论文部分内容阅读
本文介绍了近年来使用神经网络构造分子体系势能面的进展,在以前的研究基础上完善了一套完整的分子构型选择方案,在严格的势能面评价标准下,完全解决了势能面构造过程中分子构型选择的难题.基于此方法,采用一系列拟合技巧,结合大量反应动力学计算,成功构造了一系列重要体系的高精度从头算势能面,并能得到可靠的化学动力学结果.
In this paper, we introduce the progress of using neural networks to construct the potential energy surface of molecular systems in recent years. Based on the previous studies, we have completed a complete set of molecular configuration options. Under the strict standard of potential energy surface evaluation, Based on this method, a series of fitting techniques are combined with a large number of reaction kinetic calculations to successfully construct a series of important ab initio potential energy surfaces and to obtain reliable chemical kinetic results .