Learning Non-Local Molecular Interactions via Equivariant Local Representations and Charge Equilibration
The Charge Equilibration Layer for Long-range Interactions (CELLI) is a new building block for equivariant Graph Neural Networks. It generalizes the charge equilibration method to efficiently model non-local interactions, like electrostatic forces and charge transfer, which traditional local models miss. CELLI integrates these long-range effects into local potentials, improving accuracy in molecular simulations.