Fully generic sequences and a multiple-term return-times theorem (Q1385169): Difference between revisions

Let \((X,{\mathcal F},\mu, T)\) be a dynamical systems consisting of a measure-preserving transformation defined on a standard probability space. A bounded sequence \(\{a_i\}\) of complex numbers are called universal weights (for the pointwise ergodic theorem) if for any dynamical system \((Y,{\mathcal G},\nu, S)\) and \(g\in L^1(\nu)\), \({1\over N} \sum^{N-1}_{j= 0}a_ig(S^j y)\) converges. The following theorem was proved by \textit{J. Bourgain}, \textit{H. Furstenberg}, \textit{Y. Katznelson} and \textit{D. S. Ornstein} [Publ. Math., Inst. Haut. Étud. Sci. 69, 5-45 (1989; Zbl 0705.28008)] (and a joinings proof was given by \textit{D. J. Rudolph} [Ergodic Theory Dyn. Syst. 14, No. 1, 197-203 (1994; Zbl 0799.28010)]): For any dynamical system \((X,{\mathcal F},\mu,T)\) and \(f\in L^\infty(\mu)\) for \(\mu\) a.e. \(x\), the sequence of values \(f(T^ix)\) are universal weights. The author proves a multi-term version of the above result first proposed by I. Assani, i.e., involving averages of the form \[ {1\over N} \sum^{N-1}_{i= 0} f_1(T^i_1 x_1)f_2(T^i_2 x_2)\cdots f_k(T^i_k x_k), \] where all the \(f_i\) are bounded and where for each \(j<k\), the points \(x_1,x_2,\dots, x_j\) guaranteeing convergence can be chosen universally, without knowledge of the transformations and functions to follow. The author is able to avoid the \(L^2\) orthogonality arguments in the proof of Bourgain et al., which requires the splitting of the \(L^2\) space into Kronecker functions and their orthocomplement, by using the notion of pointwise genericity. In particular, a disjointness result on joinings avoids the need to identify distinguished factor algebras for the higher term averages. A disadvantage of this approach is that delicate information about characteristic factors is not available (as in, for example, the work of \textit{D. J. Rudolph} [Lond. Math. Soc. Lect. Note Ser. 205, 369-432 (1995; Zbl 0877.28012)], concerning the Conze-Lesigne algebra).