Unit 5 Special Functions-II

UNIT 5
SPECIAL FUNCTIONS-II
Structure

5.1 Introduction 5.4 Sturm-Liouville Problem

Expected Learning Outcomes 5.5 Expansion in Orthogonal Functions
5.2 Hermite Polynomials 5.6 Summary
Generating Function and Recurrence 5.7 Terminal Questions
Relations for Hermite Polynomials 5.8 Solutions and Answers
Orthogonality Relations for Hermite
Polynomials
5.3 Laguerre Polynomials
Generating Function and Recurrence
Relations for Laguerre Polynomials
Orthogonality Relations for Laguerre Polynomials

5.1 INTRODUCTION
In Unit 4, you have learnt about Legendre polynomials, their generating function,
recurrence relations and orthonormality. You have also learnt about spherical
harmonics and hypergeometric functions. You have solved their respective
differential equations and problems based on these special functions.
In this unit, we will discuss the special functions Hermite and Laguerre
polynomials. You will learn about their generating functions, recurrence relations
and orthogonality property. We will also discuss the Sturm-Liouville problem.
Finally, we will explain how to expand a given function in terms of orthogonal
functions.
Expected Learning Outcomes
After studying this unit, you should be able to:
 solve the differential equations for Hermite and Laguerre polynomials;
 write generating functions of Hermite and Laguerre polynomials and obtain
recurrence relations for Hermite and Laguerre polynomials.
 state orthogonality properties of Hermite and Laguerre polynomials and
use them to solve physical problems;
 define a Sturm-Liouville problem, and state and apply the properties of a
Sturm-Liouville problem to solve ODEs; and
 expand any given function in terms of orthogonal functions.

5.2 HERMITE POLYNOMIALS

You may have studied Hermite’s differential equation and its solutions in UG
physics or mathematics courses (refer to SAQ 3 in Unit 3 of PHE-05). It is
given as: 109
Block 1 Mathematical Methods in Physics
d 2y dy
 2x  2ny ( x )   (5.1)
dx 2 dx
Hermite’s polynomials find applications in quantum theory of harmonic
oscillators. You can obtain the general solution of Hermite’s differential
equation by solving SAQ 1. It is:
 ( 2) n 2 ( 2) 2 n (n  2) 4
y  a0 1  x  x
 2! 4!
( 2) 3 n(n  2) (n  4) 6 
 x  ...
6! 
 ( 2) (n  1) 3 ( 2) 2 (n  1) (n  3) 5 
 a1  x  x  x  ...
 3! 5! 
(5.2)
The constants a 0 and a 1 may take arbitrary values.

SAQ 1

Solve Eq. (5.1) and obtain its general solution given by Eq. (5.2).

If n is a non-zero negative integer, the series given in Eq. (5.2) will be an

infinite series. You may now like to know: What happens if n is zero or an even
positive integer? To understand this, let us consider the case corresponding to
n  6. You can see that the fifth and all subsequent terms of the series (not
explicitly shown) in the first square bracket in Eq. (5.2) will be zero as their
numerator will have a factor (n – 6). It means that for n  6, the first series will
( 2)3 6  4  2 6
terminate at x .
6!
To extend this discussion for the general case, let us write n  2m,
m  0, 1, 2... Then the series in first square bracket of Eq. (5.2) will terminate
at the term:
( 2) m (2m ) (2m  2) 2m ( 1) m m!
x  (2x ) 2m
(2m )! (2m )!

and all subsequent terms will have their numerators equal to zero. The series
in the second square bracket of Eq. (5.2) will however, remain an infinite
series. But if we choose a1  0, we will obtain a particular solution of Eq. (5.1),
which is a polynomial of degree 2m in x. However, you should note that such a
polynomial will contain only even powers of x and the coefficient a0 will still be
arbitrary. For the series to appear more systematic, we choose the constant
a0 to be
(1) m (2m)!
a0 
m!
The polynomial so obtained is called Hermite polynomial of degree 2m and
is denoted by the symbol H2m ( x ) :

( 1) m (2m)! (1) m1(2m)!

H 2m ( x )   ...
m!
(1)(2m)!
 (2x )2m 2  (2x )2m (5.3)
110 (2m  2)!
Unit 5 Special Functions-II
Note that the term containing the highest power of x is (2x ) 2m .

Similarly, when n is a positive odd integer, say 2m  1; m  0, 1, 2,..., the

second series in Eq. (5.2) will terminate at:
( 1) m m! (2 x ) 2m 1
(2m  1)! 2

and the first series remains an infinite series. As before, we get a particular
polynomial solution of Hermite’s differential equation by putting a 0  0 and
choosing a1 as

( 1) m (2m  1)!

a1  2
m!
This leads to Hermite’s polynomial of degree 2m  1:
( 1) m (2m  1)! ( 1) m 1(2m  1)!
H 2m 1( x )  2x  (2x ) 3
m! 3! (m  1)!

( 1)(2m  1)!
 (2x )2m 1  (2 x )2m 1 (5.4)
(2m  1)!

We would like to point out here that Eq. (5.4) contains only odd powers of x;
the term containing the highest power of x is (2x ) 2m 1.

We can now combine Eqs. (5.3) and (5.4) by taking n to be any positive
integer including zero, and define the Hermite polynomial of degree n as: Any solution of
Eq. (5.1) can be
n(n  1)
H n ( x )  (2x ) n ( 2 x ) n 2 multiplied by an
1! arbitrary constant to
obtain another
n(n  1) (n  2) (n  3) ( 1) n / 2 n!
 (2x ) n 4  ...  solution. Then the
2! (n / 2)! question arises: Why
are we taking the
(if n is even) (5.5a) constants a0 and a1
n(n  1) in this particular
H n ( x )  (2x ) n  ( 2 x ) n 2 manner? This is
1!
because Hermite
n 1 polynomial of degree
n(n  1) (n  2) (n  3) ( 1) 2 n!
 (2 x ) n 4  ...  2x n is defined in such a
2!  n  1! way that the term
 
 2  containing the highest
power of x is ( 2 x )n .
(if n is odd) (5.5b)
where we have written the terms in decreasing powers of x.
From Eq. (5.5) we can write the first few Hermite polynomials as follows:
H0 ( x )  1

H 1( x )  2 x

H 2 ( x )  4x 2  2

H 3 ( x )  8 x 3  12x

H 4 ( x )  16x 4  48x 2  12 (5.6)

111
Block 1 Mathematical Methods in Physics
A plot of some of these polynomials is shown in Fig. 5.1. For ease in scaling,
H n ( x )(n  0)s are divided by n 3 .

n=0 Hn(x) n=1 n=2

n=3
n=4

n=5

H (x)
Fig. 5.1: Plots of Hermite polynomials H0 ( x ), n (n  1 to 5).
n3
The series representation of Hermite polynomials becomes somewhat
unwieldy, particularly when we have to evaluate integrals involving Hermite
polynomials. In such situations and for the derivations of many other
properties of Hermite polynomials, it is convenient to use the generating
function for Hermite polynomials. So, we now state the generating function for
the Hermite polynomials and derive the recurrence relations for Hermite
polynomials of different orders and their derivatives.
5.2.1 Generating Function and Recurrence Relations for
Hermite Polynomials
The generating function for Hermite polynomials is:

tn
g ( x, t )  e 2 xt  t 
2
 Hn ( x ) n! (5.7)
n 0

As in the case of Bessel functions and Legendre polynomials, the generating

function for Hermite polynomials is expanded in powers of t and the
x-dependent coefficients are related to the special function. Therefore, we
rewrite g(x, t) as
g ( x, t )  e 2 xt  e t
2

and express the exponential functions in their respective power series to get

(2xt ) j  t 2m
g ( x, t )   j! m0
 ( 1) m
m!
j 0
 
(2x ) j 2m j
   (1)m j! m!
t
j 0 m 0

To obtain the coefficient of t n , we put j  2m  n. You should note that this

equality can be satisfied for various combinations of the values of j and m,
namely j  n, m  1; j  n  2, m  1; j  n  4, m  2; and so on. If n is
n n 1
even, we will have j  0, m  and if n is odd, we will have j  1, m  .
2 2
Thus, the coefficient of t n is:
( 2 x ) n ( 2 x ) n 2 (2x ) n 4
   ...
n!0! (n  2)!1! (n  4)!2!
112
Unit 5 Special Functions-II
n n 1
( 1) 2 ( 1) 2
The last term in this series is for even n and 2 x for odd n.
 n  n  1
0!  ! 1! !
2  2 
tn
Thus the coefficient of , i.e. H n (x ) will be obtained by multiplying the above
n!
expression by n!. You should recognize that the resultant expression is
identical to that given in Eq. (5.5). Let us use the generating function to relate
Hermite polynomials of different kinds.
Example 5.1
Establish the relation between H n (x ) and H n (  x ) using the generating
function.
Solution : In Eq. (5.7) for the generating function, we change x to  x and t to
 t .This will leave the exponential function unchanged so that
( t ) n tn
e 2 xt t 
2
 H n (x )
n!
  ( 1) n H n (  x )
n!
n

On comparing this expression with that given in Eq. (5.7), you will get
H n ( x )  ( 1) n H n (  x )

or H n (  x )  ( 1) n H n ( x )

From this we note that if we change the sign of x, the Hermite polynomials for
even positive integral values of n do not change whereas those with odd
positive integral values of n just change sign. This result is, of course, obvious
from the fact that Hermite polynomials contain only even (odd) powers of x
when n is even (odd).

Now we will use the generating function to obtain the recurrence relations for
Hermite polynomials.
Recurrence Relations
When we differentiate both sides of Eq. (5.7) partially with respect to t, we get:

t n 1
(2 x  2t )e 2 xt t 
2
 H n ( x ) (n  1)!
n 1

Replacing e 2xt t on the left hand side by the right hand side of Eq. (5.7), we
2

get:
 
tn t n 1
(2 x  2t )  H n ( x ) n!   H n ( x ) (n  1)!
n 0 n 1

Equating coefficients of t n 1 from the two sides, we have:

H n 1( x ) H ( x ) H n 2 ( x )
2x 2 n 
(n  1)! n! (n  1)!

On multiply throughout by (n  1)!, we can write

H n  2 ( x )  2 xH n 1( x )  2(n  1) H n ( x ) (5.8) 113
Block 1 Mathematical Methods in Physics
You should note that this recurrence relation connects Hermite polynomials of
three successive orders.
We will now illustrate how the generating function and this recurrence relation
can be used to obtain expressions of some lower order Hermite polynomials.
You should go through the following the example carefully.
Example 5.2
Starting from the generating function for the Hermite polynomials, obtain
expressions for H 0 ( x ) and H1( x ) and then use the recurrence relation given
in Eq. (5.8) to obtain expressions for H 2 ( x ), H 3 ( x ) and H 4 ( x ).

Solution : We first write the exponential in the generating function given in

Eq. (5.7) as a power series in 2 xt  t 2 :

(2 xt  t 2 ) 2 tn
e 2 xt t  1  (2 xt  t 2 ) 
2

2!
 ...  Hn (x)
n!
n 0

You should note that H 0 ( x ) corresponds to a term independent of t. on the left

side, the term independent of t is 1. Hence, we can write:
H0 (x)  1
Further, H1( x ) occurs as the coefficient of t. On the left hand side, the
coefficient of t is 2x. Therefore,
H 1( x )  2 x
From the recurrence relation in Eq. (5.8) with n  0, we get:
H 2 ( x )  2 xH 1( x )  2H 0 ( x )
On substituting for H1( x ) and H 0 ( x ), we obtain:

H 2 ( x )  4x 2  2
If we repeat this procedure for n  1 and n  2, we get:

H 3 ( x )  2xH 2 ( x )  4H1( x )  2x( 4x 2  2)  4(2x )  8x 3  12x

and H 4 ( x )  2xH 3 ( x )  6H 2 ( x )  16x 4  48x 2  12

Note that these expressions for Hermite polynomials are the same as given in
Eq. (5.6) and obtained from the series given by Eq. (5.5). Proceeding in this
way, you can calculate higher order Hermite polynomials. You are advised to
obtain expressions of a couple of these.

You can obtain another recurrence relation by partially differentiating both

sides of Eq. (5.7) with respect to x:
H n 1( x )  2( n  1)H n ( x ) (5.9)

where the prime denotes differentiation with respect to the argument x. We

leave the derivation of this result as SAQ 2 for you.

SAQ 2
Differentiate the generating function for Hermite polynomials partially with
respect to x and obtain the recurrence relation given in Eq. (5.9).
114
Unit 5 Special Functions-II
On combining the recurrence relations given in Eqs. (5.8) and (5.9), we get:
H n 2 ( x )  2 x H n 1( x )  H n 1( x ) (5.10)

By changing (n  1) to n in Eq. (5.10), we can write:

H n 1( x )  2 xH n ( x )  H n ( x ) (5.11)

Again differentiating both sides of this equation with respect to x, we get:

H n 1( x )  2H n ( x )  2 xH n ( x )  H n ( x )

On combining this with Eq. (5.9), we get:

2(n  1)H n ( x )  2H n ( x )  2 xH n ( x )  H n ( x )

We can rearrange this equation and write:

H n ( x )  2 x H n ( x )  2nHn ( x )  0 (5.12)

Do you recognise this equation? It is Hermite’s differential equation (for

positive integral or zero value of n) for H n (x ) . It means that if the polynomials
H n (x ) satisfy the recurrence relations in Eqs. (5.8) and (5.9), they must
satisfy Hermite’s differential equation.
The generating function has many other uses. You can utilize it to obtain
Rodrigues’ formula, which gives a compact expression for Hermite
polynomials.
Rodrigues’ Formula
We write Eq. (5.7) in expanded form:
t2 tn
e 2xt t  1  t H1( x )  H 2 ( x )  ...  H n ( x )  ...
2

2! n!
Successive partial differentiation of both sides with respect to t yields:
nt n1
t
e
 2xt t 2
 2t
 H1( x )  H 2 ( x )  ... 
2! n!
H n ( x )  ...

e 
 2 2 xt t 2

 H2 (x) 
32
H 3 ( x )  ... 
n(n  1) n 2
t H n ( x )  ...
t 2 3! n!

e 
 n 2 xt t 2

 Hn (x) 
(n  1) n(n  1)...2
(n  1)!
t H n 1( x )  ...
t n
For t  0, this expression simplifies to:

Hn (x)   e 
  n 2 xt t 2 
 
 t n  t 0

Since e 2xt t  e x e ( x t ) , we can rewrite this expression as:

2 2 2

2   2
n
Hn (x)  e x  e ( x t ) 
 t n
 t 0

We note that e ( x t ) is a function of x – t, and for such a function the partial

2

derivative with respect to t can be obtained from the partial derivative with
115
Block 1 Mathematical Methods in Physics
respect to x by just changing the sign. So for nth order partial derivative, the
sign will change n times. Hence, we can write:

n   2
n
H n ( x )  e x ( 1)  e ( x t ) 
2

 x n
 t 0

dn
 ( 1) n e x (e  x )
2 2
(5.13)
dx n

This is Rodrigues’s formula for the Hermite polynomials.

You may like to apply Rodrigues’s formula for a simple case. Solve SAQ 3.

SAQ 3
Use Rodrigues’s formula to evaluate H 4 ( x ) .

Yet another interesting application of the generating function for the Hermite
polynomials is in evaluation of integrals involving their product with suitable
polynomials. Of particular importance is the result that the integral over x from
  to   of the product of two Hermite polynomials of different degrees with
e  x is zero. (The function e  x is called the weight function.) These are
2 2

termed the orthogonality relations of Hermite polynomials. You will now

learn how to obtain as well as apply these relations. Along with the values of
similar integrals for Hermite polynomials of the same degree, we can calculate
the expansion coefficients when a function is expanded in terms of Hermite
polynomials.
5.2.2 Orthogonality Relations for Hermite Polynomials
To obtain orthogonality relations for Hermite polynomial, we can start with
Hermite differential equation as in the case of Legendre and Bessel
polynomials, or the generating function for the Hermite polynomials. For the
sake of variety here we consider the latter option:

tn
e 2 xt t 
2
 Hn (x)
n!
n 0

Next we change t to u and express the expansion as a power series in u:


um
e 2 xu u 
2
 Hm (x)
m!
m 0

Now we multiply these equations and the resultant expression by e  x . This

2

gives:
 
t nu m
e  x 2 xt t 2 xu u 
2 2 2 2

e  x H n ( x )H m ( x )
n! m!
n 0 m 0

You can rewrite the left hand side as:

e 2tu e ( x t u 2xt 2xu2tu )  e 2tu e ( x t u )

2 2 2 2

Proceeding further, we integrate both sides with respect to x from   to  

and, on the right hand side, interchange the order of the summations and
116 integration. The leads to
Unit 5 Special Functions-II
    t nu m
  
Consider the integral:
 e ( x t u ) dx  e  x H n ( x )H m ( x )dx
2 2
e 2tu
n 0 m  0
n! m! 
 z 2 dz
(5.14) I e

Do you recognise the integral on the left side? By changing the variable of

integration from x to z through the substitution x  t  u  z, so that dx  dz,
 
 2 e z dz

 e z dz 0
which is just (1/ 2) and is equal to
2
you can rewrite it as  (read
 We put z 2  p so that
the margin remark). So, the left hand side of the above equation reduces to 2z dz  dp
1/ 2 e 2tu . On expanding the exponential in a power series, we can write or 2dz 
dp
z
  
 t nu m (2tu ) n  p 1/ 2dp
 
e  x H n ( x )H m ( x ) dx
2

n! m!
 
n!
 Hence,
n 0 m 0 n 0

 (2tu )1 (2tu ) 2 (2tu ) n  

  1    ...  ... 1/ 2  p
 1! 2! n!  I p e dp
0
On equating the coefficients of t n u m , we get:  (1/ 2)

 e x H n ( x )H m ( x )dx  0 if n  m
2


 e x H n ( x ) H m ( x ) dx  2 n n! 1/ 2 if n  m
2
and

On combining these results, we can write:


 e x H n ( x ) H m ( x ) dx  2 n n! 1/ 2  nm
2
(5.15)

where  nm is the Kronecker delta.

In words we can say that the Hermite polynomials of different degrees are
orthogonal to each other on the interval ( ,  ) with weight function e  x .
2

You may recall that for the Legendre polynomials, the weight function is unity
and the range of integration varies from  1 to  1.
You may now like to solve SAQ 4 on evaluation of integral involving Hermite
polynomials.

SAQ 4
Use the generating function for Hermite polynomials to evaluate the integral


 xe  x
2
H n ( x ) H m ( x ) dx


We hope that you have understood the properties of Hermite polynomials. We

can apply this knowledge to solve the one-dimensional harmonic oscillator
problem in quantum mechanics. This enables us to understand the nature of
small vibrations of atoms in molecules and quantization of the electromagnetic
field where energy of the field may be written as a sum of harmonic oscillator
type energy terms. You will study this in the first semester course on Quantum
Mechanics.
117
Block 1 Mathematical Methods in Physics
We will now discuss Laguerre polynomials.

5.3 LAGUERRE POLYNOMIALS

Laguerre’s differential equation is a linear second order ODE:
d 2y dy
x  (1  x )  ny ( x )  0 (5.16)
dx 2 dx
where n is a parameter. You can see that x  0 is a regular singular point of
this equation. The solution around this point can be obtained using the power
series method (refer to Unit 3, Block 1 of PHE-05):

y(x)   a j x a  j , a0  0 (5.17)
j 0

If n is not a positive integer or zero, this will remain an infinite series. However,
for particular case when n is a positive integer or zero, the series will terminate
at the (n 1)th term and reduce to an nth degree polynomial in x. If we further
choose a0  1, the resultant expression defines the Laguerre polynomial,
Ln (x ) :
n n(n  1) 2 n(n  1)...1 n
Ln ( x )  1  x x  ...  ( 1) n x
(1! ) 2 (2! ) 2 (n! ) 2

x x2 xn
 1  n C1  n C2  ...  (1) n n Cn (5.18)
1! 2! n!
From this equation it readily follows that Laguerre polynomials for the first few
values of n are:
L0 ( x )  1

L1( x )  1  x

x2
L2 ( x )  1  2x 
2
3 2 x3
L3 ( x )  1  3x  x 
2 6
2 3 x4
and L4 ( x )  1  4 x  3 x 2  x  (5.19)
3 24
Fig. 5.2 shows a plot of Laguerre polynomials versus x.

Ln(x) L2(x) L5(x)

L0(x)

x
L4(x)

L1(x) L3(x)

118 Fig. 5.2: Laguerre polynomials Ln (x ) for n  1 to n  5.

Unit 5 Special Functions-II
The Rodrigues’s formula for Ln (x ) enables us to express it in a compact
form:

ex d n
Ln ( x )  ( x ne x ) (5.20)
n! dx n

You may now like to solve SAQ 5 for practice.

SAQ 5

Show that the Laguerre polynomial defined in Eq. (5.20) is identical with that
given in Eq. (5.18).

5.3.1 Generating Function and Recurrence Relations for

Laguerre Polynomials
The generating function for Laguerre polynomials is:

e  xt /(1t )
g ( x, t ) 
1 t
  Ln ( x )t n t 1 (5.21)
n 0

If we differentiate both sides of Eq. (5.21) partially with respect to t, we get:


e  xt /(1t )
(1  t  x )
(1  t ) 3
  Ln ( x ) nt n 1
n 1

On combining this result with Eq. (5.21), we can write:

 
(1  t  x )  Ln ( x )t n  (1  t ) 2  Ln ( x )nt n 1
n 0 n 1

On equating the coefficients of t n 1 from the two sides and rearranging terms,
we obtain a recurrence relation which connects Laguerre polynomials of three
successive degrees:
(n  2)Ln 2 ( x )  (2n  3  x ) Ln 1( x )  (n  1)Ln ( x ) (5.22)
We now make a direct expansion of g ( x, t ) given in Eq. (5.21) in powers of t: The binomial series of e x
and (1  x) p (all x and p)
 xt  2  1 
   
xt 1 are
g ( x, t )  1    ...  
 1  t  2!   1  t    1  t  x2 x3
e x  1 x    ...
2! 3!
 1   ... 1  t  1
xt
 1 t  (1  x ) p  1  px 
p( p  1) 2
x
2!
 1  1  xt  2 xt 2  ... p( p  1) ( p  2) 3
 x  ...
3!
From this we note that the coefficient of t 0 and t 1 are 1 and 1 – x,
respectively. So we can say that L0 ( x )  1 and L1( x )  1  x . If we now put
n  0 in Eq. (5.22) and substitute for L0 ( x ) and L1( x ), we obtain
x2
L2 ( x )  1  2x . Proceeding in the same way we can successively
2
generate the Laguerre polynomials of all degrees.
In order to get another recurrence relation for the Laguerre polynomials, we
differentiate both sides of Eq. (5.21) partially with respect to x. This gives: 119
 Block 1 Mathematical Methods in Physics

t   xt 

dLn n
exp    t
(1  t ) 2  (1  t )  n 0 dx

This can be rewritten as:

 
 
dLn n
t Ln ( x ) t n  (1  t ) t
n 0 n 0
dx

On equating the coefficients of t n 1 on both sides, we get:

dLn 1 dLn
  Ln ( x ) (5.23)
dx dx
Using binomial expansion The recurrence relations given in Eqs. (5.22 and 5.23) can be combined to
we can write
obtain other relations. However, we will not go into these details.
 
 ... (1 t )1
xt
1  You have earlier learnt the orthogonality relations for Legendre, Bessel and
 1  t 
Hermite functions and their importance in understanding various physical
 (1  xxt (1  t )1  ...)(1  t )1 problems. In the following section, we obtain orthogonality relations of
  t2   Laguerre polynomials.
 1  xt 1  t   ...  ...
  2!   5.3.2 Orthogonality Relations for Laguerre Polynomials
   
 2  There are different ways of obtaining orthogonality relations for Laguerre
1  t  t  ...
 2!  polynomials. We start with the differential equations satisfied by Laguerre
 
polynomials of degrees n and k:
 t3 
 1  xt  xt 2  x  ... d 2 Ln dLn
 2!  x  (1  x )  nLn ( x )  0 (5.24a)
  dx 2 dx
(1  t  ...) d 2 Lk dLk
x  (1  x )  kLk ( x )  0 (5.24b)
xt 3 dx 2 dx
 1  xt  xt 2   ... 
2! We multiply Eq. (5.24a) by e  x Lk (x ) and Eq. Eq. (5.24b) by e  x Ln (x ) and
t  xt 2  xt 3  ... subtract the latter from the former. Then we can write the resultant expression
as:
d  x 
x e Lk ( x ) n  Ln ( x ) k 
dL dL 
   (n  k ) e  x Ln ( x )Lk ( x )  0
dx   dx dx 
We integrate this expression over x from 0 to  :
 
x e  x Lk ( x ) n  Ln ( x ) k   (n  k ) e  x L ( x )L ( x ) dx  0
dL dL
 dx dx 
0
 n k
0

Note that the expression within the square bracket is zero for both the limits (at
 because of the exponential factor and at zero because of the x factor).
Hence, we obtain:



(n  k ) e  x Ln ( x ) Lk ( x ) dx  0
0

Thus, if n  k


 e  x Ln ( x ) Lk dx  0 (5.25)
0
120
Unit 5 Special Functions-II
Thus, Laguerre polynomials of different degrees are orthogonal to each other
on the interval (0,  ) with weight factor e  x .
To obtain the orthogonality relation for n  k, we take the products of the two
sides of Eq. (5.21) with themselves. This gives:
 
e 2 xt /(1t )
(1  t ) 2
  Ln ( x ) t n  Lk ( x )t k
n 0 k 0

As before, we multiply both sides of this equation by e  x and integrate from 0

to  . This gives:
 1t   
 x

1
(1  t ) 2  e 1t dx  
t n t k e  x Ln ( x ) Lk ( x ) dx (5.26)
0 n 0k 0 0

Note that we have changed the orders of summations and integration on the
right hand side. For n  k, the right side of above equation reduces to
 
 t 2n  e  x L2n ( x ) dx.
n 0 0

e x
To proceed further, we use the formula  e ax dx  
a
. Thus, the left hand

side of Eq. (5.26) can be written as:

1t 
1 t  
 ( )
1 1 1
 e 1t  
(1  t ) 2 1 t  (1  t ) (1  t ) 1  t 2
0

For t  1, we have:

 t 2n
1
 1  t 2  t 4  ... 
1 t 2 n 0

  
so that  t 2n   
t 2n e  x L2n ( x ) dx
n 0 n 0 0

On comparing the coefficients of t 2n for all n, we obtain:


x 2
e Ln ( X ) dx 1 (5.27)
0

We now combine Eqs. (5.25 and 5.27) to write the orthonormality relation for
Laguerre polynomials as:


 e  x Ln ( x ) Lk ( x ) dx   nk (5.28)
0

Let us now discuss the Sturm-Liouville problem.

5.4 STURM-LIOUVILLE PROBLEM

You have learnt how to express a given function as a linear combination of an
orthogonal set of functions, for example, Legendre polynomials in Unit… The
question is: How do we determine or identify such orthogonal sets of 121
Block 1 Mathematical Methods in Physics
functions? When is an orthogonal set of functions complete a complete set?
Sturm Liouville theory helps us find the answers. So, in the final section of this
unit and block, we discuss the Sturm-Liouville problem, as it is referred to in
many texts.
We begin by stating: What is a Sturm-Liouville problem? The Sturm-Liouville
problem comprises a second order linear differential equation along with
certain boundary conditions. The Sturm-Liouville ODE on a finite interval [a, b]
is of the form:
d  d 
p( x ) y  q( x )  r ( x )y  0, x  [a, b] (5.29a)
dx  dx 

In brief, we write it as:

[ p( x )y ]  [q( x )  r ( x )]y  0 (5.29b)

Note that in the Sturm-Liouville equation (5.29a or 5.29b), p, q and r are

specific functions of x, and  is a parameter. Since  is a parameter, it can be
replaced by other variables or expressions. So, we can recast many ODEs
that occur when we separate PDEs into ODEs by separation of variables in
Sturm-Liouville form.
A Sturm-Liouville problem consists of:
 Sturm-Liouville equation (5.29a or 5.29b) on an interval [a, b] and

 boundary conditions, which, as you know, specify the behaviour of y at

x  a and x  b.
It is also assumed that the functions p, p, q and r are continuous and p > 0 on
(at least) the open interval a < x < b. This ensures that solutions of Eq. (5.29a
or b) exist.
Thus, the regular Sturm-Liouville problem can be stated as follows:

[ p( x )y ]  [q( x )  r ( x )]y  0 (5.29b)

c1y (a )  d 1y (a )  0 (5.29c)
c 2 y (b )  d 2 y (b )  0 (5.29d)

So, in the regular Sturm-Liouville problem:

 c1, c 2  0 and d1, d 2  0 ;

 p, p , q and r are continuous on [a, b];

 p and r are positive on [a, b].

The boundary conditions (5.29b and c) are homogeneous, mixed, separated

boundary conditions.
A Sturm-Liouville differential equation on an interval [a, b] with periodic
boundary conditions and p(a)  p(b) is called a periodic Sturm-Liouville
problem/system.
A Sturm-Liouville differential equation on an interval [a, b] with any of the
122 following conditions is called a singular Sturm-Liouville system.
Unit 5 Special Functions-II
1. p(a )  0, boundary condition at x  a is dropped, boundary condition at
x  b is homogenous mixed;

2. p(a )  0, boundary condition at x  b is dropped, boundary condition at

x  a is homogenous mixed;

3. p(b)  p(a)  0, and no boundary condition; and

4. interval [a, b] is infinite.

You should note that:

1. If p(a)  0 and there is no boundary condition at x  a, then y is
considered a solution if y (a)  . This is also true for the other cases
listed at 2, 3, and 4 above.

2. If the interval is infinite, then y can be a solution only if it is square

integrable.

You must remember that:

A non-zero solution y of the Sturm-Liouville problem [Eq. (5.29b)] along with

the boundary conditions, is called an eigenfunction, and the corresponding
value of  is called its eigenvalue. The eigenvalues of a Sturm-Liouville
problem are the values of  for which non-zero solutions exist.

We can talk about eigenvalues and eigenfunctions for regular or singular

problems. Our goal is to determine all values of  for which a nontrivial
solution y exists. Let us now consider an example of a Sturm-Liouville problem
and convert a given equation into Sturm-Liouville equation.

Example 5.3

a) Determine whether the following ODEs are Sturm-Liouville equations and

whether the given boundary conditions together with the Sturm-Liouville
equations constitute the Sturm-Liouville problem:
i) y   my  0, 0  x  a,

y (0)  y (a)  0,

ii) x 2y   xy   (m2 x 2  n2 )y  0, 0  x  c,

y (0 )  0

b) Convert Legendre’s differential equation (4.1) into a Sturm-Liouville

problem.

Solution : a) i) Let us check if the ODE with the given boundary conditions
is a Sturm-Liouville equation. Comparing with Eq. (5.29b), we note
that in this ODE:
p( x )  r ( x )  1 and q( x )  0

So, p, p, q and r being constants are continuous and p, r > 0 on the
open interval 0  x  a. Therefore, its solutions exist and it is a Sturm-
Liouville problem. 123
Block 1 Mathematical Methods in Physics
ii) Dividing the ODE by x, we can write it as:
n2
xy   y   (x  )y  0, 0  x  c,
x
p( x )  r ( x )  x and q( x )  0

Comparing it with Eq. (5.29b), we note that:

n2
p( x )  r ( x )  x and q( x )  
x
Note that at x  0, p(x ) and r (x ) are not positive. The function q (x )
diverges at x  0, which means that it is not continuous at x  0.
Further, the boundary condition at x  c is not given. So, this is not a
regular Sturm-Liouville problem.
b) Legendre’s differential equation is:
(1  x 2 )y   2xy   n(n  1)y  0

Note that we can write:

(1  x 2 ) y   2xy   [(1  x 2 ) y ] :

Comparing the above equation with Eq. (5.29b), we note that:

p( x )  1  x 2 , q( x )  0, r ( x )  1,
and the parameter  is n(n  1). So, the Sturm-Liouville form of Legendre’s
equation is:
[(1  x 2 ) y ]  y  0 with   n(n  1).

We will now state a few properties of the Sturm-Liouville system of equations.

Properties of the Sturm Liouville System
We can get a lot of information about the eigen values and eigenfunctions
without actually solving the Sturm-Liouville differential equation simply by
virtue of its properties such as the following:
1. The eigen values are always real and bounded below but not above. So,
the eigen values form an increasing sequence of real numbers
 1   2   3   4  ... with lim  n  .
n 

2. If the interval [a, b] is finite, then eigen values are discrete.

3. The eigenfunction y n corresponding to  n is unique (up to a scalar
multiple), and has exactly n − 1 zeros in the interval a < x < b.
4. Eigen functions are oscillatory in nature.
5. Suppose that y m and y n are eigenfunctions corresponding to distinct
eigenvalues  m and  n , then y m and y n are orthogonal on [a, b] with
respect to the weight function r(x):
b

 y m ( x )y n ( x ) r ( x )dx  0
a

124 This is the orthogonality property of the Sturm-Liouville system.

Unit 5 Special Functions-II
Let us take up an example to illustrate the above properties.

Example 5.4

Determine the eigenvalues of the regular Sturm-Liouville problem:

y   my  0, 0  x  a,

y (0)  y (a)  0,

Do you recognise this equation? This is the ODE separated for the space part
of the one-dimensional wave equation and you know that non-zero solutions
occur only for:

n 2 2
n  (eigenvalues)
a2

The unique eigenfunctions corresponding to these eigenvalues are:

nx
y n  sin
a
for n  1, 2, 3, ... You also know that these eigenfunctions satisfy the
orthogonality property:
a
mx nx
 sin a
sin
a
dx  0 (5.30)
0

since r ( x )  1 for this problem.

Let us now discuss how to expand a given function in terms of orthogonal

functions.

5.5 EXPANSION IN ORTHOGONAL FUNCTIONS

You have already learnt how to expand a given function in terms of Legendre
polynomials in Unit 4. Let us now generalise this method for any given set of
orthogonal functions.
Suppose, the functions y n (x ) form a set of orthogonal functions on an interval
[a, b] that satisfy Eq. (5.30). Then we can expand any given function f(x) in the
same interval in terms of the set of functions y n (x ) as follows:

f (x)   cn y n (x)
n

where the coefficients c n are determined from the orthogonality property of the
functions y n (x ) [Eq. (5.30)] as:
b
1
cn 
Nn  f ( x )y m ( x ) dx
a

where N n is the normalization constant given by:

b

 y n ( x )y m ( x ) dx  N n  mn
a 125
Block 1 Mathematical Methods in Physics
This is how we can determine c n and expand a given function in terms of
orthogonal functions.
We will now summarise what you have learnt in this unit.

5.6 SUMMARY
In this unit, we have covered the following concepts:
 Hermite polynomials, their generating function, recurrence relations and
orthogonality relations.
 Laguerre polynomials, their generating function, recurrence relations and
orthogonality relations.
 Sturm-Liouville problem, properties of Sturm-Liouville problem and its
solution.
 Expansion of any given function in terms of orthogonal functions.

5.7 TERMINAL QUESTIONS

1. Two operators
1  d  1  d 
a    and a    
2 d  2 d 
m
where   x operate on the harmonic oscillator wave function

1
 2
 n  Nn e 2
H n 
1/ 2
 m 1/ 2 
N n  
1
where  
   2 n n! 1/ 2 

Show that a  n  n  n 1 and a  n  n  1  n 1

and (aa  aa)  n   n

2. The ‘zero’ line in the fundamental band of the near infrared absorption
spectrum of HCI35 gas occurs at 3.46 × 10–6m. This corresponds to a
transition from a state with vibrational quantum number zero to a state with
quantum number one. Calculate the force constant for HCI bond assuming
harmonic oscillator potential.
3. Expand the following function in terms of Legendre polynomials:
0,  1  x  0,

f (x)  
1 0 x 1

5.8 SOLUTIONS AND ANSWERS

Self-Assessment Questions
1. Solve and obtain its general solution given by Eq. (5.2).
d 2y dy
 2x  2ny ( x )  
126 dx 2 dx
Unit 5 Special Functions-II
Since x  0 is an ordinary point of Hermite’s differential equation
[Eq. (5.1)], its solution in the form of a power series in x is given by

y a j x j (i)
j 0

with
2(n  j )
a j 2   aj (ii)
( j  1) ( j  2)

You should verify Eq. (ii) on your own. Eq. (ii) tells us that for even positive
integral value of j, the coefficients a j can be expressed in terms of a0 and
for odd positive integral values of (j > 1) the coefficients can be expressed
in terms of a1 :
2n
a2   a0
12

2(n  2) ( 2) 2 n(n  2)

a4    a2   a0
34 1 2  3  4

2(n  1)
and a3   a1
23
(n  3) (2) 2 n(n  1)(n  3)
a5   a3  a1
45 23 45
Substituting these results in Eqs. (i) above, we obtain Eq. (5.2).
2. Differentiating both sides of the generating function for Hermite
polynomials partially with respect to x, we get:

tn
2t e 2 xt t 
2
 H n ( x )
n!
n 0

We can rewrite it as:

 
tn tn
2t  
Hn (x)
n! n 0
H n ( x )
n!

n 0
Equating coefficients of t n 1 from both sides, we get:
H ( x ) H n 1( x )
2 n 
n! (n  1)!

or
Hn 1( x )  2(n  1) Hn ( x )

3. According to Rodrigues’ formula:

H 4 ( x )  ( 1) 4 e x
2 d4
dx 4

e x
2

Since
dx
e 
d  x2
 2xe  x2

d 2  x2
dx 2
e  
 (2x ) 2 e  x  2e  x
2 2
127
Block 1 Mathematical Methods in Physics
d3
3
2 2

e  x  (2x )3 e  x  8 x e  x  4 x e  x
2 2

dx
 (2x ) 3 e  x  12x e  x
2 2

d4
e x   (2x ) 4 e x
2 2
 24 x 2 e  x  24x 2 e  x  12e  x
2 2 2

dx 4
 H 4 ( x )  16x 4  48x 2  12

4. From the expression for the generating function of Hermite polynomials,

we can write:


 x e x e 2 xt t  2 xu u dx
2 2 2



  
t nu m
    x e x 2
H n ( x ) H m ( x ) dx
n! m!
(i)
n 0 m 0  

Following the steps used in arriving at Eq. (5.14), we can rewrite the left
hand side as:

2
I e 2tu
 xe ( x t u ) dx


We now change the variable of integration and put z  x  t  u . Then the

above integral takes the form:

I  e 2tu  ( z  t  u ) e z
2
dz



 e 2tu 0  t   u      (t  u )
(2tu ) n
n!
n 0

   t u  t n u n 1 
2 n n 1 n
(ii)
n 0
n !

On equating the coefficients of t n u m of the series in (ii) with the

corresponding coefficients from the right hand side of (i), we obtain the
required result:


 x e x H n ( x ) H m ( x ) dx  2 n  (n  1)! if m = n + 1
2



 2n 1  n! if m = n – 1
0 otherwise

5. Show that the Laguerre polynomial defined in Eq. (5.20) is identical with
that given in Eq. (5.18).
Recall Leibnitz rule for the nth derivative of the product of two polynomials
u(x) and v(x):
dn d nu d n 1u dv d nv
(uv )  v  n C1  ...  n Cn u
128 dx n dx n dx n 1 dx dx n
Unit 5 Special Functions-II
Now we choose u( x )  x n and v ( x )  e  x . Then using Leibnitz rule, we
can write:
ex d n
x nen  
ex 

n! n!
n! e  x  n C1 x( )e  x  n C2 x 2 ( 1) 2 e  x
n! dx n n!  1! 2!

 ...  n Cn x n ( 1) n e  x 

x x2 xn
 1  n C1  n C2  ...(1) n n Cn
1! 2! n!
which is Eq. (5.18).
Terminal Questions
d n 2 dH n 
1.  N n e (1/ 2)    H n   
d  d 

 N n e (1/ 2)   H n ()  2n H n 1()

2

1  d 
 a n     n
2 d 
N
 n e (1 / 2) 2n H n1()
2

2
1/ 2
 m 1/ 2 
2n  
Nn 1
But   2n
2    2 n n! 1/ 2 

1/ 2
 m 1/ 2 
 
1
  n
   2 n 1(n  1)! 1/ 2 

a n  n N n 1e (1/ 2) H n 1()  n  n 1

2
Hence,

1  d 
Further, a  n      n
2 d 

Nn
 e (1/ 2) 2H n ()  2n H n 1()
2

2
Nn
 e (1/ 2) H n 1()
2

1  m 1/ 2 1 
 n 1  e (1/ 2) H n 1()
2
 
2 n  1    2 n! 
n 1/ 2


 n  1 N n 1e (1/ 2) H n 1()  n  1  n 1

2

The operators a and a are called raising and lowering operators (step up
and step down operators) or (in the context of quantum field theory)
creation and annihilation operators. Again,

aa   a  a n  a n  1  n 1  a  n  n 1

 n  1 n  1  n  n n  n  (n  1  n )  n

 n 129
Block 1 Mathematical Methods in Physics
2. Let  be the frequency of the ‘zero’ line. Then
c 3  10 8
  Hz
 3.46  10 6
Also
k
h  E n 1  E n 0    

2
k  2  3
   1014 
  3.46 

where k is the force constant and  is the reduced mass of H and Cl 35 ,

i.e.
1 35
  1.66  10 27 kg
1  35

6  2
 1.66  
35
Then k   10 Nm 1  479 Nm 1
36  3.46 

3. We put:

f (x)   an Pn ( x ) (i)
n 0

We have to determine the coefficients a n . For this, we multiply both sides

of Eq. (i) by Pm (x ) and integrate from  1 to  1. You know that Legendre
polynomials are orthogonal, all integrals on the RHS are zero except the
one containing a m . So, we use the orthogonality relation for Legendre
polynomials and get:

 an 1Pn ( x ) Pm ( x )dx  c m . 2m  1
1 1 2
1 f ( x )Pm ( x )dx 
n 0

Using this result, we get:

1 1
1f ( x )P0 ( x )dx  c0 1[P0 ( x )]2 dx
1 1
or 0 dx  c 0 .2, c0  ;
2
1 1
1f ( x )P1( x )dx  c1 1[P1( x )]2 dx
1 2 3
or 0 xdx  c1. 3 , c1 
4
;

1 1
1 f ( x )P2 ( x )dx  c 2 1[P2 ( x )]2 dx
1 3 1 2
or 0  2 x 2  2 dx  c 2 . 5 , c 2  0.

1 3 7 11
 f (x)  P0 ( x )  P1( x )  P3 ( x )  P5 ( x )  ...
2 4 16 32

130