A stationary rho-mixing Markov chain which is not ``interlaced'' rho-mixing (Q5952039)

Let \(X= \{X_k, k\in \mathbb{Z}\}\) be a sequence of random variables defined on a probability space \((\Omega,{\mathcal F},P)\), \[ \rho(X,n)= \sup_{j\in\mathbb{Z}} \rho(\sigma(X_k, k\leq j), \sigma(X_k, k\geq j+ n)), \] \[ \rho^*(X, n)= \sup\rho(\sigma(X_k, k\in S), \sigma(X_k, k\in T)), \] where this latter supremum is taken over all pairs of nonempty, disjoint sets \(S\) and \(T\subset\mathbb{Z}\) such that \(\inf_{s\in S,t\in T}|s-t|\geq n\) and for any two \(\sigma\)-fields \({\mathcal A}\) and \({\mathcal B}\subset{\mathcal F}\), \(\rho({\mathcal A},{\mathcal B})= \sup|\text{Corr}(f,g)|\), where the supremum is taken over all pairs of square-integrable random variables \(f\) and \(g\) such that \(f\) is \({\mathcal A}\)-measurable and \(g\) is \({\mathcal B}\)-measurable. It is proved that for any sequence of positive numbers \(\{c_n, n\geq 1\}\) there exists a strictly stationary, countable-state Markov chain \(X= \{X_k, k\in \mathbb{Z}\}\) such that \(\rho(X,n)\leq c_n\) for all \(n\geq 1\) and \(\rho^*(X,n)\equiv 1\).