Strategies for the Construction of Neural-Network Based Machine-Learning Potentials (MLPs)
This notebook is a companion to Miksch, Morawietz, Kaestner, Urban, and Artrith,
Machine Learning: Science and Technology, in press, (2021) DOI: https://doi.org/10.1088/2632-2153/abfd96.
Copyright (c) 2021 A. M. Miksch, T. Morawietz, J. Kaestner, A. Urban, and N. Artrith.
Contact: Nong Artrith nartrith@atomistic.net
Distributed under the terms of the Mozilla Public License, version 2.0 (https://www.mozilla.org/en-US/MPL/2.0/)
A Jupyter notebook demonstrating the construction of artificial neural network (ANN) interatomic potentials can be found in ./tutorial/aenet_tutorial_beginners_guide.ipynb. The notebook can be run directly using Google Colaboratory and is also shared at the following link: https://colab.research.google.com/drive/1K7uhrUxmM0g7OERWzhSudtCdiVhfX_ul?usp=sharing
In the manuscript, the failure of an initial water potential during a molecular dynamics simulations is discussed. The analysis shown in Figure 10 of the manuscript can be reproduced using another Jupyter notebook provided here in ./script-Fig10/analysis_water_trajectory.ipynb.