BIJU PATNAIK UNIVERSITY OF TECHNOLOGY,

ODISHA

Lecture Notes
On
DIFFERENCE EQUATION
AND
ITS APPLICATION

Prepared by,
Dr. Subhendu Kumar Rath,
BPUT, Odisha.
DIFFERENCE EQUATION AND
ITS APPLICATION

 Dr.Subhendu Kumar Rath

DEFINITION
An expression which expresses a
relation between an independent
variable and successive values or
Successive differences of dependent
variable is called a difference equation.
EXAMPLES

ORDER AND DEGREE

SOLUTION
 A solution of a difference equation is any
function that satisfies it.
 The general solution of a difference
equation of order n is a solution that
contains n arbitrary constants or n
arbitrary function which are periodic with
period equal to the interval of differencing.
CONTD…..
 The particular solution of a difference
equation is a solution obtained by
assigning particular values to the arbitrary
constants or functions.
LINEAR DIFFERENCE EQATION

CONTD….

APPLICATION
 Difference equation can be applicable in
the following areas.
 Numerical methods to solve partial
differential equation.
 Fourier series
 Algebra and Analysis
First Derivative Approximations
 Backward difference: (uj – uj-1) / Δx

 Forward difference: (uj+1 – uj) / Δx

 Centered difference: (uj+1 – uj-1) / 2Δx

 Taylor Expansion
u(x + Δx) = u(x) + u΄(x)Δx + 1/2 u˝(x)(Δx)
2

+ 1/6 u˝΄(x)(Δx) + O(Δx)
3 4

u(x – Δx) = u(x) – u΄(x)Δx + 1/2 u˝(x)(Δx)

2

- 1/6 u˝΄(x)(Δx) + O(Δx)
3 4
Taylor Expansion
u΄(x) = u(x) – u(x – Δx) + O(Δx)
Δx
u΄(x) = u(x + Δx) – u(x) + O(Δx)
Δx
u΄(x) = u(x + Δx) – u(x – Δx) + O(Δx) 2

2Δx
Second Derivative Approximation
Centered difference: (uj+1 – 2uj + uj-1) / (Δx)
2


 Taylor Expansion
2
u˝(x) = u(x + Δx) – 2u(x) + u(x – Δx) + O(Δx)
(Δx)
2
Function of Two Variables
u(jΔx, nΔt) ~ uj
n

 Backward difference for t and x

∂u (jΔx, nΔt) ~ (ujn – ujn-1) / Δt

∂t

∂u (jΔx, nΔt) ~ (ujn – ujn-1) / Δx

∂x
Function of Two Variables

 Forward difference for t and x

∂u (jΔx, nΔt) ~ (ujn+1 – ujn ) / Δt

∂t
∂u (jΔx, nΔt) ~ (ujn+1 – ujn ) / Δx
∂x
Function of Two Variables

 Centered difference for t and x

∂u (jΔx, nΔt) ~ (ujn+1 – ujn-1) / (2Δt)

∂t
∂u (jΔx, nΔt) ~ (ujn+1 – ujn-1) / (2Δx)
∂x
 Partial Differential
Equations
 Partial Differential Equations (PDEs).
 What is a PDE?
 Examples of Important PDEs.
 Classification of PDEs.
 Partial Differential Equations
A partial differential equation (PDE) is an
equation that involves an unknown function
and its partial derivatives.

Example :
∂ 2 u( x, t ) ∂ u( x, t )
=
∂x 2
∂t
PDE involves two or more independent variables
(in the example x and t are independent variables)
 Notation
∂ 2 u ( x, t )
u xx =
∂x 2
∂ 2 u ( x, t )
u xt =
∂x ∂t
Order of the PDE = order of the highest order derivative.
Examples of PDEs
PDEs are used to model many systems in
many different fields of science and
engineering.

Important Examples:
 Laplace Equation
 Heat Equation
 Wave Equation
 Laplace Equation
∂ 2u ( x , y , z ) ∂ 2u ( x , y , z ) ∂ 2u ( x , y , z )
+ + =0
∂x 2
∂y 2
∂z 2

Used to describe the steady state distribution of

heat in a body.
Also used to describe the steady state
distribution of electrical charge in a body.
 Heat Equation
∂ u ( x, y , z , t )  ∂ 2u ∂ 2u ∂ 2u 
= α  2 + 2 + 2 
∂t 
 ∂x ∂y ∂z 

The function u(x,y,z,t) is used to represent

the temperature at time t in a physical body
at a point with coordinates (x,y,z)
α is the thermal diffusivity. It is sufficient to
consider the case α = 1.
 Simpler Heat Equation
∂ T ( x, t ) ∂ T ( x, t )
2
= x
∂t ∂x 2

T(x,t) is used to represent the temperature

at time t at the point x of the thin rod.
 Wave Equation
∂ 2u ( x , y , z , t )  ∂ 2
u ∂ 2
u ∂ 2
u 
= c  2 + 2 + 2
2

∂t 
 ∂x ∂y ∂z
2


The function u(x,y,z,t) is used to represent the

displacement at time t of a particle whose
position at rest is (x,y,z) .
The constant c represents the propagation
speed of the wave.
Classification of PDEs
Linear Second order PDEs are important
sets of equations that are used to model
many systems in many different fields of
science and engineering.

Classification is important because:

 Each category relates to specific engineering
problems.
 Different approaches are used to solve these
categories.
Linear Second Order PDEs
Classification
A second order linear PDE (2 - independent variables)
A u xx + B u xy + C u yy + D = 0,
A, B, and C are functions of x and y
D is a function of x, y , u, u x , and u y
is classified based on (B 2 − 4 AC) as follows :
B 2 − 4 AC < 0 Elliptic
B 2 − 4 AC = 0 Parabolic
B 2 − 4 AC > 0 Hyperbolic
Linear Second Order PDE
Examples (Classification)
∂ 2u ( x , y ) ∂ 2u ( x , y )
Laplace Equation + =0
∂x 2
∂y 2

A = 1, B = 0, C = 1 ⇒ B 2 − 4 AC < 0
⇒ Laplace Equation is Elliptic
One possible solution : u( x, y ) = e x sin y
u x = e x sin y , u xx = e x sin y
u y = e x cos y , u yy = −e x sin y
u xx + u yy = 0
 Linear Second Order PDE
Examples (Classification)
∂ 2u ( x , t ) ∂ u ( x , t )
Heat Equation α − =0
∂x 2
∂t
A = α , B = 0, C = 0 ⇒ B 2 − 4 AC = 0
⇒ Heat Equation is Parabolic
______________________________________
∂ 2
u ( x , t ) ∂ 2
u ( x, t )
Wave Equation c 2
− =0
∂x 2
∂t 2

A = c 2 > 0, B = 0, C = −1 ⇒ B 2 − 4 AC > 0
⇒ Wave Equation is Hyperbolic
 Boundary Conditions for PDEs
 To uniquely specify a solution to the PDE,
a set of boundary conditions are needed.
 Both regular and irregular boundaries are
possible. 2 t
∂ u( x, t ) ∂u( x, t )
Heat Equation : α − =0
∂x 2
∂t region of
u(0, t ) = 0 interest
u(1, t ) = 0
u( x,0) = sin(π x ) 1 x
The Solution Methods for PDEs
 Analytic solutions are possible for simple
and special (idealized) cases only.

 To make use of the nature of the

equations, different methods are used to
solve different classes of PDEs.

 The methods discussed here are based

on the finite difference technique.
Parabolic Equations
 Parabolic Equations
 Heat Conduction
Equation
 Explicit Method
 Implicit Method
 Cranks Nicolson Method
 Parabolic Equations
A second order linear PDE (2 - independent variables x , y )
A u xx + B u xy + C u yy + D = 0,
A, B, and C are functions of x and y
D is a function of x, y, u, u x , and u y

is parabolic if B 2 − 4 AC = 0
 Parabolic Problems
∂ T ( x, t ) ∂ 2 T ( x, t )
Heat Equation : =
∂t ∂x 2
T (0, t ) = T (1, t ) = 0
T ( x,0) = sin(π x ) ice ice
x
* Parabolic problem ( B 2 − 4 AC = 0)
* Boundary conditions are needed to uniquely specify a solution.
 Finite Difference Methods
 Divide the interval x into sub-intervals,
each of width h
 Divide the interval t into sub-intervals,
each of width k t
 A grid of points is used for
the finite difference solution
 Ti,j represents T(xi, tj)
 Replace the derivates by x

finite-difference formulas
 Finite Difference Methods
Replace the derivatives by finite difference formulas
∂ 2T
Central Difference Formula for 2 :
∂x
∂ 2T ( x, t ) Ti −1, j − 2Ti , j + Ti +1, j Ti −1, j − 2Ti , j + Ti +1, j
≈ =
∂x 2
( ∆x ) 2
h2
∂T
Forward Difference Formula for :
∂t
∂T ( x, t ) Ti , j +1 − Ti , j Ti , j +1 − Ti , j
≈ =
∂t ∆t k
Solution of the Heat Equation
• Two solutions to the Parabolic Equation
(Heat Equation) will be presented:
1. Explicit Method:
Simple, Stability Problems.
2. Crank-Nicolson Method:
Involves the solution of a Tridiagonal
system of equations, Stable.
 Explicit Method
∂T ( x, t ) ∂ 2T ( x, t )
=
∂t ∂x 2
T ( x, t + k ) − T ( x, t ) T ( x − h, t ) − 2T ( x, t ) + T ( x + h, t )
=
k h2
T ( x, t + k ) − T ( x, t ) = 2 (T ( x − h, t ) − 2T ( x, t ) + T ( x + h, t ) )
k
h
k
Define λ = 2
h
T ( x, t + k ) = λ T ( x − h, t ) + (1 − 2 λ ) T ( x, t ) + λ T ( x + h, t )
 Explicit Method
How Do We Compute?
T ( x, t + k ) = λ T ( x − h, t ) + (1 − 2 λ ) T ( x, t ) + λ T ( x + h, t )
means

T(x,t+k)

T(x-h,t) T(x,t)
T(x+h,t)
 Convergence and Stability
T ( x, t + k ) can be computed directly using :
T ( x, t + k ) = λ T ( x − h, t ) + (1 − 2 λ ) T ( x, t ) + λ T ( x + h, t )

Can be unstable (errors are magnified )

1 h2
To guarantee stability, (1 − 2 λ ) ≥ 0 ⇒ λ ≤ ⇒ k ≤
2 2
This means that k is much smaller than h
This makes it slow.
 Example 1
∂ 2 u( x, t ) ∂ u( x, t )
− =0
∂x 2
∂t
u ( x − h , t ) − 2u ( x , t ) + u ( x + h , t ) u ( x , t + k ) − u ( x , t )
2
− =0
h k
16(u( x − h, t ) − 2u( x, t ) + u( x + h, t ) ) − 4(u( x, t + k ) − u( x, t ) ) = 0

u ( x , t + k ) = 4 u ( x − h, t ) − 7 u ( x , t ) + 4 u ( x + h, t )
 Example 1
u ( x , t + k ) = 4 u ( x − h, t ) − 7 u ( x , t ) + 4 u ( x + h, t )

t=1.0 0 0
t=0.7 0 0
t=0.5 0 0
t=0.2 0 0
t=0 0 0
Sin(0.2 Sin(0. Sin(0.75
5π) 5π) π)
x=0. x=0. x=0. x=0. x=1.
0 25 5 75 0
 Example 1
u(0.25,0.25) = 4 u(0,0) − 7 u(0.25,0) + 4 u(0.5,0)
= 0 − 7 sin(π / 4) + 4 sin(π / 2) = −0.9497

t=1.0 0 0
t=0.7 0 0
t=0.5 0 0
t=0.2 0 0
t=0 0 0
Sin(0.2 Sin(0. Sin(0.75
5π) 5π) π)
x=0. x=0. x=0. x=0. x=1.
0 25 5 75 0
 Crank-Nicolson Method
The method involves solving a Tridiagonal system of linear equations.
The method is stable (No magnification of error).
→ We can use larger h, k (compared to the Explicit Method).
 Crank-Nicolson Method
Based on the finite difference method
1. Divide the interval x into subintervals of width h
2. Divide the interval t into subintervals of width k
3. Replace the first and second partial derivatives with their
backward and central difference formulas respectively :
∂u( x, t ) u( x, t ) − u( x, t − k )
≈
∂t k
∂ 2 u ( x , t ) u ( x − h , t ) − 2u ( x , t ) + u ( x + h , t )
=
∂x 2
h2
 Crank-Nicolson Method
∂ 2 u ( x, t ) ∂ u ( x, t )
Heat Equation : = becomes
∂x 2
∂t

u ( x − h , t ) − 2u ( x , t ) + u ( x + h , t ) u ( x , t ) − u ( x , t − k )
2
=
h k
k
2
(u( x − h, t ) − 2u( x, t ) + u( x + h, t ) ) = u( x, t ) − u( x, t − k )
h
k k k
− 2 u( x − h, t ) + (1 + 2 2 ) u( x, t ) − 2 u( x + h, t ) = u( x, t − k )
h h h
 Crank-Nicolson Method
k
Define λ = 2 then Heat equation becomes :
h
− λ u( x − h, t ) + (1 + 2λ ) u( x, t ) − λ u( x + h, t ) = u( x, t − k )

u(x-h,t) u(x,t)
u(x+h,t)

u(x,t - k)
Crank-Nicolson Method
The equation :
− λ u( x − h, t ) + (1 + 2λ ) u( x, t ) − λ u( x + h, t ) = u( x, t − k )
can be rewritten as :
− λ ui −1, j + (1 + 2λ ) ui , j − λ ui +1, j = ui , j −1
and can be expanded as a system of equations (fix j = 1) :
− λ u0,1 + (1 + 2λ ) u1,1 − λ u2,1 = u1,0
− λ u1,1 + (1 + 2λ ) u2,1 − λ u3,1 = u2,0
− λ u2,1 + (1 + 2λ ) u3,1 − λ u4,1 = u3,0
− λ u3,1 + (1 + 2λ ) u4,1 − λ u5,1 = u4,0
 Crank-Nicolson Method
− λ u( x − h, t ) + (1 + 2λ ) u( x, t ) − λ u( x + h, t ) = u( x, t − k )
can be expressed as a Tridiagonal system of equations :
1 + 2λ −λ   u1,1   u1,0 + λ u0,1 
 − λ 1 + 2λ −λ  u   u 
   2,1  =  2,0 
 − λ 1 + 2λ − λ  u3,1   u3,0 
  u  u + λ u 
 − λ 1 + 2λ   4,1   4,0 5,1 

where u1,0 , u2,0 , u3,0 , and u4,0 are the initial temperature values
at x = x0 + h, x0 + 2h, x0 + 3h, and x0 + 4h
u0,1 and u5,1 are the boundary values at x = x0 and x0 + 5h
 Crank-Nicolson Method
The solution of the tridiagonal system produces :
The temperature values u1,1, u2,1, u3,1, and u4,1 at t = t0 + k
To compute the temperature values at t = t0 + 2k
Solve a second tridiagonal system of equations ( j = 2)
1 + 2λ −λ   u1,2   u1,1 + λ u0,2 
 − λ 1 + 2λ  u   
−λ  2, 2  = 
u

  2,1
 − λ 1 + 2λ −λ  u3,2   u3,1 
     
 − λ 1 + 2 λ  u
 4,2   4,1u + λ u5, 2 

To compute u1,2 , u2,2 , u3,2 , and u4,2

Repeat the above step to compute temperature values at t0 + 3k , etc.
 Example 2
Solve the PDE :
∂ 2u( x, t ) ∂u( x, t )
− =0
∂ x
2
∂t
u(0, t ) = u(1, t ) = 0
u( x,0) = sin(π x )

Solve using Crank - Nicolson method

Use h = 0.25, k = 0.25 to find u( x, t ) for x ∈ [0,1], t ∈ [0,1]
 Example 2
Crank-Nicolson Method
∂ 2 u( x, t ) ∂ u( x, t )
− =0
∂x 2
∂t
u ( x − h , t ) − 2u ( x , t ) + u ( x + h , t ) u ( x , t ) − u ( x , t − k )
2
=
h k
16(u( x − h, t ) − 2u( x, t ) + u( x + h, t ) ) − 4(u( x, t ) − u( x, t − k ) ) = 0
k
Define λ = 2 = 4
h
− 4 u ( x − h, t ) + 9 u ( x , t ) − 4 u ( x + h, t ) = u ( x , t − k )
− 4 ui −1, j + 9 ui , j − 4 ui +1, j = ui , j −1
Example 2

− 4u0,1 + 9u1,1 − 4u2,1 = u1,0 ⇒ 9u1,1 − 4u2,1 = sin(π / 4)

− 4u1,1 + 9u2,1 − 4u3,1 = u2,0 ⇒ −4u1,1 + 9u2,1 − 4u3,1 = sin(π / 2)
− 4u2,1 + 9u3,1 − 4u4,1 = u3,0 ⇒ − 4u2,1 + 9u3,1 = sin(3π / 4)
 Example 2
Solution of Row 1 at t1=0.25 sec

The Solution of the PDE at t1 = 0.25 sec is the solution

of the following tridiagonal system of equations :
 9 −4   u1,1  sin(0.25π )
 − 4 9 − 4 u  =  sin(0.5π ) 
   2,1   
 − 4 9  u3,1  sin(0.75π )
 u1,1   0.21151
  
⇒ u2,1  = 0.29912
 
u3,1   0.21151
Example 2:
Second Row at t2=0.5 sec

− 4u0,2 + 9u1,2 − 4u2,2 = u1,1 ⇒ 9u1,2 − 4u2,2 = 0.21151

− 4u1,2 + 9u2,2 − 4u3,2 = u2,1 ⇒ −4u1,2 + 9u2,2 − 4u3,2 = 0.29912
− 4u2,2 + 9u3,2 − 4u4,2 = u3,1 ⇒ − 4u2,2 + 9u3,2 = 0.21151
Example 2
Solution of Row 2 at t2=0.5 sec

The Solution of the PDE at t2 = 0.5 sec is the solution

of the following tridiagonal system of equations :
 9 −4   u1,2   u1,1   0.21151
 − 4 9 − 4 u  = u  = 0.29912
   2,2   2,1   
 − 4 9  u3,2  u3,1   0.21151
 u1,2  0.063267
 
⇒ u2,2  =  0.089473
 
u3,2  0.063267
Example 2
Solution of Row 3 at t3=0.75 sec
The Solution of the PDE at t3 = 0.75 sec is the solution
of the following tridiagonal system of equations :
 9 −4   u1,3   u1,2  0.063267
 − 4 9 − 4 u  = u  =  0.089473
   2,3   2, 2   
 − 4 9  u3,3  u3,2  0.063267
 u1,3  0.018924
 
⇒ u2,3  =  0.026763
 
u3,3  0.018924
Example 2
Solution of Row 4 at t4=1 sec
The Solution of the PDE at t4 = 1 sec is the solution
of the following tridiagonal system of equations :
 9 −4   u1,4   u1,3  0.018924
 − 4 9 − 4 u  = u  =  0.026763
   2, 4   2,3   
 − 4 9  u3,4  u3,3  0.018924
 u1,4  0.0056606
 
⇒ u2,4  = 0.0080053
 
u3,4  0.0056606
THANK YOU