[petsc-users] Using VecGetType() and VecNest

[Stefano Zampini]

The idea of the pod code was to create local vectors to perform local
products and use mpi non blocking collectives . Not sure how to fix for the
nest case. I'm on vacation now, I can take a look in a couple of weeks or so

On Sat, Aug 20, 2022, 05:30 Barry Smith <bsmith at petsc.dev> wrote:

>
>   I am not sure why the code works this way creating these sequential work
> vectors. Since VecNest needs some information about the subvectors I don't
> think just setting these work vectors to nest vectors will work properly.
>
>   I am cc:ing Stefano who wrote the code and can likely say immediately
> what the solution is.
>
>   Barry
>
>
> On Aug 19, 2022, at 5:21 PM, Wells, David <drwells at email.unc.edu> wrote:
>
> Hello PETSc experts,
>
> I am using VecNest to solve a Stokes problem and I ran into an issue using
> the POD KSPGuess routines:
>
> [0]PETSC ERROR: --------------------- Error Message
> --------------------------------------------------------------
> [0]PETSC ERROR: Unknown type. Check for miss-spelling or missing package:
> https://petsc.org/release/install/install/#external-packages
> [0]PETSC ERROR: Unknown vector type: nest
> [0]PETSC ERROR: See https://petsc.org/release/faq/ for trouble shooting.
> [0]PETSC ERROR: Petsc Release Version 3.17.0, Mar 30, 2022
> [0]PETSC ERROR: /[...]/step-32/step-32 on a  named mitral by drwells Fri
> Aug 19 16:33:08 2022
> [0]PETSC ERROR: Configure options [...]
> [0]PETSC ERROR: #1 VecSetType() at /afs/
> cas.unc.edu/users/d/r/drwells/Documents/Code/C/petsc-3.17.0/src/vec/vec/interface/vecreg.c:80
> [0]PETSC ERROR: #2 KSPGuessSetUp_POD() at /afs/
> cas.unc.edu/users/d/r/drwells/Documents/Code/C/petsc-3.17.0/src/ksp/ksp/guess/impls/pod/pod.c:118
> [0]PETSC ERROR: #3 KSPGuessSetUp() at /afs/
> cas.unc.edu/users/d/r/drwells/Documents/Code/C/petsc-3.17.0/src/ksp/ksp/interface/iguess.c:352
> [0]PETSC ERROR: #4 KSPSolve_Private() at /afs/
> cas.unc.edu/users/d/r/drwells/Documents/Code/C/petsc-3.17.0/src/ksp/ksp/interface/itfunc.c:830
> [0]PETSC ERROR: #5 KSPSolve() at /afs/
> cas.unc.edu/users/d/r/drwells/Documents/Code/C/petsc-3.17.0/src/ksp/ksp/interface/itfunc.c:1078
>
>
> It looks like VecGetType() works with VecNest while VecSetType() does not.
> I also noticed that VecNest isn't listed in VecRegisterAll().
>
> I think I can write a patch to make the POD KSPGuess work with VecNest,
> but: should VecSetType() work with VecNest? It seems like this is an
> oversight but I'd like to know if there is some fundamental reason why the
> sequence used there
>
>     PetscCall(KSPCreateVecs(guess->ksp,1,&v,0,NULL));
>     PetscCall(VecCreate(PETSC_COMM_SELF,&vseq));
>     PetscCall(VecGetLocalSize(v[0],&n));
>     PetscCall(VecSetSizes(vseq,n,n));
>     PetscCall(VecGetType(v[0],&type));
>     PetscCall(VecSetType(vseq,type));
>     PetscCall(VecDestroyVecs(1,&v));
>     PetscCall(VecDuplicateVecs(vseq,pod->maxn,&pod->xsnap));
>     PetscCall(VecDestroy(&vseq));
>     PetscCall(PetscLogObjectParents(guess,pod->maxn,pod->xsnap));
>
> can't work.
>
> Best,
> David Wells
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20220820/744a2c0c/attachment-0001.html>