Learning properties of ordered and disordered materials from multi-fidelity data

2020/05/31 by Chi Chen, Yunxing Zuo, Weike Ye +2 · 196 citations
Computer Science · Materials Science · Physics and Astronomy · #Advanced Graph Neural Networks #Band gap #Bridging (networking) #Graph #Graph theory #Machine Learning in Materials Science #Material properties #Materials informatics #Quantum many-body systems #cond-mat.dis-nn #cond-mat.mtrl-sci

paper · pdf · doi:10.1038/s43588-020-00002-x

published in Nature Computational Science 1(1), 46-53 (Nature Portfolio)

openalex publication_date 2021/01/14 · openalex created_date 2021/01/18 · arxiv created 2021/01/28 · arxiv updated 2021/01/29 · openalex updated_date 2026/08/05

Abstract

Predicting the properties of a material from the arrangement of its atoms is a fundamental goal in materials science. While machine learning has emerged in recent years as a new paradigm to provide rapid predictions of materials properties, their practical utility is limited by the scarcity of high-fidelity data. Here, we develop multi-fidelity graph networks as a universal approach to achieve accurate predictions of materials properties with small data sizes. As a proof of concept, we show that the inclusion of low-fidelity Perdew-Burke-Ernzerhof band gaps greatly enhances the resolution of latent structural features in materials graphs, leading to a 22-45% decrease in the mean absolute errors of experimental band gap predictions. We further demonstrate that learned elemental embeddings in materials graph networks provide a natural approach to model disorder in materials, addressing a fundamental gap in the computational prediction of materials properties.

Citations

• Predicting the output from a complex computer code when fast approximations are available
• Generalized Gradient Approximation Made Simple
• SchNet – A deep learning architecture for molecules and materials
• A composite neural network that learns from multi-fidelity data: Application to function approximation and inverse PDE problems
• A Performance and Cost Assessment of Machine Learning Interatomic Potentials

Cited by

• Self-Supervised Graph Neural Networks for Accurate Prediction of Néel Temperature
• Scalable Data-Driven Basis Selection for Linear Machine Learning Interatomic Potentials
• Reconstruction of nanoscale particles from single-shot wide-angle free-electron-laser diffraction patterns with physics-informed neural networks
• Cross-functional transferability in universal machine learning interatomic potentials