skip to main content

DOE PAGESDOE PAGES

This content will become publicly available on March 19, 2019

Title: Hierarchical modeling of molecular energies using a deep neural network

In this work, we introduce the Hierarchically Interacting Particle Neural Network (HIP-NN) to model molecular properties from datasets of quantum calculations. Inspired by a many-body expansion, HIP-NN decomposes properties, such as energy, as a sum over hierarchical terms. These terms are generated from a neural network—a composition of many nonlinear transformations—acting on a representation of the molecule. HIP-NN achieves the state-of-the-art performance on a dataset of 131k ground state organic molecules and predicts energies with 0.26 kcal/mol mean absolute error. Finally, with minimal tuning, our model is also competitive on a dataset of molecular dynamics trajectories. In addition to enabling accurate energy predictions, the hierarchical structure of HIP-NN helps to identify regions of model uncertainty.
Authors:
ORCiD logo [1] ; ORCiD logo [2] ; ORCiD logo [1]
  1. Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
  2. Los Alamos National Lab. (LANL), Los Alamos, NM (United States); Univ. of Florida, Gainesville, FL (United States)
Publication Date:
Report Number(s):
LA-UR-17-28890
Journal ID: ISSN 0021-9606
Grant/Contract Number:
AC52-06NA25396
Type:
Accepted Manuscript
Journal Name:
Journal of Chemical Physics
Additional Journal Information:
Journal Volume: 148; Journal Issue: 24; Journal ID: ISSN 0021-9606
Publisher:
American Institute of Physics (AIP)
Research Org:
Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
Sponsoring Org:
USDOE Laboratory Directed Research and Development (LDRD) Program
Country of Publication:
United States
Language:
English
Subject:
97 MATHEMATICS AND COMPUTING; 37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CHEMISTRY
OSTI Identifier:
1479929
Alternate Identifier(s):
OSTI ID: 1426848

Lubbers, Nicholas Edward, Smith, Justin Steven, and Barros, Kipton Marcos. Hierarchical modeling of molecular energies using a deep neural network. United States: N. p., Web. doi:10.1063/1.5011181.
Lubbers, Nicholas Edward, Smith, Justin Steven, & Barros, Kipton Marcos. Hierarchical modeling of molecular energies using a deep neural network. United States. doi:10.1063/1.5011181.
Lubbers, Nicholas Edward, Smith, Justin Steven, and Barros, Kipton Marcos. 2018. "Hierarchical modeling of molecular energies using a deep neural network". United States. doi:10.1063/1.5011181.
@article{osti_1479929,
title = {Hierarchical modeling of molecular energies using a deep neural network},
author = {Lubbers, Nicholas Edward and Smith, Justin Steven and Barros, Kipton Marcos},
abstractNote = {In this work, we introduce the Hierarchically Interacting Particle Neural Network (HIP-NN) to model molecular properties from datasets of quantum calculations. Inspired by a many-body expansion, HIP-NN decomposes properties, such as energy, as a sum over hierarchical terms. These terms are generated from a neural network—a composition of many nonlinear transformations—acting on a representation of the molecule. HIP-NN achieves the state-of-the-art performance on a dataset of 131k ground state organic molecules and predicts energies with 0.26 kcal/mol mean absolute error. Finally, with minimal tuning, our model is also competitive on a dataset of molecular dynamics trajectories. In addition to enabling accurate energy predictions, the hierarchical structure of HIP-NN helps to identify regions of model uncertainty.},
doi = {10.1063/1.5011181},
journal = {Journal of Chemical Physics},
number = 24,
volume = 148,
place = {United States},
year = {2018},
month = {3}
}