Repository navigation
Set mesh ghosting functor periodic boundaries - #4649
Merged
Merged
Conversation
We need to set PBCs automatically, but in cases where there's already a setting we'd probably best avoid overwriting it until we can handle multiple sets.
We can query from the const iterators but if we want to set_periodic_boundaries that's non-const.
This gets us periodic point neighbors, which we need to be able to examine for our current periodic constraints algorithm to assign a consistent constraint ordering.
lindsayad
approved these changes
Nov 6, 2026
lindsayad
left a comment
Member
There was a problem hiding this comment.
an infinite improvement over the previous automatic handling of periodic bc geometric ghosting in the library
roystgnr
enabled auto-merge
November 6, 2026 18:58
Member
Author
|
Huh. Auto-merge didn't wait for the dbg distributed sweeps to finish. That was unexpected. They're 3/4 of the way done now that I check, though, and that includes passing the recipe step that's been failing for devel->master merges, so probably best that it merged ASAP. |
Member
|
lol that's weird |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes #1728 for me.
This is a big improvement over the status quo. Our new DirichletConstraintValues test, instead of breaking on distributed meshes on 11 ranks (and IIRC 23 and maybe somewhere higher) now passes on at least 1 through 32 ranks (probably exhaustive, for a mesh with 32 elements). systems_of_equations_ex9 now passes on the previously failing 21 rank case (and every other count from 1 to 64). I'll start a 1-to-64 dbg sweep testing everything next.
This is also obviously a hack. It won't be sufficient in cases where we have multiple systems and there's a periodic boundary geometry that exists on later systems but not the first system; those cases will still have to fall back on our existing distributed periodic constraint calculation mechanism, "dumb luck".
But @lindsayad tells me "I’ve removed the dependence of the constraint elimination PRs on this PR so there is no rush to me to resolve this", and I've mostly been in a panic over the devel→master failures from
DirichletConstraintValuesthat should be fixed here, so I'm no longer in a rush either. Assuming this doesn't regress anything elsewhere, it should give us time to continue the discussion in #4637 and come up with the right fix for distributed-mesh PBCs. ...or the right fixes, plural? I'm beginning to think that "tell every mesh ghosting functor about every one of multiple PBC sets" and "change the PBC constraint calculations to work without periodic point neighbors" are both things we should do... probably in that order, since the first fix will be much more straightforward than the second.