Atomic Cluster Expansion for "Learning" Many-Body Interaction

The integration of machine learning (ML) into the traditional modeling workflows is replacing decades-old - often ad hoc - approximations (e.g., in constitutive laws) leading to new models that far outstrip their predecessors in accuracy and transferability. ”Pure” ML approaches are rarely successful but remarkable results can be achieved when integrated with domain knowledge. I will introduce a general formalism, the Atomic Cluster Expansion, for parameterizing many-body interaction and how it can be used for reduced-order modelling.