Jr. Frank Ayres-Theory and Problems of Differential Equations Including 560 Solved Problems-Schaum Publishing

of
DIFFERENTIAL
EQUATIONS
by FRANK AYRES, JR.
including
560
solve d
problems
Completely Solved in Detail
SCHAUM'5 OUTLINE SERIES

M~W-HILL BOOKC0l,l'ANY
SCHAUM'S OUTLINE OF
THEORY AND PROBLEMS

OF
DIFFERENr'rIAL EQUATIONS
BY
FRANK AYRES. JR .. Ph.D.

Professor and Head, Department of Mathematics
Dickinson College
SC::HAU~l'S OUTf,.fNE SElllES

l\kGT{A W-HILL BOOK COMPANY
N PW Yo~k, St. Louis. San Francisco, Toronto, Sydney
<'opyright © rn:i2 by l\kGraw-Hill. Ine. All HightH ReHPl'YP<l. Printed in tile
UnitPd Htate;; of Ameri1,a. /'\o part of this publi!'ation may be reproduced,
stored in a rPtrieval RYRtPm, or trarn,mittP<l, in auy form or by any means.
Pleetrouie. llJedianil'al. pllotot·opying, re<·or<ling. or otherwiNP, without tile prior
written )JPrmiHHiou of thP ))llbliHller.
026-'\--l
4 =i ti 7 S H O HHHH 7 2 1 0 6 H
Prefaee
This hook is designed primarily to supplement standard texts in elementary

differential equatiom. All types of ordinary and partial differential equations
found in current texts. together with the various procedures for solving
them. arc included. Since the lwginning student mnst be concerned largely
with mastering the methods of solving a variety of different type equations,
it is felt that there is need for a comprehensive prohle111 hook such as this.
It should prove also of equal service to practicing engineen and scientists who
feel the need for a review of the theory and prohlem work in this increasingly
important field.
Each chapter. except for the third which i1- entirely "xpository. begins with
a hrief l"-taternent of definitions. prirwipl1•s. and theoremr-. followed hy a set of
solved and supplementary problems. Tlwse solved problems have been selected
to make a careful study of eaeh ar,; rewarding as possible. Equal attention has
heen given to the chapters on application;.;. whi<'h include a wide variety of
problems from geometry and the physical s<'ienees.
Much more material is presented here than can he taken up in most first
courset'-. This is done not only to meet any ehoice of topics which the instructor
may make. hut also to stimulate further interest in the suhject and to provide
a handy book of reference. However. this book is definitely not a formal text•
book and. since there is always a tendency to "get on" with the problems, those
being introduced to the snhject for the first time are warned again;,;t m,ing it as
a means of avoiding a thorough study of the regular text.
The author is pltased to aeknowledge his indehted1ws~ to Mr. Louis Sand-

ler. associate editor of the publishers. for invaluable :--11ggestionf- and critical.
review of tlu• entire manuscript.
FRANK AYRES. JR.

Carlisle, Pa.
September, 1952
Contents
CHAPTER PAGE
]. ORIGIN OF DIFFERENTIAL EQUATIONS 1
2. SOLUTIONS OF DIFFERENTIAL EQUATIONS ..................................................................... 7
3. EQUATIONS OF FIRST ORDER AND FIRST DEGREE ······················· .. ············ ...- - 12
,i. EQUATIONS oF FrnsT ORDER AND FmsT DEGREE - - \ ARIABLEs SF.PA-

RABLE AND REDUCTION TO VARIABLES SEPARAJ3LE -·········--- JS
;). EQUATIONS OF FIRsT ORDER AND FIRST DEGREE~ ExACT EQUATIONS
AND REDUCTION TO EXACT EQUATIONS ·····•······························································· 24

6. EQUATIONS OF FmsT ORDER AND FmsT DEGREE - L11rnAR EQUATIONS
AND THOSE REDUCIBLE To THAT FoRM ........................................................................ 35
7. GEOMETRIC APPLICATIONS
8. PHYSICAL APPLICATIO'.'IS 49
9. EQllATIONS OF FmsT ORDER AND H1GHER DEGREE ................................................ 61
1 O. SINGULAR SoLUTIONS -~ ExTRANEous Loci ............................................................... 67
11. APPLICATIONS OF FIRST ORDER AND HIGHER DEGREE EQUATIONS............... 7.5
12. I,INEAR EQtiATIONS Of' ORDER n .......................................................................................... 78
13. HoMOGENEoLs LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS......... 82
14. l ,1NEAR EQlJ ATION s WITH CoN STANT CoEl'FICIEN TS ____ 87
1:-:i. LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS - VARIATION OF

p ARA::\'IETERS, lJ NDETF.RMINED COEFFICIENTS - - - 93
] 6. LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS - SHORT
METHODS·········------···························································-----" 99
l 7. LINEAR EQUATIONS, WITH VARIABLE COEFFICIENTS - THE CAUCHY
AND LEGENDRE LINEAR EQUATIONS ·····················-- ------- 108
18. LINEAR EQUATIONS WITH VARIABLE COEFFICIENTS - EQUATIONS OF

TffE SECOND ORDER · · - - -.... - ........................................................- - - - 111
CHAPTER PAGE
19. LINEAR EQUATIONS WITH VARIABLE COEFFICIENTS~ MISCELLANEOUS

TYPES ...................................................................................................................................................... 122
20. APPLICATIONS OF LINEAR EQUATIONS .............................................................................. I 3:1
2]. SYsTE\IS OF S11\lULTANEous LINEAR EQUATIO'.'ls ...................................................... I :;7
22. TOTAL DIFFERENTIAi. EQLATIONS ....................................................................................... 16 l
2.1. APPLICATIONS OF TOTAi. AND SIMlJI.TANEOlJS EQl 1ATIO"IS .............................. 178
24. NUMERICAL APPROXIMATIO'.'ls TO SoLllTIONs ............................................................ 186
2S. INTEGRATION IN SERIES 197
26. Il'JTEGRATION IN SERIES 206
27. THE LEGEc"IDRE. BESSEL. AND GAl ss EQllATIO"IS ................................................... 220
28. PARTIAL DIFFERENTIAL EQUATIONS .................................................................................... 2.11
29. LINEAR PARTIAL DIFFERENTIAL EQCATIONS OF ORDER ONE ........................ 238
30. NON-LINEAR PARTIAL DIFFERENTIAL EQUATIONS OF ORDER ONE ............... 2H
31. HOMOGENEOUS PARTIAL DIFFERENTIAi. EQUATIONS OF HIGIIER ORDER

"'ITH CONSTANT COEFFH:IENTS .......................................................................................... 2.s.;
32. No!'<-Ho,10GENEOLS LINEAR EQUATIONS WITH CoNSTA'\/T CoEFFI-

CIE'.'ITS 26S
3:3. PARTIAL DIFFERENTIAL EQLATIONS OF ORDER T,Yo WITH VARIABLE

CoEFFICIE N Ts 276
INDEX ..................................................................................................................................................................... 29S

CHAPTER l
Origin of Differential Equations
A DIFFERENfIAL EQUATION is an equation which involves derivatives. f'or example,
dy x2
1) X +5 5) (y"/ + (y I ) 3 + 3y
dx
d2y oz
2) + 3 dy + 2y 0 6)
clz
z + x-
dx 2 dx ox cly
3) xy'+ y = 3 2 2
cl z cl z
7) + -= x2 + Y.
,,, clx 2 oy 2
4) y + 2(y"/ + y' = COS X
If there is a single independent variable, as in l)-5), the derivatives are

ordinary derivatives and the equation is called an ordinary differential equa-
tion.
If there are two or more independent variables, as in 6)-7), the deriva-
tives are partial derivatives and the equation is called a partial di fferen-
t ial equation.
The order of a differential equation is the order of the highest derivative
which occurs. Equat1ons 1), 3), and 6) are of the first order; 2), 5), and 7)
are of the second order; and 4) is of the third order.
The degree of a differential equation which can be written as a polynomial
in the derivatives is the degree of the highest ordered derivative which then
occurs. All of the above examples are of the first degree except 5) which is
of the second degree.
A discussion of partial differential equations will be given in Chapter 28.
For the present, only ordinary differential equations with a single dependent
variable will be considered.
ORIGIN OF DIFFERENfIAL EQUATIONS.

a) Geometric Problems. See Problems 1 and 2 below.
b) Physical Problems. See Problems 3 and 4 below.
c) Primitives. A relation between the variables which involves n essential
arbitrary constants, as y = x 4 r Cx or y = Ax 2 + Bx, is called a primitive.
Then constants, always indicated by capital letters here, are called essen-
tial if they cannot be replaced by a smaller number of constants. See Prob-
lem 5.
In general, a primitive involving n essential c1rbitraryconstants will give
rise to a differential equation, of order n, free of arbitrary constants. This
equation is obtained by eliminating then constants between the (n +1) equa-
tions consisting of the primitive and the n equations obtained by differen-
tiating the primitive n times with respect to the independent variable. See
Problems 6-14 below.
2 ORIGIN OF DIFFERENTIAL EQUATIONS
SOLVED PROBLEMS
1. A curve is defined by the condition that at each of its points (x,y), y
its slope dy / dx is equal to twice the sum of the coordinates of the
point. Express the condition by means of a differential equation.
dy
The differential equation representing the condition is - = 2(x + y). X
dx
0
2. A curve is defined by the condition that the sum of the x- and y-intercepts of its tangents is
always equal to 2, Express the condition by means of a differential equation.
dy
The equation of the tangent at (x,y) on the curve is Y -y = -(X-x) and the x- and y-
• dx
dx dy
intercepts are respectively X = x -y - and Y = y-x - • The differential equation represent-
dy dx
dx ·dy dy 2 dy
ing the condition is X + Y =~x-y dy + y-x dx 2 or x(-) - (x + y - 2 ) - + y = 0.
dx dx
3. One hundred grams of cane sugar in water are being converted into dextrose at a rate which is
proportional to the amount unconverted, Find the differential equation expressing the rate of
conversion after t minutes.
Let q denote the number of grams converted in t minutes, Then ( 100 - q) is the number of grams
unconverted and the rate of conversion is given by dq = k (100 - q), k being the constant of
dt
proportionality.
4. A particle of mass m moves along a straight line (the x-axis) while subject to 1) a force pro-
portional to its displacement x from a fixed point O in its path and directed toward O and 2)
a resisting force proportional to its velocity, Express the total force as a differential equa-
tion.
dx
The first force may be represented by -k 1 x and the second by -k 2 - , where k 1 and k 2 are
dt
factors of proportionality,
The total force (mass x acceleration) is given by
5. In each of the equations a) y = x 2 +A+B, b) y =Aex+B, c) y =A+ lnBx showthatonly one

of the two arbitrary constants is essential.
a) Since A+ B is no more than a single arbitrary constant, only one essential arbitrary con-
stant is involved.
x+B x B B
b) y = Ae = Ae e , and Ae is no more than a single arbitrary constant.
c) y = A + ln Bx = A + ln B + ln x, and (A + ln B) is no more than a single constant.
6. Obtain the differential equation associated with the primitive 2

y = Ax +Bx+ C.
Since there are three arbitrary constants, we consider the four equations
2
2
y = Ax +Bx+ C,
dy
- = 2Ax +B ~ = 2A
dx ' dx 2 '
d3y
The last of these - , being free of arbitrary constants and of the proper order, is the
dx3
ORIGIN OF DIFFERENTIAL EQUATIONS
required equation.
Note that the constants could not have been eliminated between the first three of the above
equations. Note also that the primitive can be obtained readily from the differential equa-
tion by integration.
7. Obtain the differential equation associated with the primitive x y 3 x 3y 5

2
+ = C.
3 2 2 dy 2 5 3 11 dy
Differeritiating once with respect to x, we obtain ( 2xy + 3x y - ) + ( 3x y + 5x y - ) 0
dx dx
or, when xy 1 0, (2y + 3x dy) + x/ (3y + 5x dy) = O as the required equation.

dx dx
When written in differential notation, these equations are
(2xy3 dx- ..- 3x /dy) + (3x y 5 dx + 5x 3 /dy)
2 2
1) 0
and 2) (2ydx + 3xdy) + x/(3ydx + 5xdy) = O.

Note that the primitive can be obtained readily from 1) by integration but not so readily
from 2). Thus, to obtain the primitive when 2) is given, it is necessary to determine the fac- ·
tor xy 2 which was removed from 1).
8. Obtain the differential equation associated with the primitive y = A cos ax + B sin ax, A
and B being arbitrary constants, and a being a fixed constant.
dy
Here -Aa sin ax + Ba cos ax
dx
and
d\ -Aa 2 cos ax- - Ba 2 sin ax -a 2 (A cos ax + B sin ax) -a 2 y.
dx2
The required differential equation is o.
2X X
9• Obtain the differential equation associated with the primitive y = Ae +Be + C.
dy d2y d3y
Here 2Ae2x + B?, = 4Ae 2x + Bex, 8Ae 2 x + Bex.
dx dx2 dx3
2
d\ d2y d2y dy d\
Then - 4Ae 2x, - 2Ae2X, and d\ - 2(~ ~).
dx3 dx2 dx2 dx dx3 dx2 dx2 dx
2 dy
The r~quired equation is o.
dx
3X 2X X
10 • Obtain the differential equation associated with the primitive y = C1 e + C2 e + C3 e .
dy
Here
dx
The elimination of the constants by elementary methods is somewhat tedious. If three of

the equations are solved for C1 , C2 , ½, using determinants, and these substituted in the
fourth equation, the result may be put in the form (called the eliminant):
-I ORICIN OF J)JFFERENTIAL E()lATJONS
3x 2X X
e e e y 1 l 1 y
3e3X 2/X ·c
e y' 3 2 1 y'
6x 6X
e e (- 2y
Ill
+ 12y" - 22y' + 12y) o.
9e3X 4e2x e
X
y" 9 4 1 y"
27e 3X 8e2x X
e y "' 27 8 1 y "'
The required differential equation is 11 dy - 6y o.

dx
2 2
11. Obtain the differential equation associated with the primitive y Cx + C •
dy C = 1 dy y = Cx2 + C2 = _!_dy x2 + _l_(dY/.

Since dx = 2Cx, and
2x dx 2x dx 4x2 dx
dy 2 3 dy 2
The required differential equation is (- ) + 2x - - 4x y o.
dx dx
Note. The primitive involves one arbitrary constant of degree two and the resulting dif-
ferential equation is of order one and degree two.
12. Find the differential equation of the family of circles of fixed radius t with centers on the
x-axis.
The equation of the family is (x - c/ + y2 2

r '
C being an arbitrary constant.
dy X
Then (x -C) + dy 0, x-C = -y -• and the
y dx dx
2
differenti~l equation is = ,. •
13. Find the differential equation of the family of parabolas with foci at the origin and axes
along the x-axis.
y IY
,t X
(-A,0)
x2 + y 2 = (2A + x>2
2 'l
y = 4A (A+ x) y = 4A(A +x)
The equation of the family of parabolas is y 2 = 4A(A + x).

Then yy ' = 24 , A = 2yy' , and y2 = 2yy ' ( bY' + x) •
The required equation is

ORIGIN OF DIFFERENTIAL EQUATIONS ,)
2
14. Form the differential equation representing all tangents to the parabola y 2x.
At any point (A,B) on the parabola, the equation of the tangent is y- B (x -A )/B or,
since A ~ ½B 2 , By = x + ½B 2 • Eliminating B between this and By'= 1, obtained by differenti-
ation with respect to x, we have as the required differential equation 2x(y 1 ) 2 - 2yy' + 1 = 0.
SUPPLEMENTARY PROBLEMS
15. Classify each of the following equations as to order and degree.
a) dy + (xy - cos x)dx = 0 Ans. Order one; degree one
d2() q_
b) L - + R dQ + 0 Ans. Order two; degree one
2 dt C
dt
2
c) y"' + xy" + 2y(y') + xy = 0 Ans. Order three; degree one
2
d v dv
d) dv/ +
+ x(- V o· Ans. Order· two; degree one
dx2 dx. dx
3 2
e) d
(-- w/ d w
(--)
4
+ vw 0 Ans. Order three; degree two
2
dv 3 dv
f) ey "' Ans. Order three; degree does not apply

- xy" + y = 0
g) ~ = sine Ans, Order one; degree one
h) y' + X = (Y -xy,) -3 Ans. Order one; degree four
i)
d2p
2
=) p+ (-)
dp
d8
2 Ans. Order two; degree four
d8
16. Write the differential equation for each of the curves determined by the given conditions.
a) At each point (x,y) the slope of the tangent is equal to the square of the abscissa of the
point. Ans. y' = x 2
b) At each point (x,y) the length of the subtangent is equal to the sum of the coordinates of
thepoint. Ans. y/y'=x+y or (x+y)y'=y.
c) The segment joining P(x,y) and the point of intersection of the normal at P with the x-axis
dx I
is bisected by the y-axis. Ans . y + x- = 2y or yy' + 2x = 0
dy
d) At each point (p,8) the tangent of the angle between the radius vector and the tangent is
d8 1
equal to 1/3 the tangent of the vectorial angle. Ans. p -
dp
= -
3
tan e
e) The area bounded by the arc of a curve, the x-axis, and two ordinates, one fixed ans one
variable, is equal to twice the length of the arc between the ordinates.
Ans . y = 2 /2 + (y 1
/
6 ORIGIN OF DIFFERENTIAL EQCATIONS
17, Express each of the following physical statements in differential equation form.
a) Radium decomposes at a rate proportional to the amount Q present. Ans. dQ/dt = -kQ
b) The population P of a city increases at a rate proportional to the population and to the
difference between 200,000 and the population, Ans. dP/dt = kP(200,000 - P)
c) For a certain substance the rate of change of vapor pressure {P) with respect to temper-
ature (T) is proportional to the vapor pressure and inversely proportional to the square
of the temperature, Ans. dP/dT = kP/T 2
d) The potential difference E across an element of inductance L is equal to the product of L

and the time rate of change of the current i in the inductance, Ans. E = L di
dt
dv
e) Mass x acceleration= net force. Ans. m- = F
dt
18, Obtain the differential equation associated with the given primitive, A and B being arbitrary
constants.
2
a) y = Ax Ans. y' = y/x e) y = sin(x+A) Ans. (y' / 1 - y
X
b) y = Ax +B Ans. y" 0 f) y = Ae +B Ans, y" y'
x +A
c) y =e = Bex Ans. y' y g) X =A sin(y+B) Ans. y" = x(y')3
2
d) y = A sin x Ans. y' y cot x h) ln y = Ax +B
Ans. xyy" - yy r - x (Y, / 0
19. Find the differential equation of the family of circles of variable radii r with centers on
the x-axis, (Compare with Problem 12.)
2
Hint: (x-A/ + y 2 = r 2 , A and r being arbitrary constants. Ans. yy" + (y 1 ) + 1 =o
20, Find the differential equation of the family of cardiods P· = a (1 - cos 8).
Ans. (1 - cos8)dp = p sin 8 d8
21, Find the differential equation of all straight lines at a unit distance from the origin.
Ans. (xy'-y/=l+(y 1 /
22, Find the differential equation of all circles in the plane.

2
Hint: Use x +/-2Ax-2By+C = Q. Ar,s. [l+(y 1 /)y"'-3y 1 (y"/ 0
CHAPTER 2
Solutions of Differential Equations
THE PROBLEM in elementary differential equations is essentially that of recovering

the primitive which gave rise to the equation. In other words, the problem of
solving a differential equation of order n is essentially that of finding a
relation between the variables involving n independent arbitrary constants
which together with the derivatives obtained from it satisfy the differential
equation. For example:
Differential Equation Primitive

d3y
1) = 0 y =Ax 2 +Bx+ C (Prob. 6, Chap. 1)
dx 3
d3y 2 dy
2) 6d y + 11- - 6y 0 y = C1 e 3x + C2e 2x + C3 e x (Prob.10, Chap. 1)
cJx3 dx.2 dx
3) y2(dy/ + y2 r2 (x - c/ + y 2 = r 2 (Prob. 12, Chap. 1)

dx
THE CONDITIONS under which we can be assured that a differential equation is solv-
able are given by Existence Theorems.
For example, a differential equation of the form y' = g(x,y) for which
a) g(x,y) is continuous and single valued over a region R of points (x,y),
b) dg exists and is continuous at all points in R,
dY
admits an infinity of solutions f(x, y,C) = 0 (C, an arbitrary constant) such
that through each point of R there passes one and only one curve of the fam-
ily f(x ,y ,C) = 0. See Problem 5.
A PARTICULAR SOLUTION of a differential equation is one obtained from the primitive

by assigning definite values to the arbitrary constants. For example, in 1)
above y=O (A=B=C=O), y=2x+5 (A=O, B=2, C=5), and y=x 2 +2x+3 (A
= 1. B = 2, C = 3) are particular solutions.
Geometrically, the primitive is the equation of a family of curves arid a
particular solution is the equation of some one of the curves. These curves
are called integral curves of the differential equation.
As will be seen from Problem 6, a given form of the primitive may not in-
clude all of the particular solutions. Moreover, as will be seen from Prob-
lem 7, a differential equation may have solutions which cannot be obtained from
the primitive by any manipulation of the arbitrary constant as in Problem 6.
Such solutions, called singular solutions, will be considered in Chapter 10.
The primitive of a differential equation is usually called the general so-
lution of the equation. Certain authors, because of the remarks in the para-
graph above, call it a general solution of the equation.
7
8 SOLUTIONS OF DIFFERENTIAL EQUATIONS
A DIFFERENTIAL EQUATION dy = g(x,y) associates with each point (x 0 ,y 0 ) in the re-

dx
gion R of the above existence theorem a direction m = ddxyl ( ) = g(x0 , y0 ) .
xo,Yo
The direction at each such point is that of the tangent to the curve of the
family f(x,y,C) =O, that is, the primitive, passing through the point.
The region R with the direction at each y
of its points indicated is called a direc-
t ion field. In the adjoining figure, a num-
ber of points with the direction at each is
shown for the equation dy/dx = 2x. The in-
tegral curves of the differential equation
are those curves having at each of their
points the direction given by the equation.
In this example, the integral curves are
parabolas.
X
Such diagrams are helpful in that they
aid in clarifying the relation between a
differential equation and its primitive, but
since the integral curves are generally 'I <\ p p
slopec -4 / /' ) \ "'slope 4
quite complex, such a diagram does not aid
0
slope•-2 / slope• 2
materially in obtaining their equations. slope O O
SOLVED PROBLEMS
1. Show by direct substitution in the differential equation and a check of the arbitrary constants
that each primitive gives rise to the corresponding differential equation.
d2y dy
a) y (1 - x cot x) - - x dx + y 0
dx2
dy X
b) y dx + y Be
a) Substitute y = C1 sin x + C2 x, - C1 sin x in the differential equa-

tion to obtain
The order of the differential equation (2) and the number of arbitrary constants (2) agree.
b) y
X X -x 2x2 X
C1e + C2 xe + C3 e + e '
X X -x 2x2 ex 4xe
X
Y' (C1 + C 2 )e + C2 xe C3 e + +
'
y" (C1 + 2C 2 )e
X
C2 xe
X
+ C3 e
-x + 2x2 ex + Bxe
X
+ 4ex,
+
y "'
X X -x 2x2 ex + 12xex
(C1 + 3C2 )e + C2 xe C3 e + + 12ex,
and y"' - y" - y' + y = Bex, The order of the differential equation and the number of arbitrary
constants agree.
2. Show that y = 2x + Ce x is the primitive of the differential equation dy - y = 2( 1 - x) and find

dx
the particular solution satisfied byx=0,y=3 (i.e., the equationoftheintegralcurvethrough
(0, 3).
SOLUTIONS OF DIFFERENTIAL EQCATIONS 9
Substitute y 2x + Cex and dy = 2 + Ce x in the differential equation to obtain 2 + Ce x_ ( 2x+ Ce°'')

=
dx
2-2x. When x= 0, y =3, 3 = 2•0+Ce 0 and C= 3, The particular solution is y = 2x+ 3ex.
2
d y dy
is the primitive of the differential equation - - 3- + 2y
dx2 dx
2x - 3 and find the equation of the integral curve through the points (0, 0) and ( 1, 0).
-dy = C1 e x + 2C2 e 2x + 1,
dx
ferential equation to obtain 2x- 3.
X 2x
e - e
and the required equation is y = x +
2 2
e - e e - e
4. Show that (y-C/ = Cx is the primitive of the differential equation 4x(dy/ + 2xdy - y 0
dx dx
and find the equations of the integral curves through the point (1,2),
dy dy C
Here 2(y -C) - = C and
dx dx 2(y- C)
c2 C
2
C x+Cx(y-C)-y(y-C)
2
y [Cx - (y -C/]
Then 4x--- + 2x--- - y o.
4(y-C)2 2(y - C) (y-C)2 (y -C)2
2
Whenx=l, Y"'2: ( 2 - C) = C and C = 1, 4.
2 2
The equations of the integral curves through (1,2) are (Y - 1) x and (y - 4) 4x.
5. The primitive of the differential equation dy

dx
= r
X
is y = Cx. Find the equation of the in-
tegral curve through a) (1,2) and b) (0,0).
a) When x = 1, y = 2: C = 2 and the required equation is y = 2x,

b) When x = 0, y = o: C is not determined, that is, all of the integral curves pass through the
origin, Note that g(x,y) = y/x is not continuous at the origin andhence the existence theorem
assures one and only one curve of the family y = Cx through each point of the plane except the
origin.
6. Differentiating xy = C(x - 1) (Y - 1) and substituting for C, we obtain the differential e'qua-

tion
dy
C{(x - 1 ) - + y-1} xy { (x - 1/Y + y - 1}
dx (x-l)(y-1) dx
dy
or 1) x(x-l)dx + y(y-1) = 0,
Now bothy= 0 and y = 1 are solutions of 1), since, for each, dy/dx = 0 and 1) is satistied,
The first is obtained from the primitive by setting C = 0, but the second y = 1 cannot be obtained
by assigning a finite value to C. Similarly, 1) may be obtained from the primitive in the form
Bxy = (x-l){y-1). Now the solution y=l is obtained by settingB=0 while the solution y=0
cannot be obtained by assigning a finite value to B, Thus, the given form of a primitive may
not include all of the particular solutions of the differential equation. (Note that x = 1 is
also a particular solution.)
SOLUTIONS OF DIFFERENTIAL EQUATIONS
7. Differentiating y = Cx + 2C
2
, solving for C = dy
dx. and substituting in the primitive yields the
differential equation
dy 2 dy
1) 2(--'--)
dx
+ x(-) - y
dx
o.
Since y - ix satisfies 1), x + By

2
= O is a solution of 1).
Now the primitive is represented by a family of straight lines and it is clear that the
equation of a parabola cannot be obtained by manipulating the arbitrary constant. Such a solu-
tion is called a singular solution of the differential equation.
8. Verify and reconcile the fact that y C1 cos x + C2 sin x and y = A cos(x + B) are primitives
d2y
of + y = o.
dx 2
From y = C1 cos x + C2 sin x, y 1 = -C 1 sin x + C2 cos x and

y"= -C 1 cosx - C2 sinx = -y or o.
From y = A cos(x+B), y 1 = -A sin(x+B) and y"= -A cos(x+B) = -y.

Now y = A cos (x +B) A(cos x cosB - sin x sinB)
(A cos B) cos x + (-A sin B) sin x C1 cos x + C2 sin x.
2
9. Show that ln x 2 + ln L A+ x may be written as y2
x2
2
2 2 A+x A x
ln x + ln L ln / A+ x. Then y e e •e Bex.
x2
10. Show that Arc sin x Arc sin y = A may be written as x ✓i:-=? - y/27 B.
sin(Arc sin x - Arc sin y) = sin A= B.
Then sin(Arc sin x) cos( Arc sin y) - cos(Arc sin x) sin( Arc sin y) = x/1-y 2 -y/1-x 2 = B.
11. Show that ln(l+y) + ln(l+x) = A may be written as xy+x+y =·c.
ln(l+y) + ln(l+x) = ln(l+y){l+x) = A.

A
Then ( 1 + y) ( 1 + x) = xy + x + y + 1 = e =B and xy + x + y = B - 1 = C.
12. Show that sinh y + cosh y = Cx may be written as y = ln x + A,
Here sinhy + coshy = ½(ey-e-y) + ½(ey+e-y) = ey Cx.
Then y = ln C + ln x = A + In x.
SOLUTIONS OF DIFFERENTIAL EQUATIONS ]l
Show that each of the following expressions is a solution of the corresponding differential
equation. Classify each as a particular solution or general solution (primitive).
13, y = 2x2. xy' = 2y. Particular solution

2 2
14, X + y = C, yy' + X o. Primitive
'I
15, y = Cx + c\ y = xy' + (YI) • Primitive
16, (1- x)y

2
= X3 . 2x'y' = Y(Y
2
+ 3x )
2
. Particular solution
X
17, y =e (1 + x), y'' - 2y' + y = o. Particular solution
18, y = C1 x + C2 e
X
. (x- l)y 11 - xy 1 + y o. General solution
19, y = C1 e
X
+ C2 e
-x
. y" - y = o. General solution
-x
20. y = C1 e X + C2 e + X - 4, y" - y =4 - x. General solution
21. y = C1 e X + C2 e
2x
. y" - 3y' + 2y o. General solution
22. y = C1 e X + C2e
2x
+ X
2 X
e . y II _ 3y I + 2y 2ex (1- x). General solution
CHAPTER 3
Equations of First Order and First Degree
A DIFFERENTIAL EQUATION of the first order and first degree may be written in the form
1) M(x,y)dx + N(x,y)dy = 0.
EXAMPLE 1. a) dy + Y +x = O may be written as (y +x)dx + (y-x)dy = O in

cJx Y-X
which M(x,y) =y+x and N(x,y)= y-x.
2 2
b) dy = 1 + x y may be written as (1 + x y)dx -dy =0 in which
dx
M(x, y) =1+ x2y and N(x, y) = - 1.
If M(x,y)dx+N(x,y)dy is the complete differential of a function µ(x,y),
that is if
' M(x,y)dx + N(x,y)dy = dµ(x,y).
1) is called exact and µ(x,y) = C is its primitive or general solution.
EXAMPLE 2. 3x
2
l dx + 2x 3 y dy = O is an exact differential equation since
3x y dx + 2x 3 y dy = d(x 3 y2), Its primitive is x 3 y 2 = C.
2 2
If 1) is not exact but

t(x,y){M(x,y)dx + N(x,y)dy} = dµ(x,y),
t(x,y) is called an integrating factor of 1) and µ(x,y)=C is its primitive.
EXAMPLE 3. 3y dx + 2x dy = 0 is not an exact differential equation but when

multiplied by g(x,y)=x 2 y, we have 3x 2 y 2 dx+2x 3 yciy=O which is exact. Hence,
the primitive of 3y dx + 2x dy = O is x 3 y 2 = C. See Example 2.
If 1) is not exact and no integrating factor can be found readily, it may

be possible by a change of one or both of the variables to obtain an equation
for which an integrating factor can be found.
EXAMPLE 4. The transformation x = t-y, dx = dt-dy, (i.e., x+y = t),

reduces the equation ( x + y + 1 ) dx + ( 2x + 2y + 3) dy = 0
to (t +l)(dt-dy) + (2t +3)dy = 0

or (t+l)dt + (t+2)dy = O.
By means of the integrating factor _l_ the equation takes the form
t +2
dy + -t -+d
1 1
t = dy + dt - - - d t = o.
t+2 t+2
Then y + t - ln(t + 2) =C
and, since t = x + y, 2y + x - ln ( x + y + 2) = C.
Note. The transformation x + y + 1 = t or 2x + 2y + 3 = 2s is also suggested by
the form of the equation.
12
EQUATIONS OF FIRST ORDER AND FIRST DEGREE 13
A DIFFERENTIAL EQUATION for which an integrating factor is found readily has the
form
2)
By means of the integrating factor - - -1- - - , 2) is reduced to

f2(x) · g2(Y)
fi(x) dx + gi(y) dy = 0
f
2
(x) g2 (y)
whose primitive is
Equation 2) is typed as Variables Separable and in 2 1) the variables are sep-

arated.
EXAMPLE 5. When the differential equation

2 4
(3x y -xy)dx + (2x 3 y + x 3 y )dy = 0
2
2
y(3x -x)dx+ x 3 (2y +y )dy
2 4
is put in the form = 0
1
it is seen to be of the type Variables Separable. The integrating factor
yx3
reduces it to (~ - ..!_)dx + (2y + y 3 )dy = 0 in which the variables are sepa-
x x2
rated. Integrating, we obtain the primitive
3 lnx + ~ + y + iy
2 4
= C,
IF EQUATION 1) admits a solution f(x,y,C) = 0, where C is an arbitrary constant,

there exist infinitely many integrating factors t(x,y) such that
t(x,y){M(x,y)dx + N(x,y)dy} = 0
is exact. Also, there exist transformations of the variables which carry 1)

into the type Variables Separable. However, no general rule can be stated here
for finding either an integrating factor or a transformation. Thus we are
limited to solving certain types of differentialequations of the first order
and first degree, i.e., those for which rules may be laid down for determin-
ing either an integrating factor or an effective transformation.
Equations of the type Variables Separable, together with equations which
can be reduced to this type by a transformation of the variables are con-
sidered in Chapter 4.
Exact differential equations and other types reducible to exact equations
by means of integrating factors are treated in Chapter 5.
The linear equation of order one
3) dy + P(x) • y = Q(x)
dx
and equations reducible to the form 3) by means of transformations are con-
sidered in Chapter 6.
14- EQUATIONS OF FIRST ORDER AND FIRST DEGREE
These groupings are a matter of convenience. A given equation may fall into
more than one group.
EXAMPLE 6. The equation x dy - y dx = o may be placed in any one of the

groups since
a) by means of the integrating factor 1/xy the variables are separated; thus,
dy/y - dx/x = 0 so that lny - lnx = ln C or y/x = C.
b) by means of the integrating factor 1/x 2 or 1/y 2 the equation is made ex-
act; thus x dy - Y dx = 0 and K = C or x dy - Y dx = 0 and - x C
' X2 X y2 y= 1•
y
X
dy
c) when written as lX y = o, it is a linear equation of order one.
dx
Attention has been called to the fact that the form of the primitive is
not unique. Thus, the primitive in Example 6 might be given as
a) ln y - lnx = ln C, b) y/x = C, = Cx,
d) x/y = K, etc.
c) y
It is usual to accept any one of these forms with the understanding, already
noted, that thereby certain particular solutions may be lost. There is an
additional difficulty!
7. It is clear that y=0 is a particular solution of dy/dx=y or

EXAMPLE
dy-y dx = 0. When y f. 0, we may write dy/y - dx = 0 and obtain ln y - x = ln C
with Cf. 0;in turn, this may be written as y = Cex, Cf. o. Thus, to include
all solutions, we should write y = 0; y= Cex, Cf. o. But note that y = Cex,
free of the restrictions imposed on y and C, includes all solutions.
This situation will arise repeatedly as we proceed but, as is customary,
we shall refrain from pointing out the restrictions; that is, we shall write
the primitive as y = Cex, with C completely arbitrary. In defense, we offer
the following observation. Let us multiply the given equation by e-x to ob-
tain e-x dy - ye-x dx = 0 from which, by integration, we get e-x y = C or
y = Cex. In this procedure, it has not been necessary to impose any restric-
tion on y or C.
CHAPTER 4

VARIABLES SEPARABLE A'\ID REDUCTION TO
VAHIABLES SEPARABLE
VARIABLES SEPARABLE. The variables of the equation M(x,y)dx + N(x,y)dy = 0 are

separable if the equation can be written in the form
1)
The integrating factor ___l___ , found by inspection, reduces 1) to the

f2(x)•g2(Y)
form
from which the primitive may be obtained by integration.
For example, (x - 1/ydx + x 2 (y + l)dy = O is of the form 1). The integrat-

(x -1)2 (y+l)
ing factor - 1- reduces the equation to - - - dx + ~ - - dy = O in which
x2y x2 y
the variables are separated. See Problems 1-5.
HOMOGENEOUS EQUATIONS. A function f(x,y) is called homogeneous of degree n if

n
f(>-..x,>-..y) = >-.. f(x,y).
For example:
a) f(x,y) = x 4 - x3y is homogeneous of degree 4 since
f(h,>-..y) = (h) 4 - (>-..x) 3 (>-..y) = >-.4. (x4 - x 3 y) >-.4. f(x, y).
b) f(x, y) is homogeneous of degree o since

0
= e '>..y/"'Ax + tan >-..y = >-.. f(x,y),
>-..x
c) f(x,y) = x 2 + sinx cosy is not homogeneous since

2 2
f(h,>-..y) = >-.. x + sin(>-..x) cos(>-..y) -f. >-..nf(x,y).
The differential equation M(x,y)dx + N(x,y)dy = O is called homogeneous

if M(x,y) and N(x,y) are homogeneous and of the same degree. For example,
2
x int dx + :!:'._ arc sin ~ dy = o is homogeneous of degree 1, but
X X X
neither (x 2 +y 2 )dx- (xy 2 -y 3 )dy = O nor (x+y 2 )dx + (x-y)dy = O is a

homogeneous equation.
Jj
16 EQUATIONS OF FIRST ORDER AND FIRST DEGREE
The transformation
y = vx, dy = v dx + x dv
will reduce any homogeneous equation to the form
P(x, v)dx + Q(x, v)dv = 0
in which the variables are separable. After integrating, vis replaced by y/x
to recover the original variables. See Problems 6-11.
EQUATIONS IN WHICH M(x,y) AND N(x,y) ARE LINEAR BUT NOT HOMOGENEOUS.
a) The equation (a 1 x+b 1 y+c;)dx+(a 2 x+b 2 y+c 2 )dy = 0, (a 1 b 2 -a 2 b 1 0),
is reduced by the transformation
to the form P(x, t)dx + Q(x, t)dt = O
in which the variables are separable. See Problem 12.
b) The equation (a 1 x+b 1 y+ci)dx+(a 2 x+b 2 y+c 2 )dy 0, (a 1 b 2 -a 2 b 1 -1-- 0),

is reduced to the homogeneous form
(a 1 x' + b 1 y')dx' + (a 2 x' + b 2 y 1 )dy 1 0
by the transformation
x = x'+h, y = y' + k
in which x=h, y=k are the solutions of the equations

a 1 x + b1 y + c1 = 0 and a 2 x + b2 y + c 2 = 0. See Problems 13-14.
EQUATIONS OF THE FORM y. f(xy)dx + x.g(xy)dy = 0. The transformation
z x dz - z dx
xy = z, y = X-, dy =
x2
reduces an equation of this form to the form

P(x,z)dx + Q(x,z)dz 0
in which the variables are separable. See Problems 15-17.
arHER. SUBSTITUTIONS. Equations, not of the types discussed above, may be reduced
to a form in which the variables are separable by means of a properly chosen
transformation. No general rule of procedure· can be given; in each case the
form of the equation suggests the transformation. See Problems 18-22.
SOLVED PROBLEMS
VARIABLES SEPARABLE.
3
1. Solve x dx + (y+l/dy = 0,
The variables are separated, Hence, integrating term by term,
4
X
- + or
4
YARIABLES SEPARABLE 17
2 2
.I 2 • Solve x (Y + l)dx + y (x - l)dy = O.
2 2
1
The integrating factor reduces the equation to _x_ dx + _Y_ dy
x-1 y+l
o.
(y+l)(x-1)
1 1
Then, integrating (x+l + --)dx + (y-1 + --)dy = 0,
x-1 y+l
½x
2
+ x + ln(x -1) + ½/ - y + ln(y + 1) = C2 ,
2 2
x + y + 2x - 2y + 2 ln(x - 1) (y + 1)
and (x + 1/ + (y - 1/ + 2 ln(x - 1) (Y + 1) c.
v 3. Solve 4x dy - ydx =
2
x dy or ydx + (x
2
- 4x)dy = O.
1 dx dy
The integrating factor reduces the equation to +- 0 in which the
2 x(x - 4) y
y(x -4x)
variables are separated,
The latter equation may be written as *

X
dx
-4
-k dx
X
+
dy
y
= 0 or
X
dx
-4
dx + 4 dy =o.
X y
4
Integrating, ln(x -4) - ln x + 4 ln y = ln C or (x -4)y = Cx.
v 4. Solve dy 4y
or x (y - 3) dy = 4y dx,
dx x(y - 3)
1 Y.-3 4
The integrating factor reduces the equation to -·- dy = - dx,
xy y X
Integrating, y - 3 ln y = 4 ln x + ln C1 or y
4 3
This may be written as or X y
..; 5. Find the particular solution of ( 1 + x 3 )dy - x y dx = O satisfying the initial conditions x = 1,
2
y = 2.
1
First find the primitive, using the integrating factor
y(l + x 3)
2
dy x
Then- - - - dx = O, ln y - 1 ln ( 1 + x 3 ) = C1 , 3 ln y = ln( 1 + x 3) + ln C,
Y 1 + x3 3
3 3 3
Whenx=l, y=2: 2 = C{l+l), C=4, andtherequiredparticularsolutionis y =4(1+x ),
HOMOGENEOUS EQUATIONS.
J 6. When Mdx + Ndy = O is homogeneous, show that the transformation y vx will separate the
variables,
When Mdx + Ndy O is homogeneous of degree n, we may write
The transformation y ~ vx, dy = vdx + x dv reduces this to

Mi(v) dx + Ni(v){v dx + x dv} = O
or, finally, = O in which the variables are separated,

7. Solve (x
3 + y 3 )dx - 3x/dy = 0,
The equation is homogeneous of degree 3, We use the transformation y = vx, dy = v dx + x dv

to obtain 1) x 3{(1+v 3)dx - 3/(vdx + xdv)} = 0 or (l-2v 3)dx - 3v xdv = 0 2
in which the variables are separable.

2
1 dx 3v dv
Upon separating the variables, using the integrating factor - - - - - = 0,
X
1 - 2v 3
lnx+ ½ln(l-2v 3 ) =C 1 , x (1-2v 3 )

2
and 2lnx + ln(l-2)) = lnC, or C.
2 3 3
.
S rnce v = y/ x, th e pnm1
. . t 1. ve 1s
. x (1 - 2y ; x 3 ) =C or x - 2y 3 = Cx •
Note that the equation is of degree 3 and that after the transformation x 3 is a factor of
the left member of 1). This factor may be removed when making the transformation.
8. Solve xdy - ydx - ff-/ dx = 0,
The equation is homogeneous of degree 1. Using the transformation y = vx, dy v dx + x dv

and dividing by x, we have
v dx + x dv - v dx - ~ dx = 0 or x d v - ~ dx o.
1 dv dx
When the variables are separated, using the integrating factor - -- - o.
xri=--;;
Then arc sin v - In x = In C or arc sin v = ln(Cx), and returning to the original va- .
riables, using v y/x, arc sin :2:'. = ln(Cx) or Cx = e arc sin y/x
X
J 9. Solve (2x sinh ,2'. + 3y cosh ,2'.)dx - 3x cosh ,2'. dy

X X X
= 0.
The equation is homogeneous of degree 1. Using the standard transformation and dividing by
x, we have
2 sinhv dx - 3x coshv dv = 0.
dx· _ cosh v dv
Then, separating the variables, 2
x
3
sinh v
o.
Integrating, 2 lfi x - 3 ln sinh v = ln C, and X

2
C sinh3 :2:'. •
X
'- 10. Solve (2x+ 3y)dx + (y-x)dy = 0.
The equation is homogeneous of degree 1. The standard transformation reduces it to

2
(2+3v)dx + (v -l)(vdx + xdv) = 0 or (v +2v+2)dx + x(v-l)dv = 0,
dx __v_-_1_ dv dx 2v + 2 2 dv
Separating the variables, + - + ½ -2 - - - dv o.
X
v 2 + 2v + 2 x v + 2v + 2 (v + 1)2 + 1
Integrating, ln x + ½ ln (v 2 + 2v + 2) 2arctan(v+l) C1 ,
2 2
ln x
2
(/ + 2v + 2) - 4 arc tan(v + 1) = C, and ln(y + 2xy + 2x ) 4 arc tan x +Y
X
c.
1/ 11. Solve o.
VARIABLES SEPARABLE 19
The equation is homogeneous of degree O. The appearance of x/y throughout the equation
suggests the use of the transformation x = vy, dx = v dy + ydv.
Then (1+2ev)(vdy + ydv) + 2ev(l-v)dy = 0,
dy + 1 + 2ev dv
and o.
Y v + 2ev
x/y
Integrating and replacing v by x/y, ln C and X + 2ye = C,
LINEAR BUT NOT HOMOGENEOUS.

'12. Solve (x+y)dx + (3x+3y-4)dy = O.
The expressions (x + y) and (3x + 3y) suggest the transformation x + y = t.

We use y= t-x, dy=dt-dx to obtain tdx+ (3t-4)(dt-dx) 0
or (4-2t)dx + (3t-4)dt O
in which the variables are separable,
Then 2dx + ~ dt 2dx - 3dt + - - dt

2
o.
2- t 2-t
Integrating and replacing t by x + y, we have
2x-3t-2ln(2-t) = C1 , 2x-3(x +y) -2 ln (2-x -y) C1 , and x+3y+2ln(2-x-y) c.
, 13. Solve (2x-5y+3)dx - (2x+4y-6)dy = 0,
First solve 2x - Sy + 3 O, 2x + 4y - 6 = O simultaneously to obtain x = h = 1, y = k = 1.

The transformation x = x'+ h = x'+ 1, dx dx'
y=y'+k=y'+l, dy dy'
reduces the given equation to (2x' - 5y')dx' - (2x' + 4y')dy' = O
which is homogeneous of degree 1. (Note that this latter equation can be written down without
carrying out the details of the transformation.)
Using the transformation y' = vx', dy' = v dx' + x' dv,

we obtain (2-5v)dx'-(2+4v)(vdx'+ x'dv)=O, (2-7v-4v 2 )dx'-x'(2+4v)dv 0,
dx' 4 dv 2 dv
and finally - + --- + = o.
x' 3 4v -1 3 V + 2
1 2 3 2
Integrating, In x' + -h1(4v-1) + -ln(v+2) = lnC 1
3 3
or x' (4v-1) (v + 2) c..
Replacing v by y'/x', (4y'-x')(y'+ 2x 1 / = C,
2
and replacing x' by x - 1 and y' by y - 1, we obtain the primitive (4y-x-3)(y+2x-3) = C.
l 14. Solve (x-y-l)dx + (4y+x-l)dy = O.
Solving x-y-1=0, 4y+x-l=O simultaneously, we obtain x=h=l, y=k=O.

The transformation x=x'+h=x'+l, dx = dx'
y = y' + k = y' dy = dy'
reduces the given equation to (x' -y')dx' + (4y' + x')dy' = 0 which is homogeneous of de-
gree 1. (Note that this transformation x-1= x', y=y' could have been obtained by inspection,
that is, by examining the terms (x-y-1) and (4y+x-1),)
Using the transformation y'=vx', dy'=vdx'+x'dv

we obtain (1- v)dx' + (4v + 1) (v dx' + x' dv) 0.
Then
dx' 4v + 1 dv dx' Bv dv
- + + ½ dv + 0 '
x' 4v 2 + 1 x' 4v 2 + 1 4v 2 + 1
2 2
ln x' + ½ln(4/ + 1) + ½ arc tan 2v = C 1 , ln x' (4v + 1) + arc tan 2v = C,
2 2 2y I 2 2 2y
ln(4y' + x') + arc tan = C, and ln[4y + (x-1)] + arc tan - - = C:
x' X -1
I<,ORM y f(xy)dx + xg(xy)dy = O.
2
J15. Solve y(xy+l)dx+ x(l+xy+x /)dy = 0,
x dv - vdx
The transformation xy = v, y = v/x, dy
x2
v 2 x dv - v dx
reduces the equation to -(v + l)dx + x( 1 + v + v ) - - - - - 0
x x2
which, after clearing of fractions and rearranging, becomes o.

dx dv dv dv
Separating the variables, we have
V
o.
X v3
1 1
Then In x + - - + C'V 2 '
2v2 V
2
and 2x y2 ln y - 2xy - 1
16. Solve (y -x/)dx - (x + x 2y)dy = 0 or y( 1-xy)dx - x(l+ xy)dy = o.

,J
x dv- v dx
The transformation xy = v, y = v/x, dy reduces the equation to
x2
v x dv - v dx
-(1-v)dx - x(l+v) - - - -
x x2
0 or 2v dx - x(l + v)dv o.
2
Then ~ dv 0, 2 ln x - Inv - v ln C,
X
Cev, and
V v·
or
x dv - v dx
The transformation xy =v, y =v/x, dy reduces the equation to
x2
2
0 or vdx+x(v -v)dv o.
dx 2
Then + (v - l)dv 0, In x + ½v - v C, and In x
X
VARIABLES SEPARABLE 21
MISCELLANEOUS SUBSTITUTIONS.
) 18. Solve -dy = (y-4x)

2
or dy = (Y - 4x/ dx,
dx
The suggested transformation y - 4x = v, dy = 4dx+ dv reduces the equation to
4dx + dv = }dx or dx - ____±:_ = o.

2
v - 4
I V + 2 1) + 2 = v+2 -4X y - 4x + 2 = Ce -4x

Then x + 4 l n - - = C1, ln .. ln C - 4x, = Ce ' and 0
v-2 v-2 v-2 y-4x-2
2
I 19. Solve tan (x+y)dx - dy = O.
The suggested transformation x +y = v, dy = dv - dx reduces the equation to

2
tan v dx - (dv -dx) = 0, dx _ __d_v_ or dx - co/ v dv = 0.
Integrating, and 2(x-y) = C + sin 2(x+y),
2 ½ 2 ½
J 20. Solve ( 2 + 2x y )Y dx + (x y + 2)x dy = 0,
2 2
½ 2
The suggested transformation X
2
y V, y
V
dy v dv - ~ dx reduces the equation
x4 x~
to
2 2
2
(2 + 2v)~ dx + x(v + 2) ( v dv - ~ dx) 0 or v(3+v)dx - x(v+2)dv o.
x 4 x4 x~
dx 2 dv 1 dv
Then - - - - = 0, 3 ln x - 2 ln v - ln (v + 3) = ln C 1 ,
X 3v 3v+3
2 ½ 2 ½
and 1 = C1 xy(x y + 3) or xy(x y + 3) = C,
J 2 2 2 2
21. Solve (2x + 3y - 7)x dx - (3x + 2y - 8)y dy = 0,
2
The suggested transformation x = u, / = v reduces the equation to
(2.u+3v-7)du - (3u+2v-8)dv = 0
which is linear but not homogeneous,
The transformation u = s + 2, v = t + 1 yields the homogeneous equation ( 2s+ 3t)ds - (3s + 2t) dt
0, and the transformation s = rt, ds = rdt+ tdr yields 2(r 2 -l)dt + (2r+3)t dr = 0,
dt 1 dr 5 dr
Separating the variables, we have 2 dt + 2r + 3 dr 2 + - -- a.
t r2 - 1 t 2 r +1 2 r - 1
Then 4 ln t - ln(r + 1) + 5 ln (r - 1) ln C,
4
t (r-1/ ( s - t )~ (u-v-1)~ <x -l - 1)~
2
= C, and (x
2
- l - 1)~ C(x
2
+/ - 3).
r +1 s + t U+ V -3 x2 + y2 - 3
2
l 22. Solve x (x dx + y dy) + y(x dy - y dx) = 0,
2 2 2
Here xdx+ ydy =~d(x +/) and xdy-ydx = x d(y/x) suggest x +y2 = p2, y/x = tan 8,
or x = p cos 8, y = p sin 8, dx = - p sin 8 d8 + cos 8 dp, dy = p cos 8 d8 + sin 8 dp,
2 2
The given equation takes the form /cos 8 (P dp) + p sin 8 (P d8) = 0
or dp + tan 8 sec 8 d8 0.
r-;;---;; x+l 2
Then p + sec 8 ii x~ + y~ (-x-) and Cx •
22 EQLI\TIONS OF FIRST ORDER AND FIRST DEGREE
23, Determine whether or not each of the following functions is homogeneous and, when homogene-
ous, state the degree,
2
a) X -XY, homo. of degree two, e) arc sin xy, not homo.
y/x x/y
b) --2_' not homo. f) xe + ye , homo. of degree one.
X + y2
g) In x - In y or ln:. , homo. of degree zero.
__:L_, y
c) homo. of degree zero.
x2 + y2 h) /4 2 + 2xy + 3/, homo. of degree one.
d) X +y COS l'_, homo. of degree one. i) x siny + y sinx, not homo.

X
Classify each of the equations below in one or more of the following categories:
(1) Variables separable
(2) Homogeneous equations
(3) Equations in which M(x,y) and N(x,y) are linear but not homogeneous
(4) Equations of the form y f (xy)dx + x g(xy)dy = O
(5) None of the above apply.
J 24. 4y dx + x dy = 0 Ans. {1); (2), of degree one

2
25, (1+2y)dx+(4-x )dy o (1)
2 2
26. y dx - x dy = o (1); (2), of degree two
'2:7. (1 + y)dx - (1 + x)dy = o (1); (3)
2 2
28. (xy + y)dx + (x y - x)dy = 0 (4)
29. (x sin :!'. - y cos r)dx + x cos l'. dy o (2), of degree one
X X X
2
30. / (x + 2)dx + (x 3 + y 3 ) (Y dx - x dy) 0 (5)
31, y /4 2
+ y2 dx - x(x + /;2;/)dy 0 (2), of degree two
32, (x+y+l)dx + (2x+2y+l)dy = 0 (3)
33, Solve each of the above equations (Problems 24-32) which fall in categories {1)-(4),
4
Ans. 24. x y = C 28. y = Cxexy
2 2-x sin l'_ = C
25. (1 + 2y) =C- 29. X
2 +x X
26. y = x + Cxy 31. Cx - /42Jy = X ln ( /42Jy - x)

27. (1 + y) = C(l+x) 32. x + 2y + ln (x + y) = C
Solve each of the following equations,
34. (1 + 2y)dx - (4 - x)dy = o Ans. (x-4/(1+2y) = C

YARIABLES SEPARABLE 23
2 2 2
35. xy dx + (1 + x )dy = 0 Ans. y (1 + x ) = C
36. cot 8 dp + P d8 = O Ans. p = C cos 8

2
37. (x + 2y)dx + (2x + 3y)dy = O Ans, x + 4xy + 3/ C
38. 2x dy - 2ydx = /4 + 4/
2
dx
2 ~
39. (3y- 7x + 7)dx + (7y- 3x + 3)dy 0 Ans. (y - x + 1) (Y + x - 1) C
40. xydy = (y+l)(l-x)dx Ans. y + x = In Cx (y + 1)
2 2
41. (Y - x )dx + xy dy = 0 Ans. 2x2i=xll+C
2 -1/xy
42. y(l + 2xy)dx + x(l- xy)dy = O Ans. y = CX e
2 • 2 1- X
43, dx + ( 1- x ) cot y dy = o Ans. sin y = C --
1 +x
11 3
Ans. x + 4xy =C
5
45. (3x + 2y + l)dx - (3x + 2y - l)dy Ans. ln(15x+10y-1) + C
0
2(x-y) =
In each of the following, find the particular solution indicated.
46. x dy + 2y dx = O ; when x = 2, y = 1. Ans,

2 II 2 2
47. (x + / )dx + xy dy = O ; when x = 1, y = -1. Ans, x+2xy=3
48. cosy dx + (1 + e -x)sin y dy = 0; when x = 0, y = i-/4. Ans,

2 2 Ans, x ei-x/y
49. (Y + xy)dx - x dy = o; when x = 1, y = 1. =
50, Solve the equation of Problem 30 using the substitution y = vx.

2 3 - ½y 3 2
Ans. x y lnx - y + x Cx y
2
51. Solve y' = -2(2x+ 3y) using the substitution z = 2x+ 3y.
Ans.
1 + V3 (2x + 3y)
1- V3(2x + 3y)
52. Solve (x - 2 siny + 3)dx + (2x - 4 siny - 3)cosy dy = O using the substitution siny = z.
Ans. 8siny + 4x + 9 ln(4x - ssiny + 3) = C
CH·\PTER S

EXACT EQUATIONS AND REDUCTION TO EXACT EQUATIONS
THE NECESSARY AND SUFFICIENT CONDITION that
1) M( x, y) dx + N ( x, y) dy = O
be exact is
oM oN
2)
oy ox
At times an equation may be seen to be exact after a regrouping of its terms.

The equation in the regrouped form may then be integrated term by term.
For example, (x 2 -y)dx + (y 2 -x)dy = 0 is exact since
oM = ]_ (x 2 _ y) = - l = 2 (y 2 _ x) = oN.
oy oy ox ox
2
This may also be seen after regrouping thus: x dx + y 2 dy - (y cJx + xdy) = O.
This eqqation may be integrated term by term to obtain the primitive x3/3 +
y 3/3 - xy = C. The equation (y 2 - x)dx + (x 2 - y)dy = O, however, is not exact
. ~ oN
srnce - = 2y f 2x = See also Problem 1.
oy ox
IF 1) IS THE EXACT DIFFERENTIAL of the equation µ(x,y) = C,
dµ = oµ dx + oµ dy = M(x,y)dx + N(x,y)dy.
ox oy
Then oµdx = M(x,y)dx and µ(x,y) = rM(x,y)dx + cp(y),

ox
where fx indicates that in the integration y is to be treated as a constant

and ¢(y) is the constant (with respect to x) of integration. Now
2 {f x M( x, y) dx } + dcp = N ( x, y)
oy dy
from which dcp = cp'(y) and, hence, ¢(y) can be found. See Problems 2-3.
dy
INTEGRATING FACTORS. If 1) is not exact, an integrating factor is sought.
oM oN
a) If
oy ox
= f(x), a function of x alone, then e
f f(x)dx . .
1s an rntegrat-
N
ing factor of 1).
24
EXACT EQUATIONS 2,
.)
oM oN
oy ox f g(y)dy
If = - if(y), a function of y alone, then e is an inte-
.If
grating factor of 1). See Problems 4-6.
1
b) If 1) is homogeneous and Mx + Ny -f 0, then - - - is an integrating factor.
Mx+ Ny
See Problems 7-9.
c) If 1) can he written in the form y f(xy)dx + x g(xy)dy = 0, where f(xy)
-f g(xy), then
1 1 is an integrating factor.
xy{ f(xy) - g(xy)} Mx - Ny
See Problems 10-12.
d) At times an integrating factor may be found by inspection, after regroup-

ing the terms of the equation, by recognizing a certain group of terms as
being a part of an exact differential. For example:
GROUP OF TERMS INTEGRATING FACTOR EXACT DIFFERENTIAL
1 X dy- ydx
-" dy - y dx
x2 x2
1 y dx- xdy
xdy - ydx d(- :.)
y2 y2 y
1
xdy - ydx
xy
X dy -ydx
2
1 X dy-ydx X
X dy - ydx d(arc tan l)
x2 + y2 x2 + y2 X
X dy+ ydx -1
d{-----}, if n-/1
(xyt · (n -1) (xyt- 1
X dy + ydx
X dy+ y dx
d {ln(xy)}, if n =1
xy
-1
d{-------}, if n-/1
2(n - 1) (x 2 + y 2 )n- 1
xdx + ydy
X dx+ ydy
if n = 1
x2+ y2
See Problems 13-19.

e) The equation xrys(mydx+ nxdy) + xP/'(µydx+ vxdy)= 0, where r,s,m,n,
P, a,µ, v are constants and mv - nµ -:/ 0, has an integrating factor of the form
8
x~y • The method of solution usually given consists of determining a andµ
by means of certain derived formulas. In Problems 20-22, a procedure, essen-
tially that used in deriving the formulas, is followed.
SOLVED PROBLEMS
1. Show first by the use 2) and then by regrouping of terms that each equation is exact, and solve.
a) (4x3y3-2xy)dx + (3x4/-x2)dy = 0 d) 2x(yex2 -l)dx + ex2dy = 0

b) (3e 3xy-2x)dx+ e 3xdy = 0
~3 3~ 02 44
e) (6x'y +4x y )dx + (3x y +5x y )dy = 0
c) (cosy+ y cosx)dx + (sin x - x sin y)dy = O
c!M 2 c!N
a) By 2): 12x 3y - 2x = - and the equation is exact.
c!y c!x
4 2 2 4 3 2
By inspection: (4x 3y 3 dx + 3x y dy) - (2xy dx + x dy) = d(x y) - d(x y) o.
The primitive is x4y3-x2y = C.
c!M 3X c!N
b) By 2): 3e and the equation is exact.
c!Y dX
By inspection: (3e 3Xydx + e3x dy) - 2x dx d(e3x y) - d(x 2) = o.

3X 2
The primitive is e y - X = C.
c!M c!N
c) By2): sin y + cos x and the equation is exact.
c!Y dX
By inspection: (cosy dx - x sin y dy) + (y cos x dx + sin x dy)

d(x cosy)+ d(y sin x) = O. Theprimitive is x cosy+ ysinx C.
c!M c!N
d) By 2): and the equation is exact,
cJy dX
2 2
By inspection: (2xyex dx + ex dy) - 2x dx
2
X 2
The primitive is ye - X = C.
c!M c!N
e) By 2): 18x~/ + 20 X 3y 4 and the equation is exact.
cJy dX
6 2
By inspection: 3 ~ 4 4
(6x~y 3dx + 3x y dy) + ( 4x y dx + 5x y dy) = d(xoy3) + d(x 4 y ~ ) o.
6 3 4 ~
The primitive is X y + X y = c.
2. Solve 3
(2x +3y)dx+ (3x+y-l)dy = 0.
cJM 3 = c!N and the equation is exact.

c!Y c)X
Solution 1, Set µ(x,y) = fx (2x 3 + 3y)dx ½x 4 + 3xy + <:p(y).
Then c!µ = 3x+cp (y)

1
= N(x,y) = 3x+y-1, ¢' (y) = y -1, <:p(y) = ½/- y,
c!y
I 4 I 2 4 2
and the primitive is 2 X + 3xy + 2Y - Y = C1 or x + 6xy + y - 2y = C.
Solution 2. G:r;ouping the terms thus 2,:3dx+ ydy- dy + 3(ydx+xdy) = O

4
and recalling that ydx + xdy = d(xy), we obtain, by integration, ½x + ½/ - y + 3xy = C1
as before,
EXACT EQCATIONS 27
2 2
3, Solve (/ exy + 4x 3)dx + (2xyexy - 3/)dy = 0,
oM oN
and the equation is exact.
oy cJx
2
3
Set µ(x,y) = J X 2
(y e )
X
+ 4x )dx
2 2
oµ xy xy 2 2 3
Then 2xye + ¢ I (y) = 2xye - 3y , ¢'(y) = - 3y ' ¢(y) = -y '
oy
2
4
and the primitive is exy + x - y 3 = C.
2 2
The equation may be solved by regrouping thus 4x 3dx - 3y 2dy + (y 2 exy dx + 2xyexy dy) O
2 2 2
and noting that /exy dx + 2xyexy dy = d(exy ).
2
4. Solve (x +/+x)dx + xydy = O.
oM cJN
= 2y, = y; the equation is not exact.
oy OX
cJM cJN
oy ox In x
However,
2y-y = ! = f(x) and e
ff (x)dx
= e
fdx/x
e = X
N xy X
is an integrating factor, Introducing the integrating factor, we have

3 22 2 3 2 2 2
(x + xy + x )dx + x y dy = 0 or x dx + x dx + (xy dx + x y dy) O.
2
Then, noting that x/dx + x y dy we have for the primitive
4 3
:__ + X + h2/ 4 3 2 2
3x + 4x + 6x y = C.
4 3
cJM 3 y 4 y cJN 4
cJy
=
2
8xy e + 2xy e + 6xy + 1,
c)x
2xy ey - 2x/ - 3; the equation is not exact.
cJM cJN
cJM cJN 3 y 2 cJy cJx 4
However, - - - = Sxy e + 8xy + 4 and -g(y).
oy cJx M y
Then /g(y)dy = e-4Jdy/y = e-4 lny 1/y 4 is an integrating factor and, upon introduc-
ing it, the equation takes the form
2
2
(2xey + 2: + .2:.)dx + (x ey - :.... _ 3_:_)dy O and is exact.
y y3 y2 y4
2
2 y X X
Set µ(x,y) X e + + + cp(y).
y y3
2 2
clµ 2 X X
x 2 ey -
X
Then x ey - 3-+ ¢' (y) 3 ...:.. ' ¢'(y) 0, cp(y) = constant,
cJy y2 y4 y2 y4
2
2 y X X
and the primitive is X e + + - C.
y y3
28 EQCATIONS OF FIRST ORDER AND FIRST DEGREE
32 2 2 4 3 2
6• Solve (2x y + 4x y + 2xy + xy + 2y)dx + 2(y + x y + x)dy = O.
oM 3
4x y +
2
4x + 4xy + 4xy + 2,
3 7W
2( 2xy + 1); the equation is not exact.
oy c)x
oM oN
oy OX X
2
2x and the integrating factor is ef2xdx e When it is introduced, the
N
given equation becomes
2 2
3 2 2 2 4 x 3 2 x
(2x y + 4x y+2xy +xy + 2y)e dx + 2(y + x y+x)e dy 0 and is exact.
2
3 2 2 2 4
Set µ(x,y) fX (2x y + 4x y + 2xy + xy + 2y)e
X
dx
f x (2xy 2 + 2x y )e
3 2 x2
dx +
f x (2y + 4x y)~
2 x2
dx +
f x xy 4 ex2 dx
2 2 2
22X X ,4X ,;._
x y e + 2xye + 2y e + '+'(y).
oµ 2 X
2
X
2
3 X
2
,;._ I 3 2
x2
Then 2x ye + 2xe + 2y e + '+' (y) = 2 (y + x y + x) e , ¢' (y) 0,
oy
2
2 2 4 x ,
and the primitive is (2x y t 4xy + y )e = C.
1
7, Show that - -- , where Mx + Ny is not identically zero, is an integrating factor of the ho-
Mx + Ny
mogeneous equation M(x, y )dx + N (x, y )dy =0 of degree n. Investigate the case Mx + Ny= 0 iden-
tically,
We are to show that _M_dx + - -N- d y 0 is an exact equation, that is, that
Mx + Ny Mx + Ny
~(-M-)
o
-(--)·
N
cJy Mx + Ny cJx Mx + Ny
(Mx + Ny) cJM - M(x cJM cJM cJN

- + N + y cJN) Ny - - MN - My -
-cl M cJy cJy cJy cJy cJy
-{--)
cJy Mx + !Sy 2
(Mx + Ny/ (Mx + Ny)
and
cJN cJM cJN Mx cJN - MN - Nx cJM
(Mx + Ny) cJx - N (x - + M + y - )
cJ !V dx dx c)x dx
-(--)
c)x Mx +Ny 2 2
(Mx + Ny) (Mx + Ny)
...,M ...,M 0 ...,N ...,N

N(x £_ + y £_) - M(x ::_ + )
~(-Jl,-1-)
c)x cJY c)x y cJ y N(nM) - M(nN)
~(-M-) = 0
cJY Mx +Ny ox Mx + Ny (Mx + Ny)
2
(Mx + Ny/
(by Euler's Theorem on homogeneous functions).
If Mx +Ny= O identically, then ~ = - ! and the differential equation reduces to y dx - x dy

N X
= O for which 1/xy is an integrating factor.
1
The equation is homogeneous and is an integrating factor, Upon its intro-
Mx + Ny x5
EXACT EQUATIOKS 29
duction, the equation becomes 0 and is exact.
X 1 4 4
Set µ(x,y) f (-X + 'L)dx In x - ~ 'L + ¢(y).
x5 4 x4
Then
?)µ
dy
y3
x4
+ ¢' (y) - -4
y3
X
. ¢'(y) = 0, and the primitive is
4
~ 'L 4 4 4
ln x - 4 X4
C1 or y = 4x lnx + Cx ,
Note, The same integrating factor is obtained by using the procedure of a) above, The equa-
tion may be solved by the method of Chapter 4.
2dx
9• So 1 ve y + (x 2 - xy - y 2 ) ay
,
= 0.
The equation is homogeneous and is an integrating factor.

Mx + Ny
2 2
Upon introducing it the given equation becomes _Y_dx + x - xy -y dy = 0 which is exact.
x2 - / y(x2 - y2)
set µ(x,y) =
fx - -y d x
1
-f -1
(x-Y
X
- -)dx
1
1 ln x-y + ¢(y),
X
2' -Y 2 2 x+y 2 X +Y
2 2
?)µ X X - X)'-Y 1 X 1
Then - - - + ¢'(y) - - -- • ¢'(y) - • ¢(y) ln y,
dy x2 -y2 y(x2 _ y2) y x2 _ y2 y
2
and the primitive is ½ln x - Y + In y = ln C1 or (x -y)y C(x + y).
X +y
10. Show that when Mx - Ny is not identically zero, is an integrating factor for the
Mx-Ny
equation M dx + N dy = yf i(xy)dx + xf 2 (xy) dy = O. Investigate the case Mx - Ny= 0 identically.
The equation y f1 (xy) dx + x f2 (xy) dy = 0 is exact

xy{f1(xy) -f2(xy)} xy{fi(xy) -f2 (xy)}
since
f 1 ?)f2 - f2 ?)f1
dX dX
2
Y<f1-f2l
and
df (xy) ?)f (xy)

This is identically zero since y-- x--·
dy dX
:rn EQllA TIO NS OF FIRST ORDER AND FIRST DEGREE
If Mx - Ny= O, then M y and the equation reduces to x dy + y dx = 0 with solution xy = C.

N X
2 2 2 2
11 • Solve y(x y + 2)dx + x(2-2x y )dy = O.
1
The equation is of the form and is an inte-
Mx-Ny
grating factor.
2 2 2 2
Upon introducing it, the equation becomes X Y + 2 dx +2-- -2x~y yd = 0 and is exact.
3x 3 y2 3x2y3
X 2 2
Set µ(x,y) J (X Y + 2)dx J (-
X 1 2. dx
+ --)
1
ln x -
1
+ rp{y).
3x3y2 3x 3x3y2 3 3x2y2
?)µ 2 2 - 2x2/ 2 2
Then - - + ¢' (y) ¢' {y) - 3y. rp(y) = ln y,
?)y 3x2y3 3 x2 y3 3
1 1 2 2 1/ x2 y2
and the primitive is - ln x - - - - - In y = ln C1 or X = Cy e ' •
3 3x2y2 3
The equation may be solved by the method of Chapter 4.
12. Solve y(2xy+l)dx + x(1+2xy-x 3y 3 )dy = O.

1
The equation is of the form and is an inte-
Mx-Ny
grating factor.
Upon introdu~ingit, the equation becomes

is exact.
Set µ(x,y) t (2
-
x3y2
+ _l_)dx
x4y3
1
x2y2
- -1-
3x3 y3
+ rp(y).
Then
?)µ
dy
2
x2y3
+
1
x3y4
+ ¢' (y)
1
--+ --
x3y4
2
x2y3
- -·1
y
¢' (y) = - -y1 ' ¢{y) - ln y,
1 1 Ce -(3xy+1)/(3x 3y3)
and the primitive is - lny - - - - - - = C1 or y
x2y2 3x3y3
13. Obtain an integrating factor by inspection for each of the followfng equations.
2 2
a) (2x/ ey + 2xy 3 + y)dx + (x / ey - x / - 3x)dy = 0 (Problem 5)
2 3 3
b) (x y + 2y)dx + (2x - 2x /)dy = O (Problem 11)
2 4 3
c) (2x/ + y)dx + (x + 2x y - x y )dy = 0 (Problem 12)
a) When the equation is written in the form

2 2
/(2xeydx+x eydy) + 2xy 3dx - x /dy + ydx - 3xdy = 0
2
the term /(2xeydx+x eydy) = /•(an exact differential) suggests that 1/l is a possible
integrating factor. To show that it is an integrating factor, we verify that its introduction
produces an exact equation.
b) When the equation is written in the form 2 (y dx + x dy) + / y 3dx - 2x 3/ dy = 0, the term
(Y dx + x dy) suggests 1/(xyl as a possible integrating factor. An examination of the remaining
terms shows that each will be an exact differential if k = 3, i.e., 1/(xy) 3 is an integrating
factor.
EXACT EQUATIOI\S 31
4 3
c) When the equation is written in the form (xdy+ydx) + 2xy(xdy+ydx) - x y dy = 0 the
first two terms suggest 1/(xy)k. The third term will be an exact differential if k = 4;
thus, 1/(xy) 4 is an integrating factor.
2 2
14. Solve y dx + x ( 1 - 3x y ) dy = 0 or x dy + y dx - 3x 3/ dy = 0.
The terms x dy + y dx suggest 1/ (xy) k and the last term requires k = 3.
t he equation
. becomes x dy + y dx 3
Upon introducing the integrating factor -1- , - - - - - -dy 0
(xy)3 x3y3 Y
2 2
-1 /' =Ce-1 /(x y l.
whose primitive is 3 lny 6 lny ln C - or
3 2 2
15 , Solve x dx + y dy + 4y (x + y ) dy = 0.
The last term suggests 1/ (x 2 + y 2) as an integrating factor.
Introducing it, the equation becomes x dx + y dy + 4y3 dy 0 and is exact.

x2 + y2
2 2 4 2 2 24
The primitive is
I
2 ln(x + y ) + y or (x +y )e y C.
2
16. Solve x dy - ydx - (1-x )dx = O.
Here 1/x 2 is the integrating factor, since all other possibilities suggested by x dy -y dx
render the last term inexact.
X dy-y dx
Upon introducing it, the equation becomes (~ - l)dx 0 whose primitive
x2 x2
is r
X
+ - +
1
X
X = C or y + X
2
+ 1 Cx.
2 2 2
or x dx + y dy + (x + y ) dx O.
1
An integrating factor suggested by the form of the equation is Using it, we
(x2 + y2 i2
have x dx + y dy + dx 1 2 2
0 whose primitive is - ---- + x = C1 or (C + 2x) (x + y ) = 1.
(x2 + y2)2 2(x2+y2)
18. Solve x 2 dy + xy + /1 -x 2 y2 0 or x(xdy+ydx) + /1-x

2
/ dx = O.
dx
dx
The integrating factor reduces the equation to the form 0
2 X
x/1-x /
whose primitive is arc sin (xy) + ln x C.
19. y-x/-x 3 3 2 3 2
Solve dy or (x + xy -y)dx + (Y + x y+ x)dy o.
dx X + x2y + y3
2
When the equation is written thus (x + y2) (x dx + y dy) + x dy - y dx = O, the terms x dy- y dx
suggest several possible integrating factors. By trial, we determine 1/(x 2 + y 2 ) which reduces
E()F\TJO~S OF FIRST ORDER AND FIRST DEGREE
X dy -ydx
X dy-ydx x2
the given equation to the form d
xdx+yy+ X dx + y dy + == o.
x2+ y2
2
The primitive is ½x 2 +½/+arc tan~ C1 or x +/ + 2 arc tan lX = C.
3
20. Solve x(4ydx + 2xdy) + y (3ydx+ 5xdy) = 0.
a. 8
Suppose that the effect of multiplying the given equation by x y is to produce an equation
a.+1 6+1 a.+2 8 a. /3+4 a.+l /3+3
,4) (4x y dx + 2x y dy) + (3x y dx + 5x y dy) = 0
each of whose two terms is an exact differential. Then the first term of A) is proportional to
a.+2 i3+1 a.+1 8+1 a.+2 8
B) d(x y ) (a+2)x y dx + ((3+1)x y dy,
that is,
a + 2 /3 + 1
Cl and a - 2/3 = 0.
4 2
Also, the second term of A) is proportional to

a+1 8+4 a 8+4 a.+1 /3+3
D) d (x y ) (a+ l)x y dx + (/3+ 4)x y dy,
that is,
a+ l /3 + 4
El and 5a - 3/3 = 7.
3 5
Solving a-2/3 = 0, 5a-3/3 7 simultaneously, we find a= 2, /3 = 1.

When these substitutions are made in A), the equation becomes
3 5
The primitive is X
4
Y
2
+ X Y c.
2 3
21. Solve (Sy dx + 8xdy) + x y (4y dx + 5xdy) = O.
a. 8
Suppose that the effect of multiplying the given equation by x y is to produce an equation
a, 6+1 a+l 8 a,+ 2 8+4 a.+3 6+3
Al ( 8x y dx + 8x y dy) + ( 4x y dx + 5x y dy) = 0
each of whose two terms is an exact differential. Then the first term is proportional to
a.+1 /3+1 a. 8+1 a.+l /3
B) d (x y ) (a+ l)x y dx + (/3 + l)x y dy,
that is,
Cl
a + 1 /3 + 1
and a - /3 = O.
8 8
Also, the second term is proportional to

a.+3 8+4 a,+2 8+4 a.+3 8+3
D) d(x y ) (a+ 3)x y dx + (/3 + 4)x y dy,
that is,
a + 3 /3 + 4
£) and 5a - 4/3 = 1.
4 5
Solving a -/3 = O, 5a - 4/3 = 1 simultaneously, we find a = 1, /3 = 1.

E\:ACT E()l'ATIONS
When these substitutions are made in A), the equation becomes
2 2 4 5
The primitive is 4X Y + X Y = C,
Note. In this and the previous problem it was not necessary to write statements Bl and D)
since, after a little practice, the relations C) and£) may be obtained directly from A).
22. Solve x 3y 3 (2ydx+xdy) - (5ydx+7xdy) = 0,
a, 8
Multiplying the given equation by x y, we have
a.+3 ,'3+4 a.+4 /3+3 a. 8+1 a.+l 8
Al ( 2x y dx + x y ) dy - ( 5x y dx + 7x y dy) o.
If the first term of .4.) is to be exact, then

a+4 ,B + 4 and a - 2,8 = 4.
2 1
If the second term of A) is to be exact, then

a+l ,B + 1 and 7a - 5,8 = - 2.
5 7
Solving a-2,B = 4, 7a-5,B = -2 simultaneously, we find a= -8/3, ,B = -10/3.
Then, from A), 0,

each of the two terms is exact, and the primitive is
3 4/3 2/3 -5/3 -7/3 4/3 2/3 -5/3 -7/3 3 5/3 7/3
x y + 3x y = C1 , x y + 2x y C or x y3 + 2 CX y •
2
23. Select from the following equations those which are exact and solve,
x 3/3 + C
2
a) (x -y)dx - x dy = 0 Ans. xy =
2
b) y(x - 2y)dx - x dy = O
2 2
c) (x + y )dx + xy dy = O
2 2 2 3
(x + y ) dx + 2xy dy = o Ans. xy + x /3 = C
2
e) (x + ycos x)dx + sinx dy O Ans. x + 2y sin x = C
20 20 20
(l+e )dp+2pe d8=o Ans. p(l + e l = C
g) dx - ~ dy = O
2 2
h) (2x+3y+4)dx + (3x+4y+5)dy o Ans. x + 3xy + 2y + 4x + 5y = C
(4x 3y 3 + .!_)dx + (3x

4
i) / - .!_)dy 0
X y
2 2 2
j) 2(u + uv)du + (u + v )dv = O
k) (x ~ - y)dx + (y /4 2
+ y2 - x)dy 0 Ans. (x
2
+y )
2 3/2
- 3xy = C
EQUATIONS OF FIRST ORDER AND FIRST DEGREE
I) (x + y + l)dx - (x-y-3)dy = 0
2 2
m) (x + y + 1 )d.x - (y-x+3)dy =0 Ans. X + 2xy - y + 2x - 6y =C
2
n) csc e tan e dr - (r csc 8 + tan 8)d8 =0 Ans. r csc e = ln sec 8 + C
2 1 1nx+y 2
o) (y y
- + 2)d.x + [ - - + 2y (x + 1)] dy = 0 Ans. - - + (x + 1) (Y + 2) C
(x + y)
X x+y X
2 2 2 2 2 2
(2xyex y + / exy + l)d.x + (x2 ex y + 2xyexy - 2y)dy ex y + e xy + 2
p) 0 Ans. X - y C
24, solve the remaining problems above [b), c), g), I)] using the appropriate procedure of Chap. 4,
Ans. b) x/y=2lnx+C g) y = arc sin x/a + C
ln /x 2
+ / - 2x + 4y + 5 - arc tan y +
x-1
2
C
25, For each of the following, obtain an integrating factor by inspection and solve,
2 2 2 = Ce2x
, a) X dx + y dy = (x2 + /)d.x Ans, 1/(x +/); X + y
2 3
b) (2y- 3x)d.x + xdy = O Ans, x· X Y =X + C
2
'
2 2
c) (x - y )dx + 2xy dy = O Ans, 1/x ; y + x ln x =Cx
2 2 2 2 2 3
d) X dy - ydx 3x (x + y )dx Ans. 1/(x + y ) ; arc tan y/x = X + C
e) y dx - X dy + In x dx = 0 Ans. lfx2 ; y + ln x + 1 = Cx
2 2 22 2
/) (3x + y )dx - 2xy dy = 0 Ans, 1/x ; 3x - y = Cx
2 2 2 3 2
g) (xy- 2y )dx - (x - 3xy)dy 0 Ans. 1/xy x/y + ln(y /x ) = C
2 2 2 C 2 arc tan y/x
h) (x + y)dx - (x-y)dy = 0 Ans. 1/(x +/); X + Y = e
2 2
i) 2y dx - 3xy dx - x dy = 0 Ans. x/y ; x2/y - x3 = C
2 2
J) y dx + x(x y-l)dy = 0 Ans. y/x3 ; 3/ _ 2x2y3 = Cx
(y+x3y+2x2)d.x + (x + 4xy + Sy 3 )dy
4
k) 0 Ans. 1/ (xy + 2) ; ln(xy + 2) 3 + x3 + 3/ C
26- For each of the following, obtain an integrating factor and solve.
2 X
a) x dy - ydx = x exdx Ans. y = Cx + xe
2 2
b) (1 + y )dx = (x+x )dy Ans. arc tan y = ln x/(x+l) + C
2
c) (2y-x 3 )d.x + xdy = 0 Ans. X y
5
- x /5 = C
2 2
d) y dy + y dx - X dy = 0 Ans. y + X = Cy
2 2
e) (3y 3 - xy)dx - (x + 6xy )dy =0 Ans. 3/ + x ln(xy) = Cx
2 3 3 4 3
/) 3x /dx + 4(x y-3)dy =0 Ans. X y - 4y = C
2
g) y(x+y)dx - x dy =0 Ans. x/y + In x = C
2 2
h) (2y + 3x/)d.x + (x + 2x y)dy = 0 Ans. xy(l+xy) = C
2 2
i) y(/ - 2x )d.x + x(2/-x )dy = 0 Ans. x2/ ( / - x2) =C
1
27, show that - f(y/x) is an integrating factor of x dy - y dx = O.
2
X
CHAPTER 6
E<1uation~ of Fir~t Order and Fir:-;t Deo-ree

t,
Lll\EAR EQUATIONS AND THOSE REDUCIBLE TO THAT FORM
THE EQUATION 1) dy + yP(x) = Q(x),

dx
whose left member is linear in both the dependent variable and its derivative,
is called a linear equation of the first order. For example,
-~ + 3xy = sin xis called linear wliile -~ + 3xy 2 = sin x is not.
dx dx
d f P(x)dx dy f P(x)dx f P(x)dx JI'(x)dx d

Since -(ye ) = - e + y P(x) e = e (-1:' + yP(x)),
dx dx dx
e
f P(x) dx is an integrating factor of 1) and its primitive is
See Problems 1-7.
BERNOULLI'S EQUATION. An equation of the form
or dy + y-r+l P(x) = Q(x)

dx
dv
is reduced to the form 1), namely, + v{(l-n)P(x)} = (1-n)Q(x), by the
dx
transformation
y
-r+l
= v, y-n dy = 1 dv See Problems 8-12,
dx 1-n dx
OTHER EQUATIONS may be reduced to the form 1) by means of appropriate transforma-

tions. As in previous chapters, no general rule can be stated; in each in-
stance, the proper transformation is suggested by the form of the equation.
See Problems 13-18.
SOLVED PROBLEMS
LINEAR EQUATIONS.
dy
1 • Solve - + 2xy 4x.
dx
rP(x) dx x2
f P(x)dx = f 2x dx = x2 and e e is an integrating factor.
2 2 2
Then ye
X
f 4xex dx or y 2 + Ce ~x •
dy 3 2 dy 1 2
2. Solve X - y+x +3x -2x or - - -y X + 3x - 2,
dx dx x
JP(x) dx = -J dx X
- ln x and e
-ln X 1
X
is an integrating factor.
E()l.ATTOJ\S OF FIRST ORDER AND FIRST DEGREE
1 1 2
Then y-
x
J -(x +
X
3x - 2)dx J (x r 3 - ~) dx
X
or
1 2
2y x + 6x - 4x ln x + Cx.
dy 3 dy _ _1_ y
3. Solve (x-2)- y + 2(x - 2) or
dx dx X - 2
- ln(x - 2) 1
f P(x)dx -J dx -ln(x-2) and an integrating factor is e
X -2 X - 2
1 1
Then y(--) = 2 f ( x - 2 / , --dx 2J (x-2)dx (x-2/+ C or y (x-2)' + C(x-2).
x-2 x-2
COS X
4. Solve dy + y cot x 5e • Find the particular solution, given the initial conditions:
dx
x = brr , y = - 4•
Jcotx dx ln sin x
An integrating factor is e = e sin x and
_ COS X
y sin x 5JeCOSXsinx dx = - :)e c.
When x ½rr, y = -4: (-4) (1) -:i (1) + C and C ~ 1. The particular solution is
" COS X
y sin x + ;J e 1.
3 dy
5, Solve X + (2-3x )y
2
X
3
or 1.
dx
1 1
x2
- 3 ln x and an integrating factor is
~-
x e ·
2
J x'•e~;x 2
y 3 + (',X 3 e 1/x
Then or 2y X
dy
6• Solve - - 2y cot 2x = 1 - 2x cot 2x ..... 2 csc 2x.
dx
-J2 cot 2x dx - ln sin 2x
An integrating factor is e e = csc 2x.
Then y csc 2x = J (csc 2x - 2x cot 2x csc 2x - 2 csc

2
2x) dx = x csc 2x + cot 2x + C
or y = x + cos 2x + C sin 2x.
7. Solve y ln_y dx + (x - ln y)dy = 0.
dx 1 1
The equation, with x taken as dependent variable, may be put in the form + --·X
dy y ln y y
Then
efdy/(y lny) = eln(lny) ln y is an integrating factor.
Thus, x ln y f ln y dy 1 2
- ln y + K and the solution is 2x ln y ln y
2
+- C,
y 2
LINEAR EQUATIONS 37
BERNOULLI'S EQUATION.
dy 5 -5 dy -4
8. Solve dx - y = xy or y -- y x.
dx
dy 1 dv
The transformation y -4 = v, y -", dx reduces the equation to
4 dx
1 dv dv 4fdx 4X
or - + 4v = -4x. An integrating factor is e e
4 dx dx
4x 4X 4 4
Then ve -4 f xe dx -xe x + tie x + C,
-4 4X
-xe 4x + tie 4x + C, 1
y e or
y4
dy 4 -4 dy -3
9. Solve
dx
+ 2xy + xy 0 or y
dx
+ 2xy -x.
y-3 = v, -4 dy dv dv
The transformation -3y reduces the equation to 6xv 3x,
dx dx dx
2
-J6x dx -3x
Using the integrating factor e e we have
1 2
or - 2 + Ce3x •
dy 1 1 4 1
10. Solve dx + :? -(l-2x)y
3
or -(l-2x).
3
dv dv
The transformation y- 3 = v, - 3y-
4
dy reduces the equation to - V 2x - 1
dx dx dx
for which e -xis an integrating factor, Then, integrating by parts,

1
or = -1 - 2x + Cex.
y3
dy 2 . -2 dy -1
11. Solve dx + y y (cos x - srn x) or y dx + y cos x - sin x.
-1 -2 dy dv dv
The transformation y = v, -Y - reduces the equation to - v sin x - cos x
dx dx dx
for which e -x is an integrating factor, Then

1
or - sin x + Cex.
y
-3 dy 1 -2
12. Solve x dy - { y + xy 3 (l + ln x) }dx 0 or y dx -y 1 + ln x.
X
-2 -3 dy dv dv 2
The transformation y = v, -2y - reduces the equation to - + -v =-2(l+lnx)
dx dx dx X
for which e J2 dx/x = x

2
is an integrating factor. Then
2
2 2 2 4 3 X
vx = -2J(x + x lnx)dx - -x ~) lnx + C or
9 3 y2
:rn EQPATIOl\S OF FIRST ORDER AND FIRST DEGREE
MISCELLANEOUS SUBSTITUTIONS,
13. An equation of the form f' (y) : + f(.y) P(x) = Q(x) is a linear equation of the first order
dv
+ vP(x) = Q(x) in the new variable v = f(y). (Note that the Bernoulli equation
dx
-n dy + y-r,+l P(x) = Q(x)
y (-n + l)QCx) is an example,)
dx
dy
Solve + 1 or 4 sin x.
dx
y dv
In the new variable v = f(y) e , the equation becomes + v 4 sin x for which ex is
dx
an integrating factor. Then
vex= 4fex sinx dx = 2ex(sinx - cosx)+C or ey = 2(sinx - cosx) + Ce-x,
dy 2 . 2
14. Solve sin y cos x (2 cosy - sin x) or - sin y dy + cosy (2 cos x) sin x cos x.
dx dx
dv 2
In the new variable v = cos y, the equation becomes + 2v COS X sin x cosx for
dx
2rcosxdx
e ,
2 sinx
which e is an integrating factor. Then
2 s~r. x 2 S 1n X . 2 dx , 2 slnx . 2 , 2 slnx sinx

. + l.e2 slnx + C
ve fe sin x cos x 2 e sin x 2e n
or 1 • 2 • J. C -2 S 1n x
cosy 2 sin x - 2
1
sin x + n + e •
. dy sin y dy 1
15. Solve srny -
dx
cosy ( 1 - X COS y) or
2
- - -
cos y
-x.
cos y dx
d 1 sin y, 1 dv
Since - ( - - ) we take v and obtain the equation - V -x.
dy cosy 2 cos y dx
cos y
Using the integrating factor e-x, we obtain
or V secy
cos y
dy 3 2 3 2
16• Solve x - - y + 3x y - x 0 or x dy - y dx + 3x y dx - x dx o.
dx
Here (x dy - y dx) suggest the transformation :2:'. v.

X
Then x dy-ydx + 3x 2 °!..dx - dx 0 is reduced to dv + 3x2v 1 for which e

x3 is an
X
2 X dx
integrating factor.
Thus ve
X
3
or y = xe
-x3 f e x3 dx + Cxe
-x3 •
The indefinite integral here cannot be evaluated in terms of elementary functions.

LINEAR EQUATIONS 39
17. Solve
2
(4r s - 6)dr + r ds
3
= 0 or ( r ds + s dr) + 3s dr = ~ dr.
r2
The first term suggests the substitution rs = t which reduces the equation to
dt 3
dt + 3 ~ dr = ~ dr or + - t
6
r r2 dr r 2
r
Then r 3 is an integrating factor and the solution is
3 4 2 3 C
tr = r s = 3r + C or
3 2 x sin 8 d8 + cos 8 dx
18. Solve xsin 8 J8 + (x - 2x cos8 + cos 8)dx = 0 or - - - - - - - - - + 2cos8dx = x dx.
2
X
X sin 8 d8 + cos 8 dx
The substitution xy = cos 8, dy - - 2
reduces the equation to
X
dy
dy + 2xy Jx X UX or + 2.xy x.
dx
2
X
An integrating factor is e and the solution is
2 2 2 2 2
ye
X cos 8
X
e
X
= J
r
e
X
X dx
1 X
-e + K
2
or 2 cos 8 X + Cxe
-x
.
19, From the following equations, select those which are linear, state the dependent variable,
and solve.
2 2
a) dy /dx + y = 2 + 2x k) y(l + y )dx = 2(1- 2.xy )dy
2
b) dp/d8+3P=2 Z) yy' - xy + x = O
2 2 2
c) dy/dx - y = xy m) x dy-ydx = x/x - y dy
d) x dy - 2y dx = (x - 2)exdx n) ¢i< t) dx / d t + x ¢2 ( t) = 1
e) di/dt - 6i = 10 sin 2t 3
o) 2 dx/dy - x/y + x cosy = 0
2 X 2
dy /dx + y = y e p) xy I = y (1 - X tan X ) + X cos X
g) y dx + (xy + x - 3y)dy = 0 q) {2+/)dx (xy+2y+/)dy = o

2t zt
h) (2s - e )ds = 2(se - cos 2t)dt r) (1 + /)dx (arc tan y - x)dy
i) xdy+ydx=x 3y6dx s) 5
(2xy - y)dx + 2x dy = O
J) dr + (2r cot 8 + sin 28)d8 = o t) (1 + sin y)dx = [2y cosy - x(sec y + tan y)]dy
Ans.
X -6t
a) y; I.F., e • y = 2x + Ce -x e) i; I.F., e i = - ½(3 sin 2t + cos 2t) + Ce 6t
b) p; ...
I F e 30 •. 3p = 2 + Ce - 3e g) x; I.F., yey; xy = 3 (Y - 1) + Ce -y
d) y; j) r; I,F,, sin 8;
2
2r sin
2
e + sin 4 e C
2 2 2 2 2
k) x; I.F., (1 + y ) ; ( 1 + y ) x = 2 ln y + y +C
f
e ¢,2 ( t) dt/¢i( t) . f ¢ 2 (t)dt/¢i(t)
n) x; I.F.,
1 2
xe =
J ¢1\t)
e
f¢ 2 (t)dt/¢i(t) d
t + C
p) y; I.F., y = X COS X + Cx COS X

X COS X
q) x; I,F., 1/£+"/; X 2 + y
2
+ ch-;?
arc tany -arc tan y
r) x; I. F.' e ; X arc tan y - 1 + Ce
2
t) x; I,F., secy+ tany; x(secy + tany) = y + C
20. From the remaining equations in Problem 19, solve those of the Bernoulli type,
2
-1 -x 2 2
Ans. c) y v; 1/y = 1-x + Ce Z) y = v; y = 1 + Cex
-1 X -2 -2
fl y v; (C+x)ye +1= 0 o) X v· X y = cosy + y sin y + C
-5
2/y5 = Cx 5 + 5x 3 -4 2 4
i) y v; s) y v; 3x = (4x 3 + C)y
21. solve the remaining equations, h) and m), of Problem 19.

2 2t
Ans. h) s - se + sin 2t = C m) y = x sin(y + C)
22. Solve:
2
+ x 3ex
X
a) xy' = 2y subject to y = O when x = 1. Ans. y=x(e-e)
b) L ~¾ + Ri ~ E sin 2t, where L,R,E are constants, subject to the condition i =0 when t =O.
Ans. E (R sin 2t - 2L cos 2t + 2Le-Rt/L)
2 2
R + 4L
23. Solve:
2 dy
a) X cosy dx = 2x sin y - 1, using sin y z. Ans. 3x sin y = Cx 3 + 1
2 2 2 2
4x yy 1 = 3x (3/ + 2) + 2(3/ + 2) 3 ,
9 8
b) using 3y + 2 = z. Ans. 4X (C - 3x l (3y + 2)
3 3 2 X
c) (xy - y - x e ) dx + 3x/dy = o. using y 3 = vx. Ans. 2y3ex = xe 2X + Cx
2
3 3 - 1, 2 2
d) dy/dx + X (X + y) = X (X + y) Ans. 1/(x + y) = X + 1 + Cex
-x
e) (y+ey-e--,;)dx + (1+ ey)dy = o. Ans. y + ey = (x+C)e
CHAPTER 7
Geometric Applications
IN CHAPTER 1 it was shown how the differential equation
1) f(x, y, y') = 0
of a family of curves
2) g(x,y,C) = 0
could be obtained. The differential equation expresses analytically a certain

property common to every curve of the family.
Conversely, if a property whose analytic representation involves the de-
rivative is given, the solution of the resulting differential equation rep-
resents a one parameter family of curves, all possessing the given property.
Each curve of the family is called an integral curve of 1) and particular in-
tegral curves may be singled out by giving additional properties, for example,
a point through which the curve passes.
For convenience, the followiIJg properties of curves which involve the de-
rivative, are listed.
RECTANGULAR COORDINATES. Let (x,y) be a general point of a curve F(x,y) = 0.
y y
!<-,
y
!<-;~e<::>
X 0 subtangent subnonn. X
0 M N
01 T
a) dy is the slope of the tangent to the curve at (x,y),

dx
dx
b) is the slope of the normal to the curve at (x,y),
dy
c) Y-y = !<X-x) is the equation of the tangent at (x,y), where (X,Y) are the coordinates
of any point on it.
d) Y-y = - dx(X-x) is the equation of the normal at (x,y), where (X,Y) are the coordinates
dy
of any point on it.
dx dy
e) x -Y- and y-x - are the x- and y-intercepts of the tangent.
dy dx
41
42 GEOMETRIC APPLICATIONS
dy dx
j) x + y dx and y + x - are the x- and y- intercepts of the normal.
dy
g) y j1 +(dx/
dy
and xj 1 + <!) 2 are the lengths of the tangent tetween (x,y) and the x-
and y-axes.
h) y j1 +<!>2 and xj~ + (dx/

dy
are the lengths of the normal between (x,y) and the x-
and y-axes.
dx dy
i) y - and y- are the lengths of the subtangent and subnormal.
dy dx
]) ds = /(dx)2 +(dy)2 c1x/1+(!>2

= = dy
M y is an element of length of arc.
k) ydx or xdy is an element of area.
POLAR COORDINATES. Let (p,0) be a general point on a curve p f (0).
d0
l) tan tf; = p - , where tj; is the angle between the radius vector and the part of the tan-
dp
gent drawn toward the initial line.
2 d0
m) p tan tj; p - is the length of the polar subtangent.
dp
dp
n) p cot tf; is the length of the polar subnormal.
d0
2 d0
o) p sin tj; = p - is the length of the perpendicular from the pole to the tangent.
ds
p) ds j 2 d0 2
dpl+P(-) d0 )<:~/ + p
2
is an element of length
dp
of arc.
q) ½p 2 d0 is an element of area.
CEO METRIC APPLICA TJOKS
TRAJECTORIES. Any curve which cuts every member of a given family of curves at the
constant angle cu is called an w-trajectory of the family. A 90° trajectory of
tbe family is commonly called an orthogonal trajectory of the family. For ex-
ample, in f'igure (a) below, the circles through the origin with centers on the
y-axis are orthogonal trajectories uf the family of circles through the origin
with centers on the x-axis.
y
y
T C
(a)
In finding such trajectories, we shall use:
A) The integral curves of the differential equation
3) f( x,y,
y' - tan w ) = 0
1 + y'tan w
are thew-trajectories of the family of integral curves of
1) f(x,y,y') = 0.
To prove this, consider the integral curve C of 1) and an w-trajectory which

intersect at P(x, y), as shown in Figure (b) above. At each point of C for
which 1) defines a value of y'. we associate a triad of numbers (x, y; y'), the
first two being the coordinates of the point and the third being the corres-
ponding value of y' given by 1). Similarly, with each point of T for which
there is a tangent line, we associate a triad (x,y;y'), the first two being
the coordinates of the point and the third the slope of the tangent. To avoid
confusion, since we are to consider the triads associated with P as a point
on C and as a point on T, let us write the latter (associated with P on T) as
(x,y;y'). Now, from the figure, x=x, y= y at P while y'= tan e and y'= tan¢
are related by
y' = tan e = tan(¢- w) tan¢ - tanw y' - tan w
1 +tan¢ tanw l+y 1 tanw
Thus, at P (a general point in the plane) on an w-trajectory, the relation
f(x,y,y), = f(x,y, -y'- tan w ) = 0

----
1 + y'tan w
holds, or, dropping the dashes, f(x, y, y'- tan w ) o.

1 + y' tan w
B) The integral curves of the differential equation

4) f(x, y, -1/y') = 0
are the orthogonal trajectories of the family of integral curves of 1).
C) In polar coordinates, the integral curves of the differential equation
5)
are the orthogonal trajectories of the integral curves of
6) f(p,0,dp) = 0.
d/3
SOLVED PROBLEMS
1. At each point (x,y) of a curve the intercept of the tangent on the y-axis is equal to 2x/.
Find the curve,
Using e), the differential equation of the curve is y

dy 2 y dx - X dy
y - x- 2xy or 2x dx.
dx y2 I 2
I y-2xy
X 2 2 I
Integrating, - = X + C or X - X y = Cy.
y ----- X ___ _Jj
The differential equation may also be obtained di- 2xy2
dy y - 2xy2
rectly from the adjoining figure as
dx X
01
2. At each point (x,y) of a curve the subtangent is proportional to the square of the abscissa.
Find the curve if it also passes through the point (1,e).
dx
Using i), the differential equation is or where k is the
x2
proportionality factor.
Integrating, k In y = - !
X
+ C. When x = 1, y = e: k = -1 + C and C = k + 1.
1
The required curve has equation klny=-;+k+l.
3, Find the family of curves for which the length of the part of the tangent between the point of
contact (x,y) and the y-axis is equal to they-intercept of the tangent.
From g) and e), we have x j1 + (:)

2
y- x -
dx
dy
or A) X
2
= y
2 dy
- 2xy- .
dx
The transformation y = vx reduces A) to

dx 2v dv
(l+v 2 )dx + 2vxdv = 0 or + o.
X
1+ v2
2
Integrating, In x + ln(l+ v ) = ln C.
or 2
+ y
2 is the equation of the family,
Then C x = Cx
GEOMETRIC APPLICATIONS 4S
4, Through any point (x,y) of a curve which passes through the ongrn, lines are drawn parallel
to the coordinate axes. Find the curve given that it divides the rectangle formed by the two
lines and the axes into two areas, one of which is three times the other.
X X
0 0 A
(a) (b)
There are two cases illustrated in the figures.

X X X
a) Here 3(area OAP) = area OPE. Then 3_[
0
y dx = xy - _[
0
y dx or 4_[0 y dx xy.
To obtain the differential equation, we differentiate with respect to x.
4y=y+xdy
dy 3y
Thus, or
dx dx X
An integration yields the family of curves y = Cx 3 •
b) Here area OAP= 3(area OPE) and 4.[x y dx = 3xy.

0
The differential equation is dy = 1-, and the family of curves has equation y3 = Cx.
dx 3x
Since the differential equation in each case was obtained by a differentiation, extraneous
solutions may have been introduced, It is necessary therefore to compute the areas as a check.
In each of the above cases, the curves found satisfy the conditions. However, see Problem 5.
5. The areas bounded by the x-axis, a fixed ordinate x = a, a variable ordinate, and the part of a
curve intercepted by the ordinates is revolved about the x-axis. Find the curve if the volume
generated is proportional to a) the sum of the two ordinates, b) the difference of the two
ordinates.
a) Let A be the length of the fixed ordinate. The differential equation obtained by differen-
X dy
tiating 1) n.Ja /dx k(y + A) is n.y
2
= k- ,
dx
Integrating, we have 2) y (C - TT.x) = k.
When the value of y given by 2) is used in computing the left member of 1), we find
k(y - A).
C- TT.x C- TT.a
Thus, the solution is extraneous and no curve exists having the property a),
b) Repeating the above procedure with we obtain the differen-
tial equation n.y 2 = kdy

- whose solution is 2r) y(C - TT.x) = k.
dx
It is seen from 3) that this equation satisfies 1 1). Thus, the family of curves 2 1) has the
required property.
GEOMETRIC APPLICATIONS
6. Find the curve such that at any point on it the angle between the radius vector and the tan-
gent is equal to one-third the angle of inclination of the tangent.
Let 0 denote the angle of inclination of the radius vector, T the angle of inclination of
the tangent, and t/; the angle between the radius vector and the tangent.
Since t/; = T/3 = (t/; + 0)/3, then t/; = ½0 and tan t/; = tan ½0.
d0 dp
Using l). tan t/; = P- = tan ½0. so that
p
= cot ½0 d0.
dp
Integrating, ln p = 2 ln sin ½0 + ln C1 or C(l - cos 0),
7. The area of the sector formed by an arc of a curve and the radii vectors to the end points is
one-half the length of the arc, Find the curve.
Let the radii vectors be given by 0 = 01 and 0 = 0.
()
Using q) and p), 1.f
2 ()
p 2 d0
1
Differentiating with respect to 0, we obtain the differential equation
or 1) dp = ± p/p 2 - 1 d0.
2
If p = 1, 1) reduces to dp = O. It is easily verified that p = 1 satisfies the condition
of the problem.
2 dp
If p ,f 1, we write the equation in the form ± d0 and obtain the solution
p = sec(C ± 0), Thus, the conditions are satisfied by the circle p = 1 and the family of curves
p = sec(C + 0). Note that the families p = sec(C + 0) and p = sec (C - 0) are the same.
8. Find the curve for which the portion of the tangent between the point of contact and the foot
of the perpendicular through the pole to the tangent is one-third the radius vector to the
point of contact.
0
0
(a) (b)
In Figure {a): p = 3a = 3p cos(n: - t/;) = -3P cost/;, cost/; = -1/3, and tan t/; = - 2\1'2.
In Figure (b): p = 3a = 3P cost/; and tan t/; 2V2,
d0 dp
Using Z) and combining the two cases, tan t/; p- ± 2\1'2 or
p
dp
The required curves are the families p = C e 0/2/2 and p = Ce-0/2/2 •

GEOMETRIC APPLICATIO~S l7
9, Find the orthogonal trajectories of the hyperbolas xy = C.
The differential equation of the given family is x dy + y = O, obtained by differenti-

dx
ating xy = C. The differential equation of the orthogonal trajectories, obtained by replacing
dy dx dx
- by - - , is -x - + y = 0 or y dy - x dx = 0,
dx dy dy
2
Integrating, the orthogonal trajectories are the family of curves (hyperbolas) / - x = C,
y
Problem 9 Problem 10
2 2
10. Show that the family of confocal conics :_ + _Y_ 1, where C is an arbitrary consta~t,
C C-t..
is self-orthogonal,
Differentiating the equation of the family with respect to x yields ~ + ..J..f!.__ 0, where
C C-t..
dy A.X -t..py •
p = -, Solving this for C, we find C = - - so that C-t.. = When these replace-
dx X + YP X + YP
ments are made in the equation of the family, the differential equation of the family is found
to be
(x + yp) (px -y) - t..p = 0,
Since this equation is unchanged when p is replaced by -1/p, it is also the differential
equation of the orthogonal trajectories of the given family,
11. Determine the orthogonal trajectories of the family of cardiods p = C(l + sin 0),
Differentiating with respect to 0 to obtain dp = C cos 0, solving for C = _1_ dp

d0 cos 0 d0
and substituting for C in the given equation, the differential equation of the given family
is
dp p cos 0
d0 1+ sin 0
The differential equation of the orthogonal trajectories, obtained by replacing dp by

2d0 . d0
-P - lS
dp d0 cos 0
or dp + (sec 0 + tan 0)d0 o.
dp P(l + sin0) p
C cos 0
Then ln p + ln(sec 0 + tan 0) - ln cos 0 ln C or p C{l- sin 0).
sec 0 + tan 0
2
12. Determine the 45° trajectories of the family of concentric circles x + y2 = C.
The differential equation of the family of circles is x + yy' = 0,

The differential equation of the 45° trajectories, obtained by replacing y' in the above
y' - tan 45° y' - 1 y' - 1
equation by is X + Y-- = 0 or (x + y)dy + (x -y)dx = O.
1 + y'tan 45° 1 + y' • 1 + y'
Using the transformation y =vx, this equation is reduced to
(v
2
+ l)d.x + x(v + l)dv = 0 or dx + ~ dv o.
X V2 + 1
Integrating, ln x + ½ln(v 2 + 1) + arc tan v 1n K - 2 arc tan v ,
and x2 + y2 = Ke-2 arctan yfx.

2 -20 e
In polar coordinates, the equation becomes P = Ke or Pe C.
13. Find the equation of the curve for which

a) the normal at any point (x,y) passes through the origin.
b) the slope of the tangent at any point (x,y) is½ the slope of the line from the origin to
the point. Ans. y 2 = Cx
c) the normal at any point (x,y) and the line joining the origin to that point form an isos-
celes triangle having the x-axis as base. Ans. y 2 -x 2 = C
d) the part of the normal drawn at point (x,y) between this point and the x-axis is bisected
by the y-axis. Ans. y 2 + 2x 2 = C
e) the perpendicular from the origin to a tangent line of the curve is equal to the abscissa
of the point of contact (x, y). Ans. x 2 + y2 = Cx
f) the arc length from the origin to the variable point (x,y) is equal to twice the square
root of the abscissa of the point. Ans. y = ±(arc sin/2" + ~ ) + C
g) the polar subnormal is twice the sine of the vectorial angle. Ans. p = C - 2 cos 0
h) the angle between the radius vector and the tangent is½ the vectorial angle.
Ans . p = C(1 - cos 0)
e
i) the polar subtangent is equal to the polar subnormal. Ans. p = Ce
14, Find the orthogonal trajectories of each of the following families of curves.
a) X + 2y = C Ans. y-2x=K j) y = x - 1 + Ce-x Ans. x = y -1 + Ke -y
2 2 2
b) xy =C X -Y = K g) y =2x 2 {1-Cx) /+3/ ln(Ky) = 0
2
c) X+ 2/ = C y = Kx2 h) p a cos e p b sin 0
2
d) y = Ce -2x y = X +K i) p a(l + sine) p b{l-sin0)
2 -sine
e) y = x 3/ (C -x) (x2+//=K(2x2+/) ]) p a(sec e + tan 0) p be
CHAPTER 8
Physieal Applieations
MANY OF THE APPLICATIONS of this and later chapters will be concerned with the mo-
tion of a body along a straight line. If the body moves with varying velocity
v (that is, with accelerated motion) its acceleration, given by dv/dt, is due
to one or more forces acting in the direction of mot ion or in the opposite
direction. The net force on the mass is the (algebraic) sum of the several
forces.
EXAMPLE 1. A boat is moving subject to a force of 20 pounds on its sail
and a resisting force (lb) equal to 1/50 its velocity (ft/sec). If the direc-
tion of motion is taken as positive, the net force (lb) is 20- v/50.
EXAMPLE 2. To the free end of a spring of negligible mass, hanging verti-
cally, a mass is attached and brought to rest. There are two forces acting on
the mass - gravity acting downward and a restoring force, called the spring
force, opposing gravity, The two forces, being opposite in direction, are equal
in magnitude since the mass is at rest. Thus, the net force is zero.
Newton's Second Law of Motion states in part that the product of the mass
and acceleration is proportional to the net force on the mass. When the sys-
tem of units described below is used, the factor of proportionality is k = l
and we have
mass x acceleration net force.
THE U.S. ENGINEERING SYSTEM is based on the fundamental units: the pound of force
(the pound weight), the foot of length, and the second of time. The derived
unit of IT18ss is the slu~, defined by
weight in pounds
mass in slugs
g in ft/sec 2
Hence,
mass in slugs x acceleration in ft/sec 2 net force in pounds.
The acceleration g of a freely falling body varies but slightly over the
earth's surface. For convenience in computing, an approximate value g = 32
ft/sec 2 is used in the problems.
SOLVED PROBLEMS
1. If the population of a country doubles in 50 years, in how many years will it treble under the
assumption that the rate of increase is proportional to the number of inhabitants?
Let y denote the population at time t years and Yo the population at time t = O. Then
dy dy = k dt,
1) - = ky or where k is the proportionality factor.
dt y
kt
First Solution. Integrating 1), we have 2) ln y kt+ ln C or y = Ce •
kt
At time t =O, y "Yo and, from 2), C = y0 • Thus, 3) y = Yoe •
50 PHYSICAL APPLICATIONS
50 k 50 k
At t = 50, y = 2Yo, From 3), 2Yo = Yoe or e 2.
kt 350 = 50 kt 50 k t 2t
When y = 3yo, 3) gives 3 = e Then e (e ) = and t = 79 years.
Second Solution. Integrating 1) between the limits t = o. y =Yo and t = 50, y = 2Yo,
(2Yo
J,,,
Yo
dy
y
k I 0
50
dt, ln 2Yo - ln Yo = 50 k and 50k ln 2.
Integrating 1) between the limits t = 0, y =Yo and t = t. y = 3yo •
13Yo dy
Yo y
k I 0
t
dt, and ln 3 = kt.
50 ln 3
Then 50 ln 3 50kt t ln 2 and 79 years.
ln 2
2. In a certain culture of bacteria the rate of increase is proportional to the number present.
{a) If it is found that the number doubles in 4 hours, how many maY be expected at the end
of 12 hours? (b) If there are 10 11 at the end of 3 hours and 4• 10 11 at the end of 5 hours, how
many were there in the beginning?
Let x denote the number of bacteria at time hours. Then

dx
1) dx = kx or k dt.
dt X
a) First Solution. Integrating 1), we have 2) ln x = kt + ln C or

Assuming that x = x0 at time t = 0, C = x0 and
'lk
At time t = 4, x = 2x0 • Then 2x0 = x0 e and
12 k 'I k 3 3
When t = 12, x = x0 e = x 0 (e ) = x 0 (2) = 8x0 , that is, there are 8 times the original
number.
Second Solution, Integrating 1) between the limits t = o. X = Xo and t = 4, X = 2xo,
i~o
dx
X
k i 'I
dt, ln 2xo - ln x 0 4k and 4k = ln 2,
Integrating 1) between the limits t = 0, x =x 0 and t = 12, X =x,
lxXO
dx
X
k i 12
dt, and ln ..:._

Xo
12k 3(4k) 3 ln 2 ln 8.
Then X = Bxo, as before.

11 3k 10'1
b) First Solution. When = 3, X = 10\ Hence, from 2), 10 = Ce and C
3k
e
11
4• 10
_5_k_ •
e
10'1 2k k
Equating the values of C, Then e 4 and e 2.
3k
e
Thus, the original number is C bacteria.

PHYSICAL APPLICATIONS ,j l
4 4
Second Solution. Integrating 1) between the limits t=3, x=10 and t=5, x=4•10 ,
ln 4 2k and k = In 2.
Integrating 1) between the limits t = 0, x =x 0 and t = 3, x =10 4 ,
k I 3
dt, ln -
104
Xo
3k = 3 ln 2 = ln 8 and x0
104
8
as before.
3. According to Newton's law of cooling, the rate at which a substance cools in moving air is
proportional to the difference between the temperature of the substance and that of the air.
If the temperature of the air is 30° and the substance cools from 100° to 70° in 15 minutes,
find when the temperature will be 40°.
Let T be the temperature of the substance at time t minutes.

dT dT
Then = -k(T - 30) or = -k dt.
dt T - 30
(Note. The use of -k here is optional. It will be found that k is positive, but if +k is
used it will be found that k is equally negative.)
Integrating between the limits t = 0, T = 100 and t = 15, T = 70,
f 100
10 dT
T - 30
= -k I, lo
dt, ln 40 - ln 70 = -15k ln i
7
and 15k ln 2 o. 56.
4
Integrating between the limits 0, T = 100 and t = t, T = 40,
J
lDO
40
dT
T - 30
= -k I 0
t
dt, ln 10 - ln 70 = -kt, 15kt=15ln7,
15 ln 7
t=---
0.56
52 min.
4. A certain chemical dissolves in water at a rate proportional to the product of the amount un-
dissolved and the difference between the concentration in a saturated solution and the con-
centration in the actual solution. In 100 grams of a saturated solution it is known that 50
grams of the substance are dissolved. If when 30 grams of the chemical are agitated with 100
grams of water, 10 grams are dissolved in 2 hours, how much will be dissolved in 5 hours?
Let x denote the number of grams of the chemical undissolved after t hours. At this time
3 50
the concentration of the actual solution is 0-x and that of a saturaced solution is
100 100
Then
dx dx k
kx(.2Q_ _ 30 -x) kx x + 20 or - -dx- - dt.
dt 100 100 100 X X + 20 5
Integrating between 0, X 30 and t = 2, X = 30 -10 20,
120
30
dx
-
!
20
xt20 ~5 I 2
dt • and k
5
2
ln ~
6
-0.46.
X 0
Integrating between t = 0, X = 30 and t = 5, X = X,
-k
5
J"0
dt
•
ln----
3(x + 20)
5x
k = -0.46,
X
X
+ 20
0.38, and x 12. Thus, the amount dissolved after 5 hours is 30 - 12 18 grams.
PHYSH :AL APPLICATIONS
5. A 100 gallon tank is filled with brine containing 60 pounds of dissolved salt. Water runs into
the tank at the rate of 2 gallons per minute and the mixture, kept uniform by stirring, runs
out at the same rate. How much salt is in the tank after 1 hour?
Lets be the number of pounds of salt in the tank after t minutes, the concentration then
being s/100 lb/gal. During the interval dt, 2dt gallons of water flows into the tank and
2 dt gallons of brine containing ~ dt = ~ dt pounds of salt flows out.
100 50
Thus, the change ds in the amount of salt in the tank is ds - ~ dt.

50
Integrating, S
__ (_',e-t/50. At t = 0, s = 60; hence, C = 60 ands~ 60e-t/ 5o_
615
When t = 60 minutes, s . : 60e- = 60(.301) = 18 pounds.
6. The air in a certain room 150 1x50 1 x12 1 tested 0.2% CO 2 • Fresh air containing 0.05% CO 2 was
then admitted by ventilators at the rate 9000 ft3/min, Find the percentage CO 2 after 20 minutes,
Let x denote the number of cubic feet of CO 2 in the room at time t, the conceutration of
CO 2 then being x/90, 000. During the interval dt, the amount of CO 2 entering the room is
9,000(.000S)dt ft 3 and the amount leaving is 9,000 __ x_ dt tt 3 •
90,000
X -45
Hence, the change dx in the interval is dx = 9,000(.0005 - _x_)dt - --dt.
90,000 10
Integrating, 10 ln(x - 45) = - t + ln C1 and x = 45 + Ce-t/lO•

-t/10
At t = 0, X = .002(90,000) 180, Then C = 180 -45 = 135 and x = 45 + 135 e •
2 63
i\,,.;n t = 20, x = 45 + 135 e - = 63. The percentage CO 2 is then = ,0007 = 0.07%,
90,000
7, Under certain conditions the constant quantity

Q calories/ second of heat flowing through a wall
is given by
dT
Q = -kA - ,
dx
s<.)
where k is the conductivity of the material, 0
A ( cm 2 ) is the area of a face of the wal 1 perpen- 0
.....
I
dicular to the direct ion of flow, and T is the I
temperature x(cm) fromthat face such that T de- I X
t--- c..J
0
creases as x increases. Find the number of cal- 1
tn
t-
I
ories of heat per hour flowing through 1 square I
meter of the wall of a refrigerator room 125 cm I E-
I
thick for which k = O. 0025, if the temperature ,I
of the inner face is -5°C and that of the outer
face is 75 ° c. ◄
direction of- flow
Let x denote the distance of a point within

the wall from the outer face.
Integrating dT = - .!]_ dx from x 0, T = 75 to x 125, T = -5,

kA
Q j.125 80 -- -{125),
Q and Q= 80kA 80(.0025)(100/ =
16
cal •
kA o dx' kA 125 125 sec
Thus, the flow of heat per hour 3600Q = 57,600 cal.

PHYSICAL APPLICATIONS
8. A steam pipe 20 cm in diameter is protected with a

covering 6 cm thick for which k = 0.0003. (a) Find
the heat loss per hour through a meter length of
the pipe if the surface of the pipe is 200 °C and
that of the outer surface of the covering is 30°C.
(b) Find the temperature at a distance x > 10 cm from
the center of the pipe,
At a distance x > 10 cm from the center of the

pipe, heat is flowing across a cylindrical shell of
surface 'lrea 2n:x cm 2 per cm of length of pipe, From
Problem 7,
-kA dT dT dx
Q -2n:kx - or 2n:k dT -Q ~.
dx dx X
a) Integrating between the limits T = 30, x = 16

and T = 200, x 10,
2n:k J
30
200
dT -Q J16
10
dx.
X
340n:k = Q(ln 16 -ln 10) = Q ln 1.6 and
340n:k
Q = - - - cal/sec.
ln 1.6
Thus, the heat loss per nour through a meter length of pipe is 100(60/Q = 245,000cal.
340nk dx
b) Integrating 2n: k dT = between the limits T = 30, x 16 and T = T, x x,
ln 1. 6 x
oc .
l T
30
dT 170
ln 1.6
T-30
170
ln 1,6
and T= (30 +170
- - l n16
-)
ln 1. 6 x
Check. When x =10, T= 30 + ~ l n l . 6 200°C. When x = 16, T = 30 + O = 30°C.

ln 1.6
9. Find the time required for a cylindrical tank of radius 8 ft and height 10ft to empty through
a round hole of radius 1 inch in the bottom of the tank, given that water will issue from such
a hole with velocity approximately v = 4. 8 /h
ft/ sec, h being the depth of the water in the
tank.
The volume of water which runs out per second may be thought of as that of a cylinder 1 inch
in radius and of height v. Hence, the volume which runs out in time dt sec is
n:cl./ (4.8Vh)dt = ~(4. 8 ✓

h)dt.
12 144
Denoting by dh the corresponding drop in the water level in the tank, the volume of water
which runs out is also given ,by 64Tidh. Hence,
64(144) dh dh
~ ( 4. 8 ✓h) dt = - 64n: dh or dt -1920 - ·
144 4.8 ✓
h ,/j;
Integrating between t = 0, h = 10 and t = t, h = 0,
o dh 10bn
= -1920
I 10
-
,(h
and t = -38401h 3840 /io sec = 3 hr 22 min.
10. A ship weighing 48,000 tons starts from rest under the force of a constant propeller thrust of
200,000 lb. a) Find its velocity as a function of time t, given that the resistance in pounds
is 10,000v, with v = velocity measured in ft/sec. b) Find the terminal velocity (i.e., v when
t - oo) in miles per hour. (Take g = 32 ft/ sec 2.)
2
Since mass (slugs) x acceleration (ft/sec ) net force ( lb)
impetus of propeller - resistance,
48,000(2000) dv dv v 20
then 2OO,QQQ - 10,QQQ V or 1) +
32 dt dt 300 300
Integrating, ve
t/300
-
20 Je t;,oo d t 20 e t/300 + C.
300
a) When t = 0, v = O; C = -20 and v = 20 - 20e-t/ 3oo = 20(1 - e-t/ 3oo).
b) As t - co, v 20; the terminal velocity is 20 ft/sec= 13.6 mi/hr. This may also be ob-
tained from 1) since, as v approaches a limiting value, dv - O. Then v = 20, as before.
dt
11. A boat is being towed at the rate 12 miles per hour. At the instant (t = 0) that the towing
1 ine is cast off, a man in the boat begins to row in the direction of motion, exerting a force
of 20 lb. If the combined weight of the man and boat is 480 lb and the resistance (lb). is
equal to 1.75v, where v is measured in ft/sec, find the speed of the boat after½ minute.
2
Since mass (slugs) X acceleration ( ft/ sec ) net force (lb)
forward force resistance,
480 dv dv 7 4
then 20 - l, 75 V or + -v
32 dt dt 60 3
Integrating, ve
7 t/60
-4 Je 7t/60 dt 80 7t/60
-e + C.
3 7
12(5280)
When t = O, V -88 C -216
35
and V
80
+ -e
216 - 7t/6o
(60/ 5 7 35
When t 30, V -80 + -e

216 -3,5
11. 6 ft/sec.
7 35
12. A mass is being pulled across the ice on a sled, the total weight including the sled being
80 lb. Under the assumption that the resistance offered by the ice to the runners is negli-
gible and that the air offers a resistance in pounds equal to 5 times the velocity (v ft/sec)
of the sled, find
a) the constant force (pounds) exerted on the sled which will give it a terminal velocity of
10 miles per hour, and
b) the velocity and distance traveled at the end of 48 seconds.
2
Since mass (slugs) x acceleration (ft/sec ) net force {lb)
forward force - resistance,
80 dv dv
then F - 5v or + 2v ~F
5 • where F (lb) is the forward force.
32 dt dt
F -zt • = Q; then C = F -2t

Integrating, V - + Ce When t = 0, V and A) V = ~(1 e )•
5 5 ;J
F 10 (5280) 44
a) As t --+ 00, V = The required force is F = 220 lb,
5 3 3
(60/
44 -2t
b) Substituting from a) in A), v=-(1-e ).
3
44 -96 44 48
When t = 48: V = - (1 - e
3
) = - ft/ sec
3 '
and s = J0 V dt 697 ft.
PHYSICAL APPUCA TIONS
13. A spring of negligible weight hangs vertically. A mass of m slugs is attached to the other
end. If the mass is moving with velocity v 0 ft/sec when the spring is unstretched, find the
velocity v as a function of the stretch x ft.
According to Hooke's law, the spring force (force opposing the stretch) is proportional to
the stretch.
Net force on body weight of body spring force.
dv dv dx dv dx
Then m- mg - kx or m-- mv - mg - kx, since v.
dt dx dt dx dt
2 2
Integrating, mv 2mgx - kx + C.
2 2 2 2
When x = 0, V Vo, Then C = mv0 and mv 2mgx - kx + mv0 •
14. A parachutist is falling with speed 176 ft/sec when his parachute opens. If the air resis-
tance is Wv 2/256 lb, where Wis the total weight of the man and parachute, find his speed as
a function of the time t after the parachute opened.
Net force on system weight of system - air resistance.

2
Then
W dv
- -
g dt
w- Wv
256
or
dv
- -dt
8
.
v2 - 256
Integrating between the limits and t = t. v,
r
0, V 176 V
J:7:
dv
v 2 - 256
-~ l
8 o
t
dt, -321 lnV -16
--
V + 16 176
- ! It
8 0
-4t
V - 16 5 V -16 5 -4t 6 + Se
ln - ln = -4t, -e and V = 16
V t 16 6 V + 16 6 6 -Se - 4 t
Note that the parachutist quickly attains an approximately constant speed, that is, the
terminal speed of 16 ft/sec.
15. A body of mass m slugs falls from rest in a medium for which the resistance (lb) is propor-
tional to the square of the velocity (ft/sec). If the terminal velocity is 150 ft/sec, find
a) the velocity at the end of 2 seconds, and
b) the time required for the velocity to become 100 ft/sec.
Let v denote the velocity of the body at time t seconds.

Net force on body weight of body - resistance, and the equation of motion is
dv
1) m-
dt
2
Taking g 32 ft/sec 2, it is seen that some simplification ispossiblebychoosing K=2mk •
dv 2 2 dv
Then 1) reduces to 2(16 - k v ) or = -2 dt.
dt k2v2 - 16
kv -4 kv -4 Ce -16 kt •
Integrating, ln-- -16 kt + ln C or
kv + 4 kv + 4
kv- 4 -16 kt
When t 0, V o. Then C = -1 and 2) -e
kv + 4
-16 kt -150
-2 , -. 43 t
V
When t .... oo, V 150. Then e 0, k and 2) becomes -e
75 Vt 150
V -150 -,86
a) When 2, -e = -.423 and v = 61 ft/ sec.
V + 150
-1.6
b) When v 100, .2 e and 3. 7 sec.
16. A body of mass m falls from rest in a medium for which the resistance (lb) is proportional to
the velocity (ft/sec). If the specific gravity of the medium is one-fourth that of the body
and if the terminal velocity is 24 ft/sec, find {a) the velocity at the end of 3 sec and (b)
the distance traveled in 3 sec.
Let v denote the velocity of the body at time t sec. In addition to the two forces acting
as in Problem 15, there is a third force which results from the difference in specific grav-
ities. This force is equal in magnitude to the weight of the medium which the body displaces
and opposes gravity,
Net force on body weight of body - buoyant force - resistance, and the equation of
motion is
dv 1 3
m- mg - - mg - Kv mg - Kv.
dt 4 4
2 dv dv
Taking g = 32 ft/sec and K 3mk, the equation becomes 3(8- kv) or 3dt.
dt 8- kv
Integrating from t = O, v = O to t, V = V,
~ ln(8-kv)I~ = 3tl: -ln(8-kv) + ln 8 = 3kt, and kv 8(1-e- 3 kt).
-t
When t .... m, v = 24. Then k = 1/3 and 1) V = 24(1- e ),
a) When t = 3, v = 24(1 - e- 3 ) = 22.8 ft/sec.

dx -t
b) Integrating V = = 24( 1 - e ) between 0, X 0 and 3, X = x,
dt
xi: = 24( t + e - t ) 13
0
and X = 24(2 + e- 3 ) 49,2 ft.
17. The gravitational pull on a mass m at a distance s feet from the center of the earth is pro-
portional tom and inversely proportional to s 2 • a) Find the velocity attained by the mass
in falling from rest at a distance SR from the center to the earth's surface, where R = 4000
miles is taken as the radius of the earth. b) What velocity would correspond to a fall from
an infinite distance, that is, with what velocity must the mass be propelled vertically up-
ward to escape the gravitational pull? (All other forces, including friction, are to be neg-
lected.)
The gravitational force at a distance s from the earth's center is km/s 2 • To determine k,
note that the force is mg whens = R; thus mg= km/R 2 and k ~ gR 2 , The equation of motion is
dv ds dv dv 2 ds
1) m- m- - mv or v dv = -gR
dt dt ds ds s2
the sign being negative since v increases ass decreases.
a) Integrating 1) from v = 0, s = SR to v v, s = R,
2 2 1 1 4
v 2 = ~(32) (4000)(5280),
I
2v = gR (---) = -gR,
R SR 5 5
and v 2560v'l65 ft/sec or approximately 6 mi/sec.

PIIYSlCAL APPLICATIONS 57
b) Integrating 1) from v = 0, s - oo to v = v, s = R,
l0 v v dv = -gR
2
JroR -sds2 , v
2
= 2gR, v = 6400 /33 ft/sec or approximately 7 mi/sec.
18. One of the basic equations in electric

circuits is
1) Ldi + Ri = E(t), L
dt L
where L(henries) is called the induc-
tance, R(ohms) the resistance, i(am-
peres) the current, and £(volts) the
electromotive force or emf. (In this R
book, Rand L will be constants,)
a) Solve 1) whenE(t)=E0 and the ini-
tial current is i 0 •
(a)
b) Solve 1) when L
= 3 henries, R = 15
ohms, E(t) is the 60 cycle sine wave of
amplitude 110 volts, and i = O when t = O.
a) Integrating L
di
+ Ri E0 , Le
Rt/L E
-
0 f e Rt/L d t E0 Rt/L C
-e + or = Ea + Ce -Rt/L 0
dt L R R
Ea E0 ( l - e -Rt/L ) -Rt/L
When t = 0, i = i 0 • Then C = to and + i0 e
R R
Note that as t - oo, i = E0 /R, a constant.
b) Integrating 3 di + 15 i E0 sin wt 110 sin 2rt(60) t 110 sin 120rt t,

dt
5t 110 Je5t sin 120rr t dt 110 e 5t 5 sin 120rt t - 120rt cos 120rt t + C
ie
3 3 25 + 14400 rt 2
22 sin 12011:t - 24rt cos 120rtt + Ce-5t

or 0
3 1 + 576rt 2
22• 24 rt
\'/hen t =0, i = O. Then C
3(1+576rt 2 )
22 sin 120rtt - 24rt cos 120rtt + 24rt e- 5t

and
3 1 + 576rt 2
A more useful form is obtained by noting that the sum of the squares of the coefficients
of the sine and cosine terms is the denominator of the fraction above. Hence, we may define
24 rt
sin¢ and cos¢ =
(1+ 576rt 2 )½
22 176rte- 5t
so that - - - - - - (cos¢ sin 120rtt - sin¢ cos 120rt t) +
3( 1 + 576rt 2 )½ 1 + 576n: 2
22 176rte- 5 t
sin( 120rt t - ¢) +
3< 1 + 576rt
2 2
l 1 + 576rt 2
19. If an electric circuit contains a resistance R(ohms) and a con-

R
denser of capacitance C( farads) in series, and an emf E(volts),
the charge q(coulombs) on the condenser is given by
dq
R- + :l. = E.
dt• C
If R = 10 ohms, C = 10- 3 farad and E(t) = 100 sin 120rtt volts,
a) find q, assuming that q = 0 when t = O.
b) use i = dq/dt to find i, assvming that i = 5 amperes when t = o.
C
Integrating 100 sin 120rt t, we have
100 t lOOt 100 sin 120rtt - 120rt cos 120rtt

q_e 10 J/ 00
t sin 120rt t dt 10e ------------
2
--+A
10,000 + 14,400rt
lOOt 10 sin 120rt t - 12rt cos 120TTt

e + A,
100 + 144rt 2
1 -lOOt
and 1) q - - - - - - , - , - sin( 120rtt - ¢) + .4 e
(100 + 144rt 2 )½
where sin¢ and cos¢
-mot
3rt 3 rt e
a) When t =0, q= O. Then A and q sin(l20rtt - ¢) +
2 2
25 + 36rt 2( 25 + 36rt 2 )½ 25 + 36rt
b) Differentiating 1) with respect tot, we obtain
dq
dt
300rt
When t=O, t=5. Then lOOA - 5
2
25 + 36rt
300rt -lOOt
and 60rt cos( 120'tt - ¢) ( - 5)e
2
(25 + 36rt 2 )½ 25 + 36rt
20. A boy, standing in corner A of a rectangular y

pool, has a boat in the adjacent corner Bon B
the end of a string 20 feet long. He walks
along the side of the pool toward C keeping
the string taut. Locate the boy and boat when (.i-,.Y)
the latter is 12 feet from AC.

''
' ',fo
Choose the coordinate system so that AC is
along the x-axis and AB is along t1J.e y-axis.
Let (x ,y) be the position of the boat when
X /400 -l' - -,,~ 0 X
A E C
the boy has reached E, and let 8 denote the
angle of inclination of the string.
dy -y /400 - y 2
Then tan 8 or dx dy.
dx y
PHYSICAL APPLICATIOJ\"S
Integrating, x = -
✓ 400 - y2 + 20 + /400 - y2
20 ln ------=-- + C.
y
When the boat is at B, x = 0 and y = 20.
✓~--~ ·20 + /400 -y 2

Then C = 0 and x = - 400 - / + 20 ln -------'-- is the equation of the boat's path.
y
20 + /400 -y 2
Now AE X + 1400 - / 20 ln - - - - - - ' - Hence, when the boat is 12 feet from
y
AC (i.e., y 7
12), x +16 = 20 ln 3 = 22.
The boy is 22 feet from A and the boat is 6 feet from AB.
21. A substance y is being formed by the reaction of two substances a and fin which a grams of a
and b grams off form (a+ hJ grams of y. If initially there are x 0 grams of a, y 0 grams off,
and none of y present and if the rate of formation of y is proportional to the product of the
quantities of a and f uncombined, express the amount (z grams) of y formed as a function of
time t.
bz
The z grams of y formed at time t consists of
az
grams of a and grams of e.
a+ b a+ b
bz
Hence, at time t there remain uncombined (x 0 - ~-)grams of a
a+ b
and (y0 - - ) grams of
a+ b
e.
dz I< ab a+ b a+ b
Then ---(--xo z)(--Yo z)
dt (a+ a b/ b
(a+ b)y0
k (A - z) (B - z) , where k A and B
a b
There are two cases to be considered: 1) A f B, say A>B, and 2) A~ B.
dz dz dz
1) Here k dt.
(A - z) (B - z) A-BA-z A-BB-z
Integrating from t = 0, z = 0 to t t' z ~
z' we obtain
1 A -z A
A-B
A-zr
ln - -
B- z 0
= kt 1: _1_ ( ln A - z
A -B B- z
ln ~)
13·
kt,
B- z
-e
B
(A - 5 )kt
and
dz
2) Here k dt. Integrating from t 0, z 0 to t t. z -, we obtain
2
(A - z)
1
A -z
r0
kt,: ,
A -z
- 1
-
A
kt, and z =
2
A kt
---·
1 + Akt
60 PHYSICAL APPLICI\ TIO:\S
SUPPLEI\IENTARY PROBLEMS
22, A body moves in a straight line so that its velocity exceects by 2 its distance from a fixed
point of the line. It v =5 when t = 0, find the equation of motion. Ans. x - Set - 2
23. Find the time requirPd for a sum of money to double itself at 5% per annum compoundect con-
tinuously. Hint: dx_ldt = 0.05x, where xis the amount after t yearn. Ans. 13.9 years
24. Radium decomposes at a rate proportional to the amount present. If half the original amount
disappears in 1600 years, find the percentage lost in 100 years. Ans. 4. 2 %
25. In a culture of yeast the amount of active ferment grows at a rate proportional to the amount
present. If the amount doubles in 1 hour, ho¼ many times the original amount may bP antici-
pated at the end of 2~ hours? Ans, 6. 73 times the original amount
26. If, when the temperature of the air is 20°C, a certain substance cools from 100· C to 60 )C
in 10 minutes, find the temperature after 40 minutes. Ans, 25°C
27. A tank contains 100 gal of brine made by dissolving 60 lb of salt in water. Salt water con-
taining 1 lb of salt per gal runs in at the rate 2 gal/min and the mixture, kept uniform by
stirring, runs out at the rate 3 gal/min. Find the amount of salt in the tank at the end of
1 hr. Hint: dx/dt = 2 - 3x/(100-t). Ans. 37.4 lb
28. Find the time required for a square tank of side 6 ft and depth 9 ft to empty through a one
inch circular hole in the bottom. (Assume, as in Prob. 9, v = 4. Bv'h ft/sec.) Ans. 137 min
29, A brick wall (k = 0.0012) is 30 cm thick. If the inner surface is 20°C and the outer is 0°C,
find the temperature in the wal 1 as a function of the distance from the outer surface and
the heat loss per day through a square meter. Ans. T = 2x/3; 691,000 cal
30. A man and his boat weigh 320 lb. If the force exerted by the oars in the direction of motion
i:o 16 1 b and it' the resistance ( in 1 b) to the motion is equal to twice the speed (ft/sec),
find the speed 15 sec after the boat starts from rest. Ans. 7.6 ft/sec
31. A tank contains 100 gal of brine made by dissolving 80 lb of salt in water. Pure water runs
into the tank at the rate 4 gal/min and the mixture, kept uniform by stirring, runs out at
the same rate. The outflow runs into a second tank which contains 100 gal of pure water ini-
tially and the mixture, kept uniform by stirring, runs out at the same rate. Find the amount
of salt in the second tank after 1 hr.
. dx 4 -o.o~t x
!lint: - = 4(-e ) - 4 - for the second tank, Ans. 17.4 lb
dt 5 100
32. A funnel 10 in. in diameter at the top and 1 in. in diameter at the bottom is 24 in. deep.
If initially full of water, find the time required to empty, Ans. 13.7 sec
33. Water is flowing into a vertical cylindrical tank of radius 6 ft and height 9 ft at the rate
6n ft 3/min and is escaping through a hole 1 in. in diameter in the bottom. Find the time re-
quired to fill the tank. Hint: (~ - _n:_ 4.8v'h)dt = 36n:dh. Ans. 65 min
10 (24/
34. A mass of 4 slugs slides on a table. The friction is equal to four times the velocity, and
the mass is subjected to a force 12 sin 2t lb. Find the velocity as a function of t if v = O
when t = O, 3 -t
Ans. v = -(sin 2t - 2 cos 2t + 2e )
5
35. A steam pipe of diameter 1 ft has a jacket of insulating material (k = 0.00022) ft thick. !
The pipe is kept at 475°F and the outside of the jacket at 75°F. Find the temperature in the
jacket at a distance x ft from the center of the pipe and the heat loss per day per foot of
pipe. Ans, T = 75 - 400(ln x)/(ln 2); 69,000 B.T.U.
36. The differential equation of a circuit containing a resistance R, capacitance C, and emf e
E sin wt is R di/dt + i/C = de/dt. Assuming R,C,E,w to be constants, find the current i
at time t. ECw . -t/RC
Ans. - - - - ( c o s wt + RCw-srn wt) + C 1 e
2 2 2
1+R C w
CHAPTER 9
Equations of First Order and Higher Degree
A DIFFERENTIAL EQUATION of the first order has the form f(x,y, y') = 0 or f(x,y,p)=0,
where for convenience y' = :: is replaced by p. If the degree of p is greater
2
than one, as in p - 3px + 2y = 0, the equation is of first order and higher
(here, second) degree.
The general first order equation of degree n may be written in the form
It may be possible, at times, to solve such equations by one or more of

the procedures outlined below. In each case the problem is reduced to that
of solving one or more equations of the first order and first degree.
EQUATIONS SOLVABLE FOR p. Here the left member of 1), considered as a polynomial
in p, can be resolved into n linear real factors, that is, 1) can be put in
the form
(p - F 1) (p - F 2) • • • • · • • (p - Fn) = 0,
where the F's are functions of x and y.
Set each factor equal to zero and solve the resulting n differential equa-
tions of first order and first degree
dy = dy
=
dx dx
to obtain
2) fn(x,y,C) =0.
The primitive of 1) is the product

3) f 1 (x,y,C)•f2 (x,y,C)• · • • · • • • •4i(x,y,C) 0
of then solutions 2).
Note. Each individual solution of 2) may be written in any one of its sev-
eral possible forms before being combined into the product 3). See Prob.1-3.
EQUATIONS SOLVABLE FOR y, i.e., y = f(x,p).

Differentiate with respect to x to obtain
dy = P = cJf + cJf dp = F(x,p, dp),

dx cJx cJp dx dx
an equation of the first order and first degree.
Solve p = F(x,p, : ) to obtain ¢(x,p,C) = 0.
Obtain the primitive by eliminating p between y = f(x,p) and ¢(x,p,C)= 0,

when possible, or express x and y separately as functions of the parameter p.
See Problems 4-7.
61
62 EQCATIONS OF FIRST ORDER AND HIGHER DEGREE
EQUATIONS SOLVABLE FOR x, i.e., x = f(y,p).

Differentiate with respect toy to obtain
dx = ! = cJf + di dp = F(y,p,
dp
dy)
dy P cy cJp dy
an equation of the first order and first degree.
Solve !p = F(y,p, ddp)

y
to obtain ¢(y,p,C) = 0.
Obtain the primitive by eliminating p between x = f(y,p) and ¢(y,p,C)= 0,

when possible, or express x and y separately as functions of the parameter p.
See Problems 8-10.
CLAIRAUT'S EQUATION. The differential equation of the form
y = px + f(p)
is called Clairaut's equation. Its primitive is
y = Cx + f(C)
and is obtained simply by replacing p by C in the given equation.

See Problems 11-16.
SOl..VED PROBLEMS
1. Solve / - (x + 2y T l)p 3 + (x + 2y + 2xy)/ - 2xyp = 0 or p(p -1) (p -x) (p - 2y) 0.
The solutions of the component equations of first order and first degree
dy = 0 dy = 1, dy dy
- - X = 0, dx - 2y = 0
dx ' dx dx
are respectively 2 2x
y-C = 0, y-x-C = 0, 2y-x -C = O, y-Ce = o.
2
The primitive of the given equation is (y-C)(y-x-C)(2y-x 2 -C)(y-Ce x) = 0.
2. Solve xyp2 + (x 2 +xy+y2)p + x 2 +xy = O or (xp + x + y) (yp + x) = 0.
The solutions of the component equations x dy + x + y = 0 and y dy + x = O

dx dx
2 2 2
are respectively 2xy + x - C = 0 and x + y - C = 0.
2 2 2
The primitive of the given equation is (2xy+x -C)(x +y -C) = 0.
2 2 2
3. Solve (x +x)p + (x +x-2xy-y)p + / - xy = 0 or [(x+l)p-y][xp+x-y] = 0.
dy dy
The solutions of the component equations (x+l)--Y = 0 and x- + X - y = 0
dx dx
are respectively y-C(x+l) = 0 and y + X ln Cx = 0.
The primitive of the given equation is [y - C (x + 1)] [y + X ln Cx] = o.

EQUATIONS OF FIRST ORDER AND HIGHER DEGREE
2
4. Solve or 2y = px - 16 :_ •
p2
dp 32x
Differentiating the latter form with respect to x, 2p p + X -
dx p2
Clearing of fractions and combining, p(p3 + 32x) - x(p3 + 32x)dp 0

dx
or 1) (p3 + 32x) (p - X dp) o.

dx
dx
This equation is satisfied when p 3 + 32x = O or
From the latter, dp p - x dp = O. and
p dx X
p ~ l<.x. When this replacement for pis made in the given equation, we have
2 ~ C y - C3x = 0,
2 2 2 4 2 2
16x + '2K x y - f.· 3 x = 0 or
after rep 1acing I<. by 2C,
The factor p 3 + 32x of 1) wil 1 not be considered here since it does not contain the de-
rivative dp/dx, Its significance will be noted in Chapter 10.
Differentiating with respect to x, p = 2x dp + 2p + 2p 4 x + 4p 3x 2 dp

dx dx
or (p + 2xdp)(l \- 2p 3x) o.
dx
3 From p + 2xdp = 0, xp 2 = C,
The factor 1 + 2p x is discarded as in Problem 4.
dx
2
In parametric form, we have x = C/p 2 , y = 2C/p + C , the second relation being obtained
by substituting x = Cjp 2 in the differential equation.
2 2
Here p may be eliminated without difficulty between the relation xp = C or p = C/x and
4 2
the given equation. The latter may be put in the form y - p x = 2px and squared to give
4 2 2 2 2 2 2 2
(y - p x ) = 4p x • Then, substituting for p , we have (y - C ) = 4Cx,
X
6. Solve x = yp + p
2
or y = p- p.
1 xdp_dp
Differentiating with respect to x, p or o.
P p 2 dx dx
dx __
P_,
(p 3 -p)dx + x + p2 X
Then 0 or +
dp dp p3 _ p p2 - 1
The latter is a linear equation for which e

fdpj(p 3 -p) ~ is an integrating fac-
p
tor, Using it,
-f dp - ln(p + ~ ) + C
p
/7=1
Cp
and - _P_ ln (p + ~ ) + y -p - 1 _ ln (p + /p2-l) +
__ C
/pG IT-I ~
6-l EQUATIONS OF FIRST ORDER AND HIGHER DEGREE
7. Solve y = (2+p)x + p
2
•
dx
Differentiating with respect to x, p = 2 + p + (x + 2p) dp or + ½x -p.
dx dp
This is a linear equation having e½fdp = e½P as an integrating factor,
½P 1.p ½P + 4e ½P + C
Then xe -f pe 2
dp -2pe
and y = 8 - p2 + (2+p)Ce-½P
8. Solve y = 3px + 6p2y2.
2
Solving for x, 3x = l :.. 6py • Then, differentiating with respect toy,
p
3 1 y dp 2 dp
- = - - - - - 6y - - 12py and
p p P2 dy dy
The second factor equated to zero yields py 2 = C. Solving for p and substituting in the
original differential equation yields the primitive y 3 = 3Cx + 6C 2 •
2
9. Solve p
3
- 2xyp + 4y
2
= 0 or 2x L + 4y.
y p
Differentiating with respect toy,

2 1 y dp dp 2
4(- - - -) or (p - 2y -) (2y o.
p p P2 dy dy
dp
Integrating p - 2y- O and eliminating p between the solution p2 = Ky and the original
dy
differential equation, we have 16y = /f(K - 2x/, This may be put in the form 2y = C(C -x/
by letting K = 2C.
10. Solve 4x = py(/- 3).

2
4 2 2 dp dy + 3p(p - l)dp
= p(p - 3) + 3y(p -1)- or o.
p dy y (p2 - 4) (p2 + 1)
9 9 3 2
Integrating, by partial fractions, lny + - ln(p+2) + - ln(p-2) + -ln(p + 1) ln C.
10 10 5
2
C X = - Cp(p _
_ _--...:...__.:.... 3) __,..
- _
Then y
4 (p2 _ 4//10 ( / + 1/5
CLAIRAUT'S EQUATION,
11. Solve y = px + li"7, The primitive is y = Cx + A+?.

2
12. Solve (y-px) =
Here y = px ± ~ -
The primitive is (y - Cx - ~ ) (y - Cx + ~ ) 0 or (y - Cx)

2
1 + c2 •
EQll\TIONS OF FIRST ORDER AKD HIGHER DEGREE
13. Solve y = 3px + 6/p

2
• (See Problem 8,)
This may be reduced to the form of a Clairaut equation,
2 2 ~ 2
Multiply the equation by y to obtain 3y px + 6y p •
2 dv 2 dv 2
Using the transformation y 3 = v, 3y p = dv , this becomes V = X lU + 3(cu) •
dx
y3
2
The primitive is or or = 3Cx + 6C •
14. Solve cos 2 y p 2 + sin x cos x cosy p - sin y cos 2 x = 0,
cosy du
The transformation sin y = u, sin x = v, p--= - reduces the equation to
cos x dv
du du 2 2 2
u = v-
dv + <d) • Then u = Cv + C or sin y = C sin X + C •
15. Solve (px - y) (py + x) = 2p.

2 2
The transformation J = U, X = V, p reduces the equation to
v du du du du
(-- or (v u) ( - + 1) 2 - •
u½ dv dv dv dv
2
du
du dv 2C 2 2C
Then u V - and u = Cv - or Cx -
dv du l+ C l+ C
1 +
dv
2
16. Solve /x(x -2) + p(2y -2xy-x + 2) + y + y = O.
The equation may be written as (y - px + 2p)(y - px + 1) = O.

Each of y = px - 2p and y = px - 1 is a Clairaut equation.
Thus the primitive is (y - Cx + 2C)(y - Cx + 1) = 0,
Find the primitive of each of the fol lowing,

2 2 2 2 -3
17, X p + xyp - 6y = 0 Ans, (y-Cx )(y-Cx ) = 0
2 2 2
18, xp + (y-1-x )p - x(y-1) 0 Ans. (2J- X + C) (xy- x + C) 0
2 2
19. xp 2yp + 4x 0 Ans. Cy = X + c2
20, 3x ~ p 2 - xp - y 0 Ans, xy = C ( 3 Cx - 1 )
2 2 2
21. Byp 2xp + y 0 Ans, y - Cx + 2C = 0
2 2 3 2
22. y p + 3px - y = 0 Ans, y - 3Cx - C = 0
66 EQUATIONS OF FIRST ORDER AND HIGHER DEGREE
2 2
23. p - xp + y = 0 Ans, y = Cx - C
3 2
24. 16y p - 4xp + y = 0 Ans. / = C(x - C)
5 ~ 2 3 2 2 3 2
25. xp - yp + (x + l)p - 2xyp + (x + y )p - y = 0 Ans. (y- Cx - C ) (C x - Cy + 1) = 0
26, xp
2
- yp - y = 0 Ans. X = C (p + 1) e1'' y = c/ e1)
2 3
(Use y = z.) Ans, /=2Cx+C
C/✓p
2
28. p - xp - y = 0 Ans. 3x = 2p + C/lp, 3y = / -
2
29, y=(l+p)x+p Ans. x = 2(1-p) + Ce-P, y = 2 - / + C(l+p)e-P
30. y = 2p + ~ Ans. x = 2 1n p + l n (p + ~ ) + C, y = 2p + ~
2 1/2 2 2 -5/~ 3/2 2 -5/~
31. yp - xp + 3y = 0 Ans. x = Cp (p + 3) (p + 2) , y = Cp (p + 2)
CHAPTER 10
Singular Solutioni',-Extraneous Loci
THE DIFFERENTIAL EQUATION

2
1) y = px + 2p
has as primitive the family of straight lines of equation

2
2) y = Cx + 2C •
With each point (x, y) in the region of points for which x 2 + 8y > 0, equation
1) associates a pair of distinct real directions and equation 2) associates
a pair of distinct real lines having the directions determined by 1). For ex-
ample, when the coordinates (-2, 4) are substituted in 1), we have 4 = -2p + 2p 2
or p 2 - p -2 = O and then p = 2,-1. Similarly, when 2) is used, we obtain C =
2, -1. Thus, through the point (-2, 4) pass the lines y = 2x + 8 and y c::. -x + 2
of the family 2) whose slopes are given by 1). Points for which x 2 + 8y < 0
yield distinct imaginary p- and C-roots.
1 ines of family
y " ex+ 2c 2
envelope
- x 2 + By = 0
(a) (b)
Through each point of the parabola x 2 + 8y = 0 there passes but one line
of the family, that is, the coordinates of any point on the parabola are so
related that for them the two C-roots of 2) and the two p-roots of 1) are e-
qual. For example, through the point (-8,-8) there passes but one line, y =
2x + 8, and through the point (4,-2) but one line, y = -x + 2. (See Fig. a. )
It is easily verified that the line of 2) through a point of x 2 + 8y = 0

is tangent to the parabola there, that is, the direction of the parabola at
any one of its points is given by 1). Thus, x 2 +By= O is a solution of 1).
It is called a singular solution since it cannot be obtained from 2) by a
choice of the arbitrary constant, that is, since it is not a particular solu-
tion. The corresponding curve, the parabola, is called an envelope of the fam-
ily of lines 2). (See Fig, b above.)
67
68 SINCllLAR SOLlITIONS - EXTRANEOUS LOCI
Surnnary and Ex tens ion:

A singular solution of a differential equation satisfies the differential
equation but is not a particular solution of the equation.
At each point of its locus (envelope) the number of distinct directions
given by the differential equation and the number of distinct curves given
by the corresponding primitive are fewer than at points off the locus.
THE SINGULAR SOLUTIONS of a differential equation are to be found by expressing

the conditions
a) that the differential equation (p-equation) have multiple roots, and
b) that the primitive (C-equation) have multiple roots.
In general, an equation of the first order does not have singular solu-
tions; if it is of the first degree it cannot have singular solutions. More-
over, an equation f(x, y, p) = O cannot have singular solutions if f(x, y, p) can
be resolved into factors which are linear in p and rational in x and y.
The simplest express ion, called the discriminant, involving the coeffi-
cients of an equation F(X) = O whose vanishing is the condition that the equa-
tion have multiple roots is obtained by eliminating X between F(X) = O and
F'(X) = 0. The discriminant of
2 2
aX + bX + c = 0 is b - 4ac,
2
of aX 3 + bX + cX + d = 0 is
See Problem 1.
For the example above, the discriminants of the p- and C-equations are iden-
tical, being x 2 + 8y.
If E(x,y) = 0 is a singular solution of the differential equation f(x,y,p)
= 0, whose primitive is g(x,y,C) = 0, then E(x,y) is a factor of both discrim-
inants. Each discriminant, however, may have other factors which give rise to
other loci associated with the primitive. Since the equations of these loci
generally do not satisfy the differential equation, they are called extraneous.
EX1RANEOUS LOCI. (Differential equation, f(x,y,p) o; primitive, g(x,y,C) = O.)
a) Tac Locus.
Let P be a point for which two or more of then distinct curves of the fam-
ily g(x,y,C) = 0 through it have a common tangent at P. Now the number of dis-
tinct directions at P is less than n so that the p-discriminant must vanish
there. The locus, if there is one, of all such points is called a tac locus.
If T(x, y) = 0 is the equation of the tac locus, then T(x,y) is a factor of the
p-discriminant. In general, T(x,y) is not a factor of the C-discriminant and
T(x,y) = 0 does not satisfy the differential equation.
0 X
y = O is a tac locus.
SINGl,LAR SOLUTIONS~ EXTRANEOUS LOCI 69
b) Nodal Locus.
Let one of the curves of the family through P have a node (a double point
with distinct tangents) there. Since two of the n values of p are thus ac-
counted for, there can be no more than n-1 distinct curves through P; hence,
the C-discriminant must vanish at P. The locus, if there is one, of all such
points is called a nodal locus. If N(x,y) = 0 is the equation of the nodal
locus, then N(x,y) is a factor of the C-discriminant. In general, N(x,y) is
not a factor of the p-discriminant and N(x,y) = O does not satisfy the dif-
ferential equation.
c) Cusp Locus.
Let one of the curves of the family through P have a cusp (a double point
with coincident tangents) there. Since one of the p-roots is of multiplicity
two, the p-discriminant must vanish at P. Moreover, as in the case of a node,
there can be no more than n-1 curves through P and the C-discriminant must
vanish at P. The locus, if there is one, of all such points is a cusp locus.
If C(x, y) = 0 is the equation of the cusp locus, then C(x, y) is a factor of
both the p- and C-discriminants. In general, C(x,y) = O does not satisfy the
differential equation.
y = O is a nodal locus. y = O is a cusp locus.
If the curves of the family g(x,y,C) = 0 are straight lines, there are no
extraneous loci.
If the curves of the family are conics, there can be neither a nodal nor
cusp locus.
THE p-DISCRIMINANT RELATION. The discriminant of the differential equation f(x,y,p)

= O, the p-discriminant, equated to zero includes as a factor
1) the equation of the envelope (singular solution) once. See Problems 2-4.
(The singular solution satisfies the differential equation.)
2) the equation of the cuspidal locus once. See Problem 7.
(The equation of the cuspidal locus does not satisfy the differential equa-
tion unless it is also a singular solution or particular solution.)
3) the equation of the tac locus twice. See Problem 5.
(The equation of the tac locus does not satisfy the differential equation
unless it is also a singular solution or particular solution.)
THE C-DISCRIMINANT RELATION. The discriminant of the primitive g(x,y,C) 0, the

C-discriminant, equated to zero includes as a factor
1) the equation of the envelope or singular solution once.
2) the equation of the cuspidal loct1.s three times.
70 SINGULAR SOLUTIONS - EXTRANEOUS LOCI
3) the equation of the nodal locus twice. See Problem 6.

(The equation of the nodal locus does not satisfy the differential equa-
tion unless it is also a singular solution or particular solution.)
When any locus falls in two of the categories, the multiplicity of its e-
quation in a discriminant relation is the sum of the multiplicities for each
category; thus, a cuspidal locus which is also an envelope is included twice
in the p-discriminant and four times in the C-discriminant relation.
The identification of extraneous loci is, however, more than a mere count-
ing of multiplicities of factors.
SOLVED PROBLEMS
1, Find the discriminant relation for each of the following:

3 3 2 2
a) p + px - y = O, b) p x - 2p y -16x = 0, c) y = C (x - C / •
Note, These discriminant relations may be written readily using the formula given above,
We give here a procedure which may be preferred,
cJf
a) We are to eliminate p between /(x,y,p) = p 3 +px-y 0 and o. This is best
cJp
done by eliminating p between
cJf
3/ - p cJf = 3p 3 + 3px - 3y - 3/ - px = 2px - 3y = 0 and o. Solving the first
cJp cJp
2
3 27y 3 2
for p = Y and substituting in the second, we find 3/ + x + x = O or 4x + 27y o.
2x 4X
2
Note. If j(x,y,p) = O is of degree n in p, we eliminate p between nf - p cJf = 0 and d/ = 0,

cJp cJp
cJf 3 2 2 3 2
b) We are to eliminate p between 3/ - p - = 3p x - 6p y - 48.x - 3p x + 4p y = -2p 2 y - 48.x
2
= O
cJp
and cJj = 3/ X - 4py = Q, From the latter we obtain 9p

4 2
X 16p y
2 2
or 9p
4 2
X -
2 2
16p y 0
cJp 2
and from the former / = -24 :_ , S u bs t 1. t ut mg
. f or p 2 , we o bt a1· n x 2 ( 2y 3 + 27x 4 ) = O.
y
c) Here g(x,y,C) = c3 - 2
2C x + Cx
2
- y = O and we are to eliminate C between
cJg 3 2 2 3 ,..2 2 2 2
1) 3g - C - = 3C - 6C x + 3Cx 3y - 3C + 4C x - Cx = -2C x + 2Cx - 3y O and
cJC
cJg 2 2
2) - = 3C - 4 fa + x = o.
cJC
2
Multiplying 1) by 3 and 2) by 2x, and adding, we have -2Cx + 2x 3 - 9y o.
2x3 - 9y
Substituting C = - - - in 2) and simplifying, we obtain y(4x 3 - 27y) o.
2x2
2
2. Solve y = 2xp -yp and examine for singular solutions.
Solving for 2x = r
p
+ yp and differentiating with respect toy, we have
SINGULAR SOLUTIONS - EXTRANEOUS LOCI 71
2 1 __Y_ dp dp
p
+p+y- or o.
p P2 dy dy
C
Integrating p + y dp = o to obtain py = C and substituting for p in the given dif-
dy y
2 2
ferential equation, we obtain the primitive y 2Cx - C •
2 2
The p- and C-discriminant relations are X - y = 0. Since both y = x and y -x satisfy
the given differential equation, they are singular solutions.
If p is eliminated between the differential equation and the relation / - 1 = O, discarded
in this solution, the equation of the envelope x 2 - y2 = o is again obtained. The presence
of such a factor implies the existence of a singular solution but not conversely. Hence, this
procedure is not to be used in finding singular solutions.
The primitive represents a family of parabolas with principal axis along the x-axis. Each
parabola is tangent to the line y = x at the point (C,C) and to the line y = -x at the point
(C, -C). See Figure (a) below.
y
C =-1 C =1
X 0 X
Fig. (a) Prob. 2 Fig. (b) Prob. 3
Family of parabolas y 2 2Cx- c2, Family of straight lines y = Cx + c 3,

envelope y = ± x. envelope 4x3 + 27y 2 o.
3
3. Examine p +px-y = o for singular solutions.
This is a Clairaut equation, the primitive being y = Cx + c 3•

The p- and C--discriminant relation 4x 3 + 27/ = o is a singular solution since it satis-
fies the differential equation.
The primitive represents a family of straight lines tangent to the semi-cubical parabola
4x 3 + 27y2 =
o, tl}e envelope. See Figure (b) above.
4. Examine 6//+3px-y = o for singular solutions.

3 2
From Problem 13, Chapter g, the primitive is y = 3Cx + 6C .
2 3
Both the p- and C-discriminant relations are 3x + By = O. Since this satisfies the dif-
ferential equation, it is a singular solution.
2 2 2
5. Solve (x - 4 )p - 2xyp - x = O and examine for singular solutions and extraneous 1 oci.
4 X
Solving for 2y = xp - _;P p
and differentiating with respect to x, we have
72 SINGCLAR SOLUTIONS - EXTRANEOUS LOCI
p + x dp + 4p 4 dp 1 + !__ dp 2 2 2 2 dp
2p or (p X - 4p +X ) (p - X - ) o.
dx x2 X dx p p2 dx dx
dp 2 2
From p - x - o, p = Cx and the primitive is C (x - 4) - 2Cy - 1 = o. The p--discrim-
dx
2 2 2
inant relation is x (x + / - 4) = O, and the C-discriminant relation is x + /- 4 = o.
2
Now x + / = 4 occurs once in the p- and C-discriminant relations and satisfies the dif-
ferential equation; it is a singular solution. Also x = O occurs twice in the p-discriminant
relation, does not occur in the C-discriminant relation, and does not satisfy the differen-
tial equation; it is a tac locus.
2 2
The primitive represents a family of parabolas having the circle X + y 4 as envelope.
See Figure (c) below.
Note 1. The two parabolas through a point P of the tac 1ocus x = O have at P a common tan-
gent.
Note 2. A curve of the family meets the envelope in the points (±----, - C-)
/4c 2
-1 1
hence,
C
only those parabolas given by C >
2
i touch the circle.
Family of parabolas Family of cubic curves

C2 (x 2 - 4) - 2Cy - 1 = o. (Y + C )2 x (X - 1 ) 2•
Fig. (c) Prob. 5 Fig. (d) Prob. 6
2 2
6. Solve 4xp - (3X -1) 0 and examine for singular solutions and extraneous loci.
3 1/2 3/2 1/2

Solving for p = +(- x - we obtain by integration y = ± (x - x ) + C1 or
- 2
(Y + C / = x (x - 1 / . The p-discriminant relation is X (3X - 1 i2 = Q, and the C-discriminant

relation is x(x-1/ =O.
Here x = O is common to the two relations and satisfies the differential equation, that is,
dx .
x = 0, dy = 0 satis f ies the equation when written rn the f orm 4x - (3x - 1)2dx2
(--) = o. It
dy
is a singular solution.
3x - 1 = O is a tac 1ocus since it occurs twice in the p-discriminant re 1at ion, does not
occur in the C-discriminant relation, and does not satisfy the differential equation.
x - 1 = O is a nodal locus since it occurs twice in the C-discriminant relation, does not
SfNGD,AH sou:noNS ~ EXTRANEOlTS LOCI
occur in the p~discriminant relation, and does not satisfy the diffrrential equation.
The primitive represents a family of cubics obtained by moving y ~ x (x -1 )
2 2
along the
y-axis. These curves are tangent to the y-axis and have a double point at x = 1. Moreover,
through each point on x = 1/3 pass two curves of the family having a common tangent Lherc.
See Figure (d) above.
7. Solve
2
9yp + 4 = o and examine for singular solutions and extraneous loci.
Solving for gy ,,. -4// and differentiating with respect to x, we have
8
dx and X + C =
Eliminating p between this latter relation and the differential equation, the primitive
3 2
is y + (X + C) = o.
The p-discriminant relation is y = o, and the (;-discriminant relation is y3 O. Since

y O occurs once in the p-discriminant relation, three times in the C-discriminant relation,
and does not satisfy the differential equation, it is a cusp locus.
3 2
The primitive represents the family of semi-cubical parabolas obtained by moving y + x = O
along the x-axis. Each curve has a cusp at its intersection with the x-axis, and y = O is the
locus of these cusps. See the figure below.
A Aloc~]\
IY
A
Family of semicubical parabolas
y3 + (x+C/ = o
3 2 2
8. Solve x p +- x YP + 1 = 0 and examine for singular solutions and extraneous loci.
Solving for y = and differentiating with respect to x, we have
3 2 dp
(1 - X p )(2p +XJ;) 0,
2
From 2p +x: = 0, px = C and, eliminating p between this and the differential eq~ation,
2
the primitive is C + Cxy +x = o.
2
The p-discriminant relation is x 3 (x/- 4) 0, and the C-discriminant re 1ation is x (xy - 4)
o.
x/- 4 = 0 satisfies the differential equation and is a singular solution.
x = 0 is a particular solution (C = 0). Note that it occurs three times in the p-dis-
criminant relation and once in the C-discriminant relation.
SINGULAR SOLUTIONS~ EXTRANEOUS LOCI
3
9. Examine p X -
2
2p y -16x
2
0 for singular solutions and extraneous loci.
2
c 3x
2
From Problem 4, Chapter 9, the primitive is - C y - 2 = o.
The p-discriminant relation is x 2 (2y 3 + 27x
4
) " o, and the C-discriminant relation is
2y 3 + 27x 4 = Q.
3 4
Since 2y + 27x = o is common to the discriminant relations and satisfies the differen-
tial equation, it is a singular solution. At each point of the line x = O, two parabolas of
the family are tangent there (for y < O, the parabolas are real). Thus, x = O is a tac locus.
Also, x = O is a particular solution. Since it is obtained by letting C -m, it is sometimes
called an infinite solution. Note however that when the primitive is written as x 2 -Ky-2K 3
= O, this solution is obtained when K = o.
Investigate for singular solutions and extraneous loci.
2 2 2
10. y = px -2p Ans. primitive, y = Cx - 2C ; singular solution, X By.
2 2 3 2 3 2
11. y p + 3xp -y o. Ans. prim., y +3Cx-C O; s.s.' 9x + 4y o.
2 2 2 2
12. x / - 2yp + 4x = o. Ans. prim., C x - Cy+ 1 = o; s.s.' y - 4x o.
2 2 2 2 2
13. Xp - 2yp + X + 2y = Q. Ans. prim., 2x + 2C (x - y) + C O; s.s •• X + 2xy -y o.
2 2 2
14. (3y-1) p 4y. Ans. prim., (x+ C ) 2 y(y -1) s. s.' y o; t. 1.. y 1/3;
n. l., y = 1.
4 2 2
15. y "-xp +x p • Ans. prim., xy = C + C 2 x; s.s.' 1 + 4X y o; t. 1., X = o.
2 3 2 2 3
16. 2y p + 4xp. Ans. prim., (4x + 3xy+C) = 2(2x +y) ; no s.s.;
2
c. 1., 2x + y = o.
2
17. Ans. prim., (x- C) = y3(1-y); s. s.' y 1; c. 1., y o;
t. 1 •• y = 3/4.
3 3 6
18. 4
p - 4x p + Bx y O. Ans. prim., y = Cx
2
- c3 ; s. s.' 4x - 27y
2
o; t. l., X " o.
2 2 2
19. (p +l)(x-y) (x+yp) Ans. t.}., y = X.
Hint: Use x p cos 8,
y = p sin 8.
CHAPTER 11
Applications of First Order and Higher Degree Equations
IN FINDING THE EQUATION of a curve having a given property, (for example, that its
slope at any point is twice the abscissa of the point), we obtained in Chap-
ter 7 a family of curves (y = x 2 + C) having the property. In this chapter
the family of curves will frequently be a family of straight lines. In such
cases, the curve in which we are most interested is the envelope of the family.
SOLVED PROBLEMS
1. Find the curve for which:

a) the sum of the intercepts of the tangent 1 ine on
the coordinate axes is equal to k.
b) the product of the intercepts of the tangent 1 ine
on the coordinate axes is equal to k.
c) the portion of the tangent 1 ine intercepted by
the coordinate axes is of constant length k.
Let the equation of the tangent line be

y = px + /(p) •
the x- and y-intercepts being -f(p)/p and /(p) re-
spectively. X
a) Since /(p) -f(p)/p = k, /(p) = -kp/(1-p), and

the equation of the tangent line is y "' px - _!!!_ •
1 -p
This is a Clairaut equation, the primitive being
2
the family of lines y " Cx - __!£ or xC - (x + y -k)C + y " O. The required curve, the en-
velope of the family, has equa~~~ (x + y - k/ = 4xy or x½ ± y½ = k½. Note that this curve is
an envelope (singular solution) since it satisfies the differential equation and cannot be ob-
tained from the primitive by assigning a value to C.
b) Since /(p)[-f(p)/p]=k, /(p)= ± ✓-kp, and the equationofthetangentlineisy

px ± ~ - This is a Clairaut equation, the primitive being
2 2
y-Cx = ± ~ or x C + (k-2.xy)C + y2 = O,
The required curve, the envelope of the family, has equation 4xy = k.
2
c) Since [{f(p)f + {-/(p)/p} ]½ = k, /(p) = ± kp/~.
and the equation of the tangent
line is y = px ± kp/✓ i:;}. The primitive of this equation is y = Cx ± kC//1+C 2•
2 312
Differentiating with respect to C, we have O = x ± k/( 1 + C ) •
2 312 2 12 3 2 312
Then x = + k/(1+ C ) , y = Cx ± kC/(1+ C / " ± kC /(l+C ) , and the equation of
the envelope is x2/3 + /13 = k2/3/(l+C2) + k2/3 C2/(l+C2) = k2/3.

76 APPLICATIONS-FIRST ORDER AND HIGHER DEGREE EQUATIONS
2. Find the curve for which:

a) the sum of the distances of the points (a,0) and (-a, 0) from the tangent line is equal to k.
6) the sum of the distances of the points (a,0) and (0,a) from the tangent line is equal to k.
Take px -y + f (p) -- 0 as th e norma 1 f orm o f th e equa t.10n of a t angen t 1.1ne.

11+7
(a)
a) The distances of the points (a,0) and (-a,0) from the line are ap + f (p) and -ap + f (p) re-
~ ~
2 2
spectively. Thus, /(p) = k, f(p) = ½k,ll+p , and the equation of the tangent line is y
~
px + ½k ~ - The primitive of this Clairaut equation is
2 2 2 2
y = Cx + ½k/1+C 2 or (4x -k )C 2 - 8xyC + 4y - k = 0,
2 2
The required curve, the envelope of this family of lines, has as equation x + y = tk 2 •
ap + f (p) -a+ f (p)
b) The distances of the points (a,0) and (0,a) from the line are--'----'-~ and re-
~ ~
-a+ap+2f(p) = k
spectively. Thus, /(p) = ; \ - [ k ~ - ap + a]. and the equation of the
11+7 '
tangent line is y = px + ½[k ~ - ap + a]. The primitive is y = Cx+ b [k /i:c2 -aC +a].
Differentiating with respect to C, we have 0 = x + ~ [kC//1+ C 2 - a].
Then x = - ½[kC//2 + C 2 - a], y = ½[k//'l+"c2 + a], and the envelope of the family of 1 ines
2 2 2
has equation x2-+y -ax-ay = ti(k -2a ),
3. Find the curve such that the tangent line at any of

its points P bisects the angle between the ordinate
at P and the line joining P and the origin,
Let 8 be the angle of inclination of a tangent
line and¢ be the angle of inclination of OP. Then,
if Mis the foot of the ordinate through P,
angle OPM = 90° - ¢ = 2(90° - 8) = 180° - 28.
Now tan(90° - ¢) = cot ¢=tan( 180° - 28) ~ -tan 28
X
and tan¢ tan 28 = - 1.
0
Since tan¢= y/x and tan 8 =y' = p, we obtain the
APPLICATIONS~FIRST ORDER AND HIGHER DEGREE EQUATIONS 77
differential equation of the curve l · --3.E._ -1 or 2y = xp - x/p, Differentiating with

X l -p2
1 x dp 2 dp
respect to x, 2p =p + (x + -)- , p(p + 1) = x(p 2 + 1)-, and X dp - p dx = o.
p p2dx dx
Integrating, ln p = ln x + ln C or p = Cx. substituting for pin the differential equa-
tion, we obtain the family of parabolas C2x 2 - 2Cy - 1 = O.
4. Find the shape of a reflector such that

1 ight y
coming from a fixed source is reflected in
parallel rays.
Let the fixed point be at the origin of
coordinates and the reflected rays be parallel T'
to the x-axis. The reflector is then a sur-
face of revolution generated by revolving a
curve f(x,y) = 0 about the x-axis. Q
Confining ourselves to the xOy plane, let
P(x,y) be a point on the curve f (x,y) = 0,
X
TPT' be the tangent 1 ine at P, and PQ be the
reflected ray. Since the angle of incidence T
is equal to the angle of reflection, it fol-
lows that LOPT = ¢ = LQPT'.
Now p = dy = tan L OTP = tan ¢ and tan L TOP tan(rt - 2¢) = - tan 2¢ -2 tan¢"" _ l.
dx X
y _2p y
hence, or 2x = - - yp.
X l-p2 p

2 ~ - J__ dp - p - y dp and dp - dy. Then, p
C
p p p2 dy dy P y y
Eliminating p between this relation and the original differential equation, we have the family
of curves y 2 = 2Cx + C 2 • Thus, the reflector is a member of the family of paraboloids of
revolution y 2 + z 2 = 2Cx + C 2 •
5. Find the curve for which each of its tangent lines forms with the coordinate axes a triangle
of constant area a 2 • Ans. 2xy = a 2
6. Find the curve for which the product of the distances of the points (a,O) and (-a,O) from the
tangent lines is equal to k, Ans. kx 2 = (k+a 2 )(k-y 2 )
7. Find the curve for which the projection upon the y-axis of the perpendicular from the origin
upon any tangent is equal to k. Ans. x 2 = 4k(k -y)
8. Find the curve such that the origin bisects the portion of the y-axis intercepted by the tan-
gent and normal at each of its points. Ans. x 2 + 2Cy = C 2
9. Find the curves for which the distance of the tangent from the origin varies as the distance
of the origin from the point of contact.
p
2 fJ l1-k2
Hint: kp. Ans. p = Ce k
/4 2
+ (dp/d 0)2
CHAPTER 12
Linear Equations of Order n
A LINEAR DIFFERENTIAL EQUATION of order n has the form
dy
1) + '"'"' + pn-1 + Pny Q,
dx
where P0 -p 0, P 1 , P2 , • • •, • , Pn. Q are functions of x or constants.
If Q = 0, 1) has the form
2) Po
dny
+ p1
dn-ly
n-l
+ p2
dn-2y
+ ....... + Pn-1
dy
dx
+ Pny = 0
dxn dx dxn-2
and is called homogeneous to indicate that all of the terms are of the same
(first) degree in y and its derivatives.
d3y d2y 5 dy
Examples. A) x3 + 2x
2
- - xy = sin x, of order 3.
dx3 dx dx
2
d y 3 dy
B)
2 - dx
+ 2y 0, of order 2,
dx
Equation B) is an example of a homogeneous linear equation.
SOLUTIONS. If y = y 1 (x) is a solution of 2), then y = C 1 y 1 (x), where C 1 is an ar-

bitrary constant, is also a solution. If y=y 1 (x), y=y 2 (x), y=y 3 (x), .....
are solutions of 2), then y = C1 y 1 (x) + C2 y 2 (x) + C3 y 3 (x) + •• •• •• is also a
solution.
A set of solutions y=y 1 (x), y=y 2 (x), •··•, y=yn(x) of 2) is said to be
linearly independent if the equality
" " ' " · + CnYn = 0,
where the e's are constants, holds only when c 1 = c 2 = c 3 = • • • • = en = 0.
Example 1. The functions ex and e-x are linearly independent. To show· this,
0, where c 1 and c 2 are constants, and differentiate to
obtain c 1 ex - c 2 e-x = 0. When the two relations are solved simultaneously for
c 1 and c 2 , we find c. = c 2 = 0.
Example 2. The functions ex, 2ex, and e-x are linearly dependent, since
c 1 ex + 2c 2 ex + c 3 e-x = 0 when c 1 = 2, c 2 = -1, C3 = 0.
A necessary and sufficient condition that the set of n solutions be lin-

early independent is that:
78
LINEAR EQUATIONS OF ORDER n 79
Yn
I
Yn
w = fl
Yn
-/ o.
(n-1)
Yn
If y=y 1 (x), y=y 2 (x), •··· •··, y=yn(x) are n linearly independent solu-
tions of 2), then
3)
is the primitive of 2).
If y = R(x) is a particular solution, also called particular integral, of

1), then
4)
is the primitive of 1). Note that 4) contains all of 3). This part of 4) is
called the complementary function. Thus the primitive of 1) consists of the
sum of the complementary function and a particular integral.
Attention has been called to the fact that the primitive of a differential
equation is not necessarily the complete solution of the equation. However,
when the equation is linear, the primitive is its complete solution. Thus 3)
and 4) may be called complete solutions of 2) and 1) respectively.
LINEAR DIFFERENTIAL EQUATIONS with constant coefficients (equation B) above) will

be treated in Chapters 13-16. Those with variable coefficients (equation A)
above) will be considered in Chapters 17-19.
SOLVED PROBLEMS
2
d d
1. Show that the equation _J_ - ...1.. - 2y = O has two distinct solutions of the form y = eax.
dx2 dx
ax
e • for some value of a, is a solution then the given equation is satisfied when the
replacements y e
ax dy - ae ax 2 ax
are made in it,
= a e
dx
d2y - dy - 2y = ax 2
We obtain e (a - a - 2) O which is satisfied when a= -1, 2.
dx2 dx
X 2X
Thus y = e- and y = e are solutions.
is the primitive of the equation of Problem 1,
Substituting for y and its derivatives in the differential equation, it is readily checked
2
that y = C1 e-x + C2 e x is a solution. To show that it is the primitive, we note first that
the number (2) of arbitrary constants and the order (2) of the equation agree and second that
80 LINEAR EQUATIONS OF ORDER n
-x 2x
e e
2
since w y e x are linearly independent.
-x 2x
-e 2e
3
3. Show that the differential equation x3 d y - 6x dy + 12y 0 has three 1 inearly inde-
dx3 dx
pendent solutions of the form y = xr.
After making the replacements

d2y r-2 r-3
r(r-l)x , r ( r - 1) ( r - 2) x
dx2
in the left member of the given equation, we have xr (r 3 - 3/ - 4r + 12) = O which is satisfied
2 3 -2
when r 2, 3,-2. The corresponding solutions y = x , y = x , y = x are linearly independent
2 -2
X x3 X
-2x -3
2
since w 2x 3x 20 I- o. The primitive is y
-4
2 6x 6x
2
d y
4. Verify that y = - sin x is a particular integral of - d.1'. - 2y cos x + 3 sin x and
dx2 dx
write the primitive.
Substituting for y and its derivatives in the differential equation, it is found that the
equation is satisfied. From Problem 2, the complementary function is y = C1 e-x + C2 e 2x.
Hence, the primitive is y = C1 e-x + C2 e2x - sin x.
5. Verify that y = lnx is a particular integral of x3 d3y - 6x ddxy + 12y 12 lnx - 4 and
dx3
Substituting for y and its derivatives in the given equation, it is found that the equation
2 3 -2
is satisfied. From Problem 3, the complementary function is y = C1 x + C2 x + C3 x •
. ·t.1ve 1s
. 2 3 2
Hence, th e pr1m1 y = C1 x + C2 x + C3 x - + 1n x.
2
d4y - d3y dy
6. Show that - 3~ + 5 - 2y 0 has only two linearly independent sol u-
d.x4 dx3 dx2 dx
ax
tions of the form y e
4 2
Substituting for y and its derivatives in the given equation, we have e ax ( a - a 3- 3a + 5a - 2) 0
which is satisfied when a= 1, 1, 1,-2.
X X X -2x
e e e e
X -2x X X X -2e-2X
e e e e e
Since I- 0, but 0, the linearly independent
X -2x X X X -2x
e - 2e e e e 4e
X X X -2x
e e e -8e
x; -2x;
solutions are y e and y e
LINEAR EQLATIONS OF ORDER 11 81
2 2
7, Verify that y = ex, y = xex, y = x ex, and y = e- x are four linearly independent solutions
of the equation of Problem 6 and write the primitive.
2
By Problem 6, y = ex and y = e- x are solutions. By direct substitution in the given equa-
tion it is found that the others are solutions.
X X 2 X -2x
e xe X e e
X X X 2 X X -2x 1 0 0 1
e xe + e X e + 2xe -2e X 1 1 0 -2
Since w e
1 2 2 4
-54ex f 0,
X X 2 X -2x
e xe + 2ex X e + 4xex + 2ex 4e 1 3 6 -8
X X 2 X -2x
e xe + 3ex X e + 6xex + 6ex -Be
these solutions are linearly independent and the primitive is

X X 2 X -2X
y = C1 e + C2 xe + C3 x e + C4 e •
2 2 d 2y d
8. Verify that y = e - x cos 3x and y e- xsin 3x are solutions of + 4 ...l. + 13y 0 and
dx2 dx
Substituting for y and its derivatives, it is found that the equation is satisfied.
Since W= 3e-~x, 0, the solutions are linearly independent.
Hence, the primitive is y = e-2x(C 1 cos 3x + C2 sin 3x).
9. Show that each of the following sets of functions are linearly independent.
2 2
a) sin ax, cos ax C) 1, X, x e) lnx, xlnx, x lnx
b) eax sin bx, eax cos bx d) eax, ebx, ecx (a lb le)
Form the differential equation having the given primitive.

2X -3x
10. y = C1 e + C2e Ans. y" + y' - 6y = 0
2X 2X 2 2X
11. y = C1 e + C2 xe + C3 x e y'" - 6y 11 + 12y' Sy 0
X 2X 5X
12. y = C1 e + C2e + e /12 y fl - 3y , + 2y = e 5x
13, y = C1 cos 3x + C2 sin 3x + (4x cosx + sin x)/32 y fl + 9y = X COS X
x 3y'" + xy' - y = 3x4-
2
y" + (y') + 1 = 0
CHAPTER 13
Homogeneous Linear Equations with Constant Coefficients
THE HOMOGENEOUS LINEAR EQUATION with constant coefficients has the form
1)
in which P0 I= 0, P1 , P2 , • • • • ·, Pn are constants.
By a convenient change of notation, writing dy = Dy, D•Dy

2 dx
=Dy, etc., 1) becomes
d
Now D = - is an operator which acts on y, and
dx
is simply a much more complex operator. However, we shall find it very con-
venient to consider 3) at times as a polynomial in the variable D and to de-
note it by F(D). Thus, 1) may be written briefly as
4) F(D)y = 0.
It can be shown in general and will be indicated by an example that when 3)
is treated as a polynomial and factored as
5) F(D) = P0 (D-m 1 )(D-m 2 )(D-m 3 )""""'(D-mn_ 1 )(D-mn),
then
6) F(D)y = P0 (D-mi)(D-m 2 )(D-m 3 )""""'(D-mn_ 1 )(D-mn)Y = 0
remains valid, i.e., is equivalent to 1) when Dis treated as an operator.
d 3y
2
. d y d 3 2
EXAMPLE. In the D notation - - - - 4 2 + 4y = O becomes (D - D - 4D + 4)y = O
c1x3 dx 2 dx
and, in factored form, (D - 1) (D - 2) (D + 2) y = O, Now
d
(D - 1) (D - 2) (D + 2) y (D - 1) (D - 2) ( d.x + 2) y (D-l)(D-2)(dy + 2y)
dx
d dy
(D - 1) { - ( - + 2y) 2(dy + 2y)}
dx dx dx
2
l(~ - 4y)
dx2
4 dy + 4y = 0,
dx
In Problem 1 below, it will be indicated that the order of the factors here is immaterial,
82
HOl\IOCEf\EOrs Lll\EAR E()l ATIONS WITH CO:\ST,\NT COEFFICIENTS 8:{
THE EQUATION
is sometimes called the characteristic equation of 1) and the roots m1 , m2 ,
m3 ,• • • • , mn are called the character is tic roots. Note that it is never nec-
essary to write the characteristic equation since its roots can be read di-
rectly from 6).
TO OBTAIN THE PRIMITIVE of 1) we first write the equation in the form 6).
involving n linearly independent solutions of 1) with n arbitrary constants,

is the primitive.
Thus in the example above, where 4 cly + 4y = 0 or

clx 3 clx
2 dx
(D -1) (D - 2) (D + 2)y = 0, the characteristic roots are 1,2,-2 and the primi-
See also Problems 5-7.
is the primitive.
In general, to a root m occurring r times there corresponds
C1emx + C2xemx + C3x2emx + ...... + C,,,.xr-1 e1t,
in the primitive.
cl 3y 2
Thus to solve - 2 cl y - 4 cly + By = 0, write the equations as
clx3 clx 2 clx
3 2 2
(D - 2D - 4D + 8)y = (D -2) (D + 2)y = 0. The characteristic roots are 2, 2, -2
c) If the coefficients of 1) are real and if a+bi is a complex root of 6), so

also is a-bi. The corresponding terms in the primitive are
Peax sin(bx + Q) = Peax cos(bx + R),

where A,B,C 1 ,C 2 ,P,Q,R are arbitrary constants.
2
2
Thus the characteristic roots of cl Y - 4 dy + 5y = 0 or (D - 4D+5)y=0
dx2 cJx
are 2±i. Here a=2, b=l and the primitive is y = e 2x(C 1cos x + C2 sinx).

84 HOMOGENEOUS LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS
SOLVED PROBLEMS
1. Show that (D-a)(D-b)(D-c)y = (D-b)(D-c)(D-a)y.
2
dy d d
(D -a) (D - b) (D -c)y = (D-a)(D-b)(- - cy) (D-a)(-2- (b+c)..l + bey)
dx dx2 dx
3 2
d y _ (a+b+c/ Y + (ab+bc+ac/Y - abcy.
dx3 dx2 dx
2
dy d d
(D-b){D-c)(D-a)y (D-b)(D-c)(- - ay) {D-b)(-1. - (a+c).1'. + acy)
dx dx2 dx
3 2
d y _ d y dy
(a+b+c)- + (ab+ac+bc)- - abcy.
dx3 dx2 dx
We are to show that
(D-a)(D-b)(D-c)C1 eax = (D-b)(D-c)(D-a)C1 eax = (D-b)(D-c)0 = 0, and similarly for

the other two terms.
This follows since: a) (D-m/c1 /"x = (D-m/(D-m)C 1 emx = (D-m/o = 0,

mx 3 2 mx
b) (D-m) Cixe- = {D-m) C2 e = (D-m)0 = 0, and
c) (D-m/C:J/emx = 2(D-m/C3 xemx = 2(D-m)0 = 0.
4, Find the primitive of (D -m/ y = 0 (a) by assuming a solution of the form y = xr emx and (b)
by solving the equivalent pair of equations (D-m)y = v, (D-m)v = 0.
O when r = O, 1.
Thus the equation has two linearly independent solutions y = emx and y = xemx
The primitive is y = C1 emx + C2 xemx.
b) If we write (D-m)y = v, then (D-m/y (D -m) (D-m)y = (D-m)v = 0.
dy
Solving (D-m)v = 0, we obtain v = C2 emx. Since (D - m)y - - my is linear of
dx
the first order, its solution by the method of Chapter 6 is
or
DISTINCT REAL ROOTS.

2
d y dy
+ - - 6y o.
dx2 dx
2
We write the equation as (D +D-6)y = (D-2)(D+3)y = 0.
The characteristic roots are 2,-3, and the primitive is y = c1 /x + C2 e- x.

3
H()l\IOGE:\'Em;s UJ\EAR EQCATIONS WITH CONSTANT COEFFICIENTS 8S
3 2
6. Solve J y - d y - 12 dy o.
dx3 dx2 dx
3 2
~e write the equation as (D -D -12D)y = o or D(D-4)(D+3)y = 0,
The characteristic roots are 0,4,-3, and the primitive is y = C1 + c2 /x + C3 e- 3x.
2
d3y dy
7. Solve + 2 <!:.J_ - 5 - 6y = o.
dx) dx2 dx
3 2
We write the equation as (D + 2D - 5D- 6) y or (D - 2) (D + 1) (D + 3) y = o.
The characteristic roots are 2,-1,-3, and the primitive is y = C1 e2x + C2 e-x + C3 e- x,
3
REPEATED ROOTS,
3 2
8. Solve (D - 3/J + 3D - 1) y = o or
4 3
9. Solve (/) + 6D + 51/ -24/) - 36)y = 0 or (D - 2) (D + 2) (D + 3/ y = o.
4 3 2 3
10. Solve (D - D - 9D - 11 D - 4) y = o or (D+1) {D-4)y = o.
COMPLEX ROOTS,
2
11. Solve (D -2D + lO)y = O.
The characteristic roots are 1 i 3i, and the primitive is

y = ex(C1 cos 3x + C2 sin 3x) or e:i/;sin(3x+ C4 ) or C3 /cos(3x+ C6 ),
3
~ 0
2
12. SolH' (1) + 4D)y or /J(D + 4)y = O,
The characteristic roots are 0,±2i, and the primitive is y = C1 + C2 cos 2x + C3 sin 2x.
3 2 2
13, So 1ve
4
(JJ + D + '21/ - /J + 3) y = O or (D +'2D+3)(D -D+l)y = 0,
The characteristic roots are -1 ± i 12, ~ ± ~ i ff, and the primitive is

lx
2
y = e-x(C1 cos v'2x + C2 sinv'2x) + e (C3 cos~V3x + C4 sin½V3x),
4 2 2 2
14. Solve (D + :'JD - 36)y = O or (D - 4) (D + 9) y = O.
The characteristic roots are ±2, ±3i, and the primitive is

2 2
y Ae x + Be- x + C3 cos 3x + C4 sin 3x
C1 cosh 2x + C2 sinh 2x + C3 cos 3x + C4 sin 3x

2 2
since cosh 2x = !z(/x + e- x) and sinh 2x = ½(/x - e- x),
2 2
15. Solve (D - '2D + 5) y = o. The characteristic roots are 1 ±2i, 1 ±2i, and the primitive is
y ex (C1 cos 2x + C2 sin 2x) + xex (C3 cos 2x + C4 sin 2x)
ex { (C 1 + C3 x) cos 2x + (C2 + C4x) sin 2x} ,
86 H07\10GENEOUS LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS
Solve.
2
16. (D + 2D - 15)y = o Ans.
3 2
17, (D + D - 2D)y = O
2
18. (D + 6D + 9) y = O
2
19, (D~ - 6D 3 + 12D - 8D)y = o
e 2x ( C1 C sin
. 3x)
2
20. (D - 4D + 13)y = o y = cos 3x + 2
21. (D 2 + 25)Y = O y = C1 cos 5x + C2 sin 5x
22. (D
3 - D2 + 9D - 9)y = O
2
23. (D" + 4D )y = O
4 3 2
24. (D - 6D + 13D - 12D + 4)Y = o
6 2
25. (D + 9D~ + 24D + 16)y = O y = C1 cos x + C 2 sin x + (C3 + C4 x)cos 2x
+ (C5 + C8 x)sin 2x
CHAPTER 14
Linear Equations with Constant Coefficients
THE PRIMITIVE OF
where P0 -/: 0, P 1 , P2 , • • • • •, Pn are constants and Q = Q(x) -/: 0, is the sum of

the complementary function (primitive of F(D)y = O obtained in the preceding
chapter) and any particular integral of 1). (See Chapter 12.)
At times a particular integral may be found by inspection. For example,
2
y = 1x is a particular integral of (D 3 - 3D 2 + 2)y =x, since D 3 y =D y =0. Such
equations occur infrequently, however, and we proceed to consider in this chap-
ter two general procedures for obtaining a particular integral. Other pro-
cedures will be given in the next two chapters.
1
In each of the procedures below, use will be made of an operator - - de-
F(D)
1
fined by the relation --·F(D)y = y. When the operator is applied to 1) we
F(D)
obtain
1
--,F(D)y = y = _1_ Q
F(D) F(D)
or
1 1 1 _l_Q.
2) y =
D-mn
FIRST METHOD. This consists of solving a succession of linear differential equa-

tions of order one, as follows:
SET SOLVE TO OBTAIN
_l_Q du
u = - - fflnU = Q
D - Inn dx
1 dv
V = u dx - TTln-1 V = U
D - Inn-1
1
y = --w
D -m 1
As is indicated in Problem 3 below, the following formula may be established:
See Problems 1-6.
87
88 LIJ\EAR EQlIATIONS WITH CONSTANT COEFFICIENTS
1
SECOND METHOD. This consists of expressing as the sum of n partial fractions:
F(D)
N'1
+ + .......... + Then
See Problems 4-5.

In evaluating both A) and B), it is customary to discard tl1e constants of
integration as they appear; otherwise, one obtains the primitive rather than
a particular integral of the differential equation. The complementary func-
tion is then obtained by inspection and added to the particular solution to
form the primitive.
THE FOLLOWING FORMULAS will be found useful.
ibx
e cos bx + i sin bx e-ibx = cos bx - isinbx
ibx -ibx ibx -ibx
e - e e + e
sin bx cos bx
2i 2
ebx = cosh bx + sinh bx e -bx = cosh bx - sinh bx
. h bx
sin = I bx - bx
2 (e - e )
SOLVED PRORLE:\IS
2 X X
1. Solve (D - 3D + 2) y = e or (D -1) (D-2)y e •
1
The complementary function is y = C1 ex + c 2 / ~ and a particular integral is y = - - • _l_ ex.
D-1 D-2
Let u = -
D -2
1
- ex Then (D -2)u = e
X
or
du
dx
2u e
X
. ue
-2x
= Je X e-2X dx = Je-x dx = -e
-x
.
and u =-ex.
1 dy -x X
Now y=--u, (D - l)y = u or - y = -e and y e XIX
-e e-x dx -Xe
D -1 dx
The primitive is
3 2 -2x -2x
2. Solve (D + 3D - 4 )y = xe or xe
2 2
The complementary function is y C1 ex + C2 e- x + C3 xe- x, and a particular integral is
1 1 1 -2x
y --•--•--Xe
D-1 D+2 D+2
Let u = - - x e
1 -2x
Then
du
+ 2u = xe
-2x
and u = e-2x Jxe -2x ,e 2x dx =
1
- X e
2 -2x
D 1-2 dx 2
Let v
1
= -- u. Then
dv
+ 2v
1
- X e
2 -2x
and V e
-2x J -1 x 2 e -2x •e 2x dx 1 3 -2X
- X e
D+2 dx 2 2 6
LINEAR EQl ATIO~S \\ ITH CCL~STANT COEFFICIENTS 89
dy 1 3 -2x XJ 1 3 -2X -X 1 XJ x 3 e -3X dx

i'<ow .Y --//. Then dx - y - X e and y e - x e •e dx =· -e
D -1 6 6 6
1 -2x 3 2 2 2
e (x+x+-x+-).
18 3 9
2
The primitivP is y = C1 ey + C2 e- 2x + C3 xe- 2 x - _!._(x 3 + x 2 )e- x, the remaining terms of the
18
particular integral being absorbed by the complementary function.
J• Find a particular integral of {D-a) (D- b)y = Q.
A particular integral is given by y = _1_. _1_ Q.

D-a D-b
e bx f Qe-bx dx.
_1_ () u. du
7
Then - bu Q and u =
/J - b , dx
Now y O
1
--u. Then dy
dx - ay ~ u •·~ e bx f Qe -bx dx and
!) - a
y ~
5x 5X
4. Solve o/ - 3/). 2>.Y e or (D - 1) (D - 2) y = e
1 1 ',x
Th<' compkmentary funf'tion is y ~(\ex ... r:2 /x. and a particular integral is y -- •-- e
D-1 D-2
1 1 5x J (2-l)x J 5X -2x
First Method. y -----
D -1!) - 2
e e
x
e e •e (dx/
1
e"fe'fe 3x (dx/ ex J e' -1 e
3x
dx - ex J /X dx
3
1
12
e
5X
3
1
Second Method. y
(D -1) (D - 2)
e
5X
(- -1
D -1
1
+ --)e
D-2
5X
-ex J e
5x
•e
-x
dx + e
2x f e5x • e-2x cl x
1 X ~x 1 2X 3X 1 5X
4
e e + -3 e e
12
e
X 2x 1 5x
The primitive is y C1e + C2 e + e
12
2
5. Solve (D + 5D + 4)y = 3 - 2x or (D + 1) (D + 4) y = 3 - 2x.
The complementary function is y
y 1 (3-2x).
(D + 1) (D + 4)
1 1
First Method. y - - • - - (3 -2x)
D+l D+4
e-x f e-3X (-e

3 ~x 1 l\X + _1 el\X ) dx
- -xe e
-X f (-e
7 X 1 X
- -xe )dx
4 2 8 8 2
90 LINEAR EQUATIONS WlTH CONSTANT COEFFICIENTS
1 1 ( 1/3 1/3
Second Method. y --•-(3-2.x) -)(3-2.x)
D+l D+4 D+l D+4
1 e -~x J (3 -2.x)e~x dx
3
11 1
- -x.
8 2
The primitive is
3 _ 2 2 2X
6. Solve (D - ;:;D + 8D - 4) y = e
2x
or (D - 1) (D - ~) y = e
The complementary function is y C1 ex + C2 e 2x + C3 xe 2x, and a particular integral is

2x
y e
1 1 1 2x
y --•--•-- e
D-1 D-2 D-2
ex Je(2-1)x Je(2-2)x J/x e-2x(dx)3
e xr·""JJ
.Je (dx) ~ = eX Je Jx(dx) 2
X
3e
1 "J x 2 e X j
lX
The primitive is y = C1e.,. - C2 /x + C3 x/x + ¼x /X, the remaining terms of the particular
2
integral being absorbed in the complementary function.
2
/• Solve (/) , CJ I y c X COS X•
The complementary fund ion is y ° C1 cos 3x + C2 sin 3x. and a particular integral is
1
X COS X e-3ix J e(3i + 3i)x J x cos x e-3h (dx)2.
/)2 + g
I ~;X -i'X:
It will be simpler here to use COS X = 2· ( e + e ), so that:
1 -2 ix 1 -'I ix
+ -e + -e )dx
4 16
1
- e -3 J (1 ixe 11;, 1 4, iJ 1 2ix 1 2ix dx
2 2
+ -e
4
+ - ixe
4
+ 16e )
1 -3 ·:, 1 11;, 1 4-{x 1 ~ix 1 2ix 1 2ix 1 2ix

e (sxe - le - - ie + -xe + - Le -ie
2 32 16 8 16 32
1 ; - ;., 1 ix -ix
-x(e· - e -L(e -e )
16 64
1 1 . 1 1 .
-X COS X - sin x. The primitive is y = C1 cos 3x + C2 sin 3x + x cos x + s1n x,
8 32 8 32
LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS 91
2
8. Solve (D + 4) y = 2 cos x cos 3x = cos 2x + cos 4x.
The complementary function is y = C1 cos 2x + C2 sin 2x, and a particular integral is

1 1 . 1 1
y - - - ( cos 2x + cos 4x) -i(-- - --){cos 2x + cos 4x)
2
D +4 4 D + 2i D- 2i
-2ixf -2ix }
(cos 2x + cos 4x)e 2ix dx
. { 2· ·
1
;p e e tx f(cos 2x + cos 4x)e dx
1 { -2ix 1 2ix [ 1 2ix -2ix 1 ~ix -~ix ] dx

;p e Je 2(e +e ) + 2(e +e )
-e 2ix J e -2ix [ 2 1 (
e 2ix +e-2ix )+2e
1 ( ~ix
+e -~ix)] dx }
1 { -2ixf ~ ix b ix -2 ix 2ixf -~ix 2ix -bix . dx}

- e (e + 1+ e +e )i dx e (l+e +e +e )t
8
1 { -2 ix 1 ~ix . 1 b ix 1 -2 ix
- e (- e + ix + - e - - e
8 4 6 2
1 . 2ix -2ix 1( 2ix -2ix 1 ~ ix -~ ix }

{-ix(e -e )+ e +e ) -(e + e )
8 4 3
1 e ~ ix + e -~ ix
e 2 ix _ e -2 ix 2ix -2ix
1 1 e + e
-x(-----) +
4 2i 16 ( 2 12( 2
24 x sin 2x + 216 cos 2x - 212 cos 4x '

1
The primitive is y C1 cos 2x + C2 sin 2x + 2x sin 2x - cos 4x,
4 12
-3x
2 e
9, Solve (D - 9D + 18)y e
3X bx
The complementary function is y = C1 e + C2 e , and a particular integral is
1 -)X
y -----e
e-)X
e bx J e-3x J e e ,e-3x ( dx ) 2
{D-6)(D-3)
-3X -3X -3X

e bx 5e-3x (--e
1 e )dx 1 bx5e
-e e ( -e -3x)dx 1 e
-e •e
bx
3 3 9
3x 1 e - 3x bx
The complete.solution is y = C1 e + (C2 + -e )e •
9
-3x
Note, When the factors are reversed, a particular integral is y e 3x J e~x Jee •e'-bx (dx)2
Using the substitution e- 3x = v, we obtain
y = - 1515
- e V v (dv) 2 = -
1 5e V 1
(- - -)dv
1 1
-e
V
9v v2 9v v 2 v 2
9v
1 e - 3x bx
or y - e •e as before.
9
<)2 LINEAR EQUATIONS WITH CONSTA~T COEFFICIENTS
10. Evaluate, omitting the arbitrary constant.
1 X 1 X 1 2 2
a) - - e Ans. -e d) - - ( x + 1) Ans. X - 2x ~ 3
D+1 2 D~1
1 X X 1 1 .
b) - - e Ans. xe e) sin 3x Ans. --(2 Rm 3x - 3 cos 3x)
D -1 D +2 13
1 1 -2x 1 -2:>:
c) - - ( x + 1) Ans. X fl -- e sin 3x Ans. - - e cos 3x
D+1 D +2 3
solve,
2
11. (D - 4D + 3)y Ans.
2
12, (D - 4D)y = 5
2
13. (D 3 - 4D )y 5
5
14. (D - 4D 3 )y 5
15. (D 3 - 4D)y = x
2 2x
16. (D - 6D + 9)y e
2 2
17. (D + D - 2)y = 2(1 + x - x )
X -x 2
y = C1 e + C2 e + ex (x - X)
2 2 X -x 1 + --1:._ cos 2x
19. (D - l)y sin x = ½O- cos 2x) y = C1 e + C2 e - 2 10
-x
{D 2 -l)y (l+e-X)- 2 X
20, y = C1 e + C2 e - 1 + e-x ln(l + ex)
2
21. (D +l)y cscx y = C 1 cosx + C2 sinx + sinx ln sinx - x cosx
2
22. (D - 3D + 2)y " sin e-x
CHAPTER 15

VAIUATION OF PARA:HETERS, U~DETEHMINED COEFFICIENTS
TWO OTHER METHODS for determining a pArt icular integral of a linear differential
equation with constant coefficients
will be exhibited by means of examples.
VARIATION OF PARAMETERS. From the complementary function of 1),

y = C1 y 1 (x) + C 2 y 2 (x) + ........ + CnYn (x),
we obtain a basic relation
by replacing the C's by unknown functions of x, the L's, The method consists
of a procedure for determining the L's so that 2) satisfies 1).
See Problems 1-4.
UNDETERMINED COEFFICIENTS. The basic relation here is
where the functions r 1 (x), • · · • •, rt(x) are the terms of Q and those arising
from these terms by differentiation, and A,B,C,···,G are constants.
3
For example, if the equation is F(D)y = x , we take for 3)
3 2
y = Ax + Bx + Cx + D;
if the equation is f(D)y = ex+ e 3x. we take for 3)
y = Aex + Be 3 x,
since no new terms are obtained by differentiating ex and e 3x;

if the equation is F(D)y = sin ax, we take for 3)
y = A sin ax + B cos ax ;
if the equation is F ( D) y sec x, the method fails since the number of new
terms obtained by differentiating Q = sec x is infinite.
Substituting 3) in 1), the coefficients A,B,C,•· •· are foun1 from the re-
sulting identity, See Problems 5-6.
The procedure must be modified in case:

a) A term of Q is also a term of the complementary function. If a term of Q,
say u, is also a term of the complementary function correspondingto an s-fold
root m, then in 3) we introduce a term xsu plus terms arising from it by dif-
ferentiation.
2 2X 2
For example, in finding a particular integral of (D-2) (D+3)y = e +x,
93
9l LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS
the basic relation is y = Ax 2 e 2x + Bxe 2x + Ce 2x + Dx 2 + Ex +F, the first three

terms arising from the fact that the term e 2 x of Q is also a term of the com-
plementary function corresponding to a double root m = 2; hence, use is made
of x 2 e 2 x and all terms arising by differentiation. See Problems 7-8.
b) A term of Q is xru and u is a term of the complementary function. If u

corresponds to ans-fold root m, 3) must contain the term xr+su plus terms
arising from it by differentiation.
2 2 2
For example, in finding a particular integral of (D-2) 3 (D +3)y = x e x + x ,
the basic relation is
y = Ax 5 e 2x + Bx~e 2x + Cx 3 e 2x + Dx 2 e 2x + Exe 2x + Fe 2x + Gx + Hx + ],
2
the first six terms arising from the fact that e 2 x is a part of the comple-
mentary function corresponding to the triple root m = 2. See Problem 9.
SOLVED PROBLEMS
VARIATION OF PARAMETERS,
1. Show that if y = C1y 1 + C2 y 2 +. C3y3 is the complementary function of

2
F(D)y = (D 3 + P1 D + P2D + P3 )Y = Q
then
1)
where L1 , L2, L3 satisfy the conditions
L~Y1 + L~Y2 + L~y3 0
Al L~y~ + L~y~ + L~y~ 0
L~yt + L~y; + L;y:; Q.

is a particular solution of the differential equation.
We obtain, in view of A), by successively differentiating y = L1 y1 + L2y2 + L3 y3 :

Dy L1Y~ + L2Y~ + L3y~ + (L~Y1 + L~Y2 + L~ Y3) L1Y~ + L2Y~ + L3 y~
Bl D2y L1Y: + L2Y; + L3y~' + (L~y~ + L;y~ + L~y~) L1Y: + L2Yi + L3y;
D3y L1Y~ + L2Y; + L3y;' + (L~y~ + L~y; + L~y;) L1y': + L2y'; + L3y; + Q.
Then F(D)y = L1 {y~' + P1Yt + P2Y~ + P3y1} + L2{y; + P1y; + P2Y~ + P3y2}
+ L3{y; + P1y~' + P2Y~ + P3y3} + Q
0 + 0 + 0 + Q Q._
since y 1 , y 2 , y 3 are solutions of F(D)y = 0.
In using this method:
a) Write the complementary function.
b) Form the L function 1), which is to be a particular integral, by replacing the C's of the
complementary function with L's.
c) Obtain equations B) by differentiating l) as many times as the degree of the differential
equation. After each differentiation, set the sum of al 1 terms containing derivatives of·the
L's equal to zero, except in the case of the last differentiation when the sum is set equal
to Q. The equations obtained by setting the sums equal to zero and Qare the equations A).
d) Solve these equations for L~. L~. •••••• ••.
e) Obtain L 1 , L 2 , · ... • .. • by integration,
VARIATION OF PARAMETERS, UNDETERMINED COEFFICIENTS 9S
2
2. Solve (D - 2D)y = ex sin x,
2
The complementary function is y = C1 + C2 e x.
We form the relation y L1 + L2/x,

2 2
obtain, by differentiation Dy 2L 2 e x + (L{ + L;e x),
and set 1)
2 2 2 2 2
Since now Dy = 2L 2 e x, D y = 4L 2 e x + 2L~e x and we set 2L~e x = Q = ex sinx.
Thus, L~ = ½e-x sinx and L 2 = - -&e-x(sin x + cos x).
From 1), L~ = -L~/x = - ½ex sin x and L 1 = - -&ex(sin x - cos x).

A particular integral of the given equation is
3. Solve 3
(D + D)y = csc x.
The complementary function is y
From the relation y L1 + L2 cos x +L 3 sinx

we obtain Dy (-L2 sin x + L3 cos x) + (L~ + L~cos x +L;sinx)
and set 1) L~ + L~cosx + L~ sin x = o.
Then Dy -L 2 sinx + L3COS X,
D2y (-L2COS X - L3 sin x) + (-L~sin x + L~cos x).

and we set 2) -L~sin x + L~cos x = o.
Then D2y -L2COS X -L 3 sinx,
D3y (L 2 sin x - L3COS X) + (-L~cos x -L~sinx),
and we set 3) -L~cos x - L~sin x = Q = CSC X.
Adding 1) and 3), L~ = csc x and L1 = - ln(csc x + cot x).

Solving 2) and 3), L~ = -1 and L~ = - cotx, so that L 3 = -x and L 2 - ln sinx,
Thus, a particular integral of the differential equation is
- ln(cscx + cotx) - cosx ln sinx - xsinx,
and the primitive is

y = C1 + C2cos x + C3 sin x - ln(csc x + cot x) - cos x ln sinx - x sinx.
The complementary function is y = C1 e 3x + C2 xe 3x.
From the relation y L1e 3x + L 2xe 3x
we obtain Dy (3L 1 + L 2 )e 3x + 3L 2xe 3X 3

+ (L~ e 3x + L~xe x)
and set 1) L'1e 3x + L~xe 3X = o.
96 LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS
2
Then D y = (9L 1 + 6L 2 ) e 3x + 9L 2 xe 3x + (3L~ + L~) e 3x + 3L~xe 3x,
and we set 2) (3L~ + L~)e 3x + 3L~xe 3x = e 3x/x 2 ,
Solving 1) and 2), L~ = - 1/x and L~ = 1/x 2 , so that L 1 = - lnx and L2 = -1/x,
Y = L1e 3x + L xe 3x 2
and the primitive is y C 1 e 3x + C2 xe 3x - e 3x ln x,
UNDETERMINED COEFFICIENTS,
2
5. Solve (D - '2JJ)y = ex sin x,
The complementary function is y = C 1 + C 2 ex , As a particular integral, we take

y Aex sinx + Bex cos x.
X
Then Dy (A -B)e sinx + (A+B)excosx,
D2y - 2Bex sinx + 2Aex cos x,
2 X
and (D - 2D)y - 2Aex sin x 2Bex cos x e sin x Q,
Equating coefficients of 1 ike terms, -2A = 1 and -2B = O, so that A = - ½ and B 0.

Hence, a particular integral of the differential equation is
I X ,
y = Aexsinx+Bexcosx - 2e sin x
and the primitive is y = C 1 + C2 e2x - ½ex sin x.
This was solved above as Problem 2.
(D -2D+3)y = x 3 + sinx,
2
6. Solve
The complementary function is y = ex(C1 cos 12 x + C2 sin 12 x). As a particular intecral,

we take y Ax 3 + Bx 2 + Cx + E + F sin x + G cos x
2
Then Dy 3Ax + 2Bx + C - G sin x + F cos x ,
D2y 6Ax + '213 - F sin x - G cos x ,
3Ax 3 + 3(B-24)x + (3C-4B + 64)x

2 2
and (D - 2D + 3) y + (3E-2C + 2B) + 2(F + G) sin x + 2(G-/i') coil x
x 3 + sin x,
Equating coefficients of 1 ike terms, 3A = 1 and A = 1/3; B - 2A = O and B ~ 2/3;

3C-4B+64=0 and C=2/9; 3E-2C+2B=O and E =-8/27; 2(F+G) =1, G-F=o and F=G "*·
1 3 2 2 2 8 1 .
y - X + - X + - X + - ( S 1n X + COS X)
3 3 9 27 4
and the primitive is
y
VARIATION OF PARAMETERS, UNDETERMINED COEFFICIENTS 97
2
7. Solve + X •
.
Th e comp 1ementary f ,.nc t 10n .
1s y = C1 e x + C2 e-x + C3 e-2x • since
. ex occurs in Q an d also in
the complementary function corresponding to a root of multiplicity one, we take as a particu-
lar integral
1) y
X
Then Dy 2Ax + B + Exe + (E+F)ex,
X
D\ 2A + Exe + (2E +F)ex,
X
D\ Exe + (3E+F)/,
3 2 X 2
and (D + 2D - D - 2) y -2Ax2 - 2(B+A)x + (4A-B-2C) + 6Eex e + X
Equating coefficients of like terms, - 2A = 1, B + A = O, 4A-B-2C = o, 6E = 1; hence,

A=-+. B = f, C = -¾,
E = f, and Fis arbitrary. Now F should be arbitrary here, since
C1 ex is a term of the complementary function. Thus, in writing 1), the inclusion of Fex was
unnecessary.
1 2 1 5 1 X
Hence, a particular integral is y = X + - X - + - xe
2 2
-4 6
1 5 1 X
and the prirni ti ve is y + -x + -xe
2 4 6
The complementary function is y = c 1 /x + C2 x/x Now /x is a part of Q and also occurs

in the complementary function corresponding to a root of multiplicity two. As aparticular in-
tegral, we take
y
2 2
Note that terms involving xe x and e x are not included, since they appear in the complementary
function with arbitrary coefficients. Then
2 2 2
Dy 2Ax 5 /x + (5A + 2B)x 4 e x + (4B + 2C)x 3 /x + (3C + 2E)x e x + 2Ex/x,
5 2x 4 2x 3 2x 2 2x 2x 2x
4Ax e +(20A+4B)x e +(20A+16B+4C)x e +(12B+12C+4E)x e +(6C+8E)xe +2Ee ,
20Ax 3e 2x + 12Bx /x + 6Cx/x + 2E/x x 3 /x + x/x
2 2
and {D -4D+4)y
Equating coefficients of like terms, 20A 1, 12B o, 6C = 1, 2E = o; hence, A = 1/20,

B = o, C = 1/6, E = o.
1 5 2x 1 3 2X
Thus, a particular integral is y X e + - X e
20 6
5 2X 1 3 2X
and the primitive is y X e + X e
6
2 2
9. Solve (D + 4)y x sin2x,
The complementary function is y = C1 cos 2x + C2 sin 2x,

2
Since x sin 2x occurs in Q and sin 2x is a part of the complementary function correspond-
ing to a root of multiplicity one, we take as a particular integral
3 3 2 2
y = Ax cos 2x + Bx sin 2x + Cx cos 2x + Ex sin 2x + Fx cos 2x + Gx sin 2x.
Note that H cos 2x + K sin 2x is not inc 1uded, since these terms are in the complementary func-
tion. Then
2 2
Dy 2Bx 3 cos 2x - 2Ax 3 sin 2x + (3A + 2E)x cos 2x + (3B -2C)x sin 2x
+ (2C + 2G)x cos 2x + (2E - 2F)x sin 2x + F cos 2x + G sin 2x ,
98 LINEAR EQL"ATIONS WITH CONSTANT COEFFICIENTS
2
-4Ax 3 cos 2x - 4Bx 3 sin 2x + (12B-4C)x
2
cos 2x + (-12A-4E)x sin 2x
+ (6A + BE - 4F)x cos 2x + (6B- SC - 4G)x sin 2x + (2C + 4G)cos 2x + (2E -4F) sin 2x,
and
2 2 2
(D + 4)y 12Bx cos 2x - 12Ax sin 2x + (6A + 8E)x cos 2x + (68- 8C)x sin 2x
2
+ (2C+4G)cos 2x + (2E-4F)sin 2x x sin 2x.
Equating coefficients of 1 ike terms, - 12 A 1, 12B = O, 6A +SE= o, 6B - BC= o,

2C+4G=o, 2E-4F=o; hence, A=-1/12, B=o, C=O, E = 1/16, F = 1/32, G = o.
3 2 1
A particular integral is y = - -1:. x cos 2x + -1:. x sin2.x + - X COS 2.x,
12 16 32
and the primitive is y = C1 cos 2x + C2 sin 2x -1:. x 3 cos 2x + -1:. x 2 sin 2x + -1:. x cos 2x.
12 16 32
Solve, using the method of variation of parameters.
2
10. (D + l)y = csc X Ans, y =C1 cosx+C 2 sinx+sinx lnsinx -x cosx
2 2
11, (D + 4)y = 4 sec 2x Ans, y = C1 cos 2x + C2 sin 2x - 1 + sin 2x ln(sec 2x + tan 2.x)
2X
Ans, y = C1ex + C2e3x + .J.e
2 + ½(ex_ e3x) ln(l + e-x)
-x . -x -x X -x X -x
13, 1 )Y = e sine + cos e Ans, y = C1 e + C2 e - e sin e
-x -2 X -x
l)y (1+ e ) Ans, y = C1 e + C2 e - 1 + e -x ln(l + ex)
Solve, using the method of undetermined coefficients.
2
16, (D - l)y = ex sin 2x Ans, y = C 1ex + C 2 e-x - ex(sin 2x + cos 2.x)/8
y = e -x (C1 cosx+ C2 sinx)+

. 1 1 .
2 2 2
17, (D + 2D + 2)Y = x + sin x Ans. -(x-1) + -(s1nx- 2cosx)
2 5
2
y = C 1e 3x t C2 e- 3x - x/9 - e 2 x/5 +
2
18, (D - 9)y = x + e x - sin 2x Ans, .2_ sin 2x
13
3 2 2
Ax 3 + Bx + Cx.)
2
19, (D + 3D + 2D)y = x + 4X + 8 (Use
2
20. (D + 1 )Y = - 2 sin x + 4x cos x
1 3 2x
Ans, + -x e
6
CHAPTER 16

SHORT METHODS
A PARTICULAR INTEGRAL of a linear differential equation F(D)y = Q with constant

coefficients is given by y = - 1-Q. For certain forms of Q the labor in-
F(D)
valved in evaluating this symbol may be considerably shortened, as follows:
a) If Q is of the form eax,
y = _1_ e ax 1 ax
F(a)-/- 0.
e •
f(D) F(a)
See Problems 2-3 when f(a) i 0, and Problems 4-5 when F(a) 0.
b) If Q is of the form sin(ax -1- b) or cos(ax-1-6),

1 sin(ax + b) 1 sin(ax + b),
2
y = 2
= 2
F(-a ) / 0,
F(D ) F(-a )
1 1 2
y = 2
cos (ax+ b) ==
2
cos (ax+ b), F(-a ) I- 0.
F(D ) F(-a )
See Problems 7-11 when F(-a

2
) I 0, and Problem 12 when F(-a 2 ) = o.
c) If Q is of the form xm,
y = -1- X
m
(a o + a 1 D + a 2 D2 + • • • • • + a m Dm) xm ,
F(D)
obtained by expanding - 1- in ascending powers of D and suppressing all terms
F(D)
beyond Dm, since Dnxm = 0 when n > m. See Problems 13-15.
d) If Q is of the form eaxV(x), y

1 eaxv ax 1 v.
= = e
F(D) F(D + a)
See Problems 17-20.
F' (D)
e) If Q is of the form xV(x), y = - 1-xV = x-l_y 2
v.
F(D) F(D) {F(D) }
See Problems 21-23.
SOLVED PROBLEMS
1. Establish the rule in a) above,

2 2 ax r ax
D y = a e , =- a e ,
F(D)eax = I.P,,,.Dreax 1 ax 1 ax
Hence, -- e -- e
r F(D) F(a)
99
3 2 4X
2. Solve (D - 2D - 5D + 6) y = e or (D - 1) (D - 3) (D + 2) y = e4x.
The complementary function is
1 4X
A particular integral is y = --------e
(D - 1) (D - 3) (D + 2)
1 4X 1 4X 1 4X
e e e •
( 4 - 1) ( 4 - 3) ( 4 + 2) 18
Hence, the primitive is y
3 2 2X 2
3. Solve (D -2D -5D+6)y = (e + 3) •
1 2x 2
A particular integral is y = -------- ( e + 3)
(D - 1) (D - 3) (D + 2)
1 4X 6 2x 9 Ox
--------e
(D - 1 l (D - 3) (D + 2)
+ --------e
(D-l)(D-3)(D+2)
+
(D - 1) (D - 3) (D + 2)
e
4X 3/X
-1- 4X
e +
6
e
2x
+
9 e
- -- + -3
3(1) 6 1(-1)4 (-1)(-3)2 18 2 2
4X 3/X
X e 3
The primitive is y = C1e + C2e3x + C3e-2x + - +
18 2 2
3 2 3x
4. Solve (D - 2D - 5D + 6) y = e •
1 3x
A particular integral is y = -------- e • Now F(a) = F{3) O, and the short
(D-l)(D-3)(D+2)
method does not apply, However, we may write
1 e3X 1 ( 1 e3x) 1 1 3x
y = --e
{D-l)(D-3)(D+2) = D - 3 (D - 1) (D + 2) 10 D-3
1 3X
- 1 e 3xJ3x-3xdx
e e _1_e3x J dx -xe •
10 10 10
The primitive is y
3 • 2X X -X
5. Solve
2
(D -~D +8D-4)y = e +2e +3e •
1 2x 2 X 3 -x
y = ------e + e + ------ e
(D - 1) (D - 2)2 (D -1) (D - 2)2 (D-1) (D - 2/
1 (-1- /X) 3 -x
e
(D -2/ D -1 (D -1) (D - 2/
1 2x 2 X 3 -x
e + e + e
(D -2/ D -1 (-2)(-3)2
1 -x 1 2 2X 2xex 1 -x
/X JI (dx/ + 2ex J dx - e - X e + - - e
6 2 6
SHORT METHODS 101
The primitive is y
6. Establish the rule in b) above for cos(ax + b),
2 4
Since, wheny=cos(ax+b), Ii2y=-a cos(ax+b), D y
2r 2 r
D y = (-a ) cos(ax + 6), then
2 2r 2 r 2
F(D )coo(ax+b) = LPrD cos(ax+b) = LPr(-a) cos(ax+b) F(-a ) cos(ax + b).
r r
1
Hence, cos(ax + b) cos(ax + b),
2
Fcdl F(-a )
7. Solve
2
(D + 4)y = sin 3x,
The complell1€ntary function is y = C 1 cos 2x + C2 sin 2x, and a particular solution is
1
y sin 3x sin 3x sin 3x.
2 5
-(3) + 4
The primitive is y · 2x
C1COS 2x + C2Sln - ~ s1'n3x.
5
The complementary function is y = C1 cos x + C2 sin x + C3 cos 3x + C4 sin 3x, and a particu-
lar intearal is
1 1 l
y cos(2x + 3) - - - cos(2x + 3) cos(2x + 3).
(D2 + l)(D2+ 9) (-3)(5) 15
1
cos(2x+3).
15
7
9. Solve (D +3D-4)y = sin 2x.
4
The C0111Plementary function is y = C1 ex + C2 e- x, and a particular integral is
1 1
y - -- - sin 2x - - - - - - sin 2x.
2
D +3D-4 (D -1) (D + 4)
1
The operator here is not of the form - - , and the short method does not apply. However,
F(D2)
we IIIIIJ" use either of the following procedures to shorten the work,
(D+l)(D-4) 1 2
a) y sin 2x sin 2x - (D -3D·-4) sin 2x
2 2 100
(D - 1) (D +4) (D - l)(D -16)
1 1
- (-4 :sin 2x - 6 cos 2x - 4 sin 2x) - - (4 sin 2x + 3 cos 2x).
100 50
1 1 3D + 8
b) y " - -- - sin 2x sin 2x sin 2x sin 2x
2
(-4) + 3D -4 3D-8 2
D + 3D-4 9D - 64
1 1 ·2x + 3 COS 2x ) •
- - (3D+8) sin2x - - ( 6 cos 2x + 8 sin 2x) - - 1 ( 4 Sln
100 100 50
(D 3 +D +D+ l)y = sin 2x + cos 3x.

2
10. Solve
The complementary function is y = C 1 cos x + C2 sin x + C3 e-x, and a particular integral is
1 1
y - - - - - - ( sin 2x + cos 3x) sin 2x + - - - - - - cos 3x
2 2 2
(D +l)(D+l) (D +l)(D+l) (D +l)(D+l)
1 1 1 D -1 1 D -1
sin 2x cos 3x - - - - sin 2x - - - - cos 3x
3 D +1 8 D+1 3 D2 - 1 8 D2 - 1
1
~ (D - l)sin 2x + -
1
( D - 1) cos 3x
1
-(2 cos 2x sin 2x) - -(3 sin 3x T cos 3x),
15 80 15 80
The primitive is
y = C1 cos x + C2 sin x + C3 e-x + ~(2 cos 2x - sin 2x) - ~(3 sin 3x + cos 3x).
15 80
2
11. Solve (D -D + l)y = sin 2x.
.Lx
2
The complementary function is y e (C 1 cos ½/3 x + C2 sin ½/3 x), and a particular inte-
gral is
1 1 D-3
y - -- - sin 2x sin 2x sin 2x - - - sin 2x
2 2
D - D+1 (-4)-D+l D+3 D - 9
1
~(D-3)sin2x -(2 cos 2x - 3 sin 2x),
13 13
2
12. Solve (D + 4)y = cos 2x + cos 4x.
The complementary function is y = C1 cos 2x + C2 sin 2x, and a particular integral is
1
y - - ( cos 2x + cos 4x) cos 2x + .::os 4x.
2
D +4
1 2
The method of this chapter cannot be used to evaluate - - cos 2x since, when D is re-
D2 + 4
2
placed by -4, D +4 = 0. However, the following procedure may be used.
1 1 1
Consider cos(2+h)x - - - - - cos(2 +h)x - - - - cos(2 + h)x
2 2
D +4 -(2+h/+4 4h+ h
1 2
- - - ( c o s 2x hx sin 2x - ½{hx) cos 2x + • • • • • • • • • •
h(4+h)
by Taylor's theorem. The first term, cos 2x, is part of the complementary function and need
not be considered here. Hence, a particular integral is
1 1
cos(2 + h)x (hx sin 2x + ½(hx/cos 2x - .......... )
2 h(4 + h)
D +4
1
- - ( x sin 2x + ½hx 2 cos 2x - .......... )
4+h
1 1
Letting h---.o, we obtain cos 2x = ~x sin 2x. Since - -- cos 4x cos 4x,
2 4 2 12
D +4 D +4
the primitive is y = C1 cos 2x + C2 sin 2x + ~xsin 2x - 2._cos4x. (Compare this solution

4 12
with that given in Problem 8, Chapter 14.)
SHORT METHODS l 0.1
_ _1__x
The complementary function is y = e
2
(C1 cos ½15 x + C2 sin ½ ✓5 x), and a particular in-
tegral is
1 2 1 2 2 2
y (x +2.x-1) (- ~D - -D ) (x + 2x - 1)
2D2 + 2D + 3 3 9 27
1 2 2 2 1 2 2 23
-(x + 2x - 1) -(2.x+2) -(2) -x + -x - -
3 9 27 3 9 27
Note:
2
1
(-
1
3
~D
9
l:....n2
27
+ .......... ) by direct di vision,
2D + 2D + 3
-½X ~ 1 2 2 25
The primitive is y e (C1 cos
I
2 /5 x + C2 sin ½15 x) + -x + -x
3 9 27
2
The complementary function is y = C1 e- x + ex (C2 cos x + C3 sin x), and a particular in-
tegral is
1 2 3 4 4 2
y = - -- - ( x4 + 3x - 5x + 2) -D ) (x + 3x - 5x + 2)
3 64
D -2D + 4
1 4 1 3 3 2 5 7
-x + -x + -x -x -
4 2 2 4 8
7
The primitive is y
8
2
Solve =X •
The complementary function is y = C 1 + C2 ex + C3 e 3x, and a particular integral is
1 2 1 1 2 1 1
y
2
X - (
2
)x - ( - + !n + 13D2)x2
D(D -4D + 3) D D -4D + 3 D 3 9 27
1 1 2 8 26 1 4 2 26 1
- (-x + -x + - ) -x 3 + -x + -x, since -{f(x)} J f (x) dx,
D 3 9 27 9 9 27 D
X 3X 1 3 4 2 26
The primitive is y C1 + C2 e + C3 e + -x + -x + - x .
9 9 27
1 2
y = (x + 3e2x + 4 sin x)
2 2
D (D + 2D - 3)
1 2 1 2x 1
------x + 3 2 2 e + 4
2 2
sin x
D2 (D2 + 2D -3) D (D + 2D - 3) D (D + 2D - 3)
3 2x 4
----- e + sin x
4(4+4-3) (-1)(-1+2D-3)
1 1 7 D2 2 3 2x 2
- )x + -e sin x
D2 (- 3 27 20 D-2
1 1 2 3 2x
- - -(9x + 12.x + 14) + - e
27 D2 20
1 3 4 3 2 3 2x 2
--(-x+2x+7x) + e + -(cosx + 2 sinx),
27 4 20 5
The primitive is
17. Establish the rule ind) above by first showing that F(D)eaxU
Since when y = eaxU, Dy = aeaxl.i + eaxDU = eax (D + a)U,

2 2 2
D y aeax(D+a)U+ eaxD(D+a)U = eax(D +2aD+a )U = eax(D+a/U,
Dry e ax (D + a { /J, and

JX 1'" ax
1) F(D)eaxU = LP Dr(eaxU) LP,,,. e (D + a) U e F(D + a)U.
r r r
Let V = F(D + a)[! so that U 1 _ V.

__ Then, from 1),
F(D + a)
_ 1_ {F(D)eax 1 V} ax __l_y _
and e
F(D) F(D+a) F(D + a)
2 3X
18. Solve (D2 - 4)y = X e
The complementary function is y = c1 /x + C2 e- x,

2
and a particular integral is
1 2 3X 3x 1 2 3x 1 2
y --x e e X e X
2 2
D -4 (D + 3/ - 4 D + 6D + 5
2
3x 1 31 D2 2 3X X 12 62
:= e ( :- ~D + 125 )x e (- -x +
;J 25 5 25 125)
The primitive is y
2
19. Solve (D +2D+4)y = e
X
sin 2x.
The complementary function is y e-x (C 1 cos /3 x + C2 sin /3 x), and a particular integral
is
1 X X 1
y - -- - e x sin 2x e sin 2x e sin 2x
2 2 2
D + 2D +4 (D + 1) + 2 (D + 1) + 4 D +4D+7
X X
1 x 4D-3 .
ex - - sin 2x e - - - s i n 2x - :.__(4D - 3) sin 2x :.__(8 cos 2x - 3 sin 2x).
41)+3 2 73 73
16D - 9
X
The primitive is y e-x(C 1 cos /3 x + C2 sin V3 x) - :.__(8 cos 2x - 3 sin 2.x).
73
SHORT METHODS lOS
2 3X X
20. Solve (D -4D+3)y = 2.xe + 3e cos 2x.
The complementary function is y = C1 ex + C 2 e 3x, and a particular integral is
1 1
y ( 2xe 3x + 3ex cos 2.x) 2 1 xe3x + 3 - -- - ex cos 2x
2 2 2
D -4D + 3 D -4D + 3 D -4D + 3
3X 1 1 2e3x ~ • _1_ x + 1
2e ---x + 3ex - -- cos 2x 3ex - -- cos 2x
D2 + 2D D -2D
2 D D+2 -4 -2D
1 1 1
2e3X -(---D)x 3 X D-2 1 3x 1 3
- e cos 2x - e -{2.x-1) + ex (D - 2)cos 2x
D 2 4 2 D -4
2 2 D 16
1 3x 2 3 X
- e (x -x) - e (cos 2x + sin 2.x).
2 8
X 3X 1 3X 2 3 X
The primitive is y C1 e + C2e + -e (x -x) - e ( cos 2x + sin 2.x).
2 8
21. Establish the rule in e) above by first showing that F(D)xU = xF(D)U + F '(D)U,
2 2
Since when y " xU, Dy = xDU + U, D y = xD U + 2DU,
Dry = xDrU + rDr-lu xDrU + (__<}_ Dr)U, then
dD
1) F(D)xU = I P,,,.Dr (xU)
r
Let V = F(D)U so that U = _l_ V. Then, substituting in 1),

F(D)
1 1 1
F(D)x - - V '
xF(D) - - V + F(D) - -1V , xV F(D)x - - V - F'(D) _l_y
F(D) F(D) F(D) F(D) F(D) '
1 1 1 1 F'(D)
and - -xV X -- V - - - F'(D) - - V X _1_ y
2
v.
F(D) F(D) F(D) F(D) F(D) {F(D)}
2
22. Solve (D + 3D + 2) y = x sin 2x,
X -2X
The complementary function is y = C1 e- + C2 e , and a particular integral is
1 1 2D + 3
y = x sin 2x X - -- - sin 2x sin 2x
2 2
D + 3D + 2 D + 3D + 2
1 2D + 3
= x - - sin 2x sin 2x
4 2
3D-2 D + 6D 3 + 13D + 12D +4
1 2D + 3 2
x - - sin 2x - - - - - - - - - - - - - sin 2x, replacing D by -4,
3D -2 (-4/ + 6{-4)D + 13(-4) + 12D + 4
3D + 2 1 (2D+3)(3D-8) sin x
x --- sin 2x + 2
2 4 2
9D - 4 9D - 64
-x(3 cos 2x + sin 2x) 24 sin 2x + 7 cos 2x

+
20 200
-x -2x 30x -7 5x -12

The primitive is y = C1e + C2e - - --
200
cos 2x - ---
100
sin 2x.
106 Lll\EAR EQUATIONS WITH CONSTANT COEFFICIENTS
2 2 •
23 • Solve (D - l)y = x sin 3x.
The complementary function is y = C 1 ex + C2 e-x, and a particular integral is
1 1 2D
y - - x2 sin 3x x - - x sin 3x - ---- x sin 3x
2
D -1
2
D -1 (D2 -1/
3
2 1 2D 1 4D - 4D
X - - sin 3x - X sin 3x 2D {x sin 3x sin 3x}
2 4 2 4 2
D -1 (D2 -1/ D -2D +1 {D -2D +1/
2
2 1 2D 1 8D
X - - sin 3x -
2
X sin 3x - 2D {x sin 3x} +
2
sin 3x
D -1 (D2 - 1/ (D2 - 1/ (D - 1/
1
X
2
sin3x -~ X C03 3x ~ D(x sin 3x) +
9
sin 3x
10 50 50 125
1 2 3 13
X sin3x - X COS 3x + sin 3x,
10 25 250
2
25x - 13 3
C1 e + C2 e-x -
X
The primitive is y = sin 3x - - x cos 3x,
250 25
3 2 -3x
24. Solve (D - 3D - 6D + 8)y = xe
2
The complementary function is y C1 ex + c2 /x + C3 e- x, and a particular integral is
1 -3x
y ------xe
3 2
D - 3D - 6D + 8
-3x 1 -3x
By a): y e -------------x e --------x
D3 - 12D + 39D - 28
3 2
(D - 3 ) - 3 (D - 3 / - 6(D - 3) + 8
-3x 1 39 -3x 1 ~)
e (----D)x e (- - X
28 784 28 784
2
1 -3x 3D 6D - 6 -3x
-
By e): y = X ------e
3 2
--------e
3 2
D -3D -6D+8 (D -3D - 6D + 8/
2
1 -3x 3D - 6D -6 -3X 1 -3x 39 -3x
- -xe e - - xe e
28 28 784
(-28/
-3x
The primitive is y C1ex + c.i/x + C:ie-2x - e--(28x + 39).
784
25. Denote the real and imaginary parts of a complex number z by Re[z] and Im[z] respectively. An
alternate short method for Problems 9-11 makes use of sin bx= Im[eibx] and cos bx= Re[ibxJ.
Consider, for example, F(D)z (D 3 + D2 + D + 1) z for which
z
F(2i) F(3i) 3+ 6i 8+ 24 i
2 i- 1 (cos2x+ isin2x) + 3 i-l(cos3x+ isin3x)
2:_i_ + "2
15 80
is a particular integral. Then
- l...c2 sin 2x + cos 2x) .l.c3 sin 3x + cos 3x)

15 80
SHORT METHODS 107
is a particular integral of F(D)y = cos 2x + cos 3x,
1 (2 cos 2x - sin 2x) + J (3 cos 3x - sin 3x)

15 80
is a particular integral of F(D)y = sin2x + sin3x,
is a particular integral of F(D)y cos 2x + sin 3x , and
is the particular integral in Problem 10.
Find a particular integral.
X
27, = e
X 3 2X
+ x e /6
X 2X
28. = e + xe y = e
29, (D~ - l)y = sin 2x y = 2_ sin 2x

15
1
30, (D 3 + l)y = cos x y = (cos x - sin x)
2
31. (D
2
+ 4)y = sin 2x y = - !x cos 2x
4
32, (D
2
+ 5)y = cos v5 X y
= v5
10
x sin i/5x
1 X -X 1 , '
y= (e +2xe )- x(sinx+cosx)
4 4
2 2 2
34. (D - l)y = X y = - x - 2
1 6 ~
35 , D~ (D2 -l)y=x 2 y = -
360
(x + 30x )
2
36, (D
2
+ 2)y = x 3 + x
2 2
+ e- x + cos 3x = ~(x 3 +/ - 3x -1) + ~e- x - ~cos 3x
y 2 6 7
2 1 X
37. (D - 2D - l)y = ex cosx y e- - :3 e COS X
2
y - e x ln x
2 3x 1 3X
39, (D - l)y = xe y e (4x - 3)
32
2 -2x 2
40. (D +5D+6)y=e (secx)(l+2tanx)
CH \PTER 17
Linear Equal ions "ith Variabl(' Coeffkicnts

TIIE C\UHY .nn LEGK\DHE Lt\E.\H EQUATIONS
THE CAUCHY LINEAR EQUATION
in which p0 ,p1 , • • •· • ,Pn are constants, and the Legendre linear equation
n
p (ax+ b) -
i"'y + p (ax+ b) 11 - 1
11 1
d -
Y + .. • .. + p
d
(ax+ b) .1'. + PnY = Q(x),
2) __
o dxn 1 dxn-1 n-1 dx
of which 1) is the special case (a= l, b =0), may be reduced to linear equa-
tions with constant coefficients by properly chosen transformations of the
independent variable.
THE CAUCHY LINEAR EQUATION. Let x = then if f) is defined by f) - d

- dz '
dy dy dz 1 dy
Dy = =- and xDy = dy = f)y.
clx dz dx x dz dz
2
D2y = _:!__(_!_ dy) = l._(d y _ dy) and x2 D2y f)(f) - l)y •
dx x dz x
2
dz 2 dz
2 3 2
D3y = i(~ - dy) + _!.__(d y - d y)
x 3 dz 2 dz x3 cJz3 dz 2
3 2
= _!_(d y 3 cl y + 2 dy_) and x3 D3y = f)(f) -1) ({} - 2)y,
3 2
x clz 3 clz dz
After making these replacements, 1) becomes
a linear equation with constant coefficients.

see Problems 1-3.
THE LEGENDRE LINEAR EQUATION. Let ax+ b =e2 ; then
Dy __ dy dz = _a_ dy and (ax+ b)Dy

dz dx ax+ b dz
2 2 2
and (ax+b) Dy = a f)(f)-l)y,
THE CAUCHY AND LEGENDRE LINEAR EQUATIONS 109
(ax +b),-- D,--y
After making these replacements, 2) becomes
{p0 anl9(19-1)(£i-2)·"•(19-n+l) + p 1 an- 1 19(19-1)(£i-2)""(19-n+2) + . . . . . . .
+ p al9 + p }y = Q(ez-b),
n-1 n A
See Problems 4-5.
SOLVED PROBLEMS
3 3 2 2
1. Solve ( x D -,- 3x D - 2xD + 2) y = 0.
2
The transformation x = e reduces the equation to
{19cl9-1) clY- 2) -.- 319cl9-1) - :w + 2} y
3
cl9 - 3£1 + 2)y 0
Wh OSe
.
SO 1Ut1on . y = ("-1e 2
lS -1-
('
s2Z€
2
+
('
,3€
-2 2
,
Since z = 1n x. the complete solution of' the given equation is y C1 x + C2 x In x + C3 /x 2 •
3 3 2
2. Solve (x D + 2xn-2)y = x ln X + 3x.
2
2 22 2
{J()cl9-l)(l9-2) + 219 - 2}y = (19-1)(19 -2l9+2)y = ze + 3e ,
2 2
The complementary function is y = C\e + e (C 2 cos z + C3 sin z ), and a particular integral is
22 1 2
y e -------------z+ 3 1 e
3 2
(Jy + 2) - 3 ( 19 + 2/ + 4 (19 + 2) - 2 (19 - 1) ( 19 - 219 + 2)
22 1 1 2
e ------ z + 3---- e
2
19 3 + 319 + 4l9 + 2 ( £1 - 1) ( 1)
e
22 I
(:z - IY) z
1f1
+ 3e
2 J e2 • e-2 dz = e
22 I
(2 z - 1) + 3ze •
2
2 2 22 2
Thus, the solution is y C1 e + e (C2 cos z + C3 sin z) + -J,e (z -2) + 3ze
C1 x + x(C2 cos lnx + C3 sin lnx) + ½x 2 (lnx -2) + 3x lnx.
2 2
3, Solve (xD -xD+4)y = coslnx+xsinlnx,
2
2 2
{£1(£1-1) - £1 + 4} y (£1 - 2£1 + 4)y = cos z + e sin z.
2
The complementary function is y = e (C1 cos /3 z + C2 sin v'3 z), <1nd a particular solution
is
1 1 2 .
y - - - - cos z + - - - - e sin z
2 2
£1 - 219 + 4 19 - 2£1 + 4
1 2 1 1 1 2
- - cos z + e sin z -(3 cosz - 2 sinz)+ -e sinz,
3-2£1 £12+3 13 2
II 0 LINEAR EQUATIONS WITH VARIABLE COEFFICIENTS
Thus, the solution is

2 2
y = e (C1 cos /3 z + C2sin /3z) + __!_(3 cos z - 2 sin z) + ~e sin z
13 2
1 1
x(C1cosv'3•lnx + C2sinv'3•lnx) + -(3cos lnx - 2sin lnx) + -x sin lnx.
13 2
4. Solve (x + 2)
2
d\ - (x + 2) dy + y 3x + 4,
dx 2 dx
2
Put x +-2 = e ; then the given equation becomes
2
{J()(J() - 1) - 19 + i} y = ( 19 - 1/ y = 3e - 2.
2 2
The complementary function is y = C1 e + C 2 ze , and a particular integral is
1 z __1 _ eoz 3 2 z
y - - - ( 3 e -2) 2 e - 2.
=
(19- 1/ ({}-1/
2z
The solution is y = or, since z ln(x+2),
y (x + 2) [C1 + C2 ln(x + 2) + ~ In
2
(x + 2)] - 2.
2
2
The transformation 3x + 2 = e reduces the equation to
2 1 2 1 22 2 1 2z
{ 919( l9 - 1) + 9 l9 - 36 } y = 9 ( l9 - 4) y = - ( 9X + 12x + 3) -(e - 1) or ({} - 4)y = (e - 1).
3 3 27
22 , -2z 1 1 2z 1 Oz
The complete solution is y = C1e + C2e + -(-- e -2-- e )
27 192 - 4 {) - 4
1 2z
- ( z e + 1)
108
2 1
or y C1 (3x+2/ + C2 (3x+2i- + - -[(3x+2/In(3x+2) + 1].
108
Solve,
2 2
6, Ans, y = C1x + C2 x In x + X + ~ x
2 3
ln x
6
7. 2
Ans. y = C1x + C2 x + ½on2x + ln x) + -&
3 3 2 2
8. (x D + 2x D )y = x + sin(ln x) Ans, y + C2 x + C 3 lnx + X ln x
= C1
I
+ 2(COS ln x + sin In x)
4 2 4
3x Ans. y = C1 x + C2 x ln x + C3 x ln x + x /9
2
10. [ex+ 1/ D + (x + l)D - 1]y = ln(x + 1/ + X - 1
1 2
Ans, y =Ci(x+l) +C2 (x+l)- - ln(x+1) + ½cx+l)•ln(x+1)+2
11, (2X + 1/ y" - 2(2x + l)y' - 12y = 6x Ans,

CHAPTER 18
Linear Equations with Variable Coefficients

F,Ql'ATIO~S OF THE SF,CO~D ORDER
A LINEAR DIFFERENTIAL EQUATION of the second order has the form
1) + R(x) dy + S(x) y = Q(x) •

dx
If the coefficients R and S are constants, the equation can be solved by

the methods of the preceding chapter; otherwise, no general method is known.
In this chapter certain procedures are given which at times will yield a so-
lution.
CHANGE OF DEPENDENT VARIABLE. Under the transformation

y = uv, u = u(x) and V = v(x),
d2y 2 2
dy dv du d v dv du d u
= u- + v-. u- + 2 + V-,
dx dx dx dx2 dx2 dx dx dx 2
1) becomes
2)
2 du .! { d2u + R(x) du + S(x)•u} •

+ R(x), =
u dx u dx2 dx
2
d y d
a) If u is a particular integral of + R(x) --1'. + S(x) y = 0, then S 1 = 0
dx2 dx
and 2) becomes
3)
dv
The further substitution dx = p, reduces 3) to
dp
4)
dx
a linear equation of the first order. See Problems 1-6.
b) If u is chosen so that R1 (x) = 2 du + R(x) = 0 or

du
- = -
1
2R(x) dx, then
u dx u
-½fR(x)dx
u = e
111
112 LINEAR EQUATIONS WITH YARIABLE COEFFICIENTS
du , du , dR
Now = -~uR(x) and = - 2R(x) - - 2u - so that
dx dx dx
1 d 2u
R( x) d u + __ _
S(x) +
S1 (x) = S(x) + - =
u dx u dx 2
and Q1 = Q/u.
If S1 (x) = S - ~R
2
- ½: = A, a constant, 2) becomes Q/u,

2 d 2v 2
a Cauchy equation,
2) becomes x - + Av = Qx /u,
dx2
2
and the substitution x = e will reduce it to one with constant coefficients.
See Problems 7-10.
CHANGE OF INDEPENDENT VARIABLE. Let the transformation be z = 8(x). Then
dy = dy dz
dx dz dx
and 1) becomes
2 2
d y (dz/ + (dz2 + Rdz)dy + Sy = Q
dz2 dx dx dx dz
or
2
d z + R dz
d2y 2 dx dy Sy
dx _Q_,
5) + + =
dz 2 (dz/ dz (dz/ (dz/
dx dx dx
Let z = 8(x) be chosen so that dz = f±S the sign being that which makes
dx ...;-;;·
dz 2 t
real and a being any positive constant. (One may consistently take a = 1.)
dx
2
d z dz
+R- 2
dx2 dx dy+,dy n
If now = A, a constant, then 5) becomes ,.,
2
± a Y = __:s::.._,
(dz/ dz2 dz (dz/
dx dx
a linear equation with constant coefficients. See Problems 11-14.
OPERATIONAL FACTORING. It may be possible to separate the left member of

2
{P(x)D + R(x)D + S(x) }y = Q(x)
into two linear operators F1 (D) and F 2 (D) so that
Then, sP.tting F 2 (D)y = v, 6) becomes F 1 (D)v = Q(x), a linear equation ot

order one.
E(Jl'ATJO:\S OF THE SE<:O_:\T) ORDER
The factorization in this section differs from that of Chapter 13. With
possible exceptions, the factors here contain the independent variable x,
they are not commutative, and the factorization differs from that when D is
treated as a variable. For example,
2 2
{xD - (x + 2)D + x}y = {(xD-2)(D-x)}y,
since
d
{(xD-2)(D-x)}y = (xD-2)(- -x)y = (xD-2)(y'-xy)
dx
= (x_!}_-2)(y'-xy) = x(yu-y-xy')-2(y'-xy)
dx
2 2 2
xy 11 - (x + 2)y' + xy {(xD - (x + 2)D + x}y.
The factors are not commutative, since
2
{(D-x)(xD-2)}y (D-x) (xy'-2y) = xy + y' - 2y' - x y' + 2xy
11
2 2 2
xy" - (x + l)y' + 2xy :::: {xD - (x + l)D + 2x} y.
Finally, when D is treated as a variable rather than an operator,
2 2
{(xD-2)(D-x)}y = {xD -(x +2)D+2x}y. See Problems 15-17.
IN SUMMARY, the following procedure is suggested for solving

2
d y cl
+ R(x) _l' + S(x) y = Q(x),
dx2 dx
1) Find by inspection, or otherwise, a particular integral u = u(x) of the

equation when Q(x) = O. The substitution y = uv will yield a linear equa-
tion in which the dependent variable v does not appear. This equation is
of the first order in dv/dx = p.
2
2) If a particular integral cannot be found, compute S - -kR - ½c1R. If this
dx
2 - '-- f R dx
is a constant K or K/x , the trans format ion y = ve 2 reduces the given
equation to a linear equation with constant coefficients or to a Cauchy
equation.
3) If the above procedure does not apply, put dz = {±s' (choosing the sign
dx ✓~
2
d z + R dz
dx2 dx
so that the square root is real) and substitute in If this
dz 2
( -)
dx
is a constant, the transformation z = Jjg dx yields a linear equation

with constant coefficients.
4) If the left member of the equation is operationally factorable, the prpb-

lem is then reduced to that of solving two linear equations of order one.
Note. As a partial check on the work, it is desirable to know the type of equa-
tion which results when the transformations in 1)-3) are made.
II I LINEAR EQUATIONS WITH YARIABLE COEFFICIENTS
SOLVED PROBLEMS
2
1. ~
7
0r the equation (D + RD +S)y = 0, show that
a) y X is a particular integral ifR+xS=0,
X
b) y e is a particular integral i f l+R+S = 0,
-x
c) y e is a particular integral if 1-R+S=0,
mx
d) y e is a particular integral if m2 + mR + S = 0,
2 2
a) If y =xis a particular integral of (D +RD+S)y = 0 then, sinceDy=l andD y=0, R+Sx=0,
2 2
d) If y = emx is a particular integral of (D +RD+S)y = 0 then, since Dy= my and D y = m2 y,
2
(m +mR+S)y=0 and m2 +mR+S=0, b) and c) are special cases (m = 1, m = -1) of d),
2 3 3
2. Solve (D - - D + -)y 2x -1.
X X2
2 3 3
Here R+Sx = 0 and y = x is a particular integral of (D - -D + -)y 0.
X X2
2
dv 2 d v dv
The transformation y =xv, Dy= x - + v, Dy= x - + 2 - reduces the given equation to
dx dx2 dx
2
d v dv dv 3 3 1 dv 1
X - + 2- - 3- - -V + -V 2x -1 or 2 -
dx2 dx dx x x X dx X
2
d d v d dp 1 f-dx/x
Putting~= p, - p this becomes - - -p for which e = 1/x is an
dx dx2 - dx ' dx X
integrating factor, Then
p
X
= f (-2 - -)d.x
X
1
X
2
1
2 ln x + - + K,
X
p
dv
dx
2x ln x + 1 + Kx,
V :r
X
Jc2x lnx + 1+ Kx)dx x
2
lnx + x+ C1 x + C2,
2
and y = C1 x 3 + C2x + x 3 ln x + X
2
2 dy 2
x(2+4x+x )dx + (2+4x+x )y
2
x(2+4x+x )
Here, R + Sx = + X 0 and y = x is a particular integral of
2
x (x + 1)
the equation with its right member replaced by zero,
dy dv dv
The transformation y -:: xv. dx X dx + V, 2- reduces the given equation
d.x
2
2 d v dv 2 2
to X (X + 1) ( X- + 2 d.x) x(2+4x+x )(x dv + v) + (2+4x+x )xv
dx2 dx
x + 2 dv X +2
or
X + 1 d.x X + 1
. dv dp x +2 X +2 e
-Jc 1+ x+l
1
- -l c1x e
-x
Putting - = p, this becomes - - --p for which is an
dx dx X + 1 X +1 X +1
integrating factor, Then
EQUATIONS OF THE SECOND ORDER 115
-x -x
e
--p
X + 1 -J (x + 2)e
(x + 1/
dx p
dv
dx
V
;r and y
X
d\ _ d
4. Solve x (2x + 1)-1: + (x + l)y
dx2 dx
2x+l x+l
Here 1 +R +S = 1 - - x - + -x- = o and y ex is a particular integral of the equation
with its right member replaced by zero,
X dy
The transformation y = e v, reduces the
dx
1 dv 1 X
given equation to (x + 1 -)e
X dx X
1
. dv
Putting - p, this becomes
1
dp - -p (x + 1 - -)e
X
for which ~ is an integrating factor,
dx dx X X X
Then
p
X
J (ex+
xe
X
X
-e
2
X
)d.x e
X
+
e
X
X
+ K, p
dv
dx
xe
X
+ e
X
+ Kx,
V
.I_ xe
X
+ C1 x 2 + C2: and y C x 2 e x + C e X + xe 2x
X 1 2
e
dy
5. Solve (4.x-7)d.x + (4x-6)y o.
2 2 4x - 7 4x - 6 2
Here m + mR +S m - m-- + 0 when m = 2, and y = e x is a particular integral.
x-2 x-2
2 2
2x dy 2x dv 2x d y 2x d v 2x dv 2x
The transformation y = e v, e + 2e v, e + 4e + 4e v
dx dx dx2 dx2 dx
reduces the given equation to
~ ~
2
d~ d v dv
(x-2)(- + 4 + 4v) - (4x-7)(- + 2v) + (4x-6)v 0 or - - -- - o.
dx2 dx dx dx2 x-2dx
. dv dp _ 1
Putting - p, this becomes
dx x-2p
o. Then
dx
dv
p K(x - 2), V = _J_
dx 2X
e
2
d y _ d
6. Solve 2 tanx _J__ + 3y 2 sec x.
dx2 dx
2
By inspection, it is seen that y = sin x is a particular integral of (D - 2 tanx D + 3)y = 0,
The transformation y = v sin x reduces the given equation to
116 LI~EAR EQEA TIO NS WITH VARIABLE COEFFICIENTS
2
sin x dv
2(COS X - --)- 2 sec x, or 4 csc 2x,
COS X dx
dv dp
The substitution = p reduces this to - + 2(cot x - tan x )p 4 csc 2x for which an
dx dx
2
integrating factor is ti sin 2x, Then
dv
J sin2x dx = - ½ cos2x + tiK 1 , p
dx
y
V =
sin x
csc 2x + K cot 2x + C2 , and y ½ secx + Ci(cos x - ½secx) + G.,sinx,
,-.
- ~ dy + ( 1 + 2)Y
d\ 2 X
(. Solve xe
dx2 X dx X
Here R - -
2
X
. s 1 + -
2
x2
• s - tiR2 _ 1-
2
dR
dx
1 and u e
- ½JR dx e
Jdx/x
x.
dv dv
The transformation y = UV xv, X - + V, 2- reduces the given
dx d.x
X
equation to - + V e • a linear equation with constant coefficients, whose complete solu-
dx2
= ;r 1 X I X
tion is V C1 cos x + C2 sin x + - - e + 7e
X 2
D + 1
Thus, y
d2y 2
½(x + 2x)
8. Solve 2x dy + ( X 2 + 2) y e
dx
I 2
e
- l fR dx 2X
Here R = -2x, S 3, and u 2 -
e
I 2
The transformation y e2 x v reduces the equation to ex whose complete solu-
X
tion is V
I
4e •
I 2 I 2
Thus, y e x
2
(C1cos /3 x + C2 sin /3 x) + ,se 2 (x + 2 x)
3 3
9. Solve (D
2
- - D + -)y
X X2
2x - 1. (Problem 2.)
Here S - tiR 2 dR 3 9 3 -½jRdx - ½f -3dx/x 3/2

- '
2 dx and u e e - = X
2
4x
The transformation y = uv = x 3/2 v reduces the equation to
3 2x -1 3
-v or - - V a Cauchy equation.
4x
2 3/2 4
X
z
Put-ting x = e • we have
-Z/2 r• 3<'1 2
The complementary function is v = C1 e , e, 2 e , and a particular integral is
1 32/2 2 12 1 z, 2 2
V = - - - - - ( 2e - e ,, ) + e
tl _f) -3/4 J() - 3/2
3/2 • - 1;2 C 3, 2 3, 2 1/2

The complete solution is v yIx = C1 x + 2 x + x ln x + x
r 3 3 2
and y C1 x + L, 2 x + x ln x + x •
d\ dy 2 x2
10. Solve - 4x + 4x y xe
dx 2 dx
Here S - ti.R - ½ dR
2
2 and u e2 - r X Jx
e
2
dx
2
X
The transformation y ve reduces the equation to whose complete solu-
tion is
2 2
X
Then y ve ex (C1 cos i/2 x + C2 sin i/2 x) +
d2y X dy 2x 2(x + ex)

11. Solve - ( 1 + 4e ) - + 3e y e •
dx2 dx
When
dz
dx
= g fa; 2
= e
X
.
2
d z/dx
2
(dz/dx/
+ R(dz/dx) e
X
- (1 + 4ex)ex
(ex)
2
-4 A.
X
The introduction of z e as new independent variable leads to
2(x + ex)
d1.y A dy 2 Q d2y dy e 2ex 22
+ + a y or - 4 + 3y e e
dz 2 dz
(dz/dx)
2
dz 2 dz
e
2x
2 32 1 22 2 32 22
who.se complete solution is y = C1 e + C2e + e C1e + C2e - e
2
1) -419 + 3
Replacing z by ex, we have y
-4
Note. The choice of a 2 = 3 is one of convenience only. Taking a2 =1, dz = V3 ex and ,4
dx V3
1 22//3
The transformation z =~ex yields - e whose solution is
3
X X X
2
Then y = C1 ee + C2 e 3 e - e e , as before,
1
d y dy . 2
12. Solve - cot x - - srn x y COS X cos 3x.
dx2 dx
2 2
Here S =
. 2
-sin x and when -
dz
dx fl sin x,
d z/dx +R(dz/dx)
(dz/dx/
cosx + (-cotxXsinx)
-------- =
.
sin x
2
o.
11 B LINEAR EQlIATIOl\S \\"ITH YARIABLE COEFFICIENTS
Thus the introduction of z = - cos x as new independent variable leads to

2
d y
- - y = COS X = -z
dz 2
Upon replacing z by - cos x, we ha Ye y = (: 1 e -cos x t C2 e cos x - cos x.
2
13, Solve d y + ~ dy t _!_ y
dx2 X dx X4
2
dz iz/dx + H(dz/dx)
When - O. Thus the introduction of z =
dx X
(dz/dx/
d2y 2
as new independent variable leads to + y 2 + z whose complete solution is
dz 2
2
Upon replacing z by -1/x, we have y C1 cos(-l/x) + K sin(-1/x) + l/x
2
C\cos(l/x) t C2 sin(l/x) + l/x •
2 2
d y 1 dy 2 -x
14. Solve + (4x - - ) - + 4x y 3xe .
dx2 x dx.
2 2
dz d z/dx + H(dz/dx) 2 + ( 4x - 1/x ) 2x
When -
dx
Is 2x,
2
2. Thus the in-
(dz/dx) (2x/
d2y 3e-z
troduction of z x 2 as new independent variable leads to + 2~ + y whose com-
2 dz
dz 4/z
C1 e-z + C2 ze-z + 3/ 4 -z -1/2
plete solution is y e z
(D + 1/
2
3 -x
Upon replacing z by x 2 , we have + X e
15. Solve 2x -1. (Problem 2.)
2 3 3
a. The equation is equivalent to Dy - D(; y) = D(D - ;lY = 2x - 1.
2
Putting (D - ~)y = v, we have Dv = 2x -1 and v = x - x + K.
X
3 2
Now (D - ; ) y = x - x + K for which_!_ is an integrating factor. Then
x3
and y
b. The equation (xD

2
- 3D + _;)Y
3
is equivalent to (D - !) (xD - l)y = 2x -x.
2
Putting (xD - l)y v, we have

3
(D - -)v
X
2x 2 - x for which ~ is an integrating factor.
x3
~
V
Then 2 In x + +K and
X
3 2 3 2 2
(xD - l)y = v = 2x lnx + x + Kx or (D-1/x)y = 2x lnx + x + Kx •
Here 1/x is an integrating factor so that

2 2 2
y/x = J (2x lnx + 1 + Kx)dx = x lnx - ½x + x + K1 x + C2
2
and y = C1 x 3 + C2 x + x (1 + x lnx).
2
16. Solve [xD + (1-x)D- 2{1+x)]y = e-x(l-6.x).
-x
The equation is equivalent to [xD + (1 +x)] [D - 2]y e (1-6x).
1 -x 1
Putting (D-2)y = v, we have [xD+l+x]v = e-x(l-6.x) or (D + - + 1) v = e (- - 6).
X X
X X 2
Now xe is an integrating factor so that vxe = Jo -Gx)dx = x - 3x + K
-x
and (D - 2)y = V = (1-3x)e + Ke-x/x.
-2x
Here e is an integrating factor so that
-2x
ye
and y
2 2
17. Solve [(x + 3)D - (2x + 7)D + 2]y (x + 3)
X
e •
The equation may be written as [(x +3)D - 1] [D - 2]y
1 X
Putting (D - 2)y = v, we have or (D - - - ) v (x +3)e •
X +3
Using the integrating factor 1/ (x + 3), we have v/ (x + 3) = Jex dx ex + K
so that (D-2)y = v = (x+3)ex+K(x+3),

2
Using the integrating factor e- x, we have
-2x -x
ye J[ (x +3)e -x + K (x +3)e -2x] dx -Xe 4e -x + K' (-2xe -4e )+ C2
1 -2x 7 -2x
X X 2X
and y = - xe - 4e + C1 ( 2x + 7) + C2 e •
18. Show that the Riccati equation dy + y P(x) + /Q(x) = R(x), Q(x) I- O, is reduced to a
dx
1 du
linear equation of the second order by the substitution y =
Qu dx
Since dy 1 du 2
-(-) the substitution yields
dx Qu dx2 Qu2 dx
1 dQ du
or (P - - - ) - - RQu = O.
Q dx dx
dy 2 1 3 2
19. Use the procedure outlined in Problem 18 to solve dx ~ :; y + x y
2 2x
I du 2 du
The substitution y reduces the equation to
Qu dx x3u dx
120 LINEAR EQUATIONS WITH VARIABLE COEFFICIENTS
1 x3 1 du 1 2
- -•-U (} or -----XU (},
2x 2 X dx 4
In turn, the substitution : = .ff = I? = x reduces this equation to
_ l..z
1 2
- u 0 whose solution is + C.,e
4
1 (C1 e ½z
1 du 2
-~z
1 e
¾x2 ke
-nX2
- C2e ) C2
Then y
Qu dx x3 ½z _.Lz
2
X = -· -x2 +
x2 e~ ke-~-x2
where k
C1
C1e + C2e
20. Solve dy - (tan x + 3 cosx)y + / cos 2 x = - 2.

dx
1 du
The substitution y = reduces the equation to
Qu dx
du 2
(tanx -3cosx)- + 2ucosx = 0,
dx
In turn, the substitution ddxz __ ✓2 co s x

2
__ cos x, or z = sin x, reduces this equa-
2
tion to 0 whose solution is
1 du esl.n x + 2 ke2 sl.nx

Then y = sec x
Qu dx eslnx + ke2s1nx
Solve.
21. xy 11 - (x + 2)y' + 2y = 0 Ans,

2
22. (1 + X ) Y 11 - 2xy / + 2y 2
2 2
23, (x + 4)y 11 - 2xy' + 2y = 8 + X
2 2
24, (x + l)y 11 - (2x + 3)y' + (x + 2)y (x + 2x + l)e x
25, y 11 - 2 tanx y' - lOy = O
26, 3 X X X 2
y = C 1 x e + C2 xe + e (x + 2)
2
Zl, 2 3 2 2 -x /4
4x y 11
+ 4x y' + (x + 1) y = 0 y = /2" e (C 1 + C2 In x)
2 2 2
28, x y 11 + (x-4x )y' + {1-2x +4x )y
29, 3 . 2 C2COS 2
xy 11 - y' + 4x y 0 y = C1Sln X + X
EQUATIONS OF THE SECOND ORDER l 21
8 7 3
31. x y" + 4x y' + y = 1/x Ans.
32, (x sinx + cosx)y" - x cosx y' + y cosx = x
33, xy" - 3y' + 3y/x = x + 2
34, Solve Problem 21 by factoring,

2
35, [(x + l)D - (3x+ 4)D + 3]y = (3x+ 2)e 3x
37, XY II + 2y I + 4xy = 4
CHAPTER 19
Linear Equations with Variable Coefficients

MISCELLANEOUS TYPES
IN THIS CHAPTER various types of differential equations of order higher than the
first and with variable coefficients will be considered. There is no general
procedure comparable to that for linear equations. However, for the types
treated here, the procedure consists in obtaining from the given equation
another of lower order. For example, if the given equation is of order three
and if, by some means, an equation of order two, which is solvable by one of
the methods of th.e previous chapters can be obtained from it, the given equa-
tion can be solved.
DEPENDENT VARIABLE ABSENT. If the equation is free of y, that is, is of the form
1) dy x) = 0
dx' '
2
d d
the substitution dy
dx
p, 2 2
= dxP, •·· • will reduce the order by one.
dx
EXAMPLE. The equation of order three, is re-
2 2
2 d p dp 2 3 dy d
duced to x dx
2
+ 2p dx -:- 3xp + x = 0, of order two, by the substitution dx =p, dx~ =
If the given equation is of the form
2)
the substitution :_, •· •· will reduce the order by k.
See Problems 1-5.
INDEPENDENT VARIABLE ABSENT. If the equation is free of x, that is, is of the form
3)
2
the substitution dy = p d y = dp dy = P dp,
dx ' dx.2 dy dx dy
122
MISCELLANEOCS TYPES
will reduce the order of the differential equation by one.
dy d\
-
_ dp
-p _,
EXAMPLE. The substitution dx = p, reduces the
dx2 dy
2
d " dp 2 3 dn
equation yy'"-y"(y'/ - 1, of order three, to YP 2 _r + py(-) - p __!:. = 1, of order
dy2 dy dy
two,
See Problems 6-10.
LINEAR EQUATIONS WITH KNOWN PARTICULAR INTEGRAL. If a particular integral y == u(x)

of the equation
4)
is known, then the substitution y == uv will transform
5)
into an equation of the same order but with the dependent variable absent. In
turn, the order of this equation may be reduced by the procedure of the first
section of this chapter. Equation 4) is called the reduced equation of 5).
2
EXAMPLE. Since y = x is a solution of (D - xD + l)y = 0, the substitution y = vx,
2 2 2 2
dy dv d y d v dv 2 2x d v 2 -x dv e2 x
dx = X dx + V, - = x - + 2 - reduces (D -xD + l)y = e to - + -- - = -·
dx2 dx2 dx Jx2 X dx X
Here, the dependent variable v is missing and the procedure of the first section above applies.
See Problems 11-14.
EXACT EQUATIONS. The differential equation
dy
6) . . . . . . ' - , Y, x) Q(x)
dx
is called an exact equation if it can be obtained by differentiating once an

equation
7) ...... , dy
- , Y, x)
dx
of one lower order. For example, the equation

3y 2y"' + 14yy'y" + 4(y') 3 + 12y'y" == 2x
is an exact equation since it may be obtained by differentiating once the equa-
tion 3y2y" + 4y(y' / + 6 (y' / == x2 + C.
The linear equation 4) is exact provided

Pn - P~_ 1
+ p~'- 2
+ • • · • • · + (-l)nP~n) == 0, identically.
EXAMPLE. Consider the equation (x 3 - 2x)y"' + (Bx 2 - 5)y" + 15xy' + 5y = O in which P 3

5 , P2 = 15x an d P2I = 15, p 1 = Bx 2 - 5 and p"1 = 16, and P,O = x 3 - 2x and P,O"' = 6. The equa-
tion is exact since P3 - P~ + P~' - P~" = 5-15+16-6 = 0. The given equation is the exact de-
. t·1ve o f
riva ( x 3 - 2x )y ,, + (5x 2 -3)y I + 5xy= C•
12i LINEAR EQUATIONS WITH VARIABLE COEFFICIENTS
If equation 6) is not linear no simple test for exactness can be stated.

In this case, we show that 6) is exact by producing the equation of one lower
order from which it may be obtained by a differentiation.
If 6) is not exact, it may be possible to find an integrating factor.
Again, no general rule can be stated for determining an integrating factor.
See Problems 15-21.
SOLVED PROBLEMS
DEPENDENT VARIABLE ABSENT.

2
2
1. Solve 2 <.!...]_ - (dy) + 4 = O.
dx2 dx
. . dy 2dp=/-4 2dp = dx.
The subst1tut1on dx = p reduces the equation to or
dx p2 - 4
p-2
Integrating, t In - - =
p+2
x + In K; p - 2 = C1e2x,
p+2
p
and y = 2x - 2 In( l -C1 /x) + C2 •
3
2. Solve x d y - 2 d\ = o.
dx3 dx2
d2y --
The substitution q reduces the equation to o.
dx2
2
Then ln q In x + In K, and y
3. Solve 1.
The substitution reduces the equation to q dq = 1 and q2

dx
dy
Then q
dx
y or 105y
4. Solve o.
2
The substitution~ .
q re duces th e equa t 10n to dx + x dq
(dq)2 dx - q = O or q = x dq
dx + (ddxq)2 •
dx2
a Clairaut equation.
q = d2y dy
Then and
dx2 dx
y
MISCELLANEOUS TYPES 12S
d3 d2y dy -x
5. Solve (1+2x)____l + 4x
dx3 dx2
- (1-2x)-
dx
e
The transformation p = dy reduces the equation to

dx
-x
-x 1 -2x e
(1 .. 2x)p" + 4xp' - ( 1 - 2x)p e or
1 + 2x p 1 + 2x
Since 1-R +S = O, we use the substitution

p = e-xv, p' = e-x(v'- v), p" = e-x(v 11 -2v'+ v)
to obtain_(l+2x)v"-2v'=l or (1+2xi2v"-2(1+2x)v' = (1+2x), a Legendre linear equation.
The substitution 1+2x =et reduces the equation to [d)(J9-1)-d)]v = et or J9(J9-2)v = t/.
Then V
I t
K1 + K2e 2t - ~e = K1 + K2 ( 1 + 2x) 2 - ~1 ( 1 + 2x),
dy -X -x K 2 -X 1 -X
p dx = e v = K1e + 2 (1+2x) e - ~(1+2x)e ,
2
and y C1e-x + C2(4x + 12x + 13)e-x + C3 + ,t(2x + 3)e-x
-x 2 -x -x
or y Ae + B(x + 3x)e + C + ½xe •
INDEPENDENT VARIABLE ABSENT.

3
6. Solve y" = (y') + y'.
dp dp • dp=l+1.
The substitution y'= p, y" = p - reduces the equation to p - = p, + p or
dy dy dy
Then __!}f_ = dy arc tan p = y + K1 , and P = dy

p
2
+ 1
' dx
7. Solve yy" = 2(y')

2
- 2y'.
dp
The substitution y' = p, y" = p dp reduces the equation to p(y -
dy
- 2p + 2) o.
dy
Here p = O and y = C is a solution, or
..5!L 2 dy' ln(p - 1) In A y ,

2 2
P = A y
2 2
+ 1, or
dy
= dx,
p -1 y 2
l+A /
1
Then arc tan Ay X + K, arc tan Ay Ax+ B, and Ay = tan(Ax +Bf.
A
2 2
8. solve yy" - (y') y In y.
The substitution y' = p, Y " -- P dp reduces the equation to

dy
dp 2 2 In y dy,
YP - - P / In y or
dy y
dy
Then ± dx, and ±x + lnK.
2
y/2n y+C
126 Lll\EAR EQUATIONS WITH YARIABLE COEFFICIENTS
This may be written as ln y C1 e±x + C2 e +x or, finally,

since C1 and C2 are arbitrary constants.
2 2
9. Solve yyll + (y I) y •
dp dp 2 2
The substitution y' = p, yll = p - reduces the equation to py - + p = y for which y
dy dy
is an integrating factor. The solution of
2
py dp + /y dy =
3
y dy is 2// y4 + C 2
•
/24 + c2 12 1
y2 Kl2.
Now 12 p = 12: ± ---
y
whose solution is sinh- = ± 2x + Then
C
2 y2
sinh- 1 y ±12x+K, sinh(± /2 x + K)
C C
d2y 2y
10. Solve
dx2
e given that y=y'=O when x= 0,
Putting y 11 p dp, we have 2p dp = 2/Ydy whose solution is / = /Y + K.

dy
Using the initial conditions, 0 = 1 + K and K = -1. Now p = dy ± ~ which, by the
dx
2 dz
substitution e Y= z, becomes ±dx. The solution of this equation is arc tan /z-1
2z v1z="1
= ±x + C or, in the original variables, arc tan /4 2Y - 1 = ±x + C. Here, the initial con-
ditions require C=O so that /42G = tan(±x) = ± tanx and, finally,
2
e Y = sec 2x.
It should be noted that the form of the solution of the given equation depends on the sign
of the first constant of integration. If in p2 = e 2Y + K, K is positive and= A , we solve
2
dz = ± dx. and obtain

1 In ~ - A = ± x + C. Then h+A2 - A = Be±2Ax and
2z ~ 2A /z + A2 + A
A(l + B/2Ax) ~ 2
A2(l+Be2Ax)2
y'Z+A-, Since A is arbitrary, we may write z +A = and
1-Be±2Ax
4A2 B/Ax
obtain z = /Y or
(1-Be2Ax/
LINEAR EQUATIONS WITH KNOWN PARTICULAR INTEGRAL.
2
11. Solve x 3 (sinx)y"'- (3x sinx+x 3 cosx)y 11 + (6xsinx+2x cosx)y'- (6sinx+2xcosx)y
2
O,
By inspection it is seen that y =x is a particular integral.

,,, ,,,
By means of the substitution y = xv, y' = xv'+ v, Y II = XV II + 2V / , y = xv + 3v" I the
2 2
d\ d v
equation is reduced to sin x
dx3
- COS X
dx2
= o. In turn, the substitution -ddx2v = q
MISCELLANEOUS TYPES ]27
dq dq
reduces this equation to sin x - q cos x 0 or cotxdx.
dx q
Then In q = In sin x + In C, q
Thus, the solution is y
12. Solve
By inspection it is seen that y = x is a particular integral.

The substitution y =xv, y r = xv r + v, y " = xv " + 2v ' , y '" = xv ,,, + 3 V fl' y 1V = xv 1V + 4 v '"
4 2 1 4 2
reduces the equation to (x - 3x 3 + 6x - 6x)v v + (-x + 4x 3 - 12x + 24x - 24)v'" = Q.
d 3v
Putting - = q, this equation becomes
dx 3
2
3 3 dq 4 3 2 dq 4 3x - 6x + 6 )dx
x(x - 3x + 6x- 6)- + (-x + 4x -12x + 24x-24)q 0 or -+(- 1 + - - - - - - - - o.
dx q x x3 - 3x 2 + 6x - 6
2
3 2 d3v x 3 -3x +6x-6 x
Integrating, In q = x - 4 In x + ln(x - 3x + 6x - 6) + In A or q = - =A------e.
dx3 x4
3 2
x - 3x + 6x - 6 ex dx
Then
4
X
3 2
Aex _1_ ( x - 3x + 6x - 6) X 1 1 3 6 6
Ae - - (- - - + - - - ) ,
D+1 x4 D+1 x x2 x3 x4
1 2 6
Now D(~)
X
- -, D2(~) -, and D3 (~)
X 4
x2 X x3 X
1 1 3 6 6 1 2 1 3 1
so that -(--- + - - -) [~ + 3D(~) + 3D (~) + D3(~)] - ( D + l ) {-)
D +1 x x2 x3 x4 D+1 X X X X D+1 x
2 1
(D + 2D + 1) {-)
X
2 2
d v X - 2x + 2 X dv (x - l)ex
Thus, A e + B, A - - - - + Bx+ C,
dx2 x3 dx x2
X
V = r
X
= C1 :__ + C2x
X
2
+ C3x + c4, and
2 X
In this example, it is fairly easy to see that y = x, y = x , y = x , and y = e are particu-
lar integrals. Thus the complete solution could have been written down immediately.
13. Solve (2 sin x - x sin x - x cos x)y"' + (2x cos x - sin x - cos x)y" + x(sin x - cos x)y'
+ {COS X - sin x)y 2 sin x - x cos x - x sin x.
By inspection it is seen that y = x, y = ex, and y = sin x are particular integrals of the
reduced equation. We shall obtain a particular integral of the given equation using the method
of variation of parameters.
128 LINEAR EQCATIONS WITH YARIABLE COEFFICIENTS
We take
Then
and we set
Al
Now
and we set
Bl
Then
and we set
Cl 2 sin x - x cos x - x sin x.
Solving A), B), C) simultaneously, we obtain

- sin x + cos x and L 1 = cos x + sin x,
I -X
2e COS X
Thus, the complete solution is

X 1 2 3 1 . 1
y C1 x + C2 e + C3 sin x + -x sinx + -x cosx -x s1nx COS X,
2 2 2 2
2 2 2 2
14. Solve (x + x)y"' - (x + 3x + l)y" + (x + 4 + !)y' 3x (x + 1) .
By inspection it is seen that y =xis a particular integral of the reduced equation. The
substitution y = xv reduces the given equation to
2
(x + x)v"' - (x
2
- 2)v" - (x + 2)v' 3x(x + 1/
and, in turn, the substitution v' = u reduces this to
2 11 2
Al (x + x)u - (x - 2)u' - (x + 2)u 3x(x+l/.
X
Since the sum of the coefficients of the reduced equation of A) is identically zero, u=e
is a particular integral and we use the substitution
X
u = e w.
to reduce A) to
2 2 -x 2
(x + x)w" + (x + 2x + 2)w' 3xe (x + 1) •
Using the substitution w' = z, this becomes

2 1 2 -x 2
(x +x)z + (x + 2x+ 2)z 3xe (x + 1)
2 X
dz 2 1 -x X e
or -dx + ( 1 + - - --)z 3e (x+l) for which
+1
is an integrating factor,
X X + 1 X
Then z --
X
X
2 X
e
+1
J 3x
2
dx X
3
+ K1,
dw
dx
z = x(x + l)e
-x
+ K1
X +1
x2
e
-x
u
ex
= w -x e
2 -x
- 3xe
-x
- 3e-x + C1 e
-x
X
+ C2,
dv
dx
= u = -x
2
- 3x - 3 +
C1
X
+ C2e
X
.
4
X
and y xv
3
MISCELLA~Em,s TYPES 129
EXACT EQUATIONS.
15. Show that 0 is exact if and only if
P4 _ p'3 + p"
2
_ p"' p1v =
1 + 0 o.
Let the given differential equation be obtained by differentiating
R0 (x) y'" + R 1 (x) y" + R 2 (x) y 1 + R 3 (x) y = C1,
Since this differentiation yields R0 y 1v + (RI0 + R 1 )y ,,, + (RI1 + R 2 ) Y 11 + (R 2, + R 3 )y I + R31 Y 0,
we have P0 = R0 , P1 = R~ + R1 , P2 = R~ + R2 , P3 = R~ + R3 , and P 4 = R~.

Now p4 - p3' + p;' - p~" + fb1v R~ - (R; +R~) + (R~ +
1
R;) - (Rt+ R;) + R;v o.
Conversely, suppose P4 - P,3I + p2II -
p'" + P, 1V
1 0
_
- 0. since
the given differential equation is exact.
16. Solve
2
xy"' + (x + x + 3)y" + (4x + 2)y' + 2y = 0.
The equation is exact since P3 - P2I + p"

1
- P,"'
0
2-4+ 2-0 o.
2
Consider the left member xy"' + (x + x + 3)y" + (4x + 2)y' + 2y.
d
To obtain the first term we must differentiate xy". Now -(xy") = xy"' + y" and when this
dx
is removed, we have (x 2 + x + 2)y" + (4x + 2)y' + 2y, To obtain the first term of the result-
2 d 2 2
ing relation, we must differentiate (x + x + 2)y '· When dx (x + x + 2)y' = (x + x + 2)y" + (2x + l)y'
d
is removed, we have ( 2x + 1) Y I + 2y = - ( 2x + 1) Y, Thus the given equation is the exact de-
dx
rivative of
2
A) xy" + (x +x+2)y 1 + (2x+l)y
Since P2 - P; + P;' = (2x + 1) - (2x +1) + 0 = 0, we now treat the left member of' A)
precisely as we did the corresponding member of the original equation.
d 2 d 2
We remove dx(xy') =xy"+y' and have (x +x+l)y' + (2x+l)y dx(x +x + l)y.
Hence A) is the exact derivative of
½x(x+2)
a linear equation for which xe is an integrating factor,
Thus, the complete solution of the given equation is

½x(x+2)
xye
The following scheme will be found convenient.

]3() LINEAR EQUATIONS WITH YARIABLE COEFFICIENTS
,,, 2
xy + (x +x+3)y" + ( 4x + 2) YI + 2y 0
xy" xy"I + y"

2
(x + X + 2) Y 11 + (4x+ 2)y' + 2y
(x 2 +x+2)y' 2
(x + X + 2) Y 11 + (2x + l)y'
(2x + l)y' + 2y
(2x + l)y (2x + l)y' + 2y
Al xy" + (x 2 + X + 2) y I + (2x + l)y C1

xy' xy" + y'
2
(x + x + l)y' + (2x + l)y
2 2
(x +x+l)y (x +x+l)y'+ (2x + l)y
Bl xy' + (x 2 + x + l)y = C1x + C2,
3 2
17. Solve 2yd- y + 6 ~ dy 1
dx3 dx2 dx X
2
3 2
We write 2yd -) + 6 ~ dy
dx3 dx 2 dx
2 3 2
2yd-y 2y ~ + 2 ~ dy
dx2 dx3 dx2 dx
2
2(dy/ 4 ~ dy
dx dx2 dx
Thus, the given equation is exact, being obtained by differentiating

2
2yd-y
dx2
+ 2(dy/
dx
-f ~ dx
X
X
1
+ K1,
A second integration yields 2ydy

- In x + whose solution is
K1x + K2
dx
2
18. Solve (1+3xy )y"' + 9(/+2xyy')y" + 18y(y'/ + 6x(y 1 ) 3 = 6.
We write (1 + 3xy 2)y"' + 9y 2y" + 18xyy'y" + 18y(y'/ + 6x(y') 3

(1 + 3xy 2 )y 11 (1 + 3xy 2 )y'" + 3y 2y" + 6xyy'y"
6y 2y" + 12xyy'y" + 18y(y'/ + 6x(y') 3
6y2y' 6/y" + 12y(y'i2
12xyy'y" + 6y(y'i2 + 6x(y')3
6xy(y 1 / 12xyy'y" + 6y(y'/ + 6x(y 1 ) 3
The given equation is exact, being obtained by differentiating

2
(1+3x/)y" + 6y y' + 6xy(y'/ = 6x + K
(1 + 3xy2)y I (1+ 3x/)y" + 3y2y 1 + 6xy(y 1 ) 2
3y2y'
3
y 3y2y'
MISCELLANEOUS TYPES 131
2
and this equation is obtained by differentiating (1 + 3x/)y' + y 3 = 3x + Kx + C2 •
3 2
In turn, this equation is exact and we have xy 3 + y = x + C1 x + C2 x + C3 •
40x 3 - 4x 5 •
It is readily verified that this linear equation is not exact. To test whether or not it
has an integrating factor of the form xm, we multiply by xm to get
and write the condition

2 2
- (2xm + xm .. ) - 2(m + l)xm + (m + 3)xm .. + 5(m + 2) (m + l)xm - (m + 3) (m + 2) (m + l)xm
2
(m+2)xm+ + (m+2)(m-m 2 )xm = 0, for all values of x.
2
Then m = -2, and x- is an integrating factor. Using it, we have
2 2
xy"' + 5y" + (- - x)y' ( - + l)y 40x - 4x 3
X x2
xy'' XY/11 + y''
2
4y'' + (~ - x)y' - ( - + l)y
X x2
2 2
4y' + (; - x)y 4y" + (~ - x)y' - (- + l)y
X x2
2 2 4
and xy" + 4y' + (; - x)y 20x - X + K.
V 2 3 5
The transformation y reduces this equation to v 11 - v (D - l)v 20x - x + Kx,
2
X
2 X X 24 0 • s
and the complete solution is v X y C1 e + C2 e- - (l+D +D +D + 00
•)(20x,-x"+Kx)
C1 ex + C2e-x + C3 x + x 5 •
2
20. Solve 2yy"' + 2(y + 3y')y 11 + 2(y') = 2.
2
We write 2yy'" + 2yy 11 + 6y'y 11 + 2(y 1 ) 2
2yy"' + 2y'y"
2yy" + 4y'y" + 2(y'/
2
2yy' + 2(y') 2yy'' + 4y'y" + 2(y'/
and thus obtain by integration
2yy" + 2(y'/ + 2yy' 2x + K1
2yy' 2yy" + 2(y I i2
2yy'
y2 2yy'
By inspection, ex is an integrating factor; then
or
132 LINEAR EQl;ATIONS \\ITH YARIABLE COEFFICIENTS
3
21. Solve x cosy y"' - 3x siny y'y" - cosy y'' - x cosy (y') + siny (y'/ + x cosy y' - siny = O.
d sin y x cosy y' - sin y 1
Since -{--) the last two terms of the given equation suggest
dx X
X
2
X
2
as a possible integrating factor. Using it and integrating,

,, . ' 2
cosy y - sin y (y ) + sin y = C'i or cosyy
II •
-srny(y)
I 2
+ siny = C1 x.
X X
The substitution siny =z reduces this equation to z"+z=C 1 x whose complete solution
is
z = sin y = C\x + C2 cosx + C3 sin x.
solve.
2
22. y II + (y I ) + 1 = Q Ans. y = ln cos (x - Ci) + C2
2x- 3
2
23. {l+x )y 11 +2xy' y = C1 + C2 arc tan x + 1/x
2
24, xy" - y' = - 2/x - ln x Y = C1 x + C 2 + (x + 1) ln x
2 X 2 2
25, y"' + Y II = X y = C1 e - + C2 x + C3 + x (x - 4x + 12)/12
26. yyll + (y')3 0 x = C1 + C2 y + y ln y
2 2 2
27. yy" + (y') 2 y = 2x + C1 x + C2
28, yy" (y 1
/ (1 - y' cosy+ yy' sin y) x = C1 + C2 ln y + sin y
X 2X
29. (2x -3)y"' - (6x - 7)y" + 4xy' - 4y = 8 y = C1 x + C 2 e + C 3 e - 2
Hint: y =x is a particular integral of the reduced equation.
3 2 4 2
30. (2x -l)y"' - 6x y" + Gxy' = 0 y = C1 (x + 4x) + C 2 x + C3
2 2
31. yy'' - (y') = y lny
Hint: Use ln y = z •.
2 2
32, (X + 2y) Y II + 2 (y I ) + 2y 1 = 2 y(x+y) = x + C1x + C2
33, (1 + 2y + 3/ )y
111
+ 6y 1 [y'' + (y 1 i2 + 3yy 11
] = X
2 3 2
y + y + y = C1 x + C2 x +
4
C3 + x /24
34, 3x[/y'" + 6yy'y" + 2(y') 3 ] - 3y[yy" + 2(y 1 / ] - 2/x
3
Hint: 1/x
2
is an integrating factor. Ans. y = C1 x 3 + C2 x + C3 + x In x
35. yy 111 + 3y'y 11 - 2yy 11 - 2(y')
2
+ yy' = e
2x
Ans. y2 = C1 + C2 ex + C3 xex + e 2 x
Hint: e-x is an integrating factor. solve also using/= v.
2 2 2
36. 2(y+l)y 11 + 2(y') + y + 2y = O Ans. y + 2y = C1 cosx + C2 sinx
2
Hint: Use y + 2y = v.
CHAPTER 20
Appli<-ations of Lin<>ar Equations
GEOMETRICAL APPLICATIONS. In rectangular coordinates the radius of curvature R of

a curve y = f(x) at a general point on it is given by
X X
~ I
y>O, y">O y>O, y"<O
X X
y<O, y">O y<O,y"<O
Let the normal at the point be drawn toward the x-axis. It is clear from
the figures that the normal and radius of curvature at any point have the same
direction when y and d 2 y/dx 2 have opposite signs and have opposite directions
when y and d 2 y/dx 2 have the same signs.
PHYSICAL APPLICATIONS. OSCILLATORY MOTION. Consider a ball bobbing up and down at

the end of a rubber string.
If the other end of the string is held fixed and no external force is ap-
plied to the ball to keep it moving once it has been started, and if the mass.
of the string and the resistance offered by the air are such that they may be
neglected, the ball will move with simple harmonic motion
x = A cos wt + B sin wt
where xis the displacement of the ball at time t from its position of rest
or equilibrium.
APPLICATIONS OF LINEAR EQl-ATIONS
For simple harmonic motion:

2
a) The amplitude or maximum d:Lsplacement from equilibrium position is /A + B~
2 2
since when dx/dt = 0, tan wt = A/B, and x = /A +B •
b) The period or number of units (sec) of time for a complete oscillation is

2n:/w sec, since when t is changed by 2n:/w sec the values of x and dx/dt
are unchanged, while for any change of t less than this amount one (or both)
of x and dx/dt is changed.
c) The frequency or number of oscillations (cycles) per sec is w/2n: cycles/sec.
2
d x
d) The differential equ:cd ion of simple harmonic motion is m - = -kx, where 2
dt
k is a positive quantity. In the above illustration
2
d x
m - = - mw 2 (A cos wt + B sin wt) - kx
2
dt
where m is the mass of the ball and k = mw 2 •
If the above assumptions are modified so that the resistance of the air
cannot be neglected, the ball will move with free damped motion
st
x e- (A cos wt + B sin wt).
The motion is oscillatory as before but never repeats itself. Since the damp-
inR factor e- st decreases as t increases, the amplitude of each oscillation
is less than that of the preceding one. The frequency is w/2n: cycles/sec.
See Problem Sa.
If the resistance offered to the motion is sufficiently great, other cases
will arfae. See Problem Sb.
If in addition to a resistance, there is an external force acting on the
ball or the complete system is given a motion, the motion of the ball is said
to be forced. If the forcing function is harmonic with period 2n:/~. the mo-
tion of the ball is the result of two motions - a free damping motion which
dies out as time increases (called the transient phenomenon) and a simple
harmonic motion with period that of the forcing function (called the steady-
state phonomenon). See Problem 9.
HORIZONTAL BEAMS. The problem is that of determining the deflection (bending) of

a beam under given loadings. Only beams which are uniform in material and
shape will be considered. It is convenient to think of the beam as consist-
ing of fibers running lengthwise. In the bent beam shown, the fibers of the
upper half are compressed and those of the lower half stretched, the two
halves being separated by a neutral surface whose fibers are neither com-
pressed nor stretched. The fiber which originally coincided with the hori-
zontal axis of the beam now lies in the neutral surface along a curve (the
elastic curve or curve of deflection). We seek the equation of this curve.
I i X
u' 9--.:_ __
__ _,,,,_.,,.- __ .L _ _ _ _
A1---1 ....
P(x,y)
APPLICATIONS OF LINEAR EQUATIONS BS
Consider a cross section of the beam at a distance x from one end. Let AB
be its intersection with the neutral surface and P its intersection with the
elastic curve. It is shown in Mechanics that the moment M with respect to AB
of all external forces acting on either of the two segments into which the
beam is separated by the cross section (a) is independent of the segment con-
sidered and (b) is given by
A) EI/R = M.
Here, E = the modulus of elasticity of the beam and I = the moment of inertia
of the cross section with respect to AB are constants associated with the
beam, and R is the radius of curvature of the elastic curve at P.
For convenience, think of the beam as replaced by its elastic curve and
the cross sect ion by the point P. Take the origin at the left end of the beam
with the x-axis horizontal and let P have coordinates (x,y). Since the slope
dy/dx of the elastic curve at all of its points is numerically small,
[1 dy j 3/2
R
+ (-)
dx
=
1 approximately,
d2y d\
dx.2 dx.2
and A) reduces to
2
B) EI d y = M.
2
dx
The bending moment Mat the cross section (point P of the elastic curve)
is the algebraic sum of the moments of the external forces acting on the seg-
ment of the beam (segment of the elastic curve) about the line AB in the
cross section (about the point P of the elastic curve). we shall assume here
that upward forces give positive moments and downward forces give negative
moments.
EXAMPLE. Consider a 30 foot beam resting on two vertical supports, as in the figure
below, Suppose the beam carries a uniform load of 200 lb/ft of length and a load of 2000
lb at its middle,
y 30-X
!x~ ½x- {~½(30-x)- --~- ½(30-X) :1

15-x R X
0
p 2000
4000 200x 6000- 200x 4000
The external forces acting on OP are {a) an upward thrust at 0, x feet from P, equal to
one-half the total load, i.e., ½(2000+ 30,200) =4000 lb, and (b) a downward force of 200x
lb thought of as concentrated at the middle of OP and thus ½x feet from P. The bending mo-
ment at P is
2
M = 4000x - 200x(!x) = 4000x - 100x ,
To show that the bending moment at P is independent of the segment used, consider the
forces acting on PR: (a) an upward thrust of 4000 lb at R, 30 -x ft from P, (b) the load
U6 APPLICATIONS OF LINEAR EQUATIONS
of 2000 lb acting downward at the middle of the beam, 15-x ft from P, and (c) 200(30-x)
lb downward thought of as concentrated at the middle of PR, 1 (30 - x) ft from P. Then
M 4000(30 -x) - 200O(15-x) - 200(30 -x)· ~(30 -x)
4000x - 100x2, as before,
A beam is said to be fixed at one end if it is held horizontal there by

the masonry. In the example above the beam is not horizontal at O and is said
to be freely supported there.
SIMPLE ELECTRIC CIRCUITS. The sum of the voltage drops L

across the elements of a closed circuit is equal
to the total electromotive force E in the circuit.
The voltage drop across a resistance R ohms is Ri,
across a coil of inductance L henries is L di/ dt, C
and across a condenser of capacitance (capacity)
C farads is q/C. Here, the current i amperes and
the charge q coulombs are related by i = dq/ dt. We
will consider R, L, and C as constants. R
The differential equation of an electric circuit containing an inductance

L, a resistance R, a condenser of capacitance C, and an electromotive force
E ( t) is therefore
C') L di
+ Ri + g_ = E( t)
dt C
or, since i = dq/dt, di/dt = d 2q/dt 2 ,

d2q
+ R dq +
C) L g_ E( t)
2 dt C
dt
from which q = q(t) may be found.
dq
By differentiating C 1 ) and using 1, we have
dt
di i EI ( t)
D) + R - +
dt C
from which i = i(t) may be found.
SOLVED PROBLEMS
GEOMETRIC APPLICATIONS.
1. Determine the curve whose radius of curvature at any point P(x,y) is equal to the normal at P
and (a) in the same direction, {b) in the opposite direction,
3/2
[1 + (y'/] 2
a) Here or yy" + (y I) + 1 = 0,
y"
The equation is exact and an integration y ie Ids yy 1 + x - C1 = O or y dy + (x - C1 ) d.x 0.

APPLICATIONS OF LINEAR EQUATIOJ\"S
Intc>grating again, '5/,. ;\(x-C1i2 or a family of circleP

with centers on the x-axis,
I 2] ), 2
[1 ; (y ) 1,2
y [1 + (Y') 2 ] or yy " - (y ' ) 2 - 1 = o.
y"
The substitution y' p, y'' - p dp of chapter 19 reduces the equation to

dy
dp 2
YP - - p - 1 = 0 or __E__':!l!_ dy.
dy 1 + p2 y
dy
Then In y2 + In c~, ± dx.
Integrating, or
_1_ [e(±C1x+C 2 ) t e-(±C1 x+C 2 )].

y
2C1
The curves are catenaries and the equation may be written in the form
y
l
2
A [ e ( B ± X) I A + e
- (B ± X )/A l, where A and 13
MOTION OF A PENDULUM.
p
2. A pendulum, of length land mass m, suspended at P (see
figure) moves in a vertical plane through P. Disregard-
ing all forces except that of gravity, find its motion. e
Under the a,;sumptions, the center of gravity C of the
bob moves on a circle with center P and radius l. Let 0,
positive when measured counterclockwise, be the angle
which the string makes with the vertical at time t. The
only force i,; gravity, positive when measured downward,
and its component along the tangent to the path of the
bob is mg sin 0. Ifs denotes the length of arc C0 C, then
s = 10 and the acceleration along the arc i,;
2 2
d s d 0
- 2 l-
2
dt dt
2 2
d 0 d 0
Thus m• l = - mg sin 0 or l - = -g sin e.
dt 2 dt 2
d0
Multiplying by 2 dS and integrating, or
dt /2gcos0 + C1
This integral cannot be expressed in terms of elementary functions.
When A is small, sin 0 = 0, approximately. When this replacement is made in the original
~
2
differential equation, we have
d 0
dt2 + 78 0 whose solutiohis 0 =
fg
C1COSV' t + C2sinv, t.
This is an example of simple harmonic motion, The amplitude is /c~ + c; and the period is 2n: ff·
1:lg APPLICATIOJ\S OF LIJ\EAR EQUATIONS
MOTION ALONG A STRAIGHT LINE.
3. A mass m is projected vertically upward from O with initial

VPlocity v 0 • Find the maximum height reached, assuming that the Kv
resistance of the air is proportional to the velocity,
Take upward direction from U as positive, and let x denote :J m

the distance of the mHss from U at time t. The mass is acted
upon by two forcPs, the gravitational force of magnitude mg mg
X
and the resistunce of magnitude Kv = K dx each directed down.
dt
Hence, using
mass x acceleration net force,
0
-mg - K dx or k dx -g, where K= mk.

dt dt
Integrating, 1) x = ( \ + C2 e-kt - ; t, and the·n differentiating once with respect tot,
2)
When t = 0, x=0 and V = Vo, Then C1+C2=0, Vo - kC 2 -

g
k' and C1 - C2
Vo
k+ ~-
k2
1 -kt
Making these replacements in 1)' we have X = -(g + kv 0 ) (1 - e ) £ t.
k
/
-kt ~Ing+ kvo.
The maximum height is reached when 1, = O. From 2), e
k g
1 g g 1 0 g + kv 1 g_ g + kvo
Then the maximum height is x -(g+ kvolO- - - - ) -(- 1n - - ) -(Vo 1n - - - ) ·
k2 g+kvo k k g k k g
4. A mass m, free to move along the x-axis, is attracted toward the origin with a force propor-
tional to its distance from the origin. Find the motion (a) if it starts from rest at x = x 0
and (b) if it starts at x = x 0 with initial velocity v 0 moving away from the origin.
Let x the distance from the origin to the mass at time t.

2
d x 2
Then = - Kx or
2
+ /x = 0, where K = mk •
dt
Integrating, 1) x C1 sin kt + C 2 cos kt, and differentiating once with respect tot,
2) v = - kC2 sin kt+ kC 1 cos kt.
a) When t=0, x=x 0 and v=0. Then C1 =0 from 2), C 2 =x 0 from 1), and
x = x 0 cos kt.
x = Vo sin kt + x 0 cos kt.

k
In a) the motion is simple harmonic motion of amplitude x 0 and period 2IT/k.
In b) the motion is simple harmonic of amplitude

/v; + k
2
x; and period 2n/k.
k
APPLICATIONS OF LINEAR EQUATIONS
MOTION OF A COMPLEX SYSTEM.

5. A chain hangs over a smooth peg, 8 feet being on one side and 12 feet on the
other. Find the time required for it to slide off (a) neglecting friction and
(b) if the friction is equal to the ~eight of 1 foot of the chain. "
I
co
a) Denote the total mass of the chain by m and the length (feet) of the chain N
.....
which has moved over the peg at time t by x. At time t there are (8 -x) feet of
chain on one side and (12 +x) feet on the other. The excess (4 + 2x) feet on one
side produces an unbalanced force of (4 + 2x)mg pounds. Thus,
20
(4+2x)mg or
20
Solution 1.
2
Integrating -d x
- lL X
g
-5 ' we have X
dt 2 10
Differentiating once with respect to t' V =
When t "' 0, x = 0 and v =0. Then C1 =C2 =1 and x
Hence yg
(To
cosh
-1
fl In x +
2
+~
When x 8 ft has moved over the peg, t JI!- ln(5 + 2V6) sec.
dx
Solution 2, Multiplying the equation by - and integrating, we have
dt
2
10 dx -d x dx dx dx 2
= gx + 2g and 5(-) !gx2 + 2gx + C1,
dt dt 2 dt dt dt
When t =0, x; = 0 and dx/dt =0. Then C1 =0 and
dx 2 2
5(-) = ½gx + 2gx or dt
dt
(The positive square root is used here since x increases with t.)
Integrating, fwJ l<x

Jg
dx
+ 2) 2 - 4
(To
When t = 0, x = 0. Then C2 - Vg In 2, and V[iogo
g In X + 2 +2/x2 + 4x
as before.
2 2
d x mg d x
b) Here m-
2
(4+2x)mg - or 20 - (2x+3)g.
dt 20 20 dt 2
dx 2
Multiplying by and integrating, we have 10 (dxi2 gx + 3gx + C1 •
dt dt
When t = 0, x = 0 and v =0. Then C1 = 0, and

dx
dt J .fL(x 2 +3x)
10
or dt f! dx
/x 2 + 3x
Then t = [io
,Jg ln(x + ~2 + ~ ) + C2,
lW APPLICATIONS OF LINEAR EQUATIONS
When t = 0, X = 0, Then C2 = -fl- g

In~
2
and t fl- 2
In -(x + -3 + ~
3 2
X + 3x),
When x = B, t !¥ In
19 + 4/22
3
1. 4 sec.
6. A bead slides without friction along a straight

rod of neg I ig ible mass as the rod rotates with
constant angular velocity w about its midpoint 0.
Determine the motion (a) if the bead is initially
at rest at O and (b) if the bead is initially at
0 moving with velocity g/2w.
0 wt
Let the bead be x units from Oat time t. It ,..,..l;
is being acted upon by two forces, {i) gravity and
(ii) the centrifugal force mw 2 x acting along
the rod and directed away from 0. Since the rod
has rotated through an angle wt, the component
of the gravitational force along the rod ha:, magnitude mg sin wt; its direction is toward 0.
Hence
2 2
mW x - mg sin wt or W X - g sin wt.
Integrating, 1) X Differentiating once with re-
spect to t,
2) V =
a) When t = 0, X ""'0 and v = 0.

g
Then C1 + C2 = 0 from 1). C1 - C2 + _.lL = o from 2), C1 = -C2 = and
2w 2 - 4W 2
g -wt wt g g g
X = -(e - e ) + sin ltl t sinh wt + sin wt.
2
41.t1 2W 2 - 21.ti2 2w 2
b) When t =0, x=O and V = g/2W,

Then and X = g sin wt,
2w 2
SPRINGS.
7. A spring, for which k = 48 lb/ft, hangs in a vertical position with its

upper end fixed. A mass, weighing 16 lb, is attached to the lower end.
After coming to rest, the mass is pulled down 2 inches and released.
Discuss the resulting motion of the mass, neglecting air resistance.
Take the origin at the center of gravity of the mass, after coming
to rest, and let x, positive when measured downward, be the change in
position of the mass at time t. ~hen the mass is at rest the spring force
is equal to but opposite in direction to the gravitational force. The
net force at time t is the spring force -kx corresponding to the change
x in the position of the mass. Then
2
- 48x or 0, taking g 32 ft/sec •
APPLICATIONS OF LlNEAR EQUATIOl\S Ii l
Integrating, x = C1 sin v'96 t + C 2 cos v'96 t.

dx
Differentiating once with respect tot, V v'96(C 1 cos ✓96t - C2 sinv'96t).
dt
1 1
When t=0, x=-andv=0. Then and X cos 66 t.
6 6
. 27t
This represents a simple harmonic motion. The period 1s - 0.641 sec, the frequency is
196 1.56 cycles/sec, and the amplitude is~ ft.

27t 6
8. Solve Problem 7 if the medium offers a n:sistance (lb) equal to (a) v/61 and (6) 64v, where v
is expressed in ft/sec,
1 dx 1 dx
a) Here - 48x - or + - - + 96x o. Using the D nota-
64 dt 32 dt
2 1
tion, (D + - D + 96)x [D - (-0.0156+ 9.8i)] [D - (-0.0156-9.Si)]x 0.
32
-o.01set C
and X = e ( 1 cos 9.8 t + C 2 sin 9.8 t).
Differentiating once with respect tot,
-dx = v = e
-Q.0156t [ ,
(9.8C 2 -0.0156Ci)cos9.8t - (9.8C 1 + 0.0l56C 2 )sin 9.8t .
]
dt
Whent=0,v=0andx=l/6. Then C 1 =1/6, 0=9.8C 2 -0.0156C1, and C 2 =0.000265,
Thus,
0 0156
X e- • \~ cos 9,8 t + 0.000265 sin 9.8t ).
6
9 8
This represents a damped oscillatory motion. Note that the frequency=• = 1.56 cycles/sec
27t
remains constant throughout the motion, while the amplitude of each oscillation is smaller than
-Q,0156t
the preceding one due to the damping factor e At t = 0 the magnitude of the damping
-0.0156t
factor is 1. It will be 2/3 wh en e = 2/3 or after t = 26 sec. It will be 1/3 when
-0,0156t
e = 1/3 or after t = 70 sec.
2
16 d x 2
b) Here - - - 48x - 64 dx or (D + 128D + 96)x o.
32 dt"l. dt
-0, 76 t -127,24t
Integrating, X C1e + C2e

C
V = - 0 • 76 c1 e -o.7et - 12 7. 24 2 e
-121.24t
•
When t=0, x=l/6 andv=0. Then C 1 + C 2 = 1/6, -0.76C1-127.24C 2 =0,

C2 = -0.001, and
The motion is not vibratory. After the initial displacement, the mass moves slowly toward
the position of equilibrium as t increases.
I +2 APPLICATIONS OF LINEAR EQL'A TIONS
9. Solve Problem Ba if, in addition, the support of the spring is

given a motion y = cos 4t ft.
Take the origin as in Problem 8 and let x represent the change

in po.sit ion of the mass after t sec. From the figure, it is seen
that the stretch in the spring is (x -y) and the spring force
is -48(x -y) = -48(x - cos4t) lb. Hence,
0
2
16 d x 1 dx
- 4B(x - cos 4t) -
g dt 2 64 dt
2
or (D + _.!_ D + 96)x 96 cos 4t.
32
-O.Ol56t 96
Integrating, X e (C 1 cos9.8t+ C 2 sin 9. Bt) + cos 4t
D2 + D/32 + 96
-0.0156t
e (C 1 cos 9.Bt + C 2 sin 9.8t) + 0.0019 sin 4t + 1.2cos4t.

-0.0156t
v = e [(9.8C 2 - 0.0156Ci)cos9.8t - (9.8C 1 + 0.0l56C 2 )sin9.8t]
+ 0.0076 cos 4t - 4.8 sin 4t,
When t =0, v =0 and x = 1 + 1/6 = 7/6. Then C 1 = -1/30, C2 = -0.0008, and

-0.0156t .
x = e (-0.0333 cos 9.8t - 0.0008 sin 9.8t) + 0.0019 sin 4t + 1.2 qos 4t.
The motion consists of a damped harmonic motion which gradually dies away (transient phe-
nomenon) and a harmonic motion which remains (steady-state phenomenon). After a time the only
effective motion is that of the steady-state. These steady-state oscillations will have a
period and a frequency equal to those of the forcing function y = cos 4t, namely, a period of
2TT/4 = 1. 57 sec and a frequency of 4/2TT = 0. 637 cycle/sec.
2 2
The amplitude is /co.0019) + (1. 2) = 1. 2 ft.
10. A mass of 20 lb is suspended from a spring which is thereby stretched 3 inches. The upper end
of the spring is then given a motion y = 4(sin 2t + cos 2t)ft. Find the equation of the motion,
neglecting air resistance.
Take the origin at the center of gravity of the mass when at rest. Let x represent the
change in position of the mass at time t. The change in the length of the spring is (x -y),
the spring constant is 20/k = 80 lb/ft, and the net spring force is -80 (x -y). Then
2
d x
= - B0(x - 4 sin 2t - 4 cos 2t) or + 128x 512(sin 2t + cos 2t).
dt 2
Integrating, X = =
C1 cos vl28 t + C2 sin
. v128 t + -(sin =
128 . 2 t + cos 2 t).
31
256
V = - /i2s C 1 sin /i2s t + /i2s C 2 cos /i2s t + (- sin 2t + cos 2t).
31
When t =0, x =4
and v =0.
1 256
Then 4 = C 1 + ::, C 1 = -0. 129; and li2s C2 + - = o,
31
Hence, x = - 0.13 cos /i2s t - 0, 73 sin /i2s t + 4. 13(sin 2t + cos 2t).

APPLICATIONS OF LINEAR EQLATIOJ\'S
11. A mass of 64 lb is attached to a spring for which k = 50 lb/ft and brought to rest, Find the
position of the mass at time t if a force equal to 4 sin 2t is applied to it.
Take the origin at the center of gravity of the mass when at rest. The equation of motion
is then
2 2
64 d x
- + 50x 4 sin 2t or -d x2 + 25x 2 sin 2t.
32 dt2 dt
2
Integrating, X = C1 cos 5 t + C2 sin 5t + - sin 2t.
21
Differentiating once with respect tot, v = - 5C 1 sin 5t + 5C 2 cos 5t + t cos 2t.
Using the initial conditions x = 0, v = 0 when t = 0, and
x = - 0.038 sin 5t + 0.095 sin 2t.

The displacement here is the algebraic sum of two harmonic displacements of different periods.
12. A mass of 16 lb is attachpd to a spring for which .k = 48 lb/ft and brought to rest. Find the
motion of the mass if the support of the spring is given a motion y = sin /3g t ft.
Take the origin at the center of gravity of the mass when at rest and let x represent the
change in position of the mass at time t.
The stretch in the spring is (x -y) and the spring force is -48(x -y). Thus,
2
d x
48(x - sin 13g t) or - + 3gx 3g sin 13g t.
dt 2
Integrating, x = C1 cos 13g t + C 2 sin 13g t - ! 13g t cos 13g t,
and lJ - G\ ✓3g sin 13g t + C2 ✓3g cos 13g t - ! 13g cos 13g t +
3
gt sin 13g t.
2
Using the initial conditions x = 0, v = O when t = 0,
I ✓3g
X = 2 sin 13g t - - t cos 13g t.
2
The first term represents a simple harmonic motion while the second represents a vibratory
motion with increasing amplitude (because of the factor t), Ast increases, the amplitude of
the oscillation increases until there is a mechanical breakdown,
13. A cylindrical buoy 2 ft in diameter stands in water (density 62,4

lb/f"t 3 ) with its axis vertical. When depressed slightly and re-
leased, it is found that the period of vibration is 2 seconds.
Find the weight of the cylin~er.
Take the origin at the intersection of the axis of the cylin-
der and the surface of the water when the buoy is in equilibrium,
and take the downward direction as positive.
Let x (ft) denote the change in the position of the buoy at
time t. By Archimedes' Principle, a body partly or totally sub-
merged in a fluid is buoyed up by a force equal to the weight of
the fluid it displaces. Thus, the corresponding change in the
buoying force is 62. 4 7t( 1 x and i2
2009
- 62.47tx or + -- 7tx 0,
w
2
where W ( lb) is the weight of the buoy and g 32, 2 ft/sec •
APPLICATIONS OF LINEAR EQUATIONS
Integrating, x = C 1 sin /2009n/W t + C 2 cos /20091t/W t.

27t W = 2009
Since the period is 2 /nW/2009 = 2, 640 lb.
7t
/20091t/W
HANGING CABLE.
14. Determine the shape of a uniform cable which hangs under its
own weight, w lb/ft of length,
Choose the coordinate axes as in the figure, the origin

being at the lowest point of the cable. Consider the part be-
tween O and a variable point P(x,y). This part is in equilib-
rium under the action of (1) a horizontal force of magnitude
H at 0, (2) the tension T along the tangent at P, and (3) the
weight II' of OP.
Since OP is in equilibrium, all force acting horizontally
toward the right and all force acting horizontally toward the
left must be equal in magnitude, and, also, all force acting w
vertically upward and all force acting vertically downward.
Hence, T cos 0 = H, T sin 0 = II, and tan 0 = dy w

dx H
Now His constant, being due to the part OQ of the cable, while w= ws, wheres is the
length of OP. Thus,
d2y
dx2
1 dW
ff dx
w ds
Ii dx
~
Ii
/2 + (dy/dx) •
2
To solve the above equation, write!= p and obtain
dp
or --5!E_ ~ dx.
dx
~ H
Integratingbetweenthelimits x=0,p=0 and x=x,p=p,
-1 w dy
sinh p - X and p sinh ~ x.
H dx H
Integrating dy sinh ~ x dx between the limits x=0, y=0 and x=x,y=y,

H
H w
y -(cosh - x - 1), a catenary.
w H
If the origin had been taken at a distance H/w under the lowest point of the cable (thus
making H/w they-intercept of the curve) the equation of the curve would have been
H w
y - cosh - x.
w H
!\PPUC!\TIONS OF LINEAR EQliATIONS
HORIZONTAL BEAMS.
15. A horizontal beam of length 21 feet is freely supported at both ends. Find the equation of
its elastic curve and its maximum deflection when the load is w lb/ft of length.
y =0
wl wx wl
Take the origin at the left end of the beam with the x-axis horizontal as in the figure.
Let P, any point on the elastic curve, have coordinates (x,y).
Consider the segment OP of the beam. There is an upward thrust wl lb at 0, x ft from P,
and the load wx lb at the midpoint of OP, ½x ft from P. Then, since El d 2 y/dx 2 = M,
2
1) EI <!-_J_ wlx - wx(½x)
dx2
Solution 1. Integrating 1) once, EI dy ~ wlx 2 - ~ wx 3 + C1,

dx 2 6
1 3
At the middle of the beam X = l and dy/dx o. Then C1 = - - w l
3
and
2) EI dy ~ wlx 2 ~ wx 3 ~ wl 3 •
dx 2 6 3
1 1
Integrating 2), Ely ~ wlx 3 -
4
wx - wl 3x + C2, At 0, X = y = 0, Then C 2 = 0 and
6 24 3
3) y _w_(4lx 3 X
4
- sl 3x).
24 EI
Solution 2, Integrating 1) twice, Ely
At 0, x = y = 0, while at R, x = 21, y = 0. Using these boundary conditions in turn, we find

C 2 =0 andC 1 =-~wl 3 , asbefore.
3
The deflection of the beam at any distance x from O is given by -y. The maximum deflec-
tion occurs at the middle of the beam (x = l) and is, from 3),
5w z4
-y max ~
24 EI
16. Solve Problem 15 if there is in addition a load of W lb at the middle of the beam,
wl + ½W wx wl-r½W wl+ ½W wx 111l+tt

w w
l<x<2l
146 APPLICATIONS OF LINEAR EQUATIONS
Choose the coordinate system as in Problem 15. Since the forces acting on a segment OP of
the beam differ according as Plies to the left or right of the midpoint, two cases must be
considered.
When O< x < l, the forces acting on OP are an upward thrust of (w l + ½W) lb at 0, x ft from
P, and the load wx acting downward at the midpoint of OP, ½x ft fromP. The bending moment is
then 2
1) M = (wl + ½W)x - wx(½x) wlx + ½Wx - ½wx •
When l<x< 21, there is an additional force - the load W lb at the midpoint of the beam,
(x - l) ft from P. The bending moment is then
2
2) M = (wl + ½W)x - wx(½x) - W(x - l) wlx + ½Wx - ½wx - W(x - l).
2
Both 1) and 2) yield the bending moment M = ½wl + ½Wl when x = l. The two cases ·may be
treated at thA same time by noting that for 1)
w lx + ½Wx - ½wx 2 = w lx - ½wx 2 - pt ( l - x) + ½Wl
2 2
and for 2) wlx+ ½Wx - ½wx - W(x-l) = wlx - ½wx + ½W(l-x) + ½Wl. Then
2
3) El <!-_J_
dx2
wlx - ½wx
2
+½W( l -x) + ½Wl
with the understanding that the upper sign holds for O < x < l and the lower for l < x < 2 l.
Integrating 3) twice, EI y =
1
-
l 3
W X - -
1
WX
4 _ 1 w
+ - rr (
l - X)
3
+
1 •
- ,r11 X
2
+
,,
v1 X +
C2,
6 24 12 4
Using the boundary conditions x =y = O at O and x = 21, y = O at R,

3 4 4 3 3 2
C2=~Wl , 21C 1 +C 2 =-iwl + ~wl +~Wl -Wl , and C 1 =-~wl 3 -~Wl • Then
12 3 3 12 3 2
Ely ~ wlx 3 - -
1
wx
4
- ~ wl 3x -+ - 1 W( l -x) 3 + ~ Wlx 2 - -21
2
Wl x + ~ w1 3
6 24 3 12 4 12
~ wlx 3 -
1
wx
4
~ wl 3x - _!_wll-xl
3
+ ~ Wlx 2 ~ Wl 2 x + ~ w1 3
6 24 3 12 4 2 12 '
and y _w_(4lx 3 - X
4
8 l 3x) + --(3lx
W 2
- I 1- xi
3 2
6l x + l 3).
24 El 12EJ
Wl 3
The maximum deflection, occurring at the middle of the beam, is -Ymax -- + 0
24 EI 6EI
17. A horizontal beam of length l feet is fixed

at one end but otherwise unsupported, Find
the equation of its elastic curve and the X
maximum deflectioo when the uniform load is
X
w lb/ft of length.
Take the origin at the fixed end and let

P have coordinates {x,y). Consider the seg-
ment PR. The only force is the weight w( l - x) w{l-x)
lb at the midpoint of PR, ½( l -x) ft from P.
Then
2
EI<!-_]_ = -w(l-x),½(l-x) = -½w(l-x/. Integrating once, EI dy
dx2 dx
AtO: x=O <!:1.=o· C 1 =-~wl 3 and Eidy = ~w(l-x) 3 -~wl 3

'dx ' 6 dx 6 6 '
1
Integrating again, Ely --w(l-x)
4
~ wl 3 x + C2•
24 6
APPLICATIONS OF Lll\"EAR EQUATIONS 147
1 4 1 3 1 4
At o: x = y = o; _2_ wl 4
24 '
Ely = - -
24
w( l -x) - .-
6
wl x + - wl
24 '
and
The maximum deflection, occurring atR(x=l) is -y = ~ - wz4 • Note that this is not a
• max 8 EI
relative minimum as in Problem 16 but an absolute minimum occurring at an end of the inter-
val O ~x ~ l,
18. A horizontal beam of length 31 feet is fixed at one end but otherwise unsupported. There is
a uniform load of w lb/ft of length and two loads of W lb each at points 1 and 2l ft from the
fixed end. Find the equation of the elastic curve and the maximum deflection.
/~/
~
y
X ~- .,-(31-x) -~,
~
1/'.'./, ~-----x
21-x
21-x I
~t.
1/
/
/ -
/ ~-
-
. l-x
½(31-x)
½(31-x)
./ / // X
~
// X
/
0
P1
P2
~
~/
w
% (__ (31-x)w
Take the origin at the fixed end and let P have coordinates (x,y). There are three cases
to be considered according as P is on the interval (0 < x < l), ( l < x < 21), or (21 < x < 31). In
each case, use will be made of the right hand segment of the beam in computing the three bend-
ing moments.
When O < x < l, (P= P1 in the figure), there are three forces acting on P1 R: the weight
(31-x)w lb taken at the midpoint of P1 R, !(31-x) ft from P1 ; the load W lb, ( l-x) ft from
P 1 ; and the load W lb, ( 21 - x) ft from Pi. The bending moment about P 1 is
M1 -(3l-x)w•½(3l-x) - W(l-x) - W(2l-x) = - ½w(3l-x/ - W(l-x) - W(2l-x),
and 2
EI <!-_J_ W(l-x) - W(2l-x).
dx2
Integrating, EI dy = ~ w(3l-x} + ~ W(l-x/ + ~ W(2l-x/ + C1 •

dx 6 2 2
C1 = - ~ w l; - ~ Wl
2
At 0: x= 0 and dy/dx = O; then
2 2 '
dy 1 3 1 2 1 2 9 3 5 2
EI - = - w ( 31 - x ) + - W( l - x ) + - W( 21 - x ) - - w l - - Wl
dx 6 2 2 2 2 '
and
At O : x = y = O; then C2 and
1+8 APPLICATIONS OF LINEAR EQllATIONS
Al Ely
When l < x < 2 l, (P = P2 in the figure), the bending moment about f' 2 is
M2 - '
2 W(3
l - x) 2 - 11'(21 -x),
and
d2y I l 2
EI - 2 w(3 -x) W(2l-x). Integrating twice, we obtain
dx2
Ely
When x = l, B 1 ) and A) must agree in deflection and slope 1 so that C: 3 = C: 1 and C4 = C 2 • Thus,
1 4 1 3 9 3 5 2 27 4 3 3
B) Ely = --w(3l-x) - -W(2l-x) - -wl x- -Wl x + -wl t -Wl.
24 6 2 2 8 2
When 21< x< 31, (P=P 3 in the figure), the bending moment about P 3 is '
M3 = - .,-w(3 l -x) 2 ,
2
and d y
EI -
, l 2
- .,-w(3 -x) • Then
dx2
1 4 1 4 9 3 5 2 27 4 3 3
Cl Ely = - - W ( 3l - X) + •C5X + CR --w(3l-x) - -wlx--1\lx+ -wl + -Wl
24 24 2 2 8 2 '
since, when x = 21, there must be agreement with B) in deflection and slope.
A), B), C) may be written in the form
y 0 < x < l,
y _w_(12lx 3 - 54l 2 x 2 - x 4 ) + w 3 - 6 lx 2 -
2
3 l x + l 3 l' < X < 21,
24 EI 6E/x
y _w_ 12lx 3 - 541 2/ - X

4
) + ____!__ (3 l 3 -
2
51 x), 21 ~ X < 3 l.
24 EI\ 2 EI
1 4 3
The maximum deflect ion, occurring at R (x = 3 l), is -y max = -(Blwl + 4BWl ).
B EI
Note that the elastic curve consists of arcs of three distinct curves, the slopes of each
pair of arcs at a junction point being equal.
19. A horizontal beam of length l ft is fixed at both ends. Find the equation of the elastic
curve and the maximum deflection if it carries a uniform load of w lb/ft of length.
p
~
y
X
---~--~
1/~?
Re
~//~
0
½x
p
R
~
X
½wl wx ½wl
Take the origin at the left end of the beam and let P have coordinates (x, y).
The external forces acting on the segment OP are: a couple of unknown moment K exerted
by the wall to keep the beam horizontal at 0; an upward thrust ot ½wl lb at 0, x ft from P;
and the load wx lb acting downward at the midpoint of OP, ½x ft from P. Thus,
APPLICATIONS OF LINEAR EQUATIOJ\"S 149
K + ½wlx - ½wx 2
Integrating once and using x =0, dy/dx = 0 at 0, EI dy Kx + ~ wlx 2 1

- wx
3
dx 4 6
At R: x ~ l, dy/dx = 0 since the beam is fixed there. Then

1
1 2
K = - - wl , and EI dy 2
- - wl x + ~ wlx 2 1
- wx
3
12 dx 12 4 6
Integrating and using x = y = 0 at 0,

1 wx 2
Ely = -2 wl x
2 2
+ 2 w lx 3 - -wx
4
and y
2 2
- - (2lx - l - x ) •
24 EI
24 12 24
The maximum deflection, occurring at the middle of the beam (x = ½l), is -Ymax
384 EI
20. Solve Problem 19 when in addition there is a weight W lb at the middle of the beam.
~ / ----x- ~
~ ~
~~
·~:j ½x
X - ½L
~
K( 0 R
/$
X
1//'//j P1 P2
wx
½(wl + W) wx ½(wl + W)
w
Using the coordinate system of Problem 19, there are two cases to be considered: from x = 0
to x = ½l and from x = ½l to x = l.
When 0<x< ½l, the external forces on the segment to the left of Pi(x,y) are: a couple of
unknown moment K at O; an upward thrust of ½(wl + W) lb at 0, x ft from P 1 ; and the load wx
lb, ½x ft from P 1 • Thus,
2
EI <!:.....I_ 1 1 2 1 2
K + -(wl + W)x - -wx = K + ~wlx -wx +
dx2 2· 2 2 2
Integrating once and using x = 0, dy/dx = 0 at 0,
EI dy = Kx + ~ wlx 2 - ~ wx 3 + ~ Wx 2 ,
dx 4 6 4
Integrating and using x = y = 0 at 0,
A) Ely = ~ Kx
2
+ 2 wlx 3 - - wx
4
+ _2_ Wx 3 •
2 12 24 12
When ½l < X < l, there is in addition the weight W lb at the middle of the beam, (x - ½l)
ft from P2 • Thus,
2
EI<!:.....I_ 1 1 2
K + -wlx
2
-WX
2
+ ~Wx
2
- W(x - ½l). Integrating twice,
dx2
APPLICA TIOJ\S OF LINEAR EQCATIOJ\S
l 2 1 3 1 4 1 3 1 ,3
Ely = - Kx + - wlx - - wx + - Wx - - W(x - 2 1) + C1 x + C2 •
2 12 24 12 6
When x = ½l, the values of y and dy/dx for B 1) must agree respectively with those for A). Thus,
C1 = C 2 = O and
1 2 1 3 1 3
B) Ely - Kx + - wlx - wx
4
+ _!_ Wx 3 - W(x - ½l) •
2 12 24 12 6
To determine K, use x = ½l, dy/dx = O in A 1 ). Then

1 2 1
and K = - -wl - -Wl. Substituting in A) and B),
12 8
2 2 1 4 1 2
Ely - _!_ wl x + _.!_ wlx 3 - wx + _.!_ wx3 Wlx and
24 12 24 12 16
_w_ (2lx 3 l 2 x2 4 W 3 2
y - X ) + - - ( 4x - 3 lx ) , 0 ~ X ~ ½l,
24 EI 48 EI
1
Ely - _!_ 2 2
wl x + _.!_ wlx 3 - - 1 wx 4 + - Wx 3 -1 w(x - ½l )3 _.!_ wzx2 and
24 12 24 12 6 16
_w_(2lx 3 2 2
y z2x2 - x4) + _W_<l3 6l x + 9lx - 4x 3 ). ½l ~ X ~ l.
24 EI 48 EI
1 4 3
The maximum deflection, occurring at the middle of the beam, is -y max - - ( w l + 2Wl ),
384 EI
21. A horizontal beam of length l ft is fixed at one end and freely supported at the other end.
(a) Find the equation of the elastic curve if the beam carries a uniform load w lb/ft of length
and a weight W lb at the middle. {b) Locate the point of maximum deflection when l = 10 and
W = 10w,
~
'/
-~l-x
/'. ½{l-x) l-x

,/ ½{l-x)
~.- ½l-x
~ 0
P1
R X
W
{
L(l-x)w
P2
( l-x)w
s
~ -ymax.
Take the origin at the fixed end and let P have coordinates (x,y). There are two cases to
be considered.
When O < x < ½l, the external forces acting on the segment P1 R are: an unknown upward thrust
S lb at R, ( l -x) ft from P 1 ; the load w( l -x) lb at the midpoint of P1 R, ½( l -x) ft from P 1 ;
and W lb, (½l-x) ft from P 1 , Thus,
2
EI<!:_}_ = S(l-x) - w(l-x)•½(l-x) W(½l-x)
dx2
Integrating pnce and using x = 0, dy/dx = 0 at 0,
1
- -S{l-x) + -1 w(l-x) 3 + -lW'l
EI dy 2 1
C2 -x) 2 + -Sl 2
dx 2 6 2 2
APPLICATI01'S OF LINEAR EQl:ATIONS
Integrating again and using x =y =0 at 0,
When ½l < x < l, the forces acting on P 2 R are the unknown upward thrust S at R, ( l -x) ft
fromP 2 and the load w(l-x) lb, ½{l-x-) ft fromP 2 , Thus,
1 2
S( l -x) -w(l-x) and
2
1 3 1 4
Ely = - S(l-x) - w( l -x) + C1 x + C'2,
6 24
When x = ½l, the values of Ely and EI dy as given by A) and B 1 ) must agree, Hence, C 1 and C2
dx
in B 1 ) have the values of the corresponding constants of integration found in determining A),
and B 1 ) becomes
2 2 3
~wl 3 - ~Wl )x - ~sz 3 + 2-wl
4
~S(l-x) 3 -
4
B) Ely = 2-w(l-x) + (~Sl - + 2-wz •
6 24 2 6 8 6 24 48
To determine S, use x = l, y =O at R in B); then S ~wl + ~W. Making this replace-

8 16
ment in A ) and B),
y and
It is clear that the maximum deflection occurs to the right of the midpoint of the beam.
When l = 10, W= 10w, the equation immediately above becomes
w 4 3 2
y - - (- 2x + 25x + 450x - 6000x + 10000).
48 EI
Since dy Oat the point of maximum deflection, we solve

dx
3 2
8x - 75x - 900x + 6000 = 0
for the real root x = 5,6, approximately. Thus, the maximum deflection occurs at the point
approximately 5.6 ft from the fixed end.
ELECTRIC CIRCUITS.
22. An electric circuit consists of an inductance of O. 1 henry, a resistance of 20 ohms and a con-
denser of capacitance 25 microfarads (1 microfarad= 10- 6 farad), Find the charge q and the
current i at time t, given the initial conditions (a) q = 0,05
coulomb, i = dq/dt = 0 when t = 0, (b) q = 0.05 coulomb, i = _6
- 0. 2 ampere when t = O. C = 25 x 10 f
-6
Since L = o. 1, R = 20, C = 25-10 , E( t) 0,
-c:
__,
d2q q
L - +
R dq
+ E( t) 0
dt 2 dt C "
.....;i
reduces to d2q
+ 200 dq + 400,000q = o.
dt 2 dt R = 20 ohms
1S2 APPLICATIONS OF LINEAR EQUATIONS
Integrating, q e
-1oot =
(A cos 100 v39 t + B sin 100 v39 t),
=
dq 100 e -lOO t [ ( /39 B - A) cos 100/39 t - ( /39 A + B) sin 100/39 t].

dt
5
a) Using the initial conditions q = 0.05, i =O when t = 0, A = 0,05 and B = 0,0 = 0.008,
✓39
-100 t
Hence, q e (0, 05 cos 624. 5 t + O. 008 sin 624. 5 t)
and i = - 0.32 e-10ot sin 624.5 t.
b) Using the initial conditions q = 0.05, i = -0. 2 when t = O, A = 0,05 and B 0,0077.
-100 t
Hence, q e (0,05 cos 624.5t + 0,0077 sin 624.5t)
-100 t
and e (- 0,2 cos 624,5t - 32,0 sin 624.5t)
Note that q and i are transients, each becoming negligible very quickly.
23. A circuit consists of an inductance of 0, 05 henry, a resist-

ance of 20 ohms, a condenser of capacitance 100 microfarads,
and an emf of E = 100 volts. Find i and q, given the initial
conditions q = 0, i = O when t = O.
-.c:
2 lJ")
q 0
Here 0. 05 <j__J_ + 20 dq + 100 0
dt2 dt -6 ~
100· 10 "
---l "
Cl:;
2
d q d
or + 400 _J_ + 200, 000q 2000.
dt 2 dt
-6
Integrating, q = e
-200 t
(
A
cos 400t + B sin 400t) + 0,01. C= lOOX 10 f
dq -2oot[ .
i = - = 200e (-A+ 2B) cos 400t + (-B-2A) sin 400t].
dt
Using the initial conditions: A =- 0.01, -A+ 2B = 0, and B =- 0.005.
-200 t
Then q e (- 0.01 cos 400t - 0.005 sin 400t) + 0.01
200
and i = 5 e- t sin 400 t.
Here i becomes negligible very soon while q, for all purposes, becomes q 0.01.
24. Solve Problem 23 assuming that there is a variable emf of

E = 100 cos 200t
E ( t) = 100 cos 200 t.
'v
The differential equation is now
d2q -.c: I@
+ 400 dq + 200, 000q 2000 cos 200 t. Then lJ")
0
.,::
0
dt2 dt c5 0
-200 t A "
---l
N
q e ( cos 400t + B sin 400t) + 0.01 cos 200t "

Cl:;
+ 0.005 sin 200t and
-200t
i = e [(-200A + 400B) cos 400t + (-200B- 400A) sin 400t] -6
C =100 x 10 f
- 2 sin 200t + cos 200t,
APPLICATIONS OF LINEAR EQUATIOJ\S
Using the initial conditions: A = - 0.01, - 200A + 400B + 1 = 0 and B = -0.0075. Then
-200t
q e (- 0.01 cos 400t - 0.0075 sin 400t) + 0.01 cos 200t + 0.005 sin 200t
and
-200 t
i = e (- cos 400t + 5.5 sin 400t) - 2 sin 200t + cos 200t,
Here the transient parts of q and i very quickly become negligible. For this reason, when
the transients may be neglected, one needs find only the steady-state solutions
q = 0.01 cos 200t + 0.005 sin 200t and i = cos 200t - 2 sin 200t.
The frequency 200/2n cycles/sec of the steady-state solutions is equal to the frequency of
the applied emf. (See also Problem 25.)
25. For a circuit consisting of an inductance L, a resistance R, L

a capacitance C, and an emf E{t) = Ea sin wt, derive the
formula for the steady-state current
Ea R . X
-<= Ea sin(wt -- 0),
i =
z z srn wt - z- cos wt)
z
C R
where X = Lw - 2, and 0 is determined
Cw
X R
from sin 0 = - and cos 0
z z -----"v----
d2q R dq + q E = Ea sin wt
By differentiating L - + Ea sin wt
dt 2 dt C
dq
and using i - , we obtain
dt
2
d i di 2
1) L R + - (LD + RD + 1/C) i wEa cos wt.
2 + dt C
dt
The required steady-state solution is the particular integral of 1):
wEa cos wt wEa

------"----cos wt
2 1
LD + RD + 1/C RD - (Lw - -)w
Cw
wEa (RD+ Xw)
- - - - - - cos wt Eo ( R sin w t - X cos wt)
R2 + X2
Ea E
- (R . wt - X
- srn - cos wt ) _.£ sin(wt - 0).
z z z z
Xis called the reactance of the circuit; when X = O, the amplitude of i is greatest (the
circuit is in resonance). Z, called the impedance of the circuit, is also the ratio of the
amplitudes of the emf and the current. 0 is called the phase angle,
At times t n/2w, 3n/2w, • • • • • • the emf attains maximum amplitude, while at times given
=
n/2 + 0 3n/2 + 0
by wt - 0 = n/2, 3IT/2, " " " , that is, when t = --w- , w , • • • • • • the current attains
maximum amplitude, Thus the voltage leads the current by a time 0/w or the current and volt-
age are out of phase by the phase angle 0.
Note that 0 = 0 when X = O, that is, 0 = O if there is resonance,
APPLICATIONS OF LI1'EAR EQUATIONS
26. The circuit consisting of an inductance L, a condenser of

E= Ea cos wt
capacitance C, and an emf Eis known as an harmonic oscil-
lator, Find q and i when E = Ea cos wt and the initial con- .--------1 ' v i - - - - ~
ditions are q = qa, i = ia when t = O.
Since R = O, the differential equation is L C

d2 E
-3.. + .3.... ~ cos wt.
dt 2 CL L
There are two cases to be considered:

I _1_ 1
(a) w and (b) w
la la
1 1 Ea 1
a) q A cos t + B sin - - t + cos wt
la L D2 + 1/CL
la
1 1 EaC
A cos - - t + B sin - - t + cos wt
2
la /CL 1- w CL
1 1 1 Ea CW
and - ( - A sin - - t + B cos - - t) sin wt.
2
✓CL la la 1-w CL
EaC
Using the initial conditions: A qa - 2
and B la ia. Then
1-w CL
q (qa -
1
EaC ) cos - - t + ./CL ia 1
sin - - t + EaC cos wt
✓CL ✓CL
2 2
1- w CL 1- w CL
1 1 EaC ) sin -1- t EaCW

and ! = ia cos - - t --(qa - sin wt.
2 2
/cI la 1- w CL /a 1-w CL
d2q 2 Ea
b) Here - + w q cos wt.
dt 2 L
Then q A cos wt + B sin wt + Ji.2_ t sin wt

2Lw
Ea 1
and = W(- A sin wt + B cos Wt) + - (- sin wt + t cos wt).
2L w
Using the initial conditions: A = qa and B ia/w.
ia J,;
Then q qa cos wt + sin wt + _a_ t sin wt
w 2L w
Ea 1 .
and !a cos wt - qaw sin wt + -(-smwt + t
2L w
COS Wt),
Note that in (b) the frequency of the emf is the natural frequency of the oscillator, that
is, the frequency when there is no emf. The circuit is in resonance since the reactance X =
1 1 E t
Lw - - = 0 when w = - - , The presence of the term -2...... cos wt, whose amplitude increases
Cw ./ci 2L
with t, indicates that eventually such a circuit will be destroyed.
APPLICATIOJ\S OF LIJ\EAR EQUATIONS
27, Determine the curve for which the radius of curvature is proportional to the slope of the
tangent,
/, :2 2
Ans, y ± ( v',d1
k - (x + C1 ) 2 + k Inv' k - (x + C1) - k ) + C2
_
X + C1
28, A 6 inch pendulum is released with a velocity of 1/2 rad/sec, toward the vertical, from a po-
sition 1/5 rad from the vertical. Find the equation of motion,
Ans. 0 = ! cos St - 2_ sin St
5 16
29, A particle of mass m is repelled from O with a force equal to k >Q times the distance from
0. If the particle starts from rest at a distance a from 0, find its position t sec later.
Ans. X
I
= .,-a(e
Vk/m t
+ e
-/k?m t )
30. If, in Problem 29, k =m and a= 12 ft, determine a) the distance from O and the velocity when
t = 2 sec, b) when it will be 18 ft from O and its velocity then.
Ans, a) x = 45,1 ft, v = 43,5 ft/sec; b) t = 0.96 sec, v = 13.4 ft/sec
31. A chain hangs over a smooth peg, 8 ft on one side and 10 ft on the other. If the force of
friction is equal to the weight of 1 ft of chain, find the time required for it to slide off.
ln(l 7 + 12 ✓2) sec
3
Ans.
✓g
32. When the inner of two concentric spheres of radii r 1 and r 2 , r 1 < r 2 , carries an electric
charge, the differential equation for the potential V at any point between the two spheres
at a distance r from their common center is
+ o.
d/ r dr
Solve for V given V=V 1 when r=r 1 and V=V 2 when r=r 2 •
= V2 r 2 (r - ri) - V1 ri(r- r 2 )
Ans. V
33. A spring is such that it would be stretched 3 in. by a 9 lb weight. A 24 lb weight is.attached
and brought to rest. Find the equation of the motion if the weight is then
a) pulled down 4 in. and released,
b) pulled down 2 in. and given an upward velocity of 2 ft/sec.
c) pulled down 3 in. and given a downward velocity of 4 ft/sec.
d) pushed up 3 in. and released,
e) pushed up 4 in. and given an upward velocity of 5 ft/sec.
Ans, =
1
4V3t +
V3 sin
. 4✓3t,
4 cos
a) X C) X
3
5/3 . ;;;
d) X -= - 41 cos 4 /3 t, e) X
1
3 cos
;;;
4v3 t -
12- sin 4v3 t
1.)6 APPLICATIONS OF LINEAR EQUATIONS
34, A spring is such that it would be stretched 3 in. by a 30 lb weight. A 64 lb weight is at-
tached and brought to rest. The resistance of the medium is numerically equal to 8 dx/dt lb.
Find the equation of the motion of the weight if
a) it is started downward with velocity 10 ft/sec.
b) it is pulled down 6 in. and given an upward velocity of 10 ft/sec.
Ans. a) x =
5/14 e
-2t
sin 2114 t, b) x = e -2t( 1 cos 2vl4t
,.;
- 9/14 sin
. ~ )
2v14t
14 2 213
35, A spring is such that it would be stretched 6 in. by a 3 lb weight. A 3 lb weight is attached
and brought to rest. The weight is then pulled down 3 in. and released. Determine the equa-
tion of motion if
1 3 . 4 .
a) an impressed force ; sin 6t acts on the spring. Ans. x = cos St - sin St + sin 6t
4 7 7
1 1 .
b) an impressed force ; sin 8 t acts on the spring, Ans. x = (1- 4t)cos St+ sin St
4 8
36, A beam of length l ft is fixed at one end and otherwise unsupported. Find the equation of
the elastic curve and the maximum deflection if there is a uniform load of w lb/ft of length
and a load W lb at the free end,
2 1 3
Ans. y = _w_(4lx 3 -6l x -x ) + _!,_(x 3 -3lx ),
2 2 4 4
-y = --(3wl + sWl )
24E1 6E1 max 24£1
37, A beam of length 21 ft is freely supported at both ends and carries a uniform load of w lb/ft
of' length. Taking the origin at the midpoint (low point) of the beam, find the equation of
the elastic curve and the maximum deflection, Compare with Problem 15.
"1 // -u
Hint:cy - l ( l - x ) - 2/ w( l -x) 2 = I
2 w(
12 - x2) and y=y'=0 whenx o.
w 2 2 4 5wz4
Ans. Y = --(6 l X - X ) , y = --
24£1 max 24E1
38. A beam of length 31 ft is freely supported at both ends, There is a uniform load of w lb/ft
of length and loads of W lb at a distance l ft from each end. Taking the origin as in Prob-
lem 37, find the maximum deflection.
l 3l
2 <x < 2 ; and M
1 4 3
Ans. Yma = --(405wl + 36BWl )
X 384£1
39, A circuit consists of an inductance of 0.05 henry, a resistance of 5 ohms, and a condenser
of capacitance 4(10)- 4 farad. If q = i = 0 when t = 0, find q and i in terms oft when a) there
is a constant emf = 110 volts, b) there is an alternating emf = 200 cos lO0t. Find the steady
state solutions, in b).
Ans. a) q = e
- 50 t
(- -
11 - 11119 .
cos 50 /I9t - - - sin 50119 t) + __!_!_ •
44 v'l9 e - 5ot sin' 50v'!.9t
250 4750 250 19
-5ot 16 ,r.,-- 12v'l9 .
b) q e (- - c o s 50vl9t- - - s 1 n 50v'19t) + _!_(4 cos lOOt + sin lOOt),
170 1615 170
= e -5ot (- 40 cos 50vl9t

r.-:: + 1640v'l9si·n r,;:; t) + 40
50 v1::,
19 · l00t)
(cos 100t - 4 sin
17 323 17
40, solve Problem 39 after replacing the 5 ohm resistance with a 50 ohm resistance.
-53 t 947 -53t -947t
Ans. a) q - 0,047e + 0,0026e- t + 0,044, i=2,46(e -e )
-53 t -947 t
b) q - o.o1se + 0,005e + 0,034 sin lO0t + 0,014 cos lO0t,
0, 98e - 53 t - 4, 43e - 947 t + 3, 45 cos lO0t - 1,38 sin lO0t
CHAPTER 21
Systems of Sin1ultaneous Linear Equations
IN PREVIOUS CHAPTERS, differential equation.s involving only two variables have been
treated. In the next several chapters, equations involving more than two va-
riables will be considered. If but one of the variables is independent, the
equations are ordinary differential equations; if more of the variables are
independent, the equations are called partial differential equations. In this
chapter we shall be concerned with systems of ordinary linear differential
equations with constant coefficients such as
2(D-2)x + (D- l)y = et

or A') d
{ (D+3)x+ y =0, where D =
dt
and
dx dy
- + - + y = 1
dt dt
{ Dx , (D d)y c I
dx dz 1
B) - - - + 2x + z 1 or B ) (D + 2)x - (D - 1) z = 1
dt dt
(D+l)y+ (D + 2)z 0
dy dz
- + - + y + 2z 0
dt dt
in which the number of simultaneous equations is equal to the number of depen-

dent variables.
THE BASIC PROCEDURE for solving a system of n ordinary differential equations in

n dependent variables consists in obtaining, by differentiating the given equa-
tions, a set from which all but one of the dependent variables, say x, can be
eliminated. The equation resulting from the elimination is then solved for
this variable x. Each of the dependent variables is obtained in a similar
manner.
dx dy t dx
EXAMPLE. Consider system A): 1) 2- + - - 4x - y = e , 2) - + 3x + y = O.
dt dt dt
Solution 1.
First, we note that the general solution x = x(t), y = y(t) of this system will also
satisfy
dy
3) + - = 0
dt
obtained by differentiating 2). Moreover, multiplying 1) by -1, 2) by -1. 3) by 1, and add-

ing, we obtain
4) + =
t
X - e
which is also satisfied by x = x(t), y = y(t), This latter differential equation, being
free of y and its derivatives, may be solved readily; thus,
Vi7
1S8 SYSTEMS OF SIMULTANEOUS LINEAR EQCATIONS
X = C1 cos t + C 2 sin t -
2
D +1
To find yin a similar manner, we differentiate 1) to obtain
2 d2 dy
2~+---1.._4dx t
5) e
dt 2 dt 2 dt dt
and between this and equations 1),2),3) eliminate x and its derivatives. However, it is
simpler here to proceed as follows. From 2), we have
y - -dx - 3x = -(-C 1 sin t + C2 cost - -!,e t ) - 3( C 1 cost + C2 sin

.
t - 2, e t )
dt
Thus, x = C 1 cos t + C 2 sin t is the

general solution.
When the equations are written in the D notation, there is a striking similarity between
the procedures used here and the method of solving a system of n equations in n unknowns.
This is due to the fact, noted in previous chapters, that the operator D may at times be
treated as a variable (letter).
t
Solution 2. Consider the system A1 ): 1) 2(D-2)x+(D-l)y e
2) (D + 3)x + y o.
Proceeding as in the case of two equations in two unknowns x and y, we multiply 2) by
d
D-1. Actually, we operate on 2) with D-1 = (dt - 1), to get
(D-l)(D+3)x + (0-l)y = 0
and subtract 1) from it to obtain

t
-e t
2
[(D-l)(D+3) - 2(D-2))x or (D + l)x = -e .
Now this is 4) above as might have been anticipated, since operating on 2) with D -1 is
equivalent to differentiating 2) and adding -1 times 2) as in the previous solution. The
general solution is obtained as in solution 1.
system A 1 ) we obtain
2(D-2)
1 D+3
or (D
2
+ l)x
t and
-e
The first of these equations is 4) above, and the second would have heen obtained hy the
procedure rejected in solution 1. We shall now show why it was rejected. When the two equa-
tions are solved, we have
6) x = C1 cos t + C2 sin t - ½et and
We know from solution 1 that 6) and 7) contain extraneous solutions. To eliminate them (that
is, to reduce the number of arbitrary constants), we substitute in 2) and see that
(C2 + 3C1 + C3)cos t + (3C 2 -C 1 + C4 )sin t = 0
for every value oft. Thus,
C3 = - (3C1 + C 2 ) and C4 = C1 - 3C2 •

When these values are substituted in 6) and 7), we obtain the general solution found above.
SYSTEMS OF SIMULTANEOCS LINEAR EQUATIONS
THE NUMBER OF INDEPENDENT ARBITRARY CONSTANTS appearing in the general solution of

the system
f 1 (D)x + g 1 (D)y :c h 1 (t)
f 2 (D)x + g 2 (D)y :c h 2 (t)
f 1 (D)
is equal to the degree in D of the determinant 6
£2 (D)
provided 6 does not vanish identically. If 6 ~ 0, the system is dependent;

such systems will not be considered here.
2(D - 2)
For the system A1 ), 6 =
D+3
The degree (2) in D agrees with the number of arbitrary constants appearing
in the general solution.
The theorem may be extended readily to the case of n equations inn depen-
dent variables.
SOLVED PROBLEMS
1. Solve the system: 1) (D-l)x+Dy = 2t+
2) ( 2D + 1) x + 2Dy = t.
subtracting twice 1) from 2), we have 3x = - 3t - 2, substituting x = - t - 2/3 in 1), we
4 1 2 4
obtain Dy = 2t + 1 - (D - l)x = t + and y = - t + - t + C1.
3 2 3
2
The complete sulution is x=-t- ,
3
1
Note that ID- DI is of degree 1 in D and there is but one arbitrary constant.
2D + 1 2D
2. Solve the system: 1) (D + 2)x + 3y 0

2t
2) 3x + ( D + 2) y 2e •
2
Operating on 1) with D+ 2, multiplying 2) by -3, and adding: (D + 4D - 5 ) x = - 5/ t
1 , t -5t 8 2t
Then From 1), Y = - -(D + 2)x -C 1 e + C2 e + -e
3 7
3. Solve the system: 1) (D -3)x + 2(D + 2)y 2 sin t
2) 2(D + l)x + (D - l)y cos t.
Operating on 1) with D-1 and on 2) with 2(D+2), we have
3) (D-l)(D-3)x + 2{D-l)(D+2)y (D-1)[2 sint] 2 cost 2 sin t
4) 4(D+2)(D+l)x + 2(D+2){D-l)y 2(D+2)cost 4 cos t - 2 sin t •

subtracting 3) from 4) and noting that (D - 1) (D + 2) (D + 2) (D - 1), since the operators
have constant coefficients,
160 SYSTEMS OF SIMULTANEOl:S LIJ\EAR EQlJATlONS
2 2 2
[4(D +3D+2) (D -4D+3)]x (3D + 16D + 5)x 2 cos t
2
and X := -----cost
C1 e -5t + c·2 e -t/3 + ( 8 sin t + cos t) / 65.
From 2), (D - l)y cos t • 2(D + l)x
4 t
cost+ 8C 1e- 5t - 3c'2 e- l3 - (18 cost + 14 sin t )/65
-5t
8C1e 34c,,ze -t/3 + (47 cos t 14 sin t)/65.
Then ye
-t f (8C1e -bt 34 C,,ze -4t/3 +
47 cos t - 14 sin t -t
65
e )dt
- !3 C e -bt + C,ze -4t/3 + 61 sin t - 33 cos t e -t + c3

"1 130
61 sin - 33 cos t t
and y - 34 C1e -5t + C,,ze -t/3 +
130
+ C3e
since the degree of 6 is 2, the general solution has but two arbitrary constants, Hence,
when these expressions are substituted for x and yin 1) it is found that C3 = 0, Then
4 -5t C -t/3 61 sin t - 33 cos

Y = - - C1 e + 2e +
3 1~
is the general solution,
2 2t
4. Solve the system: 1) (D - 2)x - 3y = e
2
2) (D + 2)y + x : : o.
Find the particular solution satisfying the conditions x=y=l, Dx=Dy=0 when t=0,
2 4 2 2 2t
Operating on 1) with D to obtain D x-2D x-3D y=4e and making the replacements
2t 2 4 2t
= 2x + 3y + e from 1) and D y = -x - 2y from 2), we have (D - l)x 6e ,
t -t C . 2 2t and, using 1),

Then X := C1 e + C2e + , 3 cost + C4 sin t + - e
5
1 2 2t 1 2t
y -[(D - 2) x - e ] - - e
3 15
Note that x could also be obtained by the use of determinants, Thus,
2 2t
D - 2 -3 e 3
4 2t
X = or (D - l)x 6e etc,
2 2
1 D +2 0 D +2
2
When t 0, X C1 + C 2 + C3 +
5
= 1 and Dx = C1 - C2 + C4 + -4 0,
5
1 1 1 2
y - -(C1 + C2) - C3 - - = 1 and Dy = - -(C1-C2) - c4 - -
15
o.
3 15 3
Then C1 3/4, C2 7/4, C3 = -19/10, c4 1/5, and the required particular solution is
1 t - t 1 2 2t
X = -(3e +7e) --(19cost -2sint) +-e
4 10 5
1 t -t 1 1 2t
y - -(3e + 7e ) + -(19 cos t - 2 sin t) - - e
12 10 15
SYSTEMS OF SIMULTANEOUS LINEAR EQUATIONS 161
t 2 2
5. Solve the system: 1) (D+ l)x + (D-l)y = e , 2) (D +D+l)x + (D -D+l)y = t .
2
2
Operating on 1) with D +D + 1 and on 2) with D + 1, and subtracting, we have
2 t 1 2 3 t
2y = t + 2t - 3e and y = t + t - e •
2 2
2
Operating on 1) with D -D + 1 and on 2) with D -1, and subtracting, we have
2 t l 2 1 t
2x = t - 2t + e and x = - t - t + - e •
2 2
D+l D-1
Note that ::: 2 is of degree O in D; hence, there are no arbitrary
constants in the solution.
2 2
6. Solve the system: 2) D y + mx = O.
Operating on 1) with D2 and substituting D2 y = -m 2 x from 2), we obtain

~(1 ±
Then D =
-+ /2 i)
mt /./2 - mt /./i (,
C
and x = e (C 1 cos mt/12 + C2 sin mt/✓2) + e 3 cos mt/✓2 + C4 sin mt/✓2).
Substituting for x in 1) and solving,
y __!_ D 2 x =
- e mt/./i (,
C cos mt / v2
2
r,; - C, · / r,;2
1 srn mt v.::) + e
-mt/./2 (C srn
3
· mt / v.::
r,;2 - C cos mt / v2).
4
r,;
m2
7. Solve the system: 1) (D

2
+ 4)x - 3Dy = o, 2) 3Dx + (D
2
+ 4)y = o.
2
Operating on 1) with D + 4 and on 2) with 3D, and adding, we have
2 2 2 2
[(D + 4/ + 9D ]x = (D + 16) (D + l)x = o and x = C1 cos 4t + C2 sin 4t + C3 cost + C4 sin t.
2
Operating on 1) with -3D and on 2) with D + 4, and adding, we have
2 2
(D + 16)(D +l)y =0 and y = K1 cos4t + K2 sin4t + K3 cost+ K4 sint.
To eliminate the extraneous solutions, substitute for x and yin 1). We have
-12C 1 cos 4t - 12C 2 sin 4t + 3C3 cost + 3C4 sin t + 12/i. 1 sin 4t - 12K 2 cos 4t + 3K 3 sin t
- 3K 4 cost = o
The complete solution is: X = C1 cos 4t + C2 sin 4t + C3 cos t + c4 sin t'

y C 2 cos 4t - C 1 sin 4t - C4 cost+ C3 sin t.
8. Solve the system: 1) Dx + (D + 1) y = 1

2) (D+2)x - (D-l)z 1
3) (D + l)y + (D + 2)z o.
subtracting 3) from 1), we have 4) Dx - (D + 2)z = 1 which is free of y.
Operating on 2) with D and on 4) with D + 2, and subtracting, we have (5D+ 4)z = -2; then
z = - 21 +Ce-
"1 4t/5• Substituting for z in 3), (D+l)y = -(D+2)z -- 1 - _65 C1e_4t/5
_ ·, then
_
162 SYSTEMS OF SIMULTANEOUS LINEAR EQUATIONS
Substituting for yin 1), Dx 1 - (D + 1) y then X =
D D+l 0
2
Since D+2 0 -(D - 1) - (5D + 9D + 4) is of degree 2 in D, there are but two ar-
0 D+l D+2
bitrary constants in the general solution. Substituting for x and z in 2), we obtain
3
and, hence, C3 Thus,
4
is the general solution.
9. Solve the system: 1) (D+l/x + 2Dy + 3Dz = 1
2) Dx + z 0
3) X - Dy Dz = o.
F'ind the particular solution for which x ccz = 1, y =O when t =O.
First, operate on 2) with D to obtain

Next, add twice 3) to 1) and subtract 4) to get (2D + 3)x = 1; then
and
From 2), z = -Dx = 23 C,1e -3t/2 •

1 -3t/2 9 C -3t/2 1 13 C1e -3t/2
From 3), Dy x - Dz - + C1e + 4 ,1e - + then
3 3 4
1 13 C - 3t/2
y -t
3 6 1e + Cz,
2D 31)
2
Since 0 2D + 3D is of degree 2 in D, there are 2 arbitrary constants
1 -D -D
1
and the general solution is y = -t z
3
13
When t = O: y 0 and C2
9
Thus, the required particular solution is

1 13 -3t/2 13
y = - t - -e + -,
3 9 9
Note that a particular solution satisfying a given set of initial conditions cannot always
be found. For example, there is no solution satisfying the conditions x = 1, y =z =O when t =O
since x=l, y=O contradicts x=l/3 + 2z/3. Similarly, y=O, z=l, dx/dt=l when t=O con-
tradicts dx/dt = -z.
SYSTEMS OF SIMULTANEOUS LINEAR EQUATIONS l(d
SUPPLEMENTARY PROBLE:HS
Solve the following simultaneous equations.
t · t + 3e 2t;1 5
10. Dx - (D + l)y = - e Ans. x = ( C1 - C2 )cos t + (C
. 1 + C, 2 )srn
2t 2 ti t .
x+(D-l)y = e y = C1 cos t + C 2 sin t + 2e 5 + e /2
C1 -3C
2 . 3C1+C2 t t2
11. (D + 2)x + (D + l)y x = --- sin t - - - - cos - + t + 3
5 5
2 2
5x + (D +3)y t y = C 1cos t + C 2 sin t + 2t - 3t - 4
t 1 -~ t [ , 13
12, (D + 1 )x + (2D + 7)y = e +2 x = C1e cos(t+C 2 ) - sin(t+C2 )] +
2 17
t
- 2x + (D + 3)y = e - 1 -~ t . 2e t 3
y = C1e srn(t+C~) + - - +
13 17
5 2t 3 1
13, (D - l)x + (D+3)y ~ e -t - 1 X
C
= 2 1e
-1t/5
+ -e + - t
17 7 49
2t
(D+2)x + (D + l)y e +t
1 2t 1 -t 1 26
y -_ 3 C 1e -1t/5 - - e + -e + - t
17 2 7 49
2
14, (D + 16)x - 6Dy 0
2
6Dx + (D + 16)y 0
2 . 2
15, (D + 4)x + y = sin z X = C1 cos ( ✓2z + c,,.) + C3COS ( ✓3z + c4) + ½ cos 2z
2
(D2+1)y-2x cos z y = -2C 1 cos( ✓2z + C,;) - C 3 cos(v'3z + C4 ) + ½-½cos 2z
2 2 t t -t
16, (D + D + l)x + (D + l)y e x = - e - 2e - C1
2 2 -t t -t
(D + D)x + D y = e y 2e + e + C1
t t t
17, (D-l)x + (D + 2)y 1 + e x = - 1 + te /2 + C2 e
t t -2t
(D + 2)y + (D + 1 )z 2 + e y e /6 + C1 e
t
W-l)x+ (D + 1 )z 3 + e z 2 + e ti 4 + C' 3 e -t
CHAPTER 22
Total Differential Equations
THE DIFFERENTIAL EQUATIONS

A) (3x 2 / - ex z) dx + (2x 3 y + sin z )dy t (y cos z - ex)dz o.
2
B) (3xz + 2y)dx + x dy + x dz = 0,
C) y dx + dy + dz = 0,
being of the general form
P(x,y,z,•••,t)dx + Q(x,y,z,··•,t)dy + •···· + S(x,y,z,··•,t)dt = 0,
are called total differential equations.
It may be verified readily that A) is the exact differential of
f(x,y,z) = x 3 y 2 - exz + ysinz = C,
C being an arbitrary constant. Such an equation is called exact.
Equation B) is not exact, but the use of x as an integrating factor yields
(3x 2 z t 2xy)dx t x 2 dy + x 3 dz = 0
2
which is the exact differential of x 3 z + x y = C. Equations A) and B) are
called integrable.
Equation C) is not integrable; that is, no primitive
1) f(x,y,z) = C
can be found for it. It will be shown later (Problem 32) that for SUCii equa-
tions a solution 1) can be obtained consistent with any prescribed relation
g(x,y,z) = 0 of the variables.
THE CONDITION OF INTEGRABILITY of the total differential equation

2) P(x,y,z)dx + Q(x,y,z)dy + R(x,y,z)dz = 0
is
3) P(oQ - oR) + Q(~R - oP) + R(oP - oQ) 0, identically. See Problem 1.
oz oy dx oz oy ox
EXAMPLE 1. For equation B),

oP oP Q=x, oQ=l oQ=O;
P = 3x z + 2y - = 2 - = 3x· and
• oy • oz • OX • oz
3) becomes (3xz+2y)(O-O)+x(2x-3x) +x 2 (2-1) =O-x 2 +x 2 =0, The equation

is integrable.
EXAMPLE 2. For equation C),

oP oP o· oR oR
P=y, -=1, -=0; Q=1 oQ = oQ =
R = 1. ox = oy = 0 • and 3) becomes
oy oz • ox oz '
y(O -0) + 1(0- 0) + 1(1- 0) -/:- O. The equation is not integrable.

TOTAL DIFFERENTIAL EQCATIONS 165
THE CONDITIONS FOR EXACTNESS of 2) are
oQ oR oR oP
4)
oz= oy' clx cl z
EXAMPLE 3. For equation A),

oP oP
3x 2 y 2 -e 6x 2 y,
x;
p z, = -ex;
cly clz
Q 2x 3 y + sin z, oQ = 6x 2 y, oQ = cos z;
ox oz
R -- y cos z - e
X
cly
,
oR
ox
eX . oR
COS Z I
and the conditions 4) are satisfied. The equation is exact.
EXAMPLE 4. From Example 1 it is readily seen that 4) is not satisfied;

hence, equation B) is not exact.
TO SOLVE AN IN""EGRABLE TOTAL DIFFERENTIAL EQUATION in three variables:

a) If 2) is exact, the solution is evident after, at most, a regrouping of
terms. See Problem 3.
b) If 2) is not exact, it may be possible to find an integrating factor. See
Problems 4-6.
c) If 2) is homogeneous, one va:r;iable, say z, can be separated from the others
by the transformation -x =uz, y =vz. See Problems 7-10.
d) If no integrating factor can be found, consider one of the variables, say
z, as a constant. Integrate the resulting equation, denoting the arbitrary
constant of integration by ¢(z). Take the total differential of the inte-
gral just obtained and compare the coefficients of its differentials with
those of the given differential equation, thus determining ¢(z). This pro-
cedure is illustrated in Problem 13. See also Problems 14-16.
PAIRS OF TOTAL DIFFERENTIAL EQUATIONS IN THREE VARIABLES. The solution of the si-
multaneous tot~l differential equations
5) P1 dx + Q1 dy + R1 dz 0
6) P2 dx + Q2 dy + R2 dz 0
consists of a pair of relations
7) f(x, y, z)
8) g(x,y,z)
To solve a given pair of equations:

e) If 5) and 6) are both integrable, each may be solved by one or more of the
procedures a)-d). Then, 7), say, is the complete solution (primitive) of
5), and 8) is the complete solution of 6). See Problem 18.
f) If 5) is integrable but 6) is not, then 7), say, is the complete solution
of 5). To obtain 8), we use 5),6),7) to eliminate one variable and its dif-
ferential, and integrate the resulting equation. See Problem 19.
g) If neither equation is integrable, we may use the method of Chapter 21.
166 TOTAL DIFFERENTIAL EQUATIONS
treating two of the variables, say x and y, as functions of the third vari-
able z.
At times it may be simpler to proceed as follows: Eliminate in turn dy and

dz (or any other pair) between 5) and 6) to obtain
P1 01 01 R1 R1 P1 01 R1
dx - dz = 0, dx - dy = 0
P2 02 02 R2 R2 P2 02 R2
and express them in the symmetric form

dx dy dz
9) - = = - ,
X y z
01. R1 R1 P1 P1 01
where X = A.. y = A.. z = A.. A..-/:- o.
02 R2 R2 P2 P2 02
(Note that this is the procedure for obtaining the symmetric form of the
equations of a straight line when the two-plane form is given.)
Of the three equations
Y dx = X dy, Xdz = Zdx, Z dy = Y dz
given by 9), any one may be obtained from the other two. Hence, in obtaining
9), we merely replace the original pair of differential equations by an equiv-
alent pair, that is, any two of 9 1 ).
If two of 9 1 ) are integrable, we proceed as in e), See Problem 20.
If but one of 9 1 ) is integrable, we proceed as inf), See Problem 21.
If no one of 9 1) is integrable, we increase the number of possible equa-
tions. By a well known principle,
dx = dy = dz = 11 dx + m1dy + n 1 dz = 1 2 dx + m2 dy + n 2 dz
X Y Z 11 X + m1Y + n 1Z 12 X + m2 Y + n 2 Z
where the 1,m,n are arbitrary functions of the variables such that
lX + mY + nZ-/:- O.
By a proper choice of the multipliers, it may be possible to obtain an inte-
grable equation, say
dy = 1 dx + m dy + n dz a dx + b dy + c dz p dx+ q dy + r dz
or =
Y lX + mY + nZ AX + bY + cZ pX + qY + rl
If so, we proceed as in£). See Problem 22,
In actual practice, it may be simpler at times to find by means of multi-
pliers a second integrable equation, rather than to proceed as in£). See
Problems 23-24.
If lX+mY+nZ=0, then also ldx+mdy+ndz=0,
If now 1 dx + m dy + n dz = 0 is integrable, we integrate and have one of
the required relations. See Problems 25-29.
TOTAL DIFFERENTIAL EQUATIONS 167
SOLVED PROBLEMS
1. Obtain the condition of integrability of Pdx + Qdy + Rdz = 0.
Suppose that the given equation is obtained by differentiating

_, 1) /(x,y,z) = C
and, perhaps, removing a common factor µ(x,y,z), Since from 1) of dx + of dy + of dz 0,

ox oy oz
it follows that of
- = µQ, and of µR.
oy oz
Now assuming the existence and continuity conditions,
2 2
0 1 oP Poµ oQ
µ - + Q oµ 0 1
A) = µ- t-
oy ox oy c)y ox ox ox oy
B) it µ oQ + Q~ oR
µ - + R~ it
oz oy oz dz oy oy oy oz
C) it oR
µ - + Roµ =
oP
µ - + Poµ it .
ox oz ox c)x oz oz oz ox
Upon multiplying these relations by R,P,Q, respectively and adding,
µ<R oP + P oQ + Q oR) µ(R oQ + p oR + Q oP)

oy oz ox ox oy oz
and the condition P( oQ oR) + Q(oR - oP) + R(oP oQ) O follows.

oz oy ox oz oy ox
.,,1 2. If µ(x,y,z) 1 in Problem 1, the differential equation is exact. Show that this implies
oP oQ oR oP
oy oz ox oz
These relations follow from A),B),C) in Prob.1. For example, ifµ= 1, A) yields oP oQ.
oy ox
J 3. Solve (x -y)dx - x dy + zdz = O.
oP oQ oQ = oR oR oP
since - = -1 = - • 0 = - ' 0 = - ' the equation is exact.
oy ox oz oy ox oz
Upon regrouping thus xdx - (x dy + y dx) + z dz= O and integrating, we have
2 2
= K or x - 2xy + z = C.
1
4. Solve y2 dx - zdy + ydz = 0.
2 oP oP oQ = 0 oR oR
Here P =y , oy = 2y, oz 0; Q = -z, R = Y, 0, 1;
ox • ox oy
then P(oQ _ oR) + Q(oR _ oP) + R(oP _ oQ) = /(-1-1) -z(0-0) +y(2y-0) =0 andtheequa-
oz oy ox oz oy ox
tion is integrable. The integrating factor lf/ reduces the equation to Jx+ Y dz - z dy = o whose
y2
solution is x + z/y = C.
168 TOT AL DIFFERENTIAL EQCA TIO NS
.I 5. Solve (2x 3y + l)dx + x~ dy + x

2
tanz dz = 0,
The condition of integrability is satisfied since

2
(2x 3 y + 1) (0 -0) + x~ (2x tanz -0) + x tanz (2x 3 - 4x 3 ) 0.
2
The integrating factor 1/x reduces the equation to
1 2 2
(2xy + -)dx + x dy + tan z dz = 0 or (2xydx + x dy) + _2,_dx + tanz dz~ 0
x2 x2
2 1
whose solution is X y - X + In sec z C.
1 3 2
6. solve ( 2x - z ) z dx + 2x y z dy + x ( z + x) dz = 0.
The normal procedure here would be to show that the equation is integrable and then to seek
an integrating factor. By examining the preceding problems it will be found that, upon using
the integrating factor, one variable appears only in an exact differential, for example, the
variable z in tan z dz in Problem 5,
When the equation of this problem is divided by x 2 z, the variable y appears only in the term
2y dy which is an exact diffetential. Thus, we shall use 1/x 2 z as a possible integrating fac-
1 x dz - z dx 2 2 z
tor. The result is 2x dx + 2y dy + -dz + - - - - = O whose solution is x + y + In z + - = C.
Z X2 X
Of course, the separation of the variable here does not indicate that the equation is inte-
grable; for example, x dx + z dy + dz = 0 is not integrable although x appears only in an exact
differential.
"' /• show that if P dx + Qdy + R dz = 0 is homogeneous (i.e., if P, Q, R are homogeneous and of the
same degree) then the substitution x =uz, y = vz will separate the variable z from the vari-
ables u and v,
Let the coefficients P, Q, R be of degree n in the variables.
Substituting x =uz, y =vz, the given equation becomes
P(uz,vz,z) [u dz+ zdu] + Q(uz,vz,z) [v dz+ z dv] + R(uz,vz,z)dz 0,
Dividing out the common factor zn and rearranging, we have
z[P(u,v,l)du + Q(u,v,l)dv] + [uP(u,v,1) + vQ(u,v,l) + Ii(u,v,l)]dz = 0
or z(P1 du + Qi_dv) + (uP 1 + vQ 1 + Ri)dz = 0, where P1 = P(u, v, 1), etc.

p
This may be written as A) 1
- - - ~- - - du + --~Q~1 - - - dv + ~ dz = 0 in which
uP 1 + vQ1 + R1 ul\ + v(h + ll 1 z
the variable z occurs only iri the last term.
Now the condition of integrability for A), ~( ~ 01 cl p1 ) = 0'

z Ou uP 1 + vQ 1 + R 1 - Ov ul-\ + vQ 1 + R 1
is satisfied provided the original equation is integrable and, when this occurs, the sum of the
first two terms of A) is an exact differential. Moreover, since the third term is an exact dif-
ferential, A) is an exact differential equation provided only that P dx + Qdy + R dz = 0 is
integrable,
8, solve the homogeneous equation 2(y + z)dx - (x + z)dy + (2y -x + z)dz = o.

The equation is integrable since 2{y+z){-1-2) - (x+z)(-1-2) + (2y-x+z)(2+1) o.
The transformation x =uz, y =vz reduces the given equation to
2z(v+l)(udz+zdu) - z(u+l)(vdz+zdv) + z(2v-u+l)dz 0.
Dividing by z and rearranging, we have 2z(v+l)du - z(u+l)dv + (uv+u+v+l)dz O or,

2 du dv dz
dividing by z(uv+u+v+l) = z(u+l)(v+l), - -- + = o.
u+l v+l z
2
Then 2 ln(u+l) - ln(v+l) + lnz = lnK, z (u + 1) = K ( v + 1),
2 2
(x+z) = K(y+z) or y + z = C (x + z) •
J 9. Solve the homogeneous equation

2
yzdx - z dy - xydz = O.
The equation is integrable since yz(-2z+x) - z2 (-y-y) - xy(z-0) o.

The transformation x = uz, y = vz reduces it to
2 2 2
vz (u dz + z du) - z (v dz + z dv) - uv z dz = 0.
dv dz
Dividing by z 2 and rearranging, v z du - z dv - v dz = 0 or du---- o.
v z
Then u - In v - In z = In K, vz = Ceu or y = Ce x/z •
10. Solve (2y -z)dx + 2(x -z)dy - (x + 2y)dz = O.
The equation is homogeneous and, by inspection, is seen to be exact since it may be writ-
ten as
2(ydx+xdy) - (zdx+xdz) - 2(zdy+ydz) = O.
The solution is 2xy - xz - 2yz = C.
11. Show that xP + yQ + zR = C is the so:tution of Pdx + Qdy + Rdz O when the equation
is exact and homogeneous of degree n I -1.
First, we check the theorem using the equation of Problem 10. Here
xP + yQ + zR = x(2y-z) + 2y(x-z) - z(x+2y) = 2(2xy-xz-2yz)
and we obtain the solution above.
From xP + yQ + zR = C, we obtain by differentiation
oP oQ oR aP oQ cJR
A) (P + X + y - + z -)dx + (Q + X - + y - + z oR)dy + cR + x oP + y oQ + z -)dz= O.
ox ox ox oy oy oy oz oz oz
oP, oR oP oR
since the given equation is exact, -
oy ox oz oy
Making these replacements, A) becomes
B) cP + x oP + Y oP + z oP)dx + co + x oQ + Y oQ + z oQ)dy + (R + X oR + y oR + z oR)dz o.

ox oy oz ox oy oz ox oy oz
since the given equation is homogeneous, X

oP
- + y -
oP + z
oP
nP, etc., according
ox oy oz
to Euler's Formula on homogeneous functions.
Making these replacements, B) becomes

(n+l)Pdx + (n+l)Qdy + (n+l)Rdz 0
or, since n I -1, P dx + Q dy + R dz = o.
170 TOT AL DIFFERENTIAL EQUATIONS
2 2 2 2 2 2
12. Solve (y + z + 2xy + 2xz)dx + (x + z + 2xy + 2yz)dy + (x + y + 2xz + 2yz)dz 0,
The equation is homogeneous of degree 2 and is also exact since

oP oR oR oP
- = 2(y+x) 2( z + y) = 2(x+z)
oy oy ox oz
2 2 2 2 2 2
The solution is x(y +z +2xy+2xz) + y(x +z +2xy+2yz) + z(x +y +2xz+2yz) =K
2 2 2 2 2 2
or x(y +z) +y(x +z) + z(x +y) =C.
13. Solve the differential equation Pdx + Qdy + Rdz = O given only that the condition of in-
tegrability is satisfied.
Consider one of the variables, say z, as a constant for the moment and let the solution of
the resulting equation
1) P dx + Qdy = O
be
2) u(x,y,z) = rp(z).
Differentiating 2) with respect to all the variables,
3) ou dx + ou dy + ou dz = ¢, ( z) dz d¢.
ox oy oz
OU OU
Now ox = µP and oy = µQ, where µ = µ(x,y, z) is an integrating factor of 1). substi-
ou
tuting in 3), we have µPdx + µQdy + - Jz = dcp.
oz
But from the given equation µPdx + µQdy + µRdz = O so that

OU
d¢ = - dz - µRdz = µR)dz.
oz
This relation is free of dx and dy and, using 2) if necessary, can be written as a differ-
ential equation in z ~nd ¢. Solving the integral for¢ and substituting in 2), we have the
required solution.
14. Solve 2(y+z)dx - (x+z)dy + (2y-x+z)dz = O. (see Problem 8.)

dy 2 2z
We treat z as a constant and solve 2(y+z)dx - (x+z)dy = 0 or - - --y --,
dx X + Z x+z
using the integrating factor 1
e- 2 fdx/(x +z) = - - - , to obtain
(x + z/
y
A)
(x + z/
Differentiating A) with respect to all variables,
dy 2 dz 2z
Y (dx + dz) + - - - ( d x + dz) + d¢
(x + z/ (x + z ) 3 (x+z) 2 (x+z) 3
or 3
2(y + z)dx - (x + z)dy + (2y -x + z)dz + (x + z) dcp = O.
Comparing this with the given equation, it is seen that (x + z) 3 dcp = O and ¢ = C.
Since, from A), y+z =¢(x+z)2, the solution is y+z = C(x+z)2.
The equation is integrable since

(exy+e 2 )(eY-eY+ex+eYz) + (eYz+ex)(-exy-e 2 ) + (ey-exy-eYz)(ex-ex) =O.
Considering z as a constant and solving the resulting equation

2
(exy dx + exdy) + eyz dy + e dx = 0,
we have
Differentiation with respect to all variables yields
2
(exy+e 2 )dx + (eyz+ex)dy + (ey+e x)dz = dcp.
2 2 2
From the given equation, (exy + e )dx + (eyz + ex)dy + (ey + e x)dz = (exy + eyz + e x)dz.
2 2
Thus, dcp = (exy + eyz + e x)dz = ¢dz and ¢ = Ce • The required solution is
2 2
exy + eyz + e x = Ce •
3
16. Solve yzdx + (xz-yz )dy - '2xydz O.
2
The equation is integrable since yz(x-3yz +'2x) + (xz-yz 3 ){-2y-y) - '2xy(z-z) o.
Considering y as a constant and solving the resulting equation
yz dx - '2xy dz = 0 or z dx - '2x dz = o,
we obtain lnx - 2 lnz = ln¢(y) or X = cp/.
Differentiating this result and making the replacement¢ = x/z 2 , we have

dx - '2rf>zdz - /dcp = 0, dx - 2:. dz - z2d¢ = 0,
z
or yz dx - '2xydz - yz 3 d¢ = 0.
Comparing this with the given differential equation, we have

or ¢ dy + y dcp - y dy O•
Then ¢y - ½/ = K or ¢ = ½Y + K/ y, so that the solution is

2 2 2 2
x = ¢z = } (½Y + K/y) or '2xy = y z + Cz •
17. Discuss geometrically the solution of the integrable total differential equation
P dx + Qdy + R dz = o.
Let (x 0 ,y0 ,z0 ) be a general point in space for which not all of P0 = P(x 0 ,y0 ,z 0 ), Q0 =
Q(x 0 ,y0 ,z0 ), R0 = R(xo,Yo,z 0 ) are zero.
Assuming that P,Q,R are single-valued, the set (P 0 ,Q0 ,R0 ) may be considered as direction
numbers of a unique line through the point. Hence, the given differential equation ·may be
thought of as defining at each point (x 0 ,y0 ,z0 )
a line
X -Xo Y-Yo Z -Zo
Qo
and a plane normal to the line.
The solution f(x,y,z) = C of the given differential equation represents a family of sur-
faces such that through a general point (x 0 ,y0 ,z 0 ) of space there passes a single surface S0
of the family. The equation of the tangent plane n0 to this surface at the point is
and the equations of the normal line L 0 are Y-Yo Z - Zo
of of of
oxo oyo ozo
From Problem 1, of= AR. Hence, the solution of an integrable total

oz
differential equation in three variables is a family of surfaces whose tangent plane and
normal at each point are respectively the plane and line associated with the point by the
differential equation.
PAIRS OF TOTAL DIFFERENTIAL EQUATIONS IN THREE VARIABLES.
18. Solve the system: (y +z)dx + (z +x)dy + (x +y)dz = 0

(x+z)dx + ydy + xdz = O.
Both equations are integrable. The first may be written as

(ydx + xdy) + (zdy + ydz) + (xdz + zdx) 0
and the solution is xy + yz + zx = C1 •

The second may be written as x dx + ydy + (z dx + xdz) O and the solution is
2 2
x + y + 2xz = C2 •
2 2
Thus, xy + yz + zx = C1 , x + y ""2xz = C2 constitute the general solution.
Through each point in space there passes a single surface of each of the two families.
since the two surfaces on a point have a curve in common, the solution of the pair of differ-
ential equations is a family of curves. This family of curves may be given by the equations
of any two families of surfaces passing through the family of curves. For example,
2 2
xy + yz + zx = C1 , x + y + 2(C 1 -xy -yz) = C2
also constitute the general solution.
19. Solve the system: 1) yz dx + xz dy + xy dz = 0

2
2) z (dx+dy) + (xz+yz-xy)dz = O.
The first equation is integrable, with solution 3) xyz = C1 , but the second is not.
2 2
Multiply 1) by z, multiply 2) by y, and subtract to obtain z (y-x)dy + y (z -x)dz O.
Multiply this by yz, and substitute xyz=C 1 from 3). The result is
2
z (/z-C1 )dy + /(yz 2 -C 1 )dz = O or z dy + ydz - C 1 (dy + dz) O
y2 z2
whose solution is 4) yz + C1 (y+z) = C2 •
yz
Equations 3) and 4) constitute a general solution. However, 4) may be replaced by the
simpler form 4 1) xy+yz+xz = C2 , obtained from 4) by substituting for C1 •
20. Solve the system: dx + 2dy - (x + 2y)dz O

2dx + dy + (x - y)dz O.
-(x + 2y) -(x + 2y)

y = A..
X -y 1 X -y
For the choice A.. = -1/3, X = -x, Y = x + y, Z = 1, and we write the system in the symmetric form
dx _El__ dz
-X X +y 1
TOTAL DWFERENTIAL EQUATIONS 17.'{
dx dz
From the integrable equation -, we obtain z + In x C1.
-x 1
dx 2
From the integrable equation -5!1__' we obtain X + 2xy = C2,
-X X +y
2
Thus, z + In x = C1 , x + 2xy = C 2 constitute the general solution.
dx dy dz
.,, 21. Solve the system Find the equations of the integral curves through the
X X + Z -z
points a) (1, 1, 1) and b) (2, 1, 1).
dx dz dx
Consider the equations and -5!]__ • The first is integrable and yields xz =
X -z X X + Z
2
Ci- The second is not integrable but is reduced to dy = (1 + C1 /x )dx by the substitution
z = C1/x. Integrating, we have y = x - Cifx + C 2 or, substituting C 1 = xz, y - x + z = C 2•
Thus, xz = C1 , y-x + z = C 2 constitute the general solution.
The integral curve through the point (1, 1, 1) is the intersection of the hyperbolic cylinder
xz = 1 and the plane y -x + z = 1. The integral curve through (2, 1, 1) is the intersection of
the cylinder xz = 2 and the. plane y -x + z = O.
dx dy dz
.,, 22. solve
y-z z -x y-x
No equation is integrable. By means of the multipliers l = m = 1, n = 0, we obtain
dz l dx + m dy + dz n dx + dy
or dx + dy -dz o. Then
y-x l(y -z) + m(z -x) + n(y-x) y-x
A) x+y-z = C1.
dx dx dy
Using A) to eliminate z in -5!]__ , we obtain Then
y-z z-x C 1 -x y-C1
ln(x-C 1 ) + ln(y-C 1 ) = In C 2 , or (x-C 1 ){y-C1 ) = C2 , or, eliminating C 1 by means of A),
B) (z -y){z -x) = C2 •
A) and B) constitute the general solution.
dz
23. Solve
z
2
y dy
From the integrable equation or 0, we obtain
x3
6 6
A) X - Y = C1.
2
We may then use A) to eliminate x in the non-integrable equation dz y dy • However, it
z x3
is simpler to use the multipliers l = m = 1, n = 0 to obtain dz Then x 3 + y 3 = C2 z~

z
dx dy dz
24. Solve the system
2 2 2xy
X +y (x+y) 3 z
dz dx + dy dz
Using l = m = 1, n =0, we obtain or (x + y) (dx + dy). Then
(x + y) 3 z (x + y)
2 z
2
(x + y) - 2 In z = C1 ,
dx + dy dx - dy
and we obtain Then
2 2
(x + y) (x -y)
1 _1_ + K 2 2
or y = C2 (x -y ) •
x+y x-y
dz
2x- 3y
. dx dy
Th e equa t 10n - = or x dx + ydy = 0 is integrable and we obtain
y -x
Using l = 3, m = 2, n = 1, we find 3{y) + 2(-x) + (2x -3y) = 0. Hence, 3dx + 2dy+ dz= 0 and
3x + 2y + z = C,;»
-5!1._ dz
,' 26. Solve the system dx ---•
4y -3z 4x -2z 2y-3x
we seek multipliers l,m,n such that A) 1(4y-3z) + m(4x -2z) + n(2y-3x) = 0.

Rearranging A) in the form (4m-3n)x+(41+2n)y+(-3l-2m)z = 0, we see that it will be
satisfiedwhen 4m-3n=0,41+2n=0,-3l-2m 0 or l:m:n=2:-3:-4. Then
2 dx - 3 dy - 4 dz = O and 2x - 3y - 4z = C1 •
Using the arrangement 4 ( ly + m.x) + 3(- lz - nx) + 2(ny - mz) = 0 and setting ly + m.x 0,
-lz -nx = 0, ny-mz = 0, we obtain l: m: n = x: -y: -z, Then
2 2
x dx - y dy - z dz =0 and x - / - z = C2 •
p dx q dy r dz
(q-r)yz (r -p)xz (p -q)xy
Consider l(q -r)yz + m(r -p)xz + n(p -q)xy = O.

From q( lyz -nxy) + r(mxz - lyz) + p(nxy -mxz) = 0 we obtain l : m: n = x: y: z, Then
2 2 2
px dx + qy dy + rz dz = 0 and px + qy + r z = C1 •
From z( lqy -mpx) + y(npx - lrz) + x(mrz -nqy) = 0 we obtain l : m: n = px '. qy: rz. Then
2 2 2 2 2 2
and p x +q y + r z = C2 •
dx dy dz
2 2 2 2 (x -y)z
X +y -yz -x -y + xz
2 2 2 2
Using l = m = 1, n = -1, we obtain (x +y -yz) + (-x -y + xz) (x - y) z o. Then
dx + dy - dz= O and X + y - Z = C1.

2 2
Using l=xz, m=yz, n=-(x +y ), we find
2 2 2 2 2 2
xz(x +y -yz) + yz(-x -y +xz) - (x +y )(x-y)z = 0.
2 X dx + ydy dz
Then xz dx + yzdy - (x +/)dz 0 or 0
2 2 z
X + y
2
and ln(x +/) - 2 In z = In C2
2 2 2
or X + y = C2 z
dx dy dz
2x -y 4x/ - 2z
2
From dx = dy we have xy 2 = C1 • By inspection, 2/(2x)+2yz(-y)-y (4x/-2z) O;
2x -y
2
then 2/dx + 2yz dy - /dz= 0 or 2dx - Y dz - 2yzdy = 0, and
~
y
dx dy dz
30. Discuss geometrically the general solution of
p Q R
For convenience, let us assume that in solving the given system we have obtained a pair of
integrable equations
and
whose integrals are respectively
g(x,y,z) = C1 and h(x,y,z) = C2 •
Through a general point (x 0 ,Yo, z 0 ) of space there pass two surfaces ( one of each of the
above families) whose curve of intersection C0 is the integral curve of the given system through
the point. The tangent planes to the two surfaces at (x 0 ,y0 ,z0 ) are normal to the directions
(P1 ,Q 1 ,R 1 ) and (P2 ,Q2 ,R 2 ) evaluated at the point, and the line of intersection Lo of these
planes is normal to the two directions. Let (X, Y, Z) be a set of direction numbers for L0 ;
then
X I\.
are proportional to P,Q,R (all evaluated at the point).

Now L 0 is the tangent to C0 at (x 0 ,y0 ,z 0 ), since the tangent to a space curve at one of
its points 1 ies in the tangent plane at the point of any surface containing the curve, Hence,
the integral curves of the system dx = dy = dz consist of a doubly infinite system of curves
p Q R
characterized by the fact that at any point (x 0 , y 0 , z0 ) the tangent to the curve through the
point has (P 0 ,Q0 ,R0 ) as direction numbers,
31. Show that the family of integral surfaces of 1) Pdx + Qdy + Rdz 0 and the family of
dx dy dz
integral curves of 2) are orthogonal.
p Q R
This follows from the fact that at any general point (x 0 ,y0 ,z 0 ) the direction (f 0 ,Q0 ,Ro)
is: a) normal to the integral surface of 1) through the point (see Problem 17) and
b) the direction of the integral curve of 2) through the point (see Problem 30).
32. solve 1) ydx+ xdy- (x+y+2z)dz 0 consistent with a) z a, b) x + y + 2z 0,

c) X + y = 0, d) xy = a.
Equation 1) is not integrable. From each given surface we may obtain an integrable total
differential equation. Our problem then is to solve this differential equation simultaneously
with 1) using the particular solution of the former rather than the general solution as in
/) of the introduction of this chapter.
a) Here z = a, dz= 0. Substituting in 1), we obtain ydx+xdy = 0; then xy C,
Equations z =a, xy=C are said to constitute a solution of 1).
b) Substituting x + y + 2z = 0 in 1), we obtain y dx + xdy = 0 and xy C.

The solution is xy = C, x + y + 2z = 0.
2 2
c) Here y = -x, dy = -dx. substituting in 1), we obtain x dx + z dz = O and x +z C.
2
The solution is x2 + z = C, x + y = O.
d) Here xy = a, x dy + ydx = O. Equation 1) reduces to (x + y + 2z)dz 0.
Then, either x + 2y + 2z = 0 or dz.= 0 and z = C.
xy = a, x +y + 2z = O and z = C, xy = a constitute the solution.
33. Discuss geometrically the problem of solving Pdx+Qdy+Rdz = 0 consistent with the given
relation g(x,y,z) = O.
F'rom the relation g(x,y,z) = 0, we obtain dg dx + dg dy + dg dz o.

-ax dY dZ
We solve the system P dx + Q dy + R dz = O, dg dx + dg dy + dg dz 0 using the particular
dX cly dZ
solution g(x,y, z) = 0 of the latter. Let
f(x,y,z) = C, g(x,y,z) = 0
constitute the solution. The integral curves are those cut out on the surface g(x,y,z) = 0
by the system of surfaces f(x,y,z) = C. Thus, Problem 32c may be stated as: Find all curves
lying on the surface (plane) x + y = O which satisfy the differential equation
y dx + xdy - (x +·y+ 2z)dz = 0.
At a general point (x 0 ,y0 ,z0 ) on the surface g(x,y,z) = 0, the line of intersection L 0 of
the tangent planes to g(x,y,z) = O and the surface of the system f(x,y,z)=C, through the point,
is tangent to the curve of intersection of the two surfaces. Thus, we have found the family
of curves on the given surface g(x,y,z) = 0 whose tangent at any point lies in the plane, through
this point, determined by the differential equation. (See Problem 17.)
For example, consider Problem 32c. On the prescribed surface x +y = 0, choose any point
(a,---a,b). At this point, the tangent plane to x +y = O (here, the plane itself) is normal to
the direction (1, 1,0) and the tangent plane to the surface (of the family) x 2 + z 2 = a2 + b2
is normal to the direction (a,0,b). A set of direction numbers for the line of intersection
L of these planes [the tangent to the curve through (a,-a,b)] is (-b,b,a).
Now the plane through (a,-a,b) determined by the given differential equation is normal to
the direction [y,x,-(x+y+2z)](a,-a,b) = (-a,a,-2b). since (-b,b,a) and (-a,a,-2b) are normal
directions, the line L lies in the plane determined by the differential equation.
34. Solve 1) 2z dx + dy + y dz = O consistent with 2) x + y + z = O.

From 2), y,; -x-z and dy = -dx-dz, Substituting for y and dy in 1), we obtain
3) (2z-l)dx - (x+z+l)dz = O.
The transformation z = z 1 + 1/2, x = x1 - 3/2 reduces 3) to
4) 2z 1 dx 1 - (x 1 + zi)dz 1 = 0, a homogeneous equation.
2 du
The transformation reduces 4) to (u - 1 )dz 1 + 2z 1 du= 0 or + -- o.
u-1
Then In z 1 + 2 ln(u-1) =lnK

2
or zi(u-1) =K,
Replacing u by xif z 1 , x 1 by x + 3/2 and z 1 by z -1/2, this becomes
2
(x-z+2) = C(2z-1).
Test for integrability and solve when possible,
35. (y + 3z)dx + (x + 2z)dy + (3x + 2y)dz = 0 Ans, xy + 2yz + 3xz = C
X
e y + eyz + sin x = C
37. dx + (x +z)dy + dz 0 y + ln(x+z) = C
2 2
38. z 3dx + z dy - 2y dz 0 xz + y = Cz
2 2
39. x dx - z dy - xy dz = 0 Not integrable.
2 2
40, (x + z) dy + y (dx + dz) = 0 y(x+z) = C(x+y+z)
2 2 2 2 2 2 2 2 2
41. 2x(y+z)dx+(2yz-x +y -z )dy+(2yz-x -y +z )dz=O x +y +z =C(y+z)
2
42. yz dx - 2xz dy + xy dz = 0 y Cxz
2
2 2 2 2 2 2 z
43. x dx + y dy + (x + y +z + 1) z dz = 0 (X +y +z )e = C
2 2 2 2
44, z(x -yz-z )dx + xz(x+z)dy + x(z -x -xy)dz 0 (x+y)/z + (y+z)jx = C
Solve the following pairs of equations.

iy-
45. dx + dy + (x + y)dz = O
z (dx + dy) + (x + y)dz O
46, 2yz dx + x(z dy + ydz) O 2 2

2 x yz = C1 , x z + x + z = C2
y dx - x z dy + y dz 0
X dx dy dz 2 2
47. X z
3 2 3
y z X Z y
3 dx dy dz 2
- ; 48. y
yz xz xy
dx dy dz 2 2 3 3
• _ 49, z = xy + x y + x y + C2
1 1 2 2
(X + y) ( 1 + 2xy + 3x y )
dx dy dz
50.
2 2 2 2xy 2xz
X -y - Z
dx dy dz
x + 2y + 3z = C1 , X
2
+ y
2
+ Z
2
= C2
3y- 2z z -3x 2x-y
dx dy dz 2 It It It
52. xyz X + y + Z
It It It It It It
x(2y -z ) y(z - 2x ) z (x - y )
dx dy dz 2 2 2
X +y +Z
2 2 2 2 2 2
x(z - y ) y(x - z ) z(y - X )
CHAPTER 23
Applications of Total and Simultaneous Equations
WHEN A MASS m moves in a plane subject to a force F, its acceleration continues to

satisfy Newton's Second Law of Motion: mass x acceleration= force.
To obtain the equations of motion, when rectangular coordinates are used,
consider the components of the vectors force and acceleration along the axes.
The components of acceleration ax and ay are given by
and, denoting the components of the force by Fx and Fy, the equations of mo-
tion are
X 0
COMPONENTS OFF IN RECTANGULAR AND POLAR COORDINATES.
In polar coordinates, the corresponding equations are
m{2 dp d0 + p d20} = Fe,

dt dt dt2
where FP and Fe are the radial and transverse components of force, i.e., the
components along the radius vector at Panda line perpendicular to it.
SOLVED PROBLEMS
2 2 2
1. Find the family of curves orthogonal to the surfaces x + 2y + 4z = C.
2 2
Since x + 2y2 + 4z = C is the primitive of the total differential equation
x dx + 2y dy + 4z dz "' 0,
the differential equation of the family of orthogonal curves is
dx dy dz
(See Chapter 22, Problem 31.)
X 2y 4z
178
APPLICATIONS OF TOTAL AND SIMULTANEOUS EQUATIONS 179
. dx dz
So 1vrng - dy we have y = Ax 2 • Solving dy - , we have z = By 2 •
X
2y. 2y 4z
The required family of curves has equations y = Ax 2 , z = By 2 •
2. Show that there is no family of surfaces orthogonal to the system of curves

2
x -/=ay, x+y=bz.
Differentiating the given equations and eliminating the constants, we have

2 2
X-Y dx + dy = x + y dz •
2x dx - 2ydy --dy,
y z
2 2
dx !!I_. Solving it for dx, X + y- d y,
The first can be written as dx - - and substi-
x2 + y2 2xy 2xy
2
(x + y2 + l)dy = x + y dz dy dz
tuting in the second, we have or
2xy z 2xy (x + y) z
Thus, the differential equations in symmetric form of the given family of curves are
dx dy dz
2xy (x + y)z
2
Since the equation (x + /)dx + 2xy dy + (x + y)z dz = O does not satisfy the condition of
integrability, there is no family of surfaces cutting the curves orthogonally,
3. The x-component of the acceleration of a particle of unit mass, moving in a plane, is equal to
its ordinate and the y-component is equal to twice its abscissa, Find the equation of its
path, given the initial conditions x =y =O, dx/dt = 2, dy/dt =4 when t =o.
The equations of motion are y, 2x.
Differentiating the first twice and substituting from the second, 2x and
X C1e at + C2 e - at + C3 cos at + C4 sin at, where a~ -- 2.
Then,
dx
dt
dy
and
dt
Using the initial conditions:
Then
The parametric equations of the path are:
x t1(2+v2)/2(e
~ 12 t -e-12 t )
180 APPLICATIONS OF TOTAL AND SIMULTANEOUS EQUATIONS
4. A particle of mass mis repelled from the origin Oby a force varying inversely as the cube of
the distance p from 0. If it starts at p = a, e = O with velocity v 0 , perpendicular to the ini-
tial line, find the equation of the path.
The radial and transverse components of the repelling force are: FP o.
d2 m.k2 2
dp d8 d 8
Hence, m ( ____E_ - P(!~l) -, m(2- - + P-) = 0
dt 2 p3 dt dt dt 2
d2p k2 2
d8 2 d 8 dp d8
or 1) - P(-) -, 2) p - + 2- - o.
dt 2 dt p3 dt 2 dt dt
d8 d8 av0
Integrating 2), p 2 d8 = C1 When t = 0, p = a and P-= Vo; then C1 = av0 and =-·
dt • dt dt 2
p
2 2 k2
d8 d2p a v0
Substituting for - in 1) • + -• Multiplying by 2dp •
dt dt 2 p3 p3 dt
2 2 2 2 2 2 2
dp d p a v + k dp a v0 + k
2 2 ~0 - - - and
dt dt2 dt p
2
When t = O, p = a and dp = o· then and

dt '
2 2
2 2 2 1 1 22 2p-a
( a v0 + k ) ( - - - ) (a v 0 + k )--- •
a2 p2 a2p'1-
a v
2 2 2 2
(a v0 + k )p (p - a )
2 2 2 2
dp 1c2
~ v 2 + k2
Dividing by (d8)2 - -0 , (dP/ and 0
d8.
dt d8 ~ 2 2
p
~
a v0 p /427p a v0
/, 2 2 2
1 a v0 + k
Integrating,
a
arc sec f!..
a 2
e + C3,
a v0
I 2 2 2
When t= 0, P=a and 8= O; then C3 = 0 and p
v a v +k
a s e c - 0- - - e.
av 0
5. A projectile of mass m is fired into the air with initial velocity v0 at an angle e with the
ground. Neglecting all forces except gravity and the resistance of the air, assumed propor-
tional to the velocity, find the position of the projectile at time t,
In its horizontal motion, the projectile is affected

only by the x-component of the resistance. Hence,
y
dx dx
1) -K dx -m.k - or -k - ,
dt dt dt
In its vertical motion, the projectile is affected

by gravity and by the y-component of the resistance.
Hence,
2 X
2) m d y = - mg - mk dy or - g - k dy.
dt2 dt dt
APPLICATIONS OF TOTAL Al\D SIMULTANEOUS EQljATIONS 181
dx 1 -kt
Integrating 1). ~
- kx + C 1 and X = - C1 + C2e
dt k
dy 1 . -kt 1 1
Integrating 2), - gt - ky + K1 and y - K1 +K2e -g(-t -).
dt k k k2
dx
Using the initial conditions X =y =0, dt = Vo cos e. sin e when t = 0:
,
C 1 = v 0 cos · , o ·
C2 - k1 v 0 cos e; K1 = v 0 sin e, 1<.·~
£ - k1 Vo sine
1 -kt 1{ (-g . O) (1 - e -ld ) - gt } •

Thus, x = -(v 0 cos 8) (1 - e ), y - + v0 srn
k k k
6. Two masses, m1 and m2 , are separated by a spring for which k = k 2 lb/ft

and m1 is attached to a support by a spring for which k = k 1 lb/ft as in
the figure. After the system is brought to rest, the masses are displaced
a feet downward and released. Discuss their motion.
Let positive direction be downward and let x 1 and x 2 denote the dis-
placement of the masses at time t from their respective positions at rest.
The elongation of the upper spring is then x 1 and that of the lower spring
is x 2 -x 1 • The corresponding restoring forces in the springs are
-k 1x 1 + k 2 (x 2 -x 1 ) acting on m1
and -k 2 (x 2 -x 1 ) acting on m2 •
The equations of motion are
2
d2x1 d x2
m1 = - k1x1 + k2(X2 -x1) and m2 - k2(X2 -X1)
d/ d/
2
or 1) and 2) (m 2D + k 2 )x2 - k2X1 o.
2
Operating on 1) with (m 2 D + k 2 ) and substituting from 2),
2 2 2 2 2 2
(m 2D +k 2 )(miD +k 1 +k 2 )x 1 k 2 (m-iD +k 2 )x 2 = (m 2D +k 2 )(m 1D +k 1 +k 2 )x 1 - k2x 1 0
or o.
Denoting the roots of the characteristic equation by ±ia, ±i/3, where
(k 1 + k 2 + k 2/ _ 4 k1k 2
m1 m2 m1m2
and
Using the initial conditions x 1 = x 2 = a, 0 when t = 0,
a v -1
-( _.._) and
2 v-µ.
182 APPLICATIONS OF TOTAL AND SIMTLTANEOUS EQUATIONS
7. A uniform shaft carries three disks as in the ad-

joining figure. The polar moment of inertia of the
disk at either end is 1, and that of the disk at
the middle is 41. The torsional stiffness constant
of the shaft between two disks (the torque required
to produce an angular displacement difference of
one radian between successive disks) is k. Find
the motion of the disks if a torque 2 T0 sin wt is
applied to the middle disk, assuming that at t = 0
the disks are at rest and there is no twist in the
shaft.
At time t, let the angular displacement of the disk at either end be 8 1 and that of the disk
at the middle be 82 • The differences of the angular twists of the ends of the two pieces of
shaft, from left to right, are 82 - 8 1 and 8 1 ~ 8 2 • The restoring torques acting on the disks
are k((3 2 -8 1), k(8 1 -82 )-k(82 -8 1) and -k(8 1 -8 2 ) respectively, The net torque acting on a
mass when rotating is equal to the product of the polar moment of inertia of the mass about
the axis of rotation and its angular acceleration; hence the equation of motion of the middle
disk is
2
1) 41 d
82 = k(8 1 -8 2 ) - k(8 2 -8 1) + 2T0 sinwt or
2
dt
and that of either end disk is

2
2) l d 81 or
dt 2
2
Operating on 2) with (2JD +k) and substituting from 1),
or
3)
The characteristic roots are O, O, ai, -ai, where a2 3k/21, and
T0 k sinwt
4) C1 + C2 t + C3 cosat + C4 sinat +
2 2
lw (2lw - 3k)
T0 k
C1 + C2 t + C3 cos at + C4 sin at + - - - - - - sin wt.
2 2 2
21 w\w - a )
From 2),
l 2
(- D + 1)81 and
k
2
12 12.
C( l --a)s1nat+ T0 k-'fowl
5) 82 = 1+ C
C 2t+
C3 (1--a)cosat+ 4 sin wt.
k k 2 2 2
21 w (w - a )
2
From 4) and 5), we obtain by differentiation,
and
Using the initial conditions 8 1 = 82 =0, O when t = 0, 0,

APPLICATIONS OF TOTAL AND SIMULTANEOUS EQUATIONS
T0 k
C1 + C 3 (1- !a2) 0, C2 + C4 a +
2
o.
k 21 w(w 2 - a 2)
T0 w
Then C1 C3 0, c4 C2
3la(w 2 -a 2 )
T0 t a
3 sinwt-w 3 sin at),
w sin at ) and
-(- +
a(w 2 - a 2 ) 31 w aw2(w2-a2)
8. The fundamental equations of a transformer are
where i 1 (t) and i 2 (t) denote the currents, while M, L1, L2 , R 1, R2 are constants.
2
Assuming M < L 1 L 2 , show that
A)
B) - ME' (t).
Solve the system when E(t) E0 , a constant.
Differentiating 1) and 2) with respect to t.

2. 2 2. d2 l1
Md l1 d i2 R 2di2 Md l2 di 1
3) + L2 + - = 0, 4) + L1 + R1 E'(t).
dt dt
dl dl d/ dl
Multiplying 3) by Mand 4) by L 2 , and subtracting,
Sus ..
bt 1tut1ng for -di2f rom 2), we obtain A) •
dt
Multiplying 3) by L 1 and 4) by M, and subtracting,
- ME' (t).
Substituting for di 1 from 1), we obtain B).

dt
When E(t) = E0 , equation A) is
Let a, /3 denote the characteristic roots.

181 APPLICATIONS OF TOTAL AND Sll\IULTANEffCS EQUATIONS
Then
To find 12 , multiply 1) by Mand 2) by L 2 , and subtract to obtain

2 di L .
(L1L2 -M ) -1 + L2R111 - 2Eo,
dt
1 2 at Bt at Bt
Then -[(L1L2 -M ) (aC 1 e + f3C 2 e ) + L2R1(C 1e + C2e )].
MR2
2
Note that since M < L1 L2 , both a and f3 are negative. Then after a time, the primary cur-
rent becomes approximately constant= E0 /R 1 and the secondary current i 2 becomes negligible.
9. A moving particle of mass mis attracted to a fixed point O by a central force which varies
inversely as the square of the distance of the particle from 0. Show that the equation of its
path is a conic having the fixed point as focus.
Using polar coordinates with O as pole, the equations of motion are

2 mk2 2 k2
d p de 2 K d P _ P<del
1) m[- - p(-) ] or - -,
2 2 d/ dt 2
d/ dt p p p
2 2
m [ 2 dp de + d e dp de d e
2) p-] = 0 or 2 - - + p- = o.
dt dt dt dt 2
d/ dt
From 2), d ( / de) = 0 and

dt dt
de dp 1 do- de
Let a- = ~, Then and
p dt dt a- 2 de dt
2
d do- 2 2 d a-
- (- C1 - ) · -C1a- - • Substituting in 1) and simplifying, we have
dt de 2
de
a linear equation with constant coefficients. Solving,
or p
the equa-
1± e cos ( e +a)
tion of a conic having Oas focus.
APPLICATIONS OF TOTAL AND SIMULTANEOUS EQUATIONS 18S
2
10. Find the family of curves orthogonal to the family of surfaces x + / + 2/ C.
2
Ans. y = Ax, z = By
11 • Find the family of surfaces orthogonal to the f amily of curves y = C1 x, X2 + y2 + .,~ 2 2

2
Ans. z = C(x + / )
12. A particle of mass m is attracted to the origin Oby a force varying directly as its distance
from 0, If it starts at (a, 0) with velocity v 0 in a direction making an angle 8 with the hor-
izontal, find the position at time t.
cos e Vo sine
Ans. X = a cos kt + Vo sin kt, y sin kt
k k
13. The currents 11 , i,2 , i = i1 + i2

in a certain network satisfy the equations
di 2
20i + 0.1 - = 5, 4i + l1 + l000q1 = 1.
dt
Determine the currents subject to the initial conditions i = 11 = 12 O when t o.
2
Hint: Use 11 = dq 1 to obtain d q1 + 240 dq 1 + 40, 000q 1 = 0.
dt dt2 dt
1 -12ot . -12ot 1 -12ot .

Ans. -e S1Il160t, e cos 160t) + e srn 160t
4 8
14. Initially tank I contains 100 gal of brine with 200 lb of salt, and tank II contains 50 gal
of fresh water. Brine from tank I runs into tank II at 3 gal/min, and from tank II into tank
I at 2 gal/min. If each tank is kept well stirred, how much salt wil 1 tank I contain after
50 minutes?
Hint: 200, Ans. 68.75 lb
50 + t 100 - t
CHAPTER 24
Numerical Approximations to Solutions
IN MANY APPLICATIONS it is required to find the value y of y corresponding to x =

x 0 +h from the particular solution of a given differential equation
1) y' = f(x, y)
satisfying the initial conditions y == Yo when x == x 0 • Such problems have been
solved by first finding the primitive
2) y = F(x) + C
of 1), then selecting the particular solution

3) y = g (x)
through (x 0 ,y0 ), and finally computing the required value y = g(x 0 +h).
When no method is available for finding the primitive, it is necessary to

use some procedure for approximating the desired value. Integrating 1) between
the limits x=x0 ,y=y 0 and x=x,y=y, we obtain
4) y Yo+ fxf(x,y)dx.
Xo
The value of y when x == x 0 +h is then
5) Y = Yo + lXo+\(x, y) dx.
Xo
The methods of this chapter will consist of procedures for approximating 4)

or 5).
PICARD'S METHOD. For values of x near x = x 0 , the corresponding value of y = g(x)

is near Yo== g(x 0 ). Thus, a first approximation y 1 of y =g(x) is obtained by
replacing y by Yo in the right member of 4), that is,
Y1 = Yo + i x f(x,Yo)dx.
Xo
A second approximation, y 2 , is then obtained by replacing y by y 1 in the right

member of 4), that is,
Continuing this procedure, a succession of functions of x
is obtained, each giving a better approximation of the required solution than

the preceding one. See Problems 1-2.
Picard's method is of considerable theoretical value. In general, it is
unsatisfactory as a practical means of approximation because of difficulties
which arise in performing the necessary integrations.
186
NUMERICAL APPROXIl\IATTONS TO SOLUTIONS 187
TAYLOR SERIES. The Taylor expansion of y = g(x) near (xo,Yo) is
6)
From 1), y' = g'(x) = f(x,y); hence, by repeated differentiation,
y" = g"(x) = of + of dy
OX oy dx
0
7) y"' = g'" (x) = ..:!._(of+ fol) = ( - + f.i_)(of + f of)
c1x ox oy ox oy ox oy
2 2
o2f
+ 2f ~ + f( of)2 + f2 0 f •
of of
+ - - etc.
OX
2 ox oy oxoy oy oy 2
of of o2f o2f o2f

For convenience, write p = -, q = -, r = -, s = --, t = and
ox oy cJX2 ox oY oy2
let f 0 ,p0 ,q0 , · · · denote the values of f,p,q,•·· at (x 0 ,y0 ) . Substituting in

6) the results of 7) and evaluating for x = x 0 +h, we obtain
1 2 1 3 2 2
8) y = y 0 + h•f0 + t0
2h (p 0 + f 0 °9o) +
6h (r 0 +Po·%+ 2f0 •s 0 + f 0 °q 0 + f 0 • )
+ •.....••.
This series may be used to compute y; it is evident, however, that addi-
tional terms will be increasingly complex. See Problems 3-4.
FIRST DERIVATIVE METHOD. A procedure involving only first derivatives, that is,
using only the first two terms of Taylor series, follows.
y
A
C
k3
k2
k1 k
Yo - -------------- N
h1 ~ M X
0 Xo
As a first approximation of y, take the first two terms of 8)

Y ~ Yo+ hf(xo,Yo).
To interpret this approximation geometrically, let PQ be the integral curve
of 1) through P(x 0 ,y 0 ) and let Q be the point on the curve corresponding to
x = x 0 +h. Then y = MQ = y 0 +k. If e is the angle of inclination of the tangent
af P, then from 1) tan e = f(x 0 ,Yo) and the approximation
Yo+ hf(x0 ,y0 ) =LP+ htane = MN+NA = MA.
188 :\{ 1IERICAL APPR0:\11\IATH):\S TO SOL[TIONS
To obtain a better approximation, let the interval LM of width h be divided

into n subintervals of widths h 1 , h 2 , • • • • hn, (In the figure, n = 3.) Let the
line x=x 0 +h 1 meet PA in R(x 0 +h 1 ,y 0 +ki) = (x 1 ,Y1), Then
Y1 = Yo+ k1 = Yo+ h1f(xo,Yo),
Let RS be the integral curve of 1) through R, and on iLs tangent at R take T
having coordinates (x 1 +h 2 ,y 1 +k 2 ) = (x 2 ,y 2 ) . Then
Y2 = Y1 + k2 = Y1 + h2f(x1,Y1) = Y1 + h2f(xo+h1, Yo+h1fo),
After a sufficient number of repetitions, we reach finally an approximation
MC of MQ. It is clear from the figure that the accuracy will increase as the
number of subintervals is increased in such a manner that the widths of the
subintervals decrease. See Problems 5-6.
RUNGE'S METHOD. F'rom 5) and 8) we obtain

x0 +h
9) k = Y-Yo =
l Xo
f(x,y)dx
2 3 2 2
h to. + -l h ( Po + f o qo ) + -l h ( r o + Po qo + 2 f o So + f o qo + f o t o ) + · • ' · · · · ·
2 6
Assume for the moment that the values y 0 ,y 1 ,y 2 of y = g(x) corresponding to
x0, x 1 = x 0 + ½h, x 2 = x 0 + h are known. Then by Simpson's Rule,
Actually, only Yo is known. Runge's Method is based on certain approxima-

tions of y 1 and y 2 ,
Y1 ~ Yo+ ½hf(xo,Yo) = Yo+ ½hfo,
y2 ~ Yo + hf(x 0 +h, y 0 +hf0 ),
chosen so that when k, found by 10), is expanded as a power series in h t!le

first three terms coincide with those of the right member of 9). Thus 10) be-
comes
These calculations are best made as follows:

k 1 -= hf0 , k2 = hf(x 0 +fi,y 0 +k 1), k 3 = hf(x 0 +h,y0 +k 2 ), k4
Note. Since the approximation of k obtained here differs from the value as
given by 8) in the terms containing powers of h greater than 3, the approxi-
mation may be poor if £0 > l. See Problems 7-11.
KUTTA-SIMPSON METHOD. Various modifications of the Runge Method have been made by
Kutta. One of these, known as Kutta's Simpson's Rule uses the following cal-
culations:
k 1 =hf0 , k 2 = hf(x 0 +½h, Yo +½k 1 ), k3 = hf(x 0 + ½h, y 0 + ½k 2 ), k4 = hf(x 0 +h,y 0 +k 3 ) ,
1
- ( k 1 + 2k 2 + 2k 3 + k 4 ) • See Problem 12.
6
NUMERICAL APPROXIMATIONS TO SOLUTIONS \8<J
SIMULTANEOUS FIRST ORDER DIFFERENTIAL EQUATIONS. Approximations to that solution

of the simultaneous differential equations
dy dz
- = f(x,y,z), g(x,y, z)
dx dx
for which y =Yo and z = z0 when x =x 0 , may be obtained by the use of Picard's
Method, Taylor Series, Runge's Method, or Kutta-Simpson Method. The necessary
modifications of the formulas given above are made in Solved Problems 13-14.
Further extensions to three or more simultaneous first order equations may be
readily made.
DIFFERENTIAL EQUATIONS OF ORDER n. The differential equation

n
d y = f(x,y,y', y", .. ·• .. ,Yn-1)
dxn
where y' = dy, y'' = d\, ••••, may be reduced to the system of simultaneous
dx dx2
first order equations
When initial conditions x=x 0 , y=y0 , Y'=(y 1 ) 0 , y"=(y,;)0 ,••··, yn- 1 =(Yn- 1 )0
are given, the methods of the preceding paragraph apply.
2
d d
EXAMPLE. The second order differential equation _J_ + 2x ....l.. - 4y = 0 is equiva-
dx2 dx
lent to the system of simultaneous first order differential equations
dy dz
::: z, 4y - 2xz. See Problems 15-16.
dx dx
SOLVED PROBLEMS
1. Use Picard's Method to approximate y when x = 0. 2, given that y =1 when x =0, and dy/ dx = x - y.
Here f(x,y) = x-y, x 0 =0, y 0 =l. Then
Y1 Yo + [x
0
f(x,yo)dx 1 + fx
0
(x - l)dx 1 2
-x - X + 1,
2
Y2 Yo + •[X f(X,Y1)dx 1 +
fx (- -x
1 2
+ 2x - l)dx
1 3 2
- - X + X - X + 1,
0 o 2 6
Y3 Yo + fx f(x,y2)dx 1 +
ix (-x
1 3 2
-x + 2x-l)dx
4 1 3 2
- x - -x + X - X + 1,
0 0 6 24 3
x 1 4 1 3 5 3
1 + i
O 24 3
2
120
4
12 3
2
(- - x + -x -x + 2x-l)dx = - _:_ + :...__:..._+x -x+ 1
'
lo 15 1413 2
- X - - X + - X - - X + X - X + 1, •••••••••
720 60 12 3
Whenx=0.2, y0 =1, y 1 =0.82, y 2 =0.83867, y 3 =0.83740, y 4 =0.83746, y 5 =0.83746.

Thus, to five decimal places, y = 0.83746.
190 l\TMERICAL APPROXI!\JATIONS TO SOLCTJONS
Note. The primitive of the given differential equation is y = x - l+ Ce -x. The particular
solution satisfying the initial conditions x = 0,y d is y = x - 1 + 2e -x. Replacing e -x by its

2
MacLaurin series, we have y = 1 - x + x - !3 x3 + ~ 4
x - ~ x5 + _1_ xo + •·•·•·••··. Upon
12 60 360
comparing this with the successive approximations obtained above, it seems reasonable to sup-
pose that the sequence of approximations given by Picard's Method tends to the exact solution
as a limit.
2. Use Picard's Method to approximate the value of y when x = 0.1, given that y = 1 when x = 0, and
2
dy/dx = 3x + y •
Here f(x,y) = 3x+ y2, x 0 = 0, y0 = 1. Then
x
Yo + I 0
2
(3x + y0 )dx 1 + fox (3x + l)dx 3
-
2
X
2
+ X + 1,
[X 9 4 3 2 95344352
1 + J" (-x + 3x + 4x + 5x + l)dx -X + -X + -X + -X + X + 1,
O 4 20 4 3 2
[X 81 10 27 9 141 8 17 7 1157 0 136 5 125 4 23 3 2

Yn 1 + JA ( - x + - x + - x + - x + - - x + - x + - x + -x + 6x + 5x + l)dx
o 400 40 80 4 180 15 12 3
When x = 0.1, y0 = 1, y 1 = 1. 115, y 2 = 1.1264, y 3 = 1.12721.
3. If 1 x -y, use the Taylor Series Method to approximate y when:
a) x=0.2, given that y=l when x=0.

b) x=l.6, given that y=0.4 whenx=l.
a) Here y g(x) • g(xo) 1, y '" g"' (x) = -y", g"' (Xo) -2,
1v 1v 1v
y' g' (x) X -y, g' (Xo) -1, y g (x) ~ -y'", g (Xo) 2,
V V 1v
y" g"(x) = 1-y'. g"(xo) 2, y g (x) = -y gv(xo) -2, etc.,
2 1 3 1 4 1 5
and equation 6) becomes y = 1 - X + X - - X + - X - - x + ........ Then
3 12 60
1 1 1
y 1 - 0. 2 + 0,04 - -(0.008) + -(0.0016) - -(0.00032) + .. • ~ 0.83746. (See Problem 1.)
3 12 60
1v
b) Here g(x0 ) = 0,4, g'(x 0 ) 0.6, g"(xo) 0.4, g"'(xo) -0.4, g (xo) 0.4, etc.,
and equation 6) becomes
h2 h3 h4 h5
y = 0,4 + 0.6h+ 0.4 - - 0,4 + 0.4 - - 0,4 + ........ where h = x -x 0 •
2 6 24 120
When x = 1. 6, h = o. 6 and
y 0.4 + 0.6(0.6) + 0.4(0.18) - 0.4(0.036) + 0.4(0.0054) - 0.4(0.000648) + 0.4(0,0000648)

+ ......... .
~ o. 81953.
NUMERICAL APPROXIMATIONS TO SOUffIONS )<))
4. If : 3x + / , use the Taylor Series Method to approximate y when:

a) x=0.1, given that y=l whenx=0.
b) x = 1.1, given that y = 1. 2 when x = 1.
a) Here (x 0 ,y0 ) = (0, 1), g(x 0 ) = 1,
Y' = g' (x) 3x + y 2 , g' (Xo) 1,
y" g"(x) 3 + 2yy I, g"(xo) 5,

y"' g'"(x) 2(y'/ + 2yy", g'"(xo) = 12,
y 1v g1v(x) 6y'y"+ 2yy"', g1v (xo) 54,
yv 2 1v
= gv(x) 6(y") + 8y'y'" + 2yy gv (xo) 354, and 6) becomes
'
y 1 + X + ~x 2 +
2
............. When x = 0. 1,
y 1 + 0.1 + 0.025 + 0.002 + 0.00022 + 0.00003 + •••••• Ro 1.12725. (See Problem 2.)
b) Here (x 0 ,y0 ) = (1,1.2), g(x 0 ) = 1.2, g'(x 0 ) = 4.44, g"(x 0 ) = 13.656, g"'(x 0 ) = 72.202,
537.078, lcxa> = 4973, ......... . and 6) becomes
h2 h3 h4 h5
y = 1. 2 + 4. 44h + 13. 656 - + 72. 202 - + 537. 078 + 4973 + ••••••• •••••
2 6 24 120
where h = x-x 0 • When x =1.1, h = 0.1 and
y = 1.2 + 0.1(4.44) + 0.01(6.828) + 0.001(12.03) + 0.0001(22.4) + 0.00001(41) + ••• Ro 1.7270.
5. Use the First Derivative Method, with n = 4, to approximate y when x = 1.1, given that y = 1. 2
when x =1 and dy/dx = 3x+y 2 • See Problem 46.
a) (xo,Yo) (1,1.2), h 1 = 0.025, f(xo,Yo) = 4.44, k 1 = hd(xo,Yo) = 0.111;

Y1 = Yo , k1 1. 311.
b) (x1,Y1) (1.025, 1.311), h2 = 0.025, f(x1,Y1l = 4. 7937, k2 h.,f(x1,Y1) o. 1198;
Y2 = Y1 + k2 1. 4308.
c) (X2,Y2) (l.05,1.4308), h3 = 0.025, f(x2,Y2l = 5. 1972, k3 = hd(x2,Y2) = 0.1299;
Y3 = Y2 + k3 1. 5607.
d) (X3,y3) ( 1. 0'15, 1. 5607), h4 0.025, f(x3,y3) = 5.6608, k4 h4f(x3,y3) o. 1415;
y Ro Y3 + k4 = 1. 7022.
6. Use the First Derivative Method, with n 4, to approximate y when x = 1.4, given that y =0. 2
dy 2 ½
when x = 1 and dx = (x + 2y)
Here h = o. 4 and we take h 1 = h 2 = h 3 = h 4 = 0. 1.
b) (x1,Y1) ( 1. 1, o. 3183), h2 = o. 1, f(x1,Y1) 1.359, k2 h2f(x1,Yd o. 1359;

Y2 = Y1 + k2. 0.4542.
c) (x2,Y2) (1.2,0.4542), h3 = o. 1, f(x2,Y2) 1. 532, k3 h3f(x2,Y2) o. 1532;
Y3 = Y2 + k3 o. 6074.
192 ~Tl\JERICAL APPROXIMATIONS TO SOTXTIONS
(1.3,0.6074), h4 0.1, f(x 3 ,y 3 ) cc 1.704, k 4 ~ h 4 f(x 3 ,y 3 ) 0.1704;

y ::;:; y3 + k4 = 0.7778.
7. Use Runge's Method to approximate y when x = 1. 6, given that y = 0. 4 when x = 1 and dy/ dx = x - y.
(See Problem 36.)
Here (x0 ,y0 ) = (1, 0.4), h = 0.6, fo = 1-0.4 = 0.6. Then
k1 hfo = o. 36,
k,;, hf (x 0 +h, Yo +k1) 0.6[(1+0.6) -(0.4+0.36)] o. 504,
k3 hf(x0 +h, Yo+k2) 0.6[(1+0.6) -(0.4+0.504)] = 0.4176,
k4 hf(Xo+~h, Yo+½k 1) = o. 6[( 1 + o. 3) - (0.4 + o. 18)] 0.432,
1 1
k ~ -(k 1 + 4k4+ ks)
6
-[0.36+4(0.432)+ 0.4176]= 0.4176,
6
and y Yo+ k ~ 0.8176.
The difference between this approximation and that found in Problem 36 arises from the fact
that h = o. 6. In finding the value of y when x = 1.1, (that is, h = 0.1), the Taylor series gives
y = 0.4 + 0.6(0.1) + 0.4(0.005) - o.4(0.00017) + 0.4(0.000004) - •••••••••••• "" 0.46193,
while by Runge's Method
k1 0.1(0.6) = 0.06, k 2 = 0.1(1.1-0.46) = 0.064, k 3 = 0.1(1.1-0.464) = 0.0636,
1
k4 = 0.1(1.05-0.43) = 0.062, k~-(k 1 + 4k 4 + k 3 ) = 0.06193, and y:=;:;0.46193.
6
8. Use Runge's Method to approximate y when x = o. 1, given that y = 1 when x = 0 and dy/dx 3x + / .
Here (x0 ,y0 ) = (0,1), h=0.1, fo = 1. Then

k1 hf0 = 0.1,
0.1[3(0+0.1) +(1+0.1/J 0.151,

2
k3 hf(xo+h, Yo+k 2 ) 0.1[3(0+0.ll+(l+0.151) ] = o.16248,
2
k4 hf(x0 +½h, y0 +~k 1 ) = 0. 1[3(0 +0.05) + ( h 0.05) ] = 0.12525,
1
-[0.1 + 4(0. 12525) + 0.16248] = 0.12725, and y = Yo + k ~ 1. 12725.
6
(See Problems 2 and 4a.)
9. Use Runge's Method to approximate y when x = 1. 1, given that y = 1. 2 when x = 1 and dy/ dx = 3x + y2.
Here (xo,Yo) = (1, 1.2), h = 0.1, fo = 4.44. Then

k1 hf0 = 0.444,
k2 hf(xo+h, y0 +k 1 ) 0.1[3(1+0.1)+(1.2+0.444/] 0.600274,

k3 hf(xo+h, y0 +k 2 ) o.1[3(1+0.l)+(l.2+0.60027/l 0.654097,
2
k4 hf(xo+½h, y0 +½k 1 ) = 0.1[3(1+0.05)+(1.2+0.222) ] 0.517208,
~[0.444+4(0.517208) + 0.654097] o. 527822, and

6
y = Yo + k "° 1. 727822.
Comparing this result with that obtained in Problem 46, it is to be noted that the approx-
imation is better than might have been expected in view of the value f 0 = 4.44.
NUMERICAL APPHOXIl\lATIONS TO SOLlTTIONS I <n
10. Use Runge's Method to approximate y when x = o. 8 for that particular solution of dy/ dx v'x + y
satisfying y = 0.41 when x = 0.4.
Here (x0 ,y0 ) = (0,4, 0.41), h 0.4, fo = V0,81 = 0.9. Then
k1 hfo = o. 36,
k2 hf(x0 +h, Yo+k1) 0,4 ✓ 1. 37 ""

o.30120,
k3 hf(x0 +h, Yo+k2) 0.4 ✓ 1, 7112 0,52325,
k., hf(x0 +½h, Yo+"!sk1) = o. 4.IT:"19 0,43635,
and y = Yo + k ~ 0,84811.
11. Solve Problem 10, first approximating y when x =0,6 and then, using this pair of values as
(x 0 ,y0 ), approximate the required value of y.
First, (x 0 ,y0 ) = (0,4, 0.41), h = 0.2, fo = lo.si 0,9. Then
k1 hfo o. 18.,
k2 hf(:xa+h, Yo+k1) 0.2/f:19 = 0.21817,
k3 hf (:xa+h, Yo+k2) o. 2 ✓ 1. 22817 = 0. 22165,
k., hf (:xa+½h, Yo+½k1) = o. 2,
1
k ~ -(k1 + 4k 4 + k3) o. 20028, and y = Yo + k "' o. 61028.
6
Next, take (xo, Yo) = (0. 6, o. 61028), h o. 2. Then Jo = V 1. 21028 "c l.1001,
k1 hfo = o. 22002,
k2 hf (:xa+h, Yo+k1) o. 2 ✓ 1. 63030 0.25537,
k3 hf (x0 +h, Yo+k2) o. 2 ✓1. 66.565 o. 25812,

k., hf (Xo+½h, Yo+½k1) = 0.2 ✓ 1.42029 0,23836,
= o. 23860, and y = Yo + k ~ O. 84888.
12. Solve Problem 10, using the Kutta-Simpson Method.
Here (x0 ,y0 ) = (0.4, 0.41L h = 0.4, fo = ✓o.sT = 0.9. Then
k1 hfo = 0.36,
k2 hf(x0 +½h, Yo+½k1) 0.4 ✓1.19 = 0,43635,
k3 hf(x0 +~h. Yo+½k2) 0.4 ✓ 1. 22817 = 0,44329,
k., hf (x 0 +h, Yo+k3) 0.4/1.65329 = o. 51432,

~ o. 43893, and y Yo + k ~ o. 84893.
13. Use Picard's Method to approximate y and z corresponding to x = 0. 1 for that particular solu-
tion of
dz
dy = f(x,y,z) = X + Z,
dx dx
g(x,y,z) = x-/
satisfying y = 2, z = 1 when x = o.
NFI\IERICAL APPROXIMATIOJ\S TO SOLUTJOJ\S
For the first approximations,

X X
Y1 Yo + Io f(x,yo,zo)dx 2 + Io (l + x)dx 2 + X
I 2
+ 2X'
X 2
Z1 Zo + Ix g(x,Yo, Zo )dx 1 + fo (-4 + x)dx 1 - 4x +bx
0
For the second approximations,
Y2 Yo + foxf(x,y1,Z1)dx 2 + 10
X
(1- 3x + ',x )dx
2 2 + X -
2
2 1 3
~x + -x
6
{X 2 3 I 4
Z2 Zo + fo\(X,Y1,Z1)dx 1 + Jo (-4 - 3x - 3x - x - 4x ) dx
2 4
1 - 4x - ~x - x 3 - ~x ~x 5 ,
2 4 20
For the third approximations,

X
Y3 = Yo + Io f(x,y 2 ,z 2 )dx 2 +
io
x 3 2
( 1 - 3X - -X
2
- X
3
- -X
1
4
4
- -X
1 5
20
)dx
321314 15 lo
2 + x - 2x - 2x - 4x - 20 x - 120 x ,
X
Zo + Io g(x,y 2 ,z,;)dx 1 +
io
x _ 2 7 3
(-4 - 3X + :JX + -X
3
-
31
-X
12
4
+
1 5
-X
2
-
1 o
-X
36
)dx
and so on.
When x = 0.1: y1 2.105 Z1 0, 605

Y2 2.08517 Z,; 0,58397
y3 2. 08447 Z3 0,58672,
14. Use Runge's Method to approximate y and z when x = 0. 3 for that particular solution of the
dy r dz r
system - = x + vz f(x,y,z), - = y- vz = g(x,y,z) satisfying y=0.5, z=0 when x=0.2.
dx dx
Here (xo,Yo,z 0 ) = (0.2, 0,5, 0), h = 0.1, fo 0,2, go 0.5. Then

k1 hfo 0.02,
11 hgo 0.05,
k2 hf (x.o+h, Yo+k1, Zo+ /1) o. 1 (0. 3 + /o.o5) = 0.05236,
l,, hg(x.o+h, Yo+k1, Zo+ /1) 0.1(0.52 - ✓ o.o5) = o. 02964,
k3 hf (x.o+h, Yo+k2, Zo+ 12) 0.1(0.3 + ✓ 0.02964) = o. 047216,

/3 hg(x.o+h, Yo+k2, z0 + l 2 ) 0.1(0.52 - ✓ 0.02964) = 0.034784,
k4 hf(x0 +½h, Yo+ ½k1, zo+½Z1) o. l(o. 25 + ✓ o. 025 ) 0,040811,
14 h g (Xo+ ½h, Yo+ ½k1, zo+½Z 1 ) 0.1(0.51 - /0.025) 0.035189,
1
0.03841, ~ -(Z 1 + 4l4 + Z3 ) 0.03759,
6
and y Yo+ k ~ 0.53841, z z0 + l ~ 0.03759.
15. Use the Taylor Series Method to approximate the value of e corresponding tot =0,05 for that
particular solution of
d2e =
- -
.
8 sin 0 satisfying 0 = 77/4,
de
- = 1 when t = 0.
dt2 dt
Nl'MERICAL APPROXIMATIONS TO SOLUTIONS
The given differential equation is equivalent to the system

d8 d¢
dt = ¢ = f(t, 8,¢), 8 sin 8 = g(t,8,¢)
dt
with initial conditions t=0, 8=77/4, ¢=1. Then
d8
dt
8'= ¢ e'0 = 1 ¢' =- 8 sin 8 ¢~ = - 4 ✓2
e" = ¢' 8"
0 =
- 4 ✓2 ¢" = - 8 e' cos e ¢~' = - 4 ✓2
f)'" = ¢" e"'0 '.=' -4 ✓2 ¢"' = 8(8 1
/ sin 8 - 88
11
case
e1v = ¢'" e1v
0
= 4 ✓2 + 32 cp; = 4 ✓2(1 + 412)
t2 t3 t4
and 8 77/4 + t - 4 ✓2 - 4 ✓2 - + 4(8+ ✓2) - + • • .. • • • • • • = o. 82821.
2 6 24
16. Use the Kutta-Simpson Method to approximate y corresponding to x= 0.1 for that particular so-
2
1 ution of <!___J_ + 2x dy - 4y = 0 satisfying y = o. 2, dy = o. 5 when x = o.
dx2 dx dx
The given equation with initial conditions is equivalent to the system
dy - z = f(x,y,z), dz
= 4y-2xz = g(x,y,z)
dx dx
with initial conditions x = 0, y = o. 2, z = O. 5.
Here (x0 ,y0 ,z 0 ) = (0, 0.2, 0.5), h = 0.1, fo = 0.5, go 0.8. Then
k1 hfo 0.05,
I1 hfo 0.08,
k2 hf (Xo+½h, Yo+ ½k1, 2o+½l1l 0.1(0.54) 0.054,
l.2 hg(Xo+½h, Yo+ ½k1, Zo+½li> o. 1 (0. 846) 0.0846,
k3 hf (x 0 + ½h, Yo+~k2, 2o+½Z2) 0.1(0.5423) 0.05423,
/3 hg(x0 +½h, Yo+½k2, 2o+½Z2) 0.1(0.85377) 0.085377,
k4 hf(Xo+h, Yo+k3, Zo+ /3) 0.1(0. 585377) 0.0585377,
k
1
-(k1 + 2k 2 + 2k 3 + k 4 ) = o. 0541663, and -y = Yo + k ~ 0.25417.
"' 6
196 ~Tl\IERIC-\L APPH.OXIMATIOJ\"S TO S0Ll7TIONS
SUPPLEMENT ARY PROBLEMS
17. Approximate y when x = 0. 2 if dy / dx = x + / and y = 1 when x = 0, using a) Picard 's method,

b) Taylor series, and c) the First Derivative method with n = 4.
Ans. a) y1 1. 2727; b) 1.2735; c) 1. 2503
2
18. Approximate y when x = 0. 1 if dy/dx = x -y
and y = 1 when x = 0, using a) Picard 's method,
b) Taylor series, and c) the Pirst Derivativ0 method with n = 4.
Ans, a) y1 = 0,905, Y2 = 0,9143, y3 = 0,9138; b) o. 9138; c) 0, 9107
19, Use Runge's method to approximate y when x = 0.025 if dy/dx x +y and y = 1 when x = 0,
Ans. 1. 0256
20, Use Runge's method to approximate y when x=2,2 if dy/dx l+y/x and y=2 when x=2,
Ans. 2,4096
21. Use Runge's method to approximate y when x = 0. 5 if dy/dx = ✓x + 2y and y = 0, 17 when x = 0, 3,

Ans. 0. 3607
22, Solve Problem 21 using the Kutta-simpson method. Ans. 0.3611
23, lJse Runge's method to approximate y and z when x = 0.2 for the particular solution of the
system dy/dx = y+z, dz/dx = x 2 +y satisfying y=0.4, z=0,1 when x=0.1.
Ans. y ~ 0.4548, z ~ 0.1450
24, Use the Kut ta-Simpson method to approximate y when x = 0. 2 for that particular solution of
dy
satisfying y = 0. 1, dx = o. 2 when x = 0, 1. Ans, 0.1191
CHAPTER 25
Integration in Series
EQUATIONS OF ORDER ONE. The existence theorem of Chapter 2 for a differential equa-
tion of the form
1) :: = f(x, y)
gives a sufficient condition for a solution. In the proof using power series,
y is found in the form of a Taylor series
where for convenience Yo has been replaced by A0 • This series i) satisfies

the differential equation 1), ii) has the value Y=Yo when x=x 0 , and iii) is
convergent for all values of x sufficiently near x = x 0 •
A. To obtain the solution of 1) satisfying the condition y = Yo when x = 0:

a) Assume the solution to be of the form
2
Y = A0 + A 1 x + A2 x + A 3 x 3 + • • • • • • + An x n + • • · • · •
in which A0 = Yo and the remaining A's are constants to be determined.
b) Substitute the assumed series in the differential equation and proceed as
in the Method of Undetermined Coefficients of Chapter 15.
2
EXAMPLE 1. Solve y' = x + y in series satisfying the condition y = y0 when x = O.
Since f(x,y) ~ x +y is single valued and continuous while of/oy=l is continuous over
2
any rectangle of values (x, y) enc losing (0, y0 ), the conditions of the Existence Theorem are
satisfied and we assume the solution
Now, within the region of convergence, this series may be differentiated term by term
yielding a series which converges to the derivative y'. Hence,
and
A 3 )x 3 + ,. .......
2
y' - x2 - y (A 1 - A0 ) + (2A,2 - A 1 )x + (3A 3 - A2 - l)x + (4A 4 -
n-1
+ (nAn -- An_ 1 )x + .... ·.... = O,
In order that this ,;eries vanish for all values of x tn some region surrounding x = o, it is
necessary and sufficient that the coefficients of each power of x vanish. Thus,
1 1
0 and 3A 3 - A 2 - 1 0 and A 3 + yo,
3 6
1 1
0 and 4A 4 - A3 0 and A4 + y0 ,
12 24
and n ~ 4.
197
]<)8 INTEGRATION IN SERIES
This latter relation, called a recurs ion formula, may be used to compute additional coef-
ficients; thus,
1 1 1 1
60 + 120 Yo' - + 720 Yo•
360
It is also possible to obtain the coefficients as follows:

1
Since An = -An- 1 and An- 1
1
- - An- 2 ,
1
An = ---An-2• But An- 2 = --
1
%-3• ......
n n-1 n(n -1) n -2
1 1 1
hence, An - - - - - - - - A3 ----- - - - - ( 1 + 2IA 0) -(2 +Yo), n > 3.
n(n-1) (n-2) • • • • •4 n(n-1) (n-2) · • • • • 4· 3 n!
When the values of the A's are substituted in the assumed series, we have
1 2 11 3 1 1 ~ 1 n
y Yo+ YoX + - yox + (-+ -Yo)x + (-+ -yolx + ...... + -(2+yo)x +
2 3 6 12 24 n!
2
(Yo + 2) ( 1 + x + _!_x
2
+ _!_x 3 + • • • • · · -+ _!_x n + • • • • • ·) - x - 2x - 2
2! 3! n!
The given differential equation may be solved using the integrating factor e-x thus,
-x 2 -x
ye Jx e dx and
X 2
Using the initial condition, y = y0 when x = 0, C = y0 + 2, and y = (yo + 2) e - x - 2x - 2,
as before.
B. To obtain the solution of 1) satisfying the condition y = Yo when x =x0 :
a) Make the substitution x -x0 = v, that is,
X = V + Xo,
resulting in dy/dv = F(y,y).

b) Use the procedure of A to obtain the solution of this equation satisfying
the condition y = Yo when v = 0.
c) Make the substitution v = x -x0 in the solution.
2
EXAMPLE 2. Solve y1 = x - 4x + y + 1 satisfying the condition y = 3 when x = 2.
First make the substitution x = v + 2 =

V
2
+ y - 3, and obtain dy We seek the solu-
dv
tion satisfying y = 3 when v = 0; hence, we assume the series solution
y =
Then
and
dy - v2 - y + 3
dv
+ (nAn - An_ 1 )vn-l + ........ • O.
Equating the coefficients to zero, we have: A1 = 0, 2A 2 -A 1 = 0 and A2 = 0,

3A3-A 2 -1 = 0 and A3 = 1/3, 4A 4 -A 3 = 0 and A4 = 1/12, .........
INTEGRATION IN SERIES 199
1
The recursion formula An = -An-1 yields
n
1 1 1 2
An - An-1 An-2 - - - - - - - - A3 n ~ 3.
n n(n - 1) n(n - 1) (n - 2) • • .. 4 n!
Thus, y + -v
1 4
+ ......................... + -2v n + .......... ..
12 n!
2 3 2 4 2 n
3 + -(x -2) + -(x -2) + •••••• •••• •••• + - (x - 2) + •••••••
3! 4! n!
LINEAR ~UATIONS OF ORDER TWO. Consider the equation
3)
where the P's are polynomials in x. We shall call x = a an ordinary point of
3) if P0 (a) ~ 0; otherwise, a singular point.
If x = 0 is an ordinary point, 3) may be solved in series about x = 0 as
4) y = A{series in x} + B{series in x},
in which A and Bare arbitrary constants. The two series are linearly inde-
pendent and both are convergent in a region surrounding x = 0. The procedure
for equations of order one in the section above may be used to obtain 4).
See Problems 5-7.
SOLVED PROBLEMS
EQUATIONS OF ORDER ONE.
1. Solve 2x-y
dy in series satisfying the condition y = y0 when x = O.
dx 1-x
Assume the series to be y
where Ao = y0 • Then y'
Substituting,in the given differential equation (1-x)y'- 2x+y = O', we have
or
+(4A 4 -2A 3 )x 3 + .... + [(n+l)An+1-(n-l)An]xn+ ....
2
(A1+A0 )+(2A 2 -2)x+(3A 3 -A 2 )x = 0.
(Note. In finding the general term in the line immediately above, we may write a number of
terms on either side of the general term of the assumed series for y, differentiate each in
getting y', carry out the required multiplications, and pick out the terms in xn OR learn to
write the required term using the general term of the assumed series and its derivative. In
the present problem we wish the term in xn when the substitutions are made in y 1 - xy 1 - 2x + y
= O. First, we need the term in xn of y' when we have the term in xn-l We simply replace n
200 INTEGRATION IN SERIES
.
by (n + 1) 1n n Anx n- l an d ob t a1n
. (n + llnn+
A n h . . t A n A n b .
1x • T e remarn1ng erms -n nx + nx are o v1ous.
)
Equating the coefficients of distinct powers of x to zero yields

1
A1 t A0 = 0 and A1 - Ao, 3A 3 - A2 0 and A3 - A2
3 3'
1 1
2A 2 - 2 = o and A2 1, 4A4-2A 3 0 and A4 - A3
2 6'
n -1
(n+l)An-1-1 - (n-l)An = 0 and An+1 = - - An, (n ? 2).
n + 1
n-2 (n - 2) (n - 3) A (n - 2) (n - 3) (n - 4)
Now An - - An-1 n-2 - - - - - - - - An-3
n n(n- 1) n(n - 1) (n - 2)
(n - 2) (n - 3) (n - 4) • • • • • • 2 • 1 2
- - - - - - - - - - - - A2 n ? 2.
n(n -1) (n - 2) ........ • .. 4· 3 n(n - 1)
2 1 3 1 ~ 1 5 + ....... 2 n + .......
Thus, y YoO-x) + X + - X + - X + -x + ---x
3 6 10 n(n-1)
00
2 n
YoO-x) + k X •
n= 2 n(n - 1)
n-1
Using the ratio test, Ix I lim lxl •
n-oo n+1
The series converges for Ix I< 1,
Note. By means of the integrating factor 1/(1-x) the solution of the differential equation
is y = 2(1-x)ln(l-x) + 2x + C(l-x), The particular integral required is
y = Yo(l-x) + 2(1-x)ln(l-x) + 2x.
2. Solve (1-xy)y' - y = O in powers of x,
Assume the series to be y Then
(1-xy)y' - y
+ 4A 4x 3 + .....
2 2
(1 - A0 x - A1x - A2x 3 - A3 x~ - .... - Anxn+i - "")(A 1 + 2A 2x + 3A 3 x
+ A3 x 3 + •··• + Anxn + ··••)
2
+ nAnxn-l + ••••) - (A 0 + A1x + A2x
Equating to zero the coefficients of distinct powers of x,

A 1 - A0 = O and A1 = A0 ,
2A 2 - A0 A1 - A1 = O and A 2 = ½A 1 (1 + A0 ) = ½A 0 (1 + A0 ).
Thus,
WP shali not attPmpt to obtain a recursion formula here nor to test for convergence.
3. Solve xy'-y-x-1 = O in powers of (x-1).
dy
Setting x = z + 1, the equation becomes (z+l)--y-z-2 o. Since we seek its solution
dz
in powers of z, assume the series to be
y Ao f- A 1 z + A2 z
2
+ A8 z
3
+ A4 z
~
+ " ........ + Anz
n
+ .......... Then
dy
A1 + 2A 2 z t 3A 3 z
2
t 4A4Z
3
+ ........ + nAnz
n-l
+ ........ and
dz
dy
(z + 1) - - y - z - 2
dz
(z + 1) (A 1 + 2A 2 z + 3A 3 / + 4A 4 z 3 + •• • • • + nAnzn-l + • • • • •)
2 3
- z - 2 - (A 0 + A1 z + A2 z + A3 z + • • • • • + Anzn + • • • • •)
2 3
(A 1 - 2 - A0 J + (2A 2 - l)z + (3A 3 + A2 )z + (4A 4 + 2A 3 )z + • • • •· • • • • •
+ [(n + l)An+ 1 + (n - l)Anlzn + • • • •· • •• • • O.
Equating to zero the coefficients of the distinct powers of z,
A1 - 2 - A0 = 0 and A1 2 + Ao, 3A 3 + A2 = 0 and

1
A3 = - -A2
3
- -61 .
1 1 1
2A 2 - 1 = 0 and A2 2. 4A 4 + 2A 3 = 0 and A4 = ·- -A3 -,
2 12
n-1
(n + l)An+ 1 + (n - l)An = O and An+1 - - - An, n ~ 2.
n+l
(-lt (n-2)(n-3)••···•2•1 A
From Problem 1, 2
(-1) n n ~ 2,
n(n-l)••·····•·••4•3 n(n-1)
1 2 1 3 1 ~ n 1 n
and y A0 + ( 2 + .4 0 ) z + - z - -z + - z - ...... + (-1) z + ""." ••
2 6 12 n(n - 1)
Replacing z by (x -1), we have
1 2 1 3 1 ~
y A0 x ~ 2(x - 1) + -(x - 1) -(x-1) + -(x-1) - ••···••··•
2 6 12
00
~
1
A0 x + 2 (x - 1) + (- 1t - -- - ( x - 1t
n= 2 n(n - 1)
n -1
Using the ratio test, lim
r.--+ cc
Iz I lim
n-oo n+ 1
Ix -1/.
The series converges for /x - 1 / < 1.
I 2 y
4 • Solve y - x - e = 0 satisfying the condition y =0 when x =o.
In view of the initial condition, assume the series to be
Then y'
Also, + ......... .
1 + (A 1x + A2x 2 + A3 x 3 + A4 x ~ + .... ) + - 1[22

A1x + 2A 1A2 x 3 + (A 22 + 2A 1A 3)x ~ + .... " ]
2!
1 33 2 ~ 1 ~~
+ -(A 1x +3A 1A2x +• .... ) + -(A 1x +, .... ) + ..................... .
3! 4!
Substituting in the differential equation,
Equating coefficients of distinct powers of x to zero,

1 1
A1 - 1 = 0 and A1 1, 2A 2 - A1 = 0 and A2 = - A1
2 2'
1 2 2
3A 3 - 1 - A2 - ~ A~ 0 and A3
1
= -(1 + A2 + - A1) = 3.
2 3 2
1 3 1 1 3
4A 4 - A3 - A1A2 - - A1 0 and A4 = -(A3 + A1A2 + - A1) 3.
6 4 6
1 2 1 2 1 ~ 17
5A 5 - A 4 --A - A 1A3 - - A1A2 - - A1 = 0 and A5 = - ,
2 2 2 24 60
" ............
and y X +
1
-
2
X
2
+
2
-
3
X
3
+
1
-
3
X
~
+
17
-X
60
5
+ ............
LINEAR EQUATIONS OF ORDER TWO.
5, Solve 2
(l+x )y 11 + xy' -y = O in powers of x.
Here P0 (x) = 1 + x 2 , P0 (0) /c 0 and x 0 is an ordinary point,
We assume the series
+ A3X + A4X ~ + • • • • • • • • • + Anxn + • • • • • • • • • •

2 3
y A0 + A1X + A2X
Then y'
and y"
Substituting in the given differential equation,

2
+ • • • •] + x(A 1 + 2A 2 x + 3A 3 x + 4A 4 x 3 + · • • •
2 2 2
(1 + x ) [2A 2 + 6A 3 x + 12A 4 x + • • • • + n(n- l)Anxn-
+ nAnxn-i + "") - (A 0 + A1 x + A2 x 2 + A3 x 3 + A4 x~ + .... + Anxn + "") = 0,
Equating to zero the coefficients of the distinct powers of x,

1
2A 2 -A 0 =0 and A2 =-A0 , 6A 3 =0 and A3 =0, 12A 4 +3A 2 =0
2
2 n -1
(n + 2) (n + l)An .. 2 + (n - l)An = 0 and - - - An,
n+2
From the latter relation it is clear that A3 = A5 = A7 O, that is, An+ 2 = O if n

is odd. If n is even, (n = 2k), then
2k -3 ( 2k - 3) ( 2k - 5) A k+l 1•3•5·•·••(2k-3)
- - -k A 2 k-2 2 k-~ (-1) - - - - - - - - Ao,
2 2k(2k - 2)
lk!
Thus, the complete solution is
1 2 1 ~ 1 o 5 8
y = AoO + -x - -x + -X - - x + ···••)
2 8 16 128
1 2 ~
Ao [ 1 + ::/ + ..:::.,
k+1 1,3,5 .... ,(2k-3) x2k] +
(-1)
k=2 l k!
1 2
A0 [ 1 + -x
2
n -1
Here lim x
2
lim x ,
2
and the series converges for Ix I< 1,
n ..... oo n+2
6. Solve y"-xy'-y=O
2
in powers of x.
Here P0 (x) = 1 and x =0 is an ordinary point. We assume the series
y Then
y'
y" and
2
y"-xy'-y
Equating to zero the coefficients of the distinct powers of x,
12A 4 -A 1 - A2 = O and A4 = ..!._Ao+ ~A 1

24 12 '
(n - l)An- 1 + An n > 1.
0 and
(n + l)(n+ 2)
The complete solution is y
2
7. SolvP y" - 2x )' 1 + 4xy ~ X
2
+ 2x t· 2 in powers of x.
Assume the seri~s to be
2 3 4 5
y Ac, • A 1 x + A2x + A3 x + A4 x + AAx + •·•··•·· • A71 xn + • • • • • • • • Then
4A,x 3 + 5A 6 x + .. • .. ..
2 4
y' A 1 + 2A 2 x t 3A 3 x +
2.4 2 * 6Ac;x
2 3
+ 20Ae;x + • • .. • • • .... .. n(n-l)Ar;,xr,,-
2
+ •·•••••, and
y" 12A4 x
2 2
+ 20A 5 x 3 + .......
2
Y f f - · 2x )' I + 4xy - X - 2x - 2 (2A2-2) + (6A 3 +4Ao-2)x + (12.4 4 -'-2.4 1 -l)x
+ [ (n + 2) (n + l>An+ 2 - 2(n -1).4n_ 1 + 4An_ 1 ]xn + •• •• ••• " 0,
Equating the coefflcients to zero, we obtain

1
2A 2 - 2 = O and .4 2 cc 1,
12
o.
2(n - 3)
and - - - - - - An-1, n; 3.
(n+l)(n+2)
The complete solution is
y 40 (1
2
3
X ' -
2
45
X
t
- -- X
405
2 9
- ........ ) + Ai(x
6
1
X
4
-
1
63
X
J 1
- 567
-x
10
- ........ ,)
t X
2
+
1
-
3
X
3 1 4
+ -x +
12
1
45
X
b l
126
7 1
405
9
+ -x + -x + --x
1
1184
10
+ .........
8. Solve y" + (x-l)y' t y = 0 in powers of x - 2.
Put x " v +2 in the given equation and obtain (v + 1) dy + y 0 which is to be

dv
integrated in powers of v. Assume the series
Then
and
2
d d
__l'. + (v + l)_l + y
dv 2 dv
Equating the coefficients of powers of v to zero, we obtain

1 1 1 1
A2 " - - (A 0 + .41), A3 ~ - -(A 1 + 42 ) " -(Ao -A 1 ), A, ~ - - (.4 2 + A~)
2 3 6 4
0 and
Thus, noting that v ~ x - 2, the complete solution is
y
1 2 1 3 1
A0 [ 1 - -(x - 2) + -(x-2) + -(x -2)
2 6 12
4
- 1
-(x -2)
20
5
-
1 o
- ( x -2) +
180
...... ]
1 2
+ A1 [(x-2) - ;:;(x - 2) -
1 3 1 4 1 o
-(x - 2) + -(x - 2) - -(x -2) +
6 36
...... ].
6
"
2
9. Solve (1-x)y' = x -y in powers of x.
3 1 1 1 2 1•2 n
Ans. y = A0 (l - x) + x (- + - x + - x + .... • + - - - - - x + .... • )
3 6 10 (n + 2) (n + 3)
10. Solve xy 1 = 1- x + 2y in powers of x - 1. Also integrate directly,

dy
Hint: Let x -1 = z and solve (z + 1)- ~ -z + 2y in powers of z.
dz
2
Ans. y = Ao[l + 2(x-1) + (x-1)] + ~ + (x-1)
2
11. Solve y 1 = 2x + 3y in powers of x,
3
y = A0 [1 + 3x + 9x /2 + 9x 3/2 + 27x /8 + ...... ] + (2x /3 + x /2 + ..... ,)
2 4 4
Ans.
2
12. Solve (x + l)y' = x - 2x + y in powers of x.
2 3 4 5 6
Ans. y = A0 (l + x) - x + 2x /3 - x /3 + x /5 - 2x /15 + ......
13. Solve y" + xy in powers of x.

O
1
R. F. An An_ 3 , n ; 3; convergent for all x.
n(n - 1)
6
y = A0 (1 - x 3/6 + x /180 - ...... ) + Ai(x - x /12 + x /504 - ...... )
4 7
Ans.
2
14. Solve y" + 2x y = O in powers of x.
2
R.F. An = - - - - A n - 4 ; convergent for all x.
n(n - 1)
9
+ A 1 (x - x 5/10 + x /360 - ...... )
4 8
Ans. y = A0 (1 - x /6 + x /168 - ...... )
2
15. Solve y" -xy 1 + x y = O in powers of x.
R.F. n(n-l)A"! - (n-2)An- 2 + An_ 4 = 0, n ~ 4.

4 6
y = A0 (1 - x /12 - x /90 + x /3360 + .... ) + A 1 (x + x 3/6 - x 5/40 - x /144 - .... )
8 7
Ans.
( 1-x )y" - 2xy 1 + p(p + l)y ~ 0, where pis a constant, in powers of x. (Legendre Equation)
2
16. Solve
(n-2-p)(n+p-1)
- - - - - - - - An_ 2 ; convergent for / x / < 1.
n(n- 1)
Ans, y Ao(l _ P(p+l) x2 + (p-2)p(p+l)(p+3) x4 _ .......... )

2! 4!
+ Ai(x _ (p-l)(p+2) x3
3!
+ (p-3)(p-l)(p+2)(p+4) x5
51
- ......... ,)
2 2 1
17. solve y'' + X y 1 + X +x in power,; of x. R. F. An = - An- 4 ; convergent for all x.
n(n -1)
4 8 5 9
Ans. y A0 (1- x /12 + x /672 - ...... ) + Ai(x - x /20 + x /1440 - ...... )
+x 2/ 2+x 3/ 6+x/12-x
4. 6/ 60-x/252-x;672+
J R 1
......
CHAPTER 26
Integration in Series
WHEN x = a IS A SINGULAR POINT OF THE DIFFERENTIAL EQUATION
1)
in which Pi (x) are polynomials, the procedure of the preceding chapter will
not yield a complete solution in series about x =a.
2 2
EXAMPLE 1. For the equation x y 11 + (x - x)y 1 + 2y = 0, x = 0 is a singular point since
Po(0) = O. If we assume a solution of the form
2 3
( i) y = A0 + A1 x + A2 x + A3 x + •••••••••
and substitute in the given equation, we obtain
2 3
2Ao + A1 x + (2A 2 + A1 )x + (5A 3 + 2A 2 )x + ........ • = 0.
In order that this relation be satisfied identically, it is necessary that A0 = 0, A1 = 0,

A2 = 0, A3 = 0, •• • •; hence, there is no series of the form (i) satisfying the given equa-
tion.
A SINGULAR POINT x = a OF 1) IS CALLED REGULAR IF, when 1) is put in the form
yrr+ R1 (x)y'+ 0.
x-a
R1 (x) and R2 (x) can be expanded in Taylor series about x=a.
EXAMPLE 2. For the equation ( 1 + x)y 11 + 2xy 1 - 3y = 0, x = -1 is a singular point since

P 0 (-1) = 1+ (-1) = o. When the equation is put in the form
y 11 + Ri(x) y' + R2(X) 2x -3(x+l)

2 y --y' + - - - y 0,
X + 1
(x + 1) X + 1
(X + 1/
the Taylor expansions about x = -1 of Ri(x) and R 2 (x) are
R 1 (x) = 2x = 2(x + 1) - 2 and
Thus, x = -1 is a regular singular point.
EXAMPLE 3. For the equation x 3y 11

+ x2 y 1 + y = 0, x = 0 is a singular point. Writing
the equation in the form
1 1/x
y/1 + - y' + - 2 y = 0.
X X
it is seen that R 2 (x) = 1/x cannot be expanded in a Taylor series about x o. Thus, x o
is not a regular singular point.
20(1
WHEN x = 0 IS A REGULAR SINGULAR POINT OF 1), there always exists a series solution
of the form
with A 0 f- O, and we shall proceed to determine m and the A's so that 2) sat-
isfies 1).
EXAMPLE 4. Solve in series 2xy 1' + (x + l)y 1 + 3y = 0,
Here, x =O is a regular singular point. Substituting
in the given differential equation, we have
m-1 m m+1
(i) m(2m - l)A 0 x + [(m + 1) (2m + l)A 1 + (m + 3)A 0 ]x + [(m + 2) (2m + 3)A 2 + (m + 4)Ai]x
m+n-l
+ ..... + [(m+n)(2m+2n-l)An + (m+n+2)An_ 1 ]x + .......... O.
Since A0 IO, the coefficient of the first term will vanish provided m(2m -1) ~ O, that is,
provided m = O or m = ½, However, without regard tom, all terms after the first will vanish
provided the A's satisfy the recursion formula
m+n+2
- - - - - - - - An-1, n? 1.
( m + n) ( 2m + 2n - 1)
Thus, the series

m +3 (m+3)(m+4) 2
------x
(m+1)(2m+l)
+ ------------x
(m+ l)(m+2)(2m+ 1)(2m+3)
_ _ _ _ _(_m_+_4_)_(m_+_5_)_ _ _ _ _ x3 + .......... ]
(m + 1) (m + 2) ( 2m + 1) ( 2m + 3) ( 2m + 5)
satisfies the equation

( ii)
The right hand member of (ii) will be zero when m=O or m=½, When m O, we have from 2 1)
with A0 = 1, the particular solution
and when m ½ with Ao 1, the particular solution
The complete solution is then

y Ay1 + By2
2 3 3
A(l - 3x + 2x - 2x /3 + .... ,) + Bh(l - 7x/6 + 21//40 - llx /SO + 0000 • ) ,
The coefficient of the lowest power of x in (i), (also, the coefficient in the right hand
208 INTEGRATION I~ SERIES
member of (ii)), has the form /(m).4 0 • The equation f(m) =0 is called the indicwl equation.
The linearly independent solutions y 1 and y 2 above correspond to the distinct roots m ~ 0
and m " \;2 of this equation.
In the Solved Problems below, the roots of the indicial equation will be:
a) distinct and do not differ by an integer,
b) equal, or
c) distinct and differ by an integer.
The first case is illustrated in the example above and also in Problems 1-2.
When the roots m 1 and m2 of the indicial equation are equal, the solutions
corresponding will be identical. The complete solution is then obtajned as
See Problems 3-4.
When the two roots m1 < m2 of the indicial equation differ by an integer,
the greater of the roots m2 will always yield a solution while the smaller root
m 1 may or may not. In the latter case, we set A 0 = B0 (m-m 1 ) and obtain the
complete solution as
y See Problems 5-7.
The series, expanded about x = 0, which appear in these complete solutions

converge always in the region of the complex plane bounded by two circles cen-
tered at x = 0. The radius of one of the circles is arbitrarily small while that
of the other extends to the finite singular point of the differential equation
nearest x=0. It is clear that the series obtained in Example 4 converge also
at x = 0; moreover, since the differential equation has but one singular point
x = 0, these series converge for all finite values of x.
THE COMPLETE SOLUTION OF
3)
consists of the sum of the complementary function (complete solution of 1)),

and any particular integral of 3). A procedure for obtaining a particular in-
tegral when Q is a sum of positive and negative powers of xis illustrated in
Problem 8.
LARGE VALUES OF x. It is at times necessary to solve a differential equation 1) for

large values of x. In such instances the series thus far obtained, even when
valid for all finite values of x, are impractical.
To solve an equation in series convergent for large values of x or "about
the point at infinity", we transform the given equation by means of the sub-
stitution
X = 1/z
and solve, it possible, the resulting equation in series near z=0.

See Problems 9-10.
SOLVED PROBLEMS
2 2
1. Solve in series 2x y"-xy' + (x + l)y = 0.
Substituting
in the given differential equation, we obtain
Now all terms except the first two will vanish if A2 ,A 3 , • • • · · satisfy the recursion formula
1
1) - - - - - - - - - An-2, n? 2.
( m + n) ( 2m + 2n - 3) + 1
The roots of the indicial equation, (m -1) (2m -1) = 0, are m = ½, 1, and for either value
the first term will vanish. Since, however, neither of these values of m will cause the second
term to vanish, we take A1 ~ 0. Using 1), it follows that A1 = A3 = A5 = ••··· = 0. Thus,
y -------X
1 2
+
1
----------------x • .......... )
(m + 2) (2m + l) + 1 [(m+2)(2m+l) +l)[(m+4)(2m+5) +1]
satisfies
and the right hand member wil 1 be 0 when m = -!, or m = 1.

2 ~ 6
When m = ½ and A0 = 1, we have /x(l - X /6 + X /168 - X /11088 + " " " " • 00
)
2 ~ 6
and when m = 1, with A0 = 1, we have x(l - X /10 + X /360 - X /28080 t ••••••••••).
The complete solution is then
y Ay1 + By2
A/x(l - x 2 /6 + x ~ /168 - x 6 /11088 + • 00
)
2
+ Bx(l - x /10 + x /360 - x /28080 +
~ 6
00
•).
Since x = 0 is the only finite singular point, the series converge for all finite values of x.
2
2. Solve in series 3xy t/ + 2y I t X Y = 0•
Substituting for y, y 1 , and y 11 as in the problem above, we have

111-l 111 m+l 111+2
m(3m-l)A 0 x + (m+l)(3m+2)A 1 x + (m+2)(3m+5)A 2 x + [(m+3)(3m+8)A 3 + A0 ]x
+ " " " + [ (m + n) (3m + 3n - l)An + An- 3 ]xun-i + •..... = o.
All terms after the third will vanish if A3 ,A 4 ,••··· satisfy the recursion formula
1
- - - - - - - - An-3, n ~ 3.
(m+n)(3m+3n-l)
The roots of the indicial equation m(3m -1) = 0 are m = 0, 1/3. Since neither will cause the
second and third terms to vanish, we take A1 = A2 = o. Then, using the recursion formula,
A1 = A4 = A7 = • • • = 0 and A2 = A" = A8 - •• • = 0. Thus the series
_ _ _l_ _ _ x3 + _ _ _ _ _ _1_ _ _ _ _ _ XO _ •••••• )

1) y
( m + 3) ( 3m + 8) (m +3) (m +6) (3m +8) (3m+ 17)
satisfies
3 6
Form= O, with A0 = 1, we obtain from 1) 1 - x /24 + x /2448 - •····· ·
113 3 6
and form= 1/3, with A0 1, we obtain x ( 1 - x /30 + x /3420 - • · · • · •• ).
61 113 3 6
y = Ay1 + By2 = A(l - x\24 + x 2448 - •··•·) + Bx (1 - x /30 + x /3420 - •···•).
The series converge for all finite values of x.
ROOTS OF INDICIAL EQUATION EQUAL.
3. solve in series xy" + y' -y = o.

11
substituting for y, y', and y as in Problems 1 and 2 above, we obtain
2 m +n-l
+ ··••••• + [(m+n) An - An_ 1 ]x + ••··••• = 0.
All terms except the first will vanish if A1 ,A 2 , • • • · · satisfy the recursion formula
1
1) An - - - 2 A"!-1, n ? 1.
(m +n)
Thus,
1 2 _ _ _ _ _1_ _ _ _ _ x3
y - A0 xrr. ( 1 + ---x + ------x + ........ )
2 2 2 2 2 2
(m + 1) (m+l) (m+2) (m+l) (m+2) (m+3)
satisfies
2)
The roots of the indicial equation are m = 0,0. Hence, there corresponds but one series so-
lution satisfying 2) with m = O. l-'owever, regarding y as a function of the independent variables
x and m,
dY' i}Y> ~ ("JY l ?Y>'
dm dm dx dX dm dm
dy' d i_?y) d
and ~?Y> ~~?Y>
dm dm dx dX dX dm dX dX dX dm
and we have by differentiating 2) partially with respect tom,
3) X (dy"
-) + +
2
m A0 x
111-1
ln x.
dm
Frorr, 2) and 3) it follo~s that y -

1 -
yj m=O and d-1
y 2 = _J_
dm m=O
are solutions of the given dif-
ferential equ8tion. Taking A0 = 1, we find

1 2 1 x3
xm lnx [1 + - - - x + ------x
2 2
+
2 2 2
+ .......... ]
(m + 1/ (m+l) (m+2) (m+l) (m+2) (m+3)
2 2 2 2 2
---x - ( 3 2 + - - - - - - )3x - ( 3 2 2
(m + 1)3 (m + 1) (m + 2) (m+l/(m+2) (m+l) (m+2) (m+3)
+ ---------- +
2 3 2
2
2 2 3)x
2 3 - .......... ]
(m + 1) (m + 2) (m + 3) (m + 1) (m + 2) (m + 3)
2xm [ 1 1 1 )x2
y ln x X + ( 3 2 +
3 3
(m + 1) (m+l) (m+2) (m + 1/ (m + 2)
+ ( _____1_ _ _ _ + 1 1 ) x3 + •••• ] •
3 2 2
- - -2- - -3- - - +
2 2 2 3
(mi-l) (m+2) (m+3) ( m + 1) ( m + 2) ( m + 3) (m + 1) (m + 2) (m + 3)
2 3
Then Y1
ylm=O 1 + X +
X
+
X
+ ..........
( 2[)
2
(3! i2
Y2 = c,y I y 1 lnx - 2 [x
1 2
+ _l_(l + -)x
1 1 3
+ _l_(l + - + -)x + ......... ],
c,m m=o (2!
2
i2 (3! i2
2 3
and the complete solution is
1 2 1 3
y (A+Blnx)[l + x + - - x + --x + ........ ]
(2! i2 (3! i2
1 1 2
- 2B[x + - - ( 1 + -)x
(2! i2 2
The series converge for all finite va1ues of x Io.
2
4. Solve in series xy"+y 1 +x y = o.
Substituting for y, y', and y", we obtain
2 m-1 2 m 2 m+l 2 m+2

m A0 x + (m+l) A1 x + (m+2) A2 x + [(m+3) A3 + A0 ]x + ............... .
2 ] m+n-l
+ [ (m + n) An + An_ 3 x + ........ ".. = 0,
The two roots of the indicial equation are equal. We take Ao = 1, A1 = A2 O, and.the re-
l
maining A's satisfying the recursion formula An An- 3 •
2
(m + n)
o, A2 = A5 = A8 ~ , •• = o,
3 1 o _ _ _ _ _1_ _ _ _ _ 9 + " " "• • •)
y X + ------x
2 2 2 2 2
X
(m + 3) (m + 6) (m+3) (m+6) (m+9)
and, following the procedure of Problem 3 above,
1 o
y ln x + 2x m [ 1 x3 ( 31 2 + ------)x +
3
(m+3} (m+3) (m+6) (m + 3/ (m + 6)
1 1
( 3 2 2 + ___2 _ _l__ 2 +
3___ 2 2 3 )x 9 - . . . . . ].
(m+3) (m+6) (m+9) (m+3) (m+6) (m+9) (m+3) (m+6) (m+9)
Using the root m = 0 of the indicial equation,
1 o 1 9 + ......... .
1 - ---·X ---x
6
3 (3!/
-;,-, 1 3 1 1 o 1 1 1 9
and y 2 = _l = y 1 lnx + 2[- x - - - - ( 1 + -)x + - - - ( 1 + - + -)x - ······].
c,mm=O 33 3 5( 2 2 !/
3 1( 3 ! / 2 3

1 o __1__ x9 + •••••••• ]
y (A + B ln x) [ 1 ---x
6
3 (3! /
1
- - - ( 1 + - + -)x
1 1 9
- .......... ].
37(3!/ 2 3
The series converge for all finite values of x IO.
ROOTS OF INDICIAL EQUATION DIFFERING BY AN INTEGER.
5. Solve in series xy" - 3y 1 + xy = 0.
Substituting for y, y 1 , and y'; we obtain
111+n-l
+ [(m+n-4)(m+n)An + An- 2 ] x + •••••••• = 0.
The roots of the indicial equation are m = 0,4, and we have the second special case men-
tioned above since the difference of the two roots is an integer. We take A1 = 0 and choose
the remaining A's to satisfy the recursion formula
1
An - - - - - - - - - An-2, n ~ 2.
(m+n-4)(m+n)
It is clear that this relation yields finite values when m = 4, the larger of the roots, but
when m = 0, A4 - CQ. Since the root m = 0 gives difficulty, we replace A0 by B0 (m - 0) = B0 m
and note that the series
1 2 1 ~ 1 o
y = A0 xm [1 - -----x + --------- X ------------x
(m-2)(m + 2) m( m - 2) ( m + 2) ( m + 4) m(m- 2) (m + 2>2 (m + 4) (m + 6)
+ _ _ _ _ _ _2_ _
1_~-----
2
X R _ • • • •., • • • • • • ]
m(m-2)(m+2) (m+4) (m+6)(m+8)
m 2 1 ~ 1 6
-----x + --------x ----------- X
(m-2)(m+2) (m-2)(m+2)(m+4) ( m - 2) ( m + 2/ ( m + 4) ( m + 6)
+ _ _ _ _ _ _ _1_ _ _ _ _ _ _ XB _ • • • • • • • • • •. • • ]
(m - 2) (m + 2/ (m + 4/ (m + 6) (m + 8)

_,,
xy - 3y
-, -
+ xy (m - 4)mAox
m-l
(m-4)mBox
2 m-1
•
Since the right hand member contains the factor m2. it follows by the argument made in Problem 3
INTEGRATION IN SERIES
t hat y- an d oy
-, with m = o, are solutions of the given differential equation. We find
om
or y ln x +
m +4
B0 x m [ 1 + - - - - - - - x
2
2
_____ 1_ _ _ _ (-1- + 1 + _l_)x 4
om [(m-2)(m+ 2)]
2 (m-2)(m+2)(m+4) m-2 m+2 m+4
1 1 2 1 6
+ -----------(--+ --+ + --)x
(m-2)(m+2/(m+4)(m+6) m-2 m+2 m+4 m+6
1 1 2 2 1 1 8 ]
------2----2-----(--? + - - + + - - + --)x + ••••• •
(m-2)(m+2) (m+4) (m+6)(m+8) m-~ m+2 m+4 m+6 m+8
Using the root m = 0, with B0 1, we obtain
1 4 1 b 1 8
--X +----X ----- X + ••••••••••
YJrr=O 2.2.4 2·22·4•6
2 2
2· 2 • 4 • 6• 8
and
1 1 1 b
- - - - ( 1 + - + -)x
b 2 3
2 3! 1!
1 1 1 1 1 8 1 1 1 1 1 1 1 10
+ ---[(l+-+-+-)+-]x - - - [ ( 1 + -+-+-+-) + (-+-)]x + •••
28 4! 2! 2 3 4 2 210 51 3! 2 3 4 5 2 3
1 4 1 6 1 8
(A + B ln x) { - -- x + ---x ---x + ........ }
3 5 7
2 2! 2 3! 1! 2 4! 2!
1 1 1 b 1 [ 11118
+ B{1 + - - - ( 1 +- + -)X + - - - (l+-+-+-) +-]x
2b 3! 1! 2 3 28 4 ! 2! 2 3 4 2
1 1 1 1 1 1 1 10
- - - - [ ( ! + - + - + - + - ) +(-+-)]x + •••••••• ].
1
2 0 5 ! 3! 2 3 4 5 2 3
Tre series converge for all finite values of x Io.
2
6. Solve in series (x-x )y"-3y'+2y = O.
Substituting for y, y', and y ", we obtain
+ •••••••• + [(m+n-4)(m+n)An-(m+n-3)(m+n)An_ 1 ]xm+n-l + ........ = O,
m+n-3A
The recursion formula is An n- 1 so that
m + n -4
- m[ m-2 m-1 2 m
+ --x
3 m+ 1 4 m+ 2 5 m+ 3 b
+ --x + ........ ]
1) y = A0 x 1 + -- x + -- x + -- X + -- X
m-3 m-3 m-3 m-3 m-3 m-3
satisfies the differential equation

2 m-1
(x-x )y" - 3y' + 2y (m-4)mA 0 x •
The roots m = 0,4 of the indicial equation differ by an integer. However, when m = 0 the ex-
pected vanishing of the denominator in the coefficient of x~ does not occur since the factor m
appears in both numerator and denominator and thus cancels out. Note that the coefficient of
x3 is zero when m = o.
Thus, with A0 = 1,
2 ~ 5 6 J
1 + 2x/3 + x /3 + O - x /3 - 2x /3 - 3x /3 - 4x /3 - • • • • • • • •
and
2
so that y1 = (1 + 2x/3 + x /3) - y 2 /3.
2
The complete solution is y C1Y1 + C2 Y2 = Ci(l + 2x/3 + x /3) + (C 2 - Ci/3)Y2
2 x~
• = A(x + 2x + 3) + B - - -
(1-x/
There are finite singular points at x = 0 and x = 1. The series converge for Ix I< 1.
7. Solve in series xy" + (x - l)y' - y = o.
Substituting for y, y', and y", we obtain
The roots of the indicial equation are m = 0, 2 which differ by an integer. We choose the
A's to satisfy the recursion formula
m+n-2 1
- - - - - - - - - An-1 - - - An-1 •
(m+n-2)(m+n) m+n
At this point we see that no Ai - oo for m = 0, the smaller root, as in Problem 5. This is
due, of course, to the fact that the factor m + n - 2 cancels out. Thus, since
y A0 xm[ 1 - 1
--x +
m+ 1 (m+l)(m+2)
1
X
2 1
( m + 1) ( m + 2) ( m + 3)
X
3
+ ....... ]
satisfies _,, m-1
xy + (x -l)y' - y (m - 2)mA 0 x
'
we obtain, with A0 = 1 and m = 0, m = 2 respectively,
2 3 -x
1 - X + X /2! - X /3! + • • • • •• • • •• e
and
-x
2(e +x-1).
The complete solution is y convergent

for all finite values of x.
PARTICULAR INTEGRAL.
2 2
8. Solve (x - x)y" + 3y' - 2y = x + 3/x near x = 0.
Substituting for y, y', and y" as in Problem 6, we obtain the condition

m-1 m
1) m(4-m)A 0 x + [(m+l)(3-m)A 1 + (m+l)(m-2)A 0 ]x + •·········
X + 3/i.
To find the complementary function, we set the left member of 1) equal to zero and proceed
as before.
m+n-3A
The recursion formula is An n- 1 , and thus
m + n -4
m -1 2 m 3 m+ 1 ~
y + --x + -- X + --x + •••••••••• )
m-3 m-3 m-3
satisfies
2 m-1
2) (x -x)y 11 +3y 1 -2y = m(4-m)A 0 x •
The right hand member of 2) will be 0 when m = 0,4. Form= 0 with A0 = 1, we have
and form= 4 with Ao 1, we have
2
Then y1 (1 + 2x/3 + x /3) - y 2 /3 and (See Problem 6) the complementary function is
y = A(x 2 + 2x +3) + Bx~/(1-x/.
In finding a particular integral, we consider each of the terms of the right member of the
given differential equation separately. Setting the right member of 2) equal to x, that is,
m-1
m(4-m)A 0 x = x, identically,
n -1
we have m = 2 and Ao = tr. For m = 2, the recursion formula is An = - - An_ 1 ; thus, A1 = A 2
n-2
2
= A3 = ••••· = O. The particular integral corresponding to the term x is x /4.
Again, setting the right member of 2) equal to 3/x 2 , that is,
m-1 2
m(4-m)A 0 x = 3/x , identically,
n-4
we have m = -1 and A0 = -3/5. For m = -1, An = - - An- 1 ; thus,
n-5
i A0 , A4 = A6 = A6 = • • • • • = 0. The particular integral corresponding to the term 3/x
2
is
3 -1 3 1 2 1 3
- - x (1 + - x + - x + - x )• The required complete solution is
5 4 2 4
2 Bx~ 3 9 3 1 2
y A (x + 2x + 3) + - -
5x
- -X + -X
(1-x/ 20 10 10
2 Bx~ 1 2 3
= C(x + 2x + 3) + + - X
( 1-x/ 4 5x
Note. A partial check of the work is obtained by showing that the particular integral y
2
x /4 - 3/5x satisfies the differential equation.
Since x = 1 is the only other finite singular point, the series converge in the annular re-
gion bounded by a circle of arbitrarily small radius and a circle of radius one, both centered
at X = 0,
EXPANSION FOR LARGE VALUES OF x.
9, Solve
2
2x (x -l)y" + x(3x + l)y' - 2y O in series convergent near x = oo.
The substitution
2
1 dy dz 2 dy + 1 d y
X y' y"
dz dx x3 dz x~ dz2
transforms the given equation into
2 d\ dy
2(z-z ) - + (1- 5z)- - 2y = O
dz2 dz
for which z = O, the transform of x = oo, is a regular singular point. We next assume these-
ries solution
and obtain the condition
m-1 2 11
m(2m -l)Aoz + {(m + 1) (2m + l)A 1 - (2m + 3m + 2)A 0 }z + .. •· ....... •
2 m+n-l
+ {(m+n)(2m+2n-1),¾- [2(m+n; -(m+n)+l]An-dz + .... 0,
2
2(m + n) - (m + n) + 1 A
The recursion formula is An n-1, and thus the series
( m + n) ( 2m + 2n - 1)
2 2
2m + 3m + 2 2m +7m+7
y z + z2 + • • • •• • • •)
(111+1)(2m+l) ( m + 2) ( 2m + 3)
satisfies
m-l
m(2m-l)Aoz •
Form 0, with A0 1, we have

2
1 + 2z + 7z /3 + 112z 3/45 + ........
1 + -2 + ~
7
+
112
+ ........... '
X
3x 2 45x 3
and for m = ½, with A0 1, we have z½ (1 + 4z/3 + 22/ /15 + 484z 3 /315 +
22 484 + •••••••• ) •
+ -- +
2
15x 315x 3
y = Ay1 + By2 =
2 7
A(l + - + - + - - +
X 2
112
..... ) + Bx
-.L
2 4
(1 + - + - - + - - +
3x
22
2
484
3
...... ) .
3x 45) 15x 315x
II
The series in z converge for z < 1, that is, for all z inside a circle of radius 1, cen-
tered at z = 0. The series in x converge for !xi> 1, that is, for all x outside a circle of
radius 1, centered at x = o.
'I
10. Solve x'y" +- x(l --x)y' t y = o in series convergent near x = oo.
Making the substitution x = 1/z as in Problem 9, we obtain
2
d y dy
1) z - + (3 -1)- + y 0
dz2 dz
for which z O is a regular singular point. We next assume the series solution
substitute in 1), and obtain
m+n-l
+ [(m+n)(m+n+2)An - (m+n-2)An_ 1 ]z + ............ O.
The roots of the indicial equation are m = 0,-2 and differ by an integer. From the recur-
m+ n- 2
sion formula An = ------- An- 1 it is seen that A 2 - oo when m ~ -2. We replace Ao
(m'" n)(m+ n + 2)
by B0 (m • 2) and note that the series
+ ( m - 1) ( m + 2) (m-l)m 2 _ _ _(_m_-_l;....)m_ _ _ z3
(m+l)(m+3)
z + --------z
(m+l)(m+3)(m+4)
+
(m+3/(m+4)(m+5)
+
_ _ _(_m_-_1)_m_(_m_+_2_) _ _ _ z ~ + ........... .
2 2
(m+3) (m+4) (m+5)(m+6)
2
d y dv
z - + (3 -z)---"- t y
dzz dz
Hence,
[ 2m + 1 ( m - 1) ( m + 2\_1_ + _1_)] +
2
(m+ l)(m+3) (m+l)(m+3) m+l m+3
2m -1 (m-l)m 1 1
[------- ( - - + - - + _1_)]/ +
( m + 1) ( m + 3)( m t 4) (m + 1) ( m + 3) (m + 4) m + 1 m+3 m+4
[ 2m -1 (m-l)m 1
(-2- + + _1_)] 23 +
(m + 3/ (m + 4) (m + 5) (m+3/(m+4)(m+5) m+3 m+4 m+5
2
[ 3m+2m-2 (m - l)m(m + 2) ( ~ + _2_ + _1_ + _l_)]z ~
( m '" 3/ ( m T 4 / ( m + 5) ( m + 6) (m+3/(m+4/(m+5)(m+6) m+3 m+4 m+5 m+6
+ ............ } also satisfies this equation.

Using m = -2 with B0 = 1, we find
-2 2 3
Y1 = z (-3z + z ) 3 and
Ylm=-2 X
Y2
~:1 m= -2
y 1 ln z
y 1 ln}
+ z-\1 + 3z + 4/
+ x
2
+ 3x + 4 - 11/3x + l;'sx
llz 3/3 + z~/8 + ..... )
2
+ ••• ••••••• The complete solution
2 2
is y (A+ B ln.I.)(1/x - 3) + B(x + 3x + 4 - 11/3x + 1/Sx + .... , ... ).
X
The series converge for all values of x IO.

Solve in series near x=O.
2 2
11. 2(x +x3)y"- (x -3x )y' + y o.
R.F. An = - An-1
Ans. y = (A ✓
x + Bx) (1 -x + X
2
- X
3
+ .. ... ). Converges for Ix I < 1.
12. 4xy" + 2(1 -x)y' - y 0.
1
R.F. An = - - - An-1
2 (m + n)
2 3 2 3
Ans. y
X
= A (1 + - - +
X
+ --+
X
.... ) + B✓
X
x(l + - +
X
+
X
+ . ... ) .
2
2•1\ 2 •2! 2 3 -3! 1•3 1·3·5 1•3•5·7
Converges for all finite values of x.
R.F. An An-2 , n even; An 0, n odd.

(m + n -1) (2m + 2n -1)
2 ~ 6
Ans. y Ax(l + - +
X
+
X
+
X
............... )
2·5 2·4•5·9 2•4•6·5·9•13
2 ~ 6
+ B✓
xo + ~ + x + ___x _ _ _ + •··•·••·•••··••).
2•3 2•4•3•7
14. xy" + y' + xy = o.

1
R.F. An An- 2 , n even; An 0, n odd.
2
(m +n)
l ~ 6
X X X
Ans. y (A+ B lnx)(l + -2--2 - _2_2_2 + ··········)
2·4 2·4•6
Converges for all finite values of xi O.
R.F. An - - - - An- 2 , n even; An O, n odd.

(m + n -1 /
2 ~
Ans. y (A + B ln x)x (1 - :._ + x

22 2~(2!/
x2 x~ 1 x6 1 1
+ Bx[- - - - - ( 1 +- ) + - - - ( 1 +- +-) + .......... ].
2 2 2 3
2 2~(2!/ 2\3!/
Converges for all finite values of xi 0.
1
16. xy" - 2y' + y o. R.F. An - - - - - - - An-1
(m +n -3) (m +n)
Ans. y =
x3 x~
(A + B ln x) (- - + - - -
x5
12 48
2
x3 19x~
480
137x
+ ••• ) + B (1 + ~ + ~ + - - - - + - - -
5
2 4 36 576 28800
... ).
Converges for all finite values of x f 0.
1
17. xy"+2y'+xy o. R.F. An - - - - - - - An- 2 , n even; O, n odd,
(m+n)(m+n + 1)
2 ~ 2 ~
Ans. y = Ax
-1
(1 - - +
2!
X
4!
X
... ... ) + B(l
X
3!
+ - -
5!
X
...... . )
Converges for all finite values Of X f o.
2
18. x (x +l)y 11 + x(x +l)y' - y = o.
m +n -1
Singular points: x = 0,-1. R.F. An = - - - - An-1 ,
m +n + 1
Converges in the annular region bounded by a circle of arbitrarily small radius

and a circle of radius one, both centered at x = o.
1
19. 2zy" + YI - Y X + 1. R.F. An - - - - - - - - An-1
(m + n) (2m + 2n -1)
Ans. y
2 3 2 3
A(l + x + x /6 + x /90 + •••··) + B/x(l + x/3 + x /30 + x /630 + •······)
1 2 2 3
+ -X (1 + x/15 + X /420 + X /18900 + • • • • •• •) - 1.
6
Solve in series near x = oo.
3 2 1
20. 2x Y" + X YI + Y = 0. R.F. An - - - - - - - - An-1
(m+n)(2m+2n+l)
1 1 1 1 1 1
Ans. y = A ( l - - + - - - - - + •••••• ) + B✓
x(l--+ + •••••• ) •
3x 30x 2
630x 3 x
3 2 1
21. X Y rr + (X + X) YI - y o. R.F. An = - - An-1
m+n
1 11 1 1 1 1 1 1 1
Ans. y = (A + B ln-) (1 + - + - + + •• ·) + B [ - + - ( 1 + -) + -(1+-+-)+•••].
X X 2x2 6x3 X 2.)-2 2 6x3 2 3

CHAPTER 27
The Legendre, Be~sel, and Gauss Equations
THE THREE DIFFERENTIAL EQUATIONS to be considered here are solved by the methods
of the preceding chapter, The first two have important applications in math-
ematical physics. The solutions of all three have many interesting properties.
THE LEGENDRE EQUATION

2
(1-x )y"-2xy' +p(p +l)y = 0,
A solution of this equation in series convergent near x = 0, an ordinary point, was called
for in Problem 16, Chapter 25. Under certain conditions on p which will be stated later, we
shall obtain here the solution convergent near x = oo, Using the substitution x = 1/z (see
Chapter 26) the equation becomes
2
~ 2dy 3dy
( z - z ) - + 2z - + p (p + 1) y 0
dz2 dz
for which z = O is a regular singular point.
Putting we have
{ -m(m -1) .+ p(p + 1) }A 0 zm + { -m(m + 1) + p(p + l)}A 1 zm+l + { [ - (m + 1) (m + 2) + p(p + l)]A 2
2
+ m(m+l)Ao}zm+ + ..... + {[-(m+n)(m+n-l)+p(p+l)]An + (m+n-2)(m+n-l)An_ 2 }zm+n
+ ••••• = o.
We take A 1 = O and An (m + n - 2) (m+ n - l) An- , and see that

2
( m + n) ( m + n - 1) - p (p + 1)
_ _ _m_(;....m_+_l)'---- z2 m(m + 1) (m + 2) (m + 3) z~
y +
(m+ l)(m+2)-p(p + 1) [ ( m + 1) ( m + 2) - p (p + 1) ] [ ( m + 3) (m + 4) - p (p + 1)]
+ m( m + 1) ( m + 2) ( m + 3) ( m + 4) ( m + 5) /' + ••••• ]
[(m + l)(m + 2) -p(p + 1)] [(m + 3) (m + 4) - p(p + l)] [(m + 5) (m + 6)-p(p + 1)]

d2-
(z~ - z2)_J_ + 2z 3 dy + p(p ➔ l)Y (m + p) (-m + p + l)A 0 zm.
dz2 dz
For m = - p with Ao = 1, we obtain
1)
y = z-P[l _ p(p-1) z2 + p(p-l)(p-2)(p-3) z~ _ p(p-l)(p-2)(p-3)(p-4)(p-5) /'
1
2(2p-l) 2•4(2p -1) (2p - 3) 2•4•6(2p -1) (2p -3) (2p - 5)
+ .......... ]
p (p - 1) -2 ~ p(p-l)(p-2) (p-3) (p -4) (p -5) x-6

- '--'----'-- X + c.. p-"-(p_-_1);_(c.:...p_-_2....:)...:(,._p_-_3_;_) X - _
2(2p-1) 2·4(2p -1) (2p -3) 2•4•6(2p -1) (2p -3) (2p -5)
+ •••••••••• ] •
220
THE LEGENDRE. BESSEL. AND GACSS EQUATIONS 221
For m = p + 1 with Ao = 1, we obtain
y = zp+1 [l + (p+l)(p+2) z2 + (p+l)(p+2)(p+3)(p+4) /

2) 2
2(2p+3) 2•4(2p+3)(2p+5)
+ (p+l)(p+2)(p+3)(p+4)(p+5)(p+6) zo + •••••.•••• ]
2· 4• 6(2p + 3) (2p + 5) (2p + 7)
x-P-l [l + (p+l)(p+2) x-2 + (p+l)(p+2)(p+3)(p+4) x-~

2(2p + 3) 2· 4(2p + 3) (2p + 5)
+ (p + 1) (p + 2) (p + 3) (p + 4) (p + 5) (p + 6) -6 + .......... ]
X
2· 4· 6(2p + 3)( 2p + 5) (2p + 7)
Thus,
is the complete solution, convergent for lxl > 1, provided that pi 1/2, 3/2, 5/2, •••··•••
or p i -3/2, -5/2, ...... • • •
Suppose p is a positive integer including 0 and consider the solution y 1 which is a poly-
nomial, say up(x). Putting p = o, 1,2,3,··•·· in 1), we have
Uo(X) = 1, U1(X) = X, U2(X) = x2 - 1/3, U3(X) = x3 - 3x/5, ,.., ..... •• .. • .... •• ,
[h]
i (-l/ k(k-l)·•··· .. •(k-2n+l)
n=O 2n n! (2k-l) ....... (2k-2n+l)
where [½k] denotes the greatest integer ~ ½k (i.e., [½k] 3 i f k = 7, [½k] 4 i f k = 8).
The polynomials defined by
l• 3• 5 .. • • (2p -1)
3) up(x), p = 0, 1, 2,
pl
are called Legendre polynomials. The first few of these are:
Po(x) = Uo(x) 1,
P 1 (x) u1(x) x,
3 2 1
2x -2•
7• 9 ~ 5· 7 3 3· 5
- X - 2- X + -x,
2·4 2·4 2•4
7• 9· 11 o 5•7•9 ~ 3• 5· 7 2 l• 3• 5
--x 3--x + 3--x - --,
2•4• 6 2•4• 6 2•4• 6 2•4•6
1.3 .... 13
P7 (x) - - - - u 7 (x)
7!
It is clear from 3) that Pp(x) is a particular solution of the Legendre equation

2
(1-x )y" - 2xy' + p(p + 1) = o. See Problems 1-6.
222 THE LEGE~DRE, BESSEL. Al\"D GAtSS EQCA TIO NS
THE BESSEL EQUATION

2
x y
11
+ xy , + (x 2 - k2 ) y = 0.
It is evident that x ~ O is a regular singular point. To obtain the solution in series,

convergent near x ~ O, we substitute
and obtain
2 2 m 2 2 m+l 2 2 m +2
(m -k )A 0 x + {(m+l) -k }A 1 x + {[(m+2) -k ]A 2 + Ao}x + ·•···•···• ..
1
We take A1 =0 and .4 72 = - 2 2 An-2 and see that
(m+n) -k
1 2 1 ~
y A0 xm { 1 - ----- X + ----------- X
2 2 2 2 2
(m + 2) - k [(m + 2/ - k ] [(m + 4) - k ]
_ _ _ _ _ _ _ _ _1_ _ _ _ _ _ _ _ x6 + .......... }
[(m + 2) - k
2 2
]
2
[(m + 4) - k
2
] [(m + 6/ - k 2
]
satisfies the equation 2 2 2_ 2 2 m

x y" + xY' + (x - k ) y. = (m - k )Aox •
For m = k with A0 = 1, we obtain

k 1 2 4 _ _ _ _ _1_ _ _ _ _ x6 + ........ }
Y1 = X {1 - ---X + -------x
2
- 3
4(k + 1) 4 •2!(k+l)(k+2) 4 .3! (k + l)(k + 2)(k + 3)
and for m = - k with Ao = 1, we obtain
-k{ 1 ~ _ _ _ _ _1_ _ _ _ _ x6 + ........ } •

Y2 = X 1 - ---
4(1-k)
X
2
+ -------x
2
- 3
4 • 2! (1- k) ( 2 - k) 4 .3! (1-k)(2-k)(3-k)
Note that y 2 = y 1 if k = 0, y 1 is meaningless if k is a negative integer, and y 2 is meaning-

less if k is a positive integer. Except for these cases, the complete solution of the Bessel
equation is y = Ay 1 + By 2 , convergent for all x IO.
The Bessel functions of the first kind are defined by
Jk (x) --Y1
1
(-)
x k 1
2
{-
k!
1 X 2
11.( k +l).1 (-)
2
+
1
2!(k+2)!'2)
/x ~ 1
3!(k+3)! 2
(
-)
6
+ ' .... }.
2k•k!
k
J_k(x) (-1) Jk(x), where k is a positive integer including O.
1 X 2 1 X ~ 1 X 6 + ........ ..
Of these, 1 - --(-) + --(-) --H
(1!/ 2 (2! / 2 (3! / 2
and
are more frequently used.

See Problems 7-10,
THE LEGENDRE. BESSEL. AND GAUSS EQCATIONS 22.1
THE GAUSS EQUATION

2 11 1
(X - X ) y + [)' - ( q:, + /3 + 1 ) X] y - a/3 y = Q•
To obtain the solution in series, convergent near x =0, substitute
and obtain
m(m+y-l)A 0 x '1!-1 + {(m+l)(m+y)A 1 - [m(m+a+,8)+ a/3] A 0 }x "' + ......... .
m+n-1
+ {(m+n)(m+n+y-l)An - [(m+n-l)(m+n+a+/3-l)+a/3JAn- 1 }x + ·••··• O.
We take An (m+n-l)(m+n+a+/3-l)+a/3A and see that

n-1
(m+n)(m+n+y-1)
y Aoxm [l + m(m+a+/3)+a/3x + m(m+a+/3)+af3. (m+l)(m+a+/3+l)+af3 x2

( m + 1) ( m + Y) ( m + 1) ( m + Y) ( m + 2) ( m + y + 1)
+ m( m + a + /3) + a/3 • ( m + 1) ( m + a + /3 + 1) + a/3 • ( m + 2) ( m + a + /3 + 2) + a/3 x3 + •••••••• ]

(m + 1) (m + y) (m + 2) (m + y + 1) (m + 3) (m + y + 2)

2 m-1
(x-x )y" + [y- (a+ /3 + l)x]y' - af3y m(m + y - l)Aox ,
Form= O, with Ao =l, we obtain
y =
1 + a•/3x + a(a+l)/3(/3+1) x2 + a(a+l)(a+2)/3(/3+1)(/3+2) x3 + •••••••• '
1
l•y l• 2·y(y + 1) l• 2• 3•y(y + 1) (Y + 2)
and form= 1-y, y -I 1, with A0 = 1, we obtain
(a-y+ 1)(/3-Y+ 1) (a-y+ l)(a-y+ 2)(/3-Y+ 1)(/3-y+ 2) 2

-------- X + -------------------- X
1(2-Y) l•2(2-y)(3-y)
+ (a-y+l)(a-y+2)(a-y+3)(/3-y+l)(/3-y+2)(j3-y+3) x3 + •••••·•··••].
l• 2•3(2-y) (3 -y) (4 -y)
The series y 1 , known as the hypergeometric series, is convergent for )x) < 1 and is repre-
sented by
F(a, /3, y, x).
1-y
Note that x F(a-y+l, /3-y+l, 2-y, x)
is of the same type. Thus, if y is non-integral (including O), the general solution is
1-y
AF(a,/3,y, x) + Bx F(a-y+l, (3-y+l, 2-y, x).
There are numerous special cases, depending upon the values of a, (3, and y. Some of these
will be treated in the Solved Problems.
224 THE LEGENDRE. BESSEL. ASD GALSS EQL\TIO~S
SOLVED PROBLEMS
THE LEGENDRE EQUATION.
(Rodrigues' Formula)
p
""' (-1) n p! 2P-2n
By the binomial theorem, ~ _....;o__ _ X Then
n=O n! (p - n) !
[½p]
dp 2 P '-...., n p! p-2n
-(x -1) ~ (-1) ----(2p-2n)(2p-2n-l)•"•"•(p-2n-'-l)x
dxp n=O n!(p-n)!
[½p]
r
n=O
(-lt
2p (2p-1) • .. (2p-2n+ 1) (2p-2n) ( p- n-1) ••• (p- 2n+ l) (p-2n) (p -2n-1 )•" 1 •
2p(2p-1),., (2p-2n+l)
2 2
p'
(p-2n)(p-2n-1) .. ,1 n! (p-n) !
xp-2n
Now (in the denominator) 2p(2p-1)" • (2p-2n+l) 2n[p(p-1) .. (p-n+1)][(2p-1)(2p-3)"(2p-2n+l)]
and when multiplied by (p-n)! yields 2np![(2p-1)(2p-3)" .. (2p-2n+l)]. Hence,
[½p]
p-2n
""'
~ (-1)
n=O
n
---------...a--------x
2p -
{"p! [ (
(2p)!p!
1) ( 2p - 3) • • • ( 2p - 2n + 1)] (p - 2n) ! n!
[½p]
r
n=O
(-lt
( 2p) !
2n n! p!
-~p.....:(p_-_1_)•_·_•.....:(p'----2n_+_l_)_ _ xp-2n
( 2p - 1) ( 2p - 3) .. • ( 2p - 2n + 1)
( 2p) !
- - up(x)
p!
2, Show that _ _.;_(2.::...p_-_2n_)_!- - XP-2n

From Problem 1 above,
2P n! (p-n)! (p-2n)!
[~p]
""' n p! P-2n
~ (-1) -----=----(2p-2n)(2p-2n-1)•• .. (p-2n+l)x
n=o n! (p - n) !
[½p]
""' n (p - 2n) ! _ _:pc...!_ _ xp-2n
~ (-1) (2p - 2n) (2p - 2n - 1) .. • • (p - 2n + 1 ) - - -
n=O (p _ 2n) ! n!(p-n)!
[½p]
r
n=O
(-lt (2p-2'l)!p!
n!(p-n)!(p-2n)!
X
p-2n
•
[½p]
r (-lt (2p -2n)! X
p-2n
n=O 2Pn!(p-n)!(p-2n)!
THE LECK\DRE. BESSEL. A~D GATSS EQUATIONS 22S
Using Rodrigues' formula (Problem 1),

5
d,,. r d s
J l 2 2
-(x - l) • - ( x - 1) dx,
,,.+S 5
2 r I s! -1 dx,,. dx
r s r +l
d 2 r- d 2 s .L d 2 r
Let u -(x - 1) and dv -(X -1) (U, Then du - - ( x -1) dx,
0
dxr dx dxr+l
x=l x =l
and
J x=-1
ud1, UV
] x= -1
-
dr rd s-1 ]l
-(x
dx
,. 2
-1) • - - ( x
• dx
s-1 2
-1)
s
-1
d s-'" 2
Now - - . (x -1)
s Jl 0, for J 1,2, .. ,,s-1; hence, after one integration by parts,
dx 5 - J -1
r+s
2 r! s!
A second integration by parts yields
and, afters integrations by parts, we have formally
s dr +s r-
Al f
-1
l
(x
2
- 1) • - - ( x
dxr+s
2
- 1) dx,
2 21'
Suppose s > r. Then, since (x - 1{ X 0
and o. Since rands enter symmetrically, this relation holds also
when r > s. Thus, it holds when r I s.
Suppose s = r. Then A) becomes
Bl
r1 2 r d r
2
2 r
(x - 1 ) , - - ( x -1) dx.
J-l dx21'
d2r fl (x 2 -1) r dx
Now
dx2r
2
- - ( x - 1{ ( 2r) ! , Hence,
1: 2
P,,,.(x) dx
(-1{(2r)!
2 2,,,.(r!/ -1
(-1 { ( 2r) !
22r(r!)2
r
• (-1) • 2 f 0
2'-rr
sin 2r+i 8 d8
(2r) !
22r(r!/
2r+l r !
1 • 3 • • • ( 2r + 1)
2
2r + 1
using the
226 THE LEGENDRE. BESSEL, AND GA{;ss EQUATIONS
I 7T
substitution x cos 8 and Wallis' Formula f 0

2
sin2n+1 8 d8 =
2 n!
n
1. 3•,, (2n + 1)
4 3 2
4• Express f(x) = x + 2x + 2x -x-3 in terms of Legendre polynomials.
3 4 3
+ - • then x and
8 35
6 2
f(x) + - X
7
3 2 3 8 4 20 2 1 108
Now X - P3 (x)
5
+ -X
5
and f(x) = - ?4(X) + - P3(X)
35 5
+ - X + -x
7 5
- 35'
1
2 1 8 4 40
-224
2
X - P2(X)
3
+ -
3
and f (x) - P4(X) + - P3 (x) + - P2(X) + -x
21
-
105
35 5 5
8 4 40 1 224
- P4(X) + - P 3 (x) + - P2(X) + - P1(X) - P 0 (x),
35 5 21 5 105
5. Show that
Now (1-2xt + t 2 )-½ = [1- (2xt - t

2
)]-½ = 1 + ~(2xt _ t 2 ) + (1/ 2 ) <312 \2xt _ / / +
2 2
1, 3" • ( 2k - 5) 2 k- 2 l • 3 • • • ( 2k - 3) 2 k-1 l • 3 .. , ( 2k - 1) 2 k
+ k- (2xt - t ) + -----(2xt - t ) + -----(2xt - t ) +
2
2 (k-2)!
1
(k-1)! l- k! l
2 k k k
But (2xt - t ) (2x) t - ..... '
2 k-1 k~ k~ k~ k
(2xt-t) (2x) t - (k-1)(2x) t +
2 2 (2x/-2tk-2 (k- 2 )(2x/-3tk-l + (k-2~~k-3)(2x)k-4tk _ •••••, etc.
(2xt - t / -
3 2 1 2 l•3• .. (2k-l) k k
1 + xt + (- X - -) t + """ """ • + [ - - - - - 2 X
2 2 2k k!
+ •••••••••
3 2 1 2 1,3 .. ,(2k-1) k
1 + xt + (- x - -) t + .. • .... • .. • •, • • + - - - - - [x
2 2 k!
k(k-1) k-2 k(k-l)(k-2)(k-3) k-4 ] k
X + -------- X + •" • t +
2(2k -1) 2•4(2k -1) (2k -3)
6. Show that Pp(ll = 1, p = 0,1,2,3,"•·• . . . . . .

Put x = 1 in the identity established in Problem 5, Then
2 - l.2 -1 2 p
(1 - 2t + t ) = ( 1 - t) 1 + t + t + ••• ••• + t + ••••••
2
Po(l) + P1(1) t + P 2 (1) t + • •••· + Pp(l) / + identically.
Hence, P0 (1) P 1 (1) = " • ·• = Pp(l) = , •• • • = 1.
nrn LEGE~DRE, BESSEL, AND GAUSS EQUATIONS 227
THE BESSEL EQUATION.

d
7. Prove - Jo (x)
dx
O'.)
~ (-lt
1
(:/n
2 2
n=O (n!)
1 - (:/ + _1_(:)~
2
(2! i2 2
and
d X 1 X 3
- Jo(X) (-) + --(-) + .......
dx 2 l! 2! 2
O'.)
1 X 2n +l
I (-lt
n=O
(-)
n! (n + 1) ! 2
- Ji(x).
More briefly,
O'.) O'.)
d d 1 X 2n d n
-
dx
Jo(X) ~ (-lt --(-)
dx n=O
(n! 2 i2
-
dx
[1 + I
n = l
(-1) ---)
(n! /
1 (
2
2n ]
-
d
[1 - 1 (:/n+l
dx n!(n+l)! 2
8. Prove a) b)
where k is a positive integer.

O'.)
d .,_, n
a) - £.., (-1) - - - - - x 2k +2n
-k+2n
dx n=O 2 n!(k+n)!
O'.)
2k_+_2n
""' (-l)n _ _ _ _ _ x2k+2n-1
£.., k+2n
n=O 2 n! (k+n)!
2k +2n-l
---------x
k+2n-l
2 n! (k + n - 1) !
O'.)
k 1 x k+2n-1 k
X
~ (-lt
n!(k+n-1)!
(2) = X Jk_l(x),
n=O
O'.)
d -k d n 1 2n
b) -
dx
x •Jk(x)
dx n=O
I (-1)
2
k+2n
n! (k + n) !
X
- - : - - - - - - 1_ _ _ _ _ x2n+ 2 ]
2 2
2 k + n+ ( n + 1) ! ( k + n + 1) !
(1) O'.)
"-' ( - l t _ _ _ _1_ _ _ _ 2n+l

-x
-k ~ (-lt 1 (:) k+2n+l =
£.., k+2n-l-l x
n=O 2 n! (k+n+ 1) ! n:O n!(k+n+l)! 2
228 THE LECE:\Dl{E. BESSEL. Al\D CATSS EQL.ATIOJ\"S
9. Prove
where k is a positive integer.
From Problem 8,
and
Then from A),

d k
1) dx Jk(x) + - Jk(x) Jk-1 (x) ;
X
and from B),

d k
2) dx Jk(x) - Jk(x) - Jk+l(x),
X
When 1) and 2) are added, we have a); when 2) is subtracted from 1), we have b).
Note that when b) is subtracted from a), we have
or
Note also that b) is a recursion formula for Bessel functions.
1
10. Show that + -
t
J l (x) + • •• •
-
+O'.J
1
+ k J_k(x) + ........
t
e
½xct -1/tl e
½xt
•e
-x/2t
2 2 x3 t3 n n 2
xt X t X t X X
[1 + - + + + .... + + .. ,,][1 - - +
2 22 2! 23 3! 2n n! 2t
n
+ ..... + (-l)n x + ..... ] •
2n n! tn
In this product, the terms free oft are
1 -
X
(-)
2
+ _1_(:)4 --(-)
1 X 0
+ .... + (-1)
n _1_(~/n
+ . ...... Jo (x)'
2 (2! i2 2 (3! i2 2 (n! )2 2
k
the coefficient of t is
k k+1 k+2 2
X X X X X
-----·- + -----·-- - + .......
l+l(k+l)! 2 2 k+ 2 (k + 2) !
1 x/
(-
1 x k .. 2
H2 +
1
(-)
x k+4 1
(2)
x k .. o
+ ...........
k! 2 l!(k+l)! 2!(k+2)! 2 3!(k+3)!
O'.)
1
~ (-lt (:) k+ 2n and the coefficient of~ is
Jk(x) •
n=O n!(k+n)! 2 tk
THE LECE~DRE. BESSEL. AJ\D CAl;ss EQUATIONS 229
k k-1-2 2
k ' X X X X
(-1) l-- -----·-- -
k!l 2
X k-1-2 X k-1-4
(-) + H2
l!(k+l)! 2 2!(k+2) 1 3!(k+3)!
THE GAUSS EQUATION.

2 3 1
11 • Solve in series (x -x )Y
II I
+ (- - 2x)y - -y = O.
2 4
Here a+ /3 + 1 = 2, y = 3/2, a/3 = 1/4; thus a = /3 = 1/2, and y = 3/2.

2 3
1 1 3 x 3x 5x
Then Y1 = F(a,/3,y,x) = F(-, - , - , x) 1 + - + - + - + .......
2 2 2 6 40 112
1-y - .l
2
and Y2 = x F(a-y+ 1, /3-y+ 1, 2-y, x) x F(0,0,½,x)
and the complete solution is Y = AF (1- , -1, - 3, X ) + B/r

VX,
2 2 2
2
12. Solve in series (x-x )y" + 4(1-x)y' - 2y = O.
Here a+/3+1=4, Y=4, a/3=2; then a=l,/3=2,y=4 or a=2,/3=1,y=4.
For either choice, y1 F(l, 2,4,x) F(2, 1,4,x)

3
3x 5 + .......
4
X X
1 + X
+ - + + - +
2 10 .5 7 28
Since y = 4, the fourth term in y 2 has zero for denominator. However, one of a- y + 2 or
/3-Y + 2 in the third term is zero so that
-3 -3 -3
Y2 = x F(-2,-1,-2,x) x F(-1, -2,-2,x) X (1-x)
and the complete solution is

1-x
y AF(l,2,4,x) + B - - ·
3
X
13. Show that a) F(a,/3,/3,x) = (1-x)-a b) xF(l,1,2,-x) ln(l+x).
a(a+l)/3(/3+1)
a) F(a,/3,/3,x) 1 + - - - - - - x 2 + .......... .
1. 2·/3 (/3 + 1)
a(a + 1) x2 + a(a + 1) (a+ 2) x3 + ......... ..

-a
1 + ax + (1-x)
2! 3!
l•l
[ 1 + -(-x) 1•2·1•2 2 1·2•3·1•2·3 3
+ ....... ]
b) x'F(l, 1, 2,-x) x + ---(-x) + -----(-x)
1•2 1•2•2·3 1•2•3•2·3·4
1 1 2 1 4
X ( 1 - -X + -X -x + ••••••• ) ln(l+x).
2 3 4
TIIE LEGE~DRK BESSEL Al\D GAUSS EQL\TIONS
SUPPLEME!\TARY PROBLEMS
14. Compute a) P 4 (2) = 55.3750, b) J 0 (1) = 0,7652, c) J1(1) = 0.4401, d) F(l,1,10,-1)= 0.9147,
15. Verify each of the following by using the series expansion of Pp (x),
2
a) (x -l)P;(x) = (p+l)[Pp+l(x) -xPP(x)] = p[xPp(x) -Pp_ 1 (x)].
b) P/,+ 1 (x) = xPp(x) + (p+l)Pp(x).
c) (2p+l)Pp(x)
I
= Pp+i(x) -Pp_ 1 (x)
I
= x1 [ (p+l)PP+l(x)+ pPp_ 1 (x).
]
16. If P8 (2) = a and P 7 (2) = b, show that

1 1
a) P:(2) = 2(b-2a), b) P7 (2) = 2(26-a), c) P (2). = -(306- 7a),
3 3 8 8
2
18, Show that the change of independent variable x = t reduces the Legendre equation to a Gauss
equation.
19, a) Show that the change of dependent variable y = x ½z transforms y" + y = 0 into a Besse 1 equa-
tion.
b) Write the solution of the Bessel equation as y = C1 x J½(x)
½ ½
+ Cr J_½(x) and show that
J½(x) and J_½(x) may be defined as ax-½sinx and bx-½cosx respectively,
c) Show that if the relations of Problem 8 are to hold for k = ± ~, then a = b.
Note, These functions are defined with a = /2;n.
20. Use the substitution y = x 112 z and then x = (3t/2/ 13 to show that y" + xy = 0 is a special
case of the Bessel equation, and solve.
Hint: z" + tz 1 + ( / - l/9)z = 0.
3 0 9
Ans. y Ax [ 1 - X
+
X X
+ ...... ,]
22 3 2! 22 32 7 3! 22 33 7•10
0
+ B [1 - X
3
+
X X
9
+ ............ ].
3,2 2! 32 2•5 3! 33 2•5•8
2
21. Solve (x - 3x + 2)y" + 4xy 1 + 2y = 0 after reducing it to a Gauss equation by a substituti-on
of the form x = {; z + T/.
Hint: y = AF(l,2,-4,x-1) + B(x- 1)5 F(6,7,6,x-1) is not a complete solution since the-sixth
term of F(l,2,-4,x-1) becomes infinite.
Ans. y = A F(l,2,8,2-x) + B(2-x)- 7 F(-6,-5,-6,2-x)
22, Express each of the following as Gauss functions.

1
a) - = F(l, /3, /3, x) lim F(a, 1, 1, x/a)
1-x a-"'
1 1 3 2
b) arc sin x = x F(-, - , - , x ) 2
2 2 2 3 x
e) sin x lim xF(a,(3,- ,- - - ) ,
a-"' 2 4a/3
1 3 2
c) arc tan x x F(l, 2' 2' -x l /3-"'
CHAPTER 28
Partial Differt·ntial Equations
PARTIAL DIFFERENTIAL EQUATIONS are those which contain one or more partial deriv-
atives. They must, therefore, involve at least two independent variables. The
order of a partial differential equation is that of the derivative of highest
order in the equation. For example, considering z as dependent variable and
x, y as independent variables,
clz
x- + y-
clz = z
1) or 11 ) xp + yq = z
clx cly
is of order one and
cl 2 z cl 2 z cl 2 2
2) + 3-- + - = 0 or 21) r + 3s + t = 0
clx 2 clx cly cly 2
is of order two. In writing 1 1 ) and 2 1 ), use has been made of the standard
notation: p = -,
clz q = -,
clz
r =
clx cly
Partial differential equations may be derived by the elimination of arbi-

trary constants from a given relation between the variables and by the elim-
ination of arbitrary functions of the variables. They also may arise in con-
nection with geometrical and physical problems.
ELIMINATION OF ARBITRARY CONSTANTS. Consider z to be a function of two independent

variables x and y defined by
3) g(x,y,z,a,b) = 0,
in which A and bar€ two arbitrary constants. By differentiating 3) partially

with respect to x and y, we obtain
clg clg c)z clg clg

4) + = + p- = 0
clx clz clx clx clz
and
clg clg clz clg clg o.
5) + - = + q- =
cly clz cly cly clz
In general, the arbitrary constants may be eliminated from 3), 4), 5) yielding
a partial differential equation of order one
6) f(x,y,z,p,q) = 0.
2
EXAMPLE 1. Eliminate the arbitrary constants a and b from z = ax + by2 + ab,
Differentiating partially with respect to x and y, we have
clz
-=p=2ax and
clz q = 2by.
clx cly
Solving for a and b from these equations and substituting in the given relation, we obtain
2:n
PARTIAL DIFFERENTIAL EQUATIONS
2 2 2
z (1£'.)x + (11)/+ (l!?.)(l'.?) or pq + 2px y + 2qxy 4xyz,
2x 2y 2x 2y
a partial differential equation of order one.
If z is a function of x and y defined by a relation involving but one ar-

Litrary constant, it is usually possible to obtain two distinct partial dif-
ferential equations of order one by eliminating the constant.
EXAMPLE 2. Eliminate a from z = a(x + y).

Differentiation with respect to x gives p = a, so that the partial differential equation
z = p (x + y) is qbtained, Similarly, differentiation with respect to
y gives q = a and the
equation z = q (x + y).
If the number of arbitrary constants to be eliminated exceeds the number

of independent variables, the resulting partial differential equation (or
equations) is usually of order higher than the first.
EXAMPLE 3. Eliminate a,b,c from z = ax+by+cxy.

(i) pcca+cy and (ii) q=b+cx,
These, together with the given relation, are not sufficient for the elimination of three
constants. Differentiating (i) partially with respect to x, we have
2
cl cl z
-p = - = r = 0,
clx clx2
a partial differential equation of order two. Differentiating (ii) partially with respect
toy, we have
cl
-q t = o. of order two.
cly
Differentiating (i) partially with respect toy or (ii) with respect to x, we obtain
2
cl cl cl z
- p = - q = - - = s = c.
cly clx clx cl y
From (i), p = a+ sy and a= p-sy; from (ii), b = q-sx.
Substituting for a,b,c in the given relation, we obtain

z = (p-sy)x + (q-sx)y + sxy px + qy - sxy,
of order two,
Thus, we have three partial differential equations r=O, t=O, z=px+qy-sxy of the
same (minimum) order associated with the given relation.
See also Problems' 1-4.
ELIMINATION OF ARBITRARY FUNCTIONS. Let u = u(x, y, z) and v =v(x, y, z) be independent

functions of the variables x,y,z, and let
7) ¢(u,v) = 0
be an arbitrary relation between them. Regarding z as the dependent variable

and differentiating partially with respect to x and y, we obtain
8) and
PARTIAL DIFFERKVIJAL EQCATIOJ\S
9)
Eliminating c¢ and c¢ from 8) and 9)' we have

cu CV
cu + p-
cu CV
- +
CV
p-
ex CZ ex CZ cu
= (-ex
cu+ pcu)
- (-CV - (cu
cv ?:Iv
+ qCV)
- + q-) ( - + p - )
cu cu CV CV CZ cy CZ cy CZ ex CZ
cy + q-
CZ cy + qCZ-
cu CV cu CV (cu -CV - -cu CV cu CV cu CV)
= ex -cy - -cy ex + p -
cz cy cy -) +
CZ
q(-
ex CZ CZ ex = o.
Writing A.P = cu CV cu CV A.Q

cu CV cu CV cu CV cu CV
cy cz cz cy CZ ex ex CZ ex cy cy ex
this takes the form
Pp + Qq = R,
a partial differential equation linear in p and q and free of the arbitrary
function ¢(u, v).
EXAMPLE 4. Find the differential equation arising from ¢(z/x 3 , y/x) = O, where¢ is
an arbitrary function of the arguments.
We write the functional relation in the formef,(u,v) =O with u=z/x 3 and v=y/x. Dif-
ferentiating partially with respect to x and y, we have
o.
The elimination of c¢ and o;p yields

CU (JV
2
p/x 3 - 3z/x
4
-y/x
3z/x 5 + qy/x 5
4
p/x - O or px + qy = 3z,
q/x3 1/x
The arbitrary functional relation may also be given by ...:.. = j(?:.) or z = x 3 f(?:.), where
X3 X X
3
f is an arbitrary function of its argument. Using v = y/x and differentiating z = x f(v)
with respect to x and y yields
p = 3x 2 f (v) + x3 df ov 3x2 f (v) + x3 (dfl (- 1'.._) 3x 2 f(v) - xyf'(v),

dv ex dv x2
q x3 df CV x3 (df)(~) x2f'(v).
dv cy dv x
When f'(v) is eliminated from these, we have

px + qy = 3x 3 f (v) 3z
as before. See also Problems 5-8.
PARTIAL DIFFERE~TIAL EQCATIOl\S
SOLVED PROBLEMS
.. 1, Eliminate a and b from z ~ (x 2 + a)(y2 • b),
Differentiating partially with respect to x and y, p = 2x(y2-,. /J) anct q =
p 2
/ +b = - , x + a (_J__) (_I?__) or pq = 4xyz.
2x 2y 2x
2
\Ve could also eliminate a and b as follows: pq = 4xy(y2 + b) (x + a) = 4xyz.
) 2. Find the diffprential equation of the family of spheres of radius 5 with centers on the plan~
X = y.
The equation of the family of spheres is 1) (x -a/+ (Y -a/ + (z - b/ = 25, a and b being
arbitrary constants. Differentiating partially with respect to x and y, and dividing by 2, we
have
(x-a) • (z-b)p = 0 and (y-a) + (z-b)q = O.
Let z - b = -m; then x - a= pm and y - a= qm. Making these replacements in 1), we get
2
m (/+q2+1) = 25.
2
Now x-y = (p-q)m. Then m
x-y, (x - y) (/ + q2 + 1) 25, and the
p-q 2
(p -q)
2 2 2
required differential equation is (x - y) (p + q + 1) 25(p-q/.
3. Show that the partial differential equation obtained by eliminating the arbitrary constants
a,c from z = ax+ h(a) y + c, where h(a) is an arbitrary function of a, is free of the vari-
ables x,y,z.
Differentiating z = ax+ h(a) y + c partially with respect to x and y, we obtain p = a and
q The differential equation resulting from the elimination of a is q = h(p) or f (p,q)
= h(a).
= o, where f is an arbitrary function of its arguments. This equation contains p and q but
none of the variables x,y,z.
~ 4. Show that the partial differential equation obtained by eliminating the arbitrary constants a
and b from
z =ax+ by+ f(a,b),
the extended Clairaut equation, is

z = px + qy + f(p,q).
Differentiating z ax+· by+ f (a, b) with respect to x and y yields p =a and q = b, and the
required differential equation follows immediately,
- 5, Find the differential equation arising from ¢(x+y+z, x +/-z

2 2
) = o.
2 2
Let u=x+y+z, v x +/-z sothatthegivenrelationis ¢(u,v) o.
Differentiation with respect to x and y yields
qp_(l+p) + qp_(2x-2zp) = o and qp_( 1 + q) + qp_( 2y - 2zq) O. Eliminating 9P. and M, we have
OU ov OU ov OU ov
1 +p 2x - 2zp
2(y-x) + 2p(y+z) - 2q(z+x) 0 or (y + z)p - (x + z)q x - y.
1+q 2y - 2zq
PARTIAL DIFFERENTIAL EQUATIONS
6. Eliminate the arbitrary function ¢(x + y) from z = ¢(x + y).
Let x + y =u so that the given re lat ion is z = ¢(u).
Differentiating with respect to x and y yields P = d¢ = ¢' (u) and q = ¢ 1

( u) •
du
Thus, p = q is the resulting differential equation.
, 7, The equation of any cone with vertex at P0 (x 0 ,y0 ,z 0 ) is of the form ¢(x-xo, Y-Yo)
z - z0 z - z0
O.
Find the differential equation.
Let
X -Xo
-- = u,
Y -Yo
-- = V so that the given relation is ¢(u, v) o.
Z - Zo Z - Zo
Differentiating with respect to x and y, we have
c!<p(- q x -x0 ) + cxp(_l_ o.

ou (z - zo)
2 ov z - z0
Eliminating cJ<p and cJ<p we obtain p (x -x 0 ) + q(y - y0 ) Z - Zo,

OU ov
V 8. Eliminate the arbitrary functions f(x) and g(y) from z = yf(x) + xg(y).

1) p = y f 1(x) + g(y) and 2) q = j(x) + X g 1(y),
Since it is not possible to eliminate f,g,f', g' from these relations and the given one, we
find the second partial derivatives
3) r = yf"(x), s = f'(x) + g'(y), t=xg"(y).
1 1
From 1) and 2) we find f'(x) -[p -g(y)] and g'(y) = -[q-f(x)]. Hence,
y X
1 1
s = f'(x) + g'(y) = -[p-g(y)) + -[q-f(x)].
y X
Thus, xys = x[p-g(y)] + y[q-f(x)] = px + qy - [y f(x) + xg(y)] = px+ qy-z is the

resulting partial differential equation.
Note that the differential equation is of order two although, in general, a higher order is
expected. However, since one of the relations 3) involves only the first derivatives off and
g, it is possible to eliminate f, g, f 1 , g I between this relation, 1), 2), and the given relation.
J 9, Find the differential equation of all surfaces cutting the family of cones X
2
+y -a
2 2 2
Z 0
orthogonally.
Let z = f(x,y) be the equation of the required surfaces. At a point P(x,y,z) on the surface,
a set of direction numbers of the normal to the surface is [p,q,-1). Likewise, at Pa set of
direction numbers of the normal to the cone through Pis [x,y,-a 2 z]. Since these directions
are orthogonal ,
2
px + qy + a z = O.
2
The elimination of a between this and the given equation yields the required differential
equation 2
z (px + qy) + x + y2 = O.
2J6 PARTIAL DIFFEREYJ'JAL EQC\ TJOJ\"S
10. A surface which is the envelope of a one-parameter family of planes is called a developable
surface. (Such a surface can be deformed (developed) into a plane without stretching or tear-
ing. J Obtain the dift'erential equation of rlevelopable surfaces.
Let z = j(x,y) be the equation of a developable surface.

The tangent plane at a point (x 0 ,y0 ,z 0 ) of the surface has equation
1) F (X-Xo)P + (y-yo)q - (Z-Zo) = o.
Now wh<2n p anrl q sat isty a relation <P(p,q) O, 1) is a one-parameter family of planes
having z ~ j(x,y) as envelope. Thus ¢(p,q) = O or q A(p) is the required differential
equation.
The cone of Problem 9 is a clevelopable surface since p satisfies ¢(p,q)
2 2
a (p2+q ) - 1 = O.
11. Eliminate the arbitrary functions ¢ 1 and ¢ 2 from
in which m1 f m2 are fixed constants.
Differentiatin_g partially, we obtain

2 2 2 2 2, 2
2 d "¢1 2 d ¢2 , d ¢1 d ¢2 . d <P1 d ¢2.
r = m1 + m2 s = m1 + m2 +
2 2 2 2 2 2
du dv du dv du dv
2 2
Eliminating d ¢ 1 , d ¢2 we have 0
du 2 dv 2
1 1
2
12. Show that (a) z = ax 3 + by 3 and (b) z ax 3 +bx y+cx/+d//x give rise to the same dif-
ferential equation.
a) Differentiating z dx 3 + by 3 partially with respect to x and y, we have

and q = 3by2.
Thus, px + qy = 3(ax 3 + by 3) = 3z is the resultirig differential equation.
ax 3 +bx y+cx/+dy /x
2 4
b) Differentiating z = partially with respect to x and y, we have
2 4 2
+ 2cxy + 4dy 3/x.
2 2
p = 3ax + 2bxy + cy - dy /x and q = bx
px + qy = 3(ax 3 + bx y + cxy 2 + dy /x)

2 4
Thus, 3z as before,
The fact that these two equations, one with two arbitrary constants and the other with four,
give rise to the same differential equation will indicate the subordinate role which the ar-
bitrary constant will play here, In its place we will have arbitrary functions. Since (a) may
be written as
z ax'\ + by 3 = x 3 [a+ b(yjx}] = x 3 ,g(y/x),
while (b) may be written as

3
z = x 3 [a+ b(y/x) + c(y/x/ + d(y/xi4J x ,h(y/x),
3
each is a particular case of z = x •f(y/x) considered in Example 4.
PARTIAL DIFFERE~TIAL E<)lATJO:\"S
SUPPLEMENT ARY PROBLEMS
Eliminate the arbitrary constants a,b,c from each of the following equations.
2 2 2 2
13. z = (X - a) + (y-b) Ans. 4z = p -1- q
14. z = axy + b xp - yq 0
15. ax +by+ CZ = 1 ,. = 0, s = 0, or 0
, 2 2Y 2
16, z = axey + 2 a e + b q = xp + p
17, z = xy + y ~ + b pq = xp + yq
2 2 2, 2 2 2
18, x /a + //62 -1- z /C 1 xzr+xp -zp = O, yzt+yq -zq = O, or zs + pq = 0
Eliminate the arbtirary constants a,b and the arbitrary functions ¢,f,g.
Ans. 2z = xp + xq
20. xyz = <p(x + y + z) x(y-z)p + y(z-x)q z(x-y)

2 2
21. z = (x + y) <p(x - y ) yp + xq = z
y
22. z = f (x) + e g(x) t- q = 0
23. X = f(z) + g(y) ps - qr = 0
24. z =f (xy) + g (x + Y) Ans. x(y-x)r 0
25. z f(x+z) + g(x+y) Ans. qr - (l+p+q)s + (l+p)t 0

2
26. z = ax + g(y) p - xr O or s = O
2 2 2
27. z ±(a + 2)x + axy + bx + <p(y + ax) r - 2t + rt s = 2
28. Find the differential equation of all spheres of radius 2 having their centers in the xOy
2 2 2 2 2 2
plane. Hint: Eliminate a and b from (x-a) + (y-b) + z = 4. Ans. z (p + q + 1) 4
29. Find the differential equation of planes having equal x- and y-intercepts. Ans. p- q O
30. Find the differential equation of all surfaces of re vol ut ion having the z-axis as axis of
rotation. Hint: Eliminate¢ from Ans. yp -xq = 0
CHAPTER 29
Linear Partial Differential Equations of Order One
THE PARTIAL DIFFERENTIAL EQUATIONS of order one
and L,) px2 + qy = z3
are called linear to indicate that they are of the first degree in p and q.
Note that, unlike linear ordinary differential equations, there is no restric-
t ion on the degree of the dependent variable z.
All partial differential equations of order one which are not linear, as
21) p2+q2=1 and 22) p + ln q = 2z 3 ,
are called non-linear.
LINEAR PARTIAL DIFFERENTIAL EQUATIONS OF ORDER ONE. Equati-on 1 1 ) was obtained in

Chapter 28, Example 4, from the arbitrary functional relation
3) ¢(z/x 3 , y/x) = 0
or its equivalent z/x 3 = f(y/x). This solution, involving an arbitrary func-
tion, is called the general solution of 11 ).
The differential equation was also obtained (Chapter 28, Problem 12) by
eliminating the arbitrary constants from
z = ax 3 + by 3
A study of the problems of that chapter indicates that relations involving two
arbitrary constants usually yield non-linear partial differential equations
of order one, while those involving more than two arbitrary constants yield
equations of order higher than one. However, as was pointed out in Chapter 28,
Problem 12, both of these relations are particular cases of the arbitrary
functional relation 3). It is clear then that the general solution of 1) yields
a much greate~ variety of solutions than that obtained (in the case of ordi-
nary differential equations) through the appearance of arbitrary constants;
for example,
z/x 3 = A sin(y/x/ + B cos (y/x) + C ln(y/x) + Dey/x + E(y/x/ 2
is included in the general solution 3).
THE GENERAL SOLUTION. A linear partial differential equation of order one, involv-
ing a dependent variable z and two independent variables x and y, is of the
form
5) Pp + Qq = R
where P,Q,R are functions of x,y,z.
c)z
If P = 0 or Q = 0, 5) may be solved easily. Thus, the equation = 2x + 3y
clx
has as solution z = x 2 + 3xy + ¢(y), where ¢ is an arbitrary function.
2:38
LINEAR PARTIAL EQUATIONS OF ORDER ONE
Lagrange reduced the problem of finding the general solution of 5) to that

of solving an auxiliary system (called the Lagrange system) of ordinary dif-
ferential equations
dx = dy dz
6)
p Q R
by showing (see Problem 7) that
7) ¢(u,v) = 0, (¢, arbitrary)
is the general solution of 5) provided u= u(x,y, z)_= a and v =_v(x,y, zJ_-=' b are
two inaepe-naeht solutions of 6). Here, a and b are arbitrary constants and at
least one of u,v must contain z.
~ ➔ EXAMPLE 1. Find the general solution of

1) px + qy 3z.
dx dy dz
The auxiliary system is
X y 3z
dx dz dx dy
From
X 3z'
we obtain u = z/x 3 = a; and from
X
-,
y
we obtain V y/x b.
Thus, the general solution is ¢(z/x 3 , y/x) = 0, where ¢ is arbitrary.

dy dz
Of course, from - = - , we obtain z/y 3 = c, and we may write
y 3z
or A(z/y3, y/x) = O,
where if; and A are arbitrary. However, these are all equivalent and we shall call any one
of them the general solution.
The above procedure may be extended readily to solve linear first order
differential equations involving more than two independent variables.
EXAMPLE 2. Find the general solution of
CZ CZ CZ
X - + y - + t xyt,
ex cy ct
z being the dependent variable.
dx dy dt dz
X y t xyt
We obtain readily u = x/y = a, v = t/y = b.

A third independent solution may be found by using the multipliers yt, xt, xy, -3. Since
x(yt) + y(xt) + t(xy) + (xyt) (-3) 0,
yt dx + xt dy + xy dt - 3 dz O
and xyt - 3z = c.
Thus, the general solution is ¢(x/y, t/y, xyt -3z) o.

2 Hl LINEAR PARTIAL EQUATIONS OF ORDER ONE
COMPLETE SOLUTIONS. If u = a and v = b are two independent solutions of 6) and if

a,~ are arbitrary constants,
8) u = av + ~
is called a complete solution of 5). Thus, for the equation of Example 1,
z/x 3
= a(y/x) + ~
is a complete solution.
A complete solution 8) represents a two-parameter family of surfaces which
does not have an envelope, since the arbitrary constants enter linearly, It
is possible, however, to select one-parameter families of surfaces from among
8) which have envelopes. As shown in Problem 8, these envelopes (surfaces)
are merely particular surfaces of the general solution.
SOLVED PROBLEMS
.., .: 1. Find the general solution of 2p + 3q = 1.
d.x dy dz
2 3 1
From d.x = dz , d.x dy'

we have x - 2z = a; and from we have 3x - 2y b, Thus, the gen-
2 1 2 3
eral solution is q>(x -2z, 3x -2y) o.
The complete solution x-2z = a(3x-2y)+~ is a two-parameter family of planes. The one-
parameter family determined by taking~= a 2 has equation
2
A) x - 2z = a(3x - 2y) + a ,
Differentiating A) with respect to a yields 0 = 3x -2y+ 2a or a= - ½(3x -2y).

2
Substituting for a in A), we obtain the enve 1ope, a parabolic cylinder, x - 2z = - -4-< 3x - 2y) ,
This cylinder is clearly a part of the general solution.
2 2 2
2. Find the general solution of y zp - X zq X y,
The auxiliary equations are

d.x -5!I__ dz
2 2 2
y z -x z X y
2 2 d.x
From.!!:.::_ = -5!I__ or z dz+ ydy = 0, we have y + z = a; from - -dy
2-· wehave x 3 +y 3 =b,
2 2 2
X y -X Z y z -x z
2
Thus, the general solution is ¢(/+z , x 3 +y 3 ) o.
~ ' 3. Find the general solution of (y-z)p + (x -y)q = z - X,

d.x -5!I__ dz
·--..
y-z x-y z -x
Since (y-z) +(x-y) +(z-x) = 0, d.x+dy+dz = 0 and x+y+z = a,

Since x(y-z)+z(x-y)+y(z-x) =0, xd.x+zdy+ydz =0 and x 2 +2yz b,
2
Thus, the general solution is ¢(x + 2yz, x+y+z) = 0.
+ 2yz = a(x + y + z) + ~
2
The complete solution x represents a family of hyperboloids.
LINEAR PARTIAL EQUATIONS OF ORDER ONE 241
I
J 4. Find the general solution of
d.x dy dz
2 2 2 2xy 2xz
X -y -z
dy dz
From - , we obtain y/z = a.
2xy 2xz
dz x d.x + y dy + z dz x d.x + y dy + z dz dz 2(x d.x + ydy + z dz)
From or
2xz x(~2-y2-z2)+y(2xy) +z(2xz) x(x2 + y2 + z2) z x2 + y2 + z2
2 2 2
X +y +z
we obtain = b.
z
2 2 2
+y + z
Thus, the general solution is q>('!.__'
z
X
z
= o.
The complete solution x2 + y 2 + z 2 = ay + /3z consists of the spheres through the origin with
centers on the plane yOz.
;5.Solve ap+bq+cz=0.
d.x dy dz d.x
The auxiliary system is From we obtain ay- bx = A.
a b -cz a
dz dx -ex/a
If a I o, yields ln z = - ~x + ln B or z = Be , and the general solution
-CZ a a
-ex/a dz dy -cy/b
may be written as z = e ¢( ay - bx) • If b I o, -CZ
b yields z = Ce , and the
-cy/b
general solution may be written as z = e f (ay- bx).
6. Solve 1) 2p+q+z=0, 2) p - 3q + 2z = 0, 3) 2p+3q+5z=0, 4) q + 2z = 0.
1) Comparing with Problem 5 above, a= 2, b = 1, c = 1.

-02 y
The general solution is z = e q>(2y-x) or z = e- tp(2y-x),
- 2x 2y/3
2) Here, a = 1, b = -3, c = 2, The general solution is z = e q>(y + 3x) or z = e f (y + 3x).
-5x/2 -5 y/3
3) The general solution is z = e ¢ ( 2y - 3x) or z = e tp( 2y - 3x).
-2y - 2y
4) The general solution is z e q>(-x) = e tp(x).
7. · depen den t so 1 ut ions

Show that if u=u(x,y,z)=a and v=v(x,y,z)=b are t wo 1n . of -d.x =dy dz,
-=-
p Q R
where P,Q,R are functions of x,y,z, then ¢(u,v) = 0, with¢ arbitrary, is the general solu-
tion of Pp+ Qq = R.
Taking the differentials of u = a and v = b, we have
oud.x+oud
- y + -OU dz = 0, ov d.x + -ov d y + ov
- dz o.
dX oy oz ox oy oz
Since u and v are independent functions, we may solve for the ratios
d.x : dy : dz
OU 0V
(-----)
OU 0V
(-
OU 0V
- - OU 0V
- -) (- -
dU ov - -ou -ov) = p: Q: R.
oy oz oz oy oz ox ox oz ox oy oy ox
But these are the relations (see Chapter 28) defining P,Q,R in the equation Pp+ Qq =R whose
general solution is q>(u,v) = 0.
242 LINEAR PARTIAL EQUATIONS OF ORDER ONE
8. Let u = av+ /3 be a complete solution of Pp+ Qq = R. From this two-parameter family of sur-
faces, select a one-parameter family by setting /3 = h(a), where h is a given function of a,
and obtain the envelope,
The envelope of the family
1) u = av + h(a)
is obtained by eliminating a between 1) and

2) 0 = v + h' (a),
Solving 2) for a= µ(v) and substituting in 1), we have

3) u = v•µ(v) + h[µ(v)] = l\.(v),
Now 3) is a part of the general solution ¢(u,v) = O, Thus, unlike the case of ordinary dif-
ferential equations, the envelope is not a new locus,
If h(a) is taken as an arbitrary function of a, l\.(v) is an arbitrary function of v, and 3)
is the general solution, Thus, the general solution of a linear partial differential equation
of order one is the totality of envelopes of all one-parameter families 1) obtained from a com-
plete solution, It is to be noted that when h(a) is arbitrary, the elimination of a between
1) and 2) is not possible; thus, the general solution cannot be obtained from the complete
solution,
9. Show that the conditions for exactness of the ordinary differential equation
µ(x,y) M(x,y)dx + µ(x,y) N(x,y)dy = 0
is a linear partial differential equation of order one, Thus, show how to find an integrating
factor of Mdx + N dy = 0, (See Chapter 4.)
If µMdx + µNdy = O
is exact, then 0 or Moµ oN oM

-(µM) µ(---),
oy oy ox oy
This is a linear partial differential equation of order one for which the auxiliary system is
dx dy dµ
1)
-N M µ< oN _ oM)
ox oy
Any solution, involvingµ, of this system is an integrating factor of M dx + N dy = O,
Writing 1) in the form
oN oM oN oM
ox oy dx o_x_ _o_y dy dµ
2) -, it is evident that if
-N M µ
oN oM oN oM
~ - ~ =
) th enµ= eff(x)dx 1s
f( x, . t egrating f actor; or 1.fox
. an 1n - oy
----=- = g(y), µ = efg(y)dy
~ M
is an integrating factor. Moreover, if the equation is linear (that is, y' +Py= Q ), then M
= Py - Q, N. = 1 an d 2) becomes P dx p dy
-- - = dµ
- an d µ = ef p dx 1s . t egra t.1ng f ac t or.
. an 1n
Py-Q µ
(2x 3y-/)d.x
4
10. Find an integrating factor for _ (2x +xy)dy = o. (See Problem 9 above.)
oM 3 oN 3
oy = 2x - 2y, -(Sx +y),
ox
LINEAR PARTIAL EQUATIONS OF ORDER ONE 241
cb: dy dµ
We seek a solution involvingµ of
2
2x~ + xy 2x3y -Y µ(y -10x 3 )
dµ -2y dx - 3x dy -2y dx - 3xdy dµ -2y dx - 3xdy

From or
µ xy
µ(y - 10x 3 ) -2y ( 2x ~ + xy) - 3x ( 2x 3 y - / ) xy(y - 10x 3 )
we obtain ln µ = -2 lnx - 3 ln y, Thus, µ = x- 2 y- 3 is an integrating factor.
2
~)'ind the integral surface of x p + y2q + z 2 = 0 which passes through the hyperbola
xy =X + y, Z = 1.
dx dy dz
x2 y2 -z2
dx dz X+Z ciy dz y +z
From we obtain u a, and from we obtain V b.
x2 -z2 XZ y2 -z2 yz
We first e 1 iminate x 0 , Yo, z 0 between x 0 Yo = x0 +Yo, zo = 1 and u = Xo + zo = Xo +1 =a

xozo Xo
= Yo + Zo Yo+ 1
=- 1 1
and V
YoZo
- - b. Solving the latter for Xo = - - , Yo and substituting
Yo a -1 b -1
1 1
in XoYo = Xo + Yo, we obtain + or a+ b =3 as the relation which
(a-l)(b-1) a-1 Ii -1
must exist between a and b. Then the equation of the required surface is
a+b = u+v 3 or 2xy + z(x+y) 3xyz.
Find the general solution of each of the following equations.
12, p + q =z Ans. z =eyq>(x-y)

13, 3p +- 4q 2 3y - 4x =f (3z - 2x) or r/>(3y - 4x, 3z - 2x) 0
14, yq - xp cp(xy, xz) =0
2
15, xzp + yzq = xy y = xq>(xy-z )
2 2
16, X p + y q = z2 x - y = xyq>(l/x - 1/z)
2 2
17, YP - xq + X - y 0 q>(x2 + / , xy- z) 0
2 2
18, yzp - xzq = xy q>(x2 + / ' y +z ) = 0
19- zp + yq =X X + Z = y¢(x2-z2)
20, x(y-z)p + y(z -x)q = z (x - y) q>(xyz, X + y +z) = 0
2 2 2 2 2 2 2 2
21. x(y -/)p + y (z - X )q = z (x - y ) q>(xyz, X + y + z ) 0
22, Find the equation of all the surfaces whose tangent planes pass through the point (0,0,1).
Hint: solve xp+yq = z-1. Ans, z = l+xq>(y/x)
2
23, Find the equation of the surface satisfying 4yzp + q + 2y = 0 and passing through / +z = 1,
x+z =2. Ans, y 2 -tz 2 +x+z 3
CHAPTER 30
Non-linear Partial Differential Equations of Order One
COMPLETE AND SINGULAR SOLUTIONS. Let the non-linear partial differential equation
of order one
1) f(x,y,z,p,q) = 0
be derived from
2) g(x,y,z,a,b) = 0
by eliminating the arbitrary constants a and b. Then 2) is called a (or the)
complete solution of 1).
This complete solution represents a two-parameter family of surfaces which
may or may not have an envelope. To find the envelope (if one exists) we elim-
inate a and b from
og
g 0, -
ob
= o.
If the eliminant
3) A.(x,y,z) = 0
satisfies 1), it is called the singular solution of 1); if
t1.(x,y,z) = t(x y z),~(x,y,z)
and if g = O satisfies 1) while~= O does not, g = O is the singular solu-
tion. As in the case of ordinary differential equations (Chapter 10), the sin-
gular solution may be obtained from the partial differential equation by elim-
inating p and q from
f = 0, of= 0,
of o.
op oq
2
EXAMPLE 1. It is readily verified that z = ax+ by - (a 2 + 6 ) is a complete solution
of z = px + qy - (p 2 + q 2 ). Eliminating a and b from
g = z-ax-by+a +b
2 2
= 0, !! = -x+2a = 0, og
ob
= -y+2b = 0,
12 12 I 2 2 I 2 2
we have z = 2 x + 2 y - i:i(x + y ) = il(x + y ) • This satisfies the differential equation
and is the singular solution. The complete solution represents a two-parameter family of
planes which envelope the paraboloid x 2 + y 2 = 4z.
GENERAL SOLUTION. If. in the complete solution 2), one of the constants, say b, is
replaced by a known function of the other, say b = ¢(a), then
g(x,y,z,a,¢{a)) = 0
is a one-parameter family of the surfaces of 1). If this family has an envel-
ope, its equation may be found as usual by eliminating a from
g(x,y,z,a,¢(a)) = 0 and -0 g(x,y,z,a,¢(a)) 0

oa
and determining that part of the result which satisfies 1).
244
NON-LINEAR PARTIAL EQUA TIOJ\S OF ORDER ONE 2 l:i
EXAMPLE 2. Set b = ¢(a) = a in the complete solution of Example 1. The result of
eliminating a from g z - a(x + y) + 2a

2
= 0 and
og
::- = -(x + y) + 4a = 0
1
is z = -(x + y)
2
ca 8
which can be readily shown to satisfy the differential equation of Example 1. This is a pa-
rabolic cylinder with its elements parallel to the xOy plane,
The totality of solutions obtained by varying ¢(a) is called the genercil

solution of the differential equation. Thus, from Example 2, 8z = (x + y) 2 is
included in the general solution of the differential equation of Example 1.
When b = ¢(a), ¢ arbitrary, is used, the elimination of a between
g = 0 and -;,g = 0
oa
is not possible; hence, we are unable to express the general solution as a
single equation, involving an arbitrary function, as we were in the case of
the linear equation.
SOLUTIONS. Before considering a general method for obtaining a complete solution

of 1), we give special procedures for handling four types of equations.
2
TYPE I: f(p,q) = 0. Example: p2 - q = 1.
From Problem 3, Chapter 28, it follows that a complete solution is

4) z = ax + h ( a) y + c ,
where f(a,h(a)) = 0, and a and c are arbitrary constants.
The equations for determining the singular solution are
z = ax + h (a)' y + c, 0 = x + h ' (a) y, O = 1.
Thus, there is no singular solution.
The general solution is obtained by putting c == ¢(a), ¢ arbitrary, and elim-

inating a between
5) z = ax + h(a) y + ¢(a) and 0 = x + h'(a) y + ¢'(a),
The first equation of 5) for a stipulated function ¢(a) represents a one-param-
eter family of planes and its envelope (a part of the general solution) is a
developable surface. (See Problem 10, Chapter 28.)
2 2
EXAMPLE 3. solve p - q = 1,
and h(a)
A complete solution is z = ax + (a -
2 ½
1) y + c.
A neater form is obtained by putting u = sec a; then h(a) tan a and we have
z = x sec a + y tan a + c.
If we set c = ¢(a) = O, the result of eliminating a from
z = x sec a + y tan a, 0 = x tan a + y sec a or O x sin a + y

2 2 2
Z =X -y,
This developable surface (cone) is a part of the general solution of the given differential
equation.
2
Note that we might have taken h(a) = -(a - 1)½ and obtained as a complete solution
~ ½
z=ax- (a -1) :1+c,
NOl\-LINEAR PARTIAL EQl1ATIONS OF ORDER ONE
TYPE II: z = px + qy + f(p,q).
From Problem 4, Chapter 28, it follows that a complete solution is

G) z =ax+ by+ f(a,b),
This is known as tlie extended Clairaut type, for obvious reasons. This com-
plete solution consists of a two-parameter family of planes. The singular so-
lution (if one exists) is a surface having the complete solution as its tan-
gent planes.
1/3 1/3
EXAMPLE 4. solve z = px ' qy + 3p q •
A complete solution is z ~ ax + by + 3a1/3 61/3
1/3 -2/3
The derivatives with respect to a and bare X + a-2/3 61/3 -- O and y+ a 6 = 0•
1/3 1/3 -1/3 -1/3

Then ax + by = - 2a b , xy = a 6 ,
and, substituting in the complete ~olution, we obtain the singular solution

113 113
z = a 6 = 1/xy or xyz = 1.
TYPE III: f(z, p, q) o.

Assume z = F(x + ay) F(u), where a is the arbitrary constant. Then
oz dz ou dz
and q = -
dz ou
-
dz
= a - .
p = ox = du ox = du du oy du
When these are substituted in the given differential equation, we obtain
an ordinary differential equation of order one
f( dz a dz) = 0
z, du' du
whose solution is the required complete solution.
2 2
EXAMPLE 5. Solve z = p + q ,
Putz= F(x+ay) = F(u). Then p = dz/du, q adz/du, and the given equation may be
dz 2 2 dz 2
reduced to z = (-) + a (-) •
du du
Solving dz rz or
dz
= --
1
du,
1
we obtain 2vz = - - u + k = --(u+b),
1
du
/27 lz ~ ~ ~
Thus, a complete solution is 4(l+a 2 )z = (x+ay+b/, a family of parabolic cylinders.
Taking the derivatives with respect to a and b, we have
8az - 2(x+ay+b)y = O, x + ay + b = O.
The singular solution is z = O.
TYPE IV: Example: p -x

2
= q +y2 •
Set £ 1 (x, p) = a, £ 2 (y, q) = a, where a is an arbitrary constant, and solve to

obtain
and q = F 2 (y,a).
Since z is a function of x and y, dz p dx + qdy = Fi(x,a)dx + Fdy,a)dy.
Thus,
l\ON-LINEAR PARTIAL EQl:ATIOJ\S OF ORDER ONE 217
7) z = JF 1 ( x, a) dx + JF,;,_ ( y, a) dy + b,
containing two arbitrary constants, is the required complete solution.
2 2
EXAMPLE 6. solve p-q=x+y or
2 2 2
setting p-x =a, q+y =a, we obtain p=a+x2, q=a-y •
2
Integrating dz= pd.x + qdy = (a+x )d.x + (a-/)dy, the required complete solution is
z =ax+ x 3 /1 3 + ay - y 3;3
1
+ b, There is no singular solution.
TRANSF'ORMATIONS. As in the case of ordinary differential equations, it is possible

at times to find a transformation of the variables which will reduce a given
equation to one of the above four types.
The combination px, for example, suggests the transformation X=::ln x, since
then
oz oz dX 1 oz oz
p = = = and px =
ox oX dx X ox oX
Thus, q = px + oz oz oz 2
p2x2 becomes = + ( - ) , of Type I.
oy ox ox
Similarly, the combination qy suggests the transformation Y =ln y.
p q
The appearance of
2 , 2 in an equation suggests the transformation Z = ln z,
dz oZ q oZ
since then p = oz = oZ
= z - anci ez = oZ similarly, =
OX dZ ox ox OX z oy
q oZ
TliUS,
z =
cE/
z
becomes
oy
= c!!/, of Type I.
COMPLETE SOLUTION, CHARPIT'S METHOD. Consider the non-linear partial differential

equation
1) f(x,y,z,p,q) = O.
Since z is a function of x and y, it follows that
8) dz = p dx + q dy.
Let us assume p u(x,y,z,a), where a is an arbitrary constant, substitute in
1) and solve to outain q == v(x,y,z,a). For these values of p and q, 8) becomes
dz = u dx + v dy.
Now if 8 1 ) can be integrated, yielding
9) g(x,y,z,a,b) o,
this is a convlete solution of 1).
EXAMPLE 7. solve pq + qx = y.
Take p = a -x, substitute in pq + qx r y, and solve for q = y/a,
substituting in dz= p dx + qdy, we. have dz= (a-x)d.x + (y/a)dy, an integrable equa-
tion, with solution
z ~ ax - ½x
2 2 2
+ ½Y /a + k or 2az = 2a x - ax 2 + y 2 + b,
:\O~-LINEAR PARTIAL EQCATIONS OF ORDER ONE
Since the success of the above procedure depends upon our making a fortu-
nate choice for p, it cannot be suggested as a standard procedure. We turn
11rn, to a general method for solving 1). This consists in finding an equation
10) F(x,y,z,p,q) = 0
such that 1) and 10) may be solved for p=P(x,y,z) and q=Q(x,y,z), (that is,
such that
of of
op oq
11) 6 = -/; 0, identically),
oF oF
op oq
and such that for these value of p and q the total differential equation
8) dz = pdx + qdy = P(x,y,z)dx + Q(x,y,z)dy
is integrable, that is, P oQ _ 0 oP _ ~!:. + oQ = oq _ op = o.

oz oz oy ox ox oy
Differentiating 1) and 10) partially with respect to x and y, we find
12)
of + p -
of
+
of op + of oq = 0,
ox oz op ox oq ox
of of of op of oq o,
13) + q- + + =
oy oz op oy oq oy
oF oF oF op +
oF oq 0,
14) + p- + =
OX oz op OX oq ox
oF oF + oF op oF oq o.
15) + q- + =
oy oz op oy oq oy
oF oF of of
Multiplying q) by-, 13) by - , 14) by - - , 15) by - - and adding,
op oq op o-q
we obtain (noting that op= oq)
oy OX
of oF of oF of
( of + P of)oF + ( of + q of) oF (p - + q of) oF. = o.
ox oz op oy oz oq op ox oq oy op oq oz
This is a linear partial differential equation in F, considered as a function

of the independent variables x,y,z,p,q. The auxiliary system is
dp dq dx dy dz dF
16)
of of of of of of
= -.
0
+ p- + q - -(p of+ q of)
ox oz oy oz op oq op oq
Thus, we may take for 10) any solution of this system which involves p or q,
or both, which contains an arbitrary constant, and for which 11) holds.
NON-LINEAR PARTIAL EQ1:ATIONS OF ORDER Ol\E 249
2
EXAMPLE 8. solve q = - X[i t p
Here f = p2 - xp _ q so that
of ~
-p,
c!f
- 0, cf o, ')f 2p-x,
of
-1, and
ex oy oz op dq
of of of of ?Jf q ?Jf) 2
+ p- -p, -l- q __::_ 0, -(p - - 2p l- X/J + q•
ox oz oy clz clp oq
dp cfq d,: Jy dz
The auxiliary system ( 16) is
-p 0 -2p 2
+ X -2p +xp t q
dp dy -y
From = 1. we have ln p = -y + ln 11 or /J -- ae ,
-p
-axe - \'
2 2 -2:1
Using the given differential equat, ion, q -xp + P = + (1 e
-,' -~; 2 -2y
Then dz = p dx + qdy becomes dz = ae dx + (- axe +a e ) dy. Integrating,
- ') 1 2 -2_'1
z = axe - 2a e + b,
There is no singular solution.

See also Problem lS. •
SOLVED PROBLEMS
(In these solutions, the equations leading to the general solution will not be given.)
TYPE I: f(p,q) O.
2 2
1. Solve p + q 9.
2
A complete solution is z =ax+ by+ c, where a2 + 6 9.
The equations for determining the singular solution are
a
z = a x + ~ y + c, 0=x----y, 0 = 1. Thus, there is no singular solution.
/9 _ a2
2. Solve pq + p + q = 0-
a
A complete solution is z = ax+by+c, where ab+a+b 0, or z ax - - - y + c.
a+l
TYPE II: z = px + qy + f(p,q).
3. Solve z = px + qy + p 2 + pq + q 2 •
2
A complete solution is z ax+ by+ a 2 +ab+ b ,
Differentiating the complete solution with respect to a and b, we have

0 = x + 2a + b, o = y + a + 2b.
Solving to obtain a = (Y -2x)/3, b = (x - 2y)/3 and substituting in the complete solution,
the singular solution is 3z = xy - x 2 - y 2 •
2 2
4. Solve z = px + qy + p q
A complete solution is z =ax+ by+ a 2 b 2 • The equations obtained by differentiating with

2.'lo NOJ\-LINEAR PARTIAL EQl.iATIOl\S OF ORDER Ol\E
respect to a and bare 0 X + 2ab

2
and 0 = y -1-
2
2a b. Then
a= -R· -R b
2y
and the singular solution is z = -x~ y

f;J?l: +
X
--
y
= _~4 /4 x2/3 /13.
y 16
TYPE III: f(z,p,r;) =O.
dz dz
Assume z = F(x ray)= f(u), Then p = - , q a-, and the given equation becomes
du du
~ dz 2
9az (-) or 2du.
du
Integrating, /a(l+z 3 ) = u+b, and a complete solution is a(l + z 3 ) (x+ay+b/.

0
Using the results of differentiating this with respect to a and b,
1 + z3 = 2(x +ay+ b)y and O = 2(x+ay+b),
the singular solution is z 3 + 1 = O.
h. Solve p(l-l) = q(l-z).

dz dz
Assume z =F(x+ay) =F(u), Then p a - ' and the given equation becomes
du' q du
dz 2 dz 2] dz dz 2 dz 2
(-) [1-a (-) a-(1-z) or (-)[1-a+az-a (-)] o.
du du du du du
dz 2 dz 2 a dz
Then O and z = c; or 1-a+az-a (-) o, du and
du du
/1- a+ az
2
2/1 - a -r az = u + b = x + ay + b or 4(1-a+-az) = (x+ay+b).
Each of z = c and 4(1-a+az) = (x+ay+b/ is a solution; the latter is a complete so-
lution. Using it, the equations for obtaining the singular solution are
g = 4(1-a+az) - (x+ay+b)
2
= O, -
og = 4(-l+z) - 2y(x+ay+b) = 0,
og
-=-2(x+ay+b) o;
oa ob
there is no singular solution.
7. Solve 1 + p2 = qz.
dz dz
Assume z = F (x + ay) = F (u). Then p
du' q
a-, and the given equation becomes
du
dz)2 dz dz
(- az- + 1 0 or
du du
Rationalizing the left member of the latter equation, we obtain caz + /a 2 z 2 -4)dz 2 du.
2 1 az ~ I 2 2 ] = 2(u + b),
2 az + -a [2- va'z' - 4 - 2 ln(az + va z - 4)
1
whose solution is
A complete solution is then a 2 z 2 -+- az/4 2 z 2 -4 - 4 ln(az + 2 z 2 -4) 4a(x+ay+b). /4

Note that 2 2 2 2 2 2
a z - az /a z - 4 + 4 ln(az + /a z -4) = 4a(x+ ay+ b), obtained from
dz
= ½du, is also a complete solution,
NON-LINEAR PARTIAL EQL'\ TIO NS OF ORDER ONE 2Sl
TYPE IV: £1 (x, p) = £ 2 (y, q),
8. Solve vp - ✓q + 3x = 0 or Ip+ 3x = /q.

Set v'p ,- 3x = a and lq = a. Then p = (a - 3x/ and q a
2
A complete solution is
1 3
z = f p dx + f q dy + b = f (a - 3x / dx + a f dy + b
2
or z = --(a-3x)
9
+ay+b,
2
Then, is no singular solution,
9. Solve q = -px + p •
2
Set p 2 -px = a and q = a. Then p = ½ex+ /x 2 + 4a ).
A complete solution is z ½J (x + /2 2
+ 4a )dx + aJ dy + b
or z ~(x 2 + x/2 +4a) + aln(x

2
+ /2 2
+ 4a) + ay + b,
Another complete solution is obtained by the method of Charpit in Example 8,
There is no singular solution,
USE OF TRANSFORMATIONS.
,r: n 2l
10. Solve pq = X y Z or 1,
The transformation
1-l "l! +l n+l
z Y
__ , oZ oZ dx -l 1 oZ oZ dy -i 1
z = ){ -
X
y -
z p-,
m
z q-
n
1- Z m+1 n+l aX ox dX X
oY oy dY y
oZ oZ
reduces the given differential equation to 1.
oX oY
This equation is of Type I and its solution is z ax+ ~y +

a
C,
1-l m+l n+l

z X y
A complete solution of the given equation is a-- + + C,
1- Z m+ 1 a(n + 1)
2 2 2 2
11. SolvP X p + y q z.
1) The transformation
l -l I
oZ oZ dx oZ -;;
X = ln x, Y= ln y, Z = 2z 2 , pxz 2
qyz
oX ox dX oY
rPducps the given equation to Z(-)

oz 2 + Z(-)
oz 2 = Z or of Type I.
oX oY
2 2
A complete solution is Z ~ aX + bY + c or 4z (a ln x + b ln y + c/, where a + b = 1,
The singular solution is z = O.
oz oz dX 1 oz 1 oz
2) The tn,nsformation X = ln x, y = ln y, p = -- - -, q
ox ox dx X OX y oY
reduces the given differential equation to oz 2 CZ 2

(-) + (-) = z' of Type III,
oX oY
l\Ol\-Ll'.\EAR PARTIAL E<)I \TlP\" nl· Ol,i>LH O\E
cl z dz
We set • ~ F(X + aY) = F(u), Then fl-, and
r!Y du
dz 2 2 dz 2
(-) +a(-) z _:__ 1/U.
du du
Integrating, u + b ,\ + a Y + /; 1a t , a Jn y ,- b,
A complet,, solution is 4(1 • ,/)z ( ln x :tJ.i ...:1ngular soJution is
2 2
12. Solve 4xyz = 1,q 2px y 2qxy
Let X ~
,\ 2' y } ;- . Then p - oz ;,..x
?17
(),\
,IX
,h
,, ' -jL ?,z ,!Y
;w dy
Substituting in the given equation, we h:~e of Type II.
A complete solution is or
Eliminating a and b from this and O - x i- b, O ~ y 2 + 2

ll, o~tained by differentiating it
with respect to a and b, the sini;ular solution is ! ,,u,,rl l' L/1' x 2 y? - Q ..
2 2
13. Solve px ~z(z-qy),
-1Z d\ 1 ·,)z 1
- -
dZ
The transformation ln y, ln x. -- ·- ' q
X oX y ;3 y
clz 2
reduces the given equation to A) (--) C l(Z - ,)f 't;;;e Ill.
oX
lie set F(u). Then oz dz ~:lz
c-ir1cl A'; becomes z2_azdz.
oX Ju oY du
dz dz
Then
du
= ½z(/a 2 + 4 - a)• 2
z
cff+ 1 - a)J11, and ln z
2
(~ a) (u + b).
A complete solution is ln 2 2 ~ cff-~ - a) (ln x + a ln y + /1).

2 2 p 2 q 2
14. Solve fJ + q or (-) (-'-) = X + ),
z z
The transformation z ln z, p q C reduces the given equation to

c1x
X + y or (-)
oZ 2
of Type IV.
c1y
Set
c1Z 2 c1Z 2 ;3z I
;3z *
(-) -x a y-(-). Then (a +x)" and (y-a)'.
dX ;3y dX ?ly
I
A complete solution is z Jca+xl"cix + fcy-a)~dy + b
2 3112 2 3/2
or ln z -(a+ x) + -(y- a) + b.
3 3
l\ON-Lll\EAR PARTIAL EQliATIONS OF ORDER ONE
CHARPIT'S METHOD.
15. Solve l6p2z2 + 9q2z2 + 4z 2 - 4 = o.

2 + 9q2z2 + 4z 2
Let f(x,y,z,p,q) = 16/ z - 4.
Then cf = 0 =
CX
cJf.
cy
cf
CZ
2
32p z + 18q 2z + Sz,
cf
cp
= 32pz
2
. cf
cq
= 18qz
2
. and the aux-
dp dq dx dy dz
iliary system is
cf + P cf cf + q cJf _ cf cf cf + q cfi
-(p
cx CZ cy oz cp cq op cq
dp dq dx __!!L_ dz
2 2 2 2 2
2
32p 3 z + 18pq z + 8pz 32/qz + 18q 3 z + 8qz -32pz 2 -18qz -32p z - 1Sq z
Using the multipliers 4z, 0, 1, 0, 4p, we find

2 2 2 2 2 2
4z(32p 3 z + 18pq z + Spz) + l(-32pz ) + 4p(-32p z - 1Sq z ) 0
and so dx + 4p dz + 4z dp = 0.
X -a
Then x + 4pz = a and p = Substituting for p in the given differential equa-
4z
2 2 2 2
tion, we find (x -a) + 9q z + 4z - 4 = o. Using the root
_x-adx 3[z dz+ ;t(x-a)dx]

dz = p dx + qdy = + - 2 / 1-z 2 -i:i:(x-a)
I 2 dy or dy
4z 3z 2/1-z 2 -./r(x-a)
2
2 2
Then y- b = - 3/
- 1 - z 2 - -(x-a)
I 2 or
(x -a) +(y-b) +z2=l is a complete
2 ~
4 9/4
solution. This is a family of ellipsoids with centers on the xOy plane. The semi-axes of the
ellipsoids are 2 units parallel to the x-axis, 3/2 units parallel to the y-axis, and 1 unit
parallel to the z-axis. The singular solution consists of the parallel planes z = ±1.
Another complete solution may be found by noting that the equation is of Type III. Using
dz dz
F(x + ay) = F(u) and setting p = du and q = a du, the given equation ,becomes
2 dz 2 2 2 dz 2 2 z dz 2
16z (-) + 9a z (-) + 4z - 4 = O or = - - - - du, Then
du du
/1- z 2
/16 + 9a 2
2 2
--;:::==-::-(U + b) ,====-(x + ay + b),
/15 + 9a 2 /15+ 9a 2
This complete solution ( 16 + 9a 2 ) (1 - z 2 ) 4 (x + ay + represents a family of elliptic cyl- b/

inders with elements parallel to the xOy plane. The major axis of a cross section lies in
the xOy plane and tpe minor axis is 2 units parallel to the z-axis.
2'.i--1 NON-LINEAR PARTIAL EQLATIONS OF ORDER Ol\E
Find a complete solution and the singular solution (if any),

2
16, p = q
2 2
17. p +p q
18. pq =p + q ( b - 1) z = bx + b ( b - 1) y -,- c
19. z = px + qy + pq z =ax+ by-,- ab; s,s., z = -xy
2 2 2
20. p +q 4z z(l+a ) = (x+ay+b/; s.s., z = 0
2
21. pz =1 + q z2 - z/i: 2 -4a 2 + 4a 2 ln(z + /z 2 -4a 2 ) = 4(x+ay+b)
22. z
2
c/ + q 2
+ 1) (x + ay + b / ; s. s. , z
2
- 1 = O
2 2
23. p + pq = 4z (1 +a)z (x + ay + b) ; s. s. , z = 0
2 2
24, p - X = q - y 3 (Z - b)
2 2 2
25, yp - X q = X y
26, o -/)xq 2 2
+ y p = 0 (2z-ax+b)
2 2
=4a(y-1)
2
2 2 2
27. x~p -yzq-z 0 xlnz=a+(a -l)xlny+bx
Hint: Use X = 1/x, Y = ln y, Z = ln z.
2 2 2
28. x
~
p + y zq - 2z = 0 xy ln z ay + ca 2 - 2 )x + bxy
Hint: UseX=l/x, Y=l/y, Z=lnz.
~ 2 2
29. X p + y q = Q 0
30. 2p/ - /z 0 z
2 2
=ax+ ay
2
+ b
2
31. q = xp + p
2 2
32, yz = 2(axy +a/+ a + by)
dz a a
Hint: pz = a and q =
2 2 (1 - -) •
-p z l
;n. pq + 2x(y+l)p + y(y+2)q - 2(y+l)Z = 0
Ans. z =ax+ b(/+2y+a); s,s., z + x(/+2y) 0
CHAPTER :n
Homogeneous Partial Differential Equations of Higher Order
with Constant Coefficients
AN EQUATION SUCH AS
1)
o2 z t 5xy--
o2 z + x3 oz oz
+ x- + yz
ox 2 oxoy ox oy
which is linear in the dependent variable z und its partial derivatives is
called a 1 inear partial differential e'7uat ion. The ordi,r of 1) is three, be-
ing the order of the highest ordered derivative.
A linear partial diffetential equation such as
2) x
2 o3z + x yo-z- + 2--z-
3
o 3
+ o3 z
ox 3 ox 2 oY oxoy 2 oy 3
in which the derivatives involved are all of the same order, will IJe called
homogeneous, although there is no agreement among authors in the lh'>e of this
term.
HOMOGENEOUS LINEAR DIFFERENTIAL EQUATIONS WITH CONSTANT CCEFFICIENTS. Consider
3) A
oz + B -
oz = 0,
OX oy
2 2
4) A
o2 z + B ~ + C ~ = 0,
OX 2 oxoy oy 2
2
5) + B ~ +. = X t 2y,
ox 2 oxoy
in which the nwnbers A, B, C are real constants.
It will be seen as we proceed that the methcxis for solving equations 3)-5)
parallel those used in solving the linear ordinary differential equati.on
d
f(D)y = Q(x,) where D = -·
dx
0
We shall employ two operators, Dx = and Dy = :, so that equations
OX 0
3)-5) may be written as
255
256 HOMOGENEOUS PARTIAL EQt;ATIONS. CONSTANT COEFFICIENTS
Equation 3 1 ) is of order one and the general solution (Chapter 29) is

B
z = ¢(y - Ax), ¢ arbitrary.
Suppose now that z = ¢(y + mx) = ¢(u), ¢ arbitrary, is a solution of 4 1);

then substituting
D z = oz de/>
= m-,
de/> OU
X OX du du oy du
in 4 1 ) we obtain
dz,~ ( Am 2 + Rm + C) = 0.
du
Since d is arbitrary, d 2 .f/ du 2 is not zero identically; hence, m is one of the

roots m=m1 ,m 2 of Am 2 +Bm+C=O. If m1 -/=m 2 , z=¢ 1 (y+m 1 x) and z=¢2 (y+m 2 x)
are distinct solution of 4 1 ). Clearly,
z = ¢ 1 (y +m 1 x) + ¢ 2 (y +m 2 x)
is also a solution; it contains two arbitrary functions and is the general
solution.
More generally, if
6) 0
7)
is the general solution of f(Dx,Dy)z = O.
2 2
EXAMPLE 1. Solve <Dx - DxDy - 6Dy)z = <Dx + 2Dy) (Dx - 3Dy)Z = 0,
and the general solution is y = q>i(y - 2x) + ¢ 2 (y + 3x).
k
(D - m 1 D )
X Y
(D - mk
X +l
D ) .... , (D - m D ) z
y X n y o,
the part of the general solution given by the corresponding k equal factors is
2
¢i(y+m 1 x) + x¢ 2 (y+m 1 x) + x ¢ 3 (y+m 1 x) + ..... + xk-i¢k(y+m 1 x),
and the general solution of 6 1 ) is
z = ¢ 1 (y + m1 x) + x¢ 2 (y + m1 x) + ••••• + xk-l ¢k (y + m1 x) + ¢k+l (y + mk+l x)
+ "" • + ¢n(Y +mnx),
where ¢ 1 ,¢ 2 , •····¢n are arbitrary functions.

3 2 2 3 2
E XAMPLE 2 • Solve <Dx - D":J)y - BDxf)y + 12Dy)z (Dx - 2Dy) (Dx + 3Dy)z = o.
Here, m1 = m2 = 2, m~ = -3 and the general solution is z = ¢ 1 (y + 2x) + xc/>2 (y + 2x) + ¢ 3 (y - 3x).

IIOMOGENEOCS PARTIAL EQUATIONS. CO::\'STAJ\"T COEFFICIENTS 2;;7
If one of the nwnbers, say m1 , of 6) is imaginary then another, say m2 , is

the conjugate of m1 , Let m1 =a+bi and m2 =a-bi so that 6) becomes
The part of the general solution given by the first two factors is
(¢1 ,¢ 2 arbitrary, real functions), and the general solution of 6 11 ) is
z = ¢ 1 (y+ax+ibx) + ¢ 1 (y+nx-ibx) + i[¢ 2 (y+ax+ibx)-¢ 2 (y+ax-ibx)]

+ ¢ 3 ( y + m3 x) t , •• •• •• • + ¢n ( y + mn x) .
Here, m1 = m2 =1, m3 = - ½(1 + ilTT), m4 ~ - -!,(1- i/TT), and the general solution is
+ i[¢4 {y - ½Cl+ l /TT)x} -C/J {y - ½Cl-

4 i ✓TT)x}].
See also Problem 5.
The general solution of
consists of the general solution ot the reduced equation
= 0
plus any particular integral of 5 1 ). We shall speak of the general solution

of 4 1 ) as the complementary function of 5 1 ).
In setting UIJ proceclures for obtaining a particular integral of
8)
1
we define the opPrator lly the irlent ity
1
f(D , D,) - - - F(x, y) - F(x, y).
' ' f(D,, D)
The particular integral, denoted hy
z =
may be found, as in Chapter 13, !Jy solving n equations of the first order
9) U1 =
1 F(x, y), U2
l
U1' .... ' z = Un =
1
Un-1
Dx - mnDy D"' -m""_ 1 Dy Dx - m1 Dy
2::,g IIO.'\IOGEi\EOUS PARTIAL EQt;ATIONS, CONSTANT COEFFICIENTS
Note that each of the equations of 9) is of the form

10) p - mq = g(x,y)
and that we need only a solution, the simpler the better. In Problem 6 below,
the following rule is established for obtaining one such solution of 10):
Evaluate z = Jg(x,a-mx)dx, omitting the arbitrary constant of integration,
and then replace a by y + mx.
EXAMPLE 4. Solve
From Example 1, the complementary function is z = ¢1 (y - 2x) + ¢2 (y + 3x).
1 1
To obtain the particular integral denoted by z
Dx + 2Dy [Dx 3Dy (x + y)]:
1
a) Set u - - - ( x + y) and obtain a particular integral of (Dx - 3Dy )U = x + y.
DX - 3Dy
2
Using the procedure of Problem 6, we have u = f(x+a-3x)dx ax-x and, replacing a
2
by y + 3x, u = xy + 2x •
1 2
b) Set z = ---u (xy + 2x ) and obtain a particular integral of
DX +- 2Dy Dx + 2Dy
2
(DX+ 2Dy)Z xy + 2x •
1 2 4 3
Then z = J[x (a+ 2x) + 2/ ]dx -ax +- -x and, replacing a by y - 2x,
2 3
1 2 1 3
Thus the general solution is z = ¢ 1 (y-2x) + q> 2 (y+3x) f x Y + x
2 3
The method of undetermined coefficients may be used if F(x,y) involves

sin(ax +by) or cos(ax +by).
EXAMPLE 5. Solve
en: + 5D.,Dy + 5D~)z = [DX - ½(-5 + v'5)Dy] [DX - ½(-5 - v5)Dy]z x sin(3x -- 2y).
The complementary function is z = ¢i[y + ½(-5 + ✓5)x] + ¢ 2 [y + ½°(-5 - v'5)x].

Take as a particular integral
z = Ax sin(3x-2y) + Bx cos(3x-2y) + C sin(3x-2y) ,- D cos(3x-2y). Then
(6A - 9D)cos(3x - 2y) - (6B + 9C)sin (3x - 2y) - 9Ax sin (3x - 2y) - 9B x cos (3x - 2y),
(-2A+6D)cos(3x-2y) + (2B+6C)sin(3x-2y) + 6Ax sin(3x-2y) + 6Bx cos(3x-2y),
- 4D cos (3x - 2y) - 4C sin (3x - 2y) - 4Ax sin(3x - 2y) - 4Bx cos (3x - 2y),
2 2
and <Dx + 5DxDy + 5Dy)z = Ax sin(3x-2y) + Bx cos(3x-2y) + (C+ 4B) sin/3x-2y)
+ (D - 4A) cos(3x- 2y) x sin(3x -2y).
Then A = 1, B = C = O, D = 4 and the particular integral is

z x sin(3x -2y) ,- 4 cos(3x -2y). The general solution is
z = ¢i[y + ½C-5 + v5)x] + ¢2(y + ½(-5 - v'5)x] + x sin(3x-2y) + 4 cos(3x-2y).
See also Problem 10.
HOMOGEJ\EOUS PARTIAL EQUATIONS. CONSTANT COEFFICJEj';TS 2.=;<J
Short methods for obtaining particular integrals, analogous to those ot

Chapter 16, may be used.
l ux +: y
a) ---e provided f(a, b) f O.
f(c1, b)
then
r
1 1 e Ut + i..~J
u/
=
1 1 e ux +by 1 -e
X ux + by
g (A' b)
a ~
g(a, b) r!
(D.< - b
~D {
y
tJ, (D", Dy) (Dy - bDY)
b)
1 sin (ax + by) 1 sin(ax + by) and
=
f(D:,DxDy,D-~) f(-a
2
, -Ab, -b 2 )
1 1 cos (ax+ by) ,

cos (ax+ by) =
2 2 2 2
f(Dx, DxDy, Dy) f(-a , -ab, -b )
provided f(-a 2 ,-Ab,-b2 ) f 0.
zx+3y x+y .
EXAMPLE 6. Solve e +e ~srn(x-2y).
The complementary function is z = ¢iCY +x) + ¢ 2 (y +2x),
Now 1 e2x+3y _ _ _1_ _ _ e2x+3y = !/x+3Y

is one term of the
2 2 2 2 4
DX -3DX Dy + 2Dy 2 -3·2·3+2•3
particular integral, Since ¢1 (y +x) includes ex+y, we write
1 x+y 1 ,:+y x+y

------e - --e = -xe
1\-Dy
1 1 .
Also, - - - - - - - - sin(x-2y) - - srn(x-2y),
-1-3(2) +2(-1)(-2/ 15
Thus, the general solution is z
c) If F(x,y) is a polynomial, that is, F(x,y) = 2.p .. xiyj, where i , j are posi-
t;
tive integers or zero and Pij are constants, the procedure illustrated_ below
may be used.
EXAMPLE 7. X + Y• (Example 4.)

For a particular integral, write
1 1 1
-----(x+y) .. ,](x+y)} = -(x+y+-)
D
2
D D
2
DX2 DX
X 1-2-5'.::.:z'.
D D2
X X
1 1
-(x+y+x) -(2x + Y) Note that Dy(x+y) =1 and__!:__ =fdx.
D2 D2 Dx
X X

~60 IIO\IO(;Ei\EOl;S PAHTL\L E(Jl \TIO~S. CONSTANT COEFFICIENTS
~OIXED PROBLEMS
1. Solve o.
llere m1 = 1, m2 = -1, m3 = -2 and the general solution is
1+ /2, m3 ~ 1- ✓2 and the general solution is

z = ¢iCy~3x) • ¢ 2[y+(l+v'2)x] + ¢ 3 [y+(l-v'2)x].
3. SOlVP o.
Since m1 = 1, m2 = m3 = -2, the general solution is
z. ~ 'Pi(y+x) + ¢ 2 (y-2x) + x¢ 3 (y-2x), Another form of the
general ~olution is Z = ¢1(Y+X) t ¢2Cy-2x) + ycp3(y-2x).
4. solve
Here m1 = m2 = 1, m3 m4 = -1 and the general solution is

z ¢ 1 (y+x) + x¢ 2 (y+x) + ¢ 3 (y-x) + x¢4 (y-x),
5. Solve o.
since m1 = 1 + 2i, m2 = 1 - 2i, the general solution is
z = ¢iCy+x+2ix) + ¢iCy+x-2ix) + i[¢ 2 (y+x+2u) - ¢ 2 (y+x-2ix)].
where ¢ 1 , ¢ 2 are real functions,
u
If we take ¢1(u) cos u and ¢2(u) = e then since
'
i.bx 1 ibx -ibx
e cos bx + l sin bx, sin bx -(e e )•
2i
-i.bx 1 ibx -ibx
e cos bx - i sin bx, cos bx 2(e + e ) •
¢ 1 (y+x+2ix) cos(y+x) cos(2ix) sin(y +x) sin(2ix)

cos(y+x) cash 2x i sin(y +x) sinh 2x,
¢ 1 (y +x-2u) cos(y+x) cos(2ix) + sin(y+x) sin(2ix)

cos(y+x) cash 2x + i sin(y +x) sinh 2x,
y+x+2i.x y+x-2ix y+x 2ix -2i.x
e e = e (e - e ) 2iey+x sin 2x,
Thus, we obtain as a particular integral

z [cos(y+x) cash 2x - i sin(y+x) sinh 2x] + [cos(y+x)cosh2x + isin(y+x)sinh2x]
y+-r
+ i (2ie sin 2x) 2 COS(y+x) cash 2x - 2ey+x sin 2x.
Note that z is a real function of x and y.

HOMOGENEOUS PARTIAL EQUATIONS, CONSTANT COEFFICIENTS 261
6. Show that a particular integral of p-mq = g(x,y) may be found by integrating dz= g(x,a-mx)dx,
omitting the arbitrary constant of integration, and then replacing a by y + mx,
dz
The auxiliary system is dx = dy Integrating the equation formed with the
1 -m g(x,y)
first two terms, we have y + mx = a. Using this relation, the equation
dx dz dx dz
becomes
1 g(x,y) 1 g(x,a-mx)
Then z = fg(x,a-mx)dx and, in order that no arbitrary constants be involved, we replace

a by y + mx in the solution.
7. Using the procedure of Problem 6, find particular inte~rals of

3X
a) p+3q = cos(2x+y), b) p-2q = (y+l)e .
a) Here m = -3 and g(x,y) = cos(2x +y).

Then z = Jg(x,a-mx)dx = Jcos(2x +a+ 3x)dx = ! sin(5x + a) and, replacing a by y-3x, the
5
required particular integral is z = ! sin (2x + y).
5
3 1 3x 2 3x 2 3x
b) z = fg(x,a-mx)dx = fca-2x+l)e xdx = -(a+l)e --xe +-e
3 3 9
1 3X 2 3X 2 3x
Replacing a by y + 2x, we have z -(y+2x+l)e - - Xe + -e
3 3 9
8. Solve
The complementary function is z = ¢ 1 (y+2x) + ¢ 2 (y-4x).
1
To obtain the particular integral denoted by -------- ✓2x + 3y, we obtain from
(DX - 2Dy) (DX+ 4Dy)
cDx + 4Dy )u = v2x + 3y the solution u J [2x + 3(a- mx)] 112 dx = J [2x + 3(a + 4x)] 112 dx
J(14x+3a)1/2 dx =
1
-(14x+3a)
3/2 ~ ( 2x + 3y)3/2
21 21
1 3/2
and from cDx-2Dy)z = u = -(2x + 3y) the solution
21
1 5/2 1 5/2
z = - -(3a- 4x) = - -(2x + 3y) .
210 210
The general solution is z
2X+y
9. Solve e
The complementary function is z = ¢iCy+2x) + x¢ 2 (y+2x) + q> 3 (y-3x).
To obtain the particular integral denoted by
J 2x+(a+3x)dx
e
_
-
Je 5x+a dx _
-
1 5x+a
5e
262 HOMOGENEOUS PARTIAL EQUATIOJ\'S, CONSTANT COEFFICIENTS
2
~e x+y the solution v ~J /X+(a-2X) dx 1
-xe
a 1
xe
2X+ y
;
5 5 5 5
1 2 2X+y
and from Wx-2Dy)z = v = ix/x+y the solution z -x e .
10
1 2 2x +Y
The general solution is z = + -x e .
10
10. Solve ex cos 2y.
The complementary function is
Take as a particular integral z = Aex cos 2y + Bex sin 2y, Then
Aex cos 2y + Bex sin 2y, -4Aex cos 2y - 4Bex sin 2y,
-2Aex sin 2y + 2Bex cos 2y, BAex sin 2y - BBex cos 2y.
Substituting in the given differential equation, we have

(5A + lOB)ex cos 2y + (5B - lQA)ex sin 2y = ex cos 2y, so that A = 1/25 and B = 2/25.
1 X 2 X •
The particular integral is z = - e cos 2y + - e srn 2y, and the general solution
25 25
is
2 2X 3
11. Solve (Dx - 2DxDy)z = Dx (Dx - 2Dy )z e + X Y•
The complementary function is z ¢1(Y) + ¢2(y+2x).

2x 1 3
A particular integral is given by e + X Y• The first term yields
2 2
Dx -2DxDy Dx-2DxDy
1 2X 1 2x
------e -e Writing the second term
(2>2 - 2(2) (0) 4
1 3 2 3 1 3 1 ~
-(xy+-x) -(x y + - x ),
D2 Dx D2 2
X X
we obtain The general solution is z
. 3x +Y
12. Solve srn(x+2y)+e.
The complementary function is z A particular in-
tegra] is given by - - - - -- - - - --
2 2
sin (x + 2y) + -----------e 3x+y .
Wx + Dy) Wx - DxDy - 6Dy)
(Note. The separation in the first term is one of convenience, i.e., we could have written
1
sin(x+2y). The separation in the second term is necessary, how-
ever, since e 3x+y is part of the term q>3 (y + 3x) of the complementary function.)
HOMOGENEOUS PARTIAL EQUATIONS, CONSTANT COEFFICIENTS 263
1 1 1
For the first term: - - - - -- - - - - sin(x + 2y) - - - - sin(x + 2y)
2 2
<Dx +Dy) <Dx-DxDy-6Dyl Dx + Dy -1 + 2 + 24
2_ Dx- Dy sin(x + 2y) 1 1

- - (flx-D,) sin(x +2y) = - cos (x + 2y).
25 D~-D~ 25 (3) y 75
e3x+y
_ _ _ _ _ _1_ _ _ _ _ e3x+y
For the second term:
2 2
<Dx-3Dy) cDx +3DxDy +2Dy)
e3X +y 1 3X+ y
-Xe •
20
,.1, ,.1, 1 1 3X + y
The general solution is z ¢ 1 (y-x) + '-!-'2 (y-2x) + '-!-' 3 (y+3x)- -cos(x+2y) + -xe •
75 20
13. Solve cos(x-y) + x 2

+ x/ + y3
The reduced equation is that of Problem 12. A particular integral is given by
1
- - - - -- - - - - cos (x -y) +
2 2
(Dx + Dy) (Dx- DxDy- 6Dy)
(Note that cos(x-y) is part of the complementary function; hence, the corresponding fac-
tor (Dx+Dy) must be treated separately.)
1 1 1
For the first term: - - - - -- - - - - cos (x - y) = - --- cos(x -y). We must
2 2
<Dx + Dy) (Dx-D~y- 6Dy) 4 DX+ Dy
1
4cos(x-y),
obtaining u = ~Jcos[x-(a+x)]dx
4
~J
4
COS{-a) dx
1 1
-XCOS{-a) -x cos(x-y).
4 4
1 2 2 3 1 2 2 3
For the second term: - - - - - - ( x + xy + y ) 2 3 (x + xy + y )
3 2 3
n - 6Dy
DX -7Dx"y 3 Dy Dy
D(l-7--6-)
X 2 D3
DX X
2 3
1 Dy Dy 2 2 3 1 2 2 7 6
-(1 + 7 - + 6-)(x +xy +Y) - [x + xy + y3 + -(2x + 6y) + -(6))
D3 D2 D3 D3 D2 D3
X X X X X X
1 2 2 3 7 36 5 6 1 5 1 ~ 2 1 3 3
-(x + xy + y ) + -(2x + 6y) + - -x + - x (1 + 2ly) + 24 X y + 6 X y •
D3 D5 D6 72 60
X X X
The general solution is
1 ~ 2
+ 24 X y
264 HOMOGENEOUS PARTIAL EQUATIONS, CONSTANT COEFFICIENTS
Solve each of the following equations.
14. (D~ - BDxDy + 15D;)z = o.

2 2
15, (De - 2DxDy - Dy )Z = o.
2 2
16. (Dx - 4DxDy + 4Dy)z cc o.
2 2 3
17. (D3 + 2D"D,
X 1
- Dpy - 2Dy)z o.
18. (D~D; + n:n;)z ~ o.

2 2 x-y
19. (Dx + 5DxDy + 6Dy )z = e • Ans.
20.
21. Ans,
22.
3 2 2 3
23. cDx - 3DxDy - 4DxDy + 12Dy)z = sin(y +2x).
Ans. z = ¢iCy-2x) + ¢ 2 (y+2x) + q>3 (y+3x) + !xsin(y+2x)

4
24, Ans.
25,
26.
3 2 2X 2
27. (D
X
- 2D D )z
X y
= 2e + 3x y.
Ans.
28.
~on~hmnogenou~ Linear Equations with Constant Coefficients
A NON-HOMOGENEOIB LINEAR partial differential equation with constant coefficients

such as
2 2 2
f(D,,, DY) z = (D,. - Dy -r 3D'< + Dy + 2) z = (Dx + Dy + 1) (Dx - DY + 2) z = x + xy
is called reducible, since the left member can be resolved into factors each
of which is of the first degree in Dx,Dy, while
£ ( Dx , D) ) z = (Dy DY + 2 D; ) z = DY ( Dx + 2D}) z = cos ( x - 2y) ,
which cannot be so resul ved, is called irreducible.
REDOCIBLE NON-HOMOGENEOIB EQUATIONS. Consider the reducible non-homogeneous equa-

tion
where the ai,bi,ci are constants. Any solution of
2) (aiDx + biDy + ci)z = 0
is a solution of 1). From Problem 5, Chapter 29, the general solution of 2) is
3) z = e -cix/ai ¢(aiy-bix), a•t I, 0'

or
3I ) z -- e - cc,/bi y1(aiy- bix), bi f 0,
with ¢ and 'fl arbitrary functions of their argument. Thus, if no two factors
of 1) are linearly dependPnt (that is, if no factor is a mere multiple of another),
the general solution of 1) consists of the sum of n arbitrary functions of the
types 3) and 3 1 ).
EXAMPLE 1. Solve (2½ + Dy + 1) (Dx - 3Dy + 2)z = 0.
- ' ,.1, -2x ,.1,

The general solution is z "' e -''+"i(2y-x) + e '+" 2 (y+3x). Note that the ~irst term
. -~2 ~~
on the right may be replaced by e t/; 1 (2y - x) and the second by e t/;2 (Y + 3x).
EXAMPLE 2. Solve (2Dx+3Dy-5)(Dx+2Dy)(Dx-2)(Dy+2)z = O.
. 5'</2 , ,.1, 2x -2y

The general solution 1s z = e ct'i(2y-3x) + '+"2 (y-2x) + e q>3 (y) + e q>4 (x),
If
k
4) f(Dx,Dy)z = (a 1 Dx+b 1 Dy+c 1 ) (ak+IDx+b,H 1 Dy+ck+I)"•(anDx+bnDy+cn)z= 0,
where no two of the n factors are linearly dependent except as indicated, the
266 NON-HOMOGENEOUS LINEAR, CONSTANT COEFFICIENTS
part of the general solution corresponding to the k repeated factors is

e-c1x/a1[¢1(a1Y-b1x) + x¢2(a1Y-b1x) + ••• + xk-1¢k(a1Y-b1x)].
EXAMPLE 3. Solve
The general solution is z 5

= e- Ycpi(2y-x) + e--;,;[¢2 (y +2x) + xcp3 (y+2x)].
See also Problem 3.
THE GENERAL SOLUTION OF
5) f(D,: ,Dy)z = (a 1 Dx + b 1 Dy + c 1 ) (a 2 Dx + b 2 Dy + c 2 )• • • (anDx + bnDy + Cn)Z = F(x,y)
is the sum of the general solution of 1), (now called the complementary func-
tion of 5)), and a particular integral of 5),
6) z
The general procedure for evaluating 6) as well as short methcxls applica-

ble to particular forms of F(x, y) are those of the previous chapter.
X -y
(Dx -2Dy)<Dx+Dy+2)z ye + 3xe •
The complementary function is z = ¢dy +2x) -t e-2Xcp2 (y-x).
To evaluate
dx dy du
whose auxiliary system is We obtain y x +a readily and the equa-
yex - 2u
du
tion dx or du + 2u = yex = (x + a)ex. This linear equation has /x as integrating
dx
2 1 3x 1 3x 1 3x 1 3X 1 3X 1 3x
factor; hence, ue x = f(x+a)e 3xdx -xe - -e + -ae -xe - -e + -(y -x)e
3 9 3 3 9 3
1 X
We then solve <Dx - 2Dy )z = u -e obtaining the required particular in-
9
tegral (see Problem 6, Chapter 31)
z r [~(a-2x)ex
3
- ~ex]dx
9
1
- ae
3
X 2
-xe
3
X 2 X
+ -e
3
1 X
-e
9
1 X 2 X 5 X 1
-(y + 2x)e -xe + -e 3(y + ~)ex.
3 3 9 3
To evaluate (3xe -y), we solve (Dx +Dy+ 2)u = 3xe -y whose auxil-
(DX - 2Dy )(DX +Dy+ 2)
du du
iary system is Then y = x + a, and from or
3xe -y - 2u 3xe -y - 2u
NON-HOMOGENEOUS LINEAR, CONSTANT COEFFICIENTS 267
du
- + 2u = 3xe
-y
= 3(y-a)e
-y
, ue 2y = 3 J (y-a)e y dy =
' y
3(y-l-a)e y = 3(x-11e and
dy
-y 3 (x 1 )e -y, the required particular
U = 3 (X - 1 )e , Solving in turn (D,;-2Dy)z = u -
-a+2X 3 -a+2x 3 3 -y
integral is z = 3 J (x - 1 )e dx -(xe -(X - -)e ,
2 2 2
2 2
EXAMPLE 5. Solve f(DX'Dy )z (Dx - DxDy - 2Dy + 6Dx: - 9Dy + 5)z
2
(Dx; +Dy+ 5)(Dx - 2Dy + l)z = e x+y + ex+y
The complementary function is z = e-V;xq> 1 (y -x) + e-xq>2 (y + 2x),

For the particular integral corresponding to the first term of F(x,y), we use
1 ax:+by 1 ax+by
---e ---e f(a,b) f 0,
f <Dx,Dy) f(a,b)
_______1 _ _ _ _ _ _ e2x+y 1 e2x+y 1 2x +y

and obtain -e
4 - 2 - 2 + 12 - 9 + 5 8
1 x+y
In evaluating - - - e • we note that f(l,1) = 0, This means that ex+y is a part
f(Dx:,Dy)
of the complementary function. (To see this, take ¢ 2 (y+2x) = ey;.'Lx+ y;2 (y+2x); then
We write
1 x; +y 1 1 1 x+y
----e - xe •
f (DX.DY) 7 DX -2Dy + 1 7
The use of the formula
7) = V = V(x,y),
is illustrated below.
2 2x+y
EXAMPLE 6. Solve (x + 2y)e -•
,;_ , 2x ,;_
The complementary function is z = '+-'dY) + x'P2 (y) + e 'f-'3 (y-3x), A particular in-
1 2 2x + y 1 2
tegral is z - - - - - - ( x + 2y)e (x + 2y).
2 2
Dx (Dx: + 3 Dy - 2) <Dx:+ 2) <Dx+ 3Dy +3)
Setting 2 dx du
<Dx + 3Dy + 3 )u = x + 2y, the auxiliary system is
2
x + 2y -3u
268 !\0:\-HOMI lCEJ\EOl S LI~EAR. CONSTANT COEFFICIENTS
du ch.: du 2
Then y = 3x +a~ and from or + 311 "' X + 2y, we have
X
2
-r 2y - 3U
1 ch.:
3X 1 2 16 16 2 1 2 2 16 2
ue3x .. J(x 2 + 6x 3Y
+ 2n ) e · ch.: e (-x + - X -
3 9 27
+ -a)
3
and u -x - -x - - + 3 y.
3 9 27
2
Next, setting (fJ,; + 2)v = u and making use of the integrating factor e x, y being re-
. 2x 1 2 2 16 2 1 2 5 17 1 2x
garded as a constant, Je (-x3
- -x - - -,. -y)cb.:
9 27 3
= (-x - - x - -
6 18 108
+ -y)e
3
and v
Finally, setting (ll,; ... 2)w ~ v, we have
2,: 1 2 5 1 2 2 7 1 2x
Je (-x
6
- -x
18
(-x
12
- -x + -
9 216
+ -y)e
6
1 2 2 7 1
and 12x -gx + 216 t "i/'
Then and the general solution is
,.1, 2x ,.1, 1 2 2 7 1 2x +y
z = ef>i(y) • X'f-'~(y) + e '-!-'3 (3y-x) + (-x - -x + - + -y)e •
12 9 216 6
IRREDUCIBLE EQUATIONS WITH CONSTANT COEFFICIENTS. Consider the linear equation

with constant coefficients
8)
. r s + "v
ca r b s e ax + b ·J , are cons t an t s, the resul t of
!L"r.;
Srnce D-.:Dy(ce · ) where a, b, c
substituting
9)
. ,y,: + by
in 8) is cf(a,b)e =0. Thus, 9) is a solution of 8) provided
10) f(a, b) = 0,
with c arbitrary. Now for any chosen value of a ( or b) one or more values of
h (or A) are obtained by -means of 10). Thus, there exist infinitely many pairs
of numbers (Ai, bi) satisfying 10). Moreover,
11) z 0,
is a solution of 8).
If f(Dy,Dy)z = (D-x:+hDy +k)g(Dx,D'.l)z,

then any pair (a, b) for which A + hb + k = O satisfies 10). Consider all such
pairs (Ai,h,) (-hbck,bi). By 11),
-kx
z e
i=l
is a solution of 8) corresponding to the linear factor (Dx + hDy + k) of f(Dx, Dy),
This is, of course, e-kx¢(y-hx), ¢ arbitrary, used above. Thus, if f(Dx, Dy)
has no linear factor, 11) will be called the solution of 8); however, if
f(Dx,Dy) has m<n linear factors, we shall write part of the solution involv-
ing arbitrary functions (corresponding to the linear factors) and the re-
mainder involving arbitrary constants.
EXAMPLE 7.
2
The equation is irreducible. Here f (a,b) = a + a+ b = O so that for any a = ai, bi
-ai (ai + 1), Thus the solution is
(X)
"' a-x-a-(a-+I)Y
z ,.;_,, Ci e t t t • with ci and ai arbitrary constants.
i=l
EXAMPLE 8. Solve
Corresponding to the linear factors we have ¢ 1 (y-2x) and e-xq>2 (y+2x) respectively.
2 2
For the irreducible factor Dx -Di we have a- b = O or a = b •
The required solution is
z with ci and.bi arbitrary constants,
In obtaining a particular integral of f(Dx ,Dy )z F(x,y), all procedures

used heretofore are available.
EXAMPLE 9.
(X) 2
'v b-x+b-y
From Example 8, the complementary function is z ,.;_,, Ci e t t
i=l
1 2X+3y 1 2x+3y 1 2X+3y

For the particular integral: --e ---e - - e
2 7
Dx-Dy 2- (3/
THE CAUCHY (ffiDINARY) DIFFERENTIAL EQUATION f(xD)y = F(x) is transformed into a

linear equation with constant coefficients by means of the substitution x = e 2
(see Chapter 17). The analogue in partial differential equations is an equa-
tion of the form
F(x, y), crs = constant,
which is reduced to a linear partial differential equation with constant co-

efficients by the substitution
u 1J
X e • y e •
270 NON-HOMOGENEOUS LINEAR, CONSTANT COEFFICIE~TS
EXAMPLE 10. Solve

3u-2v
e
whose solution is z
Thus, the general solution (expressed in the original variables) is

3
z = cp1 (1n y) + x2¢2 (ln 1-) - 1 x
or
2 9 2
X y
SOLVED PROBLEMS
REDUCIBLE EQUATIONS.
1. Solve o.
The general solution is z = cpi(y +x) + e- 3x¢2 (y-x).
2. Solve Dx(2Dx - Dy + 1) <Dx + 2Dy - l)z = 0,
The general solution is z = ¢dy) + ey ¢ 2 (2y + x) + ex ¢ 3 (y ~ 2x).
3
3. Solve
2
( 2Dx + 3Dy - 1) (Dx - 3 Dy + 3) z = o. The general solution is
4. Solve
3 particular integral is
The complementary function is z = ¢dx) + e Y ¢ 2 (2y-x), A
3 3
- - - -- - 3 cos(3x - 2y) - - - - - cos(3x - 2y) - - cos(3x- 2y)
2
2DxDy + Dy - 3Dy 2(6)-4-3Dy 8-3Dy
3 <3 + 3Dy) cos(3x - 2y) 3

- ( 8 + 3Dy) cos(3x - 2y) ~ [ 4 cos(3x- 2y) + 3 sin(3x- 2y)].
2 100 50
64 - 9Dy
2 2
X - 4xy + 2y
The complementary function is z
1 2 2
A particular integral is denoted by z = - - - - - - - - - - ( x - 4xy + 2y ) .
Dx <Dx + Dy - 1) (Dx + 3Dy - 2)
1
To evaluate it, consider
2 2
(x - 4xy+ 2y ) ½- - - -1- - ( x 2 - 2
4xy + 2y )
DX +3Dy - 2 -1 + ½(DX + 3Dy)
2 2 2
½[-1- ½(/.),;+3Dy) - -}.(Dx+3Dy) - .... ](x -4xy+2y)
~ - ½(/-
2
½[-(x - 4xy + 2/) - (-5x + 4y) - 7/2] 4xy + 2/-5x + 4y + 7/2).
Consider next
- ½ 2
- - - - ( x -4xy + 2y -5x+ 4y+ 7/2)
2
½- - - - ( x
1 2 2
-4xy + 2y - 5x + 4y + 7/2)
Dx + Dy- 1 1-(Dx+Dy)
2 2 2 2 2
½[1 +(Dx+Dy) +(Dx+Dy) + .. ,](x -4xy+2y -5x+4y+7/2) = ½(x -4xy+2y -7x+4y+½),
½2 2 I 3 2 2 2
Finally, z -(x -4xy+2y -7x+4y+~) ~ ½(x /3-2x y+2xy -7x /2+4xy+x/2).
Dx

X 2X 1 3 2 2 2
z = c/>i(y) + e q>2 (y-x) + e q>3 (y-3x) + -(2x -12x y+12xy -2lx +24xy+3x).
12
TYPE:
The complementary function is z =
1 x+y+2
For the particular integral, ------------- e cos(2x - y)
(DX + Dy - 1) (DX + Dy - 3) (DX + Dy )
x+ y+2 1
½e ---- cos(2x -y)
DX+ Dy+ 2
1 x+y+2 Dx+Dy-2 1 x+ y+2

- e ----"---- cos(2x - y) - e (Dx+Dy-2) cos(2x-y)
2 2 2 10
Dx + 2DxDy + Dy - 4
1 x+y•t-2 .
- - e [srn(2x-y) + 2 cos(2x-y)]. The general solution is
10
X ,.1, 3X ,.1, 1 x+y+2 .

z = e '+-'1 (y-x) + e '+-'2 (y-x) + q>3(y-x) - - e [srn(2x-y) + 2cos(2x-y)].
10
x+2y( 2 2)
e X + 4y ,
The complementary function is z = c/>i(y) + ¢ 2 (y + 2x) + ¢ 3 (y - x).
For the particular integral
we first
NON-HOMOGENEOUS LINEAR, CONSTANT COEFFICIENTS
1 1 2 2
find u - - - - - - ( x +- 4y )
3
1 + ½(Dx + Dy)
1 2 2 2
-(Dx +Dy) + .... •] (x + 4y )
9
~[x2 2 2 10] 1 2 2
3 + 4y -(x + 4y) + -(9x + 36y - 6x- 24y + 10),
3 9 27
1 1
then v u - - -------u
3 1
-(2Dy -DX )
l+ 3
1 2 2
--(9x +-36y -72y-1-58),
81
2 1 2 2
and finally, z (1-l\+Dx+•••)V - -(9x +36y -18x-72y+76).
81
z =
TYPE: IRREDUCIBLE EQUATIONS.
c+)'
8. Solve f(Dx,Dy)z e
The complementary function is z from Example 9.
1 x+y
The short method for evaluating the particular integral - - - - e cannot be used, since
fCDx,Dy)
f (a, b) = f ( 1, 1) = o. We shal 1 use the method of W1determined coefficients, assuming the par-
c icular integral to be of the form z = Axex+y + Byex+y.
x+y
Now Dxz = (A +Ax+By)e ,
hence A -2B = 1. Taking A~ 1, B = 0, we have as particular integral z = xex+y; taking A= 0,
B=- ½, we have z - ½yex+y; and so on. Choosing the first, the required solution is
(X) 2
'v bic+biy x+y
z = .:;.. ci e + xe
9. Solve
The complementary function is z
1 2 1 1 (x2 _ y)
The particular integral (x _ y)
2 2 2
2Dx -Dy + Dx Dy 1
_ Dx + 2ni
D2
y
2 2
1
[1 +
Dx + 2Di + <Dx + 2Dx) + • •.] (x2 - y) - - [x
1 2
-y+---+
2x+4 _2]
D2 D2 D~ D2 D2 D~
y y y y y y
SOl\-IIO}IOGEl\EOl S Lfl\EAR. COl\STA,1\'T UlEFFICIEl\TS
1 2 2 2 't 1 4
- -(x -y+xy +2y -y /12) 12xy
1)2
~
10. Find a particular integral of sin(2x+y),
A particular integral is given by
1 1
- - - - - - - - - sin(2x + y) - - - - - - - - sin(2.x ty)
2
(1/X + Dy ) (DX - Dy - Dy ) (-4+Dy)<Dx-Dy+ 1)
1
sin(2x+ y) - - - - - sin(2x t y)
DxDy - Di - 4Dx + 5Dy - 4 5Dy-4Dx - 5
5Dy-4Dx+5 sin(2.x+y) 1
- - [ 5 sin(2.x + y) - 3 cos(2x + y)].
34
25Di- 40DxDy + 16D: - 25
The method of undetermined coefficients with z = A sin(2.x t y) + B cos(2x, y) may also he

used here,
11. Find a particular integral of
1 3x + 4)
A particular integral is - - - - - - -- - - e sin(x-2y)
2
(fl - 2Dy + 5) (DX + Dy + 3)
e3x+4y 1 sin(x-2y) 1 3X + 4 y 1
- e - - - - - sin(x -2y)
2 2
6D -11DD -2D +15D -30D 5 31), - 61\ - 4
X X y y X )
1 3X+4Y
- e
5
- __1__ e3x+ 4Y [ 15 cos(x-2y) + 4 sin(x -2y)],

1205
12. Solve or
u V 3 2 3 2
The substitution x = e • y = e • x y DxDyz = Du<nu-l)(D,,-2)D,,(n,,-l)z,
The required solutinn is
z ¢>i(v) + ¢ 2 (u) + eu¢3 (v) + ev¢4 {u) + ¢ 6 (v+u) or, intheoriginalvHriables,
z q\(ln y) + ¢ 2 (ln x) + x¢ 3 (ln y) + y¢4 {ln x) + ¢,6 {ln xy)

Y11(Y)+ Ylz(X) + X\f3{y) + Y \f4{X) + \f5{xy).
27-! NON-HOMOGENE01JS LINEAR. CONSTANT COEFFICIENTS
2 2 2 2 3
13. Solve (x D;; - 4 y Dy - 4y Dy - 1) z
2
x y ln y.
The substitution x =e'"', y =ev transforms the given equation into

2U+ 3V
ve
2u+3v
A particular integral of this equation is given by ------ve
2u+3v
e - - - - - - - - - - - v.
1 24
By inspection, a solution of (D~ - 4Di + 3Du -24Dv -35 )w = v is found to be w = - - v +
35 2
(35)
. . . 1 Zu+ 3V
Hence, t he particular rntegral 1s z = - ---e (35v- 2 4).
(35/
The required solution of the given differential equation is
(X)
z
"" aiu + biv 1 2u+3v
= ,:;_, cie -- e ( 35 v- 24 ) or, in the original variables,
i=l 1225
(X)
"" a. b- 1 2 3
z ,:;_, Ci X t y t -- x y (35 ln y - 24),
1225
i=l
Solve each of the following equations.
15,
16,
17,
2
18. (DX + WY) (DX + 2Dy + 1) (DX + 2Dy + 2) z = o.
Ans. z =c/>i(y-2x) + e-xc/>2 (y-2x) + e-Y[¢ 3 (y-2x)+y¢4(y-2x)]
19. (Dx+ Dy) (Dx+ Dy- 2)z = sin(x + 2y),
Ans. z = c/>i(y-x) + /X¢2 (y-x) + _l_ [6 cos(x+2y) - 9 sin(x+2y)]

117
20.
1 X 1 -x
21. Ans. z + -xe - -xe
2 2
NON-HOMOGENEOUS LINEAR, CONSTANT COEFFICIENTS
22. Ans. z = c/>i(y) + c/>2 (y +x) + exc/>3 (y-x) + ln x
23.
2
(3DxDy -2Dy -Dy )z = cos(3y + 2x). Ans. z = c/>1 (x) + e½Y c/>2 (3y +2x) - ½sin(3y +2x)
2X-3y at•X +b,-y 1 2x-'y 2 2
24,
2 2
(Dx + DxDy - Dy + Dx - Dy ) z = e , Ans, z = I ci e " - 6e -' • a.+
t t t t t t = 0
a.b.-!i.+a.-b.
2 2 x+y .
25. (3D;,-2H1 +Dx-l)z =3e srn(x+y).
2
Ans. z = 3a~ - 2b + a. - 1 0
t t t
26, (D; + 2flxDJ-2Dy + 3)z = ex+y cos(x + 2y),
Ans. z = 0
27.
aix+biy 1 -2x 2 2 2
Ans, z I c,, e + -27 e (9x + 18y + 18x + 12y + 16), a.t +a-b-+a.+b.
t t t t
+l = 0
2
28. e y cos '3x + ex sin 2y.
"' aix+biy 1 2y 1 x
Ans. z = .:.., ci e - e cos 3x - e (cos 2y + 3 sin 2y),
16 20
29,
30.
31.
32,
2 2
Ans. z = c/>i(x y) + c/>2 (y/x) + ½On x) ln y + -½ lnx lny
33.
CHi\PTEH :u
Par! ial Differential Ettuations of Order Two

\\ ilh Vai-iable Coeffieients
Tl!E MC.ST GENERAL LINEAR PARTIAL DH,FEREJ\i'TIAL EQUATION of order two in two indepen-
ilP n t var rn h ll!S has the form
l I Rr + Ss + Tt +Pp+ Qq + Zz = F
wlterP N,S,T,F,(),Z,F are tunctions of x and y only and not all R,S,T are zero.
1:kltJrt'. l'onsidcring the general equation, a number of special types will be
f.n,afPr!,
TYPE I.
·/z
2a) r = = F/R = F1(X, y)
2
-ax
-a 2z
'2b) s -- F/S F2(x, y)
-ax -ay
-a2 z
~:,: ) t = - 2
F/T F3(X, y),
-ay
T\i,,:-,,· an, n'ducible equations with constant coefficients (Chapter 32), but a
, ci1n't·t method of solving will be used here.
EXA~\PLE 1. Solve s = x -y,
-a2 z . t -az 2 , /,
Integrating s :::: -- = x- y with respec toy, p ox = xy -
I
2y + 't'(x), tj; arbitrary.
-ax-ay
l11t,,~ra1 ing Lliis relation with respect to x, 2 2
z -- I
2x y -
I
2 xy
,i
wh,!rl' - /1 (x) ~ f (x) and ¢.,_(y) are arbitrary tun ct ions.
,b.
TYPE II.
Rr + Pp = R -ap + Pp F
-ax
Ss + Pp = s -ap + Pp F
-ay
Ss + Qq s -aq + Qq F
-ax
Tt + Qq =
T -aq + Qq = F.
-ay
T1tl'Sl' ar,, C>ssc,ntially linear ordinary differential equations of order one in

wit ic-h p (ur q) is the dependent variable.
276
PARTIAL OF ORDER TWO. VARIABLE COEFFICIENTS 277
)Y + 2",
EXAMPLE 2. Solve xr +- 2p ( 9x +- 6) e ,
Considering µ as th(! dependent variable, x as the indevendent variable, and y as con-

2
stant, the equation is x C/! 2p - (9x + 6)e 3:c+ } for which x is an integrating factor.
ox
2 )(+2y
Integrating (9x +- 6x)e , we have
2 1 2 3d2\
X p C -(9x t- 6x)e
Dy
" 3/e3"+2y + 'P1(Y) or p
Tlwn {y) is the required solution,
TYPE III,
4a) Rr + Ss + Pp F or R op + Sop = F - Pp
0X oy
4b) Ss + Tt + Qq F or s cJq +
T cJq F - Qq.
c)X cJy
Ttese are linear partial differential equations of order one with p (or q)
as dependent variable and x,y as independent variables.
cJp cJp 2
EXAMPLE 3, Solve 2xr - ys +- 2p = 4xy2 or 2x y c!y = 4xy - 2p,
clx
dp
Using the method of Lagrange (Chapter 29), the auxiliary system is
4xl- 2p
2
r'rom the first two ratios, we obtain readily xy a,
~ 2 2
By inspection, 2y (2x) + 2py(-y) - y (4xy - 2p) ~ 0, Thus,
2
y clp - 2py cly p
2/ clx + 2py cly - / clp = 0 or 2 dx - 0, and - 2x b.
2
y
The g0ncral solution is ply

2 - 2x Then
~
t/;cx/).
Clz 2 2y 2 cJ 2 2 2
p - - 2xy 2 -1- y
2
ij1(xy ) and z = X ~ cp 1 (x/) -,- ¢2(Y), where - ¢1(xy ) y tj;(xy ).
c)x cJx
TYPE IV,
cJ2z cJz
5a) Rr +Pp+ Zz F or R + p + Zz = F
cJx2 cJx
cJ2z
5b) Tt + Qq + Zz F or T - + Q cJz + Zz = F.
cJy2 cJy
These are essentially linear ordinary differential ec1uations of order two with
x as independent variable in 5c1) and y as independent variable in 5b).
278 PARTIAL OF ORDER TWO, VARIABLE COEFFICIENTS
t - 2xq + x z = (x -2)e 3x+ Y,

2 2
EXAMPLE 4, Solve
2· 2 2 3x+2y
The equation may be written as (Dy - 2xDy + x )z = (Dy - x) z = (x - 2)e ,
The complementary function is z = exy c/>1 (x) + xex::, ¢ 2 {x) and a particular integral is
3x +2y
3x +2y x- 2 3x+2y e
- - - (x- 2 )e ----- e
2
(Dy - x) (2 -x/ X -2
3x+2y
e
Tlie required solution is + ---·
X -2
LAPLACE'S TRANSFCRMATION. This transformation on
I) Rr + Ss + Tt + Pp + Qq + Zz = G(u, VJ
consists of changing from the independent variables x,y to anew set u, v, where
6) u = u(x,y), v = v(x,y)
are to be chosen so that the resulting equation is simpler than 1). By means
of 6), we obtain
p
CZ CZ cu CZ CV
q
CZ
+ -
cx cu cx cv -cx cy
r
cp
ox
s
cp
cy
cq "
cy
Let
1' ) + Zz = F
be obtained by making the above replacements in 1) and rearranging. We shall

need only the coefficients
2 2 2 2
R I = R(ux) + Suvuy
~ + T(uy) an d r' = R( vx ) + S vx vy + T( vy ) •
We note that both are of the form
7)
i) Suppose b/a -/= f/e; then, if for u we take any solution of a~x + b~Y = 0 and
for v any solution of e(X + f(y = 0, 1) is transformed into 11) with B' = T' = D.
PARTIAL OF ORDER TWO, VARIABLE COEFFICIENTS 279
2
EXAMPLE 5. Solve a) x (y-l)r - x(/-l)s + y(y-l)t + xyp - q 0,
b) y(x+y)(r-s) - xp - yq - z = 0.
a) Here 7) is y(y-1)(!"'
~y / 0
or = (xt:' -yt:' )(xt:' _;:) = 0,

"X '°Y '°x Sy
Now x(;X -yfy =0 is satisfied by f=u=xy and xfx-fy=O is satisfied by f=v=xeY,
Moreover, it is easily shown that these solutions also satisfy the given differential equa-
tion, Hence, the required solution is
Z ¢1(Xy) + c/>z(XeY).
b) Here 7) is
Now f -
X
fy = 0 is satisfied by f =x + y and (;x =0 by f = y. However, neither of these
solutions will satisfy the given differential equation.
We take u=x +y and v = y, r = zuu s and the
given differential equation becomes
-y(x+y)zuv - XZU yzu - yzv - z 0 or uvzuv + uzu + vzv + z o.

This may be written as
1 1 1 cl c)z 1 1 CZ 1 1 c)z 1
z
UV + -z
V u
+ -z
U V
+ --Z = - ( - - + ;:;Zl + - ( - + -z)
u cv
( ~ + -)(-- + -z)
cv o.
UV cu clv V clu u V
c)z 1 c)w 1
Let -- + -z = w; then + -w 0 and u,u = y;(v). Now
CV V clu u
c)z 1
+ - z w zv and z
CV V
d 1 z = ¢1SJJ... + ¢2 (x + y) ,
where --1\.(v) v•f(v) and ¢ 1 (v) ;:; l\.(v), The required solution is
dv X + y y
2 2 2
EXAMPLE 6. Solve x r - y t + px - qy = x ,
Here 7) is o.
Now xf
x
-yfy = 0 is satisfied by f = xy and xf + y(;
X Y
= 0 by f = x/y, It is found read-
ily that these solutions satisfy the reduced equation x2 r - y t + px - qy = O; hence, the oom- 2
plementary function is z = c/> 1 (x/y) + ¢2 (xy). However, this complementary function may be
obtained along with the particular integral as followP. Take u=xy and v=x/y; then
2 2
1 X 2 1 2 X X 2x
p = yzu + yzv • q = XZU - 2ZV'
r = y zuu + 2zuv + 2ZVV • t=x 2
uu - 2 - 2 zuv + - ~ zvv + - 3 zv•
y y y y y
2 2 I
and the given equation becomes 4x zuv X or zuv = ~ .
Integrating first with respect to u, zv <j;(v) + ½u
I 2
and then with respect to v, (i:X ,
See Problems 9-10.

280 PARTIAL OF ORDER TWO. YARIABLE COEFFICIENTS
ii) Suppose b/a=f/e; then This

case is treated in Problem 11.
NON-LINEAR PARTIAL DIFFERENTIAL EQUATIONS OF CRDER TWO. One possible method for
solving a gi,ien non-linear partial differential equation of order two
8) F(x,y,z,p,q,r,s, t) = 0
is suggested lly several of the examples of linear equations above. In each of

F.xample~ 1-3, the first step consisted in finding a relation of the form
9) u = f(v), f arbitrary,
where u = u(x,y,z,p,q) and v = v(x,y,z,p,q), from which the given differen-

tial equation could be derived by eliminating the arbitrary function. Such a
relation 9) is called an _intermediate integral of 8). Forexample, p-xy+½y 2
= f(x) is an intermediate integral of s = x - y, (Example 1).
It can be shown that the most general partial differential equation having
u = f(v), f arbitrary,
where u = u(x,y,z,p,q) and v = v(x,y,z,p,q), as intermediate integral has
the form
2
10) Rr + Ss + Tt + U(rt-s ) = V,
where R,S,T,U, V are functions of x,y,z,p,q. However, it is evident from the

definitions of R,S, ···, V that not every equation of the form 10) has an in-
tern-ediate integral. The discussion below concerns Mange's method for deter-
mining an intermediate integral of 10), assuming that one exists.
TYPE: Rr + Ss + Tt = V. Consider the equation
11) Rr + Ss + Tt = V,
that is, 10) with U identically zero. Since we seek z as a function of x and
y, we have always
clz
121) dz = -- dx + -clz d y = p dx + qdy,
clx cly
clp dx
122) dp + -clp d y = r dx + s dy,
clx cly
123) dq clq dx + -clq d y = s dx + t dy.

clx cly
Solving the latter two for r = dp - s dy t = dq - s dx and substituting

dx dy
in 11), we obtain R dp - s dy + Ss + T dq - s dx = V or
dx dy
13) s[R(dy/ - Sdxcly + T(dx/] = Rdydp + Tdxdq - Vdxdy.

PARTIAL OF ORDER TWO. \ ARL\131.E COEFFICl1':\TS :2BI
The eciuations
R dy dp + T dx d(/ - V r!x dy f)
are callpd .lionge's enuntions.
Su1,11rn-;(, R\riy) 2 -- S dxrly +T(dx) 2 , (Ady, BrJx·J = 0. Jf now

2
u = t1(x,y,z,p,'?) "" :i, v - v(x,y,z,p,rJ) ~ J, :-;atisty Ll1P system
y + Bclx
0,
tLen
is an intern,0diate integntl of 11) since u=r1, v=b satisfy 13) and, hence,
11).
Suppose R(dy/- S dxdy + T(dx/ = (A 1 dy +B 1 cix)(A 2 dy +B 2 dx) = 0,

where A 1 B2 - A 2 B 1 -le O identically. We now have two systems
I A dy + B dx A2 dy + B 2 dx = 0
l
1 1 ()
and
l R dy dp + T c/x c/q - V clx cly = 0 R clydp + T clxdq - V dxdy = 0.
It PithPr systet:, i::; intPgrable, w0 are led toanintermediateintegral of 11);

if both are intc,(!.rallle, we Lave two int(!rmediate integrals at our disposal.
Procedures for find_ing a solution of ,1 given equation for which intermediate
integrals have been obtained will l,e discussed in the examples and solved
problems.
2 2
EXAMPLE 7. Solve q(yq + z)r - p(2yq -"Z)s + yp t • !' q o.
llere H=q\yq+z), S:--p(2yq+z), T=y/, V=-/q; Monge'sequationsare
H(dy)
2
- s cix dy +
,
1 (cix)
2
q(yq +Z)\dyl • p\2yq +z)cixdy + y/(cix/
(q dy-+- pdx) [(yq + z)d_v + ypdx] = O
2 2
and H dy dp 4 T dx dq - V dx dy q(yq + z)dydp .,_ yp dx <iq + p q cixdy O.
q dy + p dx ~ o
We seek first a solution of the system
l q ( yq ~ z )dy dr1 + yp dx dq + /
2
q dx dy = 0.
Combining the first r,quation and 12 1 ), we have dz~ O ancl z = a. substituting in the second
r·quation dy = -p dx q, obtained from the first, we obtain
(yq-1-z)dp -p(ydq~qdy) = o.
lie add -I· dz ~ 0 to this, obtaining
dp y dq + q dy +- dz
(yq 1-z)clp - p(ydq+qdy+dz) = 0 or
p yq + z
w1 th solution 2'-1:L:_:_ = b. Then yq + z = p• f (z) is an intermediate integral. The Lagrange

p
cix d_y dz dy dz
system for this first order equation i8 - -- From we obtain yz a,
f (z) -y z -y z
182 PARTIAL OF ORDER TWO. VARIABLE COEFFICIE~TS
and from
dx dz we obtain x =
J f (z) -dz = ¢ 1 (z) + b, Thus, the required solution is
j (z) z z
(yq + z)dy + ypdx = O

Consider next the seconrt system
[ q(yq+z)dydp + y/Jxdq + /qdxdy = o.
From the first equation, pdx + q dy = -z dy/y; then dz = -z dy/y and yz = a, Substituting
from the first equation, the second becomes
dp dq dy _
qy dp - py dq - pq dy = 0 or ------0
p q y
with solution qy/p = b, Then qy ~ p•g(yz) is an intermediate integral, The Lagrange sys-
tern is _±__ = dy • dz= O. Th en z =a · t equa t 10n

and th e firs . -dx- = dy
- h as so 1 ut 10n
.
g(yz) -y g(ya) -y
x =- J g(ya) dy = ¢ 2 (ya) + b. We thus obtain x = ¢ 1 (z) + ¢ 2 (yz) as before.

y
The solution may also be obtained by using the two intermediate integrals simultaneously,
z z,g(yz)
Upon solving them for p q =
y[f(z) - g(yz)]
and substituting in p dx + qdy = dz, we have yz dx + zg(yz)dy = yf(z)dz - yg(yz)dz.
Writing f(z) = zfi(z) and g(yz) = -yzg 1 (yz), this equation becomes
dx = f 1 (z)dz + g 1 (yz)[zdy + ydz]
and, integrating,
See also Problems 12-16,
2
TYPE: Rr + Ss + Tt + U(rt-s ) = V. Consider equation 10) with U ~ O. By
substituting r = dp- sdy, t = ~1_1_-:__sdx as in the preceding type, we obtain

dx dy
s [R ( dy / - S dx dy + T ( dx / + U ( dx dp + dy dq) ] = R dy dp + T dx dq + U dp dq - V dx dy.
Tile equations
2 2
R ( dy) - S dx dy + T ( dx) + U ( dx dp + dy dq) == 0
R dy dp + T dx dq + U dp dq - V dx dy = 0
are called Mange's equations. Note that when U = 0, these equations are 14 1 )
and 14 2 ). However, unlike 14 1 ) and 14 2 ), neither can be factored.
We shall attempt to choose r,,. = r,,.(x,y,z,p,q) so as to obtain a factorable

combination
16) r,,. [R ( dy/ - S dx dy + T ( dx )2 + U ( dx dp + dy dq)] + R dy dp + T dx dq + U dp dq - V dx dy

= (a dy + b dx + c dp) (a dy + /3 dx + y dq)
= aa(dy/ + (a/3 + ba)dx dy + b/J(dx/ + c/3 dx dp + ay dydq + ca dydp
+ by dx dq + cy dpdq = 0.
PARTIAL OF ORDER TWO. YARIABLE COEFFICIENTS
Comparing coefficients, we have

aa=Rt,, a/3+ba=-SI\.-V, b/3=Tf\, c,3=Ul\.=a,, cCt=R, b,=T, cy=U.
The fi.rst relaLion will be saLisfied by taking a= I\ and a= R; this choice de-
termines b = T/U, 3 = I\.U, c = 1, y = U. The remaining relation ap + ba = -SI\- V
2
takes the form U 1\. + TR = - SI\ - V or
u
2
1 7) u\ + SUI\ + TR + UV = O.
In general 17) will have two distinct roots /\. = /\. 1 , /\. = ,\ 2 ; thus, 16) can be
factored as
( l\. 1 Udy + Tdx + Udp) (Rdy + l\. 1 Udx + Udq) = 0 and
(l\. 2 Udy +Tdx +Udp)(Rdy + l\. 2 Udx +Udq) O.
There are four systems to be considered. The system l\. 1 Udy+Tdx+Udp = 0,

A2 Udy+Tdx+Udp = 0 implies (/', 1 - A. 2 )Udy = 0 and, hence, unless \ 1 =/\. 2 , Udy=0
identically. Similarly, the system Rdy+ l\. 1 Udx+Udq = 0, Rdy+ A- 2 Udx+Udq = 0
implies V clx = O identically. We therefore shall use only the systems
/\. 1 Udy+ T dx + U dp = 0 l\. 2 Udy + Tdx + Udp 0

19) [
and [
Rely+ l\. 2 Udx + U dq = 0 Rdy+l\. 1 Udx+Udq = O.
Each system, if integrable, yields an intermediat(: integral of 10).
2
EXAMPLE 8. Solve 3s - 2(rt - s ) = 2.
2 2
lhne, R = 0, S = 3, T = 0, u= -2, V = 2. Tlwn L /\. + SL;\+ TR+ u'V o.
\i = - ½ and ;\, = 2. We seek solutions of thP systems
f "- l'dy + Tdx + udp

1 = dy - 2dp = o
and
,\ 2 Udy + Tdx + Udp = -4dy - 2dp = 0
l Rdy + l\.2Ldx + Udq = -4dx - 2dq = 0
[
Rdy , >., 1 udx-+ /dq = dx - 2dq = O.
From the first system, y-2p-a and 2x+q=b; then (i) y-2p j(2x•q) is an intermedi-
ate integral. From the second system, 2y + p = a and x - 2q = b; then ( i-i) 2y + p = g (x - 2q)
is an intermediate integral. Since q appears in the argument ot both f and g, it is no
longer possiblP to obtain a solution of the given equation in\'olving two arbitrary functions
by solving for p and q and substituting in dz = p dx + q dy.
We shall attempt to find a solution involving arbitrary constants from the intermediate
integral y-2p = f(2x+q). To obtain an integrable equation, takp f(2x+q) = a(2x+q) + (3,
where a and /3 are arbitrary constants. The Lagrange system for
y - 2p = a(2x + q) + /3 or 2p + aq = y - 2ax - /3
is dx dy dz
2 a y-2ax-/3
From the first two members, ax 2y + f. Substituting for ax, the last two members become
dy dz
a - 3y - 2f - /3
3 2
or a dz = (- 3y - 2f - /3)dy and az - y - 2.;y - (3y + T).
5 2
2
Thus, az (2ax,+ /3)y-+ ¢ 1 (ax - 2y) is a solution of the given equation involving
=
2y -
one arbitrary function and two arbitrary constants.

281 PARTIAL OF ORDER TWO. VARIABLE COEFFICIENTS
Treating the second intermediate integral similarly, we take 2y + p = y (x - 2q) +- 8 or

p -1- 21 q = yx - 2y + 8 , where y and 8 are arbitrary constants. The corresponding Lagrange
dx dy dz
system is From the first two members, y=2yx+f. Now the first
1 2y yx - 2y + 8
dx dz
and third members become -- = and Thus,
1 - 3yx - 2f + 8
5 2 c ,-1,
z = (2y- u)x + '+' 2 (y- 2-yx) is also a solution involving one arbitrary function and
2yx -
two arbitrary constants.
A solution involving two arbitrary functions of parameters A andµ will next be found.
Set 2x+q =I\. and x-2q =µ so that x = (2/\.+µ)/5. Then (i) and (ii) become y-2p =
j(/\.) and 2y+p = g(µ), and y = [f(/\.) + 2g(µ)]/5. Now
(iii) p 2y + g(µ) and
(iv) q I\. - 2x = ~ (x - µ).
Substituting the second value of p and the first value of q in dz p dx + qdy, we have
dz [-2y + g(µ)]dx + (/\. - 2x)dy
1 1
- 2(ydx+xdy) + -g(µ)[2dl\.+dµ] + -l\.[f 1 (/\.)dl\.+2g'(µ)dµ]
5 5
- 2(ydx+xdy) + ~[l\.g 1 (µ)dµ+g(µ)dl\.] + ~[/\.j 1 (/\.)+f(/\.)]d/\. - ~f(l\.)dl\. + !g(µ)dµ

5 5 5 5
2 1
and z = - 2xy + -Ag(µ) + -1\.J(/\.) - ¢i(/\.) + ¢ 2 (µ)
5 5
2xy + l\.y - ¢di\.) + ¢2(µ).
This solution may have been obtained by using the first value of p in (iii) and the sec-
ond value of q in (iv).
SOLVED PROBLEMS
1. Solve r = x 2 e -Y or X
2
e
-Y
One integration with respect to x yields p

oz ~
3
e -Y + ¢ 1 (y), and the second inte-
ox 3
gration with respect to x yields z
2 2
2, Solve xy s 1 - 4x y.
Integrating
oz
ox
Integrating this with respect to x,
where
PARTIAL OF ORDER TWO. VARIABLE COEFFICIK\TS
2
3. Solve xys - px = y,
op
y - - p
c3y 2
Integrating with respect toy, we get F:.. = l + <}.;(x) or oz = l_ + yy.;(x),
2 X Y X OX X
y
Integrating with respect to x, we get z cj;(x),
4. Solve t - xq - Sin y - X COS y,
Integrating c3q - xq = -(sin y + x cos y), using the integrating factor e -xy, we obtain
c3y
or q = --
oz = cosy + e"Y </;(x).
oy
A second integration, with respect toy, yields z

where rpi(x) = </;(x)/x.
5, Solve sy - 2xr - 2p = 6xy,
The auxiliary system for the equation

op
2x-- -
op
y - = -6xy - 2p is
dx dy dp
ox oy 2x -y -6xy- 2p
2
From the first and second ratios, we find xy = a. By inspection,
3 2 2
2y (2x) - (2yp+2xy )(-y) + y (-6xy-2p) = 0
3 2 2
so that 2y dx - (2yp + 2xy )dy + y dp = 0,
/(dp + 2xdy + 2ydx) - 2y(p + 2xy)dy p + 2xy

or 0, and b.
~ 2
y y
2 2
Thus, we obtain as solution p + 2xy = y f (xy ) • Then
c3z 2 2 2 2
- ~ -2xy + y </;(xy ) and where y <j;(xy ),
c3x
3
6. Solve xs + yt + q = l0x y,
xoq + yoq dx dy clq

The auxiliary system for the equation is
ox oy X y ·3
l0x y- q
From the first two ratios, x/y = a, By inspection,

(q-8x\)x - 2x~(y) + x(10x 3y-q) = 0
so that (q-8x 3 y)dx - 2x~dy + xdq = 0, or
and qx = 2x ~ y + b,
~
The general solution is qx = 2x y + y;(y/x), Thus,
oz
-
3 1
::: 2x y + - </;(-)
y and z = where
oy X X
:286 PARTIAL OF ORDER TWO. VARIABLE COEFFICIENTS
7. Solve
The equation may be written as 1 1 - l)]z

[D:,,2 -Dy-:;(:; = xy 2 _ x2y2 + 2x3y _ 2x3,
·1/c y -y/x
The complempntary function is z = e ¢ 1 (x) + e ¢ 2 (x).
For a particular integral we try z = A/+ By+ C, where A ,B,C are functions of x or con-
2 11 ] 1 1 2 2 22 3 3
stants. Then [D - D, - -(--1) z = 2A-2Ay-B-(-- - -)(Ay +By+C) = xy -x y +2x y-2x,
Y _I XX 2 X
X
identically. Equating copfficients of the several powers of y, we havp
1 1
(2 - ~ )A = x ( 1 - x),
X
3
Then A= -x ' and thP rpquired solution is z
8. Solve ys + p - yq - z = (1 - x)(l + ln y).
This equation is solved rpadily by noting that it may be put in the form
--- +
c/ z 1 oz z oz
( - + -Z)
1 1-x
--(1 + ln y).
ox cly y 'ox y oy Y y
oz 1 ow 1- X -X
Set ting u = + z the equation becomes - w - - ( 1 + ln y) for which e is an
oy Y · ox y
intpgrating factor. Then
1
-----
+- ln y fx (e -x - xe
-x )dx 1 + ln y (xe -x) + </;(y) and W = X
y y
In turn, integrating
oz 1
-- + - z x ~~_l_ + ex<f;(y), using the integrating factory, we find
oy Y y
yz x f y (1 + ln y)dy + e
xfY y<f;(y) dy
LAPLACE'S TRANSFORMATION.
9. Solve t - s + p - q(l + 1/x) + z/x = 0,
2
Setting (!;'y ) - !;'x!;'y = 0 and solving, we have I;' = x and I;' x + y.
For thP choice u=x and v=x+y,
1 o oz 1 oz
stituting in the given equation, we have zU1J z'cl- + -(z - z) -(-- z) + -(- - z) o.
X 1J
ou ov X OV
oz ow w oz
Let - z = w; then - + 0 and uw = u(- - z) </;(v) •
ov OU u ov
1J
oz 1 -1J 1 e
Integrating -- - z
0V
ii </;(v), we have e z ;; ¢1M + ¢(u) or z u ¢1M + e 1J ¢(u).
PARTIAL OF ORDER TWO. VARIABLE COEFFICIENTS 287
In the original variables, 1f (x + y) + ey g(x),
2 2
10. Solve xys - x r - px - qy + z = - 2x y.
From we obtain f y and f = xy.
Using u = xy, 1' =y, p = yzu, q = xz 1,+ zv, s
differential equation becomes

1 1 1 2u
z
:.1v
1
-z
V ~u
-z
u V
+
1
--
UV
z 2u
or ~(9: - z)
V
~("z
- vz)
1)2 OU ov u ov v2
Let
oz 1
- -z w; then
ou w 2u
and
w 2u
+ <j.;(v) or w
2u
2
+ u <j;(v).
Ov V ou u - v2, u v2 v2
2 2
oz 1 2u z u
Integrating U' -z + u <j;(v), we have + U </Ji(v) + ¢2(U) or
ov V 2 V v2
V
2 2
u u
z + UV </J 1 (v) + V ¢2(U) + U /\ 1 (v) + V ¢2(U),
V V
In tJi,, original variables,
2 2
11. Solve x r - 2xys + y t - xp + 3yq = 8y /x.
!!ere ,/(;,, / - 2xyfxfy + /(fy / = (xfx - Yfy / 0, and since the factors are not dis·
t inct we obtain only f = xy,
2
\Ve SL't u ~ xy and take V = y; then p = yzu, q =X ZU + ZV, r=y zuu, s = zu + xyzuu + yz'.'lV'
2 ,_ z
t -
X z .- 2x2 and the given differential equation becomes
~...:, 1' vv
2
y z,; 1 , + 3yz 11 ~ 8y/x or
an equatim1 of the Cauchy type, However, it is seen that v is an integrating factor; hence
2
1!3Z
:,v -+ 3v zu
3
8v /u and v 3z = 2v~/u+¢(u),
1'
2
2v 1 V 1
Thrn z -u + -- ¢(u)
3
and z
u
-c/J(u) + ¢1 (u)
2
V 2v
v2 1
+ -</J(u) + ¢du)
u v2
¢1 (xy)
1
+ - </;(xy)
2
+ r
X
y
2 y 2 2
or z ¢i(xy) + X ¢2(X,Y) + -X ' where </J(xy) X y ¢2(xy) •
P.\l{TIAL OF ORDER TWO. \ ARIAHLE COEFFICIE~TS
MONG E'S .l!ETJJOD.
12. Solve
2
Thr Mongp eqtwt ions arc q dx cly + p ( clx) = 0 and
From the first c•qiwt i 011, q cly , pd:.. 0' then ciz = /! clx + ,1 cly = 0 and
Substituting r1dy = -rdx in thr; second equation yields dr1 - r/dx = O; thus 1. q , x = b
and 1. q - x = }<,z) or [x-J(z)]q = -1 is an intermPdiute integral.
The required solution is obtained by solving this first order equation; thus
2 2 2
13. Sol Ve q r - 2pqs + p t pq
2 2 2
The Monge equations are (q dy + pd:,} = O and q dy dp + p dx dq - pq dx dy O,
Prom the first equation, q dy _.. p dx = O; then dz = p dx + q dy =O and z = a,

dp dq
Substituting qdy = -pdx in the second yields -qdµ+pdq+pqdx=O or - --+-+dx = 0
p q
and ex q/p = b. Thus eY q - pf (z) =O is an intermediate integral. The Lagrange system
for this equation is dx - dy - , dz = Q,
X
- e
dx - dy
From the second equation, Z = C, Then the first becomes with solution
f (c) - e
X
e ,f(c) + y = d, As required solution, we find
2 3
14. Solve x(r + 2xs + x t) = p + 2x ,
2 2 2 2
The Monge equations are (dy) - 2x dx dy + x (dx) = (cly - x dx) 0
3 3
and x dydp + x dxdq - (p + 2x )dx dy = 0,
3
\\e seek a solution of the system dy - x dx = 0, x dydp + x 3 dxdq - (p + 2x )dx dy = O.
2
From the f'irst equation, x - 2y a, Substituting dy = x dx in the second, we get
2 3
x dp + x dq - (p-+ 2x )dx = 0,
2 3 2
Using the integrating factor 1/x , we obtain the intermediate integral p + xq = x + x J (x - 2y).
dx cly dz 2
The Lagrange system is , The first two members yield x ,- 2y = c
1 X 3
x + xf (x
2
- 2y)
dx dz
and then the first and third become Solving,
3
x +xf(c)
2 2 2
z or + ½x f (x - 2y) + cp(x - 2y).
2
15. Solve q(l+q)r - (1+2q)(l+p)s + (l+p) t o.
The Monge equations are
q(l + q) (dy/ + (1-+ 2q) (1 + p)dx dy + (1 + p/ (dx/ = [q dy + (1 + p)dx] [(1 + q)dy + (1 + p)dx] O
2
and q(l+q)dydp + (l+p) dxdq = 0,
PARTIAL OF ORDER TWO. VARIABLE COEFFICIENTS 28()
Consider first the system

q dy + ( 1 + p) dx = O
q(l+q)dydp + (l+p/dxdq = 0.
From the first equation, p dx + q dy = - dx; then dz = -dx and x + z = a. The s ubsti tut ion of
q dy = -(1 + p)dx in the second yields
-( 1 + q) dp + (1 + p) dq = o
1+p 1 +p = f (x + z)
from which we obtain b. Thus, is an intermediate integral.
1+q 1+q
Consider next the system

( 1 + q) dy + ( 1 + p) dx ~ 0
q ( 1 + q) dy dp + ( 1 + p / dx dq = o.
From the first, pdx+qdy = -(dx+dy) so that dz= -(dx+dy) and x+y+z=a. The substi-
tution of (1 + q)dy = -(1 + p)dx in the second gives -q cip + (1 J. p)dq = O which is satisfied
l+p l+p .
by - - - = b. Thus, - - = g(x + y + z) is an intermediate integral.
q q
Solving the two intermediate integrals for p = fg+f-g, q f and substituting

g-f g-f
in the relation pdx + qciy = dz, we have
(fg+f-g)dx +fdy (g-f)dz, f g dx = - f ( dx + dy + clz ) + g ( dx + dz ) ,
dx + dy + dz + dx + dz
dx and
g (x + y + z) f (x + z)
2
16. Solve (x-z)[xq 2 r-q(xJ.z+2px)s+ (z+px+pz+p 2 x)t ] (lJ.p)q (x+z).
Monge's equations are

2 2 2
xq (dy) + q(x + z + 2px)dx dy + (1 + p) (z + px) (dx) [q dy + (1 + p)dx] [xq dy + (z + px)dx] 0
2
and (x-z)[xq dydp + (l+p)(z+px)dxdq]- (l+p)/(x+z)dxdy = O.
Consider first the system

q dy + ( 1 + p) dx = 0
2 2
(x-z)xq dydp+ (l+p)(z+px)(x-z)dxdq- (l+p)q (x+z)dxdy = O.
r'rom the first equation, pdx + qdy = -dx; then dz= -dx and x+z = a. Substituting qdy =
-(1 + p)dx, z = a -x in the second, we have
i) -(2x-a)xqdp + (2x-a)(a-x+px)dq + (l+p)qadx = 0.
To solve this equation, consider x as a constant so that dx = O. Then i) becomes

-(2x-a)xqdp + (2x-a)(a-x+px)dq = 0 or x ( q dp - p dq) - ( a - x) dq = 0
and xp+a-x = </;(x). To determine </;(x), we take the differential of this relation,
q
2
q(xdp+pdx-dx) - (xp+a-x)dq = q d<f;
2
and obtain xq dp - xpdq = q clij;-pqdx + qdx J. adq - xdq.
( 2x - a ) ( a - x ) dq + ( 1 + p) qa dx (1 + p)qa dx .
From i), xq dp - xpdq (a- x)dq +
2x - a 2x- a
290 PARTIAL OF ORDER TWO. VARIABLE COEFFICIENTS
2 (1 + p)qa dx
then q d,jJ - pq dx + q dx + a dq - x dq (a- x)dq +
2x- a
2(px + a - x)
dx
2-/-;
- - - dx and
y;
b f(x+z),
dy; =
q(2x - a) 2x- a 2x- a
xp + a- x xp + z
Thus, f (x + z) is an intermediate integral.
q(2x-a) q (x - z)
Consider next the system

xqdy + (z + px)dx = 0
2
(x-z)xq dydp + (l+p)(z+px)(x-z)dxdq- (l+p)/(x+z)dxdy = 0,
From the first equation, p dx + q dy = - z dx/x; then. dz = - z dx/x and xz = a, Substituting

xq dy = - (z + px)dx, z = a/x in the second, we have
2 2 2
ii) -xq(x -a)dp + x(l+p)(x -a)dq + (l+p)q(x +a)dx = 0,
Considering x as a constant, this becomes q dp - (1 + p)dq = O and we have ~£ = y;(x), From

q
2
(1 + p)q(x + a) dx,
this n,lation we find qdp-(l+p)dq q2dy;, while from ii) qdp-(l+p)dq
2
x(x -a)
2 2
Then dy; =
(l+p)q(x + a) dx y;,(x + a) dx dx 2x -dx)Y1
(-- + - , lny;=-lnx+ln(x -a)+ ln b,
2
2 2 X 2
q2x(x - a) X (x - a) X -a
2
b (x a) 1+p
~~-
-
and y; Thus, g(xz) is an intermediate integral.
X q q(x - z)
Solving the two intermediate integrals, we find p = czg

xg-f
and q
1
xg-f'
then
dz p dx + q dy = f - zg dx + _ l _ dy or f(x+z)(dx+dz) + dy zg(xz)dx + xg(xz)dz.

xg-f xg-f
Thus, y + ¢ 1 (x + z) = ¢ 2 (xz) is the required solution.
17, Solve 3r+s+t+(rt-s2) = -9.
Here, R=3, S=T=U=l, V=-9; then
u2r-...2 +sure+ TR+ uv ~ re 2

+ re - 6 = o and
We seek solutions of the systems (see equations 19))
2 dy + dx + dp O, R dy + re 2 U dx + U dq 3 dy - 3 dx + dq b
and ½Udy+ Tdx + Udp -3dy + dx + dp 0, Rdy + ~Udx + Udq 3 dy + 2 dx + dq o.
From the first system, we have 2y+x+p=a, 3y-3x+q=b; thus, p+2y+x = f(q+3y-3x) is-
an intermediate integral. From the second system, we have - 3y + x + p = c, 3y + 2x + q = d;
thus, p - 3y + x = g (q + 3y, 2x) is an intermediate integral. Since q appears in the argument
of both f and g, it will not be possible to solve for p and q as before, and it will not be
possible to find a solution involving two arbitrary functions. We give two solutions involv-
ing arbitrary constants,
Replacing the arbitrary function f of the first intermediate integral by a(q + 3y-3x) +/3,
we obtain
p+2y+x = a(q+3y-3x) + f3 or P - aq (3a-2)y - (3a+ l)x + f3

dz dx = dy, we find
for which the Lagrange system is From
(3a - 2)y - (3a + 1 )x + /3 1 -0.
dx dz dz
y + ax = f ; then and
2
(3a -2)y- (3a + l)x +/3 -(3a +a+ 1 )x + 3af - 2f + /3
2 2
Thus, z = ½(3a - 5a - 1 )x· + (3a - 2)xy + /3x + ¢ 1 (y +ax) is a solution involving one arbi-
trary function and two arbitrary constants,
Replacing the arbitrary function g(q + 3y + 2x) of the second intermediate integral by the
1 inear function y(q + 3y + 2x) + 8, we obtain
p-3y+x =Y(q+3y+2x)+8 or p - yq = 3(Y+ l)y + (2Y-l)x + 8
dx dy = dz dx
for which the Lagrange system is From we get
1 -Y 3(y+l)y+(2y-l)x+8
dz dz
y + yx = f; then dx and
1 3(y+l)y + (2y-l);.., + 8 2
-(3y + y + 1 )x + 3yf + 3f + 8
2 2
z = --½(3y +y+l)x + (3yf+3f+8)x + TJ•
2 2
Thus, z = -½(3Y+5Y-l)x +3(Y+l)xy+8x+¢ 2 (y+yx) is also a solution.
2
18. Solve xqr + (p+q)s + ypt + (xy-l)(rt- s ) + pq = 0,
Here, R=xq, S=p+q, T=yp, U=xy-1, V=-pq; then
2 2
U !1.. + SU/\..+ TR + UV = (xy -1 / !1..
2
+ (p + q) (xy - 1 )ii.. + pq = O and ~-
xy -1
- p dy + yp dx + (xy - 1 )dp = o
Consider first the system The system is not inte-
[
xq dy - q dx + (xy - 1 )dq = 0
grable since neither equation is integrable.
Consider next the system -qdy + ypdx + (xy-l)dp O, xqdy- pdx + (xy-l)dq = O.
We multiply the second equation by y, add the first, and divide by xy -1 to obtain
q dy + dp + y dq = O and thus p + yq = a, Again, we multi ply the first equation by x, add
the second, and divide by xy -1 to obtain p dx + x dp + dq = O and thus xp + q = b, How-
ever, the form of the resulting intermediate integral xp + q = f(yq + p) or yq + p = g(xp.+ q)
does not permit a solution involving two arbitrary functions,
To obtain a solution, involving one arbitrary function and two arbitrary constants, we
replace f (yq + p) by the 1 inear function a (yq + p) + /3 in the first form of the intermedi-
ate integral above and have
(x -a)p + (1- ay)q = /3,
dx dy dz
The corresponding Lagrange system is From the first two members we
x-a 1-ay /3
a
obtain aln(x-a) + ln(l-ay) = Inf or (x-a) (1-ay) f , and from the first and third
members we get z = /3 ln(x-a) + TJ• Thus, the solution is
z = (3 ln(x-a) + ¢[(x-a/(1-ay)].
PARTIAL OF ORDER TWO. VARIABLE COEFFICIE~TS
Solve.
19. r = xy Ans, z = x ¢ 1 (y) + ¢ 2 (y) + ½x\
2 2 1 3 3
20. s = z = ¢i(x) + ¢ 2 (y) +
X + y
3(x y+ xy )
2
21. t = - X sin(xy) z = y c/Ji(x) + ¢ 2 (x) J. sin(xy)
22. xr - p = 0
2
23. xr + p 1/x
2
24, yt - q 2x y
2
25. ys - p = xy sin(xy)
-y
26. t + q = xe
2
27. r + s = 3y
2
28. xyr + x s - yp = x 3 ey
2
29. 2yt - XS + 2q = X y
2 2 3
30. xr + ys + p = 8xy + 9x + X
LAPLACE'S TRANSFORMATION.
31. 6r - s - t = 18y - 4x
2 2
32. x(xy-l)r - (x y -l)s + y(xy-l)t + (x-l)p + (y-l)q ~ 0
2
33, x(y-x)r - (/-x )s + y(y-x)t + (yJ.x)(p-q) = 2(x•y+l)
Hint: Let x+y=u, xy=v. Ans. z ~ </;i(x + y) + ¢ 2 (xy) + x - y J. ln x
2
34, (y-l)r - ( / - l)s + y(y-l)t + p - q 2ye x(l-y) 3
Ans.
Ans.
36, r - 2s + t + p - q = ex ( 2y - 3) - e y
Hint: Let x+y=u, y=v.
37. / ( r - 2s + t) - y(p-q) - z = / Ans. z
MONGE'S METHOD,
38, (ex -1) (qr- ps) = pqex I.I.:p=y;(z).

39. r-3s-lOt =-3 I.I.: p+2q = tj;i(y+5x), p-5q = tp 2 (y-2x)

G.S.: z = q\(y+5x) + ¢ 2 (y-2x) + xy
I.I.: p=qtp(z).
41. qr-(l+p+q)s+(l+p)t 0 I.I.: p-q =tj; 1 (x+z), p+l = qtj; 2 (x+y)

G.S.: z = f(x+z) + g(x+y)
2 2 1
42. ( 1 - q) r - 2 ( 2 - p - 2q + pq) s + ( 2 - p) t 0 I. I.: - q = tp(y + 2x - z)
2-p
G.S.: x + ycj)i(y+2x-z) = ¢, 2 (y+2x-z)
2
43. 5r-10s+4t-(rt-s ) ~-1
I.I.: 3y+4x-p = /(5y+7x-q), 7y+4x-p = g(5y+3x-q)

2 52 2 2 52 2
Sol.: z = 2x + 3xy+ or z =2x +7xy+
2y -2ax -(3x+¢i(y+ax)
2y +2yx -8x+q> 2 (y-yx)
I. I.: 2y + 2x + p = f (2y + 4x + q), 4y + 2x + p = g(2y + 2x + q)

2 2 2 2 2
Sol.: z =ax +(3x-(x+y) +¢ 1 (y+ax) or z = -yx +8x-x -4xy-y +¢ 2 (y+yx)
2
45. 3r-6s+4t-(rt-s) "'3
2 3 2 ,-1,
Sol.: 2x +3xy+
I.I.: 3y-'-4x-p = f(3y+3x-q). z
2y +(3x+'f'(y+ax).
2 3 2 2 2
I. I.: yp+x=f(q+y). Sol.: 6a z 2y -3a y + 6axy+6(3y + ¢(ax+½Y ).
47. xqr-(x+y)s+ypt+xy(rt-s2) 1- pq
I.I.: xp+y = j(yq+x). Sol.: z =ax+ y/a + /3 lnx+ c/J(xay).

Index
Applications, Differential equation,

geometric, 41, 75, 133, 178 1 inear,
physical, 49, 133, 178 partial, first order, 238
Approximation, numerical, 186 partial, higher order, 276
Arbitrary constant, 1, 78, 231 non-homogeneous, partial, reducible, 265
Arbitrary function, 232 non-homogeneous, partial, irreducible, 268
Auxiliary system, 239 non-linear partial, 244, 280
numerical solution, 186
Beams, 134 partial, first order, 238
Bernoulli's equation, 35 partial, higher order,
Bessel equation, 222 constant coefficients, 255, 265
Bessel functions, 222 variable coefficients, 276
solution in series, 197
Cauchy linear equation, 108 systems of, 157
C-discriminant, 69 total, 164
Characteristic equation, Direction field, 8
complex roots, 83
distinct real roots, 83 Electric circuits, 57, 136
repeated roots, 83 Exact equations, 12, 24, 123
Charpit's method, 247 Existence theorem, 7
Clairaut equation, 62 Extended Clairaut equation, 246
Complementary function, 79, 257, 266 Extraneous loci, 68
Complete solution,
of ordinary differential equation, 79 First derivative method, 187
of partial differential equation, 240, 244 Functions,
Conditions, Bessel, 222
for exactness, 24, 165 complementary, 79, 257, 266
for integrability, 164 homogeneous, 15
for linear independence, 78
Gauss equation, 223
Damping factor, 134 General solution,
Differential equation, of ordinary differential equation, 7
Bernoulli, 35 of partial differential equation, 238, 256, 265
Bessel, 222
Clairaut, 62 Harmonic motion, 134
exact, 12, 24, 123 Homogeneous equation, 15
extended Clairaut, 246 Homogeneous function, 15
first order, first degree, 12, 24, 35 Homogeneous linear equation, 78, 82, 255
first order, higher degree, 61 Hooke's law, 55
Gauss, 223 Hypergeometric series, 223
homogeneous ordinary, 15, 78, 82
homogeneous partial, 255 Indicial equation, 208
Legendre, 220 Infinite series, 197
Legendre 1 inear, 108 Integral curve, 7
1 inear, Integrating factor, 12, 24
of order one, 13, 35 Intermediate integral, 280
of order n, 78, 123
of order two, 111 Kutta's Simpson's method, 188
INDE\.
Lagrange system, 239 Particular solution, 7

Laplace's transformation, 278 p-discriminant, 69
Legendre equation, 220 Picard's method, 186
Legendre linear equation, 108 Primitive, 1
Legendre polynomial, 221
Lirwar ordinary differential equation, Recursion formula, 198
of orrler one, 13, 35 Reduction of order, 122
ot order n, 78 Regular singular point, 206
with known particular integral, 123 Runge's method, 188
Linear partial differential equation,
238, 276 Separation of variable, 15
Loci, extraneous, 68 Series,
hypergeometric, 223
Monge's method, 280 solution in, 197
Taylor, 187
Newton"s law of cooling, 51 Short methods,
Newton's second law of motion, 49 ordinary differential equation, 99
Non-homogeneous linear partial differential partial differential equation, 266
equation, Simultaneous equations, 157
reducible, 265 Singular point, 199
irreducible, 268 Singular solution, 7, 67, 244
Non-linear partial differential equation, Solutions,
244, 280 complete, 79, 240, 244
general, 7, 238
Operators, factorization of, 112 in series, 197
Order, particular, 7
of differential equation, singular, 7, 67, 244
reduction of, 122 Springs, 140
Origin, Systems of equations, 157
of ordinary differential equation, 1
of partial differential equation, 231 Taylor series, 187
Orthogonal trajectories, 43 Total differential equation, 164
Trajectories, 43
Parameters, variation of, 93
Undetermined coefficients, method of, 93
Partial differential equations, 1, 231
Partial fractions, method of, 88 Variables separable, 13, 15
Particular integral, 79, 257, 266 Variation of parameters, 93
~- --~- ---·-- - - - - ------- -- ----- ------ -~----- -------- ----1
SCHAUM'S OUTLINE SERIES
--- --- ----- - ----- - -- ----- - -- -- --- -- --- --- -- ---- -- -----------<
COLLEGE PHYSICS LAPLACE TRANSFORMS

including 625 SOLVED PROBLEMS including 450 SOLVED PROBLEMS
Edited by CAREL W. van der ME RWE, PhD. By MURRAY R. SPtEGEl, Ph.D.,
P,-ofesrnr of PhYs•cs, New Yo1k Un1ver1 ty 1 P,olessor of Mafh , Rensse, acr Polrtelh 1
Inst
COLLEGE CHEMISTRY DESCRIPTIVE GEOMETRY

including 385 SOLVED PROBLEMS including 17 5 SOLVED PROBLEMS
Edited by JEROME L, ROSENBERG. PhD., By MINOR C. HAWK. f,cod of
Professor ol C~emi~try, LJn,ver:.,ty of p,ttsbur:]h En91ricer1n9Groph,csDcpt Co,r'eJre fn~I a/Tech
First Yr_ COLLEGE MATHEMATICS ~ENGINEERING MECHANICS

including 1 850 SOLVED PROBLEMS including 460 SOLVED PROBLEMS
By FRANK AYRES, J,, Ph D., By W, G. McLEAN, B in EE., MS
ProltalOr ,,/ \ 1otherr>ol,:s, 0 rk,,150,1 Co 1e_lt' P, ofc,sor o' Lalo retie Co 1,'ege-
and E.W. NELSON.BS. in ME., M Adm. E,
COLLEGE ALGEBRA En91r1cer,nJ Superv, or, 0 Western Electrrc Co
inci°ud1ng 1940 SOLVED PROBLEMS THEORETICAL MECHANICS
By MURRAY R. SPIEGEL. PhD
including 720 SOLVED PROBtEMS
P,olencr uf ,'l,c:+f, . Rer:,:elaer Po ,te:~ 1
r,nf
By MURRAY R, SPIEGEL, Ph.D.,
TRIGONOMETRY P,0/p 01 Re,1s1clc,er Po,'rtcch lnsf
including 680 SOLVED PROBLEMS LAGRANGIAN DYNAMICS

By FRANK AYRES, J1., PhD., including 275 SOLVED PROBLEMS
p, o'"" ,, o' .1:rl' cmnf D Co,1re,H'
1
By D. A WEliS. PhD .. -
MATHEMATICS OF FINANCE P-o 1c,_01 ol Pi•ys,,,, Ura \IC';;,ty of C nc n,,nt,
i11cluding 500 SOLVED PROBLEMS STRENGTH OF MATERIALS

By FRANK AYRES PhD., 1nclud1ng 430 SOLVED PROBLEMS
Pro:t,'Or of D ,~,n,011 Co,','e:3e
By WILLIAM A Ph.D
STATISTICS Prol/Js 01 of E,·9 Lfn,"ve'S 11 of F!.Jr,do
including 875 SOLVED PROBLEMS ELECTRIC CIRCUITS

By MURRAY R SPIEGEL, Ph.D. including 350 SOLVED PROBLEMS
P,ofcs,o, o 1 '.1 :.1, Rc,1c"e'oe1 Po 1ytr-,h f,,\/
By JOSEPH A EO:!INISTER, M S.E.E
ANALYTIC GEOMETRY A'soc:: r: ot c' ~ cc f.-,) U•1•ve,1 rv of A'ron
including 345 SOtVED PROBLEMS ELECTRONIC CIRCUITS

By JOSEPH H KINDLE Ph.D including 160 SOLVED PROBLEMS
l-\o 1 (, ,0, ol l,'oti't<,T','' :a, L.),1,,(,'\ t1 ol (" ,le ,1,1,1•
By EDWIN C LOWENBERG. PhD.,
CALCULUS P'o•,::s,c, o' f,._,, f 1 0; U,, ,e,,•t;, of
including 1175 SOLVED PROBLEMS FEEOBACK & CONTROL SYSTEMS

By FRA~JK AYRES, Jr, Pl-, D including 680 SOlVED PROBtEM5
p, J', 0' \' ,, 1,, , ,-.• \ Q, k. (o/ll' iC
By J. J. D;STEFANO Ill, A. R. STUBBERUD.
DIFFERENTIAL EQUATIONS and 1. J. WILLIAMS Pt,.D
including 560 SOLVED PROBLEMS f 1>J '('i;,r 'C; De,J' Ur ,'(.') tr of (11 L A
By FRANK AYRES Jr ,PhD FLUID MECHANICS and HYDRAULICS

01 ,,,,r1"'(')1(ll D :~' lu,','c· 1 ,l
including 475 SotVED PROBLEMS
SET THEORY and Related Topics By RMJALD V. GILES, B.S MS in CE
P,ol of C v,1 E ·,i ,1ee1 Drexel :rasl of Terh
including 530 SOLVED PROBLEMS
By SEYMOUR LIPSCHUTZ, Ph D ..
A i- Pru 1 of ','otl,, itn1p,'c' U1, ,l, ,r,
FLUID DYNAMICS
FINITE MATHEMATICS Wit LIAM F HUGHtS, Ph.D
including 750 SOLVED PROBLEMS C' 8+ E"'' CC11 ,e:i·e o 1 Te,i·
and JOHN A BRIGHTOt, pI, D
By SEYMOUR L!PSCHUTZ. Ph.D .. A,'' P, o• of t,1c.:•1 r,, j
A, o: Pio' o1 Tc•'"T1c•,'e lJ,, Vl•
REINFORCED CONCRETE DESIGN

MODERN ALGEBRA including 200 SOLVED PROBLEMS
c, J. EVERARD, MSCE, PhD.,
By FRANK AYRES. Jr .. Ph.D., 01 t , , tr. ~1 A' '-Jro, .Store Col
Pro 1 eo'.O' ol ',k''•cc",",c, D,ck c,-',1lc.JC and J. l. TANNER ill, MSCE.
MATRICES Tl ,I, Cl' (on , r,1,ir T\',-,ClS fn,,L.,,lr •foS f ,,c
including 340 SOLVED PROBLEMS MECHANICAL VIBRATIONS

By FRANK AYRES, Jr, Ph.D, including 225 SOtVED PROBLEMS
Pr0lt:1sor ol A1~,t!1ernot 1c' D ::~111,0,1 Co,','l,,e
By WIILIAM W. SETO, B.S 111 M.E M.S.,
PROJECTIVE GEOMETRY A, o( Pro 1 o 1 \',:c/-, f,,c;,1, Sci,, ,oie Sro1<' Co,','cgc
including 200 SOLVED PROBLEMS MACHINE DESIGN

by FRANK including 320 SOLVED PROBLEMS
P,ofessor rJ c~ "'so,1 Col'ce,;e
By HALL, HOLOWENKO. LAUGHLIN
GENERAL TOPOLOGY P, r 1, l, ,..,, \' cil F,, I Pur,:lJP U,,,,cn,t,
including 650 SOLVED PROBLEMS BASIC ENGINEERING EQUATIONS

By SEYMOUR LIPSCHUTZ. PhD, including 1400 BASIC EQUATIONS
A,1oc Pio/ ri· \1,ir11 , Tcr•ple u,, ve o,r,
W F. HUGHES, E W. GAYLORD. Ph.D.
of Muh f•1(1 (u,ne9,e 1n5t of Tech
VECTOR ANALYSIS
including 480 SOLVED PROBLEMS ELEMENTARY ALGEBRA
By MURRAY R SPIEGEL, Ph.D., including 2700 SOLVED PROBLEMS
Pro 1e,1or of A',atr' Re,'sSe,'aer Po',t,:,-,/-, r,1-f By BARl'JETT RICH, PhD, Broo~r,n Tech HS
ADVANCED CALCULUS PLANE GEOMETRY

including 925 SOLVED PROBLEMS including 850 SOLVED PROBtEMS
By MURRAY R SPIEGEL, PhD., By BAR,JUT RICH. PhD,
Professor al A1arh I
Rensse,'ae, Polvtec'i lost r~e,1: <1/ f.',ot~ Dept 8100~1:,ri Tech HS
COMPLEX VARIABLES TEST ITEMS IN EDUCATION

including 640 SOLVED PROBLEMS 1nclurl1ng 31 00 TEST ITEMS
By MURRAY R SPIEGEL, PhD, 8yG.JMOUlY,PhD, l.E.WALTON,Ph.D.,
Professor of Nolh. 1 Rens,e'ce, Polylech. Ins/ ProlL'',,Or c of L ,,;ot,un Un,verS,ly of .J,1 ,_Jmr

Jr. Frank Ayres-Theory and Problems of Differential Equations Including 560 Solved Problems-Schaum Publishing

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Jr. Frank Ayres-Theory and Problems of Differential Equations Including 560 Solved Problems-Schaum Publishing

Uploaded by

Copyright:

Available Formats

of

SCHAUM'5 OUTLINE SERIES

THEORY AND PROBLEMS

FRANK AYRES. JR .. Ph.D.

SC::HAU~l'S OUTf,.fNE SElllES

This hook is designed primarily to supplement standard texts in elementary

The author is pltased to aeknowledge his indehted1ws~ to Mr. Louis Sand-

FRANK AYRES. JR.

2. SOLUTIONS OF DIFFERENTIAL EQUATIONS ..................................................................... 7

3. EQUATIONS OF FIRST ORDER AND FIRST DEGREE ······················· .. ············ ...- - 12

,i. EQUATIONS oF FrnsT ORDER AND FmsT DEGREE - - \ ARIABLEs SF.PA-

;). EQUATIONS OF FIRsT ORDER AND FIRST DEGREE~ ExACT EQUATIONS

AND REDUCTION TO EXACT EQUATIONS ·····•······························································· 24

9. EQllATIONS OF FmsT ORDER AND H1GHER DEGREE ................................................ 61

1 O. SINGULAR SoLUTIONS -~ ExTRANEous Loci ............................................................... 67

11. APPLICATIONS OF FIRST ORDER AND HIGHER DEGREE EQUATIONS............... 7.5

12. I,INEAR EQtiATIONS Of' ORDER n .......................................................................................... 78

13. HoMOGENEoLs LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS......... 82

14. l ,1NEAR EQlJ ATION s WITH CoN STANT CoEl'FICIEN TS ____ 87

1:-:i. LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS - VARIATION OF

] 6. LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS - SHORT

l 7. LINEAR EQUATIONS, WITH VARIABLE COEFFICIENTS - THE CAUCHY

AND LEGENDRE LINEAR EQUATIONS ·····················-- ------- 108

18. LINEAR EQUATIONS WITH VARIABLE COEFFICIENTS - EQUATIONS OF

19. LINEAR EQUATIONS WITH VARIABLE COEFFICIENTS~ MISCELLANEOUS

20. APPLICATIONS OF LINEAR EQUATIONS .............................................................................. I 3:1

2]. SYsTE\IS OF S11\lULTANEous LINEAR EQUATIO'.'ls ...................................................... I :;7

22. TOTAL DIFFERENTIAi. EQLATIONS ....................................................................................... 16 l

2.1. APPLICATIONS OF TOTAi. AND SIMlJI.TANEOlJS EQl 1ATIO"IS .............................. 178

24. NUMERICAL APPROXIMATIO'.'ls TO SoLllTIONs ............................................................ 186

2S. INTEGRATION IN SERIES 197

26. Il'JTEGRATION IN SERIES 206

27. THE LEGEc"IDRE. BESSEL. AND GAl ss EQllATIO"IS ................................................... 220

28. PARTIAL DIFFERENTIAL EQUATIONS .................................................................................... 2.11

29. LINEAR PARTIAL DIFFERENTIAL EQCATIONS OF ORDER ONE ........................ 238

30. NON-LINEAR PARTIAL DIFFERENTIAL EQUATIONS OF ORDER ONE ............... 2H

31. HOMOGENEOUS PARTIAL DIFFERENTIAi. EQUATIONS OF HIGIIER ORDER

32. No!'<-Ho,10GENEOLS LINEAR EQUATIONS WITH CoNSTA'\/T CoEFFI-

3:3. PARTIAL DIFFERENTIAL EQLATIONS OF ORDER T,Yo WITH VARIABLE

INDEX ..................................................................................................................................................................... 29S

Origin of Differential Equations

A DIFFERENfIAL EQUATION is an equation which involves derivatives. f'or example,

If there is a single independent variable, as in l)-5), the derivatives are

ORIGIN OF DIFFERENfIAL EQUATIONS.

5. In each of the equations a) y = x 2 +A+B, b) y =Aex+B, c) y =A+ lnBx showthatonly one

6. Obtain the differential equation associated with the primitive 2

7. Obtain the differential equation associated with the primitive x y 3 x 3y 5

or, when xy 1 0, (2y + 3x dy) + x/ (3y + 5x dy) = O as the required equation.

and 2) (2ydx + 3xdy) + x/(3ydx + 5xdy) = O.

The required differential equation is o.

The elimination of the constants by elementary methods is somewhat tedious. If three of

The required differential equation is 11 dy - 6y o.

dy C = 1 dy y = Cx2 + C2 = _!_dy x2 + _l_(dY/.

The equation of the family is (x - c/ + y2 2

The equation of the family of parabolas is y 2 = 4A(A + x).

The required equation is

15. Classify each of the following equations as to order and degree.

a) dy + (xy - cos x)dx = 0 Ans. Order one; degree one

f) ey "' Ans. Order three; degree does not apply

g) ~ = sine Ans, Order one; degree one

h) y' + X = (Y -xy,) -3 Ans. Order one; degree four

d) The potential difference E across an element of inductance L is equal to the product of L

22, Find the differential equation of all circles in the plane.

Solutions of Differential Equations

The integrating factor _l_ , found by inspection, reduces 1) to the