The time of bootstrap percolation in two dimensions (Q328784)

In (standard two-dimensional) bootstrap percolation, each vertex of a graph is initially occupied with probability \(p\), and at each step every unoccupied site having at least two neighbours becomes occupied. If a sequence of graphs is formed by a sequence of \(n\) by \(n\) squares, there are functions \(p_c(n)\), critical percolation thresholds, such that for sequences \(p(n)<p_c(n)\) the squares will not get fully occupied and when \(p(n)>p_c(n)\), they will be become fully occupied, both behaviours holding with high probability. The \(p_c(n)\) has been identified before by \textit{A. E. Holroyd} [Probab. Theory Relat. Fields 125, No. 2, 195--224 (2003; Zbl 1042.60065)]. In this paper, the authors consider the ``percolation time'', that is, the time it takes for the whole square to become occupied. They obtain quite sharp results, depending on how much larger than the critical threshold sequences the \(p(n)\) are. When \(p\) is sufficiently larger than \(p_c\), the percolation time is governed by the presence of unoccupied 2 by \(t\) rectangles, where \(t(n)\) is relatively large. When \(p\) is just supercritical, a more complex result is found. In the first regime, a limit value for the percolation time is found within a \(1+o(1)\) factor, in the second regime up to a constant. The results the authors find are strong improvements of what was known before, which were only results for \(p\) close to 1.