Abstract
We propose an active learning scheme for automatically sampling a minimum number of uncorrelated configurations for fitting the Gaussian Approximation Potential (GAP). Our active learning scheme consists of an unsupervised machine learning (ML) scheme coupled with a Bayesian optimization technique that evaluates the GAP model. We apply this scheme to a Hafnium dioxide (HfO2) dataset generated from a “melt-quench” ab initio molecular dynamics (AIMD) protocol. Our results show that the active learning scheme, with no prior knowledge of the dataset, is able to extract a configuration that reaches the required energy fit tolerance. Further, molecular dynamics (MD) simulations performed using this active learned GAP model on 6144 atom systems of amorphous and liquid state elucidate the structural properties of HfO2 with near ab initio precision and quench rates (i.e., 1.0 K/ps) not accessible via AIMD. The melt and amorphous X-ray structural factors generated from our simulation are in good agreement with experiment. In addition, the calculated diffusion constants are in good agreement with previous ab initio studies.
Highlights
Ab initio molecular dynamics (AIMD) simulations based on Density Functional Theory (DFT)[1,2] can provide atomistic structural descriptions of materials with quantum mechanical accuracy[3]
We begin by discussing the results of applying the Active learning (AL) workflow to HfO2 datasets generated from NVT “melt-quench” AIMD simulations
We have shown that the AL workflow can automatically sample a minimum number of training and test configuration for the Gaussian Approximation Potential (GAP) model that result in near DFT accuracy for the fitted energy and forces
Summary
Ab initio molecular dynamics (AIMD) simulations based on Density Functional Theory (DFT)[1,2] can provide atomistic structural descriptions of materials with quantum mechanical accuracy[3]. Classical molecular dynamics (MD) simulations based on interatomic potentials derived from empirical and physical approximations, on the other hand, can provide access to larger system sizes (millions of atoms) with longer timescales (~100–1000’s of ns) by sacrificing the quantum mechanical accuracy. Inverse modeling techniques such as Reverse Monte Carlo (RMC)[4], along with advanced experimental techniques such as synchrotron based high-energy X-ray diffraction, have certainly aided in improved understanding of the atomic structure of materials. Recent research showed the application of RMC in the development of a quantum mechanical-accurate model for amorphous silicon[6,7]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
