Topic 08: Power Series Solutions to the ODEs

Second Part of MA102 Mathematics II: Winter Semester of AY 2024-2025

MGPP, SN, SHB: IIT Guwahati

Sections 6.1 to 6.3

of
S. L. Ross, Differential Equations, 3rd Edition, Wiley
Learning Outcome

I Linear ODEs
I (Real) Analytic Functions
I Second Order Linear ODE with Variable/Analytic Function Coefficients
I Two Types of Points of the ODE:
I Ordinary Points
I Singular Points. It is further classified as
I Regular Singular Points
I Irregular Singular Points
I Method of Series Solutions about the Ordinary Points
I Examples: y00 + y = 0, Legendre Equation

I Method of Series Solutions about the Regular Singular Points (Frobenius Method)
I Examples: Bessel Equation, Modified Bessel Equation
I Properties of Legendre Polynomials, Bessel Functions
Linear ODEs

A linear ordinary differential equation of order n is given by

a0 (x) y(n) + a1 (x) y(n−1) + · · · + an−1 (x) y0 + an (x) y = F(x) .

We assume that ak (x), 0 ≤ k ≤ n and F(x) are continuous on an interval I = (a, b) and
a0 (x) , 0 for each x ∈ I .

If these coefficients ak (x)’s and F(x) are (real) analytic functions, then we are going to
study how to find the general solution of this ODE using power series method in this
lecture.

In particular, we study some of the famous second order equations such as Legendre
equation and Bessel equation under this topic.
Power Series
Recall from MA101: Power Series

A series of the form

∞
X
an (x − x0 )n
n=0

is called a power series about the point x = x0 .

Examples:
∞ ∞ ∞
X X (−1)n+1 (x − 1)n X xn
xn , ,
n=0 n=1
n n=0
n!

∞
X
Given a power series an (x − x0 )n , we need to determine
n=0
I The values of x for which the series converges.
I That is, determine the largest possible R (Radius of convergence) such that the
series converges in |x − x0 | < R (Domain of Convergence).
Radius of Convergence of a Power Series

Definition 1
∞
X
For a given power series an (x − x0 )n , define the number R, (0 ≤ R ≤ ∞), by
n=0

1 1
= lim sup |an | n (Cauchy-Hadamard Formula).
R n→∞

This number R is called the Radius of Convergence of the power series.

Definition of Limit Superior of a Real Sequence

The concept of limit superior (and limit inferior) is defined only for real sequences and it
is very much useful in case of non-convergence sequences.

Definition 2
If {an } is a sequence of real numbers then define

lim sup an = lim sup{an , an+1 , an+2 , · · · } = inf sup{an : n ≥ k} = sup E,

 

n→∞ n→∞ k∈N

where E is the set of all numbers x in the extended real numbers R = R {−∞, ∞}
S
such that ank → x as k → ∞ for some subsequence {ank } of {an }.
That is E contains all subsequential limits plus possibly the numbers +∞ or −∞.
Convergence of Power Series

Theorem 3
∞
X
Let an (x − x0 )n be a power series with the radius of convergence R. Then
n=0
I if |x − x0 | < R, the series converges absolutely;
I if |x − x0 | > R, the series diverges;
I if 0 < r < R, the series converges uniformly on {x : |x − x0 | ≤ r}.

The open interval/ disk {x : |x − x0 | < R} is called the Interval/ Domain of Convergence
of the series.
∞ ∞
X X xn
Examples: k x has radius of convergence R = 1/|k|.
n n
has radius of
n=0 n=0
n!
∞
X 2
convergence R = ∞. 5n xn has radius of convergence R = 0.
n=0
The radius of the convergence of a power series can be calculated sometimes from
the ratio of the coefficients as follows.
Theorem 4
∞
X
If an (x − x0 )n is a given power series with radius of convergence R, then
n=0

an
R = lim
n→∞ an+1
provided this limit exists.

Examples:
∞ ∞ n X ∞
X X x xn
I Three series n
x , , 2
have radius of convergence R = 1.
n=0 n=1
n n=1
n
∞
X xn
I The series has radius of convergence R = ∞.
n=0
n!
What happens on the boundary |x − x0 | = R?

∞
X (−2)n
Find the radius of convergence of the series (x − 3)n .
n=0
n+1
 
an (−2)n (n + 2) (n + 2) 1+ n
2
1
R = lim = lim = lim = lim  = .
n→∞ an+1 n→∞ (−2)n+1 (n + 1) n→∞ 2(n + 1) n→∞ 2 1 + 1 2
n

Therefore, the series converges absolutely for |x − 3| < 12 and diverges for |x − 3| > 12 .
Next, what happens to convergence of the power series on the boundary |x − 3| = 1/2?
At x = 5/2, the series becomes the harmonic series ∞ 1
P
n=0 n+1 , and hence diverges.
(−1)n
At x = 7/2, the series becomes an alternating harmonic series ∞
P
n=0 n+1 , and hence
converges. Thus, the power series converges for each x ∈ (5/2, 7/2]. (That is, On the
boundary, it can happen any thing! One has to analyze at each point separately).
Main Result on Power Series
Theorem 5
∞
X
Let the series an (x − x0 )n have radius of convergence R > 0. Then,
n=0
∞
X
I The function defined by f (x) := an (x − x0 )n is infinitely many times
n=0
differentiable in BR (x0 ) = {x ∈ R : |x − x0 | < R}.
I For each k ≥ 1, the (k times differentiated) series
∞
X
n(n − 1) · · · (n − k + 1)an (x − x0 )n−k has the radius of convergence R.
n=k
I If f (k) (x) is the k-th derivative of f (x) then
∞
X
f (k) (x) = n(n − 1) · · · (n − k + 1)an (x − x0 )n−k for x ∈ BR (x0 ) .
n=k
Power Series with Positive Radius of Convergence

The previous result mainly says that

I A power series with the radius of convergence R > 0 can be differentiated
term-by-term as often as desired.
I Futher the term-wise differentiated series converges to the derivative of the sum
function of the series.

Similarly, a power series can be integrated term-by-term as often as desired.

Term-wise Integration
Theorem 6
∞
X
Let the series an (x − x0 )n have radius of convergence R > 0 and
n=0
∞
X
f (x) := an (x − x0 )n for |x − x0 | < R. Then, the term-wise integration gives the power
n=0
series for the integral of f :
Z ∞
X an
f (x)dx = (x − x0 )n+1 + C for |x − x0 | < R.
n=0
n+1

A convergent power series have the following useful properties: Within its domain of
convergence,
I The power series describes a differentiable function (analytic function).
I The power series may be integrated term-by-term as often as desired.
I The power series may be differentiated term-by-term as often as desired.
These special properties make power series extremely useful.
Example
Consider
1
= 1 + x + x2 + x3 + · · · + xn + · · · for |x| < 1 .
1−x
Since d
dx {1/(1 − x)} = 1
(1−x)2
, the term-by-term differentiation of the power series yields

1
2
= 1 + 2x + 3x 2
+ 4x 3
+ · · · + nx n−1
+ ··· for |x| < 1 .
(1 − x)
Consider
1
= 1 − x 2
+ x 4
− x 6
+ · · · + (−1) x + ···
n 2n
for |x| < 1 .
1 + x2
Rx 1
Since tan x = 0 1+t2 dt, term-by-term integration of the power series yields
−1

1 3 1 5 1 7 (−1)n x2n+1
tan −1
x = x − x + x − x + ··· + + ··· for |x| < 1 .
3 5 7 2n + 1
 (Real) Analytic Functions
(Functions having Power Series Representations)
(Real) Analytic Functions

Definition 7
Let f : (a, b) ⊆ R → R and let x0 ∈ (a, b). Suppose there exists an open neighborhood
Nr (x0 ) = {x ∈ R : |x − x0 | < r} ⊂ (a, b) of x0 such that
I For each k ∈ N, the k-th derivative f (k) of f exist in Nr (x0 );
∞
X f (n) (x0 )
I The Taylor series (x − x0 )n of f (x) about the point x0 converges in
n=0
n!
Nr (x0 ) to the function g(x) (say);
I f (x) = g(x) for each x ∈ Nr (x0 ).
Then we say that f is analytic at the point x = x0 .

Examples:
I P(x) = a0 + a1 x + · · · + an xn is analytic at each point of R.
I e x , sin(x), cos(x) are analytic at each point of R.
A function f will NOT be analytic at the point x0 if any one of the following happens:
I f (x) is NOT defined at x0 and it is NOT possible to redefine it as f (x0 ) = lim f (x).
x→x0
I f (x) → ±∞ as x → x0 .
I f (k) (x0 ) does NOT exist for some k.
I The Taylor series of f at x0 does NOT converge in any neighborhood Nr (x0 ) of x0 .
I The Taylor series of f at x0 converges to g(x) (say) in some neighborhood Nr (x0 )
of x0 , but the sum function g(x) is NOT equal to f (x) in Nr (x0 ).
P(x)
Let R(x) = Q(x) be a rational function where P(x) and Q(x) are polynomials of degree m
and n respectively and they have no common zeros (That is, R(x) is written in its least
form).
Let S = {x ∈ R : Q(x) = 0} be the set of all real zeros of Q(x). Then R(x) is analytic in
R except on the set S .

x(x−3)
The function (x2 −1)(x−5)
is analytic on R except on {−1, 1, 5}.
Brief Definition of (Real) Analytic Functions
Definition 8
Let f : (a, b) ⊆ R → R and let x0 ∈ (a, b). Then, the function f is said to be analytic at
the point x0 if
∞
X
f (x) = an (x − x0 )n for |x − x0 | < r,
n=0

for some r > 0. That is, f has a power series representation about the point x0 with
some positive radius of convergence.

Observations:
I The power series in the above definition will be acutally the Taylor series of f (x)
about the point x0 . That is,
f (n) (x0 )
an = for n = 0, 1, 2, . . . .
n!
I f is analytic at each point in the neighborhood Nr (x0 ) mentioned in the above
definition.
Non-Analytic Function (even though infinitely many times differentiable)

Consider the function f : R → R given by

 1
e − x
 if x > 0,
f (x) = 

0
 if x ≤ 0.

The function f has continuous derivatives of all orders at every point x of the real line.
The formula for these derivatives is
pn (x)

 2n f (x) if x > 0,


f (n) (x) = 

 x
0 if x ≤ 0,


where pn (x) is a polynomial of degree n − 1 given recursively by p1 (x) = 1 and

pn+1 (x) = x2 p0n (x) − (2nx − 1)pn (x) for n ∈ N. for any positive integer n.
Continuation of Previous Slide
From this formula, it is not completely clear that the derivatives are continuous at 0;
this follows from the one-sided limit
1
e− x
lim =0 for m = 0, 1, 2, . . . .
x&0 xm

Since f (0) = 0 and f (n) (0) = 0 for each n ∈ N, the Taylor series of f at x0 = 0 is
∞ ∞ ∞
X f (n) (0) n X 0 n X
x = x = 0=0 for all x ∈ R
n=0
n! n=0
n! n=0

converges everywhere to the zero function. Observe that the (sum function of the)
Taylor series about x0 = 0 does NOT equal to f (x) for x > 0. Consequently, f is NOT
analytic at x0 = 0.

It also may be noted that if x0 , 0, then f is analytic at x0 .

Recall: Arithmetic Operations on Power Series

∞
X ∞
X
If f (x) = an (x − x0 )n in |x − x0 | < R1 and g(x) = bn (x − x0 )n in |x − x0 | < R2 then
n=0 n=0
I the function f (x) + g(x) is analytic in |x − x0 | < R where R = min(R1 , R2 ).
I If r∗ = min{|x − x0 | : g(x) = 0} is distance from the point x0 to the nearest zero of
f (x)
g then the function h(x) = is analytic in |x − x0 | < R where R = min(R1 , R2 , r∗ ).
g(x)

Using this result, find the domain of analyticity of each of the following functions.
e x sin x x2 (x − 1)
tan(x), ln(1 + x), , , .
1 − x x (2x − 1) sin(x)
Recall: Arithmetic Operations on Power Series

Let f (x) = n for |x − x | < R and g(x) = n for |x − x | < R .

P∞ P∞
n=0 an (x − x0 ) 0 1 n=0 b n (x − x 0 ) 0 2
I The product function f (x)g(x)
∞  ∞  ∞
X  X  X
=  an (x − x0 )n   bn (x − x0 )n  = cn (x − x0 )n for |x − x0 | < R
n=0 n=0 n=0

where
n
X
cn = ak bn−k for n ≥ 0 and R ≥ min{R1 , R2 } .
k=0

It is called the Cauchy Product of the two series.

Shifting the summation index
The index of a summation in a power series is a dummy index and hence
∞
X ∞
X ∞
X
an (x − x0 )n = ak (x − x0 )k = ai (x − x0 )i .
n=0 k=0 i=0

Shifting the index of summation is particularly important when one has to combine two
different power series.

∞
X ∞
X
n(n − 1)an xn−2 = (n + 2)(n + 1)an+2 xn .
n=2 n=0
∞
X ∞
X ∞
X
x3 n2 (n − 2)an xn = n2 (n − 2)an xn+3 = (n − 3)2 (n − 5)an−3 xn .
n=0 n=0 n=3
 ∞
X
Example: Do shifting of the summation index in n(n − 1)an xn−2 and make it as the
n=2
∞
X
series of the form bn xn .
n=n0
∞
X
Consider n(n − 1)an xn−2 and consider the transformation k = n − 2.
n=2
As n varies from 2 to ∞, the variable k varies from 0 to ∞.
k = n − 2 gives that n = k + 2. That is, in place of n, we have to substitute k + 2.
It gives that
∞
X ∞
X ∞
X
n(n − 1)an xn−2 = (k + 2)(k + 2 − 1)ak+2 xk+2−2 = (k + 2)(k + 1)ak+2 xk .
n=2 k=0 k=0

Now, just replace the dummy variable k by another dummy variable n.

∞
X
(n + 2)(n + 1)an+2 xn .
n=0
 Linear ODE
Solutions by Power Series Method
Ordinary Points and Singular Points
Ordinary Points and Singular Points

Consider a linear ODE of order n is given by

a0 (x) y(n) + a1 (x) y(n−1) + · · · + an−1 (x) y0 + an (x) y = F(x) . (1)

Definition 9
A point x0 ∈ (a, b) is said to be an ordinary point of the given ODE (1) if each of the
functions bk (x) := aa0k (x)
(x) , ( 1 ≤ k ≤ n ) and G(x) := F(x)
a0 (x) is analytic at x0 .
A point x0 ∈ (a, b) is said to be a singular point (or singularity) of the given ODE (1) if x0
is not an ordinary point of (1).

Examples:
I For the ODE y00 + 2y0 + xy = sin x, each point x ∈ R is an ordinary point.
I For the ODE xy00 + e x y0 + (x − 1)x y = 0, each point x ∈ R \ {0} is an ordinary point
and the point x = 0 is a singular point.
Regular Singular Points and Irregular Singular Points
The singular points are further classified into
I Regular Singular Points
I Irregular Singular Points
Below, we give definition in case of second order ODEs.

Definition 10
Consider the second order linear homogeneous ODE in the normalized form

y00 + P(x) y0 + Q(x) y = 0 . (2)

A point x0 ∈ R is said to be a Regular Singular Point of the ODE (2) if

I x0 is a singular point of the ODE (2),
I (x − x0 )P(x) and (x − x0 )2 Q(x) are both analytic at x0 .
A singular point x0 ∈ R is said to be an Irregular Singular Point of the ODE (2) if it is not
a regular singular point.
Example

Classify the singular points of x2 (x − 2)2 y00 + 2(x − 2)y0 + (x + 1)y = 0.

Rewriting the given ODE in the normalized form

2(x − 2) 0 (x + 1) 2 (x + 1)
y00 + 2 2
y + 2 2
y=0 =⇒ y00 + 2
y0
+ 2 2
y=0
x (x − 2) x (x − 2) x (x − 2) x (x − 2)
2
The function P(x) = 2
is analytic on R except at x = 0 and x = 2.
x (x − 2)
x+1
The function Q(x) = 2 is analytic on R except at x = 0 and x = 2.
x (x − 2)2
Therefore, x = 0 and x = 2 are singular points of the ODE.
2 x+1
Observe that xP(x) = is not analytic at x = 0 and x2 Q(x) = 2
is analytic
x(x − 2) (x − 2)
at x = 0. So, the point x = 0 is an irregular singular point.
2 x+1
Observe that (x − 2)P(x) = 2 is analytic at x = 2 and (x − 2)2 Q(x) = 2
is analytic
x x
at x = 2. So, the point x = 2 is a regular singular point.
Example
Classify the singular points of xy00 + x(x − 1)−1 y0 + (sin x)y = 0.
Rewriting the given ODE in the normalized form

x sin x
y00 + y0 + y=0.
x(x − 1) x
1
The function P(x) = is analytic on R except at x = 1.
(x − 1)
sin x
The function Q(x) = is analytic on R including x = 0, because at x = 0, we have
x
sin x x2 x4
Q(x) = =1− + − ··· for all x ∈ R .
x 3! 5!
Therefore, x = 1 is the only singular point of the ODE.
(x − 1)2 sin x
Observe that (x − 1)P(x) = 1 is analytic at x = 1 and (x − 1) Q(x) =
2
is
x
analytic at x = 1. So, the point x = 1 is a regular singular point.
Singularities at the Point at Infinity ( x = ∞)

d2 y dy
Consider the differential equation a0 (x) 2 + a1 (x) + a2 (x)y = 0. It is often desirable,
dx dx
to study the solutions of it for large values of the independent variable x. We can adapt
our previous ideas to this broader purpose by studying the solutions near the point at
infinity. This is done by making the change of variable t = 1/x and studying the
resulting equation at t = 0. The given equation will be transformed into the following
equation that can be analyzed at t = 0.

d2 y dy
+ P(t) + Q(t) y = 0 ,
dt2 dt
where    
2 a1 1t a2 1
t
P(t) = −   and Q(t) =  .
t t a0 1t
2 t4 a0 1
t
 d2 y dy
+ P(t) + Q(t) y = 0 ,
dt2 dt
where    
2 a1 1t a2 1
t
P(t) = −   and Q(t) =  .
t t a0 1t
2 t4 a0 1
t

d2 y dy
For the original equation a0 (x) 2 + a1 (x) + a2 (x)y = 0:
dx dx
I The point at infinity is an ordinary point if P(t) and Q(t) are analytic at t = 0.

I The point at infinity is a singular point if it is NOT an ordinary point.

Suppose that the point at infinity is a singular point. Then,
I The point at infinity is a regular singular point if t P(t) and t2 Q(t) are analytic at
t = 0.
I The point at infinity is an irregular singular point if either t P(t) or t2 Q(t) (or both)
is not analytic at t = 0.
Example
d 2y dy
Classify the singular point at the point at infinity for x 2
+ 6 + 3y = 0.
dx2 dx
Here a0 (x) = x , a1 (x) = 6 and a2 (x) = 3.
2

Making the change of variable t = 1/x in the ODE, we get

d2 y dy
+ P(t) + Q(t) y = 0 ,
dt2 dt
where    
2 a1 1t 2 a2 1
t 3
P(t) = −   = −6 and Q(t) =  = .
t t2 a0 1t t t4 a0 1 t2
t

Observe that t = 0 is a singular point. Since t P(t) = 2 − 6t is analytic at t = 0 and

t2 Q(t) = 3 is analytic at t = 0, the singular point t = 0 is a regular singular point of the
transformed equation.
Therefore, the singular point x = ∞ is a regular singular point of the original equation
x2 y00 + 6y0 + 3y = 0.
 Second Order Linear ODE
Power Series Solutions about the Ordinary Points
Series Solutions to y00 + P(x)y0 + Q(x)y = F(x)
Step 1: Writing Form of the Solution and its derivatives
Let x0 be an ordinary point of the given ODE.
∞
X
Solution: y(x) = an (x − x0 )n in |x − x0 | < R for some R > 0.
n=0
∞
X ∞
X
y0 (x) = nan (x − x0 )n−1 , y00 (x) = n(n − 1)an (x − x0 )n−2 .
n=1 n=2
Step 2: Expanding Coefficients of the ODE into Power Series (in terms of powers of
(x − x0 ))
Expand the coefficients P(x), Q(x) and the nonhomogeneous term F(x) (if any) into
Power Seires about the point x0 (in terms of powers of (x − x0 )).
Step 3: Substitute Power Series in the ODE
Substitute the power series of y(x) and their derivatives, P(x), Q(x) and F(x) in the
given ODE.
Note: In Steps 2 and 3, instead of normalized form, some time the original form
a0 (x)y00 + a1 (x)y0 + a2 (x)y = F(x) may give simple expressions in terms of powers of
(x − x0 ) for a0 (x), a1 (x), a2 (x) and F(x).
Continuation of Previous Slide
Step 4: Shifting the indices and Gathering together all terms of Same power of (x − x0 )
Shift the indicies, if required, and Gather together all terms involving the same power
of (x − x0 ) to arrive at an identity of the form

A0 + A1 (x − x0 ) + A2 (x − x0 )2 + · · · = 0 .
Step 5: Equating each coefficient to Zero of the above Identity
Equate each coefficient of the identity in Step 4 to zero to obtain equations

A0 = 0, A1 = 0, A2 = 0, A3 = 0, · · · · · ·
Step 6: Expressing an for n ≥ 2 in terms of a0 and a1
Express an for n ≥ 2 in terms of a0 and a1 (in case of 2nd order ODE).
Step 7: Writing the solution y(x) along with the domain of convergence
Insert obtained expressions of these coefficients in y(x) to get
∞  ∞ 
X  X 
y(x) = a0  bn (x − x0 )  + a1  cn (x − x0 ) 
 n   n

n=0 n=0
Continuation of Previous Slide

Set
∞
X ∞
X
y1 (x) := bn (x − x0 )n and y2 (x) := cn (x − x0 )n
n=0 n=0

Write the common domain of convergence |x − x0 | < R of the above two power series
as a domain of y(x).
Note that y1 (x) and y2 (x) are two linearly independent solutions of the given ODE and
y(x) = a0 y1 (x) + a1 y2 (x) is the general solution of the given ODE in the domain
|x − x0 | < R.
Step 8: Substitute Initial conditions y(x0 ) = A and y0 (x0 ) = B, if any
If the initial conditions y(x0 ) = A and y0 (x0 ) = B are given, plug them in the general
solution and determine the values of a0 and a1 .
Now substitute the values of a0 and a1 in y(x), write the unique solution to the given IVP.
Main Theorem
Theorem 11
Consider the second order, linear, homogeneous ODE in its normalized form

y00 + P(x)y0 + Q(x)y = 0 . (3)

If x0 is an ordinary point of the ODE (3) then the ODE (3) has two nontrivial linearly
independent power series solutions of the form
∞
X ∞
X
y1 (x) = cn (x − x0 )n and y2 (x) = dn (x − x0 )n
n=0 n=0

in the domain |x − x0 | < R where R is the distance from x0 to the nearest singular point
of (3). Further y(x) = Ay1 (x) + By2 (x) where A and B are arbitrary real constants, is the
general solution of (3) in |x − x0 | < R.

If the initial conditions y(x0 ) = y0 and y0 (x0 ) = y1 are given, then there exists unique
solution y∗ (x) to the given ODE (3) satisfying the given ICs in the domain |x − x0 | < R.
Example 1: Series Solution about an Ordinary Point

Find a series solution of the second order ODE y00 + y = 0.

∞
X
About the ordinary point x0 = 0, We would like to find a series solution y(x) = an x n
n=0
to the given ODE + y = 0.
y00
Step 1: Writing Form of the Solution and its derivatives
∞
X
Let y(x) = an xn in |x| < R for some R > 0.
n=0
∞
X ∞
X
y0 (x) = nan xn−1 , y00 (x) = n(n − 1)an xn−2 .
n=1 n=2

Steps 2 and 3: Substituting Power Series in the ODE

∞
X ∞
X
n(n − 1)an xn−2 + an x n = 0 .
n=2 n=0
Step 4: Shifting the indices and Gathering all terms of Same power of (x − x0 )
∞
X ∞
X ∞
X
n(n − 1)an xn−2 + an−2 xn−2 = 0 =⇒ (n(n − 1)an + an−2 ) xn−2 = 0 .
n=2 n=2 n=2

Step 5: Equating each coefficient to Zero

n(n − 1)an + an−2 = 0 for n = 2, 3, · · ·
Step 6: Expressing an for n ≥ 2 in terms of a0 and a1
−a0 −a0
a2 = =
2 2!
−a1 −a1
a3 = =
3×2 3!
−a2 −(−a0 ) a0
a4 = = =
4 × 3 4 × 3 × 2 × 1 4!
a1 −a0 −a1 a0 (−1)n a1 (−1)n a0
a5 = , a6 = , a7 = , a8 = , a2n+1 = , a2n = .
5! 6! 7! 8! (2n + 1)! (2n)!
Step 7: Writing the solution y(x) along with the domain of convergence
Insert obtained expressions of these coefficients in y(x), we get
∞  ∞ 
X (−1)n x2n  X (−1)n x2n+1 
y(x) = a0    + a1 
  .
n=0
(2n)! n=0
(2n + 1)! 

Set
∞ ∞
X (−1)n x2n X (−1)n x2n+1
y1 (x) = = cos x and y2 (x) = = sin x .
n=0
(2n)! n=0
(2n + 1)!

Note that both the series have radius of convergence R = ∞

y(x) = a0 y1 (x) + a1 y2 (x) = a0 cos x + a1 sin x for all x ∈ R .

Assume that a0 and a1 are arbitrary real constants. Therefore,

y(x) = A cos x + B sin x for all x ∈ R

where A and B are arbitrary real constants,is the general solution of the given ODE
y00 + y = 0.
Example 2: Series Solution about an Ordinary Point

Find a series solution of the second order ODE 2y00 + x y0 + y = 0.

∞
X
About the ordinary point x0 = 0, We would like to find a series solution y(x) = an x n
n=0
to the given ODE + x + y = 0.
2y00 y0
Step 1: Writing Form of the Solution and its derivatives
∞
X
Let y(x) = an xn in |x| < R for some R > 0.
n=0
∞
X ∞
X
y0 (x) = nan xn−1 , y00 (x) = n(n − 1)an xn−2 .
n=1 n=2

Steps 2 and 3: Substituting Power Series in the ODE

∞
X ∞
X ∞
X
2 n(n − 1)an xn−2 + x nan xn−1 + an xn = 0
n=2 n=1 n=0
Step 4: Shifting the indices and Gathering all terms of Same power of (x − x0 )
∞
X ∞
X ∞
X
2 n(n − 1)an xn−2 + nan xn + an xn = 0
n=2 n=1 n=0
∞
X ∞
X ∞
X
2(n + 2)(n + 1)an+2 xn + nan xn + an x n = 0
n=0 n=1 n=0
∞
X
a0 + 4a2 + (2(n + 2)(n + 1)an+2 + n an + an ) xn = 0 .
n=1
Step 5: Equating each coefficient to Zero

a0 + 4a2 = 0 .
2(n + 2)(n + 1) an+2 + (n + 1) an = 0, n≥1.
Step 6: Expressing an for n ≥ 2 in terms of a0 and a1

−1
an+2 = an n≥1.
2(n + 2)
 −1 −1
a2 = 2 a0 , a3 = a1 ,
2 2·3
−1 1 −1 1
a4 = a2 = 2 a0 , a5 = a3 = 2 a1 .
2·4 2 ·2·4 2·5 2 ·3·5
(−1)n
=⇒ a2n = 2n a0 for n = 1, 2, 3, . . . ,
2 n!
(−1)n
a2n+1 = n a1 for n = 1, 2, 3, . . . .
2 [1 · 3 · 5 · · · (2n + 1)]
Step 7: Writing the solution y(x) along with the domain of convergence
Insert obtained expressions of these coefficients in y(x), we get
∞ ∞
X (−1)n 2n X (−1)n
y(x) = a0 2n n!
x + a 1 n [1 · 3 · 5 · · · (2n + 1)]
x 2n+1
for x ∈ R .
n=0
2 n=0
2
Set
∞ ∞
X (−1)n 2n X (−1)n
y1 (x) = x and y2 (x) = x 2n+1
for x ∈ R .
n=0
2n
2 n! n=0
2 [1 · 3 · 5 · · · (2n + 1)]
n
Observe that

y1 (0) = 1, y2 (0) = 0, y01 (0) = 0, y02 (0) = 1 .

Therefore,
y1 (0) y2 (0) 1 0
W(y1 , y2 )(0) = = =1,0.
y01 (0) y02 (0) 0 1
It gives that W(y1 , y2 )(x) , 0 for all x ∈ R. Therefore, y1 (x) and y2 (x) are linearly
independent.

The general solution is given by y(x) = A y1 (x) + B y2 (x) for all x ∈ R, where A and B
are arbitrary real constants and
∞
X (−1)n 2n
y1 (x) = 2n
x for x ∈ R ,
n=0
2 n!
∞
X (−1)n
y2 (x) = x 2n+1
for x ∈ R .
n=0
2n [1 · 3 · 5 · · · (2n + 1)]
 Legendre Differential Equation
(1 − x2 )y00 − 2xy0 + p(p + 1)y = 0
Power Series Solutions about the Ordinary Point x0 = 0

Adrien-Marie Legendre (1752 - 1833) was a French Mathematician. Legendre made

numerous contributions to mathematics. Well-known and important concepts such as
the Legendre differential equation and Legendre polynomials, Legendre
transformation, Legendre symbol in number theory are named after him.
For two centuries, until the recent discovery of the error in 2005, books, paintings and
articles have incorrectly shown a side-view portrait of the obscure French politician
Louis Legendre (1752 - 1797) as that of the mathematician Legendre. The error arose
from the fact that the sketch was labeled simply “Legendre”.
Refer: http://www.ams.org/notices/200911/rtx091101440p.pdf
Example 3: Legendre Equation
Find a series solution of the Legendre equation (1 − x2 )y00 − 2xy0 + p(p + 1)y = 0 where
p is an arbitrary real constant.
Step 1: Writing Form of the Solution and its derivatives
Rewriting the given ODE in normalized form, we have

2x 0 p(p + 1)
y00
− y + y = 0.
1 − x2 1 − x2
p(p+1)
2 and 1−x2 are analytic at x0 = 0 and hence x0 = 0 is an
−2x
Observe that the functions 1−x
ordinary point. We would like to find a series solution of the form
∞
X
y(x) = an xn in |x| < R for some R > 0 .
n=0

∞
X ∞
X
y0 (x) = nan xn−1 and y00 (x) = n(n − 1)an xn−2 .
n=1 n=2
Steps 2 & 3: Expanding Coefficients into Power Series about x0 and also Substituting
Power Series of y(x) and their derivatives
∞  ∞ ∞
X X X
(1 − x2 )y00 = (1 − x2 )  n(n − 1)an xn−2  = n(n − 1)an xn .
 
n(n − 1)an xn−2 −
 
n=2 n=2 n=2
∞  ∞
X X
−2xy0 = −2x  nan xn−1  = −2 nan xn .

 

n=1 n=1
∞  ∞
X  X
p(p + 1)y = p(p + 1)  an x  =
n
p(p + 1)an xn .
n=0 n=0

We get
∞
X ∞
X ∞
X ∞
X
n(n − 1)an xn−2 − n(n − 1)an xn − 2 nan xn + p(p + 1)an xn = 0 .
n=2 n=2 n=1 n=0
Step 4: Gather together all terms of Same power of (x − x0 )
∞
X ∞
X ∞
X ∞
X
(n + 2)(n + 1)an+2 xn − n(n − 1)an xn − 2 nan xn + p(p + 1)an xn = 0 .
n=0 n=2 n=1 n=0
∞
X
(2a2 + p(p + 1)a0 ) + (6a3 − 2a1 + p(p + 1)a1 )x + (n + 2)(n + 1)an+2 xn
n=2
∞
X ∞
X X∞
− n(n − 1)an xn − 2nan xn + p(p + 1) an x n = 0 .
n=2 n=2 n=2

=⇒ (−2a2 − p(p + 1)a0 ) + (−6a3 + 2a1 − p(p + 1)a1 )x+

∞
X
((n + 2)(n + 1)an+2 − n(n − 1)an − 2nan + p(p + 1)an ) xn = 0
n=2
Step 5: Equating each coefficient to Zero
2a2 + p(p + 1)a0 = 0 .
6a3 − 2a1 + p(p + 1)a1 = 0 .
(n + 2)(n + 1)an+2 − n(n − 1)an − 2nan + p(p + 1)an = 0 for n ≥ 2 .
Step 6: Expressing an for n ≥ 2 in terms of a0 and a1

−p(p + 1)
a2 = a0 .
1×2
−(p − 1)(p + 2)
a3 = a1 .
1×2×3
−(p − n)(p + n + 1)
an+2 = an for n ≥ 2 .
(n + 1)(n + 2)
This gives that, for n ≥ 1

(−1)n (p + 2n − 1)(p + 2n − 3) · · · (p + 1)p(p − 2) · · · (p − 2n + 2) a0

a2n = ,
(2n)!
(−1)n (p + 2n)(p + 2n − 2) · · · (p + 2)(p − 1)(p − 3) · · · (p − 2n + 1) a1
a2n+1 = .
(2n + 1)!
Step 7: Writing the solution y(x) along with the domain of convergence
Set, for |x| < 1,
y1 (x) = 1+
∞
X (−1)n (p + 2n − 1)(p + 2n − 3) · · · (p + 1)p(p − 2) · · · (p − 2n + 2) a0
x2n ,
n=1
(2n)!

y2 (x) = x+
∞
X (−1)n (p + 2n)(p + 2n − 2) · · · (p + 2)(p − 1)(p − 3) · · · (p − 2n + 1) a1
x2n+1 .
n=1
(2n + 1)!

y(x) = a0 y1 (x) + a1 y2 (x) .

Assume that a0 and a1 are arbitrary real constants. Therefore,

y(x) = A y1 (x) + B y2 (x) for |x| < 1

where A and B are arbitrary real constants, is the general solution of the given
Legendre equation.
I If p ∈ R \ Z, then y1 (x) and y2 (x) are infinite series and they converge in |x| < 1.
I If p ∈ N and p is odd, then y2 (x) is a polynomial of degree p having only odd
powers of x, and y1 (x) is not a polynomial.
If p = 1, y2 (x) = x =: P˜1 (x).
If p = 3, y2 (x) = x − 35 x3 =: P˜3 (x).
If p = 5, y2 (x) = x − 14
3 x 3 + 21 x5 =: P˜ (x).
5 5
I If p ∈ N ∪ {0} and p is even, then y1 (x) is a polynomial of degree p having only
even powers of x, and y2 (x) is not a polynomial.
If p = 0, y1 (x) = 1 =: P˜0 (x).
If p = 2, y1 (x) = 1 − 3x2 =: P˜2 (x).
If p = 4, y1 (x) = 1 − 10x2 + 35 3 x =: P4 (x).
4 ˜
(−1)n (1×3×···×(2n−1)) (−1)n (1×3×···×(2n+1))
I Set K0 = 1, K1 = 1, K2n = 2×4×···×2n and K2n+1 = 2×4×···×2n .
The polynomials Pn (x) = Kn P˜n (x) for n = 0, 1, · · · are called the Legendre
Polynomials.
That is, the Polynomial Solution Pn of degree n of the Legendre equation
(1 − x2 )y00 − 2xy0 + n(n + 1)y = 0 satisfying Pn (1) = 1 is called the n-th Legendre
Polynomial.
The Legendre polynomial

To obtain a single formula which contains both the polynomials in y1 (x) and
y2 (x), let for n = 0, 1, 2, . . .
[n/2]
1 X (−1)r (2n − 2r)! n−2r
Pn (x) = n x ,
2 r=0 r!(n − r)!(n − 2r)!

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

I When n is even, it is a constant multiple of the polynomial y1 (x).

I When n is odd, it is a constant multiple of the polynomial y2 (x).

Legendre Polynomials
For n = 0, 1, · · · , the Legendre Polynomial Pn (x) is a particular solution of the ODE
(1 − x2 )y00 − 2xy0 + n(n + 1)y = 0.

P0 (x) = 1
P1 (x) = x
1 2
P2 (x) = (3x − 1)
2
1 3
P3 (x) = (5x − 3x)
2
1
P4 (x) = (35x4 − 30x2 + 3)
8
1
P5 (x) = (63x5 − 70x3 + 15x)
8
Note: Pn (−x) = (−1)n Pn (x) for x ∈ R and for each n.
Pn (1) = 1 and Pn (−1) = (−1)n for each n.
(−1)(n/2) 1·3·5···(n−1)
Pn (0) = 0 if n is odd and Pn (0) = 2·4·6···n if n is even. P0n (0) = 0 if n is even.
Rodrigues’s formula for the Legendre polynomials
Note that
n
!
(2n − 2r)! n−2r d 1 1 n
x = n x2n−2r and = .
(n − 2r)! dx r!(n − r)! n! r
Thus, Pn (x) can be expressed as
[n/2]
1 dn X r n
 
Pn (x) = n (−1) x 2n−2r
.
n
2 n! dx r=0 r

When [n/2] < r ≤ n, the term x2n−2r has degree less than n, so its nth derivative
is zero. This gives
n
1 dn X r n
  1 dn 2
Pn (x) = n (−1) x 2n−2r
= (x − 1)n
,
n
2 n! dx r=0 r n
2 n! dx n

which is known as Rodrigues’ formula.

Properties of Legendre Polynomials Pn (x)
1 dn  2 
I Rodrigues Formula: Pn (x) = n (x − 1) for n = 0, 1, . . ..
n
2 (n!) dxn
I Bonnet’s (Three Term) Recurrence Relation:

(n + 1) Pn+1 (x) − (2n + 1) x Pn (x) + n Pn−1 (x) = 0 for n ∈ N .

I Generating Function:
∞
1 X
G(x, t) := √ = Pn (x) tn .
1 − 2xt + t 2
n=0

Legendre polynomials are the coefficients of the Taylor series of G(x, t) about
t = 0.
I Orthogonality:
1
if m , n ,
Z (
0
Pm (x)Pn (x) dx =
−1
2
2n+1 if m = n .
Properties of Legendre Polynomials Pn (x)

I More Recurrence Relation:

(1 − x2 ) P0n (x) = (n + 1)x Pn (x) − (n + 1) Pn+1 (x) .

P0n+1 (x) − P0n−1 (x) = (2n + 1) Pn (x) .

I If f : [−1, 1] → R is continuous on [−1, 1] then f can be expressed as
∞
2n + 1 1
X Z
f (x) = an Pn (x) for x ∈ [−1, 1] where an = f (x)Pn (x) dx for n = 0, 1,
n=0
2 −1
2, · · · .
I Between two consecutive zeros of Pn+1 (x), there is one zero of Pn (x).
I Between two consecutive zeros of Pn (x), there is one zero of Pn+1 (x).
I Between smallest zero of Pn (x) and −1, there is one zero of Pn+1 (x). Between
largest zero of Pn (x) and 1, there is one zero of Pn+1 (x).
I All n zeros of Pn (x) lie in (−1, 1).
Graphs of Legendre Polynomials
Exercises: Power Series Solutions about Ordinary Points

I Find the power series solution of y00 + xy0 + (x2 + 2)y = 0 about the ordinary point
x0 = 0.
This is worked out example of Ross Book, Example 6.5 on Pages 225 - 228.
I Find the power series solution of (x2 − 1)y00 + 3xy0 + xy = 0 with y(0) = 4 and
y0 (0) = 6.
This is worked out example of Ross Book, Example 6.6 on Pages 229 - 231.
I Find the power series solution of the Airy’s equation y00 + xy = 0 about the
ordinary point x0 = 0.
Answer:
∞
X (−1)k
y1 (x) = 1 + x3k for x ∈ R
k=1
2 · 3 · · · (3k − 1)(3k)
∞
X (−1)k
y2 (x) = x + x3k+1 for x ∈ R
k=1
3 · 4 · · · (3k)(3k + 1)
 Linear Homogeneous ODE
Power Series Solutions about the Regular Singular Points
Solutions About Regular Singular Points
Theorem 12
Consider the second order, linear, homogeneous ODE in its normalized form

y00 + P(x)y0 + Q(x)y = 0 . (4)

If x0 is a regular singular point of the ODE (4) then the ODE (4) has at least one
nontrivial solution of the form
∞
X
|x − x0 |r cn (x − x0 )n
n=0

where r is a definite (real or complex) constant which may be determined (from the
indicial equation), and this solution is valid in some deleted/ punctured interval
0 < |x − x0 | < R (where R > 0) about x0 .
Important Note: In case of irregular singular point, we will NOT attempt to find series
solutions to the ODE in this course, because it becomes exponential functions times a
Frobenius series and its behaviour is highly complicated.
Indicial Equation

Definition 13
Let x0 ∈ R be a regular singular point of the second order, linear, homogeneous ODE
in its normalized form
y00 + P(x)y0 + Q(x)y = 0 .
Then, the indicial equation of the ODE at the regular singular point x0 is given by

r(r − 1) + p0 r + q0 = 0 ,

where
p0 := lim (x − x0 )P(x) and q0 := lim (x − x0 )2 Q(x) .
x→x0 x→x0

The roots of the inidical equation are called the exponents.

We denote the roots of indicial equation by r1 and r2 such that Re(r2 ) ≤ Re(r1 ).
Example
Find the indicial equation at the the singularity x = −1 of

(x2 − 1)2 y00 (x) + (x + 1)y0 (x) − y(x) = 0.

Here x = −1 is a regular singular point. Since

p(x) = (x + 1)−1 (x − 1)−2 and q(x) = −(x + 1)−2 (x − 1)−2 ,

we find that
1
p0 = lim (x + 1)p(x) = lim (x − 1)−2 = ,
x→−1 x→−1 4
1
q0 = lim (x + 1) q(x) = lim [−(x − 1) ] = − .
2 −2
x→−1 x→−1 4
Thus, the indicial equation is given by
1 1
r(r − 1) + r − = 0.
4 4
Hence r = 1 and r = − 14 are the exponents.
Ordinary Point Vs. Regular Singular Point in case of IVPs

In case of IVP, there is a unique solution to the IVP about a ordinary point satisfying
given IC at that ordinary point.
Whereas in case of irregular singular point, the IVP may not have a solution.
I Consider y00 + y = 0 with IC y(0) = K0 and y0 (0) = K1 has a unique solution about
the ordinary point x0 = 0, namely, y(x) = K0 cos x + K1 sin x.
I Consider the Cauchy-Euler equation x2 y00 − 2xy0 + 2y = 0 with IC y(0) = K0 , 0,
y0 (0) = K1 . Note that x0 = 0 is a regular singular point. The general solution is
y(x) = c1 x + c2 x2 . We can not find c1 and c2 so that y(x) satisfies y(0) = K0 , 0
and hence there is no solution to this IVP.
 The Method of Frobenius
to find series solutions about Regular Singular Points
Motivation for the Method of Frobenius

To motivate the procedure, recall the homogeneous Cauchy-Euler equation

ax2 y00 + bx y0 + c y = 0 where a , 0 . (5)

Rewriting it as
y00 + p(x) y0 + q(x) y = 0 , (6)
where
p0 q0 b c
p(x) = , q(x) = with p0 = and q0 = .
x x2 a a
The indicial equation is given by

r(r − 1) + p0 r + q0 = 0 . (7)

If r = r1 is a root of the indicial equation (7), then y1 (x) = xr1 is a solution to (6).
For Cauchy-Euler equations, we seek solutions of the form y(x) = xr and then try to
determine the values for r.
Set
d2 d
L := ax 2
+ bx + c I.
dx2 dx
Now

L(y)(x) = ax2 r(r − 1)xr−2 + bxrxr−1 + cxr

= {ar2 + (b − a)r + c}xr .

Thus,
y(x) = xr is a solution ⇐⇒ r satisfies
ar2 + (b − a)r + c = 0 . (8)
The equation (8) is known as the auxiliary or indicial equation for (5).
Assume that x p(x) and x2 q(x) (instead of being constants) are analytic functions.
Then
∞
X
x p(x) = p0 + p1 x + p2 x2 + · · · = pn xn ,
n=0
∞
X
x q(x) = q0 + q1 x + q2 x + · · · =
2 2
qn xn
n=0

in some neighborhood of x = 0.
Then it follows that

lim x p(x) = p0 and lim x2 q(x) = q0 .

x→0 x→0

Therefore, it is reasonable to expect that the solutions to y00 + p(x)y0 + q(x) = 0 will
p q
behave for x near 0 like the solutions to the Cauchy-Euler equation y00 + x0 y0 + x20 y = 0.
Method of Frobenius to find Solutions about Regular Singular Points

Using this method we would like to find two linearly independent solutions of the
second order homogeneous linear ODE (4) for x , x0 .
∞
X
I First we find a basic solution of the form (x − x0 )r an (x − x0 )n (where r is
n=0
determined) for 0 < x − x0 < R for some R > 0.
I The procedure is to find this basic solution is similar to ordinary points procedure.
In Step 5 (equating coefficients to zero), we get an indicial equation which is an
equation in r that needs to be solved to determine the values of r (which is called
the exponents).
I For −R < x − x0 < 0, we can substitute t = −(x − x0 ) and carry out similar
procedure.
Questions:
I Under what conditions are we assured that the ODE (4) has two LI solutions
about the regular singular point x0 of the (basic solution) form
∞
X
y(x) = |x − x0 |r an (x − x0 )n for 0 < |x − x0 | < R with R > 0?
n=0
I If the ODE (4) does not yield two LI solutions about the regular singular point x0 ,
then what is the form of another solution that is LI of the basic solution?
Outline of Answers:
I If the roots r1 and r2 of the indicial equation are real and r1 − r2 , N where N is a
nonnegative integer, then the ODE (4) has two LI solutions of the basic solution
form.
I If r1 − r2 = N where N is a nonnegative integer, then the ODE (4) has one solution
of the basic solution form and another LI solution with suitable form that we need
to find out from the basic solution.
Initial Steps of Method of Frobenius

Let x0 be a regular singular point of y00 + P(x)y0 + Q(x)y = 0.

We seek a basic solution of the following form in 0 < x − x0 < R with R > 0
∞
X
y(x) = (x − x0 )r an (x − x0 )n where a0 , 0 .
n=0

Then,
∞
X
y0 (x) = (n + r) an (x − x0 )(n+r−1) for 0 < x − x0 < R.
n=0
∞
X
y00 (x) = (n + r)(n + r − 1) an (x − x0 )(n+r−2) for 0 < x − x0 < R.
n=0
Since (x − x0 )P(x) and (x − x0 )2 Q(x) are analytic at x0 , we have
∞
X
(x − x0 )P(x) = pn (x − x0 )n for |x − x0 | < ρ1 ,
n=0
X∞
(x − x0 )2 Q(x) = qn (x − x0 )n for |x − x0 | < ρ2 .
n=0

That is,
∞
X
P(x) = pn (x − x0 )n−1 for 0 < |x − x0 | < ρ1 ,
n=0
X∞
Q(x) = qn (x − x0 )n−2 for 0 < |x − x0 | < ρ2 .
n=0
 ∞  ∞ 
X  X 
P(x) y0 =  pn (x − x0 )   (n + r) an (x − x0 )
n−1 (n+r−1) 

n=0 n=0
∞  ∞ 
X  X 
= (x − x0 ) r−2 
 pn (x − x0 )   (n + r) an (x − x0 ) 
n   n

n=0 n=0
∞  n  
X X  
= (x − x0 ) r−2    (k + r)ak pn−k  (x − x0 )  .
n

n=0 k=0
∞  ∞ 
X  X 
Q(x) y =  qn (x − x0 )   an (x − x0 )
 n−2   (n+r) 

n=0 n=0
∞  ∞ 
X  X 
= (x − x0 ) r−2  n
 qn (x − x0 )   an (x − x0 )  n

n=0 n=0
∞  n  
X X  
= (x − x0 ) r−2    ak qn−k  (x − x0 )  .
   n

n=0 k=0
Subsituting the series expressions of y00 , P(x)y0 and Q(x)y in the given ODE, we get
∞
∞  n  
X X X  
(n + r)(n + r − 1) an (x − x0 )(n+r−2) + (x − x0 )r−2   (k + r)ak pn−k  (x − x0 )n 
   
n=0 n=0 k=0
∞  n  
X X  
+ (x − x0 )r−2   ak qn−k  (x − x0 )n  = 0.
   
n=0 k=0

Collecting the terms for n = 0 in the above expression, we get

Coefficient of (x − x0 )r−2 = a0 r(r − 1) + a0 p0 r + a0 q0 = a0 F(r),

where
F(r) = r(r − 1) + p0 r + q0 .
Collecting the terms for n ≥ 1, we get
 n   n 
X  X 
Coefficient of (x − x0 )n+r−2 = an (n + r)(n + r − 1) +  (k + r)ak pn−k  +  ak qn−k 
  
k=0 k=0
 n−1
  n−1

 X   X 
= an (n + r)(n + r − 1) + (n + r)an p0 + (k + r)ak pn−k  + an q0 + ak qn−k 
k=0 k=0
 n−1   n−1 
X  X 
= an ((n + r)(n + r − 1) + (n + r)p0 + q0 ) +  (k + r)ak pn−k  +  ak qn−k 
  
k=0 k=0
n−1
X
= an F(r + n) + ak (k + r)pn−k + qn−k ,
 
k=0

where
F(r + n) = (n + r)(n + r − 1) + (n + r)p0 + q0 .
Summary
Subsituting the series expressions of y00 , P(x)y0 and Q(x)y in the given ODE, we get

∞ 
 n−1

X X  
a0 F(r)(x − x0 )r−2 + an F(r + n) + ak (r + k)pn−k + qn−k  (x − x0 )n+r−2 = 0,

n=1 k=0

where

F(r) = r(r − 1) + p0 r + q0 ,
F(r + n) = (n + r)(n + r − 1) + (n + r)p0 + q0 ,
∞
X
(x − x0 )P(x) = pn (x − x0 )n for |x − x0 | < ρ1 ,
n=0
X∞
(x − x0 )2 Q(x) = qn (x − x0 )n for |x − x0 | < ρ2 .
n=0
Some Observation

F(r + n) = (n + r)(n + r − 1) + (n + r)p0 + q0

= n2 + n(r − 1) + rn + r(r − 1) + np0 + rp0 + q0
= n2 + 2nr + np0 + r(r − 1) + rp0 + q0 = n(n + 2r + p0 ) + F(r)
Observe that
F(r) = r(r − 1) + p0 r + q0 = r2 + (p0 − 1)r + q0 .
If r1 and r2 are roots of F(r) = 0 such that Re(r2 ) ≤ Re(r1 ), then r1 + r2 = 1 − p0 and
r1 r2 = q0 .
Now,
F(r2 + n) = n(n + 2r2 + 1 − r1 − r2 ) = n [(n + 1) − (r1 − r2 )] .
If F(r2 + n) = 0 for some integer n∗ , then r1 − r2 = n∗ + 1.

If F(r + n) = 0 for some n and for r = r2 , then

r1 − r2 = N, where N is a nonnegative integer.

Final Summary

By balancing the coefficients, we get a0 F(r) = 0 and hence

Indicial Equation: F(r) = r(r − 1) + p0 r + q0 = 0 . (9)

Let r1 and r2 be the roots of indicial equation (which may be complex) such that
Re(r2 ) ≤ Re(r1 ).
I If F(r + n) , 0 for each of the roots of r, then we can determine an in terms of an−1 ,
· · · , a0 . We can obtain two LI solutions.
I If F(r + n) = 0 for some n and r = r2 (smaller root), then the process breaks down.
However we can obtain one solution for the larger root r1 . We have to find another
LI solution using any other method in this case.
The case F(r + n) , 0 for r = r1 and r = r2 leads to the case r1 − r2 , N where N is a
nonnegative integer.
The case F(r + n) = 0 for some n and r = r2 leads to the case r1 − r2 = N where N is a
nonnegative integer.
Three Possible Cases in Method of Frobenius

In the method of Frobenius, the following are possible three cases.

1. Case: r1 − r2 , N where N is a nonnegative integer
We get two series solutions y1 (x) and y2 (x) corresponding to r = r1 and r = r2
respectively. These two solutions y1 (x) and y2 (x) are linearly independent in
0 < x − x0 < R.
2. Case: r1 − r2 = N where N is a positive integer
We get two series solutions y1 (x) and y2 (x) corresponding to r = r1 and r = r2
respectively. These two solutions y1 (x) and y2 (x) may or may not be linearly
independent in 0 < x − x0 < R. If they are linearly independent, we are done with.
If they are not linearly independent, then we attempt to find another solution which
is linearly independent with y1 (x).
3. Case: r1 − r2 = 0
Since r1 = r2 , we get one series solution y1 (x) (say) corresponding to r1 . We need
to find another solution which is linearly independent with y1 (x).
 Second Order Linear Homogeneous ODE
Power Series Solutions about the Regular Singular Points
Method of Frobenuis
Case I: r1 − r2 , N (N is a nonnegative integer)
Case I: r1 − r2 , N (N is a nonnegative integer)

Theorem 14
Consider the ODE y00 + P(x)y0 + Q(x)y = 0.
Let x0 be a regular singular point of the given ODE. Let r1 and r2 where
Re(r1 ) ≥ Re(r2 ) be the roots of the indicial equation associated with x0 . Suppose
r1 − r2 , N , where N is a nonnegative integer. Then the given ODE has two nontrivial
LI solutions y1 and y2 in 0 < |x − x0 | < R for some R > 0 given by
∞
X
y1 (x) = |x − x0 |r1 cn (x − x0 )n where c0 , 0 ,
n=0

∞
X
y2 (x) = |x − x0 |r2 dn (x − x0 )n where d0 , 0 .
n=0
Example for Case I (Refer Ross Book Example 6.11)
Use the method of Frobenius to find solutions of 2x2 y00 − xy0 + (x − 5)y = 0 in some
interval 0 < x < R.
Step 1: Writing Form of the Solution and its derivatives
Since x0 = 0 is a regular singular point of the given ODE, we seek solution of the form
∞
X
y(x) = xr an x n with a0 , 0
n=0
in 0 < x < R for some R > 0.
∞
X ∞
X
y0 (x) = (n + r)an xn+r−1 , y00 (x) = (n + r)(n + r − 1)an xn+r−2 .
n=0 n=0
Steps 2: Substituting Power Series of Coefficient functions, y(x), y0 (x) and y00 (x) in the
ODE
∞
X ∞
X ∞
X ∞
X
2 (n + r)(n + r − 1)an xn+r − (n + r)an xn+r + an xn+r+1 − 5 an xn+r = 0 .
n=0 n=0 n=0 n=0
Step 3: Gather together all terms of Same power of (x − x0 )
∞
X
(2r(r − 1) − r − 5) a0 xr + [(2(n + r)(n + r − 1) − (n + r) − 5) an + an−1 ] xn+r = 0
n=1

Step 4: Equating the coefficient of the lowest power of x to Zero to get Indicial
Equation & its Roots

Indicial Equation : 2r(r − 1) − r − 5 = 0

Its roots are
5
Exponents : r1 = and r2 = −1 (Note that Re(r1 ) ≥ Re(r2 )) .
2
Note that r1 − r2 = 52 − (−1) = 72 which is not a nonnegative integer.
For each root of r, we get one series solution and hence we get two LI solutions.
Step 5: Finding Series Solution corresponding to r1 = 5/2
Step 5(a): Equating the coefficients of higher powers of (x − x0 ) to Zero

(2(n + r)(n + r − 1) − (n + r) − 5) an + an−1 = 0 where r = r1 = 5/2

n(2n + 7)an + an−1 = 0 for n ≥ 1
−an−1
an = for n ≥ 1
n(2n + 7)
Step 5(b): Expressing an for n ≥ 1 in terms of a0
−a0 −a1 a0 −a2 −a0
a1 = , a2 = = , a3 = = , ···
9 22 198 39 7722
In this case, it is difficult to get a general pattern. So, we have computed the first few
coefficients
Step 5(c): Writing the solution y1 (x)
!
1 1 2 1 3
y1 (x) = a0 x5/2 1 − x + x − x + · · · for 0 < x < ∞ .
9 198 7722
!
1 1 2 1 3
y1 (x) = a0 |x| 5/2
1− x+ x − x + · · · for 0 < |x| < ∞ .
9 198 7722
Step 6: Finding Series Solution corresponding to r2 = −1
Step 6(a): Equating the coefficients of higher powers of (x − x0 ) to Zero

(2(n + r)(n + r − 1) − (n + r) − 5) an + an−1 = 0 where r = r2 = −1

n(2n − 7)an + an−1 = 0 for n ≥ 1
−an−1
an = for n ≥ 1
n(2n − 7)
Step 6(b): Expressing an for n ≥ 1 in terms of a0
a0 a1 a0 a2 a0
a1 = , a2 = = , a3 = = , ···
5 6 30 3 90
In this case, it is difficult to get a general pattern. So, we have computed the first few
coefficients
Step 6(c): Writing the solution y2 (x)
!
1 1 2 1 3
y2 (x) = a0 x−1 1 + x + x + x + · · · for 0 < x < ∞ .
5 30 90
!
1 1 2 1 3
y2 (x) = a0 |x| 1 + x +
−1
x + x + · · · for 0 < |x| < ∞ .
5 30 90
Step 7: Writing the General Solution y(x) along with the domain of convergence

y(x) = A y1 (x) + B y2 (x)

where A and B are arbitrary real constants and
!
1 1 2 1 3
y1 (x) = |x|5/2 1 − x + x − x + · · · for 0 < |x| < ∞ ,
9 198 7722
!
1 1 1
y2 (x) = |x|−1 1 + x + x2 + x3 + · · · for 0 < |x| < ∞ .
5 30 90
Note that y1 (x) and y2 (x) are Linearly Independent solutions of the given ODE.
Note: To write the domain of convergence of each of above series, we observe that
there is no singular point other than x0 = 0 and hence the radius of convergence is the
distance from x0 to the nearest singular point that becomes ∞.
 Bessel Differential Equation of Order α
x2 y00 + xy0 + (x2 − α2 ) y = 0
Power Series Solutions about the Regular Singular Point x0 = 0

Friedrich Wilhelm Bessel (1784 - 1846) was a German Mathematician and

Astronomer. In 1824, while studying the dynamics of ‘many-body’ gravitational
systems, Bessel developed systematic analysis of solutions to this differential equation
and what are now known as Bessel functions. Critical for the solution of certain
differential equations, these functions are used throughout both classical and quantum
physics. The asteroid 1552 Bessel was named in his honour. The largest crater in the
Moon’s Mare Serenitatis is named Bessel after him.
Example for Case I (Refer Ross Section 6.3(B) on Pages 256-261)

Use the method of Frobenius to find solutions in some interval 0 < x < R of Bessel’s
equation x2 y00 + xy0 + (x2 − α2 )y = 0 of order α > 0, where 2α is not a positive integer.
Step 1: Writing Form of the Solution and its derivatives
Since x0 = 0 is a regular singular point of the given ODE, we seek solution of the form
∞
X
y(x) = xr an x n with a0 , 0
n=0

in 0 < x < R for some R > 0.

∞
X ∞
X
y0 (x) = (n + r)an xn+r−1 , y00 (x) = (n + r)(n + r − 1)an xn+r−2 .
n=0 n=0
Steps 2: Substituting Power Series of Coefficient functions, y(x), y0 (x) and y00 (x) in the
ODE
∞
X ∞
X ∞
X ∞
X
(n + r)(n + r − 1)an xn+r + (n + r)an xn+r + an xn+r+2 − α2 an xn+r = 0 .
n=0 n=0 n=0 n=0

Step 3: Gather together all terms of Same power of (x − x0 )

h i ∞ h
X i 
a0 (r − α )x + (r + 1) − α a1 x +
2 2 r 2 2 r+1
(n + r) − α an + an−2 xn+r = 0 .
2 2

n=2

Step 4: Equating the coefficient of the lowest power of x to Zero to get Indicial
Equation & its Roots

Indicial Equation : r2 − α2 = 0
Its roots are
Exponents : r1 = α and r2 = −α (Note that Re(r1 ) ≥ Re(r2 )) .
Note that r1 − r2 = α − (−α) = 2α which is not a positive integer.
For each root of r, we get one series solution. Hence we get two LI solutions.
Step 5: Finding Series Solution corresponding to r1 = α
Step 5(a): Equating the coefficients of higher powers of (x − x0 ) to Zero
 
Coeff. of xr+1 : (r + 1) − α a1 = 0
2 2
⇒ a1 = 0 .
Coeff. of xr+n : n(n + 2α)an + an−2 = 0 for n ≥ 2 .
−an−2
an = for n ≥ 2 .
n(n + 2α)
Step 5(b): Expressing an for n ≥ 2 in terms of a0
Since a1 = 0, all odd coefficients are zero.

(−1)n a0
a2n = for n ≥ 1 .
[2 · 4 · · · (2n)] [(2 + 2α)(4 + 2α) · · · (2n + 2α)]
(−1)n a0
a2n = 2n for n ≥ 1 .
2 n! [(1 + α)(2 + α) · · · (n + α)]
Step 5(c): Writing the solution y1 (x)
∞ 
n
(−1) x 2n
α
X 
y1 (x) = a0 x    for 0 < x < ∞ .
n=0
22n n! [(1 + α)(2 + α) · · · (n + α)] 
Gamma Function and Expressing y1 (x) using it
Define Z ∞
Γ(x) := e−t t x−1 dt for x > 0 .
0
Then, Γ(x + 1) = x Γ(x) for x > 0. In particular, Γ(1) = 1 and Γ(n + 1) = n!.
If α is not equal to positive integer then

Γ(n + α + 1)
(1 + α)(2 + α) · · · (n + α) = .
Γ(α + 1)
Therefore, y1 (x) can be expressed as
∞ 
n
(−1) x 2n
α
X 
y1 (x) = a0 Γ(α + 1) x   for 0 < x < ∞ .
n=0
22n n! Γ(n + α + 1) 

∞ 
(−1) n  x  2n+α 
α
X
y1 (x) = a0 2 Γ(α + 1)   for 0 < x < ∞ .

n=0
n! Γ(n + α + 1) 2
Bessel Functions of the First Kind of Order α or −α
In y1 (x), choose a0 = 1
2α Γ(α+1) so that

∞
X (−1)n  x 2n+α
y1 (x) = for 0 < x < ∞ .
n=0
n! Γ(n + α + 1) 2

The Bessel function of the First Kind of Order α is defined by

∞
X (−1)n  x 2n+α
Jα (x) = for 0 < x < ∞ .
n=0
n! Γ(n + α + 1) 2

Similarly, we get another solution for r = −α as

∞
X (−1)n  x 2n−α
J−α (x) = for 0 < x < ∞ .
n=0
n! Γ(n − α + 1) 2

General Solution: y(x) = A Jα (x) + B J−α (x) for 0 < x < ∞ .

Students to Work Out in the Class (r1 − r2 is NOT a Non-Negative
Integer)
Find the series solution of 3x y00 + y0 − y = 0 for x > 0 about the regular singular
point x0 = 0.
Answer:
Indicial Equation is r(3r − 2) = 0.
Indicial Equation Roots are
2
r1 = and r2 = 0 .
3
The two linearly independent solutions of the given equation are
 ∞ 
2 
X 1
y1 (x) = a0 |x| 3 1 + for x ∈ R .
 
n 
x 
n=1
n! (5 · 8 · · · (3n + 2))
 ∞ 
X 1
y2 (x) = a0 |x|0 1 + for x ∈ R .

 
n 
x 
n=1
n! (1 · 4 · · · (3n − 2))
Students to Work Out in the Class (r1 − r2 is NOT a Non-Negative
Integer)
Find the series solution of 2x y00 + (1 + x) y0 + y = 0 for x > 0 about the regular
singular point x0 = 0.
Answer:
Indicial Equation Roots are

1
r1 = and r2 = 0 .
2
The two linearly independent solutions of the given equation are
 ∞ n

1 
X (−1)
y1 (x) = a0 |x| 2 1 + for x ∈ R .
 n


n
x 
n=1
n! 2
 ∞ 
X (−1)n
y2 (x) = a0 |x|0 1 + for x ∈ R .

 
n 
x 
n=1
(1 · 3 · 5 · · · (2n − 1))
 Second Order Linear Homogeneous ODE
Power Series Solutions about the Regular Singular Points
Method of Frobenuis
Case II: r1 − r2 = N (N is a positive integer)
Case II: r1 − r2 = N (N is a positive integer)

Theorem 15
Consider the ODE y00 + P(x)y0 + Q(x)y = 0.
Let x0 be a regular singular point of the given ODE. Let r1 and r2 where
Re(r1 ) ≥ Re(r2 ) be the roots of the indicial equation associated with x0 . Suppose
r1 − r2 = N , where N is a positive integer. Then the given ODE has two nontrivial LI
solutions y1 and y2 in 0 < |x − x0 | < R for some R > 0 given by
∞
X
y1 (x) = |x − x0 |r1 cn (x − x0 )n where c0 , 0 ,
n=0

∞
X
y2 (x) = |x − x0 |r2 dn (x − x0 )n + A y1 (x) ln |x − x0 | where d0 , 0 .
n=0

Here the constant A may or may not be zero.

About the Second Solution, if r1 − r2 is a positive integer

If r1 − r2 = N is a positive integer, then

I Corresponding to the root r1 (having largest real part), we get one series solution
y1 (x) in 0 < |x − x0 | < R.
I Corresponding to the root r2 (having smallest real part), we get one series
∞
X
solution y∗2 (x) = |x − x0 |r2 dn (x − x0 )n in 0 < |x − x0 | < R.
n=0
I Now we have to check y1 (x) and y∗2 (x) are linearly independent or not in
0 < x − x0 < R. If they are Linearly Independent then we are through with solving
and hence we can write the general solution and stop the procedure. That is, we
are taking A = 0 in the expression of y2 (x) in the theorem mentioned in the
previous slide.
I If y1 (x) and y∗2 (x) are Linearly Dependent then we need to take A , 0 in the
expression of y2 (x) in the theorem mentioned in the previous slide. That is, y2 (x)
contains logarithm part ln |x − x0 | also.
If y1 (x) and y∗2 (x) are linearly dependent in 0 < x − x0 < R then
I Since y∗2 (x) = Ky1 (x) for some scalar K , there is no use of y∗2 (x).
I We need to obtain the second solution y2 (x) such that y1 (x) and y2 (x) are linearly
independent in 0 < x − x0 < R.
I By setting specific values to the constant a0 in y1 (x), write a particular solution
y∗1 (x) which does not have any arbitrary constants.
I Then, compute the second solution y2 (x) using the technique in the Method of
Reduction of Order R
e− P(x) dx
Z
y2 (x) = y∗1 (x) ∗ 2
dx
(y1 (x))
where y∗1 (x) is a known particular solution of y00 + P(x)y0 + Q(x)y = 0. This function
y2 (x) will contain the logarithm part ln(x − x0 ).
Example for Case II (Refer Ross Book Section 6.3(B) on Pages
256 - 262)

Use the method of Frobenius to find solutions in some interval 0 < x < R of Bessel’s
equation x2 y00 + xy0 + (x2 − p2 )y = 0 of order p, where 2p is a positive integer.
Subcases:
I 2p is an odd natural number: In this case, the obtained solutions corresponding to
r = p and r = −p are Linearly Independent. We can stop the procedure here. Or
else we can compute y2 (x) in the form with A , 0 as given in the theorem.
I 2p is an even natural number: In this case, the obtained solutions corresponding
to r = p and r = −p are Linearly Dependent. Therefore, we need to find another LI
solution by the method of reduction of order from the solution obtained for r = p
by setting the constant a0 to some specific value. The second solution obtained in
this manner is called the Bessel Functions of the Second Kind.
2p is an odd natural number ( p = n − 1
2 where n ∈ N)
I Corresponding to the indicial root r = p, we get one series solution as
∞
X (−1)k  x 2k+p
J p (x) = for 0 < x < ∞ .
k=0
k! Γ(k + p + 1) 2

It is a particular solution of the given ODE.

I Corresponding to the indicial root r = −p, we get another series solution as
∞
X (−1)k  x 2k−p
J−p (x) = for 0 < x < ∞ .
k=0
k! Γ(k − p + 1) 2

It is a particular solution of the given ODE.

I If 2p is an odd natural number, these two solutions J p (x) and J−p (x) are linearly
independent.

General Solution: y(x) = A J p (x) + B J−p (x) for 0 < x < ∞ .

2p is an even natural number ( p = n where n ∈ N)
I Corresponding to the indicial root r = n, we get one series solution as
∞ ∞
X (−1)k  x 2k+n X (−1)k  x 2k+n
Jn (x) = = for 0 < x < ∞ .
k=0
k! Γ(k + n + 1) 2 k=0
k! (k + n)! 2
This series actually converges for all x ∈ R. It is a particular solution of the given
ODE.
I Corresponding to the indicial root r = −n, we get one series solution as
∞
X (−1)k  x 2k−n
J−n (x) = for 0 < x < ∞ .
k=0
k! Γ(k − n + 1) 2
It is a particular solution of the given ODE.
These two solutions Jn (x) and J−n (x) are linearly dependent since
J−n (x) = (−1)n Jn (x) for all x ∈ R . (Exercise: Prove this identity.)
I In this case, by the method of reduction of order, the second LI solution is then
found to be the Bessel function of the second kind.
Find the second LI solution Yn (x)
Using the method of reduction of order, the another solution can be computed as
R dx
Z −
e x
Yn∗ (x) = Jn (x) 2
dx
(Jn (x))
Z
1/x
= Jn (x) dx .
(Jn (x))2

I With a suitable constant multiplied to Yn∗ (x), the function obtained is denoted by
Yn (x) and is called the Bessel function of the second kind of order n.
The function Yn (x) can also be obtained as

J p (x) cos(pπ) − J−p (x)

Yn (x) = lim .
p→n sin(pπ)
I The functions Jn (x) and Yn (x) are two LI solutions in the domain 0 < x < ∞ of the
given ODE in this case.
More Examples for Case II (Refer Ross Book Example 6.13 and 6.14)

 
Use the method of Frobenius to find solutions of x2 y00 − xy0 + x2 − 5
4 y = 0 in some
interval 0 < x < R.
Refer Ross Book, Example 6.13.

Important Example:
Use the method of Frobenius to find solutions of x2 y00 + (x2 − 3x)y0 + 3y = 0 in some
interval 0 < x < R.
Refer Ross Book, Example 6.14. Especially see, the first solution is computed in
closed form and then the second solution is computed using the method of reduction
of order.
Students to Work Out in the Class (r1 − r2 is a Positive Integer)
Find the series solution of x y00 + y = 0 for x > 0 about the regular singular point
x0 = 0.
Answer:
Indicial Equation is r(r − 1) = 0.
Indicial Equation Roots are

r1 = 1 and r2 = 0 .

Corresponding to r1 = 1, we get
∞ 
X (−1)n 
y1 (x) = a0 x  n
x  for x > 0 .
n=0
n! (n + 1)!

Corresponding to r2 = 0 also, we get the same solution y1 (x). So, we have to find out
another solution y2 (x) so that y1 (x) and y2 (x) are linearly independent.
By the Method of Reduction of Order, we can find the second linearly independent
solution.
Finding Second Solution
Using the method of reduction of order, the another solution can be computed as
y2 (x) = v(x)y1 (x), where
R 0 dx
e− x
Z
1
v(x) = dx = dx .
(y1 (x))2 (y1 (x))2
!
1 2 1 3 1 4
y1 (x) = x − x + x − x + ···
2 12 144
!
5 7 5 2 7 3
(y1 (x)) = x − x + x − x + · · · = x [1 − W] , W = x − x + x − · · ·
2 2 3 4 5 2
12 72 12 72
" # " #
1 1 h i 1 7 19 1 1 7 19
2
= 2
1 + W + W 2
+ · · · = 2
1 + x + x 2
+ x 3
+ ··· = 2 + + + x + ···
(y1 (x)) x x 12 72 x x 12 72
Z Z " #
dx 1 1 7 19 (−1) 7 19 2
v(x) = 2
= 2
+ + + x + · · · dx = + ln(x) + x + x +· · ·
(y1 (x)) x x 12 72 x 12 144
" #
(−1) 7 19 2
y2 (x) = y1 (x)v(x) = y1 (x) ln(x) + y1 (x) + x+ x + ··· for x > 0.
x 12 144
 Second Order Linear Homogeneous ODE
Power Series Solutions about the Regular Singular Points
Method of Frobenuis
Case III: r1 − r2 = 0
Case III: r1 − r2 = 0

Theorem 16
Consider the ODE y00 + P(x)y0 + Q(x)y = 0.
Let x0 be a regular singular point of the given ODE. Let r1 and r2 where
Re(r1 ) ≥ Re(r2 ) be the roots of the indicial equation associated with x0 . Suppose
r1 − r2 = 0. Then the given ODE has two nontrivial LI solutions y1 and y2 in
0 < |x − x0 | < R for some R > 0 given by
∞
X
y1 (x) = |x − x0 |r1 cn (x − x0 )n where c0 , 0 ,
n=0

∞
y2 (x) = |x − x0 |r1 +1
X
dn (x − x0 )n + y1 (x) ln |x − x0 | where d0 , 0 .
n=0
About the Second Solution, if r1 − r2 = 0

If r1 − r2 = 0, then
I Corresponding to the root r1 = r2 , we get one series solution y1 (x) in
0 < |x − x0 | < R.
I We need to obtain the second solution y2 (x) such that y1 (x) and y2 (x) are LI in
0 < x − x0 < R. The second solution y2 (x) can be computed from
R
−
e P(x) dx
Z
y2 (x) = y1 (x) dx
(y1 (x))2

where y1 (x) is a known particular solution of y00 + P(x)y0 + Q(x)y = 0.

Example for Case III (Refer Ross Book Section 6.3(A) on Pages
252 - 256)

Use the method of Frobenius to find solutions of Bessel’s equation of order zero:
x2 y00 + xy0 + x2 y = 0 in the domain 0 < x < R.
Step 1: Writing Form of the Solution and its derivatives
Since x0 = 0 is a regular singular point of the given ODE, we seek solution of the form
∞
X
y(x) = xr an x n with a0 , 0
n=0

in 0 < x − x0 < R for some R > 0.

∞
X ∞
X
y0 (x) = (n + r)an xn+r−1 , y00 (x) = (n + r)(n + r − 1)an xn+r−2 .
n=0 n=0
Steps 2: Substituting Power Series of Coefficient functions, y(x), y0 (x) and y00 (x) in the
ODE
∞
X ∞
X ∞
X
(n + r)(n + r − 1)an x n+r
+ (n + r)an x n+r
+ an xn+r+2 = 0 .
n=0 n=0 n=0

Step 3: Gather together all terms of Same power of (x − x0 )

∞ 
X 
a0 r x + (r + 1) a1 x
2 r 2 r+1
+ (n + r) an + an−2 xn+r = 0 .
2

n=2

Step 4: Equating the coefficient of the lowest power of x to Zero to get Indicial
Equation & its Roots

Indicial Equation : r2 = 0 .
Its root are
Exponents : r1 = r2 = 0 .
For this root of r = 0, we get one series solution y1 (x). So, we have to produce another
LI solution y2 (x) from the solution y1 (x).
Step 5: Finding Series Solution corresponding to r = 0
Step 5(a): Equating the coefficients of higher powers of (x − x0 ) to Zero
Coeff. of xr+1 : (r + 1)2 a1 = 0 .
Coeff. of xr+n : n2 an + an−2 = 0 for n ≥ 2 .
−an−2
an = for n ≥ 2 .
n2
Step 5(b): Expressing an for n ≥ 2 in terms of a0
Since a1 = 0, all odd coefficients are zero.

(−1)n a0
a2n = 2 2 2
for n ≥ 1 .
2 · 4 · · · (2n)
(−1)n a0
a2n = 2n 2
for n ≥ 1 .
2 (n!)
Step 5(c): Writing the solution y1 (x)
∞ 
X (−1)n x2n 
y1 (x) = a0  2n 2
 for 0 < x < ∞ .
n=0
2 (n!) 
Bessel Functions of the First Kind of Order 0

In y1 (x), choose a0 = 1 so that

∞
X (−1)n x2n
y1 (x) = 2n (n!)2
for x > 0 .
n=0
2

Actually it converges for all x ∈ R.

The Bessel function of the First Kind of Order 0 is defined by
∞ ∞
X (−1)n x2n X (−1)n x2n
J0 (x) = 2n 2
=1+ 2n 2
for all x ∈ R .
n=0
2 (n!) n=1
2 (n!)

Note that J0 (x) is a particular solution of the given ODE. We will now find another
solution y2 (x) so that J0 (x) and y2 (x) are linearly independent in x > 0.
Finding Second Solution
Using the method of reduction of order, the another solution can be computed as
R dx
e− x
Z
y2 (x) = J0 (x) dx
(J0 (x))2
Z
1/x
= J0 (x) 2
dx .
(J0 (x))
We know that
x2 x4 x6
!
J0 (x) = 1 − + − + ···
4 64 2304
1 x2 5x4 23x6
2
=1+ + + + ···
(J0 (x)) 2 32 576
2 4 6
Z ! !
1 x 5x 23x
y2 (x) = J0 (x) 1+ + + + · · · dx
x 2 32 576
x2 5x4 23x6
!
= J0 (x) ln(x) + J0 (x) + + + ···
4 128 3456
x2 3x4 11x6
!
= J0 ln(x) + − + + ···
4 128 13824
 ∞
X (−1)k+1 Hk
y2 (x) = J0 (x) ln(x) + x2k
k=1
22k (k!)2
k
X 1
where Hk = (Harmonic number). The functions y2 (x) and J0 (x) are LI solutions.
n=1
n
However a special combination of y2 (x) is taken as Y0 (x) (That is, y2 (x) multiplied with
suitable constants) as follows:
2
Y0 (x) = y2 (x) + (γ − ln(2))J0 (x)

π
where !
1 1
Euler constant γ = lim 1 + + + · · · − ln(n) ≈ 0.5772 .
n→∞ 2 3
 ∞

2 
  x   X k+1
(−1) Hk x 2k 
 
Y0 (x) =  ln + γ J0 (x) +  .
π 2 k=1
(k!)2 2
The function Y0 (x) is called the Bessel function of second kind of order 0.
The functions J0 (x) and Y0 (x) are two LI solutions of x2 y00 + xy0 + x2 y = 0.
Properties of Bessel Functions
Properties of Bessel Functions (of the First Kind)
Wherever necessary, assume that x , 0.
I Recurrence Relation:
2p
J p−1 (x) + J p+1 (x) = J p (x) for p ≥ 1 .
x
J p−1 (x) − J p+1 (x) = 2 J 0p (x) for p ≥ 1 .
I More Recurrence Relation:
p
J 0p (x) + J p (x) = J p−1 (x) for p ≥ 1 .
x
p
0
J p (x) − J p (x) = − J p+1 (x) for p ≥ 0 .
x
I If n ∈ N then J−n (x) = (−1)n Jn (x) for all x ∈ R.
I Generating Function: For n ∈ Z,
!! ∞
 x 1 X
G(x, t) := exp t− = Jn (x) tn .
2 t n=−∞
Bessel functions of the first kind of integral orders are the coefficients of the power
(Laurent) series of G(x, t) about t = 0.
Properties of Bessel Functions (continuation)

I
d n p o
x J p (x) = x p J p−1 (x) for p ≥ 1 .
dx
d n −p o
x J p (x) = −x−p J p+1 (x) for p ≥ 0 .
dx
I r
2
J 1 (x) = sin(x) for x > 0 .
2 πx
r
2
J −1 (x) = cos(x) for x > 0 .
2 πx
I If n is a natural number with n ≥ 2 then

22 Jn00 (x) = Jn−2 (x) − 2 Jn (x) + Jn+2 (x) for x ∈ R .

Graphs of Bessel Functions of the First Kind
Orthogonality of Bessel Functions J p , p ≥ 0

Let J p (x) be the Bessel function of the first kind of order p ≥ 0.

Then, J p (x) has infinitely many zeros in (0, ∞).
We label the positive zeros of J p (x) by λk for k ∈ N.

The Bessel function of the first kind of order p ≥ 0 satisfy


1 if m , n
 0
Z 
x J p (λm x) J p (λn x) dx =  .

2
0  (J p+1 (λn ))
 if m = n
2
Proofs/Derivations of Important Properties

The detailed proofs/derivations of important properties of Legendre Polynomials and

Bessel functions are available in many books and in the internet.

Proofs/derivations of these properties (like Rodrigues formula, Recurrence relations,

Identities, etc) are left as exercises to the students and it may be asked to show/prove
these properties in the tests/exams of MA102.