Ergodic theorems for the shift action and pointwise versions of the Abért-Weiss theorem (Q2303689)

Let \(\Gamma\) be a countably infinite group with identity element \(1\) and \(\alpha: \Gamma \curvearrowright(X, \mu)\) be a probability measure-preserving action. Given \(f \in L^{1}(X, \mu),\) the global average is \[ \mathbb{E}_{\mu} f:=\int_{X} f \mathrm{d} \mu, \] and the pointwise averages are \[ \mathbb{E}_{D} f(x):=\frac{1}{|D|} \sum_{\delta \in D} f(\delta \cdot x), \] where \(x \in X\) and \(D\) is a nonempty finite subset of \(\Gamma\). Theorem 2.1 gives a pointwise ergodic theorem for continuous maps \(f\) on the Bernoulli shift action \(\Gamma \curvearrowright\left([0 ; 1]^{\Gamma}, \lambda^{\Gamma}\right)\). Instead of considering the problem of when the limits of those two averages will meet as \(|D| \rightarrow \infty\), the author considers a sequence \(\left(D_{n}\right)_{n \in \mathbb{N}}\) of finite subsets of \(\Gamma\) such that \(\left|D_{n}\right| / \log n \rightarrow \infty .\) This result leads to the mean ergodic theorem for the shift, as a corollary. Theorem 2.1 follows by a concentration of measure inequality together with the Borel-Cantelli lemma. Note that the Lovász Local Lemma, a useful tool in probabilistic combinatorics, plays an important role in this paper. The author also reviews it and its measurable analogues in Section 6 and goes through one of its proofs in order to prove some results. Theorem 2.6 is a push-forward ergodic theorem for the shift. Based on the definition of residually finite and on this theorem, the author comes up with a new definition called approximately residually finite. As a strengthening of the theorem of \textit{M. Abért} and \textit{B. Weiss} [Ergodic Theory Dyn. Syst. 33, No. 2, 323--333 (2013; Zbl 1268.37006)], Theorem 2.11 gives a pointwise (Abért-Weiss) theorem. The last result, Theorem 2.14, is valid for uniformly subexponential actions. As the author mentions, it is a purely Borel version of the Abért-Weiss theorem for finitely generated groups of subexponential growth.