Is it non-determinism that the order of the row/column of the coarsen matrix when using multigrid?

[lrtfm]

I am currently using multigrid with the coarse problem solved by PCLU, and the PC failed randomly with the error below

[ 0] Error reported by MUMPS in numerical factorization phase: INFOG(1)=-9, INFO(2)=36

Now it is sure that the non-deterministic results are caused by the pivoting and the renumbering of MUMPS. See the discussion here: https://lists.mcs.anl.gov/pipermail/petsc-users/2023-March/048136.html.

I am still curious if there may be non-determinism in the coarsening/refining process, which will result in a different order of the result matrix when every time I run the code.

[dham]

The mesh refinement and hence the prolongation and restriction operators are expected to be deterministic.

[lrtfm]

Thanks for your explanation. Best wishes, Zongze

…

On Mon, 6 Mar 2023 at 21:22, David A. Ham ***@***.***> wrote: The mesh refinement and hence the prolongation and restriction operators are expected to be deterministic. — Reply to this email directly, view it on GitHub <#2795 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAJDG3EYYZWJN2NIJEZRNATW2XQIHANCNFSM6AAAAAAVPSVKOE> . You are receiving this because you authored the thread.Message ID: ***@***.*** com>