Nonadiabatic semiclassical dynamics in the mixed quantum-classical initial value representation

TL;DR

This paper extends the MQC-IVR semiclassical method to nonadiabatic systems using the MMST Hamiltonian, introducing an efficient symplectic integrator and demonstrating its accuracy and convergence in model systems.

Contribution

It develops a nonadiabatic extension of MQC-IVR with a new symplectic integration scheme and analyzes the impact of quantizing different degrees of freedom.

Findings

MQC-IVR agrees with quantum results in the quantum limit.

In the classical limit, results align with mean-field approaches.

Quantizing degrees of freedom improves convergence without losing accuracy.

Abstract

We extend the Mixed Quantum-Classical Initial Value Representation (MQC-IVR), a semiclassical method for computing real-time correlation functions, to electronically nonadiabatic systems using the Meyer-Miller-Stock-Thoss (MMST) Hamiltonian to treat electronic and nuclear degrees of freedom (dofs) within a consistent dynamic framework. We introduce an efficient symplectic integration scheme, the MInt algorithm, for numerical time-evolution of the nuclear and electronic phase space variables as well as the Monodromy matrix, under the non-separable MMST Hamiltonian. We then calculate the probability of transmission through a curve-crossing in model two-level systems and show that in the quantum limit MQC-IVR is in good agreement with the exact quantum results, whereas in the classical limit the method yields results in keeping with mean-field approaches like the Linearized Semiclassical IVR. Finally, exploiting the ability of MQC-IVR to quantize different dofs to different extents, we present a detailed study of the extents to which quantizing the nuclear and electronic dofs improves numerical convergence properties without significant loss of accuracy.