Department of Applied Chemistry, Northwestern Polytechnical University, West Youyi Road 127, 710072 Xi'an, China.
J Chem Phys. 2019 Sep 21;151(11):114302. doi: 10.1063/1.5118692.
To deeply understand the neural-network (NN) fitting procedure in constructing a potential energy surface (PES) in a wide energy range with a rather small database, based on the existing BKMP2 PES of H + H, the relationship between NN function features and the size of the database is studied using the multiconfiguration time-dependent Hartree method for quantum dynamics calculations. First, employing 3843, 3843, 2024, and 1448 energy points, four independent NN-PESs are constructed to discuss the relationship among the size of the database, NN functional structure, and fitting accuracy. Dynamics calculations on these different NN PESs give similar reactive probabilities, which indicate that one has to balance the number of energy points for NN training and the number of neurons in the NN function. To explain this problem and try to resolve it, a quantitative model between the data volume and network scale is proposed. Then, this model is discussed and verified through 14 NN PESs fitted using 3843 energy points and various NN functional forms.
为了深入了解神经网络（NN）在构建具有相当小数据库的宽能量范围势能面（PES）中的拟合过程，基于现有的 H + H 的 BKMP2 PES，使用多组态含时哈特ree 方法进行量子动力学计算，研究了 NN 函数特征与数据库大小之间的关系。首先，使用 3843、3843、2024 和 1448 个能量点，构建了四个独立的 NN-PES，以讨论数据库大小、NN 功能结构和拟合精度之间的关系。在这些不同的 NN PES 上进行动力学计算得到了相似的反应概率，这表明人们必须平衡 NN 训练的能量点数量和 NN 函数中的神经元数量。为了解释这个问题并尝试解决它，提出了一个数据量和网络规模之间的定量模型。然后，通过使用 3843 个能量点和各种 NN 功能形式拟合的 14 个 NN PES 对该模型进行了讨论和验证。
