Issues · cai4cai/torchsparsegradutils · GitHub

Labels Milestones

Change the default solver in sparse_generic_solve

#86

· tvercaut opened

on Jul 7, 2026

Add a dedicated `SparseGenericTransposeSolve` autograd function for general higher-order sparse solves

#84

· theo-barfoot opened

on Jul 7, 2026

`sparse_logsumexp` numerically stable log-sum-exp reduction for sparse tensors

#83

· tvercaut opened

on Jun 25, 2026

`test_jax_bindings.py` fails if optional jax dependency is not available

#82

· tvercaut opened

on Jun 15, 2026

Linear system solution from `minres` solver is not available for higher order derivatives due to `out=`

#81

· kaigu1997 opened

on Jun 10, 2026

`sparse_mm` should use `sparse.mm` by default for unbatched CSR matrices

#71

· theo-barfoot opened

on Jan 20, 2026

[Feature] Add dependency-free GMRES

#55

· tvercaut opened

on Sep 11, 2024

Modify custom `torch.autograd.Function`s to use `setup_context` instead of passing `ctx` as an argument to `forward`

#37

· tvercaut opened

on Jun 8, 2023

Add support for dense matrices in generic solvers

#30

· tvercaut opened

on Dec 12, 2022

Provide pre-conditioners for iterative methods

#29

· tvercaut opened

on Dec 12, 2022