Machine Learning Modeling of Materials with a Group-Subgroup Structure

Abstract

Crystal structures connected by continuous phase transitions are linked through mathematical relations between crystallographic groups and their subgroups. In the present study, we introduce group-subgroup machine learning (GS-ML) and show that including materials with small unit cells in the training set decreases out-of-sample prediction errors for materials with large unit cells. GS-ML incurs the least training cost to reach 2-3% target accuracy compared to other ML approaches. Since available materials datasets are heterogeneous providing insufficient examples for realizing the group-subgroup structure, we present the "FriezeRMQ1D" dataset with 8393 Q1D organometallic materials uniformly distributed across 7 frieze groups. Furthermore, by comparing the performances of FCHL and 1-hot representations, we show GS-ML to capture subgroup information efficiently when the descriptor encodes structural information. The proposed approach is generic and extendable to symmetry abstractions such as spin-, valency-, or charge order.