This code repository accompanies the paper "A Unified Framework for Bayesian Optimization under Contextual Uncertainty".

The Python packages used in the experiments are listed in environment.yml, from which a fresh conda environment can be created.

To run all experiments, simply run job_creator.py to create a list of jobs at jobs/jobs.txt, then run job_runner.pyto run them sequentially. Alternatively, to run a specific experiment,

python exp.py [task] [distance_name] [unc_obj] [acquisition] [seed]

where the valid strings for each argument can be found at the top of job_creator.py.