Transfer learning for atomistic simulations using GNNs and kernel mean embeddings

Interatomic potentials learned using machine learning methods have been successfully applied to atomistic simulations.