Persistent homology-based descriptor for machine-learning potential of amorphous structures
The Journal of Chemical Physics · 28.08.2023 · 159 (8)
DOI: 10.1063/5.0159349 ↗Kurz gefasst
Fachartikel in der Zeitschrift The Journal of Chemical Physics, online erschienen am 22.08.2023. Verfasst von Emi Minamitani und 3 weiteren Autor:innen. Beteiligte Einrichtung: The Institute of Scientific and Industrial Research, Osaka University, Ibaraki 567-0047, Japan. Die Arbeit wurde bisher 22-mal zitiert (Crossref) und nennt 46 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Precursory Research for Embryonic Science and Technology, Ministry of Education, Culture, Sports, Science and Technology.
Abstract (Originalsprache)
High-accuracy prediction of the physical properties of amorphous materials is challenging in condensed-matter physics. A promising method to achieve this is machine-learning potentials, which is an alternative to computationally demanding ab initio calculations. When applying machine-learning potentials, the construction of descriptors to represent atomic configurations is crucial. These descriptors should be invariant to symmetry operations. Handcrafted representations using a smooth overlap of atomic positions and graph neural networks (GNN) are examples of methods used for constructing symmetry-invariant descriptors. In this study, we propose a novel descriptor based on a persistence diagram (PD), a two-dimensional representation of persistent homology (PH). First, we demonstrated that the normalized two-dimensional histogram obtained from PD could predict the average energy per atom of amorphous carbon at various densities, even when using a simple model. Second, an analysis of the dimensional reduction results of the descriptor spaces revealed that PH can be used to construct descriptors with characteristics similar to those of a latent space in a GNN. These results indicate that PH is a promising method for constructing descriptors suitable for machine-learning potentials without hyperparameter tuning and deep-learning techniques.
Autor:innen und Institutionen
- The Institute of Scientific and Industrial Research, Osaka University, Ibaraki 567-0047, Japan.
- Institute for Molecular Science, Okazaki 444-8585, Japan.
- JST, PRESTO, Kawaguchi, Saitama 332-0012, Japan.
- Center for Artificial Intelligence and Mathematical Data Science, Okayama University, Okayama 700-8530, Japan.
- Department of Materials Engineering, The University of Tokyo, Hongo, Bunkyo 113-8656, Japan.
ORCID-Profile: Emi Minamitani, Ippei Obayashi, Koji Shimizu, Satoshi Watanabe
Förderung
- Precursory Research for Embryonic Science and Technology
- Ministry of Education, Culture, Sports, Science and Technology
Zitieren
APA
Minamitani, E., Obayashi, I., Shimizu, K., & Watanabe, S. (2023). Persistent homology-based descriptor for machine-learning potential of amorphous structures. The Journal of Chemical Physics, 159(8). https://doi.org/10.1063/5.0159349
Vancouver
Minamitani E, Obayashi I, Shimizu K, Watanabe S. Persistent homology-based descriptor for machine-learning potential of amorphous structures. The Journal of Chemical Physics. 2023;159(8). doi:10.1063/5.0159349
BibTeX
@article{Minamitani2023,
author = {Minamitani, Emi and Obayashi, Ippei and Shimizu, Koji and Watanabe, Satoshi},
title = {Persistent homology-based descriptor for machine-learning potential of amorphous structures},
journal = {The Journal of Chemical Physics},
year = {2023},
volume = {159},
number = {8},
doi = {10.1063/5.0159349}
}
EUVIMED