Rudolph E. Langer A First Course in Ordinary Differential Equations

Download as pdf or txt
Download as pdf or txt
You are on page 1of 265
At a glance
Powered by AI
The document discusses a textbook on ordinary differential equations including its preface, contents, and index.

The book is a textbook on ordinary differential equations intended for students with a year of calculus experience.

The preface discusses the importance of differential equations, the intended audience and level of the book, and the author's approach and goals in writing the book.

m

A First Course in

ORDINARY
DIFFERENTIAL
EQUATIONS

A
First
Course
in

ORDINARY
DIFFERENTIAL
EQUATIONS

Rudolph E. Longer, Ph.D.


PROFESSOR OF MATHEMATICS
UNIVERSITY OF WISCONSIN

NEW

YORK

LONDON

JOHN

WILEY

CHAPMAN

&

& S O N S , INC

HALL,

LIMITED

Copyright, 1954, By John Wiley & Sons, Inc.


All Rights Reserved

This book or any part thereof must not


be reproduced in any form without the
written permission of the publisher.
Library of Congress Catalog Card Number: 54-5326
Printed In the United States of America

PREFACE

In presenting this textbook I have indulged a long-felt desire to teach on


a wider range a subject that has been my main interest over many years.
Differential equations are so important i n their applications that no scientist and no engineer can afford to neglect them. A t the same time they are
the carriers of so many fundamental, far-reaching, and ingenious ideas
that mathematical analysis can well be regarded as centered around them.
Present-day research upon them is alive and intensive, and is being spurred
continuously by the urgent needs of the most up-to-date developments i n
science and technology.
This is a first course i n the subject, written for the student with a year's
experience i n the calculus. Its manipulative material is adjusted to that
level, and whatever goes beyond it is explained i n detail. I n connection
with each new development, examples of a l l the various cases that may
arise are given and are worked out at every step. These will serve as
guides of method, and also as models of economical procedure.
The
reader may miss a chapter on review material. Such material, in considerable amount, is nevertheless there, generally in the explanatory
examples. It has always seemed to me preferable on pedagogical grounds
to insinuate the reviews unobtrusively where the motivations for them
exist, rather than to emphasize his shortcomings to the student by fencing
off the m a i n subject by a discouraging barrier of material that may be
relevant here or there i n the book.
I n the matter of method, I have consistently played down the need for
memorization and played up the common sense and reasonable approach
w h i c h experience leads one to recognize. N o tabulations of formulas or
lists of rules are therefore to be found i n the book. Although I have not
concealed any essential difficulties (my intention is that the book should be
wholly honest) I have tried consistently to disburden the presentation of
unnecessary difficulties. T o that end I have avoided such things as
untimely appeals to existence theorems when explicit solutions are available, and where proofs of substantial length have been included I have
v

vi

Preface

stripped them of generalities not germane to the present context to reduce


them to their most elementary and perspicuous terms. I have kept close
to the subject. Although the applications to physics, chemistry, engineering, etc., have been stressed, this is still a book on differential equations,
not a potpourri of textbooks on these other subjects.
I believe that all the material of the book may be taken up profitably,
and in the order i n which it is here presented, where time permits. However, continuity w i l l not be sacrificed even by many and substantial
omissions. A shorter course may well be based upon Chapters 1, 2, and 3;
parts of Chapters 4 and 5; and Chapters 8 and 9. I have assumed that the
student will have a table of integrals at hand.
Although I have broached the significance of the differential equation,
the arbitrary constant in its solution, and such matters, even in the introductory Chapter 1, it is my belief that some substantial first-hand experience with the equations alone supplies a basis upon which more extended
discussions of these matters is easily comprehensible and truly meaningful.
I have accordingly, i n Chapters 2 and 3, hastened into the manipulative
work of finding explicit solutions and applying them. T h i s w i l l engage
the student almost at once i n operations which are already familiar to h i m
from the calculus, and will give h i m the opportunity to establish confidence
i n his ability to do something definite. A t the same time it w i l l make the
essential characteristics of the differential equation concretely familiar to
hira.
In Chapter 2, I have presented the explicit methods that are applicable
to differential equations of the first order in a definite relationship to the
two fundamental notions of the integrating factor and the change of variables. This gives organization to material that has all too often been
presented as hardly more than a collection of unrelated tricks. I n Chapter
4 the geometrical significance of the differential equation, as it is displayed
by the integral curves, is stressed. Chapter 5 centers upon the procedures
of approximation to be resorted to when explicit solutions cannot be found.
I have included at this point a proof of the existence and uniqueness of a
solution under appropriate conditions, in the belief that this reasoning w i l l
be manageable and challenging to the more able student. For others it
may be omitted. Chapters 8 and 9, on the differential equation of the
second order with constant coefficients, will be of prime importance to
students of physics and engineering. I n Chapter 1 1 , 1 have set forth a
more complete treatment of solutions by power series than is usual in books
of this kind.
It has been my steady purpose throughout the book to present the differential equation dually as a mathematical concept and as a technological
tool. Consistent with this I have tried to convey an understanding of

Preface
what such an equation is and involves, and no less to give practice i n the
commoner techniques that are available for solving it, or otherwise
extracting useful information from it.
RUDOLPH

Madison, Wisconsin
January, 1954

E .

LANGER

CONTENTS

Chapter

Page

1. I N T R O D U C T I O N
1.1.
1.2.
1.3.
1.4.
1.5.

W h a t is a differential equation?
Integrals or solutions of a differential equation
T h e direction field of a differential equation
Constants of integration
Checking a general solution. T h e differential equation of
a family of curves

2. S O M E T Y P E S O F S O L V A B L E D I F F E R E N T I A L E Q U A TIONS OF T H E FIRST ORDER


2.1.
2.2.
2.3.
2.4.
2.5.
2.6.
2.7.
2.8.
2.9.
2.10.
2.11.

Equations with separable variables


Exact differential equations
Integrating factors
Finding integrating factors
T h e linear differential equation of the first order
Changes of variables
Equations with homogeneous coefficients
Equations with linear coefficients
Simultaneous equations
T h e linear differential system with constant coefficients
T h e replacement of a single differential equation by a differential system

3. A P P L I C A T I O N S O F D I F F E R E N T I A L E Q U A T I O N S O F
T H E FIRST O R D E R
3.1.
3.2.
3.3.
3.4.
3.5.

Problems i n velocities
Frictional motion
Problems i n rates
Problems of
flow
Locus problems. Rectangular coordinates
ix

1
1
2
3
5
5

9
9
11
14
15
17
19
22
24
26
28
29

33
33
37
38
40
44

Contents

X
Chapter

3.6.
3.7.
3.8.
3.9.

Page

Polar coordinates
Some curves of pursuit
Suspension cables
T h e rope around a shaft

4. I N T E G R A L C U R V E S .
4.1.
4.2.
4.3.
4.4.
4.5.
4.6.

Integrals of a differential equation


T h e direction field of a differential system
T h e isoclines of a direction field
T r a j ectories
Singular points of a direction
field
Some elementary cases of the differential equation
'

4.7.

TRAJECTORIES

hx + ky

T h e integral curves as trajectories

5. A P P R O X I M A T E S O L U T I O N S . I N F I N I T E S E R I E S .
EXISTENCE AND UNIQUENESS OF A SOLUTION
5.1.
5.2.
5.3.
5.4.
5.5.
5.6.
5.7.

Polygonal graph approximations


Polygonal approximations for simultaneous equations
Solutions in power series. Polynomial approximations
T h e method of successive approximation
Successive approximations for differential systems
A proof of the existence of a solution
Uniqueness of the solution

6. F U R T H E R S T U D I E S O F D I F F E R E N T I A L E Q U A T I O N S
OF T H E FIRST ORDER. T H E RICCATI E Q U A T I O N
6.1.
6.2.
6.3.
6.4.
6.5.
6.6.
6.7.

Singular solutions
Factorable equations
Singular solutions again
Integrations by differentiation. T h e Clairaut equation
T h e method of difTcrcntiation with respect to j
T h e Riccati equation
Some properties of the integrals of R i c e a l i equations

7. S O M E S O L V A B L E D I F F E R E N T I A L E Q U A T I O N S O F
T H E S E C O N D O R D E R . A P P R O X I M A T I O N S . APPLICATIONS
7.1.

Differential equations of the second order

47
49
54
58
63
63
64
65
68
70

72
73
78
78
81
83
88
91
93
96
98
98
100
101
103
105
107
110

114
114

Contents

xi

Chapter

Page

7.2.
7.3.

115

Equations i n which either or x are not present


T h e differential equation of a 2-parameter family of curves.
Checking a general solution
7.4. Exact equations
7.5. Simultaneous equations
7.6. Solutions i n power series. Polynomial approximations
1.1.
Approximate integral curves
7.8. Some geometrical applications
7.9. Curves of pursuit
7.10. T h e motion of a particle on a curved path

116
118
120
122
123
127
128
130

8. T H E L I N E A R D I F F E R E N T I A L E Q U A T I O N W I T H C O N STANT COEFFICIENTS

134

8.1.
8.2.
8.3.
8.4.
8.5.
8.6.
8.7.
8.8.
8.9.
8.10.

T h e complete equation
T h e complementary function
T h e case of complex roots. Euler's formulas
Differential operators
A n operational method for finding a particular integral
A second operational method
A comparison of formulas
T h e method of undetermined coefficients
T h e method continued
Difierential equations of higher order than the second

9. A P P L I C A T I O N S O F L I N E A R D I F F E R E N T I A L
TIONS OF T H E SECOND ORDER
9.1.
9.2.
9.3.
9.4.
9.5.
9.6.
9.7.
9.8.

10.1.
10.2.
10.3.
10.4.

EQUA156

Simple harmonic motion


Free vibrations
D a m p e d vibration
Forced vibration without damping. Resonance
Forced vibration with damping
T h e forced vibration of a system with a repelling force
M o t i o n under an intermittent force
Simple electrical circuits

10. T H E L I N E A R
EFFICIENTS

EQUATION

WITH

O r d i n a r y and singular points


T h e Euler equation
Exact equations
T h e adjoint differential equation.

134
135
137
140
141
144
146
147
149
152

VARIABLE

156
159
163
166
169
172
173
176

CO180

Integrating factors

180
181
182
183

xii

Contents

Chapter

10.5.
10.6.
10.7.
10.8.
10.9.
10.10.
10.11.
10.12.
10.13.

Page

T h e existence and uniqueness of a solution. T h e R i c c a t i


equation
T h e Wronskian. Linear independence of solutions
T h e general integral
T h e change of dependent variable
Solution of an equation when an integral of the reduced
equation is known
T h e method of variation of parameters
T h e change of independent variable
Simultaneous linear equations
T h e Laplace transformation

11. S O L U T I O N S I N P O W E R S E R I E S
11.1.
11.2.
11.3.
11.4.
11.5.
11.6.
11.7.
11.8.

EQUATIONS

201
203
205
206
208
211
213
215

INVOLVING

12.1. T h e separation of variables in a partial differential equation


12.2. A problem i n the flow of heat
12.3. T h e gamma function
12.4. T h e Legendre equation
12.5. T h e Legendre polynomials
12.6. T h e hypergeometric equation
12.7. Equations that are solvable by hypergeometric functions
12.8. T h e Bessel equation
12.9. Bessel function graphs and inter-relationships
INDEX

190
193
194
196
197
201

T h e reduced equation at an ordinary point


A proof of convergence
T h e complete equation at an ordinary point
T h e ordinary point at . Solutions i n powers of Xjx
T h e reduced equation at a regular singular point
Logarithmic integrals of the reduced equation
T h e complete equation
T h e regular singular point at <

12. S O M E
DIFFERENTIAL
PARAMETERS

185
186
188
189

219
219
222
223
224
228
223
236
238
240
245

CHAPTER

1
Introduction

1.1. What is a differential equation?


The central problem of the integral calculus is this: F r o m a given function f(x)y to find a n integral
and to apply this result to the solution of a problem.

I n terms of symbols

this may be stated as follows: G i v e n the equation


(1.1)

dy/dx=f{x\

to find functions >(x) which fulfill it, and to apply them.


It is but a short step from an equation 1.1 to one of the more general
form
(1.2)

F{x,y,y')

=0,

i n which y' stands for dy/dx. A n y equation of form 1.2 that actually
involves^' is called a n ordinary differential equation of the first order. A n equation that involves the second derivative oiy but no derivative of a higher
order is called a n ordinary differential equation of the second ordery and one that
involves the nth derivative, but none of higher order, is called a differential
equation of the nth order. T h e term " o r d i n a r y " refers to the fact that
the derivatives are ordinary ones (not partial derivatives).
T h e reader will already know that equations of type 1.1 have a wide
range of scientific a n d technological applications. T h e more inclusive
type of equation 1.2 is even much more widely applicable. Indeed, the
differential equation is the primary mathematical instrument for the precise expression of nature's laws, for these laws are always statements of
relationships between variables and their rates of relative change, namely,
their derivatives. F o r that reason the development of the theory of differential equations has been hastened from the time of its very beginning by
the need for solutions of scientific problems. However, this theory also
has great interest for reasons quite apart from its applications. It is a
1

Introduction

branch of mathematical analysis that has occupied the minds of many of


the great mathematical masters, and continues, even i n the present, to be
an active field of mathematical research.

1.2. Integrals or solutions of a differential equation


A n y function that fulfills a differential equation is called an integral^ or a
solution of that equation.
T o solve a differential equation is to make these
functions known. This statement is precise only if the sense of the term
" k n o w n " is made clear. A function is " k n o w n , " of course, if it can be
expressed by a formula i n terms of standard and familiar functions. W e
shall, however, use the term i n a broader sense by admitting a number of
alternatives. I n each of these the values of the function can be found by
calculation, at least to any prescribed degree of accuracy.
(i) If a function f{x) is known, and is integrable on an interval containing the point x = a, the function

is regarded as " k n o w n . " W e shall henceforth call this function a <^uad'


rature of fix), rather than an " i n t e g r a l " of f{x).
This is to preclude confusion, for we have already agreed to use the term " i n t e g r a l " as an equivalent of "solution" for a differential equation. It is not to be required that
the quadrature i n question be one that can be "carried out." Thus, for
instance, the functions

are regarded as " k n o w n , " although they cannot be represented otherwise,


since no proper names or symbols have been assigned to them.
(ii) A function y is to be regarded as " k n o w n " if it is characterized by
an implicit relation
V(x, y) = 0.
Thus the relation

is regarded as making y known as a function of x.


(iii) A function given by a convergent infinite series is regarded as
" k n o w n " if the terms of the series can be determined.
A differential equation 1.2 from which can be explicitly found in terms
of x and y is expressible as
(1.3)

=/(x,>).

The Direction Field

Since;'' = dy/dx, such an equation is often conveniently put into the form
(1.4)

P{x, y) dx ^

Q{x,y) dy = 0.

This is accomplished, of course, by m u h i p l y i n g the equation by Q{xj y) dx,


and then writing P{x,y) i n place of ~(l{xyy)f{xyy).
Equations 1.4 and
1.3 are, however, not necessarily equivalent. Equation 1.4 may, for
instance, be fulfilled by the relation x = c. However, x = c, since it does
not characterize;', has no significance for equation 1.3. Also, equation
1.4 may be fulfilled by the relation Cl{x,y) = 0, without equation 1.3
being so.
T h e choice of letters to designate the variables i n a differential equation
is immaterial. Thus we may have an equation
F{t, s, ds/dt) = 0,

whose integrals are s = f{t) or <p{ty s) = 0. This is, i n fact, usual i n problems of physics and mechanics, since the independent variable is commonly
the time, and time is customarily denoted by t.

1.3. The direction field of a differential equation


T h e graph of a differential equation is quite different i n character from
that of a n ordinary coordinate equation i n analytic geometry. A s we
shall see, it consists not of a single curve, but of a whole family of curves.
Consider any region of the (x^y) plane i n which/(x,;*) is a real, singlevalued, continuous function. A t each point of this region / ( x , >) has a
numerical value. There is a direction i n the plane whose slope has that
vsdue. B y assigning this direction to the point, and doing so for each
point, we impose upon the region a so-called direction field. Clearly every
differential equation 1.3 defines a direction field i n any region i n which its
function f{x, y) has the properties specified above. A direction field may
be represented graphically, though somewhat crudely, by drawing short
line segments i n the prescribed directions through a number of its points.
Such a representation is shown i n F i g . 1, which applies to the differential
equation

T h e function fix^y) is i n this case continuous except on the line x = 2.


T h e slope of this field is at any point twice that of the line j o i n i n g that
point with the point (2, 1).
Direction fields are common i n our physical surroundings. T h e force
of gravity surrounds the earth with such a field, and its magnetism likewise

Introduction

does so. A n y flow of a liquid or gaa defines such a field at each instant,
the direction of the field being that of flow.
I n the instance of any direction field we may conceive of a particle as
starting from any one of its points, and then moving always i n the direction
of the field. T h e path of such a particle is a streamline of the field. E v i y

FIG. 1.

dently every point of a field lies upon a streamline. N o w a streamline is


a graph. It defines > as a function of x. A t any of its points its slope i s ^ ' .
It is, however, also /(x,
namely, the slope of the field. W e see thus
that the function^ defined by a streamline is an integral of the differential
equation to which the field applies. T h e differential equation thus has a

FIG. 2.

whole family of integrals corresponding to the family of streamlines


Figure 2 shows some streamlines for the differential equation
y = W

~ X- +^x

iy.

T h e streamlines are called the integral curves of the differential equation.

Checking a Generell Solution

1.4. Constants of integration


A differential equation of the simple type 1.1 is solved by a straight-out
quadrature. This introduces a constant of integration, namely, a constant which is arbitrary i n the sense that it may be given any value on a
suitable range. There is therefore a family of integrals. It is a 1-parameter family, because just 1 constant identifies its several members. It
appears that any equation 1.3 also has a 1-parameter family of integrals,
for, if we think of its streamlines as cut by some transversal curve, each
point of that curve is determined by a single parameter value, for instance,
by its distance along the curve from some specified point. E a c h point also
identifies the streamline through it. Thus the integral curves are identifiable by the values of 1 parameter. I n the coordinate equation for the
family an arbitrU"y constant therefore appears,
A relation that solves a differential equation 1.2 and that contains a n
arbitrary constant is called the general integral, or the general solution. A n
integral that is obtainable by giving to the constant a specific value is called
a particular integral. T h e general integral may, and often does, include a l l
the integrals of the differential equation. There are, however, cases i n
which it does not do so. A n integral that is not so included is called a
singular integral.
Example 1.

T h e differential equation
y"

+ 2{l

-y]

=0

has the general integral

However, it also has_y = 1 as an integral, y = ^, since it is not included


i n the general integral, is a singular solution.

1.5. Checking a general solution.


a family of curves

The differential equation of

T o check a relation
(1.5)

<p{x,y,c) = 0,

as the general integral of a differential equation 1.2, we proceed as follows:


Differentiate equation 1,5 implicitly with respect to x to obtain the
equation
(1.6)

tPxixyy, c) +

iPy{x,y, c)y' =

0.

I n (1.6),
and tpy denote the partial derivatives of <p with respect to x and
y. T h e liminant of c between equations 1.5 and 1.6 is an equation i n x.

Introduction

yy and y'. If it is the same as the original differential equation or the


equivalent of it, the integral is thereby checked. If equation 1.6 does not
contain f it is itself the liminant.
Example 2.

G i v e n the differcntiad equation


* + ^ | / + > - 2 x = 0.

T o check the relation


xy

\o% y x"^ ~ c

as its general integral.


T h e derivative of the given relation is
x y ' + > + - - 2 x = 0.
y

T h i s is the original differential equation.


checked.
Example 3.

T h e integral has thus been

G i v e n the differential equation

T o check the relation


y = log c tan ' x\

as its general integral.


T h e derivative of the given relation is
y

-1
{1 + x 2 ) { c -

T h e liminant of c is thus v' =


ferential equation.

tan-^j}

This is the original dif-

{1 + x-\'
T h e integral is thus checked.

T h e method described may be used to find the differential equation of


any 1-parameter family of curves. Thus, let the family have the coordinate equation 1.5. B y differentiation we may obtain equation 1.6. T h e
liminant of c is the differential equation of the family.
Example 4.

T o find the differential equation of the family of curves


log> + <:x2 -

1 =0.

Checking a General Solution

T h e derivative of this equation is


-

+ 2cx = 0

T h e eUminant of c is the relation


logjy - J

1 = 0

T h i s is the required differential equation.


ASSIGNMENTS
In each of the following cases, check the relation given as the general integral of the
respective differential equation.
1.

for

yy' + X = 0.

2. y = cx^,

for

xy' -2y

3. y = c cos X + sin x,

for

y' cos x + ^ sin x 1 0 .

+ c tan X,

4. 7 =

5. ^ + log ^

for

f = 0,

6. sin^ ^ + 2 tan x =
7.

y - 2

1 + cx
\x+cU--\

8.

9. y{x + 4| +
10.

= 0.

= 0,

1
2 cos X + c sin X

11. log (xy) = cx.


-1

2 . sin * X + sin * ^ =*

= 0.

\y' ^] sin X {7

for

y' -{-ytr" = 0.

for

y' +2

scc^ y esc i2y) 0.


y - 2

for

y'

for

y' +2xy =

for

} sec x 0.

= 0,
''\

{x + 4 } / +

{x + 5]y - 2e^ = 0.

for

y' -\- -y cot X ^ ' CSC X = 0.

for

xy' +y

= y log {xyj.
1

for

V i

- = sm * X + sm ^ y.

' V i -

PROBLEMS
In each of the following cases, find the differential equation for the given family of
curves.
i.

2,

y cx ^ x'.

i +I l112
4. U + ^ r2 +I [;r
|

3. y 2 ^ c tan x,
5.

+ Ic^ X

33 J

0,

6.
1

8, y.cx

1,
1.

Intrcxiuction
10. -y/y

9. |c + > P + 2:Jt = 0.
11- >

"i

TT Jf + 1

12.
*

-X
-

+ log \cx\ - 1

2 cos

+ c sin X

ANSWERS TO ODD-NUMBERED PROBLEMS

3. / -

\. xy' ~y ~2x*

5.

log X - ^ -

0.

7.

2xy'(/ +

11. 2(x^ -

X+

sm X cos X
'
1 + X -i- V 1

'
9. ^ -

^~^

11-0.

+ {2x -

11/ -

2x + 1.

CHAPTER

2
Some Types of Solvable
Differential Equations
of the First Order
2.1. Equations with separable variables
T h e central problem with which we shall be concerned is that of studying the integrals of given differential equations. W e may attempt to find
these integrals exactly or approximately, and we may do either of these
things analytically by the use of methods of the calculus, or by the use of
graphs, direction fields, etc. As we shall see, only certain types of differential equations can be solved explicitly. I n this chapter some such types
are considered. T h e i r integrals are applicable to many practical problems. T h e y are also typical of the integrals of many differential equations
that cannot be explicitly solved.
A differential equation
(2.1)

P{x, y) dx +

Q{x, y) dy ^

0,

i n which P(x, y) and Q(x, y) are each a product of a function of x by a function of 7, is said to have separable variables. W e may write such an equation
more explicitly as
Pi{x)p2{y) dx +

qi[x)q2{y) dy =

0.

If there is a value yo at which ptiyo) ~ 0, the equation is fulfilled by the


relation y >o since that gives dy 0. T h u s ^ =
is a solution. If
there is a value ;ifo at which q\{xQ) = 0, the equation is also fulfilled by the
relation x = xo. This, however, does not define ^ as a function of x.
For values of x a n d ^ at which ^i(x) diiidpziy) are not zero, we may write
the equation thus:
q^iy)

~
p2{y)

P\{x)

?iW

dy ^

dx.

10

Some Types of Solvable Equations

T h e variables are now separated, since o n l y ^ appears on the left and only
X on the right. E a c h member of the equation is now a differential,
namely, the differentizd of its own quadrature. Since functions with equal
differentials can differ by any constant, but cannot differ by more than a
constant, we conclude that

J p2(y)

qi(x)

T h i s is the general integral.


Example 1.

T o integrate the differential equation


2{y

\ ] dx-\- x^ sinydy

This equation has^ = 1 as an integral.

0.

W i t h separated variables it is

- rfy = - - rfx. y i

x^

/* sin y
2
/ y - \ ^ dy =
-+

Thus
is the general integral.
Example 2.

T o integrate the differential equation


xydx -

(* -

2) rfy =

0.

W i t h separated variables this equation is


dy

x dx
x~2

Thus

log> = 2 log jjv -

2) + X + A

is the general solution, with k as the arbitrary constant. B y replacing each


member of the last equation by its exponential, the relation is given the
form

T h e differential equation also has the integral y = 0.


integrals are included i n the relation
, =

w i t h the constant c arbitrary.

21V,

Clearly all these

Exact Differential Equations

11

PROBLEMS
F i n d the general solution of each of the following differential equations.
i.

{y -2\^dx

3. {x^+

- x^dy ^0.

\]ydy-\-x{y^

5. sec* xsecydx

2.

4. {2xV - r f x + 2xd> -

A\dy ^0.

sin ydy = 0.

6. y/1

+ i 1 + x^} (/y = 0.

7.

y dx-\-dy ^ 0.

~ y^ dx -\-3 -s/\

8. |x -

0.

- x^ dy -

11 cos )- rfy -= 2x sin > dx.

9. X log Xflfy+ -N/I + ^'^rfx= 0.

10. {1 - 2x y/l

- sin^ x \ dx + V 2 - sin^ x dy 0.

11. xe"*dx -\-e~'dy = 0.


13.

y/\

12. ^cos^^rfx -\-&ccydy = 0.

]- y^ dx + xy dy = Q.

= 0.

15. xf.?" + 4} dx
16. x\y -2\dx

14. xy log y dx = sec x dy.

\x -

l]\y -\-A\ dy = {2 ~ y\dx.

17. x<fx + {2 - X + 2>} dy = xy{a> - (fx!.


18. tan ydx-\- cot x rfy = 0 .

2.2. Exact differential equations


T h e diflferential of a function F{x, y) is known from the calculus to be
Fx{x,y) dx + Fy{x,y) dy,

where
and
are the partial derivatives of F with respect to x and ^.
T h i s is of the form of the left-hand member of equation 2.1. I n at least
some cases, therefore, a differential equation 2.1 is actually of the form
dF{x, y) = 0. W h e n this is so, the differential equation is said to be exict.
Because a differential is zero only when the function is constant, the exact
equation implies that
(2.2)

Fix,y)

= c.

T h i s is its general solution.


A differential equation 2.1 need not be exact. Before formula 2.2 can
be used, therefore, we must determine whether the equation is exact or not,
a n d , if it is, how the function F(x, y) is to be found. W e shall show how
that may be done provided the functions Py{x, y) and Qx{x, y) are
continuous.
If the differential equation is of the form dF = 0, it is so because
(2.3)

P{x,y)

= F,{x,y),

Q{x,y)

Fy{x,y).

12

Some Types of Solvable Equations

B y partial differentiations, then,


Pyix.y)

= Fx,y{x,y),

Qxixyy) =

Fj,,j:ix,y),

and, since the cross partial derivatives Fx,y and Fy.^ are equal, we see that
(2.4)

Py{x,y)

Qx(x,>).

This is therefore a necessary relation for exactness. I f it is not fulfilled the


differential equation is inexact.
Suppose now, conversely, that relation 2.4 is fulfilled. T h e n let F{Xfy)
be defined by the formula
(2.5)

F{x,y)

P{x,y)dx-{-R{y),

i n which R{y) is an undetermined function, XQ is any suitable constant, and


the quadrature is made as though^ were a constant. T h e first equation
(2.3) is then fulfilled, and also
=

Fyix^y)

\'Py{x,y)dx-\'R'{y).

But because of (2.4), we may write this last equation


FA>^.y) =

Q.{x,y)dx^-R'{y),

namely,

F^ix.y)

= Q{x,y)

- Q(xo,>) + R'{y).

T h e second equation 2.3 is thus also fulfilled \i R{y) is chosen so as to make


R'ij) = Q(xo,>),that is, if
m

!Qixo,y)dy.

W i t h this evaluation, therefore, the function 2.5 fulfills both equations


2.3, and the general integral 2.2 is given by the relation
(2.6)

/J P{x,y)

dx-\- f

Qixo^y) dy = c.

T h e method could have been applied with the roles of P and Q, and of
X andy
interchanged. W i t h a suitable constant
the general integral is
so found to be
(2.7)

Q{x,y) dy +

P{x,yo) dx = c.

T h e two results 2.6 and 2.7 are equivalent. They may, however, be quite
different i n form, and the quadratures called for by one may be simpler
than those for the other. It is therefore often wise to consider both
formulas, and to carry through with the one that is found to be simpler.

13

Exact Differential Equations


Example 3.

T o find the general integral of the diflferential equation


W

-^y\dx-\-{x-\-^\dy

= 0.

For this equation Py{x,y) = 1, and Qxix^y) 1. Condition 2.4 being


fulfilled, the equation is exact. W i t h XQ 0, formula 2.6 gives the
general solution
f e^dy = c,

j^W^y]dx+

that is,

^xy-\-i^

= c.

W i t h ^ o = 0) formula 2.7 gives the solution


xy-\-e" - \

-hW

=^c.

This differs from the one first obtained only by having {f: + 11 i n place
of c. T h a t , however, is immaterial since c can have any value.
Example 4.

T o find the general integral of the differential equation

6xy +

1
V v * - x^

This equation is exact.


are, respectively.

3x'

6xy +
2x

Vv'

3x^ -

W i t h XQ = 0, and^o =

0
and

dx +

2x

x-

Vy'

dy = 0.

formulas 2.6 and 2.7

dx = c.
x'

6x +

dy +

T h e former is clearly the simpler one.


solution

1
V l

-x^

dx c.

F r o m it we obtain the general

+ s i n " ^ \x/y'\ = c.

'hx-y

PROBLEMS
Test the following differential equations to determine which ones are exact, and find
the general solution for each one that is exact.
19.
20.

21.

{Ixy + 6x! dx
y

x"-

xy

[x"^ rfx +

\ \ dy = 0.

\x

1
2
*
*
3xY - -1,

s dy = 0,

~ X

dy = 0,

14

Some Types of Solvable Equations


1

22,

23,

log > +

dx

xy

dx-V

i - + 2>

- !

dy

dy

= 0,

= 0,

24. \xy + log x} rfjf = (xy + log>l dy.


cos y dy = xe"^ dy

25. ^ sin y i/x +


x dx -\- y dy
26.
~
+
x'

dx.

{Un y -\-x] dx -\-X scc^ y dy = 0.


1

27.

('-:v)'

dy =

28.

{cos

29.

{x + > cos ( x y ) ^ " ' " " U x +


rfx

30.

tan ^ sin x sec ^|

dy

xdy

^ +
X

+ {sin x sec'' y + cos x tan^ y esc y]

dy

=0.

{> + X cos (ry)*-"'f"'>} <y = 0.

y dx

,
X

dx

0,

= 0.

-t y

2.3. Integrating factors


A n exact differential equation 2.1 is usually made inexact if a factor is
removed from it. T h e restoration of the factor, of course, restores the
exactness. W e shall show that i n principle any differential equation 2.1
which has a general solution can be made exact by multiplying it by a
suitable function. Such a function (other than zero) is called an integrating
factor.
Example 5.

T h e differential equation
log ydx-\- dy/y = 0,

is inexact. After multiplication either by e^, or by 1/log^, it is exact.


These multipliers are therefore integrating factors.
Let the general solution of any given differential equation 2.1 be written
i n form 2.2. T h e differential equation of which this is the general solution
is Fi{xyy) dx + Fy{xjy) dy = 0, a n d , since this must be equivalent to
equation 2.1, it follows that
Fxjxy y) ^
P(x.y)

Fyjx.y)
~Q{x,y)'

Let ${Xj y) be the common value of these ratios.


e{x,y)P{x,y)

= F,(x,y),

e{x,y)Q(x,y)

Then
=

Fy{x,y)

Finding Integrating Factors

15

T h i s , however, means that if equation 2.1 is multiplied by 0{xyy) it assumes


the form dF = 0, that is, it becomes exact. Thus 0(x, y) is an integrating
factor.
Suppose equation 2.1 is multiplied by ^{F)d, where 4> is any integrable
function. T h e equation becomes
dF = 0, and, since the left-hand
member of this is also a differential, the equation is again exact. Thus
^{F)d is also a n integrating factor. Since the number of functions i> is
unlimited, an equation 2.1 is thus seen to have an unlimited number of
integrating factors.
Example 6.

T h e differential equation
y dx

dy = 0

is inexact. Its general integral isy/x = c. Thus


jy) is i n this case
y/x^y a n d , since P{xyy) =
we see that 0{xyy) = \/x^. Thus
1/x^, a n d i n fact ^{y/x){\/x~)y with any integrable function 4> is a n
integrating factor. W i t h appropriate choices of <> we see, for instance,
that all the functions

xy

y*-

x'

y^

x^

are integrating factors.

2.4. Finding integrating factors


T h e discussion of Section 2.3 has theoretical interest but little practical
value. T h e integrating factors it displays are expressed i n terms of the
general integral of the differential equation. T o find the integrating
factors the general integral would have to be known. T h e n , however,
there would be no need to concern oneself any longer with integrating
factors.
There is no systematic, generally applicable, method for finding a n
integrating factor. If there were, a l l equations 2.1 could be solved, and
our discussion of this equation could be terminated. T h e finding of an
integrating factor therefore remains a problem. Sometimes one is
detectable by inspection of the equation directed toward recognizing
differentials i n it. However, success in this detection is the exception, not
the rule.
I n the search for an integrating factor it is useful to observe that
(i) A n y function of just one variable, multiplied by the differential of
that variable, is a differential; and
(ii) A sum of differentials is a differential.
W i t h these facts i n m i n d we may try to collect the terms of a differential

16

Some Types of Solvable Equations

equation into groups, each of which is recognizably transformable into a


differential by some suitable multiplier.
Example 7.

T o integrate the differential equation


{xy-\-2 log *} dx +

{x^ + xe^\ dy = 0.

O n grouping the terms i n the manner


x{x dy -{-ydx]

+ 2 log x dxxe"

I
dy -

0,

it is recognizable that the multiplier X/x transforms the first group and also
the last term into a differential. A t the same time it leaves the remaining
term a differential. Thus X/x is a n integrating factor. It reduces the
differential equation to d\xy\ + d {log;cj^ +rf^*= 0. T h e general
integral is thus
xy + ( l o g x l ^ + v = c.
M a n y schemes, such as the following, can be invented to help i n the
search for an integrating factor. Suppose there is a n integrating factor
of the form
(2.8)

e{x,y) =

dx+iHv) dv

T h e differential equation, after multiplication by (2.8), is then exact, and,


by equation 2.4,
-

1P(X, V)J''<'> dx+fb(v) dyj ^ {Q(x, V)J<''rfx+/6(v)dy^

dy

dx

This equation can be reduced to the form


(2.9)

P,{x,y)

QAx^y)

= a(x)Q{x,y) -

b{y)P{x,y).

T h e reasoning is reversible; hence, if a{x) and b{y) can be found to fulfill


equation 2.9, formula 2.8 gives an integrating factor.
Example 8.

T o find a n integrating factor for the differential equation


siny\ dx + cosydy = 0.

For this, equation 2.9 is


cosy = a{x) cosy b{y){e' s i n ^ j .

A comparison of the terms i n cos^ suggests the trial of a(x) = 1. T h e


equation is then fulfilled if b{y) = 0; hence
is a n integrating factor.
Example 9.

T o find an integrating factor for the differential equation


[y + ^ + / i dx + |2x + 4 / 1 dy = 0.

17

The Linear Equation


For this case, equation 2.9 is
-1 + X+

= a(x){2x + V )

-b{y)y{\

+x-\-y'^\.

T h e term 1 on the left must be matched on the right, and this suggests
the trial of b{y)y 1, that is, b{y) = \/y. T o match the term B^"^ on the
left, we may then try a{x) = 1. These trials actually do fulfill the equation. Hence, by equation 2.8, the function e^^^*
which can be written,
ye^, is an integrating factor.
PROBLEMS
For each of the following differential equations, find an integrating factor, and, by
using it, integrate the equation.
31. ydx

dy = 0.

{2x +y\

32.

{2xy2 -y^\dx-\r

33.

\x^

34.

(4*2 ~

35.

[2x^y + > M

xy^'dy =

dy = 0.
0.

xy -h 7x -\- 2y -

dx +

sin y

36.

37.

- sec y tan y

^ ) dx

38.

{2x^ +

I] dx +

xy] dy =

X log X

{x -

6y +

~e

0.

0.

x^ COS y

dy = 0.

dx {x sec y log x] dy ~

dx

\ ] dy

" + JT sin

^ X

0.

dy = 0

2.5. The linear differential equation of the first order


A diff"erential equation which is important both theoretically and practically is that of the form
(2.10)

y'

+p(x)y

q{x).

It is called the linear differential equation of the first crder because it is linear,
that is, of the first degree, i n y and y'. For an equation of this type an
integrating factor is always easily found.
Let the equation be multiplied by ${x) dx. Its resulting form is
e{x)

dy +

{p{x)e{x)y

q{x)e{x)]

dx =

0.

By equation 2.4, the last equation is exact if 6' = pix)6y that is, if
0'/6 = p{x).
This integrates to give log $ = jp{x) dx, which is ^ =
^Jpix) di
T h u s tf/p^** dx is an integrating factor, by virtue of which the

18

Some Types of Solvable Equations

differential equation becomes

T h e general integral is thus


^ J p ( x ) dx ^

J^(x)tf/P<^> ^dx

+ C.

This method is easily remembered. It requires no memorization of


details.
Example 10. T o integrate the differential equation
y' -\- y tan x = cos x.

After multiplication by d{x) dx^ the given equation is exact if 6' =


6 tan Xy that is, it 6 = sec x. Thus sec x dx is an integrating factor. T h e
general integral found by its use is
y sec X =^ X -\- c.
Example 11.

T o integrate the differential equation


{1 +x^]y'-{-

{1 -x]^y

xe-'.

W e begin by writing the given equation i n form 2.10, that is, as


I

2x

+x'

xe

1 -\-x

T h e method then shows that 0{x) dx is an integrating factor if


e' = e

2x

1 -\-x

T h e variables i n the last equation can be separated to give


d0/e =

2x

1 -\-x^

dx

Thus, log 0 = ;t - log 1 + x'^l, that is,


=
+ x^). T h e integrating factor of the differential equation originally given is thus
e'dx/il

4-x2).

B y the use of this factor the differential equation is given the form

Changes of Variables

19

Its general integral is, accordingly.


-1
1

2{1

+xM

PROBLEMS
F i n d the general integral of each of the following differential equations,
39. >' + 2xy =
41. /

43. y'

xy

-x^

e-^\

40. y' + y cot

= 3x.

42. xy' log

3x' + 3x'

+ 1

45. y - y - ^ l

47. y'

2x

49. y'

51. / [ l

46. y'

1.

48. y'

y =

+xM

Ix +4}y'

53.
54.

{x + 5)y =

11

2x|y

{2x

1
| - cot
X

2x

sm

cos x csc^ x.

le'.

+ 1 y _ 2xy = |x* + 2x2 + 1} eos


y'

1 -

52.

2x2.

50. y' sin X + y sec x =

2
- + 4x

+y

+ y -

1.

44, y' -\- xy = x.

+logx!.

X =

X.

\ \e-^'.

2.6. Changes of variables


O n e of the best modes of dealing with a diferential equation to which
no other method seems applicable is to change one or both of its variables.
B y this means we may try to modify its form so as to make it subject to
some known method of integration.
Let X and y be replaced by new variables s and , by formulas i

These formulas are to be reversible, so that s and can also be expressed i n


terms of x and y. B y virtue of these formulas
dx

= fs {Sy u) ds

dy = g^,

+ / u (j, w)

) ds +

du.

gu(sy u) duy

20

Some Types of Solvable Equations

and the functions P and Q are replaced by functions of s and , say,


P(x,>) = i ? ( j . a ) ,

Q(x,y)

= r(j,).

Differential equation 2.1 is thus transformed into

R[f, ds + / ^1 + T\g, ds + iu du\ = 0.


If this differential equation can be integrated, and has the general integral
u, c) = 0, the general integral of the original differential equation is
obtainable from
u,c) = 0 by replacing s and w by their values i n
terms of x and)'.
Example

12.

T o integrate the differential equation


+ 11 + > V ' l rfy = 0.

B y the change of variables


X

s,

e~'Uy

the given equation is transformed into


e-'u

ds-\-\\-\-

u'^We-' du -

e-'u

ds\ =

0,

that is, into u^ ds -\- (1 + u-j rfu = 0. T h i s last equation is solvable


since its variables are separable. Its solutions are u = 0 and
j - f (1/2^) -

l o g u = c.

Since s = XfU ~ ^y, the original equation has the solution) = 0, and the
general solution is
V2>-)
Example

13.

log > = c.

T o integrate the differential equation

B y the change of variables x = su, y u, the given equation is transformed into


-u^'du

H- u^s^

u]ds

0.

W e may drop the factor


since u = 0 does not lead to an integral of the
original differential equation. W e thus find that (du/ds) u = s^. For
this last equation, which is linear, e~' is an integrating factor. T h e general
integral is ue^" = { +
2j + 2)
+
Since s = x/y, u = y, the

Changes of Variables

21

general integral of the given equation is


X

\y

ASSIGNMENT
1. Show that the differential equation
(2.11)
i n which d is a constant other than 1, is transformed into a linear difTercntial equation
by the change of variable x "
"
Differential equation 2.11 is known as a
Bemoulli equation, in honor of the Swiss mathematician James Bernoulli (1654-1705).

PROBLEMS
Solve each of the following differential equations by use of the indicated change of
variables, or as a Bernoulli equation.
55.

56.

57.

(1 + 2JC + 2y|

2 + 4* + 2> +

{2y - xylogx]

1 - e] dy - 0,

+ 12x + 2y 1

dx-\-{l

1 +x

dx -2x

+2x+>|<6'-0.

{ 3 / - x^y] dx + {x* sin y + x* -

59.

\4xy -

60.

iy +

V +

dx-\-dy0,
x^y
1

dx+{x-\-

61.

{sin~* y cos xjrfxH

62.

{y + V l -

63.

i2> + 2 x V

64. x V dx -

65.

y = u log s.
X

3xy^] dy = 0,

= S/Uy

y = s.

sm u,
y = s esc u.
X =

x^y\dy = 0,

x '~ y
j==.dy

~ 0,

AT = J ,

1^ r V .

dx + |x + x ' y } dy - 0,

{x - 2 / + 2y^] dy = 0,

y ^ s 2u.

|* ^ ^'2

Irf^+ xrfy - 0,

.-xiv

u.

X - J ,

log xdy 0,

58.

2x*

X "

X = J ,

x " s

</x = 0,

X =

j ^ - " .

22

Some Types of Solvable Equations


e^Hx-\-y-\-

66. (^+'' + 1 -

1)J dx +

{f*+>' -

1)) dy = 0,

1 + ^""(x + ^ -

x = {s + ) / 2 ,
y ^ i s - u)/2.
2
67. / - >
3JC
69. >' + ^

1 =

cot

75. /

77. /

cos X

- .

= 8v'^ sin^

COS X

2x 2\x^ - x

COS

v>.

2>
3x

76. y' -\-3y\ogx = 3>^x

X.

Ay
log

= 2{x + l}

"2- > + - > sin X


2

jy + xy

2)-

+ >

70. y + - ^ >
Jf + 1

>^

Sin

68.

CSC X .

'

3xy

jt =

3
71

2 log

log

Ma
X

2x + 1

+ \\^ ^ 2{x^ -

+ \\y

2.7. Equations witti homogeneous coefficients


A function fix^y) is said to be homogeneous of the degree k if the effect of
replacing x and y by ax and ay, respectively, is merely to multiply the function by a*. Thus
/(ax, ay) = a^f{x, y).
Example 14.

T h e functions
1
.y/xy
tan-"
X

-yh

/Mog

1 +y

are homogeneous of the respective degrees 0, 2 , and f .

For

tan~" == a t a n ^ " -
ax
X

{axp

jayP^ log

1 +

ax
ay

= fl^-;;'^ log

1 +

A n y differential equation 2.1 whose coefficients P{x,y) and Qix^y) are


homogeneous functions of the same degree is transformed by either one of

Equations with Homogeneous Coefficients

23

the changes of variable


(2.12)

X = Sj

y = sUf

X = sUf

y =u

or
(2.13)
into a n
obtained
obtained
changes,

equation with separable variables. T h e differential equation


b y the one change of variable is ordinarily different from that
by the other. It is therefore often advisable to consider both
and to integrate the simpler of the transformed equations.

Example 15.

T o integrate the differential equation


\y -

Vx^ + / 1

dx-

xdy = 0.

T h e coefficients of this equation are both homogeneous of the degree 1


T h e change of variable 2.12 transforms the equation into
V l

-\-u^ds + sdu = 0.

T h i s has separable variables, and has the solution


s{u + V l

= c.

s = Xy and u y/xy the general solution i n terms o( x andy is


y + Vx'+y'
Example 16.

= c.

T o integrate the differential equation


y'^dx-\- Ix^ -

xy-\-y^] dy = 0.

T h e coefficients of this equation are homogeneous of the degree 2. T h e


change of variables 2.12 transforms the equation into
[u + u^] ds+

{\ ~ u-^-u^sdu

= 0,

whereas the change of variable 2.13 transforms it into


uds -\- {s^ -{- \ } du = 0.

O f these the second is obviously simpler.

Its solution is

tan~^ J + log = c,
and, since s = x/y, and u = >, the general solution of the given equation is
tan"^ - + log;" = c.

24

Some Types of Solvable Equations


PROBLEMS

Integrate the following differential equations.


79.

{3x + 2y] dx + \2x +>) dy - 0.

80. {x < _
81.

_ xy
_

-_ 2x'y\dx + {2x* + xy* -\- xY

xY

+ x*y] dy - 0

{x^y -{-xy^ - >'} dx + \xy^ ~ x^] dy - 0.

82. y sec^* - dx +

X sec2 -

dy " 0.

83.

\2x^y +/}

84.

\x -ye^f^

85.

86.

dx + I V rfx

- COS - tfAT

-sin
y

M**^* ^ sin -

2je} rfy = 0.
+

y
h COS X
X

rfy

= 0.

dy - 0,

<ir + jt sin - ((y 0,


X

87. x ) r I o g - r f x +

-Mogj^

2xlog^-x+>|rfx

88.

X'

+x

dy = 0,

89. fl>^^'rfx + {7 - flx^''*! dfy = 0.


90.

>-s/x^ + / r f x -

{x V * ^ + /

+>M

= 0,

2.8. Equations with linear coefficients


Every differential equation
(2.14)

a-iX \- bxy

ci\ dx +

02* +

hiy +

^ 2 ! dy =

0,

whose coefficients are of the first degree i n x and y, can be integrated.


Such a n equation m a y be of a type we have already considered. If
h\ = 02 it is exact; if ci = C2 = Oj its coefficients are homogeneous of the
degree 1; if 2 = 0, it is of the linear type 2.10. W e may therefore suppose
now that 6 2 ^ 0.
T h e change of variables

with a n undetermined constant A, transforms the equation into


\A{s -

k) -V Bu-VC\ds-\-udu

= ^.

Equations with Linear Coefficients

25

If, here,
= 0, the variables are separable. I n that case we may assign
k the value 0. \i A 7^ 0, we may choose h so that Ah-{-C
0. T h e
equation is then one i n which the coefficients are both homogeneous of the
degree 1.
Example 17.

T o integrate the differential equation


{2y-\]dx^-{ix-y

+ 2]dy = {i.

T h e change of variables, s = x -\equation into


[ds-\-u

u = 3 ; f ^ + 2, transforms the

~ ()h-\-^\ ds ~ udu = 0.

W i t h A = ^, this is a n equation with homogeneous coefficients, whose


general integral is [u -\- 2s\'^[u 7>s\^ = c. I n terms of the original
variables, the general integral is thus \Sx y + 3j^j2jv I p = c.
Example 18.

T o integrate the differential equation


{2x -

6> + 3) rfx -

{x -

ly ~ \ \ dy

0.

T h e change of variable, j = x + A, u = x 3^ 1, transforms the


equation into {5w + 15 j ds -\- udu = 0. Here the variables are separable. W e may therefore take k ~ 0. T h e general solution, i n terms of
the original variables, is 2A: ^ log fx Sjy + 2 j = c.
PROBLEMS
F i n d the general Intcgrsd of each of the following differential equations.
91.

{2x+y

- 2\ dx + [x - y-\-2] dy = 0.

92.

{4x + 2> -

93.

{7y - 3}rfx+ {2x + 11 /y = 0.

94.

{2x -y

95.

[2x + 3>! dx

96.

I5x '\- 2y -\- \ ] dx

{x -

97.

{3x - 2> + 4} rfx -

{2x + 7> -

98.

{6x + 4> + 3irfx+ {3x + 2> + 2} rfy = 0.

99.

{;-+71rfx+

l]dx -\- {2x - y\ dy = 0.

+ 2,]dx -

{4x - 2y -\- \] dy

{ix - y -

0.

U] dy = 0.
\ \ dy = 0.
1}rfy= 0.

{ 2 x + > + 3lrfy = 0 .

100.

\4x-{-y - 2\ dx + {?>x+y - 2\dy = 0.

101.

{5x -\-Ay ~ A] dx -\- {4x -\-5y -

102.

{2x-\-y -

I] dx[2x

- 9y -

S\ dy ^ 0.
\] dy = 0.

26

Some Types of Solvable Equations

2.9. Simultaneous equations


Just as in algebra a number of unknowns may be given by an equal
number of simultaneous equations, so in the calculus a number of functions may be given by an equal number of differential equations. A pair
of equations,
z'=/2(x,:v,z),

(2.15)

constitutes such a system for two functions > and z. T h e general solution
of such a system consists of two equations in x,
and
and involves two
arbitrary constants. T o obtain these equations two quadratures have to
be made.
(i) A F I R S T I N T E G R A L . If the first equation 2.15 does not involve z,
or if the second equation 2.15 does not involve y^ we may integrate the
equation in question, and so obtain a first relation in x, y, and z. W h e n
neither equation 2.15 permits such a direct integration, we may use
undetermined multipliers a , ^3, and 7 , to obtain from 2.15 the equation
(2.16)

(a + i3/i +

adx-^^dy^-'idz^

7/2I

dx.

If a , /3, and 7 can be chosen so as to make the left-hand member of equation 2.16 a differential, say dQ (x, >>, z), and at the same time to mzJce the
right-hand member a function (possibly zero) of 9 and x times dx^ this
choice makes the equation appear as
(2.17)

dB =

F{e,x)dx.

T h e general integral of equation 2.17, (p{$, x, c\) = 0, yields one of the


required equations
(2.18)
(ii) T H E S E C O N D
second equation.

* ( x , > , z , ^ i ) = 0.
INTEGRAL.

T W O

methods are available to obtain a

(iifl) It may be possible to reduce equation 2.16 to a form 2.17 by the


use of new multipliers a, /3, 7 that are not proportional to those that were
used in obtaining the first integral. A second integral of form 2.18 which
involves an arbitrary constant c-z is then obtained.
(ii) It may be possible to eliminate either a n d ^ ' , or z and 2', between
equations 2.15 and 2.18. T h e liminant is then a differential equation
for the remaining function. T h e integration of this liminant then yields
the second integral.

Simultaneous Equations
Example 19.

27

T o integrate the differential system


+ 2xz' +

{2x + 2!e -

3 = 0,

x^z' +

Ix^ -{-x}z -

x = 0.

xy' -y-{-

B y algebraic means we may reduce this system to form 2.15


/

= - > - l ,
X

T h e first of the last two equations does not contain z; hence it can be
integrated directly to givey = x
Cix^. B y the elimination ofy between
this equation and the second one of the reduced system, we find that
1 +

z.

X
the general integral of which is xz = 1 -\- ci{\ x] -\~ C2e
general integral of the given differential system is, therefore.

y X
Example 20.

cix\

The

xz = 1 -\- ci{\ x] + C2e

T o integrate the differential system


y

xy + 2xz

xy + 2xz

T h e choice of multipliers a = 0, |8 = 1 , 7 = 2 reduces equation 2.16


i n this case to dB = (d/x) dx^ with 6 = y -\- 2z. Thus a first integral
6 = c\Xy that i s , ^ -\- 2z = cix, is obtained. T h e elimination oiy between
this result and the second equation of the given system yields the equation
z' =

2z{cix - 2z\
ClX

T h e general integral of the last equation is ^ = cix^/{4x


general integral of the given system is thus
y -\- 2z = ciXy
Example 21.

z =

C2),

and the

CiX

4x +

C2

T o integrate the differential system

y - X

z' =

\y -

x\z

T h e choice of multipliers a = 2>', ^3 = 2z, 7 = 2A: reduces equation


2.16 for this case to dB = 0, with B = y"^ z^ x^. Hence y'^ z"^
x^ =
is a first integral. W i t h the different choice of multipliers a = x.

28

Some Typ)es of Solvable Equations

/3 = z , 7 =
we obtain for (2.16) the form dB - 0 with B ^ xy - i z
T h e generzd integral of the system is thus
~

2xy _

x^ = cu

= C2.

r2

PROBLEMS
Integrate the following differential systems.
103. xy' -\- z' -2z

= Ixe',

\x^ + 4x\z' -

104.

105. y' - y 106. y'

>' -

r' +

2z

U + 4}^ - y - y = 0,

\x\\z'

{x + 3J,

2z

y - xz' -

3xV -

3x* - xy' + ;r - 0.

xy' - xy -

^ "

X +

2.

z -\- x^ = 0,

107.

z -

X**

- x' -

1 = 0.

{ 1 + 2 log xjv' - xz' -

+ -> - x r ' - ^ + 0,

z=

0.

108. 2^5' + {3 tan X 109.

-2z

110./

4yz + l
=

2.10.

,
z' =

~4xz
3 +4yr

2z = -j sec* x.

Ayz + 1
3x
3

+4^
z + 2 ~ Ax
2y

2y - 3z
*'{2x -y]

yy' - -fz' -

6x - ;y - 3

111. y

112. y

4|z = 3 sec X + sec* x,

-yly

x\y + 2z\

~3z
1

r =

-2z)
x\y + 2zi

The linear differential systems with constant coefficient

T h e method of undetermined multipliers can be used to integrate any


system
(2.19)

>' = fly H- 6z + ^i(x),

z'^hy-\-kz

+ g2(xh

i n which a, bj k, and k arc constants. W e may i n this case choose a


and /3 and 7 as constants to fulfill the equations
fljS + A7 = mfi,

b^ -\- ky = my.

= 0,

Replacement of a Differential Equation by a System 29


is possible if m 5 taken to be a root of the quadratic equation
(2.20)

a m

k m

= 0.

T h e multipliers then give equation 2.17 the form


de=

with 6 =
the integral
(2.21)

\me-\- ^gi{x) + yg2(x)] dx,

72. Since this is a linear differential equation for 6, it has

= J { ^ 5 i W + yg2{x)\e-^

{&y + yz]e-^

dx + c^^.

process can be carried out for each value of m i f equation 2.20 has
distinct roots. Otherwise the elimination of ^ or z between (2.21) a n d
an equation 2.19, yields a differential equation i n the other variable. T h e
integration of this liminant gives the second integral.
PROBLEMS
Integrate the following differential systenss.
113. y' =ly

z' = 3y - 2z.

114. y = 2z - 4y,

z' = -iz - 5y.

115. / = | y + 5r, z' =


116. /

~y + h,

117. y = z,

z' = ~h

z' = 4y ~ Sz.

i\9. y' = 2y - z + e',

121. y'

+ h-

z' = -y -\- 2z.

118. y' = 3y - 4z,

120. y -'ly

~ h-

z' = 4y ~

+ Zz'h e^',
-2y-\-

3z-{-5.

z' = 2y - Zz - 7**.

2^"'.

122. y' = -2y -\-z-{' 3 " * ' ,

= 3z + 6> + e~^.
z' = 9y - 2z- 2e'.

2.11. The replacement of a single differential equation by a


differential system
W h e n a differential equation 2.1 is given, it is sometimes advantageous
to regard x and y as functions of a third variable tj and to replace the
differential equation by the system
(2.22)

dx/dt = Q{x,y),

dy/dt =

-P(x,y).

T h e two equations that make up the general integral of this system express
X and y i n terms o( t. T h e y therefore give a parametric representation of

30 .

Some Types of Solvable Equations

the integral of the original differential equation. T h a t integral i n terms


of X and y is obtainable by eliminating / between the pair of equations
which constitutes the integral of the system.
Alternative to system 2.22 we could equally well replace differential
equation 2.1 by the system
dx/di = k{x,y)Q{x,y),

dy/di =

-k{x, y)P{x, y),

with any chosen function k{x,y). T h e integration of this system is


generally different for different choices oik{x, y). However, such different
choices only determine effectively different parameters t. W h e n t is
eliminated the same integral for (2.1) is always obtained.
Example 22.

T o integrate the differential equation


V l

dx-\- {x + 2y] dy = 0.

- /

Equivalent system 2.22 is i n this case


dx/dt = x-\-2y,

dyjdi =

V l -

y-.

F r o m the second of the last two equations we find a first integral relation
y cos \t + c\\.

W i t h y thus made known, the first equation of the system, which is linear
i n Jf, yields the second integral:
* = sin j ^ +

(Ti

cos {( +

(Ti

I + c-i.

These two integrals together represent the solution of the given differential
' equation parametriccdly i n terms of t. T h e liminant of t from them is
^^\-y'^-y-\-c

COB y

with c = c^e
This is the general integral of the differential equation
originally given.
PROBLEMS
Integrate each of the following differential equations in terms of a parameter.
123.

{2^ -y

124.

13^^ + 2^ + 4} rf* + {1 + 2e-'\ dy = 0.


1
2;- + y

125.

126.

-2}dx+

l*i dy = 0.

dx + (6* + 1 0 / -

Ij rf> = 0.

dx + {* + 2y^ + 9} rfy = 0

Answers to Problems
1

127.

dx -

X
Xhx + 6x2}

128.

{2x -

31

2x^} dy = 0.

j _ _ 3y

dx = 0.

129. 2 V ^ r f x + V 1 - * ((y -= 0.
130.

rfjr

V l -

((y = 0.

ANSWERS TO ODD-NUMBERED PROBLEMS


1. > = 2 +
5. sin^ y

1 -hex

and y = 2.

3. >r = 4 +

2 tan x = c.

1. y = log
11. fe-v'dy

9. {y + \ / l ~ T ? " | l o g x
13. ^2

_ logxp -

1.

tan * x}.

=c H-i'il

15. log ( . ' + 4 1 -

-xj.

{ x - l - l ^ - ' + f.

17. > X + log {2 - x)2 + c, and> = - 1 .


19. x^y
23.
27.

log ^ + log X

X +
X

ix^ - y = c.

>

21. Inexact.
= c.

25. e' sin > + xe~* c.


29.

+ sin ^ X = c.

- >

33.

39. ye^* =

4x^V^* -

37. cos y;

r
3x^^

43. > = |x + I I U ^ +

57.

->
0,

V^-

= c -

49.

53. y = {x^ + l}{sinx

+ Ix +^1
and x + log

59. > 0, and 2


61.

3*

= c.

sin ^ ^ sin

= x^ X

{1 -x2}W.

45. :y - < r * ( x l o g x + W

51. ^-1x2 + x * | = x ' + x + f.


55.

x sin y y log x

41.

+ c.

47. y = 2x + f

log X
X

2/4

35. x-^y-^^;

/''<''+iU2+/|

65. 2 t a n - i Ix."} + y \/1

2 log jlog x) c.
1 +
63. log X

- >^ + s i n " * > = c.

xV'*' - f

32

Some Types of Solvable Equations

67. >* = log X

75.

cx^.

69. y

sin 2x + ^ tan x.

79. 3x* + 4xy +

f.

2 cos JK + c sin X

73.

- ^V T T x

77. y -

{/^"

V l

+rIogx}*.

81. I o g x + - + - - < : .
y
X

83. -5 + l o g \xy] - c.

85. ^ sin - c.
X

xM

87. -2

J-

1
- ^

logy

91. (2x +y

89. ae'^v + log > - c.

+ log >

2}* -

3b -

2 ^ = c.

93. 7 log {2x + 11 + 2 log |7> -

n{y+2\^

95. [2x + 3>|2 99.

{> 4 - 7 | 2 l 3 x + ^

103. y = fix* -

xz

109. /

1,

111.

X +

X "

2; + 3z

115. > -

119. y =W

|if-*

123. ^ = 1 +
125. X =- 1 -

Ci*"*,

129.

X =

ii4 -

- ^ i d

8x - \0y

C2.

y + z* - ^2.
>z =

X -

X* +

^2.

-r = cie' + 3f2*~'.
-iVi^'*

- '^2'""'^^

z = Cixe* + [ci + C2l*.


C2e-'^^ - W
4x

- f^-" +
~

cie-*^ + ^2^',

1
127. - 1 + e i . - " ,

ly"^ + 2y

{x + 2 }

log X + ci (log x ) ' +

- -k^

4xv + 8x -

= fi^:^-

z =ri{x*+x} -

c i ^ ^ * + C2e-'^^, z -

117. y - cixe' + f2^,

121. > +

=- C i

ci,

113. y = cie* + ^2?"',

97. 3x^ -

101. 5x' +Sxy + 5y^ -

2C1X + 2 ^ + f2,

107. jf - ci log X ,
r +

= c.

+ 11 - ff.

105. r - < r 2 ^ - o U H - 2 | ,

3} = c.

>

6;- -

- i .

+ f2l~".

[ci - / | V 4 -

[ci -

+ 5 + f i ^ + 4^2

.-8*
z = 6 ; - ' - x^""
+ cze

f2*.

(1^ + 2 s i n - 1

CHAPTER

3
Applications of Differential Equations
of the First Order
3.1. Problems in velocities
T o solve a physical problem mathematically, it is commonly necessary
to resort to a differential equation. T h e general integral of this differential equation must be found, and from that the particular integral that
fits some given condition must be determined. T h a t is to say, the family
of integral curves must first be found, and then the curve of this family that
goes through a given point must be isolated. T h e differential equation
expresses the relevant physical law, and the particular integral applies it to
the specific situation.
For problems i n velocities and rates, we shall use t to denote the time,
and i n the case of a moving particle we shall use s to denote the distance
along its path. This distance is to be measured from some fixed point,
with one of the directions taken as positive. T h e choice of the origin and
of the positive direction may be made as seems most convenient.
Once
made, however, it must be adhered to throughout the course of the
solution.
T h e velocity v and the acceleration a of a particle in a straight path are
known from the calculus to be given by the formulas
(3.1)

V = ds/dty

a = dv/dt,

a = v dv/ds.

T h e first two of these formulas are i n effect the definitions of u and a. T h e


third is obtainable as the quotient of the second by the first.
T h e fundamental law that describes the way particles move under the
action of forces is Newton's second law of motion:
(3.2)

/ = ma,

f = m dv/dt,

f mv dv/dsy

these forms all being equivalent by (3.1). m stands for a constant called
the mass of the particle, and / represents the component of the force acting
i n the direction of the path (or the sura of such components if there is more
33

34

Applications

than one force). Force / is to be taken as positive or negative depending


on whether it acts to increase or decrease s.
W h e n a body is allowed to fall freely without resistance under the pull
of gravity, its acceleration is the acceleration of gravity, which is denoted by g.
This is the same for all bodies. Its numerical value depends upon the
units i n which distance and time are measured. I n the foot-pound-second
system it is nearly 32.2. Since law 3.2 applies to falling motion as well as
to any other motion, we see that w = mg, where w is the weight of the
particle. T h u s
(3.3)

w/g.

W h i c h of the three forms of Newton's law 3.2 is to be used in any given


case depends upon the variables i n terms of which the problem is to be
solved. If the problem calls for velocity in terms of time, the second form
of (3.2) is appropriate, for the variables are v and t. If velocity is required
i n terms of distance, or vice versa, the third form is clearly the one to be
used.
W h e n a body falls i n a resisting medium like air or water, the law by
which the force of resistance is related to the velocity varies with a number
of circumstances, such as the density of the medium, the shape of the body,
the magnitude of the velocity. I n each of the following examples we shall
assume a simple law to take a l l this into account. This would not necessarily be the law that experiment would bear out i n the case of some other
body. W e shall be solving differential equation problems, not establishing
physical facts. For numerical results, we shall take the evaluation ^ = 32
to be sufficiently accurate for our purpose.
Example 1.

A 2-lb. particle is dropped from a balloon at a great height.


D u r i n g its fall it is acted upon by air resistance. T o find the velocity of
the particle when it has fallen 1000 ft., if it is assumed that the resistance
amounts to kv^, with k = 1/20,000, i n the foot-pound-second system.
T h e path of the particle is a vertical line. A l o n g this we may take s as
mccisured from the point at which the particle is dropped, and as increasing
downward. T h e acting forces are then the weight w, which is positive,
since it is directed downward like the increasing s, and the air resistance
kv^, which is negative, since it opposes the motion and so is directed
upward. T h e total force is thus {w kv^]. T h e problem calls for v i n
terms of s; hence we choose the third form of law 3.2, and obtain from
it the differential equation
w kv^ = mv dvjds,

with w = 2,k^

1/20,000, and, by (3.3), m = i V -

Problems in Velocities

35

T h e variables i n this differential equation are separable,


integral is

T h e general

,2

- 6 2 5 log

20,000

s -\- c.

T o fit the condition ZJ = 0, when j = 0, the value of c must be c


625 log 2. T h e particular integral that applies is thus
- 625 log

40,000

s,

or, equivalently,
V = 200 V\

- e-'^^^^.

W h e n s = 1000, this equation gives v 178.6 approximately.


Example 2.

A m a n w i t h a parachute jumps from a great height. T h e i r


combined weight is 192 l b . D u r i n g the first 10 sec, before the parachute
opens, the air resistance amounts to kv^ w i t h A: =
i n the foot-poundsecond system. Thereafter, while'the parachute is open, the resistance
amounts to AID, with k = M. T o find the man's velocity of fall 15 sec.
after the j u m p .
W i t h the direction of increasing s taken as downward, the acting forces
are 192 and kv. T h e second form of law 3.2 thus gives the differential
equation
192 -

yti/ =

6 dv/dt.

which has the general integral


(3-4)

log

fl92

k-

o
fo

A n equation like 3.4 but with different values for k and 1 applies to each
part of the fall.
For the first part of the fall A: = 4, and y = 0, when t = 0. T h e particular integral that fits these conditions is
log
that is,

= 256(1

256

A t the end of this part of the fall, that is, at / = 10, this gives the velocity
Pio = 256{1

= 182 (approximately).

36

Applications

For the second part of the fall k = 12, a n d , as we have just found,
V = 182, when / = 10- T h e particular integral 3,4 that fits these conditions is
log
namely,

V -

16

166

20 - 2(,

= 16 + 166^^^"".

At t = 15, this last formula gives


Pi5 = 16 + 166tf

-10

T h e result is very nearly 16 f.p.s., since 166^~^^ amounts only to about


0.007,
PROBLEMS
Solve each of the following problems, assuming that the all data given arc expressed
in the foot-pound-second system.
1. A ball weighing H lb- is falling. T h e air resistance amounts to u/lOO.
p = 15, at / = 0, what is the formula for v in terms of

If

2. A ball weighing 1 lb. is dropped from a great height. The air resistance amounts
to p/32. Find the formula for s in terms of /, if f = 0, when j = 03. A bullet weighing H o lb. is fired vertically downward from a balloon, with a
muzzle velocity of 1000 ft./scc.
T h e air resistance amounts to f^/1,000,000. Find
the formula for v in terms of j .
4. A bullet of weight w is fired straight upward with a muzzle velocity of ro ft/.sec,
The air resistance amounts to kv^. Find the formula for s (the height), in terms of c,
while the bullet is rising.
5. In the case of the bullet of problem 4, find the formula for / in terms of v,
6. A certain particle weighs 8 lb. As it sinks in water its weight forces it downward,
but an upward force, its buoyancy, also acts upon it. This buoyancy amounts to 4 lb.
T h e water resistance amounts to v'. Find v in terms of j , if :J = 0, when j = 0,
7. When a certain body weighing 16 lb. is placed under water, it rises because of
its buoyancy, which amounts to 20 lb. T h e water resistance amounts to
Find
the velocity with which the body rises / seconds after it is released.
8- A motor boat weighing 200 lb. is going at a speed of 20 ft./sec. at / = 0. T h e
motor is then turned off. T h e water resistance amounts to -jj^^. Find the formula for D
in terms of (.
9. A 1-lb. particle moves in a straight line on a smooth horizontal plane. A force
acting upon it amounts to ^ , and is directed so as to repel it from the point s = 0.
Uv = Oy when j = 4, what is the formula for u in terms of s?
10, Find the formula for v in terms of s for the particle of problem 9, if the force
amounts to l A ^ , and is directed so as to attract the particle toward the point j = 0.

Frictional Motion

37

11. A certain vehicle weighs 3200 lb. Its air resistance amounts to 2D. If a force
of 40 lb. is applied to push it, and starts it into motion, find the formula for the distance
s it has gone, when its velocity is v.
12. If the vehicle of problem 11 is moving with the velocity 20 ft./sec. at 0 , and
a braking force of 40 lb. is thereafter applied to it, what is the formula for v in terms of

3.2. Frictional motion


W h e n a particle moves on a rough surface, its motion is opposed by
friction, which acts as a resisting force amounting to fif^fy where
(i) fa is the sum of the components of the forces acting upon the particle i n the direction perpendicular (i.e., normal) to the surface; and
(ii) / i is a constant called the coefficient of friction. T h e value of M ts a
measure of the roughness of the surface. It is zero when the surface is
perfectly smooth.
W i t h the friction taken into account, the motion of the particle is subject to Newton's law 3.2.
Example 3.

A 10-lb. particle is placed upon a plane whose slope is


T h e coefficient of friction is
and, i n the foot-pound-second system, the
air resistance amounts to v/4. T h e particle is at rest at / = 0, and is then
released. T o find its velocity at any subsequent time /.
Let <p be the angle at which the plane is inclined, and let the direction
down the plane be that of increasing s. T h e weight of the particle,
which acts vertically downward, then has the components 10 sin ip along
the plane in the direction of increasing s, and 10 cos <p perpendicular to
the plane. These forces amount to 6 and 8, respectively, since tan <p =
|. T h e friction thus amounts to f lb., and the law of motion is accordingly
8

32 dt

W h e n the variables are separated this differential equation becomes


dv

idt.

and the condition y = 0, when / = 0, is fulfilled if c = log (88/5).


T h e integral that applies is thus

38

Applications

that is,

PROBLEMS
Solve the following problems, assuming all the data to be given in the foot-poundsecond system,
13. A boy and sled together weigh 80 lb. They coast on a hill whose slope is TTJ,
and on which the coefficient of friction is -r^T h e air resistance amounts to v/l'i.
Find the formula for the velocity, i f = 0 , when ( 0,
14. What is the formula for the velocity of the boy and sled of problem 13 if the hill
is perfectly smooth?
15- A boy on skis weighs 96 lb. He slides down a hill whose slope is
which the coefficient of friction is -jfr- T h e air resistance amounts to 2v/l3.
p => 0, when J = 0, find the formula for s in terms of v.

If

16. A 16-Ib, particle slides up a plane whose slope is T, and upon which the coefficient
of friction is x- T h e air resistance amounts to v/\0.
If at / = 0 the velocity of the
particle is 64, what is the formula for v at time t?
17. A 100-Ib- boy slides on the ice. T h e coefficient of friction is 7nr> and the air
resistance amounts to v/20.
If the boy's velocity at / = 0 is 40, what is it at time (?
18. A skater weighing 160 lb. allows himself to be blown along by the wind. T h e
coefficient of friction is -j^j-, and the wind pressure upon him amounts to 2 {30 v].
If his velocity is 14 when t = 0, what is it at any later time /?
19. In the case of the skater of problem 18, find the formula for s in terms of v.
20. A 4-lb. particle moves in a straight line on a horizontal plane whose coefficient
of friction is T . T h e air resistance amounts to v^/l6. T h e force acting upon the
particle amounts to 4J, and is directed to repel it from the point J = 0. If v = 0^
when J = 4, find the formula that relates v and j .

3.3. Problems in rates


I n many applied mathematical problems the immediately known facts
concern the rate at which a certain quantity changes. Finding the
quantity itself then requires the integration of a diflferential equation.
Example 4.

A certain mixing tank of 200-gal. capacity is filled with


brine i n which 60 l b . of salt are dissolved. Beginning at / = 0 the solution i n the tank is drawn off at the rate of 5 gal-/sec., and the tank is
meanwhile refilled at the same rate with a solution that contains
lb.
of salt per gal. T o determine the amount x of salt that is i n solution i n the
tank at time t.
Since at time / there is a total of x l b . of salt i n solution, the amount of
salt per gallon at that time is x/200. T h e rate of withdrawal is thus
5x/200. T h e rate of replacement is T ^ - W e thus have the relation

Problems in Rates
dx

5A:

~ 20

It ~

39

10*

T h i s is a linear differential equation for x. Its general integral is x =


20 + ce^'^^^y and with c determined so that the condition x = 60 when
/ = 0 is fulfilled, the result is
X = 20-\- 40e-'^^^.
Example 5.

A certain compound X is formed by the combination of 2


parts of a chemical U with 3 parts of a chemical W. W h e n certain
amounts of U and W are placed together, the rate at which X is produced
is constantly proportional to the product of the amounts of U and W that
are still present at the instant. T o determine the amount of X that is
produced i n time t, if 10 lb. of U and 81b. of W are placed together at
/ = 0, and the amount of X at / = 1 is 2 lb.
Let x be the amount of compound X that is produced i n time /. T h e
amounts of U and W that have been used are ^x and %x, respectively.
T h e remaining amounts are therefore 10 f x j , and 8 f x j .
We
have thus the differential equation
dx/dt =

k{\0-%x]{B-%x].

log 18 - | x l -

log (10 - |x) = -^kt

Its general integral is


+

.:.

W i t h c determined so that x = 0, when t = 0, this equation yields


log

200 -

15x

200 -

8x

-^kt

T h e fact that x = 2, when / = 1, serves to evaluate k.


that k = l o g f I , and that therefore
.200 -

15x

200 -

8x

W e find from it

.
85
= / log
^ 92

This result can be written as


my\

2oo{i -

15 -

8(M)

PROBLEMS
21, A mixing tank contains 100 gal. of fresh water at / = 0. A solution containing
7 lb. of salt per gal. is then added to it at the rate of 3 g a L / s e c , and the resulting mixture
in the tank is drawn off at the same rate. Find the formula for the amount of salt in
solution in the tank at time /.

40

Applications

22. In a certain volume of liquid 100 lb. of salt, but no more, will eventually dissolve.
T h e rate at which it dissolves is always proportional to the amount that is still dis->
solvable. Find the amount x of salt that will dissolve in time /, if this amount is 5 lb.
when 1 = 1.
23. In a certain volume of liquid, 200 lb. of salt, but no more, will eventually dissolve. When a quantity of salt is placed in this liquid, the rate at which it dissolves is
T^JTC times the product of the undissolved amount by the amount that is still dissolv*
able. If 100 lb- of salt are placed in this liquid at / = 0, what is the amount x that is
dissolved at time /?
24. From problem 23, what is the amount of salt that is dissolved in the liquid in
time t if 200 lb. arc placed in it at ( = 0?
25. From example 5, if 10 lb. of chemical U and 15 lb. of chemical M^arc placed
together at / = 0, how much of compound X is produced at time I?
26- From example 5, if chemicals U and
were to combine in equal amounts to
produce compound x, what would be the amount of compound produced in time
interval / after u lb, of U and w lb. of IV were placed together?
27, T w o chemicals U and IV combine in equal parts to form a compound -V. The
rate at which the compound is produced is proportional to the square root of the product of the amounts of U and IV that are still present at the instant. If at ( = 0 there are
8 units of U and 2 units of W^, find the formula for the amount of compound formed at
time /,
28- If 2 parts of a chemical U combine with 3 parts of a chemical W and the rate
at which the compound is produced is always proportional to the amount of U that
is still present at the instant, what is the formula for the amount of compound produced
in time ty if this amount is M at / = 0 ?
29- T h e population ^ of a certain insect colony is
at ( = 0, In the absence of
outside influences the rate of its growth would be kp. T h e colony is, however, exposed
to an influence that causes deaths at the constant rate . Find the formula for the
population at any time t.
30. In the absence of outside influences a certain population p grows at the rate kp.
However, it is exposed to an outside influence which induces growth at the additional
rate {a + 4 ^ ] , where a and i are constants. If the population is
at time /o, what is
the formula for the population at time t?
31. T h e amount x of a certain chemical disintegrates at a rate which is always proportional to the amount that is still left. If it diminishes by 1 per cent in 10 years,
and its amount is XQ at t = 0, what is the formula for x at time t?
32. T h e amount x of a certain chemical naturally diminishes, because of disintegration, at a rale kx if left to itself. In use there is a further loss by wear at a constant
rate . If the amount is XQ at time ( = 0, what is it at time t?

3A. Problems of flow


A problem of flow is always based upon a differential equation. This
statement applies to a flow of water, or of air, or of some other fluid, and
to an electric current, or the flow of heat from a cooling body.

Problems of Flow

41

W h e n a liquid whose volume is V is allowed to escape from its container


through an opening at a distance y beneath its surface, the rate at which
it flows through that opening is k V ^ , with some constant k. T h e value
of this constant depends upon the physical circumstances of the individual
Ccise, in particular upon the shape and size of the opening, the character
of the fluid, the system of units used, etc. This rate of flow is also the rate
at which the volume V diminishes. W e draw from that fact the relation
(3.5)

-dV/dt

k\fy.

This is not a differential equation such as we have already studied, because


it involves too many variables. A differential equation i n just two variables is, however, obtainable from it, by substituting for dV its value i n
terms of y and dy. T h i s value is determinable in any case from the shape
of the volume F, and the location of the opening through which the liquid
escapes.
I n a case i n which the figure is simple, the formula for V in terms oiy may
be easily discernible. A differentiation then gives dV. It is, however,
often simpler to obtain the formula for dV directly from the figure without
first finding V. This is done by applying the reasoning of the integral
calculus, which says that dV is equal to the area of the surface of the l i q u i d
multiplied by dy.
Example 6.

A cylindrical tank of length 9 ft. and radius 5 ft. is mounted


horizontally and is filled with oil. A t / = 0, a plug at the lowest point of
the tank is removed, and a flow results to which relation 3.5 applies with
k =
T o find the depth ^ of oil i n the tank at any time t while the tank
is draining.
T h e circumference of the end of the tank is the circle
+ [y 5 j ^
= 25.

For this circle, x = +

T h e surface of the oil is thus

a rectangle whose width is 2 V ' l O ) '


and whose length is 9. Since
dV equals the area of this surface times dy, relation 3.5 yields the differential equation

T h e general integral of this equation is


12(10-;'i^ =

^+r.

T o fit the condition that> = 10, when / = 0, the value o f t must be zero.
T h e result can be given the form

42

Applications
t

V = 10 ^

180

I n this case it was simpler to find the value of dV from the figure of the
liquid, than it would have been to find the formula for V and to differentiate it, for by elementary geometry V is found to have the relatively
complicated formula
V = 9

25T -

25 cos"^

1^ +

5)

VlO^ -

If the tank had been standing on end the situation would have been different, for then the formula for V would have been recognizable at a glance.
W h e n a material body whose temperature is T is immersed in a quantity of water whose temperature is T', there is a flow of heat from the body
to the water, or vice versa. T h e law that applies to this is Newton's law of
cooling, which states that the rate at which the temperature of the body
drops, namely, dT/dt,
is proportional to the difference between the
temperatures of the body and water. Thus
(3.6)

-dT/dt

= k{T

T'].

T h e constant k depends upon the physical values involved, among them


the area over which the body is i n contact with the water, the material of
which the body is composed, and the units used.
T o obtain a differential equation i n two variables from equation 3.6, the
value of T' i n terms of T must be substituted. This value can generally
be found from the fact that the amount of heat which the body loses is the
amount that the water gains. T h e change i n the body's heat content is
always obtainable as the product of its weight by its specific heat by the
change i n its temperature.
Example 7.

A body weighs 45 lb. and is made of a metal whose specific


heat is 5 . W h i l e at a temperature of 300 degrees, it is plunged, at ^ = 0,
into 100 lb. of water at a temperature of 50 degrees. T o find the formula
for the temperature T of the body during its cooling.
T h e amount of heat that the body has lost at time / is 45(g) {300 T].
T h e amount of heat the water has then gained is 100(1) j
501 (the
specific heat of the water being 1). Since these two amounts are the
same, we have
V-1300 -

T] = 1001 r

50},

whence 7"' = 65 {T/20).


O n substituting this value into equation 3.6
we obtain the differential equation

Problems of Flow
-dTldi

43

= k

T h e integral which applies is that for w h i c h T = 300, when / = 0.


integral is
r = -y*f|13 + 50^--t2i/20)fcj_

This

F r o m this integral we can observe the temperature to which the body


w i l l eventually cool. As t increases indefinitely the exponential term
approaches zero. T h e limiting value of T is therefore ^ ^ r ^ , that is, 61.9
degrees (approximately).
PROBLEMS
33, A cylindrical tank of length 10 ft. and radius 3 ft, stands on end, and is filled
with oil. When a plug halfway up its side is removed the oil Rows in accordance with
formula 3,5, with k = ^ir- Find the formula for the depth of the oil in the tank while
the tank is more than half full.
34, A tank is 6 ft. long, and its end is a square with a 4-ft. diagonal. At / = 0 it b
half full of oil, and at that instant a plug in a lower edge is removed. T h e flow obeys
law 3.5 with k =
At what time is the tank drained, if it lies horizontally on one
side?
35, At what time is the tank of problem 34 drained if it is mounted horizontally
with a diagonal of its end in the vertical position?
36, A tank is in the form of a right circular cone with altitude 4 ft. and radius 5 ft.
At ^ = 0, it is filled with water, and the water is then allowed to escape through a hole
to which law 3.5 applies, with k = - j ^ . Find the formula for the depth of water in
the tank at time /, if the tank is mounted with its axis vertical and its apex downward,
and if the hole is at the apex.
37- If the tank of problem 36 is mounted with its apex upward, and the whole is in
the base, what is the formula for the time when the depthj^f water in the lank is^?
38. A tank is in the shape of a sphere with radius 4 ft. At /
0, it is filled with
water, and an opening on the level of the center of the sphere then permits the water to
escape. Find the formula for the time in terms of the depth k of the water in the tank,
while the tank is more than half full.
39. A cylindrical tank with radius 4 ft, stands on end, and has an opening, for which
k ^ "ffVj in its lower base. As water excapes through this opening, fresh water is run
into the tank at the rate of 1 cubic ft,/sec.
Find the formula for the time at which the
depth of the water in the tank is A, if A = 9, when / = 0.
40. A 60-lb- piece of metal whose specific heat is -5- is at a temperature of 450 degrees.
At f = 0, it is plunged into 300 lb. of water whose temperature is 60 degrees.
Find
the formula for the temperature of the body at time t.
41. A n 84b. body is made of metal whose specific heat is ^ T J . While at a temperature of 308 degrees it is plunged into 11 lb, of water at a temperature of 53 degrees.
Find the temperature to which the body eventually cools.

44

Applications

42. A 10-lb. body whose specific heat is -g- is at 0 degrees when t ^ 0. It is then
plunged into 100 lb- of water at 80 degrees.
If law 3.6 applies with k yf* what is
the temperature of the body at time t?
43. A 5-lb. body whose specific heat is TnJ" is plunged at /
0 into 50 lb. of water.
If the temperature of the body at this time is 200 degrees^ and if it eventually cools to
50 degrees, what is the formula for its temperature at time /?
44. A 50-lb. body whose specific heat is i^f, and whose temperature is 208 dc^ecs,
is plunged into 20 lb. of oil whose specific heat is 3^, and whose temperature is 50 degrees.
Find the formula for the temperature of the body at time t?
45. T h e differential equation for the current i which flows under an e.m.f. (electromotive force) in a simple electric circuit containing a constant inductance L and a
constant resistance R is
(3-7)

Li-Ri
at

E.

Find the formula for r\ if is a constant, and / = 10 when t ^ /Q.


46. F r o m problem 45, find the formula for current 1 that flows in the circuity if
I >- 0, when t ^ Oj and E ^ a sin <aty a and o being constants.
47. T h e differential equation for charge ^ on a condenser of constant capacity C
in a simple electric circuit with resistance R and e.m.f. E is

(3.8)

4 1 + ^

= ^-

F i n d the formula for q, i E is constant and q ~ go when / = fo48. From problem 47, find the formula for q on the condenser, if ^ = 0, when / 0,
and E = cos at.

3.5. Locus problems.

Rectangular coordinates

A plane curve may often be characterized most directly by a differential


equation, for the properties that define the curve may be simply expressible
i n terms of its tangent and normal lines. T h r o u g h the slopes of these lines
the derivative is involved, and the defining relationship is accordingly a
differential equation. T h e integral of this differential equation then
yields the coordinate equation of the curve.
I n the following statements of problems we shall use the terms " a r e a "
and " l e n g t h " to mean the numerical values of these quantities. They
refer, therefore, to the number of square units or linear units in the figure
referred to. A r e a and length are thus always positive or zero.
Example 8.

T o find the equation of the curve that goes through the


point (3, 1) and whose tangent and normal lines always form with the
AT-axis a triangle whose area is equal to the slope of the tangent line.
Since the slope
is to be equal to an area it must be positive, and since

Locus Problems.

Rectangular Coordinates

45

the point through which the curve is to go has a positive ordinate, we shall
consider > positive too. If {x, y) is any point of the curve, the slopes of the
tangent and normal lines are, respectively, y' and 1 / y ' , and these lines
meet the x-axis at the points (x y/y\ 0) and {x + yy\ 0). T h e triangle
therefore has a base of length [yy' + y/y' \. Since its altitude is we have
1

yy'

T h i s differenticil equation can be written i n the form y' V 2 y^ = y^


Its variables are separable, and the integral for which ^ = 3, when * = 1,
is
- /

-h 2 -

Vaiog

[V2

^]y

x.

T h i s is the required equation.


Example 9.

T o find the equation of the curve that goes through the


point (2,1), such that the area under any of its arcs and above the x-axis is
equal to the length of the respective arc.
F r o m the calculus we know that, for a curve above the x-axis, which
extends to the right from a point (XQ, >O) the area under an arc is given by
the integral /

y dx.

T h e length of the corresponding arc is

T h e curve sought is, therefore, one for which

B y differentiation this becomes y


y' = + y/y^ 1.

+ y'^j or, i n another form,

T h e general integral of this differential equation is


log \y +

V /

ij

x-^c,

and withe determined so that)" = l , w h e n x = 2, the result can be written


y +

^-

It is possible to improve upon this form.


canceling
etc., we find that

O n transposing y^ squaring,

We can now drop the ambiguous signs, since the two alternatives give the

46

Applications

same formula.

T h e equation sought is thus

This is the equation of a catenary. It is the curve i n which a rope or


chain hangs when it is suspended from two of its points. I n terms of the
hyperbolic functions, which are defined i n the calculus by the formulas
(3-9)

svnhx

r^!,

coshjc =

i l ^

^""^h

the result can be expressed i n the form^' == cosh (jt 2 ) .


PROBLEMS
49. Find the equation of the curve that goes through the point (1, 3), and for which
the tangent at any point, and the Une joining that point with the origin, have slopes
which arc the negatives of each other.
50, Find the equation of the curve that goes through the point (4, 3), and whose
tangent and normal lines always form with the x-axis a triangle whose area is equal to
the product of { - T T ) by the slope of the tangent line,
51, Find the equation of the curve that goes through the point (0, 1), and whose
tangent and normal lines alw^ays form with the :r-axis a triangle whose area is equal to
the negative of the slope of the normal line.
52. Find the equation of the curve that goes through the point (4, 2), such that the
segment of its normal line between the curve itself and the >-axi3, is always bisected
by the ^r-axis.

53- Find the equation of the curve through the point (1, 5), whose tangent and
normal lines always make with the ^-axis a triangle whose area is equal to the slope of
the tangent line54. Find the equation of the curve through the point (2, 5), whose tangent and
normal lines always make with the >-axis a triangle whose area is equal to 4 times the
slope of the normal line.
55. Find the equation of the curve through the point (-J, 0), whose tangent and
normal lines cut from the >-axis a segment whose length equals the negative of the slope
of the tangent line.
56. Find the equations of the curves through the point (3, 2), for which the tangent
and normal lines always form with the
and ^-axes a quadrilateral whose area is equal
to 2xy.
57. Find the equation of the curve through the point (4, 2), for which the tangent
line and the horizontal through the point of contact form with the x- and ^-axcs a
trapezoid having the area 6.
58- Find the equations of the curves through the point (4, 3), for which the tangent
line and the vertical through the point of contact always make with the x-axis a triangle
having the area 9,

Polar Coordinates

47

59. Find the equation of the curve through the point (1, 1), for which each tangent
line is at a distance from the origin that is equal to the abscissa of its point of contact.
60. Find the equations of the curves through the point (1, 1), for which any tangent
line and the line from the origin to the point of contact make with the>-axis a triangle
having the area ^.
61. Find the equations of the curves through the point (0, 5), for which the area
under any arc and above the jr-axis is equal to 3 times the length of the respective arc.
62. Find the equation of the curve through the point (-ff, 0), along which the arc
length and the function 2x^ increase at the same rate.

3.6. Polar coordinates


Polar coordinates are sometimes better adapted to the statement and
solution of a given geometrical problem than are rectangular coordinates.

FIG.

3.

T h e two systems are related as shown i n F i g . 3. T h e connecting relations


are
X = r cos df
y r sin Q.
F r o m the calculus, we know that the angle ^ and the differential of arc ds
are given by the formulas
(3.10)

dB

tan^ = rI
dr

and
(3.11)

ds"" =

r''dQ'^-\-dr''.

These are the essential formulas for dealing with the geometrical problems
to be considered.
Example 10.

T o find the equations of the curves through the point with


the polar coordinates (1, ir/2) for which every normal line is at the distance
1 from the pole.

48

Applications

T h e perpendicular from the pole upon the normal line has the direction
of the tangent line. It therefore makes with the radius vector the angle ^
or
i^j according as ^ is acute or obtuse. F o r the curves sought we
have thus 1 / r = cos ^, or 1 / r = cos ( r
W e can write these relations tan 1^ = V ' r ^ 1. F r o m equation 3.10 we thus obtain the
differential equations
r d$/dr =

V ^ ^ .

T h e i r general integrals are


0 = {Vr^

1 -

c o s - i (1/r)) + c,

and with c determined so that $ = T / 2 , when r = 1, the required equations are found to be
6
Example 11.

T/2

{Vr^ -

1 -

sec"^ r ) .

T o find the equations of the curves through the point

with the polar coordinates {'s/2y T / 4 ) along which the arc length increases
at twice the rate of the vectorial angle.
T h e specific property of the curves sought is ds 2 dd. B y (3.11),
this

is

dd^ \- dr^ 2 dd.

We

may

write

this i n the

form

r d6 dr. This equation has r = 2 as an integral- However,


that integral does not go through the given point. T h e generzil integral is
B = i: sin~^
+ Cj and when c is appropriately determined this gives
6 = sin~^
and 6 = s i n " ^ (r/2) + ir/2.
W e can write these equations i n the forms r = 2 sin 0 and r 2 cos 0. T h e curves are circles of
radius 2 through the pole.
PROBLEMS
63, Find the equation of the curve through the point with the polar codrdinates
(2, x/6), for which the angles 0 and ^ are always equal.
64. Find the equation of the curve through the point with the polar coordinates
(3, 3ir/2), for which 4" and 0/2 arc always equal.
65, Find the equation of the curve through the point with the polar coordinates
(1, r ) , for which 1^ is always the supplement of 0/2.
66. Find the equation of the curve through the point with the polar coordinates
(1, ir/4), for which ^ is always the supplement of 2(9.
67. Find the equations of the curves through the point with the polar coordinates
(3, T / 2 ) , for which all the tangent lines are at the distance 3 from the pole,
68, Find the equations of the curves through the point with the polar coordinates
(5, 0), for which the tangent and normal lines are always equidistant from the pole.

Some Curves of Pursuit

49

69. Find the equations of the curves through the point with the polar coordinates
(l/'\/2> ~""/4), for which the distance from the pole to the tangent line is always
equal to the square of the radius vector to the point of contact.
70. Find the equations of the curves through the point with the polar coordinates
(1, 0), for which the distance from the pole to the normal line is always equal to the
square of the radius vector to the point of contact.
71. Find the equation of the curve through the point with the polar coordinates
(1, 0), for which ^ is always equal to log r.
72. Find the equations of the curves through the point with the polar coordinates
(3, 0), along which the arc length increases at 6 times the rate of 6.
73. Find the equations of the curves through the point with the polar coordinates
(1, T / 3 ) , along which the arc length and the square of the radius vector increase at
the same rate.
74. Find the equations of the curves through the point with the polar coordinates
(4, JT), along which the arc length increases at a rate that is always 2 tan
times the
rate at which the radius vector increases.

3.7. Some curves of pursuit


W h e n a point moves so as always to be directed toward a certain goal,
its locus is a curve of pursuit. T h e goal may be fixed or moving, and outside
influences may bear upon the motion of the point itself. A simple example
of a curve of pursuit is the path of an airplane that pursues another while
always keeping its quarry i n the line of sight. W e shall consider this case
here under the following conditions:
(i) T h e pursued airplane 0 flies i n a fixed direction with a constant
ground speed whose eastward and northward components are a and b,
respectively.
(ii) T h e pursuing plane A has a constant air speed V.
(iii) There is a constant wind whose components toward the east and
north are, respectively, Vx and Vy.
It need not concern us much that condition (i) is actually quite artificial
unless 0 is fixed. W e are presently engaged in developing techniques i n
differential equations, not i n obtaining results usable i n aviation. W e
shall consider other curves of pursuit in Section 7.9.
T o the pilot of the pursued plane 0 of F i g . 4, the path of the pursuing
plane A is a curve expressible by an equation either i n the polar coordinates (r, 6) or i n the rectangular coordinates (x x*, y y*).
With
reference to the ground, A has a velocity whose eastward component dx/dt
is made up of the air-speed component V cos (x 0) and the wind component Vx. T h u s
dx/dt = Kcos 0 + Vx.

Applications

50

Similarly, A^s northward velocity component is


dy/dt = - F sin

+ Vy.

For the plane 0 we have dx*/dt = a, and dy*/dt = b.


these equations into

W e can combine

d{x -

= { - F c o s f l H- Vx - a\ dt,

d{y-y*}

= \ -Vsme

Vy-b\dt,

and by eliminating dt find that


(3.12)

[VsmB-Vy-\-b]d{x-x*\

\Vcos6

^ a}d\y -y*\

= 0.

T h r o u g h the use of the relation tan B = {y y*)/{x x*), equation 3.12


is expressible either as a differential equation i n r and B, or as one i n
[x x*\ and \y y*\.

Fio. 4.

Example 12.

A n airplane 0 has a ground speed with constant eastward


and northward components of 144 m.p.h. and 108 m.p.h., respectively.
T h e pilot observes a pursuing plane .4 at a distance of 10 miles due north.
T h e air speed of A is 270 m.p.h., and there is no w i n d . T o find an equation for the path of A as it appears from 0.
W e have i n this case a = \ AA, b = \08,
= 0, Vy = 0, and V = 270.
Equation 3.12 is therefore
270 sin ^ + m\d{x - X*] -

{270costf + 144lrf(;' ->'*1 = 0.

Since j j ; A : * ) r cos B, and {y y*] = r sin By so that d[x AT*j =


r sin 6 d9 + cos B dr^ and d{y y*\ = r cos B dB -\- sin B dr, the equation becomes

+ 4 cos

H- 3 sin B\TdB +

(4 sin 5 -

3 cos B] dr = 0.

T h e variables i n this are separable, and the general integral is


5 log

5 4 cos B 3 sin 0
4 sin 5 3 cos d

-\- log 4 sin

3 cos B] + log r =

c.

Some Curves of Pursuit

51

T o fit the condition r = 10, when 6 = i r / 2 , the value ofc must be 10


T h e resulting equation can be put into the form
r =

10 V 2 4 s i n g (5 -

y/l.

3cosg|

4 cos g - Ssintfj^^

Example 13.

A n airplane A whose air speed is 200 m.p.h. takes off for a


fixed destination 0 which is due west and at a distance of 300 miles. A
constant w i n d blows with components of 15 m.p.h., and 20 m.p.h.,
toward the west and north, respectively. T o find the path of A if it is
always directed toward 0.
Since the point 0 is fixed, we may take it to be the origin. We have,
then, X * = 0,
= 0, a = 0, i = 0,
= - 1 5 , Vy = 20, and V = 200.
Equation 3.12 is thus
{200 sin e -

20) dx -

{200 cos 6 +

15) dy = 0.

Since tan 0 = y/xj we have cos 6 = V ^ ^ ^ * M - > ^ , and sin d =


T h e equation is thus
200;Vx^

20

dx

200x

Vx'-\-y

y^

+ 15

dy =

y/'Vx^

T h i s is a differential equation i n which both coefficients are homogeneous


functions of the degree 0. T h e method of Section 2.7 therefore serves to
integrate it. T h e integral for which x = 300, w h e n ^ = 0, is found to be
10 V x ^

log {4x-\-3y\ + 8

log

+jy2 _

4x

- h

6^ _|_ 8^

= log (1200)

3y

We can write this i n the form


10 Vx^ -\-y^ -

6x + Sy]^ = 1200{4x +

3y]\

A curve of pursuit of a different kind is the locus of one end of a piece of


string when the other end is drawn along a given curve. I n F i g . 5 the
piece of string is shown as QP. Its length a is constant, and the curve
/(x*, y*) = 0, along which point P is being drawn, is given. Since a
string can pull only along its own direction, the path of Q has QP as its
tangent line. Hence y' = tan 0, and therewith the obvious relations
X* X = a cos 6, y* y = a sin tf, can be given the forms
(3.13)

x-\-

ay

V l

y*

Applications

52

W h e n these values are substituted i n the given equation/(x*,^*) = 0, the


result is the differential equation for the path of Q. T w o loci are always
to be expected, since point P may be drawn i n either direction along the
given curve. W h e n the given curve is a straight line, the locus of Q is
called a tractrix.
14. One end Q of a piece of string of the length 3 is placed at
the point (2, 2), and the other end is then drawn along the line y = x.
T o find the equation of the locus of Q.
Example

FIG.

5.

T h e given curve has the equation y* = x*.


values 3.13 transforms^* = x* into

T h e substitution of the

3/
V l

V l

/"2

which is the differential equation for the path of Q.


by making the change of variables x-{-y
= SyXy
is thereby transformed into
u V l

(du/ds)^ =

-3

It can be integrated
= u.
T h e equation

Vidu/ds,

that is, into


Vl8

One integral of this is u = 0.


-

V l 8 -

u^du

uds.

T h e general integral is

+ V l 8 log

Vis

Vis

s-\-c.

I n terms of the original variables, the integral for which y = 2 , when


X = 2, is

Some Curves of Pursuit


{x'\'y]

= A/lSlog

V2(x -

53

y)
V2

V i s -

(x

-y)\

T h r o u g h the two choices of sign, this gives the equations of twotractrices.


PROBLEMS
75. A n airplane O whose ground speed is constant at 200 m.p.h. is flying northeast.
T h e pilot observes a pursuing plane A southeast of him at a distance of 6 miles. T h e
air speed of ^ is 300 m.p.h. There is no wind. F i n d the equation of A^s path as the
path appears from O.
76. A n airplane O is flying due north with a ground speed of 200 m,p.h., when the
pilot observes a pursuing plane ^ at a distance of 14 miles directly east. T h e pursuer's
air speed is 240 m,p.h. There is no wind. Find the equation of A's path as it appears
from O.
77- A n airplane 0 is flying with a ground speed whose components are 100 m.p-h.
westward and 110 m.p.h. northward. T h e pilot observes a pursuing plane A directly
east at a distance of 10 miles. T h e pursuer's air speed is 250 m.p.h., and there is a
steady wind of 40 m.p.h, from the north. Find the equation of A*s path as it appears
from O,
78, A n airplane O has a ground speed whose components are 100 m,p,h. to the east
and 240 m.p.h. to the north. T h e pilot observes a pursuing plane ^4 at a distance of
15 miles due west. The pursuer's air speed is 300 m.p,h. There is no wind. Find
the equation of A^s path as it appears from O,
79- A n airplane A whose air speed is 250 m.p-h., takes off^ for a point 0 which is due
west and 150 miles distant. A wind of 25 m.p.h. blows from the south. T h e pilot
keeps the plane directed toward O. Find an equation for the path of the plane,
80. A n airplane A whose air speed is 250 m,p,h. takes off" for a point O which is
200 miles west and 150 miles south. A wind of 25 m.p.h. blows from the south.
F i n d the equation of the plane's path,
81, A n airplane A whose air speed is 200 m,p,h. takes off for a point 0 which is 100
miles due south, A wind of 30 m.p.h. blows from the west. Find the equation of
the plane's path,
82. A n airplane whose air speed is 200 m.p.h. takes off* for a point 100 miles due
south, A steady wind blows, with components of 20 m,p.h, from the west and 10
m.p.h. from the north- Find an equation for the plane's path,
83, One end Q of a piece of string of length 5 is placed at the point (4, 3), and the
other end is then drawn along the x-axis. Find the equations of the loci of Q,
84, One end Q of a piece of string of length 2 is placed at the point (0, 2), and the
other end is then drawn along the line ^ = 4. Find the equations of the loci of Q.
85. One end Q of a piece of string of length 13 is placed at the point (5, 2), and the
other is then drawn along the ^-axis. Find the equations of the loci of Q.

54

Applications

86. One end Q of a piece of string of length 2 is placed at the point (1, 1), and the
other end is then drawn along the line x
y = 0, Find the equations of the loci of Q ,
87. One end Q of a piece of string of length 6 is placed at the point (3, 1), and the
other end is then drawn along the line ^ + 1 0. Find the equation of the locus of Q ,
88. One end P of a piece of string PQ whose length is 2 is drawn along the circle
x^
" b^t for which b > a. Find the differential equation for the locus of
and show that
+
= 6^
is an integral of this differential equation.
89. A ferry boat whose speed is a 8 m.p.h. plies between opposite points on the banks
of a straight river 1 mile wide. T h e river flows at 4 m.p.h. T h e boat is always kept
pointed directly at its objective. Find an equation for its path.
90- If the ferry boat of problem 89 takes off for a point that is z niilc upstream on the
opposite bank, what is an equation for its path?
91. Show that if the ferry boat of problem 89 had a speed of only 4 m-p-h, it could
not reach a point directly opposite the starting point, and find how far downstream
its landing would have to be.
92, A ferryboat whose speed is V m.p-h. plies on a river K miles wide. T h e river
flows at V m.p.h. If the boat takes off for a point on the opposite bank a miles upstream^
and is kept pointed at its objective, what is the equation of its path?

3.8. Suspension cables


A piece of rope or cable, like any other body, is subject to Newton's law
of motion 3.2. W h e n it is at rest, therefore, the resultant of the forces
upon it, or on any piece of it, is zero. T h e key to the solution of many
problems concerning ropes, cables, chains, etc., lies i n this fact.
y

FIG.

6.

W h e n a cable (or rope or chain) is supported at two of its points, and


hangs between these points under the burden of some distributed load, we
call it a suspension cable. T h e cable of a suspension bridge is an example,
as is also any heavy chain burdened by its own weight. T h e curve i n
which the cable hangs is determined by the distribution of its load. W e
shall see how this curve can be found.
I n F i g . 6, OQ is any segment of a suspension cable, of which one end 0 is

Suspension Cables

55

the cable's lowest point. T h a t point is called the vertex. T h e tangent line
there is horizontal. As i n the figure, we shall take the origin at this point.
T h e forces acting upon the segment OQ are its burden or load L (which is
not indicated i n the figure, but which acts vertically downward) and the
pulls from the adjoining parts of the cable. These pulls are along the
tangents of the curve. Thus the pull P at 0 is horizontal, whereas at Q
we have tan tp y'. Since the whole segment of cable is at rest, the sum
of the components of all forces acting upon it i n the horizontal, or the
vertical, direction is zero. Thus,
p cos ^ P = 0,

^ sin ^ Z, = 0.

T h e elimination of/* from these equations yields


(3.14)

y' = L/P.

If the density of the load as referred to the horizontal is designated by p(x),


the value of L is given by the formula
(3.15)

fjp{x)dx.

Equation 3.14 is thus a differential equation for the curve i n which the
cable hangs.
Example 15.

A suspension cable hangs from the tops of two towers that


are 200 ft. apart, and its vertex is 50 ft. below the points of support. T h e
weight of the cable itself is negligible, but the cable bears a load whose
density is 100 + x^j. T o find the equation of the curve i n which the
cable hangs.
Formula 3.15 gives
100 + x^] dx = lOOx +

L =

W,

and therewith differential equation 3.14 is


/

= ^ | l 0 0 x + ^x3

T h e integral of the last equation for which y = 0, when x = 0, is


y =

50x2 + - 1 X"'

T h e curve passes through the tower tops, whose coordinates are ( 1 0 0 ,


50). T o fit this condition, the value of P must be 530,000/3. The curve

56

Applications

thus has the equation


y =

10,600

600

If the burden on the cable is not distributed horizontally, but is distributed along the cable itself, say with the density p(j), where s is the arc
length of the curve measured from the vertex, the formula for L is
=

/; Pis) ds.

W i t h this formula, equation 3.14 is


(3.16)

= ^

j\{s)ds.

T o obtain a differential equation from (3.16) we proceed as follows: Since


ds = V l

dx, the derivative of (3.16) is

(3.17)

dy' =

WT7^

dx.

If it is possible to eliminate s between (3.17) and (3.16), the result is a


differential equation for^', the integral of which, with^"' = 0, at x = 0,
is the required differential equation for the curve of the cable.
Example 16.

A chain is supported at two points on the same horizontal


level and 12 ft. apart. Its slope at a point of support is and its density
is given by the formula 1 + as\~^'^. T o find the differential equation
for the curve of the chain, and then, i n the case a = 0, to find the curve
itself.
Equations 3.16 and 3.17 are in this case
y

= -

Wx+as

~ 1),

d/

..ZL^

dx.

P V 1 + as
T h e elimination of
equation

+ asy and thus of J , yields for y' the differential

T h e required integral of the last equation is


aP[y/\

+2

log ( Z + ' s / l

=2x/P.

This is the differential equation for the curve of the chain.

Suspension Cables

57

W e are prepared to integrate the last equation only if a = 0. T h e


chain is then of uniform density. I n that case the differential equation is
log ( y + V i + / 2 }

=v^,

and since it was given that y' = T , when x = 6, we see that P has the value
6/log 2. T h e differential equation can therefore be put into the form

Its integral for which ^ = 0, when x = 0, is thus


^ -

f .(i/6)Iog2 I -Cj:^6)log2 _ 9)

Iog2 ^

A n alternative form for the last equation is


'

log2'''

PROBLEMS
93. A suspension cable of negligible weight is burdened with a load distributed
horizontally with a constant density. The supports of the cable are 400 ft, apart, and
80 ft, higher than the vertex. Find the equation of the curve in which the cable hangs.
94, A suspension cable of negligible weight is burdened with a load dbtributed
horizontally with the density jlOO + x^/100)}. T h e supports of the cable are 160
ft- apart, and 64 ft, higher than the vertex. Find the equation of the curve in which the
cable hangs.
95, A suspension cable of negligible weight is burdened with a load distributed
horizontally with the density (24 + Z t ^ ) / \ / 2 4 + x^. A t ;r =^ 1, the slope of the
cable is
F i n d the equation of the curve in which the cable hangs.
96. A suspension cable of negligible weight is burdened with a load distributed
horizontally with the density j2
-j^e'^^^}.
T h e point (10, 4) is on the curve in
which the cable hangs. Find the equation of this curve.
97, A suspension cable of negligible weight is burdened with a load distributed
horizontally with the density 35 sec^ (TAT/SOO). T h e slope of the cable at a point of
support is -y, and these points are 150 ft, apart. Find the equation of the curve in
which the cable hangs.
98. A cable of constant density hangs under its own weight without any other load.
At a point of support its slope is
and these points arc 10 ft. apart. Find the equation of the curve in which the cable hangs,
99. A chain of constant density hangs freely from two supports 2 ft. apart, and the
slope of the chain at one of them is -j-. Find the equation of the curve in which the
chain hangs.
100, T h e density of a certain string varies with the distance from one of its points O,
and is p{s) T I ^ ' +
T h e string is hung from two pegs so that the point O

58

Applications

is the vertex, and its length is adjusted until the pull upon one of the p ^ has the hori*
zontal component 1 lb. Find the equation of the curve in which the string hangs.
101. T h e density of a certain rope at distance s from one of its points 0 is p{s) =
sec^ J ,
T h e rope is hung from two pegs so that the point 0 is the vertex, and its length
is adjusted until the pull upon one of the pegs has the horizontal component ^ lb.
Find the equation of the curve in which the rope hangs102. In the case of the rope of problem 101, fmd the equation of the curve in which
it h a n ^ if its length is adjusted until the pull upon one of the pegs has the horizontal
component 2 / V 3

lb.

3.9. The rope around a shaft


W h e n a rope is passed over a shaft or around a post, a large pull P on
one end of it can be held i n check by a much smaller pull p on the other

T
Fio. 7.

end. W i t h a given value of P , the value of p depends upon the extent to


which the rope is wound around the post or shaft, namely, upon the angle
$ which the arc of contact between the rope and the shaft subtends. It
depends also upon /x, the coefficient of friction between the rope and the
shaft.
Suppose the rope is of constant density p, and that the shaft is horizontal
and of radius a. T h e n , when pull p is just enough to keep the rope from
yielding to P , the configuration is that of Fig. 7. A t any point of contact
Q the gravitational pull due to the weight of the rope has the intensity p,
and this has the component p cos ^ in the direction of the rope. Its other
component is merged in the pressure that the shaft exerts upon the rope.

The Rope Around a Shaft

59

I f we designate the normal component of the intensity of that pressure by


/ ( v ) , the corresponding frictional component amounts to nf{(p). These
force intensities are indicated i n the figure by dashed vectors. T o obtain
the components of the forces which they generate, we must integrate their
respective components with respect to s. W e note that ds = a dtp. B y
taking components horizontally to the right, and then upward, we thus
obtain the equations
9
'0

\p cos ipsvfMp -\- f{<p) cos (p ixfi<p) sin <p\a dtp p{6) sin ^ = 0,

and
P+

cos <p

C O S <p

+ f(ip) sin <p - h M

/ W COS (p]a

d<p

+ p{e) cos ^ = 0,

T h e derivatives of these equations are


ap sin dcose

-\-

(cos ^ -

ju sin

^ sin ^ dd

p{e) cos ^ = 0,

cos0 -

p{e) sin 9 = 0.

ap cos^ 0 + af{Q) {sin 8 -\- ^cosO}


dd

T h e elimination of f(6) from them gives the equation


(3.18)

{dp/dS) + up = ap cos 6.

This is the differential equation for/?. T h e requisite integral is the one for
which p = Py when ^ = 0.
W h e n the rope is i n a horizontal plane, and is passed around a vertical
post instead of a horizontal shaft, the effect of the rope's weight is eliminated because it acts along the post and not upon it. T h e differential
equation i n that case is accordingly
(3.19)

(dp/dd) + fxp = 0.

Example 17.

A rope weighing 2 lb./ft. hangs over a horizontal shaft


1 ft. in diameter upon which the coefficient of friction is ^. O n one side a
3-ft. piece of rope overhangs, and this has a weight of 50 l b . attached to it.
O n the other side a 5-ft. piece overhangs, and sustains the pull pi. T o
find what pi must be to keep the rope from moving.
In this case a = - j , p = 2, and / i = ^. Also P = 50 + 6, and when
6 = IT then p = pi -\- 10. Differential equation 3.18 is
(dp/dS) -i-ip

= cos 0

60

Applications

Its integral, for which p = 56, when $ 0^ is


A i l ^5/3

= -^^ cos 0 + A sin ^ + W ^


O n setting 6 = ir^ and p = pi

10, we find that

pi = rffi557tf"^^^ - 33) = 16-27 lb- approximately


PROBLEMS
103, A rope b passed three*fourths of a revolution around a vertical post upon
which the coefficient of friction is -ff, A pull of 100 lb, is exerted upon one end of it,
What pull must be exerted upon the other end to keep the rope from moving?
104, A man capable of exerting a pull of 100 lb. wishes to hold a pull of 300 lb, in
check by passing the rope around a vertical post upon which the coefficient of friction
Over how large an angle must he wind the rope around the post?
3
105, By passing a rope just once completely around a certain post^ it is possible to
hold a pull on one end of it in check by a pull one*fourth as great on the other end.
What is the coefficient of friction on this post?
106, A rope whose weight is negligible hangs over a horizontal shaft upon which the
coefficient of friction is -j. One overhanging end has a weight of 100 lb, attached to
it, and the other end sustains a weight of W lb. What arc the limits within which fV
must lie if the rope is not to move?
107, A rope weighing 1 lb./ft, is wound l-j times around a horizontal shaft whose
radius is 6 in, and upon which the coefficient of friction is -j. N o rope overhangs at
one end. How much rope may overhang at the other end without causing the rope to
go into motion?
108, A rope passes over a large horizontal cylindrical surface whose radius is 6 ft.
and upon which the coefficient of friction is y. At one end the rope overhangs by 5 ft,,
and to this end a weight of W lb, is attached. T h e arc of contact between the rope and
the surface subtends a right angle, and the pull exerted upon the horizontal end is
just enough to keep the rope from moving. How much larger must this pull be if the
rope weighs 5 lb,/ft, than if its weight is negligible?

ANSWERS TO ODD-NUMBERED PROBLEMS


1, ^ = 50 - 35e
tan - 1

5. / = "

7. V = 4

3, V -

9. V = \6 V l o g (j/4),

-At

11, s = 5 0 1 " 20 log [1 - v/20]\.


15, J = 468 log

90e-^'^^^^'r

tan

I - eAt
1

100 \^\0

24
24 - v\

39
2

13. V = 54(1 ^ ^-^'''^1,


17_ V = 40|2^-^'/^" -

1|

Answers to Problems
19. I - 70 log

23. * - 200

27. * -

28 -

10 -

2511

21. X

--3</ioo[

150t/

25.

_ 9,-A(/2_

99 1

31. X ^ xo

61

6kt + 1

29.P--

P0--

t/io

33. 5 +

1001

360x)

35. / = 320 \ / 2 .
37. ( = 5 T 1 H ^ -

- 2y^\.

160;.^ +

39. t = 960ir 13 -

s/h

30 -

30 log

VA

27

41. 5 8 t degrees.

43.

4 5 . . - -

47. 9 -

49. xy = 3.
S3, y =' 6 -

^2 - ^ s i n - 12. -

and

65. r =

1 cos

69. r = cos tf, and r ~ sin 6.

and r = 3.

(2r) - \ / 3 , and

1 - sec

e = sec"^ (2r) + \/3


6 \ / 2 jcos

\/4r^ -

- sin

1 - 2jr/3.
77. r =

| \ / 2 - cos e - sin ^ l ^ ^

90
5 + 4 cos

3 sin

81. u + V ^ t ^ + y i 20

79,

83.

85.

= i

cos"i

= \^AT^ -

75. r =

{x-\y'-\-y^

- iU'^^ + O^-^^^j^

63. r = 4 sin 6.
= 3,

1}.
59.

8.

61. > = i l 9 ^ ^ ' +

73.

V 2 - x\

57. 2xy -y

71. log r -

+ 1^0 - Cl*-'^*^.

51. ;r 1 -

55. ^ = - J +

67. r sin

r=50-|-150-'i*'^^*''*'.

- 4} - v/25

ii)--

- 5 log

5
2| = V 1 6 9 - .c^ _ 13

5 + V 2 5 -

,13 + \ / l 6 9 '
Sx

- 12,

1,000,000;- 17

62
87. J -

Applications
89. y -

-1.

93. y = xVSOO.

91. \ mile downstream.


95. ^ V f f ! [ 2 4 + x 2 ] H _

99

101. ^ -

(S cos (3^5) + V Z S cos2 (3JC

log J

103. 100*""'=' lb.


105. M -

[24]'^).

log 9

-4.

(20.8 lb. approximately).

( l A ) log 2

107. (3/20) {"" -1- 1 |ft.

(\/i/2)il - * | .

(0.22 approximately).
(3.62 ft. approximately).

CHAPTER

4
Integral Curves
Trajectories
4.1. Integrals of a differential equation
T h e methods of Chapter 2 for solving differential equations
(4.1)

=M>)

were found i n Chapter 3 to be applicable to many kinds of physical problems. It remains a fact, however, that all these methods are essentiedly
special. It would be easy to write down differential equations 4.1 i n great
variety, to which no one of the methods given would apply. A n d that
would remain true even if more methods were to be given, as they could
be. F o r the fact is, that, among a l l differential equations 4.1, those for
which a solving formula, other than one i n infinite series, can (or could) be
given are relatively rare. This is by no means due to a mere lack of
mathematical proficiency or skill. T h e reason lies deeper than that.
A n elementary calculus formula is a combination of certain functions
that are more or less familiar. T h e category of these functions is relatively quite restricted. It includes the exponential, the trigonometric,
and the power functions, together with their inverses, and the successive
quadratures of combinations of these. N o w a differential equation may
define its function through a relationship which, i n its very nature, is
inexpressible by such means. I n short, it may define a function proper to
itself, which is not representable by any combination of elementary functions, just as log X, for instance, is not representable by any combination
of polynomials.
N o w it is, of course, quite apropos to inquire what one shall take to be an
integral of a differential equation, when no formula for one can be given.
T h e answer is to be found i n the definition of the term "function." I n the
calculus y is defined to be a function of x upon a specified x-interval, if to
each X of that interval there corresponds a value of y. This does not
require that y be expressible by a formula. It requires only that y be i n
some way determined. Such determination may, for instance, be by a
63

64

Integral Cur\'es; Trajectories

graph, or by some process of calculation through which y is computable,


either exactly or to as high a degree of accuracy as may be desired.
This definition is also adopted i n the theory of differential equations.
I n Chapter 1 we have already considered the direction field of a differential equation 4.1. That led to the conception of the streamlines of the
field as integral curves, and thus suggested that the graph of the general
integral is a family of curves one of which passes through any chosen point
of a region. Analytically this statement implies that a differential equation is fulfilled by a family of functions, and that, subject to proper restrictions, there is a member of that family which for an arbitrarily assigned
value xo takes on a prescribed value ^o-

4.2. The direction field of a differential system


T h e considerations which led to the notion of a direction field for a single
differential equation can easily be extended to the case of a pair of simultaneous equations.
(4.2)

=fi{x,y,z),

z'

=fo{x,y,2).

Consider any region of three-dimensional (jt, y, z) space i n which the functions fi{x, yj z) and f'i{x, >, z) are real and single veilued. A t each point of
such a region these functions have numerical values, and the triple of
numbers 1, / i ( x , ^, z), f^ix^ y^ z), is thus determined there. These numbers, however, determine a direction in space, namely, the direction whose
cosines are proportional to them. By assigning this direction to the point,
and doing so for each point of the region, we impose upon the region a
three-dimensional direction field.
A particle, starting at any point of such a field, and moving thence
always i n the direction of the field, describes a streamline. For these
streamlines, as for any curve, t/x, dy^ dzy are direction numbers.
Because
these numbers also give the direction of the field, we see that along a
streamline
dx

dy

f\{x,yy z)

dz
fiixyy,

z)

This means that the functions y and z defined by the streamlines fulfilled
equations 4.2. T h e streamlines are thus integral curves.
A curve i n space is given analytically by two equations i n (v, y, z).
E a c h of these equations separately defines a surface. Together they
define the curve which is the intersection of the surfaces. E a c h i n d i v i d u a l
integral curve is thus the intersection of a surface of one family by one of
another. Since each of these families is given by an equation involving a n

The Isoclines of a Direction Field

65

arbitrary constant, the pair of equations representing the family of integral


curves involves two such constants. These observations have already
been borne out by the actual solutions of differential systems of the form
4.2 that are found i n Sections 2.9 and 2.10. We shall return to this subject again. I n the sections immediately following, however, we shall consider further the integrals of a single differential equation 4.1 only.

4.3. The isoclines of a direction field


T h e direction field of a differential equation 4.1 over a chosen region is
graphically determinable by calculating the values o f / ( x , y) at a chosen
and suitably dense set of points, and assigning to each of these points the
direction with the slope respectively calculated. W h e n considerable
accuracy is required, the number of points for which the calculation must
be made is apt to be relatively large. T h e labor involved may then be
substantial. Various methods, such as the following, are known for
reducing this labor.
For each value k which the function f{x, y) takes on in the region of the
field, the equation
(4.3)

fix, y) = k

defines a locus that traverses the region. This is the locus at each point of
which the slope of the field is k. It is thus a locus along which the field
has a single inclination. It is therefore called an isocline of the field. B y
drawing i n a number of isoclines, the plotting of a field and the sketching
of the streamlines may often be considerably facilitated. W e note especially that the isoclines for which k = 0 are the loci along which the integral curves have their maxima and m i n i m a .
Example 1.

T o plot the integral curves of the differential equation


,

2y-2

T h e isoclines are i n this case the loci {2y 2)/{x 2) = k.


the straight lines

These are

y - \ = ^ U - 2 ] ,

which pass through the point (2, 1). T h e isocline is crossed by all the
integral curves at the same slope, and that is here twice the slope of the
isocline itself. Figure 8 applies to this example. It shows in full lines
some integral curves, and in dashed lines a pair of isoclines.

Integral Curves; Trajectories

Fio. 8,

Fio. 9.

The Isoclines of a Direction Field


Example 2.

67

T o plot the integral curves of the differential equation


,

y =

1 -3{x+>P
1 +3lx+^l

T h e isoclines are i n this case the loci


1 -3ix+^p

that is, the straight lines

A l l these have the slope 1 . It is clear that k cannot have any value
greater than 1, and hence that no integral curve ever has a slope greater
than 1. Some integral curves and a few isoclines are shown i n F i g . 9.
ASSIGNMENTS
With the use of isoclines, plot the direction field and sketch in some integral curves
for each of the following differential equations.
1. y =x/y.

3. /

2. /

= ~2xy.

= ^ ^ -

A. y'

^
X

5. / = - l ^ .

6. / = \ / ? T ? .

7, Show that if :to is a value for which /C^o) = 0, the line ^ = xo is an isocline of
the linear equation
(4.4)

y'+p(x)y

=q{x).

8. Show that if
i3 any value for which p(xi) ^ 0, all the directions of the field of
differential equation 4.4 along the line x ^ x\ are concurrent in the point \x\ + l/p(xi)j
g{xi)/p(x\)]. This 5 shown for the difTercntial equation

in Fig, 10.
In the manner of Fig. 10, plot the direction field and some integral curves for each
of the following differential cquations4
9.y'^x-2xy.

U.y'=h-y-

, 0 . / =

12./ =

^ ^ + ! .

? ^ ^ .

68

Integral Curves; Trajectories


y

Fro. 10.

4.4. Trajectories
If two families of curves JFi and SF2 occupy a region of the {x^y) plane, one
curve of each family passing through any point of the region, the curves of
either family are said to be trajectories of those of the other. T h e relation
between curves and their trajectories is determined by the law of their
intersection. W h e n a family ffi is given, the trajectories which intersect
it i n the point {x, y) at the angle i/'(x, y) has a differential equation which
can be determined.
Let the differential equation of the family SJi be (4,1). T h i s can be
found by the method of Section 1.5. T h e inclination of its direction field
at (x, y) is tan~^ /(x, y). T h e inclination of the field of the trajectories is
therefore {4/{x,y) + xarT^ f{x,y)\. Hence the trajectories have the differential equation
/

This equation is

= tan {yp{xyy) + larT^ f{x,

y)].

Trajectories

69

if 1^ is a right angle everywhere, and otherwise, by the addition formula for


the tangent,
(4.6)

f{x, y) + tan yp(x, y)

It - f{x,y)

tan^(jf,>)

T h e case i n which is a right angle everywhere is especially important.


T h e two families of curves are then said to be orthogonal trajectories of each
other. Orthogonal families often have interesting a n d important
geometrical properties. Such families also occur i n many connections i n
physics. T h e lines of electric flow in a conducting plate, for instance, are
the orthogonal trajectories of the lines along which the potential is constant. A n d i n the plane flow of a n incompressible liquid, the lines of
constant pressure are the orthogonal trajectories of the streamlines.
Example 3.

T o find the equation of the trajectories that cut the hyperbolas of the family
2x^ c everywhere at the angle tan~^ (^).
T h e diff"erential equation of the family of hyperbolas is ^' = 2x/y.
Differential equation 4.6 for the trajectories is thus
y

{2x/y) -f- 1/3


1 -

2x/'hy

which has the general integral


[x-^y}'[2x-y\'

This equation gives the family of trajectories.


Example 4.

T o find the equation of the orthogonal trajectories of the

curves
4x-\-ly

= cx^'\

T h e differential equation of this given family of curves is


,
^

\2x + 25v
Ax

T h a t of the orthogonal trajectories is, by (4.5),

\2x - h 25y

T h e general integral of this equation, namely


log {Ax' + nxy - h 25y^} - - tan
2
gives the family of trajectories.

Ay

70

Integral Curves; Trajectories


PROBLEMS

1, F i n d the orthogonal trajectories of the family of parabolas

= ex.

2. F i n d the equation of the orthogonal trajectories of the family of hyperbolas

3, Find the orthogonal trajectories of the family of circles x^ + y^

= 2cy.

4, Find the trajectories that intersect the lines of the family x + y = c at the angle
45 degrees.
5. F i n d the equation of the family of trajectories that intersect the lines y ^ ax -\- c
at the angle tan~^ x.
6. Find the equation of the family of orthogonal trajectories of the ellipses x^/a^ +

7* Find the orthogonal trajectories of the family of parabolas


8. F i n d the equation of the family of trajectories that intersect the lines * + a> =
c at the angle t a n " ^
9. Find the equation of the family of trajectories that intersect the circles
+
= c
at the angle tan~^ (x* + y^\.
10. Find the equation of the family of trajectories that intersect the ellipses
y^/2S
at the angle whose tangent is
11. Find the trajectories that make with the curves of the family
tan~* (y/x) = c the angle { - t a n ~ ^ \ / x * +

12. Find the equation of the family of orthogonal trajectories of the curves
2{1
+3sin"^;'

4.5. Singular points of a direction field


F r o m the fact that each point of a direction field lies upon an integral
curve, it might seem natural to conclude that each point lies upon only one
such curve. That, however, need not be so.
Example 5.

T h e differential equation
/

= 2{x^ -\-y -

1|'^ - 2x

has the general integral

TTirough the point (2, 3) therefore passes the particular integral curve

But the curve y = \ ~ x^ is also an integral curve, and it also passes


through the point i n question.

Singular Points of a Direction Field

71

W h e n two integral curves have a point i n common they are, of course,


tangent there, for they must both conform to the single direction of the
field. A point of a field through which just one integral curve passes is
called an ordinary point. A point which is not ordinary is called a singular
point.

If we broaden our definition of a field to permit the inclusion of points


at which the function f{xjy) is not continuous, such points may be singular
points. However they are not necessarily so. F o r instance, i n the case of
an equation 1,4, namely,
(4.7)

Pi",

y)

<2(*,

y)

a point at which Q{xy y) = 0 is one of discontinuity. However, i f at that


point P(jf, y) 7^ 0, the field merely has the vertical direction there, as may
be seen by writing the differential equation i n the reciprocal form
dx/dy = Q{xjy)/P{xyy).
T h e point may then be simply ordinary.
I n the neighborhood of a singular point the integral curves may conform
to any one of a number of difTercnt patterns. For instance, i n F i g . 8, all
the integral curves pass through the singular point (2, 1), and none at all
pass through any other point of the line x = 2, although all such points are
singular. T h e whole course of the integral curves is often in large measure
determined by their character near a singular point. T o show some
possible families of integral curves we shall consider the differential
equation.

=i n which a, 6, A, and k are constants, and ak bh ^ 0. F o r this equation


the origin is a singular point.
T o a considerable extent the integral curves of equation 4.8 are typical
of those of a much broader class of differential equations 4.7, namely, of
those for which Q{x, y) and
y) are both zero at the origin. If P(x, y)
and Q(x, y) are replaced by their Taylor's series, such a differential equation appears as
ax-\-by-\y

kx-^ky-\-

P2ix,

y)

Q2(x, y),

with P2{xy y) and Q2(^, y) having no terms of a degree less than 2. N e a r


the origin, therefore, this equation closely resembles equation 4.8 if
ak bh 7^ 0. W h i l e that fact alone does not permit the inference that
the integrals of the one differential equation resemble those of the other,
that inference can nevertheless often be justified by other means.

72

Integral Curves; Trajectories


PROBLEMS

In each of the following cases find for the difTcrential equation the particular integral
that fulfills the respective condition. Then show that the function yi{x) given is also
such an integral.
13. y

{x^-\-2y -

y{-3)

l\^-x,

yi(x)~i[\

=4.

14. y' = j2** + 4x ;r(l) 15. y yi\)

16.y

>(3)

6.

x^].

1 1 + 4* +

;-i(:c) -

2** +

4,

4:..

-s/yTs
-

yiix) 0.

0.

\^-2

yi(x)

-3.

-3.

1+
^(4)

y('2)

20.

1.

>(i)

= i:r.

-iix-V^'-4>i.

18. y =

19. y

yi(x)

1.

>i(x)

=ix2
3x2,

+jrj -

-1.

>iW

y = , / ? l ^ - 3 .
:)'(4) -

-12.

yi(x)

-3x.

4.6. Some elementary cases of the differential equation


, ^ _ ox + by
^
hx + fcy
Section 2.8 shows how any differential equation 4.8 may be integrated.
T h e solving formula is, however, often such that the character of the
family of integral curves is not easily discerned from it. T w o cases are
exceptional i n this respect. If a = A = 0, and ^ = A, the differential
equation is simply y' y/x.
For this equation the family of integral
curves is the family of straight lines through the origin. I n this instance,
therefore, a l l the integral curves pass through the singular point.
If b = h, the differential equation admits {hx -\- ky\ as an integrating

The Integral Curves as Trajectories

73

factor, and the general integral found by means of this factor is


ax^ +

h\xy +

ky^ = c.

T h e integral curves are thus a family of similar conies centered at the


origin and having their axes i n common. T h e y are ellipses (or circles) if
[ak bh\ > 0, and hyperbolas together with their common asymptotes if
{ak bh\ < 0. Figure 11 illustrates these cases. I n (a), no integral
passes through the singular point; i n (b), two integral curves do so.
A n y differential equation 4.8 remains unchanged if x is replaced by mx,
andy by my, with any constant m.
Such a replacement can be interpreted

Via. 11.

either as a change of scale, or alternatively, as a magnification (or diminution) of the geometrical figure. T h e family of integral curves therefore
always appears unchanged when the figure is magnified. Figures 11
clearly have that property. T h e origin is the only singular point of the
differential equation; hence every other point has just one integral curve
through it.

47,

The integral curves as trajectories

I n cases of differential equation 4.8 other than those dealt with i n


Section 4.6, the form of the integral curves can quite readily be inferred
without any recourse to the actual solving formulas. This inference could
be made, for instance, by using the fact that the straight lines through the
singular point are isoclines. W e shall make the inference by considering
the integral curves as trajectories.

74

Integral Curves; Trajectories

T h e trajectories that cut the straight Hues through the origin at the
angle ^(x, y) have the differential equation
y

O'A) + tan ^

tan 4f

{y/x)

T h i s is equation 4.8 if
ax -\- by
hx

ky

(y/x) + tan ^
1

0/*) tan ^

that is, i f
k(y/x)^{b-\-k\(y/x)a
b(y/^)^-\-{a-k\{y/x)-h

T h e integral curves thus cut any line y = \x at the angle iff given by the
formula
^"''^

bX'^ +

[a -

k]\ - h

B y virtue of equation 3.10, we may write (4.9) alternatively i n the form


U 10
^ ^

b\^+la~k\\-h

^I'og'-I _

de

jtX2+

{b-\-k]\-\-a

where (r, 0) are the polar coordinates of the intersection of the integral
curve with the line y = }ix. Thus X = tan 0. T h e discriminants of the
quadratic forms i n the numerator and denominator of formula 4.9 are
Di=

{b + h]^-4ak,

D2 =

{a -

k]^-{-4bh.

I n this case the numerator of formula 4.9 is not


zero for any value of X; therefore the angle ^ is never zero. A l l the integral
curves thus cut across every straight line through the origin and so encircle
this point completely. If any one of them closes, all do so, for all are
similar. T h e y are then a family of ovals such as were shown in Fig. 11 (a).
If they do not close they are spirals. A l o n g these the distance from the
origin always increases or always decreases if D2 ^ 0, for the numerator of
formula 4.10 then maintains its sign. W h e n D2 > 0, the distance from
the origin alternately increases and decreases along the spirals. Figure 12
shows families of integral curves of these respective types.
C A S E 1.

Di < 0.

I n this case the numerator of formula 4.9 is a


perfect square. It therefore vanishes at just one value of X, say X i . T h e
straight line y = Xix is an integral curve. This line is therefore not
crossed by any other integral curve except at the singular point. A s
X > X i , t h e numerical value given by (4.10) increases indefinitely. Thus r
C A S E 2.

Di 0.

as Trajectories

4x-f 9y
x-Ay

^ "

75

8jc+y
7x+2y

(6)

(o)

Fio. 13.

either approaches zero or becomes infinite. If D2 ^ 0, formula 4.10


maintains its sign, and log r always increases or always decreases. If
D% > 0, there are 2 lines through the origin upon which
^) is a right
angle. O n these lines r has a relative maximum and m i n i m u m . Figure
13 shows families of integral curves of these respective types.

Integral Curves; Trajectories

76

3x+>
w Bx-2y
y -

6x+5y

(a)

f6)
Fio. 14.

2x-Zy
X

F I O . 15.

Answers to Problems

77

3. Di > 0, A N D {ak bk] > 0. I n this case the numerator of


formula 4.9 vanishes for 2 values of X, say X i and X2. T h e lines y = \\x
and y = X2* are integral curves. They are therefore not crossed by any
other integral cur\'es except at the singular point. A l l the integral curves
are tangent to one of these lines at the origin, and approach parallelism
with the other as r increases indefinitely. Figure 14 shows families of
integral curves of this type, (a) applying in an instance in which D-z < 0,
and (b) i n an instance i n which D2 > 0.
CASE

4. {ak bk\ < 0. I n this case Di is necessarily positive, a n d


there are therefore again two straight integralsjy = Xix andy = Xox. T h e
right-hand member of formula 4.9 has a derivative whose denominator is
a squared quantity, and whose numerator is a negative definite quadratic
form i n X. Thus tan ip always decreases as X increases, that is, a l l the
integral curves, except for the straight ones, are convex toward the origin.
T h e y have the straight integrals as asymptotes. Figure 15 shows a family
of integral curves of this type.
CASE

ANSWERS TO ODD-NUMBERED PROBLEMS


1. T h e family of ellipses 2x^ +
3. T h e family of circles
1

+
a

-\-+

7. The family of parabolas V


y
2 tan ' X- = r.
1

9. X- +y"
2 -

11. T h e straight lines y = ex.


13. y =
15.
17.

{>

+
X

5 , 2 + 4;,

+|.

I p = 4x.

\y^\\^^x.

19. y =^\\

2x^^ -

= 2cx.

log 11 + ax\ = c.

a2

= ^.

8x'}.

-\-

= c ~ x.

CHAPTER

5
Approximate Solutions
Infinite Series
Existence and Uniqueness
of a Solution
5.1. Polygonal graph approximations
A graph to approximate the integral curve of a differential equation
(5.1)

=My),

for which
(5.2)

= ^-o,

y{xo)

can be constructed i n the following way: Let XQ, xi, X 2 , " ' * , be any set
of abscissas such that XQ < xi < X2 < ' ' . F r o m (XQ, ^O) draw the
straight-line segment with the slope /{XQ, yo) until it meets the line x = xi.
Let the coordinates of this point be (JTI, {), and from the point draw the
straight-line segment with the slope f(xi, Yi) until it meets the line x = X2.
Give this point the coordinates (x2, K 2 ) , and from this second point draw
the straight-line segment with the slope / ( j f 2 , J^z). Continue i n this way.
T h e graph so constructed is a chain of segments, that is, an open polygon, w i t h vertices at the points (AT,-, Yi), a n d is called a polygonal graph.
W i t h YQ = yof we have for this graph
(5.3)

Yi+i

Yi^fixi,

Yi){xi+,

Xi],

I = 0, 1, 2,

. . .

T h e direction of advance along this graph coincides at each vertex w i t h


that of the field of the differential equation. T h e graph may therefore be
expected to diverge but little from a streamline if the intervals {xi^i Xi)
are short. It then gives an approximation to an integral curve; moreover,
the shorter the intervals taken, the better the approximation. T h e
approximation is always best near the point {XQ, yo). A s the construction
78

Polygonal Graph Approximations

79

advances fron^ this point the approximation becomes poorer because the
divergences accumulate. A polygonal graph extending to the left from
the point (;ifo, ^o) can be constructed in a similar way.
Example 1.

T o construct for the differential equation


1

7J

the polygonal graph to the right from the point (0, 0) with [xij^i ~ Xi\ = ^.
Also to compare this graph with the integral curve over the interval
In this instance
=
= 0, and formula 5.3 gives Yi = T ^ .
This,
together with x\ = -y, yields f{xi, Yi) = jsF o r m u l a 5.3 thus gives
^2 = i ^ '
Proceeding in this way, the values of T a b l e 1 may be
calculated:
Table 1
1

0
Yi
fi^i. Yd

0
1

1
1 7

1 17

e 07

T7

1 7

7^

1 17

1S43

T5T

ffO 7 3

2
15 4 3

5 14 1

7144

20S 5

"807T
6 14 1

^144

The graph obtained is that designated by Y i n F i g . 16.


Approximate solutions are primarily useful when the differential equation cannot be integrated. I n this example the differential equation is
linear and therefore can be integrated. We may, therefore compare the
approximation with the actual integral curve. T h e relevant integral of
the differential equation is
y =

that is.

-1

.-V2

ds.

ea:V4

y =

where I{x) stands for the probability integral


I{x)

ds.

T h e values of !{x) and of e" are obtainaljle from standard mathematical


tables. Thus the numerical values of y can be computed. T o three
significant figures, they compare with those of K,-, as Table 2 shows.

80

Approximations, Series, and Existence


Table 2

5
T

-0.083

-0.177

-0.305

-0.502

-0.837

-0.087

-0.197

-0.368

-0.676

-1.30

0
0

T h e graph of > is included i n F i g . 16. T h e approximation is, i n this


instance, at most moderately good, because the intervals taken were relatively large. W i t h smaller intervals the number of computations needed

F I G . 16.

to cover the given range would have been larger, but the approximation
would also been better.
ASSIGNMENTS
1. Construct a polygonal graph for the differential equation

y'-i-y-

through the point (0, 1), with [x,-+i xi] = T , and plot also the integral curve.
2. Construct the polygonal graph for the diflTcrential equation

through the point (0, 1), with {xi+i -

x, and plot

Xi\ =

also the integral curve.

3. Construct the polygonal graph for the differential equation

= y +

- Xy

through the point (0, 1), with \xi^\ Xi] ^ \^ and plot also the integral curve.
{Note: Fig. 10 applies to this differential equation.)

Approximations for Simultaneous Equations

81

4. Gomtruct the polygonal graph for the differential equation


>' =- sin {x +y]
through the point (0, T / 2 ) , with

x,} = -j-.

1,
(The integral in this case is

y = 2 t a n - l [e^] - x.)

5.2. Polygonal approximations for simultaneous equations


T h e method of polygonal graphs can be applied also to approximate the
solution of a differential system
(5.4)

y=fi{x,y,z),

z'

^f2{xyyyz\

for which
(5.5)

z(x(,) = ZQ.

y{xQ) = yoy

T h e integral curves of such a system are found i n Section 4.2 to be three


dimensional. T h e y may, however, be studied through their projections
upon the (x, y) and the (x, z) coordinate planes. These projections are
plane curves, one defining as a function of x and originating at the point
(jffl, ^o)) the other defining z{x) and originating at the point (xo, ^o). W e
may construct polygonal graphs to approximate these curves.
W i t h abscissas xo, x i , X2, * , chosen as i n Section 5.1, draw the
straight-line segment with the slope fi{xo,yoj ZQ) from the point (xo, >o) to
the point where it intersects the line x = x i . Give this point the coordinates (xi, K i ) . Similarly, draw the straight-line segment with the slope
fzixot yoi ZQ) from the point (XQ, ZQ) to the point with the abscissa x i , say
(xi, Z i ) . F r o m the two points thus determined, draw the straight-line
Table 3
1

Xi

Yi

0.5

0.37

0.19

Zi

1.5

1.58

1.75

/i(x.-, Vi, Z . )

-0.25

-0.36

/2(xi, Yi, Zi) .

0.17

0.34

Yi+i

0.37

0.19

Zi+i

1.58

1.75

-0.52
0.46
-0.07
1.98

-0.07
1.98
-0.63
0.70
-0.39
2.33

-0.39
2.33

82

Approximations, Series, and Existence

segments with the respective slopes


Yu ^i)
corresponding points {x2y Y^) and {x2, -^2) etc.
Example 2.

and / 2 ( ^ i , ^ i , ^ i ) to

T o construct for the differential system


y' ==2y +
z' =

z-{-

- 3 > -2z

ie^'^ -

h'^^

3,

+ 5,

the polygonal graphs for which ^(0) = ^j, and z(0) = -f, with (xi^i *,)
T h e method described calls for the computation of the values which
T a b l e 3 sets forth.

Fro. 17.

T h e graphs Y and Z as they may be plotted from these, and as they


appear upon a single plane, are indicated in F i g . 17.
T h e actual integration of the differential system is in this instance possible. B y the method of Section 2.9, the relevant solution is found to be
y ~

I se

Z =

I -t e^

-\- se

T h e graphs of these relations based upon the computed values set forth i n
T a b l e 4, are also shown i n F i g . 17.

Power Series, Polynomial Approximations

83

Table 4
1

0.5

0.34

0.14

-0.16

-0.54

1.5

1.62

1.81

2.12

2.52

5.3. Solutions in power series.

Polynomial approximations

W h e n the function f(x,y) in differential equation 5.1 is representable by


its Taylor's series about the point {XQ, yo), the solution fulfilling condition
5.2 can be obtained as a power series. T h e sum of the first {n + 1} terms
of that series is an approximating polynomial of the degree n.
By the change of variable
S

XQ.

u = y -

yo;

condition 5.2 is replaced by w(0) = 0, and the differential equation is


transformed into
(5.6)

du/ds = / i (s, u).

T h e function f\{s, u) is then expressible i n a series in powers of s and u,


namely.
(5.7)
k.3 - 0

and it is the supposition that this series converges for some range of values
of s and u. F r o m the calculus, the coefficients Aj^k are known to be given
by the formula
L

ds' du

Ju = 0

I n practice it is, however, often easier to find series 5.7 by combining


known power series for terms or factors of / i {s, u). A m o n g the most useful
of such series are the following:
1 4-r

(5.8)

' = 1 + ^ + 1
2!

^2

+ i r3 _j_
3!

84

Approximations, Series, and Existence


sin z

6_

3!

. . .

^5!

(5.8)

log \a ^ z\ = log a +
Let u be taken i n the form
(5.9)
with undetermined coefficients I, (22, as, * . W h e n this is substituted
into series 5.7 and the requisite multiplications are made, all terms i n like
powers of s may be collected, and their sum equated to the term of that
power in the series for du/ds. W e obtain thus a system of equations which
successively determine the values of a i , a^, as, * * " . The solution is
then obtained by expressing equation 5.9 in terms of the original variables.
A n assurance that this procedure actually does give a solution can be
obtained only by proving that the power series which it yields for the solution is convergent. T h a t can be done. W e shall not, however, do it
here. T h e whole matter of convergence is irrelevant if only the first n
terms of the series, instead of the whole series, are considered, that is, if the
method is applied to obtain a polynomial approximation. T h e method is
applicable also to systems of simultaneous differential equations, to obtain
either power series solutions or polynomial approximations.
Example 3.

T o find for the differential equation

1
\ X y

the polynomial of the fourth degree that approximates the solution for
which _y(l) = - 2 .
T h e change of variables s = x \, u = y -\- 2 transforms the given
differential equation into form 5.6, with
his,

u) =

{2-

is+u)\ I

By the first equation of series 5.8, therefore,

| .

Power Series, Polynomial Approximations

85

F r o m (5.9),
-1-

(ai +

1)J +

{s +

u)^ =

(aj. +

2.2 +
1) V

(^ +

)' =

(ai +

l ) 3_3
V +

025^ +

OiS^

2^2(^1 +

1).^

and thus.
a2

{ay

4
az

+ 1)"
8

, a^iax +

1)

, (11 +

1)

16

O n equating this to the series for du/ds, and comparing the coefficients of
like powers of j , we see that
1

+ 1

2ao =
a2

3(Ja = -

4^4

1)'

(ai +

^3

^2(^1 + 1)

(^1 +

1)'

16

T h e first of these equations gives a\\ the second then gives 2, the third
thereupon 03, etc. Thus,
=

+ A ^ ' + irs^

T ^ ^ '

I n terms of the original variables, the approximating polynomial of the


fourth degree is thus

Example 4.

T o find for the differential equation


y' = e''y j^x

2,

the polynomial of the sixth degree that approximates the integral for which
T h e change of variable u = y \ gives form 5.6 with s = x, and
/ i = ^*'C+i) -\-x-2.
By the second equation of 5.8, with z = x\u + 1),

86

Approximations, Series, and Existence

therefore,

T o terms of less than the sixth degree this last equation would be
/i

j( ^

-\- ^x* -\- x*u -\-

O n substituting form 5.9, and equating the result to the series


that

we find

fli = 1 ,
2^2 = 1,
3^3

= 1,

6^6 =

+ ai,

fl3

Thus

and the polynomial sought is

Example 5.

T o find for the difi'erential system


/

= ^'^^^'\

z' -

(x+jy) +

z,

the polynomials of the fourth degree that approximate the solution for
which ^(0) = 0, and (0) = 0.
T h e second and third equations of 5.8 give
^io(v+*) = 1 + s i n ( y + ^) + : i s i n 2 ( y + z) 4 - : ^ s i n ' 0 + z) +
sin (y + 2) = (y + ^) - i ( y + z)3 +
Sin^

^ (y +

- ^(y +

sin^(y + z) = (y + ^)' +
whence

,
+

. . . ,

Power Series, Polynomial Approximations

87

T h e substitutions
y = aix -\- a2X^ + a^x^ + a^x^ +
b,x

2*2 ^ ^3 ^ ^^4 +

,
. . .

therefore yield
2
02 + 2 +

and
{x+yy

+ z = bix

-f

(2 +

(01 +

1)^1*'' +
3 + 2^2(01 +

1)|X+

O n equating these series respectively to the series for u' and z', we obtain
the equations
ai =

1,

bi =

0,

2a2 = a i +

1,

2b2 = bi,
3^3 = ^2 + 2

i(oi

+ 1)',

3i-3 = 2 +

(1

4a4 = 3 +

63 +

4^4 = ^'3 +

2d2(ai +

[ai +

6 i ] [ o 2 + 2],

l),

T h e required approximations are thus found to be


yA = x-{'^x^-^

W
ix'

+ %x\

PROBLEMS
Find for each of the following differential equations or systems the polynomial
approximation of the fourth degree for the solution specified.
1. v' = > - 6e-',

3. /

1 - 2x

y(0) = 3.

y(0) = 0.

2./

= ^ f ' ' y(0)

4. / - 2 ^ + > ,

y(\) = 0 ,

88

Approximations, Series, and Existence

5. y'

?
3x > 3

7. y'

>2+8U+2}

8. y'

2""<'+i'>,

9. y'

10. y'
z'
11. y'

12. y'

\ -

>(1) -

3.

6. /

+log il

3 + 2 sin X,

:K ~

y(0)

= 0

:v(-2)=0.

j(0) =- 0.

z + Axy,

2 + 2> + 2>2 + xz^ + 2x^,


3 + 4x -x+y-2z,

2> + 4z + Axz,
y{0) - 0 .

y{0) = 0,

z(0) - 0 .

:y(0)=0,
2 + log 11 + 1.

z(0) - 0.

r ' = I + icy,

^(0) - 0.
;f(0) =- 0,

z(0) -

0.

5.4. The method of successive approximation


T h e solution of differential equation 5.1 which fulfills condition 5.2 is a
function
for which
/ W

~f{x,y{x))

^0.

B y a quadrature, therefore,
y{x) -yQ-

f'f(x,y(x))dx

^ 0.

T h e last equation says that>'(j:) is a solution of the equation


(5.10)

y=yo

f^'jix,y)dx.

Conversely, any continuous solution of equation 5.10 is a solution of 5.1


and 5.2, as is easily checked by differentiating (5.10) and setting x = XQ.
T h e single equation 5.10 therefore effectively stands for the two relations
5.1 and 5.2.
W h e n ^ i n the function /(x, y) is replaced by ^IQ, the resulting function
/(*> yo) depends only on x and can be integrated. T h e formula
yiix)

= >o + j^,/(^.>o) dx

thus defines the function yi{x).


W i t h this formula for _>'I(A:), now,
fix, yi{x)) is a function of x that can be integrated.
Hence the formula
yiix)

= >o +

fixyyiix))

dx

defines the function ^2(x). This process can (theoretically) be repeated


again and again, and thus the functions of the sequence
yMyy2ix)yyzix),

Successive Approximation

89

can be successively defined by the formula


(5.11)

=>o + fjj{x,yn-i(x))

yM

n = 1, 2, 3, - .

dx,

W e shall prove i n Section 5.6 that, when suitable hypotheses on /{x, y) are
fulfilled, the functions yn{x) converge to a limit when
>, and that
this limit function is the integral of the differential equation. T h a t proof
will show that the functions^i(x),^2(j^)j>3(^)) * * " , are approximations
to the integral. T h e y are called successive approximations.
Example 6.

T o find for the differential equation


y' =

2<l'-y

the successive approximations to the integral for which ^(0) = 1.


I n this instance XQ = 0, and^o ~ 1- T h u s /{x, yo) = le" - 1, and, by
(5.11), w i t h n = 1,
fj{2e'-\}dx^ 2e'

>iW = 1 +

X -

1.

W i t h this, n o w , / ( x , >'i(x)) = \ -\- x, and by (5.11), with n = 2,


y2{x)

= 1+

1 +

rfx

= 1 + X + ix\

Continuing i n this way, f{x, yzix)) = 2^*


(5.11), w i t h n = 3,
yM

= 1 +

2^ -

A:

1 ,

dx

(1

2e'

A: + - I X 2 ) ,

I n this instance, which is simple, the process can be continued.


that
2e' yn{x) =

\\-\-x +

\-\-x-\--x^-\-

2!

^x'^'j-^,x^-\'

3!

-\--xni

a n d , by

when

! ? when

W e find
n is odd,

n is even.

It is easy to see from this general formula how the approximation takes
place. T h e polynomial part oiynix) is the sum of the first (n -{- 11 terms
of the power series for
Hence, as n is taken larger and larger, ^ ( . 1 : )
converges to e^, which is the solution of the given differential equation and
condition.
It is only rarely possible to make all the successive quadratures as shown
in example 6. A few of them at most are ordinarily feasible, and formulas

Approximations, Series, and Existence

90

can then be found for only the first few successive approximations.
Nevertheless, the method is of considerable practical importance, for even
the first few approximations are often quite good, and further quadratures,
even if they are impossible, can still be calculated numerically by a number
of different methods.
Example 7.

T o find for the differentied equation


y

2 -

e-v

\ -\-2x'

the successive approximations to the integral for which ^(0) = 0.


Since (xo, yo) is (0, 0), we have /(x, yo) = 1/(1 + 2x), and thus
dx

; ^

= 2log|l+2.1.

Using the last equation,

1
y2(x)=log{l+2x| + | ^ - p ^ -

hence.

1.

W i t h this i n turn.
Kx, yzix))

1
l+2x

exp

11+2x}

1
VI

+2x.

and
yz(x) = l o g ( l + 2 x 1 -

1 +

1
V I + 2x

exp

1
V l

+ 2x.

+ 2

T h e quadratures that are required to carry the process farther are difficult
or impossible. L e t us, however, compare the approximations we have
found with the actual integral, which is^ = log 1 + x|. O n the interval
M ^ X ^ 2, the respective values, to three significant figures, are those
of T a b l e 5.
Tabic 5
T

1
T

yi

-0.347

0.202

0.347

yi

-0.279

0.223

-0.289

0.288

.If

5
T

7
T

0.458

0.549

0.627

0.693

0.752

0.805

0.400

0.550

0.676

0.788

0.886

0.975

1.05

0.223

0.405

0.559

0.692

0.809

0.913

1.01

1.09

0.223

0.405

0.560

0.693

0.811

0.916

1.01

1.10

Successive Approximations for Systems

91

It appears from these values that even ^ 2 W is a tolerable approximation


near (ATQ, JVO), and that yz{x) is quite a good approximation over the v/hole
interval.
PROBLEMS
13. Find the general formula for yn(x) for the differential equation y' ^ y
with the condition ^(0) = 3,

6^"^,

14. Find the general formula for ^nW for the differential equation y = ^ + ^, with
the condition >(1) = 0.
15. Find the formula for y%{x) for the differential equation y' > + 2 sin
with
the condition >(0) = 0. Find adso the integral of the difTcrcntial equation and condition, and compare its Taylor's series with y^{x),
16. Find the formula for > A W for the differential equation y' = cos x y^ with the
condition ^(0) = 1. Find also the integral, and compare its Taylor's series with

17. Find the formula for y\{x) for the differential equation
> +
^ , with
the condition ^(0) = 3. F i n d also the integral, and compare its Taylor's series with

18. Find the formula iov yz{x) for the differential equation >' = 7^ + x^y + 1, with
the condition ^(0) = 0.
19- Find the formula for yz{x) for the difTcrcntial equation y' = \y
the condition ^(1) J-^.

A i with

20, Find the formula for^'sW for the diflrcrentialcquationy =^ \y^ 2y}/{2
with the condition y(0) = 1.

2^1,

5*5* Successive approximations for difFerentiai systems


T h e method of successive approximations is also applicable to a system
of simultaneous differential equations. I n the case of a system 5.4 with
the conditions 5.5 it shapes up as follows: T h e solution of the system consists of a pair of functions y{x) and z{x)y which fulfill the equations of the
system. By quadratures they are therefore seen to fulfill the equations

z = 0+

I'f2{x,y,

z) dx.

JXa

T h e functions
yo, ZQ) and fzix, yo,
integrated. T h e equations
:yiW = >o +
zi{x)

ZQ

-\- f

ZQ)

involve only x a n d can be

vo, ^0) dx,


f2{x, yo, zo) dx

92

Approximations, Series, and Existence

therefore define the functions )'i(x) a n d zi{x).


yzix) =>o

zi{x)) dx,

j'ji{x,yi{x),

2 2 W = ^ 0 + l'ji{x,yi{x),

define the functions


Example 8.

W i t h these the equations

zi{x)) dx

and Z2W, etc.

T o find for the differential system


y' = x-\- z,

z' = \ -

y,

the successive approximations to the integral for which ^(0) = 2, a n d


z(0) - 1.
T h e formulcis that are successively found arc
2H-X,

zi = 1 +

\)dx

jy2 = 2 +

2 + x,

^2 = 1 + j '

\ -\

= 2 +

>3

^3=1

jy4 = 2

Z4

= 1

- x\

x\ dx = \

dx^2

-x-ix'^\

x-^^x\

x] dx = \ X ix^\

+
1
+

3!
1

2!

'
1
X* +

X*.

4!

T h e actual integral is i n this instance


^ = 2 + sin X ,

z cos X

X.

Existence of a Solution

93

I n power series form these formulas are


y

=2-[-X-~X^-\-~-^X^

2!

^4!""

'

'

6!

It is easy to see that approximations to these series were obtained.


PROBLEMS
21. Find the formulas for yi{x) and Zi(x) for the differential system

y - ^ - 1 ,

z'-2>-z-2,

with the conditions ;'(0) 2, z(0) = 0.


22. Find the formulas for;'8(x) and zi{x) for the differential system

with the conditions >'(0) = 1, z(0) = - 1 .


23. Find the formulas for ys{x) and z^^x) for the differential system

z'^y-2,

-z-1,
with the conditions ;r(0) = 3, z(0) = 1 .

24. Find the formulas for yh(x) and zeW for the differential system
y' = z/x,

z' ^ y -~ z,

with the conditions >(0) = 1, z(0) 0.

5.6. A proof of the existence of a solution


Thus far we have relied upon the geometrical evidence of the integral
curves for an assurance that a differential equation 5.1 has an integral
fulfilling a condition 5.2, even when no formula for such an mtegral c a n
be given. Subject to suitable hypotheses upon the function f{x, y) we
shall now prove that such an integral does i n fact exist. Such a proof
coidd be given by several different methods. W e shall give it by the
method of successive approximations.
T h e hypotheses upon /(x, y) shall be the following ones : that i n some
region of the (jt, y) plane of which (XQ, yo) is an interior point,
(5.12)

(i)
(ii)

\fix,y)\<M,
{fix,y) -

fix, y*)\

^N\y-y*l

O f these, relation (i), i n which M may be any positive constant, says .that
/(x,y) is bounded. Relation ii), i n which N may be any positive constant,
shall hold whenever both (x, y) and (x, y*) are points i n the region.

94

Approximations, Series, and Existence

Equation 5.12(ii) is known as a Lipschitz condition, i n honor of the G e r m a n


mathematician R . Lipschitz (1832-1903).
It is worth observing that any function having a partial derivative w i t h
respect to y fulfills a Lipschitz condition i n any region i n which fy{x, y) is
bounded, but does not fulfill such a condition i n a region i n which/^{Ar, y)
becomes infinite. This can be seen at once from the mean value relation
of the calculus
/(^.>)

-=fy{x,yx)\y-y*\,

i n which^1 is some value between^ a n d ^ * .


T h e relations
(5.13)

X(i\ ^

A,

y - ya\ ^

Mh,

confine the points {x, y) to a rectangle centered at (XQ, ^'O) - T h e size of the
rectangle is determined by h. It is small enough to lie within the region
i n which relations 5.12 apply, if A is small enough. We shall suppose such
a value of A to have been chosen, and shall henceforth refer to this rectangle
as R. I n it the range of variable x is
(5.14)
W e begin the proof by showing that all the graphs of the functions _y(x)
given by formula 5.11 remain within rectangle R for all x of interval 5.14.
This is easily done by mathematical induction. T h e fact is obviously so
for^o- W e need therefore only show further that if it is so f o r ^ _ i ( x ) it is
also so {oTyn{x). Suppose it is so for^n_i(x). T h e n \f{x,yn\{x))\ < A f ,
by (5.12), and, by (5.11),
(5.15)

yn{x) - yo

f^^ f{x,yn-i{x))

<

dx

X ~

XQ

Thus
which means that the graph of^n(x) is within R.
W e proceed to show now, again by mathematical induction, that for
every n
(5.16)

yn{x) -yn-x{x)\

<

MN--^

\X

XQ

This relation is valid when n = 1, by (5.15). Suppose it is so when


n = p. Since both the points (x, yp{x)) and (x, >'p_i(x)) are in R, it follows
from (5.12) and (5.16) that
J{x.yp{x)) -fix,yp-i{x))\

^ A ' i ( x ) ~yp-M\

< MN^

' "

"

Existence of a Solution

95

F r o m formula 5.11 taken with n / -f- 1, and with n =


yp+i{x) -yp{x)

= j^J{x,yp{x))

dx -

we see that
dx.

j^J{x,yp^i{x))

Thus
yp-\-\{x) -yp{x)

-J{x,y^-i{x))]dx

(MJVPW)
X

MN''

XQ

Herewith relation 5.16 has been established.


range 5.14, and for every n.
(5.17)

ynix) - yn-i(x)\

dx

<

= MNP

XQ

It follows that for all x o n

<

B y virtue of this result, the infinite series of functions


^"0 +

bnix)

-yn-l{x)\,

is seen to converge uniformly on interval 5.14. Let us denote by y{x) the


function to which the series converges. Since the sum of p terms of the
series reduces by cancellations to^p_i(jif), we have then the relation
l i m yp_i{x) = y{x)i

(5.18)

uniformly on (5.14).

CO

W e shall show that this function ^(x) is an integral of differential equation


5.1 and fulfills 5.2.
Since each function ^p_i(x) has a graph within rectangle R and is continuous, it follows from (5.18) that the graph of^(j:) is also within R and is
continuous. Therefore, by (5.12),
fix,yix))

- / ( x , : v p _ ] ( x ) ) | N\y{x) - j p - i ( x )

F r o m the relation
jy(^) - yo -

rfix,y{x))

dx = y(x) ~ y^ix) -

{f(x,y{x))

fix,yp^i{x))]

dx.

which reduces by cancellations to (5.11), it follows that


> W -yo

f^Myix))dx

>W

- ypix)\ 4X

y(x)

-yp-iix)\dx

96

ApproximationSj Series, and Existence

T h e right-hand member of the last equation is arbitrarily small when p is


sufficiently large, by (5.18). T h e left-hand member, which does not
depend on/>, is therefore zero. Thus^(x) fulfills equation 5.10, and, as we
have already seen, it therefore fulfills differential equation 5.1 and condition 5.2. By producing the integral ^(x) we have shown that the integral
exists.
T h i s same method of proof can be used to show that a differential system
5.4 has an integral fulfilling conditions 5.5, provided the functions fi{x,yy z)
and fiix, y, z) fulfill hypotheses analogous to (5.12).

57,

Uniqueness of the solution

For a differential equation that fulfills hypotheses 5.12, the integral for
which>(xo) = yo is unique. T o prove this, let both^(x) and rj(x) be such
integrals. Both are then functions that fulfill equation 5.10, and therefore
y{x) -

nix)

- /J

{f{x,y{x)) - / ( x , i,(x)) | dx.

{x, yix)) and (x, ri{x)) are points for which relation 5.12 (ii) applies upon

some X interval centered at XQ.


fix.y(x))

Thus

-f{x,v{x))\

^ N\y{x) -

v{x)

hence, for x ^ XQ,


\y{x) -r,{x)\

^ N

fj^\y(x) -r}{x)\dx

This relation implies that the function

Ll \yi^)

vix)\ dx

has a non-positive derivative. T h e function therefore does not increase


w i t h X . A t xo it has the value zero, and it clearly cannot be negative. It
is therefore identically zero, which means that |^(x) 7i(x)| = 0. T h u s
ri{x) is the same asjy(x).
Example 5 of Section 4.5 showed that the differential equation
y' = 2{x^

I P ' - 2x

has at least two integrals for which ^(2) = 3. T h i s differential equation must therefore fail to fulfill hypotheses 5.12, in a region containing the
point (2, 3). T h e function/(x, y) has in this instance a partial derivative

97

Answers to Problems

This partial derivative becomes infinite as (x, y) approaches the locus


-\- y \ = 0; hence the function f{xyy) fulfills no Lipschitz condition
in a region containing points of this locus. Since (2, 3) is such a point,
the failure of the hypotheses is evident.
ANSWERS TO ODD-NUMBERED PROBLEMS
1. yi{x) -

3 -

3. y,(x) =2x
5. yi(x)

3x +

+ 1_4
ix\

+ 4x' + i ^ ; c + ^ l V .

3 -

3U -

1| -

3U -

5{x

7. y^ix) - 4{* + 2}^ - I j x + 2}' +


^2 + x3

9. ^-i

i\. yi = X + x^ +

1
13.

ix^ + ix\

- X

yM
6*-' -

15.

2x

1
2!
1 -

;c +

1}" - - V U -

1|'

2\\

-\-x^

z i = x +

x^

X*.

1..3
ix\

3!

n!

2!

n!

x'

when rt is even.

when

x2 + |. ^3 _^ 2 ^6 _^ ^ ,7^
^ ^ sin jc cos jc.
7
m x\

17. yi(x) = l i + V ' + Vx'


> ^ + 12 x ) ^ .

T h e Taylor's scries of this coincides with ^8(;r) to terms

+ i^x' + ix* +

- 4*.

T h e Taylor's series coincide to terms in x*.

19. yz(x) = - + - log X - log^ X + log* x - ^ l o g ' x.


/
2
4**
48
480 ^
16,128 ^
1
21. >4 = 2 + x + :^x2 + - X ' ' + - X\
3!
2!
41
23.

n is odd.

= 3 -

^ x" +
x^
2
4!

26 =

24 =

2X +

- X'.
3

CHAPTER

6
Further Studies
of Differential Equations
of the First Order
The Riccati Equation
6.1. Singular solutions
A diflferential equation 5.1 may have a locus of singular points, and this
locus may fulfill the differential equation without being included i n the
general integral. T h e locus is then called a singular integral, or a singular
solution. O n the other hand a locus of singular points may not be a n
integral curve at all. Whether it is or not can be determined by testing i t
to see whether it fulfills the differential equation.
T h e function / ( x , y) i n the differential equation y'
[y x]^^ has a partial derivative fy{x, y) which becomes infinite o n the
locus ^ X 0. It therefore does not fulfill any Lipschitz condition i n
any neighborhood of this locus. Since the condition for uniqueness of a
solution of the differential equation is thus unfulfilled, the locus may be
suspected of containing singular points. T h e locus, however, is not, an
integral curve, for it does not fulfill the differential equation.
Example 1,

Example 2.

F o r the differential equation y' = 2x ~ Sly x^]^, the


locus y x^ = 0 may be suspected of containing singular points. It is
not included i n the general integral, which is
y = x'-

Wl"

<:]'.

Nevertheless it is an integral curve. It is therefore a singular integral.


T h r o u g h each of its points (xo, XQ^) passes, besides this singular integral
itself, also the particular integral
y =

x ' - W \ x - x o ] '

98

Singular Solutions

99

Figure 18 shows a number of these particular integral curves P and the


singular integral F .
Since two integral curves passing through a single point must be
tangent there, it is clear that a singular solution of a differential equation
is an envelope of the curves of the general solution.
PROBLEMS
For each of the following differential equations find a singular solution, and show
that there b also a particular integral through each of its points.

2y

3. y = i { x - V ' * ' - 4 y l

5. y

' \ / 3 7 + 7 - 3.

7. y = \/{x + 4 } b - 3 i .

9.

y =

,0. / =

-1

y=

4-

y =

6. >

8.

2 \ / c o s (x +
sin (x + ^)

1 - 2 cosM > +

tani > +

+1

- VJ

1.

r i

100

Further Equations of the First Order

6.2. Factorable equations


A difTerential equation of the first order need not be of the simple form
5.1, but may be more generally
(6.1)

F{x,y,/)=0.

T h i s may not be explicitly reducible to form 5.1 at all, or it may be reducible to a number of such equations. I n the latter case it assigns a number
of different slopes to a point (x, >), and the integral curves therefore constitute a net of intersecting loci rather than just a simple family.
Example 3.

T h e differential equation
y a _ 2yy' -

-1-^2

= 0

reduces to the two equations y' = y -\- I and y' = y \. It therefore


defines two slopes at any point, and all curves of the two intersecting
families y ~ ce' -\- \ and y = ce' ~~ \ are integral curves.
W h e n the function F(x, y, y') is factorable, and has the factors Fi{x, y, y'),
^zixj y, y'), ' , ^^(x, y, y'), a l l the solutions of the several differential
equations
(6.2)

Fiix,y,y')

= 0,

F2{x,y,y')

=0,

- ,

Ft(x,y,/)

= 0

are solutions of equation 6.1; and conversely, every solution of (6.1) is a


solution of some equation 6.2. If the general integrals of equations 6.2
are, respectively,
'Pi{x,yyc) = 0,

'P2{x,yyc) = 0 ,

tpk{x,y,c) 0,

the general integral of (6.1) can be expressed as


(6.3)

Wi{xyy,c)\{<p2{x,y,c)]

=0,

\*Pkix,yyC)\

for this product is zero if and only if one of its factors is so.
Example 4.

T o find the general integral of the differential equation


yy'^ -

{xyhy^-\-

x]/

+ x^ + xy -

0.

T h e left-hand member of this equation is factorable into the form


{/

-x-y]\yy'

x\

=0.

T h e given equation is therefore reducible to the equations y^ x -\- y,


yy' = X . Its general integral is, accordingly,
{e-^lx -^y'\-\\~c\\x^-y^-c\=

0.

Singular Solutions Again


Example 5.

101

T o find the general integral of the differential equation


i / ^

_^2j,v'

_ xy'^ +

xy^ =

0.

T h e factors of the left-hand member of this equation yield the equations


/

- : v = 0,

y'-\-y

gy-

= 0,

X ^

0.

T h e last of these may be put into the form y' = log x.


integral is thus
c\ lye' c\\y x\og X

T h e general

c] = 0 .

PROBLEMS
Find the general integrals of the following differential equations.
11. xy'^ 13.

{Ax'' + 1 1 / + 4x = 0.

y + y -

15. 2 x V ' 16.

X^

'^^x

12. e'yy''^ +
14. y - +

0.

+ 12

21..'
+ 2x^> + V t y +

[xb

+/I

U= + 2>}y2 + |2x> + [x= + 2>l[y - r']]y'

17. y'^ + 2xy' +x^l


19. y ^ -

-4/}

log

|4*^ - y'^]y' X

xlogx

-\ = 0,

= 0.
+ 2;o'l> " ^'""1 = 0.

18. y ^ - 4y2 -

= 0.

Ay = 0.

2 i y = o.

5 / = 0.

20. y i o g y - x y ' +

[> - / ! l o g y - x ; - ! ! -y\

=0.

6.3. Singular solutions again


A diff"erential equation 6.1 may have a singular solution, for its general
integral, or some part of it, may have an envelope which is not itself a
curve of that net. A t the points of a n envelope the differential equation is
fulfilled by fewer directions than at some neighboring points. This can
be seen from F i g . 19, in which the circles are the curves of a general solution. T h e i r two common tangents are envelopes, and hence are singular
solutions. A t a point Pi the differential equation is fulfilled by two directions which make the angle 6i with each other. A t a point P2 nearer the
envelope the angle 62 is smaller, and as the point approaches the envelope
at P the angle between the two directions reduces to zero; that is to say,
the two directions reduce to one.
T h e envelope is thus included among such loci as there may be upon
w h i c h the differential equation defines a reduced number of slopes,
namely, upon which it has a reduced number of roots y'. T h e condition
for that is
(6.4)

dF(x,y,y')

ay

0.

102

Further Equations of the First Order

T h e liminant oiy' between this equation and (6.1) includes any existing
envelope. It cannot be assumed that every locus given by the liminant
is an envelope, for there are other loci also upon which the differential
equation defines a reduced number of directions. T h e line of centers of
the circles in F i g . 19 illustrates this. O n this line only the perpendicular
direction fulfills the differential equation. Each locus obtained from the

FiO. 19.
liminant must therefore be tested to determine whether or not it is a
solution.
Example

6.

T h e differential equation for the family of circles i n F i g . 19

is
\x -

Viy\ny'^

\\ -

{x+yy'\^

= 0.

For this equation, equation 6.4 is


=0,

{x-Vly]^y'-y\x+y/]

and the y' liminant is


y\y -

Vlx][Vly

x]'' = 0.

T h e liminant thus gives the three l o c i ^ = 0, )" =


y = (l/\/3)x.
B y test in the differential equation the third of these loci is found not to be
a solution. It is the line of centers. T h e other two loci are solutions.

Integrations by Differentiation

103

T h e y are singular solutions, because the general solution

does not include them.


Example 7.

F o r the differential equation


{ / - ^ 2 i y + / + i =0,

y 2 _

the y' liminant is = 0. This does not fulfill the differential equation.
Thus there is no singular solution.
PROBLEMS
Find a singular solution for each one of the following differential equations.
21. / y ' - W - ' + T b ' - ' l

22. {3y -7\y^

=0.

23.

{xy' - ; y p - y 2 -

25.

{x^ - y\y''^ - xy' -\-x^ = ^.

-4{y

-2]

=0.

24. xy'* - 2yy'^ + 2x^ = 0.

1 =0.

26. jc^"' - /

+ ^ -

2 = 0

27. y 2 -

4xy' + 4> = 0.

28. >' -

29. xy' - y' cos~' (y') + V l 30. f i + / i

log \xy' - y\ - 1.

> = 0.

log {1 + y i - y i i -t-xj 4 - y -

6.4. Integrations by differentiation.

1 = 0,

The Clairaut equation

B y definition a solution of a differential equation 6.1 is a function ^(x)


tor which the expression F{x, y{x), y'ix)) is identically zero. T h e derivative of this expression is therefore also zero, that is,
F x ( ^ , > W , > ' W ) + F^{x,y{x-),y'{x-))y'{x) + F^{x,y{x),y'{x))

"

= 0.

This signifies \S\aXy{x) is a solution of the equation


(6.5)

F^yy.y')

+ Fy{x,y,y')y'

- f F^{x,y,y')

= 0,

and that it is therefore a solution of simultaneous equations 6.1 and 6.5.


T h i s fact can sometimes be used to accomplish the integration of equation
6.1.
If >' is determinable i n any manner from equation 6.5, the substitution
of that value for^i' i n equation 6.1 converts that equation into an integral.
W h e n equation 6.5 involves too many variables to permit a determination of^', we may eliminate^ between equations 6.1 and 6.5. A value of
y' determined from the;) liminant also converts (6.1) into an integral.

104
Example 8.

Further Equations of the First Order


F o r the differential equation
2/ + > -

2 / logy = 0,

equation 6.5 is
y' -

0.

2 log/idy'/dx)

This is a differential equation f o r ^ ' whose integral is *

= jlog^y')^.

F r o m this r e s u l t / = e^^^^^'y and / = e~^^^''. These values convert the


given differential equation into the respective integrals

and
Example 9.

2e^'l\

2e~^^'{\

Vx

c]y = 0

c\y = 0.

F o r the differential equation


2yy'^ + 2x^ = 0,

xy'^ -

equation (6.5) is
=0.

[4xy'' -6yy''\^+{6x^-y"]
dx

There are too many variables in the last equation to permit a determina
tion oy'. T h e liminant of y between the two equations is
dy'
'Tx

T h e factors of the last equation give, respectively, y' = ex, and y'^ = 6x^.
These values convert the given differential equation into, the general
integral
c^x^ -

2c^y + 2 - 0 ,

and the singular integral


9y^ = 4 V 6 ; f 3 _

There are two types of differential equation 6.1 to which this method is
peculiarly well adapted.
(i) T h e equation that can be made explicit for y, namely,
(6.6)

> =/(^,/);

(ii) T h e equation i n which x and y occur only i n the combination


[xy' - y\, namely,
(6.7)

f(x/

-y,y')

=0.

A n equation 6.7 is called a Clairaut equation i n honor of the French mathematician A . C . Clairaut (1713-1765).

Differentiation with Respect to y

105

For a Clairaut equation, equation 6.5 is simply {dy'/dx) Fj/(xyy, y') = 0.


T h e first factor of the last equation gives y' = c. A Clairaut equation is
thus converted into its general integral by merely replacing y' by c.
Example 10.

T h e differential equation
\ =0,

W - y \ ^ - y ^ -

is a Clairaut equation. Its general integral is therefore \cx y\^


1=0.
Equation 6.5 is i n this instance

dx

Uy

-y'

-"y]

=0.

T h e second factor of the last equation gives ^' = xy/jjc^ ~


this
value of y' the differential equation is converted into the singular integral
x'^^y''

= 1.

PROBLEMS
Solve the following differential equations.
31. xy' -y

+ / 2 = 0.

32. 1 6 / -

{xy' -y\^

33. 2y - A ~ log {y^+ 1} = 0.

34.

35. xy'^ - yy'"^ + 2 = 0 .

36. / 2 Jr\x-

37. 2y - 1

38. y

V^y"" + 1
39. y< - x.2..2
y + "^.xyy' - y

0.

y log y' = 0.
\\y' -y-\.

1 = 0.

-Ixy'
.'2 -

40. \zx - 1 l y ^ - 2{> -

41. 2x>./2 = 4 x V ' + 1.

43. 16{:c + 1

- 0.

1 l y + 4 = 0,

42. i l -\-r^']y = (y - 1]*.

- 4{4> - ! } / - !

= 0.

44. ^ + 2 + 2 1 o g (y/2} - xy' = 0.

45. Axy - 2xy'^ + ^

= 0,

46. y * + l / + > } { 4 y 2 + l | - 0 .

6.5. The method of differentiation with respect to /


If we think of ^ as the independent variable, and recall that dx/dy = 1 / y ' ,
the rccisoning of Section 6.4 shows that any solution of a differential equation 6.1 is also a solution of the equation obtained by the differentiation of
(6.1) with respect t o ^ , namely, of the equation
(6.8)

FAx,y,y')

~ + Fy{x,y,/)
y

+ FAx^y,/)

7^ = 0.
^y

A n y value o f y that can be determined from (6.8) converts (6.1) into an


integral.

106

Further Equations of the First Order

If y' cannot be obtained from (6.8) because too many variables are
involved, we may eliminate x between equations 6.1 and 6.8. A value
of ^' obtained from the x liminant also converts (6.1) into a n integral.
T h i s method is peculiarly well adapted to a differential equation that can
be made explicit for x, namely,
(6.9)

^ =/(>,>').

Example 11.

T o integrate the differential equation


ilog{l

- l o g / - x

+ 2 = 0.

T h e derivative of the given equation with respect tojy is


1

dy'
r 2 - ~

dy

+ i

and integrates to give y' = tan {c y\.


differential equation into its integral
log sin

^1

(c

=0,

This value converts the given

2 = 0.

Example 12. T o integrate the differential equation

| x 2 - 2 x l { y 2 + l) + 1 = 0 .
T h e y derivative of the given equation is
. .
lv'^ + 1
{x^-2xiy-<- + | x - i !

= 0,

dy

and the liminant of x between the two equations is


'2

+ 1

d/

-, + V y 2 + 1 = 0

F r o m this we obtain y' = v {1 + (y + f^) ^ / 1 ^ + c}, and these values


convert the given differential equation into its general integral
x^ - 2x +

= 0.

PROBLEMS
Solve the following differential equations.
47. 2xy' = y -

yy'\

49. 2xyy' = y'^ + 4y~.


51. xy' + 2 log /

+ 2 = 0.

48. x = 25>V" + ~ 5y
50. x = 5 + log \y' + y/

+ y'^\

The Riccati Equation

107

52. 3 / V l - JV'^ + 3 s i n - i 1/1 + 2JC = 0.


+ Sxy' -

53.

2> = 0.

54.

dAxy' + 32> = 0.

.. The Riccati equation


A differential equation
(6.10)

= io(^) + ii(^)> + ? 2 ( x ) / ,

i n which q2{x) ^ 0, is called a Riccati equation, i n honor of the Italian


mathematician Count Riccati (1676-1754). T h e reciprocal of any solution of this equation, w = l / y , is a solution of the associated equation
(6.11)

w' =

-q^ix)

qi{x)w qo{x)w^.

T h u s the integrals of either one of these differential equations c a n be


obtained by accomplishing the integration of the other equation.
A n equation 6.10, or 6.11, may be of a type that is integrable by a
method of Chapter 2. W h e n put into form 2.1, it may be linear, or it
may have separable variables, or it may have coefficients that are homogeneous and of the same degree. Except for these, and a few other rare
elementar)' cases, the Riccati equation is not integrable by any systematic
method other than that of power series.
It is sometimes possible to find a solution of a Riccati equation by trial.
One may try such functions as ^ = ax^, or y = a^^, with undetermined
constants a and h, or some other function that the appearance of the
differential equation itself suggests. Success i n finding a solution is
rewarding, for, if any integral can be found, the general integral is obtainable from it.
Suppose ^I(x) is any particular integral of di"erential equation 6.10.
T h e change of variable
(6.12)

y =yi{x)

with an arbitrary constant cu transforms the equation into


A _L ' ^ i " '
_L
r \
yi {x)
= qo-\- qiyiix)
u

'^^ii ^
2/ ^
+ 9 ^ 1 [x)

2ciq2yi{x)
a

^
u

B y such cancellations as can be made because yi{x) fulfills (6.10), this


transformed equation reduces to
+

{qi + 2q2yi\u = ciq2.

This differential equation for u is linear.


of a form

Its general solution is therefore

U = Ci<pi{x) + C2<p2{x),

108

Further Equations of the First Order

i n which tpi{x) and <p2{x) are specific functions, and cz is a second arbitrary
constant. B y (6.12), then,
(6.13)

y ^ y i i x ) -

Cl(pl{x) + C2<P2(x)

This is the general solution. A n y particular solution obtainable from it is


determined by the ratio of ci to C2* This ratio is therefore the single
arbitrary constant which formula 6.13 actually involves.
Example 13.

T o integrate the R i c c a t i equation

y = j1 +

X +

2x^ cos A: I

{1 + 4* cos

+ 2y^ cos x.

By the trial of> = ax^ it is found that the given equation h a s ^ i = * as a


solution. T h e change of variable y = x Ci/u transforms the equation
into
u' u = 2c\ cos X .
T h e general integral of the last equation is
u = ci{sin X cos x\ + C2^\
Hence the general solution of the given equation is
c\
>

Example 14.

f T l j s i n X cos

x|

T o integrate the R i c c a t i equation


/

=e^'+

|1

+^e'\y^y\

By the trial oiy = ae^' i t is found that


general solution as found by the use of
is

^e' is a solution.

c,e^

1^1 + c^e-^*'^^

The

A solution found by trial may be complex, say,


y, = a{x) + ;-^(x),
with i =
is then

1.

(6.14)

T h e general solution, expressed wholly i n real terms,

y = a(x) + ^(x) tan {jq2{x)0{x)

dx-\-c].

This may be proved as follows: Since a + i/3 is a solution, we have


a' +

i^' = ?o + qiW

+ m

+ 921^ + 2ia^ -

^'}.

109

The Riccati Equation

F r o m the last equation we see, by equating real and imaginary parts, that
(6.15)

a' =

90 +

qxct + q^la^ -

/S^),

^' = q,^ + 2 ^ 2 ^ .

T h e result of substituting evaluation 6.14 into equation 6.10 is


a' + /3' tan 6 + yaS^ sec^ 0 = ?o +

+ q\& tan B + q^oi^ +


2?2a/3 tan B + jaS'^ tan^ ^,

where B = {fq2{x)^{x) dx -\- c}.


6.15.
Example 15.

This is fulfilled by virtue of relations

T o integrate the Riccati equation


/

= 1 -y

e^y.

By the trial of ^ = ae^' the given equation is found to have the solution
yi = i>~*. T h i s is of the form above with a = 0, and /3 = e~'' T h e
general solution, by (6.14), is thus
y e~^ tan {e^ + c\.

PROBLEMS
Find the general integrals of the following differential equations.
55. /

V
1
5 + 4 ^ + - ^ .

3^ cos x\ Ixy cos x

56.

cos x.

= 3f*' -Vly - 12>^

57.

58. xy' => [ 6 - 4 log x} -

[5 -

4 log x\y + (1 - log x j / .

59. _y' = 3x' + - + 3x).2.


X

61. /

2 - 2x'

= 2x* +

y -Vy

^ - 1

60. ^'

62.

=> sin^ X + _)> cot X

Find the integrals of the following differential equations that fulfill the given
conditions.

63. y'

2
+X

64. y = 9x + {3 65.

66. >' -

I S x b + (9x -

tan

-7

X +

sin

y{-\)

= ^

..2

1 +x^

31/,

x l l +x^j

cos X

'

>(1) = 0.

- y"^ cot X ,

;y ( -

110
67.

Further Equations of the First Order


- 5** + 6*^7 + 2**/,

68. y' "x'

69. /

-\-

= tan

-f.

X +
S i n jr

70.

;r(0) -

cos Jc

+ /

^-(0)

cot x,

( - ^) =
\
4/

-1.

6.7. Some properties of the integrals of Riccati equations


If xo is any point of an interval
(6.16)

a-^x-^b,

upon which ?o(*), ?i(*) and 5 2 ( ^ ) are continuous functions, the point
( ' 0 ) ^ 0 ) with any ordinate ^yo lies within some region i n which the function
i{x,y)

= qo(x) +

qi(x)y + q2{x)y^

and its partial derivative


y) are bounded. T h e function therefore
fulfills a Lipschitz condition 5.12, and accordingly every point (xo, ^o) has
just one integral curve passing through it. A Riccati equation therefore
has no singular solutions.
A R i c c a t i equation may have some integrals that are continuous over
the whole of an interval 6.16. However, it always has some that are discontinuous, a fact that is easily seen. For, with any chosen XQ, differential
equation 6.11 has an integral ;o(^) for which wo{xo) = 0. T h e reciprocal
of wo{x) is an integral of equation 6.10, and it obviously has the vertical
line X = xo as an asymptote.
Example 16.

T o find the solution of the Riccati equation

which is discontinuous at x = 1.
T h e differential equation has constant coefficients.
constant solution, and by trial this is found to bejJi
solution, as found by the use of^"! = 2, is
y = 2

It therefore has a
2. T h e general

T h i s is discontinuous at x = 1 if ci + 12^ = 0.
thus
-

T h e required solution is

Integrals of Riccati Equations

111

T h e R i c c a t i equation has important relations with the equation


(6.17)

pMu"

+ pi{x)u' + p2{x)u - 0,

which is the linear differential equation of the second order. Thus if i;(x) is any
solution of the R i c c a t i equation

Po{x}^ PQ{X)

on a n interval upon which po{x) ^ 0, the formula


(6.19)

u = el"^'^'^'

gives a solution of equation 6.17. Therefore equation 6.17 can be solved


by solving equation 6.18.
Conversely, if v{x) is any solution of the differential equation
(6.20)

v" -

v' + qo{x)qi{x)v = 0,

qiix) H 7 qiix)

which is of type 6.17, the formula


(6.21)

q2{x)v{x)

gives a solution of Riccati equation 6.10.


Example 17.

T o find a solution of the differential equation


u" -

{x +

l j u '+

{x -

lju =

0.

T h e related Riccati equation 6.18 is


r,' = ~{x-\]

{x-\-\U

- ri\

It has 71 = X as a solution. By formula 6.19 the given differential equation


therefore has a solution u = e'^^^.
Example 18.

T o find a solution of the R i c c a t i equation

2H-X
x[l -\- x\

2
x\\

x^

12

T h e related equation 6.20 is

x{\ -\-x]

x\\ + x l

T h i s has f =
as a solution. By formula 6.21 the given equation therefore has a solution)' = 1 / | 1 +

112

Further Equations of the First Order


PROBLEMS

71. Find the solution of the equation


1

2
x

which 15 discontinuous at * = 4.
72. F i n d the solution of the equation
3
4 - -

1
2 - -

which is discontinuous at x

73. Find the solution of the equation

for which y{0) -a".

Determine its discontinuities, if any.

74. Find the solution of the equation


x'y + y\
for which ^(1) = 0.

Determine its discontinuities, if any,

75. Find a solution of the differential equation

76. Find a solution of the differential equation


u " + u' log X + -

=0.

11. Find a solution of the differential equation


u " u' cos X + u sin

0.
x

78. Find a solution of the differential equation


u"

2u' tan X

0.

ANSWERS TO ODD-NUMBERED PROBLEMS


1, y' = X-

T h e general solution is 5

3- x^ 4^ = 0.
5. ^ = 3x,
7. ^ + 3.

x}^^ = x^ + f.

T h e general solution is \ / x ^ 4)^ = x + c-

The general solution is 2 ' s / 3 x + ^ = log x +

T h e general solution is 3 y/y

3 =

9. ^ = (n + ^jw X, with any whole number n.


The general solution is

cos (A: +

= x + c.

+ {x + 4 p

Answers to Problems
{y - 2x^ - c][y - log X - c\

11.
13.

xy --x^

log

V-^-c

15.

h 4 log X
y

- c

2x
X

y -

113

0.
= 0,

= 0

11 J *

19. y\21y ~ 5(x


23.

= 0.

2S.y

= 1.

x2

- > = 0.

31.

^ + tf* " 0,

and ^ =

^ = 2 log cos \x + 1,

4;- + Ix -

= 0,

39. c* - c^x^ 4- 2xy 41. ^

-2^^

+ ^

= 2cx + c\

51.

59. jvi

IX,

61. yi

{1 + i l x ,

and 2y^ = x.
49. 16^ = c(2x - c\\
2\/y

y =

3x -

57. :y
fi +

tan {x^ + c].


jv = x 2

+x2tan

{-i*'
65. > =

11

- 3

69. y

~ X

* + l
tan

cot

X +

-8x
X -

4'
1

> = ' ' + ? T - i =* cxp

and 27^ = 2x*.

+ c + 2] = 0 .

67. > - 7 tan f*' -

and 14;^ + 1 p + 4|x + 1} = 0.

and 27>'* + 125x' = 0.

3 x - x 2

75.

0,

Scix^

-2x -

71. >

1 -

+ 2 H x * - v ^ -

=0,

55. y -

63.

1} -

= 0,

f ^ f V ^ + 2 \/>

3 +

and > = 1

and 1 6 / = xV

and > - 0.

53. 5 - 2 / ^ +

73

- 0,

4c{4y -

43. 16c^{x + 1} -

47.

and y = 0,

K> and 2;' = 27x.

2c - s / x

45. 4c^ Vx

and y = 2.

and 2y^ = 27x^.

35. c^x - c > + 2 = 0,


37. 4 / -

= x ' ~ l

29. y = sin x.

27.

33.

=x-V\.

21.

It has no discontinuitiefl.
77. u

lin X

1:2^'''*'

CHAPTER

7
Some Solvable
Differential Equations
of the Second Order
Approximations.

Applications

7.1. DifFerential equations of the second order


A relation
(7.1)

which actually involves^", is an ordinary differential equation of the second


order. T h e simplest type of such equation is that of the form
y"

fW-

B y successive quadratures the last equation yields

and
i n which ci and C2 are arbitrary constants. A solution of a differential
equation of the second order is called the general solution, or the general
integral, if it involves two arbitrary constants i n such a way that these constants are not replaceable by just a single constant.
A n y differential equation of the first order that is obtainable from the
given equation of the second order by an integration, and that involves a n
arbitrary constant, is called a first integral. W h e n a first integral c a n
be found, the problem of integrating the original equation is reduced to
that of integrating the first integral. F o r that, the methods and observations of the preceding chapters can be drawn upon. W e shall consider a
few types of difi"erential equations 7.1 for which first integrals can be found.
114

Equations in Which Either ^ or ;c are not Present

115

7.2. Equations in which either / or x are not present


A differential equation 7.1 in which y is not present, namely, F{x, y\y")
= 0, is i n effect F(x, y\ dy'/dx) = 0, that is, it is a differential equation of
the first order for^'. A n integration of the equation is therefore possible
and yields a first integral f{x, y\ c) = 0.
Example 1.

T o integrate the differential equation


\x + 2]y''-\-y'^

1 - 0 .

As a differential equation for^' the given equation has separable variables. A n integration gives

y + 1

that is,

> = 1 +

T h i s is a first integral.
integral

{x-\-2]'-ci

B y a quadrature of the last equation the general

[x -\- 2 + ci

+ C2

is obtained.
A differential equation 7.1 in which x is not present can be treated quite
similarly. Since
dy dy'
dx

dy

,dy'
dy

the equation Fiy, y\y") = 0 is, i n effect,


y\y' dy'/dy) = 0. This is a
differential equation of the first order for^' as a function of;'. A n integration of it as such yields a first integral f{y, y', c) = 0.
Example 2.

T o integrate the differential equation


y"2

yy"
y

By setting^" = y' dy'/dy, the form of this equation is changed into


dl
dy

dy

This is a Clairaut equation 6.7 f o r ^ ' as a function oiy. It therefore has


the first integral c\ + c i j >' = 0, and the singular integral^'
i:)'^.

116

Solvable Equations of the Second Order

Integrations of these relations yield the general integral y


and the further integral) = 4/jt + f j .

= 2^^'

i,

PROBLEMS
Find the general integrals of the following differential equations
1. xy" + y - * -= 0.

2. xy" -

3. y" - /

4. yy" + {1 + 2 / | / 2 -

V l +

5. xy'' + xy''' + /
7.

= 0.

'

2yy' = 0.

y \y' = 0.

8. 1 + xy" - y' ~ y"^ = 0 .


/

A\ - 0.

6. {;y + 1 b " + {> + 2 b ' 2 - 0.

cos ^ + sin ^ \y" + {2 cos y y

10. y 3 _ ^ y y ' +

{1 + x'^Wy'

9. 1
"

- / ^ J - 4 / / = = 0.

= 0.

7.3. The differential equation of a 2-parameter family of curves.


Checking a general solution
A family of curves whose equation
(7.2)

^{x,y,cuC2)

involves two arbitrary constants is called a two-parameter family. A n y curve


of this family characterizes)' as a function)'(x), for which the expression
^{.x, y{x), c\y C2) is identically zero. T h e first and second derivatives of
this expression are therefore also zero. Hence y{x) also fulfills the
equations
^x{xyyy ci, C2) + ^y{x,y, c i , C2)y' = 0,

(7.3)
*z.x(^,:y, ci, C2) + 24>x.v(^,>, cu C2)y' H- %,v{x,y,

2)/' = 0,

that is, y \s a solution of simultaneous equations 7.2 and 7.3. T h e liminant of ci and C2 from these three equations is the differential equation of
family 7.2.
T h e final step of elimination i n this finding of the differential equation
can be avoided by arranging the several equations so that the differentiations themselves bring about the eliminations. Thus if equation 7.2 c a n
be made explicit for one of the constants, that constant is eliminated by the
differentiation, a n d does not appear i n the first equation 7.3. If that
equation can then be made explicit for the remaining constant, the second
differentiation completes the elimination. T h e final equation 7.3 is then
the differential equation of the family.
Example 3.

T o find the differential equation for the family of a l l circles


having the radius 3.

The Differential Equation of a Family of Curves

117

T h e coordinate equation of this family is


(7.4)

= 9,

\y-C2]'

and the derived equations 7.3 are thus

= 0.

{y-c2]y"

T h e liminant of ci and cz from these three relations is the differential


equation
(7.5)

9/'2_ {l+y2j3^o

If equation 7.4 is first made explicit for C2, thus,

the first differentiation yields

V 9 -

{x-

ci\

If the last equation is now made explicit for ci, thus, x 3y'/y/l
y'^
ci, the second differentiation yields, without any further step, the result

(1

+y'T

=1.

T h i s is, of course, the same as result 7.5.


T h e graph of the general integral of a differential equation 7.1 is a twoparameter family of curves. T h e correctness of an integration of such a n
equation m a y therefore by checked by using the integral obtained to
re-derive the differential equation.
Example 4.

T o check that y^
of the differential equation

cix^y

cz = 0 is the general integral

T h e derivative of the relation given is


2y/

ci{xy2xy}

= 0.

If this is first made explicit for c i , the second differentiation yields

118

Solvable Equations of the Second Order

T h e denominator may be dropped. Since the result is then the given


differential equation, the questioned relation hcis been checked to be a n
integral.
PROBLEMS
F i n d the differential equations of the following families of curves.
11. ylogy

=cix +C2.

13. y -

12. y =- {cix

+ C2\^.

15. ciy + Ci+

log \xy\ + 0.

14. y -cie'

+ C2X^ -

16.

17. log
19. 7 -

+/}
X -

21. [xy]*

= ci t a n - i (y/x) + C2.

18.

+ ci\x - y\ + <:2 = 0.

2cix -

0.

2ciy ~ 0.

- ^

20. y =

cie'**.

+^2}^

+ 1 -

0.

,1-io

22. Aaxy-\-c^y/x

"

0,

ASSIGNMENT
1. Check the general integrals of the differential equations of odd-numbered problems 1 to 9.

7.4. Exact equations


A differential equation 7.1 is said to be exact if its left-hand member is
exactly the derivative of some function
y, y'). Because such a derivative is linear i n y", only a differential equation of the form
Pix,yyy)y"

-hQixyy,/)

=0,

that is,
(7.6)

P(xyyy/)

dy' + Qix^y^y') dx = 0,

can be exact. W e shall derive the condition under which such an equation is exact.
Let <p{x, y, y') be defined by the formula
(7.7)

iP(x,y,y')dy'

^v{x,y,y'),

i n which the quadrature is made as though y were the only variable, that
is, as though x and y were constants. T h e n Pix^y^y') = <py'{x,yty'),
which is to say that
Kxyy^y')

dy' = dip {x,y,y')

iPx(x,y,y') dx -

tpyix, y,y') dy.

Equation 7.6 may then be written as


(7.8)

d,p(x,y,y')

{Q(x,y,y')dx

tp,ix,y,y')

dx -

<Py{x,y,y') dy\ =

0.

Exact Equations

119

This is exact if, and only if, the expression within the brace is a differential,
that is, if, when y' is replaced by dy/dx, the differential equation of the
first order obtained by omitting the first term of (7.8) is exact. T h e test
for such exactness, and the integration of the equation into the form
V'i(^) y) = c if it is exact, are given i n Section 2.2. If differential equation
7.6 is exact, therefore, it has the first integral
w
<f>{x,y,y') + <p\{x,y) = ci.
Example

5. T o integrate the differential equation


\2y -

\ }y" +

+ /

- 2 > / - 2 * - 2e'] = 0.

F o r m u l a 7.7 gives (p(xy y, y') = 12>' 1 \y\ and with this evaluation
the given differential equation can be written i n the form
d\{2y -

\)y'] + ! / ~ V

- 2xe' ~ 2e'] dx = 0.

B y omitting the first term of the last equation and replacing)"' by dy/dx^
we obtain the equation
-

{2xe' + 2e'\ dx -

\2y -

1\ dy ^ 0.

This is exact and has the integral


2xe'

y^

= ci.

T h e given differential equation therefore has the first integral


{2y ~

\]y' -

2 x e ' y

y^ = ci.

This c a n be integrated again. T h e change of variable u = y^ y


transforms the last equation into u' u = ci
2xe'j which is linear.
T h e general integral of the given differential equation is thus found to be
y

e* -\- c%e

C\.

PROBLEMS
Determine which of the following differential equations are exacts and find the
general integrals of those that arc.
23. {1 -x^\y"
24.

^\\

- 3 x } y - : y = 0.
^

{1 + x\e = 0.

25. xy" + {3x/ + 2 i / + { 2 / + 2 }

=0.

26. y" sin x + Isec x + 2> cos x]y' + tan x = 0.


27. {x co&y + sin x\y" xy''^ s i n ^ + 2{cos^ + cos x\y' ^ sin x + 0.

120

Solvable Equations of the Second Order

28. xyy" + xy'"^ + yy' - - - 0.


X

1 + xy
30.

2y"
yy

31.

y'
log
y

1+
_\

y"

xy

xy

= 0,

+ - - 2 = 0 .
y

32. y" cos ^ + {cos^ - y' tin. y\y' 2x = 0.

7.5. Simultaneous equations


A differential equation
(7.9)

y"

= Six, y, y'),

is replaceable, i n an unlimited number of ways, by a pair of simultaneous


differential equations of the first order. L e t / i ( x , y, z) be chosen as any
function that actually involves z, and set^' =
y, z). T h i s last equation, its derivative equation, and equation 7.9 together make up a system
of three equations from w h i c h
and y" can (theoretically) be eliminated.
T h e liminant is an equation z' = /2(x, y, z). Equation 7.9 has thus
been replaced by the differential system
(7.10)

^fiix,y,z),

z'

^h{x,y,zY

O n e particular way of making such a replacement is to take / i = z.


T h e system obtained is then
y' = z,
Example 6.

z' ^ f{x,y, z).

T o replace the differential equation


y" cosy + {cos;; y' sin;)^' 2xy = 0

by a differential system.
T h e choice y' = z gives the derivative equation y'' = z'. and the elimination of^' andy" between these and the given differential equation yields
the liminant
z' cosy + z cosy z^ sin y 2xy = 0.

T h e system obtained is thus


y' = z,

z' 2xy sec y +

tan y z.

T h e alternative choice, y' = z sec y^ replaces the differential equation


by the system
y' = z sec y,

z' = 2xy z.

Simultaneous Equations
Example 7.

121

T o replace the differential equation

1+

1 + xy

xy

by a differential system, by setting 1 + x^y'] = z.


T h e derivative of this assigned relation is {x'^y" + 2xy'\ = z', and the
liminant o y' and y" from the three equations is {yz'/xz) log z = 0.
T h e system obtained is thus
y

= ^
X

xz

z = log z.
y

Conversely, any differential system 7.10 is replaceable by a single differential equation of the second order. T h e two equations 7.10 and the
derivative of the first of them constitute a system of three equations from
which z and z' can (theoretically) be eliminated. T h e liminant is a
differential equation 7.9 for y. If this equation is integrated, y and y'
are thereby determined. T h e first equation 7.10 is then an equation for
2. A system 7.10 can thus be integrated by replacing it by a differential
equation 7.9 and integrating that equation 7.9.
Example 8.

T o integrate the differential system


y = 2 - 3,

z' =

-z^

3x+> + l

T h e derivative of the first of these equations is j " = z', and the liminant
of z and z' from the three equations is
l 3 x + : v + l i / ' + ( 3 - / | 2 = 0.
This differential equation of the second order is integrable because it is
exact. A first integral of it is
!3x+>4-ii!3-y!

=cu

and its general integral is


{3x +JV +

1)2 =

2cix^-C2-

B y these relations )? andjy' are determined. Therewith the first one of the
given equations determines z. T h e solution of the given system so found
is

122

Solvable Equations of the Second Order


PROBLEMS

I n t r a t e the following d i
y - z.

33.

34. y'
35. /
36.

z' - 2^-" - z\
y sin X z cot

sin X
- z - 2*,
-

systems.

z' -

- ( l / x ) { 2 ; f + 2* + 4 )

>+2

37. y = z

' ^ - ^ + s e c {z - > .

38. ; r ' = e\

z' = " " .

Replace each of the following differential equations by a differential system, by


using the respectively assigned relation.
x'-y'

39. y

40.

41.

/ 2 + (2> - 1 ly** - y {1 + ^ ' ) -

y' -\-xy'' -2x

y
j 1 + tan

42. y" -y

' - ^ -^

43. y" - yr^'


44. y" ~y

y' = x^ - yz.

+ cot x\y' y tan x 0,

=0,

- 2 = 0,

V i - ,

= 0,

^' = z tan x.

= 0, y - y/* + z.

=;r + z + l.

=. log z.

j'''^ eos > =" 0,

sin z.

7.. Solutions in power seres.

Polynomial approximations

W h e n the function /(x,


y') is representable by its Taylor*s series about
the point (xo, >o, >o'), the integral of differential equation 7.9, for w h i c h
(7.11)

y(xij) =:>o,

y'(xo)

yoy

can be obtained as a power series, Polynomial approximations of the


integral are thus also obtainable. T h e method is that of Section 5.3.
W e seek an integral i n the form
7

= flo +

ailx

XQ] 4-

azlx

XQ]^ +

asix

XQ]^

Conditions 7.11 require that Jo = >o, and ai yo'. T h e remaining


coefficients are determinable by substituting the series (ory into the differential equation, a n d equating the coefficients of like powers of {x xo].
Example

9.

T o find as a power series the integral of the differential

equation
y"

, y

/2

Approximate Integral Curves

123

which fulfills the conditions


y{0) = O.>'(0) = 1.
Since XQ =

ya = 0, and yo' 1, the series sought is i n this instance

T h e substitution of this series into the differential equation converts the


equation into
2(i2 H" (ia-^ +

\7.a^x^ +

* ' '

1 4a%x

|1 -

6^3 + 4 ^ 2 ^ ! : ^ ^ +

T h e equating of coefficients of like powers of x yields the relations


202 = - 1 .
6fl3 =

4(22)

12^4 = 1 - 6fl3 + 4a2^

T h u s ai = ^, ^3 =

, and accordingly,

PROBLEMS
Find (up to terms in ;c^) the integrals of the following differential equations that
fulfill the given conditions.

45. y" = xy' sin y,

46.

y{^) = 0, >'(0) -

2.

= jy + cos {xy), ).(0) - 0,y(0) = - 1 .

47. y =

+ /

+ 4, >(o) = o,y{o) = 0.

48. / ' = { / ^ x y p .
49. y = >' log:y + X .

:v(o) - o , y ( o ) = 2 .
:yCO) = 1,>'(0) = 3.

50. y =. 1 + {jy - 2i^!'^2|^

^(oj ^ 2.y(0) = - 2 .

7.7. Approximate integral curves


W e shall give two methods by which a graphical approximation to the
integral of a differential equation 7.9 fulfilling conditions 7.11 may be
obtained.
1. T h e differential equation may be replaced by a differential system 7.10. T h e relation _)'o' = fiixo, yoi ZQ) then determines ZQ.
T h e polygonal graphs that approximate the^- and z of the resulting system
METHOD

124

Solvable Equations of the Second Order

are then constructable upon the basis of any chosen set of abscissas x b y
the method of Section 5.2. O f the two graphs so obtained, the one
through the point (xo, yo) approximates the designated integral curve.
Example 10.

T o construct a polygonal graph to approximate the


integral of the equation
/ ' + { 2 x - h 2 ! y + l 2 x - l ) > = 0,
for which

>(0) - 0,

jv'(O) = 2.

T h e given differential equation and conditions are replaceable by the


system
y = z,
z' = - ) 2 x - l b - { 2 x - | - 2 l z ,
w i t h the conditions ^(0) = 0,
^(0) = 2. O n the basis of abscissas
w i t h the differences {x.+i x,j = i , the method of Section 5.2 yields for
Yi and Z,- the calculated values of T a b l e 6.
Table 6
1
T

8
T

5
T

7
T

Yi

0.5

0.75

0.88

0.91

0.89

0.83

0.76

0.68

0.59

0.51

Zi

0.54

0.14

-0.09

-0.23

-0.30

-0.34

-0.35

-0.34

-0.30

Yi'

0.54

0.14 - 0 . 0 9

-0.23

-0.30

-0.34

-0.35

-0.34

-0.30

Zi' - 4
y

-2.25

-1.62

-0.93

-0.55

-0.30

-0.16

-0.03

0.06

0.15

0.39

0.61

0.70

0.74

0.71

0.67

0.60

0.54

0.47

0.41

T h e polygonal graph Y thus obtained is shown i n F i g . 20. T h e actual


solution is i n this instance^ = 2x~'.
Its values at the points x,- have been
included i n the table for purposes of comparison, and the integral cur\^e^
is also shown i n F i g . 20.
2. A differential equation 7.9 associates a numerical value
y" w i t h each triple of values x, y, and y'. It thus associates with these
values a curvature K, given by the calculus formula
METHOD

(7.12)

1 + y 2 | '

and a radius of curvature,

(7.13)

fl

+y''\^
ft

Approximate Integral Curves

125

A graph approximating the integral curve through the point (xo, yo) with
the slope ^^o' may therefore be constructed as follows:
A t (xo, ^o) the normal to the integral curve has the slope 1/yo'A l o n g this normal, with the direction of increasing y taken as positive,
lay off po, the value obtained from (7.13) by assigning to x, y, and y' the
values xo, yo, and yo'.
This locates the center of curvature Co of the integral
curve at (xo, ^o)- W i t h Co as center, draw a circular arc through (XQ, yo).
Let Vi and Yi be the ordinate and slope of this arc at any suitable chosen

FIG.

20.

point xij where x\ > XQ.


T h e construction may now be repeated with the
point (xi, Yi) and the slope Yi i n place of the original ones, etc. T h e
graph obtained is a chain of circular arcs which at each abscissa Xi
advances i n a manner that fulfills the differential equation. I n a certain
way it therefore approximates the integral curve. T h e approximation is
best near the point (xo, yo), and the smaller the differences {xi+i
are taken, the better is the over-all approximation.
Table 7

5
T

0.54

0.32

0.07 - 0 . 1 8

0.71

0.50

0.23 - 0 . 0 4

-0.55

-0.78

-0.96

-1.72

-3.36

-5.33

0.97

0.87

0.73

0.96

0.86

0 -0.26
-1.13

Yi
Yi'
Pi - 1

-1.07
-13.6

7
T

-0.42

-0.63

-0.80

-0.27

-0.62

-0.86

-1.09

-1.07

-1.07

-1.07

-0.99

-0.81

78.5

11.6

5.07

3.24

1.95

126

Solvable Equations of the Second Order

Example 11.

F o r the differential equation and conditions


y+JV-O,

;-(0) = 1. / ( O ) = 0 .

the solution is ^ = cos x. This solution and its approximating g r a p h


based upon the calculated values of T a b l e 7 are shown i n F i g . 21. F o r the

Fro. 21.

construction of this table the values of { and Y'i were determined by


measurements of the figure, and the values of Pi were computed b y
formula 7.13.
ASSIGNMENTS
2. ConstrucE by Method 1 a graph to approximate the integral curve of the difTerential equation

/ ' + {2* + 2 } /

+ (2x -

l b - 0.

for which >(0) = 1,>'(0) = 1.


3. Construct by Method I a graph to approximate the integral curve of the differential equation
I2x - x ^ i y for which

= hy'

2 ^ x^\y + J2 - 2xly = 0,

(0 = - A -

4. Construct by Method 2 a graph to approximate the integral curve of the differential equation
for which ^(0) = l , y C O ) = 1.

Some Geometrical Applications

127

5, Construct by Method 2 a graph to approximate the integral curve of the differ


ential equation
/'
for which >(0)

-T,/CO)

3>' + 2;- = 0,

"0-

7.8. Some geometrical applications


W h e n a curve is specified i n terms of its curvature, it is in fact characterized by a differential equation of the second order. Its coordinate
equation can then be found by integrating that differentid equation.
Example 12.

T o find the equation of the curve whose curvature is


always equal to the sine of its inclination, and that goes through the point
( 1 , 0) with the slope 1.
T h e inclination has the tangent^'. Its sine is therefore ) ' ' / \ / l +>'".
Since the curvature is given by formula 7.12, the curve is one for which
/'

y'

|i + y ' i ^

(1 4-/^1^^

This is a differential equation i n which ^ is not present, and which is therefore of the first order for^'. Its variables are separable. A first integral
is thus
log

= 2x + ci.

/2
+y

T h e condition y* = 1, when x = 1 , is fulfilled by


W i t h ci thus evaluated, the first integral yields

= 2 log 2.

e'+'
dy = .

dx.

and by a quadrature of the last equation y = sin~^ {tf*'*"VV^} + a.


W i t h C2 determined so that y 0, when x = 1, the equation of the
curve sought is found to be
^ = sin ^

TT

V5

PROBLEMS
51. Find the differential equation of the curves whose centers of curvature are
always on the x-axis.
Integrate this differential equation,
52. Find the differential equation of the curves whose centers of curvature are always
on the^-axb.
Integrate this differential equation.

128

Solvable Equations of the Second Order

53. Find the equation of the curve that goes through the point (0, 3) with the slope
zero, and whose curvature is always equal to twice the cosine of its inclination,
54- Find the equation of the curve that is tangent to the ;r-axis at the point (1, 0),
and for which the product of the slope by the curvature has the constant value 1 ,
55, Find the equation of the curve along whose normal the center of curvature and
the jr-axis are equidistant from the curve, and are on opposite sides of the curve.
56, Find the equation of the curve that goes through the point
, 0) with the slope
ff, and along whose normal the center of curvature and the ^--axis are equidistant
from the curve, and are on opposite sides of the curve,
57, Find the equation of the curve that goes through the point (1, 0) with the slope
zero, and for which the product of the curvature by the sine of the inclination has the
constant value -j,
58- Find the equation of the curve that goes through the point (0, 1) with the slope
1, and for which the product of the curvature by the cosine of the inclination has the
constant value

7.9. Curves of pursuit


W h e n an airplane P flying along a certain path is pursued by another
airplane Q that is always directed toward it and that flies at k times its
speed, the path of Q is a curve of pursuit. T h e equation of this curve is
determinable as follows:
Let the coordinates of Q be (x, y) and those of P be (x*, y*), as i n
F i g . 5. T h e n because Q is always directed toward P we have y' =
{y y*)/{x X * ) .
T h i s relation and its differential give us the two
equations
{x~x*\y'

(7.14)

jx - x*| d/ -y'dx*

=
=

[y-y*],
-dy*.

T h e equation of the path of P and its differential, and the fact that the
speed ds/dt is k times ds*/dt give us, further, the three equations
f{x*,y*)

(7.15)

fAx^y*)

= 0,

dx* + / * ( x * , > * ) dy* = 0,


Vdx^

+ dy^ = k Vdx*^

+ dy*\

F r o m equations 7.14 and 7.15 the four quantities x*, >*, dx^, and dy* can
(theoretically) be eliminated. T h e liminant is the differential equation
of the curve of pursuit.
Example 13. A n airplane P flies northeast.

twice its speed.

A pursuing plane Q has


T o find the equation of the pursuer's path.

Curves of Pursuit

129

If the origin is chosen on the path of P , equations 7.15 are


y*

X*,

dy* =

dx*,

dx =

2V2

Vl

dx*.

T h e liminant of x*, >*, dx*, and dy*, between these and equations 7.14 is
-

1P Vl

= 2V2{y

x]y".

This is the differential equation of the path of Q.


T o integrate this equation we set ^ x = u, which transforms it into
a'2
1)2 = 2 "N/Z UU",
and this equation does not contain x.
Hence, when u" is replaced by u' du'/du, the differential equation appears
as an equation of the first order i n u' and u. T h e variables i n this are
separable. A first integral is thus
V2

2u'

+ V2

V~2

Vu

which can be put into the form


ci
Vu

rr
- V i -

Vu

2(r 1

' = 2 V 2

T h e general integral is accordingly


2<:i \ 4 -

\/2 u -

u'^ = 2 V 2 X + :2,

3c 1

and, i n terms of the original variables, is thus


2ci

Vy

V2\y

x]

\y -

x]^^ = 2

x + cj.

3; I

T h e constants C\ and
are determinable when the positions of the two
planes at some instant are known.
PROBLEMS
59. A n airplane flies due east. It is pursued by a plane that has twice its speed.
Find the differential equation for the curve of pursuit. Integrate this differential
equation.
60. Find the differential equation for the curve of pursuit if the pursued plane flies
cast and the speed of the pursuing plane is only one-half as great. Integrate the difi"erential equation.

130

Solvable Equations of the Second Order

61, A pursued airplane flies north. T h e pursuing plane has the same speed.
the differential equation for the curve of pursuit, and integrate it,

Find

62, A pursued airplane flica north. T h e pursuing plane has three times the speed.
Find the differential equation for the curve of pursuit^ and integrate it.
63. A n airplane flies north on the line JC 4, A pursuing plane has 50 per cent
more speed, and passes through the point (12, 0) in the northwest direction. Find the
equation of the curve of pursuit,
64. A n airplane flics east on the line ^ = 3. A pursuing plane has the same speed
and passes through the point (5, 13) in the direction with the slope f-.
Find the equation of the curve of pursuit.

ASSIGNMENTS
6, Describe the method for obtaining the differential equation of the curve of pursuit,
if the pursuing plane has k times the speed of the plane P, and is always directed toward
the point that is at a certain fixed distance directly ahead of/*.

7.10.

The motion of a particle on a curved path

T h e motion of a heavy particle that slides without friction along a fixed


curved path is in accordance with Newton's law. This law again has form
y

FIG.

22.

3.2, if s is taken to be the arc length of the curve, and / is the tangential
component of the active force. W h e n the only force acting is the weight
of the particle, it is seen from F i g . 22 that / zi; sin
that is, / =
z/j dy/ds.
T h e n , since m = w/g^ by (3.3), the law of motion is
dv/dt =

~gdy/dSy

Motion on a Curved Path

131

which, after muUiplication by 2 ds, appears as


2v dv =

2g

dy.

A n integration thus gives


v"^ = a -

2gy,

and when ci is determined by a condition v = VQ, w h e n ^ = ^o, the result is


(7.16)

[ds/dt\^ = vo"

\-2g\yo-y].

By expressing ds i n terms of^ and dy, (7.16) can be converted into a differential equation for as a function of t. Alternatively, by expressing both
ds and y i n terms of x and dx, (7.16) can be converted into a differential
equation for x as a function of /.
Example 14.

A heavy bead slides along a smooth wire i n the shape of


the curve y = x^^. A t time / = 0, it is at the point (1, 1) and its
velocity is f V T 3 . T o find its position at any later time.
W i t h the given values oivo,yo, a n d ^ , and with g = 32, equation 7.16 is
ds = ^ y/^ + 9J:^^ dl.
ds =

Because we also have


V l

dx = V 4 + 9x^^

it follows that dx/x^ = 8 dt. Hence


= 8/ - j - C2- T h e condition
X = 1, when t = 0, requires that C2 = |. W i t h x thus determined, the
equation of the curve gives y. T h e result is

PROBLEMS
65, A smooth wire has the shape of the curve y = x"^, A bead sliding on it is at
the point (0, 0) at ( = 0, and has at that time the velocity 4.
Find the position of the
bead at any later time t.
66, A smooth wire has the shape of the curve ^ = 25 ;c'/16.

A bead sliding on

it is at the point (4, 24) at / = 1, and has at that time the velocity 8 \ / 5 ,
position of the bead at any later time
67, A smooth wire has the shape of the curve ^ = ^ {1 ~ x^^}.

A bead sliding on it

is at the point ( j , f ) at / = 0, and has at that time the velocity 2 \ / 3 0 ,


position of the bead at any later time
68, A smooth wire has the shape of the curve y =

f x^^-

= h \/8

- r

+ 4 sin^^ {y/2

V2|-

Find the

A bead sliding on it is

at the point (1, f ) at / = 0, and has at that time the velocity \/^3'^position of the bead at any later time
69, A smooth wire has the shape of the curve
X

F i n d the

Find the

132

Solvable Equations of the Seconcf Order

A bead sliding on it a at the point (0, 0) at / ~ 0, and has at that time the velocity
8\/3.

F i n d the ordinate of the bead at any later time t

70. A smooth wire has the shape of the curve


i y V 9 - / + | 8 i n - '

{>/3|.

A bead sliding on it is at the point (0, 0) at t = 0, and has at that time the velocity
8\/5.

F i n d the ordinate of the bead at any later time L

71. A smooth wire has the shape of the curve.


jr -

8{e -

sin

fl},

y = 8[cos e -

\],

A bead sliding on it is in the position for which d = 0, at / = 0.


of the bead at any later time

Find the coordinates

72. A smooth wire has the shape of the curve


2\e + sin e],

X ^

jK = 2(1 "

A t>ead sliding on it is in the position for which 9


of the bead at any later time t

cos 9 .

at f 0.

Find the coordinates

ANSWERS TO ODD^NUMBERED PROBLEMS


\. y = \x^ + a log X + C2.

3. >

2 tan ^ \ci^] + C2

5- > log {log X -\- a] -\- C2y and y " c.


9. y=

7, y %in y = c\x + rj-

C\X + f2 - 1
ClX + C2

./2

11.

{1 + l o g ; - / ' +
y

15. : r y V + / y

27.
31.

{1

+ 0.

= 0.

19. | ^ _ ; , i y ' _ ( y _ i j 2
23.

n. yy" - yy' -

= 0.

= o.

= O - ^ i l o g 11 -

X sin y + y sin x =

cix

21. 2 U + : y l / ' + {1 - / M i l + / i
25.

x].

Inexact.

1
29. ^ = - + C2 -

+ cj.

1^
-

[ x + c i l y e ' + c i .

33. > = log [x^ + cix + C2],


35. y =
37. y -

z=

{ci+

2xU-''.

CiX + C2

X'

.3

Ix + c i l s i n - ^ {x + C 1 + - \ / l -

z = ^ 4- sin

+^2,

+ i'll-

39. y' = x^ ~ yz,


43. :y' = log ^,

U +f,j2

z' = xj-.

2' = > + 2z.

41. /

z' - x ^

45. ;v = 2x + i x * +

+ /

-0,

Answers to Problems
47. y ~ 2x' -{-^x* -\- .
SI. yy"y'^

I -0;

49. ^ =- 1 + 3;t +

{x +

a]^-\-y

Ci

53. ^ = 3 -J log cos 2x.


57. y

59. V l

{V('

65. x > 4r,


67.

69. ^ =
71.

1 w

"

4)^^ -

2x)\\

cos- (3 -

2M

63. y = 6 [ V 2

'

- i

= 2"

61. v * i

-x)

1)(2

log X -

2} -

_u

^1-

4*1

^^'^ +
40

-l6/^
?5

16

{12;

= 16( -

+3

V3|** -

8 sin (2/),

3.

y =

-8

2.

+ 8 cos (2/)-

4) -

32}

fx*

CHAPTER

8
The Linear DifFerential Equation
with Constant Coefficients
8.1. The complete equation
T h e differential equation
(8.1)
i n which a,

y"

ay' + i> =

or its derivatives, is called the linear


differential equation of the second order because each of its terms is of, at most,
the first degree in y and its derivatives. If wi{x) and W2{x) are, respectively, solutions of equations of this type w i t h the same coefficients a and by
thus
(8.2)

h, f{x)

do not involve

/(x),

wi'

awi

bwi

= fi{x)y

W2"

aw2

bw2

/2(jf),

we see, upon multiplying these equations respectively by any constants ai


and 02 and adding them, that the combination
(8.3)

y =

aiwiix)

-\- a2W2(x)

is a solution of equation 8.1 with


(8.4)

/(;.)

aifi(x)

a2f2{x).

Clearly a corresponding statement can be made if there are more than two
equations 8.2. Hence a differential equation 8.1 in which the function
fix) consists of a number of terms, as in (8.4), can be integrated by integrating each of the simpler equations 8.2, and combining their solutions.
Several important conclusions may be drawn from this observation.
T o begin with, let us consider the differential equation
(8.5)

u" + au' -\-bu

= Oy

which is obtainable from (8.1) by replacing/(x) by 0. If ui(Af) and


are any two particular integreils of this equation, the combination
(8.6)

ClUlix)

134

C2U2{x)y

U2{x)

The Complementary Function

135

w i t h arbitrar)' constants ci and C2 is also an integral. It is the general


integral if the constants ci and
are not replaceable by just a single
arbitrary constant, that is, if the ratio of ui{x) to 2 ( ^ ) is not a constant.
W h e n differential equations 8.1 and 8.5 are considered together, (8.1)
is generally referred to as the complete equation, and 8.5 as the reduced equation. T h e genera] integral 8.6 of the reduced equation is called the complementary function.

If i n equations 8.2 we take fi(x) = f{x), and fiix) = 0, we may choose


for wi{x) any particular integral yi{x) of equation 8.1, and for W2{x) the
general integral of equation 8.5. T h e combination
(8.7)

y = yi{x)

ciUi{x) + czuzix)

is thus also an integral of equation 8.1. T h e general integral of a complete equation can therefore be found by finding a particular integral, and
adding the complementary function to it. Thus the task of integrating a
complete equation is resolvable into two parts, namely, the determination
of the complementary function, and the determination of a particular
integral.
T h e simplest type of equation 8.1 is that in which the coefficients a and b
are constants. Such an equation is always integrable by quadratures.
T h e case i n which = 0 is especially simple. A quadrature can then be
applied directly to obtain a first integral, and this first integral is a linear
differential equation of the first order.

8.2. The complementary function


One method of integrating the reduced equation 8.5 is already known to
us. W e found, in Section 6.7, that by setting

this equation is transformed into the Riccati equation


v' =

av v^.

This Riccati equation, since it has constant coefficients, has a constant


solution. B y the trial of y = m, it is found that this is a solution if m is a
root of the equation
(8.8)

m^ +

am -\- b =

0.

This is called the auxiliary equation. If its roots mi and mz arc distinct, we
obtain i n this way two integrals i = e^^' and 2 = e^^'. W i t h these
integrals, formula 8.6 gives the general integral. If mo = m i , we obtain
only one integral, i = " * ' ^ in this way. However, the Riccati equation
then also has the integral
= m\ -\- \/x, and &2 leads to u^ = xe^^'. T h e

136

Linear Equations with Constant

Coefficients

complementary function is thus


(8.9)

u = CKT"*!' H- C2f"'^'y

if

mu

u = citf"'* + C 2 " ' * ,

if

m2 = m i .

W i t h this result once established, a procedure for recovering it quickly


i n any instance is easily formulated. By substituting e"*' for u i n the given
differential equation, we are led at once to auxiliary equation 8.8, the
roots of which must be found. W i t h these roots, formulas 8.9 yield the
general integral. T h e particular integral for which
(8.10)

u{xo) = 0 ,

U'(XQ)

uo'

is obtainable, as usual, by determining the constants ci and C2 i n (8.9) to


fulfill these conditions.
Example 1.

T o find the integral of the differential equation


u" -

2u' -

2 = 0,

for which w(0) = 1, and u'(0) := - 1 .


T h e substitution u = e"' converts the differential equation to
f"'|m2 -

2m -

2)

=0.

B y striking out the exponential factor we obtain the auxiliary equation,


the roots of which are m i
and m2 = 1
T h e general
integral is thus
B y differentiating the general integral we find that
' = 1 + \/3)tfitf!i+V3I*+ |1 - \ / 3 | c 2 ^ U - v ^ l .

T h e conditions to be fulfilled are therefore


ci-\-C2

These yield ci = ^
integral is therefore

1,

{1

I/'N/S,

V3]ci

and

"2

{1

V3]C2

= i + i/'s/Sy

-1.

a n d the required

u =
Example 2.

T o find the integral of the differential equation


" - 1 ' + iftr = 0,

for which u ( l ) = 0, and u'{\) = 2.

Complex Roots.

Euler's Formulas

T h e auxihary equation is i n this instance


~
- f ^ = 0.
mi = f as a double root. T h e general integral is therefore
u

cye

137
It has

-\- Czxe

By differentiating the general integral we find that

T h e conditions are thus


= 0,

{ci-\-C2]e^

F r o m these conditions, ci = 2e~^^ and


integral is therefore
u =

= 2.

Ihi-^^c^U^
C2

= 2e~^^. T h e required

2{x-\]e^^^'-^K

PROBLEMS
Find for each of the following differential equations the integral that fulfills the
respective conditions.
1. u" -

3u' + 2u = 0,

2. u" + u' - TU = 0,

u(0) -

9u = 0,

4. u " -

7u' +

5. u" -

4' -

u = 0,

6.

a' + 2u = 0,

7. a " -

2u' -

80u = 0,

8. u" -

2 V ^ u ' + 3 = 0,

9. u " -

7u' + lOu = 0.

10. 9 " -

6u' + u = 0 .

11. u" + 3u' -

u(l) = 1, u'(0 = 1= 0,

u =0,

12. 2u" + u' = 0,

1.

uCO) = 1, u'(0) = 0.

3. u " -

u" -

0, o'(0) -

u ( - l ) = 0, u ' ( - l ) = 2.

u(l) = 2, u'(l) =

-1.

u(0) = 1, u'(0) = 2.

uC-1) = 0, u ' ( - l ) = 0.

uCO) => 0, u'CO) = 0.

u(0) = 7, u'(0) = 14.


u(3) = - 1 , u'C3)

u(0) = 2, u'(Q) =

4.
-3.

u ( - 4 ) = 1, u ' ( - 4 ) = 1.

8.3. The case of complex roots.

Euler's formulas

Formulas 8.9 are valid for complex as well as for real values of mi a n d
mi. However, they do not give the integral i n a real form when these
roots are complex. Alternative real formulas are therefore desirable i n
that case.
If mi = a + 1^3 with /3
0, the R i c c a t i equation has a +
as a complex solution. Its general integral, as given by formula 6.14, is then
V

= a 0

tan

{0x +

c].

138

Linear Equations with Constant Coefficients

F r o m this value of v we obtain the general integral of equation 8.5:


w - - + I o 8 c o ItfaH-cJ

T h i s result can be put into any one of the following forms:


u C2^"* cos {j8x + c i l ,

(8.11)

u = CiC^'sm

{/3x + c i | ,

sin

C2 cos ^x\.

T h e first of these forms is obvious. T h e third is obtainable from the first


by using the cosine addition formula, a n d renaming the arbitrary constants. T h e second is obtainable from the first by replacing c\ by ci i r / 2 .
W h i c h one of forms 8.11 is the most convenient to use w i l l ordinarily
depend upon the circumstances of particular problems.
Example 3.

T o find the integral of the differential equation


u" -

6 ' + 13 = 0,

for which u(0) = 2 and u'(0) = - 1 .


T h e roots of the auxiliary equation are i n this instance 3 2 i . T h u s
a = 3 a n d j3 = 2. B y the third formula 8.11 the general integral is
therefore
u = e^'{ci sin 2x + ^2 cos 2x\.

T h i s general integral has the derivative


u' = e^'{{3ci 2c2) sin 2x + (3^2 + 2ci) cos 2x\.

T h e given conditions are accordingly fulfilled if ^2 = 2 and 3c2 -\- 2ci 1.


T h e required integral is thus
u = e^'{ - i sin 2x + 2 cos 2x
Example 4.

T o find the integral of the differential equation


M" +

2u' +

4 = 0,

for which U(T/2) = 0 a n d U'(T/2)


=- 1 .
I n this instance the roots of the auxiliary equation are 1 i/2. T h e
first formula 8.11 therefore gives u = C 2 ? ~ ' c o s l(x/2) -\- ci]. This has
the derivative
w' = cze *cos

+ ci

- f 2*

Complex Roots.

Euler's Formulas

139

and the conditions therefore take the form


cos 5 + ,

C2e

cos

c^e

^ C2e

= 0,

72 sm

-1

T h e first of the last pair of equations is fulfilled if ci = 7r/4, and the second
is then fulfilled if t:2 = 2e'^^. T h e required integral is therefore

2^4
Some important formulas of mathematical analysis follow from the
results we have obtained. T h e integral of the equation
u 0, for
w h i c h u(0) = 0 and u'(0) = 1, is found from (8.11) to be sin
and from
(8.9) to be

e"*'].

Similarly, the integral for which u(0) = 1

and a'(0) 0 is found to be both cos x and i\e*' +


that
(8.12)

1
sinx = - {^"-^-"1,

W e see thus

1
2

2i

I n their converse form these evaluations are


(8.13)

e^' =

cos

X -\- i

sin

tx

x,

=_ cos

A:

4
zsm

x.

Relations 8.12 and 8.13 are called Euler^s formulas, i n honor of the Swiss
mathematician Leonard Euler (1707-1783).
PROBLEMS
F i n d for each of the following differential equations the integral that fulfills the
respccdve conditions.

13. u" - 4u' + 5u = 0,

u{0) = 4, u'(0) = 2.

14. u" + 2u' + lOu = 0,

u(l) = 0, o'(l) = 4.

15. u" - 3u' + V-u = 0,

u(0) = 1, u'(0) = 0.

16. u" - lu' + 2u - 0,


17. u" + 25u = 0,
18. u" + 2 y/l

u(2) = 1, u'(2) = 0.

u(3) = - 1 , u'(3) = 0.

u' + 4u = 0,

aI

VV2/
19. " + 2 V 3 ' + 4u = 0,

J = 0, u' (

V V i /

u(0) - 0, a'(0) = 0.

140

Linear Equations with Constant

20. u" -

12 log 2]u' +

21. u" -

4 ' + 53 -

22. u " - 2 { 1
23. " -

o(l) - 1, u'(l)

2.

0,

2ru' + 2^^* = 0,

u(l) = 0. '(1) =

(1) = 1, '(1) =

- 1 , u'

uC-3)

+ \ / 2 l u ' + | Y H - 2 \ / 2 | U =0,

24. u" + 17 = 0,

8.4.

|4 + log^ 2\u = 0,

Coefficients
^ 0.

= 0, u ' ( - 3 )

=0.

2T.

1.

Differential operators

W h e n a differentiation is denoted by D instead of by the less simple


d/dxj and
is used to denote the second derivative, such expressions as
fp'{x) 4- ctp(x)j and <p"{x) + a<p'{x) + b<p{x) appear
respectively
as
Dtp -h c<p and D^<p 4- aD(p + btp. If we now define \D 4- c]<p to stand
for D<p 4- c<py etc., we have
^'W

ip"(x) 4- V W

4-^^ =

{D-\-c]<p,

4- b<p(x) =

a) 4-

fik.

Herewith the symbols {D -\- c], and {D^ {- aD -\- b\ have been introduced. They do not stand for quantities, as is the case with literal
expressions i n algebra. Rather, they indicate operations that are to be
performed, namely, differentiations, additions, and multiplications that
are to be carried out upon the function to whose symbol they are prefixed.
For that reason they are called differential operators. D itself is the simplest
differential operator. T h e effect of a differential operator upon a function
is to convert it into another function, precisely as the operator D converts
a function into its derivative. Thus, for example,
( D 4- 3) cos

A:

[D^ 4- 4 D 4- S]xe'

sin

de'.

4- 3 cos

x,

T h e function which is obtained as the net result of first applying the


operator {D -\- c\\ to ^ ( x ) , and following that w i t h an application of the
operator \D -\- C2\,\s denoted by
(8.14)

\D +

C2\[D

ci]ip.

This is by definition the same as


{D-^C2\W{x)^ciip{x)\.

It is thus further the same as


W{X)

Ci,p{x)Y +

C2W{X) +

.i^(x)).

141

Finding a Particular Integral


which is the function

But we have agreed to write this last as


(8.15)

\D''\-{ci +

C2)D-ir

C1C2W.

Because expressions 8.14 and 8.15 are thus the same for every function
^(jf) we shall write
(8.16)

\D + C2][D

+ ci}

\D''^{cx

C2)D

ciC2\.

(8.16) is not an algebraic equation between quantities because D is not a


quantity. F r o m the way it was derived it is shown to be the statement,
expressed i n symbols, that the net effect of the two operators appearing on
the left is the same as that of the single operator appearing on the right.
F r o m relation 8.16 we now obser\'e a most important fact, namely, that
the right-hand symbol is correctly obtained if the ordinary rules of multiplication are applied on the left, even though the letters do not all stand
for algebraic quantities. T h e usefulness of differential operators is largely
due to the fact that they do combine with each other and with numbers by
the laws that are already known to us from algebra. I n particular, if mi
and m2 are again the roots of auxiliary equation 8.8, they have the sum a
and the product b. Hence relation 8.16 shows that
(8.17)

-^aD + b] = {D -

mi\{D -

m^].

This is precisely as if the left-hand expression were "factored." A n


application of the left-hand operator of 8.17, which is of the second order,
is thus replaceable by the successive applications of the right-hand operators, each of which is only of the first order.

8.5. An operational method for finding a particular integral


B y the use of relation 8.17, differential equation 8.1 may be given the
form
I D - mi]{D

m2]y

f(x).

This is equivalent to the successive equations


(8.18)

{D -

mi\Y = f(x\

and
(8.19)

{D-m2\y

Y{x).

N o w these are two linear differential equations of the first order.


tion 8.18 has a solution
(8.20)

Y(x)

= e'^''Sf{x)e-^^' dx,

Equa-

142

Linear Equations with Constant

Coefficients

and therewith equation 8.19 has a solution


(8.21)

y{x) = * " * / r ( x ) - ' / x .

T h e substitution of (8.20) into (8.21) gives the integral


(8.22)

y(x) = e'"^'je^'"'-^'^'\Sf(x)e-^^'dx]

Example 5.

dx.

T o find the general integral of the differential equation


y"

3/

']-2y

= x.

T h e roots of the auxiliary equation are i n this instance m i = 1, a n d


m2 = 2. T h e given differential equation can therefore be written as
{D^i}{D^2\y^x.

T h e successive equations by which it can be integrated are


\D -

l | r = X,

\D-2]y

Y(x).

O f these the first has a solution Y{x) = (x + 11. Therewith the second has a solution v = (x/2) +
and this is a particular integral. T h e
complementary function is cie' + 2 ^ ' . T h e general integral is therefore
> =
Example 6.

+ 1 + cie' + C2e^.

T o find the integral of the differential equation


+ 2y' + 10;r = i + 2,

/'

or which y(0) = 1 and y'{Q) = 0.


T h e roots of the auxiliary equation are 1 3i, therefore the given
equation is
\D-\-i

3i] Z) +

1 + ?>i]y =

2.

T h e successive equations by which this equation can be integrated arc


(Z) + 1 -

3i]Y

+ 2,

3i]y -

Y(x).

T h e first of these has an integral


2
2 3i

3i

and therewith the second has an integral y = - J ^ * -|- i. T h e complementary function is e~'lci cos 3x + f2 sin 3x|, T h e general integral is
therefore
y =
+ i + ^"'Ui cos 3x + cz sin 3x1.

Finding a Particular Integral

143

F r o m the general integral and its derivative we see that the conditions
to be fulfilled are
T^

+ i + '^i = 1,

3^2 = 0.

T V - ^ J +

These determine ci and C2, and the integral required is thus found to be
> = TV* + i + e-'{U
Example 7.

cos 3x-\-U

sin 3x\.

T o find a particular integral of the differential equation


iy = xe' + sin 2x + 5x.

y" -

I n place of this equation, we may consider the simpler equations 8.2,


which i n this instance are
wi"

~ iwi =

W2"

-1^2

sin 2x,

^z" i^3 = Sx.

These, respectively, have the integrals


Wi ^xe' -V"^*>

1^2 = ~ A" sin 2x,

wz = 20x.

Their sum,
4^

321
4 .
\ sm X
9I
17

2\Jx.

is a particular integral of the given diflferential equation.


PROBLEMS
Find the general integral of each of the following differential equations.
25. y" - y' -2y

= 3^

26. y" ~ Ay = x^.

27. >" + > = Ae'.

28. y" + 2y' + 5y = 3.


e~'

29. v " + 2y' + y =

31. y" -

2y' -

x+2

30. y " -

8> = 9**^ + l O * " ' .

2v' + 2y = sin x.

32. y" - y =^ - ; f + 2

x""

Find for each of the following differential equations the integral which fullills the
respective conditions.
33. y" +2y
34. y" -

= 4,

y{Q) = 3 , / ( O ) = 1.

2y' \-2y =

:y(0) = 0, ^-'(0) - 0.

144

Linear Equations with Constant Coefficients

35. y

+9> = 8sin:c,

36. y" + 2y' -f-

= -1,

> 2 cos^- -

1 +

37. >" - h ' + h =


38.

+ 2;.' -

1.

2T

sin^'

= 1, / ( I )

+ 2 U ^ yiO) = 3. / ( O ) - 4,

>(0) - 0, /(O)

3;^ -= - 3 ; c - + x + 7.

= 2.

8.6. A second operational method


A n alternative method by differential operators can be used whenever
the roots of the auxiliary equation f re unequal. T h i s method is suggested
by the following analogue from algebra. If k is any number not equal to
either mi or m2, the value of ^ that fulfills the equation
[k y =

is

mi\{k

m2\y

=/

{k mi) (A m 2 )

A resolution into partial fractions gives the last equation the form
{//('"I - ' " 2 ) 1

y = -

Thus

3 = ^1 + > 2

{k -

k m

+
,

|//(m2-mi)j

k m2

where

mi]yi

m2l>2 =

{k -

mi m2

/
m2 mi

Consider now the analogue of this procedure for differential operators


T h e relations
(8.23)

{/) - m i i j v i =

fix)

{D -

mi m2

m2]y2

fix)
m2

mi

are linear differential equations of the first order. T h e results of operating


upon equations 8.23, respectively, by \D m2l a n d [D m\] are the
equations
yi"

+ ayi' + hyi =

{fix)

m2f{x)\,

7fl\
y2

ayi + ^'2 =

-1

l/'W

- m i / W ) .

T h e sum of the last two equations shows that \y\ + ^ 2 } is an integral of


differential equation 8.1. W e can therefore find a n integral of equation

A Second Operational Method

145

8.1 by finding and adding integrals of the simpler equations 8.23. T h e


formula obtained i n this way is
(8.24)

y(x) =

Example 8.

-^^^^

-'"1'

dx +

mi m2

e-"^^ dx.
m^ mi

T o find the general integral of the differential equation


jy" + >' - 2V

T h e roots of the auxiliary


Equations 8.23 are therefore

e'

equation are mi = 1, and m2 2

3(1+^)

'

3(1-1-^}

T h e y have, respectively, the integrals


y\{x) =

log
^ll

e'

y2{x) = - 7 + J ^ - ' - ^ ^ ^ l o g 1 + . ' 1 .


6

T h e sum of these is a particular integral of the given differential equation.


A d d i t i o n of the complementary function yields the general integral
> = i (T* log

\ +en

-\e''
3

log{l +
'

+
' 3

+
cie'

Example 9.

C2e~^''

T o find a particular integral of the differential equation


y"

-ly'

^-ly

^ e sin^ x.

T h e roots of the auxiliary equation are 1 + i . Equations 8.23 are


therefore
\D-

\ ~ i\yi = ^

^ sin^ X ,
2
'

|/) -

'

1 + i\y2 ^ ^
" 2

2
sin^
x.

T h e y have as solutions
yi =
>2

{ sin^ * + 2i sin x cos x + 2},


sin^ X

2i

sin * cos j ; + 2 ) ,

from the sum of w h i c h , we obtain the desired particular integral


y = h'{2

- sin^xj.

146

Linear Equations with Constant Coefficients


PROBLEMS

Find the general integral of each of the following difTerential equations.


39. y" -

40. y" + 5 / + 6> = 12x.

8>' + 15> =

42. y" y sin x.

41. / ' +:y =


43. y" + ly' - 2y = e' +
45.

+4/

47. y" -

+ 13;- =

5>' -

44. :y" 46.

X.

2> = 3*^'

+ 4;- = cos 2x.

48. > " + > ' - 6> = X + 2x

7;> = 1.

49. y" - ly' -\-2y

51.y'-:y

r'.

x\

50. / ' + /

52. :v

-2> =

3^

4^

i l + x ^

8.7. A comparison of formulas


Formulas 8.24 and 8.22 each give a particular integral of differential
equation 8.1. But 8.24 does so by two independent quadratures, whereas
8.22 does so by two successive quadratures. Although i n any given
instance the evaluation by one of these formulas may be simpler than that
by the other, the content of the two formulas is the same. T h a t can be
shown as follows: A n integration by parts applied to formula 8.22 changes
its form into

m\

mi

m^

f{x)e-^^'dx

It is Ccisy to see that this is formula 8.24.


It is sometimes desirable to choose for the indefinite quadratures i n
formula 8.24 those that reduce to zero at a given point JCQ. T h e y are the
definite quadratures from XQ to x. W i t h the variable of integration
denoted by s they make the formula appear as
lis)

y{x) =

Jo mi

ds -

Jxo mi m2

ds.

Since X is now not the variable of integration, the factors


and
may be put under the integral signs. T h e formula can then be written
compactly with a single integral sign as
(8.25)

yM

fis)
xo

^1

ffl2

T h i s gives the particular integral for which >'(xo) = 0 and>'(xo) = 0,

The Method of Undetermined Coefficients

147

ASSIGNMENTS
1. Consider formula 8.22 when nti
the form
(8.26)

yix) "

f'f{s){x

mu and show that it can then be reduced to

- .r}*'"'^^') ds,

if m2 = mi.

2, Consider formula 8.25 when the auxiliary equation has the complex roots a t^
and show that it can then be reduced, by the use of Euler's formulas 8,13, to the form
,W -

(8.27)

s.

8.8. The method of undetermined coefficients


M a n y problems i n engineering and science depend upon differential
equations of type 8.1 in w h i c h / ( x ) is a polynomial, an exponential, a sine
or a cosine, or some product of such factors. F o r such an equation a n
integral can be found without making any quadratures.
Suppose f(x) is a polynomial of the degree n. W e shall then take y
as a polynomial of the same degree
(8.28)

yix"-^

y^g^^--\-

with undetermined coefficients. T h e substitution of (8.28) into the given


differential equation converts the latter into a n equation between polynomials. B y equating the coefficients of like powers of x we obtain a set
of relations from w h i c h the values of ^o, ?i> * ' ' ?n can be successively
found.
Example 10.

T o find a n integral of the differential equation


y" H- jy' -

6> = 2x^ +

-7x-\-2.

Substitution 8.28 with n = 3 converts the differential equation into


the equation
-6?6* +

1 - 6 ^ 1 + 3qo\x^ +

1 -6q2 - f 2qi + 6qo\x +

{-6qz

?2 +

2qi] = 2x^ + Sx^ -

7x + 2.

By equating coefficients of like powers of x we obtain the relations


-6qo

= 2,

6^1 + 3^0 = 5,
6q2 + 2qi + 6qo = 7 ,

6^3 4- 72

+ 2qi =

2.

148

Linear Equations with Constant Coefficients

F r o m these equations qo = i ; qi 1 ; q2 =

and q^ =

Thus

is an integral.
A n equation 8.1, i n which f{x) is of the form *Yi(x), is transformed by
the change of variable y = e^v into
v" +

(a + 2k\v' + | i +

If
is a polynomial,
method above.
Example 11.

+ k^\v =

can be determined as a polynomial by the

T o find an integral of the differential equation


/'

e-'^^\\2x^ -

- W

9x -

V-l-

T h e change of variable y = e~'^'^v transforms the given equation into


2v' = 12JC^ -

v" -

9x

I n the transformed equation the coefficient h is zero. W e may therefore


perform a quadrature upon the equation directly, to give it the form
= 3x* -

v' -2v

\x^ -

^x.

If we now substitute for v a polynomial 8.25 with n = 4, we are led to


the equations
=

3,

4qo =

0,

-2qo
-2qi

-2q2

+ 3^1 =

-2?3

+ 2?2
-f

-2?4

F r o m these

v = -h*

Hence

?3

= 0.

- e'^^H ~ix^

-f,

+ 2x + 1-

3x^ -

^x^2x-\-I],

is an integral.
Example 12.

T o find an integral of the differential equation


y"

1 x ^ - 1 ) ^ ^ + {3A:+

-2y'-^y=

4)^'.

A pair of equations 8.2 are i n this instance


wi" W2"

2wi'

;i -

e^^'lx^ -

2w2' -\- W2 = e'i^x

Ij,

-\- 4].

Undetermined Coefficients Continued

149

T h e respective changes of variables wi = e^vi and W2


them into
+ 2oi' +

v{'

1,

transform

v^' = 3* + 4.

T h e first of these has an integral v\ =


4x ~\r 5. T h e second, w h i c h
can be solved by two quadratures, has an integral v% ^x"^ + 2x^. By
multiplying v\ and v^. by the appropriate eponentials to obtain w\ and KJ2,
and then adding w\ and w^, we find the integral
V =

4x +

51.2^ +

\kx^ +

2^2|^.

PROBLEMS
Find, by the method of undetermined coefficients, an integral for each of the following differential equations.
53. / ' + / - ! > ' = 3*2 54. y" + /
55. y" -

6> = Jf.

ly' = - 3 5 * ^ + 76x^ -

56. y" -^-y = \2x 57. >" + y

-6y

58. y" + 2y' 59. y" - h '


60. y" -y'
61. y" -

By = 9e^^.
\20x^ + 24^2 + 5x -

iU^\

= 12*V^-.

3 / +|>

= 27'^^^

6y = 2e^ + 3^' + 5JC + 1-

Ay' + 2> =

-2y=

65.
66. y" -

42x + 6.

= \x^ -\-\U'.

-\-h

64. / ' ^y-

lAx'^ -

2}*"^

- y=

62. y" + Sy'


63. / '

8x + 1.

123*2 _ 43^ _,_ \2\r'^^ +

\\^x^ + 4x + S}*** +
+ 24 + 3e^ +

2/ -

3> = - i4x + 4}^' -

\\\x +

19

{6x + Sjir^''.

4f-^
{8x -

2)-' -

6^^' -

(5x -

6|f-K

8.9. The method of undetermined coefficients continued


T h e method of undetermined coefficients can also be applied to an
equation 8.1 i n w h i c h i s of the form
/(x) = P i W sin

7:^

+ 5) +

^ 2

cos

(7X

+ 51,

i n which 7 and 5 are constants, and F\{x) and Pi{x) are polynomials.
I n this instance we shall take y i n the form
(8.29)

y - Qi(x) sin (7^: + 6) + Q2(x) cos 7^ + S j ,

Linear Equations with Constant Coefficients

150

w i t h Q i and Q2 as polynomials with undetermined coefficients

Q2W

= ?2,0X"

?2.1""

?2.n.

T h e degree n shall be the larger of the degrees of P i and Pz, unless the
instance is one i n which a = 0 and b = 7^. I n this exceptional instance
n must be taken greater by 1.
Substitution 8.29 converts the differential equation into an equation
between sums of functions of distinctive types. B y equating the coefficients of like functions, we obtain relations from which the values of qtj
can be determined.
Example 13.

T o find an integral of the differential equation


y"

+ 2>' + 5> = 3 cos

I n this instance n = 0, 7 = 1, and 8 = T / 4 . Substitution 8.29


converts the given differential equation into the equation
H^i.o

292.0) sin

ir
4

{2qi,Q +

492.0I

cos

" - 4

3 cos ' * ~ ^
By equating the coefficients of like functions we obtain the relations
4?i.o 2^2,0 = 0,

2^1,0 + 4^2.0 = 3.

These simultaneous equations give 71,0 ^ 0 and 72,0 = %


, 3

Thus

-COS

Jf 4

is an integral.
Example 14.
y"

+>'

T o find a n integral of the differential equation


+

87

IIOJC^ +

21*

9!

sin

3x

x cos

3x.

I n this instance n = 2, 7 = 3, and 5 = 0. Substitution 8.29 with these


values converts the differential equation into the equation
l(-?i.o -

3^2.0)-^ +

(-

qi.i

372,1 4- 271.0 -

1272,0)*

( - 7 1 . 2 - 372.2 H- 71.1 - 672,1 + 271,0)) sin 3x + {(371.0 - 72.o)x^ +


(371.1 - 72.1 + 1271,0 + 272.o)x H- (371,2 72.2 + 671.1 + 72,1 +
272.0)) cos 3x =

\\0x^ + 21jr + 9| sin 3x + x cos

3x.

Undetermined Coefficients Continued

151

By equating the coefficients of like functions we obtain the relations


?i,o

3^2,0

= 10,

3?i.o -

?2,o

0;

? i . i 3 ? 2 . i + 2yi.o 12^2.0 = 21,

6^2,1 + 2yi,o = 9 ,

~ ? 1 . 2 3^2.2 +
?2,2

391,2

6^1.1

?2.1

2^2,0

= 0-

T h e first pair of these relations is a simultaneous system which yields


fji.o = 1 and 9 2 , 0 = 3 .
Therewith the second pair yields ^ i . i = 7
and q2,i = 2. T h e third pair then yields 9 1 , 2 = 1 3 and ^2,2 = ~ 1 T h e integral thus obtained is
y =

l-x"^

Example 15.

-\.Jx

13j sin 3x +

{-2,x^ +

2x -

1 i cos

3x.

T o find an integral of the diff"erential equation


_

^ e, 2 x

4y' -\-Sy

^) + 2 cos

sin(. +

A:

T h e change of variable y e^^v transforms this diff'erential equation


into
v" -\-

= sm \x

+ 2 cos

'+5

Although the polynomials i n the right-hand member of the transformed


equation are of the degree zero, we take n = 1, because this is an instance
i n which a 0 and b = 7^. B y the method described, we then find that
2?i,o ~ 2, and 2 ^ 2 . 0 =
N o relations are found for 91,1 and 9 2 , i T h e values of these coefficients are therefore arbitrary. W e may assign
to them the value zero. Thus
V

sin \x -\- -

- X

cos x-^^n

and the integral found is


2x

sm I X + ~ I

1
~

cos

PROBLEMS
Find, by the method of undetermined coefficients, an integral for each of the fol
lowing difTcrcntial equations.

152

Linear Equations with Constant Coefficients

67. / ' + 3 / +7>


68. y" -

2)-' -

15 cos

3;- = sin X

X.

12 cos x.

IT

69. y" + 2;-' + 5/ -

3 sin

70. y" -

-*'cos.

2/ + > -

71.

+ 2) sin X +

72.

4 / + 4^ = sin X -

73.

+ 6y' + 16> = sin 2x + cos

74. y

ix^ + 4x| cos

3/^'.
2JC

+ 4 sin 4x.

4y' + 5> =- 2"^ sin x

75. y" -\-Sy = - 1 5 ^ ' sin 3x -

29^"' cos x.

76. / ' -

2 / + 2> = e'{2 sin x + 4 cos x\ -

77. y

2 / + 6> = *{4 sin x + sin 2x -

jf.

78. / ' + 3y' + 4;. = } Jx^ + x| sin 2x +


79. / ' -

3 / + 3> = x ' + ^'1 x^ -

80. / ' -

4 / -\-Sy = e'\{x'^ -

lOx cos 2x.

4 sin 3xi.
\x^ + 2x2]

4xj sin X -

2x.

2x cos x|.

\2x + 1) sin 3x + ( - I B x * + 12x + 18) cos 3x1 +


2'{4sin X + 2 cos x|.

8.10. DifFerential equations of higher order than the second


A l l the methods of this chapter are applicable to linear differential
equations of higher order if these equations have constant coefficients.
T h e general form of such a differential equation if it is of the reduced
type is
u^"l + aiKt"-^l + 0 2 ' " " ' * ' +

+ au = 0,

with u'"^ standing for d^u/dx^. T o obtain the complementary function,


we substitute e"^' for w, and thereby convert the differential equation into
the auxiliary equation

For each root rrij of the auxiliary equation, the function e"^'' is a particular
integral. If all the roots
are distinct, the complementary function is

and involves the n arbitrary constants c i , (^2) * ' ' , ^n- F o r any root
m that is multiple, say of the order r, the functions r " " , x^"**, , x'^^^""
are particular integrals. If the coefficients of the differential equation
are real, as we have been supposing, a complex root m = a +
of the
auxiliary equation occurs only i n conjunction with its conjugate ih =

Equations of Higher Order

153

a i/3. B y the use of formulas 8.13, the integrals e"^' and e^' can be
replaced by the real ones, e"' cos 0x and e"' sin jS.t.
T h e complete linear equation of the order n is

and is expressible by the use of differential operators i n the form


{D -

mi] ID -

m2\ - {D -

m^ly

f{x).

A particular integral of this equation is therefore obtainable by integrating


successively the equations
-milri =

f{x\

7722 Y2
=

Viix),

\D tn^^Yz =

y2{x),

{D-~m,]y

T h e method of undetermined
and 8.9, can be applied without
higher order if its function f{x)
kind mentioned in Section 8.8.
particular integral.
Example 16,

coefficients, as described in Sections 8.8


modification to a differential equation of
is made up of terms and factors of the
This method can then be used to find a

T o find the general integral of the differential equation

T h e auxiliary equation is
m^ -

2m^ + 6m=^ -

32m + 40 = 0.

Since this equation can be given the form


[m -

2\^{m^ +

2m +

iO\ =

0,

it is seen to have m = 2 as a double root, and to have also the roots


1 3i. Thus the reduced equation has the integrals i " * and xe"^^,
along with
cos 3x and
sin 3A:. T h e complementary function is
u = cie'^ + c^xe'^ + Cz~^ cos "hx + c^e"'

sin 3A:.

T o find a particular integral of the complete equation we make the


change of variable y = e^w. This transforms the given equation into
^ [ 4 ] _^ 2^'^' + dw" -

22;' + 13J = 2x^ -

3x + 1.

154

Linear Equations with Constant Coefficients

T h e substitution
w = qox^ + qix'^ + qzx + 73

is now found to fulfill this equation if qo = 2, qi = 6, 72 = " " 3 , a n d


73 = 4. Thus
y = tf*{2x^ - 6x^ - 3x + 41
is a particular integral. B y adding the complementary function, the
general integral is obtained.
PROBLEMS
Find the general integral of each of the following differential equations.
81. u''> - 2u" - u' + 2u = 0.

82. 1^1 + u " -

83. u^^l + 6 u " + 5u' -

84. uf'I -

85.

86.

12u = 0.

2ul''J _ 9 " + 2u'

4' -

4u = 0.

6 u " + 3ii' + lOo = 0,

+8u=0.

- 21u" + u' + 20u = 0.

87. a'^J + 2uf^] + l O u " + 18u' + 9u = 0,


88. uf^J -

2u'^' + 5u" -

89.

2u" -

8u' + 4u = 0.
90.

+ u" + 8u' -

91. u'^l - f 6 u " + 8u = 0.

92.

+ 6 u " + 5u = 0.

93.

94, y^^i + 3>" -y'

+ 2/' -

3u' + lOu = 0.

3>' =

X.

95. y [31

3;-" + 3^' - ;r = 5x -

97. y^^^ -

7>" + \9y' -

98.

3y" +y'

96. ^1^1 -

lOu == 0.

-3y

=* H

Sy" + 16>' = 4x + 2,

13)- = tf2'{2 cos x + 6 sin x}.

-\-5y = ^'(4 cos 3x + 33 sin 3x}.

99. y^*i + 4^f^l + 8;-" + Sy' + 4; = 1.


100.

8>'31 + 26>" -

40;-' 4- 25> = 5,

ANSWERS TO ODD-NUMBERED PROBLEMS


3. u = if2.='f--'> + * - ^ f ' - i M

I. =
5. u = [(2 -

/2:

'\/2)x + l U " ^

9. u = i2 - x ^ 3x-9
13. u = f-'|4 cos

7. w = 0.
11.

6 sin x}

15. u = ^''^-{cos 2x -

f s i n 2xj.

17. u = cos [5x 15

19. u a 0.

21. u = ^2^^^^' cos {7x - 7 } .

23. u = - Z i ' t ' ^ - ^ ' cos i^rx -

25. y =

27. y =

+ ,i,2r

2g'

ci

sin x + 2 cos x.

Answers to Problems
29. > -

155

U + 2 } - ' log {x-\-2\ + {cix -h czU-'.

31. ^ -

+ C2f.-2*

2*-' +

2 + cos V 2 * + ^ sin

33. jy

-N/2
35. ) r - sin

+ ^ cos 3x + 2 sin 3x.

37. > ^ x V + S i x + l l * ' ^ ^


41. ^ = ^e' +

39. > =
43. y = e-

X +

^2 cos x.

1.

+ f2*

45. y - T^x

sin

+ e~^^\ei sin 3x + ci cos 3x)

47. ;y =
49-

y = ^x^ +

51, J- =

+ -s

e^{ci cos jf +

-1 + ^ log {1 +

sin

x\.

log |1 + ^} + cie' + C2fi~'.

- 4*2 _ 12.

55.

>

57. >

59.

> =

61. y

63.

53. ^ =

65. y
69. >

4x' + 24x +

Sin

+ 2x

71. >

cos

7 j sin

73- > = Tff{si" 2x -

{2x*-i-x^

=i

67, > = sin

+ 4*-'.

2x* + 3*^

=- X * -

+ \2x + \U bx

_
X +

2 cos

x.

X -

5x + 8} cos

X.

2 cos 4x1.

75. _y ^ {sin 3J: + 2 cos 3x)f* + [2 sin x 5 cos x]e


77. >

5^(sin x + sin 2JC + sin 3jej.

79. y

ijjt' +

+ 4JC + 21 + ^{2 sin X +

81. u

cos x).

83- tf - cid^ + cr-^^ + ^^^^

85.
* + C2^e * + ^3 cos 3x + Ci sin 3x,

87. a
89.

.2x sin x.
+ C2"'2x
^^ cos X + cz^^

u =- ci^

91. u ^ ci cos 2x + ^2 sin 2x + f3 cos ^2x


93. :y -

- - i x ^ - |x + o + -:2^ + -:3^"^^

95. > - x^ +
97- > =

+ ^4 sin \ / 2 x .

sin

X +

3 + cie^ + ^2^^ + c ^ x V .
^itf* +

C2fi^^ cos 2x + c^^^ sin 2x,

99, ^ = T + ci~' cos X + c^fi"^ sin X + czx$~^ cos x + axe^^ sin x.

CHAPTER

9
Applications of
Linear Differential Equations
of the Second Order
9.1. Simple harmonic motion
A particle moving on a straight line is said to be in simpU harmonic
motion if its acceleration is constantly directed toward a fixed point of the
line and is proportional to the distance (displacement) from that point.
T h e fixed point is called the central point of the motion. T o be directed
toward this point the acceleration must be negative when the displacement is positive, and vice versa. It is therefore a restoring acceleration.
W h e n the origin is taken at the central point, and the displacement is
denoted by
the acceleration is d^y/dt^ and is proportional to y with a
negative constant of proportionality, say ^. T h e differential equation
for simple harmonic motion is thus
(9.1)

d^dt^-^-'^y

0.

This is a differential equation of type 8.1, with / i n the place of x.


Its auxiliary equation is
+
= 0, and has the complex roots +i.
By (8.11), therefore, its general integral is given by any one of the formulas
y = C2 sin {t H- ci\,

(9.2)

y = C2Cos {t + ci\,
y c\ sin t + C2 cos t.

T h e constants ci and cz can be determined to give the solution w h i c h


fulfills the conditions
(9.3)

:v(/o) =>o,

v{t) = vo,

w i t h any assigned values of /QJ >0J and VQ.


It is clear from formulas 9.2, for instance from the first one, that, as /
increases, y goes through a cycle of positive and negative values, and that
156

Simple Harmonic Motion

157

this cycle is repeated again and again. T h e motion is therefore oscillatory, that is, vibratory, and periodic. Since the largest and smallest values
of a sine or a cosine are 1 and 1 , the largest value of ^ given by either
of the first two formulas 9.2 is \c2\. This m a x i m u m displacement is
called the amplitude of the motion. T h e motion completes a cycle i n the
time it takes the value [0t -\- ci\ to increase by lir. This interval, 2ir/0,
is called the period of the motion. Thus
Period = Irr/^.

(9.4)

T h e number of cycles completed in unit time is the reciprocal of the period.


This is the frequency of the motion. Thus
Frequency =

(9.5)

27r'

T h e character of the variation of y w i t h time / is shown by F i g . 23.


indicated values cj and C2 are those of the first formula 9.2.

The

Fio. 23.

Example 1.

A particle is i n simple harmonic motion with y = "3, and


y = 0, at / = 0. These same values apply to y and v again 3'^ second
later, but not before then. T o find the formulas for y and v at any time.
If the second formula 9.2 is used to describe the motion, the conditions
given take the form
C2 cos

Ci =

3,

C2^

sin

ci =

0,

C2

cos

= 3.

T h e second of these three equations is fulfilled if ci = 0. T h e first is then


fulfilled if C2 = 3, and the last if/3 = AT. T h e required formulas are thus
= 3 cos 4ir/,

v =

127r sin Airt.

Applications

158
Example 2.

A particle in simple harmonic motion passes through the


central point at ( = 1, and at intervals of
second thereafter. W h e n it
does so at / = 1, its velocity is 10, T o find the amplitude of the motion.
T h e interval between passages through the central point is a half period.
T h e period is thus % second; hence, by (9.4), /3 = 5ir, T h e first formula
9.2 therefore gives
y = 2 sin I Sirt +

],

v ~ Sirc2 cos {Sirt +

T h e conditions^(1) = 0, v{\) = 10 require that


2 sin {5ir + i]

= 0,

T h e y are fulfilled by ci = 5T, and

5ir^2 cos {Sir + ^ri) =

= 2/x.

y = (2/T) sin Sirj/ -

10.

T h e formula is thus

Ij.

and from this the amplitude is seen to be 2/T.


PROBLEMS
1. A particle in simple harmonic motion with the period T has the position > 2,
and the velocity P = 4, at ( 0, Find the amplitude of the motion,
2. A particle in simple harmonic motion with the amplitude 8 passes through the
central point with the velocities 1 0 , F i n d the frequency of the motion,
3. A particle is in simple harmonic motion with a frequency of 10 cycles/sec. and
the amplitude 7, What is its speed as it passes through the central point?
4. A particle is in simple harmonic motion. Its velocity as it passes the central
point is twice its velocity as It passes the point y =
Find the amplitude of the
motion,
5. A particle is in simple harmonic motion with the amplitude 6 and the period x / 4 .
With what velocity does it pass the point ^ = 3?
6. A particle is in simple harmonic modon with the period 4. At f
0 its position
is given b y ^ ^
and at / = 1 it is given by = ^ i Find the formula for the position
at any time.
7. A particle is in simple harmonic motion with the period p. When its position
is given by J' =
its velocity is .v\. What is the amplitude of the motion?
8. A simple harmonic motion has the frequency 10. At / = / i , the velocity is
and ji second later the velocity is
Find the formula for the velocity at any time,
9. The motion of a particle is along a straight line. Its acceleration is always
directed away from the origin on this line, and is equal to the distance &om the origin.
Find the general integral of the differential equation for this motion.
10. If in the motion described in problem 9, ^ = 2 ,
formula for the distance at any time?

and r = 2, at ^ =* 0, what is the

Free Vibrations

159

11. Ifj in the motion described in problem 9, the position and velocity of the particle
at ( = 0 are y
S and :^ = 5, find the position which the particle approaches as a
limit when the time increases Indefinitely12, If, in the motion described in problem 9, the position and velocity of the particle
at f = 0 are 7 = 3 and v ^ 0, and at / = 1 the position is ^ = 5, find the formula for
the position at any time.

9.2.

Free vibrations

M a n y mechanical systems include a heavy particle which vibrates with


respect to some coordinate when it is displaced from its equilibrium position. W i t h the varying coordinate taken to be^, Newton's law of motion
3,2 and the relation 3,3 give
- 5 = / ,

g dt-

where / is the force acting i n the jf-direction. T h e following discussions a,


bj and c apply to such mechanical systems.
a. A coil spring (or an elastic string or wire) resists stretching by a
force T which is proportional to the amount by which it is already
stretched. Thus, if its unstretched length is /o, its backpull when it is
stretched to the length / is
(9.7)

/o(,

T = k\l-

with some constant k. This value T is also the pull which must be
applied to stretch the spring to the length /, for the stretching ceases when
this pull just balances the resistance of the spring.
If the spring hangs vertically from a point of support
and a particle
of the weight w is suspended from its lower end, it stretches to a length / i ,
where
(9.8)

w -

k[h

/o|.

I n equilibrium, then, l\ is the length of the system. F r o m w and l\ the


value of k is determinable through relation 9.8. T h a t is the value of k
in (9.7).
Consider the particle now when it is displaced by the amount^ from its
equilibrium position, as shown in F i g . 24. T h e resultant of the upward
force of the spring and the downward pull of gravity is T ~ w. B y
(9.7) and (9.8), this resultant is k[l - h], that is, {w/{h - h)]\l - h\.
Since li ~ I = y, the resultant is wy/{lx /o), and the equation of
motion 9.6 is therefore
g__

160

Applications

Because this equation is 9.1 with


= g/{l\ ~ lo)j the vibration of the
suspended particle is seen to be simple harmonic, with the equilibrium
position as the central point.
Example 3.

A certain spring, shown i n F i g . 24, stretches by 6 i n . under


a pull of 4 lb. A 1-lb. particle is attached to it, and is released at / = 2,
with the velocity 0, from a point 2 i n . above the equilibrium position.
T o determine the formula for the position of the particle at any later time.
I n the foot-pound-second system, equation 9.7 holds with 7" = 4, and
{/ lo] = M ' T h e spring is therefore one for which k = 8. F o r the
S

I
1

i
Fio. 24.

1-lb. particle the equilibrium position is, by (9.8), that i n which {/i lo]
= -ff. T h e equation of motion (taking g = 32) is thus
d'^y/dt^ + 256; = 0,

Its integral, for which y{2) = i , v{2) = 0 , is


= i cos 16/ - 32}.
F r o m this evaluation of y the amplitude of the motion is seen to be -B- ft.,
and the frequency is 8/ir cycles/sec.
b. A body floating i n a liquid is acted upon by the usual downward pull
of gravity, but also by an upward force called its buoyancy. This buoyancy
is due to the supporting action of the liquid, and is equal to the weight
of the liquid which the body momentarily displaces. T h e position i n
which the body Hoats in equilibrium is that in which the buoyancy just
balances the weight. If the body is displaced from the equilibrium
position it bobs up and down with an oscillatory motion.
For a body which is so shaped and ballasted that the amount of liquid
it displaces is proportional to the depth s to w h i c h it is immersed, the
buoyancy is given by a formula B = ksj with some constant k. T h e
weight w and the depth so of immersion in equilibrium are thus connected
by the equation w = ksQ. W h e n the body is displaced so that its depth
of immersion is
as in F i g . 25, the resultant upward force is B ~ w.

Free Vibrations

161

This is k\s SQ], that is, {w/so){s SQ].


It is therefore {w/so)y,
y = SQ Sj and the equation of motion is thus
d'y/dt^

if

(g/so)y.

This determines the motion of the body to be simple harmonic.

'

^
Fio. 25.

F r o m an analysis of Fig. 26, the differential equation for the motion


of a simple pendulum of length / is found to be
d'0

(9.9)

dt

+ ^ sin
I

= 0,

where 6 represents the angle between the pendulum and the vertical.
Equation 9.9 is not of form 9.1. T h e motion of a pendulum is therefore

FIO.

26.

not simple harmonic. However, it is nearly so if the angle of swing


(measured in radians) is small, as it is, for instance, in a pendulum clock.
T o show that the motion is nearly simple harmonic we draw from the
third of equations 5.8 the relation
- sin ^ =
3!

5!

+ i : 0^ - T T 5^ +
7!
9!

162

Applications

This is an evaluation of the change that is made if sin 9 is replaced by $.


T h e relative change, namely, the ratio of {$ sin 0] to $ itself is thus
^
(9.10)
^

smB
$

1 ^
= ^2
3!

1
5!

.
e*

42

I n (9.10) each quantity within a brace is positive when 0 ^ir.


W e see,
therefore, that the left-hand member of (9,10) is less than the first term
on the right. T h e relative change made i n replacing sin 6 by $ is thus
less than 6^/31 This is less than 0-5 per cent when \d\ < 0.175 (i.e., 10),
and less than 2 per cent when 6 < 0.35 (i,e-, 20^). W i t h only such
relative errors, therefore, equation 9.9 is replaceable by the differential
equation
d^e g
dr

T h e variation of 0 i n accordance with this is simple harmonicPROBLEMS


13. A coil spring ha5 the unstretched length 2 ft. When a 4-lb. particle hangs from
it the equilibrium length is 4 ft. T h e particle is pressed down until the spring is 6 ft.
long, and is released from that position, while it is at rest, at time i = 0. Find the
formula for the position of the particle at any later time /,
14. A piece of steel wire 100 ft. long stretches by 6 in. when a 50-lb. particle is
suspended from it. What is the frequency with which this suspended particle oscillates
when displayed vertically from the equilibrium position?
15. A 1-ft. cube of wood is ballasted so that it floats with a horizontal base in water.
In equilibrium it is just half immersed. If the block is raised 1 in. from the equilibrium
position, and is there released while it is at rest, what are the amplitude and period
of its resulting oscillation?
16. If the block described in problem 15 is pressed down until its upper face is at
the water surface, and is then released while it is at rest at time f = 15, what is the
formula for its velocity at any later time i? What b the highest position to which it
will rise?
17. A simple pendulum swinging in a small arc without resistance completes
the cycle of its motion in 1 sec. What is the length of this pendulum?

}^

18. A certain coil spring is stretched 1 in. by a pull of 10 lb. A particle of weight w
lb. suspended from it oscillates with a frequency of 5 cycles/sec. What is w?
19. A buoy in the shape of a right circular cylinder is ballasted so that it floats with
its bases horizontal. Its period of vertical oscillation is 1 sec. What is the depth of
immersion at which it floats in equilibrium?
20. A simple pendulum swinging in a small arc without resistance passes through
the vertical position at t ^ 2, It has at that instant an angular velocity {dB/di) ofta
radians/sec. What is the formula for its position 6 at any later time /?

163

Damped Vibration

21. T h e length of a simple pendulum is


ft. At ( = ti, the pendulum has the
position 5 = "5 radian, and its angular velocity is zero. What is the value of 6 at the
instant % second later?
22. A pull of 5 lb. stretches a certain coil spring by 2 ft. A particle of weight 5 lb.
hangs from it in equilibrium, and while in this state it is given a hammer blow which
gives it a velocity of 6 ft./scc.
Find the formula for the position of the particle / sec.
later.

9.3. Damped vibration


W h e n a mechanical system whose free vibrations would be simple
harmonic moves in the presence of a force which resists the motion, the
motion is said to be damped. If the resistance is proportional to the
velocity it appears i n the right-hand member of the equation 9.6 as a term
rv, namely, as r dy/dt, w i t h some positive constant r. W e shall call r
the coefficient of resistance. If the restoring force of the system is ky^ the
equation of motion 9.6 is
(9.11)

-2=

-ky

g dt'

'

dt

which is
d^y

(9.12)

dy

^ 2 +^^^

+ ^^ = 0^

with
(9.13)

T h e general integral of equation 9.12 is given by formula 8.9 or 8.11,


w i t h ^ and / in the place of u and x. I n these formulas mi and m2 are the
roots of auxiliary equation 8.8, and a and /3 are the real and imaginary
parts of m i . As we shall see, the character of the motion depends upon
the relative sizes of a and 6, that is, of r and k.
CASE

(9.14)

1.

a^ < 4b.

I n this case the roots mi and

m2

are complex, with

/3 = i -N/46 - a^.

a = -a/2,

The general integral of the equation of motion is therefore given by any


one of formulas 8.11, that is, by
y = ae"'

(9.15)

cos {^t

y = ca^'^'sin

y = e"^{ci sin ^t +

C2 cos

164

Applications

T h e first of these formulas shows, through the factor cos {^t + ^1!, that
the motion is oscillatory with the frequency ^/2ir. It has i n the place of
an amplitude constant the factor cj^"'. This decreases as t increases, since
a is negative. T h e motion is therefore a vibration w i t h a fixed frequency,
but w i t h an amplitude that diminishes toward zero with incresising time.
T h e graph of such a motion is shown i n F i g . 27. T h e exponential tf"',
that is,
is called the damping factor. A n increase i n r produces a n
increase i n a and a decrease i n ^. Increasing the resistance therefore
hastens the damping, and decreases the frequency of the motion.

Fio. 27.
Example 4.

A certain coil spring is stretched 1 ft. by a pull of 3 f l b . A


5-lb. particle is suspended from it, and is set to vibrating. T h e air resists
the motion w i t h a coefficient of resistance ^. T o find the frequency of the
motion.
T h e value of k for this spring is
T h e equation of motion is accordingly
_5^^__29
ndt^

_ 1 _ ^
~ Sdt'

T h e general integral of this equation is


y = ^26-2^/5 cos {^t + Ci].
F r o m this formula for> the frequency of the motion is seen to be 12/5T.
CASE

2.

a^ ^ 46.

I n this case the roots of the auxiliary equation are

Damped Vibration

165

T h e y are, therefore, real and negative. Hence


(9.16)
V =

micif"*!* + m^2t^^\

if a > 4,

and
(9.17)
V ^

{C2 + mici + miC2/)^'"*',

if a'^ = 4b.

Since a l l the exponentials i n these formulas decrease toward zero as a limit


when t increases, the values of y and v are damped out. N o right-hand
member of a formula 9.16 or 9.17 vanishes for more than one value of
T h e moving particle therefore passes through its equilibrium position once
at most, a n d changes its direction once at most. T h e motion is clearly
non-oscillatory.
Example 5.

A certain coil spring is stretched 8 i n . by a pull of 1 l b . A


1-lb, particle is suspended from it, and the system is immersed i n oil for
which the coefficient of resistance is -y. T o determine the motion for
w h i c h ) ' = 1, and v = 0, when ( 0.
T h e equation of motion (with g 32) is

Its general integral \&y =


given conditions is

\ d^y

1 dy

32 dt^

2^

2dt

-1- C2e~^^\

y = "^e

and the integral that fulfills the

se

I n the right-hand member of this formula the first term is obviously always
larger than the second when / > 0. T h e y both approach zero as t
increases. T h e moving particle thus subsides, without any change i n the
sign of y, to the equilibrium position.
Example 6.

T o determine the motion of the system of example 5, for


which)* = 3, and v = 4 4 , when / = 0.
T h e integral of the equation of motion that fulfills these conditions is
y -

-e-*'

+ 4e-'^',

v = Ae-'' -

48^-*^'.

T h e moving particle passes through the equilibrium position ) = 0, i n


this instance when ~e~^^ + 4^~'^' 0, that is, when t = ^]og2.
It
changes its direction when Ae~^ 48^^'^' = 0, that is, when / = ^ log 12.
Thereafter it subsides to the equilibrium position.

Applications

166

PROBLEMS
Solve the following problems, assuming the units to be feet, pounds, and seconds.
Take g - 3223. A certain coil spring is stretched 2 ft. by a pull of 1 lb. For a 1-lb. particle suspended from it the coefficient of resistance is x V F i ^ d the formula for ^, if >
3,
and p = 0, when / = 0,
24. For the system of problem 23, find the time when
when ( = 0,

= 0, if;' = 0, and v = \2,

25. A certain coil spring is stretched 3 in. by a pull of 1 lb. A 2-lb. particle is suspended from it, and for this the coefficient of resistance is 1. Find the formula for
i[y(0) = ^'o, and v(0) VQ.
26. A certain coil spring has a 3-lb. particle suspended from it. This stretches it ^ ft.
T h e coefficient of resistance is x- Find thcformula for y^ if >(0) = 0 and v{0) = 7.
27. For the system of problem 26, find the formula for v if >(0)

3 and r(0) = 2,

28. For the system of problem 26, find the frequency of the motion, and find what
the frequency would be if the resistance were not present.
29. Derive the differential equation for the approximate motion of a simple pendulum swinging through a small arc, if there is a resistance of which the coefficient is r.
30. A simple pendulum is 32 ft. long, and has a bob weighing 4 lb.

What Is the

smallest coefficient of resistance for which this pendulum will not oscillate?
31. A simple pendulum has a bob weighing 1 lb. It is to swing in a small arc with
a coefficient of resistance xtIts frequency is to be -y. What must its length be?
32. A particle moves on a horizontal line against a resisting force which is proportional to its velocity. N o other force acts upon it. Find its position at any time if
y ^ yoi and v ^ VQ, ahen ( ^ 0.

9.4. Forced vibration without clamping.

Resonance

A mechanical system i n w h i c h a l l the forces are internal, as i n those


considered i n Sections 9.2 a n d 9.3, is called a free system. Its motions are
said to be free motions. A system upon w h i c h an outside force acts is called
a forced system. W h e n a force /(/) acts upon a system w h i c h i n the free
state obeys equation 9,11, the dilTerential equation for the resulting forced
motions is
g dt

dt

W i t h a and b determined by relations 9.13, this equation is


(9.18)

at

+ a

dt

J+

Forced Vibration without Damping.

Resonance

167

This is a complete differential equation. It can therefore be solved by


finding a particular integral and adding the complementar)' function to it.
T o find a particular integral, any one of the methods of Sections 8.5, 8.6,
8.8, or 8.9 may be used.
T h e case in which j{t) is a pulsating force which varies harmonically in
accordance with a formula
(9.19)

/(/) = Ksin \yt + 5}

is of special interest. F o r a system without damping, whose free motion


would accord w i t h equation 9.1, the equation of motion is then
(9.20)

-J+^^y
dt

~sm{yt-{-d

T h e character of this motion depends, as we shall see, upon the relative


values of/3 and 7, and especially upon whether or not ^" and 7^ are equal.
1. 7^
jS^. A particular integral of the differential equation
(9.20) , may be found i n this case, by the method of undetermined coefficients of Section 8.9, to be
CASE

T h e general integral is thus


(9.21)

y =

^ s i n 7^ + 51 +

sin {^t + d}.

For each possible motion, c\ and


have specific values.
A motion 9.21, for which C2 = 0, is simple harmonic. It has the same
frequency and phase as the impressed force, but ordinarily it has a diff'erent
amplitude. A motion for which C2 7^ 0, on the other hand, is compounded of two motions with different frequencies.
It is therefore less
simple. If the two frequencies arc commensurable, namely, if ^/y is a
rational number, the motion is periodic and executes a certain pattern i n
the course of a period. T h e pattern is, however, more intricate than that
of simple harmonic motion. Figure 28 shows the graph of such a motion
for which ^/y = 4. T h e period of this motion is 27r/7.
A motion for which
5^ 0, and ^ and 7 are incommensurable, is not
periodic but has an aspect of being erratic. However, even such a motion
is bounded, for neither of the sine functions in formula 9.21 takes on a value
numerically greater than 1. Hence l^^l never exceeds the value

168

Applications

T h e range of values w h i c h ^ may take on is, however, large when {/3^ 7^)
is small.
2. 7^ ==
RESONANCE.
A particular integral of differential
equation 9.20 is in this case found to be
CASE

2w

cos {t + a).

and the general integral is accordingly


(9.22)

~gKt

2w

cos \t + 51 + C2 sin {t + Cl]

T h e motions which accord with this formula are quite dissimilar from
those of Case 1, for the multiplier of the cosine term in (9.22) contains / as
a factor. T h e amplitude of the motion therefore increases indefinitely as

Fro. 28.
t increases, which is to say that the motion becomes increasingly violent.
A n actual mechanical system is i n this case pushed to the bounds of its
physical limitations, the point where it must break down. This phenomenon is called resonance. It results when the impressed force is of the same
frequency as the free vibrations of the system. A marching army is said to
break step i n crossing a bridge to forestall such strains as may result from
actual or approximate resonance between the frequency of the footbeat
and that of the free vibrations of the bridge.
T h e application of a vibrating force 9.19 to a particle suspended from a
spring, as i n F i g . 24, affords a simple instance of a system in forced vibration. A way i n which this can be realized practically is to make the point
of support S of the spring itself move vertically in a simple harmonic manner. T h u s if S is raised by the amount F, the length of the spring w i l l be

Forced Vibration with Damping


{ y + / i y]then be

169

T h e equation of motion for the suspended particle will

w at

1 ' 0

If y = A sin 7^ + 5 j , the previous equation becomes

This is clearly of form 9-20,


PROBLEMS
Solve the following problems, assuming the units to be feet, pounds, and sccondsTake g = 32,
33, A particle of weight 2 is subject to a restoring force which is numerically equal
to
of its displacement.
It has also the outside force / = sin 2l impressed upon it.
Find the motion for which ^ = 0, and v = -^^ when / = 0.
34- Find the motion of the particle in problem 33 for which)' = ^-^andv

= 0, when

/ = 3x/4.
35, A particle of weight 32 is subject to a restoring force which is numerically equal
to 9 times its displacement.
It has also the outside force / = 2 cos 3f impressed upon
it. Find the motion for which ^ = 2, and = 0, when t = 036, A particle subject to a restoring force vibrates freely with a frequency v greater
than 1, When the outside force / = cos / is impressed upon it, the motion for which
y yu and v = 0, when ( = 0, is simple harmonic. Find the value of ^ i ,
37, Find the general integral of the difTcrential equation
fy ,
.
. . . . 7
- ~ + /37 = 2 sin 7' + 5 sin ^ I,

(9.23)

assuming that the motions arc not those of resonance.


38, Show that there are two values of y for which the motions defmed by differential
equation 9.23 are resonant. Find a particular integral of the differential equation for
each of these values of 7.

9.5. Forced vibration with damping


W h e n a damping resistance is present i n a forced vibrating system the
law of motion is given by (9.18) with (9.19), that is, by the differential
equation
(9.24)

at

+ ^ ^ 4.
at

= ^ sin 7/ + |,
w

170

Applications

with a > 0, T h e roots of the auxiliary equation are then either real and
negative, or complex with a negative real part. T h e complementary
function therefore approaches zero as a limit when t increases indefinitely.
Every motion of the system accordingly subsides into the motion that is
represented by the particular integral. This is called the steady-state
motion of the forced system. As it is found by the method of undetermined coefficients, its formula is
y =

rl2

2 2>

-y') sin (T + 5) -

cos (7 +

5)1.

W e can apply to this formula the general trigonometric relation


(9.25)

^ sin i>

feos

f^sin

t a n - ^ (5/^4)1,

to give it the form


(9.26)

= ^ s i n | 7 i + ( 5 - 61)!,

with
(9.27)

K =

^'^

=y

; V ( e - T ' ) ' + '7'

5i = t a n ' ^
^--^

F o r m u l a 9.26 shows that the steady-state motion is simple harmonic, w i t h


the same frequency as the applied force. T h e motion is, however, not i n
phase with the force, for, by (9.19) / = 0, when / = (nx 5)/7, w i t h n
any interger; whereas, by (9.26)
= 0, when / = \m (6 5i)l/7,
that is, at times w h i c h are later by the amount 61/7.
T h e amplitude K of motion 9.26, though it is given by 9.27, can also be
put into the alternative form
(9.28)

K = w

If

^ 0, the value of K always decreases when 7 is increased,


that is, when the frequency of the impressed force is raised. O n the other
hand, if
6 < 0, the value of K first increases to the m a x i m u m
(9.29)

^'^

and then decreases. T h e m a x i m u m is attained when 7 = V 6


For this value of 7 the system is said to be in resonance. T h e resonance

Forced Vibration with Damping

171

amplitude 9.29 is evidently large if a is small. Unless sufficient damping


is present, therefore, the breakdown of a mechanical system at resonance
results. T h e frequency at w h i c h an applied force is i n resonance w i t h a
damped system is
1

(9.30)

^TZl.

This is somewhat less than the frequency of the free vibration of the systern, w h i c h is (1/2T) '\^b
PROBLEMS
39. F i n d the formula for y in the steady-state motion given by the differential
equation
dh

dy
+ 2 ^ + 10, = cos 2/.

40. Find the formula for y in the steady-state motion given by the difTercntial
equation
+ 2-~ + II7 = sin yt.
dt^ -dt
if 7 has a value for which the system is in resonance.
41. Find the general integral of the differential equation

^ + 3 -

2,=sm

and put it into the form that displays the phase difference 2i between the force and the
steady-state component of the motion.
42. Find the formula for y in the steady*state motion given by the differential
equation

when y has such a value that the phase difference b\ between the force and the motion
has the value ir/2,

43. Find the general integral of the differential equation

Observe that all the motions here represented subside completely-

172

Applications

44. Find the general integral of the differential equation


^
at
with k > 0.
vibration.

sin t,

Observe that as / increases all the motions subside into a simple harmooic

9.6. The forced vibration of a system with a repelling force


If a particle that is subject to a central repelling force that is proportional
to the distance is acted upon by an outside force 9.19, the differential equation of motion is again of form 9.24. However, the value of is now negative. T h e auxiliary equation therefore has real roots, and of these one is
positive and one is negative. A particular integral is again given by

Fio.

formulas 9.26 and 9.27.


formula
(9.31)

29.

T h e motions of the particle are thus given by the

Ksm{yt-\-

8 -

5i\

-\- cie""^* +

Cie""^'.

N o w one of the exponentials i n this formula increases w i t h t. A n y motion whose formula 9.31 includes that exponential is therefore unbounded.
U n d e r it the particle flies off in the positive or the negative direction. T h e
motions whose formulas 9.31 do not include the increasing exponential
are different. They subside into the simple harmonic steady-state motion
given by (9.26).
A simple example of a mechanical system of this type follows. A
smooth straight rod is mounted to rotate i n a vertical plane about one of
its points 0. It carries a heavy particle which is free to slide eilong i t
without friction or resistance. T h e system is shown i n F i g . 29, where y
denotes the angular velocity of the rod in radians per second. T h e rotation of the r o d generates a centrifugal force upon the particle, w h i c h
amounts to {ivy'/g)yj and repels it from 0. Gravity is a n outside force

Motion Under an Intermittent Force


whose component along the rod \s w sin {yt + 6].
motion of the particle along the rod is therefore
ID d\

17.3

T h e equation of

wy^
y ~ w sva. [yt -\- h\J
g

gdt'

that is,
(9.32)

= - 5 s i n ( 7 / + 5!.

dt

Some of the motions defined by this equation are unbounded. I n them


the particle flies off from one or the other end of the rod. T h e other
motions subside into the simple harmonic steady state
(9.33)

y -

- % s i n [yt ^-h]

2y

T h e polar coordinate equation r 4 sin ^ represents the circle with its


lowest point at the pole and with the radius {}'i)A. Equation 9.33 isof this
form. I n the steady-state motion the particle therefore describes a circle
in the vertical plane, the lowest point of the circle being at 0, and its radius
being g/Ay \
PROBLEMS
45. Find the formula for^ in the motion defined by equation 9.32, for which ^ = 0,
and 0 = 0, when ( 0.
46. Find the formula for D in the motion defined by equation 9.32, for which y
g/3y^, and v = g/^y^ when / = 0.
47. Find the formula for y in the general motion defined by the differential equation
(9.34)

+ 6

16> = - 3 2 sin 4t

48. Determine what the relation between ^(0) and v{0) must be in order that the
motion defined by differential equation 9.34 may subside into the steady state.
49. Find the formula for y in the general motion defined by a differential equation
9.24, in which b ^ 0 and a > 0. Observe that every motion in this case subsides into
a simple harmonic steady state, but that these do not all have the same central point.
50. Find the formula for y in the general motion defined by a differential equation
9.24, in which 6
0 and a 0. Observe that some motions are unbounded and the
others are simple harmonic.

97.

Motion under on intermittent force

A force /(() that must be described by different formulas i n different


parts of the total time interval is said to be intermittent. A particular

174

Applications

instance of an intermittent force is one which acts during only a part of the
time. T o such a force may be assigned the value zero while it does not
act. A n intermittent force may well be discontinuous. E v e n when / is
discontinuous, however, formulas 8.25, 8.26, a n d 8.27 give an integral for
diflferential equation 9.18, w h i c h is continuous and has a continuous
derivative. T h e general integral is obtainable, as usual, by adding the
complementary function.
Example 7.

F o r a certain particle the differential equation of motion is

at

2 - + 5> =
-fit),
at
w

w i t h fit) = e^ from / = 1 to / = 2, and otherwise /(/) = 0. T o find the


integral of this equation for w h i c h ^(0) = 0 and p(0) = 6.
T h e auxiliary equation has the roots 1 2i\ hence formula 8.27,
w i t h X and fix) replaced by t and ig/w)f{t), gives a particular integral. I f
we choose to = 0, since the conditions apply at that time, the formula is

2w Jo

N o w , because/(j) = 0 when j < 1, and when j > 2, this formula is more


explicitly
0,

(9.35)

yp(t) =

\2w

when

0 g / ^ 1,

ds,

when

1 ^ / g 2,

e^'-^ sin 2it - s) ds,

when

( 2.

sin 2it~~s)

j'^

2w

T h e general integral is
y = ypiO

sin 2t + C2 cos

2t],

and gives
p = ^

dt

e-^{i-ci

2c2) sin 2t +

(-^2

2ri) cos

2t].

Since yp(t) a n d its derivative are both 0 at / = 0, the conditions to be


fulfilled are
= 0, C2 + 2ci = 6. T h e solution required is therefore
y = >p(0 + 3(r~' sin 2t.

I n the instance of this example quadratures 9.35 for^p(/) can be made.


T h e resulting formulas are

Motion Under an Intermittent Force


3tf~' sin 2/,

when 0 ^ Z g 1,

^2-'[sin 2(t -

-f- {g' -

Sw

175

1) + cos 2(/ -

1)11 + 3<f-' sin 2t,

when 1 ^ Z ^ 2,

y =
g

8a;

{e*''[sm 2{t -

2) 4- cos 2{t -

2)] - '-'[sin 2{i -

cos 2{t Example 8.

1) +

1)11 + 3~' sin 2/, when t > 2.

T o find a particular integral of the differential equation


^ - y - m .

i f / ( 0 = tjromt = OtoZ = l ; / ( 0 = l , f r o m Z = 1 toZ = 2; a n d / ( 0 = 0,


when t > 2.
F o r m u l a 8.25 is i n this instance
VW = i

pis)

ds,

W-

-<:-'+] ds,

more explicitly,
i pU'~'
i /'

s{e'-' -

when

ds +

/ ; (.'-' - .-'+!

yi') = <

0 g < g 1,

ds,

when

1 g < S 2,

when

/ > 2.

Thus, by (3.9),
sinh / t,
v(/) = I sinh Z - sinh (Z - 1) -

1,

sinh Z sinh (Z - 1) cosh (z - 2),

when

0 ^ Z ^ 1,

when

1 ^ / ^ 2,

when

Z > 2.

PROBLEMS
51. In the difTcrcntial equation

+ 2

I+

2> = /(,).

the function f(t) has the value e"^ from / = 0 to / ^ 1, and is otherwise zero.
the formula for ^, if ^ = 0 and y' = 0, when / = 0,

Find

176

Applications

52. In the difTcrential equadon


dS

dy

the function /(/) has the value 1 from i


1 to / = 2, and is otherwise zero. There is a
solution for which ^ =* ^, and ^' = 1, when ( = 1,
Find the formula for this solution
when ( > 2.
53, Find a particular integral of the differential equation

dt2
if / ( O has the value sec / from ( = 0 to / = ir/4, and is otherwise zero,
54. In the differential equation
d'^y

dy

+ i :-

dt^ ' dt

6> = / ( / ) ,

the function f{t) has the values

fit)

= { 0,

when

0 ^ r ^ 1,

when

1 < / < 2,

when

/ ^ 2,

e~^\
Find a solution of the equation.

55, T h e differential equation of problem 54 has a solution for which > = 0, when
J 2, and which docs not increase indefinitely in numerical value as t increases.
F i n d the formula for this solution when 1 ^ / ^ 2.
56. In the differential equation

2 + 3 T + tv=/W.

dt^ ^ ^ dt

the function /(/) has the value / when ( > 1, and is otherwise zero. T h e equation has
a solution for which > = 0, when / = 0, and for which also y ^ 0 when t ^ 2.
Find
this solution.

9.8. Simple electric circuits


W h e n an electric circuit has an electromotive force E impressed upon it,
a current i flows and a charge q accumulates i n it. A simple circuit is
diagrammed i n F i g . 30. Its elements arc a resistance
an inductance L ,
and a capacity C. T h e values of these are constants. I n terms of an
appropriate system of units (t i n seconds, E i n volts, i i n amperes, q i n
coulombs, R i n ohms, L i n henrys, and C i n farads) the laws that apply are:

(9.36)

I =

do

Simple Electric Circuits

177

and
di

(9.37)

^ 7 / + ^ ' + = ^'

These are derived i n the elementary theory of electric flow.


If the value of i , as given by (9.36), is substituted into relation 9.37, the
result is the equation
1

dq

(9.38)

This is the differential equation for the charge q. If, alternatively, equation 9.37 is differentiated and dq/dt is then replaced by its value 9.36, the
result is the equation
(9.39)
This is the differential equation for the current i.
Differential equations 9.38 and 9.39 are of the same form as equation
9.18 for the motion of a particle. T h i s reflects an analogy between the

WW
R

30.

FIO.

phenomena of electric flow and those of mechanical motion, particularly


between those of alternating currents and those of vibrating particles.
Under an electromotive force E which pulsates in simple harmonic
fashion, there may be a manifestation of resonance, of a steady-state flow,
etc.
Example 9.

F o r a certain electric circuit L = ^Vj


~ 4, and C = -r^ffT o find the formula for q,ifq = go, and i = to at Z = 0, and = 0, when
t > 0.
Differential equation 9.38 is i n this instance
1

do

25 dt 2 + 4 dt + 676? = 0.

178

Applications

Its general integral is


q =

s i n l 2 0 / + <:2cosl20z),

and from this, by (9.36)


= ^-6o<j[_50(:i -

12OC2I

sin 120/ + [120ci -

5OC2I

T h e conditions are fulfilled if ^2 = ?0) 120iri 50^2 =


formula is therefore
'

cos 120/i.
T h e required

sin 120/ + qo cos 120/

Example 10.

F o r a certain electric circuit / , = 1, i f = 40, and C


1/20,000. T o find the steady-state current when E = 120 cos 150/.
Differential equation 9.39 is i n this instance
d^i

di

dt

dt

V 5 + 40 y + 20,000z = - 1 8 , 0 0 0 sin 150/.


It has the particular integral
.
-36 .
z = s m

150/ +

tan-i^

and the complementary function e~^'^^[c\ sin 140/ + cn cos 140/j. Since
the complementary function approaches zero as a limit when / increases
indefinitely, the steady-state formula for i is that given by the particular
integr2il.

PROBLEMS
57. For a certain electric circuit L =

R \i and C = y ^ ^ .

Find a particular

R = 40, and C = TDVO"-

Find the formula

formula for g if 5 sin 140(.


58. For a certain electric circuit L =
for ^ if 7 = iTTOirt

I = 10, at ( = 0 ,

and E 0, when t > 0.

59. For a certain electric circuit L = TTJ-* R ^


tor if - 10^-><" + 50 sin 60t.

and C T O -

Find the formula

60. For a certain electric circuit L = ^11, R = i i r , and 1/C = 0.

Find the formula

for I, if = 5 cos 200/.


^, R = 0, and C = TtnnT-

Find the formula

62. For a certain electric circuit L 5, i? = 0, and C = u^nnj"for i' if => 10 cos 1100/ + ir/A\.

Find the formula

61. For a certain electric circuit L


for I when = 10 cos 180/.

63. For a certain electric circuit Z. = ^, and C = 5a ^.

How large must R be in

order that the current may merely subside to zero without oscillating when is reduced
to zero?

Answers to Problems

179

64. For a certain electric circuit Z. = y , i2 = 20, and C = T S U value of the steady-state current when E = 5 cos 60/.

Find the maximum

ANSWERS TO ODD-NUMBERED PROBLEMS


1. 2 V 2 .

3. 140ir.

5. 2 4 V i .

7. Vyi" + 0'=^^IV4T2)11. > = 0.

9. > = cie' + C2e~'.

13. ^ = - 2 cos V(g/2) t.


17.

15. Amplitude inj, period ir \ / 2 / ^ ,

19. i f t

/ = ^ / T T ' ^ ft.

21. e = 0.1.

23. V = 4^-2* _ ^-8i_

25. H8>o +1-0]/ +>o|^-''.

27. & _= ,-4(/3 |2 cos At -

29. - 2 +
- + f e = 0.
dt^
w dt
I

31. / = 32/(T2 + 1).

33. y = - Y s i n 2/ + 11 sin /.

35. y = it sin 3/ + 2 cos 3/.

sin yt +

37. jy

39. )- =

41. > =

14 sin 4/j.

sin - ( + C2 sin {j3; + ci|.

2/ + t a n " ' I

sm

2 -\/l3

VlO

stn

/ -I- ' 4

21

tan-'3

cos | i + Uit + c s l ^ - ^

43.

45. > =

g
4T

2 sin yt -

+ ^-T'l'

47. > = I sin |4f + tan-^ J} + CK^' + C2^-^'.

49. y =

y^ +
-

51. y =

sm 7/ + 5 + tan
when

cos 1,

^ '|cos (/ 1) cos t\y


53. y

at
0 S / ^ 1,

when

i > 1-

M i n / + cos / log cos /,

when

/ ^T/4,

(ir/4) sin i -ffOog 2) cos /,

when

i > 7r/4.

55, y =
59. i" = 5 sin 60/ +

57. q = TshslA sin 140/ + 3 cos 140/}

sin 10

-\/llr+

61. j" = 40/ cos 180/ + ci sin 180/ +


63. R ^ AO.

fa cos 10 -^/u

cos 180/.

CHAPTER

10
The Linear Equation
with Variable Coefficients
10.1. Ordinary and singular points
T h e general linear differential equation of the second order, i n its complete form, is
(10.1)

po{x)y" -hPiix)/

+P2(x)y

=/,

and i n its reduced form is


(10.2)

Po{x)u" + piixW

p2{x)u =

0.

W e shall suppose that the functions po{x)j piix), and p2(x) have no common
factor. W h e n such a factor is present, it shall be divided out of the
equation.
A point X at w h i c h
(i) the coefficients poix), pi{x), and p2{x) are continuous and
(ii) poix) ^

is called an ordinary point for the differential equation. A point which is


not ordinary is called a singular point. Throughout this chapter we shall
suppose that the differential equation is being considered upon a n x-interval that does not contain any singular point.
As we have already observed i n Section 8.1, if ui{x)y and U2{x) are any
integrals of equation 10.2, the linear combination {ciUi(x) -\-czU2{x)\
with arbitrary constant coefficients ci and C2 is also an integral. If this
combination cannot be expressed with the use of only a single arbitrary
constant, it is the general integral. F o r any equation 10.2, u{x) ^ 0 is a n
integral. This, however, is rarely of any use, and is therefore called the
trivial integral. Other integrals are non-trivial.
If yi{x) is any particular
integral of the complete equation 10.1, the general integral is obtainable
from^i(j>f) by adding the complementary function.
Aside from the equations with constant coefficients, there arc only a few
minor classes of equations 10.1 or 10.2 that are solvable by methods other
than that of power series.
180

The Euler Equation

181

10.2. The Euler equation


T h e differential equation
(10.3)

xol Y

\x -

+ a\x - xo\y' + by = f{x),

in w h i c h a, b, and xo are constants, is called the Euler equation. It has a


singular point at XQ.
Hence we suppose that ;i: > J:O, or J: < XQ.
The
change of variable

and

= xo -\- e",

XQ

e*,

if

if

x <

>

XO,
XQ,

transforms differential equation 10.3 into


d^y , ,
+ a ds^ '

,,dy
11 - f +by=f{xo
' ds

e')

This equation has constant coefficients and can therefore be integrated.


Its integral, when expressed in terms of the original variable x, is an integral of (10.3).
Example 1.
\x -

T o find the general integral of the differential equation


\ \Y

4{x -

l|y -

Uy

= x^ ~

3x2 +

3;^ _

8.

T h e change of variable x = \ + e' transforms the given equation into


d^y
ds^

^ dy

5-f " ds

14> = - 1 e^\

By the method of undetermined coefficients the general integral of the last


equation is found to be
y = s + sjie

cie

-\- cze

T h e integral i n terms of x is therefore


> = i Example 2.

V:^ - 1 r +

- 1 r + C2\x ~~

r'.

T o find the integral of the differential equation


x^u" -

3xu' + 13u = 0,

for which u(l) = 1 , and u'(\) = 1.


I n this instance xo = 0, and the imposed conditions apply at a point
where x > 0. T h e appropriate change of variable is therefore x = e'.
It transforms the equation into

182

Linear Equations, Variable Coefficients

F r o m this equation the general integral is found to be


u x^{ci sin (3 log x) + C2 cos (3 log x)\.

T h e assigned conditions are fulfilled by C2 = 1 , and


required integral is thus

= 1.

The

u = x^{sin log x^ cos log x^].


PROBLEMS
Find for each of the following differential equations the integral fulfilling the given
conditions, or the general integral for the indicated values of x.
1. fjf + 2 | V ' -

20a -

2. \x + 2} V

A\x + 2 } ' -

3. \x -

3} V

U -

4. |x -

5}V'+7{jf -

5. xY

2xy' +ly

6. x-y"

*y' + 5^ -

8. , Y

+^'+

0,

* > 0.
14 = 0,

^\u' + a = 0,

2i,

X < 3.
x<Q.

5 ] ' + 9u = 0,

= flog ;r}^
3,

o(0) = 5, u'(0) = 13.

;c > 0.

<0.

= ' ^ ' ^ " + \

, > 0.

9. \x -

\\y" + 3 / +

, -

1,

,(2) -

\x ~ \\
10. 6 r V ' -Ky'

-yy=

X
12. :ry" +

ly'

6x-^

y{\) -

O.y'CD =

sin log

\'
4

--ft-

13

--y

^,y(2) 4

13

jt > 0.

10.3. Exact equations


A differential equation 10.1 is easily tested for exactness, and is readily
integrated if it is exact. After being multiplied by dx the equation can
be written i n the form
d\poy' -

poy]

d[piy\ +

\po" -

pi

+ p2\y dx = f{x) dx.

T h e term \po" pi -\- p2]y dx is not a differential unless it is zero, for


it involves)! but not dy. T h e condition for exactness is thus
(10.4)

po"(x) -pr'(x)

-\-p2{x)

=0.

The Adjoint Equation

183

I n the event that the equation is exact, a first integral is


po{x)y + pi{x)y =

po{x)y' Example 3.

J / W dx +

a.

T o find the general integral of the differential equation

X-

f1

x^

x'

Condition 10.4 is fulfilled, and the equation is therefore exact,


first integral is
1
X

y =

The

e'-\-c 1

F r o m the last equation the general integral is found to be


y =

\xe'

c\x + C2xe~''.

PROBLEMS
Find the general integrals of those of the following differential equations that are
exact.
13. * V '
14.

{3^2 JfxW

|1 + 2e' + e'^'W

-{-u = 0.
+

1 + li" + Ze'^'W + e^u = 0.

15. u" cos X + u'(csc X sin jt) u esc x cot ;c = 0.


16. >" cos X +
17. xy"

cos

{ J : ^ sin J : ) ; ' ' +

2 +

I.
[x^ cos JT + a: sin J : ) ^ = cos x.

6JC

1 20. y"

Jf =

y =

4x

X,

0.

2)2
Ax^

{1 -

1 + log ^
X log

{x log x|-2 y = 2x.

21. u" + w' tan x -\- u sec"


2JC

^" ^ lT7^^'

AT

2|1 -

0.

x^\

11 + x 2 p ^ "

4x
|1 -Tx^i-

10.4. The adjoint difFerentiol equation.

Integrating factors

After a differential equation 10.1 is multiplied by a function v{x), it is


exact, by (10.4), if
(10.5)

\pQv}" -

[pivY ^-Piv

0.

184

Linear Equations, Variable Coefficients

T h i s is a differential equation for f, which is, more explicitly,


(10.6)

po{x)v" +

\2po'{x) -

Pi{x)W

Wix)

-pi'ix)

-\-p2{x)\v

= 0.

Equation 10.6 is called the adjoint of the given differential equation 10.1 or
10.2. A function v{x) is thus a n integrating factor for a given differential
equation, if and only if it is an integral of the adjoint equation.
T h e adjoint of equation 10.6 is i n turn found to be equation 10.2;
thus each is the adjoint of the other.
Example 4.

T o find the general integral of the differential equation


\x^ -

x\/'

2*2 + 4;f -

3 1 / -h Sxy = 1.

T h e adjoint of this equation is


{x"" -

x\v" -

\2x^ -\]v'

-\- |4x -

2]v = 0,

B y the trial of A:"*, this equation is found to have x^ as an integral. Thus


x^ is an integrating factor of the given differential equation. T h e general
integral, as found by the use of this factor, is
1

y =

1
7 T +

- 5 + *^2

PROBLEMS
Find an integrating factor for each of the following differential equations, and b y
the use of it find the general integral.
4x

23. u" +

Bx -S

7 ' +
2x - \

2x - \\

u - 0.

24. {3x-\-x-\u" + |12 + 4* - X 2^ I..'


H' -

25. {x

26. u " 27. u" +

|4 + Sx\u - 0

1
X

u' tan x -

u{2
+

+x

tan x + tan^ x\ = 0.

~ u - 0.
x{\ +x\

28. 2 sin J: cos x\u" + j? sin x + 4 cos x\u' + lOu cos x = 0,


29. y

1 -

> -

30. / ' + I2x + 11/ + i2x + 2\y -

|1 - x\e-2JC

2*-^

The Existence of a Solution


10.5.

185

The existence and uniqueness of a solution.


equation

The Riccati

It was observed in Section 6.7 that any solution of the R i c c a t i equation


,
^^^H

p2{x)

pl{x)

po{x)

po{x)

^^^^^^^^^^^^^^

2
^?

yields, through the formula

an integral of differential equation 10.2, and vice versa. W e can use


this fact to show that, when XQ is any ordinary point, differential equation
10.2 has an integral for which
(10.7)

U{XQ)

Mo,

u'{xo) = ttu',

and that this integral is unique.


If U(j ^ 0, the Riccati equation has an integral
UO'/UQ.
T h e formula
u{x) = Wo exp /

T)(X),

for which

TJ{XO)

7i(x) dx

Jxo

gives a n integral of (10.2) which fulfills (10.7). If UQ = 0, let ai(x) and


U2(x) be integrals for which i(xo) = 1, i'(xo) = 0, and 2(^0) = 1 ,
"2'(^o) Wo'have just shown that such integrals exist. T h e i r sum,
u(x) = {ui{x) + "2(^)1, is a n integral that fulfills (10.7).
Suppose now that two integrals fulfill any given set of conditions (10.7).
Their difference is an integral C/(x), for which
(10.8)

U{xo) = 0,

U'ixo) = 0.

Therefore both the integrals U(x) + u\{x) and i(x) fulfill the conditions
u{xo) = 1, u'{xo) = 0. N o w each of the functions
+ ui'(x)
''

Lh{x)
Uy{x)

V'(x)

^
and

ui'(x)
U,{X)

is a n integral of the Riccati equation, which at XQ has the value zero.


There is only one such integral; therefore the two functions are equal, that
is, U(x) + ui(x) = kui{x). Since this equality applies at XQ we see that
A 1; hence that U{x) = 0. A n y two integrals fulfilling a specific set
of conditions 10.7 are thus identical; the integral is unique. W e note
especially that the integral fulfilling condition 10.8 is merely the trivial
one.
It was shown in Chapter 6 example 17, that an integral of a differential
equation 10.2 is sometimes obtainable through the use of the Riccati

186

Linear Equations, Variable Coefficients

equation. T h e method can even be given a somewhat more general


form. If i}{x) is an integral of any Riccati equation
(10.9)

7} - <Tix)7)-,

po{x)

<T{X)

in which iT(.if) is a function that is not zero anywhere on the given interval,
the formula
(10.10)

u = exp lj(T{x)ijix) dx]

yields an integral of equation 10.2.


Example

5.

T o find an integral of the differential equation


xu"

\ \u' -

xh'u

= 0.

Riccati equation 10.9 is i n this instance


xe^ -\-\

W h e n a = X, this equation has an integral e'


given differential equation has the integral

Thus by (10.10), the

ui{x) = e

PROBLEMS
By the use of a Riccati equation, find an integral for each of the following differential
equations.
31. u" -

-u'
x

9x*u = 0.

33. 2 u " + \5e' -2]u'

+ le'^^u = 0.

35. u " + I2x cos Jf + tan x\\i' +


36. x u " log X u' xu|Iog x|
38. xu" +

32. u " -

{cos x +
0.

34. xu"

+
+

.8
X -

0,
3. = 0
2 ! ' + x'u

cos" x\u = 0.
37. u

2u' -

36f*' = 0

11 X sin x\u' jsin x + x cos x|u = 0.

10.6. The Wronskian.

Linear independence of solutions

If, on a given interval, two integrals i(x) and U2{x) of a differential


equation 10.2 are such that
(10.11)

aiui{x) + azuzix) = 0,

with some constants a i and 02 which are not both zero, the integrals are
said to be linearly dependent. O n the other hand, if relation 10.11 is fulfilled only when ai and a^ are both zero, the integrals are linearly independent.

The Wronskian

187

Suppose ui{x) and uzix) are linearly dependent. A t any x, equation


10.11 a n d its derivative then yield a pair of homogeneous simultaneous
equations for the " u n k n o w n s " ai and 02. Since these unknowns are not
both zero, the determinant of the system, namely,
lW

U2{x)

l'W

U2'{X)

is zero. This is so for every x. This determinant is called the Wronskian


ofu\ and 2, i n honor of the Polish mathematician H . Wronski (1778-1853).
The symbol for it is W^(i, U2\ x)\ thus
(10.12)

W{U,,

U2\ x)

= i(x)2'(x) -

u{{x)u2{x).

We have shown that, when the integrals u\{x) and W2(^) are linearly
dependent, then W{u\, U2\ x) = 0.
Suppose now, conversely, that the Wronskian of u\ and 2 is zero at a
point XQ.
T h e system of equations
a\ti\{x^) + a22(*o) = 0,

ai"i'(^o) + a2"2'(*o) = 0
is then fulfilled by some certain pair of values ai and f72, of which at least
one is not zero. T h e combination
aiui(x) - f

a2U2{x)

with those values of a\ and 2 is an integral, and it fulfills conditions 10.8.


Therefore it is the trivial integral; that is, relation 10.11 is fulfilled.
Hence, if W{u\, U2', x) vanishes at any ordinary point xo, the integrals
i(x) a n d 2(x) are linearly dependent.
Example 6.
{2x

T h e differential equation
+

\ ]u"

( 4 x 2 +\\u'

has a singular point at x =


2 = e^ . F o r these integrals,

4x2 H- 2x +

2| =

and it has the integrals u i = e~^ a n d

W{uu U2\x) = . * " - ^ { 2 x + 1}.


This value is not zero on any interval that does not contain the singular
point; therefore the solutions are linearly independent on that interval.
Example 7.

T h e differential equation
u" -f- {tan X -

2 cot X]M' = 0

188

Linear Equations, Variable Coefficients

has the integrals ui = sin x -I sin 3x, and 2 = s i n ' x.

H'Xwi,

2 ; x) = 2

F o r these,

sin^ * cos x sin^ xjsin 3x cos x cos 3x sin x].

By trigonometric reductions this value is found to be sin^ x sin 2x


sin^ X sin 2x, that is, zero. T h e solutions are linearly dependent.
By differentiation of formula 10.12 we see that
W'iuu 2 ; x) = uiixWix)

u,"{x)u2{x).

W h e n Uz^ix) a n d ui^x) are replaced by the values assigned to them by


differential equation 10.2, this is found to be
pi

W'iui, 2 ; x) = ui{x)

po

that is, W = -{pi/po)W.


(10.13)

p2

po

pl

po

p2

po

2(x),

Thus

^/'i(x)

W{ui, U2\ x) = 1.2 exp

xo being any suitable point of the interval i n question, a n d


being a
constant. This result shows, among other things, that M^(ui, 2 ; x) is
either identically zero (if 1,2 = 0), or else is not zero at any ordinary
point (if 1.2 ^ 0).

10.7. The general integral


W h e n formulas for the integrals of a differential equation are obtainable, as they were i n Chapter 8, the question whether the combination
(10.14)

Cmix) +

C22(jf),

i n which Cl and C2 are arbitrary, is, or is not, expressible with the use of
only one arbitrary constant can be answered by inspection. T h a t is not
so when no formulas are available.
Suppose ui{x) a n d U2{x) are linearly dependent integrals, a n d that ai
and fl2 3re constants (not both zero) for w h i c h identity 10.11 maintains.
T h e n let u^{x) be the integral
3(x) = azUlix)

aiuzix).

W e can easily verify the fact that expression 10.14 is now simply
with
_ <:ia2 f2^1
~~
2 I
2 '
a i + 02

c^usix),

The Change of Dependent Variable

189

It is thus expressible with the use of only one constant, namely, Cz,
and is thus is not the general integral when ui(x) and U2{x) are linearly
dependent.
O n the other hand, when
a n d U2{x) are linearly independent,
every integral of the differential equation is obtainable from (10.14) by
giving appropriate values to the constants. T h e relation

i n which Uz{x) is any integred whatever, is fulfilled because all its terms
cancel. However, it is recognizably expressible as
W^(i,

2 ; x)uz{x) +

W{u2,

Us; x)ui{x)

W^("3,

i ; x)u2{x) = 0;

hence, by formula 10.13, it is


{*1.2"3(^) +

iC2,Ztil{x)

4- k3,lU2{x)]

CXp

= 0.

W e may divide this by Ai,2 and drop the exponential, since neither of these
factors is zero. It follows that
Uz{x)

7 l ( ^ ) - 7 "2(JC),
1,2
Al.2

that is, uz{x) is included i n (10.14).

10.8. The change of dependent variable


It is at times desirable for various purposes to transform a differential
equation 10.1 by a change of variable of the form
(10.15)

y = ip{x)w.

T h e transformed equation is then


(10.16)

poW

+ {2po<p' + pi<p]w' + {po<p" + pi'p' +

p2<p]w =

fix).

T o preclude the introduction of singular points, it is clearly requisite that


the function <p{x) be non-vanishing over the interval concerned. T h r o u g h
some choice of <p{x) a new equation which can be integrated may be
obtainable. N o rule for such a fortunate choice c a n be given. T h e
differential equation itself may suggest one.
It can be seen from equation 10.16 that if (p{x) is chosen so that 2pQ<p' +
pi(p 0, the transformed equation lacks the term i n w'. A formal simplification of a differential equation 10.1, may, therefore, be achieved by
the change of variable 10.15, with
<p(x) = exp

Piix)

2 J Po{x)

190

Linear Equations, Variable Coefficients

Example 8.

T o transform the differential equation


xu" - f 6u' +

u=

4x-h

0,

into one w h i c h lacks the first derivative term.


T h e change of variable u *p(x)w transforms the equation into
(up

T h e choice of <p{x) so that Ixtp'


equation to

6^? =

0,

w" + 4; =

ij =

that is, ^ = l / x ' , reduces the

0.

Since the general integral of this equation i n w is ci sin 2x +


the general integrad of the given equation is

u=

0.

cos 2x,

c\ sin 2x + C2 cos 2x\.

PROBLEMS
Transform each of the following differential equations into one which lacks the term
in the first derivative.
39. jjc" + 1 }u" + Axu' 4- {9x2 + 11 }u = 0.
40. x V +

{2x2 _ 2x\u' +

41. / ' + 4 /

4 +

42. {x" + 31u" -

{x" -

-2*

4xu' + 6 u = 0.

43. y" -|- 2y' sec x esc x

2y tan" x =

44. y" 2y tan x +;?{sin x 45. xY

46. xY

4xb' +
2x
*

log*

2x + i ) u = 0.

I4x* -

y +

1)

cot

X.

=1.

2x2 + 51^ = X

1
logx

y = x^,

10.9. Solution of an equation when an integral of the reduced


equation is known
There are at least two methods by which a differential equation 10.1
can be completely integrated when a particular integral of the reduced
equation is known.
M E T H O D 1.
If i(x) is the known integral, change of variable 10.15
w i t h <p(x) = ui{x) can be made on any interval upon which wiC*) 9^ 0.

Solution by a Known Integral

191

T h i s change transforms the differential equation into form 10.16.


But i n this form the term i n w is lacking because of the choice of <p (x).
T h e transformed differential equation is thus a linear equation of the
first order for w'. F r o m this w' can be found by a quadrature.
Another quadrature gives w, and y is then obtainable as ui{x)w.
Example

9.

T o find the general integral of the differential equation


1 -Ix

xy It

1 -

, , 1 - 3x + x^
y ^
.
y = {1 _ ^ j ^ .

1 -

T h e reduced equation is found by trial to have the solution u\


T h e change of variable y = e^w transforms the given equation into
X

w"

^1
X

= il

-x\'

which has the first integral


w' =

A quadrature gives
hence

{x-

\ \e-'

w = xe"^
y =

-\-ci\x -

21

. 2x

x^\e

+ ^cix^e""^' + C2'y

X + icix^e

' + Cze'^.

2. I n terms of differential operators, equation 10.1 c a n be


expressed i n the form
METHOD

(10.17)

D-\-

ut'ix)\
poix)

ui(x)J
uiix)

u^ix)
Ulix)

y =

fix)
PQ{X)

T o prove this assertion we need simply carry out the indicated differentiations. A s i n Section 8.5, therefore, the equation can be integrated by
first finding Yix) from the differential equation
(^)
Uiix) J]

\P(\

poix)

and then finding y from the equation


D
Example 10.

ui'ix)
Ulix)

y =

Yix)

T o find the general integral of the differential equation


xy''

--y

^
X

x-\-x^

Linear Equations, Variable Coefficients

192

By trial it may be found that i = x^.


tion is therefore
D -

I n form 10.17, the given e q u a -

y =

\+x

F r o m the equation
D +

-\Y=\+x\

we find that Y = \x -\- ^x"^ + c\x


D -

T h e equation
y =

Y{x)

then gives
1

\ 4

_ i L + ,^,2

4x

PROBLEMS
Find the general integral of each of the following differential equations by detecting
a solution of the reduced equation and making a change of variable.
47. u"
48.

-6
x"

+2x

{2x + 1 ) / '

x\4x^ -

u'

2x + 6

x"

2/ 2)

+2x

0.

i2x + 3]y 4x= -

50.

|i -

51.

{2x + 3|u" -

4x2 + 4x -

52.

|x' -

|2x^ + x^ -

16x*^

logxi/' + i y

x}u" -

1 }u' -

(Sx" + 16x + lOju - 0,

1 \u' + 4x'u = 0.

53. u" - 2u' cot X + u|l + 2 cot" x\ = 0.


54. xlx* -

\ \u" + [4x* - 6x' - 2\u' +

|2x' -

6x]u -

0.

ASSIGNMENTS
1. Show that if ui(x) is a solution of differential equation 10.2, another solution is

Variation of Parameters

193

2. Show that the integral ui(x) and the integral u^ix) aa given by (10.18) arc linearly
independent.

PROBLEMS
By Method 2 above, find the general integral of each of the following differential
equations.
I** + 2\u" - e'u' -

55.

56. u " + 3x"a'

2 - 0.
2

0.

x^

57. xy" + {2JC + 2 ) / + {x + 2};. - 1.


58. xit" 59. xv" -

3u' + {3 - x\u - 0.
{x -

60. / ' + 1 2 + i
61.

|x' -

62.

2x2)u" -

+ 2x)'' -

- > - 3x2.
2

y -

y -

{x + 3if-'.

{x' + 2x2 - g^j^/ _,_ j3^2 _

= 0.

2;r = 2x2 + 2.

10.10. The method of variation of parameters


W h e n two linearly independent integrals ui{x) and 1/2W of the reduced
equation are known, the integral of the complete equation may be found
as follows: Set
(10.19)

y = gl{x)ui{x)

g2{x)u2{x),

with undetermined functions ^i(;>c) and gzix).

Then

and
/ '

= glUl"+g2U2"

+52'w2i'

T h e substitution of these values into differential equation 10.1 reduo


the differential equation to
-^polglUi

PolglW

T h e last equation is fulfilled


glUi

4- ^ 2 ' 2

that is, if ^i'(^) =

+g2U2]

+ ^ 2 ' W 2 ) ' -hpllglUi


gi{x)

0,

=/(*)

and gzix) are such that


giW

-"2(^)/(;^)

po(x)W{uu 2 ; x)

+ g2U2

g2'(x) =

fix)
Poix)'
ui{x)f{x)

poix)W(ui,

U2; x)

194

Linear Equations, Variable Coefficients

W i t h the choice of any point XQ on the interval tinder consideration, therefore, we may take

^iW
(10.20)

--/

U2is)
po{s)W{uu 2 ; s)

J Xt

Therewith formula 10.19 yields an integral.


Example

11. T o find a n integral of the differential equation

{x^ - 3x +

\]y"

X -

2}y'

+ 2* -

= ^(x^ - 3x + I j ^ .

3]y

T h e reduced equation is found by trial to have the integral ui = e'.


Therefrom the second integral U2 = x x'^ c a n be found by formula
10.18. T h e n by (10.19) and (10.20),
~e'

that is,

2
1 ..3 _ 2x'-

y =

s']e-^ ds-t

[x 4JC -

x'']

4+

s ds.

4<f^

PROBLEMS
Find an integral of each of the following differential equations by using formula
10.19.
63. xy" -

{2x + 1 1 / + fx + 11> = x^e"^.

64. xY

-2xy'

-\-2y=

65. xY

- xy' +y

\x^ -

\X

Zx\y" 1
X

4x + 2 l ^ H

^ xXo^x.

66. xy" -^xy' -y=^


67-

\Ax'^ -

6|)r = x\x -

6}/ +

{3JC -

/ +

1
1 - -

y "ixe ^,

3}

10.11. The change of independent variable


If 6(x) is any differentiable function whose derivative is not zero anywhere on the interval under consideration, we may change the variable
from X to s by setting
(10.21)
B y virtue of this,

s = e{x).
y' = $'{x) dy/ds,
y"

and

= e'Hx) dYds^

e"(x) dy/ds.

The Change of Independent Variable

195

Equation 10.1 is thus, i n partially converted form.


(10.22)

/ - o W e ' ^ W --f^ + ( / , o W 9 " W + / - i W f f ' W 1

dy

+ p^{x)y = f{x)

T o complete the transformation, the value of * i n terms of s must be found


from relation 10.21, a n d must then be used to express equation 10.22
wholly i n terms of ^ and s.
Example 12.

T o transform the differential equation

1
-^U

by the change of variable j = \/x.


T h e partially transformed equation 10.22 is i n this case
1

\\d^y\dy
ds '^x'ds

X*^

x''

x^

Since X = \/Si the completely transformed equation is


n _ . ! j ^ + . J _ ,
ds

^ _ 2 _ 2 .

ds

It may be found by the trial of j * " that this equation has the particular
integral \/s. Since its reduced equation has the integrals U i = e' a n d
2 = s, it follows that its general integral is
= (V-f) + c\e' + c^sT h e given differential equation thus has the general integral
y = x-\- cie'^'

PROBLEMS
By making the indicated change of variable, transform each of the following differendal equations, and thus find its general integral.
69.

xy" -

ly' + x^y = 2x\

70. y" - y' + Ae-^y =


71. Ixy" +
12.

1 -

*V'+

73. u" +

s = e^.

log

s = x\
3

+ > = 4,

t U

Vx.

J=

llogxl
1

0,

- + - ; i " ' " r - i r s " " ^ .


X

log x)

\x log x\

f = log jr.
/

f = viog*.

196

Linear Equations, Variable Coefficients

74. x^y'' +

75.

|2** -

|1 +

76. xy" +

lx^\y' +

+ 2xy' +
i2x' -

1 }>' -

* cos - -

1,

-I

=
.
1 -1- x"*
1 + x^
24xV -

t a n - i x.

4x' sin x^,

j -

x^

10.12. Simultaneous linear equations


T h e method of Section 7.5 can be used to integrate the simultaneous
differential equations
(10.23)
It is not even necessary to begin by putting these equations into forms that
are explicit for^' and z'. Equations 10.23, together w i t h their derivatives,
constitute a system of four equations from which z", z\ and z can (theoretically) be eliminated. T h e liminant is a differential equation for y.
W h e n the integral of this differential equation has been found, the
liminant of z' between the two equations 10.23 is a n equation for z.
T h e roles oy and z can, of course, be interchanged. Thus the liminant
of y", y\ and y from the four equations is a differential equation for z,
and the liminant of^' from equations 10.23 then gives
Example 13.

T o integrate the system


ey

1 -

e'\y'

z' +

[A -

z' -

e'\y

z A y

= 0,
^ e"'

sin x.

These equations, together with their derivatives, are four equations


w h i c h have as their liminant of z " , z', and z the equation
y"

Ay = e~'{cos * sin x

T h e general integral of the last equation is


y -^^^Isin X -\- 3 cos x\ + ci sin 2x -1-

cos 2x.

T h e liminant of z' from the given equations is


z y' -\- e'y e~' sin x.

F r o m this liminant, and the evaluation of^ eilready made, it follows that
z = A{2?"' -

3} cos X - - ^ ( 6 ^ " ' + 11 sin * +


[cxe' -

2c2] sin 2x -\- { c 2 ' ^-2ci\ cos

2x.

The Laplace Transformation


Example

197

14. T o integrate the system


2xy

{x -

2x^)2' +

3xz' -\-2xy -

[2x

22 =

4-

+ 2jz =

{3x -

2]e',

2|tf^,

These equations a n d their derivatives have as the liminant of y"j


y'j and y the equation
xz" -

z' ^

[x -

\ \e'.

T h e general integral of the liminant is


2 = tf* +

Cix"^

C2.

T h e second of the given equations then yields


y ^

^-cAx''2x}

+ ^2

PROBLEMS
Integrate the following differential systems,
77.

y'
xy' -

78. {5x -

z' -2Sy z ~ x = 0.
3z' + ISy - xz + Zx = 0.
10;- + 2xz =
y' - 2xz =

1 }>' + 2 x V -

79.

u' - z' Su' - z'

80.

4u' - x^' - 6u - U -3u'+xr' + 5u+

81.

1
2

-X.

8u + 3z = 0,
2 = 0.
1

= 0.
z=0.

2 x 1 / + xh' + 4^ = 1 - x'.
'
2
y xy xz X .
~ z
2

82.

X,

z' +

1 +-

z+

y = 2t\

xy' -\-~z = 2xe'.


X

10.13. The Laplace transformation


For a differential equation 10.2 whose coefficients are linear i n x it is
sometimes possible to obtain an integral i n the form
(10.24)

u{x) =

(^\'^v{t)dt,

where x\ and X2 are constants and v{t) is a function that can be determined.

198

Linear Equations, Variable Coefficients

L e t the differentisil equation be


(10.25)

{aox + boW + {aix + bi]u' + [a^x + 6 2 } = 0,

It is converted by substitution 10.24 into the form


(10.26)

fj'xe'JVit)

dt + j'\"R{t)V{t)

dt = 0,

where
(10.27)

V{t) = [a^-^

axt \-at]v{t),

and
(10.28)

^(0 =
aot" + ait -r 02

A n integration by parts changes (10.26) into


[e'^Vm:

lJ*e"{V\t)

R(t)V(t)]

dt = 0.

This is fulfilled i f F ' - R{t)V = 0, that is, i f


(10.29)

logK(0 ^

SR{t)dt,

and if xi and X2 are roots of the equation V{t) = 0.


Example 15.

T o find a n integral of the differential equation


xu" -

{x-\\u'

[2x + 11 = 0.

F o r m u l a 10.28 gives i n this instance

t ^ - t - 2

and thus, by equation 10.29,


V{t) ^ \t- 2)'^(/ + 1 1 ^
T h e equation V{t) = 0 has the roots ATI = 1 and X2 = 2.
equation 10.27,
v{t) =

( 2 ~ / - 2

j/-2l'^[/ +l l ^

F o r m u l a 10.24 therefore gives the integral

Also, from

Answers to Problems
PROBLEMS
Find an integral for each of the following differential equations.
83. xu" ~- \x 85. xu" -

4|u' -

\Ax -

3}' +

87. x u " + 3u' -

84. xu" -

3 = 0.
l\u = 0.

f3x -

xa = 0.

89, x a " + I2x + 2 l a ' + a => 0.

i2x -

3)a' - 4u

86. xu" + 5u' -

(x + 3}a =

88. x a " -

l}a' - a

{2x -

90. 2 x a " + j2x + 3 } a ' + a

ANSWERS TO ODD-NUMBERED

PROBLEMS

1. a = f i ( x + 2 i + f 2 ( x + 2l
sin log {3 x| + ci cos log 3 x}

3. a =

+ f log X + i + cix + <:2xl

5. > = i U o g
7. y = {x -

1 i + \cx log (x -

9. > - i | x - 1} +

11. : y -

^
1

20

13

13- li ^1

10

1) + cat

{ l o g ( x - 1)

. ,
2
Sin log X *

- 1

+1).

cos loK
20
^

15. a ^ 1 CSC X +

cot x.

17. Inexact.
19. a

21. a ="

23. a

1 -

2x

1 +2x

^1

cos X log

2x -

(1 - 2 x )

1 + sin X
1 sin X

+x
\

- e'^' + cix /

31. a cxp X

+ 2 cos X .

log (x 1)

X log X 1
1

+ f2

2x - 1
,

27. a = ci

29.

i log

C2

f2
1
dx C2X33. a -

exp \ -2e']

200

Linear Equations, Variable Coefficients

35. u ^ - ^ i " " ' - ' .

37. u = cxp {3e^\.

39. w" -\-9w'0.

41. w" -\-\w

-2w

43.

3*"' + 1.

= e'.

45. ; " + -5

47. u - ^ + C 2 ^ .

49. u -

51. y "

f 1^^*"* + c^e

53. u cix sin x + f 1 sin x.

55. u -

o i l

57. > -

+ fa*.

- + 1

X
59. y '

- {x^ + 2x + 2} -\-cif' j

^2"'.

^ rfx + c j ^ .

61. u - fix** + C2'.

63. > -

Ix -

65. > - i x ( l o g x|'.

67. y -

- 4 x ^ + x + 1 - x M o g x.

X*
69. ^ 2 + f 1 sin - J +

11^"*.

x
cos

71. J, - 4 + eV^[ci sin \ / ^ + a cos


73. u - f 1 log X + .

75. y - tan * x H

V l

logx
77. > -

- ^ x +

ci-r^' +

ca*-'',

79. u
^{ci sin X +
C2 COS x|,
z ^'{(Sci j) sin X + (ci + 3^2) cos x|.
81.:)'^ .
83. u 87. u

i + cj** +

log X,

* -|-fi(2 - x ^ l +ct{l

+2Iogx

-xMogx).

85. u / / * ^ ( ' - 11

CHAPTER

11
Solutions in Power Series

11.1. The reduced equation at an ordinary point


T h e method of power series of Section 7.6 is applicable i n particular to
a linear differential equation 10.2. If the equation has an ordinary point
at xo, and has coefficients that are representable by power series i n

XQ], its integrals are also so representable.


T h e change of variable x XQ = z makes the point x ~ XQ correspond
to 2 = 0, and transforms the differential equation into another linear one.
W h e n XQ 7^ 0, we shall suppose that this change is made. I n proceeding
we shall therefore assume that XQ 0, and accordingly that
Poix) = pQ,G -\-pO,lX

(11.1)

-\-p0.2X^

Pi{x)

=pl,Q-\-pl,lX-\-pi,2X^-\-

p2{x)

= /'2.0 + / ' 2 , 1 * + / ' 2 , 2 * ^ +

^0.0 ^

0,

T h e form of an integral is then


(11.2)

u{x) = Co + cix + czx"^ +

I n the general integral, CQ and ci


UQ' i n the integral for which u(0)
T h e substitution of series 11.2
ferential equation 10.2 converts
series. By equating coefficients
system of relations

are arbitrary, whereas


= o 3.nd a
= UQ, U'{0) = UQ'.
with undetermined coefficients into dif(10.2) into an equation between power
of like powers of x, we then find the
2/^0,0^2 -\- pi,oCi + p2,QC0

2 3yjo.oC3 +

{2po.i + 2pi,o]c2 +

{pi,i

-\-p2,o]ci

i'p2,ico

= 0,
= 0,

These successively determine cj, C3, 4, * , i n terms of


and ci.
If it is deemed desirable, equation 10.2 can first be transformed into an
equation
(11.3)
w" - q{x)w = 0,
201

202

Solutions in Power Series

by the change of variable u = <p{x)w, w i t h


<p(x) =

exp

dx

I n this form, then, q{x) is representable by a power serit


(11.4)

= ?o +

q(x)

qix +

q2X^ +

which is determinable from the series for po{x) and pi{x).


F o r the simplified equation 11.3, the relations determining the coefficients of the integral
(11.5)

w(x) = Co + CiX - h C2X^ +

are, explicitly,
1
f2 = 2
C3 =

1
2-3

(11.6)
1

"

(n - l)n

{?n-2^0 +

qn3Ci

B y determining an integral 11.5, and multiplying it by the power series for


<p(x)y integral 11.2 is obtainable.
I n practice it is generally preferable to
obtain the integral directly from the differential equation i n its original
form.
T o find (up to terms of the fourth degree) the integral of
the differential equation
Example 1.

u"

cos X + w' sin ;i: + {cos x 4* sin x]u 0,

for which u(ir/4) = 2 a n d '(7r/4) = 0.


T h e change of variable * (ir/4) = z transforms the differential equation into

cos z

d^u

du

dz^

dz

sin z} ^ + (cos z + sin z) + 2u cos z = 0.

F o r this.
PQ(Z)

= 1

- i z ^ + - ,

Pl{z) = l - h

Z -

p2{z) = 2 -

z2 -}-

iz2 +

. ,

A Proof of Convergence

203

T h e substitution u = CQ -\- ciz -\- czz^ + * * , and the equating of


coefficients of like powers of z leads to the relations
2c2 -\-Ci-\6^3 +
\2c4

3c2

2CQ = 0 ,
2>ci = 0 ,

i c i

fn =

0,

F r o m these it is found that ^2 = {^o + i ^ ^ i l ; ^3


= {^CQ +
^ V i h . T h e particular integral required is determined by the
values Co = 2 and ci 0. W h e n z is replaced by {>: 7r/4j, the particular integral is found to be

PROBLEMS
Find (up to terms of the fifth degree) the integrals of the following differential equations that fulfill the respective conditions.
1. " -

{x -

2. u" +

11 -

'*

3.

ISx^lu' -

\3x + 4x^\u' +

u" -

xu' +

4. a " + a ' sin x 5. a "


6.

Vl

il -

7. a " +
8.

{1

a(0)

+2^2

I* +

u(0) =

\2 -

u' +

(9 -

6x]u = 0,
{1 -

=. CO, a'(0)

u(2) = 1, u'{2) = 0.

6x + x^Ju = 0,

= 1, a'(0)

a(0)

i^^^x

{1 + log (1 -

u(-l)

- 1 , u'{-\) = 1

= 2.

= 0,

Y^^la

-3.

a(0)

x)W

= 0, a'(0)

= 2.

u log (1 + x ) =

0,

= 2.

{4 + X + 6x2}' _
x^ja" +

(0) = 2, u'(0)

{13 +24JC + 12*2)a = 0,

+ % V ) a " -

- 3 , a'(0)

X -

3x^]u = 0,

{1 -

|1 +

|3 _
6x +

4^ _i_ 4^2j ^
5x^)u' +

{1 -

^(QJ ^ ^^^^ ^/^o)


5x +

x^lu =

o.

0,

= ci.

11.2. A proof of convergence


T o assure that a series 11.2 does in fact give an integral, it is necessary to
show that it converges upon some interval
(11.7)

-h<x

<h.

Suppose this interval is one which contains no singular point, and upon
which a l l series 11.1 are convergent. It is then an interval upon which
series 11.4 for the transformed equation 11.3 and also the power series for

204

Solutions in Power Series

the transformation function tp{x) are convergent. W e need therefore


merely show that on this interval series 11.5 is convergent.
Let a be any positive value less than h. A t x = a series 11.4 converges
absolutely; hence there is a constant M such that \qja^\ < M^, f o r ; = 0,
1, 2, 3, . T h e Formulas 11.6 accordingly show that
^0

(11.8)

(n l ) n l a

n-2

n ^ 2.

Consider now the constants 4 that arc defined by the formulas


^0

Co

ll

n - 1

M-l-

n = 1, 2, 3,

An

It is easily seen that


1 2.42

23^3

1 -2
a

(n l ; n . 4 n

AQM

Ao^

2 3
.4.
-

3 4.44

A1

.4

^n-l =

i4_2

Af+

O n multiplying thej'th of these equations by a^"*"^


equations, we find that because of cancellations

(n -

\)nAn

M +

AQM

n-2

n - 2
a

and then adding the

a
,n-3

An-2

-\-

n -

Thus
(11.9)

An>

(n

l ) n [a n-2 ~ _n-3

+
~

+ ^n-zl'

n ^ 2.

The Complete Equation

205

Now Ao ^ Co', and Ai ^ ci|, and therefore a comparison of relations


11.9 and 11.8 shows, for successive values of n, that i4n ^ c . S e r i e s l l . 5
thus converges absolutely whenever the series AQ
Aix -\- A^x^ -\- ' ' ,
does so, which is found, by the use of the test ratio, to be when
a<x<a.
Since any point i n the interior of inter\^al 11.7 is within some interval
a<x a, the convergence of series 11.5 within the interval 11.7 has
been proved.

11.3. The complete equation at an ordinary point


W h e n the method of series is applicable to the reduced equation, it can
also be applied to the complete equation 10.1, provided f{x) is of the form
(11.10)

fix) = ; ^ " ^ { / O + / I A : + / 2 X 2 +

with any constant m that is not a negative integer.


case, a series:
(11.11)

= azx'^+^ +

^3^"+^ +

T h e integral is, i n this

If m is a negative integer the method may fail. A suitable modification


of it is then applicable, as is shown i n Section 11.7.
Example 2.

T o find an integral of the differential equation


y'' + 3/-\-{\-x'\y

T h e division of [jf
fix)

= "

1 +

j by 11 + A;) gives for this example

= x[\ -

2x -h 2x^ -

2x^ +

T h e substitution 11.11 with m = 1 leads to the relations

12^3

6^2

= 1,

9^2

=-2,

20(24 4- 12(23 - f ^ 2 = 2,

30a5 + 15a4 + (23 = 2 ,

F r o m these relations the integral is found to be


y sx

-^x

-\-

-rsiix - h

PROBLEMS
Find, in powers of x (up to the term in x^), an integral for each of the following differential equations.

206

Solutions in Power Series

9. / ' +

Ux-}/

|2 -

\(>x -

Sx'^]y = 2 -

4x + lOx^iy +

to.

{2 - f x'^\y" +

|2 -

11.

{2 + 4x2};)." +

{10 + 8x + 6^2)/

12. / ' +
13. ^ " +

{8 + 4x + x2}_y =
2 -

{1 + x=b = 1 + 6x -

12 + x l ^ ' + ^

cos

= sin

Ix'^.

{2 +

x^jf^

Sx^ljy = 4 + 18x + Ix"^ + 36x=.

2x -

3x'.

14. y"

+ jr' tan x + ;> sec x =

11.4,

The ordinary point at <. Solutions in powers of

sec x.

1/x

A differential equation 10.1 whose coefficients are representable by


series in powers of 1 / x of the precise forms

(10.12)
X

is said to have an ordinary point at <x>. By the change of variable x 1/z,


such a differential equation is transformed into one which has an ordinary
point at z = 0. T h a t accounts for the special forms of series 10.12
If the series
(10.13)

M = Co + " +
X\

-2 +
X

x^

is substituted into the reduced equation, and the coefficients of the several
powers of 1/x are then equated to zero, a system of relations from w h i c h
(^2,
C4, * , are determinable is obtained. T h e values of co and c\ are
left arbitrar)'. F o r the complete equation the method determines the
coefficients of a particular integral
(10.14)

, = p +

T h e series so obtained for the integrals converge upon any range x > H
upon which all of series 10.12 are convergent, and/>o(x) 9^ 0.

207

Solutions in Power of
Example 3.

T o find the general integral of the differential equation


2]

Ax- -\-2x-\- 10

i n powers of 1 /x.
W h e n the given equation is divided by x"^ its coefficients have the forms
10.12. I n the case of the reduced equation the relations obtained by
substituting series 10.13 are
2c2 H- 2c(i = 0,
6cz + 4ci = 0,

12c4 4- 10<:2 = 0,
T h e complementary function is thus found to be
^ _
x~

2_
3x'~^

T h e substitution of series 10.14 into the complete equation leads to the


relations
2a2 - 4,

2,

603

12^4 + 10(22 = 10,

T h e particular integral is therefore


1
+

^ = - 2

3x

6x

PROBLEMS
F i n d the general integrals of the following differential equations in powers of 1 /x.
15.

16.

fx* -

x^ju" + [2x' + 4x2 _(_ 4^]/ _|_ 2u = 0.


1
+ -

17.

1 + 3+73

1
x^

18. x^u" + 2 x2../


V + u log

u" +

=0.

= 0,

208

Solutions in Power Series

Find a particular integral for each of the following differential equations.

19.

{x* -

20. xY
21. y

x^\y" +

2x* + x ^ i y +

2x^^

22.

1 -

\2x' + 2 x 1 / -

|2x* + 3x -

2x'^

1
4

x^

101> -

{x -\-\]\

2x

X*

2
4
- + +
X

2;- = 2 -

x^

2_

X*

x'

11.5. The reduced equation at a regular singular point


I n the case of a differentia! equation 10.2 having a singular point at
jc = 0, it is found convenient to multiply the equation by such a power of
X as gives it the form
(11.15)

x%{x)u"

w i t h po{0) 7^ 0.
thus,

the point

xqiixW

q2{x)u =

0,

If the coefficients are then representable by power series,

po{x)

= po,o

qi{x)

91.0 +

po,ix H- po^zx"^ +
qi.ix

qi,2X^

po,o

7^

0,

0 is said to be a regular singular point. Otherwise it is a n


irregular singular point. A discussion of irregular singular points is beyond
the scope of this book. However many important differential equations
have regular singular points.
T h e substitution of the form
(11.16)

J: =

U^x'^ll

CiX

C2X^

with undetermined constants k, a, ^2 " " ' , into differential equation


11.15 converts it into an equation of the form AQX'^ -\- A ix*^^ ^ + A 2X*''"^ +
* = 0 , which is fulfilled if Aj = 0, f o r ; = 0, 1, 2, . T h e relation .4o = 0 is explicitly
(11.17)

po,o\f^ 1 !A + ?i,o +

92,0

= 0.

This is called the indicial equation at the singular point x = 0. Its roots ki and
2 are called the indices, or the exponents, at this point.
W h e n k is assigned
one of the index values, the relations . 4 , = 0, f o r ; = 1, 2, 3, * , form
a system of equations associated w i t h that index, from w h i c h the values

The Regular Singular Point


^1)
^3) ' , may or may not be determinable.
several cases.

209
We must consider

1. T h e indices are different but do not differ by an integer.


The relations Aj = 0 then serve, in the instance of each index, to determine
the values of ^ i , 2, f.i, . T w o integrals of form 11.16 are thus
obtained.
CASE

2. [ki 2) = m, a positive integer. T h e relations Aj = 0


then determine the values c i , C2, C3, , in the instance of the larger
index A;i. Thus one integral is obtained. In the instance of the smaller
index 2, however, the relation .4^ = 0 is found not to involve Cm- Because
of this, two alternatives present themselves.
CASE

SUBCASE 2a.

T h e relation

Ai=

for the index

^2

is fulfilled by the

previously determined values c i , fr2, " ' ' ,<^m-i- W e may then assign to
Cm any value (in particular zero) and proceed to determine c^+i, Cm+2y
, from the respective subsequent relations, A second integral of
form 11.16 is thus obtained.
T h e relation A,n 0 for the index 2 is not fulfilled by the
previously determined values ci, C2, " ,^mi- There is then no integral
of form 11.16 associated with 2.
SUBCASE 2b.

3. 1 2 = 0. T h e relations Aj = 0, with k = A i , determine


the values cu ^2, ^3 ' ' ' But, since 2 is not different from A i , the
method affords no procedure for obtaining a second integral.
A method for obtaining a second integral i n the events of Subcase 2b and
Case 3 is given i n Section 11.6.
CASE

Example 4,

T o find a pair of integrals of the differential equation


1 8 x V ' H-x|15 + 3x]u' -

|1 + 10:^1 = 0,

about the point x 0.


T h e indicial equation is in this instance ISA^ 3A 1 = 0 .
The
exponents are therefore \ and
These do not differ by an integer.
T h e instance is therefore one of Case 1. W i t h k ~ \, the relations
Aj = 0 are
27ci - 9
90c2 -

= 0,

6ci = 0,
I

T h e y yield the integral m = x ^ ( l +


the relations are

+ T V * ^ +*}.

With k =

210

Solutions in Power Series


9^1 54c2 -

and yield the integral U2 = x~^{l


Example 5.

= 0,
=

^ 1

0,

-\-ix -\-

T o find a pair of integrals of the differential equation

2(1 -

x]u" +

1 4- x}u' -\- Ix -

3 -

(x -

l)Vlu = 0

about the singular point x = \.


T h e change of variable x \ = z transfers the singular point to the
origin, and gives the given differential equation the form
2^-

dz^

du

z[2^-z]-+

dz

[2z -

z"- + ez^ +

\u = 0

T h e indicial equation for this equation is 2k' 4 A = 0. T h e indices are


therefore 2 and 0, and we accordingly have a n instance of Case 2. T h e
relations associated w i t h k ~ 2 are
(>c\ = 0,
16^2 f 1 1

30^3 -

2^2 -

^1

=0,

+ = 0,

18^
^^7
240

T h e y yield the integral ui =

{1 +
16 H

The

relations associated w i t h k ^ are


~2ci

+ 2 = 0,
-

1 = 0,

c\ \- e =

0,

T h e first of these relations yields c\ = \. T h e second one does not involve


c%. However, with ci = 1, the relation is fulfilled. This example is
therefore an instance of Subcase 2a. O n assigning to
the value zero, we
find from the third relation that 3 = (1 rf)/6, etc. Thus we obtain
3

1 +^

Logarithmic Integrals

211

I n terms of the original variable the integrals are


i W = Ix - i p i l + 1 (x -

U2(X)

= |1 + (x -

1) +

1)^ +

(x -

(x -

1)' + 4 4

PROBLEMS
For each of the following differential equations find two integrals about the indicated
singular point.
23. * V '

-x\i

-x-\-

4x^\u' -\- {x-

24. x V - xli + 4x + 4x^W +


25. lOxu" + 12' -

{2 + x} j5 -

x^\u = 0,
+ * +

xa = 0.
=0,

3xlu - 0,

xo = 0.

XQ = 0.

26. x2{l - x j u " + x{2 + xju' + [| + i x - -Jx^Iu = 0,


27. 16x{x + l i a " 28. 2 (x -

8{x + 2|u' -

1 \tt" - o' +

{x'^ -

= 0,

29. x ^ l + 4 x + l l x ^ l u " + x |1 -

4} =0,

x -

XQ - 0.
xo = - 1 .

x = 1.
11x^1' -

J T + 3X + * 2 }

xo = 0.
30. |x + 2IM2O + 27x + 8x2)" ^
50 + 46x + 12x='j = 0, xo =

^ 2) 46 + 76x + 24x^1'
-2.

11.6. Logarithmic integrals of the reduced equation


W h e n only one integral of form 11.16 is obtainable, due to the fact that
m, w h i c h is 1 21J is a positive integer or zero, the change of variable
a =;r -

(11.18)

bmUl{x){o^X

transforms differential equation 11.15 into


(11.19)

xVo(^)>" +

- i - g2{x)y = /(x),

with
(11.20)

fix) = b{2xpo{x)ui'{x)

[qiix)

-po{x)]ur(x)].

T h e form
fix) = x * ' ! / o + / i * + M ' +

-1

is obtained by replacing ui(x) i n equation 11.20 by its series 11.16. T h e


substitution
(11.21)

y -

x'^^ibo + bix-\-bzx''-\-

-1

212

Solutions in Power Series

leads to a system of equations w h i c h successively determine 1, 2, 3


* , in terms of 0Example 6. T o find a pair of integrals of the differential equation
x^\

x^W

x{l

-\-2x-{- Sx^jtt' +

(x + Sx^lu = 0

about the point XQ = 0.


T h e indices are in this case 1 = 2 and kz = 0.
lW

+ x +

ix2

F o r ki the integral

-I,

may be found, but the instance is then discovered to be one of Subcase 2b.
T h e c h a n g e o f v a r i a b l e l l . l S l e a d s t o e q u a t i o n 11.19with/(jr) = bz{2x^
2jt' + ' ) and substitution 11.21 thus leads to the relations
~bi

+ 5^0

2*2 +

3*3 -

bo =

261 =

0,
2A21

2*2,

O n assigning to bo the value 1, and then determining 1, 2, 3, * ' ' ,


we find that > = 1 + x + 2x^ +
- f . B y (11.18), the second
integral is thus
uzix) =
Example 7.

(1 +

+ 2x=^ +

- . ) -

2ui{x) l o g x .

T o find a pair of integrals of the differential equation


x V

+ x[2 + x]u' +

( i + Ix -

2x^1 = 0

about the point XQ = 0.


T h e indices are i n this case equal, and have the value H integrcil found is
1 = x-^{\

- x

+ x2-|x3+

The

.}.

B y virtue of this evaluation of u i , formula 11.20 yields


fix)

box'H-i

+ 3 x - i x 2 +

-1,

and the substitution of form 11.21 into equation 11.19 leads to the relations
1 +

0 =

2bi -

2bo =

9bz -f- 3*2 -

2bi =

462 +

bo,
3bo,
-ibo,

The Complete Equation

213

O n assigning to 0 the value 1, the second integral is found to be


2 = * - ^ l l - 2x + Ix^ -

-1 - iW l o g x .

PROBLEMS
Find two integrals for each of the following differential equations about the point
*o = 0.
31. x V - x{2 32. x^l\ -

5x\u' + \x -

x^\u" + *(3 ~2x

33. x^{2 + 3x^\u" + x[2 34. u" sin^

u' &in

6x^\u = 0.

~ 3x^W 4x -

xu

\3x + 3x^ + 2x^]u = 0.

3x'^}u' -

12 + 2x -

5x^ -

ix^]u = 0.

=0.

35. x^u" - xe'u' + = 0.


36. 3 x V ' + 6xu' + ( i + Sx^la = 0.

ny.

The complete equation

I n the case of the complete equation 11.19, i n w h i c h


fix)

X^\fo + / l X + / 2 X 2 +

- - -1,

fo9^

0,

the form of a particular integral depends upon h.


1. Neither index exceeds {h
is i n this case an integral of the form
CASE

(11.22)

y = x^{ao +

a^x

1 j by a positive integer.

There

a^x'' +),

the coefficients of which c a n be determined.


2. Just one index exceeds (A 1) by a positive integer. L e t us
say 1 = A +
where m is zero or a positive integer, a n d 2 h] is not
zero or an integer. T h e change of variable
CASE

(11.23)

>-=>*-

amUi(x)

log

X,

then transforms the differential equation into one for


which has a n
integral of form 11.22 whose coefficients can be determined.
3. E a c h index exceeds (A 1 ( by a positive integer. L e t us
say 1 = A - f m i , and A2 = A - f ^2. There are then the following
alternatives:
CASE

3a. T h e reduced equation has two integrals of form 11.16.


T h e change of variable,
SUBCASE

y = yif -

amiUiix)

log x -

a^^zix)

log x.

214

Solutions in Power Series

transforms the equation into one for


11.22.

w h i c h has a n integral of form

T h e reduced equation has an integral ui{x) of form 11.16,


and another of the form M 2 W = x^^v{x) bui{x) log x, w i t h v{x) a power
series. T h e change of variable,
SUBCASE 3b.

y = y*

am.MlW log X

OmiUzix)

log X

iflms^UiW {log x ) ^

then yields for^-* a differential equation w i t h an integral of form 11.22.


Example 8.

T o find a particular integral of the differentigd equation


2x^>" + xy' + xy = sin X + 2x^ cos x

about the point XQ = 0,


T h e indices are i n this instance ^-2, and 0, a n d
fix) = x i l + 2x - i x ^ - x^ - h -1.

This is therefore an instance of Case 1.

Example 9.

T h e integral is found to be

T o find a particular integral of the differential equation


xy

+ 3 x V - 2> = x'^{2 -

7 x + 4x^1

about the point XQ = 0.


T h e indices are k\ = 2 and
= 1 . Since A = 2, this example is a n
instance of Case 2. F o r the reduced equation we may find the integrzil
l(x) = X 2 { 1 - | ; c + |Jx^ +

-1.

T h e change of variable 11.23, w i t h m = 0, transforms the given equation


into
x ^ " +

3x2,*'

= 2 +

3(2olx2 -

+ fflol^' +
!4 + f ^ a o ) * * + ,

which has the particular integral


= x'^{% -\- rkjix'^ +).
integral of the given differential equation is thus
:v

- ; c 2 { - | +

Example 10.
^xY

T^i7x'+

) + V { 1 - | ; c + K x 2 +

The

- .}logx.

T o find a particular integral of the differential equation


-

x[\

+ x\y' -

about the point xo = 0.

I I + |xl> = x^^{4 + 2x + x^)

215

The Regular Singular Point at oo

T h e indices are 1 = 4 and kz = h


T h i s example is therefore a n
instance of Case 2, w i t h m = 1. T h e reduced equation has the integral

iW

= x^l\

^2

f
T h e change of variable 11.23 transforms the given differential equation
into
3xW

- '{1 +

{I + ixljK* = 4x^ + {2 + SmW'

which is found to have the integral


=
2 + ix^ +
this instance it happens that ai = 0; hence, by (11.23), y
integral of the given differential equation is thus

+
). I n
y*.
The

I,
and does not involve log x.
W e shall not consider any instances of the more intricate Case 3.
PROBLEMS
Find a particular integral of each of the following differential equations about the
point *o = 0.

2
37. x M l + 4x-\y" + xy
38. 5 x V ' + (ixy' -

39. 2 x V '

-ly^lx

4> =

+*}/

3x{l + x | l

+ U

40. Axy" - lOxV -Sxy^


41. x|2

42.

x V y

1-^
4

+ 3{I - x } /

+x}> = x^^{2 - 14x|

x^4{3 - 5x + x^}.
- x>Ml

\x-k-i^x^\.

+ / sin X - 1 + x)> = .r^

43. x^\l -

x]y" +x)r'

+ xy = l + 2x -

3x1

44. x ^ ; - " + 4x^'y' + 2> = ^ ! 1 + - x=


X
3
45. x V + x'-^^ =. X - M 6 + 2x + 3x2 46. 3x'>" -

llxy' -

j^3j

5;F cos X = x-^^|38 + 4x + 6x2|.

11.8. The regular singular point at oo


W h e n the coefficients of differential equation 11.19 are representable
by power series i n l/x, thus:

216

Solutions in Power Series


poix)

q\{x) =

?i,o H

?2W

?2,0

fix)

= pO,(i +

12~

with/Jo.o 5^ 0,

+
f:

*r"

the differential equation has a regular singular point at infinity


reduced equation, the substitution
1

_L '^l _ L ^2 ,

Co H

with Co

For the

0,

leads to an indicial equation for A, and yields for each index a system of
relations i n the coefficients CQ, CU C2, ' ' ' F o r the complete equation
the substitution
1
' = = 7 r ^

leads to a system of relations i n the coefficients aoj oi, 02, ' ' ' I n
general, two integrals of the reduced equation and a particular integrzil of
the complete equation are so determinable. I n the exceptional cases i n
which these integrals are not obtainable, the given differential equation
may be transformed by the change of variable x = \/z. T h e transformed
equation has a singular point at z = 0, and for this transformed equation
the integrals are obtainable by the methods of Sections 11.6 and 11.7.
PROBLEMS
For each of the following differential equations find an integral in powers of \/x.
47. 2x\"

1 +

1
u' -\--u=

48. Ix'^u" + 5xu' -

1
+ -

u = 0,

49. 9x-y" + \Sxy' + 2> = 1 +

-+ 4
X

50. x'y
2..H

1
X

0,

8 +

2
X

1
^
X

X*

Answers to Problems
51. xY

1
-

5 -

+x

f
y

7 -

1
ll
+-2

52. x^

2
-

ANSWERS TO ODD-NUMBERED PROBLEMS


1. u =* 2 -

3x

3. a = - 1 +

X*

- x + 4x* -

U + 1) +

2^5 +

jx + I p -

U + ll'*+

5. u = 2x -j-x"^ + X* + Ox'' -\-


7. = , o l l + | : c 2 - f x ' + ^ * - K x ^ +

- !.

9. y
U.y^x'-W-ih'^hUx'-^-

n.

1 4

> = Ifx Y^x^ YTUX +

15. u = CO 1

17. u

CO

1
1 +-5

3A:^

x^ '

X*

+ C1

1
2
12-13

1
-

1 + 1 + 4 +

2 U y

23.

6x

2x^

3x'

x^

ui
(1 + 2 x -

3*2 +

25. a i i.

27.

fl + T ( * + 1) +

u2-(x-\-

D^Mi -

29. 01 - * ^ { 1
U2

31. ui

x'(l i l

}>

+ 1) + ^ ( *

+ 2 * +4*'' +

X-^{1

+Jt2 + * 4
4x + 9*2 +
32

1.

13

3 I

1 + T r i W log X.

33. ui = *(1 + x + i*2 4-


"2

"Ml

1.

+ x + 3*2 +

4 *

} - 3m(x) log X.

37^ 37^

Solutions in Power Series


1,
+

U - *+

}-

+ -1 - {1 - * + i ' ' +

- + 2 + 3x -

3* +

2x

5x^

Sx^

V**

X*

Ki(jf) log X.

2x'

+
41

CHAPTER

12
Some Differential Equations
Involving Parameters
12.1. The separation of variables in a partial differential
equation
T h e mathematical expression of a physical law of nature generally
requires the use of a partial differential equation. T h e course of any
chain of events which is subject to that law is then governed by an integral
of that equation, that is, by a particular integral that is characterized by
certain conditions which the chain of events fulfills. A n important
method for determining the requisite integral depends upon ordinary differential equations. W e shall indicate some elements of that method.
T h e partial differential equation

a^r

d^T

a^r

governs a variety of physical phenomena.


It is known as Laplace's equo'
tion, i n honor of the French mathematician P . S. Laplace (1749-1827).
A product of a function of x by a function of > by a function of
namely,
u{x)v{y)w{z)y is a n integral of (12.1) if
u" {x)v(y)w{z) + u{x)v" {y)w{z) + u{x)v{y)w" {z) = 0,

T h a t is, if
v"{y)

^ w"{z)

v{y)

w(z)

u"{x)
u(x)

I n this equation the left-hand member does not vary w i t h x. T h e righthand member does not vary w i t h y or z. T h e i r common value is therefore a constant. O n designating this constant by A, we have

u{x)

'

v(y)

w{z)

I n the second of these equations the members are again a constant, say
/. W e see thus that u{x)v(y)w{z) is a n integral of the partial differential
219

220

Differential Equations with Parameters

equation 12.1, if and only if its factors are integrals of the respective o r d i
nary differential equations
u'\x)

huix) =

v"(y)

h{y)

(12.2)
w"{z)

{h-

0,

= 0,

l\w{z) =

0.

I n these differential equations k and / do not depend upon x, y, or


T h e y are therefore constants. But their values are not specified. T h e y
can be assigned various values. They are therefore called parameters.
This method of referring a partial differential equation to a set of ordinary

F i o . 31.

differential equations involving parameters is called the separation of


variables. W e have applied it to differential equation 12.1, namely, to
the Laplace equation i n the rectangular coordinates x, y, z. It can be
applied also i n other coordinate systems, as we shall show.
T h e cylindrical space coordinates (r, tp, z) indicated i n F i g . 31 are
related to x, y, z by the formulas
X

r cos ip^

r sm (p.

z =

z.

These define a change of variables which transforms differential equation


12.1 into
d^T
1 dT
1 a^r
d^T
dr

r dr

r^ d<p'

dz'-

It is now found for this last equation that a product u{r)v{<p)w{z) is an


integral, if its several factors fulfill the respective ordinary differential

Separation of Variables

221

equations
r V ' ( r ) + m ' ( r ) + {hr^ ~ l\u{r) = 0 ,
(12.3)

v"{<p) + IvM
w"iz)

= 0,

hw{z) = 0.

T h e first of these equations is of a type that we consider i n Section 12.8.


T h e other two are elementary.

F i o . 32.

T h e spherical space coordinates (r, tp, 6) indicated i n F i g . 32 are related


to X, Zj by the formulas
X =

T sin

$ cos (fj

y = r sin 6 sin <pj

z r cos 9.

These define a change of variables which transforms differential equation


12.1 into

For equation 12.4 we could seek an integral of the form u{r)v{B)w{*p). W e


can, however, also seek one of the simpler form u{T)v{d). T h a t is found
to require that
T\"{r) + 2r'(r) - hu{r) = 0,
(12.5)
v"{e) + v'{e) cot 6 + hv{e) = o.
The first of these equations is a n Euler equation of type 10.3.
is of the type which we consider i n Section 12.4.

T h e second

222

Differential Equations with Parameters

12.2. A problem in the flow of heat


W h e n the surface temperatures of a material sphere are held at prescribed values w h i c h depend only on the variable fl, say at the values o f a
function f{B), heat flows through the sphere. I n time this flow becomes
steady, and the temperatures at points within the sphere are then g i v e n
by an integral of partial differential equation 12.4. N o w we have seen
that there are particular integrals of this equation w h i c h are of the f o r m
u{r)v{B)y and that the factors of these integrals fulfill the respective o r d i n a r y
differential equations 12.5. Every integral of the first of equations 12.5
is a combination of the functions
and

r - * - ^ -

These functions, however, are single valued and continuous at r = 0 o n l y


if the exponents are non-negative integers. T o obtain an acceptable
integral we must therefore assign to A a value h = k{k + 1 j , where k is
zero or a positive integer. T h e acceptable integral is then w = r*. W e
have thus been led to the products r'^Vk{B)y where, for each integer ky
vit{d) is a n integral of the differential equation
(12.6)

v"{e) + v'{e) cot e-{-klk-\-

\ \v{e) = o.

Because any linear combination of integrals of a differential equation


12.4 is also an integral, we consider now the possibility of representing
the integral giving the temperatures within the sphere by a form
(12.7)

r = ^

Ckr%{d\

each cjc being a constant. T h i s form must then comply with the conditions
upon the sphere's surface, say where r = a; that is to say, the relation
(12.8)

CO

^ka'vi,{e)

k-Q

must be fulfilled. T h i s clearly poses the questions: (1), whether the function f{d)y though it was arbitrarily chosen, can be represented i n this way
by a series of integrsds of the ordinary differentisd equation 12.6, and (2),
if so, how the constants CQ, ci, <r2 ' " " > are to be determined to yield such
a representation. W i t h such constants, formula 12.7provided it is
convergentyields the required integral of partial differential equation
12.4, and therefore yields the temperatures within the sphere.
W e shall not pursue this large subject further at this point. W e shall,
however, touch upon it again at appropriate points below, i n our analysis
of some differential equations which involve parameters.

The Gamma Function

223

12.3. The gamma function


T h e gamma function, whose symbol is T{s)y is defined when j > 0 by
the formula
r(.) =

(12.9)

a'-'e-"

doc.

It can be seen directly that r ( l ) = 1, and it can be shown that V{i) =


v..

T h r o u g h an integration by parts, formula 12.9 can be given the form


+
a=0

If-.'
- /

aU-"" da.

s Jo

I n this the integrated part of the right-hand member is zero since a'e "
vanishes at a = 0, and approaches zero as a * . Hence V{s) =
- r ( j + 1), and, by a n m-fold iteration of this formula,
s

(12.10)

ns) =

^
s{s +

l]{s

-\- 2]

m -

- r ( ^ + ;7i).
11

T h i s relation serves to define T{s) for negative values of j , except the


negative integral values, for when m is large enough to make {J + m\
positive, the right-hand member of equation 12.10 is defined, and gives
V{s) its value. T{s) is left undefined at j = 0, 1 , 2 , * . It
becomes infinite as these points are approached.
I n other forms, relation 12.10 is
s{s +

i][s

+ 2]

[s + m-U

^^'f^'

(12.11)
+

U + 2m -

2] =

^^^^^^

T h e first of these is precisely (12.10). T o obtain the second from i t ,


we need merely replace s by j / 2 , and then multiply each of the m factors
by 2.
F r o m (12.11), with j = 1, it is seen that, for any integer m,
(12.12)

ml = T{m - f 1).

W e can now write formulas for the products of any set of successive
integers, or of successive even or odd integers, for if q is any integer,
formulas 12.11 and 12.12 yield directly the first two of the relations

224

Differential Equations with Parameters

(12.13)

2? + 2)

+ 4| [2q + 2m\ =

l2,+ l J i 2 , 4 - 3 ) . . - l 2 , + 2 ; - l i =

'

f2? + 2 . ! ! , !

2'"l2q]\{q-\-m\\

T h e third relation can be verified by observing that its product by the


second one can be evaluated by the first one.

12.4. The Legendre equation


T h e change of variable x = cos 6 transforms differential equation 12.6
into
(12.14)

{1 - x^]u" - 2xu' + k[k + \]u = 0.

This is called the Legendre equation, i n honor of the French mathematician


A . M . Legendre (1752-1833). Its parameter k may be regarded as
eligible to take any constant value. T h e differential equation 12.14 has
singular points a t x = l , x = 1 , and x =
and each of these is a
regular singular point.
T h e point x = 0 is an ordinary point. I n powers of x, therefore, the
integrals of (12.14) have the form u = ^

Cnx".

F o r them,

n-O

n~0

and

n= 0

n=-0

By the substitution of these series, differential equation 12,14 is converted


into the equation

l(n +

+ 2 ) f + 2 -in

\)nc. -

2nCn + k{k + l ) f | x = 0,

n-O

and, by equating the coefficients of each power of x to zero, we obtain


the relations
{n + l ) { n + 2 k + 2

{{n -

1) + 2n - k(k + 1)|<: - 0.

The Legendre Equation

225

Thus
(12.15)

{k~n]lk

Cn+2

\-\-n]

= - , r r r r ; r - : 7 i '^^^ = 0 , 1 , 2,
\n +

l]{n-\-2\

Formulas 12.15 show that, if CQ = 1 and ci = 0, all coefficients c with


an o d d subscript are also zero, and
^( + 11

i A - 2 ) A i A + i n A + 3|
4!

C2m

(-1)

2m + 2|{A - 2m + 4j
{k ~ 2]k{k + 1|{^ + 3| 1^ + 2m -

(2m

If Co 0 and ci = 1, all coefficients


zero and
\k-l]{k

c^ =

1},

with an even subscript are also

2]

3-,

[k -

i\{k -\]{k

+ l][k

+ A}

[k-2m

+ l]{k-2m

3]

' 2 m + 1}!
W e thus obtain the two particular integrals
00

0.1

= 1 + 2 /

^17 !^ -

2m + 2j (A - 2m + 4}

m=l

^J2.i6)

^
V
"^^ = * +

k\k +

\\\k + Z\ {A + 2 m - l ) ; . 2 ^ ,

(-1)"*
* V

2/ {^;;rw!
+ n
- 2m + 3 !
{ A : - l ) A i A + 2| {A + 2 m ) x 2 - + ^

226

Differential Equations with Parameters

A t the singular point x = t (and similarly at * = 1) both indices


are zero. There is therefore only one integral of the form

For this it is found that


{h-n\{k

l+n\

2|n + l p

" =

^' 2>

T h e choice CQ 1 thus yields


ci =

C2 =

{k-

\]k{k-\-\\{k-\-2\

and the integral, i n powers of


(12.17)

1 ], is accordingly

1.1 = 1 +
{k-n

\\[k~n-{-2]

\k-\-n\

2"(n!p

n =l

{x -

i r

T h e second integral about this point involves log {x 1), and is of the
form
to

(12.18)

1.2 =

2 ^f'-

1 ) " - huiAx)

log

11, bo^ 0.

n-O

T h e coefficient bo may be assigned the value 1, and the remaining coefficients bti are then determinable by the method of Section 11.6.
A t the singular point x
the indices are k and {k + 1). T h e
form
1

u =
leads to the relations 2kci
{n-2k-

\]ncn = {A -

-ft

I
n -0

[X

0, a n d

n + l l l i - + 2l<:_2,

n = 2. 3, 4,

These cannot be fulfilled if 24 + 1) is an even positive integer. I n every


other case they can be fulfilled by assigning to each c w i t h a n odd sub-

The Legendre Equation


script the value zero, and to CQ the value 1.
C-2m ~

\k-2m

227

Then

l][k-2m-\-2]

2m{2k -

2m +

^2m2

1|

W e m a y repeat this deduction with k replaced by (A -f- M - T h e


relations thus obtained can be fulfilled if {2A: + 11 is not an even negative
integer. W e thus obtain the two integrals

w 1

1 +

( - i r ( A - 2m + I H A - 2m + 2}
(12.19)

m-l

2"'m!(2A: - 2m + l\{2k + 2m + 3]

.2

1
X

1} [x

k+1

1 +

{A + l ) { ^ + 2|
m-l

{2k -

2m

\k-\-2m\

{2A: + 2m + l l

2"'m!{2A 4- 3112A - f 5}

2m"

unless {24 + 1) is an even integer. I n the excepted cases one of the


formulas 12.19 gives an integral, but the other integral involves log x.
PROBLEMS
1. Find an integral of the dllTerential equation

Jt^u" + ; t V -

[kx + 2\u = 0,

in powers of *.
2. F i n d an integral of the difierential equation
(12.20)

x^u" +

[1 -

x}u' + Au = 0,

in powers of x. [This is called the Laguerre equation, in honor of the French mathematician E . Laguerre (1834-1866)1.
3. Find two integrals of the differential equation
(12.21)

u" -

2xu' + 2ku =- 0.

in powers of x.
fThu is called the Hermite equation, in honor of the French mathematician a Hermite (1822-1905).]
4. F i n d two integrals of the differential equation
4:cV'+4xV-

|4Jt2-l

in powers of Xy assuming that 2k is not an integer.

-2x\u

0,

228

Differential Equations with Parameters

5. Find two integrals of the differential equation

in powers of

+ 1).

6. F i n d two integrals of the differential equation


l-r -

l l V-

2k{x -

l l u ' + {k(k + 1) + (x -

\)^\u = 0.

in powers of {x 1}.
7. Find two integrals of the differential equation

3k
2x

u = 0.

in powers of 1 /x.
8. Find two integrals of the diflferential equation

in powers of 1 /x.

12.5. The Legendre polynomials


W h e n the parameter k i n the Legendre equation is zero or a positive
integer, every term i n integral 12.17 subsequent to that i n {x 1)*
vanishes. This integral is therefore a polynomiEil of the degree k. It is
called the Ath Legendre polynomial, a n d is denoted by Pk{x)- Its formula
12.17 can be written compactly by the use of evaluations 12.13 as
k

(12.22)

= ^
n-O

2''{1'.^"^'
{k-n]\

~ ^' "

Similarly every term i n the integral u^^-[{x) of equation 12.19 subsequent


to that i n (l/jc)* vanishes. This integral is therefore also a polynomial.
B y the use of evaluations 12.13 its formula can be written
It/21
.12 23)

(12.23)

..i-x

{ - ^ r m 2 k - 2 m ] l

2^^^^^2k\l{k-m]l{k~2m}l'

'

m-O

where lk/2] stands for the largest integer contained i n k/2.


W e can conclude at once that the two integrals 12.22 and 12.23 are not
linearly independent, for if they were, every integral of the differential
equation would be a combination of them, and so would be a polynomial,
whereas that is not so of integral 12.18, which involves l o g {x 1],
Thus Pk{x) is a multiple of u^j{x). T o determine what multiple, we
need only compare the coefficients of any single power of x i n the two

The Legendre Polynomials

229

functions. T h e coefficient of x* i n equation 12.22 is


whereas i n equation 12.23 it is 1. Thus

{2k}\/2''lk\}\

{2k\\

2*{A!!
T h i s yields the expression of Pk{x) i n powers of x, namely,
[*/2)
P
_ 1 V
(-ir{2A-2ml!
^*W-2*Z/m!{A-ml!lJt-2m)!*

n 2 24^
(12.24)

T h e first seven Legendre polynomials are thus seen to be


- 1 ,

11.

- 30** + 3 ) ,
- 7 0 * ' + 15x),
' - 315x* + 105*=^ - 51.
Since every Legendre polynomial hzis the value 1 at x = 1, by equation
12.22, and since Pk{x) involves only even powers of x when k is even, and
only odd powers of x when k is odd, it follows that
(12.25)

P * ( l ) - 1,

P*(-l) = (-1)*

A compact formula for Pkix) is


(12.26)

^*W =

,1*^-1)'.

T o prove this we differentiate the binomial expansion


k

(y2 _ 1 Ifc =
*

>
m-O

^ 1)"*A!

2fc-2m

m\lk-m}\

k times, to get
[k/2\

. ,
rfx* '

V
'

(-l)"*|2A-2m}!

.k-2m

1/ ml\k - m]\{k - 2m\\

m-0

A comparison of the last equation with equation 12.24 proves the formula

230

Differential Equations with Parameters

T h e Legendre polynomials are interconnected by the relations


xPk'ix) Pk'(x) -

(12.27)

Pk+x'{x) {x^ {k +

\]Pk+iix)

xPk-i'{x)

Pk-i'ix)

\\Pk'{x) -

{2k +

l]xPk{x} +

These are called recurrence formulas.


that an equivalent form of it is

kPkix) =

0,

0,

kPt-i(x)

kxPkix) +

{2k +

Pk-i'ix)

\]Pk{x) = 0,
=

0,

APA_I(X) =

0.

kPk-dx)

T o prove the first one we observe

B y substituting for Pk{x) and Pk_i{x) their forms 12.24, and performing
the indicated differentiations, the formula is proved. T h e second formula
12.27 can be proved in the same way, and the remaining ones can then
be inferred from the first two.
Some important quadratures of Legendre polynomials can be evaluated.
Thus, to begin with,
(12.28)

2,
0,

j',Pk{x)dx

if
if

>t = 0,
A > 0.

This is obvious if A = 0, since Po{x) = 1. W h e n A > 0, a quadrature


of the first relation 12.27 from 1 to 1, w i t h an integration by parts of the
terms i n Pk{x) and Pk-i{x), yields the equation
[xPkix) -

Pj^i{x)]l,

{/: +

1 j / _ \ Pkix) dx =

0.

N o w , by (12.25), the integrated part of the last equation is zero.


result (12.28) then follows.
A more general evaluation is
[0,
(12.29)

\\x^Pkix)

dx =

The

if ffl < i ,

I 2^+M/:!P

This also is obvious when k = 0. T o prove it for other values of A, let


the first formula 12.27 be multiplied by x"^ and then integrated. T h a t
yields the equation
[x^+'Pk(x) -

x"^Pk-i{x)]l,

{A +

m +

1 i / _ \ x'-Pkix) dx

m f\x"'''Pk-i{x)dx

= 0,

The Legendre Polynomicds


wherein the integrated part vanishes, by (12.25).

-1

x^Pkix) dx ==

k -\- m -\- \ J ~\

231

Hence

x - - i P f c _ i ( x ) dx.

A n m-fold iteration of this relation yields


- 1 x'^Pkix) dx = -r
{ A - m + 3 l { / t - m + 5 j - - - l y t + m-|-lj
1
Pk-m{x) dx.
- 1

Evaluation 12.29 then follows, by equations 12.28 and 12.13


W e may now show finally that
[0,
(12.30)

|_\ PH{x)Pk{x) dx =

2
2k + 1

i{

k ^ k,

if

k = k.

T o do so we replace the polynomial Ph{x) or Pk{x) that is of the lower


degree, or one of them if h = k, by its form (12.24) i n powers of x. Quadrature 12.30 then appears as a sum of quadratures of type 12.29, a n d
evaluation 12.30 follows.
Result 12.30 is applicable to the discussion of Section 12.2. T h e
function there denoted by vic($) is Pk(x) when x = cos 6. Thus relation
12.8 is, effectively,!
CO

f{x) =

cyz^Pk^x),

f{x) being an arbitrary function. T o determine the coefficients Ck we


multiply the above relation by Ph{x), with any A, and integrate it term
by term from 1 to 1. Every resulting quadrature on the right except
that for which k = k then vanishes, by quadrature 12.30. T h e integrated
relation therefore reduces to
^

f{x)PH(x)dx

2
CKa^-,

and this determines Ch*


ASSIGNMENTS
1. By the use of formula 12.24, show that

ax
and thus derive the second of the recurrence formulas 12.27,

232

Differential Equations with Parameters

2. From the first and second of the recurrence formulas 12.27, derive the remaining
ones,
3, Show, by the use of formulas 12.16, that
( 1)*^*

; '

^ uo.iW)

when

k is even,

_r
2

Pkix)

no,jW,

when A U o d d .

4. Show that when k is zero or a positive integer the Laguerre equation 12,20 has
the polynomial solution
k

(,2.M)

^'W =

I*^I^X^'^=
n =0

(This is called the kxh Laguerre polynomial.)


5. Show that

(12.32)

iiW

= X -

Liix)

1,
-

4x + 2,

Lz{x) x ' -

U{x)

= X* -

+ 18x -

6,

16x' + 72x* -

96x + 24.

6. Show that when k is zero or a positive integer, the Hermitc equation 12.21 has
the polynomial solution
f*/2I
(12.33)

2^

H.(x)

c-ir

(This is called the i t h Hermite polynomial.)


7. Show that
//o(^) = 1,
Hi{x) - 2x,
Ha(x) = 4x= ff3(^)

= 8x' -

HA{X)

i/6(:f) -

2,
12x,

16X* - 48x2 _j.


32x^ -

Hiiix) = 64x" -

160x + 120x,
480x* + 720x -

120,

The Hypergeometric Equation

233

8. Show that
d
dx
and. from this, that
0,
9. By the use of formulas 12.32, show that

when A -

0, 1, 2, 3, 4.

10. Show that

0,
1,

e-'Lk{x) dx =

if k
if

0,
A = 0.

12.6. The hypergeometric equation


T h e differential equation
(12.34)

x{\ - x]u" + (T -

(a + /3 + 1)^1' - a0u = 0,

i n w h i c h a , |9, and y are parameters, is called the hypergeometric equation.


Its singular points are at x = 0, x = 1, and x = oc, and are a l l regular.
T h e indices at x = 0 are zero a n d 1 y\. O n substituting the form
as

u X * ^ rx" into differential equation 12.34, the relations


n-O
(12.35)

7 + i + 1 1 + A +

{a + A + | (i3 + A + n]c, = 0,
n = 0, 1, 2, ,

are obtained. W h e n k 0, these can be fulfilled, except possibly when


7 1) is a negative integer, and with that possible exception,
(12.36)

Cn+i

, w
, 1 Cn.
{1 + nj{7 + n]

T h e power series w i t h these coefficients, when CQ is assigned the value 1,


is denoted by F ( a , /3; 7; x). It is called the hypergeometric function a n d has
the formula
(12.37)

F(a,P;y;x)
go

= 1 +

a j g + l i ( a + M g ( g + l ! ' ' ' !j3 + M + ,


l " + l } ! 7 l 7 + l | ' ' {y + n\

234

Differential Equations with Parameters

T h e series is convergent Vk^hen 1 < x < 1, and is unchanged when a a n d


0 are interchanged. Differential equation 12.34 thus has the integral
(12.38)

o.i(x) -

F{a,p;y;x).

W h e n i = 1 7, equations 12.35 can be fulfilled, except possibly when


7 1) is a positive integer, and with that possible exception,
(12.39)

r
^
1 + 1 1 2 - 7 + l

Cn+l =

= 0, 1, 2,
T h e series w h i c h results from this can be found i n the usual way. It c a n ,
however, also be obtained by observing that relation 12.39 is derivable
from 12.36 merely by making the replacements ( a 7 + l j for a;
1/3 7 - f 1! for j3; and 2 7) for 7. T h e second integral i n powers of
X is thus
(12.40)

uo,2{x) = x'-^Fia

- 7 + 1, /3 - 7 + 1; 2 - 7; ;c).

W h e n {7 1 j is an integer, the system of equations 12.35 may include a n


equation that is not fulfilled. Whether it does so or not depends upon the
values of a and /3. W h e n it does so, there is only one integral i n powers of
X, the second one then involving log x.
T o obtain the integrals i n powers of {A: 1), we transform differential
equation 12.34 by the change of variable \ x = z into
d^u

^1 - ^1 ^

+ {(a + ^ - 7 + 1) -

du

(a + i3 + 1)^) ^ - a^M = 0.

This is recognizably again a hypergeometric equation, i n fact one that is


obtainable from equation 12.34 by making the replacements z for x\ a n d
{a + /3 7 + 1) for 7. T h e integrals, i n terms of the original variable,
are therefore, by equations 12.38 and 12.40,
(12.41)

ui,x{x) -

F ( a , /3; a + /3 -

ui.2{x) = 1 -

x]''-''-'F{y

7 + 1; 1 -

^, 7 -

x\

;7 + 1 -

a -

^; 1 -

^)-

T o obtain the integrals in powers of 1 /x, we make the change of variables


u x~*'w,

X \/Zj

which transforms differential equation 12.34 into


-

^1 ^

+ {(a - /3 + 1) -

(2a - 7 + 2)z\ J

235

The Hypergeometric Equation

T h i s is again a hypergeometric equation, and is seen to be obtainable from


equation 12.34 by making the replacements w for ; z for x] (a 7 + 1}
for j3; and {a j8 + 1) for 7. Formulas 12.38 and 12.40 may thus be
d r a w n upon for the integrals w, and from these formulas it follows that
(12.42)

.i(^) = ! i M ' " ^ ( a , a - 7


=

u^Ax)

13 -

ir/x]^F{^,

+ i;a-)9 + i ; lA),

7 +

1; ^ -

a +

1;

M a n y familiar functions are of the hypergeometric type.


by the following list.
1 -

= P{-a,

x]"

= -i^(l, l;2;x).

-log{l - * |
X

1
2x

log

1; l ; x ) .

+x

1 -

1
- sm 1* X
X

1
_ i
- t a n ^X

2)H
1 - x M

\2

cos \k sin ^ x|

2' 2'

^ V i ' T ' 2 '

It can also be shown that


e' =

lim F

0* 00

COS X

- sin X = l i m F
X

<^ *

V'

2'

lA).

This is shown

236

Differential Equations with Parameters


ASSIGNMENTS

11. Find integrals 12.41 for equation 12,34 without using a change of variable,
12. Find integrals 12.42 for equation 12.34 without using a change of variable.
13. By the use of scries 12,37, show that, when {7 11 is not a negative integer,
F'ia,

&x >;

- F ( + 1, ^ + 1; 7 + l ; x).
7

14. Show that, when of is a negative integer, differential equation 12.34 has a
polynomial solution, this being uo,2W if 7 is one of the numbers 0, 1 2 , ,
ja + 1}, and otherwise uo^iW.
15. Show that when a => 1, or ^ + 1, differential equation 12.34 has two integrals
in powers of jc, even when (7 1} is a positive integer.
16. Show that the change of variable
-

{1 - x\'^-'^^w

transforms differential equation 12.34 into the hypcrgeomctric equation


x[\ - x W

[7 -

(27 -

- ^ + l)xl;' -

h -

a\{y-p\w

= 0.

17. By repeatedly differentiating equation 12.34, show that every derivative of


F (a, ^; 7; x) also fulfills a hy|>ergeometric equation.
18. By use of the result of assignment 17, show that F"{OL 2, /3 2; 7
2; JV) is a
linear combination of the functions uo,iW ^ " d uo.2(^), when 7 is not an integer.
Then, by comparing coefficients of like powers of
show that

2. , -

2, . -

2; .) = l ^ - 2 | | a - H . - 2 H , - l )
\y ~ 2\\y - \ ]

12.7. Equations that are solvable by hyper geometric functions


Some differential equations are transformable into the hypergeometric
type by suitable changes of variables. S u c h equations are therefore solvable by hypergeometric functions. A n y differential equation of the f o r m
(12.43)

a{x -

n l \x ~

r2l" +

b{x -

r^W + c -

i n w h i c h a, b, c, r i , r 2 , and rs are constants, and


this way. For the change of variable
(12.44)

x =

(r2-ri}z

0,

9^ r j , can be solved i n

r1

transforms (12.43) into differential equation 12.34 with values a, /3, and 7*,
obtainable from the relations
a + )3 + 1 = b/a,

a& = c/a,

7 = ^-j^^^'^*
a\ri - ra

Equations Solvable by Hypergeometric Functions


T h e singular points x = ri and x =
respectively.
Example 1.

237

then appear as z = 0 and z

T o find the integrals of the differential equation


4 ) u " H- [Ix + l\u' -

{2x^ - 2 x -

3u = 0

about the singular point x = 2.


I n factored form the coefficient of u" is 2[x 2)|x + 1). W i t h
T\ 2 and rz 1 , the change of variable 12.44 is x= 3 ^ + 2, which
transforms the given equation into
d^

1-1
2~
2^\

dz

, ^

+ - = 0.

This is the hypergeometric equation w i t h a = 3, j8


therefore has the integrals
0.1 =

FO,

- h h

It

z),

-3;

uo.2 = z-^F{h

z).

-I;

I n terms of the original variable x, these integrals are

"o.iW ^f(3

V
2-

W0.2W

Example 2.

'-I-

2-

2'2'

x\

V2'

" 2' ~ 3 ~

T o find the integrals of the differentiail equation


{3^2

15x + 1 8 i " -

6x]u'

+ 1 = 0,

i n forms that are appropriate to large values of x.


T h e change of variable 12.44 is x = z + 2, and the transformed equation is hypergeometric with a = i, 0 ^, y = ^. Large values of z
correspond to large values of x; therefore w^,i and ^ , 2 are appropriate.
These integrals are
1

c-/^
2

X -

Example 3.

^V3'2'5^"^/

1
X

^ 2

/2
2

\3

6'3'x-2)

T o find the integrals of the diflferential equation


1 - e']u"

-\-W

by making use of the change of variable

e'u =

= z.

238

Differential Equations with Parameters

This change of variable transforms the equation into the hypergeometric


type, and makes x = 0 correspond to z = 1. F r o m formulas 12.41,
therefore,

PROBLEMS
9, Find the integrals of the differential equation
{x^ -

\ \u" +

|5JC +

4]u' + 3u = 0

about the point x = 1,


10, Find the integrals of the differential equation

about the point x = ^ ,


11, T h e Legendrc equation 12.14 is of the hypergeometric form.

Assuming that

2k is not an integer, find for it a pair of integrals as hypergeometric functions that arc
appropriate to large values of jr,
12, Find an integral of the differential equation
-

2W'

{1 -

+ u -

about the point JC = \ / 2 .


13, Find two integrals of the differential equation

about the point JT = 0, by using the change of variable x =

"sfz.

14, F i n d two integrals of the differential equation


{1 -

*-2^Ju" -

3u' + Ar'^'u = 0,

in a form that is suitable for use with large positive values of x, by making the change
of variable e~^' = z.

12.8. The Bessel equation


T h e differential equation
(12.45)

x\" + xu' +

- jt^jw = 0

is called the Bessel equation, i n honor of the G e r m a n mathematician a n d


astronomer F . W . Bessel (1784-1846).
It has singular points at = 0 a n d
X = CO, but only .v = 0 of these is regular. W e shall therefore consider
only the solutions i n powers of x.

The Bessel Equation

239

T h e indices at x = 0 and k and ~k. F o r the integral associated w i t h


the index A, the relations giving the coefficients are
1 ^2k\ci = 0,
n{n + 2k]c Cn~i = 0,

n = 2, 3, 4,

W h e n n is even, say n 2m, these relations are


-1

2'^m{m + A}
and yield the evaluations
m

"^"^

(-1)
22"m!{A + l ) { A + 2 l jA + m l " * ' '

A l l the coefficients w i t h odd subscripts may be taken to be zero.


CQ is assigned the value l / [ 2 * r ( A + 1)], the series obtained is
_

(12.46)

Jk{x)

When

fc+2m

'

'

T h i s is called the Bessel function of the order k, of the first kind. T h e integral
associated with the index A is similarly obtainable. W e find thus that
(12.47)

i(x) = /*(*),

U2{x) =

J_j,(x).

T h e integrals 12.47 are linearly independent unless A is an integer. W h e n


A is a n integer, a second integral, linearly independent of MI(X), involves
log X.
T h e values of the Bessel functions are obtainable, for many values of A,
from tables that are i n common use; hence they are known i n the same
sense as the values of log x or cos x. W e may therefore regard a differential equation as solved if its integrals are expressed i n Bessel functions.
Example 4.

T o find a pair of integrals of the differential equation


" -1- 9x^u = 0.

T h e change of variables
u = 's/x w,

transforms the differential equation into

A/IZ

240

Differential Equations with Parameters

This is the Bessel equation w i t h k \. Thus wi =


I n terms of the original variables, therefore,

J^(z),

and

1V2

PROBLEMS
15. F i n d the general integral of the differential equation
1

4
9*'

a = 0

by making the change of variable u = "x/x w.


16. Find the general integral of the differential equation

x'u" - xu' +

{x2 + | ) u

= 0

by making the change of variable u ^ xw.


17. Find the general integral of the differential equation

by making the change of variable 4 i ' = z^.


18. Find a pair of integrals of the differential equation
4 u " + 2Sx\ = 0
by making the change of variables u = \/x w.
19. Find a pair of integrals of the differential equation

as power scries in jr, assuming that 2k is not an integer.


20. Find a pair of integrals of the differential equation

as power series in x, assuming that k is not an integer.

12.9. Bessel function graphs and inter-relationships


Formula 12.46 shows that every* Bessel function Jjt(x) of a real order k
is real for positive values of x. W e shall show that it is an oscillatory function for such J , that is, that its graph has the shape of a wave.
T h e change of variable u = w/'\/x transforms the Bessel equation i n t o
(12.48)

w"

1 -

k^-V

w ^ 0

241

Bessel Function Graphs

L e t xi be any value greater than


and set a = v ' l k^/xi^. T h e
function V = sin a(x xi) is an integral of the differential equation
v" + a^v = 0. It follows from this equation, and equation 12.48, that
{vw'

v'wY

x^ ^

4x^

vw.

and hence, by a quadrature from x\ to x\ -f- ir/a, that


aw

4- aw(j:i)

x^

Ax

w{x) sin

x\) dx.

T h i s woxild be impossible if w{x) maintained its sign over the whole interval
from x\ to x\ + v/a, for then the two members of the equation would have
opposite signs. Therefore w{x) changes its sign on any interval of the
length ir/a to the right of the point x i , and the function Jk{x) w{x)/'\/lc
is accordingly oscillatory. T h e graphs of JQ{X),
J\{X)J
and Jzix), are
shown i n F i g . 33.

Fio.

33.

Let the zeros of Jk{x) with any chosen real value k be a i , 2 , 3 ,


T h e functions v(x) = Jkiocnx) and w{x) = Jkiax), in which a is a parameter, are integrals of the respective differential equations
.2

xv" +

xw"

-\-w'

v'-\-

-\-

Oln X
k^

V = 0,

w; = 0.

N o w the result of multiplying these equations respectively by w and by

242

Differential Equations with Parameters

Df and adding them, is the relation


{x{v'w -

vw')]' - h {cxn^ -

a^}xvw = 0.

A quadrature from 0 to 1 gives this relation the form


'(1M1)

+ {a2 -

j^xvwdx

= 0,

or, more explicitly,


(12.49)

a7jb'(a)7fc(a) = (a^ - a^^j

xJk{ar,x)Jk{ax) dx.

T h e derivative of this w i t h respect to a is the relation


(12.50)

aJfc'(an)^* (a) = 2 a \^ xJ^MJ^iwc)


-

dx +
x'JkMJk'iax)

n')

If a is now assigned any one of the values


reduces to

dx.

other than a , relation 12.49

^ = fo xJk{anx)Jk(a^)

dx.

O n the other hand, as a > an, the limiting form of relation 12,50 is
anJk\<Xn)

= 2a

xJk\otnX ) dx.

W e have thus found that


(12.51)

xJk{anx)Jk{ax)

dx =

0,

if a 5^ any

iA'\an)y

if

a = a.

Evcduations 12.51 are applicable to the discussion of Section 12.2. T h e


first of equations 12.3 is transformable into the Bessel type. If a function
f{x) is representable as a series of Bessel functions, thus,
(12.52)

f(x) =

2 Cm-Jk{ctmX),
m-1

a determination of the coefficients Cm is made possible by evciluations 12.51.


For, if relation 12.52 is multiplied by xJk{anx)y w i t h any n, a n d is then
integrated term by term, the resulting quadratiu"cs, except one, vanish, by
(12.51).

T h e result is the relation


0

xf{x)Jk(anX) dx =

CnWk^(0Cn\

and thereby the coefficient c is determined.

243

Answers to Problems
ASSIGNMENTS
19. Show, by use of formula 12.46 and the evaluation r ( ^ ) = -yjr, that
/2

/2
J}^{x) ^ /

COS X.

20. Show, by use of formula 12.46, that

dx
that is, that
Jk\x) "-Jkix)

(12.53)

-Jjt+iW-

21. Show, by use of formula 12.46, that

dx
that is, that
Jk'ix) -\--Jk(x)

(12.54)

Jk-i{x).

x
22- From relations 12.53 and 12.54, derive the further recurrence formulas

Jk-iix)

+ Jk+i(x) -

2k
- Jkix),
X

Jk^iix)

Jk+lix) = 2Jk'{x).

23, Show that in the product of the two series


^ 2
^2=2! ^ 2 ' 3 !
.-x/2t

1 _

2z^2W

^
^

2^3

the total term in z is Joix), and the total term in z is zJi(x).


24. Show that
f^^ xJk{fimx)Jk{M

dx = 0.

if /3i, /32, 03, , are the values of ;c for which Jk'ix) = 0, and /3m 7* p.

ANSWERS TO ODD-NUMBERED PROBLEMS


6[-t n - i

2H* -

3} . . . {A -

n-

1}

244

Differential Equations with Parameters


m

l 2 m } ! r C i * - m + 1)
m>0

( - i r 2 ^ " ' r ( i A + |)

U2

|2m +

m-0

li!r(iA+i-m)

.Sftt+l

m-l

(-l)n2'"2-5-8

i3m + l i !

00

7. Ul -

1
n-0

I2n + l l !

U2

n-0

9. Ul = F

"2

11.

Ul

\2'2'2'

F l A + l , A + l ; 2 A + 2;

13. U l

''F(i,i;hx\

15. a = ^/x

UlJ^iix) +

f2-/-4a(*)i-

17. a = fiJ^,(2<r'^^) + .2y-?i.(2/''").


19. ai -

Qjt(x),

U2 = Q-kix), with

0
.n+k

+
n~0

1)

13m -

1|

INDEX

Acceleration, 33
of gravity, 34
Addition of solutions, 134, 145, 153, 180
Adjoint equation, 183
Airplane problems, 49, 53, 128, 130
Amplitude, 157
Approximate integrals (solutions), of differential equations of first order, 78,
83, 88
of diflTercntial equations of second order,
122
of simultaneous differential equations,
81, 86, 91
Approximation of integrals (solutions), by
graphs, 78, 81, 123
by polynomials, 83, 84, 122
by successive approximations, 88, 91
Arbitrary constants, 5, 26, 65, 114, 152,
180, 188, 201
Arbitrary function, expansion

of, 222,

231, 242
A r c length, 45
Area, 45
Auxiliary equation, 135, 152
complex solutions, 137, 152
Bernoulli, James (1654-1705), 21
equation, 21, 22
Bessel, F . W_ (1784-1846), 238
equation, 238
functions, 239
graphs of, 241
Buoyancy, 36, 160
Cable, suspension, 54
Catenary, 46
Center of curvature, 125
Centrifugal force, 172
Changes of variables, 19, 107, 148, 189,
194, 240
Checking a solution (integral), 5, 116

Circuit, electric, 44, 176


Qairaut, A , Q (1713-1765), 104
equation, 104
Coefficient, of friction, 37
of resistance, 163
Coefficients, in expansions of arbitrary
functions, 222, 231, 242
linear, 24, 72
method of undetermined, 84, 122, 147,
149, 154, 201
Comparison of formulas, 146, 147
Complementary function, 135, 152, 180
Complete equation, 135, 180, 205, 213
Complex solutions, of auxiliary equation,
137, 152
of Riccati equation, 108, 137
Compounds, problems in, 39
Constants of integration, 5, 26, 65, 114,
152, 180, 188, 201
Convergence proofs, 93, 203
Coohng, Nevvton*s law of, 42
problems in, 42, 222
Coordinates, cylindrical, 220
polar, 47
spherical, 221
Curvature, 124
center of, 125
problems in, 127
radius of, 124
Curve, motion along a, 130
Curves, families of, 5, 116
integral, 4, 64, 65, 73, 78, 81, 98, 123
normals to, 44
of purniit, 49, 128
Cylindrical coordinates, 220
Damped vibrations, 163, 169
Damping factor, 164
Degree of a homogeneous function, 22
Dependence, linear, 186

246

Index

Differential equation of a family of curves,


5, 116
Differential equations, adjoint, 183
exact, 11, 118, 182
explicit for x or y, 104, 106
factorable, 100
family of curves, 5, 116
hypcrgcometric, 233
in which X or ^ does not occur, 115
involving a parameter, 219
linear, 17, 67, 134, 152, 180, 196
linear fractional, 72
of Bernoulli, 21, 22
of Bessel, 238
of Clairaut, 104
of Euler, 181
" of Hermite, 227
of higher order than the second, 152
of Laguerrc, 227
of Laplace, 219
of Legendre, 224, 238
of Riccati, 107, 110, 135, 137, 185
ordinary point of, 180, 201, 205
partial, 219
removal of a term, 189, 202
simultaneous, 26, 64, 81, 86, 91, 120,
196
singular point, 108, 208, 215
solvable, 9, 98, 114, 134, 152, 181, 182
with homogeneous coefficients, 22
with linear coefficients, 24, 72
with separable variables, 9
Differential of volume, 42
Differential operators, 140, 144, 153, 191
Differential systems (simultaneous equations), 26, 64, 81, 86, 91, 120, 196
Direction field, 3
isoclines of, 65
of simultaneous equations, 64
singular points of, 70
Discontinuous force, 174
Discontinuous solutions of a Riccati equation, 110
Discriminants, 74
Disintegration, problems in, 40
Electric circuits, 44, 176
Eliminants, 5, 26, 102, 121, 196
Envelope, 101
Equation, sfe also Differential equations

Equation, auxiliary, 135, 137, 152


complete, 135, 180, 205, 213
implicit, 2
indicial, 208
reduced, 135
replacement by a system, 29, 120
Euler, U (1707-1783), 139
equation, 181
formulas, 139
Exact differential equations, 11, 118, 182
Existence of solutions (integrals), 93, 185,
203
Expansion of arbitrary functions,
231, 242

222,

Exponents at a singular point, 208


F (a, ffi y; x), 233
Factor, damping, 164
integrating, 14, 17, 183
Factorable differential equations, 100
Falling, problems in, 34
Falling motion, 33
Family of curves, differential equation of,
5, 116
trajectories of, 68
Field, direction, 3
isoclines of, 65
of simultaneous equations, 64
singular points of, 70
First integral, 26, 114
Floating bodies, problems in, 160, 162
Flow, problems in, 40, 42, 222
Force, 33
centrifugal, 172
discontinuous, 174
intermittent, 173
repelling, 172
restoring, 159
Forced vibrations, 166, 169, 172
Formulas, comparison of, 146, 147
recurrence, 230, 231, 243
Fractional linear differential equation, 72
Free vibrations, 159, 163
Frequency, 157
Friction, coefficient of, 37
Frictional motion, 37
Function, arbitrary, expansion of, 222,
231, 242
complementary, 135, 152, 180
gamma, 223

Index
Function, homogeneous, 22
hyperbolic, 46
hypcrgcomctric, 233
Functions of, Bessel, 239
Hermite, 232
Lagucrre, 232
Legendrc, 228
G a m m a function, 223
General integral (solution), 5, 26, 100,
114, 135, 152, 180, 188
Geometrical applications, 44, 47, 127
Graphs, approximating integrals, 78, 81,
123
of Bcsscl functions, 241
Gravity, acceleration of, 34
Harmonic motion, simple, 156
Heat flow, 42, 222
Hermite, Q (1822-1905), 227
equation, 227
polynomials, 232
Higher degree, difTcrcntial equation of,
152
Homogeneous coefficients,
differential
equations with, 22
Homogeneous functions, 22
Hyperbolic functions, 46
Hypcrgeometric equation, 233
Hypcrgcometric polynomials, 236

247

Integrals (solutions), first, 26, 114


general, 5, 26, 100, 114, 135, 152, 180,
188
logarithmic, 211, 213
non-uniqueness of, 70, 72, 99
of simultaneous equations, 26, 64, 81,
86, 91, 120, 196
parametric representation, 29
particular, 5, 107
probability, 79
singular, 5, 98, 101
trivial, 180, 185
uniqueness of, 96, 185
Integrating factors, 14, 17, 183
Integration, constants of, 5, 26, 65, 114,
152, 180, 188, 201
Integration of a differential equation, by
differential operators, 140, 144, 153
by differentiation, 103
by power series, 83, 122, 201
by successive approximation, 93
by use of a known integral, 190, 191
by variation of parameters, 193
by undetermined coefficients, 84, 122,
147, 149, 154, 201
in parametric form, 29
Intermittent force, 173
Isoclines, 65
Jkix\ 239

Imaginary roots of auxiliary equation, 137


Imaginary solutions of Riccati equation,
108

Known function,

Implicit equations, 2
Independence, linear, 186
Indices at a singular point, 208
Indicial equation, 208
Initial conditions, 33, 78, 81, 122, 136
Integral curves, 4, 64, 65, 73, 78, 81, 98,
123

equation, 227
polynomials, 232
Laplace, R S. (1749-1827), 219
equation, 219
transformation, 197
Lependre, A . M - (1752-1833), 224
equation, 224, 238
polynomials, 228
Linear coefficients, differential equations
with, 24, 72
Linear dependence, 186
Linear differential equations, of first
order, 17, 67
of higher order, 152
of second order, 134, 180
simultaneous, 26, 196

Integrals (solutions), 2, 63, 114


addition of, 134, 145, 153, 180
approximation of
by graphs, 78, 81, 123
by polynomials, 83, 122
by successive approximations, 88, 91,
93
checking, 5, 116
existence of, 93, 185, 203

Laguerre, E , (1834-1866), 227

248

Index

Linear fractional differential equations, 72


Linear independence, 186
U p s c h i u , R- (1832-1903), 94

Period, 157
Periodic motion, 156
Phase, 170

condition, 94
Liquids, flow of, 40
Locus problems, 44, 47, 127
Logarithmic integrals, 211, 213

Point, ordinary, 180, 201, 205, 206

Mass, 33
M a x i m a and minima of integral curves,
65
Mbcturc problems, 38
Motion, against resistance, 34, 37, 163,
169
falling, 33
ictional, 37
Newton's law of, 33
on a curved path, 130
simple harmonic, 156
steady state, 170
under repelling force, 172
vibratory, 157
Multipliers, undetermined 26
Newton's law, of cooling, 42
of motion, 33
Non-uniquencss of integrals, 70, 72, 99
Normals to curves, 44
Oblique trajectories, 68
One-parameter family of curves, 5
Operators, differential, 140, 144, 153, 191
Order, 1
Ordinary differential equation, 1
Ordinary point, at infinity, 206
of a differential equation, 180, 201, 205,
206
of a direction field, 71
Orthogonal trajectories, 69
Oscillatory motion, 156
Pn(x), 228
Parameters, 29, 220
differential equation involving, 219
variation of, 193
Parametric representation of intcgrab, 29
Partial differential equation, 219
Partial fractions, 144
Particle, motion of a, 33, 130
Particular integral (solution), 5, 107
Pendulum, 161, 166

regular singular, 180, 208, 215


Polar coordinates, 47
Polygonal approximations, 78, 81, 123
Polynomial approximations, 83, 84, 122
Polynomials, hypergcometric, 236
of Hermite, 232
of L ^ ^ c r e , 232
of Legcndrc, 228
Population problems, 40
Post, rope around a, 58
Power series, 83, 84
solutions, 83, 122, 201, 213
solutions of simultaneous equations, 86
Probability integral, 79
Pursuit, curves of, 49, 128
Quadrature, 2
Radius of curvature, 124
Rates, problems in, 38
Real solutions from complex, 108
Recurrence formulas, 230, 231, 243
Reduced equation, 135
Regular singular point, 208, 215
Removal of a term of a differential equation, 189, 202
Replacement of an equation by a system,
29, 120
Resistance, coefficient of, 163
motion against, 34, 37, 163, 169
Resonance, 166, 170
Restoring force, 159
RiccaU, Count (1676-1754), 107
Riccati equation, 107, 110, 135, 137, 185
discontinuous solutions of, 110
general integral from particular, 107
real integral from complex, 108
reciprocal of integral, 107
relation to differential equations of second order. 111, 135, 137, 185
Repelling force, motion under, 172
Rope around a post or shaft, 58
Separable variables, 9, 219
Scries, expansions of arbitrary functions,
222, 231, 242

Index
Scries, in powers of jt,83, 84, 86,122, 201,
213
in powers of 1/x, 206, 215
Shaft, rope around, 58
Simple harmonic motion, 156
Simultaneous difTerential equations, 26,
29, 64, 81, 86, 91, 120, 196
direction field of, 64
Singular integral (solution), 5, 98, 101
Singular point, of a differential equation,
180, 208, 215
of a direction field, 70
Sliding particles, problems i n , 38, 131
Slope of a direction field, 3
Solvable differential equations, of first
order, 9, 98
of higher order, 152
of second order, 114, 134, 181, 182
Spherical coordinates, 221
Spring, vibrating, 159, 166, 168
Steady state, 170
Streamline, 4, 64
Successive approximation, 88, 91, 93
Suspension cable, 54
Systems of simultaneous equations, 26, 29,
64, 81, 86, 91, 120, 196
replacement of an equation by, 29, 120
Tangent lines, problems in, 44
Taylor's series, 83

249

Tractrix, 52
Trajectories, 69, 73
oblique, 68
of family of curves, 68
Transformation, by change of variable,
19, 107, 148, 189, 194, 240
Laplace, 197
Trivial solution (integral), 180, 185
Two-parameter family of curves, 116
Undetermined coefficients, method of, 84,
122, 147, 149, 154, 201
Undetermined multipliers, 26
Uniqueness of solution (integral), 96, 185
counter examples, 70, 99
Variables, change of, 19, 107, 148, 189,
194, 240
separation of, 9, 219
Variation of parameters, 193
Velocity, problems in, 33
Vibrating spring, 159, 162, 168
Vibration, damped, 163, 169
forced, 166, 169, 172J
free, 159, 163
Vibratory motion, 157
Volume, differential of, 42
Wronski, R (1778-1853), 187
Wronskian, 186

k rI
i

9015 00450 7557

You might also like