Lecture On Coefficient of Variables
Lecture On Coefficient of Variables
Rainville
University of Michigan
Phillip E. Bedient
Professor of Mathematics
It is the belief of the present author that the success of the previous editions
of book has been largely due to Professor Rainville's ability to write in
this
a style that was readily accessible to the serious student. The present edition
attempts to maintain that accessibility while making some changes in content
and emphasis. In particular, the chapter on existence theorems has been
expanded to include a proof of an existence and uniqueness theorem.
As in earlier editions, I have attempted to exhibit both the techniques for
obtaining solutions and the basic ideas and theories behind these techniques.
A number of exercises of a more theoretical nature have been added to assist
in relating these techniques with the theory.
The omission of infinite series methods has made possible, even within the
bounds of a small book, a thorough treatment of topics essential to a first
course. A longer book, Elementary Differential Equations, Fourth Edition,
contains an extensive treatment of both power series and Fourier series
methods and an introduction to the application of the Laplace transform tech-
nique to boundary value problems involving partial differential equations.
vi Preface
The author is indebted to many persons who have influenced his thinking
in many ways. The most profound influence has come from the late Professor
Rainville, whose reputation as a superb teacher Hves in the minds of all his
former students. Especial thanks are also given to Esther Rainville for her
interest and encouragement.
I am grateful for the many comments and suggestions made by my students
and colleagues at Franklin and Marshall College. 1 am pleased to acknowl-
edge in particular Professor Paul W. Berg of Stanford University and Pro-
fessor Howard Sherwood of Illinois Institute of Technology for their very
valuable suggestions.
Phillip E. Bedient
Lancaster, Pennsylvania
Contents
33. Introduction 97
34. The auxiliary equation; distinct roots 97
Contents ix
CHAPTER 12 Applications
Index 275
A Short Course in Dilferential Equations
CHAPTER ±
Definitions, Elimination of
Arbitrary Constants
tial equations; that is, to find the unknown function, or functions, which
occur.
Differential equations arise frequently in physics, engineering, and chemis-
try, and on occasion in such subjects as biology, physiology, and economics.
The solution of differential equations plays an important role in the study of
the motions of heavenly bodies such as planets, moons, and artificial satellites.
The following are examples of differential equations:
dy
-— = cos X, (1)
ax
+ k'y = 0, (2)
1?
Definitions, Elimination of Arbitrary Constants [Ch. 1
; cut, (5)
(6)
(7)
(8)
(9)
(10)
(11)
L
d^i
—
dt
+5 i? — —C
di
dt
I
1
i = Eco cos cot (5)
may be written
§ 2] Definitions
ax
or
-, T dx
dy
we may consider either variable to be dependent, the other being the inde-
pendent one.
Oral Exercise
2. Definitions
^ + 2h[^] +y = () (1)
dx^ \dx
equation."
More generally, the equation
/"'=f{x,y,y',...,/"-'\ (3)
For the purposes of book we shall assume that this is always possible.
this
differential equation (3) providing the n derivatives of the function exist on the
y = e^^
dx^
^ + ^-6y
dx
= 0. (4)
We substitute our tentative solution into the left member of equation (4)
and find that
^+^_
dx dx
6j = 4e^^ + 2e'^ - 6e'^ = 0,
All of the equations we shall consider in Chapter 2 are of order one, and
hence may be written
dx
Fix,y,y',...,/"^)^0
is called linear if the function F is a linear function of the variables y, y', . .., y^"\
For example, equation (1) above is nonlinear, and equation (4) is linear.
The equation
x^y" + xy' + (a-- - n^)y = 4x^
is also linear. The manner in which the independent variable enters the
equation has nothing to do with the property of linearity.
§ 3] The elimination of arbitrary constants 5
Oral Exercises
For each of the following, state whether the equation is ordinary or partial,
linear or nonlinear, and give its order.
1.
Definitions, Elimination of Arbitrary Constants [Ch. 1
y= c^e~^''^C2e^''. (1)
/' + 2/ = 15c2e3";
Hence
y" + 2y' = 3(>'' + 2y),
or
e'
-/
-y"
§ 3] The elimination of arbitrary constants 7
d"y d"-^y dy
''d?^'^d^^---^'"-^d-x^'"' =
''^
in which the coefficients Oq, a^,..., a„ are constants. The study of such
differential equations will receive much of our attention.
(jc - af + y^ = a^.
2{x — a) + 2yy' = 0,
from which
a = X + yy'.
Therefore, using the original equation, we find that
>'^ = x^ + lxyy\
{x - af + y^ = a^
x^ + y^ - lax = 0,
or
— la.
X
or
as desired.
It is interesting to speculate here about the significance of a- = upon the
argument just used. The student should draw a few of the members of this
family of circles and ask himself what is peculiar about their behavior at
x = 0.
—
dx
dt
= —0)B sin {cot + a), (6)
—
d^x
dt-
^ = - orB cos (cot + a). (7)
-— + tolv - 0.
dt^
cxy + c^x + 4 = 0.
At once we get
c(y + xy') + c^ = 0.
Since c ^ 0,
c= -iy + xy')
x^iy'f + x\vy' + 4 = 0.
Exercises
In each of the following, eliminate the arbitrary constants.
§ 4] Families of curves 9
1. — 7>x^y = c.
x^ ANS. (x — 2y) dx — x dy = 0.
2. — xy^ = c.
y sinx ans. y(cos x — y) dx + (sin x — 2xy) dy = 0.
3. PV = C.
4. Families of curves
or
FIGURE 1
center on the line y = x and each passing through the origin. Figure 1 shows
several elements, or members, of this family.
constant and eliminated as in the preceding section, the result is called the
= lc
X +y
we find that
(4)
dx — Ixy — y^
§4] Families of curves 11
or
y = (-l-v/2)x
y = {-l + s^)x
FIGURE 2
In Figure 2, the straight lines (5) and (6) appear along with the family (1).
It is seen that the lines (5) and (6) cut the members of the family of circles in
precisely those points of vertical tangency.
For a two-parameter family of curves, the differential equation will be of
order two, and such a simple geometric interpretation is not available.
FIGURE 3
x^ = Aay. (7)
Then from
= 4a
2y dx - X dy = 0, (9)
Since a member of the family of circles of this example may have its
center anywhere on the j-axis and its radius of any magnitude, we are
dealing with the two-parameter family
x^ + (y- bf = /'
(10)
At once
x + (y- b)y' = 0,
from which
x + yy'
= b.
y
Then
FIGURE 4
14 Definitions, Elimination of Arbitrary Constants [Ch. 1
Exercises
In each exercise, obtain the differential equation of the family of plane curves
described and sketch several representative members of the family.
2. Straight lines through the fixed point (h, k). The h and k are not to be eliminated.
ANS. {y — k)dx — {x — h) dy = 0.
3. Straight lines with slope and >'-intercept equal.
ANS. y dx —{x + 1) dy = 0.
4. Straight lines with slope and x-intercept equal. ans. {y'Y = xy' —y.
5. Straight lines with algebraic sum of the intercepts fixed as k.
ANS. {xy' — y){y' — \) + ky' = 0.
6. Straight lines at a fixed distance p from the origin.
ANS. {xy' - yY = p^{\ + {yyi
7. Circles with center at the origin. ans. xdx + y dy = Q.
8. Circles with center on the x-axis. ANS. yy" + {y')'^ +1=0.
9. Circles with fixed radius r and tangent to the x-axis.
ANS. {y±ry{y'y + y^±2ry = 0.
10. Circles tangent to the x-axis. ans. [1 + {yVY = \yy" + + {y'YY.
1
11. Circles with center on the line y = —x, and passing through the origin.
ANS. {x^ — Ixy — y^) dx + {x^ + Ixy — y^) dy = 0.
12. Circles of radius unity. Use the fact that the radius of curvature is one.
ANS. {yy = u+{y'rr.
13. All circles. Use the curvature. ans. y"'[l + {y'Y] = 3y'{y")^.
14. Parabolas with vertex on the x-axis, with axis parallel to the j-axis, and with
distance from focus to vertex fixed as a. ans. a{y'y = y.
15. Parabolas with vertex on the j-axis, with axis parallel to the x-axis, and with
distance from focus to vertex fixed as a. ans. x{y')^ = a.
16. Parabolas with axis parallel to the j-axis and with distance from vertex to focus
fixed as a. ans. lay" — 1
17. Parabolas with axis parallel to the x-axis and with distance from vertex to focus
fixed as a. ans. lay" + {y'Y = 0.
d^x
18. Work Ex. 17, using differentiation with respect to y. ans. la —^ = 1.
dy^
19. Use the fact that
x^ y^
a^ + X'^ b^ + X^^
§ 4] Families of curves 15
31. Circles through the intersections of the circle x^ y^ = -\-and the line y = x.
\
x^' + y^ -\+k{y-x) = 0.
ANS. (x^ -2xy-y^+\)dx + {x^ + 2a:j ->'2 -\)dy== 0.
32. Circles through the fixed points {a, 0) and {—a, 0). Use the method of Ex. 31.
ANS. 2xy dx + {y^ + a^ — x^) dy = 0.
33. The circles r == 2o(sin 9 — cos 9).
37. The trisectrices of Maclaurin r = a(4 cos 9 — sec 9). (See Ex. 30.)
ANS. cos 9(4 cos^ 9 -~l)dr + r sin ^(4 cos^ ^+1)^/^ = 0.
CHAPTER ^
Equations of Order One
In this chapter we shall study several elementary methods for solving first-
^=f(x,y). (1)
ax
a region with the point {xq , jfo) at its center. Suppose that / and df/dy are
continuous functions of x and y in T.
16
§ 6] Separation of variables 17
only function that has all of the properties (a), (b), and (c).
The interval \x — Xq\ ^h may or may not need to be smaller than the inter-
val \x — Xq\ ^a over which conditions were imposed upon/(jic, y).
In rough language, the theorem states that iff(x, y) is sufficiently well be-
g =/(...) (1)
has a solution that passes through the point (xq , >'o) ^^^ that solution is
6. Separation of variables
Mdx + Ndy = 0,
where M and N may be functions of both x and y. Some equations of this
type are so simple that they can be put in the form
that is, the variables can be separated. Then a solution can be written at once.
For it is only a matter of finding a function F whose total differential is the
dy 2y
-r = —, for x>0 and v > 0. (2)
dx X
We note that for the function in equation (2), the theorem of the previous
section applies and assures the existence of a unique continuous solution
through any point in the first quadrant. By separating the variables we can
write
18 Equations of Order One [Ch. 2
dy 2 dx
y X
Hence we obtain a family of solutions
In Ij'I
= 2ln\x\ + c (3)
y = e'^x^. (4)
written
y = c2x\ (7)
where Cj is taken to be an arbitrary real number. Indeed this form for the
solutions incorporates the special case y = 0. Thus we say that the family of
curves given in Figure 5 represents a family of solution curves for the differ-
ential equation (2).
FIGURE 5
§ 6] Separation of variables 19
dx dy
1+x^ \^-y^
In the set of solutions (9), each "Arctan " stands for the principal value of
the inverse tangent and is subject to the restriction
Suppose next that we wish to sketch the graph of (10). Resorting to a device
of trigonometry, we take the tangent of each side of (10). Since
tan (Arctan x) =x
and
tan A -\- tanB
tan {A-\-B) =
1 — tan A tan B
20 Equations of Order One [Ch. 2
X +y
= -1,
I — xy
or
xy-x-y-\=0. (11)
Hence x < 1, and equation (10) represents only one branch of the hyperbola
(1 1). In Figure 6, the solid curve is the graph of equation (10); the solid curve
and the dotted curve together are the graph of equation (11).
^ = — 1 when X = 0.
One distinction between the equations (10) and (11) can be seen by noting
that a computing machine, given the differential equation (8) and seeking a
FIGURE 6
§ 7] The exp u notation 21
solution that passes through the point (0, —1), would draw only the left
2x dx
-('-JTl)'"-
y
'*-'
Integrating, we get a family of solutions given implicitly by
x^=y-\n\y+\\+c. (13)
Since we seek a member of this family that passes through the point
(0, — 2), we must have
0= -2-ln|-l| + c,
or
c = 2.
x^ =y-\n\y+\\+2.
The reader should note how the theorem of Section 5 applies to this prob-
lem to indicate that we have found implicitly the unique solution to the
initial value problem which is continuous for 7 < — 1.
The function e" enters our work frequently, and sometimes the exponent u
is complicated. Even a simple example such as e^ ^^
may be undesirable in
print, because exponents are usually set in small type. It is customary in ad-
vanced mathematics to use the notation
That is, exp is used to denote the exponential function in much the same way
that sin is used to denote the sine function.
Most people retain the notation e" when u simple, even though in the same
equation the exp notation may have been used elsewhere. Equation (2) is
A few minutes of practice should make the reader feel at home with the
symbol exp u. Some familiarity with this notation will result from verification
exp (x) •
exp (y) = exp {x + y), (4)
expg)expg)=exp(^), (7)
We shall use the exp notation whenever it seems to add to the clarity of
Exercises
In Exs. 1-26, obtain a family of solutions and indicate for what values of x they
are valid.
7. —
dP
=
P
. ANS. PV = C.
8. e%y - + 2(e^ + A) dy = 0.
1) Jx ANS. {y - + 4) = c^
\Y{e'
9. dr = e(r sin- cos 6 dr).
d dd ANS. + e cos 6) = c.
/•(!
= a sec y -V
/ c
25. a^ dx =XV JC ^ — a^ dy. ans. jc .
a
16. y\nx\r\y dx + dy =^0. ans. jcln jc + In |in>'| =x + c.
In Exs. 27-33, obtain the particular solution satisfying the initial condition
indicated.
dr
11. —= —Art; when / = 0, r = ro ans. r = ro exp (—2/^).
28. 2x7/ = + >'^ when x = 2, 7 = 3.
1 ; ans. 7^ = 5x — 1.
29. xyy' = + 7^ when x = 2, 7 = 3.
1 ; ans. 5x^ — 27^ = 2.
30. / =xexp(jv — x^); when X = 0, 7 = 0. ans. 2e"* = + exp (— x^).
1
8. Homogeneous functions
x^ — 3x7 + 4j^,
x^ + y\ (1)
x'^y + ly^,
are called homogeneous polynomials. We wish now to extend that concept of
homogeneity so it will apply to functions other than polynomials.
If we assign a physical dimension, say length, to each variable x and y in
the polynomials in (1), then each polynomial itself also has a physical dimen-
sion, length to some power. This suggests the desired generalization, If, when
certain variables are thought of as lengths, a function has physical dimension
length to the /:th power, then we shall call that function homogeneous of
24 Equations of Order One [Ch. 2
Vx^+/
is homogeneous of degree zero in x and y.
vAx/ Xx + 3A_y
f{Xx,Xy) = X^f{x,y);
Theorem 1 : If M(x, y) and N(x, y) are both homogeneous and of the same
degree, the function M(x, y)lN(x, y) is homogeneous of degree zero.
in which the x is now playing the role taken by X in the definition (3) above.
By (4), fix, y) depends on v alone as stated in Theorem 2.
§ 9] Equations with homogeneous coefficients 25
Oral Exercises
Determine in each exercise whether the function is homogeneous or not. If it is
7. exp
&
26 Equations of Order One [Ch. 2
Since the coefficients in (4) are both homogeneous and of degree two in
from which the factor x^ should be removed at once. That done, we have to
solve
(1 - V -{ v^) dx — v{v dx + X dv) = 0,
or
(1 — v) dx — XV dv = 0.
—
dx
X
+
V
V dv
—
-
I
= 0.
Then from
dx
—X
+ 1 +
V-
1
1
dv =
or
X{V -])€" = C.
- '^p '•
'(x ') (9
or
(y - x) exp f-j = c.
Again the coefficients in the equation are homogeneous and of degree two.
We could use y — vx, but the relative simplicity of the dx term in (5) suggests
that we put
X= vy.
or
vy dv + {2v^ + \) dy = 0, (6)
or
y\2v^ + l) = c.
(7-)-^
y
that is,
Since the left member of equation (7) cannot be negative, we may, for sym-
metry's sake, change the arbitrary constant to c^*, writing
It is worth-while for the student to attack equation (5) using y = vx. That
method leads directly to the equation
Exercises
In Exs. 1-19 obtain a family of solutions.
1. (x 2y) + (2x + y) dy = 0,
dx ANS. In (jc^ + >'^) + 4 Arctan {yjx) = c.
2. 2{2x^ dx - xy dy = 0.
-V y"") ans. x'^ = c''{Ax'' + y^).
3. xy dx — {x^ + 3>'^) dy = 0. ans. x^ = 6y^ In \ylc\.
4. x^y' = 4x^ + Ixy + 2y^. ans. x^(y + 2x) = c^y + x).
5. 3xy dx + (x^ + y^) dy = 0.
6. {x — y)(4x -^y)dx + xi5x —y)dy^ 0. ans. x(y + xY =c(y — 2x).
28 Equations of Order One [Ch. 2
the form
ANS. 4(2^ + x) In J = 2^ — X.
27. xy dx + 2(x^ + 2y^) dy = 0; when x = 0, = ans. >'*(3x^ + 4y^) = 4.
3; 1
34. From Theorems 1 and 2, page 24, it follows that if F is homogeneous of degree
km X and y, F can be written in the form
(A)
=-K9-
Use (A) to prove Euler's theorem that if F is a homogeneous function of degree
kinx and y,
dF 8F
In Section 6 it was noted that when an equation can be put in the form
A(x) dx + B{y) dy = 0,
Then certainly
F{x,y) = c (3)
dF=0,
or, in view of (2),
Mdx + Ndy = 0,
as desired.
Two things, then, are needed: first, to find out under what conditions on
M and A'^ a function F exists such that its total differential is exactly M dx +
N dy; second, if those conditions are satisfied, actually to determine the func-
tion F. If there exists a function F such that
Mdx + Ndy
is exactly the total differential of F, we call equation (1) an exact equation.
30 Equations of Order One [Ch. 2
If the equation
aF df
dF = ^
ox
dx + ^
,
dy
,
dy,
so
_8F _8F
dx' dy
dM d^F
dy dy dx
and
dN _ d^F
dx dx dy
d^F d^F
dy dx dx dy
— =—
dM dN
dy dx
• (4)
d^_
dx
dy dx dy
§ 10] Exact equations 31
-^ =— (5)
dx dy dx
Let us integrate both sides of this last equation with respect to x, holding y
fixed. In the integration with respect to the x, the "arbitrary constant" may
be any function of y. Let us call it B'(y), for ease in indicating its integral.
dS
-f
= N + B'(y). 6)
8y
F=(l>(x,y)-B(y),
for which
dd) d(j)
dF = ^dx + ^dy- B'(y) dy
dx dy
= M dx + N dy.
Hence, equation (1) is exact. We have completed a proof of the theorem
stated below.
dM _ dN
dy dx
Furthermore, the proof contains the germ of a method for obtaining a set
dM
— ^^ =
—= ^ 2
Sx-^ and^ -— . 2
Sx"^,
dy dx
32 Equations of Order One [Ch. 2
dF
—
ex
= M = 3x^y - 6x, (8)
and
8F
—
dy
= N = x' + 2y. (9)
x^ + T'(y) = x^ + ly,
T'iy) = 2y.
ny) = y\
and from (10)
Here
dM dN
dF
— = 2x^- xy^ -2y + 3 (12)
§ 10] Exact equations 33
and
dF
— =-x'y-2x. (13)
Because (13) is simpler than (12), and for variety's sake, let us start the
determination of F from equation (13).
At once, from ( 1
3)
F= -^xV-2xj;+ Q(x),
Q'ix) = 2x^ + 3.
Therefore
Qix) = ix^ + 3.V,
Exercises
Test each of the following equations for exactness and solve the equation.
The equations that are not exact may, of course, be solved by methods discussed
in the preceding sections.
which we choose as the standard form for the linear equation of order one.
For the moment, suppose that there exists for equation (2) a positive inte-
grating factor v(x) > 0, a function of x alone. Then
dy (3)
v(x) = v{x) •
Q{x)
dx
must be an exact equation. But (3) is easily put into the form
Mdx + Ndy =
with
M=vPy- vQ,
and
N=v,
in which v, P, and Q are functions of x alone.
Therefore, if equation (3) is to be exact, it follows from the requirement
36 Equations of Order One [Ch. 2
yexpi^Pdx);
the right member of (6) is a function of x only. Hence equation (6) is exact,
exponent (
J P dx) any function whose derivative is P.
Because of the great importance of the ideas just discussed and the frequent
occurrence of linear equations of first order, we summarize the steps involved
dx
Note in integrating the exact equation, that the integral of the left member
is always the product of the dependent variable and the integrating factor used.
2(y - 4x^) dx + X dy = 0.
Next apply the integrating factor to (7), thus obtaining the exact equation
x' ^
dx
+ 2xy = 8JC^ (8)
This can be checked. From (10) we get (8) by difTerentiation. Then the
original differential equation follows from (8) by a simple adjustment. Hence
(10) defines a set of solutions of the original equation.
y dx -V (3a- — xy -\-T)dy = 0.
Since the product y dy occurs here, the equation is not linear in y. It is,
—
dx
dy
/3
+ --1
\y
\
/
x = —
-2
y
, for y¥^0. (11)
Now
j\--ljdy = 3\n\y\-y + c„
It follows that for j^ > 0, y^e~^ is an integrating factor for equation (11) and
for j^ < 0, —y^e~^ serves as an integrating factor. In either case we are led to
the exact equation
xy^e-y= -l^y^-'dy
= 2y^e-' + Aye'" + Ae'" + c.
^
ax
+ P(x)y = Q{x). (1)
number, there exists a unique solution y = y{x) of the differential equation (1)
>'(^o) = J'o •
Moreover, this solution satisfies equation (1) throughout the entire interval
a <X< b.
The proof of this theorem has essentially been obtained in Section 11.
yv = jvQdx + c.
It is also easy to see that given any a'o on the interval a <x< b together
with any number jo we can choose
, the constant c so that j = Vq when x — Xq.
The effect of our argument is that every equation of the form of equation
(1), for which P and Q have some common interval of continuity, will have
is called the general solution of equation (1). The word " general " is intended
to mean that we have found all possible solutions that satisfy the differential
equation on the interval a <x< b.
§ 12] The general solution of a linear equation 39
Exercises
In Exs. 1-25, find the general solution.
1. (x^ + dx — xdy = 0.
3y) = x^ + cx^.
ans. 2y
2. 2{2xy + 4y-3) dx + {x + ly dy = 0. ans. >' = 2(x ++ c(a: + 2)"*.
2)"'
3. y = x — 2y. ANS. 4>' = 2a: — + ce~^'. 1
dy
14. my = Cie""", where Ci and m are constants, ans. y = (cix + C2)e""'.
dx
dy
15. — —m2y = Cie
m X
'
, where Ci, mi, ma are constants and rtii ^mz.
dx
Ci
ANS. y =^ C3 e""^" + Ca e"^*, where C3
— m2
mi
dv + Pvdx =
40 Equations of Order One [Ch. 2
dy + Pydx = Q dx.
In Exs. 27-33 find the particular solution indicated.
ANS. / = — — exp — 1 (
i['-"''(-T!
30. L — + Ri = E sin cjot; when / = 0, = 0. /
dt
32. Find that solution oi y' ^2{2x — y) which passes through the point (0, 1).
ANS. >' = 2x- +2e-2^ 1
Miscellaneous Exercises
In each exercise, find a set of solutions, unless the statement of the exercise
stipulates otherwise.
dy
7. cos X = cos X tan^ y. ans. 2 sin x = + sin
>' >' cos y + c.
dx
8. cos X —=
dy
1 — J — sin X. ans j(1 + sin x) = (x + c) cos x.
Jx
dr
9. sin ^ —= — — 1 2/- cos 6. ans. r sin^ = c + cos ^.
dtf
dy
11. = sec^ X sec' y. ANS. 3 tan X +c= 3 sin J'
— sin' y.
dx
12. yilx^ -x^y-\-y^)dx-x{2x^ +y^)dy=-0. ans. 2x^7 In |cx1 =4a:' ->''.
13. (1 + jc^)>'' = x'^y'^. ANS. x'j'' = y^{c + 3x - 3 Arctan x).
+ 1
(1,-?)
FIGURE 7
42 Equations of Order One [Ch. 2
2S. ydx = (2x + l)idx - dy). ans. 3y = {Ix + 1) -f c{2x + 1) " ''^
In solving Exs. 29-33, recall that the principal value Arcsin x of the inverse sine
function is restricted as follows:
31. Solve the equation of Ex. 29 with the added condition that when x = 0,
y = — ^V3. ANS. Arcsin X + Arcsin >>=— ^77, or that arc of the ellipse
x^ + xy + y^ = J, that is indicated by a light solid line in Figure 7.
32. Show that after the answers to Exs. 30 and 31 have been deleted, the remaining
arcs of the ellipse
x^ + xy + y^ = i
Vl -y^dx + V\ - x^ dy = 0.
For this purpose consider the sign of the slope of the curve.
33. For the equation
Vl -y^dx-Vl -x^dy =
state and solve four problems analogous to Exs. 29-32 above.
34. V = {e" + luv - 2m) dv.
du ans. v^u = ce^" — {v+ \)e\
2>5.ydx = {2>x + y^-y^)dy;v.'\\cnx^\,y=~\. ans. x =^^[1 + 7 In (-j)].
36. y^ dx - {xy + 2)dy = 0. ans. xy = cy^ -I.
37. (x^ - Ixy - y"") dx - {x^ -f Ixy - y^) dy = 0.
ANS. (x + y)(x^ — 4xy + y^) = c^.
38. y^ dx + (xy + y^ - I) dy = 0; when x= -\,y = l.
ANS. y^ + 2xy + = 2 In 1 >'.
44
§ 13] Velocity of escape from the earth 45
from which
gR'
a = — -
We wish to express the acceleration in terms of the velocity and the distance.
dv dr
We have a = — and v^—. Hence
dt dt
dv
a = — = dr dv = —
dv
v
dt dt dr dr
dv
_ _ gR^
^-=--T-
dr r
(1)
IgR^
= -^— +
,
v^ C.
r
Suppose the particle leaves the earth's surface with the velocity ^o • Then
V = t'o when r = R, from which the constant C is easily determined to be
C=Vo'- 2gR.
For, if the inequality (3) is satisfied, the velocity given by equation (2) will
remain positive, since it cannot vanish, is continuous, and is positive at
r = R. On the other hand, if (3) is not satisfied, then Vq^ — 2gR < 0, and there
will be a critical value of r for which the right member of equation (2) is
zero. That is, the particle would stop, the velocity would change from positive
A particle projected from the earth with a velocity Vq such that Vq ^ y/2gR
will escape from the earth. Hence, the minimum such velocity of projection,
r, = ^2gR, (4)
Of course, the gravitational pull of other celestial bodies, the moon, the
sun, Mars, Venus, etc. has been neglected in the idealized problem treated
here. It is not difficult to see that such approximations are justified, since we
are interested in only the critical initial velocity v^. Whether the particle
actually recedes from the earth forever or becomes, for instance, a satellite
It must be realized that the formula v^ = JlgR applies equally well for
the velocity of escape from the other members of the solar system, as long
equation
du
-^-k(u-\0), (1)
and
equation (1).
u = 10 + 60e-'". (4)
The value of k will be determined now by using the condition (3). Putting
/ = 3 and u = 25 into equation (4) we get
25= 10 + 60e--'\
that is, let .y = Xq at / = 0. Then the amount x at any time / > is determined
by the differential equation
A- = Ce-"'.
x^xoe-''. (2)
thirds of the original amount Xq has already been converted. Let us determine
how much unconverted substance remains at / = 60 (sec).
When two-thirds of the substance has been converted, one-third remains
unconverted. Hence a- = ^a'q when t = 30. Equation (2) now yields the
relation
3X0 — Ao^
from which k is easily found to be -j- In 3. Then with t measured in seconds,
x = Xoexpi--^t\n3). (3)
At t = 60,
X = Xq exp (-2 In 3) = .Vo(3)"^ = }xq .
Exercises
1. The radius of the moon is roughly 1080 miles. The acceleration of gravity at
the surface of the moon is about 0. 65g, where g is the acceleration of gravity
at the surface of the earth. Determine the velocity of escape for the moon.
ANS. 1.5 mi. /sec.
§ 15] Simple chemical conversion 49
2. Determine to two significant figures the velocity of escape for each of the
celestial bodies listed below. The data given are rough and g may be taken to
be 6.1(10)-^ mi./sec.^
Venus
50 Elementary Applications [Ch. 3
11. Two substances, A and B, are being converted into a single compound C. In
the laboratory it has been shown that, for these substances, the following law
of conversion holds: the time rate of change of the amount x of compound Cis
proportional to the product of the amounts of unconverted substances A and
B. Assume the units of measure so chosen that one unit of the compound C is
formed from the combination of one unit of A with one unit of B. If at time
/ = there are a units of substance A, b units of substance B, and none of the
compound C, present, show that the law of conversion may be expressed by the
equation
dx
— =k(a-x)(b-x).
Solve this equation with the given initial condition.
ab[e\p {b — d)kt —
U b ^ a, X = —
1]
ANS. -; -; ;'\ib = a,x = a^kt/(akt + 1 ).
b exp {b — a)kt — a
12. In the solution of Ex. 1 1 above, assume that /: > and investigate the behavior
ofjcas/->oo. ANS. \i b^ a, x->a\'\i b -^ a, x-^b.
13. Radium decomposes at a rate proportional to the quantity of radium present.
Suppose that it is found that in 25 years approximately 1.1 per cent of a certain
quantity of radium has decomposed. Determine approximately how long it will
take for one-half the original amount of radium to decompose.
ANS. 1600 years.
14. A certain radioactive substance has a half-life of 38 hours. Find how long it
controls being exerted, should B become 100 times what it was at noon?
ANS. About 8:22 p.m.
16. In the motion of an object through a certain medium (air at certain pressures is
w
—dv
r
gdt
= M' — kv'^,
where kv^ is the magnitude of the resisting force furnished by the medium.
17. Solve the differential equation of Ex. 16 with the initial condition that v — Vo
when t = 0. Introduce the constant a^ = wjk to simplify the formulas.
o J
ANS.
a
a
+ v = a + Vo
— V a — Vo exp I
\
—
/2^^\
I
18. List a consistent set of units for the dimensions of the variables and parameters
of Exs. 16-17 above. ans. / in sec. g in ft./sec.^
V in ft./sec. k in (lb.)(sec.^)/(ft.^)
w in lb. a in ft./sec.
§15] Simple chemical conversion 51
There are mediums that resist motion through them with a force proportional
to the first power of the velocity. For such a medium, state and solve problems
analogous to Exs. 16-18 above, except that for convenience a constant b = wjk
may be introduced to replace the a^ of Ex. 17. Show that b has the dimensions
of a velocity.
— b + KVo —
, ^
ANS. V b) exp
-f
FIGURE
downward under the action of gravity alone. Find how long it will take the
weight to reach the bottom of the board and determine the terminal speed.
ANS. 1.9 sec. and 10.5 ft./sec.
Add to the conditions of Ex. 20 above a retarding force of magnitude kv,
where v is the velocity. Determine v and x under the assumption that the weight
starts from rest with x = Xo. Use the notation a = kg/w.
ANS. y = a~'^sin a(l — e'"'); x =^ Xo + a~^g sin (x.(— I + e'"' + at).
A man, standing at O in Figure 9, holds a rope of length a to which a weight is
attached, initially at P^o • The man walks to the right dragging the weight after
him. When the man is at M, the weight is at W. Find the differential equation
of the path (called the tractrix) of the weight and solve the equation.
a + Va^-,
ANS. a In Va^-
y
24. A tank contains 80 gallons of pure water. A brine solution with 2 pounds of
salt per gallon enters at 2 gallons per minute, and the well-stirred mixture
leaves at the same rate. Find (a) the amount of salt in the tank at any time, and
(b) the time at which the brine leaving will contain 1 pound of salt per gallon.
ANS. (a) s = 160[1 - exp (- ^/40)] (b) ^ = 40 In 2 min. ;
52 Elementary Applications [Ch. 3
FIGURE 9
25. For the tank in the previous problem, determine the limiting value for the
amount of salt in the tank after a long time. How much time must pass before
the amount of salt in the tank reaches 80% of this limiting value?
26. A certain sum of money P draws interest compounded continuously. If at a
certain time there is Po dollars in the account, determine the time when the
principal attains the value IPo dollars, if the annual interest rate is (a) 2%, or
(b) 4 % ANS. (a) 50 In 2 years (b)
;
1 00 In 2 years.
/(.x,j,c) = 0, (1)
g{x,y,k) = Q (2)
such that at any intersection of a curve of the family (2) with a curve of the
family (1) the tangents to the two curves are perpendicular. The families
(1) and (2) are then said to be orthogonal trajectories* of each other.
* The word orthogonal comes from the Greek opd-q (right) and ytuvta (angle); the word
trajectory comes from the Latin trajectus (cut across). Hence a curve that cuts across certain
others at right angles is called an orthogonal trajectory of those others.
§ 16] Orthogonal trajectories; rectangular coordinates 53
slopes of the curves must be negative reciprocals of each other. That fact
leads us to a method for finding orthogonal trajectories of a given family of
curves. First we find the differential equation of the given family. Then
replacing dyjdx by —dxjdy in that equation yields the diftcrential equation
of the orthogonal trajectories to the given curves. It remains only to solve
the latter differential equation.
So far we have solved differential equations of only one form,
Mdx + Ndy = 0.
For such an equation
dy _ M
'dx~ ~~N'
dy _ N
dx M
or
TV dx - M dy = 0.
y^ = 4ax. (3)
Hence, from
— =4fl,
X
2.V dy - y dx = 0. (4)
Therefore the orthogonal trajectories of the family (3) must satisfy the
equation
2x dx + y dy = 0. (5)
are certain ellipses (6) with centers at the origin. See Figure 10.
54 Elementary Applications [Ch. 3
FIGURE 10
Exercises
In each exercise, find the orthogonal trajectories of the given family of curves.
Draw a few representative curves of each family whenever a figure is requested.
x = c exp iy^).
§ 17] Orthogonal trajectories; polar coordinates 55
17. ax^ -\- y^ = 2acx, with a held fixed. ans. l(a^2,{2 — a)x^ -\- y'^ = dy".
If a = 2, x^ = —y^ In {czy).
18. >'(x^ + c) + 2 = 0. ANS. y^=-Z\n\kx\.
19. x" + y" = a", with n held fixed and n^2. ans. x^"" — >'^~" = c.
20. ^'^ = x^{\ — ex). ANS. x^ + 3y^ = Ci>'.
21. y^ = 4x^1 - ex). ANS. 2x^ = 37^(1 - c^y^
22. j'^ = ax^{\ — ex), with a held fixed.
If a = 2, x^ = —3y^ In |c2>'|.
23. y(x^+l) = ex. ans. >'^ - x^' + 2 In |A:(x^ - 1)|.
24. >' = 3x- +ce-3\ 1 ANS. 27x = 9y~l+ke-^''.
25. >'2(2a:^ + j;^) = e\ ans. j^^ = 2x^ In |A:a:|.
dO
tan ^/ = r —
dr
If two curves are orthogonal, as they are shown in Figure 12, then
1A2 = lAi + ^/2 and thus tan ipz — ~ cot i/^i = — 1/tan i/zj. Therefore, if two
curves are to cut one another at right angles, then at the point of intersection
FIGURE II
56 Elementary Applications [Ch.3
FIGURE 12
for one curve must be the negative reciprocal of the value of that product
for the other curve.
When polar coordinates are used, suppose a family of curves has the
differential equation
Pdr+QdO = 0. (1)
Then
dO_ _P
'd'r~ Q
so
d9 Pr
(2)
dr
Hence the family of the orthogonal trajectories of the solutions of (1) must
be solutions of the equation
de Q
dr Pr
or
Q dr - r^P dO = 0. (3)
From
— a
1 + cos
§17] Orthogonal trajectories; polar coordinates 57
58 Elementary Applications [Ch. 3
Exercises
In each exercise, find the orthogonal trajectories of the given family of curves.
Draw the figures in Exs. 1-8.
1. = b sin 6
ANS. r^
= b sin 2d
ANS. r'*
ANS. r = 2b cos 6
Order One
In Section 1 1 we found that any linear equation of order one can be solved
with the aid of an integrating factor. In Section 19 there is some discussion of
tests for the determination of integrating factors.
At present we are concerned with equations that are simple enough to
enable us to find integrating factors by inspection. The ability to do this de-
pends largely upon recognition of certain common exact differentials and
upon experience.
Below are four exact differentials which occur frequently:
,/x\ y dx — X dy
59
60 Additional Topics on Equations of Order One [Ch. 4
c^
/
\
Arctan-
v\
x/
=
X dy
X
{ ——+ 4-.y dx
y
(4)
A differential involving only one variable, one such as x"^ dx, is an exact
differential.
Let us group the terms of like degree, writing the equation in the form
(y dx + X dy) + x^y^ dy = 0.
function of the product .vv will not disturb the integrability of that term. But
the other term contains the differential dy, hence should contain a function
of y alone. Therefore, let us divide by (xy)^ and write
d(xy) d_y^^
(xy)' y
or
2.vVlnk>i = l.
Recalling that
dx — x dy
<;)=^ y
2
'
§ 18] Integrating factors found by inspection 61
,3,/M_f((^^0. (9)
k + 3=-n. ^
(11)
In a similar manner, from the coefficient of d(xy) it follows that we must put
k = n-l. (12)
From equations (11) and (12) we conclude that k = —2, n = —I. The
desired integrating factor is x~^y~^ and (10) becomes
y \y/ X y
2 \y/ xy 2
x^ + 2y = cxy^.
3x^y dx + (/ - x^) dy = 0.
Since two terms in the coefficients of dx and dy are of degree three and the
other coefficient is not of degree three, let us regroup the terms to get
or
y d{x^) - x^ dy -\- y'*' dy = 0.
62 Additional Topics on Equations of Order One [Ch. 4
The form of the first two terms now suggests the numerator in the differential
of a quotient, as in
'm\ V du — u dv
VI v^
y d(x^) — x^ dy
+ y' dy = 0,
/
or
di—j +y^dy = 0.
>'
^ 3
~ 3
or
Exercises
Except when the exercise indicates otherwise, find a set of solutions.
Let US see what progress can be made on the problem of the determination
of an integrating factor for the equation
uM dx + uN dy = (2)
dy ox
dM dN
+ M — = u-— + N —
du du
u^
dy dy ox ox
,
or
/dM dN\ _ du du
\dv dxl dx dv
64 Additional Topics on Equations of Order One [Ch. 4
Not much has been gained, since we have developed no methods for
attacking an equation such as (3). Therefore, we turn the problem back into
the realm of ordinary differential equations by restricting w to be a function
of only one variable.
First let w be a function of x alone. Then cu/dy = and dujdx becomes
dujdx. Then (3) reduces to
IdM
—
dN\
=iV—
dx dx
du
or
1 JdM dN\
N \dy
\dv
~ Jx)
ox u
1 /dM
N (--f)=/(.),
\dy
1 (dM dN\
If — j
1 = /(.y), a function of .v alone, then exp (j f{x) dx) is an
N\dy dx)
integrating factor for the equation
we can say only that the equation does not have an integrating factor that is a
§ 19] The determination of integrating factors 65
function of ^ or >^ alone. For example, the student should show that the above
criteria fail in the case of Example (a) of Section 18, even though (xy)~^ is an
integrating factor for the differential equation.
Hence
1 (dM
/dM dN\
8N\ _ 2x + 4y 2
Returning to the original equation (7), we insert the integrating factor and
obtain
or
x^{4xy + 4y^ — x) = c
follows at once.
First we form
dM
-— dN
= x + 2y+l, -- = 2x + 3y + 2.
dy dx
66 Additional Topics on Equations of Order One [Ch. 4
dM _dN__
dy dx
so
1 (^ _ ^\ - _ '^ + 3'+ 1
N \dy dx/ x{x + 3y + 2)
1 /dM
/dM _^\
8N\ _ _ x + y+l 1
Therefore exp (In |>'|) = |>'| is the desired integrating factor for (9).
or
{xy^ dx + x^y dy) + {y^ dx + Ixy^ dy) + {y^ dx + 2xy dy) = 0.
or
xy^{x + 2>' + 2) = c.
From —
dM =
dy
X+ 2v, —
dN
dx
= 1, we conclude at once that
or
y(x +y- \) = ce'"".
§ 19] The determination of integrating factors 67
Exercises
Solve each of the following equations.
^ ^
ox
+J—
cy
=kF.
Use Euler's theorem to prove the result that if M and A'^ are homogeneous
functions of the same degree, and if Mx + Ny ^0, then
Mx + Ny
is an integrating factor for the equation
16. In the result to be proved in Ex. 15 above there is an exceptional case, namely,
when Mx + Ny = 0. Solve equation (1) when Mx + Ny = 0. ans. y = ex.
Use the integrating factor in the result of Ex. 15 above to solve each of the equa-
tions in Exs. 17-20.
M dx + N dy^O
may not yield at once (or at all) to the methods of Chapter 2. Even then the
usefulness of those methods is not exhausted. It may be possible by some
change of variables to transform the equation into one of a type which we
know how to solve.
A natural source of suggestions for useful transformations is the differential
equation itself If a particular function of one or both variables stands out
in the equation, then it is worth while to examine the equation after that
function has been introduced as a new variable. For instance, in the equation
the presence of both sin y and its differential cos y dy, and the fact that y
appears in the equation in no other manner, leads us to put sin y = \v and
to obtain the differential equation in u' and .r. See Example (b) below.
Then
dx = dv — 2 dy
or
(v - 1) dv + (r + 2) dy = 0.
^ fly
v + 2
70 Additional Topics on Equations of Order One [Ch. 4
But the differential of j~""^' is (1 — n)y'" dy, so equation (2) may be simpli-
fied by putting
y-"^'=z,
from which
(1 - n)y~" dy = dz.
2x dy - y{x + 1) dx + 6y^ dx = 0.
Now it can be seen that the equation is a Bernoulli equation, since it in-
volves only terms containing respectively dy, y, and y" (n = 3 here). Therefore,
A- dv + v{x + \) dx — 6 dx,
or
- -
dv = r( 1 + A '
) dx = 6a 1 dx. (4)
Since
xve^ = 6e^ + c.
§ 21] Bernoulli's equation 71
y^(6 + ce'"") = X.
In (6), the variable x appears only in the combinations x^ and its differential
3.v^ dx. Hence a reasonable choice for a new variable is w= x^. The equation
in u' and y is
2y^ dw — w{2w + y) dy = 0,
2y dz - z{2z - \) dy = 0,
dy
2z
4 dz
— 1
2 dz
z
- — = 0.
y
Therefore, we have
21n|2z- 1| -21n|z| - In b| = In |r|,
or
(2z - \Y = cyz^.
Exercises
In Exs. 1-21 solve the equation.
2. sin yix + sin y) dx + 2x^ cos y dy = 0. ans. x^ sin^ y = c(3x + sin yY.
dy
3. -f- = (9x + 4y+ \y. ans. 3 tan (6x + c) = 2(9x + 4y + 1).
dx
72 Additional Topics on Equations of Order One [Ch. 4
6y^ dx - x{2x^ + y) dy =
dii
21. The equation of Ex. 20 for the values of A and n not included there.
ANS. If // = and A r^ 0, x' = A In
1 \cy/x
U n = and k ^0,y = cx^.
1
24. 2xyy' = y^ — 2x^. Find the solution that passes through the point (1, 2).
ANS. y^ = x(5 — x^).
25. (>•* — 2x>') dx +3x^ dy = 0; when x = 2, j = 1 ans. x^ = :>/^(x + 2).
26. (2>'^ — x^) dx + 3xy^ dy = 0; when x = \,y= 1. Solve by two methods.
ANS. 5x^y^ = x^ + 4.
27. (x^ + 6y^) dx — 4x>' dy ^ 0; when x = I, y = 1. Solve by three methods.
ANS. 2y^ = x^(3x — 1).
§ 22] Coeflicients lineur in the two variables 73
in which the «'s, 6's, and t"'s are constants. We know already how to solve
the special case in which Ci = and C2 — 0, for then the coefficients in (1)
are each homogeneous and of degree one in x and y. It is reasonable to
attempt to reduce equation (1) to that situation.
They may be parallel or they may intersect. There will not be two lines if a^
and b^ are zero, or if aj and 62 ^re zero, but equation (1) will then be hnear
in one of its variables.
If the lines (2) intersect, let the point of intersection be (/?, k). Then the
translation
X — u + h, (3)
y — V -[ k
will change the equations (2) into equations of lines through the origin of the
uv coordinate system, namely,
+ Z)ii' = 0,
fliW (4)
a2U + b2V — 0.
X = u + h,
y = v + k,
where (//, A) is the point of intersection of the lines (2), will transform the
differential equation (1) into
If the lines (2) do not intersect, there exists a constant k such that
The recurrence of the expression (a^x + biy) in (6) suggests the introduction
of a new variable u' = OiX + biy. Then the new equation, in u' and .v or in
w and y, is one with variables separable, since its coefficients contain only
w and constants.
The lines
X + 2y - 4 = 0,
2x+y-5 =0
intersect at the point (2, 1). Hence put
X = u + 2,
j = r+l.
Then equation (7) becomes
which has coefficients homogeneous and of degree one in u and t'. Therefore
let u — vz, which transforms (8) into
dv (z + 2) dz
V z^-\
With the aid of partial fractions, we can write the above equation in the form
Hence we get
—+
2 dv
—— 3 dz
-
dz
=»
2 1n|r|+31n|j- 1| - In |r + 1| = ln|ci
or
{vz — v)^ = c(vz + v).
(u — vY = c(u + v).
§ 22] Coefficients linear in the two variables 75
But u =X— 1 and v =y— 1. Therefore the desired result in terms of a: and y
is
The lines
2x + 3y-\=0
and
2x + 3>; + 2 =
are parallel. Therefore we proceed, as we should have upon first glancing at
the equation, to put
2x + 3y = V.
or
{v -\)dv-iv- 7) dy = 0. (10)
y -;; + c + 6 In |y - 7| =0.
2x + 2y + c= -6 In \2x + 3y - 7|.
Exercises
In Exs. 1-18 solve the equations.
-
x-l
ANS. In [{x 1)^ + (>' + 2)2] - 8 Arctan = c.
y +2
76 Additional Topics on Equations of Order One [Ch. 4
7. Solve the equation of Ex. 6 by using the fact that the equation is linear in x.
8. {x -y+ 2) dx + Z dy = 0. ans. a- +c= 3 In |x - + >' 5|.
X- 1
ANS. 3(^-2)= -2(a:- 1) In——.
20. The equation of Ex. 19 but with the condition: when x = —\,y =2.
1 -X
ANS. 3(j-2)= -2(a:- l)ln—^.
21. (a: + — 4) dx — Ox — y — A) dy = Q\
>' when a- = 4, > = 1 .
22. The equation of Ex. 21 but with the condition: when x = Z,y=l.
—X
ANS. >' — 5x +8= 2{y — x) In y—-—
23. Prove that the change of variables
X = cxiu -\- (XiV, y —u -\- V
into an equation in which the variables // and v are separable, if a, and a^ are
roots of the equation
and if a2 ^ ai.
Note that this method of solution of (A) is not practical for us unless the roots
of equation (B) are real and distinct.
Solve Exs. 24-29 by the method indicated in Ex. 23.
24. Ex. 4 above. As a check, the equation (B) for this case is
a^ - 5a + 6 - 0,
Note that this method is not practical for us unless the roots of equation (B)
are real. But they need not be distinct as they had to be in the theorem of Ex.
23. The method of this exercise is particularly useful when the roots of (B) are
equal.
31. Ex. 16 above. The only possible ai is ( — 2). Then /3 may be chosen to be any-
thing else.
you choose ai = 2, for example, then you make ^ anything except 2. Of course,
if you choose ai = 2 and jS = 3, then you are reverting to the method of Ex. 23.
^
.2, - '
exp (-x^) dx I
dx x tan x dx
. 2 J
sin A- C dx
sin X dx I
dx
X J In X
, I
cos X C dx
cos x" dx flx
y' - 2xy = 1
^V — 2xy dx = dx,
(-l)V"
exp(-xO= L
« =
-
n
—
1
r
exp(-x') dx = c+}^
7 ^ (-\)"x^"*'
.
J „ =o nl{2n + 1)
>'exp(-x'-) = c+ X -TTT—TV-
„ = o n\{2n + 1)
00 \n„2n+
(—\Yx' 1
„ = n (2n
! + 1
erf X = 4- f
""exp ( - /?') cip (3)
— - erf X = —^ exp ( — x ),
^^ Vtt
Writing a solution in the form of (6) implies that the definite integral is to
Exercises
In each exercise, express the solution with the aid of power series or definite
integrals.
3. / = - 4x^y.
1 ans. j^ = exp (— x*) c + j
exp (/3^) dp
80 Additional Topics on Equations of Order One [Ch. 4
Miscellaneous exercises
In each exercise, find a set of solutions unless the statement of the exercise
stipulates otherwise.
3. (x + 3>' - 5) Jx - (X - J - 1) dy = 0.
ANS. 2(>'- l) = (x + j-3)ln|c(x + >;-3)|.
4. (x^ — y^) dx + 2x>' dy = 0. ans. x' + Ay^ = ex.
5. {Ix + y- 4) dx+{x-^y+\l)dy = 0.
ANS. 2x2 _^ 2xy — 33^2 - 8x + 24^ = c.
6. Solve in two ways the equation y' = ax + by + c with b ^Q.
ANS. = Cic'"' — a6x — a — cb. /j^j'
15. (x-37 + 4)^x + 2(x->^-2)i/y-0. ans. (x + - 8)-^ = r(x- 23^ + 1). >;
xy' = yfixy)
to an equation in which the variables are separable.
28. (x + 4y + 3) dx - {2x -y- 3) dy = 0.
ANS. 3(>'+l) = U + >')ln|c(x + >')l.
29. i3x-3y-2)dx-(x~y+\)dy = 0.
ANS. 2(y — 3x + c) = 5 In |2x — 2>' — 3|.
ANS. >'3(2x-l)(5x-4)=l.
36. Kx- — (x^— 2x-2;')^>' = 0.
l)i/x ans. x^ - 2x - 4^ = cj'^.
37. i6xy - 3y^ + 2y) dx + 2(x - y) dy = 0. ans. >'(2x -y) = ce' ^^
38. >''= X — + 2. Solve by two methods.
>» ans. ln|x — >'+1| = c — x.
39. (x + - 2) ^x - (x - 4>' - 2) ^>' - 0.
>>
x-2
ans. In [(x — 2)2 + 4^^] + Arctan —-— = c.
^y
x-y + 2
40. 4 ^x + (x - J + 2)2 dy = 0. ans. y + 2 Arctan = c.
CHAPTER 5
Linear Differential Equations
First we prove that if j^^ and yj are solutions of the homogeneous equation
* It is perhaps unfortunate that the word homogeneous as it is used here has a very
different meaning from that in Sections 9 and 10.
82
§ 25] Linear independence 83
y = c^y^ + C2y2
and
Now let us multiply each member of (3) by c^, each member of (4) by C2 , and
add the results. We get
•••
bolcyy^ + C2yri + b,lc,y["-'^ + €2/2"' ''li +
+ b„-ilciy[ +C2y'2l + b„[_Ciy, +C2>'2] =0. (5)
Since c^y'i + C2y'2 = (ciy^ + C2>'2)', etc., equation (5) is neither more nor
less than the statement that c^y^ + f2>'2 is a solution of equation (2). The
proof is completed. The special case c 2 = is worth noting; that is, for a
homogeneous linear equation any constant times a solution is also a solution.
In a similar manner, or by iteration of the above result, it can be seen that
if j^,; / = 1, 2, . .
.
, A:, are solutions of equation (2), and if c^; / = 1, 2, . .
.
, A:,
yi> yi^ •••> yn- The theorem just proved can thus be stated: Any linear
combination of solutions of a linear homogeneous differential equation is also a
solution.
cMx) + C2f2{x) + • • •
+ cj„{x) = (1)
identically in some interval a ^x^ b, then the functions/1,/2 ,...,/„ are said
to be linearly dependent. If no such relation exists, the functions are said to
be linearly independent. That is, the functions f, f2, •••,/„ are linearly
independent when equation (1) implies that q= C2 = ••• = c„ = 0.
84 Linear Differential Equations [Ch. 5
It should be clear that if the functions of a set are linearly dependent, then
at least one of them is a linear combination of the others; if they are linearly
independent, then none of them is a linear combination of the others.
Let P^, Pj, ..., P„, and R be functions that are continuous on an interval
a < X < b. Suppose that Xq is a real number in the interval and that j'q, yy,
. . ., y„-i are n arbitrary real numbers. Then there exists a unique function
y = y(x) defined on the interval a <x< b, which is a solution of the equation
We observe that sin x and cos .v are solutions of (2), so that for arbitrary
Cj and Cj
that Ci sin + <:2 cos = and c j cos — f2 sin = 1. This can be done in
only one way, namely by choosing" Cj — 1 and Ci = 0. We find that the
function sin x is a solution of the initial value problem (2). Moreover, since
the problem satisfies the conditions required in the foregoing theorem, no
matter what interval we choose, sin x is the only solution to the problem
given in (2).
a<x<b.
§27] The Wronskian 85
cJl + C2 fz + •
r(n-l)
cjr'' + c2fr''
'2 + "T"
•
+ (^nfn = 0.
/l
86 Linear Differential Equations [Ch. 5
C, + f2-Y+ •••
+ C„X""^ =0
can have, at most, {n — I) distinct roots and so cannot vanish identically in
Exercises
1. Obtain the Wronskian of the functions
\, X, x^, .. .
, x""' for « > 1.
2. Show that the functions e"", e^", e^" are linearly independent.
ANS. W^le^^^O.
3. Show that the functions e", cos x, sin x are linearly independent.
ANS. W=2e'' ¥- 0.
4. By determining constants Ci, d, d, C4, not all zero and which are such that
Ci /i + C2 /2 + C3 /s + C4 /4 = identically, show that the functions
/i = x,f2 = e\ fi = xe\ = {l — 3x)e' f4.
5. Show that cos(a)/ ~ /S), cosa»/, sin ojt are linearly dependent functions of /.
In part (c) you must show that if Ci /i(x) + Ci fzix) + c 3 fiix) = for allx in
— ^ 1 X^ 1, Ci = C2 = C3 = 0. Use a: = — 1, 0, 1 successively to obtain three
equations to solve for Ci, Cz, and C3
10. Given any interval a <x <b with Xo a fixed number in the interval and suppose
>' is a solution of the homogeneous equation
/' + Py' +Qy = 0. (A)
§ 28] General solution of a homogeneous equation 87
Further, suppose that y(xo) = y'(xo) = 0. Use the existence and uniqueness
theorem of Section 26 to prove that y(x) = for every x in the interval
a <x <b.
11. Suppose that ji and yz are solutions of equation (A) of Ex. 10 and suppose the
Wronskian of yi and J2 is identically zero on a <x <b. Show that for Xo in
the interval a <x <b, there must exist constants c and Cz not both zero such 1
that
ciyi(xo) + Ciyiixo) = 0,
and
Ci>^i(-^o) + Ciy'zixo) = 0.
12. Consider the function defined by
where Ci and f 2 are the constants determined in Ex. 11. Show that this function
is a solution of equation (A) above and that it follows that y(x) = on a < x < b.
(Use the results of Ex. 10.)
13. Combine the results of Exs. 10-12 to obtain a proof of the necessity condition
of Theorem 4 in the case when n = 2. Also note that the sufficiency condition
was established in the text of this section.
Let yi, y2, ••, yn t>e linearly independent solutions of the homogeneous
equation
where Cj , ^2 , . .
.
, c„ are arbitrary constants.
a special case (some choice of the c's) of the general solution (2). The basic
ideas needed for a proof of this important result are exhibited here for the
equation of order two. No additional complications enter for equations of
higher order.
Consider the equation
y=fix) (4)
88 Linear Differential Equations [Ch. 5
be any particular solution of equation (3), the solution being valid in some
interval a <x< b in which ^o(-^') does not vanish and in which the functions
bijbo and /^a/^o possess derivatives of all orders.
f(x) is sufficiently well behaved to insure that the Taylor series for/(.v),
- /""(a)(.x - ^f
/(^) = L= Ti ' (^^
fc
K-
we can divide by /?o(-v), equation (3) yields v" in terms of j, v', and the
since
known coefficients in the equation. Then y"iyi) =f'\i) can be determined.
Successive differentiation of the equation for y" will yield the higher deri-
vatives. Thus P''\oc) can be computed for k = 2, 3, . .
.
, from the values of
/(a) and/'(oc). Then, using Taylor's series (5), we see that/(.Y) is determined
by the differential equation and the two values /(a) and /'(a).
Now let>'i(.v), yjix) be any two linearly independent solutions of equation
(3). Further, let a be chosen as some point in the interval, a < oc < b, where
the Wronskian of >'i
and J2 does not vanish.
>'i(a) >'2(a)
^ 0. (6)
>''i(a) y'lici)
We wish to choose c\ and C2 so that the v of equation (7) will become the
solution /(x). But we can surely force that v to be such that v(a) —fii) and
y(a) =/'(a). That is, the equations
Ci>''i(a) + C2 3'2(a)=/'(a),
can be solved for Cj and C2 because of the inequality (6), the nonvanishing of
the Wronskian. Then/(.Y) and y{x) have the same initial values, so that by
the uniqueness theorem of Section 26 they are identical.
It is necessary to keep in mind that the above discussion used the fact
that ^oW 7^ '" ^^^ interval a < x < b. It is easy to see that the linear
equation
xy' -2y =
§ 29] General solution of a nonlioinogeneous equation 89
has the general solution y = cx^ and also such particular solutions as
y, ^x\ ^ X,
= -4;c^ X <0.
The solution yi is not a special case of the general solution. But in any
interval throughout which bQ{x) =x^ 0, this particular solution is a special
case of the general solution. It was, of course, made up by piecing together at
X = two parts, each drawn from the general solution.
bo/"' + b,/"-'' + • • •
+ b„^,y' + b„y = 0. (2)
Then
y = yc + yp (3)
is a solution of equation (1). For, using the y of equation (3) we see that
U y^, y2, ..., y„ are linearly independent solutions of equation (2), then
3'c
= ^iyi + ^2^2 + -**
+ c„j'„, (4)
in which the c's are arbitrary constants, is the general solution of equation (2).
The right member of equation (4) is called the complementary function for
equation (1).
We note that since /and yp are both solutions of the nonhomogeneous equa-
tion (1),/— >'p is a solution of the homogeneous equation (2). Hence by the
theorem of Section 28
for some particular choice of the Cy, . . ., c„. This establishes what we wished
to show.
90 Linear Differential Equations [Ch. 5
We first observe that the functions 1 and .y are linearly independent on any
interval and are solutions of the homogeneous equation y" = 0. Hence, the
complementary function for equation (5) is
yc = Ci + C2 X.
On the other hand, the function 2x^ is a particular solution of equation (5).
y ^ Ci + C2.V + 2.Y^.
y^ = Cie'' + C2e-\
Thus the complementary function for equation (6) is c^e"" + Cje'"" and a
particular solution of (6) is yp = — 4. Hence the general solution of equation
(6) is
D'y = ^,-
The expression
A=aoD" + a,D"-' + --'
+ a„_,D + a„ (1)
d"y d"-'y dy
^'^
^'^'''d?'-'^d7^^^---^'"-^d-x^'"'-
* The function y is assumed to possess as many derivatives as may be encountered in
whatever operations take place.
§ 30] Differential operators 91
The coefficients Oq, a^, ..., a„m the operator A may be functions of .v, but
in this book the only operators used will be those with constant coefficients.
Two operators A and B are said to be equal if, and only if, the same result
is produced when each acts upon the function y. That is, ^ =^ if, and only
if, Ay = By for all functions y possessing the derivatives necessary for the
operations involved.
The product AB of two operators A and B is defined as that operator which
produces the same result as is obtained by using the operator B followed by
the operator A. Thus ABy — A(By). The product of two differential operators
always exists and is a differential operator. For operators with constant
coefficients, but not usually for those with variable coefficients, it is true that
AB = BA.
Then
By = 0D-\)y = 3^-y
ax
and
A{By) = {D-\-2)(2,^^-y^
d^y dy ^dy
= 3-^--^_ + 6^_-2j ^
dx^ dx dx
—
= ^-6. +
dx^
+ 5^
^^-^y
dx
B(Ay) = (3D
-'S-^)
d^y ^dy dy
= ^-A + 6-r.--r.-2y
dx^ dx dx
d^y ^dy
= 3-^ +
dx^
5^-2j
dx
Hence
= X — X — + — — 2y
^-x
-7-3 2
dx'^ dx dx
^
= x-rj + i2-x)i^-2y.
(2-x)^'
dx^ dx
so
GH = xD^ + (2 - x)D - 2.
dx \ dx J \ dx
d^y
^ .2 — -X —
dx^ dx dx dx
^
= x-r^ + (3
dx^
_ x) ^' -2y;
(3-x)—
dx
that is,
HG = xD^ + (3 - x)D - 2.
the factors. On this topic see also Exs. 17-22 in the next section.
The sum of two differential operators is obtained by expressing each in the
form
GoD" + ayD"~^ + •••
+ a„_iZ) + fl„
A = 3>D^ - D+ x-2
and
B = x^D^ + 4Z) + 7,
then
A+B= + x^)D^ + 3D + A- + 5.
operator, if c^ and c, ^^e constants, and if/i and/2 are any functions of x
each possessing the required number of derivatives, then
A +B= B+ A.
{A + B) + C=A+ (B + C).
(AB)C = A(BC).
If A and B are operators with constant coefficients, then they also satisfy
AB = BA.
Therefore, differential operators with constant coefficients satisfy all the laws
of the algebra of polynomials with respect to the operations of addition and
multiplication.
If m and n are any two positive integers, then
D'"D" = /)"+",
Exercises
Perform the indicated multiplications in Exs. 1^.
6. 2D^-5D- 12.
1. D'-2D^-5D + 6. ANS. (£)-l)(D + 2)(Z)-3).
8. 4D^-4D^- ll£> + 6.
9. D* - 4D^. ANS. D\D - 2){D + 2).
10. i)^-3£)^ + 4.
11. I>^ - 21 D + 20. ANS. (D - 1)(Z) - 4)(I> + 5).
12. 2D'- D^- 13/) -6.
13. 2D*+ 11Z>3+ 18Z)^ + 4Z)-8. ans. (D + 2)^(2Z> - 1).
£,fc^mx
^ ^k^mx^
(1)
it is easy to find the effect an operator has upon e'"''. Let f{D) be a polynomial
in D,
f(D) = aoD'' + a,D"-' + ---
+ a„_,D + a„. (2)
Then
f{D)e""' =aQm"e'"' + a^nf'^e"''' + • • •
+ a„_i/ne'"-'' + a„e'"^
so
f{D)e'"' = e'"Y(»i). (3)
/(Z^y"^ = 0.
Next consider the effect of the operator D—a on the product of e""" and a
function y. We have
(Z) - aWy) = D(e''y) - ae^y
= e'^'Dy,
§ 32] Some properties of differential operators 95
and
= e'^'D^y.
e'"f(D)y=f(D-a)le'"y'\. (5)
The relation (5) shows us how to shift an exponential factor from the left
of a differential operator to the right of the operator. This relation has many
uses, some of which we will examine in Chapter 6.
EXAMPLE (a): Let f(D) = 2D^ + 5D- 12. Then the equation /(w) =
is
2m^ + 5m - 12 = 0,
or
(m + 4)(2m - 3) = 0,
In equation (5) we let f(D) = {D — m)" and y = x*. Then using the ex-
ponential shift we obtain
{D + 3fy = 0. (7)
e^'^iD + 3)V = 0.
D\e^y) = 0.
e^'^y = Ci + C2 X + C3 x^ + c^x^,
and finally,
Note that each of the four functions e~^^ xe~^^ x^e^^"", and .v^e"^^ is a
solution of equation (7). This of course is assured by the theorem of equation
(6) of Example (b).
If we now show that the four functions are linearly independent, equation
(8) gives the general solution of equation (7). See Ex. 5 below.
Exercises
In Exs. 1^ use the exponential shift as in Example (c) above to find the general
solution.
33. Introduction
d"y d"-\ dy
97
98 Linear Equations with Constant Coefficients [Ch. 6
f{D)y = Q, (2)
f{D)e'"' = 0,
which means simply that y = e'"'' is a solution of equation (2). The equation
f{m) = (3)
m„ . If these roots are all real and distinct, then the n solutions
are linearly independent and the general solution of (1) can be written at once.
It is
Repeated roots of the auxiliary equation will be treated in the next section.
Imaginary roots will be avoided until Section 37, where the corresponding
solutions will be put into a desirable form.
^_^
dx^ dx^
dy
dx
m^ - Am^ + 777 + 6 = 0,
and its roots are m = 0, — 2, ^. Using the fact that e°* = 1 , the desired solution
may be written
—dr - 4x =
d'^x
2
m^ -4 = 0,
with roots m = 2, —2. Hence the general solution of the differential equation
is
X = qe^' + €26''^'.
dt
= Cj + C2,
3 = 2ci - 2c2 .
x = i(e''-e-''),
X = f sinh (2/).
Exercises
In Exs. 1-22, find the general solution. When the operator D is used, it is implied
that the independent variable is x.
6. iD^-3D^- lOD)y = 0.
7. (D^ + 6D^+ llD + 6)y = 0. ans. y = de'' + CzC-'''' + de-^".
8. (Z)^ + 3D^ -4D-\2)y = 0.
9. (4/)^ -7D+ 3)y = 0. ans. y = Cie" + Cz exp (^x) + c^ exp (- §x).
10. (4Z)^- UD-(>)y = 0.
— ^dx
2 — = 0. +,,,-,,
^. c/^jc f/^x ^
11. H ANS. jc = Ci C2 e' + C3 e- 2'.
12.^-19^ + 30;c = 0.
dt^ dt
13. (9£)^ -7D + l)y = 0. ans. >' = Cie-' + Cz exp {\x) + Ca exp (fjc).
23. (D^- 2D- 3)7 = 0; when = 0,7 = 0, and/ = -4. ans. J = e-^-e^^
a:
In Exs. 25-29, find for x = 1 the y value for the particular solution required.
KD)y = (1)
§ 35] The auxiliary equation; repeated roots 101
the operator/(Z)) has repeated factors; that is, the auxihary equation/(w) =
has repeated roots. Then the method of the previous section does not yield
the general solution. Let the auxiliary equation have three equal roots
Wj = W2 = b, /W3 = b. The corresponding
b, part of the solution yielded by
the method of Section 34 is
m^ = 1712 = • — m„ — b.
Then the operator f{D) must have a factor {D — b)". We wish to find n
linearly independent >''s for which
{D - b)"y = 0. (4)
Turning to the result (6) near the end of Section 32 and writing m = b,
we find that
(D - bfix^'e^'') = for A: = 0, 1, 2, ...,(« - 1). (5)
y = de'"' + C2 xe^"" + • • •
+ c„ x"" ^e*"^. (6)
fiD)y = (1)
can be written
g{D){D - bfy = (7)
where ^(Z)) contains all the factors off(D) except (D — b)". Then any solution
of
(D - bfy = (4)
and corresponding to n equal roots niy, m2 , ..., ni„, each equal to b, there
Cie^'',C2xe''\...,c„x"-'^e''\ (9)
The collection of solutions (9) has the proper number of elements, a number
equal to the order of the differential equation, because there is one solution
corresponding to each root of the auxiliary equation. The solutions thus
obtained can be proved to be linearly independent.
With the aid of synthetic division, it is easily seen that the auxiliary equation
or
y = Cie"" + (C2 + C3 A- + C4 x^)e^''.
d^
dx'*'
2^
dx^
^=
dx^
m'^ + 2m^ + m^ = 0,
Exercises
In Exs. 1-20 find the general solution.
4. (Z)^-8Z)^+ \(>D)y = 0.
5. (/)^ + + 9Z)^)>' = 0.
6Z)3 ANS. 7 = Ci + Cz X + (C3 + C4 x)e - ^\
6. (Z)^-3Z)^ + 4)>' = 0.
7. (4Z)^ - 3Z) + \)y = 0. ans. 7 = c^e-'' + (c2 + c^x) exp (ix).
8. (/)* - 3>D^ - 6D^ + 2SD - 24)y = 0.
or = Ci + C2 X + C3 x^ + ce cosh X + ci sinh x.
^^
ANS. J = (ci + C2 x)e " + C3 exp [(1 + V3)x] + C4 exp [(1 — V3)a:].
^"^
ANS. 7= (1 + x)e~^*.
22. The equation of Ex. 21 with the condition that the graph of the solution pass
through the points (0, 2) and (2, 0). ans. y = {X — x)e-^''
23. (D^ -3D-2)y = 0; when x = 0, y= 0, y' = 9, y" = 0.
In Exs. 27-29, find for = 2 the y value for the particular solution required
a;
Since the auxiliary equation may have imaginary roots, we need now to lay
down a definition of exp z for imaginary z.
104 Linear Equations with Constant Coefficients [Ch. 6
Let z = a + //? with a and P real. Since it is desirable to have the ordinary
laws of exponents remain valid, it is wise to require that
Now consider e''^, P real. In calculus it is shown that for all real x
X x~ x^ X"
*
-'+T! + 2! + 3!
+ - + ;^ + -' ('>
or
^^=Z^- (2)
Separating the even powers of P from the odd powers of P in (3) yields
2! ^ 4! i2k)\
ip i'p' i'>^^'p'
or
:
y -^- + y
A'o (2/c)! fcfo
- ——
(2/c +
'^
l)!
.
(4)
2!
^4!^ (2/v)!
p p^ i-D^p^"^^
+ /
_1_ • . . _J_ _L
(5)
1! 3! (2/c+l)!
or
But the series on the right in (5) are precisely those for cos P and sin P as
The student should realize that the manipulations above have no meaning
in themselves at this stage (assuming that infinite series with complex terms
are not a part of the content of elementary mathematics). What has been
accomplished is this: the formal manipulations above have suggested the
meaningful definition (6). Combining (6) with (1) we now put forward a
reasonable definition of exp (a + ifi), namely,
It is interesting and important that with the definition (7), the function e^
for complex z retains many of the properties possessed by the function e"" for
real x. Such matters are often studied in detail in books on complex variables.*
y = exp (a + ib)x,
(D-a- ib)y - 0.
The result desired follows at once by differentiation, with respect to x, of the
function
y = e^^Ccos bx + / sin bx).
nil = a + ib
is a root of the equation /(m) — 0, with a and b real and b ^0, then
ni2 — a — ib
is also a root of f{m) = 0. It must be kept in mind that this result is a con-
sequence of the reality of the coefficients in the equation/(m) — 0. Imaginary
roots do not necessarily appear in pairs in an algebraic equation whose
coefficients involve imaginaries.
* For example, R. V. Churchill, Complex Variables and Applications (2nd edition; New
York: McGraw-Hill Book Co., 1960), pp. 46-50.
106 Linear Equations with Constant Coefficients [Ch. 6
f{D)y = (1)
Taking .v to be real along with a and b, we get from (2) the result
m^ - 2>m^ + 9m +13 = 0,
(,„2 ^ 4)2 ^ Q^
so its roots are seen to be m— ±2i, ±2i. The roots mi = 2? and mj = —2/
occur twice each. Thinking of 2/ as + and recalling
2/ that e°^ = 1, we
write the solution of the differential equation as
Exercises
Find the general solution except when the exercise stipulates otherwise.
[(/) - ay + b^]y = 0.
2. (D^ — 2D + 5)y = 0. Verify your answer.
ANS. y = Cic" cos 2x + C2 e" sin 2x.
3. {D^ -2D + 2)y = 0.
4. (D^ + 9)y = 0. Verify your answer. ans. y=Ci cos 3a: + ci sin 3x.
5. {D^ — 9)y = 0. ANS. y=Ci cosh 7>x + d sinh 2>x.
6. (D^ + AD + 5)y = 0. Verify your answer.
7. (D^ - 6£) + 25);^ = 0. ANS. y = e^^Ci cos Ax + Ci sin 4jc).
8. (Z)3 + D^ + 4Z) + 4)>' = 0. Verify your answer.
9. (D^ — \)y = 0; when x = 0,y = yo and y = 0. ans. y = yo cosh x.
10. (£)^ + !)>' = 0; when x = 0, 7 = jo and / = 0. ans. y = yo cos x.
11. (/)* + 2Z)3 - 11 D^ - 52£>)>' = 0.
12. (Z)3 + 5Z)^ + 17Z) + 13)>' = 0; when x ^0, y = Q,y =\, and /' = 6.
ANS. = e""* — e"^'' cos 3a:. >'
108 Linear Equations with Constant Coefficients [Ch. 6
d^x dx
19. ^-^ + /:^a: = 0, k real; when / = 0, x = and —r=Vo- Verify your result
dt^ dt
completely. ans. x = {volk) sin kt.
20. (/)' + Z)^ + 4Z) + 4)>' = 0; when x = 0, j = 0, / = - 1, and >'"= 5.
ANS. y = e~'' — cos2x.
d^x dx dx
21. -— + 2b — + k^x = 0, k > b >Q; when / = 0, x = and —= !;o
dt^ dt dt
Miscellaneous exercises
Obtain the general solution unless otherwise instructed. Your answer can be
checked by direct substitution.
7. AD^ - 21 D + 21)y = 0.
8. 10Z)^ + Z)^ - 7D + 2)>' = 0.
9. D^ + 7Z)2 + 19Z) + 13)>' = 0; when = 0, = 0, / = 2, and /' = - 12.
a- >'
11. Z)'' + 6D^ + 9D^)>' = 0; when x = 0, 7 = 0, j' = 0, and j'" = 6, and as x -> cx),
''->!. For this particular solution, find the value of >' when x = 1.
ANS. y=\—e~^.
12. D^ + 6/)" + 12D + 8);' = 0; when x = 0, j = 1, /= -2, and /' = 2.
13. Z)' + 3Z)^ + 3Z)+l)j = 0.
14. £)4 _ 2£)3 _ i3£)2 ^ 38£, _ 24)^ = 0.
Undetermined Coefficients
is
where y^, the complementary function, is the general solution of the homo-
geneous equation
Various methods for getting a particular solution of (1) when the bo,
110
§ 38] Construction of a homogeneous equation 111
bi, ..., b„ are constants will be presented. In preparation for the method of
undetermined coefficients it is wise to obtain proficiency in writing a homo-
geneous differential equation of which a given function of proper form is a
solution.
Recall that in solving homogeneous equations with constant coefficients,
a term such as Cic"^ occurred only when the auxiliary equation /(m) = had
a root m= a, and then the operator/(Z)) had a factor (D — a). In like manner,
appeared only when f{D) contained the factor
€2X6'"' {D — a)-^, c^x^e"'' only
when f{D) contained (D — a)^, etc. Such terms as ce"'' cos bx or ce""" sin bx
correspond to roots m= a ± ib, or to a factor [{D — a)^ + b^].
y = le^"" + 2x.
First note that the coefficients (7 and 2) are quite irrelevant for the present
problem, so long as they are not zero. We shall obtain an equation satisfied
by > = Cjg^* + C2X, no matter what the constants c^ and Cj may be.
A term c^e^'' occurs along with a root m — Z of the auxiliary equation. The
term Cj x will appear if the auxiliary equation has m = 0, 0; that is, a double
root m = 0. We have recognized that the equation
D\D - l)y = 0,
or
(D^ - 3D^)y = 0,
The term 6 is associated with m = 0, the term Sxe"" with a double root
m= 1, 1, and the term ( — cos x) with the pair of imaginary roots m — 0±i.
m{m- \)\m^ + 1) = 0,
or
m^ — 2m* + 2w^ — 2m^ + w = 0.
The desired equation must have its auxiliary equation with roots
m= I ±2i, \ ± 2/. The roots m= \ ± 2/ correspond to factors (m — 1)^ + 4,
or
Note that in all such problems, a correct (but undesirable) solution may be
obtained by inserting additional roots of the auxiliary equation.
Oral exercises
In Exs. 1-14, obtain in factored form a linear differential equation, with real,
constant coefficients and which is satisfied by the given function.
In Exs. 15-36, list the roots of the auxiliary equation for a homogeneous linear
equation with real, constant coefficients and having the given function as a parti-
cular solution.
Before proceeding to the theoretical basis and the actual working technique
of the useful method of undetermined coefficients, let us examine the under-
lying ideas as applied to a simple numerical example.
Consider the equation
D\D - \)y = 3e* + sin x. (1)
Jc = <^i + ^2 ^+ C3 e\ (3)
where y^ is as given in (3) and yp is any particular solution of (1), all that
remains for us to do is to find a particular solution of (1).
114 Nonhomogeneous Equations: Undetermined Coefficients [Ch. 7
with constant coefficients, because we know how to solve any such equation.
But, by (6), the operator {D — \){D^ + 1) will annihilate the right member of
(1). Therefore, we apply that operator to both sides of equation (1) and get
Any solution of (1) must be a particular solution of (7). The general solution
of (7) can be written at once from the roots of its auxiliary equation, those
roots being the values w= 0, 0, 1 from (2) and the values m' = 1, +/ from (5).
y = yc + yp^ (9)
where
>'c = f1 + f2 -V + f3 e\
the Tj, C2 , C3 being arbitrary constants as in (8). Thus there must exist a
Using different letters as coefficients to emphasize that they are not arbitrary,
we conclude that (1) has a particular solution
Since (11) is to be an identity and since e^, sin x, and cos x are linearly
independent, the corresponding coefficients in the two members of (1 1) must
be equal; that is,
A = 3
B+C= 1
B-C = 0.
A careful analysis of the ideas behind the process used shows that to arrive
at the solution (12), we need perform only the following steps:
(a) From (1) find the values of w and m' as exhibited in (2) and (5);
(b) From the values of m and m' write y^. and yp as in (3) and (10);
(c) Substitute yp into (1), equate corresponding coefficients, and obtain
the numerical values of the coefficients in yp\
(d) Write the general solution of (1).
Let us examine the general problem of the type treated in the preceding
section. Let/(Z)) be a polynomial in the operator D. Consider the equation
y = yc + yp (3)
116 Nonhomogeneous Equations: Undetermined Coefficients [Ch. 7
where y^ can be obtained at once from the values of /» in (2) and where
y = y^is any particular solution (yet to be obtained) of (1).
Now suppose that the right member R{x) of (1) is itself a particular solution
9{D)R = 0, (4)
Recall that the values of i»' in (5) can be obtained by inspection from Rix).
The differential equation
g{D)f(D)y - (6)
has as the roots of its auxiliary equation the values of m from (2) and m' from
(5). Hence the general solution of (6) contains the y^ of (3) and so is of the
form
But also any particular solution of (1) must satisfy (6). Now, if
thenf{D)yg = Rix) because /(Z))>'^. = 0. Then deleting the y^ from the general
solution of (6) leaves a function y^ which for some numerical values of its
coefficients must satisfy (1); that is, the coefficients in y^ can be determined
Here we have
m= \, —2
and
w' = 0, 0, ±2i.
yc = <^i^'' + <^2^~^^
yp = A + Bx + C cos 2x + E sin 2x,
Since
Dyp = B — 2C sin 2.v + 2£ cos 2x
and
D^yp = -4C cos 2x - AE sin 2x,
B-2A=0.
The above equations determine A, B, C, and E. Indeed, they lead to
A^-\, C= 6,
B=-\, E^-2.
Since the general solution of (7) is y = yc + >'p , we can now write the
desired result,
y^ — c^ cos X + f 2 sin x,
yp = Ax cos X + Bx sin x.
Now
y'p = A( — x cos X — 2 sin x) + B( — x sin x + 2 cos x),
y ^ Axe'"" + Be^''.
Now
2x
J-;
= Ai-xe-'' + e-"") + 2Be^\
y'L = A(xe-'' - 2^-^) + ABe^\
y"' = Ai-xe'"" + 3e"") + SBe Ix
Then
y";-y'^ = lAe-^-^(^Be^\
We must determine Tj, c 2 ^ <"3 so (11) will satisfy the conditions that when
X = 0, >' = 0, >•' = — L and y" — 2.
and
We put .Y = in each of (11), (12), and (13) to get the equations for the
=
+ r2 + f 3 + i
f1
- = C2 - f 3 + 3,
1
2 = C2 + f3 - 2,
general solution, the y of (11), that must be made to satisfy the initial con-
ditions.
§ 40] The method of undetermined coefficients 119
Exercises
In Exs. 1-35, obtain the general solution.
dx d^x
41. jc + 4x + 5x = 8 sin /; when r = 0, = at and x = 0. Note that a: = ^- x
, = —-r
is a common notation when the independent variable is time.
X == (1 + e~ ^') sin r — (1 — e~^') cos ANS. /.
In Exs. 45-48, obtain from the particular solution indicated the value of y and
the value of >'' at x = 2.
51. {D^ -r l)^ = 2 cos x; when x = 0, y = 0, and when x = tt, >> = 0. Show that this
initial value problem has an unlimited number of solutions and obtain them.
ANS. y = {c -{- x) sin x.
52. For the equation {D^ + D^)y = 4, find the solution whose graph has at the
origin a point of inflection with a horizontal tangent line.
ANS. J = 4 — 4x + 2x^ — 4^" *.
53. For the equation (D^ — D)y = 2 — 2x, find a particular solution which has at
some point (to be determined) on the x-axis an inflection point with a horizon-
tal tangent line.
by inspection.
For example, if R(x) is a constant Rq and if b„ ^ 0,
Ro
(2)
is a solution of
Now D'^x'' = k\, a constant, so that all higher derivatives of x* are zero.
Thus it becomes evident that (4) has a solution
yp = TTz
k\b„_k
Rqx''
— '
(5)
y^ = Cie"" + C2 e^"".
=
yp 2
+4— =
—-5+4-5
d'y d'y ^
7. (7)
dx^ dx
7x^ 7x^
y,
3 ! 4 24
, Ix^ 7 •
6
(D^
^ + AD^) = + 4 •
= 7.
'
lA 24
Examination of
y = A^mZx (9)
we get
D^y = -9 A sin3x,
us to write
and so to write
When the >'p of (10) is substituted into (8), it is found, of course, that
A= -i,B^O.
In contrast, consider the equation
possible. For this equation, m= —2, —2,m'= ±3/, and in the particular
solution
it must turn out that B^ 0. No labor has been saved by the inspection
In more complicated situations such as
is a solution.
Finally, note that if Ji is a solution of
f{D)y = R,ix)
and ^2 is a solution of
f{D}y = R,(x),
then
yp^yi+ yj
is a solution of
f{D)y = R,(x) + R2{x).
fiD)y = Rix)
yi = -ie""
124 Nonhomogeneous Equations: Undetermined Coefficients [Ch. 7
is a particular solution of
{D^ - 9)y2 =X
and that
y^ = TT sin 4x
satisfies
(D^ - %3 = -sin4x.
Hence
yp= - 1^''
-i^ + Tj sin 4x
m' = ±2i,
we put
Exercises
1. Show that if b ^a, then
+ a^)y — sin bx
(D^
42, Introduction
coefficients.
In this chapter we shall study two methods that carry no such restrictions.
126
§ 43] Reduction of order 127
y = yiv (3)
/^ + y'lV,
yiv'
y" = yiv" + 2y\v' + y'lv,
or
By the usual method (integrating factor) we can find vv from (6). Then we
can get r from r' = iv by an integration. Finally y = yiV.
Note that the method is not restricted to equations with constant co-
depends only upon our knowing a particular solution of equation
efficients. It
(2); that upon our knowledge of the complementary function. For practical
is,
purposes, the method depends also upon our being able to effect the inte-
grations.
128 Variation of Parameters [Ch. 8
We shall take the particular solution e"" and use the method of reduction of
order by setting
y = ve^.
Then
y' = ve'' + y'e"",
and
y" = ve^ -{ Iv'e^ -\- v"e''.
and hence
V = c^^e'^^ + C2 + jx,
Of course, the solution to equation (7) could have been obtained by the
method of undetermined coefficients. Let us now solve a problem not solvable
by that method.
y^ = Ci cos X + C2 sin x. (1 1)
§ 43] Reduction of order 129
We may use any special case of (11) as the y^ in the theory above. Let us
then put
y = V sin x.
We find that
and
y" = v" sin x + 2v' cos x — j; sin x.
or
y" + 2r' cot jc = csc^ X. (12)
w sin^ X = X,
or
Hence
V =\ X csc^ X dx,
or
Exercises
Use the method of reduction of order to solve the following equations.
(x- l)y"-x/ + y = 0.
(x ~ \)y" — xy' + y= \.
2x^y" + xy' — j =
and find the general solution. For what values of x is the solution valid?
ANS. y = CiX + CzX'^'^.
14. In Chapter 19 we shall study Bessel's differential equation of index zero
xy" + y' + xy = 0.
Suppose that one solution of this equation is given the name Jo{x). Show that a
second solution takes the form
r dx
ax
(1 - x^)y" - 2xy' + 2y =
l+x
is y = X. Find a second solution. ans. y = — 2 + In
l-x
Suppose now that we know the general solution of the homogeneous equa-
tion (1). That is, suppose
y = Ay,+ By 2 (4)
and try to determine A{x) and B{x) so that Ay^^ + By 2 is a solution of equa-
tion (2).
that we are involved with two unknown functions A{x) and B{x) and
Note
that we have only insisted that these functions satisfy one condition: the
function in (4) is to be a solution of equation (2). We may therefore expect to
impose a second condition on A{x) and B{x) in some way which would be to
our advantage. Indeed, if we simply impose the condition B{x) = 0, then we
will be dealing with the method of reduction of order. Actually we impose
Rather than become involved with derivatives of ^4 and B of higher order than
the first, we now choose some particular function for the expression
A'yi + By 2 .
Technically, we could let this function be sin x, e^, or any other suitable
function. For simplicity we choose
^>i+5>2=0. (6)
Since y was to be a solution of (2) we substitute from (4), (5), and (7) into
But ji and jj are solutions of the homogeneous equation (1), so that finally
Equations (6) and (8) now give us two equations that we wish to solve for
^'i yz
y'l y'l
does not vanish. But this determinant is precisely the Wronskian of the
functions y^ and yi , which were presumed to be linearly independent on the
interval a <x< b. Therefore, the Wronskian does not vanish on that
and we can find A' and B' By integration we can now find A and B.
interval .
method that prohibits the linear differential equation involved from having
variable coefficients.
Of course,
y^ = Ci cos X + C2 sin x.
from which
y' = —A sin x + B cos x + A' cos x + B' sin x.
Next set
so
Then
y" = —A cos A — 5 sin x — A' sin a + B' cos a. (12)
§ 44] Variation of parameters 133
Next we eliminate y by combining equations (10) and (12) with the original
equation (9). Thus we get the relation
B' = tan X,
so
5= ln|secx|, (14)
in which the arbitrary constant has been disregarded because we are seeking
only a particular solution to add to our previously determined complementary
function y^
From equations (13) and (11) it also follows easily that
A' = — tan^ X.
Then
/I = — J
tan^ X dx = J(l — sec"^ x) dx,
so
A=X— tan X, (15)
Here
y, = Cie'' + C2 e^\
134 Variation of Parameters [Ch. 8
SO we put
Combining (18), (20), (21), and the original equation (17), we find that
1 +e-^
Elimination of B' from equations (19) and (22) yields
Then
Similarly,
so
B
or
§ 45] Solution of j" + J = f(x) 135
and if we choose
A' cos X + B' sin x — 0, (3)
we obtain
y" = —A cos X — 5 sin a: — /I' sin X + B' cos x. (4)
Equations (3) and (5) may be solved for A' and B', yielding
for any x « ^ x ^ 6. It
in is here that we use the integrability of/(x) on the
interval a ^ ^ b.
x
The A and B of (6) and (7) may be inserted in (2) to give us the particular
solution
Hence we have
Exercises
In Exs. 1-18, use variation of parameters.
17. j'" — = X.
>''
18. >'"'
+ y' = tan x.
19. Observe that x and ^"^ are solutions of the homogeneous equation associated
with
(1 - x)y" + xy' -y= 2(x - lye''.
Let W(x) be the Wronskian of >'i and yz and assume iV(x) , 7^ on the interval
a <x <b. Show that a particular solution of equation (A) is given by
f"/()3)L)'.(ft/«W - y,My2m dp
''']. WW)
•
<^>
and
If we multiply equation (C) by yz and equation (D) by yi and then subtract the
two equations, we obtain
Miscellaneous exercises
1. iD^-l)y=2e-%l + e-^'')-^. ans. y = yc- xe'"' - ie-' \n(l + e-^").
2. (D^-l)y = (l-e^'')-''\ ans. y = y,-(l-e^^y'\
3. (D^ — l)y = e^%3 tan e^ + e'' sec^ e"). ans. y = yc + e''ln |sec e"].
4. (D^ + 1)^ = sec^ X tan x.
ANS. y = yc+Hanx+icosx\n |sec x + tan x\.
5. Do Ex. 4 by another method.
6. (D^ + 1);^ = CSC X. Use variation of parameters.
ANS. See Example (b) of Section 43.
7. (Z)^ -3D + 2)y = sec^ (e-"). ans. _>; = Ci e' + C2 e^^ + e^^ in |sec (e-^)|.
138 Variation of Parameters [Ch. 8
12. (D^ — 3D + 2)y = sin e" '. ans. = - e^^ sin e"
j^ >'c *.
13. (D^ — l)y = llie^" + 1). ans. y = yc—i — cosh x Arctan e'".
14. y" + y = sec^ x tan x. ans. y ^ yc + ^ sec x tan x.
15. >'" + = sec X tan^ x. Verify
>' your answer.
16. /' + 4y' + 3>' = sin e". ans. y = yc — e'^'' sin e" — e~ ^* cos e".
17. y" + y = csc^ x cot x. ans. = >'c + i cot x csc x.
^^
CHAPTER !/
e"YiD)y=fiD-a)[e''y], (1)
139
140 Inverse Diflferential Operators [Ch. 9
>'c
= CyC^ cos 2x + C2 e^ sin 2x. (3)
which can be obtained by the method of Chapter 7. But the task of obtaining
the derivatives of Vp and finding the numerical values of A, B, C, and £ is a
little tedious. It can be made easier by using the exponential shift (1).
D\e-^''yp)^6Ax + 2B,
+ 85=
\2A 0,
6/1 + 85 + 8C = 0,
2B + 4C+SE= 0,
Therefore
or
y. = {2x^ - 3x^ + f;c)e'^
§ 46] The exponential shift 141
The complementary function could also have been obtained from (5).
Here the immediate use of the exponential shift would do no good, because
removing the e'' factor from the first term on the right would only insert a
factor e"* in the second and third terms on the The terms 7a; — 2 on
right.
the right give us no trouble as they stand. Therefore we break (7) into two
problems, obtaining a particular solution for each of the equations
and
(Z)' - 2Z) + 1)V2 = 7a- - 2. (9)
e-\D - l)Vi =X
to
D\e-yy) = X. (10)
e'^'yi = ix\
so
y,=ix'e\ (11)
y2 = Ax + B.
Then Dy2 = A, and from (9) it is easily found that A= 1, B= \2. Thus a
particular solution of (9) is
Using (1 1), (12), and the roots of the auxiliary equation for (7), the general
solution of (7) can now be written. It is
e^\D + Afy^ = 6,
y = Ci + C2 X + (cj + C4 X + 3x^)e~'*''.
Exercises
In Exs. 1-12, use the exponential shift to find a particular solution.
y=j^^RM (2)
and to try to define an operator l/f(D) so that the function v of (2) will have
meaning and will satisfy equation (1).
After all, the only thing that we require of our evaluation is that
fiD)-j^^R(x)^Rix). (3)
and
iD-a)"ix"e'"') = n\e''\ (2)
1 e"'
^"=7Z^' /(a)^0. (3)
/(£>) /(a)
f{a) f{a)
Hence (3) is verified.
Now suppose that f{a) = 0. Then f{D) contains the factor {D — a).
Suppose that the factor occurs precisely n times in /(/)); that is, let
/(Z)) = 0(Z)X/)-a)";0(a)^O.
With the aid of (2) we obtain
Therefore we write
1 xV""
(t>(D){D-ay
^"=^T^,
nl(f>{a)
<t>(a)^0, (5)
Note that formula (3) is included in formula (5) as the special case, n = 0.
/(/)) = (Z)^ + 1)
and
/(2) 9^ 0.
•^p
£)' + 1 2^ + 1 5
§ 49] (D2 + a2)-i gin ax and (D^ + a^y cos ax 145
1
._ xV _ 1 3^
^^ {D-{)'D'(D+l)^ 12-2 ^
'
3! n""
Then the general solution of (7) is
1 . . sin bx
sin bx = -2 7-2', b ^ a,
D^ + a^ a'-b'
and a similar result holds for the expression (D^ + a^)~^ cos bx, when b ^ a.
1
sin ax. (1)
j)2_^^2
The formulas of the preceding section can be put to good use here, since
sin ax = .
2/
Then
-,
D^ + a-
1.1 ^ sni ax =
2i(D-ai){D +
1
ai)
(e""' - e""'*)
~ 2f ll!2a/
~ l!(-2a/)/
,aix ^ — aix
X e"'-' +
1
2a 2
146 Inverse DiflFerential Operators [Ch. 9
SO
1 X
— = sin ax = — —- cos ax. (2)^
D^ + a^ 2a ^
1 X
_, 7 cos ax = -— sin ax, (3)
^
D^ + a^ 2a
Oral Exercises
Give the general solution of each equation.
Exercises
1. Verify formula (2) of Section 49.
2. Obtain and verify formula (3) of Section 49.
4. {D^^'iD-A)y = 2\e^\
5. (Z)^ + 3Z) - A)y = ISe". ans. y = de'*" + (cj + 3.vK
6. (£>^ + 3Z)-4)>' =20e-*\
7. (Z)^ - 3£) + 2)j =
e" + e^\ ans. y = (ci- x)e'' + (cz + x)e'''
8. (4D^- !):>;= e^/2+ \2e\
9. Z)HZ)-2)3>' = 48e'^ ANS. >' = Ci + C2A: + (c3 + C4A: + f5X^ + 2A:^)e'2^
10. (Z)* - 18Z)^ + 81)7 =- 36e^^
11. {D^ + 16)^ = 14 cos 3a:. ans. y = Ci cos 4a- + Ci sin 4a' + 2 cos 3Ar
13. y" + 16j = 24 sin 4a:. ans. y={ci — 3x) cos 4x + Ci sin 4Ar
14. >»"+ 6>' = 48 cos 4a:.
1
ANS. y = e~ "(ci cos 2Ar + Cz sin 2jc) — ixe~ " cos 2^.
19. Prove that if /(a) = (at - ay(f>{x), then /'"'(o) = «! (/)(o). Then use equation (5)
of Section 48 above to prove that
/(D) /'"'(a)'
— D) sin ax
= f{
1
24. —^ sin ax
J^-^J^—-^
.nam-ai) ^ 0.
25. — 1
cos ax =
fi—D) cos OAT
J^^r^j^Z-f ^fiaOfi-ai) ^ 0.
26. — 1
sinh ax =
/(-/)) sinhflx
-J^-^J^—^
; /(«)/("«) ^ 0.
27. — 1
cosh ax =
f{ — D) cosh oa:
^(^)^(_^) ; mn-a) ^ 0.
1 a:"
^^-
rm~^ — i7« cos ^^ = ,-, ,„ ,
cos (ax - ^rt77).
The reader is already familiar with some operators that transform functions
148
§ 51] Definition of the Laplace transform 149
Given a function K(s, t), called the kernel of the transformation, equation
(1) associates with each F{t) of the class of functions for which the above
integral exists a function /(j') defined by (1). Generalizations and abstractions
of (1), as well as studies of special cases, are to be found in profusion in
mathematical literature.
K{s,t) = Q, for/<0,
= e~'\ for t ^ 0,
Let F{t) be any function such that the integrations encountered may be
legitimately performed on F{t). The Laplace transform of F{t) is denoted
by L{F{t)] and is defined by
The integral in (1) is a function of the parameter s; call that function /(5).
We may write
for 5 > 5o . In extreme cases the integral may converge for all finite s.
* If 5 is not to be restricted to real values, the convergence takes place for all s with real
part greater than some fixed value.
150 The Laplace Transform [Ch. 10
Using elementary properties of definite integrals, the student can easily show
the validity of equation (3).
We shall hereafter employ the relation (3) without restating the fact that
-(s-k)t
dt.
L{/'}= I
e-"-'^'dt
•'o
— e -(s-fc)t-i OC
s-k
1
= +
s — k
Thus we find that
r e"(a
e {a sin mx — m cos mx)
e sm mx ax = ^
^ h C.
J m
a^ + m^
Since
.00
L{sin kt} = \
e ^'
sin kt dt.
§52] Transforms of elementary functions 151
it follows that
For positive s,e ^' -> as / -> oo. Furthermore, sin kt and cos kt are bounded
as / ^ 00. Therefore (3) yields
1(0 - k)
L{sin kt) =
s' + k''
or
The result
By definition
Let us attack the integral using integration by parts with the choice exhibited
in the table.
t" e-^'dt
1
nt"-^ dt e
s
We thus obtain
,Q0 00 M 00
-t"e' n r^
/»
For s > and n>0, the first term on the right in (6) is zero, and we are left
with
or
n
L{t"
so
§ 52] Transforms of elementary functions 153
Thus
L{H(t)} = -
s
1^ +
s
+ 4-0 + ^^-
s s
^+
Exercises
5. Evaluate L{2e^'-e-^'}.
s
6. Show that L {cosh A:/} ==-; —,s>\k\.
s^ — k^
k
7. Show that L{sinh = -; —
A:^}
5^ — A:^ ,^>l^l•
8. Use the trigonometric identity cos^ A= ^{\ -\- cos lA) and equation (5),
——,
_y2 + 2k^
Section 52, to evaluate L {cos ^ A;/}. ans. ,s>0.
154 The Laplace Transform [Ch. 10
'"'
{h+'-y '-{?+-)'''''''"'
16. Find £{5(0} where
Bit) = sin 2/, < < / TT,
= 0, t>7T.
2 (1 -e-")
ANS. ; : , 5 > 0.
j^ +4
Because our major interest lies in the applicability of the Laplace trans-
formation to the solving of differential equations, it seems natural to seek
information about the transforms of functions that are derivatives of other
functions.
Suppose, for example, we wish to solve the initial value problem
^=
at
e'\ yiO) = i. (1)
gives
(3) yields
§ 53] Transforming initial value problems 155
For the moment, we shall accept (4) and apply it to equation (2). We obtain
sL{y(t)}-y{0) = L{e''}.
sL{y{t)]-- = -, s>2,
2 s — 2
or
XO = \e^'. (6)
It is a simple matter to prove that this function is indeed the solution of the
initial value problem of (1).
or
We shall presume throughout that a solution of (8) exists and that the
solution and its derivatives are sufficiently well behaved to insure the existence
of the needed transforms. Of course, we will need to prove that the function
we obtain on the basis of these assumptions actually is a solution of (8).
L{y" + y}^0,
or
L{y"} + L{y} = 0.
s'L{yU)}-\+L{y{t)}=0,
an equation that can easily be solved for L{y{t)}. We have
L{y(t)]=-^. .
(9)
s + 1
25^+1 _ 1
Exercises
In Exs. 1-10, solve the problem by the Laplace transform method. Verify that
your solution satisfies the differential equation and the initial conditions.
1. / = e';XO)-l.
2. y' = 3e';yiO) = l.
3. y'-y=-e-';y(0) = l.
158 The Laplace Transform [Ch. 10
range / ^ 0.
For such functions, the existence of the integral (1) can be endangered only
at points of discontinuity of F{t) or by divergence due to behavior of the
integrand as ?-» oo.
In elementary calculus we found that finite discontinuities, or finite jumps,
of the integrand did not interfere with the existence of the integral. We there-
fore introduce a term to describe functions that are continuous except for
such jumps.
If the integral of e""F(/) between the limits and t^ exists for every finite
positive to , the only remaining threat to the existence of the transform
e-"F(t)dt (1)
•'0
We know that
e-"dt (2)
-'o
converges for c > 0. This arouses our interest in functions F(t) that are, for
large /(/ ^ /q), essentially bounded by some exponential e'" so that the inte-
grand in (1) will behave like the integrand in (2) for s large enough.
to mean that F(t) is of exponential order, the exponential being e*", as / -> oo.
If F(/) is of exponential order, F{t) = 0{e^'), the last integral in equation (5)
exists because from the inequality (3) it follows that for s > b,
For s > b, the last member of (6) approaches zero as /q -> oo. Therefore the
last integral in (5) is absolutely convergent* for s > b. We have proved the
following result.
Theorem 5: If the integral of e~^'F{t) between the limits and to exists for
every finite positive /q, and if F{t) is of exponential order, F(t) = 0{e'") as
/ ^ 00, the Laplace transform
integrable over that interval. This leads us to the following useful special
case of Theorem 5.
range t ^ 0, and if F{t) is of exponential order, F{t) = (9(e'") as t-> cx), the
Laplace transform L{F{t)} exists for s > b.
exists, the function F(t) is of exponential order, indeed of the order of e'^'.
To see this, let the value of the limit (8) be A' # 0. Then, for / large enough,
\e~'"F(t)\ can be made as close to A^ as is desired, so certainly
\F{t)\<Me", (9)
with M = IK. If the limit in (8) is zero, we may write (9) with M= \.
the function F{t) is not of exponential order. For, assume b exists such that
k"^'"/"(/)| KMe-"',
*3
f^ 3/2 6t 6
= Lim —v- = Lim
. .
= Lim —t-t- =
. . . .
t^ = 0{e'"), t-*oo,
for any fixed positive b.
Consider
exp (t')
Lim —- -. (13)
,^00 e\p{bt)
= Lim .
Thus, no matter what fixed b we use, the limit in (13) is infinite and exp (t^)
classic example showing that functions other than those of class A do have
Laplace transforms is
Fit) = rK
This function is not sectionally continuous in every finite interval in the range
/ ^ 0, because F(t)^ oo as t -^0*. But f^-^ is integrable from to any
positive Iq. Also t~^^- -*0 as /-^ oo, so t~^'^ is of exponential order, with
M= 1 and Z? = in the inequality (3), page 159. Hence, by Theorem 5,
-'o
Therefore
= (-) , .>0,
.00
e-"F(t)dt <M\Jn
X
e-"-e"dt =
M
J^
s>c.
,
(5)
-'n '0 S
s — C
c
Since the right member of (5) approaches zero as 5 -> 00, we have proved the
following useful result.
§ 57] Transforms of derivatives 163
Lim/(s) = 0.
s-*co
From (5) we may also conclude the stronger result that the transform /(5)
of a function F(t) of class A must be such that sf{s) is bounded as 5 -> go.
Exercises
1. Prove that if Fi(t) and FzO) are each of exponential order as t->co, then
FiU) Fzit) and Fi{t) + Fzit) are also of exponential order as / -^ co.
•
2. Prove that if Fi(r) and FjCO are of class A, page 161, then Fi(0 + F^^t) and
Fi{t) FiiO are also of class A.
3. Show that z'' is of exponential order as ^ -> oo for all real x.
In Exs. 4-17, show that the given function is a class A. In these excercises, n
denotes a non-negative integer, k any real number.
4.
164 The Laplace Transform [Ch. 10
, . exp (f')
Lim = 00
,^^ exp (60
for any real b. Since the factors It cos [exp (/^)] do not even approach zero
as / ->• 00, the product F{ exp (
— c/) cannot be bounded as ? -> 00 no matter
how large a fixed c is chosen.
Therefore, in studying the transforms of derivatives, we shall stipulate that
F'{t) dt
-se-"dt Fit)
or
L{F'(t)}=-F(0) + sL{F(t)}. (2)
or
L{F"(t)} = s'fis) - sF(0) - r(0), (4)
L{F^"\t)] = s"f{s) - \
k =
s"- '
-'F^^XO). (5)
Thus
L{F^^\t)] = s^fis) - s^FiO) - sF'(0) - F"(0),
L{F^*\t)} = s'^f(s) - s^F(O) - s^F'iO) - sF"(0) - F<^>(0), etc.
Theorem 10 is basic in employing the Laplace transform to solve linear
differential equations with constant coefficients. The theorem permits us to
transform such differential equations into algebraic ones.
The restriction that F{t) be continuous can be relaxed, but discontinuities
in F{t) bring in additional terms in the transform of F'{t). As an example,
consider an F(t) that is continuous for / ^ except for a finite jump at
t = /i, as in Figure 15. If F(t) is also of exponential order as /-> oo, and if
L{F'(t)}= re-'"F'(t)dt
•'o
FIGURE 15
166 The Laplace Transform [Ch. 10
-exp(-st,)F(t,^)
L{F{t)} ^f{s)
it follows that
If F{t) has more than one finite discontinuity, additional terms, similar to
the last term in (6), enter the formula for L{F'{t)].
it follows that
The integral on the right in (2) is the transform of the function { — t)F{t).
L{F(t)} =f(s)
that
Theorem 13: If F{t) is of class A, it follows from L{F{t)} =f{s) that for
any positive integer n,
to add to our list of transforms with very little labor. We know that
-2ks
L{-t sin kt}.
(s^ + k^)
Thus we obtain
t
= LI- sin kt]. (6)
(s^ + /c^)2 \2k
s
= L{cos kt}
5^ + k^
k'-s'
= L{-t cos kt}. (7)
(s^ + kY
Let us add to each side of (7) the corresponding member of
1
= L (1- sm kt
s^ + k^ [k
to get
s^ + k^ + k^-s^ (\
L - sin kt — t cos kt].
(s^ + k^f \k
(s^
1
+ k^f
L — 3 (sin kt - kt cos kt)\. (8)
168 The Laplace Transform [Ch. 10
.00
Since a' > 0, ^"^ ^ as ^ -> 0. Since x is fixed, e'^P"" -> as /? -> oo. Thus
r(/7 + 1) = nY{n)
= n{n- \)T{n- 1)
r(l)= f"e-TJiS=[-e-^l" = l.
In the integral for r(x + 1) in (2), let us put P = st with s> Q and / as the
new variable of integration. This yields, since / -»• as ^^ and / -^ oo as
P -^ <X),
5^^^^i^
s
= f
''o
e- V c/r, s>0,x>-l,
./2, _ r(«
S
n\)=^n. (7)
The function is completely determined by (1) once the nature of F{t) through-
out one period, ^ ?< co, is given. If F(t) has a transform,
^{^(0}=Z Qxp(-sna>-sl^)F(P+nco)dp.
n= ''o
The integral on the right in (4) is independent of n and we can sum the
series on the right
00 00
J
5;exp(-s«ft>)= i; [exp(-sco)]"
= -
n = n = V — e r^.
170 The Laplace Transform [Ch. 10
Theorem 16: If F{t) has a Laplace transform and if Fit + to) = F{t),
e-"'FiP)dp
L{F(t)} (5)
1 - e"^"
Next suppose that a function //(/) has a period 2c and that we demand
that H{t) be zero throughout the right half of each period. That is.
= 0, c^t <2c.
Then we say that //(/) is a half- wave rectification of ^(/). Using (5) we may
conclude that for the Hit) defined by (6) and (7),
(cxpi-sP)giP)dp
L{Hit)} = '^
. (8)
1 — exp (
— 2cs)
= 0, c < t <2c;
Ht + 2c, c) = il'it, c). (10)
^M
VMMMM W^M¥i'/i'M
2C 3C 4C
FIGURE 16
Jq S
§60] Periodic functions 171
to conclude that
1 1 — exp — sc)
( 1 1
Lmt, c)} (11)
si— exp ( — 2sc) s I + exp ( — sc)
= — 1, c < t <2c;
Q{t + 2c, c) = Qit, c). (13)
QM
2C 3C 4C
-1
FIGURE 17
Exercises
= — l-l
,^15 /Try
2. Show that L{r *'-} , s>0.
3. Use equation (4), page 165, to derive L{sin kt}.
4. Use equation (4), page 165, to derive L{cos kt}.
5. Check the known transforms of sin kt and cos kt against one another by using
Theorem 9, page 164.
6. If/? is a positive integer, obtain L{/V*^'} from the known L{e'"} by using Theorem
§ 60] Periodic functions 173
F(t) = smojt, ^ / ^- ,
CO
= 0, -</< —
7T
CD
2tT
CO
;
ANS.
s^^-oj'
— exp
1
(-a
18. Find L{F(/)} where F{t) = / for < <w / and F(? + co) - F{t).
1 CO exp(— i'Cu) 1 c 0}S\
ANS. -= —+— |l-coth
s^ s 1— exp(— 5co) s^ 2s
Fit)
19. Prove that if L{F(/)} =f{s) and if -^ is of class A,
.p)^i;/<.)*
Hint: Use Theorem 12, page 166.
CHAPTER XX
Inverse Transforms
F(t)=L-'{f(s)}. (2)
174
§ 61] Definition of an inverse transform 175
re-^'F(t)dt=f(s), (3)
•'o
Fi(t) and ^2(0 are identical except at a discrete set of points and differ at
these points, the value of the integral in (3) is the same for the two functions;
their transforms are identical.
Let us employ the term null function for any function N(t) for which
rN{t)dt = (4)
•'o
for every positive /q . Lerch's theorem (not proved here) states that if
L{Fi(/)} = L{F2{t)}, then Fi(0 - F2U) = ^(0- That is, an inverse Laplace
transform is unique except for the addition of an arbitrary null function.
The only continuous null function is zero. If an f{s) has a continuous
inverse F{t), then F{t) is the only continuous inverse of/(5). If /(-s) has an
inverse Fi{t) continuous over a specified closed interval, every inverse that
is also continuous over that interval is identical with Fi(/) on that interval.
Essentially, inverses of the same f{s) diff^er at most at their points of dis-
continuity.
In applications, failure of uniqueness caused by addition of a null function
is not vital, because the effect of that null function on physical properties of
the solution is null. In the problems we treat, the inverse F{t) is required
either to be continuous for / ^ or to be sectionally continuous with the
values of F(t) at the points of discontinuity specified by each problem. The
F{t) is then unique.
A crude, but sometimes effective, method for finding inverse Laplace
transforms is to construct a table of transforms (page 199) and then to use
it in reverse to find inverse transforms.
'
L{cos kt} = (5)
S I /v
Therefore
We shall refine the above method, and actually make it quite powerful,
by developing theorems by which a given f{s) may be expanded into com-
ponent parts whose inverses are known (found in the table). Other theorems
176 Inverse Transforms [Ch. 11
will permit us to write /(5) in alternate forms which yield the desired inverse.
The most fundamental of such theorems is one that states that the inverse
transformation is a linear operation.
Next let us prove a simple, but extremely useful, theorem on the manipula-
tion of inverse transforms. From
we obtain
= Ce-\e'"F{t)-]dt.
''O
L-'{fis-a)} = e'"F{t),
or
Equation (8) may be rewritten with the exponential transferred to the other
side of the equation. We thus obtain the following result.
L-' L ]^
W+ + 4s
^
13)
] = 5L-'l-
\{s + If +
^-^—\ = 5e-"L-'l^^] = 5e-"
9) \s^ + 9/
sin 3/,
s+
EXAMPLE (b): Evaluate L
1
s^ + 6s + 25
We write
s + 1 s+ 1
L-' L-'
s^ + 6s + 25 (s + 3r + 16
Then
s+ 1 5-2
e-^'L-
s^ + 6s + 25 5^+ 16
s
= e L-'
is' + 16) ^ is' + 16
= e~^'(cos4t - ^ sin4r).
Exercises
In Exs. 1-10, obtain L~^{f(s)} from the given f(s).
1
1.
s^ + 2s + 5'
1
2.
3.
4.
5.
6.
7.
8.
9.
"••(.+ 2)-
11. Show that for n a non-negative integer
178 Inverse Transforms [Ch. 11
1 re-
L
{s^aT*') r(m+l)'
13. Show that
1
) 1
L
(5 + ay + o^l o
s \ 1
+ b^) ,
15. For a > 0, show that from L' '{/{$)) = F(0 it follows that
L-Hf{as)) = -F(-
a \a
16. For fl > 0, show that from L-'ifis)} = F{t) it follows that
L-M/(«5 + Z>)}-^exp(-H/'(^
d(i)
V,W-W/>W'/<.W/M';'
FIGURE 18
§ 62] A step function 179
= 1, t^c.
The a function permits easy designation of the result of translating the
graph of F{t). If the graph of
FIGURE 19 FIGURE 20
of the negative argument in a(r — c). Notice that the values of F{x) for
negative jc have no bearing on this result because each value is multiplied by
zero (from the a); only the existence of F for negative arguments is needed.
The Laplace transform of a{t — c)F{t — c) is related to that of F{t).
Consider
FIGURE 21
§ 62] A step function 181
Although the above integrations are not difficult, we prefer to use the a
function.
Since a{t - 2) = for / < 2 and a(/ - 2) = 1 for f ^ 2, we build the yit)
in the following way. The crude trial
works for </< 2, but we wish to knock out the t^ when / > 2. Hence we
write
y2 = t^ - t^a{t - 2).
This gives /^ for / < 2 and zero for / > 2. Then we add the term 6a(/ — 2)
and finally arrive at
The y of (7) is the y of our example and, of course, it can be written at once
after a little practice with the a function.
Unfortunately, the y of (7) is not yet in the best form for our purpose. The
theorem we wish to use gives us
,
L{y(t))
_ = -,-^^-^^^.
2 le'^' 4e-^' le'^'
^"1^)"'^^^"^^°'^^"^^
182 Inverse Transforms [Ch. 11
and
L-^{^^}=4(r-3)a(r-3).
0^/< 1
Equations (10), (11), and (12) are equivalent to equation (9). The graph of
g{t) is shown in Figure 22.
FIGURE 22
Exercises
In Exs. 1-7 sketch the graph of the given function for / ^ 0.
1. a(r-c).
2. ccit - 1) + 3a(/ - 3) - 6a(r - 4).
§ 62] A step function 183
3. (/ - 2)a(/ - 2).
4. sin (t — tt) ait —
•
tt).
5. (/ — TTytxit — tt).
6. t^-(t-2ya(t-2).
7. t^-tMt-2).
In Exs. 8-15, express Fit) in terms of the a function and find L{F(/)}.
8. F(0 = 2, 0</<l, 2 /I 1\
r>l. ANS. - + .-^^---j.
9. F(/) = 6, < < 4,
/ 6
-2/+1, />4. ANS. - +^
i"n)-
10. F(0-/', 0</<l, 2
= 4, />!. ^>^^- -.
+ ^-\- --.--.)
11. Fit)^t\ 0<t<2,
= 4, 2^t^4,
= 0, />4.
2
--^—
ANS.
/4 2\
(^- + -j--e
4
—
12.F(0=/^ 0</<2 2 /3 2\ /53 n
= .-1, 2<.^<3, .ws. --.-^^(- + - + -)+.-3.(___j.
13.
14.
F(/)-e-',
Fit)
^0,
= sin 3t,
0</<2,
t>2.
< < / ^tt,
^^'-
l-exp(-2^-2)
3
7:n
—
+ ^ exp (-^775)
= 0, t>^7r. ANS. ^7-p^ .
s s
17. Evaluate L-'j-—T77- ans. K^ -4)^ exp [-2(/ -4)] a(/ -4).
Fit)=i
(s+l)'
evaluate F(2), F(5), F(7). ans. F(2) = 0, F(5) = 2e-\ Fil) = 8e " *.
19. If Fit) is to be continuous for / ^ and
evaluate F(l), F(3), F(5). ans. F(1) = 1, F(3) = -1, F(5) = -4.
00
Example (a). Section 60. Note that for any specific /, the series is finite; no
question of convergence is involved.
184 Inverse Transforms [Ch. 11
21. Obtain the transform of the half-wave rectification F(/) of sin / by writing
(
— l)"sin / = sin (/ — mi).
Check your result with that in Ex. 17, page 173.
in which F{t) and G(r) are assumed to be functions of class A, we shall obtain
a formula for
L-'{f{s)g{s)}. (2)
On the right in (5) we encounter the product e~^^f{s). By Theorem 19, page
180, we know that from
L-'{f{s)}=Fit) (6)
it follows that
in which a is the step function discussed in Section 62. Equation (7) means
that
With the aid of (8) we may put equation (5) in the form
Since a(/ - j8) = for < < / i? and a(/ - P) ^ \ for t ^ (i, equation (9)
may be rewritten as
fis)g(s)= r re-'^'G(P)F(t-P)dtdp.
•'B
(10)
•'O
In (10), the integration in the /j5-plane covers the shaded region shown in
Figure 23. The elements are summed from t = fi to t = oo and then from
j9 = to i?
= 00 .
FIGURE 23
or
(G(P)F(t - P) dp,
-'o
we have arrived at the desired result, which is called the convolution theorem
for the Laplace transform.
Theorem 20: rfL-'{f(s)} = F(t), ifL-'{g(s)] = G(t), and if F(t) and G(t)
are functions of class A, page 161, then
'{As)g{s)]= (G(l3)F(t-p)dp.
It is easy to show that the right member of equation (12) is also a function
of class A.
Of course F and G are interchangeable in (12), since / and g enter (12)
symmetrically. We may replace (12) by
..(iU.,
Lr.{^>}=fy(«.//i.
Exercises
In Exs. 1-3 find the Laplace transform of the given convolution integral.
2
1.
[
(t-p) sin 2p dp. ANS.
s{s^ + 4)
s
2. !
€-<'-"' cos
p dp. ANS. - .,, , ,-
Jo (5+l)(5^+l)
3.
f
Jq
((-pye''dp. ANS. —
s^is -
-.
1)
In Exs. 4-6, find an inverse transform of the given f(s) using the convolution
theorem.
§ 64] Simple initial value problems 187
5. — 1
s\s —
-. ANS. e'-/-l,
1)
1
^« .,,.., • ANS. Ksin / - / cos 0-
at its best when the boundary conditions are actually initial conditions, those
giving the value of the function and its derivatives at time zero.
Aoi
s u(s) — s + B u(s) = , ,
from which
"(^) = 2
+
, o2
/?"
+ /„2
(s^ + ,
^^ +
o2^r.2 , ,,2^
w')
• (2)
We need the inverse transform of the right member of (2). The form of
that inverse depends upon whether P and co are equal or unequal.
If CO # p,
_ s Aoj I 1 \
I
_ s A / (Dp cop \
A
y(t) = cos pt + -;;7^2 2^ (P ^in cot - (o sin pt). (3)
s Ap
"(^) = ,2 , o2 + , 2 , d2x2 •
C"^)
'
= (sin pt - pt cos pt).
1(5^ + P'-fj IP'
Hence, for w= P,
A
>•(/) = cos pt + ^^ (sin pt - pt cos ^50- (5)
2p'
It is a simple matter to show that this function is indeed the solution of the
given initial value problem.
Note that the initial conditions were satisfied automatically by this method
when Theorem 10 was applied. We get, not the general solution with arbitrary
constants still to be determined, but that particular solution which satisfies
the desired initial conditions. The transform method also gives us some insight
into the reason that the solution takes different forms according to whether
CO and P are equal or unequal.
3
s^y(s) - 4s - 2 + 2[sv(s) - 4] + y(s) =
(s + ly
or
4s + 10 3
We may write
4(5 + 1) + 6 ,
(5 + 1)^ (s + iy
§ 64] Simple initial value problems 189
,.4 6 ^
and thus verify that the x of (8) satisfies both the differential equation and
the initial conditions of the problem (6). Such verification not only checks
our work but also removes any need to justify temporary assumptions about
the right to use the Laplace transform theorems on the function x(t) during
the time that the function is still unknown.
= 4, / > 1.
equation has discontinuous derivatives makes the use of the classical method
of undetermined coefficients somewhat awkward, but such discontinuities do
not interfere at all with the simplicity of the Laplace transform method.
In attacking this problem, let us put L{x{t)} = h(s). We need to obtain
L{il/{t)}. In terms of the a function we may write, from (10),
4 4e~'
s s
4 4e-'
s^'his) -S-0 + 4/i(s) = ^ -
from which
s 4 4e
S2 + 4 5^(5^ + 4) ^2(^2 ^ 4)
Now
5^(5^ + 4) s^ s^+4'
so (12) becomes
x{t) = cos 2r + - / i sin It - [(/ - 1) - i sin 2(/ - 1)] y.{t - 1). (14)
Therefore x(0) = 1 and x'(0) = 0, as desired. Also, from (14) and (16), we get
,, As + B f(s)
U(S) = -^ r +
^
-^ ? . (lo)
§ 64] Simple initial value problems 191
To get the inverse transform of the last term in (18), we use the convolution
theorem. Thus we arrive at
B 1
c(t) =
x{t)= A cos kt + - sin kt +- f F(t - (i) sin A/? dp,
k k Jq
or
B
x(t) =A cos kt +- sin kt +- \
F((i) sin k{t - ft) d^. (19)
k k ^0
Verification of the solution (19) is simple. Once that check has been per-
formed, the need for the assumption that F{t) has a Laplace transform is
removed. It does not matter what method we use to get a solution (with
certain exceptions naturally imposed during college examinations) if the
validity of the result can be verified from the result itself.
In this example the boundary conditions are not both of the initial condi-
tion type. Using x, rather than /, as independent variable, let
We know vv(0) = — 3, but we also need w'{0) in order to write the transform
of w"{x). Hence we put
h'(0) =B (22)
from which
,, -3(s+l) + B-3 1
1 2 12 1
so
g(s)
_= 1
-2
2 1
-— B-2
+ ——rr2, (24)
s s s + 1 (s + 1)^
We have yet to impose the condition that n(l) = — 1. From (25) with x= I,
we get
-1 = 1 -2-e-* +{B-2)e-\
so 5= 3.
Exercises
In Exs. 1-14, solve the problem by the Laplace transform method. Verify that
your solution satisfies the differential equation and the initial conditions.
3. y"(,t) - y{t) =- 5 sin 2t\ y{0) = 0, y'{Qi) = 1. ans. y{t) = 3 sinh — sin 2t. /
4. y'XO + 6/(0 + 9X0 = 6t^e-^'; y(0) = 0, /(O) = 0. ans. y{t) = i/^e" ^'.
ANS. M(0=4+5e-'-2e-^'.
9. y'Xx) + 9yix) = 40^^; ^(0) = 5, ^'(0) = -2.
ANS. y{x) ^ Ae"" + cos 3x — 2 sin 3;c.
10. y"{x) + ^(jc) = Ae" ><0) = 0, ^'(O) = 0.
;
ans. y{x) = 2(e* - cos x - sin x).
11. x"{t) + 3x'(0 + 2x{t) = 4t^; x{0) = 0, a-'(O) = 0.
ans. x(0 = 2r^-6/ + 7-8e-' + e-^'.
12. jc"(0 - 4a:'(0 - Ax{t) = 4 cos 2r; .r(0) = 2, x'(0) = 5.
ANS. x(0 = 2e^'(l + 0-2 sin 2t.
13. x"(/) + x(r) = F(f); x(0) = 0, a-'(O) = 0, in which
Fit) =4, ^ / ^ 2,
= + 2, t t>2.
ANS. x(/) = 4 - 4 cos + / [(t - 2) — sin (r — 2)]a(r — 2).
14. x"(0 + xiO = H(0; xiO) - 1, x'(0) - 0, in which
Hit) = 3, ^ ^ 4, /
= 2r - 5, > 4. /
CF(/) = 1, 0^r<7r/2,
= 0, / ^ 77/2.
§ 64] Simple initial value problems 193
17. Compute yilTr) and y{2 + ^tt) for the function y{x) that satisfies the initial
value problem
18. Compute x(_l) and x(4) for the function x{t) that satisfies the initial value prob-
lem
In Exs. 23-42, use the Laplace transform method with the realization that these
exerciseswere not constructed with the Laplace transform technique in mind.
Compare your work with that done in solving the same problems by the methods of
Chapter 7. See pages 119-120.
L{F{t)] =f{s)
and
L{G{t)} ^ g{s),
then
LUF(p)G{t-^)dp\=f{s)g(s)
The relation (1) suggests the use of the Laplace transform in solving equations
that contain convolution integrals.
The integral in (2) is in precisely the right form to permit the use of the
convolution theorem. Let
L{F(t)} =/(5).
Then, since
L{smt} = -j-—,
s^ + 1
§ 65] Special integral equations 195
4 3/(5)
f(') = -2-^^-
S^ 5^+1 (3)
We need to obtain/(5) from (3) and then F{t) from/(5). From (3) we get
(i+7TtV(^^ = ?'
or
4(5^+1) 1 ^
s\s^ + 4) s^ s' +4
Therefore
F(t) = L-'{\ +
or
F(t) = t+ fsin2/. (4)
That the F{t) of (4) is a solution of equation (2) may be verified directly.
Such a check is frequently tedious. We shall show that for the F of (4), the
right-hand side of equation (2) reduces to the left-hand side of (2). Since
RHS = 4t-3 f
()6 + f sin 2p) sin (/ - p) dfi,
from which
+ 9 f
cos 2p cos {t - p) dp.
196 Inverse Transforms [Ch. 11
or
RHS = - r
f sin 2r - 3[sin (t - p)] + f [cos {t + P) + cos (t - 3P)\dp.
J
RHS = - f
f sin 2f + 3 sin r + f [sin (t + P) - ^ sin (t - 3p)\
1-
-'o
or
RHS = + ism2t = F(t) = LHS,
t
as desired.
It is important to realize that the original equation
An essential ingredient for the success of the method being used is that
the integral involved be in exactly the convolution integral form. We must
have zero to the independent variable as limits of integration, and an inte-
Again the integral involved is one of the convolution type with x playing
the role of the independent variable. Let the Laplace transform of g{x) be
some as yet unknown function h(z)
/7(z)=-3--2-,
z z
§ 65] Special integral equations 197
from which
(-M Kz) = \,
or
=
1 z^ + l-z'
/i(z)
z(z^ + 1) + 1)
ziz"" z z^ + 1
Then
or
g(x) = 1 - cos X. (7)
= i^^ - f(^
- y){y - sin y)] - ! (y - sin >;) dy
Jo *'o
Exercises
In Exs. 1^, solve the given equation and verify your solution.
1. F(t) = +
1 2 f F(t-^)e-^^ d^. ANS. Fit) = 1+2/.
t
2. F(r) = +
1
f
F((8) sin (t - ^) d^. ans. F(0 = 1 + i^'.
In Exs. 5-8, solve the given equation. If sufficient time is available, verify your
solution.
8. Fit) = +2
1 f F(/ - iS) cos ]8 d^. ans. Fit) = + 2te'.
1
'o
11. gix) - e-^ - 2 [ gi^) cos (x - jS) JjS. ans. ^(x) = ^-^(1 - x)\
•'o
12. K0 = 6r + 4f i^-tYyi^)d^.
F'(/) = +
/ f Fit-^) cos ]8 ^^. ANS. Fit) = 4 + f r^ + 2^ r*.
14. Solve the following equation for Fit) with the condition that F(0) == 0:
r
F'it) = sin +
/ f Fit - iS) cos ^3 d^. ans. Fit) - it\
15. Show that the equation of Ex. 3 above can be put in the form
Differentiate each member of (A) with respect to t and thus replace the integral
equation with a differential equation. Note that F(0) = 0. Find Fit) by this
method.
f Fit-^)e-^d^^t
by two methods; use the convolution theorem and the basic idea introduced in
Ex. 15. Note that no differential equation need be solved in this instance.
TABLE OF TRANSFORMS
Whenever /; is used, it denotes a non-negative integer.
The range of validity may be determined from the appropriate text
material. Many other transforms will be found in the examples and exercises.
Table of transforms 199
As -a)
200 Inverse Transforms [Ch. 11
fXs)^F{LU)}
CHAPTER LJi
Applications
Q= kc (1)
201
202 Applications [Ch. 12
^
§ 66] Vibration of a spring 203
w w
- x"(t) + bx'(t) + kx(t) = - F(t). (3)
9 g
At time zero, let the weight be displaced by an amount Xq from the equi-
librium point and let the weight be given an initial velocity t'o • Either or
both of Xq and Vq may be zero in specific instances. The problem of deter-
mining the position of the weight at any time t becomes that of solving the
initial value problem consisting of the differential equation
w w
- x"(0 + hx'{t) + kx{t) = - F(0, for t > 0, (4)
9 9
w w
We may choose ^ > and we know y ^ 0. Note that 7 = corresponds to a
negligible retarding force.
and obtain from the problem (5) and (6) the transformed problem
"(s) =
SX0
s^
2
+ V0 + 27x0
+ ^2ys +n^T-
.
P""
—+
s^ +
/(s)
2ys + p^
'
which we write as
We shall now consider separately the various situations that arise according
to the choice of parameters P and y and the forcing function F{t) in this
F(t)^A, 0<t<to,
= 0, t>to;
Fit) = 0, 0<t<to,
= A, to<t<ti,
= 0, t>ti;
all of our present work carries over to simple circuit problems with new
meanings attached to the parameters and functions involved. In circuit theory
we shall add to the forms of Fit) listed above both Fit) equal to a constant
and F(t) equal to the square-wave function that appeared in Section 60.
§ 67] Undamped vibrations 205
Vq
Xq cos Pt + -f:
sin pt,
we i^refer to obtain x(/) without a convolution integral for the specific types
of F{t) which occur often. Equation (2) was easily obtained but it leaves all
the labor of simplification to the unfortunate user who has a particular F{t)
with which to deal.
EXAMPLE (a): Solve the spring problem with no damping but with a
forcing function F{t) =A sin cot.
w w
- x"(0 + ^^(0 = - ^ sm cot
9 9
and is readily put into the form
X(0) = Xo ,
X'(0) = Vq .
Aco
s^m(s) — sxq — Vo + P u(s) = 2'
S^ + O)
206 Applications [Ch. 12
or
The last term in (5) will lead to different inverse transforms according to
whether to = /? or w# ^. The case oj = jS leads to resonance, which will be
discusssed in the next section.
U CO ^ P, equation (5) yields
-1 n Aio ^
+ ——^
.
That the x of (7) is a solution of the problem (3) and (4) is easily verified. A
study of (7) is simple and leads at once to conclusions such as that x(/) is
bounded, etc. The first two terms on the right in (7) yield the natural har-
monic component of the motion, the last two terms form the forced com-
ponent.
for the mass. We wish to use the value g = 2>2 ft. per sec. per sec. and we must
use consistent units, so we put all lengths into feet.
First we determine the spring constant k from the fact that the 12-pound
weight stretches the spring 6 inches, \ ft. Thus \2 ^ \k so that k = 2A lb.
per ft.
At time zero the weight is 4 inches (3 ft.) below the equilibrium point, so
x(0) = \. The initial velocity is negative (upward), so x'(0) = — 2. Thus our
problem is that of solving
from which
u(s) = -^ .
Then
4/) = 1 cos 8/ - i sin 8/. (10)
VW+df = A
that is, the weight oscillates between points 5 inches above and below E.
The period is ^n sec.
EXAMPLE (c): A spring, with spring constant 0.75 lb. per ft., lies on a
long smooth (frictionless) table. A 6-lb. weight is attached to the spring and
is at rest (velocity zero) at the equilibrium position. A 1.5-lb. force is applied
to the support along the line of action of the spring for 4 seconds and is then
removed. Discuss the motion.
We must solve the problem
in which
g
s^u(s) + 4uis) = - (I -e"^0,
s
or
8(1 - e-'^)
U(S) — ;5
^ ^
sis^ + 4)
Of course, the solution (13) can be broken down into the two relations
From (13) or (14) we see that in the range < < / 4, the maximum deviation
of the weight from the starting point is x = 4 ft. and occurs SLt t = ^n — 1.57
sec. At / = 4, .Y = 2.29 ft. as shown above. For / > 4, equation (15) takes over
and thereafter the motion is simple harmonic with a maximum x of 3.03 ft.
= 2^2 Jl - cos 8
= 272.2910 = 3.03.
Example (c) is one type of problem for which the Laplace transform
technique is particularly useful. Such problems can be solved by the older
classical methods, but with much less simplicity and dispatch.
68. Resonance
The solution of (1) was obtained by putting L{x{t)} = u{s) and getting
SXq + Vq Aco
"(•^) = .2 , 02 + /.2 , o2u- 2 , ..2> • (2)
sxq + Vq Ap
K^) = -?i^T^ + 7Tr^?^- (4)
i-'|(jr^| = 2j3(si"/"-/"cos/!/).
Therefore (4) leads us to the solution
V A
x(t) = Xo cos Pt + ^ sin pt + —^ (sin pt - pt cos pt). (5)
Exercises
1. A spring is such that a 5-pound weight stretches it 6 inches. The 5-pound weight
is attached, the spring reaches equilibrium, then the weight is pulled down 3
inches below the equilibrium point and started off with an upward velocity of
6 ft. per sec. Find an equation giving the position of the weight at all subsequent
times. ANS. x = i(cos 8/ — 3 sin 8/).
2. A spring is stretched 1.5 inches by a 2-pound weight. Let the weight be pushed
up 3 inches above E and then released. Describe the motion.
ANS. X = — icos 16/.
3. For the spring and weight of Ex. 2, let the weight be pulled down 4 inches
below E and given a downward initial velocity of 8 ft. per sec. Describe the
motion. ans. x = cos 16/ + ^ sin 16/.
J
4. Show that the answer to Ex. 3 can be written x = 0.60 sin (16/+ ^) where
= Arctan f
(f)
9. For the motion of Ex. 8, find the first four times at which stops occur and find
the position at each stop. ans. t = ttjS, -njA, 1, 37r/8 (sec.)
pulled down to a point 4 inches below the equilibrium point and given an
initial downward velocity of 4 ft. per sec. An impressed force of 360 cos At
pounds is applied. Find the position and velocity of the weight at time t = tt/S
seconds. ans. At r = 77/8(sec.), x = —f (ft.), y = —8 (ft. /sec.).
12. A spring is stretched 3 inches by a 5-pound weight. Let the weight be started
from E with an upward velocity of 12 ft. per sec. Describe the motion.
ANS. a: = —1.06 sin 11.3/.
13. For the spring and weight of Ex. 12, let the weight be pulled down 4 inches
below E and then given an upward velocity of 8 ft. per sec. Describe the
motion. ans. x = 0.33 cos 1 1.3/ — 0.71 sin 1 1.3/.
14. Find the amplitude of the motion in Ex. 13. ans. 0.78 ft.
15. A 20-pound weight stretches a certain spring 10 inches. Let the spring first be
compressed 4 inches, and then the 20-pound weight attached and given an
initial downward velocity of 8 ft. per sec. Find how far the weight would drop.
ANS. 35 in.
16. A spring is such that an 8-pound weight would stretch it 6 inches. Let a 4-
pound weight be attached to the spring, pushed up 2 inches above its equilibrium
point and then released. Describe the motion. ans. x ^ —\ cos 11.3/.
17. If the 4-pound weight of Ex. 16 starts at the same point, 2 inches above E,
but with an upward velocity of 15 ft. per sec, when will the weight reach its
lowest point ? ans. At / = approximately 0.4 sec.
18. A spring is such that it is stretched 4 inches by a 10-pound weight. Suppose
the 10-pound weight to be pulled down 5 inches below E and then given a
downward velocity of 15 ft. per sec. Describe the motion.
ANS. X = 0.42 cos 9.8/ +1.53 sin 9.8/
= 1.59 cos (9.8/ — ^), where = Arctan 4> 3.64.
19. A spring is such that it is stretched 4 inches by an 8-pound weight. Suppose the
weight to be pulled down 6 inches below E and then given an upward velocity
of 8 ft. per sec. Describe the motion.
ANS. x = 0.50 cos 9.8/ -0.82 sin 9.8/.
20. Show that the answer to Ex. 19 can be written jc= 0.96 cos (9.8/ + where <f))
<f>
= Arctan 1.64.
21. A spring is such that a 4-pound weight stretches it 6 inches. The 4-pound
weight is attached to the vertical spring and reaches its equilibrium point. The
weightis then (/ = 0) drawn downward 3 inches and released. There is a simple
harmonic exterior force equal to sin 8/ impressed upon the whole system. Find
§ 69] Dumped vibrations 211
the time for each of the first four stops following / = 0. Put the stops in chrono-
logical order. ans. / -^ 7r/8,
J, 7r/4, Stt/S (sec.).
22. A spring is stretched 1.5 inches by a 4-pound weight. Let the weight be pulled
down 3 inches below equilibrium and released. If there is an impressed force
8 sin 16/ acting upon the spring, describe the motion.
ANS. X = 1(1 — 8/) cos 16/ + J sin 16/.
23. For the motion of Ex. 22, find the first four times at which stops occur and find
the position at each stop. ans. / = i, 77/I6, Tr/8, 377-/16 (sec.) and
X = +0.11, +0.14, -0.54, +0.93 (ft.), respectively.
and obtained
P'-y' = 5\ (3)
Then (2) becomes
and we arrive at
in which iAi(/) is an inverse transform of the last term in (4). The function
i//i(/) can always be written in terms of F(t) by the convolution theorem,
but that leads to undesirable complications for the most common choices
of F{t). In the solution (5), the presence of the damping factor e~^' shows
that the natural component of the motion approaches zero as / -> 00.
212 Applications [Ch. 12
"(i) = r—
(s+
— 71
7)2
+
(s + y)'
(6)
. .
= Xo(s + r) + % + 7^0 ^
fis)
"(5) 7^; h (9)
(5 + 7)^ - <r'
:
(s + 7)2-(t''
in which 1/^3(0 is an inverse of the last term in (9). By (8) we see that y > a,
so that once more the natural component of the x{t) in (10) approaches zero
as /^ 00.
Suppose for the moment that we have F{t) = 0, so the natural component
of the motion is all that is under consideration. If /? > 7, equation (5) holds
and the motion is a damped oscillatory one. If /? = 7, equation (7) holds and
Critically damped
FIGURE 26
§69] Damped vibrations 213
EXAMPLE: Solve the problem of Example (b). Section 67, with an added
damping force of magnitude 0.6 \v\. Such a damping force can be realized by
immersing the weight B in a thick liquid.
The initial value problem to be solved is
^-^-^
40^ ^
^L-^f
^
]
\(s + 0.8)2 _^ 53 3^1
5-5.2
= iexp(-0.80L"
[s~ + 63.36)
FIGURE 27
214 Applications [Ch. 12
Exercises
1. A certain straight-line motion is determined by the diflferential equation
- + 2y-+169.=0
d^x dx
weight starts \ foot below the equiUbrium point with an imparted upward
velocity of 3 ft. per sec. Find a formula for the position of the weight at time t.
downward velocity of 5 ft. per sec. The motion takes place in a medium which
furnishes a damping force of magnitude \\v\ at all times. Find the equation
describing the position of the weight at time /.
5. A spring is such that a 4-pound weight stretches the spring 0.4 feet. The 4-pound
weight is attached to the spring (suspended from a fixed support) and the system
is allowed to reach equilibrium. Then the weight is started from equilibrium
position with an imparted upward velocity of 2 ft. per sec. Assume that the
motion takes place in a medium which 'furnishes a retarding force of magnitude
numerically equal to the speed, in feet per second, of the moving weight. Deter-
mine the position of the weight as a function of time.
ANS. X = — it-""*' sin 8/.
6. A spring is stretched 6 inches by a 3-pound weight. The 3-pound weight is
attached to the spring and then started from equilibrium with an imparted
upward velocity of 12 ft. per sec. Air resistance furnishes a retarding force
equal in magnitude to 0.03| v\ . Find the equation of motion.
ANS. x= -1.5e-°**'sin 8r.
released from a point 3 inches below the equilibrium point, find its position for
/>0. ANS. JC = (i + 4/>"^' — icos 8r.
§ 69] Damped vibrations 215
9. For Ex. 8, fmd the times of the first three stops and the position (to the nearest
inch) of the weight at each stop.
ANS. /i = 0.3 sec, xi = — 12 in. tz = 0.8 sec,
;
14. Show that whenever > (sec), the solution of Ex. 13 can be replaced (to the
/ 1
per sec. Assuming a damping force twice the magnitude of the velocity, des-
cribe the motion and sketch the graph at intervals of 0.05 sec. for ^ / ^ 0.3
(sec). ANS. x = ^e~^'(l —6t).
19. An object weighing w pounds is dropped from a height /; feet above the earth.
At time /(sec) after the object is dropped, let its distance from the starting point
be jc (ft.), measured positive downward. Assuming air resistance to be negligible,
show that X must satisfy the equation
w d^x
=w
gdt^
as long as X < h. Find x. ans. x = Igf^.
20. Let the weight of Ex. 19 be given an initial velocity Do . Let v be the velocity at
time /. Determine v and x. ans. v ^ gt -\- v^^^x — \gt^ + vq t.
21. From the results in Ex. 20, find a relation that does not contain / explicitly.
ans. y^ = Vq- + Igx.
216 Applications [CIi. 12
22. If air resistance furnishes an additional force proportional to the velocity in the
motion studied in Exs. 19 and 20, show that the equation of motion becomes
w d^x dx
g dt^ dt
when / =: 0, X = = vo.
and v (B)
d^x dx
+
^
6 — + 25x = 0.
dt' dt
If the particle started at x = with an initial velocity of 12 ft. per sec. to the
left, determine: (a) x in terms of /; (b) the times at which stops occur; and (c)
vertical plane. Let a weight B (the bob) of w pounds be attached to the free
end of the rod, and let the weight of the rod be negligible compared to the
Figure 28, of the rod at time / (sec). The tangential component of the force
w (lb.) is IV sin 9 and it tends to decrease 0. Then, neglecting the weight of the
rod and using S = CO as a measure of arc length from the vertical position,
fe + ^g sin = 0. (2)
2
~d?^C
§70] The simple pendulum and the deflection of beams 217
FIGURE 28
+ ^0 = 0;
" P^=^. (3)
dt""
'
'
'
C
The solution of (3) with pertinent initial conditions gives usable results
whenever those conditions are such that 9 remains small, say \0\ < 0.3
(radians).
29. Denote distance from one end of the beam by x, the deflection of the
beam by ;•. If the beam is subjected to a vertical load lV(x), the deflection y
must satisfy the equation
Such problems can be solved by integration with the use of a little algebra.
There are however two reasons for employing our transform method in such
problems. Frequently the load function, or its derivative, is discontinuous.
Beam problems also give us a chance to examine a useful device in which a
problem over a finite range is solved with the aid of an associated problem
over an infinite range.
FIGURE 29
The student should verify that the right member of (2) is the stipulated load
function
w,
c
for O^xS c, (5)
To apply the transform technique, with x playing the role for which we
usually employ t, we need first to extend the range of x so it will run from
to 00. That is, instead of the problem (2), (3), (4), we shall solve the problem
consisting of
and the conditions (3) and (4). In (6) the function H(x) is to be chosen by
us except that H(x) must agree with IV (x) over the range < a: < 2c. The
solution of the problem (6), (3), (4) will then be used only in the range
^ A" ^ 2c. Of the various choices for Hix), it seems simplest to use
c
c)oc{x - c)], for < X < 00. (7)
That is, in practice we ordinarily retain the equation (2) and merely extend
the range from <x< 2c to < a" < oo. The student must, however, keep
in mind that we cannot apply the Laplace operator to the function lV(x) of
(5), since that function is not defined over the entire range <x< oo. We
shall solve (6) and conclude that the solution is valid for (2) on the range
^X^ 2c, over which (2) and (6) are identical.
Let
L{EIy(x)} = uis);
To transform EIy^^\x) we need to use the values of EIy{x) and its first three
derivatives at x = 0. From (3) we know that
ElyiO) = 0, Ely'(0) = 0.
Put
Ely"(0) = A, EIy"'iO) = B. (9)
L{H(x)] - — L{c - X +
c
(x - c)a(x - c)}
— [---.+
Thus the differential equation (6) is transformed into
VV'o /c 1 c~^
^
s'^u(s) -s^-0-s^-0-s-A-B = —\
^ -J
3+ -T
c \s s^ s'^
220 Applications [Ch. 12
^^ A vvo/cB e-«\ 1
"(s) = ^+-i+—c \s---5+ -^
s"* s s^ s /
• (10)
Ely'ix) = Ax + ifix^ + ^
24c
[4cx^ - x^ + (x - 4)*a(x - c)], (12)
With the aid of equations (1 1) through (14) we can now determine A and B
to make the y satisfy appropriate conditions at 2c, whether the x — beam be
free, imbedded, or pin-supported there. In our example the beam is to be
Using (13) and (14), and a Httle work, we find that (4) requires A = i\yoC^^
B = —^WqC. We are thus led to the solution
Exercises
1. A clock has a 6-inch pendulum. The clock ticks once for each time that the
pendulum completes a swing, returning to its original position. How many times
does the clock tick in 30 seconds? ans. 38 times.
§ 70] The simple pendulum and the deflection of beams 221
y'(0-4>(0-2x'(0=-2, (2)
Let L{x{t)} = u(s) and L{y(t)} = v{s). Then, application of the Laplace
operator transforms the problem into that of solving a pair of simultaneous
algebraic equations
222
§ 71] The Laplace transform method 223
-25^ + 4
"''^ - s(s' + l)(s' + 4)-
('>
Seeking the inverse transforms of u and r, we first expand the right members
of (6) and (7) into partial fractions
.
u(s)
s
=
1
T
-\ — 5
^
4
, (8)
1 2s s
K^)-T-:^T-r
s s^ + I
+ :2-^-
s^ +
4' (9)
Since x{t) = L~^{u(s)} and y(t) = L~^{r(5)}, we get the desired result
which is easily verified by direct substitution into (1), (2), and (3).
On the other hand, the use of the transform theory can help us gain insight
into the general theory of systems of linear differential equations. We illus-
dy
= x + G(0. (13)
y^
Here we presume that the transforms of the functions x{t), y{t), F{t), and
G{t) all exist and are given by u{s), v{s), f(s), and g(s), respectively. We then
have
su{s) - fi = lis) +f(s), (14)
where Cj and Cj represent the initial values of x{t) and y{t). We can rewrite
equations (14) and (15) in the form
Using Cramer's rule, the solution of the algebraic system (16) and (17)
may be written
ci+f(s) -1 f(s) -1
=
C2 + g(s) S 9(s) s
u{s)
s'-l
v{s) =
§711 The Laplace transform method 225
s c,
v(s) =
226 Systems of Equations [Ch. 13
5. The equations of Example (a) of Section 71 with initial conditions x{0) =0,
jc'(0) = 0,j(0) = l,/(0)-0.
x'(0) = 2,>'(0)-l,/(0)-0.
ANS. x(t) = —t+ 15 cos — 5 sin — 6 cos 2/ + 4 sin 2r,
/ /
7. + yV)-y{t) = 0,
x'V)
2xV)- xit) + z'(/) - ziO - 0,
xV) + 3xit) + /(/) - 4y{t) + 3z(0 = 0;
x(0) = 0, x'(0) = y(0) - 0, 2(0) = 0.
1 ,
14. For the initial value problem of Ex. 13, state a reasonable definition for a first-
xXO + 2yit) = 0,
x"(t) + 2y'(t) + 2yit) = 2e';
x(S)) = l,x'(0)=0,y(0) = 0.
(a) Show that the Laplace transform method produces
X = 3 — 2e', y = e\
(b) Verify that these functions satisfy the differential equations but do not
satisfy the initial conditions.
(c) By elementary elimination, show that a solution of the system of diff'eren-
tial equations has the form
X = Ci — 2e\ y = e\
and thus the initial conditions given are not compatible with the system of
equations.
2y' - v" + Av = 2,
or
or
(Z)^ + 5/)' +A)y= -Ax. (4)
and
y— —x + b^ cos X + 6, sin x + 63 cos 2x + 64 sin 2a'. (6)
The a's and 6's have yet to be chosen to make (5) and (6) satisfy the original
equations, rather than just the equations (3) and (4), which resulted from the
original ones after certain eliminations were performed.
Combining (by substitution) the v of (5) and the y of (6) with the first
-2b 2- = 0, 5^1
-5^3+ 10^4 = 0,
-5^>4- 10^3 = 0.
Relations between the a's and b\ equivalent to the relations (8) follow from
substitution of the r of (5) and the y of (6) into the second equation of the
system (2).
y——x+ |-«2 cos A — yfl'i sin A + 2^4 cos 2a — 2^3 sin 2a',
(D' - 1) 5D (D^-l) A
2D -{D^-A) 2D 2
{D^-\){2)-2D{x),
§ 72] The differential operator method 229
The V of equation (5) may now be used in (12) to compute the y which was
given in the solution (9). In this method of solution there is no need to
obtain (4), (6), (7), or (8).
Both methods outlined here are available whenever the system is linear
with constant coefficients. Such systems appear naturally in the study of
electric networks which will be discussed in the next chapter.
Systems of nonlinear equations and systems of linear equations with
variable coefficients will not be studied in this book.
Exercises
In Exs. \-4, use the differential operator method.
i2D+l)y + iD^~4)v=0.
ANS.^ + Oie" + aze'" + 03 cos x + a4. sin x,
V 1
z =- 7 + 7x + 2aie* - 1^3 e- ^^
10. (/)' + I>^ - 1)« + (/)' + 2Z)^ + 3Z) + l)y = 3 - a: ,
73. Circuits
t (sec. ) = time
Q (coulombs) = quantity of electricity; e.g., charge on a capacitor
/ (amperes) = current, time rate of flow of electricity
E = electromotive
(volts) force or voltage
R — resistance
(ohms)
L (henrys) = inductance
C (farads) = capacitance.
By the definition of Q and / it follows that
I(t) = Q'(t).
231
232 Electric Circuits and Networks [Ch. 14
(a) The sum of the currents into {or away from) any point is zero,
and
(b) Around any closed path the sum of the instantaneous voltage drops in a
specified direction is zero.
m
mO)
"^C
FIGURE 30
For a circuit, Kirchhoff's current law (a) indicates merely that the current
is the same throughout. That law plays a larger role in networks, as we shall
see later.
To apply Kirchhoff's voltage law (b), it is necessary to know the contribu-
tions of each of the idealized elements in Figure 30. The voltage drop across
the resistance is RI, that across the inductance is Ll'it), and that across the
capacitor is C~^Q{t). The impressed electromotive force E{t) is contributing
a voltage rise.
without charge, ^(0) = 0. From Kirchhoff's law (b), we get the differential
equation
in which
/(O = QV)- (2)
§ 73] Circuits 233
From (5) and (6) we find either or both of i(s) and q(s),
se(s)
i(s) = —, 5lj!
(7)
q{s) = —, — ^-^
J . (8)
Then I(t) and Q(t) are obtained as inverse transforms from (7) and (8).
H.'
FIGURE 31
234 Electric Circuits and Networks [Ch. 14
Figure 31, let the switch be closed at / = 0. At some later time, t — 1q, the
Let the transform of /(/) be i{s). We know the transform of E{t). Therefore
from which
£[1 -expC-fps)]
i(s) (13)
s{sL + R)
Now
= i(i
1
sisL + R) R\s s
'—I
+ RL-'J
RL~
so
E /I
i{s) = [1 -Q\p(-toS)'].
R\s s + RL
Therefore
E R R
I(t) =- I - a(f - t
o)
- exp -- f
j
+ exp -- (r - to)\cc{t - t^) (14)
^
R I
The student should verify (14) and show that it can be written
Rt
ForOSt^to, ^(0 = — exp
-^ 1
T (15)
EXAMPLE (a): Determine the character of the current /,(/) in the net-
work having the schematic diagram in Figure 32, under the assumption that
when the switch is closed the currents are each zero.
E-=
FIGURE 32
h=h + h. (1)
Application of the voltage law to the circuit on the left in Figure 32 yields
RJ,+L,l',^E. (2)
Still another equation can be obtained from the circuit on the right in
Figure 32:
Equation (4) also follows at once from equations (2) and (3); it may be used
instead of either (2) or (3).
Let L{I^{t)) = ii,{s) for each of A' = 1, 2, 3. Then use of the operator L
transforms our problem into the algebraic one of solving the equations
(7)
s
Since we desire only i^is), let us use determinants to write the solution
§ 74] Simple networks 237
h(s) = — +—+
5 s ai
+ --
S + ^2
, (12)
so
/i(r) = Aq + Ai exp(-fliO + ^2 exp(-a2 0-
EXAMPLE (b): For the network shown in Figure 33, set up the equations
for the determination of the currents /j, I2, I3, and the charge Q^ . Assume
that when the switch is closed all currents and charges are zero. Obtain the
transformed problem.
FIGURE 33
(13)
(16)
'
dt
Our problem consists of the four equations (13) through (16) with the
initial conditions that
Eco
/?i/i+sL2/2=^——5, (19)
s + (a
1 £to
Rxh + «3 '3 + TT ^3 = -T—-T' (20)
C3 5 + OJ
h= sgi. (21)
1. For the RL circuit of Figure 3 1 , page 233, find the current / if the direct current
element E is not removed from the circuit.
ANS. I = ER-'[l -expi-RtL-')].
2. Solve Ex. 1 if the direct-current element is replaced by an alternating-current
element Ecos tot. For convenience, use the notation
Z' = R^ + (coC)-',
FIGURE 34
§ 74] Simple networks 239
6. In Figure 30, page 232, find /(/) if E(t) = 60 volts, /? - 40 ohms, C= 5(10)"'
farad, L = 0.02 henry. Assume /(O) = 0, Q{0) = 0.
ANS. / = 3000/ exp (- 1000/)(amp.).
7. In Ex. 6, find the maximum current. ans. /max = 3e"' (amp.).
In Exs. 8-11, use Figure 30, page 232, with £"(0 == ^sin cot and with the follow-
ing notations used to simplify the appearance of the formulas:
R
^=2Z'
240 Electric Circuits and Networks [Ch. 14
FIGURE 35
£-=
FIGURE 36
15. For the network in Figure 36, set up the equations for the determination of the
charge Q3 and the currents h, I2, h- Assume all four of those quantities to be
zero at time zero. Transform the problem into algebraic form.
E-=r
FIGURE 37
§ 74] Simple networks 241
16. For the network inFigure 37 set up the equations for the determination of the
currents. Assume currents to be zero at time zero. Find ii(s) =L{Ii(t)} and
all
erties (variables separable, exactness, etc.), that may or may not be possessed
by an individual equation. It is intuitively plausible that no collection of
methods can be found that would permit the explicit solution, in the sense
242
§ 76] An existence and uniqueness theorem 243
^=f{^,yy
ax
(1)
(d) 0(a') is unique on the interval |,y — Xq\ ^ // in the sense that it is the
only function that has all of the properties (a), (b), and (c).
The interval |a- — a'oI ^ h may or may not need to be smaller than the
interval |a — Xq\ ^ b over which conditions were imposed on /and df/dy.
In rough language, the theorem states that if /(a-, y) is sufficiently well
behaved near the point (aq, Jo)' the differential equation (1) has a solution
that passes through the point (xq , yo), and that solution is unique near
(^0 yo)-
A proof of this fundamental theorem is presented in the next three sections.
In essence the proof involves showing that a certain sequence of functions
has a limit and that the limiting function is the desired solution. The sequence
considered will be defined as follows:
yo(x) = >'o
y2(x)^yo + \ At,y,(t))dt,
yn(x) = yo +j f(t,y„-^{t))dt.
So that the proof may appear more reasonable, we first consider some
examples of the proof for special differential equations.
244 The Existence and Uniqueness of Solutions [Ch. 15
dy
§77] A Lipschitz condition 245
of (4).
Exercises
In each of the following exercises, determine the limit of the sequence defined in
(2) above. Verify that the function you obtain is a solution of the initial value
problem.
1. y' = x; Xo = 2, yo = 1.
is bounded there. That is, there exists a number A^ > such that
df
^K,
dy
df
\fix,y,)-fix,y2)\ = (X, >'*) bl ->'2U
dy
and
l/(^,>',)-/(A-,y2)l^^|}'l->'2l, (1)
The inequality (1) is called a " Lipschitz condition " for the function/. We
have shown that under the hypotheses of our existence theorem, the Lip-
schitz condition (1) holds for every pair of points (x, y^) and {x, jj) in T.
continuous in T.
246 The Existence and Uniqueness of Solutions [Ch. 15
>'nW = }o+ f
fit,y„-i{t))dt (1)
\yn(x)-yo\ ^b,
— Xq\ ^ h we have
if \x
\yi(x) -
§ 78] A proof of the existence theorem 247
Lemma 3 :
7/" |.v - XqI ^ h, then
MK"-'\x-Xo\" MK"-'h"
\yn(x) -y„- l(x)| ^ ^ ;
,
for n = \, 2, 3, . . . .
\yiix)-yo\ ^M\x-xo\.
Assuming that
b'n-M - JV2WI ^ ,
(n — ,„"' , (2)
1)!
^ MK"-'\x-Xo\"
\yM - y„-i(x)\„ s
,
,
, ,
; .
We will prove this for the case a'o ^x^ Xq + h. From Lemma 2 we have
^ f\f(t,yn-i(t))-fit,y„-2(mdt
^Kf\y„_,(t)-y„_2it)\dt.
or
\yn(x) - yn-l(x)\ S
ni
— \X - Xol". (3)
248 The Existence and Uniqueness of Solutions [Ch. 15
iLyM-yn-Ml and J]
!^1!L^^
n=l /.= ! nl
S[>'„W->-„-iW] (4)
n= 1
- = - + [J2W - + ••
n=l
I lyn(x) y„-i(x-)] [>'i(x) jo(x)] yi(x)']
+ [>'fcW->'fc-iW],
we see that
t
That is, the statement that the series (4) converges absolutely and uniformly
is equivalent to the statement that the sequence y„(x) converges uniformly
on the interval
\x - Xo\ ^ h.
If we now define
4){x) = lim y„{x)
and recall from the definition of the sequence y„{x) that each >'„Cv) is con-
tinuous on \x — Xq\^ h, it follows (since the convergence is uniform) that
^(x) is also continuous and
Because of the continuity of/ and the uniform convergence of the sequence
y„(x), we may interchange the order of the two limiting processes to show
that (p(x) is a solution of the integral equation
Finally, since we have shown in Lemma 1 that \y„(x) — yo\ ^ b for each n
and for \x — Xq\ ^ /?, it follows that the same inequality must hold for
({)(x) = lim y„(x). That is, if \x — Xq\ ^ h, then |0(x) - jol S b.
n-* 00
Thus we have completed the proof of parts (a), (b), and (c) of the existence
theorem of Section 76.
We must now show that the function (j){x) obtained in Section 78 is unique.
Suppose there is another function Y{x), such that dY/dx =f[x, Y(x)],
^(•^o) = >'o ' and Y(x) I
— Jo ^ I
b for \x — Xq\ ^ h. Then we may write
We shall now show that as n -^ co the integral on the right side of (1)
approaches zero for \x - Xo\ ^ /?. It will then follow that Y(x) = \imy„{x),
For any a- on the interval \x - Xq] ^ h it is true that [x, Y(x)] and
[x, y„-i{x)] are in the rectangle R, hence the Lipschitz condition of Section 77
will allow us to change (1) into
\Y(x)-y,ix)\^K f\Y{t)-yo\dt
^ Kbix - xo).
in.)-.„_.w,.aii^
(n - 1)!
250 The Existence and Uniqueness of Solutions [Ch. 15
„w_,„w,.^:^(ifM. (3)
This will complete an inductive argument for the relation (3). We have for
Xq ^ ^' ^ Xq + h^
^K f\Y{t)-y„_,it)\dt
-'xo
K"b r^
(n - 1) J.^
\)\
K"b
< {x - Xof,
|y(x)->'„(A)l^—^. (4)
As A? ->• 00 the expression on the right side of relation (4) approaches zero.
Hence it follows that for |a- — Xq\ ^ h, y„(x) -^ Y(x). Thus Y(x) must be the
same function (f){x) we obtained in Section 78. That is, the solution (j){x) is
unique.
Theorem: If the function f of equation (1) and its partial derivatives with
then there exists an interval \x — Xq\ -^ h and a unique function (j)(x) such that
(l)(x) is a solution of {\) for all x in the interval |x — XqI ^h, (J)(xq) = Jq , and
0'(^o) = y'o •
A proof of this theorem that is quite similar to the proof given in Sections
78 and 79 can be found in Ince.* The generahzation of the theorem to equa-
tions of higher order is direct.
Nonlinear Equations
252
§ 82] Factoring the left member 253
arise with nonlinear equations and a few techniques that will find solutions
it may be possible to factor the left member. The problem of solving (1) is
then replaced by two or more problems of simpler type. The latter may be
capable of solution by the methods of Chapters 2 and 4.
Since y' will be raised to powers in the example and exercises, let us simplify
the printing and writing by a common device, using/? for y'
dy
dx
The left member of equation (2) is readily factored. Thus (2) leads to
{xp + \){yp + I) = 0,
yp+\=Q (3)
or
xp + 1 = 0. (4)
y dy -\- dx —
it follows that
/=-2(x-q). (5)
X dy -{- dx =
from which, for x 9^ 0,
^^
X
254 Nonlinear Equations [Ch. 16
SO
y= - \n\c2xl (6)
We say, and it is very rough language, that the solutions of (2) are (5) and
(6). Particular solutions may be made up from these solutions; they may be
drawn from (5) alone, from (6) alone, or conceivably pieced together by
using (5) in some intervals and (6) in others. At a point where a solution
from (5) is to be joined with a solution from (6), the slope must remain
continuous (see Ex. 21 below), so the piecing together must take place along
the line y= x. Note (see Ex. 24) that the second derivative, which does not
enter the differential equation, need not be continuous.
The existence of these three sets of particular solutions of (2), that is,
solutions from (5), from (6), or from (5) and (6), leads to an interesting
phenomenon in initial value problems. Consider the problem of finding a
solution of (2) such that the solution passes through the point ( — j, 2). If the
result is to be valid for the interval —1 < x < — ^, there are two answers,
which will be found in Ex. 25, following. If the result is to be valid for
— < X <^,
1 there is only one answer (Ex. 26), one of the two answers to
Ex. 25. If the result is to be valid in — <x<
1 2, there is only one answer
(Ex. 27).
Exercises
In Exs. 1-18, find the solutions in the sense of (5) and (6) above.
Exs. 19-27 refer to the example of this section. There the diflferential equation
and
y= -ln|c2x|. (6)
19. Show that of the family (5) above, the only curve that passes through the point
(1, 1) is >> = (3 — IxY'^ and that this solution is valid for x < f.
20. Show that of the family (6) above, the only curve that passes through the point
(1, 1) is >'= 1 — In X and that this solution is valid for < x.
21. Show that the function defined by
y = (2-xy" forx^I,
y = —In X for x ^ 1,
FIGURE 38
256 Nonlinear Equations [Ch. 16
24. Show for the solution given in Ex. 23 that /' is not continuous at a: = 1. Show
that as x-^ I ~,y" and as x-> 1 *,/'-> + 1.
-> —1,
25. Find those solutions of (2) which are valid in — 1 < jc < —i and each of which
has its graph passing through the point (— i, 2).
ANS. y = (3 -IxY'^; and j = 2 — In 2 - In (-x).
26. Find that solution of (2) which is valid for — 1 < x < ^ and has its graph passing
through the point ( — i, 2). ans. y ^(3 —IxY'^.
27. Find that solution of (2) which is valid for — 1 < x < 2 and has its graph passing
through the point ( — i, 2). ans. y = {3 —IxY'^ for x ^ I,
>> = 1 — In X for 1 ^ X.
Here
yp= ±s/a^-y\
so we may write
ydy
dx, (2)
Va^/
or
= dx, (3)
^a'-y'
or (if the division by ^Ja^ — y^ cannot be effected)
a^-y^= 0. (4)
X = c2 + VJ'a' - y\
^ (6)
y = a OX y = —a. (7)
Graphically, the solutions (5) are left-hand semicircles with radius a and
centered on the x-axis; the solutions (6) are right-hand semicircles of radius
a, centered on the x-axis. We may combine (5) and (6) into
{x-cf-^y^ = a\ (8)
§84] The c-discriminant equation 257
FIGURE 39
(8) and also shows the two lines representing y = a and y = —a. At each
point of either line, the line is tangent to an element of the family of circles.
(l>(x,y,c) = 0. (2)
—
d(b
—
dx
dx dc
+ ——+ —=
dy
dy dc
d(b dd)
dc
0. (3)
The slope of the envelope and the slope of the family element concerned
must be equal at the point of contact. That slope can be determined by
differentiating equation (2) with respect to x, keeping c constant. Thus it
follows that
— + —— =
d(b dy dd)
0. (4)
dx dy dx
Equations (3) and (4) both hold at the point of contact and from them it
follows that
dd)
^=
dc
0. (5)
0(x,>;,c) = O, (2)
* The c-discriminant equation may contain a locus of cusps of the elements of the general
solution and a locus of nodes of those elements, as well as the envelope which aroused our
interest in it.
§ 85] The p-discriminant equation 259
is one.
f(x,y,p) = (1)
P= 9ix,y). (2)
If at a point {xq , Jq) the equation (1) has, as an equation in p, all its roots
distinct, then near {xq yo) there ,
will be n distinct equations of the type of
equation (2). Near (xq , >'o) the right members of these n equations will be
single-valued and may satisfy the conditions of the existence theorem de-
scribed in Chapter 15. But if at {xq, Jq) equation (1) has at least two of its
roots equal, then at least two of the n equations like (2) will have right
members assuming the same value at (xq yo). For such equations there is ,
no region, no matter how small, surrounding (xq yo) in which the right ,
(x, y) for which (1) has at least two of its roots equal.
The condition that equation (1) have at least two equal roots as an equation
in p is that both /= and df/dp = 0. These two equations in the three
variables x, y, and p are parametric equations of a curve in the xj-plane
with p playing the role of parameter. The equation that results when p is
* For more detail on singular solutions and the discriminants see E. L. Ince, Ordinary
Differential Equations (London: Longmans, Green and Co., 1927), pp. 82-92.
260 Nonlinear Equations [Ch. 16
Exercises
1. For the quadratic equation
B^-4AC = 0.
2. For the cubic p^ + Ap + B = 0, show that the /^-discriminant equation is
4A^ + 21B^=0.
3. For the cubic p^ + Ap^ + B = 0, show that the /^-discriminant equation is
Bi4A^ + 27B)=0.
4. Set up the condition that the equation x^p^ + x^yp + 4 = have equal roots as a
quadratic in p. Compare with the singular solution xy^ = 16.
5. Show that the condition that the equation xyp^ + ix + y)p +1=0 of the
example of Section 82 have equal roots in /? is (x — yY = and that the latter
equation does not yield a solution of the differential equation. Was there a
singular solution?
6. For the equation y^p'^ — a^ + 7^ = of Section 83, find the condition for equal
roots inp and compare with the singular solution.
7. For the differential equation of Ex. 6 show that the function defined by
y-W
§ 86] Eliminating the dependent variable 261
f(x,y,p) = 0; p = ^, (1)
ax
is of a form such that we can readily solve it for the dependent variable y
and write
y = g(x,p). (2)
.(x.p,g=0 (3)
involving only x and p. If we can solve equation (3), we will have two equations
relating x, y, and p, namely, equation (2) and the solution of (3). These
together form parametric equations of the solution of (1) with p now
considered a parameter. Or, if p be eliminated between (2) and the solution
of (3), then a solution in the nonparametric form is obtained.
Rewrite (4) as
3y = xp + 9x^lp. (5)
Then differentiate both members of (5) with respect to x, using the fact that
dy/dx = p, thus getting
18x / 9x^ dp
3p =p+ + x--^ /,
p \ p^ / dx
262 Nonlinear Equations [Ch. 16
or
\ PV \
or
§ 87] Clairaut's equation 263
It is not difficult to show that equation (II) defines two solutions of the
differential equation. These solutions are not special cases of the general
solution (10). They are singular solutions. Equation (II) has for its graph
the envelope of the family of curves given by equation (10). The solutions
defined by (11) are also easily obtained from the /^-discriminant equation.
y=px+f{p), (1)
where /(/?) contains neither x nor y explicitly, can be solved at once by the
method of Section 86. Equation (1) is called Clairaut's equation.
p=p^{x^f'{p)]p',
[^+/'(P)]^ = 0. (2)
Then either
dp
or
X +r(p) = 0. (4)
y = cx +f(c), (5)
a result easily verified by direct substitution into the differential equation (I).
^v = -f'(p), (6)
y=fip)~pf'{p). (7)
264 Nonlinear Equations [Ch. 16
represents a straight line for each value of c, the solution (6) and (7) cannot
be a special case of (5); it is a singular solution.
y=px+p\ (8)
y = ex + c'^
at once.
Then using (6) and (7) we obtain the parametric equations
FIGURE 40
§ 87] Clairaut^s equation 265
Rewrite (11) as
{y-xpf-\-p^ = () (12)
is the general solution of (11). The solution (13) is composed of two families
of straight lines,
and
y = C2X-J\^c^. (15)
From the /^-discriminant equation for (11) we obtain at once the singular
solutions defined by
x^+/ = l. (16)
¥!ixercises
to get p in terms of a and then show that y —px —f(p) vanishes identically.
2. Prove that if/"(a) J^ 0, then (17) above is not a special case of the general solu-
tion y = cx+f(c).
Hint: show that the slope of the graph of one solution de-
pends upon x whereas the slope of the graph of the other does not depend
upon X.
In Exs. 3-30, find the general solution and also the singular solution, if it exists.
4. p^ + 4x^p — llx'^y = 0. ANS. 12j^= c(c + 4x^); sing, sol., 3>' —a:^.
5. 2xp^ — 6yp^ + X* ^0. ans. 2c^x^ = — 6c^y; sing, sol., 2y = x^.
1
11. 3x'*^p^ — xp — y = 0. ANS. x^' = c(3cx — 1); sing, sol., Hjc^j' = —1.
12. A-/?^ + (x - + - = 0.
;^);p 1 :^'
14. x*/7^ — 'ixp — 3^ = 0. ANS. 2)xy = c{xc^ — 3); sing, sol., 9x^y^ = 4.
15. J x^p^ — xp.
= ANS. xy = c{c^x — 1); sing, sol., llx^y^ = 4.
16. xp"" - lyp^ + \2x^ = 0. ANS. 20^7 = c*^^ + 12; sing, sol., 3y^ = ±%x^.
17. xp^ —yp^+\=Q. ANS. xc^ — >'c^ + = 0; sing, sol., Ay^ = 21x^.
1
/xV \""'
I y
ANS. >> = cj:+ c"; Sing, sol., (
-I =— I -I
30. 8>' = 3x^ + p\ ANS. (/? - 3x)^ =c(p-x) and 8j = 3x=' + p^.
f(x,y',y") = 0, (1)
which does not contain the dependent variable y explicitly. Let us put
y =p.
Then
. dp
dx
and equation (1) may be replaced by
§ 88] Dependent variable missing 267
an equation of order one in p. If we can find p from equation (2), then y can
be obtained from y' =p hy an integration.
xy"-{y'?-y' = Q (3)
Since y does not appear explicitly in the differential equation (3), put
y' = p. Then
~ dp
y
dx'
x —
^P -p'-p
dx
=n 3
0.
dp dx
p{p^ + 1)
~ T
or
dp p dp dx
'
p p^ + I X
from which
follows.
Equation (4) yields
c, V= ^^(l + p'),
P =^ x^— X
C,
2-
dy=+ ,-,
X dx
— •
(6)
268 Nonlinear Equations [Ch. 16
y-c2= +{c,^-x^Yi\
or
Equation (7) is the desired general solution of the differential equation (3).
Note that in dividing by p early in the work we might have discarded the
solutions y = k{p = 0), where k is constant. But (7) can be put in the form
with new arbitrary constants c^ and c^. Then the choice C3 = 0, C4 = — \jk
yields the solution y = k.
A second-order equation
/(>',/, y') = o (1)
in which the independent variable x does not appear explicitly can be reduced
to a first-order equation in y and y' . Put
y'=P,
then
„ dp dy dp dp
'
dx dx dy dy
so equation (1) becomes
/(..P,p|)=0. (2)
Since the independent variable does not appear explicitly in equation (3),
. dp
y =p-r
dy
§ 89] Independent variable missing 269
Pdp dy
2 < ' ~ ^»
p" + 1
.
from which
so
p'+\=c,'y-\ (5)
P=± y
Therefore
dy
_ ,
{c,'-yy
dx ~ y
or
±y{c,^ -y^y'i^dy^dx.
Then
+ (ci2_/)i/2=^_C2.
from which we obtain the final result
{x-C2f+y^ = c,^.
Exercises
1. /' = xi/y. ANS. X ^Ci sin (y + Cz).
2. x^y" + (y'y — Ixy' = 0; when x = l,y =^5 and y' = —4.
ANS. J = ix^ + 3x - 3 + 9 hi (3 - x).
3. x'^y" + {y'Y — Ixy' = 0; when x = 2, y = 5 and y' = 2.
ANS. x^ = l{y — 3).
4. jj'" + (7')^ = 0. ANS. See Ex. 20, page 14.
5. >' V
+ {y'Y = 0. ANS. X= cij' - In C2 I A'l
6. + \)y" = {y'Y.
(j' ANS. y+\=c2 /'\
7. 2a/' + {y' y = 0. ANS. (y - CiY = 4a(x - Ci).
8. Do Ex. 7 by another method.
9. x/' =/ + .Jc'; when x = I, y = ^ and /= 1. ans. 24>' = x^ + 9x^ + 2.
270 Nonlinear Equations [Ch. 16
22. Show you can perform the integrations encountered, then you can solve
that if
3j'
>» = + X^ 1
= x^ + C2 — 3 ln|x + Ci|.
-m-
39. {y'y — 2y" + {y'y — 2xy' + x^ = 0; when x = 0,y =
and / == 1 i
2^ = 1 + x^ + 2
ANS. sin x.
40. {y")^ — xy" + y' =0. ans. General solution 2y = CiX^ — 2ci ^x + ci',:
FIGURE 41
will sag and there will be a lowest point V as indicated in the figure. We wish
to determine the curve formed by the suspended cable. That curve is called
the catenary.
Choose coordinate axes as shown in Figure 42, the j-axis vertical through
the point V and the x-axis horizontal and passing at a distance >'o (to be
chosen later) below V. Let s represent length (ft.) of the cable measured
from V to the variable point P with coordinates Then the portion of
(x, y).
the cable from K to P is subject to the three forces shown in Figure 42. Those
272 Nonlinear Equations [Ch. 16
forces are: (a) the gravitational force ws (lb.) acting downward through the
center of gravity of the portion of the cable from Kto P, (b) the tension T^ (lb.)
acting tangentially at P, and (c) the tension T2 (lb.) acting horizontally
(again tangentially) at V. The tension T^ is a variable; the tension T2 is
constant.
Since equilibrium is assumed, the algebraic sum of the vertical components
of these forces is zero and the algebraic sum of the horizontal components
of these forces is also zero. Therefore, if is the angle of inclination, from
the horizontal, of the tangent to the curve at the point (x, y), we have
Ti sin e - ns = (1)
and
Ti cos e-Tj^O. (2)
tan 9 = ^. (3)
cix
We may eliminate the variable tension T^ from equations (1) and (2) and
obtain
ws
tanO = — (4)
The constant Tjl^v has the dimension of a length. Put T2iw = a (ft.). Then
equation (4) becomes
tan = - (5)
a
a ax
Now we know from calculus that since s is the length of arc of the curve,
then
d^y
From (6) we get
1
a dx
ds
= —
dx
j , so the elimination of s yields the differential
equation
^=y-er-
dx
§ 90] Miscellaneous exercises 273
The desired equation of the curve assumed by the suspended cable is that
solution of the differential equation (8) which also satisfies the initial con-
ditions
dy
when X = 0, y = yo and -— = 0. (9)
ax
Equation (8) fits into either of the types studied in this chapter. It is left
as an exercise for the student to solve the differential equation (8) with the
conditions (9) and arrive at the result
y = a cosh - + yQ — a. (10)
a
y = a cosh -
a
Miscellaneous exercises
1. x^p^ + x^yp + 4=0. ANS. cxy + 4x + c^ xy^ = 16.
=0; sing, sol,
2. 6xp^ — (3x + 2y)p + j = 0. = CiX, 2y = x + C2.
ans. y^
3. 9p^ + 3xy*p + ^'^ = 0. ans. cy^{x — c) = sing, sol., x^y^ = 4.
1 ;
ans. x^c^ — (2x>' + l)c + >'^ + = 0; sing, sol., 4x^ — 4x>' — 1=0.
1
21. x"/?^ = 8(2y + xp). ANS. c^x^ = 8(2x^>' - c); sing, sol., a:> = - 1.
22. x^p^ = (x — yy. ANS. x(x — 2y) = Ci,y= —x In C2 xl |
274 Nonlinear Equations [Ch. 16
ANS. py = cexp (/?"') andpx ^ yip^ —p+ 1); sing, sol., = >' jc.
Index
Abel's formula, 137 boundary conditions pertaining to, 217
Alpha function, 178-182 Bernoulli's equation, 69-71
Applications, 44-58, 201-221, 229-241, Bessel's equation, 130
271-273
beams, 217-220
catenary, 271-273 Cardioids:
chemical conversion, 48 differential equation of, 15
compound interest, 52 self-orthogonal family of, 56-57
cooling, Newton's law, 46—47 Catenary, 271-273
electric circuits, 231-234 c-discriminant equation, 258
electric networks, 234-241 Change of dependent variable, linear
mixture problems, 51-52 equation of order two, 127-129
orthogonal trajectories, 52-57 Change of form of solution, 18-20
pendulum, 216-217 Change of variable in equation of order
tractrix, 51-52 one, 25, 28, 68, 73-76
277
278 Index