Abstract:Ultra-cold Fermi gases exhibit a rich array of quantum mechanical properties, including the transition from a fermionic superfluid Bardeen-Cooper-Schrieffer (BCS) state to a bosonic superfluid Bose-Einstein condensate (BEC). While these properties can be precisely probed experimentally, accurately describing them poses significant theoretical challenges due to strong pairing correlations and the non-perturbative nature of particle interactions. In this work, we introduce a Pfaffian-Jastrow neural-network quantum state featuring a message-passing architecture to efficiently capture pairing and backflow correlations. We benchmark our approach on existing Slater-Jastrow frameworks and state-of-the-art diffusion Monte Carlo methods, demonstrating a performance advantage and the scalability of our scheme. We show that transfer learning stabilizes the training process in the presence of strong, short-ranged interactions, and allows for an effective exploration of the BCS-BEC crossover region. Our findings highlight the potential of neural-network quantum states as a promising strategy for investigating ultra-cold Fermi gases.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to more accurately describe the transition from the Fermi superfluid (Bardeen - Cooper - Schrieffer, BCS) state to the Bose - Einstein Condensate (BEC) state in ultracold Fermi gases. This transition process involves strong pairing correlations and non - perturbative particle - particle interactions, which pose significant challenges to theoretical research. Specifically, the paper introduces a method based on the Pfaffian - Jastrow neural - network quantum state (NQS), which efficiently captures pairing and backflow correlations through a message - passing architecture. This method aims to overcome the limitations of the existing Slater - Jastrow framework and the state - of - the - art Diffusion Monte Carlo (DMC) method in dealing with strong short - range interactions and improve the effectiveness of exploration in the BCS - BEC crossover region. The key innovation of the paper lies in using neural - network quantum states, especially the message - passing neural network that combines Pfaffian and Jastrow factors and backflow transformation, to enhance the ability to capture pairing and backflow correlations. This method not only improves computational efficiency and accuracy but also shows consistent performance across different particle numbers and effective ranges, especially near the unitarity limit, i.e., when the scattering length \(a\) diverges, the properties of the system become particularly complex and interesting. By benchmarking against the existing DMC method, the paper demonstrates the superiority of the proposed method in energy prediction and its ability to more effectively explore the BCS - BEC crossover region. In addition, the paper also explores pairing distribution functions, which provide a quantitative description of the spatial correlations and pairing phenomena between fermions, further validating the effectiveness and accuracy of the new method.

Neural-network quantum states for ultra-cold Fermi gases

Neural-network quantum states for ultra-cold Fermi gases

Competing Pairing States for Ultracold Fermions in Optical Lattices with an Artificial Staggered Magnetic Field

Neural-Network Quantum States: from Condensed Matter to Quantum Computing

Neural Network Quantum States for the Interacting Hofstadter Model with Higher Local Occupations and Long-Range Interactions

Autoregressive neural quantum states of Fermi Hubbard models

Neural network approach to quasiparticle dispersions in doped antiferromagnets

Ab-initio study of interacting fermions at finite temperature with neural canonical transformation

Implementing quantum electrodynamics with ultracold atomic systems

Message-Passing Neural Quantum States for the Homogeneous Electron Gas

Low-depth circuit ansatz for preparing correlated fermionic states on a quantum computer

Machine learning one-dimensional spinless trapped fermionic systems with neural-network quantum states

Entanglement transition in deep neural quantum states

Learning the ground state of a non-stoquastic quantum Hamiltonian in a rugged neural network landscape

Neural Wave Functions for Superfluids

Polynomially efficient quantum enabled variational Monte Carlo for training neural-network quantum states for physico-chemical applications

Efficient solutions of fermionic systems using artificial neural networks

Quantum many-body simulations of the two-dimensional Fermi-Hubbard model in ultracold optical lattices

Neural-network quantum states for many-body physics