MATHEMATICS - III

BITS Pilani Prof. Manoj Kumar Pandey

Department of Mathematics
K K Birla Goa Campus

BITS Pilani, K K Birla Goa Campus

Power Series Solutions
and
Special Functions

BITS Pilani, K K Birla Goa Campus

Power Series Solutions

 Finding a general linear differential equation solution depends on

determining a fundamental set of solutions of the homogeneous
equation.

 We have established a systematic method for generating

fundamental solutions in cases where the equation features
constant coefficients.

 However, our approach must expand when dealing with a

broader range of equations characterized by variable coefficients.
In such instances, we must explore solutions that go beyond the
realm of conventional elementary functions in calculus.
BITS Pilani, K K Birla Goa Campus
Power Series Solutions

 A key strategy is to express a given function using a power series

representation. This tool enables us to tackle equations with
varying coefficients more comprehensively, paving the way for a
deeper understanding of their solutions.

BITS Pilani, K K Birla Goa Campus

Elementary and Special Functions
Elementary Functions: The elementary functions consist
of algebraic functions, rational functions, transcendental
functions that are trigonometric and inverse trigonometric
functions, exponential, logarithms functions, any
combination (adding, subtracting, multiplying, dividing ) or
compositions (forming function of function) of these
functions.

Beyond these functions, there exists another class of functions

called Special functions.
Examples: Bessel functions, Gamma functions, Zeta functions,
etc.
BITS Pilani, K K Birla Goa Campus
 A large class of special functions arise as a solution of second-
order linear ODEs.

 If we wish to solve y  y  0 then the familiar functions

y  sin x and y  cos x are already available for this purpose
in elementary calculus.

 But the situation is different in the case of

xy   y   xy  0

BITS Pilani, K K Birla Goa Campus

Purpose

 Explore the solution of second-order linear equations featuring

variable coefficients through power series analysis.

 Utilize these series to establish special functions.

BITS Pilani, K K Birla Goa Campus

Power Series

 An infinite series of the form



 n
a
n 0
x n
 a 0  a1 x  a 2 x 2


is called a power series in x.

 The series 

 na (
n 0
x  x0 ) n
 a0  a1 ( x  x0 )  a 2 ( x  x0 ) 2


is called a power series in x  x0 .

The second series can be transformed into the first by translating or

changing the coordinate system.
BITS Pilani, K K Birla Goa Campus
We shall confine our discussion to the power series of the
form 

a
n 0
n x n  a0  a1 x  a 2 x 2   (1)

The series (1) is said to converge at a point x if the limit

m
lim
m

n0
an x n

exists, and in this case, the sum of the series is the value of
this limit. The series (1) always converges at x = 0.

BITS Pilani, K K Birla Goa Campus

All power series in x fall into one or another of three
major categories



n0
n ! x n  1  x  2 x 2  3! x 3  

 n 2 3
x x x

n0 n!
 1 x 
2

3!


 x
n 0
n
1  x  x 2
 x 3


BITS Pilani, K K Birla Goa Campus

• The first series converges only for x=0 and diverges for
all x  0.

• The second series converges for all x.

• The third series converges for x < 1 and diverges for
x >1

BITS Pilani, K K Birla Goa Campus

Remark: The first is of no use, and the second is the most
straightforward to work with all others are roughly similar
to the third series in the sense that to each series of this
kind, there exists a number R, called the radius of
convergence such that the series converges for x < R and
diverges for x >R.

BITS Pilani, K K Birla Goa Campus

How to find Radius of Convergence (R)
In many important cases, the value of R can be found as
follows. Let 
u
n 0
n  u 0  u1  u 2  

be a series of non-zero constants. We recall by elementary

calculus that if the limits (Ratio test)
u n 1
lim  L
n  un

exist, then the ratio test asserts that the series converges if
L< 1 and diverges L >1.
BITS Pilani, K K Birla Goa Campus
In the case of power series
a n 1 x n 1 a n 1
lim n
 lim xL
n  an x n a
n

this series will converge if L< 1 and diverges L >1. These

considerations yield the formula for the radius of
convergence
an
R  lim
n  a
n 1

BITS Pilani, K K Birla Goa Campus

Example:

BITS Pilani, K K Birla Goa Campus

Remark: Endpoints need to be checked separately.
The series converges at x=15/8 but diverges at x=17/8. (Check)
BITS Pilani, K K Birla Goa Campus
Problem:

BITS Pilani, K K Birla Goa Campus

Suppose that the power series


 n
a
n 0
x n
 a 0  a1 x  a 2 x 2


converges for x < R, R > 0, and denote its sum by f (x)

f ( x )   an x n  a0  a1 x  a2 x 2  
n0

Then f (x) is continuous and has derivatives of all orders

for x < R (Abel’s theorem). Also, the series can be
differentiated term by term in the sense that

f ( x )   na n x n 1  a1  2 a 2 x  3a3 x 2  
n 1

BITS Pilani, K K Birla Goa Campus

 
f ( x )   n ( n  1) a n x n  2  2 a2  3  2 a3 x  
n2

And so on, and each of the resulting series converges for

x < R.

Let us have a second power series in x that converges to a

function g(x) for x < R so that

g ( x )   bn x n  b0  b1 x  b2 x 2  
n0

then,
BITS Pilani, K K Birla Goa Campus
 
f ( x )  g ( x )   ( an  bn ) x n  ( a0  b0 )  ( a1  b0 ) x  
n0


f ( x ) g ( x )   cn x n , where cn  a0 bn  a1bn 1    a n b0
n 0

That is they can be added subtracted multiplied as if they

were polynomials.

BITS Pilani, K K Birla Goa Campus

Series Solution of First-Order Equations
We consider the equation
dy
 y,  (* )
dx

We assume that this equation has a power series solution

of the form
y  a0  a1 x  a2 x   an x  
2 n
(**)

That converges for x < R, R> 0; that is, we assume that
(*) has the solution that is analytic at the origin.
BITS Pilani, K K Birla Goa Campus
A power series can be differentiated term by term in its
interval of convergence
y   a1  2a2 x  3a3 x    (n  1) an 1 x   (***)
2 n


  n
na
n 1
x n 1

Since
dy
 y
dx

these two series must have the same coefficients

BITS Pilani, K K Birla Goa Campus

 a 1  a 0 , 2 a 2  a1 , 3 a 3  a 2 ,  , ( n  1) a n  1  a n , 

Hence, we have the following solution

a0 2 a0 3 a0 n
y  a0  a0 x  x  x  x 
2! 3! n!
 x2 x3 xn 
 y  a0  1  x    
 2! 3! n! 
where a0 is an arbitrary constant.
BITS Pilani, K K Birla Goa Campus
Remark: We need to validate this solution by showing
that the above series has a positive radius of convergence.
The ratio test says that the above series converges for
every x; hence, it is the desired solution.

BITS Pilani, K K Birla Goa Campus

Finding Power Series Expansion

The method of power series is a useful method of finding

power series expansion of many functions.

Example: Find the power series expansion of

(1  x) p
and write the region of convergence.

BITS Pilani, K K Birla Goa Campus

Second-Order Linear Equations

Let us consider the general second order linear equation

d2y dy
2
 P (x)  Q (x) y  0  (1)
dx dx
it is occasionally possible to solve such an equation in
terms of familiar elementary functions.

Remark: The behaviour of its solution near a point x0

depends on the behaviour of its coefficient functions P(x)
and Q(x) near this point.

BITS Pilani, K K Birla Goa Campus

Ordinary Point: A point x0 is called an ordinary point of
(1) if P(x) and Q(x) are analytic at x0, which means that
each of these functions has a power series expansion valid
in some neighborhood of this point.

If x0 is an ordinary point of (1) then the solution of (1) is

also analytic at x0.

If x0 is not an ordinary point, then it is a singular point of

(1).

BITS Pilani, K K Birla Goa Campus

Examples:

(i) y  y  0
(ii) (1  x ) y  2xy  p( p  1) y  0, p isa constant
2

BITS Pilani, K K Birla Goa Campus

Solution

is constant

Since x = 0 is an ordinary point of the differential equation,

we expect a solution of the form


y   an x n , x  R , R  0
n0

BITS Pilani, K K Birla Goa Campus

 
 y  
n  0
(n  1)a n 1 x n
,

y   
n  0
(n  2 )(n  1)a n  2 x n
,

 2 xy  
n  0
2na n x n
,

 x 2
y   
n  0
 (n  1)n a n x n
,

p ( p  1) y  
n  0
p ( p  1)a n x n
,

BITS Pilani, K K Birla Goa Campus

Substituting these in the differential equation, we get the
following Recursion formula (by comparing the coefficient
of the general term)

( p  n)( p  n  1)
an  2  an , n  0,1, 2, 
( n  1)( n  2)

BITS Pilani, K K Birla Goa Campus

Hence, by computing different coefficients, we get

ya
1  p( p  1) x 2

p( p  2)( p  1)( p  3)
x 
4 
0
 2! 4! 

a
 x  ( p  1)( p  2) x 3

( p  1)( p  3)( p  2)( p  4)
x 
5 
1
 3! 5! 

as our formal solution.

BITS Pilani, K K Birla Goa Campus

When p is not an integer, each series in the bracket has a
radius of convergence R =1, hence this is the general
solution of Legendre’s equation and is called Legendre
functions. In general, these functions are not an elementary
function.

If p is a non-negative integer, one of the series terminates

and is thus a polynomial, while the other remains an
infinite series. This leads to a particular solution of the
Legendre equation known as Legendre polynomials.

BITS Pilani, K K Birla Goa Campus

Theorem:
Let x0 be an ordinary point of the differential equation
d2y dy
2
 P (x)  Q (x) y  0
dx dx
and let a0 and a1 be arbitrary constants. Then, there exists a unique
function y(x) that is analytic at x0 , is a solution of the above
equation in a certain neighborhood of this point and satisfies the
initial conditions y(x0) = a0 and y  ( x 0 )  a 1 . Furthermore, if the
power series expansions of P(x) and Q(x) are valid on an interval,
|x-x0| <R, R>0, then the power series expansion of this solution is
also valid on the same interval.
BITS Pilani, K K Birla Goa Campus
Examples

(1) Find the general solution of the following equation in

terms of power in x

(i) (1  x ) y  2 xy  2 y  0,

2

(ii) y  2 xy  2 py  0, p is a constant

(2) Find the general solution of the following equation in

terms of power in (x-1)

BITS Pilani, K K Birla Goa Campus

 Method of Frobenius
(Frobenius series Solutions)

BITS Pilani, K K Birla Goa Campus

Regular Singular Points

BITS Pilani, K K Birla Goa Campus

 For Rational Functions
If P(x), Q(x), and R(x) are polynomials, then a point x0 is
called a regular singular point if and only if the following
limits exist and are finite.

BITS Pilani, K K Birla Goa Campus

Example

BITS Pilani, K K Birla Goa Campus

Examples:

BITS Pilani, K K Birla Goa Campus

 Example

Both the limits exist.

BITS Pilani, K K Birla Goa Campus

Example

BITS Pilani, K K Birla Goa Campus

 Method of Frobenius

Frobenius series Solutions: Solution near the

regular singular points.

Let us consider the differential equation

BITS Pilani, K K Birla Goa Campus

The type of solution we are aiming at for (1) near a
regular singular point is a “quasi-power series” of
the form:

BITS Pilani, K K Birla Goa Campus

 Motivation

The motivation comes from the Euler-Cauchy

Equation:

This equation has a solution of the form:

A Frobenius series solution is motivated by the
above equation. The Euler-Cauchy equation is the
most straightforward, with x=0 as a regular singular
point.
BITS Pilani, K K Birla Goa Campus
A most general second-order equation with x=0 as a
regular singular point is:

Replacing p and q with the power series. Then it is

natural to guess that the corresponding solution
might be accomplished by taking a solution of the
form:

BITS Pilani, K K Birla Goa Campus

 = ( )

≠ 0.

Solutions of this kind are called the Frobenius series

solution, and the method is called the Method of
Frobenius.

BITS Pilani, K K Birla Goa Campus

 Example

Find the general solution of the following equation

near x=0

BITS Pilani, K K Birla Goa Campus

 Frobenius Series Solution

Let us consider a general second-order differential

equation

near the regular singular point x=0.

BITS Pilani, K K Birla Goa Campus

We assume a solution of the form:

= ( ), ≠ 0.

(2)

Since x=0 is a regular singular point, xP(x) and

are analytic at x=0.

BITS Pilani, K K Birla Goa Campus

 and

We have the solution:

BITS Pilani, K K Birla Goa Campus

BITS Pilani, K K Birla Goa Campus
BITS Pilani, K K Birla Goa Campus
Substituting these in the equation (1) and comparing the
respective coefficients of x we get:

BITS Pilani, K K Birla Goa Campus

Let f(m) = then

BITS Pilani, K K Birla Goa Campus

since ≠ 0, we have f(m) =0, that is

f(m) =

the above equation is called the indicial, and its roots,

say are the exponents of the differential
equation (1) at the regular singular point x=0.

Now, we have the following theorem.

BITS Pilani, K K Birla Goa Campus

 Theorem
Assume that x = 0 is a regular singular point of the
differential equation
y   P ( x )  Q ( x ) y  0
and that the power series expansions of xP(x) and x2Q(x)
are valid on an interval |x| < R with R > 0. Let the indicial
equation
m(m  1)  mp0  q0  0.
having roots m1 and m2 with m2  m1 . Then, the equation
has at least one solution

BITS Pilani, K K Birla Goa Campus

on the interval 0 < x < R, where an are determined in terms of a0 by
the recursion formula
n1
an (m  n)(m  n 1)  (m  n) p0  q0   ak (m  k ) pnk  qnk   0
k 0


with m replaced by m1, and the series

 n
a
n 0
x n

converges for |x| < R . Furthermore, if m1 - m2 is not zero or a

positive integer, then the equation has a second independent solution

BITS Pilani, K K Birla Goa Campus

on the same interval, where, in this case, the an are determined
in terms of a0 by the recursion formula with m replaced by
m2, and again the series



n0
an xn

converges for |x| < R.

BITS Pilani, K K Birla Goa Campus

 Following Cases
The theorem fails to answer the question of how to find the
second linearly independent solution when the difference
m1 - m2 is zero or a positive integer.

We have the following three cases:

CASE A: If m1 = m2, there can not exist a second Frobenius
series solution.
For the other two cases, we see the formula replacing m by
m2

BITS Pilani, K K Birla Goa Campus

In the other two cases, in both of which m1 - m2 is a positive
integer will be easier to understand if we insert m = m2, in
the recursion formula to get

an f (m2  n)  a0 (m2 pn  qn )    an 1[(m2  n  1) p1  q1 ]

Difficulty in calculating an arises because f(m2+n)=0

for some positive integer n.

BITS Pilani, K K Birla Goa Campus

CASE B: If the right side of the previous equation is not
zero when f(m2+n) = 0, then there is no possible way of
continuing the calculation of the coefficients, and there
cannot exist a second Frobenius series solution.

CASE C: If the right side of the equation happens to be

zero when f(m2+n) = 0, then an is unrestricted and can be
assigned any value. Hence, in this case, a second
Frobenius series solution exists.

BITS Pilani, K K Birla Goa Campus

 Examples

Find the Frobenius series solutions for the following

differential equations:

BITS Pilani, K K Birla Goa Campus

Problem-1

BITS Pilani, K K Birla Goa Campus

Problem-2

BITS Pilani, K K Birla Goa Campus