On the cluster size distribution for percolation on some general graphs (Q986616)

Percolation is perhaps the most widely studied statistical physics model for modeling random media. The first theorem of the paper states that for any Cayley graph, the probability that the cluster of the origin has size \(n\) decays at a well-defined exponential rate. The authors' method combines a randomized version of the usual method using subadditivity (as in for \(\mathbb{Z}^d\)) together with a proof that any two finite subgraphs of \(G\) have disjoint translates that are at distance \(\leq\delta\) from each other, under \(\delta\) is an appropriate function of the sizes of the subgraphs.