0.6.12

New Features

• Add support for cubic lattice embeddings on Zephyr to make_origin_embedding. See #295.

Fixes

• Remove ambiguous xrefs to reduce build warnings. See #294.
• Improve cubic lattice embedding on Zephyr, strengthen tests. See #296.

0.6.11

New Features

• Add Python 3.12 support. See #289.

• Add support for Kings graph in large neighborhood search. See #290.

• Enable optional metadata tracking in LNLS (LatticeLNLS and LatticeLNLSSampler). See #292.

Fixes

• Fix docs build. See #291.

• Use more specific assertions and assertEqual. See #293.

Upgrade Notes

• Drop Python 3.7 support and refresh CI. See #288.

0.6.10

New Features

• Add Python 3.11 support. See #285.

Fixes

• Adapt SA concurrency test to new Neal default schedule. See #278.

• Add lattice ref solver to docs. See #282.

• Stop using deprecated numpy type aliases. See #283.

• Fix async multiprocess queue logging test. See #286.

0.6.9

New Features

• Add LatticeLNLS reference workflow for larger-than-QPU lattice-structured problems.

• Add LatticeLNLSSampler, a dimod-compatible hybrid decomposition sampler for problems of lattice structure.

• Add make_origin_embeddings decomposer that creates optimal embeddings for a QPU or cubic lattice.

Fixes

• Support unorderable variables in energy based decomposer (fixes #273).

0.6.8

Fixes

• Support all structured samplers in SubproblemCliqueEmbedder (#271)

0.6.7

Fixes

• Fixed State.from_sample to accept SampleSet. Fixed KerberosSampler to accept SampleSet for initial state (#268)

Upgrade Notes

• Drop Python 3.6 support (#269)

0.6.6

New Features

• Support dimod<0.13.0 (#266)

Fixes

• Docs refresh (#262)

0.6.5

New Features

• Add hybrid.Log, a block to log user-define state metrics to a JSON file, python logger, or in-memory queue or list. See #257.

• Add convenience JSON encoders for commonly used NumPy (hybrid.utils.NumpyEncoder) and Ocean types (hybrid.utils.OceanEncoder). See #257.

Fixes

• Clarify docs, #142.

Upgrade

• Drop dimod 0.8.x support. Oldest supported dimod version is now 0.9.6. See #259.

0.6.4

New Features

• make_origin_embeddings problem decomposing utility that creates optimal embeddings for a lattice.
  Embeddings are compatible with the topology and shape of D-Wave QPU samplers (#252).
• SublatticeDecomposer that produces lattice-structured subproblems (#252).

Fixes

• SteepestDescentProblemSampler now actually uses greedy.SteepestDescentSolver (#250).

0.6.3

Fixes

• Support dimod 0.10 (#248)