Evaluating the diversity and utility of materials proposed by generative models

TL;DR

This paper evaluates a physics-guided generative model for crystal structures, revealing limitations in its input space smoothness and stability predictions, and discusses potential improvements for inverse material design.

Contribution

It provides a critical assessment of PGCGM's effectiveness in inverse design, highlighting areas for enhancement in generative modeling for materials science.

Findings

PGCGM's input space is not smooth, hindering optimization.

Most generated structures are thermodynamically unstable.

Out-of-domain data affects property prediction accuracy.

Abstract

Generative machine learning models can use data generated by scientific modeling to create large quantities of novel material structures. Here, we assess how one state-of-the-art generative model, the physics-guided crystal generation model (PGCGM), can be used as part of the inverse design process. We show that the default PGCGM's input space is not smooth with respect to parameter variation, making material optimization difficult and limited. We also demonstrate that most generated structures are predicted to be thermodynamically unstable by a separate property-prediction model, partially due to out-of-domain data challenges. Our findings suggest how generative models might be improved to enable better inverse design.