Approximate but accurate quantum dynamics from the Mori formalism: II. Equilibrium correlation functions

TL;DR

This paper extends the Mori formalism to efficiently compute equilibrium correlation functions in quantum systems, demonstrating improved accuracy over traditional methods using the spin-boson model.

Contribution

It introduces a self-consistent Mori formalism approach for equilibrium correlation functions, utilizing a Dyson expansion and mean-field Ehrenfest dynamics for enhanced accuracy.

Findings

Achieves remarkable accuracy improvements over Ehrenfest dynamics

Successfully applies the method to the spin-boson model

Shows robustness to kernel closure choices and initial density variations

Abstract

The ability to efficiently and accurately calculate equilibrium time correlation functions of many-body condensed phase quantum systems is one of the outstanding problems in theoretical chemistry. The Nakajima-Zwanzig-Mori formalism coupled to the self-consistent solution of the memory kernel has recently proven to be highly successful for the computation of nonequilibrium dynamical averages. Here, we extend this formalism to treat symmetrized equilibrium time correlation functions for the spin-boson model. Following the first paper in this series [A. Montoya-Castillo and D. R. Reichman, J. Chem. Phys. $\bf{144}$, 184104 (2016)], we use a Dyson-type expansion of the projected propagator to obtain a self-consistent solution for the memory kernel that requires only the calculation of normally evolved auxiliary kernels. We employ the approximate mean-field Ehrenfest method to demonstrate the feasibility of this approach. Via comparison with numerically exact results for the correlation function $\mathcal{C}_{zz}(t) = \mathrm{Re}\langle \sigma_z(0)\sigma_z(t)\rangle$, we show that the current scheme affords remarkable boosts accuracy and efficiency over bare Ehrenfest dynamics. We further explore the sensitivity of the resulting dynamics to the type of kernel closures and the accuracy of the initial canonical density operator.