[petsc-dev] Gmsh files - PETSc label names and values for physical groups

Jed Brown jed at jedbrown.org
Tue May 3 09:45:49 CDT 2022

See -dm_plex_gmsh_use_regions.


There are a couple issues. One is that a number of functions take a label and a list of values. For example, we have meshes of a bunch of sand grains (e.g., segmented from CT images), where each grain is a different label value because postprocessing wants to distinguish it, but the same material. Also, it would be nice to be able to iterate over surfaces and volumes specifically, rather than all labels. 

"Aagaard, Brad T via petsc-dev" <petsc-dev at mcs.anl.gov> writes:

> PETSc developers,
> Is there a reason why the PETSc Gmsh reader does not just set the PETSc label name to the name of the physical group and set the PETSc label value to 1? 
> From a user perspective, it would be nice to only have to specify one identifier for a physical group (preferably a name) rather than always specify two identifiers (name and value). This is what we do for our PyLith Exodus reader (nodeset names are mapped to label names and the label value is set to 1).
> PETSc maps the Gmsh physical group name and tag into a label name and label value. This seems redundant because from what I can tell, physical groups in Gmsh at a given dimension need both a unique name and a unique tag. That is, no two physical groups at a given dimension can have the same name and different tags nor can they have different names and the same tag.
> Thanks,
> Brad

More information about the petsc-dev mailing list