UNIT 1 FIRST AND HIGHER ORDER

EQUATIONS
Structure Page No.
1.1 Introduction 7
Objectives
1.2 Initial and Boundary Value Problems 7
1.3 Existence and Uniqueness of Solutions for Initial Value Problems 10
1.4 Linear Differential Equations 20
Solutions of Homogeneous, Linear DEs with Constant Coefficients
Solutions of Non-Homogeneous, Linear DEs with Constant Coefficients
1.5 Summary 33
1.6 Solutions/Answers 35

1.1 INTRODUCTION
From your knowledge of differential equations which you must have studied at the
undergraduate level, you are familiar with various methods of solving first/second or
even higher order linear, ordinary differential equations (ODEs) with constant
coefficients. Also, for linear ODEs upto second order with variable coefficients you
know certain transformations of the dependent and independent variables, which
reduce the equation to a form, which is solvable by the known methods. However, in
all the cases, solutions to ODEs were obtained in terms of known elementary
functions. To this day, no method of getting a solution of general first order initial
dy
value problem (IVP) of the form = f ( x , y) , in closed form or, in terms of
dx
elementary functions exists. In this unit we shall give a method of finding an
approximate solution to this general first order IVP.
In Sec.1.2 we shall start with defining the initial and boundary value problems and
illustrate them through various examples giving geometrical interpretation of their
solutions. In Sec.1.3, we have discussed the Picard’s method of successive
approximation for solving a first order initial value problem (IVP), y ′ = f ( x , y)
subject to y( x 0 ) = y 0 . We have also stated and proved Picard’s theorem on existence
and uniqueness of solution of IVP here. We have shown, through examples, how the
nature of solutions of first order IVPs can be examined even without solving them. In
Sec 1.4, we shall do a quick recap of methods of solving linear differential equations
of any order with constant coefficients and illustrate them through a number of
examples.
Objectives
After studying this unit you should be able to
• identify initial value problems and boundary value problems;
• obtain an approximate solution of an IVP of the first order;
• describe the nature of solution of first order differential equation with reference to
the existence and uniqueness of solutions;
• solve linear differential equations of first/second/higher orders with constant
coefficients.

1.2 INITIAL AND BOUNDARY VALUE PROBLEMS

Consider the first order linear differential equation
y ′ = 2 x (1 + y 2 )
By the usual process of integration, it can be seen that the solution of the given
differential equations is
Ordinary Differential y( x ) = tan ( x 2 + c) ,
Equations
where c is an arbitrary constant. Since c is arbitrary, for each value of c , we get a
distinct solution of the given equation. Hence, a one-parameter family consisting of
an infinite number of solutions is obtained. Geometrically, this equation represents a
one-parameter family of curves, called integral curves of the given equation. Each
integral curve is the geometric representation of the corresponding solution of the
differential equation. In physical applications, we often require a specific solution out
of this family representing a particular physical phenomenon. Specifying a particular
solution is equivalent to picking out a particular integral curve from the one-parameter
family. It is usually convenient to do this by prescribing a point ( x 0 , y 0 ) through
which the integral curve must pass; that is, we seek a solution such that
y( x 0 ) = y 0 .
Such a condition is called an initial condition. A first order differential equation
together with an initial condition form an initial value problem. This terminology is
suggested by the fact that the independent variable often denotes time, the initial
condition defines the situation at some fixed instant, and the solution of the initial
value problem describes what happens later. In the case of the equation considered
above we may be interested in a solution which satisfies the requirement
y ( 0) = 0.
This condition means that we obtain that solution y( x ) which passes through the point
(0, 0) . Thus,
y(0) = tan (0 + c) = tan c.
and y(0) = 0 gives c = 0.
The desired solution is given by
y( x ) = tan x 2 , (0 < x < π / 2 ) .
Next consider the differential equation of order two, namely,
dy d2y
y′′ + 5y ′ + 6 y = 0. ( y′ = , y′′ = 2 ) .
dx dx
It can be verified that for the above equation
y ( x ) = c1 e −2 x + c 2 e −3x
is a solution, existing on − ∞ < x < ∞ . Here c1 and c 2 are arbitrary constants. For
each choice of c1 and c 2 , we get a distinct solution. Hence, c1 e −2 x + c 2 e −3 x is a two
parameter family of solutions. To find a specific solution of this family, we require
two conditions, the value of the solution y( x 0 ) and its derivative y′( x 0 ) at a point x0.
Thus to determine uniquely an integral curve of a second order equation it is necessary
to specify not only a point through which it passes, but also the slope of the curve at
that point. Suppose for the above second order equation the conditions are
y(0) = 0 and y ′(0) = 1
Then, it follows that
y ( 0) = c 1 + c 2 = 0
y ′(0) = −2 c1 − 3 c 2 = 1
These are simultaneous equations in c1 and c 2 and solving them we get
c1 = 1, c 2 = −1 . The specific solution is, therefore, given by
y( x ) = e −2 x − e −3 x , − ∞ < x < ∞.
We thus observe that to determine a particular solution of a first order equation, we
need one condition while in the case of a second order equation, we need two
conditions at the same value of x . Similarly, we conclude that for the nth order
equation of the form
P0 ( x ) y ( n ) ( x ) + P1 ( x ) y ( n −1) ( x ) + L + Pn −1 ( x ) y′ + Pn ( x ) y = G ( x )

8
where the functions P0 , P1 , K Pn and G are functions of x only on some interval First and Higher Order
Equations
x 1 < x < x 2 , we need n conditions
y(a ) = α 0 , y ′(a ) = α1 ,K , y ( n −1) (a ) = α n −1 ,
where, x 1 < a < x 2 and α 0 , α 1 K , α n −1 is any set of prescribed real constants. These
conditions are referred to as initial conditions.
A differential equation together with initial conditions is called an Initial Value
Problem (IVP).
You may observe that an IVP is determined by the differential equation and the initial
conditions. A change in either of them would mean a different IVP.
Next, consider a second order differential equation
y ′′ + y = 0
It can be verified that solutions of this equations is given by
y( x ) = c1 cos x + c 2 sin x , − ∞ < x < ∞.
where c1 and c 2 are arbitrary constants. Let us assume that the following pairs of
conditions are required to be satisfied by the solutions
(i) y(0) = 0, y (π / 2) = 0
(ii) y(0) = 0, y ′( π / 2) = 0
(iii) y(0) = 0, y ′(π / 2) = 1
Since the general solution involves two constants c1 and c 2 , two conditions are
required to determine them. It may be noted that in this case the two conditions in
π
each of the pair above are given at two different points, namely, x = 0 and x = . The
2
conditions given in (i), (ii) and (iii) are referred to as boundary conditions stated at
π
x = 0 and x = .
2
A differential equation together with boundary conditions is called Boundary-
Value Problem (BVP).
If you apply these pair of conditions to the general solution
y( x ) = c1 cos x + c 2 sin x , then you would see that corresponding solutions for the
three cases are given respectively, by
(i) y( x ) = 0 (ii) y( x ) = c 2 sin x (iii) No solution, existing on − ∞ < x < ∞.
You may note that in (i) above there is only one solution viz., trivial solution y = 0 .
In (ii), since c 2 in an arbitrary constant, we get an infinite number of solutions. In
(iii), y(0) = 0 gives c1 = 0 and y ′( π / 2) = 1 gives 1 = c 2 0 = 0, which is a
contradiction and hence no solution. A boundary-value problem may thus have
infinite solution, single solution or no solution at all.

Similarly, for an initial value problem (IVP) the following questions may come to your
mind
i) Does an initial value problem always have a solution?
ii) Can it have more than one solution?
iii) Is the solution valid for all x, or only for some restricted interval about the initial
point?
Such questions are answered by the existence and uniqueness theorem for the solution
of an IVP which we shall be discussing in the next section, but before that you may try
this exercise:

E1) Discuss the solution of the boundary value problem

y ′′ + λy = 0
with boundary conditions y(0) = 0 and y(π) = 0 .

9
Ordinary Differential Let us now discuss the existence and uniqueness of solutions of IVPs.
Equations
1.3 EXISTENCE AND UNIQUENESS OF SOLUTIONS
FOR INITIAL VALUE PROBLEMS
We know that only a few simple type of differential equations can be solved explicitly
in terms of known elementary functions. Some of these types are discussed in Blocks
1 and 2 of MTE-08. However, many differential equations fall outside this category
and nothing has been done so far to suggest a procedure that might work in such cases.
Let us examine the initial value problem (IVP).
dy
= f ( x , y), y( x 0 ) = y 0, (1)
dx
where f ( x , y) is an arbitrary function defined and continuous in some neighborhood
of the point ( x 0 , y 0 ) . Geometrically, we want to obtain a function y = y( x )
satisfying the differential equation y ′ = f ( x , y) in some neighborhood of x 0 and
whose graph passes through the point ( x 0 , y 0 ) (see. Fig.1).

y We know that elementary procedures will not work here. We will have to evolve a
process to construct a sequence of functions that converges to a limit function
(x 0 , y 0 ) satisfying the IVP. This process is provided by Picard’s method of successive
y = y( x ) approximations.
Picard’s methods for solving differential equations is quite different from any method
that you might have studied so far. This method consists in replacing the initial value
o problem (1) by the equivalent integral equation
x0 x x

Fig.1 ∫
y( x ) = y 0 + f [ t, y ( t ) ] dt (2)
x0

Eqn.(2) is called an integral equation because the unknown function occurs under the
integral sign.
It can be easily seen that Eqns.(1) and (2) are equivalent. For if y( x ) is a solution of
Eqn.(1), then y( x ) is continuous and so is the right hand side of
y ′( x ) = f [ x , y( x ) ] (3)
Now we integrate Eqn.(3) w.r.t. x, from x 0 to x and obtain
x

∫
y( x ) = C + f [ t , y( t ) ] dt.
x0

Using the initial condition y( x 0 ) = y 0 , we obtain

x0

∫
y( x 0 ) = C + f [ t , y( t ) ] dt = y 0
x0

implies C = y 0 . Thus,
x
y( x ) = y 0 + ∫ f [t, y(t) ] dt,
x0

which is our Eqn.(2). Note that t is a dummy variable here.

Thus any solution of Eqn.(1) is a continuous solution of Eqn.(2).
Conversely, if y( x ) is a continuous solution of Eqn.(2), then y( x 0 ) = y 0 because the
integral vanishes when x = x 0 . Also by differentiation of Eqn.(2) w.r.t. x we recover
the differential equation
y ′( x ) = f [ x , y( x ) ].
We can thus say that the integral Eqn.(2) and the IVP(1) are equivalent.
10
We now try to solve Eqn.(2) by a method of successive approximation i.e., we begin First and Higher Order
with a guess or approximation to the solution of Eqn.(2) and improve it step-by-step Equations
by applying a repeatable operation, which we hope will bring us as close as we please
to an exact solution. The primary advantage that Eqn.(2) has over Eqn.(1) is that the
integral equation provides a convenient mechanism for carrying out this process as we
see below.
The first or a rough approximation to a solution of Eqn.(2) is given by the constant
function y( x 0 ) = y 0 , which is simply a horizontal straight line through the point
( x 0 , y 0 ) . We put this approximation in the right hand side of Eqn.(2) to obtain a new
approximation y 1 ( x ) as follows:
x

∫
y1 ( x ) = y 0 + f [ t , y 0 ( t ) ] dt.
x0

We hope that this new function is a better approximation to the solution. We then use
y1 ( x ) to generate another and perhaps even better approximation y 2 ( x ) in the same
way and obtain
x

∫
y 2 ( x ) = y 0 + f [ t , y 1 ( t )] dt.
x0

In this manner we obtain the iterants i.e., a sequence of functions

y1 ( x ), y 2 ( x ), y 3 ( x ), K whose nth term is defined by the relation
x

∫
y n ( x ) = y 0 + f [ t , y n −1 ( t ) ] dt , n = 1, 2, 3K (4)
x0

The repetitive use of formula (4) is known as Picard’s method of successive

approximations/iterations.

We shall now show how this method works by means of an example.

Example 1: Apply Picard’s iteration method to the initial value problem
y ′ = x 2 y, y ( 0) = 1
and compare with the exact solution, if possible.
Solution : For an initial value problem
y ′ = x 2 y, y(0) = 1,
the corresponding integral equation is
x

∫t
2
y( x ) = 1 + y( t ) dt
x0

Picard’s iterations are

y0 (x ) = 1
x
x3
∫
y1 ( x ) = 1 + t 2 1dt = 1 +
3
0
2

x
t3  x3 x6 x3 1  x3 
∫
y 2 ( x ) = 1 + t 1 +  dt = 1 + +  
2
+ =1+
 3  3 18 3 2!  3 
0
x
 t3 t6  x3 x6 1 x9
∫
y 3 ( x ) = 1 + t 2 1 + +  dt = 1 +
 3 18 
+ + .
3 18 18 9
0
2 3
x3 1  x3 
 1  x3
=1+ +   + 


3 2!  3 
 3!  3
KKKKKKKKKKKKKKKKKKKKKK
KKKKKKKKKKKKKKKKKKKKKK
11
Ordinary Differential 2 3 n
x3 1  x3  1  x3  1  x3 
Equations y n (x) = 1 + +   +   + L +  
3 2!  3  3!  3  n!  3 
The exact solution of the given problem can easily be obtained as
 x3 
 
 3 
 
y=e ,
to which the above approximate solution converges when n → ∞ .
***
We would like to remark here that you should not be deceived by the relative ease
with which the iterants y n ( x ) were obtained in Example-1. In general, it is seen that
the integration involved in generating each y n ( x ) can become complicated very
quickly. Even if it is possible to generate y n ( x ) , it may not always converge to an
explicit function or the exact solution. Let us consider the following example.
Example 2: Apply Picard’s method of successive approximations to the initial value
problem.
y ′ = x ( y − x 2 + 2) , y ( 0) = 1 .
Solution : The integral equation, equivalent to the given initial value problem is
x

∫
y( x ) = 1 + t ( y( t ) − t 2 + 2) dt.
0
The approximate solutions are
y0 (x ) = 1
x
3 2 1 4
∫
y1 ( x ) = 1 + t (3 − t 2 ) dt = 1 +
2
x − x
4
0
x
t2 t4 3x 2 x 4 x 6
y 2 ( x ) = 1 + t (3 + ∫ 2
− ) dt = 1 +
4 2
+
8
−
24
0
x
t2 t4 t6
y 3 ( x ) = 1 + t (3 + ∫ 2
+
8
−
24
) dt
0
3x 2 x 4 x 6 x 8
= 1+ + + −
2 8 48 192
and so on.
The exact solution of the given problem can be easily obtained as
x2
2
y=x + e2
x2 x4 x6 x8
= x2 +1+ + + + +L
2 8 48 192
The approximate solution in this case differs from the exact solution.
***
After going through Examples 1 and 2, you might ask the question:
Is Picard’s method a practical means of solving a first order equation y ′ = f ( x , y)
subject to y( x 0 ) = y 0 ? In most cases the answer is no. The question then arises, what
is it good for? Picard’s methods of iteration is a theoretical tool used in the
consideration of existences and uniqueness of solutions of differential equations.
Under certain conditions on f ( x , y) it can be shown that as n → ∞ , the sequence
{y n ( x )} defined by Eqn.(4) converges to a function y( x ) that satisfies the integral
Eqn.(2) and hence the IVP (1) and the solution obtained is unique. The theorem that
makes precise assertions of this kind is called Existence and Uniqueness Theorem
which we shall discuss now but before that you may try the following exercise.
12
E2) Apply Picard’s iteration method to the following initial value problems and First and Higher Order
Equations
compare the results obtained with the exact solutions wherever possible.
Perform at least three iterations.
1
(a) y ′ = x + y, y(0) = 1 (b) y ′ = x − y 2 , y(0) =
2
(c) y ′ = 2x (1 + y), y(0) = 0 (d) y ′ = y, y(0) = 1
(e) y ′ = y 2 , y ( 0) = 1.

We shall now state and prove Picard’s theorem on existence and uniqueness of
solution of IVP.
∂f
Theorem 1: Let f ( x , y) and be continuous functions of x and y on a closed
∂y
rectangle R with sides parallel to the axes (see. Fig.2). If ( x 0 , y 0 ) is an interior point
of R , then there exists a number h > 0 with the property that the initial value problem
y ′ = f ( x , y), y( x 0 ) = y 0 (5)
has one and only one solution y = y( x ) on the interval x − x ≤ h.

0
y

R
(x 0 , y 0 )
y = y( x )
R1

x
o x0 − h x0 + h
Fig.2

Proof: We know that every solution of IVP (1) is also a continuous solution of the
integral equation
x

∫
y( x ) = y 0 + f [ t , y( t ) ] dt (6)
x0

and conversely. We can thus say that Eqn.(5) has a unique solution on an interval
x − x 0 ≤ h , if and only if, Eqn.(6) has a unique continuous solution on the same
interval.
By Picard’s method of successive approximations, the sequence of functions y n ( x )
defined by
y 0 (x ) = y 0
x

∫
y1 ( x ) = y 0 + f [ t , y 0 ( t ) ] dt
x0
x

∫
y 2 ( x ) = y 0 + f [ t , y1 ( t ) ] dt
x0
................................................
x

∫
y n ( x ) = y 0 + f [ t , y n −1 ( t ) ] dt. (7)
x0

13
Ordinary Differential may converge to a solution of Eqn.(6) . Here you may observe that y n ( x ) is the nth
Equations
partial sum of the series of functions
∞
y 0 (x) + ∑ [y
n =1
n (x ) − y n −1 ( x )] = y 0 ( x ) + [y1 ( x ) − y 0 ( x )] + [y 2 ( x ) − y1 ( x )]

+ L + [y n ( x ) − y n −1 ( x )] + L (8)
and thus the convergence of the sequence (7) is equivalent to the convergence of series
(8). In order to prove Theorem 1, we are now required to produce a number h > 0
defined on the interval x − x 0 ≤ h , and show that on this interval, the following
statements are true
(i) the series (8) converges to a function y(x)
(ii) y(x) is a continuous solution of Eqn.(6)
(iii) y(x) is the only continuous solution of Eqn.(6).

We shall now prove the above three statements one by one but first of all let us try to
produce a positive number h.
∂f
From the hypothesis of Theorem 1 we know that f ( x , y) and are continuous
∂y
functions on the rectangle R . Now since R is closed (as it includes its boundary) and
bounded, each of these functions is necessarily bounded on R . Thus there exist
constants M and N such that
f ( x , y) ≤ M (9)
and
∂
| f ( x , y) | ≤ N (10)
∂y
for all points ( x , y) is R . Now if ( x , y1 ) and ( x , y 2 ) are two distinct points in R
with the same x-coordinate, then the mean value theorem guarantees that
∂
f (x, y1 ) − f (x, y 2 ) = f ( x , y ∗ ) y1 − y 2 (11)
∂y
∗
for some number y between y1 and y 2 . It is clear from Eqns.(10) and (11) that
f ( x , y1 ) − f ( x , y 2 ) ≤ N y 1 − y 2 (12)

for any points ( x , y1 ) and ( x , y 2 ) in R lying on the same vertical line. Let us choose
h to be any positive number such that
N h <1 (13)
and the rectangle R 1 defined by the inequalities | x − x 0 | ≤ h and | y − y 0 | ≤ Mh is
contained in R . As ( x 0 , y 0 ) is an interior point of R , we can say that such an h
exists.
Now after choosing h we shall now prove statements (i) – (iii) above. Here onwards,
we shall confine our attention to the interval | x − x 0 | ≤ h .
In order to prove (i), it is sufficient to show that the series
| y 0 ( x ) | + | y1 ( x ) − y 0 ( x ) | + | y 2 ( x ) − y1 ( x ) | + L + | y n ( x ) − y n −1 ( x ) | + L (14)
converges and to achieve this, we estimate the terms | y n ( x ) − y n −1 ( x ) | . First we
show that the graph of each of the functions y n ( x ) lies in R 1 and hence in R .
Obviously, y 0 ( x ) = y 0 , and so the point [ t , y 0 ( t ) ] lies in R 1 . Eqn.(9) gives
| f ( t , y 0 ( t ) ) | ≤ M, and
x
y1 ( x ) − y 0 = ∫ f [t, y 0 ( t ) ] dt ≤ Mh ,
x0

14
which proves that the point y1 ( x ) lies in R 1 . It follows consequently from this First and Higher Order
Equations
inequality, that the point [ t , y1 ( t ) ] is in R 1 , so that f [ t , y 1 ( t ) ] ≤ M and
x

∫
| y 2 ( x ) − y 0 | = | f [ t , y1 ( t ) ] dt | ≤ Mh.
x0

Similarly, we can say that

x

∫
| y 3 ( x ) − y 0 | = | f [ t, y 2 ( t ) ] dt | ≤ Mh.
x0

and so on. Now we have estimated the terms | y n ( x ) − y n −1 ( x ) | .

We know that a continuous function has a maximum on a closed interval. Since
y1 ( x ) is continuous, we can define a constant ' α' by α = max | y1 ( x ) − y 0 | and write
| y 1 ( x ) − y 0 | ≤ α.
Next, the point [ t , y1 ( t ) ] and [ t , y 0 ( t ) ] lie in R 1 and so relation (12) yields
f [ t , y1 ( t ) ] − f [ t , y 0 ( t ) ] ≤ N y1 ( t ) − y 0 ( t ) ≤ Nα,
and we have
x
| y 2 ( x ) − y1 ( x ) | = | ∫ { f [t, y (t ) ] − f [ t, y
x0
1 0 (t ) ] } dt | ≤ Nαh = α( Nh )
Similarly, we obtain
| f [ t , y 2 ( t ) ] − f [ t , y 1 ( t ) ] | ≤ N | y 2 ( t ) − y 1 ( t ) | ≤ N 2 αh
and
x
| y 3 (x) − y 2 (x) | = ∫ {f [ t, y
x0
2 (t ) ] − f [ t , y1 ( t ) ] }dt

≤ ( N 2 α h ) h = α( N h ) 2
Continuing this way, we find that
y n ( x ) − y n −1 ( x ) ≤ α ( Nh ) n −1 ,
for every n = 1, 2, K . You can thus observe that each term of the series (14) is less
than or equal to the corresponding term of the series of constants
y 0 + α + α ( Nh ) + α ( Nh ) 2 + L + α ( Nh ) n −1 + L
But relation (13) guarantees that this series converges. So series (14) converges by the
comparison test and hence Eqn.(8) converges to a sum, which we denote by y( x ) , and
thus y n ( x ) → y( x ) . Also, since the graph of each y n ( x ) lies in R 1 , the graph of
y( x ) also has this property. This completes the proof of statement (i).

Next, we prove statement (ii). From the proof of (i) it is clear that not only y n ( x )
converges to y( x ) in the interval, but also that this convergence is uniform. This
means that by choosing n to be sufficiently large, we can make y n ( x ) as close as we
please to y( x ) for all x in the interval. This means that if we are given ∈> 0 , then
there exists a positive integer n 0 such that, for n ≥ n 0 , we have y( x ) − y n ( x ) <∈ ,
for all x in the interval. Since each y n ( x ) is continuous, this uniformity of the
convergence implies that the limit function y( x ) is also continuous. To prove that
y( x ) is actually a solution of Eqn.(6), we must show that
x

∫
y( x ) − y 0 − f [ t, y( t ) ] dt = 0, (15)
x0

From relations (7) we know that

15
Ordinary Differential x
Equations y n (x) − y 0 − ∫ f [t, y n −1 ( t ) ] dt = 0, (16)
x0

Subtracting the left side of Eqn.(16) from the left side of Eqn.(15) we get
x
y( x ) − y n ( x ) + ∫ { f [t, y n −1 ( t ) ] − f [ t , y( t ) ] }dt = 0,
x0

We can thus write

x x

∫
y( x ) − y 0 − f [ t , y( t ) ] dt = y( x ) − y n ( x ) + ∫ {f [t, y n −1 ( t ) ] − f [ t , y( t ) ] } dt.
x0 x0

Taking modulus of both sides, we obtain

x x
y( x ) − y 0 − ∫ f ( t , ( t ) ) dt ≤ y( x ) − y n ( x ) + ∫ {f (t, y n −1 ( t ) ) − f ( t , y( t )) }dt
x0 x0

Since the graph of y( x ) lies in R 1 and hence in R , using relations (12) we can write
x

∫
| y( x ) − y 0 − f ( t , y( t ))dt | ≤ | y( x ) − y n ( x ) | + Nh. max | y n −1 ( x ) − y( x ) |
x0
(17)

The uniform convergence of y n ( x ) to y( x ) implies that the right side of inequality

(17) can be made as small as we please by taking n large enough. The left side of
inequality (17) must, therefore, be equal to zero and this completes the proof of
statement (ii).
To prove statement (iii), let us assume that y( x ) is also a continuous solution of
Eqn.(6) on the interval | x − x 0 | ≤ h. We shall then be through, if, we show that
y( x ) = y( x ) for every x in the interval. As a first step we shall try to show that y( x )
lies in R 1 and hence in R . If possible, let us suppose that the graph of y( x ) leaves
R 1 see Fig. 3. y
y( x )
R1

Mh
y0

o x0 − h x0 x1 x 0 + h x
Fig.3

Now since y( x ) , is continuous and the fact that y( x 0 ) = y 0 (being a solution) implies
that there exists an x 1 such that | x 1 − x 0 | < h , | y( x 1 ) − y 0 | = Mh and
| y( x ) − y 0 | < Mh if | x − x 0 | < | x 1 − x 0 | . For such an x1, it follows that
| y( x 1 ) − y 0 | Mh Mh
= > = M.
| x1 − x 0 | | x1 − x 0 | h

Thus by the mean value theorem, there exists a number x ∗ between x 0 and x 1 , such
that
| y( x 1 ) − y 0 |
= | y′ ( x * ) |
| x1 − x 0 |
16
But, | y( x * ) | = f [ x * , y( x * ) ] ≤ M, since [x * , y( x * ) ] lies in R 1 . First and Higher Order
Equations
| y( x 1 ) − y 0 |
∴ ≤M
| x1 − x 0 |
which is a contradiction. This shows that no such point x 1 exists and the graph of
y( x ) lies in R 1 . Finally, to show that the two solutions y( x ) and y( x ) of Eqn.(6) are
equal we write
x
| y ( x ) − y( x ) | = ∫ {f [t, y( t ) ] − f [t, y(t ) ] }dt
x0

Since the graphs of y( x ) and y( x ) both lie in R 1 , relation (12) yields

| y( x ) − y( x ) | ≤ Nh max y( x ) − y( x ) .
∴ max | y( x ) − y( x ) | ≤ Nh max | y( x ) − y( x ) |
This implies that max. | y( x ) − y( x ) | = 0, for otherwise we would have 1 ≤ Nh in
contradiction to relation (13). This proves that y( x ) = y( x ) for every x in the interval
| x − x 0 | ≤ h , and Picard’s Theorem is completely proved.
We will now make the following remarks.
Remark 1: Picard’s theorem can be strengthened in various ways by weakening its
∂f
hypotheses. For example, our assumption that is continuous on R is stronger than
∂y
the proof requires, and is used only to obtain the inequality (12). We can, therefore,
introduce this inequality (12) into the theorem as an assumption which replaces the
∂f
assumption of continuity of on R . This would lead us to a stronger form of the
∂y
theorem as in practice, there are number of functions that lack a continuous partial
derivative but satisfy relation (12) for some constant N . This inequality viz;
| f ( x , y 1 ) − f ( x , y 2 ) | ≤ N | y 1 − y 2 | , which says that the difference quotient
f ( x , y1 ) − f ( x , y 2 )
y1 − y 2
is bounded on R is called a Lipschitz condition in the variable y. The constant N is
called the Lipschitz constant.
Remark 2: If we drop the Lipschitz condition, and assume only that f ( x , y) is
continuous on R , then it is still possible to prove that the IVP (5) has a solution. This
result is known as Peano’s theorem named after Italian logician and mathematician
Guiseppe Peano (1858 – 1932). But with only this assumption the solution whose
existence this theorem guarantees may not necessarily be unique. Consider for example,
2
the problem y′ = 3y 3 , y(0) = 0 and let R be the rectangle x ≤ 1, y ≤ 1. Here
2
f ( x , y) =
3y 3is continuous on R . Also y1 ( x ) = x 3 and y 2 ( x ) = 0 are two different
solution of this problem valid for all x . The given problem therefore has a solution but it
is not unique. The reason being that f ( x , y) does not satisfy the Lipschitz condition on
the rectangle R , because the difference quotient
2
f (0, y) − f (0, 0) 3y 3 3
= = 1
y−0 y
y3
is unbounded in every neighborhood of the origin.
Remark 3: Theorem 1, is called a local existence and uniqueness theorem because it
guarantees the existence of a unique solution only on some interval x − x 0 ≤ h , where h

17
Ordinary Differential may be very small. However, there are several important cases in which this restriction
Equations can be removed. For example, consider the first order linear equation
y ′ + P( x ) y = Q( x ),
where P ( x ) and Q( x ) are defined and continuous on an interval a ≤ x ≤ b.
If we compare this equation with the IVP(5), then we have
f ( x, y) = −P( x ) y + Q( x ),
and if N = max P( x ) for a ≤ x ≤ b, it is clear that
f ( x , y1 ) − f ( x , y 2 ) = − P ( x ) ( y1 − y 2 ) ≤ N y1 − y 2
The function f ( x , y) is therefore continuous and satisfies a Lipshitz condition with
Lipshitz constant N = max P( x ) , on the infinite vertical strip defined by
a ≤ x ≤ b and − ∞ < y < ∞ . Under these circumstances, the IVP
y ′ + P( x ) y = Q( x ), ( x 0 ) = y 0
has a unique solution on the entire interval a ≤ x ≤ b.
We shall now take up a few examples to illustrate the existence and uniqueness
theorem.
Example 3: Show that f ( x , y) = x 2 y satisfies a Lipschitz condition on the
∂f
rectangle x ≤ 1 and y ≤ 1, but fails to exist at many points of this rectangle.
∂y
Solution: Here rectangle R is defined by R : x ≤ 1 and y ≤ 1.
Now,
| f ( x , y1 ) − f ( x , y 2 ) | = | x 2 | y1 | − x 2 | y 2 | |
= | x 2 | | | y1 | − | y 2 | |
≤ | x 2 | | y1 − y 2 |
≤ | y 1 − y 2 | in R ,
Hence f ( x , y) satisfies a Lipschitz condition, with Lipschitz constant as 1.
Let (a , 0) be any point in R for a ≠ 0.
f (a , 0 + h ) − f (a , 0) a2 h − a20 a2 h
Here, lim = lim = lim , which does not exist.
h →0 h h →0 h h →0 h
∂f
Hence, does not exist for many points (a , 0) in R for a ≠ 0.
∂y
***
2
Example 4: Examine the IVP y ′ = e − x y 2 sin x , for Lipschitz condition on the
region R where, R is the strip 0 ≤ y ≤ 2 in the xy -plane.
2
Solution: The function f ( x , y) = e − x y 2 sin x is continuous for all x and y .
Let ( x , y1 ) and ( x , y 2 ) be the two points of R , then
2
f ( x , y 2 ) − f ( x , y 1 ) = e − x sin x y 2 + y1 y 2 − y1

≤ 4 y 2 − y1
2
because e − x sin x ≤ 1 for all x and y 2 + y1 ≤ 4 , if y1 and y 2 are both in the
interval [ 0, 2 ] . Thus f ( x , y) satisfies the Lipschitz condition with N = 4 in the strip
R.
***
Example 5: Discuss the nature of solutions of the initial value problem
y′ = | y | , y(0) = 0 , on the rectangle | x | ≤ 1 and | y | ≤ 1.
18
Solution: The function f ( x , y) = | y | is continuous for all values of y . Taking First and Higher Order
Equations
y1 = 0 and y 2 > 0 , as two points, we get
| f ( x, y 2 ) − f ( x , y 1 ) | | y2 | 1
= =
| y 2 − y1 | | y2 | y2
1
Now can be made as large as we please by choosing the sufficiently small
y2
values of y2. Therefore, there exists no N , independent if x , y such that
| f ( x , y 2 ) − f ( x , y1 ) | < N | y 2 − y1 | .
⇒ Lipschitz condition is not satisfied in any region containing y = 0 .
x2
It may be observed that given IVP has solutions y( x ) = 0 and y( x ) = . Thus
4
continuity of f ( x , y) is not sufficient for the unique solution of the IVP.
***
Example 6: Discuss the nature of solution of the IVP y ′ = y 2 , y(0) = b > 0 on the
infinite strip R consisting of the points ( x , y) for which 0 ≤ x ≤ 1 and y is arbitrary.

Solution: Function f ( x , y) = y 2 is continuous everywhere. Taking ( x , y1 ) and

(x, y 2 ) as two points of the region R , we get
f ( x , y 2 ) − f ( x , y1 ) = y 22 − y12
= y 2 + y 1 y 2 − y1
Because y 2 + y1 can be made arbitrarily large, it follows that f ( x ,y) does not satisfy
Lipschitz condition on the infinite strip R . However, when we solve the given IVP
by separation of variable, we get
b
y( x ) =
1 − bx
1
Now because the denominator vanishes for x = , the above solution of IVP is only
b
1
valid for x < . In particular, if b is large, then we have a solution only on a very
b
small interval to the right of x = 0 .
***
You may now try some exercises.
E3) For IVP
 2y
 , x>0 y ( 0) = 0
y′ =  x
0 , x = 0.
show that the Lipschitz condition is not satisfied in any closed rectangle
containing (0, 0).
E4) Examine the existence and uniqueness of solution of IVP
y ′ = f ( x , y), y(0) = 1,
where, (a) f ( x , y ) = x
(b) f ( x , y) = − y .

E5) Show that f ( x , y) = xy 2

(a) satisfies a Lipschitz condition on any rectangle a ≤ x ≤ b and c ≤ y ≤ d.
(b) does not satisfy a Lipschitz condition on any strip a ≤ x ≤ b and
−∞< y<∞.
19
Ordinary Differential ( y − 1)
Equations E6) Show that y ′ = , y(0) = 1 has an infinity of solutions.
x
 4x 3 y
 when x & y are not both zero
E7) Examine IVP y ′ =  x 4 + y 2 , y ( 0) = 0
0, when x = y = 0

for uniqueness of solutions.
In the next section we shall quickly recall various properties of solutions of linear
differential equations. We shall also give here a quick recap of various methods of
finding solutions of these equations when the coefficients are constants. For details,
you can refer Block-2 of MTE-08. The case when the coefficients are variable will be
discussed in Unit 2.

1.4 LINEAR DIFFERENTIAL EQUATIONS

The most general linear, non-homogeneous differential equation of order n is of the
form
a 0 ( x ) y ( n ) + a 1 ( x ) y ( n −1) + L + a n −1 ( x ) y′ + a n ( x ) y = f ( x ) (18)
where a 0 ( x ) ≠ 0. Further, if the functions a 0 ( x ), a 1 ( x ),K , a n ( x ), f ( x ) are
continuous on an interval [a , b] , where the interval [a , b] can be of finite or infinite
length or, it can be open or closed at either end, then a solution y( x ) of Eqn.(18) over
[a , b] together with its derivatives y′( x ), y′′( x ),K , y ( n −1) ( x ) will be continuous on this
interval. From Eqn.(18) it follows that y ( n ) ( x ) will also be continuous on the [a , b]
(ref. Unit 5, MTE-08).
We shall now state a theorem (without proof) which gives the conditions under which
we can expect a solution of Eqn.(18).
Theorem 2: Let x = x 0 be a point of the interval [a , b] and let b 0 , k 1 ,K , k n −1 be
arbitrary set of n real constants. If the functions a 0 ( x ), a 1 ( x ), K , a n ( x ), f ( x ) are
continuous on [a , b] and a 0 ( x ) does not vanish at any point of the interval then there
exists one and only one solution y( x ) of Eqn.(18) with the property
y( x 0 ) = k 0 , y ′( x 0 ) = k 1 , K , y ( n −1) ( x 0 ) = k n −1 .
Further, this solution is defined over the entire interval [a , b] .
For example, let the interval [a , b] be 0 ≤ x ≤ 1 and that n = 3 . Then at the point
1
x = , we may prescribe y = 2 , y ′ = π 2 , y ′′ = 1010 . Then the theorem asserts the
3
1
existence of one and only one solution y( x ) taking on these values at x = . This
3
solution will, further, be defined at every point of 0 ≤ x ≤ 1 .
If f ( x ) ≡ 0 on [a , b] Eqn.(18) becomes
a 0 ( x ) y ( n ) + a 1 ( x ) y ( n −1) + L + a n −1 ( x ) y′ + a n ( x ) y = 0 (19)
Eqn.(19) is called a homogeneous equation. Before talking about the properties of the
solutions of Eqn.(18) and (19) let us quickly recap linear dependence and
independence of set of functions on a given interval.
Consider a set of n functions, namely y1 ( x ), y 2 ( x ), K , y n ( x ) of x defined over the
interval [a , b] . These n functions are said to be linearly dependent in an interval
[a , b] if for every x in the interval, there exists a relation
n
c1 y 1 ( x ) + c 2 y 2 ( x ) + L + c n y n ( x ) = ∑ c y (x) = 0,
i =1
i i (20)

20
where the constants c i ' s are not all zero. First and Higher Order
Equations
If such a relation does not exist, then the functions are said to be linearly independent
in [a , b] , i.e., none of the functions can be expressed as a linear combination of the
others.
For example, the functions cosh x , e x and e − x are linearly dependent, since
1 1 1
cosh x = (e x + e − x ) or , cosh x − e x − e − x = 0 . Similarly, function
2 2 2
y1 ( x ) = sin 2 x and y 2 ( x ) = sin x cos x are linearly dependent on the interval
] ∞, ∞ [ since c1 sin 2x + c 2 sin x cos x = 0 is satisfied for every real x with
1
c1 = and c 2 = −1 .
2
Note that in the consideration of linear dependence or linear independence, the
interval on which the functions are defined is important. You can check this for
yourself that the functions y1 ( x ) = x and y 2 ( x ) = | x | are linearly independent on the
interval ] − ∞, ∞ [ whereas, they are linearly dependent on ]0, ∞ [.
The above procedure of examining the linear dependence or independence of a set of
functions appears to be quite involved. We, therefore, give below a sufficient
condition for examining the linear independence of a set of n functions.
Suppose that y1 ( x ), y 2 ( x ), K , y n ( x ) are n functions defined on an interval [a , b]
possessing derivatives upto (n − 1) th order. If the determinate
y1 y2 K yn
y1′ y ′2 K y ′n
W ( y1 ( x ), y 2 ( x ), K , y n ( x ) = KKKKKKKKK
KKKKKKKKK
y1( n −1) y (2n −1) K y (nn −1)

is not zero for at least one point in the interval [a , b] , then the functions
y1 ( x ), K , y n ( x ) are linearly independent on the interval. The determinant
W ( y1 ( x ), y 2 ( x ), K , y n ( x )) is called the Wronskian of the functions. It is named
after a Polish mathematician Josef Maria Hoene Wronski (1778-1853).
The functions y1 ( x ) = e kx , y 2 ( x ) = e − kx and y 3 ( x ) = sinh kx , for instance are linearly
dependent on ] − ∞, ∞ [ because
e kx e − kx sinh kx
kx − kx
W ( y1 , y 2 , y 3 ) = ke − ke k cosh kx = 0 .
2 kx 2 − kx
k e k e k 2 sinh x
Remember that in the above condition the non-vanishing of the Wronskian at a point
in the interval provides only a sufficient condition. In other words, if
W ( y1 , y 2 , K , y n ) = 0 for some x in the interval [a , b] then it does not necessarily
mean that the functions are linearly dependent on the interval. For example, if
y1 ( x ) = x 2 and y 2 ( x ) = x | x | then W ( y1 ( x ), y 2 ( x )) = 0 for every real number
whereas, y1 ( x ) and y 2 ( x ) are linearly independent on ] − ∞, ∞ [ .
With the above background let us now come back to the properties of solutions of
Eqns.(18) and (19).
We can think of the forms of the solutions of Eqn.(19) by making use of the following
elementary theorems.
Theorem 3: If y = y1 is a solution of Eqn.(19) on an interval [a , b] , then y = c y1 is
also its solution on [a , b] , where c is any arbitrary constant.
21
Ordinary Differential Theorem 4: If y = y1 , y 2 , K , y n are n solutions of homogeneous differential
Equations
Eqn.(19) on an interval [a , b] , then y = c1 y1 + c 2 y 2 + L + c n y n is also a solution of
Eqn.(19) on [a , b] , where c1 , c 2 , K , c n are arbitrary constants.
Theorem 4 is known as the superpostion principle. Theorems 3 and 4 represent
properties that non-linear differential equations, in general, do not possess. This will
become more clear to you after doing E8).
Let us now consider the following definition which involves a linear combination of
solutions.
Definition: Let y1 , y 2 , K , y n be n linearly independent solutions of homogenenous
linear differential Eqn.(19) of order n on an interval [a , b] . Then
y = c1 y 1 ( x ) + c 2 y 2 ( x ) + L + c n y n ( x ) ,
where c i , i = 1, 2, K , n are arbitrary constants, is defined to be the general solution of
Eqn.(19) on [a , b] .
The above definition generates our interest in knowing when n solutions
y1 , y 2 , K , y n of the homogeneous differential Eqn.(19) are linearly independent.
Surprisingly, the non-vanishing of the Wronskian of a set of n such solutions on an
interval [a , b] is both necessary and sufficient for linear independence. That is, if
y1 , y 2 , K , y n be n solutions of homogeneous linear n rh order differential Eqn.(19)
on [a , b] , then the set of solutions is linearly independent on [a , b ] , if and only if
W ( y1 , y 2 , K , y n ) ≠ 0
For every x in the interval. Such a set y1 , y 2 , K , y n of n linearly independent
solutions of Eqn.(19) on [a , b] is said to be a fundamental set of solutions on the
d2y
interval. For instance, the second order equation − k 2 y = 0 , has two solutions
dx 2
y1 = e kx and y 2 = e − kx since

e kx e − kx
W (e kx , e − kx ) = = −2k ≠ 0
ke kx − ke − kx
for every value of x . Functions y1 and y 2 form a fundamental set of solution on
] − ∞, ∞ [ . The general solution of the differential equation on the interval is
y = c1 e kx + c 2 e − kx .
We now state one more theorem wich pertains to the solution of non-homogeneous
linear differential Eqn.(18).
Theorem 5: If y = Y0 ( x ) is any solution of differential Eqn.(18) on an interval [a , b]
and if y = Y ( x ) is the general solution of the corresponding homogeneous Eqn.(19)
on the interval, then
y = Y0 ( x ) + Y ( x )
is the general solution of Eqn.(18) on the given interval.
We shall not prove Theorems 3, 4 and 5 here but illustrate them through various
examples. If you are interested in the proofs of these theorems then you can refer to
Unit 5, Block 2 of MTE-08.
Let us consider the following examples.
Example 7: Show that if y1 = x 2 and y 2 = x 2 ln x are both solutions of the
equation x 3 y ′′′ − 2 xy ′ + 4 y = 0 , on the interval ] 0, ∞ [ . Then y = c1 x 2 + c 2 x 2 ln x
is also a solution of the equation on the given interval.
22
Solution: We have y = c1 x 2 + c 2 x 2 ln x First and Higher Order
Equations
∴ y ′ = 2c1 x + 2c 2 x ln x + c 2 x ,
2c
y ′′ = 2c1 + 2c 2 ln x + 3c 2 , y ′′′ = 2 .
x
 2c 
∴ x 3 y ′′′ − 2 xy ′ + 4 y = x 3  2  − 2 x (2c1 x + 2c 2 x ln x + c 2 x ) + 4c1 x 2 + 4c 2 x 2 ln x
 x 
=0
Thus y = c1 x + c 2 x 2 ln x is also a solution of the equation.
2

***
Example 8: Show that linearly independent solutions of
y′′ − 2 y ′ + 2 y = 0
on any interval are e x sin x and e x cos x . What is the general solution? Find the
solution y( x ) with the property y(0) = 2, y ′(0) = −3.

Solution : It can be verified that y1 = e x sin x and y 2 = e x cos x satisfy the given
equation and hence, are solutions of given equation.
Here
y1 y 2 e x sin x e x cos x
W ( y1 , y 2 ) = = x = − e x ≠ 0.
′ ′
y1 y 2 x
e cos x + e sin x x x
e cos x − e sin x
Hence e x sin x and e x cos x are linearly independent solutions of the given equation.
The general solutions can then be written as
y( x ) = c1 e x sin x + c 2 e x cos x.
where, c1 and c 2 are arbitrary constants.
Here ( ) (
y ′( x ) = c1 e x sin x + e x cos x + c 2 e x cos x − e x sin x )
Now y(0) = 2 and y ′(0) = −3 imply
2 = c 2 and − 3 = c1 + c 2
i.e, c1 = −5, c 2 = 2
Hence, in this case the solution is y( x ) = −5 e x sin x + 2 e x cos x.
***
You may now try the following exercises.

E8) Functions y1 = 1 and y 2 = ln x are solutions of the non-linear differential

equation y ′′ + ( y ′) 2 = 0 , on the interval ] 0, ∞ [ . Then
a) is y1 + y 2 a solution of the equation?
b) is c1 y1 + c 2 y 2 , a solution of the equation, where c1 and c 2 are arbitrary
constants?
E9) Show that sin ax and cos ax are linearly independent solutions of y′′ + a 2 y = 0 ,
a being a positive constant on the interval ] − ∞, ∞ [ . Obtain the general solution
of the equation on the interval.
1 1
E10) Show that on the interval 0 < x < ∞, sin and cos are linearly independent
x x
4 3
solutions of the equation x y ′′ + 2x y ′ + y = 0. Find the solution y( x ) of the
1 1
differential equation with the property y  = 1 and y ′  = −1.
 
π  π

In practice, equations of the form (18) and (19), where coefficients are functions of x ,
do not usually have solutions expressible in terms of elementary functions, and even
23
Ordinary Differential when they do, it is very difficult to find them. However, if coefficients in Eqns.(18)
Equations and (19) are constants then they are called linear equations with constant
coefficients, and their solutions in terms of elementary functions can be obtained. In
the next section we shall quickly recall some of the methods of finding these solutions.
1.4.1 Solutions of Homogeneous, Linear Differential Equations with
Constant Coefficients
Consider the differential equation
dn y d n −1 y dy
a 0 n + a 1 n −1 + L + a n −1 + any = 0 (21)
dx dx dx
where a 0 , a 1 ,K , a n are all constants and a 0 ≠ 0.
Suppose that a possible solution of Eqn.(21) is
y = e mx (22)
Differentiating y w.r.t. x , n times we get
dy d2y dn y
= m e mx , 2 = m 2 e mx , K, n = m n e mx .
dx dx dx
With these values, Eqn.(21) takes the form
a 0 m n e mx + a 1 m n −1e mx + L + a n −1 me mx + a n e mx = 0 (23)
mx mx
Since e ≠ 0 for all m and x , we can divide Eqn.(23) by e to obtain
a 0 m n + a 1 m n −1 + L + a n −1 m + a n = 0 (24)
mx
Now each value of m, for which Eqn.(24) holds, will make y = e a solution of
Eqn.(21). Eqn(24) is called the auxiliary equation (A.E.) or, the characteristic
equation (C.E.) of Eqn.(21). Observe that Eqn.(24) is an algebraic equation in m of
degree n and, therefore, by the fundamental theorem of algebra, it has at least one and
not more than n distinct roots. We denote these roots by m1 , m 2 , K , m n , where m' s
need not all be distinct or real.
The following three cases arise while solving the A.E. (24).
1. All the roots of A.E. (24) are real and distinct.
2. All the roots of A.E. (24) are real but some are repeated.
3. Some of the roots of A.E. (24) are complex.
We, now, discuss these three cases one by one.
Case I: If the n roots m1 , m 2 , K, m n of A.E.(24) are distinct, then n solutions of
Eqn.(21) are
y1 = e m1x , y 2 = e m2 x , K, y n = e mn x
But these n solutions are different and linearly independent and, thus, the general
solution of Eqn.(21) is
y c = y = c1e m1x + c 2 e m 2 x + L + c n e mn x (25)
Here y c is known as the complementary function.
Let us look at an example now.
d3y d2 y dy
Example 9: Solve 3
− 2 −6 = 0.
dx dx dx
Solution: A.E. is
m3 − m2 − 6 m = 0
i.e., m(m 2 − m − 6) = 0 , or , m = 0,−2, 3
Thus the general solution is
y = c1e 0 x + c 2 e −2 x + c 3 e 3 x = c1 + c 2 e −2 x + c 3 e 3 x .
***
You may now try the following exercises:
24
E11) Solve y ′′′ + 6 y ′′ + 11y ′ + 6 y = 0 First and Higher Order
Equations
E12) Solve y′′ − 3y′ + 2 y = 0 with y = 0 and y′ = 0 when x = 0.

Case II: If A.E. (24) has a root m1 , which repeats r times, then the part of the solution
corresponding to m = m1 is
(c 1 )
+ x c 2 + x 2 c 3 + L + c r x r −1 e m1x .
Now if A.E.(24) has r roots each equal to m 1 and the remaining (n – r) roots are all
distinct, then solution of Eqn.(21) is
( )
y c = y = c1 + c 2 x + c 3 x 2 + L + c r x r −1 e m1x + c r +1e mr +1x + L + c n e mn x
We now illustrate this case with the help of the following examples.
d2y dy
Example 10: Solve 2
− 2a + a 2 y = 0, a being a positive constant.
dx dx
Solution: Here A.E. is
m 2 − 2am + a 2 = 0
and has a double root m = a . The general solution is
y( x ) = (c1 + c 2 x ) e ax .
***
Example 11: Solve y′′′ − 3 y′ + 2 y = 0.
Solution: The A.E. is
m 3 − 3m + 2 = 0 ⇒ (m + 2) (m 2 − 2m + 1) = 0
which gives m = 1, 1, − 2 and the general solution is given by
y( x ) = (c1 + c 2 x ) e x + c 3 e −2 x
***
We now discuss the case when some of the roots of A.E.(24) are complex.
Case III: We know from the theory of equations that if all the coefficients of Eqn.(21)
are real, then any complex root it may have must occur in conjugate pairs. Thus, if
α + i β is one root, then α − i β must be another root. If α ± i β are the two complex
roots of an A.E.(24), then the corresponding part of the general solution for constants
A and B is given by
((((
xxxx

A e α +i β) + B e (α −i β) x
= e αx [ A (cos β x + i sin β x ) + B (cos β x − i sin β x ) ]
= e αx [ ( A + B) cos β x + i (A − B) sin β x ] = e αx [ c1 cos β x + c 2 sin β x ],
where c1 = A + B and c 2 = i (A − B) are the new constants.
If α + i β and α − i β each occurs twice as a root of A.E. Eqn.(24), then the
corresponding part of the general solution is
e αx [ (c1 + c 2 x ) cos β x + (c 3 + c 4 x ) sin β x ].
Let us consider the following examples to illustrate this case.
2 2
 d y 
2
 dy
Example 12: Solve  − y   2 + y  = 0
 dx   dx 
Solution: The A.E. is
(m − 1) 2 (m 2 + 1) 2 = 0 ,
which gives m = 1,1, ± i, ± i . Then the general solution is
y( x ) = (c1 + c 2 x ) e x + (c 3 + c 4 x ) cos x + (c 5 + c 6 x ) sin x.

***
25
Ordinary Differential d3y
Equations Example 13: Solve + y = 0.
dx 3
Solution : The A.E. is m 3 + 1 = 0 , or, (m + 1) (m 2 − m + 1) = 0
1± i 3
which gives m = −1, and the solution is
2
−x
1
x  3 3 
y ( x ) = c 1e + e2  c 2 cos x + c 3 sin x .
 2 2 
***
How about trying the following exercises now?
E13) Solve the following differential equations:
a) 16 y ′′ + 24 y ′ + 9 y = 0
b) y ( 4 ) − 2a 2 y ′′ + a 4 y = 0 , a being a constant.
c) y ′′′ + y ′′ = 0 with y(0) = 1, y ′(0) = 0, y ′′(0) = 1.
E14) Solve the following differential equations.
d4y d2y
a) + 8 + 16 y = 0
dx 4 dx 2
b) y ′′ + 4 y = 0
d3y dy
c) 3
−2 + 4y = 0
dx dx
d) y (6 ) + 9 y ( 4 ) + 24 y ( 2 ) + 16 y = 0.

We next take up the solution of the non-homogeneous linear DEs with constant
coefficients.
1.4.2 Solution of Non-Homogeneous Linear Differential Equations
with Constant Coefficients
The general solution of a non-homogeneous linear DE
dn y d n −1 y dy
a 0 n + a 1 n −1 + L + a n −1 + a n y = f ( x ), (26)
dx dx dx
where a 0 ≠ 0, f ( x ) ≠ 0 and a 0 , a 1 , K, a n are constants is y( x ) = y c + y p .
Here y c and y p are respectively, known as the complementary function (C.F.) and
particular integral (P.I.) or, particular solution. In Sec.1.4.1, we discussed, for
f ( x ) = 0 , the methods of finding the general solution y = y c . It now remains to find
the P.I. of a given differential equation.
Before going into the details of the procedures of finding the P.I. of a given DE, let us
recall the definition of the differential operator.
Definition: A mathematical device by means of which we can convert one function
into another is known as an operator.
The operation of differentiation is an operator as it converts a differentiable function
f ( x ) into a new function f ′( x ) . The letter D , which we shall use to denote the
differentiation, is called the differential operator.
If y is an nth order differentiable function, then
D 0 y = y, Dy = y′, D 2 y = y′′, K, D n y = y ( n ) (27)
If F(D) is a polynomial operator of order n defined by
F(D) = a 0 D n + a 1 D n −1 + L + a n −1 D + a n , a 0 ≠ 0, (28)
th
and y is an n order differentiable function, then
26
 F(D) y = (a 0 D n + a1D n −1 + L + a n −1D + a n ) y First and Higher Order
Equations
= a 0 D n y + c1D n −1y + L + a n −1Dy + a n y
= a 0 y n + a1 y n −1 + L + a n −1 y′ + a n y, (29)
and Eqn.(26) can be written as
F(D) y = f ( x ), a 0 ≠ 0.
If F(D) is defined by the relation (28), then
F(D) = a 0 (D − m1 ) (D − m 2 ) K (D − m n ), (30)
where m1 , m 2 , K, m n are the real or complex roots of A.E. corresponding to
F(D) y = 0. We can thus say that a polynomial operator with constant coefficients
can be factored just the way we factor an ordinary polynomial.
Remark 1: F(D) is a polynomial operator given by Eqn.(28) , then
F(D) = F1 (D). F2 (D),
where F1 (D) and F2 (D) may be composite factors of F(D) , i.e., F1 and F2 may be
products of factors of Eqn.(30).
Remark 2: The polynomial operator satisfies the commutative law for multiplication,
viz; (D − m1 ) (D − m 2 ) = (D − m 2 ) ( D − m1 ).
Remark 3: If F(D) is a polynomial operator defined by Eqn.(28) and g ( x ) is an nth
order differentiable function of x , then
F(D) [g ( x ) e ax ] = e ax F(D + a ) g ( x ),
where ‘a’ is a constant.
We are now in a position to give a method for solving Eqn(26) by making use of the
polynomial operator. The procedure is illustrated by means of the following example
Example 14: Solve the differential equation y ′′′ + 2 y′′ − y ′ − 2 y = e 2 x
Solution : In the operator notation, the given DE can be written as
( D 3 + 2 D 2 − D − 2) y = e 2 x
or, (D − 1) (D + 1) (D + 2) y = e 2 x (31)
Let u = (D + 1) (D + 2) y (32)
Then Eqn.(31) becomes
(D − 1) u = e 2 x , (33)
which is a linear differential equation and its solution is
u = e 2 x + c1 e x .
Putting this value of u in Eqn.(32), we get
(D + 1) (D + 2) y = e 2 x + c1e x (34)
Let v = ( D + 2) y (35)
Then Eqn.(34) becomes
(D + 1) v = e 2 x + c1e x
which is a linear equation and its solution is
1 c
v = e 2x + 1 e x + c 2 e −x .
3 2
putting the value of v in Eqn.(35), we obtain
1 c
Dy + 2 y = e 2 x + 1 e x + c 2 e − x ,
3 2
which is again, a linear equation with the solution as
1 c
y = e 2x + 1 e x + c 2 e −x + c 3 e −2 x
12 6
1 2x c
Here e is a P.I. (free from constant) and 1 e x + c 2 e − x + c 3 e − 2 x is C.F.
12 6
***
27
Ordinary Differential In general, if the non-homogeneous linear differential equation
Equations
a 0 y n + a 1 y n −1 + L + a n −1 y + a n = f ( x ), a 0 ≠ 0,
of order n is written as
(D − m1 ) ( D − m 2 ) K (D − m n ) y = f1 ( x ), (36)
where m1 , m 2 , K, m n are the roots of A.E., then a general solution can be obtained as
follows:
Let u = (D − m 2 ) (D − m 3 ) K (D − m n ) y (37)
Then Eqn.(36) takes the form
(D − m1 ) u = f1 ( x ), (38)
which is a linear equation in u . Obtain its solution and put it in Eqn.(37) to get
(D − m 2 ) (D − m 3 ) K (D − m n ) y = u ( x ) (39)
Now, let v = (D − m 3 ) (D − m 4 ) K ( D − m n ) y (40)
Then Eqn.(39) becomes
(D − m 2 ) v = u ( x ), (41)
which is linear in v . Obtain its solution and put it in Eqn.(40) to get
(D − m 3 ) (D − m 4 ) K (D − m n ) y = v ( x )
Repeat this process (n − 2) times to get the solution for y .
We shall now define the inverse operator and use it to find the P.I. We shall also give
some shortcut methods to find P.I. when the right hand side, i.e., the non-
homogeneous term of the given equation is a polynomial in x, an exponential function,
sine or cosine function or, combination of these functions.
Inverse Operation
We start with the definition of inverse operator
Definition : Let F(D) y = f ( x ), where F(D) is the polynomial operator defined as
F(D) = a 0 D n + a 1 D n −1 + L + a n −1 D + a n
and f ( x ) is a function of x . The inverse operator of F(D) , written as F −1 (D) or
1
, is then defined as an operator, which when operated on f ( x ) , gives the P.I. y p
F(D)
of F(D) = f ( x ) , i.e.,
1
F −1 (D) f ( x ) = y p or y p = f (x )
F(D)
Remark 1: From the above definition, we conclude that D − n f ( x ) will mean
integration of f ( x ) n-times by ignoring constants of integration.
Remark 2: If F(D) y = 0, then
1
yp = (0) = 0.
F(D)
Remark 3: From the above definition and general method stated above,
1
(D − m1 ) (D − m 2 ) K (D − m n ) ∫ (∫
f ( x ) = e m n −1 e ( m n −1 − m n ) x e ( m n − 2 − m n −1 ) x K ∫
(∫ e ( m1 − m 2 ) x
(∫ e − m1x
) )
f ( x )dx L dx dx

Remark 4: If we write P.I. of F(D) y = f ( x ) as

1 1
yp = f (x ) = f ( x ) , then following the ordinary
F( D ) (D − m1 ) (D − m 2 ) K (D − m n )
rules of algebra, for distinct m1 , m 2 , K m n , we can write
 α1 α2 αn 
y p =  + +L+ f ( x )
 D − m1 D − m 2 D − mn 
28
 ∫ ∫ ∫
= α 1e m1x e − m1x f ( x )dx + α 2 e m 2 x e − m 2 x f ( x )dx + L α n e m n x e − m n x f ( x )dx , where First and Higher Order
Equations
coefficients α1 , α 2 ,K, α n can be obtained by a simple algebraic manipulations (as in
partial fractions).
In case a root m1 of A.E. is repeated r times, the corresponding partial fraction will be
α1 α2 αr
+ 2
+L+
D − m1 ( D − m1 ) ( D − m1 ) r
and the corresponding terms in the integral will be
∫
α1e m1x e − m1x f ( x )dx + α 2 e m1x ∫∫ e f (x ) (dx) + L
− m1x 2

∫∫K ∫ e f (x) (dx) .

− m1x
L + α r e m1x r

We illustrate the above theory and remarks through the following examples:
d2y
Example 15: Find the particular integral of + y = sec 2 x.
dx 2
Solution : Here,
1 1
P.I. = y p = 2
sec 2 x = sec 2 x
D +1 ( D + i) ( D − i)
1  1 1 
= − sec 2 x
2i  D − i D + i 


=
2i
[
1 ix −ix
∫
e e sec 2 x dx − e −ix e ix sec 2 x dx ∫ ]
1 cos x − i sin x cos x + i sin x 
= eix
2i  ∫ 2
cos x
dx − eix
cos 2 x∫ dx 

=
1 ix
2i
[ ∫ ∫
e (sec x − i sec x tan x ) dx − e ix (sec x + i sec x tan x ) dx ]
=
1 ix
2i
[( )∫ (
e − e −ix sec x dx − i e ix + e −ix sec x tan x dx )∫ ]
1
= [(2i sin x ) ln sec x + tan x − (2i cos x ) sec x ]
2i
= sin x ln sec x + tan x − 1
***
Example 16: Find the particular integral of (D − 1) 2 (D + 1) 2 y = e x .
Solution: The particular integral is
1 1  −1 1 1 1  x
yp = 2 2
ex =  + 2
+ + e
(D − 1) (D + 1) 4  D − 1 (D − 1) D + 1 (D + 1) 2 

=
1
4
[ ∫
− e x e − x e x dx + e x ∫ (∫ ) ∫
e − x e x dx dx + e − x e x e x dx + e − x ∫ (∫ e
x x
) ]
e dx dx

1 x2 x 1 x 1 x 
= − x e x + e + e + e .
4 2 2 4 
***
You may now try the following exercises.
E15) Find the complete solution of the following equations:
a) y ′′′ + y ′ = x 3 + cos x
b) (D 3 + 3D 2 − D − 3) y = cosh x
c) (D 3 + D 2 + 4D + 4) y = sin 2 x
d) y ′′ + n 2 y = sec nx
29
Ordinary Differential e) (D 3 − D 2 − 8D + 12) y = X( x ).
Equations

The general method of computing a particular integral as given above is quite

laborious and requires a lot of calculations as can be seen from Examples 15 and 16
above. However, in certain cases, the P.I. can be obtained by shorter methods. We
shall now consider these shorter methods.
Short Methods of Finding Particular Integrals
Consider the general nth order linear DE, namely,
F(D) y = (a 0 D n + a 1 D n −1 + L + a n −1 D + a n ) y = f ( x ), a 0 ≠ 0,
where a 0 , a 1 ,K, a n are constants.
For certain particular forms of the non-homogeneous term f ( x ) , shorter methods of
finding P.I. are available which we are stating below.

Case I: When f(x) = e α x , α constant.

1 αx 1 αx
e = e , F(α) ≠ 0
F(D) F(α)
When, F(α) = 0 and F(D) = (D − α) p φ(D), φ(α) ≠ 0 for some p , then
1 αx 1
e = p
e αx , φ(α) ≠ 0
F(D) ( D − α) φ(D)
x p 1 αx
= . e .
p! φ(α)

Case II: when f(x) = cos (ax + b), or, sin (ax + b).
If f (D) = φ (D 2 ) ,
1 1
then 2
cos (ax + b) = 2
cos (ax + b), if φ (−a 2 ) ≠ 0
φ( D ) φ(−a )
1 1
and, 2
sin (ax + b) = 2
sin (ax + b), if φ (−a 2 ) ≠ 0
φ( D ) φ ( −a )
If, φ (−a 2 ) = 0 and φ (D 2 ) = (D 2 + a 2 ) p ψ (D 2 ) for some p and ψ (−a 2 ) ≠ 0
1 1
Then, 2
cos (ax + b) = 2 cos (ax + b)
φ (D ) (D + a ) ψ(D 2 )
2 p

1  1 
= 2  2 2 p
cos (ax + b)  if , ψ( −a 2 ) ≠ 0.
ψ ( −a )  ( D + a ) 
1 1
and, 2
sin (ax + b) = 2 sin (ax + b)
φ (D ) (D + a ) ψ(D 2 )
2 p

1  1 
= 2  2 2 p
sin (ax + b)  if , ψ( −a 2 ) ≠ 0.
ψ ( −a )  ( D + a ) 

Note that the terms in the brackets above are evaluated by the general method.
Cosine and sine functions in Case II above can also be dealt as exponential functions
Symbols Re and Im by writing them in the form:
are read as ‘real part
of’ and ‘imaginary part cos (ax + b) = Re e i ( ax + b ) ,
of’ respectively. and sin (ax + b) = lm e i (ax + b )

Case III: When f(x) is a polynomial in x.

Let F(D) be a polynomial in D of degree n and let f ( x ) be a polynomial in x of

degree say, K . Then
30
 1 1 First and Higher Order
f (x) = n n −1
f (x) Equations
F(D) a 0 D + a 1 D + L + a n −1 D + a n
−1
1  a a a 
= 1 + n −1 D + L 1 D n −1 + 0 D n  f ( x )
an  an an an 
( )
= c 0 + c1 D + c 2 D 2 + L + c k D k f ( x ) + 0 (D k +1 ) f ( x )
= (c 0 + c1 D + c 2 D 2 + L + c k D k )f (x ) , the second term on r.h.s. is zero.
Case IV: When f(x) = e α x V(x), α constant.
1  αx 1
e V ( x )  = e αx V(x )
F(D)   
 F(D + α)
This result is known as shifting theorem.
Remark 1: We may remark that Euler’s equation
(a 0 x n D n + x n −1a 1 D n −1 + L + a n −1 x D + a n ) y = f ( x )
can be reduced to an equation with constant coefficients with the help of
substitution x = e z .
Remark 2: Differential Equation
[a 0 (ax + b) n D n + a 1 (ax + b) n −1 D n −1 + L + a n −1 (ax + b)D + a n ] y = f ( x )
can be either reduced to Euler’s equation by the substitution ax + b = z , or, it can be
reduced to an equation with constant coefficients by the transformation ax + b = e z .
We, now, take up some examples to illustrate the shorter methods of finding P.I. stated
above.
Example 17: Find P.I. of y ′′′ + y ′′ + y′ + y = x 4 + 2 x + 1
1
Solution: Here P.I. = ( x 4 + 2x + 1)
1 + D + D2 + D3
1− D 4
= ( x + 2 x + 1)
1 − D4
[
= (1 − D 4 ) −1 x 4 + 2 x + 1 − 4x 3 − 2 ]
4 8 4 3
= (1 + D + D + L) ( x − 4 x + 2x − 1)
= ( x 4 − 4 x 3 + 2x − 1) + 24
= x 4 − 4 x 3 + 2 x + 23
***
Example 18: Find P.I. of (D 4 − 2D 3 + D 2 ) y = x
1
Solution : Here P.I. = x
D − 2D 3 + D 2
4

1
= 2
[1 − (2D − D 2 )]−1 x
D
=
D
1
2
[
1 + 2D − D 2 + L x ]
1
= [x + 2]
D2
1  x2 
=  + 2 x 

D 2 
x3
= + x2.
6
***
31
Ordinary Differential Example 19: Solve (D 2 − 2D + 1) y = x sin x
Equations
Solution : A.E. is m 2 − 2m + 1 = 0 ⇒ m = 1,1
∴ C.F. = y c = (c1 + c 2 x ) e x
1 1
P.I. = 2 x e x sin x = e x 2
x sin x
D − 2D + 1 ( D + 1) − 2(D + 1) + 1
= e x D − 2 ( x sin x )
= e x [− x sin x − 2 cos x ]
∴General solution is, y( x ) = (c1 + c 2 x )e x + e x (− x sin x − 2 cos x ).
***
Example 20: Find the particular integral of (D 2 + 1) y = x 2 sin 2 x.
1 1
Solution : Here P.I. = 2
x 2 sin 2 x = Im 2 ( x 2 e 2ix )
D +1 D +1
1 1
= Im of e 2ix 2
x 2 = Im of e 2ix 2 x2
(D + 2i) + 1 D + 4iD − 4 + 1
−1
1   4iD + D 2  2
1 −   x
2 ix
= Im of e
− 3   3 

 4iD + D 2 16 2
e 2ix 
= Im of 1 + − D + L x 2
−3 3 9 
2 ix
e  4i 13 2 3  2 e 2ix  2 8i 13 
= Im of 1 + D − D + 0( D ) x = Im of  x + 3 x − 9 .2 
− 3  3 9 
 −3  
 cos 2 x + i sin 2 x   2 26 8  8 1 26 
= Im of  x − + i x  = − x cos 2 x −  x 2 −  sin 2 x
 −3  9 3  9 3 9 

=−
1
27
[
24 x cos 2 x + (9 x 2 − 26) sin 2 x . ]
***
Example 21: Find P.I. of (D 4 − 1) y = sin x.
1 1
Solution: Here P.I. = 4 sin x = sin x
D −1 (D − 1) (D 2 + 1)
2

1 1
= 2
sin x
(−1 − 1) D + 1
1 x  x
= −  − cos x  = cos x
2 2  4
***
d2y dy
Example 22: Solve (1 + x ) 2 2
+ (1 + x ) + y = 2 sin [ ln (1 + x ) ]
dx dx
d
Solution: Let 1 + x = e t , t = ln (1 + x ) and D ≡ . Then the given equation
dt
becomes
[D (D − 1) + D + 1] y = 2 sin t , i.e.,
(D 2 + 1) y = 2 sin t ,
which is a linear equation with constant coefficients with
C.F. = y c = c1 cos t + c 2 sin t
1
and P.I. = y p = 2
2 sin t = − t cos t
D +1
32
Hence, the complete solution is First and Higher Order
y = y c + y p = c1 cos t + c 2 sin t − t cos t Equations

= c1 cos [ ln (1 + x )] + c 2 sin [ ln (1 + x )] − ln (1 + x ) cos [ ln (1 + x ) ] .

***
Example 23: Solve x 3 D 3 y + 3x 2 D 2 y + x Dy + y = x + ln x
d
Solution: Let x = e t , ≡ D1 , so that the given equation becomes
dt
[D1 (D1 − 1) (D1 − 2) + 3D1 (D1 − 1) + D1 + 1 ] y = e t + t
i.e., (D13 + 1) y = e t + t
1 3
A.E. is, m 3 + 1 = 0, which has the roots m = −1, ± i . Thus,
2 2
t
 3 3 
C.F. = y c = c1e − t + e 2 c 2 cos t + c 3 sin t
 2 2 
  3   3 
= c1 x −1 + x c 2 cos  ln x  + c 3 sin  ln x  
 2   2 
     
1 1 1 t
P.I. = y p = 3 et + 3 t= e + (1 − D13 + L) t
D1 + 1 D1 + 1 1 + 1
1 t 1
= e + t = x + ln x
2 2
Therefore, the required solution is
  3   3  1
y = y c + y p = c1 x −1 + x c 2 cos ln x  + c 3 sin  ln x   + x + ln x.
 2   2  2
     
***
You may now try the following exercises.
E16) Solve the following differential equations
a) (D 4 − 2D 3 + 2D 2 − 2D + 1) y = 0
b) (D 4 + 2D 3 − 3D 2 + 4D + 4) y = 0
dy
c) (D 2 − 2D + 5) y = 0 , given that y = 0 and = 4, when x = 0.
dx
E17) Obtain the general solution of the following differential equations
a) y ′′′ + 3y ′′ + 3y ′ + y = e − x (2 − x 2 ).
x 3x
b) (D 2 + 1) (D 2 + 4) y = cos cos .
2 2
5 x
c) (D − D) y = 12e + 8 sin x − 2 x.
d) (D 2 − 1) y = e − x + cos x + x 3 + e x cos x.
e) ( x + 3) 2 y ′′ − 4( x + 3) y ′ + 6 y = ln ( x + 3).
f) (D 2 + a 2 ) y = tan ax

We now end this unit by giving a summary of what we have covered in it.

1.5 SUMMARY
In this unit, we have shown you the following:
1. The differential equation together with the initial conditions is called an initial
value problem (IVP).

33
Ordinary Differential 2. The differential equation together with the boundary conditions is called
Equations boundary value problem (BVP).
dy
3. IVP, = f ( x , y), y( x 0 ) = y 0 is equivalent to the integral equation
dx
x
y( x ) = y 0 + ∫ f [t, y(t) ]dt.
x0

4. Picard’s Method of successive Approximations for IVP,

dy
= f ( x , y), y( x 0 ) = y 0 , is given by
dx
x
y n (x) = y 0 + ∫ f [t, y
x0
n −1 ( t ) ]dt, n = 1, 2, 3, K .
∂f
5. If f ( x , y) and be continuous functions of x and y on a closed rectangle
∂y
R and f ( x 0 , y 0 ) is an interior point of R , then there exists a number h > 0
dy
with the property that the initial value problem = f ( x , y), y( x 0 ) = y 0 has one
dx
and only one solution y = y( x ) on the interval x − x 0 ≤ h . It is called the
Picard’s theorem on existence and uniqueness of solution of IVP.
f ( x , y1 ) − f ( x , y 2 )
6. The inequality < N (a constant), is called a Lipschitz
y1 − y 2
condition in the variable y and the constant N is called a Lipschitz constants.
7. Continuity of f ( x , y) guarantees a solution of IVP, y ′ = f ( x , y), y( x 0 ) = y 0 ,
∂f
whereas, Lipschitz condition or, continuity of ensures a unique solution.
∂y
8. For a linear differential equation of order n of the form
a 0 ( x ) y ( n ) + a 1 ( x ) y ( n −1) + L + a n −1 ( x ) y ′ + a n ( x ) y = f ( x ), a 0 ( x ) ≠ 0 ,
let x = x 0 be a point of interval [a , b] and let k 0 , k 1 ,K , k n −1 be an arbitrary set
of n constants. Then there exists one and only one solution y( x ) of the
equation in [a , b] , with the property
y( x 0 ) = k 0 , y′( x 0 ) = k 1 , K , y ( n −1) ( x 0 ) = k n −1,
where continuity of a 0 ( x ), a 1 ( x ), K, a n ( x ) in [a , b] has been assumed.
9. The ‘n’ functions y1 ( x ), K, y n ( x ) defined on [a , b] are said to be linearly
dependent in [a, b] if, for all x in that interval, there exists a relation
c1 y 1 + c 2 y 2 + L + c n y n = 0
in which the ci are constants and not all are zero. If such a relation does not
exist, then the functions are said to be linearly independent in [a, b].
10. If a 0 ( x ), a 1 ( x ), K , a n ( x ), f ( x ) are continuous functions of x on [a, b] and
a 0 ( x ) ≠ 0, then

a) for a homogeneous equation

a 0 ( x ) y ( n ) + a 1 ( x ) y ( n − `1) + L + a n −1 ( x ) y ′ + a n ( x ) y = 0 ,
if, y1 ( x ), y 2 ( x ), K , y n ( x ) are n linearly independent solutions of the
equation in [a, b ] then y( x ) = c1 y 1 ( x ) + c 2 y 2 ( x ) + L + c n y n ( x ) is the
general solution of the equation for constants c1 , c 2 , K , c n .
b) for a non-homogeneous equation
a 0 ( x ) y ( n ) + a 1 ( x ) y ( n −1) + L + a n −1 ( x ) y ′ + a n ( x ) y = f ( x )
If Y( x ) is a particular integral and y1 , y 2 ,K, y n are n linearly independent
34
 solutions of the corresponding homogeneous equation then, First and Higher Order
y( x ) = Y( x ) + c1 y 1 ( x ) + L + c n y n ( x ) is the general solution for constants Equations

c1 , c 2 , K , c n
11. Solution y, of an nth order linear DE
a 0 y ( n ) + a 1 y ( n −1) + L + a n −1 y ′ + a n y = 0, a 0 ≠ 0,
with constant coefficients a 0 , a 1 ,K , a n having n roots m1 , m 2 , K, m n when
a) roots are all real and distinct, is
y = c1e m1x + c 2e m2 x + L + c n e mn x .
b) roots are real and repeated, say m1 = m 2 = L = m r , is
y = (c1 + c 2 x + L + c r x r −1 ) e m1x + c r +1e m r +1x + L + c n e m n x .
c) roots are complex and one such pair is α ± i β , then corresponding part of
solution is y = e αx (c1 cos βx + c 2 sin βx ).
12. For a non-homogeneous equation
a 0 y ( n ) + a 1 y ( n −1) + L + a n −1 y ′ + a n y = f ( x ), a 0 ≠ 0 , with constant coefficients
a 0 , a 1 ,K, a n
a) the complete solution of the corresponding homogeneous part is called its
complementary function (C.F.).
b) particular solution of the non-homogeneous part involving no arbitrary
constant is called its particular integral (P.I.).
c) complementary function and particular integral together constitute its
general solution.
13. A mathematical device by means of which we can convert one function
d
into another is known as an operator, e.g., D = is differential operator.
dx
14. If F(D) is a polynomial operator of order n , then it is defined as
F(D) = a 0 D n + a 1 D n −1 + L + a n −1 D + a n , a 0 ≠ 0.
A polynomial operator with constant coefficients can be factored like an
ordinary polynomial.
15. Euler’s equation (a 0 x n D n + a 1 x n −1 D n −1 + L + a n −1 x D + a n ) y = f ( x )
can be reduced to an equation with constant coefficients by substituting x = e z .
16. D.E. [a 0 (ax + b) n D n + a 1 (ax + b) n −1 D n −1 + L + a n −1 (ax + b)d + a n ] y = f ( x )
can be either reduced to Euler’s equation by the substitution ax + b = z or it can
be reduced to an equation with constant coefficients by using ax + b = e z .

1.6 SOLUTIONS/ANSWERS
E1) If λ is negative say λ = −α then the solution is
y = c1 e − − α x + c 2 e − α x
when y(0) = 0 and y(π) = 0 , we get c 2 = 0 = c1
Hence only solution of given problem is the trivial solution y = 0 .
If λ = 0 , then general solution is y( x ) = c1 x + c 2
Again y(0) = 0 and y(π) = 0 give the trivial solution y = 0 .
If λ is positive, the general solution is
y( x ) = c1 sin λ x + c 2 cos λ x .
For y(0) = 0 , we get 0 = c 2 and the solution reduces to y( x ) = c1 sin λ x .
For the second b.c. y(π) = 0 , we get 0 = c1 sin λ π
For the non-trivial solution, we must have sin λ π = sin nπ for some positive
integer n , so that λ = n 2 , i.e, λ must be one of the numbers 1, 4, 9, K .
Hence the non-trivial solution of given problem are
35
Ordinary Differential y( x ) = a 1 sin x , or, a 2 sin 2 x , or, a 3 sin 3x , or K .
Equations
And the solution vanishes at the end points 0 and π of interval [0, π] .
dy
E2) a) Here = x + y, y(0) = 1 . Thus x 0 = 0, y 0 = 1 , f ( x , y) = x + y
dx
x x
x2
∫ ∫
∴ y 1 = y 0 + f ( x , y 0 ) dx = 1 + ( x + 1) dx = 1 + x +
2
0 0
x x
 x2  x3
y2 = y0 + ∫ f ( x , y1 ) dx = 1 + ∫ x +1+ x +


 dx = 1 + x + x 2 +
2  6
0 0
x x
 x3 
y3 = y0 + ∫ f (x, y ) = 1 + ∫2
 x + 1 + x + x 2 +  dx

 6 
0 0
3 4
x x
= 1 + x + x2 + +
3 24
dy 1 1
b) = x − y 2 , y(0) = . Here x 0 = 0, y 0 = , f ( x , y) = x − y 2
dx 2 2
x x
1  1 1 1 x2
∫
∴ y 1 = y 0 + ( x − y 02 ) dx = +
2 ∫ x −

 dx = − x +
4 2 4 2
0 0
x
1
x
 1 1 x4 x2 1 1  
∫
y 2 = y 0 + ( x − y 12 ) dx =
2
+ ∫ 

x − 
 +
 4 16
x +
4
+ − x − x 3   dx
2 4 4  
0 0

1 1 19 2 x 3 1 4 x 5
= − x+ x − + x −
2 4 32 6 16 20
x

∫
y 3 = y 0 + ( x − y 22 ) dx
0

1 1 5 21 3 89 4 1531 x 5 67 6 1303 7
= − x + x2 − x + x − . + x − x
2 4 8 96 768 3072 5 1440 80640
77 8 77 9 1 x 11
+ x − x + x 10 − .
7480 8640 1600 4400
dy
c) = 2x (1 + y), y(0) = 0 . Here x 0 = 0, y 0 = 0, f ( x , y) = 2 x (1 + y)
dx
x x

∫
∴ y 1 = y 0 + f ( x , y) dx = 0 + 2 x dx = x 2 ∫
0 0
x
x4
∫
y 2 = y 0 + 2x (1 + x 2 ) dx = 0 + x 2 +
2
0
x
 x4  x4 x6

∫
y 3 = y 0 + 2x 1 + x 2 +  dx = x 2 +
2  2
+
6
0
dy
d) = y, y(0) = 1 . Here x 0 = 0, y 0 = 1, f ( x , y) = y
dx
x x

∫
y1 = y 0 + f ( x , y 0 ) dx = 1 + dx = 1 + x ∫
0 0
x x
x2
∫ ∫
y 2 = y 0 + f ( x , y1 ) dx = 1 + (1 + x ) dx = 1 + x +
2
0 0
x x
 x2  x2 x3
∫
y 3 = y 0 + f ( x , y 2 ) dx = 1 + 1 + x +
 2 ∫
 dx = 1 + x +
2
+
6
.
0 0
36
 dy First and Higher Order
e) = y 2 , y(0) = 1 . Here x 0 = 0, y 0 = 1, f ( x , y) = y 2 Equations
dx
x

∫
y1 = y 0 + f ( x , y 0 ) dx = 1 + x
0
x
x3
∫
y 2 = y 0 + f ( x , y1 ) dx = 1 + x + x 2 +
3
0
x
2x 4 x 5 x 6 x 7
∫
y 3 = y 0 + f ( x , y 2 ) dx = 1 + x + x 2 + x 3 +
3
+
3
+
9
+
63
.
0

 2y
 , for x > 0 , y(0) = 0
E3) Here f ( x , y) =  x
 0 , for x = 0
2
y1 − y 2
f ( x, y1 ) − f ( x , y 2 ) x 2
Thus, = = .
y1 − y 2 y1 − y 2 x

Hence f ( x , y) doesn’t satisfy Lipschitz conditions in any closed rectangle

containing (0, 0) .
∂f
E4) a) f ( x , y) = x and = 0 are continuous on a whole plane and in every
∂y
rectangle. Thus f ( x , y) satisfies the Lipschitz condition in every such
x2
rectangle and there exists a unique solution y = + 1.
2
b) f ( x , y) = − | y | is not continuous and solution does not exist at any point in
the neighbourhood of origin.
∂f
E5) Here f ( x , ) = xy 2 , = 2xy and,
∂y
| f ( x, y1 ) − f ( x , y 2 ) | | xy12 − xy 22 | | x ( y1 − y 2 ) ( y1 + y 2 ) |
= = = | x | | y1 + y 2 |
| y1 − y 2 | | y1 − y 2 | | y1 − y 2 |
and the right hand side is ≤ N for , a ≤ x ≤ b, c ≤ y ≤ d but is not less than equal
to some finite constant for a ≤ x ≤ b, − ∞ < y < ∞ . Hence Lipschitz’s condition
is satisfied for any rectangle a ≤ x ≤ b, c ≤ y ≤ d but is not satisfied for any strip
a ≤ x ≤ b, − ∞ < y < ∞ .
y −1 ∂f 1
E6) Here f ( x , y) = . Thus = , which does not exist at the origin. Also
x ∂y x
y1 − 1 y 2 − 1 1
− | y1 − y 2 |
| f ( x , y1 ) − f ( x , y 2 ) | x x x 1
= = = , which is
| y1 − y 2 | | y1 − y 2 | | y1 − y 2 | x
unbounded in every neighbourhood of origin.
Hence the Lipschitz condition is not satisified. The given problem may have an
infinity of solutions.
4x 2 y
E7) Here f ( x , y) = , when x and y are not both zero
x4 + y2
=0, when x = y = 0
Hence function f ( x , y) is continuous function of x and y (prove it.)
On the other hand,
37
Ordinary Differential 4 x 3 ( x 4 − y1 y)
Equations | f ( x , y 1 ) − f ( x , y) | = ( y1 − y)
( x 4 + y 2 ) ( x 4 + y12 )
4(1 − pq) ( y 1 − y)
=
2 2
, where, y = px 2 , y1 = qx 2
(1 + p ) (1 + q ) x
| 1 − pq | |y − y|
=4 2 2
. 1
| (1 + p ) (1 + q ) | | x |
and therefore the Lipschitz condition is not satisfied in any region containing the
origin. The equation admits the solution y = c 2 − x 4 + c 4 , where c is an
arbitrary real constant. Thus there is an infinity of solutions satisfying the intial
conditions x = 0, y = 0 .
E8) a) y1 + y 2 is a solution of the equation
b) c1 y1 + c 2 y 2 (for arbitrary c1 , c 2 ) is not a solution of the equation when
c 2 ≠ 0,1 .
E9) It can be easily verified that = sin ax and y 2 = cos ax satisfy the equation
y ′′ + a 2 y = 0
and, hence, are the solutions of the given equation on the interval ] − ∞, ∞ [ .
y1 y2 sin ax cos ax
Here W ( y1 , y 2 ) = =
y1′ y ′2 a cos ax − a sin ax
= −a sin 2 ax − a cos 2 ax = −a ≠ 0 .
Hence sin ax and cos ax are linearly independent solutions of the given
equation. The general solution can then be written as y = c1 sin ax + c 2 cos ax .
1 1 1 1 1 1
E10) Let y1 = sin and y 2 = cos , y1′ = − 2 cos , y ′2 = 2 sin
x x x x x x
1 1 2 1 −1 1 2 1
y1′′ = − 4 sin + 3 cos , y ′2′ = 4 cos − 3 sin
x x x x x x x x
4 3
′′ ′
Here x y1 + 2x y1 + y1
 1 1 2 1  1 1 1
= x 4  − 4 sin + 3 cos  + 2 x 3  − 2 cos  + sin = 0
 x x x x  x x x
Hence y1 satisfies given equation for 0 < x < ∞ .
Similarly x 4 y ′2′ + 2 x 3 y ′2 + y 2 = 0 , thus y 2 is also a solution of given equation.
1 1
y y2 sin cos
Here W ( y1 , y 2 ) = 1 = x x = 1 ≠ 0 , for 0 < x < ∞ .
y1′y ′2 − 1 cos 1 1 sin 1 x 2
x2 x x2 x
Hence y1 and y 2 are linearly independent solutions of given equation and the
general solution can be written as
1 1 1 1 1 1
y = c1 sin + c 2 cos and y ′ = −c1 2 cos + c 2 2 sin
x x x x x x
1
Now y  = 1 ⇒ 1 = c1 sin π + c 2 cos π = −c 2 ⇒ c 2 = −1
 π
1 1
y ′  = −1 ⇒ −1 = −c1 π 2 cos π + c 2 π 2 sin π ⇒ −1 = c1 π 2 ⇒ c1 = − 2
 
π π
1 1 1
Hence the required solution is y = − 2 sin − cos .
π x x

E11) A.E. is m 3 + 6m 2 + 11m + 6 = 0 , m = −1, − 2, − 3

38
 Thus the general solution is, y = c1 e − x + c 2 e −2 x + c 3 e −3 x . First and Higher Order
Equations
E12) A.E. is m 2 − 3m + 2 = 0 , m = 1,2
Thus the general solution is, y = c1 e x + c 2 e 2 x
Satisfying the conditions y = 0 and y ′ = 1 when x = 0 , we get c1 = −1, c 2 = 1
Hence the solution of given problem is y = −e x + e 2 x

E13) a) A.E. is 16m 2 + 24m + 9 = 0 , m = −3 / 4, − 3 / 4

−3
x
y = (c 1 + c 2 x )e 4
b) A.E. is m 4 − 2a 2 m 2 + a 4 = 0 , m = a , a , − a , − a .
Hence the solution is
y = (c1 + c 2 x )e ax + (c 3 + c1 x )e − ax .
c) A.E. is m 3 + m 2 = 0 , m = 0,0, − 1
Thus, solution is y = (c1 + c 2 x ) + c 3 e − x
Initial conditions y(0) = 1, y ′(0) = 0, y ′′(0) = 1 give c1 = 0, c 2 = 1, c 3 = 1
Hence required solution is y = x + e − x .

E14) a) A.E. is m 4 + 4m 2 + 16 = 0 , m = 2i, 2i, − 2i , − 2i

Hence the solution is
y = (c1 + xc 2 ) cos 2 x + (c 3 + xc 4 ) sin 2 x .
b) y = c1 cos 2 x + c 2 sin 2 x .
c) y = c1 e −2 x + e x (c 2 cos x + c 3 sin x )
d) A.E. is m 6 + 9m 4 + 24m 2 + 16 = 0 , m = ± i, ± 2i, ± 2i
Hence the solution is
y = (c1 cos x + c 2 sin x ) + (c 3 + xc 4 ) cos 2 x + (c 5 + xc 6 ) sin 2x .

E15) a) A.E. is m 3 + m = 0 , m = 0, m = ± i
C.F. = c1 + c 2 cos x + c 3 sin x
1 1
P.I. = 3 ( x 3 + cos x ) = ( x 3 + cos x )
D +D D ( D + i ) ( D − i )
1 1 1 1 1  3
= + . − .  ( x + cos x )
D 2 D−i 2 D+i
1 1
∫ ∫ ∫
= ( x 3 + cos x ) dx + eix e − ix ( x 3 + cos x ) dx − e − ix eix ( x 3 + cos x ) dx
2 2
4
x 1
= − 3x 2 − x cos x
4 2
x4
Hence y = c1 + c 2 cos x + c 3 sin x + − 3x 2 + x cos x
4
b) C.F. = c1 e x + c2 e − x + c 3 e −3 x
1 1 1 1 1 1 1 1 x −x
P.I. = cosh x =  . − . + . . (e + e )
(D − 1) (D + 1) (D + 3)  8 D − 1 4 D + 1 8 D + 3  2
1 x −x x 1 1
∫ ∫ ∫
= e e (e + e − x ) dx − e − x e x (e x + e − x ) dx + e −3 x e 3x (e x + e − x ) dx
16 8 16

1 1 3 x
= x e x − x e −x − e
16 8 64
1 1
Hence y = c1e x + c 2 e − x + c 3 e −3x + x e x − x e −x
16 8
39
Ordinary Differential c) C.F. = c1e − x + c 2 cos 2x + c3 sin 2x
Equations
1 1 1 1 1 1 1 
P.I. = 2
sin 2 x =  . − . +  sin 2x
(D + 1)(D + 4)  5 D + 1 4(i + 2) D + 2i 4(i − 2) D − 2i 
1 1 1
5 ∫
= e − x e x sin 2x dx −
4(i + 2) ∫
e − 2ix e 2ix sin 2 x dx +
4(i − 2) ∫
. e 2ix e − 2ix sin 2x dx

1 1 2  1
= e − x  e x sin 2 x − e x cos 2x  −
5  5 5  8 i (i + 2 ) ∫
e 2ix e 2ix (e 2ix − e − 2ix ) dx

1
+
8i(i − 2) ∫
e 2ix e − 2ix − e − 2ix ) dx

−x
= (cos 2 x + 2 sin 2 x )
20
x ln cos(nx )
d) y = c1 cos nx + c 2 sin nx + sin(nx ) + . (cos nx )
n n2
e) C.F. = (c1 + c 2 x ) e 2 x + c 3 e −3 x
1
P.I. = X( x )
(D − 2) 2 (D + 3)
 1 1 1 1 1 1 
=  − . + . + . X
 25 D − 2 5 (D − 2)
2
25 D + 3 
1 1 1 −3x 3 x
25 ∫ 5 ∫∫
= − e 2 x e − 2 x X dx + e 2 x e − 2 x X(dx ) 2 +
25
e e X dx ∫
1 2 x −2 x 1
Hence, y = (c1 + c 2 x )e 2 x + c 3 e −3 x −
25
e ∫ ∫∫
e Xdx + e 2 x e − 2 x X(dx ) 2
5
1 −3 x 3x
+
25
e ∫e X dx.

E16) a) y = (c1 + c 2 x ) e x + c 3 cos x + c 4 sin x

b) y = (c1 + c 2 x ) e x + (c 3 + c 4 x ) e −2 x
c) y = 2e x sin 2 x

E17) a) C.F. = (c1 + c 2 x + c 3 x 2 )e − x

1
P.I. = 3
e − x (2 − x 2 )
(D + 1)
1 1
= e −x 3
(2 − x 3 ) = e − x 3 (2 − x 2 )
[(D − 1) + 1] D

1 x3  1 x4   x3 x5 
= e −x  2x −  = e −x  x 2 −  = e −x  

D2 3  D  12   3 − 60 
     
x 3
x 
5
Hence, y = (c1 + c 2 x + c 3 x 2 ) e − x + e − x  − .

 3 60 
b) C.F. = c1 cos x + c 2 sin x + c 3 cos 2 x + c 4 sin 2x
x  1 
P.I. =  sin x − sin 2 x 
12  2 
c) y = c1 + c 2 e − x + c 3 e x + c 4 cos x + c 5 sin x + 2 x 2 + 2 x sin x + 3x e x .

x −x 1 1
d) y = c1 e − x + c 2 e x − e − cos x − x 3 − 6 x − e x (cos x − 2 sin x )
2 2 5
40
 d First and Higher Order
e) Let x + 3 = e t , = D , then the given equation becomes Equations
dt
(D 2 − 5D + 6) y = t
A.E. is m 2 − 5m + 6 = 0 ⇒ m = 2, 3
C.F. = c1 e 2 t + c 3 e 3 t = c1 ( x + 3) 2 + c 2 ( x + 3) 3
1 1 5 1
P.I. = 2
t = (1 − D + D 2 ) −1 t
6 − 5D + D 6 6 6
1 5
= (1 + D + L) t
6 6
1 5 1 5 1 5
= t +  = t + = ln ( x + 3) +
6 6 6 36 6 36
5 1
Hence y = c1 ( x + 3) 2 + c 2 ( x + 3) 3 + + ln ( x + 3)
6 6
f) C.F. = c1 cos ax + c 2 sin a x
1 1  1 1 
P.I. = tan ax =  −  tan ax
D +a2 2
2ia  D − i a D + i a 
1 iax −iax 1 −iax iax
=
2ia
.e e ∫ tan ax dx −
2ia
e ∫
e tan ax dx

=
1 iax
2ia
.e [∫
(sin ax − i sec ax + i cos ax ) dx ]
−
1 −iax
2ia
e [∫
(sin ax + i sec ax − i cos ax ) dx ]
1 1 1
= − 2 sin ax cos ax + 2 sin ax cos ax − 2 cos ax ln sec ax + tan ax
a a a
1
= − 2 cos (ax ). ln sec (ax ) + tan (ax )
a
1
Hence y = c1 cos ax + c 2 sin ax − 2 cos (ax ) . ln sec (ax ) + tan (ax ) .
a
—x—

41