Absence of a stationary distribution for the edge process of subcritical oriented percolation in two dimensions (Q1103280)

The main result of this paper is as follows: Consider a percolation structure occurring on \(L=\{(m,n):\) \(m+n\) is even, \(n\geq 0\}\), in which each oriented arc from (m,n) to \((m+1\), \(n+1)\) or (m-1, \(n+1)\) is open with probability p. Let \(L_ n=\{(m,n):\) \(m+n\) is even, n fixed\(\}\) and \(\mu\) be any distribution on non-positives in 2\({\mathbb{Z}}\) which has unbounded support. If \(\mu\) is the distribution of percolation points on \(L_ 0\), then: for subcritical percolation, the distribution of \(\xi_ n^{\mu}\), the points of \(L_ n\) reached by the percolation process, can not be stationary, i.e., \(\mu\) can not be the distribution of all \(\xi_ n^{\mu}\), \(n=0,1,2...\), when the \(\xi_ n^{\mu}\) are viewed from their rightmost point. The reductio ad absurdum proof leads to an inequality and a Borel- Cantelli argument based the following observation: if d is large enough and n small enough, the events \(r=\sup \xi_ n^{\mu}\) and \((r-d,n)\in \xi_ n^{2Z}\) are independent, since they depend on different parts of the percolation structure.