Cutoff for polymer pinning dynamics in the repulsive phase

TL;DR

This paper analyzes the mixing time of a polymer model's Glauber dynamics in the repulsive phase, revealing an abrupt cutoff at a specific time scale for certain interaction strengths.

Contribution

It provides new precise results on the cutoff phenomenon for the polymer's Glauber dynamics when the wall interaction parameter is less than 2.

Findings

For λ ≤ 1, the total variation distance drops sharply at (L^2 log L)/π^2.

For λ in (1,2), cutoff occurs from extremal initial conditions at the same time.

Results improve previous bounds on mixing times for this model.

Abstract

We consider the Glauber dynamics for model of polymer interacting with a substrate or wall. The state space is the set of one-dimensional nearest-neighbor paths on $\mathbb{Z}$ with nonnegative integer coordinates, starting at $0$ and coming back to $0$ after $L$ ($L\in 2\mathbb{N}$) steps and the Gibbs weight of a path $\xi=(\xi_x)^{L}_{x=0}$ is given by $\lambda^{\mathcal{N}(\xi)}$, where $\lambda \geq 0$ is a parameter which models the intensity of the interaction with the substrate and $\mathcal{N}(\xi)$ is the number of zeros in $\xi$. The dynamics we consider proceeds by updating $\xi_x$ with rate one for each $x=1,\dots, L-1$, in a heat-bath fashion. This model was introduced in [CMT08] with the aim of studying the relaxation to equilibrium of the system. We present new results concerning the total variation mixing time for this dynamics when $\lambda< 2$, which corresponds to the phase where the effects of the wall's entropic repulsion dominates. For $\lambda \in [0, 1]$, we prove that the total variation distance to equilibrium drops abruptly from $1$ to $0$ at time $(L^2 \log L)(1+o(1))/\pi^2$. For $\lambda \in (1,2)$, we prove that the system also exhibit cutoff at time $(L^2 \log L)(1+o(1))/\pi^2$ when considering mixing time from "extremal conditions" (that is, either the highest or lowest initial configuration for the natural order on the set of paths). Our results improves both previously proved upper and lower bounds in [CMT08].