Course code: MAT 208 Course Title: Engineering Mathematics II Target Groups: SoENG

Department: Physics, Mathematics and Computer Science Lecturer: Nixon Ronoh

Topic: 4 Title: Linear Second Order Homogeneous Differential Equations Date: 21/05/2025

4.1. Introduction

4.1.1. Linear Second Order Homogeneous Differential Equations

Linear second order homogeneous differential equations are equations of the form

d2 y dy
p(x) 2
+ Q(x) + R(x)y = 0 (4.1)
dx dx

where P , Q and R are functions of x only. It is important to consider Equation (4.1) at specific points x = x0 ,
with the following definitions

Definition 4.1.1 (Ordinary point). If at the point x = x0 , the coefficient p(x) 6= 0, then x0 is said to be an
ordinary point of Equation (4.1)

Definition 4.1.2 (Regular Singular Point). If at the point x = x0 , the coefficient p(x) = 0, and the expressions

Q(x) R(x)
x and x2 (4.2)
P (x) P (x)

have convergent Taylor series, then the point x = x0 is a regular singular point of Equation (4.1)

Definition 4.1.3 (Singular Point). If at the point x = x0 , the coefficient p(x) = 0, and the expressions

Q(x) R(x)
x and x2 (4.3)
P (x) P (x)

have no convergent Taylor series, then the point x = x0 is a singular point of Equation (4.1)

Theorem 4.1.4. If the point x = x0 is an ordinary point of Equation (4.1), then the power series solution will
be of the form
X1
y= ar xr , ar 6= 0 (4.4)
r=0

Theorem 4.1.5. If the point x = x0 is a regular singular point of Equation (4.1), then the power series solution
will be of the form
X1
y= ar xr+c , ar 6= 0 (4.5)
r=0

where c is some arbitrary constant.

In this section we look at some functions that are of interest in Applied Mathematics and Engineering. We
shall limit ourselves to Bessel and Legendre functions.

15
4.2. Bessel Function

The Bessel equation, named after German mathematician and astronomer Friedrich Wilhelm Bessel finds
utility in many Engineering applications, especially those that involve cylindrical coordinates. The Bessel
equation is a linear second order ordinary differential equation of the form

x2 y 00 + xy 0 + (x2 ⌫ 2 )y = 0 (4.6)

The parameter ⌫ is the order of the Bessel equation. Friedrich, in 1824, showed that Equation (4.6) could be
solved through a special class of functions called cylinder or Bessel functions.

Example 4.2.1. The task is to find a power series solution to the Bessel equation of order 0. This is the
equation

x2 y 00 + xy 0 + x2 y = 0 (4.7)

As seen before in Section 3.1, we first assume that Equation (4.7) has a series solution of the form (3.1), and
that the first and the second derivatives y, y 0 and y 00 conform to the forms (3.3) and (3.4), namely

1
X
y= an x n (4.8)
n=0
X1
y0 = n · an x n 1
(4.9)
n=1
X1
y 00 = n · (n 1) · an xn 2
(4.10)
n=2

Substituting the power series expressions of y, y 0 and y 00 into Equation 4.7 leads to the following;

1
! 1
! 1
!
X X X
x2 n · (n 1) · an xn 2
+x n · an x n 1
+ x2 an x n =0 (4.11)
n=2 n=1 n=0

We next apply the left distributive law of multiplication to open the brackets;

1
X 1
X 1
X
n · (n 1) · an x + n
n · an x +n
an xn+2 = 0 (4.12)
n=2 n=1 n=0

We now proceed to

P1 P1
• shift the second term by one term; n=1 n · an x n = a1 x + n=2 n · an xn

• re-index the third term so that the powers of x are similar in all terms. This is done through the replace-
ment n = n 2.

This leads to

16
 1
X 1
X 1
X
n · (n 1) · an xn + a1 x + n · an x n + an 2 x n = 0 (4.13)
n=2 n=2 n=2
1
" 1 1
#
X X X
a1 x + n · (n 1) · an x + n n
n · an x + an 2 x n
=0 (4.14)
n=2 n=2 n=2
1
X
a1 x + [n · (n 1) · an + n · an + an 2 ] xn = 0 (4.15)
n=2
X1
⇥ ⇤
a1 x + (n2 n) · an + n · an + an 2 xn = 0 (4.16)
n=2
1
X ⇥ ⇤
a1 x + n 2 an + an 2 xn = 0 (4.17)
n=2

We now equate the coefficients of the powers of x to obtain

a1 = 0 for n = 1 (4.18)
and n an + an 2 = 0 for all n 2
2
(4.19)
1
or an = an 2 for all n 2 (4.20)
n2

Using Equation (4.20), we deduce that;

1 1
for n = 2; a2 = a0 = a0 (4.21)
4 22
1
for n = 3; a3 = a1 = 0 (4.22)
9
1 1 1 1
for n = 4; a4 = a2 = · a0 = 2 2 a0 (4.23)
16 16 ✓ 4 ◆ 4 · 2
1 1 1
for n = 5; a5 = a3 = · a1 = 0 (4.24)
25 25 9
1 1 1 1 1
for n = 6; a6 = a4 = ·( · a0 ) = a0 (4.25)
36 36 16 4 62 · 42 · 22
for n = 7; a7 = 0 =) for all odd n , an = 0 (4.26)
1 1 1 1 1 1
for n = 8; a8 = a6 = ·( · · a0 ) = 2 2 2 2 a0 (4.27)
64 64 36 16 4 8 ·6 ·4 ·2
Generally, for all odd n, the coefficients an = 0 since a1 = 0. Notice that for all even n,

( 1)k
a2k = a0 (4.28)
2(2k) (k!)2

The solution to Equation (4.7) is thus

1
X X1
( 1)k 2k
y= a2k x 2k
= a0 x (4.29)
n=0 n=0
2(2k) (k!)2

17
Figure 4.1: Plot showing Bessel functions of orders 0,1 and 5

18
 Course code: MAT 208 Course Title: Engineering Mathematics II Target Groups: SoENG
Department: Physics, Mathematics and Computer Science Lecturer: Nixon Ronoh
Topic: 5 Title: Legendre Functions Date: 28/05/2025

5.1. Legendre Function

Consider the ordinary differential equation

d2 y dy
(1 x2 ) 2
2x + n(n + 1)y = 0 (5.1)
dx dx
In this equation, we note that p(x) = 1 x2 , and further that P (0) = 1 0 = 1 at the point x = 0. This is
the Legendre Differential Equation. We further note, using Definition (4.1.1), that x = 0 is an ordinary point
of Equation (5.1). The task is to solve this equation using power series.

We first let
1
X= =x 1
(5.2)
x
dX
which leads to = x 2
= X2 (5.3)
dx
dy
and seek expression for y 0 = dx
in terms of X;

dy dy dX
= · (5.4)
dx dX dx
dy
from Equation (5.3) = · ( X 2) (5.5)
dX
dy dy
OR = X2 · (5.6)
dx dX
2
d y
With Equation (5.6), we can obtain an expression for y” = dx 2 , in terms of the new variable X;

✓ ◆
d2 y d dy
= · (5.7)
dx2 dx dx
✓ ◆
d dy
from Equation (5.6) = · X ·2
(5.8)
dx dX
✓ ◆
d dy dX
= · 2
X · · (5.9)
dX dX dx
✓ ◆
d dy
= · X2 · · ( X 2) (5.10)
dX dX
✓ ◆
d2 y dy
= X · 2
2X · · X2 (5.11)
dX 2 dX
d2 y d2 y dy
OR 2
= X 4
· 2
+ 2X 3 · (5.12)
dx dX dX
We substitute the expressions in Equations (5.2), (5.6) and (5.12) into equation (5.1), so that the variable x is

19
now fully replaced by X.
d2 y dy
x2 ) 2 2x + n(n + 1)y = 0
(1 (5.13)
✓ 2
◆ ✓dx dx
◆
dy dy dy
(1 2 4
X )· X · + 2X ·3
2X 1
X ·2
+ n(n + 1)y = 0 (5.14)
dX 2 dX dX
d2 y dy
(X 4 X 2 ) · + 2X 3 2X + 2X · + n(n + 1)y = 0 (5.15)
dX 2 dX

We now have an Equation devoid of x, and fully in terms of X, and which we can now proceed to solve using
a power series solution. That is;

d2 y dy
(X 4 X 2) · 2
+ 2X 3 · + n(n + 1)y = 0 (5.16)
dX dX

To see this, first note that at the point X = 0, we have the following;
P (X) = X 4 X2 (5.17)
P (X = 0) = 0 0 = 0 (5.18)
Q(X) R(X)
With P (X0 ) = 0, we proceed to check convergence of the term X · P (X)
and X 2 · P (X)
;

Q(X) 2X 3 2X 2
X· =X· 4 = = 2X 2 (1 X 2) 1
(5.19)
P (X) X X2 X2 1
= 2X 2 (1 + X 2 + X 4 + X 6 + · · · ) (5.20)

Equation (5.20) is a series convergent for 1 < X < 1. Next, we check the convergence of the term X 2 · PR(X)
(X)
;

R(X) n(n + 1) n(n + 1)

X2 · = X2 · 4 = = n(n + 1) · X 2
· (1 X 2) 1
(5.21)
P (X) X X2 X 2 (1 X 2 )
= n(n + 1) · X 2
· (1 + X 2 + X 4 + X 6 + · · · ) (5.22)
Equation (5.20) is a series convergent for 1 < X < 1.

With equations (5.20) and (5.20), since P (X0 ) = 0, we note that X0 = 0 is a regular singular point of Equation
(5.16). The series solution will be of the form of Equation (4.5) (see Theorem (4.1.5)). We proceed as follows;
1
X
y= ar X r+c , ar 6= 0 (5.23)
r=0
X1
= ar x r c
, ar 6= 0 (5.24)
r=0

Equation (5.24) is called the descending solution of Equation (5.1). We can write out a few terms of the series;
y = a0 · x c
+ a1 · x 1 c
+ a2 · x 2 c
+ a3 · x 3 c
+ ··· (5.25)
where the coefficients ar , r = 0, 1, 2, · · · have to be determined ( using the frobenious or any other appropriate
method). We now proceed to find expressions for y 0 .
1
dy X
= ( r c) ar x r c 1
(5.26)
dx r=0
1
X
= (r + c) ar x r c 1
(5.27)
r=0

20
The derivative of Equation (5.27) leads to the expression for y 00 ;
1
d2 y X
= ( r c) ( r c 1) ar x r c 2
(5.28)
dx2 r=0
1
X
= (r + c) ( r c 1) ar x r c 2
(5.29)
r=0
X1
= (r + c) (r + c + 1) ar x r c 2
(5.30)
r=0

Equations (5.24), (5.27) and (5.30) can now be substituted into Equation (5.1);

1
X 1
X 1
X
(1 x2 ) · (r + c) (r + c + 1) ar x r c 2
2x · (r + c) ar x r c 1
+ n(n + 1) ar x r c
=0
r=0 r=0 r=0
(5.31)
Opening up the brackets on the first term;
1
X 1
X 1
X 1
X
r c 2 r c r c r c
(r+c) (r+c+1) ar x (r+c) (r+c+1) ar x 2· (r+c) ar x +n(n+1) ar x =0
r=0 r=0 r=0 r=0
(5.32)
Collecting the like terms in the powers of x;
1
X 1
X
(r + c) (r + c + 1) ar x r c 2
[(r + c) (r + c + 1) 2(r + c) n(n + 1)] ar x r c
= 0 (5.33)
r=0 r=0

This expression can be simplified further;

1
X 1
X
(r + c) (r + c + 1) ar x r c 2
[(r + c) (r + c 1) n(n + 1)] ar x r c
=0 (5.34)
r=0 r=0

The first term of Equation (5.34) can be rewritten with the replacement r = r 2, so as to match the powers
of x.

1
X 1
X
((r 2)+c) ((r 2)+c+1) a(r 2) x (r 2) c 2
[(r +c) (r +c 1) n(n+1)] ar x r c
= 0 (5.35)
r 2=0 r=0

We note here that after this replacement, the summation index now begins at r = 2;

1
X 1
X
(r + c 2) (r + c 1) ar 2 x r c
[(r + c) (r + c 1) n(n + 1)] ar x r c
=0 (5.36)
r=2 r=0

The expanded equation follows, where an initial evaluation is made for r = 0 and for r = 1, and the summation
now begins from r = 2.
[c (c 1) n(n + 1)] a0 x c [(c + 1) c n(n + 1)] a1 x 1 c

X1
r c
[(r + c) (r + c 1) n(n + 1)] ar x
r=2
1
X
r c
+ (r + c 2) (r + c 1) ar 2 x =0
r=2

21
 (5.37)
Assume (rightfully so) that the right hand side of the expression in Equation (5.37) is a sum of zeroes. Since
a0 6= 0, we start by equating the coefficients of x c to 0;
[c (c 1) n(n + 1)] = 0 (5.38)
OR c (c 1) n(n + 1) = 0 (5.39)

The expression in Equation (5.39) is called the indicial equation. We have

c (c 1) n(n + 1) = c2 c n2 n = 0 (5.40)
OR c2 n2 (c + n) = 0 (5.41)
OR (c + n)(c n) (c + n) = 0 (5.42)
OR (c + n)(c n 1) = 0 (5.43)
This implies that either c= n (5.44)
OR c=n+1 (5.45)
Equating the coefficients of x 1 c
to 0 (see the second term of Equation (5.37));
[(c + 1) c n(n + 1)] a1 =0 (5.46)
OR [c (c + 1) n(n + 1)] a1 =0 (5.47)
OR [c2 + c n2 n] a1 =0 (5.48)
OR [c2 n2 + c n] a1 =0 (5.49)
OR [(c + n)(c n) + (c n)] a1 =0 (5.50)
OR [(c n)(c + n + 1)] a1 =0 (5.51)

Substituting equation (5.44) into (5.51);

[(c n)(c + n + 1)] a1 = 2n · 1 · a1 = 0 (5.52)
=) a1 = 0 (5.53)

Substituting equation (5.45) into (5.51);

[(c n)(c + n + 1)] a1 = (n + 1 n) · (n + 1 + n + 1) = 2(n + 1) · a1 = 0 (5.54)

=) a1 = 0 (5.55)

So far, we made evaluations for Equation (5.37) at r = 0 and at r = 1. Equating the coefficients of the
remaining terms of the powers of x to 0;

1
X 1
X
[(r + c) (r + c 1) n(n + 1)] ar x r c
+ (r + c 2) (r + c 1) ar 2 x r c
=0 (5.56)
r=2 r=2
1
X
OR ([(r + c) (r + c 1) n(n + 1)] ar + (r + c 2) (r + c 1) ar 2 ] x r c
=0 (5.57)
r=2

Since x r c
6= 0, we have for all r 2
[(r + c) (r + c 1) n(n + 1)] ar + (r + c 2) (r + c 1) ar 2 =0 (5.58)
(r + c 2) (r + c 1)
=) ar = ar 2 (5.59)
(r + c) (r + c 1) n(n + 1)

22
Equation (5.59) is a recurrence relation. Without loss of generality, and to allow us to clearly see the recurrence
relationship, we let
(r + c 2) (r + c 1)
coef f = (5.60)
(r + c) (r + c 1) n(n + 1)
Notice that ar = coef f · ar 2 , so that each subsequent expression of ar is a coef f multiple of ar 2 ;
Recurrence ar = coef f · ar 2 (5.61)
For r = 0 ar = a0 (5.62)
From Equation (5.55) r = 1 ar = a1 = 0 (5.63)
For r = 2 a2 = coef f · a0 (5.64)
For r = 3 a3 = coef f · a1 = 0 (5.65)
For r = 4 a4 = coef f · a2 (5.66)
For r = 5 a5 = coef f · a3 = 0 (5.67)
For r = 6 a6 = coef f · a4 (5.68)
··· (5.69)
··· (5.70)
From the above expressions (see Equations (5.55), (5.65) and (5.67)), we can deduce that since a1 = 0 for all
r odd, the equivalence relation ar = 0 holds. That is
a1 = a3 = a1 = a5 = · · · = a2k 1 = 0, k = 1, 2, 3, · · · (5.71)
It now remains to evaluate for the ar , r 2 for r even. We proceed as follows;

Suppose c = n (see Equation (5.43));

(r n 2) (r n 1)
ar = ar 2 (5.72)
(r n) (r n 1) n(n + 1)
( n) (1 n)
At r = 2 : a2 = a0 (5.73)
(2 n) (1 n) n(n + 1)
( n) (1 n)
= a0 (5.74)
2 3n + n2 n2 n
( n) (1 n)
= a0 (5.75)
2 4n
n (n 1)
= a0 (5.76)
2(2n 1)
(2 n) (3 n)
At r = 4 : a4 = a2 (5.77)
(4 n) (3 n) n(n + 1)
(2 n) (3 n)
= a2 (5.78)
12 7n + n2 n2 n
(2 n) (3 n)
= a2 (5.79)
12 8n
(n 2) (n 3)
= a2 (5.80)
4(2n 3)
(4 n) (5 n)
At r = 6 : a6 = a4 (5.81)
(6 n) (5 n) n(n + 1)
(4 n) (5 n)
= a4 (5.82)
30 9n + n2 n2 n
(4 n) (5 n)
= a4 (5.83)
30 12n
(n 4) (n 5)
= a4 (5.84)
6(2n 5)

23
We can now generalize the recurrence relation by considering the series for a0 , a2 , a4 and a6 .

n (n 1)
a2 = a0 (5.85)
2(2n 1)
(n 2) (n 3) (n 2) (n 3) n (n 1)
a4 = a2 = · a0 (5.86)
4(2n 3) 4(2n 3) 2(2n 1)
n (n 1)(n 2) (n 3)
= (5.87)
1 · 2 · 22 (2n 1)(2n 3)
(n 4) (n 5) (n 4) (n 5) n (n 1)(n 2) (n 3)
a6 = a4 = · (5.88)
6(2n 5) 6(2n 5) 1 · 2 · 22 (2n 1)(2n 3)
n (n 1)(n 2) (n 3)(n 4) (n 5)
= (5.89)
1 · 2 · 3 · 23 (2n 1)(2n 3)(2n 5)
... (5.90)
n (n 1)(n 2) (n 3)(n 4) (n 5) · · · (n 2m + 1)
a2m 2 = ( 1)m a0 (5.91)
m! · 2m (2n 1)(2n 3)(2n 5) · (2n 2m + 1)

With the coefficients determined by Equation (5.91), it can now be shown that for the case where c = n, the
solution to Equation (5.1) is
[n]
X ( 1)m (2n 2m)! · xn 2m
yn (x) = · (5.92)
m=0
m! 2m (n m)! (n 2m)!
usually denoted as Pn (x), and called The Legendre Function.

The case where c = n + 1 (see Equation (5.44)) is of little interest in Applied mathematics and Engineering,
and is denoted as Qn (x).

Students are free to analyse this case at own time! Start with the expression trhat follows;

(r + n 1) (r + n)
ar = ar 2 (5.93)
(r + n + 1) (r + n) n(n + 1)

24