Table of Contents
Fetching ...

Entanglement membrane in the Brownian SYK chain

Márk Mezei, Harshit Rajgadia

Abstract

There is mounting evidence that entanglement dynamics in chaotic many-body quantum systems in the limit of large subsystems and long times is described by an entanglement membrane effective theory. In this paper, we derive the membrane description in a solvable chaotic large-$N$ model, the Brownian SYK chain. This model has a collective field description in terms of fermion bilinears connecting different folds of the multifold Schwinger-Keldysh path integral used to compute Rényi entropies. The entanglement membrane is a traveling wave solution of the saddle point equations governing these collective fields. The entanglement membrane is characterised by a velocity $v$ and a membrane tension ${\cal E}(v)$ that we calculate. We find that the membrane has finite width for $v<v_B$ (the butterfly velocity), however for $v > v_B$, the membrane splits into two wave fronts, each moving with the butterfly velocity. Our results provide a new viewpoint on the entanglement membrane and uncover new connections between quantum information dynamics and scrambling.

Entanglement membrane in the Brownian SYK chain

Abstract

There is mounting evidence that entanglement dynamics in chaotic many-body quantum systems in the limit of large subsystems and long times is described by an entanglement membrane effective theory. In this paper, we derive the membrane description in a solvable chaotic large- model, the Brownian SYK chain. This model has a collective field description in terms of fermion bilinears connecting different folds of the multifold Schwinger-Keldysh path integral used to compute Rényi entropies. The entanglement membrane is a traveling wave solution of the saddle point equations governing these collective fields. The entanglement membrane is characterised by a velocity and a membrane tension that we calculate. We find that the membrane has finite width for (the butterfly velocity), however for , the membrane splits into two wave fronts, each moving with the butterfly velocity. Our results provide a new viewpoint on the entanglement membrane and uncover new connections between quantum information dynamics and scrambling.

Paper Structure

This paper contains 41 sections, 252 equations, 20 figures.

Figures (20)

  • Figure 1: The $n$-th Rényi entropy of sites to the left of site $x$ at $t = T$ is related to a membrane of velocity $v$ that intersects the chain at $t = 0$ at site $x- vT$. The velocity of the membrane is determined by the minimization in equation \ref{['eq:entanglement_growth_conjecture']}.
  • Figure 2: A maximally entangled state in $\mathcal{H}_R \otimes \mathcal{H}_L$. The dashed lines illustrate maximal entanglement between sites in $R$ and $L$ at $T = 0$. $S^{(n)}(x,y,T)$ is the entanglement entropy of the region $A \cup B \cup \bar{A}$. (Note that the bar stands for the subsystem in the $L$ chain and not for complement.) We illustrate the entanglement membrane by a black line of slope $v$ connecting $x$ and $y$.
  • Figure 3: Left: Path integral contour for preparation of the state $\rho(T)_{A \cup B \cup \bar{A}}$. The solid arcs at the bottom of the contour illustrate maximally entangled states between the subregions of the chains $R$ and $L$. The grey boxes illustrate the joint unitary evolution of fermions in chain $R$. The open legs on the left and right denote the ket and bra states of the density matrix $\rho(T)_{A \cup B \cup \bar{A}}$ respectively. Right: We consider two copies of $\rho(T)_{A \cup B \cup \bar{A}}$ and cyclically contract all the open legs in $A \cup B \cup \bar{A}$. This defines the Schwinger-Keldysh contour for $\text{Tr}\,\rho^2(T)_{A \cup B \cup \bar{A}}$.
  • Figure 4: Left: Contour for the second Rényi entropy ignoring $A$. Right: The OTOC contour. The second Rényi contour resembles the OTOC contour when $A$ is ignored.
  • Figure 5: Saddle point solution for $v >v_B$. The solution consists of two fronts: (i) originating from the domain wall $A|B$ at $t = 0$, (ii) originating from $B|C$ at $t = T$. Both fronts propagate with the butterfly velocity towards region $B$, one forward and one backward in time as indicated by the arrows.
  • ...and 15 more figures