Five point scheme Finite Difference Method

Click For Summary
SUMMARY

The discussion focuses on the five-point scheme for solving the Poisson equation $$u_{xx}+u_{yy}=f$$ using finite difference methods. The general form of the scheme is presented as $$a_{1}U_{i,j-1}+a_{2}U_{i-1,j}+a_{3}U_{i,j}+a_{4}U_{i+1,j}+a_{5}U_{i,j+1}=f_{i,j}$$. An alternative formulation is proposed, involving a rotated grid and approximations of neighboring points, which may lead to different computational characteristics. The concept of separating equations for grid points based on a chessboard pattern is also introduced, suggesting potential benefits depending on the application.

PREREQUISITES
  • Understanding of finite difference methods
  • Familiarity with the Poisson equation
  • Knowledge of numerical approximation techniques
  • Basic concepts of grid-based computational methods
NEXT STEPS
  • Research the implementation of the five-point finite difference scheme in MATLAB
  • Explore the effects of grid rotation on numerical stability and accuracy
  • Study the chessboard patterning technique in numerical simulations
  • Investigate alternative finite difference schemes for solving partial differential equations
USEFUL FOR

Mathematicians, computational scientists, and engineers involved in numerical analysis and simulation of partial differential equations, particularly those working with finite difference methods.

sigh1342
Messages
30
Reaction score
0
For possion equation $$u_{xx}+u_{yy}=f$$
I know the general five point scheme is in the form
$$a_{1}U_{i,j-1}+a_{2}U_{i-1,j}+a_{3}U_{i,j}+a_{4}U_{i+1,j}+a_{5}U_{i,j+1}=f_{i,j}$$
But , is there have the form
$$a_{1}U_{i-1,j-1}+a_{2}U_{i-1,j+1}+a_{3}U_{i,j}+a_{4}U_{i+1,j+1}+a_{5}U_{i+1,j-1}=f_{i,j}$$?
 
Physics news on Phys.org
It would be easy enough to invent one. Just approximate ##U_{i-1,j} = (U_{i-1,j-1}+U_{i-1,j+1})/2## etc.

Whether that would be any good in practice is another question, of course.

Alternatively, imagine your grid is rotated through 45 degrees, and use your original formula with ##h## replaced by ##h\sqrt 2##.

One feature of it would be: color the grid points red and black, in a pattern like a chess board. Except for the boundary conditions, you have one set of equations linking only the red points, and another set linking only the black points.

That may or may not be a good thing, depending on what use you make of it.
 
Last edited:
  • Like
Likes   Reactions: 1 person

Similar threads

  • · Replies 8 ·
Replies
8
Views
5K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 8 ·
Replies
8
Views
3K
  • · Replies 36 ·
2
Replies
36
Views
8K
  • · Replies 30 ·
2
Replies
30
Views
2K
Replies
5
Views
3K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 12 ·
Replies
12
Views
2K
  • · Replies 19 ·
Replies
19
Views
3K