[petsc-dev] DMPlex overlap redistribution

[Adrian Croucher]

hi Matt,

On 5/03/24 1:40 am, Matthew Knepley wrote:
> Let's start with the distribution. I think the problem may be here. 
> Even with dual porosity cells, I don't think we should have cells in 
> the pointSF, just the internal face between cells. Do you agree?

You mean the re-distribution? If you are removing the overlap cells and 
then recreating them after redistribution, then I agree, it should be 
just faces in the pointSF.

After I've done the initial distribution (with overlap 1), then added 
the dual porosity cells, faces etc., I create a new pointSF for the 
resulting DMPlex:

https://urldefense.us/v3/__https://github.com/waiwera/waiwera/blob/8bd5e1df1b3eaa728c689d2ec4d8ed35a7ca64f6/src/mesh.F90*L3412__;Iw!!G_uCfscf7eWS!b5nsxRn6Fs2vSvKgF7HaqlN5nQ3ywLAgC1i_djUVTvPYTHRiuQ1yy4mSMNTshFhfhgzBaRDRm7gKT3voRf5jCMqS3DNJK1A$ 

Mostly that is just updating the original pointSF because the original 
DMPlex points have been shifted to make space for the new cells, faces etc.

But if there is something not quite right in there, perhaps that could 
cause trouble during the subsequent redistribution?

- Adrian

-- 
Dr Adrian Croucher
Senior Research Fellow
Department of Engineering Science
Waipapa Taumata Rau / University of Auckland, New Zealand
email:a.croucher at auckland.ac.nz
tel: +64 (0)9 923 4611
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20240305/cabfe359/attachment.html>