English

Quantum chaos challenges many-body localization

Strongly Correlated Electrons 2021-01-01 v4 Disordered Systems and Neural Networks Statistical Mechanics High Energy Physics - Theory Quantum Physics

Abstract

Characterizing states of matter through the lens of their ergodic properties is a fascinating new direction of research. In the quantum realm, the many-body localization (MBL) was proposed to be the paradigmatic ergodicity breaking phenomenon, which extends the concept of Anderson localization to interacting systems. At the same time, random matrix theory has established a powerful framework for characterizing the onset of quantum chaos and ergodicity (or the absence thereof) in quantum many-body systems. Here we numerically study the spectral statistics of disordered interacting spin chains, which represent prototype models expected to exhibit MBL. We study the ergodicity indicator g=log10(tH/tTh)g=\log_{10}(t_{\rm H}/t_{\rm Th}), which is defined through the ratio of two characteristic many-body time scales, the Thouless time tTht_{\rm Th} and the Heisenberg time tHt_{\rm H}, and hence resembles the logarithm of the dimensionless conductance introduced in the context of Anderson localization. We argue that the ergodicity breaking transition in interacting spin chains occurs when both time scales are of the same order, tThtHt_{\rm Th} \approx t_{\rm H}, and gg becomes a system-size independent constant. Hence, the ergodicity breaking transition in many-body systems carries certain analogies with the Anderson localization transition. Intriguingly, using a Berezinskii-Kosterlitz-Thouless correlation length we observe a scaling solution of gg across the transition, which allows for detection of the crossing point in finite systems. We discuss the observation that scaled results in finite systems by increasing the system size exhibit a flow towards the quantum chaotic regime.

Keywords

Cite

@article{arxiv.1905.06345,
  title  = {Quantum chaos challenges many-body localization},
  author = {J. Šuntajs and J. Bonča and T. Prosen and L. Vidmar},
  journal= {arXiv preprint arXiv:1905.06345},
  year   = {2021}
}
R2 v1 2026-06-23T09:07:47.849Z