Malhimnticd Me&& in and v = x.

The general solution is, therefore,

Physics -11
y = Clcosx+C2sinx+ln~cosx~cosx+xs~~x a

Note that the method of undetermined coefficients could not be used to obtain yp
because secx is not a solution of the homogeneous linear differential equation.
(ii) The corresponding homogeneous equation y" - y = 0 has the g e ~ ~ e rsolution
al
y = C1 ex+C2 e-'. Because of the nature of the driving function, y, could not be
found by the methad of undetermined coefficients. In the method of variation of
parameters, we put yl = exand y2 = e-' in the expressions for u' and v' to obtain

x e x (2r
Therefore, u' = x / 2 and v' = from which it readily follows that
2
u = $4 and v -i - (xek/4 ) + ( e"h ). The general solutio~~
is then
1 ex
1 2ex- E e X + -
y = Clex+C2e-x+-X
4 4 8

Finally, we note that C le x and x 1 x

s a can be combined as(Cl t 8-) ex = Ce and the
general solutio~imay be written as .

y =
1
C1 ex+ C2e-'- -xeX + -x
1 2
ex
4 4
 UNIT 3 SECO
E
E
Structure
3.1 Introduction
Objectives
3.2 Some Terminology
3.3 Power Series Method
3.4 The Frobenius' Method
3.5 Summary
3.6 Terminal Questions
3.7 Solutions and Answers

3.1 INTRODUCTION
In Unit 2, you have learnt how to solve second order ODEs with constant coefficients. The
solutions of these equations are simple exponentials, trigonometric or hyperbolic fuilctions
known from calculus. But in many physical and engineering problems, we have to solve
second order ordinary differential equations with variable coefficients. For example, we
have to solve such equations to study the field distribution around a charged sphere or a
cylinderland energy production in a reactor. Similarly, when we wish to know how high
a vertical column of uniform cross-section can be extended upward until it buckles hnder its
own weight, we have to solve a second order ODE wit. variable coefficients. In such cases,
simple algebraic or transcendental solutions do not exist and methods discussed in Unit 2 do not
work. We, therefore; look for other methods.
One of the most elegant and efficient methods of solving such ODEs is the power series
method. This is so particularly because it facilitates numerical computations. Even so, it
has limited utility when coefficients of the given differential equation are not well
defined at some point. In such cases, we use an extension of the power series method,
called the Frobenius' method, You will leant these two methods in this unit. The
f
I properties of power series and certain other mathematical concepts are given in an
Appendix at tbe end of this unit, It would be better if you study the appendix before
studying this unit.
1
I

Objectives
After studying this unit, you should be able to

define ordinary and singular points

@ locate and classify the type of singularity

'
I
use power series method to solve a second order ODE about an ordinary point

@ use Frobenius' method to solve a second order ODE about a regular slngular point.
I

I 3.2 SOME TERMINOLOGY

While studying first and second order ODESwith constant coefficients, you have learnt
I
some basic terminology. You would come across some more common terms, which you do
a not know as yet, in reference to second order ODESwith variable coefficients. This section
is intended to familiarise you with these concepts,
I
 Mathematical Methods in Analytic.Functions
-
Phydcs I1
A function is said to be analytic atx = xo if it has a Finite value atx = xo . Mathematically,
it can be expressed as an infinite power series of the form

where ao,a l , ... are constants (also see Appendix A).

The tenn"radiusofconvergence'is For the particular case of xo 0, familiar examples of analytic functions are given in
defined in the Appendix A at the
end of this unit. Table 3.1.
Table 3.1: Power Series Expansion of Some Simple Functions Analytic at x = 0
Function Power Series Radius of Convergence

-
1 Ix I
1 -x ex.- l + x + x2 +.....
n-0

2 'j 5 -I+.+-+22! ... Ix l cap

n-0

sin x X-
2
X + -- ..a.
IxI<m
3! 51

COS X 1 x2 --
- -+ x4 . ..., IXI
2! 4 !
Ordinary and Singular Points
A second order ODE of the form

has an ordinary point at x = xo if the functionsp (x) and q(x) are analytic a t x = xo. On
the other hand, x = xois a singular point of Eq. (3.2) if it is a singular point of either
p (x) or q (x), ie., when either function is not analytic at x '= xo. For example, x = 0 is a
-
singularity of l/x as well as lnx, and in the equationx2y" + xy' 2y = O,p(x) and q(x)
-
have a singular point at x 0. Similarly, in the equation ( 1 x 2 ) y" - 2xyt + 6y = 0,
.p

p (x) and q(x) have singular points at +1and -1.

Regular Singularity
A singular pointx = xo of the differential equation (3.2) is said to be regular whenp(x) 1
-
and/or q(x) diverge but?&%{ x - x, ) p (x) andJimxo(x xo )' q(x) remain finite. If a singular
point is not regular, it is called irregular. That is, whenx = xo continues to be a singularity
-
of ( x xo ) p (x) or ( x - xo )' q (x) it is irregular. These definitions hold for finite values of
x, (The analysis for x -. m leads to hypergeometric series.) The following example
illustrates these concepts.

Example 1
Locate and classify the singular points of the equation ,
I

where n is a positive integer. This equation arises in many physical problems in electrostatics
and quantum mechanics particularly in boundary value problems modelled in spherical
polar coordinates, It is known as Legendre's equation.
Solution
On comparing the given equation with Eq. (3.2), you will note that

P(X) e
,
--1 - x 2 and q (x) = n ( ; t + l )
1-3
 Are these functions analytic atx = 2 1 ? YOUwill readily recognise thatx = + 1 and Second Order Ordinary
DifFerential Qustions w i l h
x = - 1 are singular points of the given equation. To determine whether the singularities Variable CoeLficients
are regular or not, let us first compute
lim (x-x0)p(x) a n d ~ $ $ x - x ~ ) ~ ~ ( x )
x-rxo

, atx = 1 . The result is

and

lim ( x - 1 )'q(x) = ( x - i ) ' n ( n + l )

x-rl 1 -x2
(ii)

From (i) and (ii), you can say thatx 3 +1 is a regular singularity. What aboutx = -1 7 It is
also a regular singularity. However, you should convince yourself before proceeding further.

I
I SAQl Spend
10 min
To ensure that you have understood these concepts, we would like you to determine which
of the following equations have regular singular point(s).
a) x2y" + 3xy1+y' = 0
b) ~ ~ y " - ~ ' + 2=x0~ y

Now you are familiar with the basic terminology. ~ e t ' u stherefore,
, learn first to solve
homogeneous ODESof secoiid order with variable coefficients by the so-called power
s'eries method. It gives solutions in the form of an infinite power series.

3.3 POWER SERIES METHOD

The basic idea of the power series method for solving ODESwith variable coefficients
about an ordinary point is very simple. We will illustrate the procedure by considering
specific examples. The steps used in this procedure are listed below.
The first step of the power series method,is to assume a solution of the-form

If x = 0 is an ordinary point, we obtain a power series in powers ofx only :

Next, we substitute Eq. (3.3) and its derivatives into the given differential equation. Then
we obtain the values of an's by equating the coefficients of each power ofx to zero. This
technique is illustrated in the following example.

Example 2
\
Solve the differential equation y" + x2y = 0 by assuming a power series solution about the
pointx = 0, ,
 Mathematical Mabods in
-
Physics 11
Solution
On comparing the given equation with Eq. (3.2), you will note tbatp (x) = 0 and q (x) = x 2
so that x = 0 is an ordinary point. So we assume that

and differentiate it with respect to x. This gives

In the summation notation, we can write

The lower limit on the summation index has been shifted from n = 0 to n = 1. This is
because the term corresponding ton a 0 in (i) is constant and its derivative with respect to
xis zero. Similarly, you can write

(ii)

Substituting fix y and y" in the given differential equation, we find that
( 2 a 2 + 6 a 3 x + 1 2 a 4 x 2 + 2 0 a 5 x 3 + ...) +(sox 1 + a l x 3 t a z x 4 + ...) = 0

Next, let us collect like powers of x. This gives

Since the RHS is identically zero, we equate the coefficient of every power ofx on the LHS
Here 'coeff' is being used as an to Zero.
abbrevation of 'coefficient'.
, , x0
~ o e f fof : 2a2 = 0 +a2 = 0
!
Coeff. of a" : 6ag=O+a3=0
I
a0
I Coeff. of 2 : 12a4+ a. = O + a 4 = - ~

Coeff. of x3 : 20 a5+ al = 0 * as - -
a1
20
5

Coeff. of x 4 : 30 a6 + a2 = 0 * a6 = 0
Coeff. of x 5 ; 42 a, + a3 = 0 4 a7 - 0

. xn
~ o e f fof + : ( n + 3 ) ( n + 4 ) ~ , + ~ + a =, 0

or an+4 = - ( n + 3 )1( n t 4 ) a, for ,,n2 0

-
With a2 0, it follows that alternative even coefficients beyond a4 ( a6 , a10 , ) will be zero.
Similarly,since as = 0, it readily follows that a7 = all = ... = 0. Hence, you can write
y(x)= a o + a l x + a 4 x 4 +a5x5 + a g x8 + a 9 x9 +....
\L
 Second Order Ordinary
Dmerentinl Equation4 with
Vmiable Coefficients

Are these solutions linearly independent?'you should'check it by computing their

Wronskian.

You may now ask : Can we not use the power series method to solve first or second order
ODES with constant coefficients? The answer is: Yes, we can use the above steps as
outlined for second order equations with variable coefficients. You can easily convince
yourself about this by solving the following SAQ.
. .
Spend
SAQ 2 15 min
Use power series method to solve the following equations.

As pointed out earlier, the equation given in Example 1 is of particular interest in physics
and nuclear engineering. It is known as Legendre's equation. Let us obtain its two linearly
independent solutions using the power series method.

Example 3
Obtain series solution of the Legendre's equation given :qExample 1

Solution
To solve ~ e ~ e n d r eequation
's. using the power series method, we first rewrite it as

From Example 1, you would recall that the functionsp (x) = - 2x and

4(x) P( rn
1- x
+ :
) have regular singularities at - 1- x
+I.However, they arc analytic at

x = 0 and we can, therefore, use power series method to solve Legendre's equation in the
ral!ge -1 < x < 1. Let us write

(ii)'

Substituting this and its derivatives

and

in Legendre's equation, we get

Where we have put m ( m + 1 ), = k,

 I

" I

Mathematid Mdhodsin
I ' -
Physics I1
You can rewrite it as

In the expanded form, you can write

(2a2+6ajx+12a4x2+...)-( 2azn2+6a3x3+12a4x4+...)
2
- 2 ( a l x + 2 a 2 x 2 + 3 a 3 x 3 + . . . ) + k ( n , , + a l x + a 2 x + ...) = 0

As before, let us collect the coefficients of each power of x. This gives

(2q+kao)+(6~-2al+kal)x+(12cb-2q-4az+kq)x2+ ... I0
Again the coefficient of each power ofx is zero. Thus,

In general, for the ~t..khpowerofx, you can write

You can readily verify by putting k I m(m + 1 ) that the numerator can be written as
-
( m n ) ( n + m + 1 ). Hence, this expression takes the form
(vii)

This equality enables us to determine each expansion coefficient in terms of the second one
preceding it, except for a. and a1 (which are arbitrary). Such a relatiqp between the
coefficients is called a recurrence relation or recursion Formula.
The recurrence relation (vii) implies that coefficients with even subscripts can be expressed
in terms of ao and those with odd subscripts in terms of a1 .That is,

and so on.
By i&rting these values for the coefficients in (ii), you can write ,

Y(4 .
I aoy~
(XI+a1 y2 ( 4 (viii)
where

and 1.
 are two solutions over the interval -1 < x < 1. You may now again ask: Are yl and y2 Second Order Onllnuy
Diffem6iPI Equations wi!h
linearly independent? To answer this question, we note that yi contains only even powers of VPriPbleCoellldents
x whereas y2 contains only odd powers of x. As a result, the ratio yilyz will not be constant,
implying that y l and y2 are linearly independent solutions. And (viii) is a general solution of
Legendre's equation.
-
Spend
SAQ 3 15 min
The equation
yl'-2ry'*2,17y =0
II plays a particularly important role in statistics. Its solutions are known as Hermite '

polynomials. We would like you to obtain the coefficients of the power series solution of
,this equation.

Before proceeding, let us summarise the steps you should follow to solve a second order
ODE for which x = 0 is an ordinary point.

power Series Method

Step1: Assume a power series solution of the form
0

yo11 =
a
2 anxn
n-0
Step2: Substitute the agsumed solution and its derivatives into the given differential
equation.
Step 3 : Equate the coefficients of each power ofx to zero for a homogeneous equation.
This results in a recursion relation, which helps us in determining successively
the coefficients occurring in the power series in terms of two 'arbitrary
constants.
Step 4 : , Write the explicit form of the series solution which satisfies the given
equation.

So far we have refrained from Mathematical rigour. It may, however, be pointed out here
that the mathematical justification of power series method is contained ili Fuchs' Theorem.
We state it without giving proof:
If x = x i is an ordinary poht of the equation
Y" + P ( 4 ~+' 9 =0
then there exists a unique function y (x) which is analytic and satisfies the given equation in
the neighbourhood pf xo as well as the initial conditions y (%) = a. and y' (%) rn al where
a. and a1are two arbitrary constants.
Several second order ODES with variable coefficients appearing in many important

-
physical problems have coefficients which are not'analytic functions. In particular,
these equations may have a regular singularity at x xb For such equations, power
series solution of the form given by Eq. (3.2) is not physically acceptable and we use an
extended power series which always provides at least one solution around a regular
singular point:
m

= an(x-q)".' (3.4)
n-0

Such a series is also called Frobenius series with index r. Here r may betany(real or
complex) number so,that a0 0. You will note that the series given in Eq. (3.4) reduces to a
P w e r series (Eq. (3.3)) if r is a non-negative integer.
 Mnthcmnllul biethod~h
-
Phyaica II 3.4 THE FROBENIUS' METNOD
We illustrate the use of Frobenius' method when x 0is a regular singular point of
Eq.(3.2):

--
i
Y" +p(x)yl + q(4y 0
Sincep (x) and q (x) are not analytic at x 0 butx 0 is ii regular singularity, let us
rewrite this equation in terms of b(x) = xp(r) and c(x) xZq(x),which will be analytic
atx 0. So, we multiply throughout by 2 to obtain

I Since b(x) and c (x) are analytic at x = 0,we can write

and

You will note that if it happens that bo co - - cl 0, then x 0 defines an ordinary

point rather than a regula~singular point.
Differentiating the series expansion given in Eq: (3.4) forxo - 0 term by tenn, you will get

and

Substituting for y (x), y' (x), y"(x), b (x) and c (x) in Eq43.5)' we get

As before, let us equate the sum of the coefficients of each power of x to zero. This yields a

-
system of equations involving the unknown coefficients an's. The coefficient ofx ', which is /
the lowest power ofx, is obtained fiom then 0 tenn. Equating this coefficient to zero,
you will obtain
Coefficient of lr : [ r ( r -1 ) + r bo + co ] a0 = 0
Since a0 d 0, this equality will be satisfied if
r 2 + ( b o - l ) r + c o= 0 (3s7)
Eq.(3.7) is called the Indlcial equation c o ~ e s ~ o n dtoi nthe
~ given differential equation.
Since the indicia1equation is quadratic, it'will have two roots. This means that there should
be two Frobenius seriessolutions. Will these solutions always be linearly independent? Not
necessarily. In fact, the roots of the'indicial equation give us some idea of the nature of
solutions of the ODE of interest. In practice, the desired solutions are obtained using the *
steps outlined for tbe power series method. However, before plunging into these details, we
would like you to go through the following example.
 SceolldOrdcrOnlhury
Example 4 ~emtklEq~~~Wrrfth
V*ble.Ce
Determine the rootsof the indicial equation around the origin for the differential equation

Solution
To be able to compare the given equation with Eq. (3.2), we dihde throughout by x '. This gives

You would readily recognise that this equation has a singularity at x = 0. Is it regular or
irregular? To discover this, compare (i) with Eq. (3.2). You will note thatp(x) 1 -
X

x2-
1
-
9
and q(x) = 7sothatxp(x) = 1 andx2q(x)= x2-(1/9).Inthelimitx-,O,
i J

-
xp (x) = 1 and ~ ~ ~= ( ( 01/9) ). That is, b(x) and c (x) are analytic at x = 0 Thus, .
x = 0 is a regular singular point.
We can, therefore, assume a solution of the form

Differentiating it with respect to x, we get

and ytt(x) -2 am(m'+r)(m+r-l)xm+r-2

On substituting these in the given equation, we get

To arrive at the indicial equation, we equate the coefficients of the lowest dower of x, i.e., x'
to zero. This gives

For a. 0, the indicial equation takes the form

r2- I a o
9
which has roots r = * 1.That is, the roots of the indicial equatian are distinct. Moreover,
3
they do not differ by ax! integer.

'
Let us now pause for a minute and ask: Will the roots always be distinct? Certainly not.
Moreover, even when they are distinct, they may differ by an iuteger. In fact, there are
possibilities. The indicial equation may have (i) distinct roots not differing by an Integer,
(ii) double roots, and (iii) distinct roots differing by an integer. To discover these, you
should solve the following SAQ.

SAQ 4
Spend
Determine-the roots of the indicial equations correspondiug to the following ODESabout 10 min
x = 0.
 (a) x(x-l)y"+(3x-l)y'+y =0
-
(b) ~ ~ ( ~ ~ - 1 ) ~ ~ ~ - ( ~ 0~ + 1 ) ~ ~ + ( ~ ~ + 1 ) ~

You now knaw that there are three possible cases. But, only one of these, viz. when the
roots of the indicial equation are distinct and do not differ by an integer is important from
the view-point of physics. So we shall consider it in detail. For repeated roots or for distinct ' .
roots differing by an integer, we shall just quote the results and you wiU not be tested for,
these (in TMA/CMA/term-end examination).Let us learn to obtain solutions for this case.
Case 1: Distinct Roots of the Indiclal Equatlon not Differing by an Integer
When the two roots (rl, rz) of the indicial equation are distinct and do not differ by
an integer, we expect two linearly independent Frobenius series iolutions of the form

and

As such, this is the simplest case. We have illustrated the procedure of obtaining the two
linearly independent soIutions using Frobenius' method in the following example.

Example 5
Determine the two Frobenius series soIutionsaround x - 0 for the ODE

k t US firSt rewrite the given ODE in the standard form on dividing by x2

Here p (x) = 0 and q (x) -

' x2+(5/36)
x2 "
.
You d l easily recognise that the pointx = 0 is a
x ) you may logically mnclude thatx = 0 is a
singularity. Further, since l i m ~ ~ ~=(5/36, i
x--0
regu!ar shqular point of the given differential equation.
By/assuming a solution of the form

+ a d substituting it in (I), you will obtain

, To obtain the indicial equation, we equate-thesum of the coehcients of x' to zero. This
yields
 is the required indicia1 equation. You can readily verify that the two distinct roots are 516 Second Order Ordinary
Dilferential Equatlona with
and 116. Moreover, these do not differ by an integer. Variable Coemcients
To determine an corresponding to r = 516, let us equate the sum of the coefficientsof
$+l,d+', tozero:

Since the bracketed term does not vanish for r = rl = 516, this equality will hold only if we
choose a1 = 0. Similarly

On substituting r = rl = 516, you will find that

.

l
am = - a, -2
m(m.5) (ii)

Since a1 = 0,this recurrence relation implies that all odd coefficients will vanish.
To evaluate even coefficients, let us introduce the change m - 2p. This means that

azp -- azp-2 = - -3 a*-2

4 ~ ( 3 ~ + 1 )

Hence, with

Similarly, you can write

a0
p ! 1 x 4 x 7 x l O .....( 3 p + 1 )
Hence, one of the solutions of the given equation is

-
Y I ( X ) aox
'
1 --x.3 2++,9
w [ ' 16 16 2 x 4 ~ 7
x*
~ 1 1 x 4 ...(
~ 73 p + 1 ) 1
I
I
This can be expressed in a compact form as
I
 Matbematid Methods in
Physlca I1-

For r = r2 = 1/6 also, all odd coefficientswill vanish and we leave it as an exercise for
you to verify that

where

Case 2 : Double Root of the Indicia1 Equation

When roots of the indicial equation are equal, we cannot obtain two linearly independent
solutions using the procedure outlined above. In fact, there can be only one Frobenius series
solution. To determine this solution, we first find r from Eq. (3.7):

quadratic equation ax2+ bx + c

-
is b2-4ac 0.Thismeans that the
-
The condition for equal roots of a
0
Using the condition for equal roots, you will find that the common root is - bO - k!%!.
term under the radical sign will
2 2
vanish. Hence, ft readily follows from Eq. (3.4) that one of the solutions will be of the form

where an's are unknown constants. I

The second linearly independent solutioii is I

where Ai is some other constant. 1

Case 3 : Roots of the Indicia1Equation Differing by an Integer I

-
When the roots (rl, rz) of the indicia1 equation are distinct but differ by an integer, we can
always determine the first Frobenius series solution as before. If rl ( = r ) and r2 ( r - p,
wherep is a positive integer) are the two roo&, then
,
I

yl (x) = xr' C
A-0
a,,xu

' The other linearly independent Frobenius solution is

' 00

Let us now summarise what you have studied in this unit.

 3.5 SUM Y DiffcrcntLI Equations d(h
Va~inbleCoeffldents

o A second order ODE with variable coefficients

Y" + P ( ~ ) Y+' q(x)y - 0
-
is said to have an ordinary point atx = xo ifp(x)and q (x)are analyticatx .q,. Otherwise,
the point is said .to be singular. The singularity is regular if lim ( x -xo )p(x) and

-
*4Xp

lim ( x -% )2q(x) are finite at x xo.

x4*0

a We can solve a second order ODE with variable coefficients around an ordinary point at
m

x = 0 by assuming a power series solution of the fonn y (x) = 1 a,, xn. The constants
n-0
an's are determined using the recursion relation.

o The solution around a regular singularity atx = 0is obtained using Frobenius method by
assuming a solution of the form

o The indicial equation is'obtained by equating the sum of the coefficients of the lowest
power ofx to zero.

o The roots of the indicial equation give us an idea of the nature of solutions of the ODE.
. The roots of the indicial.equation may be distinct, repeated or differ by an integer:

e For distinct roots, two linearly independent solutions are

. and
00

3.6 TERMINAL QUESTIONS

1, Like Legendre's equation, another ODE that arises in advanced studies in physics and
. applied mathematics is the Bessel's equation of order m:
' 2 y 1 1 + x y 1 + ( ~ 2 - m 2 ) X = .~
In Example 4, we solved Bessel's equation of order 1/3.Use Frobenius' method to
solve this equation.
,2. Solve the ODE

around the point x = 0.

3.7 SOLUTIC)NSAND ANSWERS

To compareiit'with the standard fonn ( y" +p(x)yl + q@)y = 0 ), you should divide it
by 2.This
M?themrtiul Mdbodf in
-
Physics 11

3 and q(x) =
s o that p (x) = - ,
1 Both functions diverge at x = 0. To determine
X x2
whether the singularity is regular or not, let u s compute!@o xp(x) and?^^
x2q(x).
These are respectively equal to 3 and 1. So we can say that x = 0 is a regular
singularity.

@> xZy" - y' + 2c2y = 0

Dividing throughout by x2,we get

which shows thatp(x) = - 1- is not analyticatx= 0. In fact, lim x p (x) = 1

x-0
-
x2 x
and the function diverges at x = 0. Sox * 0 is an irregular singularity of the given
equation.

(4 2x2y"-v'+(x-5)y =0 - I

Dividing by k2throughout, you will obtain

On comparing with d e standard form, you will recognise tHatp (x) = --

1 and

q(x) = xL 5 which diverges at x = 0.. However, lim xp (x)

-2r2 x-0
- -1 2
and
2x I

lim x2q(x) = - -SO

5 thatx =. 0 isa regularsingularity.
x-0 2 . .

Assume that

and differentiate it with respect to x. The result is

and

Substituting these in the given equation, we fmd that .

In the expanded form, you can write

...)+ d ( a a + o 1 i + a 2 ~ ? +...)
(2az+6a~x+12a4x2+ - 0
Collecting the coefficients of like powers of x, we find that
( 2 a z + d a o ) + ( 6 a ~ + o ~ a 1 ) x + ( 1 2 a 4 + d & ) x ~ +=. , 0
.
'1
 Equating the coefficient of each power of x to zero, we have SecoodOrder Olrdinvy
" DiEemtinl Equ~tlonswith
Varbblc CocMdents

That is, az, a4, ... can be expressed in tenns of ao and a3, as, ... etc., can be expressed
in terms of al.Hence

Collecting the coefficients of ao and al, we fwd that

You would recognise that the coefficient of a0 defines cos ofi whereas the
coefficient of a1 defines sin wo x. Hence, the general solution can be written as

From Unit 2, you may recall that cos wo x and sin o o x are linearly independent
.solutions.

@) The given differential equation has an ordinary point atx = .

0 So there exists a power
series solution of the form
m

Substituting the power series for y and y' in the given equation, we obtain

In the expanded form, we can write

Collecting the coefficients of each power of+ on LHS, we get

The coefficient a0 is arbitrary. To find the other coefficients, we proceed by equating

coefficieilts of like powers of x on both sides of this equation:

Coeff. ofxO : a1=0-

Coeff. ofx l : & + a D = -2 =? a2 = --a02+ 2

 Coeff.of 2'- : ' mn+a,,-2 E 0 s an -- 1
--an-2 for n a 4
n

Iteration of this formula for n 2 4 yields

ao + 2

a5 9 ---a3
5
-- 1
5x3

1 a0+2
-za4= --6 x 4 ~ 2
a7
1
-ia5 --7ao+2
xSx3

ao+2
" g a16 8 ~ 6 ~ 4 x 2

Thesolution can then be written as

3. youwould radily recognise that x = 0 is an ordinary point of the given equation So

we assume a solution of the form
I

Then

and .
 Substituting these in the given equation, we find that Secoqd Order Ordinmy
Diffcrmtial EquaUons 4th
Variable CoeMcienb

- In the expanded form,

( 2 a 2 + 6 a 3 x + 1 2 a 4 x 2 +...+ n ( n - l ) a , , ~ " - ~...)
+

' Collecting the coefficients of each power of x, we get

(2a2+2mao)+(6a3-2al+2mal)x

Next we equate the coefficient of each power ofx to zero. The result is

Coeff. of x' : 6a3 + ( 2m - 2 ) al = 0 +a3 = 2 ( 1 - m ) a1

3 x1

Let us Brst rewritelthisequation in the standard form by dividing throughout by

x ( x - 1):

You can readily identify that

Both functions diverge at x = 0 as well asx - 1. But lim xp(x)= I, l i r n ~ ~ ~0,( ~ ) =

*-rl
-
lim ( x 1 )p(x) -
2 and lim ( x 1 )'q(x)
x-rl
- x-ro
0 SO thatx
x-ro
0 andx - 1 are
regular singularities. Forx = 0, let us, therefore, assume a solutionof the form
m

and y"(x) -2 n-0

(n+r)(n+r-1 ) a , , ~ " + ~ ~

On substituting fory(x), y' (x) and yrl(x)in the given equation, we find that
 t j 2 (n+r)anxn+'-2
m m

(ntr)anxn+'-' + 5 u , , x " + ~- 0
n-0 U-0 n-0

YOU Will note that the lowest power of x is xr-" By equating the sum of its
coefficients to zero, we have

Since a,, 0, we must have r2 1 ~

Hen=, this indicia1 equation has a double root: r a 0.

@I
us divide t~oughoutby ( x 2 - 1 ) x" to put it in tie standard form ((Eel. (3.2)):

- 2
yff -yfx t 1 , t i t 1 y - 0
x ( x 2- 1 ) 2(x2-1) I

You rudily remgnise thatx = 0 is a regular singular p i n t of the @en ODE.Let

us, therefore, assume a solution of the form

and

Substituting these in the given equation, we fmd that .

performing the multiplication, we find that I

Combring the first, third and fifth series together, and second and last serks
together, we fmd that fi
[I
 SecmdOrder 0rxUnm-y
Diorermli.l Equations with
Variable CocBdenb

On simplification, we find that

The lowest power of x in this equation is 2 . By equating its power to zero, you will
obtain the required indicial equation:
(r+l)(r-1)=0
whose roots are rl = 1 and r2 = -1. You will readily recognise that these roots differ
by an integer.
Terminal Questions
1. The family of ODES

is known as Bessel's equations. The parameter m is real and non-negative. You would
readily note that x = 0 is a regular singular point of the equation. So we assume that

Substitute y(x) and its derivatives in the given equation. This yields

Changing the first summation so that the exponent on x is n+r and collecting other
series, we have
m m

[ ( n + r ) ( n + r - l ) + ( n + r ) - m 2 ] a n X " " + an-2xn+r

~ =0 (ii)

The smallest power of x is 2 ( n = 0 ). Equating the coefficient o f d to zero, we get

[r(r-l)+r-m2]a0 = 0
Since a. d 0,we have the indicial equation

which has roots rl m and r2 = -m.

0

Depending on the value of m, the solutions can differ vastly:

and

To find yl, let us write (ii) in the expanded form

x'[(r2-n2)ao+((r+ ~ ) ~ - r n ~ J a ~ x
. + ( ( 1 + 2 ) 2 - m 2 ) a 2 2 + . . +. ( ( n + r ) 2 - d l a n $ + ...I
, + 2 [ a o x2. + a l x 3 + a 2 x 4 + + . . . + a f l - 2 Y...I
+ =0
. ,
M.thrm.tkrl Mabods in
-
Phgslcs Jl
Equating the coefficient of each power of x to zero, we get
Coeff.ofY : ( 3 - m 2 ) a o .r 0

For r = m, we find that a1 = 0since the bracketed quantity does not vanish. Then
recursion relation implies that as = as = a7 = 0 .:.... That is, all odd subscripted
coefficients vanish. For even subscripted coefficients, we fmd that

a2 -- 00

( m t 212-m 2

Similarly,

In general, .. ' t

The solution corresponding to r2

m is not an integer.
-m is found by simply replacing m by -mprovided
-
m
2. Letting y = 2 a,, x", ws.fid ttut
a-0

and

From Table 3.1, you would recall that series expansion for ex is

qubstituting thegiven differential equation, we obtain .

1/
Rearranging the terms, you will obtain Fi

Since the right-hand side of the equality is identically zero, we equate each coefficient
of every poyer ofx on the left-hand side to zero. To evaluate an, the exponents in the
two series mustbe made the same by shifting the index. Here, we choose to replace n
with n -2 in the second series. Thus,
 3
n- 2
n(n-l)anx"-2+
' n- 2
Second Ordcr Ordiauy
DifPerenlW Equations with
Variable Coeffidents
; :
, We combine the two series to obtain ,

5
n-2
The coefficients o f f -2 for n r 2 must be equal to zero. Thus,
n(n-l)an+an-2- =O
(n- 2)!
Solving for an, we obtain the recursion formula
a, ---
1 1
n ! n(n-1)
an-2, n = 2 , 3 , 4 ...

iteration of this formula yields

1
a4 = - - 1
- 'a2 = -ao
41 ( 4 x 1 3 ) 41

and so on

Since y(x) - n-0

w
anx n, and no and a1 are arbitrary, we get

- ao~l(x)+a1~2(~)+~p
where yl and yz are, respectively, infinite series of even and odd terms, and

-
Since this differential equation is linear and of the second order, the form of the
solution is y yc + yp where y, consists of a linear combination of two linearly
1 independent functions and yp is a particular solution to the given nonhomogeneous
,
Mathematical Mahodn in
Physics 11 APPENDIX A: POWER SERIES

A power series about a point xo is an infinite series of the form

.. ...
where the numbers a. , a1 , (12, . , an, are called the coefficients of the power series. A
power series does not include terms with negative or fractional powers.
We say that the power series converges if

exists. The value of the limit is called the sum of the power series at the point x = xo, If Ife
limit does not exist, the power series is said to diverge. The interval of values of x for which
a power series converges is called the interval of convergence and is denoted as
-
Ix xo I 4 R, where xo is called the centre of the power series and R is called the radius of
convergence. If R = 0, the series converges only at xo ;if R = oo ;the series converges for
all values of x.
Within a common interval of convergence, two power series may be added term by term.
That is,

Further, within the interval of convergence, the power series represenls a function whose
derivative and integral may be found from term-by-term differentiation and integration.
That is, if

then

and
X

A power series expansion of f(x) around x = 0 is called a Maclaurin series. The

Maclaurin series off (x)is given by

f (x) -2
n-0
fl0)
n!
where f "(0) means the value of the nth derivative off at x = 0. Recall that f(0)
O!-1
- f and
 T 4 SO APPLIGATIONS OF
ODEs IN PNYSICS
Structure
4.1 Introduction
Objectives
4.2 Mathematical Modelling
4.3 ~ i r sOrder
t OD% in Physics
Applications in Newtonian Mechanics
Simple Electrical Circuits
4.4 Second Order ODEs in Physics
Rotational Mechanical Systems
Planetary Orbits
4.5 . ~ o u ~ l e d ' ~ i f f e r e n tEquations
ia1
Coupled Oscillators
Coupled Electrical Circuits
Charged Particle Motion in Electric and Magnetic Fields
4.6 Summary
4.7 Terminal Questions
4.8 Solutions and Answers

4.1 INTRODUCTION
p:
In Units 1 to 3 you have learnt various techniques of solving first and second order OD&.
R e d l that we have studied these techniques so that we are able to answer real-life questions
such s the following: Does the quantity of fuel burned by a rocket affect its velocity? How
long f
oes it take a polluted gulf to return to its natural state, once man-made pollution is
+, :+.:-

mpped? What is the response of an LCR circuit to an applied signal? You are now on the
verge of being able to answer such questions. ?his unit on.applications of ODEs will further
help you in this respect.
This unit is primarily concerned not with how to do mathematics but with how to use it.
Here we shall be applying the techniques you have learnt so far to solve a variety of real
problems involving differential equations. These problems have both their origin and
solutions outside mathematics. Doing the mathematics is only a part of the process called
mathematical modelling. Today, as scientists seek to further our understanding of nature,
the technique of representing our "real- worldn in mathematical terms has become an
invaluable tool. Indeed, theprocess of mimicking reality by using the language of
mathematics is known as mathematical madelling.
In this unit, you will first learn what is involved in the process of mathematical modelling,
especially with ODEs. As the unit progresses, you will realise that mathematical modelling .
with ODEs finds immense use in various areas of physics. We hope that having studied the
unit, you will be able to answer the kind of questions posed above and many others you are
likely to come across as you delve deeper in physics.
In this block you have studied various methods of solving first and second order ordinary
. differential equations and their applications in physics. In the next block, we shall discuss
, partial differential equations (PDEs). You will learn how to model physical systems with
PDEs and various methodsof salving them.

Objectives
After studying this unit you should be able to