Skip to content

Conversation

Ig-dolci
Copy link
Contributor

@Ig-dolci Ig-dolci commented May 5, 2025

Description

This PR handles the issue with the incorrect checkpointing of Dirichlet boundary conditions. Currently, during tape recomputation, the user-defined boundary condition (DirichletBC._original_arg) is being updated. This leads to the issue #4206.
In that issue, James suggested an approach to checkpoint the Dirichlet boundary condition. This PR implements that suggestion and includes a test to verify the fix.
This change fixes issue #4206.

@connorjward
Copy link
Contributor

To fix CI I think you need to rebase on top of current master.

@dham dham enabled auto-merge (squash) May 7, 2025 15:17
@angus-g
Copy link
Contributor

angus-g commented May 21, 2025

It still seems like this should pass the CI. It would be nice to see it merged to help unblock us a bit, if that's possible?

@dham dham merged commit c579a94 into master May 21, 2025
7 checks passed
@dham dham deleted the dolci/checkpoint_bcs branch May 21, 2025 09:44
angus-g pushed a commit that referenced this pull request Jun 4, 2025
Ensure that restoring a BC from checkpoint returns a new BC and doesn't incorrectly modify the program variable BC.

---------

Co-authored-by: David A. Ham <[email protected]>
Co-authored-by: Pablo Brubeck <[email protected]>
connorjward pushed a commit that referenced this pull request Jun 4, 2025
Ensure that restoring a BC from checkpoint returns a new BC and doesn't incorrectly modify the program variable BC.

---------

Co-authored-by: Daiane Iglesia Dolci <[email protected]>
Co-authored-by: David A. Ham <[email protected]>
Co-authored-by: Pablo Brubeck <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

BUG: Disk checkpointing with tape.timestepper() causes incorrect BC evaluation at taping.
5 participants