Artificial neural networks for density-functional optimizations in fermionic systems

Abstract

In this work we propose an artificial neural network functional to the ground-state energy of fermionic interacting particles in homogeneous chains described by the Hubbard model. Our neural network functional was proven to has an excellent performance: it deviates from numerically exact calculations by less than $0.15\%$ for all the regimes of filling factors and magnetizations, and for a vast regime of interactions. When compared to analytical functionals, the neural functional was found to be more precise for all the regimes of parameters, being particularly superior at the weakly interacting regime: where the analytical parametrization fails the most, $\sim 7\%$, against only $\sim 0.1\%$ for the neural functional. We have also applied our homogeneous functional to finite, localized impurities and harmonically confined systems within density-functional theory (DFT) methods. The results show that while our artificial neural network approach is substantially more accurate than other equivalently simple and fast DFT treatments, it has similar performance than more costly DFT calculations and other independent many-body calculations, at a fraction of the computational cost.

8 pages, 5 figures, submitted to Scientific Reports

References in corpus (11)

• Entanglement in spatially inhomogeneous many-fermion systems
• Entanglement entropy and entanglement witnesses in models of strongly interacting low-dimensional fermions
• Density-functional theory for systems with noncollinear spin: orbital-dependent exchange-correlation functionals and their application to the Hubbard dimer
• The Hubbard model as an approximation to the entanglement in nanostructures
• Interacting fermions in 1D disordered lattices: Exploring localization and transport properties with lattice density-functional theories
• Feasibility of approximating spatial and local entanglement in long-range interacting systems using the extended Hubbard model
• Uniqueness of density-to-potential mapping for fermionic lattice systems
• Entanglement and exotic superfluidity in spin-imbalanced lattices
• Testing density-functional approximations on a lattice and the limits of the related Hohenberg-Kohn-type theorem
• Metric-space approach for distinguishing quantum phase transitions in spin-imbalanced systems
• Energy of bond defects in quantum spin chains obtained from local approximations and from exact diagonalization

Cited by in corpus (9)

• Quantum Machine Learning for Chemistry and Physics
• Machine Learning for Condensed Matter Physics
• Machine learning configuration interaction for ab initio potential energy curves
• Scalable neural networks for the efficient learning of disordered quantum systems
• Grad DFT: a software library for machine learning enhanced density functional theory
• Efficient Learning of a One-dimensional Density Functional Theory
• Fast and stable deep-learning predictions of material properties for solid solution alloys
• Density Functional Theory of the Hubbard-Holstein Model
• Supervised learning of few dirty bosons with variable particle number