HI,all I am trying to implement a Poisson solver(firstly 2D) using FFT. All boundaries are von Neumann. I employ a cosine FFT along x direction and then solve tridiagnal systems along y for each wave number K. But you know the matrix for K=0 is singular. While some articles say that it can be decided by the properties of the physical problem,but I do not know what the tricks are. Any comments are greatly appreciated! Linfeng

Try Jos Stam's paper on the topic: Jos Stam, "A Simple Fluid Solver based on the FFT", Journal of Graphics Tools Volume 6, Number 2, 2001, 43-52

Avaiable here: http://www.dgp.toronto.edu/people/st...earch/pub.html

Cheers... -ben houston http://www.exocortex.org/ben

The paper is interesting,but it says little about the FFT boundary treatment.Thank you,anyway.

Linfeng

Well, there is some discussion as well in Numerical Recipes about FFT and boundary conditions. But its been a while since I've looked at it.

Yeah!thanks,Ben.I read through the chapters there and got a much more clearer concept about it. And then, I made it!

Sincerely,