Graph dynamical networks for unsupervised learning of atomic scale dynamics in materials

Abstract

Understanding the dynamical processes that govern the performance of functional materials is essential for the design of next generation materials to tackle global energy and environmental challenges. Many of these processes involve the dynamics of individual atoms or small molecules in condensed phases, e.g. lithium ions in electrolytes, water molecules in membranes, molten atoms at interfaces, etc., which are difficult to understand due to the complexity of local environments. In this work, we develop graph dynamical networks, an unsupervised learning approach for understanding atomic scale dynamics in arbitrary phases and environments from molecular dynamics simulations. We show that important dynamical information, which would be difficult to obtain otherwise, can be learned for various multi-component amorphous material systems. With the large amounts of molecular dynamics data generated every day in nearly every aspect of materials design, this approach provides a broadly applicable, automated tool to understand atomic scale dynamics in material systems.

Predicting materials properties without crystal structure: deep representation learning from stoichiometry

“…Soft-attention builds upon this concept by allowing the function that produces the attention coefficients to be learnt directly from the data. The softattention mechanism is the crux behind many state-of-the-art sequence-to-sequence models used in machine translation and language processing 33,34 and it has recently shown good results on graphs 35 and in some material science applications 36,37 . In this domain, the attention mechanism allows us to capture important materials concepts beyond the expressive power of older approaches e.g.…”

Dynamic graphical models of molecular kinetics

“…Machine learning has had a significant impact on molecular modeling in recent years (54)(55)(56)(57)(58)(59)(60)(61)(62)(63). In particular, deep learning with its ability to learn complicated nonlinear functions given sufficient available data (64) has seen multiple applications.…”

Big-Data Science in Porous Materials: Materials Genomics and Machine Learning

“…section 4.1.0.1). And in fact some of the earliest applications of these techniques were to analyze 45 − 47 and then speed up molecular simulations. 48 , 49 The challenge with molecular simulations is that we explore a 3 N dimensional space, where N is the number of particles.…”