TY - JOUR

T1 - In operando active learning of interatomic interaction during large-scale simulations

AU - Hodapp, M.

AU - Shapeev, A.

N1 - Publisher Copyright:
© 2020 IOP. All rights reserved.

PY - 2020/9/17

Y1 - 2020/9/17

N2 - A well-known drawback of state-of-the-art machine-learning interatomic potentials is their poor ability to extrapolate beyond the training domain. For small-scale problems with tens to hundreds of atoms this can be solved by using active learning which is able to select atomic configurations on which a potential attempts extrapolation and add them to the ab initio-computed training set. In this sense an active learning algorithm can be viewed as an on-the-fly interpolation of an ab initio model. For large-scale problems, possibly involving tens of thousands of atoms, this is not feasible because one cannot afford even a single density functional theory (DFT) computation with such a large number of atoms. This work marks a new milestone toward fully automatic ab initio-accurate large-scale atomistic simulations. We develop an active learning algorithm that identifies local subregions of the simulation region where the potential extrapolates. Then the algorithm constructs periodic configurations out of these local, non-periodic subregions, sufficiently small to be computable with plane-wave DFT codes, in order to obtain accurate ab initio energies. We benchmark our algorithm on the problem of screw dislocation motion in bcc tungsten and show that our algorithm reaches ab initio accuracy, down to typical magnitudes of numerical noise in DFT codes. We show that our algorithm reproduces material properties such as core structure, Peierls barrier, and Peierls stress. This unleashes new capabilities for computational materials science toward applications which have currently been out of scope if approached solely by ab initio methods.

AB - A well-known drawback of state-of-the-art machine-learning interatomic potentials is their poor ability to extrapolate beyond the training domain. For small-scale problems with tens to hundreds of atoms this can be solved by using active learning which is able to select atomic configurations on which a potential attempts extrapolation and add them to the ab initio-computed training set. In this sense an active learning algorithm can be viewed as an on-the-fly interpolation of an ab initio model. For large-scale problems, possibly involving tens of thousands of atoms, this is not feasible because one cannot afford even a single density functional theory (DFT) computation with such a large number of atoms. This work marks a new milestone toward fully automatic ab initio-accurate large-scale atomistic simulations. We develop an active learning algorithm that identifies local subregions of the simulation region where the potential extrapolates. Then the algorithm constructs periodic configurations out of these local, non-periodic subregions, sufficiently small to be computable with plane-wave DFT codes, in order to obtain accurate ab initio energies. We benchmark our algorithm on the problem of screw dislocation motion in bcc tungsten and show that our algorithm reaches ab initio accuracy, down to typical magnitudes of numerical noise in DFT codes. We show that our algorithm reproduces material properties such as core structure, Peierls barrier, and Peierls stress. This unleashes new capabilities for computational materials science toward applications which have currently been out of scope if approached solely by ab initio methods.

KW - Active learning

KW - Atomistic simulation

KW - Dislocation

KW - Machine-learning potential

KW - Multiscale modeling

UR - http://www.scopus.com/inward/record.url?scp=85102882594&partnerID=8YFLogxK

U2 - 10.1088/2632-2153/aba373

DO - 10.1088/2632-2153/aba373

M3 - Article

AN - SCOPUS:85102882594

VL - 1

JO - Machine Learning: Science and Technology

JF - Machine Learning: Science and Technology

SN - 2632-2153

IS - 4

M1 - 045005

ER -