English

Key Observable for Linear Thermalization

Statistical Mechanics 2023-08-10 v3

Abstract

For studies on thermalization of an isolated quantum many-body system, the fundamental issue is to determine whether a given system thermalizes or not. However, most studies tested only a small number of observables, and it was unclear whether other observables thermalize. Here, we study whether `linear thermalization' occurs for all additive observables: We consider a quantum many-body system prepared in an equilibrium state and its unitary time evolution induced by a small change Δf\Delta f of a physical parameter ff of the Hamiltonian, and examine whether \emph{all} additive observables relax to the equilibrium values in a manner fully consistent with thermodynamics up to the linear order in Δf\Delta f. We find that the additive observable conjugate to ff is key for linear thermalization in that its linear thermalization guarantees, under physically reasonable conditions, linear thermalization of all additive observables. Such a linear thermalization occurs in the timescale of O(Δf0)\mathcal{O}(|\Delta f|^0), and lasts at least for a period of o(1/Δf)o(1/\sqrt{|\Delta f|}). We also consider linear thermalization against the change of other parameters, and find that linear thermalization of the key observable against Δf\Delta f guarantees its linear thermalization against small changes of any other parameters. Furthermore, we discuss the generalized susceptibilities for cross responses and their consistency between quantum mechanics and thermodynamics. We demonstrate our main result by performing numerical calculations for spin models. The present paper offers an efficient way of judging linear thermalization because it guarantees that examination of the single key observable is sufficient.

Keywords

Cite

@article{arxiv.2303.03868,
  title  = {Key Observable for Linear Thermalization},
  author = {Yuuya Chiba and Akira Shimizu},
  journal= {arXiv preprint arXiv:2303.03868},
  year   = {2023}
}

Comments

28 pages, 12 figures and 1 table

R2 v1 2026-06-28T09:05:27.575Z