Mixing properties of the generalized \(T,T^{-1}\)-process (Q1385426)

Let \((S_n)_{n\in\mathbb N_0}\) and \((S_{-n})_{n\in\mathbb N_0}\) be independent copies of a general random walk on \(\mathbb Z^d\), starting at \(S_0=0\), and let \((C_z)_{z\in\mathbb Z^d}\) be an independent, i.i.d.\ coloring of the sites, taking values \(\pm 1\) with equal probability each. The authors consider the so-called \(T,T^{-1}\)-process \((Z_n)_{n\in\mathbb Z}\) where \(Z_n=(X_n, C_{S_n})\) is the pair consisting of the walker's step \(X_n\) at time \(n\) and the color of its present site. They derive sufficient and necessary conditions for this process to satisfy certain mixing properties in terms of intersection properties of the walk. The main mixing properties considered and explained in the paper are `weak Bernoulli' (WB, also sometimes called \(\beta\)-mixing), `very weak Bernoulli' (VWB) and `Bernoulli' (B; isomorphy to an i.i.d.\ process). For a general stationary process having a countable state space, these properties imply each other in this order, but for a \(T,T^{-1}\)-process, the latter two are shown to be even equivalent. Let \(I\) be the (random) intersection set of the two walks, i.e., the set of sites which are visited by both walks \((S_n)_{n\in\mathbb N_0}\) and \((S_{-n})_{n\in\mathbb N_0}\). One of the main results is that the process \((Z_n)_{n\in\mathbb Z}\) is WB if and only if \(| I| <\infty\) a.s. For a large class of random walks, this is even equivalent to the finiteness of the expectation of \(| I| \). A second main result states that \((Z_n)_{n\in\mathbb Z}\) is VWB if the walk is transient, and that under some mild conditions also the converse holds. These results lead in particular to a complete classification for the case of simple random walk in all dimensions and for all one-dimensional walks whose step distribution lies in the domain of attraction of a stable random variable. The paper is motivated by, and the ideas of the proofs are based on, a paper by \textit{S. A. Kalikow} [Ann. Math., II. Ser. 115, 393-409 (1982; Zbl 0523.28018)] which settled the case of one-dimensional simple random walk.