A Neural Network Potential with Rigorous Treatment of Long-Range Dispersion

Abstract

Neural Network Potentials (NNPs) have quickly emerged as powerful computational methods for modeling large chemical systems with the accuracy of quantum mechanical methods but at a much smaller computational cost. To make the training and evaluation of the underlying neural networks practical, these methods commonly cutoff interatomic interactions at a modest range (e.g., 5~\AA), so longer-range interactions like London dispersion are neglected. This limits the accuracy of these models for intermolecular interactions. In this work, we develop a new NNP designed for modeling chemical systems were dispersion is an essential component. This new NNP is extended to treat dispersion interactions rigorously by calculating atomic dispersion coefficients through a second NN, which is trained to reproduce the coefficients from the quantum-mechanically derived exchange-hole dipole moment (XDM) model. Calculation of the dispersion component of intermolecular interactions through this scheme provides results in very good agreement with the QM data, with a mean absolute error (MAE) of 0.6 kcal/mol and a coefficient of determination (R2) of 0.98. The dispersion components of these intermolecular interactions are predicted in excellent agreement with the QM data, with a mean absolute error (MAE) of 0.02 kcal/mol and an R2 of 1.00. This combined dispersion-corrected NNP, called ANIPBE0-MLXDM, predicts intermolecular interaction energies for complexes from the DE370K test set with an MAE of 0.5 kcal/mol and an R2 of 0.94 relative to high-level ab initio results (CCSD(T)/CBS), but with a computational cost that is billions of times smaller. The ANIPBE0-MLXDM method is effective for simulating large-scale dispersion-driven systems, like gas adsorption in porous materials, molecular liquids, and nanostructures, on a single computer workstation.

Exploring the frontiers of condensed-phase chemistry with a general reactive machine learning potential

“…Certain applications may require more direct treatment of long-range effects. Future work could investigate incorporating recent developments, such as explicit long-range terms 63 , charge equilibration schemes 64 or graph NN models 5,6,[13][14][15][16] that can implicitly account for long-range interactions. A recent advancement in ML for natural language processing is the concept of foundational models, that is, large, general models usually trained with unlabelled data that can be specialized to specific tasks quickly with very small amounts of data 65 .…”

Scalable hybrid deep neural networks/polarizable potentials biomolecular simulations including long-range effects

“…Such performances should continue to improve thanks to further Deep-HP optimizations, TorchANI updates and GPUs hardware evolutions. Deep-HP will enable the implementation of the next generation of improved MLPs [84][85][86] and has been designed to be a place for their further development. It will include direct neural networks coupling with physics-driven contributions going beyond multipolar electrostatics and polarization through inclusion of many-body dispersion models.…”

Exploring the frontiers of condensed-phase chemistry with a general reactive machine learning potential

“…A more optimal solution would be to add an explicit dispersion correction to ANI-1xnr that captures long-range interactions while maintaining an accurate description of the local environment. 45 We also compare the ANI-1xnr lattice constants with those from ANI-2x, 46 a model explicitly trained to small organic molecules as a baseline. ANI-2x performs poorly at predicting the lattice constants for both diamond cubic and graphite.…”