130 ● CHAPTER 4 HIGHER-ORDER DIFFERENTIAL EQUATIONS

Discussion Problems (e) By the superposition principle, Theorem 4.1.2,

both linear combinations y ! c1 y1 & c2 y2 and
37. Let n ! 1, 2, 3, . . . . Discuss how the observations
Y ! c1Y1 & c2Y2 are solutions of the differential
D nx n"1 ! 0 and D nx n ! n! can be used to find the gen-
equation. Discuss whether one, both, or neither of
eral solutions of the given differential equations.
the linear combinations is a general solution of the
(a) y# ! 0 (b) y$ ! 0 (c) y (4) ! 0 differential equation on the interval ("', ').
(d) y# ! 2 (e) y$ ! 6 (f) y (4) ! 24
40. Is the set of functions f1(x) ! e x&2, f2(x) ! e x"3 lin-
38. Suppose that y1 ! e x and y2 ! e"x are two solutions of early dependent or linearly independent on ("', ')?
a homogeneous linear differential equation. Explain Discuss.
why y3 ! cosh x and y4 ! sinh x are also solutions of
the equation. 41. Suppose y1, y2, . . . , yk are k linearly independent solu-
tions on ("', ') of a homogeneous linear nth-order
39. (a) Verify that y1 ! x 3 and y2 ! !x!3 are linearly differential equation with constant coefficients. By
independent solutions of the differential equation Theorem 4.1.2 it follows that yk&1 ! 0 is also a solution
x 2 y# " 4xy% & 6y ! 0 on the interval ("', '). of the differential equation. Is the set of solutions
(b) Show that W( y1, y2) ! 0 for every real number x. y1, y 2 , . . . , yk , yk&1 linearly dependent or linearly inde-
Does this result violate Theorem 4.1.3? Explain. pendent on ("', ')? Discuss.
(c) Verify that Y1 ! x 3 and Y2 ! x 2 are also linearly 42. Suppose that y1, y 2 , . . . , yk are k nontrivial solutions of
independent solutions of the differential equation a homogeneous linear nth-order differential equation
in part (a) on the interval ("', '). with constant coefficients and that k ! n & 1. Is the set
(d) Find a solution of the differential equation satisfy- of solutions y1, y 2 , . . . , yk linearly dependent or linearly
ing y(0) ! 0, y%(0) ! 0. independent on ("', ')? Discuss.

4.2 REDUCTION OF ORDER

REVIEW MATERIAL
● Section 2.5 (using a substitution)

● Section 4.1

INTRODUCTION In the preceding section we saw that the general solution of a homogeneous
linear second-order differential equation
a2(x)y# & a1(x)y% & a0 (x)y ! 0 (1)
is a linear combination y ! c1 y1 & c2 y2, where y1 and y2 are solutions that constitute a linearly inde-
pendent set on some interval I. Beginning in the next section, we examine a method for determining
these solutions when the coefficients of the differential equation in (1) are constants. This method,
which is a straightforward exercise in algebra, breaks down in a few cases and yields only a single
solution y1 of the DE. It turns out that we can construct a second solution y2 of a homogeneous equa-
tion (1) (even when the coefficients in (1) are variable) provided that we know a nontrivial solution
y1 of the DE. The basic idea described in this section is that equation (1) can be reduced to a linear
first-order DE by means of a substitution involving the known solution y1. A second solution y2 of
(1) is apparent after this first-order differential equation is solved.

REDUCTION OF ORDER Suppose that y1 denotes a nontrivial solution of (1) and

that y1 is defined on an interval I. We seek a second solution y 2 so that the set consist-
ing of y1 and y2 is linearly independent on I. Recall from Section 4.1 that if y1 and
y2 are linearly independent, then their quotient y2 "y1 is nonconstant on I —that is,
y 2(x)"y1(x) ! u(x) or y2(x) ! u(x)y1(x). The function u(x) can be found by substituting
y2 (x) ! u(x)y1(x) into the given differential equation. This method is called reduction
of order because we must solve a linear first-order differential equation to find u.
 4.2 REDUCTION OF ORDER ● 131

EXAMPLE 1 A Second Solution by Reduction of Order

Given that y1 ! e x is a solution of y" # y ! 0 on the interval (#$, $), use reduction
of order to find a second solution y2.

SOLUTION If y ! u(x)y1(x) ! u(x)e x, then the Product Rule gives

y' ! uex & exu', y" ! uex & 2ex u' & ex u",
and so y" # y ! ex (u" & 2u') ! 0.

Since e x % 0, the last equation requires u" & 2u' ! 0. If we make the substitution
w ! u', this linear second-order equation in u becomes w' & 2w ! 0, which is a
linear first-order equation in w. Using the integrating factor e 2x, we can write
d 2x
[e w] ! 0. After integrating, we get w ! c1e#2x or u' ! c1e#2x. Integrating
dx
again then yields u ! #12 c1 e#2x & c2. Thus
c1 #x
y ! u(x)ex ! # e & c2 ex. (2)
2
By picking c2 ! 0 and c1 ! #2, we obtain the desired second solution, y2 ! e#x.
Because W(e x, e #x) % 0 for every x, the solutions are linearly independent on
(#$, $).

Since we have shown that y1 ! e x and y2 ! e#x are linearly independent solu-
tions of a linear second-order equation, the expression in (2) is actually the general
solution of y" # y ! 0 on (#$, $).

GENERAL CASE Suppose we divide by a2(x) to put equation (1) in the standard
form
y" & P(x)y' & Q(x)y ! 0, (3)
where P(x) and Q(x) are continuous on some interval I. Let us suppose further that
y1(x) is a known solution of (3) on I and that y1(x) % 0 for every x in the interval. If
we define y ! u(x)y1(x), it follows that
y' ! uy'1 & y1u', y" ! uy"1 & 2y'1u' & y1u "
y" & Py' & Qy ! u[ y1" & Py1' & Qy1] & y1u" & (2y1' & Py1)u' ! 0.

zero

This implies that we must have

y1u " & (2y'1 & Py1)u' ! 0 or y1w' & (2y'1 & Py1)w ! 0, (4)

where we have let w ! u'. Observe that the last equation in (4) is both linear and
separable. Separating variables and integrating, we obtain
dw y'
& 2 1 dx & P dx ! 0
w y1

ln# wy21 # ! #! P dx & c or wy21 ! c1e#"P d x.

We solve the last equation for w, use w ! u', and integrate again:

u ! c1 ! e#"P d x
y21
dx & c2.
132 ● CHAPTER 4 HIGHER-ORDER DIFFERENTIAL EQUATIONS

By choosing c1 ! 1 and c2 ! 0, we find from y ! u(x)y1(x) that a second solution of

equation (3) is

y2 ! y1(x) ! e#"P(x) d x
y21(x)
dx. (5)

It makes a good review of differentiation to verify that the function y2(x) defined in
(5) satisfies equation (3) and that y1 and y2 are linearly independent on any interval
on which y1(x) is not zero.

EXAMPLE 2 A Second Solution by Formula (5)

The function y1 ! x 2 is a solution of x 2 y" # 3xy$ % 4y ! 0. Find the general solu-

tion of the differential equation on the interval (0, &).

SOLUTION From the standard form of the equation,

3 4
y" # y$ % 2 y ! 0,
x x

we find from (5) y2 ! x2 ! e3"d x /x

x4
dx 3
; e3"d x /x ! eln x ! x 3

! x2 ! dx
x
! x 2 ln x.

The general solution on the interval (0, &) is given by y ! c1 y1 % c2 y2; that is,
y ! c1x 2 % c2 x 2 ln x.

REMARKS

(i) The derivation and use of formula (5) have been illustrated here because this
formula appears again in the next section and in Sections 4.7 and 6.2. We use (5)
simply to save time in obtaining a desired result. Your instructor will tell you
whether you should memorize (5) or whether you should know the first princi-
ples of reduction of order.
(ii) Reduction of order can be used to find the general solution of a nonhomo-
geneous equation a2 (x)y" % a1(x)y$ % a 0 (x)y ! g(x) whenever a solution y1 of
the associated homogeneous equation is known. See Problems 17 – 20 in
Exercises 4.2.

EXERCISES 4.2 Answers to selected odd-numbered problems begin on page ANS-4.

In Problems 1 –16 the indicated function y1(x) is a solution

7. 9y" # 12y$ % 4y ! 0; y1 ! e 2x/3
of the given differential equation. Use reduction of order or
formula (5), as instructed, to find a second solution y2(x). 8. 6y" % y$ # y ! 0; y1 ! e x/3
1. y" # 4y$ % 4y ! 0; y1 ! e 2x 9. x 2 y" # 7xy$ % 16y ! 0; y1 ! x 4
2. y" % 2y$ % y ! 0; y1 ! xe#x 10. x 2 y" % 2xy$ # 6y ! 0; y1 ! x 2
3. y" % 16y ! 0; y1 ! cos 4x 11. xy" % y$ ! 0; y1 ! ln x
4. y" % 9y ! 0; y1 ! sin 3x 12. 4x 2 y" % y ! 0; y1 ! x 1/2 ln x
5. y" # y ! 0; y1 ! cosh x 13. x 2 y" # xy$ % 2y ! 0; y1 ! x sin(ln x)
6. y" # 25y ! 0; y1 ! e 5x 14. x 2 y" # 3xy$ % 5y ! 0; y1 ! x 2 cos(ln x)
 4.3 HOMOGENEOUS LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS ● 133

y2 % em2 x or of the form y2 % xem1 x, m1 and m2

15. (1 ! 2x ! x 2)y" # 2(1 # x)y$ ! 2y % 0; y1 % x # 1
constants.
16. (1 ! x 2)y" # 2xy$ % 0; y1 % 1
(c) Reexamine Problems 1 – 8. Can you explain why the
In Problems 17 –20 the indicated function y1(x) is a solution statements in parts (a) and (b) above are not
of the associated homogeneous equation. Use the method contradicted by the answers to Problems 3 – 5?
of reduction of order to find a second solution y2(x) of the 22. Verify that y1(x) % x is a solution of xy" ! xy$ # y % 0.
homogeneous equation and a particular solution of the given Use reduction of order to find a second solution y2(x) in
nonhomogeneous equation. the form of an infinite series. Conjecture an interval of
17. y" ! 4y % 2; y1 % e !2x definition for y2(x).
18. y" # y$ % 1; y1 % 1
Computer Lab Assignments
19. y" ! 3y$ # 2y % 5e 3x; y1 % e x
23. (a) Verify that y1(x) % e x is a solution of
20. y" ! 4y$ # 3y % x; y1 % e x
xy" ! (x # 10)y$ # 10y % 0.
Discussion Problems
(b) Use (5) to find a second solution y2(x). Use a CAS to
21. (a) Give a convincing demonstration that the second- carry out the required integration.
order equation ay" # by$ # cy % 0, a, b, and c con- (c) Explain, using Corollary (A) of Theorem 4.1.2, why
stants, always possesses at least one solution of the the second solution can be written compactly as
form y1 % em1 x, m1 a constant.
10
1
(b) Explain why the differential equation in part (a) y2(x) % ! xn.
n%0 n!
must then have a second solution either of the form

4.3 HOMOGENEOUS LINEAR EQUATIONS

WITH CONSTANT COEFFICIENTS
REVIEW MATERIAL
● Review Problem 27 in Exercises 1.1 and Theorem 4.1.5
● Review the algebra of solving polynomial equations (see the Student Resource
and Solutions Manual)

INTRODUCTION As a means of motivating the discussion in this section, let us return to first-
order differential equations—more specifically, to homogeneous linear equations ay$ # by % 0,
where the coefficients a & 0 and b are constants. This type of equation can be solved either by
separation of variables or with the aid of an integrating factor, but there is another solution method,
one that uses only algebra. Before illustrating this alternative method, we make one observation:
Solving ay$ # by % 0 for y$ yields y$ % ky, where k is a constant. This observation reveals the
nature of the unknown solution y; the only nontrivial elementary function whose derivative is a
constant multiple of itself is an exponential function e mx. Now the new solution method: If we substi-
tute y % e mx and y$ % me mx into ay$ # by % 0, we get
amemx # bemx % 0 or emx (am # b) % 0.
Since e mx is never zero for real values of x, the last equation is satisfied only when m is a solution or
root of the first-degree polynomial equation am # b % 0. For this single value of m, y % e mx is a
solution of the DE. To illustrate, consider the constant-coefficient equation 2y$ # 5y % 0. It is not
necessary to go through the differentiation and substitution of y % e mx into the DE; we merely have
to form the equation 2m # 5 % 0 and solve it for m. From m % !52 we conclude that y % e!5x/2 is a
solution of 2y$ # 5y % 0, and its general solution on the interval (!', ') is y % c1e!5x/2.
In this section we will see that the foregoing procedure can produce exponential solutions for
homogeneous linear higher-order DEs,
an y(n) # an!1 y(n!1) # ( ( ( # a2 y" # a1 y$ # a0 y % 0, (1)
where the coefficients ai , i % 0, 1, . . . , n are real constants and an & 0.