The Uniform Random Walk on graphs, loop processes and graphings

TL;DR

This paper introduces the Uniform Random Walk (URW) on graphs as a limit of uniform walks, explores its existence, uniqueness, and phase transitions on various graph structures, and connects it to spectral theory and graphings.

Contribution

It defines URW as a limit object, analyzes its properties on different graph classes, and links finite graph behavior to spectral theory of limiting graphings, including phase transition phenomena.

Findings

URW equals MERW on finite graphs

URW exists and is unique in the delocalized phase

Localization of MERW is governed by adjacency norm

Abstract

We define the Uniform Random Walk (URW) on a connected, locally finite graph as the weak limit of the uniform walk of length $n$ starting at a fixed vertex. When the limit exists, it is necessarily Markovian and is independent of the starting point. For a finite graph, URW equals the Maximal Entropy Random Walk (MERW). We investigate the existence and phase transitions of URW for loop perturbed regular graphs and their limits. It turns out that for a sequence of finite graphs, it is the global spectral theory of the limiting graphing that governs the behavior of the finite MERWs. In the delocalized phase, we use a "membrane argument", showing that the principal eigenfunction of an expander graphing is stable under a small diagonal perturbation. This gives us: 1) The existence of URW on leaves; 2) The URW is a unique entropy maximizer; 3) The MERW of a finite graph sequence Benjamini-Schramm converges to the URW of the limiting graphing. In the localized phase, the environment seen by the particle takes the role of a finite stationary measure. We show that for canopy trees, the URW exists, is transient and maximizes entropy. We also show that for large finite graphs where most vertices have a fixed degree, localization of MERW is governed by the adjacency norm.