拓展 DeePMD：学习激发态能量、力及非绝对耦合
化学物理
2024-10-17 v2
摘要
我们扩展了 DeePMD 神经网络架构，用于预测进行非绝对动力学模拟所必需的电子结构属性。虽然学习激发态能量和力遵循 DeePMD 针对基态能量和力方法的直接扩展，但如何学习非绝对耦合向量（NACV）与 DeePMD 局部化学环境描述之间的映射则较为不那么直接。大多数基于机器学习的非绝对动力学实现本质上都会近似 NACV，假设能量差比例 NACV 为保守场。我们克服了这一近似方法，实现了 Richardson 于 2023 年在《J. Chem. Phys. 158 011102》中介绍的方法，该方法学习能量差比例 NACV 的对称配对。我们展示了该神经网络架构的效率和准确性，通过甲基亚尼ium阳离子 CHNH 作为示例。
关键词
引用
@article{arxiv.2407.10881,
title = {Exciting DeePMD: Learning excited state energies, forces, and non-adiabatic couplings},
author = {Lucien Dupuy and Neepa T. Maitra},
journal= {arXiv preprint arXiv:2407.10881},
year = {2024}
}
备注
Presentation of the 3 challenges in learning NACVs was clarified, with references added. A supplementary material section was added to clarify some points made in main text about methods' properties, together with more tests and comparisons of the different methods. Main results of these comparisons are summarized in main text with new figures. Ancillary files with code and data were added