Skip to content

Conversation

@ChrisRackauckas-Claude
Copy link
Contributor

No description provided.

@ChrisRackauckas ChrisRackauckas force-pushed the claude/open-linearsol-pr-019f2nZ8STSAH5PaVG3PVNep branch 2 times, most recently from f7663de to a293e0a Compare December 11, 2025 06:37
…ctorizations

- Add `has_concretization` import from SciMLOperators to enable checking operator concretization support
- Add `init_cacheval` methods for `AbstractSciMLOperator` types with sparse factorization algorithms:
  - KLUFactorization
  - UMFPACKFactorization
  - LUFactorization
  - CHOLMODFactorization
  - GenericFactorization
  - GenericLUFactorization
  - QRFactorization
  - NormalCholeskyFactorization
- When an operator has `has_concretization`, it is converted to an AbstractMatrix and the appropriate sparse solver is used
- Add tests for MatrixOperator and WOperator with sparse matrices
- Bump SciMLOperators lower bound to v1.13 (WOperator was added in v1.13.0)
@ChrisRackauckas ChrisRackauckas force-pushed the claude/open-linearsol-pr-019f2nZ8STSAH5PaVG3PVNep branch from a293e0a to b8cfa09 Compare December 11, 2025 06:40
Replace plain error() calls with a custom exception type when operators
that don't support concretization are passed to factorization algorithms.
Import has_concretization directly from SciMLOperators instead of
LinearSolve, since LinearSolve doesn't re-export it.
The init_cacheval methods need to return nothing when concretization
is not supported, so that the default algorithm selection can fall
back to other methods like Krylov solvers.
@ChrisRackauckas ChrisRackauckas merged commit 06d9f78 into SciML:main Dec 11, 2025
134 of 142 checks passed
@ChrisRackauckas ChrisRackauckas deleted the claude/open-linearsol-pr-019f2nZ8STSAH5PaVG3PVNep branch December 11, 2025 09:52
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.

3 participants