通过主动学习加快强非谐材料机器学习原子间位势的训练并提高其可靠性

Kisung Kang, Thomas A. R. Purcell, Christian Carbogno, Matthias Scheffler
{"title":"通过主动学习加快强非谐材料机器学习原子间位势的训练并提高其可靠性","authors":"Kisung Kang, Thomas A. R. Purcell, Christian Carbogno, Matthias Scheffler","doi":"arxiv-2409.11808","DOIUrl":null,"url":null,"abstract":"Molecular dynamics (MD) employing machine-learned interatomic potentials\n(MLIPs) serve as an efficient, urgently needed complement to ab initio\nmolecular dynamics (aiMD). By training these potentials on data generated from\nab initio methods, their averaged predictions can exhibit comparable\nperformance to ab initio methods at a fraction of the cost. However,\ninsufficient training sets might lead to an improper description of the\ndynamics in strongly anharmonic materials, because critical effects might be\noverlooked in relevant cases, or only incorrectly captured, or hallucinated by\nthe MLIP when they are not actually present. In this work, we show that an\nactive learning scheme that combines MD with MLIPs (MLIP-MD) and uncertainty\nestimates can avoid such problematic predictions. In short, efficient MLIP-MD\nis used to explore configuration space quickly, whereby an acquisition function\nbased on uncertainty estimates and on energetic viability is employed to\nmaximize the value of the newly generated data and to focus on the most\nunfamiliar but reasonably accessible regions of phase space. To verify our\nmethodology, we screen over 112 materials and identify 10 examples experiencing\nthe aforementioned problems. Using CuI and AgGaSe$_2$ as archetypes for these\nproblematic materials, we discuss the physical implications for strongly\nanharmonic effects and demonstrate how the developed active learning scheme can\naddress these issues.","PeriodicalId":501234,"journal":{"name":"arXiv - PHYS - Materials Science","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Accelerating the Training and Improving the Reliability of Machine-Learned Interatomic Potentials for Strongly Anharmonic Materials through Active Learning\",\"authors\":\"Kisung Kang, Thomas A. R. Purcell, Christian Carbogno, Matthias Scheffler\",\"doi\":\"arxiv-2409.11808\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Molecular dynamics (MD) employing machine-learned interatomic potentials\\n(MLIPs) serve as an efficient, urgently needed complement to ab initio\\nmolecular dynamics (aiMD). By training these potentials on data generated from\\nab initio methods, their averaged predictions can exhibit comparable\\nperformance to ab initio methods at a fraction of the cost. However,\\ninsufficient training sets might lead to an improper description of the\\ndynamics in strongly anharmonic materials, because critical effects might be\\noverlooked in relevant cases, or only incorrectly captured, or hallucinated by\\nthe MLIP when they are not actually present. In this work, we show that an\\nactive learning scheme that combines MD with MLIPs (MLIP-MD) and uncertainty\\nestimates can avoid such problematic predictions. In short, efficient MLIP-MD\\nis used to explore configuration space quickly, whereby an acquisition function\\nbased on uncertainty estimates and on energetic viability is employed to\\nmaximize the value of the newly generated data and to focus on the most\\nunfamiliar but reasonably accessible regions of phase space. To verify our\\nmethodology, we screen over 112 materials and identify 10 examples experiencing\\nthe aforementioned problems. Using CuI and AgGaSe$_2$ as archetypes for these\\nproblematic materials, we discuss the physical implications for strongly\\nanharmonic effects and demonstrate how the developed active learning scheme can\\naddress these issues.\",\"PeriodicalId\":501234,\"journal\":{\"name\":\"arXiv - PHYS - Materials Science\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-09-18\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - PHYS - Materials Science\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2409.11808\",\"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 - Materials Science","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.11808","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

采用机器学习原子间位势(MLIPs)的分子动力学(MD)是对原子分子动力学(ab initi-molecular dynamics,aiMD)的一种高效、急需的补充。通过对原子间位势进行训练,这些位势的平均预测结果可以显示出与原子间位势方法相当的性能,而成本仅为原子间位势方法的一小部分。然而,训练集不足可能会导致对强非谐波材料动力学的描述不当,因为临界效应可能会在相关情况下被忽略,或只是被错误地捕捉,或在实际上并不存在的情况下被 MLIP 幻化。在这项工作中,我们展示了一种将 MD 与 MLIP(MLIP-MD)和不确定性估计相结合的主动学习方案,可以避免此类问题预测。简而言之,高效的 MLIP-MD 可用于快速探索构型空间,而基于不确定性估计和能量可行性的获取函数可用于最大化新生成数据的价值,并将重点放在最不熟悉但可合理访问的相空间区域。为了验证我们的方法,我们筛选了超过 112 种材料,找出了 10 个遇到上述问题的例子。我们以 CuI 和 AgGaSe$_2$ 作为这些问题材料的原型,讨论了强谐波效应的物理意义,并展示了所开发的主动学习方案如何解决这些问题。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
Accelerating the Training and Improving the Reliability of Machine-Learned Interatomic Potentials for Strongly Anharmonic Materials through Active Learning
Molecular dynamics (MD) employing machine-learned interatomic potentials (MLIPs) serve as an efficient, urgently needed complement to ab initio molecular dynamics (aiMD). By training these potentials on data generated from ab initio methods, their averaged predictions can exhibit comparable performance to ab initio methods at a fraction of the cost. However, insufficient training sets might lead to an improper description of the dynamics in strongly anharmonic materials, because critical effects might be overlooked in relevant cases, or only incorrectly captured, or hallucinated by the MLIP when they are not actually present. In this work, we show that an active learning scheme that combines MD with MLIPs (MLIP-MD) and uncertainty estimates can avoid such problematic predictions. In short, efficient MLIP-MD is used to explore configuration space quickly, whereby an acquisition function based on uncertainty estimates and on energetic viability is employed to maximize the value of the newly generated data and to focus on the most unfamiliar but reasonably accessible regions of phase space. To verify our methodology, we screen over 112 materials and identify 10 examples experiencing the aforementioned problems. Using CuI and AgGaSe$_2$ as archetypes for these problematic materials, we discuss the physical implications for strongly anharmonic effects and demonstrate how the developed active learning scheme can address these issues.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
期刊最新文献
Anionic disorder and its impact on the surface electronic structure of oxynitride photoactive semiconductors Accelerating the Training and Improving the Reliability of Machine-Learned Interatomic Potentials for Strongly Anharmonic Materials through Active Learning Hybridization gap approaching the two-dimensional limit of topological insulator Bi$_x$Sb$_{1-x}$ Sampling Latent Material-Property Information From LLM-Derived Embedding Representations Smart Data-Driven GRU Predictor for SnO$_2$ Thin films Characteristics
×
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
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 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 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1