Legendre, Hermite and

UNIT 3 LEGENDRE, HERMITE AND Laguerre Polynomials

LAGUERRE POLYNOMIALS
Structure Page No.
3.1 Introduction 83
Objectives
3.2 Legendre Polynomials 84
Rodrigue’s Formula
Generating Function
Recurrence Relations
Orthogonality Property
Legendre Function of the Second Kind
3.3 Hermite and Laguerre Polynomials 100
3.4 Applications to Physical Situations 106
3.5 Summary 111
3.6 Solutions/Answers 112

3.1 INTRODUCTION
In Unit 2, we discussed methods of finding series solutions of second order, linear,
homogeneous differential equations with variable coefficients. In this unit and the unit
to follow, we shall apply these methods to a few special second order equations
occurring frequently in applied mathematics, engineering and physics. These
equations are Legendre’s, Bessel’s, Hermite’s and Laguerre’s equations. The solution
to these equations, that occur in applications, are referred to as special functions.
They are called ‘special’ as they are different from the standard functions like sine,
cosine, exponential, logarithmic etc. In this unit, we shall concentrate on Legendre,
Hermite and Laguerre polynomials which are polynomial solutions to Legendre’s,
Hermite’s and Laguerre’s differential equations.

Legendre polynomials first arose in the problem of expressing the Newtonian potential
of a conservative force field in an infinite series involving the distance variable of two
points and their included central angle. Other similar problems dealing with either
gravitational potential or electrostatic potentials and steady-state heat conduction
problems in spherical solids, also lead to Legendre polynomials. Other polynomials
which commonly occur in applications are the Hermite and Laguerre. They play an
important role in quantum mechanics, and in probability theory.

We have started the unit by obtaining the power series solutions of Legendre’s
differential equation in Sec.3.2 and introduced Legendre polynomials and Legendre
functions of both first and second kind. Various properties of Legendre polynomials
are also discussed in this section. Polynomials solutions of Hermite’s and Laguerre’s
equations and their properties are discussed in Sec.3.3. Applications of the Legendre
and Hermite polynomials to physical situtations are discussed in Sec.3.4.
Objectives
After studying this unit you should be able to
• obtain the power series solutions of Legendre’s differential equation;
• derive Rodrigue’s formula, for Legendre polynomials;
• obtain Legendre polynomials through generating function;
• use recurrence relations for Legendre polynomials and its orthogonality property
in various applications;
• derive Rodrigue’s formula, generating function, recurrence relations and
orthogonal property of Hermite and Laguerre polynomials and use them in various
applications.
83
Ordinary Differential
Equations 3.2 LEGENDRE POLYNOMIALS
The differential equation of the form
d2y dy
(1 − x 2 ) 2
− 2x + n (n + 1) y = 0 (1)
dx dx
is called Legendre’s differential equation or simply Legendre’s equation, where ‘n’
is a real constant. We can also write Legendre’s equation in the form
d  2 dy 
(1 − x )  + n ( n + 1) y = 0 (2)
dx  dx 
In Eqn.(1), if we divide by (1 − x 2 ) throughout then we can write the equation as
d2y 2x dy 1
2
− + n (n + 1) y = 0, x ≠ ±1. (3)
dx 1 − x dx 1 − x 2
2

2x n (n + 1)
Here − 2
and are analytic functions for x = 0 . Also if we use
1− x 1− x2
Binomial expressions, then both these converge for | x | < 1 . Hence x = 0 is an
ordinary point of Legendre’s Eqn.(1) and this suggests that Eqn.(1) has a power series
solution about x = 0 .
Assume the series solution
∞
y( x ) = ∑c
k =0
kx
k
(4)

Differentiting Eqn.(4) w.r. to x , we get

∞
y ′( x ) = ∑c
k =1
kk x k −1
∞
and, y ′′( x ) = ∑c
k =2
k k(k − 1) x k − 2

Substituting for y( x ), y ′( x ) and y ′′( x ) in Eqn.(1), we get

∞ ∞ ∞
(1 − x 2 ) ∑c
k =2
k k (k − 1) x k − 2 − 2 x ∑c
k =1
kk x k −1 + n (n + 1) ∑c
k =0
kx
k
=0

∞ ∞ ∞ ∞
i.e., ∑c
k =2
k k (k − 1) x k − 2 − ∑c
k =2
k k (k − 1) x k − 2 ∑c
k =1
k kx
k
+ n (n + 1) ∑c
k =0
kx
k
=0

∞ ∞ ∞
i.e., ∑c
k =0
k + 2 (k + 2) (k + 1) x k − ∑c
k =0
k + 2 (k + 2) (k + 1) x k + 2 − 2 ∑c
k =0
k +1 ( k + 1) x k +1

∞
+ n (n + 1) ∑c
k =0
kx
k
=0 (5)

Equating the coefficients of various powers of x on both sides of Eqn.(5), we get

n ( n + 1)
For x 0 : c 2 (2.1) + n ( n + 1) c 0 = 0 i.e., c 2 = − c0 (6)
2
2 − n (n + 1)
For x 1 : c 3 (3.2) − 2 c 1 + n (n + 1) c1 = 0 i.e., c 3 = c1 (7)
3.2
For x k : c k + 2 (k + 2) (k + 1) − c k k (k − 1) − 2 c k k + n (n + 1) c k = 0
k (k − 1) + 2k − n (n + 1)
i.e., c k + 2 = c k , for k ≥ 0
(k + 2) (k + 1)
( k − 2) (k − 1) − n (n + 1)
i.e., c k = c k − 2 , for k ≥ 2
k (k − 1)
− [n − (k − 2) ][n + (k − 1) ]
= c k − 2 , for k ≥ 2 (8)
k (k − 1)
84
From recurrence relation (8), alongwith Eqns.(6) and (7), we observe that coefficients Legendre, Hermite and
Laguerre Polynomials
c k for k even are multiples of c 0 while those for k odd are multiples of c1 . Thus, in
general, we have
− (n − 2k + 2) (n + 2k − 1)
c 2k = c 2 k − 2 , for k ≥ 1
2k (2k − 1)
(n − 2k + 2) (n − 2k + 4) K (n − 2). n (n + 1) (n + 3) L (n + 2k − 1)
= (−1) k c0 ,
(2k )!
for k ≥ 1 (9)
− (n − 2k + 1) (n + 2k )
and, c 2 k +1 = c 2 k −1 , for k ≥ 1
(2k + 1) (2k )
(n − 2k + 1) (n − 2k + 3) K (n − 1) (n + 2) (n + 4) K (n + 2k )
= (−1) k c1 ,
(2k + 1)!
for k ≥ 1 (10)
Substituting from relations (9) and (10) in relation (4), the power series solution of
Eqn.(1) can be written as
 n (n + 1) 2 (n + 3) (n + 1) n (n − 2) 4 
y( x ) = c 0 1 − x + x − L
 (n )! (4)! 
 (n + 2) (n − 1) 3 (n + 4) (n + 2) (n − 1) (n − 3) 5 
+ c1  x − x + x − L
 (3)! (5)! 

 ∞
(n − 2k + 2) ( n − 2 k + 4) L ( n − 2).n ( n + 1) ( n + 3) K (n + 2k − 1) 2 k 
= c 0 1 +

∑ (−1)
k =1
k
( 2k )!
x 


 ∞
(n − 2k + 1) (n − 2 k + 3) L ( n − 1).( n + 2) ( n + 4) K (n + 2k ) 2 k +1 
= c1  x +

∑ (−1)
k =1
k
(2 k + 1)!
x 


= c 0 y 1 ( x ) + c1 y 2 ( x ) (11)

where,
∞
( n − 2k + 2) (n − 2k + 4) K (n − 2).n ( n + 1) ( n + 3) K ( n + 2k − 1) 2 k
y1 (x) = 1 + ∑ (−1)
k =1
k
( 2k )!
x 12)

and
∞
(n − 2k + 1) (n − 2k + 3)K( n − 1).n ( n + 2) ( n + 4)K ( n + 2k ) 2 k +1
y2 (x ) = x + ∑ (−1)
k =1
k
( 2k + 1)!
x (13)

when n is not an integer both y1 ( x ) and y 2 ( x ) converge for | x | < 1 .

Since c 0 and c1 are arbitrary constants, and y 1 and y 2 are linearly independent
solutions of Eqn.(1), solution (11) can be considered as the general solution of
Legendre’s Eqn.(1). The functions defined by Eqns.(12) and (13) are called
Legendre Functions.
In case, n is a non-negative integer, then y 1 ( x ) reduce to a polynomial of degree n , if
n even; and the same is true for y 2 ( x ) if n is odd. For example, for n even, we
have
For n = 0 : y1 ( x ) = 1
For n = 2 : y1 ( x ) = 1 − 3x 2
35 4
For n = 4 : y1 ( x ) = 1 − 10x 2 + x
3
and so on. Thus y 1 ( x ) reduces to a polynomial of even powers. For these values of
n, y 2 ( x ) remains an infinite series.
85
Ordinary Differential In general, for n = 2m , where m is an integer,
Equations
m
(2m − 2k + 2) ( 2m − 2k + 4) K 2m( 2m + 1) ( 2m + 3)K( 2m + 2k − 1) 2 k
y1 ( x ) = 1 + ∑ (−1)
k =1
k
( 2k )!
x (14)

and for k = m + 1, k = m + 2, K etc. the terms in the summation on the R.H.S. of

Eqn.(14) are zero.
When n is odd, y 1 ( x ) remains an infinite power series and y 2 ( x ) reduces to
polynomial of odd powers. For example,
For n = 1 : y 2 ( x ) = x ,
5
For n = 3 : y 2 ( x ) = x − x 3 .
3
14 21
For n = 5 : y 2 ( x ) = x − x 3 + x 5 .
3 5
In general, for n = 2m + 1 , we may obtain

∞
(2m − 2k + 2) (2m − 2k + 4)K2m.(2m + 3) (2m + 5)K(2m + 2k + 1) 2 k +1 . (15)
y 2 (x ) = x + ∑ (−1)
k =1
k
( 2k + 1)!
x

These polynomials multiplied by suitable constants are called Legendre polynomials.

The Legendre polynomials are denoted by Pn ( x ) where n denotes the order of the
polynomial. Therefore, when n takes positive integral values, one of the linearly
independent solutions of Eqn.(1) is a Legendre polynomial and the second solution is
an infinite series. In order to explicitly write the expressions for the Legendre
polynomials, we need to evaluate the multiplicative constants. The values of the
multiplicative constants are obtained by setting Pn (1) = 1 . It is too cumbersome to use
the recurrence relation given by Eqn.(8) or the polynomials given above to determine
the multiplicative constants. It is easy to use the Rodrigue’s formula which we shall
discuss now to find the expressions for the Legendre Polynomials.
3.2.1 Rodrigue’s Formula
We shall now show that a polynomial of degree n , obtained by applying the binomial
theorem to ( x 2 − 1) n and differentiating it n times, is a solution of the Legendre’s
differential equation. The expression for Pn ( x ) can then be obtained by using
Pn (1) = 1 .
Consider the function
f ( x ) = ( x 2 − 1) n . (16)
Differentiating it w.r. to x , we get
f ′( x ) = n ( x 2 − 1) n −1 .2x
⇒ ( x 2 − 1) f ′( x ) − 2n f ( x ) x = 0 . (17)
Differentiating Eqn.(17), (n + 1) times by using the Leibnitz rule, we get
 2 ( n + 2) n (n + 1) 
( x − 1) f ( x ) + (n + 1) (2 x ) f ( n +1) ( x ) + .2 f ( n ) ( x ) 
 2 
− 2n[ x f n +1 ( x ) + (n + 1).1. f ( n ) ( x ) ] = 0 ,
or, ( x 2 − 1) f ( n + 2 ) ( x ) + 2 x f ( n +1) ( x ) − n (n + 1) f ( n ) ( x ) = 0 ,
d2 d (n)
or, (1 − x 2 ) 2
[f ( n ) ( x ) ] − 2 x [f ( x ) ] + n ( n + 1) [f ( n ) ( x ) ] = 0 .
dx dx
dn dn
This shows that f ( n ) ( x ) = [f ( x )] = n [ ( x 2 − 1) n ] satisfies Lagendre’s Eqn.(1).
dx n dx
86
 Legendre, Hermite and
dn
Thus Legendre polynomial Pn ( x ) must be a constant multiple of [( x 2 − 1) n ] , i.e., Laguerre Polynomials
dx n
dn
Pn ( x ) = A [( x 2 − 1) n ] , where A is a constant. The constant A is determined by
dx n
setting Pn (1) = 1.
d
Now Pn ( x ) = A D n [ ( x 2 − 1) n ] = A D n [(x − 1) n ( x + 1) n ], where D =
dx
n
=A ∑
r =0
n
C r [D n − r { ( x − 1) n } ].[D r { ( x + 1) n ]

= (n )! A( x + 1) n + terms with ( x − 1) as factor.

Thus, Pn (1) = 1 ⇒ A. (n )!. 2 n = 1

1
i.e., A= .
(n )! 2 n
1 dn
Hence, Pn (x) = n n
{ (x 2 − 1) n } . (18)
(n)!2 dx

It is called Rodrigue’s Formula for Legendre Polynomials.

Using the Binomial theorem we can write
n n
(−1) r n! x 2 n − 2 r
( x 2 − 1) n =
r =0
n
∑
C r ( x 2 ) n − r (−1) r =
r =0
∑
r!(n − r )!
Differentiating the above expression n times and substituting in Eqn.(18), we get
1 n (−1) r n! d n ( x 2 n − 2 r )
Pn ( x ) = ∑
n! 2 n r =0 r!(n − r )! dx n
1 N
( −1) r (2n − 2r)! x n − 2r
=
2n
∑
r = 0 r! (n − r)! (n − 2r)!
(19)

where, N = n / 2 or (n − 1) / 2 whichever is an integer. In notation we usually write

 n / 2 , if n is even
n  
N = [n / 2] , where   =  n − 1 .
2  , if n is odd
 2
This function Pn ( x ) as given by Eqn.(19) is called Legendre function of the first
kind or Legendre polynomial of degree n.
Example 1: Show that
 1.3.5.K (n − 1)
(−1) n / 2 , if n is even
Pn (0) =  2 .4 .K n .
0 , if n is odd
Solution: From Rodrigue’s formula for Legendre polynomials ,
1 dn  2
Pn ( x ) = n n 

x −1 
n
(
)
2 n! dx
1 dn
= n
2 n! dx n
[
(x + 1)n (x − 1)n ]
1  n n!
= n  n!( x + 1) n + C1 n ( x + 1) n −1 ( x − 1) n! + n C 2 n (n − 1) ( x + 1) n − 2 ( x − 1) 2
2 n!  2
+ L + n!( x − 1) n ]
1
[ n n
= n n!( x + 1) n + ( C1 ) 2 ( x + 1) n −1 ( x − 1) n! + ( C 2 ) 2 ( x + 1) n − 2 ( x − 1) 2 n!
2 n!
+ L + n!( x − 1) n ]
87
Ordinary Differential Putting x = 0 , we get
Equations
1
[ ]
Pn (0) = n n! ( n C 0 ) 2 − ( n C1 ) 2 + ( n C 2 ) 2 + L + (−1) n ( n C n ) 2
2 n!
1 0 , if n is odd n n
= n n!  n/2 n [Q C k = C n − k ]
2 n! (−1) . C n / 2 , if n is even
∴ Pn (0) = 0 if n is odd
1
and Pn (0) = n (−1) n / 2 n C n / 2 , if n is even
2
1 n!
= n ( −1) n / 2
2 n n
  !  !
2 2
1.3.5.K (n − 1)
= (−1) n / 2
2.4.6.K n
***
You may now try the following exercises.
E1) Using Rodrigue’s Formula, show that
1 1
(−1) n
a) ∫ f ( x )Pn ( x )dx = n
2 n! ∫ ( x 2 − 1) n f ( n ) ( x )dx.
−1 −1
1 n +1
2 ( n!) 2
b) ∫ x n Pn ( x )dx =
(2n + 1)!
.
−1

E2) Prove that all the roots of Pn ( x ) = 0 are real, distinct and lie between − 1 and
+1.
1
E3) Show that ∫ Pn ( x ) dx = 0 , except when n = 0 , in which case, the value of the
−1
integral is 2.
Thus you have seen that Legendre polynomials are obtained either as polynomial
solutions of Legendre’s equation or by Rodrigue’s formula. We shall now discuss the
historical discovery of the Legendre polynomials through expansion of the generating
function into a particular type of series. We shall show that these polynomials can
also be obtained as coefficients of t n in the expansion of (1 − 2 xt + t 2 ) −1 / 2 .

3.2.2 Generating Function

The subject of potential theory is concerned with the forces of attraction due to the
presence of a gravitational field. Central to the discussion of problems of gravitational
attraction is Newton’s law of universal gravitation.
Every particle of matter in the universe attracts every other particle with a force
whose direction is that of the line joining the two, and whose magnitude is directly as
the product of their masses and inversely as the square of their distance from each
other.
The force field generated by a single particle is usually considered to be conservative.
That is, there exists a potential function V such that the gravitational force F at a
point of free space (i.e., free of point masses) is related to the potential function
according to
F = −∇V (20)
where the minus sign is conventional. If r denotes the distance between a point mass
and a point of free space, the potential function can be shown to have the form
88
 k Legendre, Hermite and
V(r ) = (21) Laguerre Polynomials
r
where k is a constant whose numerical value does not concern us. Because of
spherical symmetry of the gravitational field, the potential function V depends on
only the radial distance r .
To obtain Legendre’s results, let us suppose that a particle of mass m is located at
point P , which is ‘a’ units from the origin of our coordinate system (see Fig.1).
P
r

a Q

φ b
O Fig.1
Let the point Q represent a point of free space r units from P and b units from the
origin O . Let us assume that b > a , then, from the law of cosines, we find the
relation
r 2 = a 2 + b 2 − 2ab cos φ (22)
where φ is the central angle between segments OP and OQ . By rearranging the
terms and factoring out b 2 , it follows that
 a a 
2
r 2 = b 2 1 − 2 cos φ +    , a < b (23)
 b  b  
We introduce the parameters
a
t = , x = cos φ (24)
b
and taking the square root of Eqn.(23), we obtain

(
r = b 1 − 2 xt + t 2 )
1/ 2
(25)
Finally, substituting Eqn.(25) into Eqn.(21), we obtain the potential function
k
(
V = 1 − 2 xt + t 2
b
−1 / 2
)
, 0 < t <1 (26)

We refer to the function w ( x , t ) = (1 − 2 xt + t 2 ) −1 / 2 as the generating function of the

Legendre polynomials. We shall now develop w ( x, t ) in a power series in the
variable t and show that
∞
1
1 − 2 xt + t 2
=
n =0
∑t n Pn ( x ), t ≠1 (27)

We write
(1 − 2 xt + t 2 ) −1 / 2 = [1 − (2 x − t ) t ] −1 / 2 = (1 − u ) −1 / 2
where, u = (2 x − t ) t
Expanding in a binomial series, we get
1 (1 / 2) (3 / 2) 2 (1 / 2) (3 / 2) (5 / 2) 3
(1 − u ) −1 / 2 = 1 + u + u + u +L
2 2! 3!
1 2! 4! 6! (2n )!
= + 2
u+ 2 4
u2 + 2 6
u 3 +L+ 2 2n
u n +L
2 (1! ) 2 (2! ) 2 (3 ! ) 2 (n! ) 2

Substituting u = (2 x − t ) t , we get 89
Ordinary Differential 2! 4!
Equations (1 − 2xt + t 2 ) −1 / 2 = 1 + 2 2
(2x − t ) t + 2 4
(2x − t ) 2 t 2 + L
(1) 2 (2!) 2
(2n − 2r )! (2n )!
+ 2 2n−2r
(2x − t ) n − r t n −r +L+ 2 2n
(2x − t ) n t n + L (28)
[ (n − r )!] 2 (n!) 2
Since we are interested in the coefficients of t n in the above series, we consider the
(n − r ) th term of the series and expand the product (2 x − t ) n − r t n − r in powers of t .

We have
(n − r ) ! 2 n −2r
t n − r [ n − r C r ( − t ) r (2x ) n −2 r ] = (−1) r t n x n −2 r
r !( n − 2 r ) !
The (n − r ) th term of series (28) then becomes
(2n − 2r ) ! (n − r ) !2 n − 2 r (−1) r (2n − 2r )! t n x n − 2 r
. (−1) r t n x n − 2 r = n L (29)
[ (n − r ) !] 2 2 2 n − 2r r ! ( n − 2r ) ! 2 (n − r ) ! r!(n − 2r ) !

n (n − 1)
Summing the r.h.s of Eqn(29) for r = 0, 1, K , N , where N = or , whichever
2 2
is an integer, we get on using Eqn.(19)
N
(−1) r (2n − 2r ) ! x n − 2 r n
∑
r =0 2 n
r! ( n − r ) ! ( n − 2 r )!
t = Pn ( x ) t n (30)
∞
Therefore, (1 − 2xt + t 2 ) −1 / 2 = ∑ P (x)t
n =0
n
n
,t ≠1

Thus Pn ( x ) can be obtained by the generating function. You may observe here that
when n is an even number the polynomial Pn ( x ) is an even function; and when n is
odd, the polynomial is an odd function and the relation
Pn ( − x ) = (−1) n Pn ( x ) , (31)

holds for n = 0, 1, 2, K .
In Fig.2 below, we have given the graphs of Pn ( x ), n = 0, 1, 2, 3, 4, over the interval
−1≤ x ≤1.
Pn ( x )

1 P0

P1
P3
P4

x
−1 1

P2

−1
Fig.2: Graph of Pn ( x ), n = 0, 1, 2, 3, 4

Further, considering Eqn.(26) with x = cos φ and t = a / b , we find that the potential
function has the series expansion
90
 n Legendre, Hermite and
k ∞ a
V=
b n =0
∑
Pn (cos φ)   , a < b
b
(32) Laguerre Polynomials

In terms of the argument cos φ , the Legendre polynomials can be expressed as

trigonometric polynomials of the form shown below
P0 (cos φ) = 1
P1 (cos φ) = cos φ
1 1
P2 (cos φ) = (3 cos 2 φ − 1) = (3 cos 2φ + 1)
2 4
1 1
P3 (cos φ) = (5 cos 3 φ − 3 cos φ) = (5 cos 3φ + 3 cos φ)
2 8
Let us now take up the following examples.
Example 2: Show that
(a) Pn ( − x ) = (−1) n Pn ( x )
(b) Pn (1) = 1
(c) Pn (−1) = (−1) n

Solution: From Generating Function for Legendre Polynomials, we get

∞ −1
(a) ∑ P (− x ) h
n =0
n
n
= (1 + 2xh + h 2
)2
−1

[
= 1 − 2x (−h ) + (−h ) 2
] 2

∞
= ∑ P ( x ) (−h )
n =0
n
n

∞
= ∑ (−1)
n =0
n
Pn ( x ) h n

∴Pn ( − x ) = (−1) n Pn ( x )
∞ 1
−
(b) ∑
n =0
Pn (1) h n = (1 − 2h + h 2 ) 2

= (1 − h ) −1
=1+ h + h2 + h3 +L + hn + L
∞
= ∑n =0
hn

∴ Pn (1) = 1 .
∞ 1
−
(c) ∑
n =0
Pn (−1) h n = (1 + 2h + h 2 ) 2

= (1 + h ) −1
∞
= ∑ (−1)
n =0
n
hn

∴ Pn (−1) = (−1) n
***

Example 3: Using the generating function for the Legendre polynomial show that
3x 2 − 1 1
P0 ( x ) = 1, P1 ( x ) = x , P2 ( x ) = , P3 ( x ) = (5x 3 − 3x )
2 2
1
P4 ( x ) = (35x 4 − 30 x 2 + 3).
8 91
Ordinary Differential Solution: From generating function for Legendre polynomials we know that
Equations ∞ 1
−
∑h
n =0
n 2
Pn ( x ) = (1 − 2hx + h ) 2 = [ 1 − h (2 x − h ) ] −1 / 2

1 1 .3 2 1.3.5 3
=1+ h ( 2x − h ) + h (2x − h ) 2 + h (2x − h ) 3
2 2 .4 2 .4 .6
1.3.5.7 4
+ h (2x − h ) 4 + L
2.4.6.8
⇒ P0 ( x ) + h P1 ( x ) + h 2 P2 ( x ) + h 3 P3 ( x ) + h 4 P4 ( x ) + L
1 1 1
= 1 + x.h + (3x 2 − 1)h 2 + (5x 3 − 3x )h 3 + (35x 4 − 30 x + 3)h 4 + L
2 2 8
Equating the coefficients of like powers of h , we get
1 1
P0 ( x ) = 1, P1 ( x ) = x , P2 ( x ) = (3x 2 − 1), P3 ( x ) = (5x 3 − 3x )
2 2
1
P4 ( x ) = (35x 4 − 30x 2 + 3).
8
***
Example 4: Express f ( x ) = x + 3x + 4x 2 − x + 2 in terms of Legendre
4 3

polynomials.
Solution: We have shown in Example 3, that
3x 2 − 1 1
P0 ( x ) = 1, P1 ( x ) = x , P2 ( x ) = , P3 ( x ) = (5x 3 − 3x )
3 2
1
P4 ( x ) = (35x 4 − 30x 2 + 3).
8
The above results can also be obtained by substituting n = 0, 1, 2, 3, and 4 in relation
(19).
From these relations we can write
P 2 2 3 2 3
x 2 = 0 + .P2 ( x ), x 3 = P3 ( x ) + x = P3 ( x ) + P1 ( x ) and
3 3 5 5 5 5
4 8 6 2 3
x = P4 ( x ) + x −
35 7 35
8 4 1
= P4 ( x ) + P2 ( x ) + P0
35 7 5
4 3 2
Thus, x + 3x + 4 x − x + 2
 8 4 P  2 3  2 P 
=  P4 ( x ) + P2 ( x ) + 0  + 3 P3 ( x ) + P1 ( x )  + 4 P2 ( x ) + 0  − P1 ( x ) + 2P0 ( x )
 35 7 5  5 5  3 3 
8 6 68 4 53
= P4 ( x ) + P3 ( x ) + P2 ( x ) + P1 ( x ) + P0 ( x ).
35 5 21 5 15
***
You may now attempt the following exercises.
E4) Prove that
 1  1 1  1 1  1 1
Pn  −  = P0  − P2 n   + P1  −  P2 n −1   + L + P2 n  − P0   .
 2   2    2  2  
2  2 2
E5) Show that
1
a) Pn′ (1) =
n (n + 1)
2
1
b) Pn′ (−1) = (−1) n +1 n (n + 1)
2
E6) Show that
a) P2 n +1 (0) = 0 , n = 0, 1, 2, K
92
 Legendre, Hermite and
(−1) n 2n! Laguerre Polynomials
b) P2 n (0) = , n = 0, 1, 2, K
2 2 n (n!) 2
E7) Verify that
a) P0 (cos φ) = 1
b) P1 (cos φ) = cos φ
1
c) P2 (cos φ) = (3 cos 2φ + 1)
4
1
d) P3 (cos φ) = (5 cos 3φ + 3 cos φ)
8
We now give some relations between Legendre polynomials of different order.
3.2.3 Recurrence Relations
We shall now establish a few useful relations involving Legendre polynomials of
different order or between Legendre polynomials and their derivatives, which are
called recurrence relations.
From Rodrigue’s formula, we have
1 d
Pn +1 ( x ) = n +1 D ( n +1) [ ( x 2 − 1) n +1 ] , where D = (33)
2 (n + 1) ! dx
Differentiating Eqn.(33) w.r. to x , we get
1
Pn′ +1 ( x ) = n +1 D n +1 [ (n + 1) ( x 2 − 1) n . 2x ] (34)
2 (n + 1) !
1
= n D n [ ( x 2 − 1) n + 2nx 2 ( x 2 − 1) n −1 ]
2 (n ) !
1
= n D n [ ( x 2 − 1) n + 2n ( x 2 − 1 + 1) ( x 2 − 1) n −1 ]
2 (n ) !
1 1 n
= n D [ ( 2n + 1) ( x 2 − 1) n + 2n ( x 2 − 1) n −1 ]
2 n!
2n + 1 n 2 1
= n D ( x − 1) n + n −1 D{D n −1 ( x 2 − 1) n −1 }
2 n! 2 ( n − 1) !
= (2n + 1) Pn ( x ) + Pn′ −1 ( x )
Thus Pn′ +1 (x) = (2n + 1) Pn (x) + Pn′ −1 (x) (35)
Now we use Leibnitz rule on the right hand side of Eqn.(34) to obtain
1
Pn′ +1 ( x ) = n [ {D n +1 ( x 2 − 1) n } x + (n + 1){D n ( x 2 − 1) n }.1 ]
2 (n ) !
 1  1
= x. D  n D n { ( x 2 − 1) n }  + (n + 1). n {D n ( x 2 − 1) n }
 2 (n )!  2 (n ) !
= x Pn′ ( x ) + (n + 1)Pn ( x )

⇒ Pn′ +1 (x) = x Pn′ (x) + (n + 1) Pn (x) (36)

Subtracting Eqn.(36) from Eqn.(35), we get

n Pn (x) = xPn′ (x) − Pn′ −1 (x) (37)
Changing (n + 1) to n in relation (35) and using it in relation (37), we get

n Pn ( x ) = x [ (2n − 1) Pn −1 ( x ) + Pn′ − 2 ( x ) ] − Pn′ −1 ( x )

= (2n − 1) x Pn −1 ( x ) + [ x Pn′ − 2 ( x ) − Pn′ −1 ( x ) ]
= (2n − 1) x Pn −1 ( x ) − (n − 1) Pn − 2 ( x ) ( using relation (37))
93
Ordinary Differential We can thus write
Equations
(n + 1) Pn +1 (x) = (2n + 1)x Pn (x) − n Pn −1 (x) (38)
which is a three-term recurrence relation.

From Eqn.(36) and (37) following relations can also be obtained

(1 − x 2 ) Pn′ (x) = n{ Pn −1 (x) − x Pn (x) } (39)
2
and (1 − x ) Pn (x) = (n + 1) {x Pn (x) − Pn +1 (x) } (40)
We are leaving them here for you to do it yourself.
E8) Establish relations (39) and (40).
E9) Derive the identity
(1 − x 2 ) Pn′ ( x ) = (n + 1) [ x Pn ( x ) − Pn +1 ( x ) ] , n = 0, 1, 2, K .

Legendre polynomials belong to an important class of orthogonal polynomials which

we shall discuss now.
3.2.4 Orthogonality Property
We shall now show that the Legendre polynomials Pn ( x ) satisfy the following
orthogonal property
+1 0 , if m ≠ n

∫ Pm ( x ) Pn ( x ) dx =  2
 , if m = n
' (41)
−1  2n + 1
Consider the Legendre equation
d2y dy
(1 − x 2 ) 2 − 2 x + n (n + 1) y = 0.
dx dx
This can be written as
d  dy 
 (1 − x 2 )  + n (n + 1) y = 0 (42)
dx  dx 
Legendre Polynomials Pm ( x ) and Pn ( x ) satisfy Eqn.(42), thus
d
dx
[ ]
(1 − x 2 ) Pn′ = − n (n + 1) Pn (43)

and
d
dx
[ ]
(1 − x 2 ) Pm′ = − m(m + 1) Pm (44)
Multiplying Eqn.(43) by Pm and Eqn.(44) by Pn and substracting the resulting
expressions, we get
d
dx
[ ]
(1 − x 2 ) (Pm Pn′ − Pn Pm′ ) = [m(m + 1) − n (n + 1)] Pn Pm
Integrating both sides w.r. to x from x = −1 to x = 1 , we get
1
+1
(1 − x 2 ) (Pm Pn′ − Pn Pm′ )
−1 ∫
= [m (m + 1) − n (n + 1)] Pn ( x ) Pm ( x ) dx
−1
The left hand term vanishes at x = ±1
+1
∴ ∫
(n + m + 1) (m − n ) Pm ( x ) Pn ( x )dx = 0
−1
Since m and n are non-negative integers, ∴ n + m + 1 ≠ 0 . Thus if m ≠ n , i.e.,
m − n ≠ 0 , then from above
+1

∫P
−1
m ( x ) Pn ( x )dx = 0, for m ≠ n.

To prove the property for m = n , we use Rodrigue’s formula

94
 Legendre, Hermite and
1 dn
Pn ( x ) = { ( x 2 − 1) n } Laguerre Polynomials
(n )!2 n dx n
Let f ( x ) be any function with at least n continuous derivatives on the interval
− 1 ≤ x ≤ 1 and consider the integral
1

∫
I = f ( x ) Pn ( x ) dx.
−1
Using Rodrigue’s formula
1
1 dn
I= n
2 n! −1
f ∫
( x )
dx n
( x 2 − 1) n dx

Integrating by parts, we get

1
1  d n −1 ( x 2 − 1) n 
1
1 d n −1 ( x 2 − 1) n
I = n f ( x )
2 n!  dx n −1
 − n
 −1 2 n! −1
f ′( x )
dx n −1 ∫ dx

The first expression in the above equation on the right hand side, vanishes at both
limits, so
1
1 d n −1 ( x 2 − 1) n
I=−
2 n n! −1∫f ′( x )
dx n −1
dx

and continuing to integrate by parts, we obtain

1
(−1) n
∫f
(n)
I= ( x ) ( x 2 − 1) n dx.
2 n n! −1
2n!
Now if we put f ( x ) = Pn ( x ) in the above integral then since Pn( n ) ( x ) = , it
2 n n!
follows that
1 1
2n!
∫
I = Pn2 ( x ) dx = ∫ (1 − x
2 n
) dx
−1
2 ( n!) 2
2n
−1
1
2(2n )!
∫ (1 − x
2 n
= ) dx
2 2 n (n!) 2 0
Using the substitution x = sin θ , and recalling the reduction formula,
1 2n
∫
cos 2 n +1 θ dθ =
2n + 1
cos 2 n θ sin θ +
2n + 1
cos 2 n −1 θ dθ . ∫
Then the definite integral in I becomes
1 π/2 π/2
2n
∫ ∫ cos 2 n +1 θ dθ = ∫ cos
2 n −1
(1 − x 2 ) n dx = θdθ
0 0
2n + 1 0
π/2
2n 2n − 2 2
= L
2n + 1 2 n − 1 3 ∫ cos θ dθ
0

2 n! n
2 2 n (n! ) 2
= =
1.3 K (2n − 1) (2n + 1) (2n )! (2n + 1)
Thus, we conclude that
1
2(2n )!
∫ (1 − x
2 n
I= 2n 2
) dx
2 (n! ) 0

2(2n )! 2 2 n (n!) 2 2
= 2n 2
=
2 (n!) ( 2 n )! ( 2 n + 1) 2 n +1
1
2
∫P
2
or, n ( x ) dx = (45)
−1
2n + 1
95
Ordinary Differential Example 5: Assuming the orthogonality property of Legendre polynomials, show
Equations that
1 0 , if m ≠ n

∫
(1 − x 2 )Pn(1) ( x ) Pm(1) ( x ) dx =  2n (n + 1)
 2n + 1 , if m = n
−1
where m and n are integers.
1
Solution: L.H.S. = ∫ (1 − x 2 ) Pn(1) ( x ) Pm(1) ( x ) dx
−1
1
1
d 
= (1 − x 2
) Pm(1) ( x ). Pn ( x ) − ∫ Pn ( x )  { }
(1 − x 2 ) Pm(1) ( x )  dx
−1
−1
 dx 
Since Pm ( x ) is a solution of Legendre’s equation, thus
d
dx
[ ]
(1 − x 2 ) Pm(1) ( x ) = −m (m + 1) Pm ( x )

1 0 , if n ≠ m

∫
1
∴ L.H.S. = m(m + 1) Pn ( x ) Pm ( x ) dx = 
−1 

∫
n (n + 1) Pn2 ( x ) dx , if n = m
−1

0 , if n ≠ m

=  2n (n + 1) [ using result (41)]
 2n + 1 , if n = m

***
Many problems of potential theory depend on the possibility of expanding a given
function in a series of Legendre polynomials. It is easy to see that this can always be
done when the given function is itself a polynomial. For example, any third-degree
polynomials p( x ) = b 0 + b1 x + b 2 x 2 + b 3 x 3 can be written as
1 2  3 2 
p( x ) = b 0 P0 ( x ) + b1 P1 ( x ) + b 2  P0 ( x ) + P2 ( x ) + b 3  P1 ( x ) + P3 ( x ) (see Example 4)
3 3  5 3 
 b   3b  2b 2b
=  b 0 + 2  P0 ( x ) +  b1 + 3  P1 ( x ) + 2 P2 ( x ) + 3 P3 ( x )
 3   5  3 5
3
= ∑
n =0
a n Pn ( x )

More generally, since Pn ( x ) is a polynomial of degree n for every positive integer

n , it can easily be seen that x n can always be expressed as a linear combination of
P0 ( x ), P1 ( x ), K , Pn ( x ) , so any polynomial p( x ) of degree k has an expansion of the
form
k
p( x ) = ∑
n =0
a n Pn ( x ) (46)

But when the given function is not a polynomial then it may not be easy to expand an
‘arbitrary’ function f ( x ) in series of the form:
∞
f (x) = ∑
n =0
a n Pn ( x ) (47)

We then need a new procedure for calculating the coefficients a n in Eqn.(47), and the
key lies in formula (41). We multiply Eqn.(47) by Pm ( x ) and integrate term by term
from − 1 to 1 and obtain
1 ∞ 1

∫ f ( x ) Pm ( x ) dx = ∑
n =0
an ∫ Pm ( x ) Pn ( x ) dx
−1 −1

Using formula (41), it reduces to

96
 1 Legendre, Hermite and
2a m
∫ f ( x ) Pm ( x ) dx =
2m + 1
Laguerre Polynomials

−1
Coefficients a n in Eqn.(47) are then obtained as
1
 1
an = n +


2 ∫ f ( x ) Pn ( x ) dx (48)
−1
The above manipulations are easy to justify if f ( x ) is known in advance to have a
series expansion of the form (47) and this series is integrable term by term on the
interval − 1 ≤ x ≤ 1 . Both the conditions are obviously satisfied when f ( x ) is a
polynomial. However, we shall not discuss here the conditions under which the
representation of form (47) and (48) is valid for any other type of functions. For now
it suffices to say that for certain functions the series (47) will converge throughout the
interval − 1 ≤ x ≤ 1 , even at points of finite discontinuities of the given function.
Series of this type are called Legendre series.
In practice, the evaluation of integrals like (48) is performed numerically. However, if
the function f is not too complicated, we can sometimes use various properties of
Legendre polynomials to evaluate such integrals in closed form. We illustrate this
through the following example.
Example 6: Find the Legendre series for
− 1 , − 1 ≤ x < 0
f (x) =  .
1 , 0 < x ≤ 1
Solution: The given function f is an odd function. Hence f ( x ) Pn ( x ) is an odd
function when n is even, and in this case
1
 1
 2 ∫
a n =  n +  f ( x ) Pn ( x ) dx = 0 , n = 0, 2, 4, K
−1
For n odd, the product f ( x ) Pn ( x ) is an even function, and therefore
1
 1
an = n + 
 2 ∫ f ( x ) Pn ( x ) dx
−1
1 1

∫ ∫
= (2n + 1) f ( x ) Pn ( x ) dx = (2n + 1) Pn ( x ) dx , n = 1, 3, 5, K
0 0
Now let us use the identity (35)
1
Pn ( x ) = [Pn′ +1 (x ) − Pn′ −1 ( x )]
2n + 1

and set n = 2k + 1 for k = 0, 1, 2, K

1
a 2 k +1 = (4k + 3) ∫ P2 k +1 ( x ) dx
0
1

∫
= [ P2′ k + 2 ( x ) − P2′ k ( x )] dx
0
1
= [P2 k + 2 ( x ) − P2 k ( x )]
0
= P2 k (0) − P2 k + 2 (0) [Q Pn (1) = 1 for all n ]
Using the result of E6), we have

(−1) k (2k ) ! (−1) k +1 (2k + 2) !

a 2 k +1 = −
2 2 k (k!) 2 2 2 k + 2 [(k + 1) !]
2

97
Ordinary Differential
(−1) k (2k )!  (2k + 2) (2k + 1) 
Equations = 1 + 
2 2 k (k!) 2  2 2 (k + 1) 2 
(−1) k (2k )!  2k + 1  (−1) k (2k )!(4k + 3)
= 1 + =
2 2 k (k!) 2  2k + 2  2 2 k +1 k!(k + 1)!
and thus
∞
(−1) k (2k )!(4k + 3)
f (x) = ∑
k =0 2 2 k +1 k!(k + 1)!
P2 k +1 ( x ) , − 1 ≤ x ≤ 1 .

***
You may try a few exercises now.
E10) If f ( x ) is a polynomial of degree less then n , prove that
1

∫ f ( x ) Pn ( x ) dx = 0 .
−1

E11) Express f ( x ) = x 3 + a x 2 + b x + c as a linear combination of Legendre

polynomials.
E12) Show that (2n + 1) ( x 2 − 1) Pn′ = n (n + 1) (Pn +1 − Pn −1 ) .
1
Pn ( x ) 2t n
E13) Prove that ∫ 1 − 2x t + t 2
dx =
2n + 1
.
−1

1 1  θ θ
E14) Prove that 1 + P1 (cos θ) + P2 (cos θ) + L = ln 1 + sin  sin  . /
2 3  2 2
E15) Make the change of variable x = cos φ in the DE
1 d  dy 
 sin φ  + n (n + 1) y = 0
sin φ dφ  dφ 
and show that it reduces to Legendre’s Eqn.(1).
E16) Determine the values of n for which y = Pn ( x ) is a solution of
a) (1 − x 2 ) y ′′ − 2 xy ′ + n (n + 1) y = 0, y(0) = 0, y(1) = 1
b) (1 − x 2 ) y ′′ − 2 xy ′ + n (n + 1) y = 0, y ′(0) = 0, y(1) = 1 .

3.2.5 Legendre Function of the Second Kind

We have shown that for positive integral values of n the Legendre polynomials Pn ( x )
represents one of the linearly independent solutions of Legendre’s Eqn.(1), viz.,
(1 − x 2 ) y ′′ − 2 xy ′ + n (n + 1) y = 0 .
Because the equation is of second order, we know from the theory of differential
equations that there exists a second linearly independent solution Q n ( x ) (say) , such
that
y = c1 Pn ( x ) + c 2 Q n ( x ) (49)
where c1 and c 2 are arbitrary constants, is a general solution of Eqn.(1).
Also from the theory of second-order linear differential equations it is well known that
if y1 ( x ) is a non-trivial solution of
y ′′ + a ( x ) y ′ + b( x ) y = 0 (50)
then a second linearly independent solution can be defined by

∫
exp[− a ( x ) dx]
y 2 ( x ) = y1 ( x ) ∫ y12 ( x )
dx [ref. Unit 7, MTE-08]. (51)
98
Hence, if we express Eqn.(1) in the form Legendre, Hermite and
Laguerre Polynomials
2x n (n + 1)
y ′′ − 2
y′ + y=0
1− x (1 − x 2 )
and let y1 ( x ) = Pn ( x ) , it follows that
dx
y 2 ( x ) = Pn ( x ) ∫ 2
(1 − x ) [Pn ( x )]
2
(52)

is a second solution, linearly independent of Pn ( x ) . Because any linear combination

of solutions is also a solution of a homogeneous differential equation, we defined the
second solution of Eqn.(1) as follows:
 dx 
Q n ( x ) = Pn ( x ) A n + B n

∫ 2 2 
(1 − x ) [Pn ( x )] 
, (53)

where A n and B n are constants to be chosen for each n . We refer to Q n ( x ) as the

Legendre function of the second kind of integral order.
Accordingly, when n = 0 , we choose A 0 = 0 and B 0 = 1 , and hence
dx
Q 0 (x) = ∫
1− x2
which leads to
1 1+ x
Q 0 ( x ) = ln , | x | <1 (54)
2 1− x
For n = 1 , we set A 1 = 0 and B1 = 1 , from which we obtain
dx
Q1 ( x ) = x ∫
(1 − x 2 ) x 2
 1 1 
=x ∫ 
1 − x
2
+ 2  dx
x 
1 1+ x
= x ln −1 (55)
2 1− x
or, Q1 ( x ) = x Q 0 ( x ) − 1 , | x | < 1 (56)
If we continue in this fashion then we will have to evaluate more difficult integrals.
Since it is assumed that all solutions of Legendre’s equation satisfy the recurrence
formulas for Pn ( x ) , hence we select the Legendre functions Q n ( x ) so that,
2n + 1 n
Q n +1 ( x ) = x Q n (x ) − Q n −1 ( x ) (57)
n +1 n +1
for n = 1, 2, 3, K with Q 0 ( x ) and Q1 ( x ) already defined. The substitution n = 1
into Eqn.(57) yields
3 1
Q 2 ( x ) = x Q1 ( x ) − Q 0 ( x )
2 2
1 3
= (3x 2 − 1) Q 0 ( x ) − x
2 2
3  1 2 
or, Q 2 ( x ) = P2 ( x ) Q 0 ( x ) − x , | x | < 1 Q P2 ( x ) = (3x − 1)  (58)
2  2 
For n = 2 , we find
5 2
Q 3 ( x ) = P3 ( x ) Q 0 ( x ) − x 2 + , | x | < 1 (59)
2 3
whereas in general, we state the result
[( n −1) / 2 ]
(2n − 4k − 1)
Q n ( x ) = Pn ( x ) Q 0 ( x ) − ∑ k =0 (2k + 1) (n − k )
Pn − 2 k −1 ( x ) , | x | < 1 (60)

for n = 1,2, 3, K .
Because of the logarithmic term in Q 0 ( x ), Q n ( x ) has infinite discontinuities at
x = ± 1 . However, within the interval − 1 < x < 1 , these functions are well defined.
99
Ordinary Differential Without giving the details, we state that the Legendre functions Q n ( x ) satisfy all
Equations
recurrence relations given in Sec.3.2.3 for Pn ( x ) . In addition, there are several
relations that directly involve both Pn ( x ) and Q n ( x ) . For example, if | t | < | x | , then
∞
1
=
x − t n =0
∑ (2n + 1) Pn ( t ) Q n ( x ) (61)

From this result, it can be easily shown that

1
1 Pn ( t )
Q n (x) =
2 ∫x−t
dt , n = 0, 1, 2, K (62)
−1
which is called the Neumann formula.
You may now try the following exercises.
E17) Verify result (62).
E18) Find the general solution of the following differential equations in terms of
Pn ( x ) and Q n ( x )
a) (1 − x 2 ) y ′′ − 2xy ′ = 0
b) (1 − x 2 ) y ′′ − 2xy ′ + 12 y = 0
c) (1 − x 2 ) y ′′ − 2 xy ′ + 2 y = 0
d) (1 − x 2 ) y ′′ − 2xy ′ + 30 y = 0 .
1 1+ x
E19) Given P0 ( x ) = 1 and Q 0 ( x ) = ln , find W [P0 ( x ), Q 0 ( x )] where , W is
2 1− x2
the wronskian.
E20) Using Eqn.(53), show that the wronskian of Pn ( x ) and Q n ( x ) is given by
Bn
W (Pn ( x ) , Q n ( x ) ) = , n = 0, 1, 2, K .
1− x2
Other polynomials which commonly occur in applications are the Hermite and
Laguerre polynomials which we shall be discussing now.

3.3 HERMITE AND LAGUERRE POLYNOMIALS

Let us first consider the Hermite polynomials.
Hermite Polynomials
The Hermite polynomials play an important role in problems involving Laplace’s
equations in cylinderical coordinates, in various problems in quantum mechanics and
in probability theory.
The Hermite polynomials are polynomial solutions to Hermite’s equation
y ′′ − 2 xy ′ + 2n y = 0 (63)
where n is a constant.
Using the power series method it can be easily shown that y( x ) = a 0 y 1 ( x ) + a 1 y 2 ( x )
is the general solution of Eqn.(63), where
2n 2 (n − 2) 4 2 3 n (n − 2) (n − 4) 6
y1 ( x ) = 1 − x + 22 n x − x +L (64)
2! 4! 6!
and
2(n − 1) 3 2 2 (n − 1) (n − 3) 5 2 3 (n − 1) (n − 3) (n − 5) 7
y 2 (x ) = x − x + x − x + L (65)
3! 5! 7!
and both the series converge for all x . We shall not be going into the details of the
series solution here and leave it for you to verify it yourself.
100
If n is a non-negative integer, then one of these series terminates and is thus a Legendre, Hermite and
Laguerre Polynomials
polynomial- y1 ( x ) if n is even, and y 2 ( x ) if n is odd, while the other remains an
infinite series. It can be easily verified that for n = 0, 1, 2, 3, 4, 5, these polynomials
are
2 4 4 4
1, x , 1 − 2x 2 , x − x 3 , 1 − 4x 2 + x 4 , x − x 3 + x 5 , respectively.
3 3 3 15
The polynomial solutions of Hermite’s Eqn.(63) are constant multiples of these
polynomials. The constant multiples with the property that the terms containing the
highest powers of x are of the form 2 n x n are denoted by H n ( x ) and called the
Hermite polynomials.
Let us now define the Hermite polynomials
Generating Function
We define the Hermite polynomials H n ( x ) by means of the relation
∞
tn
∑
2
e (2xt −t ) = , for all finite x and t .
H n (x) (66)
n=0 n!
The function on the left hand side is called the generating function of the Hermite
polynomials. This definition of Hermite polynomials is used often in statistical
applications.
We have
 ∞ (2 xt ) m   ∞ (− t 2 ) k 
∑ ∑
2 2
e ( 2 xt − t
= e 2 xt .e − t = 
)
 
m =0 m!   k = 0 k! 
∞ [ n / 2]
(−1) k (2 x ) n − 2 k n
= ∑ ∑
n =0 k = 0 k! (n − 2k ) !
t (67)

where the last step follows from the index change m = n − 2k and [n / 2] is the
standard notation to the greatest integer ≤ n / 2 . From Eqns.(66) and (67) it follows
that
[n/2]
( −1) k n!
H n (x) =
k =0 k! (n
∑ − 2k) !
(2x)n − 2 k (68)

n
, if n is even
n   2
where   = 
 2   n − 1 , if n is odd
 2
Eqn.(68) shows that H n ( x ) is a polynomials of degree n in x and that

H n (x ) = 2 n x n + π n −2 ( x ) ,
where π n − 2 ( x ) is a polynomial of degree (n − 2) in x . Further, it follows that it is an
even function of x for even n and an odd function of x for odd n . Thus, it follows
that
H n ( −x) = ( −1) n H n (x) (69)
The first few Hermite polynomials are listed in Table-1.
Table 1
H 0 (x) = 1
H1 (x ) = 2x
H 2 (x ) = 4x 2 − 2
H 3 ( x ) = 8x 3 − 12
H 4 ( x ) = 16x 4 − 48x 2 + 12
H 5 ( x ) = 32 x 5 − 160x 3 + 120x
101
Ordinary Differential In addition to series (68), the Hermite polynomials can be defined by the Rodrigues
Equations formula also.
The Rodrigue’s Formula
2 dn 2
We prove the formula H n ( x ) = (−1) n e x n
(e − x ) (70)
dx
We have the relation
∞
H n (x) n
∑
2 2
−( t − x )2
e 2 tx − t = e x =
t
n =0 n!
If we examine this relation in the light of Maclaurin’s theorem, we see that
∂ n 2 tx − t 2 x2 ∂
n
2

n
( e ) = e n
(e − ( t − x ) ) = H n (x )
∂t t =0 ∂t t =0

∂n −( t − x )2 2
or, [ e ] = e −x H n (x)
∂t n t =0
∂ ∂
If we introduce a new variable z = x − t and use the fact that =− , then since
∂t ∂z
t = 0 corresponds to z = x , the above expression reduces to
2
dn 2 d n (e − x ) 2
(−1) n
n
(e − z ) = (−1) n
n
= e −x H n (x)
dz z=x dx
2 d n −x2
∴H n (x) = ( −1) n e x (e )
dx n
which proves the Rodrigue’s formula (70).
Hermite polynomials satisfy a number of properties, we list some of these here
without proving them. You can verify them yourself.
Recurrence Relations
Using the generating function for Hermite polynomials it can be shown that
x H ′n (x) = n H ′n −1 (x) + n H n (x) (71)
The relation
∞
H n (x)t n
∑
2
e ( 2 xt − t )
= (72)
n =0 n!
on differentiating with respect to x , yields
∞
H ′n ( x ) t n
∑
2
2 t e ( 2 xt − t ) = (73)
n =0 n!
From Eqns.(72) and (73), we get
∞
2H n ( x ) t n +1 ∞
H ′n ( x ) t n
∑
n =0 n!
=
n =0 n!
∑
which with a shift of index on the left, yields H ′0 (x) = 0 , and for n ≥ 1 ,
H ′n (x) = 2n H n −1 (x) (74)
Combining relations (71) and (74), we obtain
H n (x) = 2x H n −1 (x) − H ′n −1 (x) (75)
Further, using relations (74) and (75), we obtain the relations
H n (x) = 2x H n −1 (x) − 2 (n − 1) H n − 2 (x) (76)
and Hermite’s differential equation
H ′n′ (x) − 2x H ′n (x) + 2n H n (x) = 0 (77)

102 Relation (77) can be used to prove the orthogonality property of Hermite polynomials.
Orthogonality Legendre, Hermite and
Laguerre Polynomials
The Hermite polynomials H n ( x ) satisfy the following orthogonality property
∞
0 , if m ≠ n
∫e
−x 2
H n (x) H m (x) dx =  n (78)
−∞ 2 n ! π , if m = n
Eqn.(77) can be written in the form

[e − x H ′n ( x ) ]′ + 2n e − x H n ( x ) = 0
2 2
(79)

Similarly, for a polynomial H m ( x ) , we can write

[e − x H ′m ( x )]′ + 2m e − x H m ( x ) = 0
2 2
(80)
Multiplying Eqn.(79) by H m ( x ) and Eqn.(80) by H n ( x ) and adding the two, we get
2(n − m) e − x H ( x )H ( x ) = H ( x )[e − x H ′ ( x )]′ − H ( x )[e − x H ′ ( x )]′
2 2 2
n m n m m n
−x2
= [e {H n (x ) H ′m ( x ) − H ′n ( x ) H m (x )}]′
Therefore, it follows that
∞
2 2

∫
2(n − m) e − x H n ( x ) H m ( x ) dx = [e − x {H n ( x ) H ′m ( x ) − H ′n ( x ) H m ( x ) } ]∞− ∞
−∞
(81)

2
Since the product of any polynomial in x by e − x → 0 as x → ∞ or as x → −∞ , we
conclude that
∞
−x 2
∫e
−∞
H n ( x ) H m ( x ) dx = 0, m ≠ n (82)

This proves the first part of formula, that is, the Hermite polynomials form an
2
orthogonal set over the interval ] − ∞, ∞ [ with weight function e − x .
To establish the second part of (78), we use Rodrigue’s formula (70) and integrate
∞ ∞ 2
2 d n e −x
∫
−∞
e − x H n ( x ) H n ( x ) dx = (−1) n ∫
−∞
H n (x)
dx n
dx

by parts with u = H n ( x ), du = H ′n ( x ) dx
dn −x 2 d n −1 − x 2
dv = ( e ) dx , v = (e )
dx n dx n −1
2
Since uv is the product of e − x and a polynomial, it vanishes at both limits and
∞ ∞
2 d n −1 2

∫
−∞
∫
e − x [H n ( x ) ] 2 dx = (−1) n +1 H ′n ( x )
−∞
dx n −1
(e − x ) dx

∞
d n−2 2

∫
n+2
= (−1) H ′n′ ( x ) n −2
(e − x ) dx
−∞
dx
∞
2
= LL = (−1) 2 n ∫H e − x dx
(n)
n (x)
−∞

Now the term containing the highest power of x in H n ( x ) is 2 n x n , so H (nn ) ( x ) = 2 n n!

and the last integral is
∞ ∞
2 2

∫ ∫
2 n n! e − x dx = 2 n n! 2 e − x dx = 2 n n! π
−∞ 0

which gives the desired result

∞
−x 2
∫e
−∞
H 2n ( x ) dx = 2 n n! π

103
Ordinary Differential These orthogonality properties can be used to expand an arbitrary function f ( x ) in a
Equations
series
∞
f(x) = ∑
n=0
a n H n (x) (83)
2
The coefficients a n can be found by multiplying Eqn.(83) by e − x H m ( x ) and
integrating term by term from − ∞ to ∞ . Using formula (78), this gives
∞ ∞ ∞

∫e
−x 2
H m ( x ) f ( x ) dx = ∑
n =0
an ∫e
−x2
H m ( x ) H n ( x ) dx = a m 2 m m! π
−∞ −∞

Replacing m by n , in the above equation, we get

∞
1
∫e
−x2
an = n
H n (x) f(x) dx (84)
2 n! π − ∞
The series (83), with its coefficients calculated by formula (84), is called the Hermite
series and is useful in the theory of orthogonal functions.
Let us consider the following example.
Example 7: Express f ( x ) = e 2 bx in a Hermite series, and use this result to deduce the
value of the integral
∞
2

∫ e −x + 2 bx
H n ( x ) dx .
−∞
Solution: Set t = b in the generating function (66) to obtain
∞
bn
∑
2
e 2 bx − b = H n (x )
n =0 n!
We can then write the series
∞ ∞
bn b n b2
∑ ∑
2
f ( x ) = e 2 bx = e b H n (x) = a n H n ( x ) , where, a n = e
n =0 n1 n =0 n!
Also, from formula (84), we obtain
∞
1 2

∫ e −x + 2 bx
an = n
H n ( x ) dx, n = 0,1, 2, K
2 n! π −∞

Thus, it follows that

∞ 2
− x 2 + 2 bx n! b n e b 2

∫e
n
H n ( x ) dx = π 2 = π ( 2b) n e b , n = 0, 1, 2, K
−∞
n!
***

You may now try the following exercises.

E21) Verify that y1 ( x ) and y 2 ( x ) given by Eqns.(64) and (65) respectively, are two
linearly independent solutions of Hermite’s Eqn.(63).
E22) Use the generating function for the Hermite polynomials to find
H 0 ( x ), H 1 ( x ), H 2 ( x ) and H 3 ( x ) .

E23) Prove the following relations for Hermite polynomials H n ( x ) .

a) x H ′n ( x ) = n H ′n −1 ( x ) + n H n ( x )
b) H n ( x ) = 2x H n −1 ( x ) − 2(n − 1) H n − 2 ( x )
c) H ′n′ ( x ) − 2x H ′n ( x ) + 2n H n ( x ) = 0
104
E24) Show that Legendre, Hermite and
Laguerre Polynomials
2n!
a) H 2 n (0) = (−1) n b) H 2 n +1 (0) = 0
n!
( 2 n + 2) !
c) H ′2 n (0) = 0 d) H ′2 n +1 (0) = (−1) n
(n + 1) !
E25) Use the generating function (66) to derive the relation
[ n / 2]
n !H n −2k ( x )
xn = ∑ n
k = 0 2 k! ( n − 2 k ) !
, n = 0, 1, 2, K

∞
E26) Expand f ( x ) = x 3 − 3x 2 + 2x in a series of the form ∑
n =0
a n H n (x) .

Let us now consider the Laguerre polynomials.

Laguerre Polynomials
The Laguerre polynomials L n ( x ) are polynomial solutions to Laguerre’s equation
xy ′′ + (1 − x ) y ′ + ny = 0 (85)
where n is a constant. These polynomial solutions are obtained when n is a non-
negative integer. An important application involving Laguerre polnomials in
quantum mechanics is to find the wave function associated with the electron in a
hydrogen atom.
We now define the Laguerre polynomials and list some of their useful properties.
We shall not be proving these properties here. We shall leave them for you to verify
yourself.
Generating Function
We define the Laguerre polynomials L n ( x ) by means of the relations
− xt ∞
(1 − t) −1 e 1− t = ∑L
n =0
n (x) t
n
, | t |< 1, 0 ≤ x ≤ ∞ (86)

The function on the left hand side is called the generating function of the Laguerre
polynomials.
We have
− xt ∞
(−1) k
(1 − t ) −1 e 1− t = ∑
k =0 k!
( xt ) k (1 − t ) − k −1

∞
(−1) k ∞
 − k − 1
= ∑
k =0 k!
( xt ) k ∑ 
m=0  m 
 ( −1) m t m (87)

 − k − 1 (− k − 1) (− k − 2) (− k − 3) K (− k − m)
But since   =
 m  m!
(k + 1) (k + 2) K (k + m ) k + m
= (−1) m = (−1) m   ,
m!  m 
Eqn.(87) becomes
− xt ∞ ∞
(−1) k (k + m)! x k k + m
−1 1− t
(1 − t ) e = ∑ ∑
m = 0 k =0 (k!) 2 m!
t (88)

where we have reversed the order of summation.

Using the change of index m = n − k in Eqn.(88) and comparing with Eqn.(86), the
Laguerre polynomials are defined by
105
Ordinary Differential ∞
( −1) k n! x k
Equations L n (x) = ∑
k =0 (k! ) 2 (n − k)!
(89)

In Table-2, we have listed the first few polynomials L n ( x )

Table 2

L 0 (x) = 1
L1 ( x ) = − x + 1
1 2
L 2 (x) = ( x − 4 x + 2)
2!
1
L 3 ( x ) = (− x 3 + 9 x 2 − 18x + 6)
3!
1
L 4 ( x ) = ( x 4 − 16 x 3 + 72x 2 − 96 x + 24)
4!

The Rodrigue’s Formula

The Rodrigue’s formula for Laguerre polynomials L n ( x ) is given by
ex dn
L n (x) = n
(x n e − x ) , n = 0, 1, 2, K (90)
n! dx
which can be verified by the application of the Leibniz formula
dn n
 n  d n −k f d k g
dx n
(fg) = ∑ k =0
 
 k  dx
n −k
dx k
, n = 1, 2, 3, K (91)

Recurrence Relations
We now list a few important properties satisfied by Laguerre polynomials
(n + 1) L n +1 (x) + (x − 1 − 2n) L n (x) + n L n −1 (x) = 0 (92)
L ′n (x) − L ′n −1 (x) + L n −1 (x) = 0 (93)
x L ′n (x) = nL n (x) − n L n −1 (x) (94)
and x L ′n′ (x) + (1 − x) L ′n (x) + n L n (x) = 0 (95)
Relation (95) can be used to prove the following orthogonality property of Laguerre
polynomials.
Orthogonality
The Laguerre polynomials L n ( x ) satisfy the orthogonality property
∞

∫
0
e − x L n (x) L m (x) dx = 0, if m ≠ n (96)

You may now try the following exercises.

E27) Determine the Laguerre polynomials L 0 ( x ), L1 ( x ), L 2 ( x ) and L 3 ( x ) .

E28) Prove the recurrence relations (92) – (95).
∞

∫e
−x
E29) Prove that L n ( x ) L m ( x ) dx = 0 , if m ≠ n .
0

We shall now discuss some applications of these polynomials to physical situations.

3.4 APPLICATIONS TO PHYSICAL SITUATIONS

We shall first illustrate the role played by Legendre polynomials in solving certain
boundary value problems of mathematical physics.
106
Steady-state temperature in a sphere Legendre, Hermite and
Laguerre Polynomials
Consider a solid sphere of radius a . Let origin be its centre. Let the surface of the
sphere be held at a prescribed temperature say, g (φ) , which is assumed to be
independent of coordinate θ , until within the sphere a steady state for temperature
T (ρ, φ) is produced by the flow of heat. Here (ρ, θ, φ) are the spherical polar
coordinates and we wish to find the steady-state temperature T (ρ, φ) which satisfies
Laplace equation.
You may recall that Laplace equation for any function T ( x , y, z) is given by
∂ 2T ∂ 2T ∂ 2T
+ +
= 0.
∂x 2 ∂y 2 ∂z 2
and in the spherical polar coordinates (ρ, θ, φ) where ,
x = ρ sin φ cos θ, y = ρ sin φ sin θ, z = ρ cos φ , it reduces to

1 ∂  2 ∂T  1 ∂  ∂T  1 ∂ 2T
 ρ  + 2  sin φ  + 2 =0 (97)
ρ2 ∂ρ  ∂ρ  ρ sin φ ∂φ  ∂φ  ρ sin 2 φ ∂θ 2

In the given situation since T does not depend on θ , Eqn.(97) reduces to

∂  2 ∂T  1 ∂  ∂T 
 ρ  +  sin φ  = 0 (98)
∂ρ  ∂ρ  sin φ ∂φ  ∂φ 
We have to find the solution of Eqn(98) subject to the boundary condition
T (a , φ) = g (φ) (99)
To solve Eqn.(98) , we use the method of separation of variable and take the product
solution of the form
T (ρ, φ) = R (ρ) P(φ) (100)
Using Eqn.(100), Eqn.(98) becomes
d  2 dR  1 d  dP 
P  ρ  + R  sin φ  = 0
dρ  dρ  sin φ dφ  dφ 

1 d  2 dR  1 1 d  dP 
⇒  ρ  = −  sin φ  (101)
R dρ  dρ  P sin φ dφ  dφ 
The left side of Eqn.(101) is a function of ρ only whereas right hand side of Eqn.(101)
is a function of φ only. This is possible only when each side is equal to the same
λ

constant, say . Then Eqn.(101) is equivalent to the equations.

d 2R dR
λ

ρ2 2
+ 2ρ − R =0 (102)
dρ dρ
1 d  dP 
λ

and,  sin φ  + P = 0 (103)

sin φ dφ  dφ 
Eqn.(102) is an Euler equation, its auxiliary equation is
λ

λ λ

m (m − 1) + 2m − = 0, or, m 2 + m − = 0.
−1± 1+ 4
⇒m= .
2
General solution of Eqn.(102) can be written as
1 4

1 4

−1 −1
λ

+ + − +
R = c1 ρ 2 + c 2ρ 2 (104)
107
Ordinary Differential In order that R is single-valued and bounded near ρ = 0 , we put in Eqn.(104)
Equations
1 1

λ
c 2 = 0 and − + + = n , where n is a non-negative integer. It then follows that
2 4

λ
= n (n + 1) and Eqn.(104) reduces to
R = c 1ρ n (105)
2
cos φ dPd P
and Eqn.(103) becomes 2
+
+ n (n + 1)P = 0 .
dφ sin φ dφ
In this equation if we let x = cos φ, then we get
d 2P dP
(1 − x 2 )
2
− 2x + n (n + 1) P = 0 (106)
dx dx
which is Legendre’s equation.
By the physics of the problem, the function P must be bounded for 0 ≤ φ ≤ π i.e.,
− 1 ≤ x ≤ 1 and we know that, the solution of Eqn.(106) are constant multiples of
Legendre polynomials, Pn ( x ) . If we combine this result with Eqn.(105), then we
have particular solution of Eqn.(98) of the form
α n ρ n Pn (cos φ) , for each n = 0, 1, 2, 3K

where α n are arbitrary constants.

The general solution of Eqn.(98) can be written as
∞
T (ρ, φ) = ∑α
n =0
nρ
n
Pn (cos φ) (107)

The boundary condition (99) now requires

∞
g (φ) = ∑α
n =0
n a n Pn (cos φ)

or equivalently,
∞
g (cos −1 x ) = ∑α n =0
n a n Pn ( x )

Multiplying both sides of the above equation by Pm ( x ) and integrating w.r. to x from
x = −1 to x = 1 and using orthogonality property of Legendre polynomials, we get
P 1
1  1
∫
−1
αn = n
 n +  g(cos x )Pn ( x ) dx (108)
a  2  −1
Hence relation (107) gives the required solution where coefficients are given by
r2 r1 relation (108).
r
Let us look at another application of Legendre polynomials.

θ Electrostatic Dipole Potential

Consider two point charges of equal magnitude, say q , but of opposite sign.
−q a 0 a q
Fig.3 Let these charges be placed in a polar coordinate system (see Fig.3). With suitable
units of measurement, the potential U at the point P is
q q
U= − (109)
r1 r2
where, r1 = r 2 + a 2 − 2a r cos θ

and, r2 = r 2 + a 2 + 2a r cos θ
From generating function relation for the Legendre’s polynomials, we have
∞
(1 − 2 xt + t 2 ) −1 / 2 = ∑t
n =0
n
Pn ( x ) (110)
108
when r > a , we can use Eqn.(110) to write Legendre, Hermite and
n Laguerre Polynomials
1 1 1 1 ∞ a
=
r1 r 1 − 2(a / r ) cos θ + (a / r ) 2
=
r n =0
∑
Pn (cos θ)  
r
(111)

and, similarly,
n
1 1 1 1 ∞ a
=
r2 r 1 − 2(a / r ) cos (−θ) + (a / r ) 2
=
r n =0
∑
Pn (− cos θ)  
r
(112)

From Eqns.(109), (110) and (111), we get

∞ n
q a
U=
n =0r
∑
[ Pn (cos θ) − Pn (− cos θ) ]  
r
(113)

From the properties of Legendre’s polynomial Pn ( x ) , we know that Pn ( x ) is even if

n is even and it is odd if n is odd. Thus [Pn (cos θ) − Pn (− cos θ) ] equals 0
or, 2Pn (cos θ) according as n is even or odd. Thus relation (113) becomes
∞ 2 n +1
2q a
U=
n =0r
∑
P2 n +1 (cos θ)  
r
2q  
3
a a
= P1 (cos θ)   + P3 (cos θ)   + L . (114)
r  r r 
When r is large compared with a , we assume that all terms except the first can be
neglected and since P1 ( x ) = x , relation (114) becomes
 cos θ 
U = 2a q 2 
 r 
Physicists use this approximation for the dipole potential.
We shall now consider the application of Hermite polynomials and understand the role
played by the Hermite polynomials H n ( x ) and the corresponding Hermite function
2
e −x /2
H n ( x ) in the theory of the linear harmonic oscillator in quantum mechanics.
Simple Harmonic Oscillator
In wave mechanics, the basic equation for describing the location of a particle
attracted by an energy potential V (z) is the SchrÖdinger’s (time-independent)
equation
d 2 ψ 8mπ 2
+ [V(z) − E ] ψ = 0 (115)
dz 2 h2
where m is the mass of the particle, E is the total energy, and h is Plank’s constant.
The unknown quantity ψ is called the wave function i.e. the amplitude of the wave
whose intensity gives the probability of finding the particle at any given point in
space. A fundamental problem in wave mechanics concerns the motion of a particle
bounded in a potential well. It has been established that bounded solutions of
Schrodinger’s equation for such problems are obtained only for certain discrete energy
levels of the particle within the well. A particular example of this important class of
problems is the linear oscillator (also called the simple harmonic oscillator), the
solution of which lead to Hermite polynomials.
If the restoring force on a particle displaced a distance z from its equilibrium position
is − kz , where k can be associated with the spring constant of the classical oscillator,
1
then its potential energy is V (z) = kz 2 . Substituting this expression into Eqn.(115)
2
and introducing the dimensionless parameters
1/ 4
 4 m kπ 2  4πE m 4πE
x = z 2

 , λ= =
 h  h k hω
109
Ordinary Differential
where ω = k / m is the angular frequency of the classical oscillator, we find that
Equations
Eqn.(115) becomes
ψ ′′ + (λ − x 2 ) ψ = 0 , − ∞ < x < ∞ (116)
where the primes denote differentiation with respect to x . In addition to Eqn.(116),
the wave function ψ must satisfy the boundary condition
lim ψ ( x ) = 0 (117)
| x |→∞
In order to look for bounded solutions of Eqn.(116), we start with the observation that
λ becomes negligible compared with x 2 for large values of x . Thus, asymptotically
we expect the solution of Eqn.(116) to behave like
2
ψ(x ) ~ e ± x /2
, | x |→ ∞
where only the negative sign in the exponent is appropriate in order that Eqn.(117) be
satisfied. Based on this observation, we make the assumption that Eqn.(116) has
solutions of the form
2
ψ ( x ) = y( x ) e − x /2
(118)
for suitable y . The substitution of Eqn.(118) into Eqn.(116) yields the differential
equation
y ′′ − 2 xy ′ + (λ − 1) y = 0 (119)
The boundary condition (117) suggests that whatever functional form y assumes, it
2
must either be finite for all x or approach infinity at a rate slower than e − x / 2
approaches zero. It is seen that the only solution of Eqn.(119) satisfying this condition
are those for which λ − 1 = 2n , or
λ = λ n = 2n + 1, n = 0,1, 2, K (120)
called eigenvalues . In terms of E , we find
 1
 n +  hω
En = 
2
, n = 0,1, 2, K (121)
2π
with E 0 = hω / 4π being the lowest or minimum energy level. With λ so restricted, we
see that Eqn.(119) becomes
y ′′ − 2 xy ′ + 2ny = 0
which is Hermite’s equation with solution y = H n ( x ) . Hence, we conclude that to
each eigenvalue λ n given by Eqn.(120) there corresponds the solution of Eqn.(116)
(called an eigenfunction) given by
2
ψ n (x ) = H n (x) e − x /2
, n = 0, 1, 2, K (122)
You may now try the following exercises.
2
E30) Show that the functions ψ n ( x ) = e − x / 2 H n ( x ) satisfy the relations.
a) 2n ψ n −1 ( x ) = x ψ n ( x ) + ψ ′n ( x )
b) 2 x ψ n ( x ) − 2n ψ n −1 ( x ) = ψ n +1 ( x )
c) ψ ′n ( x ) = x ψ n ( x ) − ψ n +1 ( x )
2
E31) By writing ψ n ( x ) = c n e − x /2
H n ( x ) , show that the normalized eigen function,
∞

∫ [ψ
2
that is n (x) ] dx = 1 , assumes the form
−∞

ψ n (x) = [ π 2 n!]
n
−1 / 2
e −x
2
/2
H n ( x ), n = 0, 1, 2, K .

We now end this unit by giving a summary of what we have covered in it.
110
 Legendre, Hermite and
3.5 SUMMARY Laguerre Polynomials

In this unit, we have learnt the following points:

1. The equation (1 − x 2 ) y ′′ − 2xy ′ + n (n + 1) y = 0 is called Legendre’s Equation
where n is a real constant. The polynomial solutions of this equation for
integral values of n are Legendre polynomials .
2. Legendre polynomial of degree n , having value 1 at x = 1 , is denoted by
Pn ( x ) and is called Legendre function of the first kind and is given by
[n / 2]
(−1) k (2n − 2k ) !
Pn ( x ) = ∑k =0 2 n
( k ) ! ( n − k ) ! ( n − 2 k ) !
x n −2k

n / 2 , if n is even

where [n / 2] =  n − 1
 2 , if n is odd.
3. Legendre function of the second kind is denoted by Q n ( x ) , when n is an
integer and is given by
 dx
Q n ( x ) = Pn ( x ) A n + B n

∫
(1 − x ) [Pn ( x )] 2
2

where A n and B n are constants.

4. Rodrigue’s formula for Legendre’s polynomials is given by
1 dn
Pn ( x ) = n [ ( x 2 − 1) n ].
2 (n ) ! dx n
5. Orthogonality property of the Legendre’s polynomials Pn ( x ) is
1 0 , if m ≠ n

∫
−1
Pn ( x ) Pm ( x ) dx =  2
 2n + 1 , if m = n
6. The function on the left side of
1
= P0 ( x ) + t P1 ( x ) + t 2 P2 ( x ) + L + t n Pn ( x ) + L | x | < 1
2
1 − 2xt + t
is called the generating function of the Legendre polynomials.
7. Some of the recurrence relations for Legendre polynomials are
Pn′ +1 ( x ) = (2n + 1) Pn ( x ) + Pn′ −1 ( x )
Pn′ +1 ( x ) = x Pn′ ( x ) + (n + 1)Pn ( x )
nPn ( x ) = x Pn′ ( x ) − Pn′ −1 ( x )
(n + 1)Pn +1 ( x ) = (2n + 1) xPn ( x ) − nPn −1 ( x )
(1 − x 2 ) Pn′ ( x ) = n [Pn −1 ( x ) − x Pn ( x ) ]
(1 − x 2 ) Pn ( x ) = (n + 1) [ x Pn ( x ) − Pn +1 ( x ) ]

8. The Hermite polynomials H n ( x ) are polynomial solutions to Hermite’s

equation y ′′ − 2 xy ′ + 2ny = 0 , where n is a constant, and are given by
[n / 2]
(−1) k n!
H n (x) = ∑ k =0 k! (n − 2k )!
(2x ) n − 2k

where [n / 2] is the greatest integer ≤ n / 2 .

∞
tn
∑
2
9. The function on the left side of equation e ( 2 xt − t )
, for all = H n (x )
n =0 n!
finite x and t is the generating function of the Hermite polynomials H n ( x ) .
10. Some of the recurrence relations of Hermite polynomials are
111
Ordinary Differential x H ′n (x) = n H ′n −1 (x) + n H n (x)
Equations
H ′n (x) = 2n H n −1 (x)
.
H n (x) = 2x H n −1 (x) − H n −1 (x)
H n (x) = 2x H n −1 (x) − 2(n − 1) H n − 2 (x)
11. The Rodrigues formula for Hermite polynomials is given by
n
2 d 2
H n ( x ) = (−1) n e x n
(e − x )
dx
12. Orthogonality property of the Hermite polynomials H n ( x ) is
∞
−x2 0 , if m ≠ n
∫e
−∞
H n ( x ) H m ( x ) dx =  n
2 n! π , if m = n
13. The Laguerre polynomials L n ( x ) are polynomial solutions to Laguerre’s
equation x y ′′ + (1 − x ) y ′ + ny = 0 , where n is a constant, and are given by
∞
( −1) k n! x k
L n (x ) = ∑
(k!) 2 (n − k )!
k =0
14. The function on the left hand side of the equation
− xt ∞
(1 − t ) −1 e 1− t = ∑
n =0
L n ( x ) t n , | t | < 1, 0 ≤ x < ∞

is the generating function of the Laguerre polynomials.

15. The Rodrigues formula for the Laguerre polynomials L n ( x ) is
ex dn
L n (x ) = n
(x n e −x )
n! dx
16. Some of the recurrence relations of Laguerre polynomials are
(n + 1) L n +1 (x) + (x − 1 − 2n) L n (x) + n L n −1 (x) = 0
L ′n (x) − L ′n −1 (x) + L n −1 (x) = 0
x L ′n (x) = n L n (x) − n L n −1 (x)
17. Orthogonality property of the Laguerre polynomials L n ( x ) is
∞

∫e
−x
L n ( x ) L m ( x ) dx = 0 , if m ≠ n .
0

3.6 SOLUTIONS/ANSWERS
1
E1) a) Let I = ∫ f (x) P (x) dx
n
−1
1
1 dn
= n
2 n! ∫ f (x )
dx n
[ ( x 2 − 1) ] dx
−1

1  
1 1
d n −1 d n −1
∫
= n f ( x ) n −1 {( x − 1) } − f (1) ( x ) n −1 {( x 2 − 1) n } dx 
2 n

2 n!  dx dx 
 −1 −1 
1
(−1) d n −1
= n
2 n! ∫ f (1) ( x )
dx n −1
{ ( x 2 − 1) n } dx
−1

Proceeding like this

1
(−1) n
∫f
(n)
I= n ( x ). ( x 2 − 1) n dx .
2 n! −1

b) Here f ( x ) = x n , f ( n ) ( x ) = n!
112
 1 1 Legendre, Hermite and
( −1) n
∴ ∫ x n Pn ( x ) dx =
2 n n! ∫ n!. ( x 2 − 1) n dx Laguerre Polynomials

−1 −1
n 1
(−1)
∫ (x
2
= − 1) n dx
2n
−1
2 n +1 (n!) 2
=
(2n + 1)!

E2) By Rodrigue’s formula

1 dn
Pn ( x ) = n n
{( x 2 − 1) n }
2 n! dx
since ( x − 1) vanishes n times at x = 1 and x = −1 , thus its n th derivative
2 n

and hence , Pn ( x ) have all real zeroes lying between − 1 and + 1 .

∴ the roots of Pn ( x ) = 0 are real and lie between − 1 and + 1 .
Let, if possible, the equation
Pn ( x ) = 0
has a repeated root, say ' α' . Then
Pn (α) = 0 and Pn′ (α) = 0 (123)
since Pn ( x ) is a solution of Legendre’s equation, therefore,
(1 − x 2 ) Pn( 2 ) ( x ) − 2xPn(1) + n (n + 1)Pn ( x ) = 0 (124)
Differentiating m times, by Leibnitz’s theorem, we get
(1 − x 2 )Pn( m + 2) ( x ) − 2x (m + 1)Pn( m +1) ( x ) + (n − m) (n − m + 1)Pn( m ) ( x ) = 0, (125)
Putting x = α in Eqn.(124) and using Eqn.(123), we get
Pn( 2) (α) = 0
Putting m = 1, 2, 3, K in Eqn.(125), we get
Pn(3) (α) = 0 = Pn( 4) (α) = Pn(5) (α) = L = Pn( n ) (α )
2n
But from definition, Pn( n ) (α) = n ≠ 0 .
2 n!
Thus Eqn.(123) is not valid and hence Pn ( x ) = 0 cannot have repeated roots.
Hence all the roots of Pn ( x ) = 0 are distinct.

E3) From Rodrigue’s formula, we have

1 dn
Pn ( x ) = n {( x 2 − 1) n }
2 n! dx n
1 1
1 dn
∴ ∫ Pn ( x ) dx = n
2 n! ∫ dx n
( x 2 − 1) n dx
−1 −1
+1
1 d n −1
= n {( x 2 − 1) n }
2 n! dx n −1
−1
+1
n −1
1 d
= n
2 n! dx n −1
[
( x − 1) n ( x + 1) n ]
−1
1
= n [0 − 0] = 0 if n ≠ 0
2 n!

Now when n = 0 , then P0 ( x ) = 1

1 1
+1
∴ ∫ P0 ( x ) dx = ∫ 1.dx = x −1
=2.
−1 −1
113
Ordinary Differential E4) We know that
Equations ∞
(1 − 2xh + h 2 ) −1 / 2 = ∑
n =0
Pn ( x ) h n (126)

1
Putting x = in Eqn.(126), we get
2
∞
1
(1 − h + h 2 ) −1 / 2 =
n =0
∑
Pn   h n
2
(127)

1
Again, putting x = − in Eqn.(126), we get
2
∞
 1
(1 + h + h 2 ) −1 / 2 =
n =0
∑
Pn  −  h n
 2
(128)

Now put h 2 for h in Eqn.(128), we get

∞
 1
(1 + h 2 + h 4 ) −1 / 2 =
n =0
∑
h 2 n Pn  − 
 2
(129)

But (1 + h 2 + h 4 ) −1 / 2 = {(1 + h 2 ) 2 − h 2 }−1 / 2

= (1 + h + h 2 ) −1 / 2 (1 − h + h 2 ) −1 / 2
Using Eqns.(127), (128) and (129) in the above, we get
∞
∞ n  ∞ n 
∑h
n =0
2n
Pn (−1 / 2) =  ∑
 n =0
h Pn (−1 / 2)
 n = 0
∑
h Pn (1 / 2)

= [P0 (−1 / 2) + h P1 (−1 / 2) + h 2 P2 (−1 / 2) + L] × [P0 (1 / 2) + h P1 (1 / 2) + h 2 P2 (1 / 2) + L]
Now equating the coefficient of h 2 n on both sides, we get
 1  1 1  1 1  1 1
Pn  −  = P0  − P2 n   + P1  −  P2 n −1   + L + P2 n −1  −  P1  
 2  2 2  2 2  2 2
 1 1
+ P2 n  −  P0   .
 2  2
E5) Since Pn ( x ) satisfies Legendre’s equation, we have
(1 − x 2 )Pn′′ ( x ) − 2 xPn′ ( x ) + n (n + 1)Pn ( x ) = 0 (130)
a) putting x = 1 in Eqn.(130), we get
− 2Pn′ (1) + n (n + 1) Pn (1) = 0
1
i.e., Pn′ (1) = n (n + 1) Pn (1)
2
1
= n ( n + 1) [Q Pn (1) = 1]
2
b) putting x = −1 in Eqn.(130), we get
2Pn′ (−1) + n ( n + 1) Pn (−1) = 0
1
i.e., Pn′ ( −1) = − n (n + 1) Pn (−1)
2
1
= (−1) n −1 (n ) (n + 1) [Q Pn (−1) = (−1) n ]
2
E6) a) Putting x = 0 in Eqn.(27), we obtain
∞
(1 + t 2 ) −1 / 2 = ∑ P ( 0) t
n =0
n
n
(131)

Expending left hand side of Eqn.(131) in binomial series

∞
(1 + t 2 ) −1 / 2 = ∑ ( )t
n =0
−1 / 2
n
2n
(132)

114
 Comparing terms on the right in series (131) and (132), we see that series Legendre, Hermite and
Laguerre Polynomials
(132) has only even powers of t . Thus, Pn (0) = 0 for n = 1, 3, 5, K or,
equivalently, P2 n +1 (0) = 0, n = 0,1, 2, K .
b) Since all odd terms in series (131) are zero, replace n by 2n in the series
and compare with (132), we obtain
−1 −1   −1 
 − 1 K  − n + 1
( )
P2 n (0) = −1n/ 2 =
2  2   2 
n!
11  1 
(−1) n  + 1 K  + n − 1
22  2 
=
n!
 1  1  1 1
(−1) n  + n − 1 + n − 2  K  + 1
2  2  2 2
=
n!
 1
1  (−1) n Γ n + 
= (−1) n  2
+ n − 1
= (−1) n n −1n/ 2 = ( ) 2
   
1
 n  n!Γ 
 2
(−1) n (2n )!   1 2 n!  1  
= n . Q Γ n + 2  = 2n Γ 2   .
   2 n!   
2
2 (n!)

E7) In the expressions for P0 ( x ), P1 ( x ), P2 ( x ), P3 ( x ) in terms of x , as obtained in

Example 3, put x = cos φ and obtain the required results.

E8) a) Multiplying Eqn.(37) by x and then subtracting it from Eqn.(36), we get

(1 − x 2 )Pn′ ( x ) = n{Pn −1 ( x ) − xPn ( x ) } .
b) Multiplying (36) by x and then subtracting it from (37), we get
(1 − x 2 ) Pn′ ( x ) = (n + 1) {xPn ( x ) − Pn +1 ( x ) } .

E9) From relation (39), we have

(1 − x 2 ) Pn′ ( x ) = n{Pn −1 ( x ) − x Pn ( x )}
= (2n + 1) x Pn ( x ) − (n + 1) Pn +1 ( x ) − nx Pn ( x ) [ using (38)]
= (n + 1) [ x Pn ( x ) − Pn +1 ( x )]

E10) We know that

1.3.K (2n − 1)  n n (n − 1) n − 2 
Pn ( x ) = x − x + L
n!  2(2n − 1) 
Thus we see that P0 ( x ) is a constant, P1 ( x ) is a polynomial of degree 1, P2 ( x )
is a polynomial of degree 2 and so on.
Thus we can express x m in terms of Legendre forms P0 ( x ), P1 ( x ), K Pm ( x ) , and
hence a polynomial f ( x ) of degree k < n can be expressed as
f ( x ) = A 0 P0 ( x ) + A 1 P1 ( x ) + L + A k Pk ( x )
1 k 1
∴ ∫ f ( x )Pn ( x )dx = ∑
r =0
Ar ∫ P ( x) P ( x) dx
r n
−1 −1
k
= ∑ A .0 ,
r =0
r as r ≠ n

= 0.

E11) Proceeding as in Example 4. 115

Ordinary Differential
f ( x ) = x 3 + ax 2 + bx + c
Equations
2 3  P (x ) 2 
= P3 ( x ) + P1 ( x ) + a  0 + P2 ( x ) + b P1 ( x ) + c P0 ( x )
3 5  3 3 
2 2 3  a 
= P3 ( x ) + a P2 ( x ) +  + b  P1 ( x ) +  + c  P0 ( x ).
3 3 5  3 

E12) From E8) (a) and (b), we have

(1 − x 2 ) Pn′ ( x ) = n {Pn −1 ( x ) − x Pn ( x ) }
and (1 − x 2 ) Pn′ ( x ) = (n + 1) {x Pn ( x ) − Pn +1 ( x )}
Substituting for xPn ( x ) from the first relation in second relation, we get
 (1 − x 2 )Pn′ ( x )  
(1 − x 2 )Pn′ ( x ) = (n + 1)  Pn −1 ( x ) −  − Pn +1 ( x )
  n  
 n + 1
or, (1 − x 2 ) 1 + Pn′ ( x ) = (n + 1) [Pn −1 ( x ) − Pn +1 ( x )]
 n 
or, (2n + 1) (1 − x 2 ) Pn′ ( x ) = n (n + 1) [Pn −1 ( x ) − Pn +1 ( x ) ] .

E13) From generating function of Legendre Polynomials, we have

∞
[1 − 2 xt + t 2 ] −1 / 2 = ∑
m=0
t m Pm ( x )
1 1
Pn ( x )  ∞ m 
∴ ∫ 1 − 2 xt + t 2
dx = ∫ Pn ( x ) 
m=0
∑
t Pm ( x ) dx

−1 −1
1

∫ [P
n
=t n ( x )] 2 dx , [using formula (41)]
−1
2
= tn. .
2n + 1

E14) We know that

∞
(1 − 2xt + t 2 ) −1 / 2 = ∑
n =0
t n Pn ( x )

Integrating both sides w.r.t t between the limit 0 to 1, we get

1 ∞ 1

∫ (1 − 2 xt + t 2 ) −1 / 2 dt = ∑
n =0
∫
Pn ( x ) t n dt
0 0
Putting x = cos θ , we get
1 ∞ 1
dt t n +1
∫ 1 − 2 cos θ.t + t 2
= ∑
n =0
Pn (cos θ).
n +1
0 0
1 ∞
dt 1
or, ∫ ( t − cos θ) 2 + sin 2 θ
= ∑ P (cos θ). n + 1
n =0
n
0
1 ∞
Pn (cos θ)
or, ln ( t − cos θ) + sin 2 θ + ( t − cos θ) 2  = ∑
  0 n +1
n =0

 (1 − cos θ) + sin 2 θ + (1 − cos θ) 2  ∞ P (cos θ)

or, ln 
 1 − cos θ
= n
∑
 n =0 n + 1
 
2 + 1 − cos θ ∞ Pn (cos θ)
or, ln
1 − cos θ
=
n =0
∑
n +1
116
 Legendre, Hermite and
θ Laguerre Polynomials
2 + 2 sin 2 ∞
Pn (cos θ)
or, ln
θ
2
= ∑
n =0 n +1
2 sin 2
2
θ
1 + sin
or, ln 2 = 1 + 1 P (cos θ) + 1 P (cos θ) + 1 P (cos θ) + L .
1 2 3
θ 2 3 4
sin
2

dx dy dy
E15) x = cos φ , therefore, = − sin φ = − 1 − x 2 and = − sin φ
dφ dφ dx
1 d  dy  −1  2 dy
∴  sin φ  + n (n + 1) y =  2x 1 − x dx
sin φ dφ  dφ  1− x2 

+ (1 − x 2 )
d2y 
− 1− x2   + n ( n + 1) y
dx 2   
d2y dy
= (1 − x 2 ) 2
− 2x + n (n + 1) y .
dx dx

E16) a) y = Pn ( x ) satisfies the given equation and the boundary conditions are
y(0) = 0 ⇒ Pn (0) = 0 and y(1) = 1 ⇒ Pn (1) = 1 , (which is true).
Now Pn (0) = 0 , for n odd (see Example 1)
∴ y = Pn ( x ) is a solution for n = 1, 3, 5, K .
b) y = Pn ( x ) satisfies the given equation and the condition y(1) = 1 [Q Pn (1) = 1]
y ′(0) = 0 ⇒ Pn′ (0) = 0 , but Pn′ (0) = n Pn −1 (0) [using (39)]
Pn′ (0) = 0 ⇒ Pn −1 (0) ⇒ n = 0, 2, 4, K .
∴ y = Pn ( x ) is a solution for n = 0, 2, 4, K .

E17) Multiplying both sides of Eqn.(61) with Pm ( t ) and integrating from − 1 to 1

w.r.t. t , we get
1 ∞ 1
1
∫ x−t
Pm ( t )dt =
n =0
∑ ∫
Q n ( x ) (2n + 1) Pn ( t ) Pm ( t ) dt
−1 −1
1
Pm ( t ) 2
∴ ∫ x−t
dt = Q n ( x ) (2n + 1)
2n + 1
−1
1
1 Pn ( t )
or, Q n (x) =
2 ∫ x−t
dt .
−1

E18) a) Comparing the given equation with Eqn.(1), we see that given equations is a
Legendre’s equation with n = 0 , so its general solution is
y = C1 P0 ( x ) + C 2 Q 0 ( x ) .
b) Comparing the given equation with Eqn.(1), we find that n (n + 1) = 12 ,
therefore, n = 3, − 4 . Since n is non-negative, the general solution of the
given equation is y = C1 P3 ( x ) + C 2 Q 3 ( x ) .
c) y = C1 P1 ( x ) + C 2 Q1 ( x ) .
d) y = C1 P5 ( x ) + C 2 Q 5 ( x ) .

P0 Q0 1
E19) W (P0 , Q 0 ) = = P0 Q ′0 − Q 0 P0′ = .
P0′ Q ′0 1− x2
117
Ordinary Differential E20) W [Pn ( x ), Q n ( x ) ] = Pn Q ′n − Q n Pn′
Equations
Substituting in the above equation for Q n and Q ′n from Eqn.(53), we get
 dx  1
W [Pn ( x ), Q n ( x )] = Pn Pn′ A n + B n

2 ∫
(1 − x ) [Pn ( x )] 2 
+ [Pn ( x )] 2 B n
(1 − x ) [Pn ( x )] 2
2

 dx 
− Pn Pn′  A n + B n

∫ 
(1 − x 2 ) [Pn ( x )]2 
Bn
= , n = 0, 1, 2, L .
1− x2
∞
E21) Consider y = ∑
m=0
c m x m , c 0 ≠ 0 . Substitute for y, y ′ and y ′′ in Eqn.(63),

simplify and equate the coefficients of equal power of x to get two linearly
independent solutions y1 ( x ) and y 2 ( x ) [ref. E9) Unit 2].

∞
H n (x ) t n H (x ) 2 H 3 (x) 3
∑
2
E22) e 2 tx − t = = H 0 (x ) + H1 (x ) t + 2 t + t +L
n =0 n! 2! 3!
2 (2 tx − t 2 ) 2 (2 tx − t 2 ) 3
Now, e 2 tx − t = 1 + (2 tx − t 2 ) + + +L
2! 3!
 4x 3 − 6x  3
= 1 + (2 x ) t + (2 x 2 − 1) t 2 +   t +L

 3 
Comparing the two series, we have
H 0 ( x ) = 1, H 1 ( x ) = 2 x , H 2 ( x ) = 4x 2 − 2, H 3 ( x ) = 8x 3 − 12 x .

∞
H n (x) n
∑
2
E23) a) Let w = e ( 2 xt − t )
= t
n =0 n!
∂w ∂w
then, ( x − t ) −t =0
∂x ∂t
∞ ∞ ∞
H ′n ( x ) n n H n (x)t n n H ′n −1 ( x ) n
or, ∑
n =0
x
n!
t −
n =0 n!
∑ =
n =0 n!
∑ t

Comparing coefficients of t n on both the sides, weget

x H ′n ( x ) − nH n ( x ) = n H n −1 ( x ).
b) From relation (75), we get H n ( x ) = 2 x H n −1 ( x ) − [2(n − 1) H n − 2 ( x ) ]
[using relation (74)].
c) H ′n′ ( x ) = 2n H ′n −1 ( x ) [using Eqn.(74)]
= 2n [ 2 x H n −1 ( x ) − H n ( x ) ] [using Eqn.(75)]
= 2 x H ′n ( x ) − 2n H n ( x ) [using Eqn.(74)].

E24) a) We have from Eqn.(68)

n
(−1) k 2n !(2x ) 2 n − 2 k
H 2n (x) =
k =0
∑
k! (2n − 2k )!
The last term in the above summation is a constant and therefore, we have
(−1) n 2n!
H 2 n ( 0) = .
n!
b) From Eqn.(68), we have
n
(−1) k 2n + 1!(2 x ) 2 n +1− 2 k
H 2 n +1 ( x ) =
k =0
∑
k! (2n + 1 − 2k )!
The last term in the above summation is a function of x , hence we get
H 2 n +1 (0) = 0 .
118
 c) Differentiating both sides of Eqn.(68), we obtain Legendre, Hermite and
 n −1 
Laguerre Polynomials
 2 
 
( −1) k n!(2 x ) n − 2 k −1 .2
H ′n ( x ) =
k =0
∑
k!(n − 2k − 1)!
from which it follows that
n −1
(−1) k (2n )!.2.(2 x ) 2 n − 2 k −1
H ′2 n ( x ) =
k =0
∑
k! (2n − 2k − 1)!
Putting x = 0 , we get
H ′2 n (0) = 0
n
(−1) k (2n + 1)!.2.(2 x ) 2 n − 2 k
d) H ′2 n +1 ( x ) = ∑ k =0 k! (2n − 2k )!
(−1) n (2n + 1)!.2 (−1) n (2n + 2)!
H ′2 n +1 (0) = = .
n! (n + 1)!

E25) From Eqn.(66), we have

∞
tk
∑
2
e ( 2 xt − t )
= H k (x)
k =0 k!
∞
tk
∑
2
e 2 xt = e t H k (x )
k =0 k!
∞ n n ∞ ∞
(2x ) t t 2m tk
∴ ∑n =0 n!
=
m=0 m !
∑ ∑
k =0
H k (x)
k!
∞ [ n / 2]
H n − 2k ( x ) t n [by interchanging m and k
= ∑ ∑
m =0 k =0 k !(n − 2k ) ! and setting k = n − 2 k ].

Comparing the coefficients of t n on both the sides, we have

[ n / 2]
n ! H n −2k (x )
xn = ∑
k =0 2 n k !(n − 2k ) !

−3 7 3 1
E26) Using E22) required result is, H 0 + H1 − H 2 + H 3 .
2 4 4 8

dn
E27) We have, L n ( x ) = e x ( x n e − x ) , then
dx n
d
L 0 ( x ) = 1, L1 ( x ) = e x (x e −x ) = 1 − x,
dx
2
d
L 2 (x) = e x 2
(x 2 e −x ) = 2 − 4x + x 2
dx
d3
L 3 (x) = e x 3
( x 3 e − x ) = 6 − 18x + 9x 2 − x 3
dx

 − xt 
E28) Let w ( x , t ) = (1 − t ) −1 exp  . It satisfy the identity
1 − t 
∂w
(1 − t ) 2 + (x − 1 + t) w = 0
∂t
Using Eqn.(86), the above equation reduces to
∞

∑ [(n + 1) L
n =1
n +1 ( x ) + ( x − 1 − 2n ) L n ( x ) + n L n −1 ( x ) ] t n = 0

Equating the coefficients of t n to zero, relation (89) is proved.

119
Ordinary Differential ∂w
Equations Similarly, using Eqn.(86) into the identity (1 − t ) + tw = 0 , relation (90) is
∂x
obtained.

Differentiating relation (89), using relation (90) in it and then eliminating

L ′n+1 ( x ) and L ′n−1 ( x ) from it relation (91) is obtained.

Differentiating relation (91) and using relation (90), relation (92) is obtained.

E29) From Laguerre’s differential equation we have for any two Laguerre
polynomials L m ( x ) and L n ( x )
x L ′m′ + (1 − x ) L ′m + mL m = 0
x L ′n′ + (1 − x ) L ′n + n L n = 0
Multiplying these equations by L n and L m respectively and subtracting, we
obtain
x [L n L ′m′ − L m L ′n′ ] + (1 − x ) [L n L ′m − L m L ′n ] = (n − m) L m L n
d (1 − x ) ( n − m)
or, [L n L ′m − L m L ′n ] + [L n L ′m − L m L ′n ] = LmLn
dx x x
d
or, {x e − x [L n L ′m − L m L ′n ]} = (n − m)e − x L m L n , where we have multiplied
dx
by I.F. e ∫
(1− x ) / x dx
= e ln x − x = x e − x
Integrating from 0 to ∞
∞

∫ {
(n − m) e − x L m ( x )L n ( x )dx = x e − x [L n L ′m − L m L ′n ] }∞
0
=0
0
∞

∫e
−x
If m ≠ n , L m ( x ) L n ( x ) dx = 0 .
0
2 2 2
E30) a) Putting ψ n ( x ) = e − x / 2 H n ( x ) and ψ ′n ( x ) = e − x / 2 H ′n ( x ) − x e − x / 2 H n ( x ) ,
in the given equation, it reduces to
2nH n −1 ( x ) = H ′n ( x ) which is relation (74)
b) The given equation reduces to
2 xH n ( x ) − 2n H n −1 ( x ) = H n +1 ( x )
replacing n by (n − 1) we get relation (76).
c) Substituting for ψ n ( x ) and its derivatives in the given equation, relation
(75) is obtained.

∞ ∞
2

∫ [ψ ] ∫
2
E31) n (x ) dx = 1 ⇒ c 2n e − x [ H n ( x ) ] 2 dx = 1
−∞ −∞
1
⇒ c 2n 2 n n! π = 1 or, c n = [using result (78)]
[2 n
n! π ]1/ 2

1
−
∴ψ n ( x ) = [ π 2 n!] n 2 −x 2
e H n ( x ), n = 0,1, 2, K .
—x—

120