Under- and over-independence in measure preserving systems (Q2303693)

For a measure-preserving system \((X,\mathcal{B},\mu,T)\) on a standard probability space, fundamental properties like ergodicity, weak-mixing, and mixing can all be formulated as convergence properties of correlation sequences \((\mu(A\cap T^{-n}A))\) for \(A\in\mathcal{B}\) with \(\mu(A)>0\). In attempting to further refine the Cesàro sequential convergence, the second author [Lond. Math. Soc. Lect. Note Ser. 228, 1--61 (1996; Zbl 0846.05095)] asked the following question: does any such (invertible) system have a set \(A\in\mathcal{B}\) with \(\mu(A)>0\) such that \(\mu(A\cap T^nA)<\mu(A)^2\) for all \(n\ge1\), or a set \(A\in\mathcal{B}\) with \(\mu(A)>0\) such that \(\mu(A\cap T^nA)>\mu(A)^2\) for all \(n\ge1\)? In the terminology adopted here, are there always sets exhibiting under- or over-independence? \textit{M. Boshernitzan} et al. [Isr. J. Math. 222, No. 2, 815--840 (2017; Zbl 1379.37005)] showed that there are mixing systems with no under-independent sets, that ergodic systems of positive entropy have under-independent sets, and that there are mixing systems with both under- and over-independent sets. The first author here showed that any mixing system has an over-independent set. Here questions of a similar sort are studied and extended in multiple directions: to higher-order (i.e., multiple) correlations along polynomial times; to densities of times along which under- or over-independence occurs; to over- and under-independence along moving Cesàro averages; and to over- and under-independence for Cesàro averages infinitely often.