google-deepmind/qtqp v0.0.4

v0.0.4

Repository: google-deepmind/qtqp

Tag: v0.0.4

Published: 2026-04-10T15:03:38Z

Prerelease: no

Release notes:

What's Changed

• always do 1 iterative refinement step by @bodono in https://github.com/google-deepmind/qtqp/pull/14
• remove alpha and sigma columns by @bodono in https://github.com/google-deepmind/qtqp/pull/15
• Equilibrate tweaking by @bodono in https://github.com/google-deepmind/qtqp/pull/17
• Extra testing by @bodono in https://github.com/google-deepmind/qtqp/pull/18
• switch to alternative tau solve, add test of equivalence by @bodono in https://github.com/google-deepmind/qtqp/pull/19
• reduce min static regularization to 1e-8 by @bodono in https://github.com/google-deepmind/qtqp/pull/20
• fix equilibration for zero rows and cols, add test by @bodono in https://github.com/google-deepmind/qtqp/pull/21
• silence cuda error by @bodono in https://github.com/google-deepmind/qtqp/pull/22
• simplified mu calculation by @bodono in https://github.com/google-deepmind/qtqp/pull/23
• Fix reproducibility issue in tests by @matteosantama in https://github.com/google-deepmind/qtqp/pull/24
• Fix sigma calculation by @bodono in https://github.com/google-deepmind/qtqp/pull/25
• Bump actions/checkout from 5 to 6 by @dependabot[bot] in https://github.com/google-deepmind/qtqp/pull/26
• Bump actions/upload-artifact from 5 to 6 by @dependabot[bot] in https://github.com/google-deepmind/qtqp/pull/27
• Bump actions/upload-artifact from 6 to 7 by @dependabot[bot] in https://github.com/google-deepmind/qtqp/pull/30
• llm optimizations by @bodono in https://github.com/google-deepmind/qtqp/pull/31
• Add UMFPACK and dense LU solvers, clean up LinearSolver API by @bodono in https://github.com/google-deepmind/qtqp/pull/32
• Expand test coverage by @bodono in https://github.com/google-deepmind/qtqp/pull/33
• Precompute constants and deduplicate step multiplication by @bodono in https://github.com/google-deepmind/qtqp/pull/34
• Add per-solver stats and timing to CI test output by @bodono in https://github.com/google-deepmind/qtqp/pull/36
• Optimize CuDSS solver: single GPU copy with in-place updates by @bodono in https://github.com/google-deepmind/qtqp/pull/38
• Add DENSE_LDLT and CUPY_DENSE solver backends by @bodono in https://github.com/google-deepmind/qtqp/pull/37
• Fix MUMPS solver and re-enable tests by @bodono in https://github.com/google-deepmind/qtqp/pull/41
• Use Gram/Cholesky reduction for dense solver backends by @bodono in https://github.com/google-deepmind/qtqp/pull/40
• Split solver backends into separate modules by @bodono in https://github.com/google-deepmind/qtqp/pull/42
• Add weekly CI cron job by @bodono in https://github.com/google-deepmind/qtqp/pull/43
• Tune Pardiso iparms for numerical stability by @bodono in https://github.com/google-deepmind/qtqp/pull/39
• attempt to hack in iparms before first factor/solve for pardiso by @bodono in https://github.com/google-deepmind/qtqp/pull/47
• Fix MUMPS solver with complex PETSc builds by @bodono in https://github.com/google-deepmind/qtqp/pull/50
• Add ACCELERATE linear solver backend (macOS) by @bodono in https://github.com/google-deepmind/qtqp/pull/49
• Enable Windows PARDISO test coverage by @bodono in https://github.com/google-deepmind/qtqp/pull/48
• Drop Python 3.10; require scipy>=1.16 by @bodono in https://github.com/google-deepmind/qtqp/pull/51
• Add pip ecosystem to Dependabot by @bodono in https://github.com/google-deepmind/qtqp/pull/53
• Replace pydiso with py-mkl-pardiso for PARDISO solver by @bodono in https://github.com/google-deepmind/qtqp/pull/52
• Use triangular KKT storage in direct solver by @bodono in https://github.com/google-deepmind/qtqp/pull/54
• Split set_kkt into set_kkt + update_diag by @bodono in https://github.com/google-deepmind/qtqp/pull/55
• Improve README linear solver guidance by @bodono in https://github.com/google-deepmind/qtqp/pull/56

New Contributors

• @matteosantama made their first contribution in https://github.com/google-deepmind/qtqp/pull/24

Full Changelog: https://github.com/google-deepmind/qtqp/compare/v0.0.3...v0.0.4