Entanglement flow in the Kane-Fisher quantum impurity problem

TL;DR

This paper investigates entanglement entropy differences in the Kane-Fisher quantum impurity problem, revealing persistent effects and resonance behaviors linked to interactions, using numerical DMRG and analytical conformal perturbation theory.

Contribution

It introduces a novel entanglement-based perspective on the Kane-Fisher impurity problem, connecting entanglement entropy differences to impurity fixed points and resonance phenomena.

Findings

The entropy difference δS remains finite in the thermodynamic limit.

Resonance curves interpolate between -ln 2 and 0 depending on interactions.

δS can be analyzed via conformal perturbation theory near fixed points.

Abstract

The problem of a local impurity in a Luttinger liquid, just like the anisotropic Kondo problem (of which it is technically a cousin), describes many different physical systems. As shown by Kane and Fisher, the presence of interactions profoundly modifies the physics familiar from Fermi liquid theory, and leads to non-intuitive features, best described in the Renormalization Group language (RG), such as flows towards healed or split fixed points. While this problem has been studied for many years using more traditional condensed matter approaches, it remains somewhat mysterious from the point of view of entanglement, both for technical and conceptual reasons. We propose and explore in this paper a new way to think of this important aspect. We use the realization of the Kane Fisher universality class provided by an XXZ spin chain with a modified bond strength between two sites and explore the difference of (Von Neumann) entanglement entropies of a region of length $\ell$ with the rest of the system - to which it relates to a modified bond - in the cases when $\ell$ is even and odd. Surprisingly, we find out that this difference $\delta S\equiv S^e-S^o$ remains of $O(1)$ in the thermodynamic limit, and gives rise now, depending on the sign of the interactions, to "resonance" curves, interpolating between $-\ln 2$ and $0$, and depending on the product $\ell T_B$, where $1/T_B$ is a characteristic length scale akin to the Kondo length in Kondo problems. $\delta S$ can be interpreted as a measure of the hybridization of the left-over spin in odd length subsystems with the "bath" constituted by the rest of the chain. The problem is studied both numerically using DMRG and analytically near the healed and split fixed points. Interestingly - and in contrast with what happens in other impurity problems - $\delta S$ can, at least to lowest order, be tackled by conformal perturbation theory.