English

Resonance-induced growth of number entropy in strongly disordered systems

Disordered Systems and Neural Networks 2022-06-06 v3

Abstract

We study the growth of the number entropy SNS_N in one-dimensional number-conserving interacting systems with strong disorder, which are believed to display many-body localization. Recently a slow and small growth of SNS_N has been numerically reported, which, if holding at asymptotically long times in the thermodynamic limit, would imply ergodicity and therefore the absence of true localization. By numerically studying SNS_N in the disordered isotropic Heisenberg model we first reconfirm that, indeed, there is a small growth of SNS_N. However, we show that such growth is fully compatible with localization. To be specific, using a simple model that accounts for expected rare resonances we can analytically predict several main features of numerically obtained SNS_N: trivial initial growth at short times, a slow power-law growth at intermediate times, and the scaling of the saturation value of SNS_N with the disorder strength. Because resonances crucially depend on individual disorder realizations, the growth of SNS_N also heavily varies depending on the initial state, and therefore SNS_N and von Neumann entropy can behave rather differently.

Keywords

Cite

@article{arxiv.2112.12987,
  title  = {Resonance-induced growth of number entropy in strongly disordered systems},
  author = {Roopayan Ghosh and Marko Žnidarič},
  journal= {arXiv preprint arXiv:2112.12987},
  year   = {2022}
}

Comments

19 pages, 20 figures, version published in Phys. Rev. B