Machine Learning Hamiltonians are Accurate Energy-Force Predictors

Recently, machine learning Hamiltonian (MLH) models have gained traction as fast approximations of electronic structures such as orbitals and electron densities, while also enabling direct evaluation of energies and forces from their predictions. However, despite their physical grounding, existing Hamiltonian models are evaluated mainly by reconstruction metrics, leaving it unclear how well they perform as energy–force predictors. We address this gap with a benchmark that computes energies and forces directly from predicted Hamiltonians. Within this framework, we propose QHFlow2, a state-of-the-art Hamiltonian model with an SO(2)-equivariant backbone and a two-stage edge update. QHFlow2 achieves $40\%$ lower Hamiltonian error than the previous best model with fewer parameters. Under direct evaluation on MD17/rMD17, it is the first Hamiltonian model to reach NequIP-level force accuracy while achieving up to $20\times$ lower energy MAE. On QH9, QHFlow2 reduces energy error by up to $20\times$ compared to MACE. Finally, we demonstrate that QHFlow2 exhibits consistent scaling behavior with respect to model capacity and data, and that improvements in Hamiltonian accuracy effectively translate into more accurate energy and force computations.

Predicting how molecules move and react is important for designing new medicines, batteries, and materials, but accurate quantum chemistry calculations can be very slow. Many machine learning models try to speed this up by directly predicting final quantities such as a molecule’s energy or the forces on its atoms. In this work, we take a different approach: we train AI to predict a key part of the molecule’s electronic structure, called the Hamiltonian, which describes how electrons behave in the molecule. Our model, QHFlow2, predicts Hamiltonians more accurately than previous methods. From this prediction, we can compute energies and forces, while also accessing electronic information such as molecular orbitals, electron density, and chemical reactivity trends. This makes the model useful not only for molecular simulation, but also for understanding why a molecule is stable or reactive. Our experiments show that QHFlow2 reaches force accuracy comparable to strong energy-force models, while giving much more accurate energies on several molecular benchmarks. It can also provide better starting points for conventional quantum chemistry software, reducing expensive iterative computation. Overall, our work suggests that teaching AI to predict electronic structure through Hamiltonians can lead to molecular modeling tools that are both accurate and more informative for chemical analysis.