作者: Jun Li , Jun Chen , Dong H. Zhang , Hua Guo
DOI: 10.1063/1.4863138
关键词:
摘要: A permutationally invariant global potential energy surface for the HOCO system is reported by fitting a larger number of high-level ab initio points using newly proposed permutation polynomial-neural network method. The small error (∼5 meV) indicates faithful representation over large configuration space. Full-dimensional quantum and quasi-classical trajectory studies title reaction were performed on this surface. While results suggest that differences between an earlier neural fits are small, discrepancies with state-to-state experimental data remain significant.