Skip to content
Preprint

Evaluating Predicted Densities, Hamiltonians, and Density Matrices as Periodic SCF Initializers

Sep 2026 · 0 citations · 23 references
Physics

Abstract

Learned electronic states are usually evaluated by prediction error, even though their intended use is to accelerate the self-consistent-field (SCF) loop of density functional theory (DFT). We ask whether lower offline error actually yields a better SCF initializer. We construct $\rho$HD-43K, a 43,851-crystal DFT corpus with aligned charge-density, Hamiltonian, and density-matrix labels, and extend the solver workflow to inject all three predicted states. The closed-loop benchmark compares direct converged-state prediction against residual prediction from solver-native references on a frozen test set of non-magnetic crystals, using paired convergence, iteration, and SCF-loop timing measurements. Injecting the exact converged density matrix or Hamiltonian as an oracle upper bound cuts the median SCF count from 16 to one, revealing substantial acceleration headroom. Among learned inputs, direct charge-density prediction (Charge3Net-E3) accelerates about 91% of paired crystals, saves a median of three SCF iterations, and yields a $1.18\times$ SCF-loop speedup. In contrast, the tested residual-density and matrix initializers do not consistently improve over the standard superposition-of-atomic-densities baseline. These results establish that target-space accuracy alone is insufficient: a learned initializer must also be compatible with the nonlinear solver trajectory, and its utility must be measured in the loop.

View source

Similar papers

#artificial intelligence Preprint Oct 2026

Equivariant Flow Matching for Electron Density Prediction

OrbFlow is developed, an $\mathrm{SE}(3)$-equivariant generative model that predicts Gaussian-type orbital (GTO) coefficients via flow matching that retains the efficiency of a compact atom-centered basis while replacing pointwise regression with a learned probability path over the full coefficient space.

Chen-Xing Liang, Cheng-Dong Wang, Yu-Ching Lin et al. · 0 citations
Open access Sep 2026

Structural benchmarks overlook density-of-states errors in machine-learned interatomic potentials

Universal machine-learned interatomic potentials are widely used as fast surrogates for density functional theory in structural modeling. Standard benchmarks emphasize energies, forces, and relaxed geometries. Whether the resulting structures also preserve the density of states remains largely unexplored. We examine th...

Shao-Yu Tseng, Guang-Shuai Han, Tian-Hao Li et al. · 0 citations
Open access Oct 2026

Accelerating SCF Convergence Through Equivariant Density-Matrix Learning and Analytic Refinement.

We present dm-PhiSNet, a physically constrained PhiSNet-based equivariant model that predicts one-electron reduced density matrices (1-RDMs) directly from molecular geometries in an atomic-orbital (AO) basis to accelerate self-consistent-field (SCF) convergence. Training follows a two-stage schedule with progressively...

Zuriel Y. Yescas-Ramos, Andrés Álvarez-García, Huziel E. Sauceda · 0 citations
Oct 2026

Stable and transferable exchange-correlation potentials via machine-learned density-independent mapping.

We present a machine-learning framework, density-independent mapping for exchange-correlation, for incorporating XC effects into density functional theory via a density-independent mapping from the external potential to the XC potential. By predicting the ground-state XC potential in a one-shot manner and holding it fi...

Min Chen, Yu Xie, M. Pavanello et al. · 0 citations
Preprint Sep 2026

Orbital-Free Surrogate Functionals Yield Transferable Interatomic Potentials and Electron Densities

Orbital-free density functional theory seeks to compute the energy of an electronic system directly from its electron density, avoiding one-electron wave functions and thereby offering a route to scalable electronic structure calculations. Machine-learned orbital-free density functionals have recently achieved promisin...

Simon Wagner, Marc K. Ickler, M. V. Klockow et al. · 2 citations
#machine learning Preprint Oct 2026

Physics-Aligned Electronic Ground-State Learning Improves Generalization

Machine-learned interatomic potentials (MLIPs) excel at in-distribution tasks, accelerating drug and material development, yet they struggle to generalize out-of-distribution. We propose to push the cost-accuracy Pareto frontier by designing observable-agnostic electronic ground-state descriptor models (GSMs) with comp...

Eike S. Eberhard, Xaver Kainz, Viktor Kotsev et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.