Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

RF module-constant error after few thousands of iterations

Please login with a confirmed email address before reporting spam

Hi,
I am solving in RF module for scattering field geometry made of metallic particles on substrates and PMLs.
Sometimes (reason unknown) GMRES and other iterative solvers cannot solve my models, and above all, after thousands of iterations initial error of 400 stays constant all the time. If I create similar model, similar number of meshing elements and similar geometry it might works.

Any idea?

Comsol support didn't manage to solve one of problematic models, nor they offered any explanation.

2 Replies Last Post 06.10.2010, 10:53 GMT-4
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 06.10.2010, 09:15 GMT-4
Hi

iterative solvers may get fooled by local optimums, or too large sacaling differences between the variables solved for. Have you checked the expected variable values and how they relate ?

--
Good luck
Ivar
Hi iterative solvers may get fooled by local optimums, or too large sacaling differences between the variables solved for. Have you checked the expected variable values and how they relate ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 06.10.2010, 10:53 GMT-4

Hi

iterative solvers may get fooled by local optimums, or too large sacaling differences between the variables solved for. Have you checked the expected variable values and how they relate ?

--
Good luck
Ivar

Hi, thanks for the quick reply.

I am not sure if I get what you mean, I am still rookie. However, there should be some high gradient of the fields, but I mesh nicely there.
Difference between two models, where one work and other is "flat line" convergence is minimum. Actually, once, COMSOL solved model where I have "singular" line of contact between 2 scatterers (which should be tricky for solving, unstable solution, etc) and couldn't solve when scatterers were separated by some particular distance, meaning that if I change that distance slightly, it solves it again. BTW, system was off the resonance.

Thanks
[QUOTE] Hi iterative solvers may get fooled by local optimums, or too large sacaling differences between the variables solved for. Have you checked the expected variable values and how they relate ? -- Good luck Ivar [/QUOTE] Hi, thanks for the quick reply. I am not sure if I get what you mean, I am still rookie. However, there should be some high gradient of the fields, but I mesh nicely there. Difference between two models, where one work and other is "flat line" convergence is minimum. Actually, once, COMSOL solved model where I have "singular" line of contact between 2 scatterers (which should be tricky for solving, unstable solution, etc) and couldn't solve when scatterers were separated by some particular distance, meaning that if I change that distance slightly, it solves it again. BTW, system was off the resonance. Thanks

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.