9

Legendre Polynomials
PART I : LEGENDRE FUNCTION OF THE FIRST KIND
9.1. Legendre’s equation and its solution. [Bilaspur 1996; Meerut 1994]
2
The differential equation of the form (1 – x )y – 2xy + n(n + 1)y = 0 ...(1)
is called Legendre’s equation, where n is a positive interger. We now solve (1) in series of descending
powers of x. Let the series solution of (1) be

y   cm x k  m , where c0  0 . ...(2)
m 0
Differnetiating (2) and then putting the values of y, dy/dx and d2y/dx2 into (1), we have
  
(1 – x2)  cm (k – m) (k – m – 1) xk – m – 2–2x  cm (k – m) xk – m – 1 + n(n + 1)  cm xk – m = 0
m 0 m 0 m 0
 
or  cm (k – m) (k – m – 1)xk – m – 2 –  cm {(k – m) (k – m – 1)+ 2(k – m) – n(n + 1)}xk – m = 0 ....(3)
m 0 m 0
Now, (k – m) (k – m – 1) + 2 (k – m) – n(n + 1)
= (k – m)2 – (k – m) + 2(k – m) – n(n + 1) = (k – m)2 – n2 + (k – m) – n
= (k – m + n) (k – m – n) + (k – m – n) = (k – m – n) (k – m + n + 1).
Hence (3) may be re–written as
 
 cm (k – m) (k – m – 1)xk – m – 2 –  cm (k – m – n) (k – m + n + 1)xk – m = 0. ...(4)
m 0 m 0
(4) is an identity. To get the indicial equation, we equate to zero the coefficient of the highest
power of x, namely xk in (4) and obtain
c0(k – n) (k + n + 1) = 0 or (k – n) (k + n + 1) = 0, as c0  0.
...(5)
So the roots of (5) are k = n, –(n + 1). They are unequal and differ by an integer. The next
lower power of x is k – 1. So we equate to zero the coefficient of xk – 1 in (4) and obtain
c1(k – 1 – n) (k + n) = 0. ...(6)
For k = n and –(n + 1), neither (k – 1 – n) nor (k + n) is zero. So from (6), c1 = 0. Finally,
equating to zero the coefficient of xk – m in (4), we have
cm – 2(k – m + 2) (k – m + 1) – cm(k – m – n) (k – m + n + 1) = 0
(k  m  2) (k  m  1)
cm = c . ...(7)
(k  m  n) (k  m  n  1) m  2
Putting m = 3, 5, 7, ... in (7) and noting that c1 = 0, we have
c1 = c3 = c5 = c7 = ..... = 0, ...(8)
which hold good for both k = n and k = –(n + 1).
To obtain c2, c4, c6 .... etc., we consider two cases.

9.1

Created with Print2PDF. To remove this line, buy a license at: http://www.software602.com/
 9.2 Legendre Polynomials

(n  m  2) (n  m  1)
Case I. When k = n. Then, (7) becomes cm = – cm  2 . ...(9)
m(2n  m  1)
Putting m = 2, 4, 6 .... in (9), we have
n(n  1) (n  2) (n  3) n(n  1) (n  2) (n  3)
c2 = – c , c4 =  c2 = c0
2(2n  1) 0 4(2n  3) 2  4  (2n  1)(2n  3)
and so on. Re–writting (2), we have for k = n
y = c0xn + c1xn – 1 + c2xn – 2 + c3xn – 3 + c4xn – 4 + ...... ...(10)
Using (8) and the above values of c2, c4, c6. etc., (10) becomes (after replacing c0 by a)
 n ( n  1) n  2 n( n  1) ( n  2) ( n  3) n  4 
y = a  xn  x  x  .... . ...(11)
 2(2n  1) 2  4  (2 n  1) (2 n  3) 
(n  m  1) (n  m)
Case II. When k = –(n + 1). Then, (7) becomes cm = cm  2 . ... (12)
m(2n  m  1)
Putting m = 2, 4, 6, ... in (12), we have

(n  1) (n  2) (n  3) (n  4) (n  1) (n  2) (n  3) (n  4)
c2 = c0 , c4 = c2 = c0
2(2n  3) 4(2n  5) 2  4  (2n  3) (2n  5)
and so on. For k = – (n + 1), (2) gives
y = c0 x–n – 1 + c1 x–n – 2 + c2 x–n – 3 + c3 x–n – 4 + c4 x – n – 5 + ... ...(13)
Using (8) and the above values of c2, c4, c6....etc., (13) becomes (after replacing c0 by b)

y= b x
LM
n  1

(n  1) (n  2) n  3 (n  1) ( n  2) (n  3) ( n  4)  n  5
x  x

 ... . ...(14)
N 2(2n  3) 2  4  (2 n  3) (2n  5) 
Thus, two independent solutions of (1) are given by (11) and (14). If we take
a  [1  3  5...(2n  1)] / n !, the solution (11) is denoted by Pn(x) and is called Legendre’s function of
the first kind or Legendre’s polynomial of degree n. Notice that (11) is a terminating series and so
it gives rise to a polynomial of degree n. Thus Pn(x) is a solution of (1), Again, if we take
b  n /[1  3  5...(2n  1)] the solution (14) is denoted by Qn(x) and is called Legendre’s function of
the second kind. Since n is a positive integer, (14) is an infinite or non–terminating series and
hence Qn(x) is not a polynomial. Thus Pn(x) and Qn(x) are two linearly independent solutions of
(1). Hence the general solution of (1) is
y = A Pn(x) + B Qn(x), where A and B are arbitrary constants. ...(15)
Remark 1. When there is no confusion regarding the variable x, we shall use a shorter
notation Pn for Pn(x) and Pn for dPn(x)/dx, Qn for Qn(x) and Qn for dQn(x)/dx etc.
Another form of Legendre’s polynomial Pn(x)
Legendre’s polynomial of degree n is denoted and defined by

1  3  5....(2 n  1)  n n( n  1) n  2 n( n  1) (n  2) ( n  3) n  4 
Pn(x) = x  x  x  .... ...(1)
n!  2  (2 n  1) 2  4(2 n  1) (2 n  3) 
We now re–write (1) in a compact form. The general term of polynomial (1) is given by
1  3  5...(2n  1) n(n  1)...(n  2r  1)
. (1)r xn  2r . ...(2)
n 2  4  2r (2n  1) (2n  3) ...(2n  2r  1)

Created with Print2PDF. To remove this line, buy a license at: http://www.software602.com/
 Legendre Polynomials 9.3

Now, 1  3  5...(2n  1),

1  2  3  4  5  6...(2n  1) . 2n (2n)! (2 n)! (2n )!
= = = n = n . ...(3)
2  4  6 ... 2n (2  1) (2  2) (2  3) ...(2  n) 2  1  2  3 ...n 2  n!
Also, n(n – 1) ...(n – 2r + 1)
n(n  1) (n  2r  1) (n  2r ) (n  2r  1) ... 3  2  1 n!
= = ...(4)
(n  2r ) (n  2r  1) ...3  2  1 (n  2r) !

and 2  4  6 ... 2r  (2  1) (2  2) (2  3) ... (2  r )  2r  r !. ...(5)

Finally, (2n – 1) (2n – 3) ...(2n – r + 1)
(2n) (2n  1) (2n  2) (2n  3) ... (2n  2r  2) (2n  2r  1) (2n  2r) !
= 
(2n) (2n  2) (2n  4) ...(2n  2r  2) (2 n  2 r ) !
(2n) (2n  1) (2n  2) ... (2n  2r  1) (2n  2r ) (2n  2r  1) ... 3  2  1
=
2  n  2(n  1)  2(n  2) ...2 (n  r  1) . (2n  2r ) !

(2n ) !
= n
2  n( n  1) ( n  2) ...(n  r  1) (2 n  2 r ) !

(2 n) ! ( n  r ) ( n  r  1) ... 3  2  1
= n

2 (2n  2 r ) ! n ( n  1) (n  2) ... ( n  r  1) (n  r ) (n  r  1) ... 3  2  1
(2n)! (n  r)!
=  . ... (6)
2n (2n  2r)! n!
Using (3), (4), (5) and (6), the general term (2) becomes

(2n) ! n! 1 2n (2n  2r ) ! n ! n  2 r
(1)r  r  x
2n n ! (n  2r ) ! 2 r ! (2n) ! (n  r ) !

(2 n  2 r ) !
i.e. ( 1)r n x n  2r . ...(7)
2 r ! (n  r ) ! (n  2 r ) !
Since (1) is polynomial of degree n, r must be chosen so that n – 2r 0, i.e., r  n/2.
Thus, if n is even, r goes from 0 to 1 n , while if n is odd, r goes from 0 to 1 (n  1) ; that is,
2 2

for the complete polynomial (1), r goes from 0 to 1 n , where

2

n / 2, if n is even
1
n = (n  1) / 2, if n is odd
2 {
Hence the Legendre polynomial of degree n is given by
 n / 2
(2 n  2 r ) !
Pn(x) =  (1)
r0
r
r
2 r ! (n  r ) ! (n  2r ) !
x n  2r . ...(8)

9.2. Legendre’s function of the first kind or Legendre’s polynomial of degree n.

[Kanpur 2011, Ranchi 2010]
Definition. Legendre’s polynomial of degree n is denoted and defined by
1  3  5 ...(2n  1)  n n ( n  1) n  2 n( n  1) (n  2) ( n  3) n  4 
Pn(x) = x  x  x  ... ...(1)
n!  2(2n  1) 2  4  (2n  1) (2 n  3) 

Created with Print2PDF. To remove this line, buy a license at: http://www.software602.com/
 9.4 Legendre Polynomials

n/2
(2n  2r) !
or Pn(x) =  (1)
r 0
r
2 n r ! (n  r) ! (n  2r) !
x n  2r , ...(2)

where {
n / 2, if n is even
[n/2] = (n  1) / 2, if n is odd ...(3)

Legendre’s function of the second kind. Definition. This is denoted and defined by
Qn(x) =
n!   ( n  1) (n  1) (n  2) ( n  3) (n  1) (n  2) (n  3) (n  4)  ( n  5) 
x  x  x  ... . ...(4)
1  3  5 ... (2n  1)  2(2n  3) 2  4  (2n  3) (2n  5) 
Determination of first few Legendre’s polynomials.
Putting n = 0, 1, 2, 3, 4, 5, ..... in result (1), we have
1 3  2 2 1 0  1
P0(x) = 1 x 0 = 1, P1(x) = 1 x1 = x, P2(x) = x  x   (3 x 2  1),
0! 1! 2!  23  2
[Bhopal 2010]
1 3 5  3 3  2 1 1
P3(x) = x  x   (5 x 3  3 x),
3!  25  2

1  3  5  7  4 4  3 2 4  3  2 1 0  1
P4(x) = x  x  x   (35 x 4  30 x 2  3),
4!  27 247 5  8

1 3 5  7  9  5 5  4 3 5  4  3  2 1 1
P5(x) = x  x  x   (63 x 5  70 x 3  15 x) [Kanpur 2004]
5!  2  9 2  4  9  7  8
Ex. 1. Express 2 – 3x + 4x2 in terms of Legendre polynomials. [Bangalore 1995]
Sol. We have 1 = P0(x), x = P1(x), (3x2 – 1)/2 = P2(x) x2 = [2P2(x) + 1]/3 ...(1)
Now, 2 – 3x + 4x2 = 2P0(x) – 3P1(x) + (4/3) × [2P2(x) + 1], by (1)
= 2P0(x) – 3P1(x) + (8/3) × P2(x) + (4/3) × P0(x) = (10/3) × P0(x) – 3P1(x) + (8/3) × P2(x).
Ex. 2. Show that P0(x) = 1, P1(x) = x, P2(x) = (3x2 – 1)/2, P3(x) = (5x3 – 3)/2 and also expand
x + 2x3 + 2x2 – x – 3 in terms of Legendre’s polynomials.
4
[Bhopal 2004, Kakitiya 1997]
3
Ex. 3. Show that x = (2/5) × P3(x) + (3/5) × P2(x). [Nagpur 1996]
9.3. Generating function for Legendre polynomials. [Nagpur 2005]

Theorem. To show (1 – 2xz + z2)–1/2 =  z n Pn ( x ), | x |  1, | z |  1 [M.D.U. Rohtak 2006,
n0

Delhi Physics (H) 2000; Kanpur 2005; Banaglore 1994; Nagpur 2003; Meerut 1994, 96]
Or To show that Pn(x) is the coefficient of zn in the exapansion of (1 – 2xz + z2)–1/2 in assending
powers of z. [Garhwal 2004; Meerut 1998, Ravishankar 2010; Ranchi 2010]
[Note : (1 – 2xz + z2)–1/2 is called the generating function for Legendre polynomial Pn(x)].
Proof. (a) since | z | < 1 and | x |  1, we have
(1 – 2xz + z2)–1/2 = [1 – z(2x – z)]–1/2
1 3 2 1  3 ...(2n  3) n  1
=1+ 1
2
z(2x – z) + z (2x – z)2 + .... + z (2x – z)n – 1
24 2  4 ...(2n  2)
1  3 ... (2n  1) n
+ z (2x – z)n +... ...(1)
2  4 ...(2n)

Created with Print2PDF. To remove this line, buy a license at: http://www.software602.com/