Find the general solution of the equation y" + 4y = 0.
15. Consider the equation y" + y = 0.

(b) Solve the initial-value problem, given the initial conditions y(0) 1,
y'(0) = 0.
(c) Solve the initial-value problem, given the initial conditions y(0) 0,
y'(0) = 1.
16. Find the general solution of the equation x 2y" + 2xy' - 6y = 0.
17. Find the general solution of the equation 6x2 y" + 7xy' - 2y = 0.
18. Solve the initial-value problem x2 y" - 7xy' + l6y = 0, y(l) = 0, y'(l) = 2, and
determine y(-1) and y(e).
19. Find the general solution of the equation 9x2 y" - 3xy' + 4y = 0.
20. Employ three different methods to solve the equation x2 y" + xy' = 0.
21. Find the general solution of the equation -2x 2 y" + 2xy' - 5y = 0.
22. Solve the initial-value problem x2 y" - 3xy' + 13y = 0, y(l) = 0, y'(l) = 6.
23. Find the general solution of the equation x2 y" + xy' + 4y = 0.
3.3. LINEAR NONHOMOGENEOUS EQUATIONS 57
3.3 Linear Nonhomogeneous Equations

Theorem 3.3. The general solution of the nonhomogeneous equation
a(x)y" + b(x)y' + c(x)y = g(x) 3

( .7)
is y = YP + Yh, where yP is any particular solution, and Yh = C1Y1 + C2Y2 is the general
solution of the associated homogeneous equation
a(x)y" + b(x)y' + c(x)y = 0, (3.8)
with y 1 and y2 any two linearly independent solutions.
Proof. y = YP + Yh is a solution of Equation 3
( .7) because
a(x)y" + b(x)y' + c(x)y a(x)(yp + Yh)" + b(x)(yp + Y1i)' + c(x)(yp + Yh)
= [a(x)yi + b(x)y� + c(x)UE] + [a(x)y� + b(x)y� + c(x)yh]
= g(x) + 0 = g(x) .
Since y contains two arbitrary constants, it is the general solution of Equation (3.7).
The solutions y 1 and y2 of Equation (3.8) are called the homogeneous solutions,
and g(x) in Equation 3 ( .7) is called the nonhomogeneous term.
In view of Theorem 3.3, given a nonhomogeneous equation, it is necessary to find

a particular solution yP in order to obtain the general solution, the determination of
two independent solutions of the associated homogeneous equation already having
been discussed in Section 3.2.
3.3.1 The Method of Undetermined Coefficients

Consider a nonhomogeneous equation
ay" +by'+ cy = g(x), (3.9)
where a, band care constants, i.e., the associated homogeneous equation
ay" +by'+cy =0 (3.10)
has constant coefficients.
I\ If g(x) consists of certain, particular forms, then the form of a particular solution
yp , containing unknown constants, of Equation 3 ( .9) can be deduced. The unknown
constants in YP are then determined by forcing YP to be a solution of Equation (3.9).
This is called the method of undetermined coefficients.
If g(x) = C ef3x , where C and /3 are constants, then YP = Aef3x , provided that ef3x
is not a solution of Equation (3.10).
Example 3.19. Consider the nonhomogeneous equation
y II -y-
I x
2y= 3e.
The associated homogeneous equation is
y II-yI - 2y= 0
and y = erx =? r 2-r -2 = (r -2)(r + 1) = 0:::} y1 = e2x, Y2 = e-x, and the general
solution of the homogeneous equation is Yh = c1 e2x + c 2 e- x .
Since g(x) = 3ex and ex is not a solution of the homogeneous equation, YP = Aex .
Then y'p = Aex yp11= Aex and
)
)
3 3
Thus, YP =
- ex and y = YP + Yh = -2 ex + c1e2x + c2 e-x is the general solution of
2
the nonhomogeneous equation.
y11 + y = 4ex , y(O)= l, y'(O) = 2.

yll + y= 0,
and y = erx ⇒ r2 + l = 0 ⇒ r = ±i ⇒ y1 = cos(x), Y2 = sin(x), and the general

solution of the homogeneous equation is Y1t = c1 cos(x) + c 2 sin(x).
Since g(x) = ex , which is not a solution of the homogeneous equation, YP = Aex .

- Aex, and Yp11 + YP -
Then Yp'- Aex , Yp11 - - 4ex 1"f and on1y 1"f A- - 2ex ,
- 2 . H ence, YP-
and the general solution of the nonhomogeneous equation is
Then y(O) = 1 ⇒ c 1 =-1 , and y'(O) = 2 ⇒ c2 = 0, and y = 2ex - cos(x) is the

solution of the initial-value problem.
If ef3x is a solution of Equation (3.10), then so is Aef3x and, hence, it cannot be

a solution of Equation (3.9). In that case, YP = Axef3x, provided that xef3x is not a
solution of Equation (3.10).
y" - 4y = 8e2x .
y" - 4y = 0,
and y = erx ⇒ r2 - 4 = 0 ⇒ r = ±2 ⇒ Yi = e x, y2 = e- x, and the general solution
2 2
of the homogeneous equation is Yh = c1e 2x + c2e- 2x .
Since g(x) = 8e 2x and e 2x is a solution of the homogeneous equation but xe 2x is not,

YP = Axe2x . Then y; = A(e 2x + 2xe 2x ) = Ae 2x (l + 2x), yi = Ae 2x (4 + 4x), and
yi - 4yp = 8e 2x if and only if Ae 2x (4 + 4x) - 4Axe2x = 8e 2x, i.e., A= 2.
Thus, YP = 2xe2x and y = YP + Yh = 2xe2x + c1 e2x + c2e-2x is the general solution of

the nonhomogeneous equation.
If both ef3x and xef3x are solutions of Equation (3.10), then yP = x2 ef3x . In this
case, x2 ef3x cannot be a solution of Equation (3.10) because a linear, second-order
equation has precisely two independent solutions (Theorem 3.2).
y" - 6y ' + 9y = 4e3x .

y" - 6y ' + 9y = 0,
and y = erx ⇒ r 2 - 6r + 9 = (r - 3) 2 = 0 ::::} Yi = e 3x, Y2 = xe3x, and the general
solution of the homogeneous equation is Yh = c1 e3x + c2 xe3x .
Since g(x) = 4e 3x and both e 3x and xe 3x are solutions of the homogeneous equation,
YP = Ax2e 3x . Then y; = Ae3x (2x + 3x2 ), yi = Ae3x (2 + 12x + 9x2 ), and
YpII - 6YpI + 9YP = 4e3x

if and only if
Ac3x [(2 + 12x + 9x2 ) - 6(2:r; + 3x 2 ) + 9x2 ] = 4e3x , 1.e., A= 2.
Thus, YP = 2x2 e3x and y = YP + Yh = 2x2 e3x + cie3x + c2 xc3x is the general solution
of the nonhomogeneous equation.
The case where g(x) = Ce13x can be summarized by the following:
Theorem 3.4. A particular solution YP of the nonhomogeneous equation
ay" +by'+ cy = Cef°3x ,

where a, b, c, C and /3 arc constants, is given by
l. YP = Ae13x , provided that e13x is not a solution of the associated homogeneous
equation.
2. YP = Axe13x, if ef3x is a solution of the associated homogeneous equation and

xe13x is not.
3. YP = Ax2e/3x , if both ef3x and xef3x are solutions of the associated homogeneous
equation.
If g(x) = Ccos(f3x) + D sin(/Jx), where C, D and /3 are constants, then

YP = Acos(f3x) + B sin(/Jx),
provided that cos(/Jx) and sin(/Jx) are not solutions of Equation (3.10 ).
y" + 4y' + 4y = 6sin(3x).

y" + 4y' + 4y = 0,
and y = erx ⇒ r 2 + 4r + 4 = (r + 2)2 = 0 ⇒ Y1 = e- 2x , Y2 = xe- 2x, and the general
solution of the homogeneous equation is Yh = c 1 e- 2x + c2 xe- 2x.
Since g(x) = 6sin(3x) and sin(3x) and cos(3x) are not solutions of the associated
homogeneous equation, YP = Acos(3x) + Bsin(3x). Then
y� = -3A sin(3x) + 3B cos(3x), vi = -9A cos(3x) - 9B sin(3x),

and yi + 4y; + 4yp = 6 sin ( 3x) if and only if
[-9A cos(3x) - 9B sin(3x)] + 4[-3A sin(3x) + 3B cos(3x)]
+4[A cos(3x) + Bsin(3x)] = 6sin(3x),

1.e.'
(12B - 5A) cos(3x) - (12A + 5B) sin(3x) = 6 sin(3x),
which holds if and only if A and B satisfy the system { ��: �2/} : !6 } , with
72 30 72 30 .
the solution A =- and B =- . Thus, yP =- cos(3x) - sm(3x) and
169 169 169 169
72 ) 30 . ( )
Y = YP + Yh = - cos ( 3x + c 1 e-2x + c2 xe-2x
- sm 3x
169 169
is the general solution of the nonhomogeneous equation. Note that YP includes both
a cosine term and a sine term, even though g(x) includes no cosine term.
If cos((Jx) and sin((Jx) are solutions of Equation (3.10), then
YP = x[Acos((3x) + Bsin((3x)].
In this case, xcos((3x) and x sin((3x) cannot be solutions of Equation (3.10) because a
linear, second-order equation has precisely two independent solutions (Theorem 3.2).

y" + y = 2cos(x).
y" + y = 0,
and y = erx =} r 2 + 1 = 0 =} r = ±i =} y 1 = cos(x),
y2 = sin(x), and the general
solution of the homogeneous equation is Yh = c 1 cos(x) + c2 sin(x).
Since g(x) = 2cos(x) and cos(x), sin(x) satisfy the associated homogeneous equation,
YP = x[Acos(x) + Bsin(x)].
Then
y; = [Acos(x) + Bsin(x)] + x[-Asin(x) + Bcos(x)],

y: = 2[-Asin(x) + B cos(x)] + x[-A cos(x) - B sin(x)],
and Yi+ YP = 2cos(x) if and only if
2[-Asin(x) + Bcos(x)] = 2cos(x),
i.e., A= 0 and B = l. Thus, YP = xsin(x) and
Y= YP + Yh = X sin(X) + C1 cos(X) + C2 sin(X)
is the general solution of the nonhomogeneous equation.
The case where g(x) = C cos(/3x) + D sin(/3x) can be summarized by the following:
Theorem 3.5. A particular solution YP of the nonhomogeneous equation
ay" +by'+ c:y = Ccos(f3x) + Dsin(/3x),

where C, D and /3 are constants, is given by
l. YP = A cos(/3x) + B sin(/Jx), provided that cos(/3x) and sin(/3x) are not solutions
of the associated homogeneous equation.
2. YP = x[Acos(/3x) + Bsin(/3x)], whenever cos(/3x) and sin(/3x) are solutions of

the associated homogeneous equation.
If g(x) is a polynomial of degree n 2: 0, i.e.,
g(x) =Co+C1x + C2 x2 + · · · + Cnxn ,
then YP is a polynomial of degree n,
YP =Ao+A1x + A2 x2 + · · · + Anxn ,
provided that no term in YP satisfies Equation (3.10).
y" - Sy' + 6y = 3 + 12x.
y" - Sy'+ 6y = 0,
and y = erx ⇒ r 2 - Sr+ 6 = (r - 2)(r - 3) = 0 ⇒ Y1 = e 2x , Y2 = e 3x , and the general

solution of the homogeneous equation is Yh = c1e 2x + c2e 3x .
Since g(x) = 3 + 12x and no term in a first-degree polynomial is a solution of the

homogeneous equation, YP =A+Bx. Then y; = B, yi = 0, and
y; - sy; + 6yp = 3 + 12x if and only if - SB+ 6(A +Bx)= 3 + 12x,
13 13 .
which gives B = 2 and A=
. .
6. + 2x, and the general solut10n of the
Hence, YP
6 =
13
e 2x + c2e3x .
nonhomogeneous equat10n 1s y = YP + Yh =
6 + 2 + c1
x
---
If any term in Ao+A 1 x + A 2x2+· · ·+An xn is a solution of Equation (3.10), then
provided that no term in YP is a solution of Equation (3.10).
y"-2y' = 2-3x.
y"-2y' = 0,
and y= erx ⇒ r 2 -2r= r(r - 2) = 0 ⇒ y1 = 1, y2 = e 2x, and the general solution

of the homogeneous equation is Yh = c1 + c2 e 2x.
Since g(x) = 2 -3x and a constant is a solution of the homogeneous equation,

YP i= A+Bx, but YP = x(A +Bx)=Ax+ Bx2. Then y; =A+2Bx, y�=2B, and
y:-2y�=2-3x if and only if 2B-2(A+2Bx)=2-3x,
which gives B= l and A= -1- Hence, YP = -lx lx2

+ , and the general solution
of the nonhomogeneous equation is y = YP +Yh =-ix+ �x2 +c1 +c2 e2x .
If both Ao +A1x + A 2 x2 + · · · + A n xn and Aox+ A1x2 + A 2 x3 + · · · + An xn+l

contain terms which are solutions of Equation (3.10), then
The latter cannot contain any terms which satisfy Equation (3.10), because the only
polynomial which can satisfy a homogeneous equation with constant coefficients has
degree zero or one, and every term in the above polynomial has degree at least two.
y" = 24-36x + 12x2.
y" = 0,
and y = erx ⇒ r2 = 0 ⇒ Y1 = 1, Y2 = x, and the general solution of the homogeneous

equation is Yh = C1 + C2X.
Since g(x) = 24 - 36x + 12x2 and both a constant and x satisfy the homogeneous
equation, YP i- A+Bx+Cx2 and YP i- x(A +Bx+Cx2 ), but
YP = x2 (A+Bx+Cx2 ) = Ax2 + Bx3 + Cx4.

Then y� = 2Ax + 3Bx2 + 4Cx3, yi = 2A + 6Bx + 12Cx2 , and
y� = 24 - 36x + 12x2 if and only if 2A + 6Bx + 12Cx2 = 24 - 36x + 12x2 ,

which gives A = 12, B = -6 and C = 1, i.e., YP = 12x2 - 6x3 + x4 . The general
solution of the nonhomogcneous equation is then
Of course, this equation can be solved much more easily simply by two integrations:
y" = 24 - 36x + 12x2 ⇒ y' = 24x - 18x2 + 4x3 + k1
as above.
The case where g(x) = C0 + C1x + C2 x2 + · · · + Cn xn can be summarized by the

following:
Theorem 3.6. A particular solution of the nonhomogeneous equation
ay" + by'+cy = Co + C1x + C2 x2 + · · · + Cn xn

is given by
l. YP = Ao + A1x + A2 x2 + · · · + An xn , provided that no term in YP is a solution

of the associated homogeneous equation.
2. YP = x(Ao + A1x + A 2 x2 + · · · + An xn ) = Aox + A1x2 + A2 x3 + · · · + An xn+l

if Ao+A 1 x + A2 x2 + · · · + An xn contains a term which satisfies the associated
homogeneous equation and x(Ao + A1 x + A2 x2 + · · · + A n xn ) does not.
3. YP = x2 (Ao + A1x + A2 x2 + · · · + An xn ) = Aox2 + A1x3 + A2x4 + · · · + An xn+2

if both Ao+ A1x + A2x2 + · · · + An xn and x(Ao + A1x + A2 x2 + · · · + An xn )
contain terms which satisfy the associated homogeneous equation.
Theorem 3. 7. If YPi and yp2 are particular solutions of

ay" + by' + cy = 91(x) and ay" + by' + cy = 92(x),
respectively, then YP = YPi + yp2 is a particular solution of
ay" +by'+ cy = 91 (x) + 92(x).
More generally, if YP; is a particular solution of
ay" +by'+cy = 9i(x)

for 1 � i � n, then YP = YP i + yp2 + · · · + YP is a particular solution of
n
ay" +by'+cy = 91(x) + 92(x) + · · · + 9n(x).

Proof. For n = 2,
b
+ Yp + cyp a(YP1 + YP2 ) + b(YP1 + YP2 )
11 1
+ c(YP1 + YP2 )
II I
ayP
(ay:1 + by� 1 + cyp1 ) + (ay:2 + by�2 + cyp2)
91(x) + 92(x).
The general case is proved similarly.
Example 3.28. Consider the nonhomogencous equation

y" + 4y = 4x2 - 5ex .
= 0, y" + 4y
and y = erx ⇒ r2 + 4 = 0 ⇒ r = ±2i ⇒ y1 = cos(2x), y2 = sin(2x), and the general
solution of the homogeneous equation is Yh = c 1 cos(2x) + c 2 sin(2x).
For y" + 4y = 4x2, YP i =A+ Bx+Cx2. Then y� 1 = B + 2Cx, yi1 = 2C, and
y:1 + 4yp 1 = 4x2 if and only if 2C + 4(A +Bx+Cx ) = 4x ,
2 2
1 1
which gives C = 1, B = 0, and A = - . Hence, Yp 1 = - + x2 .
2 2
For y" + 4y = -5ex ' yP2 ' = De ' y"P2 = De ' and
= Dex . Then yP2 x x
y:2 + 4yp2 = -5ex if and only if Dex +4Dex = -5ex ,
which gives D = -1. Hence, yp2 = -ex , and YP = YPi + yp2 = -� + x2 - ex , and
Y = YP + Yh = - 1 + x2 - ex +c1 cos(2x) + c2 sin(2x)

2
If g(x) consists of products such as

P(x)ef3x , P(x)[Ccos(,x) + Dsin(,x)], e-8x [Ccos(,x) + Dsin(,x)],
or
P(x)elh [Ccos(,x) + Dsin(,x)],
where P is a polynomial and C, D, /3 and I are constants, then YP must have the
corresponding forms
Q(x)ef3 , Q(x) cos(,x) + R(x) sin(,x), ef3 [A cos(1x) + Bsin(,x)],

x x
or
ef3x [Q(x) cos(,x) + R(x) sin(,x)],
multiplied by x or x2 if necessary, in order that no term in YP be a solution of the
associated homogeneous equation, where Q and R are polynomials of the same degree
as that of P.
y" - 2y' + 2y = ex cos(x).

y" - 2y' + 2y = 0,
and y = erx ⇒ r2 - 2r + 2 = 0::::} r = 1 ± i::::} y1 = ex cos(x), Y2 = ex sin(x), and the
general solution of the homogeneous equation is Yh = c1ex cos(x) + c2 ex sin(x).
Since g(x) = ex cos(x) and ex cos(x) is a solution of the associated homogeneous

equation, YP = xex [A cos(x) + Esin(x)]. Then
y� = ex [A cos(x) + Bsin(x)] + xex [(A + B) cos(x) + (B - A) sin(x)],

= ex [2(A + B) cos(x) + 2(B - A) sin(x)] + xex [2B cos(x) - 2A sin(x)],
y�
and yi - 2y� + 2y = ex cos(x) if and only if

p
ex [2Bcos(x)-2A sin(x)] = ex cos(x),
which gives A= 0 and B = t- Hence, YP = txex sin(x), and
y = YP + Yh = �xex sin(x) + c1ex cos(x) + c2 ex sin(x)

A more efficient method for the determination of particular solutions in examples

such as Example 3.29 will be discussed next.
3.3.2 Variation of Parameters

Consider a nonhomogeneous equation
+ p(x)y' + q(x)y = J(x)

11
y (3.11)
in standard form. The associated homogeneous equation is
+ p(x)y' + q(x)y = 0.
11
y (3.12)
Let y1 and y2 be any two linearly independent solutions of Equation (3.12). Seek a
particular solution of Equation (3.11) in the form
(3.13)
where u 1 (x) and u2 (x) are functions to be determined by forcing YP to be a solution

of Equation (3.11). Then, by the product rule,
(3.14)
The requirement that YP be a solution of Equation (3.11) imposes one constraint upon
the two functions u1 and u2. We may therefore impose a second constraint, chosen
in order to simplify the procedure. Thus, impose the condition
(3.15)
Then Equation (3.14) reduces to

Yp = U1Y1 + U1Y1 + U2Y2 + U2Y2·
II I I II I I 11
Hence, y� + p(x)y; + q(x)yp = f(x) if and only if

[u�y� + U1Y� + u;y� + U2Y�] + p(x)[u1y� + U2Y�] + q(x)[u1y1 + U2Y2] = f(x).
Rearranging terms, the latter condition becomes
Since Y1 and y2 are solutions of Equation (3.12),
y� + p(x)y� + q(x)y1 = 0 and y; + p(x)y� + q(x)y2 = 0,

and condition (3.16) reduces to
(3.17)
Combined with condition (3.15), u� and u; must satisfy the system
YiU'i + Y2U2 = 0
I
(3.18)
y� u� + y� u; = f ( x)
of algebraic equations. Employing Cramer's rule, we obtain
- 0 Y2
, I
I J(x) y; yif(x)
u, � I �; �; I
The quantity
I Yi Y2 I = YiY2t - Y1Y2
t
y� y�
is denoted by W[y1, Y2](x) or simply by W(x) and called the Wronskian of y1 and y2.
Thus,
-yd(x) yif(x)
ui =J dx, u2 =J dx, and YP = uiyi + U2Y2-
W(x) W(x)
This method for the determination of a particular solution of Equation (3.11) is called
variation of parameters or variation of constants.
Note that the above formulas for u i and u2 are defined provided that W(x) i= 0.
This is guaranteed by the following:
Theorem 3.8. If p and q are continuous on an interval I, and Yi and y2 are solutions
=
of y" + p(x)y' + q(x)y = 0, then either W 0 on I or W(x) i= 0 for any x in J. The
solutions Yi and y2 are linearly independent on I if and only if W(x) i= 0 on I. (See
Chapter 3 Exercises, Exercise 15, on page 85.)
y - 3yI+ 2y=e.
ff 2x
y" - 3y' + 2y = 0,
and y = e rx ⇒ r2 - 3r + 2 = (r - l)(r - 2) = 0 ⇒ Yi = e , Y2 = e , and the
x 2x
general solution of the homogeneous equation is Yh = Ci e + c2 e . The equation is

x 2x
in standard form with f(x) = e 2x, and the Wronskian of Yi and Y2 is
Yi Y2 ex e 2x = 3x
w (X) =
I Yi Y 2
I I
I I
= ex 2 2x
e e .
I
Hence,
2x.e2x
J () X =J-e 3 x dX =!-exdX
- y 2](x)d
=-e,x
WX e
Y1 J ( x)d e . e2x d
x
d
J () X =J x X = J 1 X = X,
Wx e 3
xe2x - e2x + C1ex + C2e2x

xe2x + c1ex + (c 2- l)ex2
Xe2x + C1ex + C3e2x
Note that the term -e2x = -y2 has been absorbed into the term c2e2x in the general
solution above. Thus, any term in YP which satisfies the homogeneous equation may
be removed from YP · In fact, a particular solution of a nonhomogeneous equation is
not unique because, if YP is a particular solution, then so is YP + Yi, where Yi is any
solution of the associated homogeneous equation. For this reason, arbitrary constants
of integration k 1 and k2 need not be added to u 1 and u 2, since this would merely add
the term k1Y1 + k2 Y2 to yp, and these may be removed.
4y" - y = 4e½ x
, y (0) = 0, y' (0) = 3.
The associated homogeneous equation, 4y "- y = 0, has the two independent solutions
x x
Y1 = e½ and Y2 = e-½, with the Wronskian W(x) = -1. The standard form of the
. . yII - 1 y = e2lx . Hence, · x) = e2I x ,
non l1omogcncous equat10n 1s f(
4
and YP = xe2x - e2 x , or YP = xe2x, since -e2x = -y1 satisfies the homogeneous

I 1 I l
equation and may be removed from YP · The general solution of the nonhomogeneous
equat10n 1s y = YP + Yh = xe2x + c1e2 x + c2e-2 x . Then y( 0 ) = 0 ⇒ c2 = -c1 ⇒
. . l l I
and y'(O) = 3 ⇒ c1 = 2. Hence, the solution of the initial-value problem is

y" - 2y ' + 2y = ex cos(x),

which has been solved in Example 3.29 by the method of undetermined coefficients.
Here, we shall solve it by variation of parameters. Two independent solutions of the
associated homogeneous equation were found to be y1 = ex cos(x) and y2 = ex sin(x).
With f(x) = ex cos(x) and
Y1 Y2 I=I ex cos(x) ex sin(x)

I = e2x '
I
W(x) = y � y� e cos(x) - e sin(x) e sin(x)+ ex cos(x)
x x x
u1 - j sin(x) cos(x) dx = -� j sin(2x) dx = i cos(2x),
u2 j cos 2(x) dx = � j l + cos(2x) dx = }x+ 1 sin(2x),
and
YP }cos(2x)[ex cos(x)]+ [ix+}sin(2x)] [ex sin(x)]
}xex sin(x)+ }ex [cos(2x) cos(x)+ sin(2x) sin(x)]

1 x 1
xe sin(x)+ :::tx cos(x).
2
1 1
Since ex cos(x) =
satisfies the homogeneous equation, it suffices to take as a
4
1
4 y1
particular solution YP = xex sin(x), in agreement with the result of Example 3.29.
2

2ln(x)
+ 6y, + 9y = xe3x , > 0.
11
y X
The associated homogeneous equation has the two independent solutions y 1 = e- 3x

21 x
and y2 = xe-3x , with f(x) = n(
e3x
) and
x
I =e -6x
Then, by integration by parts and by substitution,

xe-3x 2ln(x)
-J -6- dx = -2 J ln(x) dx = -2x ln(x)+ 2x,
e- x xe3x
e-3x 2ln(x) ln(x)
J 6 dx = 2f dx = [ln(x)] 2,
e- x xe3 x X
and
since 2xe- 3x = 2y2 . Thus,
y = YP + Yh = xe-3x [ln(x)] 2 - 2xe-3x ln(x) + c1e-3x + c2 xe- 3x

is the general solution of the nonhomogcneous equation.
x2y" - 6xy' + 12y = x5 ex .

The associated homogeneous equation is the Euler equation
x2y" - 6xy' + 12y = 0

and, for x > 0, y = xr ⇒ r 2 - 7r + 12 = (r - 3)(r - 4) = 0 ⇒ Y1 = x3 and Y2 = x4.
Since lxl3 = ±x3 and lxl4 = x4 , these solutions are defined for all real x, with the
I x3 x4
Wronskian
I
6
W (x) = 3x2 4x3 = x .
6 I 12
- -y + 2 y = x3 e
• II X
The nonhomogcneous equation in standard form 1s y Thus,
X X
f(x) =x e . By integration by parts,
3 x
and YP = (ex - xex)x3 + exx4 = x3 ex . Hence, y = YP + Yh = x3ex + c1x3 + c2x4 is the

general solution of the nonhomogeneous equation.
+ 3y = 4x5 ex
2
x2y" - 3xy' .
x2y" - 3xy' + 3y = 0
and, for x > 0, y = x r ⇒ r2 - 4r + 3 = (r - l)(r - 3) = 0 ⇒ Y1 = x and Y2 = x3 .
Since lxl = ±x and lxl3 = ±x3 , these solutions are defined for all real x, with the
Wronskian
x x3
I
W (x) = 1 3x2 = 2x .
I
3
3
The standard form of the nonhomogeneous equation is y " - � y' + = 4x3 ex , with
2
y
X X2
f (x
)
= 4x3 e x2 . Thus, by integration by parts,
J x . 4x e
dx
J x2 · 2 xe x dx
3 3 x
- ---
2
2x3 -
=- 2
- x 2e x
2
+
J 2 xe x2 dx = - x 2ex2 + ex ,
2
J 4x e
J
3 x2
X .
dx = 2 xe
x2
dx = ex2
2x3
and YP = xe x\1 - x2) + x3 ex = xex . The general solution of the nonhomogeneous

2 2
equation is then y = YP + Yh = xe x + c 1x + c2x3.

2
x
2
y" - 3xy ' + 5y = x2, x > 0.
x y" - 3xy ' + 5y = 0,

2
and y = xr ⇒ r 2 4r + 5 = 0 ⇒ r
4±\/-4 = 2 ± i
= --- ⇒ y1 = x2 cos[ln(x)] and
2
-
y2 = x2 sin[ln(x)], with the Wronskian

x2 cos[ln(x)] x2 sin[ln(x)]
- ·
( )-
I 2xcos[ln(x)] - xsin[ln(x)] 2xsin[ln x)] + xcos[ln(x)] I-
3
x x
W - (
5
The nonhomogeneous equation in standard form is y" - �y' + 2 = 1, and f(x) = 1.
X X
du 1
By integration by substitution with u = ln(x) and - = -,
-! x2 sin[ln(x)] d =
x
-!
sin[ln(x)]
dx = cos[ln(x)],
dX X
x 3 X
J x cos[ln(x)]
dx = cos[ln(x)]
J
dx = sin[ln(x)],
2
x 3 X
and YP =x 2
cos2[ln(x)] + x2 sin2[ln(x)] = x2. Hence,
y = YP + Yh = x2 {1 + c1 cos[ln(x)] + c2 sin[ln(x)]}
x 2 y" - xy' + y = 27x4 ln(x), x > 0, y(l) = 3, y'(l) = -3.

x 2 y" - xy' + y = 0,
and y = xr ⇒ r 2 - 2r + 1 = (r - 1)2 = 0 ⇒ Y1 =x and Y2 = x ln(x), with the
Wronskian
x =
I x x ln(x) I = x.
W( ) 1 ln(x) + 1
1 1
The nonhomogeneous equation in standard form is y" - -y' + 2y = 27x2 ln(x), and
X X
f(x) = 27x 2 ln(x). By integration by parts,
u1 - J x ln(x). 7x 2 ln(x)
! dx =- J 27x 2 [ln(x)] 2 dx
-9x3 [ln(x)]2 + J 18x 2 ln(x) dx
-9x3 [ln(x)]2 + 6x3 ln(x) - J 6x2 dx
-9x 3 [ln(x)]2 + 6x 3 ln(x) - 2x 3,
U2 J
x · 27x 2 ln(x)
dx = 27x 2 ln(x) dx J
J
x
9x3 ln(x) - 9x2 dx
9x3 ln(x) - 3x3,
and
YP { -9x [ln(x)] + 6x ln(x) - 2x }x + [9x ln(x) - 3x ]x ln(x)

3 2 3 3 3 3
3x4 ln(x) - 2x4 .
Hence, the general solution of the nonhomogeneous equation is
Then y(l) = 3 ⇒ c1 = 5, y' = 12x 3 ln(x) - 5x 3 + 5 + c2 [ln(x) + 1], and y'(l) = -3 ⇒

c2 = -3. Hence,
y = 3x4 ln(x) - 2x4 + 5x - 3x ln(x)
4x2 y" + 4xy' - y = x.
4x2 y" + 4xy' - y = 0
and, for x > 0, y = xr ⇒ 4r(r - 1) + 4r - 1 = 4r2 - 1 = O ⇒ r = ±½ ⇒ y1 = x 1 12

and y2 = x- 1/2, with the Wronskian
The nonhomogeneous equation in standard form is y " + � y ' - -;

X �
=�
�
and, hence,
x 1
J( ) = x. Then
4
- J x-1/2 .
-x- 1
J
4� dx = ! x-1/2 dx = !x1/2 '
4 2
J J
1 f 2 · .l_
x 1 1
___4�x dx = -- x1/2 dx = --x3/2 '
-x- 1 4 6
and
1 1/2 . 1/2 _ -1 3/2 1
YP - =
x x x . x-1/2 = -x.
2 6 3
Since YP is a particular solution of the nonhomogeneous equation for any real x, the
general solution for any x =I= 0 is
The absolute values are required only in Yh·
Exercises 3.3
In Exercises 1-17, solve the equation or the initial-value problem by the method of
undetermined coefficients.
1. y" + 4y' = -sex

2. y" - y = 6e2x, y(O) = 2, y'(O) = 4
3. y" - 5y' + 6y = 4e-2x

4. y" + 3y' - 4y = -lOex
5. y" + 6y' + 9y = 6e- 3x
6. 6y"+y' - 2y=65sin(x)
7. y"+2y' +y=50cos(3x), y(0) = 0, y'(0) = 2
8. y"+y=4sin(x)
9. y"+4y = 8 cos(2x)+12sin(2x)
10. y"+4y' +4y = 4sin(x) - 3 cos(x)
11. y" - y = 4
12. y" - y' - 12y=3+12x
13. y" + 2y' + 3y = -9x
14. y" - 9y = 2 - 3x + x2
15. y"+y' =1+x, y(0)=1, y'(0) = 1
16. y" - y' = 4x3
17. y" - y=-2x+4ex +2sin(x), y(0)=0, y'(0)=9
In Exercises 18-37, solve the equation or the initial-value problem by variation of

parameters .
18. y"+y' - 2y=9ex
19. y" - 4y' +4y=6xe2x
20 . y"+y' = 1, y(0)=2, y'(0)=4
21. y"+9y=6sin(3x)
22. y"+y = sec(x)
23. y"+6y' +13y =l6e- 3x cos(2x)

9e 2x
24. y
II
- 4y' +13y =
COS(3X )
25. y" - 2y' +y = ex ln ( x), x > 0
26. y"+3y' +2y =sin(ex )
27. x2 y" - 6y=10x5

28. x2 y" - 4xy' + 6y = x 4 ex

29. x2y" + 5xy' + 3y = 4xex
2
, y(l) = e, y'(l) =e
30. x2 y" - 2xy' + 2y = x3 cos(x)
31. x2 y" + xy' - y = x2 sin(x)
ln(x)
32. x2 y" + 3xy' + y = - ,- x> 0
X
33. x2y" - 3xy' + 4y = x2 ln(x), x>0
34. x2y" - 3xy' + l3y = 9x 2

4
35. x2y" + 7xy' + l3y = x>0
x 3 cos2 [2ln(x)]
36. x2y" - 5xy' + l3y = 8x3 sin[2ln(x)], x>0
37. 4x2 y" + 4xy' - y = x2
3.4 Equations Reducible to First-Order Equations

Consider a second-order equation in which the dependent variable y does not appear
explicitly. For example,
y'y" + x2 = 0.
Let y'(x) = z(x). Then y" = z', and the equation becomes zz' + x2 = 0, which is of
the first order. If the first-order equation can be solved for z, then y is obtained by
y = j y' dx = j z dx.
The general solution z of the first-order equation includes one arbitrary constant,
and its integral y contains a second arbitrary constant, and is therefore the general
solution of the second-order equation.
y" l
--=-+x.
y' + l X
Since y does not appear explicitly, let z(x) = y'(x) to obtain the first-order equation
z' l
--=-+x
z+ l X
3.4. EQUATIONS REDUCIBLE TO FIRST-ORDER EQUATIONS 77
which is separable. Then
1 1 2
J--dz=J!+xdx
Z + 1
⇒ ln lz+ll=lnlxl+- x +c1
2
J J
X
⇒ z + 1= k1xe x
212
⇒ y= zdx = k1xe x
212
- 1 dx = k1ex ! 2
2
- X + C2
x./xy" =./xy' + 2x,/yi, x > 0.
Since y does not appear explicitly, let z(x) =y'(x) to obtain the first-order equation
xvx z' = VX z + 2xyz,

z
which is homogeneous (also a Bernoulli equation). Then u = -
X
⇒
z 1 1
u+xu' =z'= -+ 2 �-= u+ 2--./u
X X
⇒ 2y r,;uY' = -
X
⇒ VU= ln(x)+c1
⇒ u= [ln(x)+c1] 2 ⇒ z = x[ln(x) + c1] 2 ⇒ y= J zdx
and, by integration by parts,

-1
y" =-----2 - 1
(x + y' + 1)
Since y does not appear explicitly, let z(x) =y'(x) to obtain the first-order equation
-1
z'=-----1,
(x + z + 1) 2
which has the form z' = f(x + z + 1). By Exercise 11 in the Chapter 2 Exercises on
page 39, let u = x + z + l to obtain
1
uI = 1 +zI = -
u2
⇒ u = (k1 - 3x) 1 13 ⇒ z = (k1 - 3x) 1 13 - x-1
⇒ y =
J z dx =
J ( k1 - 3x) 1/3 - x - l dx = - 1 ( k1 - 3x)4/3 - 21 x2 - x + c2
4
y'y" = x.
Since y does not appear explicitly, let z(x) = y'(x) to obtain the first-order equation
zz' = x,
The value of the last integral depends upon the sign of k. If k = 0, then
If k > 0, let k = µ2 and make the trigonometric substitution
dx ✓x2 + µ2 = µsec(t),
2,
1r 1r
x = µ tan(t), - <t< dt = µsec2 (t),
2
J✓ J
to obtain
y=± x 2 + µ2 dx = ±µ2 sec3(t) dl.
By integration by parts,
= j sec(t) · sec (t)2

dl
sec(t) tan(l) - J tan(t) · sec(l) tan(t) dl
sec(l) tan(l) - j sec(t) tan (t) 2

dt
sec(t) tan(t) - J sec(l)[sec2 (t) - 1] dt
sec(t) tan(t) - j sec (t) + j sec(t)

3
dl dt
⇒ 2 j sec (t)
3
dl scc(t) tan(l) + ln \ sec(t) + tan(l)\ + C3
⇒ j sec (l)
3
dt
1
2 [ sec(t) tan(l) + ln I sec(t) + tan(t)\] +c4.
l �--
Hence, employing tan(t) = - and sec(t) =-✓x 2 +µ2 ,
X
l
µ µ
y=
µ2 x x 2 +µ2
±- [ + ln -'--
✓
-
x 2 +µ2 x
-- +- +c5.
✓
2 2
µ µ µ
If k < 0, let k = -µ2 and make the trigonometric substitution
x = µsec(t), 0�l<
w
or w � t <
3w dx
= µsec(t) tan(t), ✓x 2 - µ2 = µ tan(t),
2 2, di
to obtain
y ±j ✓x -µ = ±µ j sec(t) tan (t) = ±µ j sec (t) - sec(t)

2 2 dx 2 2
dt 2 3
dt
±µ { t [ sec(t) tan(t) + ln \ sec(t) + tan(t)I] -ln \ sec(t) + tan(t)I} +

2
ct;
µ2
±
2 [ sec(t) tan(l) -ln I sec(l) + tan(t)I] +c6.
l �--
Hence, employmg sec(t) = - and tan(t) =-✓x 2 -µ2 ,
.
l
X
/1, µ
µ2 x ✓x 2 -µ2
y=±- [-----ln
x x 2 -µ2
-+---
✓ +c5.
2 µ2 µ µ
Consider a second-order equation in which the independent variable x does not

appear explicitly. For example,
y'y" + y3 = 0.
Let y'(x) = v (y). Then, by the chain rule,
dv dy ,
y"(x) = vv
= dy dx '
and the equation becomes v2 v' + y3 = 0, which is of the first order. The general
solution v(y) of the latter equation contains one arbitrary constant. Once v(y) is
found, y' = v(y) is a separable equation, and
y' = v(y) => �

vt )
=1 => / ty dy
v )
= J l dx = x + c,
yielding a second arbitrary constant and, therefore, the general solution of the second
order equation.
y" = (y') 2 ( 2y - t)
Since x does not appear explicitly, let v(y) = y'(x) to obtain the first-order equation
vv' = v2 ( 2y - t)
which is separable. One solution is v = 0 and gives y' = 0, i.e., y = c 1. If v #- 0, then
=> J 2
ye-Y dy = J k 1 dx => -1e-y
2
= k1x + k2
=> -y2 = ln(k3x + k4) => y= ± ✓- ln(k3x + k4)

y" + (1 + 2y)(y') 3 = 0.
vv' + (1 + 2y)v3 = 0,
which is separable. One solution is v = 0 and gives y' = 0, i.e., y = c1. If v i- 0, then
v'
-- = 1 + 2y
v2
=} !V = y + y2 + C2 =} (y + y2 + C2)y' =1
1 2 1
⇒ J y + y2 + c 2 dy = J 1 dx =}
2,Y
+ 3Y3 + C2Y = X + C3
yy" + 3y2(y')3 + (y')2 = 0.

yvv' + 3y2 v3 + v2 = 0.
One solution is v = 0 and gives y' = 0, i.e., y = c 1. If vi- 0, then
yv' + v = -3y2v2
is a Bernoulli equation with a= 2. Hence, u = v- 1 ⇒ v = u- 1 ⇒ v' = -u- 2 u', and
the equation for v transforms into
-yu- 2 u' + u- 1 = -3y2u-2, or - yu' + u = -3y2,
which is linear. Its standard form is
1
uI - -u = 3y,
y
with the integrating factor J(y) ef -t dy = e- In IYI ±y- 1 and, with

J(y) = y- 1, the equation becomes
1
(-y u)
1
!u' - -2:.._u =3 1.e., = 3.

y y2
Hence,
'* j 3y2 + c2y dy = j 1 dx ⇒ y3 + �c2y2 = x + c3

y" + (y') 2 = y'. (3.19)
Since x does not appear explicitly, let v (y) = y' (x) to obtain the first-order equation
vv' + v2 = v,
which is separable. One solution is v = 0 and gives y' = 0, i.e., y = c1 . If v =/- 0, then
v' + v = l => -v'

- = 1 => - ln\ 1-v \ = y + c2
1-v
---=
y' 1
1-ke-Y
=> J �
1- e Y
_ dy = J l dx => j __!!____ dy = x +
eY - k
c3
Note that y also does not appear in Equation (3.19). It may therefore also be solved
by letting z(x) = y'(x). The equation then becomes
z' + z 2 = z,
1
_z_'_ = 1 => (- - + �) z' = 1 => -ln\1-z\ +lnlz\ = x+c1
z- z 2 l - z z
z z kex
=> ln 1- -l =x + => _ _ = kex
=> z =
1- z 1- z kex + l
C1
is the general solution, and is equivalent to the solution obtained earlier because
Exercises 3. 4
Find the general solution of the given equation.
y" 2
1. --=-+3x2
y' -2 X
2. X2 yll = (y')2
3. y" - x(y')3 = 0
4. y'y" + X = 0
5. xy" = y'ln(y') - y'ln(x), x > 0
6. y" + 2y' = 2e-x .,/iJ
7. x4 + (y') 2 - xy'y" = 0
8. y" = (2x - y' + 3)2 + 2
9. y" + (2 - 6y + 12y2 )(y')3 = 0
10. y" = (y')2 tan(y)
11. yy" = y'(y' -1) ln(y' - 1)
12. sin(y)y" + cos(y)(y')2 = (y')3
13. y" = 2yy'
14. y" -(y')2 = 1
15. yT+y'y" = 1
16. eY' y" = 1
17. cos(y')y" = 1
18. y" = y' ✓1 - (y') 2

Chapter 3 Exercises
1. Consider the equation x2 y" + xy' + (x2 - } ) y = 0, x > 0. Given that one
. . cos(x)
so1ut10n 1s Y1 = , find a second, independent solution y2 by reduction of
VX
order.
2. Consider the equation y" + xy' + y = 0. Given that one solution is y1 = e-x / 2 ,
2
find a second, independent solution y2 by reduction of order. Note that y2

includes an integral which cannot be evaluated in terms of elementary functions.
3. Consider the equation 2x2 y" + (2x2 + x)y' - y = 0, x > 0. Given that one
solution is y1 = x-1/ 2 e-x, find a second, independent solution y2 by reduction
of order. Note that Y2 includes an integral which cannot be evaluated in terms
of elementary functions.
4. Consider the equation y" + y = cos(x).

(a) Solve the equation by the method of undetermined coefficients.
(b) Solve the equation by variation of parameters.
5. Consider the equation y" - y' - 2y = 8x2 .

6. Consider the equation y" - y = cos(x) - 5 sin(2x).

7. Consider the equation y" - 2y' - 8y = 24(x2 + l)e4x .

8. Consider the equation a(x - x0) 2y" + b(x - x0)y' + cy = 0, where a, b, c and Xo
are constants.
(a) Show that the change of variables t = x - x0 and y(x) = z(t) transforms
the equation into an Euler equation for z(t).
(b) Solve the equation 2(x - 3) 2 y" + 3(x - 3)y' + y = 0 for x-=/= 3.
9. Consider the Euler equation ax2 y" + bxy' + cy = 0, x > 0.

(a) Show that the change of variables t = ln(x) and y(x) = z(t) transforms the
Euler equation for y(x) into an equation for z ( t) with constant coefficients.
(b) Show that the indicial equation for z = e rl is identical to the one for y = xr.
(c) Express the three cases of solution z = c 1 e rit + c2er2l , z = erl (c1 + c2t) and
z = e°'t [cL cos(,6l) + c2 sin(,6t)] in terms of x to obtain the three cases of
solution for y(x).
10. Consider the equation x2 y" - xy' + y = � , x > 0.

ln x)
(a) Solve the equation by variation of parameters.

(b) Let l = ln(x) and y(x) = z(l) to transform the equation for y(x) into an
equation for z(l) (see Exercise 9).
(c) Solve the equation obtained in part (b) for z(t).
(d) Obtain y(x) from z(t) and compare with the result of part (a).
11. Consider the equation 4x2 y" + 4xy' - y = x2 lxl 1 1 2.

(a) Solve the equation for x> 0.
(b) Solve the equation for x < 0. (You will need to employ the fact that
lxl = -x for x < 0, and that �(-x)°' = -a(-x)°'-1, by the chain rule.)
d
(c) Combine the solutions in parts (a) and (b) into a solution which is valid
for any x i= 0.
12. Consider the equation y" + y' = 2x.

(c) Solve the equation by letting z(x) = y' ( x).
13. Consider the equation x2y" + xy' = l, x > 0.
(a) Solve the equation by variation of parameters.

(b) Solve the equation by letting z(x) = y'(x).
14. Consider the equation 3y'y" = l.
(a) Solve the equation by letting z(x) = y'(x).

(b) Solve the equation by letting v(y) = y'(x).
15. Suppose that the functions p and q are continuous on an interval I and let y 1
and Y2 be solutions of the equation y" + p(x)y' + q(x)y = 0.
(a) Show that the Wronskian W(x) = W[y1, y2](x) = Y1Y; - y�y2 satisfies the
first-order equation W'(x) + p(x)W(x) = 0. (Hint: Multiply the equation
y� + p(x)y; + q(x)Y2 = 0 by Y1, y� + p(x)y� + q(x)Y1 = 0 by Y2, and subtract
the latter from the former.)
(b) Solve the first-order equation in part (a) for W(x).
(c) Deduce from the result of part (b) that either W(x) = 0 on I, or W(x) =/= 0
for any x in I.
(d) Prove that y 1 and y2 are linearly dependent on I if and only if W (x) =0
on I.
(e) Deduce from the result of part (d) that y 1 and y2 are linearly independent
on I if and only if W(x) =/= 0 on I.
16. Consider the equation x 2y" - xy' = 0 on the interval I= (-l, 1).
(a) Find two linearly independent solutions y 1 and Y2 •

(b) Compute the Wronskian W(x).
(c) Show that there is a point x0 in I with W(x0) = 0.
(d) W hy does the result of part (c) not contradict the result of part (e) in
Exercise 15?
Chapter 4
Higher-Order Linear Equations
The definitions and certain of the methods of solution discussed in Chapter 3 can be
extended to equations of any order n 2'. 2.
Definition 4.1. The general solution of an equation of order n 2'. 2 is the solution
which contains n arbitrary constants which cannot be combined. It is an n-parameter
family of solutions, i.e., a set of solutions such that every solution corresponds to
particular values of the arbitrary constants. A solution in which the constants arc
assigned particular values is called a particular solution.
Definition 4.2. An initial-value problem for an equation of order n 2'. 2 consists of

the equation, together with n initial conditions, which specify the values of
I II
y, y' y' . . . ' y(n-1)
at a single point x and, thereby, determine a particular solution.
Definition 4.3. An equation of order n 2'. 2 is linear if it has the form
Otherwise, it is nonlinear. The standard form of Equation (4.1) is
y(n) + Pn-1(x)y(n-l) + · · · + P2(x)y + P1(x)y + Po(x)y = f(x).

11 1
(4.2)
The equation is called homogeneous if g(x) = 0 (or f(x) = 0). Otherwise, it is

nonhomogeneous.
87
88 CHAPTER 4. HIGHER-ORDER LINEAR EQUATIONS
4.1 Homogeneous Equations

Theorem 4.1. If y 1, Y2, · · ·, Yn are solutions of the linear, homogeneous equation
then any linear combination
is also a solution.
Theorem 4.2. The homogeneous equation (4.3) has precisely n linearly independent
solutions Y1, Y2, · · · , Yn , and the general solution is
where c1, c2, · · ·, Cn are arbitrary constants.
4.1.1 Equations with Constant Coefficients

Consider the equation
an y( n ) + an -lY( n -l) + ... + a2y" + a1y' + aoy = 0, (4.4)
where a,, a2, · · ·, an are constants.
As in the case n = 2 discussed in Chapter 3, seek solutions in the form y = e rx.

Then
and substitution into Equation (4.4) yields
or, upon division by erx # 0, the indicial equation
P(r) = an ,n + an -lrn-l + ... + a2r2 +air+ ao = 0. (4.5)
In the case n = 2 discussed in Chapter 3, there are three possibilities for the
roots of the polynomial P in Equation (4.5), namely, two distinct real roots, one real
repeated root, or two complex-conjugate roots.
In the cases n 2: 3, there are more possibilities, such as repeated complex roots,
or real roots which are repeated more than once, but the forms of the solutions are
analogous to the ones for n = 2, and are given by Theorems 4.3-4.5.
Theorem 4.3. If P has n distinct real roots ri, 1 :S: i :S: n, i.e.,
where r i f= rj if if= j, then n independent solutions of Equation (1.4) are given by
and the general solution is y = C1Y1 + c2 y2 + · · · + CnYn·

4 1/1 4 /I 0
y - y -y y= .
I+
P(r) = 4r3 - 4r2 - r + 1 = 0.

Since r = 1 is a root, r -1 must be a factor. Long division of r -1 into 4r3 - 4r2 - r + 1
gives the other factor 4r2 - 1. Alternatively,
4r3 - 4r2 - r+1 (r - 1)(4r2 + ar - 1)
4r3 + (a - 4)r2 - (a+ l)r + 1 ⇒ a= 0 ⇒
2
4r 3
- 4r - r+1 (r - 1)(4r2 - 1),
as above. Thus,
P(r) = 4r3 - 4r2 - r + 1 = (r - 1)(2r - 1)(2r + 1) = 0
y"' - 4y" + y' + 6y = 0, y(0) = 0, y'(0) = 0, y"(0) = 12.
P(r) = r3 - 4r2 + r + 6 = 0.
Since r = -1 is a root, r+1 must be a factor. Long division of r+1 into r3 -4r2+r+6
gives the other factor r2 - 5r + 6. Alternatively,
r3 - 4r2 +r+6 (r+ l)(r2 + ar +6)
r 3 +(a+ l)r2 +(a+ 6)r+ 6 ⇒ a= -5 ⇒
r3 - 4r2 + r+ 6 (r + l)(r2 - 5r + 6) = (r + l)(r - 2)(r - 3) =0
is the general solution · Then y' = -c1 e- x +2c2 e2x +3c3 e3x , y" = c1 e- x +4c2 e2x +9c3 e3x ,
and the initial conditions y(0) = 0, y'(0) = 0 and y"(0) = 12 require that
The sum of the first two equations gives 3c2 + 4c3 = 0, and the sum of the last two
gives 6c2 + 12c3 = 12. The solution of the system
3c2 + 4c3 - 0
{ }
6c2 + 12c3 = 12
is c2 = -4 and c3 = 3. Then c1 = 1, and
y= e- x - 4e2x + 3e3x
y( 4) - 5y"' + 5y" + 5y' - 6y = o.
P(r) = r4 - 5r3 + 5r 2 + 5r - 6 = 0.
Since r = 1 and r = -1 are roots, r - 1 and r + 1 must be factors. Long division of
(r - l)(r + 1) = r2 - 1 into r4 - 5r3 + 5r2 + Sr - 6 gives the other factor r2 - 5r + 6.
Alternatively,
r4 - 5r3 + 5r2 + 5r - 6 (r2 - l)(r 2 + ar + 6)

r4 + ar3 + 5r2 - ar - 6 ⇒ a = -5 ⇒
r4
- 5r 3
+ 5r + 5r - 6
2 2 2
(r - l)(r - 5r + 6)
(r - l)(r + l)(r - 2)(r - 3) =0
⇒ r1 = 1, r2 = -1, r = 2 and r4 = 3, and y1 = e
3
x
, Y2 = e- x , y3 = e2x and Y4 = e x,
3
and

Definition 4.4. Let P be a polynomial and suppose that
P(r)= Q(r)(r - ror,
where Q is a polynomial with Q(r0) =I- 0 and m 2: 1 is an integer. Then r0 is a root

of P(r) and m is called the multiplicity of the root r0.
For example, if P(r) = r3 - 3r + 2 = (r + 2)(r - 1)2 , then r0 = -2 is a root of

multiplicity 1 and r1 = 1 is a root of multiplicity 2.
If P(r) = r4 - r3 - 3r2 + 5r - 2 = (r2 + r - 2)(r - 1)2 , then the multiplicity of

the root r0 = 1 is not 2 because Q(r) = r2 + r - 2 and Q(r0) = Q(l) = 0. Since
P(r)= (r+2)(r-1)3 and Q(r)= r+2 with Q(l) = 3 =/- 0, the multiplicity of r0 = 1
is m= 3.
Thus, the multiplicity of a root r0 of a polynomial P is the largest power of r - r0

which occurs in the factorization of P.
Theorem 4.4. Let r0 be a root of the indicial equation (4.5) obtained by letting
y = erx in Equation (4.4). If r0 has multiplicity m 2: 1, i.e., P(r) = Q(r)(r - r0)m
with Q(r0) =I- 0, then m linearly independent solutions of Equation (4.4) are given by
Proof. Define the linear operator L by
L[y] = any (n) + an JY (n-l) + · · · + a2y" + a1y' + aoy,

-
i.e., L assigns to a function y the expression on the left-hand side of Equation (4.4).
Thus, y is a solution of Equation (4.4) if and only if L[y] = 0. Then
(an 1.n + an lrn-l + · · · + a2 r2 + a1 r + ao)e TX
-
P(r)e TX = Q(r)(r - roreTX = 0 when r= ro, if m 2: 1,
L [!!,_erx ] = !!,_L eTX ] = !!,_ Q(r)(r - rore rx l
dr dr [ dr [
(r - ror!!,_ [ Q(r)erxl + m(r - ror- 1 [Q(r)erx l = 0

dr
when r= r0, if m 2: 2,
L [.!._erx] = .!._L[e rx l = .!._[Q(r)(r - rore rx l
dr2 dr 2 dr2
d2 rx .1 d rx
(r - ror-[Q(r)e l + 2m(r - ror- -[Q(r)e l
dr 2 dr
+m(m - l)(r - ror-2 [ Q(r)e TX ] = 0 when r = ro, if m 2: 3,
and, in general, L[xk e rx ] = 0 when r = ro, if m > k. Thus, for k = 0, l, · · · , m - 1,

xk e rox is a solution. Note that the order of differentiation with respect to r and x has
been reversed. This is permissible by Theorem 2.1 in Section 2.4.1 since the partial
derivatives of all orders of e rx are continuous everywhere.
By Theorem 4.4, every real root of multiplicity m yields m independent solutions

of Equation (4.4). The solutions which correspond to the complex roots of P(r) are
discussed in Theorem 4.5.

111 - 4 " + 5 ' - 2y = 0.
y y y
P(r) = r3 - 4r2 + 5r - 2 = 0.
Since r = l is a root, r - l is a factor. Long division of r - l into r3 - 4r2 + 5r - 2
gives the other factor r2 - 3r + 2. Hence,
r3 - 4r2 + 5r - 2 = ( r - l)(r2 - 3r + 2) = ( r - l) 2 (r - 2) =0
⇒ r 1 = 1 is a root of multiplicity 2 and r2 = 2 is a root of multiplicity 1. Three
linearly independent solutions are therefore given by Y1 = ex , Y2 = xex and y3 = e 2x ,
and

111
y
-
3y" + 3y' - y = 0, y(0) = 1, y'(0) = 2, y"(0) = 9.
y = e rx ⇒
P(r) = r3 - 3r2 + 3r - 1 = 0.
Since r = l is a root, r - l is a factor. Long division of r - l into r3 - 3r2 + 3r - 1
gives the other factor r 2 - 2r + 1. Hence,
r3 - 3r2 + 3r - 1 = (r - l)(r2 - 2r + 1) = (r - 1)3 = 0

⇒ r1 = 1 is a root of multiplicity 3, and three linearly independent solutions are
given by Y1 = ex , Y2 = xex and y3 = x2 ex , and
is the general solution. Then
y' C1ex + c2 (e x + xex) + c3 (2xex + x2ex)

(c1 + c2)ex + (c2 + 2c3)xex + c3 x2ex ,
y" (c1 + c2)ex + (c2 + 2c3)(ex + xex ) + c3(2xex + x2ex )
(c1 + 2c2 + 2c3)ex + (c2 + 4c3 )xex + C3X2ex ,
y (0)= c1 = 1, y '(0) = c1 + c2 = 2 ⇒ c2 = 1, y"(0) = c1 + 2c2 + 2c3 = 9 ⇒ C3 = 3 ⇒
x 2 x
y = e + xe + 3x e is the solution of the initial-value problem.
x
y
( 4)
- 4y "' + 3y " + 4y ' - 4y = 0.
P(r) = r4 - 4r3 + 3r2 + 4r - 4 = 0.

Since r = 1 and r = -1 arc roots, r - 1 and r + 1 are factors. Long division of
(r - l)(r + 1) = r2 - 1 into r4 - 4r3 + 3r2 + 4r - 4 gives the other factor r2 - 4r + 4.
Hence,
r4 - = (r2 - l)(r2 - 4r + 4) = (r - l)(r + l)(r - 2)2 = 0
4r3 + 3r2 + 4r - 4
⇒ r 1 = 1 and r2 = -1 are roots of multiplicity 1, and r3 = 2 is a root of multiplicity
2. Four independent solutions are therefore given by y 1 = ex , y2 = e-x , y3 = e2x and
y4 = xe2x , and
Example 4. 7. Consider the equation

4
y ) - 2y "' - 3y " + 4y ' + 4y
(
= 0.
P(r) = r4 - 2r 3 - 3r2 + 4r + 4 = 0.
Since r = - l and r = 2 are roots, r + l and r - 2 are factors. Long division of
(r + l)(r - 2) = r2 - r - 2 into r4 - 2r3 - 3r2 + 4r + 4 gives the other factor r2 - r -2.
Hence,
r4 - 2r3 - 3r2 + 4r + 4 = (r2 - r - 2)2 = (r + 1)2 (r - 2)2 = 0
⇒ r1 = -1 and r2 = 2 are roots of multiplicity 2. Four independent solutions are
therefore given by Yi = e-x , Y2 = xe-x , y3 = e2x and y4 = xe2x , and
Y = C1e-x + C2Xe-x + C3 e 2x + C4Xe2x
y (4 ) - 5y'" + 6y" + 4y' - 8y = 0.
P(r) = r4 - 5r3 + 6r2 + 4r -8 = 0.

Since r = -1 and r = 2 are roots, r + 1 and r - 2 are factors. Long division of
(r + l)(r - 2) = r2 - r -2 into r4 -5r3 +6r2 + 4r -8 gives the other factor r2 -4r +4.
Hence,
r4 - 5r3 + 6r2 + 4r - 8 = (r + l)(r -2)(r2 - 4r + 4) = (r + l)(r -2)3 = 0

=> r1 = -1 is a root of multiplicity 1 and r2 = 2 is a root of multiplicity 3. Four
independent solutions are therefore given by y1 = e- x , y2 = e 2x , y3 = xe2x and
Y4 = x 2 e 2x, and
y ( 4) - 8y111 + 24y" - 32y' + 16y = 0.
P(r) = r4 - 8r3 + 24r2 - 32r + 16 = 0.

Since r = 2 is a root, r-2 is a factor. Long division of r-2 into r4 -8r3 +24r2 -32r+16
gives the other factor r3 - 6r2 + 12r -8. Since r = 2 is a root of the cubic, r - 2 is a
factor. Long division of r -2 into r3 - 6r2 + 12r -8 gives r2 - 4r + 4. Hence,
r4 - 8r3 + 24r2 - 32r + 16 = (r - 2)2 (r2 - 4r + 4) = (r - 2)4 = 0

=> r1 = 2 is a root of multiplicity 4. Four independent solutions are therefore given
by Y1 = e2x' Y2 = xe2x' y3 = x 2e2x and Y4 = x3 e2x' and
is the general solution. Note that, since no root of P(r) other than r1 = 2 could be
found, it may be beneficial to determine whether or not r1 = 2 is also a root of P'(r).
P'(r) = 4r3 - 24r2 + 48r - 32,

and r1 = 2 is a root. Since r1 = 2 is a root of both P(r) and P'(r), (r - 2)2 must
be a factor of P(r), and long division needs to be performed only once in order to
determine the other factor r2 - 4r + 4.
In general, if P(r) is a polynomial of degree n 2: 1 and 1 ::::; k :::; n, then (r - r0)k

is a factor of P(r) if and only if r0 is a root of P(r), P'(r), ..., and p(k-ll(r).
Theorem 4.5. Let r0 =a+ i/3 and r0 = a - i/3 be two complex-conjugate roots of
the indicial equation (4.5) obtained by letting y = e rx in Equation (4.4). Then two
real, linearly independent solutions of Equation (4.4) are given by
Y1 = e °'x cos(/Jx) and Y2 = e °'x sin(/Jx).

More generally, if r0 and r0 have multiplicity m 2: 1, i.e.,
then 2m real, linearly independent solutions of Equation (4.4) are given by
e °'x cos(f]x), xe°'x cos(f3x), x 2 e °'x cos(f3x),
e °'x sin(/Jx), xe °'x sin(/Jx), x 2 e °'x sin(/Jx),

The remaining n - 2m independent solutions of Equation (4.4) are determined by the
roots of the polynomial Q of degree n - 2m.
Proof. By the proof of Theorem 4.4, the complex-conjugate functions xk e rox and
xk erox , 0 :::; k :::; m - 1, are solutions. As in the case n = 2 derived in Section
3.2.1, from every one of the m pairs of complex-conjugate solutions Zk = xk e ro x and
Zk = x e o , 0 :::; k :::; m - 1, two real solutions
k r x
Zk + Zk Zk - Zk
= xk e°'x cos(/3x) and = xk e°'x sin(/Jx)
2 2i
are obtained, for a total of 2m real, independent solutions. By the first part of this
theorem and/or Theorem 4.4, an additional n - 2m real, independent solutions are
determined by the roots of Q.
y111 - 8y" + 25y' - 26y = 0.
P(r) = r3 - 8r2 + 25r - 26 = 0.

Since r = 2 is a root, ,,. - 2 is a factor. Long division of r - 2 into r3 - 8r 2 + 25r - 26
gives the other factor r2 - 6r + 13, and
r2 - 6r + 13 =0 =} r = 6 ± J35 - 52 = 3 ± 2i.
2
Hence, P has the real root r1 = 2 of multiplicity 1 and the complex-conjugate roots
3 ± 2i of multiplicity 1. Three real, independent solutions are therefore given by
Y1 = e 2x, Y2 = e 3x cos(2x) and y3 = e sin(2x), and
3x
y = c1e 2x + c2 e 3x cos(2x) + C3 e3x sin(2x)


y ( 4) - 7y"' + 27y" - 47y' + 26y = 0.
P(r) = r4 - 7r3 + 27r2 - 47r + 26 = 0.

Since r = 1 and r = 2 are roots, (r - 1) and (r - 2) are factors. Long division of
(r - l)(r - 2) = r2 - 3r + 2 into r4 - 7r3 + 27r2 - 47r + 26 gives the other factor
r2 - 4r + 13, and
4±
r2 - 4r + 13 =0 ⇒ Jl6 -
r
52
= = 2 ± 3i.
2
Hence, four real, independent solutions are given by y1 = ex , Y2 = e2x, y3 = e 2x cos(3x)
and y4 = e 2x sin(3x), and
y = c1ex + c2 e 2x + C3 e2x cos(3x) + C4e 2x sin(3x)

4
y ( ) - 8y"' + 26y" - 32y' + 13y = 0.
P(r) = r4 - 8r3 + 26r2 - 32r + 13 = 0.

Since r = 1 is a root of both P(r) and P'(r) = 4r3 - 24r2 + 52r - 32, (r - 1)2 is a
factor of P(r). Long division of (r - 1)2 = r2 - 2r + 1 into r4 - 8r3 + 26r2 - 32r + 13
gives the other factor r2 - 6r + 13, and
r2 - 6r + 13 =0 ⇒ = 6 ± J35 - 52 = 3 ± 2i.
r
2
Hence, r1 = 1 is a root of multiplicity 2 and 3 ± 2i are two complex-conjugate
roots
of multiplicity 1. Four real, independent solutions are therefore given by y1 = ex ,
Y2 = xe , y3 = e cos(2x) and Y4 = e3x sin(2x), and
x 3x
y = c1ex + C2Xex + C3 e3x cos(2x) + C4e3x sin(2x)


4
y( ) + y" + y = 0.
y =e ⇒
rx
P(r) = r4 + r 2 + 1 = 0.
Since no real roots are immediately evident, write
r4 + r 2 + 1 (r 2 + ar + l)(r 2 + br + 1)
r4 +(a+ b)r3 +(ab+ 2)r 2 +(a+ b)r + 1
and solve for a and b to obtain a= ±1 and b = =fl. Choosing a= 1 and b = -1 gives
r4 + r 2 + 1 = (r2 + r + l)(r2 - r + 1).
Choosing a = -1 and b = 1 gives the same result.
r2 + r + 1 = 0 => r= -l±Jf=4 =-!±i \/'3 and

2 2 2 '
r2 - r+1
l±vT=°4
= 0 => 1· = --- - = -±i-.
1 .v'3
2 2 2
Hence, P(r) has two pairs of complex-conjugate roots of multiplicity 1. Four real,
independent solutions are therefore given by
y 1 = e-lx cos ( '7x)

and the general solution is y = C1Y1 + C2Y2 + c3y3 + C4y4.
y ( 4) - 4y111 + 14y" - 20y 1 + 25y = 0.
P(r) = r4 - 4r3 + 14r2 - 20r + 25 = 0.

Since no real roots are immediately evident,
r4 - 4r3 + 14r 2 - 20r + 25 (r 2 + ar + 5)(r2 + br + 5)

r4 +(a+ b)r3 +(ab+ 10)r 2 + 5(a + b)r + 25
⇒ a+b = -4 and ab+ 10 = 14 ⇒ b = -4- a ⇒ -a2 - 4a = 4 ⇒ a2 + 4a + 4 = 0

⇒ (a+ 2)2 = 0 ⇒ a= -2 ⇒ b = -2 ⇒
r4 - 4r3 + 14r2 - 20r + 25 = (r2 - 2r + 5)2,
and
2 ± ✓4- 20
r 2- 2r + 5 = 0 ⇒ r = ----- = l ± 2,;"·
2
Hence, 1 ± 2i are complex-conjugate roots of multiplicity 2. Four real, independent
solutions are therefore given by
Y1 = ex cos(2x), Y2 = xex cos(2x), y3 = ex sin(2x), y4 = xex sin(2x),

and the general solution is y = c 1 y 1 +c2y2+c3y3+c4 y4. Note that, if P(r) is expressed
as
r4 - 4r3 + 14r2 - 20r + 25 (r2 + ar + l)(r2 + br + 25)
r4 +(a+ b)r3 +(ab+ 26)r2 + (25a + b)r + 25,
then the equations a+b = -4, 25a + b = -20 and ab+26 = 14 are inconsistent and
lead to a contradiction.
4.1.2 Cauchy-Euler Equations

Definition 4.5. An equation of the form
an xn y ( n ) + an-1X n-l y ( n- l) + · · · + a2x2y " + a1xy ' + aoy = 0, n 2: 2, (4.6)
where ai, 0 :::; i :::; n, are constants, is called a Cauchy-Euler or Euler equation.
Definition 4.5 is a generalization of Definition 3. 7 given in Section 3.2.2 for n = 2

to equations of order n 2: 2. The solutions are analogous to the ones obtained therein
for n = 2, and are given by the following:
Theorem 4.6. Let P(r) = 0 denote the indicial equation which results from the
substitution of y = xr into Equation (4.6) for x > 0. If r = r0 is a real root of
multiplicity m 2: 1, then m independent solutions of Equation 4.6 for x > 0 are given
by
xr0 , xr0 ln(x), xr0 [ln(x)]2, •··, xr0 [ln(x)J m -l_
If a± i/3 are complex-conjugate roots of multiplicity m 2: 1, then 2m independent
solutions of Equation 4.6 for x > 0 are given by
xa cos[/3 ln(x)], xa ln(x) cos[/3 ln(x)], x0 [ln(x)] m -l cos[/3 ln(x)],
xa sin[/3 ln(x)], xa ln(x) sin[/3 ln(x)], x0 [ln(x)]m-t sin[/3 ln(x)] .
The solutions for x =/- 0 are obtained from the ones for x > 0 by replacing x by lxl.
Proof. For x > 0, make the change of variables l = ln(x) and y(x) = z(l). Then, by
the chain rule,
dy dz dl l dz
-- = -- ⇒ xy (x) = z'(t),
,
]
dx dl dx x dt
d2 y
dx2
1 d2 z
x2 dt2
l dz
x2 dl
=
l d2 z dz
[
x2 dl2 dt
-
⇒ x2 y"(x) = z"(l) z'(t),
d3 y 1 d3 z d2 z 2 d2 z dz l d3 z d2 z dz
[ ] [ ] = [ 3 + 2 ]
dx3 x dt
3 3 dt 2 x dt
3 2 dt x dl
3 3 dl 2 dt
3 111 =
⇒ x y (x) z (t) - 3z"(l) + 2z'(t),
111
(
etc., i.e., every term xk y( k) x) in the Euler equation transforms into an expression with
constant coefficients. Thus, the Euler equation for y(x) transforms into an equation
for z(t) with constant coefficients. Moreover, the indicial equation P(r) = 0 for y(x)
is identical to the one for z(t) because z(t) = e rl if and only if y(x) = e r In(x) = xr .
The forms of the solutions for y(x) then follow from the corresponding forms of the
solutions for z(l) as given by Theorems 4.4 and 4.5 in Section 4.1.1. Replacing x by
lxl then gives the solutions for both x > 0 and x < 0.
For x > 0, y = xr ⇒
P(r) r(r - l)(r - 2) + 3r(r - 1) - 2r + 2
r ( r - 1) ( r - 2) + 3r ( r - 1) - 2 (r - 1)
(r - 1) [r ( r - 2) + 3r - 2]
(r - 1) ( r2 + r - 2)
(r - 1) 2 (r + 2) = 0
⇒ r1 = 1 is a root of multiplicity 2 and r2 = -2 is a root of multiplicity 1. Hence,

Yi= x, Y2 = x ln(x) and y3 = x- 2 for x > 0, and
is the general solution for x =/ 0.
x3 y
111 - 4x2 y" + l3xy' - l3y = 0.
For x > 0, y = xr ⇒
P(r) r(r - l)(r - 2) - 4r(r - 1) + 13r - 13
r(r - l)(r - 2) - 4r(r - 1) + 13(r - 1)
(r - l)[r(r - 2) - 4r + 13]
( r - 1)(r2 - 6r + 13) = 0
6 ± J36 - 52
r - 6r + 13
2
=0 ⇒ r = = 3 ± 2i.
2
Hence, Y1 = x, Yz = x3 cos[2 ln(x)] and y3 = x3 sin[2 ln(x)] for x > 0, and
y = c1x + c2 x3 cos (2 ln lxl) + c3x3 sin(2 ln lxl)
is the general solution for x # 0.
x3 y'" - x2 y" - 2xy' + 6y = 0, y(l) = 0, y'(l) = 1, y"(l) = 16.

For x > 0, y = xr ⇒
P ( r) r(r - 1)(r - 2) - r ( r - 1) - 2r + 6
r(r - l)(r - 2) - (r + r - 6)
2
r(r - l)(r - 2) - (r + 3)(r - 2)

(r - 2)[r(r - 1) - (r + 3)]
(r - 2)(r2 - 2r - 3)
(r - 2)(r - 3)(r + 1) = 0
⇒ y = c1x2 + c2 x3 + c3x-1 is the general solution for x # 0. Then
y' 2c1x + 3c2 x2 - c3 x-2 ,

y" = 2c1 + 6c2 X + 2C3X-3 '
and y(l) = 0, y'(l) = 1 and y"(l) = 16 ⇒

y ( 1) Cl + C2 + C3 = 0,
y'(l) 2c1 + 3c2 - c3 = 1,
y"(l) 2c 1 + 6c2 + 2c3 = 16.
Subtracting twice the first equation from third gives c2 = 4. Adding the first two
gives 3c1 + 4c2 = 1 ⇒ c1 = -5, and the first equation then gives c3 = 1. Hence,
y = -5x2 + 4x3 + x-1 is the solution of the initial-value problem.
x4y ( 4) - x3 y"' + 4x2 y" - 8xy' + 8y = 0.

For x > 0, y = x r ⇒
P(r) = r(r - l)(r - 2)(r - 3) - r(r - l)(r - 2) + 4r(r - 1) - 8r + 8
= r(r - l)(r - 2)(r - 3) - r(r - l)(r - 2) + 4r(r - 1) - 8(r - 1)
= (r - l)[r(r - 2)(r - 3) - r(r - 2) + 4r - 8]
= (r - l)[r(r - 2)(r - 3) - r(r - 2) + 4(r - 2)]
= (r - l)(r - 2)[r(r - 3) - r + 4]
= (r - l)(r - 2)(r2 - 4r + 4)
(r - l)(r - 2)3 = 0
x4y (4) + 7x3 y"' + 13x 2 y" - 26xy' + 26y = 0.

For x > 0, y = xr ⇒
P(r) r(r - l)(r - 2)(r - 3) + 7r(r - l)(r - 2) + 13r(r - 1) - 26r + 26
r(r - l)(r - 2)(r - 3) + 7r(r - l)(r - 2) + 13r(r - 1) - 26(r - 1)
= (r - l)[r(r - 2)(r - 3) + 7r(r - 2) + 13r - 26]
= (r - l)[r(r - 2)(r - 3) + 7r(r - 2) + 13(r - 2)]
(r - l)(r - 2)[r(r - 3) + 7r + 13]
(r - l)(r - 2)(r2 + 4r + 13) = 0
-4 ± ✓16 - 52
r2 + 4r + 13 = 0 ⇒ r = ----- = -2 ± 3i.
2
Hence, y = c1x + c2x2 + c3x-2 cos(3 ln !xi) + c4x- 2 sin(3 ln lxl) is the general solution
for X-=/ 0.
Exercises 4.1
1. F ind the general solution of the equation y"' - 6y" + lly' - 6y = 0.
2. Solve the initial-value problem y"' - 2y" - y' + 2y = 0, y(0) = 6, y'(0) = 0,
y"(0) = 0.
9 111
3. Find the general solution of the equation y + 9y" - y' - y = 0.
4. Find the general solution of the equation y111 -
5y" + 8y' - 4y = 0.
5. Find the general solution of the equation y"' - 6y" + l2y' - 8y = 0.
6. Solve the initial-value problem y"' + y" - y' - y = 0, y(0) = 0, y'(0) 3,
y"(0) = 2.
7. Find the general solution of the equation y"' - 5y" + l 7y' - l3y = 0.
8. Find the general solution of the equation y ( 4) - 5y" + 4y = 0.
9. Solve the initial-value problem y( 4) - 2y'" - y" + 2y' = 0, y(0) = 0, y'(0) = 2,
y"(0) = 0, y'"(0) = 8.
10. Find the general solution of the equation y ( 4) - 7y111 + l7y" - l7y' + 6y = 0.
11. Find the general solution of the equation y(4) - y"' - 3y" + 5y' - 2y = 0.
12. Find the general solution of the equation y( 4) - 4y111 + 6y" - 4y' + y = 0.
13. Find the general solution of the equation y( 4) - 6y111 + l3y" - 12y' + 4y = 0.
14. Find the general solution of the equation y( 4) + 2y"' + 4y" - 2y' - 5y = 0.
15. Find the general solution of the equation y( 4) + lOy"' + 4ly" + 76y' + 52y = 0.
16. Find the general solution of the equation y ( 4) + 5y" + 4y = 0.
17. Find the general solution of the equation y ( 4) + 4y = 0.
18. Find the general solution of the equation y( 4) - 8y"' + 26y" - 40y' + 25y = 0.
19. Find the general solution of the equation x3 y"' + x2 y" - 2xy' + 2y = 0.
20. Find the general solution of the equation x3 y"' - 2x2 y" + 3xy' - 3y = 0.
21. Find the general solution of the equation x3y"' + xy' - y = 0.
22. Solve the initial-value problem x3 y"' + 8x2 y" + l3xy' - l3y = 0, x > 0, y(l) = 0,
y'(l) = 6, y"(l) = -2.
23. Find the general solution of the equation x4 y( 4 ) + x3 y"' + x2 y" - 2xy' + 2y = 0.
24. Find the general solution of the equation x4 y(4 ) - x3y"' + 5x2 y" - lOxy' + lOy = 0.
25. Find the general solution of the equation x4y( 4) + 6x3 y111 + 6x2 y" - 2xy' + 2y = 0.
26. Find the general solution of the equation x4y( 4) + 2x3 y"' + 3x2 y" - 3xy' + 4y = 0.
4.2. NONHOMOGENEOUS EQUATIONS 103
4.2 Nonhomogeneous Equations

Theorem 4. 7. The general solution of the linear, nonhomogeneous equation of order
n 2'. 2,
is y = YP + Yh, where YP is any particular solution, and Yh = C1Y1 + C2Y2 + · · · + C,,,iYn

is the general solution of the associated homogeneous equation
where Yk, 1 ::; k ::; n, are n linearly independent solutions.
Theorem 4.7 is a generalization of Theorem 3.3 given in Section 3.3 for n = 2,

and the proof is similar. The general solution of Equation (4.8) is obtained by the
methods discussed in Sections 4.1.1 and 4.1.2 for equations with constant coefficients
and Cauchy-Euler equations. A particular solution of Equation (4.7) is obtained by
variation of parameters or, where applicable, the method of undetermined coefficients,
suitably generalized to equations of order n 2'. 2.
4.2.1 Variation of Parameters

Definition 4.6. The Wronskian of n solutions y 1, Y2, · · ·, Yn of Equation (4.8) is
Y1 Y2 Yn
y� y; y�
W(x) = W[y1, Y2, · · · , Yn ](x) =
n- (n-l n-l )
Y1( l ) Y2 ) Yn(
Theorem 4.8. Consider the standard form of Equation (4.7),
y(
n)
+ Pn-1(x)y (n-l ) + · · · + P2(x)y" + P1(x)y' + Po(x)y = f(x), (4.9)
and the associated homogeneous equation
y(
n)
+ Pn-1 (x)y (n-l) + · · · + P2(x)y" + P1(x)y' + Po(x)y = 0. (4.10)
=
If Pk, 0 ::; k ::; n - 1, are continuous on an interval 1, and Yk, 1 ::; k ::; n, are solutions
of Equation (4.10), then either W 0 on I or W(x) -=I= 0 for any x in J. The solutions
Yi, Y2, · · ·, Yn are linearly independent on I if and only if W -=I= 0 on I.
The method of variation of parameters for equations of order n � 2 is given by

the theorem which follows, and is an extension of the one for n = 2.
Theorem 4.9. L et y1, y 2, · · · , Yn be n linearly independent solutions of Equation

(4.10) and let W(x) be the Wronskian. Let �j (x), 1 :S j :Sn, denote the Wronskian
with its j th column replaced by the column vector [0, 0, • • • , 0, J(t)J t , i.e.,
0
0
(n-1) (n-1)
Y1 Y2 J(t) Y
n
(n-1)
9
P of Equation (4 . ) is given by
Then a particular solution Y
where

/ 2y= 27e -X .
y - 3y-
II/
For the associated homogeneous equation
yII/ - 3y/ - 2y= 0 '
y = erx ⇒ P(r) = r3 - 3r - 2 = (r + l)(r 2 - r - 2) = (r + l)2 (r - 2) = 0 ⇒ Y1 = e-x ,

y2 = xe- x and y
3 = e
2x
are three independent solutions. W ith J(x) = 27e- x ,
e- x xe-x e2x
W(x) = -e - x -
e - xe
x - x
2e2x
e - -2e- + xe-x 4e 2X
x x
e- (sex - 6xex ) - xe-x (-6ex)
x
+ e2x ( e- 2x) = 9,
0 xe- x e 2x
0 e-x - xe-x 2e2x = 27e- x (3xex - ex ) = 27(3x - 1),
x x
27e- -2e- + xe- 4e2x x
e- x 0 e 2x
-e - x
0 2e 2x = -27e-x (3ex ) = -81,
x x
e- 27e- 4e 2x
e- x xe- x 0
-e - x -
e - xe- x
x
0 = 27e-x ( e-2x) = 27e-3x .
x
e- -2e- + xe-x 27e- x
x
J
Then
u1 3(3x - 1) dx = ;x2 - 3x,

u2 J -9dx =
J
-9x,
U3 3e- 3x dx = -e-3x ,
and
YP ( ;x2 - 3x) e-x - 9x(xe-x ) - (e-3x e2x )
9
--x2 e-x
2
- 3xe- x - e- x
or YP = -;x2 e- x , since -3xe-x - e- x is a solution of the homogeneous equation.

Hence,
Y = --9 x2 e-x + C1e-x + C2 Xe-x + C3e2x
2
y"' - y' = 8xex , y(0) = 0, y' (0) = 0, y" (0) = 0.

y"' -y' = 0,
y=e
rx
⇒ P(r) = r3 - r = r(r2 -
1) = r(r - l)(r + 1) = 0 ⇒ Y1 = 1, Y2 = e
x
and
y3 = e- are three independent solutions. With J(x) = 8xe ,
x x
1 ex e-x
W(x) = 0 ex = -e-x 2,
0 ex
e-x
0 ex e -x
0 e -e-x
x
= -16xex ,
8xe x x
e e -x
1 0 e-x
0 0 -e-x = 8x,
0 8xex e-x
1 ex 0
0 ex 0 = 8xe2x .
0 ex 8xex
Then
J -8xexdx = -8xex + 8ex,

J
U1
4xdx = 2x2 ,
J
u2
U3 4xe 2xdx = 2xe 2x - e 2X ,
and
or YP = 2x2 e x - 6xex, since 7ex is a solution of the homogeneous equation. Hence,
is the general solution of the nonhomogeneous equation. Then
y' 2x2 e x - 2xex - 6ex + C2 ex - C3e- x ,

y" 2x2 ex + 2xe x - 8e x + C ex + C3e-x, 2
and y (0) = 0, y '(0) = 0 and y"(0) = 0 ⇒
⇒ c1 = -8, c2 = 7, c3 = 1, and y = 2x2 ex - 6xe x - 8 + 7ex + e-x is the solution of

the initial-value problem.
y
111
- 3y " + 7y ' - 5y = 32e x sin(2x).
y
111
- 3 y" + 7 y ' - 5y = 0,
y = e rx ⇒ P(r) = r3 - 3r2 + 7r - 5 = (r - l)(r 2 - 2r + 5) = 0 ⇒ r1 = 1 and
2±F1]
r = --- - = l ± 2i ⇒ y1 = ex, Y2 = e x cos (2x) and y3 = e x sin(2x) are three
2
independent solutions. With J(x) = 32ex sin(2x),
ex e x cos(2x) e x sin(2x)
W(x) e x
e cos(2x) - 2e sin(2x)
x x
e sin(2x) + 2ex cos(2x)
x
e -3e cos(2x) - 4e sin(2x) 4ex cos(2x) - 3ex sin(2x)

x x x
e x (10 e 2x ) - e x (4e 2x ) + e x (2e 2x ) = 8e 3x,

0 �coo(2x) e x sin(2x)
0 e cos(2x) - 2ex sin(2x)
x
e x sin(2x) + 2ex cos(2x)
32ex sin(2x) -3ex cos(2x) - 4ex sin(2x) 4ex cos(2x) - 3ex sin(2x)
32ex sin(2x) ( 2e 2x ) = 64e3x sin(2x),
ex O e x sin(2x)
ex O e sin(2x) + 2ex cos(2x)
x
ex 32ex sin(2x) 4ex cos(2x) - 3e x sin(2x)

-32ex sin(2.1:) [2e 2x cos(2x)] = -64e 3x sin(2x) cos(2x),
ex e x cos(2x) 0
e x
e cos(2x) - 2ex sin(2x)
x
0
e x -3ex cos(2x) - 4ex sin(2x) 32e sin(2x)
x
32ex sin(2x) [-2e 2x sin(2x)] = -64e 3x sin 2(2x).

Then
u1 J 8sin(2x)dx = -4cos(2x),
u2 J -8sin(2x) cos(2x)dx = -4 J sin(4x)dx = cos(4x),
u3 J -8sin 2(2x)dx = 4 J cos(4x) - 1 dx = sin(4x) - 4x,
and
YP -4ex cos(2x) + ex cos(2x) cos(4x) + [sin(4x) - 4x]ex sin(2x)

-4ex cos(2x) - 4xex sin(2x) + e x [cos(2x) cos(4x) + sin(2x) sin(4x)]
-4ex cos(2x) - 4xe x sin(2x) + ex cos(2x)
-3ex cos(2x) - 4xe x sin(2x),
or YP = -4xex sin(2x), since -3ex cos(2x) is a solution of the homogeneous equation.

Hence,


y(4) - 6y111 + 1ly" - 6y' = 36e3x .
y(4) - 6y"' + 1ly" - 6y' = 0,
y = e rx =} P(r) = r4 - 6r3 + 1lr 2 - 6r = 0 =}
r(r3 - 6r 2 + 1lr - 6) r(r - l)(r2 - 5r + 6)

r(r - l)(r - 2)(r - 3)
0
Y1 = 1, Y2 = =e =
x, y3
=} e 2x
and y4 e 3x
are four independent solutions, with the
Wronskian
l ex e2x e3x
0 e x 2 2X 3e3X
W (x) = e
0 ex 4e2x 9e3x
0 ex 8e2X 27e3X
Computing the determinant along the fourth row, we obtain
1 e2x e3x 1 ex e3x l ex e2x
W(x) = e O 2 e
x 2x 3e 3x - 8e 2x O ex 3e3x + 27e3 x
0 e x 2 e 2x
0 4 e2x 9e3x O ex 9e3x 0 ex 4 e2x
ex (6e5 ) - 8e2 (6e4 ) + 27e3 (2e3x ) = 12e6 _
x x x x x
With f(x) = 36e3x ,

0ex e2x e3x
e x e 2x e3x
0ex 2e 2x 3e3X
0ex 4e 2x 9e 3x
= - f(x) ex 2 e2x 3e3x
9e3x
f(x) ex 8e2x 27e3x
- f(x) [ex (6e5x ) - ex (5e5x ) + ex ( e5 x
)] = -2 e6x f(x) = -72 e9x ,
1 0 e2x e3x
3e 3x 1 e2x e3x
0 0 2 e2x
0 0 4 e2x 9e3x = f(x) 0 2e2 3e3xx
= 6e5x f(x) = 216e 8x
,
0 4 e2 9e3x
x
0 J(x) 8e 2x 27e 3X
1 ex 0 e3x
0 ex 0 3e3x
fl 3 (x)
0 ex 0 9e3x
= -6e4x f(x) = -216e7x ,
0 ex J(x) 27e3x
1 e
"'
e2
x 0
x 2 2x 1 ex e2x
0 e 0
0
e
e x 4 e 2x 0
= f(x) 0 ex 2 e2x = 2e3x f(x) = 72e6 x
_
0 e x 4 e 2x
0 ex 8e2x f(x)
J
Then
U1 -6 e3x dx = -2e3x ,
u2 18 je 2x
dx = 9e 2x ,
-18 j ex dx = -18e x,
J
U3
U4 6 1 dx = 6x.
Hence, YP = -2e3X + 9e2X ex - 18ex e2X + 6xe3x = -1le3x + 6xe3x, or YP = 6xe3x , since
-lle3x is a solution of the homogeneous equation. Thus,
+ 6xy' - 6y = 8x ln(x), x > 0.

3
x3 y
111 - 3x2 y"
x3 y
111 - 3x 2 y" + 6xy' - 6y = 0,
for which y = xr =} P(r) = r(r -l)(r - 2) - 3r(r - 1) + 6r - 6 = 0 =}
(r - l)[r(r - 2) - 3r + 6] = (r - l)(r 2 - 5r + 6) = (r - l)(r - 2)(r - 3) = 0
⇒ Y1 = x, Y2 = x2 and y = x 3
3
are three independent solutions. With f(x) = 8ln(x),
x x2 x3
W(x) = 1 2x 3x2 = x (6x2 ) - 4x3 = 2x3,
0 2 6x
0 x2 x 3
�1(x) 0 2x 3x2 = f(x) [x4 ] = 8x4 ln(x),
f(x) 2 6x
X 0 x3
1 0 3x2 = - f(x) [2x3] = -16x3 ln(x),
f(x) 6x
2
X x 0
1 2x 0 = f(x) [x2 ] = 8x2 ln(x).
0 2 f(x)
Then
u1 41 xln(x)dx=2x ln(x)-x 2 2
,
u2 -8 j ln(x) dx = -8x ln(x)+ 8x,
u3 4
j ln x)
� dx = 2[ln(x)] 2 .
Hence,
YP [2x2 ln(x) - x2 ] x+ [-8xln(x)+8x] x2 +2[ln(x)] 2 x3

2
-6x3 ln(x) + 7x3 +2x [ln(x)] ,
3
or YP = 2x3 [ln(x)] 2 - 6x3 ln(x), since 7x 3 is a solution of the homogeneous equation.

Thus,
is the general solution of the nonhomogeneous equation for x > 0.
x3 y'" + 2xy' - 2y = x, x > 0.
x3 y"' + 2xy' - 2y = 0,
for which y = xr =} P(r) = r(r - l)(r - 2) +2r - 2 = 0 =}
(r - l)[r(r - 2) + 2] = (r - l)(r2 - 2r + 2) = 0
2± \1-4
⇒r1 =1and r2 -2r+2 = 0⇒r = = l±i⇒y 1 = x,y2 = xcos [ln(x)]
2
1
and y3 = xsin[ln(x)] are three independent solutions. With f(x) = 2,
X
x xcos[ln(x)] xsin[ln(x)]
W(x) = 1 cos[ln(x)] - sin[ln(x)] sin[ln(x)] + cos[ln(x)]
0 -¾ ¾
sin[ln(x)] - cos[ln(x)] ¾ cos[ln(x)] - ¾ sin[ln(x)]
X [�] - 1[1] = 1,
0 xcos[ln(x)] xsin[ln(x)]
0 cos[ln(x)] - sin[ln(x)] sin[ln(x)] + cos[ln(x)]
f (x) -� sin[ln(x)] - � cos[ln(x)] � cos[ln(x)] - � sin[ln(x)]
1
J(x)[x] = -,
X
x O xsin[ln(x)]
1 0 sin[ln(x)] + cos[ln(x)]
0 J(x) ¾ cos[ln(x)] - ¾ sin[ln(x)]
cos[ln(x)]
- J(x){x cos[ln(x)]} = ----,
X
x xcos[ln(x)] 0
6._3(X) 1 cos[ln(x)] - sin[ln(x)] 0
0 -¾
sin[ln(x)] - ¾ cos[ln(x)] J(x)
sin[ln(x)].
f(x){-xsin[ln(x)]} = - X
Then
u1 .
/ 1 dx = ln(x),
cos[l (x)]
u2 -/
. : dx = - sin[ln(x)],
sin[l (x)]
= cos[ln(x)].
- ./
u3 : dx
Hence, yP = xln(x) - xcos[ln(x)]sin[ln(x)] +xsin[ln(x)]cos[ln(x)] = xln(x), and
y = x ln(x) + c1x + c2xcos[ln(x)] + C3X sin[ln(x)]

is the general solution of the nonhomogeneous equation for x > 0.
4. 2. 2 Undetermined Coefficients-Annihilators
In Section 3.3.1, the method of undetermined coefficients was employed to determine
particular solutions YP of second-order nonhomogeneous equations . The same method
can be applied to the higher-order equations considered in the present chapter, with
a slight difference. Instead of listing the form of YP for every type of nonhomogeneous
term g(x), the form of YP will be determined by g(x). The unknown constants in YP
are then determined by substitution into the nonhomogeneous equation, as before.
Definition 4. 7. Let D denote the linear operator defined by
DJ= n[J] = f = J,
d� '
where f is a differentiable function. Let n2 denote the linear operator defined by
D2 f = n 2 [J] = D[n[f ]] = n[J'] = f",
and, more generally, for any integer n 2': 1,
the derivative of f of order n. If n = 0, n ° f is defined by n° f = f. The abuse of

notation n[J(x)] is customary, and denotes the value of the function nf = f' at x,
i.e.'
n[J(x)] = (nf)(x) = J'(x),
and, more generally,
Let P(r) be a polynomial of degree n 2': 0, i.e.,
where ak, 0 � k � n, are constants. The linear operator P(n) is defined by
(an nn + an -ln -l + · · · + a2 n + a1n + aon ) f

n 2 °
P(n)J
an nn f + an -l nn-l f + · · · + a2 n2 f + a1nf + aon° f
l
anf( n ) + an _if ( n- ) + · · · + a2J" + aif ' + aof.
If P(n)J(x) = 0 for all x, then P(n) is called an annihilator of the function J, and
the function f is said to be annihilated by P(n).
Example 4.26. Let P(r) = r - 1. Then P(n) = n - 1, and (n - l)f = f' - f for
any differentiable function f. If f ( x) = ex , then
Thus, n - 1 is an annihilator of the function e . x

Example 4.27. Let P(r) = r2 + 4. Then P(D) = D2 + 4, and (D2 + 4)J = f" + 4J
for any differentiable function f. If f(x) = sin(2x), then
P(D)[sin(2x)] = (D 2 +4) sin(2x) = D2 sin(2x)+4 sin(2x) = -4 sin(2x)+4 sin(2x) = 0.
Thus, D2 + 1 is an annihilator of the function sin(2x).
Example 4.28. Let f(x) = cos(3x). Then

Df(x) = -3 sin(3x) andD2 f(x) = -9cos(3x) = -9f(x).
Hence, (D2 + 9) cos(3x) = 0, i.e., P(D) = D2 + 9 is an annihilator of the function
f(x) = cos(3x).
Example 4.29. Let f(x) = e3x sin(2x). How can an annihilator off be determined?
Recall that e 3x cos(2x) and e3x sin(2x) are solutions of a second-order homogeneous
equation for which the indicial equation has the roots r = 3 ± 2i. Then
r - 3 = ±2i => (r - 3)2 = -4 => P(r) = r2 - 6r + 13 = 0
is the indicial equation. Hence, e3x cos(2x) and e 3x sin(2x) are the solutions of the
equation y" - 6y' + 13y = 0, i.e., they are annihilated by P(D) = D2 - 6D + 13.
Example 4.30. Let J(x) = x2 e2x . Recall that e 2x , xe 2x and x2 e 2x are solutions of a
third-order homogeneous equation for which the indicial equation has the root r = 2
of multiplicity 3, i.e., the indicial equation is P(r) = (r-2)3. Hence, P(D) = (D-2)3
is an annihilator of x2 e 2x , as well as e 2x and xe 2x .
Consider a nonhomogeneous equation of order n � 2 with constant coefficients,
an y ( ) + an -lY ( -J) + · · · + a2y" + a1y' + aoy = g(x), (4.11)

n n
where the nonhomogeneous term is denoted by g(x) since the equation need not be
in standard form. With
Equation (4.11) can be expressed as

P(D)y = g(x).
If Q ( D) is an annihilator of g ( x), then
P(D)y = g(x) Q(D )P(D)y = Q(D)g(x) = 0,
=>
i.e., if y is a solution of the nonhomogeneous equation P(D )y = g(x), then it is a
solution of the homogeneous equation Q(D)P(D)y = 0. From the solutions of the
latter equation, the form of yP can be deduced by means of the following:
Theorem 4.10. Consider the nonhomogeneous equation (4.11), and let Q be a

polynomial of the least degree m � 1 such that Q(D) is an annihilator of g. Let
be n+m independent solutions of the equation Q(D)P(D)y = 0 of order n+m, where

y 1, y2, · · · , Yn are independent solutions of P(D)y = 0. Then there exist constants
c1, c2 , · · ·, Cm such that
(4.12)
Equation (4.12) gives the form of Yp, and the constants Ci, 1 � i � m, are determined
by substitution of yP into Equation (4.11).
Proof. Since Q(D)P(D)zi = 0, 1 � i � m, P(D)zi are solutions of Q(D)y = 0. They
arc linearly independent because if d1P(D)z1 + d2P(D)z2 + · · · + dmP(D)zm = 0, then
P(D)(d1z1 + d2 z2 + · · · + dm zm ) = 0, i.e., z = d1z1 + d2 z2 + · · · + dmZm is a solution
of P(D)y = 0. Hence, there exist constants b1, b2 , · · ·, bn such that
Then
d1z1 + d2z2 +· · · + dmzm = b1Y1 + b2Y2 + · · · + bn Yn
⇒ d1z1 + d2z2 + · · · + dm zm - b1 Y1 - b2 y2 - · · · - bnYn = 0
⇒ di = 0, bj = 0, 1 � i � m, 1 � j � n,
smce
are linearly independent. Thus, P(D)zi, 1 � i � m, are m linearly independent

solutions of Q(D)y = 0. Since g(x) is a solution of Q(D)y = 0, there exist constants
c1, c2, · · · , Cm such that
g(x) c1P(D)z1 + c2P(D)z2 + · · · + em P(D)zm

P(D )(c1z1 + C2Z2 + · · · + CmZm ),
which shows that YP = c1z1 + c2 z2 + · · · + CmZm satisfies P(D)yp = g(x).
y
111 -
3y' - 2y = 27e-x,
which has been solved by variation of parameters in Section 4.2.1, Example 4.20. The
indicial polynomial is P(r) = (r + 1)2 (r - 2). Hence, the nonhomogeneous equation
lS
P(D )y = (D + 1)2 (D - 2)y = 27e-x.

The annihilator of g(x) = 27e-x of minimal degree is Q(D) = D + l. Thus,
Q(D)P(D)y = (D + 1)3 (D - 2)y = (D + 1)[27e-x ] = 0
Since e-x , xe-x and e2x are solutions of the homogeneous equation P(D)y 0,
YP = C3x 2 e-x, or YP = cx2 e- x , where c is to be determined. Then
y� c[2xe-x - x2 e-x ],
Y; c[2e-x - 4xe-x + x2e-x],
y;' c[-6e-x + 6 xe-x - x2 e-x ],
⇒ -6c = 27 ⇒ c = - 29 ⇒ YP = - 29 x2e -x ⇒
9
y = --x2 e-x + c 1 e-x + C2 Xe-x + C4e 2x
2
y"' - y' = 8xex ,
indicial polynomial is P(r) = r(r - l)(r + 1). Hence, the nonhomogeneous equation
IS
P(D)y = D(D - l)(D + l)y = 8xex .

The annihilator of g(x) = 8xex of minimal degree is Q(D) = (D - 1)2. Thus,
Q(D)P(D)y (D - 1)2 D(D - l)(D + l)y

D(D - 1)3 (D + l)y = (D - 1)2 [8xex] = 0
Since 1, ex and e-x are solutions of the homogeneous equation P(D)y = 0,

where a and bare to be determined. Then
y� (2ax + b)ex + (ax2 + bx)ex

[ax2 + (2a + b)x + b]ex ,
y� [2ax+ 2a + b]ex + [ax2 + (2a + b)x + b]ex
[ax2 + (4a + b)x + (2a + 2b)]ex,
y�' [2ax+ 4a + b]ex + [ax2 + (4a + b)x + (2a + 2b)]ex
[ax2 + (6a + b)x + (6a + 3b)]ex ,
[ax2 + (6a + b)x + (6a + 3b)]e x - [ax2 + (2a + b)x + b]ex = 8xex
⇒ 4ax + ( 6a + 2b) = 8x
⇒ a = 2 and b= -6 ⇒ YP = (2x2 - 6x)ex ⇒
y
111
- 3y" + 7y' - 5y = 32ex sin(2x),
which has been solved by variation of parameters in Section 4.2.1, Example 4 2. 2.
The indicial polynomial is P(r) = (r - l)(r 2 - 2r + 5) . Hence, the nonhomogeneous
equation is
P(D)y = (D - l)(D2 - 2 D + 5)y = 32ex sin(2x).
Since ex cos(2x) and ex sin(2x) are solutions of a homogeneous equation for which
the indicial equation has the roots r = 1 ± 2i, r - 1 = ±2i ⇒ (r - 1)2 = -4 ⇒
r2 - 2r + 5= 0 is the indicial equation. Hence, the annihilator of g(x) = 32ex sin(2x)
of minimal degree is Q(D) = D2 - 2D + 5. Then
Q(D)P(D)y ( D2 2D + 5)(D - l)(D2 - 2D + 5)y

-
(D - l)(D2 - 2D + 5)2 y = (D2 - 2D + 5)[32ex sin(2x)] =0

⇒ y = c 1 ex + c2 cx cos(2x) + c3ex sin(2x) + C4Xex cos(2x) + C5Xex sin(2x).
Since c1 ex +c2 ex cos(2x) +c3ex sin(2x) satisfies the homogeneous equation P(D)y = 0,
where a and b are to be determined. Then

y� =ex[acos(2x) + bsin(2x)] + xex [acos(2x) + bsin(2x)]
+xex[-2asin(2x) + 2bcos(2x)]
e"[acos(2x) + bsin(2x)] + xex[(a+ 2b) cos(2x) + (b- 2a) sin(2x)],
x x
y; - e [acos(2.T) + bsin(2x) ] + e [-2asin(2x) + 2bcos(2x)]
+ex [(a + 2b) cos(2x) + (b- 2a) sin(2x)]
+xex [(a+ 2b) cos(2x) + (b- 2a) sin(2x)]
+xex [(-2a- 4b) sin(2x) + (2b- 4a) cos(2x)]
ex[(2a + 4b) cos(2x) + (2b- 4a) sin(2x)]
+xex[(4b- 3a) cos(2x) + (-4a- 3b) sin(2x)],
Ill
Yp ex[(2a + 4b) cos(2x) + (2b- 4a) sin(2x)]
+e x[(-4a- 8b) sin(2x) + (4b- 8a) cos(2x)]
+ ex[(4b- 3a) cos(2x) + (-4a- 3b) sin(2x)]
+xex[(4b- 3a) cos(2x) + (-4a- 3b) sin(2x)]
+xex[(-8b + 6a) sin(2x) + (-8a- 6b) cos(2x)]
ex[(12b - 9a) cos(2x) + (-12a- 9b) sin(2x)]
+xex[(-lla- 2b) cos(2x) + (2a- llb) sin(2x)],
11 1
Yp - 3 Yp + 7 Yp - 5 YP = ex cos(2x)[(12b- 9a) - 3(2a + 4b) + 7a]
Ill
+ ex sin(2x)[(-12a- 9b) - 3(2b- 4a) + 7b]

+xex cos(2x)[(-lla - 2b) - 3(4b- 3a) + 7(a + 2b) - 5a]
+xex sin(2x)[(2a - llb) - 3(-4a- 3b) + 7(b- 2a) - 5b]
32 ex sin(2x)
-8aex cos(2x) - 8bex sin(2x)
=} = 32 ex sin(2x)
⇒ a= 0 and b = -4 ⇒ YP = -4xex sin(2x) ⇒
y = -4xex sin(2x) + c1ex + c2ex cos(2x) + c 3 ex sin(2x)

y ( 4) - 6y111 + lly" - 6y' = 36 e3x )
indicial polynomial is P(r) = r(r - l )(r - 2 )(r - 3 ). Hence, the nonhomogeneous
equation is
P(D)y = D(D - l)(D - 2)(D - 3)y = 36e3x.
The annihilator of g(x) = 36e3x of minimal degree is Q(D) = D - 3. Thus,
Q(D)P(D)y (D - 3)D(D - l)(D - 2)(D - 3)y

D(D - l)(D - 2)(D - 3)2y = (D - 3)[36e3x ] =0
=? Y = C1 + C2ex + C3e 2x + C4e 3x + C5Xe3x .
Since c 1 + c2ex + C3 e 2x + c4e3x is a solution of the homogeneous equation P(D)y = 0,
YP = c5xe , or YP = axe , where a is to be determined. Then
3x 3x
Y� a[e3x + 3xe3x ],
Y; a[6e3x + 9xe3x ],
y;' a[27e3x + 27xe3x ],
4 a[108e 3x + 81xe 3x ],
Yt )
4 =
Yt ) - 6y�' + lly � - 6y� a[108e3x + 81xe3x ] - 6a[27e 3x + 27xe3x ]
+lla[6e 3x + 9xe3x ] - 6a[e3x + 3xe 3x ]
36e 3x
⇒ 6a = 36 ⇒ a = 6 ⇒ Yv = 6xe 3x ⇒
Exercises 4.2
In Exercises 1-16, employ variation of parameters to find the general solution or to
solve the initial-value problem.
1. y "' - 3y " - y ' + 3y = 8e 2x
2. y "' - 3y " + 2y' = 8xe 2x
3. y "' - 2y" + y ' = -2e x cos(x)

4. y "' - y " + 2y = lOex
5. y111 - y" + y' - y = 4sin(x) + 4cos(x)
6. y '" - y " = 20x3 , y (0) = 20, y '(0) = 0, y "(0) = -20
7. y"' - 2y" = 5sin(x)

8. y( 4) - y" =2
9. y( 4) + y" = 6x
10. x 3y'" - x 2y" + 2xy' - 2y = x, x >0
1
11. x3 y"' + 5x2 y" + 2xy' - 2y = -, x >0
12. x3 y"' + 2x2 y" = 1, x > 0, y(l) = 5, y'(l) = 1, y"(l) = 0

13. x3 y"' + x2 y" - xy' = 4x2 , x >0
14. :r;3 y'" + 2x2 y" - 2xy' = x3 cos(x), x >0
15. x3 y111 + 3x2 y" + 2xy' = 1, x > 0
16. x 3 y111 + 3x 2 y" = x 2 sin(x), x >0
In Exercises 17-25, employ annihilators to determine a particular solution.
1 7. y"' - 3y" - y' + 3y = 8e2x ( see Exercise 1)
18. y"' - y" + 2y = lOex ( see Exercise 4)

19. y'" - y" = 20x3 (see Exercise 6)
20. y"' - 2y" = 5 sin(x) (see Exercise 7)

21. y 111 - 3y" + 2y' = 8xe 2x (see Exercise 2)
22. y111 - y" + y' - y = 4sin(x) + 4 cos(x) (see Exercise 5)
23. y"' - 2y" + y' = -2ex cos(x) (see Exercise 3)
24. y(4) - y" =2 (see Exercise 8)
25. y(4) + y" = 6x (see Exercise 9)
Chapter 4 Exercises
1. Consider the equation y"' - y" - 2y' = 12.
(a) Employ variation of parameters to solve the equation.

(b) Employ annihilators to solve the equation.
(c) Integrate the equation once and solve the resulting second-order equation.
2. Solve the equation x3 y"' + x2y" = x, x > 0.

4 . 2) .
3
3. Solve the equation x y
/II
+ x2 y/I = - , x > 0 ( see Exerc1se
X
4. Integrate the equation y"' - y" = 20x3 twice and solve the resulting first-order
equation (see Exercise 6 in Section 4.2 on page 118).
5. Solve the equation x4y( 4) - 4x3 y"' + 12x2 y" - 24xy' + 24y = x4, x > 0.
6. Solve the equation x3 y111 - x2 y" + lOxy' = 156x2 sin[3 ln(x)], x > 0.
7. Let P(r) be a polynomial with real coefficients. Show that if z is a complex

root, then its complex conjugate z is also a root.
8. Let p(x) and q(x) be continuous on the interval I = (-1, 1). Is it possible for
the equation y" + p(x)y' + q(x)y = 0 to have the solutions y 1 = 1, y2 = x and
y3 = x on I?
3
9. Solve the equation y"' + 3y" + 3y' + y = 2xe-x cos(x) by variation of parameters.
10. Employ annihilators to solve the equation in Exercise 9.

Chapter 5
Linear Systems
Definition 5.1. A set of n first-order differential equations which involve n unknown

functions of an independent variable l, where n 2'. 2, is called a system of first-order
equations. A system is linear if it has the form
x: au (l)x1 + a12(l)x2 + · · · + a1 n (t)xn (l) + !1 (t)

x; a21(t)x1 + a22(t)x2 + · · · + a2n (l)xn(t) + h(l)
(5.1)
=
where x 1, x 2, · · · , X n are unknown functions oft, and aij (l), 1 :S i,j :Sn, and Jj (t),
1 :S j :S n, are given functions of t. A linear system is homogeneous if fj (t) 0,
1 :S j :Sn. Otherwise, it is nonhomogeneous. It has constant coefficients if aij (t) = aij
is constant, 1 :S i, j :S n.
For example, if n = 2, writing x 1 = x and x 2 = y for convenience, the system
x' 3x + 2y
y' 2x -y
is linear and homogeneous with constant coefficients, whereas
X
1
X - 3y + t 2
y' 6x + 4y + e l
is linear and nonhomogeneous with constant coefficients.
Definition 5.2. A solution of a system is a set of n differentiable functions x 1 (t),

x2 ( t), · · · , X n ( l) which satisfy every equation in the system for all t in an interval I.
121
122 CHAPTER 5. LINEAR SYSTEMS
For example, x(t) = 2e t , y(t) = et is a solution of the system

x' 3x - 4y
y' x-y (5.2)
because
x'(l) 2i = 3(2et ) - 4(et ) = 3x(t) - 4y(t)

y' ( t) et = 2et - et = x (l) - y ( t)
for all t in the interval I = JR.
In the following sections, methods will be developed for the determination of all
solutions of linear systems with constant coefficients. This is achieved most efficiently,
especially for large n, by expressing a linear system in matrix form. For example, with
the system (5.2) can be expressed as
and the solution x(t) = 2e t , y(t) = et can be expressed as the vector (or vector-valued)
function
Similarly, the system of three nonhomogenous equations
x' 2x + 3y - z + t 2
y' x - 4z - et
z' -y + 6z + 5
can be expressed as x' =Ax+ f(t) by letting
More generally, the system (5.1) can be expressed as x' = A(t)x + f(t) by letting
5.1. HOMOGENEOUS SYSTEMS 123
au(l) a 1 2(l)
a2 1(t) a2 2(t)
For a homogeneous system, f(t) =0, and, for a system with constant coefficients,
A (l) = A has constant entries aij, 1 :S i, j '.S n.
5.1 Homogeneous Systems

5.1.1 General Theory
Theorem 5.1. Let A(t) be an n x n matrix with entries aij(t) and consider the
homogeneous system x' = A(t)x. If x1(l), x2(t), · · ·, xk(t), k � 1, are solutions, then
any linear combination
is also a solution.
x'(l) (c1XJ (t) + C2 X2 (t) + · · · + Ck Xk(t))'

C1X� (t) + C2x;(t) + · · · + Ckx�(t)
c 1 A(t)x1 (t) + c2A(t)x2 (t) + · · · + ckA(t)xk(t)
A(t)(c1x1(l) + c2x2(t) + · · · + ckxk(l))
A(t)x(t).
Definition 5.3. The vector-valued functions x 1 (l), x2 (t), · · ·, xk(t), k � 1, are

linearly dependent on an interval I if there exist constants c 1, c2 , · · ·, Ck, not all zero,
such that
for all t in I. Otherwise, they are linearly independent on I.
Definition 5.4. Let x1 (l), x2 (t), · · ·, xn (l) be n-vector-valued functions, i.e.,
1 :S i :S n.
The Wronskian W(t) = W[x1(t),x2(t), · · · ,x (l)] is defined by

n
xu(t) X1 2(t)
Xn(t) X22(t)
W(t) = det(x 1 (t) x2(t) · · · Xn(t)) =
Xnn (l)
Theorem 5.2. Suppose that A(t) = (aij(t)) is continuous on an interval I, i.e.,
=
aij(t) is continuous for 1 � i,j � n, and let x1 (t),x2(l), · · · ,xn(t) be solutions of
x' = A(t)x. Then either W(t) 0 on I, or W(t) i= 0 for any t in I. The solutions
x1(t),x2(t), · · · ,xn (t) are linearly independent on I if and only if W(t) i= 0 on I.
Definition 5.5. Let x1 (t),x2(l), · · · ,xn (t) be n linearly independent solutions of

x' = A(t)x. The set {x1 (t),x2(t), · · · ,xn (t)} is called a fundamental set of solutions.
The matrix
xu(t) X12(t)
X2 1 (t) X22(t)
Xnn(t)
l l·
is called a fundamental matrix for the system. The general solution of the system is
x(t) = C1 X1 (t) + C2X2(t) + · · · + Cn Xn (t),
where c 1 , c2, · · ·, Cn are arbitrary constants, and may be expressed more concisely as
x(t) = (x 1 (l) x2 (l) · · · xn(t)) ( : = X(t)c, where c = (:
A solution in which c 1 , c2, · · ·, Cn are assigned particular values is called a particular

solution.
Theorem 5.3. A fundamental matrix X(t) for a system x' = A(t)x, where A(t) is
continuous on an interval I, is nonsingular on I, i.e., det(X(t)) i= 0 on I, and satisfies
the matrix equation X'(t) = A(t)X(t), where X'(t) = (x�/t)).
Proof. Since x1 (t), x2 (t), · · ·, xn (t) are linearly independent, det(X(t)) = W(t) i= 0
on I, and, since x� = A(t)xi, 1 � i � n,
X' ( t) ( x't ( t) x; ( t) · · · x� ( t))
(Ax1(l) Ax2(t) · · · Axn(t))
A(x1(l) X2(t) · · · Xn (l))
AX(t).
Theorem 5.4. Let A(l) be continuous on an interval J. If X(t) is a fundamental

matrix for the system x' = A(l)x, then X(t)C is also a fundamental matrix for any
nonsingular n x n matrix C. If X ( t) and Y(t) arc any two fundamental matrices,
then there exists a nonsingular constant n x n matrix C such that
Y(l) = X(l)C.
Equivalently, X(t) = Y(t)c- 1.
Proof. Z(t) = X(t)C satisfies the matrix equation Z'(t) = AZ(t) because
Z'(l) = [X(l)C]' = X'(t)C = [AX(l)]C = A[X(l)C] = AZ(t).
Since det(X(l)) =I= 0 and det(C) =I= 0, det(Z(t)) = det(X(t)) det(C) =I= 0. Hence,
the
columns of Z(l) arc linearly independent, and each column Zj is a solution of the
system because
(z1(t) z2 (l) · · · Z n (l))'
A(z1(t) z2(t) · · · Z n (l))
(Az1(l) Az2(l) · · · Az n (t))
⇒ z�(t) = Azj(l), 1 :::; j :::; n.
Thus, Z ( l) is a fundamental matrix. If X ( l) and Y ( t) are any two fundamental
matrices, let C(t) = X(t)- 1 Y(t). Then X(t)C(t) = Y(t)
⇒ X' (t)C(l) + X(t)C'(t) = Y' (l)
⇒ A(l)X(l)C(t) + X(l)C'(t) = A(l)Y(t)
==> A(l)Y(l) + X(t)C'(t) = A(t)Y(t)
⇒ X(t)C'(t) = 0
= X(l)- 1 X(t)C'(t) = 0.
==> C '(l)
Hence, C(t) = C is a constant matrix, and Y(t) = X(t)C.
Definition 5.6. An initial-value problem for a homogeneous system consists of a

system x' = A(t)x, together with an initial condition x(lo) = x0, which requires that
the solution x(l) be equal to x0 at l = lo, and determines the arbitrary constant
vector c.
Theorem 5.5. Let A(t) be continuous on an interval J containing the point t0 in

its interior. The solution of the initial-value problem
x' = A(t)x, x(lo) = xo,
is given by x(t) = X(t)X(lo)- 1
x0, where X(t) is a fundamental matrix.
Proof. x(t) = X(t)c and x(to) = xo ==> x0 = x(lo) = X(l0)c ==> c = X(t0 )-1 x0 ==>
x(t) = X(t)c = X(l)X(tot 1 xo.
5.1.2 Systems with Constant Coefficients

Let A = ( a ij) be an n xn matrix, where a ij, 1 � i, j � n, are constants, and consider
the system
x'= Ax. (5.3)
Analogous to single equations with constant coefficients, seek solutions in the form
(5.4)
=
where A is a scalar and v =/- 0 is a vector, to be determined. The case v = 0 is
omitted because it yields only the trivial solution x(l) 0. Then
Al
x'(t) = Ae v,
and x'(t) = Ax(t) if and only if Ae At v = Ae Alv, or Av = Av, i.e., A is an eigenvalue

of A and v is a corresponding eigenvector.
The equation Av= Av is equivalent to
(AI - A)v = 0, (5.5)
where I is the n x n identity matrix. Equation (5.5) is a homogeneous algebraic

system, and admits nontrivial (i.e., nonzero) solutions v if and only if the coefficient
matrix Af - A is singular, i.e.,
det(AI - A) = 0. (5.6)
The eigenvalues of A are determined by solving Equation (5.6) for A, and for each
eigenvalue Ai , the eigenvectors which correspond to Ai are obtained by setting A = Ai
in Equation (5.5) and solving for v.
If then x n matrix A admitsn real, linearly independent eigenvectors v1, v2, · · ·,

vn , corresponding to the eigenvalues A1 , A2 , · · · , An (not necessarily distinct), then n
linearly independent solutions of the system x' = Ax are given by
A2 l
X1(l) = e A1
lV1, X2 (t) = e V2, · · · , Xn (t) = e An lVn .
This is automatically true in case A possesses n real, distinct eigenvalues, because

eigenvectors which correspond to distinct eigenvalues are linearly independent. The
cases where A has complex eigenvalues or fewer than n independent eigenvectors will
be di cussed later.
Example 5 .1. Consider the system
{ :: : �: - : �� } .
(5.7)
5.1. HOMOGENEOUS SYSTEMS 1 27
This system may be expressed in matrix form as x' = Ax, where
x= ( : ) and A = ( _ � � ) .
Then
A
; (A- 2)(,\.-7)+4=,\. 2 -9,\.+18
I A-_\ l=
2
det(>J - A)
(,\. - 3)(,\. - 6) = 0
=} ,\. 1 = 3 and ,\.2 = 6 are the eigenvalues of A.
For A1 = 3, (,\. 1 f - A)v= 0 with v= ( � ) ⇒
⇒ a - 2b = 0, and b = l ⇒ v= v1 = ( � ) is an eigenvector which corresponds to

the eigenvalue ,\.1 = 3.
⇒ 2a -b = 0, and a= l ⇒ v= v2 = ( � ) is an eigenvector which corresponds to

the eigenvalue ,\. 2 = 6.
Thus, two independent solutions of the system (5.7) are given in vect9r form by
x 1 ( t) = e 3t (�) 2 t
= ( �: )
e
and x2 ( t) = e 6t (�) = ( ;:;t ),
2e e 6t
and X(t) = (x1(t) x2(t)) = ( 3 6 ) is a fundamental matrix. The general
3t
et 2e t
solution in vector form is
from which the general solution x(t) = 2c 1 e 3t + c2e 6t and y(t) =c1 e 3t + 2c2e 6t of the
system (5.7) is obtained in scalar form. Note that, if vis-an eigenvector of A, then
so is cv for any constant c =/- 0. Thus, the choice of an eigenvector is not unique,
4 2 4e3t 2e6t
and we may take v 1 = ( ) and v2 = ( ) . Then Y(t) = ( 3 ) .1s also
2 4 2e t 4e 6 t
a fundamental matrix. By Theorem 5.4, X(t)- 1 Y(t) = ( � � ) is a constant matrix.
Example 5.2. Consider the system

y
{ x ; :_ -x + 3 }
y (5.8)
-4x + 6y
Then
det(>-.J - A) I ,\: 1 ,\ -=-\ I = (>-. + 1)(,\ - 6) + 12 = ,\ 2 _ 5,\ + 6

(,\ - 2)(,\ - 3) = 0
⇒ ,\ 1 = 2 and ,\2 = 3 are the eigenvalues of A.

For ,\ 1 = 2, (,\if - A)v = 0 with v = ( � ) ⇒
⇒ 3a 3b = 0 , and a= l ⇒ v = v 1 = ( � ) is an eigenvector which corresponds

�
to the eigenvalue ,\ 1 = 2.
For ,\2 = 3, (,\2 J - A)v = 0 with v = ( � ) ⇒
( : =� ) ( � ) = ( � )
⇒ 4a - 3b = 0, and a = 3 ⇒ v = v2 = ( � ) is an eigenvector which corresponds

to the eigenvalue >-.2 = 3.
Thus, two independent solutions of the system (5.8) are given in vector form by
1 e2t ) ( 3 3e3t )
x1(t) = e2t (
1
) = (
e2t
and x2 (l) = e3t 4
) = (
4e 3t ,
e2t 3e3t
and X (t) = (x 1 (l) x2 (l)) = (
e 2 t 4e 3 t
) is a fundamental matrix. The general
from which the general solution x(t) = c 1 e2t + 3c2 e3t and y(t) = c 1 e2t + 4c2 e3t of the
system (5.8) is obtained in scalar form.
(5.9)
Then
=I >.; 1 � I = (>. + 1)
det(>.J - A)
>. 1 =0
2
⇒ >. 1 = -1 is the only eigenvalues of A, with multiplicity 2.

For >.1 = -1, (>. 1 1 - A)v = 0 with v = ( � ) ⇒
which states that both a and b are free. Hence, a = 1 and b = 0 gives the eigenvector
v1 = ( t ), and a = 0 and b = 1 gives the eigenvector v2 = ( � ) . Since v1 and
v2 are linearly independent, two independent solutions of the system (5.9) are given
in vector form by
t
e � ) is a fundamental matrix. The general
( ;
e t
from which the general solution x(t) = c 1 e-t and y(t) = c2 e-t of the system (5.9) is
obtained in scalar form.
Note that, even though the 2 x 2 matrix A has only one eigenvalue A 1, the general
solution is obtained as in the preceding examples because there are two independent
eigenvectors which correspond to ,\ 1 .
A system such as (5.9) is said to be decoupled because each equation involves only one
of the unknown functions, and can be solved independently of the other(s). Thus,
=
x' = -x ⇒ x(t) = c1 e-t, and y' = -y ⇒ y(t) = c2 e-t, as above.
{ x;
y -
3x +
- 4X -
2y } ' x(0)
y
= 1, y(0) = 5. (5.10)
This problem may be expressed in matrix form as x' = Ax, x(0) = xo, where
X
= (: ) ' A= ( -� -� ) and Xo
=( !).
Then
I ,\ � I = (>. - 3)(>. + 2) + 4 = ,\2 - A - 2

3
det(,X.J - A) ,\�\
(>. + 1)(>. - 2) = 0
⇒ A1 = -1 and >.2 =2 are the eigenvalues of A.
⇒ 4a + b = 0, and a= l ⇒ v = v1 = ( _! ) is an eigenvector which corresponds

to the eigenvalue >-1 = -1.
⇒ -a - b = 0, and a = l ⇒ v = v2 = ( -i ) is an eigenvector which corresponds

to the eigenvalue >. 2 = 2.
Thus, two independent solutions of the system (5.10) are given in vector form by
e -t e2t )
and X(t) = (x 1 (l) x2(l)) =
( - 4e -t
-e 2t
is a fundamental matrix. The solution
of the initial-value problem (5.10) is then given by x(t) = X(l)X(o)- 1 xo.
Note that the inverse of a 2 x 2 matrix ( � � ) is obtained simply as
( a b )-
c d
1
1
- det(A)
(
d -b )
-c a -
1
(
d -b )
ad - be -c a ·
Thus, X(o)- 1 = ( _! -� )- 1 ( -! 1
=
-
�),and
x(i) = (
-1 -1
4 1
) (
1
5 )
= (
e -t e2t
-4e-t -e2t
)(- 2
3 )
i.e., x(t) = -2e-t + 3e2t and y(t) = 8e-t - 3e2t .
{ -=
x' - 2y + 2z
y' -3x + y + 3z } (5.11)
z' = -x + y + 3z
-
and A= ( -� � � ) .
-1 1 3
Then
>- 2 -2
det(,H - A) 3 >- - 1 -3
1 -1 >- - 3
>-[(>- - 1)(>- - 3) - 3] - 2[3(>- - 3) + 3] - 2[-3 - (>- - 1)]
>-[>-2 - 4>-] - 4), + 16
>-2 (>- - 4) - 4(>- - 4)
(>- - 4)(>-2 - 4)
(>- - 4)(>- - 2)(>- + 2) = 0
⇒ >-1 = 2, >-2 = -2 and A3 = 4 are the eigenvalues of A.
For>.1 = 2, (>. 1 1 - A)v = 0 withv = ( � ) =>
Row reduction of the coefficient matrix (R2 - R2 - 3R3 and R3 - R1 2R3, where
� denotes row i) then gives
-
=> b = 0 and a + b- c = 0 => a = c, and a = 1 => v1 = ( i ) is an eigenvector
which corresponds to the eigenvalue >- 1 = 2.
For>., = -2, (>.2 1-A)v=Owithv= ( �) =>
Row reduction of the coefficient matrix gives

⇒ c = 0 and -a + b - c = 0 ⇒ a= b, and a= 1 ⇒ v2 = ( i) is an eigenvector
which corresponds to the eigenvalue >-. 2 = -2.
For ,\3 = 4, (,\3/ - A)v =0 with v = (�) ⇒
=� ) ----+ ( � � =� ) ----+ ( � -� -� )
(1 �
1 -1 1 1 -1 1 0 3 -3
⇒
0 -i :DU) 0)
⇒ -b+c = 0 and 2a+b-c = 0 ⇒ a= 0, and b = 1 ⇒ v3 = (:) is an eigenvector
which corresponds to the eigenvalue >-.3 = 4.
Thus, three independent solutions of the system (5.11) are given in vector form by
e2t e- 2t o)
and X(l) = (x1(t) x2(l) x3 (t)) = 0 e- 2
e4t is a fundamental matrix. The
,c�
( t
e 2t
O e4t
general solution of the system (5.11) is
x(t) C1X1(t) + C2X (t) + C3X3(i)
2
e�' )( �:)
X(t)c
e- 2 l c,e"He-")
e- 2 t C2 e- 2l + C3e 4l
,
e t
2
0 e 4l
C3 C1 e 2l
+ C3e
4t
{ -
x'
y' = (5.12)
z' =
(n n
(J
0
x= and A= 1
0
Then
,\-1 0 0
det(,\J - A) = 0 ,\-1 0 = (,\ - 1) 2 (,\ - 2) = 0
0 ,\-2
=? ,\ 1 = 1, of multiplICity 2, and ,\ 2 = 2 are the eigenvalues of A.
For A 1 = 1, (A 1 I - A)v = 0 with v = ( � ) =>
=> a - c = 0 => a = c and b is free. Thus, a = 1 and b = 0 => v 1 = ( � ) , and
a = O and b = 1 => v 2 = ( [ ) axe independent eigenvectors which correspond to
the eigenvalue ,\ 1 = 1.
For A 2 = 2, (A 2 J - A)v = 0 with v = ( � ) =>

=> a - 0 and b - 0 , and c - I => v 3 - (�) is an eigenvector which corresponds
to the eigenvalue A2 = 2.
Thus, three independent solutions of the s ystem (5.12) are given in vector form by
et 0 0 )
and X(l) = (x1(l) x 2(l) x3(t)) = ( 0 et O is a fundamental matrix. The
et 0 e2t
general solution of the system (5.12) is
x(t)
Complex Eigenvalues
Let A be an n x n matrix with real entries, and suppose that A = a+ i(3, (3 =/- 0, is
a complex eigenvalue. Then "X = a - i(3 is also an eigenvalue since det( AI - A) is a
polynomial with real coefficients . If v_is an eigenvector corresponding to A, then vis
an eigenvector which corresponds to A because
Av= AV ⇒ Av= Av ⇒ Av= "Xv ⇒ Av= Av.
Then z(t) = e>.tv and z(t) = e>-t v are two complex solutions of the system x' = Ax.
Let v = a + ib, where a and b are real vectors. Then v = a - ib and, employing
Euler's identity, we obtain
e(a+i,B)t[a + ib] = e e ,Bt [a + ib]

°'t i
z(l)
e°'t [cos(,6t) + i sin(,6t) ][a + ib]
e°'t{[cos(,6t)a - sin(,6t)b] + i[sin(,6t)a + cos(,6t)b]},
z(l) °
e 't{[cos(,6t)a - sin(,6t)b] - i[sin(,6t)a + cos(,6L)b]}.
Since the system is linear and homogeneous, any linear combination of solutions is a
solution, by Theorem 5.1. Thus,
z(t) + z(t)
= �(z(t)) = e'�t [cos(,Bt)a - sin(,Bt)b],
2
z(t) z(t)
Xz(t) = Ss(z(t)) = e0t [sin(,8t)a + cos(,Bt)b]
;
are two real solutions which are linearly independent.
Example 5. 7. Consider the system

}
{ :_x
y;
2x
-3x
+
+
3y
2y
(5.13)
Then
>- 2 -
det(>J - A) = \ ; -\ I = (>- - 2) 2 + 9 = >-2 4>- + 13 = 0
>-
-
⇒ >-, >- = 4 ± J=36

-
= 2 ± 3·i are the comp1ex e1genva1ues of A.
.
2
For >- = 2 +. 3i, (>-I - A)v = 0 with v = ( � ) ⇒
⇒ 3ia - 3b = 0, and a= l ::::::}
is an eigenvector which correspo,nds to the eigenvalue >- = 2 + 3i. Then
z(t) e2t [cos(3t) + i sin(3t)] [ ( � ) +i ( � )]
e 2t { [cos(3t) ( � ) - sin(3t) ( � ) ] +i [sin(3t) ( � ) + cos(3t) ( � )] }

1 7
is a complex solution, and
x1(l) R(z(t)) = 3
e 2t [cos( t) ( �) -sin( l) ( �)] 3
cos( t) 3e 2t cos(3t)
e 2t (
- sin( l) ) = (
3
-e2t sin( t) ) ' 3
x 2(l) �(z(t))= e 2 t [sin( t) ( �) 3 + cos(3 t) ( �) ]
t sin(3 l)
= ( e2 sin( t )
2t
3
e2 (
cos( t) ) 3e tcos(3l) )
are two real, independent solutions of the system (5.13).
= cos(3l) sin( t ) = e 2tcos(3l) 3 3

e 2t sin( l)
X()t (x i()l x 2(t))= e 2t (
-sin( l) cos( t ) ) 3
( -e 2t sin(3t) 3 3
e 2tcos( t) )
is a fundamental matrix, and the general solution is
x(t)
. .
c1x1(t) + c 2x 2(t) = X(l)c = e 2t (
cos( t) sin( t) c1 3 3
-sin(3t) cos(3l) ) ( c 2 )
3
c1 cos( t) + c 2 sin(3 l) c1e 2t cos(3t) + c 2 e 2t sin( t) 3
e 2t (
-c1 sin(3 t) + c 2 cos(3t) ) = (
3
-c1 e 2t sin( t) + c 2 e 2t cos( t) ) ' 3
i.e., x(t) = e l[c1 cos(3 t) + c2 sin(3 t)] and y(t) = e t[-c1 sin(3 t ) + c
2 2
3 t)].
2 cos(

+ 13y .
{ ;: : �: } (5.14)
This system may be expressed in matrix form as x'= Ax, where
Then
det(>J - A)= I
>-
�
4
-1 3 1 3
= >-(>- - 4) + 1 = >- 2 - 4 >- + 13 = 0
⇒ -
>-, >- =
4± \/-36 = 2 ± 3i are the complex eigenvalues of A.
2
For >- = 2 + 3i, (>-[ - A)v= 0 with v = ( � ) ⇒
( -2 + 3i -1 3 ) (
a
)= (
0)
1 2+ 3i b 0
⇒ a+ (2 + 3i)b = 0, and b = -1 ⇒
is an eigenvector which corresponds to the eigenvalue ,\ = 2 + 3i. Then
z(t) e 2l[cos(3t) + i sin(3t)] [ ( !1 ) + i ( � ) ]

e 2l {[ !1 ) - sin(3t) ( � ) ]
cos(3t) (
+i [sin(3t) ( ! ) + cos(3t) ( � ) ]
1 }
x 1 (t) = �(z(t)) = e 2l [cos(3t) ( !1 ) -sin(3t) ( �)]

e2 t( 2cos(3t) - 3sin(3l)
=
e 2l [2cos(3l) - 3sin(3t)]
-cos(3t) ) ( -e 2l cos(3t) )'
x 2(t) = �(z(t)) = e 2l [sin(3t) ( !1 ) + cos(3t) ( � ) ]

e2 t( 2sin(3t) + 3cos(3t)
) = (
e 2t[2sin(3t) + 3cos(3t)]
)
-sin(3t) -e 2t sin(3l)
are two real, independent solutions of the system (5. 1 4).
X(t) (x1(t) x 2(t))
2cos(3t) - 3sin(3l) 2sin(3l) + 3cos(3t)
e 2t ( )
-cos(3t) -sin(3t)
e 2t[2cos(3t) - 3sin(3t)] e 2t[2sin(3t) + 3cos(3t)]
= ( )
-e 2t cos(3t) -e2t sin(3t)
is a fundamental matrix, and the general solution is
(
x(l) c1 x1(t) + c 2 x 2(t) = X(t)c
2cos(3t) - 3sin(3t) 2sin(3t) + 3cos(3t) c1
e 2l ) ( c2 )
(
-cos(3l) -sin(3t)
ci[2cos(3t) - 3sin(3t)] + c 2 [2sin(3t) + 3cos(3l)]
e2 t )'
-c1 cos(3l) - c 2 sin(3t)
i.e.'
x(t) e 2l [(2c 1 + 3c 2) cos(3t) + (2c 2 - 3c1) sin(3t)],

y( t) = e 2l[-c1 cos(3t) - c2 sin(3t)].
{ X
y'
1
=
= 5
+
_ x + y}
5
X y
, x(0) = 1,
y
(0) = 3. (5.15)
This problem may be expressed i n matrix form as x' = Ax, where
x = ( ; ) , A= ( � � ) and x(O) = ( � ) .
-
The n
>. 1 1
det(>J - A) = ; -_ = (>. - 1)(>. - 5) + 5 = >.2 - 6>. + 10 = 0
>. 5
I 1
⇒ >., ->. = 6±R

2
= 3 ± i are the complex eigenvalues of A.
For >. = 3 + i, (>.I - A) v = 0 with v = ( � ) ⇒

2+i
(
-1.)(a o
5 -2+i 0
)=( )
b
⇒ (2 + i)a - b = 0, and a = 1 ⇒
i s an eigenvector which corresponds to the eigenvalue >. = 3 + i. Then
z(t) = e3t [co s(l) +i sin(t)] [ ( � ) + i (� )]
e3t {[cos(t) ( � )- sin(t) ( � )] +i [ sin(l) ( �) +co s(t) ( �) }

]
x 1 (t) = R(z(l)) = e 3t [co s(t) ( � ) - si n(t) ( �)]
e 3t
( 2cos(t) � si n(l)
s )
)
'
x 2 (t) �(z(t)) = e [sin(t) ( � ) + co s(t) ( �)]

3t
e
3t
( 2sin(�\ f �os(t)
n
)
are two real, independent solutions of the system (5.15) . Thus,
_ (xi (t ) x2 (t) ) _
X(t) - - e 3t
( cos(t ) sin(t )
2cos(t) - sin(t ) 2sin(t ) + cos(t ) )
is a fundamental matrix,
X(0) = ( � � ) , X(0)- 1 = ( � � ),
-
and x(O) = ( ; ) ⇒
x(t) X(t)X(o)- 1 x(0)

= e 3t (
cos(t) sin(t ) 1 0 1
2cos(t ) - sin(t ) 2sin(t ) + cos(t ) ) ( -2 1 ) ( 3 )
- e3t ( 2cos (t) � sin(t ) 2sin(�� f �os(t )

s ) n
) ( )
�
= e3
t ( cos(t) + sin(t )
3cos(t) + sin(t) ) '
i.e., x(t) = e3t [cos(t ) + sin(t )] and y(t) = e3t [3cos(t ) + sin(t ) ].
(5.16 )
{ �;z' =: -4x
�: +: 4y4� +: 3:z } .
Then
A+ l -4 -3
det(AI - A) = -2 A - l -1
4 -4 A - l
= (A+ l)[(A - 1)2 - 4] + 4[-2(.A. - 1) + 4] - 3[8 - 4(.A. - l)]
(A+ 1)[.A.2 - 2,,\ - 3] + 4A - 12
(A+ l)(A - 3)(A + 1) + 4(.A. - 3)
(A - 3)[(.A. + 1)2 + 4]
= (A - 3)[.A. 2 + 2.A. + 5] = 0
- -2± J=R
=} A1 = 3 and A2, A2 = = -1 ± 2i are the eigenvalues of A.
2
For .\ 1 = 3, (.\,I - A)v = 0 with v = ( � ) ⇒
⇒ c = 0 and a-b = 0, and a= 1 ⇒ v 1 = ( i) is an eigenvector which corresponds
to the eigenvalue A 1 = 3. Hence,

3l
e
( e 3l )
0
is one solution of the system (5.16).
For.\2 = -1+2i, (.\2 /-A)v = Owithv =

(
!) ⇒
( 2i -4
-2 2i- 2
4 -4
( 2i -4 -3 ) 2i -4
(
-2 2i - 2 -1 ---+ 0 -2i- 6
4 -4 2i - 2 0 -8i-4
==> 2b + c = 0 ==> c = -2b and 2ia - 4b - 3c = 0 ==> 2ia + 2b = 0, and b = 1 ==> c = -2

and a= i ==>
is an eigenvector which corresponds to the eigenvalue A.2 = -1 + 2i. Then

- sin(2t) -e-l sin(2t)
x 2(t) = R(z(t)) = e-t ( cos(2t) ) = ( e-tcos(2t) ) ,
-2 cos(2t) -2e-l cos(2t)
cos(2t) e-tcos(2l)
x3(t) = S'(z(t)) = e-l ( s�n(2t) ) = ( e-t s�n(2t) )
-2 sm(2t) -2c t sm(2t)
are two real, independent solutions of the system (5.16). Thus, three independent
solutions are x1(t), x2(t) and x3(t),
e3t -e-l sin(2t)
X(t) = (x1(t) x 2 (t) x3 (l)) = ( e3l e-tcos(2t)
0 -2e-tcos(2l)
is a fundamental matrix, and
x(l)
i.e.'
x(l) c 1 e3l - c2 e-t sin(2t) + c3e-l cos(2l),

3l
y(l) c 1 e + c2 e-l cos(2l) +c3 e-t sin(2t),
z(l) -2c2 e-l cos(2t) - 2c3e-t sin(2l)
is the general solution of the system (5.16).
(5.17)
U
( i) -�)
0 -4
0 0
X= and A=
0 0 0
8 0 0
Then
,\0 4 0
,\0 2 0 ,\ 2
0 A 0 2
det(,\J - A) =.A. 0 ,\ 0 +4 -1 0 0
-1 0 A 0
-8 0 ,\ 0 -8 ,\
0 -8 0 ,\
,\[,\(,\ 2 + 16)] + 4[,\2 + 16]
(,\2 +16)(.A.2 +4) = 0
=> .A. 1, >: 1 = ±2i and A 2 , >:2 = ±4i arc the eigenvalues of A.
For>. 1 = 2i, (A 1 J - A)v = 0 with v = ( ; ) =>
(t -1
0
2i
0
4
0
2i
0 -8 0
(
0 4 0 2i O 4 0 2i 0 4
2i O 2 0 2i O 2 0 2i 0
=>
-1 0 2i O --t O O O O 0 0 0
) ( ) (
�0 -8 0 2i O O O 6 0 0 0
=> d = 0 => b =0 and 2ia + 4c = 0, and c = 1 => a = 2i =>
0 n l
is an eigenvector of A which corresponds to the eigenvalue >. 1 = 2i. Then
(t) = [cos(2t) + sin(2t)]
! )- !
i
Z1
i
[ ) + (
= [cos(2t) ( sin(2t) ( � ) ] + i [sin(2t) ( ) + cos(2t) ( � ) ]
l(
0 2 -2 sin(2l)
l
!!l(z 1 ( t)) = cos(2t) - sin(2t) (
= cos!2t)
( [ ) ( � ) ) ,
!
[
0 2 2 cos(2t)
'J(z 1 (t)) = sin(2t) + cos(2t) = sin:2t) )
( ) ( � )
[
ForA2 = 4i, (,\ 2 1 - A)v = 0 with v = (; ) =>


4i o 4
0 4i 0
( -1
0 4i
0 -8 0
* u �
u n1
is an eigenvector of A which corresponds to the eigenvalue >.2 = 4i. Then
z,(l) - [coo(4l) + isin(4l)] [ ) +i(
- [cos(4l) (�)-sin(4l) (;)1 +i [sin(4t) (�)+co (4t)

(;) 1
)1
1
� si�(4t)
x 3(t) = �(z2(l)) = [cos(4t) ( )-sin(4t) ( � = (- ),
2 0 2cos(4t)
� coo�4t)
x4(t) = 8'(z2(l)) = [sin(4t) ( ) + cos(4t) ( � ) = ( )
2 0 2sm(4t)
are two real, independent solutions of the system (5.17). Thus, x 1 (t), x2(t), x3(l) and
X4 (l) are four real, independent solutions,
-2sin(2t) 2cos(2l) 0 0 )
X(t) - O
- ( cos(2t)
0 -sin(4l) cos( 4t)
sin(2l) 0 0
0 0 2cos(4t) 2sin(4t)

x(t) C1X1(l) + C2X2(t) + C3X3(t) + C4X4(t)
X(l)c
(
-2 sin(2t) 2 cos(2t) 0 0
0 0 - sin(4t) cos(4t)
cos(2t) sin(2t) 0 0
0 0 2 cos(4l) 2 sin(4t)
i.e.,
x(l) -2c1 sin(2t) + 2c2 cos(2t),
y(t) -c3 sin(4t) + c4 cos(4t),
z(t) c1 cos(2l) + c2 sin(2t),
w(t) 2c3 cos(4l) + 2c4 sin(4t)
is the general solution of the system (5.17).
Generalized Eigenvectors
Consider the system x' = Ax, where A is a 2 x 2 matrix. If A possesses two distinct
eigenvalues >-1 and >-2 , with corresponding eigenvectors v 1 and v2, then two linearly
independent solutions of the system are
.\ 2 t
x1 (t) = e
.\ 1 t
v1 and x2(l) = e v2.
If A possesses only one eigenvalue >.1 of multiplicity two, i.e., det(>.I -A)= (>. - >.1) 2 ,
and if A has two independent eigenvectors v1 and v2 corresponding to >.1, then two
independent solutions of the system are
x1(t) = e .\ 1 tv1 and x2(t) = e .\ 1 t v2.
More generally, if an n x n matrix A possesses n independent eigenvectors (real or
complex), then n independent solutions of the system x' = Ax can be determined by
the methods already discussed.
If an n x n matrix A admits only k linearly independent eigenvectors, where k < n,

which occurs whenever an eigenvalue of A of multiplicity m > 1 has fewer than m
independent eigenvectors, then the methods employed thus far yield only k linearly
independent solutions of the system x' = Ax, and additional independent solutions
are required in order to obtain all n independent solutions.
Given an n x n maLrix A, det(>J - I\) is a polynomial of degree n, called the

characteristic polynomial of A. If >.1, >.2, ·· ·, Ar are the distinct eigenvalues of A,
with multiplicities m 1, m2, ···, mr , respectively, then
with
m1 + m2 + · · · + mr = n.
Thus, n independent solutions of the system x' = Ax can be determined provided
that, for every eigenvalue Ai of multiplicity mi, a total of mi independent solutions of
the system can be found. Several cases are possible, and will be considered for n = 2,
n = 3 and n = 4, from which the suitable procedures for any n 2:: 2 can be deduced.
Let A be a 2 x 2 matrix with only one eigenvalue A 1 of multiplicity 2, and with

only one independent eigenvector v1. Then one solution of the system x' = Ax is
x1(t) = e >-1t v1. A second, independent solution x2(l) is sought in the form
(5.18)
where u0 and u 1 arc vectors to be determined. Then
and, hence, x;(t) = Ax2 (t) if and only if
A 1 e >-1t (tuo + u 1 ) + e >-1t uo = Ae>- 1t (tuo + u 1 ).

Division by e >-1t =f. 0 yields the condition
which holds for all real l if and only if
(5.19)
The first condition in (5.19) is equivalent to
i.e., uo is an eigenvector of A corresponding to the eigenvalue A 1 . We may, therefore,

take uo = v 1 , or uo = cv 1 for any constant c =f. 0. The second condition in (5.19) is
equivalent to
(5.20)
A solution u1 of Equation (5.20) is called a generalized eigenvector of the matrix A
corresponding to the eigenvalue A 1 .
Note that Equation (5.20) is a nonhomogeneous algebraic system with a singular

coefficient matrix A - A 1 I, and such a system may have no solution, in general. In
the present case, however, it can be shown that, since u0 is an eigenvector which
corresponds to A 1 of multiplicity m 1 > 1, a solution u 1 exists.
Thus, two independent solutions of the system x' = Ax are given by

x1(t) = eA 1 t v1 and x2(t) = e A 1 t (tu0 + u 1),
where uo is an eigenvector and u 1 is a generalized eigenvector corresponding to the
eigenvalue ,\ 1 of multiplicity 2.

{ x: -
y =-
5x
4x
-+ }
y
y
(5.21)
Then
I ,\�/ ,\ � 1 I
det(>.I - A) = (>. - 5)(>. - 1) + 4 = ,\2 _ 6,\ + 9
(,\. - 3) 2 = 0
=} ,\ 1 = 3 is the only eigenvalue of A, of multiplicity 2.
=} -2a + b = 0, and a = l =} v = v 1 = ( � ) is an eigenvector which corresponds

to the eigenvalue ,\ 1 = 3. Thus, one solution of the system (5.21) is
x 1(t) = e3l (�) = ( ;::t ).

Since A has only one independent eigenvector which corresponds to the eigenvalue ,\ 1
of multiplicity 2, a second, independent solution of the system must be sought in the
form
x2(t) = e 3t (tuo + u1),
where u0 = v 1 is an eigenvector corresponding to ,\ 1 and u 1 is a solution of the
algebraic system (A - >. 1 I)u 1 = u0, i.e., a generalized eigenvector. Employing the
fact that (A - >-1!) = -(>.1I - A), (A - >-1I)u1 = uo with u1 = ( � ) =}
=> 2a - b= l, and a= 0 => b= -l => u 1 = ( � ) . Hence,

l
x�(t)= e 3t [t ( � ) + ( �l ) ] = e
3l
( 2l � 1 )
.
Thus, two independent solutions of the system (5.21) are given by
and
is a fundamental matrix. The general solution is
x(l) c1x1(t) + c2x2(l) = X(l)c
(� )'
l 3l
e
3
t 2t � 1 ) ( �� )
=( 2c1e�i� c:( �; �
t
l e
3
t
)
{ x'
y'
=
=
-5x
-x -
+ 4y } .
y
(5.22)
Then
A
det(>J - A) I \ l=(,\.+5)(,\.+1)+4=,\.2 +6,\.+9
S
;
A�
(,\. + 3) 2 = 0
=> ,\.1 = -3 is the only eigenvalue of A, of multiplicity 2.
For ,\. 1 = -3, (,\.1/ - A)v= 0 with v = ( � ) =>

⇒ a - 2b = 0, and b = l ⇒ v = v 1 = ( � ) is an eigenvector which corresponds to

the eigenvalue A1 = -3. Thus, one solution of the system (5.22) is
Since A has only one independent eigenvector which corresponds to the eigenvalue ). 1
form
x2(t) = e- 3t(tuo + u1),
where u0 = v1 is an eigenvector corresponding to ). 1 and u 1 is a solution of the
algebraic system (A- ,,\ 1 I)u 1 = u0, i.e., a generalized eigenvector. With u 1 = ( � ),
⇒ -a+ 2b = 1, and b = 0 ⇒ a= -1 ⇒ u 1 = ( �
l
). Hence,
Thus, two independent solutions of the system (5.22) are given by
and
If A is a 3 x 3 matrix with fewer than three independent eigenvectors, then there

are three possibilities, depending upon the number of distinct eigenvalues and the
number of independent eigenvectors which correspond to each eigenvalue.
Let A be a 3 x 3 matrix with only two distinct eigenvalues, >.. 1 of multiplicity

m 1 = 1 and >.. 2 of multiplicity m 2 = 2, and with only two independent eigenvectors v1
and v2 , corresponding to >.. 1 and >.. 2, respectively. Then three independent solutions
of the system x' = Ax are given by
x1(l) = e >qt v1, x2(t) = e >- 2t v2 and x3(t) = e >- 2 t(tuo + u1),
where u0 is an eigenvector which corresponds to >..2 and u 1 is a generalized eigenvector
which corresponds to >.. 2, i.e., u 1 is a solution of the algebraic system (A->..2 J)u 1 = u0.
{
x' = 2x
y' =
+ 3y
3y -- 9z
4z } . (5.23)
z' = y z
=� ) .
and A = ( � �
0 1 -1
Then
>.. -2 -3 9
det(>..I - A) 0 >.. - 3 4 = (>.. - 2)[(>.. - 3)(>.. + 1) + 4]
0 -1 >.. + 1
2
(>.. - 2)(>.. - 2>.. + 1) = (>.. - 2)(>.. - 1)2 = 0
⇒ >.. 1 = 2 of multiplicity 1 and >..2 = 1 of multiplicity 2.
For,\, = 2, (,\1J - A)v = 0 with v = ( ; ) cc>
cc> -b + 4c = 0 and -b + 3c = 0 cc> b = c = 0, and a= l cc> v = v1 = ( �) 1s
an eigenvector which corresponds to the eigenvalue >.. 1 = 2. Thus, one solution of the
system (5.23) is
For >. 2 = I, (>.2 1 - A)v = 0 with v = ( �) =>
⇒ -b + 2c = 0 and -a - 3b + 9c = 0 ⇒ b = 2c and -a+ 3c = 0, and c = 1 ⇒

v = v, = ( � ) is an eigenvector which corresponds to the eigenvalue >.2 = I. Thus,
a second, independent solution of the system (5.23) is
Since A has only one independent eigenvector which corresponds to the eigenvalue .A. 2
of multiplicity 2, a third, independent solution of the system must be sought in the
form
where u0 = v2 is an eigenvector corresponding to >-2 and u 1 is a solution of the
algebraic system (A- A2 J)u1 = Uo, i.e., a generalized eigenvector. With u, = ( � ) ,
=> a+ 3b - 9c =3 and b - 2c = I, and c =0 => b = I and a= 0 => u, = ( !}

Hence,
Thus, three independent solutions of the system (5.23) are given by

and
x(t) =
Let A be a 3 x 3 matrix with only one eigenvalue )q of multiplicity m 1 = 3, and

with only two independent eigenvectors v 1 and v2 corresponding to ,\. 1 . Then two
independent solutions of the system x' = Ax are given by
A third, independent solution is sought in the form
where, as before, uo is an eigenvector which corresponds to ,\. 1 and u 1 is a generalized

eigenvector which corresponds to A 1, i.e., u1 is a solution of the algebraic system
(A - ,\. 1 I)u 1 = u0. In the present case, since there are two independent eigenvectors
v 1 and v 2 corresponding to A 1 , we must set u0 = av 1 + /3v2, and choose a and /3 in
order that the nonhomogeneous algebraic system
be consistent and have a solution u 1 .
r
= X + y - 2z
y' = - 3x + 5y - 6z } (5.24)
z' = -x + y
=� ) .
and A = ( -� �
-1 1 0
Then
,\-1 -1 2
det(>J - A) 3 ,\-5 6
1 -1 ,\
(,\-1)[,\(,\ - 5) + 6] + [3,\ - 6] + 2[-3 - (,\ -5)]
(>. - 1)(,\ 2 - 5,\ + 6) + (,\ - 2)
(,\ - 1)(,\ - 2)(,\ - 3) + (,\ - 2)
(>. -2)[(>. - 1)(,\ - 3) + 1]
(,\ -2)[>.2 - 4,\ + 4] = (,\ - 2)3 = 0
⇒ A1 = 2 is the only eigenvalue of A, of multiplicity 3.
For ,\ 1 = 2, (A 1 I - A)v = 0 with v = ( � ) =>
=> a - b + 2c = 0. c = 0 => a = b => v = v 1 = ( i), and a = 0 => b = 2c
=> v = v 2 = ( � ) are two independent eigenvectors which correspond to the
eigenvalue ,\1 = 2. Thus, two independent solutions of the system (5.24) are
Since A has only two independent eigenvectors which correspond to the eigenvalue ,\ 1
form
!)
where u 0 = av 1 + /Jv 2 is an eigenvector corresponding to ,\1 and u 1 is a solution of the
algebraic system (A-,\ 1 l)u 1 = uo, i.e., a generalized eigenvector. With u 1 = (

a= c =0 ⇒ b = 1 ⇒ u, =
(
n
which is consistent if and only if a = (3. Then -a+ b - 2c = a, and a
and uo =
(
�) Hence,
= 1 and
x 1 ( t) = e" ( i), x, ( t) = e" ( !) and x, ( t) = e" ( 3t � 1 ) ,
and
X (t) = (x 1 ( t) x, (t) X3 ( t)) = e21 ( i; 3t 7 I )
x(t) C1X1(t) + C2X2 (t) + C3X3(t) = X(t)c
= e2t
(� � 3t � 1 ) ( �: ) = e
2t
( c + 2� 1
: � t3t
3
+ l) ) ,
0 1 l C3 C2 + C3 l
1
i.e., x(l) = (c1 + c3t)e2 t. y(t) = [c1 + 2c2 + c 3 (3t + l)]e2t and z(t) = (c2 + c3t)e2t .
{ �: : -� +
This system may be expressed in matrix form as x'
y + J
= Ax, where
(5.25)
Then
,\-1 0
det(>.I - A) =
0
0 ,\ - 1 0 = (,\ - 1) 3 = 0
1 -1 ,\-1
⇒ >-1 = 1 is the only eigenvalue of A, of multiplicity 3.
For .\1 = I, (.\ 1 1 - A)v = 0 with v = ( � ) =>
=> a - b 0. a = I and c = 0 => v = v 1 = ( i ), and a O and c = I
=> v v, are two independent eigenvectors which correspond to the
n
= = (�)
eigenvalue ,\1 = 1. Thus, two independent solutions of the system (5.25) are
x1(t) = e' ( D and x,(t) = e' (
Since A has only two independent eigenvectors which correspond to the eigenvalue ,\ 1
form
where uo = av1 + f3v2 is an eigenvector corresponding to ,\1 and u 1 is a solution o� the
algebraic system (A- ,\if)u 1 = u0, i.e., a generalized eigenvector. With u 1 = ( � )

which is consistent if and only if a= 0. Then -a+ b = (3, and (3 = 1 and a= c = 0
o) n.
=> b = 1 => u, = 0 ) and Uo = ( � ) Hence,
x(, t) = e' [l ( D + J = e' (
and
0 0
X(t) = (x1(t) x2(t) x,(t)) = e' ( � 0 1 )
1 t
Let A be a 3 x 3 matrix with only one eigenvalue >. 1 of multiplicity m1 = 3, and

with only one independent eigenvector v 1 corresponding to >. 1 . Then one solution of
the system x' =Axis given by
A second, independent solution is sought in the form
where, as before, u0 is an eigenvector which corresponds to ,\ 1 and u 1 is a generalized

eigenvector which corresponds to >. 1, i.e., u 1 is a solution of the algebraic system
(A - >. 1 I)u1 = u0. A third, independent solution is sought in the form
where uo, u1 and u2 are vectors to be determined. Then
xI3(t) = A 1eA1l (2 1
t 2 uo + tu1 + u2 ) + e .), t( tu0 + u1 ) ,
1
and, hence, x;(t) = Ax3(t) if and only if
>-1e >.1t ( }t uo + tu1 + u ) + e

2
2
>. 1 t
(tuo (
+ u1) = Ae >.it }t 2u0 + tu1 + u2 ).
Division by e >.it =/- 0 yields the condition
>-1 (�t 2uo + tu1 + u ) + (tuo + u1) = A (�t 2uo + tu1 + u ),

2 2
which holds for all real t if and only if

>-1 uo = Auo, >-1u1 + uo = Au1 and >-1u2 + u1 = Au2. (5.26)
- -
The first two conditions in (5.26) are equivalent to
(>.1I A)uo = 0 and (A >.iI)u1 = uo,
respectively, as before. The third condition in (5.26) is equivalent to
(A - >-1I)u2 = u1,
a solution u2 of which is also called a generalized eigenvector corresponding to the
eigenvalue >.1 of multiplicity 3.

{ X
y'
1
= X + 2y -
= -x + 4 y -
z
z
} (5.27)
z' = y + z
x-(n A-(-i =D
and
i
-
Then
>. 1 -2 1
det(>.I - A) 1 >.-4 1
-
0 -1 >.-1
-
(>. 1)[(>.-4)(>.- 1) + 1] - [-2(>. - 1) +1]
(>. 1)(>.2 - 5>. + 5) + (2>. - 3)
-
>.3 - 6>.2 + 12>.- 8
(>. 2) 3 =0
⇒ >. 1 = 2 is the only eigenvalue of A, of multiplicity 3.
For ,\ 1 = 2, (,1 1/ - A)v = 0 with v = ( � ) ⇒
⇒ -b + c = 0 and a-2b + c = 0 ⇒ b = c and a = c, and c = 1 ⇒ v = v 1 = ( : ) 1s
the only independent eigenvector which corresponds to the eigenvalue >. 1 = 2. Thus,
one solution of the system (5.27) is
Since A has only one independent eigenvector which corresponds to the eigenvalue >. 1
form
x2(l) = e2\luo + u1),
!),
where u0 = v1 is an eigenvector corresponding to >.1 and u 1 is a solution of the
algebraic system ( A-,\ 1 I) u 1 = no, i.e., a generalized eigenvector. With u 1 = (
(A - >-d)u1 = uo ⇒
-1 2 -1 a 1
(-1 2 -1 ) ( b ) ( 1 )
i)
0 1 -1 C 1
⇒ -a + 2b - c = 1 and b - c = 1, and c = 0 ⇒ b = 1 and a = 1 ⇒ U1 = ( ⇒
A third, independent solution must be sought in the form
x3 (t) = e2 t ( tl2 uo + tu1 + u2 ) ,

where Uo = v1 = ( : ) and u 1 = ( i) are as above, and u 2 is a solution of the
algebraic system (A - A 1 I)u2 = u 1. With u2 = ( � ) , (A - A 1 I)u2 = u 1 cc>
n
=> b - c = 0 and -a+ 2b - c = 1 => b = c, and c = 1 => b = l and a = 0 =>
u, = ( cc>
t2 t
½2 +
t+l
x 2 (t)=e2t ( t+l ) and x3 (t)=e 2t ( ½t +t+l ) ,
t lt
2
2
+1
and

i.e.,
x(l) [c1+c2 (t +1)+c3 (�t2 +t)] e 2t

,
y(t) [c1+c2 (l+l)+c3 (�t2 +t+1)] e 2

t,
z(l) [c1+c2 t+c3 (�t +1)] e

2 2t
.
{ :: : =�: : �� = �; } .
z' = -x + y - 4z
(5.28)
(n 5
c
2
-3)
and A= -7 3 -8
x= -1 1 -4
Then
>- + 5 -2 3
det(>.I - A) 7 >- -3 8
1 -1 >-+4
(>-+5)[(>- -3)(>-+4)+ 8] +2[7(>-+4) - 8] +3[-7 - (>- -3)]
(>-+5)(>-2 +>- - 4)+1U +28
(>.3 +6>. 2 +>- - 20)+ lU +28
>.3 +6>-2 +12>. +8
3
(>-+2) =0
>. 1 = -2 is the only eigenvalue of A, of multiplicity 3.
⇒
o =n H n o ) o -! =D u ) o ) ⇒
c; b-3c = 0 and 3a-2b+3c = 0 c> b = 3c and a= c, and c = 1 c> v = v1 = ( � ) is
the only independent eigenvector which corresponds to the eigenvalue >-1 = -2. Thus,
one solution of the system (5.28) is
Since A has only one independent eigenvector which corresponds to the eigenvalue >. 1
form
where u0 = v 1 is an eigenvector corresponding to >.1 and u 1 is a solution of the
algebraic system ( A - A 1 I) u 1 = Uo, i.e., a generalized eigenvector. With u 1 = ( � }
(A - >.1I)u1 = uo ⇒
A third, independent solution must be sought in the form
where u0 = v 1 = ( � ) and u 1 = ( � ) are as above, and u2 is a solution of the
algebraic system (A- A,J)u2 = u 1. With u2 = ( � ) , (A - A,I)u, = u, =>

⇒ -b + 3c = 1 and -3a + 2b - 3c = 1, and a = 0 ⇒ b = 2 and c = 1 ⇒ u2 =

( 0�)
and
l+l
3t + 2
l
i.e.'
x(t) 2
[c1 + c (t + 1) + c3 (tt2 t)]
+ 2
e- i,
y(l) 2 2
[3c1 + c (3l + 2) + c3 (�l + 2l+ 2)] e- 2 t,
z(t) 2 2
[c1 + c l+ c3 (�l + 1)] e- 2 t.
Let A be a 4 x 4 matrix which has only one pair of complex-conjugate eigenvalues

>.1and "X1 of multiplicity 2, and with only one pair of independent eigenvectors v 1 and
v1, corresponding to >.1 and °X1 , respectively. Then two complex-conjugate solutions
of the system x' = Ax are given by
from which two real, independent solutions can be obtained as
as demonstrated earlier. In the present case, it is necessary to obtain two additional

real, independent solutions. Hence, a second pair of complex-conjugate solutions is
required, and they are obtained by means of complex generalized eigenvectors. Thus,
seek a complex solution in the form
where, as before, uo is an eigenvector and u 1 is a solution of the algebraic system

(A - >. 1 J)u 1 = u0, i.e., a generalized eigenvector. Then z 2 and z2 are two complex
solutions, and provide the two additional real, independent solutions
for a total of four real, independent solutions.
Example 5.19. Conside the system
r, }
= -y + z
y' = X + w
= -w (5.29)
z'
w' = z
Then, expanding the 4 x 4 determinant along the fourth row, we obtain

>. 1 -1 0
-1 >. 0 -1
det(>.I - A)
0 0 >. 1
0 0 -1 >.
>. 1 0 >. 1 -1
-1 >. -1 + >. -1 >. 0
0 0 1 0 0 >.
2 2
(>. + 1) + >.[>.(>. + 1)]
(>. + 1) 2 = 0
2
⇒ .\ 1 = i and .\ 1 = -i are the only eigenvalues of A, of multiplicity 2.
For )'1 = i, (!. 1 1 - A)v =0 with v = (;) =>
i 1 -1 0 i 1 -1 0 i 1 -1
-1 i 0 -1 0 0 -1 -i --) 0 0 -1
( ) --) ( ) (
00 i 1 00 i 1 0 0 0
00 -1 i 00 0 0 0 0 0
⇒ d =0 ⇒ c =0 and ia + b =0, and a= i ⇒ b = 1 ⇒
is an eigenvector which corresponds to the eigenvalue .\ 1. Then

are two complex-conjugate solutions and, hence,
- sin(t) cos(t)
cos t sin t
x 1 (t) = !R(z 1 (l)) = ( �( ) ) and x2 (t) = '.:r(z 1 (t)) = ( �( ) )
A generalized eigenvector corresponding to )11 is a solution u 1 of the algebraic system
(A� .\ 1 I)u 1 = tJo = v 1, i e , with u 1 = ( ; ) ,
Row reduction of the augmented matrix gives
I ;)
c� c� 2:)
-1 1 0 -1 1 0
I )
-1, 0 1 I 1 0 1
0 -1, -1 I o 0 -1, -1 0 -
0 1 I o 0 0 0 0
c�
-1,
-1 1 0
0 1 i I 2:
0 0 -2 I -2
0 0 0 I o
⇒ d = l, c +id= 2i and -ia - b + c = i ⇒ c = i and ia + b = 0, and b = 0 ⇒ a = 0

!)
=> u1 = ( is a generalized eigenvector, and
and z2 (t) are two additional complex-conjugate solutions. Hence,
are two additional real, independent solutions of the system 5

( .29),
- sin(l) cos(t) -tsin(t) tcos(t)

cost sin(t) t cos(t) tsin(t)
X(t) = (x 1 (t) x2 (t) x 3 (t) x.,(t)) = (
0 - sin(t) cos(t)
)
�( ) 0 cos( t) sin( t)
x(t) C1X1(t) + C2X2 (t) + C3X3(t) + C4X4(t)

X(t)c
-sin(t) cos(t) -tsin(t) icos(t) c1
cos(t) sin(t) tcos(t) tsin(t)
( ) ( :: )
0 0 -sin(t) cos(t) c
0 0 cos(t) sin(t) 4
-C1 sin(t) + C2 cos(t) -C3 t sin(t) + c4t cos(t)
c1 cos(t) + c2 sin(t) + c3t cos(t) + c4 t sin(t)
( )
-c3 sin(t) + c4 cos(t)
C3 cos(t) + C4 sin(t)
i.e.'
x(t) -c1 sin(t) + c2 cos(t) - c3t sin(t) + c4 t cos(t),

· y( t) c 1 cos(t) + c2 sin(t) + c3 t cos(t) + c4 t sin(t),
z(t) -c3 sin(t) + c4 cos(t),
w(t) c3 cos(t) + C4 sin(l)
Exercises 5 .1
1. Find the general solution of the system { x;
= -5x + 6Y } .
y=- 3x+ 4 y
2. Find the general solution of the system { :: : ��; y } .
3. Find the general solution of the system x' = Ax, where A= ( t � ).

4. Find a fundamental matrix for the system x' = Ax, where A = ( � .
� ) ,
5. Find a fundamental matrix for the system x' = Ax, where A= ( � .

� )
6. Solve the initial-value problem { x;

= x - Y } , (O) = 2, y(O) = 1.
y = 2x+ 4 y x
7. Find a fundamental matrix for the system x' = Ax, where A = ( _ �

-� ) .
8. Find the general solution of the system

{ x
' = 2y
y' = 2x
}.
t: }
=� ! ) .
9. Find the general solution of the system { �2x + 3y
z' = -x+y+ 2z
10. Find a fundamental matrix for the system x' = Ax, where A = ( -!
-1 -3 5
1
1 . Find a fundamental matrix for the system x' = Ax, where
A
=
(
=! ! � ) .
=� ) .
1 2. = 3
Find the general solution of the system { x; - x + Y } .
y=- 4x- 3y
1 3. Find the general solution of the system x' = Ax, where A= ( �

1
14. Find a fundamental matrix for the system x' = Ax, where A= ( i -� ) .
1 5. Solve the initial-value problem x' = Ax, x(O) = x0, where xo = ( !1 ) and
A= ( � -� ).
x' =x
1 6. Find the general solution of the system { y' = 2x + 2y - 2z } .
Z = X +y
1
furr:r
0 0
1 7. Find the general solution of the system x' = Ax, where A=
(
0 1
-1 2
1 8. Find a r trix for the system x' = Ax, where
( )
A= 0 1 1 0 .
-1 .o O 0
19. Find the general solution of the system

{ X
y'
1
=
=y
X + 3y }
170
=1 ; ) .
CHAPTER 5. LINEAR SYSTEMS
20. Find the general solution of the system x' = Ax, where A= (
21. Find a fundamental matrix for the system x' = Ax, where A= � � .
( )
x' = 3x -y }
22. Find the general solution of the system { y' = .1: + 2y - z
z' = 3y - z
23. Find the general solution of the system x' = Ax, where A= ( �
i �).
24. i daf matrix for the system x' = Ax, where
:: �rr�r
( )
r-r�1
-1 1 -4
25. i a matrix for the system x' = Ax, where

::
( )
2 -1 -4
26. Find the general solution of the system ( ;: )

z'
( =� � � ) ( ; ) .
0 0 -1 z
27. Find the general solution of the system x' = Ax, where A= (-� .
-1-0� 0
� )
28. Find the general solution of the system x' = Ax, where A = ( �-� � ).
1 1 1
-
29. i d th n of the system x' = Ax, where
:� �
r ·
( r-rr)
0 0 -1 1
5.2. NONHOMOGENEOUS SYSTEMS 171
5.2 Nonhomogeneous Systems

Theorem 5.6. The general solution of the nonhomogeneous system
x' = A(t)x + f(t) · (5.30)
is x(t) = xv(i) + x1i(t), where xv(t) is any particular solution of (5.30), and x1i(t) is
the general solution of the associated homogeneous system
x' = A(l)x, (5.31)
X1i(l) = C1X1(l) + C2X2(l) + · · · + CnXn (l),

where x1(l), x2(t), · · ·, xn (l) are n independent solutions of (5.31).
Proof. x(l) = xv(l) + x1i(t) is a solution of the system (5.30) because
x'(l) [xv(t) + x1i(t)]'

x�(t) + x;i (t)
A(t)xv(t) + f(i) + A(l)x1i(t)
A(l)[xv(l) + x1i (l)] + f(l)
A(l)x(t) + f(t).
Since x(t) contains n arbitrary constants, it is the general solution of (5.30).
Let X(t) be a fundamental matrix for the homogeneous system (5.31), and seek
a particular solution of the nonhomogeneous system (5.30) in the form
xv(t) = X(t)u(t), (5.32)
where u(t) is a vector function to be determined. Then
x�(l) = X'(t)u(l) + X(t)u'(l),
and, hence, (5.32) is a solution of (5.30) if and only if
X'(t)u(l) + X(t)u'(l) = A(t)X(t)u(t) + f(t). (5.33)
Since X'(t) = A(t)X(t) by Theorem 5.3 in Section 5.1.1, (5.33) is equivalent to
A(l)X(t)u(l) + X(t)u' (t) = A(t)X(l)u(l) + f(t),
i.e.'
X(t)u'(t) = f(t),
left multiplication of which by X(t)- 1 gives u'(t) = X(tt 1 f(t) and, hence,
u(t) = J X(t)- 1 f(t) dt. (5.34)
Thus, Equation (5.32), with u(t) defined by (5.34), gives
(5.35)
and determines a particular solution xp (t) for the nonhomogeneous system (5.30).
Analogous to single nonhomogeneous equations, this method is called variation of
parameters for systems.
Note that the integrand in (5.35) is an n-vector. The integral of a vector-valued

function v(t) is defined as the vector-valued function the components of which are
the integrals of the components of v(t), i.e.,
f V1 (t) dt
v(t) = => j v(t) dt =
f v2(t) dt
Vn (t)
The general solution of the nonhomogeneous system x' = A(t)x + f(t) is then
x(l) Xh(t) + X p(t)
X(t)c + X(t) J X(t)- 1 f(l) dt
X(l) [c + J X(lt 1 f(t) dl]
As in variation of parameters for single equations, the addition of arbitrary constants

of integration is not necessary, since they may be incorporated into the arbitrary
constant vector c.
Example 5.20. Consider the nonhomogeneous system
{ x'
y'
= -5x + 4y + 2t
= -4x + 5y + t } (5.36)
This system may be expressed in matrix form as x' =Ax+ f(t), where
( 2t ) .
x=
( ) X
y
, A=
( -5 _
4)
4 5
and f(t) =
t
For the associated homogeneous system x' = Ax,

det(-XI - A) = I A: 5 A
-=._\ I = (-X + 5)(-X - 5) + 16 = ,X 2
- g
= (-X - 3)(-X + 3) =0
⇒ -X1 = 3 and A2 = -3 are the eigenvalues of A.
⇒ 2a - b = 0, and a= l ⇒ v = v1 = ( � ) is an eigenvector which corresponds to

the eigenvalue -X1 = 3.
⇒ a - 2b = 0, and b = l ⇒ v = v2 = ( � ) is an eigenvector which corresponds to

the eigenvalue -X2 = -3.
Hence, two independent solutions of the homogeneous system are given by

1 e3t 2 2e-3t
x1 (t) = e3t ( ) = ( 3 ) and x2(t) = e-3t ( ) = ( ) ,
2 2e t 1 e-3t
and
is a fundamental matrix. Then

2e-3l
),
-e3t
2e-3t 2t O
X(t)- 1 f(t) ) ( l ) = ( 3t ) '
-e3t te
J X(t)- 1 f(l) dt dt = (
0
½te3t - ! e 3t
)
= g
e
1
3l (
3t - 1
)
0
::::} Xp(l) X(t) J e3t 2e-3t

X(t)-1f(t) dt = ( 3t
2e 3
e- t 9
1
) e3t (
O
3t _ 1
)
� 3t 2(3t - l)e-3t � ( 6t - 2 )
e ( 3 ) =
g (3t - 1 )e- t 9 3t - 1 '
and the general solution of the nonhomogeneous system (5.36) is
=� ) '
x(t) xh(t) + Xp (t) = X(t)c + Xp (t)
( ;::t
2e t
e _�:
) ( �: ) + t ( �!
i.e.'
x(t)
y(t)
Example 5.21. Consider the nonhomogeneous system
x' = x - 2y + e t cos(2t)
{ y' = 2x + y (5.37)
+ e t sin(2l) }
This system may be expressed in matrix form as x' =Ax+ f(l), where
x=
( )
y
X
, A=
( 2
1 -2
1
) and f(t) = e
t ( COS (2t)
sin(2t) ) .
det(.U - A) I ,,\-- 1 2 I = (,,\ - 1)2 + 4

2 ,,\-1
,,\
2
- 2,,\ + 5 = 0
-
⇒ ,,\, ,,\ =
2±FI6 = 1 ±2i are the eigenvalues of A.
2
For,,\= 1 +2i, (,,\1 J - A)v = 0 with v = ( � ) =>
=> ia + b = O,· and a= 1 ⇒ v = ( � ) is an eigenvector which corresponds to the

i
�igenvalue ,,\ = 1 +2i .
. Hence, a complex solution of the homogeneous system is given by
z(l) = e t [cos(2t) +i sin(2t)] [ ( �) +i ( �l ) ] ,

two real, independent solutions are given by
i [cos(2l) ( � ) - sin(2t) ( �l ) ] =e
t
( ��;g:j ) ,
( _ ;�:g�� ) ,
x 1 (l)
+ cos(2l) ( �l ) ] = e
t
x2 (t) e
l
[ sin(2i) ( � )
and
X(t) .
= (xi(l) x2 (i)) = e
l ( cos(2t) sin(2t)
sin(2t) -cos(2l) )
_ (- -t cos(2t) - sin(2l)
cos(2t) )
=
-l ( cos(2t) sin(2t)
sin(2t) - cos(2l) ) '
( (
e - sin(2t) e
-t cos(2t) sin(2l) l cos(2l) 1
X(l)- 1 f(l) e ) ) = ( )
J J ( �)
sin(2t) -cos(2l) e sin(2t) 0 '
X(l)- 1 f(l) di dl = ( �)
⇒ Xp(t) X(t) J X(tt 1 f(l) dt = e

l ( c�s(2t) sin(2t)
sm(2t) -cos(2l) )
t
( )
(
0
t l cos(2t)
e t sin(2t) ) '
and the general solution of the nonhomogeneous system (5.37) is
( (
x(i) x1i (i) + Xp (l) = X(t)c + Xp(t)
t cos(2t) sin(2t) c1 t cos(2t)
) ( c ) t
e sin(2t) -cos(2t) 2 + e sin(2t) ) '
1.e.'
x(l) l
[c1 cos(2t) + c2 sin(2t) + t cos(2t)],
et
y( l) e [c1 sin(2l) - c2 cos(2t) + t sin(2l)].
Example 5.22. Consider the nonhomogencous system
{
x' = -5x + 9y +
y' = -4x + 7y +
e
e
t
t
} (5.38)
=� � )
This sys tem may be expressed in matrix form as x' =Ax+ f(t), where
x = ( � ) , A= ( and f(t) = e
t ( �) .
det(,U -A)
I >, ;
5
;,
-_\ I = ( A + 5) (A - 7) + 36 = ,),2 - 2A + 1
(>--1)2 =0
==> >-1 = 1 is the only eigenvalue of A, of multiplicity m 1 = 2.
==> 2a - 3b = 0, and a= 3 ==> v = v1 = ( � ) is an eigenvector which corresponds

to the eigenvalue A 1 = 1.
Hence, one solution of the homogeneous system is given by
A second, independent solution takes the form
where u0 = v 1 is an eigenvector which corresponds to >- 1 , and u1 is a solution of the

algebraic system (A - >- 1 I)u1 = uo. With u 1 = ( � ) , (A - >-1I)u1 = uo ==>
==> 3b-2a = 1, and a = l ==> u 1 =( �) ==>
and
X(l) = (x1(t) x2(t)) = et ( 3 3i + l
2 2t + 1
)
X(l)- 1 f(t)
J X(t)- 1 f(l) dt
and the general solution of the nonhomogcneous system (5.38) is
�s:/ ) '
x(t) x1i(t) + xp(t) = X(t)c + xp (t)
=
i ( � �� : � ) ( �: ) + e
t
(
i.e.'
x(l) el [3c1 + c2 (3t + 1) + �t 2 + t] ,

y(t) e l [2c 1 + c2(2t + 1) + t2 + t] .
Definition 5. 7. An initial-value problem for a nonhomogeneous system consists of a

system x' = A(t)x+f(l), together with an initial condition x(t0 ) = x0, which requires
that the solution x(t) be equal to x0 at t = t0, and determines the arbitrary constant
vector c.
Theorem 5. 7. Let A ( l) and f ( l) be continuous on an interval J containing the point

t0 in its interior. The solution of the initial-value problem
x' = A(l)x + f(t), x(to) = xo,

is given by
x(t) X(t)X(t0)-1 x0 + X(l) 1l X(s)- 1 f(s) ds
1:
lo
X( l) [ X ( t 0 )- x0 +
1
X ( s )- 1 f ( s) ds]
where X(t) is a fundamental matrix for the associated homogeneous system.

Proof. The general solution of the nonhomogeneous system may be expressed in

terms of a definite integral as
1
x(t) = X(t)c + X(t) [ X(s)- 1 f(s) ds.
lto
Then x(to) = xo ==>- xo = X(lo)c ==>- c = X(t0 )-1x0.
Note that the definite integral of a vector-valued function is defined in a manner

analogous to the definition of the indefinite integral:
1: v1(s) ds
v(t) =
1: v2(s) ds
Vn (t)
{
x ' = -x + 2Y + et } x(O) = 2 y(O) = 1. (5.39)
y' = y + et ' '
This problem may be expressed in matrix form as x' =Ax+ f(t), x(O) = xo, where
I >,; 1 -_2 I = (>- + 1)(>- - 1) = 0

det(,U - A) =
>, 1
⇒ >- 1 = 1 and >- 2 = -1 are the eigenvalues of A.
⇒ a= b, and a= 1 ⇒ v = v 1 = ( � ) is an eigenvector which corresponds to the

eigenvalue >. 1 = 1.
⇒ b = 0, and a = 1 ⇒ v = v 2 = (
eigenvalue >-2 = -1.
t) is an eigenvector which corresponds to the
Hence, two independent solutions of the homogeneous system are given by
and

Q -e -t Q -l
-( ) = ( el � et ) '
- el el
( �t ��l ) ( :: )
t
( t)'
1
t
( t) i: ds = ( ; ) =
( � ),
( � -i ) ( � ) i ) ' =
(
(i)+(i)=( : )
t l
t
⇒ x(t) X(l) [x(o)- 1 x(O) + 1 X(s)- 1 f(s)ds]
: t 1
( : ; ) ( ; )
t
e
(t + l)et + e -t
( (t + l)et )
i.e., x(t) = (l + l)et + e-t, y(t) = (l + l)et is the solution of the initial-value problem
(5.39).

x' = -2x + 4y + 4e-2 t
{ y' = -x - 2y + 2e- 2t } '
x(O) =2 ' y(O) = -1. (5.40)
This problem may be expressed in matrix form as x' =A x+ f(t), x( O) = x 0, where
x= ( ) y
X , A= ( -2
_ 1 _ 42), f ( t) =
e-2t
2
(42) and xo =(_ 1) .
- 4±FT6
⇒ ,\, ->,. = ----
2
= -2± 2i are the eigenvalues of A.
For,\= -2 + 2i, (>-.J -A)v = 0 with v = � ⇒

( )
⇒ a+ 2ib = O, and b = i ⇒ v = : is an eigenvector which corresponds to the

( )
eigenvalue,\= -2 + 2i.
Hence, a complex solution of the homogeneous system is given by
z(l) = e-2t[cos(2t) + i sin(2t)] [(� ) + i

( � ) ] ,
O 2cos(2 ) l
x 1 (t) e
_2 t[cos(2·t)(02).
-sm(2t)
(1 )]
=e _2t
(_ sin(2l))
,
x 2 (t) e- 2 t [sin(2t) � + cos(2t)(� )] = e

() -
2
t ( ��:g�� )
2
arc two real, independent solutions, and
t( 2cos(2t) 2sin(2t)
X(l) = (x1(t) x2(t)) = e _2
-sin(2l) cos(2t))
�e2t ( cos(2l) -2sin(2t)
2 sin(2l) 2cos(2t)) '
= �e2 t
2
(
cos(2t) -2sin(2t)
sin(2t) 2 cos(2t))
e-2t (
4
2)
2 cos(2t) - 2sin(2t)
( 2sin(2t) + 2cos(2t))
lt( 2cos(2s)-2sin(2s)
d8
0 2 sin(2s) + 2cos(2s))
sin(2s) + cos(2s) t
( -cos(2s) + sin(2s))
I0
sin(2t) + cos(2t) - 1
( -cos(2t) + sin(2t) + 1 )
1 ( � � ) ( !1 ) = ( !1 ) '
X(o)-1x(O) + 1t X(st1f(s) ds
sin(2t) + cos(2t) - 1
1 ) + ( -cos(2t) + sin(2t) + 1 )
( -1
sin(2t) + cos(2t)
= ( - cos(2t) + sin(2t))
⇒ x(t) X(t) [ X(o)- 1 x(O) + 1t X(s)-1f(s) ds]
2cos(2t) 2sin(2t) sin(2t) + cos(2t)

e _2t
-cos(2t) + sin(2t))
(
- sin(2t) cos(2l)) (
e-2t
( !1 ) ,
i.e., x(t) = 2e-2t, y(t) = -e- 2 t is the solution of the initial-value problem (5.40).
{ :: : x _ �� : � } , x(O) = 3, y(O) = 2. (5.4 )

1
=� ) '
This problem may be expressed in matrix form as x' =Ax+ f(t), x(O) = x0, where
X
= ( :) ' A = ( � f(t) = ( � ) and Xo
= ( �) .
det(,\J - A)
I :1 ,\ !2 /
= ,\(,\ + 2) + 1 = ,\2 + 2,\ + 1
(,\ + 1) = 0
2
=> A.1 = -1 is the only eigenvalue of A, of multiplicity m1 = 2.
For A.1 = -1, (,\if - A)v = 0 with v = ( � ) =>
=> a= b, and a= l => v = v 1 = ( � ) is an eigenvector which corresponds to the

eigenvalue ,\ 1 = -1.
Hence, one solution of the homogeneous system is given by
A second, independent solution takes the form
where u0 = v 1 is an eigenvector which corresponds to ,\ 1 , and u 1 is a solution of the

algebraic system (A - ,\ 1 J)u 1 = uo. With u 1 = ( � ) , (A - A1I)u1 = uo =>
=> a - b = l, and b = 0 => u 1 = ( � ) =>
and
X(l)- 1 = -el ( �
1
- -l
\ )
= et ( � ),l t
�/
l l
( i ( ; �: ) ,
X(l)- 1 f(l) =
i � �/ ) ( � )
(s2 -s)e8 ds (s2 -s)e8 [- (2s-l)e8 ds

l l
1 1
= (t -t)e l -(2s-l)e8
2
i: +2 o e8 ds
f
t
= (l2 - l)e l- (2t-l)el -1 + 2e8 i:

= (t2 -
t)et (2t-l)el -1 +2et -2
-
-
(l2 - 3l 3)el 3,
+
1
l
(2 - s)e8 ds (2 - s)e8 i: + fo\s ds
(2-t)el 2 - + - el 1
(3-t)e l -3,
1
l
X(s)- 1 f(s) ds
1
l
e8 ( � �: ) ds
(
(l 2 +
--
3l 3)el 3
(3 - l)et 3 .
-
)
X (o)- 1 x(O) = =
( � -� ) ( � ) ( � ) ,
2
( 1 )
+ (
(t2 3t + 3)e l 3
(3-t)e l 3
-
) - -
( ( t - 3l + 3)el
-- 1
2
(3 - l)e
t 2 )
x(l) = X(t) [x(o)- 1 x(O) +1 X(s)- 1 f(s) ds]

t
⇒
= e -l ( 1 t+
t
1
) (
(l2 3t + 3)et -1
(3 - t)e l 2
-
) -
- -
1
( 6 - l 3e- t 2te- t
3 - e l - 2lc-t
' )
i.e, x(t) = 6- l 3e- l
problem (5.41).
- - -
2le- t , y(t) = 3- e- t 2te- l is the solution of the initial-value
{
X =1
X
y' = 3x - z + tcos(t) , x(O) = 1, y(O) = 2, z(O) = 1. (5.42)
z' = -x + y + tsin(t) }
This problem may be expressed in matrix form as x' =Ax+ f(l), x(O) = x0, where
x= : , A=
( )
z
(
� � -� , f(t) =
-l 1 0
) (
tco�(l)
tsin(l)
)
and Xo =
(
�l) ·
>. - 1 0 0
det(>.J - A) -3 >. 1
1 -1 >.
⇒ >.1 = 1 and >.2 , "X2 = ±i are the eigenvalues of A.
For ,\ 1 = 1, (,\ 1/ - A)v = 0 with v =

( � ) cc>
cc; b = 2c and a = c, and c = 1 cc> v = v 1 = cc>

(� )
0)
n
cc>
Jl 0) 0)
(i-1
-3
1
⇒ a = 0 and b = ic, and c = 1 ⇒ v = v2 = (: ) ⇒
z(l) = [cos(t)
+ isin(t)] [ ( � )
+
i(
! )]
is a complex solution of the homogeneous system,
x2 (t) cos(t) ( � ) - sin(t) ( � ) = (- s�n(t) ) ,

1 0 cos(t)
�
x3 (l) = sin(t) ( � ) + cos(l) ( � ) = ( c� (l) )
1 0 sm(t)
are two real, independent solutions, and
is a fundamental matrix. Then (see Appendix A.3 for inversion of matrices)
0 �
1
X(lt = ( 2sin(t): cos(t) - sin(t) co (l) ) ,
- sin(l) - 2cos( t) cos(t) sin(t)
0
X(l)- 1 f(l) = ( 2sin(t): cos (t ) - sin(t) co�(l) ) ( t co�( t) ) = ( � ) ,
- sin(t) - 2cos(t) cos(t) sin(t) l sin(t) t
186 CHAPTER 5. LINEARSYSTEMS
⇒ x(t)
1 1
1.e., x(t) = et , y(t) = 2et + t2 cos(t), z(t) =et + t2 sin(t) is the solution of the
2 2
initial-value problem (5.42).
Exercises 5. 2
' = 2x + 3e2t
1. Find the general solution of the system { x,
}.
y =x- y+et
x'=4y+4cos(2 t) }
2. Find the general solution of the system { , . .
y = -X - 2 Slll(2t )
x' = -x + 4y + t2 et
3. Find the general solution of the system { ,
y = -x + 3y + tet }·
4. Find the general solution of the system { x; = 2x - 2Y }.
y =x+4y+ 2
x =x+e
5. Solve the initial-value problem { ; t
y = 2y+ 2e2
t } , x(0) = 3, y(0) = 2.
x; = -y 1
6. Solve the initial-value problem { + } , x(0) = 2, y(0) = 1.
Y = x+ 1
}
n' n'
x' = -2x - y + e- 4 t
7. Solve the initial-value problem { , _ t x(0) = 0, y(0) = 1.
y = 4x- 6y+ 2e 4
,
X
= ( A= (
-
� -[ f(l) = e' UJ
8. Solve the initial-value problem x' =Ax+ f(l), x(0) =xo, where
and Xo = ( D
Chapter 5 Exercises
1. A differential equation of order n 2:: 2 for the unknown function y(i) can be
transformed into a system of n first-order equations for the unknown functions
X1(t), X2(t), · · ·, Xn (l) by letting X1 = y, X2 = y', X3 = y", ·· ·, Xn = y( n -l)_
Consider the equation y" - 2 y'+y =0.
(a) Find two independent solutions y 1 (l) and y2 (t) of the equation, and the
general solution y(l).
(b) Transform the equation into a system of two first-order equations.
(c) Find two independent solutions x 1 (l) and x2 (l) of the system in part (b),
and the general solution x 1 (l), x 2 (l). Confirm that x 1 (t) is equivalent to
y( l).
(d) Compute the Wronskians W[y1 (l) y2 (t)] and W[x 1 (t) x2 (l)] and compare
the results.
2. Let X (t) be a fundamental matrix for the system x' = Ax.
(a) Find a fundamental matrix Y(l) such that Y(O) = I, the identity matrix,
and express the solution of the initial-value problem x' = Ax, x(O) = x0,
in terms of Y (t).
(b) Find a fundamental matrix Z (l) such that Z ( t0 ) = I, where t0 is a real
number, and express the solution of the initial-value problem x' = Ax,
x(t 0) = xo, in terms of Z(l).
3. Consider the system

{ x'
y
,
= 2x
= x- 2 y
}
.
(a) Find the general solution of the system.

(b) Solve the first equation in the system for x(t), substitute into the second
equation for x, and solve the second equation for y(l). Confirm that your
answer is equivalent to the one in part (a).
(c) Differentiate the second equation in the system and eliminate x and x' in
order to obtain a second-order equation for y(t).
(d) Solve the equation in part (c) for y(t), determine x(l), and confirm that
your solution is equivalent to the one in part (a).
=:
4. An n x n matrix A is diagonalizable, i.e., there exists a nonsingular matrix
P such that p- 1 AP = D is a diagonal matrix, if and only if A admits n
independent eigenvectors. Then D has the eigenvalues of A along its diagonal,
and the columns of Pare the eigenvectors of A. Consider the system x' = Ax,
1
where A = ( � ).

(b) Determine a new unknown y in terms of x such that the system x' = Ax
for x transforms into an equivalent system y' = Dy for y.
(c) Solve the system in part (b) for y(l).
(d) Determine x(t) from y(t) in part (c) and confirm that your answer 1s
equivalent to the one in part (a).
0 0
5. Find a fundamental matrix for the system x' � Ax, where A � ( �
1 0
0 1
0 0
by generalizing the method employed in Examples 5.15 and 5.16.
6. Find a fundamental matrix for the system x' = Ax, where A = (

0 0 0 1
H
by generalizing the method employed in Examples 5.17 and 5.18 and taking
n ),
In general, the subscripts of the vectors from left to right, are n = 0, 1, 2,
tn
Un,
3, · · •, and their coefficients, from right to left, are , n = 0, 1, 2, 3, · · ·. The

1
n.
latter sequence of functions has a special significance, and will be encountered
in Part II of this book.
7. Let >. = a + i/3 be a complex eigenvalue of an n x n matrix A, with the

corresponding complex eigenvector v = a + ib. Show that the corresponding
two real solutions
are linearly independent on R

Part II
Infinite Series
189
Chapter 6
Sequences and Series
Infinite sequences and series arise in such numerous applications that neither science
nor engineering would be possible in its present form without them. Sequences and
series are defined in the present chapter. More specialized forms of series and some
of their applications will be discussed in subsequent chapters.
All of the linear differential equations solved in Part I of this book are equations
with constant coefficients, with the exception of Cauchy-Euler equations, which are
of such a special form that they can be transformed into equations with constant
coefficients. In general, an equation with variable coefficients such as
y" + xy' + y = 0 (6.1)
can be solved only by means of infinite series. In the final chapter of this book, we
shall have gathered sufficient knowledge to solve Equation (6.1).
6.1 Sequences
Definition 6.1. Let f be a real-valued function and k an integer. The ordered set
of values off at the integers k, k + l, k + 2, · · ·, i.e.,
{J(k), f(k + 1), J(k + 2), · · ·},

is called a sequence of real numbers. The integer k is the initial value of n. It is
customary to let J(n) = an (or bn or en ), called the general term of the sequence, and
to express the sequence' as
or, more concisely, as {an } �=k .
191
192 CHAPTER 6. SEQUENCES AND SERIES
Example 6.1. The ordered set of real numbers
{1 ! ! ! ... }
' 2' 3' 4'
00
is a sequence with a n = � and k = 1. It may be expressed as { � }
n n n=l
1 1
is a sequence with an = n and k = 0. It may be expressed as { n } .
2 2 n=O
{ -4, -2, 0, 2, 4, 6, · · ·}
is a sequence with an = 2n and k = -2. It may be expressed as { 2n}�=- 2.
{2, 3, 4, 5, · · ·}
is a sequence with an = n and k = 2, or an = n + 1 and k = 1, or an = n + 2 and

k = 0. Thus, the representation of a sequence as {an}�=k is not unique. Hence,
but { n}�=2 is the simplest.
{1 ' -1 ' 1 ' -1 ' 1 ' -1 ' · · ·}
is a sequence with an = (-lt and k = 0, or a n= (-lt and k = 2, or an = (-1r- 1

and k = 1.
n
Example 6.6. The sequence {an}�=o, where an = v'n+I' is
n+l
6.1. SEQUENCES 193
Example 6. 7. The Fibonacci sequence is defined by
The first few terms are

{1, 1, 2, 3, 5, 8, 13, 21, 34, · · · }.
Such a sequence, where the terms arc not given explicitly in terms of n, but in terms
of preceding terms, is said to be defined inductively.
Given a sequence { an }�=k' it is frequently the case that the initial value k of n is
irrelevant. In that case, the sequence is expressed simply as {an }. One such instance
is when it is necessary to determine whether or not the terms an approach a unique,
finite number as n is increased indefinitely.
Definition 6.2. Given a sequence { an },

lim an= L
n-+oo
means that, the absolute value of the difference (i.e., the distance) between an and
L can be made arbitrarily close to O by taking n sufficiently large. More rigorously,
for any E > 0, however small, there exists an integer N, which depends upon E, such
that
n 2:'. N => Ian - LI < E.
If lim an= L, the sequence { an } is said to converge to L, and Lis called the limit of
n-+oo
the sequence. The notation an ---+ L as n ---+ oo is also employed. If no such L exists,
then the sequence is said to diverge, and lim an does not exist.
n-+oo
Example 6.8. It is clear that the number I_ can be made arbitrarily close to O by
n
taking n sufficiently large. For example, in order to make I_ < 10- 1 0 , take n > 10 10 .
n
In order to make I_ < 10- , take n > 10 , etc. Hence,
100 100
n
. 1
hm - = 0.
n-+oo n
Formally, for any E > 0, I I_ - 0 I = � < E if and only if n > �. Hence, it suffices to
n n E
let N be any integer such that N > !. Then
1 1
n > - <E =?
E n
which proves that the sequence { ¾} converges to 0.
Notice the manner in which this proof would fail if we made the false assumption
that lim a n= L where L-/=- 0. For example, if we take L = 17, then
n�oo
/¾- 11/ � 16
for all n -/=- 0, and, hence, I¾- 171 < c is impossible for any c � 16.
Example 6.9. Let c be a constant. In the present context, a constant is any number
independent of the "variable" n. The sequence
{ C} = { C, C, C, · · · }
converges to c, i.e., lim c = c since, for any c > 0, le - cl = 0 < c for all n � 1.
n->oo
Example 6.10. As in Example 6.8, the sequence { �2 } converges to O since �2 can

be made arbitrarily close to O by taking n sufficiently large. More generally,
hm -=0
. 1
n->oo nP
for any real number p > 0 smce, for any c > 0, let N > l/p. Then
. 1
c
n� N
If p =0, then lim � = lim 1 = 1. If p < 0, then lim � does not exist because
n-+oo nP n-+oo n-+oo nP
1
-P = P
n- , with -p > 0, increases indefinitely as n increases indefinitely. Thus, the
n
sequence { :P} converges for p � 0 and diverges for p < 0.
Example 6.11. Consider the sequence { ( -1 )"} of Example 6.5.
lim (-lt
n-+oo
doeti not exist because a n = ( -1 )" takes both of the values 1 and -1 regardless of
how large n is and, hence, the terms an do not approach a unique number. Thus, the
sequence { (-1 )"} diverges.
6.1. SEQUENCES 195
Example 6.12. lim sin(n) does not exist because sin(n) oscillates and does not
n......,oo
remain close to any one number regardless of how large n is. Hence, the sequence
{sin(n)} diverges.
1
Example 6.13. The number can be made arbitrarily close to O by taking n
2n
sufficiently large. Hence,
. 1
hm - = 0,
n---+oo 2 n
1
i.e., the sequence { n} converges to 0.
2
Example 6.14. Let r be a real number, and consider the sequence {rn }�=O· We
shall determine those values of r for which the sequence converges, and find the limit.
If r > 1, then rn increases indefinitely as n increases indefinitely; hence, the sequence

{rn } diverges.
If r = 1, then rn = 1 for all n 2:: 0; hence, the sequence {rn} converges to 1.
If lrl < 1, i.e., -1 < r < 1, then rn can be made arbitrarily close to 0 by taking n
sufficiently large; hence, the sequence {rn } converges to 0.
If r � -1, then rn alternates between positive and negative values, with rn 2:: 1 if n
is even and rn � -1 if n is odd; hence, the sequence { rn} diverges.
Theorem 6.1. Let {a n} and {bn} be convergent sequences, with lim an = L and
n......,oo
lim bn = M. Then:
n-,oo
1. The sequence {an + b,i} converges, and lim (an + bn )
n......,
= L + M.
2 . The sequence {can} converges for any constant c, and lim can = cL.
n......,oo
3. The sequence { an bn } converges, and lim an bn = LM.
n......,oo
an . an L
4. The sequence {- } converges if M =I- 0, and lim -
n......,oo bn
= -.
bn M
Example 6.15. By Example 6.10, lim � = 0 and lim --; = 0. Hence, by item 2
,i-,oo n ,i-,oo n
of Theorem 6.1 with c = -1, lim (--;) = 0 and, by item 1, lim (� - -;) = 0.
n......,oo n n......,oo n n
Example 6.16. Since the sequence { ¾} converges to 0 by Example 6.8 and the
¾}
sequence {1} converges to 1 by Example 6.9 with c = 1, it follows by Theorem 6.1
that the sequence { 1 - converges to 1.
Example 6.17. If either {an } or {bn } diverges, then Theorem 6.1 is not applicable,
and sums, differences, products and quotients of such sequences may converge or may
diverge, depending upon the particular sequences.
Both of the sequences {n2 } and { n3 } diverge because their terms increase indefinitely
as n increases indefinitely, and { n 3 - n2} diverges. The limit is not 0.
Both of the equences { vn+T} and { fa} diverge, but
� � vln+l+fa 1
vn-t-1-fa= ( v n+l-fa) =
vn+1
n+l+fan vnTI
n+l+ fa
n
converges to 0 since the denominator can be made arbitrarily large and, hence, the
fraction arbitrarily close to 0, by taking n sufficiently large.
The sequence { ¾} converges to 0 and the sequence { n2} diverges. The product
{ ¾n2} = {n} diverges.
The sequence { :3 } converges to 0 and the sequence { n2 } diverges. The product
{ 3
2
: n } ={ ¾} converges to 0.
n2 + 1n3 -
Example 6.18. Consider the sequence { }· In order to determine the
2n3 + n -l
limit of such an expression, divide the numerator and the denominator by nr , where
r is the greatest exponent of n in the denominator. In the present case, r = 3, and
n3 n2 + 1
-
2n3 + n -l
1 1 1
Since lim - = 0, lim 2 = 0, lim 3 = 0 and lim c = c for any constant c,
n->oo n n->oo n n->oo n n->oo
lim (1 - � + �) = 1
-= n �
and
�=
lim (2 + �� - �� ) = 2
n3 - n2 + 1 1
-
by Theorem 6.1. Hence, lim
n->oo 2 n3 + n - l 2
6.1. SEQUENCES 197
2n + 1
Example 6.19. Consider the sequence { J }· Divide the numerator
3n2 + 3n + 5
and the denominator by n = Jni to obtain
2n + 1 2+1n
J3n2 + 3n + 5
2
which converges to J3 as n --+ oo.
Definition 6.3. Given a sequence {an },

lim an = oo
n->oo
means that an can be made arbitrarily large by taking n sufficiently large. More
rigorously, for any K > 0, however large, there exists an integer N, which depends
upon K, such that
Similarly,
lim an = -oo
n->oo
means that lan l can be made arbitrarily large with a n < 0 by taking n sufficiently
large. More rigorously, for any K > 0, however large, there exists an integer N, which
depends upon K, such that
n�N ==} a n < - K.
In either case, the sequence { a n} diverges.
Example 6.20. The terms of the sequence { n3 } can be made arbitrarily large by
taking n sufficiently large. Formally, for any K > 0, let N > K 1 13. Then
n� N ⇒ n > K 113 ⇒ n3 > K

==} lim n3 = oo and the sequence { n3 } diverges. Similarly, lim (-n3 ) = -oo and the
n---+oo n---+oo
sequence { -n3 } diverges.
Definition 6.3 is useful in distinguishing sequences { an } with lim a n = ±oo from

n->oo
other divergent sequences such as {sin(n)} and {(-lt}.
In general, it is difficult to determine the convergence or divergence of a sequence
directly from first principles, especially for more complicated sequences, and several
theorems are available in order to facilitate this task.
Theorem 6.2. For any sequence {an },
lim a n
n-+oo
=0 if and only if lim lan l
n-+oo
= 0.
Proof. an is close to 0 if and only if lan l is close to 0.
Example 6.21. Since

lim I (-ir1 = n->oo
lim � = 0,
n->oo n n
it follows, by Theorem 6.2, that lim --(-lt = 0.

n->oo n
Theorem 6.3. Let J be a real-valued function and let an = J(n). Then
limf(x) = L
x-+oo
⇒ liman= L.
n--+oo
In other words, if lim J(x) = L, then it is also true that lim an= L.
x-+oo n-+oo
The converse of Theorem 6.3 is false, i.e., if lim a n = L, then it may or may not
n->oo
be true that lim J(x) = L. For example, let J(x) = sin(1rx). Then an= sin(1rn) = 0
X->00
for all n and, hence, lim an = 0, but lim J(x) does not exist.
n-+oo x-+oo
ln n) ·
Example 6.22. Consider the sequence { � } In order to determine the limit of
ln(x)
the sequence, let J(x) = , and employ L'Hopital's rule to determine lim f(x).
X X->00
ln(x) l. ln(n)
lim -- = X->00
lim =0 ⇒
.:£ lim = 0'
X->00 X 1 n->oo n
by Theorem 6.3.
Example 6.23. Consider the sequence { ( 1 + ¾) n}. In order to determine the
limit of the sequence, let J(x) = ( 1 + �) x' and employ the fact that
f(x) = eln[J( x)] = ln

e [(1+¾rJ = e xln (l+¾)_
Since the exponential function is continuous,

l im [x ln(l+ )] '
1.1m ex In(l+l) = e x -oo
.l
X
X
X->00
6.1. SEQUENCES 199
and
ln (1 + l)
lim x ln 1+ -X1) = lim x ,
X->00 ( X->00 -1
which is of the form i and, hence, L'H6pital's rule can be applied. Thus,
ln(l + l) 1�1
( - ;2 ) 1
lim x = lim = lim --1 = 1
X->00 lX X->00 x- -
1
�
X->00 1+ -
X
⇒ ⇒
n
lim J(x) = e = e
X->00
1
lim
n->OO
(1 + ]:_)
n
= e,
by Theorem 6.3.
Theorem 6.4. Let {an }, {bn } and {en } be sequences, and suppose that
an :s; bn :s; Cn for all n � k,
where k is an integer. Then
lim an = lim Cn = L =} lim bn = L.

n->oo n->oo n->oo
This theorem is called the squeeze theorem because the terms of the sequence {bn }
are "squeezed" between the corresponding terms of the sequences {an } and {bn }-
sm n )
00
.
Example 6.24. Consider the sequence { . ( } Since -1 :s; sin(n) :s; 1, we
-n- n=l •
obtain the inequalities
1 sin(n) 1
--<--
- <-.
-
n n n
sin(n)
By Theorem 6.4, lim (-]:_) = 0 and lim (]:_) = O ⇒ lim = O.
n->oo n n->oo n n->oo n
Alternatively, since O :s; I sin(n)I :s; 1, we obtain the inequalities
I sin(n) = 0 and, hence,

I
By Theorem 6.4, lim O = 0 and lim (�) = 0 ⇒ lim
n->oo n->oo n n->oo n
sin(n)
lim -- = 0 by Theorem 6.2.
n-oo n
2n
Example 6.25. Consider the sequence {an }�=1, where an = . Then
1
n.
2-2·2···2· 2 2 2 2 2 2
an = -------- = - . - . -···--. -·
1-2·3···n ( -l)·n 1 2 3 n-1 n
4 4
If n = l, then a1 = 2:::; -. If n = 2,then a2 = 2:::; -. If n :2:: 3,then
n n
2 -···--
-· 2 2 <1 2 -·
-· 2 -···--
2 2 <2
⇒
2 3 n- l - 1 23 n-1 -
Hence,
4
0:::; an :::; - for all n :2:: 1.
n
4
Since lim O = 0and lim - = 0,it follows by Theorem 6.4 that lim an = 0.
-->oo
n n-->oo n n-->oo
nl
Example 6.26. Consider the sequence {an }�=l,where an =
�- Then
n
1 ·2·3 ···(n-1) n
· l 2 3 n- l n
an = -------- = - . - . -•••--. -,
n·n·n···n·n n n n n n
a1 = 1 and, for n :2:: 2,
2 -···--·
3 n 1n -
-· -<1 ⇒
n n n n-
Hence,
1
0:::; a n :::; - for all n :2:: 1.
n
1
Since lim 0= 0and lim - = 0,it follows by Theorem 6.4 that lim an = 0.
n-->oo n-->oo n n-->oo
Definition 6.4. A sequence { a n } is n

i creasing for n :2:: N, where N is an integer, if
an+l :2:: an for all n :2:: N.
It is decreasing for n :2:: N if
an+l:::; an for all n :2:: N.
A sequence which is either increasing or decreasing is called monotone.
Example 6.27. The sequence {an } = { 1} is decreasing for n 2: 1 because
1 1
an+ l = -- < - = an for all n 2: 1.
n+ 1 n
6.1. SEQUENCES 201
Example 6.28. The sequence { an } = { 1 - �} is increasing for n 2: 0 because

2
1 1
->--
n+l
=>
2n 2
i.e., an < an +l for all n 2: 0.
n2
Example 6.29. Consider the sequence { an}�= 1, where an = ne-s. In order to
2
x
determine whether or not this sequence is monotone, let f(x) = xe- 8. Then
J'(x)
2
= e_xs +
2
xe_xs (-�) = e_x: ( 1 - :
2
) < 0 for x > 2.
It follows that the function f is decreasing for x > 2 and, hence, the sequence {an }
is decreasing for n 2: 3 > 2.
Example 6.30. The sequences {(-l)n} and {sin(n)} are not monotone because
none of the inequalities in Definition 6.4 can hold for all n 2: N for any integer N.
Definition 6.5. A sequence { an }�=k is bounded above if there exists a real number
K such that an :::; K for all n 2: k. It is bounded below if there exists a real number
M such that an 2: M for all n 2: k. It is bounded if it is bounded above and bounded
below, i.e., there exist real numbers K and M such that M :::; an :::; K for all n 2: k.
A sequence which is not bounded is called unbounded. K is called an upper bound,
and M a lower bound of the sequence.
00
n
Example 6.31. The sequence {- -} is bounded because
n + l n=O
n
0- > 0.
< -- < 1 for all n -
n+ 1 -
Here, M = 0 and K = l. These numbers are not unique since any M :::; 0 and K 2: 1
are also lower and upper bounds of the sequence, respectively.
Example 6.32. The sequence {2n }�=l is bounded below because
2n 2: 2 for all n 2: 1.
It is not bounded above because lim 2n = oo, i.e., 2n can be made arbitrarily large
n--+oo
by taking n sufficiently large and, hence, it is impossible to have 2n ::; K for all n 2: 1,
for any real number K. Thus, {2n }�=l is unbounded.
Example 6.33. The sequence {1- n}�=l is bounded above because

n 2: 1 ⇒ -n S -1 ⇒ an = 1 - n S 0.
It is not bounded below because lim (1 - n) = -oo. Thus, it is unbounded.
n --+oo
Theorem 6.5. A sequence { an }�=k is bounded if and only if the sequence {/an /}�=k
is bounded above.
Proof. If { an }�=k is bounded, then there exist real numbers M and K such that
MSa n SK for all n 2: k. Let L = max{/M/, /K/}. Then
M Sa n SK =? -L S an S L =? /an / S L,
i.e., {/an /}�=k is bounded above. Conversely, if {/an /}�=k is bounded above, then
there exists a real number K such that /an /S K for all n 2: k, and
/an /SK =? -KS an SK,
i.e., {a n}�=k is bounded.
Example 6.34. The sequence {sin(n)}�=O is bounded because -1 S sin(n) S 1,

which is equivalent to I sin(n)/S 1, i.e., the sequence {I sin(n)/}�=O is bounded above.
Theorem 6.6. If a sequence {an }�=k is increasing for n 2: N, where N is an integer,

and bounded above, then it converges. If it is decreasing for n 2: N and bounded
below, then it converges. If it is monotone for n 2: N and bounded, then it converges.
00
1 1 l
Example 6.35. The sequence { - } is decreasing for n 2: 1 since -- < - for
n n= n+1 n
l
1
all n 2: 1, and it is bounded below since - > 0 for all n 2: 1. Hence, by Theorem 6.6,
n
it converges, a result which has already been established in Example 6.8.
Example 6.36. The sequence { 1 - e-n } :'=o is increasing for n 2: 0 because
e n+l > en =? e-n > e-(n+l) =? a n = 1 - e- n < 1- e-(n+l) = an +l,

and it is bounded above since
e-n > 0 ⇒ -e-n < 0 ⇒ a n = 1 - e- n < 1 for all n 2: 0.

Hence, by Theorem 6.6, it converges. Note that the sequence { e-n } :'=o converges to
0 by Example 6.14 with r = e- 1. Hence, { 1 - e-n } :'=o converges to 1 by Theorem 6.1.
6.1. SEQUENCES 203
Theorem 6. 7. Let P(n) denote a statement or a formula which involves an integer

n. If P(l) is true and the implication P(n) =} P(n + 1) is true for all n 2 1, then
P(n) is true for all n 2 1. More generally, if P(k) is true, where k is an integer, and
P(n) =} P(n + 1) is true for all n 2 k, then P(n) is true for all n 2 k. This is called
the principle of induction.
In order to prove that the implication P(n) =} P(n + 1) is true, assume P(n) is
true and employ P(n) to show that P(n + 1) is also true. The assumption that P(n)
is true is called the inductive hypothesis.
n
Definition 6.6. Let a 1, a2, · · · , an be real numbers. The symbol Lai, read "The
i=l
sum of ai as i ranges from 1 to n," is defined by
n
Lai = a1 + a2 + a3 + · · · + an-I + an,
i=l
and is called sigma notation.
Example 6.37. Consider the sum

n
Li= 1 + 2 + 3 + · · • + (n -1) + n.
i=I
We wish to prove the formula
�i= n(n+l)
� 2
(6.2)
i=l
for all n 2 1. Let P(n) denote the formula (6.2), and apply the principle of induction.
If n = 1, then P(l) states that 1 = 1, which is true. In order to prove that the
implication P(n) =} P(n + 1) is true, assume P(n) is true and employ P(n) to show
that P(n + 1) is also true. Thus,
n+l
Li 1+2 + 3 + · · • + n + (n + 1)
i=l
n
i=l
n(n + 1)
+ (n + 1) (by the inductive hypothesis)
2
n(n+l) 2(n+l)
+
2 2
(n + l)(n + 2)
2
which is P(n + 1), i.e., (6.2) with n replaced by n + 1. Thus, the formula (6.2) has
been proven by induction.
Example 6.38. Consider the sequence { an }�=l, defined inductively by
We shall employ induction to prove that the sequence is increasing for n 2: 1 and is
bounded above, and, hence, that it converges, by Theorem 6.6.
The first four terms of the sequence are
Since -/7 < 3, we find that
a1 = 1 < 3, a2 = v'7 < 3, a3 = J + v'7 <

6 3, a4 = ✓ + J + v'7 <
6 6 3,
and it seems that an < 3 for every n 2: 1. In order to confirm this, we shall employ
induction, where P(n) is the statement "an < 3."
Since a 1 < 3, P(l) is true. Assuming that P(n) is true (the inductive hypothesis),
i.e., an < 3, we obtain
which is P(n + 1). Thus, it is proved that an < 3 for all n 2: 1, and the sequence is
bounded above.
Since a 1 < a2 < a3 < a4, it seems that the sequence is increasing. Let P(n) denote
the statement "an < an+1•"
Since a 1 < a2, P(l) is true. Assuming that P(n) is true, i.e., an < an+ i, we obtain
which is P(n + 1). Thus, it is proved that an < an+ l for all n 2: 1, and the sequence
is increasing.
By Theorem 6.6, the sequence converges. Let L = n->oo

lim an. Then
an+l = J6 + an => lim an+l

n---+oo
= n---+(X)
lim J6 + an = ✓+ 6 lim an ,
n---+oo
6.1. SEQUENCES 205
where the last equality follows by the continuity of the function g(x) = J6 + x. Since
n ---. oo ⇒ n + l ---. oo,
lim a n+ l = n+lim an+l = lim an = L.

n--+oo 1--+oo n--+oo
Hence, L = J6 + L ⇒ L 2 = 6 + L ⇒ L 2 - L- 6 = 0 ⇒ (L- 3)(L + 2) =0⇒L=3

or L = -2. Since an > 0 for all n 2'.: 1, L 2'.: 0 and, hence, L = 3.
Exercises 6.1
In Exercises 1 8, express the given sequence as {an }�=k.
1. {5, 7, 9, 11, · · ·} 2. {-6, -3, 0, 3, ··· }
1, -}, . ··}
1 1 1 1
3 {1, 4. {1, }
. 2' 4' 9' 1 6' ...
·}
1 1 1 .. 1
5. {1, 6. { 0, 1-- 1-� 1-� ···}
3' 9' 27' 2' 3' 4'
.
9. Wnte down the first five terms of the sequence
{ Jn2 + 1
--- }
2n + 1
00
.
n=O
10. Write down the first five terms of the inductively-defined sequence a1 2,
a2 = 3, an = a n-1 - an -2 for n ;::: 3.
11. Let {an }�=O and {bn}�=l be sequences such that bn = an+ J +; for all n 2'.: 1.
Determine bn+ l for n 2'.: 0 and bn-l for n 2'.: 2.
1 1 1 1
3 5, 7, 9, · · ·
, as {an}�=k in three different ways.
12. Express the sequence { }
In Exercises 13-48, determine whether or not the given sequence { an} converges. If
it converges, find the limit. If it diverges, determine whether or not lim an = ±oo.
n--+oo
1
13. an= -1r
n
3
20. a n = r,;:;
yn
101,000,ooo,ooo,ooo 2 5
21. a n =
no.0000000001 22. a n = 3 - -2 + -n
n 4
1 3n2 2n - 3
n+
-
23. an = 24. an
+ 3n + 1
=
3n 2n2
3n3 - 2n - 3 3n2 - 2n - 3
25. a n 26. an
4n2 + n+ 1
=
+
=
2n3 2n -1
3n2 - 2n - 3 3n2 - 2n - 3
27. a n = ----;:::==== 28. an = ----;:::====
v2n4 + 3n + 1 v2n5 - 2n + 1
3n - 2n - 3
2
29. an = ----;::==== 30. a n = n2 - 2y'n
v'2n 3
- n+1
31. an = Jn2 +n- Jn2 - n 32. an = n - Jn2 +n
33. an = Jn3 +1-n
1r
(-- -1 1 - (-1r
35. a n = - 36. an = -- -
n n
ln(n )] 2
37. an = [
n
n P
40. a n = [ln( )] , p > 0, q > 0
nq
42. an = [ln(n)] 1/ln(n)
l ln( n)
43. a n = [ln(n)] /n
1
44. an = n /
6.1. SEQUENCES 207
cos(n) sin(n2 )
45. an =-- 46. an=
n fo
en (2n)!
-
47. an = - --
48. an -
n! 2nn2 n
In Exercises 49 54, determine whether or not the given sequence { a11 } is increasing
or decreasing for n 2::: N.
oo
49. 50. {2-�}

n=l
51. 52. { ln�n)} �=

2
53.
{ n ln(n)
00
54.
sin n)
{ � }
00
en } n =] n n =l
In Exercises 55-58, determine whether or not the given sequence { an }�=l is bounded
above (an :S K), bounded below (an 2::: M), or bounded (M :S an :S K). In each case,
determine Kand/or M, if it exists (the bounds K and Marc not unique).
n 1
t 55. an =
vnTI
56. an = -
2n
- ln(n)
-1
57. an =
1 58. an=n2 e-n
1--
n+]
. . � n(n + 1)(2n + 1)
59. Prove, by mduct10n, that � i 2 = .
t=l 6
60. Prove, by induction, that the sequence { an }�=l, defined inductively by a1 = 1

and an+l = 1 + .ja;,, for n 2::: 1, is increasing for n 2::: 1 and bounded above,
hence convergent, and find the limit.
61. Consider the sequence { an }�=l, defined inductively by a1 = 1 and an+l = 1-2_
an
for n 2::: 1. Does the sequence converge? Justify your answer.
62. Consider the sequence {an }�=l , defined inductively by a1 =2 and an+l =2 - 2_
an
for n 2::: 1. Prove, by induction, that 1 :S an :S 2 and that {an }�=l is decreasing
for n 2'. 1, and find the limit.
6.2 Series
Definition 6. 7. Let { an }�=l be a sequence. Define the sequence { sn }�=l by
S1 a1
s2 a1 + a2
s2 a1 + a2 + a3
Lai.
n
Sn a1 + a2 + a3 + · · · + an =·
i=l
The sequence
{S1 , S2 , 83 , ··· , Sn, ···}
{ a1, a1 + a,, a1 +a,+ a3,
is denoted by the symbol
La
00
n = a1 + a2 + a3 + · • • (6.3)
n= l
00
and called an infinite series, or simply a series. The series Lan is said to converge
n=l
if the sequence { sn } converges. Otherwise, the series diverges. If lim Sn = s, then s
n-->oo
is called the sum of the series, and
oo n
� an = lim Sn = lim � ai = S.
L n----+oo n----+ooL
n=l i=l
La .
00
The number Sn is called the n th partial sum of the series n

n=l
Since it is impossible to perform an infinite number of additions, Definition 6. 7

gives precise meaning to the concept of an "infinite sum" as the limit of a sequence
of finite sums.
Note that the symbol n in (6.3) is a "dummy index," analogous to a "dummy

variable" in a definite integral, and may be replaced by any other symbol whatsoever.
Thus,
L a = L ai = L= ak = La, = a1 + a2 + a3 + · · ·
00 00 00 00
n
n=l i=l k l (=l
6.2. SERIES 209
all represent one and the same series, determined entirely by the sequence {an } n=l.
In Definition 6.7, the sequence {an } was employed with n = l, 2, 3, · · · for the sake
of definiteness. More generally, if the initial value of n is k, then the corresponding
senes 1s
CX)
Lan = ak + ak+l + ak+2 + · · ·

n=k
For example,
La
CX) CX)
n = ao + a1 + a2 + · · · , and Lan = a11 + al8 + a19 + · · ·

n=O n=17
Since a series is a special type of sequence, the properties of sequences given by

Theorem 6.1 in Section 6.1 are also shared by series. The first two are relevant for
senes:
L L
CX) CX) CX) CX)
Theorem 6.8. If Lan and bn converge, with Lan = s and bn = t, then:

n=l n =l n=l n =l
CX) CX)
1. The series L(an + bn ) converges, and L(an + bn ) = s + l.

n =l n =l
L ca L ca
CX) CX) CX)
2. The series n converges for any constant c, and n = c Lan = cs.

n=l n =l n=l
L
CX) CX) CX)
Note that, if Lan converges and bn diverges, then L(an + bn ) must diverge
n=l n =l n=l
L (a
CX)
because, if n + b11 ) converges, then

n =l
CX) CX) CX)
Lbn = L[(an + bn ) - an ]= L(an + bn ) - Lan

n=l n =l n =l n =l
converges, which is a contradiction.
Example 6.39. Let r be a real number, and consider the sequence {rn }�=o· The
' corresponding series is
CX)
Lrn ,
n =O
called a geometric series. In order to determine the values of r for which the series
converges and to find the sum, consider the nth partial sum
Sn = Lr = 1 + r + r + r3 + • • • + rn-l + rn.
i=O
i 2
Then
rs n r (1 + r + r2 + r3 + • • • + rn-l + rn)
r + r 2 + r3 + ... + rn-1 + rn + rn+l
and, hence,
1 + r + r 2 + r 3 + ... + r n-l + r n
-r_r 2 _ r 3_ ..._r n-1 _ r n _ r n+l
1 - rn+l
⇒ (1-r)s n 1 - rn+l _
If r-=/ 1, then
S n= ---
1-r
By Example 6.14 in Section 6.1, the sequence {rn } converges to O if -1 < r < 1 and
diverges for r ::; -1 or r > 1. Hence,
1-r n+l
lim Sn = lim --- if -1 < r < 1,
1
1-r
and { s n} diverges for r ::; -1 or r > 1. If r = 1, then rn = 1 for all n, and
n-+oo n-+oo 1 - r
Sn = 1 + 1 + 1 + ··· + 1 = n+1
⇒ { s n} diverges. In summary,
�r = - lrl < 1,
�
00
1-r'
1
n=O
n
and the series diverges for lrl � 1.
Example 6.40. Consider the series

1 l
2n I:
n
2
00 00
I:
=O =O ( )
=
n
1 1 1
n
This series is geometric with r = - and, since - = - < 1, the series converges and
2 2
11 2
�2n- 1_.!-
00
1 1
n =O
� - -2
2
6.2. SERIES 211

00 2n 00 ( 2 n
)
�
(-3)n = � -3
This series is geometric with r = -f and, since 1-11 = f < 1, the series converges
and

oo 8 n
9
008 n 00
8 ( ) � 9) '
(
n n
� -2- .3-2 = � -2 =-2
where the last equality follows by item 2 of Theorem 6 8 This series is geometric
. .
i
with r = and, since Ii I i
= < 1, the series converges and
00
'""""' -2 · 8n · 3 -2n = -2--
1
= -18.
� 1- -98
n=O
�5 (-3) 2n+> r 3n = � -15 (�)" = -15� (ff

Th.1s senes
. 1s
. geometric
. wit
. hr and, smce
. 2': 1, the senes
. d.1verges
9 191 9 .
= =
8 8 8
Example 6.44. Consider the geometric series
00
where the initial value of n is 1 instead of 0. Then
n=l
r + r2
00
+ r3 + • • •
-1 + ( 1 + r + r2
n=l
00 + r3 + ... )
n=O
1
-1+--
1-r
r
1-r
More generally, for any k � l,

00
n=k
-1 - r - r2 + (1 + r + r2 + • • • + r k -l + r k + r k+l + ... )
- · · · - r k -l
- (1 + r + r2 + · · · + r k -1) + (1 + r + r 2 + • • •)
-L Lr
k-1 oo
r +
n n
n=O n=O
l -r k l
(by Example 6.39)
-� + 1_r
rk
l-r
For example,
and

00 00
1 (1 1 )
-
� n(n + 1) = � ;, n + l
where the right side is obtained from the left by partial fractions. Then the n th partial
sum of the series is
n (1 1 )
Sn
= � k- k+1
!2 ) + (!2 _ !3 ) + (!3 _ !4 ) +... + (- 1 _ 2:.) + (2:. __ 1
(1 _ )
n-1 n n n +l
1
1---.
n+l
Hence, lim Sn = l, i.e., the series converges and
n-+oo
1
00
� n(n + 1) = 1.
A series in which such cancellations occur in Sn is often called a telescoping series.

6.2. SERIES 213
Example 6.46. Consider a series with only a finite number of nonzero terms, i.e.,
00
with an = 0 for n > m � 0. Then the partial sums Sn of the series are given by
S1 a1
s2 a1 + a2
s3 a1 + a2 + a3
L ak
m
Sm a1 + a2 + a3 + · · · + am =
k=l
L ak
m
Sm+1 a1 + a2 + a3 + ... + am + 0 =
k=l
L ak,
m
Sm+2 a1 + a2 + a3 + ... + am + 0 + 0 =
k=l
L ak for all n � m and, hence, lim S L ak. Thus,

m m
etc., i.e., Sn =
n =
n-+oo
k=l k=l
oo m
"ak = lim Sn = L
"ak,
L n--+oo
k=l k=l
i.e., the series reduces to an ordinary finite sum. Such a series is called a finite series.
The convergence or divergence of a geometric or a telescoping series and, in the

case of convergence, the sum of the series, has been established without any great
difficulty. For other types of series, these properties cannot, in general, be determined
by first principles, and various theorems must be employed.
00
Theorem 6.9. If Lan converges, then lim an = 0. Equivalently, if lim an =/= 0,

n--+oo n--+oo
n=l
00
then Lan diverges. The latter version is often called the nth -term test.
n=l
oo n
"ak is the n th partial sum of the series.
Proof. Lets= Lan = lim Sn , where Sn = L
n--+oo
n=l k=l
Then
Sn a1 + a2 + · · · + an-1 + an and
Sn-l a1 + a2 + · · · + an-1
Hence,
lim an = lim (sn - Sn-1) lim Sn - lim Sn-1 = s - s = 0.
= n---+oo
n---+oo n---+oo n---+oo
The second assertion in the theorem is the contrapositive of the first, and the two are
logically equivalent.
00
Note that the proof fails if Lan diverges for, then, the numbers does not exist,
n=l
and the term s - s has no meaning.
00
l n 1
Example 6.47. The series� ( ) converges since it is geometric with r
3 =
3 and
n
l!I < 1, and lim (!) = 0, as guaranteed by Theorem 6.9.
3 n->oo 3

00
n=l
with the nth partial sum
0, if n is even
Sn l-1+1 -1 + 1 - 1 + ... + (-l) n -1 = { }.
1, if n is odd
=
Thus, {s n}�=l = {1, 0, 1, 0, 1, 0, · · ·} diverges and, hence, the series diverges.

This result can be established much more easily by means of Theorem 6.9. Since
lim ( - 1
n---+oo
r-
1
does not exist, lim ( - 1 1
n---+oo
r-
=/- 0 and, hence, the series diverges.
n
Since lim --;::== = 1 =/- 0, the series diverges by the n th-term test (Theorem 6.9).
n->oo ✓n 2 + 1
6.2. SERIES 215
The reader is warned that the converse of Theorem 6.9 is false, i.e., if lim an = 0,
n-+oo
00
then there is no guarantee that the series Lan converges. It may converge, or it
n=l
may diverge, depending upon the particular series. Examples of both will be given
shortly.
6.2.1 The Integral Test

Let f be positive, continuous and decreasing for x 2:: 1, and let a n = J(n).
y y
a
II
2 3 n-1 n X 2 3 n-1 n x
Figure 6.1: Rectangles of lesser area (left) and rectangles of greater area (right).
Since every rectangle has unit width, its area is equal to its height. Hence, the areas
of the rectangles on the left are a 2, a3, · · · , an , and the areas of the ones on the right
are a 1, a 2, · · · , an -J, as depicted in Figure 6.1. Employing the fact that the area
n
under each curve for 1 :S: x :S: n is i f ( x) dx, we obtain the inequalities
(6.4)
00
Consider the series Lan , and let { sn } n=l be the sequence of n th partial sums. From
n=l
(6.4), we obtain the inequalites

n
Sn a1 + (a2 + a3 +···+an) :S a1 + 1 J(x) dx, (6.5)
n
Sn = (a1 + a2 + a3 + · · · + an-1) + an 2: an + 1 J(x) dx. (6.6)
Since an 2: 0 for n 2: 1, the sequence {s n }�=l is increasing because
00
If / f(x) dx < oo, then, by (6.5) and the fact that f(x) > 0 for x 2: 1,
Sn :S a1
n
+ l J(x) dx :S a1 + 100 J(x) dx < oo
⇒ { s n }�=l is bounded above and, hence, it converges by Theorem 6.6.
If 100 J(x) dx = oo, i.e., the improper integral diverges, then, by (6.6), ·
n n
S n 2: an + J J(x) dx 2: 1 J(x) dx
n 00
⇒ lim Sn 2: lim J J(x) dx = J f(x) dx = oo
n->oo n->oo 1 1
00
⇒ n__.,oo
lim S n oo, i.e., the sequence {sn }�=l and, hence, the series""
= an , diverges.
n= l
L,_;
In summary, we have derived the following:
Theorem 6.10. Let f be positive, continuous and decreasing for x 2: 1, and let
an = f(n). Then the series
00
converges if and only if

j00 J(x) dx < oo.
More generally, if J is positive, continuous and decreasing for x 2: k and an = J(n),

then the series
00
6.2. SERIES 217
converges if and only if

00
This theorem is known as the integral test.
1 J(x) dx <oo.
1
and let f ( x) = 2. Since f is positive, continuous and decreasing for x > l and
X
an= J(n), the integral test applies. Thus,
00
1 1 00
⇒ 1
oo
J
2 d x =--1 =l<oo
1 X X 1 I: n2
n=l
converges by Theorem 6.10. We shall have to wait until Chapter 8 before we shall be
able to determine the sum of this series.

00
1
I: n
-,
n=l
1
and let f ( x) = -. Since f is positive, continuous and decreasing for x > l and
X
a n = J ( n), the integral test applies. Thus,
oo
1
oo 00
1
- dx = ln lxl = oo
ll
⇒
JI X
I:
n=l
n
-
diverges by Theorem 6.10.
Note that, in Example 6.50,

00
1
lim � =0 and converges,
n---+oo n n 2
I:
n=l
whereas in Example 6.51,

00
1 1
lim - = 0 and I:- diverges.
n---+oo n n=l
n
Thus, as stated earlier following the n th -term test, lim an = 0 does not guarantee
n-+oo
00
the convergence of the series Lan . Comparing the two series above,
n=l
00 1 1 1 1 1 1 1 1 1 ...
Ln 2 1+-+-+-+-+-+-+-+-+ (6.7)
n=l 4 9 16 25 36 49 64 81
00 1 1 1 1 1 1 1 1 1
I:�
n=l
1+-+-+-+-+-+-+-+-+···
2 3 4 5 6 7 8 9
(6.8)
we find that, although the terms in (6.8) approach 0, they do so much more slowly
than the ones in (6.7), and that is the reason for its divergence. Whereas the sum in
(6. 7) approaches a finite number, the sum in (6.8) increases indefinitely.
Example 6.52. Let p be a real number, and consider the p-series

00 1
I:
n=l nP.
If p < 0, then -p > 0 and lim � = lim n-P = oo-/= 0. Hence, the series diverges
n-+oo nP n-+oo
by the nth-term test.
1
If p = 0, then lim - = lim 1 = 1 -/= 0. Hence, the series diverges by the nth -term
n-+oo nP n-+oo
test.
1
If p > 0, let f(x) = -. Since f is positive, continuous and decreasing for x �land
xP
an = f(n), the integral ·test applies. The series diverges if p = l by Example 6.51
and, for p -/= 1,
,
00 1 , x1 -P 1 00 oo, if p < l
J J
00
x P dx=-- = { 1 ' if p >l}

_
-dx= .
1 xP l 1-p i p-1
Hence, the series converges for p > land diverges for p ::; 1 by Theorem 6.10.

00 1
I::--
n=lfo
1 1 1 1
Since
fo n /2
=l , this is a p-series with p =
2 and, since
2 ::; 1, the series diverges
by Example 6.52.
6.2. SERIES 219
1
00
�nvn·
3 3 .
S.1ncel = - l -, th.1s 1s
· a p-senes
· wit
· hp = - and, smce
. - > 1, the senes converges
nyn
r,;;; n31 2 2 2
by Example 6.52.
1
00
� nln(n)"
1
Let J(x) = . Since f is positive, continuous and decreasing for x > 2 and
x 1 nx( )
an = J(n), the integral test applies. Thus, making the substitution u = ln(x),
12
00
- 1-dx =
X ln( X)
f00 ..!. du= ln 1u1l
J1n(2) U
00
= 00.
Jn(2)
Hence, the series diverges by Theorem 6 10.

.
Approximations of Series
00
Let Lan be a convergent series. If the sum s of the series cannot be determined,
n=l
then it can be approximated. By definition of limit, since
oo n
s = � an = lim = lim � ak,
L n-+oa
Sn
n-?OO L
n=l k=l
the terms Sn can be made arbitrarily close to s by taking n sufficiently large. Thus,
Sn is approximately equal to s, written Sn � s, and the larger we take n, the better
the approximation. Whenever an approximation is made, knowledge of the maximum
possible error is essential. The error made in approximating s by Sn is
DO n 00
n=l k=l k=n+l
called the remainder of the series.

00
In the present discussion, we restrict our attention to series Lan where an = f(n)
n=l
and f is positive, continuous and decreasing for x 2". 1. In order to determine upper
and lower bounds for Rn, we appeal to Figure 6.2.
From the graph on the left, we obtain
Rn= an +I + an+2 + · · · � 1 J(x) dx, (6.9)
and, from the one on the right, we find
Rn= a n+I + an +2 + · · · 2': 100

n+l
J(x) dx. (6.10)
y y
2 3 n n+l n+2 X 2 3 n+l n+2 n+3 X
Figure 6.2: Upper and lower bounds for the remainder Rn .
Combining (6.9) and (6.10), we obtain the inequalities
1 00
n +I
f(x) dx�Rn� 100
n
J(x) dx. (6.11)
Hence, the error m approximating s by

00
Sn is at least 100 n +l
f (x) dx and at most
1 J(x) dx.
6.2. SERIES 221
Example 6.56. Consider the p-series
Since p = 3 > 1, the series converges by the integral test (Theorem 6.10). If the sum
of the series is approximated by s2 , i.e., n = 2, then
=
s=L 3 � s2 = L k23 = 2 + 41 = 2.25,
2
n
2
n=l k=l
and
1 1 1
= =
2
x 13 2
x1 1 2
⇒
.
-- :S R2 :S - - - < R2 < -.
9 - - 4
1 1
Thus, the minimum error is � 0.111 and the maximum error is = 0.25. If, instead,
the sum is approximated by s5, i.e., n = 5, then
9 4
2 2 1 2 2 2
= � -3 ,.:;::; =�- = 2 + -4 + - + - + - ,.::::; 2.37'
� k3
= 5
�n 27 64 125
S S5
and
1 1 1 1
<R5<-'
=
x2 1 5
=
x2 36 - - 25
⇒
16
-- :S R5 :S - - -
1 1
i.e., the minimum error is - ::::::; 0.028 and the maximum error is -
36 25 = 0.04, with a
much better approximation.

=
� n[ln(n)]2 ·
1
1
Let f(x) = x[ln(x)]2. Since f is positive, continuous and decreasing for x � 2, the
integral test applies, and the series converges because
1 1 1
= -ln(x) I =
r= .
J2 x[ln(x)]2 dx ln(2) < oo
2
Suppose that an approximation of the sum s of the series is required such that the
error is no greater than 0.1. How large mustn be taken? Since
1 1 1 loo 1
00
Rn< ---dx = ---

- n x[ln(x)]2 ln(x) n ln(n)'
1
the required precision is guaranteed if - ::; 0.1, which requires that ln(n) > 10,
ln(n)
i.e.,n :2: e 10 � 22,026.47. Hence,n :2: 22,027.
3n2
00
I:=l
n
en
3
·
3x2
Let f ( x) = � = 3x2 e-x . Then f is positive and continuous for x :2: 1, and
3
for x :2: 1 ⇒ f is decreasing for x :2: 1. Hence, the integral test applies, and the series
converges because
Suppose that an approximation of the sum s of the series is required such that the
error is less than 0.0001. Note that R n= s - Sn> 0 since {s n}�=l is increasing. How
large mustn be taken?
1 3x2 1 1
00
Rn< = -- I = - < 0.0001

CX)
- -dx
n ex3 ex3 n e n3
is achieved by taking n such that e

n
3
> 10,000, i.e.,n > [ln(lO,000)] 1 1 3 � 2.096.
Hence, n :2: 3 .
Comparing the results of Examples 6.57 and 6.58, we notice that a large degree of
accuracy is obtained in the latter by taking only 3 terms of the series, whereas fairly
poor accuracy is obtained in the former by taking as many as 22,027 terms. Such
discrepancy is due to the fact that the terms of the series in Example 6.58 approach
0 much more rapidly than the ones in Example 6.57, and the addition of more terms
has a small effect upon the final sum.
6.2. SERIES 223
6.2.2 The Comparison Tests

It is frequently possible to determine whether a given series converges or diverges
by comparing its terms to those of series the convergence or divergence of which is
already known. Two such comparison tests are given in the present section.
Theorem 6.11. Let {a n } and {bn } be sequences, and suppose that
for all n � k, where k is an integer. Then:
Lb La
00 00
1. If n converges, then n also converges.

n=l n=l
Lb
00 00
2. If Lan diverges, then n also diverges.

n=l n=l
This theorem is known as the comparison test.
Note that the initial value of n is given as 1 for definiteness, and may be replaced
by any other integer, and the initial values of n in the two series need not be the
same. The reason is that, for any integer m,
00
m-1 00
n=l n=l n=m

00 00
and Lan converges if and only if Lan converges because a finite number of terms
n=l n=m
La
m-1
such as n do not affect convergence.
n=l
Since
1 1
O<--<
- n2 + 1 - n2
for all n � 1 and the series L 2n1 is known to converge (it is a p-series with p = 2 > 1),
00
n=l
1 1
it follows by the comparison test (Theorem 6.11) with an= -2-- and bn = 2 that
n +1 n
1
00
� --- converges.
Ln
n=O
2 +1
1
00
Ln-1·
n=2
Since
L n-1 is known to diverge (it is a p-series with p = 1 :S 1),

00
for all n 2: 2 and the series
1
n=l
it follows by the comparison test (Theorem 6.11) with an _!__ and bn = - - that
n n-1
=
1
00
� -- diverges.
Ln-1
n=2
Since
00 2 n
for all n 2: 0 and the series ( ) is known to converge (it is a geometric series
� 3
2 2 2
with r = and 1 1 = < 1), it follows by the comparison test (Theorem 6.11) with
3 3 3
2n 2 n 00 2n
an = + and bn = ( ) that + converges.
3n 1 3 � 3n 1
Since
6.2. SERIES 225
3
= n
for all n 2'. 2 and the series � ( ) is known to diverge (it is a geometric series
2
with r =�and �
, , =�2: 1), it follows by the comparison test (Theorem 6.11) with
3 n 3n = 3n
an = ( ) and bn = that � n _ diverges.
2 2
n _
3 2 3
(X) 1
Ln+1·
n=O
The inequality
1 1
O<--<
- n+ 1 - n
for all n 2'. 1 is of no use, because L -n1 diverges. Instead, employ the inequality
(X)
n=l
1 1 1
-->--=->0
n + 1 - n+ n 2n -
=
= 1 1 1 = 1
for all n 2'. 1. Since � - = - � - diverges, � -- diverges by the comparison
L2n
n=l
2Ln
n=l
Ln+l
n=O
test.
f ln(l
n
+ ¼).
n=l
In order to make use of the fact that lim

n-->(X)
(1 + n�) n
= e, express the given series as
� ln(l + ¼) = � nln(l + ¾) =
� ln[(l + ¾) ]
n
.
L n L n2 L n2
n=l n=l n=l
Since the logarithmic function is continuous,

By definition of limit, the quantity ln [ ( 1 + ¾) n] can be made arbitrarily close to

1 and, hence, less than 2, by taking n sufficiently large. Thus, there exists an integer
N such that

ln(l + ¾) _ ---�-
ln[(l + ¾rl 2
< - n � N.
n n2 - n2 '
L 2 1
L ln(l + .!.n )
L
00 00
Since 2 =2 2 converges, it follows by the comparison test that
n n n
converges.
n=l n=l n=l
� [ln(n)] 6
L n2
n=l
[ln(x)] P [ln(n)J P
Since lim = 0 for any p > 0, q > 0, by L'Hopital's rule, lim = 0 by
X->00 Xq n->OO nq
Theorem 6.3. Thus,
[ln(n)] 6 [ln(n)] 6 1
n2 nl/2 n3/2 '
[ln 1 6
and lim tl = 0 =} there exists an integer N such that
n->00 n
[ln(n)] 6 [ln(n)] 6 1
nl/2 <
1 =} 0 :s: n2 :s: n3/2 '
and, since f
n=l
; converges,
n /2
f
n=l
[ln ) 6
�� ] converges by the comparison test.
5
00 00
Lan =
n=l
L
n=l
n3 - ;: 2n + 3 ·
In order to determine which inequalities to employ, it is necessary to deduce the

"large-n behaviour" of an . In the limit as n - oo, n + 5 ~ n, read "n + 5 behaves as
n," and n3 - n2 + 2n + 3 ~ n3 . The reason is that the term with the highest power
of n dominates the expression for large n. Hence,
n+5 n l
n3 - n2 + 2n + 3 rv n = n 2'
3
6.2. SERIES 227
L n-; converges, we must show that a

00
and, since n S --; for all sufficiently large n,

n=l
n
where c is a constant. Thus,
n+5 n+5 n+n 4

-------<---<---- n ::::: 5,
n3 - n + 2n + 3
2 n n
3 - 2 - n3 - l.2 n3 n2'
00
1
since n + 5<
- n + n for n -
> 5 and --n3
2 <
2
> 2. Hence, Lan converges
- -n for n -
n=l
by the comparison test.
As seen in Example 6.66, certain of the inequalities required for the comparison
test may become too tedious. An alternative test, which is especially suitable for
series such as the one in Example 6.66, is given by the following:
Theorem 6.12. Let {an } and {bn } be sequences with an > 0 and bn > 0 for n::::: k,
where k is an integer, and let
. an
L = hm -.
n ---+oo bn
Lb
00 00
1. If O<L<oo, then either both Lan and n converge, or they both diverge.
n=l n=l
Lb
00 00
2. If L = 0 and n converges, then Lan converges.

n=l n=l
Lb
00 00
3. If L = oo and n diverges, then Lan diverges.

n=l n=l
As in the comparison test, the initial value 1 of n may be replaced by any other
integer. This test is called the limit comparison test..
Example 6.67. Consider, once again, the series in Example 6.66,
L n - ;:52n + 3 ·
00 00
Lan = 3
n=l n=l
Since we have deduced in Example 6.66 that a n rv -;, let bn = -;. Then
n n
a n3 + 5n2
L = lim _!I: = lim ------- = 1
n ---.oo bn n ---.oo n3 - n2 + 2n + 3
00 00
and, since O < 1 < oo and Lb n converges, Lan converges by the limit comparison
n=l n=l
test (Theorem 6.12).

00 00
1
'°"'an =
L n1+-n
'°"'-1-
L
n=1 n=l
Since lim x 1 lx = 1 by L'Hopital's rule, lim n 1 /n =1 by Theorem 6.3. Hence,

X---+00 n---+CX)
1 1 1 1
1
for large n. Let b n = -. Then
n
an 1
lim - = lim -- =1
n---+oo bn n---+oo n / n
l
00 00
and, since O < 1 < oo and Lb n diverges, Lan diverges by the limit comparison
n=l n=l

00 00
n2 - 2n + 1
�an � .Yn 10 + 3n2 + n ·
=
Since, for large n,

n
2 - 2n + 1 n
2 n
2 1
.Yn 10 + 3n2 + n rv � = nl0/3 = n /3'
4
1
let bn = 4 . Then
n /3
n l0/3 _ 2n7/3 + n4/3 1- ln + ...!...

n2 =1
lim --;;-;::======-- lim
n---+oo .Yn 10 + 3n2 + n n---+oo 3/l
+ --1_
n8 + ...!...
n9
V
00 00
and, since O < 1 < oo and Lb n converges, Lan converges by the limit comparison
n=l n=l
6.2. SERIES 229
Example 6. 70. Consider the series

� � [ln(n)] 2
a _
-
L n L n 3/2
n=l n=l
oo
L
1 [ln(n)] 2
Since - converges for any p > 1 and lim = 0 for any q > 0, choose any
n- oo �
.:.....--'---c._
n=l
�
3 5 1 1
p > 1 and q > Osuch that p + q = , e.g., p = and q = , and let bn = 5 4. Then
2 4 4 n1
an [ln(n)] 2 =0
L = nlim = nlim
-oo bn -oo n l /4
Lb
00 00
and, since L = 0 and n converges, Lan converges by the limit comparison test
n=l n=l
(Theorem 6.12).
Example 6. 71. Consider the series

00 00
1
�an =
� Jn+T[ln(n-1)]7'
1
and let bn = -. Then
n
a n n 1/2
L = lim ___!1:.
n -oo bn
= lim -------
n -oo Jn+T[ln(n - 1)] 7
=
F+I
lim ------ -
n- 00
- = oo
[ln(n _ 1)]7
00 00
[ln(n - 1)] 7
because lim .:;____
n -oo nl/2
= 0. Since L = oo and "b
_c._
L n diverges, L
"an diverges by
n=l n =3
the limit comparison test (Theorem 6.12).
6.2.3 Alternating Series

Definition 6.8. A series of the form
00
I)-l)nbn = bo - b1 + b2 - b3 + b4 - b5 + · · · ,
n =O
where bn > 0 for all n � 0, is called an alternating series. More generally, for any
integers k and m,
n =k
with bn > 0 for all n � k, is an alternating series.
Thus, an alternating series is one where the terms are alternately positive and
negative.
Example 6. 72. The series
1
> 0.
is an alternating series with bn = - - > 0 for n -
n+ 1
f (-1r- = 1 __1 + _1 _� + ...

1
n=l y'n y12 y'3 2
1
is an alternating series with bn = Jn > 0 for n 2 1.
I)-1r sin(n)
n=l
is not an alternating series because there are values of n for which bn = sin(n) f 0.
Theorem 6.13. Consider the alternating series

00
If {bn } is decreasing for n 2 k, where k is an integer, and lim bn = 0, then the series
n->oo
converges. This theorem is called the alternating series test.
Example 6. 75. The alternating series
(-1r 1 1 1
L--
00
= -1+- -- +- - ...
n
n=l
2 3 4
converges by the alternaLing t>eries test (Theorem 6.10) since { �} �= is decreasing

l
1
for n 2 1 and lim - = 0.
n->oo n
6.2. SERIES 2 31
Example 6. 76. Consider the alternating series

ln (n)
I)-1r .
n
n=2
ln(x). 1 - l ( x)
Let J(x) = Then f'(x) = �
X
< 0 for x > e ⇒ J is decreasing for x 2:: e.
X
ln (n) ln(n) .
Hence, ·{ -- }
00
is decreasing for n 2:: 3 > e, an d, since lim --

n-+oo
= 0, the senes
n n=2
n
converges by the alternating series test (Theorem 6.13).
Approximations of Alternating Series

Theorem 6.14. Suppose that b n > 0 and {bn } is decreasing for n 2:: 0, and that
lim bn = 0, so that the alternatin g series
n-+oo
00
n=O
converges by the alternating series test. If the sum

00
n=O
is approximated by the nth partial sum

n
I)-llbk = bo - b1 + b2 - · · · + (-ltb n ,
Sn =
k =O
then the absolute value of the error is less than the absolute va lue b n+l of the first
neglected term (which is ±b n + 1). In other words,
IR..I
= Is - sn l < bn +l,
L (-l) b is the remainder, as defined in Section 6.2.1.

00
where Rn= k
k
k=n+l
Proof.
00
S - Sn
k= n +l
(-1r+ 1 bn+ l + (-l)n +2b n +2 + (-1r+3 b n+3 + (-1r+4 bn+4 + ...

(-1r+1 (bn+l - bn+2 + bn+3 - bn+4 + b n +5 - · · ·)
b n +l - bn +2 + bn +3 - b n +4 + bn +5 - · · ·
bn +l - (bn +2 - bn +3) - (bn +4 - b n +5) - · · ·
< bn +l
because {bn } is decreasing for n 2: 0 => -(bn+m - bn+m+1 ) < 0 for all m 2: 0.

00 (-1r-1
I:
n=l
n4
Since {bn} = { �4 } is decreasing and bn > 0 for n 2: 1, and J�� bn = 0, the series
converges by the alternating series test (Theorem 6.13). Suppose that the sum s of
the series is approximated by s3, i.e.,
(-1r-1 (-l)k-1 1 1
s L L 1- � 0.9498.
oo 3
= n4 � S3 = k4 = 16 +
81
n=l k=l
Then
1 1
IR3I = Is - s3I = 0.003 ,
44 256 � 9
<
i.e., the error, in absolute value, is at most 0.0039.

(-1r
00
� ln(n) ·
Since {bn } is decreasing and bn > 0 for n 2: 2, and J�� b = 0, the series
= { lntn)} n
converges by the alternating series test (Theorem 6.13). How large must n be in order
that the error in the approximation of s by Sn be less than 0.0001 in absolute value?
Since
lerro rl = IRnl = Is - sn l < bn+ l,
the required accuracy is achieved by requiring that
1
bn+ l = l '.S 0.0001,
ln(n + )
which holds if and only if
ln(n + 1) 2: 10,000,
i.e.'
n> e 10 ,ooo - 1
- '
1
which is an astronomically large number. The reason is that the terms - - - 0
ln n)
(
far too slowly compared to, say, -;- , and such a large number of terms is required in
n
order for the finite sum Sn to become sufficiently close to the infinite sum s.
6.2. SERIES 233
6.2.4 Absolute and Conditional Convergence

00 00 00
Theorem 6.15. If L la l converges, then
n Lan converges. Equivalently, if Lan
n=l n=l n=l
00
diverges, then L la l diverges.
n
n=l
00 00 00
If L lan l converges, then L 2lanl = 2 L lanl converges and, by the comparison
n=l n=l n =l
00
test, L(an + lan l) converges. Hence,
n=l
00 00
Lan = L[(an + lan l) - lanl]
n =l n=l
converges by Theorem 6.8. The second statement is the contrapositive of the first,
and is logically equivalent to it.
Example 6.79. Since f-\-

n=l n
converges (it is a p-series with p = 2 > 1) and
00 (-1r 00 _!_
L 1 2 l= L 2
n=l
n n' n=l
L(--
-1r
00
it follows by Theorem 6.15 that converges. Of course, this result can also
n2
-
n=l
be established independently by the alternating series test.
00
Example 6.80. The series L(-1r
--n
converges by the alternating series test, but
n=l
the series
001(-1rl 00 i
I:-
n =I:-n
n=l n=l
diverges (it is a p-series with p = 1 � 1).
00
Examples 6.79 and 6.80 demonstrate that, for certain series, both La n and
n=l
00
L L
00 00
lanl converge, whereas for others, La n converges but lanl diverges. Of course,
n=l n=l n=l
00 00
if L lanl converges, then La n converges by Theorem 6.15.
n=l n=l
00
Definition 6.9. A series La n converges absolutely if both
n=l
00 00
n=l n=l
00 00 00
converge. If La n converges but L lanl diverges, then the series Lan converges
n=l n=l n=l
conditionally.
00
If a n 2: 0 for all n 2: 1, then lanl = a n and, hence, La n converges if and only if
n=l
it converges absolutely. Thus, there is a difference between absolute and conditional
00
convergence only for series La n where the a n take both positive and negative values.
n=l
00
It must be emphasized that, whether La n converges absolutely or conditionally,
n=l
it converges. The distinction between absolute and conditional convergence pertains
00
only to whether or not the series L
lanl converges.
n=l
00
Example 6.81. By Example 6.79, the series L--
(-1r
n2
converges absolutely and,
-
n=l
00
by Example 6.80, the series L (--
-1r
n
converges conditionally. Note that both series
n=l
converge. The distinction arises because, in the former,
00 00
L
(-1r _
1 n2 l - Ln2
2-
n=l n=l
converges, whereas in the latter,

00 00
(-1r i
1
L-=L- \
n=l
n n
n=l
6.2. SERIES 235
diverges.
This series converges by the alternating series test. However,
1
I (-1r I (X)
� nln(n)
(X) = � nln(n)
oo
(-l)n
diverges by the integral test. Hence,� converges conditionally.
nln(n)
L�-
00
( 2r -
or r
n=O
�J�= "I = t, �: = t,
The series
)
oo
. .
converges smce 1t 1s geometnc . h I r I < 1. Hence,
. wit
2 L (-
5n converges abso 1 ute1y.
--
. n=O
Example 6.84. The series
(-1r ✓n2 + 1
diverges by the n th-term test because lim - - -- =f. 0 (the limit does not
n-->oo n+l
exist ). It follows by Theorem 6 15 that
f
.
(-1r ✓n2 +1 = f Jn + 2 1
n+ l n +l
n=O n=O
Jn2 + 1 = 1 =f. 0.
diverges, a result which can be established independently since lim
n-->oo n+l
Theorem 6.16. Let {an } be a sequence and suppose that
L = lim
n---->
exists. Then:
00
1. If L < 1, then Lan converges absolutely.

n =O
00
2. If L > 1, then Lan diverges.

n=O
3. If L = 1, then no conclusions can be drawn regarding convergence.
As usual, with regard to convergence, the initial value O of n can be replaced by any
other integer. This theorem is called the ratio test.
Note that, since the ratio test involves lan l and not an, it cannot be employed as
a test for conditional convergence.
an+l = 1 3 2n ! 1 3 !
I an I n---->oo 2n+l n 2 n---->oo n
L = lim lim (n + ) 3 = lim (n +3 ) = < 1
n---->oo 2
00
⇒ Lan converges absolutely by the ratio test (Theorem 6.16). Thus, both
n =O
3
L L=O lanl = L=O ;
00 00 00 00
(-1rn3
Lan = 2n
and n
n=O n=O n n
converge.
6.2. SERIES 237
an 3n+l n4 + 1 n4 + 1
I an I n--->oo (n + 1)4 + 1 3n
L= lim _ _+_l = lim ----- = 3 lim ---- - =3 > 1
n--->oo n--->oo (n + 1)4 + 1
00
=? Lan diverges by the ratio test (Theorem 6.16). Thus, both

n=O
diverge.
Example 6.87. For the series
which diverges,
L= h. m lan+ll
-- = h. m-- n
=1
n--->oo an n--->oo n + l
and, for the series
which converges,
I + I = lim
an l 2
n
L = lim = 1.
n--->oo a n n--->oo ( n + l ) 2
Thus, convergence or divergence cannot be determined by the ratio test if L = l.

(X) 00
'
� an = � !!..:_.
� �nn
n=l n=l
. n n +l)!
S1nce an > 0, I anI = an= -n! and I an+l I = ( ( + 1 )n+l
n n
⇒
a_+l_ (n+ l) nn = l m _(_ n +l)n

L lim 1- n lim _ _ _ n_ ! _ _ _ _ _ l im
n
i
n
I= =
n-+oo ( n +l)n+l
n
n-+oo n-+oo ( n +l) +l n! n-+oo ( n + l)n

n
an
. n . 1 lm 1 1
1 lffi ( -- ) 1lffi - i --- = - < 1.
n-+oo n+ 1 n-+oo ( !l )n n-+oo (1 + e
¾t
--
n
> 0, convergence and absolute

00
Hence, Lan converges by the ratio test. Since an

n=l
convergence are synonymous.
(2n+ 2)! 2n(n!)2 = (2n + 2)(2n+ 1) n!

2
[ ]
2n+ 1 [(n+ 1)!]2 (2n)! 2 (n+ 1)!
(n
+ 1)(2n+ 1) (-1-)n+l
2 2n+ 1
n+l
an l n+ 1
+ I = 1.lffi 2---
⇒ L 1.lffi I -- = 2 > 1.
n-+oo an n-+oo n + 1
Hence, Lan diverges by the ratio test.

00
n=O
� _ � 1 ·3 · 5 ···(2n + 1) .
L an - L 3nn !
n=O n=O
1·3·5·· · (2n+ 1)(2n + 3) 3nn ! 2n+3

3n + 1 (n+l)! 1·3·5···(2n+l) 3( n + 1)
2n 3
⇒ L lim I n+l I = lim + = � < 1.
n---+oo 3( n + 1)
a
n---+oo an 3
00
Hence, Lan converges by the ratio test.

n=O
6.2. SERIES 239
Theorem 6.17. Let { an } be a sequence and suppose that
L = nlim lan l l /n
---+=
exists. Then:

n=O
(X)

n=O
Again, the initial value O of n can be replaced by any other integer. This theorem is
called the root test.
As for the ratio test, the root test cannot be employed as a test for conditional
convergence.
= = [ln(n)] 5n
Lan L
=
n3
n=l n= l n
Since
ln n 5
L = nlim lanll/n = lim [ ( )] = 0 < 1,
---+
(X) n ---+(X) n3
Lan converges by the root test (Theorem 6.17).

n=l
.£)
Since
n
f, = lim lan l l /n = lim (1 + = e > 1,
n�oo n�oo n
(X)
Lan diverges by the root test (Theorem 6.17).

n= l
Exercises 6. 2
00
1
1. Write down the n th
partial sum Sn of the series � -
L..t 2n
for n = 0, l, • • • , 4.
n=O
2. Write down the nth partial sum Sn of the series f�

n=l n
for n = l, 2, · · · , 5.
3. Let { sn }�=l denote the sequence of partial sums of the series Lan . Determine
n=l
3n - 1
the sum s of the series if Sn = + for n � l.
2n 1
In Exercises 4-11, determine the sum of the given series.

00 00
1 2n
4. L5n 5.
n=O � (-5) n
oo 2. 3n 00
1
6. I: 7. L3n
n=l
�
n= 2
8. L 5. 2
00
3n
3-2n 9. L5 . 3
00
2n
2-4n
n=O n= l
L 5 (-2)-3 L 4n +8n+3
00 00
n n 2
10. 11. 2
n=2 n=O
In Exercises 12-23, determine whether the given series converges or diverges.
12.
� 2Jn2 +3
L..t 13.
f cos(n)
n n=O 2 + sin (n)
2
n=l
n3 + 2n2 - 1
00 00
(-1ri 15.
11. I:[1 - �- -
L..t n2 +n+l
n=O n=O
00
1 �vn
16.
L
n=l n2 vn
17.
L..t n2
n=l
00
1
L=l � r-1
'
CX)
18. r+l , r>0 19. L..t
n
n
n n=l
6.2. SERIES 241
1
00 00
20. 21. L2ne-n

2
L2n-l
n=l n=l
1 1
00 00
22. P' p < 1 23. p>l

� n[ln(n)] � n[ln(n)]P'
24. Consider the series f 3

n=l n
4
.
(a) Approximate the sums of the series by s 2 and determine lower and upper
bounds for the error R2.
(b) How large must n be taken in order that Sn approximate s with error no
greater that 0.001?
25. Consider the series f

n=l
1
nyn
..
r,;:;
(a) Approximate the sums of the series by s 3 and determine lower and upper
bounds for the error R3.
(b) How large must n be taken in order that Sn approximate s with error no
greater that 0.0001?
In Exercises 26-45, determine whether the given series converges or diverges.
1 1
00 00
26. � 27.
nyri,+3 �yn-1
3n 3
00 00
28. 29. L 2
L4n+1
n=O n=O 2n -3
2n + 1 2n + n
oo 00
30. L�
n=O
31.
L
n=O
en
1
f vn-1
00
32. L 33.
n=l vn+n n=l n+l
34.
f n3 +2n-1
3 +2
35. L 2
00
-{jn2 +1
n=O n +n n=l n +n-l
4
2 42 CHAPTER 6. SEQUENCES AND SERIES
36.
�
00
n 213
1
ln(n)
37.
f ln(n)
n2
n =2
38.
f ln(l + ¾)
39.
f [ln(n)] 3
nfo
n=l vn n=2
oo (-)l n (-r
I::
1
40. 41. �
n=l n 2/3 +
2 sin(n)
00
(-1r 00
(-r l n
42. 43.
�Jln(n) � en ln(n)
44.
f (-1r - 1
n
45.
f 1+(-r
n 2
l n
n=l n=l
L (-n1r-1
00
46. Consider the series 2

n=l
(a) Approximate the sums of the series by s 3 and determine an upper bound
for IRnl-
(b) How large must n be taken in order that Sn approximate s with error less
than 0.0001 in absolute value?
L n(-+r1 1
00
47. Consider the series 3

n=O
(a) Approximate the sums of the series by s 2 and determine an upper bound
for IRnl-
(b) How large must n be taken in order that Sn approximates with a maximum
1
error of ?
1, 001
-
±
In Exercises 485
- 3, determine whether the given series converges absolutely, converges
conditionally, or diverges.
(-1r (-1r+1
00 CX)
48. L n -1 49.
n=2
� n Jn
50.
f (-l)n vn
51.
00
(-2r
n+l L�
n=O n=l
52. f (-l)n [ln(n)] 2

n
53.
00
(-2r
L 3n _ 1
n=2 n=I
In Exercises 54 63, determine whether the given series converges or diverges.
2n(n + 1)3 3n-l n 2

L
00 CX)
54. 3nn 2 55. � n 1

n=l
2 + (n + 1)4
n 3 + 2n2 - 1 2n
L=O L
00 CX)
56 . 57.
n, (n + 1)!
n n=O
(n!)2 nl
!;
CX) CX)
58. 59.
n 2 (2n)! � 3 · 6 · 9 ·. · · (3n)
60.
(2n + 1)!
00
� n !(n + 1)!
61.
f n !(n+ 3)!
(2n + 3)!
n=O
CX)
(-n r (-1rn2n
CX)
62. � [ln(n)J 2n 63.

� 2n (n 2 + l)n
Chapter 6 Exercises
1 - rn+l
1. Prove, by induction, the formula Sn --- for the n th partial sum of the
1 -r
=
Lrn.
CX)
geometric series
n=O
La
CX) CX)
2. Prove that Lan converges if and only if n converges for every integer
n=l n=ni
m 2': 1.
3. Determine whether or not the series

f
n=l6
n ln(
L
n){ [ln(n)]} c
onverges.
4. Determine whether or not the series '°'

00
1
L1+2+3+···+n
n=l
converges.
5. Determine whether or not the series '°'

00
n
L1+4+9+···+n
n=l
2
converges.
6. Prove that every finite series converges absolutely.
7. Consider the Fibonacci sequence a1 = a2 = 1 and an = an -I + an -2 for n 2: 3.

The latter relation is called a difference equation, which is the discrete version
of a differential equation. Find two solutions in the form an = rn , take a linear
combination, and impose the "initial conditions" a 1 = a2 = 1 to determine an
explicitly in terms of n for every n 2: 1.
Chapter 7
Taylor Series
The subject of Chapter 6 has been the study of sequences and series of numbers. The
topic of the present chapter is the study of series which contain a variable x, i.e.,
series of functions. Thus, the convergence or divergence of the series, as well as its
sum, will depend upon the value of x, making the infinite series a function of x.
7.1 Power Series

Definition 7.1. Let a be a real number. A series of the form
00
2:::C n (x - a)11 =Co+ c1(x - a)+ c2 (x - a) 2 + c3(x - a)3 + · · · (7.1)

n=O
is called a power series about a or centred at a, and the numbers Cn, n = 0, l, 2, · · ·

are called the coefficients of the power series. In the special case where a = 0, the
series reduces to
L
00
CnX
n
= Co+ C1X + C2X + C3X + · · · ,
2 3
n=O
i.e., a power series about 0.
Note that, at x = a, the power series (7.1) reduces to the single term Co and,
hence, converges absolutely. However, only those power series which converge for all
x in an interval arc of any practical use.
Example 7.1. The power series

00
Lnnx n
n=O
245
246 CHAPTER 7. TAYLOR SERIES
is centred at 0 and has the coefficients Cn =n n

. By the root test (Theorem 6.17), the
series converges absolutely if
lim lan l l / n lim ln

= n---+oo n
x l /
n l n
lim nlxl <
= n-+oo 1,
n---+oo
which is possible only if x = 0. Thus, this series converges only at x = 0.
Example 7.2. The geometric series (with r = x)

00
n=O
is a power series about 0 with the coefficients = 1, n 2: 0. It converges to - -

Cn
1-x
for lxl < 1 and diverges for lxl 2: 1, as demonstrated in Section 6.2, Example 6.39.
For lxl < 1, it converges absolutely because
L lx l = L lxl
00 00
n n
n=O n=O
and llxll = lxl < 1.
Example 7.3. Consider the power series

oo
L -x
n!
n=O
1 n
l xn
about 0 with the coefficients Cn = , and let an Since
1
= -•
n. n.1
an+ ·l x +l n! l xl
lim I 1 lim I --- lim -- = 0 < 1
n
= n-+ I = n-+
n-+oo an oo (n + l) ! x n
oo n +l
for all x, the series converges absolutely for all x by the ratio test (Theorem 6.16).
The preceding three examples indicate that, given a power series about a, the
series may converge only at a, it may converge for all x in a finite interval, or it may
converge for all real x.
Definition 7.2. If the power series
LCn(x - af
n =O
7.1. POWER SERIES 247
converges absolutely for Ix - al < R and diverges for Ix - al > R, then R is called
the radius of convergence of the power series. If the series converges only at x = a,
then R = 0. If the series converges absolutely for all real x, then R = oo.
Note that Ix - al is the distance between x and a, and Ix- al < R is the interval of
all points x such that -R < x -a < R, i.e., a - R < x < a+ R. The term "radius" is
employed because, in the event that x be a complex number, the set Ix -al < Ris a
disc of radius R centred at a. We shall, however, restrict our attention to real numbers.
The endpoints of the interval Ix - al < R are the two points Ix - al = R, i.e.,
x =a± R. At these values of x, the series may converge absolutely, it may converge
conditionally, or it may diverge. The interval J on which the series converges, either
absolutely or conditionally, may therefore be any one of
[a - R, a + R], [a - R, a + R), (a - R, a + R] and (a - R, a + R),
and is called the interval of convergence of the power series.
In Example 7.1, R = 0 and I = [O, O], in Example 7.2, R = 1 and I = (-1, 1),
and, in Example 7.3, R = oo and I= (-oo, oo).
Theorem 7.1. The radius of convergence of the power series

00
is given by
R= lim I -- ,
Cn
n-->oo Cn l
+ I
provided that the limit exists.
Proof. Let an = Cn (X - ar. By the ratio test (Theorem 6.16), the series converges
absolutely if
.n-->oo I a-- l I = Ix - a I 1·1m I -- l I < 1,
n+ Cn+
lim
an n-->oo Cn
i.e.,
Ix - al<
1�1,
1
;� I �n l I
= nlim
--> oo Cn +1
and diverges if
Ix - al> lim
1�1-
n -->oo Cn l
+
Hence,
R= hm -n-
. I I
C
n->oo C n+ 1
is the radius of convergence.
Once R has been determined, the series converges absolutely at least for
a-R < x < a+ R.
In order to determine convergence or divergence at the two endpoints Ix - al = R,

and, hence, the entire interval of convergence, a test other than the ratio test must
be employed.

oo n
Ln:l
n=O
about O with the coefficients en= - -. The radius of convergence is

1
n+l
. I-
R= hm
Cn I . n+2
- = hm --=1.
n->oo Cn+ 1 n->oo n + l
Hence, the series converges absolutely for lxl < 1, i.e., -1 < x < l.
At x =l, the series becomes L --,

00 1
n+1
which diverges.
n=O
(-1r
00
At x = -l, the series becomes�--, which converges conditionally. Thus, the

Ln+l
interval of convergence is I = [-1, 1).
n=O
1
about 3 with the coefficients Cn = -, n � l. This is a power series with Co = 0. The
n
radius of convergence is
1
R = lim =1.
n
= lim +
n->oo C n+l
1� 1
n->oo n
Hence, the series converges absolutely for lx-31 < 1, i.e., -1 < x-3 < 1, or 2 < x < 4.
At x = 2, the series becomes f

n=l
(-)"
n
l , which converges conditionally.
L=l n-,1 which diverges. Thus, the interval of convergence

00
At x = 4, the series becomes
n
is I= [2, 4).
� (3 - 2x)"
L
n=2
n-1
Before the point a at which the series is centred, as well as its coefficients Cn, can be
determined, the series must be placed in the standard power series form
Loo=2 ( n-1
n
3 - 2x)"
=
oo (-2)"
Ln-1
=2 n
�
(x - )2
n
'
3 (-
from which we find a= - and c11 = - 2)",-n 2". 2. This power series has Co = c1 = 0.
2 n-1
The radius of convergence is
.
R = hm - -
. I Cn I = hm
. 211 n
---n+-
1
= -.
n-> n->oo n-1 2 l 2
oo Cn+l
Hence, the series converges absolutely for
I �1 < !
x-
2 '2
i.e.
'
- < x -� <
2
! 2
! '2 or 1<x< .2
00
1
At x = 1, the series becomes� --, which diverges.
Ln-1
n=2
1"
(- ) which converges conditionally. Thus, the
00
At x = ,2 the series becomes � --,
Ln-
n=2
1
interval of convergence is J = ( 1, ]2 .
f
n=O
(-3)"( x
2
Jn+T
4)
+ "= f J(-6+)"
n=O
n l
(x + )2 "
centred at a= -2 with the coefficients �- The radius of convergence is

n+l
= nlim 1�1 = lim foTI = 6n !_

n -+oo vn+l 5n +l
R
- oo
+ Cn+l 6
Hence, the series converges absolutely for
1 1 1 13 11
Ix+ 21 i.e., - or -
< 6, 6 < x +2 < 6, 6 < x < - 6.
L Jn+T,
00
13 1
At x = - -,the series becomes which diverges.
6 n+l n=O
At x = -- , the series becomes L -Jn+T' which converges conditionally.

00
11 (-1r
Thus,
6 n+l n=O
1 3 1 1
the interval of convergence is J = (- ,- ].
6 6

CX)
centred at 1. Since only even powers of x - l are present, the coefficient of every odd
power of x - l is 0, i.e., the coefficient of (x - 1)2n+ 1 is c2n+l = 0 for n � 0, and the
coefficient of (x - )1 2n is C2n = 2n .
In such a power series, where every second term is 0, Theorem 7. 1 cannot be applied
because the expression 1 � 1 is undefined whenever k = 2n. In this case, the radius
Ck+l
of convergence can be found directly by the ratio test. Thus, with an = 2n (x - 1)2n,
a l 2 +1 (x- 1) + =
I an+n m 2lx -
n 2n 2
lim I = nlim 1 I nli-+oo 112 = 2lx- 11 2 < 1
n -+oo -+oo 2n ( X - l) 2n
1 1
for Ix - 11 < v12· The radius of convergence is therefore R = J2' and the series
converges absolutely for
1 1 1 1
-- <x-l
v'2 J2'
<- i.e.' 1- -
v'2
< x < l+-.
v'2
At the endpoints x = l ± �' the series becomes L 1, and diverges. Thus, the
CX)
n=O
interval of convergence is J = ( 1 - �, 1 + �) .
Exercises 7 .1
In Exercises 1-6, write down the first 5 terms of the given power series.
1.
00
L4n
x
n
2.
f (x - 2r
n =O JnTI
n =O
3.
f (x+1r
4.
f (x+2r
n=O
(-3t n=l n2
00
2n oo (x - 3)3n
5. 6. �
x
�v'n+l nfo
L2 ( L n+ l ( L C (X - 1r.
00 00 00
7. Determine Cn, n 2: 0, if - 1r+

1
n
x x - 1r = n
n =O n =O n =O
L a (X - ar + /3 L b (X - ar = L e,i(
00 00
8. Determine Cn, n 2: 0, if a n n x - ar.

n =O n =O n =O
Lb L e,ix .
00 00
9. Determine Cn , n 2: 0, if 3 + 2x - 5x + 2
nx
n
= n
n=O n =O
L b (X - 3t = L C,i(
00 00
10. Determine c11 n 2: 0, if 2 + (x - 3) - (x - 3)

,
4
+ n x - 3t.
n =O n =O
11. Express LC X
as a sum of two power series, one of which contains only even
n
n
n=O
powers of x, and the other only odd.
In Exercises 12-23, find the radius and interval of convergence of the given power
senes.
2 n xn
00 00
12. 13.
n
x
Ln+l Ln 2 +1
n =O n =O
fo(x - 6)11 (-1r(x - 1)71

00 00
14. I:
n =O
n3 +1
15.
L=O
n
2n+l
(x+1)11 2n (2x - 4)
00 00
16. 17. 11
L 2n
n =O � y'n+l
= 2n(6- 3xt =
18. L n+l
19.
L
(x - 3)2n
2n
n=O n =O
x3n
L(3x - 2)2n
(X) (X)
20. 21.
� Jn2 +1 n=O
= 3n + 1 (n2 + 1) = 22n(x -3t

22. � 2n v1n (x +2r 23.
L
n=O
n,
In Exercises 24 and 25, determine the radius of convergence of the given series.
-1)!]2 (x - )2n
24.
� (2n)! n
� (n!)2 x 25. L [(n(2n)!
1
n=O
(�
n=l
26. (a) Determine c,,, n 2 0, if (� a,.xn ) n

bn X ) = � c,,x n
.
(b) Determine Cn , n 2: 0, if
7. 2 Representations of Functions by Power Series
Lx
(X)
As seen earlier, the geometric series n

converges for l x l < 1, and the sum of the
n=O
. . 1 .
senes 1s --, 1.e.,
1-x
1
(X)
�
�X
n
=-, lxl < 1.
n=O
1-x
1
This equation also shows that the function f (x) = -- can be expressed as a power
1-x
series about 0 for l x l < 1, i.e.,
1
(X)
f(x) =- = �x n ,
1-x � l x l < 1. (7. 2)
n=O
In numerous problems, it is necessary to express a given function f(x) as a power

series about a given point a,
L=O en(x - a)",

(X)
J(x) = Ix -al < R,

n
7.2. REPRESENTATIONS OF FUNCTIONS BY POWER SERIES 253
where R is the radius of convergence of the series. The solution of a problem often
appears as a power series, and it is then necessary to identify the function which is
represented by that series. The study of such issues is the content of the present
section.
Example 7.9. Consider the function
3x2
f(x) = .
l+x
In order to express f(x) as a power series about 0, employ the geometric series,
00
1
- - -..;:' tn ltl < 1,
1 l-L...., '
- n=O
with t = -x. Then

3 2 00
f(x) = : =
2
3x L(-xt = L3(-ltx
n+ 2
, lxl <
= ltl 1.
1 x
n=O n=O

1
f(x) = -.
3-x
In order to express J(x) as a power series about 0, we may write J(x) as
1 1 1 11
f(x) =
3- x = 31- �3 = 31- l'
3.
X
where l = Employing (7.2) in the variable t, we obtain
oo oo
11 l l oo 1 n
J(x) - -- - - -..;:' t n - - -..;:' (-) - -..;:-" -x
x n
.
-1 3 t - 3 L...., - 3 L...., 3 L...., 3n+l
- n=O n=O n=O
Since the geometric series in the variable t converges absolutely for Ill < 1, the above
series representation of f(x) converges absolutely for
Thus, the radius of convergence of the series is R
ltl 1,
<
= 3.
Iii=
i.e., for lxl < 3.
Example 7.11. Consider, again, the function

1
-.
f(x) = -3-x
In order to express f(x) as a power series about 2, f(x) must first be expressed as a
function of x-2, since the power series must contain powers of x-2. Thus,
1 1 1 1
f(x) = - = --- -
3-X 3 - (x - 2) - 2 1- (x-2) 1- l'
where l = x -2. Employing the geometric series in the variable t for ltl < 1, we
obtain
I> n
00 00
1
J(x) = 1-t = =
I)x - 2r,
n=O n=O
which converges absolutely for Ix -21 = ltl < 1. The radius of convergence of the
series is therefore R = 1.
1
f(x) =
8-2x
In order to express f(x) as a power series about 2, write f(x) as
1 1 1 1 1 1 1
f(x) = 41 - x-2 41- t'
8- 2x 8 - 2(x - 2) - 4 4-2(x- 2) 2
x-2
where t = - -.Then
2
2
which converges absolutely for Ix; \ = itl < 1, i.e., for Ix-21 < 2. The radius of
convergence of the series is therefore R = 2.
. 4
J(x) = 4 + x2.
In order to express f(x) as a power series about 0, write J(x) as
. 4 1 1
( +
j X) = 4 x = 1 + x4 = 1- t'
2 2
x2
where t = -- Then
4·
2
The series converges absolutely for I- : 1 =!ti< 1, i.e., for lxl < 2. Thus, the radius
of convergence of the series is R = 2.

1 1
J(x) =- +--
1 +x l - 2x
With t = -x,
with radius of convergence R 1 = 1, and, with t = 2x,
1
with radius of convergence R2 = . It follows that
2
The radius of convergence is

1
.
2'
1
which is the smaller of R 1 = 1 and R2 = -.
It can be shown, in general, that the
2
radius of convergence R of a linear combination of two power series is at least the
smaller of the radii of convergence R 1 and R2 of the two series.
Theorem 7.2. Suppose that
J(x) = Lcn(x - at, Ix - al< R,

n=O
with the radius of convergence R > 0. Then the derivative of the series can be
obtained by term-by-term differentiation of the series, i.e.,
with radius of convergence R. Since the first term (n = 0) in the series on the right
of (7.3) is 0, the series may be expressed as
00 00
n=O n=l
and, making the change of index k = n - l, we obtain
00 00
n=l k=O
which is the more standard form of a power series.

1
J(x) = (1 - x)2
1
Since f(x) is the derivative of--, we obtain, by Theorem 7.2,
l-x
00 00 00 00
1 -5!:___1_ = -5!:__ L Xn = I:nxn-l = I:nxn-1 = L(k+ l)x k ,
=
(1 - x) 2 dx l - X dx n=O n=O n=l k=O
where k = n - 1, with radius of convergence R = 1.
Theorem 7.3. Suppose that

00
J(x) = I:en (x - at, Ix - al< R,

n=O
with the radius of convergence R > 0. Then the integral of the series can be obtained
by term-by-term integration of the series, i.e.,
oo oo oo
Cn ( X - a)n+l
J
f(x)dx= f Lcn(x-atdx=
J
[en(x-at]dx= L n+l +C,
L
n=O n=O n=O
(7.4)
with radius of convergence R. Making the change of index k = n + l in the series on

the right of (7.4), the series may be expressed as
oo (x -
� Cn a )n+l �oo
Ck-1 (x - a )k C
+ C = + '
�
n=O
n+ 1 �
k=l
k
where C is an arbitrary constant of integration.

f(x) = ln(l - x), \x\ < 1.
1
Since J(x) is an antiderivative of---, we obtain, by Theorem 7.3,
l-x
00 00
l xn+l
ln(l- x) = - J -- dx = - J � x n dx = - � _ _l + C
l-x �
n=O
�n+
n=O
oo
xk
= C-L k'
k=l
where k = n + l. Since both the series and ln(l-x) have the value 0 at x = 0, C = 0
and, hence,
00
ln(l-x) =
1 k
x , L� (7.5)
k=l
with radius of convergence R = l. Since the series converges for any x with \xi < 1,
and I - x\ = \x\, replacing x by -x in (7.5) gives
oo oo oo (
ln(l + x) = L �1 (-xl = L - (�l)k x = L -l)kk-1 x k k
(7.6)
k=l k=l k=l
as the power series representation off (x) = ln(l + x) about 0 with R = l.
In all of the preceding examples, the representation of J(x) as a power series has
relied upon manipulations of the geometric series, i.e., Equation (7.2). This is not
always adequate, and more general methods are required.
Theorem 7.4. If
J(x) = L Gri(x - at,
n=O
\x - a\ < R,
with the radius of convergence R > 0, then the coefficients Cn are given by
(7.7)
Proof. By Theorem 7.2, the power series may be differentiated term-by-term. Thus,
J(x) 2 4
co+ c1(x-a)+ c2 (x - a) + c3 (x - a)3 + c4 (x - a) + c5(x - a) + · · · ,
5
J '(x) 2 3
c1 + 2c2 (x - a)+ 3c3 (x-a) + 4c4 (x-a) + 5c5 (x - a) + · · · ,
4
f"(x) 2c2 + 2 · 3c3 (x - a)+ 3 · 4c4 (x - a) 2 + 4 • 5c5 (x - a)3 + • • • ,

f" '(x) 2 · 3c3 + 2 · 3 · 4c4 (x -a) + 3 · 4 · 5c5 ( x - a) 2 + · · · ,
/4l(x) 2 · 3 · 4c4 + 2 · 3 · 4 · 5c5 ( x - a) + · · · ,
etc. It follows that
etc., and, hence,
j(0 l(a) J (a) f"(a)

, C1 = J '(a) = -' -, C2 = '
Co = J(a) =
0 ! 1, 2!
j(n)(a)
and, in general, Cn = for all n 2:: 0.
n.1
Example 7.17. Let us confirm the formula (7.7) for the geometric series
1
CX)
J(x) = - = � xn
l-x L
n=O
about a = 0, which has Cn = 1 for all n 2:: 0.

1
J(x) --
l-x
⇒ J(0) = 1 = 0!,
1
f '(x)
(1-x)2
⇒ j'(0) = 1 = 1!,
2
f"(x)
(1-x)3
⇒ j"(0) = 2 = 2!,
2-3
J" '(:r)
(1-x)4
⇒ J"'(0) = 3!,
2-3-4
j(4l(x) ⇒ JC4l(o) = 4!,
(1 - x)5
j(n)(0)
etc., and, in general, j( n)(0) = n!. Hence, Cn = = 1, as expected.
n.1
Definition 7.3. Given a function J, the series

is called the Taylor series of f about a or centred at a. In the special case where
a= 0, the Taylor series off about 0 is also called the Maclaurin series off.
By Theorem 7.4 and Definition 7.3, the power series which represents a function
about a point a is its Taylor series about a. A distinction between the two arises,
however, due to the fact that there are functions J for which the Taylor series T(x)
is not equal to f (x) on an interval Ix - al < R, R > 0, and is therefore not a power
series representation of J(x). We shall see later how to prove that T(x) = f(x) for
functions for which it is true.
It follows by Theorem 7.4 that the power series representation of a function f

about a point a is unique, i.e., if
L d (X - ar,
00 00
J(x) = :z::::>n(X - ar = n Ix - al < R, R > 0,

n=O n=O
j(n)(a)
then Cn = = dn , n 2:: 0. Thus, as observed in Example 7.17, different methods
n.1
of computing the series representation of a function must yield identical results.
It follows also by Theorem 7.4 that if
J(x) = :z::= en(x - ar

n=O
= o,
then en = 0 for all n 2:: 0. This fact is essential in the solution of differential equations
by infinite series.
Example 7.18. Let f(x) = ex and a= 0. Then
1
n!
and, hence, the Taylor series of ex about 0, i.e., its Maclaurin series, is
with the radius of convergence
R = lim = nlim ( n + l) ! lim + l) = oo.

n-+oo 1 �
Cn+l 1 -+oo nl n-+oo ( n
Thus, the series converges absolutely for all x. We shall prove later, but assume for
the present, that T(x) = f(x) = ex for all x, i.e.,
1
ex = � -xn for all x.
CX)
L..t n!
n=O
Example 7.19. Let f(x) = e 2x and a= 0. Since ex = L 1n.1 x

CX)
n
for all x, replacing
x by 2x gives
n=O
00 00
e 2
x = L 1
,
n.
(2xr = L,
2
n.
x
n
n
for all x.
n=O n=O
2n j( n)(o)
Notice that = = , in accord with Theorem 7.4.
Cn
1
n. n.1
Example 7.20. Let f(x) = ex and a = 3. Then j(n)(x) = ex for all n ?: 0 and,
hence,
f(n) (3)
Cn =
n.1
=
e3
1
n.
and T( x) =
oo e3
1
n.
(x -3 L t
n=O
L1
1
CX)
is the Taylor series of e about 3. An alternative is to employ e

x x
= x n
for all x
. n.
n=O
and to replace x by x -3 to obtain
00
1
00 e3
e =e e -
x 3 x 3
= e3
-(x-3t = � -(x-3r,
L..t n! L..t n!
�
n=O n=O
which is the power series representation of e about 3 and, hence, its Taylor series x
about 3, as above.
.
1
Example 7.21. Let f(x) tan- 1 (x), - <tan- 1 (x) < Since f' (x)
=
2. = +
7f 7f
2 1 x2
-2x
and J"(x) = the computation of f( n)(x) for every n?: 0 is tedious. A much
(l + x2)2,
better alternative is to employ the geometric series (once again!) and Theorem 7.3.
Thus,
1
1::t'\ itl< 1,
CX)
1-t n=O
CX) CX)
1 + x2 n=O n=O
1
00 00 (-l)n
J --d x= �
J L....t (-l t x2n
=c+� x2n+l , lxl<l.
1 + x2 L..t 2n + 1
n=O n=O
Since both the series and f(x) have the value 0 at x = 0, C = 0. Hence,
is the power series representation of f about 0, and the series on the right is its
Taylor series about 0. This alternative derivation of T(x) automatically shows that
T(x) = f(x) for !xi < 1.
7f 1 7f
Note, in Example 7.21, the restriction -
2 < tan- (x) < 2 on the range (i.e., set
of values) of f(x) = tan- 1 (x). Without such a restriction, the inverse tangent would
be multi-valued, and not a proper function. Had a different "branch" of the inverse
31r
< tan-1(x) < , the value of the constant C would
7f
tangent been chosen, e.g.,
2 2
have been different (C = tan-1(0) = 1r for the latter choice).
Example 7.22. Let f(x) = ln(x), x > 0, and a= 1. Then f(l) = 0,

J'(x) = ..!.,
X
f" (x) = --;, f"'(x) = �,
X X
l4 l(x) = - 2X}, l 5\x) = 2 ·X3 · 4, 5
(-1) n- 1 ( n - 1) !
and, in general, l nl(x) = n
for n 2: 1. Hence, the Taylor coefficients
off are given by
l JC n (l) (-1 -1 (n- l)!

(-1r- 1
co =0 and Cn = --- = - r---- = ---, n 2: 1,
n! n! n
and its Taylor series about 1 is
T(x) = L00 ( -1r-1 (x - 1r.

n=l
Alternatively, employing (7.5) in Example 7.16, i.e.,
ln(l - x) = L00 =-x

n
1
, n
!xi < 1,
n=l
and, replacing x by 1 - x, we obtain, for Ix - 11 = I1 - xi < 1, the power series

representation of ln(x) about 1, i.e.,
00 00
ln(x) = ln[l-(1-x)] = L 1
=--(1-xt =
n
( 1r
n
L--
(x-1r =
00 ( 1r-1
n
(x-1r, L-
n=l n=l n=l
and the series on the right is the Taylor series of ln(x) about 1. This alternative
derivation of T(x) automatically shows that T(x) = f(x) for \x - 1\ < 1.
Example 7.23. Let f(x) = sin(x) and a= 0. Then
J'(x) = cos(x), J"(x) = - sin(x), J'"(x) = - cos(x), j< l(x) = sin(x)

4
= f(x)
(4
⇒ f(0) = 0, f'(0) = 1, J"(0) = 0, !"'(0) = -1, ! ) (0) = 0,
and, in general,
Hence, the Taylor coefficients of sin(x) about Oare given by

j( 2k)(o) j( 2k+ l)(o) (-l)k
c2k = ( 2k)! =0 and c2k+1 = , k 2: 0.
(2k + l)! = (2k + l)!
The Taylor series of sin(x) about Ois therefore
00 00 00 00
( - l)k 2k+l
T(x) = L
� Cn xn = �
L C2k x
2k
+ � C2k+ l x2k+l =� x .
n =O k=O
L L (2k + 1)!
k=O k=O
1 x2k+1
The radius of convergence can be determined by the ratio test. With ak = (-2kt
( + 1)! ,
x2k+3 (2k + 1)! l xl
2
lim la k+l
I = lim I I= lim =0< 1
k--+oo ak k--+oo (2k + 3)! x2k+l k--+oo (2k + 3)(2k + 2)
for all x. Hence, R = oo, i.e., the series converges absolutely for all x. We shall prove
later, but assume for the present, that T(x) = sin(x) for all x. Thus,
(-l)k 2k+1
sin(x) L=O
oo x3 x5 x1
= x = x - 3! + 5! - 7! + ...
k
(2k + l)!
for all x.
Example 7.24. Let f(x) = cos(x) and a= 0. In order to determine the Taylor series
off about a, the procedure employed in Example 7.23 can be repeated. However,
a more efficient method is to employ Theorem 7.2 and the result of Example 7.23.
Thus,
cos(x) � sin(x) = �
dx
f
(-lt 2k+1
dx = (2k + 1)!
x = f
(-lt(2k + 1) 2k
(2k + 1)!
x
k O k=O
oo ( l)k 2
- x2 x4 x6
L
k=O
(2k)!
x = l - 2! +
k
4!
- 6! + ...
for all x.
Example 7.25. Let J(x) = sin(x) and a= .

7f In order to determine the Taylor series
2
of f about a, the procedure employed in Example 7.23 can be repeated. However,
a more efficient method is to employ the result of Example 7.24 and the fact that
sin(x) = cos (x- �). Thus, replacing x by x in -�
L ((2k)!) x2k
00 -l k
cos(x) =
k=O
sin(X) = cos ( 7f) = � --

X- -
2
(-l) k
00
� (2k)!
7r 2k
( X - -)
2
k=O
for all x.
The Binomial Series

The binomial theorem states that, for any two real numbers a and b, and any integer
n 2:: 0,
where
n n!
( )
k - k!(n - k)!
is read "n choose k," and is the number of different ways in which k objects can be
selected from a set of n objects. Since
it suffices to consider the simpler case where a= 1 and b = x, i.e.,
from which the more general formula can be reconstructed, if required.
Our goal at present is to develop a binomial theorem for
J(x) = (1 + x)'\
_)
where a is any real number, and not necessarily a nonnegative integer. This will, of
course, involve an infinite power series rather than a finite sum. The advantage of
determining a representation of f for general a is that the functions
1
J(x) = Jf+x, J(x) = Jf+x' J(x) = (1 + x) 31 2 ,
l+x
among numerous others, have the form f(x) = (1 + x)°', and having a general formula
for all a eliminates the necessity of finding the Taylor series of every such function
separately. Thus,
f(x) (l+x)'\
J'(x) a(l+x)°'- 1,
J"(x) a(a - 1)(1 + x)c,- 2,
J'"(X) a(a - l)(a - 2)(1 + x)c,- ,
3
a(a - l)(a - 2) ···(a - k + 1)(1+x)°'-k,
and, hence,
f(O) = 1, j'(O) = a, J"(O) = a(a - 1), J"'(O) = a(a - l)(a - 2),
j(k)(o) = a(a - l)(a - 2) •••(a - k+ 1).

The Taylor coefficients off about O are therefore given by
Co J(O) = 1,
J(k)(O) a(a - l)(a - 2) •••(a - k+ l)
k! k!
Define the symbol ( �) by
a _ a _ a(a - l)(a - 2) ···(a - k+ 1)

(O) - 1 and (k) - for k 2:: 1.
k!
Then the Taylor series off about O is
T(x) = �
k a a(a - l)(a - 2) ···(a - k+ 1) k
L (k) X
_
- l+ L
�
k! X
k=O k=l
a(a - 1) a(a - l)(a - 2) a(a - l)(a - 2)(a - 3)
1 +ax+ x 2+ x3+ x4
2! 3! 4!
+··· '
with radius of convergence
R = lim ,�,
. I a(a - l)(a
k-+oo Ck+l
- 2) •••(a - k + l) (k + 1)!
l)(a - k) I
hm
k-+oo k! a(a - 1)(a - 2) · · · (a - k +
lim I I
k+ l k+l
= lim =1
k-+oo a - k k-+oo k - a
since la -kl = k - a for all k 2:'. a. It can be shown that T(x) = J(x) for lxl < 1.
Thus,
(7.8)
The series on the right of (7.8) is called the binomial series, and this result is also
called the binomial theorem.
Note that, in the special case where a = n 2:'. 0 is an integer,

f(k)(x) = n(n - l)(n - 2) · · · (n - k + 1)(1 + xr-k = n!
is a constant when k = n and, hence, J(k)(x) = 0 for all k 2:'. n + l, in which case, the
binomial series reduces to the finite series
with
(n) = n(n- l)(n - 2) · · · (n - k + l) (n - k)! = n!
k k! (n-k)! k!(n-k)!'
as before.
1
Example 7.26. The Taylor series of J(x) = -- about O can be obtained by means
l+x
of the geometric series,
1 00
1 00 00
- n
= 2)-xr = I)-1rxn , lxl < 1.
1- l I:> ⇒ 1+X
n=O n=O n=O
1
Alternatively, since -- = (1 + x)- 1, we may also employ the binomial series with
l+x
a= -1. Thus, with a= -1, (�) = 1 and, fork 2:'. 1,
(�) = a(a- l)(a - 2)(a � 3) •••(a - k + l)

k
(-1)(-2)(-3) · · · (-k) (-l) kk!
= (-l).
k
kl = kl
Hence,
as above.
1
Example 7.27. Consider the function J(x) = _ = (1 - x )- 3. Its Taylor series
(l x)3
about O (Maclaurin series) is the binomial series with a = -3 and with x replaced
by -x. Thus, with a= -3, (�) = 1 and, fork � 1,
a(a - l)(a - 2)(a - 3) ···(a - k + 1)

k!
(-3)(-4)(-5)(-6) · · ·(-k - 2)
k!
(-l) · 3 · 4 · 5 · 6 · · · (k + 2)
k
k!
In order to place the last expression into a more concise form, multiply and divide by
2. Then
(-l)k · 3 · 4 · 5 · 6 · · · (k + 2) 2
k! 2
(-l)k (k + 2)!
2k!
(-l)k (k + l)(k + 2)
2
Hence,
(-x)'
1
(1 - )
X 3
t, (�)
1+ f (-l)k (k
�
l)(k + 2)
(-1/ xk
k=l
� (k + l)(k + 2) x k ,
� 2 lxl < 1,
k=O
(k + l)(k + 2) = 1 when k = 0.
since
2
Alternatively, employing the geometric series and Theorem 7.2,

00
1
Xk
l- x L
k =O
1 �-1- = � � xk = � kxk -1
(1 - ) dx l- x dx � �
::::}---
X 2
k=O k =O
� 1 = � � kxk-l =�k(k -l) xk -2 = � k(k -l) xk-2

::::} (1 - X ) 3 dx (l - x) 2 dx �k=O
� �
k =O k=2
since k(k - 1) = 0 when k = 0 and k = l. Making the change of index n = k -2, we

obtain
1 _ � (n + 2)(n + 1) X n
- I X I < l,
(1 - X )3
'
� 2
as above.
1
Example 7.28. Consider the function f (x) = y11+x = (1 + x )- 2. Its Taylor series
I
l+ x
1 1
about O (Maclaurin series) is the binomial series with a= -2" Thus, with a= -
2,
(�) = 1 and, fork :2: 1,
a(a - l)(a -2 )(a - 3) · · ·(a -k + l)

k!
(-½)(-�)(-�)(-�)... (-¥)
k!
(-1) k · 1 · 3 · 5 · 7 · · · (2k- 1)
2 k!
k
In order to place the last expression into a more concise form, multiply and divide by
2 · 4 · 6 · · · (2k). Then
( -1)k · l · 3 · 5 · 7 · · · (2k- 1) 2 · 4 · 6 · · · (2k)
2 kk! 2 · 4 · 6 · · · (2k)
(-l) (2k) !
k 1
2 k! 2 · 1 · 2 · 3 · · · k
k k
(-l)k(2k)!
22k(k!)2 ·
Hence,
1 =�(a) X k = 1 � (-lk) (2k) !X k = � (-l) (2k)! X k
k k
lxl < l,
k + � 22 (k!)2 � 22k(k!)2
k =O k =O
�
)
k=l
vl1+x
. (-l)k(2k)!
smce
22 k(k!)2 = 1 when k = 0.
Example 7.29. Consider the function J(x) = �- In order to employ the
8+x 3
2
binomial series, express f(x) as
X
f(x)=-;;-;===2=---2 =- l+-
X X ( x2 )-½
\1/8 + x 2{/1 + x 2 8
8
2
The Taylor series off about O is then the binomial series of (1 + t)°' with t = � and
a= _!,3 multiplied by �-2 Thus, with a= _!,3 (a) =

0
1 and, for k -
> l,
(�) =
a(a - l)(a - 2)(a 3)•·•(a - k + l)
�
k
(-½)(-½)(-D(-�) ... (-¥)
k!
(-l)k · 1 · 4 · 7 · 10··· (3k - 2)
Hence,
(1 + t) -½ _
- L..,
f--- (a) x k _
- l + L...,
f---. (-l) · 1 · 4· 7· 10··· (3k - 2) t ,
k
k
ltl < 1
k 3kk!
k=O k=l
� l + f---(-l) -1-4-7-10···(3k - 2)
k (x2 )k]
[
2 L..., 3kk! 8
k =l
x
+ f---
(-l)k · 1 · 4 · 7 · 10··· (3k - 2) 2 k+l
x
2 L..., 23 k+l3k k!
k=l
= ltl < 1, i.e., for lxl < 2yl2.
2
valid for 1
I �
Example 7.30. Consider the function J(x) = v1f+x = (1 +x)½. Its Taylor series
about O is the binomial series with a=�- Thus, (�)=1, (7) =a=�' and, for
k "?. 2,
a(a - l)(a - 2)(a - 3)···(a - k + 1)
k!
(½)(-½)(-�)(-�) ... (-¥)
k!
(-1) - k 1
· 1 · 3 5. ··(2k - 3)
·
2kk!
noting that there are k factors in the numerator, only k - 1 of which have minus
signs. Multiplying and dividing by 2 · 4 · 6 · · · (2k -2), we obtain
(-l)k-l · 1 · 3 · 5 · · · (2k - 3) 2 · 4 · 6 · · · (2k-2)

' 2k k! 2 · 4 · 6 · · · (2k - 2)
k 1 1
(-l) - (2k-2)!
2k k! 2k-l · 1 · 2 · 3 · · · (k. - 1)
(-l)k-1(2k-2)!
k>2- )
22 k-lk!(k- 1)! )
which is also correct for k = 1. Hence,
=
L
(a)
00
k 1 +
00
L
(-1t-1(2k-2)! k
lxl < 1.
v1f+x k x = 22k-lk!(k- 1)! x '
k=O k=l
Taylor Polynomials and Approximations

Definition 7.4. Consider the Taylor series of a function J about a point a, i.e.,
oo
f( n) (a )
T(x) =
L
n=O
n.I
(x - ar.
T he nth partial sum Tn (x) of the Taylor series is
L
n f(k) ( )
a (x - ) k
a
k=O k!
f"(a) ' f "(a)
J(a) + J'(a)(x - a)+ - -(x - a) 2 + (x - a) 3
21 �
f( n l(a)
+···+ (x -a)'\
n.1
and is a polynomial of degree n, called the nth -degree Taylor polynomial off about
a. If the Taylor series of a function converges to the function on an interval, i.e.,
T(x) = f(x) for Ix - al <Rand R > 0, then J(x) can be approximated by its Taylor
polynomial Tn (x), and the error is
oo f(k) ( a )
11,,-,(x) = f(x) - Tn (x) = L
k=n+l
kl
(x - a)k ,
called the remainder.

The product of two Taylor series, or their quotient, about the same point a, is
again a Taylor series about a, but the formula for the coefficients of the product is
complicated ( see Exercise 26 in Section 7.1), and none exists for the quotient. In such
cases, their Taylor polynomials may still furnish useful information.
Example 7.31. Consider the function f(x) = ex cos(x). In order to determine its
4th -degree Taylor polynomial about 0, the Taylor polynomials of suitably high degree
of the functions ex and cos(x) must be multiplied. Since
12 l3 14
eX = 1 + X + -X + -X + -X + · · ·
2 6 24
and
1 1
= 1 - -x2 + -x + · · · '
4
cos(x)
2 24
and the product is required only up to the fourth power of x, the terms xk for k � 5
are neglected. Thus,
ex cos(x) = (1 + x + !x
2
+ !x3 + 2-x4)
2 6 24
13 1 4
1+X - -X - -X + ··· .
3 6
Hence, T4 (x) = 1 + x - !x3 - !x4 for f(x) = ex cos(x) about 0.

3 6
Example 7.32. Consider the function J(x) = tan(x). In order to determine its
5th -degree Taylor polynomial about 0, the Taylor polynomial of suitably high degree
of the function sin( x) must be divided by that of cos( x). Thus,
1 3 1 5
sin(x) =x- x +
120
x +···
6
and
1 1
= 1 - -x + -x + · · · '
2 4
cos(x)
2 24
and, dividing the former by the latter and neglecting the terms xk for k � 6, we
obtain
sin(x) x - lx
6
3
+ _1
120 x
5
cos(x) 1- lx
2
2
+ ....!...
24x
4
1 3 2
X + -X +-x5 + · · ·.
3 15
Hence, T5 (x) = x + !x3 + �x5 for tan(x) about 0.

3 15
Example 7.33. Consider the function J(x) = (x2 -1)ln(x). Its Taylor series about
1 can be determined by multiplying the Taylor series of ln(x) about 1 by the Taylor
series of x2 - 1 about 1. Since x2 - 1 is a polynomial of degree 2, its Taylor series is
also a polynomial of degree 2, and the multiplication can be performed without great
difficulty.
By Example 7.22,
00
ln(x) = L ( -1nr-1(x - 1r, Ix -11 < 1.
n =l
The Taylor series of g(x) = x2 - 1 about 1 can be obtained by Equation (7.7). T hus,
g'(x) = 2x, g"(x) = 2, and g(n) (x) = 0 for n 2:: 3 ::::}
g(n) (l)
g, (l) = 2, c2
g"(l)
co = g(l) = 0, C1 = = -- = 1, and Cn = = 0 for n 2:: 3,
2 n.1
and the Taylor series of x2 - 1 about 1 is T(x) = 2(x - 1)+ (x - 1)2 . As can be
readily verified,
T(x) = 2(x - 1)+(x - 1)2 = x2 - 1 for all x.
Then the product is
00 (
(x - 1)ln(x)
2
= [2(x - 1) + (x - 1) ]
2
L - 1nr-1(x -1r
n=l
Making the change of index n - n+l in the first series on the right, we obtain
(x2 - l)ln(x) = 2(x -1)2 +L� 2(-lt (x - 1r+ 2 + � (-1r-1 (x - 1r+ 2

n+l L n
n =l n =l
� [2(-1)" + (-1r-1] (x - 1r+2

2(x -1)2 + L
n =l
n+l n
� [2(-l) - (-l) ] (x - 1r+2

n n
2(x -1)2 + L
n =l
n+l n
2(.r: - 1)2 + f - )n (
n ( n + 1)
l)
[( l n - ] (x - 1)
"+2
f
n =l
(-l)n (n - l) +2
2(x - 1)2 + [ ] (x- 1r , Ix - 11 < 1.
n=2
n(n + 1)
The last series on the right begins at n = 2 because n - l = 0 when n = l. The

series converges absolutely for Ix - 11 < 1 since the Taylor series of ln(x) converges
absolutely for Ix - 11 < 1. Truncation of the series by neglecting the terms beyond
f
(x - 1r yields
2(x - 1)2 + ( )k (k - 1)
[ -l ] (x - 1t+2
k(k + 1)
k=2
i.e., the Taylor polynomial Tn (x) of degree n.
Example 7.34. Consider the function f (x) = e-x 2

. Assuming that the Taylor series
of ex converges to ex for all x, we obtain
= n = ( x2 )n = (-l)n x 2n
ex = � :._ e-x = � -
2
=}
= � --
Ln =O
n! L
n =O
n! Ln =O
n!
Since this is an alternating series for x =/- 0, if f(x) is approximated by its nth-degree
Taylor polynomial Tn (x), the error, in absolute value, will be less than the absolute
2n
= -1 ,
x
value of the first neglected term. Thus, with bn (x )
n.
x 2n+2
IRn(x)I = lf(x) -Tn (x)I < bn+1(x) = ·
(n + l)!
If, for instance, f (1) = e-¼ is approximated by T2 (1), then
and
2
1 (-l) k (-!-2) 2k 1
2
l 1
4
25
=
L = 1 - (2) + 2 (2) = = 0 . 78125.
1
e-4 � T2 ( )
2 k=O
kl 32
The value of e-¼ obtained by calculator is approximately 0.7788. The difference

between the exact value and the approximate value is -0.00245, and
I - 0.002451 = 0.00245 < 0.0026,

as guaranteed by the general theory. The difference between the exact value and the
approximate value is negative because the first neglected term is negative and, hence,
the approximate value is greater than the exact value.
Example 7.35. Consider the definite integral
r½ e-x dx.
2
lo
Since f(x) = e-x does not have an elementary function as an antiderivative, we must
2
appeal to infinite series. Employing the Taylor series of e-x in Example 7.34 and
2
Theorem 7.3, we obtain
r½ e-x2 dx r½ 00
(-1rx2n 00
(-1rx2n+1 ½ 00
(-1r
dx = 1 = 2n 1
lo = lo � n! � (2n+ l)n! 0 � 2 + (2n+ l)n!·
I)-1 tb
00
The sum s of the alternating series n on the right is the exact value of the
n=O
integral. If the series is approximated by its nth partial sum S n , then
1
IRnl = Is - sn l < bn+l = 2n+3
2 (2n+ 3)(n+ l)! •
If, for instance, an approximation of the integral is required with error less than 0.01
in absolute value, then n must be sufficiently large in order that IRnl < 0.01. With
n = l,
1 1
IR1I < b2 = 5 = - � 0.003 < 0.01,
2 ·5·2 320
1
and
1 1 11
e -x2 dx � s1 = - - --
2
= - � 0.4583 '
0 2 2 .3
3 24
with the required accuracy.
Example 7.36. Consider the definite integral
/ ¼ --;==
lo vl
l
=
+ x6
dx.
By Example 7.28 with x replaced by x6, we obtain
1 00
(-l) k (2k)! 6k
Vl + x6 = � 22k (k!)2 x ' lxl < 1.
Hence, by Theorem 7.3,
1o
-- 1
¼ Jl + -
x6
dx 1 ¼ � (-l) (2k)! x6kdx
0 �
k
22k (k!)2
=
�
�
(-l)k (2k)!x6k+1 ¼
I
22k (k!)2 (6k + 1) 0
00 (-l)k (2k)!
� 22k (k!) 2(6k + 1)4 6k+l
00 ( -1)k ( 2k)!
� (k!)2(6k + 1 )4 7k+l
00
is an alternating series, and

(2k + 2)!
IRkl < bk+l = .
[(k + 1)!]2(6k + 7)4 7 k+8
2 1 1
If k = 0, then IRol < b1 = 8 = 15 = 229,3 � 0.00000 4359. Hence,
.
7 4 .
7 2 76
1 ¼ �
1
o V 1 + x6
dx � s0 = -1 = 0.25
4
with error less than 0.00000 5 in absolute value.
Taylor polynomials can be employed in the evaluation of limits, even in cases

where L'Hopital's rule is inapplicable or too tedious .
Example 7.37. Consider the limit

. (2x2 +3)sin 2 (x)
1 Im--2----.
x---->O x cos (x)
Employing the Taylor series of sin (x) and cos(x) about O and retaining only the first
two terms, we obtain
. 2
(2x- +3)sin 2
(x) 2
. (2x +3)(x -l 6
x3 +-··)2
1 11n - -- -- hm
x---->O x)
x2 cos ( x---->O x 2 (1 - ½x2 + • • •)
2 + ... 2
. (2x2 +3)x2 (1 -l 6x
)
hm
x---->O x (1 - ½x + • • •)
2 2
2 +
. (2x 3)(1 - l 6
x2 + ... )2
hm
x---->O ( 1 - ½ x 2 + ... )
since x k ----, 0 for all k � 1 as x ----, 0.
Example 7.38. Consider the limit
. (ex - x - 1) cos(x)
--
1 1m--- --
x--->O sin2 (x)
x2 x3
Since e x = 1 + x+ - + - + · · •
2 6
. (ex - x - 1) cos(x)
(-
2
x x +... ) ( 1 - -
+- 3
1 x2 +... )
1Im----=---- -- lim 2 6 2
x--->O si1i(x) x--->O (x _ ½x3 +... )2
x 2 ( l2 + £126 +...) ( 1 - l2 x 2 +...)
lim
x--->O x2 (1 - ½x 2 + • • • )2
(l + £12 + .. ·. )(1 - l2 x2 + · · ·)
lim 2 6
x--->O (1 - ½x2 + ... ) 2
1
2
Example 7.39. Consider the one-sided limit
. e x sin(x)
1 1m -;::::===----,
x--->O- v'2x2 + x3 cos2 (x)
i.e., x----, 0 with x < 0. Then
e x sin(x) . (1 + x + • • • )(x - l6 x3 +...)

lim --;::::===-- hm
x--->O- J2x2 + x3 cos 2 (x) x--->O- jxjJ2 + x(1 - ½x 2 +... ) 2
. X (1 + X + · · · )(1 - l x2 +...)
hm - 6
x--->O- lxl J2 + x(1 _ ½x2 +... )2
· · ) ( 1 - ½x 2 + · · · )
(1 + X + ·
lim
x--->O- J2 + x(1 - ½x2 +... )2
1
J2
since vlx2 = lxl = -x for x < 0.
Theorem 7.5. Suppose that the function f has the Taylor series
oo j(n)( )
T(x) = � a (x -
6 n! at
n=O
about the point a, and let
L
n j(k)( a)
Tn (x) = (x - a) k
k!
k=O
be its n th -degree Taylor polynomial. T hen the remainder Rn(x) = f(x) - Tn(x) is
given by
f (n+l)(z )
Rn (x) = (x - a)n+1 (7.9)
(n + 1)!
where z is a number between a and x. Equation (7.9) is known as Taylor's remainder
formula.
Example 7.40. Consider the function J(x) = ex . By Example 7.18, its Taylor series
about 0 is
and converges absolutely for all x. If ex is approximated by its n th-degree Taylor

polynomial
Tn(x) =
n 1 k
k!
x , L
k=O
then, employing Theorem 7.5 and the fact that j( n+l)(z) = ez for all n � 0, we obtain
ez
Rn (x) = ---!xn+l
(n + 1)
and, hence,
2
e
IRn(x) I = (n lxl n+l ·
+ 1)!
Suppose that we wish to approximate ve = e 112 with error less than 0.001 in absolute
1 1 1
value. Since x = and z is between a= 0 and x = , 0 :S z :S < 1. Then
2 2 2
1) n+
( 21) I
l
I
z
e e
Rn = --- ( < 0.001
( n + l) ! 2 < ( n + l) ! 2n+ 1 -
if (n + 1)! 2 n+l
� lO00e � 2718.28, which is satisfied for n � 4. Taking n = 4, we
obtain
4
L
1 1 1 1 1
1+ +8+ +
-le� k=O k 2! k =
2 48 384
� 1.6484·
The value of ft obtained by calculator is approximately 1.6487, with a difference of

0.0003 < 0.001, as required.
Theorem 7.6. Suppose that the function f has the Taylor series
oo (nl
J (a)
T(x) = "°""'
L n!
--(x - at
n=O
about the point a, with radius of convergence R > 0. Let
L J k!( a) (x - al
n (k)
Tn (x) =
k=O
be its n th -degree Taylor polynomial, and let Rn (x) = J(x) -Tn (x) be the remainder.
Then T(x) = J(x) for Ix - al < R if and only if
lim Rn(x) = 0, Ix - al < R.

n-+oo
Proof. T(x) = f(x) if and only if J(x) - T(x) = 0 if and only if
Rn (x) = J(x) - Tn (x) - 0 as n - oo
smce lim Tn (x) = T(x).

n-+oo
Example 7.41. In order to prove that the Taylor series of f(x) ex about 0
converges to ex for all x, it must be shown that
By Theorem 7.6, it suffices to show that Rn(x) = f(x) - Tn (x) - 0 as n - oo. By

Theorem 7.5,
f( n+l) (z) n+l -,------�
ez
Rn(x) = x = xn+l '
(n+l)! (n+l)!
where z is a number between 0 and x.
Since
x
n+ l - l I I n+ 1
� e x �
� (n +l)! I X I - e � (n +l)! X
ex
converges (absolutely) by the ratio test, lim lx/ n+l = 0 by Theorem 6.9.
(n + l ) !
n-+oo
Hence, lim /Rn(x)I = 0 by Theorem 6.4 and, by Theorem 6.2, lim Rn(x) = 0.
n-.oo n__.oo
Similarly, if x< 0 , then x :::; z :::; 0 ⇒ ex :::; e z :::; 1 ⇒

ez 1
- I.Rn(x)I = --lxl
0<
(n + 1)!
n+l <
--lxl n+l --t 0
- (n + 1)!
00 1
⇒ lim Rn(x) = 0. Thus, ex ="' -xn for all x.
n---+oo L
n=O
n!
More generally, for any real number a,
i.e., the Taylor series of ex about any real number a converges to ex for all x.
Example 7.42. In order t o prove that the Taylor series of f (x) = sin(x) about 0
converges t o sin(x) for all x, it must be shown that
OO
T(x) = I:---
(-ll = sin(x) = f(x).
l x2k+I
(2k
k=O
+ )!
By Theorem 7.6, it suffices t o show that Rn(x) = f(x) - Tn (x) --t Oas n --too. By
Theorem 7.5,
f (n+l)( z ) n+l
x
Rn ( ) = x
(n + 1)!
where z is a number between O and x.
Since j(n+I )(z) = ±sin(z) or ±cos(z), lf (n+ l)(z)I:::; 1 for all real z and n 2: 0. Hence,
l (n+l) (z)I 1
- IRn(x)I = f(n + 1)! lxln+l <
0<
- (n + 1)!
lxln+ l --t 0
(-l)k 2k+I
oo
⇒ lim Rn(x) = 0. Thus, sin(x) ="' (
x for all x.
n---+oo L
k=O
2 k + 1 ) .1
( l)k 2k
L
oo
It follows by Example 7.24 that cos(x) = ; x for all x.
( k)!
k=O
Definition 7.5. Let f be a funct ion and a a real number. The funct ion f is analytic
at a if f has a Taylor series T(x) about a, with radius of convergence R > 0, and the
Taylor series converges t o J(x), i.e., T(x) = f(x), for all x with Ix -al< R. If f does
not have a Taylor series about a, or if the Taylor series does not converge to f(x) for
all x with Ix - al < R, then the point a is called a singularity or singular point of the
function f.
Example 7.43. The function J(x) = ex is analytic at a for every real number a, as
shown by Example 7.41.
Example 7.44. The functions sin(x) and cos(x) are analytic at 0, as shown by
Example 7.42.
Example 7.45. The function
is analytic at 0. However, it is not analytic at 1 because J(l) is undefined and, hence,

J does not have a Taylor series about 1. Thus, 1 is a singularity off.
For any a =I= 1, J is analytic at a since, for Ix - al < 11 - al,
__
l = 1 = _1_ 1 =� 1
-a
(x - at.
1 -x 1 - (x - a) - a 1 - a 1 - xl-a �
n=O
( 1 - a)n+l
Example 7.46. The function J(x) = ,Ix = x 1 ! 2 is not analytic at 0 because

1
J'(x) = ;;:;: => f'(O) is undefined and, hence, f does not have a Taylor series about
2yx
0. Thus, 0 is a singular point of f.
Theorem 7. 7. Suppose that J and g are analytic at a. Then:
1. cif ( x) + c2g (x) is analytic at a for any constants c1 and c2.
2. J(x)g(x) is analytic at a.
3. ��:? is analytic at a provided that g(a) =I= 0.
The concept of analytic function is required in the solution of differential equations

with variable coefficients by infinite series, as will be seen in Chapter 9.
Exercises 7. 2
In Exercises l 10, express the given function J cIB a power series about 0 and determine
its radius of convergence.
= 1
l. f (X)
1 + 2x
2. f(x) = - 6-
2-x
1
3.f(x)=
3+x
4. f(x) = �
4-x
1
5. f(X) = +
6 2x
. 1
6. j(x) = - 2 -
l-x
1
7. J(x) = --3
2+x
1 1
8· f(x) = - -
1 x l - 3x
1 1
9. f(x) = - - +
2 +x 3 - 2X
x2 +1
10. f(x) = -
X 2-
- 1
In Exercises 11 28, find the Taylor series of the given function f about the given
point a and determine its radius of convergence.
11. J(x) = e-x , a= 0

12. f(x) = e-3x, a= 0
13. J(x)=ex , a=2

14. J(x) = ex, a= -2
15. f(x) = e- 3x , a= 0
2
16. J(x) = e- 3x, a= l
17. J(x) = sin(2x), a= 0
18. f(x) = cos(3x), a= 0
19. J(x) = sin(x), a= 1r

20. J(x) = cos(x), a= 1r
22. f(x) = ln(x-1), a= 2

23. J(x) = ln(x), a= 2
24. J(x) = tan- 1 (x - 2), -� < tan-1(x) < �' a= 2
4
25. J(x) = ft+x' a= 0
x2
26. f(x) = ' a= 0
ijf+x
27. f(x)=xJl+x2 , a=0

X
28. f(x) = , a=0
(l-x3)312
29. Find the 4th-degree Taylor polynomial of J(x) = ex sin(x) about 0.

. cos(x)
30. Fmd the 4 th -degree Taylor polynomial of f(x) = . about 0.
1 + sm( x )
31. Find the 4th -degree Taylor polynomial of f(x) = (x2 + 1) cos(x) about 0.
. x ln(x)
32. Find the 4th -degree Taylor polynomial of j(x) = -- about 1.
2-x
33. Approximate sin ( 1) with error less than 0.001 in absolute value, and confirm
your answer with a calculator.
34. Approximate 11 cos( Jx) dx with error less than 0.001 in absolute value.
35. Approximate 12
I
Jl +x8
1
dx with error less than 0.001 in absolute value.
. x( e 3x - 1)2 cos2 (x)

36. Evaluate hm . 3( )
x--+0 Sln X
. (x2 - l)(e-x + x - 1) cos(x)

37. Eva1uate 1 1m . .
x--+O X Sln (3X)
. (x+2)3 ln(l-x)
38. Evaluate hm -'--------'--�.
x--+O sin(2x)cos(3x)
xJ4x4 - x 5
39. Evaluate lim - 3- - - -) .
-
x--+O- tan (2x) cos2 (3x2
40. Determine how large n must be taken in order to approximate ffe = e 1 13 by

the Taylor polynomial of x
about O at x = with error less than 0.0001 in
e
3
absolute value, and give the approximate value. Confirm your answer with a
calculator.
41. Determine how large n must be taken in order to approximate ln(0.9) by the
Taylor polynomial of ln(l - x) about O at x = O.l with error less than 0.001 in
absolute value, and give the approximate value. Confirm your answer with a
calculator.
42. Show that f (x) = sin(x) is analytic at every real number a.
43. Show that f (x) = cos(x) is analytic at every real number a.
44. Employ power series to prove Euler's identity: e

ix
= cos(x) + isin(x).
45. (a) Show that, for any two real numbers a and b, ab::::; lallbl.
(b) Show that, for any two real numbers a and b, la+ bl ::::; lal + lbl. This
inequality is known as the triangle inequality.
(c) Show that, if
00
n=O n=O
have radii of convergence R 1 > 0 and R 2 > 0, respectively, then the radius
of convergence R of
00
f(x) + g(x) = I)an + bn)Xn

n=O
is at least the lesser of R 1 and R2.
Chapter 8
Fourier Series
The topic of Chapter 7 was the repre entation of function by power erie . Variou
s s s s s
problem require the repre entation of function by eries which contain trigonometric
s s s s
functions. The content of the pre ent chapter i the tudy of such serie .
s s s s
8.1 Fourier Series of Periodic Functions

Definition 8.1. A function f i periodic if there exi t T > 0 uch that
s s s s
f (x + T) = f (x) for all real x.
Such a function f i said to have periodT (or to be T-periodic). The least (positive)
s
period off i called it fundamental period.

s s
Example 8.1. The function J(x) = sin(x) i periodic with period T = 2mr for any
s
integer n 2:'. 1 ince in(x + 2mr) = in(x) for all x. The fundamental period of in(x)
s s s s
i 21r.
s
Let f be periodic with period T = 2L, L > 0, and suppo e that f is represented
s
by a trigonometric series, i.e.,
ao � n1rx . n1rx
= [a n COS ( L ) +b (8.1)
J(x)
2+L
( )]
n=l
11 sm
L
where a n , n 2:'. 0, and bn , n 2:'. 1, are constants. The fir t term in the erie , which
s s s
correspond to n = 0, i denoted by �o in tead of a0 merely for convenience, in order

s s s
that a 0 and a n for n 2:'. 1 can be defined by a single formula in tead of two eparate s s
one .
s
283
284 CHAPTER 8. FOURIER SERIES
Given that Equation (8.1) holds, the coefficients an , n ;::=: 0, and bn, n ;::=: 1, will be
determined formally, i.e., without justification of the procedures employed in their
derivation. Subsequently, the conditions under which and the values of x for which
such a series converges to f ( x) will be given.
. .
In order to determine the coefficients an , n ;::=: 0, let m ;::=: 0 be any integer, multiply
m1rx
Equat10n (8.1) by cos ( L) , and mtegrate from -L to L to obtam
1L f ( x) cos (m1rx
-L £) dx = 1 a ° cos (m1rx
L
L) dx + -L
1L f--, [
-L
n1rx m1rx n1rx
2
m1rx
L an cos (L) cos ( L) dx + bn sin (L) cos ( L dx.
)]
n=l
Reverse the order of integration and summation to obtain
1L J(x)cos (m1rx) dx = a 1L cos (m1rx) dx +

-L
0
-L
L 2 L
f--, [a 1 cos (n1rx
L
n L) dx + bn 1 sin (n1rx
L) cos (m1rx
-L
L
L) dx] .
L) cos (m1rx -L
�
If m = 0, then
1
-L
L
m1rx
cos ( -- ) d.T
L
=
1L
-L
1 dx = 2L,
and
L
a
Hence, j J(x) dx = o2L = a0 L, from which it follows that
-L 2
ao =
l 1L f(x) dx. (8.2)
L -L
If m ;::=: 1, then
1 L
-L
m1rx
cos ( L) dx =
L
m1r
m1rx
sin ( L)
L
= 0.
-L
8.1. FOURIER SERIES OF PERIODIC FUNCTIONS 285
Employing the trigonometric identities
cos(a + b) = cos(a) cos(b) - sin(a) sin(b) ⇒

{ cos(a - b) = cos(a) cos(b) + sin(a) sin(b) }
1
cos(a) cos(b) =
2 [cos(a - b) + cos(a + b)]
n1rx m1rx
with a= Land b = L' we obtain
1 £
-L
n1rx m1rx
cos ( L ) cos (L ) dx
12 1_£ { cos [(n - Lm)1rx] + cos [(n + Lm)1rx]} dx

£
0, n i- m
{ L, n=m }
since, for any integer k i- 0,
Employing the trigonometric identities
sin(a + b) = sin(a) cos(b) + cos(a) sin(b)

{ sin(a - b) = sin(a) cos(b) - cos(a) sin(b) } ⇒
sin(a) cos(b) = 1 [sin(a - b) + sin(a + b)]

n1rx
with a= L and b = L'
m1rx
we obtam
.
1 £
-£
n1rx m1rx
sin ( L) cos (L ) dx
11: { sin [(n -;i) ]

1rx
+ sin
+ )
[(n ;i 1rx]} dx
0
siuce, for any integer k i- 0,
1L-£ sin
(brx)
L L k1rx
dx = - kn cos (L)
L
= 0,
-L
£
brx £
and, if k = 0, then 1- sin ( L ) dx = 1- O d x = 0. Thus, for any m � 1,
L L
1: m x
f(x ) cos ( ; ) dx = amL, i.e., am = ± 1: m x
f(x ) cos ( ; ) dx. (8.3)
Combining Equations (8.2) and (8.3), we obtain
am = ± 1: m x
f(x ) cos ( ; ) dx , m � 0. (8.4)
m x
Similarly (see Exercise 3), multiplication of Equation (8.1) by sin ( ; ) , m � 1,
and integration from - L to L yields
£
bm = L11-L f(x ) sin (mn
--y-x ) dx, m � I. (8.5)
Definition 8.2. Suppose that f is periodic with period T = 2L. The trigonometric
senes
ao � nnx . nnx
+ [anCOS ( L ) +bnSIIl ( )]
2 L
n=l
L
with an, n 2". 0, and bn, n 2". 1, defined by Equations (8.4) and (8.5), respectively, is
called the Fourier series off. The coefficients an, n 2". 0, are called the Fourier cosine
coefficients, and bn, n � 1, the Fourier sine coefficients off. The integral formulas
(8.4) and (8.5) which define the coefficients an and bn are called the Euler-Fourier
formulas.
Since f has period 2£, the above integrals can be evaluated on any interval of
length 2L, and one should select the interval on which an explicit formula for f(x ) is
available.
Example 8.2. Consider the function defined by
f(x ) = x 2 + x, -1 ::; x < 1, f(x + 2) = f(x) for all x.
T
Then f is periodic with period T = 2 and, with L =
2 = 1, its Fourier series takes
the form
Note that the formula f ( x) = x 2 +x is valid only for x in the interval [-1, 1). For
y
2
3
X
Figure 8.1: The graph of J in Example 8.2.
any x outside the interval [-1, 1), the 2-periodicity off must be employed in order
to determine the value f(x). The graph of f is depicted by Figure 8.1, where three
periods are displayed. The curve from -1 to 1 repeats over the entire real line.
By Equations (8.4) and (8.5), the coefficients an , n 2: 0, and bn , n 2: 1, are given by
3 2 2
1>x2 +x)dx = (� +�) [
3'
1� (x + x) cos(mrx)dx
1
11
2
-(x 2 + x) sin(mrx) -- (2x + 1) sin(mrx)dx

1
1 1 1
11
n1r -1 n1r -1
2(2x + 1) cos(n1rx) 1 2 cos(n1rx)dx

- 22
1 1
1
2 7r -1 n 7r -1
4(-lr
n
n2 1r 2 '
n 2: 1,
1� (x2 + x) sin(n1rx)dx
--(x 2 (2x + 1) cos(n1rx)dx

1 1
1
+ x) cos(n1rx) 1 +- 1
1
n1r -1 n1r -1
2(- )n
1 1 11
-l + 2(2x + 1) sin(n1rx) - 2 2 sin(n1rx)dx
n 1r _1 n 1r _ 1
11
2 2
- --
n1r
2(-lr-l 2
11_1
--- + 33 cos(n1rx)
2(-1r-1
n1r n 1r
n 2: 1.
Thus, the Fourier series of f is
1 00
4(-lt 2(-1r- 1 ]
-3 + L
n=l
[
n2 1r 2
cos(mrx) + --sin(mrx)
n1r
- .
The following two definitions are required in order to state a convergence theorem
for Fourier series.
Definition 8.3. A function f is piecewise continuous on an interval [a, ,BJ if [a, ,BJ
can be partitioned into a finite number of subintervals such that f is continuous on
each open subinterval and approaches a finite limit as x approaches an endpoint of
any of the subintervals. A function is piecewise continuous if it is piecewise continuous
on every finite interval.
Thus, a piecewise continuous function may have finite jump discontinuities on

[a, ,BJ, but no other types of discontinuity. The graph of a typical piecewise continuous
function on [a, ,BJ is displayed in Figure 8.2.
a X
Figure 8.2: The graph of a piecewise continuous function on [a, ,BJ.
Definition 8.4. The left-hand limit f (x-) of a function f at x is defined by
lim J(y).
J(x-) = y.-x-
The right-hand limit f ( x+) of a function f at x is defined by
J(x+) = y.-x+
lim J(y).
The average value of f at x is defined by

1
2[J(x+) + f(x- )].
If J is continuous at x, then f (x-) = f (x+) = f (x ), and the average value off

at x is f(x). The one-sided limits and the average value off at x are useful notions
if J has a jump discontinuity at x.
Example 8.3. The function f in Example 8.2 is piecewise continuous. It has jump
discontinuities at every odd integer x =2n + 1, where n is an integer. At x = l,
1
f(l-)=2, f(l+)=O, and [f(l+)+J(l-)]=1
2
is the average value.
Theorem 8.1. Suppose that J is periodic with period T = 2L and that f and f'
are piecewise continuous. Then, at any point x, the Fourier series of f converges to
tu(x+) + J(x-)],
i.e., the average value off at x. If f is continuous at x, then the Fourier series off
at x converges to f ( x). This theorem is called the Fourier convergence theorem.
In certain books, a discontinuous function is redefined at a point x of discontinuity

1
by J(x) = f(x [ +)+ f(x-)], in which case, the Fourier series of J converges to J(x)
2
for all x.
Example 8.4. Let J(x) = 1 - x for O � x < l and f(x + 1) = f(x) for all x. The
graph of f is displayed in F igure 8.3.
-2 -I 0 2 3 X
F igure 8.3: The graph off in Example 8.4.
The function J has finite jump discontinuities at the integers, and no other types of
discontinuity. Hence, f is piecewise continuous. Its derivative f' ( x) = -1 everywhere
except at the integers, where it is undefined, but lim J'(x) = -1 is finite for every
X--->n±
integer n. Hence, f' is piecewise continuous, and the Fourier convergence theorem
(Theorem 8.1) applies.
T 1
Since J has period T = 1, L =
2 2, and the Fourier series off takes the form
=
ao
2+�
L...,[an cos(2mrx) + bn sm(2mrx)],
n=l
and, since an explicit formula for f(x) is available on the interval [0, 1), the integrals
in Equations (8.4) and (8.5) must be evaluated from 0 to 1. Thus,
fo\1
1fo
ao 2 - x) dx = (2x - x2 ) I:= 1,
an 2 (1 -x) cos(2mrx) dx
1
-(1 - x) sin(2mrx)
n1r
1
o
1 +-
1 11
n1r o
sin(2mrx) dx
1
1
fo1
-- 2 2 cos(2n1rx)
2n 7r 1o
0, n 2: 1,
bn 2 (1 - x) sin(2n1rx) dx
1 1
--(1 -x) cos(2n1rx) - -
n1r o n1r o
cos(2n1rx) dx 1
1
11
1
n 2: 1.
n1r
)
Hence, for all x except the integers,
= - + I:- sin(2n1rx).
00
1 1
f(x)
2 n=l
n1r
1
At x = m, where m is an integer, the series converges to the average value
2, as is
evident from the series since sin(2n1rm) = 0 for every n 2: 1.
1-
The number to which the series converges can be determined for any real number x.
For example, at x = �' the series converges to J (�) = At x = 731, employing
the 1-periodici ty of J, the series converges to
1 1 1
2 [!(731-) + f(731+)] = 2 [!(0-) + f(O+)] = 2,
and, at x=731.2, the series converges to f (731.2) = f (0.2) =1 - 0.2=0.8.
Example 8.5. Let f(x)=

{0<X<1
X,
l, 1 � x < 2
} and f(x +2) = f(x) for all x. The
graph of f is displayed in Figure 8.4.
-6 -5 -4 -3 -2 -] 0 2 3 4 5 6 X
Figure 8.4: The graph off in Example 8.5.
The function f is piecewise continuous, with discontinuities at the even integers 2m,
and f' is piecewise continuous, with discontinuities at all of the integers. The average
1
value of f at an even integer is . Thus, the Fourier series of f converges to f(x) at
2
every x except at x= 2m, where m is an integer.
Since f has period T = 2, L = 1 and the Fourier series of f takes the form
with
ao 122 f(x)dx= 11 xdx+ f1 2

1dx=1+1=�,
an 1 J(x) cos(mrx)dx= 1 x cos(mrx)dx + j\os(mrx) dx
1 11 1
1 2
� sin(mrx) - - sin(mrx)dx +- sin(mrx)

1
n1r o n1r o n1r 11
22 cos(n1rx) = 22[(-lt - 1], n 2:: 1,

1 11 1
n 1r n 1r
1 11
O
2
bn f(x)sin(n1rx)dx= xsin(n1rx)dx+ j2 sin(n1rx)dx
1 11 1
1 2
-� cos(n1rx) + - cos(n1rx)dx - - cos(mrx)

1
n1r o n1r o n1r 11
(-1r 1 . 1 (-1r 1
1
1
--- + --sm(n 1rx) - -+ -- = --, n 2:: 1.
n1r n 1r
2 2
0 n1r n1r n1r
Hence, for all x except the even integers,
f(x)
3
= -4 + L00 { r(-1nr2 2- 11 cos(mrx) - -
n=l
1r
1
n1r
sin(mrx) .
}
(8.6)
1
At every even integer x = 2m, the series converges to
2.
The number to which the series converges can be determined for any real number x.
For example, at x = l, the series converges to f(l) = 1. At x = 273, employing the
2-periodicity off, the series converges to J(273) = J(l) = 1. At x = 28.3, the series
converges to f(28.3) = f(0.3) = 0.3.
1
An interesting result emerges at x = 0. Since the series converges to at x = 0,
2
setting x = 0 in (8.6) gives
and, since (-1 t - 1 = 0 for n = 2k even and (-1 t - 1 = -2 for n = 2k + l odd, we

obtain
00 -2 1 00 1
L (2k + 1)
2
=
L (2k + 1) = s·
7r
i.e.,
k=O
2
1r
2 -4, k=O
2
Thus, Fourier series can be employed to determine the exact sums of certain series.
Having completed a general discussion of Fourier series, we shall consider next

certain important special cases.
Definition 8.5. A function f defined on an interval I of the form [-L, L] or (-L, L),
L > 0, is odd if J(-x) = -J(x) for all x in J.
Example 8.6. For any L > 0, the function J(x) = x3 is odd on [-L, L] because
J(-x) = (-x)3 = -x3 = - J(x) for all x in [-L, L].
Definition 8.6. A function f defined on an interval I of the form [-L, L] or (-L, L),
= f(x) for all x in J.
L > 0, is even if f(-x)
Example 8.7. For any L > 0, the function J(x) = x2 is even on (-L, L) because
f(-x) = (-x) 2 = x2 = J(x) for all x in (-L, L).
Example 8.8. The function J(x) = ex is neither even nor odd on any interval
I= [-L, L] since J(-x) = e- x -/- ±ex = ±J(x), except at the single point x = 0.
Example 8.9. The function f(x) = x2 is not even on [0, 1] beca use [0, 1] is not of
the form [-L, L] or (-L, L).
Theorem 8.2. If J is odd on [-L, L] or (-L, L), then
1: J(x) dx = 0.
1: 1:
Proof.
L
f(x) dx = J(x) dx + 1 J(x) dx.
Let y = -x in the first integral on the right to convert it into

o L £
l J(-y)(-dy) = 1 J(-y) dy = -1 J(y) dy
since J(-y) = - J(y). Thus,
1 £
-L
L
f(x) dx = - ( f(y) dy + ( f(x) dx = 0.
lo lo
L
Theorem 8.3. If J is even on [-L, L] or (-L, L), then
1
-L
£
J(x) dx
L
= 2 { J(x) dx.
lo
1: 1:
Proof:
£
J(x) dx = J(x) dx + 1 J(x) dx.
Let y = -x in the first integral on the right to convert it into
o L L
J J(-y)(-dy) = ( J(-y) dy = ( J(y) dy
1, lo lo
since J(-y) = J(y ). Thus,
1£
-L
f(x) dx = [1
lo
"' L
J(y) dy + ( f(x) dx = 2 ( J(x) dx.
lo lo
L
Theorem 8.4. Suppose that f is a 2L-periodic function. If f is odd, then the

Fourier series off reduces to
with
£
bn = 11 f( x ) sin ( � ) dx,
n x
n 2'. 1.
This series is called the Fourier sine series ( or simply the sine series) of J.
Proof. The Fourier series off is
ao � n 1r x . n 1r x
2 + L [an cos ( L ) + bn s ( L )]
m
n=l
with
Ll 1-L f( x ) cos ( L
£ n1r x 1 £ n 1r x
1-L J( x ) sin (
an ) dx, n 2'. 0, bn = L ) dx, n 2'. 1.
=
L
Since cos ( � ), 2'. 0, is an even function and f is an odd function, their product
n x
n
is an odd function and, hence, an = 0 for n

2'. 0, by Theorem 8.2. Since sin ( ; ),
n x
n 2'. 1, is an odd function and J is an odd function, their product is an even function
and, hence,
bn =± 1: f( x ) sin ( ; ) dx
n x
= 11 L f( x ) sin ( ; ) dx,
n x
n 2'. 1,
by Theorem 8.3.
Example 8.10. Let J( x ) = x on the interval [-2, 2) and f( x + 4) = f( x ) for all

T
x. This function is 4-periodic and odd. Hence, by Theorem 8.4 with L =
2 = 2, the
Fourier series of J reduces to the sine series
with
Thus, the Fourier series of f is the sine series

4(-1r- 1 . n 1r x
oo
L ----sm ( -- ) '
n 1r 2
n=I
and converges to f( x ) for all x except x = ±2, ±6, •••,where it converges to 0. The
graph of J is displayed in Figure 8.5.
y
2
-6 -4 -2 0 2 4 6 X
-2
Figure 8.5: The graph of f in Example 8.10.
Theorem 8.5. Suppose that f is a 2£-periodic function. If f is even, then the

Fourier series of J reduces to
ao � n 1r x
an cos ( L ) ,
2+L
n=l
with
21 L n 1r x
an = J ( x ) cos ( L ) dx, n 2'.'. 0.
L O
This series is called the Fourier cosine series ( or simply the cosine series) of f.
Proof. The Fourier series of J is
a0 �[ n 1r x n1rx
,
2 + L an cos L + n sin L
( ) b ( )]
n=l
.
with
1 1L n 1r x 1 1L n 1r x
-L J( x ) cos ( L ) dx, J( x ) sin (
an 2: 0, bn = L L ) dx, n 2: 1.
= n
L -L
Since sin ( � ), n 2: 1, is an odd function and f is an even function, their product

n x
is an odd function and, hence, bn = 0 for n 2: 1, by Theorem 8.2. Since cos ( l ),

n x
n 2: 0, and f are even functions, their product is an even function and, hence,
1 L n 1r x 2 L n 1r x
an = L 1 J ( x ) cos ( L ) dx = 1 J ( x ) cos (
L 0 L ) dx, n 2'.'. 0,
-L
by Theorem 8.3.
2 -2 < - X < -1
Example 8.11. Let J(x) = { 1: -1 ::; x < 1 } on [-2, 2) and f(x + 4) = f(x)
2, 1 :s; X < 2
for all x. The graph off is displayed in Figure 8.6. This function has period T = 4
y
2
-5 -4 -3 -2 -] 0 2 3 4 5 6 7 X
Figure 8.6: The graph off in Example 8.11.
T
and is even. Hence, by Theorem 8.5 with L =
. . 2 = 2, its Fourier series reduces to the
cosme senes
ao � n1rx
2 + �an COS (- -) '
n=l
2
with
212 J(x) cos (n1rx
1 cos n1rx- dx 12
- - ) dx
2
1 0 2
n1rx
(- ) + 2cos ( - -) dx, n � 0,
0 2 1 2
11
ao 1 dx + f 2 dx = 3,
2
2 n1rx 1 4
1
n1rx 1
2 2 n1r 4 . n1r
an
n1r
sin ( - -)
2 + n1r
sin (- -)
2 = n1r
sin ( )
2 - n1r sm (2)
0 1
n1r
_ _3_ sin ( ) ' n > - 1.
n1r 2
Thus, the Fourier series of f is the cosine series
00
3 2 n1r n1rx
-+ L--sin (-) cos ( --) ,
2 n=l n1r 2 2
and converges to f ( x) at every x except the odd integers, where it converges to �.
Exercises 8 .1
1. Show that if f is periodic with period T, then f (x + nT) = f (x) for every
integer n.
2. Let J(x) = x 2 + 1 for -1::::; x < 2 and f(x + 3) = J(x) for all x. Determine the
values f(x) off at x = 5, 7, -4, 783 and -291.
3. Let f be 2L-periodic, and suppose that f is represented by a trigonometric

series, i.e.,
ao � n1rx n1rx
f(x) = + L [an COS (L ) + bn sin ( )]
2 n=l
L
11:
The coefficients an have been determined in the text. In a similar way, show
(formally) that bn = n x
f(x) sin ( � ) dx for every n 2'. 1.
4. Let f(x) = { �: -� � �: � } and f(x+4) = f(x) for all x . Find the Fourier
series off.
O O<x<l . .
5. Let J(x) = {
x: 1 � x < 2
} and f(x + 2) = f(x) for all x. Fmd the Founer
series off.
6. Let f( x) = 2x - 3 for 2::::; x < 5 and f(x + 3) = J(x) for all x. F ind the Fourier
series off.
1
7. Let J(x) =- for O < x :::; 1 and f(x + 1) = f(x) for all x. Is f piecewise
X
continuous?
0 -1 < X < 0 }
8. Let f(x) = { - '.
x 0�x<1
, and f(x + 2) = j(x) for all x.
(a) Find the Fourier series of f.

(b) Sketch the graph of f.
1
(c) Determine the numbers to which the series converges at x = 0,
2, 1, 18,
18.5 and 19.
9. Let f(x) = x2 for O::::; x < 2 and J(x + 2) = f(x) for all x.
(a) Find the Fourier series of f.

(b) Sketch the graph of f.

( c) Determine the numbers to which the Fourier series off converges atx = 3,
8 and -10.
10. Let J(x) = x3 for -1 :::; x< 1 and f(x + 2) = J(x) for all x. Find the Fourier
series off.
- -
1 1. Let f ( X) = { �: � � : � � } , and f( x + 4)= f ( x) for all x.
( a) Sketch the gr aph off.

(b) Fin d the Fourier series of J.
12. Let J(x) =x 2 for -1 :::; x< 1 and f (x+2) = J(x) for all x. Find the Fourier
series off.
1 -x 2 Q :::; X :::; 1
13. Letf(x)= { 2 }, and f(x+2)=J(x)for allx.
l-(x-2) , l<x< 2
( a) Sketch the graph of J.

(b) Find the Fourier series of f. Determine the points x where the series
converges to f( x).
( c) Select a suitable value of x in order to deduce the sum of the series
L00 (-1r-1
n=l
n2
14. Consid er the 3-periodic function f which has the graph d epictedby Figure 8. 7.
-4 -3 -2 -I 0 2 3 4 5 6 7 8 X
Figure 8. 7: The graph off.
(a) Explain why 18 5

J(x) dx = j J(x) dx.
2
-1
8.2. FOURIER SERIES OF FUNCTIONS ON FINITE INTERVALS 299
(b) Explain why 1: f(x) dx = .l3 f(x) dx.
15. Prove that if J is T-periodic, then, for any real number a,

1a+T 1
f(x) dx = T J(x) dx.
a 0
Hence, deduce that for any two real numbers a and /3,
1a
+T J(x) dx = 1/3+T J(x) dx.
a 13
Thus, the integral of a T-periodic function over any interval of length T is

the same. (Hint: Write 1 +T
a
J(x) dx = 1° J(x) dx + l
a +T
J(x) dx and let

y = x + T in the first integral on the right.)
-cos(x), -1r :s; x < 0
16. Let J(x) = {
cos(x), 0 :s; x < 1r
} and J(x + 21r) = f(x) for all x. Find
the Fourier series of f.
8.2 Fo urier Series of Functions on Finite Intervals

Suppose that a function f is defined only on a finite interval [a, bl, and it is necessary
to express f(x) as a Fourier series, a sine series, or a cosine series for a :s; x :s; b.
Such problems arise, for example, in the solution of ordinary and partial differential
equations. The Fourier series, including the special cases of sine and cosine series,
of such a function J are obtained as the Fourier series of certain periodic functions
which are suitable extensions of f to the entire real line JR = ( -oo, oo).
Definition 8. 7. Suppose that J is defined on an interval [a, b], and let I be an

J
interval containing [ a, b]. A function defined on I is called an extension of f from
[a, b] to I if i(x) = J(x) for all x in [a, b].
J
Of course, can be defined in any manner whatsoever outside the interval [a, b].
The only restriction upon the function ](x) is that it be equal to J(x) for all x in
[a, b]. Thus, there is an infinite number of possible extensions of a given function J
from [a, b] to I. However, only three very particular types of extension are important
in the study of Fourier series.
Definition 8.8. Let J be defined on [a, b] and let T = b- a. Define the function j
on JR by
](x) = J(x) for all x in [a, b] and ](x + T) = ](x) for all x.
The function f is called the T-periodic extension off to R
Example 8.12. Let f(x) = 1 - x on [O, 1]. Then the 1-periodic extension off to J
JR is defin�d by i(x) = f(x) = 1- x for O:::;; x:::;; 1 and ](x + 1) = ](x) for all x. The
graph off is displayed in Figure 8.8.
-2 -I 0 2 3 X
Figure 8.8: The graph of the 1-periodic extension f off to R
J
Note that, in Example 8.12, is not well defined at integral values of x since, on
the one hand, ](1) = f(l) = 0 but, on the other hand, ](1) = i(o) = J(0) = 1. In
general, if f is defined on a closed interval of the form [a, b] and T = b- a, then the
T-periodic extension off to JR is not well defined at the points nT, n = 0, ±l, ±2, · · ·,
J
unless J(a) = f(b). Such points, where an extension is not well defined, are points
of discontinuity. In the study of Fourier series, the value �f a function at a point x
of discontinuity is irrelevant, because the Fourier series off converges to the average
J
value of at x, regardless of how ](x) is defined, and we shall pay no more heed to
this matter.
J
Definition 8.9. Let f be defined on [a, b], let T = b - a and let be the T-periodic
T
extension of f to JR. Then, with L = , the Fourier series off on [a, b] is the Fourier
_ 2
series off, i.e.,
ao f-.. n1rx . n1rx
+ bn sm ( L )]
2+L n=l
[an COS (
L ) (8.7)
lb f-(x)cos (n1rx 1 1• b .
with
1
- - ) dx = - J (x) cos (n1rx
- ) dx, n 2: 0, (8.8)
l l - 1 l .
L a L L a L
b n1rx b n1rx
f(x) sin ( ) dx = J (x) sin ( ) dx, L n 2: 1, (8.9)
L a L L a
employing the fact that ](x) = J(x) for all x in [a, b].
The Fourier series off can be determined directly by Equations (8. 7), (8.8) and
(8.9), without any reference to the extension j. Knowledge of is required only if J
the sum of the series at points x ::; a or x 2:: b needs to be determined.
Example 8.13. The function f in Example 8.12 is the same as the periodic function
fin Example 8.4 on page 2§9, where its Fourier series was derived. Since and}' are J
piecewise continuous and f is continuous for O < x < 1, its Fourier series converges
to ](x) = f(x) for O < x < 1. Thus,
00
1 1
1-x = - + �-sin(2mrx), 0 < x < 1,
2 Lmr
n=l
and the series on the right (including the constant term) is the Fourier series of
J(x) = 1 - x on [O, 1].
Example 8.14. Let f(x) = x on [O, 1]. The graphs off and its 1-periodic extension
j to IR are displayed in Figure 8.9.
y y
0 X -2 -I 0 2 X
Figure 8.9: The graphs of J(x) = x on [O, 1] and its 1-periodic extension J to IR.
The Fourier series off on [O, 1] is the Fourier series of the periodic function f with
1
L = ,
2 1.e.,
a
2
° + L [a cos(2mrx) + b
n n sin(2mrx)] ,
fo 1
n=l
with
ao = 2 x dx = 1,
and,for n � l,
an 2 11 x cos(2mrx)dx = nn
0
- sin(2mrx)
X ll
O
--
1
nn
11 sin(2mrx)dx = 0,
0
[ xsin(2nnx)dx = _ __:_cos(2nnx)l + _!_ [ cos(2nnx)dx = --1-.

1 1 1
bn 2
lo nn O nn lo nn
Since J and ]' are piecewise continuous and J is continuous for O < x < l,
1 1
00
x = - - "" - sin(2nnx), 0 < x < l.

2 n=l
Lnn
�xample 8.15. Let f(x) = x + l on [-1,2]. The graph of the 3-periodic extension
f off to � is displayed in F igure 8.10.
y
3
-4 -I 0 2 5 8 X
F igure 8.10: The graph of the 3-periodic extension f off to R
The Fourier series off on [-1, 2] is the Fourier series of the periodic function J with
L =�,i.e.,
2
a0
2+�
�[ 2nnx
anCOS (--)
2nnx
+ bnsin (- -)]
3 3
with
2
3 4m 3 2mrx 1
-sin (-r) +-cos ( -)
mr 3 2n 7r
2 2 3 _1
1._ sin (4n7r) + _3 4 2m
[cos ( n7r) _ cos ( r)]
n 7r 3 2n 7r
2 2 3 3
1._ sin (4n7r) , n � l,
11:
n1T' 3
2n x
12
bn ( x + l)sin ( ; ) dx
1 2n7rx 1
+- 1
2 2n 7rx
--( x + l)cos ( -- ) cos (-- ) dx
n 7r 3 _ 1 n 7r _ 1 3
2
3 4n7r 3 . 2n 7rx 1
--cos ( - ) +- sm ( - )
n 7r 3 2n 7r
2 2 3 _1
4 3 4 2
_1._ cos ( n7r) + _ [sin ( n7r) + sin ( n7r)]
n 7r 3 2n 7r
2 2 3 3
4
_1._ cos ( n7r) , n � l.
n 7r 3
Thus,
x+l
� - � 1._ sin [2n7r(x + l)] , -1 < x < 2.

2 6 n=l
n1T' 3
Suppose that J is defined on an interval of the form [O, L], L > 0, and that a
Fourier sine series representation of f on [O, L] is required. Since a sine series is the
Fourier series of an odd function, the extension f of f to IR must be an odd function.
Definition 8.10. Let f be defined on [0, L], and let ii be an extension of f to

[- L, L]. If Ji is an odd function, then ii is called the odd extension off to [-L, L].
Definition 8.11. Let j be defined on [0, L], let ii be the odd extension of J to
[-L, L], and let j be the 2L-periodic extension of fi to R Then J is called the odd,
2L-periodic extension off to R
Example 8.16. Let f(x) = 1 - x on [0, 1]. The graphs of the odd extension ii off
to [-1, 1] and the odd, 2-periodic extension J of J to IR are displayed in Figures 8.11
and 8.12, respectively.
y y
I I
0 X -I 0 X
-I
Figure 8.11: The graphs of J and its odd extension ii to [-1, l].
y
I
-2 -I 0 2 3 4 5 6 X
-I
Figure 8.12: The graph of the odd, 2-periodic extension j of J to R
The odd extension ii of f to [-1, 1] is defined by
. -1 - X, -1 S XS O
J 1 ( X) = { }
1 - X, 0S X S1 '
with the understanding that Ji is not well defined at x = 0.

Definition 8.12. Let f be defined on [0, L]. The Fourier sine series off on [0, L]
is the Fourier series of the odd, 2L-pcriodic extension J of J to IR, i.e.,
(8.10)
with
£
bn = 11£ f( x ) sin ( ; ) dx
n x
=
i1 f( x ) sin ( ; ) dx ,
n x
n � l, (8.11)
employing the fact that f(x) = f (x) for 0 ::; x ::; L.
The sine series of f on [0, L] can be determined directly by Equations (8.10) and
(8.11), without any reference to the extension J.
Knowledge off is required only if
the sum of the series at points x ::; 0 or x � L needs to be determined.
Example 8.17. The sine series of J(x) = l-x on [0, 1] is
2::)n sin(mrx),
n=l
with
1
bn 21 (1 - x) sin(mrx) dx
2 2 1
--(1 -x) cos(mrx) - -
1 1
1
cos(mrx) dx
nK O nK 0
-, n � l.
nK
Since f is continuous for 0 < x < l,
00 2
l-x = I:-sin(nKx), 0 < x < l,
n=l
nK
and also at x = l since J(l) = 0, which is also the sum of the series at x = l.
Example 8.18. The sine series of f(x) = x on [0, 1] is
Lb n sin(nKx),
n=l
with
1 1
bn 2 [ xsin(mrx)dx = _2_xcos(mrx)l
mr
+ 2_
mr
[1 cos(mrx)dx
lo O lo
2(-1r-1
n � 1.
Since f is continuous for O < x < 1,
L-
oo -l)n l
x=
2(
-
n1r
--
sin(mrx), 0 < x < 1,
n=l
and also at x = 0 since f (0) = 0, which is also the sum of the series at x = 0. The
graphs of the odd extension Ji of f to [-1, 1] and the odd, 2-periodic extension of J
J to IR. are displayed in Figures 8.13 and 8.14, respectively.
y
I
0 I X -I 0 I X
-I
Figure 8.13: The graphs of f(x) = x on [O, 1] and its odd extension !1 to [-1, l].
y
I
-3 -2 -I 2 3 X
-I
Figure 8.14: The graph of the odd, 2-periodic extension J of f to R

The odd extension Ji of J to [-1, 1] is defined by fi(x) = x.
1 0<x<l} . . .
Example 8.19. Let J(x) = { o: . The sme senes of J on [O, 2] 1s
1�x�2
12 11
with
n x n x
bn J(x ) sin ( ; ) d x = sin ( ; ) d x
n x
- : cos ( ; )
1r
i:
n
� [1 - cos (
n
2
7r)] , n � 1.
Since f is continuous for 0 < x < 2 except at x = 1, and f (2) = 0,
. 1
At x = 1, the senes converges to the average value off.
2
Suppose that f is defined on an interval of the form [0, L], L > 0, and that a
Fourier cosine series representation of f on [0, L] is req�ired. Since a cosine series is
the Fourier series of an even function, the extension f of f to IR must be an even
function.
Definition 8.13. Let f be defined on [0, L], and let Ji be an extension of J to

[-£, L]. If Ji is an even function, then Ji is called the even extension ofJ to [- L, L].
Definition 8.14. Let f be defined on [0, L], let Ji be the even extension off to
J
[-£, L], and let be the 2L-periodic extension of Ji to R Then is called the even, J
2L-periodic extension off to R
Definition 8.15. Let f be defined on [0, L]. The Fourier cosine series off on [0, L]
is the Fourier series of the even, 2£-periodic extension off to IR, i.e., J
ao � n 1r x
+ L..tan ( ) (8.12)
2 n=l
COS
L
with
an = f 1L n x
i(x) cos ( ; ) dx = f 1L f(x) cos ( ; ) dx,
n x
n � 0, (8.13)
employing the fact that i( x) = f(x ) for 0::; x :=; L.

The cosine series of J on [O, L] can be determined directly by Equations (8.12)

and (8.13), without any reference to the extension J. Knowledge of j is required only
if the sum of the series at points x :S: 0 or x 2:'. L needs to be determined.
Example 8.20. Let j(x) = 1 - x on [O, 1]. The graphs of the even extension Ji of
J to [-1, 1] and the even, 2-periodic extension J of f to IR. are displayed in Figures
8.15 and 8.16, respectively.
y y
0 X -1 0 X
Figure 8.15: The graphs of f and its even extension Ji to [-1, 1] .
-3 -2 -1 0 2 3 4 5 X
Figure 8.16: The graph of the even, 2-periodic extension J of f to R
l+x -1:S:x:S:0
The even extension Ji of J to [-1, 1] is defined by fi(x) = {
1 _ x'.
}
0 :S: x :S: 1 ·
The cosine series off on [O, 1] is
ao
cos(mrx),
2 +�an
"
n=l
with
ao 2 1\ 1 - x) dx = l,
1 1 1
an 21 (1 - x) cos(mrx) dx = 2_(1 - x) sin(mrx) 1 + 2-1 sin(mrx)dx
o n1r o n1r o
2 1
1
- 22 cos(n1rx)
n 7r o
2
[1- (-1)],
n
n 2:: 1.
n 7f
22
Since J is continuous for O � x � 1,

00
1 2[1 ( 1ri
l -x = +
2 L �2
n=l
7r�
cos(n1rx), 0 � x � l.
Example 8.21. Let J(x) = x on [O, 1]. Th_e graphs of the even extension Ji off to
[-1, 1] and the even, 2-periodic extension J of f to IR are displayed in Figures 8.17
and 8.18, respectively.
y y
I X -I X
Figure 8.17: The graphs off (x) = x on [O, 1] and its even extension Ji to [-1, 1].
-3 -2 -I 0 2 3 X
Figure 8.18: The graph of the even, 2-periodic extension f of f to R

The even extension Ji off to [-1, 1] is defined by fi (x) = lxl.

The cosine series off on [O, 1] is
1
fo
with
ao x dx = 1,
2
1 1 1
an 21 x cos(mrx) dx = _3_x sin(mrx) _3_ 1 sin(mrx) dx
o n1r o 1 - n1r o
2
1
2 cos(n1rx) 1
n 7r
2 o
2
(
27r 2 [ -lt-1],
n � 1.
n
-
Since f is continuous for O ::::; x :::; 1 ,

00
1 2[( l)n - 1]
x=-+� -2 2 cos(n1rx) ' o::::;x::::;1.
�
2 n=l n 1r
Since ( -1 t- 1 = 0 if n = 2k is even and ( -1 t - 1 = -2 if n = 2k + 1 is odd, the

last equation becomes
1 -4
00
x=
2 + L
k=O
(2k + l)21r2 cos[(2k + l)1rx], 0::::; x:::; 1.
Setting x = l and employing the fact that cos[(2k + l)1r] = -1 for all k � 0, we
obtain
which gives
In addition, since
1
00 00
1
00
1 1
00
1
00
1
� n2 =
n-l
L
k=l
(2k)2
+ L (2k +
k=O
1) 2 = 4L
k=J
k2
+ L (2k + 1) '
k=O
2
we obtain
3 00
1
L
7r 2
i.e.,
4 n2 = 8'
n=l
2, 0 < x < l } . . .
Exam p 1 e 8 . 22 . Let ;'( x ) = { . Its cosme senes 1s
l, 1 ;; x :::; 2
ao � (n1rx)
2 + Lan cos -2-
n=l
with
Since f is continuous on [O, 2] except at x 1 and f is an even extension, J is

continuous on [O, 2] except at x = l. Hence,
3
At x = 1, the series converges to the average value of J. This is evident from the
_ _ 2
senes, smce
Exercises 8. 2
1. Suppose that the function f is defined on the interval [O, 6]. Determine the form
of its Fourier series.
2. Suppose that the function f is defined on the interval [-3, 5]. Determine the
form of its Fourier series.
O< x< 1 . senes
3. Let f(x) = { � F'md t he Founer . of f .
: 1 ;; x :::; 2 } .
4. Let f(x) = { �: O< x<

<- X <-
7f - 27f
1r } . .
. F'md the Founer senes of f· .
5. Let J(x) = x on [2, 4]. Find the Fourier series off.

6. Let f(x) = x3 on [1, 3]. Find the Fourier series off.
of its Fourier sine series.
8. Let f (x) = { �: �� ; � � }· F ind the Fourier sine series off.
_
9. Let f (x) =
{ X, 0::; X < 1
2 _ x, 1::; x::; 2
} . F ind the Fourier sine series of f.
10. Let f(x) = x2 for O::; x::; 2.
(a) F ind the Fourier sine series off.

(b) Sketch the graph of the odd, 4-periodic extension J of f to R
(c) Determine the points x in [O, 2] where the series converges to f(x).
(d) Determine the numbers to which the sine series of f converges at x = 3,
x = 8 and x = -10.
of its Fourier cosine series.
12. Let f (x) = { �'. ½ �: � l }· F ind the Fourier cosine series off.
13. Let f(x) = x3 on [O, 1]. Find the Fourier cosine series off.
14. Let f(x) = sin(x) on [O, 3]. F ind the Fourier cosine series off.
15. Let J(x) = x2 on [-2, 2].
(a) Sketch the graph of the 4-periodic extension J off to JR.

(b) Determine all points x where the Fourier series off converges to J(x).
(c) F ind the Fourier series of f.
16. Let f(x) =

{ X, 0::; X < 1
l, 1::; x::; 2
}.
(a) Sketch the graph of the odd, 4-periodic extension J off to R

(b) F ind the Fourier sine series of f. Determine the points x in [O, 2] where
the series converges to f (x).
(c) Determine the numbers to which the sine series of J converges at x = 0,
x = 1, x = 2, x = 78 and x = 79.
(d) Sketch the graph of the even, 4-periodic extension J of f to R
(e) Find the Fourier cosine series of f. Determine the points x in [0, 2] where
the series converges to f ( x).
(f) Determine the numbers to which the cosine series off converges at x = 0,
x = 1, x = 2, x = 78 and x = 79.
11 x <
17. Let f be defined on the interval [0, 1r] by J(x) = { � �< < � }.
0 ' 2-X_Jf
(a) Find the Fourier sine series off. Determine the points x in [0, 1r] where
(b) Find the Fourier cosine series of f. Determine the points x in [0, 1r] where
(c) Find the Fourier series of f. Determine the points x in [0, 1r] where the
series converges to f ( x).
(d) Simplify the series in part (c) and select a suitable value of x in order to
oo (-l)k
deduce the sum of the series
k=O
L
2k + 1
.
{
Chapter 8 Exercises
x-3
1. Let J(x) = 1, '
6-x,
(a) Sketch the graph of the 3-pcriodic extension J off to R

(b) Find the Fourier series of f.
(c) Determine all points x where the Fourier series of f converges to !(x).
2. Suppose that f and J' are continuous on [0, L], and let F(x), S(x) and C(x)
denote the Fourier series, the sine series and the cosine series of f, respectively.
J
By analyzing the suitable extensions of f to JR in each case, determine the
conditions under which:
(a) F(x) = J(x) at x = 0; F(x) = J(x) at x = L.

(b) S(x) = f(x) at x = 0; S(x) = f(x) at x = L.
(c) C(x) = f(x) at x = 0; C(x) = J(x) at x = L.
n x
3. (a) Let L > 0 and Yn (x) = sin ( � ), n � 1. Show that
L
1 Yn (x)ym (x) dx = 0 if n f m.
In this case, the functions Yn and Ym with n #- mare said to be orthogonal

on [O, L], and the set of functions { yn, n 2: 1} is called an orthogonal family
on [O,L].
n x
(b) Let L > 0 and Yn(x ) = cos ( � ), n 2: 0. Show that the set of functions
{ yn , n 2: O} is an orthogonal family on [O, L].
(c) At the beginning of Section 8.1 and in Exercise 3 in Section 8.1, the Euler
Fourier formulas for the coefficients an and bn of a Fourier series were
determined by employing the fact that the set of functions
is an orthogonal family on [-L,L]. Employ part (a) to show (formally)

that if
f(x) = Lb n
n
sin ( � ) ,
x
n=l
i1
then
L
bn = n
f(x ) sin ( � ) dx,
x
n 2: 1.
(d) Similarly, employ part (b) to show that if
ao n1r x
f(x) = +� cos (
2 L..,. a
n=l
n
L)
then
4. Suppose that f and g are continuous and 2L-periodic, and that f' and g' are
piecewise continuous. Let
ao � n 1r x . n 1r x
cos (L ) + bn sm (
2 + L..,.
n=l
[a n )]
L
be the Fourier series of J, and
co � n1r x n 1r x
+ L..,. [Cn cos ( ) + dn sin ( )]
2 n=l
L L
the Fourier series of g. Prove that if the two series are equal for all x, then
a n = Cn for all n 2: 0 and bn = dn for all n 2: 1. (The analogous results hold for
sine series if f and g are odd, and for cosine series if they are even.)
5. (a) Let
. (7fX)
J.( X ) = 2 sin . (47rX)
3 - 3 sin . (57fX)
- - + 4sm 3
3
Determine the fundamental period of J and find its Fourier sine series by
computing its sine coefficients, employing the appropriate Euler-Fourier
formulas and Exercise 3(a). F ind the sine series of J by employing the
result of Exercise 4.
(b) Let
3 x x x
J(x) = 3 - cos ( 1r ) + 2cos (61r ) + 5cos (71r ) + 3cos (41rx).
4 4 4
Determine the fundamental period of J and find its Fourier cosine series.
(c) Let
J(x) =
1
- 2cos
(1rx) + cos (31rx 7 x
- - ) + 3sin ( - 1r - )
9x
- 5sin ( - 1r - ) .
2 2 2 2 2
Determine the fundamental period off and find its Fourier series.
Chapter 9
Series Solutions of Differential

Equations
9.1 Solutions About Ordinary Points

Definition 9.1. Consider the second-order, linear, homogeneous equation
y11 + p(x)y' + q(x)y = 0,
and let x0 be a given point. If both of the functions p and q arc analytic at x0, then
the point x 0 is called an ordinary point of the equation. If at least one of p and q is
singular at xo, then xo is called a singular point or singularity of the equation.

y11 - xy' + e x y = 0,
with p(x) = -x and q(x) = ex . Since both p and q are analytic at every point x0,
every point is an ordinary point of the equation.

(X -
2) yII + eX yI - -y
1 = 0.
In standard form, the equation becomes
ex I 1
II
+ - = 0,
y -.-2y xx-
x- ( 2) y
ex 1
with p(x) = -- and q(x) = - . Since pis analytic at everyx0 I- 2, and q
x- 2 (
xx- 2)
is analytic at every x0 I- 0 or 2, the pointsx0 = 0 and x0 = 2 are singular points, and
all other points are ordinary.
317
318 CHAPTER 9. SERIES SOLUTIONS OF DIFFERENTIAL EQUATIONS
Theorem 9.1. If x0 is an ordinary point of the differential equation
y" + p(x)y' + q(x)y = 0,
then the general solution is analytic at x0, and is therefore given by
Y=Lan(X-xor, lx-xol<R,
n=O
with two of the coefficients (usually a 0 and a 1) being arbitrary. Moreover, the radius
of convergence R is at least as great as the distance from x 0 to the singular point x 1
of the equation which is closest to x0. If the equation has no singular points, then
R= oo.
1
y
II
-2yI +--y = 0
x-1
has a singular point at x1 = 1, and all other points are ordinary. The series solution
about x0 = 0 has radius of convergence R 2". 1, the distance from the expansion point
x0 = 0 to the nearest singular point x1 = 1, by the last part of Theorem 9 .1. The
series solution
y = �bn x-
= ( l)
n
about x0 = i has radius of convergence R 2". �, the distance from the expansion point
x0 = !4 to the nearest singular point X1 = 1.

Example 9.4. Complex singularities cannot be ignored. The equation
y" + _l _y' +y =0
x2 + 1
has singular points at x1 = ±i. The series solution
9.1. SOLUTIONS ABOUT ORDINARY POINTS 319
about x0 = 0 has radius of convergence R 2:: 1, the distance from 0 to ±i. The series
solution 00
about x0 = 2 has radius of convergence R 2:: /5, the distance from 2 to ±i. We
cannot conclude that R = oo due to the absence of any real singularities.
y" + xy' + y = 0.
Since p(x) = x and q ( x) = 1 are analytic at every point x0 and, in particular, at
x0 = 0, the general solution can be expressed as
= I: an xn ,
00
y
n=O
by Theorem 9.1. By Theorem 7.2, the derivatives of y may be obtained by termwise

differentiation of the series, and substitution of y, y' and y" into the differential
equation will result in a relation amongst the coefficients an , n 2:: 0, from which every
an and, hence, y, may be determined. Thus,
= I: nan x L n(n - l)a x -

00 00 00
y = I:anX n
=} y' n
-l y" = 11
n 2
,
n=O n=O n=O
and y is a solution of the differential equation if and only if
L n(n - l)a x -2 +XL na X -l + L a x

00 00 00
n
n
n
n
n
n
= 0,
n=O n=O n=O
i.e.'
L n(n - 1)a X -2 + L na X + L a x
00 00 00
n
n
n
11
n
n
= 0. (9.1)
n=O n=O n =O
In the first series, the term n(n - 1) = 0 when n = 0 and when n = 1. Hence,
00 00
n=O n=2
Make the change of index n --+ n + 2 to obtain

00 00 00
n=2 n+2=2 n =O
Condition (9.1) then becomes
2)n + 2)(n + l)an+2Xn + L nanxn + L anxn = 0.

00 00 00
n=O n=O n=O

Since all three series now have precisely the same form (i.e., they have the same power
of x and the same initial value of n), they may be combined into a single series by
adding their coefficients. Thus,
00 00 00
n=O n=O n=O
By the statement preceding Example 7.18 on page 259, it follows that

(n + 2)(n + l)an+2 + (n + l)an = 0
for all n � 0. Dividing by n + l -=I= 0 and solving for an+2 in terms of an, we obtain
an
an+2 = - , n � 0,
n+2
called the coefficient recursion (or recurrence) relation. This relation must be solved
in order to determine every a n explicitly in terms of n. The form of the relation
shows that the an with n even are determined by a 0, and the ones with n odd are
determined by a 1. For the coefficients a n with n = 2k even,
ao
n=0 ⇒ a2 = - ,
2
a2 ao ao
n= 2 ⇒ a4 = - 4 = • 4 = 2
2 2 ( 1 • 2)'
a4 ao ao
n =4 ⇒ a 6 = - 6 = - . 4 . 6 = - 3
2 2 ( 1 · 2 · 3)'
a6 ao ao
n=6 ⇒ as = - 8 = . 4 • 6 • = 4
2 8 2 ( 1 · 2 · 3 · 4) '
and the emerging pattern indicates that
(-l)k ao
a2k = 2k k! ,
k � 0.
For the coefficients a n with n = 2k + l odd,

a1
n=l ⇒ a3 = -3,
a3 a1
n=3 ⇒ a 5 = -5 = 3 · '
5
a5 a1
n=5 ⇒ a 7 = -7 = - 3
· 5 · 7'
and, in general,
(-l)ka1
a2k+1 = 3. 5 . 7 .. . ' k 2'.'. 1.
(2k + 1)
2 · 4 · 6 · · · (2k)
In order to express a2 k+ l in a more concise form, multiply it by = 1
2. 4. 6 ... (2k)
to obtain
The solution is therefore given by
L L L
00 00 00
y = anX
n
= a2kX
2k
+ a2k+ 1X
2k+l
n =O k=O k=O
� (-l)k 2k � (-l)k2kk! 2k+l

ao
L 2kk! x + a 1 L (2k + 1)! x
k=O k=O
aoy1 + a1 y 2,
where
Y � (-l)k � (-l)k 2k k! 2k+l
1 = L k x 2k and Y2 -
- L
2 k! k
(2k + 1)! x .
k� �
Since there are no restrictions upon the constants a 0 and a 1, they are both arbitrary.
Setting ao = 1 and a 1 = 0 gives the solution y = y 1, and setting a0 = 0 and a 1 = 1
gives the solution y = Y2• Moreover, since Y1(0) = 1 =/- 0 and Y2(0) = 0, Y1 and Y2 are
linearly independent. Thus, y = aoy1 + a 1 y2 is the general solution, as guaranteed by
Theorem 9.1. Since the equation has no singularities, R = oo.
Note that the solution
is an elementary function, but y2 is not.
Since
it follows by Theorem 7.4 that

y( ) (0)
n
an = , n 2'.'. 0.
n.I
In particular,
ao = y(0) and a1 = y'(0).
Hence, if initial conditions are given, then a0 and a1 are determined.
For example, if y(0) = 2 and y'(0) = 0, then a0 = 2 and a1 = 0, from which it

follows by the recursion relation that a2k+l = 0 for all k � 0, and the solution of the
initial-value problem is y = 2y1 = 2e-x2
(x2 + l)y" + xy' - 4y = 0.
Since
X -4
p(x) = -- and q(x) =- -
x 2
+1 x 2
+1
are analytic at x 0 = 0, the point 0 is an ordinary point of the equation and, hence,
Theorem 9.1 applies. Thus,
L
00
⇒ y" = n (n - l)a n x -z,

n
n =O n =O n =O
and y is a solution of the differential equation if and only if
L +XL L
00 00 00
(1 + x 2 ) n(n - l)anX -Z
n
nanX -l
n
-4 an X
n
= 0,
n =O n =O n =O
i.e.'
L L L L
00 00 00 00
n( n - l)an X -Z
n
+ n( n - l)an X
n
+ na n X
n
- 4anX
n
= 0.
n =O n =O n =O n =O
Combining the last three series, we obtain the condition

00 00
n =O n =O
and, making the change of index n ---t n + 2 in the first series, we obtain
00 00
n =O n =O
i.e.'
L [(n + l)(n + 2)a +2 + (n - 2)(n + 2)a ]

00
n n x
n
= 0.
n =O
The coefficient recursion relation is therefore
(n + l)(n + 2)a n+ 2 + (n - 2)(n + 2)an = 0,
or, since n + 2 -/= 0 for n � 0,
(2 - n)an
n - 2)an - ---
a +2 - -(---
- - n >_ 0 .
n n+l n+l '
For the coefficients an with n even,
n = 0 ⇒ a2 = 2ao,
n = 2 ⇒ a4 = 0,
n = 4 ⇒ a6 = 0,
etc.; hence, a2k = 0 for k � 2.
For the coefficients an with n odd,
a1
n=l ⇒ a =
3 2,
-a3 -
n=3 ⇒ a5 - - a1
- 4 - 2. 4'
-3as 3a1
n=5 ⇒ a7- ---
- 6 - 2-4·6'
-Sa1 -3· Sa1
n=7 ⇒ ag=-- =
8 2·4·6·8'
- 7a9 3 · 5 · 7a1
n=9 ⇒ au = -- =
10 2-4-6-8-10'
etc., and, in general,
(-1 )k -l1· 3· 5 · 7··· (2k - 3) a1

2 · 4 · 6· 8 · 10 · · · (2k)
(-1) - 1 1. 3. 5 • 7··· (2k - 3)a 1
k
2k k!
(-l) -t1 • 3· 5 · 7 · · · (2k - 3) a 1 2· 4 • 6 ·8· 10 ··· (2k - 2)
k
2k k! 2·4·6·8· 10 ··· (2k - 2)

1
(-1) - (2k - 2)!a 1
k
, k � l.
22 k -lk!(k - 1)!
Hence,
is the general solution, where
� (-1t- (2k - 2)! 2k +l

1
Y1 = 1 + 2x 2 and Y2 = x + L 2k x
2 -lk!(k - 1)!
k=l
Thus, y1 is an elementary function, but y2 is not.
Suppose that the initial conditions y(0) = -3 and y'(0) = 6 are given. Since y(0) = a0
and y'(0) = a 1 , the solution of the initial-value problem is
Of course, if y(0) = 0, then a 0 = 0 and y 1 need not be computed, and if y'(0) = 0,

then a 1 = 0 and Y2 need not be computed.
By Theorem 9.1, since the only singularities of the equation are x 1 = ±i, and the
distance from x0 = 0 to x 1 = ±i is 1, the radius of convergence of the general solution,
as well as that of any particular solution, is R 2:: 1.
Since y 1 is a polynomial, i.e., a finite Taylor series, its radius of convergence is R1 = oo.
In order to determine the precise value of the radius of convergence R2 of y2, we employ
the ratio test. Thus, with Y2(x) = x + L ck(x),
k=l
Ck 1 ( x) r (2k)!x2 k+3 22k -lk!(k - 1)!

lim I + I = I I
k--+oo Ck (X) k.:.� 2 2 k+ 1 (k + l)!k! (2k - 2)!x 2k+l
2 (2k)(2k - 1)
lxl lim 2
k --+oo 2 (k + l)k
lxl < 1
2
if lxl < 1. Hence, the radius of convergence R2 of y2 is precisely 1.

In equations where the coefficient recursion relation cannot be solved explicitly

for all n � 0, approximate solutions of the differential equation can be obtained by
retaining a finite number of terms of the series.
Example 9. 7. Consider the equation
y" + cos(x)y = 0,
with p(x) = 0 and q(x) = cos(x). Since p and q are analytic at x0 = 0, the general
solution is given by
Since
L= (-lt x k = 1 - 2!x + 4!x - 6!x + · · ·
2 4 6
cos(x) = 2
(2k)!
k=O
is an infinite series and not a polynomial, we shall retain only the first six terms in
y. Thus,
y ao + a1x + a2 x 2 + a3x.3 + a 4 x4 + a5x5 + · · ·,

y' a1 + 2a2x + 3a3x2 + 4a4 x3 + 5a5x4 + · · · ,
y" 2a2 + 6a3x + 12a4x2 + 20a5x3 + · • •
and y is a solution of the equation if and only if
Performing the multiplication and collecting terms, we obtain the condition
which leads to the relations
i.e., a partial set of recursion relations for the coefficients an . It follows that
and ao and a 1 are arbitrary. Thus,
where
12 1 4
Y1 = 1- x
2 + 12x + · and Y2 = x - 13
x +
1 5
x +···
·· 6 30
are two linearly independent solutions.
If a solution of the equation
y" + p(x)y' + q(x)y = 0 (9.2)

is sought about an ordinary point x0 -=f. 0, then the change of variables
t = x - x0, y(x) = z(t), p(x) = P(t), q(x) = Q(t),

transforms the equation into the equivalent equation
z" + P(t)z' + Q(t)z = 0, (9.3)

with t = t 0 = 0 as the ordinary point corresponding to x = x0. Once the solution
z(t) of Equation (9.3) is determined, the solution y(x) of Equation (9.2) is obtained
by y(x) = z(t) = z(x - xo).
If the solution of a differential equation is sought in a neighbourhood of a singular

point x 1, then Theorem 9.1 does not apply and the methods employed in the foregoing
examples fail. In order to obtain such a solution, a different procedure is required,
called the Frobenius method, a topic which is beyond the scope of this book.
Exercises 9 .1
1. Determine all points x0 where the given function is analytic.
3
(a) f(x) = -
x+2
x + 1
2
(b) J(x) = x 2 -4
(c) f(x) = tan(x)
x+l
(d) J(x) = -;;---
e -1
(e) f(x) = Jx + 2, x 2:'. -2
x2 - x + 1 .
2. Compute the Taylor series of f (x) = ---- about xo = 0 by d1v1·d·mg 1 + x
x+l
into 1 - x + x2 .
La x .
00
n
3. Consider the series n
n= 2
(a) Write down the first four terms.

(b) Let n = m + 2 and express the series as one involving the index m. Write
down the first four terms.
(c) Replace m by n in the series obtained in part (b) and write down the first
four terms.
(d) Show that step (b) may be avoided simply by replacing n by n + 2 in the
original series.
L na x + L a x +
00 00
4. Determine an , n 2:'. 0, if n
n
n
n 2
= 0.
n=O n=O
5. Consider the differential equation y" + xy' + 2y = 0.
(a) Find the coefficient recursion relation for the general series solution about
Xo = 0.
(b) Solve the recursion relation, i.e, find a n explicitly in terms of n, and thereby
obtain the general solution and two linearly independent solutions.
(c) Write down the first four terms of each one of the two series solutions.
(d) Express one of the series solutions as an elementary function.
(e) Deduce the radius of convergence of each series solution.
(f) Solve the initial-value problem for the given equation if y(0) 0 and
y'(0) = 2.
6. Consider the differential equation (1 + x 2 )y" - 4xy' + 6y = 0.

Xo = 0.
(b) Solve the recursion relation and thereby determine the general solution
and two linearly independent solutions.
(c) Find y satisfying the given equation if y(0) = -3 and y'(0) = 1.
7. Find two linearly independent solutions of the differential equation
y" + xy' - 2y = 0.
8. Find two linearly independent solutions of the differential equation
(2 + x2 )y" + 2xy' - 2y = 0.
9. Consider the equation
(a) Find the 5 th -degree Taylor polynomial T5 (x) of the general solution.
(b) Find T5 (x) for each one of two independent solutions.
10. Consider Legendre's equation of order v,
(1 - x2 )y" - 2xy' + v(v + l)y = 0,
where v is a constant.
Xo = 0.
(b) Deduce from the recursion relation that if v = m is a nonnegative integer,
then one of the two series solutions is a polynomial Pm (x) of degree m (i.e.,
show that am+2 = am+4 = am+6 = · · · = 0).
(c) Impose the condition Pm (l) = 1 (which specifies the arbitrary constant
within the polynomial solution) and find Pm (x) for 0 :S m :S 3. T he
polynomials Pm (x), with Pm(l) = 1, are called Legendre polynomials.
9.2 Periodic Solutions

Consider the nonhomogeneous equation
y" + )..y = j(x), -00 < X < 00, (9.4)
where J is 2L-periodic and continuous, f' is piecewise continuous, and >.. is a real
constant. Suppose that a 2L-periodic solution y of equation (9.4) is sought. By the
Fourier convergence theorem (Theorem 8.1),
ao � [ (n1rx) (n1rx)]
J (x) = + � an cos + bn sin < X < 00,
2 n=l
L L -00
9.2. PERIODIC SOLUTIONS 329
where an and bn are the Fourier cosine and sine coefficients off, respectively. Given
the form off, a 2£-periodic solution y is sought in the form of a 2L-periodic Fourier
senes, 1.e.,
Y=
co �
+ L [Cn COS (
n 1r x
) + d nsm (
n1r x
)]
.
2 n=l
L L
Once the coefficients Cn and dn have been determined formally, it can be shown
that the resulting series is the Fourier series of a function y, that y' and y" can be
determined by the termwise differentiation of the series, and that y, y' and y" are
continuous. Thus,
and y" + >.. y = f( x ) requires that

- t, [ n
::\n cos ( n� ) + n:: dn sin ( n�x )]
x
2
+>-{ � t, [en + cos ( n:x ) + dnsin c:x ) l}

�o + f
n=l
[an cos ( � ) + bn sin ( � )],
n x n x
which is satisfied if
and
n
2 2
If >.. -/- V for any integer n 2'. 0, then
1r
and dn -
- bn
_ n2 2, n 2'. 1,
>.. 1r
£2
and the solution y is determined.
If>.. = 0, then there is no periodic solution unless

£
ao =L f( x )dx = 0,
11-L
. . . k
m wh.1ch case, co 1s arb'1trary, and 1f A'
2 2
=V 1r
for an integer k 2'. 1, then there is no
periodic solution unless
L
l L k1r x l k1r x
ak z 1-LJ( x )cos (L ) dx =0 and bk 1-Lf(x)sin (
L ) dx =0,
= =
L
330 CHAPTER 9. SERIES SOLUTIONS OF DIFFERENTIA L EQUATIONS
in which case, Ck and dk are arbitrary. The case where no 2£-periodic solution exists
is one of resonance, which occurs if the forcing term f contains a term proportional
to a periodic solution of the homogeneous equation y" + >..y = 0. The solution in that
case is not periodic, but contains a term proportional to x.
Example 9.8. Let f(x) = x3 - 2x2 + 1 on [O, 2] and f(x + 2) = f(x) for all x. Since
f is continuous on [0, 2] and f(0) = 1 = f(2), f is continuous everywhere and f' is
piecewise continuous. Hence,
-00 < X < 00.
A 2-periodic solution of the equation
y" + 3y = f(x)
is therefore sought in the form
y= � + I)cn cos(mrx) + dn sin(mrx)], -oo < x < oo.

n=l
Since >.. = 3 -=f. n2 1r2 for any integer n 2 0, and L = 1,
, n2 1,
an bn
=
Cn = 3 - n 27r 2
, n 2 0, and d n
3 - n27r 2
and it remains to determine the Fourier coefficients of f. Thus,
2
3'
12 2x2 + 1) cos(n1rx) dx
2 + 1) sin(n1rx) 2 - -1 1 2(3x2
an (x3 -
1
1 -
2 o n1r1 1o 2
-(x 3 - 2x 4x) sin(n1rx) dx
n1r
1
-- (3x2 - 4x) cos(n1rx) 1 - 22
2 2
(6x - 4) cos(n1rx) dx
n 1r o n 7r o
2 2
�- �(6x - 4)sin(n1rx)\ + 6 [ sin(n1rx)dx
n 1r n 1r n 3 1r3 Jo
2
O
1
22
4 6 4
,
-44cos(n1rx) n21,
22
n 1r n 1r O n 1r
2
\x - 2x + 1) sin(mrx) dx
3
bn
f o
2 2
--(x 1 3 - 2x 2 + 1) cos(mrx) 1 + - 1 1 (3x 2 - 4x) cos(mrx) dx
1
nn nn 0
2
O
2
(3x2 - 4x) sin(nnx) 1 - 22
1 1
22
(6x - 4) sin(nnx) dx
nn O n n 0
1 2 2
6 1 cos(nnx) dx
33(6x - 4) cos(nnx) 1 - 33
nn O n n 0
12
n "2 1.
Hence,
1
y= - +
L
oo [ 4cos(nnx)
- - --- +- - ---
12sin(nnx)
] , <X<
9 n 2n2(3 - n 2n2 ) n3n3(3 - n 2n2 )
-OO 00.
n=l
Example 9.9. Let f be an odd, 2-periodic function with f(x) = x - x2 on [O, 1].
Since f is continuous on [0, 1] and J(O) = f(l) = 0, f is continuous everywhere and
J' is piecewise continuous. Hence,
00
J(x) = L b sin(nnx),
n -oo < x < oo.
n=l
y" + y = J(x)
00
n=l
Since >. = 1 =/= nn for any integer n "2_ 0, and L = 1,
2 2
bn
2 2'
dn n"2_1.
-nn
=
l
The Fourier sine coefficients of f are
bn = 2 fo\x - x2 ) sin(nnx) dx
2
--(x - x2) cos(nnx) 1 + -2
1
11
(1 - 2x) cos(nnx) dx
11
nn O nn 0
2
4
1
(1 -2x) sin(nnx) 1 + 22 sin(nnx) dx

22
n n O n n 0
4 1
4
- 3 3 cos(nnx) = 3 3 [1 - (-lt]' n "2 1.
nn 10 nn
Hence,
4[1-(-lr]
L
00
Y= sm( mrx), -oo < x < oo.

n=l
n3 1r3 ( l-n21r 2)
Example 9.10. Let f be even and 2-periodic, with J(x) = x on [O, 1]. Since f is
continuous on [O, 1] and f is even, it is continuous everywhere and f' is piecewise
continuous. Hence,
00
ao
f(x) = +�
L..,ancos(n1rx), < < 00.
2 n=l
-00 X
y" + y = f(x)
00
Co �
y= + L....encos(n1rx), < < 00.
2 n=l
-OO X
Since >. = 1 i= n2 1r2 for any integer n 2'. 0, and L = l,

an
Co = a0 and = , n 2'. 1.
1-n2 7r 2
Cn
The Fourier cosine coefficients of f are

1
2 fo xdx = l,
1 1 1
21 xcos(n1rx) dx = 2-xsin(n1rx)l - 2_ 1 sin(n1rx)dx
o n1r o n1r o
2 11 2
cos(n1rx) = [(-lt - 1], n 2'. 1.
n 1r
22 O n 1r
22
Hence,
1 2[(-1r-11
L n21r2( l-n21r2) cos(n1rx),
00
y = +
-2 -OO < X < 00.
n=l
Exercises 9. 2
1. Let f(x) = x2 - x + 2 on [O, 2] and .f(x + 2) = f(x) for all x. Find a 2-periodic
solution of the equation y" + 3y = J(x).
2. Let J(x) = x2 - 2x on [-1, 3] and J(x + 4) = J(x) for all x. Find a 4-periodic
solution of the equation y" + 4y = f(x).
3. Let J be an odd, 2-periodic function, with f(x) = x - x2 on [O, 1]. Find a

2-periodic solution of the equation y" - 3y = f(x).
4. Let J be an odd, 4-periodic function, with J(x) ={ 2

�'
x,
� � : � ; } . Find
a 4-periodic solution of the equation y" + 3y = f(x).

5. Let J be an even, 2-periodic function, with J(x) = 1 - x on [O, 1]. Find a
2-periodic solution of the equation y" + y = f(x).
6. Let J be an even, 4-periodic function, with f(x) ={ �: � �; �; }· Find a

4-periodic solution of the equation y" + 4y = J(x).
7. Let J(x) = 2x - x2 on [O, 2] and f(x + 2) = f(x) for all x. Find a 2-periodic
solution of the equation y" + 2y = J(x).
8. Let f be an even, 2-periodic function, with J(x) = sin(21rx) on [O, 1]. Find a
2-periodic solution of the equation y" = f(x).
Appendix A
Formulas and Techniques
A. l Trigonometric Identities
cos 2(0) + sin 2(0) = 1
cos(a + b) = cos(a) cos(b) - sin(a) sin(b)
cos(a - b) = cos(a) cos(b) + sin(a) sin(b)
sin(a + b) = sin(a) cos(b) + cos(a) sin(b)
sin(a - b) = sin(a) cos(b) - cos(a) sin(b)
sin(20) = 2sin(0) cos(0)
cos(20) = cos 2(0) - sin 2(0) = 2cos 2(0) - 1 = 1 - 2sin 2(0)
1
cos 2(0) = [1 + cos(20)]
2
2
sin (0) = �[1- cos(20)]
scc 2(0) = tan 2(0) + 1
csc 2(0) = cot2(0) + 1
A.2 Integration Techniques

Integration by Parts
J f' ( x)g ( x) dx = J ( x)g ( x) - J J ( x)g' ( x) dx.
Proof. By the product rule of differentiation,
[f(x)g(x)]' = f'(x)g(x) + J(x)g'(x) ⇒ J(x)g(x) = j J'(x)g(x) dx + j J(x)g'(x) dx

⇒ J J'(x)g(x) dx = J(x)g(x) - J J(x)g'(x) dx.
335
336 APPENDIX A. FORMULAS AND TECHNIQUES
Integration by Substitution ( Change of Variable)

J J(g(x))g'(x) dx = J J(u) du, where u = g(x).
Proof. By definition of the indefinite integral as an antiderivative,
d�
J J(g(x))g'(x) dx = J(g(x))g'(x) and �
d
J f(u) du= f(u).
Hence, by the chain rule,

d
J J(u) du=
( d
J J(u) du
) du du .
= J(u) dx = J(g(x))g'(x).
J
dx du dx
Since both J(g(x))g'(x) dx and / f(u) du have the same derivative with respect
to x, they are equal (up to the addition of an arbitrary constant).
Trigonometric Integrals
1. J sin°' (x) cosn ( x) dx, n � 1 an odd integer and a any real number.
Factor out cos(x) and convert cosn - 1 (x), with n- 1 even, into sines, employing
cos2 ( x) = 1 - sin2 ( x), to express the integral as
j f(sin(x)) cos(x) dx.
Then make the substitution u = sin(x) to obtain J J(u)du.
2. J cos°' (x) sinn (x) dx, n � 1 an odd integer and a any real number.
Factor out sin(x) and convert sinn- 1 (x), with n-1 even, into cosines, employing
sin2 ( x) = 1 - cos2 ( x), to express the integral as
J f(cos(x)) sin(x) dx = - J J(cos(x))[- sin(x)] dx.
Then make the substitution u = cos(x) to obtain - / f(u) du.
3. J cosm (x) sinn (:r;) dx, m � 0 and n � 0 both even integers.

Employ the identities
1 1
= =
cos2 (x)
2 [1 + cos(2x)] and sin2 (x)
2 [1 - cos(2x)],
and repeat if necessary until only first powers of cosines remain.
A.2. INTEGRATION TECHNIQUES 337
4. For any real numbers a and /3,
j cos(ax) cos({Jx) dx =� j cos[(a - {J)x] + cos[(a + {J)x] dx.
j sin(ax) sin({Jx) dx = � j cos[(a - {J)x] - cos[(a + {J)x] dx.
j sin(ax) cos({Jx) dx =� j sin[(a - {J)x] + sin[(a + {J)x] dx.
7. j tan°' (x) see n (x) dx, n 2". 2 an even integer and a any real number.
Factor out sec2 ( x) and convert secn -2 ( x), n - 2 even, into tangents, employing
sec2 ( x) = tan2 ( x) + 1, to express the integral as
j J(tan(x)) sec2 (x) dx.
Then make the substitution u = tan(x) to obtain j f(u) du.
8. j tanm (x) sec°' (x) dx, m 2". 1 an odd integer and a any real number.
Factor out sec(x) tan(x) and convert tanm-l (x), with m - 1 even, into secants,
employing tan2 (x) = sec2 (x) - 1, to express the integral as
j f(sec(x)) sec(x) tan(x) dx.
Then make the substitution u = sec(x) to obtain j f ( u) du.
9. j tanm ( x) seen ( x) dx, m 2". 0 an even integer and n 2". 1 an odd integer.
Convert tanm(x) into secants employing tan2 (x) = sec2(x) - 1, to obtain an
integral containing a sum of constants times odd powers of sec(x). Then for
any k 2". 3 an odd integer, evaluate
j seck (x) dx = j seck -2 (x) sec2 (x) dx
by integration by parts.
Trigonometric Substitution
1. j(a2 - x 2 )� dx, nan odd integer, a> 0 any real number.
2, and employ the identity

7f 7f
Make the substitution x = asin(t), - S: t S:
2
= a cos (t) to obtain
J J
a2
- a sin (t)
2 2 2 2
(a2 - x2 )� dx = an+l cosn+l (t ) dt.
2. J (a2 + x 2 )� dx, nan odd integer, a> 0 any real number.
Make the substitution x = a tan(t), -� < i < �' and employ the identity
a2 + a2 tan2 (t) = a2 sec 2(t) to obtain
j(a2 + x2)� dx = J an+ l secn+2(t) dt.
3. j(x2 - a2 )� dx, nan odd integer, a> 0 any real number.
2, and employ the

7f 31r
Make the substitution x = a sec(t), 0 S: t < or S: t <
2 1r
= a2 tan 2(t) to obtain

J J
identity a2 sec 2 (t) - a2
(x2 - a2 )� dx = an+ l tann+ 1 (t) sec(t) dt.
4. J (a+ bx - x 2 )� dx, n an odd integer, a and b real numbers, b2 + 4a > 0.

Complete the square to obtain
2
a + bx - x = ( � + a) - (x - �)
2
,
and make the substitution
x-
b�
2
=
4
y 7f
+ a sin(t), -- -
7f
< -.
< t-
J
2 2
5. (a+ bx+ x 2 )� dx, n an odd integer, a and b real numbers, b2 - 4a < 0.

2
a + bx + x 2 = ( x + �) + ( a - � ) ,
and make the substitution
x+
b
2
= ya
/b2
-
4 tan(l),
7f
-- < t < -.
2
7f
2
A.2. INTEGRATION TECHNIQUES 339
6. j(a+bx+x 2)� dx, n an odd integer, a and b real numbers, b2 - 4a > 0.

2
a + bx + x2 = (x + �) - ( � _ a) ,
Partial Fractions
Partial fractions are required in order to evaluate integrals of rational functions, i.e.,
j ��:� dx, where P and Qare polynomials. First, suppose that deg(P) < deg(Q).
1. If Q factors into distinct linear factors, i.e.,
then there exist constants A 1, A 2, · · ·, An , such that

P(x) A1 2 A A
-- = --- + --- + ...+ ---,
n
Q(x) a1x+b1 a2 x+b2 an x+bn
called the partial fraction decomposition of ��:�.
2. If Q contains a repeated linear factor ( ax + b?, r � 2, then the partial fraction

decomposition of ��:� must include the r terms
B1 2B Br
---+----+···+- - -
-
ax + b ( ax + b)2 ( ax + b)r '
where B 1, B 2, ···,Br are constants. These terms must be included in addition

to the terms which correspond to the distinct linear factors of Q.
3. If Q contains the distinct, irreducible, quadratic factor ax2 +bx+ c, then the
partial fraction decomposition of��:� must include the term
Ax+B
ax2 +bx+c'
where A and B are constants, in addition to the terms which correspond to
the other factors of Q. A quadratic ax2 + bx + c is irreducible if its roots are
complex, which occurs if and only if b2 - 4ac < 0.
4. If Q contains the repeated, irreducible, quadratic factor (ax2 +bx+cf, r 2:: 2,

then the partial fraction decomposition of��:� must include the terms
A1x+B1 A2 x+B2 Ar x+Br

-----
ax +bx+ c + ---
2 (ax +bx+
2 - c)-- + ... + -
2
(ax- --cY'
2 +bx+ -
where Ai and Bi, 1 :S i :S r, are constants, in addition to the terms which
correspond to the other factors of Q.
If deg(P) 2:: deg(Q), then perform long division to express ��:� as
P(x) P1 (x)
(x)
Q(x) = R + Q(x)'
where Ris a polynomial and deg(Pi) < deg(Q).
A.3 Matrix Inversion

Let A be an n x n matrix. Two methods of computing A- 1 are given below.
1. Write down the augmented matrix (Alf), where I is then x n identity matrix,
and row reduce the augmented matrix in order to transform A into I. The
resulting augmented matrix is then (IIA- 1).
2. Let Mij denote the determinant of the submatrix obtained from A by deleting
the i-th row and j-th column, called the ij-minor of A. Let Aj = (-1)i +j Mij ,
called the ij-cofactor of A. The transpose of the matrix with entries Aj is called
the classical adjoint of A, and denoted by adj(A). Then
adj(A)
A_ 1 =
det(A)'
i.e.'
In particular, if A = ( : � ) is a 2 x 2 matrix, then
A_ 1 = 1 ( d -b )
ad- be -c a
Appendix B
Proofs of Selected Theorems
B.1 The Comparison Test

Theorem 6.11 Let {an } and {bn } be sequences, and suppose that
for all n 2='. k, where k is an integer. Then:
Lb La
00 00
1. If n converges, then n also converges.

n=l n=l
Lb
00 00
2. If Lan diverges, then n also diverges.

n=l n=l
n
Proof. Without loss of generality, assume O � an � bn for all n 2='. 1. Let S n = I: ak
k=l
n
and t n = I: bk. Since an 2='. 0 and bn 2='. 0, { s n } and { t n } are increasing sequences. In
k=l
addition, 0 � Sn � l n because O � an � bn .
00
L-t bn converges, then lim t n = t exists. Hence, S n � l n � t < oo ⇒ { s n } is

1. If�
n-+oo
n=l
00
bounded above. Since it is increasing, it converges. Thus, Lan converges.

n=l
lim l n = oo, i.e., L bn diverges.

00 00
2. If�
L-t an diverges, then lim S n
n-+oo
= oo ⇒ n-+(X)
n=l n=l
341
342 APPENDIX B. PROOFS OF SELECTED THEOREMS
B.2 The Limit Comparison Test

Theorem 6.12 Let {an } and {bn } be sequences with an > 0 and bn > 0 for n � k,
where k is an integer, and let
. a
L = 11m -n . n-+oo bn
Lb
00 00
1. If O < L < oo, then either both Lan and n converge, or they both diverge.
n =l n =l
Lb
00 00
2. If L = 0 and n converges, then Lan converges.

n =l n =l
Lb
00 00
3. If L = oo and n diverges, then Lan diverges.

n =l n =l
Proof. Without loss of generality, assume k = l.
1. If O < L < oo, let E

> 0 with E < L. Then there exists an integer N such
that n � N ⇒ I:: - LI < E. Hence, for all n > N, -E < :: - L < E, i.e.,
L - E < :: < L + E. Let a = L - E and /3 = L + E. Then O < a < f3, and
Lb
00
a< : < f3 ⇒ abn < an < /3b11 • By the comparison test, if converges,
n
n
n n=l
Lb
00 00 00
then Lan converges, and if n diverges, then Lan diverges. In addition,

n=l n=l n=l
Lb Lb
00 00 00 00
if La n converges, then n converges, and if Lan diverges, then n

n=l n=l n=l n=l
diverges.
2. If L = 0, let E
> 0. Then there exists an integer N such that n � N ⇒
I:: I < E, i.e., -E < :: < E, or -Ebn < an < Ebn. By the comparison test, if
Lb La
00 00
n converges, then n converges.

n=l n =l
b
3. If L =oo, then nlim � = 0. Thus, there exists an integer N such that n � N
an
I bn I . ->oo bn
⇒ - < E, 1.e., -E < - < E, or -Ean < bn < Ean . By the comparison test, if
an an
Lb
00 00
n diverges, then Lan diverges.

n=l n =l
B.3. THE ALTERNATING SERIES TEST 343
B.3 The Alternating Series Test

Theorem 6.13 Consider the alternating series
If {bn } is decreasing for n � k, where k is an integer, and lim bn

n->oo
= 0, then the series
converges.
Proof. Without loss of generality, assume {bn } is decreasing for n � 0. Consider the
sequence { sn} with n = 2k even, i.e.,
2k
S2k I)-l) i bi = b1 - b2 + b3 - b4 + b5 - · · · + b2k-l - b2k
i=O
Since bi - bi+ 1 > 0 for all i � 0, the sequence { s2k} is increasing. In addition,
2k
S2k I)-l) i bi = bi - (b2 - b3) - (b4 - b5) - · · · - (b2k-2 - b2k-1) - b2k
i=O
< b1,
i.e., the sequence { s2k} is bounded above and, hence, it converges. Thus, lim s 2k
k->oo
=s
exists. Next, consider the sequence { sn } with n = 2k + l odd. Then
S2k+1 = S2k - b2k+1 ⇒ lim s2k+1

k->oo
lim S2k - lim b2k+1 = s.
= k->oo k->oo
Thus, for all E > 0, there exists an integer K such that k � K ⇒ ls2k - sl < E: and
ls2k+1 - sl < E:. It follows that lsn - sl < E: for all n � N = 2K + 1, i.e., lim Sn = s.
n->oo
2:) - 1tb
00
Thus, n converges.
n=O
B.4 The Ratio Test

Theorem 6.16 Let { an} be a sequence and suppose that
I an+ 1
I
L = n->oo
lim
an
exists. Then:
00

n =O
00

n=O
Proof. Since lim

-- n-+oo
I aann+l I = L � 0 exists, there exists an integer k such that an =J. 0
. an+l
for all n � k because, otherwise, - - would be undefined. Assume, without loss of
an
generality, that k = 0.
1. If L < 1, there exists K > 0 such that L < K < 1. Let c = K - L > 0. Then
there exists an integer N such that
n � N ⇒ 11 a:: 1 1- LI < E ⇒ -E: < I a:: 1 I-L<E

⇒ L- E < I a:: 1 I < L+E= K
⇒ lan+ 1 I < Kian i-
It follows that la1I < Klaol, la 2 I < Kla1 I < K2 laol, la3I < Kla2 I < K3 laol, etc.,
00 00
and, in general, lan l < K laol for all n � 1. Since L K laol = laol L K is
n n n
n =O n =O
L Ja i converges by the comparison test, i.e.,

00
a convergent geometric series, n

n=O
00
Lan converges absolutely.

n=O
2. If L > 1, there exists K > 0 such that 1 < K < L. Let E = L- K > 0. Then
00
Hence, lan l > K laol

n
---+ oo ⇒ n___.CX)
lim lan l =J. 0 ⇒ lim an =J. 0 ⇒ Lan diverges
n=O
n-.oo
by the n tl'-term test.
3. If L = 1, the test is inconclusive, as shown by Example 6.87.

B.5. THE ROOT TEST 345
B.5 The Root Test

Theorem 6.17 Let {an } be a sequence and suppose that
L = lim lan l l / n
n---->oo
exists. Then:
00
1. If L < l, then Lan converges absolutely.

n=O
00
2. If L > l, then Lan diverges.

n=O
3. If L = l, then no conclusions can be drawn regarding convergence.
Proof.
1. If L < l, there exists K > 0 such that L < K < 1. Let E = K - L > 0. Then
L Ia I converges by the comparison test since L K

00 00
⇒ Ian I < K n
⇒ n
n
is a
n=O n=O
00
convergent geometric series. Thus, Lan converges absolutely.

n=O
2. If L > l, there exists K > 0 such that 1 < K < L. Let E = L - K > 0. Then
00
⇒ la n l > K n
-----+ oo ⇒ n-+oo
lim lan l /: 0 lim an i= 0
⇒ n--400 =} Lan diverges by the
n=O
L 2n1 converges and

00
.
3. If L = l, then the test is inconclusive For example,
L = lim (-;) l
/n 1
= , and f ..!_ diverges and L = lim
n=O
(..!..) l
/n
=
1.
n ---->oo n n n-oo n
n=O
B.6 The Binomial Theorem

For any real number a, (1 + x)" = t, (:)x• for !xi< I.
t (:)x•, lxl
Proof. Since the series converges absolutely for !xi < 1, let
f(x) = t, (:)x• = 1 + < I.
If f(x) = (1 +x)°', then

( ) = a(l+x)° - 1 =;, (1+x)J'(x) = a(l+x) = af(x).
°
f' x
( ) = af(x), then - - = - - =;,

( )
f'x a
Conversely, if (1+x)f'x
( )
fx l+x
= a ln(l+x)+ c = ln[(l+x)° ] + c =;, f(x) = k(l +x)°
ln lf(x)I
and, since (1+x)°' = 1 = f(O) at x = 0, k = 1 and f(x) = (1+x)°' . Thus, we must
show that (1+x)f'(x) = af(x). By Theorem 7.2, since f(x) = 1+ f
t
(�)xk ,
k=l
J'(x) = k (:) x._, = t, ( k+1) (

k
: l) x'
⇒ (1+x)J' (x) = f
k=O
(k+1) ( : )xk +x
k 1 k=l
f
k (:)xk -l
= t,(k+ l)( : )x• + tk(:)x•
t
k 1
=
(7) +
[
(k+
l)c: 1) +k(:) l x'
= c, +at (:)x• = c, [1 + t (:)x'] = c,j(x),

because ( 7) = a and
(k+l)( : ) +k(:)
k l
a(a :-- 1) •••(a - k+l)(a - k) a(a - 1) ···(a - k+1)
= (k+ 1) +k
(k+1)! k!
_ a(a - 1) •••(a - k+1) a(a - 1) ···(a - k+1) = a
= (a k) k!
+k
k!
0( )
k .
Appendix C
Solutions
C.l Chapter 1 Solutions

Exercises 1.1 - page 7
1. y' + 2x = xex ⇒ y = J xex - 2xdx = xex - ex - x2 + C.
1
2. -y' = 4cos(2x) ⇒ y = J 4xcos(2x)dx = 2xsin(2x) - J 2sin(2x)dx
X
= 2xsin(2x) + cos(2x)+ c.
3. * = 2x cos(x2 ) ⇒ y = J 2xcos(x2 )dx = sin(x2 ) + c.
y = ± J 2xex dx = ±ex + c.
2
y' = ±2xex
2
(*)2 = 4x2 e2x
2 2
4. =} =}
5. co1(x ) y' = 2sin(x) =* y = J 2sin(x)cos(x)dx = J sin(2x)dx = -½ cos(2x) + c.
6. � = ln(l) ⇒ y = f ln(l) dt = t ln(t) - t + c, y(l) = 2 ⇒ c = 3.

7. y' = 4cos2 (x) ⇒ y = J 4cos2 (x)dx = J 2 + 2cos(2x)dx = 2x + sin(2x) + c,
y (�) = � =} C = -1.
8. y' = xL�3_6 y = ¼ J �3 - x!2 dx = ¼[6 ln Ix - 31 - ln Ix+ 21] + c.
=}
x
9. �! = 12e x sin(3x) ⇒ y = J 12e2x sin(3x)dx = 6e2x sin(3x) - J 18e2x cos(3x)dx
2
= 6e2x sin(3x) - 9e2x cos(3x) - J 27e2x sin(3x)dx =} J 39e2x sin(3x)dx

= 6e2x sin(3x) - 9e2x cos(3x) + c1 ⇒ y = 1� [6e2x sin(3x) - 9e2x cos(3x)] + c.
10. 1Jt = 3y ⇒ � = 3 ⇒ lnlvl = 3t + c1 ⇒ y = ce3t , y(O) = 2 ⇒ c = 2.
11. 1Jt = -2y ⇒ � = -2 ⇒ ln IYI = -2t + c1 ⇒ y = ce- 2 t, y(O) = 3 ⇒ c = 3.
12. y' + cos(x)y = 0 =* 'Ji..
y
= -cos(x) ⇒ lnlyl = -sin(x) + c1 ⇒ y = ce-sin(x),
y(1r) = -3 ⇒ c= -3.
347
348 APPENDIX C. SOLUTIONS
J
13. 3y2 y' = 1 ⇒ 3y 2 y' dx = 1 dx ⇒ 3y2 dy = J J J 1 dx ⇒ y
3
= x+c ⇒
y = (x + c) 1
1 3
= �x + c, y(O) = 2 ⇒ c = 8.
C.2 Chapter 2 Solutions

Exercises 2 .1 - page 14
1· y ' = 3x2 y2 ⇒ JL
y2 = 3x2 ⇒ _.!y = x3 + c ⇒ = y x--=.L
3+c' y(O) = .!2 ⇒ c = -2.
2. y
' = 7⇒y 3
y
' = Jx+1 ⇒ ¼Y4 = �(x+1)3 / 2+c 1 ⇒ = y
3 2 1 4
rn(x + 1) / + cJ / _
3. (x2 + 1)!� = 2;; 1 ⇒ 3y y' = ;;!�

⇒ y3 = ln(x2 + 1) + tan-1(x) + c
2
⇒
y = [ln(x
2
+ 1) + tan -1(x) + c] 1 1 3 , y(O) = 3 ⇒ c = 27.
y'
4. 4x 2 +3 - - 4 Y3 + - 4X 3 + 3X2 - 2X
x -2 - 5 4 -4y3+1 ::::} (5 Y4 1)YI -
X =}
_ y4 + y = x4 + x3 _ x2 + c.
y
5
5. xy' - ln� )
x
= 0 ::::} yy' = ln�x) ::::} ½Y2 = ½[ln(x)] 2 + C1::::} y = ± ✓[ln(x)] 2 + c,
y(l) = ⇒ c = 9 and y = - ✓[ln(x)] 2 + 9.
-3
6. xy ' = (x + l)eY ⇒ e-Y ' = x!l = 1 + � ⇒ -e-Y = x + ln lxl + c ⇒

y
y= - ln(-x - ln lxl - c).

? ' xy+x =} _]f__ ' = ___3:__ ::::} (1 __l_)y' = 1 + _1_ ::::}
=
· y
xy-y y +l
y x- 1 y+l x- 1
y - l n IY + 11 = x + ln Ix - 11 + c, y(2) = -2 ⇒ c = -4.
8. cos(x) ln(y)� = y sin(x) ⇒ In�y) y' = tan(x) ⇒ ½[ln(y)] 2 = ln I sec(x)I + c 1 ⇒

✓
ln(y) = ± 2 ln I sec(x)I + c::::} y = e ± J2 Inlsec(x)l+c , (O) = e2 ::::} c = 4 an d
y
= J2 Inlsec(x)l+4_
Y e
' = � ⇒ yy' = x2 ⇒ ½y 2 = ½x3 +c1 ⇒

2
9. y y = ±j�x3 + c is the general solution.
The orthogonal trajectories:
y' = --;z::::} �
= - ;2 ⇒ ln IYI = � + C1 ⇒ Y = ce l /x_
10. y
' = 2x(y2 + 1) ⇒ )'� 1 = 2x ⇒ tan- 1 (y) = x2 + c ⇒ y = tan(x2 + c) is the
general solution. The orthogonal trajectories: y' = 2x (;l+ l ⇒
)
(y2 + l)y' = ;; ⇒ ½Y3 + y = -½ ln lxl + c1, or 31n lxl + 6y + 2y3 = c.
11. y = x3 + c ⇒ y ' = 3x2 . The orthogonal trajectories: y ' = ;\ ⇒ = y 3� + c.

12. y = kx4 ⇒ x- 4 y = k ⇒ -4x- 5
y + x-4 y' = 0 ⇒
= ; . The orthogonal
y'
trajectories: y' = �: ⇒4 yy ' = -x ⇒ 2y2 = -½x2 + c1 ⇒ = ±Jc - ¼x2 . y
C.2. CHAPTER 2 SOLUTIONS 349
13. y = ecx > 0 ⇒ x- 1 ln(y) = c ⇒ -x - 2 ln(y) + x- 1 - 1 ' =0 y y

⇒ y
'= Y 1 :(y). The
orthogonal trajecto ries: ' = yl�(y) ⇒ ln( ) '=-x ⇒
y y y y
½y 2 ln(y) - ¾Y 2 = -½x 2 + c1, or 2x 2 - y 2 + 2y 2 ln(y) =c.
Exercises 2. 2 - page 18
+ ⇒ ⇒
1. y
' = x X y =1 + '11.X, u = 'Ji.
X
⇒ u + xu' = 1 + u ⇒ u'= X1 u = ln /x/ + c
y
=x(ln /x/ + c) , y(l) = 2 ⇒ c= 2.
2. x '= y2 - x 2 ⇒ '= ;; - �, u = ;; ⇒ u +xu' = u - ⇒ uu'=- ¼ ⇒
yy y
t
½u 2 =- ln/xl +c1 ⇒ u = ±Jc-2 ln/x/ ⇒ = ±xJc-2ln /x/, y(-1) = 2 y
⇒ c =4 and y =-xJ4 - 21n Ix/ .

3· ./Y + x'Jl. ' u =x'11. ⇒ u + xu' = _!_
' = ,/x ,/u +
u ⇒ u1 l 2 u'=x.!. ⇒ 3lu 3! 2 =ln/xi+c1 ⇒
rn
y
u= (� ln/xi+ c)213 ⇒ y =X ln /xi+ c)213, y( l) =9 ⇒ C=27.

4 . y '- '11.+ x - 'Jl.+ l _ ___,__
' U - 'Ji. -----,,, U
+ XUt_
- U
+ 1/3 1 1/ 2 1 -----,,,
____._
-
x y
3� x 2 y+�+x
-
x 3 !Ti"
V �+y �+l
/Ti x u +u +
(u 113 +u112 +1)u'= ¼ ⇒ ¾u413 +�u 1 3 2 +u=ln
/x/ +c ⇒
¾ (;;)'1 1 3 ( 3 12
+� ;;) +;; =ln /xi + c.
5. y
+xsec (;;)- x ' =0 ⇒ ' =;; + sec(;;), u=;; ⇒ u +xu'=u +sec(u)
y y
⇒ cos(u)u'= ¾ ⇒ sin(u) =ln /xi+c ⇒ u=sin- 1 (ln /x/ + c)

⇒ y = x sin- 1 (ln /xi+ c), y(l) = f ⇒ c = {!-.
6. y
'=
.
2 x+3y
3 X- 2 y
2
= 3 +32 : ) u = XV.
- X
⇒ u +xu' = 2 + 3u
3 - 2U
⇒ xu'= 32 + 2 u ⇒
- ,._
2
()U
3- 2 �
1 +u
u'= 'J.X ⇒
3tan- 1 (u) - ln(l +u2 ) = 21n/xi + c ⇒ 3tan- 1 (;;) - ln (1 + ;�) =21n/xi+c.
8. yy' = 6x + ⇒ ' = 5yx + 1 ) u ='Ji.

y
X
⇒ u +xu' = QU + 1 ⇒ xu' = 5+uU-u2 ⇒
y
_u_ u' = - 1 ⇒ (1 -1- -1-) u' = - 1 ⇒ 1 ln / u-3/+l ln/ u+21 =-ln /x/+c
2
u - u-6 x 5 u-3
+ 5l u+ 2 x 5 5
⇒ }lnl;;-31 +glnl;; +21 =-ln/xl +c, y(l) = 4 ⇒ c=gln(6).
l. y'+y = e-x, I(x) = efldx =ex ⇒ (ex y )'= 1 ⇒ ex y
=x+c ⇒ y
=(x+c)e-x ,
y(O) = -2 ⇒ c = -2.
2. y
' _ 3 =1, J(x) =ef-3dx = e- x3 ⇒ ( e- 3 x y )' = e- x3 ⇒ e-3x = -:}e-3x + C
y y
=?
y = ce x3 - ½-
3. exy ' + 2ex y 1 - = 0::::} y' + 2y = e-x , I(x)

⇒ =
⇒
ef 2dx = e2x (e2x y)' =ex =}
::::}
e2Xy = ex + C y = e-x + ce-2x ) y(O) = -3 C = -4.
4. y' + 2xy = 2x, I(x) = ef2xdx = ex

2
⇒ ( )' =
ex
2
y 2xex
2
⇒ ex
2
y = ex
2
+C ⇒
x2
y = l+ ce- .
5. xy' + y = � =} + � y = x\, I(x) = ef½ dx = e1n( x) =

⇒ ⇒
y
' x =} (xy)' 1.X ⇒
xy = ln(x) + c y = �[ln(x)+ c], y (l) = 2 c = 2.
7. x2y'+xy =- 1- l. -1-
In( x) ::::} y'+ x y = x2 ln( x) ' I(x) = ef ½ dx = e1n( x) = x ⇒ =- 1-
⇒ ⇒
(xy)' x ln (x )
xy = ln I ln(x)I+ c y = �(ln I ln(x)I + c).
8. cos(x)y' + sin(x)y = cos4 (x)

I(x) = ef tan( ) =e (
⇒
y' + tan(x)y = cos (x),
3
I = I sec(x)I = ±sec(x), and I(x) = sec(x)

⇒
x dx 1n lsec x )
2
[scc(x)y]' = cos (x)
J J
⇒ sec(x)y = cos2 (x) dx = ½ l+ cos(2x) dx = ½+ ¼ sin(2x)+ c
⇒y = cos(x) [½+ ¼ sin(2x)+ c], y (1r) = � c = -1r. ⇒
9. xy' + 2y = e-x =} y' +� y = e x", I(x) = ef f dx = e2Inlxl = lxl2
(x2y)' = xe-x ⇒
x2y = -xe-x e-x + c - ⇒ y = ;2 (c - e-
x
xe-x).-
10. xy' - = x�l ::::} y' - � = xX:1 , I(x) = ef -½ dx = e-In lxl = l;I = ±�,
⇒¾
y
= x - ln Ix+ 11 + c ::::}
y
and I(x) = ¾ ::::} (¾ y )' = x:i = 1 - x!i y

y = x ( x - ln Ix+ l I+ c).
12. X ln(x)y' + Y = x;:_ l ::::} y' + x l;(x) Y = ( x2 12) ln( x) '

-
J(x) = ef xlt�(x) = e1n l ln( x )I = lln(x)I = ±ln(x), and I(x) = ln(x)
dx
::::} (ln(x)y)' = x2� 1 = x�l x!i ::::}

-
ln(x)y = ln Ix - 11 - ln(x+ 1)+ c ::::}
Y = ln(x) [ln Ix - 11 - ln(x+ 1)+ c].
13. 3y' - y = y\ is a Bernoulli equation with a = -2. u = y 1-0 = y3 y = u 11 3 ⇒

⇒ l 213u'
y' = 31u- ⇒
u-213u' u 1 13 = u-2/ 3 - u = l ) which is linear , ⇒ -
⇒ ⇒ ⇒
u'
and I(x) = e- (e- u)' = e- x e- x u = -e-x+ c u = cex -
⇒
x x
1 ::::}
y = u
1 13
= (ce x -
1 13
1) , y(O) = 2 c = 9.
14. y' + 4y = 2e-x .jy is a Bernoulli equati on with a ½- u = y 1 - 0 = y 1 1 2

⇒ y = u ⇒ y ' = 2uu' ⇒ 2uu' + 4u = 2e- u ⇒ u' + 2u = e-x , which is
2 2 x
linear , and J(x) = e 2x ⇒ (e2x u)' = ex ⇒ e2x u = ex +C ⇒ U = e-x + ce-2x ⇒

y = u2 = (e-x + ce-2x )2 .
15. 2x y' - 2xy = -y sin(x) is a Bernoulli equati on with a = 3. u = y - = y-

2 3 1 0 2
⇒ y = u-1/ ⇒ y' = -½u- 1 u' ⇒ -x u- 1 u' - 2xu- ! = -u- 1 sin(x) ⇒

2 3 2 2 3 2 1 2 3 2
x2 u' +2xu = sin(x) ⇒ (x2 u)'=sin(x) ⇒ x2 u = - cos (x) + c ⇒ u = c-:i ( ) ⇒

x
y = u- 1 /2 = ± 2 = 7r ⇒ C = _!4 and y = ✓¼-cos(x) = ✓l-4cos(x) ·

x x 2x
✓c-cos(x)' y ('!!_)
16. xyy' + y - xe = 0 ⇒ xy' + y = e is a Bernoulli equati on with a = -1.
2 x x x
u = y l -a = y2 ⇒ y = u l /2 ⇒ y' = ½u- 1/2 u' ⇒ �u-1f2u' +u l /2 = xex u-1/2 ⇒

u' + �u = 2ex , I(x) = e2 1nl x l = x2 ⇒ (x2 u)' = 2x2 ex ⇒
x2 u = J 2x2 ex dx = 2x2 ex - 4xex +4ex + c ⇒ u = x\ (2x2 ex - 4xex +4ex + c) ⇒
y = u 112 = ±� ✓2x2 ex - 4xex + 4ex + c.
17. xln(x)y' +y = 2y.,Jy is a Bernoulli equati on with a= lu = y 1 - 0 = y- 1 / 2 ⇒
y
= u-2 ⇒ y' = -2u- 3 u' ⇒ -2xln(x)u-3u'+u-2 = 2u-3 ⇒ u' __ u_ = �
2x l n(x) x ln(x)'
I(x) = ef-2xl�( x ) d = e-½ ln(ln(x)] = [ln(x)J-1/2 ⇒ {[ln(x)J-l/2 u}' = _ (In(xr 1 ⇒
x 3 2
[ln(x)J-1l2 u = 2[ln(x)J-1l2+c ⇒ u = 2+cJ!n(x) ⇒ y = u- 2 = [ 2+cJ!n(x)J-2.

1. J(x,y)=x3 - y2 +x2 y 2 +1.
(a ) f(l,0) = 2, f(3,1) = 36, f(l,3) = 2, f(2, -1) = 1 2.

(b) fx (x,y) = 3x2 +2xy2 ,
fx (l,0) = 3, fx (3,1) = 33, J�(l,3) = 21, fx (2, -1) = 16.
fy (x,y) = -2y +2x2 y,
]y (l,0) = 0, fy (3,1) = 16, fy (l,3) = 0, fy (2,-1) = -6.
(c) fxx (x,y) = 6x + 2y2 ,
fxx (l,0) = 6, fxx (3,1) = 20, fxx (l,3) = 24, fxx (2, -1) = 14,
fxy (x,y) = fyx (x,y) = 4xy,
fxy (l,0) = 0, fxy (3,1) = 12, fxy (l,3) = 1 2, fxy (2, -1) = -8.
fyy (x,y) = 2x2 - 2,
f�y (l,0) = 0, fyy (3,1) = 16, fyy (l,3) = 0, fyy (2, -1) = 6.
2. f(x,y) = x2 ln(y) - y 3 ex +x3 - y2 +1.
fx = 2xln(y) - y3 ex +3x2 , fy = xy - 3y2 ex - 2y,
2
fxx = 2ln(y) - y3 ex +6x, f�y = fyx = 2: - 3y2 ex , fyy = -:� - 6yex - 2.
3. J(x,y) = x4 + 3x2 y2 +y4 , x(t) = t2 +2t, y(i) = t3 - 2t.

ftf(x(t),y(t)) = fx �� +J�� = (4x +6xy2 )(2t +2) +(6x2 y +4y3 )(3l 2 - 2).
3
4. f(x,y) = x2 -cos(y) + xsin(y), x(t) = 2t2 + 3t - 1, y(l) =et + 2t.

1tf(x(t),y(t)) = [ 2x + sin(y)](4t + 3) + [sin(y) + xcos(y)](e t + 2).
5. f(x,y) = sin(xy) + �' x(t) =et + 1, y(t) = ln(t).

1tf(x(t),y(t)) = [ycos(xy) + 'Jx] e + [xcos(xy) + 1] ½-
2
t
2
6. J(x,y) = x4 + 3x2 y2 + y4, y(x) = x3 + 6x2 - x.

fxJ(x,y(x)) = fx + f/!;; = 4x + 6xy + (6x y + 4y )(3x + 1 2x - 1).
3 2 2 3 2
7. J(x,y) = eY - x2 + y3 ln(x), y(x) = sin(x).

fxJ(x,y(x)) = -2x + Y: + [eY + 3y ln(x)] cos(x).
2
8. f(x,y) = x2�y2 + xe- Y, y(x) = sin(2x) .

d!J(x,y(x)) = (x2-:;_�)2 + e-Y+ 2
[(x2�;2)2 - xe- Y ] [2cos(2x)].
9. f ( X' y) = X - y - 1.
4 4
dy
f(X ' y) = 0 =} Jx + fY ddxy = 0 = - fyfx = - -4y = xy3 .
4x 3 3
=}
dx 3
10. f(x,y) = x3 + y2 - eY.

J(x ' y) = 0 ⇒ fx + fY :EL.=
dx 0 ⇒ :EL.=
dx _b.
fy
= -�.
2y -eY

1. 3x2 + y + 1 - (3y2 - x + l)y' = 0, Py = 1 = Qx ⇒ the equation is exact.
fx = P = 3x + y + 1 ⇒ J(x,y) = x + xy + x + g(y),
2 3
fy = Q =} X + g'(y) = X - 3y2 - 1 ::::} g'(y) = -3y2 - 1

⇒ g(y) = -y3 - y + c ⇒ f ( x,y) = x3 + xy + x - y3 - y + c, and the general
solution is J(x,y) = c 1, or x3 + xy + x - y3 - y = k, where k = c 1 - c.
2. xy + x + 1 + (½x2 + y + 1) :t
= 0, Py = x = Qx ⇒ the equation is exact.
fx = P = xy + x + 1 ⇒ J(x, y) = ½x2 y + ½x2 + x + g(y),
fy = Q ⇒ ½x + g'(y) = ½x + y + 1 ⇒ g'(y) = y + 1 ⇒ g(y) = ½Y 2 + y + c
2 2
⇒ f(x,y) = ½x2y + ½x2 + x + ½Y 2 + y + c =} ½x2 y + ½x2 + x + ½Y2 + y = k.

3. 3x2 y2 - 2x + (2x3 y + 3y2)y' = 0, Py = 6x2 y = Qx ⇒ the equation is exact.
fx = P = 3x2 y - 2x ⇒ f(x,y) = x3 y2 - x + g(y),
2 2
fy = Q ⇒ 2x3 y + g'(y) = 2x3y + 3y2 ⇒ g'(y) = 3y2 ⇒ g(y) = y3 + c

⇒ f(x,y) = x3y2 - x2 + y3 + c ⇒ x3y2 - x2 + y3 = k.
4. y3 -3x2 y2+4x3 +(3xy2 -2x3 y-4y3)y' = 0, Py = 3y2 -6x2 y = Qx ⇒ the equation
is exact. fx = P = y3 - 3x2 y2 + 4x3 ⇒ J(x,y) = xy3 - x3 y2 + x4 + g(y), fy = Q
⇒ 3xy2 - 2x3 y 3+ g'(y) = 3xy2 - 2x3 y - 4y3 ⇒ g'(y) = -4y3 ⇒ g(y) = -y4 + c
⇒ f(x,y) = xy - x y + x4 - y4 + c ⇒ xy3 - x3 y2 + x4 - y4 = k. y(0) = 1 ⇒
3 2
k = -1.
u- xy3 ⇒ 2x -3x
5 · y' = 3xx2uy2--2x
2 2
y + y + (3x y - 2x y + x)y' = 0 is exact because
3 2 3 2 3 2 2 2
3 +x y
y
Py = 6xy2 - 6x2 y + 1 = Qx . fx = P = 2xy -3x y + y =} 3 2 2
J(x,y) = x 2 y3 - x 3y2 + xy + g(y), fy = Q ⇒

3x2y - 2x3 y + x + g'(y) = 3x y - 2x y + x ⇒ g'(y) = 0 ⇒ g(y) = c ⇒
2 2 2 3
J(x,y) = x y - x y + xy + c ⇒ x y - x y + xy = k. y( 2) = 1 ⇒ k = -2.
2 3 3 2 2 3 3 2
6. /}x + y + ( 21 + x) �� = 0, P y
=
4)xu
⇒ the equation is exact.
+ 1 = Qx
fx = P = /'}x +y ⇒ f(x,y) = ft;y+xy + g(y), fy = Q ⇒
21 +x
+ g'(y) = 21 +x ⇒ g'(y) = 0 ⇒ g(y) = c ⇒ f(x,y) = ft;y+xy +c
⇒ ft;y + xy = k. y( 1) = 4 ⇒ k = 6.
7. (6xy + eY)y' + xex + 3y2 = 0, Py = 6y = Qx ⇒ the equation is exact.
fx = P = xe + 3y ::::} J(x,y) = xe - e + 3xy + g(y), fy = Q =}
x 2 x x 2
6xy + g'(y) = 6xy + eY ⇒ g'(y) = e Y ⇒ g(y) = e Y + c

=} J(x, y) = xe -e + 3xy + eY + c ⇒ xe - e + 3xy + eY = k.
x x 2 x x 2
8. 2xe x +Y + 4x + ( 2yex - 2y) � = 0, Py = 4xye x +Y = Qx ⇒ the equation

2 2 2 2 2 2
3 +Y
is exact. fx = P = 2xex +Y + 4x3 ::::} J(x,y) = ex +Y + x4 + g(y), ] = Q

2 2 2 2
y
::::} 2yex +y + g'(y) = 2yex +y - 2y ::::} g'(y) = -2y ::::} g(y)
2 2 2 2
= -y2 + c ::::}
J(x,y) = ex +y + x4 - y2 + c::::} e x +y + x4 - y2 = k.
2 2 2 2
9. ex+y +x+y+cos(x)+(ex+y +x-y)y' = 0, Py = ex+y +l = Qx ⇒ the equation is

exact. fx = P = ex+y +x+y+cos(x) ⇒ J(x,y) = ex+y +½x2+xy+sin(x)+g(y),
fy = Q =} e +y +x +g'(y) = ex+y +x- y ⇒ g'(y) = - y =} g(y) = -½y2 +c::::}
x
J(x,y) = e +½x2 +xy+ sin(x)- ½y2 +c ⇒ ex+y + ½x2 +xy+sin(x)-½y2 = k.

x +y
- 0 , Fsy -
- (x2+y2
10 · x2 x2 1 - 2X· + ( x2 Y1P l + 2Y ) ddxy - -2xy - Qx =} the equaf10n
+y + + + + 1) 2 -
is exact. fx = P = x2 +:2 + 1 - 2x ::::} J(x,y) = ½ ln(x + y + 1) - x + g(y),

2 2 2
fy = Q =} x2 +y2 + 1 + g'(y) = x2 y2 +l + 2y ::::} g'(y) = 2y ::::} g(y) = y + C::::}

2
+
J(x,y) = ½ ln(x2 + y2 + 1) - x2 + y2 + c ⇒ ½ ln(x2 + y2 + 1) - x2 + y2 = k.
y(O) = 0 ⇒ k = 0.
-(2x+ 2 )(2y-3)
11 · x2 +y22+x2+x-
2
+ l + x2 +y22+y2-x-
3
_ 0' r
<!3J.. -
3y +3 x
Dy -
_
(x +y2 + 2x- 3y +3)2
= Qx =} the
3y +3
-
2
. . p - d
2 x+ 2
equat10n 1s exact. fx - - x2+ 2 +2x-3 +3 + 1
⇒ J(x,y) = ln(x2 + y2 + 2x -2 3y + 3) + x + g(y), fy = Q ⇒
y y
2 y- 3 I _ y- 3 _
x2 +y2 +2x-3y+3 + 9 ( Y ) - x2 +y2 +2x-3y+3 ::::} gI ( Y ) - 0 =} 9 ( Y ) - _ C =}
J(x,y) = ln(x +y + 2x - 3y + 3) + x + c ⇒ ln(x + y + 2x - 3y + 3) + x = k.
2 2 2 2
12. J x2 2 + 1 + ( J 2y 2 + 1) ddxy = 0.
x y x +y
+
(b) fx = = �P x +y +
1 ⇒ f(x, y) = Jx2 + y2 + x + g(y), fy =
2 2
Q ⇒
+ g'(y) = � x +y2
+ 1 =? g'(y) = 1 =? g(y) = y + C
� 2
⇒ f(x,y) = Jx2 + y2 + x + y + c ⇒ Jx2 + y2 + x + y = k.

J
(c) f(x, y) = x2 + y2 + x + y + c is a potential function, hence the equation
is exact in R, even though the region R is not simply connected.
13. � - x2:y2 y' = 0.

2 2
(a) Py =
(:2 ;;2 ) 2 = Q x on R = {(x,y): (x,y) =/= (0,0)}.
(b) fx = P = x !y * f(x,y) = J x2!y2 dx = J (c£)\+1 i dx = tan-1(;) + g(y),

2 2
fy = Q =? (:)1+1 ( - ;2) + g'(y) = x2--;!y + g'(y) = - x2:y2 =? g(y) = C =?

2
J(x,y) = tan-1 (;) + c ⇒ tan-1 (;) = k ⇒; = tan(k) = k1.

(c) f(x, y) = tan- 1 (:E) + c is multi-valued in R, hence not a (well-defined)
y
function. Hence, the equation is not exact in R. This is possible because
R is not simply con.nected, despite the fact that Py = Qx in R.
14. x + 6y2 + 4xyy' = 0, Py = 12y, Q x = 4y, Py =/= Qx ⇒ the equation is not exact.
Py Q
Q x = 4
8;Y =
� is independent of y ⇒ an integrating factor I(x) exists, and
�(�i = � ⇒ ln II(x)I = 21n lxl ⇒ I(x) = ±x and, with I(x) = x , the equation
2 2
becomes x3 + 6x2 y2 + 4x3 yy' = 0, which is exact. Then fx = x3 + 6x2 y2

⇒ J(x,y) = ¼x4 + 2x3 y2 + g(y), fy = 4x3 y + g'(y) = 4x3 y ⇒ g(y) = c ⇒
= ¼x4 + 2x3 y2 + c ⇒ = =0⇒ k=4 ⇒ = �;3
4 1 x4
f(x,y) ¼x4 + 2x3 y2 k. y(2) y2
⇒ y = ±J;3 - �-
15. x3 + 2y2 + xyy' = 0, Py = 4y, Qx = y, Py =/= Qx ⇒ the equation is not
exact. PyQQx = � = � is independent of y ⇒ an integrating factor I(x) exists,
and �(�i = � ⇒ ln II(x)I = 31n lxl ⇒ I(x) = ±x3 . With I(x) = x3 , the
equation becomes x6 + 2x3 y2 + x4 yy' = 0, which is exact. Then Jx = x6 + 2x3 y2
⇒ f(x,y) = ½x7 + ½x4 y2 + g(y), fy = x4 y + g'(y) = x4y ⇒ g(y) = c ⇒
f(x, y) = ½x7 + ½x4 y2 + c ⇒ ½x7 + ½x4 y2 = k.
16. e- x -cos(y) +sin(y)y' = 0, Py = sin(y), Qx = 0, Py =/= Q x ⇒ the equation is not

exact. Pv-;}x = :;�i�\
= 1 is independent of y ⇒ an integrating factor I(x) exists,
and � (:} = 1 ⇒ ln II(x)I = x ⇒ I(x) = ±ex . With I(x) = ex , the equation
(
becomes 1-ex cos(y) + ex sin(y)y' = 0, which is exact. Then fx = 1- ex cos(y)
⇒ f(x, y) = x -ex cos(y) + g(y), fy = ex sin(y) + g'(y) = ex sin(y) ⇒ g(y) = c
⇒ f(x, y) = x - ex cos (y) + c ⇒ x - ex cos(y) = k.
17. xex + xy2 + yy' = 0, Py = 2x , Q x = 0, Py =/ Q x =} the equation is not exact.

2
Py Qx =
Q � = 2x is independent of y =} an integrating factor I(x) exists, and
� �i
( = 2x ⇒ ln II(x)I = x =} I(x) = ±e . With I(x) = e , the equation
2 x2 x2
becomes xe2x + xex y2 + ex yy' = 0, which is exact. Then fx = xe2x + xex y2

2 2 2 2 2
=} f(x, y ) = ¼e
2x2
+ ½ex y2 + g(y ), fy = ex y + g'(y ) = ex =} g(y) = c =}
2 2 2
f(x, y ) = ¼e2x + ½ex y2 + c ⇒ ¼e2x + ½e x2 2 = k.

2 2 2
y
18. jx + y2 + 4xyy' = 0, Py = 2y, Q x = 4y , Py =/ Q x =} the equation is not exact.

Py Qx =
Q
�;t
= ;; is independent of y =} an integrating factor J(x) exists, and
�(�] = ;; =} ln II(x)I = -½
ln(x) =} I(x) = ±x-112. With I(x) = x-112, the
equation becomes 1 + x-112 y2 + 4x 112 ' = 0, which is exact. Then
yy
fx= 1 + x-1/2 y2 =} f(x, y) = x + 2x 1/2y2 + g(y), fy = 4x l f2y + g'(y) = 4x l f2 y
⇒ g(y) = c =} f(x, y) = x + 2x 1l2 y2 + c ⇒ x + 2-jx y2 = k.

19. x + y2 + 3xyy ' = 0, Py = 2y, Q x = 3y, Py =/ Q x =} the equation is not exact.
PyQQx = 3�� = 3; is independent of y ⇒ an integrating factor J(x) exists, and
-½
I '
;(�i = 3; ln II(x)I = ln lxl =} J(1;) = ±x- 1 13. With J(x) = x- 1 13, the
1
=}
equation becomes x213 + x-1 13y2 + 3x213 ' = 0, which is exact. Then
yy
fx = x2/3 + x-I/ y 2 =} f(x, y) = ¾x5/3 + �x2f3y2 + g(y ),

3
j� = 3x f y + g'( ) = 3x f3y =} g(y ) = c =} f(x, y ) = ¾x5/3 + �x2f3y 2 + c =}

23 2
= k, or 2x513 + 5x213y2 = k1, where k1 = 130 k.

y
J5 x + J2 x
5 13 21 23
20. 4xy + (6x2 + )y ' = 0, Py = 4x, Q x = 12x, Py =/ Q x =} the equation is not

y
exact. PyQQx = 6;;}�y is not independent of y =} J(x) does not exist. However,
Qx /Y = 8: = � is independent of x =} an integrating factor I( ) exists,
; 4 Y y
and � �] = � =} ln II(y)I = 2 ln IYI =} I(y ) = ±y . With J( ) = y2, the

2
y
(
equation becomes 4xy + (6x2 y2 + 3 ) ' = 0, which is exact. Then fx = 4xy3
3
y y
=} J(x, y ) = 2x y + g(y ), f = 6x y + g'(y ) = 6x y + y =} g'(y ) = y

2 3 2 2 2 2 3 3
y =}
2 3 ¼ 4
g(y ) = ¼Y +c =} f(x, y ) = 2x y + y +c =} 2x y +¼Y = k, or 8x2 +y4 = k1 .
4 2 3 4 3
y(l) = 1 ⇒ k1 = 9.
21. sin(x) + [e- Y - cos(x)]y' = 0, Py = 0, Q x = sin(x), Py =/ Q x =} the equation

is not exact. Qx;Py = ::��;\ = 1 is independent of x =} an integrating factor
I(y) exists, and ��i
( = 1 ⇒ ln II(y)I = y ⇒ J(y) = ±e . With I(y) = e , the
Y Y
equation becomes sin(x)eY + [1 - cos(x)eY ]y' = 0, which is exact. Then

fx = sin(x)eY =} J(x, y) = - cos(x)eY + g( ), y
fy = - cos(x)eY + g'(y) = 1 - cos(x)eY =} g'( ) = 1 =} g(y) = y + c =} y
f(x, y) = - cos(x)eY + y + c =} - cos(x)eY = k. y
22. y
+(1+2x+xy)y ' = 0, Py = 1, Q x = 2+ , Py =/ Q x =} the equation is not exact.
y
Qx /Y =
; � = 1 + t is independent of x =} an integrating factor I (y) exists,
and �(�i = 1 +; ⇒ ln JJ(y)J= y +ln JyJ ⇒ J(y) = ±ye Y . With J(y) =ye Y, the
equation becomes y2 e Y +ye Y (l+2x+xy)y1 = 0, which is exact. Then fx = y2 e Y
⇒ f(x, y) = xy2 e Y + g(y), fy = (xy2 + 2xy)e Y + g1 (y) = ye Y (l + 2x + xy)
⇒ g1 (y) = ye Y ⇒ g(y) = ye Y - e Y + c ⇒ f(x,y) = xy2 e Y + ye Y - e Y + c ⇒
(xy2 + y - l)e Y =k.
23. : = x �«�)�� x ::::} yln(y) + [ xln(y)+ 2x]y1 = 0, Py= ln(y) + 1, Q x =ln(y) + 2,

P
Py =/- Q x ⇒ the equation is not exact. Qx; y = l;( ) is independent of x ⇒
y y
an integrating factor J(y) exists, and j(�i = yl;( y) ⇒ ln JI(y)J =ln Jln(y)J ⇒
I(y) = ± ln (y). With J (y) = ln(y), the equation becomes
y[ln(y)] 2 + {x[ln(y)]2 + 2xln(y)}y =0, which is exact. Then fx = y[ln(y)] ⇒
1 2
f(x,y) = xy[ln(y)] 2+g(y), fy= x[ln(y)] 2+2xln (y)+g (y) =x[ln(y)]2+2xln(y)

1
⇒ g(y)= c ⇒ f(x, y)=xy[ln(y)] +c ⇒ xy[ln(y)] = k.

2 2
Chapter 2 Exercises - page 38

2
1. (a) xy + y2 - x 2 y1 = 0 ⇒ y1 = ; + (;) , u = ; ⇒ u + xu1 = u + u2 ⇒
2 1 --'-
U - U I = X - -,r -U
-1 = 1 n JX J + C U=I -
1
::::} Y = lxl+c.
-x
=} n lxl+c In
(b) xy + y2 - x 2 y1 = 0 ⇒ y1 - �y = ;;-, u = y- 1 ⇒ y = u-1 ⇒ y1 = -u-2 u1

2
::::} -u-2 u1 - ¾u- 1 = u �2 ⇒ u1 + ¾u = ;J, I(x) = ef¾d x = eln lxl = JxJ,

x
I(x) = x ⇒ (xu) = -l ::::} xu = -ln JxJ + c⇒ u = - l l l ::::} y = �.
1 n x +c
x x In lxl-c
2. (a) y =0 ⇒ _!_
y2 y = _!_ ⇒ -y- 1 = -x-l +c::::} y = _
2
1 1 _
x
.
'1/__ -
x2 x2 1-cx
y = 0 ⇒ y = (�) , u =� ⇒ u+ xu = u ⇒ xu =u - u ⇒
2 2 1 2 1 2
(b)
1 1
:t'
;2 -
( u �l - t) u = � ⇒ ln Ju - lJ - ln Jul=ln JxJ + c⇒ 1 =c1JxJ⇒

1 1
1 - l_ k __.__ _ 1 __.__ _
y -
x
�- X ----,- U - 1-kx ----,- 1-kx.
(c) ;� - y = 0⇒ y = ;�, u =y- 1 ⇒ y =u- 1 ⇒ y =-u- u ⇒ -u- u = ;/

1 1 1 2 1 2 1 u
=? U = x2 =? U = .!
x +C
::::} y = _x_·.
1 -l
l+cx
y2 ( ) _!_ I = 0 is
y2 ⇒ _!_
Qx-Py
(d) x2
- y = 0,
l ⇒ I'J( yy) = -2
p = -2
yy
⇒ I(y) = _!_ x2 - y2 Y
exact. fx = }2 ⇒ f(x,y) = � + g(y), fy= g (y) = - Y\ ⇒ g(y) =; +c
1 1
k
=} y - x = ⇒ Y = l+kx ·
1 1 x
(e) The orthogonal trajectories are given by y

1
= - yx� ⇒ y y
2 1
= -x2 ⇒
½Y 3 = -½x3 + c⇒ x3 + y3 =k.
3. (a) x + y - xy1 = 0 ==> y 1 = 1 + �' u = � =;, u + X'U1 = 1 + u ==> u' = � ==>
u =ln JxJ+c ⇒ y = x(ln JxJ + c).
(b) x + y- xy1 =0 =} y1 - �y = 1, I(x)= ef-fdx = e-ln lxl = JxJ- 1, J(x) = �
⇒ (�y) = �⇒ �y=lnJxJ+c⇒ y=x(ln JxJ+c).
1
(c) X + y- xy' = Q ' Py-QQx = -2 ⇒ II('(xx)) = -2 ⇒ J(x) = ..l..

X x
⇒ 1. + JL
Xx
- 1.y' = Q
2 X 2 X
is exact. fx = �+ ⇒ J(x, y) = ln lxl - �+g(y), fy = -�+g'(y) = -�

;2
⇒ g(y) = c ⇒ lnlxl- � = k ⇒ y = x (ln lxl - k).

(d) The orthogonal trajectories are given by y' = x�xy = 1 u = � ⇒ �¥,
u + xu' = __=._!_ ⇒ xu' = l+u ⇒ J -2'c±l._
-l-u-u
u +u+l ' dx = J x dx
-l
= -ln Ix! + c.
2
l+u
u 2
2
-2'c±1._ _ (u+2l+2 _ .! 2 (u+2) 73
l 1 1
u +u+l - (u+l)
2 2 +;1 - 2 (u+l) +;l
2
+ __.!._
J3 ( 73
2 u+ I ) 2 +1 ⇒
2 4 2 4 73
J u u+ = 2 1n [ ( u+2
l 1 1 2 3 1
) +4 ] + v'3
t an -1 1
+ v'3 ) ⇒
�J
+u+l du
2
v1:3U
2 (
.2! ln [(u+.!)
X 2
2
+4 +__1._
v'3 tan
-1 (2.u v'3 ) = -ln !xi+c.
v'3x+__1._
4. (a) 3xy'+ 2y = 0 ⇒ 1:;- = -lx ⇒ ln IYI = -� lnlxl+c ⇒ y = kx- 2!3_

(b) 3xy' + 2y = 0 ⇒ y' = - l3 1lx ' u = 1l ⇒ u + xu' = - lu 3
X
⇒ xu' = -Q3 u ⇒
!{ = _Q.!. ⇒ lnlul = _Q lnlxl+c ⇒ u = kx- / ⇒ y = kx- 2 13.
5 3
U 3 X 3
(c) 3xy'+ 2y = 0 ⇒ y'+3�y = 0 ⇒ I(x) = ef txdx = ei 1 nl x l = lxl2/3 = x213

⇒ (x 2f3y)' = o ⇒ x 2f3y = c ⇒ y = cx-2/3_
(d) 3xy' + 2y = 0 ' Py-QQx = -1 I'( x ) -1 I
3x ⇒ I( x ) = 3x ⇒ (x) = x
-1/3 ⇒
2x- 1 y+3x 1 y' = 0 is exact. fx = 2x- 1 y ⇒ f(x, y) = 3x213y+g(y),
1 3 2 3 1 3
fy = 3x 2 13 + g'(y) = 3x 213 ⇒ g(y) = c ⇒ 3x2 13y = k ⇒ y = k 1 x- 2 l3 _
(e) The orthogonal trajectories are given by y' = �: ⇒ 2yy' = 3x ⇒
y 2 = �x 2 + c ⇒ 3x 2 - 2y 2 = k.
5. 2x - 2e- x - y 2 =
k ⇒ 2+2e- x - 2yy' = 0 ⇒ y' = l+e x . The orthogonal
f
y
trajectories are given by y' = 1 .;/-x ⇒ = 1 �1 x = e��1 ⇒
ln IYI = -ln(ex+1)+C ⇒ Y = ex: 1 •
6. y = 2x lnlxl+ex ⇒ � - 2 ln lxl = c ⇒ -;2+� - � = 0 ⇒ y' = �+ 2. The

- 12 ⇒
orthogonal trajectories arc given by y' = 11.-+12, u = 1lX ⇒ u+xu' = u+
x
1 _ -l- 2u-u2
XU - ⇒ u+2 1 _ -1 ⇒
u+2
l
(u+l)2 U - x J u+l
+ 1 d _ -1 d ⇒ (u+l) 2 U - Jx X
lnju+11- u�l = -lnlxl+c ⇒ ln[� + 1[- x : y = -lnlxl+c.
7. J(x, y) = 0 defines y as a function of x. Hence, fx+fy �� = 0 ⇒ : = - � :
J(x, y) = 0 also defines x as a function of y. Hence, fx :+fy = 0 ⇒ : = - �:,
and <!:11.-..l...
dX - dx ·
dy
8. (a) 3x - 2y+(y - 2x)y' = 0 is exact since Py = -2 = Qx - fx = 3x - 2y ⇒

J(x, y) = �x 2 - 2xy+g(y), fy = -2x+g'(y) = y - 2x ⇒ g'(y) = y ⇒
g(y) = ½Y 2+c ⇒ �x 2 - 2xy+ ½Y 2 = k ⇒ y 2 - 4xy+3x 2 - k 1 = 0 ⇒
y = 2x ± Jx 2+k 1.
(b) 3x - 2y + (y - 2x)y' = 0 => y'= 2 v-3x 21L_3

.J__
u-3
= ll_ 2 ' u = Jf..x => u + xu' = 2u-
-2x X 2
2
+4u-3 => u-2 u = -1 =? -1- _1_
=> xu' = -u u-
y
- 2 =?
2 u2 -4u+3 ' x (u-1 + u-3 ) u' = x
lnlu-ll+lnlu-31 = -2ln lxl+c => (u-l)(u-3) = x\ => u2 -4u+3- x\ = 0
=>u=2±J1 +;2 =>y=2x± ✓x2 +k.
9. (a) xex - xln(x) + y + xy' = 0 is exact since Py = l = Qx , fv = x =>
f(x, y)=xy+g(x), fx=y+g'(x)=xex -xln(x)+y => g'(x) = xex -x ln(x)
=> g(x) � xex -ex -½x2 ln(x) +¼x2 +c=> xy+xex -ex -½x2 ln(x)+¼x2 = k
=>y = !:..... 2 ln(x) - 4 x + !E.
x 1
X - e + .!x X
(b) xex -xln(x)+y+xy' = 0 => xy'+y=xln(x)-xe =>(xy)' = xln(x)-xex
=>xy = ½x2 ln(x) - ¼x2 - xex +ex + c =>y = e: - ex + ½xln(x) - ¼x + �-
10. (a) g(y)y'= f(x) => f(x) - g(y)y' = 0 and [f(x)Jv = 0 = [- g(y)] x .
n y)
(b) y+ xln(x)ln(y)y'= 0 => l � y' = x l�(x) => ½[ ln(y)]2 = -lnlln(x)I + c =>
± ✓k-2 lnlln(x)I_
y=e
(c) ln(yy) y' = --=.!._
x ln(x)
=> - 1-
xln( x)
+ ln(yy) y'= 0 is exact. fx = - 1-
x ln( x ) =>
1 Y)
f(x, y)=ln I ln(x)I + g(y), fv = g'(y)= "� => g(y) = ½[ln(y)] + c =>
2
ln I ln(x)I + ½[ln(y)] 2=k.

11. (a) y' = J(ax +by +c), b-/- 0, u(x) =ax +by +c => u'=a +by'= a +bf(u)
=> a+t; (u) =1, which is separable.
(b) y' = (x + y + 1)2 , u= x + y + l =>u' = 1 + y'= 1 + u2 => 1��2 = 1 =>
tan-1(u)=x+ c =>u= tan(x + c) => y = u - x - l= tan(x + c) - x - l.
1 _ y'u+l y'u
C yI = Jx+iF°2' U=x +y - 2 ⇒ uI=1 +yI=1+ Ju - .ju ⇒ y'u+luI - 1
( ) 1 _
=> J � 1 du = J 1 dx = x + c, and v= fa =>

J }£_1 du = J -:;;h 2v dv = 2 J v - l + :;¼i- dv = v 2 - 2v + 2 ln(v + 1) =>
x + y - 2 - 2 ✓x + y - 2 + 2ln( ✓x + y - 2 + 1)=x+ c.
(d) y' = (x + y) ln(x + y) - 1, u = x+ y => u' = 1 + y' = u ln(u) => ul:(u)u' = 1
=> ln I ln(u)I = x + c => I ln(u)I = ex+c => ln(u) = kex => u = e ke =>
x
y = u - X = e ke - X.
X
12. (a) a(y) + [b(y)x - c(y)]y' = 0, Qx;Py = b(y�c:;(y) is independent of x; hence,

I (y) exists.
(b) a(y) + [b(y)x - c(y)]y' = 0 => a(y) + [b(y)x - c(y)];;, = 0 =>
a(y)x' + b(y)x = c(y), which is linear for x(y).
13. (a) y + [(2y2 + l)x - 2y]y' = 0, Qx; y = 2y => �(�i = 2y => I(y) = eY =>
P 2
2
ye Y +[(2y2 +1)x-2y]eY y'=0 is exact. fx=yeY => f(x, y)=xyeY +g(y),
2 2 2
2 2 2
fy = (2y2 + l)xeY + g'(y) = [(2y2 + l)x - 2 y]eY ⇒ g'(y) = -2 yeY ⇒
2
g(y) = -eY + c ⇒ (x y - l)eY = k.
2
(b) y+ [(2y +l)x - 2y]y' = 0 ⇒ y + [(2y2 +l)x - 2y]� = 0 ⇒

2
yx' + (2y + l)x = 2 y ⇒ x' + ( 2 y +

2
t)
x = 2. The integrating factor is
J(y) = eJ2+ n Y = ± y eY2, and I(y) = yeY2 ⇒ ( yeY2x)' = 2yeY2

Y dY = eY +111
Y l 2
⇒ y eY
2
x = eY2 + c ⇒ (xy - l)eY
2
= c.
14. +(3xy+2)y' = 0, Qx; y =; = �(�i ⇒ I(y) = y ⇒ y3+(3xy2+2 y)y' = 0
2 P
(a) y
is exact. fx = y3 ⇒ J(x, y) = xy3 + g(y), fy = 3xy2 + g'(y) = 3xy2 + 2y

⇒ g(y) = y + c ⇒ xy + y = k.
2 3 2
(b) y2 + (3xy + 2)y' = 0 ⇒ y2 + (3xy + 2)� = 0 ⇒ y

2
x' + 3yx = -2 ⇒
x' + �x = ;'1, J(y) = ef f y = e3 Inlyl = ± y3 , I(x) =y x)' = -2y ⇒
d 3
⇒ (y 3
y x = -y + c ⇒ x y + y = c.
3 2 3 2
1 5. 1 + (xy - x- 3 y3)y' = 0 ⇒ 1 + (xy - x-3 y3 )¾

= 0 ⇒ x'+ yx = y3 x-3, which is a
Bernoulli equation for x(y) with a= -3. u = x4 ⇒ x = u 114 ⇒ x' = ¼u- 314 u'
⇒ ¼u- f4u' + yu / = y3u- /4 ⇒ u' + 4 yu = 4y , I(y) = e f y dy = e y ⇒
3 l 4 3 3 4 2 2
' 2
( e2Y u) = 4y3 e2Y ⇒ e2Y u = J y2 • 4ye2Y dy = y2 e2Y - ½e2Y + c (by parts) ⇒
2 2 2 2 2
/4
u = y2 - ½ + c e-2y2 ⇒ x = (y2 - ½ + ce-2y2)1 .
16. (a) y' = 2/y=l ⇒ 2.J'i=r = 1 ⇒ jy=1 = x + c ⇒ y = (x + c)2 +1.

(b) The function y = 1 satisfies y' = 0 = jy=I and is a singular solution.
(c) At every point (x 0, 1) on the line y = 1, the line is tangent to the parabola
y = (x - x0)2 + 1 at its vertex. Hence, the line is the envelope of the
one-parameter family of parabolas.
= l-y2y y
' =1 ⇒ =f ✓l - y2
2
17. (a) (y')2 ⇒
±� y
= X +C ⇒ (x + c)2 + y2 =1
⇒ J(x, y, c) = (x + c) 2 + y2 -1 = 0.
(b) y = ±1 are two singular solutions.
(c) f(x, y,c) = (x+c)2 +y2 -l = 0 and fc (x, y,c) = 2(x+c) = 0 ⇒ y2 -1 = 0
⇒ y = ±1.
(d) The general solution defines a one-parameter family of circles of radius 1

with centres at (-c, 0) on the x-axis. At every point (x 0, ±1) on the lines
y = ±1, the lines are tangent to the circle centred at (x0, 0). Hence, each
line is an envelope of the family of circles.
(a) y - ✓m2 + 1 ⇒ y' -- m ⇒ (y') 2 +1 m2 +1 -

18. - mx ± ( y- xy')2 - m + -xm)2 1.
( mx ±�
(b) y = mx ± Jm2 + 1 ⇒ y - mx = ±Jm 2 + 1 ⇒ (y - mx)2 = m2 + 1 ⇒

f(x,y,m) = (y-mx) 2 -m2 - 1 = 0. fm (x,y,m) =-2x(y-mx)-2m = 0
⇒ x(mx-y) =m ⇒ x2 = (mx-y m2 = 2
)2 m�+1 ·
(c) x(mx - y) = m::::;, y = m(x;-l) ::::;,

m 2(x2-1 ) 2
= (m 2 + 1) (---=L _ 1_
2 2
Y = x2 m 2+1 ) = m 2+1 ·
(d) x2 + Y2 = m�:1 + m/+1 = 1 ⇒ g(x,y) = x2 + y2 - 1 = 0 is a singular
solution.
(e) The graph of the singular solution is the unit circle centred at the origin.
At every point on the circle, the circle is tangent to one of the lines defined
by the general solution. This can be confirmed as follows: x2 + y2 = 1 ⇒
2x + 2yy' = 0 ⇒ y' = -�. Thus, the slope of the tangent to the circle at
(xo ,Yo) is m1 = -�- The line y = mx ± Jm 2+ 1 goes through (x0, y0) if
and only if Yo = mxo ± Jm + 1. Then (y0 - mx0) 2 = m 2 + 1 and, solving
2
for m, we obtain m =-xo

YO
= m1 . Thus, the line is tangent to the circle at
(xo ,Yo)-
19. (a) ⇒ 'Ji.. = .!. ⇒ y = kx.

y' = Ji.X y X
(b) y = kx ⇒ y(0) = 0 for all values of k. Hence, y(0) = 1 is impossible. The

reason is that f(x,y) = � is not continuous at x = 0 and, hence, it is not
continuous in a rectangular region containing (0, 1).
20. ⇒ y- 1 13y' = 3x ⇒ �y2!3 = �x2 + c ⇒ y= (x2 + k) 312.
(a) y' = 3xy 1 ! 3
(b) y =(x2 + k) 3 1 2 and y(0) = 0 ⇒ k = 0 ⇒ y =x 3 .
(c) y = 0 is a singular solution.
=
(d) Both y= x3 and y 0 are solutions of the initial-value problem. Hence,
the solution is not unique. The reason is that, with j(x,y) = 3xy 1 1 3,
fv(x,y) = xy- ! 3 is not continuous at y = 0 and, hence, it is not continuous
2
in a rectangular region containing (0, 0).

Exercises 3 .1 - page 46
1. y" + 6x2 + sin(x) = 2 ⇒ y" = 2 - 6x2 - sin(x) ⇒ y' = 2x - 2x3 + cos(x) + c1
⇒ y = x2 - ½x4 + sin(x) + c1x + c2.
2. xy" = ln(x) ⇒ y" = In� ) ::::;, y' = ½[ln(x)] 2 + c1 ::::;, y = ½ f[ln(x)] 2dx + C1X + C2.
x
J J
f[ln(x)] 2 dx = 1 · [ln(x)] 2 dx = x[ln(x)] 2 - 2 ln(x)dx
= x[ln(x)] 2 - 2x ln(x) + J 2dx = x[ln(x)] 2 - 2x ln(x) + 2x ⇒
y = ½x[ln(x)] 2 - x ln(x) + x + c1x + c2 = ½x[ln(x)] 2 - x ln(x) + c3 x + c2.
3. y"-4 cos2(x) = 2 ⇒ y" = 4 cos2(x)+2 = 4+2cos(2x) ⇒ y' = 4x+sin(2x)+c1

⇒ y = 2x2 - ½ cos(2x) + c1x + c2,y(0) = 0 ⇒ c2 =½,and y'(0) = 1 ⇒ c 1 = 1
⇒ y = 2x2 - ½ cos(2x) + x + ½-
4. y" = xe
x
⇒ y' = xex - ex + C 1 ⇒ y = xex - 2ex + C1 X + C2,y(0) = 1 ⇒ C2 = 3,
and y'(0) = 1 ⇒ C 1 = 2 ⇒ y = xex - 2ex + 2x + 3.
5. (a) xy" - y' + 2y = 0 is linear and homogeneous.

(b) 2y"+ ln(x)y = sin(x) is linear and nonhomogeneous.
(c) y" + y' + (x2 + y 2) = 0 is nonlinear.
(d) y"+ xy' = y 3 is nonlinear.
(e) x2y"+ xy' = 2y ⇒ x2y"+ xy' - 2y = 0 is linear and homogeneous.
(f) y" + y' + y + x = 0 ⇒ y" + y' + y = -x is linear and nonhomogeneous.
(g) y"+ yy' + y = 0 is nonlinear.

6. If y 1 = er i x and y2 = er2x are linearly dependent, then there exist constants C 1
and c2, not both zero, such that c 1 erix + c2er2 x = 0 for all x. Then x = 0 ⇒
c 1 + C 2 = 0 ⇒ c1(er , x - er2x ) = 0 for all x,and Ci -/- 0 (otherwise C2 = -Ci = 0)
⇒ eri x = er2x ⇒ r1 x = r2x,and x = l ⇒ ri = r2. Conversely,if ri = r2,then
Yi = Y2 ⇒ Yi and Y2 are linearly dependent.
7. If y1 = xa and y2 = xf3 are linearly dependent, then there exist constants c 1

and c2,not both zero,such that cixa + c2 xf3 = 0 for all x in J. Then x = l ⇒
ci +c2 = 0 ⇒ c1(xa -xf3) = 0 for all x in J, and c 1 -/- 0 (otherwise c2 = -ci = 0)
⇒ xa = xf3,and x = 2 ⇒ 2a = 2/3 ⇒ a= (3. Conversely,if a= /3, then Y 1 = Y2
⇒ Yi and Y2 are linearly dependent.
8. If Yi = xr and Y2 = esx are linearly dependent, then there exist constants c 1
and c2,not both zero,such that c 1 xr + c2esx = 0 for all x in J. Then x = l ⇒
C1 + C2e5 = 0,X = 2 ⇒ 2r c 1 + C2e 25 = 0,and X = 3 ⇒ 3r ci + C2e35 = 0. These
three equations give s = r ln(2) and 2s = r ln(3), which is impossible unless
r = s = 0. Conversely,if r = s = 0,then Yi = Y2 = 1 ⇒ Y 1 and Y2 are linearly
dependent.
9. If Yi,Y2, · · ·,Yn are linearly dependent,then there exist constants
ci, c2, · · ·, Cn , not all zero, such that C 1 Yi + C 2Y2 + · · · + CnYn = 0. If c1 -/- 0,
then yJ· = -!:l Cj
yi - £?..
Cj 2
y - · · ·- fn
Cj yn'
where' of course' the term - 1
Cj yJ
· is omitted
from the right-hand side. Conversely, if y1 = C 1 Yi + C2Y2 + · · ·+ Cn Yn , then
C1Y 1 + C2Y2 + · · · + Cn Yn - l · y1 = 0, with Cj =-1 -/- 0.
10. The functions y 1 = x,y2 = ex and y3 = e-x are linearly independent. Suppose
that c 1 x + c2 e + c3 e-x = 0 for all x. Then x = 0 ⇒ c2 + c3 = 0, x = l ⇒
c 1 + c2(e - e-i) = 0,and x = -l ⇒ -ci + c2(e-i - e) = 0. The unique solution
of this system of three equations is ci = c2 = c3 = 0.
1 1. The functions Y1 =x, y2 =e x and y3 =2x are not linearly independent because
2 · Y1 + 0 · Y2 + (-1) · y3 =0 for all x, with c1=2 -/- 0.
12. If y =
0, then 1 · y = 0 for all x, with c1 = 1 -/- 0. Hence, y is linearly
dependent. If y ¢. 0, then there exists x0 such that y(x0) -/- 0. If cy =0 for all
x, then cy(x0) = 0 ⇒ c=0, i.e.,y is linearly independent. Thus, y is linearly
independent if and only if y ¢. 0.

1. y
11
- y' - 2 y =0,y=e rx=}r2 - r-2=(r- 2)(r+l)=0 ⇒ y=c1e 2x + c2e-x .
2. y"+ y' -6 y=0, y =e rx=}r2 +r- 6=(r-2)(r+3)=0=}y=c1 e 2x +c2 e- 3x _
y (0) =0 =} C2 =-Ci=} y= c1(e
2x
-e-3x),y'=c1(2e 2x +3e-3x), and y'(0)= 10
=} C1=2=}y =2(e 2x -e-3x). Then y(l)=2(e 2 -e-3) and y'(l) =2(2e 2 +3e-3).
3. y"+5 y' +4 y=0, y=e rx =} r2 +5r+4=(r+l)(r+4) =0 =} y =c1e-x +c2 e-4x _

x 4x x 4x
y (0) = 0 =} C2 = -c1 =} y = c1(e- - e- ), y' = c1(-e- + 4e- ), and
)
y'(0 =-6 =} C1 =-2=} y =-2(e-x - e-4x ).
4. 4y" + y' -3 y = 0, y= e rx=} 4r2 + r - 3 = 0 =} r= -1/7 =} r1= ¾, r2 = -1

3
=} Y=c1e 4 x + c2e-x .
3 ..rsx
5. y" -3 y' + y = 0, y=e
rx
=} r2 - 3r + l = 0=} r = 3 ±2v'5 =} Y1 = e \ and
�
Y2 = e 2 x
6. y"- y=0, y =e rx =} r2 - 1 =0 =} r = ±1 =} y =C1e x +c2 e- x , y' =C1e x - C2 e-x ,

y (0) = 1 =} C1 + C2 = 1, y' ( 0) = 2 =} C1 - C2 = 2 =} C1 = t
C2 = -½ =}
Y= 1e x x
- .2!.e- · Then y (l)= 23 e - 1...
and y (-1 ) = .1.. �
- 2·
2 2e 2e
7. 9y" - 24y' + 16y =0, y = e rx =} 9r2 - 24r + 16 =(3r - 4)2 = 0=} r= ! =}

ix
Y1 =e ix
3
, Y2 =xe
3
8. y" - l0 y' + 25 y=0, y=e

rx
=} r2 - lOr + 25 =(r - 5)2 =0 =} r=5=}
y =e (c1 + c2 x). y(0)=l=}C1 = 1 =} y =e (l + c2 x),
5x 5x
y'=5e (l + C2 X) + C2 e , y'(0) = 1 =} C2 =-4=} y=e 5x (l

5x 5x
- 4x).
1
9. 4y" - 4y' + y = 0, y = e rx =} 4r2 - 4r + 1 = (2r - 1)2 = 0 =} r 2
=}
y =e½ x (c1 + c2x).
10. y" =0, y = e rx =} r2 = 0 =} r = 0 =} Y1 = e 0 = l, Y2 = XY1 = x, and

y=c1 + c2 x. Alternatively,y"=0 ⇒ y'=k1 =} y=k1x + k2 .
1 1. y" + y' + y=0, y =e rx =} r2 + r + l =0 =} r = -1±2,/=3 = -½ ± i °';' =}

y=e-½ x [c1 cos (-1,;x) + C sin (-1,;x)].
2
12. y" - 2y' + 2y = 0, y = e rx ⇒ r2 - 2r + 2 = 0 ⇒ r = 2±t=4" = 1 ± i ⇒

y = e x [c 1 cos(x) + C2 sin(x)]. y(0) = 0 ⇒ c 1 = 0 ⇒ y = c2 ex sin(x),
y' = c2 [e x sin(x) + ex cos(x)], and y'(0) = 1 ⇒ C2 = 1 ⇒ y = ex sin(x).
13. 2y" - 4y' + 3y = 0, y ⇒ 2r2 - 4r + 3 = 0 ⇒ r = 4±� = 1 ± i4 ⇒

= erx
y = ex [C1 cos ( 1 x) + C2 sin ( 1 x) ] .
14. y" + 4y = 0, ⇒ r2 + 4 = 0 ⇒ r = ±2i ⇒ y = c1 cos(2x) + C2 sin(2x).

y = e rx
15. (a) y" + y = 0, y = erx ⇒ r2 + 1 = 0 ⇒ r = ±i ⇒ y = C1 cos(x) + C2 sin(x).

(b) y(0) = 1 and y'(0) = 0 ⇒ y = cos(x).
(c) y(0) = 0 and y'(0) = 1 ⇒ y = sin(x).
16. x2 y"+2xy'-6y = 0. For x > 0, y = xr ⇒ r(r-1)+2r-6 = 0 ⇒ r2 +r-6 = 0

⇒ (r + 3)(r - 2) = 0 ⇒ y = c1lxl-3 + c2lxl2 = c3x-3 + c2x2 for x =/- 0.
17. 6x2 y" +7xy' -2y = 0. For x > 0, y = xr ⇒ 6r(r -1) + 7r -2 = 6r2 + r -2 = 0
⇒ r = -�t7 = ½, -i ⇒ y = c1lxl 1 /2 + c2x-2 l3 for x =/- 0.
18. x2 y" -7xy'+ 16y = 0. For x > 0, y = xr ⇒ r(r-1)-7r+16 = r2 -8r+16 = 0
⇒ (r - 4)2 = 0 ⇒ y = x4 [c1 + c2 ln lxl] for x =/- 0. y(l) = 0 ⇒ c 1 = 0 ⇒
y = c2x4 ln lxl, y' = c2 [4x3 ln lxl + x3 ], and y'(l) = 2 ⇒ c2 = 2 ⇒ y = 2x4 ln lxl.
Then y (-1) = 0 and y ( e) = 2e 4.
19. 9x2 y"-3xy'+4y = 0. For x > 0, y = xr ⇒ 9r(r-1)-3r+4 = 9r2 -12r+4 = 0

⇒ (3r - 2)2 = 0 ⇒ y = x213 [c1 + c2 ln lxl] for x =/- 0.
20. x2 y"+xy' = 0. (1) For x > 0, y = xr ⇒ r(r-l)+r = r2 = 0 ⇒ y = Ci +c2 ln lxl
for x =/- 0. (2) For x =/- 0, x2 y" + xy' = 0 ⇒ xy" + y' = 0. Integrate by parts
J J
to obtain xy' - y' dx + y' dx = k 1 ⇒ y' = � ⇒ y = k1 ln lxl + k2. (3) For
x =/- 0, 7 = -¾
⇒ ln IY'I = - ln lxl + C1 ⇒ y' = � ⇒ y = k1 ln lxl + k2.
21. -2x2 y" + 2xy' - 5y = 0 ⇒ 2x2 y" - 2xy' + 5y = 0. For x > 0, y = xr ⇒
2r(r -1) -2r + 5 = 2r2 - 4r + 5 = 0 ⇒ r = 4±ij24 = 1 ± i 'I,; ⇒
y = x [c 1 cos ( 4 ln lxl) + c sin ( 'I,; ln lxl)] for x =/- 0.
2
22. x2 y"-3xy'+ 13y = 0. For x > 0, y = xr ⇒ r(r-1) -3r+ 13 = r2 -4r+13 = 0

⇒ r = 2 ± 3i ⇒ y = x2 [c1 cos(3 ln Ix!)+ c2 sin(3 ln lxl)] for x =/- 0. Then y(l) = 0
⇒ c1 = 0 ⇒ y = c2 x2 sin(3 ln lxl), y' = c2[2x sin(3 ln lxl) + 3x cos(3 ln lxl)], and
y'(l) = 6 ⇒ c2 = 2 ⇒ y = 2x2 sin(3 ln \x\).
23. x2 y" + xy' + 4y = 0. For x > 0, y = xr ⇒ r(r - 1) + r + 4 = r2 + 4 = 0 ⇒

r = ±2i ⇒ y = c 1 cos(2 ln lxl) + c2 sin(2 ln lxl) for x =/- 0.
,.I
Exercises 3. 3 - page 7 4
l. y" + 4y' = -5e x . For y" + 4y' = 0, y = e rx ⇒ r2 + 4r = r(r + 4) = 0 ⇒ y1 = 1
and Y2 = e-4x _ YP = Ae x ⇒ y� = Ae x , yi = Ae x , and yi + 4y� = -5e x ⇒
A= -1 ⇒ YP = -ex . The general solution is y =-e x + c 1 + c2 e-4x_
2. y" - y = 6e 2x , y(0) = 2, y'(0) = 4. For y" - y = 0, y = e rx ⇒ r2 - 1 = 0 ⇒

Yl = e x and Y2 = e-x _ YP = Ae 2x ⇒ y� = 2Ae 2x , Yi= 4Ae 2x , and Yi- yp = 6e 2x
⇒ A = 2 ⇒ YP = 2e2x . The general solution is y = 2e 2x + c1ex + c2 e-x.
y(0) = 2 +C1 +C2 = 2 and y'(0) = 4 + C1 - C2 = 4 ⇒ c 1 = C2 = 0 ⇒ y = 2e .
2x
3. y" - 5y' + 6y = 4e- 2x . For y" - 5y' + 6y = 0, y = e rx ⇒

r - 5r + 6 = (r - 2)(r - 3) = 0 ⇒ Y1 = e 2x and Y2 = e 3x . YP = Ae-2x ⇒
2
YP, = -2Ae-2x ' yP" = 4Ae-2x ' and yP" -5yP' +6yP = 4e-2x ⇒ A=.!.5 ⇒ yP = l
5
e-2x.
The general solution is y = ½e- 2x + c 1 e 2x + c3 e 3x .
4. y" + 3y ' - 4y = -lOe x . For y " + 3y ' - 4y = 0, y = e rx ⇒

r +3r-4 = (r-l)(r+4) = 0 ⇒ y1 = e x and Y2 = e-4x _ Since e x is a solution of
2
the homogeneous equation , YP = Axe x . Then y� = Ae x(l+x), yi = Aex(2+x),

and yi + 3y� - 4yp = - lOe x ⇒ A = -2 ⇒ YP = -2xe x . The general solution is
Y = -2xe x + C1ex + C2 e-4x _
5. y" + 6y' + 9y = 6e-3x _ For y " + 6y' + 9y = 0, y = e rx

⇒ r2 + 6r + 9 = (r + 3)2 = 0 ⇒ Y1 = e- 3x and Y2 = xe- 3x _ Since e-3x
and xe- 3x are solutions of the homogeneous equation , YP = Ax2 e-3x _ Then
y� = Ae- (2x - 3x ), yi = Ae- (2- l2x + 9x ), and yi + 6y� +9yp = 6e- ⇒
3x 2 3x 2 3x
A= 3 ⇒ YP = 3x e- _ The general solution is y = 3x e- + c1 e- + c2 xe- x_

2 3x 2 3x 3x 3
6. 6y " + y ' - 2y = 65sin(x). For 6y" + y' - 2y = 0, y = e rx ⇒ 6r2 + r - 2 = 0 ⇒

Y1 = e½ and Y2 = e-� . YP = Acos(x) +B sin(x) ⇒ y� = -Asin(x) +Bcos(x),
x x
yi = -Acos(x) - Bsin(x), and 6yi +y� - 2yp = 65sin(x) ⇒

A + 8B = -65 .
{ } ⇒ A= -1 and B = -8 ⇒ YP = -cos(x) - 8sm(x).
-SA + B = 0
The general solution is y = -cos(x) - 8sin(x) + c1e½ x +c2 e-� x .
7. y" + 2y' + y = 50cos(3x), y(0) = 0, y '(0) = 2. For y" + 2y' + y = 0, y = e rx ⇒

r +2r+l = (r+l)2 = 0 ⇒ Y1 = e- x and Y2 = xe-x . YP = Acos(3x)+Bsin(3x)
2
⇒ y� = -3Asin(3x) + 3Bcos(3x), yi = -9Acos(3x) - 9 Bsin(3x), and

Yi + 2y� + YP = (-8A + 6B) cos(3x) +(-6A - 8B)sin(3x) = 50cos(3x) ⇒
-8A + 6B = 50 .
{ } ⇒ A = -4 and B = 3 ⇒ YP = 3sm(3x) - 4cos(3x).
_6A _ SB = 0
The general solution is y = 3sin(3x) - 4cos(3x) +c1e-x + c2 xe-x . y(0) = 0 ⇒
c1 = 4, y'(0) = 2 ⇒ C2 = -3 ⇒ y = 3sin(3x) - 4cos(3x) + 4e-x - 3xe-x .
8. y" + y = 4sin(x). For y" + y = 0, y = e rx ⇒ r 2 + 1 = 0 ⇒ y1 = cos(x) and

y2 = sin(x). Since cos(x) and sin(x) are solutions of the homogeneous equation,
YP = x[Acos(x) + Bsin(x)]. Then
y� = [Acos(x) + Bsin(x)] + x[-Asin(x) +Bcos(x)],
yi = [-2A sin(x) + 2B cos(x)]+x[-Acos(x) - Bsin(x)], and yi +YP = 4sin(x)
⇒ A= -2 and B = 0 ⇒ YP = -2xcos(x). The general solution is
y = - 2xcos(X) + C1 cos(X) + C2 sin(X).
9. y" + 4y = 8cos(2x) + 12sin(2x). For y" + 4y = 0, y = e rx ⇒ r 2 + 4 = 0 ⇒

y1 = cos(2x) and y2 = sin(2x). Since cos(2x) and sin(2x) are solutions of the
homogeneous equation, YP = x[Acos(2x) + Bsin(2x)]. Then
y� = [Acos(2x) + B sin(2x)] + x[-2Asin(2x) + 2B cos(2x)],
yi = [-4A sin(2x) + 4B cos(2x)] + x[-4A cos(2x) - 4B sin(2x)], and
yi + 4yP = 8cos(2x) + 12sin(2x) ⇒ A= -3 and B = 2
⇒ YP = x[2sin(2x) - 3cos(2x)]. The general solution is
y = x[2sin(2x) - 3cos(2x)] + c1cos(2x) + c2 sin(2x).
10. y" + 4y' + 4y = 4sin(x) - 3cos(x) . For y" + 4y' + 4y = 0, y = e rx ⇒

r 2+4r+4 = (r+2) 2 = 0 ⇒ Y1 = e- 2x and Y2 = xe- 2x. YP = Acos(x)+Bsin(x)
⇒ y� = -Asin(x) +Bcos(x), yi = -Acos(x) - Bsin(x), and
yi +4y� +4yp = (3A+4B) cos(x) +(-4A+3B) sin(x) = 4sin(x) - 3cos(x) ⇒
3A + 4B = -3
{ } ⇒ A= -1 and B = 0 ⇒ yP = - cos(x) . The general
_4A+ 3B = 4
solution is y = - cos(x) + c 1 e- 2x + c2 xe- 2x.
11. y" - y = 4. For y" - y = 0, y = e rx :::::} r 2 - 1 = (r - l)(r + 1) = 0 :::::} Y1 = e x

and Y2 = e-x. YP = A ⇒ y� = yi = 0, and yi - YP = 4 ⇒ A = -4 ⇒ YP = -4.
The general solution is y = -4 + c 1 e x + c2 e- x .
12. y" - y' - 12y = 3 + 12x. For y" - y' - 12y = 0, y = e rx ⇒

r 2 - r - 12 = (r - 4)(r + 3) = 0 ⇒ Y1 = e4x and Y2 = e- 3x_ YP = A+ Bx ⇒
y� = B, yi = 0, and yi-y� - 12yp = (-12A- B) - 12Bx = 3+ 12x ⇒ A=-½
and B = -1 ⇒ YP = -½-x. The general solution is y = -½-x+c1 e 4x +c2 e-3x _
13. y" + 2y' + 3y = -9x. For y" + 2y' + 3y = 0, y = e rx ⇒ r 2 + 2r + 3 = 0 ⇒

r = -l±v'2i ⇒ Y1 = e- x cos(v'2x) and Y2 = e-xsin(v'2x). YP = A+Bx ⇒
y� = B, yi = 0, and yi + 2y� + 3yp = (2B + 3A) + 3Bx = -9x ⇒ A= 2 and
B = -3 ⇒ YP = 2 - 3x. The general solution is
y = 2 - 3x + e-x[c1cos(v'2x)+ C2 sin(v'2x)] .
14. y" - 9y = 2 - 3x+ x2. For y" - 9y = 0, y = e rx ⇒ r 2 - 9 = (r - 3)(r + 3) = 0
⇒ Y1 = e 3x and Y2 = e- 3x_ YP = A+Bx+Cx2 ⇒ y� = B+2Cx, vi= 2C, and
20
YP" - 9yP = (2C - 9A) - 9Bx - 9 Cx2 = 2 - 3x + x2 ⇒ A = - 81' B = .3!. and
20
C = _.!.9 ⇒ yP = -81 + lx
3 - lx9
2
. The general solution is
_ 20
Y = 81 + 3 x 9 x + c1e + c2 e-3x
.!. _ l 2 3x
15. y"+ y ' = 1 + x, y(0) = 1, y '(0) = 1. For y"+ y ' = 0, y = erx =}

r2 + r = r(r + 1) = 0 ⇒ y1 = 1 and y2 = e-x . Since 1 is a solution of the
homogeneous equation, yP = x(A+Bx)= Ax+ Bx2. Then y�=A+ 2Bx,
y; = 2B, and y ; + y� = 1 + x⇒ A=0 and B = ½ ⇒ YP = ½x2 . The general
solution is y = ½x2 + c1 + c2e- x . y(0) = 1 ⇒ c1 + c2 = 1, and y'(0) = 1 ⇒
C2 = -1 ⇒ C1 = 2 ::::} y = ½x2 + 2 - e- x .
16. y" - y' = 4x3 . For y" - y' = 0, y = erx ⇒r2 - r = r(r - 1) = 0⇒ Y1 = 1 and
y2 = ex . Since 1 is a solution of the homogeneous equation,
YP=x(A+Bx+ Cx2 + Dx3 )=Ax+ Bx2+ Cx3+ Dx4 . Then
Y,p =A+ 2Bx+ 3Cx2 + 4Dx3 ' y"p = 2B + 6Cx+ 12Dx2 ' and yp" - y'p = 4x3 ⇒
(2B - A)+ (6C - 2B)x+ (12D - 3C)x2 - 4Dx3 = 4x3 ⇒A=-24, B = -12,
C = -4 and D = - l⇒ YP = -24x - 12x2 - 4x3 - x4. The general solution is
y = -24x - 12x2 - 4x3 - x4 + c 1 + c2 ex .
17. y" - y = -2x + 4ex + 2 sin(x), y (0) = 0, y'(0) = 9. For y" - y = 0, y = erx
⇒ r2 - 1 = (r - l)(r + 1) = 0 ⇒ y1 = ex and Y2 = e- x . For y" - y = -2x,
Yp 1 =A+Bx⇒A=0 and B = 2 ⇒YP i = 2x. For y " - y = 4ex , Yp2 = Axex
⇒ A = 2 ⇒ Yp2 = 2xex . For y" - y = 2 sin(x), yp3 = A cos(x) + Bsin(x)
⇒ A = 0 and B = -l ⇒ yp3 = - sin(x). Hence, YP = 2x + 2xex - sin(x).
The general solution is y = 2x + 2xex - sin(x) + c1ex + c2 e-x . y(0) = 0 ⇒
c 1 + c2 = 0, and y'(0) = 9 ⇒ 3 + c1 - c2 = 9 ⇒ c1 = 3 and c2 = -3 ⇒
:=�:
y = 2x+ 2xex - sin(x)+ 3ex - 3e- x .
18. y"+y'-2y = 9ex . For y"+y'-2y = 0, y = erx⇒r2+r-2 = (r-l)(r+2) = 0⇒

Y1 = ex and Y2 = e-2x, with W(x) = I :: _ I = -3e-x and f(x) = 9ex .
2
Then u 1 = J 3dx = 3x, u 2 = J-3e3x dx = -e3x , and YP = 3xex - ex , or
YP = 3xex . The general solution is y = 3xex + c1ex + c2e- 2x.
19. y" -4y'+4y = 6xe2x. For y" -4y'+4y = 0, y = erx⇒r2 -4r+4 = (r- 2)2 = 0
e2x xe2x 4x
=} y1 = e and Y2 = xe2x, with W ( x ) = I 2x 2x 2x I = e and
2x .
2e e + 2xe
f(x) = 6xe2x . Then u 1 = J-6x2 dx = -2x3 , u 2 = J 6xdx = 3x2 , and
YP = -2x3 e2x+3x3 e2x = x3 e2x . The general solution is y = x3 e2x+c1e2x+c2 xe2x.
20. y"+y' = 1, y(0) = 2, y'(0) = 4. For y"+y'

= 0, y = erx⇒r2+r = r(r+ 1) = 0
⇒y 1 = 1 and Y2 = e- x , with W(x) = _: x = -e-x and J(x) = 1. Then
1 1
0 --x I
u 1 = f ldx = x, u 2 = J-ex dx = -ex , and YP = x - l, or YP = x. The general
solution is y = x+ c1 + c2e-x. y (0) = 2 ⇒c1 + c2 = 2, y'(0) = 4 ⇒1 - C2 = 4
⇒ c2 = -3, C1 = 5, and y = x+ 5 - 3e-x .
21. y" + 9y = 6 s in( 3x). For y" + 9y = 0, y = erx =} r2 + 9 = 0 =} y1 = cos( 3x) and
. .
y2 = sm( 3x), with W(x) = _ I cos( 3x) sin( 3x) I .
= 3 and J(x) = 6 sm( 3x).
3 s in(3x) 3 cos( 3x)
Then
u1 = - J 2 sin2( 3x) dx = - J l - cos(6x) dx = -x + ¾ sin( 6x),
u2 = J2 sin( 3x) cos( 3x) dx = J sin( 6x) dx = -¼cos(6x), and
YP = [-x + ¼ sin( 6x]) cos( 3x) - ¼ cos( 6x) sin( 3x) = -x cos(3x) + ¼ sin( 3x), or
YP = -x cos( 3x). The general solution isy = -x cos( 3x) +c1 cos(3x) +c2 sin(3x).
22. y" + y = sec(x). For y" + y = 0, y = erx ⇒ r2 + 1 = 0 ⇒ Y1 = cos(x) and

y2 = sin( x), with W(x) =
I c?s( x) sm(x) I = 1 and J(x) = sec( x). Then
-Sm ( X ) COS( X )
u1 = - J tan( x) dx = - lnI sec(x)I, u2 = J l dx = x, and
YP = -cos(x) InI sec( x)I + xsin( x). The general solution is
y = xsin( x) - cos(x) InI sec( x)I + c1 cos(x) + c2 sin( x).
23. y"+6y'+l3y = l6e- 3 x cos( 2x). For y"+6y'+l3y = 0, y = erx =} r2 +6r+13 = 0

⇒ r =-3 ± 2i ⇒ Y1 = e- 3x cos(2x) and Y2 = e- 3x sin(2x), with W(x) =
I
e- 3x cos( 2x) e- 3x sin( 2x)
I -3e- 3x
cos( 2x) - 2e- 3x .
sm 2x) -3e-
.
3x .
sm ( 2x) + 2e- 3x
cos 2x)
= 2e-6x and
( (
J(x) = l6e-3 x cos( 2x). Then
u1 = -8 J sin( 2x) cos(2x) dx =-4 J sin( 4x) dx = cos( 4x),
u2 = 8 J cos2(2x) dx = 4f l + cos( 4x) dx = 4x + sin( 4x), and
YP = cos( 4x)e- 3 x cos(2x) + [4x + sin( 4x)Je- 3x sin( 2x)
= 4xe- 3x sin(2x) + e-3x cos( 2x), or YP = 4xe-3x sin( 2x).
The general solution is y = 4xe- 3x sin( 2x) + c 1 e- 3x cos( 2x) + c2e-3x sin( 2x).
24. y" - 4y' + l3y = �. cos(3x) For y"- 4y' + l3y = 0 ' y = e =} r - 4r + 13 = 0 =}
rx 2
r = 2 ± 3i ⇒ Y1 = e2x cos( 3x) and Y2 = e2x sin( 3x), with

W(x ) -
2e cos(3x)- 3e2x s in( 3x) 2e2x sin( 3x) + 3e2x cos(3x) I -
_ , 2x
e2x cos(3x) e2x sin( 3x)
- 3 e4x and
J(x) = c!se(;x )· Then U1 = -3J tan(3x)dx =- lnlsec(3x)I, U2 = 3f ldx = 3x,
and YP = -e2x cos( 3x) lnI sec( 3x)I + 3xe 2x sin( 3x). The general solution is
y = 3xe2x sin(3x) - e2x cos( 3x) lnI sec(3x)I + c 1e2x cos( 3x) + c2e2x sin(3x).
25. y"- 2y' + y = ex ln( x), x > 0. For y"- 2y' + y = 0, y = erx =}
r2- 2r + 1 = (r- 1) 2 = 0 ⇒ Y1 = ex and Y2 = xex, with

W(x) = I ee: x
xex
e +xex
/ = e2x and f(x) = ex ln(x). Then
u1 = - J = J ln( x) dx = x ln( x)- x, and
2
x ln( x) dx = -; ln( x) + x:, u2
YP = [ x: - x2 ln( x)] ex + [xln( x) - x]xex = ex [½x2 ln( x)-¾x2]. The general
2
solution is y = ex [½x2 ln(x)-¾x2] + c1ex + c2 xex .

26. y " + 3y' + 2y = sin(ex ). For y" + 3y' + 2y = 0, y = e rx ⇒

r2 + 3r + 2 = (r + l)(r + 2) = 0::::} y1 = e -x and Y2 = e - 2x , with
e e
W(x) = I =: =:: I = -e -3x and f(x) = sin(ex ). Then
-e - 2e
U1 = Jex sin(ex ) dx and u2 = - J e 2x sin(ex )dx. The substitution u = ex ,
:'. = ex , then gives u1 = Jsin(u)du = - cos(u) = - cos(ex ) and, by integration
by parts, u2 = - Ju sin(u)du = ucos(u) - sin(u) = ex cos(ex ) - sin(ex ). Then
YP = - e - cos(e ) + e - [ e cos(ex ) - sin(e )] = - e -2x sin(e ), and the general
x x 2x x x x
solution is y = -e-2x sin(ex ) + c 1 e-x + c2 e-2x .
27. x2 y" - 6y = l0x5.

For x2 y " - 6y = 0 and x > 0, y = xr =} r2 - r - 6 = (r - 3)(r + 2) = 0 ⇒
y1 = x3 and y2 = x-2 . These solutions are defined for all x =I= 0, with
W(x) = / 3;: _2;=:
I = -5 and J(x) = l0x3. Then u1 = 2Jxdx = x2,
u2 = -2 Jx6 dx = -�x 7
7
'
and yP = x2 x3 - �x
7
7
x-2 = Qx
7
5
. The general solution
is y = �x5 + C1X 3 + c2x- 2 for X =I= 0.
28. x2 y" - 4xy' + 6y = x4 ex .

For x2 y " - 4xy' + 6y = 0 and x > 0, y
= xr ⇒ = (r - 2)(r -3) = 0
r2 - 5r + 6
⇒ y1 = x2 and 2 = x3 , defined for all real x, with W(x) = I ;; 3;: I = x4
y
and f(x) = x2 x . Then U1 = - Jx x dx = x - x x , u2 = J x dx = x , and

e e e e e e
YP = ( x - xex)x2 + x3 x = x2 x . The general solution is = x x + c 1 x + c2x

2 2 3
e e e e
y
for all real x.
29. x2 y " + 5xy' + 3y = 4xex , y (l) = e , y'(l) = e .

2
For x2 y" + 5xy'+3y = 0 and x > 0, y = xr ⇒ r2 +4r+3 = (r+ l)(r+3) = 0 ⇒

x x
y1 = x-1 and y2 = x-3, defined for all x f. 0, with W(x) = I =: =: I =
-X -3X
-2x-5 and f(x) = 4 e; • Then u1 = 2Jxex dx = ex and U2 = -2Jx3 ex dx.

2
2 2 2
Integration by parts gives

u2 = - Jx2 · 2xex dx = -x2 ex + J2xex dx = -x2 ex + ex . Hence,
2 2 2 2 2
1 + ex \1 - x2 )x-3 = x-3ex 2 . The general solution for x f. 0 is

YP = e x-
x2
y
= x-3 ex +c1x-1+c2 x- 3. y(l) = e ⇒ C 1 +c2 = 0 ⇒ y = x-3ex2 +c1(x-1 -x- 3)
2
4ex2+2x-2 ex2 +c (-x- 2+3x- 4), and y'(l) = e ⇒ c1 = e ⇒ c2 = -e

⇒ y' = -3x- 1
⇒ y = x-3 ex + e(x_, - x-3 ).
2
30. x2 y" - 2xy' + 2y = x3 cos(x).

For x2 y" - 2xy ' + 2y = 0 and x > 0, = xr = (r - l)(r - 2) = 0
r2 - 3r + 2
;:
y
=}
⇒ y1 = x and 2 = x2 , defined for all real x, with W(x) = I I

y I = x2
and f(x) = xcos(x). Then u 1 = - Jxcos(x)dx = -x sin(x) - cos(x), u2 =
J cos(x) dx = sin(x), and Y = -[ xsin(x) +cos(x)]x +[sin(x)]x2 = -xcos(x).

The general solution is y = c 1 x + c 2 x2 - x cos(x) for all real x.
P
31 . x 2 y" +xy' - y = x2 sin(x).

For x2y" +xy' - y = 0 and x > 0, y = xr
and y2 = x- 1 , defined for all x =/: 0, with W(x) =
f(x) = sin(x). Then u 1 =
J
½J
sin(x) dx =
⇒
-½
cos(x), u 2 =
½x cos(x) - x cos(x) dx = ½x cos(x) - x sin(x) - cos(x ), and
2 2
I� _:=:
r2 - 1 = 0 ⇒ r = ±l
I
-½ J
⇒ Y1
x2 sin(x) dx =
=x
= -2x-1 and
YP = -½x x
cos(x) +x- 1 [½x2 cos(x) - sin(x) - cos(x)] = - sin(x) - co� ) The
x .
general solution is y = - sin(x) - co�x ) + c1 x + c 2 x-1 for x =/: 0.
32. x2 y" + 3xy' + y = ln� ), X > 0.

⇒r 2 +2r + 1 = (r + 1 )2 = 0 ⇒ Y1 = x- 1 and
x
For x2 y" +3xy' +y = 0, y = xr

1
. W ( x) = I x-_ 2 _ 2 I = x and
x-11n(x) -3
Y 2 = x-1 ln ( x) , with _
-x -x 1 n(x ) +x
2
J
f(x) = 1,�C:). Then U1 = - [ln�)] dx = -½[ln(x)]3 , u 2 = J 11; ) dx = ½[ln(x)] ,
2 1 x 2
1 3 1 3 1
and YP = -½ x- [ln(x)] +½x- [ln(x)] = ½x- [ln(x)] . The general solution for
3
x > 0 is y = ¼x-1[ln(x)]3 + c 1 x- 1 + c 2 x-1 ln(x).
33. x2 y" - 3xy' +4y = x2 ln(x), x > 0.

For x2 y" - 3xy' +4y = 0, y = xr ⇒ r2 - 4r +4 = (r - 2)2 = 0 ⇒ y 1 = x2 and
Y 2 = x2 ln(x), with W(x) = I
x2 x2 ln(x)
2x 2xln(x) +x
I = x3 and f(x) = ln(x). Then
U 1 = - [In�)]2 dx = -½[ln(x)]3 , u 2 = 11; ) dx = ½[ln(x)] , and
J J
1 x 2
yP = -½ x2[ln(x)]3 +½x2[ln(x)]3 = ½x2[ln(x)]3 . The general solution for x > 0 is

y = ¼x2[ln(x)]3 + c1 x2 + c 2 x2 ln(x).
34. x2y" - 3xy' + l3y = 9x2 .

For x 2 y" - 3xy' +l3y = 0 and x > 0, y = x,. ⇒ r2 - 4r +13 = 0 ⇒ r = 2 ± 3i
⇒ Yi = x2 cos[3ln(x)] and Y 2 = x2 sin[3ln(x)], with W(x) =
x2 cos[3ln(x)] x2 sin[3ln(x)]
I - 3x ' and
I 2xcos[3ln(x)] - 3xsin[3ln(x)] 2xsin[3ln(x)] +3xcos[3ln(x)] _ 3
J
J(x) = 9. Then U1 = -3 si [ !n(x)]
n 3
dx = cos[3ln(x)],
3 n
u 2 = 3f cos[ � (x)] dx = sin[3ln(x)], and YP = x2 cos 2[3ln(x)] +x2 sin 2[3ln(x)] =
x2 . Since yP is a particular solution of the nonhomogeneous equation for any
real x, the general solution for x =/: 0 is y = x2[1 +c1 cos(3ln lxl)+c 2 sin(3ln lxl)].
35. x2y" + 7xy' + l3y = X3 COS212111 (X )l, x > 0.

For x2 y" +7xy' + l3y = 0, y = x1· ⇒ r2 +6r + 13 = 0 ⇒ r = -3 ± 2i ⇒
Yi = x- 3 cos[2ln(x)] and Y 2 = x- 3 sin[2ln(x)], with W(x) =
x- 3 cos[2ln(x)] x- 3 sin[2ln(x)]
I -3x- 4 cos[2ln(x)] - 2x- 4 sin[2ln(x)] -3x- 4 sin[2ln(x)] +2x- 4 cos[2ln(x)] I
= 2x- 7 and f(x) = xs cos212111 x ]' The substitution u = 2ln(x), �� = �, then

( )
gives
u1 = -2J x scoin[s22[12nln(x(x)] ] dx - -J cos
s in( u)
2 (u) du -
__ l_ - 1
cos(u) - cos[ 2 ln(x)]'
J J
)
U 2 = 2 x cos[i ln(x) ] dx = sec(u) du= ln I sec(u) +tan(u)I =
ln I sec[2ln(x)] +tan[2ln(x)]I, and
3 3
YP = -x- + x- sin[2ln(x)]ln I sec[2ln(x)] +tan[2ln(x)]I. The general solution
for x > 0 is y =
x- 3 {sin[2ln(x)]ln I sec[2ln(x)]+tan[2ln(x)] l-1+c1 cos[2ln(x)]+c 2 sin[2ln(x)]} .
36. x2 y" - 5xy' + l3y =8x3 sin[2ln(x)], x > 0.

For x 2 y" - 5xy' + l3y = 0, y = xr ⇒ r 2 - 6r +13 = 0 ⇒ r = 3 ± 2i ⇒
Y 1 =x cos[2ln(x)] and y 2 =x sin[2ln(x)], with
3 3
x cos[2ln(x)]
3
x3 sin[2ln(x)]
W )-1 - 3x cos[2ln(x)] - 2x sin[2ln(x)] 3x sin[2ln(x)] + 2x2 cos[2ln(x)] I
( x 2 2 2
= 2x5 and f(x) = 8xsin[2ln(x)]. The substitution u = 2ln(x), �� =�'then

gives
J J J
u1 = -4 si [ 2�n(x)] dx = -2 sin 2(u) du= cos(2u) -1 du= ½sin(2u) - u
n2
= ½ sin[4ln(x)] - 2ln(x),
u2 = 4 J sin[ 2 In (x)]xcos[ 2 In(x)] dx = 2 J sin(u) cos(u) du= J sin(2u) du=-½ cos(2u)
= -½ cos[4ln(x)], and
YP = {½ sin[4ln(x)] - 2ln(x)}x cos[2ln(x)] - ½ cos[4 ln(x)]x sin[2ln(x)]
3 3
= ½x {sin[4ln(x)]cos[2ln(x)] - cos[4ln(x)]sin[2ln(x)]} - 2x3 ln(x)cos[2ln(x)]

3
= ½x3 sin[2ln(x)] - 2x3 ln(x)cos[2ln(x)], or YP = -2x3 ln(x)cos[2ln(x)]. The

general solution is y = x3 { c1 cos[2ln(x)] + c 2 sin[2ln(x)] - 2ln(x) cos[2ln(x)]},
X > 0.
37. 4x 2 y" +4xy' - y = x2 .

For 4x 2 y" +4xy' -y = 0 and x > 0, y = xr ⇒ 4r(r -1) + 4r -1 = 4r 2 - 1 = 0
⇒ y1 = xt /2 and Y2 = x- 1 12, with W(x) = I 1 x_/112 1 x-_31/ 2 I = -x-1 and
1 2 1 2
2X -2X
J J
J(x) = .4!. · Then ul = .4!. x1 12 dx = l6 x31 2 ' u 2 = _.!.4 x312 dx = -l 10 x
512
' and
YP = /5 x . Since YP is a particular solution of the nonhomogeneous equation for
2
all real x, the general solution for x =/- 0 is y = 1�x2 + c 1 lxl1 / 2 + c 2 lxl-112 .

1. /� 2 = �+3x2 and z(x) =y'(x) ⇒ z�2 = �+3x2 ⇒ ln lz- 21 = 21n lxl+x3 +c1
⇒ z - 2 =k 1 x2 ex3 ⇒ y =f k1x2 ex3 + 2 dx = ½k1ex3 + 2x + k 2 .
2. x2 y" = (y') 2 and z(x) = y'(x) ⇒ x2 z' = z 2 ⇒ ;� = ;2 ⇒ -� = -� + c ⇒
z =_ 1-cx =} y = J 1-cx dx = _.!.
_
x --x
c
-. 1- dx = -"'c _ l..
J l - 1-cx c2
-J. 0.
ln 11-cxl +k ' if c 1
2
If c = 0, then z = X =} y = 2 + k 1.
x
3. y" -x(y')3 =0 and z(x) = y'(x) ⇒ z' -xz3 = 0 ⇒ z- 3 z' =x ⇒ -½z- 2 = ½x2 +c
⇒ z2 = k �x2 ⇒ z = ✓:�x2 and is real if and only if k = µ2 > 0. Then
x = µsin(t) ⇒ y = ±J .� 1
µ,2-x2 dx = ±f ldt = ±l+c1 = ±sin - µ, +c1. (.=£)
4. y'y" + x =0 and z(x) = y'(x) ⇒ zz' + x = 0 ⇒ ½z2 = -½x2 + c ⇒
z = ±Jk - x2 and is real if and only if k = µ2 > 0. Then x = µsin(t) ⇒
J
y = ±J µ2 - x2 dx = ±µ2 J cos 2(t)dl = ±� J 1 + cos (2t)dl
= ¥ [t+ ½ sin(2t)] +c1 = ¥[t+sin(l) cos(t)] +c1 =
± � [sin - 1 (!) + x7] + c 1.
5. xy" = y'ln(y') -y'ln(x) and z(x) = y'(x) ⇒ xz' =zln(z) - zln(x) ⇒

z' = �ln( �), which is homogeneous, and u = � ⇒ u + xu' = u ln( u) ⇒
xu' = u[ln(u) - 1] :::} u[ln(�)-l] =;; :::} ln I ln(u) - 11 =ln(x) +c:::}
ln(u) -1 = kx ⇒ u = e kx+l ⇒ z =xekx+1, and k = ±ec -=J 0. Hence,
Y =J xekx+ l dx = lk xekx+ l - l....
k2
ekx+ l + kl ·
6. y" + 2y' = 2e-x # and z(x) = y'(x) ⇒ z' +2z = 2e-xy'z, which is a Bernoulli
equation with a=½- Then u = z1l2 ⇒ z = u2 ⇒ z' = 2uu' ⇒
2uu' + 2u2 = 2e- x u ⇒ u' +u = e- x , which is linear, with the integrating factor
I(x) = ex . Then (ex u)' = 1:::} ex u = x+c ⇒ u = (x+c)e-x ⇒ z =(x+c)2e- 2x
⇒ y =J(x +c)2e-2xdx = -½(x+c)2e-2x + f(x +c)e-2xdx
= -½(x +c)2e- 2x - ½(x+c)e- 2x + ½ J e- 2x dx
= -½(x+c)2e- 2x - ½(x+c)e- 2x - ¼e- 2x + k.
7. x4 +(y')2 - xy'y" =0 and z(x) = y'(x) ⇒ x4 + z2 - xzz' = 0, with

= _;!x = I'( x) I
I(x) :::} (x) = x- is an
3
P(x ' z) = x4 + z2 ' Q(x ' z) = -xz ' Pz-Qx
Q
integrating factor. Then x + x- 3 z2 - x- 2 zz' = 0 is exact. fx = x + x-3 z2 ⇒
J(x, z) =½x2 - ½x- 2 z2 +g(z), and fz = -x- 2 z ⇒ f(x, z) = ½x2 - ½x-2z2 + c
and ½x2 - ½x- 2 z2 =c1, or x2 - x- 2 z2 =k, and z = ±xJx2 - k. Then
y = ±J xJx2 - kdx = ±½(x2 - k)31 2 + k1 .
8. y" = (2x - y' + 3 )2 + 2 and z(x) = y'(x) ⇒ z' = (2x -z + 3 )2 + 2. Let
u = 2x - z + 3. Then u' = 2 - z' = 2 -(u2 + 2 ) = -u2 , which is separable.
Then - Uu� = 1 :::} Ul =x+c ⇒ u = -+
1
:::} 2x - z+ 3 = -+ :::} z= 2x+ 3 - -+
1 1
⇒ y =J 2x+3 - x�c dx = x2 + 3x - ln Ix + cl +k.

X C X C X C
9. y" +(2 -6y +12y2) (y')3 = 0 and v(y) = y'(x) ⇒ vv' +(2 - 6y + 12y2 )v3 = 0 . If
v = 0, then y = c. If v -=J 0, then - �� = 2 -6y +12y 2 ⇒ � = 2y-3y2 +4y3 +c1
:::} (2y - 3y2 + 4y3 + c1)y' = 1:::} y2 - y3 +y4 + C1Y = X + C2.
10. y" = (y')2 tan(y) and v(y) = y'(x) ⇒ vv' = v2 tan(y). Ifv = 0, then y = c. If
v -=J 0, then�= tan(y) ⇒ ln lvl = ln I sec(y)l+c1 ⇒ v = ksec(y) ⇒ cos(y)y' = k
11. yy" =y'(y' -1) ln(y' -1) and v(y) =y'(x) ⇒ yvv' = v(v-l) ln(v-1), with v-/= 0
since ln(-1) is undefined. Then (v-l)f�(v-l) =t ⇒ ln I ln(v -1)1 = ln IYI+ c ⇒
ln(v - 1) =ky ⇒ v = ek + y l:::::} kY'y =l:::::} e kYy' = land k= ±ec ...J. 0
e + l l+ckY r ·
+
Then -¾ ln(l +e- ) =x+k1 :::::} 1+e- y =e (x ki) :::::} -ky =ln[e- k(x ki ) -1]
k y k - k +
+
:::::} y = -¾ ln[e- k(x k1 ) - 1].
12. sin(y)y" + cos(y)(y') 2 = (y')3 and v(y) = y'(x) ⇒ sin(y)vv' + cos(y)v2 = v3.
If v = 0, then y = c. If v -/= 0, then sin(y)v' + cos(y)v = v2, which is a
Bernoulli equation with a = 2. Then u = v- 1 ⇒ v =u- 1 ⇒ v' = -u- 2 u'
⇒ - sin(y)u-2 u1 + cos(y)u-1 = u-2 , or -sin(y)u' + cos(y)u = 1, which is
linear. The standard form is u'-cot(y)u= - csc(y), with the integrating factor
J(y)= e-fcot()d y y =e1n lc sc(
y)I =lcsc(y) , and with I(y) = csc(y), the equation
I
becomes csc(y)u' -csc(y) cot(y)u =- csc2 (y), i.e., [csc(y)u]'= - csc 2 (y). Then
csc(y)u = cot(y)+ k ⇒ u= cos(y)+ ksin(y):::::} v= s )ksin
co (y) *
[cos(y)+ ksin(y)]y'= 1 ⇒ sin(y) - kcos(y)= x + k1 .
(y)
13. y" =2yy' and v(y) = y'(x) ⇒ vv' =2yv. If v =0, then y = c. If v -/= 0,
J
then v' = 2y ⇒ v = y2 + k ⇒ /�k = l⇒ Y2�k dy = x + c1. If k =0, then
_.!.y = x + c1 ⇒ y = X-+1C1 . If k =µ2 > 0, then y =µ tan(t) ⇒
J 2 + 2 dy =
1
y µ
J (u)
lµ dl = lµ = .l tan-1 µ ⇒ y = µtan[µ(x+c1)]. If k = -µ2 < 0,
µ
then J� y -µ
dy =_ 1
2µ J y µ
-1- - _1 dy = 1 (ln I
-
+y µ _
2µ Y- µI -ln IY+µI)= _
1 ln v-µ I =
2µ I +µ
y
2µx
X + C1 :::::} � = k1 e
2 µx
:::::} y(l - k1 e )= µ(k1 e +1) :::::} Y= µ�1__ :;:2t;.
2 µx e
1
14. y" - (y') 2 = 1 and z(x) = y'(x) ⇒ z' - z2 =l⇒ l�1 = 1 ⇒ tan- (z) = x+ c
J
⇒ z =tan(x+ c) ⇒ y = tan(x+ c) dx =ln I sec(x+ c)I + k.
15. v1f+Tl y" = land z(x) = y'(x) ⇒ v'1+z z' =l⇒ �(1 + z)31 2 =x + c ⇒
(1+ z)3 1 2 = �x+ c1 ⇒ z= (�x+ c1) 1 - 1
2 3
⇒ y =J (�x+ c1) 213 - 1 dx= � (�x+ c1) 513 - x+ c2.

16. e Y'y" =land z(x) = y'(x) ⇒ e z z' = l ⇒ e z =x+c ⇒ z = ln(x+ c) ⇒
J
y = ln(x+ c) dx. Let u =x+ c and integrate by parts to obtain
J l- ln(u) du= u ln(u) - u+ c1. Hence, y =(x+ c) ln(x+ c) - ln(x+ c)+ C1-
17. cos(y')y" =1 and z(x) =y'(x) ⇒ cos(z)z' = 1 ⇒ sin(z) = x+ c ⇒

z= sin-1(x+ c) ⇒ y = sin- 1 (x+ c) dx. Let u = x+ c and integrate by parts
J
to obtain l -sin-1(u) du= u sin- 1 (u)- �du= u sin- 1 (u)+✓1 - u2+c 1.
J J
Hence, y = (x+ c) sin- (x+ c)+ J1 - (x+ c)2 + c 1.
1
18. y" = y'jl - (y' ) 2 and v(y) = y'(x) => vv' = vvT"=v2. If v = 0, then y = c. If
v =/= 0, then ✓iv�v2 = 1 => sin- 1(v) = y+c1 => v = sin(y+c1) => csc(y+c1)y' = 1
=> - ln I csc(y + c,) + cot(y + c1)1 = x +k => csc(y + c1) +cot(y + c1) = k 1 e- . x

1. The standard form of the equation is y" + ¼Y' +(1 - 4;2 )y = 0, with p(x) = ¼
and Y1 = cosr) => u'(x) = �e- -4-- X Xl = sec 2(x)
J ¾ dx = co
yX Y1 s c)
=> = sec
u J2
(x) dx = tan(x) => y2 = u y
0 s )
1 = tan(xt 1x) = i� .
2 . y" +xy' +y = 0, p(x) = x, Y1 = e-x /2 ⇒ u'(x) = -tr e -f x dx = _ x2 / =e /

2 2 2
x 2 x 2
e e
=? U = J e ! dx =? Y2 = UY1 = e - x / J ex / dx.
2 x 2
2 2 2 2
3. The standard form of the equation is y" + (1 + 2x 1

)y' - 2; y = 0, with p(x) =
2
1 + .1...
� and Y1 = x - 1 / -x
2
e => u' ( x) = �
�
e - J l+t;, dx = xe 2xe - x -½ ln(x) = x1 / e => 2 x
u = J x1 / 2ex dx => Y2 = uy1 = x- 1 / e -x J x1 / e dx.

2 2 x
4. y" + y = cos(x). For y" + y = 0, y1 = cos(x) and Y 2

= sin(x).
(a ) YP = x[Acos(x) + Bsin(x)] =>
y; = [Acos(x) + B sin(x)] + x[-A sin(x) + B cos(x)],
y� = [-2Asin(x) +2Bcos(x)] + x[-Acos(x) - Bsin(x)],
y� + YP = -2Asin(x) + 2Bcos(x) = cos(x) => A = 0 and B 1
2*
YP = ½xsin(x) => y = ½xsin(x) + c1 cos(x) + c2 sin(x).
(b) W(x) = 1, f(x) = cos(x),
u 1 = - J sin(x)cos(x) dx = J sin(2x) dx = ¼cos(2x),
-½
J J
u = cos (x) dx = ½ 1 +cos(2x) dx = ½x + ¼sin(2x),
2
2
YP = ¼cos(2x)cos(x) + [½x + ¼sin(2x)]sin(x) = ½xsin(x) + ¼cos(x), or

YP = ½xsin(x) => y = ½xsin(x) + c1 cos(x) + c2 sin(x).
5. y" - y' - 2y = 8x2 . For y" - y' - 2y = 0, Y1 = e -x and Y 2 = e
2x
.
6. v" -V = cos(x) - 5sin(2x). For v" - v = 0, Vi= ex and v2 = e-x _
(a) For v" - v = cos(x), VPi =Acos(x)+Bsin(x), vi1 =-Acos(x)- Bsin(x),

vi1 - VPi = -2Acos(x) - 2Bsin(x) = cos(x) =} A = -½, B = O ⇒
Vp 1 =-½cos(x). For v" - v =-5sin(2x), Vp2 = Acos(2x) + Bsin(2x),
vi2 =-4Acos(2x) - 4B sin(2x),
vi2 - Vp2 = -5Acos(2x) - 5B sin(2x) =-5sin(2x) ⇒ A = 0, B = l ⇒
Vp2 =sin(2x) =}VP -½cos(x) +sin(2x) =}
=
V = -½cos(x)+ sin(2x) +c 1 ex + c2 e-x .

(b) W(x) = -2, f(x) =cos(x) - 5sin(2x),
u1 =½Je-x [cos(x) - 5sin(2x)] dx =½Je-x cos(x) dx- � Je-x sin(2x) dx,
u2 = -½Jex [cos(x) - 5sin(2x)] dx =-½ Jex cos(x) dx+� Jex sin(2x) dx.
J e-xcos(x) dx
= -e-x cos(x) - J e-x sin(x) dx =-e-x cos(x) + e-x sin(x) - J e-xcos(x)
=} J e- cos(x) dx=½e- [sin(x) -cos(x)],
x x
Jexcos(x) dx
=ex cos(x) +Jex sin(x) dx = ex cos(x) + ex sin(x) -Jex cos(x)
=}Jex cos(x) dx =½ex [cos(x)+sin(x)],
Je-x sin(2x) dx =-e- x sin(2x) +2Je- x cos(2x) dx
= -e-x sin(2x) - 2e-x cos(2x) - 4 Je-x sin(2x)
=} J e- sin(2x) dx =-ge- [sin(2x) + 2cos(2x)],
x x
x sin(2x) dx
Je
= ex sin(2x) - 2Jex cos(2x) dx=ex sin(2x) - 2ex cos(2x) - 4 Jex sin(2x)
=} Je sin(2x) dx = ge [sin(2x) - 2cos(2x)],
x x
u1 = ¼e-x [sin(x) -cos(x)] +½e- x [sin(2x) + 2cos(2x)],

u2 = -¼ex [cos(x) +sin(x)]+ ½ex [sin(2x) - 2cos(2x)],
VP=-½cos(x) +sin(2x), and v=-½cos(x) + sin(2x) + c1ex + c2e-x .
7. v" - 2v' - Sv = 24(x2 + l)e4x .

For v" - 2v' - 8v = 0, V1=e- 2x and V2 = e4x .
(a) VP =x(A +Bx+Cx2 )e4x=(Ax+ Bx2 + Cx3 )e4x ,

v� =[ A+ (4A+ 2 B)x+ (4B + 3C)x2 +4Cx3]e4x ,
vi= [(SA+2B) +(16A +16B+6C)x +(l6B +24C)x2 + l6Cx3]e4x,
vi - 2v; - 8vP= [(6A+ 2B) + (12B+ 6C)x+ 18Cx2 ]e4x = 24(x2 +l)e4x
=} C = :! B =-�, A 38 =}
VP = ( 398x _ �3 x2 + :!3 x3 ) e4x
3' 9
=
=} V = (3 8X - �X + 1x ) e
3
9
2 3 4x
+ C1e-2x +C2 e4x _
-
(b) W(x)=l : �: ::: 1=6e 2x ,u1 = -4f(x2 +l)e6x dx
-2 4
1
=-�(x2 + l)e6x + Jxe6x dx = -�(x2 + l)e6x + ixe6x - Je6x dxi
=-�3 (x2 + l)e6x + �9 xe6x _ _!27 19 ) e6x
_e6x = -�3 x2 + �9 x _ 27
t
u2 =4Jx + 1 dx = 1x + 4x, VP = (- �x + ix - ��) + ( 1x3 +4x)] e4x,
'
2 3 2
8. (a) a(x - x0) 2 y" + b(x - x0)y' + cy = 0, l = x - x0 and y(x) = z(t) ⇒

dz dl dz i!JJ. d2 z dt d2 z.
dx = dl dx = dl and dx2 = dl 2 dx = dt 2 Hence ' the equation for y(x)
rJ:1!.
transforms into the Euler equation ai 2 z" + btz' + cz = 0 for z(t).

(b) t = x - 3 and y(x) = z(t) transforms 2(x - 3)2 y" + 3(x - 3)y' + y = 0 into
2l 2 z"+3tz'+z = 0. For t> 0, z = tr ⇒ 2r(r-1)+3r+l = 2r2 +r+l = 0 ⇒
r = -l�iv'7 = -¼ ±i ./} ⇒ z = ltl-1/4 [C 1 cos ( ./} ln ILi) + c2 sin ( ./} ln ltl)]
fort -I- 0 ⇒ y = Ix - 31- 1 /4 [c 1 cos ( ./} ln Ix - 31) + c2 sin ( ./} ln Ix - 31)]
for X-=/- 3.
9. (a) ax2 y" +bxy' +cy = 0, x> 0, l = ln(x) and y(x) = z(l) ⇒ � = = �1: �1 ;,
�:¾ = �:�(;) + �1(�}). Hence, the Euler equation for y(x) transforms into
2
az" + (b - a)z' + cz = 0, which has constant coefficients.

(b) z = erl ⇒ ar2 + (b - a)r + c = 0, and y = xr ⇒ ar(r - 1) + br + c = 0 ⇒
ar2+ (b - a)r + c = 0.
e = er ln(x) = e'n(x ) = xr ⇒ z = C 1 er 1 t + C2er2 t = C 1 Xr1 + C2 Xr2 = Y,
r
(c) rt
z = ert (c1 + c2 t) = xr [c 1 + c2 ln(x)] = y, and

z = eat [c1 cos(,Bl) + c2 sin(,Bt)] = x0 { c1 cos[,8 ln(x)] + C2 sin[,8 ln(x)]} = y.
10. x2 y11 - xy' + y = l n(x) , X> 0.
(a) For x2 y" - xy' + y = 0, y 1 = x and Y2 = x ln(x).

(
W(x) = I � l:(�) �l / = x,J(x) =x ,;(x)' u 1 = -J;dx = -ln(x),
u 2 = J xl;(x) dx = ln I ln(x)I, YP = -x ln(x) + x ln(x) ln I ln(x)I, or
YP = x ln(x) ln I ln(x)I ⇒ y = x ln(x) ln I ln(x)I + c1x + C2 X ln(x).
(b) t = ln(x) and = z(l) ⇒ z" - 2z' + z =
y(x) t·
(c) For z" - 2z' + z = 0, z = ert ⇒ r2 - 2r + 1 = (r - 1)2 = 0 ⇒ z1 = et
e
and z2 = leL . W(t) = I : t
Let
e e + teL
I = e2t , J(t) = u 1 = - J 1 dt = -t, t,
u 2 = Ji dt = ln ltl, zp = -tet + tet ln Ill, or Zp = tet ln ltl
⇒ z = let ln ltl + c1et + c2teL .
(d) y(x) = z(l) = z(ln(x)) = x ln(x) ln I ln(x)I + c 1 x + c2x ln(x), as in part (a).
11. 4x2 y" + 4xy' - y = x2lxl 1 1 2.
(a) For 4x2 y" + 4xy' - y = 0, x > 0, y = xr ⇒ 4r(r - 1) + 4r - 1 = O �

4r 2 - 1 = 0 ⇒ Y1 = x 1 /2 and y2 = x-1/2 .
I x� 1 /2 x- 1 /32 I = -x-1, (x) = x�12'
W(x) = - 1 /2 - x- /2 J
½ ½
- .4!. Xd X - x8, U2 -
U1 - J J
-4l X 2dX - - x1 2 ,
2
- 3
YP = x8 x l /2 _ x12 X-1/2 = x241 ' y = x241 + C 1 xl/2 + C2 X- 1/2 ·

2 3 5 2 5 2
12. y" + y'=2x.
(a) For y" + y' = 0, Y1 = 1 and Y2 = e-x . YP = x (A + Bx) = Ax + Bx2,

y� = A+2Bx, vi=2B, vi+y� = (A+2B)+2Bx = 2x ⇒ B = 1, A=-2
⇒ YP =-2x + x2 ⇒ y=x2 - 2x + C1 + c2e-x .
e-x
1
(b) W(x) =
0 -e-x = -e- , J(x) = 2x, u1 = 2J xdx = x ,
1
x 2
J
U2 = -2 xexdx = -2xex +2ex , Yp = x2 + (-2xex +2ex )e-x = x2 -2x+2,
I
or YP = x2 - 2x, and y=x2 - 2x + c1 + c2e-x.

(c) z(x) = y'(x) =} z' + z = 2x, J(x) = ef ldx =ex =} (exz)' = 2xex
⇒ exz = 2 xexdx = 2xex - 2ex + c ⇒ z = 2x - 2 + ce-x
J
⇒ y = x2 - 2x - ce-x + k.
13. x2y" + xy' = 1, x > 0.
1 and Y2 I� ln )
I
(a) For x2y" + xy' = 0, Y 1 = = ln(x). W(x) = t =
¾,
J
J(x) = }2 , U1 = - Ini ) dx = -½[ln(x)]2, u 2 = ¾ dx = ln(x),
x
J
YP = -½[ln(x)] 2 + [ln(x)] 2 = ½[ln(x)] 2 =} y = ½[ln(x)] 2 + c 1 + c2ln(x).
(b) z(x) = y'(x) =} x2z'+xz = 1 =} z'+¾z = ;2 , I(x) = ef½dx = eln(x) = x =}

(xz)' = ¾ =} xz = ln(x) + c =} z = Ini ) + � =} y = ½[ln(x)] 2 + cln(x) + k.
x
14. 3y'y" = 1.
(a) z(x) = y'(x) ⇒ 3zz' = 1 =} �z2 = x+c1 =} z2 = �x+c2 ⇒ z = (�x + c2 )

112
⇒ y = J (�x + c2) 1 dx = (�x + c2) 1 + c.

1 2 3 2
(b) v(y) = y'(x) ⇒ 3v2v' = 1 ⇒ v3 = y+k1 ⇒ v = (y+k1)1 l 3 ⇒ (y+k� ) 113 = 1

⇒ � (y + k 1 )2 l3 =x + k2 ⇒ y= (�x + k3)3 12
- k1.
15. (a) y�+ p(x)y� + q(x)y2 = 0 and y� + p(x)y� + q(x)y1 = 0

==> yi[y� + p(x)y� + q(x)y2] = 0 and Y2 [Y� + p(x)y� + q(x)y 1] = 0
==> (Y1Y� - y�y2 ) + p(x)(Y1Y� - Y�Y2 ) = 0 ==> W'(x) + p(x)W(x) = 0.
(b) I(x) = efp(x)dx ⇒ [efp(x)dxw(x)]' = 0 ⇒ efp(x)dxw(x) = c
==> W(x) = ce-fp(x)dx_
=
(c) If c = 0, then W (x) 0 on I, and if c-/- 0, then W(x)-/- 0 for any x in I
since e-fp(x)dx > 0 for any continuous function p(x).
(d) Suppose that Y1 and Y2 are linearly dependent on I. Then there exist
constants c1 and c2 , not both 0, such that c1y1 + c2 y 2 = 0 on I. If c1 -/- 0,
then Y1 = - :Y2 ==> y� = -:y� ==> W (x) = Y1Y� - Y�Y2 = 0. Similarly
if c2 -/- 0. Conversely, if W(x) = y1y� - y�y2 = Y2 = Yi ==>
0 on I, then Y2
YI
ln IY2I = ln IY1I + c ==> Y2 = ky1 , i.e., Y1 and Y2 are linearly dependent on I.
(e) Suppose that Y1 and Y2 are linearly independent on I. If W (x) 0 on =
I, then, by part (d), y 1 and Y2 are linearly dependent on I, which is a
contradiction. Hence, W(x) -I- 0 on I. Conversely, suppose that W(x) -I- 0
on I. If y 1 and y 2 are linearly dependent on I, then, by part (d), W(x) 0 =
on I, which is a contradiction. Hence, y 1 and y 2 are linearly independent
on I.
16. x2y" - xy' = 0 on the interval I= (-1, 1).

(a) For x > 0, y = xr ==> r 2 - 2r = 0 ==> y1 = 1 and Y2 = x2 , valid for all x.
(b) W ( x) = I � ;; I = 2x.
(c) At x0 = 0, W(0) = 0.
(d) The result of part (c) does not contradict the result of part (e) in Exercise
15 because the standard form of the equation is y" -¾y' = 0, and p(x) = -¾
is discontinuous at x = 0, hence not continuous on I.

Section 4.1 Exercises - page 101
1. y"' - 6y" + lly' - 6y = 0, y = erx ==> P(r) = r3 - 6r 2 + llr - 6 = 0 ==>
(r - l)(r2 - 5r + 6) = (r - l)(r - 2)(r - 3) = 0 ==> y = c1ex + c2e2x + C3e3x .
- 2y" - y' + 2y = 0, y = erx ==> P(r) = r3 - 2r - r + 2 = 0 ==>

2
2. y"'
(r - l)(r - r - 2) = (r - l)(r + l)(r - 2) = 0 ==> y = c1ex + c2 e- x + c3e2x
2
⇒ y' = C1ex - C2e-x + 2c3e2x , y" = C1ex + C 2e-x + 4c3e2x , y (0) = 6, y'(0) = 0,
C1 + C2 + C3 =
y " (0) = 0 =} { C1 - C 2 + 2c3 = � } a> c1 = 6, c, = 2, c, = -2, and
C1 + C2 + 4c3 =
y = 6ex
+ 2e- x
- 2e2x .
3. 9y111 + 9y" - y' - y = 0, y = erx P(r) = 9r3

=} + 9r2 - r - 1 = 0 =}
(r + 1)(9r2 - 1) = (r + 1)(3r -1)(3r + 1) = 0 =} y = c1e- + c 2 e½ + c3 e-½ .
x x x
4. - 5y " + 8y ' - 4y
y "' = 0, y = erx =} P(r) = r3 - 5r2 + 8r - 4 = 0 =}
(r - l)(r2 - 4r + 4) = (r - l)(r - 2)2 = 0 =} y = c1ex + c2e2x + c3 xe2x.
5. - 6y " + 12y' - 8y = 0, y = erx =} P(r) = r3 - 6r2 + 12r - 8 = 0
y '" =}
(r - 2)(r2 - 4r + 4) = (r - 2)3 = 0 =} y = c1e2x + c2 xe2x + C3X2 e2x .
6. y "'+ y " - y' - y = 0, y = erx =} P(r) = r3 + r2 - r - 1 = 0 =}

(r - l)(r2 + 2r + 1) = (r - l)(r + 1)2 = 0 =} y = c1ex + c2 e- x + C3 Xe-x =}
y ' = C1e - C 2 e- + C 3(e- - xe- ) y
x x x x
C1ex + 2e x + c3(-2e- x + xe-x),
� ; c: � �
1
y(0) = 0, y'(0) = 3, y "(0) = 2 =} { C1 - C2 + C3 = 3 } =} C1 = 2, C2 = -2,
C1 + C 2 - 2c3 = 2
C3 = -1, and y = 2ex - 2e-x - xe-x.
7. - 5y " + 17y' - 13y = 0, y = erx =} P(r) = r3 - 5r 2 + 17r -13 = 0

y "' =}
(r - l)(r 2 - 4r + 13) = 0 ⇒ r 1 = l and r2 , r2 = 4±f36 = 2 ± 3i =}

y = c1ex + e2x[c 2 cos(3x) + c3 sin(3x)].
8. y(4) - 5y " + 4 y = 0, y = erx =} P(r) = r4 - 5r2 + 4 = (r2 - l)(r 2 - 4) =0 =}
(r - l)(r + l)(r - 2)(r + 2) = 0 =} y = C1ex + c 2 e- x + C3 e2x + C4e-2x.
10. y(4) _ 7y "' + 17y" -17y' +6y = 0, y = erx =} P(r) = r4 -7r3 +17r2 -17r+6 =0
=} (r - l)(r - 2) (r - 4r + 3) = (r - 1) (r - 2)(r - 3) = 0
2 2
=} Y = C1e + C2Xe + C3 e + C4e .

x x 2x 3x
11. y(4) - y "' - 3y " + 5y' - 2y = 0, y = erx =} P(r) = r4 - r3 - 3r2 + 5r - 2 = 0 =}

(r-l)(r+2)(r 2 -2r+l) = (r-1)3 (r+2) = 0 =} y = c1ex +c2 xex+c3 x2 ex +c4e-2x .
12. y( ) - 4y "' + 6y " - 4y ' + y = 0, y = erx ⇒ P(r) = r4 - 4r3 + 6r2 - 4r + 1 = 0.

4
Since r1 = 1 is a root of both P(r) and P'(r) = 4r3 - 12r2 + 12r - 4, (r - 1)2
is a factor of P(r). Hence, P(r) = (r - 1)2 (r2 - 2r + 1) = (r - 1)4 = 0
x 2 x
=} y= C1e + C2Xe + C3X e + C4 X e .
x 3 x
13. y( ) _ 6y111+ 13y" -12y'+4y = 0, y = erx ⇒ P(r) = r4 - 6r3 + 13r2 - 12r+4

4
=0
⇒ (r - l)(r - 2)(r2 - 3r + 2)= (r - 1)2(r - 2)2 = 0
=} Y = C1 e + C2Xe + C3 e
x x 2x
+ C4Xe2x .
14. y( ) + 2y'" + 4y " - 2y' - 5y = 0, y = erx ⇒ P(r) = r4 + 2r3 + 4r2 - 2r - 5 = 0 ⇒

4
(r - l)(r + l)(r2 + 2r + 5) = 0 ⇒ r1 = 1, r2 = -1, r3 , r3 = - 2±F15 = -1 ± 2i

⇒ y = c,ex + c2e-x + C3 e-x cos(2x) + C4 e- x sin(2x).
15. y( ) + lOy '" + 41y" + 76y ' + 52y = 0, y = erx =}

4
P(r) = r4 + 10r3 + 41r2 + 76r + 52 = 0. Since r 1 = -2 is a root of both

P(r) and P'(r) = 4r3 + 30r2 + 82r + 76, (r + 2)2 is a factor of P(r). Hence,
P(r) = (r + 2)2 (r2 + 6r + 13) = 0 ⇒ r1 = -2, r2 , r2 = -B±F15 = -3 ± 2i ⇒
y = c1e-
2x
+ c2xe-2x + C3 e-3x cos(2x) + C4 e-3x sin(2x).
16. y( ) + 5y " + 4y = 0, y = erx ⇒ P(r) = r4 + 5r2 + 4 = (r2 + l)(r2 + 4) = 0 ⇒

4
r1, r1 = ±i, r2 , r2 = ±2i ⇒ y = c 1 cos(x) + c2 sin(x) + C3 cos(2x) + C4 sin(2x).
17. y( ) + 4y = 0, y = erx ⇒ P(r) = r4 + 4 = 0. P(r) = (r2 + ar + 2)(r2 + br + 2)

4
⇒a= ±2, b = =F2 ⇒ P(r) = (r2 + 2r + 2)(r2 - 2r + 2) = 0 ⇒ r 1 , r 1 = -1 ± i,

r2 , r2= 1 ± i ⇒ y= c 1 e-x cos(x) + c2 e-x sin(x) + c3ex cos(x) + C4ex sin(x).
18. y( -8y '"+26y "-40y '+25y = 0, y = erx ⇒ P(r) = r4 -8r3 +26r2 -40r+25 = 0.
4l
P(r) = (r2 +ar+ 5)(r2 +br+ 5) ⇒a= b = -4 ⇒ P(r) = (r2 - 4r + 5)2 = 0 ⇒

r1, r1= 2 ± i ⇒ y= c1e2x cos(x) + c2 e2x sin(x) + c3 xe2x cos(x) + C4Xe2x sin(x).
19. x3y 111 + x2y " - 2xy ' + 2y = 0. For x > 0, y = xr ⇒

P(r) = r(r - 1)(r - 2) + r(r - 1) - 2r + 2 = ( r - 1)[r(r - 2) + r - 2] =
(r - l)[r2 - r - 2 ] = (r - l)(r - 2)(r + 1) = 0 ⇒ y = c 1 x + c2x2 + c3 x- 1, x-/= 0.
20. x3y '" - 2x2 y" + 3xy ' - 3y= 0. For x > 0, y= xr ⇒
P(r) = r(r - l)(r - 2) - 2r(r - 1) + 3r - 3 = (r - l)[r(r - 2) - 2r + 3] =
(r - 1)2(r - 3) = 0 ⇒ y = c1x + c2x ln lxl + c3 x3, x-/= 0.
21. x3y 111 + xy ' - y = 0. For x > 0, y = xr ⇒ P(r) = r(r - l)(r - 2) + r - 1 = 0 ⇒
(r - l)[r(r - 2) + 1] = (r - 1)3 = 0 ⇒ y = c 1 x + c2x ln lxl + C3X(ln lxl)2, x-/= 0.
22. x3y111 + 8x2y " + 13xy ' - 13y = 0. For x > 0, y = xr ⇒
P(r) = r(r - l)(r - 2) + 8r(r - 1) + 13r - 13 = (r - l)[r(r - 2) + 8r + 13]
= (r - l)(r2 + 6r + 13) = 0 ⇒ r1 = 1 and r2, r2 = - 5 ±F15 = -3 ± 2i ⇒
y = c 1 x + c2 x-3 cos[2 ln(x)] + c3 x-3 sin[2 ln(x)],
y ' = c 1 + (2c3 - 3c2 )x- cos[2 ln(x)] - (2c2 + 3c3 )x- sin[2 ln(x)],
4 4
y " = [-4(2c3 - 3c2 ) - 2(2c2 + 3c3)Jx-5 cos[2 ln(x)]

+ [4(2c2 + 3c3) - 2(2c3 - 3c2)Jx-5 sin[2 ln(x)], or
5 5
y " = (8c2 - 14c3 )x- cos [2 ln (x)] + (14c2 + 8c3 )x- sin[2 ln(x)]. Then y (l) = 0,
y'(l) = 6 and y"(l) = -2 ⇒ c1 +c2 = 0, c1-3c2 +2c3 = 6 and 8c2- 14c3 = -2 ⇒

c1 = 2, c2 = -2 and c3 = -1. Hence, y = 2x - 2x-3 cos[2 ln(x)J- x-3 sin[2 ln(x)].
23. x4 y( 4 ) + x3 y + x2 y" - 2xy ' + 2y = 0. For x > 0, y = xr ⇒

m
P(r) = r(r - l)(r - 2)(r - 3) + r(r - l)(r - 2) + r(r - l) - 2r + 2 = 0 ⇒

(r - l)[r(r - 2)(r - 3) + r(r - 2) + r - 2] = (r - l)(r - 2)[r(r - 3) + r + 1] =
(r - l)3(r - 2) = 0 ⇒ y = c1x + C2X In lxl + c3 x(ln lxl) 2 + c4 x2 , x =/= 0.
24. x4y( 4 ) - x3 y + 5x2 y " - lO xy ' + lOy = 0. For x > 0, y = xr ⇒

m
P(r) = r(r - l)(r - 2)(r - 3) - r(r - l)(r - 2) + 5r(r - l) - lOr + 10 = 0 ⇒

(r - l)[r(r - 2)(r - 3) - r(r - 2) + 5r - 10] = (r - l)(r - 2)[r(r - 3) - r + 5] =
(r - l)(r - 2)(r2 - 4r + 5) = 0 ⇒ r1 = 1, r2 = 2 and r3, r3 = 4±'(-4 = 2 ± i
⇒ y = c1x + c2x2 + c3x2 cos(ln lxl) + C4X2 sin(ln lxl), x =/= 0.
25. x4 y(4 ) + 6x3 y"' + 6x2 y" - 2xy' + 2y = 0. For x > 0, y = xr ⇒
P ( r) = r(r - l)(r - 2)(r - 3) + 6r(r - l)(r - 2) + 6r(r - l) - 2r + 2 = 0 ⇒
(r - l)[r(r - 2)(r - 3) + 6r(r - 2) + 6r - 2] = (r - l)(r3 + r2 - 2)
= (r - 1)2 (r2 + 2r + 2) = 0 ⇒ r1 = 1 and r2 , r2 = - 2±2vGI = -1 ± i
⇒ y = c1x + c2x ln lxl + c3x-1 cos(ln lxl) + c4x-1 sin(ln lxl), x =/= 0.
26. x4 y (4 ) + 2x3 y "' + 3x2 y" - 3xy ' + 4y = 0. For x > 0, y = xr ⇒
P(r) = r(r - l)(r - 2)(r - 3) + 2r(r - l)(r - 2) + 3r(r - 1) - 3r + 4 = 0.
P(r) = r4 - 4r3 + 8r2 - 8r + 4 = (r2 + ar + 2)(r2 + br + 2) ⇒ a= b = -2
⇒ r4 - 4r3 + 8r2 - 8r + 4 = (r2 - 2r + 2)2 = 0 ⇒ r1, r1 = 2±-(-4 = 1 ± i
⇒ y = c1x cos(ln lxl) + c2 x sin(ln lxl) + c3x ln lxl cos(ln lxl) + C4X ln lxl sin(ln lxl)-
Section 4.2 Exercises - page 118

l. y"' - 3y " - y ' + 3y = 8e 2x . For y - 3y" - y' + 3y = 0, y = erx =}
m
P(r) = (r2 -l)(r -3) = 0 ⇒ Y1 = ex , Y2 = e -x and y3 = e3x . With J(x) = 8e 2x ,

e -x
x 3x
e e
W(x) = e -e - 3e
x x 3x
= ex [-9e 2x - 3e2x ] - e -x[9e4x - 3e4x ] + e3x[2] =
x 9 3x
e-
x
e e
3x
-l 6 e '
x 3x
0 e- e
�1(x) = 0 -e- 3e3x = J(x)[3e2x + e 2x ] = 4e 2x J(x) = 32e4x ,
x
x 9 3x
f(x) e- e
3x
0 e
0 3e 3x = -f(x)[3e4x_e4x ] = -2e4xf(x) = -l6e6x ,
e
x
f(x) 9e3x
ex e-x 0
� 3(x)= e -e-x O =f(x)[-1 - 1]= -2f(x)= -16e2x ,
x
ex e-x f(x)
U1 = -2Je dx = -2ex , U2 = Je3x dx= ½e3x , U3 =Je-x dx=-e-x
x ⇒
Yp = -!e2x ⇒ y =-!e2x + C1ex + C2 e-x + C3 e3x _
2. y"' - 3y" + 2y' =8xe 2x . For y'" - 3y" + 2y' =0, y = erx ⇒
P(r) = r(r - l)(r- 2)= 0 ⇒ Y1 = 1, Y2 =ex and y3 =e2x . With J(x)=8xe2x ,
1 ex e2x
W(x)= 0 ex 2e2x = 2e3x,� 1 (x)= e3x f(x)= 8xe5x ,
0 ex 4e2x
� 2 (x)= -2e2x f(x)=-16xe4x ,� 3(x) = ex f(x)= 8xe3x ,
UJ =4Jxe2x dx = 2xe2x - e2x , U2 = -8Jxex dx=-8xex + 8ex ,
u3 = 4Jx dx = 2x2 ⇒
YP = (2xe2x - e2x ) + (8ex - 8xex )ex + 2x2 e2x = 7e2x - 6xe2X + 2x2 e2x , or
YP = (2x2 - 6x)e2x ⇒ y =(2x2 - 6x)e2x + C1 + c2 ex + c3 e 2x.
3. y'" - 2y" + y' =-2ex cos(x). For y'" - 2y" + y' =0, y =erx ⇒
P(r) =r(r-1)2 = 0 ⇒ Y1 = 1, Y2 = ex and y3 =xex . With J(x)= -2ex cos(x),
1 ex xe x
W(x)= 0 ex ex + xex =e2x ,� 1 (x)=e2x f(x)=-2e3x cos(x),
0 ex 2ex + xex
� 2 (x) =-(1 + x)ex f(x) = 2(1 + x)e2x cos(x),� 3 (x)= ex J(x)=-2e2x cos(x),
u1 =-2Jex cos(x) dx= -2ex cos(x) - 2Jex sin(x) dx
= -2ex cos(x) - 2ex sin(x) + 2Jex cos(x) dx
⇒ -4J ex cos(x) dx = -2ex [cos(x) +sin(x)] ⇒ u1 = -ex [cos(x) + sin(x)],
u2 =2J(1 + x) cos(x) dx = 2(1 + x)sin(x) + 2 cos(x),
u3 =-2Jcos(x) dx= -2sin(x)
⇒ YP= -ex [cos(x) + sin(x)] + [2(1 + x)sin(x) + 2 cos(x)]ex - 2xex sin(x)
= ex [cos(x) + sin(x)] ⇒ y =ex [cos(x) +sin(x)] + c 1 + c2 ex + c3xex .
4. y'" - y" + 2y = l0ex . For y'" - y" + 2y= 0, y = erx ⇒
P(r) = (r + l)(r-2 - 2r + 2) = 0 ⇒ r- 1 = -1 and r = 1 ± i ⇒ Y1 = e-x ,
Y2 = ex cos(x) and Y2 =ex sin(x). With J(x)= l0ex ,
e-x ex cos(x) ex sin(x)
W(x)= -e- e cos(x) - e sin(x) e sin(x) + ex cos(x)
x x x x
e-x -2ex sin(x) 2ex cos(x)

=e-x [2e2x ] + e-x [2e2x ] + e-x [e2x]=5ex ,� 1 (x)= e2x J(x)= 10e3x ,
� 2 (x)=-[2sin(x) + cos(x)]f(x) =-10ex [2sin(x) + cos(x)],
� 3(x) = [2 cos(x) -sin(x)]J(x) = 10ex [2 cos(x) -sin(x)],
u1 = 2Je2x dx= e2"', u2 =-2J2sin(x) + cos(x) dx =4cos(x) - 2sin(x),
u3 =2J2 cos(x) -sin(x) dx=4sin(x) + 2 cos(x) ⇒
YP = e2x [e-x ]+[4 cos(x)-2sin(x)][ex cos(x)]+[4sin(x)+2 cos(x)][exsin(x)] = 5ex
⇒ y = 5ex + c1e-x + Czex cos(x) + c3 ex sin(x).
5. y"'- y" + y' - y = 4sin(x) + 4cos(x). For y"' - y" + y' - y = 0, y = erx =}
P(r) = (r-l)(r2 + 1) =} y1 = ex , Y2 = cos(x) and y3 = sin(x). With
ex cos(x) sin(x)
J(x) = 4[sin(x) + cos(x)], W(x) = e -sin(x)
x
cos(x) = 2ex ,
e - cos(x) -sin(x)
x
�1(x) = f(x) = 4[sin(x) + cos(x)],

� 2 (x) = -ex [cos(x)-sin(x)]f(x) = -4ex [cos(x) -sin(x)l[sin(x) + cos(x)]
= -4ex [cos2 (x) -sin2 (x)] = -4ex cos(2x),
� 3 (x) = -ex [sin(x) + cos(x)]f(x) = -4ex [sin(x) + cos(x)] 2 = -4ex [l + sin(2x)],
u1 = 2 J e- x [sin(x) + cos(x)] dx = -2e- x cos(x),
u2 = -2 J cos(2x) dx = - sin(2x),
u 3 = -2 J l + sin(2x1 dx = -2x + cos(2x) ⇒
YP = -2cos(x) -sin(2x) cos(x) + [-2x + cos(2x)] sin(x)
= -2cos(x)- 2xsin(x) + [sin(x)cos(2x)-cos(x)sin(2x)]
= -2cos(x) - 2xsin(x)-sin(x), or YP = -2xsin(x)
⇒ y = -2xsin(x) + c1 ex + c2 cos(x) + c3 sin(x).
6. y"' - y" = 20x3 . For y"' - y" = 0, y = erx P(r) = r2 (r - 1) = 0 =}
=}
l X ex
Y1 = 1, Y2 = x and y3 = e . With J(x) = 20x , W(x) = 0 1 ex = ex ,
x 3
0 0 ex
�1 (x) = (xe - e )f(x) = 20(x - x )e , � 2 (x) = -e f(x) = -20x e ,
x x 4 3 x x 3 x
�3(x) = f(x) = 20x , u1 = 20 J x4 - x3 dx = 4x5 - 5x4 ,

3
u2 = -20 J x3 dx = -5x4 , u 3 = 20 J x3 e-x dx = -20x3 e- x + 60 J x2 e- x dx =

-20x3 e- x - 60x2 e- x + 120 f xe- x dx = -20x3 e-x - 60x2 e-x - l20xe- x-l20e- x
⇒ YP = (4x5 - 5x4 ) - 5x5 - [ 20x3 + 60x2 + 120x + 120]
= -x5 - 5x4 - 20x3 - 60x2- 120x- 120, or YP = -(x5 + 5x4 + 20x3 + 60x2 ) ⇒
y = -(x5 + 5x4 + 20x3 + 60x2 ) + c1 + c2 x + c3 ex is the general soluti on. Then
y' = -(5x4 + 20x3 + 60x2 + 120x) + C2 + c3 ex ,
y" = -(20x3 + 60x2 + 120x + 120) + c3 ex , and y(0) = 20, y'(0) = 0 and
y"(0) = -20 ⇒ c1 + c3 = 20, c2 + c3 = 0 and -120 + C3 = -20 ⇒ c1 = -80,
c 2 = -100 and c3 = 100 =} y = -(x5 + 5x4 + 20x3 + 60x2 ) - 80 - lO0x + lO0ex .
7. y"'-2y" = 5sin(x). For y"'-2y" = 0, y = erx =} P(r) = r2 (r-2) = 0 =} Y1 = 1,

l X e 2x
Y2 = x and y3 = e2x . With J(x) = 5sin(x), W (x) = 0 1 2e2x = 4e2x ,
0 0 4e2x
� 1 (x) = [ 2xe2x - e2x ]f(x) = 5(2x -l)e2x sin(x),
�2 (x) = -2e2xf(x) = -10 e2x sin(x), � 3 (x) = f(x) = 5sin(x),
J
u1 = ¾ (2x - 1)sin(x) dx = -¾(2x- 1) cos(x) + � sin(x),
J
u2 = -� sin(x) = � cos(x),
J J
u3 = ¾ e- 2x sin(x) dx = -¾e- 2x cos(x) - � e- 2x cos(x) dx
J J
= -¾e- 2x cos(x) - �e- 2x sin(x) - 5 e- 2x sin(x) dx =} 245 e- 2x sin(x) dx =
- e-2 x [�sin(x) + ¾ cos(x)] ⇒ u3 = -e-2x [½sin(x) + ¼cos(x)] ⇒

YP = -¾(2x- l)cos(x) + �sin(x) + �xcos(x)- [½sin(x) + ¼cos(x)]
= 2sin(x) + cos(x) ⇒ y = 2sin(x) + cos(x) + c1 + c2 x + c3 e2x .
8. y(4) - y" = 4
2. For y( ) - y" = 0, y P(r) = r2 (r2 - 1) = 0 ⇒ Y 1 = 1,
= erx ⇒
l X ex e-x
O l e -e-x
x
Y2 = x, y3 = ex and y4 = e- x ⇒ W(x) = = -2, and with
0 0 ex e-x
0 0 ex -e-x
J(x) = 2, 61(x) = -2xf(x) = -4x, 62 (x) = 2f(x) = 4,
63 (x) = -e- x f(x) = -2e-x , 64 (x) = ex f(x) = 2ex , U 1 = 2J xdx = x2 ,
U2 = - J 2dx = -2x, U3 = Je- x dx = -e-x , U4 = - Jex dx = -ex ⇒
YP = -x2 - 2, or YP = -x2 ⇒ y = -x2 + C 1 + C2 X + c3 ex + C4e-x.
9. y(4) + y" = 6x. For y( 4) + y" = 0, y = erx =? P(r) = r 2 (r2 + 1) = 0 =? Y1 = 1,
Y2 = x, y3 = cos(x) and y4 = sin(x) ⇒ W(x) =
l x cos(x) sin(x)
0 1 - sin(x) cos(x) . .
( ) . ( ) = -sm (x)[-sm(x)] -cos(x)[-cos(x)] = 1
0 0 - COS X - Slll X
0 0 sin(x) -cos(x)
(along the fourth row) and, with f(x) = 6x, 6 1 (x) = -f(x)[x - O] = -6x2 ,
62 (x) = f(x)[l] = 6x, 63 (x) = -f(x)[-sin(x)] = 6xsin(x),
64(x) = f(x)[-cos(x)] = -6xcos(x), u1 = -6J x2 dx = -2x3 ,
u2 = 6J xdx = 3x2 , u3 = 6J xsin(x)dx = -6xcos(x) + 6sin(x),
U4 = -6J xcos(x)dx = -6xsin(x) - 6cos(x) ⇒
YP = x3 + [6sin(x) - 6x cos(x)] cos(x) - [6cos(x) + 6x sin(x)] sin(x) = x3 - 6x,
or YP = x3 ⇒ y = x3 + c 1 + c2 x + c3 cos(x) + C4 sin(x).
10. x3 y111 - x2 y" + 2xy' - 2y = x. For x3 y"' - x2 y" + 2xy' - 2y = 0, x > 0, y = xr ⇒

P ( r) = ( r - l )2 (r - 2) = 0 ⇒ Y 1 = x, Y2 = x ln(x) and y3 = x2 . With f ( x) = x\ ,
x x ln(x) x2
W(x) = 1 ln(x) + 1 2x = x[ 2ln(x)] - [ 2xln(x) - x] = x,
0 l 2
X
6 1 (x) = f(x)[x ln(x) - x2 ] = ln(x) - 1, 62 (x) = - f(x)[ x2 ]

2
= -1,
63 (x) = J(x)[x] = �, u 1 = J In(�-l dx = ½[ln(x)] 2 - ln(x),
u2 = - J � dx = -ln(x), U3 = J ;2 dx = -� ⇒
YP = {½[ln(x)] 2 - ln(x)}x - ln(x)[xln(x)] - �[x2 ] = - ½x[ln(x)] 2 - xln(x) - x,
or YP = -½x[ln(x)] 2 ⇒ y = -½x[ln(x)] 2 + c1x + c2 xln(x) + c3 x2 .
11. x3 y111 + 5x2 y" + 2xy' - 2y = �- For x3 y"' + 5x2 y" + 2xy' - 2y = 0, x > 0, y = x
r
⇒ P(r) = (r - l)(r + l)(r + 2) = 0 ⇒ Yi = x, Y2 = x- and y3 = x- . With

1 2
x x-1 x-2
J(x) = 2\, W(x) = 1 -x-2 -2x-3 = x[-2x-6] - [4x-5] = -6x- 5,
0 2x- 3 6x- 4
61(x) = f(x)[-x-4] = -x- 8, 62 (x) = -f(x)[-3x- 2 ] = 3x-6,

63 (x) = f(x)[-2x-1] = -2x- 5, u1 = ¼ J x-3 dx = -/2 x- 2 ,
u2 = -½ J x-1 dx = -½ ln(x), u3 = J ½ dx = ½x
1 1 1 1
=} Yv = -/2 x- - ½ ln(x)[x- ] + ½x- , or Yv = -½x- ln(x)
1 1
=} Y = -½x- ln(x) + C1X + C 2 x- + C3 x- .
2
12. x3 y
111
+2x2 " = 1. For x3
y y
111
+2x2 " = 0, x > 0,
y y = xr =;,- P(r) = r 2 (r-l) = 0 =;,-
1 ln(x) x
Y1 = 1, Y2 = ln(x) and 3 = x. With f(x) = }3
y , W(x) = 0 1. 1 - ;2 ,
xl
0 - x2 0
1
6 1 (x) = f(x)[ln(x)- 1] = ln(:1- , 6 2 (x) = -f(x)[l] = - ;3 ,
63 (x) = f(x)[¾] = ;4, u1 = J In(�-l dx = ½[ln(x)] 2 - ln(x),

u 2 = - J ¾ dx = - In(x), U3 = J x\ dx = -¼
=} Yv = ½[ln(x)] - ln(x)- [ln(x)] - 1, or Yv = -½[ln(x)]
2 2 2
=;,- y = -½[ ln(x)] + c1 + C 2 ln(x) + c3x is the genera l so luti on. Then
2
y' = -� ln(x) + c2 ¾ + c3 , y" = x\ ln(x) - ;2 - c2 ;2 , and y(l) = 5, y'(l) = 1 and
y" ( 1) = 0 =} C1 + C3 = 5, C 2 + C 3 = 1, -1 - C 2 = 0 =} C1 = 3, C 2 = -1 and C 3 = 2
=;,- y = -½[ln(x)] 2 + 3 - ln(x) + 2x.
13. x3 + x2 " - x ' = 4x2 . For x3 + x2 " - x ' = 0, x > 0, = x

r
111 111
=;,
!,
y y y y y y y
P(r) = r (r - 2) =} Y1 = 1, Y2 = ln(x) and 3 = x . With f(x) =

2 2
y
1 ln(x) x2
W(x) = 0 1. 2x = �, '6. 1 (x) = f(x)[2xln(x) - x] = 8ln (x) - 4,
xl
0 x2 2
62 (x) (x) =
= - f [ ] -8, 63(x) = f(x)[¼] = ;2,
2x
u1 = J 2x ln(x) - x dx = x2 ln(x) - x2 , u 2 = - J 2x dx = -x2 ,
U3 = J;; dx = ln(x) =} Yv = x ln (x)- x , or YP = x ln(x) =}
2 2 2
y = x2 ln(x) + c1 + c2 ln(x) + c3 x2 .
14. x3 "' + 2x2 y" - 2x ' = x3 cos(x). For x3 "' + 2x2 y" - 2x ' = 0, x > 0, = xr
y y y y y
=;,- P(r) = r(r + l)(r - 2) = 0 =;,- Y1 = 1, Y2 = x-t and 3 = x2 . With y
1 x- 1 x2
f(x) = cos(x), W(x) = 0 -x- 2 2x = -6x- 2 , '6.1(x) = f(x)[3] = 3cos(x),
0 2x- 3 2
i6. 2 (x) = - f(x)[2x] = -2xcos(x), i6. 3 (x) = J(x)[-x- ] = -x- cos(x),
2 2
u 1 = -½ J x cos(x) dx = -½x sin(x) + J xsin(x) dx

2 2
= -½x2 sin(x)- x cos(x) + sin(x),

u 2 = ½ J x3 cos(x) dx = ½x3 sin(x) - J x2 sin(x) dx
= ½x3 sin(x) + x2 cos(x) - 2 J xcos(x) dx
= ½x3 sin(x) + x2 cos(x) - 2xsin(x) - 2cos(x), u 3 = ½ J cos(x) dx = ½sin(x) =}
Yv = [-½x2 sin(x) - xcos(x) + sin(x)] +
[½x3 sin(x) + x2 cos(x)- 2xsin(x)- 2cos(x)] x- 1 + [½sin(x)] x2
= -sin(x)- 2x- 1 cos(x) =;,- y = -sin(x) - 2x-1 cos(x) + c1 + c2 x- 1 + c3 x2 .
C.4. CHAPTER 4 SOLUTIONS 38 5
15. x3y111 + 3x2 y " + 2xy'= 1. For x3 y"' + 3x2 y" + 2xy ' =0, x > 0, y = xr => P(r) =
r(r2 + 1) = 0 => y 1 = 1, Y2 =cos[ln(x)] and y3 =sin[ln(x)]. With f(x) = x\,
1 cos[ln(x)] sin[ln(x)]
W(x) = 0 sin[ln(x)] -¾ ¾
cos[ln(x)] - ...L'
- x
3
0 }2 sin[ln(x)] - }2 cos[ln(x)] - x\ cos[ln(x)] - x\ sin[ln(x)]
{¾
� 1 (x)= f(x) [¾] = ;4, � 2 (x)= -f(x) cos[ln(x)l} = -}4 cos[ln(x)],
� 3 (x) =f(x) {-� sin[ln(x)l} = - }4 sin[ln(x)], u1= ¾ dx= ln(x), J
J J
u2 = - cos[;( )j dx= -sin[ln(x)], U 3 = - sin[l;(x)] dx= cos[ln(x)]
x
=> YP =ln(x) => y =ln(x) + c1 + c2 cos[ln(x)] + c3 sin[ln(x)].
16. x3y111 + 3x2y" = x2 sin(x). For x3y111 + 3x2y"= 0, x > 0, y =xr =>
P(r) = r(r2 - 1) =0 => Y1 = 1, Y2 =x and y3 = x-1 . With J(x) =sin?),
1 X X-l
W(x) = 0 1 -x- 2 = 2x-3 , � 1 (x) =f(x)[-2x-1] =-2x- 2 sin(x),
0 0 2x-3
2 1
� 2 (x) =-J(x)[-x- ] = x-3 sin(x), � 3 (x)= f(x)= x- sin(x),
J
u1 = - J xsin(x) dx= xcos(x) -sin(x), u2 = ½ sin(x) dx= -½cos(x),
J
u3 = ½ x2 sin(x) dx =-½x2 cos(x) + xcos(x) dx J
=-½x2 cos(x) + xsin(x) + cos(x) =>
YP =[xcos(x) -sin(x)] - ½x cos(x) + [-½x2 cos(x) + x sin(x) + cos(x)] x-1
=x- 1 cos(x) => y =x- 1 cos(x) + c 1 + c2 x + c3 x-1.
17. y '" - 3y" - y' + 3y = 8e2 x . By Exercise 1, P(r) = (r - l)(r + l)(r - 3)
⇒ (D - l)(D + l)(D - 3)y = 8e2x ⇒ (D - 2)(D - l)(D + l)(D - 3)y =
(D - 2)[8e2x ] = 0 ⇒ y = C 1 e2x + C2 ex + C 3 e-x + C4 e3x ⇒ YP = ae2 x , y� = 2ae2x ,
Y"P = 4ae2x ' yP"' = 8ae2 x ' and yP111 - 3yP" - yP' + 3y = 8e2x => -3a = 8 => a= _§.3
⇒ YP = -ie2x _
18. y
111
- = lOex . By Exercise 4, P(r) = (r + l)(r2 - 2r + 2) =>
y" + 2y
(D+l)(D -2D+2)y = lOex => (D-l)(D+l)(D 2 -2D+2)y = (D-l)[l0ex ] = 0
2
=> y = c1ex + c2 e-x + c3ex cos(x) + C4ex sin(x) => YP = aex , � =yi =yi' = aex ,
y
and yi' - yi + 2y =lOex => 2a = 10 => a= 5 => YP = 5ex .

p
19. y
1 11
= 20x3 . By Exercise 6, P(r) = r2 (r - 1) => D2 (D - l)y = 20x3 =>
- y"
D D (D - l)y = D 4 [20x3 ] = 0 => D 6 (D - l)y = 0 ⇒
4 2
Y= c1 + c2 x + C 3 X 2 + C 4 X 3 + c5x4 + c6x5 + C7ex => YP = ax2 + bx 3 + cx4 + dx5,

y� = 2ax + 3bx + 4cx + 5dx , yi = 2a + 6bx + 12cx + 20dx ,
2 3 4 2 3
111 =
Yp
6b + 24cx + 60dx2 , and y - yp" = 20x3 ⇒
111
p
[6b+24cx+60dx2 ]-[2a+6bx+12cx2 +20dx 3 ]= 20x 3 ⇒ d= -l, 60d-12c =0
=> c = -5, 24c - 6b = 0 => b= -20, 6b - 2a = 0 => a= -60 =>
YP = -60x2 - 20x3 - 5x4 - x5 .
20. y"' - 2y" = 5sin(x). By Exercise 7, P(r) = r2 (r - 2) ⇒ D2 (D - 2)y = 5sin(x)

⇒ (D2 + l)D2 (D - 2)y = (D 2 + 1)[5sin(x)] =0 ⇒
y= c1 + c2 x + C3 e2x + C4 cos(x) + c5 sin(x) ⇒ YP = acos(x) + bsin(x),

y� = -asin(x)+bcos(x), yi = -acos(x)-bsin(x), yi' = asin(x)-bcos(x), and
yi' - 2yi = 5sin(x) ⇒ [asin(x) - bcos(x)] - 2[-acos(x) - bsin(x)] = 5sin(x)
⇒ 2a - b = 0 and a + 2b = 5 ⇒ a= l and b = 2 ⇒ YP = cos(x) + 2sin(x).
21. y "' -3y " + 2y' = 8xe 2x . By Exercise 2, P(r) = r(r - l)(r - 2) ⇒
D(D - l)(D- 2)y = 8xe2x ⇒ (D - 2)2 D(D- l)(D - 2)y = (D- 2)2 [8xe2x] = O
⇒ D(D - l)(D - 2)3 y = 0 ⇒ y = C1 + c2 ex + c3e2x + c4xe2x + c5 x2 e2x ⇒
YP = (ax+bx2)e2x , y� = (a+2bx)e2x+2(ax+bx2)e2x = [a+(2a+2b)x+2bx2]e2x,
Yi= [(2a + 2b) + 4bx]e2x + 2[a + (2a + 2b)x + 2bx2 ]e2x
= [(4a + 2b) + (4a + 8b)x + 4bx2 ]e2x,
yi'= [(4a + Sb) + 8bx]e + 2[(4a + 2b) + (4a + 8b)x + 4bx ]e
2x 2 2x
= [(12a + 12b) + (Sa + 24b)x + 8bx ]e , and y;' - 3y; + 2y� = 8xe2x ⇒
2 2x
[(12a + 12b) + (Sa + 24b)x + 8bx2 ]e2x - 3[(4a + 2b) + (4a + 8b)x + 4bx2 ]e2x+
2[a+(2a+2b)x+2bx2 ]e2x = 8xe2x ⇒ 2a+6b+4bx = 8x ⇒ b = 2 and 2a+6b = 0
⇒ a= -6 ⇒ YP = (2x2 - 6x)e2x .
22. y "' - y " + y ' - y 4sin(x) + 4cos(x). By Exercise 5, P(r) = (r - l)(r2 + 1)
=
⇒ (D - l)(D + l)y = 4sin(x) + 4cos(x) ⇒ (D2 + l)(D - l)(D2 + l)y =
2
(D2 + 1)[4sin(x) + 4cos(x)] = 0 ⇒ (D - l)(D2 + 1)2 y = 0 ⇒

y = c1ex + c2 cos(x) + c3 sin(x) + C4Xcos(x) + c5 x sin(x) ⇒
YP = axcos(x) + bxsin(x) = x[acos(x) + bsin(x)],
y� = [acos(x) + b sin(x)] + x[-asin(x) + bcos(x)],
y; = [-2asin(x) + 2bcos(x)] + x[-acos(x) - bsin(x)],
y;' = [-3acos(x) - 3bsin(x)] + x[asin(x) - bcos(x)], and
y;' - y; + y� - YP = 4sin(x) + 4cos(x) ⇒
(-2a - 2b)cos(x) + (2a - 2b)sin(x) = 4sin(x) + 4cos(x) ⇒ 2a - 2b = 4 and
-2a - 2b = 4 ⇒ a = 0 and b = - 2 ⇒ YP = - 2xsin(x).
23. y111 - 2y" + y' = -2ex cos(x). By Exercise 3, P(r) = r(r - 1)2 ⇒
D(D - l) 2 y= -2ex cos(x) ⇒
(D2 - 2D + 2)D(D - 1)2 y = (D2 - 2D + 2)[-2ex cos(x)] = 0 ⇒
y = c1 + c2 e + C3 Xe + C4e cos(x) + C5e sin(x) ⇒ YP = e [acos(x) + bsin(x)],
x x x x x
y� = e [(a + b)cos(x) + (b - a) sin(x)],

x
x x
y;= e [(a + b)cos(x) + (b - a) sin(x)] + e [-(a + b)sin(x) + (b - a)cos(x)]
= e [2bcos(x) - 2asin(x)], y i' = e [(2b - 2a)cos(x) - (2a + 2b)sin(x)], and
x x
yi'- 2y i+y� = -2e cos(x) ⇒ e [(-a - b)cos(x)+(a - b)sin(x)] = -2e cos(x)

x x x
⇒ a - b = 0 and -a - b = -2 ⇒ a= b = l ⇒ YP = e [cos(x) +sin(x)].

x
24. y( )
4
y" = 2. By Exercise 8, P(r) = r2 (r2 - 1) ⇒ D2 (D2 - l)y = 2 ⇒
-
D (D- l)(D+ l)y = D[2] = 0 ⇒ y = C1 + C2 X + C3X2 +c4ex +cse-x ⇒ YP = ax2 ,

3
Y'P = 2ax ' yP" = 2a ' yP"' = yP( ) = 0 ' and yP( ) - yP" = 2 ⇒ -2a = 2 ⇒ a= -l ⇒
4 4
YP = -x .
2
25. y(4) + y" = 6x. By Exercise 9, P(r) = r2 (r2 + 1) ⇒ D2 (D2 + l)y = 6x ⇒

D4 (D2 + l)y = D2 [6x] = 0 ⇒ y = c1 + c2 x + c3x2 + c4x3 + c5cos(x) + c5sin(x)
⇒ yP = ax2 + bx3' y'P = 2ax + 3bx2 ' y"P = 2a + 6bx ' y"'P = 6b,py ( ) = 0' and
4
Yt) + yi = 6x ⇒ 2a + 6bx = 6x ⇒ a= 0 and b = 1 => YP = x3.

1. y"' - y" - 2y' = 12.
(a) For y y" - 2y' = 0, y = e
111
- => P(r) = r(r + l)(r - 2) = 0 => Y1 = 1,
rx
1 e-x e 2x
y2 = e- x and y3 = e2x, and W(x) = 0 -e-x 2e 2x = -6ex . With
0 e-x 4e 2x
J(x) = 12, 6. 1 (x) = J(x)[3e ] = 36e , 6. 2 (x) = - f(x)[2e ] = -24e x,
x x 2x 2
b. 3 (x) = J(x)[-e-x] = -12e-x , U 1 = - f6dx = -6x,

U2 = 4 Jexdx = 4ex , U3 = 2 J e- 2xdx = -e-2x => YP = 3 -6x,or YP = -6x
=? Y = -6x + C1 + C2 e-x + C3e 2x.
(b) P(r) = r(r + l)(r - 2) ⇒ D(D + l)(D- 2)y = 12 ⇒ D2 (D + l)(D- 2)y =
D[l2] = 0 => y = C1 + C2 X + c3e-x + C4e2x => YP = ax,y� = a,Yi= yi' = 0,
and yi' - yi - 2y� = 12 ⇒ -2a = 12 ⇒ a = -6 ⇒ YP = -6x ⇒
Y = -6x + C1 + C3e-x + C4e 2x .
(c) y"' - y" - - -
2y' = 12 => y" y' 2y = 12x + c 1 . For y" - y' - 2y = 0,
y = erx => P(r) = r2 - r - 2 = (r - 2)(r + 1) = 0 => Y1 = e2x and
Y2 = e-x . YP = ax + b => y� = a, yi = 0 and yi y� 2yp - - = 12x + C1 =>
-a - 2[ax + b] = 12x + c1 ⇒ a = -6 and b = 3 - � = c2 ⇒ YP = -6x + c2
=? Y = -6x + C2 + C3e 2x + C4e-x .
2. x3 y111 + x2 y" = x. For x3 y"' + x 2y" = 0, x > 0, y = xr => P(r) = r(r - 1)2 = 0
1 x x ln(x)
=> Y1 = 1, Y2 = x and y3 = x ln(x), and W(x) = 0 1 ln(x) + 1 1
X
0 0 .!.
With J(x) = :2 , 6.1(.1:) = J(x)[x] = ¾, 6. 2 (x) = - f(x)[ln(x) + l] = - n(;l+ ,
I 1
6.3(x) = J(x)[l] = :2 , u1 = J1 dx = x,u2 = - J n(�+I dx = -½[ln(x)] 2 -ln(x),

I
U3 = J ¾ dx = ln(x) ⇒ YP = x - { ½[ln(x)] + ln(x)} x + ln(x)[x ln(x)], or

2
YP = ½x[ln(x)] 2 ⇒ y = ½x[ln(x)]2 + c1 + c2 x + c3xln(x).
3. x3 y"'+x2 y" =�-By Exercise 2, W(x) = ¾- f(x) = x� => 6.1(x) = f(x)[x] = x�'
6.2(x) = - J(x)[ln(x) + l] = _4 In(;� +l, 6.3 (x) = J(x)[l] = �,
x
u1 = J ..i..2 dx = - 1x , u2 = -4f In(x3) +l dx = 2x- 2 ln(x)+x-2 +2x- 2 = 2 ln(2x ) +2-
x x 1n x
2-]
x x2 ,
U3 = J ..xi.. dx = _-1_
3 x 2=> yP = _ _'!x + [2 x ) + x x - 2
2
(
2 x
2 [x ln(x)] = _ _lx =>
y = -¾ + c1 + c2 x + c3x ln(x).
4. Y111 - y " = 20x3 => y " - y ' = 5x4 + C1 => y ' - y = x5 + c1 x + c2 . I(x) = e-x =>
(e-xy )' = (x5 + C 1 X + c2 )e-x => e-xy = J(x5 + C 1 X + c2 )e-x dx
= -(x5 + C1X + c2)e-x +f(5x4 + c1 )e-xdx
= -(x5 + C1 X + c2 )e-x - (5x4 + c1 )e-x +f20x3 e-x dx
= -(x5 + C1 X + c2 )e-x - (5x4 + c1 )e-x - 20x3 e-x +J60x2 e-x dx
= -(x5 + C1 X + c2 )e-x - (5x4 + c1 )e-x - 20x3 e-x - 60x2 e-x +J l20xe-x dx =
-(x5 +c 1 x+c2 )e-x - (5x4 +c 1 )e-x- 20x3 e-x- 60x2 e-x - l20xe-x -12Qe-x +c3
=> y = -(x5 + c1x + c2 ) - (5x4 + c1 ) - 20x3 - 60x2 - 120x - 120 + c3 ex
= -x5 - 5x4 - 20x3 - 60x2 + C5 + c4 x + C3 ex .
5. x4 y ( 4) - 4x3 y"' + 12x2 y" - 24xy' + 24y = x .
4
For x y - 4x y + 12x y" - 24xy' + 24y = 0, x > 0, y = xr =>

4 (4) 3 111 2
P(r) = (r- l)(r- 2)(r- 3)(r- 4) = 0 =>Yi = x, Y2 = x2 , y3 = x3 and y4 = x4 ,

x x2 x 3 x4
x x2 x4 x x2 x3
1 2x 3x2 4x3
and W(x) = = -6 1 2x 4x + 24x 1 2x 3x2
3
0 2 6x 12x2
0 2 12x2 0 2 6x
0 0 6 24x
= -6[x(16x3 ) - 10x4 ] + 24x[x(6x ) - 4x3 ] = 12x4 . With J(x) = 1,
x2 x3 x4
6 1 (x) = - 2x 3x2 4x3 = -[x2 (12x4 ) - x3 (l6x3 ) + x4 (6x2 )] = -2x6 ,
2 6x 12x2
x x3 x4
6 2 (x) = 1 3x2 4x3 = x(l2x4 ) - 6x5 = 6x5, 6 3 (x) = -6x , 6 4 (x) = 2x ,
4 3
0 6x 12x2
u1 = -lfx 2
dx = _l.18..x3 ' u2 = l.Jxdx = lx2
' u3 = -lfldx = -lx
i
6 2 4 2 2 '
U4 = ½ J � dx = ln(x) => YP = -;�x + ¼x ln(x), or YP = ¼x4 ln(x) =>
4 4
y = ½x ln(x) + C 1 X + C2X + C3X + C4X .

4 2 3 4
6. x3y "' - x2 y" + lOxy' = 156x2 sin[3ln(x)]. For x3 y111 - x2 y " + lOxy' = 0, x > 0,
y = x => P(r) = r(r - 4r + 13) = 0 => r 1 = 0 and r = 2 ± 3i => Y 1 = 1,
r 2
Y2 = x2 cos[3ln(x)] and y3 = x2 sin[3ln(x)]. With J(x) = 155 [ � (xl], W(x) =

sin 3 n
1 x2 cos[3ln(x)] x2 sin[3ln(x)]
0 2xcos[3ln(x)] - 3xsin[3ln(x)] 2xsin[3ln(x)] + 3xcos[3ln(x)] = 39x,
0 -7cos[3ln(x)] - 9sin 3ln(x)] 9cos[3ln(x)] - 7sin[3ln(x)]
61 (x) = f (x)[3x 3
] = 468x2
sin[3ln(x)],
6 2 (x) = - J(x){2xsin[3ln(x)] + 3xcos[3ln(x)]}
= -312sin2 [3ln(x)] - 468sin[3ln(x)] cos[3ln(x)]
= -156 + 156cos[6ln(x)] - 234sin[6ln(x)],
6 3 (x) = J(x){2xcos[3ln(x)] - 3xsin[3ln(x)]}
= 312sin[3ln(x)]cos[3ln(x)] - 468sin2 [3ln(x)]
= 156sin[6ln(x)] - 234 + 234cos[6ln(x)],
u1 = 12Jxsin[3ln(x)] dx. x = et => u1 = 12 Je2t sin(3t)dt, and integration by
partsg ivesu1 = i�e2t sin(3l)-i�e2t cos(3t) = i�x2 sin[3ln(x)]-��x2 cos[3ln(x)],
u2 = J -� + 4cos[6�n (x)] - 5 sin [6�n(x)] dx = -4ln(x) + �sin[6ln(x)] + cos[6ln(x)],

u3 = J 4sin[6�n(x)] - � + 5 cos[6�n(x)] dx = -� cos[6ln(x)] - 6ln(x) + sin[6ln(x)] =>
YP = i!x2 sin[3ln(x)] - f�x2 cos[3ln(x)]+
x2 cos[3ln(x)] {-4ln(x) + �sin[6ln(x)] + cos[6ln(x)l} +
x2 sin[3ln(x)] {-�cos[6ln(x)] -6ln(x) + sin[6ln(x)l}. Ignoring the first two
terms because they are homogeneous solutions,
YP = �x2 { sin[6ln(x)] cos[3ln(x)] - cos[6ln(x)] sin[3ln(x)]}
+x2 { cos[6ln(x)] cos[3ln(x)] + sin[6ln(x)] sin[3ln(x)]}
-4x2 ln(x) cos[3ln(x)] - 6x2 ln(x) sin[3ln(x)]
= �x2 sin[3ln(x)] + x2 cos[3ln(x)] - 4x2 ln(x) cos[3ln(x)] -6x2 ln(x) sin[3ln(x)],
oryp = -4x2 ln(x) cos[3ln(x)]-6x2 ln(x) sin[3ln(x)] =>y = c 1 +c2 x2 cos[3ln(x)]+
c3 x2 sin[3ln(x)] - 4x2 ln(x) cos[3 ln(x)] - 6x2 ln(x) sin[3ln(x)].
7. Let P(r) = a0 +a1r+a2 r 2 +· · -+an rn , with ai, 0 :s; i :s; n, real. Then P(z) = 0
=> ao + a1z + a2 z 2 + · · · + an zn = 0 => ao + a1z + a2 z 2 + · · · + an zn = 0 =>
a0 + a1z + a2z2 + · · · + an:zn = 0 since ai = ai, 0 :s; i :s; n . Thus, P(z) = 0.
1 X x3
8. No, because W(x) = 0 1 3x2 = 6x = 0 at x = 0 and W(x) =I- 0 for x =/- 0,
0 0 6x
in contradiction of Theorem 4.8.
9. y"' + 3y" + 3y' + y = 2xe- x cos(x). For y"' + 3y" + 3y' + y = 0, y = erx =>
P(r) = (r + 1) 3 = 0 =>Yi= e-x , Y2 = xe-x and y3 = x2 e-x , and
e-x xe-x x2 e-x
W(x) = -e- x
e- - xe-
x x
2xe-x - x2 e-x
e-x -2e-x + xe- x 2e-x - 4xe-x + x2 e-x
1 X x2
= e-3x -1 1 - x 2x - x2
1 x - 2 2 - 4x + x2
= e-3x[(l - x)(2 - 4x + x2 ) - (x - 2)(2x - x2 ) + x(2 - 4x + x2 ) - x 2 (x - 2) +
x(2x - x2 ) - x2(1- x)] = 2e-3x_ With f(x) = 2xe-x cos(x),
� 1 (x) = f(x)[x(2x - x2 ) - x2(1- x)Je-2x = x2 e-2x f(x) = 2x3 e-3x cos(x),
� 2(x) = - f(x)[(2x - x2 ) + x2 Je-2x = -2xe- 2x f(x) = -4x2 e-3x cos(x),
� 3(x) = J(x)[(l - x) + x]e-2x = e- 2x J(x) = 2xe-3x cos(x),
u1 = J x3 cos(x) dx = x3 sin(x) -3 J x2 sin(x) dx
= x3 sin(x) + 3x2 cos(x) -6 J xcos(x) dx
= x3 sin(x) + 3x2 cos(x) - 6xsin(x) -6cos(x),
u2 = -2 J x2 cos(x) dx = -2x2 sin(x) + 4 J x sin(x) dx
= -2x2 sin(x) - 4xcos(x) + 4sin(x),
u3 = J xcos(x) dx = xsin(x) +cos(x) =>
YP = [x3 sin(x) + 3x2 cos(x) - 6x sin(x) -6cos(x)Je-x+
[-2x2 sin(x) - 4xcos(x) + 4sin(x)] xe-x + [xsin(x) +cos(x)]x2 e-x
= -2xe-xsin(x)- 6e-xcos(x) =>

y = -2xe-x sin(x) - 6e-x cos(x) + c 1 e-x+ c2 xe-x+ c3x e-x.
2
10. By Exercise 9, P(r) = (r + 1) 3 . xe-x cos(x) and xe-x sin(x) are solutions of a
fourth-order equation for which the indicial equation has the roots r = -1 ± i
of multiplicity 2 => (r + 1) 2 = -1 => r2 + 2r + 2 = 0 and (r2 + 2r + 2) 2 = 0
is the indicial equation => (D 2 + 2D + 2)2 is the annihilator of xe-x cos(x) of
minimal degree. Hence, (D + l)3y = xe-x cos(x) =>
(D2+2D+2) 2 (D+l) 3 y = (D 2+2D+2)2 [2xe-x cos(x)] = 0 =>y = c 1 e-x cos(x)+
c2 e-x sin(x) + C3Xe-x cos(x) + C4Xe-x sin(x) + cse-x+ c5xe-x+ C7x2 e-x =>
YP =(a+ bx)e-x cos(x) + (c+ dx)e-x sin(x) ,
v; = be-x cos(x) - (a+ bx)e-xcos(x)- (a+ bx)e-xsin(x) + de-x sin(x)
(c+ dx)e-x sin(x)+ (c+ dx)e-x cos(x)
= [(b- a+ c) + (d- b)xJe-xcos(x) + [(d- a - c) - (b+ d)xJe-x sin(x),
v;= (d - b)e-xcos(x)- [(b- a+ c) + (d- b)x]e-xcos(x)-
[(b- a+ c)+ (d - b)x]e-xsin(x) - (b+ d)e-x sin(x)-
[(d- a- c) - (b+ d)xJe-x sin(x)+ [(d - a - c) - (b+ d)xJe-x cos(x)
= [(2d - 2b - 2c) - 2dxJe-x cos(x) + [(2a- 2b- 2d)+ 2bxJe-x sin(x) .
v;' = -2de-x cos(x) - [(2d - 2b- 2c) - 2dx]e-x cos(x)-
[(2d- 2b - 2c) - 2dx]e-x sin(x) + 2be-x sin(x)-
[(2a - 2b - 2d)+ 2bx]e-x sin(x) + [(2a - 2b - 2d)+ 2bxJe-xcos(x)
= [(2a+ 2c- 6d)+ (2b+ 2d)xJe-x cos(x) + [(6b+ 2c- 2a)+ (2d-2b)xJe-x sin(x) ,
and v;'+ 3y;+ 3y; + YP = 2xe-x cos(x) =>
[(2a+ 2c- 6d)+ (2b+ 2d)x]e-xcos(x)+ [(6b+ 2c- 2a)+ (2d-2b)x]e-xsin(x) +
3[(2d - 2b - 2c) - 2dxJe-x cos(x) + 3[(2a - 2b- 2d)+ 2bx]e-xsin(x)+
3[(b - a+ c) + (d - b)x]e-x cos(x) + 3[(d- a - c) - (b+ d)x]e-x sin(x)+
(a+ bx)e-x cos(x) + (c+ dx)e-x sin(x) = 2xe-x cos(x) =>
[-(3b+ c) - dxJe-x cos(x) + [(a- 3d)+ bxJe-xsin(x) = 2xe-x cos(x) => b = 0,
d = -2, a - 3d = 0 =>a = -6, 3b + c = 0 =>c = 0 =>
YP = -6e-x cos(x) - 2xe-x sin(x) =>
y = -6e-x cos(x) - 2xe-x sin(x) + cse-x+ C5Xe-x+ C7X e-x.
2

x = -5x+ 6y -5 6 I ,\+ 5 -6 I = (,\ _ l) (,\+ 2) = 0
1. { y ;
= -3x + 4y } => A= ( -3 4 ) ' 3 ,\ - 4
=>,\1 = 1, A2 = -2.
For ,\ 1 = 1, ( � =� ) ( � ) = ( � )=>a= b => v1 = ( � ).
For ,\2 = -2, ( � =� ) ( � ) = ( � ) =>a= 2b => v2 = ( � ).
x1(t) = el ( � ), x2(t) = e-2l ( � ), x(l) = C1X1(t) + C2X2(t) ⇒

x(t) = ciel + 2c2e-2t, y(t) = c 1 e l + c2e-2t.
A; 3 1
2. { ;: : ��; y } ⇒ A= ( -� � ), I � I = (A - l)(A - 2) = O ⇒
A 1 = 1, A2 = 2.
- -
For A1 = 1, ( � � ) ( �)= ( � ) ⇒ b = -2a ⇒ vi = ( !2).
- -
For A2 = 2, ( � � ) ( �) = ( � ) ⇒ b =
-a ⇒ v2 = ( !l ).
x1(l) et ( , x2(l) = e2l ( ! ), x(t)= C1X1(t) + C2X2(l)
!2)
=
1
⇒
x(t) = c1e + c2e , y(i) = -2cie - c2e2L.
l 2l t
6.
{ =X y -
, x = 2, y(O) = 1 ⇒ A = (
y' = 2x + 4y } (O)
X
1
2 4
, x(O) =
1 -1 ) ( 2 ).
1
A 1 = 2, A2 = 3.
I _::-2 A� 41 = (A - 2)(A - 3) = 0 ⇒ A1
9. { �; : � 2x + 3y ' } ⇒ A = ( -� � � ) ,
z' = -x + y + 2z -1 1 2
u
,\-1 0 0
= (,\ - 1)(,\- 2)(,\ - 3) = 0 ⇒
=�
2 ,\ - 3 0 = = 2, = 3.
0
>-1 1, >-2 ,\3
0 n
1 -1 ,\-2
⇒ a = b and a - b - c = 0 ⇒
0)
ForA1 = 1, ) ) = )
J
v, = ⇒ x1(t) = e' (
-1 0 2 ).+1 0 -2
11. A= ( -2 1 2 ) , 2 ). - 1 -2 = (). - 1) 2 (). + 1) = 0 ⇒ ). 1 = 1,
0 0 1 0 0 .A. -1
0 -1 l
= >.2 +6>.+ 13 = 0 ⇒ >., >.= -3 ± 2i.
A
13. A = ( ) I I
- .
13 _6 , _ 13 >.+6
-
For >. = -3+ 2 , ( � t/
3
i i 1 2i
) ( � ) = ( � ) ⇒ b = (3 - )a 2i =}
v= ( �
3 2
) i ⇒ z(t) = e-3t [cos(2t)+i sin(2t)]
[( � )+ ( � )] ⇒
2
i
x 1 ( t ) = e _3 t ( cos(2t)
· =e _3
x (t)
3cos(2t)+2sin(2t) ) ' 2
t sin(2t) (
3sin(2l)- 2cos(2t) ) '
x(t) = C1X1(l) + C2X2(t).
14. A = (
2 -2
), I
A -2 2
1 = >. - 2>.+2= 0
2
⇒ >., ->.= 1 ± i..
l O -l >.
i
For>.= 1+ , ( i -
�t
l !i ) ( � )= ( � ) ⇒ a= (l + i)b ⇒
v = ( 1 �i) ⇒ z(t) = et [cos(t) + isin(t)] i
[ ( � ) + ( �) ⇒
]
x1 (t) =e
t ( cos(t)-sin(t)
()
cos t
()
) , x2 t =e
t ( sin(l)+cos(l)
.
sm (·l) ),
cos(l) - sin(l) sin(l) + cos(l)

X(t) = et ( _ ).
cos(t) sm(l)
( 0 0 1 ) ,\ 0 -1
17. A = 0 1 0 , 0 ,\-1 0
-1 2 0 1 -2 ,\
0 0 0 1 ,\ 0 0 -1
0 1 -1 0 0 ,\ - 1 1 0
lB. A = (
0 l 1 0 ) ' _ 0 -1 ,\ - 1 0
±(\ ,\2 ,12

(a) O)
-1 0 0 0 1 0 0 ,\
⇒ A , °Xi =
i = l
t i. -
l
i 1
For ,\ 1 = i , � _=-
1 i �l
�
)
: = (� ⇒ b= c = 0, d=ia ⇒
(
r( n
1 0 0 i d 0
v, = �) ⇒ z,(t) = [cos(t)+ i sin{t)] +i �

( )]
⇒
cos(t) sin(t)
(a)
X1(t)= ( � ) ,x2(i)= ( �
).
- sin(t) cos(t)
l+ i -1 0
O O
r ( D u) J
ForA2 = l+i,
(
� � �
l
�
)
: = ( �) ⇒ a=d=0,b=ic
n
1 0 0 l+ i d 0
⇒ v, = ( ⇒ z,(t) = e•[cos(t)+ i sin(t)] +i ⇒

0 0
sin(l) cos(l)
X3(l) = e ( co (l) ) ' X4(t) = e ( sin(t) ) '
l - l
�
Q
cos(t) sin(t) 0 0
0 0 -e l sin(t) e l cos(t)
X(t) = ( )
0 0 e l cos(t) e t sin(t) ·
- sin(t) cos(t) 0 0
19. { t:;
m1 = 2.
+ 3Y }
=> A= ( � { ), I A� l
>,
-=_\ I = (>- - 1) 2
= 0 => A1 = 1,
For A1 = 1, ( � -�) ( �) = ( �) => b = 0 => v1 = ( � ) =>
x1(t) = e t ( � ) , x2(l) = e l (tuo + u1), uo = vi, (A - >-1I)u1 = uo =>
( � � ) ( �) = ( � ) => 3b = 1 => ll1 = ( 1�3 ) => Xz(t) = e ( 1;3 ) ,

t
x(l) = c1xt(l) + c2x2(l) => x(l) = e t (c1 + Czl), y(l) = ½cze l .

z' = 3y - z
x
22. { ;: : ! +-2i - z } ⇒A= ( i -� -� ) ,
O 3 -1
=i j D ( n 0)
>.- 3 1 0
-1 >. - 2 1 = (>.- 1)2 (>.- 2) = 0 ⇒ >.1 = 2, = 1, = 2.
0 -3 >. + 1
D
>-2 m2
FoTA, = 2, ( = => 0 =b= C => v, = (
n
=> x 1 (t) = e" ( D
FoTA, = 1, (
=! =i O ) 0 )
= e'
= => b = 2a , c = 3 a
= e'(t11o + u1), u0 = v2,

=>
v, = => x,(t) ( � ) , x,(t)
U -i =D O ) U )
(�)
(A - A,I)u, = uo => = => u

,
=
0) =>
x3(t) = e' ( ;l � \ ), x(t) = c,x1(t) + c,x2(t) + c3x3(t) =>

x(t) = c1e2t + et [c 2 + c3(t + 1)], y(t) = c1e2t + et [2c 2 + c3(2t + 1)],
z(t) = c1e2t + et (3c 2 + 3c3t).
0 � n O) 0)
= => c = 0, 2b = 1 => u 1 = ( 1:2) =>
(
=>
x3 (t) = e" 1�2) , x(t) = c1 x 1 (t) + c2x2 (t) + c3 x3 (t).

>-+2 -1 0
1 >- 0
0 0 >-+1
u) u) n
-2 0 1) >- + 2 0 -1
27. A = ( 1 -1 0 , -1 >- + 1 0
-1 0 0 1 0 >-
m1 = 3.
FofA, = -1, ( -: � : D = = c = 0 => v, = (
!),
=> a
n
=> x 1 (t) = e-' ( x,(t) = c'(tuo +u,), u0 = v,, (A - A,I)u, = Uo =>
(] � D O) = ( =>
a
= =
C I
=>
u,
= (
D =>
x2 (t) = e_, ( : ) , x3 (t) = e-• (lt2 u0 + tu 1 + u2 ), (A - A 1 J)u2 = u 1 =>

( 1 1 0 1) A - 1 -1 0 -1
-1 1 -1 0 1 A-1 1 0
29· A=
0 0 1 1 ' 0 0 A - 1 -1
0 0 -1 1 0 0 1 .A.-1
_ O 4 _ _ 4cos(2t) 4
( -1 0 ) , (l)- (-2sin(2t)) ' I ,\1 -,\ , _ ->-2 +4- 0⇒,\,>--±2i,
2. A- f
_ - _ ·
v = � l , z(t) = [cos(2t) + i sin(2t)] [ �l + i �

i
( ) ( ) ( )] ,
X(t) = -2sin(2t) 2cos(2t) X(t)_ 1 = .! -sin(2t) -2cos(2t)
( -cos(2t) -sin(2t)' ) 2 ( cos(2t) -2sin(2t) '
)
X(t)-lf(t) = .! -sin(2t) -2cos (2t) 4cos(2t) = O
2 ( cos(2t) -2sin(2t) ) (-2sin(2t)
) ( 2'
)
J X(l)- 1 f(t) dl = ( it ) ,
x(l) = ( -2 sin(2t) 2c�s (2t) ) [( c 1 ) +( 0
-cos(2l) - sm(2t) c2 2i )]
= ( -2c 1 sin(2t) + 2c2 cos(2l) + 4tcos(2t) )
-c 1 cos (2t) - c2 sin(2l) - 2l s in(2i)
_ (t + 3)et
- ( 2(t + l)e 2t
) .
6. A = ( �
-
�), f(t) = ( i ), Xo = ( �), I :
1
; I = A'+ 1 = 0 c>
A,X = ±i, v = (; ), z(t) = [cos(t) +isin(t)] [ ( �) +i ( �)].
X(t) = - sin(t) c?s(t) X -l - sin(t) c?s(t) 1

( cos(t) sm(t) ) ' (t) f(l) = ( cos(t) sm(t) ) ( 1 ) =
cos(t) - s� n(t) f s� n(s) + cos(s) t
( cos(t) + sm(t) ) '}0 X(s)-if(s) ds = ( sm(s) - cos(s) ) =
sin(t) + cos(t) - 1 _1 _ 0 1 2 _ 1
0 I
( sin(t) - cos(t) + l ) ' X (O) x0 - ( 1 O ) ( 1 ) - ( 2 ) '
- sin(t) cos(t) 1 sin(t) + cos(l) - 1
x(t) =
( cos(t) sin(t) ) [( 2 ) + ( sin(l) cos(t) + 1 )]
-
- sin(t) co s (t) sin(t) + co s (t) 3cos(t) - 1
( cos(t) sin(t) ) ( sin(t) - cos(t) + 3 ) - ( 3sin(t) + 1 ) ·
I I -(
-
_ -2 _
7. A - ( 4
-1
6)
, f(·t) _
- e -4 t ( 12 ) , x0
_
-
o ),
( 1
). + 2 1
_4 ). + 6 - >. + 4)2
= 0 c> A1 = -4, v1 = ( ; ), (A - A1/)u1 = Uo = v, c> u, = ( �l ),

1 ( t
X(t) = e-4 t 2 2t - ), X(t)-lf(t) = e4t
1- 2t l e-4t
1( = (
1: i ) ,
) 2)
( � ), l
1 2 -1
X(s J-' f(s) ds = ( ; ) = ( X (0)- 1 xo = ( ; _ � ) ( � )
= ( �l ) ' x(t) = e-« ( ;

21 � 1 ) [ ( �l )
+
( t )] = e- ( )
4' � ·
(
C.5. CHAPTER 5 SOLUTIONS 4 05
t t
= 0 ),
O 2e- l - 2
l. y" - 2y' + y = 0.
(a) y = e rt ⇒ 2r - 2 r + 1 = (r - 1) 2 = 0⇒ r = 1 ⇒ Y1()l = e t, 2Y (t) = te l,

y ( l) = 1
c te + 2c et t .
(b) X1 = y and x 2 = y' ⇒ x� = y' = x 2 and x; = y" = 2y' - y = 2x 2 - x1.
c ) A
( = ( _ � � ), I ; ,\ -=_\ I = ( ,\ - 1 )2 = 0 =? ,\ 1 = 1, v1 = (�)
⇒ x1(t) = e t ( � ) , x 2(t) = e l(tuo + u1 ), uo = vi, (A - A1I ) u1 = uo
⇒ U1 = ( �) ⇒ x 2(t) = e t ( l1 l) ⇒ x(t) = C1X1(t) + C 2X 2(t) ⇒

x1(t) = (c1 + 2c t) e t, x 2(t) = [ 1c + + l)]e t, and x1(i) = y ( t).
2c (t
I -- e
I
t
Y ( t) Y 2()l l
el
1-1
(d) W [ 1Y (t) Y2(]l) = e 2t
and
Y�1 ( ·t) y�(t) - e (t+ l) e l
t
W[x1(t) x 2(t)] =I x1(t) x 2(t) I = I :: (t: : ) e t

e
I= e 2t = W[y1(t) Y2(t)].
2. Let X(t) be a fundamental matrix for the system x' = Ax.

(a) Since X(o)- 1 is nonsingular, Y()l = X(t)X(o)- 1 is a fundamental matrix
by Theorem 5.4, and Y(0) = X(0)X(o)- 1 = I. The solution of x' = Ax,
x(0) = xo, is x(t) = Y(t)x o.
(b) Since X(to )- 1 is nonsingular, Z(t) = X(t)X(to)- 1 is a fundamental matrix
by Theorem 5.4, and Z(t0) = X(ol )X(io )- 1 = I. The solution of x' = Ax,
x(to) = xo, is x(t) = Z(l)x o. Part (a) is the special case where ot = 0.
(a) A= ( �
-�
), I
,\
�/ 12 1 = (,\ - 2)(,\ + 2) = 0 ⇒ ,\ 1 = 2,
(�
,\
= -2, v 1 = ( � ), v 2 = ( � ),x 1 (t) = e2t

,\2 ),x 2 (t) = e- 2t ( � ),
x(t) = c 1x 1 (t) + c2x2 (t) ⇒ x(t) = 4c 1 e2t, y(t) = c 1 e2t + c2e- 2t.
(b) x' = 2x ⇒ x(l) = c1 e2t ⇒ y' = c1 e2t -2y ⇒ y'+2y = c 1 e2t ⇒ (e2t y)' = c 1 e4t
⇒ e2ty = ¼c 1 e4t + c2 ⇒ y(t) = ¼c1e2t + c2 e- 2t and, replacing the arbitrary
constant c 1 by 4c 1 , the solution coincides with the one in part (a).
= x - 2 y ⇒ y" = x' - 2y' = 2x - 2(x - 2 y) = 4y ⇒ y" - 4y = 0.
( c) y'
(d) y = e ⇒ r2 - 4 = (r - 2)(r + 2) = 0 ⇒ y(t) = c 1 e2t + c2e-2t ⇒
rt
x(t) = 2y(t) + y'(t) = 2[c1e2t + c2e- 2t] + [ 2c1e2t - 2c2e- 2t] = 4c 1 e2t, as in
part (a).
-7 12
4. x' = Ax, A = (
_4 )
7 .
lt2 lt3
2 6
t lt2
1 )
t
0 1
=
7 If c1x1(t)+c2x2(t) 0, then t = 0 ⇒ c1a+c2b = 0, and l = 213
.
⇒ c2a-c1b =0
⇒ (er+ c�)b =0 ⇒ C1 = C2 = 0 since b-/- 0.

1. {5, 7, 9, 11, · · ·} = {2n + 3}�=l
2. {-6, -3, 0, 3, · · ·} = {3n}�=- 2
3. { 1, -21 , 3'
1 _ _! . } = { (- 1:-1}OO
4, .. n=l
1 .!.9' 116'
4. { 1, 4, }
... = { n\} :;=l
1 l 1
5. {1, 3' 9' 27' ... } = { 3�} :;=0
6. {o' 1 - 12' 1 - .3'
00
! 1 - 14' ...} = {1 - 1n} =l
n
oo
2
9. {../2nn++1 l} ={1, /22 v'55' v'Io
10 \/'17
17 ···}
n=O 3 5 7 , 9 '
_ 1 =? bn+l -
11 . bn = an+l + 2nn fOr a11 n > n+ 1
_ 0 and bn-1 - + n- l
- an+2 + 2n+1 f01· n > - an 2'11
for n 2:: 2.
12 · { 3l ' 5'
1 l 1. _ 1 oo _ l oo _ l oo
7' g, · · · } - { 2n+l } n=l - { 2n+3 } n=O - { 2n-l } n=2
13. an = n1,, ---t 0 =} { an} converges.
14. an = n-\
_ = -n3 ---t -oo =} { an} diverges.
15. an = }n ---t 0 =} { an} converges.
16. an = 2-n = � ---t 0 =} { an} converges.

e90 r ---to=⇒ {a } converges.
2
11. an = n
< � = (-�r ---t O ⇒ {a } converges.

n
18. an = -; n
19. an = ;
( ! )n = (-!r, and n-+oo
lim a n does not exist⇒ {an} diverges.
20. an = }n ---t O =} { an} converges.

21. an = 10n0
1, 000 , 000 , 000, ooo
000000000 1 ---t O =} {an} COnVerges.
22. an = 3 - ;2 + }n ---t 3 =} { an} converges.

23. an = n + 3n1 ---t 00 =} { an} diverges.
3n2 -2n-3 _ 3-¾-�
24 · an = 2n2+3n+l , - 2+1+ J
n �
---t 23 =? {an} Converges.
3 -2n-3
25. an = 3n
4n2 +n+l
26. an = 3n -2n-3
2
2n3 +2n-l
3n2 -2n-3
27. an = J2 n4+3n+l
---t 1 =} { an} converges.
3 - 2 - 3
3n2 -2n-3
28. an = J2n5-2n+l ;r72 � �
✓ 2-3'-+;:!s-
---t O =} { an} converges.
n n
3n2 -2n
29. an= J2 n3-n+l
31. an= ✓n2 + n - ✓n2 -n = ( ✓n2 + n - ✓n2 - n) J

Jnz+n+JnT=n =
n2+ n+Jn2 -n
2 n
Jn + n+Jn2 -n 2
-+ 1 => { an } converges.
Ji+¾! ✓i="¼
32. an = n- ✓n
2 +n = (n - ✓n
2 + n ) nn++J
v'n2+n
n2 + n
-n
n+Jn2+n
-1
-t
i+Ji+¾
{an } converges.
n3/2+_1_2 _ nl/2
A
Jn3"+l+n 3
33. an = ✓n
3 + 1 - n = ( ✓n3 + 1 - n) M,
n +l+ n
= �
n2
n +l+n
- i+.J,,.+ _1_
n� nl/2
-2 n n . n
34. an -
- (3nn)2 -- ( - 32 ) 1n2 -t 0 by Theorem 6.1 since ( - 32 ) -t O and 1n2 -t 0.
Hence, { an } converges.
-l n
35. an - -- 0 by Theorem 6.2 smce I ( )n 11 -- _
l � 0. Hence, {an }
_ (-lr I
n-
-t
n
converges.
l - -l n .!. - (-l) n - n
36. an = (n ) n n
-t O by Theorem 6.1 since .!.
n
-t O and ( nl) -t O by
Theorem 6.2. Hence, { an } converges.
n)] 2 . = [ln(xX )] 2 . By L'Hopital's rule, lim [ln(xX )]2 = lim [2 ln(x)J¾
37. an = [ln(
n
Let f(x)
X--->00 X--->00 1
2 In(x) [In(nll 2 = 0 by Theorem 6.3.
= lim
X-+00
X
= lim �
X-+00 l
= lim lX
X-+00
= 0=> lim
n-+oo
n
Hence,
{an } converges.
38. Let f (x) X

x . By L'Hopital's rule, lim � lim _!_ X
0=>
e x-+oo e X---+00 e
lim �
n-+oo e
=0 by Theorem 6.3. Hence, { an} converges.
= = = lim 3� = lim 6;
2
39. an n
!. Let f(x) x:. By L'Hopital's rule, lim x:
e e x-+oo e x-+oo e x-+oo e
lim � = 0=> lim n! = 0 by Theorem 6.3. Hence, {an } converges.
x-+oo e n-+oo e
·f l p n
40. p > 0 , q > 0 . an 0 1•f' and on1 y 1 {an/ } - lnn(
-t q/p) -t 0 . Let
· � · �
ln(xIP) - 11m ·
B y L'Hop1ta
A
· l's ru 1 e, 11m
X---+00 X
l/x
X---+00 !l.x P
= X--->00
11m --
P
qxq/p
= 0=>
lim [In(
n
�)]P = 0 by Theorem 6.3. Hence, { an} converges.
n-+oo
= x 1 I = e-x-.By L'Hopital's rule, lim � = lim 1- = 0

ln(x)
41. an = n 1 In_ Let f(x) x I ( )
X
X-+00 X-+00 X
=> lim e¥
x-+oo
lim n 1/n = 1 by Theorem 6.3. Hence, {an } converges.
= e0 = 1 => n---+oo
42. an = [ln(n)] 1/ 1 u(n)_ Let f(x) = [ln(x)]1 /ln(x) = elnfxi ln [ln(x)]_ By L'Hopital's
. . I I
. . Tii(xj I l _._
ru 1e, 11m J.n[Inln(x
· )] _
- 11m ---r- x _- 11m r -
1
- 0 _._
_
.....,- 11m e l n [ n(x)l = l .....,-
ln(x)
X-+00 ()
X
X---+00 x x-+oo
n()
X
X---+00
lim [ln(n)]1 /ln(n) = 1 by Theorem 6.3. Hence, {an } converges.

n---+oo
43. an = [ln(n)]1/n_ Let f(x) = [ln(x)]1/x = e ½I n (In(x)]_ By L'Hopital's rule,

lim I n[lnX (x)] = lim -l 1(-) = 0 =} lim [ln(x)]l/x = 1 =} lim [ln(n )Pin 1 by
X->(X) X ->(X) x n x X->(X) n->oo
Theorem 6.3. Hence, { an } converges.
n) n) l n (n) = e for all n � 2 ⇒ hm an = lim e =e.
4 4. an= n1/1 n (n) =e ] n (n1/ ln( ) = e -
1
ln(-
n---+oo n---+oo
Hence, {an } converges.
45. an = cosn(n). -1 - < 1 =} _l

< cos(n) - < cosn(n) -
n -
<ln
, and lim (-.!.)
n
= lim n.!. =0
n-+oo n---+oo
⇒ lim cos(n) =0 by Theorem 6. 4. Hence, {a } converges. Alternatively,
n n
n-+oo
< I cosn(n) I -
0 - < n.!. and lim 0 = lim l = 0 =} lim I cos(n) I = 0 by Theorem 6.4
n---+oo n-+oo n n---+oo n
and, hence, lim co s (n)

=0 by Theorem 6.2.
n---+oo n
46. an = sn
i �2 ).
vn
0 :S I sinv�n ) I :S v�
2
n
and lim O = lim v�
n
=0
n---+oo
=} lim I si n �
v n
2
) I =0
n---+oo n---+oo
2
by Theorem 6.4 and, hence, lim sin�n ) = 0 by Theorem 6.2. Hence, {an}
n---+oo v
converges.
_
47. an - _ ne e·e·..e ·e·e � . £ . £ ... _e_ . � < £:. . � = .£: for n -
> 4 =}
- 1-2-3-.
n! . (n-l)·n 1-2 3 4 n-1 n - 2 n 2n
3e 3e
0 :S an :S 2n and lim O = lim 2n = 0 ⇒ lim a n = 0 by Theorem 6.4. Hence,
n---+oo n---+oo n---+oo
{ an } converges.
_ (2n)! _ 1 . 1·2·3--·(n-l)·n . (n+l)·(n+2)---2n < l . 1 . 2n _ 1 =} 0 < a < :;:;-1 and
48. a n - 2n n2 n - 2n n·n-n--·n·n n·n·n--·n·n - 2n :;:;- - :;:;- - n-
lim O = lim .n!. =0 ⇒ lim an = 0 by Theorem 6.4. Hence, {a n} converges.
n---+oo n---+oo n---+oo
49. an = 23n > 2n3+ 1 =an+1 ⇒ { an} is decreasing for n � 1.
50. an = 2 - ¾ < 2 - n!l = an+l ⇒ {an} is increasing for n � 1.
51. n + 1 > n ⇒ -n+l < n

1
- .!. ⇒ --1
n+l- > _.!_ ⇒ 1 - -- > 1 - l ⇒
n
1
n+l n
-1 :::1 :::1}
1_ n+l > 1 n ⇒ { an} = { 1 n is increasing for n � 2.
l (n (x)-1
52 . an _- l nn(n)' Let f(X ) -_ l nx(x)' Then f'(X) _
-
l n (x)]2 > 0 c10r X > e =} f lS
increasing on (e, oo) ⇒ {an} is increasing for n � 3 > e.
nl n) x l x)
53. a n = ;� . Let f(x) = ;! = xe-x ln(x). Then f'(x) =e-x[ln(x)-x ln(x)+l]
= e-x[(l - x) ln(x) + 1], f'(l) > 0, f'(2) > 0, and f'(x) < 0 for x � e ⇒ {a n }
is decreasing for n � 3 > e (but not for n � 2).
54. an = s,��n) =} { an} is not monotone.
55. a n = ✓nn+l � 0 ⇒ { a n} is bounded below by 0. lim �

n->oo v n+11
= lim � =
n->oo V l+,;;n
oo
⇒ { a n} is not bounded above ⇒ {an} is unbounded.
56. an = 2� - ln(n) is decreasing since both L�}

and {- ln(n)} are decreasing.
Since a1 = ½, an �½ for all n 2'. 1 ⇒ {an} is bounded above by ½. Since
lim an = -oo, {an} is not bounded below ⇒ {an } is unbounded.
n--+oo
57. an = 1_-\ ⇒ {an} is increasing for n 2'. 1 by Exercise 51 with n replaced by

n+l
n +1. a1 = - 2 and lim an -1 ⇒ -2 �an �-1 ⇒ {an} is bounded.

=
n--+oo
58. an = n2 e-n. Let f(x) = x2 e-x. Then f'(x) = e-x[2x - x 2 ] = xe-x(2 - x) ⇒

f'(x) > 0 on (0, 2) and J'(x) < 0 on (2,oo) ⇒ f(x) � f(2) = 4e-2 on (0,oo) ⇒
0 �an � 4e-2 for n 2'. 1 ⇒ {an } is bounded.
1 n
n(n+1)(2n+l)
59. If n = l, then 1 = L i2 1(1+1)(2+1)
6
Assume '\""'
u i --
2
6
Then
i= l i=l
n+l
L i 2 = L i2 + (n + 1)2
n
n(n+l (2n+l ) n(n+l)(2n+ )+6(n+l ) 2 _
� 2
+ (n + l) = ;
i=l i=l
(n+l )[n(2n+l)+6(n+l)] _ (n+l)[2n2 +7n+6] -_ (n+l )(n+2)(2 n+3) _ (n+l)[(n+l)+l)[2(n+l)+l]
6 - 6 6 - 6
60. a1 = 1 < 3, and if an < 3, then an+l = 1 + Fn < 1 + v3 < 3 ⇒ an < 3 for all
n 2'. 1, by induction. Hence, {an}�=l is bounded above. a1 = 1 < 1 +JI= a2 ,
and if an < an+l, then Fn < � ⇒ 1 + Fn < l + � ⇒ an+l < an+2
⇒ {an }� =l is increasing for n 2'. 1, by induction. Since {an }� =l is increasing
and bounded- above, it converges by Theorem 6.6. Let L = lim an . Then
n�oo
an+l = 1 + Fn =? L = l+ v'I ⇒ (L - 1) 2 = L ⇒ L 2 - 3L + 1 = 0 =?
L = 3
±-/5_
2
Since an_ >
> l for all n - > 1 and ' hence' L = 3+2Js_
l' L -
l
61. If lim an = L, then an+l = 1- .an
1... ⇒ L =1- -£2 ⇒ L2 -L+ 2 =0⇒ L= ±p
n�oo
is complex, which is impossible since every an is real. Hence, L does not exist,
i. e. , {an} diverges.
62. a1 =2 ⇒ 1 � a1 � 2, and if1 �an � 2, then½ � a� � 1 ⇒ -1 �- a� � -½
=? 1 � 2 - � � !
⇒ 1 � an+1 � 2 ⇒ 1 � an � 2 for all n 2'. 1, by
. < - 1- ⇒ _.1...
> -- 1- ⇒
a
induction. a1 = 2 > �2 = a2 , and an > an+l ⇒ .1..
an an+l an an+l
2 - .1...
an
> 2- - 1
an+!
- ⇒ an+l > an+2 ⇒ an > an+l for all n 2'. 1, by induction.
Hence, {an } covergcs by Theorem 6.6. Then L lim an ⇒ L = 2 - z ⇒

n--+oo
L2 - 2L + 1 = 0 ⇒ (L - 1) 2 = 0 ⇒ L =1.
L ¼, 83 = 1+½+ ¼+ ½, 84 = 1+½+¼+ ½ + /6 •
00
1. 2
1
n, SQ = 1, S1 = 1 +½, 82 = 1 +½+
n=O
L= ¼, 81
00
2. =1, 82 = 1+½, 83 = 1+½+½, 84 = 1+½+½+¼, 85 = 1+½+½+¼+½-
n l
. S
3 . S = 1lffi = 1·lffi 3n-l = 2- .
3
n-+oo n n-+oo 2n+l
7. �
L; _!_
3n L; (.!3 ) n = �
= n� 1--
3
= l6.
n= 2 =2
I: 5. 23n 3-2n = 5 I: (�r = 51�!!9 = 45.

00 00
8.
n=O n=O
9. L=l5. 3
00
n
2n
2 -4 n = 5L
00
n=l
C96 r= 9
5l�"._g__
16
= 4;.
L=O 4
00 00
11.
n
n2}8n+3 = nL
=O
( 2n� l - 2 n�3 ), Sn =
(l - ½)+(½- ¼)+· · ·+(2n�l - 2n� 3) =
l - 2 n�3 => L 4n 2 }8n+3
n=O
= E.� S n = 1.
2+ 2+
12. I: 2 ✓nn 3 diverges by then th-term test because lim 2 ✓nn 3 = 2 -=/= 0.
n=l n->oo
I:
00
cos(n)
+cos(n)
. 2c n ) diverges by thenth -term test because hm 2 +sm . 2( n ) -=/= 0.
·
13. sm
•
n=O
2 n - > oo
L=O [1 - (-l)
00
14. n
] diverges by then l /1-term test because lim [1 - (-lt]-=/= 0.
n n->oo
L=l L
00 00
16. n2Jn = ni/ 2
converges because it is a p-series with p = � > 1.
n n=l
L=l :Ji = L=l

00 00
17. n
i/ 2 converges because it is a p-scries with p = � > 1.
n n
I:
00
18. nr�i , r > 0, converges because it is a p-series with p = r + 1 > 1.
n=l
00
L
00
19. I: nr-l = n/ r, r � 0, diverges because it is a p-series with p = l - r :S 1.
n=l n=l
ft 2x1:_
00
20.
I: 2n1:_ 1
diverges by the integral test because 1
dx = ½ ln(2x-1)/� = oo.
n=l
ft 2xe-x
00
2 2
21. I: 2ne-n converges by the integral test because dx =
n=l
2
- e-x ,� = e-1 < 00.
L n[ln(n)]P, p < l, diverges by the integral test because ft x[ln(x)]P dx =

00
22.
n=2
1 P
1�P [ln(x)] - l;i = oo.
ft x[ln(x)]P dx =
00
23. I: n(In(n)]P, p> l, converges by the integral test because
n=2
1 �P [ln(x)]
1 P
- l;i = P
� 1 [ln(2)] 1 -P < oo.
= 3 + 136 = f!, ft ft
00
24. (a) s = L
n=l
.;;4 � s2 ;4 dx :S R2 :S ;4 dx ⇒ 2\ :S R2 :S ½-
(b) Rn :S f; ;4 dx = ;3 :S 0.001 if n � 10.
J4oo x-3/2 dx -< R J3oo x- /2 dx ⇒

00
25. (a) s = � _1_
u n,/n
� S3 = 1 + 1
__
2v'2
+ _1_
3v13'
<
3 -
3
1, i.e., 1 :S R3 :S 1·
n=l
A :S R3 :S
(b) Rn :S f; x-3/2 dx = }n :S 0.0001 if n � 400, 000, 000.
L nfn+3 converges by the comparison test because nfo+3 L n}/

00 00
26. :S n}/ 2 and 2
n=O n=l
converges since it is a p-series with p =�> l.
)n _1 L
00
27. I: diverges by the comparison test because fo-1
> n1\2 and n/12
n=2 n=l
diverges since it is a p-series with p = ½ :S 1.
!: 1 converges by the comparison test because 4!:1

00
28. I: 4
n=O
converges since it is a geometric series with r = ¾ and lrl < 1.
J_ J_
00
29. I: converges by the comparison test because :S L 2 .;;2 for
n=O
2 3 2 3 2n n
00
n � 2 and I: ;2 converges since it is a p-series with p = 2 > 1.
n=l
E 2;t 1 converges by the comparison test because 2;t 1 �'t for n � 0

00
30. � 2 ;;;,2
n n
= 2
n=O
and E
n=O
00
rnr converges since it is a geometric series with r = � and lrl < 1.
31. E
00 2n
e!
n
converges by the comparison test because
2n n
e!
2n+2n
� � = --;n
2 2n
for
I: (�r
n=O
n � 0 and converges since it is a geometric series with r = � and

n=O
lrl < 1.
E
00
32. -fo+n diverges by the comparison test because -fo+n > n!n 2
� for n � 1
n=l
E
00
and ¼ diverges since it is a p-series with p = 1 � 1.
n=l
00
00
.jn - -
33. " lm an bn - 1 ' 0 < 1 < 00, and "
- -jn-l 1 -
L.,, n+l · an - n+l rv n -
-
y'n l
vcn - bn -->...
--r 1·--->oo L.,, bn
n=l n n=l
⇒ E an diverges by the limit comparison test.
00
diverges
n=l
·
34. " n3 +2n-l an
OO
L.,, n4 +n3 +2· = n3 +2n-l rv !I:_3

n4 +n3 +2 n4
= l = bn ⇒ lim ab n = 1, 0 < 1 < oo, and
n=O
n n--->oon
E bn diverges ⇒ E an diverges by the limit comparison test.

00 00
n=l n=O
35 " �
OO
- � rv n2 / 3 - 1- - - 1, 0 < 1 < 00, and
b -
.3� 3�
· L.,, n2 +n -l · a n - n2 +n -l n4/3 - bn l'lm !!n.

- -
n2 =}
n-+oo n
n=l
E bn converges ⇒ E an converges by the limit comparison test.

00 00
n=l n=l
n l /3
00
36. " 1 . a
� n2/3 Jn(n) n
= 1
n2 /3 Jn(n). Lct bn = 1 1· a
n. Then n.:_,� b: = 1·
. :_, � ln(n ) oo and
n 2
n
Eb ⇒ E an diverges by the limit comparison test.

00 00
n diverges
n=l n=2
ln(n) Let b ln(n) Eb
00 00
l
T hen 1.1m b _ l'1m """""I/2
an - 0 and
37. E ��)- n2 . n n n
n=2 n-+oo n n-+oo n=l
Ea
00
converges⇒ n converges by the limit comparison test.
n=2
= + ln t] = ln(e) = 1, 0 < 1 < oo, and Eb

ex,
lim �n lim ln[(l n converges ::::}

n-+oo n n-+oo n=l
Ea
00
n converges by the limit comparison test.

n=l
39 � [ln(n)] - [ln(n)
an - Let bn - 4/3
1 . Then 1·1m 1·1m [ln(n)] 0 and
3 3
] !!n -
3
n3/2 .
. L., n 'n . -n b -
n=2 v ·,t, n-+oo n n-+oo nI/6
00 00 _
L bn converges ⇒ Lan converges by the limit comparison test.
n=l n=2
40. f (;;,2t converges by the alternating series test since

n=l
bn
decreasing for n � l and lim bn = 0.

n-+oo
. 1· (-l) n
(- ) n
d"1verges by the n th- term test smce n-=-.� _j_ 0
41 . �
� 2+si
l
n(n) 2+sin(n) 1 .
42. f (� converges by the alternating series test since bn = � > 0, {bn} is

n=2 y ln(n) n(n)
yl
decreasing for n � 2 and lim
n-+oo
bn = 0.
43. f
n=2
�;:-�l�n) converges by the alternating series test since bn = en l� n)
(
> 0, { bn} is
decreasing for n � 2 and lim
n-+oo
bn = 0.
44. f
n=l
(-i� -i
n
diverges since f
n=l
(-�)
n
converges and f ¾ diverges.
n=l
45. f l +(
:;
n
) n
converges since both f ,;2 and f (-�) n
converge (Theorem 6.8).
n=l n=l n=l
(a) L
00
(-l) n -l
_ _ 31 _ 1
36 I R3I <
46. -n- ~
2 - ~S3- 1 - 41 + 91 - , b4 - 16 .
n=l
(b) IRnl < bn+l = (n1l)2 � 0.0001 if n � 99.
47. (a) L
OO (-l) n ~ 1 _
l _ 11
I R2 I <b3-_ 21s·
n3+1 ~S2-l-2+9-18'
n=O
(b) IRnl < bn+l = (n+l\3+1 � l,�Ol if n � 9.
48. f (;;,�t converges by the alternating series test but f I (;;,�t I = f

n=2 n=2 n=2
n� l
oo ( l)n
diverges. Hence, L ;;,_ 1
n=2
converges conditionally.
00 00
( 1) n+1 00 1) n +1
49. L -
n..fii,
converges absolutely since L I (� I = L
1
n..fii,
converges.
n=l n=l n=l
..fii,
50. I: (-�: ../n

n=O
1
converges by the alternating series test but f I (-�: ..fo I
n=O
1 -
00 ..fii,
- L n+l
n=O
diverges. Hence, f (-�:1../n converges conditionally.

n=O
51. I:
n=l
(-�r diverges by the n
n
th
-term test since lim (-n
2
n->oo
r =J. 0.
52. ( l)n ( n)
converges by the alternating series test since bn = [ln�)]2 > 0,
2
I: - �n j
-i t n 2
f
n=2
{bn} is decreasing for n::::: 8 and bn = 0. Since I( � ( nl l I =

[ln�)j2
��
1 n�2 2
diverges, (-l) (n)

converges conditionally.
n 2
I: !�n )
n=2
53 . � (-
� 3
n_
2)n
1 converges abSO1 ute 1y smce
. �
;;:l
I (-2) n
3
n_
1 I-�
-
;;:l 3
2n
n_
1
converges.
54.
00
55. � 2 n +i(n+l)4 diverges by the
3n I n2
ratio test since lim
I I=
an+
i
n->oo an
n=O
3"(n+l) 2 2n +1(n+l)4 _ �
rlffi n +2
n->oo 2 (n+2)4 3n 1 n 2 - 2 > 1.
00
56. converges by the ratio test since lim
3 2 2 l ,
E n
+ � -
n. n----+oo I I=a,,+i
an
n=O
l" (n+l) 3 +2(n+l)2 -l n! _
0 < 1·
n!....� (n+l)! n3 +2n2 -l -
57. I:
(n�l)!
converges by the ratio test since lim
n->oo
I aan+i I =
n
n=O
lim 2(n+l) (n+l)! 0 = <1
n->oo (n+2)! 2n
58.
I: n
n
� ({:)! converges by the ratio test since lim
n-----+oo
I aan
,,
i
+
I=
r r
n=l
= < 1.
[(n+1)!) 2 n 2 (2n)! n2 l
-
n!....� (n+l) (2n+2)! (n!) n !....� (2n+2)(2n+l) - 4
2 2
00
59. � 3_6 _t'-( 3n) converges by the ratio test since n----+oo
lim I a n
:
+1
I=
r
n=l
lim (n+l)! 3-6-9---(3n)
n!
= n+l - l
< 1·
n->oo 3-6-9···(3n)(3n+3) n!....� 3n+3 - 3
60. I: �� ::�\\ diverges by the ratio test since lim

( n->oo
I ana+i I =
n=O
n
> 1.
l" (2n+3)! n!(n+l)! _ l" (2n+3)(2n+2) _
n!....� (n+l)!(n+2)! (2n+l)! - n!....� (n+2)(n+l) - 4
61. I:
�
%�::;/ converges by the ratio test since lim I -
a n +i
¾
I=
l" (n+l)!(n+4)! (2n+3)! _ l" (n+l)(n+4) _ 1
n!....� (2n+5)! n!(n+3)! - n!....� (2n+5)(2n+4) - 4 < l·
(-n )n
L
00
62. [l n (n)J2n diverges by the root test since lim /an / l f n lim [I n (nn l2 = oo > 1.
= n-+oo
n=2 n-+oo )
;p..
( n n2n
63. u 2n-l) 2
(n2+l)n converges by the root test since lim /an / 1/ n =. lim 2(nn2+l ) = 21 < 1.
n =O n---+oo n---+oo

= O, = 1 and 1�r-r = 1 and th e formu1a is true. A ssume Sn 1-rn+l -
n +l
1. At n so =�
Then s n+l = sn + rn+ l = 1-rn+l + rn+l = 1-rn+l+(l-r)rn+l 1-rn+2
1-r 1-r �
n
L ak and,
n
2. Let Sn =
k=l
for n 2: m, t n = L ak .
If m = 1, then Sn = t n . If
k m
=
m 2: 2, then Sn = Sm -1 + t n for every n 2: m. Hence, lim Sn exists if and only
n---+oo
if lim ( Sm -1 + t n) exists, if and only if lim t n exists, since Sm -l is a constant.
n�oo n�oo
Ea Ea
00 00
Thus, n converges if and only if n converges for every integer m 2: 1.
n=l n =m
3. The function f(x) =

xl n (x){i1n [In (x ]} is positive, decreasing and continuous for
)
x 2: 16 > e and J1c;:' f(x) dx = ln{ln[ln(x)]}I� = oo. Hence, the series diverges
e
by the integral test.
00 00 00
4. '°' = 2 = an converges by the limit comparison test with
6 1+2+;+· +n u -(
'°'
n n+l) u
'°'
n=l n=l n=l
bn = n2 ·
1
El an converges by the limit comparison test

00 00 00
S. L
n =l
H4+9':-.-+n2 L
n=l
(n+1)(2n+l)
n=
with bn = ;2•
Ea
00
6. Let be a finite series, i.e., an = 0 for n 2: m, where m 2: 1 is an integer.
n
n=l
m
L L
n
Then Sn = /akl = /akl = Sm for all n 2: m =} lim Sn = lim Sm = Sm=?
k=l k=l n---+oo n---+oo
E E an converges absolutely.
00 00
/an / converges and, hence,
n =l n=l
7. an = Tn ⇒ Tn = rn -l + rn 2 ⇒ r2 - r - l = 0 ⇒ r1' r2 = l±2V5' an = C1rf + C2 r2

⇒ C1 r1 + C2 r2 = a1 = 1 and C1rr + C2r� = a2 = 1 ⇒ C1 = � and C2 = J5 ⇒
n n
a = _l_ (l+V5) - _l_ (l-V5) = _l_ [(l + vfs) n - (1 - vfs)n] ' n > 1.
n V5 2 v'5 2 2n y15 -
C. 7. CHAPTER 7 SOLUTIONS 419

� (x- 2) = + x-2 + (x- 2) + (x-2) + (x- 2) •

n 2 3 4
2. L, Jn+l l J3
y'2 2 y'5
n=O
1 _ x+l + (x+l) _ (x+l) + (x+l)

2 3 4
� (x+l)
n
3. L, ( -3)" = 3 9 27 81 .
n=O
� (x+2) _
n
(x+2) 2 (x+2) 3 (x+2) 4 (x+2) 5
4. L.., - ( X + 2) + + + 16 + 25 .
n2 4 9
n= l
oo
5. L x2" x2 x4 x6 xs
Jn+l = l + y'2 + J3 + 2 + v15·
n =O
� (x-3)
3n
_ 3 (x- 3) 6 (x-3) 9 (x-3) 12 (x-3) 15
6- L, nfo = (X 3) + v'2 + 3✓3 +
2 8 +
5 v'5 .
n=l
00 00 00
7. L=O 2 n
( x - l)71 + L=O n l (
� x - l)71 = L [2 n
+ n�l ] (x - l )71.
n n n= O
00 00 00
8. a La n (x - a)'1 + (3 Lb n (x - a)'1 = L= (aa n + f3b )(x
n - a)'1.
n=O n =O n O
00 00
9. 3 + 2x - 5x +
2
Lb nx
n
= L= C n X
n
=? Co = 3 + bo, C1 = 2 + b1, C 2 = -5 + b 2,
=O
n n O
00 00
10. 2+(x-3)-(x-3) +
4
L= b (x-3)
n (x-3)1 1
n
>i
= LC ⇒ Co=2+bo, C1 = l+b1,
O n n=O
C 2 = b 2) C3 = b3 ' C4 =-1 + b4) Cn = bn) n 2". 5.
00
11. LC n
n 2 3 4
X =Co+C1 X+C 2X +C3X +C4 X + C 5 X 5 + · ··= (Co+ C 2X + C 4 X 4 + · · ·)+
2
n =O
00 00
(c1x+ c 3x + C 5 X + ···) =
3 5
I: c 2k x
2k
+ I: C k+1x k+l_
2
2
k =O k=O
12
.
� 2"xn R - 1·lffi
L, n+l ,
n=O
-
n--,oo Cn+1
1· I� I -
� n+2 - 1 1 1 1
- lffi n+l 2n+l - 2. j X j < 2 =? -2 < X < 2. At
n__,oo
00 00
x _ 12,
--
=
(-l) n
L .
n+l converges, and at x
_ 1
2,
1 ·
n+l diverges I _ [
1 1
)
2, 2 . - I: ⇒ - -
n O n=O
13. E
n=O
nf+1, R = n-+oo
lim /_fn._/ = lim (n+
Cn +1
;l 2 +1 = 1. Ix/< 1 ⇒ -1 < x < l. At
n-+oo n +l
oo ± n
x = ±l, E ( l) converges⇒ I= [-1, 1].
n2 +i
n=O
� vn(x-6) n , R - I'lill
14 · L.J - 1lill
I _fn._I I - · vn (n+ 1 ) 3 + 1 - 1. / - 6/ < 1
x ⇒
n3 +l n-+oo Cn+ n-+oo 3 + 1 v n+ 1
n
e7, -
n=O
oo
-1 < X -6 < 1 ⇒ 5 < X < 7. At X - 5, E yn(-l)n
n3 +l converges, and at x = 7,
n=O
00
E n�l converges ⇒ I = [5, 7].
n=O
15. f (-l2n(+x-l l )
n
)
n
' R = lim l_fn._l
n-+oo en+ !
= lim 2n+3
n-+oo 2n+l
= 1. Ix - 11 <1 ⇒ -1 < X - l < 1
n=O
00
⇒ 0 < x < 2. E 2n�
00 (
At x = 0, l diverges, and at x = 2, E )n
2�� 1 converges ⇒
n=O n=O
1=(0,2].
16. E (x1,�
n=O
)
n
, R = Iim I _fn._ I = lim
n-+oo Cn+J n-+oo
2
;! 1 = 2. / X + lI <2 ⇒ -2 < X +l<2 ⇒
00 00
-3 < x < l. At x = -3, E (-1r diverges, and at x = l, E 1 diverges ⇒
n=O n=O
I= (-3, 1).
17 · �
2n (2x-4)n �
= LJ 4n (x-2) n R = lim I _fn._ I = lim ___±'.'._ n+2 = .!. . /x - 21 < .!.
✓
LJ Jn+l C77 '
n-+oo Cn +J n-+oo Jn+l 4n +J 4 4
n=O n=O yn-r-,
⇒ _.!,
4
< x - 2 < .!.4 ⇒ ?..
4
< x < g4. At x = ?..4' �
D
(-
l)n
Jn+l
converges ' and at x = g4'
n=O
2n (6-3x) n = � (-6) n (x-2) n R = Iim _fn__ = lim ..E_ n �� = 1._ /x 2 < 1

18. � - / _6
n-+oo l Cn+ 1 l
LJ n+1 D n
n=O n=O
n+ 1 ' n-+oo n+ 1 6 6
00
⇒ - 61 < x - 2 < 61 ⇒ 6 11
<x< 6 13
. At x = 6 11 """ 1 . _ 13,
, L.J n+l d1verges, and at x - 6
n=O
� (-l) n converges
u n+l ⇒ I= (1.1. 13
6' 6 ] '
n=O
19. �
n=O
(x-3) 2" lim an ! = lim (x-3)
LJ 2n ' n-+oo an n-+cx:> 2 n +I
(x
I I
3)
+
2
2
I
_-2n_2n = lx-31 < 1 if and onl y if
2n +2
-I
/x - 31 < v'2 ⇒ R = v'2. /x - 3/ < v'2 ⇒ -v'2 < x - 3 < v'2 ⇒
00
3- v'2 < x < 3 + v/2. At x = 3 ± v'2, E 1 diverges⇒ 1 = (3 - v'2, 3 + v/2).
n=O
20. E vn-+l
�' n-+oo
n=O
lim I a a i I = lim
n-+oo
n+
n I
J(
x3
"+32
n+l) +l
✓:�: 1 = /x/ 3 < 1 if and only if /xi < 1
1
⇒ R = l. lxl < 1 ⇒ -1 < x < l . At x = -l, I:

=O
n
)-�)
n+l
n
converges, and at
00
x = l, E ;
✓n +1
diverges ⇒ I= [-1, 1).
n =O
21. I: 2n
( 3x - 2 ) ,
an
lim I +i I
-->oo an
= lim I
(3x-�)2::2
n-->oo (3x 2)
1 = l3x - 21
2 < 1 if and only if
½ ⇒ R = ½. Ix - � I < ½ ⇒ - ½ < ½⇒
n =O
n
I 3x - 21 < 1, i.e., Ix - � I < x- � <

½< = ½ and at = l, E 1 diverges ⇒ I= (½, 1).
00
x < l. At x x
n =O
22.
23.
'\""'
u
2 2n (x-3)n R
nl ' -
_ l'lill I I - 1·
_fn_
c - lill
22n (n+l)! _
I 2n+2 - lill
1· n+l _ ---'-- J _ -00 CX) •
4 - CX) ----r -( , )
n=O · n-+oo n+l n-+oo n. 2 n-+oo
24. u
� (2
(n l)2
n=O .
n)!
X
n
, - 1·1m
R-
n
I 1-
....fn_
-->oo Cn +I
-
1·lill
n
(2n)! [(n+l)!j2
-->oo ( 1. )2 (2n+2)I.
n
·
- 1lill
(n+1)2
-->oo (2n+2 )(2n+l)
n
- 1
- -4 .
� ((n-1)!] (x - 1 2n lim an+1 l m I (n!)2 (x-1) 2nt2

=
2
=
{2n)!
25. � (2n)! ) ' I I i I
n -- >oo an -->oo (2n+2)!
n
[(n-1)!) 2 (x-1)2n n
1. •f • I I< ⇒ =
n2 lx-ll 2 _ lx-11
<
2
n-=-.m (2n+2)(2n+l) - -- 4 1 1 and only 1f X - l 2 R 2.
26. (a) (fo a X

n
) (�b n
) = fa Cn X
n
, Cn = 1 akbn-k = 1 an-kbk.
[t
x
n n
(b) [J; an(x - ar] bn(x - ar] fa en(x - ar,

0
n n
Cn = E akb
k=O
n -k = E an-kbk.
k=O
⇒ lxl < ½ = R.
00 00
1 . f(x ) = 1)2x = E (-2xr = E (-2r x , I - 2xl
n
<1
n=O =O
n
(�r = I:
00 00
2. J(x) = 2�x = l �;f
2
= 3 I: /n x ,
n
l�I < 1 ⇒ lxl < 2 = R.
n =O n =O
00 00
3 · j'(x) - 1 -
- 3+x
- 1 1 - 1 '\""' ( - x)
3 1+;£ - 3 3 u
n
-
'\""' (-l)n
- U
3n+1
x
X > I -3 I
n
<1 ==? X I I < 3 = R·
3 n=O n=O
4 22 APPENDIX C. SOLUTIONS
L2,._ L=O (Jr= L=O ⇒

(X)
4. f(x) = 4�x =
(X)
=� 22l+1X
n+1
, l¼I < 1 lxl < 4 = R.
4
n n
- 6 1 - 6
1 +ie. - 1 '\' (-l) n n
. X) - ⇒ I X I < 3= R ·
(X) (X)
5 . f( +2x l 3 - 6 6 (-3 ) - u 2 . 3n +l X '

1 x n - '\'
/-3 /
x
< 1
n =O n=O
6. f(x) = 1lx 2 = L=O(x r = L=O x , lx2 1 < 1 ⇒ lxl < 1 = R.

(X) (X)
2 2n
n n
7. f(x) = - 1 = _1 � = l. � =� (-l)+ n 3n I < 1 ⇒ R = 21 / 3

n
(- x ) ,_x
3 3
2+x 2 1+L 3
2 LJ 2 LJ 2n 1 X ' 2
n =O 2 n=O
L=O x L (3xr for lxl < 1-

(X) (X)
8. f(x) = l.!3x = - 1 and l3xl < 1, i.e., lxl <

n
l�x -
n n=O
Hence, J(x) = I: (1 - 3 )x , with R = lim
n =O
n
->(X) l_fn_l
C +1 = lim
n
->(X) n n n
1 � �:1
3 = ½-
L1,._ L-
1� =½I: (-�r +½I: (2;r for lxl < �-
(X) (X)
9. f(x) = 2�x + 3l2x = 2

+ 3 n =O n=O
l ) "hR- -1"-> _fn_l-
Hence, f() - u [(-
X -� lffi l
n
n
n
2n H + 3n+l ] X , Wlt - .
2 3
n = Cn +1 2
n=O
: �i = -(x +1) L=O x = - L x2n _ L x2n+2 = - L x2n _ L x2n =

(X) (X) (X) (X) (X)
10. f(x) = 2
2
2 2n
n n =O n=O n =O n=l
L x2n - L x2n = -1 - L 2x2n , jx 2 1 < 1 ⇒ R = 1.

(X) (X) (X)
-1-
n=l n =l n=l
12. f(x) = e-3 x =

n =O
I: c-:r x , R = oo.
I: ;h(-3xt = n=O n
e2ex -2 = e2 L=O ;h(x - 2r = L=O �(x - 2r,

(X)
13. J(x) =e = R = 00.

2
x
n n
15. f(x ) = e-3x =

2
n
I:=O ;h(-3x t = I:=O c :r
- x2
n
2n
, R = oo.
16. J(x ) = L �[-3(x - l)] n

(X)
e - 3x = e - 3e - 3 ( x -l ) e-3
n=O
R= oo.
( 1) 2k +1
=� (-1) 22k+1 2k+1 R = oo.
u (2-k+l)! (2x)
17. f( x) - sin(2x) =�
k k
u (2k+ l)! x '
k=O k=O
(-l) k ( x. 2k _ (-l}k3 2k
x2k , R·. -
_ oo.
18. J(x) _ _
- cos(3x) - oo
I: (2k)!
k=O
3 ) - L
oo
k=O
( k}!
2
19. J(x) = sin(x) = sin[1r + (x -1r)] = sin(1r) cos(x -1r) + cos(1r) sin(x -1r)
· � (- 1 ) k +1 2k _
-sm(X - 7r) _
- L.., (2k l}! (X - 7r ) + l , R - 00.
+
k=O
20. J(x) = cos(x) = cos[1r + (x -1r)] = cos(1r) cos(x -1r) -sin(1r) sin(x - 1r) =
00 ( ) k +l
- cos(x -1r) = L..,
"\"' ....=]:__(x
(2k}! -1r)2k ' R
· = oo.
k=O
21. tan-1(x) = f ��'x

n=O
2n+1
, jxj < 1, ⇒ tan -1(2x)= f ��;(2x)
n=O
2n+1
=
� (-l} n22ntJ x 2n+l l
L.., 2n+1 ' 2xj < 1 ⇒ lxl < 21. = R·
n=O
00
22. x�l
= 1+(!-2) = L=O(-lt(x -2r, Ix - 21 < 1, ⇒
00 ( l}
n
ln(x -1) = J -1
x- dx = "\"'
- (x - 2)n l + C X= 2 ⇒ C= 0 ⇒ n
+
L.., n+l 1 ' n=O
ln(x - 1) = f
n=l
(-l�n -i (x
- 2) n, R = 1.
_1 � -x-2 n = � (-l)n
23. l
x
1 _ 1 1
(x-2} - 2 l+x 22
2+ 2 L..,
( 2 ) (
L.., 2n +i X -
2) n ' I -x-2-
-21 < 1, ⇒
n=O n=O
00
J
ln(x)= .!.x dx = L.., (-l} n
l
"\"' 2n+l (n+l} (x -2) + C ' x = 2 ⇒ C = ln(2) ⇒
n+
n=O
ln(x) = f=
n=O
2J;}l: 1 ) (x -2r+ 1 + ln(2), jx -21 < 2 = R.
00 00
24. 1+(L2)2 = L=O [-(x - 2) t = L=O(-lt(x - 2)
2 2n
, jx - 2j < 1, ⇒
00
n n
tan-1 (x - 2) _
- J l+(x-1 ) 2 2
_
dx - L=O (-l}
2 l (x -2)
n+
n
2 _2 ⇒C_
n+l + C, x - -0 ⇒
n
tan-1(x -2) = L..,

� 2)2n+ l ' R = 1.
n
(-l} (x -
2n+ l
n=O
26 . f(x) = x 2
= x2(l+x)-l/3 ' a=_.!.3 ' (ak )= (-1) ,1.4.7
k ... (3k-2} k > l ⇒ x = 2
3/f+x 3 k! ' - ' lfJ:+x

� (-l) k ·l-4-7---(3k-2} k _ .2 � (-1 )k , 1 ,4.7,.. (3k-2) k +2 I _
[1 + X ] - X + , X I < 1 - R.
2
X L.., 3kk! L.., 3kk! X
k=l k=l
00
( 2k +l)! x 3k+l
I: 2k(k!)2 , l x l < 1= R.
k=O
2
29. e x =l+ x+½x 2+ ¾x 3+·· · and sin(x)= x - ¾x 3+··· =>ex sin(x)=

( 1+ x+ ½ x 2+ ¾x 3+· · ·) ( x - ¾x 3+··· ) => T4 (x) = T3 (x)= x+ x 2+ ½ x 3.
00 00
32. 1.x = l+( x1-.1 )= � n n ( l)n - l)n+l C
� -
LJ (-l) (x - l) => ln(x) = J .!x dx= LJ n +l (x + '
n=O n=O
x = l => C= 0 => ln(x) = L -=¼-(x - 1r, x = l + (x - 1) => x ln(x) =
OO ( )n I
n=l
[1 + (x - l)][(x - 1) -½(x - 1) 2 + ½(x - 1) 3 - ¼(x - 1) 4 + . . •]=
(x - 1) +½(x - 1) 2 - ¾(x - 1) 3+ ;2 (x - 1) 4 +·· ·,
00
n x ( x)
_1_ =
2-x
1 .
1-( x-l)
= � 2 3
In x =
u (x - l) = 1+ (x - 1) + (x -1) + (x - 1) + . . . => 2-
n =O
[(x-l)+½(x -1) 2-¾(x-1) 3+ 1\(x-1) 4 + • • l[l+(x -l)+(x -1) 2+(x -1) 3+· · ·]
=> T4 (x) = (x - 1) + �(x - 1) + !(x - 1) 3 + g(x - 1) .
2 4
oo (-lj k 2k+l s· 1) _ oo (-l) k k · ·

33. s·m(x) _
_ �(
- I: ( 2k +l)! x => m( 2 - I: 2 2 k+1( 2k +l)! - 0 -1) bk 1s alternatmg,
k=O k=O k=O
IRk l < bk+l= 2 2k+3( 2k +3 )!, k= l ⇒ IR 1 I < 2i51 � 0.00026 < 0.001 ⇒
1
sin(½) � s 1 = ½ - }8= �� 0.47916. The value by calculator is 0.47942, with

an absolute difference of 0.00026 < 0.001.
� (-l)k 2k � (-l)k k fl
34. COS (X) = 0 ( 2 k )! X => COS( yX r;;.) dX =
r;;.) = 0 ( 2k )! X => Jo COS( yX
k=O k=O
� (-l)k x k +11l - � (-l)k - � ( )k . alternatm .
u ( k +l)( 2k)! O - 0 ( k +1)( 2k )! - 0 -1 bk lS g, IRkI < bk+l
k=O k=O k=O
1
l
( k +2 )( 2k +2 )" k= 2 => IR2 I < 2 80 � 0.0003 < 0.001 =>
J�1 cos( vx) dx � S 2 = l - ¼+ i2= �� 0.7638, with the required accuracy.
k
L (2k+2)!
00
(-l) (2k)!_ � _
- L..., (-1) bk i· s alternat·mg, I Rk I < bk+l - (sk+9) 210k+11[(k+l)!]2,
k
(Sk+l)210k+l(k !) 2
k=O k=O
k = 0 ⇒ IRol < b1 =
g.�10 = 92\6 � 0.000108 < 0.001 ⇒
J
1
2
0 ✓i�xs dx � so = ½ = 0.5, with the required accuracy.
3x - 1)2 COS2(X) _ 11· x ( 3x+-29 x 2+... )2(l --2J x 2 +.. )

2
. X(e sm
36. 11m . 3x - m
x-+O ( ) x-+0 1 3+..
( x -6x .
)
3+2 x +.. ) (·l -21 x +·.. )

2
2 2
lim ( 9
3 = 9.
x-+O ( J-6x
1 2 +.. )
. )3 Jn(l-x) 1· (x+2)3 (-x-½x 2 -... )

38. 11m (x+2
. = 1mO (2 x -6x = -4.
x-+O Stn(2 x) cos(3x) x-+ s 3+... ) ( l -2X
9 2+ · )
1
4·
40. I 1P < _e
(x)I - •- l n+l ' x = l ⇒ 0 < z < l ⇒ 0 < e z < e l /3 < e ⇒
(n+l)! lx - -3 - -
IRn (½) I
Ln 3
< 3 n+l(n+l)! ::; 0.0001 if 3n+1 (n + 1)! � lOOOOe � 27,182.8, and n = 4
⇒ 3n+l (n + 1)! = 35 5! = 29,160 > 27,182.8 ⇒ ffe � 1 + ½ + }g + 6-17 + 24\ 1 �
1.395576. The value by calculator is 1.395612, with an absolute difference of
0.000036 < 0.0001.
L
-;/xn and x = 1� ⇒ ln(0.9) = L n�b n · f(n+l)(z) =
(X) (X)
41. f(x) = ln(l - x) =

n=l n=l
-n! P _ f(n+l)(z)x n+l _ -x n+l p 1 _ -1
(1 -z)n+l ⇒ Ln
1 ( X ) - (n+l)! - (l -z)n➔ 1 (n+l) ⇒ 1 Ln ( 10 ) - ( l -z)n+ 1 (n+l)1Qn+l ·
0::; z::; /o ⇒ 0::; l�z ::; 190 ⇒ Uo) IRn I ::; n+l
gn+l(n+l) < 0.001 if 9 (n + 1) >
1,000, and n = 2 ⇒ 9n+l (n + 1) = 93 · 3 = 2,187 > 1,000 ⇒ ln(0.9) �
-/0 - 2�0 = - 22010 = -0.105. The value by calculator is -0.10536, with an
absolute difference of 0.00036 < 0.001.
42. J(x) = sin(x) = sin[a + (x - a)] = sin(a) cos(x - a) + cos(a) sin(x - a)

( 2k)! a L ( 2k+l)! - a) 2k+l -
oo (-l)k (x - ) 2k + cos(a) oo (-J)k (x
s·m(a) L _ L oo e (x - )n , with
n a
k=O k=O n=O
_ (- l) sin(a) and
k _ (-l) cos(a)
k
C2k - (2k)! C2k+l - (2k+l)! , k � O , and R - _ 00.
43. J(x) = cos(x) = cos[a + (x - a)] = cos(a) cos(x - a) - sin(a) sin(x - a)

cos(a) L (-l) 2k - s · (a) oo (-1)" (x - ) 2k+l _ L
L
00
k
oo c (x - ) n with
( 2k)! (x - a) m ( 2k+l)! a - n a ,
k=O k=O n=O
_ (-l) cos(a) and
k _ (-l) sin(a)
k 1
> 0, and R -
C2k+1 - (2k+l)! , k _
1
C2k - (2k)! _ 00.
'°"'
00
44. eix = L.J .ln!.(ix)n = L.J (_ '°"'
00
l (ix)2 k +
_
2 k)!
'°"'
00
_1 - i 2 k+1 =
L.J ( 2k+l)! ( x)
n=O k=O k=O
� (-l) k 2 k � i(-l) k 2 k+l _
L.J ( 2 k)! X + L.J ( 2k+l)!x - cos(X) + ism · · (X ) .
k=O k=O
4 5. (a) If ab� 0, then ab= labl =lallbl. If ab< 0, then ab= -labl < labl= lalJbl.
Hence, ab :s; lalJbl.
(b) (a+b)2= a 2 +2ab+b 2 :s; a 2 +21allbJ +b2= lal2 +2lallbl +Jbl2 =(Jal +lbl)2
==? J(a + b) :s; J(lal + lbl) ==? la + bl :s; lal + lbl.
2 2
E la x E Jb x
00 00
(c) Since n
n
l converges for lxJ < R 1 and n
n
l converges for
n=O n =O
E (la x l + lb x l) converges for lxl <

00
lxl < R 2 , n
n
n
n
min{R1, R 2 }- Since
n =O
L=O l(a
00
l(an +bn)xn l= lan xn +bn xn l :s; lan xn l+lbn xn l by part (b), n +bn )x l
n
n
converges for Ix! < min{R1, Rd by the comparison test. Hence,
R � min{R1, R2 }-

1. Iff is periodic with period T, then J(x+T)= J(x) and f(x) =f((x- T)+T) =
f(x - T), f(x ± 2T) = f((x ± T) ± T) = f(x ± T) = f(x), etc. Hence,
f ( x + nT) = f ( x) for every integer n.
2. f(x) =x2 + 1 for -1 :s; x< 2 and f(x + 3) = f(x) for all x. Since f has period
3, f(5) = f(-1) = 2, J(7) = f(l) =2, J(-4) =J(-1) = 2, J(783) =f(0) =1
and J(-291) = J(0)= 1.
00
3. Multiplyf(x) = T+I°:: [an cos( nzx ) + bn sin( nlx )] by sin( mzx ),wherem� 1
n =l
is any integer, integrate from - L to L and reverse the order of summation and
integration to obtain J:
L f(x) sin( z ) dx= T L sin( z ) dx
m x m x
J:
+ f1 [an J:L cos( n x
l ) sin( z ) dx + bn
m x
J:L sin( l ) sin( mzx ) dx].
n x
The first integral on the right is L sin( mzx ) dx=J: cos( mzx ) l�L = 0. -::7r
Employing the trigonometric identity cos(a) sin(b) = ½ [sin(a + b) - sin(a - b)]
with a= n1rLx and b = mL1rx ' we obtain
J!L cos( l ) sin( mzx ) dx = ½

n x
J:L { sin [ ( +7)1rn x
] - sin [ ( n -7)1rx]} dx = 0. If
n =/- m, then the trigonometric identity sin(a) sin(b) = ½ [cos(a - b) - cos(a + b)]
with a= nt and b= mzx gives
J_�L sin ( n
l )
x
sin ( mz x
) dx = {
½ f!:L cos [ (n-7) 1rx] - cos [ (n+7) 1rx] } dx = 0.
If n = m 2'. 1, then, employing the trigonometric identity sin 2 (a) = ½ [1-cos(2a)]
with a = mz , we obtain f!:L sin 2 ( z ) dx = ½ f!:L [1 - cos (2�7rX )] dx
x m x
2 2 m1r L '
l [x - 2-__ sin ( 2 m 1rx)] I L = L. Thus b m = l.L J-LL f(x) sin (mL1r ) dx m > 1.
-L
x
' -
4. f(x) = { �: -� �:: � } and J(x + 4) = f(x) for all x. T = 4, L = 2,
an t
= ½ 2 f(x) COS ( ; ) dx = ½ }� COS ( ; ) dx = nln sin ( ; ) I� = 0, n
2n x n x n x
2'. 1,
dx = _....!... (-l) l.
cos (nn2 ) 1 0 = [l-n
n
a = l r 1dx = 1 b n = l r sin ( The
2 2 nn ) 2
' n7r 7r
x x
0 2 Jo 2 Jo
f
2
Fourier series of J is ½ + 11-�-:,1 ) n] sin ( ; ). n x
n=l
5.
J(x) = {0 0<x<l
x'. 1 � x < 2
} and f(x + 2) = J(x) for all x. T = 2, L = 1,
an =
f02 f(x) cos(mrx)dx = t xcos(mrx)dx = : sin(mrx)I� -
2 _ 1 -l n _ 3,
- n21r2 cos ( n1rx)1 1 - -n(21r 2) , n 2'. 1, ao -J1 x dx -
. (n1rx)dx _
1 J 2 sin _ x2 1 2 -
1r
1 2
nn 1 2 1 2
bn = Ji2 x sin(mrx)dx = -: cos(n1rx)I� + �
n 1 cos(n1rx) dx = -,;1r + n� cos(n1r)
2
1r
f
= (-�;- 2
, n 2'. 1. The Fourier series of J is
¾+ fi [ 1
�t;) cos(n1rx) + (-�;- sin(n1rx)] .
n 2
6. f(x) = 2x - 3 for 2 ::; x < 5 and J(x + 3) = f(x) for all x. T = 3, L = !,

a =
f5 5
j 2 (2x-3)cos ( 2n1r3 x) dx = � (2x-3)sin ( 2 ; ) I� -n1r2 2 sin ( n 2 1rx)
dx = n x
J
5
n
3
1 n
n
....!...n [7 sin ( 3 ) - sin (4 3 )] + n_32 n_2 COS ( 2 3n ) 12 = _§_ sin ( 4nn

0 1r nn
3
) + n x
3 1 0n1r) 4n 1r )] = _§_ sin (4n1r) n > 1 a = � r 5 (2x - 3)dx =

n n1r
n-2 -2 [cos ( 3 - cos ( 3 nn 3 ' - ' 0 3 J2

f
1r
j[x - 3x]I� = 8, bn = j 25(2x -3)sin ( 3 x) dx = -;1r (2x -3)cos ( 3 x) I�+

2 2 1r
n 2 1r
n
..1... r cos (2 1r ) dx = _....!... [7 cos (1on 1r ) - cos (4n 1r )] + _23_ sin ( 2 ) 1 5 =

5
n
n x n1rx
nn J2 3 n
4 1r
3 3 3
-
2 2
:1r cos ( � ) . The Fourier series of f is

n 1r
4 + � [ ,;1r sin ( � ) cos ( 3 x) - cos ( ; ) sin ( 2n; )] .

00
4 1r 2 1r
n 4 1r x
n=l
:
11
7. f(x) = � for O < x ::; 1 and J(x + 1) = J(x) for all x. f is not piecewise
continuous because lim J(x) = oo. The discontinuity at O is not a finite jump.
x->0+
0, -1 < X <0 }
8. f(x) = { and f(x + 2) = f(x) for all x.
-x, 0�x<1
(a) T = 2, L = 1, a n
=
f01 -x cos(n1rx) dx = - :
1
sin(n1rx)I�+ n � 0 sin(n1rx)dx f
-½,
1r
[ ��-:,;) l, n 2'. l, ao f0 -xdx = _x;I� =

n
= -n2� 2 cos(n1rx)I� = l
= 1
f 1
bn = 0 -x sin(mrx)dx = nx1r cos(mrx)I� - n� 0 cos(mrx) dx = ( ��f 1 n
. The
Fourier series off is -¼+ fi [
1 ��;) n n
cos(mrx) + (�� sin(mrx)].
(b) The graph of f is displayed in Figure C.1.
-2 -I 0 2 3 4 5
-I
F igure C. l: The graph of f.
(c) At x = 0,the series converges to f(0) = 0.

At x =½,the series converges to f (½) = -½.
At x = 1,the series converges to ½[f(l+) +J(l-)] = -½
At x = 18, the series converges to f(18) = f(0) = 0.
At x = 18.5, the series converges tof(18.5) = f (½) = -½
At x = 19, the series converges to ½ [f(19+) + f(19-)] =
½[f(l+) +f(l -)] = -½.
9. f(x) =x2 for 0 � x < 2 and f(x + 2) = f(x) for all x.
(a) T = 2, L = 1, an= f0 x2 cos(mrx)dx = �1r x2 sin(mrx)I� -

2
2
n1r 0f
2 2
f
x sin(mrx)dx = n 2�2 x cos(mrx)I� - n 221r2 0 cos(mrx)dx = n2�2 ,
n?: 1,ao = f0 x2dx = ; =
2 3
1� i,
bn = f0 x2 sin(mrx)dx =
2
- n�x cos(mrx)I� + n� f0 xcos(mrx)dx = - n� + n 221r2x sin(mrx)I� -

2 2
2 2
--f-,x f sin(n1rx)dx = _--1..
n 1r Jc0 ntr
+ n}1r3 cos(n1rx)l 0 = _ --1.. > 1. The Fourier
n1r , n -
I:
00
series off is;+ [n 2 �2 cos(n1rx)- n� sin(n1rx)].
n=l
(b) The graph off is displayed in F igure C.2.
(c) At x = 3,the series converges to f(3) =f(l) =1.
At x = 8,as at x = 0, the series converges to ½ f
[ (0+) + f(0-)] = 2.
At x = -10, as at x = 0,the series converges to ½ J[ (0+) + f(0-)] = 2.
10. f(x) = x3 for -1 :S .r, < 1 and J(x + 2) = J(x) for all x. T = 2, L = 1, and f
is odd. Hence,its Fourier series is a sine series,with bn = 2 f0 x3 sin(n1rx) dx =
1
- ;1rx3 cos(n1rx) I�+ n� f0 x2 cos(n1rx) dx = - 2 (��) + n 261r2 x2 sin(n1rx)
1 n
I� -
f x sin(n1rx)dx = - 2(- f cos(n1rx)dx = - (-l)"
1 n 1 1 2
n 21r2 Jo l) +_Q_
n 3 7r3 x cos(n1rx) 0 - _Q_1
n 3 1r3 Jo n1r
_Q_ n1r
+ 12nC3-V7'.
n
� [ 1 2n(;n1
The Fourier series off is the sine series 0 .
t- 2 l) n ]
(-
nn
sin(mrx).
n=l
y
4
-4 -2 0 2 4 6 X
Figure C.2: The graph of f in Solution 9.
-1 -2 < X < 0
11. f(x) = { } and J(x + 4) = J(x) for all x.
1: 0�x<2
( a) The graph off is displayed in Figure C.3.
y
I
-4 -3 -2 -I 0 2 3 4 5 6 X
--------, - I
Figure C.3: The graph off.
(b) T = 4, L = 2 and f is odd. Hence, its Fourier series is a sine series, with
n
r 2 sin (n1rx) dx = _.1_ cos (n1rx) 1 = .1..[1--
bn = Jo ( l) ] ' n -
> l.2
f
2 mr 2 O mr
The Fourier series of f is the sine series 2[1 -�� l ) n)

sin ( n;x).
n=l
12. J(x) = x2 for -1 � x < l and J(x + 2) = J(x) for all x. T = 2, L = 1 and J is
even. Hence, its Fourier series is a cosine series, with an= 2 x 2 cos(mrx) dx = Jd
2 2
n4 1r x sin(mrx) I� -
1
;, ] x sin(mrx) dx = n2� 2 x cos(mrx) n2�2
1 cos(mrx)
dx = I� - f0
l
n 0
�;�(, n 2: 1, ao = 2 0 x dx =
1 2
f
The Fourier series of J is the cosine series
l
3 +I: n2
oo 4(-l)n
1r
2
(
cosmrx).
n=l
1-x2 0SXS 1
13 . f(x) = { _ _ 2 < < } and f(x+ 2) = f(x) for all x.
1 (x 2) , 1 x 2
(a) The graph off is displayed in F igure C.4.
-4 -3 -2 -1 0 2 3 4 5 6 X
F igure C.4: The graph off.
(b) T = 2, L = l and f is even . Hence, its Fourier series is a cosine

series, with an = 2f0 ( l-x2 )cos(n1rx)dx = }rr(l-x2 )sin(mrx)[�+
1
1
n� fo X sin( mrx) dx = - n2�2x cos(mrx)I�+ n2�2 fo
cos(mrx) dx = - 4�;�r,
1
n 2'. 1, ao = 2f0 1-x2 dx = 2x-�x3[� =

1
The Fourier series off is the !-
cosine series�+ I:
OO
4( )n
�;1r 2- cos(mrx). The series converges to J(x) for all
1
n=l
x since f is continuous everywhere and J' is piecewise continuous.
L L = 72 -
OO 4( )n 1 OO ( )n 1 2
(c) x = 0 ⇒ 1 = f(O) = � + ⇒
n=l n=l
�;1r2- -�2 -
14. (a) The area under y = J(x) for -1 S x S 2 is the same as the area under
y = f(x) for 5 S x S 8.
(b) The area under y = J( x) for -1 S x S O is the same as the area under
y = J(x) for 2 s s
x 3, and the interval O S x S 2 is common to both
t
integrals: J� 1 f(x) dx = 1 J(x) dx+ 0 f(x) dx =
2
J
f
3
J
2 f(x) dx + 0 J(x) dx = f0 J(x) dx.
2 3
15. For any real number a, employ the T-periodicity of f and make the change
of variable y = x + T to obtain J: J(x) dx = J; f(x) dx + J0°+T J(x) dx =
+r
J:+ J(y-T) dy+ J0° T J(x) dx = J:+ J(y) dy+ J;' T f(x) dx = J:+r J(x) dx+
+ +
T T
J0°+T J(x) dx = ft J(x) dx. It follows that J0o T J(x) dx = fo J(x) dx =
+ r
J: T f(x) dx for any real a and {3.

+
{ - cos( x), - x O
16. f(x) = o7f <s < } and f(x + 21r) = J(x) for all x. T = 21r,
COS( X), _ X < 1f
L = 1r and f is odd since J(-x) = -f(x). Hence, the Fourier series off is a sine
series, with bn = ¾ J;
cos(x) sin(nx) dx = ¼ f0 sin[(l + n)x] - sin[(l-n)x] dx 1r
l. { cos[( l -n)x] _ cos[( l +n)x]

7r (-l)"-1-1 -1- __1_) 2n[(- l ) n -l_ l ].
= 1r 1-n l +n
} 1 = 1r
( 1-n l+n = 1r (l-n 2 )
O
The Fourier series of f is the sine series f:
n=l
2n n 1
l
l�( �� 2 - ] sin(nx).
n )

00
1. The Fourier series off on [0, 6] has the form � + I: [ ancos ( n;x ) + bnsin ( n;x )] .
n= l
2. The Fourier series of f on[-3, 5] has the form

00
a°
2 + I: [ anCOS ( n;x ) + bnsin ( n;x )] .
n=l
3. The Fourier series of f(x)= { �: � �: � � } is
�+I: [ancos(mrx) + bnsin(mrx)], oa = f01 1dx + f12 2dx = 3,

00
n= l
an= f0 cos(mrx)dx + J/ 2cos(mrx)dx= 0, n � 1,
1
bn= f0 sin(mrx)dx + Ji2 2sin(mrx)dx= -;1r cos(mrx)j� - n': cos(mrx)j� =

1
;1r [l-(-lt]- ;1r[l-(-lt] = n�[(-lt-1], n � 1.
The Fourier series is � + f:

l(-�;-11 sin(mrx).
n=l
.
4. The Founer senes of f(x)=
. {2 ' 0<x<1r
- - } is
0
1r<x_< 27r
00 '
a2o + I: [ ancos(nx) + bnsin(nx)], ao = ¾ f0 J(x)dx= ¾ J01r 2 dx= 2,
2 1r
ft
n=l
1r
an = ¼ J(x) cos(nx)dx = ¾ J0 2cos(nx) dx = n� sin(nx)I: = 0, n � 1,
2 1r
J J
bn = ; 0 J(x) sin(nx)dx= ¾ 0 2 sin(nx) dx = - n': cos(nx) I: =
1r
2
[
n7r
l (-1ri, n � 1. The Fourier series is 1 +
- 2 1- l)n
[ �� ] sin(nx). f:
n= l
00
5. The Fourier series of f(x)= x on [2,4] is a2o + I: [ancos(n1rx) + bnsin(n1rx)],
n= l
ao = f2 xdx= ; I�= 6, f2 xcos(n1rx)dx=
4 x 4
an=
�xsin(n1rx)I� - ;1r f2 sin(n1rx)dx = n2�2 cos(n1rx)j� = 0, n � 1,
4
n
f4
bn = 2 xsin(n1rx) dx= -n\xcos(n1rx) I� + n1 2 cos(n1rx) dx= 1r
f4
n1r + �
4
- ..1... n 1r sin(n1rx) 1 2 = ..1...
- , n-
n1r
> 1. The Fourier series is 3 - L...J
� ..1...
n1r
sin(n1rx).
n=l
6. The Fourier series of f(x) = x3 on [1, 3] is

00
�+I:[an cos(n1rx) +bnsin(n1rx)], a0 = Ji3x3 dx= xJ� = 20,
n=l
a n= Ji3x3 cos(mrx)dx = n�x3 sin(mrx)I:- }1r Ji3x2 sin(mrx)dx =

3
n2,r2 X cos( n1rx )1l - n2 ,r2 Jl X cos (n1rx) dx = �
3 6 24(-l) n
3 2 - n36,r3 X sin
• ( n1rx )13 +
l
6 3 . ( ) _ n- 2 4(-l) n
6
47 cos( n1rx )131 - _ -- 4 (-lr >
n1r J1 sm n1rx dx- -
2
33 2 1r -- -, n 1,
-
2 n1r 2
n 1r 2
bn = J3 x3 sm(n1rx) dx = -;;;:x3 cos(n1rx) 131 + n1r 3 J3 x2 cos(n1rx)dx = 2 6(-l) -1
n
· 1
+
1: - 1: -
1 1
n2 1r2 Ji3 x sin( n1rx)dx =

n1r
x2 sin( n1rx) 6 2 6( n
-;.,� -i + n3:3 x cos( n1rx)
n}1r2
3
n li1 cos (n1rx ) x =
.
n > 1. The Founer senes 1s
6 2 6(-l) n l 1 2(-l) n . .
00
d +-
33 n1r n-3 -,
J; {
1r 3 1r -
2 4( l) n
10 + cos( n1rx) + [ 2 6(-;.,J -l + 1 2�;1rl)3 ] sin( n1rx) .
n n 1
n;1r2
}
7. The sine series of f on [O, 6] has the form

00
I: b sin ( n;x ).
n
n=l
. 0 '.S X < 1 .
• senes
8 . The sme of f· ( x ) = { } s � bn sm
X
, .
( -2- ) ,
n1rx
O, 1 :S x '.S 2
1
n�
bn = r l X sin ( n1rx ) dx = - _1_ X cos ( n1rx ) I 1 + _1_ r cos ( n1rx ) dx
l
Jo 2 mr 2
O
n1r Jo 2
= - _1_ ) + -
n21r-2 sin ( 2 ) 0 = _1_
4 - n1r cos ( n1r2 ) + -4 >
1
1 ( n1r )
cos ( n1r n21r-2 sin 2 ' n- 1 .
00
n1rx
n1r 2
The sine series is I: [ n2�2 sin ( n21r ) - n21r cos ( �1r )] sin ( n;x ).
n=l
9. The sme. sen. es of f ( x ) = { X, 0 < X < 1 } 1s. ,t;

� b sm
n
.
( -2- ) ,
n1rx
2 _ x, 1 ;; x '.S 2 1
bn =
2
f ) dx = f xsin ( n;x ) dx + Ji2(2 - x) sin ( n;x ) dx
1
=
I: I: -
0 f(x) sin (
n x
;
0
2
- n1r X COS ( n x
; )
1
f
+ n� 0 COS ( n;x ) dx- :1r (2 - X) COS ( n;x )
2
n1r
Ji2 cos ( ; ) dx
n x
=- 2 cos ( nr)
n1r
+ n2�2 sin ( n x
; ) 1: + ;
1r
cos ( n21r ) -
2
4 · n1rx
Sln ( -2-)
4 · n1r = 4 · n1r _ 8 · n 1r
2 1r2 Sln ( 2 ) + n2 1r2 Sln ( 2) - n2 1r2 Sln ( 2) , n 2: 1.
1
n2 1r2 l n
00
The sine series is I: n281r2 sin ( n;) sin ( n;x ).
n=l
10. f(x) = x2 for O :S x '.S 2.
(a)
00 = Jro2 x2 sin ( ) dx = _ _1_7rx2 cos ( 1r )
2
+
u bn sin ( 1r2 ) , bn
n x n1rx n x
.._..... 2 2 1O
1: -
n=l
n
;1r f0 xcos (
2 n x
; ) dx = -:1r cos(n1r) + }1r2 xsin n x
( ; )
8 2
n2 1r2 f0 sin ( ; ) dx = -:1r (-1r + n½:3 cos ( ; ) =
n x n x
1:
n + 16 cos(n1r)- __!_§_ = _§_(-l)n-l + _J&_ [(-l)n-1] , n?. 1.
- _§_(-l)
n1r n3 ,r3
00 {
n3 ,r3 n1r n3 ,r3
The sine series is L :1r (-1r-1 + n��3 [(-lr- 1]} sin ( n;x ).
n=l
(b) The graph of the odd, 4-periodic extension f of f to � is displayed in

Figure C.5.
y
4
-6 -4 -2 0 2 4 6 8 10 X
-4
Figure C.5: The odd, 4-periodic extension f off to R
(c) The series converges to f(x) on [O, 2). At x = 2, the series converges to
0 =/= f(2).
(d) At x � 3, the series converges to ](3) = f(-1) = -f(l) = -f(l) = -1
since f is 4-periodic and odd.
At x = 8, the series converges to ](8) = ](O) = f(O) = 0 since f is 4-
periodic.
At x = -10, by 4-periodicity, the series converges to
½ [l(-10+) + i(-10-)] = ½ [!(2+) + ! (2-)] = ½(-4 + 4) = o.
L an cos ( n�x ).
00
11. The cosine series of J on [O, 6] has the form�+
n=l
12. The cosine series of J(x) = { �: ½ �; � 1} is�+ f1 a n cos(mrx),
ao = 2 J�½ 1 dx + 2 JI 2dx = 3, a
2
n = 2 Ji cos(mrx)dx + 2 JI 2 cos(mrx)d
2
x =
-;1r sin (�1r ), n � 1.

l
n� sin(mrx); + n� sin(mrx)l =
11 1
2
L n1r sin ( n1r) cos(mrx).

00
2
The cosine series is � - 2
n=l
L an cos(mrx),
00
3
13. The cosine series of J(x) = x on [O, 1] is�+
n=l
J;
ao = 2 J0 x 3 dx = ½, an = 2 x3 cos(mrx)dx =
1
I� -
1
:1r ]0 x2 sin(mrx) dx = n261r2 x2 cos(mrx) I� -
�;�r -
2 3
sin(mrx)
I�
n x
J;
1r
n;;3 X sin( mrx) + n 3 fo sin(mrx) dx =

6 1
n�;2 X cos(mrx) dx = ;;
6(-lr 12 cos(n1rx) 11 _
_ 7""4" 6(-l) n 12 [1- (-1)n], n -
+ nn >
_ 2_2 - nn
nn - - 2--
nn 7""4" l.
fi { �;-�r +
O 2
The cosine series is ¾ + 6 12[1��--:; 1)n]}

cos(n1r x).
14. The cosine series of J(x) = sin(x) on [0,3] is T +La n cos (n;x ),
n=l
ao = i f03 sin(x)dx = -icos(x)I� = i [1- cos(3)],

i f0 sin(x) cos ( ; ) dx = ½ ]� { sin [(1+ 31r) x] + sin [(1 - =
3 3
an t) x]} dx
n x n n
=
3
3
_ .! cos[(l+ n3")x] cos[(1- nt)x]
1 {cos(3+n1r) -l cos(3-n1r)-l }-
1
3
{ l+- n,r + 1--n,r }-
- - -��-3+ n 1r
+-��3- n 1r
3 3
0 0
_ {(-l)n cos(3)-l
+ (-l) 3-cos(1r3)-l} = [l _ (-l)n cos(3)] (- 1- + __)
n
1 =
3+ n 1r n 3+ 1r 3- 1r n n
B[l-(-l)n cos(3)l n > l The cos

9-n21r2 , - . ine series is
� 6[1-(-l)
.3! [1- cos(3)] + L..,
n
cos(3)] COS (n1rx).
9-n2 1r2 3
n=l
15. J(x) = x2 on [-2,2].
(a) The graph of the 4-periodic extension J off to JR is displayed in Figure

C.6.
y
4
-6 -4 -2 0 2 4 6 8 JO X
Figure C.6: The 4-periodic extension f of f to R
(b) The Fourier series of f converges to (x) for all x because J J is continuous
everywhere and ]' is piecewise continuous.
- 00
(c) Since f is even, the Fourier series off is the cosine series T+La n cos ( n ;x ),
n=l
2
a0 = r
Jo
2
x dx
2 = x 1
3 2
3 O
= §3' an = r x cos
Jo
2 2
(
n 1rx
2
) dx = ..2. x2 sin ( n21rx ) 1 -
n 1r
O
16(-l)n '
2
__i_ r x sin
n 1r Jo
z
(n ) dx
1rx
2
= __ s
n27r2 x cos (
n 1rx
2
) _ r2 cos
1 0 - _s
n27r2 Jo (
n 1rx
2
)dx = n27r2
n 2 1. The cosine series is f + ;� I: (
n=l
-,,T cos ( . n
; )
x
16· f(x) = { X, 0 :s;

1, 1 :s;
X
X
<
:s; 2
1 }
.
(a) The graph of the odd, 4-periodic extension f of f to IR. is displayed in

Figure C.7.
y
I
-6 -5 -4 -3 -2 -I 0 2 3 4 5 6 X
-1
Figure C. 7: The graph of the odd, 4-periodic extension J of J to R
00
(b) The sine series off is � b n sin ( ; ), bn
n x
= f02 J(x) sin ( n x
; ) dx =
n=l
1
f0 X sin ( ; ) dx + Ji2 sin ( ; ) dx = - �! cos ( n; )
n x n x
4-
x
-
I: + ;1r
2 cos(mr) + 2 cos (n1r)
1
f0 cos ( n; ) dx - x
+-
2 1
2 cos n1rx
( 2 )
1 = _..1_ cos n1r
( 2 ) n 1r
sin ( 2 )
n1rx
1
2
1
( -1 r, n � 1.
2 2
n 1r n1r
0
n 1r n 1r
= n2�2 sin ( n;r) - ,;1r The sine series is
fi [ �
n2 2
sin ( nt) - 2
(��) ]
n
sin ( n;x ). The series converges to f (x) on [O, 2).
(c) At x = 0, the series converges to ](O) = f(0) = 0.

At x = 1, the series converges to 1(1) = f(l) = 1.
At x = 2, the series converges to½ [](2+) + !(2-)] =½(-1 + 1) = 0.
At x = 78, employing the 4-periodicity of J, the series converges to
½ [1(78+) + J(78-)] = ½ [n2+) + J(2-)] = o.
At x = 79, since j is 4-periodic and odd, the series converges to
f(79) = ](-1) = -](1) = - f(l) = -1.
(d) The graph of the even, 4-periodic extension f of f to IR. is displayed in
Figure C.8.
00
(e) The cosine series of J is � + � an cos ( n;x ), ao =
n=l
f (x) dx = f02
f01 x dx + f12 1 dx = ½ + 1 = an = f02 f ( x) cos ( n;x dx = f01 x cos ( n;x )2 dx
!, )
+ J12 cos (n1r2 x ) dx = n2x sin ( n1r2 x ) 1 - ..1_ + ..1_
1
rl
JO sin ( 2 ) dx
n1rx
sin (n1r2 x ) 1 =
n 1r n
0 1r 1r 1
2 sin
n 1r
(n1r
2 )
+-
n 1r
4-
cos (
2 2
n1rx
2 )
1
1
- 2 sin
n1r
n1r
( 2 )
= -24-
n 1r2
[cos 1r
( 2 )
n - 1] >
n>
-
1.
0
L ) - 1] cos ( n;
00
The cosine series is ¾ + �
n2 2
[cos ( n21r x
), and converges to
n=l
f(x) for all x in [0, 2].
-6 -5 -4 -3 -2 -1 0 2 3 4 5 6 X
Figure C.8: The graph of the even, 4-periodic extension J off to IR.
(f) At x = 0, the series converges to f(0) = f(O) = 0.

At x= l, the series converges to ](1) = f(l) = 1.
At x= 2, the seri�s converges to ](2) = f(2) = 1.
At x = 78, since f is 4-perioclic, the series converges to
! (78) = !(2)= f (2)= 1.
At x = 79, since J is 4-periodic and even, the series converges to
](79)= ](-1)= !(1)= f(l) = 1.
1 x
17.f(x)={ ' �:S; <� ·
0' 2::Sx:S;1r }
Lb !
00
(a) The sine series of J is n sin(nx), bn = J01r: J(x) sin(nx) dx=
n=l
! f0� sin(nx) dx = -;1r: cos(nx)/: = ;1r: [1 - cos ( nr)], n 2". 1.
L :1r:
00
The sine series is [1 - cos ( nr)] sin(nx), and converges to f(x) on
t
n=l
(0, 1r] except at
La
00
(b) The cosine series off is af + n cos(nx), a o = ¾ J01r:f(x) dx=
n=l
¾ fo� 1 dx= l, an = ¾ J01r: f(x) cos(nx) dx= ¾ fo� cos(nx) dx = :1r: sin(nx) 1:
; 1r), n 2". 1. L �)
00
= 1r sin ( n2 The cosine series is ½ + � sin ( 1r cos(nx), and
n
n=l
converges to f(x) on [O, 1r] except at �.
L [a
00
(c) The Fourier series off is�+ n cos(2nx) + bn sin(2nx)],
n=l
ao = ½ J0 J(x) dx =!Ji 1 dx = l, an ½ J0 f(x) cos(2nx) dx=
1r
= 1r
t J1r: J(x) sin(2nx) dx

2 2
¾ f0� cos(2nx) dx = n� sin(2nx)1: = 0, n 2". 1, bn = 0

= ¾ f0� sin(2nx)dx = - � cos(2nx) 1: = � [1 - cos(mr)] = ; [1 - (-It],
E i-��W sin(2nx), and converges to J(x)

n n 1r
n 2:'. 1. The Fourier series is½+

n=l
on (0, 1r) except at �-
=
(d) The series may be simplified to ½ +I: ( 2k!i)1r sin[2(2k + l)x]. Then
k=O
_
X - 4 ⇒ 1 _
1r
-J r
4 - 21
(7 ) _ L
+ k=O
(X) 2(-l)k
(2k+l)1r L (-l)
⇒ k=O
(X)
2k+l
k
_
-
4.
1r

x 3,
1. J(x) = { �,
6- x,
(a) The graph of the 3-periodic extension J off to IR is displayed in Figure

C.9.
-3 -2 -I 0 2 3 4 5 6 X
Figure C.9: The graph of the 3-periodic extension f off to R
(b) Since f is even and 3-periodic, the Fourier series of f on [3, 6] is the same ·
as the cosine series of g on [O, 3], where g(x) = { � �: : � } . �:

2 :S X :S 3 3 - X,
Thus, the Fourier series off on [3, 6] is the cosine series a;+ L a cos ( n; ),
(X)
x
n
1� +
n=l
with ao = i f0 g(x) dx = i ]0 xdx + i Ji2 dx + i ]2 (3 - x) dx = ½x2
3 1 3
i f0 g(x) cos ( ; ) dx =
3
ix[�+ (2x - ½x ) I�=!, a
2 =
n x
J:
n
2 n x
)
i fol X cos ( ; dx + i f1 COS ( ; ) dx + i (3 - X) cos ( ; ) dx =
n x n x
..2.:r;
n 1r
sin (
n 1rx
3 ) 1
1
O
- ..2.
1m
J/ sin
O
n
( 1r
3 )
x
dx+ ..2.
mr
sin ( 3 ) 1
n1rx
2
1
+..2.1r (3-x) sin
n
( 3 ) 1
n1rx
3
2
+
..21r. f2 sin ( n1r3 ) dx + n-6-

1
= ..2.1r sin ( cos ( ..2. sin ( 2n31r)

3
1 + n1r
x n 1r n 1rx
-
n n 3 ) 1r
2 2
3 )
0
..2.1r sin
n
n1r
( 3 ) - ..2. sin (
n 1r
2n1r - -6-
3 ) n 1r
2 2
cos (
n1rx
3
) 1
3
=- 6-
n 1r 2 2 cos (
n
3 )
1r -- 6- -
n2 1r2
2
6
cos(mr) + n267r2 cos (2�71" ) = n267r2 [cos
n2 7r2 e;n
- 1 - cos(mr) + cos (2�71" )]
c�n
= n2 1r2 {[1 + (-lt] [cos (�1r ) -1] }, n 2:: 1, employing
6
cos + cos (2�71" ) = cos ( n37r) + cos ( - 2�71" ) = cos (7�71" ) + cos ( �71" - n1r) =
cos ( n31r) + cos ( n31r ) cos(n1r). Since a2k+l = 0 for k 2:: 0, the cosine series,
l
00
with n = 2k, k 2:: 1, is�+ ;2 L 2 [cos (2;1r ) - 1] cos ( 2k;x ). Note that,
k=l
since f has period 3 (as well as period 6), if we had taken the cosine series
L Cn cos (2� t the exact same result would have
~ 00
off as � + 7rX
), with L =
n=l
been obtained.
(c) The series converges to ](x) for all x since j is continuous everywhere and
]' is piecewise continuous.
2. (a) F(x) = f(x) at x = 0 and at x = L if J(0) = f(L).
(b) S(x) = f(x) at x = 0 if J(0) = 0; S(x) = f(x) at x = L if f(L) = 0.
(c) C(x) = f(x) at x = 0 and at x = L regardless of the values f(0) and f(L).
3. (a) fo sin ( nzx) sin ( m;x ) dx = ½ foL { cos [ ( n -7)1rx ]

L
- cos [ (n+7)7rX] } dx = 0 if
n # m, for any m, n 2:: 1.
n n
(b) foL COS ( nlx ) COS ( m;x ) dx = ½ J/ { COS 1rx
[( -7) ] + COS [( +7)7rX]} dx 0
if n # m, for any m, n 2:: 0.
I: bn sin ( nzx), then, for any integer m 2: 1,
00
(c) If f(x) =
n =l
00
fo f(x) sin ( mzx ) dx = Lbn fo sin ( nlx ) sin ( m;x ) dx =
L L
n=l
bm fo sin2 ( m;x ) dx = ½bm =} bm = fo f(x) sin ( mzx ) dx.
L L
f
00
(d) If J(x) =
+Lan cos ( nlx ), then, for any integer m 2:: 0,
ao
2
n =l
fo f(x) cos ( m;x ) dx = � foL cos ( mzx ) dx +

L
oo
L
Lan fo cos ( nlx ) cos ( m;x ) dx = ½am =}
n =l
L
i
am = fo f(x) cos ( mzx ) dx.
4. Sincef and g are continuous andf' and g' are piecewise continuous, the Fourier
series off and g converge to J(x) and g(x), respectively, for all x, by Theorem
8.1. Since the two series are equal, f(x) = g(x) for all x. It follows, that
t t
an = J.!:L J(x)cos ( nlx ) dx = J_!:L g(x)cos ( nlx ) dx = Cn , n 2:: 0, and
L ( n1rx ) dx = .!. JL g(x) sin ( n1rx ) dx = d n > 1
L J-L f(x) sin
bn = .!. L L -L · L n, - .
I: bn sin ( n;x ),
00
5. (a) The fundamental period off is 6. The sine series off is
n=l
with bn = � f0 3
[2 sin ( 1r3x ) - 3 sin ( 4 x
; ) + 4 sin x
(5; )] sin ( ; ) dx, n 2:: 1.
n x
By Exercise 3(a), bn = 0 if n =J 1,4,5, b 1 = � 0 2 sin 2 dx = 2, f3 (7•n

�
b4 = 3 Jo
3
1rx) � 3
r -3 sin ( 3 dx = -3, and b5 = 3 Jor 4 sin ( 3 dx = 4 · Thus,
2 4 2 5
1rx)
the sine series of f is f ( x) itself. This is due to the fact that f ( x) has the
form of a finite sine series, namely, EC n sin ( n;x ), with c1 = 2, C4 = -3,
n=l
EC
00
Cs = 4, and Cn = 0 otherwise. By Exercise 4, f ( x) = n sin (n;x ) =
n=l
Lb
00
n sin ( n;x ) ⇒ bn = Cn for all n � l, and, hence, f ( x) is its own sine
n=l
senes.
(b) The fundamental period off is 8. As in part (a), the cosine series of J is
f(x).
(c) The fundamental period of J is 4. As in part (a), the Fourier series of f is
J(x).

1. (a) f(x) = x!2 is analytic at all points except x = -2.
(b) f ( x) = �� is analytic at all points except x = ±2.
(c) f(x) = tan(x) is analytic at all points except x = (2n + lH.
(d) J ( x) = e�::.\ is analytic at all points except x = 0.
(e) J(x) = Jx + 2, x � -2, is analytic at all points x > -2.
2. Division of 1 + x into 1 - x + x2 gives J(x) = 1 �::x =

2
E 3(-ltx
00
1 - 2x + 3x 2 - 3x3 + 3x4 - · · · = 1 - 2x + n
.
n=2
Ea
00
3. (a) nx
n
= a2 x2 + a3x3 + a4x4 + asx5 + · · ·.
n=2
La L=2 a La
00 00 00
(b) nx
n
= m+2 X
m+2
m+2 X
m+2
=
n=2 m+2 m=O
a2 + a3 + a4x4
x 2
x 3
+ asx5 + · · ·.
00 00
(C) La
1n=O
m+ 2 x
m +2
= La
n=O
n+2 x
n
+2 = a2x2 + a3 x3 + a4 X4 + a5x5 + · · · .
I: an xn = I:
00 00
= I: an+2 xn+2 = a2 x2 +a3 x3 +a4 x4 +a5x5 +· · ·.
00
(d) a n+2 xn+2
n=2 n+ 2=2 n=O
L na x Lax+ =0
00
L na x
00
Lax
00 00
4. n
n
+ n
n 2
=} 0 + a1x + n
n
+ n
n+ 2
= 0 =}
n=0 n=0 n=2 n=0
L [(n + 2)an+2 + a ] x + = 0 ⇒ a1 = 0 and (n+2)a + +a = 0, n � 0,

00
a1x+ n 2
n n 2 n
n =0
⇒ a1 = 0 and an+2 = �:2, n � 0. Then a1 = 0 a3 = 0 a5 = 0, etc., i.e., ⇒ ⇒
a2 k+ 1 = 0 for all k -> 0. Also, n = 0 a2 = .::!:!.u
a
2 '
n = 2 ⇒
a4 = �4 = .QcQ.
2-4'
⇒
n - 4 __,__ - =:Q1. -
- -,, a6 - 6 - 2.4.6 - 23. 1.2.3, and, m
o - ( l
3
(-
-ao l ) a
. genera1 , a2k - 2'ck!a0, k � 0, and
- l
ao is arbitrary.
5. y" + xy' + 2y = 0.
Lax L na x L n(n- l)a x

00 00 00
(a) Y = n
n
, y' = n
n
-l, y" = n
n
-2, y"+xy'+2y =0
n=0 n=0 =0 n
L n(n - l)a x - L na x + L 2a x = 0
00 00 00
=} n
n 2
+ n
n
n
n
=}
n=0 n=0 n=0
L [(n + 2)(n + l)an+2 +na

00
= 0 =} n + 2an] xn
n=0
(n + 2)(n + l)an+2 + (n + 2)an = 0 ⇒ an+2 = �:1, n � 0.
- 0 ⇒ a2 -
(b) n - - 2
- .=Q:o. ' n - ⇒ a4 -- � - .Qcll..
- 1-3' n - - 4 ⇒ a6 - - =:Q1. - -ao
I 3 5 - 1-3·5'
.
n -
- 6 __,__ -,, a8 - genera1 , a2 k -
( l) k o
- .=Qfi.
7
- ___Qo_ , et c., and, m
- 1_3_5_7 - 1_3_-5___ (2ak-l)
(-l) k ·2·4·6···(2k)ao _ (-1) k 2 k k!ao k >
(2k)! - (2k)! ' - 0·
n - 1 ⇒ a3 - - -, n - ⇒ a5 _
_ _ -a1 _ 3
2 - � - _ a1. , n _- 5 ⇒ a7 _
4 -
24
� _ - -a1 6 2-4·6'
· genera1, a2k+l _ (-lla1 > _ �

_ (-1) a1 k
k
_ 0 . y - 6 anx -
n _
and, m - 2_4_6 ___(2k) - �,
n=0
2 k+l _
� 2k �
u a2kx + u a2k+lx - a0 6
� (-1) k 2 k k!x2k
(2k)! + al 6 �
� (-l) k 2 k+l
x =
k=0 k=0 k=0 k=0
a0 y 1 + a 1 y2 is the general solution, and y 1 and Y2 are two independent
solutions.
(c) y1 = 1 - x2 + .!x
3
4
- 1...!_x
5
6
+ · · · ' y2 = x - .!x
2
3
+ lx
8
5
- ...!_x
48
7
+ · · ·.
k
(-l) k 2k +l - � ..!_
= xe-\ .
2 2
( d ) Y2 - �
6 �x - X 6 k! (-x2 )
k=0 k=0
(e) Since the equation has no singular points, R = oo for both series.
(f) = 0 ⇒ a0 = 0 and y'(O) = 2 ⇒ a1 = 2. = 2y2 = 2xe-2.
x2
y(O) Thus, y
6. (1 + x2)y" - 4xy' + 6y = 0.
Lax L na x -i, y" = L n(n- l)a x

00 00 00
(a) y = n
n
, y' = n
n
n
n
-2,
n=0 n=0 n =0
(1 + x )y" - 4xy' + 6y = 0 ⇒
2
(1 + x 2 ) L n(n - l)anxn - 2 - 4x L na x -l + 6 E a x
00 00 00
n
n
n
n
=0 =}
n=0 n=0 n=0
L n(n - l)a L n(n - l)a L 4na L 6a

00 00 00 00
n 2
nx - + nx
n
- nx
n
+ nx
n
= 0 =}
n=O n =O n =O n=O
L
00
[(n + 2)(n + l)an+2 + n(n - l)an - 4nan + 6an] xn = 0 =}
n=O
L [(n + 2)(n + l)an+2 + (n2 - 5n + 6)a

00
n ]x
n
= 0 =}
n =O
(n + 2)(n + l)an+ 2 + = 0 ⇒ an+2 = -�:��)�:�3{t, n 2:: 0.

(n - 2)(n - 3)an
( b) n = 0 ⇒ a2 = -3a0 , n = 2 ⇒ a4 = 0, n = 4 ⇒ a6 = 0, etc. Hence,
a2 k = 0 fork 2:: 2. n = 1 ⇒ a3 = -½a1, n = 3 ⇒ a5 = 0, n = 5 ⇒ a7 = 0,
etc. Hence, a2k+1 = 0 for k 2:: 2. Thus, y = a0 + a1x + a2x2 + a3x3 =
a0 +a1x-3a0x2 -½a1x3 = a0(1-3x2)+a1 (x - ½x3 ) is the general solution,
with a0 and a1 arbitrary, and y1 = 1-3x and y2 = x-½x3 are two linearly 2
independent solutions.
(c) y(0) = -3 ⇒ ao = -3 and y'(0) = 1 ⇒ a 1 = 1. Hence,
y = -3(1 - 3x2) + (x - ½x3) = -3 + x + 9x2 - ½x3 .
Eax E na E n(n - l)a

00 00 00
7. y" + xy' - 2y = 0. y = n
n
, y' = nx
n
-l, y" = nx
n 2
- '
n=O n=O n=O
L n(n - l)a + L nan xn - L 2an xn = 0

00 00
y" + xy' - 2y = 0::::} nx
n 2
- =}
n=O n=O n =O
L n(n - l)a L (n - 2)a

00 00
nx
n 2
- + nx
n
= 0 =} (n + 2)(n + l)an+ 2 + (n - 2)an =0
n =2 n =O
⇒ an+2 = - ( n� ��a.;2 ), n 2:: 0. n = 0 ⇒ a2 = ¥.'# = ao, n = 2 ⇒ a4 = 0, n = 4
⇒ a6 = 0, etc. ence, a2 k = 0 for k 2:: 2. n = 1 ⇒ a3 = ;.� = �t, n = 3 ⇒
�
_ 5 ___,_ 3a _ _ 7 ⇒ a _ -5a7 _ -3-5a1 _ g
-�
a5 _ 4- 5 - 5! , n -
_ .=Qcl
----,,,- a7 _
- -6-7s - 3a1
7! , n - 9 - 8-9 - 9! , n -
___,_ _- 7ag _3-5-7a1 · genera1 , a2k+l _
----,,,- au - 10_11 - (1 l)! , etc., and , m -
(-l)k+l.1. 3.5.7...(2 k-3) a1 _ (-l)k+L.1-3 -5-7---(2k- 3) a 1 2-4-6---(2k-2) _ (-l)k+1(2k-2)! a 1
(2k+l)! - (2k+l)! 2-4-6···(2k-2) - (2k+l)!2 k -1(k-l)!'
k 2:: 1. The general solution is y = a0 y 1 + a 1 y2, where y1 = 1 + x2 and
(-l)k+1(2k-2)!
_
Y2 - x +
oo
L
(2k+l)!2k-1(k-l)!x
2k+l . ·
are two lmearly mdependent solutions. ·
k=l
La L=O na
00 00
8. (2 + x2 )y" + 2xy' - 2y = 0. y = n
n x , y' = nx
n
-l,
n=O n
L n(n - l)a
00
y" = nx - , (2 + x2 )y" + 2xy' - 2y
n 2
= 0::::}
n=O
L 2n(n - l)a L n(n - l)a + L 2nan Xn - L 2anXn = 0 ⇒

00 00 00 00
nx
n 2
- + nX
n
n=O n=O n=O n =O
L 2n(n - l)an x L (n
00 00
n 2
- + 2
+ n - 2)an x n
= 0 =}
n =O n=O
2(n + 2)(n + l)an+2 + (n + 2)(n - l)an = 0 ::::} an+2 = -;(:)i)n, n 2:: 0.
n = 1 ⇒ a3 = 0 ⇒ a5 = 0, etc., and a2k+l = 0 for k 2:: 1.
- 0 ___,_
----,,,- a2 - - 4 ⇒ a6 -
n - - 2 ___,_ - -=!!c2. - -2.5
Qll - .=Q11 3a1
- 2. , n - ----,,,- a4 - 2_3 - 22. 3 , n -
9. y" + exy = 0.
(a) Y = ao + a1x + a2 x 2 + a3 x3 + a4 X4 + a5 x5 + · · ·, y' = a1 + 2a2 x + 3a3 x 2 +
4a4 x3 + 5a5x4 + · · ·, y" = 2a2 + 6a3 x + 12a4x 2 + 2Oa5x 3 + • • •,
ex = l + x + l2 x 2 + l6 x3 +. _!._
24
x4 + -
12 0 x +
1 5 ...
e y = ao + (ao + a1)x + (½ao + a1 + a2 ) x + ( ¾ao + ½a 1 + a2 + a3 ) x 3 + • • • ,
x 2
y" + e xy = 0 =} (2a2 + ao) + (6a3 + ao + a1)x + (12a4 + ½ao + a1 + a2 ) x 2 +

( 2Oa5 + ¾ao + ½a1 + a2 + a3 ) x3 + · · · = 0 =} a2 = =?, a3 = -ao5-ai,
a4 = -;_;1 , a5 = � - �b :::::} y = ao + a1x - �x 2 - ¾(a0 + a 1 )x3 - n-x4 +
(.'!Q _
a1 5 ...
4 0 60) X +
(b) ao = 1 and a1 = 0 ⇒ Y1 = 1 - ½x 2 - 16 x3 + _!_x 40 + · · · , a0 = 0 and a 1 = 1
5
⇒ Y2 = x - -x 1
6
3
- -x
1
12
4
- -x1
60
5 + ...
10. Legendre's equation of order v, (1 - x2 )y" - 2xy' + v(v + l)y = 0.

La x L na x -l, y" = L n(n - l)a x -2:::::}
00 00 00
(a) Y = n
n
, y' = n
n
n
n
n=0 n=0 =0 n
L n(n- l)a x - L n(n- l)a x L=0 2na x L v(v + l)a x = 0

00 00 00 00
n
n 2
- n
n
- n
n
+ n
n
n =0 n =0 n n=0
I: n(n - l)anx - + L [v(v + 1) - n(n + 1)] anx = 0
00 00
n 2 n
=}
n=0 n=0
:::::} L {(n + 2)(n + l)an+2 + [v(v + 1) - n(n + 1)] an } xn = 0

00
n =0
---'- _ n(n+l)-v(v+l) an , n _
----r an+2 - > O.
(n+l)(n+2 )
(b) If v = m � 0 is an integer, then the recursion relation becomes
an+2 _
n(n+l)-v(v+l) _ n(n+l)-m(m+l)
- (n+l)(n+2) an - (n+l)(n+2 ) an , n > - 0'
and n = m ⇒ am+2 = 0, n = m + 2 ⇒ am+4 = 0, n = m + 4 ⇒ am+6 = 0,
etc., which shows that one of the series solutions truncates at amxm . If m
is even, then y 1 = ao + a2x2 + · · · + amxm is a polynomial, and if m is odd,
then Y2 = a1x + a3 x3 + · · · + amxm is a polynomial.
(c) If m = 0, then P0(x) = a0, and P0 (1) = 1 :::::} a0 = l. Hence, P0(x) = l.
If m = 1, then Pi(x) = a1x, and P1(1) = 1 ⇒ a1 = l. Hence, P1(x) = x.
If m = 2, then P2 (x) = a0 + a 2x 2, with a 2 given by the recursion relation
with n = 0 : a2 = -;_�;}ao = -3ao :::::} A(x) = ao - 3aox2 = ao (1 - 3x 2).
Then A (1) = 1 ⇒ -2a0 = 1 :::::} a0 = -½ :::::}
P2 (x) = -½
+ !x2 = ½ (3x2 - 1).
If m = 3, then P3(x) = a1x + a3 x3 , with a3 given by the recursion relation
�t
with n = l: a3 = 1· 22 4 a1 - -ia1 :::::} P3 (x) = a1 (x - ix3 ). Then
A(l) = 1 :::::} a1 = -! :::::}
A(x) = ½ (5x3 - 3x).

1. f(x) = x 2 - x + 2 on [O,2] and f(x + 2) = f(x) for all x. f(x) =
2
°+ I::[an cos(mrx)+bn sin(mrx)],ao = f0 (x 2 -x+2)dx = (; - ; + 2x) /
2 3 2
a2
n=l O
14,a = f 2(x 2 - x + 2) cos(mrx) dx = ,;'If (x 2 - x + 2) s i n(mrx)I� -
n 0
1 f 2 (2x-l) sin(mrx) dx = 2�2 (2x-l) cos(mrx)l�- l lf2 f 2 cos(mrx)dx = 2�2 ,
3
n7f 0 n n ' 0 n
n 2'. 1, bn = f0\x 2 - x + 2) sin(mrx) dx = - n� (x 2 - x + 2) cos (mrx) I� +
n� J0 (2x-1) cos(mrx) dx = - n1f + n2� 2 (2x-1) sin(mrx)I�-)7f2 J0 sin(mrx) dx
2 2 2
=
-;'If + n}'lf 3 cos(mrx)I� = -;'If , n 2'. 1. y" + 3y = J(x) =}

�+ L [cn cos(mrx)+dnsin(mrx)],co = � = \4,Cn = _�11 7f2 =
00
n2 7r2 ( � n2 7r2 )'

y =
n=l
3 3
3- n27f2 = n7r(-2- n2 7r2).

dn = ___Qn__
3
2. f(x) = x - 2x on [-1,3] and f(x + 4) = f(x) for all x. J(x) =
2
�+ � [an cos(n;x)+ bn sin(n;x)],ao= ½f_3 1 (x 2 - 2x)dx= (x6 _x ) 3

00
3 2
, 1 2
1 -l
�3' an = .!.2 f-1 (x 2 - 2x) cos(n'lf2 x)dx = ...L(x - 2x) sin( " )
-1 -
3 3 1
n,r
2 n x
2
4 n - 4
...L f (2x - 2) sin( "x) dx - ,r2 - l) cos( "2 ) -1 - n-
n2-(x 2 ,r2 f-1 cos (�) dx =
3 3
n,r -1
n x 3
= 1
2 2
_ (n ") >
n2 2 cos 2 ' n - 1 , b n = .!.2 f-1(x - 2x) sin
_!_§ (n'lf2 x)dx =
3 2
-...L(x +...L 4 ( n "2 x) 1 3-1

-2x) cos (n"x) n,r J-1 (2x-2) cos( 2" )dx = n2,r2 (x-1) sin
,r
3
n,r
2 3
n x
1
4 f3 sin(n "x)dx = -lB sin(n") , n > 1. y" + 4y = f(x) =}

2 -1
- n-2 - 2
,r -1 2 n2 2 2
,r -
�+ L [Cn ; ) + dn s in ( ; )] , Co = 1 = ¼,
00
y = n x n x
COS (
n=l
a 64cos(!!f) bn -6si 4 n(!!f)
Cn = _4 n2n rr2 1 n- 2,r 2),d n = 4_ n2rr2 = n2 ,r2 (16- n2 2 ) ·
= n2 ,r2(6 ,r
4 4
3. f(x) = x - x 2 on [O,1], odd, 2-per i odic. f(x) = L bn sin(mrx),

00
n=l
1
bn = 2 JO (x - x 2) s in(mrx) dx = -; (x - x 2) cos(mrx)I� +
;" JO1 (1 - 2x) cos (mrx) dx = 1r2 (1" - 2x) sin(mrx) I� + n2�2 J; sin(mrx) dx
i
1 4(1-(-l)nj , n _
- n3,r4 3 COS( n1rx) 1 O - 3 . ( n1rX),
00
- 3
n ,r 3 > 1 . y /I
- y = .
J ( X ) L..,
=} y = ,;;:-" dn Sln
n=l
4. f(x) = { �' �
2 x, � � �; }, odd, 4-per i odic. f(x) = 1 bn s in ( ; ),
n x
f,
bn = n2�2 sin ( nr) by Sect ion 8.2, Exercise 9. y" + 3y = f(x) ⇒
� . n,rx b 2 in(¥)
s
Y = L.., d n sin ( -2- ), d n = - ,� rr2 = n2,r2(1 2 - n2 ,r2 ) ·
3
n=l 4 3
00
5. J(x) = 1 - x on [0, 1], even, 2-periodic. f(x) = � + Lan cos(mrx), a0 = 1
:I�J l l , n 2 1, by Example 8.20 in Section 8.2.

= n n= l
and an 2 [1
y" + y = f(x) =>-
Y= 3 + L Cn cos(mrx), Co = ao = 1, Cn = a17r2
1 -n
=
n;l;(
(- 1
1 -n(t )
.
n=l
1r 1r
X, 0 � X < 1
6. f( x) = { l } , even, 4_pen.od"1c. J( x) _
- QcO. +�
�
an cos (n1r2 x ) ,
, 1�x�2 2
ao = �' an = n2�2 [cos ( nr) - 1], n 2 1, by Exercise 16(e) in Section 8.2.

00
Y
11
+ 4y = f(x) =>- Y = T + L CnCOS ( ; n x)
, Co = �o = �' Cn = 4_:'2"2
n= l 4
[cos 1r
(� ) - 1] = n2 1r2(l��n21r2) [cos 1r
(� ) - 1].
n2 1r2(4��)
7. f(x) = 2x - x2 on [O, 2] and f(x + 2) = f(x) for all x. Since f is even,

J(x) = �+ f 1
an cos(mrx), ao = 2 f0 (2x - x2 ) dx
2
= 2 ( x2 - x;) i: i,
an = 2 f0\2x - x2 ) cos(mrx) dx = n1r2 (2x - x2 ) sin(mrx) I� -
2
f
n1r 0 (2- 2x) sin(mrx) dx = n2�2 (1- x) cos(mrx) I�+ n2�2 0 cos(mrx) dx =
2
00
2
f n;-!2,
n 2 1. y" + 2y = J(x) =>- y = T + L Cn cos(mrx), Co = �=!,

n=l
00
8. f(x) = sin(2nx) on [0, 1], even, 2-periodic. J(x) = � + Lan cos(nnx),
n=l
ao = f1
2 0 sin(2nx) = 0. If n =/= 2, an = f1
2 0 sin(2nx) cos(nnx) dx =
1
Jfo { sm
l
• [( 2 _ n ) nx] }dx -
• [( 2 + n ) nx] + sm
__ { cos[(2 +n)1rx]
( 2 +n)1r + cos[( 2-n)1rx]
( 2 -n)1r } 1
0
+ = If n = 2 Jro sin(2nx) cos(2nx) dx
n
[ 1 -(-l) ] 4 -(-l) n
n
[ 1 -(-l) ] 1
(2 +n) 1r ( 2 -n)7r
[l
1r(4-n2 )
]. = 2' a2
f01 sin(4nx) dx = 0. y" = f(x) =>- y = T + n=I: Cn cos(nnx), 0 ·Co = ao = 0 =>- co

00
l
· _ a2 _ _
r2
_ -4 (1-(-l) ] n _j_
n
1s arb"t
1 rary, C2 - _4 1r2 - 0 , Cn - 2
an
2 - n 3( 4-n2) ,
-n 1r 2 1r
Index
Absolute convergence, 234 Complex eigenvalues, 135

Adjoint Conditional convergence, 234
classical, 340 Connected region, 30
Alternating series, 229 Continuous
approximations, 231 function of two variables, 26
remainder, 231 Contrapositive, 214, 233
test, 230 Convergence
Alternating series test interval of, 247
proof, 343 of a sequence, 193
Analytic function, 278 of a series, 208
Annihilator, 112 ·radius of, 247
Approximations Converse, 215
of alternating series, 231 Cramer's rule, 68
of series, 219
Associated homogeneous equation, 57, 103 Decreasing sequence, 200
Associated homogeneous system, 171 Dependent variable, 3
Average value of a function, 288 Diagonalizable matrix, 187
Difference equation, 244
Bernoulli equation, 21 Direction field, 5
Binomial Divergence
series, 265 of a sequence, 193
theorem, 263, 265 of a series, 208
Binomial theorem Domain, 24
proof, 346
Bounded sequence, 201 Eigenvalue, 126
complex, 135
Cauchy-Euler equation, 52, 98 Eigenvector, 126
Chain rule, 27 complex generalized, 164
special case of, 27 generalized, 147, 158
Change of index, 256, 319 Elementary function, 321
Characteristic polynomial, 146 Envelope, 40
Classical adjoint, 340 Equation
Cofactor, 340 associated homogeneous, 57, 103
Comparison test, 223 Bernoulli, 21
proof, 341 Cauchy-Euler, 52, 98
Complex conjugate, 50 difference, 244
445
446 INDEX
Euler, 52, 98 average value, 288

exact, 29 elementary, 321
homogeneous, 15, 44, 87 even, 292
indicial, 48, 52, 88 extension, 299
Legendre's, 328 odd, 292
linear, 18, 44, 87 of two variables, 24
nonhomogeneous, 44, 87 periodic, 283
nonlinear, 18, 44, 87 periodic extension, 300
of order n, 3 piecewise continuous, 288
ordinary point of, 317 potential, 29
separable, 9 singular point of, 278
with constant coefficients, 48 singularity of, 278
Euler equation, 52, 98 vector, 122
Euler's identity, 50 vector-valued, 122
Euler-Fourier formulas, 286, 315 Fundamental
Even matrix, 124
extension, 307 period, 283
function, 292 set, 124
Exact equation, 29
General solution, 4, 43, 87, 124
Existence and uniqueness theorem
General term
Picard's, 41
of a sequence, 191
Existence theorem
Generalized eigenvector, 147, 158
Peano's, 41
complex, 164
Extension, 299
Geometric series, 210
even, 307
even, periodic, 307 Homogeneous
odd, 304 equation, 15, 44, 87
odd, periodic, 304 solution, 57
periodic, 300 system, 121
F ibonacci sequence, 193 Identities

F inite series, 213 trigonometric, 285
Fourier Imaginary part, 51
convergence theorem, 289 Implicit solution, 17
cosine coefficients, 286 Increasing sequence, 200
cosine series, 295, 307 Independent variable, 3
series, 286 Independent variables, 24
series on [a, b], 300 Indicial equation, 48, 52, 88
sine coefficients, 286 Indicial polynomial, 114
sine series, 294, 305 Induction
Frobenius method, 326 principle of, 203
Function Inductive hypothesis, 203
analytic, 278 Inductively-defined sequence, 193
INDEX 447
Infinite series, 208 system, 121

Initial condition, 5, 43, 87, 125, 177 term, 57
Initial-value problem, 5, 43, 87, 125, 177 Nonlinear equation, 18, 44, 87
'- Integral test, 217
Integrating factor, 19, 33 Odd
Integration extension, 304
by partial fractions, 339 function, 292
by parts, 335 One-parameter family
by substitution, 336 of circles, 11
techniques, 335 of curves, 11
Interval of convergence, 247 of ellipses, 14
Inverse of a matrix, 340 of lines, 11
Irred�1cible quadratic, 339 of parabolas, 13
of solutions, 4
Left-hand limit, 288 Open region, 26
Legendre polynomials, 328 Operator
Legendre's equation, 328 linear, 91, 112
Limit comparison test, 227 Ordered set, 191
proof, 342 Ordinary differential equation, 3
Limit of a sequence, 193 Ordinary point, 317
Linear Orthogonal, 12
equation, 18, 44, 87 family, 314
operator, 91, 112 functions, 314
system, 121 trajectory, 12
Linearly dependent
functions, 44, 45 p-series, 218
vector-valued functions, 123 Parameter, 4, 11
Linearly independent Partial derivative
functions, 44, 45 second-order, 25
vector-valued functions, 123 with respect to x, 25
Lower bound, 201 with respect to y, 25
Partial differential equation, 3
Maclaurin series, 259
Partial fraction decomposition, 339
Matrix inversion, 340
Partial fractions, 339
Method of undetermined coefficients, 57
Partial sum of a series, 208
Minor, 340
Particular solution, 4, 43, 87, 124
Monotone sequence, 200
Period, 283
Multiplicity, 91
fundamental, 283
n choose k, 263 Periodic
nth -term test, 213 extension, 300
n-parameter family of solutions, 87 function, 283
Nonhomogeneous solution, 328
equation, 44, 87 Piecewise continuous function, 288
448 INDEX
Potential function, 29 inductively defined, 193

Power series, 245 initial value of n, 191
coefficients, 245 limit of, 193
interval of convergence, 247 lower bound, 201
radius of convergence, 247 monotone, 200
uniqueness, 259 unbounded, 201
Principle of induction, 203 upper bound, 201
Series, 208
Radius of convergence, 247 absolute convergence, 234
Ratio test, 236 alternating, 229
proof, 343
--
approximations, 219, 231
Real part, 51 binomial, 265
Recurrence relation conditional convergence, 234
coefficient, 320 convergence, 208
Recursion relation cosine, 295, 307
coefficient, 320 divergence, 208
Reduction of order, 47 finite, 213
Region Fourier, 286
connected, 30 geometric, 210
open, 26 infinite, 208
simply connected, 30 Maclaurin, 259
Relation, 24 p-, 218
Remainder partial sum, 208
formula, 276 power, 245
of a series, 219 remainder, 219
of a Taylor series, 269 sine, 294, 305
of an alternating series, 231 sum, 208
Resonance, 330 Taylor, 259
Right-hand limit, 288 telescoping, 212
Root test, 239 trigonometric, 283, 286
proof, 345 Sigma notation, 203
Simply connected region, 30
Separable equation, 9 Singular point
Sequence, 191 of a function, 278
bounded, 201 of an equation, 52, 317
bounded above, 201 Singular solution, 40
bounded below, 201 Singularity
convergence, 193 of a function, 278
decreasing, 200 of an equation, 52, 317
divergence, 193 Solution, 3
Fibonacci, 193 general, 4, 43, 87, 124
general term, 191 homogeneous, 57
increasing, 200 implicit, 17
INDEX 449
particular, 4, 43, 87, 124

periodic, 328
singular, 40
Solution curve, 5
Squeeze theorem, 199
Standard form, 18, 44, 103
Sum of a series, 208
System of equations, 121
decoupled, 130
homogeneous, 121
in matrix form, 122
linear, 121
nonhomogencous, 121
with constant coefficients, 121
Taylor
polynomial, 269
remainder formula, 276
series, 259
Telescoping series, 212
Transpose, 340
Triangle inequality, 282
Trigonometric
identities, 335
identity, 285, 426, 427
integrals, 336
series, 283, 286
substitution, 338
Two-parameter family of solutions, 43
Unbounded sequence, 201

Undetermined coefficients
method of, 57
Upper bound, 201
Variation of constants, 68
Variation of parameters, 68, 104
for systems, 172
Wronskian, 103, 124

'

Sam Melkonian, MATH1005 - Differential Equations and Infinite Series

Uploaded by

Copyright:

Available Formats

Sam Melkonian, MATH1005 - Differential Equations and Infinite Series

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sam Melkonian, MATH1005 - Differential Equations and Infinite Series

Uploaded by

Copyright:

Available Formats

Ordinary Differential Equations

COPYRIGHT@ 2016 by No part of this work covered ISBN-13: 978-0-17-671614-1

Appendix A contains formulas and techniques from trigonometry, calculus and

I Ordinary Differential Equations 1

4 Higher-Order Linear Equations 87

5 Linear Systems 121

II Infinite Series 189

6.2.4 Absolute and Conditional Convergence 233

7 Taylor Series 245

8 Fourier Series 283

9 Series Solutions of Differential Equations 317

A Formulas and Techniques 335

B Proofs of Selec ted Theorems 341

Ordinary Differential Equations

1.1 Basic Concepts

Example 1. 1. y' + y = x2 + ex is a first-order ordinary differential equation.

Example 1.2. y111 = 2y2 + y' + 1 is a third-order ordinary differential equation.

Definition 1.2. A solution of a differential equation is any differentiable function f

Example 1.3. The function f (x) = e 2x is a solution of the equation y' - 2y = 0

where c is an arbitrary constant of integration. Thus, f(x) = --

= - x , J(x) = - x + 1 and f(x) = -2 - 2

Figure 1.1: Particular solutions of y' + x = 0 and directions.

In Example 1.4, y + c is the general solution, and

Definition 1.4. The set of directions defined by a first-order differential equation

Definition 1.5. An initial-value problem for a first-order equation consists of a first­

If we attempt to solve the equation as in Example 1.4, we obtain

y' = ky * ty' =k * J ty' dt = J kdt.

By the substitution rule,

Evaluating both integrals, we obtain

where c1 and c2 are arbitrary constants of integration and the combination c2 - c1

where c = ±ec3 is an arbitrary constant since c3 is arbitrary. At t = 0, we obtain

3. Find the general solution of the equation !� = 2x cos(x2 ).

13. Solve the initial-value problem 3y2 y' = 1, y(O) = 2.

2.1 Separable Equations

g(y) :� = f(x) ⇒ J g(y) :� dx = J J(x) dx ⇒ J g(y) dy = J J(x) dx,

Example 2.3. Consider the initial-value problem

=> y2 = ln { �[ln(x)] 2 + c} => y = ± In { �[ln(x)] 2 + c}

is the general solution, and

y(l) = 2 => ±� = 2 => ln(c) = 4 => c = e4 => y = In { �[ln(x)] 2 + e4 }

is the solution of the initial-value problem.

2.1.1 Orthogonal Trajectories

More generally, a relation between x and y which includes a parameter (constant)

Definition 2.2. A one-parameter family of curves is a set of curves defined by an

The relation g(x, y, c) = 0 my take more particular forms, such as y = cf(x),

Definition 2.3. A curve is an orthogonal trajectory of a one-parameter family of

Given a one-parameter family of curves, it is often necessary to determine its

The equation is separable, and

and noting that x = 0 is a solution.

In Example 2.4, the one-parameter family of curves was obtained by solving a

Example 2.5. Find the orthogonal trajectories of the one-parameter family of

Figure 2.2: The parabolas y = x2 + c and their orthogonal trajectories.

Example 2.6. Find the orthogonal trajectories of the one-parameter family of

Figure 2.3: The parabolas y = kx 2 and their orthogonal trajectories.

6. Find the general solution of the equation xy' = (x + l)e Y .

8. Solve the initial-value problem cos(x) ln(y) 1� = ysin(x), Y > 0, y(O) = e 2.

2.2 Homogeneous Equations

For example, the equations

A homogeneous equation is solved by making the substitution (also called a change

u= � ⇒ y = xu ⇒ y' = u + xu' ⇒ u + xu' = y' =f (;-) = f (u)

Definition 1.5. An initial-value problem for a first-order equation consists of a first