Search papers, labs, and topics across Lattice.
This paper introduces grappa-fullFF, a unified framework that simultaneously learns both bonded and nonbonded parameters for classical molecular mechanics force fields directly from ab initio data. By incorporating physically inspired regularization, including electrostatic potential supervision and a charge equilibration mechanism, the model achieves state-of-the-art accuracy on geometry optimization and reproduces conformational sampling without relying on empirical nonbonded parameters. This approach bridges the gap between the efficiency of classical force fields and the accuracy of ab initio methods.
Classical force fields can now learn both bonded and nonbonded parameters consistently from first principles, achieving ab initio accuracy without empirical nonbonded terms.
Classical force fields (FFs) remain the workhorse for large-scale simulations even as machine-learned interatomic potentials (MLIPs) approach ab initio accuracy. They decompose total configuration energies into simple effective interactions whose parameters are traditionally assigned based on atom or bond types, enabling efficient simulations but also limiting their ability to adapt across configurations. Recent machine learning approaches have improved the accuracy and transferability of bonded parameters in these FFs by inferring them as functions of local atomic environments, but still rely on empirical nonbonded parameters for practical simulations. In this work, we introduce a unified approach, \texttt{grappa-fullFF}, which learns both bonded and nonbonded parameters \emph{consistently} and simultaneously from ab initio reference data. By incorporating physically inspired regularization via supervision of the electrostatic potential and an architecture that facilitates charge equilibration, our model recovers accurate electric response properties, achieves state-of-the-art accuracy on geometry optimization benchmarks, and reproduces the conformational sampling of both classical and existing machine-learned FFs, without relying on externally assigned nonbonded parameters.
