Unfortunately, there is not much you can do besides increase the L parameter and the number of iterations.
Increasing the diameter of a simulation relative to the wavelength increases the condition number, which worsens the convergence of iterative solvers. The general way to improve this is to implement a more sophisticated iterative solver that employs preconditioners. Preconditioning wave equations (Helmholtz-like equations) is notoriously difficult to do well, but some possible strategies are discussed in https://github.com/NanoComp/meep/issues/548 <https://github.com/NanoComp/meep/issues/548> > On Sep 24, 2019, at 4:24 PM, Alexis Hotte <alexis-hottekilb...@hotmail.com> > wrote: > > Hello to the meep community, > > > I was hoping to use the frequency domain solver for my own application, > but I've encountered a stumbling block. > > I've played around with the ring resonator tutorial example, and it > seems that when I increase the domain size (let's say a radius of 10 > micrometers) or resolution (to about 30-40), the residual keeps > increasing over time! Or, the terminal/notebook simply crashes. > > Has anyone here been able to use the frequency domain solver > effectively? Does it simply boil down to playing with the tol, maxiters, > and L parameters? > > > Thank you, > > Alexis HK. > > _______________________________________________ > meep-discuss mailing list > meep-discuss@ab-initio.mit.edu > http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
_______________________________________________ meep-discuss mailing list meep-discuss@ab-initio.mit.edu http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
