-
Notifications
You must be signed in to change notification settings - Fork 579
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Tpetra::Map should check for duplicate GIDs on a calling process #157
Comments
Periodic boundary conditions in Nalu are implemented by a "pseudo-node" approach: each slave node knows the global identifier of its' master, and only master (and regular) nodes contribute rows to the system matrix. When assembling the right-hand-side or matrix, contributions to slave nodes are redirected to the rows/columns associated with their master nodes. When there is more than one slave node referencing a master node, the code was introducing a new local ID for each slave node. This change checks if the master node has already been visited and re-uses the existing local ID (previously, multiple global ID/local ID pairs were sent to Tpetra which could cause a failure). The new code is a bit simpler in that it directly sorts the global ID vectors instead of the vector of nodes, from which the global ID vectors were then extracted. * add TGV IC for more detailed testing of periodic bcs
We've debated this back and forth. I've even heard arguments that Tpetra::Map should support duplicate GIDs on a single process, because that would make some kinds of code (e.g., periodic boundaries) the same no matter how GIDs are divided among processes. In any case, if we're building a hash table anyway, it's not hard to check for duplicates. |
@trilinos/tpetra |
Please refer also to discussion of the original issue in Bugzilla (Bug 6475): https://software.sandia.gov/bugzilla/show_bug.cgi?id=6475 I've closed that Bug; further discussion should take place here. Thanks Stefan! |
I'm subsuming Bugzilla Bug 6365 ("Multiple (duplicate, repeat, repeated) local indices per global index on a process don't work") into this issue. |
This issue has had no activity for 365 days and is marked for closure. It will be closed after an additional 30 days of inactivity. |
This issue was closed due to inactivity for 395 days. |
See spdomin/Nalu@b11ced4
The text was updated successfully, but these errors were encountered: