Vous êtes sur la page 1sur 9

# Using Series to Solve Differential Equations

Many differential equations cant be solved explicitly in terms of nite combinations of simple familiar functions. This is true even for a simple-looking equation like
1

2xy

But it is important to be able to solve equations such as Equation 1 because they arise from physical problems and, in particular, in connection with the Schrdinger equation in quantum mechanics. In such a case we use the method of power series; that is, we look for a solution of the form y f x
n 0

cn x n

c0

c1 x

c2 x 2

c3 x 3

The method is to substitute this expression into the differential equation and determine the values of the coefcients c0 , c1, c2 , . . . . Before using power series to solve Equation 1, we illustrate the method on the simpler y 0 in Example 1. equation y
EXAMPLE 1 Use power series to solve the equation y
SOLUTION We assume there is a solution of the form 2

0.

c0

c1 x

c2 x 2

c3 x 3
n 0

cn x n

## We can differentiate power series term by term, so y c1 2c2 x 3c3 x 2

n 1

ncn x n

2c2

2 3c3 x
n 2

nn

1 cn x n

In order to compare the expressions for y and y more easily, we rewrite y as follows:
By writing out the rst few terms of (4), you can see that it is the same as (3). To obtain (4) we replaced n by n 2 and began the summation at 0 instead of 2.

y
n 0

2 n

1 cn 2 x n

Substituting the expressions in Equations 2 and 4 into the differential equation, we obtain n
n 0

2 n

1 cn 2 x n
n 0

cn x n

or
5
n 0

2 n

1 cn

cn x n

If two power series are equal, then the corresponding coefcients must be equal. Therefore, the coefcients of x n in Equation 5 must be 0: n 2 n 1 cn
2

cn

## 2 USING SERIES TO SOLVE DIFFERENTIAL EQUATIONS

cn

cn 1 n

0, 1, 2, 3, . . .

Equation 6 is called a recursion relation. If c0 and c1 are known, this equation allows us to determine the remaining coefcients recursively by putting n 0, 1, 2, 3, . . . in succession. Put n 0: c2 c0 1 2 c1 2 3 c2 3 4 c3 4 5 c4 5 6 c5 6 7 c0 1 2 3 4 c1 2 3 4 5 c0 4! 5 6 c1 5! 6 7 c0 4! c1 5! c0 6! c1 7!

Put n

1:

c3

Put n

2:

c4

Put n

3:

c5

Put n

4:

c6

5:

c7

c0 2n !
n

c1 2n 1!

## Putting these values back into Equation 2, we write the solution as y c0 c0 1 c1 x x2 2! c1 x c0

n 0

c2 x 2 x4 4! x3 3! x 2n 2n !

c3 x 3 x6 6! x5 5! c1
n 0

c4 x 4

c5 x 5 1
n

x 2n 2n ! 1
n

x7 7! 1
n

x 2n 1 2n 1 !

x 2n 1 2n 1 !

Notice that there are two arbitrary constants, c0 and c1. We recognize the series obtained in Example 1 as being the Maclaurin series for cos x and sin x. (See Equations 8.7.16 and 8.7.15.) Therefore, we could write the solution as

NOTE 1

y x

c0 cos x

c1 sin x

But we are not usually able to express power series solutions of differential equations in terms of known functions.

## USING SERIES TO SOLVE DIFFERENTIAL EQUATIONS 3

EXAMPLE 2 Solve y

2xy

0.

y
n 0

cn x n ncn x n
n 1 1

Then

and

y
n 2

nn

1 cn x n

2 n 0

2 n

1 cn 2 x n

## as in Example 1. Substituting in the differential equation, we get n

n 0

2 n

1 cn 2 x n 1 cn 2 x n

2x
n 1

ncn x n

1 n 0

cn x n cn x n
n 0

n
n 0

2 n

2ncn x n
n 1

2ncn x n
n 1 n 0

2ncn x n
n 0

2 n

1 cn

2n

1 cn x n

## This equation is true if the coefcient of x n is 0: n

7

2 n

1 cn

2n n

1 cn

cn

2n 1 cn n 1 n 2

0, 1, 2, 3, . . .

## We solve this recursion relation by putting n Put n 0: c2 1 c0 1 2 1 2 3 3 3 4 c1

0, 1, 2, 3, . . . successively in Equation 7:

Put n

1:

c3

Put n

2:

c4

c2

3 1 2 3 4 1 5 c1 2 3 4 5 3 7 c0 4! 5 6 1 5 9 c1 5! 6 7 3 7 11 c0 8!

c0

3 c0 4! 1 5 c1 5! 3 7 c0 6!

Put n

3:

c5

5 c3 4 5 7 c4 5 6 9 6 7 c5

Put n

4:

c6

Put n

5:

c7

1 5 9 c1 7!

Put n

6:

c8

11 c6 7 8 13 c7 8 9

Put n

7:

c9

1 5 9 13 c1 9!

## and the odd coefcients are given by c2n The solution is y c0 c0 1 c1 x c2 x 2 1 2 x 2! c1 x or

8

1 5 9
1

2n

4n 1!

c1

c3 x 3 3 4 x 4!

c4 x 4 3 7 6 x 6! 1 5 5 x 5! 3 7 11 8 x 8! 1 5 9 7 x 7! 1 5 9 13 9 x 9!

1 3 x 3!

c0 1

1 2 x 2! c1 x

3 7
n 2

4n 2n ! 2n 4n 1!

x 2n 3 x 2n
1

1 5 9
n 1

NOTE 2 In Example 2 we had to assume that the differential equation had a series solution. But now we could verify directly that the function given by Equation 8 is indeed a solution. NOTE 3 Unlike the situation of Example 1, the power series that arise in the solution of Example 2 do not dene elementary functions. The functions

y1 x

1 2 x 2!

3 7
n 2

4n 2n ! 4n 1! 3

x 2n
1

2 _2

T 2

and

y2 x

x
n 1

1 5 9 2n

x 2n

T _8

are perfectly good functions but they cant be expressed in terms of familiar functions. We can use these power series expressions for y1 and y2 to compute approximate values of the functions and even to graph them. Figure 1 shows the rst few partial sums T0, T2, T4, . . . (Taylor polynomials) for y1 x , and we see how they converge to y1 . In this way we can graph both y1 and y2 in Figure 2.
NOTE 4

FIGURE 1
15

## we would observe that

c0
_2.5 2.5

y0

c1

y 0

This would simplify the calculations in Example 2, since all of the even coefcients would be 0. The solution to the initial-value problem is y x x
n 1

_15

1 5 9 2n

FIGURE 2

4n 1!

x 2n

Exercises
111

10. y 11. y

x2y x y

0, xy

y0 0,

1, y0

y 0 0,

0 y 0

## Use power series to solve the differential equation. y x y xy 1y xy xy y 0, y0 1, y 0 0 y xy 0 y 0

2

1. y 3. y 5. y 7. x 2 8. y 9. y

2. y 4. x 6. y

xy 3y y 2y 0

## 12. The solution of the initial-value problem

x2y

xy

x2y

y0

y 0

is called a Bessel function of order 0. (a) Solve the initial-value problem to nd a power series expansion for the Bessel function. (b) Graph several Taylor polynomials until you reach one that looks like a good approximation to the Bessel function on the interval 5, 5 .

## 6 USING SERIES TO SOLVE DIFFERENTIAL EQUATIONS

S

c0
n 0

1.

xn n!
n

c0 e x

3. c0
n 0

x 3n 3nn!

c0 e x
1

5. c0
n 0

1 n 2n x 2 n! c1 x c0 ex
2

c1
n 0

2 n n! 2n x 2n 1 ! 1
n 1

7. c0 9.
n 0

x2 2
2

c0
n 2

2n 2

2n n! n

3 ! 2n x 2!

x 2n 2nn!

11. x
n 1

1 n225 2 3n 1!

3n

x 3n

## Solutions: Using Series to Solve Differential Equations

1. Let y(x) =
P P

cn xn . Then y 0 (x) =
P

n=0

n=1

n=1

ncn xn1

## ncn xn1 and the given equation, y 0 y = 0, becomes

n=0 P

n=0

(n + 1)cn+1 xn

n=0

so

n=0

cn 1 c0 1 1 1 c0 1 c0 , n = 0, 1, 2, . . . . Then c1 = c0 , c2 = c1 = , c3 = c2 = c0 = , c4 = c3 = , and n+1 2 2 3 3 2 3! 4 4! c0 in general, cn = . Thus, the solution is n! X X c0 n X xn y(x) = x = c0 = c0 ex cn xn = n! n! n=0 n=0 n=0 cn+1 = 3. Assuming y(x) = x2 y =
n=0 P P n=0 P

[(n + 1)cn+1 cn ]xn = 0. Equating coefcients gives (n + 1)cn+1 cn = 0, so the recursion relation is

cn xn = 0,

cn xn , we have y 0 (x) =
P

n=1

n=0

cn xn+2 =
n=2 P

## cn2 xn . Hence, the equation y0 = x2 y becomes

P

ncn xn1 =

n=0

(n + 1)cn+1 xn and

n=2

(n + 1)cn+1 xn

cn2 xn = 0 or c1 + 2c2 x +

n=2

## gives c1 = c2 = 0 and cn+1 =

cn2 for n = 2, 3, . . . . But c1 = 0, so c4 = 0 and c7 = 0 and in general n+1 c0 c0 c0 c3 = = 2 , c3n+1 = 0. Similarly c2 = 0 so c3n+2 = 0. Finally c3 = , c6 = 3 6 63 3 2! c0 c0 c6 c0 c9 = = = 3 , . . . , and c3n = n . Thus, the solution is 9 963 3 3! 3 n! X X X c0 X x3n X x3/3 n 3 n 3n 3n y (x) = x = c0 = c0 = c0 ex /3 cn x = c3n x = 3n n! 3n n! n! n=0 n=0 n=0 n=0 n=0 P
n n=0 cn x

5. Let y (x) =

## (n + 2)(n + 1)cn+2 + (n + 1)cn = 0, thus the recursion relation is cn+2 =

P P y 0 (x) = ncn xn1 and y 00 (x) = (n + 2)(n + 1)cn+2 xn . The n=1 n=0 P P P differential equation becomes (n + 2)(n + 1)cn+2 xn + x ncn xn1 + cn xn = 0 or n=0 n=1 n=0 P P P n n n since . Equating coefcients gives n=0 [(n + 2)(n + 1)cn+2 + ncn + cn ]x n=1 ncn x = n=0 ncn x cn (n + 1)cn = , (n + 2)(n + 1) n+2 c0 c2 c4 c0 c0 n = 0, 1, 2, . . . . Then the even coefcients are given by c2 = , c4 = = , c6 = = , 2 4 24 6 246 and in general, c2n = (1)n c7 = c1 c5 c1 (2)n n! c1 = , and in general, c2n+1 = (1)n = . The solution is 7 357 3 5 7 (2n + 1) (2n + 1)!
X (1)n 2n X (2)n n! 2n+1 . x + c1 x 2n n! (2n + 1)! n=0 n=0

y (x) = c0

## 8 USING SERIES TO SOLVE DIFFERENTIAL EQUATIONS

P P P 7. Let y (x) = cn xn . Then y00 = n (n 1) cn xn2 , xy 0 = ncn xn and n=0 n=0 n=0 2 P P x + 1 y00 = n (n 1) cn xn + (n + 2) (n + 1) cn+2 xn . The differential equation becomes n=0 n=0 P
n=0

## [(n + 2) (n + 1) cn+2 + [n (n 1) + n 1] cn ] xn = 0. The recursion relation is cn+2 =

(n 1) cn , n+2

n = 0, 1, 2, . . . . Given c0 and c1 , c2 =

c2n = (1)n1

1 3 (2n 3) c0 (2n 3)! c0 (2n 3)! c0 = (1)n1 n n2 = (1)n1 2n2 for 2n n! 2 2 n! (n 2)! 2 n! (n 2)! 0 c1 =0 3 c2n+1 = 0 for n = 1, 2, . . . . Thus the solution is

## 3c0 c0 c2 3c4 c0 , c4 = = 2 , c6 = = (1)2 3 , ..., 2 4 2 2! 6 2 3!

n = 2, 3, . . . . c3 =

y (x) = c0 + c1 x + c0

## 9. Let y(x) = y 00 (x) =

P

n=0 P

cn xn . Then xy 0 (x) = x

n=1

ncn xn1 =

n=1

ncn xn =

ncn xn ,

n=0

n=0

## (n + 2)(n + 1)cn+2 xn , and the equation y00 xy0 y = 0 becomes

n=0

[(n + 2)(n + 1)cn+2 ncn cn ]xn = 0. Thus, the recursion relation is ncn + cn cn (n + 1) cn = = for n = 0, 1, 2, . . . . One of the given conditions is (n + 2)(n + 1) (n + 2)(n + 1) n+2
X

cn+2 =

## y(0) = 1. But y(0) =

n=0

cn (0)n = c0 + 0 + 0 + = c0 , so c0 = 1. Hence, c2 =

1 1 c0 c2 = , c4 = = , 2 2 4 24

c4 1 1 c6 = = , . . . , c2n = n . The other given condition is y0 (0) = 0. But 6 246 2 n! X c1 y 0 (0) = = 0, c5 = 0, . . . , ncn (0)n1 = c1 + 0 + 0 + = c1 , so c1 = 0. By the recursion relation, c3 = 3 n=1 c2n+1 = 0 for n = 0, 1, 2, . . . . Thus, the solution to the initial-value problem is
X n X 2n

y(x) =

cn x =

c2n x

n=0

n=0

P

## 11. Assuming that y(x) = x2 y 0 = x2

P

n=0

ncn xn1 =

cn xn , we have xy = x
P

cn xn =

n=0

n=0

ncn xn+1 ,

cn xn+1 ,

n=1

n=0

y 00 (x) =

n=2

P

n=1

## (n + 3)(n + 2)cn+3 xn+1

[replace n with n + 3]

= 2c2 +

n=0

## USING SERIES TO SOLVE DIFFERENTIAL EQUATIONS 9

and the equation y00 + x2 y0 + xy = 0 becomes 2c2 + So c2 = 0 and the recursion relation is cn+3 =

n=0

## [(n + 3)(n + 2)cn+3 + ncn + cn ] xn+1 = 0.

But c0 = y(0) = 0 = c2 and by the recursion relation, c3n = c3n+2 = 0 for n = 0, 1, 2, . . . . Also, c1 = y0 (0) = 1, so c4 = 25 22 52 2c1 5c4 2 = , c7 = = (1)2 = (1)2 , ..., 43 43 76 7643 7! 22 52 (3n 1)2 . Thus, the solution is (3n + 1)! y(x) =
X

c3n+1 = (1)n

cn xn = x +

n=0

n=1

(1)n