Stability of approximate group actions: uniform and probabilistic (Q6057872)

The authors are motivated by a classical question asked by Ulam in 1940, called \textit{Ulam's stability problem}, which asks if an approximate homomorphism from a group \(\Gamma \) to a group \(G\) is close to a homomorphism. Here, of course, the answer depends on the groups \(\Gamma , G\) as well as on our notions of approximate homomorphism and proximity of functions. In the case when \(\Gamma \) is amenable and \(G = U(\mathcal{H})\) is the group of unitary operators on a Hilbert space \(\mathcal{H}\), \textit{D. Kazhdan} [Isr. J. Math. 43, 315--323 (1982; Zbl 0518.22008)] has given a positive answer to this question as follows: for small enough positive \(\delta \) (namely, for \(0 < \delta < \frac{1}{200}\)) and for any function \(f:\Gamma \to U(\mathcal{H})\) satisfying \(||f(xy)-f(x)f(y)|| < \delta \) for all \(x,y\in \Gamma \), there exists a homomorphism \(h: \Gamma \to U(\mathcal{H})\) such that \(||f(\gamma )-h(\gamma )|| < 2\delta \) for every \(\gamma \in \Gamma \). Here, \(||.||\) denotes the operator norm. Kazhdan's result encourages to study the question for various other versions of \(G\) while \(\Gamma \) is still an amenable group (either a general amenable group or from a more restrictive subclass). When \(G = \mathrm{Sym}(n)\), \textit{L. Glebsky} and \textit{L. M. Rivera} [Taiwanese J. Math. 13, No. 2A, 493--500 (2009; Zbl 1203.20003)] studied this problem for finite groups and obtained the following result: Theorem: Let \(\Gamma \) be a finite group and \(f:\Gamma \to \mathrm{Sym}(n)\) be a function, \(n\in \mathbb{N}\). Then there is a group homomorphism \(h:\Gamma \to \mathrm{Sym}(n)\) such that \(d^H(h(\gamma ),f(\gamma ))\leq C\mathrm{def}_{\infty }(f)\) for all \(\gamma \in \Gamma \) where \(C\) depends on the group \(\Gamma \), but not on \(n\). Here, \(d^H\) is the \textit{normalized Hamming metric} and \(\mathrm{def}_{\infty }(f)\) is the so-called \textit{uniform local defect} of \(f\). The first main result in the paper (Theorem 1.1.) shows that the above theorem of Glebsky and Rivera does not hold in the case when \(\Gamma \) is an infinite cyclic group. This answers a question asked by Lubotzky in 2018. Indeed, the result (Theorem 1.1) is stated and proved not just for the group \(\mathbb{Z}\), but for any group which acts transitively on the set \([n]\). The second main result (Theorem 1.2) is in the positive direction. It is motivated (guided) by the results by \textit{W. T. Gowers} and \textit{O. Hatami} [Sb. Math. 208, No. 12, 1784--1817 (2017; Zbl 1427.20022); translation from Mat. Sb. 208, No. 12, 70--106 (2017)] and its generalization by \textit{M. De Chiffre} et al. [Mathematika 65, No. 1, 98--118 (2019; Zbl 1416.22008)] where the problem is studied in the case when \(\Gamma \) is a discrete amenable group and \(G\) is the unitary group \(U(n), n\in \mathbb{N}\). In Theorem 1.2, the authors prove that when \(\Gamma \) is a discrete amenable group and \(f:\Gamma \to \mathrm{Sym}(n)\) is any function, then for all natural \(N\) in the interval \([n, (1+2018\mathrm{def}_{\infty }(f))n]\) there exists a homomorphism \(h:\Gamma \to \mathrm{Sym}(N)\) such that \(d_{\infty }(h,f)\leq 2039\mathrm{def}_{\infty }(f)\) with \(d_{\infty }(h,f)\) being the so-called \textit{uniform distance} (from \(h\) to \(f\)). This result answers a question by \textit{G. Kun} and \textit{A. Thom} [``Inapproximability of actions and Kazhdan's property (T)'', Preprint, \url{arXiv:1901.03963}] (as well as a slight variation of Lubotzky's original question) in the affirmative.