Atomistic Line Graph Neural Network for improved materials property predictions

Abstract Graph neural networks (GNN) have been shown to provide substantial performance improvements for atomistic material representation and modeling compared with descriptor-based machine learning models. While most existing GNN models for atomistic predictions are based on atomic distance inform...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Kamal Choudhary, Brian DeCost
Formato:	article
Lenguaje:	EN
Publicado:	Nature Portfolio 2021
Materias:	Materials of engineering and construction. Mechanics of materials TA401-492 Computer software QA76.75-76.765
Acceso en línea:	https://doaj.org/article/61a150a846574ac2a1dffb7b2b7fc925
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:61a150a846574ac2a1dffb7b2b7fc925
record_format	dspace
spelling	oai:doaj.org-article:61a150a846574ac2a1dffb7b2b7fc9252021-11-21T12:13:31ZAtomistic Line Graph Neural Network for improved materials property predictions10.1038/s41524-021-00650-12057-3960https://doaj.org/article/61a150a846574ac2a1dffb7b2b7fc9252021-11-01T00:00:00Zhttps://doi.org/10.1038/s41524-021-00650-1https://doaj.org/toc/2057-3960Abstract Graph neural networks (GNN) have been shown to provide substantial performance improvements for atomistic material representation and modeling compared with descriptor-based machine learning models. While most existing GNN models for atomistic predictions are based on atomic distance information, they do not explicitly incorporate bond angles, which are critical for distinguishing many atomic structures. Furthermore, many material properties are known to be sensitive to slight changes in bond angles. We present an Atomistic Line Graph Neural Network (ALIGNN), a GNN architecture that performs message passing on both the interatomic bond graph and its line graph corresponding to bond angles. We demonstrate that angle information can be explicitly and efficiently included, leading to improved performance on multiple atomistic prediction tasks. We ALIGNN models for predicting 52 solid-state and molecular properties available in the JARVIS-DFT, Materials project, and QM9 databases. ALIGNN can outperform some previously reported GNN models on atomistic prediction tasks by up to 85% in accuracy with better or comparable model training speed.Kamal ChoudharyBrian DeCostNature PortfolioarticleMaterials of engineering and construction. Mechanics of materialsTA401-492Computer softwareQA76.75-76.765ENnpj Computational Materials, Vol 7, Iss 1, Pp 1-8 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Materials of engineering and construction. Mechanics of materials TA401-492 Computer software QA76.75-76.765
spellingShingle	Materials of engineering and construction. Mechanics of materials TA401-492 Computer software QA76.75-76.765 Kamal Choudhary Brian DeCost Atomistic Line Graph Neural Network for improved materials property predictions
description	Abstract Graph neural networks (GNN) have been shown to provide substantial performance improvements for atomistic material representation and modeling compared with descriptor-based machine learning models. While most existing GNN models for atomistic predictions are based on atomic distance information, they do not explicitly incorporate bond angles, which are critical for distinguishing many atomic structures. Furthermore, many material properties are known to be sensitive to slight changes in bond angles. We present an Atomistic Line Graph Neural Network (ALIGNN), a GNN architecture that performs message passing on both the interatomic bond graph and its line graph corresponding to bond angles. We demonstrate that angle information can be explicitly and efficiently included, leading to improved performance on multiple atomistic prediction tasks. We ALIGNN models for predicting 52 solid-state and molecular properties available in the JARVIS-DFT, Materials project, and QM9 databases. ALIGNN can outperform some previously reported GNN models on atomistic prediction tasks by up to 85% in accuracy with better or comparable model training speed.
format	article
author	Kamal Choudhary Brian DeCost
author_facet	Kamal Choudhary Brian DeCost
author_sort	Kamal Choudhary
title	Atomistic Line Graph Neural Network for improved materials property predictions
title_short	Atomistic Line Graph Neural Network for improved materials property predictions
title_full	Atomistic Line Graph Neural Network for improved materials property predictions
title_fullStr	Atomistic Line Graph Neural Network for improved materials property predictions
title_full_unstemmed	Atomistic Line Graph Neural Network for improved materials property predictions
title_sort	atomistic line graph neural network for improved materials property predictions
publisher	Nature Portfolio
publishDate	2021
url	https://doaj.org/article/61a150a846574ac2a1dffb7b2b7fc925
work_keys_str_mv	AT kamalchoudhary atomisticlinegraphneuralnetworkforimprovedmaterialspropertypredictions AT briandecost atomisticlinegraphneuralnetworkforimprovedmaterialspropertypredictions
_version_	1718419125862137856

Atomistic Line Graph Neural Network for improved materials property predictions

Ejemplares similares