Alexandre Boucher, Cameron Beevers, Bertrand Gauthier, Alberto Roldan
{"title":"加速结构稳定性无能量计算器","authors":"Alexandre Boucher, Cameron Beevers, Bertrand Gauthier, Alberto Roldan","doi":"arxiv-2408.14577","DOIUrl":null,"url":null,"abstract":"Computational modeling is an integral part of catalysis research. With it,\nnew methodologies are being developed and implemented to improve the accuracy\nof simulations while reducing the computational cost. In particular, specific\nmachine-learning techniques have been applied to build interatomic potential\nfrom ab initio results. Here, We report an energy-free machine-learning\ncalculator that combines three individually trained neural networks to predict\nthe energy and atomic forces of particulate matter. Three structures were\ninvestigated: a monometallic nanoparticle, a bimetallic nanoalloy, and a\nsupported metal crystallites. Atomic energies were predicted via a graph neural\nnetwork, leading to a mean absolute error (MAE) within 0.004 eV from Density\nFunctional Theory (DFT) calculations. The task of predicting atomic forces was\nsplit over two feedforward networks, one predicting the force's norm and\nanother its direction. The force prediction resulted in a MAE within 0.080 eV/A\nagainst DFT results. The interpretability of the graph neural network\npredictions was demonstrated by underlying the physics of the monometallic\nparticle in the form of cohesion energy.","PeriodicalId":501259,"journal":{"name":"arXiv - PHYS - Atomic and Molecular Clusters","volume":"8 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-08-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Accelerated structure-stability energy-free calculator\",\"authors\":\"Alexandre Boucher, Cameron Beevers, Bertrand Gauthier, Alberto Roldan\",\"doi\":\"arxiv-2408.14577\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Computational modeling is an integral part of catalysis research. With it,\\nnew methodologies are being developed and implemented to improve the accuracy\\nof simulations while reducing the computational cost. In particular, specific\\nmachine-learning techniques have been applied to build interatomic potential\\nfrom ab initio results. Here, We report an energy-free machine-learning\\ncalculator that combines three individually trained neural networks to predict\\nthe energy and atomic forces of particulate matter. Three structures were\\ninvestigated: a monometallic nanoparticle, a bimetallic nanoalloy, and a\\nsupported metal crystallites. Atomic energies were predicted via a graph neural\\nnetwork, leading to a mean absolute error (MAE) within 0.004 eV from Density\\nFunctional Theory (DFT) calculations. The task of predicting atomic forces was\\nsplit over two feedforward networks, one predicting the force's norm and\\nanother its direction. The force prediction resulted in a MAE within 0.080 eV/A\\nagainst DFT results. The interpretability of the graph neural network\\npredictions was demonstrated by underlying the physics of the monometallic\\nparticle in the form of cohesion energy.\",\"PeriodicalId\":501259,\"journal\":{\"name\":\"arXiv - PHYS - Atomic and Molecular Clusters\",\"volume\":\"8 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-08-26\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - PHYS - Atomic and Molecular Clusters\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2408.14577\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - PHYS - Atomic and Molecular Clusters","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2408.14577","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
摘要
计算建模是催化研究不可或缺的一部分。随着它的发展,人们正在开发和实施新的方法,以提高模拟的准确性,同时降低计算成本。特别是,特定的机器学习技术已被用于根据原子序数结果建立原子间势。在此,我们报告了一种无能量机器学习计算器,它结合了三个单独训练的神经网络来预测微粒物质的能量和原子力。我们研究了三种结构:单金属纳米粒子、双金属纳米合金和支撑金属晶体。通过图神经网络预测了原子能量,与密度函数理论(DFT)计算结果相比,平均绝对误差(MAE)在 0.004 eV 以内。预测原子力的任务由两个前馈网络分担,一个预测力的标准,另一个预测力的方向。力预测结果与 DFT 结果的最大误差在 0.080 eV/Aagainst 范围内。图神经网络预测的可解释性通过以内聚能形式为基础的单金属粒子物理学得到了证明。
Computational modeling is an integral part of catalysis research. With it,
new methodologies are being developed and implemented to improve the accuracy
of simulations while reducing the computational cost. In particular, specific
machine-learning techniques have been applied to build interatomic potential
from ab initio results. Here, We report an energy-free machine-learning
calculator that combines three individually trained neural networks to predict
the energy and atomic forces of particulate matter. Three structures were
investigated: a monometallic nanoparticle, a bimetallic nanoalloy, and a
supported metal crystallites. Atomic energies were predicted via a graph neural
network, leading to a mean absolute error (MAE) within 0.004 eV from Density
Functional Theory (DFT) calculations. The task of predicting atomic forces was
split over two feedforward networks, one predicting the force's norm and
another its direction. The force prediction resulted in a MAE within 0.080 eV/A
against DFT results. The interpretability of the graph neural network
predictions was demonstrated by underlying the physics of the monometallic
particle in the form of cohesion energy.