Representing Born effective charges with equivariant graph convolutional neural networks

Abstract Graph convolutional neural networks have been instrumental in machine learning of material properties. When representing tensorial properties, weights and descriptors of a physics-informed network must obey certain transformation rules to ensure the independence of the property on the choic...

Full description

Saved in:
Bibliographic Details
Main Authors: Alex Kutana, Koji Shimizu, Satoshi Watanabe, Ryoji Asahi
Format: Article
Language:English
Published: Nature Portfolio 2025-05-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-025-01250-5
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849326731988566016
author Alex Kutana
Koji Shimizu
Satoshi Watanabe
Ryoji Asahi
author_facet Alex Kutana
Koji Shimizu
Satoshi Watanabe
Ryoji Asahi
author_sort Alex Kutana
collection DOAJ
description Abstract Graph convolutional neural networks have been instrumental in machine learning of material properties. When representing tensorial properties, weights and descriptors of a physics-informed network must obey certain transformation rules to ensure the independence of the property on the choice of the reference frame. Here we explicitly encode such properties using an equivariant graph convolutional neural network. The network respects rotational symmetries of the crystal throughout by using equivariant weights and descriptors and provides a tensorial output of the target value. Applications to tensors of atomic Born effective charges in diverse materials including perovskite oxides, Li3PO4, and ZrO2, are demonstrated, and good performance and generalization ability is obtained.
format Article
id doaj-art-b33f8e113a5e408f96803ebfe804df64
institution Kabale University
issn 2045-2322
language English
publishDate 2025-05-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-b33f8e113a5e408f96803ebfe804df642025-08-20T03:48:05ZengNature PortfolioScientific Reports2045-23222025-05-0115111010.1038/s41598-025-01250-5Representing Born effective charges with equivariant graph convolutional neural networksAlex Kutana0Koji Shimizu1Satoshi Watanabe2Ryoji Asahi3Nagoya UniversityNational Institute of Advanced Industrial Science and Technology (AIST)The University of TokyoNagoya UniversityAbstract Graph convolutional neural networks have been instrumental in machine learning of material properties. When representing tensorial properties, weights and descriptors of a physics-informed network must obey certain transformation rules to ensure the independence of the property on the choice of the reference frame. Here we explicitly encode such properties using an equivariant graph convolutional neural network. The network respects rotational symmetries of the crystal throughout by using equivariant weights and descriptors and provides a tensorial output of the target value. Applications to tensors of atomic Born effective charges in diverse materials including perovskite oxides, Li3PO4, and ZrO2, are demonstrated, and good performance and generalization ability is obtained.https://doi.org/10.1038/s41598-025-01250-5Equivariant graph convolutional neural networksPhysics-informed neural networksTensor of atomic Born effective chargesLinear responseOxides
spellingShingle Alex Kutana
Koji Shimizu
Satoshi Watanabe
Ryoji Asahi
Representing Born effective charges with equivariant graph convolutional neural networks
Scientific Reports
Equivariant graph convolutional neural networks
Physics-informed neural networks
Tensor of atomic Born effective charges
Linear response
Oxides
title Representing Born effective charges with equivariant graph convolutional neural networks
title_full Representing Born effective charges with equivariant graph convolutional neural networks
title_fullStr Representing Born effective charges with equivariant graph convolutional neural networks
title_full_unstemmed Representing Born effective charges with equivariant graph convolutional neural networks
title_short Representing Born effective charges with equivariant graph convolutional neural networks
title_sort representing born effective charges with equivariant graph convolutional neural networks
topic Equivariant graph convolutional neural networks
Physics-informed neural networks
Tensor of atomic Born effective charges
Linear response
Oxides
url https://doi.org/10.1038/s41598-025-01250-5
work_keys_str_mv AT alexkutana representingborneffectivechargeswithequivariantgraphconvolutionalneuralnetworks
AT kojishimizu representingborneffectivechargeswithequivariantgraphconvolutionalneuralnetworks
AT satoshiwatanabe representingborneffectivechargeswithequivariantgraphconvolutionalneuralnetworks
AT ryojiasahi representingborneffectivechargeswithequivariantgraphconvolutionalneuralnetworks