FINITE DIFFERENCE

AND PDE
Haryo Tomo
Numerical methods: properties
Finite differences - time-dependent PDEs
-> robust, simple concept, easy to
parallelize, regular grids, explicit method

- static and time-dependent PDEs

Finite elements -> implicit approach, matrix inversion, well founded,
irregular grids, more complex algorithms,
engineering problems

Finite volumes - time-dependent PDEs

-> robust, simple concept, irregular grids, explicit
method
Other numerical methods
Particle-based - lattice gas methods
- molecular dynamics
methods - granular problems
- fluid flow
- earthquake simulations
-> very heterogeneous problems, nonlinear problems

Boundary element - problems with boundaries (rupture)

- based on analytical solutions
methods - only discretization of planes
-> good for problems with special boundary conditions
(rupture, cracks, etc)

Pseudospectral - orthogonal basis functions, special case of FD

- spectral accuracy of space derivatives
methods - wave propagation, ground penetrating radar
-> regular grids, explicit method, problems with
strongly heterogeneous media
What is a finite difference?
Common definitions of the derivative of f(x):

f ( x  dx)  f ( x)
 x f  lim
dx0 dx
f ( x)  f ( x  dx)
 x f  lim
dx0 dx
f ( x  dx)  f ( x  dx)
 x f  lim
dx0 2dx

These are all correct definitions in the limit dx->0.

But we want dx to remain FINITE

What is a finite difference?
The equivalent approximations of the derivatives are:

 f ( x  dx)  f ( x)
x f  forward difference
dx

 f ( x)  f ( x  dx)
x f  backward difference
dx

f ( x  dx)  f ( x  dx)
x f  centered difference
2dx
The big question:


How good are the FD approximations?

This leads us to Taylor series....

Taylor Series
Taylor series are expansions of a function f(x) for some
finite distance dx to f(x+dx)

2 3 4
dx dx dx
f ( x  dx)  f ( x)  dxf ' ( x)  f '' ( x)  f ''' ( x)  f '''' ( x)  ...
2! 3! 4!

What happens, if we use this expression for

f ( x  dx)  f ( x)
x f  ?
dx
Taylor Series
... that leads to :

f ( x  dx)  f ( x) 1  dx 2
dx 3

 dxf ( x) 
'
f ( x) 
''
f ( x)  ...
'''

dx dx  2! 3! 
 f ' ( x)  O(dx)

The error of the first derivative using the forward

formulation is of order dx.

Is this the case for other formulations of the derivative?

Let’s check!
Taylor Series
... with the centered formulation we get:

f ( x  dx / 2)  f ( x  dx / 2) 1  dx 3

 dxf ( x) 
'
f ( x)  ...
'''

dx dx  3! 
 f ' ( x)  O(dx 2 )

The error of the first derivative using the centered

approximation is of order dx2.

This is an important results: it DOES matter which formulation

we use. The centered scheme is more accurate!
Higher order operators
(2dx) 2 (2dx)3
*a | f ( x  2dx)  f  (2dx) f ' f ' ' f '''
2! 3!
(dx) 2 (dx)3
*b | f ( x  dx)  f  (dx) f ' f ' ' f '''
2! 3!
(dx) 2 (dx)3
*c | f ( x  dx)  f  (dx) f ' f ' ' f '''
2! 3!
(2dx) 2 (2dx)3
*d | f ( x  2dx)  f  (2dx) f ' f ' ' f '''
2! 3!

... again we are looking for the coefficients a,b,c,d with which
the function values at x±(2)dx have to be multiplied in order
to obtain the interpolated value or the first (or second) derivative!

... Let us add up all these equations like in the previous case ...
Problems: Stability
c 2 dt 2
p (t  dt )  2
 p( x  dx)  2 p( x)  p( x  dx)
dx
 2 p (t )  p (t  dt )  sdt 2

Stability: Careful analysis using harmonic functions shows that

a stable numerical calculation is subject to special conditions
(conditional stability). This holds for many numerical problems.
(Derivation on the board).

dt
c   1
dx
Problems: Dispersion
c 2 dt 2
p (t  dt )  2
 p( x  dx)  2 p( x)  p( x  dx)
dx
 2 p (t )  p (t  dt )  sdt 2

Dispersion: The numerical

approximation has
artificial dispersion,
in other words, the wave
speed becomes frequency
dependent (Derivation in
the board).
You have to find a
True velocity
frequency bandwidth
where this effect is small.
The solution is to use a
sufficient number of grid
points per wavelength.
Finite Differences - Summary

 Conceptually the most simple of the numerical methods and

can be learned quite quickly
 Depending on the physical problem FD methods are
conditionally stable (relation between time and space
increment)
 FD methods have difficulties concerning the accurate
implementation of boundary conditions (e.g. free surfaces,
absorbing boundaries)
 FD methods are usually explicit and therefore very easy to
implement and efficient on parallel computers
 FD methods work best on regular, rectangular grids
Partial Differential Equations

by Lale Yurttas, Texas A&M

University
Part 8 14
PERSAMAAN DIFERENSIAL PARSIAL
• Persamaan Umum

 2  2  2  
a 2 b c 2 d e  f  g  0
x xy y x y
• Menyatakan bagaimana variabel tak bebas Ø berubah
terhadap variabel bebas x,y. Disini a,b,c,d,e,f, dan g
mungkin merupakan fungsi dari Ø
Jenis2 PDP
• Ditentukan oleh harga b2-4ac
< 0, eliptic
= 0, parabolic
> 0, hyperbolic
• Adveksi…
• Difusi….
• Gelombang…
JENIS-
JENIS
PDP
by Lale Yurttas, Texas A&M
University
18
The Laplacian Difference Equations/
 2T  2T
 2 0
x 2
y Laplace Equation
 2T Ti 1, j  2Ti , j  Ti 1, j O[(x)2]

x 2 x 2
 2T Ti , j 1  2Ti , j  Ti , j 1
 O[(y)2]
y 2
y 2

Ti 1, j  2Ti , j  Ti 1, j Ti , j 1  2Ti , j  Ti , j 1

 0
x 2
y 2

x  y
Ti 1, j  Ti 1, j  Ti , j 1  Ti , j 1  4Ti , j  0 Laplacian difference
equation.

by Lale Yurttas, Texas A&M

Holds for all interior points
Chapter 29
University
by Lale Yurttas, Texas A&M
University
Chapter 29 20
• In addition, boundary conditions along the edges must be
specified to obtain a unique solution.
• The simplest case is where the temperature at the boundary is
set at a fixed value, Dirichlet boundary condition.
• A balance for node (1,1) is:
T21  T01  T12  T10  4T11  0
T01  75
T10  0
 4T11  T12  T21  0
• Similar equations can be developed for other interior points to
result a set of simultaneous equations.

by Lale Yurttas, Texas A&M

University
Chapter 29 21
• The result is a set of nine simultaneous equations with nine
unknowns:
4T11  T21  T12  75
 T11  4T21  T13  T22 0
 T21  4T31  T32  50
 T11  4T12  T22  T13  75
 T21  T12  4T22  T32  T23 0
 T31  T22  4T32  T33  50
 T12  4T13  T23  175
 T22  T13  4T23  T33  100
 T32  T23  4T33  150
h

Diffusion

x
h  2h
D 2 Diffusion Equation
t x
 h
ud   D
x
x
h

ud
ud ud  x
x

x
ud
h  x   xt
x x
 ud
h  x   xt
x
t  0
h ud h
 ud   D
t x x

h h 2
D 2 Diffusion Equation
t x
Numerical Solution of Diffusion Eq.
h
h(i  1)
h(i)

h(i  1)

x
 h  h(i )  h(i  1)  h  h(i  1)  h(i)
     
 x 1 x  x 2 x

 2 h 1  h   h  
      
x 2
x  x 2  x 1 
 Numerical Calculation of Diffusion Equation
t

J
h(i  1, j  1)

j+
h(i, j )

j
t

2 ・・ j-1
1
x
x
1 2 ・・・・・ i-1 i i+1 ・・・・ N
 h h(i, j  1)  h(i, j )

t t
 h D  h(i  1, j )  h(i, j ) h(i, j )  h(i  1, j ) 
2
D 2    
x x  x x 
h(i  1, j )  2h(i, j )  h(i  1, j )
D
Unknown  x 2

h(i, j  1)  h(i, j ) Known

Dt
 2 {h(i  1, j )  2h(i, j )  h(i  1, j )}
x
 Numerical Solution of Diffusion Equation
t Boundary Condition (Given)

J
j+
j
t

2 ・・ j-1
1
x
x
Initial Condition
1 2 ・・・・・ i-1 i i+1 ・・・・ N (Given)
 Dt
h(i, j  1)  h(i, j )  2 {h(i  1, j )  2h(i, j )  h(i  1, j )}
x

do i  2,N-1
t
hnew (i )  hold (i )   hold (i  1)  2hold (i )  hold (i  1)  2 D
x
end do
do i  2,N-1
hold (i )  hnew (i )
end do
Contoh

d2y dy
2
 3  2 y  4 x 2
, y(1)  1, y(2)  6
dx dx

• Cari solusinya dengan step size 0,2

• Jumlah titik solusi n=((2-1)/0,2)-1=4
• Kita dapatkan 4 persamaan, satu untuk tiap titik yang
dicari.
Penyelesaian dengan Beda Hingga
Persamaannya:

yi 1  2 yi  yi 1 yi 1  yi 1
3  2 yi  4 xi
2
2
h
Buat persamaan untuk semua 2htitik, mulai dari i=1,
hingga i=4

x0=a x1 x2 x3 x4 x5=b
Domain Solusi

y5=6

y0=1

x0=1 x5=2
Penyelesaian dg Finite Diff.
• Dengan h = 0,2

32,5 yi 1  48 yi  17,5 yi 1  4 xi
2

• Buat persamaan untuk semua titik, mulai dari i=1,

hingga i=4
• Masukkan nilai-nilai x1=1,2 hingga x4 =1,8 dan kondisi
batas y0 = 1 dan y5 = 6
Kondisi batas
• Dirichlet atau fixed boundary,
misal C(0) = Co
• Neuman atau natural boundary,
misal dC/dx = 0
• Robin/ Cauchy boundary condition,
misal dC/dx + C = 0

• Penerapan dalam finite difference dengan menambahkan

imaginary node
penyelesaian
persamaan
parabolik
dengan
skema
eksplisit
SEHINGGA…
persamaan diferensial,

ditulis dalam bentuk metode beda hingga,

 penyelesaian
persamaan
laplace
….equation
eliptik
 Stabilitas skema eksplisit

k
penyelesaian
persamaan
parabolik
dengan
skema
implisit