Bootstrap percolation in high dimensions

TL;DR

This paper analyzes the critical probability for bootstrap percolation on high-dimensional grids, providing bounds and sharp thresholds for the case when r=2, especially as the dimension grows large.

Contribution

It establishes precise bounds and a sharp threshold for the critical probability in high-dimensional bootstrap percolation when r=2, extending understanding to arbitrary functions n and d with d \, log n.

Findings

Derived bounds for p_c([2]^d,2) involving the root of a specific equation

Established sharp thresholds for p_c([n]^d,2) for functions n(d) with d \, log n

Extended previous results to high dimensions and arbitrary growth functions

Abstract

In r-neighbour bootstrap percolation on a graph G, a set of initially infected vertices A \subset V(G) is chosen independently at random, with density p, and new vertices are subsequently infected if they have at least r infected neighbours. The set A is said to percolate if eventually all vertices are infected. Our aim is to understand this process on the grid, [n]^d, for arbitrary functions n = n(t), d = d(t) and r = r(t), as t -> infinity. The main question is to determine the critical probability p_c([n]^d,r) at which percolation becomes likely, and to give bounds on the size of the critical window. In this paper we study this problem when r = 2, for all functions n and d satisfying d \gg log n. The bootstrap process has been extensively studied on [n]^d when d is a fixed constant and 2 \leq r \leq d, and in these cases p_c([n]^d,r) has recently been determined up to a factor of 1 + o(1) as n -> infinity. At the other end of the scale, Balogh and Bollobas determined p_c([2]^d,2) up to a constant factor, and Balogh, Bollobas and Morris determined p_c([n]^d,d) asymptotically if d > (log log n)^{2+\eps}, and gave much sharper bounds for the hypercube. Here we prove the following result: let \lambda be the smallest positive root of the equation \sum_{k=0}^\infty (-1)^k \lambda^k / (2^{k^2-k} k!) = 0, so \lambda \approx 1.166. Then (16\lambda / d^2) (1 + (log d / \sqrt{d})) 2^{-2\sqrt{d}} < p_c([2]^d,2) < (16\lambda / d^2) (1 + (5(log d)^2 / \sqrt{d})) 2^{-2\sqrt{d}} if d is sufficiently large, and moreover we determine a sharp threshold for the critical probability p_c([n]^d,2) for every function n = n(d) with d \gg log n.