Conversation
m4 doesn't use subshells to keep local variables local, and I think we were getting false positives from one macro's commonly-named "local" variable infecting another.
Even with the prefix we're not really safe unless we make sure to ignore any pre-existing variable settings.
This was useful for debugging
I don't want to rewrite 1.3.2 history since it was *almost* working, everything save for libMesh distcheck. Since we haven't actually changed any APIs we don't bother updating libtool version info this time.
This prepares all the build system changes from the new FE exceptions code and configuration.
GiudGiud
changed the title
When a tensor of AD Reals has every component sharing one derivative sparsity pattern (the usual case: all components depend on the same element DOFs), the sparsity union is identical for every component. Compute it once for the whole tensor instead of once per component, eliminating the per-component union scans and temporaries. Works for DynamicSparseNumberArray and SemiDynamicSparseNumberArray (MOOSE's ADReal derivative type) derivatives. - NumberArray/NumberVector::operator*= now route through a component_multiply_assign customization point (default: the per-component loop). shared_sparsity_tensor_multiply.h adds specializations for DualNumber over DynamicSparseNumberArray and SemiDynamicSparseNumberArray that use the batched one-union path, with a safe per-component fallback when the patterns are not shared. - shared_sparsity_tensor_contract.h provides fused_shared_sparsity_contract<DIM> for the matrix product C = A*B (e.g. a RankTwoTensor exposing operator()(i,j)): one union for the whole contraction, no temporaries, with a plain_tensor_contract fallback. - dualdynamicsparsenumberarray.h includes the multiply header last, so the component-wise fused path is active by default for AD sparse tensors. Per-thread reusable scratch keeps the batched paths allocation-free in steady state. Co-Authored-By: Claude <noreply@anthropic.com>
…ty AD tensors Lets a generic tensor operator (e.g. libMesh TypeTensor::operator*) use the shared-sparsity fast path: fused_shared_sparsity_contract and fused_shared_sparsity_rotate are safe to call for any entry type. A has_sparse_derivatives trait dispatches to the fused one-union path for AD entries backed by a dynamic sparse derivative array and to the plain implementation otherwise, with a run-time fallback to plain when component patterns are not shared. - New sparse_pattern_compare.h holds same_index_pattern so consumers can use these routines without pulling in the DualNumber/container headers. - shared_sparsity_tensor_contract.h: trait + tag-dispatched plain/fused contract (plain accumulates in the output/supertype so mixed-type products are correct) and fused_shared_sparsity_rotate for A = R*A*R^T, one union over P_R u P_A. Co-Authored-By: Claude <noreply@anthropic.com>
…rank4) fused_shared_sparsity_rank4_times_rank2 (C_ij = A_ijkl b_kl) and fused_shared_sparsity_rank4_times_rank4 (C_ijkl = A_ijpq B_pqkl) contract over the flat N^2 storage and, when every entry of each operand shares one derivative sparsity pattern, compute the union once per product instead of once per scalar multiply. Same trait-based dispatch as the rank-2 routines: fused for sparse-AD entries, plain otherwise, with a run-time fallback when patterns are not shared. Co-Authored-By: Claude <noreply@anthropic.com>
Member
|
Isn't that what draft is for? 😄 |
GiudGiud
marked this pull request as draft
September 18, 2026 20:50
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
3 participants
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.
written by Claude.
I ll report on the timings and find out if it compiles soon