Development of Machine-Learning for Interatomic Potential Models

1.3 Statistics, Machine Learning, Data Assimilation, Algorithms, Mathematical Foundation, Data Mining
2.1 Material (Material Informatics, Remote Experiment, Laboratory Automation, etc.)

Satoshi Watanabe

Graduate School of Engineering

Professor

In this project, interatomic potential models enabling high accuracy and low computational load simulations of atomic structures of materials and their atomic dynamics are developed using first-principles calculation data and machine learning techniques.

Comparison of first-principles density functional theory (DFT) calculation and neural network potential (NNP) for energy prediction of Li3PO4 structures used (Left) and not used (Right) in the machine-learning process.
J. Chem. Phys. 147 (2017) 214106

Research collaborators

・Institute for Molecular Science
・National Institute of Advanced Industrial Science and
Technology
・Seoul National University, Korea

Related publications

Satoshi Watanabe, Wenwen Li, Wonseok Jeong, Dongheon Lee, Koji Shimizu, Emi Mimanitani, Yasunobu Ando, and Seungwu Han, “High-dimensional neural network atomic potentials for examining energy materials: some recent simulations,” Journal of Physics: Energy Vol. 3, No. 1, 012003, Jan. 2021.