which is the numerical heat equation, right? Do I just need to finetune the c? Or maybe it's not actually a problem and my coloring takes small changes too seriously, after a while it looks nice...
(I already fixed the central horizontal distortion?)
@noneuclideandreamer Do you have the stability conditions satisfied? von Neumann stability analysis says you need (c\leq\frac{1}{4}). In normal units this is (\Delta t\leq \frac{\Delta x^2}{4\mu}) with \mu being the diffusion coefficient. If that is satisfied then it should work!
@noneuclideandreamer It's the +/-1 in the f(x+1,y,t) terms that makes it diamond-shaped -- you're running on Manhattan distance. I'm not sure how to fix that behaviour in a CA-like model (where state is quantised to a grid and effects between cells are purely local to the neighbours).
@noneuclideandreamer not exactly. Compare with the Laplacian in polar coordinates. Or written in a more coordinate-free way as the divergence of the gradient. Mathematically the same, but a way different naive conversion to code. You left out influences from the diagonal neighbors so of course you get a diamond pattern.
Add comment