BO in a mixed, multiobjective domain

This project aims to optimize a reaction with a mixed input domain and multiple objectives.

Code associated with the paper Multi-objective Bayesian optimisation using q-noisy expected hypervolume improvement (q NEHVI) for the Schotten–Baumann reaction.

Setup

1. Clone the repo

   git clone https://github.com/sustainable-processes/bo_mixed.git

2. Create a virtual environment and install the dependencies

   python3 - m venv .venv
   source .venv/bin/activate
   pip install -r requirements.txt

3. Run the code

   python main.py
   

You can see a full list of commands by running python main.py --help. Some examples:

# Use TSEMO as a strategy
python main.py --strategy MOBO

# Use TSEMO as a strategy
python main.py --strategy TSEMO

# Pass your own initialization data and don't run LHS initialization
python main.py --initialization-data-path data.csv --num-initial-experiments=0