Multi-objective optimization algorithm for expensive-to-evaluate function

Thompson sampling efficient multiobjective optimization (TSEMO) algorithm

This repository contains the source code for “Thompson sampling efficient multiobjective optimization” (TSEMO) algorithm [1].
The algorithm is designed for global multi-objective optimization of expensive-to-evaluate black-box functions. For example, the algorithm has been applied to the simultaneous optimization of the life-cycle assessment (LCA) and cost of a chemical process simulation [2]. However, the algorithm can be applied to other black-box function such as CFD simulations as well. It is based on the Bayesian optimization approach that builds Gaussian process surrogate models to accelerate optimization. Further, the algorithm can identify several promising points in each iteration (batch sequential mode). This allows to evaluate several simulations in parallel.
[1] Bradford, E., Schweidtmann, A.M. & Lapkin, A. J Glob Optim (2018). https://doi.org/10.1007/s10898-018-0609-2
[2] D. Helmdach, P. Yaseneva, P. K. Heer, A. M. Schweidtmann, A. A. Lapkin, ChemSusChem 2017, 10, 3632. https://doi.org/10.1002/cssc.201700927