You must login before you can run this tool.
Category
Published on
Abstract
This program is meant to be a simple, practice simulation tool on the basics of how to run molecular dynamic (MD) simulations within LAMMPS using different statistical ensembles, such as NVE, NPT, NVT, and NPH. Statistical ensembles are different system states that keeps certain equilibrium state variables fixed to mimic experimental conditions during the MD simulation.
NVE: Microcanonical: Constant Energy, Volume
NPT: Isothermal-Isobaric: Constant Pressure, Temperature
NVT: Canonical: Constant Temperature, Volume
NPH: Isentropic-Isobaric: Constant Pressure, Enthalpy
(Background information on molecular dynamics can be found and referenced in Prof. Alejandro Strachan's lectures here [Lectures 1,3, & 4]: https://nanohub.org/courses/FATM)
The forcefield being used for these MD simulations will be a neural network reactive forcefield (NNRF). The neural network reactive force field (NNRF) is a machine learning based force field to describe atomistic behavior of complex chemical processes. The fully connected neural network is designed to learn and predict the relationship between atomic structures (using weighted gaussian symmetry functions) and a potential energy landscape of chemical reactions. The coordinates of local atomic structure are converted to 4 two-body (radial) and 20 three-body (angular) weighted gaussian symmetry functions. These 24 input nodes are fed into the atomic neural network potential consist of two hidden layers with 30 nodes per each hidden layer and one output node to predict the atomic energy in the given chemical environment. The total energy is predicted from the sum of the atomic energies in the system.
The architecture can be represented as 24 x 30 x 30 x 1: input nodes x hidden layer1 x hidden layer2 x atomic energy
(More infomation on NNRF can be found and referenced here: Yoo, P., Sakano, M., Desai, S. et al. Neural network reactive force field for C, H, N, and O systems. npj Comput Mater 7, 9 (2021). https://doi.org/10.1038/s41524-020-00484-3 )
Homework Assignment: Download PDF
Sponsored by
We acknowledge support from the US Department of Defense [Contract No. W52P1J2093009]. This work is funded by the National Science Foundation, Network for Computational Nanotechnology Cyberplatform, Award EEC 1227110. Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.