Does a distinct quasi many-body localized phase exist? A numerical study of a translationally invariant system in the thermodynamic limit

TL;DR

This study investigates whether a distinct quasi many-body localized phase exists in a translationally invariant system by analyzing the dynamics and entanglement properties in a spin ladder model with varying heavy particle hopping.

Contribution

It provides numerical evidence that, unlike true MBL, the system exhibits a crossover regime with unique dephasing and entanglement dynamics, challenging the existence of a distinct quasi-MBL phase.

Findings

Exponential polarization decay depends on heavy particle hopping amplitude J'

Entanglement entropy transitions from constant/logarithmic to power-law growth with J'

No clear regime showing characteristics of a true MBL phase was observed

Abstract

We consider a quench in an infinite spin ladder describing a system with two species of bosons in the limit of strong interactions. If the heavy bosonic species has infinite mass the model becomes a spin chain with quenched binary disorder which shows true Anderson localization (AL) or many-body localization (MBL). For finite hopping amplitude $J'$ of the heavy particles, on the other hand, we find an exponential polarization decay with a relaxation rate which depends monotonically on $J'$. Furthermore, the entanglement entropy changes from a constant (AL) or logarithmic (MBL) scaling in time $t$ for $J'=0$ to a sub-ballistic power-law, $S_{\textrm{ent}}\sim t^\alpha$ with $\alpha< 1$, for finite $J'$. We do not find a distinct regime in time where the dynamics for $J'\neq 0$ shows the characteristics of an MBL phase. Instead, we discover a time regime with distinct dephasing and entanglement times, different both from a localized and a fully ergodic phase.