Quasisymmetric dimension distortion of Ahlfors regular subsets of a metric space (Q2628928)

Given a mapping between metric spaces each of which is equipped with the Hausdorff measure corresponding to its dimension, one can ask questions about dimension distortion. The first question that probably comes to mind concerns Luzin's condition~(N). That is, we can ask if sets of measure zero are mapped to sets of measure zero. The second conglomerate of questions is how much the mapping can increase a set of a given dimension. These questions have been extensively studied in the case that the metric spaces are ``nice'' and the mappings are Sobolev mappings or quasiconformal mappings. The paper under review investigates questions that belong to a third category. Namely, one can ask whether given the family of sets of a certain dimension~\(k\), the dimension of their images is smaller or equal than~\(k\) for almost every member of the family, with the term ``almost every'' suitably interpreted. The paper under review considers quasisymmetric and quasiconformal mappings. Let me give a better idea of the proven results by showcasing Theorem~1.1. It states that if \(f\) is a quasiconformal mapping on \(\mathbb{R}^N\), \(N\geq 2\), and \(E\subset \mathbb{R}^N\) is a bounded Ahlfors regular set, then \[ \dim_H f(y+E)=\dim_H(E) \] for \(\mathcal{H}^n\)-a.e.\ \(y\in \mathbb{R}^N\). To put this result into context, quasiconformal mappings of \(\mathbb{R}^N\) are absolutely continuous on almost all lines for \(N\geq 2\). This setting corresponds to the statement of the theorem when \(E\) is a fixed line. Note that, identifying \(y\) with the line parallel to \(E\) through~\(y\), almost every can be looked at as a statement concerning an amount of lines. In the theory of quasiconformal mappings and the one of Sobolev spaces on metric measure spaces, the modulus of curve families is often used as an (outer) measure on rectifiable curves. Actually, already in \textit{B. Fuglede}'s paper [Acta Math. 98, 171--219 (1957; Zbl 0079.27703)] moduli of more dimensional objects are studied. The paper under review makes extensive use of this more general notion of modulus. Often, the results are of the form that the dimension of the image of a set is bounded by the dimension of the set (or equals the dimension of the set) unless the set is contained in the family of exceptional sets, which has modulus~\(0\). The sharpness of the results is investigated as well. However, this is done only in the \(2\)-dimensional space. The reason for restricting to the plane lies in the fact that there is an abundance of conformal mappings in the plane, contrasting the situation in higher dimensions. The approach that the authors take is very general and quite abstract. Proving an abstract result first, they then show that the result can be applied to certain specific situations. The constructions of the counterexamples is quite involved and takes several pages. The authors support the reader with some nice figures of their constructions.