Higher order Sobolev homeomorphisms and mappings and the Lusin \((N)\) condition (Q1635698)

Luzin's condition (N) for a map \(f: \mathbb{R}^n\to \mathbb{R}^n\) means that sets of \(n\)-dimensional measure zero are mapped to sets of measure zero. This condition is important in physical models as it prevents the creation of mass. Another important application are area formulas. Luzin's condition (N) has been extensively studied for first-order Sobolev mappings. For example, in [Bull. Am. Math. Soc. 79, 790--795 (1973; Zbl 0275.49041)], \textit{M. Marcus} and \textit{V. J. Mizel} proved that a continuous \(f\in W^{1,p}(\mathbb{R}^n,\mathbb{R}^n)\) satisfies Luzin's condition (N) if \(p>n\). The result is not true for \(n=p\) by \textit{J. Malý} and \textit{O. Martio} [J. Reine Angew. Math. 458, 19--36 (1995; Zbl 0812.30007)] (and \textit{L. Cesari} in the case \(n=2\) [Ann. Mat. Pura Appl. (4) 21, 157--188 (1942; Zbl 0028.21004)]); see also [Manuscr. Math. 100, No. 1, 87--101 (1999; Zbl 0976.26004)] by \textit{J. Kauhanen} et al. for an improvement where Lorentz spaces are considered. For Sobolev homeomorphisms, the bound is \(n=p\) (see [\textit{Yu. G. Reshetnyak}, Sib. Math. J. 7, 704--732 (1966; Zbl 0158.32701); translation from Sib. Mat. Zh. 7, 886--919 (1966)] for the positive result and [\textit{S. P. Ponomarev}, Sov. Math., Dokl. 12, 1788--1790 (1971; Zbl 0255.26016); translation from Dokl. Akad. Nauk SSSR 201, 1053--1054 (1971)] for the sharpness). The author considers the case of higher-order Sobolev mappings. For Sobolev homeomorphisms \(f\) from an open subset \(\Omega\) of \(\mathbb{R}^n\) to \(\mathbb{R}^n\), Luzin's condition (N) holds if \(f\in W^{k,p}(\Omega,\mathbb{R}^n)\) for \(p\geq \frac{n}{k}\) and can fail for \(1\leq p<\frac{n}{k}\). Assume now that \(k,m,n\in \mathbb{N}\), \(m\geq n>k\), \(p>\frac{n}{k}\) and that \(\Omega\subset \mathbb{R}^n\) is a domain. If \(f\in W^{k,p}(\Omega,\mathbb{R}^m)\) and \(p>\frac{n}{k}\) or \(1=p=\frac{n}{k}\), then Luzin's condition (N) holds. It can fail if \(1\leq p\leq \frac{n}{k}\) and \(n\not=k\). The focus of the paper is mainly on constructions of examples showing the sharpness of the positive results. This is due to the Sobolev embedding theorem, which, together with the first-order case, gives the results. In the case of a Sobolev mapping in \(W^{n,1}\), the author needs to go to the Lorentz scale as well. The constructions of the homeomorphic counterexamples is done in several steps. First, two Cantor sets are constructed; one with measure zero and the other with positive measure. The map is then constructed as a limit of mappings that map each step in the construction of the first Cantor set to the same step in the construction of the second Cantor set. The construction of the nonhomeomorphic counterexamples follows the steps from Malý and Martio [loc.ċit.]. However, some functions in their construction are replaced.