Pointwise convergence of some multiple ergodic averages (Q1644982)

The convergence of multiple averages of the form \[ {1\over N} \sum^{N-1}_{n=0} f_1(T^n_1 x)f_2(T^n_2 x)\cdots f_d(T^n_d x),\tag{1} \] where \((X,{\mathcal X},\mu)\) is a probability space and \(T_1,\dots, T_d\) are measure-preserving transformations on \(X\), has been studied through different approaches leading to the most general result for the \(L^2\) convergence of multiple ergodic averages by \textit{M. N. Walsh} [Ann. Math. (2) 175, No. 3, 1667--1688 (2012; Zbl 1248.37008)] under the assumption of nilpotency of the group generated by \(T_1,\dots, T_d\). Furthermore the pointwise convergence of (1) has been studied by \textit{W. Huang} et al. [J. Differ. Equations 263, No. 1, 779--810 (2017; Zbl 1373.46063)], through the usage of topological models for \(T_i=T^i\), \(i=1,2,\dots,d\) under the assumption that \((X,\mu,T)\) is distal. The aim of the present paper is to investigate the pointwise convergence of multiple averages and to obtain partial results. The authors develop new methods which combine the existence of sated and magic extensions with ideas on topological models and manage to prove a pointwise limit for an easier multiple average. In particular they find out that for every ergodic system \((X,\mu,\, T_1,\dots, T_d)\) with commuting transformations, the average \[ {1\over N^{d+1}} \sum_{0\leq n_1,\dots, n_d\leq N-1}\;\sum_{0\leq n\leq N-1} f_1\Biggl(T^n_1 \prod^d_{j=1} T^{n_j}_j x\Biggr) f_2\Biggl(T^n_2 \prod^d_{j=1} T^{n_j}_j x\Biggr)\cdots f_d\Biggl(T^n_d \prod^d_{j=1} T^{n_j}_j x\Biggr) \] converges for \(\mu\)-a.e. \(x\in X\) as \(N\to\propto\). Moreover they deduce an expression for the \(L^2\)-limit of a multiple average which implies that if \((X,\mu,\,T_1,\dots, T_d)\) is a distal ergodic system with commuting transformations the average \[ {1\over N} \sum^{N-1}_{n=0} f_1(T^n_1 x) f_2(T^n_2x)\cdots f_d(T^n_d x) \] converges for \(\mu\)-a.e. \(x\in X\) as \(N\to\propto\).