For Chapter 5, the following questions should be attempted (please refer to the

Part IV of the Book).

Ex. 2.2 7, 15, 19, 21,

Ex. 2.3 11, 15, 17, 21, 25(a)
Ex. 2.4 9, 13, 17, 25, 29
Ex. 2.6 11, 13, 15, 29, 31, 46
Ex. 4.2 5, 7, 17, 19, 39, 41, 43
Ex. 4.3 5, 7, 11, 27
Ex. 4.4 9, 11, 13, 15, 21, 27, 29, 31, 35
Ex. 4.5 3, 5, 7, 19, 21, 27, 31, 33, 35, 39
478 490 FIRST-ORDER DIFFERENTIAL EQUATIONS

16.

EXERCISES 2.2
In Problems 1-6, determine whether the given differen­
dx = x 2( I + y) ,
25. dy y (0 ) =3
tial equation is separable.

dy dy 26. Vydx + (I + x)dy = 0 , y(O) = I

17. 1. dx = 41 - 3y + I 2. dx - sin(x + y) = 0
27. Solutions Not Expressible in Terms of Elemen­
tary Functions. As discussed in calculus, certain
3 dy = x2
indefinite integrals (antiderivatives) such as f e dx
'dx
cannot be expressed in finite tenus using elementary
5. s2 + ds = s + I functions. When such an integral is encountered while
dt st
18. solving a differential equation, it is often helpful
6. (xy2 + 3l)dy 2x dx =0 to use definite integration (integrals with variable
upper limit). For example, consider the initial value
In Problems 7-16. solve the equation. problem
dy
7. dx :::; y(2 + sinx) 8. dxdv : :; dy =
y(2) :::; I .
dx
2
dy = I - x dx 2
9. dx 10. dt :::; 3xt The differential equation separates if we divide by I
and multiply by dx. We integrate the separated equa­
dv sec2y dv _ I - 4v 2
11• ....::...:::;-­ 12 . x dx ­ tion from x :::; 2 to X Xl and find
dx I + x2
Mat X=X, eX 2 Ix=x'd2'v
Won dx =
ExpJ Ix=2 x=2 Y
Mat 15. y-l dy + yeCos.t sin x dx = 0 -1.lx=.l:1
How 16. (x + x:l)dx + ery dy = 0 (d.;v'd~ b "< , Y .11=2
Expl
= __1_+ _1_
Mat: y(XI) y(2)'
Visu In Problems 17-26, solve the initial value problem.
Verii 17. y' r(l - y) , y(O) 3 If we let t be the variable of integration and replace
by x and y(2) by 1, then we can express the solu­
= ~;;:
Xl
dy
18. dx =(
I +' y-
)tan x , y ()
0 V 3 tion to the initial value problem by
dy r-:--:
19. dx
<

= 2V Y + 1 cos x , o
dy 3x 2 + 4x + 2
20. dx = + y(O) :::; Use definite integration to find an explicit solution to
the initial value problems in parts (a)-(c).
dy . x2
21. de = y sm e , y( 1T) :::; - 3 (a) dv/dx = e , y(O) 0
(b) dy/dx = e-<2y-2 , y(O) 1
22. x 2 dx + 2y dy = 0 , y( 0) =2 (c) dy/dx Vl+sinx(1+y2), y(O) 1
dy
23. dt = 2tcos y 2
, yO
( )
1T/4 ~ (d) Use a numerical integration algorithm (such as
Simpson's rule, described in Appendix B) to
approximate the solution to part (b) at x = 0.5 to
y(l) :::; 0
three decimal places.
 INl
498 FIRST-ORDER DIFFERENTIAL EQUATIONS

.et

Je th,
~awi
y(2) e-2+1(4) + e- 2 r
Inserting the given value of y( 1) and solving, we express

eX\!t + cos2x dx .

Using Simpson's rule, we find that the definite integral is approximately 4.841, so
:ntiat
J I(l. y(2) "'" 4e- 1 + 4.84le-2 "'" 2.127. •
men
:',pn In Example 3 we had no difficulty expressing the integral for the integrating factor
f.L{x) = ef 1 dx = eX. Clearly, situations will arise where this integral, too, CarInot be expressed
with elementary functions. In such cases we must again resort to a numerical procedure such as
Euler's method (Section 1.4) or to a "nested loop" implementation of Simpson's rule. You are
Supp' invited to explore such a possibility in Problem 27.
regiOJ Because we have established explicit formulas for the solutions to linear first-order differ­
nonn ential equations, we get as a dividend a direct proof of the following theorem.
throu

Theorem 1. Suppose p{x) and Q(x) are continuous on an interval (a, b) that contains
the point Xo. Then for any choice of initial value Yo, there exists a unique solution y{x)
on (a. b) to the initial value problem

hem dy
(15) dx + p{x)y Q(x),
"in (
Lore it
In fact, the solution is given by (8) for a suitable value of C.
hem
,Can
Lore iJ The essentials of the proof of Theorem I are contained in the deliberations leading to equa­
:hem tion (8); Problem 34 provides the details. This theorem differs from Theorem 1 on page 12 in
laiiVI that for the linear initial value problem (15), we have the existence and uniqueness of the
fy the solution on the whole interval (a, b), rather than on some smaller unspecified interval about Xo.
The theory of linear differential equations is an important branch of mathematics not only
because these equations occur in applications but also because of the elegant structure associ­
ated with them. For example, first-order linear equations always have a general solution given
by equation (8). Some further properties of first-order linear equations are described in Prob­
lems 28 and 36. Higher-order linear equations are treated in Chapters 4, 6, and 8.

EXERCISES 2.3
In Problems 1-6, determine whether the given equation y dv
3. (r " + I )ddt =e l

is separable, linear. neither, or both.

= yt y 4. 3t dt + yIn t
dy dx dr
+ t 2x
3
1. .x2 dx + sin x y= 0 5. x dt sin t 6. 3r = de - e
r
"
iI
PART IV: FUNDAMENTALS OF DIFFERENTIAL EQUATIONS AND BOUNDARY VALUE PROBLEMS 499

t In Problems 7-16, obtain the general solution to the y the term (185!4)e- Zt eventually dominates (has

I equation.
dy
7. dx Y - e3J: = 0

dr
9. dO + r tan 0
dy Y
8. dx ~+2x+l

sec 0
dv
10. xix + 2y x- 3
greater magnitude for t large).
(a) Redo Example 2 taking k = 20/sec. Now which
term in the solution eventually dominates?
(b) Redo Example 2 taking k = 1O/sec.

I
25. (a) Using definite integration, show that the solution
dy to the initial value problem
11. (t+y+ l)dt-dy=O 12. dx x 2e -4x - 4y
: + 2xy = 1, y(2) I,
0- dx 3
,(13.' y-d
Y + 2x = 5y can be expressed as
dy
14. xix + 3y + 2x- = x + 4x
'3
y(x) = e-
XZ
(e 4 + J: etC dt)
15. (~ + I): + xy - x =0 ~. (b) Use numerical integration (such as Simpson's
rule, Appendix B) to approximate the solution at
dy x = 3.
16. (Xl + 1) dx = x 2 + 2x - I - 4xy
26. Use numerical integration (such as Simpson's rule,
Appendix B) to approximate the solution, at x = 1,
In Problems 17-22, solve the initial value problem. to the initial value problem
dy y < dy sin 2x 2 Y y(O) 0 .
17. dx - ~ = xe" y(l) =e- I -+ I,
dx 2(1 + sin x)
dy -x 4 Ensure your approximation is accurate to three deci­
18. dx + 4y - e =0 , y(O) = ­3 mal places.
27. Consider the initial value problem
19. t 3 ;:; + 3t l x = t , x(2) = 0

dy
: + V I + sin2x Y = X , y( 0) = 2 .
20. -d
x
+ x
+ 2 = 3x , y( 1) =I
(a) Using definite integration, show that the inte­
dy. , grating factor for the differential equation can be
21. cos x dx + y sm x = 2x cos-x , written as

y(;) 32
p.(x) = exp(tt VI + Sin2tdt)
' dy+ . and that the solution to the initial value problem is
22• sm x dx y cos x = x sm x ,
y(x)
I
p.(x)
IX0 p.(s) s ds + p.(x)
2
.
23. Radioactive Decay. In Example 2 assume that the
rate at which RA I decays into RA2 is 40e -20t kg/sec .$" (b) Obtain an approximation to the solution atx = 1
and the decay constant for RA z is k = 5/sec. Find the by using numerical integration (such as Simp­
mass y(t) of RA z for t ~ 0 if initially y(O) = 10 kg. son's rule, Appendix B) in a nested loop to esti­
24. In Example 2 the decay constant for isotope RA I was mate values of p.(x) and, thereby, the value of
IO/sec, which expresses itself in the exponent of the
rate term 50e -lOr kg/sec. When the decay constant
LP.(S)Sds.
for RA2 is k = 2/sec, we see that in formula (14) for

l
 PART IV: FUNDAMENTALS OF DIFFERENTIAL EQUATIONS AND BOUNDARY VALUE PROBLEMS 509

Remark. Since we can use either procedure for finding F(x, y), it may be worthwhile to con­
sider each of the integrals !M(x, y)dx and !N(x, y)dy. If one is easier to evaluate than the
other, this would be sufficient reason for us to use one method over the other. [The skeptical
reader should try solving equation (15) by first integrating M(x, y).]

Example 4 Show that

(16. (x + 3x3siny)dx + (x 4cosy)dy =0

is not exact but that multiplying this equation by the factor x-I yields an exact equation. Use
this fact to solve (16).
Solution In equation (16). M = x + 3x 3sin y and N = x 4cos y. Because

aM = 3x3cos y ~ 4x 3cos Y = aN ,
ay ax
equation (16) is not exact. When we multiply (16) by the factor X-I, we obtain

417) (1 + 3x2siny)dx + (x 3cosy)dy = 0 .

For this new equation, M = I + 3x2sin y and N = x 3cos y. If we test for exactness, we now
find that
t aM
~ =
,
3x~cosy =
aN
,
ay ax
f and hence (17) is exact. Upon solving (17), we find that the solution is given implicitly by

l x + x 3sin y = C. Since equations (16) and (17) differ only by a factor of x, then any solution
to one will be a solution for the other whenever x =1= O. Hence the solution to equation (16) is
given implicitly by x + x 3sin y = C. •

I In Section 2.5 we discuss methods for finding factors that, like

inexact equations into exact equations.
in Example 4, change

EXERCISES 2.4
In Problems 1-8, classify the equation as separable, lin­ 6. xy dx + dy = 0
ear, exact, or none of these. Notice that some equations 7. e dr + (3r e - l)de 0
may have mon? than one classification.

II
8. [2x + ycos(xy)]dx + [xcos(xy) - 2y]dy 0
1. {x 2y + x 4cos x)dx x 3 dy = 0
2. (x lO/ 3 - 2y)dx + x dy = 0 In Problems 9-20, determine whether the equation is
3. {ye;ty + 2x)dx + (xe'ty 2y)dy = 0 exact. If it is, then solve it.
4. V-2y - idx + (3 + 2x - x 2 )dy 0 9. (2xy + 3)dx + (x 2 l)dy = 0

I
~
o y2 dx + (2xy + cosy)dy =
Q.(l'cf ~ Uf'..u, '/ --"i-h.:1
0 10. (2x + y)dx + (x 2y)dy = 0
 510 FIRST-ORDER DIFFERENTIAL EQUATIONS

INTEGRP

11. (cos x cos Y + 2x)dx - (sin x sin y + 2y)dy = 0 (a) Show that this equation is not exact.
~t (b) Show that mUltiplying both sides of the equation
2
12. (eXsiny - 3x )dx + ktcosy + y-2!3/3)dy = 0
by y - 2 yields a new equation that is exact.
13. (t/y)dy + (1 + lny)dt = 0 (c) Use the solution of the resulting exact equation
14. et(y - t)dt + (1 + et)dy = 0 to solve the original equation.
e the gra: (d) Were any solutions lost in the process?
IS. cos fJ dr - (r sin fJ - e")dfJ = 0
,aw in Set
ntiable ve 16. (ye;CY - l/y)dx + (xeX)' + x/y2)dy = 0 30. Consider the equation
f f(X, y, • 17. (l/y)dx - (3y - x/y2)dy == 0 (5x 2y + 6x 3yZ + 4xy2)dx
unctions ( 18. [2x + y2 + y)] dx
- cos(x + (2x3 + 3x4y + 3x 2y)dy = 0 ;
-:, prove tl
+ [ay - cos(x + y) - eY]dy = 0 (a) Show that the equation is not exact.
(b) Multiply the equation by xnym and determine
19. (2x + I +Yx 22)
Y
dx + (
1 + x-y
2- 2)Jdy = 0
X 1 values for n and m that make the resulting equa­
tion exact.
3uppose
:egion D
(lormal 11
througho!
20. [h +I - x2
YCOS(XY)]dx
(c) Use the solution of the resulting exact equation
to solve the original equation.
31. Argue that in the proof of Theorem 2 the function g
+ [xcos(xy) - y-l/3]dy =0 can be taken as

In Problems 21-26, solve the initial value problem.

21. (l/x + 2ylX}dx + (2yx 2 - cos y)dy = 0 ,
g(y) = r Yo
N(x, t)dt - rr [:t
yo·lQ
M(s, t)dS]dt ,

which can be expressed as

y(l) = 7T
:hemati 22. (yeX)' - l/y)dx + (xeX)' + X/yZ)dy = 0 , g(y) = IY N(x, t)dt - Iot M(s, y)ds
k in Con y( 1) == 1 Yo Xo

lore inte! ~i (ety + tety)dt + (te t + 2)dy = 0 , yeO) = -1 + IX M(s, yo)ds .

:hemat:
I' Can Yo
24~ (e1x + l)dt + (e l - l)dx =0 , x(l) = 1 Xo

lore inte 25. (y2sinx)dx + (l/x - y/x)dy = 0, y(7T) = 1 This leads ultinnately to the representation
themat 26. (tany - 2)dx + (xsec2y + l/y)dy = 0 ,
!lalizing yeO) == 1 (18) F(x, y) = 1.:: N(x, t)dt + [M(S, yo)ds .
ify the D
Evaluate this formula directly with Xo = 0, Yo = 0
27. For each of the following equations, find the most
to rework
general function M(x, y) so that the equation is exact.
(a) Example 1.
(a) M(x,y)dx + (sec~ - x/y)dy == 0
(b) Example 2.
(b) M{x,y)dx + (sinxcosy - xy - e-Y)dy == 0
(c) Example 3.
28. For each of the following equations, find the most
general function N(x, y) so that the equation is exact. 32. Orthogonal Trajectories. A geometric problem
occurring often in engineering is that of finding a
(a) [ycos(xy) + eX]dx + N(x,y)dy == 0
family of curves (orthogonal trajectories) that inter­
(b) (yeCY - 4x 3y + 2)dx + N(x,y)dy == 0
sects a given family of curves orthogonally at each
29. Consider the equation point. For example, we may be given the lines of
(y2 + ay)dx - x 2dy = 0 . force of an electric field and want to find the equation
 PART IV: FUNDAMENTALS OF DIFFERENTIAL EQUATIONS AND BOUNDARY VALUE PROBLEMS 523

When we substitute back in for z, U, and v, we find

(v/U)2 + 2(v/u) 3 = Cu- 2 ,
2
v + 2uu 3u 2 = C ,
(y + 3)2 + 2(x - 1)(y + 3) - 3(x 1)2 =C .
This last equation gives an implicit solution to (16). ..

EXERCISES 2.6
In Problems 1-8, identify (do not solve) the equation as Use the method discussed under "Bernoulli Equations"
homogeneous. Bernoulli, linear coefficients, or of the to solve Problems 21-28.
fonny' = G(ax + by).
1. (y 4x 1)2 dx - dy = 0 21. : +~ 2
x y2
2. 2tx dx + (t 2 x 2 )dt = 0 dy
3. (t + x + 2)dx + (3t - x - 6)dt = 0 22. dx
4. dy/dx + y/x = x 3y2 dy
23· dx
5. (ye- 2x + l)dx - e- 2x dy = 0
6. e dy y de VIiY de 24. dy +
dx x
7. (y3 - el)de + 2
2e y dy = 0
dv ?
8. cos(x + y)dy = sin(x + y)dx 25. dx + tx 3 + ! = 0 26. dx + y = eXy-­
dt t
Use the method discussed under "Homogeneous Equa­ 27. dr = r2 + 2re dy 3
tions" to solve Problems 9-16. 28. +yx+y 0
de dx
9. (3x -y'l)dx + (.xy
2 3
x y-l)dy = 0
10. (.xy + y2)dx - x 2 dy = 0 Use the method discussed under "Equations with Linear
11. (l- xy)dx + x dy = 0 2 Coefficients" to solve Problems 29-32.
2
12. (x + l)dx + 2.xy dy = 0 29. (-3x + y - I)dx + (x + Y + 3)dy = 0
30. (x + y l)dx + (y - x - 5)dy = 0
13. dx = =-=--~~=---.:....:.::....
dt tx 31. (2x - y)dx + (4x + y - 3)dy = 0
dy e sec(y/e) + y 32. (2x + y + 4)dx + (x - 2y - 2)dy = 0
14. de e
In Problems 33-40, solve the equation given in:
dy x 2 - y2 dy y(Jny lnx+l)
15 - = - ­ 16. dx 33. Problem I. 34. Problem 2 .
. dx 3xy x
35. Problem 3. 36. Problem 4.
37. Problem 5. 38. Problem 6.
Use the method discussed under "Equations of the Fonn
39. Problem 7. 40. Problem 8.
dy/dx = G(ax + by) .. to solve Problems 17-20.
17.dy/dx= I 18.dy/dx=(x+y+2)2 41. Use the substitution v = x - y + 2 to solve equa­
19. dy/dx = (x y + 5)2 20. dy/dx = sin(x - y) tion (8).

l ... _...__......_ - - ­
526 FIRST-ORDER DIFFERENTIAL EQUATIONS

In Problems 1-30, solve the equation.

24. (vYTx + cosx)dx + (Vx7Y + siny)dy = 0
dy eX +Y dy ~
1. dx Y_ 1 2. dx - 4y = 32x­ 25. y(x y 2)dx + x(y - x + 4)dy =0
dv
3. (x 2 2y-3)dy + (2xy - 3x 2)dx =0 26. dx + xy 0
dy 2 27. (3x y - 5)dx + (x - y + l)dy = 0
4. dx + X x 4x +3
dy
5. [sin(xy) + xy cos(xy)]dx + [1 + x 2cos(xy)]dy=0 28· dx x+y+
6. 2xy3 dx (1 x 2)dy 0 29. (4xy3 9l
+ 4xy2)dx
7. t 3l dt + t 4y-6 dy = 0 s,l'P + (3x 2 l
6xy + 2x 2y)dy = 0

8.
dx
+ x = 30. : = (x + Y + 1)2 (x + Y - 1)2
9. (x 2 + i)dx + 3xydy = 0
10. [1 + (1 + x 2 + 2xy + It 1 ] dx In Problems 31-40, solve the initial value problem.
+ [y-l/2 + (1 + x 2 + 2xy + y2)-1]dy = 0
31. (x 3 y)dx+xdy 0, y(l)=3
11. dx
dt = 1 + cos-' (
t )
= (~+~).
x
32. : y(l) = -4
12. (y3 + 4e Xy)dx + (2e t + 3i)dy = 0
33.(t+x+3)dt+dx 0, x(O) = 1
dy
13. dx -
y
x = x-sm
" .
2x
dy 2y 2
34. dx - -X- = x cos x , y(1T) = 2
dx x 2 2
14. dt - t"=l = t + 35. (21 + 4~)dx - xy dy 0, y(I)=-2
dy /36) [2cos(2x + y) - 2
x ]d;t
15. dx = 2 ­
.~ + [cos(2x + y) + eY]dy 0, y(l) = 0
dy . 37. (2x - y)dx + (x + y - 3)dy 0, y(O) = 2
16. dx + ytanx + smx o
38. vYdx + (xl + 4)dy 0, y(O) = 4
dy 2
17. dO + 2y = Y

18. : = (2x + Y 1)2 dy

40. dx - 4y = 2xy
2
y(O) -4
19. (x 2 - 3y2)dx + 2xy dy = 0 41. Express the solution to the following initial value
dy Y _1 problem using a definite integral:
20. dO + (j = -40y ­
dy I
21. (y - 2x - l)dx + (x + Y 4)dy 0 dt = 1 + t 2 - Y , y(2) 3 .
22. (2x - 2y - 8)dx + (x 3y 6)dy =0 Then use your expression and numerical integration
23. (y - x)dx + (x + y)dy = 0 to estimate y(3) to four decimal places.
~IIUt~ It.. ., ....'-'t_." •.---­

t
550 LINEAR SECOND-ORDER EQUATIONS

Solution If we try to find solutions of the fonn y = en, then, as with second-order equations, we are led
to finding roots of the auxiliary equation
(15) ? + 3,.1 - r - 3 = 0 .
We observe that r = 1 is a root of the above equation, and dividing the polynomial on the
left-hand side of (15) by r - I leads to the factorization
(r - 1)(,.1 + 4r + 3) = (r 1)(r + 1)(r + 3) = 0 .
Hence, the roots of the auxiliary equation are 1, - 1, and - 3, and so three solutions of (14) are
e, e-I 3r
, and e- • The linear independence of these three exponential functions is proved in

Problem 40. A general solution to (14) is then

(16) y(t) = cle + C2e-r + C3e-3r . •

So far we have only seen exponential solutions to the linear second-order constant coeffi­
cient equation. You may wonder where the vibratory solutions that govern mass-spring oscilla­
tors are. In the next section, it will be seen that they arise when the solutions to the auxiliary
equation are complex.

EXERcrSES 4.2
In Problems 1-12, find a general solution to the given 21. First-Order Constant-Coefficient Equations.
differential equation. (a) Substituting y = en, find the auxiliary equation
1. 2y" + 7y' - 4y = 0 2. y" + 6y' + 9y = 0 for the first-order linear equation
3. y" + 5y' + 6y = 0 4. y" - y' - 2y = 0 ay' + by = 0 ,
5. y" + 8y' + 16y = 0 6. y" - 5y' + 6y =0 where a and b are constants with a O.'*
(b) Use the result of part (a) to find the general
7. z" + z' - Z =0 8. 6y" + y' - 2y =0 solution.
9. y" - y' - lly == 0 10. 4y" 4y' + Y =0 In Problems 22-25, use the method described in Problem
11. 4w" + 2Ow' + 25w == 0 21 to find a general solution to the given equation.
12. 3y" + lly' - 7y 0 22. 3y' - 7y = 0 23. 5y' + 4y = 0
24. 3z' + lIz = 0 25. 6w' - 13w = 0
In ProNems 13-20, solve the given initial value problem.
13. y" + 2y' - 8y = 0; yeO) 3, y' (0) = -12 26. Boundary Value Problems. When the values of a
soh.1tion to a differential equation are specified at
14. y" + y' = 0; y(O) = 2 , y'(O) = 1
two different points, these conditions are called
15.y"-4y' 5y=0; y(-I)=3 ,y'(-1)=9 boundary conditions. (In contrast, initial conditions
16. y" - 4y' + 3y = 0; y(O) 1, y'(O) = 1/3 specify the values of a function and its derivative at
17. z" - 2z' - 2z = 0; z(O) = 0 . z,(O) = 3 the same point.) The purpose of this exercise is to
show that for boundary value problems there is no
18. y" - 6y' + 9y = 0; y(O) 2. y'(O) = 25/3
existence-uruqueness theorem that is analogous to
19. y" + 2y' + Y = 0; y(O) = 1 , y'(O) = -3 Theorem 1. Given that every solution to
20. y" 4y' + 4y = 0; y(I) = I , y'(l) = 1 t17) y"+ y =O

1
 PART IV: FUNDAMENTALS OF DIFFERENTIAL EQUATIONS AND BOUNDARY VALUE PROBLEMS 551

is of the fonn (c) Show that if YI(t) and Y2(t) are any two differen­
y(t) = CI cos t + C2 sin t , tiable functions that are linearly dependent on I,
where CI and c2 are arbitrary constants, show that then their Wronskian is identically zero on 1.
(a) There is a unique solution to (17) that satisfies the 35. Linear Dependence of Three Functions. Three
boundary conditions y( 0) = 2 and y( 7r/2) = O. functions YI (t), yz(t), and Y3(t) are said to be linearly
(b) There is no solution to (17) that satisfies y(O) = 2 dependent on an interval 1 if, on I, at least one of
andy( 7r) = O. these functions is a linear combination of the
(c) There are infinitely many solutions to (17) that remaining two [e.g., if YI (t) = ClY2(t) + CZY3(t)].
satisfyy(O) = 2andY(7r) = -2. Equivalently (compare Problem 33), YI> yz, and Y3
are linearly dependent on I if there exist constants
C .. Cz• and C3, not all zero, such that
In Problems 27-32, use Definition 1 to determine
CIYI (t) + CZY2(t) + C3Y3(t) = 0 for all t in I.
whether the functions YI and Y2 are linearly dependent
Otherwise, we say that these functions are linearly
on the interval (0, I).
independent on I.
27. YI (t) cos t sin t , Y2(t) = sin 2t
For each of the following, determine whether the
28. YI(t) = e3t , Y2(t) = e­ 4t
given three functions are linearly dependent or lin­
2t
29. YI (t) = te , Y2(t) e 2t
early independent on (-00,00);
30. YI (t) = rcos (Int) , Y2(t) rsin (lnt) (a) YI(t) = I. Y2(t) = t, Y3(t) r2.
31. YI(t) = tan 2t - sec2t , yz(t) = 3 (b) YI (t) -3, Y2(t) = 5 sin 2 t , Y3(t) cos2 t .
t t
32. Yl (t) == 0 , Y2(t) = et (c) YI(t) = e , Y2(t) = te , Y3(t) = rei.
33. Explain why two functions are linearly dependent on (d) YI(t) et , yz(t) = e- I , Y3(t) cosh t .
an interval I if and only if there exist constants CI 36. Using the definition in Problem 35, prove that if
and c2, not both zero, such that rI, rz, and r3 are distinct real numbers, then the func­
CIYI(t) + C2Y2(t) = 0 for all t in I . tions e'tt, e"t, and e'3t are linearly independent on
34. Wronskian. For any two differentiable functions (-00,00) . [Hint: Assume to the contrary that, say,
YI and Y2, the function e't l = cle"t + czer,t for all t. Divide by e'21 to get
e(rt- r2)1 = Cl + cze(r'-'iJt and then differentiate to
tIS) W[YI,Y2]U) y 1U)y2(t) - Yi(t)y/t)
deduce that e(rt- r2)t and e(r;-r2)1 are linearly depen­
is called the Wronskian l'
of YI and Y2- This function
dent. which is a contradiction. (Why?)]
plays a crucial role in proof of Theorem 2.
(a) Show that W[y., Y2] can be conveniently In Problems 37-42, find three linearly independent solu­
expressed as the 2 X 2 determinant tions (see Problem 35) of the given third-order differen­
tial equation and write a general solution as an arbitrary
W[v Y ] (t)
• " 2
= IYI(t) Y2(t)
il(t) Y2(t)
I. linear combination of these.
37. ym + y" + 4y = 0
6y'
(b) Let YI(i), Y2(t) be a pair of solutions to the ho­ 38. ym 6y" - Y' + 6y = 0
mogeneous equation ay" + by' + cy = 0 (with
39. zm + 2z" 4z' - 8z = 0
a :;; 0) on an open interval I. Prove that Y I(t) and
40. ym - 7y" + 7y' + 15y = 0
Y2(t) are linearly independent on I if and only if
their Wronskian is never zero on 1. [Him: This is 41. ym + 3y" - 4y' - 12y = 0
just a refonnulation of Lemma 1.] 42. Will + w" - 4w' + 2w = 0

HIHUIl( "I FeuZlIOI<' The Wronskian was named after the Polish mathematician H. Wronski (1778--1863).
 f l

552 LINEAR SECOND-ORDER EQUATIONS

43. Solve the initial value problem: hyperbolic sine, sinh t, is defined as the solution of
ym _ y' = 0; y(O) 2, this equation subject to the initial values: yeO) = 0
y'(O) = 3, y"(O) -1 and y'(O) 1.
(a) Solve these initial value problems to derive
44. Solve the initial value problem:
explicit formulas for cosh t, and sinh t. Also
ylll 2y" - y' + 2y = 0 ;
y( 0) = 2 , y' (0) = 3 , y" (0) = 5 . show that dt cosh t sinh t and ; sinh t =
~ 45. By using Newton's method or some other numerical cosh t.
procedure to approximate the roots of the auxiliary (b) Prove that a general solution of the equation
equation, find general solutions to the following y" y = Oisgivenby y = Cl cosh t + c2 sinh t.
equations: (c) Suppose a.. b, and c are given constants for which
(a) 3y'" + 18y" + 13y' 19y = 0 . ar2 + br + c = 0 has two distinct real roots. If the
(b) yiv - 5y" + 5y 0 two roots are expressed in the form a - {3 and
(c) yV 3yiv - 5y'" + 15y" + 4y' - 12y = 0 . a + {3, show that a general solution of the equation
46. One way to define hyperbolic functions is by means ay" + by' + cy 0 is y = Cleat cosh(J3t) +
of differential equations. Consider the equation C2eat sinh(J3t).
y" y = O. The hyperbolic cosine, cosh t, is (d) Use the result of part (c) to solve the initial value
defined as the solution of this equation subject to problem: y" + y' - 6y 0, yeO) 2, y' (0)
the initial values: yeO) = I and y'(O) = O. The 1712.

The simple harmonic equation y" + Y = 0, so called because of its relation to the fundamental
vibration of a musical tone, has as solutions Yt(t) cos t and yz(t) = sin t. Notice, however,
that the auxiliary equation associated with the harmonic equation is r2 + 1 = 0, which has
imaginary roots r = ::!:i, where i denotes v=I.t
In the previous section, we expressed the
solutions to a linear second-order equation with constant coefficients in terms of exponential
functions. It would appear, then, that one might be able to attribute a meaning to the forms eit
and e -it and that these "functions" should be related to cos t and sin t. This matchup is accom­
plished by Euler's formula, which is discussed in this section.
When bZ - 4ac < 0, the roots of the auxiliary equation
(1, ar2 + br + c = 0
associated with the homogeneous equation
(2) ay" + by' + cy = 0

tElectrical engineers frequently use the symbol j to denote

 560 LINEAR SECOND-ORDER EQUATIONS
l

We also remark that a complex differential equation can be regarded as a system of two
real differential equations since we can always work separately with its real and imaginary
parts. Systems are discussed in Chapters 5 and 9.

EXERCISES 4.3
In Problems 1-8, the auxiliary equationfor the given dif­
29. Find a general solution to the following higher-order
ferential equation has complex roots. Find a general equations.
solution. (a) y'" y" + y' + 3y = 0
1. y" + 9y = 0 2. y" + y = 0 (b) y'" + 2y" + 5y' 26y = 0

3. z" 6z' + lOz = 0 4. y" - lOy' + 26y 0 (c) yiV + l3y" + 36y = 0
S. w" + 4w' + 6w = 0 6. y" - 4y' + 7y = 0 30. Using the representation for e(a+i/3)t in (6), verify the
differentiation formula (7).
7. 4y" 4y' + 26y = 0 8. 4y" + 4y' + 6y = 0
31. Using the mass-spring analogy, predict the behavior
In Problems 9-20, find a general solution. as t -+ + 00 of the solution to the given initial value
problem. Then confirm your prediction by actually
9. y" - 8y' + 7y = 0 10. y" + 4y' + 8y 0
solving the problem.
11. z" + IOz' + 25z = 0 12. u" + 7u = 0
(a) y" + 16y = 0; y(O) 2, y'(O) 0
13. y" + 2y' + 5y 0 14. y" - 2y' + 26y = 0 (b) y" + lOOy' + y = 0; y(O) = I. y'(O) = 0
15. y" + lOy' + 41y = 0 16. y" - 3y' lly = 0 (c) y" - 6y' + 8y = 0; y(O) = 1. y'(0) = 0
17. y" - y' + 7y = 0 18. 2y "+ 13y'-7y = 0 (d) y" + 2y' - 3y = 0; y(O) = -2, y'(O) = 0
19. y'" + y" + 3y' - 5y = 0 20. y'" y" + 2y = 0 (e) y" y'-6y=0; y(O) = 1, y'(O)== I
32. Vibrating Spring without Damping. A vibrating
In Problems 21-27, solve the given initial value problem. spring without damping can be modeled by the initial
21. y" + 2y' + 2y = 0 ; y(O) = 2, y'(O) = I value problem (11) in Example 3 by taking b = O.
22. y" + 2y' + 17y = 0; y(O) = 1, y'(O) =-1 (a) If m = 10 kg, k = 250 kg/sec2 , y(O) = 0.3 m,
23. w" - 4w' + 2w = 0; w(O) =0, w'(O) = I and y' (0) = -0.1 mJsec, find the equation of
24. y" + 9y = 0 ; y(O) = 1, y'(O) 1 motion for this undamped vibrating spring.
25. y" - 2y' + 2y = 0; y(7T) err, y'(7T)=O (b) When the equation of motion is of the form
displayed in (9), the motion is said to be oscil­
26. y" 2y' + Y = 0 ; y(O) = 1, y'(O) = -2
latory with frequency fJ/27T. Find the fre­
27. y"'-4y"+7y'-6y=0; y(O) = I, y'(O) =0, quency of oscillation for ~he spring system of
y"(O) = 0 part (a).
28. To see the effect of changing the parameter b in the
33. Vibrating Spring with Damping. Using the model
initial value problem
for a vibrating spring with damping discussed in
y" + by' + 4y 0; y(O) = I , y'(O) = 0 , Example 3:
solve the problem for b = 5, 4, and 2 and sketch the (a) Find the equation of motion for the vibrating spring
solutions. with damping if m = 10 kg. b = 60 kg/sec,

-
 PART IV: FUNDAMENTALS OF DIFFERENTIAL EQUATIONS AND BOUNDARY VALUE PROBLEMS 569

EXERCISES 4.4
In Problems 1-8, decide whether or not the method of 20. y" + 4y = 16t sin 2t
undetennined coefficients can be applied to find a panic­
21. x"{t) - 4x'{t) + 4x{t) = te 2t
ular solution of the given equation. 22. x"{t) - 2x' (t) + x{t) = 24t 2e t
1. y" + 2y' - y t-Je t
23. y"{O) - 7y'(0) = 0 2
2. 5y" 3y' + 2y = t 3 cos 4t 24. y"{X) + y{x) 4x cos x
3. x" + 5x' 3x = 3t 25. y" + 2y' + 4y = llle 2t cos 3t
4. 2y"(x) - 6y'{x) + y{x) = (sinx)/e 4x 26. y" + 2y' + 2y 4te - I cos t
5. y"{O) + 3y'(0) - y{O) = sec 0
6. 2w/{x) 3w(x) = 4x sin 2x + 4xcos 2x In Problems 27-32. detennine the fonn of a particular
7. ty" - y' + 2y = sin 3t solution for the differential equation. (Do not evaluate
8. 8z'(x) - 2z{x) = 3x lOOe 4x cos 25x coefficients.)
27. y" + 9y = 4t 3 sin 3t
In Problems 9-26, find a panicular solution to the differ­
28. y" + 3y' - 7y t 4 e t
ential equation.
29. y" - 6y' + 9y = 5t6e 3t
9. y" + 3y = -9
30. y" 2y' + Y = te t cos t
10. y" + 2y' - y = 10
31. y" + 2y' + 2y = 8t 3e- t sin t
11. 2z" + z = ge 2t 32. y" - y' - 12y = 2t6e­ 31
12. 2x' + x = 3P
13. y" y' + 9y = 3 sin 3t In Problems 33-36, use the method of undetennined
14. y"{x) + y{x) = 2 x coefficients to find a panicular solution to the given
d 2y dy x higher-order equation.
15. dx 2 5 dx + 6y = xe
33. yltl - y" + Y sin t
16. O"{t) - O{t) = t sin t 34. 2y'" + 3y" + y' - 4y = e- t
17. y" - 2y' + Y = 8e t 35. ylll + y" - 2y = te t
18. y" + 4y = 8 sin 2t 36. /4) - 3y" - 8y = sin t
19. 4y" + lly' - 3y = -2te- 3t

The next theorem describes the superposition principle, a very simple observation which
nonetheless endows the solution set for our equations with a powerful structure. It extends the
applicability of the method of undetermined coefficients and enables us to solve initial value
problems for nonhomogeneous differential equations.

-------------------------- - - .. ...~---~
 PART IV: FUNDAMENTALS OF DIFFERENTIAL EQUATIONS AND BOUNDARY VALUE PROBLEMS 575

(If -1 were a triple root, we would need t 3 (A\t + Ao)e- t for 7fe- t .) Of course, we have to
rename the coefficients, so the general form is
Yp(t) == Ae-tcost + Be-1sint + tC + t 2 (Dt + E)e- t . •

EXERCISES 4.5
1. Given that YI(t) == cos t is a solution to 13. 2y" + 3y' 4y = 2t + sin 2t + 3
y" - Y' + Y == sin t 14. y" 2y' + 3y == cosht
andyAt) == e 2t/3 is a solution to 15. y" + ety' + y == 7 + 3t
y" y' + Y = e 2f , 16. 2y" - y' + 6y == t 2e -( sin t - 8f cos 3t + lOt
use the superposition principle to find solutions to
In Problems 17-22. find a general solution to the differ­
the following differential equations:
ential equation.
(a) y" - y' + Y = 5 sin t .
17. y" - y == -lIt + 1
(b) y" - y' + Y == sin t - 3e 2t •
18. y" - 2y' 3y == 3t2 - 5
(c) Y" - y' + Y = 4 sin t + 18e 2t •
19. y"(x) 3y'(x) + 2Y(x) == ~ sin x
2. Given that YI(t) (l/4) sin 2t is a solution to 20. y"((J) + 4y((J) sin e - cos 8
Y" + 2y' + 4y == cos 2t and that Y2(t) = t/4 1/8
is a solution to y" + 2y' + 4y = t , use the super­ 21. y"(e) + 2y'((J) + 2y(e) == e- 6 cos e
position principle to find solutions to the following: 22. y"(x) + 6y'(x) + 10y(x)
(a) y" + 2y' + 4y == t + cos 2t . = lOx 4 + 24x 3 + 2x 2 - 12x + 18
(b) y" + 2y' + 4y 2t - 3 cos 2t . In Problems 23-30, find the solution to the initial value
(c) y" + 2y' + 4y == Ilt - 12 cos 2t . problem.
In Problems 3-8, a nonhomogeneous equation and a 23. y' - y == 1 , y( 0) 0
panicular solution are given. Find a general solution for 24. y" == 6t; y(O) 3, y'(O) = -1
the equation. 25. z"(x) + z(x) == 2e- x ; z(O) == 0, z'(O) 0
3. y" - y == t, yp(t) == - t 26. y" + 9y = 27; y(O) 4, y'(O) 6
4. y" + y' = 1 , Yp(t) t 27. Y"(x) - y'(x) - 2Y(x) = cos x sin 2x ;
5. (J" - (J' - 28 == 1 - 2t, 8p (t) = t y(O) == -7/20, y'(O) == 1/5
6. y" + 5y' + 6y == 6x2 + lOx + 2 + l2e X , 28. y" + y' - 12y = e t + e 2t
Yp(x) = eX + x 2 y(O) 1, y'(0) = 3
7. y" == 2y' - Y + 2e x , yp(x) = x 2ex 29. y"(e) - y(e) == sin e - e 26 ;
8. y" == 2y + 2 tan 3x, Yp(x) = tan x y(O) == 1 , y'(O) == -1
30. y" + 2y' + Y = t 2 + 1 - e t ;
In Problems 9-16 decide whether the method of undeter­
y(O) = 0, y'(O) = 2
mined coefficients together with superposition can be
applied to find a particular solution of the given equa­
In Problems 31-36, determine the form of a panicular
tion. Do not solve the equation. solution for the differential equation. Do not solve.
9. 3y" + 2y' + 8y == t 2 + 4t - t 2e t sin t 31. y" + Y == sin t + t cos t + lOt
10. y" - y' + Y == (e t + t)2 32. y" - y == e 2t + te 2t + t 2e 2t
11. y" - 6y' - 4y 4 sin 3t - e 3tt 2 + 1ft 33. x" x' - 21: = e t cos t t 2 + t +1
12. y" + y' + ty == e t + 7 34. y" + 5y' + 6y == sin t - cos 2t
576 LINEAR SECOND-ORDER EQUATIONS

35. y" - 4y' + 5y == e SI + t sin 3t - cos 3t (b) Discuss the long-term behavior of this system.
36. y" - 4y' + 4y = r2 e 2t - e 2f [Hint: Consider what happens to the general
solution obtained in part (a) as t --+ +00].
In Problems 37-40, find a particular solution to the
given higher-order equation. 43. A mass-spring system is driven by a sinusoidal exter­
nal force g (t) = 5 sin t. The mass equals I, the spring
37. ylll 2y" y' + 2y == 2t 2 + 4t - 9
constant equals 3, and the damping coefficient equals
38. y(4) - 5y" + 4y == 10 cos t - 20 sin t 4. If the mass is initially located at y( 0) == 1/2 and at
39. ylll + y" - 2y == tel + I rest, i.e. y' (0) 0, find its equation of motion.
40. y(4) - 3y'" + 3y" - y' == 6t 20
44. A mass-spring system is driven by the external force
g(t) = 2 sin 3t + 10 cos 3t. The mass equals 1. the
41. Discontinuous Forcing Tenn. In certain physical spring constant equals 5, and the damping coeffi­
models, the nonhomogeneous term, or forcing cient equals 2. If the mass is initially located at
term, g(t) in the equation y(O) == -1. with initial velocity y'(O) == 5, find its
equation of motion.
ay" + by' + cy == g(t)
may not be continuous but may have a jump disconti­ 45. Speed Bumps. Often bumps like the one depicted
nuity. If this occurs, we can still obtain a reasonable in Figure 4.11 are built into roads to discourage
solution using the following procedure. Consider the speeding. The figure suggests that a crude model of
initial value problem the vertical motion y(t) of a car encountering the
y" + 2y' + 5y == g(t}; y(O) == 0 , y'(O) == 0 , speed bump with the speed V is given by
where y(t) = 0, fort:S -L/2V •

g (t )
== {100 ift>
if 0:S t:S 31T/2
31T/2 . cos(1Tt) for ItI < L/2V
(a) Find a solution to the initial value problem for
my" + ky =
{o for L/2V :S t .
o :S t:S 31T/2.
(b) Find a general solution for t > 31T/2. (The absence of a damping term indicates that the
car's shock absorbers are not functioning.)
(c) Now choose the constants in the general solution
from part (b) so that the solution from part (a) and
the solution from part (b) agree, together with their
first derivatives, at t == 31T/2. This gives us a con­
tinuously differentiable function that satisfies the
differential equation except at t = 31T/2.

42. Forced Vibrations. As discussed in Section 4.1. a

vibrating spring with damping that is under external Figure 4.11 Speed bump
force can be modeled by
(15) my" + by' + ky = g(t) , (a) Taking m = k = I and L = 1T for convenience,
where m > 0 is the mass of the spring system. b > 0 solve this initial value problem. Thereby show that
is the damping constant, k > 0 is the spring constant, the formula for the oscillatory motion after the car
g(t) is the force on the system at time t, and y(t) is Iws traversed the speed bump is y(t) == Asint,
the displacement from the equilibrium of the spring where the constant A depends on the speed V.
system at time t. Assume b2 < 4mk. (b) Plot the amplitude IA I of the solution y(t) found
(a) Determine the form of the equation of motion for in part (a) versus the car's speed V. From the
the spring system when g(t) == sin f3t by finding graph, estimate the speed that produces the most
a general solution to equation (15). violent shaking of the vehicle.

J
580 LINEAR SECOND-ORDER EQUATIONS
1

Example 1 Find a particular solution on (-Til2, Ti/2) to

d 2y
(6) - +Y tan t + 3t - 1 .
dt 2

Solution With g(t) = tan t + 3t - 1, the variation of parameters procedure will lead to a solution. But it
is simpler in this case to consider separately the equations

(17)

d 2y
OS) dt 2 + Y = 3t
and then use the superposition principle (Theorem 3, page 196).

In Example 1 we found that

Yq(t) - (cost) In(sec t + tan t)
is a particular solution for equation (17). For equation (18) the method of undetennined coeffi­
cients can be applied. On seeking a solution to (18) of the formYr(t) = At + B, we quickly obtain
y,(t) = 3t 1.
Finally, we apply the superposition principle to get
Yp(t) = Yq(t) + y,(t)
= - (cos t) In (sec t + tan t) + 3t - 1
as a particular solution for equation (16). •

Note that we could not have solved Example 1 by the method of undetermined coeffi­
cients; the nonhomogeneity tan t is unsuitable. In Section 4.7 we'll see that another important
advantage of the method of variation of parameters is its applicability to equations whose coef­
ficients are functions of t.

EXERCISES 4.6
In Problems 1-10, find a general solution to the differen­8. y" + 9y = sec2 (3t} 9. y" + 4y = csc 2 (2t}
tial equation using the method a/variation a/parameters. 10. y" + 4y' + 4y 2t
e­ In t
1. y" + 4y = tan 2t 2. y" + Y = sec t
3. 2x" - 2x' - 4x = 2e 3t 4. y" y = 2t + 4 In Problems 11-18, find a general solution to the
5. Y" - 2y' + Y = t-Iet 6. Y" + 2/ + Y e- t differential equation.
7. y"(e) + 16y(8) sec 48 11. y" + Y = tan 2 t

..l
 PART IV: FUNDAMENTALS OF DIFFERENTIAL EQUATIONS AND BOUNDARY VALUE PROBLEMS 581

12.+ Y tan t + e 31
y" 20. Use the method of variation of parameters to show that
v" + 4v = sec4 (2t)
13.
y(t) Clcost+C2sint+ J/{s)sin(t-S)dS
y"(e) + y(e) = se2 e
14.
y" + y = 3 sec t - P +
15. is a general solution to the differential equation
~y" + 5y' + 6y = lSP y" + Y J{t),
'-..-/ 1 1 where J(t) is a continuous function on (0).
17. 2"Y" + 2y = tan 2t - 2"e l [Hint: Use the trigonometric identity sin (t s) =
18. y" - 6y' + 9y = t- 3e 31 sin t cos S sin S cos t .J
~ . 21. Suppose y satisfies the equation y" + lOy' + 25y =
~. 19. Express the solution to the initial value problem ell subject to yeO) = 1 and y' (0) - 5. Estimate
y(0.2) to within ±O.OOOI by numerically approxi­
y" - y = t1 . y( 1) =0 , y' ( 1) - 2, mating the integrals in the variation of parameters
formula.
using definite integrals. Using numerical integration
(Appendix B) to approximate the integrals. find an
approximation for y(2) to two decimal places.

The techniques of Sections 4.2 and 4.3 have explicitly demonstrated that solutions to a linear
homogeneous constant-coefficient differential equation,
(11 ay" + by' + cy 0,
are defined and satisfy the equation over the whole interval (-00, +(0). After all, such solu­
tions are combinations of exponentials, sinusoids. and polynomials.
The variation of parameters formula of Section 4.6 extended this to nonhomogeneous
constant-coefficient problems,
(2) ay" + by' + cy g(t).
yielding solutions valid over all intervals where get) is continuous (ensuring that the integrals
in (10) of Section 4.6 containing get) exist and are differentiable). We could hardly hope for
more; indeed, it is debatable what meaning the differential equation (2) would have at a point
where get) is undefined, or discontinuous.
Therefore, when we move to the realm of equations with variable coefficients of the form
(3) a2(t)y" + al(t)y' + ao(t)y g(t) ,
the most we can expect is that there are solutions that are valid over intervals where all four
"governing" functions-- a2(t). al (t), ao(t), and get) -are continuous. Fortunately, this expecta­
tion is fulfilled except for an important technical requirement~namely, that the coefficient
function a2(t) must be nonzero over the interval. t

tlndeed, the whole nature of the equation-reduction from second-order to firsl-order--changes at points where a2(1)
is zero.