Extremes of the 2d scale-inhomogeneous discrete Gaussian free field: Convergence of the maximum in the regime of weak correlations

TL;DR

This paper studies the maximum of the 2D scale-inhomogeneous discrete Gaussian free field in the weak correlation regime, proving convergence to a randomly shifted Gumbel distribution and analyzing the structure of local maxima.

Contribution

It establishes the convergence in law of the maximum to a shifted Gumbel distribution and characterizes the random shift and the geometry of high maxima.

Findings

Maximum converges to a randomly shifted Gumbel distribution.

The random shift depends on variances at different scales.

High local maxima form finite clusters separated by large distances.

Abstract

We continue the study of the maximum of the scale-inhomogeneous discrete Gaussian free field in dimension two. In this paper, we consider the regime of weak correlations and prove the convergence in law of the centred maximum to a randomly shifted Gumbel distribution. In particular, we obtain limiting expressions for the random shift. As in the case of variable speed branching Brownian motion, the shift is of the form CY, where C is a constant that depends only on the variance at the shortest scales, and Y is a random variable that depends only on the variance at the largest scales. Moreover, we investigate the geometry of highest local maxima. We show that they occur in clusters of finite size that are separated by macroscopic distances. The poofs are based on Gaussian comparison with branching random walks and second moment estimates.