Our paper on FALCON, our active-learning framework for machine learning potentials, has been published open access in npj Computational Materials.

FALCON trains a machine learning potential while a molecular dynamics simulation is running. The model uses its own uncertainty to decide when a new quantum chemical calculation is needed, so expensive calculations are only done where they matter. This brings molecular dynamics with quantum chemical accuracy to larger systems and longer time scales.

Congratulations to Noah Felis, who developed FALCON during his research internship in our group! The code is open source:

pip install falcon-md