Вы находитесь на странице: 1из 14

Chapter 6

Series Solutions of Differential


Equations

In this chapter we consider methods for solving differential equations using power series.
Sequences and infinite series are also involved in this treatment. Sequences, infinite series,
and power series are topics which are typically studied in Calculus II. See the Appendices
for a brief review of that material.

Section 6.1. The Taylor Series Method.

Recall that if a function g has derivatives of all orders on some open interval I and a I,
then the Taylor series expansion of g in powers of (x a) is:
X g (n) (a)
g 00 (a) g 000 (a)
g(x) = g(a) + g 0 (a)(x a) + (x a)2 + (x a)3 + = (x a)n .
2! 3! n!
n=0

Suppose we are given the initial-value problem

y 0 + p(x)y = f (x), y(a) = .

We assume that the solution y = y(x) has the Taylor series expansion
X y (n)(a)
y 00 (a) y 000 (a)
y(x) = y(a) + y 0 (a)(x a) + (x a)2 + (x a)3 + = (x a)n .
2! 3! n!
n=0

on some interval I = (a R, a + R). The initial condition, y(a) = , determines the first
term in the series expansion. From the differential equation, we have

y 0 (x) = f (x) p(x)y(x) and so y 0 (a) = f (a) p(a)y(a)

217
and this determines the second term of the expansion. Differentiating the differential equa-
tion (assuming that p and f are differentiable functions on I), we get

y 00 (x) = f 0 (x) p(x)y 0 (x) p0 (x)y(x) and so y 00 (a) = f 0 (a) p(a)y 0 (a) p0 (a)y(a).

We continue by calculating successive derivatives, followed by evaluating at x = a. Note


that the functions p and f will have to be infinitely differentiable on I in order to carry
this method out completely.

Example 1. Given the initial-value problem

y 0 y = 2ex , y(1) = 0.

Here p(x) 1 and q(x) = 2ex are infinitely differentiable on I = (, ).

Assuming that the solution y = y(x) has the Taylor series expansion

y 00 (1) y 000 (1)


y(x) = y(1) + y 0 (1)(x 1) + (x 1)2 + (x 1)3 + ,
2! 3!
we have y(1) = 0 and y 0 (1) = 2e1 + y(1) = 2e1 .

From the differential equation,

y 00 (x) = 2ex + y 0 (x) and so y 00 (1) = 2e1 + y 0 (1) = 2e1 + 2e1 = 0.

Continuing this process,

y 000(x) = 2ex + y 00 (x) and y 000 (1) = 2e1

y (4)(x) = 2e1 + y 000 (x) and y (4)(1) = 0

y (5)(x) = 2e1 + y (4)(x) and y (5)(1) = 2e1

and so on.

In general, we get

0, n even
y (n) (x) = (1)n+1 2ex + y (n1) (x) and y (n) (1) =
2e1 , n odd

Therefore,

2e1 3 2e1 5
X 2e1
y(x) = 2e 1
(x 1) + (x 1) + (x 1) + = (x 1)2n+1 .
3! 5! (2n + 1)!
n=0

Of course, we could have solved this initial-value problem using the method of Section
2.1. Using that method, the solution is y(x) = ex2 ex . You can verify that the Taylor
series expansion in powers of x 1 of this function is the series given above. 

218
This method can be applied equally well to higher order equations. For example, suppose
we are given the second order linear initial-value problem

y 00 + p(x)y 0 + q(x)y = f (x), y(a) = , y 0 (a) = .

As above, we assume that the solution y = y(x) has the Taylor series expansion

X y (n)(a)
y 00 (a) y 000 (a)
y(x) = y(a) + y 0 (a)(x a) + (x a)2 + (x a)3 + = (x a)n .
2! 3! n!
n=0

on some interval I = (aR, a+R). The initial conditions, y(a) = , y 0 (a) = , determine
the first two terms in the series expansion, and from the differential equation, we have

y 00 (x) = f (x) p(x)y 0 (x) q(x)y(x) and so y 00 (a) = f (a) p(a)y 0 (a) q(a)y(a)

This determines the third term of the expansion. Differentiating the differential equation
(again, assuming that p, q and f are differentiable functions on I), we get

y 000(x) = f 0 (x) p(x)y 00(x) p0 (x)y 0 (x) q(x)y 0 (x) q 0 (x)y(x)

and so
y 000 (a) = f 0 (a) p(a)y 00(a) p0 (a)y 0 (a) q(a)y 0 (a) q 0 (a)y(a).

We obtain the higher ordered derivatives by successive differentiation followed by evaluation


at x = a.

Example 2. Given the initial-value problem

y 00 + x2 y 0 2y = 0, y(0) = 1, y 0 (0) = 1.

Here p(x) x2 , q(x) 2, f (x) 0 are infinitely differentiable on I = (, ).


Assuming that the solution y = y(x) has the Taylor series expansion

y 00 (0) 2 y 000 (0) 3


y(x) = y(0) + y 0 (0)x + x + x + ,
2! 3!
we have y(0) = 1, y 0 (0) = 1, and from the differential equation,

y 00 (x) = x2 y 0 (x) + 2y(x) and so y 00 (0) = 2.

Continuing this process,

y 000 (x) = x2 y 00 2xy 0 + 2y 0 and y 000 (0) = 2

y (4)(x) = x2 y 000 4xy 00 2y 0 + 2y 00 and y (4)(1) = 6

y (5)(x) = x2 y (4) 6xy 000 6y 00 + 2y 000 and y (5)(0) = 16

and so on.

219
Therefore, we find that
2 2 2 6 16
y(x) = 1 x + x x3 + x4 x5 +
2! 3! 4! 5!
1 1 2
= 1 x + x2 x3 + x4 x5 + 
3 4 15

The Taylor Series method can also be applied to nonlinear initial-value problems.

Example 3. Consider the initial-value problem

y0 = 1 + y2, y(0) = 0.

Assume that the solution y has a Taylor series expansion in powers of x which converges
for x (R, R) for some R > 0. Then

y 00 (0) 2 y 000 (0) 3


y(x) = y(0) + y 0 (0)x + x + x + .
2! 3!
We will calculate the first several terms of this series. From the initial condition, y(0) = 0,
and from the differential equation y 0 (0) = 1. We calculate successive derivatives of y using
the differential equation, and we evaluate the derivatives at x = 0.

y 00 = 2yy 0 , y 00 (0) = 2y(0)y 0 (0) = 0

y 000 = 2yy 00 + 2(y 0 )2 , y 000 (0) = 2

y (4) = 2yy 000 + 2y 0 y 00 + 4y 0 y 00 = 2yy 000 + 6y 0 y 00 , y (4)(0) = 0

y (5) = 2yy (4) + 2y 0 y 000 + 6y 0 y 000 + 6(y 00)2 = 2yy (4) + 8y 0 y 000 + 6(y 00 )2 , y (5)(0) = 16

and so on. Therefore, we have


2 3 16 5 1 2
y(x) = x + x + x + = x + x3 + x5 +
3! 5! 3 15

Exercises 6.1.

Use the Taylor series method to find a power series expansion of the solution of the initial-
value problem. Find at least three non-zero terms beyond those given by the initial condi-
tions.

1. (1 x2 )y 0 xy = 1, y(0) = 0

2. y 0 + xy = ln x y(1) = 1

3. y 00 + xy 0 2y = 0, y(0) = 1, y 0(0) = 0

4. y 00 + xy 0 2y = 0, y(0) = 0, y 0(0) = 1

220
5. y 00 + ex y 0 xy = 0, y(0) = 1, y 0 (0) = 0

6. y 00 2xy 0 + x2 y = 0, y(0) = 1, y 0 (0) = 1

7. y 00 2xy = 0, y(2) = 1, y 0(2) = 0

8. y 00 2xy = x2 , y(1) = 0, y 0(1) = 2

The equations in Exercises 9 14 are nonlinear. Use the Taylor series method to find
a power series expansion of the solution of the initial-value problem. Find at least three
non-zero terms beyond those given by the initial conditions.

9. y 0 = x2 + y 2 , y(1) = 1

10. y 0 = 1 x2 y 2 , y(0) = 0

11. y 0 = sin y, y(1) = /2

12. y 0 = ln(y x), y(1) = 2


1
13. yy 00 + 3y 02 = 0, y(0) = 1, y 0(0) = 4

14. Find a Taylor series solution of

yy 0 = 1, y(1) = 1.

The differential equation is separable. Solve this initial-value problem using the
method of Section 2.2. Compare the two results.

Section 6.2. The Power Series Method; Ordinary Points

In this section we discuss a method for finding power series solutions of linear differential
equations. In Section 3.7 we stated that an nth -order linear differential equation is an
equation that can be written in the form

y (n) + pn1 (x)y (n1) + pn2 (x)y (n2) + + p1 (x)y 0 + p0 (x)y = f (x) (L)

where the functions p0 , p1 , . . . , pn1 (called the coefficients) and f (called the forcing
function or nonhomogeneous term) are continuous on some interval I. We will be seeking
solutions of (L) having the form

X
y(x) = cn (x a)n (T)
n=0

for some real number a I.

221
In the preceding section we saw that we could find a series solution provided the functions
involved in the equation were infinitely differentiable on the interval I. We will need to
elaborate on this restriction in this section. Recall that a function g defined on an interval
I containing a point x = a is analytic at x = a if g has a Taylor series expansion in
powers of (x a) with radius of convergence R > 0.

For the purposes of this section, we assume that the coefficients p0 , p1 , . . . , pn1 and
f are analytic functions. That is, if we are seeking a power series solution in the form (T),
then we assume that the coefficient functions P, Q, R and G each have a Taylor series
expansion in powers of (x a) with a positive radius of convergence. Before we present
the power series methods, we need to introduce some concepts which relate the coefficients
of a linear equation with the existence, uniqueness and behavior of solutions.

Singular Points, Ordinary Points In Section 2.1 we studied first order linear differential
equations written in the standard form

y 0 + p(x)y = f (x) (S)

where p and f were assumed to be continuous functions on an interval I. In many


examples and exercises in that section, we started with an equation in the form

P (x)y 0 + Q(x)y = G(x) (L)

Our first step, then, was to divide the equation (L) by P (x) to obtain an equation in
the form (S). We then considered (S) on intervals which did not contain points at which
P (x) = 0. A point c is a singular point of (L) if P (c) = 0.

Example 1. Consider the differential equations:

(a) (1 + x2 )y 0 + xy = 2x
Here P (x) = 1 + x2 , Q(x) = x, G(x) = 2x are continuous functions on the interval
J = (, ). Since 1 + x2 = 6 0 for all x J, this equation has no singular points.

Using the solution method presented in Section 2.1, the general solution of this equa-
tion is
C
y(x) = 2 + .
1 + x2

(b) (x 1)y 0 2y = 2
Here P (x) = x 1, Q(x) = 2, G(x) 0 are continuous functions on the interval
J = (, ). Since x 1 = 0 when x = 1, 1 is a singular point.

From Section 2.1, the general solution of this equation is

y(x) = 1 + C(x 1)2 .

222
(c) (4 x2 )y 0 4xy = 4x
Here q(x) = 4 x2 , r(x) = x, g(x) = 2x are continuous functions on the interval
J = (, ). Since 4 x2 = 0 when x = 2, 2 and 2 are singular points.

The general solution of this equation is


C
y(x) = 4 + . 
4 x2

Remarks. Equations (b) and (c) in this example illustrate possible effects on the solutions
of an equation with singular points. In (b), the one-parameter family of solutions is defined
for all x, but every member of the family has the value 1 at x = 1. Therefore, every
member of the family is a solution of the initial-value problem

(x 1)y 0 2y = 2, y(1) = 1

but none of the initial-value problems

(x 1)y 0 2y = 2, y(1) = , 6= 1

has a solution.

On the other hand, in (c), the only member of the one-parameter family of solutions
which is defined at either of the singular points is y 4 when C = 0. All other members
of the family are undefined at the singular points. 

The ideas presented above for first order linear equations also apply to second order
linear equations. The general second order linear differential equation is an equation of the
form
q0 (x)y 00 + q1 (x)y 0 + q2 (x)y = f (x)
The points at which q0 (x) = 0, if any, are called singular points of the equation

Example 2. Consider the first-order linear differential equation

y0 y = 0

As we saw in Chapter 1 and in Section 2.1, the one-parameter family of solutions is y = Cex .

Assume that the differential equation has a power series solution of the form

X
2 3
y = a0 + a1 x + a2 x + a3 x + = an xn (1)
n=0

which converges on some interval I = (R, R), 0 < R . Since a power series can be
differentiated term-by-term to produce a new series with the same radius of convergence,
we have

X
y 0 = a1 + 2a2 x + 3a3 x2 + = nan xn1 (2)
n=1

223
Substituting (1) and (2) into the differential equation, we get

X
X
n1
nan x an xn = 0 (3)
n=1 n=0
or
(a1 + 2a2 x + 3a3 x2 + ) (a0 + a1 x + a2 x2 + a3 x3 + ) = 0
which can be written

a1 a0 + (2a2 a1 )x + (3a3 a2 )x2 + = 0

Since the powers of x are linearly independent functions, it follows that

a1 a0 = 0, 2a2 a1 = 0, 3a3 a2 = 0,

and, in general, nan an1 , n = 1, 2, 3, . . .. We can use this formula to express each
coefficient an , n 1 in terms of a0 .
a1 a0 a2 a0 an1 a0
a1 = a0 , a2 = = , a3 = = , . . . an = = , ...
2 2 3 23 n n!
Setting a0 = C, an arbitrary constant, and substituting these equations into (1), we have

X C n X xn
y(x) = x =C . (4)
n! n!
n=0 n=0

Of course, we recognize the power series in (4) as the Maclaurin series expansion of ex so
we get, as expected, the solution y = Cex . 

The calculations involved in substituting the power series for y and y 0 into the differ-
ential equation can be organized using summation notation. In the example we obtained
equation (3)
X X
n1
nan x an xn = 0
n=1 n=0
The index of summation, n, is a dummy variable, comparable to the variable of integration
in a definite integral, and so changes of indices are possible. In particular, if we let m = n1,
(which means n = m + 1), then the first summation in (4) can be written

X
X
nan xn1 = (m + 1)am+1 xm .
n=1 m=0

The second summation can, of course be, written as



X
X
n
an x = am xm
n=0 m=0

Thus (3) becomes



X
X
(m + 1)am+1 xm am xm = 0
m=0 m=0

224
or

X
[(m + 1)am+1 am ] xm = 0
m=0
from which it follows that

(m + 1)am+1 am = 0, m = 0, 1, 2, . . . . (5)

Thus, the coefficients in the power series expansion of the solution can be obtained by
a0
solving (5), a so-called first order recursion formula whose solution is am = .
m!

The form of the power series in Example 1, that is, powers of x, was chosen for
convenience. In general, the form of the assumed power series is usually dictated by the
particular equation or by an initial-value problem whose solution is sought.

Example 3. Consider the initial-value problem

y 0 y = 2ex , y(1) = 0.

which we solved in Section 6.1.

Assume that the solution has a power series expansion of the form

X
y= an (x 1)n . (1)
n=0

Note that we chose an expansion about x = 1 since the initial value is specified at 1. The
derivative of y is:
X
0
y = nan (x 1)n1 . (2)
n=1
x
The Taylor series expansion of e in powers of (x 1) is:

X (1)n e1
ex = (x 1)n. (3)
n!
n=0

Substituting (1), (2), and (3) into the differential equation, we get

X X X (1)n e1
nan (x 1)n1 an (x 1)n = (x 1)n . (4)
n!
n=1 n=0 n=0

The first summation on the left can be written equivalently as



X
(n + 1)an+1 (x 1)n .
n=0

Thus, (4) becomes



X X X (1)n e1
(n + 1)an+1 (x 1)n an (x 1)n = (x 1)n
n!
n=0 n=0 n=0

225
which can be written as

X X (1)n e1
[(n + 1)an+1 an ] (x 1)n = (x 1)n.
n!
n=0 n=0

From this equation we get the recursion formula


(1)n
(n + 1)an+1 an = 2e1 , n = 0, 1, 2, 3 . . ..
n!
Successively setting n = 0, 1, 2, . . . in the recursion formula, we have

2e1 2 1 2e1
a0 a1 = 2e1 , 2a2 a1 = , 3a3 a2 = 2!, 4a4 a3 = , ...
1! e 3!
Therefore,

a0 a0 2e1 a0
a1 = a0 + 2e1 , a2 = , a3 = + , a4 = , ...
2! 3! 3! 4!
Assuming that the pattern continues, we find that
a0
, n even
n!
an = 1
a0 + 2e

n odd
n! n!
Applying the initial condition y(1) = 0 in (1), we get

X
0 = y(1) = an (1 1)n = a0
n=0

so a0 = 0. We can now conclude that




0, n even
an = 1
2e

n odd
n!
and

X 2e1
y(x) = (x 1)2n+1
(2n + 1)!
n=0
which is the result we obtained using the Taylor series method in Section 6.1. The radius
of convergence of this series is R = ; the series converges for all x (, ). 

In the next example we find a powers series solution of a second order equation.

Example 4. Consider the second order equation

y 00 + y = 0.

As you know from Chapter 3, the general solution of this equation is y = C1 cos x+C2 sin x.

226
Assume that the differential equation has a power series solution of the form

X
y = a0 + a1 x + a2 x2 + a3 x3 + a4 x4 + = an xn (1)
n=0

which converges on some interval I = (R, R), 0 < R . Differentiating twice, we have

X
y 0 = a1 + 2a2 x + 3a3 x2 + 4a4 x3 + = an xn1 ,
n=1

X
y 00 = 2a2 + 2 3a3 x + 3 4a4 x2 + = n(n 1)an2 xn2 , (2)
n=2
Substituting (1) and (2) into the differential equation, we get

X
X
n(n 1)an2 xn2 + an xn = 0 (3)
n=2 n=0

To add these two series we change the index of summation in the first series by setting
m = n 2, and we set and m = n in the second series. The result is

X
X
m
(m + 2)(m + 1)am+2 x + am xm = 0
m=0 m=0
or

X
[(m + 2)(m + 1)am+2 + am ] xm = 0.
m=0
From this equation we get the second order recursion formula

(m + 2)(m + 1)(n + 1)am+2 + am = 0, m = 0, 1, 2, 3 . . ..

Successively setting m = 0, 1, 2, . . . in the recursion formula, we have


a0 a1 a2 a0 a3 a1
a2 = , a3 = a4 = = , a5 = = ,
21 32 43 4321 54 5432
Assuming that the pattern continuous (and returning to the original variable n), we find
that
(1)n a0


, n even
(2n)!

an =
(1)n a1
, n odd



(2n + 1)!

Therefore,
a0 2 a1 3 a0 4 a1 5
y = a0 + a1 x
x x + x + x . . ..
2! 3! 4! 5!
Rearranging the terms and factoring out a0 and a1 , we get
x2 x4 x3 x5
   
y = a0 1 + . . . + a1 x + ...
2! 4! 3! 5!
which is

X (1)n 2n
X 1)n
y = a0 x + a1 x2n+1 = a0 cos x + a1 sin x. 
(2n)! (2n + 1)!
n=0 n=0

227
In the preceding examples, the differential equation had constant coefficients. The power
series method can also be applied to equations with nonconstant coefficients.

Example 5. Find power series solutions of

y 00 xy = 0

in powers of (x 1).

Set

X
y(x) = an (x 1)n . (1)
n=0
We differentiate twice and substitute into the differential equation to get

X
X
n2
n(n 1)an (x 1) x an (x 1)n = 0.
n=2 n=0

Before we can manipulate the series and combine the coefficients of the powers of (x 1),
we must write the coefficient x in terms of x 1; x = (x 1) + 1. We now have

X
X
n2
n(n 1)an (x 1) [(x 1) + 1] an (x 1)n =
n=2 n=0

X
X
X
n(n 1)an (x 1)n2 (x 1) an (x 1)n an (x 1)n =
n=2 n=0 n=0

X
X
X
n2 n+1
n(n 1)an (x 1) an (x 1) an (x 1)n = 0
n=2 n=0 n=0

By changing indices, this equation can be written



X
(2a2 a0 ) + [(n + 2)(n + 1)an+2 an an1 ] (x 1)n = 0.
n=1

Equating the coefficients of corresponding powers of (x 1), we get


a0 a1 a0 a0 a1 a0 a1
a2 = , a3 = + , a4 = + , a5 = + ,
2 6 6 24 12 30 120
Substituting these coefficients into (1), we have
 
1 2 1 3 1 4 1 5
y(x) = a0 1 + (x 1) + (x 1) + (x 1) + (x 1) + + (2)
2 6 24 30
 
1 3 1 5 1 5
a1 (x 1) + (x 1) + (x 1) + (x 1) +
6 12 120

It can be verified that the two series


1 1 1 1
y( x) = 1 + (x 1)2 + (x 1)3 + (x 1)4 + (x 1)5 +
2 6 24 30

228
and
1 1 1
y2 (x) = (x 1) + (x 1)3 + (x 1)5 + (x 1)5 +
6 12 120
are linearly independent. Therefore (2), with a0 and a1 being two independent arbitrary
constants, is the general solution of the given differential equation. 

Exercises 6.2

Find power series representations of the general solution of each of the following
differential equations. Express your powers series solutions in terms of elemen-
tary functions whenever possible.

1. y 00 + 4y = 0 in powers of x.

2. y 00 + xy = 0 in powers of x.

3. y 00 x2 y 0 y = 0 in powers of x.

4. y 00 + (x 1)y 0 + 3y = 0 in powers of (x 1).

5. y 00 (x + 1)y 0 y = 0 in powers of (x + 1).

6. (1 + x2 )y 00 + 2xy 0 = 0 in powers of x.

7. (x2 1)y 00 xy 0 + y = 0 in powers of x.

8. x2 y 00 xy 0 + 2y = 0 in powers of (x 1).

9. (1 + x2 )y 00 4xy 0 + 6y = 0 in powers of x.

10. (1 x2 )y 00 4xy 0 2y = 0 in powers of x.

Find power series solutions of the following initial-value problems.

11. y 00 3y 0 + 2y = 0, y(0) = 0, y 0 (0) = 1.

12. y 00 + xy 0 2y = 0, y(0) = 1, y 0 (0) = 0.

13. y 00 + xy 0 2y = 0, y(0) = 0, y 0 (0) = 1.

14. x2 y 00 + xy 0 4y = 0, y(1) = 0, y 0 (1) = 1.


Find power series solution of the following nonhomogeneous differential equations

15. y 00 y = ex in powers of x.

16. x2 y 00 2xy 0 + 2y = x + 1 in powers of (x + 1).

229
1
17. y 00 2xy 0 + 2y = in powers of x.
1x
18. y 00 xy 0 = ln(1 + x) in powers of x.

19. Consider the linear homogeneous equation

(ax2 + 1)y 00 + bxy 0 + cy = 0.

where a, b and c are constants.

(a) Give a necessary and sufficient condition for the equation to have a polynomial
solution.
(b) Give a necessary and sufficient condition for the equation to have exactly one
polynomial solution and one power series solution.
(c) Give a necessary and sufficient condition for the equation to have only polynomial
solutions.

230

Вам также может понравиться