In this presentation, we report new exciting progresses in our development of multi-scale modeling approaches in two application areas: enzyme reactions and molecular polaritons. For enzyme simulations, we trained machine learning potentials (MLP) and delta machine learning potentials (DeltaMLP) to reproduce combined density functional theory molecular mechanics (DFT/MM) energy functions, and employed them to accelerate DFT/MM-quality free energy simulations. For chorismate mutase catalysis, we acquired DFT/MM-quality free energy profiles with a 32-46 fold speedup in computer time over conventional DFT/MM simulations. For molecular polaritons, we rederived the cavity electrodynamics time-dependent density functional theory (cQED-TDDFT) method, which now involves a Hermitian Hamiltonian matrix. Our cQED-TDDFT model allows us to properly account for all excited states of the (isolated) photochrome, and thus paves the way for accurate simulations of electronic processes involving molecular polaritons.

Multi-scale modeling of enzyme reactions and molecular polaritons

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.