Accurate force field for molybdenum by machine learning large materials data
Abstract
In this work, we present a highly accurate spectral neighbor analysis potential (SNAP) model for molybdenum (Mo) developed through the rigorous application of machine learning techniques on large materials data sets. Despite Mo's importance as a structural metal, existing force fields for Mo based on the embedded atom and modified embedded atom methods still do not provide satisfactory accuracy on many properties. We will show that by fitting to the energies, forces and stress tensors of a large density functional theory (DFT)-computed dataset on a diverse set of Mo structures, a Mo SNAP model can be developed that achieves close to DFT accuracy in the prediction of a broad range of properties, including energies, forces, stresses, elastic constants, melting point, phonon spectra, surface energies, grain boundary energies, etc. We will outline a systematic model development process, which includes a rigorous approach to structural selection based on principal component analysis, as well as a differential evolution algorithm for optimizing the hyperparameters in the model fitting so that both the model error and the property prediction error can be simultaneously lowered. We expect that this newly developed Mo SNAP model will find broad applications in large-scale, long-time scale simulations.
- Authors:
-
- Univ. of California, San Diego, CA (United States). Dept. of NanoEngineering
- Publication Date:
- Research Org.:
- Lawrence Berkeley National Lab. (LBNL), Berkeley, CA (United States)
- Sponsoring Org.:
- USDOE Office of Science (SC)
- OSTI Identifier:
- 1559139
- Grant/Contract Number:
- AC02-05CH11231
- Resource Type:
- Accepted Manuscript
- Journal Name:
- Physical Review Materials
- Additional Journal Information:
- Journal Volume: 1; Journal Issue: 4; Journal ID: ISSN 2475-9953
- Publisher:
- American Physical Society (APS)
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 36 MATERIALS SCIENCE
Citation Formats
Chen, Chi, Deng, Zhi, Tran, Richard, Tang, Hanmei, Chu, Iek-Heng, and Ong, Shyue Ping. Accurate force field for molybdenum by machine learning large materials data. United States: N. p., 2017.
Web. doi:10.1103/physrevmaterials.1.043603.
Chen, Chi, Deng, Zhi, Tran, Richard, Tang, Hanmei, Chu, Iek-Heng, & Ong, Shyue Ping. Accurate force field for molybdenum by machine learning large materials data. United States. doi:10.1103/physrevmaterials.1.043603.
Chen, Chi, Deng, Zhi, Tran, Richard, Tang, Hanmei, Chu, Iek-Heng, and Ong, Shyue Ping. Fri .
"Accurate force field for molybdenum by machine learning large materials data". United States. doi:10.1103/physrevmaterials.1.043603. https://www.osti.gov/servlets/purl/1559139.
@article{osti_1559139,
title = {Accurate force field for molybdenum by machine learning large materials data},
author = {Chen, Chi and Deng, Zhi and Tran, Richard and Tang, Hanmei and Chu, Iek-Heng and Ong, Shyue Ping},
abstractNote = {In this work, we present a highly accurate spectral neighbor analysis potential (SNAP) model for molybdenum (Mo) developed through the rigorous application of machine learning techniques on large materials data sets. Despite Mo's importance as a structural metal, existing force fields for Mo based on the embedded atom and modified embedded atom methods still do not provide satisfactory accuracy on many properties. We will show that by fitting to the energies, forces and stress tensors of a large density functional theory (DFT)-computed dataset on a diverse set of Mo structures, a Mo SNAP model can be developed that achieves close to DFT accuracy in the prediction of a broad range of properties, including energies, forces, stresses, elastic constants, melting point, phonon spectra, surface energies, grain boundary energies, etc. We will outline a systematic model development process, which includes a rigorous approach to structural selection based on principal component analysis, as well as a differential evolution algorithm for optimizing the hyperparameters in the model fitting so that both the model error and the property prediction error can be simultaneously lowered. We expect that this newly developed Mo SNAP model will find broad applications in large-scale, long-time scale simulations.},
doi = {10.1103/physrevmaterials.1.043603},
journal = {Physical Review Materials},
number = 4,
volume = 1,
place = {United States},
year = {2017},
month = {9}
}
Search WorldCat to find libraries that may hold this journalWeb of Science
Works referenced in this record:
Self-Consistent Equations Including Exchange and Correlation Effects
journal, November 1965
- Kohn, W.; Sham, L. J.
- Physical Review, Vol. 140, Issue 4A, p. A1133-A1138
UFF, a full periodic table force field for molecular mechanics and molecular dynamics simulations
journal, December 1992
- Rappe, A. K.; Casewit, C. J.; Colwell, K. S.
- Journal of the American Chemical Society, Vol. 114, Issue 25, p. 10024-10035
Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set
journal, October 1996
- Kresse, G.; Furthmüller, J.
- Physical Review B, Vol. 54, Issue 16, p. 11169-11186
Projector augmented-wave method
journal, December 1994
- Blöchl, P. E.
- Physical Review B, Vol. 50, Issue 24, p. 17953-17979
Generalized Gradient Approximation Made Simple
journal, October 1996
- Perdew, John P.; Burke, Kieron; Ernzerhof, Matthias
- Physical Review Letters, Vol. 77, Issue 18, p. 3865-3868
A climbing image nudged elastic band method for finding saddle points and minimum energy paths
journal, December 2000
- Henkelman, Graeme; Uberuaga, Blas P.; Jónsson, Hannes
- The Journal of Chemical Physics, Vol. 113, Issue 22, p. 9901-9904