Insightful classification of crystal structures using deep learning

Abstract

Classifying crystal structures using their space group is important to understand material properties, but the process currently requires user input. Here, the authors use machine learning to automatically classify more than 100,000 simulated perfect and defective crystal structures.