Knight Chris, Maupin C Mark, Izvekov Sergei, Voth Gregory A
Department of Chemistry, James Franck Institute, and Computation Institute, University of Chicago, 5735 South Ellis Avenue, Chicago, Illinois 60637, United States, and Center for Biophysical Modeling and Simulation and Department of Chemistry, University of Utah, Salt Lake City, Utah 84112, United States.
J Chem Theory Comput. 2010 Oct 12;6(10):3223-32. doi: 10.1021/ct1004438. Epub 2010 Sep 23.
In this report, a general methodology is presented for the parametrization of a reactive force field using data from a condensed phase ab initio molecular dynamics (AIMD) simulation. This algorithm allows for the creation of an empirical reactive force field that accurately reproduces the underlying ab initio reactive surface while providing the ability to achieve long-time statistical sampling for large systems not possible with AIMD alone. In this work, a model for the hydrated excess proton is constructed where the hydronium cation and proton hopping portions of the model are statistically force-matched to the results of Car-Parrinello Molecular Dynamics (CPMD) simulations. The flexible nature of the algorithm also allows for the use of the more accurate classical simple point-charge flexible water (SPC/Fw) model to describe the water-water interactions while utilizing the ab initio data to create an overall multistate molecular dynamics (MS-MD) reactive model of the hydrated excess proton in water. The resulting empirical model for the system qualitatively reproduces thermodynamic and dynamic properties calculated from the ab initio simulation while being in good agreement with experimental results and previously developed multistate empirical valence bond (MS-EVB) models. The present methodology, therefore, bridges the AIMD technique with the MS-MD modeling of reactive events, while incorporating key strengths of both.
在本报告中，提出了一种通用方法，用于使用来自凝聚相从头算分子动力学（AIMD）模拟的数据对反应力场进行参数化。该算法允许创建一个经验反应力场，该力场能够准确再现潜在的从头算反应表面，同时提供对大型系统进行长时间统计采样的能力，而这是单独使用AIMD无法实现的。在这项工作中，构建了一个水合过量质子模型，其中模型的水合氢离子和质子跳跃部分在统计上与Car-Parrinello分子动力学（CPMD）模拟结果进行力匹配。该算法的灵活性还允许使用更精确的经典简单点电荷柔性水（SPC/Fw）模型来描述水-水相互作用，同时利用从头算数据创建水中水合过量质子的整体多态分子动力学（MS-MD）反应模型。由此得到的系统经验模型定性地再现了从头算模拟计算得到的热力学和动力学性质，同时与实验结果和先前开发的多态经验价键（MS-EVB）模型高度吻合。因此，本方法将AIMD技术与反应事件的MS-MD建模联系起来，同时融合了两者的关键优势。
