论文部分内容阅读
在密度泛函理论的基础上,根据软、硬酸碱原理,通过孤对电子计算得到的各种化学性质,我们建立了一种定量研究有机化合物亲核性/碱性的方法。这些化学性质描述符包括全局柔性,Fukui方程,局部柔性,局部Mulliken电荷等,它们是通过PC Spartan Pro软件,在密度泛函理论SVWN/DN~*层面上计算得到的量。在本研究工作中,选择了28个化合物,基于计算得到的密度将它们按照亲核性/碱性进行了分类。神经网络中的BP算法被用来研究所选取描述符与亲核性/碱性之间的关系。通过交叉验证避免了在神经网络训练中可能会出现的过度拟合问题。结果显示基于所建立神经网络的预测结果与已知的实验结果符合的很好。局部柔性与Fukui方程与亲核性之间有着相当好的对应关系,而局部Mulliken电荷更好地反映了碱性。我们期望所建立和训练得到的BP网络模型可以被用来预测未知化合物和功能基团的亲核性/碱性。
Based on the density functional theory, we have established a method for the quantitative study of the nucleophilic / basicity of organic compounds based on the chemical and physical properties calculated by the lone pair of electrons based on the principles of soft and hard acids and bases. These chemical descriptors include global flexibility, Fukui equation, local flexibility, local Mulliken charge, etc., which are calculated by density functional theory (SVWN / DN ~ *) using PC Spartan Pro software. In this work, 28 compounds were selected and they were classified according to nucleophilic / basicity based on the calculated density. The BP algorithm in the neural network is used to study the relationship between the selected descriptors and nucleophilic / basicity. Cross-validation avoids over-fitting problems that may occur in neural network training. The results show that the prediction results based on the established neural network are in good agreement with the known experimental results. The local flexibility and the Fukui equation have a good correspondence with the nucleophilicity, whereas the local Mulliken charge better reflects the alkalinity. We expect that the BP network model built and trained can be used to predict the nucleophilic / basicity of unknown compounds and functional groups.