[Getdp] getdp 2.0.0 : petsc solver options ?

[Lars Rindorf]

Dear Christophe

If understand it correctly you wish to solve, say, 
A*v + B[v]*v = b,
And then use different solvers for matrix A and B? If you use PETSc non-linear solver, SNES, then this is possible using the usual command-line options, but Getdp does not use SNES. It is still possible to use different solvers and settings, but you have to make some alterations in the getdp source. Should be quite easy though.

KR Lars


 



-----Oprindelig meddelelse-----
Fra: getdp-bounces at ace20.montefiore.ulg.ac.be [mailto:getdp-bounces at ace20.montefiore.ulg.ac.be] På vegne af Christophe Geuzaine
Sendt: 17. oktober 2010 12:03
Til: christophe.trophime at lncmi.cnrs.fr
Cc: getdp at geuz.org
Emne: Re: [Getdp] getdp 2.0.0 : petsc solver options ?

On 15/10/10 15:53, trophime wrote:
> Is it possible to store the solver options (in the right format wheter
> we are using sparskit or petsc) in a file which
> is defined in the "resolution" section like that :
>
> Resolution {
>    { Name NL_Elec_Th_Mag_3D ;
>    System {
>      { Name Sys_Th ; NameOfFormulation NL_Heat_Equation ; Solver
> "Sys_Th.par"; }
>
> It is working right when getdp is compiled with sparskit
> But it does not seem to work when getdp is compiled with petsc...

Hi Christophe - No, it's not supported with PETSc. I just had a quick 
look at how PETSc manages its option database and I'm actually not sure 
how we could/should handle different options per linear system...

Any ideas on the list from petsc specialists?



>
> Best regards,
> Christophe
>
>
>
> _______________________________________________
> getdp mailing list
> getdp at geuz.org
> http://www.geuz.org/mailman/listinfo/getdp
>


-- 
Prof. Christophe Geuzaine
University of Liege, Electrical Engineering and Computer Science
http://www.montefiore.ulg.ac.be/~geuzaine

_______________________________________________
getdp mailing list
getdp at geuz.org
http://www.geuz.org/mailman/listinfo/getdp