Solutions to Problem Set 7 FYS3140

Problem 7.1
The method of variation of parameters states that if two linearly independent solutions y1 , y2 to a ho-
mogeneous second order differential equation is known, then a particular solution to a non-homogeneous
equation is given by ˆ ˆ
y2 (x)R(x) y1 (x)R(x)
yp (x) = −y1 dx + y2 dx (1)
W (x) W (x)
where R(x) is the right hand side of the DE in canonical form and W (x) is the Wronskian
 
 y1 (x) y2 (x) 
W (x) =  0 (2)
y1 (x) y20 (x) 
 

a)
The DE is
x2 y 00 − 2xy 0 + 2y = x ln x (3)
which in canonical form becomes
2 0 2 ln x
y 00 −
y + 2y = (4)
x x x
We are given solutions y1 = x, y2 = x2 to the homogeneous DE. The Wronskian is
 
 x x2 
W (x) =   = x2 (5)
1 2x 
and the particular solution becomes
ˆ ˆ
x2 ln x 2 x ln x
yp (x) = −x dx + x dx
x2 x x2 x
ˆ  ˆ 
ln x 1
= −x u du + x2 − + dx
x x2
  (6)
x ln x + 1
= − ln2 x + x2 −
2 x
 
1 2
= −x ln x + ln x + 1
2
The general solution to the DE is therefore
 
2 1 2
y(x) = c1 y1 (x) + c2 y2 (x) + yp (x) = c1 x + c2 x − x ln x + ln x + 1 (7)
2

b)
The DE is
(x2 + 1)y 00 − 2xy 0 + 2y = (x2 + 1)2 (8)
which in canonical form becomes
2x 0 2
y 00 − y + 2 y = (x2 + 1) (9)
x2 + 1 x +1

1
We are given solutions y1 = x, y2 = 1 − x2 . The Wronskian is
 
 x (1 − x2 ) 
W (x) =   = −2x2 + x2 − 1 = −x2 − 1 = −(x2 + 1) (10)
1 −2x 

We find
ˆ ˆ
(1 − x2 )(x2 + 1) 2 x(x2 + 1)
yp (x) = −x − 2
dx + (1 − x ) − 2 dx
(x + 1) (x + 1)
ˆ ˆ
= x 1 − x2 dx − (1 − x2 ) x dx
(11)
1 1 1
= x − x4 − x2 + x4
2
3 2 2
1 2 1 4
= x + x
2 6
The general solution is then
1 1
y(x) = c1 x + c2 (1 − x2 ) + x2 + x4 (12)
2 6

Problem 7.2

12.1.8)

We are to solve the DE

(x2 + 2x)y 00 − 2(x + 1)y 0 + 2y = 0 (13)
by an elementary method, and then by a series solution. The focus will be on the latter method. For
the elementary method, we want to make a guess at one solution u(x). A simplifying choice is one
that has u00 (x) = 0, for instance a linear function u(x) = Ax + B. Inserted, we find

−2(x + 1)A + 2Ax + 2B = 0 (14)

which gives
A=B (15)
That is, one solution is y1 = x + 1. The second solution is found by letting y2 = u(x)v(x). Inserted
into the original DE, this gives us a new DE,

(x + 2x)u00 − 2(x + 1)u0 + 2u v + (x2 + 2x)u v 00 + 2(x2 + 2x)u0 − 2(x + 1)u v 0 = 0

 2     
(16)

The part in front of v vanishes because u was a solution to the original DE. Changing variables to
w = v 0 and inserting u0 = 1 and u = x + 1, we find

(x2 + 2x)(x + 1)w0 + 2(x2 + 2x) − 2(x + 1)2 w = 0

 
(17)

which is a first order DE. Working it out we find

(x2 + 2x)(2x + 2)w0 − 4w = 0 (18)

which is separable. The solution is

x2 + 2x
w(x) = c1 (19)
(x + 1)2

2
from which we find
ˆ
x2 + x + 1
y2 = u(x)v(x) = (x + 1) w(x) dx = c1 (x + 1) = c1 (x2 + x + 1) (20)
(x + 1)
Finally, the general solution is

y(x) = c1 (x + 1) + c2 (x2 + x + 1) = C1 (x + 1) + C2 x2 (21)

Let us now turn to the series expansion solution. We will assume the solution y to be a power series
of the Frobenius form:
X∞
y(x) = an xn+s (22)
n=0

Inserting into the DE gives:

X X
(n + s)(n + s − 1)an xn+s + 2 (n + s)(n + s − 1)an xn+s−1
X X X (23)
−2 (n + s)an xn+s − 2 (n + s)an xn+s−1 + 2 an xn+s = 0

where all the sums run from n = 0. Since all the different powers of x are linearly independent, the
coefficients of all powers of x must be 0. In the sums where xn+s−1 appears, we change the dummy
summation indices to n0 , and then rewrite them in terms of n. As an example, the part
∞
X
(n + s)(n + s − 1)an xn+s−1 (24)
n=0

becomes
∞
X 0
(n0 + s)(n0 + s − 1)an0 xn +s−1 (25)
n0 =0

We want the powers of x to look the same across all the sums, so we require n0 − 1 = n, that is,
n0 = n + 1. Therefore we get
X∞
(n + s + 1)(n + s)an+1 xn+s (26)
n=−1

The full DE becomes, after dividing everywhere by 2,

∞  
X 1
(n + s)(n + s − 1)an + (n + s + 1)(n + s)an+1 − (n + s)an − (n + s + 1)an+1 + an xn
n=0
2
+ (s(s − 1)a−1 − sa−1 ) xs−1 = 0
(27)

where we have extracted the terms with n = −1 for clarity. The coefficient of xs−1 gives us the allowed
values for s:
s(s − 2) = 0 ⇒ s = 0 ∨ s = 2 (28)
For s = 0, the DE then requires that
 
1
n(n − 1) − n + 1 an = [−n(n + 1) + (n + 1)] an+1 (29)
2
which simplifies to
(n − 1)( 21 n − 1) 2−n
an+1 = an = an (30)
(1 + n)(1 − n) 2 + 2n

3
That is, a1 = a0 , a2 = 14 a0 , a3 = 0, an>3 = 0.
For s = 2, the DE requires
 
1
(n + 2)(n + 1) − (n + 2) + 1 an = [−(n + 2)(n + 3) + (n + 3)] an+1 (31)
2
which simplifies to
−n
an+1 = an (32)
6 + 2n
That is, a1 = 0, an>1 = 0. The full solution is therefore, renaming the two lowest coefficients to c1 , c2 :
1
y(x) = c1 (1 + x + x2 ) + c2 x2 = C1 (1 + x) + C2 x2 (33)
4
like we got with the elementary method.

12.11.2)
Solve
x2 y 00 + xy 0 − 9y = 0 (34)
using the Frobenius method. Note however that this in an Euler-Cauchy equation, so we expect a
solution of the form y(x) = C1 xm1 + C2 xm2 or similar. Inserting power series we find
∞
X
[(n + s)(n + s − 1)an + (n + s)an − 9an ] xn+s = 0 (35)
n=0

The lowest power of x occurs for n = 0, yielding

s(s − 1) + s − 9 = 0 ⇒ s = ±3 (36)

We easily see that all the coefficients an>0 = 0 for both values of s. Therefore the general solution is
simply
y(x) = C1 x0+3 + C2 x0−3 = C1 x3 + C2 x−3 (37)
which seems to agree with our initial estimation.

12.11.6)
Solve
3xy 00 + 3xy 0 + y 0 + y = 0 (38)
Inserting a generalised power series gives
∞ ∞
X X 0
[3(n + s)an + an ] xn+s + [3(n0 + s)(n0 + s − 1)an0 + (n0 + s)an0 ] xn +s−1 = 0 (39)
n=0 n0 =0

equivalent to
∞
X ∞
X
n+s
[3(n + s)an + an ] x + [3(n + s + 1)(n + s)an+1 + (n + s + 1)an+1 ] xn+s = 0 (40)
n=0 n=−1

For n = −1 we find
2
3s(s − 1) + s = 0 ⇒ s=0 ∨ s= (41)
3

4
For s = 0 we find
1
an+1 = − an (42)
1+n
which can be written as
(−1)n
an = a0 (43)
n!
For s = 2/3 we find
3
an+1 = − an (44)
5 + 3n
which gives
3 9
a1 = − a0 a2 = a0 ... (45)
5 40
The general solution is therefore
∞
(−1)n n
 
X 3 9
y(x) = c1 x + c2 x2/3 − x5/3 + x8/3 + · · · (46)
n=0
n! 5 40

The part with c1 in front is identified as e−x .

12.11.10)
Solve
2xy 00 − y 0 + 2y = 0 (47)
Inserting a generalised power series gives
∞ ∞
X 0 X
[2(n0 + s)(n0 + s − 1)an0 − (n0 + s)an0 ] xn +s−1 + 2an xn+s = 0 (48)
n0 =0 n=0

which is equivalent to
∞
X ∞
X
[2(n + s + 1)(n + s)an+1 − (n + s + 1)an+1 ] xn+s + 2an xn+s = 0 (49)
n=−1 n=0

n = −1 gives
3
2s(s − 1) − s = 0 ⇒ s=0 ∨ s= (50)
2
For s = 0 we find
2
an+1 = − an (51)
2n2 + n − 1
which gives
4
a1 = 2a0 a2 = −2a0 a3 = a0 ... (52)
9
For s = 3/2 we find
2
an+1 = − an (53)
2n2 + 7n + 5
which gives
2 4 8
a1 = − a0 a2 = a0 a3 = − a0 . . . (54)
5 5 · 14 5 · 14 · 27
Therefore the general solution is
   
2 4 3 3/2 2 5/2 4 7/2
y(x) = c1 1 + 2x − 2x + x + · · · + c2 x − x + x + ··· (55)
9 5 5 · 14