All notable changes to Hyperactive are documented in this file.
- Add missing torch entries in README and documentation (#239)
- Update landing page and styling
- Change formatting and key features in README
- Rework entire README with mermaid diagram and updated links
- Fix links, add overview diagram hover effect
- Hide toc parents
- Fix
_score_paramsand add tests; fix errors masked bytry... exceptblock (#237) - Add
warningsto exception and fix bug - Override
_predict_proba
- Fix formatting
- Remove unused dependency
- Remove workflow
- Update CONTRIBUTING.md
- Update PULL_REQUEST_TEMPLATE.md
- Change
n_iterin examples
- Preload font, add animation of right sidebar
- Add explanations, change title styling, add links
- Rework user-guide, add diagrams
- Change API reference optimizer pages
- Add light/dark logo, fix logo font/text
- Cleanup README, update links
- Add link to legacy docs
- Add nav-bar logo, change title and sidebar
- Rework landing page: Quick Install, Examples, Integrations, Optimization backends
- Add multiple pages and small corrections
- Separate examples pages
- Add Python examples via
literalincludewith tests - Dynamically get supported Python versions, update CSS
- Add static content and readthedocs skeleton
- Add keywords and classifiers to
pyproject.toml - Ensure tests are not skipped
- Remove custom Docker container from CI, add Dockerfile
- Move doctests, add dependency, fix doctest
- Add
--no-cache-dirflag, run targeted tests first - Add auto-pr-label and draft-changelog workflows
- Add PR template
- Python 3.14 compatibility and Python 3.9 end-of-life (#202)
- Add PyTorch Lightning integration (#203)
skprointegration (#200)
- Fix cmaes optimizer tests
- Bump pytest from 9.0.1 to 9.0.2 (#218)
- Bump pytest from 8.4.2 to 9.0.1 (#206)
- Fix
TSCOptCVintegration for metric function input (#190)
- Add optuna optional import (#187)
- Minor improvements to README: white background for logo, link table formatting (#186)
- Complete v5 API redesign (#185)
- Change
BaseOptimizer.runmethod toBaseOptimizer.solve(#159) - Rename
toymodule tobench(#164) - Experiments: uniform call signature and terminology (#178)
- Refactor callbacks/catch parameter
- Remove
add_searchmethod
sktimeintegration for time series classification (#173)sktimeintegration for forecasters (#157)optunaoptimizer interface (#155)- Add optimization algorithms from GFO (#127)
- Parallelization backends for grid and random search (#150, #162)
- Allowing old function API to be passed as experiment (#152)
- Sign handling in experiments and optimization (#142)
- Quick testing utility
check_estimator(#130) - Implement sklearn-approved experiment class
- Add composite optimizer class (enabling parallel via
addmethod)
- Fix selection direction, scorer handling, and fit kwargs; resolve sktime doctest (#182)
- Fix
paramsignored inSktimeForecastingExperiment(#175) - Fix docstrings of benchmark toy experiments (#171)
- Fix sklearn v1.7 compatibility (#138)
- Lint entire repository using
pre-commit(#156) - Code quality job (#151)
- Raise
scikit-learnbound to<1.8(#145, #149) - Cleanup v4 (#165)
- Change
mastertomain(#166) - Restore sklearn integration tests (#167)
- Change examples to v5 (#168)
- Bump pytest from 8.4.0 to 8.4.2 (#129, #137, #181)
- Add support for numpy v2 and pandas v2
- Add testing for Python 3.12
- Transfer
setup.pytopyproject.toml - Change project structure to src-layout
- Read version from pyproject file
- Add Genetic Algorithm and Differential Evolution optimizers
- Add constrained optimization example to README
- Add linter
- Remove Python 3.5, 3.6, 3.7 from supported versions
- Update requirement files
- Add constrained optimization support (constraints parameter in API)
- Add constraint class with support in optimization strategies
- Add examples: constrained optimization, grid search, Lipschitz optimization, stochastic hill climbing, direct algorithm, downhill simplex, Powell's method, pattern search, spiral optimization
- Add early stopping support in optimization strategies
- Print additional results from the objective function
- Pass
early_stoppingparameter to optimization backend - Add type hints for basic API
- Log warning if getting max score fails (when all scores are NaN)
- Fix verbosity for progress-bar
- Remove empty lines on no verbosity (#71)
- Add optimization strategy feature with custom strategies
- Add tqdm progress-bar with description and postfix
- Add SpiralOptimization and LipschitzOptimizer
- Add Direct Algorithm
- Add Ray support for parallelization
- Add
timesparameter to.search_data()for eval and iter times - Add callbacks feature
- Add exception handling (catch parameter)
- Add
pass_throughparameter - Add SMBO warm start support in optimization strategies (#54)
- Create deepcopy of optimizer (#62)
- Extend Python versions to 3.11
- Add contributing.md
- Bump GFO requirement version
Initial v4 release.