Conversation
pearsonca
left a comment
There was a problem hiding this comment.
Minor items; going to try it out w/ flepimop-demo
Perhaps these changes would be better suited for a separate PR. The failing CI is making it a bit annoying to verify the the actual code changes pass quality checks. Also, I do not think the premise for this change is true.
|
TimothyWillard
left a comment
There was a problem hiding this comment.
Minor code edits, larger issues about the CI changes. Is there any documentation for these mixing kernels? I do not really understand what they are or how to use them?
|
Addressed reviewer comments including reverting the |
This pull request introduces improvements to the operator wiring and kernel parameter handling in the op_engine, as well as enhancements to the test suite and developer tooling. The main changes include merging precomputed mixing kernels into the right-hand side (RHS) parameters, initial support for passing operator specifications (especially for IMEX methods), and updates to the development workflow to better separate provider and core tasks.
Core engine improvements:
mixing_kernelsexposed by a system into the RHS parameters, allowing systems to inject kernel values directly into the computation.Testing enhancements:
Added
_KernelStepper,_KernelSystem, and_ImexSystemtest helpers, along with new tests to verify that mixing kernels are passed correctly and that IMEX operator specs are accepted and functional.Developer tooling and workflow:
Updated the
justfileto better separate provider and core tasks, add "nosync" variants for faster CI, and useuvxfor tool-only tasks. The CI aggregate step is now more efficient and explicit.Minor cleanup:
reshape_for_axis_solveinmodel_core.py.Closes #8 in prep for adding multiple demos to flepi demo.