Efficient crystal structure prediction based on the symmetry principle

IF 12 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS
Yu Han, Chi Ding, Junjie Wang, Hao Gao, Jiuyang Shi, Shaobo Yu, Qiuhan Jia, Shuning Pan, Jian Sun
{"title":"Efficient crystal structure prediction based on the symmetry principle","authors":"Yu Han, Chi Ding, Junjie Wang, Hao Gao, Jiuyang Shi, Shaobo Yu, Qiuhan Jia, Shuning Pan, Jian Sun","doi":"10.1038/s43588-025-00775-z","DOIUrl":null,"url":null,"abstract":"Crystal structure prediction (CSP) is an evolving field aimed at discerning crystal structures with minimal prior information. Despite the success of various CSP algorithms, their practical applicability remains circumscribed, particularly for large and complex systems. Here, to address this challenge, we show an evolutionary structure generator within the MAGUS (Machine Learning and Graph Theory Assisted Universal Structure Searcher) framework, inspired by the symmetry principle. This generator extracts both global and local features of explored crystal structures using group and graph theory. By integrating an on-the-fly space group miner and fragment reorganizer, augmented by symmetry-kept mutation, our approach generates higher-quality initial structures, reducing the computational costs of CSP tasks. Benchmarking tests show up to fourfold performance improvements. The method also proves valid in complex phosphorus allotrope systems. Furthermore, we apply our approach to the diamond–silicon (111)-(7 × 7) surface system, identifying up to 42 metastable structures within an 18 meV Å−2 energy range, demonstrating the efficacy of our approach in navigating challenging search spaces. This study presents a symmetry principle-biased crystal structure prediction scheme within the MAGUS framework, achieving up to a fourfold performance improvement compared with state-of-the-art methods.","PeriodicalId":74246,"journal":{"name":"Nature computational science","volume":"5 3","pages":"255-267"},"PeriodicalIF":12.0000,"publicationDate":"2025-02-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Nature computational science","FirstCategoryId":"1085","ListUrlMain":"https://www.nature.com/articles/s43588-025-00775-z","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS","Score":null,"Total":0}
引用次数: 0

Abstract

Crystal structure prediction (CSP) is an evolving field aimed at discerning crystal structures with minimal prior information. Despite the success of various CSP algorithms, their practical applicability remains circumscribed, particularly for large and complex systems. Here, to address this challenge, we show an evolutionary structure generator within the MAGUS (Machine Learning and Graph Theory Assisted Universal Structure Searcher) framework, inspired by the symmetry principle. This generator extracts both global and local features of explored crystal structures using group and graph theory. By integrating an on-the-fly space group miner and fragment reorganizer, augmented by symmetry-kept mutation, our approach generates higher-quality initial structures, reducing the computational costs of CSP tasks. Benchmarking tests show up to fourfold performance improvements. The method also proves valid in complex phosphorus allotrope systems. Furthermore, we apply our approach to the diamond–silicon (111)-(7 × 7) surface system, identifying up to 42 metastable structures within an 18 meV Å−2 energy range, demonstrating the efficacy of our approach in navigating challenging search spaces. This study presents a symmetry principle-biased crystal structure prediction scheme within the MAGUS framework, achieving up to a fourfold performance improvement compared with state-of-the-art methods.

Abstract Image

基于对称原理的高效晶体结构预测。
晶体结构预测(CSP)是一个以最小先验信息识别晶体结构为目标的新兴研究领域。尽管各种CSP算法取得了成功,但它们的实际适用性仍然有限,特别是对于大型和复杂的系统。在这里,为了解决这一挑战,我们展示了一个受对称原理启发的MAGUS(机器学习和图论辅助通用结构搜索器)框架内的进化结构生成器。该生成器使用群和图理论提取所探索晶体结构的全局和局部特征。通过集成动态空间群挖掘器和碎片重组器,再加上保持对称的突变,我们的方法产生了更高质量的初始结构,降低了CSP任务的计算成本。基准测试显示性能提高了四倍。该方法同样适用于复杂的磷同素异形体体系。此外,我们将我们的方法应用于金刚石-硅(111)-(7 × 7)表面体系,在18 meV Å-2能量范围内识别了多达42个亚稳结构,证明了我们的方法在导航具有挑战性的搜索空间中的有效性。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
CiteScore
11.70
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信