Table of Contents
Fetching ...

Exponential convergence to equilibrium in supercritical kinetically constrained models at high temperature

Laure Marêché

Abstract

Kinetically constrained models (KCMs) were introduced by physicists to model the liquid-glass transition. They are interacting particle systems on $\mathbb{Z}^d$ in which each element of $\mathbb{Z}^d$ can be in state 0 or 1 and tries to update its state to 0 at rate $q$ and to 1 at rate $1-q$, provided that a constraint is satisfied. In this article, we prove the first non-perturbative result of convergence to equilibrium for KCMs with general constraints: for any KCM in the class termed "supercritical" in dimension 1 and 2, when the initial configuration has product $\mathrm{Bernoulli}(1-q')$ law with $q' \neq q$, the dynamics converges to equilibrium with exponential speed when $q$ is close enough to 1, which corresponds to the high temperature regime.

Exponential convergence to equilibrium in supercritical kinetically constrained models at high temperature

Abstract

Kinetically constrained models (KCMs) were introduced by physicists to model the liquid-glass transition. They are interacting particle systems on in which each element of can be in state 0 or 1 and tries to update its state to 0 at rate and to 1 at rate , provided that a constraint is satisfied. In this article, we prove the first non-perturbative result of convergence to equilibrium for KCMs with general constraints: for any KCM in the class termed "supercritical" in dimension 1 and 2, when the initial configuration has product law with , the dynamics converges to equilibrium with exponential speed when is close enough to 1, which corresponds to the high temperature regime.

Paper Structure

This paper contains 9 sections, 17 theorems, 34 equations, 4 figures.

Key Result

Theorem 3

If $d=1$ or 2, for any supercritical update family $\mathcal{U}$, for any $q' \in ]0,1]$, there exists $q_0=q_0(\mathcal{U},q') \in [0,1[$ such that for any $q \in [q_0,1]$, for any local function $f: \{0,1\}^{\mathds{Z}^d} \mapsto \mathds{R}$, there exist two constants $c=c(\mathcal{U},q')>0$ and $

Figures (4)

  • Figure 1: Illustration of a dual path $\Gamma$ of length $t'$ starting at $(x,t)$ for $d=1$ and $\rho=2$. Each horizontal line represents the timeline of a site of $\mathds{Z}$, the $\times$ representing the clock rings. $\Gamma$ is the thick polygonal line; it starts at $t$ and ends at $t-t'$. It can jump only when there is a clock ring, and never at a distance greater than $\rho=2$.
  • Figure 2: Illustration of proposition \ref{['prop_Bollobas']}. The $\ast$ represent the sites $x_1,\dots,x_m$. If we start with the sites of $R$ at zero and there are successive 0-clock rings at $x_1,\dots,x_m$ while there is no 1-clock ring in $R \cup \{x_1,\dots,x_m\}$, these clock rings will put $x_1,\dots,x_m$ at zero, hence the sites of $a_1u+R$ will be put at zero.
  • Figure 3: The proof of proposition \ref{['prop_Bollobas']}. (a) The mechanism for $d=1$; the $\bullet$ represent zeroes and the $\circ$ represent ones. (b) The shape delimited by the solid line is the droplet of Bollobas_et_al2015; if it is infected in the bootstrap percolation dynamics, the infection can grow to the shape delimited by the dashed line. (c) $R$ contains the original droplet (dashed line), hence if $R$ is infected, the infection can propagate to a bigger droplet (in gray) that contains $a_1u+R$ and is contained in $R \cup (a_1u+R) \cup (2a_1u+R)$.
  • Figure 4: An illustration of proposition \ref{['prop_transfer_zeroes2']} with $n^{y,k} = 3$ and $r_0=1$. The figure at the left represents the bonds of the oriented percolation process $\zeta^{y,k}$; the open bonds are the thick ones, and the path of open bonds allowing $\zeta_{n^{y,k}}^{y,k}(r_0)=1$ is outlined by arrows. The figure at the right represents the consequences on the KCM process; each vertical strip represents the state of $\bigcup_{i \in \mathds{Z}}(y+ia_1u+R)$ at a certain time. If at time $t-(k+3)K$ the rectangle $y+\frac{1-n^{y,k}}{2}a_1u+R = y-a_1u+R$ is at zero (in gray), since the bond $(0,2) \rightarrow (1,3)$ (bond $b_1$) is open, $y-a_1u+R$ is still at zero at time $t-(k+2)K$. Moreover, since $(1,1) \rightarrow (0,2)$ (bond $b_2$) is open and $y-a_1u+R$ is at zero at time $t-(k+2)K$, $a_1u+(y-a_1u+R) = y+R$ is at zero at time $t-(k+1)K$. Finally, since $(0,0) \rightarrow (1,1)$ (bond $b_3$) is open and $y+R$ is at zero at time $t-(k+1)K$, $y+R$ is still at zero at time $t-kK$.

Theorems & Definitions (31)

  • Definition 1
  • Definition 2
  • Theorem 3
  • Remark 4
  • Proposition 5
  • Lemma 6
  • proof : Sketch of proof.
  • Proposition 7
  • Proposition 8
  • Lemma 9
  • ...and 21 more