A path-integral approach to real-time quantum dynamics is presented which is suitable to treat the dynamics of vibronic coupling or spin boson models. In these models the vibrational dynamics is nonseparable as a consequence of the electronic inter-state coupling. The sum over all possible paths in electronic-state space generated by the usual Trotter procedure is expressed in terms of single-mode averages over classes of paths and statistical mode correlations. The averages for classes of a given length can be calculated iteratively from averages over shorter paths. This expansion is formally exact and finite for a finite number of modes. Usually only a limited number of terms has to be evaluated in order to obtain converged results. The scaling of the computational effort with respect to the number of time steps and the number of modes is given by a low-order power law, depending on the chosen class structure and the order of the expansion. The usual time-dependent wave-packet propagation and the full path enumeration, which exhibit an exponential scaling behavior with respect to either the number of modes or the number of time steps, can be considered as opposite limiting cases of the correlation expansion (CE) of the path integral. The convergence of the CE is tested by application to a two-state four-mode model representing S1-S2 vibronic coupling in pyrazine, for which exact references (time-dependent correlation functions) are available. The potential of the CE approximation for the treatment of multi-mode problems is demonstrated by application to an extended 24-mode vibronic-coupling model. This model is suitable to provide a microscopic description of ultrafast optical dephasing processes in large molecules. © 1995 American Institute of Physics.
CITATION STYLE
Krempl, S., Winterstetter, M., & Domcke, W. (1995). Path-integral treatment of multi-mode vibronic coupling. II. Correlation expansion of class averages. The Journal of Chemical Physics, 102(16), 6499–6510. https://doi.org/10.1063/1.469364
Mendeley helps you to discover research relevant for your work.