Installation#

Layer 1: Use

Requirements#

See pyproject.toml for the exact versions of the dependencies.

Install with pip as follows. This installs only the minimal dependencies.

pip install saealib

To install with all dependencies:

pip install "saealib[all]"

See here for the available options.

Specify a version as follows.

pip install "saealib==X.Y.Z"
# example:
pip install "saealib==0.1.0"

Specify the latest pre-release version as follows.

pip install --pre saealib

Warning

The API of pre-release versions may change or be removed without notice.

Some packages are enabled by adding extra dependencies. Install these dependencies at the same time by specifying options as follows.

pip install "saealib[opt1,opt2,...]"
# example:
pip install "saealib[sklearn,parallel]"

To install all dependencies:

pip install "saealib[all]"

All options are shown in the following table.

Option	Adds
`sklearn`	scikit-learn-based surrogate models
`xgboost`	XGBoost surrogate model
`lightgbm`	LightGBM surrogate model
`torch`	PyTorch-based components
`pymoo`	`PymooCrossover`/`PymooMutation`/`PymooAlgorithm`/`PymooProblem` adapters
`parallel`	Parallel evaluation via joblib
`deap`	DEAP algorithm and operator adapters
`viz`	Matplotlib-based result and history plots
`tqdm`	tqdm progress bars for experiment sweeps
`rich`	Rich progress bars for experiment sweeps
`hdf5`	HDF5 trial-result storage and loading via h5py
`all`	All optional dependencies listed above

Once installation is complete, check the version with the following command.

python -c "import saealib; print(saealib.__version__)"