A Python package to perform sequential Bayesian experimental design for implicit models via mutual information, as in https://arxiv.org/abs/2003.09379.
Create and activate a virtual environment, then pip install the package. For example, with conda:
conda create -n env python=3.6
conda activate env
Then to install the package, change directory to the root of the package and:
pip install .
conda install -c conda-forge glmnet
conda install scikit-learn=0.21
Unfortunately, glmnet is currently importing a deprecated module of scikit-learn, which is why an older version of scikit-learn needs to be installed.