Physically Motivated Recursively Embedded Atom Neural Networks: Incorporating Local Completeness and Nonlocality

Recent advances in machine-learned interatomic potentials largely benefit from the atomistic representation and locally invariant many-body descriptors. It was, however, recently argued that including three-body (or even four-body) features is incomplete to distinguish specific local structures. Utilizing an embedded density descriptor made by linear combinations of neighboring atomic orbitals and realizing that each orbital coefficient physically depends on its own local environment, we propose a recursively embedded atom neural network model. We formally prove that this model can efficiently incorporate complete many-body correlations without explicitly computing high-order terms. This model not only successfully addresses challenges regarding local completeness and nonlocality in representative systems, but also provides an easy and general way to update local many-body descriptors to have a messagepassing form without changing their basic structures.

Automated summary

Recent advancements in machine-learned interatomic potentials have benefited greatly from the use of atomistic representations and locally invariant many-body descriptors. A new recursively embedded atom neural network model has been proposed, which efficiently incorporates complete many-body correlations without explicitly computing high-order terms. This model successfully addresses challenges regarding local completeness and nonlocality in representative systems and offers a simple and general way to update local many-body descriptors into a message-passing form without altering their basic structures.