Unified Representation of Molecules and Crystals for Machine Learning

Abstract

Accurate simulations of atomistic systems from first principles are limited by computational cost. In high-throughput settings, machine learning can potentially reduce these costs significantly by accurately interpolating between reference calculations. For this, kernel learning approaches crucially require a single Hilbert space accommodating arbitrary atomistic systems. We introduce a many-body tensor representation that is invariant to translations, rotations and nuclear permutations of same elements, unique, differentiable, can represent molecules and crystals, and is fast to compute. Empirical evidence is presented for energy prediction errors below 1 kcal/mol for 7k organic molecules and 5 meV/atom for 11k elpasolite crystals. Applicability is demonstrated for phase diagrams of Pt-group/transition-metal binary systems.