物理化学学报 >> 2016, Vol. 32 >> Issue (1): 119-130.doi: 10.3866/PKU.WHXB201512011

综述 上一篇    下一篇

基于新一代密度泛函和神经网络势能面的量子反应动力学计算

苏乃强1,陈俊2,徐昕1,*(),张东辉2,*()   

  1. 1 复旦大学化学系,分子催化与功能材料上海市重点实验室,物质计算科学教育部重点实验室,上海 200433
    2 中国科学院大连化学物理研究所,分子反应动力学国家重点实验室,理论与计算化学研究中心,辽宁大连 116023
  • 收稿日期:2015-09-30 发布日期:2016-01-13
  • 通讯作者: 徐昕,张东辉 E-mail:xxchem@fudan.edu.cn;zhangdh@dicp.ac.cn
  • 基金资助:
    国家自然科学基金(91427301, 91221301, 21433009, 21133004);国家重点基础研究发展规划项目(973)(2013CB834601, 2013CB834606);中国科学院资助

Quantum Reaction Dynamics Based on a New Generation Density Functional and Neural Network Potential Energy Surfaces

Neil-Qiang SU1,Jun CHEN2,Xin XU1,*(),Dong-H. ZHANG2,*()   

  1. 1 Department of Chemistry, Shanghai Key Laboratory of Molecular Catalysis and Innovative Materials, MOE Key Laboratory of Computational Physical Sciences, Fudan University, Shanghai 200433, P. R. China
    2 State Key Laboratory of Molecular Reaction Dynamics, Center for Theoretical and Computational Chemistry, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, Liaoning Province, P. R. China
  • Received:2015-09-30 Published:2016-01-13
  • Contact: Xin XU,Dong-H. ZHANG E-mail:xxchem@fudan.edu.cn;zhangdh@dicp.ac.cn
  • Supported by:
    the National Natural Science Foundation of China(91427301, 91221301, 21433009, 21133004);National Key BasicResearch Program of China (973)(2013CB834601, 2013CB834606);Chinese Academy of Sciences

摘要:

介绍了近年来发展起来的新一代密度泛函XYG3及利用神经网络构造分子体系势能面的最新进展。以H3和CH5等体系为实例,表明基于高效准确的密度泛函电子结构计算,与神经网络高精度势能面构造的理想结合,可以得到可靠的化学动力学结果,并有望用于更大更复杂的体系。

关键词: 密度泛函, 势能面, 神经网络, 第一性原理, 反应动力学

Abstract:

Recent progresses on a new generation density functional XYG3 and the construction of potential energy surfaces using neural networks are reviewed in this article. Using H3 and CH5 systems as illustrative examples, it is concluded that highly reliable dynamics results can be obtained from the combination of electronic structure calculations based on efficient and accurate density functionals and accurate potential energy surfaces using neural networks. It holds promise for future applications in larger and more complicated systems.

Key words: Density functional, Potential energy surface, Neural network, First principles, Reaction dynamics

MSC2000: 

  • O643