Multidimensional Tunneling Dynamics Employing Quantum-Trajectory Guided Adaptable Gaussian Bases
Multidimensional Tunneling Dynamics Employing Quantum-Trajectory Guided Adaptable Gaussian Bases
复制标题
采用量子轨迹引导的自适应高斯基的多维隧道动力学
DOI：
10.1021/acs.jpca.0c07168
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Garashchuk, Sophya
中科院分区：
文献类型：
--
作者：
Dutra, Matthew;Wickramasinghe, Sachith;Garashchuk, Sophya
含时波函数的有效基表示对于高维大振幅运动分子系统的理论研究是必不可少的。对于完全耦合的非谐系统，一般波函数的复杂性与系统大小成指数关系;因此，出于实际原因，需要将基函数调整为与时间相关的波函数。在寻求最小基表示的过程中，经常会使用依赖于时间的高斯函数，部分原因是它们在组态和动量空间中的局部化，也因为它们与经典和半经典动力学的直接联系，指导基函数参数的演化。在这项工作中，量子轨道引导的自适应高斯（QTAG）基方法[J. Chem. Theory Comput.2020，16，18 - 34]被推广到包括相关的，即，不可因式分解的基函数，QTAG动力学的性能在多达20个维度的基准系统/浴隧道模型上进行评估。对于流行的初始条件的选择描述隧道之间的反应物/产品威尔斯，最小的“半经典”的描述浴模式基本上使用一个单一的多维基函数结合的多高斯表示的隧道模式示出捕捉动态的主要特征，在一个高效的方式。
An efficient basis representation of time-dependent wavefunctions is essential for theoretical studies of high-dimensional molecular systems exhibiting large-amplitude motion. For fully coupled anharmonic systems, the complexity of a general wavefunction scales exponentially with the system size; therefore, for practical reasons, it is desirable to adapt the basis to the time-dependent wavefunction at hand. Often times on this quest for a minimal basis representation, time-dependent Gaussians are employed, in part because of their localization in both configuration and momentum spaces and also because of their direct connection to classical and semiclassical dynamics, guiding the evolution of the basis function parameters. In this work, the quantum-trajectory guided adaptable Gaussian (QTAG) bases method [J. Chem. Theory Comput.2020, 16, 18−34] is generalized to include correlated, i.e., non-factorizable, basis functions, and the performance of the QTAG dynamics is assessed on benchmark system/bath tunneling models of up to 20 dimensions. For the popular choice of initial conditions describing tunneling between the reactant/product wells, the minimal “semiclassical” description of the bath modes using essentially a single multidimensional basis function combined with the multi-Gaussian representation of the tunneling mode is shown to capture the dominant features of dynamics in a highly efficient manner.