Library-Based LAMMPS Implementation of High-Dimensional Neural Network Potentials.

By A. Singraber, J. Behler, Christoph Dellago. Journal of Chemical Theory and Computation, 2019

295 citations recorded by the source corpus.

Neural networks and other machine learning approaches have been successfully used to accurately represent atomic interaction potentials derived from computationally demanding electronic structure calculations. Due to their low computational cost, such representations open the possibility for large scale reactive molecular dynamics simulations of processes with bonding situations that cannot be described accurately with traditional empirical force fields. Here, we present a library of functions developed for the implementation of neural network potentials. Written in C++, this library incorporates several strategies resulting in a very high efficiency of neural network potential-energy and force evaluations. Based on this library, we have developed an implementation of the neural network potential within the molecular dynamics package LAMMPS and demonstrate its performance using liquid water as a test system.

Subfields: Computational Theory and Mathematics, Materials Chemistry, Molecular Biology

View Paper: https://www.semanticscholar.org/paper/2782e068099d09660e8cb04de78f46172314c1cd

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

1. Florian Hase
2. Pascal Friederich
3. Jonny Proppe
4. Chi Chen
5. Shyue Ping Ong
6. Alán Aspuru-Guzik
7. Tommi S. Jaakkola
8. Regina Barzilay

In Materials Chemistry

1. Chi Chen
2. Florian Hase
3. Shyue Ping Ong
4. Pascal Friederich
5. Jonny Proppe
6. Alán Aspuru-Guzik
7. Tommi S. Jaakkola
8. Regina Barzilay

In Molecular Biology

1. Alán Aspuru-Guzik
2. Chi Chen
3. Jonny Proppe
4. Shyue Ping Ong
5. Pascal Friederich
6. Florian Hase

Citations in the corpus

• Machine-learned potentials for next-generation matter simulations (2021, methodology): “Seminal examples of such efforts include the integration of the high-dimensional neural network potentials in LAMMPS 142 , the TensorMol 0.1 model available as a Python package 34 .”
• A universal graph deep learning interatomic potential for the periodic table (2022, background): “…harmonics function; σ is the sigmoid activation function; 3 is the cut-off function ensuring that the functions vanished smoothly at the neighbor boundary 47 ; g ( x ) = xσ ( x ) is the nonlinear activation function 48 ; and ̃ e ij is a vector of length n max l max , expanded by indices l = 0,…”