Bayesian Force Fields from Active Learning for Simulation of Inter-Dimensional Transformation of Stanene

TL;DR

This paper introduces an accelerated Gaussian process-based force field model using active learning, enabling large-scale, accurate, and uncertainty-aware molecular dynamics simulations of materials like stanene.

Contribution

The authors develop a low-dimensional feature mapping for force and uncertainty functions, significantly speeding up Gaussian process models for interatomic forces with active learning.

Findings

Discovered a novel phase transformation mechanism in stanene involving ripple-induced defect nucleation.

Achieved near-quantum accuracy in large-scale molecular dynamics simulations.

Demonstrated constant evaluation cost suitable for simulating millions of atoms.

Abstract

We present a way to dramatically accelerate Gaussian process models for interatomic force fields based on many-body kernels by mapping both forces and uncertainties onto functions of low-dimensional features. This allows for automated active learning of models combining near-quantum accuracy, built-in uncertainty, and constant cost of evaluation that is comparable to classical analytical models, capable of simulating millions of atoms. Using this approach, we perform large scale molecular dynamics simulations of the stability of the stanene monolayer. We discover an unusual phase transformation mechanism of 2D stanene, where ripples lead to nucleation of bilayer defects, densification into a disordered multilayer structure, followed by formation of bulk liquid at high temperature or nucleation and growth of the 3D bcc crystal at low temperature. The presented method opens possibilities for rapid development of fast accurate uncertainty-aware models for simulating long-time large-scale dynamics of complex materials.