Parallel and Distributed Thompson Sampling for Large-scale Accelerated Exploration of Chemical Space

TL;DR

This paper introduces a scalable parallel and distributed Thompson sampling method that significantly improves large-scale Bayesian optimization for accelerating chemical space exploration, outperforming existing methods in scalability and efficiency.

Contribution

The paper presents PDTS, a scalable parallel and distributed Thompson sampling approach that effectively handles large-scale Bayesian optimization for chemical space exploration.

Findings

PDTS performs comparably to parallel expected improvement in small problems.

PDTS outperforms other scalable baselines in large-scale settings.

PDTS enables efficient large-scale parallel Bayesian optimization.

Abstract

Chemical space is so large that brute force searches for new interesting molecules are infeasible. High-throughput virtual screening via computer cluster simulations can speed up the discovery process by collecting very large amounts of data in parallel, e.g., up to hundreds or thousands of parallel measurements. Bayesian optimization (BO) can produce additional acceleration by sequentially identifying the most useful simulations or experiments to be performed next. However, current BO methods cannot scale to the large numbers of parallel measurements and the massive libraries of molecules currently used in high-throughput screening. Here, we propose a scalable solution based on a parallel and distributed implementation of Thompson sampling (PDTS). We show that, in small scale problems, PDTS performs similarly as parallel expected improvement (EI), a batch version of the most widely used BO heuristic. Additionally, in settings where parallel EI does not scale, PDTS outperforms other scalable baselines such as a greedy search, $\epsilon$-greedy approaches and a random search method. These results show that PDTS is a successful solution for large-scale parallel BO.