[petsc-dev] DMPlex VecView to HDF5

[Matthew Knepley]

On Mon, Feb 12, 2018 at 10:57 PM, Adrian Croucher <a.croucher at auckland.ac.nz
> wrote:

> hi,
>
> I have a DMPlex which I pass in to DMCreateGlobalVector() to create a
> vector, and then use VecView() to view the vector.
>
> If the DM happens to contain no points on one process, this still works
> fine if I view the vector using PETSC_VIEWER_STDOUT_WORLD.
>
> However if I create an HDF5 viewer using PetscViewerHDF5Open() and try to
> view the vector using that, it crashes with the error message "Could not
> classify input Vec for VTK".
>
> It goes into VecView_Plex_Local_HDF5_Internal(), which calls
> DMPlexGetFieldType_Internal() (plex.c:93). This has some conditionals
> testing if vStart < pEnd or cStart < pEnd, which are never satisfied
> because pEnd = 0. This is what leads to it raising the error.
>
> This function returns the point range sStart, sEnd and field type ft. To
> handle the case where there are no points on the process, should it perhaps
> just return sStart, sEnd = 0,0 and a default (or maybe null) value for ft?


Crap, this is a problem. I will have to put some parallel code in there
because everyone must agree on the ft.

   Matt


>
> - Adrian
>
> --
> Dr Adrian Croucher
> Senior Research Fellow
> Department of Engineering Science
> University of Auckland, New Zealand
> email: a.croucher at auckland.ac.nz
> tel: +64 (0)9 923 4611
>
>


-- 
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

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20180212/9f1cd3fb/attachment.html>