Machine learning-driven investigation of the structure and dynamics of the BMIM-BF4 room temperature ionic liquid

We demonstrate a learning-on-the-fly procedure to train machine-learned potentials from single-point density functional theory calculations before performing production molecular dynamics simulations.