Universally \(L^1\)-bad arithmetic sequences (Q542042)

Let \((X,{\mathcal A},\mu,T)\) be a dynamical system. Let \((n_k)_{k \geq 1}\) be a subsequence of \(\mathbb N\), and consider for \(f \in L^1(X)\) the subsequence averages \[ A_N f(x) = \frac{1}{N} \sum_{k=1}^N f(T^{n_k} x). \] A sequence \((n_k)_{k \geq 1}\) is called universally \(L^p\)-good if for every ergodic system \((X,{\mathcal A},\mu,T)\) and every \(f \in L^p(X,\mu)\) the limit \(\lim_{N \to \infty} A_N f(x)\) exists for almost every \(x\in X\). Similarly, \((n_k)_{k \geq 1}\) is called universally \(L^p\)-bad if, for every non-atomic ergodic dynamical system \((X,\mathcal{A},\mu,T)\) there exists an \(f \in L^p(X,\mu)\) such that the sequence \((A_N f(x))_{N \geq 1}\) diverges on a set of positive measure. By classical results of \textit{J. Bourgain} [Publ. Math., Inst. Hautes Étud. Sci. 69, 5--45 (1989; Zbl 0705.28008); Lect. Notes Math. 1317, 204--223 (1988; Zbl 0662.47006)] and \textit{M. Wierdl} [Isr. J. Math. 64, No. 3, 315--336 (1989; Zbl 0695.28007)], any sequence of polynomial values and the sequence of primes are universally \(L^p\)-good for every \(p > 1\). Very few results were known for the case \(p=1\), until very recently a breakthrough was achieved by \textit{Z. Buczolich} and \textit{R. D. Mauldin} [Contemporary Mathematics 430, 41--56 (2007; Zbl 1122.37003); Ann. Math. (2) 171, No. 3, 1479--1530 (2010; Zbl 1200.37003)], who showed that the sequence of squares is universally \(L^1\)-bad. In the paper under review, the author generalizes the result of Buczolich and Mauldin and shows that also the sequence of \(d\)-th powers (for \(d > 1\)) and the sequence of primes are universally \(L^1\)-bad. The proof of these results, which is inspired by the original proof of Buczolich and Mauldin, is complicated, but very pleasant reading.