University College London

Department of Physics and Astronomy

2B21 Mathematical Methods in Physics & Astronomy
Suggested Solutions for Problem Sheet M7 (2003–2004)

1. We are given that Z +1

Pm (x)Pn (x) dx = δnm .
−1

Starting with the cases where n 6= m, since P1 (x) is odd and P0 (x) and P2 (x)
are even, it follows immediately that
Z +1 Z +1
P0 (x)P1 (x) dx = P2 (x)P1 (x) dx = 0 . [1]
−1 −1

Furthermore,
Z +1 Z +1  +1
P2 (x)P0 (x) dx = 1
2
(3x2 − 1) dx = 1
2
x3 − x =0. [2]
−1 −1 −1

For n = m = 0, Z +1 Z +1
[P0 (x)] dx = 2
dx = 2 . [1]
−1 −1

For n = m = 1, Z +1 Z +1
[P1 (x)] dx = 2
x2 dx = 2
. [1]
3
−1 −1

For n = m = 2,
Z +1 Z +1 h i+1
2
[P2 (x)] dx = 1
4
(9x4 − 6x2 + 1) dx = 1
4
9 5
5
x − 2x3 + x = 2
5
. [1]
−1 −1 −1

If P3 (x) = a[x3 + bx2 + cx + d], orthogonality with P0 (x) requires that

Z +1
[x3 + bx2 + cx + d] dx = 32 b + 2d = 0 . [1]
−1

Similarly, orthogonality with P1 (x) necessitates

Z +1
[x4 + bx3 + cx2 + dx] dx = 2
5
+ 23 c = 0. . [1]
−1

1
 Finally, imposing orthogonality with respect to P2 (x) means that
Z +1 Z +1
5 4 3 2
3
2
[x + bx + cx + dx ] dx + 1
2
[x3 + bx2 + cx + d] dx = 0 .
−1 −1

Now the second integral vanishes because of the orthogonality to P0 (x). Hence
3
5
b + 12 d = 0 . [1]

This result is incompatible with the previous relation obtained between b and
d. Hence b = d = 0. Students might get this result by claiming that P3 (x) is
an odd function. Though true, this was not given in the question and would
only receive a maximum credit of three marks.
Using the result for c, the Legendre polynomial reduces to
h i
P3 (x) = a x3 − 35 x . [1]

The easiest way of determining the value of a is from the condition that the
Legendre polynomials are normalised by Pn (x = 1) = 1. Therefore
h i
3
1=a 1− 5
= 25 a ,

and a = 52 . [3]
Alternatively, For n = m = 2,
Z +1 Z +1  
2 2
[P3 (x)] dx = a x6 − 65 x4 + 9 2
25
x dx
−1 −1

h i+1
= a2 1 7
7
x − 6 5
25
x + 3 3
25
x = 8 2
175
a = 2
7
.
−1

Hence a2 = 25/4 and we find once more that a = 52 . [3]

2. Expanding the left hand side of

∞
exp (−xt/(1 − t)) X
g(x, t) = = Ln (x) tn
1−t n=0

in powers of t gives

g(x, t) = (1 + t + t2 ) exp(−xt(1 + t)) + O(t3 ) ≈ (1 + t + t2 )(1 − xt − xt2 + 21 x2 t2 ) [1]

= 1 + t(1 − x) + t2 ( 21 x2 − 2x + 1) + O(t3 ) .
Hence L0 (x) = 1, L1 (x) = 1 − x, [1]
and L2 (x) = 12 (x2 − 4x + 2). [1]

2
Differentiating the generating function with respect to x gives
∞
∂g(x, t) t
 
L0n (x) tn .
X
=− g(x, t) = [1]
∂x 1−t n=0

Expanding g(x, t) as a power series and multiplying through by (1 − t) results

in ∞ ∞
Ln (x) tn = (1 − t) L0n (x) tn .
X X
t
n=0 n=0
∞ ∞ ∞
X
Ln (x) tn+1 =
X
L0n (x) tn −
X
L0n (x) tn+1 . [1]
n=0 n=0 n=0
Now change the summation index n → n + 1 in the first term on the right
hand side and compare the coefficients of tn+1 ;

Ln (x) = L0n+1 (x) − L0n (x) . [2]

Now ∞
exp (−xt/(1 − t)) X
g(x, t) = = Ln (x) tn
1−t n=0
∞
exp (−xu/(1 − u)) X [1]
g(x, u) = = Ln (x) un
1−u n=0
Z ∞ 1 Z ∞  
t u

−x
I= e g(x, t) g(x, u) dx = exp −x 1 + + dx
0 (1 − t)(1 − u) 0 1−t 1−u
∞
" !#
1 Z ∞
1 − ut 1 [2]
un tn .
X
= exp −x dx = =
(1 − t)(1 − u) 0 (1 − t)(1 − u) 1 − ut n=0

Using the expansion formulae, the integral must also be given by

∞ X
∞ Z ∞
tn tm e−x Ln (x) Lm (x) dx .
X
I= [1]
n=0 m=0 0

On the left hand side the power of u is always equal to the power of t. Hence
only n = m is non-vanishing on the RHS so that the required integral is
proportional to δnm . [1]
Then, comparing powers of un tn on both sides, this shows that
Z ∞
e−x Ln (x) Ln (x) dx = 1 . [1]
0

This gives the normalisation integral for the Coulomb wave functions that is
of use in the 2B22 course.