Convergence of ergodic averages on lattice random walks (Q1320950)

The author investigates a question of H. Furstenberg which can be given as follows. For \(d\geq 1\), consider the standard random walk on the lattice \(\mathbb{Z}^ d\), i.e., if \(n(j)\in \mathbb{Z}^ d\) in the \(j\)th position, then \(n(j+1)\) takes one of the values \(n(j)\pm e_ 1\), \(n(j)\pm e_ 2,\dots, n(j)\pm e_ d\) with equal probability, where \(e_ j\) stands for the \(j\)th unit vector. Consider a probability space \((\Omega,\mu)\) and \(d\) commuting invertible measure preserving transformtions \(T_ 1, T_ 2, \dots, T_ d\) of \(\Omega\). Given a measurable function \(f\) on \(\Omega\), define \[ A_ k f= {1\over k} \sum_{j=1}^ k T_ 1^{n(j,1)} T_ 2^{n(j,2)} \dots T_ d^{n(j,d)} f, \tag{*} \] where \(n(j)= (n(j,1),\dots, n(j,d))\). The author studies the convergence properties of the averages (*).