Divergence of weighted square averages in \(L^1\) (Q2237362)

The authors study the convergence and divergence of ergodic averages with polynomials weights along squares. More precisely, they study the set of reals \(\theta\) such that for every ergodic system \((X,\mu,T)\) there is a function \(f \in L^1\) for which the weighted average \[ \frac{1}{N} \sum_{n=1}^N e(P(n) \theta) T^{n^2} f \] is divergent on a set of positive measure. Here \(P\) is a fixed polynomial with integer coefficients. In the language of ergodic theory, the authors study those \(\theta\) for which the pair of sequences \((e(P(n) \theta))_{n \geq 1}\) and \((n^2)_{n \geq 1}\) is \(L^1\)-universally bad. The authors show that the set of such \(\theta\) is residual and contains rationals as well as a dense set of Liouville numbers. These results should be compared with the corresponding convergence results for the case of \(L^p,~p>1,\) in [\textit{J. Bourgain}, Isr. J. Math. 61, No. 1, 73--84 (1988; Zbl 0642.28011); Isr. J. Math. 61, No. 1, 39--72 (1988; Zbl 0642.28010); \textit{T. Eisner} and \textit{B. Krause}, Ergodic Theory Dyn. Syst. 36, No. 7, 2172--2202 (2016; Zbl 1375.37015)] and divergence results for \(L^1\) in the unweighted case in [\textit{Z. Buczolich} and \textit{R. D. Mauldin}, Ann. Math. (2) 171, No. 3, 1479--1530 (2010; Zbl 1200.37003)].