PND: Physics-informed neural-network software for molecular dynamics applications

PND: Physics-informed neural-network software for molecular dynamics applications

We have developed PND, a differential equation solver software based on a physics-informed neural network (PINN) for molecular dynamics simulators. Based on automatic differentiation technique provided by PyTorch, our software allows users to flexibly implement equation of motion for atoms, initial...

Full description

Bibliographic Details
Main Authors: Taufeq Mohammed Razakh, Beibei Wang, Shane Jackson, Rajiv K. Kalia, Aiichiro Nakano, Ken-ichi Nomura, Priya Vashishta
Format: Article
Language:English
Published: Elsevier 2021-07-01
Series:SoftwareX
Subjects:
Molecular dynamics
Machine learning
Differential equation solver
Online Access:http://www.sciencedirect.com/science/article/pii/S2352711021000972
id doaj-503282e096cf44d9ac4c9c76ac562c7c
record_format Article
spelling doaj-503282e096cf44d9ac4c9c76ac562c7c2021-08-10T04:05:05ZengElsevierSoftwareX2352-71102021-07-0115100789PND: Physics-informed neural-network software for molecular dynamics applicationsTaufeq Mohammed Razakh0Beibei Wang1Shane Jackson2Rajiv K. Kalia3Aiichiro Nakano4Ken-ichi Nomura5Priya Vashishta6Collaboratory for Advanced Computing and Simulations, University of Southern California, Los Angeles, CA 90089-0242, USA; Department of Computer Science, University of Southern California, Los Angeles, CA 90089-0781, USACollaboratory for Advanced Computing and Simulations, University of Southern California, Los Angeles, CA 90089-0242, USA; Department of Physics & Astronomy, University of Southern California, Los Angeles, CA 90089-0484, USACollaboratory for Advanced Computing and Simulations, University of Southern California, Los Angeles, CA 90089-0242, USA; Department of Physics & Astronomy, University of Southern California, Los Angeles, CA 90089-0484, USACollaboratory for Advanced Computing and Simulations, University of Southern California, Los Angeles, CA 90089-0242, USA; Department of Computer Science, University of Southern California, Los Angeles, CA 90089-0781, USA; Department of Physics & Astronomy, University of Southern California, Los Angeles, CA 90089-0484, USA; Department of Chemical Engineering & Materials Science, University of Southern California, Los Angeles, CA 90089-12111, USACollaboratory for Advanced Computing and Simulations, University of Southern California, Los Angeles, CA 90089-0242, USA; Department of Computer Science, University of Southern California, Los Angeles, CA 90089-0781, USA; Department of Physics & Astronomy, University of Southern California, Los Angeles, CA 90089-0484, USA; Department of Chemical Engineering & Materials Science, University of Southern California, Los Angeles, CA 90089-12111, USA; Department of Quantitative & computational Biology, University of Southern California, Los Angeles, CA 90089-2910, USACollaboratory for Advanced Computing and Simulations, University of Southern California, Los Angeles, CA 90089-0242, USA; Department of Chemical Engineering & Materials Science, University of Southern California, Los Angeles, CA 90089-12111, USA; Corresponding author at: Collaboratory for Advanced Computing and Simulations, University of Southern California, Los Angeles, CA 90089-0242, USA.Collaboratory for Advanced Computing and Simulations, University of Southern California, Los Angeles, CA 90089-0242, USA; Department of Computer Science, University of Southern California, Los Angeles, CA 90089-0781, USA; Department of Physics & Astronomy, University of Southern California, Los Angeles, CA 90089-0484, USA; Department of Chemical Engineering & Materials Science, University of Southern California, Los Angeles, CA 90089-12111, USAWe have developed PND, a differential equation solver software based on a physics-informed neural network (PINN) for molecular dynamics simulators. Based on automatic differentiation technique provided by PyTorch, our software allows users to flexibly implement equation of motion for atoms, initial and boundary conditions, and conservation laws as loss function to train the network. PND comes with a parallel molecular dynamic engine in order to examine and optimize loss function design, and different conservation laws and boundary conditions, and hyperparameters, thereby accelerating PINN-based development for molecular applications.http://www.sciencedirect.com/science/article/pii/S2352711021000972Molecular dynamicsMachine learningDifferential equation solver
collection DOAJ
language English
format Article
sources DOAJ
author Taufeq Mohammed Razakh
Beibei Wang
Shane Jackson
Rajiv K. Kalia
Aiichiro Nakano
Ken-ichi Nomura
Priya Vashishta
spellingShingle Taufeq Mohammed Razakh
Beibei Wang
Shane Jackson
Rajiv K. Kalia
Aiichiro Nakano
Ken-ichi Nomura
Priya Vashishta
PND: Physics-informed neural-network software for molecular dynamics applications
SoftwareX
Molecular dynamics
Machine learning
Differential equation solver
author_facet Taufeq Mohammed Razakh
Beibei Wang
Shane Jackson
Rajiv K. Kalia
Aiichiro Nakano
Ken-ichi Nomura
Priya Vashishta
author_sort Taufeq Mohammed Razakh
title PND: Physics-informed neural-network software for molecular dynamics applications
title_short PND: Physics-informed neural-network software for molecular dynamics applications
title_full PND: Physics-informed neural-network software for molecular dynamics applications
title_fullStr PND: Physics-informed neural-network software for molecular dynamics applications
title_full_unstemmed PND: Physics-informed neural-network software for molecular dynamics applications
title_sort pnd: physics-informed neural-network software for molecular dynamics applications
publisher Elsevier
series SoftwareX
issn 2352-7110
publishDate 2021-07-01
description We have developed PND, a differential equation solver software based on a physics-informed neural network (PINN) for molecular dynamics simulators. Based on automatic differentiation technique provided by PyTorch, our software allows users to flexibly implement equation of motion for atoms, initial and boundary conditions, and conservation laws as loss function to train the network. PND comes with a parallel molecular dynamic engine in order to examine and optimize loss function design, and different conservation laws and boundary conditions, and hyperparameters, thereby accelerating PINN-based development for molecular applications.
topic Molecular dynamics
Machine learning
Differential equation solver
url http://www.sciencedirect.com/science/article/pii/S2352711021000972
work_keys_str_mv AT taufeqmohammedrazakh pndphysicsinformedneuralnetworksoftwareformoleculardynamicsapplications
AT beibeiwang pndphysicsinformedneuralnetworksoftwareformoleculardynamicsapplications
AT shanejackson pndphysicsinformedneuralnetworksoftwareformoleculardynamicsapplications
AT rajivkkalia pndphysicsinformedneuralnetworksoftwareformoleculardynamicsapplications
AT aiichironakano pndphysicsinformedneuralnetworksoftwareformoleculardynamicsapplications
AT kenichinomura pndphysicsinformedneuralnetworksoftwareformoleculardynamicsapplications
AT priyavashishta pndphysicsinformedneuralnetworksoftwareformoleculardynamicsapplications
_version_ 1721212982779707392

Similar Items