This repository was archived by the owner on Jan 31, 2025. It is now read-only.
Firedrake #20
Merged
Firedrake #20
Conversation
This file contains 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
My approach really does seem to be about the right thing to do. Substantial cleanup is required and there are a couple of fixes I need to do.
Still facing convergence issues in Firedrake.
This means that we are correctly generating sparsity code.
Things still work in Firedrake-land.
Firedrake assign works OK. Can now call assignment operations as well as loops.
Trying to generate code without needing the data structures to exist. The generated code looks right for the moment. The final piece for the single-cell wrapper (I think) is to enable one to pass loop indices in as arguments. I think all that is required for that to work is to do a bit extra when we unwrap loop contexts. Loop indices should be arrays of integers rather than single ints because we can happily pass around multi-indices.
* About 8 tests currently failing, but they are non-essential for Firedrake. * Substantially improve PetscMat implementation. Row and column axes must now be distinctly labelled. This avoids confusing failures. * Add `Pack` class because sometimes `__getitem__` isn't enough (for example if you want a specific DoF layout).
* Made some changes to get high-order quads working in Firedrake. Specifically it is convenient to label maps so that the resulting set of axes is addressable. * I have almost definitely broken a large number of tests, but fixing them should be straightforward. * I have also definitely broken the code that previously allowed for interchanging axes. This should be easy enough to fix but now we don't simply do the naive thing.
* Always renumber temporaries now. * That bit of the code is nicer now, but I should reimplement the renaming expression transformation as we could reduce the number of arguments that get passed about.
This will let us do things with, for example, interior facets. We want to be able to address this subset of all facets as if it were a full axis component itself.
* Make the difference between indices and loop_exprs more explicit. * About to try dropping target_path.
The maps for PETSc are now malformed. Not sure how to approach it.
Previous bits of Firedrake that worked still do, but I am having a lot of trouble getting the matrix tests to pass. I think I am building my maps badly in some way.
Need to clean up source and target bits and pieces as a matter of urgency.
In particular sparsity construction is now definitely right. I just have to fix the interior facet integral issue.
Also error if appropriate.
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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.
No description provided.