pymc-devs / pymc-devs/pytensor
Add Ops for specialized dot products with structured matrices
Open
Nobody has claimed this yet.
graph rewriting
help wanted
linalg
performance
- Dominant language
- Python
- Stars
- 644
- Forks
- 208
- Avg merge
- 2d 14h
- Merged PRs (30d)
- 16
Description
Description
BLAS/LAPACK offer a number of specialized dot products when matrices have special structure:
- Triangular matrix-vector product and matrix-matrix product
- Symmetrical matrix-vector product and matrix-matrix product
- Banded matrix-vector product
We can implement these as non-user facing ops, and rewrite into them when we can infer the special structure of matrices in a dot Op.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
No file or test entry point is named. Start by locating the existing dot Op and its rewrite or optimization path, then compare the linked BLAS/LAPACK operations for triangular, symmetric, and banded matrices. Done means non-user-facing operations exist for the listed structures and dot expressions rewrite to them when the structure can be inferred.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- performance
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100