0% found this document useful (0 votes)
43 views305 pages

Lecture On Coefficient of Variables

Uploaded by

simonpatdu2
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
43 views305 pages

Lecture On Coefficient of Variables

Uploaded by

simonpatdu2
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 305

Earl D.

Rainville

Late Professor of Mathematics

University of Michigan

Phillip E. Bedient

Professor of Mathematics

FrankUn and Marshall College


A Short Course
in Differential
Equations FOURTH EDITION

The Macmillan Company CoUier-Macmillan Limited, London


© Copyright, The Macmillan Company, 1969

All rights reserved. No part of this book may be reproduced or


transmitted in any form or by any means, electronic or mechanical,
including photocopying, recording or by any information storage and
retrieval system, without permission in writing from the Publisher.

Second Printing, 1970

This book appears as the first 16 chapters of Elementary Differential


Equations, Fourth Edition, © copyright 1969 by The Macmillan
Company. Earlier editions of Elementary Differential Equations copyright
1949 and 1952, ©1958, and ©
copyright 1964 by The Macmillan Company.
Some material in this book is from The Laplace Transform; An
Introduction, © copyright 1963 by Earl D. Rainville.

Library of Congress catalog card number: 69-10390

The Macmillan Company


Collier-Macmillan Canada, Ltd., Toronto, Ontario

Printed in the United State of America


Preface to

the Fourth Edition

It is the belief of the present author that the success of the previous editions
of book has been largely due to Professor Rainville's ability to write in
this

a style that was readily accessible to the serious student. The present edition
attempts to maintain that accessibility while making some changes in content
and emphasis. In particular, the chapter on existence theorems has been
expanded to include a proof of an existence and uniqueness theorem.
As in earlier editions, I have attempted to exhibit both the techniques for
obtaining solutions and the basic ideas and theories behind these techniques.
A number of exercises of a more theoretical nature have been added to assist
in relating these techniques with the theory.
The omission of infinite series methods has made possible, even within the
bounds of a small book, a thorough treatment of topics essential to a first
course. A longer book, Elementary Differential Equations, Fourth Edition,
contains an extensive treatment of both power series and Fourier series
methods and an introduction to the application of the Laplace transform tech-
nique to boundary value problems involving partial differential equations.
vi Preface

The author is indebted to many persons who have influenced his thinking
in many ways. The most profound influence has come from the late Professor
Rainville, whose reputation as a superb teacher Hves in the minds of all his

former students. Especial thanks are also given to Esther Rainville for her
interest and encouragement.
I am grateful for the many comments and suggestions made by my students
and colleagues at Franklin and Marshall College. 1 am pleased to acknowl-
edge in particular Professor Paul W. Berg of Stanford University and Pro-
fessor Howard Sherwood of Illinois Institute of Technology for their very
valuable suggestions.

Phillip E. Bedient

Lancaster, Pennsylvania
Contents

CHAPTER 1 Definitions, elimination of arbitrary constants

1. Examples of differential equations 1


2. Definitions 3
3. The elimination of arbitrary constants 5
4. Families of curves 9

CHAPTER 2 Equations of order one

5. An existence theorem for equations of order one 16


6. Separation of variables 17
7. The exp u notation 21
8. Homogeneous functions 23
9. Equations with homogeneous coeflficients 25
10. Exact equations 29
viii Contents

11. The linear equation of order one 34


12. The general solution of a linear equation 38
Miscellaneous exercises 40

CHAPTER 3 Elementary applications

13. Velocity of escape from the earth 44


14. Newton's law of cooling 46
15. Simple chemical conversion 48
16. Orthogonal trajectories; rectangular coordinates 52
17. Orthogonal trajectories; polar coordinates 55

CHAPTER 4 Additional topics on equations of order one

18. Integrating factors found by inspection 59


19. The determination of integrating factors 63
20. Substitution suggested by the equation 68
21. Bernoulli's equation 69
22. Coefficients linear in the two variables 73
23. Solutions involving nonelementary integrals 77
Miscellaneous exercises 80

CHAPTER 5 Linear diflferential equations

24. The general linear equation 82


25. Linear independence 83
26. An existence and uniqueness theorem 84
27. The Wronskian 84
28. General solution of a homogeneous equation 87
29. General solution of a nonhomogeneous equation 89
30. Differential operators 90
31. The fundamental laws of operation 93
32. Some properties of differential operators 94

CHAPTER 6 Linear equations with constant coefficients

33. Introduction 97
34. The auxiliary equation; distinct roots 97
Contents ix

35. The auxiliary equation ; repeated roots 100


36. A definition of exp z for imaginary z 103
37. The auxihary equation; imaginary roots 105
Miscellaneous exercises 108

CHAPTER 7 Nonhomogeneous equations:


undetermined coefficients

38. Construction of a homogeneous equation from a specified


solution 110
39. Solution of a nonhomogeneous equation 113
40. The method of undetermined coefficients 115
41. Solution by inspection 120

CHAPTER 8 Variation of parameters

42. Introduction 126


43. Reduction of order 127
44. Variation of parameters 130
45. Solution of >-" + y =f(x) 135
Miscellaneous exercises 137

CHAPTER 9 Inverse differential operators

46. The exponential shift 139


47. The operator l/f(D) 143
48. Evaluation of [\lf(D)V 143
49. Evaluation of (D^ + a^)~^ sin ax and (Z)^ + a^)~^ cos ax 145

CHAPTER 10 The Laplace transform

50. The transform concept 148


51. Definition of the Laplace transform 149
52. Transforms of elementary functions 150
53. Transforming initial value problems 154
54. Sectionally continuous functions 157
55. Functions of exponential order 159
56. Functions of class A 161
X Contents

57. Transforms of derivatives 163


58. Derivatives of transforms 166
59. The gamma function 168
60. Periodic functions 169

CHAPTER 11 Inverse transforms

61. Definition of an inverse transform 174


62. A step function 178
63. A convolution theorem 184
64. Simple initial value problems 187
65. Special integral equations 194

CHAPTER 12 Applications

66. Vibration of a spring 201


67. Undamped vibrations 205
68. Resonance 208
69. Damped vibrations 211
70. The simple pendulum and the deflection of beams 216

CHAPTER 13 Systems of equations

71. The Laplace transform method 222


72. The differential operator method 227

CHAPTER 14 Electric circuits and networks

73. Circuits 231


74. Simple networks 234

CHAPTER 15 The existence and uniqueness of solutions

75. Preliminary remarks 242


76. An existence and uniqueness theorem 243
77. A Lipschitz condition 245
78. A proof of the existence theorem 246
Contents xi

79. A proof of the uniqueness theorem 249


80. Other existence theorems 250

CHAPTER 16 Nonlinear equations

81. Prehminary remarks 252


82. Factoring the left member 253
83. Singular solutions 256
84. The c-discriminant equation 257
85. The /j-discriminant equation 259
86. Eliminating the dependent variable 261
87. Clairaut's equation 263
88. Dependent variable missing 266
89. Independent variable missing 268
90. The catenary 271
Miscellaneous exercises 273

Index 275
A Short Course in Dilferential Equations
CHAPTER ±
Definitions, Elimination of

Arbitrary Constants

1. Examples of differential equations

A differential equation is one that involves at least one derivative of an


unknown function. As in equation (3) below, a derivative may be involved
implicitly through the presence of differentials. Our aim is to solve differen-

tial equations; that is, to find the unknown function, or functions, which
occur.
Differential equations arise frequently in physics, engineering, and chemis-
try, and on occasion in such subjects as biology, physiology, and economics.
The solution of differential equations plays an important role in the study of
the motions of heavenly bodies such as planets, moons, and artificial satellites.
The following are examples of differential equations:

dy
-— = cos X, (1)
ax

+ k'y = 0, (2)
1?
Definitions, Elimination of Arbitrary Constants [Ch. 1

(x^ + y^) dx - 2xy dy = 0, (3)

; cut, (5)

(6)

(7)

(8)

(9)

(10)

(11)

When an equation involves one or more derivatives with respect to a


particular variable, that variable is called an independent variable. A variable

is called dependent if a derivative of that variable occurs.


In the equation

L
d^i

dt
+5 i? — —C
di

dt
I
1
i = Eco cos cot (5)

i is the dependent variable, / the independent variable, and L, R, C, E, and


CO are called parameters. The equation

has one dependent variable V and two independent variables.


Since the equation

{x^ + y^) dx - 2xy dy = (3)

may be written
§ 2] Definitions

ax
or
-, T dx
dy

we may consider either variable to be dependent, the other being the inde-

pendent one.

Oral Exercise

Identify the independent variables, the dependent variables, and the


parameters in the equations given as examples in this section.

2. Definitions

The order of a differential equation is the order of the highest-ordered

derivative appearing in the equation. For instance,


;2.. / J.A 3

^ + 2h[^] +y = () (1)
dx^ \dx

is an equation of "order two." It is also referred to as a "second-order

equation."
More generally, the equation

F(x, ;;,/,..., y^) = (2)

is called an ";7th-order" ordinary differential equation. Under suitable


restrictions on the function F, equation (2) can be solved explicitly for j^"* in

terms of the other n + 1 variables x, y, y', y^"'^^, to obtain


. .
.,

/"'=f{x,y,y',...,/"-'\ (3)

For the purposes of book we shall assume that this is always possible.
this

Otherwise, an equation of the form of equation (2) may actually represent


more than one equation of the form of equation (3).
For example, the equation

x{y'Y + Ay' - 6x^ =


actually represents the two different equations.

-2 + J4 + 6x^ -2- V4 + 6x-


,
y' = ^^
or y
,
= ^^
4 Definitions, Elimination of Arbitrary Constants [Ch. 1

A function 0, defined on an interval a < x < b, is called a solution of the

differential equation (3) providing the n derivatives of the function exist on the

interval a < x <b and

for every .x in a <x< b.

For example, let us verify that

y = e^^

is a solution of the equation

dx^
^ + ^-6y
dx
= 0. (4)

We substitute our tentative solution into the left member of equation (4)
and find that

^+^_
dx dx
6j = 4e^^ + 2e'^ - 6e'^ = 0,

which completes the desired verification.

All of the equations we shall consider in Chapter 2 are of order one, and
hence may be written

dx

For such equations it is sometimes convenient to use the definitions of ele-

mentary calculus to write the equation in the form

Mix, y) dx + N{x, y) dy = 0. (5)

A very important concept is that of linearity or nonlinearity of a differential


equation. An equation

Fix,y,y',...,/"^)^0

is called linear if the function F is a linear function of the variables y, y', . .., y^"\

Thus the general linear equation of order n may be written

^oW dx + ^ b^ix) Tir^


dx
+ + K- i(x)
ax
+ ^ b„{x)y = R(x). (6)

For example, equation (1) above is nonlinear, and equation (4) is linear.

The equation
x^y" + xy' + (a-- - n^)y = 4x^

is also linear. The manner in which the independent variable enters the
equation has nothing to do with the property of linearity.
§ 3] The elimination of arbitrary constants 5

Oral Exercises

For each of the following, state whether the equation is ordinary or partial,
linear or nonlinear, and give its order.

1.
Definitions, Elimination of Arbitrary Constants [Ch. 1

EXAMPLE (a): Eliminate* the arbitrary constants Cj and Cj from the


relation

y= c^e~^''^C2e^''. (1)

Since two constants are to be eliminated, obtain the two derivatives,

/= -2Cle-2* + 3c2e3^ (2)

/' = 4cie-2* + 9c2e^^ (3)

The elimination of Cj from equations (2) and (3) yields

/' + 2/ = 15c2e3";

the elimination of Ci from equations (1) and (2) yields

y' + 2y = 5c2 e^"".

Hence
y" + 2y' = 3(>'' + 2y),

or

y" — y' — 6>' = 0.

Another method for obtaining the differential equation in this example


proceeds as follows. We know from a theorem in elementary algebra that
the three equations (1), (2), and (3) considered as equations in the two un-
knowns c^ and C2 can have solutions only if

e'
-/
-y"
§ 3] The elimination of arbitrary constants 7

y = c, e'"'-^ + C2 e"""" + •• + €„ e"-"^

will always lead to a linear differential equation

d"y d"-^y dy
''d?^'^d^^---^'"-^d-x^'"' =
''^

in which the coefficients Oq, a^,..., a„ are constants. The study of such
differential equations will receive much of our attention.

EXAMPLE (b): Eliminate the constant a from the equation

(jc - af + y^ = a^.

Direct differentiation of the relation yields

2{x — a) + 2yy' = 0,

from which

a = X + yy'.
Therefore, using the original equation, we find that

(y/y +y^ = (x + yy'f,


or

>'^ = x^ + lxyy\

which may be written in the form

(jc^ - y^) dx + 2xy dy = 0.

Another method will be used in this example as an illustration of a device


which is frequently helpful. The method is based upon the isolation of an
arbitrary constant.
The equation

{x - af + y^ = a^

may be put in the form

x^ + y^ - lax = 0,

or

— la.
X

Then differentiation of both members leads to

x(2x dx + ly dy) — (x^ + y^) dx


—5 = u.
8 Definitions, Elimination of Arbitrary Constants [Ch. 1

or

(x^ - y^) dx + 2xy dy = 0,

as desired.
It is interesting to speculate here about the significance of a- = upon the
argument just used. The student should draw a few of the members of this
family of circles and ask himself what is peculiar about their behavior at

x = 0.

EXAMPLE (c): Eliminate B and y. from the relation

X= J5 cos (oj/ + a), (5)

in which w is a parameter (not to be eliminated).


First we obtain two derivatives of .v with respect to /:


dx
dt
= —0)B sin {cot + a), (6)


d^x
dt-
^ = - orB cos (cot + a). (7)

Comparison of equations (5) and (7) shows at once that

-— + tolv - 0.
dt^

EXAMPLE (d): Eliminate c from the equation

cxy + c^x + 4 = 0.

At once we get

c(y + xy') + c^ = 0.

Since c ^ 0,

c= -iy + xy')

and substitution into the original equation leads us to the result

x^iy'f + x\vy' + 4 = 0.

Our examples suggest that in a certain sense the totality of solutions of an


nth-order equation depends on n arbitrary constants. The sense in which this

is true will be stated in Sections 5 and 12.

Exercises
In each of the following, eliminate the arbitrary constants.
§ 4] Families of curves 9

1. — 7>x^y = c.
x^ ANS. (x — 2y) dx — x dy = 0.
2. — xy^ = c.
y sinx ans. y(cos x — y) dx + (sin x — 2xy) dy = 0.
3. PV = C.

4.x^y = l+cx. ANS. (x^y + \) dx + x^ dy =0.


5. cy^ =x^ -'r y. ans. 2xy dx — (y + 2x^) dy = 0.
6. X ^A sin (cot + j8) : at a parameter, not to be eliminated.
d^x
ANS. -r-r + co^X = 0.
d'x
7. X = Ci cos oj/ + C2 sin co/; oj a parameter. ans. -tt + ^^-"^ = 0.
8. J'
= ca: + c^ + 1 ans. 7 = ^3^' + (y'y + 1.
/7 /i

9. y= mx + — a parameter, ; // //? to be eliminated. ans. y = xy' + —


m y
0. y^ = 4ox. ans. —y dx = 0.
2x dy
1. = «x^ + 6x + c.
:^' ANS. = 0. >'"'

2. j^= Ci + C2 e^''. ANS. j" — 2y' = 0.


3. ^^= 4 + Cie^"". ANS. y' — 2y=~%.
4. = Ci + C2 e ~ ^^
;^ ANS. /' + Zy' = 0.
5. y = Cie-^ + Cie'^". ANS. /' + 4/+3;' = 0.
6. J =x + Cie~'' + C2e~^*. ANS. /' + 4/4- 3^ = 4 + 3j:.
1. y=^Ci,e'' -V Cze'^". ANS. /' + /— 2^ = 0.
%. y=x^ + Cie' + C2e-^\ ans. /' + / - 2>' = 2(1 + - x^*). ;c

9. J'= Cie~'^ + C2Jce'~\ ANS. y" + 2y' + y = 0.


20. J = ^e^^ + Bxe^\ ANS. /' — 4/ + 4;^ = 0.
21. J = Cif ^^ cos 3x + C2 e^" sin 3x. ans. /' —Ay'+ \7>y = 0.
22. 7 = Cie'"' cos 6a" + Ci e"" sin ftx; o and b are parameters.
ans. - 2ay' {a^ + b^)y = Q.
y" -\-

23. y = CiX + €26'". ans.(x + l)y" + xy' ~ y = 0.


24. y = x^ + CiX + Cie'". ans. (x + \)y" + xy' — y = x^ + 2x + 2.
25. J = cix^ + C2 e^^. ANS. x(l - x)>'" + (2a;^ - 1)^' - 2(2a; - l)y - 0.

4. Families of curves

A relation involving a parameter, as well as one or both of the coordinates


of a point in a plane, represents a family of curves, one curve corresponding
to each value of the parameter. For instance, the equation

(x - cf +iy- cy = 2c\ (1)

or

x^+y^- 2c(x + y) = 0, (2)


10 Definitions, Elimination of Arbitrary Constants [Ch. 1

FIGURE 1

may be interpreted as the equation of a family of circles, each having its

center on the line y = x and each passing through the origin. Figure 1 shows
several elements, or members, of this family.

If the constant c in equation (1) or in equation (2) is treated as an arbitrary

constant and eliminated as in the preceding section, the result is called the

differential equation of the family represented by equation (1). In this example,

the elimination of c is easily performed by isolating c, then differentiating


throughout the equation with respect to x. Thus, from

= lc
X +y
we find that

(x + >>)(2x dx + ly dy') - (x^ + y'^){dx + dy)


(X + yf
Therefore
(x^ + 2x>^ - y^) dx - (x^ - 2x>' - v^) dy = (3)

is the differential equation of the family of circles represented by equation (1).


Note that equation (3) associates with each point (x, y) in the plane a
definite slope
dy + 2xy - y'^

(4)
dx — Ixy — y^
§4] Families of curves 11

except where the denominator on the right in (4) vanishes. An examination


of the method that was used to get (4) from (1) shows that at any point (x, y),

equation (4) gives the slope of that curve of the family ( 1


) which passes through
the point in question.
A simple and convenient check is now revealed. When the denominator on
the right of equation (4) vanishes, the curve passing through that point must
have a vertical tangent. From
x^ -2xy- y^ ^0
we see that

J = (72 - \)x (5)

or

y=-{y/2 + \)x. (6)

y = (-l-v/2)x

y = {-l + s^)x

FIGURE 2

In Figure 2, the straight lines (5) and (6) appear along with the family (1).

It is seen that the lines (5) and (6) cut the members of the family of circles in
precisely those points of vertical tangency.
For a two-parameter family of curves, the differential equation will be of
order two, and such a simple geometric interpretation is not available.

EXAMPLE (a) : Find the differential equation of the family of parabolas


(Figure 3), having their vertices at the origin and their foci on the j-axis.
12 Definitions, Elimination of Arbitrary Constants [Ch. 1

FIGURE 3

From analytic geometry, we find the equation of this family of parabolas


to be

x^ = Aay. (7)

Then from

= 4a

the a may be eliminated by differentiation. Thus it follows that

2xy dx - x^ ciy = 0. (8)

We may write the differential equation of the family (7) as

2y dx - X dy = 0, (9)

because x = is still a solution of (9), and nothing is lost in removing the


factor X from the left member of (8).

EXAMPLE (b): Find the differential equation of the family of circles


(Figure 4) having their centers on the j-axis.
§4] Families of curves 13

Since a member of the family of circles of this example may have its
center anywhere on the j-axis and its radius of any magnitude, we are
dealing with the two-parameter family

x^ + (y- bf = /'
(10)

We shall eliminate both b and r and arrive, of course, at a second-order


differential equation for the family (10).

At once

x + (y- b)y' = 0,
from which

x + yy'
= b.
y
Then

y'[.l + yy" + (y'f] - y"(x + yy')


= 0,

so the desired differential equation is

FIGURE 4
14 Definitions, Elimination of Arbitrary Constants [Ch. 1

Exercises
In each exercise, obtain the differential equation of the family of plane curves
described and sketch several representative members of the family.

1. Straight lines through the origin. ans. y dx — x dy =^ 0.

2. Straight lines through the fixed point (h, k). The h and k are not to be eliminated.
ANS. {y — k)dx — {x — h) dy = 0.
3. Straight lines with slope and >'-intercept equal.
ANS. y dx —{x + 1) dy = 0.
4. Straight lines with slope and x-intercept equal. ans. {y'Y = xy' —y.
5. Straight lines with algebraic sum of the intercepts fixed as k.
ANS. {xy' — y){y' — \) + ky' = 0.
6. Straight lines at a fixed distance p from the origin.
ANS. {xy' - yY = p^{\ + {yyi
7. Circles with center at the origin. ans. xdx + y dy = Q.
8. Circles with center on the x-axis. ANS. yy" + {y')'^ +1=0.
9. Circles with fixed radius r and tangent to the x-axis.
ANS. {y±ry{y'y + y^±2ry = 0.
10. Circles tangent to the x-axis. ans. [1 + {yVY = \yy" + + {y'YY.
1

11. Circles with center on the line y = —x, and passing through the origin.
ANS. {x^ — Ixy — y^) dx + {x^ + Ixy — y^) dy = 0.
12. Circles of radius unity. Use the fact that the radius of curvature is one.
ANS. {yy = u+{y'rr.
13. All circles. Use the curvature. ans. y"'[l + {y'Y] = 3y'{y")^.
14. Parabolas with vertex on the x-axis, with axis parallel to the j-axis, and with
distance from focus to vertex fixed as a. ans. a{y'y = y.
15. Parabolas with vertex on the j-axis, with axis parallel to the x-axis, and with
distance from focus to vertex fixed as a. ans. x{y')^ = a.
16. Parabolas with axis parallel to the j-axis and with distance from vertex to focus
fixed as a. ans. lay" — 1

17. Parabolas with axis parallel to the x-axis and with distance from vertex to focus
fixed as a. ans. lay" + {y'Y = 0.
d^x
18. Work Ex. 17, using differentiation with respect to y. ans. la —^ = 1.
dy^
19. Use the fact that

d^x_ d_ ldx\ dx d /dx\ dx d_ (dy\-' _ -y"


'di^^Ty\dy)^ly'dx\dy}~dy'dx\dx) ^{pY
to prove that the answers to Exs. 17 and 18 are equivalent.
20. Parabolas with vertex and focus on the x-axis. ans. yy" + {y')^ = 0.
21. Parabolas with axis parallel to the x-axis. ans. y'y" — 2>{y"y = 0.
22. Central conies with center at the origin and vertices on the coordinate axes.
ANS. xyy" + x{y'y — yy' = 0.
23. The confocal central conies

x^ y^
a^ + X'^ b^ + X^^
§ 4] Families of curves 15

with a and b held fixed. ans. {xy' — y){yy' + x) ^ (a^ — b^)y'-


24. The cubics cy^ = x^(x — a) with a held fixed.
ANS. 2x(x — a)y' = y{2)X — la).
25. The cubics of Ex. 24 with c held fixed and a to be eliminated.
ANS. 2cy{xy' —y)= x^.
26. The quartics c^y^ = x(x — aY with a held fixed.
ANS. 2x{x — a)y' = y{Ax — a).
27. The quartics of Ex. 26 with c held fixed and a to be eliminated.
ANS. c^ilxy' — yy = llx'^y.
= x^(a + x)
28. The strophoids y^ . ans. (x"^ - 4x^>'^ - j^*) c^x + 4x^y dy = 0.
a —a:
x^
29. The cissoids 3'^ = . ans. Ix^y' = y{y^ + 7>x^).
30. The trisectrices of Maclaurin y^{a + x) ^ x^{2>a — x).
ANS. (3a:* - ex^y"" - y'') dx + ^x^y dy = 0.

31. Circles through the intersections of the circle x^ y^ = -\-and the line y = x.
\

Use the "« + ^z;" form; that is, the equation

x^' + y^ -\+k{y-x) = 0.
ANS. (x^ -2xy-y^+\)dx + {x^ + 2a:j ->'2 -\)dy== 0.
32. Circles through the fixed points {a, 0) and {—a, 0). Use the method of Ex. 31.
ANS. 2xy dx + {y^ + a^ — x^) dy = 0.
33. The circles r == 2o(sin 9 — cos 9).

ANS. (cos 9 — sin ^) dr + Kcos ^ + sin 6) d9 = 0.


34. The cardioids r = a(l — sin 9). ans. (1 — sin 9) dr + r cos 9 d9 = 0.
35. The cissoids r = a sin ^ tan ^. (See Ex. 29.)
ANS. sin 9 cos ^ ^r — r(l + cos^ 9) d9 = 0.
dr
36. The strophoids r = a(sec 9 + tan 9). ans. —= a- sec ^.

37. The trisectrices of Maclaurin r = a(4 cos 9 — sec 9). (See Ex. 30.)
ANS. cos 9(4 cos^ 9 -~l)dr + r sin ^(4 cos^ ^+1)^/^ = 0.
CHAPTER ^
Equations of Order One

5. An existence theorem for equations of order one

In this chapter we shall study several elementary methods for solving first-

order differential equations. To make what we do more plausible, we now


state an important theorem which will be discussed in more detail in Chapter
15.

Consider the equation of order one

^=f(x,y). (1)
ax

Let T denote the rectangular region defined by

\x — Xq\ <a and \y — >'ol ^ b,

a region with the point {xq , jfo) at its center. Suppose that / and df/dy are
continuous functions of x and y in T.

16
§ 6] Separation of variables 17

Under the conditions imposed on f{x, y) above, there exists an interval


about Xq, \x — Xo\ ^ h, and a function y{x) which has the properties:

(a) y = y(x) is a solution of equation (1) on the interval \x — Xq\ ^ h;


(b) On the interval \x — Xq\ ^ h, y{x) satisfies the inequality \y{x) — yo\ ^b;
(c) Atx = Xo,y = y{xo)=yo;
(d) y(x) is unique on the interval — Xq\ ^ h in the sense that it is the
|.v

only function that has all of the properties (a), (b), and (c).

The interval \x — Xq\ ^h may or may not need to be smaller than the inter-
val \x — Xq\ ^a over which conditions were imposed upon/(jic, y).
In rough language, the theorem states that iff(x, y) is sufficiently well be-

haved near the point (xq yo) then the ,


differential equation

g =/(...) (1)

has a solution that passes through the point (xq , >'o) ^^^ that solution is

unique near (xq , jo)-

6. Separation of variables

We begin our study of the methods for solving first-order equations by


studying an equation of the form

Mdx + Ndy = 0,
where M and N may be functions of both x and y. Some equations of this
type are so simple that they can be put in the form

A{x)dx + B(y)dy = 0; (1)

that is, the variables can be separated. Then a solution can be written at once.
For it is only a matter of finding a function F whose total differential is the

left member of (1). Then F = c, where c is an arbitrary constant, is the desired


result.

EXAMPLE (a) : Solve the equation

dy 2y
-r = —, for x>0 and v > 0. (2)
dx X
We note that for the function in equation (2), the theorem of the previous
section applies and assures the existence of a unique continuous solution
through any point in the first quadrant. By separating the variables we can
write
18 Equations of Order One [Ch. 2

dy 2 dx

y X
Hence we obtain a family of solutions

In Ij'I
= 2ln\x\ + c (3)

or, because we are in the first quadrant,

y = e'^x^. (4)

If we now put Cj = e^ we can write

y = Tix , Cy >0. (5)

EXAMPLE (b): Solve the equation of the previous example for x # 0.

The argument now must be taken in two parts. First, if j ?£ 0, we can


proceed as before to equation (3). However, equation (5) must be written
|>'| = ClX^ Ci > 0. (6)

Second, if j = 0, we see immediately that since jc # 0, >> = is a solution


of the differential equation (2).

As a matter of convenience the solutions given by equation (6) are usually

written
y = c2x\ (7)

where Cj is taken to be an arbitrary real number. Indeed this form for the
solutions incorporates the special case y = 0. Thus we say that the family of
curves given in Figure 5 represents a family of solution curves for the differ-
ential equation (2).

FIGURE 5
§ 6] Separation of variables 19

We must be cautious, however, The function defined by

g{x) = x^, jc>


= -Ax^, x<0,
obtained by piecing together two different parabolic arcs could also be con-
sidered a solution of the differential equation even though this function is not
included in the family of equation (7). The uniqueness statement in the
theorem of Section 5, indicates that as long as we restrict our attention to a
point {xq , >'o) with Xq ?^ and consider a rectangle with center at (jcq , >'o)

containing no points at which x = 0, then in that rectangle there is a unique


solution that passes through {xq , Jq) and is continuous in the rectangle.

EXAMPLE (c) : Solve the equation

(l+y^)dx + i\+x^)dy = 0, (8)

with the " initial condition " that when x = 0, y = — 1.


From the differential equation we get

dx dy
1+x^ \^-y^

from which it follows at once that

Arctan x + Arctan y = c. (9)

In the set of solutions (9), each "Arctan " stands for the principal value of
the inverse tangent and is subject to the restriction

— \n< Arctan x < \ii.

The initial condition that j^ = — 1 when a: = permits us to determine the


value of c that must be used to obtain the particular solution desired here.
Since Arctan = and Arctan (— 1) = —\n, the solution of the initial value
problem is

Arctan x + Arctan y = —\n. (10)

Suppose next that we wish to sketch the graph of (10). Resorting to a device
of trigonometry, we take the tangent of each side of (10). Since

tan (Arctan x) =x
and
tan A -\- tanB
tan {A-\-B) =
1 — tan A tan B
20 Equations of Order One [Ch. 2

we are led to the equation

X +y
= -1,
I — xy
or
xy-x-y-\=0. (11)

Now (11) is the equation of an equilateral hyperbola with asymptotes

X= 1 and y = \. But if we turn to (10), we see from


Arctan x = —in — Arctan y
that, since ( — Arctan y) < \n,

Arctan x < ^n.

Hence x < 1, and equation (10) represents only one branch of the hyperbola
(1 1). In Figure 6, the solid curve is the graph of equation (10); the solid curve

and the dotted curve together are the graph of equation (11).

Each branch of the hyperbola (11) represents a solution of the differential


equation, one branch for jc < 1, the other for x > 1. In this problem we were
forced onto the left branch, equation (10), by the initial condition that

^ = — 1 when X = 0.

One distinction between the equations (10) and (11) can be seen by noting
that a computing machine, given the differential equation (8) and seeking a

FIGURE 6
§ 7] The exp u notation 21

solution that passes through the point (0, —1), would draw only the left

branch of the curve in Figure 6. The barrier (asymptote) at x = 1 would


prevent the machine from learning of the existence of the other branch of the
hyperbola (11).

EXAMPLE (d) : Solve the initial value problem

2x{y+l)cix- ydy = 0, (12)

where x= and y= —2.


Separating the variables in equation (12), we obtain

2x dx
-('-JTl)'"-
y
'*-'
Integrating, we get a family of solutions given implicitly by

x^=y-\n\y+\\+c. (13)

Since we seek a member of this family that passes through the point
(0, — 2), we must have

0= -2-ln|-l| + c,
or
c = 2.

Thus the solution to the problem is given implicitly by

x^ =y-\n\y+\\+2.
The reader should note how the theorem of Section 5 applies to this prob-
lem to indicate that we have found implicitly the unique solution to the
initial value problem which is continuous for 7 < — 1.

7. The exp u notation

The function e" enters our work frequently, and sometimes the exponent u
is complicated. Even a simple example such as e^ ^^
may be undesirable in
print, because exponents are usually set in small type. It is customary in ad-
vanced mathematics to use the notation

exp u — e". (1)

That is, exp is used to denote the exponential function in much the same way
that sin is used to denote the sine function.

EXAMPLE (a) : For e^^^"' we write exp {x^^y^).


22 Equations of Order One [Ch. 2

EXAMPLE (b) : The meaning of the following relation should be clear

(y+l) exp (x^) = C3 exp (y). (2)

Most people retain the notation e" when u simple, even though in the same
equation the exp notation may have been used elsewhere. Equation (2) is

sometimes written in the form

iy+ l)exp(jc') = C3e^ (3)

A few minutes of practice should make the reader feel at home with the

symbol exp u. Some familiarity with this notation will result from verification

of simple identities such as the following:

exp (x) •
exp (y) = exp {x + y), (4)

[exp (x)]*' = exp (/ex), (5)

exp (In y) = y, (6)

expg)expg)=exp(^), (7)

exp (3 In x) = [exp (In x^)] = x^ (8)

exp(x-21nx) = x-V. (9)

We shall use the exp notation whenever it seems to add to the clarity of

the printed result.

Exercises
In Exs. 1-26, obtain a family of solutions and indicate for what values of x they
are valid.

1. (4 + x)y' = y\ ANS. 2y^ In + x)| = -1.


|c(4

2. exp (y^) dx + x^y dy = 0. ANS. xexpi-y^) + 2 = ex.


3. cos X cos y dx + sin x sin ydy = 0. ans. sin x = c cos y.

4. 3y dx = 2x dy. ans. x^ = cy^.


5. my dx = nx dy. ans. x" - cy".
6. y' = xy^. ANS. y(x^ + c) + 2 = 0.

7. —
dP
=
P
. ANS. PV = C.
8. e%y - + 2(e^ + A) dy = 0.
1) Jx ANS. {y - + 4) = c^
\Y{e'
9. dr = e(r sin- cos 6 dr).
d dd ANS. + e cos 6) = c.
/•(!

10. (xy -x)dx + (xy + y) dy = 0. ans. (y - 1) exp ix + y) = c(x + 1).


11. (y+\)dx^ Ixy dy. ans. e^^ = cx{y + 1)^
12. x^ dx + y{x -\)dy^ 0.
13. (x>' + x) Jx = {x^y"" + x^ + >'^ + 1) dy.

ANS. In (x^ + l) =7^ -27 + 4^10(7+1)1.


14. X cos^ 7 c/x + tan y dy = ^. ans. x^ + tan^ y = c^.
§ 8] Homogeneous functions 23

15. x^yy' = e\ ans. jc(>; + 1) = (1 + cx)e\


16. tan^ y dy^ sin' x dx. ans. cos' — 3 cos x = 3(tan y — y + c).
x
17. / = cos^ cos a: >'. ans. 4 In |sec;' + tan j| = 2x + sin 2a: + c.
18. y ^y sec x. ans. = c(sec x + tan jc).
>'

19. dx = t{\ + t^) sec^ a: c//. ans. 2x + sin 2x = c + (1 + t^y.


20. (e^^ + 4)7' = 3^. ANS. y\l + 4e " ^') = c^
21. ocd^ + ^d(X + aj3(3 Ja + = 0. f//3) ANS. caj8 = exp (-3a - j3).
22. (1 + lnx)</A: + (1 + ln>')d> = 0. ANS. xlnx + y In y = c.
23. X dx-Va^ -x^ dy - 0.
ANS. J — c = — Va^ — x^, the lower half of the circle x^ + (y — cY = a^.

24. JC^x + Va^-x^ dy = 0.


ANS. — c = Va^ — x^, the upper half of the circle x^ + (y ~ cY = a}.
J'

= a sec y -V
/ c
25. a^ dx =XV JC ^ — a^ dy. ans. jc .

a
16. y\nx\r\y dx + dy =^0. ans. jcln jc + In |in>'| =x + c.
In Exs. 27-33, obtain the particular solution satisfying the initial condition
indicated.

dr
11. —= —Art; when / = 0, r = ro ans. r = ro exp (—2/^).
28. 2x7/ = + >'^ when x = 2, 7 = 3.
1 ; ans. 7^ = 5x — 1.
29. xyy' = + 7^ when x = 2, 7 = 3.
1 ; ans. 5x^ — 27^ = 2.
30. / =xexp(jv — x^); when X = 0, 7 = 0. ans. 2e"* = + exp (— x^).
1

31. xy^ t/x + e'f/y = 0; when x^ 00,7^^. ans. 7 = eV(2e* — x — 1).


32. (20^ - r^) dr = r' sin ^ dd; when ^ = 0, r = a.

33. V — ^g; when x = Xo , u = Uo ans. u^ — Uq^ = 2^(x — Xo).

8. Homogeneous functions

Polynomials in which all terms are of the same degree, such as

x^ — 3x7 + 4j^,

x^ + y\ (1)

x'^y + ly^,
are called homogeneous polynomials. We wish now to extend that concept of
homogeneity so it will apply to functions other than polynomials.
If we assign a physical dimension, say length, to each variable x and y in
the polynomials in (1), then each polynomial itself also has a physical dimen-
sion, length to some power. This suggests the desired generalization, If, when
certain variables are thought of as lengths, a function has physical dimension
length to the /:th power, then we shall call that function homogeneous of
24 Equations of Order One [Ch. 2

degree k in those variables. For example, the function

/(.,,) = 2/ expg)-^, (2)

is of dimension (length)^ when x and y are lengths. Therefore that function


is said to be homogeneous of degree 3 in x and y.

We permit the degree k to be any number. The function \l x+ Ay'vs, called

homogeneous of degree ^'\n x and y. The function

Vx^+/
is homogeneous of degree zero in x and y.

A formal definition of homogeneity is: The function f{x, y) is said to be


homogeneous of degree k in x and y if, and only if

f{Xx,Xy) = l'f{x,y). (3)

The definition is easily extended to functions of more than two variables.


For the function f{x, y) of equation (2), the formal definition of homo-
geneity leads us to consider

f{Xx, ky) = Ik^y^ exp '


"^

vAx/ Xx + 3A_y

But we see at once that

f{Xx,Xy) = X^f{x,y);

hence /(x, y) is homogeneous of degree 3 in x and y, as stated previously.

The following theorems prove useful in the next section.

Theorem 1 : If M(x, y) and N(x, y) are both homogeneous and of the same
degree, the function M(x, y)lN(x, y) is homogeneous of degree zero.

Theorem 2 : Iff{x, y) is homogeneous of degree zero in x and y, f{x, y) is a


function ofy/x alone.

Proof of Theorem 1 is left to the student.


Proof of Theorem 2: Let us put y = vx. Then Theorem 2 states that if

f{x, y) is homogeneous of degree zero, fix, y) is a function of v alone. Now


fix, y) =fix, vx) = xV(l, V) =/(l, V), (4)

in which the x is now playing the role taken by X in the definition (3) above.
By (4), fix, y) depends on v alone as stated in Theorem 2.
§ 9] Equations with homogeneous coefficients 25

Oral Exercises
Determine in each exercise whether the function is homogeneous or not. If it is

homogeneous, state the degree of the function.

1. 4x^ — 3xy + y^. 2. x^ — xy + y^.


3. 2y + Vx^ + y\ 4.
5. e\

7. exp
&
26 Equations of Order One [Ch. 2

EXAMPLE (a) : Solve the equation

{x^ -xy^- y^) dx - xy dy = 0. (4)

Since the coefficients in (4) are both homogeneous and of degree two in

X and y, let us put y = vx. Then (4) becomes

(x^ - x^v + x^v^) dx - x^viv dx + X dv) = 0,

from which the factor x^ should be removed at once. That done, we have to
solve
(1 - V -{ v^) dx — v{v dx + X dv) = 0,

or
(1 — v) dx — XV dv = 0.

Hence we separate variables to get


dx
X
+
V
V dv

-
I
= 0.

Then from
dx
—X
+ 1 +
V-
1

1
dv =

a family of solutions is seen to be

ln|Ai + r + ln|t7- 1| = ln|f|,

or
X{V -])€" = C.

In terms of the original variables, these solutions are given by

- '^p '•
'(x ') (9
or

(y - x) exp f-j = c.

EXAMPLE (b): Solve the equation

xy dx + {x^ + y^) dy = 0. (5)

Again the coefficients in the equation are homogeneous and of degree two.
We could use y — vx, but the relative simplicity of the dx term in (5) suggests

that we put
X= vy.

Then dx = v dy + y dv, and equation (5) is replaced by

vy\v dy + y dv) + (v^y^ + y^) dy = 0,


§ 9] Equations with homogeneous coefficients 27

or

v(v dy + y dv) + (v^ + I) dy ^ 0.

Hence we need to solve

vy dv + {2v^ + \) dy = 0, (6)

which leads at once to

ln(2y2+ l) + 41n|>'| = lnc,

or

y\2v^ + l) = c.

Thus the desired solutions are

(7-)-^
y
that is,

y'(2x' + y') = c. (7)

Since the left member of equation (7) cannot be negative, we may, for sym-
metry's sake, change the arbitrary constant to c^*, writing

It is worth-while for the student to attack equation (5) using y = vx. That
method leads directly to the equation

(v^ + 2v) dx -f- x{v^ + \)dv = 0.

Frequently in equations with homogeneous coefficients, it is quite im-


material whether one uses y = vx ov x = vy. However, it is sometimes easier
to substitute for the variable whose differential has the simpler coefficient.

Exercises
In Exs. 1-19 obtain a family of solutions.

1. (x 2y) + (2x + y) dy = 0,
dx ANS. In (jc^ + >'^) + 4 Arctan {yjx) = c.
2. 2{2x^ dx - xy dy = 0.
-V y"") ans. x'^ = c''{Ax'' + y^).
3. xy dx — {x^ + 3>'^) dy = 0. ans. x^ = 6y^ In \ylc\.
4. x^y' = 4x^ + Ixy + 2y^. ans. x^(y + 2x) = c^y + x).
5. 3xy dx + (x^ + y^) dy = 0.
6. {x — y)(4x -^y)dx + xi5x —y)dy^ 0. ans. x(y + xY =c(y — 2x).
28 Equations of Order One [Ch. 2

7. (Sv -u)du + i3v - 7«) dv = 0. ans. (,3v + w)' = c(y - «).


8. {x^ + Ix;' - 4y^) dx - (x^ - ^xy - 4j^) a> = 0.
ANS. + 4^^ = C{X +
X^ J').

9. {x^ + >'^) ^x - ciy = 0. ATjK ANS. y^ = 2x^ In \x/c\


10. dx + xiv - 4x) dv = 0.
y^^ ans. xu^ = c(v - 2x).

11. {2x + yydx = xy dy. ans. x\x + y) = c exp 0/x).


12. ydx = ix + Vy^ - x^) dy. ans. Arcsin (x/y) = In \y/c\.
13. (3x^ - 2x:y + 3y^) dx - 4xy dy. ans. (j' - x)0' + 3xy = cx\
14. [x CSC (7/x) - j] f/x + X Jy = 0. ans. In |x/c| = cos (j/x).

15. X dx + sin 2 {ylx)\y dx -x dy] = 0.


ANS. 4x In \x/c\ — 2^ + X sin (2>'/x) = 0.
16. {x-y\ny + y In x) dx + x(ln y -\nx)dy = 0.
ANS. (x — >') In X + y In = ex +
^^ >-.

17. [x - >' Arctan (y/x)] dx + x Arctan (y/x) dy = 0.


ANS. 2y Arctan (y/x) = x In [c\x^ + y^)/x*].
18. Ki^^ + u^) du + u{v^ - u^) dv = 0.
19. Ck^
- 4x>'^ - 2x2) ^ ^ ^2(2_^ ^_ ^) j_y ^ 0.
ANS. x^iy - 2xY = c^(x^ + j'^).
20. Prove that with the aid of the substitution y = vx, you can solve any equation of

the form

y"fix) dx + H{x, y)(y dx - x dy) = 0,


where //(x, y) is homogeneous in x and y.

In Exs. 21-33 find the solution indicated.

21. (x->')^x + (3x + >')</>' = 0; whenx = 2,>'= -1.


+ 2>') + (x + >')ln|x + =0.
ANS. 2(x >'|

22. (y -Vx^ + y^) dx-xdy = 0; when x = V3, y = l. ans. x^ = 9 - 6y.

23. O + Vx^ + y^) dx -xdy = 0; when x = V3, = 1. ans. x^ = 2>'+1.


>'

24. [x cos^ (>'/x) - c^x + X d>


J']
= 0; when x = = 77/4. 1 ,
>'

ans. tan (y/x) = In (e/x).


25. Cv^ + 7x>' + 16x^) dx + x^ dy=^ 0; when x = 1, = 1. >'

ANS. x —y = 5(y + 4x) In x,


26. y^ (/x + (x^ + 3xj' + 4y^) dy = 0; when x = 2, = 1. ^^

ANS. 4(2^ + x) In J = 2^ — X.
27. xy dx + 2(x^ + 2y^) dy = 0; when x = 0, = ans. >'*(3x^ + 4y^) = 4.
3; 1

28. >'(2x^ -xy + y^) dx - x\2x -y)dy = 0; when x = l,y =h


ANS. j'^ In X = 2y^ + x>' — x^.
29. >'(9x — 2y) dx — x(6x — y) dy = 0; when x = 1, = 1. j'

ANS. 3x2 _ ^2y __ 2y2 ^ 0.


30. :v(x^ + y^) dx + x(3x^ - 5y^) dy = 0; when x = 2,y = \.
ANS. 2y^ - 2x2>'2 + 3x = 0.
31. (16x + Sj') dx + (3x + c(v 0;= the
J')curve to pass through the point (1, —3).
ANS. y + 3x =^(y + 4x) In (y + 4x).
32. v(3x + 2v) dx - x^ dv = 0; when x = 1, y = 2. ans. 2x^ + 2x^v - 3y = 0.
33. (3x^ - 2y^)y' = 2xy; when x = 0,y= -L ans. x^ = 2y^(y + 1).
§ 10] Exact equations 29

34. From Theorems 1 and 2, page 24, it follows that if F is homogeneous of degree
km X and y, F can be written in the form
(A)
=-K9-
Use (A) to prove Euler's theorem that if F is a homogeneous function of degree
kinx and y,
dF 8F

10. Exact equations

In Section 6 it was noted that when an equation can be put in the form

A(x) dx + B{y) dy = 0,

a set of solutions can be determined by integration; that is, by finding a


function whose differential is A{x) dx + B{y) dy.
That idea can be extended to some equations of the form

M{x, y) dx + N{x, y)dy =Q (1)

in which separation of variables may not be possible. Suppose that a


function F{x, y) can be found that has for its total differential the expression

M dx + N dy; that is,

dF=Mdx + Ndy. (2)

Then certainly

F{x,y) = c (3)

is a set of solutions of (1). For, from (3) it follows that

dF=0,
or, in view of (2),

Mdx + Ndy = 0,
as desired.
Two things, then, are needed: first, to find out under what conditions on
M and A'^ a function F exists such that its total differential is exactly M dx +
N dy; second, if those conditions are satisfied, actually to determine the func-
tion F. If there exists a function F such that

Mdx + Ndy
is exactly the total differential of F, we call equation (1) an exact equation.
30 Equations of Order One [Ch. 2

If the equation

Mdx + Ndy^Q (1)

is exact, then by definition F exists such that

dF= Mdx + N dy.


But, from calculus.

aF df
dF = ^
ox
dx + ^
,

dy
,
dy,

so

_8F _8F
dx' dy

These two equations lead to

dM d^F
dy dy dx

and

dN _ d^F
dx dx dy

Again from calculus

d^F d^F
dy dx dx dy

provided these partial derivatives are continuous. Therefore, if (1) is an exact


equation, then

— =—
dM dN
dy dx
• (4)

Thus, for (1) to be exact it is necessary that (4) be satisfied.


Let us now show that if condition (4) is satisfied, then (1) is an exact equa-
tion. Let 0(a:, y) be a function for which

d^_
dx

The function is the result of integrating M dx with respect to x while holding


y constant. Now
d^cf) _ dM ^

dy dx dy
§ 10] Exact equations 31

hence, if (4) is satisfied, then also

-^ =— (5)
dx dy dx

Let us integrate both sides of this last equation with respect to x, holding y
fixed. In the integration with respect to the x, the "arbitrary constant" may

be any function of y. Let us call it B'(y), for ease in indicating its integral.

Then integration of (5) with respect to x yields

dS
-f
= N + B'(y). 6)
8y

Now a function Fcan be exhibited, namely,

F=(l>(x,y)-B(y),

for which

dd) d(j)
dF = ^dx + ^dy- B'(y) dy
dx dy

= Mdx + lN + B'(y)'] dy - B'(y) dy

= M dx + N dy.
Hence, equation (1) is exact. We have completed a proof of the theorem
stated below.

Theorem 3 : If M, N, dM/8y, and dN/dx are continuous functions ofx and y,


then a necessary and sufficient condition that

Mdx + Ndy = (1)

be an exact equation is that

dM _ dN
dy dx

Furthermore, the proof contains the germ of a method for obtaining a set

of solutions, a method used in Examples (a) and (b) below.

EXAMPLE (a): Solve the equation

3x{xy -2)dx + (x^ + 2y) dy = 0. (7)

First, from the fact that

dM
— ^^ =
—= ^ 2
Sx-^ and^ -— . 2
Sx"^,
dy dx
32 Equations of Order One [Ch. 2

we conclude that equation (7) is exact. Therefore, its solution is F= c, where

dF

ex
= M = 3x^y - 6x, (8)

and

8F

dy
= N = x' + 2y. (9)

Let us attempt to determine Ffrom equation (8). Integration of both sides


of (8) with respect to x, holding y constant, yields

F=x^y-3x^ + T(y), (10)

where the usual arbitrary constant in indefinite integration is now necessarily


a function T(y), as yet unknown. To determine T(y), we use the fact that the
function i^ of equation (10) must also satisfy equation (9). Hence

x^ + T'(y) = x^ + ly,

T'iy) = 2y.

No arbitrary constant is needed in obtaining T{y), since one is being intro-


duced on the right in the solution F= c. Then

ny) = y\
and from (10)

F= x^y — 3.v^ + y^-


Finally, a set of solutions of equation (7) is defined by

x^y — 3jv:^ -\- y^ = c.

EXAMPLE (b) : Solve the equation

{2x^ - xy^ -2y + 3)dx- {x^y + 2.v) dy^{). (11)

Here

dM dN

SO equation (11) is exact.


A set of solutions of ( 1 1 ) is F= c, where

dF
— = 2x^- xy^ -2y + 3 (12)
§ 10] Exact equations 33

and
dF
— =-x'y-2x. (13)

Because (13) is simpler than (12), and for variety's sake, let us start the
determination of F from equation (13).

At once, from ( 1
3)

F= -^xV-2xj;+ Q(x),

where Q(x) will be determined from (12). The latter yields

-xy^ -2y+ Q'(x) = 2x^ - xy^ - 2j^ + 3,

Q'ix) = 2x^ + 3.
Therefore
Qix) = ix^ + 3.V,

and the desired set of solutions of (1 1) is

-iX^y^ - 2xy + ix* + 3x = ^c,


or
X* — x^y^ — 4xy + 6x = c.

Exercises
Test each of the following equations for exactness and solve the equation.
The equations that are not exact may, of course, be solved by methods discussed
in the preceding sections.

1. (x -^y)dx + (x— y) dy ^ 0. ANS. + 2xy —y^=c.


x^
2. {6x + y^) dx + y{2x - 3^) dy = 0. ans. 3x^ + xy^ - y^ = c.
3. {2xy - 3x^) dx + (x^ + y) dy = 0. ans. x^y - x^ + ^y^ = c.
4. (y^ - 2xy + 6x) dx - (x^ - 2xy + 2) dy = 0.
ANS. xy^ — x^y + 3x^ —2y = c.
5. (Ixy + y)dx+ (x^ - x) dy = 0. y = cx{x - 1) " ^
ans.
6. (x - 2y) dx + 2{y - x) dy = 0. ans. x^ + 2y^ = 4xy + c.
7. Do Ex. 6 by another method.
8. i2x - 3y) dx + {2y - 3x) dy = 0. ans. x^ + y^ = 3xy + c.
9. Do Ex. 8 by another method.
10. v{2uv'- -3)du + Ou^v^ -3u + 4v) dv = 0.
ans. v{u^v^ — 3u + 2v) = c.
11. (cos - 3x^y^) dx + (cos 2y - 2x sin 2y - 2x^y) dy = 0.
2y
ANS. i sin 2y + X cos 2y — x^y^ = c.
12. (1 + y^) dx + (x^y + y) dy = 0. ans. 2 Arctan + In (1 + y^) = c. jc

13. (1 + ^^ + xy^) dx + (x^y + y + 2xy) dy = 0. ans. 2x + y\\ + xY = c.


14. {w^ + wz^ -z)dw + (z^ + w^z — w)dz = 0. ans. (w^ + z^y = 4wz + c.
15. i2xy — tan y) dx + (x^ — x sec^ y) dy = 0. ans. x^y — tan = c. a: j'

16. (cos X cos y — cot x) dx — sin x sin y dy = 0.


ans. sinxcosj' = ln |csinx|.
34 Equations of Order One [Ch. 2

17. (r + sin ^ - cos 6) dr + Ksin 9 + cos 6) dd = 0.


ANS. r^ + 2r(sin 6 — cos 6) = c.
18. xOxy - 4^^ + €)dx + (x^ - Gat^j'^ - 1) ^^^ = 0.
ANS. x^y — Ix^y^ + 3x^ — = c. j'

19. (sin e ^ 2r cos^ d) dr + r cos ^(2/- sin + 1) (/^ = 0.


ANS. — r^ cos^ ^ = c.
r sin

20. [2a: + y cos (xj')] ^jc + cos (xy) dy = 0.


Jr + sin {xy) = c.
ans. x^
21. 2x7 ^-^ + (>'^ + x^) dy = 0. ans. y(3x^ + y^) = c.
22. 2x>' dx + (y^ - x^) dy = 0. ans. x^ + y^ = cy.
23. (x^^ + y — x)dx + x(xy + I) dy ^ 0. ans. x^y^ + Ixy — x^ = c.
24. 3y(x^ -l)dx + (x^ + Sy- 3x) ^7 = 0; when x = 0, = 1.
>'

ans. xyix^ — 3) = 4(1 — y^).


25. (1 -x>')-^^x+b' + ^^(l - xy)- ^] dy = 0; when x= 2, y = l.

ANS. xy'^ —y^ + 5xy — 3x = 5.


26. (3 + +
>- 2y^ sin^ x) c/x + (x + 2x>' - y sin 2x) (/y = 0.
ANS. y^ sin 2x =c+ 2x(3 + y + y^).
27. 2x[3x + y~yexp ( -x^)] c/x + [x^ + 3>'' + exp -x^)] dy = 0.
(

ANS. x^y + y^ + 2x' + y exp ( — x^) = c.


28. (xy^ + X — 2y+ 3) dx + x^y dy = 2(x + y) dy; when x = 1 , j^ = 1

ANS. (x>;-2)^ + (x+3)^=2j^+ 15.

11. The linear equation of order one

In Section 10 we studied first-order differential equations that were exact.


If an equation is not exact, it is natural to attempt to make it exact by the
introduction of an appropriate factor, which is then called an integrating
factor. Indeed, in Section 6 we multiplied by an integrating factor to separate
the variables and thereby obtain an exact equation.
In general, very little can be said about the theory of integrating factors

for first-order equations. In Chapter 4, we shall prove some theorems that


will give some assistance in a few isolated situations. There is one important
class of equations, however where the existence of an integrating factor can
be demonstrated. This is the class of linear equations of order one.
An equation that is linear and of order one in the dependent variable y
must by definition (Section 2) be of the form

A(x)-^ + Bix)y = C(x). (1)


dx

By dividing each member of equation (1) by Aix), we obtain

^ + P(x)y = Q(x), (2)


§ 11] The linear equation of order one 35

which we choose as the standard form for the linear equation of order one.
For the moment, suppose that there exists for equation (2) a positive inte-
grating factor v(x) > 0, a function of x alone. Then

dy (3)
v(x) = v{x) •
Q{x)
dx

must be an exact equation. But (3) is easily put into the form

Mdx + Ndy =
with

M=vPy- vQ,
and

N=v,
in which v, P, and Q are functions of x alone.
Therefore, if equation (3) is to be exact, it follows from the requirement
36 Equations of Order One [Ch. 2

Let us apply the factor throughout (2), obtaining

exp (I Pdx)-^ + P exp (J P dx)y =Q exp (J


P dx). (6)

The left member of (6) is the derivative of the product

yexpi^Pdx);
the right member of (6) is a function of x only. Hence equation (6) is exact,

which is what we wanted to show.


Of course one integrating factor is sufficient. Hence we may use in the

exponent (
J P dx) any function whose derivative is P.

Because of the great importance of the ideas just discussed and the frequent
occurrence of linear equations of first order, we summarize the steps involved

in solving such equations

(a) Put the equation into standard form

dx

(b) Obtain the integrating factor exp (


J P dx);
(c) Apply the integrating factor to the equation in its standard form;
(d) Solve the resultant exact equation.

Note in integrating the exact equation, that the integral of the left member
is always the product of the dependent variable and the integrating factor used.

EXAMPLE (a) : Solve the equation

2(y - 4x^) dx + X dy = 0.

The equation is hnear in y. When put in standard form it becomes

-l + -y = Sx, when x ¥^ Q. (7)


dx X
Then an integrating factor is

exp I 1 = exp (2 In |x|) = exp (In x^) = x^.

Next apply the integrating factor to (7), thus obtaining the exact equation

x' ^
dx
+ 2xy = 8JC^ (8)

which may be immediately rewritten as

^ix'y) = Sx\ (9)


dx
§ 11] The linear equation of order one 37

By integrating (9) we find that

x^y = Ix" + c. (10)

This can be checked. From (10) we get (8) by difTerentiation. Then the
original differential equation follows from (8) by a simple adjustment. Hence
(10) defines a set of solutions of the original equation.

EXAMPLE (b) : Solve the equation

y dx -V (3a- — xy -\-T)dy = 0.

Since the product y dy occurs here, the equation is not linear in y. It is,

however, linear in x. Therefore we arrange the terms as in

y dx-\- {^ — y)x dy = —2dy

and pass to the standard form,


dx
dy
/3
+ --1
\y
\

/
x = —
-2
y
, for y¥^0. (11)

Now
j\--ljdy = 3\n\y\-y + c„

so that an integrating factor for equation (11) is

exp (3 In \y\ - y) = exp (3 In \y\) •


e'^
= Qxp{\n\y\')-e-y
= \y\'e-\

It follows that for j^ > 0, y^e~^ is an integrating factor for equation (11) and
for j^ < 0, —y^e~^ serves as an integrating factor. In either case we are led to
the exact equation

y^e-y dx + /(3 - y)e-''x dy = -ly^e'^ dy

from which we get

xy^e-y= -l^y^-'dy
= 2y^e-' + Aye'" + Ae'" + c.

Thus a family of solutions is defined implicitly by

xy^ = 2y^ + 4y + 4 + ce^.


38 Equations of Order One [Ch. 2

12. The general solution of a linear equation

At the beginning of this chapter we stated an existence and uniqueness


theorem for first-order differential equations. If the differential equation in

that theorem happens to be a linear equation we can prove a somewhat


stronger statement.
Consider the linear differential equation

^
ax
+ P(x)y = Q{x). (1)

Suppose that P and Q are continuous functions on the interval a <x< b,

and that x = Xq is any number in that interval. If Jq is an arbitrary real

number, there exists a unique solution y = y{x) of the differential equation (1)

that also satisfies the initial condition

>'(^o) = J'o •

Moreover, this solution satisfies equation (1) throughout the entire interval
a <X< b.

The proof of this theorem has essentially been obtained in Section 11.

Multiplication of equation (1) by the integrating factor v = exp {j P dx) and


integration gives

yv = jvQdx + c.

Since y # 0, we can write

y = v~^ jvQdx + cv'K (2)

It is a simple matter to show that since r # and r is continuous on


a <X< b, (2) is a family of solutions of equation (1).

It is also easy to see that given any a'o on the interval a <x< b together
with any number jo we can choose
, the constant c so that j = Vq when x — Xq.
The effect of our argument is that every equation of the form of equation
(1), for which P and Q have some common interval of continuity, will have

a unique set of solutions containing one constant of integration, and which


can be obtained by introducing the appropriate integrating factor. Because
we are assured of the uniqueness of these solutions, we know that any other
solution obtained by any other method must be one of the functions in our
one parameter family of solutions. It is for this reason that this set of solutions

is called the general solution of equation (1). The word " general " is intended
to mean that we have found all possible solutions that satisfy the differential
equation on the interval a <x< b.
§ 12] The general solution of a linear equation 39

Exercises
In Exs. 1-25, find the general solution.

1. (x^ + dx — xdy = 0.
3y) = x^ + cx^.
ans. 2y
2. 2{2xy + 4y-3) dx + {x + ly dy = 0. ans. >' = 2(x ++ c(a: + 2)"*.
2)"'
3. y = x — 2y. ANS. 4>' = 2a: — + ce~^'. 1

4. (y +\)dx + (4x -y)dy = 0. ans. 20x = 4>' - + c{y + 1) "*. 1

5. udx + {I — 3u)x du = 3u^e^" du. ans. xu = («' + c)e^".


6. M f/x + (1 — 3«)x du = 3u du. ans. xu = ce^" — m — i.
7. y ^ X — Axy. Solve by two methods. ans. 4;^ = + c exp (— 2a:^).1

8. y =cscx-\-y cot X. ans. = c sin x — cos x.


9. y' = CSC x — 7 cot X. ans. sin = + c. j' jc a:

10. {2xy + x^ + x'^) - (1 + x^) dy = 0.


flTx

ANS. + x^)(c + jc — Arctan x).


>» = ( 1

11. {y — cos^ x) f/x + cos xdy = Q. ans. + tan x) = c + x — cos x.


>'(sec x
12. >''= X — 2^ cot 2x. ans. Ay sin 2x = c + sin 2x — 2x cos 2x.
13. {y — x-\- xy cot x) dx ^- x dy = 0. ans. x^ sin x — c + sin x — x cos x.

dy
14. my = Cie""", where Ci and m are constants, ans. y = (cix + C2)e""'.
dx

dy
15. — —m2y = Cie
m X
'
, where Ci, mi, ma are constants and rtii ^mz.
dx
Ci
ANS. y =^ C3 e""^" + Ca e"^*, where C3
— m2
mi

16. f (/x + +1 —vx)dv = 0.


(2x ANS. xi;^ = y + + ce". 1

17. x(x^ + + 2y = (x^ + 1)\


1);^' ans. x^y = i(x^ + 1)^ + dx^ + 1).
18. 2y(y^ — x) dy = dx. ans. x = j'^ — + c exp —y^). 1 (

19. (1 + xy) dx-(l+ x^) d^ = 0. ans. = x + c(1 + x^'^. j'

20. 2y dx = (x^ - l)(^x - dy). ans. (x - 1)>' = (x + l)[c + x - 2 hi [x + 1 1].

21. ^x — (1 + 2x tan y) dy = 0. ans. 2x cos^ y =^y + c + siny cos y.


22. (1 + cos x)y' == sin x(sin x + sin x cos x — y).

ans. = (1 + cos x)(c + X — sin x).j'

23. y' = + 3;^ tan x.


1 ans. 3y cos^ x = c + 3 sin x — sin^ x.
24. (x + a)y' = bx — ny; a,b, n are constants with n^O,n^ — 1

ans. «(« + 1);; = binx — a) + c(x + a) "".


25. Solve the equation of Ex. 24 for the exceptional cases « = and « = — 1.
ans. If n = 0, y ^ bx + c — ab \n \x + a\.
If n = —l,y = ab + c(x + a) + bix + a) In |x + a|
26. In the standard form dy Py dx = Q dx put y = vw, thus obtaining
-\-

w{dv + Pv dx) + V dw = Q dx.

Then, by first choosing v so that

dv + Pvdx =
40 Equations of Order One [Ch. 2

and later determining >v, show how to complete the solution of

dy + Pydx = Q dx.
In Exs. 27-33 find the particular solution indicated.

27. (2jc + 3)/=>'+(2x + 3)"2; when x = -1,7 = 0.


ANS. 2>' = (2x+3)''Mn(2A:+3).
28. y' = x^ — 2xy; when x = l,y = 1. ans. 2y =x^ — I +2 exp(1 — x^).

29. L — + Ri = E, where L, R, and E are constants; when = 0, = 0. / /


dt

ANS. / = — — exp — 1 (
i['-"''(-T!
30. L — + Ri = E sin cjot; when / = 0, = 0. /

dt

ANS. Let Z = /?^ + oj^L\


2 Then

i = EZ R sin a»r — a>L cos cor + coL exp


?)
31. Find that solution of y' = 2(2x —7) which passes through the point (0, —1).
ANS. y =2x— \.

32. Find that solution oi y' ^2{2x — y) which passes through the point (0, 1).
ANS. >' = 2x- +2e-2^ 1

33. (1 T- 1^) ds + 2t{st^ - 3(1 + t^y] dt==0; when = 0,s = 2.


t

ANS. = (l +.y r^)[3 -exp(-/^)].

Miscellaneous Exercises
In each exercise, find a set of solutions, unless the statement of the exercise
stipulates otherwise.

1. >•' = exp (2x — y). ans. = e^' + c.


26"

2. (x + y)dx + xdy = 0. ans. + 2y) = c.


A:(jr

3. j'^ dx - x(2x + 3y) dy = 0. ans. y\x + = ex. >-)

4. (x^ + 1) (/x + x^y^ dy = 0. ans. xy^ = 3(1 + ex - x^*).


5. (x^ + >'^) dx + y\'3x + ^>') dy =0; k is constant.
ans. A:^* + 4x7^ + X* = c.

6. >'' = x^ — 2x>'; when x = 1, y = 2. ans. 2y = x^ — +4 exp (1 — x^).


I

dy
7. cos X = cos X tan^ y. ans. 2 sin x = + sin
>' >' cos y + c.
dx

8. cos X —=
dy
1 — J — sin X. ans j(1 + sin x) = (x + c) cos x.
Jx
dr
9. sin ^ —= — — 1 2/- cos 6. ans. r sin^ = c + cos ^.
dtf

10. ^(x + 3>') </x + x^ dy = 0. ans. x^'^ = c(2x + 3y).


Miscellaneous Exercises 41

dy
11. = sec^ X sec' y. ANS. 3 tan X +c= 3 sin J'
— sin' y.
dx
12. yilx^ -x^y-\-y^)dx-x{2x^ +y^)dy=-0. ans. 2x^7 In |cx1 =4a:' ->''.
13. (1 + jc^)>'' = x'^y'^. ANS. x'j'' = y^{c + 3x - 3 Arctan x).
+ 1

14. yO-^2xy^)dx+3{x^y^^-x-\)dy=0. ans. x^y^ == ^{c + y - xy).


15. 0-x^ — 2xy - y^) dx + xy dy = 0. ANS. x^ == c(y — x) exp (>'/x).
16. jCx^ + y^) dx + x(3x2 - 5;^^) dy = 0; when x = 2, 7 = 1

ANS. 2y' - 2x^y^ + 3a: = 0.


17. y' -{- ay ^b;a and 6 are constants. Solve by two methods.
ANS. y ^b\a-\- ce-'"'.
18. {x —y^dx — {x^r y) dy = 0. Solve by two methods.
ANS. x'^ — 2xy —y^=c.
dx
19. — = cosA:cos^r.
dt
.
ANS.
^
4
.
In
,

secx+ tanx =2/ + sm 2/ + c.


20. (sin — sin x) dx + (cos x + x cos y) dy = 0.
^^ >' ans. xsiny y cos x = c. -\-

21. (1 + Axy — Ax^y) dx + (x^ — x') dy = 0; when x = 2, = i. >'

ANS. 2xV = x^ + 2x + 2 In (x - 1).


22. 3x'j' = 2y{y — 3). ans, y = c{y — 3) exp (x~ ^).
23. (2>' cos X + sin* x) dx = sin x dy; when x = \iT,y = \.

ANS. 7 = 2 sin X sin ix. ^ ^

24. xyidx — dy) =^ x^ dy + y^ dx. ans. x = In |cx>'|. j'

25. a^idy — dx) = x^ dy + y^ dx;a is constant.


ANS. 2 Arctan (y/d) = In |c(x + a)/(x — fl)|.

26. (y — sin^ x) ^x + sin x dy = 0. ans. >'(csc x — cot x) = x + c — sin x.


27. (x — y) dx + (3x + y) dy = 0; when x = 2, =— j' 1

ANS. 2(x + 2y) + (x + In (x + y) = 0. >')

(1,-?)

FIGURE 7
42 Equations of Order One [Ch. 2

2S. ydx = (2x + l)idx - dy). ans. 3y = {Ix + 1) -f c{2x + 1) " ''^

In solving Exs. 29-33, recall that the principal value Arcsin x of the inverse sine
function is restricted as follows:

^\7T < Arcsin x < hn.

29. Vl -y^ dx + \^\-x^ dy = 0.


ANS. Arcsin x + Arcsin j = c , or a part of the ellipse
x^ + Icixy + >'^ -f Ci^ 1 = 0, where Ci = cos c.
30. Solve the equation of Ex. 29 with the added condition that when x = 0,
y — jV3. ANS. Arcsin x + Arcsin y = i-rr, or that arc of the ellipse
x^ + xy + y^ = i that is indicated by a heavy solid line in Figure 7.

31. Solve the equation of Ex. 29 with the added condition that when x = 0,
y = — ^V3. ANS. Arcsin X + Arcsin >>=— ^77, or that arc of the ellipse
x^ + xy + y^ = J, that is indicated by a light solid line in Figure 7.

32. Show that after the answers to Exs. 30 and 31 have been deleted, the remaining
arcs of the ellipse

x^ + xy + y^ = i

are not solutions of the differential equation

Vl -y^dx + V\ - x^ dy = 0.
For this purpose consider the sign of the slope of the curve.
33. For the equation
Vl -y^dx-Vl -x^dy =
state and solve four problems analogous to Exs. 29-32 above.
34. V = {e" + luv - 2m) dv.
du ans. v^u = ce^" — {v+ \)e\
2>5.ydx = {2>x + y^-y^)dy;v.'\\cnx^\,y=~\. ans. x =^^[1 + 7 In (-j)].
36. y^ dx - {xy + 2)dy = 0. ans. xy = cy^ -I.
37. (x^ - Ixy - y"") dx - {x^ -f Ixy - y^) dy = 0.
ANS. (x + y)(x^ — 4xy + y^) = c^.
38. y^ dx + (xy + y^ - I) dy = 0; when x= -\,y = l.
ANS. y^ + 2xy + = 2 In 1 >'.

39. yiy^ - 3x^) dx + x^ dy = 0. ans. 2x^ = y^ix^ + c).


40. y' = cos — J sec .x; when x = 0,y =
a: 1.

+ sin x) = cos x(x + 2 — cos x).


ANS. >(1
41. Find that solution of y' = 3x + y — 1, 0).
which passes through the point (

ANS. y = —3(x + 1).


42. Find that solution of / = 3x + which passes through the point — 1, 1).
>- (

ANS. = exp (x-f 1) — 3(x+ 1). >'

43. y' =ytanx + cos x. ans. 2y = s'mx + (x + c) sec x.


44. (x^ -1+ 2y) dx + il - x^) dy^O; when x = 2,y = l.
ans. (x + l)y = - l)[x + + 2 In (x - 1)]. (a: 1

45. {x^ - 3xy^) dx^{y^ - 3x^y) dy = 0. ans. x^ - 6x^y^ + y* = c.


46. (I -x^)y' = -xy-3x^ + 2x''.
1 ans. y = x - x^ + c(l - x^Y'^.
47. (y^ + y)dx ~ {y^ + 2xy + x)dy = 0; when = 3, = ans. 2y^ + y =x. a- >' 1 .
Miscellaneous Exercises 43

48. (>'^ — x^) dx = xy{x dx + y dy). ans. = cx^ + Ixy — y^.


2x'^ In |x + >'|

49. y =secx —y tan x. = s\nx + c cos x.


ans. y
50. x^y' =y(\ — x). ans. x In = — 1. |ca:>'|

51. ;>r/ =x — + x>' tanx.


>' ans. cos x = c + cos x + x sin
a:>' a:.

52. {2)X*y — \)dx + x^dy = 0; when x = \,y = \. ans. ;c*>' = 2x — 1

53. y^ dx + x^ dy = Ixy dy. ans. y^ = x{y + c).


54. (sin sin
jc + tan x) dx — cos x cos y dy =^0. ans. cos x sin = In |c sec x|.
>' >'

55. (3x7 ~4y-\)dx + x(x -2)dy = 0; when x^\,y = l.


ans. 2x^(a: - 2)>' = x^ - 5.
CHAPTER O
Elementary Applications

13. Velocity of escape from the earth

Many physical problems involve differential equations of order one.


Consider the problem of determining the velocity of a particle projected
in a radial direction outward from the earth and acted upon by only one
force, the gravitational attraction of the earth.
We shall assume an initial velocity in a radial direction so that the motion
of the particle takes place entirely on a line through the center of the earth.
According to the Newtonian law of gravitation, the acceleration of the
particle will be inversely proportional to the square of the distance from the
particle to the center of the earth. Let r be that variable distance, and let R
be the radius of the earth. If / represents time, v the velocity of the particle, a
its acceleration, and k the constant of proportionality in the Newtonian law,
then
_dv _k_

44
§ 13] Velocity of escape from the earth 45

The acceleration is negative because the velocity is decreasing. Hence the


constant k is negative. When r = R, then a = —g, the acceleration of gravity
at the surface of the earth. Thus

from which

gR'
a = — -

We wish to express the acceleration in terms of the velocity and the distance.
dv dr
We have a = — and v^—. Hence
dt dt

dv
a = — = dr dv = —
dv
v
dt dt dr dr

so the differential equation for the velocity is now seen to be

dv
_ _ gR^
^-=--T-
dr r
(1)

The method of separation of variables applies to equation (1) and leads


at once to the set of solutions

IgR^
= -^— +
,
v^ C.
r

Suppose the particle leaves the earth's surface with the velocity ^o • Then
V = t'o when r = R, from which the constant C is easily determined to be

C=Vo'- 2gR.

Thus a particle projected in a radial direction outward from the earth's


surface with an initial velocity ^o will travel with a velocity v given by the
equation

v' = ^^ + Vo'-2gR. (2)


r

It is of considerable interest to determine whether the particle will escape


from the earth. Now at the surface of the earth, at r — R, the velocity is

positive, V = Vq. An examination of the right member of equation (2) shows


that the velocity of the particle will remain positive if, and only if,

Vo' - 2gR ^ 0. (3)


46 Elementary Applications [Ch. 3

For, if the inequality (3) is satisfied, the velocity given by equation (2) will
remain positive, since it cannot vanish, is continuous, and is positive at

r = R. On the other hand, if (3) is not satisfied, then Vq^ — 2gR < 0, and there
will be a critical value of r for which the right member of equation (2) is

zero. That is, the particle would stop, the velocity would change from positive

to negative, and the particle would return to the earth.

A particle projected from the earth with a velocity Vq such that Vq ^ y/2gR
will escape from the earth. Hence, the minimum such velocity of projection,

r, = ^2gR, (4)

is called the velocity of escape.


The radius of the earth is approximately R= 3960 miles. The acceleration
of gravity at the surface of the earth is approximately ^f = 32.16 feet per
second per second, or^ = 6.09(10)"^ miles per second per second. For the
earth, the velocity of escape is easily found to be v^ = 6.95 miles per second.

Of course, the gravitational pull of other celestial bodies, the moon, the
sun, Mars, Venus, etc. has been neglected in the idealized problem treated
here. It is not difficult to see that such approximations are justified, since we
are interested in only the critical initial velocity v^. Whether the particle

actually recedes from the earth forever or becomes, for instance, a satellite

of some heavenly body, is of no consequence in the present problem.


If in this study we happen to be thinking of the particle as an idealization
of a ballistic-type rocket, then other elements must be considered. Air
resistance in the first few miles may not be negligible. Methods for over-
coming such difficulties are not suitable topics for discussion here.

It must be realized that the formula v^ = JlgR applies equally well for

the velocity of escape from the other members of the solar system, as long

as R and g are given their appropriate values.

14, Newton's law of cooling

Experiment has shown that under certain conditions, a good approxima-


tion to the temperature of an object can be obtained by using Newton's law
of cooling: The temperature of a body changes at a rate which is proportional
to the diff"erence in temperature between the outside medium and the body
itself. We shall assume here that the constant of proportionality is the same
whether the temperature is increasing or decreasing.
Suppose, for instance, that a thermometer, which has been at the reading
70°F inside a house, is placed outside where the air temperature is 10°F.
§ 14] Newton's law of cooling 47

Three minutes later it is found that the thermometer reading is 25°F. We


wish to predict the thermometer reading at various later times.

Let u (°F) represent the temperature of the thermometer at time / (min.),


the time being measured from the instant the thermometer is placed outside.
We are given that when / = 0, w = 70 and when / = 3, w = 25.

According to Newton's law, the time rate of change of temperature, du/dt,


is proportional to the temperature difference (u — 10). Since the thermometer
temperature is decreasing, it is convenient to choose {-k) as the constant
of proportionality. Thus the u is to be determined from the differential

equation

du
-^-k(u-\0), (1)

and the conditions that

when t^0,u = 10 (2)

and

when t — 3, u = 25. (3)

We need to know the thermometer reading at two different times because


there are two constants to be determined, the k in equation (1) and the
"arbitrary" constant that occurs in the solution of the differential

equation (1).

From equation (1) it follows at once that

Then condition (2) yields 70 = 10 + C from which C— 60, so we have

u = 10 + 60e-'". (4)

The value of k will be determined now by using the condition (3). Putting
/ = 3 and u = 25 into equation (4) we get

25= 10 + 60e--'\

from which e'^"" = I, so k = ^ In 4.

Thus, the temperature is given by the equation

?/= 10 + 60exp(-iMn4). (5)

Since In 4 = 1.39, equation (5) may be replaced by

z/= 10 + 60 exp(- 0.460, (6)

which is convenient when a table of values of e~^ is available.


48 Elementary Applications [Ch. 3

15. Simple chemical conversion

It is known from the results of chemical experimentation that, in certain


reactions in which a substance A is being converted into another substance,
the time rate of change of the amount x of unconverted substance is pro-
portional to X.
Let the amount of unconverted substance be known at some specified tim.e;

that is, let .y = Xq at / = 0. Then the amount x at any time / > is determined
by the differential equation

and the condition that x — Xq when


S— / = 0. Since the amount x is decreasing
(')

as time increases, the constant of proportionality in equation (1) is taken to


be (-A).
From equation (1) it follows that

A- = Ce-"'.

But A' = Xq when t = 0. Hence C= Xq . Thus we have the result

x^xoe-''. (2)

Let us now add another condition, which will enable us to determine k.


Suppose known that at the end of half a minute, at r = 30 (sec), two-
it is

thirds of the original amount Xq has already been converted. Let us determine
how much unconverted substance remains at / = 60 (sec).
When two-thirds of the substance has been converted, one-third remains
unconverted. Hence a- = ^a'q when t = 30. Equation (2) now yields the
relation
3X0 — Ao^
from which k is easily found to be -j- In 3. Then with t measured in seconds,

the amount of unconverted substance is given by the equation

x = Xoexpi--^t\n3). (3)
At t = 60,
X = Xq exp (-2 In 3) = .Vo(3)"^ = }xq .

Exercises
1. The radius of the moon is roughly 1080 miles. The acceleration of gravity at
the surface of the moon is about 0. 65g, where g is the acceleration of gravity
at the surface of the earth. Determine the velocity of escape for the moon.
ANS. 1.5 mi. /sec.
§ 15] Simple chemical conversion 49

2. Determine to two significant figures the velocity of escape for each of the
celestial bodies listed below. The data given are rough and g may be taken to
be 6.1(10)-^ mi./sec.^

Accel, of gravity Radius Ans. in

at surface in miles mi. /sec.

Venus
50 Elementary Applications [Ch. 3

11. Two substances, A and B, are being converted into a single compound C. In
the laboratory it has been shown that, for these substances, the following law
of conversion holds: the time rate of change of the amount x of compound Cis
proportional to the product of the amounts of unconverted substances A and
B. Assume the units of measure so chosen that one unit of the compound C is

formed from the combination of one unit of A with one unit of B. If at time
/ = there are a units of substance A, b units of substance B, and none of the
compound C, present, show that the law of conversion may be expressed by the
equation
dx
— =k(a-x)(b-x).
Solve this equation with the given initial condition.
ab[e\p {b — d)kt —
U b ^ a, X = —
1]
ANS. -; -; ;'\ib = a,x = a^kt/(akt + 1 ).
b exp {b — a)kt — a
12. In the solution of Ex. 1 1 above, assume that /: > and investigate the behavior
ofjcas/->oo. ANS. \i b^ a, x->a\'\i b -^ a, x-^b.
13. Radium decomposes at a rate proportional to the quantity of radium present.
Suppose that it is found that in 25 years approximately 1.1 per cent of a certain
quantity of radium has decomposed. Determine approximately how long it will
take for one-half the original amount of radium to decompose.
ANS. 1600 years.
14. A certain radioactive substance has a half-life of 38 hours. Find how long it

takes for 90% of the radioactivity to be dissipated. ans. 126 hrs


15. A bacterial population B is known to have a rate of growth proportional to B
itself. between noon and 2 p.m. the population triples, at what time, no
If

controls being exerted, should B become 100 times what it was at noon?
ANS. About 8:22 p.m.
16. In the motion of an object through a certain medium (air at certain pressures is

an example), the medium furnishes a resisting force proportional to the square


of the velocity of the moving object. Suppose a body falls, due to the action of
gravity, through such a medium. Let / represent time, and v represent velocity,
positive downward. Let g be the usual constant acceleration of gravity and let
w be the weight of the body. Use Newton's law, force equals mass times accelera-
tion, to conclude that the differential equation of the motion is

w
—dv
r
gdt
= M' — kv'^,

where kv^ is the magnitude of the resisting force furnished by the medium.
17. Solve the differential equation of Ex. 16 with the initial condition that v — Vo
when t = 0. Introduce the constant a^ = wjk to simplify the formulas.

o J
ANS.
a
a
+ v = a + Vo
— V a — Vo exp I
\

/2^^\
I

18. List a consistent set of units for the dimensions of the variables and parameters
of Exs. 16-17 above. ans. / in sec. g in ft./sec.^
V in ft./sec. k in (lb.)(sec.^)/(ft.^)
w in lb. a in ft./sec.
§15] Simple chemical conversion 51

There are mediums that resist motion through them with a force proportional
to the first power of the velocity. For such a medium, state and solve problems
analogous to Exs. 16-18 above, except that for convenience a constant b = wjk
may be introduced to replace the a^ of Ex. 17. Show that b has the dimensions
of a velocity.
— b + KVo —
, ^
ANS. V b) exp
-f

FIGURE

w pounds, sliding down an inclined plane which makes


Figure 8 shows a weight,
an angle a with the horizontal. Assume that no force other than gravity is
acting on the weight; that is, there is no friction, no air resistance, etc. At
time / = 0,let X = Xo and let the initial velocity be vo Determine x for / > 0. .

ANS. x= Igt^ smoi.-\- Vot -\- Xo.


A long, very smooth board is an angle of 10° with the horizontal.
inclined at
A weight starts from rest ten feet from the bottom of the board and slides

downward under the action of gravity alone. Find how long it will take the
weight to reach the bottom of the board and determine the terminal speed.
ANS. 1.9 sec. and 10.5 ft./sec.
Add to the conditions of Ex. 20 above a retarding force of magnitude kv,
where v is the velocity. Determine v and x under the assumption that the weight
starts from rest with x = Xo. Use the notation a = kg/w.
ANS. y = a~'^sin a(l — e'"'); x =^ Xo + a~^g sin (x.(— I + e'"' + at).
A man, standing at O in Figure 9, holds a rope of length a to which a weight is
attached, initially at P^o • The man walks to the right dragging the weight after
him. When the man is at M, the weight is at W. Find the differential equation
of the path (called the tractrix) of the weight and solve the equation.
a + Va^-,
ANS. a In Va^-
y
24. A tank contains 80 gallons of pure water. A brine solution with 2 pounds of
salt per gallon enters at 2 gallons per minute, and the well-stirred mixture
leaves at the same rate. Find (a) the amount of salt in the tank at any time, and
(b) the time at which the brine leaving will contain 1 pound of salt per gallon.
ANS. (a) s = 160[1 - exp (- ^/40)] (b) ^ = 40 In 2 min. ;
52 Elementary Applications [Ch. 3

FIGURE 9

25. For the tank in the previous problem, determine the limiting value for the
amount of salt in the tank after a long time. How much time must pass before
the amount of salt in the tank reaches 80% of this limiting value?
26. A certain sum of money P draws interest compounded continuously. If at a
certain time there is Po dollars in the account, determine the time when the
principal attains the value IPo dollars, if the annual interest rate is (a) 2%, or
(b) 4 % ANS. (a) 50 In 2 years (b)
;
1 00 In 2 years.

16. Orthogonal trajectories; rectangular coordinates

Suppose that we have a family of curves given by

/(.x,j,c) = 0, (1)

one curve corresponding to each c in some range of values of the parameter c.

In certain applications it is found desirable to know what curves have the


property that wherever any one of them intersects a curve of the family (1)

it does so at right angles.

That is, we wish to determine a family of curves with equations

g{x,y,k) = Q (2)

such that at any intersection of a curve of the family (2) with a curve of the
family (1) the tangents to the two curves are perpendicular. The families

(1) and (2) are then said to be orthogonal trajectories* of each other.

If two curves are to be orthogonal, then at each point of intersection the

* The word orthogonal comes from the Greek opd-q (right) and ytuvta (angle); the word
trajectory comes from the Latin trajectus (cut across). Hence a curve that cuts across certain
others at right angles is called an orthogonal trajectory of those others.
§ 16] Orthogonal trajectories; rectangular coordinates 53

slopes of the curves must be negative reciprocals of each other. That fact
leads us to a method for finding orthogonal trajectories of a given family of
curves. First we find the differential equation of the given family. Then
replacing dyjdx by —dxjdy in that equation yields the diftcrential equation
of the orthogonal trajectories to the given curves. It remains only to solve
the latter differential equation.
So far we have solved differential equations of only one form,

Mdx + Ndy = 0.
For such an equation

dy _ M
'dx~ ~~N'

so the differential equation of the orthogonal trajectories is

dy _ N
dx M
or

TV dx - M dy = 0.

EXAMPLE: Find the orthogonal trajectories of all parabolas with vertices


at the origin and foci on the .t-axis.

The algebraic equation of such parabolas is

y^ = 4ax. (3)

Hence, from

— =4fl,
X

we find the differential equation of the family (3) to be

2.V dy - y dx = 0. (4)

Therefore the orthogonal trajectories of the family (3) must satisfy the

equation

2x dx + y dy = 0. (5)

From (5) it follows that

2x^ +y^ = b\ (6)

where b is the arbitrary constant. Thus the orthogonal trajectories of (3)

are certain ellipses (6) with centers at the origin. See Figure 10.
54 Elementary Applications [Ch. 3

FIGURE 10

Exercises
In each exercise, find the orthogonal trajectories of the given family of curves.
Draw a few representative curves of each family whenever a figure is requested.

X — 4y = c. Draw the figure. ANS. 4x + y = k.


x^ + _V^ = c^. Draw the figure. ANS. y = kx.
x^ — y^ = Ci. Draw the figure. ANS. xy = Ci.
Circles through the origin with centers on the x-axis. Draw the figure.
ANS. Circles through the origin with centers on the j-axis.
5. Straight lines with slope and >'-intercept equal. Draw the figure.
ANS. {x^\Y +y^^a''.
= cx^. Draw the figure. ans. Ix^ + 2>y^ = k^.
ANS.
y = Ci(sec X + tan x).
x^ = 3{y — c). Draw the figure.

x = c exp iy^).
§ 17] Orthogonal trajectories; polar coordinates 55

17. ax^ -\- y^ = 2acx, with a held fixed. ans. l(a^2,{2 — a)x^ -\- y'^ = dy".
If a = 2, x^ = —y^ In {czy).
18. >'(x^ + c) + 2 = 0. ANS. y^=-Z\n\kx\.
19. x" + y" = a", with n held fixed and n^2. ans. x^"" — >'^~" = c.
20. ^'^ = x^{\ — ex). ANS. x^ + 3y^ = Ci>'.
21. y^ = 4x^1 - ex). ANS. 2x^ = 37^(1 - c^y^
22. j'^ = ax^{\ — ex), with a held fixed.

ANS. If a ^ 2, - 2)x^ = 37^(1 - c^y" " %


(fl

If a = 2, x^ = —3y^ In |c2>'|.
23. y(x^+l) = ex. ans. >'^ - x^' + 2 In |A:(x^ - 1)|.
24. >' = 3x- +ce-3\ 1 ANS. 27x = 9y~l+ke-^''.
25. >'2(2a:^ + j;^) = e\ ans. j^^ = 2x^ In |A:a:|.

26. 7* = c\x^ + 4>'^). ANS. x\2x^ + 5y^) = k\


11. x\Ax^ + 7,y^) = c\ ANS. y^ = k\Zx^ + 2y^).
28. For the family x^ + 2)y^ = cy, find that member of the orthogonal trajectories
which passes through (1, 2). ans. y^ = x\3x + \).

17. Orthogonal trajectories; polar coordinates

Consider a curve whose equation is expressed in polar coordinates. In


calculus it is shown that the angle xj/ (Figure 11), measured positive in the

counterclockwise direction from the radius vector to the tangent line at a


point, is given by

dO
tan ^/ = r —
dr

If two curves are orthogonal, as they are shown in Figure 12, then
1A2 = lAi + ^/2 and thus tan ipz — ~ cot i/^i = — 1/tan i/zj. Therefore, if two
curves are to cut one another at right angles, then at the point of intersection

FIGURE II
56 Elementary Applications [Ch.3

FIGURE 12

the value of the product

for one curve must be the negative reciprocal of the value of that product
for the other curve.
When polar coordinates are used, suppose a family of curves has the
differential equation

Pdr+QdO = 0. (1)

Then
dO_ _P
'd'r~ Q
so

d9 Pr
(2)
dr

Hence the family of the orthogonal trajectories of the solutions of (1) must
be solutions of the equation

de Q
dr Pr
or

Q dr - r^P dO = 0. (3)

EXAMPLE: Find the orthogonal trajectories of the family of cardioids


r = a{\ + cos 0). A few curves of each family are shown in Figure 13.

From

— a
1 + cos
§17] Orthogonal trajectories; polar coordinates 57
58 Elementary Applications [Ch. 3

Exercises
In each exercise, find the orthogonal trajectories of the given family of curves.
Draw the figures in Exs. 1-8.

1. = b sin 6
ANS. r^
= b sin 2d
ANS. r'*

ANS. r = 2b cos 6

ANS. r = 2b{sm 9 — cos 6)


ANS. r = bil — sin 6)

ANS. r^ ^ b cos 29,


ANS. /-^(l + sin^ 9) = b^
ANS. r^ sin 29 = C2
ANS. r^ sin^ = /)(1 + cos 0)
ANS. r sin^ = 6(1 + cos 9y
11. /• = A:/(l + £ cos 9) with e fixed. What are /: and e geometrically?
ANS. r' sin' ^ ^
9 = b(\ + COS 9)
12. r=a(l-2 sin ^). ans. r^ = b cos ^(1 + sin 9)
13. The strophoids, r = a(sec 9 + tan 9). ans. r = 6e"''"*

14. The cissoids, r = o sin ^ tan 9. ans. /-^= 6(1 + cos^ 6)


15. The trisectrices of Maclaurin, r = a(4 cos 9 — sec 9).

ANS. r' = 6 sin^ ^(4 cos^ 9 + \)

16. J- = a(l + sin^ 9). ans. /-^ = 6 cos cot 9


CHAPTER 4
Additional Topics on Equations of

Order One

18, Integrating factors found by inspection

In Section 1 1 we found that any linear equation of order one can be solved
with the aid of an integrating factor. In Section 19 there is some discussion of
tests for the determination of integrating factors.
At present we are concerned with equations that are simple enough to
enable us to find integrating factors by inspection. The ability to do this de-
pends largely upon recognition of certain common exact differentials and
upon experience.
Below are four exact differentials which occur frequently:

d(xy) = X dy + y dx, (1)

,/x\ y dx — X dy

Jy\ ^ xdy - ydx


(3)
\xj x^

59
60 Additional Topics on Equations of Order One [Ch. 4

c^
/

\
Arctan-
v\

x/
=
X dy
X
{ ——+ 4-.y dx
y
(4)

Note the homogeneity of the coefficients of dx and dy in each of these


differentials.

A differential involving only one variable, one such as x"^ dx, is an exact
differential.

EXAMPLE (a): Solve the equation

ydx + {X + X V) iiy = 0- (5)

Let us group the terms of like degree, writing the equation in the form

(y dx + X dy) + x^y^ dy = 0.

Now the combination (y dx + x dy) attracts attention, so we rewrite the


equation, obtaining

d(xy) + A-V' dy = 0. (6)

Since the differential of xy is present in equation (6), any factor that is a

function of the product .vv will not disturb the integrability of that term. But
the other term contains the differential dy, hence should contain a function
of y alone. Therefore, let us divide by (xy)^ and write

d(xy) d_y^^
(xy)' y

The equation above is integrable as it stands. A family of solutions is

-T-2-^ + lnbi = -ln|c|,


2x y

or

2.vVlnk>i = l.

EXAMPLE (b): Solve the equation

y(x' - y) dx - x(x' + y) dy = 0. (7)

Let us regroup the terms of (7) to obtain

x'iy dx — X dy) — yiy dx + x dy) =^ 0. (8)

Recalling that

dx — x dy
<;)=^ y
2
'
§ 18] Integrating factors found by inspection 61

we divide the terms of equation (8) throughout by y^ to get

,3,/M_f((^^0. (9)

Equation (9) will be made exact by introducing a factor, if it can be found,


to make the coefficient of d{x/y) a function of (x/y) and the coefficient of
d{xy) a function of{xy). Some skill in obtaining such factors can be developed
with a little practice.
There is a straightforward attack on equation (9) which has its good points.
Assume that the integrating factor desired is x'^y", where k and n are to be
determined. Applying that factor, we obtain

x'^^^y" d(-\ - A-y-^ d(xy) = 0. (10)

Since the coefficient of d{x/y) is to be a function of the ratio (x/y), the


exponents of x and y in that coefficient must be numerically equal, but of
opposite sign. That is,

k + 3=-n. ^
(11)

In a similar manner, from the coefficient of d(xy) it follows that we must put

k = n-l. (12)

From equations (11) and (12) we conclude that k = —2, n = —I. The
desired integrating factor is x~^y~^ and (10) becomes

y \y/ X y

of which a set of solutions is

2 \y/ xy 2

Finally, we may write the desired solutions of equation (7) as

x^ + 2y = cxy^.

EXA MPLE (c) : Solve the equation

3x^y dx + (/ - x^) dy = 0.

Since two terms in the coefficients of dx and dy are of degree three and the
other coefficient is not of degree three, let us regroup the terms to get

(3jc V dx - x^ dy) + y"^ dy = 0,

or
y d{x^) - x^ dy -\- y'*' dy = 0.
62 Additional Topics on Equations of Order One [Ch. 4

The form of the first two terms now suggests the numerator in the differential

of a quotient, as in

'm\ V du — u dv
VI v^

Therefore, we divide each term of our equation by y^ and obtain

y d(x^) — x^ dy
+ y' dy = 0,
/
or

di—j +y^dy = 0.

Hence a solution set of the original equation is

>'
^ 3
~ 3

or

3.v^ + y'^ = cy.

Exercises
Except when the exercise indicates otherwise, find a set of solutions.

1. yilxy + \) dx — xdy = 0. ans. x{xy + 1) = cy.


2. y{y^ — x)dx + x{y^ + x) dy ^ 0. ans. Ixy^ — x^ = cy^.
3. {x^y^ + 1) Jjc + x'^y^ dy = 0. ans. x^y^ = -3 In |cx|.
4. 2tds + sil + s^t)dt=id. ANS. \+s^t=cs^t^.
5. y{x* - y"") dx + x^x^ + y"") dy = 0. ans. yi3x* + y^) = cx\
6. y(y^ + I) dx + x(y^ - I) dy = 0. ans. x(y^+l) = cy.
7. Do Ex. 6 by a second method.
8. y(x^ - y*) dx — xix^ + 7^) dy = 0. ans. x* = y\c + ^xy).
^. y{x^-y^ +\)dx- x{x^ -y"^ -\)dy = ^. ans. x"" ^ cxy + y^ =\.
10. {x^ + xy^ ^y)dx\ {y"" + x^y + x) dy = 0.
ANS. (x^ + y'^y = c — Axy.
11. y{x^ + y^-\)dx + x{x^ + y^+\)dy = 0.
ANS. xy + Arctan {yjx) = c.
12. (x^ + xy^ -y)dx + (y"" + x^y + x) dy = 0.
ANS. 2 Arctan (y/x) = c — x^ — y^.
13. yix^e"" -y)dx + x{y + x V^*') c/f = 0. ans. 2x V^^ + j'^ = cx^
14. xj^(>'^ + 1) Jx + (x^^'^ - 2) c/;' = 0; when x = 1, >' - 1

ans. x^(j^+ l) = 2 + 41n>'.

15. y^iX — x^) dx + x(xV + 2x + f/v = 0. J') ans. + x + = cxy^.


x^y ^^

16. y{x^y^ -\)dx + x{x^y^ + \)dy = 0. ans. x^y"- = 2 In Icx/^l.


17. xV = — x^>' — CSC {xy). ans. 2x^ cos (x^) = cx^ — 1.
§ 19] The determination of integrating factors 63

18. [1 tan (xy)] dx + x tan {xy) dy = 0.


+ >' ans. cos {xy) = ce".
19. y(x^y^ — m) dx + x(x^y^ + n) dy = 0. ans. x^y^ = 2\n\cx"'ly"\.
20. x(x^ — y^ — x) dx — y(x^ — y^) dy = 0; when x = 2, y = 0.
ANS. 3ix^ - y^y = 4(a:^ + 4).
21. j(x^ -\-y)dx + xix^ - 2y) dy = 0; when x=\,y = 2.
ANS. x^y — y^ + 2x = 0.
22. y(x^y^ + 2x^ — y) dx + x^(xy^ — 2) dy =^0; when x= l,y= 1.
ANS. x^y^ + 4x^ — Ixy + 2y = 0.
23. y{2 — lixy) dx — x dy ^ 0. ans. x^(\ — xy) = cy.
lA. y{2x + y^) dx + x{y^ — x) dy = 0. ans. x{x + y^) = cy.
25. ydx + 2{y'^ - x) dy ^ 0. ans. y'- ^x = cy^.
26. yiSx^ - X + y) dx + x\\ - x^) dy = 0. ans. yln\cx\=xil-x^).
27. 2x'y' = j'CSx'* + y^). ans. = 7^(1 + ex). a:'^

28. ix"y" ^' +ay)dx + (x" + 'y" + bx) dy = 0.


ans. l{ n ^0,x"y" = n\n\cx~''y~''\.
\{ n = 0, xy ^ CiX~''y~''.
29. {x" + + oj) dx + (x">;" + + ox) (/y = 0.
'>'" 1

ANS. If « ^ 1, - l)(x>')"-'(x^ +^2 - c) = 2a.


(a;

If = \,x^ y^ — c= —2a In \xy\.


// -\-

19. The determination of integrating factors

Let US see what progress can be made on the problem of the determination
of an integrating factor for the equation

Mdx + Ndy = 0. (1)

Suppose u, possibly a function of both x and y, is to be an integrating factor


of (1). Then the equation

uM dx + uN dy = (2)

must be exact. Therefore, by the result of Section 10,

dy ox

Hence u must satisfy the partial differential equation

dM dN
+ M — = u-— + N —
du du
u^
dy dy ox ox
,

or

/dM dN\ _ du du
\dv dxl dx dv
64 Additional Topics on Equations of Order One [Ch. 4

Furthermore, by reversing the argument above, it can be seen that if u


satisfies equation (3), then u is an integrating factor for equation (1). We have
"reduced" the problem of solving the ordinary differential equation (1) to

the problem of obtaining a particular solution of the partial differential


equation (3).

Not much has been gained, since we have developed no methods for
attacking an equation such as (3). Therefore, we turn the problem back into
the realm of ordinary differential equations by restricting w to be a function
of only one variable.
First let w be a function of x alone. Then cu/dy = and dujdx becomes
dujdx. Then (3) reduces to

IdM

dN\
=iV—
dx dx
du

or

1 JdM dN\
N \dy
\dv
~ Jx)
ox u

If the left member of the above equation is a function of x alone, then we


can determine u at once. Indeed, if

1 /dM
N (--f)=/(.),
\dy

then the desired integrating factor is m = exp (J/(-v) dx).

By a similar argument, assuming that w is a function of ^^ alone, we are led

to the conclusion that if

1 (dM dN\

then an integrating factor for equation.(l) is w = exp (-^ g(y) dy).

Our two results are expressed in the following rules.

If — j
1 = /(.y), a function of .v alone, then exp (j f{x) dx) is an
N\dy dx)
integrating factor for the equation

M ^.v + iV ^>' = 0. (1)

If — = g{y), a function of v alone, then exp {—\g{y) dy) IS


i an
M\cyI I
dx)
integrating factor for equation (1).
It should be emphasized that if neither of the preceding criteria is satisfied,

we can say only that the equation does not have an integrating factor that is a
§ 19] The determination of integrating factors 65

function of ^ or >^ alone. For example, the student should show that the above
criteria fail in the case of Example (a) of Section 18, even though (xy)~^ is an
integrating factor for the differential equation.

EXAMPLE (a): Solve the equation

(4xy + 3y^ - x) dx + x(x + 2y) dy = 0. (7)

Here M = 4xy + 3y^ — x, N = x^ + 2xy, so


dM dN
-^ = 4x + 6y - (2x + 2y) = 2x + 4y.
dy ox

Hence

1 (dM
/dM dN\
8N\ _ 2x + 4y 2

N \dy dxj x{x + 2y) x

Therefore an integrating factor for equation (7) is

exp 1 2 — I = exp (2 In |x|) = x^.

Returning to the original equation (7), we insert the integrating factor and
obtain

(4jcV + 3x^y^ - x^) dx + {x"" + 2^-^) dy = 0, (8)

which we know must be an exact equation. The methods of Section 10 apply.


We are then led to put equation (8) in the form

(4jc V dx + jc* dy) + Qx^y^ dx + 2x^y dy) - x^ dx = 0,

from which the solution set

x'^y + x^y^ - ^x'^ = Ic,

or

x^{4xy + 4y^ — x) = c

follows at once.

EXAMPLE (b) : Solve the equation

y{x + y+l)dx + x{x + 3y + 2) dy = 0. (9)

First we form
dM
-— dN
= x + 2y+l, -- = 2x + 3y + 2.
dy dx
66 Additional Topics on Equations of Order One [Ch. 4

Then we see that

dM _dN__
dy dx
so

1 (^ _ ^\ - _ '^ + 3'+ 1
N \dy dx/ x{x + 3y + 2)

is not a function of x alone. But

1 /dM
/dM _^\
8N\ _ _ x + y+l 1

M \dy dx] y{x + y + 1) y

Therefore exp (In |>'|) = |>'| is the desired integrating factor for (9).

It follows that for y >Q, y itself is an integrating factor of equation (9)

and for jf < 0, —y is an integrating factor. In either case (9) becomes

{xy^ + j^ + y^) dx + (.y^>' + 3.vj^ + 2xy) dy = 0,

or
{xy^ dx + x^y dy) + {y^ dx + Ixy^ dy) + {y^ dx + 2xy dy) = 0.

Then a set of solutions of (9) is defined implicitly by

^x^y^ + xy^ + xy^ = \c,

or
xy^{x + 2>' + 2) = c.

EXA MPLE (c): Solve the equation

y{x-\-y)dx + {x^2y-\)dy = 0. (10)

From —
dM =
dy
X+ 2v, —
dN
dx
= 1, we conclude at once that

1 /dM _ dN\ _ x + 2y-\


^^
N \dy dxj x + 2y -\
Hence e"" is an integrating factor for (10). Then

{xye'' + y^e"") dx + (a'^'"' + 2ye'' - e"") dy =


is an exact equation. Grouping the terms in the following manner,

[xye"" dx + (xe'' - e'') dy] + (y^e'' dx + 2ye'' dy) = 0,

leads us at once to the family of solutions defined by

e\x — 1 )>' + y^e"" = c,

or
y(x +y- \) = ce'"".
§ 19] The determination of integrating factors 67

Exercises
Solve each of the following equations.

1. (x^ + y^ + \)dx + xix — 2y) dy = 0. ans. x^ — y'^ xy — = ex. -\- \

2. 2y{x^ — y + x)dx^-{x^ — 2y) dy = 0. ans. y(x^ —y) = ce-^\


2>. y{lx-y+\)dx + xOx-Ay + 'i)dy = 0. ans. xy^{x - y + \) = c.
4. y{Ax y)dx — 2(x^ — y) dy ^ 0.
-]- ans. 2x^ + xy + 2y In = cy. |>'l

5. (xy+\)dx + x(x + 4y—2)dy = 0. ans. xy + \n\x\ + 2y^ — 2y=c.


6. {2y^ + Zxy - 2y + 6x) dx + xix + 2y - \) dy = 0.
ans. x^iy^ + xy — y + 2x) = c.
7. yiy + 2x — 2)dx— 2(x + y) dy = 0. ans. yi2x + y) = ce\
S. y^dx + (3xy + y^-l)dy = 0. ans. y\y^ + 4xy - 2) = c.
9. 2y(x + y + 2)dx + (y^-x^-4x-l)dy = 0.
ans. x^ + 2xy + y^ + 4x + =^ cy. I

10. 2(2y^ + 5xy - 2y + 4) dx + ;c(2x + 2y - \) dy = 0.


ans. x'^iy^ + 2xy — y + 2) = c.
11. 3(a:^ + y^) dx + xix^ + 3y^ + 6y) dy = 0. ans. x(x^ + 3y^) = ce-\
12. yi%x - 9y) dx + 2x{x - 3y) dy = 0. ans. x^y{2x - 3y) = c.
13. Do Ex. 1 2 by another method.
14. y(2x^ — xy +\) dx + (x — y) dy =^ 0. ans. yi2x — y) = c exp (— x^).
15. Euler's theorem (Ex. 34, page 29) on homogeneous functions states that if F
is a homogeneous function of degree k in x and y, then

^ ^
ox
+J—
cy
=kF.

Use Euler's theorem to prove the result that if M and A'^ are homogeneous
functions of the same degree, and if Mx + Ny ^0, then

Mx + Ny
is an integrating factor for the equation

Mdx + Ndy = 0. (1)

16. In the result to be proved in Ex. 15 above there is an exceptional case, namely,
when Mx + Ny = 0. Solve equation (1) when Mx + Ny = 0. ans. y = ex.

Use the integrating factor in the result of Ex. 15 above to solve each of the equa-
tions in Exs. 17-20.

17. xy dx — {x^ + 2^^) dy = 0. ans. x^ = 4y^ In \y/c\


18. v^ dx + x(x + v) dv^ 0. ans. xv^ = c(x + 2v).
19. v(u^ + v^) du - u{u^ + 2v^) dv = 0. ans. u^ = 2v^ In \cv^/u\.

20. (x^ + y^) dx -xydy = 0. (Ex. 9, page 28.)


21. Apply the method of this section to the general linear equation of order one.
68 Additional Topics on Equations of Order One [Ch. 4

20. Substitution suggested by the equation

An equation of the form

M dx + N dy^O
may not yield at once (or at all) to the methods of Chapter 2. Even then the
usefulness of those methods is not exhausted. It may be possible by some
change of variables to transform the equation into one of a type which we
know how to solve.
A natural source of suggestions for useful transformations is the differential
equation itself If a particular function of one or both variables stands out
in the equation, then it is worth while to examine the equation after that

function has been introduced as a new variable. For instance, in the equation

(x + 2y - 1 ) ^.v + 3(.v + 2y) dy = (1)

the combination {x + 2y) occurs twice and so attracts attention. Hence


we put
x + 2y ^ V,

and since no other function of x and y stands out, we retain either .y or y


for the other variable. The solution is completed in Example (a) below.
In the equation

( 1 + 3.Y sin y) dx - x^ cos y dy = 0, (2)

the presence of both sin y and its differential cos y dy, and the fact that y
appears in the equation in no other manner, leads us to put sin y = \v and
to obtain the differential equation in u' and .r. See Example (b) below.

EXAMPLE (a): Solve the equation

{X + 2>' - 1 ) dx + 3(a- + 2y) dy = 0. (1)

As suggested above, put


X + 2y — V.

Then

dx = dv — 2 dy

and equation (1) becomes

{v- l)(^r-2^>') + 3r^.v = 0,

or

(v - 1) dv + (r + 2) dy = 0.
^ fly
v + 2
70 Additional Topics on Equations of Order One [Ch. 4

If n = 1 in (1), the variables are separable, so we concentrate on the case


/7 7^ 1. Equation (1) may be put in the form

y-"dy + Py-"'-' clx = Q dx. (2)

But the differential of j~""^' is (1 — n)y'" dy, so equation (2) may be simpli-
fied by putting
y-"^'=z,
from which
(1 - n)y~" dy = dz.

Thus the equation in z and x is

Jz + (1 - n)Pz dx = {\ -n)Q dx,

a linear equation in standard form. Hence any Bernoulli equation can be


solved with the aid of the above change of dependent variable (unless // = 1,

when no substitution is needed).

EXA MPLE (a) : Solve the equation

y(6y^ -x-\)dx + 2x dy = 0. (3)

First let us group the terms according to powers of >', writing

2x dy - y{x + 1) dx + 6y^ dx = 0.

Now it can be seen that the equation is a Bernoulli equation, since it in-

volves only terms containing respectively dy, y, and y" (n = 3 here). Therefore,

we divide throughout by >'^, obtaining

2xy~^ dy - y~^{x + 1) ^a" = -6 dx.

This equation is linear in y~^, so we put y~^ = v, obtain dv = —2y~^ dy,

and need to solve the equation

A- dv + v{x + \) dx — 6 dx,

or
- -
dv = r( 1 + A '
) dx = 6a 1 dx. (4)

Since

expCA' -I- In |a|) = Ia'I e""

is an integrating factor for (4), the equation

xe'' dv + ve\x + \) dx ^ ee"" dx

is exact. Its solution set

xve^ = 6e^ + c.
§ 21] Bernoulli's equation 71

together with v — y~^, leads us to the final result

y^(6 + ce'"") = X.

EXAMPLE (b): Solve the equation

6y^ dx - x(2x^ + y) dy = 0. (5)

This is a Bernoulli equation with x as the dependent variable, so it can be


treated in the manner used in Example (a). That method of attack is left for
the exercises.
Equation (5) can equally well be treated as follows. Note that if each
member of (5) be multiplied by .y^, the equation becomes

6y^x^ dx - x\2x^ +y)dy = 0. (6)

In (6), the variable x appears only in the combinations x^ and its differential

3.v^ dx. Hence a reasonable choice for a new variable is w= x^. The equation
in u' and y is

2y^ dw — w{2w + y) dy = 0,

an equation with coefficients homogeneous of degree two in y and vv. The


further change of variable vv = zy leads to the equation

2y dz - z{2z - \) dy = 0,

dy
2z
4 dz
— 1
2 dz

z
- — = 0.
y
Therefore, we have
21n|2z- 1| -21n|z| - In b| = In |r|,

or

(2z - \Y = cyz^.

But z = wjy — x^jy, so the solutions we seek are

(2x^ - y)^ = cyx^.

Exercises
In Exs. 1-21 solve the equation.

1. (3a: -2y+\)dx + (3x - 2y + 3) dy = 0.


ANS. 5(x + +>' c) = 21n|15A:- 10^+ 11|.

2. sin yix + sin y) dx + 2x^ cos y dy = 0. ans. x^ sin^ y = c(3x + sin yY.
dy
3. -f- = (9x + 4y+ \y. ans. 3 tan (6x + c) = 2(9x + 4y + 1).
dx
72 Additional Topics on Equations of Order One [Ch. 4

4. y' = y — xy^e'^". ans. e^'' = y^(x^ + c).


dy
5. —= + y).
sin {x ans. x + c = tan (x + >') — sec (x + >').

6. j:>' f/jr + (x^ — 3>') dy = 0. ans. x^y^ = 2y^ + c.


7. (3 tan x — 2 cos y) sec^ x ^/jc + tan x sin j' Jy = 0.
ans. cosj tan^ X = tan^ X + c.
8. ix + 2y- l)dx + i2x + 4y ~ 3) dy = 0. Solve by two methods.
ans. ix + 2y- ly = 2y + c.
9. Solve the equation

6y^ dx - x{2x^ + y) dy =

of Example (b) above by treating it as a Bernoulli equation in the dependent


variable x.
10. 2x^y' = y(y^ + 3x^). Solve by two methods. ans. y^(c — x) = x
11. (3 sin y— 5x) dx + 2x^ cot y dy = 0. ans. x^(sin — xY = c sin ^ y
y
12. >'' = + 1 6x exp (x — y). ans. exp {y — x) = 3x^ + c
dv
13. —= (// — vY — 2(u — v) — 2. ans. {u —V— 3) exp (4//) = c{n — + 1) i;

dii

\4. 2y dx + x{x^\x\y — \)dy = Q). ans. ^(1 + x- — x^ Inj) = cx^


15. cos >- sin 2x dx -j- (cos^ j^ — cos^ x) dy = 0.

ANS. cos^ a(1 + sin y) = cos y{y + c — cos y)


16. (A-e"^' - II) dii = 2e^%e^'' + ku) dv.
ANS. 2k Arctan (iie~^'') = In \c{u^ + e*")]
17. >-' tan x + cos^
sin Ij^ = sin^ ans. (sin^ x + 3 cos^ y) sin x = c
x >'.

18. (x + 2:f - 1) t/x - (x + 2>' - 5) J;^ = 0.


19. >'(x tan X + lny) dx + tan x = 0. ans. sin x In j = x cos x — sin x + c
fl(y

20. xy' — y = x'^y", where n ^ and + « ^ 1 A- 1

ANS. + - \)y'-"=^ (1 - + ex''" (A- /; /;)x*^

21. The equation of Ex. 20 for the values of A and n not included there.
ANS. If // = and A r^ 0, x' = A In
1 \cy/x
U n = and k ^0,y = cx^.
1

If/? 7^ 1 but k + n= l,>;i-" = (l -/0-x:'""ln|cx

In Exs. 22-27, find the particular solution required.

22. 4(3x + - 2) ^x - (3x +


>; J') ^>'
= 0; when x = 1, >; = 0.
21x + 7j-8
ANS. 7(4x — J — 4) = 8 In
13
23. y' = 2(3x + yY — \; when x = 0,y= 4 Arctan (3x + y) = 8x
1 . ans. + tt.

24. 2xyy' = y^ — 2x^. Find the solution that passes through the point (1, 2).
ANS. y^ = x(5 — x^).
25. (>•* — 2x>') dx +3x^ dy = 0; when x = 2, j = 1 ans. x^ = :>/^(x + 2).
26. (2>'^ — x^) dx + 3xy^ dy = 0; when x = \,y= 1. Solve by two methods.
ANS. 5x^y^ = x^ + 4.
27. (x^ + 6y^) dx — 4x>' dy ^ 0; when x = I, y = 1. Solve by three methods.
ANS. 2y^ = x^(3x — 1).
§ 22] Coeflicients lineur in the two variables 73

22. Coefficients linear in the two variables

Consider the equation

(fljA- + biy + Ci) dx + {a2X + b2y + C2) dy^Q), ( 1

in which the «'s, 6's, and t"'s are constants. We know already how to solve
the special case in which Ci = and C2 — 0, for then the coefficients in (1)
are each homogeneous and of degree one in x and y. It is reasonable to
attempt to reduce equation (1) to that situation.

In connection with (1) consider the lines

a^x + b^y + fi = 0, (2)

a2X -{- b2y -\- C2 =0.

They may be parallel or they may intersect. There will not be two lines if a^

and b^ are zero, or if aj and 62 ^re zero, but equation (1) will then be hnear
in one of its variables.

If the lines (2) intersect, let the point of intersection be (/?, k). Then the
translation

X — u + h, (3)

y — V -[ k

will change the equations (2) into equations of lines through the origin of the
uv coordinate system, namely,

+ Z)ii' = 0,
fliW (4)

a2U + b2V — 0.

Therefore, since ^.v = du and dy = dv, the change of variables

X = u + h,
y = v + k,
where (//, A) is the point of intersection of the lines (2), will transform the
differential equation (1) into

(ajW + biv) du + (^2 w + ^2 ^) <^^' — 0» (5)

an equation which we know how to solve.

If the lines (2) do not intersect, there exists a constant k such that

a2X + b2y = kia^x + b^y),

so that equation (1) appears in the form

{a^x + b^y + Ci) dx + [kia^x + b^y) + C2] dy = 0. (6)


74 Additional Topics on Equations of Order One [Ch. 4

The recurrence of the expression (a^x + biy) in (6) suggests the introduction
of a new variable u' = OiX + biy. Then the new equation, in u' and .v or in
w and y, is one with variables separable, since its coefficients contain only

w and constants.

EXAMPLE (a): Solve the equation

(x + 2y - 4) dx - (2a- +y-5)dy = 0. (7)

The lines

X + 2y - 4 = 0,

2x+y-5 =0
intersect at the point (2, 1). Hence put

X = u + 2,
j = r+l.
Then equation (7) becomes

(w + 2v) du - {2u + v) dv = 0, (8)

which has coefficients homogeneous and of degree one in u and t'. Therefore
let u — vz, which transforms (8) into

(_- + 2)(_- dv + V dz) - {2z +\)di=Q,


or
(r'- l)^r+r(r + 2)^/.- = 0.

Separation of the variables r and z leads us to the equation

dv (z + 2) dz
V z^-\
With the aid of partial fractions, we can write the above equation in the form

Hence we get
—+
2 dv
—— 3 dz
-
dz

2 1n|r|+31n|j- 1| - In |r + 1| = ln|ci

from which it follows that

v\z - \)' = c{z + 1),

or
{vz — v)^ = c(vz + v).

Now vz = w, so a set of solutions appears as

(u — vY = c(u + v).
§ 22] Coefficients linear in the two variables 75

But u =X— 1 and v =y— 1. Therefore the desired result in terms of a: and y
is

{x-y-\f = c{x + y-Z).


For other methods of solution of equation (7), see Exs. 23 and 30 below.

EXAMPLE (b): Solve the equation

{2x -\-ly-\)dx + {2x + 3y + 2) dy = 0, (9)

with the condition that y = 3 when x = 1

The lines

2x + 3y-\=0
and

2x + 3>; + 2 =
are parallel. Therefore we proceed, as we should have upon first glancing at
the equation, to put
2x + 3y = V.

Then 2 dx = dv — 3 dy, and equation (9) is transformed into

(v - l)idv -3dy) + 2{v + 2) dy = 0,

or

{v -\)dv-iv- 7) dy = 0. (10)

Equation (10) is easily solved, leading us to the relation

y -;; + c + 6 In |y - 7| =0.

Therefore a solution set of (9) is

2x + 2y + c= -6 In \2x + 3y - 7|.

But y — 3 when x — \, so c = —8 — 6 In 4. Hence the particular solution


required is

x+y-4= -3\n [i(2x + 3y - 7)].

Exercises
In Exs. 1-18 solve the equations.

1. iy-2)dx-{x-y-l)dy^0. ans. x - 3 = (2 ->') In 1c(>'- 2)1.


2. (x - 4;^ - 9) dx + i4x + y- 2) dy - 0.

-
x-l
ANS. In [{x 1)^ + (>' + 2)2] - 8 Arctan = c.
y +2
76 Additional Topics on Equations of Order One [Ch. 4

3. (2x -y)dx + i4x + y- 6) dy = 0. ans. (x +y- 3y = c{2x + j - 4)^


4. (.V - 4>' - 3) ^x - (x - 6J - 5) dy = 0.
ANS. (x-2y—iy^c(x—3y — 2).
5. (2x+ S^' - 5) + (3a: - - 2) dy = 0. Solve by two methods.
rfjf >'

6. ldx + {2x — y+3i)dy = 0. Use a change of variable.


ANS. y + c= — ln|2A: — + 4|. >'

7. Solve the equation of Ex. 6 by using the fact that the equation is linear in x.
8. {x -y+ 2) dx + Z dy = 0. ans. a- +c= 3 In |x - + >' 5|.

9. Solve Ex. 8 by another method.


\Q. {x-^y-\)dx + 0.x + 2y+\)dy = Q. ans. x + 2y + c = 7>\n\x + y + 2\.
11. (3a: + 2^ + 7) ^jc + (2a: — y) dy = 0. Solve by two methods.
12. {x-2)dx + A{x + y-\)dy = 0.
ANS. 2(^+1)= -(A: + 2>')ln|c(x + 2>')|.
13. {x-7>y + 2) dx + 3(a: + 2,y - A) dy = 0.
X- 1
ANS. In [(a- - 1 )^ + 9(y - 1 )^] - 2 Arctan = c.
Xy- 1)
14. (6a: - 3;^ + 2) ^a: - (2a: - - j; 1) ^>' = 0.
ANS. 3Ar — +c=
>' 5 In |2Ar — + 4|.
>'

15. (9a: - 4>' + 4) ^a: - (2x - + ^^ 1) J^' = 0.


ANS. J'- 1 = 3(>'-3a:- l)ln|c(3A:->'+ 1)|.

16. (a: + 3^ - 4) ^a: + (a: + 4^ - 5) dy = 0.


ANS. >'- = (x + 2;'- 3)ln |c(a: + 2>'-3)|.
1

17. {x + 2y - \) dx - (2x + y - 5) dy =- 0. - y - AY = c{x + y - 2).


ans. {x
18. (a:- 1)^x-(3a:-2>'-5)^>' = 0. ans. {2y - x + 7>y = c{y - x + 2).

In Exs. 19-22 obtain the particular solution indicated.

19. (2a: -y + A)dx-^3(,x-\)dy = 0; when x^7,,y = 2.

X- 1
ANS. 3(^-2)= -2(a:- 1) In——.
20. The equation of Ex. 19 but with the condition: when x = —\,y =2.
1 -X
ANS. 3(j-2)= -2(a:- l)ln—^.
21. (a: + — 4) dx — Ox — y — A) dy = Q\
>' when a- = 4, > = 1 .

ANS. 2{x + 2^ — 6) = 3(x — y) In ———y


X
-

22. The equation of Ex. 21 but with the condition: when x = Z,y=l.
—X
ANS. >' — 5x +8= 2{y — x) In y—-—
23. Prove that the change of variables
X = cxiu -\- (XiV, y —u -\- V

will transform the equation

{a,x + b,y + c,)dx + {a2X + bi y + Cj) dy = (A)

into an equation in which the variables // and v are separable, if a, and a^ are
roots of the equation

flia' + (o2 + /),)« + 62 = 0, (B)


§ 23] Solutions involving nonelementary integrals 77

and if a2 ^ ai.
Note that this method of solution of (A) is not practical for us unless the roots
of equation (B) are real and distinct.
Solve Exs. 24-29 by the method indicated in Ex. 23.

24. Ex. 4 above. As a check, the equation (B) for this case is

a^ - 5a + 6 - 0,

so we may choose a, = 2 and 0.2 = 3.


The equation in u and v turns out to be
{v- \)du-2iu + 2)dv = 0.

25. Ex. 3 above. 26. Ex. 10 above.


27. Ex. 17 above. 28. Ex. 18 above.
29. Example (a) in the text of this section.
30. Prove that the change of variables
X = (Xiu + ^v, y=u+v
will transform the equation

(oix + biy + Ci) dx + (azx + b2y + C2) dy = (A)


into an equation that is linear in the variable u, if ai is a root of the equation

a,oc' + ia2 + b,)<x + b2 = 0, (B)

and if ^ is any number such that ^ ^ oci.

Note that this method is not practical for us unless the roots of equation (B)
are real. But they need not be distinct as they had to be in the theorem of Ex.
23. The method of this exercise is particularly useful when the roots of (B) are
equal.

Solve Exs. 31-35 by the method indicated in Ex. 30.

31. Ex. 16 above. The only possible ai is ( — 2). Then /3 may be chosen to be any-
thing else.

32. Ex. 12 above. 33. Ex. 15 above. 34. Ex. 18 above.


35. Ex. 4 above. As seen in Ex. 24, the roots of the "a-equation" are 2 and 3. If

you choose ai = 2, for example, then you make ^ anything except 2. Of course,
if you choose ai = 2 and jS = 3, then you are reverting to the method of Ex. 23.

23. Solutions involving nonelementary integrals

In solving differential equations we frequently are confronted with the


need for integrating an expression that is not the differential of any elementary*

* By an elementary function we mean a function studied in the ordinary beginning


calculus course. For example, polynomials, exponentials, logarithms, trigonometric, and
inverse trigonometric functions are elementary. All functions obtained from them by a
finite number of
applications of the elementary operations of addition, subtraction, multi-
plication, division, extraction of roots, and raising to powers are elementary. Finally, we
include such functions as sin (sin x), in which the argument in a function previously classed
as elementary is replaced by an elementary function.
78 Additional Topics on Equations of Order One [Ch. 4

function. Following is a short list of nonelementary integrals:

^
.2, - '

exp (-x^) dx I
dx x tan x dx

. 2 J
sin A- C dx
sin X dx I
dx
X J In X

, I
cos X C dx
cos x" dx flx

Integrals involving the square root of a polynomial of degree greater than


two are, in general, nonelementary. In special instances they may degenerate
into elementary integrals.
The following example presents two ways of dealing with problems in

which nonelementary integrals arise.

EXAMPLE: Solve the equation

y' - 2xy = 1

with the initial condition that when x = 0, y = 1.

The equation being linear in we write


)',

^V — 2xy dx = dx,

obtain the integrating factor exp ( — x^), and prepare to solve

exp ( — x^ ) dy — 2xy exp ( — x^) ^x = exp ( — x^ ) dx. ( 1

The leftmember is, of course, the differential of >'exp( — x^). But


the right member is not the differential of any elementary function; that is,

J exp ( — x^) dx is a nonelementary integral.


Let us turn to power series for help. From the series

(-l)V"
exp(-xO= L
« =
-
n

1

obtained in calculus, it follows that

r
exp(-x') dx = c+}^
7 ^ (-\)"x^"*'
.

J „ =o nl{2n + 1)

Thus the differential equation (1) has the general solution

>'exp(-x'-) = c+ X -TTT—TV-
„ = o n\{2n + 1)

Since v = 1 when x — 0,c may be found from


1 = c + 0.
§ 23] Solutions involving nonelementary integrals 79

Therefore, the particular solution desired is

00 \n„2n+
(—\Yx' 1

„ = n (2n
! + 1

An alternative procedure is the introduction of a definite integral. In


calculus, the error function defined by

erf X = 4- f
""exp ( - /?') cip (3)

is sometimes studied. Since, from (3),

— - erf X = —^ exp ( — x ),
^^ Vtt

we may integrate the exact equation (1) as follows:

y exp ( — x^) — ^^n erf x + c. (4)

Since erf = 0, the condition that y = 1 when x = yields c = 1. Hence, as


an alternative to (2) we obtain

j^exp(-x^) = 1 + ^^/tt erf X. (5)

Equation (5) means the same as

j; exp ( - x') - 1 + f "exp ( - j?^) dp. (6)

Writing a solution in the form of (6) implies that the definite integral is to

be evaluated by power series, approximate integration such as Simpson's


rule, mechanical quadrature, or any other available tool. If it happens, as in
this case, that the definite integral is itself a tabulated function, that is a great
convenience, but it is not vital. The essential thing is to reduce the solution
to a computable form.

Exercises
In each exercise, express the solution with the aid of power series or definite
integrals.

1. y' = y[l — exp(—x^)]. ans. In |c>'| = x— ^VTrerf x.


r'' sin w
2. (xy — sin x) dx + x^ dy ^ 0. ans. xy =^ c + dw,
Jq w
00 V^"
or>^=cx-'+2/-l)"(2,+
i)(2„+i)r

3. / = - 4x^y.
1 ans. j^ = exp (— x*) c + j
exp (/3^) dp
80 Additional Topics on Equations of Order One [Ch. 4

4. (y cos^ X — x sin x) dx + sin x cos x dy = Q.


X

ANS. >» sin X =c+ /3 tan /3 d^.

5. (1 + xy) dx — X dy = 0; when x = ],y = 0. ans. j = e"^ ^ ^iS.

A- exp c^x + X (/y = ; when x = I, y = 2.


g) 2

ANS. lnx= e\p{—^^)d^.


y/x
1. x{ly + x)dx — dy = Q\ when x = Q,y= 1

ANS. 2J = 2 exp (x^) —X+ ^Vtt exp (x^) erf x.

Miscellaneous exercises
In each exercise, find a set of solutions unless the statement of the exercise
stipulates otherwise.

1. (7^ - 3>' - x) ^x + (2^ - 3) ^j' = 0. ANS. y^-3y- x+\ = ce-\


2. {.y^ +y+ 1) dx + x(x — 2>y^ — \)dy = 0. ans. y^ — xy y + = ex. -\- \

3. (x + 3>' - 5) Jx - (X - J - 1) dy = 0.
ANS. 2(>'- l) = (x + j-3)ln|c(x + >;-3)|.
4. (x^ — y^) dx + 2x>' dy = 0. ans. x' + Ay^ = ex.
5. {Ix + y- 4) dx+{x-^y+\l)dy = 0.
ANS. 2x2 _^ 2xy — 33^2 - 8x + 24^ = c.
6. Solve in two ways the equation y' = ax + by + c with b ^Q.
ANS. = Cic'"' — a6x — a — cb. /j^j'

7. j^ sec^ X ^x — (1 — 2j2 tan x) dy = 0. ans. y^ tan x = In |f>'|.

8. x^y dx + (3x* - y^) dy = 0. ans. ISx^^''^ = 4y'' + c.


9. (aix + ky + c) c/x + (A:x + /^z^' + Cj) dy = 0.
ANS. OiX^ + 2A'xj + biy^ + IciX + Iczy = c.
10. (x - 43^ + 7) ^x + (x + 2j + 1) ^^^ = 0.
ANS. {x-y + AY = c(x - 2;; + 5)^
11. xy dx + {y* - 3x^) dy = 0. ans. x^ = >'^(1 + cy^).
12. (x + 2^ - 1) c/x - (2x + 7 - 5) dy = 0.
13. (5x + 3eO t/x + 2xe^ dy = 0. ans. x^(x + e^y = c.
14. {3x + y- 2) t/x + (3x + J + 4) dy = 0. ans. x + j;+c=3 1n|3x + + 7|. >'

15. (x-37 + 4)^x + 2(x->^-2)i/y-0. ans. (x + - 8)-^ = r(x- 23^ + 1). >;

16. (x -2)dx + 4ix + y- 1) dy = 0.


ans. 2(>'+ 1)= -(x + 2>')ln|c(x + 2>')|.
17. J c/x = x(l + x>'*) dy. ans. j(5 + xj^*) = ex.
18. 2x^y + y(2 + i'2jc)(/x = 0; whenx= 1, y = i. ans. xy^CSx— 1)=1.
19. 2(x->')^x + (3x->'- l)f/v = 0. ANS. (x + >'- ir = c(4x-2>'- 1).
20. (2x -5y+ 12) dx + (7x -4y+\5)dy = 0.
ANS. (x + 2^ — 3)^ = c(x — + 3). J'

21. J'^x + x(x2>' - \)dy 0. -= ANS. >'2(2x2>' - 3) = cx^.


22. ^^'/(/x = tan y cot x — sec j cos x. ans. sin + sin x In |c sin x| = 0. 3^
Miscellaneous exercises 81

23. [1 + (X + yy] dx+[l+ x(x + y)] dy = 0.


ANS. y^ = {x-\- yY -\-2\n\x + y\-\r c.
24. {x — 2y—\)dx — {x—2))dy = 0. Solve by two methods.
ANS. {x — 3)^(x — 37) = c.
25. {lx-Zy+ \)dx- (3x + 2y - 4) dy ^ 0. Solve by two methods.
ANS. x^ +X— 3xy — y^ + 4y = c.

26. (4;c + 3y-7)dx + (4x+3y+l)dy = 0.


ANS. x + + >' c-81n|4jc + 3;^ + 25|.
27. Find a change of variables that will reduce any equation of the form

xy' = yfixy)
to an equation in which the variables are separable.
28. (x + 4y + 3) dx - {2x -y- 3) dy = 0.
ANS. 3(>'+l) = U + >')ln|c(x + >')l.

29. i3x-3y-2)dx-(x~y+\)dy = 0.
ANS. 2(y — 3x + c) = 5 In |2x — 2>' — 3|.

30. ix-6y + 2) dx + 2(x + 2;^ + 2) ^;^ = 0.

ANS. 4y = — (x — 27+2) In \c(x — 2y + 2)|.


31. (x* — 4x^y^ — y*) dx + 4x^y dy = 0; when x= l,y = 2.
ANS. >'^(5 - 3x) = x\5 + 3x).
32. ix-y-l)dx-2iy-2)dy = 0. ans. (x + - 5)'(x- 2^+ 1) = c.
>'

33. {x-3y + 3) dx + i3x + y + 9) dy - 0.

ANS. In [(x + 3)2 + = c + 6 Arctan [(x + 3)ly]. ;^;2]

34. (2x + 4y- l)dx-ix + 2y-3)dy = 0. ans. In |x + 2;^ - = - 2x + c. 1| >'

35. 4y dx + 3(2x - l)(dy + y* dx) = 0; when x - 1, = 1. >;

ANS. >'3(2x-l)(5x-4)=l.
36. Kx- — (x^— 2x-2;')^>' = 0.
l)i/x ans. x^ - 2x - 4^ = cj'^.
37. i6xy - 3y^ + 2y) dx + 2(x - y) dy = 0. ans. >'(2x -y) = ce' ^^
38. >''= X — + 2. Solve by two methods.
>» ans. ln|x — >'+1| = c — x.
39. (x + - 2) ^x - (x - 4>' - 2) ^>' - 0.
>>

x-2
ans. In [(x — 2)2 + 4^^] + Arctan —-— = c.
^y
x-y + 2
40. 4 ^x + (x - J + 2)2 dy = 0. ans. y + 2 Arctan = c.
CHAPTER 5
Linear Differential Equations

24. The general linear equation

The general linear differential equation of order n is an equation which


can be written

The functions R{x) and 6,-(-y); / = 0, 1, .. .


, /?, are to be independent of the
variable y. If R{x) is identically zero, equation (1) is said to be linear and
homogeneous*; if R{x) is not identically zero, equation (1) is called linear

and nonhomogeneous. In this chapter we shall obtain some fundamental


and important properties of linear equations.

First we prove that if j^^ and yj are solutions of the homogeneous equation

boix)/"' + b,{x)/"-'^ + •••


+ b„_,{x)y' + b„{x)y = 0, (2)

* It is perhaps unfortunate that the word homogeneous as it is used here has a very
different meaning from that in Sections 9 and 10.

82
§ 25] Linear independence 83

and if q and C2 are constants, then

y = c^y^ + C2y2

is a solution of equation (2).

The statement that y^ and yj are solutions of (2) means that

^0/1"^ + b,yr'^ + + b„.,y\ + b„y, = (3)

and

bo/2' + b,/r'' + ---


+ b„.,y'2-^ b„y2 = 0. (4)

Now let us multiply each member of (3) by c^, each member of (4) by C2 , and
add the results. We get
•••
bolcyy^ + C2yri + b,lc,y["-'^ + €2/2"' ''li +
+ b„-ilciy[ +C2y'2l + b„[_Ciy, +C2>'2] =0. (5)

Since c^y'i + C2y'2 = (ciy^ + C2>'2)', etc., equation (5) is neither more nor
less than the statement that c^y^ + f2>'2 is a solution of equation (2). The
proof is completed. The special case c 2 = is worth noting; that is, for a
homogeneous linear equation any constant times a solution is also a solution.
In a similar manner, or by iteration of the above result, it can be seen that
if j^,; / = 1, 2, . .
.
, A:, are solutions of equation (2), and if c^; / = 1, 2, . .
.
, A:,

are constants, then

y = Ciyi + c2y2 + •• + c^yk (6)

is a solution of equation (2).

The expression in equation (6) is called a linear combination of the functions

yi> yi^ •••> yn- The theorem just proved can thus be stated: Any linear
combination of solutions of a linear homogeneous differential equation is also a
solution.

25. Linear independence

Given the functions /i, ...,/„, if constants Ci, C2 , . .


.
, c„ , not all zero, exist
such that

cMx) + C2f2{x) + • • •
+ cj„{x) = (1)

identically in some interval a ^x^ b, then the functions/1,/2 ,...,/„ are said
to be linearly dependent. If no such relation exists, the functions are said to
be linearly independent. That is, the functions f, f2, •••,/„ are linearly
independent when equation (1) implies that q= C2 = ••• = c„ = 0.
84 Linear Differential Equations [Ch. 5

It should be clear that if the functions of a set are linearly dependent, then
at least one of them is a linear combination of the others; if they are linearly
independent, then none of them is a linear combination of the others.

26. An existence and uniqueness theorem

In Section 12 we stated an existence theorem for an initial value problem


involving a first-order linear differential equation. The generalization of this
theorem to nth order linear equations can be stated as follows:

Let P^, Pj, ..., P„, and R be functions that are continuous on an interval
a < X < b. Suppose that Xq is a real number in the interval and that j'q, yy,

. . ., y„-i are n arbitrary real numbers. Then there exists a unique function
y = y(x) defined on the interval a <x< b, which is a solution of the equation

/"^ + P,/"-'^ + ---+P„y = R (1)

on the interval and which satisfies the initial conditions

EXAMPLE: Find the unique solution of the initial value problem

/'+>' = 0,>^(0) = 0,y(0)=l. (2)

We observe that sin x and cos .v are solutions of (2), so that for arbitrary
Cj and Cj

y = (?! sin X + f 2 cos X

is also a solution by the theorem of Section 24.

Because of the initial conditions in (2), we are led to choose Cj and Cj so

that Ci sin + <:2 cos = and c j cos — f2 sin = 1. This can be done in
only one way, namely by choosing" Cj — 1 and Ci = 0. We find that the

function sin x is a solution of the initial value problem (2). Moreover, since
the problem satisfies the conditions required in the foregoing theorem, no
matter what interval we choose, sin x is the only solution to the problem
given in (2).

27. The Wronskian

With the definitions of Section 25 in mind, we shall now obtain a sufficient


condition that n functions be linearly independent over an interval

a<x<b.
§27] The Wronskian 85

Let us assume that each of the functions /i,/2 ••••./„ is differentiable at


,

least (n — \) times in the interval a ^x


^ b. Then from the equation
Ci/i +C2f2 + "- + cJ„ = 0, (1)

it follows by successive differentiation that

cJl + C2 fz + •

r(n-l)
cjr'' + c2fr''
'2 + "T"

+ (^nfn = 0.

Considered as a system of equations in Cj, C2 , . .


.
, c„ , the n linear equations
directly above will have no solution except the one with each of the r's equal
to zero, if the determinant of the system does not vanish. That is, if

/l
86 Linear Differential Equations [Ch. 5

for all t. Indeed, one set of such constants is Cj = sin a, Cj = cos a, ^3 = — 1.


One of the best-known sets of n linearly independent functions of x is the
set 1, -v, x^, . . ., .v"~'. The linear independence of the powers of .y follows at
once from the fact that \{ c\, Cj, . , c„ are not all zero, the equation

C, + f2-Y+ •••
+ C„X""^ =0
can have, at most, {n — I) distinct roots and so cannot vanish identically in

any interval. See also Ex. 1 below.

Exercises
1. Obtain the Wronskian of the functions

\, X, x^, .. .
, x""' for « > 1.

ANS. W^= 0!1! 2! •••(«- 1)!.

2. Show that the functions e"", e^", e^" are linearly independent.
ANS. W^le^^^O.
3. Show that the functions e", cos x, sin x are linearly independent.
ANS. W=2e'' ¥- 0.

4. By determining constants Ci, d, d, C4, not all zero and which are such that
Ci /i + C2 /2 + C3 /s + C4 /4 = identically, show that the functions
/i = x,f2 = e\ fi = xe\ = {l — 3x)e' f4.

are linearly dependent.


ANS. One such set of c's is: Ci = 0, C2 = — 2, C3 = 3, C4 = 1.

5. Show that cos(a)/ ~ /S), cosa»/, sin ojt are linearly dependent functions of /.

6. Show that 1, sin x, cos x are linearly independent.


7. Show that 1, sin^ jc, cos^ x are linearly dependent.
8. Show that two nonvanishing differentiable functions of x are linearly dependent
if,and only if, their Wronskian vanishes identically. This statement is not true
for more than two functions.
9. Let /:(x) = + x^ for ^ 0, fi{x) =
1 a: 1 for jc ^ 0;
f2{x) = for X ^ 0, /aU) = + x^
1 1 for X^ 0;
/3(x) = 3 + jc^ for all x.
Show that (a) /, /',/" are continuous for all x for each of /i, /z , /j;
(b) the Wronskian of /i, /z , /a is zero for all x\
(c) /i,/2,/3 are linearly independent over the interval — ^ 1 x ^ 1.

In part (c) you must show that if Ci /i(x) + Ci fzix) + c 3 fiix) = for allx in
— ^ 1 X^ 1, Ci = C2 = C3 = 0. Use a: = — 1, 0, 1 successively to obtain three
equations to solve for Ci, Cz, and C3

10. Given any interval a <x <b with Xo a fixed number in the interval and suppose
>' is a solution of the homogeneous equation
/' + Py' +Qy = 0. (A)
§ 28] General solution of a homogeneous equation 87

Further, suppose that y(xo) = y'(xo) = 0. Use the existence and uniqueness
theorem of Section 26 to prove that y(x) = for every x in the interval
a <x <b.
11. Suppose that ji and yz are solutions of equation (A) of Ex. 10 and suppose the
Wronskian of yi and J2 is identically zero on a <x <b. Show that for Xo in
the interval a <x <b, there must exist constants c and Cz not both zero such 1

that

ciyi(xo) + Ciyiixo) = 0,
and

Ci>^i(-^o) + Ciy'zixo) = 0.
12. Consider the function defined by

y(x) = Ciyi{x) + czyzix),

where Ci and f 2 are the constants determined in Ex. 11. Show that this function
is a solution of equation (A) above and that it follows that y(x) = on a < x < b.
(Use the results of Ex. 10.)

13. Combine the results of Exs. 10-12 to obtain a proof of the necessity condition
of Theorem 4 in the case when n = 2. Also note that the sufficiency condition
was established in the text of this section.

28. General solution of a homogeneous equation

Let yi, y2, ••, yn t>e linearly independent solutions of the homogeneous
equation

bo(x)/"' + b,{x)/"-'' + + b„_,(x)y' + b„(x)y = 0. (1)

Then the general solution of equation (1) is

y = ciyi + C2y2 + " + c„y„, (2)

where Cj , ^2 , . .
.
, c„ are arbitrary constants.

In a sense each particular solution of the linear differential equation (1) is

a special case (some choice of the c's) of the general solution (2). The basic
ideas needed for a proof of this important result are exhibited here for the
equation of order two. No additional complications enter for equations of
higher order.
Consider the equation

boix)y" + b,{x)y' + b2{x)y = 0. (3)

Let the function

y=fix) (4)
88 Linear Differential Equations [Ch. 5

be any particular solution of equation (3), the solution being valid in some
interval a <x< b in which ^o(-^') does not vanish and in which the functions
bijbo and /^a/^o possess derivatives of all orders.

We assume (it can be proved) that at x = a, within the stipulated interval,

f(x) is sufficiently well behaved to insure that the Taylor series for/(.v),

- /""(a)(.x - ^f
/(^) = L= Ti ' (^^
fc
K-

converges to/(.v) in some interval about .v = a.

Then the solution / (a) is completely determined by knowledge of /(a) and


of/ and/' taken at some point a within the interval. For,
/'(a), the values

we can divide by /?o(-v), equation (3) yields v" in terms of j, v', and the
since
known coefficients in the equation. Then y"iyi) =f'\i) can be determined.
Successive differentiation of the equation for y" will yield the higher deri-
vatives. Thus P''\oc) can be computed for k = 2, 3, . .
.
, from the values of
/(a) and/'(oc). Then, using Taylor's series (5), we see that/(.Y) is determined
by the differential equation and the two values /(a) and /'(a).
Now let>'i(.v), yjix) be any two linearly independent solutions of equation
(3). Further, let a be chosen as some point in the interval, a < oc < b, where
the Wronskian of >'i
and J2 does not vanish.

>'i(a) >'2(a)
^ 0. (6)
>''i(a) y'lici)

Consider the solution

y = ciyiix) + C2y2(x), (7)


from which
y' = Ciy'i{x) + C2y2{x).

We wish to choose c\ and C2 so that the v of equation (7) will become the
solution /(x). But we can surely force that v to be such that v(a) —fii) and
y(a) =/'(a). That is, the equations

Ci>'i(a) + C2>'2(a)=/(a), (8)

Ci>''i(a) + C2 3'2(a)=/'(a),

can be solved for Cj and C2 because of the inequality (6), the nonvanishing of
the Wronskian. Then/(.Y) and y{x) have the same initial values, so that by
the uniqueness theorem of Section 26 they are identical.
It is necessary to keep in mind that the above discussion used the fact
that ^oW 7^ '" ^^^ interval a < x < b. It is easy to see that the linear
equation
xy' -2y =
§ 29] General solution of a nonlioinogeneous equation 89

has the general solution y = cx^ and also such particular solutions as

y, ^x\ ^ X,
= -4;c^ X <0.

The solution yi is not a special case of the general solution. But in any
interval throughout which bQ{x) =x^ 0, this particular solution is a special
case of the general solution. It was, of course, made up by piecing together at
X = two parts, each drawn from the general solution.

29. General solution of a nonhomogeneous equation

Let yp be any particular solution (not necessarily involving any arbitrary


constants) of the equation

bo/"' + b,/"-'' + •••


+ b„^,y' + b„y = R{x) (1)

and let y^ be a solution of the corresponding homogeneous equation

bo/"' + b,/"-'' + • • •
+ b„^,y' + b„y = 0. (2)
Then
y = yc + yp (3)

is a solution of equation (1). For, using the y of equation (3) we see that

ho/"^ ^••• + b„y = {bo/^^-¥--- + b„y,)

+ (bo/;' + --- + b„yp) = + R(x) = R(x).

U y^, y2, ..., y„ are linearly independent solutions of equation (2), then

3'c
= ^iyi + ^2^2 + -**
+ c„j'„, (4)

in which the c's are arbitrary constants, is the general solution of equation (2).

The right member of equation (4) is called the complementary function for

equation (1).

The general solution of the nonhomogeneous equation (1) is the sum of


the complementary function and any particular solution. To justify this usage
of the term "general solution" we must show that if/ is any solution of
equation (1) then f=y^ + yp for some particular choice of the c^, .., c„. .

We note that since /and yp are both solutions of the nonhomogeneous equa-
tion (1),/— >'p is a solution of the homogeneous equation (2). Hence by the
theorem of Section 28

f-yp = cx}\ + C2>^2 + ••• + f«7«

for some particular choice of the Cy, . . ., c„. This establishes what we wished
to show.
90 Linear Differential Equations [Ch. 5

EXAMPLE (a): Find the general solution of


y" = 4. (5)

We first observe that the functions 1 and .y are linearly independent on any
interval and are solutions of the homogeneous equation y" = 0. Hence, the
complementary function for equation (5) is

yc = Ci + C2 X.

On the other hand, the function 2x^ is a particular solution of equation (5).

Hence, the general solution of equation (5) is

y ^ Ci + C2.V + 2.Y^.

EXAMPLE (b): Find the general solution of the equation


/'->• = 4. (6)

It is easily seen that y = —4 is a solution of equation (6). Therefore the yp in


equation (3) may be taken to be ( — 4). As we shall see later, the homogeneous
equation
y"-y = o
has as its general solution

y^ = Cie'' + C2e-\
Thus the complementary function for equation (6) is c^e"" + Cje'"" and a
particular solution of (6) is yp = — 4. Hence the general solution of equation

(6) is

y = c^e"" + C2e "" — 4,

in which c^ and C2 are arbitrary constants.

30. DiflFerential operators

Let D denote differentiation with respect to a', D^ differentiation twice

with respect to a-, and so on that ; is, for positive integral k.

D'y = ^,-
The expression
A=aoD" + a,D"-' + --'
+ a„_,D + a„ (1)

is called a differential operator of order n. It may be defined as that operator


which, when applied to any function* y, yields the result

d"y d"-'y dy
^'^
^'^'''d?'-'^d7^^^---^'"-^d-x^'"'-
* The function y is assumed to possess as many derivatives as may be encountered in
whatever operations take place.
§ 30] Differential operators 91

The coefficients Oq, a^, ..., a„m the operator A may be functions of .v, but
in this book the only operators used will be those with constant coefficients.
Two operators A and B are said to be equal if, and only if, the same result
is produced when each acts upon the function y. That is, ^ =^ if, and only
if, Ay = By for all functions y possessing the derivatives necessary for the
operations involved.
The product AB of two operators A and B is defined as that operator which
produces the same result as is obtained by using the operator B followed by
the operator A. Thus ABy — A(By). The product of two differential operators
always exists and is a differential operator. For operators with constant
coefficients, but not usually for those with variable coefficients, it is true that
AB = BA.

EXAMPLE (a): Let ^ = Z) + 2 and 5= 3Z) - 1.

Then

By = 0D-\)y = 3^-y
ax

and

A{By) = {D-\-2)(2,^^-y^

d^y dy ^dy
= 3-^--^_ + 6^_-2j ^

dx^ dx dx


= ^-6. +
dx^
+ 5^
^^-^y
dx

= (3D^ + 5D- 2)y.

Hence AB = {D + 2)(3Z) - 1) = 3D^ + 5D-2.


Now consider BA. Acting upon y, the operator BA yields

B(Ay) = (3D
-'S-^)
d^y ^dy dy
= ^-A + 6-r.--r.-2y
dx^ dx dx

d^y ^dy
= 3-^ +
dx^
5^-2j
dx

Hence

BA = 3D^ + 5D-2 = AB,


92 Linear Differential Equations [Ch. 5

EXAMPLE (b): Let G = xD + 2, and H^ D- 1. Then

G(H>0 = (A-D + 2)(|£->'j

= X — X — + — — 2y
^-x
-7-3 2
dx'^ dx dx

^
= x-rj + i2-x)i^-2y.
(2-x)^'
dx^ dx
so
GH = xD^ + (2 - x)D - 2.

On the other hand

H(G>;) = (D-l)(x£ + 2>'j

dx \ dx J \ dx

d^y
^ .2 — -X —
dx^ dx dx dx

^
= x-r^ + (3
dx^
_ x) ^' -2y;
(3-x)—
dx
that is,

HG = xD^ + (3 - x)D - 2.

worthy of notice that here we have two operators G and


It is (one of them H
with variable coefficients), the product of which is dependent on the order of

the factors. On this topic see also Exs. 17-22 in the next section.
The sum of two differential operators is obtained by expressing each in the
form
GoD" + ayD"~^ + •••
+ a„_iZ) + fl„

and adding corresponding coefficients. For instance, if

A = 3>D^ - D+ x-2
and
B = x^D^ + 4Z) + 7,

then
A+B= + x^)D^ + 3D + A- + 5.

Differential operators are linear operators; that is, if A is any differential

operator, if c^ and c, ^^e constants, and if/i and/2 are any functions of x
each possessing the required number of derivatives, then

A{cJ^ + f 2 ,/2) = c^Af, + f 2 Af^ .


§ 31] The fundamental laws of operation 93

31. The fundamental laws of operation

Let A, B, and C be any differential operators as defined in Section 30. With


the above definitions of addition and multiplication, it follows that differential
operators satisfy the following
(1) The commutative law of addition:

A +B= B+ A.

(2) The associative law of addition:

{A + B) + C=A+ (B + C).

(3) The associative law of multiplication:

(AB)C = A(BC).

(4) The distributive law of multiplication with respect to addition:

A(B +C) = AB + AC.

If A and B are operators with constant coefficients, then they also satisfy

(5) The commutative law of multiplication

AB = BA.

Therefore, differential operators with constant coefficients satisfy all the laws
of the algebra of polynomials with respect to the operations of addition and
multiplication.
If m and n are any two positive integers, then

D'"D" = /)"+",

a useful result which follows immediately from the definitions.


Since for purposes of addition and multiplication the operators with
constant coefficients behave just as algebraic polynomials behave, it is

legitimate to use the tools of elementary algebra. In particular, synthetic


division may be used to factor operators with constant coefficients.

Exercises
Perform the indicated multiplications in Exs. 1^.

1. (4D +\)iD- 2). ANS. 4D^ -ID -2.


2. (2Z) - 3)i2D + 3). ANS. - 9.
4£>^
3. (D + 2)(D^ -2D + 5). ans. D^ D
+ + 10.
4. (D - 2)(£> + 1)^ ANS. D^-3D- 2.
94 Linear Differential Equations [Ch. 5

In Exs. 5-16, factor each of the operators.

5. 2D^^3D-2. ANS. (D + 2)i2D-\).

6. 2D^-5D- 12.
1. D'-2D^-5D + 6. ANS. (£)-l)(D + 2)(Z)-3).
8. 4D^-4D^- ll£> + 6.
9. D* - 4D^. ANS. D\D - 2){D + 2).
10. i)^-3£)^ + 4.
11. I>^ - 21 D + 20. ANS. (D - 1)(Z) - 4)(I> + 5).
12. 2D'- D^- 13/) -6.
13. 2D*+ 11Z>3+ 18Z)^ + 4Z)-8. ans. (D + 2)^(2Z> - 1).

14. 8D* + 36£)3 - 66/)^ + 35D - 6.


15. D* + D'- 2D^ + AD- 24. ans. (Z) - 2)(£> + 3)(Z)2 + 4).
16. Z)3 - Z) - 20.
1 1 ANS. (Z) - 4)(Z)^ + 4Z) + 5).
Perform the indicated multiplications in Exs. 1 7-22.

17. (Z) - x){D + x). ANS. Z>^ + - x^


1

18. (Z) + x){D - x). ANS. Z)^ - - x\


1

19. D(xD- 1). ANS. a:Z)^


20. (xD-l)Z). ANS.xD^-D.
21. ixD + 2)ixD- I). ANS. x^Z)^ + 2a:D - 2.
22. (xZ) - l)ixD + 2). ANS. x^Z)^ + 2xD - 2.

32, Some properties of dififerential operators

Since for constant m and positive integral k,

£,fc^mx
^ ^k^mx^
(1)

it is easy to find the effect an operator has upon e'"''. Let f{D) be a polynomial
in D,
f(D) = aoD'' + a,D"-' + ---
+ a„_,D + a„. (2)
Then
f{D)e""' =aQm"e'"' + a^nf'^e"''' + • • •
+ a„_i/ne'"-'' + a„e'"^
so
f{D)e'"' = e'"Y(»i). (3)

If m is a root of the equation /(w) = 0, then in view of equation (3),

/(Z^y"^ = 0.

Next consider the effect of the operator D—a on the product of e""" and a
function y. We have
(Z) - aWy) = D(e''y) - ae^y
= e'^'Dy,
§ 32] Some properties of differential operators 95

and

(£) - 0)^6"^) = {D- a)ie'"'Dy)

= e'^'D^y.

Repeating the operation, we are led to

{D - aTie'^'y) = e'"'D"y. (4)

Using the linearity of differential operators, we conclude that when f(D) is a


polynomial in D with constant coefficients, then

e'"f(D)y=f(D-a)le'"y'\. (5)

The relation (5) shows us how to shift an exponential factor from the left

of a differential operator to the right of the operator. This relation has many
uses, some of which we will examine in Chapter 6.

EXAMPLE (a): Let f(D) = 2D^ + 5D- 12. Then the equation /(w) =
is

2m^ + 5m - 12 = 0,
or
(m + 4)(2m - 3) = 0,

of which the roots are Wi = — 4 and m2 = f .

With the aid of equation (3) above it can be seen that

(2D^ + 5D- 12)^-*^ =


and that

(2Z)2 + 5D-12) exp (fx) = 0.


In other words, y^ = e"^'' and J2 = exp (f x) are solutions of

(2D^ + 5D- I2)y = 0.

EXAMPLE (b): Show that

(D - m)"(x''e"") = for k = 0, 1, ..., (n - 1). (6)

In equation (5) we let f(D) = {D — m)" and y = x*. Then using the ex-
ponential shift we obtain

But D"x'' = for A: = 0, \,2, . .. ,n — 1, which gives us equation (6) directly.


The results obtained in equations (3), (5), and (6) are of fundamental
importance to the solving of linear differential equations with constant co-
efficients which we consider in Chapter 6.
96 Linear Differential Equations [Ch. 5

EXAMPLE (c): As an illustration of the use of the exponential shift we


solve the differential equation

{D + 3fy = 0. (7)

First we multiply equation (7) by e^'' to obtain

e^'^iD + 3)V = 0.

Applying the exponential shift as in equation (5) leads to

D\e^y) = 0.

Integrating four times gives us

e^'^y = Ci + C2 X + C3 x^ + c^x^,

and finally,

y = {ci + C2X + c^x^ + c^x^)e'^''. (8)

Note that each of the four functions e~^^ xe~^^ x^e^^"", and .v^e"^^ is a
solution of equation (7). This of course is assured by the theorem of equation
(6) of Example (b).

If we now show that the four functions are linearly independent, equation
(8) gives the general solution of equation (7). See Ex. 5 below.

Exercises
In Exs. 1^ use the exponential shift as in Example (c) above to find the general
solution.

l,iD-2yy = 0. ANS. y = ict + C2X + CiX^)e^\


2. (D+\yy = 0.
3. (2D-iyy^ 0. ANS. y = {ci + C2 x) exp (^x).
4. {D + iyy = 0.
5. To show that the four functions in Example (c) above are linearly independent
on any interval, assume that they are linearly dependent, and show that this
leads to a contradiction of the results obtained in Ex. 1 of Sec. 27.
6. Prove that the set of functions

e"", xe"", x'^e"", . .


.
, x"~^e°''

is a linearly independent set on any interval. See Ex. 5.


CHAPTER O
Linear Equations

with Constant Coefficients

33. Introduction

Two methods for solving linear differential equations with constant


coefficients are presented in this book. A classical technique is treated in this
and the next chapter. Chapters 10 and 11 contain a development of the
Laplace transform and its use in solving differential equations. Each method
has its advantages and its disadvantages. Each is theoretically sufficient; both
are necessary for maximum efficiency.

34. The auxiliary equation; distinct roots

Any linear homogeneous differential equation with constant coefficients,

d"y d"-\ dy

97
98 Linear Equations with Constant Coefficients [Ch. 6

may be written in the form

f{D)y = Q, (2)

where f(D) is a linear differential operator. As we saw in the preceding


chapter, if m is any root of the algebraic equation f{m) = 0, then

f{D)e'"' = 0,

which means simply that y = e'"'' is a solution of equation (2). The equation

f{m) = (3)

is called the auxiliary equation associated with (1) or (2).


The auxiliary equation for (1) is of degree n. Let its roots be m^, nij , • •,

m„ . If these roots are all real and distinct, then the n solutions

yi = exp (m^x), y^ = exp (/7?2 x),...,y2 = exp {m„ x)

are linearly independent and the general solution of (1) can be written at once.

It is

y ^ Ci exp (myx) + C2 exp (m2 -y) + • • •


+ c„ exp ifn„x),

in which c^, C2, • • • , c„ are arbitrary constants.

Repeated roots of the auxiliary equation will be treated in the next section.

Imaginary roots will be avoided until Section 37, where the corresponding
solutions will be put into a desirable form.

EXAMPLE (a): Solve the equation

^_^
dx^ dx^
dy
dx

First write the auxiliary equation

m^ - Am^ + 777 + 6 = 0,

whose roots 777 = — 1, 2, 3 may be obtained by synthetic division. Then the


general solution is seen to be

y = qe"-^ + C2 e^"" + C3 e^^.

EXAMPLE (b) : Solve the equation

OD^ + 5D^ - 2D)y = 0.

The auxiliary equation is

3/71^ + 5m^ - 2777 =


§ 34] The auxiliary equation; distinct roots 99

and its roots are m = 0, — 2, ^. Using the fact that e°* = 1 , the desired solution
may be written

y = Ci+C2e~^'' + C3 exp i^x).

EXAMPLE (c): Solve the equation

—dr - 4x =
d'^x
2

with the conditions that when / = 0, :\r = and dxjdt = 3.

The auxiliary equation is

m^ -4 = 0,
with roots m = 2, —2. Hence the general solution of the differential equation
is

X = qe^' + €26''^'.

It remains to enforce the conditions at / = 0. Now

dt

Thus the condition that jc = when / == requires that

= Cj + C2,

and the condition that dx/dt = 3 when / = requires that

3 = 2ci - 2c2 .

From the simultaneous equations for Cj and C2 we conclude that q = | and


C2 = — |. Therefore

x = i(e''-e-''),

which can also be put in the form

X = f sinh (2/).

Exercises
In Exs. 1-22, find the general solution. When the operator D is used, it is implied
that the independent variable is x.

1. (D^ + 2D - 3)y = 0. ans. 7= Cie' + Cae"^'.


2. (Z)2 + 2D)y = 0.
3. (D^ + D-6)y = 0. ans. y = Cic^" + C2e-^\
4. (D^ -5D + 6)y = 0.
5. iD^ + 3D^-4D)y = 0. ans. y = Ci + Cze" + C3e-*''.
100 Linear Equations with Constant Coefficients [Ch. 6

6. iD^-3D^- lOD)y = 0.
7. (D^ + 6D^+ llD + 6)y = 0. ans. y = de'' + CzC-'''' + de-^".
8. (Z)^ + 3D^ -4D-\2)y = 0.
9. (4/)^ -7D+ 3)y = 0. ans. y = Cie" + Cz exp (^x) + c^ exp (- §x).
10. (4Z)^- UD-(>)y = 0.

— ^dx
2 — = 0. +,,,-,,
^. c/^jc f/^x ^
11. H ANS. jc = Ci C2 e' + C3 e- 2'.

12.^-19^ + 30;c = 0.
dt^ dt
13. (9£)^ -7D + l)y = 0. ans. >' = Cie-' + Cz exp {\x) + Ca exp (fjc).

14. {AD^ -2\D- 10)3^ = 0.


15. (D^- 14D + 8)>' = 0.
ans. j = Cie - *^ + C2 exp [(2 + V'2)x] + Cs exp [(2 - ^2)x].
16. {D^ - D^-AD- l)y = 0.
17. (4D* - 8Z)^ - + 6)>' = 0.
7Z)^ + 1 1 Z)

y = Cie~'' + Cie^" + d exp (— ix) +


ANS. C4 exp (fx).
18. (AD* -\6D^ + 7D^ + AD - 2)y = 0.
19. (4D* + 4/)^ - 13D^ - 7D + 6)y = 0.
20. (4/)^ - 8Z)* - 17Z)' + 12D^ + 9D)y - 0.
21. (D^ - 4a£) + 3a^)y = 0, a real ^ 0.
22. [D^ — (a + 6)D + a^];' = 0, a and ^ real and unequal.

In Exs. 23-24, find the particular solution indicated.

23. (D^- 2D- 3)7 = 0; when = 0,7 = 0, and/ = -4. ans. J = e-^-e^^
a:

24. (D^ — D— €)y = 0; when x = 0, 7 = 0, and when x= \,y = e^.


ans. y = {e^'' — e~^'')l{\—e~~).

In Exs. 25-29, find for x = 1 the y value for the particular solution required.

25. (Z)^ - 2Z) - 3)7 = 0; when jc = 0, 7 =4 and 7' = 0.


Whenx= 1,7 = ^3 + 3^-' = 21.2.
ans.
26. {D^ - AD)y = 0; when x = 0, 7 = 0, 7' = 0, and 7" = 2.
ANS. When = 1,7 = sinh^ jc 1.

27. (D^ - Z) - 6)7 = 0; when x = 0, 7 = 3 and 7' = - 1.


ANS. When x = 1, 7 = 20.4.
28. {D^ + 3D- 10)7 = 0; when x = 0, 7 = 0, and when x = 2, 7 = 1.

ANS. When x= 1, 7 = 0.135.


29. (D^ - 2D^ -5D + 6)7 = 0; when x = 0, 7 = 7' = -7, and 7" = -1. 1,

ANS. When X- 1,7= -19.8.

35. The auxiliary equation; repeated roots

Suppose that in the equation

KD)y = (1)
§ 35] The auxiliary equation; repeated roots 101

the operator/(Z)) has repeated factors; that is, the auxihary equation/(w) =
has repeated roots. Then the method of the previous section does not yield
the general solution. Let the auxiliary equation have three equal roots
Wj = W2 = b, /W3 = b. The corresponding
b, part of the solution yielded by
the method of Section 34 is

y = {c,+c^ + c,)e'\ (2)

Now (2) can be replaced by


y = c^e'^ (3)

with C4 = Cj + C2 + C3 . Thus, corresponding to the three roots under con-


sideration, this method has yielded only the solution (3). The difficulty is

present, of course, because the three solutions corresponding to the roots

mi = W2 = W3 = 6 are not linearly independent.


What is needed is a method for obtaining n linearly independent solutions
corresponding to n equal roots of the auxiliary equation. Suppose the
auxiliary equation /(m) = has the n equal roots

m^ = 1712 = • — m„ — b.

Then the operator f{D) must have a factor {D — b)". We wish to find n
linearly independent >''s for which

{D - b)"y = 0. (4)

Turning to the result (6) near the end of Section 32 and writing m = b,
we find that
(D - bfix^'e^'') = for A: = 0, 1, 2, ...,(« - 1). (5)

The functions y,, = x'^e''"'; = 0,


^ 1, 2, ...,(/?— 1) are linearly independent
because aside from the common factor e'"' they contain only the respective
powers x°, x\ x^, ...,x"~\ (See Ex. 4, Sec. 32.)

The general solution of equation (4) is

y = de'"' + C2 xe^"" + • • •
+ c„ x"" ^e*"^. (6)

Furthermore, if /(/)) contains the factor {D — b)", then the equation

fiD)y = (1)
can be written
g{D){D - bfy = (7)

where ^(Z)) contains all the factors off(D) except (D — b)". Then any solution
of
(D - bfy = (4)

is also a solution of (7) and therefore of (1).


102 Linear Equations with Constant Coefficients [Ch. 6

Now we are in a position to write the solution of equation (1) whenever


the auxihary equation has only real roots. Each root of the auxiliary equation
is either distinct from all the other roots or it is one of a set of equal roots.
Corresponding to a root m, distinct from all others, there is the solution

>;,• = c,- exp (m,or), (8)

and corresponding to n equal roots niy, m2 , ..., ni„, each equal to b, there

are the solutions

Cie^'',C2xe''\...,c„x"-'^e''\ (9)

The collection of solutions (9) has the proper number of elements, a number
equal to the order of the differential equation, because there is one solution
corresponding to each root of the auxiliary equation. The solutions thus
obtained can be proved to be linearly independent.

EXAMPLE (a): Solve the equation

(£)4_7£)3 ^ 18^2 _2o/) + 8)>' = 0. (10)

With the aid of synthetic division, it is easily seen that the auxiliary equation

m^ - 7m^ + 18m^ - 20m + 8 =


has the roots m= 1, 2, 2, 2. Then the general solution of equation (10) is

y = de"" + C2 e^"" + C3 xe^* + c^ x^e^-\

or
y = Cie"" + (C2 + C3 A- + C4 x^)e^''.

EXAMPLE (b): Solve the equation

d^
dx'*'
2^
dx^
^=
dx^

The auxiliary equation is

m'^ + 2m^ + m^ = 0,

with roots m = 0, 0, — 1, — 1 . Hence the desired solution is

J = Ci + C2X + c^e'"' + C4xe~^

Exercises
In Exs. 1-20 find the general solution.

1. (Z)'-6Z) + 9)^ = 0. ANS. y={ci + C2x)e^\


2. (D^ + 4D + 4)y = 0.
3. (4£)' + 4D^ + D)y = 0. ans. y = Ci + ic2 + C3 x) exp (- ix).
§ 36] A definition of exp z for imaginary z 103

4. (Z)^-8Z)^+ \(>D)y = 0.

5. (/)^ + + 9Z)^)>' = 0.
6Z)3 ANS. 7 = Ci + Cz X + (C3 + C4 x)e - ^\
6. (Z)^-3Z)^ + 4)>' = 0.
7. (4Z)^ - 3Z) + \)y = 0. ans. 7 = c^e-'' + (c2 + c^x) exp (ix).
8. (/)* - 3>D^ - 6D^ + 2SD - 24)y = 0.

9. i4D* - 4D^ - 23 D^ + \2D + 36)y = 0.


ANS. y = ici + ci xye^" + (ca + C4 x) exp (— f x).
10. (Z)^ + 6Z)2 + 12D + 8)>' = 0.
11. {D^ — D^)y = 0. ANS. = Ci + C2X + CsX^ + C4e* + Cs^"*;
>'

or = Ci + C2 X + C3 x^ + ce cosh X + ci sinh x.
^^

12. (D'- l6D^)y^0.


13. (4/)'^ + 4D^ -3D^-2D+\)y = 0.
14. (54Z)^ - 21 D^ - 9Z)2 + 7D - l)^ = 0.
15. (D* - 2Z)^ - 3Z)^ + 4Z) + 4)7 = 0. ans. 7 - (ci + Czx)^-" + (C3 + C4x)e^^
16. (27/)* - 18/)^ + 8Z) - l)j; - 0.
17. (4Z)^ + 4Z)* - 9D^ - 11£)^+ D + 3)j = 0.
18. (Z)^ - 15D' + lOZ)^ + 60D - 72)>' = 0.
19. (Z)* + 2Z)3 - 6D^ - 16Z) - 8)>' = 0.

ANS. J = (ci + C2 x)e " + C3 exp [(1 + V3)x] + C4 exp [(1 — V3)a:].
^"^

20. (/)* - 2i)^ - 5D^ + 2)>' = 0.

In Exs. 21-26, find the particular solution indicated.

21. {D^ + 4Z) + 4)y = 0; when a: = 0, = j' 1 and /= - 1.

ANS. 7= (1 + x)e~^*.
22. The equation of Ex. 21 with the condition that the graph of the solution pass
through the points (0, 2) and (2, 0). ans. y = {X — x)e-^''
23. (D^ -3D-2)y = 0; when x = 0, y= 0, y' = 9, y" = 0.

ANS. y ^ 2e^'' + {3x — 2)e~


24. {D^ + 3D^ + 2D^)y = 0;
= 0, y = Q, y' ^ 4, y" = -6, y'" = 14.
when x
ANS. y = 2{x + e-'' — e~^'')
25. The equation of Ex. 24 with the conditions when x = Q, y = 0, y' ^ 3, y" — — 5
:

y"'^9. ANS. = 2 — e"*— e~^* j;

26. {D^ + D^ - D- \)y 0; when x 0, y=\, when x 2, = 0, and also as


= = = >'

x^ oo,y->0. ANS. 7=^(2 — A:)e~*

In Exs. 27-29, find for = 2 the y value for the particular solution required
a;

27. {4D^ - 4Z) + \)y = 0; when x^Q, y = -2, y' = 2.


ANS. When x = 2, = 4e j'

28. {D^ + 2Z)^)>' = 0; when = 0, = -3, / = 0, /' = 12.


a: >'

ANS. When x = 2, >»= 3e~* +


29. (Z)^ + 5D^ + 3D-9)y = Q\ when x = 0,y-^ -\, when x = 1,>' = 0, and also
as x^ 00, >'->0. ANS. When x = 2, 7 = e~^.

36. A definition of exp z for imaginary z

Since the auxiliary equation may have imaginary roots, we need now to lay
down a definition of exp z for imaginary z.
104 Linear Equations with Constant Coefficients [Ch. 6

Let z = a + //? with a and P real. Since it is desirable to have the ordinary
laws of exponents remain valid, it is wise to require that

exp (a + //?) = e'' e'". (1)

To e" with a real, we attach the usual meaning.

Now consider e''^, P real. In calculus it is shown that for all real x

X x~ x^ X"
*
-'+T! + 2! + 3!
+ - + ;^ + -' ('>

or

^^=Z^- (2)

If we now tentatively put x = ifi in (2) as a definition of e'^, we get

., ip i'p^ i'P^ i^P" i"P"


^
^
1! 2! 3! 4! n\

Separating the even powers of P from the odd powers of P in (3) yields

2! ^ 4! i2k)\

ip i'p' i'>^^'p'

or

:
y -^- + y
A'o (2/c)! fcfo
- ——
(2/c +
'^

l)!
.
(4)

Now / = (— 1) , so we may write

2!
^4!^ (2/v)!

p p^ i-D^p^"^^
+ /
_1_ • . . _J_ _L
(5)
1! 3! (2/c+l)!

or

,% i2ky. fct-o (2/c + l)!

But the series on the right in (5) are precisely those for cos P and sin P as

developed in calculus. Hence we are led to the tentative result

e'^ = cosp + i sin p. (6)


§ 37] The auxiliary equation; imaginary roots 105

The student should realize that the manipulations above have no meaning
in themselves at this stage (assuming that infinite series with complex terms
are not a part of the content of elementary mathematics). What has been
accomplished is this: the formal manipulations above have suggested the
meaningful definition (6). Combining (6) with (1) we now put forward a
reasonable definition of exp (a + ifi), namely,

exp (a + iP) = e°'(cos j5 + / sin P), a and P real. (7)

Replacing jS by ( — j?) in (7) yields a result which is of value to us in the next


section,
exp (a — ifi) = e°'(cos P — i sin /?).

It is interesting and important that with the definition (7), the function e^

for complex z retains many of the properties possessed by the function e"" for

real x. Such matters are often studied in detail in books on complex variables.*

Here we need in particular to know that if

y = exp (a + ib)x,

with a, b, and x real, then

(D-a- ib)y - 0.
The result desired follows at once by differentiation, with respect to x, of the
function
y = e^^Ccos bx + / sin bx).

37. The auxiliary equation; imaginary roots

Consider a differential equation /(D)>^ = for which the auxiliary equation


/(m) = has real coefficients. From elementary algebra we know that if the
auxiliary equation has any imaginary roots, those roots must occur in con-
jugate pairs. Thus if

nil = a + ib

is a root of the equation /(m) — 0, with a and b real and b ^0, then
ni2 — a — ib

is also a root of f{m) = 0. It must be kept in mind that this result is a con-
sequence of the reality of the coefficients in the equation/(m) — 0. Imaginary
roots do not necessarily appear in pairs in an algebraic equation whose
coefficients involve imaginaries.

* For example, R. V. Churchill, Complex Variables and Applications (2nd edition; New
York: McGraw-Hill Book Co., 1960), pp. 46-50.
106 Linear Equations with Constant Coefficients [Ch. 6

We can now construct in usable form solutions of

f{D)y = (1)

corresponding to imaginary roots of/(/?7) = 0. For, since /(w) is assumed to


have real coefficients, any imaginary roots appear in conjugate pairs

mj —a -\- ib and m^ — a — ib.

Then, according to the preceding section, equation (1) is satisfied by

y= Cy exp [{a + ib)x] + C2 exp [{a — ib)x\. (2)

Taking .v to be real along with a and b, we get from (2) the result

y = Cye°^{cos bx + / sin bx) + Ci e^^Ccos bx — i sin bx). (3)

Now (3) may be written

y — {c^ -\- C2)e'"' cos bx + i(ci — C2)e°^ sin bx.

Finally, let Cj + C2 = C3 , and i{Ci — C2) = Q, where fj and c^ are new


arbitrary constants. Then equation (1) is seen to have the solutions

y= c^e """cos bx + c^e """sin bx, (4)

corresponding to the two roots m^ — a + ib and /Wj = ^ — ib, (b # 0) of the


auxiliary equation.
The reduction of the solution (2) above to the desirable form (4) has been
done once and that is enough. Whenever a pair of conjugate imaginary roots
of the auxiliary equation appears, we write down at once in the form given on
the right in equation (4) the particular solution corresponding to those
two roots.

EXAMPLE (a): Solve the equation

(/)'-3Z)2 + 9Z)+ 13)>' = 0.

For the auxiliary equation

m^ - 2>m^ + 9m +13 = 0,

one root, Wj = — 1, is easily found. When the factor (m + 1) is removed by


synthetic division, it is seen that the other two roots are solutions of the
quadratic equation
tn^ - 4m + 13 = 0.

Those roots are found to be m2 =2 + 3/ and m^ = 2 — 3/. The auxiliary


equation has the roots m= — 1, 2 + 3/. Hence the general solution of the
differential equation is

y = Cie~^ + C2 e^^ cos 3x + c^e ^* sin 3x .


§ 37] The auxiliary equation; imaginary roots 107

Repeated imaginary roots lead to solutions analogous to those brought


in by repeated real roots. For instance, if the roots m = a ± ib occur three
times, then the corresponding six linearly independent solutions of the
differential equation are those appearing in the expression

(c, + C2 X + C3 x^)e'"' cos bx + {c^ + C5 x + Cg x^)e'"' sin bx.

EXAMPLE (b): Solve the equation

(Z)^ + 8Z)2 + 16)>' = 0.

The auxiliary equation n-i^ + 8/»^ + 16 = may be written

(,„2 ^ 4)2 ^ Q^

so its roots are seen to be m— ±2i, ±2i. The roots mi = 2? and mj = —2/
occur twice each. Thinking of 2/ as + and recalling
2/ that e°^ = 1, we
write the solution of the differential equation as

y = (ci + C2 x) cos 2x + (C3 + C4 A') sin 2x.

In such exercises as those below a fine check can be obtained by direct


substitution of the result and its appropriate derivatives into the differential
equation. The verification is particularly effective because the operations
performed in the check are so different from those performed in obtaining
the solution.

Exercises
Find the general solution except when the exercise stipulates otherwise.

1. Verify directly that the relation

y = C3 e""^ cos bx + Ca e'" sin bx (4)

satisfies the equation

[(/) - ay + b^]y = 0.
2. (D^ — 2D + 5)y = 0. Verify your answer.
ANS. y = Cic" cos 2x + C2 e" sin 2x.
3. {D^ -2D + 2)y = 0.
4. (D^ + 9)y = 0. Verify your answer. ans. y=Ci cos 3a: + ci sin 3x.
5. {D^ — 9)y = 0. ANS. y=Ci cosh 7>x + d sinh 2>x.
6. (D^ + AD + 5)y = 0. Verify your answer.
7. (D^ - 6£) + 25);^ = 0. ANS. y = e^^Ci cos Ax + Ci sin 4jc).
8. (Z)3 + D^ + 4Z) + 4)>' = 0. Verify your answer.
9. (D^ — \)y = 0; when x = 0,y = yo and y = 0. ans. y = yo cosh x.
10. (£)^ + !)>' = 0; when x = 0, 7 = jo and / = 0. ans. y = yo cos x.
11. (/)* + 2Z)3 - 11 D^ - 52£>)>' = 0.
12. (Z)3 + 5Z)^ + 17Z) + 13)>' = 0; when x ^0, y = Q,y =\, and /' = 6.
ANS. = e""* — e"^'' cos 3a:. >'
108 Linear Equations with Constant Coefficients [Ch. 6

13. (Z)5 - - 2£)^ - 3Z) - 2)y = 0.


2Z)^
14. (£>* - + ID^ -2D+ l)y = 0. Verify your answer.
2D^
15. (Z)^+ 18Z)^ + 81)>' = 0. ANS. = (c, + C2A:)cos3x +
j' (c3 + C4Jc)sin3A:.
16. (2D^ +l\D^- AD^ - 69D + 34)j = 0. Verify your answer.
17. (Z)^ + 6/)* + 9D^ + A)y = 0.
18. (16I>^- \\D-5)y = Q.

d^x dx
19. ^-^ + /:^a: = 0, k real; when / = 0, x = and —r=Vo- Verify your result
dt^ dt
completely. ans. x = {volk) sin kt.
20. (/)' + Z)^ + 4Z) + 4)>' = 0; when x = 0, j = 0, / = - 1, and >'"= 5.
ANS. y = e~'' — cos2x.
d^x dx dx
21. -— + 2b — + k^x = 0, k > b >Q; when / = 0, x = and —= !;o
dt^ dt dt

ANS. x = {vold)e''" sin a/, where a = Vk^ — b^.

Miscellaneous exercises
Obtain the general solution unless otherwise instructed. Your answer can be
checked by direct substitution.

1. 9Z)* + 6Z)^ + D^)y = 0.


2. AD^- nD + €)y = Q.
3. D^ + 2D^- l5D)y = 0.
4. D^ + 2D^ + D + 2)y - 0.
5. D^-2D^ -2>D)y = 0.
6. £)3 4. 3£)2 _ 4)_^ _ 0.

7. AD^ - 21 D + 21)y = 0.
8. 10Z)^ + Z)^ - 7D + 2)>' = 0.
9. D^ + 7Z)2 + 19Z) + 13)>' = 0; when = 0, = 0, / = 2, and /' = - 12.
a- >'

10. D^ - D-e)y = 0; when x = Q,y = 2 and = 1. >''

11. Z)'' + 6D^ + 9D^)>' = 0; when x = 0, 7 = 0, j' = 0, and j'" = 6, and as x -> cx),

''->!. For this particular solution, find the value of >' when x = 1.

ANS. y=\—e~^.
12. D^ + 6/)" + 12D + 8);' = 0; when x = 0, j = 1, /= -2, and /' = 2.
13. Z)' + 3Z)^ + 3Z)+l)j = 0.
14. £)4 _ 2£)3 _ i3£)2 ^ 38£, _ 24)^ = 0.

15. 06 ^ 9£,4 ^ 24Z)^ + \&)y = 0.


16. 8D^-4Z)2-2D+ 1)>' = 0.
17. D^ + D^- AD^ - AD)y = 0.
18. D^ - 2D^ + 5D^ -SD + A)y = 0.
19. D^ + 2D^ + 1)7 = 0.
20. D* + 5D^ + A)y = 0.
21. D* + 3Z)' - AD)y = 0.
22. £)5 + £)4 _ 9£)3 _ 13£)2 ^ 8Z) + 2)>; = 0. I

23. Z)* - 11 D^ + 36Z)^ - 16Z) - 64)j = 0.


24. D^ + 2D + 5)y = 0.
25. /)4 4. 4£>3 ^ 2Z)^ -8D- 8)y = 0.

26. 4Z)* - 2AD' + 35D^ + 6D - 9)y = 0.


Miscellaneous exercises 109

27. 4D* + 20£)3 + 35Z)^ + 25D + 6)y = 0.

28. D*-7D^+ IID^ + 5D- 14)7 = 0.


29.
30. D^-2D^ + D- 2)y = 0.
31. D^- D^+ D- 1)7 = 0.
32. D^ + 4D^ + 5D)y = 0.
33. £>*- 13/)^ + 36)7 = 0.
34. /)4 _ 5^3 j^ 52)2 + 5/) _ 6)7 = 0.

35. 4D^ + SD^- 11Z) + 3)7 = 0.


36. D'+ D^- 16/)- 16)7 = 0.
37. D^ - Z)3 - 3Z)^ + Z) + 2)7 = 0.
38. /)3_2Z)2_3£)+ 10)7 = 0.
39. Z)^ + Z)'^ - 6Z)^)7 = 0.
40. 4D^ + 2SD^ + 61 D + 37)7 = 0.
41. 4D^ +nD^+l3D+ 10)7 = 0.
42. 18Z)3 - 33D^ + 20D - 4)y = 0.
43. 4/)^- l5D^ + 5D + 6)y = 0.
44. D' + D^-ID^- 11/)2_8Z)- 12)7 = 0.
45. Z)* + 3Z)2 - 6Z)^ - 28Z> - 24)7 = 0.
46. 4D^ - 4D^ - 23D^ + 12D + 36)7 = 0.
47. 4D' - 23D^ - 33D^ - 17D - 3)7 = 0.
CHAPTER T
Nonhoniogeneous Equations

Undetermined Coefficients

38. Construction of a homogeneous equation from a


specified solution

In Section 29 we saw that the general solution of the equation

iboD" + b.D"-' + + b„_,D + b„)y = R{x) (1)

is

where y^, the complementary function, is the general solution of the homo-
geneous equation

{bo D" + b,D"-' + -" + b„_,D + b„)y = (2)

and >^p is any particular solution of the original equation (1).

Various methods for getting a particular solution of (1) when the bo,

110
§ 38] Construction of a homogeneous equation 111

bi, ..., b„ are constants will be presented. In preparation for the method of
undetermined coefficients it is wise to obtain proficiency in writing a homo-
geneous differential equation of which a given function of proper form is a
solution.
Recall that in solving homogeneous equations with constant coefficients,
a term such as Cic"^ occurred only when the auxiliary equation /(m) = had
a root m= a, and then the operator/(Z)) had a factor (D — a). In like manner,
appeared only when f{D) contained the factor
€2X6'"' {D — a)-^, c^x^e"'' only
when f{D) contained (D — a)^, etc. Such terms as ce"'' cos bx or ce""" sin bx
correspond to roots m= a ± ib, or to a factor [{D — a)^ + b^].

EXAMPLE (a): Find a homogeneous linear equation, with constant


coefficients, which has as a particular solution

y = le^"" + 2x.

First note that the coefficients (7 and 2) are quite irrelevant for the present

problem, so long as they are not zero. We shall obtain an equation satisfied

by > = Cjg^* + C2X, no matter what the constants c^ and Cj may be.

A term c^e^'' occurs along with a root m — Z of the auxiliary equation. The
term Cj x will appear if the auxiliary equation has m = 0, 0; that is, a double
root m = 0. We have recognized that the equation

D\D - l)y = 0,

or
(D^ - 3D^)y = 0,

has y = c^e^"^ + Cj x + c 3 as its general solution, and therefore that it also

has y = le^^ + 2x as a particular solution.

EXAMPLE (b): Find a homogeneous hnear equation with real, constant


coefficients that is satisfied by

^^ = 6 + 2)xe'^ — cos X. (3)

The term 6 is associated with m = 0, the term Sxe"" with a double root
m= 1, 1, and the term ( — cos x) with the pair of imaginary roots m — 0±i.

Hence the auxiliary equation is

m{m- \)\m^ + 1) = 0,

or
m^ — 2m* + 2w^ — 2m^ + w = 0.

Therefore the function in (3) is a solution of the differential equation

(D^ - 2Z)^ + 2D^ - 2D^ + D)y = 0. (4)


112 Nonhomogeneous Equations: Undetermined Coefficients [Ch. 7

That is, from the general solution

y= Ci + (cj + C3 x)e^ + C4 cos X + C5 sin x

of equation (4), the relation (3) follows by an appropriate choice of the


constants: c^ — 6, C2 = 0, c-j = 3, q= — 1, C5 = 0.

EXAMPLE (c): Find a homogeneous linear equation with real, constant


coefficients which is satisfied by

y — Axe'' sin 2x.

The desired equation must have its auxiliary equation with roots
m= I ±2i, \ ± 2/. The roots m= \ ± 2/ correspond to factors (m — 1)^ + 4,

so the auxiliary equation must be

[(m - 1)2 + 4]2 = 0,

or

m* - 4m^ + \4m^ - 20m + 25 = 0.

Hence the desired equation is

(£)4 _ 4/)3 + i4£)2 _ 20/) + 25)y = 0.

Note that in all such problems, a correct (but undesirable) solution may be
obtained by inserting additional roots of the auxiliary equation.

Oral exercises
In Exs. 1-14, obtain in factored form a linear differential equation, with real,
constant coefficients and which is satisfied by the given function.

1. y = 4e^^ + 3e- ANS. iD-2)(D+l)y = 0.

2. y=l-2x+ he^\ ans. D\D - A)y = 0.


3. 7 = -2a: + ie*^ ans. D\D - A)y = 0.
4. y = x^-5 sin 3x. ans. D\D^ + 9)y = 0.
5. >- = 2e^ cos 3x. ANS. (Z)- l-3/)(/)- l + 3i> = 0;
or [(£) - 1)^ + 9]>' = 0; or (Z)^ - ID + \0)y = 0.
6. y = x^ + Ae".
7. 7=3x^^-2^-^
8. y=6 + 2e^\
9. y = 2x^+ 1.
10. y = sin 2x.
11. >»= sin 2x + 3 cos 2x.
12. y = cos /:x.
13. y = x sin 2x.
14. y = 2 sinh x.
§ 39] Solution of a nonhomogeneous equation 113

In Exs. 15-36, list the roots of the auxiliary equation for a homogeneous linear
equation with real, constant coefficients and having the given function as a parti-
cular solution.

15. y = 3xe^''. ans. m = 2,2.


16. y = x^e'" + 46". ans. m = — 1, — 1, — 1, 1.
17. y = e~'' cos 4x. ans. m = — ± 4/.
1

18. >'= 3e"'cos4x+ 15e~* sin4A:. ans. m=— lih4/.


19. y = xie^" + 4). ans. m -= 0, 0, 2, 2.

20. y=3 + Se-^\


21. y = 2x^-e-^\
22. y = 4 + 2x^-e-^\
23. J'
= xe"".
24. >"= xe" + 5e*.
25. y = 4 cos 2x.
26. J= 4 cos 2a: — 3 sin 2x.
27. y= X cos 2x — 3 sin 2a:.

28. y= x cos 2x.


29. y= e~^''cos 3a:.

30. >'= e~ ^*(cos 3Ar + sin 3a:).


31. y = x^ — x + e'^ix + cos x).
32. y = x^ sin x.
33. >'= a:^ sin + cos
a: a: a:.

34. >'= 8 cos 4a: + sin 3a:.


35. y = sin^ X. Use the fact that sin^ x = i(3 sin x — sin 3a:),

36. y = cos^ A-.

39. Solution of a nonhomogeneous equation

Before proceeding to the theoretical basis and the actual working technique
of the useful method of undetermined coefficients, let us examine the under-
lying ideas as applied to a simple numerical example.
Consider the equation
D\D - \)y = 3e* + sin x. (1)

The complementary function may be determined at once from the roots


w = 0, 0, 1 (2)

of the auxiliary equation. The complementary function is

Jc = <^i + ^2 ^+ C3 e\ (3)

Since the general solution of (1) is

where y^ is as given in (3) and yp is any particular solution of (1), all that
remains for us to do is to find a particular solution of (1).
114 Nonhomogeneous Equations: Undetermined Coefficients [Ch. 7

The right-hand member of (1),

R{x) = 3e-^ + sin x, (4)

is a particular solution of a homogeneous linear differential equation whose


auxiliary equation has the roots

w'- 1, ±/. (5)

Therefore the function 7? is a particular solution of the equation

{D - \)(D^ + \)R = 0. (6)

We wish to convert (1) into a homogeneous linear differential equation

with constant coefficients, because we know how to solve any such equation.
But, by (6), the operator {D — \){D^ + 1) will annihilate the right member of
(1). Therefore, we apply that operator to both sides of equation (1) and get

(Z) - 1 )iD^ + )D\D -1 1 )>' = 0. (7)

Any solution of (1) must be a particular solution of (7). The general solution
of (7) can be written at once from the roots of its auxiliary equation, those
roots being the values w= 0, 0, 1 from (2) and the values m' = 1, +/ from (5).

Thus the general solution of (7) is

y = Ci + C2X + c^e^ + C4, xe"^ + f5 cos x + c^, sin x. (8)

But the desired general solution of (1) is

y = yc + yp^ (9)
where
>'c = f1 + f2 -V + f3 e\

the Tj, C2 , C3 being arbitrary constants as in (8). Thus there must exist a

particular solution of (1) containing at most the remaining terms in (8).

Using different letters as coefficients to emphasize that they are not arbitrary,
we conclude that (1) has a particular solution

yp = Axe" + B cos x + C sin x. (10)

We now have only to determine the numerical coefficients A, B, C by direct

use of the original equation

D\D - \)y = 3e^ + sin x. (1)

From (10) it follows that

Dyp = Aixe"" + e^) — B ?,\n x + C cos x,

D^yp = A^xe"" + le"") — B cos x — C sin x,


D^y = A{xe^ + 3e^) + B sin x - C cos x.
§ 40] The method of uiidetermined coefficients 115

Substitution o( yp into (1) then yields

Ae"" + (B + C) sin x + (B - C) cos x = 3e'' + sin x. (11)

Since (11) is to be an identity and since e^, sin x, and cos x are linearly
independent, the corresponding coefficients in the two members of (1 1) must
be equal; that is,

A = 3

B+C= 1

B-C = 0.

Therefore A = 3, B= ^, C= ^. Returning to (10), we find that a particular


solution of equation (1) is

jVp = 3xe'' + \ cos X + ^ sin x.

The general solution of the original equation

D\D - \)y = 3e'' + sin X (1)

is therefore obtained by adding to the complementary function the yp found


above

y — c^ + C2X + c^e"" + 3xe'' + \ cos x + | sin x. (12)

A careful analysis of the ideas behind the process used shows that to arrive
at the solution (12), we need perform only the following steps:

(a) From (1) find the values of w and m' as exhibited in (2) and (5);
(b) From the values of m and m' write y^. and yp as in (3) and (10);
(c) Substitute yp into (1), equate corresponding coefficients, and obtain
the numerical values of the coefficients in yp\
(d) Write the general solution of (1).

40. The method of undetermined coefficients

Let us examine the general problem of the type treated in the preceding
section. Let/(Z)) be a polynomial in the operator D. Consider the equation

f{D)y = R{x). (1)

Let the roots of the auxiliary equation /(aw) = be

m = nil, m2, . .., m„. (2)

The general solution of (1) is

y = yc + yp (3)
116 Nonhomogeneous Equations: Undetermined Coefficients [Ch. 7

where y^ can be obtained at once from the values of /» in (2) and where
y = y^is any particular solution (yet to be obtained) of (1).

Now suppose that the right member R{x) of (1) is itself a particular solution

of some homogeneous linear differential equation with constant coefficients,

9{D)R = 0, (4)

whose auxiliary equation has the roots

m' = m\,m'2,---,m'k. (5)

Recall that the values of i»' in (5) can be obtained by inspection from Rix).
The differential equation

g{D)f(D)y - (6)

has as the roots of its auxiliary equation the values of m from (2) and m' from
(5). Hence the general solution of (6) contains the y^ of (3) and so is of the
form

But also any particular solution of (1) must satisfy (6). Now, if

f(D)(y, + y,) = Rix),

thenf{D)yg = Rix) because /(Z))>'^. = 0. Then deleting the y^ from the general
solution of (6) leaves a function y^ which for some numerical values of its

coefficients must satisfy (1); that is, the coefficients in y^ can be determined

so that yq = yp- The determination of those numerical coefficients may be


accomplished as in the examples below.
It must be kept in mind that the method of this section is applicable when,
and only when, the right member of the equation is itself a particular solution
of some homogeneous linear differential equation with constant coefficients.

EX A M P L E (a) : Solve the equation

iD^ + D- 2)y = 2jc - 40 cos 2x. il)

Here we have
m= \, —2
and
w' = 0, 0, ±2i.

Therefore we may write

yc = <^i^'' + <^2^~^^
yp = A + Bx + C cos 2x + E sin 2x,

in which q and C2 are arbitrary constants, while A, B, C, and E are to be


determined numerically so yp will satisfy the equation (7).
§ 40] The method of undetermined coefficients 117

Since
Dyp = B — 2C sin 2.v + 2£ cos 2x
and
D^yp = -4C cos 2x - AE sin 2x,

direct substitution of yp into (7) yields

— AC cos 2x — AE sin 2x + B — 2C sin 2x + 2^ cos 2x — 2


- 2Bx - 2C cos 2.V - 2E sin 2x = 2a- - 40 cos 2x. (8)

But (8) is to be an identity in x, so we must equate coefficients of each of


the set of hnearly independent functions cos 2x, sin 2x, x, 1 appearing in the

identity. Thus it follows that

-6C + 2E= -40,


-6E-2C = 0,
-2B = 2,

B-2A=0.
The above equations determine A, B, C, and E. Indeed, they lead to

A^-\, C= 6,

B=-\, E^-2.
Since the general solution of (7) is y = yc + >'p , we can now write the
desired result,

y — c^e^ + C2 e~^^ — ^ —X+ 6 cos 2x — 2 sin 2x.

EXAMPLE (b): Solve the equation

{D^ + \)y = smx. (9)

At once m= ±i and m' = ±i. Therefore

y^ — c^ cos X + f 2 sin x,

yp = Ax cos X + Bx sin x.
Now
y'p = A( — x cos X — 2 sin x) + B( — x sin x + 2 cos x),

so the requirement that yp satisfy equation (9) yields

— 2A sin X + 2B cos x = sin x,

from which A = —j and B = 0.

The general solution of (9) is

y = Ci cos X + <:2 sin X — ^x cos x.


118 Nonhomogeneous Equations: Undetermined Coefficients [Ch. 7

EXAMPLE (c): Determine y so that it will satisfy the equation

j'"-/ ==4e~" + 3^2^


(10)

with the conditions that when x = 0. >' = 0, y' = — 1. and y" = 2.

First we note that m= 0, 1. — 1, and m' = —1.2. Thus

y ^ Axe'"" + Be^''.

Now
2x
J-;
= Ai-xe-'' + e-"") + 2Be^\
y'L = A(xe-'' - 2^-^) + ABe^\
y"' = Ai-xe'"" + 3e"") + SBe Ix
Then
y";-y'^ = lAe-^-^(^Be^\

so that from (10) we may conclude that A =1 and B= \.

The general solution of (10) is therefore

3; =d + ^2^'' + ^3^""^ + Ixe'"" + \e^''. (11)

We must determine Tj, c 2 ^ <"3 so (11) will satisfy the conditions that when
X = 0, >' = 0, >•' = — L and y" — 2.

From (11) it follows that

y= f,e- - ^3^-" - Ixe-"" + 2^-^ + ^2^ (12)

and

/ = f2 e^ + ^3 e"^ + Ixe-"" - 4^-' + le^\ (13)

We put .Y = in each of (11), (12), and (13) to get the equations for the

determination of c,, c 2 ^ arid '^i


These are

=
+ r2 + f 3 + i
f1

- = C2 - f 3 + 3,
1

2 = C2 + f3 - 2,

from which Cj = — f, £"2 = 0' <^3 = 4. Therefore, the final result is

J = _ I + 4e~^ + 2.V6'-" + if^-^

An important point, sometimes overlooked by students, is that it is the

general solution, the y of (11), that must be made to satisfy the initial con-

ditions.
§ 40] The method of undetermined coefficients 119

Exercises
In Exs. 1-35, obtain the general solution.

1. (D^ + D)y = —cos X. = Ci + de'" + ^ cos x — ^ sin x.


ANS. y
2. (D^ -6D + 9)y = e\ ans. y = {cv^ C2 x)e''' + \e'.
3. (Z)^ + 3D+ 2)^= 12x^ ANS. y = Cie-' + C2e-^''-V6x^-\%x + l\.
4. {D^ + l>D + 2)y=\^3x + x^. ans. y = Cie-'' + CiC-^" + \x^.
5. (D^ + 9)y = Se'' — 162x. ans. y^Ci cos 3x + Cz sin 3a: + ^e'' — 18x.
6. (D^ + 9)y = Se'' - I62x\ ans. = c, cos 3x + C2 sin 3x + ^e-* - 18^^ + 4.
j'

7. y" — 3/ — 4y= 30^"^. ans. = Cie*'' + C2 e" — Se"".


>» "^

8. y" - 3y' - Ay = 30e*^ ans. y = {ci + 6x)e^=' + C2 e" *.

9. {D^ - A)y = e'^ + 2. ans. = ce' '^ + (cz + \x)e^^ - \.


>'

10. (Z)'- Z)-2)>'=6x + 66'-^ ANS. = c,e-^ + C2e^"-3A:+3/2-2xe-\


>'

11. y — 4>'' + 3^ = 20 cos X. ANS. y = Cie" + C2 e^"" + 2 cos x — 4 sin x.


12. y — 4j' + 3>' = 2 COS X + 4 sin x. ans. = de"" + C2 e^"" + cos x.
j'

13. >-" + 2>'' + = 7 + 75 sin Ix.


>'

ANS. = e~ + C2 x) + 7 — 12 COS 2x — 9 sin Ix.


>^ ''(ci

14. (Z)^ + 4D + 5)^ = 50a- + Ue^\


ANS. J = e" ^"^(d cos X + C2 sin + lOx — 8 + \e^''. a')

15. (D^ + \)y = cos AT. ANS. y = Ci cos x + C2 sin x + ^x sin x.


16. (Z)^ - 4i) + 4)j - e^\ ANS. 7 = e'"(ci + C2 x + ^x^).
17. (Z)^ —!)>' = e-^(2 sin X + 4 cos x).
18. (D' - 1)>' = 8xe\ ANS. y = cxc-" + c^Ccz - 2x + 2x^).
19. (D^ — D)>' = X. ANS. y = Ci+C2e''-^Cie~''—^x^.
20. {D^ - D^ + D - \)y = 4 sin x.
ANS. y = Cic" + {c2 + x) COS X + (Ci — x) sin x.
21. {D' + D^-4D- 4)y = 3^"^ - 4x - 6.
ANS. = Ci^^'' + Cze'^" + {c3 — x)e~ " + x + i.
>'

22. (Z)'^- l)j = 7x^


23. (D"* — 1)>' = e"*. ANS. y = Cic" + (cz — ix)e~* + C3 cos x + c 4 sin x.
24. (Z)^ — \)y = 10 sin^ x. Use the identity sin^ x = ^{\ — cos 2x).
ANS. = Ci^'' + 02^"'' — 5 + cos 2x.
>'

25. (D^ + \)y= 12 cos^ x. ans. y = Ci cos x + C2 sin x + 6 — 2 cos 2x.


26. (D^ + 4)y = 4 sin^ x. ans. y = Ci cos 2x + C2 sin 2x + ^{l — x sin 2x).
27. y" - 3>'' — 4;^ = 6x - 50 cos 2x.
1

28. (D^ -3D- 2)y = 100 sin 2x.


29. /' + 4/ + 3y^ 15e'^ + e-\
30. y" — y = — 4.
e''

31. y" — y' — 2y = 6x + Se'".


32. y" + 6/ + 13>' - 60 cos x + 26.
33. (Z)^ - 3D^ + 4)j - 6 + 80 cos 2x.
34. iD' + D~ I0)y=29e^\
35. (D^ + D^-4D- 4)y = 8x + 8 + 6e-\

In Exs. 36^4, find the particular solution indicated.

36. (Z)^ + 1)>^ = lOf'^''; when x = 0, j = and / = 0.


ans. y = 2(e^* — cos x — 2 sin x).
120 Nonhomogeneous Equations: Undetermined Coefficients [Ch. 7

37. (Z)' -4)y = 2-8x; when x = 0,y = and y' = 5.

ANS. y^e^''—^e~^'' + 2x—l.


38. (Z)' + 3D)y = - 18x; when x = 0, j' = and / = 5.
ANS. y=\+2x—3x^~e~^''.
39. {D^ + 4D + 5)y = lOe"^^; when a: = 0, = 4 >> and / = 0.
d^x dx dx
40. -TT +4—+ 5a: = 10; when r = 0, = a: and —= 0.
dt^ dt dt
ANS. x = 2(1 — e~^' COS — r 2e~^' sin /)•

dx d^x
41. jc + 4x + 5x = 8 sin /; when r = 0, = at and x = 0. Note that a: = ^- x
, = —-r
is a common notation when the independent variable is time.
X == (1 + e~ ^') sin r — (1 — e~^') cos ANS. /.

42. y" -{-9y = ^\x^ + 14 cos 4x; when x = 0, j = and y' = 3.


43. {D^ + 4/)^ + 9D+ \0)y = -24£'''; when x = 0, j = 0, / = -4 and /' = 10.
44. y" + 2y' + 5y = Se'"; when x == 0, >' = and y' = 8.

In Exs. 45-48, obtain from the particular solution indicated the value of y and
the value of >'' at x = 2.

45. >'" + 2>'' + = ^ ^^ ; at X = 0, = — 3, J' and at x = \,y= — 1

ANS. At x = 2, = e~^ j' and y= 1.

46. >>" + 2y' + J = X ; at X == 0, >' = —2 and y' = 2.


ANS. At X = 2, J = 2e~ ^ and y' = — e~^.
I

47. 4y" +y^ 2; at X = TT, y = and y' = \.

^^ - -0.7635 and / = +0.3012.


ANS. At X = 2,

48. 2y" - 5>'' - 3>' = -9x^and y' = 0.- 1 ; at X = 0, J = 1

ANS. At X = 2, ^^ == 5.64 and y' = 5.68.


49. (D^ + D)y = X + 1 when x = 0,y= 1, and when x=\,y= I. Compute the
;

value of >' at X = 4. ans. Ai x ^ A,y =% — e~^ — e~^ — e~^.


50. {D^ + 1)>' = x^ when x = 0, j = 0, and when x = 77, 7 = 0. Show that this
;

initial value problem has no solution.

51. {D^ -r l)^ = 2 cos x; when x = 0, y = 0, and when x = tt, >> = 0. Show that this
initial value problem has an unlimited number of solutions and obtain them.
ANS. y = {c -{- x) sin x.
52. For the equation {D^ + D^)y = 4, find the solution whose graph has at the
origin a point of inflection with a horizontal tangent line.
ANS. J = 4 — 4x + 2x^ — 4^" *.

53. For the equation (D^ — D)y = 2 — 2x, find a particular solution which has at
some point (to be determined) on the x-axis an inflection point with a horizon-
tal tangent line.

ANS. The point is (1,0); the solution is>' = x^ + — 1 2exp (x — 1).

41. Solution by inspection

It is frequently easy to obtain a particular solution of a nonhomogeneous


equation
{b^D" + b,D"-' + •••
+ b„^,D + b„)y = R(x) (1)
§ 41] Sulutiun by inspection 121

by inspection.
For example, if R(x) is a constant Rq and if b„ ^ 0,

Ro
(2)

is a solution of

(^0 D" + byD"-^ + •••


+ b„)y ^ R^;b„^0, 7?o constant, (3)

because all derivatives of yp are zero, so

{boD" + b^D"-' +- + b„)yp = b„ Ro/b„ = Ro


Suppose that b„ = in equation (3). Let D'^y be the lowest-ordered derivative
that actually appears in the differential equation. Then the equation may
be written

(bo /)" + •••


+ ^„_fc D'')y = /?o ;
^«-fc ^ 0, ^0 constant. (4)

Now D'^x'' = k\, a constant, so that all higher derivatives of x* are zero.
Thus it becomes evident that (4) has a solution

yp = TTz
k\b„_k
Rqx''
— '
(5)

for then {boD" + -'- + b„_, D')yp = b„_,Rokl/ik\b„.k) = Ro •

EXAMPLE (a): Solve the equation

(i)' -3D + 2)y = 16. (6)

By the methods of Chapter 6 we obtain the complementary function,

y^ = Cie"" + C2 e^"".

By inspection a particular solution of the original equation is

=
yp 2

Hence the general solution of (6) is

y — c^e"" + (72 e^'' +

EXA M PL E (b) : Solve the equation

+4— =
—-5+4-5
d'y d'y ^
7. (7)
dx^ dx

From the auxiliary equation m^ + 4m^=0 we get m — 0, 0, 0, +2/


Hence
yc = Ci + ^2 ^ + ^3 ^^ + ^4 COS 2x + C5 sin 2x.
122 Nonhomogeneous Equations: Undetermined Coefficients [Ch. 7

A particular solution of (7) is

7x^ 7x^
y,
3 ! 4 24

As a check, note that

, Ix^ 7 •
6
(D^
^ + AD^) = + 4 •
= 7.
'
lA 24

The general solution of equation (7) is

J = (?i + C2 -v + f 3 .Y- + 24.v^ + C4 cos 2.Y + Ts sin 2x,

in which the c\, . . . , c^ are arbitrary constants.

Examination of

{D~ + 4)>' = sin 3.\'


(8)

leads us to search for a solution proportional to sin 3.v, because if j is pro-


portional to sin 3a', so is D^y. Indeed, from

y = A^mZx (9)

we get
D^y = -9 A sin3x,

so (9) is a solution of (8) if

(-9 + 4M = 1;^ = -1.

Thus (8) has the general solution

y = Ci cos 2.V + C2 sin 2x — j sin 3.y.

a result easily obtained mentally.


For equation (8), the general method of undetermined coefficients leads

us to write

tn = +2/, m' = +3/,

and so to write

>p =A sin 3.Y + B cos 3x. (10)

When the >'p of (10) is substituted into (8), it is found, of course, that

A= -i,B^O.
In contrast, consider the equation

{D^ + 4D + 4)y = s\n3x. (11)


§ 41] Solution by inspection 123

Here any attempt to find a solution proportional to sin 3x is doomed to


failure because, although D^y will also be proportional to sin 3a:, the term
Dy will involve cos 3x. There is no other term on either side of (11) to com-
pensate for this cosine term, so no solution of the form y =A sin 3x is

possible. For this equation, m= —2, —2,m'= ±3/, and in the particular
solution

yp —A sin 3a- + B cos 3a:,

it must turn out that B^ 0. No labor has been saved by the inspection
In more complicated situations such as

(D^ + 4)y = X sin 3x — 2 cos 3a:,

the method of inspection will save no work.


For the equation

(Z)2 + 4)y = e^\ (12)

we see, since (Z)^ + 4)e^'' = IQe^"", that

is a solution.
Finally, note that if Ji is a solution of

f{D)y = R,ix)
and ^2 is a solution of

f{D}y = R,(x),
then

yp^yi+ yj
is a solution of
f{D)y = R,(x) + R2{x).

It follows readily that the task of obtaining a particular solution of

fiD)y = Rix)

may be split into parts by treating separate terms of R{x) independently,


if convenient. See the examples below. This is the basis of the " method of
superposition," which plays a useful role in applied mathematics.

EXAMPLE (c): Find a particular solution of

{D^ - 9)y = 3e^ +X- sin 4x. (13)

Since {D^ — 9)e'' — — Se"", we see by inspection that

yi = -ie""
124 Nonhomogeneous Equations: Undetermined Coefficients [Ch. 7

is a particular solution of

{D^ - 9)y^ = Ze\

In a similar manner, we see that yi— —\x. satisfies

{D^ - 9)y2 =X

and that

y^ = TT sin 4x
satisfies

(D^ - %3 = -sin4x.

Hence

yp= - 1^''
-i^ + Tj sin 4x

is a solution of equation (13).

EXAMPLE (d): Find a particular solution of

(Z)^ + %= sin A- + sin 2a-. (14)

At once we see that Ji = i sin x is a solution of

{D^ + 4)ji = sin A'.

Then we seek a solution of

{D^ +4)y2 = sin2x (15)

by the method of undetermined coefficients. Since m= ±2/ and

m' = ±2i,

we put

y^ = Ax sin 2a- + Bx cos 2x

into (15) and easily determine that

4 A cos 2x — 4B sin 2x = sin 2x,

from which A =0, B = — i.


Thus a particular solution of (14) is

y =^sm X — ^x cos 2a:.


§41] Solution by inspection 125

Exercises
1. Show that if b ^a, then

+ a^)y — sin bx
(D^

has the particular solution y = (a^ — b^) sin bx. '

2. Show that the equation

(D^ + a^)y = sin ax


has no solution of the form >' = ^4 sin ax, with A constant. Find a particular

solution of the equation. ANS. y=—-T- cos ax.


la

In Exs. 3-50, find a particular solution by inspection. Verify your solution.

3. {D^ + A)y= 12. 4. (Z)^ + 9)7= 18.


5. (Z)^ + 4Z) + A)y = 8. 6. (Z)^ + 2Z) - 3)7 = 6.
7. (£)^ - 3D + 2)>' - -7. 8. (Z)* + 4Z)=' + 4)7=-20.
9. (D^ + 4Z))>'= 12. 10. (Z)3 - 9D)y = 27.
11. (£>-'+ £))>'= 15. 12. {D^ + D)y= -8.
13. {D* - AD^)y = 24. 14. (Z)* + Z)^)7= -12.
15. (Z)5 - /)')>' = 24. 16. {D' - 9D^)y = 27.
17. (Z)^ + 4);^ = 6 sin x. 7 = 2 sin X.
ANS.
18. (Z)' + 4)>'= 10cos3x. = — 2 cos 3x.
ANS. 7
19. (D^ + 4)>'=8x+ - 15e\ 1 ANS. 7 = 2j: + i — 3e'.
20. {D^ + D)y = (>+'ie^\ ANS. 7 = 6a: + i^^*.
21. (D^ + 3Z) - 4)>' = ISe^^ ANS. 7 = Se^"".
22. (£)^ + 2Z) + 5)7 = 4e^ - 10. ANS. 7 = ie'^ — 2.
23. (D^- = le^\
\)y 24. (Z)^- 1)7 = 2a: + 3.
25. (Z)^- = cos2x.
l)>' 26. (Z)^- 1)7 = sin 2a:.
27. (D^+ l)7 = e^+3x. 28. (Z)^+ 1)7 = 5e-^^
29. (Z)^ + 1)>' = -2x + cos Ix. 30. {D^ + \)y = 4e-^\
31. (Z)2 + l)j= 10 sin 4a:. 32. (£)2+l)7= -6e-3^
33. (Z)^ + 2Z)+ 1)7= 12e\ 34. {D^ + 2D+ \)y^le-^\
35. (Z)^-2Z)+ 1)7= 12^-^ 36. (D^-2Z)+ 1)7= 6e-^\
37. (Z)^ -ID- 3)7 = e^ 38. (Z)^ - 2Z) - 3)7 = e^\
39. (4Z)^+ 1)7= 12 sin X. 40. (4Z)^+ 1)7= -12 cos a:.
41. (4Z)^ + 4Z) + 1)7 = ISe^ - 5 42. {AD^ + AD +\)y = !€-=' + 1.

43. (Z)2- 1)7 = e-^ 44. (D^- 1)7 = 4 -3a:^


45. (Z)^ - Z))7 = e^\ 46. (Z)* + 4)7 = 5e^\
47. (Z)^ + 4)7 = 6 sin 2x. 48. (Z)^ + 4)7 = cos 2a:.
49. (Z)^ - Z))7 = 5 sin 2x. 50. (Z)^ - Z))7 = 5 cos 2x.
CHAPTER 8
Variation of Parameters

42, Introduction

In Chapter 7 we solved the nonhomogeneous hnear equation with constant


coefficients

(bo D" + b,D"-' + • + b„_,D + b„)y = R(x) (1)

by the method of undetermined coefficients. We saw that this method would


be applicable only for a certain class of differential equations: those for which
R{x) itself was a solution of a homogeneous linear equation with constant

coefficients.

In this chapter we shall study two methods that carry no such restrictions.

In fact, much of what we do will be applicable to linear equations with


variable coefficients.
We begin with a procedure by D'Alembert that is often called the method
of reduction of order.

126
§ 43] Reduction of order 127

43. Reduction of order

Consider the general second-order linear equation

y" + py' + qy = R. (1)

Suppose we know a solution y = yi of the corresponding homogeneous


equation

y" + py' + qy = 0. (2)

Then the introduction of a new dependent variable v by the substitution

y = yiv (3)

will lead to a solution of equation (1) in the following way.


From (3) it follows that

/^ + y'lV,
yiv'
y" = yiv" + 2y\v' + y'lv,

so substitution of (3) into (1) yields

yiv" + 2y\v' + >'> + pyy + py\v + qy^v = R,

or

yiv" + (2y\ + pyi)v' + (y" + py\ + qyi)v = R. (4)

But y = yi is a solution of (2). That is,

y'l + py'i + gj'i =


and equation (4) reduces to

yy + (2y\ + pyi)v' = R. (5)

Now let v' = w so equation (5) becomes

yy +(2y\ + pyi)w^R, (6)

a linear equation of the first order in w.

By the usual method (integrating factor) we can find vv from (6). Then we
can get r from r' = iv by an integration. Finally y = yiV.
Note that the method is not restricted to equations with constant co-
depends only upon our knowing a particular solution of equation
efficients. It

(2); that upon our knowledge of the complementary function. For practical
is,

purposes, the method depends also upon our being able to effect the inte-
grations.
128 Variation of Parameters [Ch. 8

EXAMPLE (a): Solve the equation

y" -y = e"". (7)

The complementary function of (7) is

We shall take the particular solution e"" and use the method of reduction of
order by setting

y = ve^.

Then
y' = ve'' + y'e"",

and
y" = ve^ -{ Iv'e^ -\- v"e''.

Substituting into equation (7) gives

r" + 2r'-l. (8)

Equation (8) is a first-order linear equation in the variable v' . Applying


the integrating factor e^^ yields

e2^(r" + 2v') = e^\


Thus
e^^v' = ^e^'^ + c, (9)

where c is an arbitrary constant. Equation (9) readily gives

and hence
V = c^^e'^^ + C2 + jx,

where Cj and c 2 ^''^ arbitrary constants.

Remembering that y = re*, we finally have

y = c^e'^^ + CtC^ + jxe^.

Of course, the solution to equation (7) could have been obtained by the
method of undetermined coefficients. Let us now solve a problem not solvable
by that method.

EX A M P L E (b) : Solve the equation

(D^ + l)y^cscx. (10)

The complementary function is

y^ = Ci cos X + C2 sin x. (1 1)
§ 43] Reduction of order 129

We may use any special case of (11) as the y^ in the theory above. Let us
then put

y = V sin x.

We find that

y' = v' sin x + v cos x

and
y" = v" sin x + 2v' cos x — j; sin x.

The equation for v is

v" sin X + 2v' cos x = esc x,

or
y" + 2r' cot jc = csc^ X. (12)

Put v' = w; then equation (12) becomes

H'' + 2m' cot X = CSC^ X,

for which an integrating factor is sin^ x. Thus

sin^ X dw + 2w sin x cos x dx — dx (13)

is exact. From (13) we get

w sin^ X = X,

and if we seek only a particular solution, we have

or

Hence

V =\ X csc^ X dx,

or

V = —X cot A' + In |sin x\,

a result easily obtained using integration by parts.


Now
y = V sin x,

so the particular solution which we sought is

yp= —X cos X + sin x In |sin x\.

Finally, the complete solution of (10) is seen to be

y = c^ cos X + Cjsmx — X cos x + sin x In |sin x|


130 Variation of Parameters [Ch. 8

Exercises
Use the method of reduction of order to solve the following equations.

l.(D'—\)y = x. ANS. y^Cie^ + Cze'^ — x.


2. (D^-3D + 2)y=--e''. ans. y = Cie^" + Cze" - xe".
3. (D^ -6D + 9)y = e\ ans. See Ex. 2, Section 40.
4. (Z)' + 9)y = sin x.
5. Use the substitution y = v cos x to solve the equation of Example (b) above.
6. Use y = ve'" to solve the equation of Example (a) above.
7. (D^ + 1)7= sec X.
8. (D^ + l)y = sec^jc. Use ^^ = sin x. jj

9. (D^ +!)>' = csc^x. Take a hint from Ex. 8.


10. (D^ + 2D+ l)y = ie^-l)-\ ans. y =^ e-'id + CzX - ln\l - e'^H
11. (Z)^ -3D + 2)y = (1 + e^^)-"\
12. Verify that j = e* is a solution of the equation

(x- l)y"-x/ + y = 0.

Use this fact to find the general solution of

(x ~ \)y" — xy' + y= \.

ANS. y = CiX + Cze" + \.

13. Observe that y = xisa particular solution of the equation

2x^y" + xy' — j =
and find the general solution. For what values of x is the solution valid?
ANS. y = CiX + CzX'^'^.
14. In Chapter 19 we shall study Bessel's differential equation of index zero

xy" + y' + xy = 0.
Suppose that one solution of this equation is given the name Jo{x). Show that a
second solution takes the form
r dx
ax

15. One solution of the Legendre differential equation

(1 - x^)y" - 2xy' + 2y =
l+x
is y = X. Find a second solution. ans. y = — 2 + In
l-x

44. Variation of parameters

In the previous section we saw that if ji is a solution of the homogeneous


equation
y" + p(x)y' + qix)y = 0, (1)
§ 44] Variation of parameters 131

then we can use it to determine the general solution of the nonhomogeneous


equation

y" + p(x)y' + q(x)y = R{x). (2)

In using the method of reduction of order we proceeded as follows. Since

7, is a solution of (1), the function c^yi is also a solution for an arbitrary


constant c^. We replaced the constant c, by a function v{x) and considered
the possibility of the existence of a solution of equation (2) of the form
V •
yi- This led us to a first-order linear equation in the variable v' that we
were able to solve.

Suppose now that we know the general solution of the homogeneous equa-
tion (1). That is, suppose

yc^ciyi + c2y2 (3)

is a solution of (1), where y^ and y2 are linearly independent on an interval


a < X <b. Let us see what happens if we replace both of the constants in (3)
with functions of x. That is, we consider

y = Ay,+ By 2 (4)

and try to determine A{x) and B{x) so that Ay^^ + By 2 is a solution of equa-
tion (2).

that we are involved with two unknown functions A{x) and B{x) and
Note
that we have only insisted that these functions satisfy one condition: the
function in (4) is to be a solution of equation (2). We may therefore expect to
impose a second condition on A{x) and B{x) in some way which would be to
our advantage. Indeed, if we simply impose the condition B{x) = 0, then we
will be dealing with the method of reduction of order. Actually we impose

a somewhat different condition on A and B.


From (4) it follows that

y' = Ay'i + By'2 + A'y^ + B'y^ (5)

Rather than become involved with derivatives of ^4 and B of higher order than
the first, we now choose some particular function for the expression

A'yi + By 2 .

Technically, we could let this function be sin x, e^, or any other suitable
function. For simplicity we choose

^>i+5>2=0. (6)

It then follows from (5) that

y" = Ay'i + Byl + A' y\ + B' y'2 .


(7)
132 Variation of Parameters [Ch. 8

Since y was to be a solution of (2) we substitute from (4), (5), and (7) into

equation (2) to obtain

My"i + py'i + qyi) + 5(/2 + py'i + qyi) + A'y\ + B'y'^ = r(x).

But ji and jj are solutions of the homogeneous equation (1), so that finally

A'y\ + B'y', = /?(x). (8)

Equations (6) and (8) now give us two equations that we wish to solve for

A' and B'. This solution exists providing the determinant

^'i yz

y'l y'l

does not vanish. But this determinant is precisely the Wronskian of the
functions y^ and yi , which were presumed to be linearly independent on the
interval a <x< b. Therefore, the Wronskian does not vanish on that
and we can find A' and B' By integration we can now find A and B.
interval .

Once A and B are known, equation (4) gives us the desired y.


This argument can easily be extended to equations of order higher than
two, but no essentially new ideas appear. Moreover, there is nothing in the

method that prohibits the linear differential equation involved from having
variable coefficients.

EXAMPLE (a): Solve the equation

(Z)^ + \)y = sec A' tan x. (9)

Of course,

y^ = Ci cos X + C2 sin x.

Let us seek a particular solution by variation of parameters. Put

y = Acos X + B sinx, (10)

from which
y' = —A sin x + B cos x + A' cos x + B' sin x.

Next set

y4' cos A + jB' sin A- = 0, (11)

so

y' = —A sin X + B cos x.

Then
y" = —A cos A — 5 sin x — A' sin a + B' cos a. (12)
§ 44] Variation of parameters 133

Next we eliminate y by combining equations (10) and (12) with the original
equation (9). Thus we get the relation

— A' sin X + B' cos x = sec x tan x. (13)

From (13) and (11), A' is easily eliminated. The result is

B' = tan X,
so

5= ln|secx|, (14)

in which the arbitrary constant has been disregarded because we are seeking
only a particular solution to add to our previously determined complementary
function y^
From equations (13) and (11) it also follows easily that

A' = — sin X sec x tan x,


or

A' = — tan^ X.
Then
/I = — J
tan^ X dx = J(l — sec"^ x) dx,
so
A=X— tan X, (15)

again disregarding the arbitrary constant.


Returning to equation (10) with the known A from (15) and the known B
from (14), we write the particular solution

yp = (x — tan x) cos x + sin x In |sec x\,


or

yp = X cos X — sin X + sin x In |sec x|.

Then the general solution of (9) is

y = Ci cos X + C3 sin X + X cos x + sin x In |sec x|, (16)

where the term (— sin x) in yp has been absorbed in the complementary


function term c^ sin x, since C3 is an arbitrary constant.
The solution (16) can, as usual, be verified by direct substitution into the
original differential equation.

EXAMPLE (b) : Solve the equation

(D'-3D + 2)y = -^^ (17)

Here
y, = Cie'' + C2 e^\
134 Variation of Parameters [Ch. 8

SO we put

y = Ae'' + Be^\ (18)


Since

y' = Ae"" + IBe^"" + A'e"" + B'e^\

we impose the condition

A'e"" + E'e^"" = 0. (19)


Then
y' = Ae"" + lBe^\ (20)

from which it follows that

y" = Ae'' ^-ABe^"" + A'e"" + 2B'e^\ (21)

Combining (18), (20), (21), and the original equation (17), we find that

1 +e-^
Elimination of B' from equations (19) and (22) yields

Then

Similarly,

so

B
or
§ 45] Solution of j" + J = f(x) 135

45. Solution of y" + y =f(^x)

Consider next the equation

iD' + \)y=f{x), (1)

in which all that we require off{x) is that it be integrable in the interval on


which we seek a solution. For instance, /(x) may be any continuous function
or any function with only a finite number of finite discontinuities on the
interval a ^ x ^ b.

The method of variation of parameters will now be applied to the solution


of(l). Put

y =A cos X + B sin x. (2)


Then
y' — —A sin x + B cos x -^ A' cos x -\- B' sin x,

and if we choose
A' cos X + B' sin x — 0, (3)
we obtain
y" = —A cos X — 5 sin a: — /I' sin X + B' cos x. (4)

From (1), (2), and (4) it follows that

— A's\rvx + B'cosx=f{x). (5)

Equations (3) and (5) may be solved for A' and B', yielding

A' = —/(.y) sin X, B' =f{x) cos x.

We may now write

A^- ffiP) sin p dp, (6)

B= ffiP) cos pdl3, (7)


•'a

for any x « ^ x ^ 6. It
in is here that we use the integrability of/(x) on the
interval a ^ ^ b.
x
The A and B of (6) and (7) may be inserted in (2) to give us the particular
solution

yp= -cos X f fiP) sin fi d^ + sin x I f{P) cos p dp


^a *'a

= /(/^)[sin X cos p - cos x sin jS] ^j9. (8)


J
136 Variation of Parameters [Ch. 8

Hence we have

y^= ff(P) sin {x-P) dp, (9)

and we can now write the general solution of equation (1):

y = CiCOsx + Cj sin x + f /(/?) sin (x - j?) d^. (10)


•'a

Exercises
In Exs. 1-18, use variation of parameters.

1. {D^ + \)y = CSC X cot X.


ANS. 7 = Ci cos X + C2 sin x — sin — cos x In |sin
AT jr Ar|

1. {D^ + \)y = coix. ANS. = — sin X In |cscx + cot x|.


>' j'c

3. (Z)^ + 1)>' = sec X.

ANS. >' = Ci cos X + C2 sin X + X sin X + cos x In |cos x|.


4. (Z)^ + = sec^ X.
!)>' ANS. j = — + sin x in |sec x tan x|.
>'c 1 -t-

5. {D^ +\)y = sec^ x. ans. y = yc+\sQC x.


6. (D^+ l)>' = sec''x.
ANS. y — yc— h+ i sec^ x + ^ sin x In |sec x + tan x|.
7. (Z)^ + !)>'= tan X. ans. ;^ = >'c — cosx In |secx + tan x|.
8. (Z)^ + l)y = tan^x.
ANS. y = yi — 1, where ji is the solution of Ex. 4 above.
9. (Z)^ + = sec X CSC x.
I);'

ANS. y = yc — cos x In |sec x + tan x| — sin x In |csc x + cot x|.


10. (D^ + 1)>' = sec^ X CSC x. ans. = Jc — sin x In jcsc 2x + cot 2x|
^^

11. (D^ -1D+ \)y = e^^ie" + 1)" \ ans. y = yc + e']ni\+ e").


12. (D^ -3D + 2)y = e^^/(l + e^^.
ans. y = yc + e" Arctan {e'") — le^" In (1 + e~ ^^).
13. (Z)^-3Z) + 2);' = cos(e-^). ans. y = yc- e^'' cos {€'%
14. {D^- 1)>' = 2(1 -e-^^)-^'^
ANS. y= Cic" + Ci^"'' — e" Arcsin {e'") — (1 — e'^^'Y'^.
\S. {D^ — \)y = e~ sin e~ ^''
""
ans. y = yc — sin e'"" — cos e'"'. e''

16. {D - 1)(I> - 2)(Z) - 3)>' = e\ ans. = + ixe^ >' >'c

17. j'" — = X.
>''

18. >'"'
+ y' = tan x.
19. Observe that x and ^"^ are solutions of the homogeneous equation associated
with
(1 - x)y" + xy' -y= 2(x - lye''.

Use this fact to solve the nonhomogeneous equation.


20. Solve the equation
y" — y = e"
by the method of variation of parameters, but instead of setting A'yi + B'yz =
(6), Section 44, choose A'yi + B'yz = k, for constant k.
as in equation
21. Apply the suggestion of Ex. 20 to Ex. 5 above.
Miscellaneous exercises 137

22. Let yi and yz be solutions of the homogeneous equation associated with

y" + p(x)y' + q(x)y = f(x). (A)

Let W(x) be the Wronskian of >'i and yz and assume iV(x) , 7^ on the interval
a <x <b. Show that a particular solution of equation (A) is given by

f"/()3)L)'.(ft/«W - y,My2m dp
''']. WW)

<^>

23. The conditions of Ex. 22 imply that

y'i + py'i +qyi = (C)

and

y'l + py'i + qyi = 0. (D)

If we multiply equation (C) by yz and equation (D) by yi and then subtract the
two equations, we obtain

(yz y'i — yiy'i) + piyi y'l — yiy'i) = 0.


From this equation show that the Wronskian of >'i and yz can be written

W{x) = cQ\p[-\p dx], (E)

where c is constant. Equation (E) is known as Abel's formula.


24. Conclude from Ex. 23 that if W{xo) = for some Xo on the interval a <x <b,
then Wix) = for all a < a: < b.
25. Solve the initial value problem

y" + y =f(x); when x = xo,y = yo, and y' = yo.


Hint: Show that the constant a in equations (6) and (7) above could have
been chosen to be Xq. Determine the Ci and Cz of equation (10) by using the
form of yp in equation (8).

ANS. y = yo cos (x — Xo) + y'o sin (x — Xo)+ |


/(jS) sin (x — /3) d^.

Miscellaneous exercises
1. iD^-l)y=2e-%l + e-^'')-^. ans. y = yc- xe'"' - ie-' \n(l + e-^").
2. (D^-l)y = (l-e^'')-''\ ans. y = y,-(l-e^^y'\
3. (D^ — l)y = e^%3 tan e^ + e'' sec^ e"). ans. y = yc + e''ln |sec e"].
4. (D^ + 1)^ = sec^ X tan x.
ANS. y = yc+Hanx+icosx\n |sec x + tan x\.
5. Do Ex. 4 by another method.
6. (D^ + 1);^ = CSC X. Use variation of parameters.
ANS. See Example (b) of Section 43.
7. (Z)^ -3D + 2)y = sec^ (e-"). ans. _>; = Ci e' + C2 e^^ + e^^ in |sec (e-^)|.
138 Variation of Parameters [Ch. 8

8. Do Ex. 7 by another method.


9. (D^- \)y = 2Kl +e^).
ANS. y = yc — — xe" + \ (e"" — e~'') In (1 + e").
10. (Z)^ + D)y ^ sec^ x. Hint: integrate once first.

ANS. y = Ci + C2 cos jc + C3 sin X — cos x In |sec x + tan x\.

11. (D^- l)y = 2lie''-e-').


ANS. y = yc — xe'" + — e'") In — e'^"].
lie" |1

12. (D^ — 3D + 2)y = sin e" '. ans. = - e^^ sin e"
j^ >'c *.

13. (D^ — l)y = llie^" + 1). ans. y = yc—i — cosh x Arctan e'".
14. y" + y = sec^ x tan x. ans. y ^ yc + ^ sec x tan x.
15. >'" + = sec X tan^ x. Verify
>' your answer.
16. /' + 4y' + 3>' = sin e". ans. y = yc — e'^'' sin e" — e~ ^* cos e".
17. y" + y = csc^ x cot x. ans. = >'c + i cot x csc x.
^^
CHAPTER !/

Inverse Differential Operators

46. The exponential shift

In Chapter 5 we studied some of the properties of the algebra of linear


differential operators with constant coefficients. We found this algebra
useful in finding solutions of homogeneous linear equations. In this chapter
we show briefly how differential operators may be used to find particular
solutions for nonhomogeneous linear equations.
As a first illustration we make use of the exponential shift theorem that
was derived in Section 32

e"YiD)y=fiD-a)[e''y], (1)

where /(Z)) is a linear differential operator with constant coefficients.

EXAMPLE (a): Solve the equation

(Z)2 -2D + 5)y = 16jc3e'^ (2)

139
140 Inverse Diflferential Operators [Ch. 9

Note that the complementary function is

>'c
= CyC^ cos 2x + C2 e^ sin 2x. (3)

We can conclude also that there is a particular solution,

yp = Ax^e^"" + Bx^e^"" + Cxe^"" + Ee^\ (4)

which can be obtained by the method of Chapter 7. But the task of obtaining
the derivatives of Vp and finding the numerical values of A, B, C, and £ is a
little tedious. It can be made easier by using the exponential shift (1).

Let us write (2) in the form

e-^^D^ -2D + 5)y= 16x^


and then apply the relation (1), with a= — 3. In shifting the exponential
e'^"" from the left to the right of the differential operator, we must replace

D by (Z) + 3) throughout, thus obtaining

[(D + 3)' - 2{D + 3) + 5]{e-^y) = \6x\


or
(D^ + 4D + S)ie-^y) = 16a-^ (5)

In equation (5), the dependent variable is (e'^'^y). We know at once that


(5) has a particular solution of the form

e'^'^yp = Ax^ + Bx^ + Cx + E. (6)

Successive differentiations of (6) are simple. Indeed,

Die-^'^yp) = 3Ax^ + 2Bx + C,

D\e-^''yp)^6Ax + 2B,

so from (5) we get

eAx +2B+ \2Ax^ + %Bx + 4C + %Ax^ + ^Bx^ + %Cx + 8£ = 16x^


Hence
M= 16,

+ 85=
\2A 0,

6/1 + 85 + 8C = 0,

2B + 4C+SE= 0,

from which ^ = 2, 5= -3, C= f , £= 0.

Therefore

or
y. = {2x^ - 3x^ + f;c)e'^
§ 46] The exponential shift 141

and the general solution of the original equation (2) is

y = Cic'' cos 2x + C2 e^ sin 2x + {2x^ — 3x^ + ix)e^''.

The complementary function could also have been obtained from (5).

EXA M PLE (b) : Sol ve the equation

(Z)' -2D+ ])y = xe"" + Ix - 2. (7)

Here the immediate use of the exponential shift would do no good, because
removing the e'' factor from the first term on the right would only insert a
factor e"* in the second and third terms on the The terms 7a; — 2 on
right.

the right give us no trouble as they stand. Therefore we break (7) into two
problems, obtaining a particular solution for each of the equations

(D' - 2Z) + \)y^ = jce" (8)

and
(Z)' - 2Z) + 1)V2 = 7a- - 2. (9)

On (8) we use the exponential shift, passing from

e-\D - l)Vi =X
to

D\e-yy) = X. (10)

A particular solution of (10) is easily obtained:

e'^'yi = ix\
so

y,=ix'e\ (11)

Equation (9) is treated as in Chapter 7. Put

y2 = Ax + B.

Then Dy2 = A, and from (9) it is easily found that A= 1, B= \2. Thus a
particular solution of (9) is

y2 = lx+ 12. (12)

Using (1 1), (12), and the roots of the auxiliary equation for (7), the general
solution of (7) can now be written. It is

y = {ci-\- C2 x)e'' + ^x^e"" + 7x + 12.

EXAMPLE (c): Solve the equation

D\D + Afy = 96^-'^*. (13)


142 Inverse Differential Operators [Ch. 9

At once we have m = 0,0, —A, —4 and m' = — 4. We seek first a particular


solution. Therefore we integrate each member of (13) twice before using the
exponential shift. From (13) it follows that

(Z) + 4)Vp = 6e-^^ (14)

the constants of integration being disregarded because only a particular


solution is sought. Equation (14) yields

e^\D + Afy^ = 6,

Thus the general solution of (13) is seen to be

y = Ci + C2 X + (cj + C4 X + 3x^)e~'*''.

The exponential shift is particularly helpful when applied in connection


with terms for which the values of m' (using the notations of Chapter 7) are
repetitions of values of m.

Exercises
In Exs. 1-12, use the exponential shift to find a particular solution.

1. (D - 3) V = e^"- ANS. y = ix^e^'


2. {D-iyy = e\
3. iD + 2yy=^l2xe-^\ ans. y^lx^e-^"
4. {D+\yy=3xe-\
5. (D- lyy = 6xe^\ ans. y = ^x^e^"
6. (D+ 4yy^Sxe-''\
7. iD+ 3yy= l5x^e-^\ ans. y^^x^e-^"
8. (D-4yy= \5x^e*\
9. D^iD-2yy=l6e'"'. ans. y = 2x^e^''
10. D\D + 3yy^9e-^\
11. (D^- D- 2)y = 18xe-'. ans. y = -(3x^ + 2x)e-''
12. {D^- D- 2)y = 36xe^^
In Exs. 13-18, find a particular solution, using the exponential shift in part of your
work, as in Example (b) above.

13. (D - 2yy = 20 - 3x^^^ ans. y=5- hx^e^\


14. (Z) - 2)V =- 4 - 8x + exe^".
15. y" -9y = 9(2x - 3 + Axe^"). ans. >' =3- 2x + (3x^ - x)e^\
16. /' + Ay' + 4^ = 4x - 6^" ^* + 3e\
17. (£>+ l)^>' = e-^+3x. ans. y = kx^e'" + "ix - 6.
18. (£)^-4)>'= 16xe-^^ + 8x + 4. ans. >'= -(2x + l)(xe-^* + 1).
§ 48] Evaluation of [1 //(£>)] e"' 143

In Exs. 19-28, find the general solution.

19. y" — Ay ^ %xe^''. ans. y = Cie~ ^^


+ (c2 — ix + x^)e^''.
20. y"-9y= -12xe-''\
2\. D(D+\yy = e-\ ans. y = c, + {d^ dx - iiX^)e-\
22. D\D-l)^y=^le^\
23. J'"+ 2>'' + = 48e~*cos4jc.
>' ans. y = {ci + C2X—'icosAx)e~''.
24. /' + Ay' -^Ay= ISe" ^^ cos 3x.
25. (Z) — O^j' = e" sec^ x tan jc. ans. y = e^id + CiX + i tan x).
26. {D^ + AD + A)y= -x-^e-^\ ans. y = e'^'id + C2X + \n\x\).
27. (Z) - aYy = e'Tix). ans. >^ = e'"'[c, + C2 x + /(x)].
28. {D^ + lD+n)y^ e- ^^ sec^ x{\ + 2 tan jc).

ans. y = Cie~'^''-\-e~^''{c2 + i^nx).

47. The operator 1//(D)

In seeking a particular solution of

f{D)y = R(x), (1)


it is natural to write

y=j^^RM (2)

and to try to define an operator l/f(D) so that the function v of (2) will have
meaning and will satisfy equation (1).

Instead of building a theory of such inverse differential operators, we shall

adopt the following method of attack. Purely formal (unjustified) manipu-


lations of the symbols will be performed, thus leading to a tentative eval-
uation of

After all, the only thing that we require of our evaluation is that

fiD)-j^^R(x)^Rix). (3)

Hence the burden of proof will be placed on a direct verification of the


condition (3) in each instance.

48. Evaluation of [l//(D)]e°^

We proved (Section 32) with slightly different notation that

fiD)e''^ = e-Yia) (1)


144 Inverse Diflferential Operators [Ch. 9

and
iD-a)"ix"e'"') = n\e''\ (2)

Equation (1) suggests

1 e"'
^"=7Z^' /(a)^0. (3)
/(£>) /(a)

Now from (1) it follows that

f{a) f{a)
Hence (3) is verified.

Now suppose that f{a) = 0. Then f{D) contains the factor {D — a).

Suppose that the factor occurs precisely n times in /(/)); that is, let

/(Z)) = 0(Z)X/)-a)";0(a)^O.
With the aid of (2) we obtain

<t){D){D - ar(.vV^) = (f){D)n\e''\

from which, by (1), it follows that

(l){D){D - a)"{x"e"'') = n\(f)ia)e''\ (4)

Therefore we write

1 xV""
(t>(D){D-ay
^"=^T^,
nl(f>{a)
<t>(a)^0, (5)

which is easily verified. Indeed,

<j>(D)(D - of ——- = ^\ = e-.


/7 10(a) ni(p{a)

Note that formula (3) is included in formula (5) as the special case, n = 0.

See also Ex. 19, page 147.

EXAMPLE (a): Solve the equation

(Z)2 + \)y = e^\ (6)

Here the roots of the auxiliary equation are m= ± /. Further,

/(/)) = (Z)^ + 1)

and
/(2) 9^ 0.

Hence, using (3),


1 , e^^ 1 ,

•^p
£)' + 1 2^ + 1 5
§ 49] (D2 + a2)-i gin ax and (D^ + a^y cos ax 145

SO the solution of (6) is

y = Ci cos X+ C2 sin x+ ^e^''.

EXAMPLE (b): Solve the equation

D^(D- l)\D+ l)y = e\ (7)

Here m= 0, 0, 1, 1, 1, -1. To get a particular solution of (7), we use


formula (5) with « = 1, a7 = 3, 0(Z)) = D\D + 1). Then
0(1) =1^-2

and a particular solution of (7) is given by

1
._ xV _ 1 3^
^^ {D-{)'D'(D+l)^ 12-2 ^
'

3! n""
Then the general solution of (7) is

y = Ci + C2 X + C3 e""" + (C4 + C5 X + C6 X^ + y 2-X^K.

49. Evaluation of (D^ + a^)"* sin ax and (D^ + a^)"* cos ax

No special device is needed for the evaluation of

(D^ +a^y^ sinbx


when b j^ a. \n fact, it is easy to show that

1 . . sin bx
sin bx = -2 7-2', b ^ a,
D^ + a^ a'-b'
and a similar result holds for the expression (D^ + a^)~^ cos bx, when b ^ a.

Consider next the evaluation of

1
sin ax. (1)
j)2_^^2

The formulas of the preceding section can be put to good use here, since

sin ax = .

2/
Then

-,
D^ + a-
1.1 ^ sni ax =
2i(D-ai){D +
1

ai)
(e""' - e""'*)

~ 2f ll!2a/
~ l!(-2a/)/
,aix ^ — aix
X e"'-' +
1

2a 2
146 Inverse DiflFerential Operators [Ch. 9

SO

1 X
— = sin ax = — —- cos ax. (2)^
D^ + a^ 2a ^

The verification of (2) is left as an exercise.

Another useful result,

1 X
_, 7 cos ax = -— sin ax, (3)
^
D^ + a^ 2a

can be obtained in the same way.

Oral Exercises
Give the general solution of each equation.

1. (D^ —!)>' = ^^^ y = Cie'' + Cze-'' + ^e^\


ANS.
2. (D^ — \)y = e\ y = Cie' + C2e~'+^xe\
ans.
3. (D^ + l)y = sin x. ans. y = Ci cos x + Cisinx — ^x cos x.
4. (D^ + 4)y = cos 2x. ans. >' = Ci cos 2x + Cz sin 2x + Jx sin 2x.

5. (Z)^ ^ 9)y = e^^ 6. (Z)^ + 4)y = e'\


7. (4D^ + \)y=^e-^\ 8. Z)(Z) - 2)>' = e-^
9. D{D-2Yy = e^\ 10. Z)(Z) + S)^^ = ^-3-.
11. (D^ + A)y = cos 3a:. 12. {D^ + 9)^ = cos 3x.
13. (Z)^ + 4)>^ = sin 2x. 14. (D^ + 7>€)y - sin 6a:.
15. (Z)^ + 9)>' = sin 3x. 16. {D^ + 26)y = cos 6jc.

Exercises
1. Verify formula (2) of Section 49.
2. Obtain and verify formula (3) of Section 49.

In Exs. 3-18, find the general solution.

3. (Z)' + 3Z)-4)>'= 12e^\ ans. y = Cie-*^ + C2e'= + 2e^''

4. {D^^'iD-A)y = 2\e^\
5. (Z)^ + 3Z) - A)y = ISe". ans. y = de'*" + (cj + 3.vK
6. (£>^ + 3Z)-4)>' =20e-*\
7. (Z)^ - 3£) + 2)j =
e" + e^\ ans. y = (ci- x)e'' + (cz + x)e'''
8. (4D^- !):>;= e^/2+ \2e\
9. Z)HZ)-2)3>' = 48e'^ ANS. >' = Ci + C2A: + (c3 + C4A: + f5X^ + 2A:^)e'2^
10. (Z)* - 18Z)^ + 81)7 =- 36e^^
11. {D^ + 16)^ = 14 cos 3a:. ans. y = Ci cos 4a- + Ci sin 4a' + 2 cos 3Ar

12. (4Z)2 + !):>' - 33 sin 3a:.

13. y" + 16j = 24 sin 4a:. ans. y={ci — 3x) cos 4x + Ci sin 4Ar
14. >»"+ 6>' = 48 cos 4a:.
1

15. y" + y= 12 cos 2x — sin x. ans. J'


= (ci + ^x) cos at + Cz sin a: — 4 cos 2x
§ 49] (D2 + a2)-» sin a;c and (D^ + a^)' cos ax 147

16. y" + y = sin 3a: + 4 cos x.


17. (Z)^ — 2D + 5)y = e" cos 2x. Hint: use the exponential shift followed by
formula (3) of Section 49.
ANs. y = €"(€1 cos 2x + cz sin 2jr) + iJCf-' sin 2x.
18. (Z)^ + 2Z> + 5)>' = e-'' sin 2x.

ANS. y = e~ "(ci cos 2Ar + Cz sin 2jc) — ixe~ " cos 2^.
19. Prove that if /(a) = (at - ay(f>{x), then /'"'(o) = «! (/)(o). Then use equation (5)
of Section 48 above to prove that

/(D) /'"'(a)'

where n is the smallest non-negative integer for which /<"'(a) ^ 0. See C. A.

Hutchinson, " Another Note on Linear Operators," Amer. Math. Monthly, 46


(1939), 161.

In Exs. 20-23, use the formula of Ex. 19 above.

20. Ex. 5. 21. Ex. 6.


22. Ex. 9. 23. Ex. 10.

In Exs. 24-29, verify the formulas* stated.

— D) sin ax
= f{
1

24. —^ sin ax
J^-^J^—-^
.nam-ai) ^ 0.

25. — 1

cos ax =
fi—D) cos OAT
J^^r^j^Z-f ^fiaOfi-ai) ^ 0.

26. — 1

sinh ax =
/(-/)) sinhflx
-J^-^J^—^
; /(«)/("«) ^ 0.

27. — 1
cosh ax =
f{ — D) cosh oa:
^(^)^(_^) ; mn-a) ^ 0.

1 a:"
^^-
rm~^ — i7« cos ^^ = ,-, ,„ ,
cos (ax - ^rt77).

In Exs. 30-33 use the formulas of Exs. 24-25 above.

30. Ex. 1, page 119. 31. Ex. 11, page 119.


32. Ex. 12, page 1 19. 33. Ex. 32, page 119.

* The formulas of Exs. 24-27 were obtained by C. A. Hutchinson, "An Operational


Formula," Amer. Math. Monthly, 40 (1933), 482-483; those of Exs. 28-29 were given by
C. A. Hutchinson, "Note on an Operational Formula," Amer. Math. Monthly, 44 (1937)
371-372.
CHAPTER lU
The Laplace Transform

50. The transform concept

The reader is already familiar with some operators that transform functions

into functions. An outstanding example is the differential operator D which


transforms each function of a large class (those possessing a derivative) into
another function.
We have already found that the operator D is useful in the treatment of

linear differential equations with constant coefficients. In this chapter we


study another transformation (a mapping of functions onto functions) which
has played an increasingly important role in both pure and applied mathe-
matics in the past few decades. The operator L, to be introduced in Section
51, is particularly effective in the study of initial value problems involving

linear differential equations with constant coefficients.

One class of transformations, which are called integral transforms, may be


defined by

T{F{t)}= C K(s,t)F(t)dt=f(s). (1)


''-X

148
§ 51] Definition of the Laplace transform 149

Given a function K(s, t), called the kernel of the transformation, equation

(1) associates with each F{t) of the class of functions for which the above
integral exists a function /(j') defined by (1). Generalizations and abstractions
of (1), as well as studies of special cases, are to be found in profusion in
mathematical literature.

Various particular choices of A'(^, /) in (1) have led to special transforms,


each with its own properties to make it useful in specific circumstances. The
transform defined by choosing

K{s,t) = Q, for/<0,
= e~'\ for t ^ 0,

is the one to which this chapter is devoted.

51. Definition of the Laplace transform

Let F{t) be any function such that the integrations encountered may be
legitimately performed on F{t). The Laplace transform of F{t) is denoted
by L{F{t)] and is defined by

L{F(0}= re-^^F{t)dt. (1)


•'o

The integral in (1) is a function of the parameter s; call that function /(5).
We may write

L{F(0}= Ce-^'F{t)dt=f{s). (2)


•'o

It is customary to refer to/(5'), as well as to the symbol L{F{t)], as the trans-


form, or the Laplace transform, of F(/).
We may also look upon (2) as a definition of a Laplace operator L, which
transforms each function F{t) of a certain set of functions into some function
f{s).
It is easy to show that if the integral in (2) does converge, it will do so for
all s greater than* some fixed value Sq. That is, equation (2) will define /(5')

for 5 > 5o . In extreme cases the integral may converge for all finite s.

It is important that the operator L, like the differential operator D, is a


linear operator. If Fi(/) and Fjit) have Laplace transforms and if q and Cj

are any constants,

L{c,F,{t) + c^F^it)) = cMPm + c,L{FAt)). (3)

* If 5 is not to be restricted to real values, the convergence takes place for all s with real
part greater than some fixed value.
150 The Laplace Transform [Ch. 10

Using elementary properties of definite integrals, the student can easily show
the validity of equation (3).

We shall hereafter employ the relation (3) without restating the fact that

the operator L is a linear one.

52. Transforms of elementary functions

The transforms of certain exponential and trigonometric functions and of


polynomials will now be obtained. These results enter our work frequently.

EXAMPLE (a): Find Lie*'}.


We proceed as follows:

-(s-k)t
dt.

For ^ k, the exponent on e is positive


s or zero and the integral diverges.
For > k, the integral converges.
s

Indeed, for s > k,


.CO

L{/'}= I
e-"-'^'dt
•'o

— e -(s-fc)t-i OC

s-k
1
= +
s — k
Thus we find that

L{e'''}=-^, s>k. (1)


s-k
Note the special case A: = :

L{1]=-, 5>0. (2)


s

EXAMPLE (b): Obtain L{sin A:r}.

From elementary calculus we obtain

r e"(a
e {a sin mx — m cos mx)
e sm mx ax = ^
^ h C.
J m
a^ + m^
Since
.00

L{sin kt} = \
e ^'
sin kt dt.
§52] Transforms of elementary functions 151

it follows that

€'"'( —s sin kt — k cos kt)


L{sin kt] (3)
s^ + k'

For positive s,e ^' -> as / -> oo. Furthermore, sin kt and cos kt are bounded
as / ^ 00. Therefore (3) yields

1(0 - k)
L{sin kt) =
s' + k''
or

L{sin kt] = s>0. (4)


s^ + k'

The result

L{cos kt} — s>0. (5)


s'+k''
can be obtained in a similar manner.

EXAMPLE (c): Obtain L{t"} for n a positive integer.

By definition

L{t"} = e-'Y dt.

Let us attack the integral using integration by parts with the choice exhibited
in the table.

t" e-^'dt
1
nt"-^ dt e
s

We thus obtain

,Q0 00 M 00
-t"e' n r^

e-'Ydt = +- e-'Y-'dt. (6)


ftJn
S-i

For s > and n>0, the first term on the right in (6) is zero, and we are left

with

e-'U"dt =- e-"t"-' dt, s>0.


s Jo

or

L{t"}=-L{t"-'}, s>0. (7)


s
152 The Laplace Transform [Ch. 10

From (7) we may conclude that, for /? > 1,

n
L{t"

so
§ 52] Transforms of elementary functions 153

Thus

L{H(t)} = -
s
1^ +
s
+ 4-0 + ^^-
s s

^+

Exercises

1. Show that L{cos kt} = — ,s>0.


s^ + k^
2. Obtain L{t^-3t + 5}. ANS

3. Obtain L{ht^ + t^ - 1).

4. Evaluate L{e-*' + 3^-2'}. ans.

5. Evaluate L{2e^'-e-^'}.

s
6. Show that L {cosh A:/} ==-; —,s>\k\.
s^ — k^
k
7. Show that L{sinh = -; —
A:^}
5^ — A:^ ,^>l^l•
8. Use the trigonometric identity cos^ A= ^{\ -\- cos lA) and equation (5),

——,
_y2 + 2k^
Section 52, to evaluate L {cos ^ A;/}. ans. ,s>0.
154 The Laplace Transform [Ch. 10

14. Find L{<^(0} where


<^(r) = l, 0<r<2,
= t, t>2.
1 e-^' e-^'
ANs. -
S
H
S
h —S''
, s>0.
15. Find £{/!(/)} where
Ait)=0, 0<t<l,
= t, \<t<2,
= 0, t>2.

'"'
{h+'-y '-{?+-)'''''''"'
16. Find £{5(0} where
Bit) = sin 2/, < < / TT,

= 0, t>7T.

2 (1 -e-")
ANS. ; : , 5 > 0.
j^ +4

53. Transforming initial value problems

Because our major interest lies in the applicability of the Laplace trans-
formation to the solving of differential equations, it seems natural to seek
information about the transforms of functions that are derivatives of other
functions.
Suppose, for example, we wish to solve the initial value problem

^=
at
e'\ yiO) = i. (1)

Multiplying equation (1) by e^" and integrating on the interval </< co

gives

l'-^ - Ue"] (2)

We need information about the Laplace transform of dy/dx.


By definition, we have

if this integral exists. An integration by parts performed on the integral in

(3) yields
§ 53] Transforming initial value problems 155

Now if ye'"' approaches zero as / -^ oo, we have

In Section 55 we establish sufficient conditions on the function y for the


existence of L{y(t)}. We shall also see that under the same kinds of conditions,
equation (4) is valid.

For the moment, we shall accept (4) and apply it to equation (2). We obtain
sL{y(t)}-y{0) = L{e''}.

Referring now to equation (1) of Section 52 and remembering that y(0) = {,


we find that

sL{y{t)]-- = -, s>2,
2 s — 2
or

M^'COI^^'^' ^>2. (5)

The foregoing example demonstrates the essence of the Laplace trans-


formation as it applies to linear initial value problems. The problem (1) has
been replaced by a new problem, posed in (5). Instead of looking for a func-
tion that satisfies (1), we seek a function whose Laplace transform is given
in (5). Hopefully these two functions will be one and the same.
In the present example, we can again call upon equation (I) of Section 52
to see that one function that satisfies equation (5) is

XO = \e^'. (6)

It is a simple matter to prove that this function is indeed the solution of the
initial value problem of (1).

An extension of these ideas to equations of order two can easily be made


by replacing the function y in equation (4) with the function dyjdt. We thus
have

M-74 = sL\-^\ - /(O).

Another application of (4) yields

or

l|^| = s'L{y(t)] - sy(0) - y'{0). (7)


156 The Laplace Transform [Ch. 10

EXAMPLE (a): Solve the initial value problem

y" +y= 0; j(0) = o, j'(0) = i. (8)

We shall presume throughout that a solution of (8) exists and that the
solution and its derivatives are sufficiently well behaved to insure the existence
of the needed transforms. Of course, we will need to prove that the function
we obtain on the basis of these assumptions actually is a solution of (8).

Applying the linear operator L to (8) gives

L{y" + y}^0,
or

L{y"} + L{y} = 0.

An application of equation (7) above gives us

s'L{yU)}-\+L{y{t)}=0,
an equation that can easily be solved for L{y{t)}. We have

L{y(t)]=-^. .
(9)
s + 1

From equation (4) of Section 52 we find that ,v = sin / is a function that


satisfies (9), and it is easy to verify that sin / is indeed the solution of the
initial value problem (8).

EXAMPLE (b); Solve the initial value problem

/'+;-= 1; .v(0) = 2, v'(0) = 0. (10)

Applying the Laplace operator L to equation (10) gives

s^L{v] -2s + L{v}^-, s>0. (11)


s

We solve (1 1) for L{y] and find that

25^+1 _ 1

^(>'(0} = 7772—7T = - + -TT^ (12)


s(s^ + \)~~s s^ + 1

Equations (2) and (5) of Section 52 lead us to

y(t) = 1 + cost. (13)

Again it is a simple matter to show that 1 + cos t is the desired solution of


the initial value problem (10).
The observant student will note that the ability to find the unknown func-
tion y in equation (12) depended to a great extent on our breaking a rational
§ 54] Sectionally continuous functions 157

function of s into a sum of two simpler rational functions. The process is

commonly used in elementary calculus and is often referred to as a partial


fractions expansion. In some of the following exercises the student will find
the use of such expansions helpful.

Exercises
In Exs. 1-10, solve the problem by the Laplace transform method. Verify that
your solution satisfies the differential equation and the initial conditions.

1. / = e';XO)-l.
2. y' = 3e';yiO) = l.

3. y'-y=-e-';y(0) = l.
158 The Laplace Transform [Ch. 10

particularly initial value problems in differential equations. Therefore we do not


hesitate to restrict our study to functions F{t) that are continuous or even
differentiable, except possibly at a discrete set of points, in the semi-infinite

range / ^ 0.

For such functions, the existence of the integral (1) can be endangered only
at points of discontinuity of F{t) or by divergence due to behavior of the
integrand as ?-» oo.
In elementary calculus we found that finite discontinuities, or finite jumps,
of the integrand did not interfere with the existence of the integral. We there-

fore introduce a term to describe functions that are continuous except for
such jumps.

Definition : 77?^ function F{t ) is said to be sectionally continuous over the


closed interval a ^ t -^ b if that interval can be divided into a finite number
of subintervals c ^ t ^ d such that in each subinterval:

1 F{t ) is continuous in the open interval c < t < d,

2. F{t) approaches a limit as t approaches each endpoint from within the


interval; that is, Lim F{t) and Lim F{t) exist.
§ 55] Functions of exponential order 159

counterexamples to any such notion. The concept of sectionally continuous


functions will, in Section 56, play a role in a set of conditions sufficient for the
existence of the transform.

55. Functions of exponential order

If the integral of e""F(/) between the limits and t^ exists for every finite
positive to , the only remaining threat to the existence of the transform

e-"F(t)dt (1)
•'0

is the behavior of the integrand as / -> oo.

We know that

e-"dt (2)
-'o

converges for c > 0. This arouses our interest in functions F(t) that are, for
large /(/ ^ /q), essentially bounded by some exponential e'" so that the inte-
grand in (1) will behave like the integrand in (2) for s large enough.

Definition: The function F(t) is said to be of exponential order as t -* co

if there exist constants M and b and a fixed t-value tQ such that

\F{t)\<Me'\ fort^t^. (3)

If b is to be emphasized, we say that F{t ) is of the order of e*' as / ->^ oo. We


also write
F{t) = 0{e'"\ t-^oo, (4)

to mean that F(t) is of exponential order, the exponential being e*", as / -> oo.

That is, (4) is another way of expressing (3).

The integral in (I) may be split into parts as follows:

J.0O -fo ,00

e-''F(t) dt = e-"F(t) dt + e-"F(t) dt. (5)


''o 'o ^to

If F(/) is of exponential order, F{t) = 0{e^'), the last integral in equation (5)
exists because from the inequality (3) it follows that for s > b,

f" .-F(01 d, < M fe- e- d, = ^"P^-'f-"" (6)

For s > b, the last member of (6) approaches zero as /q -> oo. Therefore the
last integral in (5) is absolutely convergent* for s > b. We have proved the
following result.

* If complex s is to be used, the integral converges for Rq{s) > b.


160 The Laplace Transform [Ch. 10

Theorem 5: If the integral of e~^'F{t) between the limits and to exists for
every finite positive /q, and if F{t) is of exponential order, F(t) = 0{e'") as
/ ^ 00, the Laplace transform

L{F(0}= re-''F(t)dt=f(s) (7)

exists for s > b.

We know that a function that is sectionally continuous over an interval is

integrable over that interval. This leads us to the following useful special
case of Theorem 5.

Theorem 6: If F{t) is sectionally continuous over every finite interval in the

range t ^ 0, and if F{t) is of exponential order, F{t) = (9(e'") as t-> cx), the
Laplace transform L{F{t)} exists for s > b.

Functions of exponential order play a dominant role throughout our work.


It is therefore wise to develop proficiency in determining whether or not a
specified function is of exponential order.
Surely if a constant b exists such that

Limle-" \F(t)\:\ (8)


t-*<x>

exists, the function F(t) is of exponential order, indeed of the order of e'^'.

To see this, let the value of the limit (8) be A' # 0. Then, for / large enough,
\e~'"F(t)\ can be made as close to A^ as is desired, so certainly

\e-'"F{t)\ < 2K.

Therefore, for t sufficiently large,

\F{t)\<Me", (9)

with M = IK. If the limit in (8) is zero, we may write (9) with M= \.

On the other hand, if for every fixed c,

Lim [_e-''\F{t)\] = oo, (10)


f->oo

the function F{t) is not of exponential order. For, assume b exists such that

\F{t)\<Me'\ t^to; (11)

then the choice c — 2b would yield, by (11),

k"^'"/"(/)| KMe-"',

so e~^'"F(t) -> as / ^ 00, which disagrees with (10).


§ 56] Functions of class A 161

EXAMPLE (a): Show that /^ is of exponential order as /^ oo.

We consider, with b as yet unspecified,

*3

Lim(e-'"f^) = Lim4- (12)


f-> 00 f-« 00 ^

If Zj > 0, the hmit in (12) is of a type treated in calculus. In fact,

f^ 3/2 6t 6
= Lim —v- = Lim
. .
= Lim —t-t- =
. . . .

Lim -r- —x-r, 0.


r-^co e"' r-oo be"' t^oo b^e" ,-.oo b^e"

Therefore / ^ is of exponential order,

t^ = 0{e'"), t-*oo,
for any fixed positive b.

EXAMPLE (b) : Show that exp (/ ^) is not of exponential order as / -* oo.

Consider
exp (t')
Lim —- -. (13)
,^00 e\p{bt)

Ub^O, the limit in (13) is infinite. U b > 0,


exp (t^)
,
Lim
.

= Lim .

exp it{t — b)] = oo.


,-.00 exp(6r) f^oo

Thus, no matter what fixed b we use, the limit in (13) is infinite and exp (t^)

cannot be of exponential order.


The exercises at the end of the next section give additional opportunities
for practice in determining whether or not a function is of exponential order.

56. Functions of class A


For brevity we shall hereafter use the term "a function of class A" for any
function that

(a) is sectionally continuous over every finite interval in the range / ^


and
(b) is of exponential order as /-^ go.

We may then reword Theorem 6 as follows.

Theorem 7: IfF(t) is a function of class A, L{F(t)} exists.

It is important to realize that Theorem 7 states only that for L{F{t)} to


exist, it is sufficient that F(t) be of class A. The condition is not necessary. A
162 The Laplace Transform [Ch. 10

classic example showing that functions other than those of class A do have
Laplace transforms is

Fit) = rK
This function is not sectionally continuous in every finite interval in the range
/ ^ 0, because F(t)^ oo as t -^0*. But f^-^ is integrable from to any
positive Iq. Also t~^^- -*0 as /-^ oo, so t~^'^ is of exponential order, with
M= 1 and Z? = in the inequality (3), page 159. Hence, by Theorem 5,

page 160, L{t~^'^} exists.

Indeed, for s > 0,

-'o

in which the change of variable st = i' leads to

L{r''^}=2s-"^ Cexpi- y^)dy, s > 0.

In elementary calculus we found that exp ( — v^) dy = ^\'n.

Therefore

= (-) , .>0,

even though r~^^^->Goas/^0'^. Additional examples are easily constructed


and we shall meet some of them later in the book.
If Fit ) is of class A, F{t ) is bounded over the range ^ t ^ to,

\F(t)\<M„ O^t^to. (2)

But F(t) is also of exponential order,

\F(t)\<M,e'\ t^to. (3)

If we choose M as the larger of M^ and M2 and c as the larger of b and zero,


we may write
\F(t)\<Me'\ r^O. (4)

Therefore, for any function F{t) of class A,

.00

e-"F(t)dt <M\Jn
X
e-"-e"dt =
M
J^
s>c.
,
(5)
-'n '0 S
s — C
c

Since the right member of (5) approaches zero as 5 -> 00, we have proved the
following useful result.
§ 57] Transforms of derivatives 163

Theorem 8 : // F(t ) is of class A and ifL{F(t)}-^ /(s),

Lim/(s) = 0.
s-*co

From (5) we may also conclude the stronger result that the transform /(5)
of a function F(t) of class A must be such that sf{s) is bounded as 5 -> go.

Exercises
1. Prove that if Fi(t) and FzO) are each of exponential order as t->co, then
FiU) Fzit) and Fi{t) + Fzit) are also of exponential order as / -^ co.

2. Prove that if Fi(r) and FjCO are of class A, page 161, then Fi(0 + F^^t) and
Fi{t) FiiO are also of class A.
3. Show that z'' is of exponential order as ^ -> oo for all real x.

In Exs. 4-17, show that the given function is a class A. In these excercises, n
denotes a non-negative integer, k any real number.

4.
164 The Laplace Transform [Ch. 10

is such that F2 is of the order of exp (0 •


t), but F2 is not of exponential order.
From Example (b), page 161,

, . exp (f')
Lim = 00
,^^ exp (60

for any real b. Since the factors It cos [exp (/^)] do not even approach zero
as / ->• 00, the product F{ exp (
— c/) cannot be bounded as ? -> 00 no matter
how large a fixed c is chosen.
Therefore, in studying the transforms of derivatives, we shall stipulate that

the derivatives themselves be of class A.


If /"(/) is continuous for / ^ and of exponential order as /-> 00, and if

F'{t) is of class A, the integral in

L{F'(0}= Ce-''F'{t)dt (1)

may be simplified by integration by parts with the choice exhibited in the


table.

F'{t) dt

-se-"dt Fit)

We thus obtain, for 5 greater than some fixed Sq ,

e-"F'(t)dt=\e-"F(t)\ +s\ e-"F(t)dt,


Jq L Jq •'0

or
L{F'(t)}=-F(0) + sL{F(t)}. (2)

Theorem 9: If F{t) is continuous for / ^ and of exponential order as


t -* CO, and if F'{t) is of class A, page 161, it follows from L{F(t)} =f{s) that

L{F'(t)} = sf(s) - F(0). .


(3)

In treating a differential equation of order n, we seek solutions for which


the highest-ordered derivative present is reasonably well behaved, say
sectionally continuous. The integral of a sectionally continuous function is

continuous. Hence, we lose nothing by requiring continuity for all derivatives


of order lower than /?. The requirement that the various derivatives be of
exponential order is forced upon us by our desire to use the Laplace trans-
form as a tool. For our purposes, iteration of Theorem 9 to obtain transforms
of higher derivatives makes sense.
From (3) we obtain, if F, F', F" are suitably restricted,

L{F"(t)} = sL{F'(t)} - F'(0),


§57] Transforms of derivatives 165

or
L{F"(t)} = s'fis) - sF(0) - r(0), (4)

and the process can be repeated as many times as we wish.

Theorem 10: If F{t), F'(t), . . . , F^"~^\t) are continuous for t ^ and of


exponential order as t -* oo, and if F^"^{t) is of class A, then from
L{F{t)] =f{s)
it follows that

L{F^"\t)] = s"f{s) - \
k =
s"- '
-'F^^XO). (5)

Thus
L{F^^\t)] = s^fis) - s^FiO) - sF'(0) - F"(0),
L{F^*\t)} = s'^f(s) - s^F(O) - s^F'iO) - sF"(0) - F<^>(0), etc.
Theorem 10 is basic in employing the Laplace transform to solve linear
differential equations with constant coefficients. The theorem permits us to
transform such differential equations into algebraic ones.
The restriction that F{t) be continuous can be relaxed, but discontinuities
in F{t) bring in additional terms in the transform of F'{t). As an example,
consider an F(t) that is continuous for / ^ except for a finite jump at

t = /i, as in Figure 15. If F(t) is also of exponential order as /-> oo, and if

F'(t) is of class A, we may write

L{F'(t)}= re-'"F'(t)dt
•'o

= e'^'F^t) dt + e-''F\t) dt.


•'o •'ti

FIGURE 15
166 The Laplace Transform [Ch. 10

Then, integration by parts applied to the last two integrals yields

L{F'(t)}=\e-"Fit)T + s (' e-''F{t)dt + [e-^'F(/)l'' + s C e-"F{t)dt


L Jo •'o L Jfi -'r,

= s f e-"F{t) dt + e\p(-sti)F(ti-) - F(0) +


Jq

-exp(-st,)F(t,^)

= sL{F(t)} - F(0) - exp (-sr,)[i^(r/) - F(tni


In Figure 15 the directed distance AB is of length [F(ri"^) — F(/i~)].

Theorem 1 1 : IfF(t) is of exponential order as t -^ co and F(t) is continuous


for / ^ except for a finite jump at t = t^, and if F\t) is of class A, then from

L{F{t)} ^f{s)

it follows that

L{FV)} = ^f(s) - f (0) - exp (-5ri)[F(/,+ ) - F(t, -)]. (6)

If F{t) has more than one finite discontinuity, additional terms, similar to
the last term in (6), enter the formula for L{F'{t)].

58. Derivatives of transforms

For functions of class A, the theorems of advanced calculus show that it

is legitimate to differentiate the Laplace transform integral. That is, if F{t)


is of class A, from

/(s)= fe-'^nOdt (1)

it follows that

f'(s)= r(-t)e-'"F(t)dt. (2)


*'o

The integral on the right in (2) is the transform of the function { — t)F{t).

Theorem 12: If F{t) is a function of class A, it follows from

L{F(t)} =f(s)

that

f'{s) = L{-tF(t)}. (3)


§ 58] Derivatives of transforms 167

When F(t) is of class A, { — t)''F{t) is also of class A for any positive


integer k.

Theorem 13: If F{t) is of class A, it follows from L{F{t)} =f{s) that for
any positive integer n,

£-J(s) = L{(-trF(t)]. (4)

These theorems are useful in several ways. One immediate application is

to add to our list of transforms with very little labor. We know that

L{sin kt], (5)


s' + k'

and therefore, by Theorem 12,

-2ks
L{-t sin kt}.
(s^ + k^)

Thus we obtain

t
= LI- sin kt]. (6)
(s^ + /c^)2 \2k

From the known formula

s
= L{cos kt}
5^ + k^

we obtain, by differentiation with respect to s,

k'-s'
= L{-t cos kt}. (7)
(s^ + kY
Let us add to each side of (7) the corresponding member of
1
= L (1- sm kt
s^ + k^ [k

to get

s^ + k^ + k^-s^ (\
L - sin kt — t cos kt].
(s^ + k^f \k

from which it follows that

(s^
1

+ k^f
L — 3 (sin kt - kt cos kt)\. (8)
168 The Laplace Transform [Ch. 10

59. The gamma function

For obtaining the Laplace transform of nonintegral powers of t, we need


a function not usually discussed in elementary mathematics.
The Gamma function r(.Y) is defined by

r(x)= e-^p""-' dp, x>0. (1)


-'o

Substitution of {x + 1) for x in (1) gives

.00

r(x + l)= e-'^P'^dp. (2)

An integration by parts, integrating e~^ dp and differentiating ^^ yields

r(x + l)= -g-^i^H +x\ e'^P'^-'dp. (3)

Since a' > 0, ^"^ ^ as ^ -> 0. Since x is fixed, e'^P"" -> as /? -> oo. Thus

r(x + 1) = X Ce-i'p''-' dp = xr(x). (4)


•0

Theorem 14: For x > 0, r(x + 1) = xr(x).

Suppose n is a positive integer. Iteration of Theorem 14 gives us

r(/7 + 1) = nY{n)
= n{n- \)T{n- 1)

= /?(//- l)(/?-2)---2- 1 -YiX)


= //!r(l).
But, by definition,

r(l)= f"e-TJiS=[-e-^l" = l.

Theorem 15: For positive integral n, r{n +!) = ;?!.

In the integral for r(x + 1) in (2), let us put P = st with s> Q and / as the
new variable of integration. This yields, since / -»• as ^^ and / -^ oo as

P -^ <X),

r(x + 1) = f e'^'s'^fs dt = s*+^ Ce-^'t'' dt, (5)


§ 60] Periodic functions 169

which is vaHd for x + 1 > 0. We thus obtain

5^^^^i^
s
= f
''o
e- V c/r, s>0,x>-l,

which in our Laplace transform notation says that

^{n = ^^^, s>0,x>-l. (6)

If in (6) we put x = —j, we get

./2, _ r(«
S

But we already know that L{t '^^} = l~l • Hence

n\)=^n. (7)

60. Periodic functions

Suppose the function F{t) is periodic with period w:

F{t + co) = F(t). (1)

The function is completely determined by (1) once the nature of F{t) through-
out one period, ^ ?< co, is given. If F(t) has a transform,

L{F(t)} = Ce-'^'Fit) dt, (2)


•'o

the integral can be written as a sum of integrals,


00 An+ l)a)

L{F{t)}= X e-^'F{t)dt. (3)


n = •'nco

Let us put t — noi + p. Then (3) becomes


00 -co

^{^(0}=Z Qxp(-sna>-sl^)F(P+nco)dp.
n= ''o

But F(P + no) = F{P), by iteration of (1). Hence


00 -co

L{F{t)} = S= exp {-snoj) exp i-sP)F(P) dp. (4)


n •'O

The integral on the right in (4) is independent of n and we can sum the
series on the right

00 00
J
5;exp(-s«ft>)= i; [exp(-sco)]"
= -

n = n = V — e r^.
170 The Laplace Transform [Ch. 10

Theorem 16: If F{t) has a Laplace transform and if Fit + to) = F{t),

e-"'FiP)dp
L{F(t)} (5)
1 - e"^"

Next suppose that a function //(/) has a period 2c and that we demand
that H{t) be zero throughout the right half of each period. That is.

Hit + 2c) = H(t), (6)

Hit) = git), 0^t<c, (7)

= 0, c^t <2c.
Then we say that //(/) is a half- wave rectification of ^(/). Using (5) we may
conclude that for the Hit) defined by (6) and (7),

(cxpi-sP)giP)dp
L{Hit)} = '^
. (8)
1 — exp (
— 2cs)

EXAMPLE (a): Find the transform of the function ij/it, c) shown in

Figure 16 and defined by


(/^(/, c)=l, 0<t<c, (9)

= 0, c < t <2c;
Ht + 2c, c) = il'it, c). (10)

^M

VMMMM W^M¥i'/i'M
2C 3C 4C

FIGURE 16

We may use equation (8) and the fact that

Jq S
§60] Periodic functions 171

to conclude that

1 1 — exp — sc)
( 1 1
Lmt, c)} (11)
si— exp ( — 2sc) s I + exp ( — sc)

EXAMPLE (b): Find the transform of the square-wave function Q{t, c)

shown in Figure 17 and defined by

Q(t,c)=\, 0<t<c, (12)

= — 1, c < t <2c;
Q{t + 2c, c) = Qit, c). (13)

QM

2C 3C 4C

-1

FIGURE 17

This transform can be obtained by using Theorem 16, but also

Q(t, c) = 2il/(t, c)-l; (14)


hence, from (11),
1 1 — exp ( — sc)
L{Q{t, c)} = - - 1 (15)
s 1 + exp ( — sc) s 1 + exp ( — sc)
By multiplying numerator and denominator of the last fraction above by
exp (^sc), we may put (15) in the form
1 cs
L{Q(t,c)} = -tanh-. (16)
s 2

Exercises

1. Show that L{ri'^} ' s>0.


2s[s)
172 The Laplace Transform [Ch. 10

= — l-l
,^15 /Try
2. Show that L{r *'-} , s>0.
3. Use equation (4), page 165, to derive L{sin kt}.
4. Use equation (4), page 165, to derive L{cos kt}.
5. Check the known transforms of sin kt and cos kt against one another by using
Theorem 9, page 164.
6. If/? is a positive integer, obtain L{/V*^'} from the known L{e'"} by using Theorem
§ 60] Periodic functions 173

16. Define the function ^C/) by

Sit) = l-t, 0^/<l,


S(t+l) = Sit), t^O.
Sketch the graph of S(t) and find its Laplace transform.
17. Sketch a half-wave rectification of the function sin cjt, as described below, and
find its transform.

F(t) = smojt, ^ / ^- ,
CO

= 0, -</< —
7T

CD
2tT

CO
;

ANS.
s^^-oj'
— exp
1
(-a
18. Find L{F(/)} where F{t) = / for < <w / and F(? + co) - F{t).
1 CO exp(— i'Cu) 1 c 0}S\
ANS. -= —+— |l-coth
s^ s 1— exp(— 5co) s^ 2s
Fit)
19. Prove that if L{F(/)} =f{s) and if -^ is of class A,

.p)^i;/<.)*
Hint: Use Theorem 12, page 166.
CHAPTER XX
Inverse Transforms

61. Definition of an inverse transform

Suppose the function F{t) to be determined from a differential


is

equation with initial The Laplace operator L is used to


conditions.
transform the original problem into a new problem from which the transform
f{s) is to be found. If the Laplace transformation is to be effective, the new
problem must be simpler than the original problem. We first find f{s) and
then must obtain F{t) from f{s). It is therefore desirable to develop methods
for finding the object function F(t) when its transform /(5) is known.
If

L{F(/)} =f(s), (1)

we say that F{t) is an inverse Laplace transform, or an inverse transform,


of /(5) and we write

F(t)=L-'{f(s)}. (2)

174
§ 61] Definition of an inverse transform 175

Since (1) means that

re-^'F(t)dt=f(s), (3)
•'o

it follows at once that an inverse transform is not unique. For example, if

Fi(t) and ^2(0 are identical except at a discrete set of points and differ at

these points, the value of the integral in (3) is the same for the two functions;
their transforms are identical.

Let us employ the term null function for any function N(t) for which

rN{t)dt = (4)
•'o

for every positive /q . Lerch's theorem (not proved here) states that if

L{Fi(/)} = L{F2{t)}, then Fi(0 - F2U) = ^(0- That is, an inverse Laplace
transform is unique except for the addition of an arbitrary null function.
The only continuous null function is zero. If an f{s) has a continuous
inverse F{t), then F{t) is the only continuous inverse of/(5). If /(-s) has an
inverse Fi{t) continuous over a specified closed interval, every inverse that
is also continuous over that interval is identical with Fi(/) on that interval.

Essentially, inverses of the same f{s) diff^er at most at their points of dis-

continuity.
In applications, failure of uniqueness caused by addition of a null function
is not vital, because the effect of that null function on physical properties of
the solution is null. In the problems we treat, the inverse F{t) is required
either to be continuous for / ^ or to be sectionally continuous with the
values of F(t) at the points of discontinuity specified by each problem. The
F{t) is then unique.
A crude, but sometimes effective, method for finding inverse Laplace
transforms is to construct a table of transforms (page 199) and then to use
it in reverse to find inverse transforms.

We know from Ex. 1, page 153, that

'
L{cos kt} = (5)
S I /v

Therefore

= cos kt. (6)


s^ + k'

We shall refine the above method, and actually make it quite powerful,
by developing theorems by which a given f{s) may be expanded into com-
ponent parts whose inverses are known (found in the table). Other theorems
176 Inverse Transforms [Ch. 11

will permit us to write /(5) in alternate forms which yield the desired inverse.
The most fundamental of such theorems is one that states that the inverse
transformation is a linear operation.

Theorem 17: If c\ and Cj are constants,

L-'{cJy{s) + cj,{s)} = qL-H/i(5)} + c,L-'{f,{s)).

Next let us prove a simple, but extremely useful, theorem on the manipula-
tion of inverse transforms. From

/(.s)= re-^^F{t)dt, (7)

we obtain

f{s- a)= Ce-^'-"^'F{t)dt


•'o

= Ce-\e'"F{t)-]dt.
''O

Thus, from L"^{/(5)} = F(t) it follows that

L-'{fis-a)} = e'"F{t),

or

L-'{f{s-a)} = e'"L-'{ns)}. (8)

Equation (8) may be rewritten with the exponential transferred to the other
side of the equation. We thus obtain the following result.

Theorem 18: L-'{f{s)} = e'^'L-^ifis - a)}.

EXAMPLE (a): FindL-'(^ 1.

First complete the square in the denominator.

ls' + 4s + 13j \(5 + 2)' + 9J'

Since we know that L~^ -5 -^ — sin kt, we proceed as follows:


W+ k^j

L-' L ]^
W+ + 4s
^
13)
] = 5L-'l-
\{s + If +
^-^—\ = 5e-"L-'l^^] = 5e-"
9) \s^ + 9/
sin 3/,

in which we have used Theorem 18.


§61] Definition of an inverse transform 177

s+
EXAMPLE (b): Evaluate L
1

s^ + 6s + 25

We write

s + 1 s+ 1
L-' L-'
s^ + 6s + 25 (s + 3r + 16

Then

s+ 1 5-2
e-^'L-
s^ + 6s + 25 5^+ 16

s
= e L-'
is' + 16) ^ is' + 16

= e~^'(cos4t - ^ sin4r).

Exercises
In Exs. 1-10, obtain L~^{f(s)} from the given f(s).

1
1.
s^ + 2s + 5'
1
2.

3.

4.

5.

6.

7.

8.

9.

"••(.+ 2)-
11. Show that for n a non-negative integer
178 Inverse Transforms [Ch. 11

12. Show that for w> -1,

1 re-
L
{s^aT*') r(m+l)'
13. Show that

1
) 1
L
(5 + ay + o^l o

14. Show that

s \ 1

S2 .2 == 7 ^"'" (6 cos bt-a sin 6/).


,
(5 + ay
,

+ b^) ,

15. For a > 0, show that from L' '{/{$)) = F(0 it follows that

L-Hf{as)) = -F(-
a \a

16. For fl > 0, show that from L-'ifis)} = F{t) it follows that

L-M/(«5 + Z>)}-^exp(-H/'(^

62. A step function

Applications frequently deal with situations that change abruptly at


specified times. We need a notation for a function that will suppress a given
term up to a certain value of / and insert the term for all larger t. The function
we are about to introduce leads us to a powerful tool for the constructing of
inverse transforms.

d(i)

V,W-W/>W'/<.W/M';'

FIGURE 18
§ 62] A step function 179

Let us define function a(/) by


a(0 = 0, / < 0, (1)
= 1, /^O.
The graph of cc(t) is shown in Figure 18.
The definition (1) says that a(/) is zero when the argument is negative and
«(/) is unity when the argument is positive or is zero. It follows that

a(/ - c) - 0, t <c, (2)

= 1, t^c.
The a function permits easy designation of the result of translating the
graph of F{t). If the graph of

y^Fit), t^O, (3)

is as shown in Figure 19, the graph of

y= oi{t- c)F{t -c), t^ c, (4)

FIGURE 19 FIGURE 20

is that shown in Figure 20. Furthermore, if F{x) is defined for — c ^ x < 0,


F{t — c) is defined for ^ / < c and the y of (4) is zero for ^ < c because
/

of the negative argument in a(r — c). Notice that the values of F{x) for
negative jc have no bearing on this result because each value is multiplied by
zero (from the a); only the existence of F for negative arguments is needed.
The Laplace transform of a{t — c)F{t — c) is related to that of F{t).
Consider

L{a(r - c)F{t - c)) = Ce-''oL{t - c)F(t - c) dt.


180 Inverse Transforms [Ch. 11

Since a(r - c) = for ^ < r c and a(/ — c) = 1 for / ^ c, we get

L{a(t - c)F(t -c)}= \ e-^'Fit - c) dt.


^c

Now put t — c = V m the integral to obtain

L{a{t - c)F{t - c)] = r e-'^'^^^Fiv) dv


= e-" ! e-'^Fiv) dv.
•'o

Since a definite integral is independent of the variable of integration,

Ce-'^Fiv) dv = Ce'^Fit) dt = L{F(t)} =f(s).

Therefore we have shown that

L{a(/ - c)F{t - c)} = e-"L{F{t)} = e-"f{s). (5)

Theorem 19: If L~'^{f{s)} = F(t), if c^O, and if F(t) be assigned values


(no matter )\hat ones) for —cSt<0,
L-'{e-"f(s)} = F{t - c)y.{t - c). (6)

EXAMPLE (a): Find L{;<0} where (Figure 21)

y{t) = t^, 0</<2,


= 6, t>2.

FIGURE 21
§ 62] A step function 181

Here, direct use of the definition of a transform yields

L{y(t)}= t^e-'"dt+ 6e-"dt.


•'0 •'2

Although the above integrations are not difficult, we prefer to use the a
function.
Since a{t - 2) = for / < 2 and a(/ - 2) = 1 for f ^ 2, we build the yit)
in the following way. The crude trial

works for </< 2, but we wish to knock out the t^ when / > 2. Hence we
write
y2 = t^ - t^a{t - 2).

This gives /^ for / < 2 and zero for / > 2. Then we add the term 6a(/ — 2)
and finally arrive at

y(t) = t^- t\{t - 2) + 6a(r - 2). (7)

The y of (7) is the y of our example and, of course, it can be written at once
after a little practice with the a function.
Unfortunately, the y of (7) is not yet in the best form for our purpose. The
theorem we wish to use gives us

L{F{t - c)a{t - c)) = e-"f{s).

Therefore we must have the coefficient of a{t — 2) expressed as a function of


(t — 2). Since

-t^ + 6= -(/2 _ 4/ + 4) - 4(t - 2) + 2,

y(t) = /2 _ (^ _ 2)2^(1 - 2) - 4{t - 2)a(/ - 2) + 2a(/ - 2), (8)

from which it follows at once that

,
L{y(t))
_ = -,-^^-^^^.
2 le'^' 4e-^' le'^'

EXAMPLE (b): Find and sketch a function g(t) for which

,(3 4e-' 4e-^']

We know that L"^^| = 4/. By Theorem 19 we then get

^"1^)"'^^^"^^°'^^"^^
182 Inverse Transforms [Ch. 11

and

L-^{^^}=4(r-3)a(r-3).

We may therefore write

g{t) = 3 - 4(r - \Mt - 1) + 4(/ - 3)a(/ - 3). (9)

To write g{t) without the a function, consider first the interval

0^/< 1

in which a(/ - 1) = and a(/ - 3) = 0. We find

g(t) = 3, OSt<l. (10)

For 1 ^ < / 3, a(/ - 1) = 1, and a(/ - 3) = 0. Hence

^(0 = 3-4(/- l) = 7-4/, l^/<3. (11)

For / ^ 3, a(/ - 1) = 1 and a(r - 3) = 1, so

^(0 = 3-4(/-l) + 4(r-3)= -5, /^3. (12)

Equations (10), (11), and (12) are equivalent to equation (9). The graph of
g{t) is shown in Figure 22.

FIGURE 22

Exercises
In Exs. 1-7 sketch the graph of the given function for / ^ 0.

1. a(r-c).
2. ccit - 1) + 3a(/ - 3) - 6a(r - 4).
§ 62] A step function 183

3. (/ - 2)a(/ - 2).
4. sin (t — tt) ait —

tt).

5. (/ — TTytxit — tt).

6. t^-(t-2ya(t-2).
7. t^-tMt-2).
In Exs. 8-15, express Fit) in terms of the a function and find L{F(/)}.

8. F(0 = 2, 0</<l, 2 /I 1\
r>l. ANS. - + .-^^---j.
9. F(/) = 6, < < 4,
/ 6
-2/+1, />4. ANS. - +^
i"n)-
10. F(0-/', 0</<l, 2
= 4, />!. ^>^^- -.
+ ^-\- --.--.)
11. Fit)^t\ 0<t<2,
= 4, 2^t^4,
= 0, />4.
2
--^—
ANS.
/4 2\
(^- + -j--e
4

12.F(0=/^ 0</<2 2 /3 2\ /53 n
= .-1, 2<.^<3, .ws. --.-^^(- + - + -)+.-3.(___j.
13.

14.
F(/)-e-',

Fit)
^0,
= sin 3t,
0</<2,
t>2.
< < / ^tt,
^^'-
l-exp(-2^-2)

3
7:n

+ ^ exp (-^775)
= 0, t>^7r. ANS. ^7-p^ .

15. F(/) = sin 3/, < < / tt, 3(1 + e""")


= 0, t>7T. ANS.
s^ +9
16. Find and sketch an inverse Laplace transform of

s s

ANS. Fit) - 5a(/ - 3) - (xit - 1).

17. Evaluate L-'j-—T77- ans. K^ -4)^ exp [-2(/ -4)] a(/ -4).

18. If Fit) is to be continuous for ? ^ and

Fit)=i
(s+l)'
evaluate F(2), F(5), F(7). ans. F(2) = 0, F(5) = 2e-\ Fil) = 8e " *.
19. If Fit) is to be continuous for / ^ and

evaluate F(l), F(3), F(5). ans. F(1) = 1, F(3) = -1, F(5) = -4.
00

20. Prove that ipit,c)=


n
2= — l)"a(^ — «c)
( is the same function as was used in

Example (a). Section 60. Note that for any specific /, the series is finite; no
question of convergence is involved.
184 Inverse Transforms [Ch. 11

21. Obtain the transform of the half-wave rectification F(/) of sin / by writing

F(t) = sin t ip{t, tt)

in terms of the */> of Ex. 20 above. Use the fact that

(
— l)"sin / = sin (/ — mi).
Check your result with that in Ex. 17, page 173.

63. A convolution theorem

We now seek a formula for the inverse transform of a product of transforms.


Given
L-'{f{s)} = F{t), L-'{g{s)} = G{t), (1)

in which F{t) and G(r) are assumed to be functions of class A, we shall obtain

a formula for

L-'{f{s)g{s)}. (2)

Since /(5) is the transform of F{t), we may write

f{s)= fe-^'FiOdt. (3)


•'o

Since g(s) is the transform of G(t),

g(s)= fe-^^Crndp, (4)

in which, to avoid confusion, we have used p (rather than /) as the variable


of integration in the definite integral.
By equation (4), we have

f(s)g(s)= fe-^^mcmd^^. (5)


•'o

On the right in (5) we encounter the product e~^^f{s). By Theorem 19, page
180, we know that from

L-'{f{s)}=Fit) (6)
it follows that

L-'{e-'l^f(s)} = F{t-PHt-P), (7)

in which a is the step function discussed in Section 62. Equation (7) means
that

e-''f{s) = Ce-^'Fit - PMt - P) dt. (8)


•In
§ 63] A convolution theorem 185

With the aid of (8) we may put equation (5) in the form

mgis) = e -"'CmFit - P)a{t - P) dt dp. (9)

Since a(/ - j8) = for < < / i? and a(/ - P) ^ \ for t ^ (i, equation (9)

may be rewritten as

fis)g(s)= r re-'^'G(P)F(t-P)dtdp.
•'B
(10)
•'O

In (10), the integration in the /j5-plane covers the shaded region shown in
Figure 23. The elements are summed from t = fi to t = oo and then from
j9 = to i?
= 00 .

FIGURE 23

By the methods of advanced calculus it can be shown that, because F(t)


and G(t) are functions of class A, it is legitimate to interchange the order of
integration on the right in equation (10). From Figure 23 we see that in the
new order of integration, the elements are to be summed from P = to

P = t and then from f = to / = oo. We thus obtain

/(5)6f(s)= f* ( e-'GiP)F(t - P) dp dt.

or

f(s)g(s) = f"e-4 (Gip)F{t - P) dp\ dt. (11)


•'o L*'o J
186 Inverse Transfonns [Ch. 11

Since the right member of (11) is precisely the Laplace transform of

(G(P)F(t - P) dp,
-'o

we have arrived at the desired result, which is called the convolution theorem
for the Laplace transform.

Theorem 20: rfL-'{f(s)} = F(t), ifL-'{g(s)] = G(t), and if F(t) and G(t)
are functions of class A, page 161, then

'{As)g{s)]= (G(l3)F(t-p)dp.

It is easy to show that the right member of equation (12) is also a function

of class A.
Of course F and G are interchangeable in (12), since / and g enter (12)
symmetrically. We may replace (12) by

L-'{f(s)g(s)} = (F(P)G(t - P) dp, (13)

a result which also follows from (12) by a change of variable of integration.

EXAMPLE EvaluateL"^ — '

Let L-'{f{s)]^ F(t). Since

..(iU.,

we use Theorem 20 to conclude that

Lr.{^>}=fy(«.//i.

Exercises
In Exs. 1-3 find the Laplace transform of the given convolution integral.

2
1.
[
(t-p) sin 2p dp. ANS.
s{s^ + 4)
s
2. !
€-<'-"' cos
p dp. ANS. - .,, , ,-
Jo (5+l)(5^+l)

3.
f
Jq
((-pye''dp. ANS. —
s^is -
-.
1)

In Exs. 4-6, find an inverse transform of the given f(s) using the convolution
theorem.
§ 64] Simple initial value problems 187

5. — 1

s\s —
-. ANS. e'-/-l,
1)

1
^« .,,.., • ANS. Ksin / - / cos 0-

64. Simple initial value problems

As we have seen in Section 53, the Laplace operator will transform a


differential equation with constant coefficients into an algebraic equation in
the transformed function. Several additional examples will now be treated
in detail so that we can get some feeling for the advantages and disadvantages
of the transform method. One fact is apparent from the start: this method is

at its best when the boundary conditions are actually initial conditions, those
giving the value of the function and its derivatives at time zero.

EXA M PLE (a) : Solve the problem

y'\0 + i^^KO = A sin iot; >;(0) = 1, /(O) = 0. (1)

Here A, P, w are constants. Since fi = would make the problem one of


elementary calculus and since a change in sign of j9 or co would not alter the

character of the problem, we may assume that P and co are positive.

Let L{yit)} = u{s).

Then L{y'{t)} = su{s) - \,

L{y"{t)} = s^u{s) -5-1-0,


and application of the operator L transforms the problem (1) into

Aoi
s u(s) — s + B u(s) = , ,

from which

"(^) = 2
+
, o2
/?"
+ /„2
(s^ + ,

^^ +
o2^r.2 , ,,2^
w')
• (2)

We need the inverse transform of the right member of (2). The form of
that inverse depends upon whether P and co are equal or unequal.
If CO # p,

_ s Aoj I 1 \
I

_ s A / (Dp cop \

+ p^ P(p^ - (O^) W + 0)^ s^ +P


188 Inverse Transforms [Ch. 11

Now >'(/) = L~^{u(s)} so, for oj ^ (i,

A
y(t) = cos pt + -;;7^2 2^ (P ^in cot - (o sin pt). (3)

If a> = P, the transform (2) becomes

s Ap
"(^) = ,2 , o2 + , 2 , d2x2 •
C"^)

We know from equation (8) of Section 58 that

'
= (sin pt - pt cos pt).
1(5^ + P'-fj IP'

Hence, for w= P,

A
>•(/) = cos pt + ^^ (sin pt - pt cos ^50- (5)
2p'

It is a simple matter to show that this function is indeed the solution of the
given initial value problem.
Note that the initial conditions were satisfied automatically by this method
when Theorem 10 was applied. We get, not the general solution with arbitrary
constants still to be determined, but that particular solution which satisfies
the desired initial conditions. The transform method also gives us some insight

into the reason that the solution takes different forms according to whether
CO and P are equal or unequal.

EXAMPLE (b): Solve the problem

x"(t) + 2x'it) + x(t) = 3te-'; x(0) = 4, x'(0) = 2. (6)

Let L{x(t)} = y(s). Then the operator L converts (6) into

3
s^y(s) - 4s - 2 + 2[sv(s) - 4] + y(s) =
(s + ly
or

4s + 10 3

We may write

4(5 + 1) + 6 ,

(5 + 1)^ (s + iy
§ 64] Simple initial value problems 189

,.4 6 ^

S+l (5 + 1)2 (s+1)**


Employing the inverse transform, we obtain

x(t) = (4 + 6t + ii>~'. (8)

Again the knowledge of initial conditions contributed to the efficiency of


our method. In obtaining and in using equation (7), those terms that came
from the initial values x(0) and a:'(0) were not combined with the term that
came from the transform of the right member of the differential equation.
To combine such terms rarely simplifies and frequently complicates the task
of obtaining the inverse transform.
From the solution

x(0 = (4 + 6r + if3)e-' (8)

the student should obtain the derivatives

x'(0 = (2-6t + it^ - it')e-\

x"(t) = (-S + 9t - 3t^ + ^t')e-',

and thus verify that the x of (8) satisfies both the differential equation and
the initial conditions of the problem (6). Such verification not only checks

our work but also removes any need to justify temporary assumptions about
the right to use the Laplace transform theorems on the function x(t) during
the time that the function is still unknown.

EXAMPLE (c) : Solve the problem

x"(t) + 4x(0 = ij/(t); x(0) = 1, x'(0) = 0, (9)

in which ip(t) is defined by

il/(t) = 4t, O^tSh (10)

= 4, / > 1.

We seek, of course, a solution valid in the range t ^ in which the right

member i/^(0 is defined.


In this problem another phase of the power of the Laplace transform
method begins to emerge. The fact that the function i//(/) in the differential

equation has discontinuous derivatives makes the use of the classical method
of undetermined coefficients somewhat awkward, but such discontinuities do
not interfere at all with the simplicity of the Laplace transform method.
In attacking this problem, let us put L{x{t)} = h(s). We need to obtain
L{il/{t)}. In terms of the a function we may write, from (10),

^(/) = 4/ - 4(/ - l)a(/ - 1), t^O. (11)


190 Inverse Transforms [Ch. 11

From (11) it follows that

4 4e~'
s s

Therefore the application of the operator L transforms problem (9) into

4 4e-'
s^'his) -S-0 + 4/i(s) = ^ -
from which
s 4 4e
S2 + 4 5^(5^ + 4) ^2(^2 ^ 4)
Now

5^(5^ + 4) s^ s^+4'
so (12) becomes

h{s) = -j^ +\- -^—- -[\- -2 ^e-\


+ 4 5^ s^ + 4 \s^ s^ + 4/
s^
(13)

Since x(/)) = L~^{h{s)}, we obtain the desired solution

x{t) = cos 2r + - / i sin It - [(/ - 1) - i sin 2(/ - 1)] y.{t - 1). (14)

It is easy to verify our solution. From (14) it follows that

x'{t) = -2 sin 2/ + 1 - cos 2/ - [1 - cos 2(/ - 1)] a(/ - 1), (15)

x"{t) = -4 cos 2r + 2 sin 2/ - 2 sin 2{t - 1) a(/ - 1). (16)

Therefore x(0) = 1 and x'(0) = 0, as desired. Also, from (14) and (16), we get

x"(0 + 4x(0 = 4f - Ait - 1) a(r - 1) = i//(r), r ^ 0.

EXA MPLE (d) : Sol ve the problem

x"{t) + ex{t) = F{t); x(0) = ^ x'(0) = B. (17)

Here k. A, B are constants and F{t) is a known but unstipulated function.


For the time being, think of F{t) as a function whose Laplace transform
exists. Let

L{x{t)} = u{s), L{F(t)}^f(s).

Then the Laplace operator transforms problem (17) into

s^s) -As-B + k^u(s) =f(s),

,, As + B f(s)
U(S) = -^ r +
^

-^ ? . (lo)
§ 64] Simple initial value problems 191

To get the inverse transform of the last term in (18), we use the convolution
theorem. Thus we arrive at

B 1
c(t) =
x{t)= A cos kt + - sin kt +- f F(t - (i) sin A/? dp,
k k Jq
or
B
x(t) =A cos kt +- sin kt +- \
F((i) sin k{t - ft) d^. (19)
k k ^0

Verification of the solution (19) is simple. Once that check has been per-
formed, the need for the assumption that F{t) has a Laplace transform is

removed. It does not matter what method we use to get a solution (with
certain exceptions naturally imposed during college examinations) if the
validity of the result can be verified from the result itself.

EXAMPLE (e): Solve the problem

h/'(a-) + 2\v\x) + w{x) = x; w{0) = -3, ir(l) = - 1. (20)

In this example the boundary conditions are not both of the initial condi-
tion type. Using x, rather than /, as independent variable, let

L{w(x)} = g(s). (21)

We know vv(0) = — 3, but we also need w'{0) in order to write the transform
of w"{x). Hence we put
h'(0) =B (22)

and hope to determine B later by using the condition that vt'(l) = — 1.


The transformed problem is

s'g(s) -s(-3)-B + 2lsg{s) - (-3)] + g(s) = -,


s

from which
,, -3(s+l) + B-3 1

But, by the usual partial fractions expansion,

1 2 12 1

s\s +iy s s^ s + 1 (s + 1)^

so

g(s)
_= 1
-2
2 1
-— B-2
+ ——rr2, (24)
s s s + 1 (s + 1)^

from which we obtain


w{x) = jc - 2 - e"' + (5 - 2)xe-\ (25)
192 Inverse Transforms [Ch. 11

We have yet to impose the condition that n(l) = — 1. From (25) with x= I,

we get
-1 = 1 -2-e-* +{B-2)e-\
so 5= 3.

Thus our final result is

yi'(x) = jc - 2 - e~^ + xe'"". (26)

The problem in Example (e) may be solved efficiently by the methods of


Chapter 7. See also Exs. 23-44 below.

Exercises
In Exs. 1-14, solve the problem by the Laplace transform method. Verify that
your solution satisfies the differential equation and the initial conditions.

1. x'V) + 4xy) + 4x(0 = 4e- ^' ; x(0) = - 1 , x'{0) - 4.


ANS. xit) = e-^'i2t^ + 2t-l).
2. x"{t) + x{t) - 6 cos 2t; x{0) - 3, x'(0) = 1.

ANS. x{t) = 5 COS + sin — 2 cos 2t. r /

3. y"(,t) - y{t) =- 5 sin 2t\ y{0) = 0, y'{Qi) = 1. ans. y{t) = 3 sinh — sin 2t. /

4. y'XO + 6/(0 + 9X0 = 6t^e-^'; y(0) = 0, /(O) = 0. ans. y{t) = i/^e" ^'.

5. x'XO + 440 = 2r - 8 x(0) = 1, x'(P)


; = 0.

ans. x(t) = 3 cos 2/ — i sin 2t + it — 2.


6. x"{t) + 2x'(0 = 8/; jc(0) = 0, x'(0) = 0. ans. x{t) ^2t^ -2t + - e-^'. I

7. m"(0 + 4m(/) = 1 w(0) =


Se-' ; u'{0) = 3. 1 , ans. «(r) = 3e' - 2 cos 2r.
8. «"(/) + 4«'(r) + 3w(r) - 12; m(0) = 7, m'(0) = 1.

ANS. M(0=4+5e-'-2e-^'.
9. y'Xx) + 9yix) = 40^^; ^(0) = 5, ^'(0) = -2.
ANS. y{x) ^ Ae"" + cos 3x — 2 sin 3;c.
10. y"{x) + ^(jc) = Ae" ><0) = 0, ^'(O) = 0.
;
ans. y{x) = 2(e* - cos x - sin x).
11. x"{t) + 3x'(0 + 2x{t) = 4t^; x{0) = 0, a-'(O) = 0.
ans. x(0 = 2r^-6/ + 7-8e-' + e-^'.
12. jc"(0 - 4a:'(0 - Ax{t) = 4 cos 2r; .r(0) = 2, x'(0) = 5.
ANS. x(0 = 2e^'(l + 0-2 sin 2t.
13. x"(/) + x(r) = F(f); x(0) = 0, a-'(O) = 0, in which

Fit) =4, ^ / ^ 2,
= + 2, t t>2.
ANS. x(/) = 4 - 4 cos + / [(t - 2) — sin (r — 2)]a(r — 2).
14. x"(0 + xiO = H(0; xiO) - 1, x'(0) - 0, in which
Hit) = 3, ^ ^ 4, /

= 2r - 5, > 4. /

ANS. xit) == 3 - 2 cos + 2[/ - 4 - sin (r - 4)]a(/ — 4).


r

15. x'Xt) + x(0 = Git); xiO) = 0, xXO) - 1, in which

CF(/) = 1, 0^r<7r/2,
= 0, / ^ 77/2.
§ 64] Simple initial value problems 193

ANS. xit) ^1 — COS + sin — a(/ — 77/2)(l — sin 0-


/ f

16. x'XO + 4x(0 = M(t); x(P) = x'(0) = 0, in which


M(/) = sint — ait — Itt) sin {t — 2tt).

ANS. x{t) = J[l — ix{t — 2tt)](1 sin — sin 2t). /

17. Compute yilTr) and y{2 + ^tt) for the function y{x) that satisfies the initial
value problem

y"{x) + y{x) = (x- 2)a(x - 2); >'(0) = 0, >^'(0) = 0.


ANS. yih-n) = 0, j(2 + Itt) = ^77 - 1

18. Compute x(_l) and x(4) for the function x{t) that satisfies the initial value prob-
lem

x'XO + 2x'{t) + x{t) = 2 + (/ - 3)a(/ - 3); a:(0) = 2, a:'(0) - 1.

ANS. jc(l) = 2 + e-', a:(4) = + 1 3e-» +4e-*.


19. Solve the problem

x"{t) + 2x'(0 + xit) = F{ty,x{0) = 0, x'ifS) = 0.

ANS. x{t) = \ ^e-Tit - P) d^.


*'o

20. Solve the problem

y'V) - k'yit) = Hity,yiO) = 0, >''(0) = 0.


1 r'
ANS. y(t) =- H{t-^) sinh k^ d^.
k Jq
21. Solve the problem

y'V) + 4yV) + 13X0 - F(0; .v(O) = 0, /(O) = 0.


22. Solve the problem

x"iO + 6xXt) + 9x(t) = Fit); x(0) = A, x'{0) = B.


t

ANS. xit)^e-^'[A + iB+3A)t]+ ! ^e'^Tit - ^) d^.

In Exs. 23-42, use the Laplace transform method with the realization that these
exerciseswere not constructed with the Laplace transform technique in mind.
Compare your work with that done in solving the same problems by the methods of
Chapter 7. See pages 119-120.

23. Ex. 1. 24. Ex.2.


25. Ex. 3. 26. Ex. 11.
27. Ex. 14. 28. Ex. 20.
29. Ex. 21. 30. Ex. 22.
31. Ex. 23. 32. Ex. 36.
33. Ex. 37. 34. Ex. 38.
35. Ex. 39. 36. Ex. 40.
37. Ex. 41. 38. Ex. 42.
194 Inverse Transforms [Ch. 11

39. Ex. 43. 40. Ex. 44.


41. Ex. 45. 42. Ex. 46.
43. Solve the problem

x'V) - 4x'(t) + 4x(r) = e" ; x'{0) = 0, x(l) = 0.


ANS. x(t) = Kl - tye^'.
44. Solve the problem

x'V) + 4x(t) = -8/^ x(0) = 3, xiiv) = 0.


ANS. x{t) = 2 cos 2t + (Itt^ - 1) sin 2/ + 1 - 2f ^

65. Special integral equations

A difFerential equation may be loosely described as one that contains a


derivative of a dependent variable; the equation contains a dependent variable
under a derivative sign. An equation that contains a dependent variable under
an integral sign is called an integral equation.

Because of the convolution theorem, the Laplace transform is an excellent


tool for solving a very special class of integral equations. We know from
Theorem 20 that if

L{F{t)] =f{s)
and

L{G{t)} ^ g{s),
then

LUF(p)G{t-^)dp\=f{s)g(s)

The relation (1) suggests the use of the Laplace transform in solving equations
that contain convolution integrals.

EXAMPLE (a): Find F{t) from the integral equation

f(t) = 4/ - 3 f F(p) sin (t - /?) dp. (2)

The integral in (2) is in precisely the right form to permit the use of the
convolution theorem. Let

L{F(t)} =/(5).
Then, since

L{smt} = -j-—,
s^ + 1
§ 65] Special integral equations 195

application of Theorem 20 yields

L{ F(P) sin (t-fi)dfi\ = As)


ii:

Therefore, the Laplace operator converts equation (2) into

4 3/(5)
f(') = -2-^^-
S^ 5^+1 (3)

We need to obtain/(5) from (3) and then F{t) from/(5). From (3) we get

(i+7TtV(^^ = ?'
or
4(5^+1) 1 ^
s\s^ + 4) s^ s' +4
Therefore

F(t) = L-'{\ +
or
F(t) = t+ fsin2/. (4)

That the F{t) of (4) is a solution of equation (2) may be verified directly.
Such a check is frequently tedious. We shall show that for the F of (4), the
right-hand side of equation (2) reduces to the left-hand side of (2). Since

RHS = 4t-3 f
()6 + f sin 2p) sin (/ - p) dfi,

we integrate by parts with the choice shown in the table.

(p + i sin 2P) sin (/ - ^) dp

(1 + 3 cos 2p) dp cos {t - P)

It thus follows that

RHS = 4f - 3[(i? + t sin ip) cos (t - j8)]|^

+ 3 f (1 + 3 cos 2P) cos (t - P) dp.

from which

RHS = 4t-3{t + i sin 20 + 3 f cos {t - p) dp


•'o

+ 9 f
cos 2p cos {t - p) dp.
196 Inverse Transforms [Ch. 11

or

RHS = - r
f sin 2r - 3[sin (t - p)] + f [cos {t + P) + cos (t - 3P)\dp.
J

This leads us to the result

RHS = - f
f sin 2f + 3 sin r + f [sin (t + P) - ^ sin (t - 3p)\
1-
-'o

=t- f sin 2r + 3 sin t + f sin 2r + I sin 2f - f sin r + f sin t,

or
RHS = + ism2t = F(t) = LHS,
t

as desired.
It is important to realize that the original equation

F(t) = 4t-3 (f(P) sin (t - P) dp (2)

could equally well have been encountered in the equivalent form

F{t) = 4f - 3 \ F{t- P) sin p dp.

An essential ingredient for the success of the method being used is that
the integral involved be in exactly the convolution integral form. We must
have zero to the independent variable as limits of integration, and an inte-

grand that is the product of a function of the variable of integration by a


function of the difference between the independent variable and the variable
of integration. The fact that integrals of that form appear with significant
frequency in physical problems keeps the topic of this section from being
relegated to the role of a mathematical parlor game.

EXAMPLE (b): Solve the equation

g{x) = ix^ - f(x - y)g(y) dy. (5)

Again the integral involved is one of the convolution type with x playing
the role of the independent variable. Let the Laplace transform of g{x) be
some as yet unknown function h(z)

L{gix)} = h{z). (6)

Since L{\x^] = -3 and L{x} = -j , we may apply the operator L throughout

(5) and obtain

/7(z)=-3--2-,
z z
§ 65] Special integral equations 197

from which

(-M Kz) = \,
or

=
1 z^ + l-z'
/i(z)
z(z^ + 1) + 1)
ziz"" z z^ + 1

Then

g(x) = L-^i-- '


[z z^ + 1

or
g(x) = 1 - cos X. (7)

Verification of (7) is simple. For the right member of (5) we get

RHS = ix^ - f (x - y)(l - cos y) dy

= i^^ - f(^
- y){y - sin y)] - ! (y - sin >;) dy
Jo *'o

= ix' - - hy^ + cos ^'l"


^ Jo

= ^x^ — |x^ — COS X + 1 = 1 — cos X = LHS.

Exercises
In Exs. 1^, solve the given equation and verify your solution.

1. F(t) = +
1 2 f F(t-^)e-^^ d^. ANS. Fit) = 1+2/.
t

2. F(r) = +
1
f
F((8) sin (t - ^) d^. ans. F(0 = 1 + i^'.

3. Fit) = t+ f Fit-^)e-'' d^. ANS. F(0 =/+ if ^.

4. F(0 = 4r2 - f F(r - jS^-" ^jS. ans. F(0 = -1 + 2/ + 2/^ + e"^'.

In Exs. 5-8, solve the given equation. If sufficient time is available, verify your
solution.

5. Fit) = t'+ \ Fi^) sin (/ - j8) d^. ans. Fit) = t^ + ^ot'.


t

6. Fit) = 8f ^ - 3 Fi^) sin (/ - ^) d^.


J
198 Inverse Transforms [Ch. 11

7. Fit) = r^ - 2 f F{t-^) sinh 1^ d^. ans. F{t) = t^-lt\


I

8. Fit) = +2
1 f F(/ - iS) cos ]8 d^. ans. Fit) = + 2te'.
1

'o

In Exs. 9-12, solve the given equation.

9. Hit) = 9e^' -l\ Hit-^)cos^d^.


•'o

10. Hiy)=y^+ ! Hix) sin iy - x) dx. ans. Hiy)^y^^-^ y''.


-'o

11. gix) - e-^ - 2 [ gi^) cos (x - jS) JjS. ans. ^(x) = ^-^(1 - x)\
•'o

12. K0 = 6r + 4f i^-tYyi^)d^.

ANS. >»(/) = ^2. — ^-'(cosVSr— VSsinVs/).


13. Solve the following equation for Fit) with the condition that F(0) = 4:

F'(/) = +
/ f Fit-^) cos ]8 ^^. ANS. Fit) = 4 + f r^ + 2^ r*.

14. Solve the following equation for Fit) with the condition that F(0) == 0:
r

F'it) = sin +
/ f Fit - iS) cos ^3 d^. ans. Fit) - it\

15. Show that the equation of Ex. 3 above can be put in the form

e'Fit) = te'+ f ei'Fi^) d^. (A)

Differentiate each member of (A) with respect to t and thus replace the integral
equation with a differential equation. Note that F(0) = 0. Find Fit) by this
method.

16. Solve the equation

f Fit-^)e-^d^^t

by two methods; use the convolution theorem and the basic idea introduced in
Ex. 15. Note that no differential equation need be solved in this instance.

TABLE OF TRANSFORMS
Whenever /; is used, it denotes a non-negative integer.
The range of validity may be determined from the appropriate text
material. Many other transforms will be found in the examples and exercises.
Table of transforms 199

/(.)= L{Fit)} Fit)

As -a)
200 Inverse Transforms [Ch. 11

fXs)^F{LU)}
CHAPTER LJi

Applications

66. Vibration of a spring

Consider a steel spring attached to a support and hanging downward.


Within certain elastic limits the spring will obey Hooke's law: if the spring is

stretched or compressed, its change in length will be proportional to the force


exerted upon the spring and, when that force is removed, the spring will return
to its original position with its length and other physical properties unchanged.
There is, therefore, associated with each spring a numerical constant, the ratio
of the force exerted to the displacement produced by that force. If a force of
magnitude Q pounds stretches the spring c feet, the relation

Q= kc (1)

defines the spring constant k in units of pounds per foot.

Let a body B weighing w pounds be attached to the lower end of a spring


(Figure 24) and brought to the point of equilibrium where it can remain at

201
202 Applications [Ch. 12

^
§ 66] Vibration of a spring 203

Suppose an additional vertical force, due to the motion of the support or to


presence of a magnetic field, etc., is imposed upon the system. The new,
impressed force, will depend upon time and we may use F(t) to denote the
acceleration which it alone would impart to the weight B. Then the impressed

force is — F{t) and equation (2) is replaced by


ff

w w
- x"(t) + bx'(t) + kx(t) = - F(t). (3)
9 g

At time zero, let the weight be displaced by an amount Xq from the equi-
librium point and let the weight be given an initial velocity t'o • Either or
both of Xq and Vq may be zero in specific instances. The problem of deter-
mining the position of the weight at any time t becomes that of solving the
initial value problem consisting of the differential equation

w w
- x"(0 + hx'{t) + kx{t) = - F(0, for t > 0, (4)
9 9

and the initial conditions

xiO) = Xo, x'{0) = vo. (5)

It is convenient to rewrite equation (4) in the form

x"(t) + 2yx'(t) + p^x(t) - F(t), (6)


in which we have put

w w
We may choose ^ > and we know y ^ 0. Note that 7 = corresponds to a
negligible retarding force.

Impressed forces so ill-behaved that their Laplace transforms do not


exist are hardly worth considering. We therefore let

L{x(t)} = uis), L{F(t)}=f(s), (7)

and obtain from the problem (5) and (6) the transformed problem

s^u(s) - sxq - Vo + 2ylsu(s) - Xq] + p^u(s)=f(s),

"(s) =
SX0
s^
2
+ V0 + 27x0
+ ^2ys +n^T-
.

P""
—+
s^ +
/(s)

2ys + p^
'

which we write as

_ xpjs + y) 4-1^0 -F yxp f{s)


^^^~ ^^
(s + yf + p'-y' is + y)' + p'-y''
204 Applications [Ch. 12

The desired jc(/) is the inverse transform of u(s).


Here we begin to reap benefits from our use of the Laplace transform.
Even before we get a'(/) we can see that its form will depend upon whether
the denominator in (8) has distinct real linear factors, or equal factors, or
is the sum of two squares. That is, the form of x(t) will depend upon whether
P <y, li = y,or P > y. Furthermore, if y 7^ 0, we can predict the presence of a
damping factor e~^' in x(t).
The inverse of the u(s) of equation (8) can be obtained with our standard
methods, including the convolution theorem. When the Fit) is reasonably
simple, one of the types which occur frequently in practice, the convolution
theorem leads to forms less desirable than those obtained by the other methods
in Chapter 1 1

We shall now consider separately the various situations that arise according
to the choice of parameters P and y and the forcing function F{t) in this

problem of the vibrating spring.

In treating spring problems we are particularly interested in impressed


forces that lead to F{t) of forms such as those listed:

(a) No impressed force: F{t) = 0;

(b) Simple harmonic forcing function:

F(t) = Ai sin ojt + A2 cos cot;

(c) Temporary constant force

F(t)^A, 0<t<to,
= 0, t>to;

(d) A delayed form of (c)

Fit) = 0, 0<t<to,
= A, to<t<ti,
= 0, t>ti;

(e) Fit) = the half-wave rectification of a sine curve.

become apparent when we study electric circuits in Chapter 14 that


It will

all of our present work carries over to simple circuit problems with new
meanings attached to the parameters and functions involved. In circuit theory

we shall add to the forms of Fit) listed above both Fit) equal to a constant
and F(t) equal to the square-wave function that appeared in Section 60.
§ 67] Undamped vibrations 205

67. Undamped vibrations

If y = in the problem of Section 66, the transform becomes

-(^) = ^r^Y + ^r^2- (1)

The inverse transform of the first term on the right in (1) is

Vq
Xq cos Pt + -f:
sin pt,

which contributes to x(t) a simple harmonic motion, called the natural


component of the motion, natural because it is independent of the forcing
function Fit). From the last term in (1) we obtain the forced component, the
form of which is dependent upon F{t).
Instead of jumping from (1) to the general solution

x(0 = Xo cos pt + VoP~^ smpt + p~^ f F(t - v) sin pv dv, (2)

we i^refer to obtain x(/) without a convolution integral for the specific types
of F{t) which occur often. Equation (2) was easily obtained but it leaves all
the labor of simplification to the unfortunate user who has a particular F{t)
with which to deal.

EXAMPLE (a): Solve the spring problem with no damping but with a
forcing function F{t) =A sin cot.

The differential equation of motion is

w w
- x"(0 + ^^(0 = - ^ sm cot
9 9
and is readily put into the form

x"(0 + P^x(t) =A sin wt (3)

with the introduction of P^ = kg/w. We shall assume initial conditions

X(0) = Xo ,
X'(0) = Vq .

Let L{x(t)] = uis). Then (3) and (4) yield

Aco
s^m(s) — sxq — Vo + P u(s) = 2'
S^ + O)
206 Applications [Ch. 12

or

"(5) = .2 . 01 + ..2 , o2v 2 ,


— r>

(5)

The last term in (5) will lead to different inverse transforms according to
whether to = /? or w# ^. The case oj = jS leads to resonance, which will be
discusssed in the next section.
U CO ^ P, equation (5) yields

sxo + vo Acq ( 1 1 \ ...

From (6) it follows at once that

-1 n Aio ^
+ ——^
.

x(0 = xo cos pt + VoP sm pt -^r sm ^t ^ ^ sm cor. (7)

That the x of (7) is a solution of the problem (3) and (4) is easily verified. A
study of (7) is simple and leads at once to conclusions such as that x(/) is

bounded, etc. The first two terms on the right in (7) yield the natural har-

monic component of the motion, the last two terms form the forced com-
ponent.

EXAMPLE (b): A spring is such that it would be stretched 6 inches by a


12-pound weight. Let the weight be attached to the spring and pulled down
4 inches below the equilibrium point. If the weight is started with an upward
velocity of 2 feet per second, describe the motion. No damping or impressed
force is present.
We know that the acceleration of gravity enters our work in the expression

for the mass. We wish to use the value g = 2>2 ft. per sec. per sec. and we must
use consistent units, so we put all lengths into feet.

First we determine the spring constant k from the fact that the 12-pound
weight stretches the spring 6 inches, \ ft. Thus \2 ^ \k so that k = 2A lb.

per ft.

The differential equation of the motion is therefore

i|x"(r) + 24x(r) = 0. (8)

At time zero the weight is 4 inches (3 ft.) below the equilibrium point, so
x(0) = \. The initial velocity is negative (upward), so x'(0) = — 2. Thus our
problem is that of solving

x"(0 + 64x(0 = 0; x(0) = i, x'(0)=-2. (9)

We let L{x{t)] = 1/(5) and conclude at once that

s^u(s) -^s + 2 + 64u(s) = 0,


§ 67] Undamped vibrations 207

from which

u(s) = -^ .

Then
4/) = 1 cos 8/ - i sin 8/. (10)

A detailed study of the motion is straightforward once (10) has been


obtained. The amplitude of the motion is

VW+df = A
that is, the weight oscillates between points 5 inches above and below E.
The period is ^n sec.

EXAMPLE (c): A spring, with spring constant 0.75 lb. per ft., lies on a
long smooth (frictionless) table. A 6-lb. weight is attached to the spring and
is at rest (velocity zero) at the equilibrium position. A 1.5-lb. force is applied
to the support along the line of action of the spring for 4 seconds and is then
removed. Discuss the motion.
We must solve the problem

^x"(t) + ix{t) = H(ty, x(0) = 0, x'(0) = 0, (11)

in which

//(/)= 1.5, 0</<4,


= 0, t>4.
Now H(t)^ 1.5[1 — a(/ - 4)] in terms of the a function of Section 62.
Therefore we rewrite our problem (11) in the form

x"(0 + 4a-(/) = 8[1 -a(/-4)]; x(0) = 0, x'(0) = 0. (12)

Let L{x{t)} = u{s). Then (12) yields

g
s^u(s) + 4uis) = - (I -e"^0,
s

or
8(1 - e-'^)
U(S) — ;5
^ ^
sis^ + 4)

The desired solution is

x(t) = 2(1 - cos 20 - 2[1 - cos 2{t - 4)]a(/ - 4). (13)


208 Applications [Ch. 12

Of course, the solution (13) can be broken down into the two relations

For ^ ^ r 4, x(t) = 2(1 - cos 2t), (14)

For t > 4, x(0 = 2[cos 2(r - 4) - cos 2Q, (15)

if those forms seem simpler to use.


Verification of the solution (13), or (14) and (15), is direct. The student
should show that

Lim x(0 = Lim x(0 = 2(1 - cos 8) = 2.29


t->4- f-»4 +
and
Lim x'(r) = Lim x'(t) = 4 sin 8 = 3.96.
r-4- t-4+

From (13) or (14) we see that in the range < < / 4, the maximum deviation
of the weight from the starting point is x = 4 ft. and occurs SLt t = ^n — 1.57
sec. At / = 4, .Y = 2.29 ft. as shown above. For / > 4, equation (15) takes over
and thereafter the motion is simple harmonic with a maximum x of 3.03 ft.

Indeed, for / > 4,

max |x(OI = 2^(1 - cos 8)^ + sin^ 8

= 2^2 Jl - cos 8

= 272.2910 = 3.03.

Example (c) is one type of problem for which the Laplace transform
technique is particularly useful. Such problems can be solved by the older
classical methods, but with much less simplicity and dispatch.

68. Resonance

In the problem of undamped vibrations of a spring, one of the examples


we encountered was
x"(t) + P^x{t) = A sin cot\ a-(0) = Xq , x'(0) = v^ .
(1)

The solution of (1) was obtained by putting L{x{t)} = u{s) and getting
SXq + Vq Aco
"(•^) = .2 , 02 + /.2 , o2u- 2 , ..2> • (2)

From we see that the inverse transform differs according to whether


(2) co =P
or CO^ p. The latter has been studied. Let us now assume w = p.
Our problem is to solve
x"(t) + p^x(t) -A sin pt; x{0) = Xo , x'(0) = v^ , (3)
§ 68] Resonance 209

with the aid of

sxq + Vq Ap
K^) = -?i^T^ + 7Tr^?^- (4)

We already know, from page 167, that

i-'|(jr^| = 2j3(si"/"-/"cos/!/).
Therefore (4) leads us to the solution

V A
x(t) = Xo cos Pt + ^ sin pt + —^ (sin pt - pt cos pt). (5)

That (5) satisfies all the conditions of (3) is readily verified.


In the solution (5) the terms proportional to cos pt and sin pt are bounded,
but the term with pt cos Pt can be made as large as we wish by proper choice
of /. This building up of large amplitudes in the vibration when co = /? in (1)
is called resonance.

Exercises
1. A spring is such that a 5-pound weight stretches it 6 inches. The 5-pound weight
is attached, the spring reaches equilibrium, then the weight is pulled down 3
inches below the equilibrium point and started off with an upward velocity of
6 ft. per sec. Find an equation giving the position of the weight at all subsequent
times. ANS. x = i(cos 8/ — 3 sin 8/).
2. A spring is stretched 1.5 inches by a 2-pound weight. Let the weight be pushed
up 3 inches above E and then released. Describe the motion.
ANS. X = — icos 16/.
3. For the spring and weight of Ex. 2, let the weight be pulled down 4 inches
below E and given a downward initial velocity of 8 ft. per sec. Describe the
motion. ans. x = cos 16/ + ^ sin 16/.
J

4. Show that the answer to Ex. 3 can be written x = 0.60 sin (16/+ ^) where
= Arctan f
(f)

5. A spring is such that a 4-pound weight stretches it 6 inches. An impressed force


i cos 8/ is acting on the
4-pound weight is started from the equi-
spring. If the
librium point with an imparted upward velocity of 4 ft. per sec, determine the
position of the weight as a function of time. ans. x = l(t — 2) sin 8/.
6. A spring is such that it is stretched 6 inches by a 12-pound weight. The 12-pound
weight is pulleddown 3 inches below the equilibrium point and then released.
If there is an impressed force of magnitude 9 sin 4/ pounds, describe the motion.
Assume that the impressed force acts downward for very small /.

ANS. X = i cos 8/ — i sin 8/ + i sin 4/.


7. Show that the answer to Ex. 6 can be written

X = iV2 cos (8/ + 77/4) + i sin 4/.


210 Applications [Ch. 12

8. A spring is such that a 2-pound weight stretches it I ft. An impressed force


i sin 8r is acting upon the spring. If the 2-pound weight from a point is released
3 inches below the equilibrium point, determine the equation of motion.
ANS. x = Kl — cos 8/ + 5^ sin St (ft.).

9. For the motion of Ex. 8, find the first four times at which stops occur and find
the position at each stop. ans. t = ttjS, -njA, 1, 37r/8 (sec.)

and A--= -0.15, ^0.05, -^0.03, +0.04 (ft.), respectively.


10. Determine the position to be expected, if nothing such as breakage interferes, at
the time of the 65th stop, when r = Stt (sec), in Ex. 8.
-6.0 (ft.).
ANS. x-
11. A spring is such that a 16-pound weight stretches it The weight is
1.5 inches.

pulled down to a point 4 inches below the equilibrium point and given an
initial downward velocity of 4 ft. per sec. An impressed force of 360 cos At
pounds is applied. Find the position and velocity of the weight at time t = tt/S
seconds. ans. At r = 77/8(sec.), x = —f (ft.), y = —8 (ft. /sec.).

12. A spring is stretched 3 inches by a 5-pound weight. Let the weight be started
from E with an upward velocity of 12 ft. per sec. Describe the motion.
ANS. a: = —1.06 sin 11.3/.
13. For the spring and weight of Ex. 12, let the weight be pulled down 4 inches
below E and then given an upward velocity of 8 ft. per sec. Describe the
motion. ans. x = 0.33 cos 1 1.3/ — 0.71 sin 1 1.3/.

14. Find the amplitude of the motion in Ex. 13. ans. 0.78 ft.
15. A 20-pound weight stretches a certain spring 10 inches. Let the spring first be
compressed 4 inches, and then the 20-pound weight attached and given an
initial downward velocity of 8 ft. per sec. Find how far the weight would drop.

ANS. 35 in.

16. A spring is such that an 8-pound weight would stretch it 6 inches. Let a 4-
pound weight be attached to the spring, pushed up 2 inches above its equilibrium
point and then released. Describe the motion. ans. x ^ —\ cos 11.3/.

17. If the 4-pound weight of Ex. 16 starts at the same point, 2 inches above E,
but with an upward velocity of 15 ft. per sec, when will the weight reach its
lowest point ? ans. At / = approximately 0.4 sec.
18. A spring is such that it is stretched 4 inches by a 10-pound weight. Suppose
the 10-pound weight to be pulled down 5 inches below E and then given a
downward velocity of 15 ft. per sec. Describe the motion.
ANS. X = 0.42 cos 9.8/ +1.53 sin 9.8/
= 1.59 cos (9.8/ — ^), where = Arctan 4> 3.64.
19. A spring is such that it is stretched 4 inches by an 8-pound weight. Suppose the
weight to be pulled down 6 inches below E and then given an upward velocity
of 8 ft. per sec. Describe the motion.
ANS. x = 0.50 cos 9.8/ -0.82 sin 9.8/.
20. Show that the answer to Ex. 19 can be written jc= 0.96 cos (9.8/ + where <f))

<f>
= Arctan 1.64.

21. A spring is such that a 4-pound weight stretches it 6 inches. The 4-pound
weight is attached to the vertical spring and reaches its equilibrium point. The
weightis then (/ = 0) drawn downward 3 inches and released. There is a simple

harmonic exterior force equal to sin 8/ impressed upon the whole system. Find
§ 69] Dumped vibrations 211

the time for each of the first four stops following / = 0. Put the stops in chrono-
logical order. ans. / -^ 7r/8,
J, 7r/4, Stt/S (sec.).
22. A spring is stretched 1.5 inches by a 4-pound weight. Let the weight be pulled
down 3 inches below equilibrium and released. If there is an impressed force
8 sin 16/ acting upon the spring, describe the motion.
ANS. X = 1(1 — 8/) cos 16/ + J sin 16/.
23. For the motion of Ex. 22, find the first four times at which stops occur and find
the position at each stop. ans. / = i, 77/I6, Tr/8, 377-/16 (sec.) and
X = +0.11, +0.14, -0.54, +0.93 (ft.), respectively.

69. Damped vibrations

In the general linear spring problem of Section 66, we were confronted


with

x"(t) + 2yx'{t) + p'xiO = Fit); x(0) = Xo , x'(0) = i;o , (1)

in which y ^ —w and (^^ = — , B > 0. We let


w
L{x(t)} = u(s), L{F(t)}=f{s)

and obtained

. . ^ ^o(^ + y) + + yxp t^o f(s)


"^'^ ^^^
(s + y)' + P'-y' '^(s + y)' + p'-y'-
We are now concerned with 7 > 0; a non-negligible retarding force is present.
We know that the form of the inverse transform of i/(5) depends upon whether
> y, P = y, or P <y.
f^

If p > y, P^ - y^ > 0, so let us put

P'-y' = 5\ (3)
Then (2) becomes

and we arrive at

x(0 = e'^'lxo cos St + (vq + yxo)S~^ sin ^r] + i]/i{t) (5)

in which iAi(/) is an inverse transform of the last term in (4). The function
i//i(/) can always be written in terms of F(t) by the convolution theorem,
but that leads to undesirable complications for the most common choices
of F{t). In the solution (5), the presence of the damping factor e~^' shows
that the natural component of the motion approaches zero as / -> 00.
212 Applications [Ch. 12

If in (1) and (2) we have [i — y, the transform

"(i) = r—
(s+
— 71
7)2
+
(s + y)'
(6)

leads us to the solution

x(0 = e-'X^o + (vo + 7-Vo)0 + ^2(t), (7)

in which ipiiO is an inverse of {s + y)~ ^f{s) and is easily determined once


F{t) is stipulated. Again the natural component has the damping factor
,->•'
in It.

If in (1) and (2) we have /? < y, 7^ — /?^ > 0, and we put

y'-p' = a'. (8)

Then (2) becomes

. .
= Xo(s + r) + % + 7^0 ^
fis)
"(5) 7^; h (9)
(5 + 7)^ - <r'
:

(s + 7)2-(t''

which leads to the solution

.y(/) = e ''[xq cosh at + (vq + 7a-o)o' ^


sinh at] + xj/ii^). (10)

in which 1/^3(0 is an inverse of the last term in (9). By (8) we see that y > a,

so that once more the natural component of the x{t) in (10) approaches zero
as /^ 00.

Suppose for the moment that we have F{t) = 0, so the natural component
of the motion is all that is under consideration. If /? > 7, equation (5) holds
and the motion is a damped oscillatory one. If /? = 7, equation (7) holds and

Critically damped

FIGURE 26
§69] Damped vibrations 213

the motion is not oscillatory; it is called critically damped motion. If /? < y,


(10) holds and the motion is said to be overdamped; the parameter y is larger
than it needs to be to remove the oscillations. Figure 26 shows a representative
graph of each type of motion mentioned in this paragraph, a damped os-

cillatory motion, a critically damped motion, and an overdamped motion.

EXAMPLE: Solve the problem of Example (b). Section 67, with an added
damping force of magnitude 0.6 \v\. Such a damping force can be realized by
immersing the weight B in a thick liquid.
The initial value problem to be solved is

iix"(t) + 0.6x'(t) + 24x(t) = 0; x(0) = i, jc'(O) = -2. (11)

Put L{x(/)} = u(s). Then (11) yields

(s^ +\.6s + 64)u{s) = Us - 4.4),

from which we obtain

^-^-^
40^ ^
^L-^f
^
]
\(s + 0.8)2 _^ 53 3^1

5-5.2
= iexp(-0.80L"
[s~ + 63.36)

Therefore the desired solution is

x(t) = exp (-0.80(0.33 cos 8.0/ - 0.22 sin 8.0/), (12)

a portion of its graph being shown in Figure 27.

FIGURE 27
214 Applications [Ch. 12

Exercises
1. A certain straight-line motion is determined by the diflferential equation

- + 2y-+169.=0
d^x dx

and the conditions that when / == 0, x = and y = 8 ft. per sec.


(a) Find the value of y that leads to critical damping, determine x in terms of
t, and draw the graph for ^ / ^ 0.2.
ANS. y = 13(l/sec.), x = 8/e-'^'.
(b) Use y = 12. Find x in terms of / and draw the graph.
ANS. X = 1.6e~*^' sin 5^
(c) Use y = 14. Find x in terms of / and draw the graph.

ANS. x = 0.77(e-««' -f--'"-^').


2. A spring is such that a 2-pound weight stretches it J foot. An impressed force
i sin 8/ and a damping force of magnitude r| are both acting on the spring. The
|

weight starts \ foot below the equiUbrium point with an imparted upward
velocity of 3 ft. per sec. Find a formula for the position of the weight at time t.

X =z2e-^'0 — 8/) — ii cos 8/.


ANS.
3. A spring is such that a 4-pound weight stretches it 0.64 feet. The 4-pound weight
is pushed up 1 foot above the point of equilibrium and then started with a

downward velocity of 5 ft. per sec. The motion takes place in a medium which
furnishes a damping force of magnitude \\v\ at all times. Find the equation
describing the position of the weight at time /.

ANS. X = Je~'(2sin 7/ — cos 7/).


4. A spring is such that a 4-pound weight stretches The weight is it 0.32 feet.

attached to the spring and moves in a medium which furnishes a damping


force of magnitude l\v\. The weight is drawn down i foot below the
equilibrium point and given an initial upward velocity of 4 ft. per sec. Find the
position of the weight thereafter. ans. = \e~^\A cos 8/ — sin 8/). jc

5. A spring is such that a 4-pound weight stretches the spring 0.4 feet. The 4-pound
weight is attached to the spring (suspended from a fixed support) and the system
is allowed to reach equilibrium. Then the weight is started from equilibrium
position with an imparted upward velocity of 2 ft. per sec. Assume that the
motion takes place in a medium which 'furnishes a retarding force of magnitude
numerically equal to the speed, in feet per second, of the moving weight. Deter-
mine the position of the weight as a function of time.
ANS. X = — it-""*' sin 8/.
6. A spring is stretched 6 inches by a 3-pound weight. The 3-pound weight is
attached to the spring and then started from equilibrium with an imparted
upward velocity of 12 ft. per sec. Air resistance furnishes a retarding force
equal in magnitude to 0.03| v\ . Find the equation of motion.
ANS. x= -1.5e-°**'sin 8r.

7. A spring is such that a 2-pound weight stretches it 6 inches. There is a damping


force present, with magnitude the same as the magnitude of the velocity. An
impressed force (2 sin 8/) is acting on the spring. If, at / = 0, the weight is

released from a point 3 inches below the equilibrium point, find its position for
/>0. ANS. JC = (i + 4/>"^' — icos 8r.
§ 69] Damped vibrations 215

8. A spring is 4-pound weight. The weight is started 6


stretched 10 inches by a
inches below the equilibrium point with an upward velocity of 8 ft. per sec.
If a resisting medium furnishes a retarding force of magnitude \\ v\ describe the ,

motion. ans. x £-"'[0.50 cos 6.1/ — 1.23 sin 6.1/].

9. For Ex. 8, fmd the times of the first three stops and the position (to the nearest
inch) of the weight at each stop.
ANS. /i = 0.3 sec, xi = — 12 in. tz = 0.8 sec,
;

X2 = +6 in.; /s= 1.3 sec, Xs = —4 in.


10. A is stretched 4 inches by a 2-pound weight. The 2-pound weight is
spring
started from the equilibrium point with a downward velocity of 12 ft. per sec.
If air resistance furnishes a retarding force of magnitude 0.02 of the velocity,
describe the motion. ans. x = 1.22e~°'^' sin 9.8/.
11. For Ex. 10, find how long it takes the damping factor to drop to one-tenth its
initial value. ans. 14.4 sec.
12. For Ex. 10, find the position of the weight at : (a) the first stop; (b) the second
stop. ANS. (a)jc = 1.2ft.;(b)A:= -1.1 ft.
13. Let the motion of Ex. 8, page 210, be retarded by a damping force of magnitude
0.6| v\ . Find the equation of motion.
ANS. = O.SOi'-* cos 6.4/ + 0.22£'-'^-'' sin 6.4/ - 0.05 cos 8/ (ft.).
X **'

14. Show that whenever > (sec), the solution of Ex. 13 can be replaced (to the
/ 1

nearest 0.01 -0.05 cos 8/.


ft.) by ;c =
15. Let the motion of Ex. 8, page 210, be retarded by a damping force of magnitude
I
v\ Find the equation of motion and also determine its form (to the nearest 0.01
.

ft.) for / > 1 (sec).


X =3-2(8/ + l)e-^' — sV cos 8/ (ft.); for / > 1, x = — ^ cos 8/.
ANS.
16. Let the motion of Ex. 8, page 210, be retarded by a damping force of magnitude
3 v\
1 Find the equation of motion.
.

ANS. X = 0.30e-'«'3^' - 0.03e- ^*' - 0.02 cos 8/.


17. Alter Ex. 6, page 209, by inserting a damping force of magnitude one-half that
of the velocity and then determine x.
ANS. X = exp (-f/)(0.30 cos 8.0/ - 0.22 sin 8.0/) - 0.05 cos 4/ + 0.49 sin 4/.
18. A spring is stretched 6 inches by a 4-pound weight. Let the weight be pulled
down 6 inches below equilibrium and given an initial upward velocity of 7 ft.

per sec. Assuming a damping force twice the magnitude of the velocity, des-
cribe the motion and sketch the graph at intervals of 0.05 sec. for ^ / ^ 0.3
(sec). ANS. x = ^e~^'(l —6t).
19. An object weighing w pounds is dropped from a height /; feet above the earth.
At time /(sec) after the object is dropped, let its distance from the starting point
be jc (ft.), measured positive downward. Assuming air resistance to be negligible,
show that X must satisfy the equation
w d^x
=w
gdt^
as long as X < h. Find x. ans. x = Igf^.
20. Let the weight of Ex. 19 be given an initial velocity Do . Let v be the velocity at
time /. Determine v and x. ans. v ^ gt -\- v^^^x — \gt^ + vq t.

21. From the results in Ex. 20, find a relation that does not contain / explicitly.
ans. y^ = Vq- + Igx.
216 Applications [CIi. 12

22. If air resistance furnishes an additional force proportional to the velocity in the
motion studied in Exs. 19 and 20, show that the equation of motion becomes
w d^x dx
g dt^ dt

Solve equation (A) given the conditions

when / =: 0, X = = vo.
and v (B)

Use a ^ bg/w. ans. x = a~^gt + a-\avo — g)il — e'").


23. To compare the results of Exs. 20 and 22 when a = bgjw is small, use the power
series for e"" in the answer for Ex. 22 and discard all terms involving a" for
n ^ 3. ANS. x = ^gt^ + Vot- ^at ^(3vo + gt) +T4a^/^(4yo + gt).
24. The equation of motion of the vertical fall of a man with a parachute may be
roughly approximated by equation (A) of Ex. 22. Suppose a 180-pound man
drops from a great height and attains a velocity of 20 miles per hour after a
long time. Determine the implied coefficient b of equation (A).
ANS. 6.1 (lb.)(sec.) per ft.

25. A particle is moving along the x-axis according to the law

d^x dx
+
^
6 — + 25x = 0.
dt' dt

If the particle started at x = with an initial velocity of 12 ft. per sec. to the
left, determine: (a) x in terms of /; (b) the times at which stops occur; and (c)

the ratio between the numerical values of x at successive stops.


ANS. (a)x = —3e~^' sin 4/,
(b) t = 0.23 + inn, = 0, rt 1, 2, 3, . .
. ,
(c) 0.095.

70. The simple pendulum and the deflection of beams

A rod of length C feet is suspended by one end so it can swing freely in a

vertical plane. Let a weight B (the bob) of w pounds be attached to the free
end of the rod, and let the weight of the rod be negligible compared to the

weight of the bob.


Let (radians) be the angular displacement from the vertical, as shown in

Figure 28, of the rod at time / (sec). The tangential component of the force

w (lb.) is IV sin 9 and it tends to decrease 0. Then, neglecting the weight of the
rod and using S = CO as a measure of arc length from the vertical position,

we may conclude that


wifS
2 = — vv sin 9. (1)
9 dt
Since S — C6 and C is constant, (1) becomes

fe + ^g sin = 0. (2)
2
~d?^C
§70] The simple pendulum and the deflection of beams 217

FIGURE 28

The solution of equation (2) is not elementary; it involves an elliptic in-


tegral. If is small, however, sin and are nearly equal and (2) is closely
approximated by the much simpler equation

+ ^0 = 0;
" P^=^. (3)
dt""
'
'
'
C
The solution of (3) with pertinent initial conditions gives usable results
whenever those conditions are such that 9 remains small, say \0\ < 0.3
(radians).

As a further example, we consider a beam of length 2c, as shown in Figure

29. Denote distance from one end of the beam by x, the deflection of the
beam by ;•. If the beam is subjected to a vertical load lV(x), the deflection y
must satisfy the equation

EI = W(x), for < X < 2c, (1)


d^
in which E, the modulus of elasticity, and /, a moment of inertia, are known
constants associated with the particular beam.
The slope of the curve of deflection isjX-^')^ the bending moment is EIy"{x),
and the shearing force is EIy"'(x). Common boundary conditions are of the
following types:
(a) Beam imbedded in a support: y = and y' — at the point;
(b) Beam simply supported: y = and y" = at the point;
(c) Beam free: v" = and ;'"' = at the point.
Problems in the transverse displacement of a beam take the form of the
differential equation (1) with boundary conditions at each .-^d of the beam.
218 Applications [Ch. 12

Such problems can be solved by integration with the use of a little algebra.
There are however two reasons for employing our transform method in such
problems. Frequently the load function, or its derivative, is discontinuous.
Beam problems also give us a chance to examine a useful device in which a
problem over a finite range is solved with the aid of an associated problem
over an infinite range.

FIGURE 29

EXAMPLE: Find the displacement ;• throughout the beam of Figure 29,


in which the load is assumed to decrease uniformly from Uq at x = to

zero at X = c and to remain zero from x = c to x— 2c. The weight of the


beam is to be negligible. The beam is imbedded at x = and free at x— 2c.

We are to solve the problem

£/—^ = ^[c-x + (.\--c)a(.\--c)], for 0<a-<2c; (2)


ax c

>-(0) = 0, >''(0) = 0; (3)

y"i2c) = 0, y"'(2c) = 0. (4)

The student should verify that the right member of (2) is the stipulated load
function
w,

c
for O^xS c, (5)

= 0, for c < X < 2c.


§ 70] The simple pendulum and the deflection of beams 219

To apply the transform technique, with x playing the role for which we
usually employ t, we need first to extend the range of x so it will run from
to 00. That is, instead of the problem (2), (3), (4), we shall solve the problem
consisting of

EI -j-^ = Hix), for < X < 00, (6)


dx'

and the conditions (3) and (4). In (6) the function H(x) is to be chosen by
us except that H(x) must agree with IV (x) over the range < a: < 2c. The
solution of the problem (6), (3), (4) will then be used only in the range
^ A" ^ 2c. Of the various choices for Hix), it seems simplest to use

H(x) = — lc-x + {x-


VVf,

c
c)oc{x - c)], for < X < 00. (7)

That is, in practice we ordinarily retain the equation (2) and merely extend
the range from <x< 2c to < a" < oo. The student must, however, keep
in mind that we cannot apply the Laplace operator to the function lV(x) of

(5), since that function is not defined over the entire range <x< oo. We
shall solve (6) and conclude that the solution is valid for (2) on the range
^X^ 2c, over which (2) and (6) are identical.

Let
L{EIy(x)} = uis);

u(s)^EI re-'''y(x)dx. (8)

To transform EIy^^\x) we need to use the values of EIy{x) and its first three
derivatives at x = 0. From (3) we know that

ElyiO) = 0, Ely'(0) = 0.

Put
Ely"(0) = A, EIy"'iO) = B. (9)

The constants A and B must be determined by using the conditions (4).

By our usual methods we obtain, for the H{x) of (7),

L{H(x)] - — L{c - X +
c
(x - c)a(x - c)}

— [---.+
Thus the differential equation (6) is transformed into

VV'o /c 1 c~^
^
s'^u(s) -s^-0-s^-0-s-A-B = —\
^ -J

3+ -T
c \s s^ s'^
220 Applications [Ch. 12

from which we get

^^ A vvo/cB e-«\ 1
"(s) = ^+-i+—c \s---5+ -^
s"* s s^ s /
• (10)

Now L"^{z/(5)} = Elyix). Hence

£/>'(x) = ^Ax' + iBx' + -^ [5c.x^ - A-^ + (X - c)^a(x - c)]. (11)

From (11) we obtain

Ely'ix) = Ax + ifix^ + ^
24c
[4cx^ - x^ + (x - 4)*a(x - c)], (12)

EIy"(x) = A + Bx + —6c [3cx^ - x^ + (x - c)3a(x - c)], (13)

£/v"'(x) =B+ —2c [2cx - x^ + (x - c)Mx - c)]. (14)

By differentiating both members of equation (14), we can see that the y of


(1 1) is a solution of (6) over the infinite range and, more important, a solution
of (2) over the range <x< 2c.

With the aid of equations (1 1) through (14) we can now determine A and B
to make the y satisfy appropriate conditions at 2c, whether the x — beam be
free, imbedded, or pin-supported there. In our example the beam is to be

free at x = 2c; the solution is to satisfy the conditions

v"(2c) = 0, v"'(2c) = 0. (4)

Using (13) and (14), and a Httle work, we find that (4) requires A = i\yoC^^
B = —^WqC. We are thus led to the solution

Ely(x) = -^w, c'x' - -fLwo ex' + ^ [5cx^ - x^ + (x - c)^a(x - c)], (15)

for ^X^ 2c.

The student should verify by differentiations and appropriate substitutions


that the y of (15) satisfies the original differential equation (2) and boundary
conditions (3) and (4).

From (15) we can obtain whatever information we wish. For example, at

X= ^c the bending moment is

£/>''(ic) = Wo C'li - 1 + i(-i + I + 0)] = ^H'o c\

Exercises
1. A clock has a 6-inch pendulum. The clock ticks once for each time that the
pendulum completes a swing, returning to its original position. How many times
does the clock tick in 30 seconds? ans. 38 times.
§ 70] The simple pendulum and the deflection of beams 221

2. A 6-inch pendulum is released from rest at an angle one-tenth of a radian from


the vertical. Using ^ = 32 (ft. per sec. per sec), describe the motion.
ANS. = 0.1 cos 8/ (radians).
3. For the pendulum of Ex. 2, find the maximum angular speed and its first time of
occurrence. ans. 0.8 (rad. per sec.) at 0.2 sec.
4. A 6-inch pendulum is started with a velocity of one radian per second, toward
the vertical, from a position one-tenth radian from the vertical. Describe the
motion. ans. 6 = iV cos 8/ — i sin 8/ (radians).
5. For Ex. 4, find to the nearest degree the maximum angular displacement from
the vertical. ans. 9°.

6. Interpret as a pendulum problem and solve:

—+ ^^6 = 0:^'=^, (A)

when / = 0, ^ = ^0 and co = — = ojo (B)


dt

ANS. ^ = ^0 COS /3r + /3- 'coo sin j8/ (radians).


7. Find the maximum angular displacement from the vertical for the pendulum
of Ex. 6. ANS. d^,,^ido^ + ^-^aJoy\
In Exs. 8-11, find the y that satisfies equation (1), page 217, with the given load
function W{x) and the given conditions at the ends of the beam. (See (a), (b), (c),

page 217.) Verify your solutions.

8. W(x) as in the example introduced on page 218; beam imbedded at both x =


and X = 2c.
ANS. EIy{x) =xioWQ c^x^ — -ioWo cx^

+ .!^ [5cx* -x' + (,x- cYocix - c)].


120c
9. Wix) =0, for < < a: he,

= Wo , for he <x < Ic,


= 0, for|c<x<2c;
beam imbedded at x = 0, free at x = 2c.
ANS. EIy{x) = hwoc^x^ — iwocx^
+ AwoKx — ic)*a(x — he) — (.X — |c)'*a(jc — |c)],
10. H^(jc) = H'o[l — oc(x — c)] (describe the load); beam to be imbedded at a: =
and pin-supported (simply supported) at x = 2c.
ANS. Elyix) = Awo e^x^ —rhwo ex^ +^Wo[x'* — {x — cYa.{x — c)].
= — {2c-x\
Wo
11. W{x) iorO<x<c,
c
= ^0, forc<A:<2c;
beam to be imbedded at x = and free at x = 2c.
ANS. EIy(x) =\iwo c^x^ — Awo ex^ +^^0 x*
+ Z^ [Sex"" -x' + (x- cYocix - c)].
120c
CHAPTER 13
Systems of Equations

71. The Laplace transform method

The Laplace operator can be used to transform a system of linear differ-

ential equations with constant coefficients into a system of algebraic equations.

EXAMPLE (a) : Solve the system of equations

x"it)-xit) + 5y'(t)=t, (1)

y'(0-4>(0-2x'(0=-2, (2)

with the initial conditions

.v(0) = 0, A-'(0) = 0, >'(0) = 0, /(0) = 0. (3)

Let L{x{t)} = u(s) and L{y(t)} = v{s). Then, application of the Laplace
operator transforms the problem into that of solving a pair of simultaneous
algebraic equations

222
§ 71] The Laplace transform method 223

{s^-\)ii(s) + 5sv(s) = -2, (4)


s

-2suis) + (s^ - 4)v(s) = --. (5)


5

We solve equations (4) and (5) to obtain

-25^ + 4
"''^ - s(s' + l)(s' + 4)-
('>

Seeking the inverse transforms of u and r, we first expand the right members
of (6) and (7) into partial fractions

.
u(s)
s
=
1
T
-\ — 5
^
4
, (8)

1 2s s
K^)-T-:^T-r
s s^ + I
+ :2-^-
s^ +
4' (9)

Since x{t) = L~^{u(s)} and y(t) = L~^{r(5)}, we get the desired result

A-(0 = -^ + 5 sin / - 2sin2/, (10)

y{t) = 1 -2cos/ + cos2/, (11)

which is easily verified by direct substitution into (1), (2), and (3).

The foregoing procedure is simple in concept, but of course its practical


use will depend on our ability to find the inverse transforms of u{s) and v{s).

On the other hand, the use of the transform theory can help us gain insight
into the general theory of systems of linear differential equations. We illus-

trate this idea in the following example.

EXAMPLE (b): Solve the system

^^=>' + /^(0, (12)

dy
= x + G(0. (13)
y^

Here we presume that the transforms of the functions x{t), y{t), F{t), and
G{t) all exist and are given by u{s), v{s), f(s), and g(s), respectively. We then
have
su{s) - fi = lis) +f(s), (14)

siis) - C2 = u(s) + gis), (15)


224 Systems of Equations [Ch. 13

where Cj and Cj represent the initial values of x{t) and y{t). We can rewrite
equations (14) and (15) in the form

su{s)-iis) = c,+f{s\ (16)

- u{s) + sv{s) = C2 + g{s). (17)

Using Cramer's rule, the solution of the algebraic system (16) and (17)
may be written
ci+f(s) -1 f(s) -1
=
C2 + g(s) S 9(s) s
u{s)
s'-l

v{s) =
§711 The Laplace transform method 225

s c,

v(s) =
226 Systems of Equations [Ch. 13

3. xV) - 2xit) - yV) - XO = 6e^',


2xV) - 3x(r) + yV) - 3y(t) = 6e'' 40) = ; 3, y(0) = 0.
ANS. xit) = (1 + 2t)e' + 2e^',
v(0 = (1 - t)e' - e^'.
4. x'Xt) + 2x{t) - /(/) = 2/ + 5,

xV) - x(t) + yV) + yiO = -2/ - 1 ; x(0) = 3, x'(0) = 0, ^(0) - -3.


ANS. x(t) = + 2 + e-^' + sint, t

y{t) = - — 3e-^' — cos


1 t t.

5. The equations of Example (a) of Section 71 with initial conditions x{0) =0,
jc'(0) = 0,j(0) = l,/(0)-0.

ANS. a:(0 = — — f sin + f sin 2/,


^ r

y{t) = + I cos — f cos 2t.


1 /

6. The equations of Example (a) of Section 71 with initial conditions x(0) = 9,

x'(0) = 2,>'(0)-l,/(0)-0.
ANS. x(t) = —t+ 15 cos — 5 sin — 6 cos 2/ + 4 sin 2r,
/ /

y(t) = + 2 cos + 6 sin — 2 cos 2/ — 3 sin 2/.


1 / ?

7. + yV)-y{t) = 0,
x'V)
2xV)- xit) + z'(/) - ziO - 0,
xV) + 3xit) + /(/) - 4y{t) + 3z(0 = 0;
x(0) = 0, x'(0) = y(0) - 0, 2(0) = 0.
1 ,

8. x'Xt) - x(t) + sj'co m,


y'V) - 4>'(r) - 2xV) = 0,
in which ^{t) = 6t, 0^t^2,
= 12, />2;
x(0) - 0, x'(0) = 0, >'(0) = 0, ^'(0) = 0.
ANS. x{t) = — 2(3r — 5 sin + sin 2t) /

+ 2[3(/ - 2) - 5 sin {t-2) + sin 2(/ - 2)]a(/ - 2),


XO = 3 — 4 cos + COS 2/ r

- [3 - 4 cos (t-2) + COS 2(/ - 2)]a(/ - 2).


9. Write the solution of the system of Example (b) in Section 71 in terms of
convolution integrals.
10. Use the results of Ex. 9 to obtain the solution of Example (c) in Section 71.

In Exs. 11 and 12, write the solution in terms of convolution integrals.

11. x'{t)-2yit) = F{t),


yV) + = Git); xiO) - 1, yiO) = 0.
xit)
12. 2x'it) + 3yit) = Fit),
y'it) + 2xit) = Git); xiO) = 2, ^0) = 1.
13. Consider the initial value problem
x'it) = ax + by + fit),
y'it) ^cx + dy + git); x(0) = Ci,yiO) = C2 ,

where a, b, c, d, and Ci, Ci are constants. Use an argument similar to that of


Example (b) of Section 71 to show that the solution, if it exists, should have the
form
xit) = Xcit) + Xpit),
yit) = ydt) + ypit),
where xdt) and ydt) depend on ci and Cz whereas Xpit) and ypit) depend on/(0
and git).
§ 72] The difTerential operator method 227

14. For the initial value problem of Ex. 13, state a reasonable definition for a first-

order homogeneous linear system with constant coefficients. a Now state


theorem that expresses the results of Ex. 13 for the solution of a nonhomo-
geneous system.
15. Consider the initial value problem

xXO + 2yit) = 0,
x"(t) + 2y'(t) + 2yit) = 2e';
x(S)) = l,x'(0)=0,y(0) = 0.
(a) Show that the Laplace transform method produces

X = 3 — 2e', y = e\
(b) Verify that these functions satisfy the differential equations but do not
satisfy the initial conditions.
(c) By elementary elimination, show that a solution of the system of diff'eren-
tial equations has the form

X = Ci — 2e\ y = e\
and thus the initial conditions given are not compatible with the system of
equations.

72. The differential operator method

In solving systems of simultaneous linear equations, the differential

operator notation is another convenient tool. Consider the system

y" -y + 5r' = A-, (1)

2y' - v" + Av = 2,

with independent variable x and dependent variables y and r. It is natural


to attack the system (1) by eliminating one dependent variable to obtain a
single equation in the other dependent variable. Using D— djdx, write the
system (1) in the form

{D^ -\)y + 5Dv = x, (2)

2Dy - {D^ - A)v = 2.

Then, elimination of one dependent variable is straightforward. We may,


for instance, operate upon the first equation with ID and upon the second
equation with {D^ — 1), and then subtract one from the other, obtaining

[10Z>^ + (i)^ -\){D^ - 4)]r = 2Dx - (D^ - 1)2,

or

(D* + 5D^ + 4)r = 4. (3)


228 Systems of Equations [Ch. 13

In a similar manner r may be eliminated; the resultant equation for y is

[(Z)2 - 1)(Z)2 - 4) + \OD^]y = (D^ - 4)x + 5D{2),

or
(Z)^ + 5/)' +A)y= -Ax. (4)

From equations (3) and (4) it follows at once that

r = 1 + «, cos .Y + fl2 sin x + a^ cos 2x + a^ sin 2x (5)

and
y— —x + b^ cos X + 6, sin x + 63 cos 2x + 64 sin 2a'. (6)

The a's and 6's have yet to be chosen to make (5) and (6) satisfy the original

equations, rather than just the equations (3) and (4), which resulted from the
original ones after certain eliminations were performed.
Combining (by substitution) the v of (5) and the y of (6) with the first

equation of the system (2) leads to the identity

X— 2/?! cos X — 2b2 sin — 563 cos 2x —


a" 5b^ sin 2a' (7)

— 5a, sin X + 5a2 cos x — \Q)a^ sin 2a + 10^4 cos 2a = x.

That (7) be an identity in a demands that

-2^1+ 5^2 = 0, (8)

-2b 2- = 0, 5^1

-5^3+ 10^4 = 0,
-5^>4- 10^3 = 0.

Relations between the a's and b\ equivalent to the relations (8) follow from
substitution of the r of (5) and the y of (6) into the second equation of the
system (2).

We conclude that a set of solutions of the system (2) is

r = 1 + a, cos A + ^2 sin a+ a^, cos 2a + a^. sin 2a, (9)

y——x+ |-«2 cos A — yfl'i sin A + 2^4 cos 2a — 2^3 sin 2a',

in which a^, Qj, o^, a 4. are arbitrary constants.


The equations (3) and (4) for v and ;' can be written with the aid of deter-
minants. From the system (2) above we may write at once

(D' - 1) 5D (D^-l) A
2D -{D^-A) 2D 2

which reduces to equation (3) above, if care is used in the interpretation of


the right-hand member. The determinant on the right is to be interpreted as

{D^-\){2)-2D{x),
§ 72] The differential operator method 229

not as the differential operator 2{D^ — 1) — x{2D).


Determinants are extremely useful in any treatment of the theory of
systems of linear equations. For many simple systems which arise in practice,
no such powerful tool is needed.
There are other techniques for treating a system such as (2). For example,
we may first obtain (3) and from it v as in equation (5). Next we wish to
find an equation giving y in terms of v\ that is, we seek to eliminate from the
system (2) those terms that involve derivatives o^ y. From (2) we obtain the
two equations
{2D^-2)y+\Q)Dv = 2x, (10)

2D^y-(D^ -AD)v = Q, (11)

the latter by operatyt? with D on each member of the second equation of


the system (2). From nO) and (11) it follows at once that

2y- Dh^-6Dv= -2x,


or
y= -x + ^Dh + 3Dv. (12)

The V of equation (5) may now be used in (12) to compute the y which was
given in the solution (9). In this method of solution there is no need to
obtain (4), (6), (7), or (8).

Both methods outlined here are available whenever the system is linear
with constant coefficients. Such systems appear naturally in the study of
electric networks which will be discussed in the next chapter.
Systems of nonlinear equations and systems of linear equations with
variable coefficients will not be studied in this book.

Exercises
In Exs. \-4, use the differential operator method.

1. Ex. 1, Section 71. 2. Ex. 2, Section 71.


3. Ex. 3, Section 71. 4. Ex. 4, Section 71.

In Exs. 5-17, use either of the two methods of this chapter.

5. v' -2v + 2w' = 2- 4e^\


2v' — 3v + 3w' — w — 0.

6. (3D + 2)v + iD-6)w = 5e'


(4Z) + 2)v + (D- S)w = 5e'

(D + 2)y + (Z) — 2)v = 2. ans. y = Ci^^^ + d cos 2x + Ca sin 2x,


V = —1 — SciC^" + C3 cos 2x — C2 sin 2x.
230 Systems of Equations [Ch. 13

8. D^y - (2D - l)v ^ 1,

i2D+l)y + iD^~4)v=0.
ANS.^ + Oie" + aze'" + 03 cos x + a4. sin x,
V 1

=4 + Qie" — 3a2 e~' + (a^ — 2a^) cos x + Q-Oz + o*) sin x.


y
9. (Z)^ - 3Z))>' - (Z) - 2)z = 14x + 7,
(Z)-3);^+ I>2= 1. ANS. = 2 + aie^ + 02^^^4-03 f-^^
>'

z =- 7 + 7x + 2aie* - 1^3 e- ^^
10. (/)' + I>^ - 1)« + (/)' + 2Z)^ + 3Z) + l)y = 3 - a: ,

(Z) — 1)m + (D + l)y = 3 — ans.jr. m = 2jr + Oi + fli cos x + 03 sin x,


y = + ai — 03 cos x Oi sin
jc -\- at.

11. (Z)^ + Oj' + 4(Z) - 1)«; = 4^^


(i) - \)y + (Z) + 9)y = 0; when x = 0, = 5, / = 0, = i
j' i;

ANS. y = 26" + 2e"* + e'^'' (cos x + 2 sin x),


y = ^e~' + e'^" sin x.
d^x dy
12. —j + x-2 — = 2t,
dt^ dt
dx dy
2 — ~x+—-ly =
^^ ^^
l. ANS. x = 2t-2-{^b,e' ^-bze-' ^-bie-^\
y=-t-l+bxe'-b2e-' -Ib^e-^'.
13. 2(D +l)y+(D-\)w = x+l,
{D + 3)>' + (Z) + l)w = 4a: + 14.
ANS. y = x-\-2 + Cie"'cos 2a: + Ca^'^sin 2x,
w = X + 6 + (c2 — Ci)e"* cos 2x — (ci + 02)^"' sin 2x.
14. (Z) + + (Z) - 4)y - 6 cos x,
!)>'

[d - !)>' + (D^ + 4)t; = -6 sin x.

ANS. = 3 cos X + 3 sin X + Oi + a2 cos 3x + a^ sin 3x,


>'

V = \ai — 5(^2 — 303) cos 3x — i(3o2 + 03) sin 3x.


15. 2Du + (D- l)v + (Z) + 2)w - 0,
(Z> + 2)u + (2D - 3)?; - (D - 6)w = 0,
2Z)i/ - (D + 3)y - ZJiv = 0. ANS. u = 3cie^'' + 3c2e-'' + 3c3e"^,
V =4cie^'' — Scze'" + c^e"'^,
w = — 4cie^* — 4c2 e'" — C3 e"''^.

16. D^y + (D- l)v = 0,


(2D-l)y + (D-l)w = 0,
(D + 3)y + (D- 4)v + 3h' = 0.
ANS. y = ai + 026" + 3036"^",
V = bze" — 02 xe" — 1603 e*",
w = —Qi + (62 — a2)e'' — 02 xe'' — 103 e"*".
17. Solve the problem of Section 72 by the method of Section 71.
CHAPTER 14
Electric Circuits and Networks

73. Circuits

The basic laws governing the flow of electric current in a circuit or a


network will be given here without derivation. The notation used is common
to most texts in electrical engineering and is:

t (sec. ) = time
Q (coulombs) = quantity of electricity; e.g., charge on a capacitor
/ (amperes) = current, time rate of flow of electricity
E = electromotive
(volts) force or voltage
R — resistance
(ohms)
L (henrys) = inductance
C (farads) = capacitance.
By the definition of Q and / it follows that

I(t) = Q'(t).

231
232 Electric Circuits and Networks [Ch. 14

The current at each point in a network may be determined by solving the


equations that resuh from applying Kirchhoff's laws:

(a) The sum of the currents into {or away from) any point is zero,

and
(b) Around any closed path the sum of the instantaneous voltage drops in a
specified direction is zero.

A circuit is treated as a network containing only one closed path. Figure 30


exhibits an '"RLC circuit" with some of the customary conventions for
indicating various elements.

m
mO)
"^C

FIGURE 30

For a circuit, Kirchhoff's current law (a) indicates merely that the current
is the same throughout. That law plays a larger role in networks, as we shall

see later.
To apply Kirchhoff's voltage law (b), it is necessary to know the contribu-
tions of each of the idealized elements in Figure 30. The voltage drop across
the resistance is RI, that across the inductance is Ll'it), and that across the
capacitor is C~^Q{t). The impressed electromotive force E{t) is contributing
a voltage rise.

Assume that at time t = the switch shown in Figure 30 is to be closed.


At = / there is no current flowing, 7(0) = 0, and if the capacitor is initially

without charge, ^(0) = 0. From Kirchhoff's law (b), we get the differential
equation

Lr(t) + RI(i) + C-'Q(t) = E(t), (1)

in which
/(O = QV)- (2)
§ 73] Circuits 233

Equations (1) and (2), with the initial conditions

7(0) = 0, Q{0) = 0, (3)

constitute the problem to be solved.


To retain the conventional symbol L for the number of henrys inductance
of the circuit, we shall in this section denote by L, the Laplace operator for
which L is used in all other sections of the book.
Let the transforms of /(/), Q{f), ^(0 t>e denoted by lower case letters:

L,{I{t)} = i(s), L,{Q(t)} = q(s), L,{E{t)} = e(s). (4)

The problem (1), (2), and (3) is then transformed into

L si(s) + Ri(s) + C " V(5) = eis), (5)

lis) = sq(s). (6)

From (5) and (6) we find either or both of i(s) and q(s),

se(s)
i(s) = —, 5lj!
(7)

q{s) = —, — ^-^
J . (8)

Then I(t) and Q(t) are obtained as inverse transforms from (7) and (8).

From (7), (8), or the differential equation

LQ"(t) + RQ'{t) + C-'Q(t) = E(t), (9)

it follows that the circuit problem is equivalent to a problem in damped


vibrations of a spring (Section 69). The resistance term RQ'{t) parallels the
damping term in vibration problems. The analogies between electrical and
mechanical systems are useful in practice.

H.'

FIGURE 31
234 Electric Circuits and Networks [Ch. 14

EXAMPLE: In the RL circuit with the schematic diagram shown in

Figure 31, let the switch be closed at / = 0. At some later time, t — 1q, the

direct current element, the constant E, is to be removed from the circuit,

which remains closed. Find the current for all / > 0.

The initial value problem to be solved is

Ll'{t) + Rl{t) = E{t); /(0) = 0, (10)

£(0 = £[l-a(?-fo)]- (11)

Let the transform of /(/) be i{s). We know the transform of E{t). Therefore

we obtain the transformed problem

sLi(s) + Ri(s) =- [1 - exp (-fos)], (12)


s

from which
£[1 -expC-fps)]
i(s) (13)
s{sL + R)
Now
= i(i
1

sisL + R) R\s s
'—I
+ RL-'J
RL~
so
E /I
i{s) = [1 -Q\p(-toS)'].
R\s s + RL
Therefore

E R R
I(t) =- I - a(f - t
o)
- exp -- f
j
+ exp -- (r - to)\cc{t - t^) (14)
^
R I

The student should verify (14) and show that it can be written

Rt
ForOSt^to, ^(0 = — exp
-^ 1
T (15)

For t>to, /(r) = /(/o) exp --(/-/o) (16)

74. Simple networks

Systems of differential equations occur naturally in the application of


KirchhofF's laws to electric networks. The use of the Laplace operator makes
the solution of such systems a straightforward, though often laborious,
process. For the general theory of networks the Laplace operator is a tool of
great value. We restrict ourselves to extremely simple situations which should
be sufficient to indicate the procedure commonly used.
§74] Simple networks 235

EXAMPLE (a): Determine the character of the current /,(/) in the net-
work having the schematic diagram in Figure 32, under the assumption that
when the switch is closed the currents are each zero.

E-=

FIGURE 32

In a network, we apply Kirchhoff 's laws, page 232, to obtain a system of


equations to determine the currents. Since there are three dependent variables,

A' A h^ we need three equations.


'

From the current law it follows that

h=h + h. (1)

Application of the voltage law to the circuit on the left in Figure 32 yields

RJ,+L,l',^E. (2)

Using the voltage law on the outside circuit, we get

RJ,+R,h + L,r, = E. (3)

Still another equation can be obtained from the circuit on the right in
Figure 32:

R,i, + L,r,-L,r^ = Q. (4)

Equation (4) also follows at once from equations (2) and (3); it may be used
instead of either (2) or (3).

We wish to obtain /^(O from the initial value problem consisting of


equations (1), (2), (3), and the conditions /i(0) = 0, /2(0) = 0, Ij,{Qi) = 0. One

of the three initial conditions is redundant because of equation (1).

Let L{I^{t)) = ii,{s) for each of A' = 1, 2, 3. Then use of the operator L
transforms our problem into the algebraic one of solving the equations

'i -h- '3 = 0> (5)


236 Electric Circuits and Networks [Ch. 14

/?,/i + SL2/2 =-' (6)


s

(7)
s

Since we desire only i^is), let us use determinants to write the solution
§ 74] Simple networks 237

The right member of equation (1 1) has a partial fractions expansion

h(s) = — +—+
5 s ai
+ --
S + ^2
, (12)

so
/i(r) = Aq + Ai exp(-fliO + ^2 exp(-a2 0-

EXAMPLE (b): For the network shown in Figure 33, set up the equations
for the determination of the currents /j, I2, I3, and the charge Q^ . Assume
that when the switch is closed all currents and charges are zero. Obtain the
transformed problem.

FIGURE 33

Using Kirchhoflf's laws we write the equations

(13)

R^Ii + L2 —r- = E sm cot, (14)


at

/?i/i +R^l3 + — Q3 = Esmwt;


1
(15)

and the definition of current as time rate of change of charge yields

(16)
'
dt

Our problem consists of the four equations (13) through (16) with the
initial conditions that

7,(0) = 0, IM = 0, ^3(0) = 0. (17)


238 Electric Circuits and Networks [Ch. 14

Let L{/fc(0} = I'kis), k = 1, 2, 3, and ^{^3(0} = qzis). Then the transformed


problem is the problem of solving the algebraic system

/"i- '2-/3=0, (18)

Eco
/?i/i+sL2/2=^——5, (19)
s + (a
1 £to
Rxh + «3 '3 + TT ^3 = -T—-T' (20)
C3 5 + OJ
h= sgi. (21)

1. For the RL circuit of Figure 3 1 , page 233, find the current / if the direct current
element E is not removed from the circuit.
ANS. I = ER-'[l -expi-RtL-')].
2. Solve Ex. 1 if the direct-current element is replaced by an alternating-current
element Ecos tot. For convenience, use the notation

in which Z is called the steady-state impedance of this circuit.


ANS. I = EZ - ^[ojL sin cot + Rcos cat — Rexp (—/?/£"')].
3. Solve Ex. 2, replacing ^cos cot with Esin tot.

4. Figure 34 shows an RC circuit with an alternating-current element inserted.


Assume that the switch is closed at ? = at which time Q =0 and / = 0. Use the
notation

Z' = R^ + (coC)-',

where Z is the steady-state impedance of this circuit. Find / for t > 0.


ANS. I = EZ -^[R sin cnt + {coC)-' cosw/ — (coC)-' e\p{-tR-'C-')].

FIGURE 34
§ 74] Simple networks 239

5. In Figure 34, replace the alternating-current element with a direct-current


element £'=50 volts and use R = \0 ohms, C= 4(10)"* farad. Assume that
when the switch is closed (at / ^ 0) the charge on the capacitor is 0.01 5 coulomb.
Find the initial current in the circuit and the current for / > 0.

ANS. /(O) = 1.25(amp), I(t) = 1.25 exp (-250/)(amp.).

6. In Figure 30, page 232, find /(/) if E(t) = 60 volts, /? - 40 ohms, C= 5(10)"'
farad, L = 0.02 henry. Assume /(O) = 0, Q{0) = 0.
ANS. / = 3000/ exp (- 1000/)(amp.).
7. In Ex. 6, find the maximum current. ans. /max = 3e"' (amp.).

In Exs. 8-11, use Figure 30, page 232, with £"(0 == ^sin cot and with the follow-
ing notations used to simplify the appearance of the formulas:

R
^=2Z'
240 Electric Circuits and Networks [Ch. 14

FIGURE 35

£-=

FIGURE 36

15. For the network in Figure 36, set up the equations for the determination of the
charge Q3 and the currents h, I2, h- Assume all four of those quantities to be
zero at time zero. Transform the problem into algebraic form.

E-=r

FIGURE 37
§ 74] Simple networks 241

16. For the network inFigure 37 set up the equations for the determination of the
currents. Assume currents to be zero at time zero. Find ii(s) =L{Ii(t)} and
all

discuss the character of hit) without explicitly finding the function.


CHAPTER JLo

The Existence and Uniqueness


of Solutions

75. Preliminary remarks

The methods of Chapter 2 are strictly dependent on certain special prop-

erties (variables separable, exactness, etc.), that may or may not be possessed
by an individual equation. It is intuitively plausible that no collection of
methods can be found that would permit the explicit solution, in the sense

of Chapter 2. of all first-order differential equations. We may seek solutions


in other forms, employing infinite series or other limiting processes; we may
resort to numerical approximations.
Confronted with this situation, a mathematician reacts by searching for
what is known as an existence theorem. He seeks to determine conditions
sufficient to insure the existence of a solution that has certain properties.

In Chapter 2 we stated such a theorem that we now wish to examine more


closely.

242
§ 76] An existence and uniqueness theorem 243

76. An existence and uniqueness theorem

Consider the equation of order one

^=f{^,yy
ax
(1)

Let T denote the rectangular region defined by

|x — XqI ^ a and \y — yQ\-^b,


a region with the point {xq , y^) at its center. Let the function/in equation (1)
and the function dfjdy be continuous at each point in T. Then there exists
an interval, |a' — Xq\ ^ /?, and a function (j){x) that has the following properties:

(a) y= (j)(x) is a solution of equation (1) on the interval \x — XqI ^ h;


(b) On the interval |a' — a'oI ^ h, ^(x) satisfies the inequality |0(a:) — >'oI ^ ^5
(c) ({)(xo) = yo;

(d) 0(a') is unique on the interval |,y — Xq\ ^ // in the sense that it is the
only function that has all of the properties (a), (b), and (c).

The interval |a- — a'oI ^ h may or may not need to be smaller than the
interval |a — Xq\ ^ b over which conditions were imposed on /and df/dy.
In rough language, the theorem states that if /(a-, y) is sufficiently well

behaved near the point (aq, Jo)' the differential equation (1) has a solution
that passes through the point (xq , yo), and that solution is unique near
(^0 yo)-
A proof of this fundamental theorem is presented in the next three sections.
In essence the proof involves showing that a certain sequence of functions
has a limit and that the limiting function is the desired solution. The sequence
considered will be defined as follows:

yo(x) = >'o

yi(x) = yo+ f fit,yo{t))dt,


''xo

y2(x)^yo + \ At,y,(t))dt,

yn(x) = yo +j f(t,y„-^{t))dt.

So that the proof may appear more reasonable, we first consider some
examples of the proof for special differential equations.
244 The Existence and Uniqueness of Solutions [Ch. 15

EXAMPLE (a) : Show that the sequence of functions defined in equations

(2) converges to a solution for the initial value problem

dy
§77] A Lipschitz condition 245

Clearly the limit of this sequence is


— — -, and this function is a solution

of (4).

Exercises
In each of the following exercises, determine the limit of the sequence defined in
(2) above. Verify that the function you obtain is a solution of the initial value
problem.

1. y' = x; Xo = 2, yo = 1.

2. / =y; Xo =0,yo =2.


3. / = 2y; Xo = 0, >'o = 1.

4. y' = x + y;xo ^O,yo = l.

77. A Lipschitz condition

We have assumed in the hypothesis of the foregoing existence theorem


that the function /and its derivative df/dy are continuous in the rectangle T.
Thus if {x, y^) and (x, y2) are points in T, the mean value theorem applies
to / as a function of y. Hence, there exists a number y* between yi and ^2
such that

fix, >',) -fix, yz) = —


dy
(x, y*)(yi - y2).

The assumption that dfjdy is continuous in T allows us to assert that dfjdy

is bounded there. That is, there exists a number A^ > such that

df
^K,
dy

for every point in T. Since (x, y*) is in T, it follows that

df
\fix,y,)-fix,y2)\ = (X, >'*) bl ->'2U
dy
and
l/(^,>',)-/(A-,y2)l^^|}'l->'2l, (1)

for every pair of points {x, y^) and {x, yj) in T.

The inequality (1) is called a " Lipschitz condition " for the function/. We
have shown that under the hypotheses of our existence theorem, the Lip-
schitz condition (1) holds for every pair of points (x, y^) and {x, jj) in T.

In the proof in Section 78 we shall actually use the Lipschitz condition


rather than the hypothesized continuity of dfjdy. Thus we could restate the
existence theorem in terms of condition (1) instead of assuming dfjdy is

continuous in T.
246 The Existence and Uniqueness of Solutions [Ch. 15

78. A proof of the existence theorem

One hypothesis of the existence theorem of Section 76 is that/is continuous


in the rectangle T. It follows that /must be bounded in T. Let A/ > be a
number such that \f{x, y)\ ^ M for every point in T. We now take h to be the
smaller of the two numbers a and bjM, and define the rectangle R to be the
set of points (jv, y) for which

\x-Xo\-^h and \y-yo\^b.


Clearly i? is a subset of T.
As indicated in Section 76, we now consider the sequence of functions

>'nW = }o+ f
fit,y„-i{t))dt (1)

and prove the following lemma:

Lemma 1 : If\x — Xo\-^h then

\yn(x)-yo\ ^b,

forn =1,2,3, ... .

The proof of this lemma will be accomplished by induction. First of all,

— Xq\ ^ h we have
if \x

\yi(x) -
§ 78] A proof of the existence theorem 247

Lemma 1 may be staled in a slightly different way: if \x — Xq\ ^ h, then


the points [x, y„{x)], a; = 0, 1, 2, ... , are in R. The Lipschitz condition of
Section 77 may now be used to deduce the following lemma.

Lemma 2: ff\x - Xq\ ^ h then

\fix, y„ix)) -fix, y„-^(x))\ S K\y„(x) - y„-y(x)\,


for « = 1, 2, 3, . . . .

We are now in a position to give an inductive proof of still another lemma.

Lemma 3 :
7/" |.v - XqI ^ h, then

MK"-'\x-Xo\" MK"-'h"
\yn(x) -y„- l(x)| ^ ^ ;
,

for n = \, 2, 3, . . . .

For the case « = 1, we have from the proof of Lemma 1,

\yiix)-yo\ ^M\x-xo\.
Assuming that

b'n-M - JV2WI ^ ,
(n — ,„"' , (2)
1)!

we now must show that

^ MK"-'\x-Xo\"
\yM - y„-i(x)\„ s
,
,
, ,
; .

We will prove this for the case a'o ^x^ Xq + h. From Lemma 2 we have

\yn(x)- y„-i(x)\ = U(t,yn-,{t))-f(t,y„-2(mdt


f

^ f\f(t,yn-i(t))-fit,y„-2(mdt

^Kf\y„_,(t)-y„_2it)\dt.

Using the hypothesis (2) we conclude that

\yn(x) - y„-i(x)\ s (t - Xo)"- dt,


_ '

or

\yn(x) - yn-l(x)\ S
ni
— \X - Xol". (3)
248 The Existence and Uniqueness of Solutions [Ch. 15

For the case Xq —


h ^ x ^ Xq, the same type of argument will yield the same
result. The proof of Lemma 3 is thus complete.
To utilize the results of Lemma 3 we now compare the two infinite series

iLyM-yn-Ml and J]
!^1!L^^
n=l /.= ! nl

The second of these series is an absolutely convergent series. Moreover, by


Lemma 3, the second series dominates the first series. Hence, by the Weier-
strass M test the series

S[>'„W->-„-iW] (4)
n= 1

converges absolutely and uniformly on the interval \x — Xq\ ^ /;. If we


consider the kth partial sum of the series (4)

- = - + [J2W - + ••
n=l
I lyn(x) y„-i(x-)] [>'i(x) jo(x)] yi(x)']

+ [>'fcW->'fc-iW],
we see that
t

E C>'«(^) - >'n-l(^")] = >'k(^)-


n= 1

That is, the statement that the series (4) converges absolutely and uniformly
is equivalent to the statement that the sequence y„(x) converges uniformly
on the interval
\x - Xo\ ^ h.

If we now define
4){x) = lim y„{x)

and recall from the definition of the sequence y„{x) that each >'„Cv) is con-
tinuous on \x — Xq\^ h, it follows (since the convergence is uniform) that
^(x) is also continuous and

(pix) = lim y„(x) = Vq + Hm \ f{t, y„-,(t)) dt.


n-»oc n->oc "
xo

Because of the continuity of/ and the uniform convergence of the sequence
y„(x), we may interchange the order of the two limiting processes to show
that (p(x) is a solution of the integral equation

0(A-) = >'o+ ffit,Ht))dt. (5)

It follows immediately upon differentiation of equation (5) that (f)(x) is a solu-


tion of the differential equation dyjdx =f(x, y) on the interval \x — Xo\ ^ h.

Furthermore, it is clear from equation (5) that ^(.Yq) = Jo •


§ 79] A proof of the uniqueness theorem 249

Finally, since we have shown in Lemma 1 that \y„(x) — yo\ ^ b for each n
and for \x — Xq\ ^ /?, it follows that the same inequality must hold for
({)(x) = lim y„(x). That is, if \x — Xq\ ^ h, then |0(x) - jol S b.
n-* 00

Thus we have completed the proof of parts (a), (b), and (c) of the existence
theorem of Section 76.

79. A proof of the uniqueness theorem

We must now show that the function (j){x) obtained in Section 78 is unique.
Suppose there is another function Y{x), such that dY/dx =f[x, Y(x)],
^(•^o) = >'o ' and Y(x) I
— Jo ^ I
b for \x — Xq\ ^ h. Then we may write

Y(x) = >'o + ff(t, 7(0) dt.

If we compare Y{x) to the functions of the sequence y„{x) of Section 78 we


see that

\Y(x)-y„{x)\S fu(t,Y{t))-f{t,y„.,(t))ldt (1)

We shall now show that as n -^ co the integral on the right side of (1)
approaches zero for \x - Xo\ ^ /?. It will then follow that Y(x) = \imy„{x),

so that finally Y{x) = (/)(a') on the interval \x — Xq\ ^ Ii.

For any a- on the interval \x - Xq] ^ h it is true that [x, Y(x)] and
[x, y„-i{x)] are in the rectangle R, hence the Lipschitz condition of Section 77
will allow us to change (1) into

\Y(x) - y„{x)\ ^K f\Y(t) - y„.,(t)\ dt. (2)

We now proceed by an inductive proof and limit our attention to values


of X greater than Xq. (A similar argument obtains the same result for
Xq — h -^ X -^ Xq .) For 77 = 1 , we have

\Y(x)-y,ix)\^K f\Y{t)-yo\dt

^ Kbix - xo).

We wish also to show that the assumption

in.)-.„_.w,.aii^
(n - 1)!
250 The Existence and Uniqueness of Solutions [Ch. 15

leads to the conclusion

„w_,„w,.^:^(ifM. (3)

This will complete an inductive argument for the relation (3). We have for
Xq ^ ^' ^ Xq + h^

\Y{x)-y„{x)\S (\f{U Y{t))-f{Uy„.,{t))\dt

^K f\Y{t)-y„_,it)\dt
-'xo

K"b r^

(n - 1) J.^
\)\

K"b
< {x - Xof,

thus completing the proof of relation (3).

For \x — Xo\ S h we have from the inequality (3),

|y(x)->'„(A)l^—^. (4)

As A? ->• 00 the expression on the right side of relation (4) approaches zero.
Hence it follows that for |a- — Xq\ ^ h, y„(x) -^ Y(x). Thus Y(x) must be the
same function (f){x) we obtained in Section 78. That is, the solution (j){x) is

unique.

80. Other existence theorems

The existence theorem we have proved in the preceding sections for a


first-order equation can be extended to equations of higher order. The
simplest such extension is to equations of second order that can be written
in the form
y"=f(x,y,y'). (1)

It is natural to expect the theorem to involve continuity requirements on the


function/and its partial derivatives. The theorem may be stated as follows:

Theorem: If the function f of equation (1) and its partial derivatives with

respect to y and y' are continuous functions in a region T defined by


\x-Xo\^a, \y-yo\^b, \y'-y'o\Sc,
§ 80] Other existence theorems 251

then there exists an interval \x — Xq\ -^ h and a unique function (j)(x) such that
(l)(x) is a solution of {\) for all x in the interval |x — XqI ^h, (J)(xq) = Jq , and
0'(^o) = y'o •

A proof of this theorem that is quite similar to the proof given in Sections
78 and 79 can be found in Ince.* The generahzation of the theorem to equa-
tions of higher order is direct.

* E. L. Ince, Ordinary Differential Equations (London: Longmans, Green and Co.,


1927), Chapter 3.
CHAPTER J-O

Nonlinear Equations

81. Preliminary remarks

The existence and uniqueness theorem of Chapter 15 made no distinction


between hnear and nonhnear differential equations. We know from our
study in the earlier chapters of this book, however, that the methods we have
found for actually determining solutions of a given equation often depend
on the equation being a linear one. For example, in Chapter 2 we found that
certain particular kinds of first-order nonhnear equations can be solved;
that is, if the equation is exact, separable, homogeneous, and so on. On
the other hand, if a first-order equation is linear, we have a method which
can produce all possible solutions of the differential equation.
The fact is, there is no general method for solving first-order nonlinear

differential equations, even if the existence of such solutions can be shown


by the theorems of Chapter 15. Indeed, the determination of such solutions is

often difficult if not impossible.


In this chapter we shall briefly discuss a few of the special diflRculties that

252
§ 82] Factoring the left member 253

arise with nonlinear equations and a few techniques that will find solutions

for certain particular types of equations.

82. Factoring the left member

To give illustration to the kind of complexity that may arise in nonlinear

situations,we consider first a relatively simple complication. For an equation


of the form
f{x,y,y') = Q, (1)

it may be possible to factor the left member. The problem of solving (1) is

then replaced by two or more problems of simpler type. The latter may be
capable of solution by the methods of Chapters 2 and 4.

Since y' will be raised to powers in the example and exercises, let us simplify
the printing and writing by a common device, using/? for y'

dy
dx

EXAMPLE: Solve the differential equation

xyp^ + (x + j)/J + 1 = 0. (2)

The left member of equation (2) is readily factored. Thus (2) leads to

{xp + \){yp + I) = 0,

from which it follows that either

yp+\=Q (3)

or

xp + 1 = 0. (4)

From equation (3) in the form

y dy -\- dx —
it follows that

/=-2(x-q). (5)

Equation (4) may be written

X dy -{- dx =
from which, for x 9^ 0,

^^
X
254 Nonlinear Equations [Ch. 16

SO

y= - \n\c2xl (6)

We say, and it is very rough language, that the solutions of (2) are (5) and
(6). Particular solutions may be made up from these solutions; they may be
drawn from (5) alone, from (6) alone, or conceivably pieced together by
using (5) in some intervals and (6) in others. At a point where a solution
from (5) is to be joined with a solution from (6), the slope must remain
continuous (see Ex. 21 below), so the piecing together must take place along
the line y= x. Note (see Ex. 24) that the second derivative, which does not
enter the differential equation, need not be continuous.
The existence of these three sets of particular solutions of (2), that is,

solutions from (5), from (6), or from (5) and (6), leads to an interesting
phenomenon in initial value problems. Consider the problem of finding a
solution of (2) such that the solution passes through the point ( — j, 2). If the
result is to be valid for the interval —1 < x < — ^, there are two answers,
which will be found in Ex. 25, following. If the result is to be valid for
— < X <^,
1 there is only one answer (Ex. 26), one of the two answers to
Ex. 25. If the result is to be valid in — <x<
1 2, there is only one answer
(Ex. 27).

Exercises
In Exs. 1-18, find the solutions in the sense of (5) and (6) above.

1. x^p^ —y^^O. y = dx, xy = Cz


ANS.
1. xp^ — {Ix + 'iy)p + 6>' = 0. ANS. = c,x^ y =2x + Ci
y
3. x^p^ — 5xyp + 6y^ ^ 0. ans. y = ClX^ y = C2X^
4. x^p^ + xp —y^ —y = 0. ans. y = CiX, xiy + I) ^ Cz
5. xp^ + (1 — x^y)p — xy = 0. ans. y ^ Ci exp (^x^), y = —In Ci x\ |

6. /?^ — (jc^ + 3);? + 3a:V == 0. ans. y = 7)X + Ci, x^ = 2i\n\c2y\


1. xp^ —{\ + xy)p + y = 0. ans. j'.-- hi |ciA-|, = ln|c2>'| j:

%. p^ -x^y^ =Q. ANS. x^ =l\n\ciy\,x^ = -l\n\c2y\


9. {x + yYp^ = y^. ans. x=y\n\ Ciy\ y(2x + y) = c2
,

10. yp^ + (x — y^)p — xy^Q. ans. x^ + y^ ^Ci^,y^ C2 e*


11. p^ — xy(x + y)p + x^y^ =^0. ans. y(x^ + Ci) = —2, x^ =3ln\c2y\
12. (Ax - y)p^ + 6(x - y)p + 2x-5y^0.
ans. x + y = Ci,(2x + yy=C2(j> — x).
13. {x — yYp^ =y^. ans. x = —yln\ciy\,y(2x — y) = C2
14. xyp^ + (xy^ — l)p — y = 0. ans. j^ = 2 In Cix\ x == —In C2y\ | , |

15. (x^ + y^Yp^ = 4x^y^. ans. y^ — x^ = Ciy, y(3x^ + y^) = C2


16. (y + xYp^ + C2y^ + xy - x')p + y(y-x)=0.
ans. y^ + 2xy = Ci, y^ + 2xy — x^ = C2
17. xy(x' + y')ip^ - 1) ^pix* + x'y^ + y^).
ans. y\y^ + 2x^) = Ci,y^=2x^ln\c2x\
§82] Factoring the left member 255

18. xp^ - (x^ + X + y)p^ + (x^ + xy + y)p -xy^O.


ANS. y = Cix,y = X + Cz, x^ =2(y — C3).

Exs. 19-27 refer to the example of this section. There the diflferential equation

xyp^ + (x + y)p+ 1=0 (2)

was shown to have the solutions

y' = -2(x - c.) (5)

and

y= -ln|c2x|. (6)

19. Show that of the family (5) above, the only curve that passes through the point
(1, 1) is >> = (3 — IxY'^ and that this solution is valid for x < f.

20. Show that of the family (6) above, the only curve that passes through the point
(1, 1) is >'= 1 — In X and that this solution is valid for < x.
21. Show that the function defined by

y = (2-xy" forx^I,
y = —In X for x ^ 1,

is a solution of equation (2) for all x 7^ 1, but fails to have a derivative at x = 1

and is therefore not a solution there.


22. Show that if a solution of (2) is to be pieced together from (5) and (6), then the
slopes of the curves must be equal where the pieces join. Show that the pieces
must therefore be joined at a point on the line y = x.
23. Show that the function determined by

>^ = (3-2x)"^ forx^l,


y = — In X
1 for 1 ^x
is a solution of equation (2) and is valid for all x. The interesting portion of this
curve is shown in Figure 38.

FIGURE 38
256 Nonlinear Equations [Ch. 16

24. Show for the solution given in Ex. 23 that /' is not continuous at a: = 1. Show
that as x-^ I ~,y" and as x-> 1 *,/'-> + 1.
-> —1,
25. Find those solutions of (2) which are valid in — 1 < jc < —i and each of which
has its graph passing through the point (— i, 2).
ANS. y = (3 -IxY'^; and j = 2 — In 2 - In (-x).
26. Find that solution of (2) which is valid for — 1 < x < ^ and has its graph passing
through the point ( — i, 2). ans. y ^(3 —IxY'^.
27. Find that solution of (2) which is valid for — 1 < x < 2 and has its graph passing
through the point ( — i, 2). ans. y = {3 —IxY'^ for x ^ I,
>> = 1 — In X for 1 ^ X.

83. Singular solutions

Let us solve the differential equation

y^p^ -a^ +y^ =0. (1)

Here

yp= ±s/a^-y\
so we may write

ydy
dx, (2)
Va^/
or

= dx, (3)
^a'-y'
or (if the division by ^Ja^ — y^ cannot be effected)

a^-y^= 0. (4)

From (2) it follows that

x = c,- ^/a' - y\ (5)

while from (3) that

X = c2 + VJ'a' - y\
^ (6)

and from (4) that

y = a OX y = —a. (7)

Graphically, the solutions (5) are left-hand semicircles with radius a and
centered on the x-axis; the solutions (6) are right-hand semicircles of radius
a, centered on the x-axis. We may combine (5) and (6) into

{x-cf-^y^ = a\ (8)
§84] The c-discriminant equation 257

which we might be tempted to call the "general" solution of (1). However,


from either of equations (7) we get/? = 0, so that y = a and y = -a arc both
solutions of equation (i), but neither of these functions is a special case of (8).
We therefore see that the use of the term "general" solution for the
functions defined implicitly by (8) is not consistent with the usage of the
term as applied to linear differential equations. For linear equations any
solution was a particular case of the general solution. It is perhaps unfor-
tunate that the words "general solution" are used for the one-parameter
family of solutions defined by (8). The particular solutions y = a and y = —a
are called singular solutions. (It should be clear that linear equations cannot
have singular solutions.)
A singular solution of a nonlinear first-order differential equation is any
solution that

(a) is not a special case of the general solution, and


(b) is, at each of its points, tangent to some element of the one-parameter
family that is the general solution.

Figure 39 shows several elements of the family of circles given by equation

FIGURE 39

(8) and also shows the two lines representing y = a and y = —a. At each
point of either line, the line is tangent to an element of the family of circles.

A curve which at each of its points is tangent to an element of a one-parameter


family of curves is called an envelope of that family.

84. The c-discriminant equation


Consider the differential equation of first order,

/(x, y, p) = 0; P = i-, (1)


ax
258 Nonlinear Equations [Ch. 16

in which the left member is a polynomial in x, y, and p. It may not be possible


to factor the left member into factors which are themselves polynomials in

X, y, and p. Then the equation is said to be irreducible.


The general solution of (1) will be a one-parameter family,

(l>(x,y,c) = 0. (2)

A singular solution, if it exists, for equation (1) must be an envelope of the


family (2). Each point on the envelope is a point of tangency of the envelope
with some element of the family (2) and is determined by the value of c that
identifies that element of the family. Then the envelope has parametric
equations, x = x(c) and y = y(c), with the c of equation (2) as the parameter.
The functions x(c) and y(c) are as yet unknown to us. But the x and y of the
point of contact must also satisfy equation (2), from which we get, by differ-
entiation with respect to c, the equation


d(b

dx
dx dc
+ ——+ —=
dy
dy dc
d(b dd)

dc
0. (3)

The slope of the envelope and the slope of the family element concerned
must be equal at the point of contact. That slope can be determined by
differentiating equation (2) with respect to x, keeping c constant. Thus it

follows that

— + —— =
d(b dy dd)
0. (4)
dx dy dx
Equations (3) and (4) both hold at the point of contact and from them it

follows that

dd)
^=
dc
0. (5)

We now have two equations, = and d(f)/dc = 0, which must be satisfied

by X, y, and c. These two equations may be taken as the desired parametric


equations. They contain any envelope which may exist for the original
family of curves, = 0. Fortunately, there is no need for us to put these

equations into the form x= x{c),y = y(c).


The equation that results from the elimination of c from the equations
(f)
— and d(l)ldc = is called the c-discriminant equation* of the family

= 0. It is a necessary and suflficient condition that the equation

0(x,>;,c) = O, (2)

considered as an equation in c, have at least two of its roots equal.

* The c-discriminant equation may contain a locus of cusps of the elements of the general

solution and a locus of nodes of those elements, as well as the envelope which aroused our
interest in it.
§ 85] The p-discriminant equation 259

There is nothing in our work to guarantee that the c-discriminant equation,


or any part of it, will be a solution of the differential equation. To get the
c-discriminant equation we need the general solution. During the process of
obtaining the general solution, we find also the singular solution, if there

is one.

85. The j9 -discriminant equation

Suppose that in the irreducible differential equation

f(x,y,p) = (1)

the polynomial/is of degree n in p. There will be n roots of equation (1), each


yielding a result of the form

P= 9ix,y). (2)

If at a point {xq , Jq) the equation (1) has, as an equation in p, all its roots
distinct, then near {xq yo) there ,
will be n distinct equations of the type of
equation (2). Near (xq , >'o) the right members of these n equations will be
single-valued and may satisfy the conditions of the existence theorem de-
scribed in Chapter 15. But if at {xq, Jq) equation (1) has at least two of its

roots equal, then at least two of the n equations like (2) will have right
members assuming the same value at (xq yo). For such equations there is ,

no region, no matter how small, surrounding (xq yo) in which the right ,

member is single-valued. Hence the existence theorem of Chapter 15 cannot


be applied when equation (1) has two or more equal roots as an equation
in p. Therefore we must give separate consideration to the locus of points

(x, y) for which (1) has at least two of its roots equal.

The condition that equation (1) have at least two equal roots as an equation
in p is that both /= and df/dp = 0. These two equations in the three
variables x, y, and p are parametric equations of a curve in the xj-plane
with p playing the role of parameter. The equation that results when p is

eliminated from the parametric equations /= and df/dp = is called the


p-discriminant equation.
If an envelope of the general solution of /= exists, it will be contained
in the p-discriminant equation. No proof is included here.* For us the p-
discriminant equation is useful in two ways. When a singular solution is

obtained in the natural course of solving an equation, the /j-discriminant

* For more detail on singular solutions and the discriminants see E. L. Ince, Ordinary
Differential Equations (London: Longmans, Green and Co., 1927), pp. 82-92.
260 Nonlinear Equations [Ch. 16

equation furnishes us with a check. If none of our methods of attack leads


to a general solution, then the p-discriminant equation offers functions which
may be particular (including singular) solutions of the differential equation.
Then the p-discriminant equation should be tested for possible solutions of
the differential equation. Such particular solutions make, of course, no con-
tribution toward finding the general solution.
The /7-discriminant equation may contain singular solutions, solutions that
are not singular,and functions that are not solutions at all.

Exercises
1. For the quadratic equation

with A, B, C functions of x and y, show that the p-discriminant obtained by


eliminating p from /= and Sfjdp = is the familiar equation

B^-4AC = 0.
2. For the cubic p^ + Ap + B = 0, show that the /^-discriminant equation is
4A^ + 21B^=0.
3. For the cubic p^ + Ap^ + B = 0, show that the /^-discriminant equation is
Bi4A^ + 27B)=0.
4. Set up the condition that the equation x^p^ + x^yp + 4 = have equal roots as a
quadratic in p. Compare with the singular solution xy^ = 16.
5. Show that the condition that the equation xyp^ + ix + y)p +1=0 of the
example of Section 82 have equal roots in /? is (x — yY = and that the latter
equation does not yield a solution of the differential equation. Was there a
singular solution?
6. For the equation y^p'^ — a^ + 7^ = of Section 83, find the condition for equal
roots inp and compare with the singular solution.
7. For the differential equation of Ex. 6 show that the function defined by

y-W
§ 86] Eliminating the dependent variable 261

12. p^-]-4x^p-l2x*y = 0. ans. (a)x\x^ + 3y) ^0;(h)3y = -x^.


13. 4y^p^ -4xp + y^0. ans. (a) (y^ - xXy"^ + x) = 0; (b) Same as (a).
14. 4y''p^ + 4xp + y = 0. See also Ex. 1 3. ans. (a) (y^^ - x)iy^ + a:) = 0;
(b) None.
15. p^ + xp^ - y =^ 0. ANS. (a)y(4x'' -27y)^0;ib)y = 0.
16. y'^p^ - 6xp + 2y = 0. ans. (a) y\y^ - 2x)(y^ + 2xy^ + 4x^) = 0;
(b) y = 0,y^= 2x.
17. For the differential equation of Ex. 4 above, the general solution will be found
to be cxy + 4x + c^ ^ 0. Find the condition that this quadratic equation in c
have equal roots. Compare that condition with the singular solution.

86. Eliminating the dependent variable

Suppose the equation

f(x,y,p) = 0; p = ^, (1)
ax

is of a form such that we can readily solve it for the dependent variable y
and write
y = g(x,p). (2)

We can differentiate equation (2) with respect to x and, since dy/dx = p,


get an equation

.(x.p,g=0 (3)

involving only x and p. If we can solve equation (3), we will have two equations
relating x, y, and p, namely, equation (2) and the solution of (3). These
together form parametric equations of the solution of (1) with p now
considered a parameter. Or, if p be eliminated between (2) and the solution
of (3), then a solution in the nonparametric form is obtained.

EXAMPLE: Solve the differential equation

xp^ - 3yp + 9x^ = 0, for x > 0. (4)

Rewrite (4) as

3y = xp + 9x^lp. (5)

Then differentiate both members of (5) with respect to x, using the fact that
dy/dx = p, thus getting

18x / 9x^ dp
3p =p+ + x--^ /,
p \ p^ / dx
262 Nonlinear Equations [Ch. 16

or

\ PV \

or
§ 87] Clairaut's equation 263

It is not difficult to show that equation (II) defines two solutions of the
differential equation. These solutions are not special cases of the general
solution (10). They are singular solutions. Equation (II) has for its graph
the envelope of the family of curves given by equation (10). The solutions
defined by (11) are also easily obtained from the /^-discriminant equation.

87. Clairaut's equation

Any differential equation of the form

y=px+f{p), (1)

where /(/?) contains neither x nor y explicitly, can be solved at once by the
method of Section 86. Equation (1) is called Clairaut's equation.

Let us differentiate both members of (1) with respect to x, thus getting

p=p^{x^f'{p)]p',

[^+/'(P)]^ = 0. (2)

Then either

dp

or

X +r(p) = 0. (4)

The solution of the differential equation (3) is, of course, p = c, where c

is an arbitrary constant. Returning to the differential equation (1), we can


now write its general solution as

y = cx +f(c), (5)

a result easily verified by direct substitution into the differential equation (I).

Note that (5) is the equation of a family of straight lines.


Now consider equation (4). Since /(/?) and f'{p) are known functions ofp,
equations (4) and (1) together constitute a set of parametric equations giving
X and y in terms of the parameter;?. Indeed, from equation (4) it follows that

^v = -f'(p), (6)

which, combined with equation (1), yields

y=fip)~pf'{p). (7)
264 Nonlinear Equations [Ch. 16

If f(p) is not a linear function of p and not a constant, it can be shown


(Exs. 1 and 2 below) that (6) and (7) are parametric equations of a nonlinear
solution of the differential equation (1). Since the general solution (5)

represents a straight line for each value of c, the solution (6) and (7) cannot
be a special case of (5); it is a singular solution.

EXAMPLE (a): Solve the differential equation

y=px+p\ (8)

Since (8) is a Clairaut equation, we can write its general solution

y = ex + c'^

at once.
Then using (6) and (7) we obtain the parametric equations

x=-3p\ y=-2p\ (9)

of the singular solutions. The parameter/? may be eliminated from equations

(9), yielding the form


27/ = -4x^ (10)

for the singular solutions. See Figure 40.

FIGURE 40
§ 87] Clairaut^s equation 265

EXAMPLE (b): Solve the differential equation

(jc^ - \)p^ - Ixyp +/ - 1 = 0. (11)

Rewrite (11) as

x^p^ - Ixyp + j'^ - 1 - p^ = 0.

Then it is clear that the equation is of the form

{y-xpf-\-p^ = () (12)

and so could be broken up into two equations, each of Clairaut's form.


Then the general solution of (11) is obtained by replacing/? everywhere in it
by an arbitrary constant c. That is,

{x^ - \)c^ - Ixyc +/ - 1 = (13)

is the general solution of (11). The solution (13) is composed of two families
of straight lines,

y = c,x^ yr+T? (14)

and

y = C2X-J\^c^. (15)

From the /^-discriminant equation for (11) we obtain at once the singular
solutions defined by

x^+/ = l. (16)

¥!ixercises

1. Let a be a parameter and prove that if /"(a) exists, then

x=-f'{a\ y=fioc)-ocf'ioc) (17)

is a solution of the equation y = px + f{p). Hint: use dx and dy


diflFerential

to get p in terms of a and then show that y —px —f(p) vanishes identically.
2. Prove that if/"(a) J^ 0, then (17) above is not a special case of the general solu-
tion y = cx+f(c).
Hint: show that the slope of the graph of one solution de-
pends upon x whereas the slope of the graph of the other does not depend
upon X.

In Exs. 3-30, find the general solution and also the singular solution, if it exists.

3. p^ + x^p — 2x^y = 0. ANS. c^+ cx^ =2>';sing. sol., 8>'= — a:*.

4. p^ + 4x^p — llx'^y = 0. ANS. 12j^= c(c + 4x^); sing, sol., 3>' —a:^.
5. 2xp^ — 6yp^ + X* ^0. ans. 2c^x^ = — 6c^y; sing, sol., 2y = x^.
1

6. p^ — xp + y = 0. ANS. y = cx — c^; sing, sol., x^ = Ay.


266 Nonlinear Equations [Ch. 16

7. y =px + kp^. ANS.= cx-\- kc^ sing, sol., x^ = —4ky.


y ;

8. xV^ + 3xp + 9y =0. ans. + c^) + c = 0; sing, sol., 4x^y = 1.


x^(y
9. xV^ + 2x^^/7 -4 = 0. ANS. x\l + cy) = c^.
10. xp^ — 2yp + 4x = 0. ANS. x^ = c(>' — c); sing, sol., = 2x and y = —2x.
>'

11. 3x'*^p^ — xp — y = 0. ANS. x^' = c(3cx — 1); sing, sol., Hjc^j' = —1.
12. A-/?^ + (x - + - = 0.
;^);p 1 :^'

ANS. xc^ + ix — y)c + — y = 0; sing,


1 sol., (jc + yy = Ax.
13. p{xp - + A:) +
>» a = 0.
ANS. c(xc — + ^) + a = 0; sing, sol., {y — kY = Aax.
;^

14. x*/7^ — 'ixp — 3^ = 0. ANS. 2)xy = c{xc^ — 3); sing, sol., 9x^y^ = 4.
15. J x^p^ — xp.
= ANS. xy = c{c^x — 1); sing, sol., llx^y^ = 4.
16. xp"" - lyp^ + \2x^ = 0. ANS. 20^7 = c*^^ + 12; sing, sol., 3y^ = ±%x^.
17. xp^ —yp^+\=Q. ANS. xc^ — >'c^ + = 0; sing, sol., Ay^ = 21x^.
1

18. y=px + p"JoTn j!:0,n7^ I.

/xV \""'
I y
ANS. >> = cj:+ c"; Sing, sol., (
-I =— I -I

19. p^ — xp — y = 0. ANS. 3x =^2p + cp'^'^ and 3y =p^ — cp^'^.


20. 2p^ + xp-2y = 0. ans. x = 2p(3p + c) and y =p\Ap + c).
21. 2;?^ + xp — 2;' = 0. ans. x = 4/? ln|/7c| and j =p^[l + 2 ln|/7c|],
22. p^ + 2xp —y = Q. ans. 4jc = —3p^ + cp~^ and 2;^ = — /7^ + c/j -
23. Axp^ -3yp + 3=0. ans. 2x = 3;?- ^ + c/?-'* and 3y = 9p-^ + 2c/?-
24. /?^ — x/? + 2>' = 0. ans. X =/7(c — 3/?) and 2>' =p^(c — 4/7).
25. 5p^ + 6xp — 2y = 0. ans. p\x +py ^c and 2>' = 6xp + 5p^.
26. 2xp^ + {2x-y)p+l -y=0. ans. /j^x = (1 +/7)-^ + ln|c(l +/?)! and
py ^l+il +/,)-! + 2 In |c(l +/7)|.
27. 5/7^ + 3x/7 - = 0.
>- ANS. /7^(x + 2py = c and y = 3xp + 5p^.
28. /7^ + 3x/? - = 0.
>' ANS. p\5x + 2py = c and j = 3x/7 + /7^
29. y = xp + x^p^. ANS. x^ = cp~*'^ — 2/?"^ and = x/> + x^p^. j'

30. 8>' = 3x^ + p\ ANS. (/? - 3x)^ =c(p-x) and 8j = 3x=' + p^.

88. Dependent variable missing

Consider a second-order equation,

f(x,y',y") = 0, (1)

which does not contain the dependent variable y explicitly. Let us put

y =p.
Then
. dp
dx
and equation (1) may be replaced by
§ 88] Dependent variable missing 267

an equation of order one in p. If we can find p from equation (2), then y can
be obtained from y' =p hy an integration.

EXAMPLE: Solve the equation

xy"-{y'?-y' = Q (3)

of Example (b), page 12.

Since y does not appear explicitly in the differential equation (3), put
y' = p. Then

~ dp
y
dx'

so equation (3) becomes

x —
^P -p'-p
dx
=n 3
0.

Separation of variables leads to

dp dx
p{p^ + 1)
~ T
or

dp p dp dx
'
p p^ + I X

from which

\n\p\ - ^\n(p' + I) + \n\cA = \n\x\ (4)

follows.
Equation (4) yields

Cip{p' + \)-"'=x, (5)

which we wish to solve for p. From (5) we conclude that

c, V= ^^(l + p'),

P =^ x^— X
C,
2-

But p = y', so we have

dy=+ ,-,
X dx
— •
(6)
268 Nonlinear Equations [Ch. 16

The solutions of (6) are

y-c2= +{c,^-x^Yi\
or

x^ + {y-c2f = c,\ (7)

Equation (7) is the desired general solution of the differential equation (3).

Note that in dividing by p early in the work we might have discarded the
solutions y = k{p = 0), where k is constant. But (7) can be put in the form

Czix^ +y^) + c^y^ 1=0, (8)

with new arbitrary constants c^ and c^. Then the choice C3 = 0, C4 = — \jk
yields the solution y = k.

89. Independent variable missing

A second-order equation
/(>',/, y') = o (1)

in which the independent variable x does not appear explicitly can be reduced
to a first-order equation in y and y' . Put

y'=P,
then

„ dp dy dp dp
'

dx dx dy dy
so equation (1) becomes

/(..P,p|)=0. (2)

We try to determine p in terms of j from equation (2) and then substitute


the result into y' = p.

EXAMPLE: Solve the equation

>'/' + (/)' + 1=0 (3)

of Ex. 8, page 14.

Since the independent variable does not appear explicitly in equation (3),

we put y' =p and obtain

. dp
y =p-r
dy
§ 89] Independent variable missing 269

as before. Then equation (3) becomes

yp^^ + p' + 1=0, (4)

in which the variables p and y are easily separated.


From (4) it follows that

Pdp dy
2 < ' ~ ^»
p" + 1
.

from which

iln(/;2 + l) + ln|j;| = ln|cJ,

so

p'+\=c,'y-\ (5)

We solve equation (5) for p and find that

P=± y
Therefore

dy
_ ,
{c,'-yy
dx ~ y
or

±y{c,^ -y^y'i^dy^dx.
Then
+ (ci2_/)i/2=^_C2.
from which we obtain the final result

{x-C2f+y^ = c,^.

Exercises
1. /' = xi/y. ANS. X ^Ci sin (y + Cz).
2. x^y" + (y'y — Ixy' = 0; when x = l,y =^5 and y' = —4.
ANS. J = ix^ + 3x - 3 + 9 hi (3 - x).
3. x'^y" + {y'Y — Ixy' = 0; when x = 2, y = 5 and y' = 2.
ANS. x^ = l{y — 3).
4. jj'" + (7')^ = 0. ANS. See Ex. 20, page 14.
5. >' V
+ {y'Y = 0. ANS. X= cij' - In C2 I A'l

6. + \)y" = {y'Y.
(j' ANS. y+\=c2 /'\
7. 2a/' + {y' y = 0. ANS. (y - CiY = 4a(x - Ci).
8. Do Ex. 7 by another method.
9. x/' =/ + .Jc'; when x = I, y = ^ and /= 1. ans. 24>' = x^ + 9x^ + 2.
270 Nonlinear Equations [Ch. 16

10. xy" 4- / + ^ = 0; when x = 2, y= — 1 and /=— i.

ANS. y= — ix^ + In (^x).


11. /' = 2X/)^- ANS. y^= 3ic2— X — Ciy).
12. yy" + {/y - (y'y = 0. ans. x =y - Ciln\c2y\.
13. y" + P^y = 0. Check your resuh by solving the equation in two ways.
14.yy" + iyy=0. ans. x = Ci+ y\n\c2y\.
15. /'cosx=/. ANS. y = C2 + Ci In(l — sinx).
16. y" = xiy'Y; when x ^ 2, y ^ Iv and y' ^ — i. ans. x = 2 cot y.
2 + X
17. /' = xiy'Y; when x = 0, >' = 1 and /= J. ans. >'= + ^ In
1 .

18. >'"=— when X = 3, = and


e"^*"; = >' >'' 1. ans. = ln(x — 2).
>'

19. /' = — f"^*'; when X = 3, = and =^ — 1. >' j^' ans. = In (4 — x).


;^

20. 2y" = sin 2y; when x = 0, y^ tt/I and y' = \.


ANS. X = — hi (csc y + cot y).
21. 2y" = sin 2^; when x = 0, y ^ —7t/2 and >'' =^ 1

ANS. X = ki — csc>' — C0t>').


(

22. Show you can perform the integrations encountered, then you can solve
that if

any equation of the form y" ^f(y).


23. xV xV
— = 3 — x^ ANS. y^x-^ + x + CiX^ + C2.
2A. y" ={y'y. ans. y = —\n\c2ici — x)\;ov x = Ci + CzC^
25. /' = e^iy'Y. ans. c^y + C2 = -In ce"' |
- 1|

26. 2y" = {y'Y sin 2x; when x = 0, j' = 1 and y' = \.

ANS. =\ -\-\n (sec x + tan x).


y
27. x^y" + {y'Y = 0. ans. Ci = CiX + In CzCc iX - 1)|
^ >'
|

28. y" = + {y'y.


\ ans. e' cos (x + Ci) == C2
29. Do Ex. 28 by another method.
30. /' = [1 + (y'yy^- Solve in three ways, by considering the geometric significance
of the equation, and by the methods of this chapter.
31. yy" = (y'yU — y' sin y — yy' cos y]. ans. x = Ci In C2 — cos y. |
>'i

21. {\+y^)y" + {y'y+y' =0. ans. = C2 + cj - (1 + c,^)lnb + Ci|. x


33. [yy" + + {y'yV = [1 + iy'yV.
1 ans. {y - c,y + (x - C2)' = c,^
34. xV =/(2x — /); when x = —1, = 5 and / = 1. ;^

ans. 27 - = (x - 2)^ + 8 In (x + 2). 1

35. X V = y'Ox - 2y'). ans. 2j = x^ + C2 - c, hi x^ + Ci| |


.

36. x/' = /(2 — 3x/). ANS. 3j = C2 + ln|x^ + Ci|.


37. x*y" =y'{y' + x^); when x =1, y = 2 and = 1. j''

38. /' = 2x + (x^ — /)^ ANS.


ANS.

3j'
>» = + X^ 1

= x^ + C2 — 3 ln|x + Ci|.
-m-
39. {y'y — 2y" + {y'y — 2xy' + x^ = 0; when x = 0,y =
and / == 1 i
2^ = 1 + x^ + 2
ANS. sin x.

40. {y")^ — xy" + y' =0. ans. General solution 2y = CiX^ — 2ci ^x + ci',:

family of singular solutions: \2y = x^ + k.


41. {y"y = \2y'{xy" — 2y'). ans. General solution y = Ci(x — cO^ + C2 :

family of singular solutions: 9>' = x* + k.


42. 3yy'y" - {y'Y - 1 ans. 27ci(>' + c,y = 8(x + CiY.
43. 4y{y'yy" = {y'Y + 3. ans. 256ci(>' - c,y = 243(x - C2)*.
§90] The catenary 271

90. The catenary

Let a cable of uniformly distributed weight h- (lb. per ft.) be suspended


between two supports at points A and B as indicated in Figure 41. The cable

FIGURE 41

will sag and there will be a lowest point V as indicated in the figure. We wish
to determine the curve formed by the suspended cable. That curve is called
the catenary.
Choose coordinate axes as shown in Figure 42, the j-axis vertical through
the point V and the x-axis horizontal and passing at a distance >'o (to be
chosen later) below V. Let s represent length (ft.) of the cable measured
from V to the variable point P with coordinates Then the portion of
(x, y).

the cable from K to P is subject to the three forces shown in Figure 42. Those
272 Nonlinear Equations [Ch. 16

forces are: (a) the gravitational force ws (lb.) acting downward through the
center of gravity of the portion of the cable from Kto P, (b) the tension T^ (lb.)
acting tangentially at P, and (c) the tension T2 (lb.) acting horizontally
(again tangentially) at V. The tension T^ is a variable; the tension T2 is

constant.
Since equilibrium is assumed, the algebraic sum of the vertical components
of these forces is zero and the algebraic sum of the horizontal components
of these forces is also zero. Therefore, if is the angle of inclination, from
the horizontal, of the tangent to the curve at the point (x, y), we have
Ti sin e - ns = (1)

and
Ti cos e-Tj^O. (2)

But tan 6 is the slope of the curve of the cable, so

tan 9 = ^. (3)
cix

We may eliminate the variable tension T^ from equations (1) and (2) and
obtain

ws
tanO = — (4)

The constant Tjl^v has the dimension of a length. Put T2iw = a (ft.). Then
equation (4) becomes

tan = - (5)
a

From equations (3) and (5) we see that

a ax

Now we know from calculus that since s is the length of arc of the curve,
then

d^y
From (6) we get
1

a dx
ds
= —
dx
j , so the elimination of s yields the differential

equation

^=y-er-
dx
§ 90] Miscellaneous exercises 273

The desired equation of the curve assumed by the suspended cable is that
solution of the differential equation (8) which also satisfies the initial con-
ditions

dy
when X = 0, y = yo and -— = 0. (9)
ax

Equation (8) fits into either of the types studied in this chapter. It is left

as an exercise for the student to solve the differential equation (8) with the
conditions (9) and arrive at the result

y = a cosh - + yQ — a. (10)
a

Then, of course, the sensible choice yQ — a is made, so the equation of the


desired curve (the catenary) is

y = a cosh -
a

Miscellaneous exercises
1. x^p^ + x^yp + 4=0. ANS. cxy + 4x + c^ xy^ = 16.
=0; sing, sol,
2. 6xp^ — (3x + 2y)p + j = 0. = CiX, 2y = x + C2.
ans. y^
3. 9p^ + 3xy*p + ^'^ = 0. ans. cy^{x — c) = sing, sol., x^y^ = 4.
1 ;

4. 4y^p^ — 4xp + = 0. >» ans. = 4c{x — c); sing, sol., y^ = x.


y'*^

5. x^p^ — 2xp —4y = 0. ans. x^(y — c^) = c; sing, sol., 4x> = — 1

6. 5p^ + 6xp — 2y = 0. ans. x =^ cp~^'^ — p and 2y = 6cp~^'^ — p^.


7. Do Ex. 6 by another method.
8. y^p^ —y(x+ l)p + X = 0. x^ — y^ = Ci, y^ =2(x — C2).
ans.
9. 4x'p^ + \2x^yp + 9 = 0. ans. x\2cy - 1) = c^ sing, sol., x^y"- = 1.
10. 4y^p^ —2xp + y = 0. ans. y^ = 2c(x — 2c^); sing, sol., 8x^ = 27^*.
U. p* + xp - 3y = 0.
12. Do Ex. 1 1 by another method.
ANS. 5x = 4p^ + cp"^ and 15>^ = + cp^""- V
13. xp^ + ik — X — y)p + J = 0, the equation of Ex. 5, page 14.
ANS. + — x—y)c + y = 0;
xc^ (Jc

sing, sol., (jc -yY - 2k(x + y) + k^ =0.


14. x^p^ - 2xyp^ -\-y^p+ 1^0.
ANS. x^c^ — 2xyc^ + y^c +1=0; sing, sol., 27a: = — 4y^.
15. \6xp^ + + >;6 ^ 0.
%yp ANS. y\c^x + 1) = 2c; sing, sol., xy* -=\.
16. x/7^ - (x^ + l)p + = 0. A- ANS. x^ = 2{y - Ci), = In C2 x| >- I

\1. p^ -2xp-y = Q. ANS. 8x = 3/?' + cp-2/3and4>'=/7^ -c/7"3.


18. Do Ex. 17 by another method.
19. 9xy'^p^ — 3y^p — 1=0. ans. cy^ = c^x — sing, sol., y^ = — 4x. 1 ;

20. x^p^ - {2xy + l)/7 + j^ + = 0. 1

ans. x^c^ — (2x>' + l)c + >'^ + = 0; sing, sol., 4x^ — 4x>' — 1=0.
1

21. x"/?^ = 8(2y + xp). ANS. c^x^ = 8(2x^>' - c); sing, sol., a:> = - 1.
22. x^p^ = (x — yy. ANS. x(x — 2y) = Ci,y= —x In C2 xl |
274 Nonlinear Equations [Ch. 16

23. x/7^ - 2yp^ + 4^2 = 0. See Ex. 10 above.


ANS. x^ = 4c(j> — 8c^); sing, sol., 8;^^ = 21 x*.
24. (p+iyiy-px) = l.

ANS. (c + 1)^(>' — cx) = l; sing, sol., 4(jc + yY ^ 21 x^.


25. p^ -p^ + xp-y = 0.
ANS. y = cx + c^ — c^\ sing. sol. with parametric equations,
and y =
X = 2a — 3a^
— 2a^. <x^

26. xp^ -^ y{\ — x)p —y^ —0. ans. xy = Ci,x = \n\c2y\.


27. >'/7^ - (a: + + = 0.
J')/? J'

ANS. py = cexp (/?"') andpx ^ yip^ —p+ 1); sing, sol., = >' jc.
Index
Abel's formula, 137 boundary conditions pertaining to, 217
Alpha function, 178-182 Bernoulli's equation, 69-71
Applications, 44-58, 201-221, 229-241, Bessel's equation, 130
271-273
beams, 217-220
catenary, 271-273 Cardioids:
chemical conversion, 48 differential equation of, 15
compound interest, 52 self-orthogonal family of, 56-57
cooling, Newton's law, 46—47 Catenary, 271-273
electric circuits, 231-234 c-discriminant equation, 258
electric networks, 234-241 Change of dependent variable, linear
mixture problems, 51-52 equation of order two, 127-129
orthogonal trajectories, 52-57 Change of form of solution, 18-20
pendulum, 216-217 Change of variable in equation of order
tractrix, 51-52 one, 25, 28, 68, 73-76

velocity of escape, 44--46


Chemical conversion, 48
Churchill, R. V., 105
vibration of spring, 201-213
Circles, differential equation of all, 14
Auxiliary equation, 98
Cissoids:
distinct roots, 97-99
differential equation of, 1
imaginary roots, 105-107
orthogonal trajectories of, 54, 58
repeated roots, 100-102
Clairaut's equation, 263-265
singular solution of, 264-265
Class A:
Beams, 217-220
functions of, 161-163
bending moment of, 217 transform, 161

277
278 Index

Coefficients linear in both variables, 73- Euler's theorem on homogeneous func-


75 29
tions,
Complementary function, 89 Exact equations, 29-33
Component: necessary and sufficient condition for,
forced. 205 31
natural, 205 Existence of solutions, 242-251
Compound interest, 52 Existence theorem, first-order equations,
Conies, central, differential equation of, 243
14 Exp notation, 21-22
Constant coefficients, linear equation Exponential function, Laplace transform
with, 97-147 of, 150
Convolution theorem, Laplace transform, Exponential function with imaginary ar-
186 gument, 103-105
Cooling, Newton's law of, 46-47 Exponential order, function of, 159-161,
Critical damping, 213 163
Current, electric, 231 Exponential shift, 94-96, 139-142
Current law, Kirchoff's, 232

Factoring in equations of order one, 253-


Damped vibrations, 211-213 254
Damping, critical, 213 Families of curves:
Damping factor, 204, 211 differential equation of, 9-13
Deflection of a beam, 217-220 orthogonal trajectories of, 52-57
Dependence, linear, 83 Forced component, 205
Dependent variable, defined, 2 Forcing function, 204
Dependent variable missing, equation Forms of solution, equation of order one,
with, 266-268 18-20
Differential equations, systems of, 222-
230, 234-241
Differential operators, 90-96 Gamma function, 168-169
exponential shift, 94-96, 139-142 General solution:
inverse, 143-147 homogeneous linear equation, 87-89
laws of operation, 93 nonhomogeneous linear equation, 89-
products of, 91-93 90
special properties of, 94—96 nonlinear equation, 257

Electric circuits, 23 1-234 Half-wave rectification, 170


Electric current, 231 of sine function, 173
Electric networks, 234-241 Homogeneous coefficients, equation with,
Eliminating the dependent variable, 261- 25-27, 67
263 Homogeneous functions, 23-24
Elimination of arbitrary constants, 5-8 defined, 24
Envelope, 257 degree of, 24
Equations, system of, 222-230 of degree zero, 24
Equations of order one, 16-18, 252-268 Euler's theorem on, 29
Equations of order two, nonlinear, 266- Homogeneous linear equation, 82
270 with constant coefficients, 97-109
Escape, velocity of, 44--46, 48-49 with variable coefficients, 82-90
Index 279
Hooke's law, 201 of derivatives, 163-166
Hutchinson, C. A., 147 of discontinuous function, 152, 170-
171, 178-182
of elementary functions, 150-153
Impedance, steady state: existence, 160-161
in RC circuit, 238 of functions of class A, 161-163
in RL circuit, 238 integral of, 173
in RLC circuit, 239 of integral, 186
Impressed force, 203-204 inverse of, 174
Ince, E. L., 251, 259 linearity, 149, 176
Inclined plane, motion on, 51 of periodic functions, 169-171
Independence, linear, 83-86 of power, 151-152, 169
Independent variable, 2 table of, 199-200
Independent variable missing, equation Legendre's equation, 130
with, 268-269 Lerch's theorem, 175
Initial value problems, 19-21, 38, 77-79, Linear coefficients, equation with, 73-75
118, 187-192, 201-221, 231-241, Linear dependence, 83
243-245 Linear equation, 4
systems of equations, 222-225 change of variable in, 127-129
Inspection solutions, linear equations, homogeneous with constant coefficients,
120-124 97-109
Integral equations, special, 194-197 homogeneous with variable coefficients,
Integral of transform, 173 82-89
Integral transform, 148 Laplace transform methods, 187-192
Integrating factor, 35-37, 59-66 nonhomogeneous with constant coeffi-
for equation with homogeneous coeffi- cients, 110-147
cients, 67 nonhomogeneous with variable coeffi-
for linear equation of order one, 36 cients, 82-90
Inverse differential operator, 143-146 operational methods, 139-146, 187-192
applied to exponential, 143-148 of order n, 82-90
applied to hyperbolic functions, 147 of order one, 34-38
applied to sine and cosine, 145-146 standard form, 34
Inverse Laplace transform, 174 systems of, 222-230
convolution theorem, 186 undetermined coefficients, 115-118
linearity of, 176 variation of parameters, 130-136
199-200
table of, Linear independence, 83-86
theorems on, 176, 180, 186 of polynomials, 86
Linearity:
of differential operators, 92
Kernel of an integral transform, 149 of inverse Laplace transform, 176
Kirchhoff's laws, 232 of Laplace transform, 149
Lipschitz condition, 245

Laplace operator, 149


Laplace transform, 149 Mixture problems, 51-52
application of, 187-197, 201-227
convolution theorem, 186
derivatives of, 166-167 Networks, electric, 234-241
280 Index

Newton's law of cooling, 46-47 RLC circuit, 232, 239


Nonelementary integrals, solutions in- Reduction of order:
volving, 77-79 linear equation, 127-129
Nonhomogeneous linear equation, 82 nonlinear equation, 226-269
change of variable 127-129
in, Resonance, 208-209
constant coefficients, 110-147, 187-192 Retarding force, 202
differential operator methods, 139-147
Laplace transform methods, 187-192
reduction of order, 127-129 Sectionally continuous functions, 157-159
undetermined coefficients, 1 15-118 Separation of variables, 17-21
variable coefficients, 82-90 Shearing force, beams, 217
variation of parameters, 130-136 Simple pendulum, 216-217
Nonlinear equation, 4 Singular solutions, 256-260
of order one, 16-33, 59-79, 242-245, Solution of a differential equation defined,
252-265 4
of order two or more, 266-269 Solution, existence of, 242-251
Null function, 175 Spring, vibration of, 201-216
Spring constant, 201
Square wave function, 171
Object function, 152
Steady-state impedance:
Operational methods:
RC circuit, 238
139-146
differential,
RL circuit, 238
transform, 187-192
RLC circuit, 239
Operator;
Strophoids:
inverse Laplace, 174
differential equation of, 15
Laplace, 149
orthogonal trajectories of, 58
Operators, differential, 90-96
Superposition of solutions, 123-124
inverse, 143-146
Systems of equations, 222-230
laws of operation, 93
differential operator method, 227-229
products of, 90-93
Laplace transform method, 222-225
special properties of, 94—96, 139-142
Order of a differential equation, 3
Order one, differential equations of, 16-81 Table, of Laplace transforms, 199-200
Ordinary differential equation, 3 Temperature determined by Newton's law
Orthogonal trajectories, 52-57 of cooling, 46-47
polar coordinates, 55-57 Tractrix, 51-52
rectangular coordinates, 52-54 Transform, integral, 148
Overdamped motion, 213 Transform, Laplace {see Laplace Trans-
form)
Transverse displacement of beam, 217-
Parameters, variation of, 130-136 220
Parametric solutions, 261, 263-265 Triangular wave function, 172
p-discriminant equation, 259-260 Trisectrices of Maclaurin:
Pendulum, simple, 216-217 differential equation of, 15
Periodic functions, transform of, 169-171 orthogonal trajectories of, 54, 58

RC circuit, 238 Undamped vibrations, 205-208


RL circuit, 238 resonance in, 208
Index 281

Undetermined coefficients, 115-118 critically damped, 213


damped, 211-213
forced, 203-204
Variable:
overdamped, 213
dependent, 2
resonance in, 208-209
independent, 2
undamped, 205-208
Variables, separation of, 17-21
Voltage drop, 232
Variation of parameters, 130-136
Voltage law, Kirchhoff's, 232
Velocity of escape, 44—46, 48-49
from Earth, 46
from Ganymede, Jupiter, Mars, Moon,
Sun, Venus, 49 Wronskian, 84-86
Vibration of an elastic spring, 201-216 Abel's formula for, 137
39783

You might also like