A Caltech Library Service

Fast Near Ab Initio Potential Energy Surfaces Using Machine Learning

Lu, Fenris and Cheng, Lixue and DiRisio, Ryan J. and Finney, Jacob M. and Boyer, Mark A. and Moonkaen, Pattarapon and Sun, Jiace and Lee, Sebastian J. R. and Deustua, J. Emiliano and Miller, Thomas F., III and McCoy, Anne B. (2022) Fast Near Ab Initio Potential Energy Surfaces Using Machine Learning. Journal of Physical Chemistry A, 126 (25). pp. 4013-4024. ISSN 1089-5639. doi:10.1021/acs.jpca.2c02243.

[img] PDF - Supplemental Material
See Usage Policy.


Use this Persistent URL to link to this item:


A machine-learning based approach for evaluating potential energies for quantum mechanical studies of properties of the ground and excited vibrational states of small molecules is developed. This approach uses the molecular-orbital-based machine learning (MOB-ML) method to generate electronic energies with the accuracy of CCSD(T) calculations at the same cost as a Hartree–Fock calculation. To further reduce the computational cost of the potential energy evaluations without sacrificing the CCSD(T) level accuracy, GPU-accelerated Neural Network Potential Energy Surfaces (NN-PES) are trained to geometries and energies that are collected from small-scale Diffusion Monte Carlo (DMC) simulations, which are run using energies evaluated using the MOB-ML model. The combined NN+(MOB-ML) approach is used in variational calculations of the ground and low-lying vibrational excited states of water and in DMC calculations of the ground states of water, CH₅⁺, and its deuterated analogues. For both of these molecules, comparisons are made to the results obtained using potentials that were fit to much larger sets of electronic energies than were required to train the MOB-ML models. The NN+(MOB-ML) approach is also used to obtain a potential surface for C₂H₅⁺, which is a carbocation with a nonclassical equilibrium structure for which there is currently no available potential surface. This potential is used to explore the CH stretching vibrations, focusing on those of the bridging hydrogen atom. For both CH₅⁺ and C₂H₅⁺ the MOB-ML model is trained using geometries that were sampled from an AIMD trajectory, which was run at 350 K. By comparison, the structures sampled in the ground state calculations can have energies that are as much as ten times larger than those used to train the MOB-ML model. For water a higher temperature AIMD trajectory is needed to obtain accurate results due to the smaller thermal energy. A second MOB-ML model for C₂H₅⁺ was developed with additional higher energy structures in the training set. The two models are found to provide nearly identical descriptions of the ground state of C₂H₅⁺.

Item Type:Article
Related URLs:
URLURL TypeDescription
Cheng, Lixue0000-0002-7329-0585
DiRisio, Ryan J.0000-0003-1272-0112
Lee, Sebastian J. R.0000-0001-7006-9378
Deustua, J. Emiliano0000-0001-8193-2229
Miller, Thomas F., III0000-0002-1882-5380
McCoy, Anne B.0000-0001-6851-6634
Additional Information:© 2022 American Chemical Society. Received 1 April 2022. Revised 27 May 2022. Published online 17 June 2022. Published in issue 30 June 2022. A.B.M. acknowledges support from the Chemistry Division of the National Science Foundation (CHE-1856125 for scientific work and OAC-1663636 for HPC code development). The development of PyVibDMC was supported by a fellowship to R.J.D. from The Molecular Sciences Software Institute under NSF grant OAC-1547580. T.F.M. also thanks the support and funds from the U.S. Army Research Laboratory (W911NF-12-2-0023), the U.S. Department of Energy (DE-SC0019390), the Caltech DeLogi Fund, and the Camille and Henry Dreyfus Foundation (Award ML-20-196). Parts of this work were performed using the Ilahie cluster at the University of Washington, which was purchased using funds from a MRI grant from the National Science Foundation (CHE-1624430). This work was also facilitated through the use of advanced computational, storage, and networking infrastructure provided by the Hyak supercomputer system and funded by the STF at the University of Washington. This research also used resources of the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231 using NERSC award BES-ERCAP0020502. Author Contributions. F.L., L.C., and R.J.D. contributed equally to this work. The authors declare the following competing financial interest(s): Thomas F. Miller is an employee of Entos, Inc.
Funding AgencyGrant Number
Army Research LaboratoryW911NF-12-2-0023
Department of Energy (DOE)DE-SC0019390
Caltech De Logi FundUNSPECIFIED
Camille and Henry Dreyfus FoundationML-20-196
Department of Energy (DOE)DE-AC02-05CH11231
Department of Energy (DOE)BES-ERCAP0020502
Subject Keywords:Chemical calculations, Energy, Mathematical methods, Potential energy, Zero point energy
Issue or Number:25
Record Number:CaltechAUTHORS:20220721-8916000
Persistent URL:
Official Citation:Fast Near Ab Initio Potential Energy Surfaces Using Machine Learning Fenris Lu, Lixue Cheng, Ryan J. DiRisio, Jacob M. Finney, Mark A. Boyer, Pattarapon Moonkaen, Jiace Sun, Sebastian J. R. Lee, J. Emiliano Deustua, Thomas F. Miller III, and Anne B. McCoy The Journal of Physical Chemistry A 2022 126 (25), 4013-4024 DOI: 10.1021/acs.jpca.2c02243
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:115744
Deposited By: George Porter
Deposited On:22 Jul 2022 21:21
Last Modified:22 Jul 2022 21:21

Repository Staff Only: item control page