Connectivity in fractal percolation (Q1200250)

Fractal percolation refers to one of the simplest examples of a statistically self similar set \(A\). Let \(0<p<1\), and choose a base \(b\). The set \(A\subset \mathbb{R}^ 2\) is an intersection of sets \(A_ n\) which are defined recursively. The set \(A_{n+1}\) is obtained by deleting independently with probability \(1-p\) the \(b^ 2\) \((n+1)^{th}\) order subsquares in each \(n^{th}\) order subsquare of \(A_ n\). The recursion starts from \(A_ 0=[0,1]^ 2\). It has been shown that for \(p\) close enough to 1, the set \(A\) ``percolates'', which has been defined as \(A\) contains with positive probability a connected set which intersects opposite sides of the unit square. The author proves here that this definition is probabilistically equivalent to the more natural definition where one requires \(A\) to contain an arc connecting opposite sides of \([0,1]^ 2\). He also proves that the number of connected components larger than one point of a percolating \(A\) is infinite but countable a.s., and that such an \(A\) contains a.s. uncountably many isolated points.