A Differential Equation is a mathematical equation that relates some functions with its

derivatives. Differential equations play a prominent role in many disciplines including

engineering, physics, economics and biology. In biology and economics, differential
equations are used to model the behaviour of complex systems. Many fundamental laws of
physics and chemistry can be formulated as differential equations.

In mathematics, differential equations are studied from several different perspectives,

mostly concerned with their solutions, the set of functions that satisfy the equation.

Differential Equation:

An equation involving derivatives of one or more dependent variable with respect to

one or more independent variable is called a Differential Equation.

For example,

is differential equation where is a constant, is an independent variable and is a

dependent variable.

Differential Equation are of two types

1. Ordinary differential equation

2. Partial differential equation

Ordinary differential equation:

A differential equation involving derivatives with respect to a single independent

variable is called an ordinary differential equation.

For example,

= 2 sin , (0, ], > 0

is an ordinary differential equation where is an independent variable and is a dependent

variable.

Partial differential equation:

A differential equation involving derivatives with respect to more than one

independent variable is called an partial differential equation.

For example,

2 2
+ 2 = 0 where = (, ) and (, ) = (0,1) (0,1)
2

is a partial differential equation where the domain of definition is the open unit rectangle of
dimension 2.

Order:

The order of a differential equation is the order of the highest order derivative
occurring in it.

For example,

2 3
+ 5 ( ) 4 =
2

is the second order ordinary differential equation.

Degree:

The degree of a differential equation is the degree of the highest order derivative
which occurs in it, after the differential equation has been made free from radicals and
fractions as far as the derivatives are concerned.
For example,

3
3 2 5
+ ( 2) + ( ) + = 7
3

is the second order differential equation of degree one.

Linear differential equation:

A linear differential equation is that no products of the function () and its

derivatives and neither the function nor its derivative occur to any power other than first
power.

A general form of linear order differential equation is

() + 1 () 1 + + 1 () = () ..(1)

For example,

+ = , (0, ], > 0 is linear in y.

If a differential equation cannot be written in the form (1) then it is called a non-linear

differential equation.

For example,

+ = sin , (0, ], > 0 is a non-linear differential equation.

Homogeneous differential equation:

When () = 0, (1) becomes,

() + 1 () 1 + + 1 () = 0

This equation is called homogeneous.

If () 0, then the equation is called non- homogeneous.

For example,

+ = 0 is a homogeneous equation.

+ 2 = is a non- homogeneous equation.

Initial value problem:

An initial value problem is an ordinary differential equation together with a specified

value (initial condition) of the unknown function at the initial point in the domain of the
solution.

On some interval I containing the point 0 ,

= (, , , 1 )

and (0 ) = 0 , (0 ) = 1 , 1 (2)

where 0 , 1 , 1 are arbitrarily specified real constants.

The values of () and its first 1 derivatives at a single point 0 ,(0 ) = 0 ,

(0 ) = 1 , 1 (0 ) = 1 are called initial conditions.

For example,

= , (0, ], > 0 with (0) = 1 is the first order initial value problem.

Boundary value problem:

A boundary value problem is a differential equation together with a set of additional

constraints prescribed at more than one point, called the boundary conditions.

There are four important kinds of boundary conditions. They are

1. Dirichlet or 1 kind
2. Neumann or 2 kind
3. Robin or 3 kind or Mixed kind

Dirichlet condition:
 The specification of the unknown function at the boundaries of the domain of the
independent variable is known as a dirichlet boundary condition.

Neumann condition:

If the derivative is specified, then this is known as a neumann boundary condition.

Mixed condition:

When the boundary condition is an equation that involves both the value of the
function and the value of its derivative, it is known as a mixed boundary condition.

Example:

Consider the 2 order linear differential equation

" + () + () = (), <<

Dirichlet or 1 kind : () = 1 and () = 2

Neumann or 2 kind : () = 1 and () = 2

Robin or 3 kind or Mixed kind :1 () + 2 () = 1 and 1 () + 2 () = 2

Well posed problem:

A boundary or initial value problem is said to be well-posed if

a. there exists a solution

b. the solution is unique
c. the solution is stable (i.e) the solution depends continuously on the given data.

Solution of a differential equation:

Any relation between the dependent and independent variables, when substituted
in the differential equation, reduces it to an identity is called a solution or integral of the
differential equation
Methods of solving differential equations:

Some of the analytical methods to get an exact solution are

Variable separable method
The method of undetermined co-efficients
The method of variation of parameters
Power series method and
Laplace transform methods.

In reality, engineering problems are often formulated in terms of an ordinary (or) a

partial differential equation. For example, the mathematical formulation of a falling body
involves an ordinary differential equation and the problem of determining the steady-state
distribution of temperature on a heated plate is formulated in terms of a partial differential
equation. In most cases, obtaining exact solutions are complicated and a numerical method
has to be adopted to solve those complicated mathematical models of the real world.

The numerical methods for the solution of a differential equation are the algorithms
which will produce a table of approximate values of the solution at certain equally spaced
points called grid, nodal (or) mesh points defined on the domain.

The numerical methods for finding solution of the initial value problem of equation

= (, ), (0 ) = 0 may be broadly classified into the following two types.

1. Single step methods

2. Multi step methods

Single step methods:

In single step methods, the solution at any points is obtained using the solution at only
the previous point.
Thus, a general single step method can be written as

+1 = + (+1 , , +1 , , )
Some of the important methods are
Euler methods
Runge kutta methods
Implicit Runge kutta methods
Extrapolation methods

Multi step methods:

These methods are based on recurrence relations, which express the function value
() at +1 in term of the values of the function () and the derivative () at +1 and
at previous mesh points.

Some of the important methods are

Predictor - corrector methods
Implicit multistep methods
Explicit multistep methods
Hybrid methods

Sources of Errors:

The solution of a problem obtained by a numerical method contains some errors. To

minimize the errors, it is most essential to identify the causes of the sources of the errors and
their growth and the propagation in numerical computation.

The difference between true value and approximate value is called error.

In order to determine the accuracy in an approximate solution to a problem, either we

find the bound of the relative error or of the absolute error; given by

|Error|
Relative error = |True value|

Absolute error = |Error|

Inherent error:

Inherent error is the quantity which is already present in the statement of the problem
before its solution.
 The inherent error arises either due to the simplified assumptions in the mathematical
formulation of the problem or due to the errors in the physical measurements of the
parameters of the problem.

The Round-off Error:

It occurs because of the computing device's inability to deal with certain numbers.
Such numbers need to be rounded off to some near approximation which is dependent on the
word size used to represent numbers of the device.

The Truncation Error:

Often an approximation is used in place of an exact mathematical expression (or)

value. For instance, consider the Taylor series expansion of

3 5 7
sin = + +
3! 5! 7!

Practically we cannot use of all the infinite number of terms in the series for
computing the sine of angle . We usually terminate the process after a certain number of
terms. The error that results due to such a termination (or) truncation is called as truncation
error.

Usually in evaluating logarithms, exponentials, trigonometric functions, hyperbolic

functions etc.., an infinite series of the form =
=0 is replaced by a finite series

= =0 . Thus a truncation error of =

=+1 is introduced in the computation.

Finite difference methods are numerical methods (or) discretization method for
solving differential equations by approximating them with difference equations, in which
finite differences are used to approximate the derivatives.

Three important characteristics of a finite difference scheme:

Consistency:
 A finite difference approximation is considered consistent if by reducing the step size,
the truncation errror terms could be made to approach zero. In that case the solution of the
difference equation would approach the true solution.

Stability:

A finite difference approximation is stable if the errors [truncation, round-off 7 etc.,]

decay as the computation proceeds from one marching step to the next. Stability of a finite
difference approximation is assessed using Von- Neumaan stability analysis.

Convergence:

Convergence means that the solution to the finite difference approximation

approaches the true solution when the mesh is refined.

Perturbation methods:

Perturbation problems depend on a small positive parameter. This parameter affects

the problem in such a way that the solution varies rapidly in some region of the problem
domain and slowly in other parts.

For problems of this type, perturbation methods could be approximate solution to

problems that have no closed form analytical solution.

Depending upon the nature of perturbation, a perturbed problem can be divided into
two categories. They are

Regularly perturbed problem

Singularly perturbed problem

Consider a family of boundary value problems , depending on a small positive

parameter , where 0 < 1 . Under some conditions , a solution y () of (), can be
constructed by the well-known method of perturbation; that is as a power series in with its
first term 0 being the solution of the problem 0 which is obtained by putting = 0 in the
problem (). This series can not often pre- asssumd to uniformly convergence,
particularly for small values of , in the entire interval. When such a power series expansion
converges as 0 uniformly in , it is a regular perturbation problem. When () does
not have a uniform limit in as 0 such a problem is known a singular perturbation
problem.

Singular perturbation problems (SPPs) arise in various fields of science and

engineering which include fluid mechanics, fluid dynamics, quantum mechanics, control
theory, semiconductor device modeling, chemical reactor theory, elasticity, hydrodynamics,
gas porous electrodes theory, etc.

Singularly perturbed differential equation:

Singularly perturbed differential equation is a differential equation that contains a

small positive parameter that pre multiples the highest order derivative.

Due to the presence of the perturbation parameter , the solutions of the singularly
perturbed differential equations and/or their derivatives behave non-smoothly in some portion
of the domain of definition of the problems. In such sub-domains the solutions of the problem
exhibit boundary or interior layers.

For example,

"() + ()() = (), (0,1),

with (0) = , (1) =

where > 0 is a small parameter, and are sufficiently smooth functions, such that
() > 0 on [0,1] is a singularly perturbed boundary value problem of second order.

Many phenomena in physics, biology, chemistry, engineering, bionomics and so on,

can be described by singular perturbation problems associated with various types of
differential equations.

These singularly perturbed problems arise in the modeling of various modern

complicated processes, such as fluid flow at high Reynolds numbers, chemical reactor theory,
electromagnetic field problem in moving media, electro-analytical chemistry, water quality
problems in rivers networks, convective heat transport problem with large Peclet numbers,
drift diffusion equation of semiconductor device modeling, financial modeling of option
pricing, turbulence model, simulation of oil extraction from under-ground reservoirs, theory
of plates and shells, atmospheric pollution, ground-water transport etc.
Singular perturbation problems are of two types. They are Convection-diffusion type
and Reaction-diffusion type.

Convection-diffusion:

A singular perturbation problem is said to be of convection-diffusion type if the order

of the differential equation is reduced by one when the perturbation parameter is set equal
to zero.

For example,

() + () = 0, (0,1)

with (0) = 0, (1) = 1

is a convection diffusion problem.

Reaction-diffusion:

A singular perturbation problem is said to be of reaction-diffusion type if the order of

the differential equation is reduced by two when the perturbation parameter is set equal to
zero.

For example,

"() + ( + sin 3 )() = cos + 2 + 1, (0,1)

with (0) = 0, (1) = 2

is a reaction-diffusion problem.

Boundary layer:

A boundary layer is a term borrowed from physics. A boundary layer is defined to be

a region of the independent variable over which the dependent variable changes rapidly.
Consider the following singularly perturbed initial value problem on the unit interval
= (0,1)

1 () such that (0) = 0

and, for all , () + () = 0

where 0 is some given constant and the singular perturbation parameter is

assumed to be non-negative. Normally, is considered over a finite range of values, such as
0 < 1. When, 1 the problem is singularly perturbed and a boundary layer may
appear depending on the given value 0 . Since this problem is so simple the precise situation
may be determined by solving the problem explicitly. When > 0 the exact solution of the
problem is

() = 0 , if > 0

Putting = 0 in the differential equation gives the reduced trivial differential

equation 0 () = 0 for all . Since its solution 0 () 0 is already completely
determined, it is not possible to impose any initial condition at = 0. This singularly
perturbed initial value problem is a simple example of the kind of problem that can arise, for
example, in models of chemical reactions if there is a fast reaction rate or in electrical circuits
if there is a fast transient.

The solution of the problem, when > 0, has the value 0 at the point = 0 due
to the initial condition, whereas the solution of the problem, when = 0, has the value 0 at
the point = 0. It follows therefore that these differ in all cases except for the case when
0 = 0. Excluding this case, it follows that there is a boundary layer in a small
neighbourhood of = 0 when 0 < 1. The problem corresponding to the value = 0 is
known as the reduced problem and its solution as the reduced solution.
Linear reaction-diffusion equation:

Consider a problem for a second order differential equation with a first order
derivative term on the unit interval = (0,1)

2 () such that (0) = 0 , (1) = 1

and, for all , () + () = 0

where 0 , 1 are given constants and < 1.

The exact solution of this self adjoint problem is a linear combination of the exponential
(1)

functions { , }. The reduced differential equation is of zero order and consequently

no boundary conditions may be imposed on its solution. Its exact solution is the trivial one
0 = 0.

Therefore a boundary layer should be expected at = 0, unless the boundary value 0 of

(0) agrees with the value of the reduced solution 0 at = 0, hat is unless 0 = 0.
Likewise there will be a boundary layer at = 1 unless 1 = 0.

Because the exponential functions in the solution have the argument , the solution

changes rapidly in the subinterval(0, ) but not for (, 1). The above discussion has
shown that at most two boundary layers of width can be expected in the solution of this
problem, but for special choices of the boundary conditions one or no boundary layers may
occur. The boundary layers are much thicker than before (because whenever
0 < 1) so numerical difficulties will arise only if is much smaller than in the previous
problem. This kind of problem arises when only diffusion processes are present.

Linear convection-diffusion equation:

Consider a problem for a second order differential equation with a first order
derivative term on the unit interval = (0,1)

2 () such that (0) = 0 , (1) = 1

and, for all , () + () = 0

where 0 , 1 are given constants and < 1.

The exact solution of this non-self adjoint problem is a lilnear combination of the fuctions
(1)
{1, }. The differential equation of the reduced problem is of first order and hence just

one boundary condition can be imposed. It is not immediately obvious which of the two
possible boundary conditions should be imposed. Since its characteristic direction is along
the -axis in the positive sense this shows that no boundary condition may be imposed at the
point = 1. The reduced problem is thus

0 1 () such that 0 (0) = 0

and, for all , 0 () = 0

and its solution is 0 () 0 . Therefore a boundary layer can be expected at = 1 unless

the boundary value of at = 1 agrees with the value of the reduced solution 0 at = 1.
In this case therefore no boundary layer occurs if 1 = 0

Because the exponential functions in the solution have the argument , the boundary

layer is of width . The above considerations have shown that at most one boundary layer can
occur in . When it occurs it is located in a neighbourhood of = 1 and it is of the width .
Problems of this type arise when linear convection and diffusion processes are
simultaneously present.

Methods of solving singular perturbation problem:

Most of the SPPs exclude exact solution or closed form solutions. Hence one has to
look for methods to get good approximation of the actual solutions. There are two major
methods of getting approximate solutions to these problems. They are

Asymptotic methods
Numerical methods

Numerical Methods:

Classical numerical methods do not give satisfactory results when singular

perturbation parameter gets nearer to zero. And, the point-wise errors in numerical methods
based on centered or up-winded differences on uniform meshes depend inversely on a power
of .

In early days, singularly perturbed differential equations were solved numerically by

using a standard finite difference operator on a uniform mesh and then refining the mesh
more in order to capture the boundary and interior layers as the singular perturbation
parameter is decreased in magnitude. Thus, the methods were found to be insufficient even
for a problem in one dimension. So the methods were constructed in such a way that it should
behave uniformly irrespective of how small the perturbation parameter is.

Methods obtained by replacing standard finite difference operator by a finite

difference operator which reflects the singular perturbation nature of the differential operator
are referred as fitted operator methods.

The parameter-uniform methods that use appropriate meshes condensing at layer

regions are referred as fitted mesh methods.

Of all the numerical methods suggested for SPPs, the most popular methods based on
finite differences are fitted operator methods and fitted mesh methods. A fitted mesh method
uses a classical finite difference operator on a piecewise-uniform mesh fitted on the domain
of definition of the differential equation.
Delay differential equation:

Delay differential equations are a type of differential equation in which the derivative
of the unknown function at a certain time is given in terms of the value of the function at
previous times.

Delay differential equations (DDEs) with constant lags > 0 for = 1, , have
the form

() = (, (), ( 1 ), , ( )).

An initial value () is not enough to define a unique solution of the above equation on an
interval . The function () = () must be specified for t a so that
( ) is defined when + .The function () is called the history of the
solution.

For example,

"() + ()() ()() ()( 1) = (), (, 2)

with () = (), [1,0], (2) = 1

is a delay problem with delay at = 1.

Delay differential equations are widely used as mathematical models in many fields,
engineering, physics, etc. Delay models are becoming more common, appearing in many
branches of biological modelling.

They have been used for describing several aspects of infectious disease dynamics:
primary infection [1], drug therapy [2] and immune response [3], to name a few. Delays have
also appeared in the study of chemostat models [4], circadian rhythms [5], epidemiology [6],
the respiratory system [7], tumor growth [8] and neural networks [9].

Statistical analysis ecological data ([10], [11]) has shown that there is evidence of
delay effects in the population dynamics of many species.
Singularly perturbed delay differential equation:

A singularly perturbed delay differential equation is a differential equation in which

the highest derivative is multiplied by a small positive parameter and involving at least one
delay term.

For example,

"() + ( ) () = 0, (0,1)

with () = 1, < 0 , (1) = 1

is a singularly perturbed delay differential equation with a history function near = 0 and a
terminal value.

The solution of the singularly perturbed delay differential equations exhibit initial and
interior layers due to the presence of the singular perturbation parameter and the presence
of the delay term.

Such types of differential equations arise frequently in the first exit time problem in modeling
of the activation of neuronal variability [12] in a variety of models for physiological
processes or diseases [13] to describe the human pupil-light reflex [14] variational problems
in control theory and depolarization in Steins model [15].

Shishkin mesh:

Shishkin mesh is a piecewise uniform mesh. Piecewise uniform fitted mesh is a union
of a finite number of uniform meshes with mesh transition parameters. It was first introduced
by Shishkin. The main difference between a Shishkin mesh and any other piecewise uniform
mesh is the choice of the so-called transition parameter, which is the one at which the mesh
size changes.

Most of the researchers use piecewise uniform Shishkin meshes to solve singular
perturbation problems numerically.
The Shishkin mesh have some interesting features.

The mesh consists of a transition parameter which is defined so as to separate the layer
region and the outer region.
Layer region is a region where the solution varies fastly and outer region is a region
where the solution smoothly.

The mesh is dense inside the layer region where more information is needed and
coarse in the outer region.

The mesh is uniform inside the layer region and inside the outer region

The mesh is so defined that it becomes a uniform mesh when the transition parameter
assumes a location that is away from the boundary.
Literature survey:

A brief literature survey of works related with SPDEs is presented.

Investigation of boundary value problems for singularly perturbed linear second-order

differential difference equations was initiated by Lange and Miura [12, 16,17]; they proposed
an asymptotic approach in study of linear second-order differential-difference equations in
which the highest order derivative is multiplied by small parameters.

An initial value problem for a system of singularly perturbed ordinary differential equations
is considered in [18]. A parameter robust computational method is constructed and it is
proved that it gives essentially first order parameter-uniform convergence in the maximum
norm.

A boundary value problem for a second-order singularly perturbed delay differential equation
is considered in [19]. The solution of this problem exhibits boundary layers at = 0 and
= 2 and interior layers at = 1. A numerical method composed of a classical finite
difference scheme applied on a piecewise-uniform Shishkin mesh is suggested to solve the
problem. The method is proved to be first-order convergent in the maximum norm uniformly
in the perturbation parameter.

A singularly perturbed linear system of second order ordinary differential equation of

reaction-diffusion type with given boundary conditions is considered in [20]. The leading
term of each equation is multiplied by a small positive parameter. These singular perturbation
parameters are assumed to be distinct. The components of the solution exhibit overlapping
layers. Shishkin piecewise uniform meshes are introduced, which are used in conjunction
with a classical finite difference discretization, to construct a numerical method for solving
this problem. It is proved that the numerical approximations obtained with this method are
essentially second order convergent uniformly with respect to all of the parameters.

In [21], a class of delay differential equations with a perturbation parameter is examined. A

hybrid finite difference scheme on an appropriate piecewise uniform mesh of Shishkin-type is
derived. It is shown that the scheme is almost second-order convergent, in this discrete
maximum norm, independent of singular perturbation parameter.
In [22] Munyakazi and Patidar proposed a fitted operator finite difference method on uniform
mesh to solve the system of singularly perturbed reaction-diffusion equations and the same
idea was extended and proposed a nonstandard finite difference scheme by Munyakazi [23]
for a system of convectiondiffusion equations to get the first order uniform convergence.

In [24], a coupled system of two singularly perturbed convection-diffusion ODEs is

examined. A numerical method is constructed for this system which involves an appropriate
piecewise-uniform Shishkin mesh. The numerical approximations are shown to converge to
the continuous solutions uniformly with respect to the singular perturbation parameters.

In the paper [25], a system of two coupled singularly perturbed convection-diffusion ordinary
differential equations is examined. The equations are coupled through their convective terms.
A numerical method consisting of simple upwinding and an appropriate piecewise-uniform
Shishkin mesh are shown to generate numerical approximations that are essentially first order
convergent, uniformly in the small parameter, to the true solution in the discrete maximum
norm.

In [26], the authors presented some uniformly convergent non standard finite difference
methods for solving class of singularly perturbed differential difference equations where
there is small delay in the convection term.

In [27], the author presented Uniform finite difference methods constructed via non standard
finite difference methods for the numerical solution of singularly perturbed quasilinear initial
value problem for delay differential equations. A numerical method is constructed for this
problem which involves the appropriate Bakhvalov meshes on each time subinterval. The
method is shown to be uniformly convergent with respect to the perturbation parameter.

In [28], the author analyzed the upwind finite difference scheme on arbitrary meshes to solve
the system of singularly perturbed convectiondiffusion equations in which he considered
systems with arbitrary number of equations.

ndhum
Motivated by these works, in this dissertation, a singularly perturbed delay problem with
discontinuous source term has been treated numerically. Computer codes are developed for
the numerical method suggested and convergence of the numerical approximations are
illustrated by means of examples.

Statement of the problem:

Consider a following BVP for singularly perturbed convection-diffusion type second

order delay differential equation with discontinuous source term

"() + ()() ()() ()( 1) = (), ((0, ) (, 2))

Subject to the condition

() = (), [1,0] , (2) = 1

1 (), [0, )
() = {
2 (), (, 2]

where 0 < < 1 , () > 0 and (+) ().

Let = (0,2), = (0, ), = [0,2].

+ = (, 2). Then (), (), () are
sufficiently smooth functions on [0,2]. () is a sufficiently smooth functions on \{}. A
simple discontinuity is assumed to occur at a point (0,2). () is a smooth function on
[-1,0] and 1 is a given constant which is independent of .

The cases (i) (0,1) (ii) (1,2) and (iii) = 1 are dealt with
separately.

In case (i), the above BVP exhibits a strong boundary layer in the neighborhood of
= 0 and due to the presence of delay term an interior layer is exhibited to the right of =
1. Also due to the presence of the point of discontinuity = (0,1), an interior layer is
exhibited to the right of (0,1) and due to the presence of delay term an interior layer is
exhibited to the right of = 1 + (1,2)

In case (ii) , the above BVP exhibits a strong boundary layer in the neighbourhood of
= 0 and due to the presence of delay term an interior layer is exhibited to the right of =
1. Also (1,2). So that an interior layer is exhibited to the right of = (1,2).

In case (iii), the above BVP exhibits a strong boundary layer in the neighbourhood of
= 0 and due to the presence of the delay term an interior layer is exhibited to the right of
= 1.

A numerical method composed of a classical finite difference scheme applied on a

piecewise uniform Shishkin mesh is suggested to solve the problem.
The Shishkin mesh:

A fitted mesh for the above problem is now described.The Shishkin mesh =

= [0,2] as follows.
{ } =0 is constructed on

Case (i): (0,1)

Therefore the domain

is subdivided into 8 sub intervals

[0, ] (, ] (, + ] ( + , 1] (1,1 + ] (1 + , 1 + ]

(1 + , 1 + + ] (1 + + , 2]

The parameters and which denote the points separating the uniform meshes, are
defined by

1
= min{ 2 , log } and = { , log }
2

Then on each sub intervals [0, ], (, ], (, + ], ( + , 1], (1,1 + ], (1 + , 1 +

], (1 + , 1 + + ] and(1 + + , 2], a uniform mesh of mesh points is placed.
8

As layers occur to the right of = 0, = , = 1 and = 1, the mesh should be fine in

these layer regions and coarse elsewhere.

Further we denote the mesh size in the region, [0, ] & [1,1 + ] by 1 , [, ] &[1 +

, 1 + ] by 2 , [, + ] & [1 + , 1 + + ] by 3 , [ + , 1] & [1 + + , 2] by

4

The Shishkin mesh = { } =0 for the case (ii) (1,2) is same as that for the case(i).
Case (iii) = 1

= [0,2] as follows.
Therefore, the Shishkin mesh = { } =0 is constructed on

The domain
is subdivided into 4 sub intervals

[0, ] (, 1] (1,1 + ] (1 + , 2]

The parameter which denotes the point separating the uniform meshes, is defined by

= min{ 2 , log }

Then on each sub intervals [0, ], (, 1], (1,1 + ] and (1 + , 2] a uniform mesh of 4

mesh points is placed.

Further we denote the mesh size in the region, [0, ] & [1,1 + ] by 1 and [, 1] &

[1 + , 2] by 2

Finite difference method:

If the mesh points of an arbitrary non uniform mesh with N sub intervals

= (1 , ), 1 are denoted by = { } =0 , then the width of the sub

intervals are written as = 1 for 1

The first and second order finite differences are defined by

+1
+ = +1

1
=
 ( + - )
2 =

(+1 + )
Where, for 1 1 and = 2

The discrete problem:

The BVP is discretized using the finite difference scheme on the piecewise uniform
. The discrete problem for the case(i) is
mesh

2 + ( )+ ( )( ) ( )( 1) = ( ) , 1

With + () = (),

(+ )
+ ( )+ ( )( ) ( )( 1) = ( )

 +1 1
( ) +1
+1
+ () ( ) ( )( ) ( )( 1) = ( )
+1

+1 +1 + 1 +1
( ) +1
+1
+ () ( ) ( )( ) ( )( 1)
+1 + +1
2

= ( )

2 +1 +1 + 1 +1 +1
( ) + () ( ) ( )
+ +1 +1 +1

( )( 1) = ( )

2 2 ( ) 2 ( )
( ) 1 ( + + ( )) + ( + ) +1
( + +! ) +1 +1 +1 ( + +! ) +1

( )( 1) = ( ) . ()

Putting = 1 in (),one can get

2 (1 ) (1 )
( 2 ) (0 ) ( 2 + + (1 )) (1 ) + ( 2 + ) (2 ) (1 )(1 1)
1 1 1 1 1

= (1 )

2 3 4 5 6 7
Let 1 = 8 , 2 = , 3 = , 4 = , 5 = , 6 = , 7 = ,
8 8 8 8 8 8

putting = 2,3, 1 1 in (), one can get

 2 ( ) ( )
( 2 ) (1 ) + ( 2 + + ( )) ( ) + ( 2 ) (+1 ) + ( )( 1)
1 1 1 1 1

= ( )

Putting = 1 in (*),one can get

2 2 (1 )
( ) (1 1) + ( + + (1 )) (1 )
1 (1 + 2 ) 1 2 2

2 (1 )
+( ) (1 + 1) + (1 )(1 1) = (1 )
2 (1 + 2 ) 2

Putting = 1,2, 1 1 in (), one can get

2 (1 + )
( 2 ) (1 +1 ) + ( 2 + + (1 + )) (1 + )
2 2 2

(1 + )
+( 2 ) (1++1 ) + (1 + )(1 + 1) = (1 + )
2 2

At = , the continuity of U in the discrete sence is stated as

(2 ) = + (2 )

1 1 1 1
( ) (21 ) + (( ) + ( )) (2 ) + ( ) (2+1 ) = 0
2 2 3 3

Putting = 1,2, 1 1 in () ,one can get

 2 (2 + )
( 2 ) (2+1 ) + ( 2 + + (2 + )) (2+ )
3 3 3

(2 + )
+( 2 ) (2++1 ) + (2 + )(2 ++1 ) = (2 + )
3 3

putting = 3 in() one can get

2 2 (3 )
( ) (31 ) + ( + + (3 )) (3 )
3 (3 + 4 ) 3 4 4

2 (3 )
+( ) (3 +1 ) + (3 )(3 1) = (3 )
4 (3 + 4 ) 4

putting = 1,2, 1 1in (), one can get

2 (3 + ) (3 + )
( 2 ) (3+1 ) + ( 2 + + (3 + )) (3+`) + ( 2 ) (3+1 )
4 4 4 4 4

+ (3 + )(3 + 1) = (3 + )

At = 1, the continuity of U in the discrete sence is stated as

(4 ) = + (4 )

1 1 1 1
( ) 4 1 + (( ) + ( ) + (4 )) 4 + ( ) 4 +1 = 0
4 4 1 1

putting = 1,2, 1 1 in() one can get

 2 (4 + )
( 2 ) (4 +1 ) + ( 2 + + (4 + )) (4 + )
1 1 1

(4 + )
+( 2 ) (4 ++1 ) + (4 + )(4 + 1) = (4 + )
1 1

putting = 5 in (), one can get

2 2 (5 )
( ) (5 1 ) + ( + + (5 )) (5 )
1 (1 + 2 ) 1 2 2

2 (5 )
+( ) (5 +1 ) + (5 )(5 1) = (5 )
2 (1 + 2 ) 2

putting = 1,2, 1 1 (*) one can get

2 (5 + )
( 2 ) (5 +1 ) + ( 2 + + (5 + )) (5 + )
2 2 2

(5 + )
+( 2 ) (5 ++1 ) + (5 + )(5 + 1) = (5 + )
2 2

putting = 6 in () one can get

2 2 (6 )
( ) (6 1 ) + ( + + (6 )) (6 )
2 (2 + 3 ) 2 3 3

2 (6 )
+( ) (6 +1 ) + (6 )(6 1) = (6 )
3 (2 + 3 ) 3
putting = 1,2, 1 1 (*), one can get

2 (6 + )
( 2 ) (6 +1 ) + ( 2 + + (6 + )) (6 + )
3 3 3

(6 + )
+( 2 ) (6 ++1 ) + (6 + )(6 + 1) = (6 + )
3 3

Putting = 7 in () one can get

2 2 (7 )
( ) (7 1 ) + ( + + (7 )) (7 )
3 (3 + 4 ) 3 4 4

2 (7 )
+( ) (7 +1 ) + (7 )(7 1) = (7 )
4 (3 + 4 ) 4

putting = 1,2, 1 2 (*), one can get

2 (7 + )
( 2 ) (7 +1 ) + ( 2 + + (7 + )) (7 + )
4 4 4

(7 + )
+( 2 ) (7 ++1 ) + (7 + )(7 + 1) = (7 + )
4 4

Putting = 1 in (), one can get

2 (1 ) (1 )
( 2 ) (2 ) + ( 2 + + (1 )) (1 ) + ( 2 ) ( )
4 4 4 4 4

+ (1 )(1 1) = (1 )
The discrete problem for the case (ii) (1,2) is same as that for the case(i).

Now, the discrete problem for the case (iii) is

2 ( )+ + ( )( ) + ( )( 1) = ( )

(+ )
( ) + + ( )( ) + ( )( 1) = ( )

+1 1
( ) +1
+1
() ( ) + ( )( ) + ( )( 1)
+1

= ( )

+1 +1 + 1 +1
( ) +1
+1
() ( ) + ( )( )
+1 + +1
2

+ ( )( 1) = ( )

2 +1 +1 + 1 +1 +1
( ) () ( ) + ( )( )
+ +1 +1 +1

+ ( )( 1) = ( )

2 2 ( ) 2 ( )
( ) 1 + ( + + ( )) + ( ) +1
( + +! ) +1 +1 +1 ( + +! ) +1

+ ( )( 1) = ( )

2 3
Let 1 = 4 , 2 = 4
, 3 = 4
,
Putting = 1 in (),one can get

2 (1 ) (1 )
( 2 ) (0 ) + ( 2 + + (1 )) (1 ) + ( 2 ) (2 ) + (1 )(1 1)
1 1 1 1 1

= (1 )

putting = 2,3, 1 1 in (), one can get

2 ( ) ( )
( 2 ) (1 ) + ( 2 + + ( )) ( ) + ( 2 ) (+1 ) + ( )( 1)
1 1 1 1 1

= ( )

Putting = 1 in (*),one can get

2 2 (1 )
( ) (1 1) + ( + + (1 )) (1 )
1 (1 + 2 ) 1 2 2

2 (1 )
+( ) (1 + 1) + (1 )(1 1) = (1 )
2 (1 + 2 ) 2

Putting = 1,2, 1 1 in (), one can get

2 (1 + )
( 2 ) (1+1 ) + ( 2 + + (1 + )) (1 + )
2 2 2

(1 + )
+( 2 ) (1 ++1 ) + (1 + )(1 + 1) = (1 + )
2 2
At = 1, the continuity of U in the discrete sence is stated as

(4 ) = + (4 )

1 1 1 1
(1/2 )(2 1 ) + (( ) + ( ) + (2 )) (2 ) + (( ) + ( )) (2 +1) = 0
1 2 1 2

putting = 1,2, 1 1 in (), one can get

2 (2 + )
( 2 ) (2 +1 ) + ( 2 + + (2 + )) (2 + )
1 1 1

(2 + )
+( 2 ) (2 ++1 ) + (2 + )(2 + 1) = (2 + )
1 1

putting = 3 in (), one can get

2 2 (3 )
( ) (3 1 ) + ( + + (3 )) (3 )
1 (1 + 2 ) 1 2 2

2 (3 )
+( ) (3 +1 ) + (3 )(3 1) = (3 )
2 (1 + 2 ) 2

putting = 1,2, 1 2 in (), one can get

 2 (3 + )
( 2 ) (3 +1 ) + ( 2 + + (3 + )) (1 + )
2 2 2

(3 + )
+( 2 ) (3 ++1 ) + (3 + )(3 + 1) = (3 + )
2 2

putting = 1 in (), one can get

2 (1 ) (1 )
( 2 ) (2 ) + ( 2 + + (1 )) (1 ) + ( 2 ) ( )
2 2 2 2 2

+(1 )(1 1) = (1 )
References:

1. S. M. Ciupe, B. L. de Bivort, D. M. Bortz, and P. W. Nelson. Estimates of kinetic

parameters from HIV patient data during primary infection through the eyes of three
different models. Math. Biosci. in press.

2. P. W. Nelson, J. D. Murray, and A. S. Perelson. A model of HIV-1 pathogenesis that

includes an intracellular delay. Math. Biosci., 163:201215, 2000.

3. K. Cooke, Y. Kuang, and B. Li. Analyses of an antiviral immune response model with
time delays. Canad. Appl. Math. Quart., 6(4):321354, 1998.

4. T. Zhao. Global periodic solutions for a differential delay system modeling a

microbial population in the chemostat. J. Math. Anal. Appl., 193:329352, 1995.

5. P. Smolen, D. Baxter, and J. Byrne. A reduced model clarifies the role of feedback
loops and time delays in the Drosophila circadian oscillator. Biophys. J., 83:2349
2359, 2002.

6. K. L. Cooke, P. van den Driessche, and X. Zou. Interaction of maturation delay and
nonlinear birth in population and epidemic models. J. Math. Biol., 39:332352, 1999.

7. B. Vielle and G. Chauvet. Delay equation analysis of human respiratory stability.

Math. Biosci. 152(2):105122, 1998.

8. M. Villasana and A. Radunskaya. A delay differential equation model for tumor

growth. J. Math. Biol., 47(3):270294, 2003.

9. S. A. Campbell, R. Edwards, and P. van den Driessche. Delayed coupling between

two neural network loops. SIAM J. Appl. Math., 65(1):316335, 2004.
10. P. Turchin. Rarity of density dependence or population regulation with lags. Nature,
344:660 663, 1990.

11. P. Turchin and A. D. Taylor. Complex dynamics in ecological time series. Ecology,
73:289305, 1992.

12. C. G. Lange and R. M. Miura, Singular perturbation analysis of boundary value

problems for differential-difference equations. V: small shifts with layer behavior,
SIAM Journal on Applied Mathematics, vol. 54, no. 1, pp. 249272, 1994.

13. M. C. Mackey and L. Glass, Oscillations and chaos in physiological control system,
Science, vol. 197, pp. 287289, 1997.

14. Longtin and J. G. Milton, Complex oscillations in the human pupil light reflex with
mixed and delayed feedback, Mathematical Biosciences, vol. 90, no. 1-2, pp. 183
199, 1988.

15. V. Y. Glizer, Asymptotic analysis and solution of a finite-horizon H control

problem for singularly-perturbed linear systems with small state delay, Journal of
Optimization Theory and Applications, vol. 117, no. 2, pp. 295325, 2003.

16. C. G. Lange and R. M. Miura, Singular perturbation analysis of boundary value

problems for differential-difference equations, SIAM Journal on Applied
Mathematics, vol. 42, no. 3, pp. 502531, 1982.

17. C. G. Lange and R. M. Miura, Singular perturbation analysis of boundary value

problems for differential-difference equations.VI: small shifts with rapid oscillations,
SIAM Journal on Applied Mathematics, vol. 54, no. 1, pp. 273283, 1994.
18. S. Hemavathi, T. Bhuvaneswari, S. Valarmathi, and J.J.H. Miller. A parameter-
uniform numerical method for a system of singularly perturbed ordinary differential
equation. Applied Mathematics and Computation, 191:111, August 2007.

19. M. Manikandan, N.Shivaranjani, J.J.H.Miller, and S.Valarmathi: A Parameter-

Uniform Numerical Method for a Boundary Value Problem for a Singularly Perturbed
Delay Differential Equation.

20. M.Paramasivam, S.Valarmathi and J.J.H.Miller: Second order parameter-uniform

convergence for a finite difference method for a singularly perturbed linear reaction-
diffusion system Math.Commun., Vol. 15, No. 2, pp. 587-612(2010) .

21. Cen,Z. : A Hybrid Finite Difference Scheme for a Class of Singularly Perturbed
Delay Differential Equations. Neural, Parallel, and Scientific Computations 16 303-
308 (2008).

22. InMunyakazi, J.B., Patidar, K.C.: An efficient numerical method for a system of
singularly perturbed reactiondiffusion equations. Neural Parallel Sci. Comput. 20,
173190 (2012)

23. Munyakazi J.B.: A uniformly convergent nonstandard finite difference scheme for a
system of convection diffusion equations, Comp. Appl. Math. (2014). doi:
10.1007/s40314-014-0171-6

24. S. Bellew and E. ORiordan. A parameter robust numerical method for a system of
two singularly perturbed convection-diffusion equations. Applied Numerical
Mathematics, 51 (2-3):171186, 2004.

25. Eugene ORiordan and Martin Stynes. Numerical analysis of a strongly coupled
system of two singularly perturbed convection-diffusion problems. Adv. Comput.
Math., 30 (2):101121, 2009.
26. Patidar, K. C. and Sharma, K. K. 2006. e -Uniformly Convergent Non-Standard
Finite Difference Methods for Singularly Perturbed Differential Difference
Equations with Small Delay. Applied Mathematics and Computation,175, 864-890.

27. Fevzi Erdogan , A Parameter Robust Method for Singularly Perturbed Delay
Differential Equations 10.1155/2010/325654.

28. Lin, T.: Analysis of an upwind finite-difference scheme for a system of coupled
singularly perturbed convectiondiffusion equations. Computing 79, 2332 (2007)