[petsc-users] Loading labels from a hdf5 mesh file in parallel

Matthew Knepley knepley at gmail.com
Wed Nov 30 07:55:43 CST 2022


On Thu, Mar 3, 2022 at 10:33 PM Yi Jiang <yi.ethan.jiang at gmail.com> wrote:

> Dear Petsc developers,
> We are trying to use the HDF5_XDMF parallel I/O feature to read/write
> unstructured meshes. By trying some tests, we found that the topology and
> geometry data (i.e., cells and vertices) can be efficiently loaded in a
> scalable way, which is very impressive! However, we also found that all
> labels (such as `cell sets', `face sets') in the .h5 file are ignored (the
> .h5 file was converted from an EXODUSII mesh, by using a PetscViewer with
> PETSC_VIEWER_HDF5_XDMF format). Hence, we are wondering, does the latest
> Petsc also support to import these labels in parallel? In particular,
> we would like to redistribute the mesh after it is parallelly loaded by the
> naive partition. If so, could you please show me where to find an example
> to learn the techniques?
> Thank you very much for devoting the continuous efforts to the community
> and keep developing these wonderful features.
>

Sorry, this mail got lost in my inbox. I believe we have fixed this now,
but I would be willing to incorporate any tests that
are needed for the operations you want.

  Thanks,

     Matt


> Best regards,
> YJ
>


-- 
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.cse.buffalo.edu/~knepley/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20221130/5e89325d/attachment.html>


More information about the petsc-users mailing list