A comparative analysis of surface hopping acceptance and decoherence algorithms within the neglect of back-reaction approximation

We have implemented a Python-based software package within the Libra software for performing nonadiabatic molecular dynamics (NA-MD) within the neglect of back reaction approximation (NBRA). Available in the software are a wide variety of proposed hop acceptance (PHA) and decoherence methodologies. Using Libra, a comparative analysis of PHA schemes and decoherence methods is performed to examine thermal equilibrium in NA-MD simulations within the NBRA. The analysis is performed using 3 model systems, each of which highlights the effects of the different decoherence methods and PHA schemes on NA transitions. We find that the interplay between decoherence and PHA schemes is important for achieving detailed balance in the NBRA and discuss the conditions by which the detailed balance is achieved for each model. We discuss the qualitative features of NA dynamics computed using various combinations of decoherence and PHA schemes for a wide range of model and condition parameters such as temperature, energy gap magnitude, and dephasing times. Furthermore, we extend the analysis to include the Boltzmann corrected Ehrenfest methodology of Bastida and co-workers and compare the dynamics produced with it with that obtained using the surface hopping-based approach.