Source code for "Efficient Bayesian Experimental Design for Implicit Models" - https://arxiv.org/abs/1810.09912
The easiest way to install all the required packages is via Anaconda.
Create an anaconda environment, called <env_name>, that uses Python 3.6 and not 3.7, as gpyopt currently fails to work with that version.
conda create <env_name> python=3.6
conda activate <env_name>
Then install the relevant packages via the anaconda distribution and pip:
conda install numpy scipy matplotlib
conda install joblib==0.11
pip install glmnet
pip install gpyopt
All the critical required packages are:
- python v3.6
- numpy, scipy, matplotlib
- joblib v0.11
- glmnet
- gpyopt