Theoretical and Computational Chemistry

One-shot trajectory learning of open quantum systems dynamics



Nonadiabatic quantum dynamics are important for understanding lightharvesting processes, but their propagation with tradition methods can be rather expensive. Here we present a one-short trajectory learning approach that allows to directly make ultra-fast prediction of the entire trajectory for a new set of such simulation parameters as temperature and reorganization energy. The whole 2.5 ps long propagation takes 50 milliseconds as we demonstrate on the comparatively large quantum system, the Fenna–Matthews–Olsen (FMO) complex. Our approach also significantly reduces time and memory requirements for training.


Thumbnail image of one_shot_traj.pdf

Supplementary weblinks

Dral's group
The goal of research is accelerating and improving computational chemistry with artificial intelligence / machine learning.