Hello everyone,
I have been solving some 3D RF Frequency domain problems recently, using the the default FGMRES solver.
I just encountered a weird problem: the solver iteration number shown by the Progress tab lags behind the iteration number shown on the convergence plot. When the convergence plot shows that the solver has reached the desired tolerance, I can see that the resource usage by COMSOL drops close to zero, consistent with the fact that the solver has stopped. However, the Progress tab still shows a residual error above tolerance, and I have to wait for it to catch up to the convergence plot. As I have to do many such 3D calculations, this extra time is quite annoying.
Has anybody encountered similar problems before, or have an idea of what might be causing the lag? I have been doing a bunch of similar calculations for the past few weeks and the problem only appeared recently.