skip to main content
DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

This content will become publicly available on April 4, 2020

Title: A universal density matrix functional from molecular orbital-based machine learning: Transferability across organic molecules

Abstract

We address the degree to which machine learning (ML) can be used to accurately and transferably predict post-Hartree-Fock correlation energies. Refined strategies for feature design and selection are presented, and the molecular-orbital-based machine learning (MOB-ML) method is applied to several test systems. Strikingly, for the second-order Møller-Plessett perturbation theory, coupled cluster with singles and doubles (CCSD), and CCSD with perturbative triples levels of theory, it is shown that the thermally accessible (350 K) potential energy surface for a single water molecule can be described to within 1 mhartree using a model that is trained from only a single reference calculation at a randomized geometry. To explore the breadth of chemical diversity that can be described, MOB-ML is also applied to a new dataset of thermalized (350 K) geometries of 7211 organic models with up to seven heavy atoms. In comparison with the previously reported Δ-ML method, MOB-ML is shown to reach chemical accuracy with threefold fewer training geometries. Finally, a transferability test in which models trained for seven-heavy-atom systems are used to predict energies for thirteen-heavy-atom systems reveals that MOB-ML reaches chemical accuracy with 36-fold fewer training calculations than Δ-ML (140 vs 5000 training calculations).

Authors:
ORCiD logo [1];  [1]; ORCiD logo [2]; ORCiD logo [1]
  1. California Inst. of Technology (CalTech), Pasadena, CA (United States)
  2. Univ. of Basel (Switzerland)
Publication Date:
Research Org.:
Lawrence Berkeley National Laboratory-National Energy Research Scientific Computing Center (NERSC)
Sponsoring Org.:
USDOE Office of Science (SC)
OSTI Identifier:
1529903
Alternate Identifier(s):
OSTI ID: 1505017
Grant/Contract Number:  
AC02-05CH11231
Resource Type:
Accepted Manuscript
Journal Name:
Journal of Chemical Physics
Additional Journal Information:
Journal Volume: 150; Journal Issue: 13; Journal ID: ISSN 0021-9606
Publisher:
American Institute of Physics (AIP)
Country of Publication:
United States
Language:
English
Subject:
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CHEMISTRY

Citation Formats

Cheng, Lixue, Welborn, Matthew, Christensen, Anders S., and Miller, Thomas F. A universal density matrix functional from molecular orbital-based machine learning: Transferability across organic molecules. United States: N. p., 2019. Web. doi:10.1063/1.5088393.
Cheng, Lixue, Welborn, Matthew, Christensen, Anders S., & Miller, Thomas F. A universal density matrix functional from molecular orbital-based machine learning: Transferability across organic molecules. United States. doi:10.1063/1.5088393.
Cheng, Lixue, Welborn, Matthew, Christensen, Anders S., and Miller, Thomas F. Thu . "A universal density matrix functional from molecular orbital-based machine learning: Transferability across organic molecules". United States. doi:10.1063/1.5088393.
@article{osti_1529903,
title = {A universal density matrix functional from molecular orbital-based machine learning: Transferability across organic molecules},
author = {Cheng, Lixue and Welborn, Matthew and Christensen, Anders S. and Miller, Thomas F.},
abstractNote = {We address the degree to which machine learning (ML) can be used to accurately and transferably predict post-Hartree-Fock correlation energies. Refined strategies for feature design and selection are presented, and the molecular-orbital-based machine learning (MOB-ML) method is applied to several test systems. Strikingly, for the second-order Møller-Plessett perturbation theory, coupled cluster with singles and doubles (CCSD), and CCSD with perturbative triples levels of theory, it is shown that the thermally accessible (350 K) potential energy surface for a single water molecule can be described to within 1 mhartree using a model that is trained from only a single reference calculation at a randomized geometry. To explore the breadth of chemical diversity that can be described, MOB-ML is also applied to a new dataset of thermalized (350 K) geometries of 7211 organic models with up to seven heavy atoms. In comparison with the previously reported Δ-ML method, MOB-ML is shown to reach chemical accuracy with threefold fewer training geometries. Finally, a transferability test in which models trained for seven-heavy-atom systems are used to predict energies for thirteen-heavy-atom systems reveals that MOB-ML reaches chemical accuracy with 36-fold fewer training calculations than Δ-ML (140 vs 5000 training calculations).},
doi = {10.1063/1.5088393},
journal = {Journal of Chemical Physics},
number = 13,
volume = 150,
place = {United States},
year = {2019},
month = {4}
}

Journal Article:
Free Publicly Available Full Text
This content will become publicly available on April 4, 2020
Publisher's Version of Record

Save / Share:

Works referenced in this record:

Density‐functional thermochemistry. III. The role of exact exchange
journal, April 1993

  • Becke, Axel D.
  • The Journal of Chemical Physics, Vol. 98, Issue 7, p. 5648-5652
  • DOI: 10.1063/1.464913

Ab Initio Calculation of Vibrational Absorption and Circular Dichroism Spectra Using Density Functional Force Fields
journal, November 1994

  • Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.
  • The Journal of Physical Chemistry, Vol. 98, Issue 45, p. 11623-11627
  • DOI: 10.1021/j100096a001