Skip to content

build(deps): bump quimb from 1.14.0 to 1.16.0 - #731

Open
dependabot[bot] wants to merge 1 commit into
mainfrom
dependabot/uv/quimb-1.16.0
Open

dependabot[bot] wants to merge 1 commit into
mainfrom
dependabot/uv/quimb-1.16.0

Conversation

@dependabot

@dependabot dependabot Bot commented on behalf of github Nov 16, 2026

Copy link
Copy Markdown
Contributor

Bumps quimb from 1.14.0 to 1.16.0.

Release notes

Sourced from quimb's releases.

v1.16.0

Highlights for this release include: 2D periodic local expectations, fermionic and abelian compression method support and a 1D compression guide.

Breaking changes

Decomposition and compression defaults

  • array_split and low-level rank-revealing decompositions now default to cutoff_mode="rel" instead of "rsum2". 1D zipup compression also defaults to "rel".
  • Randomized SVD and batched decompositions reject active cumulative cutoff modes. Use "abs" or "rel" instead.
  • tensor_network_1d_compress with an arbitrary-geometry method always returns a 1D network, inserting identities along long-range bonds. To keep the geometry, call tensor_network_ag_compress directly, as 2D boundary contraction now does for periodic boundaries, which therefore need an arbitrary-geometry method.

Hilbert spaces and Hubbard models

  • HilbertSpace: site ordering is now immutable. set_ordering raises TypeError. Use with_ordering to create a space with a different ordering.
  • fermi_hubbard_from_edges: default to order="interleaved", alternating spins at each coordinate to avoid MPO bond dimensions growing with system size, with slightly slower matrix-vector products. Rebuild anything keyed by rank or flat configuration. Use order="blocked" for the old layout.

Partial traces and local expectations

Dependencies

  • Require autoray v0.10.0 or newer and cotengra v0.8.3 or newer.

Enhancements

Tensor decompositions

  • qr_via_cholesky supports repeated orthogonalization with passes.
  • Add method="qr:auto", which tries Cholesky QR for suitable matrices and uses standard QR if it fails or loses too much orthogonality.
  • Split functions accept cutoff="auto": 1e-10 for exact decompositions, disabled for randomized SVD.
  • svd_rand_truncated: support "abs" and "rel" cutoffs on the sketched spectrum, and noise_dist="normal" or "rademacher" for the random sketch.
  • Batched "abs" and "rel" cutoffs treat blocks as one spectrum and retain a common bond dimension. Both modes support renormalization, including for batched decompositions.

1D compression and MPS

  • Add successive deterministic compression (method="sdc"), sdc-oversample, and randomized variants sdcr and sdcr-oversample, supporting fermionic networks, based on https://arxiv.org/abs/2601.19650.
  • direct, dm, zipup, sdc, sdcr, src, srcmps, fit, and their oversampling variants handle long-range bonds directly, supporting symmray abelian and fermionic tensors (for srcmps and fit, with a supplied tn_fit).
  • dm uses QR-SVD when a site's density matrix would exceed its rank bound, reducing layered PEPS norm boundary contraction costs (https://arxiv.org/abs/2406.09769), including for fermionic networks.
  • src, srcmps, fit, and their oversampling variants accept a seed or random generator and match the backend's device and dtype. By default, use the backend's global random state.
  • tensor_network_1d_compress defaults to cutoff="auto", choosing the cutoff for each method: 0.0 for src and srcmps, disabled for sdcr with its default randomized SVD, and 1e-10 otherwise.
  • zipup, sdc, and sdcr oversampling accept cutoff_mode_oversample. Zipup also accepts cutoff_oversample="auto". All 1D oversampling methods accept compress_opts_final to configure the final direct sweep separately.
  • fit with bsz=1 now supports fermionic tensor networks, including odd-parity tensors.
  • Add MatrixProductState.from_product, also used by MPS_product_state, with support for block-sparse single-site vectors.
  • Compression functions accept tag groups in site_tags; see parse_site_tag_groups. The 1D, 2D, and arbitrary-geometry wrappers also accept custom method callables.

... (truncated)

Changelog

Sourced from quimb's changelog.

v1.16.0 (2026-11-09)

Breaking changes (1.16.0)

Decomposition and compression defaults

  • array_split and low-level rank-revealing decompositions now default to cutoff_mode="rel" instead of "rsum2". 1D zipup compression also defaults to "rel".
  • Randomized SVD and batched decompositions reject active cumulative cutoff modes. Use "abs" or "rel" instead.
  • tensor_network_1d_compress with an arbitrary-geometry method always returns a 1D network, inserting identities along long-range bonds. To keep the geometry, call tensor_network_ag_compress directly, as 2D boundary contraction now does for periodic boundaries, which therefore need an arbitrary-geometry method.

Hilbert spaces and Hubbard models

  • HilbertSpace: site ordering is now immutable. set_ordering raises TypeError. Use with_ordering to create a space with a different ordering.
  • fermi_hubbard_from_edges: default to order="interleaved", alternating spins at each coordinate to avoid MPO bond dimensions growing with system size, with slightly slower matrix-vector products. Rebuild anything keyed by rank or flat configuration. Use order="blocked" for the old layout.

Partial traces and local expectations

Dependencies (1.16.0)

  • Require autoray v0.10.0 or newer and cotengra v0.8.3 or newer.

Enhancements (1.16.0)

Tensor decompositions

  • qr_via_cholesky supports repeated orthogonalization with passes.
  • Add method="qr:auto", which tries Cholesky QR for suitable matrices and uses standard QR if it fails or loses too much orthogonality.
  • Split functions accept cutoff="auto": 1e-10 for exact decompositions, disabled for randomized SVD.
  • svd_rand_truncated: support "abs" and "rel" cutoffs on the sketched spectrum, and noise_dist="normal" or "rademacher" for the random sketch.
  • Batched "abs" and "rel" cutoffs treat blocks as one spectrum and retain a common bond dimension. Both modes support renormalization, including for batched decompositions.

1D compression and MPS

  • Add successive deterministic compression (method="sdc"), sdc-oversample, and randomized variants sdcr and sdcr-oversample, supporting fermionic networks, based on https://arxiv.org/abs/2601.19650.
  • direct, dm, zipup, sdc, sdcr, src, srcmps, fit, and their oversampling variants handle long-range bonds directly, supporting symmray abelian and fermionic tensors (for srcmps and fit, with a supplied tn_fit).
  • dm uses QR-SVD when a site's density matrix would exceed its rank bound, reducing layered PEPS norm boundary contraction costs (https://arxiv.org/abs/2406.09769), including for fermionic networks.
  • src, srcmps, fit, and their oversampling variants accept a seed or random generator and match the backend's device and dtype. By default, use the backend's global random state.
  • tensor_network_1d_compress defaults to cutoff="auto", choosing the cutoff for each method: 0.0 for src and srcmps, disabled for sdcr with its default randomized SVD, and 1e-10 otherwise.
  • zipup, sdc, and sdcr oversampling accept cutoff_mode_oversample. Zipup also accepts cutoff_oversample="auto". All 1D oversampling methods accept compress_opts_final to configure the final direct sweep separately.
  • fit with bsz=1 now supports fermionic tensor networks, including odd-parity tensors.
  • Add MatrixProductState.from_product, also used by MPS_product_state, with support for block-sparse single-site vectors.
  • Compression functions accept tag groups in site_tags; see parse_site_tag_groups. The 1D, 2D, and arbitrary-geometry wrappers also accept custom method callables.

Partial traces, environments, and 2D contraction

... (truncated)

Commits
  • faee02f update changelog for v1.16.0
  • 2e65cc5 Respect tol parameter in decomp print mode
  • 43a3c80 add tn_fit kwarg to fit-oversample method
  • 7742414 fix some equalize_norms handling
  • 1a08bec informative wrong compress method error + remove cotengra version mention
  • 111e105 Bump the dependencies group with 2 updates
  • 1b64353 fix petsc/slepc call for newer versions
  • a79e66a pixi update incl cotengra 0.8.3
  • 4132cd1 tn2d compute_environments: add check_zero for dense sweep
  • 67ea9a8 tn.contract: expose check_zero option and fix some exponent handling
  • Additional commits viewable in compare view

Dependabot compatibility score

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.


Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

  • @dependabot rebase will rebase this PR
  • @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
  • @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
  • @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
  • @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
  • @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

Bumps [quimb](https://github.com/jcmgray/quimb) from 1.14.0 to 1.16.0.
- [Release notes](https://github.com/jcmgray/quimb/releases)
- [Changelog](https://github.com/jcmgray/quimb/blob/main/docs/changelog.md)
- [Commits](jcmgray/quimb@v1.14.0...v1.16.0)

---
updated-dependencies:
- dependency-name: quimb
  dependency-version: 1.16.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
@dependabot dependabot Bot added dependencies Pull requests that update a dependency file preview-benchmark-results Tag a PR to have it run benchmarks and compare to baseline. labels Nov 16, 2026

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

dependencies Pull requests that update a dependency file preview-benchmark-results Tag a PR to have it run benchmarks and compare to baseline.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

0 participants