| 刘志荣.化学通报,2021,84(11):1205-1209. |
| 机器学习在团簇研究中的可能应用 |
| Possible applications of machine learning in cluster research |
| 投稿时间：2021-03-19 修订日期：2021-04-24 |
| DOI： |
| 中文关键词: 团簇 机器学习 力场 第一性原理计算 结构预测 |
| 英文关键词:Cluster machine learning force field first-principles calculation structure prediction |
| 基金项目:国家自然科学基金项目(21773002)资助 |
|
| 摘要点击次数: 1355 |
| 全文下载次数: 0 |
| 中文摘要: |
| 团簇的结构-能量关系非常复杂,存在大量的局部能量极小点,寻找全局稳定结构是个很艰巨的任务。对于中大尺寸团簇的计算,采用纯量子力学计算方法将面临非常重的计算负担,而如果采用传统力场的方法则会面临精度不足的困难。近年来崛起的机器学习浪潮,开始渗透到包括化学在内的各个学科领域。基于机器学习的方法,有可能提供一条介乎量子力学与传统力场之间的中间新路线。发展基于机器学习的团簇势能模型,用于寻找中大尺寸团簇的稳定结构,有望为团簇的理论与计算研究提供新思路与新手段。 |
| 英文摘要: |
| The structure-energy relationship of clusters is complicated as there are a lot of local energy minima, so it is extremely difficult to predict their global stable structure. For the calculation of large/medium clusters, first-principles calculations have heavy computational burden, while the traditional force-field methods are not accurate enough. In recent years, the rising tide of machine learning began to penetrate into various academic disciplines including chemistry. The method based on machine learning may provide a new way between first-principles approach and traditional force-field model. The development of cluster potential energy model based on machine learning is expected to provide a new tool for the theoretical and computational research of clusters. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |