[petsc-dev] who broke petsc-dev

[Matthew Knepley]

On Sun, Nov 13, 2011 at 10:36 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:

0) Me

PetscErrorCode  KSPSetOperators(KSP ksp,Mat Amat,Mat Pmat,MatStructure flag)
> {
>  MatNullSpace   nullsp;
>  PetscErrorCode ierr;
>
>  PetscFunctionBegin;
>  PetscValidHeaderSpecific(ksp,KSP_CLASSID,1);
>  if (Amat) PetscValidHeaderSpecific(Amat,MAT_CLASSID,2);
>  if (Pmat) PetscValidHeaderSpecific(Pmat,MAT_CLASSID,3);
>  if (Amat) PetscCheckSameComm(ksp,1,Amat,2);
>  if (Pmat) PetscCheckSameComm(ksp,1,Pmat,3);
>  if (!ksp->pc) {ierr = KSPGetPC(ksp,&ksp->pc);CHKERRQ(ierr);}
>  ierr = PCSetOperators(ksp->pc,Amat,Pmat,flag);CHKERRQ(ierr);
>  if (ksp->setupstage == KSP_SETUP_NEWRHS) ksp->setupstage =
> KSP_SETUP_NEWMATRIX;  /* so that next solve call will call PCSetUp() on new
> matrix */
>  if (ksp->guess) {
>    ierr = KSPFischerGuessReset(ksp->guess);CHKERRQ(ierr);
>  }
>  ierr = MatGetNullSpace(Amat, &nullsp);CHKERRQ(ierr);
>  if (nullsp) {
>    ierr = KSPSetNullSpace(ksp, nullsp);CHKERRQ(ierr);
>  }
>  PetscFunctionReturn(0);
> }
>
>
> 1) It is legitimate to call KSPSetOperators with a null Amat so this is a
> bug that crashes large parts of PETSc
>

What does a NULL A even mean here. I can understand a NULL P, which might
mean A == P, but what would it mean
to have no A?


> 2) Do you really want to do this? I am not sure I like it.
>

Yes, I definitely do. How else will the solver get information about the
null space of an operator.

   Matt

-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20111113/0e23b495/attachment.html>