Targeted materials discovery using Bayesian algorithm execution

TL;DR

This paper introduces a flexible Bayesian optimization framework for targeted materials discovery, enabling efficient navigation of complex design spaces with user-defined goals, demonstrated on nanoparticle and magnetic materials datasets.

Contribution

The authors develop a novel framework translating user-defined filtering goals into three efficient, parameter-free sequential data acquisition strategies for materials design.

Findings

Framework outperforms existing methods in efficiency

Effective for discrete search spaces with multiple properties

Validated on nanoparticle synthesis and magnetic materials datasets

Abstract

Rapid discovery and synthesis of new materials requires intelligent data acquisition strategies to navigate large design spaces. A popular strategy is Bayesian optimization, which aims to find candidates that maximize material properties; however, materials design often requires finding specific subsets of the design space which meet more complex or specialized goals. We present a framework that captures experimental goals through straightforward user-defined filtering algorithms. These algorithms are automatically translated into one of three intelligent, parameter-free, sequential data acquisition strategies (SwitchBAX, InfoBAX, and MeanBAX). Our framework is tailored for typical discrete search spaces involving multiple measured physical properties and short time-horizon decision making. We evaluate this approach on datasets for TiO$_2$ nanoparticle synthesis and magnetic materials characterization, and show that our methods are significantly more efficient than state-of-the-art approaches.