Title: Large Electron Model: A Universal Ground State Predictor

Abstract:

We present the Large Electron Model, a singular neural network framework capable of generating variational wavefunctions for interacting electrons across the complete spectrum of Hamiltonian parameters. This architecture utilizes Fermi Sets, a universal formalism for representing many-body fermionic wavefunctions, which is additionally conditioned on particle count and specific Hamiltonian parameters. In the context of interacting electrons confined within a two-dimensional harmonic potential, our trained model successfully predicts ground-state wavefunctions with high precision. It demonstrates robust generalization capabilities across previously unseen coupling strengths and particle-number regimes, delivering accurate real-space charge densities and ground-state energies for systems containing up to 50 particles. These findings lay the groundwork for a foundational model approach to material discovery rooted in the variational principle, offering a method to accurately handle strong electron correlations that exceed the limitations of density functional theory.

Source: arXiv Generated at: 2026-06-02 00:00:00 UTC