Adaptive Machine Learning Approximations for Quantum Simulations of Hydrogen Bond Dynamics

Supervisors: Prof. Jochen Küpper (CFEL, DESY, UHH), Dr. Andrey Yachmenev (CFEL, DESY), Prof. Armin Iske (UHH)

First principles simulations of the hydrogen-bond dynamics in molecular complexes with water is one of the most interesting, computationally challenging, and accuracy demanding problems in the field of computational physics and chemistry. Accurate simulations of molecular and chemical dynamics in weakly-bound complexes with water will help to advance our understanding of many organic and biochemical processes. The development of rigorous variational approaches to nuclear motion dynamics of polyatomic molecules enables the accurate theoretical modeling of simple cluster systems. However the exponential scaling of the computational burden with the number of degrees of freedom, i.e., cluster size, renders the computational costs of existing variational methods unaffordable for clusters composed of more than two molecules. The goal of this project is to explore and implement the paradigm of machine-learning to mitigate or break the curse of dimensionality in variational calculations of larger clusters.