0% found this document useful (0 votes)
72 views

Laplace Finite Difference

The document describes the finite difference method for numerically solving the Laplace equation using a grid-based approach. It involves: 1) Discretizing the domain into a grid and using centered difference approximations to represent the partial derivatives at grid points. 2) Iteratively calculating the function values at interior grid points based on the values at neighboring points, in a process called relaxation. 3) The iterative process, known as Jacobi iteration, continues updating the interior points until the solution converges after many iterations.

Uploaded by

Farid Morabet
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
72 views

Laplace Finite Difference

The document describes the finite difference method for numerically solving the Laplace equation using a grid-based approach. It involves: 1) Discretizing the domain into a grid and using centered difference approximations to represent the partial derivatives at grid points. 2) Iteratively calculating the function values at interior grid points based on the values at neighboring points, in a process called relaxation. 3) The iterative process, known as Jacobi iteration, continues updating the interior points until the solution converges after many iterations.

Uploaded by

Farid Morabet
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Numerical Solution to Laplace Equation:  Finite Difference Method

[Note:  We will illustrate this in 2D.  Extension to 3D is straightforward.]

Suppose seek a solution to the Laplace Equation subject to Dirichlet boundary conditions :
2  2  ( x, y )  2  ( x , y )
  ( x, y )   0 subject to  specified on the boundary
x y
We discretize the (x,y) region of interest into a grid, with equal Δx = Δy = h grid spacings.
y
h
h

As we showed on HW #1, centered difference approximations to the partial derivatives at 
As we showed on HW #1 centered difference approximations to the partial derivatives at
a grid point (i,j) are :

 2  ( x, y )  i 1, j  2 i , j   i 1, j  2  ( x, y )  i , j 1  2 i , j   i , j 1
 
x 2
(i , j )
h 2
y 2
(i , j )
h 2
Thus, the discretized approximation to the Laplace Equation becomes :

  i 1, j  2 i , j   i 1, j    i , j 1  2 i , j   i , j 1 
 2
   2
  0
 h   h 

 i, j
1

  i 1, j   i 1, j   i , j 1   i , j 1
4

Thus, we see that under this 
approximation the value of  at grid  Numerical Analysis:  
point (i,j) depends on the values of  at  “five‐point stencil”
its four nearest neighboring grid points.
y

“Boundary value problem” becomes :
y p
[1]   “boundary values” specified on 
grid points on the boundary
x
[2]  Want to be able to numerically 
calculate the values of  at the interior grid 
points.
points

Because  at grid point (i,j) depends on its four neighbors, we can iteratively solve 
f  at the interior grid points via the following iterative scheme (“relaxation”) :
for  h i i id i i h f ll i i i h (“ l i ”)
[0a] Fix the initial values of  on the grid boundaries subject to the boundary values.
[0b] Set/choose initial values for the interior grid points
[0b] Set/choose initial values for the interior grid points.
[1] Successively sweep through all of the interior grid points, where on the (m+1)th
sweep (iteration) through all of the grid points :
1
 im, j 1   im, j 
4
   
 im1, j   im1, j   im, j 1   im, j 1  4 im, j 
  im, j   im, j

value from previous  residual of mth
iteration iteration
y
Basic “Jacobi Iteration” scheme :
Step 0:  Fix the boundary values; choose initial 
guesses for the interior points.
f h i i i
Iteration 1: Calculate  at all of the interior grid points
according to the formulas below.  The values of  at
x all of the interior grid points have been “updated”
all of the interior grid points have been “updated”.
Iteration 2: Re‐calculate  at all of the interior grid
points using the “updated” values from Iteration 1.
Again update the values of  at each grid point.
Again update the values of  at each grid point
Continue N times …

 m 1
i, j 
1
   
    im1, j   im1, j   im, j 1   im, j 1  4 im, j
m
i, j
4
  im, j   im, j After many iterations the residual will
After many iterations, the residual will 
become small (i.e., the solution will 
value from previous  residual of mth
iteration iteration
have “relaxed” to its true value).

Notes :
[1] Under this scheme, the values of  on the boundary are not modified.
[2] The accuracy of the solution will, in general, depend on the grid spacing h.
[3] The accuracy will, in general, improve with the number of iterations N, but is 
subject to [2].  (The CPU time will also, in general, scale with N.)
Extension to 3D is straightforward.  Have a 7‐point stencil, where the value of 
g p , at 
(i,j,k) depends on its six neighboring points:
(i+1,j,k),  (i‐1,j,k),   (i,j+1,k),  (i,j‐1,k),  (i,j,k+1),  (i,j,k‐1)

Jacobi iteration is the simplest/most‐elementary approach to a numerical solution of 
the Laplace Equation via relaxation.

More sophisticated methods (e.g., Gauss‐Seidel, Successive Overrelaxation, Multigrid
Methods, etc.) exist which improve both the accuracy and speed towards 
h d ) h h b h h d d d
convergence.
See, for example, Numerical Recipes in C++.

You might also like