Neural-network quantum states for ultra-cold Fermi gases

• DOI: 10.1038/s42005-024-01613-w
• Publication Date: 2024-05-08T11:01:53Z
• Authors: Jane Kim; Gabriel Pescia; Bryce Fore; Jannes Nys; Giuseppe Carleo; Stefano Gandolfi; Morten Hjorth-Jensen; Alessandro Lovato
• Keywords: Quantum Physics; Nuclear Theory; Physics; QC1-999; Systems; FOS: Physical sciences; Disordered Systems and Neural Networks (cond-mat.dis-nn); Condensed Matter - Disordered Systems and Neural Networks; Astrophysics; 01 natural sciences; QB460-466; Nuclear Theory (nucl-th); Quantum Gases (cond-mat.quant-gas); 0103 physical sciences; Condensed Matter - Quantum Gases; Quantum Physics (quant-ph)

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