Issues

• Roadmap: PR order, numerical feature backlog, and release checkpoints

  cai4cai/torchsparsegradutils#130 · Filter by author theo-barfoot opened on Oct 8, 2026

• Move iterative solvers and preconditioners out of utils into a solvers package

  enhancement
  cai4cai/torchsparsegradutils#129 · Filter by author tvercaut opened on Oct 8, 2026

• RFC: Consolidate dependency management in pyproject.toml and transition to uv

  cai4cai/torchsparsegradutils#119 · Filter by author aymuos15 opened on Sep 25, 2026
  2

• Validate BiCGSTAB initial_guess shape, dtype and device consistently

  cai4cai/torchsparsegradutils#107 · Filter by author theo-barfoot opened on Sep 22, 2026

• Fix repository-layout tests in installed-wheel Docker validation

  cai4cai/torchsparsegradutils#104 · Filter by author theo-barfoot opened on Sep 2, 2026

• Ignore max_tridiag_iter when n_tridiag is zero

  cai4cai/torchsparsegradutils#102 · Filter by author theo-barfoot opened on Sep 2, 2026

• Reduce linear_cg device synchronization in breakdown checks

  cai4cai/torchsparsegradutils#101 · Filter by author theo-barfoot opened on Sep 2, 2026

• Set up a GitHub App for guarded release automation

  cai4cai/torchsparsegradutils#98 · Filter by author theo-barfoot opened on Aug 20, 2026

• Benchmark torch.compile for linear CG update helpers

  cai4cai/torchsparsegradutils#96 · Filter by author theo-barfoot opened on Aug 19, 2026

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

  cai4cai/torchsparsegradutils#71 · Filter by author theo-barfoot opened on Jan 20, 2026
  1

  

• Modify custom torch.autograd.Functions to use setup_context instead of passing ctx as an argument to forward

  cai4cai/torchsparsegradutils#37 · Filter by author tvercaut opened on Jun 8, 2023
  1

• Provide pre-conditioners for iterative methods

  cai4cai/torchsparsegradutils#29 · Filter by author tvercaut opened on Dec 12, 2022
  1