Interpolating numerically exact many-body wave functions for accelerated molecular dynamics
The wave function contains all the information about the electrons in chemical systems. In this work, this quantum variable is interpolated through the space of chemical trajectories, allowing us to ‘follow’ the electronic state as atoms move and react. This then retains the electronic information of the system as opposed to more traditional approaches to machine learning interatomic force fields, and enables accurate interpolation of all properties of the system, including excited states. This also naturally fits with training from correlated electronic structure methods, and we show the importance of this over thermalized ensembles of Zundel cations as a minimal unit for proton transfer.
Authors: Yannic Rath & George Booth
