Methods.JPG

Simulation algorithms

We extensively use enhanced sampling simulations combined with force matching ab-initio derived potentials. Specifically we use well-tempered, (Barducci et al. Phys Rev. Lett. 2008, 100, 020603), parallel bias (Pfaendtner, Bonomi J. Chem Theory Comput. 2015, 11, 5062–5067) metadynamics (Laio, Parrinello Proc. Natl. Acad. Sci. USA 99, 12562-12566) within PLUMED (Bonomi et al. Nature Methods 2019, 16, 670–673).

From the free-energy minima ensembles, we predict a variety of properties (i.e. absorbance, emission, dichroism, reactivity) within ab-initio methods including DFT and TD-DFT framework. The property predictions can be oppurtunately coupled with learning-based algorithms.