Power Series Solutions

UNIT 2 POWER SERIES SOLUTIONS

Structure Page No.
2.1 Introduction 43
Objectives
2.2 Power Series and Radius of Convergence 44
2.3 Ordinary and Singular Points of an Equation 47
2.4 Series Solution Methods 49
Power Series Solution about an Ordinary Point
Series Solution about a Regular Singular Point: Frobenius Method
Series Solution about a Point at Infinity
2.5 Summary 67
2.6 Solutions/Answers 68

2.1 INTRODUCTION
In Unit 1, we recalled the methods of finding explicit solutions of linear differential
equations with constant coefficients and Euler-Cauchy equations which are reducible
to equations with constant coefficients by a change of independent variable and
illustrated these methods through various examples. The class of functions, appearing
as solution of these equations involved elementary functions, i.e., exponential,
trigonometric, logarithmic and polynomial functions, all of which can be written in
terms of exponential functions. In many engineering, mathematical and physical
problems, the equations governing the physical situations turn out to be homogeneous,
linear differential equations with variable coefficients. Solutions of such equations
cannot be expressed in terms of elementary functions. But since elementary functions
can be expressed in terms of power series; this gives us an idea that for a general
homogeneous, linear differential equation with variable coefficients, we might try a
function defined by a general power series
∞
y= ∑a
n =0
n (x − x 0 ) n

as a solution, where the series on the right being meaningful only when it is
convergent. We shall see that this gives a powerful method of solving linear
homogeneous differential equations with variable coefficients and leads to the
introduction of some new ‘standard functions’, commonly known as ‘higher
transcendental functions’ or ‘special functions’ such as Legendre, Bessel,
Hypergeometric, Hermite functions etc. The appearance of these special functions as
solutions of differential equations describing various phenomena in physics and
engineering indicates that these functions are as ‘standard’ as our earlier ‘elementary
functions’. In this unit we shall discuss methods of finding power series solutions of
linear homogeneous differential equations with variable coefficients.
In Sec.2.2, we have started the discussion with the introduction of power series and its
radius of convergence. We have discussed ordinary and singular points of a
differential equation in Sec.2.3. In Sec.2.4 we have discussed power series solutions
of a differential equation about an ordinary point and series solutions about a regular
singular point – in particular, the Frobenius method. We shall be discussing the
applications of these power series methods to various physical situations, in Units 3
and 4.
Objectives
After studying this unit you should be able to
• describe the need for a series solution of homogeneous linear differential equation
with variable coefficients;
• define the radius of convergence and interval of convergence, of a power series;
• distinguish between an ordinary point, a regular singular point and an irregular
singular point of a differential equation;
43
Ordinary Differential • obtain a power series solution of a given differential equation about an ordinary
Equations point;
• obtain a series solution of a given differential equation about a regular singular
point using Frobenius method.

2.2 POWER SERIES AND RADIUS OF

CONVERGENCE
You are already familiar with the series of constant terms i.e., series of the form,
∞

∑ n =1+ 2 + 3 + L
n =1
∞
1 1 1 1
∑2
n =1
n
=
+ + +L
2 4 8
If the terms of a series are variable, say, function of a variable x , then they assume
definite values when x is given a fixed value. The basic idea of the power series
method for solving differential equations, which is our concern in this unit, is very
simple and natural. Given a differential equation, we represent all the functions in the
equation by a power series. This process involves various operations like
differentiation, addition and multiplication of power series. Therefore, to justify the
method let us first consider the theoretical basis of these operations.
A power series in powers of ( x − a ) is an infinite series of the form
∞

∑a
n =0
n (x − a ) n = a 0 + a 1 (x − a ) + a 2 (x − a ) 2 + L (1)

where x is a variable, a 0 , a 1 K are constants, called the coefficients, and a is a

constant, called the centre of the series. For discussion in this unit we shall assume
that all variables and constants are real.
In particular, if a = 0 then we get a power series in powers of x as
∞

∑a
n =0
n x n = a 0 + a1 x + a 2 x 2 + L (2)

Remark: The term ‘power series’ usually refers to series of the form (1), including the
case (2), but does not include series of negative powers of x such as
c1 x −1 + c 2 x −2 + L or, series involving fractional powers of x .
The convergence behaviour of a power series can be characterized in a simple way.
We know from d’Alembert’s ratio test for absolute convergence of a series that series
(2) is absolutely convergent or divergent according as
| anxn | an 1
lim n +1
= lim
n →∞ | a
n +1 x | n →∞ a n +1 x

is greater or less than one, provided the above limit exists.

an
Let R = lim , (3)
n →∞ a n +1
then subject to the existence of this limit, series (2) converges absolutely if
R
> 1, x ≠ 0 , i.e., | x | < R
|x|
R
and diverges if , < 1, i.e., | x | > R
|x|
Thus the region of convergence is the interval − R < x < R within which the series (2)
is absolutely convergent and outside this interval it is divergent. Accordingly, series
(1) converges in the interval.
a −R< x < a +R (4)
44
 Power Series Solutions
with centre x = a . The number R is known as the radius of convergence of series
(1) and interval (4) is known as its interval of convergence. Here R is always non-
negative and may assume the value 0 or ∞ , or, any value in between.
Usually we put R equal to 0 when the series converges only for x = 0 and equal to
∞ when it converges for all x . This convention allows us to cover all possibilities in
a single statement: each power series in x has a radius of convergence R , where
0 ≤ R ≤ ∞ with the property that the series converges if | x | < R and diverges if
| x | > R . You may note that if R = 0 then no x satisfies | x | < R , and if R = ∞ then
no x satisfies | x | > R . We can thus say that all power series in x fall into one, or,
another, or three major categories which are typified by the following examples
∞

∑ n! x
n =0
n
= 1 + x + 2! x 2 + 3! x 3 + L (5)
∞
xn x2 x3
∑ n! = 1 + x +
n =0
+
2! 3!
+L (6)

∑x
n =0
n
= 1+ x + x2 + x3 + L (7)

Series(5) diverges (i.e. fails to converge) for all x ≠ 0 . Series (6) converges for all x
and series (7) converges for | x | < 1 and diverges for | x | > 1 . How R is calculated for
the above three series will be clear to you after going through the following examples.

Example 1: Find the radius of convergence of the series

∞

∑n
n =0
n
x n = 1 + x + 2 2 x 2 + 33 x 3 + L

Solution: The given series has radius of convergence R given by

nn 1
R = lim = lim =0
n →∞ (n + 1) n +1 n →∞
 1
n
(n + 1) 1 + 
 n
Hence the series does not converge for any non-zero x. However, it obviously
converges for x = 0 .
***
∞
Example 2: Find the interval of convergence of the series ∑n x
n =0
n
.

Solution: The radius of convergence of the given series is

n 1
R = lim = lim
=1
n →∞ 1
n + 1 n →∞
1+
n
Thus the given series converges for | x | < 1 , i.e., − 1 < x < 1 and diverges for | x | > 1 .

***
Let us now suppose that series (2) converges for | x | < R with R > 0 , and denote its
sum by f ( x ) then ,
∞
f (x) = ∑a
n =0
n x n = a 0 + a1 x + a 2 x 2 + L (8)

Now f ( x ) is obviously continuous and has derivatives of all orders for | x | < R . Also
the series can be differentiated termwise in the sense that
∞
f ′( x ) = ∑na n =1
n x n −1 = a 1 + 2a 2 x + 3a 3 x 2 + L

45
Ordinary Differential ∞
Equations f ′′( x ) = ∑ n (n − 1)a
n =2
nx
n −2
= 2a 2 + 3 2a 2 x + L

and so on, and each of the resulting series converges for | x | < R . The formula linking
the a n to f ( x ) and its derivatives can be obtained from these successively
differentiated series as follows:
f ( n ) ( 0)
an = (9)
n!
Just as we have differentiated series (8) it can also be integrated termwise provided the
limits of integration lie inside the interval of convergence. Further, if we have a
second power series in x that converges to a function g ( x ) for | x | < R , so that
∞
g(x ) = ∑bn =0
nx
n
= b 0 + b1 x + b 2 x 2 + L (10)

then series (8) and (10) an be added or subtracted termwise as

∞
f ( x ) ± g(x ) = ∑ (a
n =0
n ± b n ) x n = (a 0 ± b 0 ) + (a 1 ± b 1 ) x + L

They can also be multiplied like polynomials, such that

∞
f ( x )g ( x ) = ∑c
n =0
nx
n

where c n = a 0 b n + a 1 b n −1 + L + a n b 0 . If series (8) and (10) converge to the same

functions, so that f ( x ) = g ( x ) for | x | < R , then according to formula (8) they have the
same coefficients: a 0 = b 0 , a 1 = b 1 , K

In particular, if f ( x ) = 0 for | x | < R then a 0 = 0, a 1 = 0, K .

Now let us suppose that f ( x ) be a continuous function that has derivatives of all orders
for | x | < R with R > 0 . Then you may wonder whether f ( x ) can be represented by a
power series or not. If we use formula (9) to define the a n then we can hope that the
expansion
∞
f ( n ) ( 0) n f ′′(0) 2
f (x ) = ∑
n =0 n!
x = f (0) + f ′(0) x +
2!
x +L (11)

will hold throughout the interval. This is often true but sometimes it is false. To
check the validity of this expansion for a specific point x in the interval we can use
Taylor’s formula:
n
f ( k ) (0)
f (x) = ∑
k =0 k!
+ R n (x)

where the remainder R n ( x ) is given by

f ( n +1) (ξ) n +1
R n (x ) = x
(n + 1) !

for some point ξ between 0 and x . To verify (11), it suffices to show that
R n ( x ) → 0 as n → ∞ . The above procedure can be used quite easily to obtain the
following familiar expansions, which are valid for all x .
∞
xn x2 x3
ex = ∑
n =0 n!
=1+ x +
2!
+
3!
+L (12)

∞
x 2 n +1 x3 x5
sin x = ∑ n =0
(−1) n
(2n + 1)!
=x− +
31 5!
L (13)

∞
x 2n x2 x4
cos x = ∑
n =0
(−1) n
(2n )!
=1−
2!
+
4!
L (14)
46
 Power Series Solutions
You may now try the following exercises.
E1) Verify that R = 0, R = ∞ and R = 1 , for the series (5), (6) and (7), respectively.
xn ∞
E2) Discuss the convergence of the series ∑ .
n = 0 n!

E3) Obtain the radius of convergence of the following series

(a) 1 − 2 x + 3x 2 − 4 x 3 + L
∞
( kn )! n
(b) ∑
n = 0 ( n! )
k
x , k is a positive integer.

As we have mentioned earlier our aim here is to find the general solution of a linear
homogeneous differential equation with variable coefficients. In particular, we shall
try to find the solution of the equation
d2y dy
a 0 (x) 2 + a 1 (x) + a 2 ( x ) y = 0, a 0 ( x ) ≠ 0.
dx dx
or,
d2y dy
2
+ P( x ) + Q( x ) y = 0 (15)
dx dx
a (x) a (x)
where, P( x ) = 1 , Q( x ) = 2 and a 0 ( x ), a 1 ( x ) and a 2 ( x ) are continuous
a 0 (x) a 0 (x)
functions of x on some interval I . You have studied some properties of Eqn.(15) in
Unit 1 when a 0 , a 1 , a 2 are constants and obtained solutions of such equations in terms
of elementary functions. But for equations with variable coefficients when solutions
cannot be obtained in terms of these elementary functions, power series solutions
come to our rescue. For studying the power series method we need to study some
definitions, which we shall take up next.

2.3 ORDINARY AND SINGULAR POINTS OF AN

EQUATION
Here we start with recalling the definition of an analytic function which you have
studied in your complex analysis course (MMT-005).
Analytic Function: A function f ( x ) with the property that a power series expansion
of the form
∞
f (x ) = ∑ a n (x − a ) n (16)
n =0
is valid in some neighbourhood of the point a is said to be analytic at a. In this case
the a n ' s are given by
f ( n ) (a )
an = , n = 0, 1, 2, K
n!
and Eqn.(16) is called the Taylor series of f ( x ) at x = a . Thus, Eqns. (12), (13) and
(14) show that e x , sin x and cos x are analytic at x = 0 , and the given series are the
Taylor series of these functions at this point. You must also be familiar with the
following facts about analyticity:
i) Polynomials and the functions e x , sin x and cos x are analytic at all points.
ii) If f ( x ) and g ( x ) are analytic at a, then f ( x ) + g ( x ), f ( x ) g ( x ) , and
f ( x ) / g ( x ) [if g (a ) ≠ 0] are also analytic at a.
iii) The sum of a power series is analytic at all points inside the interval of
convergence.
We shall now use this definition to introduce some more definitions which will be of
use for our further discussions. 47
Ordinary Differential Ordinary Point: A point a ∈ I is called an ordinary point of Eqn.(15) if , P ( x ) and
Equations
Q( x ) are analytic at x = a . Ordinary point is also called a regular point of the
equation.

Equation y ′′ − 2xy ′ + 2py = 0 , where p is a constant is called Hermite’s equation and

has an ordinary point at x = 0 , since − 2 x and 2p are analytic at x = 0 .
The point x = 2 is not an ordinary point for the equation ( x − 2) y ′′ + y = 0 , because
1
the function does not admit a power series around 2.
x−2
Singular Point: Any point of I which is not an ordinary point of Eqn.(15) is said to be
a singular point of it. Thus for x = a to be a singular point of Eqn.(15), at least one
of the coefficients P( x ), Q( x ) must fail to be analytic at x = a .
Regular Singular Point: A singular point a of Eqn.(15) viz.,
a (x) a (x)
y ′′ + 1 y′ + 2 y = 0, a 0 ( x ) ≠ 0 ,
a 0 (x) a 0 (x )
a (x) a (x)
is said to be regular singular point, if ( x − a ) 1 and ( x − a ) 2 2 are analytic at
a 0 (x) a 0 (x )
x =a.
In other words, we say that in such cases, the singular points are ‘weak’, in the sense
that the coefficient functions are only mildly non-analytic and, with simple
modifications, can be made analytic and can have a Taylor series expansion about the
singular point.
A singular point, which is not regular singular point is called an irregular singular
point.
As a simple example, the origin is clearly a regular singular point of the equation
2 2
y ′′ + y ′ − 2 y = 0 .
x x
Consider the equation
(1 − x 2 ) y ′′ − 2 xy ′ + p(p + 1) y = 0 ,
where p is a constant. This equation is Legendre’s equation about which we shall be
studying in detail in the next unit. The equation can be put in the form
2x p(p + 1)
y ′′ − 2
y′ + y =0.
1− x 1− x2
It is clear that x = 1 and x = −1 are its singular points. The point x = 1 is regular
because
 2x  2x
( x − 1) P( x ) = ( x − 1) − 2 
= ,
 1− x  x +1

 p(p + 1)  ( x − 1) p(p + 1)
and ( x − 1) 2 Q( x ) = ( x − 1) 2  2 
=− ,
 1− x  x +1
are analytic at x = 1 . For similar reasons x = −1 is also a regular singular point.

As another example, consider Bessel’s equation of order p, where p is a non-negative

constant
x 2 y ′′ + x y ′ + ( x 2 − p 2 ) y = 0 .
Writing the equations in the form
1 x 2 − p2
y ′′ + y ′ + y =0,
x x2
it is apparent that origin is a regular singular point of the equation because
48
 Power Series Solutions
1 x 2 − p2
x P( x ) = x = 1 and x 2 Q( x ) = x 2 , = x 2 − p2 ,
x x2
are analytic at x = 0 .
You may now try the following exercises.
E4) Find the regular and singular points of the following DEs. Also, specify the
nature of singular points.
(a) y ′′ + 4x y ′ + y = 0 .
(b) x 3 ( x − 2) y ′′ + x 3 y ′ + 6 y = 0 .
(c) ( x + 1) y ′′′ + x y ′′ + x 2 y ′ + x 3 y = 0 .
(d) x 4 y ′′ + x 2 y ′ + 2 y = 0 .
(e) x y ′′ + (sin x ) y ′ + e x y = 0 .

In the next section we shall discuss the series solution methods for solving differential
equations with variable coefficients about an ordinary point and about a regular
singular point.

2.4 SERIES SOLUTION METHODS

Let us consider Eqn.(15) viz.,

y′′ + P( x ) y ′ + Q( x ) y = 0

a1 (x) a (x)
where, P( x ) = , Q( x ) = 2 and a 0 ( x ), a 1 ( x ) and a 2 ( x ) are continuous
a 0 (x) a 0 (x)
functions of x over some interval I and a 0 ( x ) ≠ 0 .

The series solution methods of solving such equations are classified into two
categories: Power series method and general series solution method-in particular, the
Frobenius method. We first discuss the power series solution of Eqn.(15) about an
ordinary point.
2.4.1 Power Series Solution About an Ordinary Point
We start with a sufficient condition for the existence of a power series solution of
Eqn.(15).
Theorem 1: Let x = a be an ordinary point (regular point) of Eqn.(15). Then, every
solution of the equation is analytic at x = a and hence has a power series expansion
about the point x = a of the form
∞
y( x ) = ∑c
m =0
m (x − a) m (17)

where c 0 , c1 are constants and c k for k ≥ 2 are obtained in terms of c 0 and c1 .

The proof is obvious since a 0 ( x ) ≠ 0 , P ( x ) and Q( x ) are analytic at x = a . Hence

y ′(a ), y ′′(a ), y ′′′(a ), K exist and the Taylor expansion of y( x ) , that is, power series
solution about x = a exists. Further, note that every function which is analytic in the
∞
region | x − a | < R admits a converging power series representation ∑c
m =0
m (x − a ) m in

the region.
Let us for the sake of convenience, restrict our argument to the case in which a = 0 .
This allows us to work with power series in x rather than x − a , and involves no real
49
Ordinary Differential loss of generality. With this slight simplification, Eqn.(15) will now have solution of
Equations the form
∞
y( x ) = ∑c
m =0
mx
m
(18)

where c m ' s are constants to be determined. Also now P ( x ) and Q( x ) are analytic at
the origin and therefore have power series expansions of the form
∞
P(x ) = ∑p
m =0
mx
m
(19)

∞
and, Q( x ) = ∑q
m =0
mx
m
(20)

where p m ' s and q m ' s are known constants. The above expansion will converge on
an interval | x | < R for some R > 0 .
In order to find a solution of Eqn.(15), term by term differentiation of Eqn.(18) yields
∞ ∞
y′ = ∑
m =1
mc m x m −1 = ∑ (m + 1)c
m =0
m +1 x
m
= c1 + 2c 2 x + 3c 3 x 2 + L

∞ ∞
and y ′′ = ∑
m=2
m(m − 1)c m x m − 2 = ∑ (m + 1) (m + 2) c
m =0
m+2 x
m

2
= 2c 2 + 2. 3c 3 x + 3.4c 4 x + L (21)
From the rule for multiplying power series, it follows that
 ∞  ∞ 
 m=0
∑   m =0
∑
P ( x ) y ′ =  p m x m   (m + 1) c m +1 x m 

∞
m  m
= ∑∑
 p m − k (k + 1) c k +1  x
m =0  k =0 
(22)

 ∞   ∞

and
 m=0
∑
Q( x ) y =  q m x m   c m x m 
  m =0
∑ 
∞
 m  k
= ∑∑
 q m−k c k  x
m =0  k = 0


(23)

On substituting Eqns(21), (22) and (23) into Eqn. (15) and adding the series term by
term, we obtain
∞
 m m

∑ 
m =0 
( m + 1) ( m + 2) c m+ 2 +
k =0
p m−k ( ∑
k + 1) c k +1 +
k =0 
∑
q m−k c k  x m = 0 .

Thus we have the following recursion formula for c m :

m
(m + 1) (m + 2)c m + 2 = − ∑ [(k + 1) p
k =0
m − k c k +1 + q m−k c k ] (24)

For m = 0,1, 2,K this formula becomes

2c 2 = − ( p 0 c 1 + q 0 c 0 )
2.3c 3 = −(p1c1 + 2p 0 c 2 + q 1c 0 + q 0 c1 )
3.4c 4 = −(p 2 c1 + 2p1c 2 + 3p 0 c 2 + q 2 c 0 + q 1c1 + q 0 c 2 )

These formulas determine c 2 , c 3 , K in terms of c 0 and c1 , and the series (18), which
satisfies Eqn.(15), is uniquely determined.
We illustrate the above method by means of the following examples.
Example 3: Solve y ′′ − x y ′ + 3y = 0 (25)
Solution: We observe that Eqn.(25) is a linear equation and its coefficients are
analytic everywhere. In particular, x = 0 is an ordinary point of Eqn.(25) and hence
50 we try power series solution
 ∞ Power Series Solutions
y( x ) = ∑c
m =0
mx
m
(26)

where the unknown constants c m ' s are to be determined. Here

∞
y ′( x ) = ∑mc
m =1
mx
m −1

∞
(27)
y ′′( x ) = ∑ m (m − 1) c
m=2
mx
m−2

Substituting from Eqns.(26) and (27) the values of y, y ′, y ′′ into Eqn.(25), we get
∞ ∞ ∞

∑
m=2
m(m − 1) c m x m − 2 − x ∑
m =1
mc m x m −1 + 3
m=0
∑c mx
m
= 0.

∞ ∞ ∞
or, ∑
m=0
(m + 2) (m + 1)c m + 2 x m − ∑
m =1
m cmx m + 3 ∑c
m =0
mx
m
=0 (28)

Equating coefficients of various powers of x on both sides of Eqn.(28), we get,

2.1 c 2 + 3c 0 = 0 (29)
and, (m + 2) ( m + 1) c m + 2 − (m − 3) c m = 0, m ≥ 1 (30)
From Eqns.(29) and (30), we get
c 2 = −(3 / 2) c 0 (31)
and, c m + 2 = [ (m − 3) /(m + 1) (m + 2) ] c m for m ≥ 1 (32)
Setting m = 1, 2, 3, K , we get
c 3 = −(1 / 3) c1 (33)
1 1  3 1
c4 = − c2 = −  −  c0 = c0 (34)
4 .3 4 .3  2  8
c5 = 0 (35)
We observe that once c 5 = 0 , all the coefficients
c7 = c9 = L = 0 . (36)
Remaining even indexed coefficients c 6 , c 8 , K can be determined in terms of c 0
using Eqn.(32). Recurrence relation (32) yields
2r −1
c 2 r+4 = c 2 r+2
(2 r + 4) ( 2 r + 3)
(2 r − 1) (2 r − 3)
= . c 2r
(2 r + 4) ( 2 r + 3) (2 r + 2) (2 r + 1)
(2 r − 1) (2r − 3) ( 2r − 5)
= c 2 r − 2 = LL
(2r + 4) (2r + 3) (2r + 2) (2r + 1) 2r (2r − 1)
(2r − 1) (2r − 3) L.3.1
= c4
(2r + 4) (2r + 3) L 7.6.5.
(2r − 1) ( 2r − 3) L 3.1 1 3
= . . c0
(2r + 4) (2r + 3) L 7.6.5 4.3 2
3(2r − 1) (2r − 3) L.3.1
= c 0 , for r = 1, 2, K. (37)
(2r + 4)!
Thus from Eqn.(26), the solution becomes
∞
3(2r − 1) (2r − 3) K .3.1
y( x ) = c 0 + c 2 x 2 + c 4 x 4 +
r =1
∑ (2r + 4)!
c 0 x 2 r + 4 + c1 x + c 3 x 3

 3 1 ∞
(2r − 1) (2r − 3) K 3.1 2 r + 4   1 3
= c 0 1 − x 2 + x 4 + 3
 2 8 r =1 ( 2 r
∑+ 4 )!
x  + c1  x − x 
  3 
(38)
51
Ordinary Differential The radius of convergence R of the series, obtained by using ratio test, is
Equations
(2r − 1) (2r − 3) K 3.1 (2r + 6)!
R = lim .
r →∞ (2r + 4)! (2r + 1) (2r − 1) K.3.1
(2r + 5) (2r + 6)
= lim →∞
r →∞ (2r + 1)
Therefore the series in Eqn.(38) converges for all x and the solution is valid for all x .
***
Example 4: Find the power series solution about the origin of the equation
(1 − x 2 ) y ′′ − 4x y ′ + 2 y = 0 (39)
Solution: We know that x = 0 is an ordinary point of Eqn.(39) and therefore the
power series solution exists. Substituting the expressions for y, y ′, y ′′ from Eqns.(26)
and (27) in Eqn.(39), we get
∞ ∞ ∞
(1 − x 2 ) ∑
m=2
m(m − 1) c m x m − 2 − 4 x ∑
m =1
m c m x m −1 + 2 ∑c
m =0
mx
m
=0.
∞ ∞ ∞ ∞
or, ∑
m=2
m(m − 1)c m x m − 2 − ∑
m=2
m(m − 1)c m x m − 4 ∑
m =1
mc m x m + 2 ∑c
m =0
mx
m
=0

∞ ∞
or, 2c 2 + 6c 3 x + ∑
m=4
m(m − 1)c m x m − 2 − ∑ m(m − 1)c
m=2
mx
m
− 4c 1 x
∞ ∞
−4 ∑
m=2
mc m x m + 2(c 0 + c1 x ) + 2 ∑c
m=2
mx
m
=0 (40)

In the third term of Eqn.(40), setting m − 2 = t or, m = t + 2 , we get

∞
2(c 2 + c 0 ) + 2(3c 3 − c1 ) x + ∑ (t + 2) (t + 1) c
t =2
t+2 xt
∞ ∞ ∞
− ∑
m=2
m(m − 1) c m x m − 4 ∑
m=2
m cm x m + 2 ∑c
m=2
mx
m
=0

Since t is dummy variable, we can write

∞
2(c 2 + c 0 ) + 2(3c 3 − c 1 ) x + ∑ [ (m + 2) (m + 1) c
m=2
m+2 − (m 2 + 3m − 2) c m ]x m = 0

Setting the coefficients of successive powers of x to zero, we obtain

c 2 + c 0 = 0, 3c 3 − c 1 = 0, (m + 2) (m + 1) c m + 2 = ( m 2 + 3m − 2)c m , m ≥ 2
Solving, we obtain
1 (m 2 + 3m − 2)
c 2 = −c 0 , c 3 = c1 , c m + 2 = cm , m ≥ 2
3 (m + 2) (m + 1)
where c 0 and c1 are arbitrary constants. We thus have
2 −2 4 4
c4 = c2 = c 0 , c 5 = c 3 = c1 ,
3 3 5 15
13 − 26 19 76
c6 = c4 = c0 , c7 = c5 = c1 , K
15 45 21 315
The power series solution is
 2 26 6   x3 4 76 7 
y( x ) = c 0 1 − x 2 − x 4 − x − L + c1  x + + x5 + x + L (41)
 3 45   5 15 315 
***
Remarks: In Example 4, x = ±1 are the singular points of Eqn.(39). Thus, the power
series solution cannot exist in any interval I which contains + 1 or − 1 . Therefore the
52 interval of largest length, in which the power series solution (41) holds is ] − 1, 1 [
 Power Series Solutions
which is its interval of convergence. Further, since the radius of convergence R is the
distance between the point x = a , about which the power series solution is sought and
the nearest singularity of the equation therefore, in this case, R = 1 .
You may now try some exercises.

E5) Solve ( x 2 − 2 x + 2) y ′′ − 4( x − 1) y ′ + 6 y = 0 , about the point x = 1 .

E6) Solve y ′′ + x y′ + 3y = 0 , by power series method about the point x = 0 .
E7) Find power series solution in powers of ( x − 1) of the initial value problem
x y ′′ + y ′ + 2 y = 0
.
y(1) = 1, y ′(1) = 2
E8) Find the power series solution of the equation y ′′ + xy ′ + y = 0 , about x = 2 .
E9) The equation y ′′ − 2 xy ′ + 2ny = 0 , where n is a positive integer, is called
Hermite’s equation of order n . Using the power series method show that if
∞
z n (x) = ∑a
k =0
kx
k

is a solution of Hermite’s equation, then

 ∞
2 k ( − n ) ( − n + 2) L ( − n + 2 k − 2) 2 k 
z n ( x ) = a 0 1 +∑ (2k )!
x 
 k =1 
 ∞
2 k (1 − n ) (1 − n + 2) L (1 − n + 2k − 2) 2 k +1 
+ a 1 x + ∑ (2k + 1)!
x .
 k =1 

So far, we have discussed a method of finding series solution of a differential equation

about an ordinary point, that is, power series solution of the equation. Power series is
inadequate to represent a solution about a regular singular point x = a , as the solution
may also contain negative powers and/or fractional powers of ( x − a ) . We shall now
discuss the Frobenius method for obtaining a series solution about a regular singular
point of the equation.
2.4.2 Series Solution about a Regular Singular Point: Frobenius
Method
The method we are going to discuss here is due to Ferdinand George Frobenius (1849-
1917), a German mathematician who made several valuable contribution to the theory
of elliptic functions and differential equations. We shall be restricting ourselves to
second order equations only because most of the important equations, in physical
science and engineering, such as Bessel’s equation, Legendre’s equation, Hermite’s
equation or, the hypergeometric equations for which the method works are of second
order only. Also, to simplify the matter we shall assume that the singular point a is
located at the origin, for if it is not, then we can always move it to the origin by
changing the independent variable from x to x − a .
Method of Frobenius
Consider linear differential equation of order two of the form
d2y dy
f ( x ) 2 + g(x ) + r(x) y = 0 (42)
dx dx
Let x = 0 , be a regular singular point of Eqn.(42).

The method of Frobenius to solve Eqn.(42) consist of the following steps:-

Step I: Take a trial solution of Eqn.(42) of the form

y( x ) = x k (c 0 + c1 x + c 2 x 2 + L + c m x m + L)
53
Ordinary Differential ∞
Equations i.e., y( x ) = x k ∑c
m =0
mx
m
.
∞
= ∑c
m =0
mx
m+k
, where k is any real number and c 0 ≠ 0 (43)

Step II: Differentiate relation (43) and obtain

∞

y ′( x ) = ∑ ( m + k )c
m=0
mx
m + k −1


∞  (44)
and y ′′( x ) = ∑
m =0
(m + k ) (m + k − 1)c m x m + k − 2 

Using Eqns.(43) and (44), Eqn.(42) reduces to an identity.
Step III: Equate to zero the coefficient of the smallest power of x in the identity in
Step II above. This would yield a quadratic equation in k . The quadratic equation so
obtained is called the indicial equation. The two roots of the indicial equation are
called exponents of Eqn.(42) and are the only possible values of k in Eqn.(43).
Step IV: Solve the indicial equation. The following cases are possible:
1) Indicial roots are distinct and do not differ by an integer.
2) Indicial roots are distinct and differ by an integer.
3) Indicial roots are equal.
Step V: Equate to zero the coefficient of general power of x (e.g. x k + m or x k + m −1 ,
whichever may be the lowest) in the identity obtained in Step II. The equation so
obtained is called the recurrence relation because it connects the coefficients
c m , c m − 2 or, c m , c m −1 , etc.

Step VI: If the recurrence relation connects c m and c m − 2 , then, in general, determine
c1 by equating to zero the coefficient of the term having next higher power in x i.e.,
the term in x next to the term already used for getting the indicial equation. On the
other hand, if the recurrence relation connects c m and c m −1 , this step may be omitted.
Step VII: Obtain all the coefficients with the help of Steps V and VI above.
Substitute them in relation (43) and obtain the required solutions corresponding to two
roots of the indicial equation.
We shall illustrate various steps involved in the method through examples. Depending
upon the nature of the indicial roots listed in Step IV, different types of solution
procedures are followed. We shall now illustrate these procedures one-by-one. You
may notice here that the two series solutions obtained in each case are linearly
independent.
Case 1: Indicial roots are distinct and do not differ by an integer.
Example 5: Obtain the series solution about the origin of the equation
9x (1 − x ) y ′′ − 12 y ′ + 4 y = 0 (45)
Solution: Evidently x = 0 is a regular singular point of Eqn.(45).
∞
Let y( x ) = ∑c
m =0
mx
k+m
, c0 ≠ 0 .

Differentiating y( x ) twice w.r.t. x , we get

∞
y ′( x ) = ∑ ( k + m) c
m =0
mx
k + m −1

∞
and y ′′( x ) = ∑ (k + m) (k + m − 1) c
m =0
mx
k + m −2

Substituting the above values of y, y ′, y ′′ in Eqn.(45), we get

54
 ∞ ∞ Power Series Solutions
9x (1 − x ) ∑ (k + m) (k + m − 1) c
m =0
mx
k +m−2
− 12 ∑ (k + m) c
m =0
mx
k + m −1

∞
+4 ∑c
m=0
mx
k+m
=0

∞ ∞
or, 9 ∑
m =0
(k + m) (k + m − 1) c m x k + m −1 − 9 ∑ (k + m) (k + m − 1) c
m =0
mx
k+m

∞ ∞
− 12 ∑ ( k + m)c
m =0
mx
k + m −1
+4 ∑c
m=0
mx
k+m
=0 (46)

Combining the first and third term and the second and fourth term of Eqn.(46), we get
∞ ∞
3
m=0
∑ (k + m) (3k + 3m − 7)c mx
k + m −1
− ∑ [9(k + m)
m =0
2
− 9(k + m) − 4]c m x k + m = 0
∞ ∞
or, 3 ∑
m =0
(k + m) (3k + 3m − 7)c m x k + m −1 − ∑ (3k + 3m − 4) (3k + 3m + 1)c
m =0
mx
k +m
=0
∞ ∞
or, 3 ∑
m = −1
( k + m + 1) (3k + 3m − 4)c m +1 x k + m − ∑ (3k + 3m − 4) (3k + 3m + 1)c
m=0
mx
k +m
=0

∞
or, 3k (3k − 7)c 0 x k −1 + ∑ x [3(k + m + 1) (3k + 3m − 4)c
m =0
k+m
m +1

− (3k + 3m − 4) (3k + 3m + 1)c m ] = 0 (47)

k −1
Equating to zero the coefficient of smallest power of x (i.e., x ) in identity (47), we
get indicial equation as
3k (3k − 7) = 0 ⇒ k = 0, 7 / 3 .
Clearly roots of indicial equation are distinct, not differing by an integer. Equating to
zero the coefficient of general power of x , i.e., x k + m , in identity (47), we get the
recurrence relation as

3(k + m + 1) (3k + 3m − 4) c m +1 = (3k + 3m − 4) (3k + 3m + 1) c m , for m ≥ 0

(3k + 3m − 4) (3k + 3m + 1) 3k + 3m + 1
i.e., c m +1 = cm = cm (48)
3(k + m + 1) (3k + 3m − 4) 3(k + m + 1)

3m + 1
For k = 0 , c m +1 = cm , m ≥ 0
3(m + 1)
Putting m = 0, 1, 2, 3, K
1 4 1 .4 7 1 .4 .7
c1 = c 0 , c 2 = . c1 = c0 , c3 = c 2 = c0 ,L
3 6 3.6 9 3.6.9
Therefore, series solution corresponding to k = 0 is
 1 1.4 2 1.4.7 3 
y 1 ( x ) = c 0 1 + x + x + x + L , (49)
 3 3 .6 3 .6 .9 
For k = 7/3 , we get from Eqn.(48),
3m + 8
c m +1 = C m , for m ≥ 0
3m + 10
Putting m = 0,1, 2, K , , we get
8 11 11.8 14 14.11.8
c1 = c 0 , c 2 = c1 = c0 , c3 = c 2 = c 0 ,L
10 13 13.10 16 16.13.10
Therefore, series solution corresponding to k = 7 / 3 is
55
Ordinary Differential  8 8.11 2 8.11.14 3 
Equations y 2 ( x ) = c 0 x 7 / 3 1 + x + x + x + L (50)
 10 10.13 10.13.16 
Since the solutions y 1 ( x ) and y 2 ( x ) given by Eqns.(49) and (50) are linearly
independent, the general solution of Eqn.(45) may be written as
y( x ) = α 1 y1 ( x ) + α 2 y 2 ( x ) ,

where α 1 and α 2 are arbitrary constants.

***
Note: In general, if indicial equation has two distinct roots, differing not by an integer,
two linearly independent solutions are obtained by substituting the values of k in the
series for y( x ) .
Let us look at another example.
Example 6: Find a series solution about x = 0 , of the equation
2 x 2 y ′′ − xy ′ + (1 − x 2 ) y = 0 .
Solution: Observe that the given equation
2 x 2 y ′′ − x y ′ + (1 − x 2 ) y = 0 (51)
has x = 0 as a regular singular point.
∞
Let y( x ) = x k (c 0 + c 1 x + c 2 x 2 + L) = ∑c
m =0
mx
m+k
(52)

be a solution of Eqn.(51).

Differentiating Eqn.(52) w.r.t. x , we get

∞
y ′( x ) = ∑ ( k + m )c
m =0
mx
m + k −1
(53)
∞
and y ′′( x ) = ∑ (k + m) (k + m − 1) c
m =0
mx
k + m −2
(54)

Putting the values of y, y ′, y ′′ from Eqns.(52), (53) and (54) in Eqn.(51), we get
∞ ∞
2x 2 ∑
m =0
(k + m) (k + m − 1)c m x k + m − 2 − x ∑ ( k + m )c
m =0
mx
k + m −1

∞
+ (1 − x 2 ) ∑c
m =0
mx
k +m
=0
∞ ∞ ∞
or, ∑ 2(k + m) ( k + m − 1)c m x k + m − ∑ ( k + m)c m x k + m + ∑c mx
k +m

m =0 m=0 m=0 (55)

∞
− ∑c
m =0
mx
k+m+2
=0

Equating the coefficient of lowest power of x , i.e., of x k to zero in identity (55), we

get the indicial equation as
2c 0 k (k − 1) − k c 0 + c 0 = 0, i.e., c 0 [2k (k − 1) − k + 1] = 0
i.e., c 0 (k − 1) (2k − 1) = 0 ⇒ k = 1, 1 / 2 , i.e., roots of indicial equation are distinct, but
difference is not an integer.
To get a recurrence relation, we equate the coefficients of general power of x , i.e.,
x k + m in identity (55) equal to zero and obtain
[2 (k + m) (k + m − 1) − (k + m) + 1] c m = c m − 2
c m−2
i.e., cm = for all m ≥ 2 . (56)
(k + m − 1) (2k + 2m − 1)
56
 Power Series Solutions
Again, equating the coefficient of x k +1 in Eqn.(55) to zero , we get
2(k + 1) kc1 − (k + 1)c1 + c1 = 0 ⇒ c1 [ 2k (k + 1) − (k + 1) + 1] = 0
⇒ c1 k (2k + 1) = 0 .
Now for k = 1, 1 / 2, k (2k + 1) ≠ 0, hence c 1 = 0 (57)
From Eqns.(56) and (57), we get
c1 = 0 = c 3 = c 5 = c 7 = K , = c 2 m +1 = K
Putting m = 2, 4, 6,K in relation (56), we get
1
c2 = c0.
(2k + 3) (k + 1)
1 1
c4 = c2 = c0 .
(k + 3) (2k + 7) (2k + 3) (2k + 7) (k + 1) (k + 3)
1 1
c6 = = c0 .
(k + 5) ( 2k + 11) (2k + 3) (2k + 7) (2k + 11) (k + 1) (k + 3) (k + 5)
LLLLLLLLLLLLLLLLLLLLLLLLLLL
Putting these values of c 0 , c1 , K in Eqn.(52), we get
 c0 c0
y( x ) = x k c 0 + x2 + x4 +
 (2k + 3) (k + 1) (2k + 3) (2k + 7) (k + 1) (k + 3)
c0 
+ x 6 + L .
[(2k + 3) (2k + 7) (2k + 11) (k + 1) (k + 3) (k + 5)] 

Putting k = 1 , we get
 x2 x4 x6 
y1 = x c 0 1 + + + + L = c 0 u (58)
 2.5 2.4.5.9 2.4.6.5.9.13 
Putting k = 1/2 in y( x ) , we get
 x2 x4 x6 
y 2 = c 0 x 1 / 2 1 + + + + L = c 0 v, (59)
 2.3 2.3.4.7 3.7.11.2.4.6 

Hence the general solution of Eqn.(51) is

y( x ) = A 1 y1 ( x ) + B1 y 2 ( x )
(60)
= Au + Bv
where A = A 1 c 0 and B = B1 c 0 are arbitrary constants and

 1 2 1 1 
u = x 1 + x + x4 + x 6 + L (61)
 2 .5 2.4.5.9 2.4.6.5.9.13 
 1 2 1 1 
v = x 1 / 2 1 + x + x4 + x 6 + L (62)
 2.3 2 .4 .3 .7 2.4.6.3.7.11 
***
You may now try the following exercises.

E10) Find a series solution about x = 0 , of the following equations

a) 2 x (1 − x ) y ′′ + (1 − x ) y ′ + 3y = 0 .
b) (2 x + x 2 ) y ′′ − y ′ − 6 x y = 0 .

Case 2: Indicial roots are distinct and differ by an integer

Let the roots of the indicial equation be k 1 , k 2 such that k 1 − k 2 = n , an integer. In
this case two types of problems arise. In one type two linearly independent solutions
57
Ordinary Differential are obtained by substituting k = k 1 and k = k 2 in y( x ) as in Case 1. This case is
Equations
illustrated through the following example.

Example 7: Find a series solution about x = 0 , of the equation

x 2 y ′′ + x 3 y ′ + ( x 2 − 2) y = 0 (63)
Solution: Evidently x = 0 , is a regular singular point of Eqn.(63).
∞
Let y= ∑c
m =0
mx
k+m
, c 0 ≠ 0. (64)

∞ ∞
∴ y′ = ∑
m=0
(k + m)c m x k + m −1 and y ′′ = ∑c
m =0
m (k + m) (k + m − 1) x k + m − 2

Putting the values of y, y ′, y ′′ in Eqn.(63), we get

∞ ∞
x2 ∑c
m =0
m (k + m) (k + m − 1) x k + m − 2 + x 3 ∑ ( k + m) c
m=0
mx
k + m −1

∞
+ ( x 2 − 2) ∑c
m =0
mx
k+m
=0

∞ ∞
i.e., ∑
m =0
c m [(k + m) (k + m − 1) − 2] x k + m + ∑c
m=0
m (k + m + 1) x k + m + 2 = 0 . (65)

Equating to zero the coefficients of smallest power of x , namely x k , Eqn.(65) gives

c 0 [k (k − 1) − 2] = 0 ⇒ k 2 − k − 2 = (k − 2) (k + 1) = 0, or k = 2,−1, c 0 ≠ 0 .

These are distinct roots of indicial equation, which differ by an integer. Equating to
zero the coefficient of next power of x , namely x k +1 , we get
[(k + 1) k − 2] c1 = 0, or c1 = 0 for k = 2,−1 .
The remaing terms are given by
∞ ∞

∑
m=2
[ (k + m) (k + m − 1) − 2] c m x k + m + ∑ (k + m + 1) c
m =0
mx
k + m+2
=0

From here we get the recurrence relation

[(m + k + 1) (m + k + 2) − 2] c m + 2 + (m + k + 1) c m = 0, m ≥ 0
(m + k + 1) c m
or, c m+2 = − , m≥0 (66)
(m + k + 1) (m + k + 2)
Putting m = 1, 2, 3,K , we get
− (k + 1) c 0 − ( k + 2) c 1 − (k + 3) c 2
c2 = , c3 = = 0, c4 = , L.
(k + 1) (k + 2) − 2 (k + 2) (k + 2) − 2 (k + 3) (k + 4) − 2
Therefore from Eqn.(64), we get

 (k + 1) (k + 3) (k + 1) 
y( x ) = c 0 x k 1 − x2 + x 4 − L
 (k + 1) (k + 2) − 2 [(k + 1) (k + 2) − 2] [(k + 3) (k + 4) − 2 ] 
(67)
For k = 2 , we get
 3 3 4 
y( x ) = c 0 x 2 1 − x 2 + x − L = y1
 10 56 
For k = −1 , coeffieicents c 2 = 0, c 3 = 0, c 4 = 0, K and
c
y( x ) = 0 = y 2 .
x
Hence the derived solution is y = a y1 + b y 2 , where a and b are arbitrary constants.
***
58
 Power Series Solutions
In the second type of problems again two cases arise. We now discuss them as I and
II and give the rules for solving such problems.
I: Indicial roots are distinct, differ by an integer and make a coefficient of y
indeterminate:
In this situation the following rule is used to obtain the two linearly independent
solutions of the given problem.
Rule: Let the indicial equation has two roots k 1 and k 2 such that k 2 − k 1 = n , an
integer. Now if one of the coefficients of y becomes indeterminate when k = k 1
(say), the complete solution is given by putting k = k 1 in y , which then contains two
arbitrary constants. The result of putting k = k 2 in y merely gives a numerical
multiple of one of the series contained in the first solution hence producing a linearly
dependent solution, which is rejected.
We illustrate the method for this case through the following example:
Example 8: Obtain a series solution about x = 0 , of the equation
x 2 y ′′ + 6 xy ′ + (6 + x 2 ) y = 0 (68)
Solution: We find x = 0 is a regular singular point of Eqn.(68).
∞
Let y( x ) = ∑c
m=0
mx
m+k
, c0 ≠ 0 . (69)

Substituting for y, y ′, y ′′ in Eqn.(68), we get

∞ ∞
x2 ∑
m =0
(m + k ) (m + k − 1) c m x m + k − 2 + 6x ∑ (m + k ) c
m=0
mx
m + k −1

∞
+ (6 + x 2 ) ∑c
m=0
mx
m+ k
=0
∞ ∞ ∞
or, ∑
m =0
(m + k ) (m + k − 1) c m x m + k + 6 ∑
m =0
(m + k ) c m x m + k + 6 ∑c
m =0
mx
m+ k

∞
+ ∑c
m =0
mx
m+k +2
=0.
∞ ∞
or, ∑ [(m + k ) (m + k + 5) + 6]c m x m +k + ∑ c m x m+k +2 = 0 .
m=0 m =0
(70)

Setting the coefficient of lowest degree term, namely x k to zero, we obtain

[k (k + 5) + 6] c 0 = 0, or (k + 2) (k + 3) = 0, c 0 ≠ 0 (71)

The indicial roots are k = −2, − 3 . Setting the coefficient of x k +1 to zero, we get
[ (k + 1) ( k + 6) + 6 ] c1 = 0 .
For k = −2, c1 = 0 and for k = −3, c1 is indeterminate. Therefore, k = −3 gives the
complete solution, which contains two arbitrary constants. Equating the coefficients
of general term to zero, we get
− cm
c m+2 = , m≥0 (72)
( m + k + 2) ( m + k + 7 ) + 6
Putting m = 1, 2, 3, K in Eqn.(72), we get
− c0 − c1
c2 = c3 =
( k + 2) ( k + 7 ) + 6 (k + 3) (k + 8) + 6

59
Ordinary Differential − c2 c0
Equations c4 = =
(k + 4) (k + 9) + 6 [( k + 2) (k + 7) + 6 ] [(k + 4) (k + 9) + 6 ]
and so on.
Therefore from Eqn.(69), we get

 1 1 
y( x ) = c 0 x k 1 − x2 + x 4 − L
 ( k + 2) ( k + 7 ) + 6 [(k + 2) (k + 7) + 6 ] [(k + 4) (k + 9) + 6 ] 
 1 1 
+ c1 x k x − x3 + x 5 − L
 ( k + 3) (k + 8) + 6 [(k + 3) (k + 8) + 6 ] [(k + 5) (k + 10) + 6 

For k = −3 , we get

 x2 x4   x3 x5 
y( x ) = c 0 x −3 1 − + − L + c1 x −3  x − + − L
 2! 4!   3! 5! 
 cos x   sin x 
= c 0  3  + c1  3  = c 0 y 1 ( x ) + c1 y 2 ( x ) (73)
 x   x 

For k = −2 , we know c1 = 0 = c 3 = c 5 and so on.

 x2 x4   x3 x5 
∴ y( x ) = c 0 x − 2 1 − + − L = c 0 x −3  x − + − L = c 0 y 2 ( x ) (74)
 3! 5!   3! 5! 

which is a linearly dependent solution.

***
You may now attempt a few exercises.

E11) Obtain the series solution about x = 0 , of the equation

(2 + x 2 ) y ′′ + xy ′ + (1 + x ) y = 0 .

E12) Find a series solution about x = 0 , of the differential equation

(1 − x 2 ) y ′′ + 2 xy ′ + y = 0 .

We, now take up II for the second type of problem.

II: Indicial roots are distinct, differ by an integer and make a coefficient of y
infinite:
In this case, the following rule is used:
Rule: Let the indicial equation has two roots k 1 and k 2 such that
k 1 − k 2 = n , an integer. Now if one of the coefficients of y becomes infinite for
k = k 2 (say), then the form of y in Eqn.(43) is modified by replacing c 0 by
d 0 (k − k 2 ) where d 0 ≠ 0 . Two linearly independent solutions are then obtain by
∂y ∂y 
substituting k = k 2 in y and . The solution corresponding to  k = k will
∂k ∂k  2
always contain a logarithmic term. The result of putting k = k 1 in y merely gives a
solution that is a numerical multiple of the one obtained for k = k 2 , leading to a
linearly dependent solution of the given equation.

Let us take up an example to understand this case.

60
 Power Series Solutions
Example 9: Find the series solution about x = 0 , of the differential equation
( x + x 2 + x 3 ) y ′′ + 3x 2 y ′ − 2 y = 0 (75)
Solution: Evidently, x = 0 , is a regular singular point of Eqn.(75).
∞
Let y= ∑c
m =0
mx
k +m
, c0 ≠ 0 . (76)

∞ ∞
Then y ′ = ∑
m =0
(k + m)c m x k + m −1 and y ′′ = ∑ (k + m) (k + m − 1)c
m=0
mx
k + m −2
(77)

Substituting for y, y ′, y ′′ in Eqn.(75), we get

∞ ∞
(x + x 2 + x 3 ) ∑
m =0
c m (k + m) (k + m − 1) x k + m − 2 + 3x 2 ∑c
m =0
m (k + m) x k + m −1

∞
−2 ∑c
m =0
mx
k +m
=0

∞ ∞
i.e., ∑
m =0
c m (k + m) (k + m − 1) x k + m +1 + ∑c
m =0
m 3( k + m) x k + m +1

∞ ∞
+ ∑
m =0
c m [(k + m) (k + m − 1) − 2] x k + m + ∑c
m =0
m (k + m) (k + m − 1) x k + m −1 = 0.

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

∞
+ ∑c
m = −1
m +1 ( k + m + 1) (k + m) x k + m = 0

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

∞
+ ∑ (k + m + 1) (k + m) c
m = −1
m +1 x
m+k
=0

i.e., k (k − 1) c 0 x k −1 + [ (k + 1) k c1 + { k (k − 1) − 2} c 0 ] x k
∞
+ ∑ [ (k + m + 1) (k + m) c
m =1
m +1 + ( m + k − 1) (m + k + 1) c m −1

+ {( k + m) (k + m − 1) − 2} c m ]x m + k = 0 (78)

This is an identity. Here the indicial equation obtained by equating to zero the
coefficients of x k −1 is
c 0 k (k − 1) = 0 ⇒ k = 0, 1 (Q c 0 ≠ 0). (79)

Equating to zero the coefficients of x k and x m + k in Eqn.(78), we get

(k + 1)k c 1 + {k (k − 1) − 2} c 0 = 0 (80)
and, (k + m + 1) (k + m) c m +1 + (m + k + 1) (m + k − 1) c m −1
+ { (k + m) ( k + m − 1) − 2} c m = 0, for m ≥ 1 (81)
k (k − 1) − 2
From Eqn.(80), we get c1 = − c0
k ( k + 1)
Putting m = 1, 2, 3, 4, K in Eqn.(81), we get the following:
For m = 1
(k + 2) (k + 1) c 2 + k (k + 2) c 0 + {k (k + 1) − 2} c1 = 0
61
Ordinary Differential k [k (k − 1) − 2] [k (k + 1) − 2] 3k − 2
Equations i.e., c2 = − c0 + 2
c0 = − c0
k +1 k (k + 1) (k + 2) k (k + 1)
For m = 2
(k + 3) (k + 2) c 3 + (k + 3) (k + 1) c 1 + { (k + 2) (k + 1) − 2} c 2 = 0
k +1 k 2 + 3k k +1 (3k − 2)
i.e., c3 = − c1 − c2 = − c1 + c0
k+2 (k + 3) (k + 2) k+2 ( k + 1) (k + 2)
 k (k − 1) − 2 3k − 2  k 3 + 3k 2 − 5k − 2
= + c
 0 = c0
 k ( k + 2) k +1  k (k + 1) ( k + 2)
and so on.
Substituting the values of c1 , c 2 , c 3 , K in Eqn.(76), we get

 k2 − k − 2 3k − 2 2 k 3 + 3k 2 − 5k − 2 3 
y( x ) = c 0 x k 1 − x− x + x − L − (82)
 k (k + 1) k (k + 1) k (k + 1) (k + 2) 
It is seen that for k = 0 , the coefficients in solution (82) become infinite. So we,
replace c 0 by k r0 (r0 ≠ 0) . Thus Eqn(82) reduces to

 k2 − k − 2 3k − 2 2 k 3 + 3k 2 − 5k − 2 3 
y( x ) = r0 x k k − x− x + x − L − (83)
 k +1 k +1 (k + 1) (k + 2) 
Substituting from Eqn.(83) in L.H.S. of Eqn.(75), we get
( x + x 2 + x 3 ) y ′′ + 3x 2 y ′ − 2 y = r0 k 2 (k − 1) x k −1

(where the right hand side is simply the indicial equation). Here the presence of k 2
∂y
shows that y and satisfy Eqn.(75) for k = 0 .
∂k
From Eqn.(83), we get
∂y  k2 −k −2 3k − 2 2 k 3 + 3k 2 − 5k − 2 3 
= r0 x k ln x  k − x− x + x − L
∂k  k +1 k +1 (k + 1) (k + 2) 
  2k − 1 k − k − 2   3
2
3k − 2  2
+ r0 x k 1 −  − 2 
x− − x
  k + 1 (k + 1)   k + 1 (k + 1) 2 
 3k 2 + 6k − 5 k 3 + 3k 2 − 5k − 2 k 3 + 3k 2 − 5k − 2  3 
+ − −  x − L (84)
 (k + 1) (k + 2) (k + 1) 2 (k + 2) ( k + 1) (k + 2) 2  
Putting k = 0 in Eqns.(83) and (84), we get
y = r0 [2 x + 2 x 2 − x 3 − L] = y 1 , say (85)
∂y
and = (ln x ) y 1 + r0 [1 − x − 5x 2 − x 3 − L] = y 2 , say (86)
∂k

Putting k = 1 in Eqn.(83), we get

 1 1 
y = r0 x 1 + x − x 2 − x 3 − L = y 3 , say (87)
 2 2 
Evidently, 2 y 3 = y1 .

Hence, out of three solutions y 1 , y 2 and y 3 obtained above, only two are linearly
independent. Hence general solution is
y = αy 1 + β y 2 ,
where α and β are arbitrary constants.
***
62 You may now try the following exercise.
 Power Series Solutions
E13) Find a series solution of the following differential equations about x = 0 .
a) xy ′′ + (3 − x ) y ′ + y = 0 .
b) x (1 − x ) y ′′ − 3xy ′ − y = 0 .

We shall now take up the case when the roots of indicial equation are equal.
Case 3: Indicial roots are equal
In this case, the following rule works:
Rule: If indicial equation has two equal roots k 1 = k 2 , obtain two linearly
∂y
independent solution by substituting this value of k in y and .
∂k
We illustrate this rule through an example.
Example 10: Solve, in series, the differential equation
xy′′ + y ′ + xy = 0 (88)
Solution: Here x = 0 is a regular singular point.
Let the series solution of Eqn.(88) be of the form
∞
y= ∑c
m =0
mx
k +m
, c0 ≠ 0 (89)

∞ ∞
Thus, y ′ = ∑
m =0
c m (k + m) x k + m −1 and y ′′ = ∑c
m =0
m (k + m) (k + m − 1) x k + m − 2 (90)

Putting these values of y, y ′, y ′′ in Eqn.(88), we get

∞ ∞ ∞

∑c
m =0
m (k + m) (k + m − 1) x k + m −1 + ∑c
m=0
m (k + m) x k + m −1 + ∑c
m =0
mx
k + m +1
=0

∞ ∞
or, ∑c
m =0
m (k + m) 2 x k + m −1 + ∑c
m =0
mx
k + m +1
= 0, (91)

which is an identity. Equating to zero the coefficient of least power x , in identity (91)
indicial equation is obtained as
c 0 k 2 = 0 ⇒ k = 0, 0 (Q c 0 ≠ 0) (92)
Thus indicial equation has equal roots.
Again from Eqn.(91), the recurrence relation is obtained as
c m ( k + m) 2 + c m − 2 = 0
1
i.e., cm = − c m −2 (93)
(k + m) 2
Again equating the coefficient of x k in Eqn.(91) to zero, we get
c1 (k + 1) 2 = 0, but since k = 0, k + 1 ≠ 0 and hence c1 = 0 . (94)
For c1 = 0 , we get from Eqn.(93), c1 = c 3 = 0 = c 5 = c 7 = L .
For m = 2, 4, 6,K Eqn.(93) yields
1
c2 = − c0 ,
( k + 2) 2
1 1
c4 = − c =
2 2
c0
( k + 4) (k + 2) (k + 4) 2
2

1
c6 = − c0
( k + 2) ( k + 4) 2 ( k + 6) 2
2

and so on.
Putting these values of c' s in Eqn.(89), we get
63
Ordinary Differential  x2 x4
Equations y = c 0 x k 1 − 2
+
 ( k + 2) (k + 2) 2 (k + 4) 2
x6 
− + L  (95)
( k + 2) 2 ( k + 4) 2 ( k + 6) 2 
Putting k = 0 in Eqn.(95), we get
 x2 x4 x6 
y = c 0 1 − 2 + 2 2 − 2 2 2 + L = c 0 u , say (96)
 2 2 .4 2 .4 .6 

To get another independent solution, substituting Eqn.(95) in L.H.S. of Eqn.(88) and

simplifying, we get
x 2 y ′′ + y ′ + xy = c 0 k 2 x k −1 (97)
Differentiating both sides of Eqn.(97) partially w.r.t. to k gives
∂ 2 c ∂ 2 k
∂k
[
x y ′′ + y ′ + xy = 0]
x ∂k
c
[
(k x ) = 0 2kx k + k 2 x k ln x
x
] (98)

The right hand side of Eqn.(98) vanishes for k = 0 , this suggests that a second
∂y 
solution of the given equation is  , where y is given by Eqn.(95).
∂k  k = 0
Differentiating Eqn.(95) partially w.r.t. to k , we get
∂y  x2 x4
= c 0 x k ln x 1 − 2
+
∂k  ( k + 2) (k + 2) 2 (k + 4) 2
x6 
− 2 2 2
L
( k + 2) ( k + 4) ( k + 6) 
 (−2) x 2
x 4
 −2 2 
+ c 0 x k − + 2 2 
− 
 (k + 2)3 ( k + 2) (k + 3)  k + 2 k + 4 
x6  2 2 2  
− 2 2 − 2
− −  + L (99)
( k + 2) ( k + 4) ( k + 6)  k + 2 k + 4 k + 6  

Putting k = 0 in Eqn.(99), we get

∂y   x2 x4 x6 
 = c 0 ln x 1 − 2 + 2 2
− 2 2 2
+ L 
∂k  k =0  2 2 .4 2 .4 .6 
x2 x4  1 x6  1 1 
+ c0  2 − 2 2 1 +  + 2 2 2 1 + +  + L
2 2 .4  2  2 .4 .6  2 3 
 x2 x4  1 x6  1 1 
= c 0 u ln x +  2 − 2 2 1 +  + 2 2 2 1 + +  + L
 2 2 .4  2  2 .4 .6  2 3 
= c 0 v, say (100)
Thus the required general series solution of the given equation is,
y = A1c0 u + B1c0 v
= Au + Bv
Where A 1 , B1 , A and B are arbitrary constants, such that A 1c 0 = A and B1c 0 = B .
***
How, about trying a few exercises now?

E14) Find a series solution about x = 0 , of the differential equation

x (1 − x ) y ′′ + (1 − x ) y ′ − y = 0 .
64
 Power Series Solutions
E15) Solve in series, xy ′′ + (p − x ) y ′ − y = 0 when,
(a) p = 1 (b) p is not an integer.
E16) Find a series solution about x = 0 , of the following equations:
a) xy ′′ + (1 + x ) y ′ + 2 y = 0 .
b) ( x − x 2 ) y ′′ + (1 − 5x ) y ′ − 4 y = 0 .
c) 4( x 4 − x 2 ) y ′′ + 8x 3 y ′ − y = 0 .

In the next subsection we discuss the case when for an equation of the type (15), we
are required to find a series solution for large values of the independent variable.
2.4.3 Series Solution about a Point at Infinity
Often we come across physical situations where we need to find the solution of
equations of the type
y ′′ + P( x ) y ′ + Q( x ) y = 0 (101)
for large values of the independent variable x . For instance, if the variable is time ,
we may want to know how the physical system described by Eqn.(101) behaves in the
distant future, when transient disturbances have faded out. In such cases, we can use
our previous ideas and broaden them up by studying solutions near a point at infinity.
If we change the independent variable in Eqn.(101) from x to t by the relation
1
t= ,
x
then large x ' s correspond to small t ' s . The transformed equation can then be solved
1
near t = 0 and then replacing t by in the solution obtained, we get solution of
x
Eqn.(101) that are valid for large values of x . To carry out this, we need the formulas
dy dy dt dy  − 1  2 dy
y′ = = =  2  = −t (102)
dx dt dx dt  x  dt
d  dy  d  dy  dt  2 d 2 y dy 
and y ′′ =  =   =  − t 2
− 2 t  (− t 2 ) (103)
dx  dx  dt  dx  dx  dt dt 

When these expressions for y ′ and y ′′ are substituted in Eqn.(101) then it becomes
 2 P(1 / t )  Q(1 / t )
y ′′ +  − 2  y′ + y=0 (104)
 t t  t4
where primes in Eqn.(104) denote derivatives w.r. to t .
We then say that Eqn.(101) has x = ∞ as an ordinary point, a regular singular point
with exponents k 1 and k 2 , or an irregular singular point, if the point t = 0 has the
corresponding character for the transformed Eqn.(104).
We now illustrate the above procedure through an example.
Example 11: Obtain a series solution of the following equation for large values of
x.
x 2 y ′′ + (3x − 1) y ′ + y = 0 . (105)
Solution: Here x = 0 , is an irregular singular point and Frobenius method can not
be applied.
1
We change the independent variable x to t by putting x =
t
dy dy dt dy
Then = . = −t 2
dx dt dx dt
2 2
d y 2 d  2 dy  4 d y dy
and 2
= − t  − t  = t 2
+ 2t 3
dx dt  dt  dt dt
65
Ordinary Differential Substituting the above values in the given Eqn.(105), we get
Equations
1  4 d2y dy   3   2 dy 
t + 2t 3  +  − 1  − t +y =0
2 
t  dt 2
dt   t  dt 
d2y dy
i.e., t 2 2 − t (1 − t ) +y=0 (106)
dt dt
Now t = 0 is a regular singular point of the above equation and we look for its series
solution.
∞
Let y( t ) = ∑c
m =0
mt
m+k
, ( c o ≠ 0) (107)
∞
Then y ′( t ) = ∑ (m + k ) c
m =0
mt
m + k −1
(108)

∞
and y ′′( t ) = ∑ (m + k ) c
m =0
mt
m+ k −2
(109)

On substituting the above expression for y, y ′ and y ′′ in Eqn.(106), we get

∞ ∞

∑
m =0
(m + k − 1) 2 c m t m + k + ∑ (m + k − 1) c
m =1
m −1 t
m+k
=0 (110)

The indicial equation is c 0 (k − 1) 2 = 0 ⇒ k = 1, 1 .

Equating coefficients of t m + k in Eqn.(110) to zero, we get the recurrence relation as
(−1) c m −1
cm = , for m ≥ 1. (111)
m + k −1
(−1) m c 0
⇒ cm = , for m ≥ 1.
k (k + 1) L (k + m − 1)

Substituting the values of c 0 , c1 , K etc. in Eqn.(107), we obtain

(−1) m c 0 t m + k
∞
y( t ) = c 0 t k + ∑
m =1 k ( k + 1) L ((k + m − 1)
. (112)

Putting k = 1 in the above expression for y( t ) we get one solution of Eqn.(106).

Since roots of indicial equation are equal, to find second linearly independent solution,
we need to put k = 1 in
1 1 1 
∞
(−1) m  + +L+ 
∂y  k k +1 k + m − 1
∂k
= y ln t −
m =1
∑ k (k + 1) L (k + m − 1)
c 0 t m+k (113)

Thus, we get two linearly independent series solution of Eqn.(106) as

∞
(−1) m t m +1
y( t ) k =1 = c 0 t + ∑
m =1
c0
(m)!
= u , say

∂y  ( −1) m H m t m +1
∞
and  = u ln t − c 0
∂k  k =1 m =1 (m)!
∑ = v, say

1 1 1 m
1
where, H m = 1 + + +L+ =
2 3 m
∑  p  .
p =1

Therefore, two linearly independent solutions of Eqn.(105) are

∞ ∞ 1
(−1) m 1 c (−1) m − m c 0 x
u=
m =0
∑
c0
( m)! x m +1
= 0
x
∑
m = 0 ( m )!
x =
x
e

1
∞
(−1) m H m 1 m
1
and v = u ln   − c 0
 x  m =1
∑
(m)!
. m +1 , where H m =
x
∑  p  .
p =1
66
 Power Series Solutions
The general solution of Eqn.(105) is then given by y( x ) = au + bv, where a and b are
arbitrary constants.
***
You may now try this exercise.
E17) Find a series solution of the following equations for large values of x .
a) 4 x 3 y ′′ + 6 x 2 y ′ + y = 0 .
b) 2 x 2 (1 − x ) y ′′ − 5x (1 + x ) y ′ + (5 − x ) y = 0 .

In this unit we discussed methods of finding series solutions of second order linear,
homogeneous differential equations with variable coefficients. In the next two units,
we shall apply these methods to solve some differential equations, which occur
frequently in physical and engineering problems. In particular, we shall solve
Legendre’s, Bessel’s, Laguerre’s and Hermite’s differential equations.
We now end this unit by giving a summary of what we have covered in it.

2.5 SUMMARY
In this unit, we have learnt the following points:
1. Second order linear homogeneous differential equation with variable
coefficients cannot be, in general, solved by methods giving closed form
solutions and there is a need for series solutions for such equations.
∞
cn 1
2. For a power series, ∑
m =0
c m x m , if R = lim
n →∞ c
n +1
.
x
exists, then the power

series converges if | x | < R and diverges if | x | > R and the region of

convergence is − R < x < R . Here R is known as radius of convergence of the
power series.
3. A power series can be differentiated term by term any number of times without
affecting its radius of convergence.
∞
4. For the power series f ( x ) = ∑c
m=0
m ( x − a ) m , the r-times differentiated series
r
d [ f ( x )]
has the sum , for each positive integer r and the coefficients c m are
dx r
f ( m ) (a )
completely determined by c m = .
( m) !
5. A function f ( x ) is said to be analytic at x = a if, f ( x ) can be expanded in a
∞
power series ∑c
m =0
m (x − a ) m with a positive radius of convergence.

6. A point x = a is called an ordinary point of the differential equations

y ′′ + d 1 ( x ) y ′ + d 2 ( x ) = b1 ( x ) if d 1 ( x ) and d 2 ( x ) are analytic at x = a.
7. If x = a is not an ordinary point of differential equation
y ′′ + d 1 ( x ) y ′ + d 2 ( x ) y = 0, then it is called a singular point of the differential
equation. Further if ( x − a ) d 1 ( x ) and ( x − a ) 2 d 2 ( x ) are analytic at x = a ,
then x = a is called a regular singular point of the differential equation. A
singular point, which is not regular, is called an irregular singular point.
8. If x = 0 is an ordinary point of the differential equation
y′′ + d1 ( x ) y ′ + d 2 ( x ) y = 0 ,
∞
then power series solution is assumed as y( x ) = ∑c
m=0
m x m . Expressions for

y, y ′ and y ′′ are substituted in the given equation and c1 , c 2 K are determined

67
Ordinary Differential by comparing the coefficient of different power of x .
Equations 9. If x = 0 is a regular singular point of the differential Eqn(42), viz.,
f ( x ) y ′′( x ) + g ( x ) y ′( x ) + r ( x ) y = 0
then series solution of the equation is obtained by using the Frobenius
method where the series solution assumed is of the form
∞
y( x ) = ∑c
m=0
m x m + k , c 0 ≠ 0.

10. In the Frobenius method, equation obtained by equating to zero the coefficient
of smallest power of x is a quadratic equation in k , called the indicial
equation.
11. When the roots of indicial equation k 1 , k 2 (say) are unequal and do not differ
by an integer then two linearly independent series solutions of Eqn.(42) are
obtained corresponding to two values of k viz., k 1 , k 2 in the expression for
y( x ) .
12. When the roots of indicial equation k 1 , k 2 (say) are unequal, differ by an
integer and make a coefficient of y indeterminate for k = k 2 (say) then two
linearly independent solutions of Eqn(42) are obtained by putting k = k 2 in
y( x ) , which then contains two arbitrary constants.
13. When the roots of indicial equation k 1 , k 2 (say) are unequal, differ by an
integer and make a coefficient of y infinite for k = k 2 (say) , then the form of
y is modified by replacing c 0 by d 0 (k − k 2 ) . Two linearly independent
∂y
solutions are then obtained by substituting k = k 2 in y( x ) and .
∂k
14. When the roots of indicial equation are equal, then two linearly independent
solution of Eqn.(42) are obtained by substituting this value of k in y( x ) and
∂y
.
∂k

2.6 SOLUTIONS/ANSWERS
n! n
E1) For series (5), R = lim = lim =0.
n →∞ (n + 1)! n → ∞ n +1
1
For series (6), R = lim n! = lim (n + 1)! = lim n + 1 → ∞
n →∞ 1 n →∞ n! n →∞

( n + 1)!
1
For series (7), R = lim = 1.
n →∞ 1

1
n
x
E2) For the series ∑ n!
, R = lim
n →∞
n! = lim (n + 1) → ∞ .
1 n →∞

(n + 1)!
Hence the given series converges for all x .

E3) a) We have, 1 − 2 x + 3x 2 − 4x 3 + L = ∑ (−1) n (n + 1) x n

 1
n − 1 + 
(−1) (n + 1)  n
R = lim = lim =1.
n →∞ ( −1) n +1 ( n + 2) n →∞  2
1 + 
 n
68
 Power Series Solutions
(kn )! [(n + 1)!] k
b) We have, R = lim .
n →∞ ( n! ) k [ k ( n + 1)]!

(n + 1) k (n + 1) k −1
= lim = lim → ∞ , for k > 1
n →∞ k ( n + 1) n →∞ k
(n + 1) k −1
and lim = 1 for , k = 1 .
n →∞ k

E4) Comparing the given equations with a 0 ( x ) y ′′ + a 1 ( x ) y ′ + a 2 ( x ) y = 0 ,

a) Here a 0 ( x ) = 1, a 1 ( x ) = 4x , a 2 ( x ) = 1
a (x) a (x )
Now ( x − a ) 1 = ( x − a ) 4 x and ( x − a ) 2 2 = ( x − a ) 2 , are analytic at
a 0 (x) a 0 (x)
x = a . Hence x = a is an ordinary point of given differential equation for all
real finite a , including a = 0 .

b) Here a 0 ( x ) = x 3 ( x − 2), a 1 ( x ) = x 3 , a 2 ( x ) = 6
a 1 (x) (x − a ) x 3
Now ( x − a ). = , is analytic for a = 2
a 0 ( x ) x 3 ( x − 2)
a 2 ( x ) ( x − a ) 2 .6
Also, ( x − a ) 2 . = is analytic for a = 2 and singular for x = 0 .
a 0 ( x ) x 3 ( x − 2)
Hence x = 0 is an irregular singular point, and x = 2 is a regular singular point
for the given differential equation.
c) x = −1 , is singular point.
a 2 (x) 2
d) x = 0 , is an irregular singular point, since x 2 = x 2 . 4 → ∞ at x = 0 .
a 0 (x ) x
sin x ex
e) x = 0 , is a regular singular point, since x and x 2 . are analytic at
x x
x =0.
E5) The given D.E. is
( x 2 − 2 x + 2) y ′′ − 4( x − 1) y ′ + 6 y = 0
In this case, we change the independent variable x to u by the relation
x − 1 = u , the given D.E. becomes
(u 2 + 1) y ′′ − 4 uy ′ + 6 y = 0 , (114)
where y ′ and y ′′ now denote the derivatives of y w.r.t. u . We are now
required to solve Eqn.(114) about the point u = 0 .
4u 6
We see that the coefficients of y ′ and y , i.e., − 2 and 2 are
u +1 u +1
analytic everywhere. Hence we assume solution of Eqn.(114) as
∞
y= ∑
m =0
cm u m (115)

Substituting for expressions for y, y ′, y ′′ from (115) into Eqn.(114), we get

∞ ∞ ∞
(u 2 + 1) ∑
m=2
c m m(m − 1)u m − 2 − 4u ∑
m =1
c m m u m −1 + 6 ∑c
m=0
mu
m
=0

or,
∞
(2c 2 + 6c 0 ) + (2c1 + 6c 3 )u + ∑ [(m + 2) (m + 1)c
m=2
m+ 2 ]
+ ( m 2 − 5m + 6) c m u m = 0

Setting the coefficients of successive powers of u to zero, we get

2c 2 + 6c 0 = 0, 2c1 + 6c 3 = 0 and (m + 2) ( m + 1) c m + 2 + (m − 2) (m − 3) c m = 0
for m ≥ 2 . 69
Ordinary Differential where c 0 , c1 are arbitrary constants. Solving, we get
Equations
− c1 (m − 2) (m − 3)
c 2 = −3c 0 , c 3 = and c m + 2 = − c m , for m ≥ 2 .
3 ( m + 2) (m + 1)
2.1 3 .2
⇒ c 4 = 0, c 5 = 0, c 6 = − c 4 = 0, c 7 = − c5 = 0
6 .5 7 .6
∴ The solution of given D.E. is
 1 
y = c 0 + c1 u + c 2 u 2 + c 3 u 3 = c 0 (1 − 3u 2 ) + c1  u − u 3 
 3 
 
[ ] 1
= c 0 1 − 3( x − 1) 2 + c1 ( x − 1) − ( x − 1) 3  .
 3 

E6) The given D.E. is

y ′′ + xy ′ + 3y = 0 (116)
For the given equations, x = 0 is an ordinary point and we shall obtain the
solution about x = 0 .
∞
Let y( x ) = ∑
m=0
cmx m .

Then Eqn.(116) becomes

∞ ∞ ∞

∑
m=2
m (m − 1)c m x m − 2 + x ∑
m =1
m c m x m −1 + 3 ∑c
m=0
mx
m
=0

∞
or, (2c 2 + 3c 0 ) + ∑ {(m + 2) (m + 1) c
m =1
m+ 2 + (3 + m) c m =0 } x m

Equating the coefficients of similar powers of x , we get

−2
2c 2 + 3c 0 = 0 ⇒ c 2 = c0
3
−2
6c 3 + 4c 1 = 0 ⇒ c 3 = c1
3
Indeed, (m + 2) ( m + 1) c m + 2 + (m + 3) c m = 0 , for m ≥ 1 .
5 5 .2
∴ c4 = − c 2 = (−1) 2 . c0
4 .3 4.3.3
6 2 .6
c5 = − c 3 = (−1) 2 c1
5 .4 3.4.5
7 7.5.2
c6 = − c 4 = (−1) 3 c0
6 .5 6.5.4.3.2
8 8.6.2
c7 = − c 5 = (−1) 3 c1
7 .6 7.6.5.4.3
LLLLLLLLLLLLLL
where c 0 and c1 are arbitrary constants.
Hence the general solution is
 2 5  2 2 7 .5  2  3 
y( x ) = co 1 − x + (−1) 2   x + (−1)
3
  x + L
 3 4 .3  3  6 .5 .4 .3  3  
  2 6 2 4 8.6  2  5 
+ c1  x −   x 3 + (−1) 2   x + (−1)
3
  x + L
  3 4 .5  3  7.6.5.4  3  

E7) The given initial value problem is

xy ′′ + y ′ + 2 y = 0

y(1) = 1  (117)
y ′(1) = 2 

70
 Power Series Solutions
Evidently, x = 1 is an ordinary point.
∞
Let y= ∑c
m=0
m (x − 1) m (118)

Differentiating twice in succession, we get

∞
y′ = ∑ mc
m =1
m (x − 1) m −1 (119)
∞
and y ′′ = ∑ m(m − 1) c
m =2
m (x − 1) m − 2 (120)

Substituting for y, y ′, y ′′ in Eqn.(117), we get

∞ ∞ ∞
x ∑
m=2
m(m − 1)c m ( x − 1) m − 2 + ∑
m =1
m c m ( x − 1) m −1 + 2 ∑c
m=0
m (x − 1) m = 0
∞
or, ( 2c 2 + c 1 + 2c 0 ) + ∑ {m(m + 1)c
m =1
m +1 + (m + 2) (m + 1) c m + 2

+ (m + 1)c m +1 + 2c m }( x − 1) m = 0
 c 
∴ 2c 2 + c1 + 2c 0 = 0 i.e., c 2 = − c 0 + 1  (121)
 2
and (m + 1) mc m +1 + (m + 2) ( m + 1) c m + 2 + (m + 1) c m +1 + 2c m = 0 , for m ≥ 1
i.e., (m + 1) (m + 2) c m + 2 + (m + 1) 2 c m +1 + 2c m = 0 , for m ≥ 1 (122)
using initial conditions given by (117) in relations (118) and (119), we get
c 0 = 1 and c1 = 2
From (121) and above, c 2 = −2
For m = 1, 2, 3, K in Eqn.(122) and using above values, we get
6c 3 + 4c 2 + 2c 1 = 0 ⇒ c 3 = 2 / 3
12c 4 + 9c 3 + 2c 2 = 0 ⇒ c 4 = −1 / 6
20c 5 + 16c 4 + 2c 3 = 0 ⇒ c 5 = 1 / 15.
and so on.
2 1 1
Hence y = 1 + 2( x − 1) − 2( x − 1) 2 + ( x − 1) 3 − ( x − 1) 4 + ( x − 1) 5 + L .
3 6 15

E8) The given D.E. is y ′′ + xy ′ + y = 0 (123)

Here x = 2 , is an ordinary point.
∞
Let y = ∑c
m=0
m (x − 2) m (124)

Substituting for y, y ′, y ′′ in Eqn.(123), we get

∞ ∞ ∞

∑
m=2
c m ( x − 2) m − 2 m(m − 1) + x ∑
m =1
m c m ( x − 2) m −1 + ∑c
m =0
m (x − 2) m = 0

∞
or, (2.1.c 2 + 2.1.c1 + c 0 ) + ∑ { (m + 2) (m + 1) c
m =1
m+2 + m cm

+ 2(m + 1)c m +1 + c m }( x − 2) m = 0 .
1 
∴ 2c 2 + 2c 1 + c 0 = 0 ⇒ c 2 = − c 1 − c 0 
2  (125)
and ( m + 2) (m + 1) c m + 2 + 2(m + 1) c m +1 + (m + 1)c m = 0 , for m ≥ 1 
For m = 1, 2, 3, K and using Eqn.(125), we get
1
3.2.c 3 + 2.2 c 2 + 2c1 = 0 ⇒ c 3 = (c1 + c 0 )
3
1 1
4.3.c 4 + 2.3 c 3 + 3c 2 = 0 ⇒ c 4 = c1 − c0
3 .4 2 .4 .3
71
Ordinary Differential 1 1
Equations 5.4.c 5 + 2.4 c 4 + 4 c 3 = 0 ⇒ c 5 = c1 + c 0
5 .6 12
and so on. Hence Eqn.(124) yields
 1 1 1 1 
y( x ) = c 0 1 − ( x − 2) 2 + ( x − 2) 3 − ( x − 2) 4 + ( x − 2) 5 + L
 2 3 24 12 
 1 1 1 
+ c1 ( x − 2) − ( x − 2) 2 + ( x − 2) 3 + ( x − 2) 4 + ( x − 2) 5 + L .
 3 12 30 

E9) The given Hermite equation of order n is

y ′′ − 2 xy ′ + 2ny = 0 (126)
∞
If z n ( x ) = ∑a
k =0
kx
k
is a solution of Eqn.(126), then substituting for y, y ′, y ′′

in Eqn.(126), we get
∞ ∞ ∞

∑a
k=2
k k (k − 1) x k − 2 − 2x ∑ka
k =1
kx
k −1
+ 2n ∑a
k =0
kx
k
=0
∞
or, 2.1.a 2 + 2n a 0 + ∑ [(k + 2) (k + 1) a
k =1
k+2 − 2k a k + 2n a k ] x k = 0

Equating coefficients of successing powers of x , we get

2a 2 + 2 n a 0 = 0 ⇒ a 2 = − n a 0
and (k + 2) (k + 1) a k + 2 − 2 (k − n ) a k = 0 , for k ≥ 1
putting k = 1, 2, 3, K in the above relation, we get
2(1 − n )
3.2. a 3 − 2(1 − n ) a 1 = 0 ⇒ a 3 = a1
2 .3
2( 2 − n ) 2n (2 − n )
4 .3 a 4 − 2 ( 2 − n ) a 2 = 0 ⇒ a 4 = a2 = − a0
4 .3 4.3
2( 3 − n ) 2 2 (1 − n ) (3 − n )
5.4 a 5 − 2(3 − n ) a 3 = 0 ⇒ a 5 = a3 = a1
5 .4 2.3.4.5
2( 4 − n ) 2 2 n (2 − n ) (4 − n )
6 .5 a 6 − 2 ( 4 − n ) a 4 = 0 ⇒ a 6 = a4 = − a0
6 .5 6.5.4.3
and so on.
Hence the solution is
 2(− n ) (2 − n ) 4 2 2 (− n ) (2 − n ) (4 − n ) 6 
z n ( x ) = y( x ) = a 0 1 + (−n ) x 2 + x + x + L
 4 .3 6.5.4.3 
 2(1 − n ) 3 2 2 (1 − n ) (3 − n ) 5 
+ a 1 x + x + x + L
 2.3 2.3.4.5 
 2(−n ) 2 2 2 ( −n ) (2 − n ) 4 2 3 (−n ) (2 − n )(4 − n ) 6 
= a 0 1 + x + x + x + L
 2 4.3.2 6.5.4.3.2 
 2
2(1 − n ) 3 2 (1 − n ) (3 − n ) 5 
+ a 1 x + x + x + L
 2.3 2.3.4.5 
 ∞
2 (−n ) (−n + 2) K (−n + 2k − 2) 2 k 
k
= a 0 1 + ∑ ( 2k )!
x 
 k =1 
 ∞
2 (1 − n ) (1 − n + 2) K (1 − n + 2k − 2) 2 k +1 
k
+ a 1 x + ∑ (2k + 1) !
x 
 k =1 

E10) a) The given equation

2 x (1 − x ) y ′′ + (1 − x ) y ′ + 3y = 0 (127)
has x = 0 , as a regular singular point
72
 ∞ Power Series Solutions
Let y( x ) = ∑
m =0
cm x k +m

Putting the values of y, y ′, y ′′ in Eqn.(127), we get

∞

∑ {3 − (k + m) − 2(k + m) (k + m − 1)}c
m =0
mx
k +m

∞
+ ∑ {2 (k + m − 1) + 1}(k + m)c
m =0
mx
k + m −1
=0 (128)

Equating the coefficients of lowest power of x to zero in Eqn.(128), we get the

indicial equation as c 0 (2k − 2 + 1) k = 0 i.e., c 0 k (2k − 1) = 0 ⇒ k = 0 and
k = 1 / 2 , i.e., the roots of indicial equation are distinct, and do not differ by an
integer. To get the recurrence relation, we equate coefficients x k + m in
Eqn.(128) to zero and obtain
[3 − (k + m) − 2(k + m) (k + m − 1)]c m + {2(k + m + 1 − 1) + 1}(k + m + 1)c m+1 = 0
2(k + m) (k + m − 1) + (k + m) − 3
i.e., c m +1 = cm
(k + m + 1) ( 2k + 2m + 1)
Putting m = 0, 1, 2, 3, K
2k 2 − k − 3
c1 = c0
(k + 1) (2k + 1)

c2 =
{2 (k + 1) k + (k + 1) − 3}{2k (k − 1) + k − 3} c
0
(k + 1) ( k + 2) (2k + 1) (2k + 3)

c3 =
{2(k + 2) (k + 1) + (k + 2) − 3}{2(k + 1)k + (k + 1) − 3}{2k (k − 1) + k − 3} c
0
(k + 1) (k + 2) (k + 3) ( 2k + 1) (2k + 3) (2k + 5)
∞
Putting values of c1 , c 2 , c 3 , K in y( x ) = ∑
m =0
c m x k + m , we get

 2k − k − 3 2
(2k 2 + 3k − 2) (2k 2 − k − 3)
y ( x ) = x k c 0 + c0 x + c0 x 2
 ( k + 1 ) ( 2 k + 1) ( k + 1) ( k + 2 ) ( 2 k + 1 ) ( 2 k + 3 )
(2k 2 + 7 k + 3) (2k 2 + 3k − 2) (2k 2 − k − 3) 
+ c 0 x 3 + L (129)
(k + 1) (k + 2) (k + 3) (2k + 1) (2k + 3) (2k + 5) 
Putting k = 0 in Eqn.(129), we get
 (−2) (−3) 2 3( −2)(−3) 3 
y1 = c 0 1 − 3x + x + x + L
 1 .2 . 1 . 3 1 . 2 . 3 . 1 . 3 . 5 
 3 2 3 
= c 0 1 − 3x + x + x 3 + L = c 0 u , say (130)
 1 .3 1 .3 .5 

Putting k = 1/2 in Eqn.(129), we get

y 2 = c 0 x 1 / 2 [1 + (− x ) + 0 + 0 + L] = c 0 x 1 / 2 (1 − x ) = c 0 v , say (131)

Hence the general solution is y( x ) = A1 y + B1 y 2 = Au + Bv

where A = A 1c 0 , B = B1c 0 are arbitrary constants and u and v are given by
Eqn.(130) and Eqn.(131) respectively.
b) The given equation
(2 x + x 2 ) y ′′ − y ′ − 6xy = 0 (132)
has x = 0 as a regular singular point.
∞
Let y( x ) = ∑
m =0
c m x k + m be a solution of Eqn.(132). Putting for y, y ′, y ′′ in

Eqn.(132), we get
73
Ordinary Differential ∞ ∞
Equations −6 ∑
m =0
c m x k + m +1 + ∑ (k + m) (k + m − 1) x
m =0
k +m

∞
+ ∑
m =0
(k + m) ( 2k + 2m − 3) c m x k + m −1 = 0 (133)

Equating the coefficient of lowest power of x to zero in Eqn.(133), we get

the indicial equations as k (2k − 3) c 0 = 0 ⇒ k = 0 and k = 3 / 2 , i.e., the
roots of indicial equation are distinct, and do not differ by an integer. To
get the recurrence relation, equate the coefficient of x k + m +1 to zero is
relation (133) and obtain
− 6c m + (k + m + 1) (k + m) c m +1 + (k + m + 2) (2k + 2m + 1) c m + 2 = 0 (134)
Equating coefficient of x k in (133) to zero, we get
k (k − 1)
k (k − 1) c 0 + (k + 1) ( 2k − 1) c1 = 0 ⇒ c1 = − c0
(k + 1) ( 2k − 1)
Putting m = 0, 1, 2, 3, K in Eqn.(134), we get
6(2k − 1) + k 2 (k − 1)
c2 = c0
(k + 2) (2k − 1) (2k + 1)
6k (k − 1) (k + 1) [ k 3 − k 2 + 12k − 6 ]
c3 = − c0 − c0
(k + 1) (2k − 1) (k + 3) (2k + 3) (k + 3) (2k − 1) (2k + 1) (2k + 3)
and so on.
For k = 0 , we get
−6 1(−6) 2
c1 = 0, c 2 = c 0 = 3 c 0 , c3 = − c 0 = − c 0 ,L
2(−1) (1) 3(−1).1.3 3
For k = 3/2 , we get
3 15
c1 = c 0 , c 2 = c 0 ,L
4 .5 32
 2 
Hence y1 = c 0 1 + 3x − x 2 + L = c 0 u , say (135)
 3 
 3 15 2 
and y 2 = c 0 x 3 / 2 1 + x+ x + L = c 0 v , say (136)
 4 .5 32 
Hence the general solution is y( x ) = A 1 y1 + B1 y 2 = Au + Bv , where
A = c 0 A 1 and B = c 0 B1 are arbitrary constant and u and v are given by
relations (135) and (136) respectively.

E11) The given D.E. is

(2 + x 2 ) y ′′ + xy ′ + (1 + x ) y = 0
∞
Let y = ∑
m=0
c m x k + m , c 0 ≠ 0 , be a series solution of the given equation. Then

putting the values of y, y ′, y ′′ in the given equation, we get

∞ ∞
(2 + x 2 ) ∑ (m + k) (m + k − 1) c
m=2
mx
k +m−2
+x ∑ (m + k ) c
m =1
mx
m + k −1

∞
+ (1 + x ) ∑c
m =0
mx
m+k
=0

⇒ 2c 0 k ( k − 1) x k − 2 + 2c1 k (k + 1) x k −1 [2c 2 (k + 2)(k + 1) + c 0 k (k − 1) + c 0 k + c 0 ] x k

∞
+ ∑ [2c
m =3
m (k + m) (k + m − 1) + c m − 2 (k + m − 2) (k + m − 3)

+ c m − 2 (k + m − 2) + c m − 2 + c m −3 ] x k + m −3 = 0
Equating coefficients of lowest powers of x , we get indicial equation as
74
 Power Series Solutions
2c 0 k (k − 1) = 0 ⇒ k = 0,1 as c 0 ≠ 0
Equating to zero coefficients of other powers of x , we get
2c1 (k + 1) k = 0
This shows that c1 becomes indeterminate for k = 0 .
∴ 2c 2 (k + 2)(k + 1) + c 0 k (k − 1) + c 0 k + c 0 = 0
k2 +1
c2 = − c0
2(k + 1)(k + 2)
The general recurrences relation is
[ ]
2c m (k + m) (k + m − 1) + c m − 2 (k + m − 2) 2 + 1 + c m −3 = 0 for m ≥ 3
Putting m = 3, 4, 5, K , we get
[ (k + 1) 2 + 1 ] c1 c0
c3 = − −
2(k + 3) (k + 2) 2( k + 3) (k + 2)
c1 ( k 2 + 1) (k 2 + 4k + 5) c 0
c4 = − + (−1) 2 2
2(k + 3) (k + 4) 2 ( k + 1) (k + 2) (k + 3) (k + 4)
and so on.
Putting these values in the assumed solution, we get
k

y = x c 0 + c 1 x −
(k 2 + 1) c 0 
x2 − 
c0
+
{
( k + 1) 2 + 1 c1  3
x
}
 2(k + 1) ( k + 2)  2(k + 2)(k + 3) 2( k + 3) (k + 2) 

+ −
c1
+ 2
{
(k 2 + 1) (k + 2) 2 + 1  } 
c 0 x 4 + L
 2(k + 4) (k + 3) 2 (k + 1) ( k + 2) (k + 3) (k + 4)  
Putting k = 0 , we get
 1 1 5 
y = c 0 1 − x 2 − x3 + 2 x 4 + L
 2 2 .2 .3 2 .1.2.3.4 
 1 1 
+ c1  x − x 3 − x 4 + L
 6 2 .3 .4 
as the required solution, where c 0 and c1 are arbitrary constants.

E12) The given equation is (1 − x 2 ) y ′′ + 2xy ′ + y = 0 .

Let the series solution of given equation be
∞
y( x ) = ∑c
m=0
mx
m+ k
, c0 ≠ 0 (137)

substituting for y, y ′, y ′′ in the given equation, we get

∞
c 0 k (k − 1) x k − 2 + c1 k (k − 1) x k −1 + ∑ [− (m + k ) (m + k − 1) c
m =0
m + 2( m + k ) c m +

c m + (m + 2 + k ) (m + k + 1) c m + 2 ] x m + k = 0
The indicial equation is c 0 k ( k − 1) = 0 ⇒ k = 0, 1 as c 0 ≠ 0 . (138)
Thus indicial roots are distinct and differ by an integer.
Equating the coefficients of x k −1 to zero, we get
c1 k (k + 1) = 0 (139)
When k = 0, c1 becomes indeterminate (0 / 0) . But in this case, we get the
identity c1 .0 = 0 , which is satisfied by every values of c1 . Thus, we can also
take c1 as arbitrary constant.
The recurrence relation is
(m + k + 2) (m + k + 1) c m + 2 + [1 − (m + k ) (m + k − 1) + 2 (m + k ) ]c m = 0
75
Ordinary Differential (m + k ) 2 − 3(m + k ) − 1 ( m + k ) (m + k − 3) − 1
Equations ⇒ c m+ 2 = cm = cm
(m + k + 2) (m + k + 1) (m + k + 2) (m + k + 1)
Putting m = 0, 1, 2, K , we get
k 2 − 3k − 1 k (k − 3) − 1
c2 = c0 = c0
(k + 2) (k + 1) (k + 2) (k + 1)
(k + 1) (k − 2) − 1
c3 = c1
(k + 3) (k + 2)

c4 =
{ (k + 2) (k − 1) − 1 }{k (k − 3) − 1} c
0
(k + 4) (k + 2) 2 (k + 1)

c5 =
{ (k + 3) (k ) − 1}{ (k + 1) (k − 2) − 1} c
1
(k + 5) (k + 3) 2 (k + 2)
LLLLLLLLLLLLLLLLLLL
Substituting these values of coefficients in series for y( x ) , we get
 k (k − 3) − 1 ( k + 1) (k − 2) − 1
y = x k c 0 + c 1 x + c0x 2 + c1 x 3
 (k + 2) (k + 1) (k + 3) (k + 2)

+
{ (k + 2) (k − 1) − 1}{k (k − 3) − 1} c x 4
o
(k + 4) (k + 2) 2 (k + 1)

+
{ (k + 3)k − 1}{(k + 1) (k − 2) − 1} c x 5 + L
1  (140)
(k + 5) (k + 3) 2 (k + 2) 

Substituting k = 0 in this, we get

 1 3   1 (−3) 5 
y = c 0 1 − x 2 − x 4 + L + c1  x − x 3 + x − L (141)
 2 16   2 90 
where c 0 and c1 are arbitrary constants.
This is the required general series solution.
Note that if we substitute k = 1 in the series for y( x ) , we get from relation
(139) that c1 = 0 and hence c 3 , c 5 , K all will vanish and then relation (140)
gives
  1  3  
y = x c 0 +  −  x 2 c 0 +  −  x 4 c 0 + L (142)
  2  90  
which is the first series in Eqn.(141) and hence it is not a new series.

E13) a) The given equation is

xy ′′ + (3 − x ) y ′ + y = 0
Here x = 0 , is a regular singular point.
∞
Let y( x ) = ∑
m=0
c m x m+ k , c 0 ≠ 0 . (143)

Substituting for y, y ′, y ′′ in the given equation, we get

∞
[k (k − 1) + 3k ]c 0 x k −1 + ∑ [{ (m + k + 1) (m + k ) + 3(m + k + 1)}c m+1
m =0

− (m + k − 1) c m ] x m + k = 0
Indicial equation is
k (k − 1) + 3k = 0 , as c 0 ≠ 0 , k = 0,−2
The recurrence relation is
[ (m + k + 1)(m + k ) + 3 (m + k + 1)]c m+1 − (m + k − 1) c m = 0, m ≥ 0
m + k −1
or, c m +1 = cm , m ≥ 0
(m + k + 1) (m + k + 3)
76
 Power Series Solutions
Putting m = 0, 1, 2, 3, K
k −1
c1 = c0
(k + 1) (k + 3)
k (k − 1)
c2 = c0
(k + 1) (k + 2) ( k + 3) (k + 4)
(k − 1)k
c3 = c0
(k + 2) (k + 3) 2 (k + 4) (k + 5)

(k − 1) k
c4 = c0
(k + 3) (k + 4) 2 (k + 5) (k + 6)
2

and so on.
 k −1 (k − 1) k
∴ y( x ) = c 0 x k 1 + x+ x2
 (k + 1) )k + 3) (k + 1) ( k + 2) (k + 3) (k + 4)
(k − 1) k
+ x3
(k + 2) (k + 3) 2 ( k + 4) (k + 5)
(k − 1) k 
+ 2 2
x 4 + L
(k + 3) (k + 4) ( k + 5) (k + 6) 

For k = −2 , coefficients of x 2 and x 3 become infinite. Replacing c 0 by

a (k + 2) , the above series solution takes the form
 (k − 1) (k + 2) (k − 1) k
y( x ) = ax k (k + 2) + x+ x2
 ( k + 1) ( k + 3) ( k + 1) ( k + 3) ( k + 4 )
(k − 1) k (k − 1) k (k + 2) 
+ 2
x3 + 2 2
x 4 + L (144)
(k + 3) (k + 4) (k + 5) (k + 3) (k + 4) (k + 5) (k + 6) 
∂y  ( k − 1) (k + 2) 
∴ = ax k ln x (k + 2) + x + L
∂k  (k + 1) (k + 3) 
  2k + 1 (k − 1) (k + 2)  1 1 
+ ax k 1 +  −  +  x
  (k + 1) (k + 3) (k + 1) (k + 3)  k + 1 k + 3  
 2k − 1 k (k − 1)
+ −
 (k + 1) (k + 3) ( k + 4) (k + 3) (k + 4) ( k + 1)
 1 1 1  2
 + +  x
 k + 1 k + 3 k + 4 
 2k − 1 k (k − 1)
+ 2
− 2
 (k + 3) (k + 4) (k + 5) (k + 3) (k + 4) (k + 5)
 2 1 1  3 
 + +  x + L (145)
 k + 3 k + 4 k + 5  

Putting k = −2 in Eqns.(144) and (145), we get

y = ax −2 [ − 3x 2 + x 3 ] = a y1 , say
∂y  11 1 
= a (ln x ) y 1 + ax − 2 1 + 3x + 4 x 2 − x 3 + x 4 − L = ay 2 , say
∂k  3 8 
∴ The general solution is
y( x ) = c1 y 1 + c 2 y 2 .

b) Given equation is
77
Ordinary Differential x (1 − x ) y ′′ − 3xy ′ − y = 0
Equations
Here x = 0 , is a regular singular point.
∞
Let y( x ) = ∑
m=0
c m x m+ k , c 0 ≠ 0

Substituting for y, y ′, y ′′ in the given equation, we get

∞ ∞

∑c
m=0
m (m + k ) (m + k − 1) x m + k −1 − ∑ (m + k + 1)
m=0
2
c m x m+ k = 0

Equating to zero the coefficients of lowest power of x , i.e., x k −1 , the indicial

equation is
c 0 k (k − 1) = 0 or k = 0,1 (Q c 0 ≠ 0)
There are unequal roots, differing by an integer.
The recurrence relation is
c m ( m + k + 1) 2 − c m +1 ( m + k + 1) (m + k ) = 0 for m ≥ 0
m + k +1
∴c m +1 = c m for m ≥ 0
m+k
Putting m = 0, 1, 2, 3, K , we get
k +1 k+2 k+3
c1 = c0 , c2 = c0 , c3 = c0
k k k
and so on.
Putting these values in the series solution for y( x ) , we get
 k +1 k+2 k+3 
y( x ) = x k c 0 + c0 x + c0x 2 + c 0 x 3 + L
 k k k 
2 3
If we put k = 0 , we find that coefficients of x , x , x , K become infinite.
To remove this difficulty, we put c 0 = k d 0 in the above series, which then
[
becomes y( x ) = d 0 x k 1 + (k + 1) x + (k + 2) x 2 + (k + 3) x 3 + L . ]
Putting k = 0 , and replacing d 0 by a , we get
y( x ) = a ( x + 2x 2 + 3x 3 + L) = a u 1 , say.
If we put k = 1 to obtain second solution, we obtain
y( x ) = c 0 ( x + 2 x 2 + 3x 3 + L) , which is not distinct from above solution.
 ∂y 
In such a case, second independent solution is given by   .
 ∂k  k =0
∂y
Now = d 0 x k ln x{k + (k + 1) x + (k + 2) x 2 + L] + d 0 x k [1 + x + x 2 + L]
∂k
Putting k = 0 and replacing d 0 by b in the above, we get
 ∂y 
  = b[ln x ( x + 2 x 2 + 3x 3 + L) + (1 + x + x 2 + L)]
 ∂k  k = 0
= b [u ln x + (1 + x + x 2 + L)] = bv , say
and required solution is y( x ) = au + bv , where a and b are arbitrary constants.

E14) The given differential equation is

x (1 − x ) y ′′ + ((1 − x ) y ′ − y = 0 (146)
Here x = 0 , is a regular singular point of the above equation.
∞
Let y( x ) = ∑
m=0
c m x m+ k , c 0 ≠ 0 .

Substituting for y, y ′, y ′′ in the given equation, we get

∞
k 2 c 0 x k −1 + ∑ [(m + k + 1)
m =0
2
{ } ]
c m +1 − (m + k ) 2 + 1 c m x m + k = 0

78
 Power Series Solutions
Equating coefficients of least power of x , the indicial equation is
c 0 k 2 = 0 ⇒ k = 0, 0 (Q c 0 ≠ 0) .
The recurrence relation is
(m + k ) 2 + 1
c m +1 = cm , m ≥ 0 .
(m + k + 1) 2
k2 +1
Hence c1 = c0
(k + 1) 2
(k 2 + 1) [ (k + 1) 2 + 1 ]
c2 = c0
(k + 1) 2 (k + 2) 2
(k 2 + 1) [ (k + 1) 2 + 1 ] [(k + 2) 2 + 1 ]
c3 = c0
(k + 1) 2 (k + 2) 2 (k + 3) 2
and so on.
 k2 +1 (k 2 + 1) {( k + 1) 2 + 1} 2
∴ y( x ) = c 0 x k 1 + 2
x + x
 (k + 1) (k + 1) 2 (k + 2) 2
(k 2 + 1) {(k + 1) 2 + 1}{ (k + 2) 2 + 1} 3 
+ 2 2 2
x + L (147)
(k + 1) (k + 2) (k + 3) 
Hence solution for k = 0 is
 1.2 1 .2 .5 
[ y( x )] k =0 = c 0 x 1 + x + 2 2 x 2 + 2 2 2 x 3 + L = y1 , say (148)
 1 .2 1 .2 .3 
with y given by Eqn.(147), left hand side of Eqn.(146) becomes k 2 c 0 x k −1
∂
∴ [x (1 − x ) y ′′ + (1 − x ) y ′ − y] = 2k c 0 x k −1 + c 0 k 2 x k −1 ln x
∂k
 d2 d  ∂y
or,  x (1 − x ) 2 + (1 − x ) − 1 = 2kc 0 x k −1 + c 0 k 2 x k −1 ln x
 dx dx  ∂k
∂y
Therefore, for k = 0, is a second solution of Eqn.(146).
∂k
From Eqn.(147), we get
 ∂y   
 
1 1 .2 1 .2 .5
[
= c 0 ln x 1 + 2 x + 2 2 x 2 + 2 2 2 x 3 + L + c 0 − 2 x − x 2 − L ]
  k =0
∂k  1 1 .2 1 .2 .3 
[ ]
= y1 ln x + c 0 − 2 x − x 2 − L = y 2 , say
since y1 and y 2 are linearly independent, hence the general solution is
y = α1 y1 + α 2 y 2 ,
where α 1 , α 2 are arbitrary constants.

E15) a) Proceeding as in E14) when p = 1 , y = au + bv

 x x2 x3 
where, u = 1 + + + + L
 1! 2 ! 3! 
x x 
2
1  x3  1 2 
and v = u ln x −  + 1 +  + 1 + +  + L .
 1 2!  2  3!  2 3 
b) When p is not a integer, y = AU + BV
x x2 x3
where, U = 1 + + + +L
p p(p + 1) p(p + 1) (p + 2)
 x x2 x3 
V = x 1− p 1 + + + + L .
 1! 2 ! 3! 

E16) a) Proceeding as in E14), the solution is y = au + bv ,

79
Ordinary Differential 3 2 4 3
Equations where, u = 1 − 2x − x − x −L
2 3
  1 3 1 1 4 1 1 1 
and v = u ln x + 2 2 −  x −  2 + −  x 2 −  2 + + +  x 3 + L
  2 2!  2 3 3!  2 3 4 

b) The given equation is

( x − x 2 ) y ′′ + (1 − 5x ) y ′ − 4 y = 0
Evidently x = 0 , is a regular singular point of the equation.
∞
Let y( x ) = ∑c
m =0
mx
m+k
, c0 ≠ 0 . (149)

Substituting for y, y ′, y ′′ in the given equation, we get

∞ ∞

∑
m =0
c m (m + k ) 2 x m + k −1 − ∑c
m =0
m (m + k + 2) 2 x m + k = 0 ,

Equating to zero, the coefficients of x k −1 , the indicial equation is

c 0 k 2 = 0 or k = 0, 0 (Q c 0 ≠ 0) .

(m + k + 1) 2
The recurrence relation is c m = c m −1 , for m ≥ 1 .
(m + k ) 2
( k + 2) 2 (k + 3) 2
Putting m = 1, 2, 3, K , we get c1 = c 0 , c 2 = c0
( k + 1) 2 (k + 1) 2
( k + 4) 2 (k + 5) 2
c3 = c 0 , c 4 = c 0 ,K
(k + 1) 2 (k + 1) 2
Putting the values of coefficients c 0 , c1 , c 2 , K in solution (149), we get
 ( k + 2) 2 (k + 3) 2 2 (k + 4) 2 3 
y( x ) = x k c 0 1 + 2
x + 2
x + 2
x + L (150)
 (k + 1) (k + 1) (k + 1) 
Putting k = 0 in Eqn.(150), we get
y( x ) k =0
[ ]
= c 0 1 + 2 2 x + 3 2 x 2 + 4 2 x 3 + L = c 0 u , (say) (151)

Differentiating Eqn.(150) partially, w.r.t. k and putting k = 0 , the second

solution is obtained as
 ∂y 
  [
= c 0 ln x 1 + 2 2 x + 3 2 x 2 + 4 2 x 3 + L ]
 ∂k  k = 0
 2  2  
+ c 0  2 2 x (1 − 2 ) + 3 2 x 2  − 2  + 4 2 x 3  − 2  + L
 3  3  
[ 2 3
= c 0 u ln x − c 0 1.2 x + 2.3x + 3.4x + L = c 0 v , (say)] (152)
Hence the required general solution is
y = au + bv .
c) Proceeding as in E14), the general solution is y = au + bv ,
 1 .3 1.3.5.7 
where, u = x 1 / 2 1 + 2 x 2 + 2 2 x 4 + L
 4 4 .8 
1 .3  1 1  1.3.5.7
and v = u ln x + 2x 1 / 2  2 1 + −  x 2 + 2 2
4  3 2  4 .8
 1 1 1 1 1 4 
1 + − + + −  x + L
 3 2 5 7 4 
80
 Power Series Solutions
E17) a) The given equation is
4 x 3 y ′′ + 6 x 2 y ′ + y = 0 (153)
Here x = 0 , is an irregular singular point. We change variable x to t by
1
x = . Then
t
d2y dy
4t 2 + 2 + y=0 (154)
dt dt
Here t = 0 , is a regular singular point.
∞
Assuming y = ∑c
m =0
mt
m+k
and using Frobenius method, the solution of D.E.

(154) is obtained as
 t t2   1 1 1 
y = a 1 − + − L + bt 1 / 2 1 − t + t 2 − t 3 + L
 2! 4 !   3! 5 ! 7! 
1
Putting t = in the above solution, we obtain
x
 x −1 x −2   x −1 x −2 x −3 
y( x ) = a 1 − + − L + bx −1 / 2 1 − + − + L ,
 2! 4!   1! 5! 71 
which is the required series solution of the given equation.

b) The given equation is

2 x 2 (1 − x ) y ′′ − 5x (1 + x ) y ′ + (5 − x ) y = 0 (155)
1
Since we have to find solution for large values of x , putting x = ,
t
Eqn.(155) becomes
d2y dy
2 t 2 ( t − 1) 2 + t (9 t + 1) + (5t − 1) y = 0 . (156)
dt dt
Clearly t = 0 is a regular singular point. Using Frobenius method, the
solution of Eqn.(156) will be
∞  ∞ 
y = a ∑
 n =0
(n + 1) (2n + 3) (2n + 5) t n +1  + b 
  n =0
∑
(n + 1) (n + 2)(2n + 1) t n +1 / 2 

Hence solution of given equation (155) is
∞ ∞
y=a ∑n =0
(n + 1) (2n + 3) (2n + 5) x − n −1 + b ∑ (n + 1) (n + 2) (2n + 1) x
n =0
− n −1 / 2
.

—x—

81