Faster Molecular Dynamics with Neural Network Potentials via Distilled Multiple Time-Stepping and Non-Conservative Forces

Following our previous work (J. Phys. Chem. Lett., 2026, 17, 5, 1288-1295), we propose the DMTS-NC approach, a distilled multi-time-step (DMTS) strategy using non conservative (NC) forces to further accelerate atomistic molecular dynamics simulations using foundation neural network models. There, a dual-level reversible reference system propagator algorithm (RESPA) formalism couples a target accurate conservative potential to a simplified distilled representation optimized for the production of non-conservative forces. Despite being non-conservative, the distilled architecture is designed to enforce key physical priors, such as equivariance under rotation and cancellation of atomic force components. These choices facilitate the distillation process and therefore improve drastically the robustness of simulation, significantly limiting the "holes" in the simpler potential, thus achieving excellent agreement with the forces data. Overall, the DMTS-NC scheme is found to be more stable and efficient than its conservative counterpart with additional speedups reaching 15-30% over DMTS. Requiring no finetuning steps, it is easier to implement and can be pushed to the limit of the systems physical resonances to maintain accuracy while providing maximum efficiency. As for DMTS, DMTS-NC is applicable to any neural network potential.

FeNNol: an Efficient and Flexible Library for Building Force-field-enhanced Neural Network Potentials

Neural network interatomic potentials (NNPs) have recently proven to be powerful tools to accurately model complex molecular systems while bypassing the high numerical cost of ab-initio molecular dynamics simulations. In recent years, numerous advances in model architectures as well as the development of hybrid models combining machine-learning (ML) with more traditional, physically-motivated, force-field interactions have considerably increased the design space of ML potentials. In this paper, we present FeNNol, a new library for building, training and running force-field-enhanced neural network potentials. It provides a flexible and modular system for building hybrid models, allowing to easily combine state-of-the-art embeddings with ML-parameterized physical interaction terms without the need for explicit programming. Furthermore, FeNNol leverages the automatic differentiation and just-in-time compilation features of the Jax Python library to enable fast evaluation of NNPs, shrinking the performance gap between ML potentials and standard force-fields. This is demonstrated with the popular ANI-2x model reaching simulation speeds nearly on par with the AMOEBA polarizable force-field on commodity GPUs (GPU=Graphics processing unit). We hope that FeNNol will facilitate the development and application of new hybrid NNP architectures for a wide range of molecular simulation problems.

Generalized Many-Body Dispersion Correction through Random-phase Approximation for Chemically Accurate Density Functional Theory

We extend our recently proposed Deep Learning-aided many-body dispersion (DNN-MBD) model to quadrupole polarizability (Q) terms using a generalized Random Phase Approximation (RPA) formalism enabling to include van der Waals contributions beyond dipole. The resulting DNN-MBDQ model only relies on ab initio-derived quantities as the introduced quadrupole polarizabilities are recursively retrieved from dipole ones, in turn modelled via the Tkatchenko-Scheffler method. A transferable and efficient deep-neuronal network (DNN) provides atom in molecule volumes, while a single range-separation parameter is used to couple the model to Density Functional Theory (DFT). Since it can be computed at negligible cost, the DNN-MBDQ approach can be coupled with DFT functionals such as as PBE/PBE0 or B86bPBE(dispersionless). DNN-MBQ-PBE/PBE0 reaches chemical accuracy exhibiting superior accuracy compared to other dispersion-corrected models, especially at near-equilibrium ranges where errors are lowered by nearly 25% compared to our dipole-only approach while gains reach nearly 50% compared to other corrected schemes.