A framework for machine-learning-augmented multiscale atomistic simulations on parallel supercomputers

TY - JOUR

T1 - A framework for machine-learning-augmented multiscale atomistic simulations on parallel supercomputers

AU - Caccin, Marco

AU - Li, Zhenwei

AU - Kermode, James R

AU - De Vita, Alessandro

PY - 2015/8/1

Y1 - 2015/8/1

N2 - Recent advances in quantum mechanical (QM)-based molecular dynamics (MD) simulations have used machine-learning (ML) to predict, rather than recalculate, QM-accurate forces in atomic configurations sufficiently similar to previously encountered ones. Here, we discuss how ML approaches can be deployed within large-scale QM/MM materials simulations on massively parallel supercomputers, making QM zones of 1000 atoms routinely attainable. We argue that the ML approach allows computational effort to be concentrated on the most chemically active subregions of the QM zone, significantly improving the overall efficiency of the simulation. We thus propose a novel method to partition large QM regions into multiple subregions, which can be computed in parallel to achieve optimal scaling. Then we review a recently proposed QM/ML MD scheme (Z. Li, J.R. Kermode, A. De Vita Phys. Rev. Lett., 2015, 114, 096405), discussing how this could be efficiently combined with QM-zone partitioning.

AB - Recent advances in quantum mechanical (QM)-based molecular dynamics (MD) simulations have used machine-learning (ML) to predict, rather than recalculate, QM-accurate forces in atomic configurations sufficiently similar to previously encountered ones. Here, we discuss how ML approaches can be deployed within large-scale QM/MM materials simulations on massively parallel supercomputers, making QM zones of 1000 atoms routinely attainable. We argue that the ML approach allows computational effort to be concentrated on the most chemically active subregions of the QM zone, significantly improving the overall efficiency of the simulation. We thus propose a novel method to partition large QM regions into multiple subregions, which can be computed in parallel to achieve optimal scaling. Then we review a recently proposed QM/ML MD scheme (Z. Li, J.R. Kermode, A. De Vita Phys. Rev. Lett., 2015, 114, 096405), discussing how this could be efficiently combined with QM-zone partitioning.

KW - fracture

KW - HPC

KW - machine learning

KW - partitioning

KW - quantum mechanics/molecular mechanics

UR - http://www.scopus.com/inward/record.url?scp=84936847600&partnerID=8YFLogxK

U2 - 10.1002/qua.24952

DO - 10.1002/qua.24952

M3 - Article

AN - SCOPUS:84936847600

SN - 0020-7608

VL - 115

SP - 1129

EP - 1139

JO - INTERNATIONAL JOURNAL OF QUANTUM CHEMISTRY

JF - INTERNATIONAL JOURNAL OF QUANTUM CHEMISTRY

IS - 16

ER -