[Getdp] Convergence of magnetodynamics

Vadims Geža vadims.geza at gmail.com
Tue Mar 27 11:09:00 CEST 2018


Hi all!

Recently I faced with necessity of iterative solver use for
magnetodynamics. I started with this case:

https://gitlab.onelab.info/doc/models/wikis/Magnetodynamics-with-cohomology-conditions

I used getdp with petsc, and I could not get to convergence.
I have tried:

getdp indheat.pro -msh indheat.msh -solve MagDynAVComplex -solver solver.par

solver.par:

-pc_type icc
-ksp_type gmres
-ksp_max_it 1000
-ksp_rtol 1e-12

I have tried also other combinations of preconditioner and solver (icc +
cg, ilu + gmres, bjacobi + gmres, bjacobi + cg ...).
So far the best convergence was with icc+gmres, which was only to 1e-9
after 1000 iterations, but the results were unphysical. Direct solver gives
proper results.

Do you have any clue why is there such behavior and is there a way to get
iterative solver work?

Best Regards, Vadims
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://onelab.info/pipermail/getdp/attachments/20180327/6d639759/attachment.html>


More information about the getdp mailing list