MA 102 (Mathematics II - Ordinary Differential Equations)

Department of Mathematics, IIT Guwahati

Semester: January-April, 2024

Solution of Tutorial Sheet No. 4 Date of Discussion: April 12, 2024

Lectures 10-14: Singular points, Power series solution for second-order ODE,
Bessel functions, Legendre polynomials

Attention students: Please keep in mind that it is not possible to discuss all problems
during the tutorial. You can consider the problems not discussed as your practice problems.

1. Determine the radii of convergence of the following power series:

∑
∞
n2 ∑∞
3 ∑∞
2−n
(a) n
(x + 2) n
; (b) 3
(x − 2)n
; (c) (x − 1)n .
n=0
2 n=1
n n=0
n+1
Solution:
The radii of convergence can be easily found by using ratio test: (a) R = 2, (b) R = 1,
(c) R = 2.

2. Classify the singular points of the following differential equations:

(a) ex y ′′ − (x2 − 1)y ′ + 2xy = 0; (b) (x2 + x)y ′′ + 3y ′ − 6xy = 0;

(c) (sin x)y ′′ + (cos x)y = 0; (d) ln(x − 1)y ′′ + (sin 2x)y ′ − ex y = 0, for x > 1.
Solution:
(a) The given differential equation can be written as

y ′′ − (x2 − 1)e−x y ′ + 2xe−x y = 0.

Here, P (x) = −(x2 − 1)e−x , Q(x) = 2xe−x . Both are analytic at all x ∈ R. Therefore,
all x ∈ R are ordinary points. Hence, this differential equation does not have any
singular point.
(b) The given differential equation can be written as
3 6
y ′′ + y′ − y = 0.
x(x + 1) x+1
3 6
Here, P (x) = , Q(x) = − . It is obvious that x = 0 and x = −1 are the
x(x + 1) x+1
singular points. Using the expressions (x − x0 )P (x) and (x − x0 )2 Q(x) for both x = 0
and x = −1, it can be found that both are regular singular points.
(c) The given differential equation can be written as
cos x
y ′′ + y = 0.
sin x

1
 cos x
Here, P (x) = 0, Q(x) = . Hence, Q(x) is not analytic at x = nπ, n ∈ Z.
sin x
Therefore, {nπ, n ∈ Z} is the set of singular points.
cos x
Since lim (x − nπ) exists, x = nπ are regular singular points for all n.
x→nπ sin x
(d) The given differential equation can be written as
sin 2x ′ ex
y ′′ + y − y = 0.
ln(x − 1) ln(x − 1)
sin 2x ex
Here, P (x) = , Q(x) = − . Hence, Q(x) is not analytic at x = 2,
ln(x − 1) ln(x − 1)
making it a singular point. Check it to get it as a regular singular point.

3. Find the indicial equation and their roots of the following differential equations:

(a) (x2 − x − 2)y ′′ + (x2 − 4)y ′ − 6xy = 0; (b) x2 y ′′ + 2(x − 3x2 )y ′ + ex y = 0;

(c) x2 y ′′ + (sin x)y ′ + (cos x)y = 0.
Solution:
(a) The given differential equation can be written as
x+2 ′ 6x
y ′′ + y − y = 0.
x+1 (x − 2)(x + 1)

It is obvious that x = −1 and x = 2 are regular singular points. We know that the
indicial equation is given by

m(m − 1) + mp0 + q0 = 0.

Now, since we have two regular singular points, the indicial equation will be different
corresponding to each singular point.
The indicial equation in the neighbourhood of x = 2 is

m(m − 1) = 0 ⇒ m = 0, m = 1.

This is because
(x + 2)
p0 = lim (x − 2) = 0,
x→2 x+1
(x + 2)(−6x)
q0 = lim (x − 2)2 = 0.
x→2 (x + 1)(x + 2)

Similarly, in the neighbourhood of x = −1, we have

(x + 2)
p0 = lim (x + 1) = 1,
x→−1 x+1
(−6x)
q0 = lim (x + 1)2 = 0.
x→−1 (x + 1)(x + 2)

Hence, the indicial equation in the neighborhood of x = −1 is

m(m − 1) + m + 0 = 0 ⇒ m = 0, m = 0.

2
 (b) The given differential equation can be written as
( )
′′ 1 − 3x 1 + x + x2 /2! + · · ·
y +2 y′ + y = 0.
x x2
It is clear that x = 0 is a regular singular point and also that p0 = 2, q0 = 1. The
1 3i
indicial equation is m(m−1)+2m+1 = 0 or, m2 +m+1 = 0 which gives m = − ± .
2 2
(c) The given differential equation can be written as
sin x ′ cos x
y ′′ + y + 2 y = 0.
x2 x
By expressing sin x and cos x as infinite series x, we can rewrite the differential equation
as
′′ x − x3 /3! + x5 /5! + · · · ′ 1 − x2 /2! + x4 /4! + · · ·
y + y + y = 0.
x2 x2
It is clear that x = 0 is a regular singular point and also that p0 = 1, q0 = 1. The
indicial equation is m(m − 1) + m + 1 = 0 or, m2 + 1 = 0 which gives m = ±i.

4. Find a series solution about the ordinary point of each of the following differential
equations subject to the given initial conditions:

(a) (4 − x2 )y ′′ + 2y = 0, y(0) = 0, y ′ (0) = 1;

(b) y ′′ − x2 y ′ + y sin x = 0, y(0) = 0, y ′ (0) = 1.
Solution:
(a) The given equation is (4 − x2 )y ′′ + 2y = 0. It is clear that x = 0 is an ordinary
point. Hence, we can assume a solution of the form
∑
∞
y= ak xk .
k=0

Finding the expression of y ′′ and putting it, along with y, back into the equation, we
get
∑∞ ∑∞ ∑
∞
y= ak k(k − 1)xk−2 − ak k(k − 1)xk + 2 ak xk = 0.
k=2 k=2 k=0

By making all series summations from 0 to ∞ and equating the coefficient of xk , we

get the recursion relation as
2 − k(k − 1)
ak+2 = − ak .
4(k + 2)(k + 1)
This gives us
2 1
a2 = − a0 = a0
4·2 4
2 1
a3 =− a1 = − a1
4·3·2 12
a4 =0
4 1
a5 = a3 = − a1
4·5·4 240
a6 = a8 = · · · = 0.

3
 Using the initial condition y(0) = 0 gives a0 = 0 and hence a2 = 0.
Using the initial condition y ′ (0) = 1 gives a1 = 1.
Hence the solution can be written as
1 3 1 5
y = x− x − x − ···
12[ 240 ]
1 ( x )3 1 ( x )5
= x−2 + + ···
3 2 15 2
∑∞
1 ( x )2n+1
= x−2 .
n=1
4n2 − 1 2

(It is not always necessary to write the solution in a series form.)

(b) The given equation is y ′′ − x2 y ′ + y sin x = 0. It is clear that x = 0 is an ordinary
point. Hence, we can assume a solution of the form
∑
∞
y= ak xk .
k=0

Finding the expression of y ′′ , y ′ and putting them, along with y, back into the equation,
we get
∑
∞ ∑∞ ∑∞ ∑∞
xk+3 ∑ xk+5
∞
(k+2)(k+1)ak+2 x − (k+1)ak+1 x +
k k+2
ak x −
k+1
ak + ak +· · · = 0,
k=0 k=0 k=0 k=0
3! k=0
5!

x3 x5
where we have used the series expression of sin x = x −
+ + · · · . Using the given
3! 5!
conditions, we get a0 = 0 and a1 = 1. Equating the coefficients of xk , we get the
following (infinite) multi-term recursion relation:
ak−3 ak−5
(k + 2)(k + 1)ak+2 + (2 − k)ak−1 − + − · · · = 0.
3! 5!
Now,

k = 0 ⇒ a2 = 0
k = 1 ⇒ a3 = 0
k = 2 ⇒ a4 = 0
k = 3 ⇒ a5 = 0
a1 1
k = 4 ⇒ (6 · 5)a6 − ⇒ a6 =
3! 180
k = 5 ⇒ a7 = 0
a1 1
k = 6 ⇒ (8 · 7)a8 + ⇒ a8 = −
5! 6720
and so on. The solution can be written as
x6 x8
y =x+ − + ··· .
180 6720

5. Find a series solution about the regular singular point of each of the the following
differential equations:

4
 (a) xy ′′ + 4y ′ − xy = 0; (b) 4x2 y ′′ + 2x2 y ′ − (x + 3)y = 0.
Solution: (a) For this differential equation, x = 0 is a regular singular point. Hence,
we need to use Frobenius series solution:
∑
∞
y= ak xm+k .
k=0

Putting y, y ′ and y ′′ in the differential equation, we get

∑
∞ ∑
∞ ∑
∞
ak (m + k)(m + k − 1)x m+k−1
+4 ak (m + k)x m+k−1
− ak xm+k+1 = 0.
k=0 k=0 k=0

The indicial equation can be obtained as

m2 + 3m = 0,

giving m = 0, −3.. The recursion relation can be obtained as

ak−2
ak = , k ≥ 2.
(m + k)(m + k + 3)
Case 1: m = 0
The recursion relation is
ak−2
ak = , k ≥ 2.
k(k + 3)
It can be found that a1 = 0. The general expression for ak is
a0
a2k = n .
2 n!(2n + 3)(2(n − 1) + 3) · · · 7 · 5
Subsequently, the series solution can be obtained as
( )
x2 x4
y = a0 1 + + + ··· .
10 280
Case 2: m = −3
The recursion relation is
ak−2
ak = , k ≥ 2.
(k − 3)k
Proceed similarly as in Case 1.
(b) Proceed similarly as above by using Frobenius series solution. The indicial equation
3 1
can be obtained as 4m2 − 4m − 3 = 0 giving m = , − . Then, proceed in the usual
2 2
way to find the solution.

6. Use
∫ 1 the orthogonality property of Legendre polynomials to show that
g(x)Pn (x)dx = 0 for every polynomial g(x) with deg(g(x)) < n.
−1
Solution:
∑
m
Since g(x) is a polynomial of degree m with m < n, we can write g(x) = ak Pk (x).
k=0
Therefore,
∫ 1 ∑
m ∫ 1
g(x)Pn (x) dx = ak Pk (x)Pn (x) dx = 0,
−1 k=0 −1

since k ≤ m < n.

5
7. Find the specific solution to the differential equation (1−x2 )y ′′ −2xy ′ +12y = 0 subject
to y ′ (0) = 4 and the function y(x) is well-behaved at x = 1.
Solution: The given differential equation has the solution

y = AP3 (x) + BQ3 (x).

Since Q3 (x) is not bounded when x → ±1, we take B = 0 for a bounded solution.
Therefore, the solution can be written as
1
y = AP3 (x) = A (5x3 − 3x).
2
Using the given condition y ′ (0) = 4, we get A = −8/3. Hence, the specific solution is
4
y = − (5x3 − 3x).
3

8. Using the substitution cos θ = x, show that the differential equation

( )
1 d dy
sin θ + n(n + 1)y = 0
sin θ dθ dθ

can be converted to the standard form of Legendre equation.

Solution: The given differential equation can be rewritten as

d2 y dy
2
+ cot θ + n(n + 1)y = 0.
dθ dθ
dx √
Since x = cos θ, we have = − sin θ = − 1 − x2 . This gives us
dθ
dy √ dy
= − 1 − x2 ,
dθ dx
d2 y d2
y dy
2
= (1 − x2
) 2
−x .
dθ dx dx
Putting these back in the given differential equation, we get

d2 y dy
(1 − x2 ) 2
− 2x + n(n + 1)y = 0.
dx dx

9. Find a general solution to each of the following differential equations in terms of Bessel
functions:

(a) 4x2 y ′′ + 4xy ′ + (4x2 − 1)y = 0; (b) x2 y ′′ + xy ′ + x2 y = 0;

(c) x2 y ′′ + xy ′ + (16x2 − 1/9)y = 0; (d) x2 y ′′ + xy ′ + ((4/9)x2 − 9/4)y = 0.
Solution:
(a) y = c1 J1/2 (x) + c2 J−1/2 (x)
(b) y = c1 J0 (x) + c2 Y0 (x)
(c) y = c1 J1/3 (4x) + c2 J−1/3 (4x)
(d) y = c1 J3/2 ((2/3)x) + c2 J−3/2 ((2/3)x)

6
10. The differential equation x2 y ′′ + (x2 − 2)y = 0, x > 0 does not look like a Bessel’s
√
equation. However, show that, by making a substitution y = v(x) x of the dependent
variable, its solution can be expressed in terms of Bessel functions.
Solution: The given differential equation is x2 y ′′ + (x2 − 2)y = 0, x > 0. The given
√
substitution is y = v(x) x. This will lead to

dy dv 1 −1/2
= x1/2 + x v,
dx dx 2
d2 y 2
1/2 d v dv 1 −3/2
2
= x 2
+ x−1/2 − x v.
dx dx dx 4
Putting all these back in the differential equation and after some simplification, we get
2
( )
2d v dv 9
x +x + x − 2
v = 0,
dx2 dx 4

which is Bessel’s equation in the dependent variable z. Its solution is

v(x) = c1 J3/2 (x) + c2 J−3/2 (x).

Writing in the original variables,

y(x) = x1/2 (c1 J3/2 (x) + c2 J−3/2 (x)).

The End