University College London

Department of Physics and Astronomy

2B21 Mathematical Methods in Physics & Astronomy
Suggested Solutions for Problem Sheet M5 (2003–2004)

1. Rewrite the equation in the form

dy x3 x
=2 = 2x − 2 ,
y 1 + x2 1 + x2
which can be integrated to give
`n(y) = x2 − `n(1 + x2 ) + C . [2]

The boundary condition that y = 1 when x = 0 means that the integration

constant C = 0, and so the solution is
2
y = (1 + x2 )−1 ex · [2]

Look now for a solution of the form

∞
an xn+k ,
X
y =
n=0
∞
y0 = an (n + k) xn+k−1 ,
X

n=0

with a0 6= 0. Inserting this into

dy
(1 + x2 ) = 2x3 y ,
dx
we find
∞ ∞ ∞
n+k−1 n+k+1
an xn+k+3 .
X X X
an (n + k) x + an (n + k) x =2 [1]
n=0 n=0 n=0

The lowest power of x comes from the first term with n = 0. Hence a0 k = 0
but, since a0 6= 0, the indicial equation gives k = 0 as the unique solution. [1]
Therefore ∞ ∞ ∞
an n xn−1 + an n xn+1 = 2 an xn+3 .
X X X

n=1 n=1 n=0

The only x0 term only exists in the first sum, which means that a1 = 0 and
in general all the odd coefficients vanish. There is an x1 term also only in the
first sum so the coefficient a2 = 0 as well. [2]
Now change the dummy index n so that one sees the same power of x in all
three sums:
∞ ∞ ∞
an+4 (n + 4) xn+3 + an+2 (n + 2) xn+3 = 2 an xn+3 ,
X X X

n=−3 n=−1 n=0

1
which leads to the recurrence relation

(n + 4)an+4 + (n + 2)an+2 = 2an . [2]

We have proved that a2 = 0 and, since y = 1 when x = 0, we know that

a0 = 1. Putting n = 0 into the recurrence relation, we find that a4 = 21 and so
y ≈ 1 + 12 x4 + O(x6 ). [2]
Expanding the two factors in the exact solution as power series,
   
y ≈ 1 − x2 + x4 + O(x6 ) 1 + x2 + 12 x4 + O(x6 ) ≈ 1 + 12 x4 + O(x6 ) , [2]

which agrees with the earlier result.

2
2. Look for a solution of the second order differential equation

d2 y dy
(2x + x2 ) 2
+ (1 + x) − p2 y = 0
dx dx
in the form
∞
an xn+k ,
X
y =
n=0
∞
y0 = an (n + k) xn+k−1 ,
X

n=0
∞
y 00 = an (n + k)(n + k − 1) xn+k−2 .
X
[1]
n=0

Inserting these into the equation, we obtain

∞ ∞
n+k−1
an (n + k)(n + k − 1) xn+k
X X
2an (n + k)(n + k − 1) x +
n=0 n=0

∞ ∞ ∞
+
X
an (n + k) xn+k−1 +
X
an (n + k) xn+k − p2
X
an xn+k = 0 . [1]
n=0 n=0 n=0

Grouping like powers together, this simplifies to

∞ ∞ h i
an (n + k)(2n + 2k − 1) xn+k−1 + an (n + k)2 − p2 xn+k = 0 .
X X
[2]
n=0 n=0

If this is to be true for a range of values of x, it must be true power by power

in x. The lowest power comes from n = 0 in the first term. Since there is no
xk−1 power in the second term, we demand that

a0 k(2k − 1) = 0 .

However, by definition, a0 6= 0 so that k = 0 or k = 12 . [2]

To get the recurrence relation, change the dummy index so that we have the
same powers of x everywhere by putting n → n + 1 in the first term:
∞ ∞ h i
an+1 (n + k + 1)(2n + 2k + 1) xn+k + an (n + k)2 − p2 xn+k = 0 .
X X
[2]
n=−1 n=0

This gives us immediately the recurrence relation:

3
 an+1 (n + k)2 − p2
=− ,
an (n + k + 1)(2n + 2k + 1)
[2]
with k = 0 or k = 12 .
The series converges if, when n → ∞,
 
a n+1 
 n+1 x  an+1 
 

=   |x| < 1 . [1]
 an x n 
  
an 
This means that
 
 (n + k)2 − p2 
1
 |x| → |x| < 1 ,
 
2

 (n + k + 1)(2n + 2k + 1) 

i.e. |x| < 2. [2]

On the other hand, if p is a positive integer the recurrence relation tells us for
the k = 0 solution that
ap+1 (p2 − p2 )
=− =0. [2]
ap (p + 1)(2p + 1)
Since there are only two terms in the recurrence relation, all subsequent an
vanish and the series terminates to give the polynomial Tp (x). [1]
Given that Tp (0) = 1, i.e. a0 = 1, the recurrence relation leads to a1 = p2 a0 =
p2 so that, to order x, the k = 0 solution is

Tp (x) ≈ 1 + p2 x . [2]

Therefore

2Tp (x) Tq (x) ≈ 2(1 + p2 x)(1 + q 2 x) ≈ 2 + 2(p2 + q 2 )x . [1]

Looking at the other side,

Tp+q (x) + Tp−q (x) ≈ 1 + (p + q)2 x + 1 + (p − q)2 x

= 2 + (p2 + 2pq + q 2 )x + (p2 − 2pq + q 2 )x = 2 + 2(p2 + q 2 )x . [1]

Thus the identity is satisfied at least to first order in x.