v3.2.0

What's Changed

New features

• multi-objective optimization by NDS and ParEGO in #87
• added utilities for benchmarking in #86

Improvements

• Update Variable in #82
• initial_data argument of range.Policy in #83

Full Changelog: v3.1.1...v3.2.0

v3.1.1

Changed from v3.1.0

Fixed a bug

• initial_data argument of range.Policy caused a runtime error (#83)
• range.Policy.write failed to find the maximum fx when multiple actions were given at the first call (#83)

v3.1.0

v3.1.0

Minor Updates

• Improvement of performance is achieved in pure python code that we implemented in Cython before [#79]
  • physbo-core-cython is no longer used even when it is installed
  • Thanks to @aoymt
• odat-se is added as optional-dependencies in pyproject.toml [#80]
  • python3 -m pip install 'physbo[odat-se] also installs ODAT-SE

v3.0.0

A New Major Version of PHYSBO!

Major Changes

PHYSBO v3 introduces three major changes:

• Cython code moved to external package: Code written with Cython has been moved to the external package physbo-core-cython [#77]
  • This simplifies the installation process of PHYSBO, particularly on Windows
  • The performance (speed) of PHYSBO is slightly reduced compared to PHYSBO v2
  • Users can achieve the original performance by installing the physbo-core-cython package
• Class names capitalized according to PEP8: Many class names have been capitalized to follow PEP8 (Python code style guide) [#73]
  • For example, the physbo.search.discrete.policy class becomes physbo.search.discrete.Policy
  • The old names are marked as deprecated, so users can still use them with warnings
• NumPy2 support: Full support for NumPy2 has been added [#70]
  • While PHYSBO v2 code worked with NumPy2, the required NumPy version was capped

New Features

• New permutation importance (PI) calculation methods [#72]
  • PI helps identify which features (parameters) are most influential for the objective functions
• New policy classes: physbo.search.range.Policy and range_multi.Policy [#76]
  • These policies perform Bayesian optimization directly in continuous parameter spaces
  • Since optimizing the acquisition function in continuous space is more challenging than in discrete space, we recommend using discrete optimization when possible

Other Updates

• Tutorial notebooks: Jupyter notebooks used as tutorials have been uploaded to ISSP datarepo [#75]
  • See the data/tutorial directory

Full Changelog: v2.2.0...v3.0.0

v2.2.0

What's Changed

New feature

• physbo.gp.core.sfs is added
  • a wrapper of physbo.gp.core.model with the scikit-learn compatible API (fit and predict) (#53)

performance improvement

• Optimized pareto.py for 3 objective TS search (#68)

v2.1.0

What's Changed

New features

• Enable get_post_fcov methods to return posterior covariance matrices of Gaussian Process by passing diag=False (#66, #67)

Full Changelog: v2.0.2...v2.1.0

v2.0.2

What's Changed

Bugfixes

• policy.load fails with MPI (#60, #61)
• discrete_multi.policy.random_search fails when initial_data was given (#64, #65)

Improvement of documentation

• Improvement of README (#62)
  • Fixed broken links, added links to tutorials, ...

Full Changelog: v2.0.1...v2.0.2

v2.0.1

Change from v2.0.0

Bugfixes

• To load a saved history fails (#52, #59)
• To write the last action to policy fails (#57, #58)

Full Changelog: v2.0.0...v2.0.1

v2.0.0

Important Notice

From v2.0.0, PHYSBO is distributed under the Mozilla Public License (MPL) v2.
v2.0.0より、 PHYSBO は Mozilla Public License (MPL) v2 のもとで公開・頒布されます。

This change makes it clear that code importing PHYSBO will not be affected by copyleft. However, if you modify and redistribute PHYSBO itself, copyleft will still apply, and you will need to redistribute it under MPL v2.
これにより、PHYSBO を import するコードについては、コピーレフトの影響を受けないことが明確になりました。一方、PHYSBO そのものを修正して再頒布する場合には、これまでと同様にコピーレフトが働き、MPL v2 の下で再頒布する必要があります。

What's Changed

• Change License to MPLv2 (#55)
• Set the upper bound of numpy version (#56)

Full Changelog: v1.1.1...v2.0.0

v1.1.1

Changed from v1.1.0

• Bug fixes
  • Use numpy<1.20 for installing physbo for python<=3.9 (#49)
    • This solves the problem of mismatches of the size of numpy objects
  • get_post_fmean, get_post_fcov, and get_score now work just after policy.write (interactive mode) (#42, #51)
  • Remove evaluated actions from loaded policy (#44, #45)
  • Remove actions on writing history (policy.write) (#43, #46)
    • Thanks @k-kitai for making a PR!

Full Changelog: v1.1.0...v1.1.1