Releases: dwavesystems/dwave-samplers
Releases · dwavesystems/dwave-samplers
1.7.0
1.6.0
1.5.0
New Features
- Add
PathIntegralAnnealingSampler, aSimulatedAnnealingSampler
style wrapper for the dwave-pimc repository. This supports
simulation of path-integral dynamics and equilibrium sampling of
quantum Boltzmann distributions (King et al., 2021). - Add
RotorModelAnnealingSampler, aSimulatedAnnealingSampler
style code for rotor model simulation. This supports a standard
classical algorithm for approximation of spin dynamics (Shin et
al., 2014).
Bug Fixes
- Add all keyword args of
SimulatedAnnealingSampler.sampleto
SimulatedAnnealingSampler.parameters. This allows effective use of
dimod.Sampler.remove_unknown_kwargs
1.4.0
New Features
- Support Python 3.13.
Upgrade Notes
- Drop support for Python 3.8.
1.3.0
New Features
- Support, and build with, NumPy 2.0
Upgrade Notes
- Require
dimod>=0.12.13. - Require
NetworkX>=3.0.
1.2.0
New Features
- Build C++ extensions with Cython 3.
- Build wheels for Python 3.12.
Upgrade Notes
- No longer build wheels for Python 3.7.
1.1.0
New Features
- Add timing information for
SteepestDescentSolver. - Add timing information for
SimulatedAnnealingSampler. - Add
randomize_orderkeyword argument to theSimulatedAnnealingSampler.sample()method. This controls whether the variables are updated in a random order or not. - Add
proposal_acceptance_criteriakeyword argument to theSimulatedAnnealingSampler.sample()method. This controls whether the sampler will use Gibbs or Metropolis updates.
Upgrade Notes
- Build extensions with C++17.
Bug Fixes
- Fix
ImportErrorbeing raised on Windows when otherdwavenamespace packages are installed afterdwave-samplers. See #57.
1.0.0
New Features
- Build wheels for Python 3.11
Upgrade Notes
- Change the default value of the
TreeDecompositionSampler.sample()betakeyword argument to 1.
Bug Fixes
- Fix greedy and simulated annealing performance in wheels compiled for linux aarch64 systems.
- Fix the energy in sample sets returned by
PlanarGraphSolver.
- Fix several errors and inconsistencies in the calculation of the simulated annealing algorithm's beta schedule. Note that for very small problems this may result in slower execution, because the schedule no longer does a discretization of 1000 sweeps at each beta. See dwavesystems/dwave-neal#91.
1.0.0.dev2
New Features
- Add
PlanarGraphSolver, an exact solver for planar Ising problems with no linear biases.
1.0.0.dev1
Bug Fixes
- Allow
num_sweeps=0for simulated annealing. This was incorrectly disabled in1.0.0.dev0.