[petsc-dev] Recent error checking in DMCreateDefaultSF breaks parallel DMs

[Matthew Knepley]

I pushed a fix to next. It will be in master in a day or two.

  Thanks,

     Matt


On Wed, May 28, 2014 at 10:16 AM, Lawrence Mitchell <
lawrence.mitchell at imperial.ac.uk> wrote:

> 1cc89ce (DM: Added check for erroneous input when creating default SF)
>
> Checks that the constrained point does not constrain more dofs than the
> unconstrained point has.  However, this breaks in parallel, since
> PetscSectionCreateGlobalSection says:
>
> "  Note: This gives negative sizes and offsets to points not owned by
> this process"
>
> Hence the following example breakage:
>
> $ cat > foo.py <<\EOF
> import petsc4py
> import sys
> petsc4py.init(sys.argv)
> from petsc4py import PETSc
>
>
> bdy = PETSc.DMPlex().create()
> bdy.setDimension(1)
> bdy.createSquareBoundary([0, 0], [1, 1], [3, 3])
>
> dm = PETSc.DMPlex().generate(bdy)
>
> dm.distribute()
> sec = dm.createSection(1, [1], [1, 0, 0])
>
> dm.setDefaultSection(sec)
> gsec = dm.getDefaultGlobalSection()
>
> dm.createDefaultSF(sec, gsec)
> EOF
>
> $ mpiexec -n 2 python foo.py
> Traceback (most recent call last):
>   File "foo.py", line 19, in <module>
>     dm.createDefaultSF(sec, gsec)
>   File "DM.pyx", line 268, in petsc4py.PETSc.DM.createDefaultSF
> (src/petsc4py.PETSc.c:179320)
> petsc4py.PETSc.Error: error code 63
> [0] DMCreateDefaultSF() line 3065 in
> /data/lmitche1/src/petsc/src/dm/interface/dm.c
> [0] Argument out of range
> [0] Point 9 has 0 constraints > -2 dof
>
>
> Maybe this patch?
>
> diff --git a/src/dm/interface/dm.c b/src/dm/interface/dm.c
> index d830c0c..71bf7ad 100644
> --- a/src/dm/interface/dm.c
> +++ b/src/dm/interface/dm.c
> @@ -3062,7 +3062,7 @@ PetscErrorCode DMCreateDefaultSF(DM dm,
> PetscSection localSection, PetscSection
>
>      ierr     = PetscSectionGetDof(globalSection, p, &gdof);CHKERRQ(ierr);
>      ierr     = PetscSectionGetConstraintDof(globalSection, p,
> &gcdof);CHKERRQ(ierr);
> -    if (gcdof > gdof) SETERRQ3(PETSC_COMM_SELF,
> PETSC_ERR_ARG_OUTOFRANGE, "Point %d has %d constraints > %d dof", p,
> gcdof, gdof);
> +    if (gcdof && gcdof > gdof) SETERRQ3(PETSC_COMM_SELF,
> PETSC_ERR_ARG_OUTOFRANGE, "Point %d has %d constraints > %d dof", p,
> gcdof, gdof);
>      nleaves += gdof < 0 ? -(gdof+1)-gcdof : gdof-gcdof;
>    }
>    ierr = PetscMalloc1(nleaves, &local);CHKERRQ(ierr);
>
> Cheers,
>
> Lawrence
>



-- 
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/20140528/b80c09bf/attachment.html>