[petsc-dev] DMPlexCreateBoxMesh Simplex Ordering?

Hapla Vaclav vaclav.hapla at erdw.ethz.ch
Wed Sep 16 08:48:33 CDT 2020



On 15 Sep 2020, at 19:38, Jacob Faibussowitsch <jacob.fai at gmail.com<mailto:jacob.fai at gmail.com>> wrote:

Hello all,

The ascii diagram showing the point numbering for DMPlexCreateBoxMesh seems off? There are duplicated numberings, and it does not seems to stack up against -dm_view ascii::ascii_info_detail. For example for a 2x2 2D mesh the bottom half is


11----6---12----7----13


 |\        |\        |


 | \    1  | \     3 |


10   0    11    1    12


 | 0   \   | 2   \   |


 |       \ |       \ |


 8----4----9----5----10


But the corresponding support output from dm_view is

Supports:
[0] Max support size: 6
[0]: 8 ----> 17
[0]: 8 ----> 19
[0]: 9 ----> 17
[0]: 9 ----> 18
[0]: 9 ----> 23
[0]: 9 ----> 27
[0]: 10 ----> 24
[0]: 10 ----> 25
[0]: 10 ----> 27
[0]: 11 ----> 18
[0]: 11 ----> 19
[0]: 11 ----> 21
[0]: 11 ----> 22
…

Which at least implies that face 4 = 17, face 10 = 19. Is this as a result of using a particular generator for the mesh? It was my understanding they all generate numberings in lexicographical CW ordering so should all produce the same numbering?

It's definitely wrong. Two points can have the same number only if they are on different processes. It would be more illustrative to have e.g. a 1 rank and 2 ranks cases there.

BTW GitLab issues should work again.

Vaclav


Best regards,

Jacob Faibussowitsch
(Jacob Fai - booss - oh - vitch)
Cell: (312) 694-3391


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20200916/b095f63d/attachment.html>


More information about the petsc-dev mailing list