> till some value and then blow up. I tried to experiment a bit, and I have > figured out that the process can be made stable by disabling subpixel > averaging (I hope the following table could be usefull):
Yes, right now the subpixel averaging occasionally exacerbates instabilities. > P.S. I also noticed that setting phase difference to get periodic boundary > conditions to 0 instead of 2\pi decreases remarkably memory usage. Yes, if the Bloch wavevector k is non-zero Meep is forced to use complex fields, which roughly doubles the memory usage. Of course, we could try to detect any integer multiple of 2\pi as being equivalent to 0, but this didn't seem worth it because (i) you might as well just set k=0 and (ii) because of floating-point errors, sin(2\pi) is not exactly 0 any more. Steven _______________________________________________ meep-discuss mailing list [email protected] http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
