Universidad Nacional de Quilmes, Roque Saénz Peña 352, B1876BXD Bernal, Argentina.
Theoretical Division, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, United States.
J Phys Chem Lett. 2021 Mar 25;12(11):2970-2982. doi: 10.1021/acs.jpclett.1c00266. Epub 2021 Mar 17.
Direct atomistic simulation of nonadiabatic molecular dynamics is a challenging goal that allows important insights into fundamental physical phenomena. A variety of frameworks, ranging from fully quantum treatment of nuclei to semiclassical and mixed quantum-classical approaches, were developed. These algorithms are then coupled to specific electronic structure techniques. Such diversity and lack of standardized implementation make it difficult to compare the performance of different methodologies when treating realistic systems. Here, we compare three popular methods for large chromophores: Ehrenfest, surface hopping, and multiconfigurational Ehrenfest with multiple cloning (MCE-AIMC). These approaches are implemented in the NEXMD software, which features a common computational chemistry model. The resulting comparisons reveal the method performance for population relaxation and coherent vibronic dynamics. Finally, we study the numerical convergence of MCE-AIMC algorithms by considering the number of trajectories, cloning thresholds, and Gaussian wavepacket width. Our results provide helpful reference data for selecting an optimal methodology for simulating excited-state molecular dynamics.
直接原子模拟非绝热分子动力学是一个具有挑战性的目标，它可以深入了解基本物理现象。已经开发了各种框架，从原子核的完全量子处理到半经典和混合量子经典方法。然后将这些算法与特定的电子结构技术耦合。这种多样性和缺乏标准化的实现使得在处理实际系统时比较不同方法的性能变得困难。在这里，我们比较了三种用于大发色团的流行方法： Ehrenfest、表面跳跃和多组态 Ehrenfest 与多重克隆（MCE-AIMC）。这些方法在 NEXMD 软件中实现，该软件具有通用的计算化学模型。所得比较结果揭示了人口松弛和相干振动动力学的方法性能。最后，我们通过考虑轨迹数、克隆阈值和高斯波包宽度来研究 MCE-AIMC 算法的数值收敛性。我们的结果为选择模拟激发态分子动力学的最佳方法提供了有帮助的参考数据。
