A tomographic interpretation of structure-property relations for materials discovery

TL;DR

This paper introduces a tomographic framework for understanding how different material representations relate to their properties, using information theory to analyze and compare various property-augmented representations in materials discovery.

Contribution

It proposes a novel information-theoretic interpretation of structure-property relations, bridging the gap between simple and complex material representations.

Findings

Property-augmented representations encode complementary information.

Different properties contribute uniquely to prediction performance.

The framework clarifies the role of structural and compositional data.

Abstract

Recent advancements in machine learning (ML) for materials have demonstrated that "simple" materials representations (e.g., the chemical formula alone without structural information) can sometimes achieve competitive property prediction performance in common-tasks. Our physics-based intuition would suggest that such representations are "incomplete", which indicates a gap in our understanding. This work proposes a tomographic interpretation of structure-property relations of materials to bridge that gap by defining what is a material representation, material properties, the material and the relationships between these three concepts using ideas from information theory. We verify this framework performing an exhaustive comparison of property-augmented representations on a range of material's property prediction objectives, providing insight into how different properties can encode complementary information.