Differentiability of Lipschitz functions in Lebesgue null sets (Q2257728)

\textit{H. Rademacher} [Math. Ann. 79, 340--359 (1920; JFM 47.0243.01)] proved that a Lipschitz map between Euclidean spaces is differentiable almost everywhere. A natural question to ask is to what extent does a converse of this theorem hold. One possible question is: given a set of measure zero \(N\subset \mathbb{R}^n\) is there a Lipschitz map \(f: \mathbb{R}^n\to \mathbb{R}^m\) that is not differentiable at any point in \(N\)? It turns out that the answer depends on the relationship between the dimensions \(m\) and \(n\). \textit{Z. Zahorski} [Bull. Soc. Math. Fr. 74, 147--178 (1946; Zbl 0061.11302)] proved the following stronger result for \(m=n=1\): a set \(N\subset \mathbb{R}\) is the set of points of non differentiability of a Lipschitz function if and only if \(N\) is a \(G_{\delta\sigma}\) set of measure zero (note that each set of measure zero is contained in a \(G_{\delta\sigma}\) set of measure zero). On the other hand, \textit{D. Preiss} proved in [J. Funct. Anal. 91, No. 2, 312--345 (1990; Zbl 0711.46036)] that there is a \(G_{\delta\sigma}\) set of measure zero in the plane such that each real-valued Lipschitz function has a point of differentiability in \(N\). As he remarks in the same paper, an analogous result is true for \(n>m=1\). \textit{G. Alberti} et al. deal with the two-dimensional case in [in: Proceedings of the international congress of mathematicians (ICM 2010), Hyderabad, India, 2010. Vol. III: Invited lectures. Hackensack, NJ: World Scientific; New Delhi: Hindustan Book Agency. 1379--1394 (2011; Zbl 1251.26010)]. Additionally, they provide an ingredient for the proof of announced results by \textit{M. Csörnyei} and \textit{P. Jones} [``Product formulas for measures and applications to analysis and geometry'', Preprint] stating that the converse of Rademacher's theorem is true for \(m\geq n\). The paper under review completes the picture. The authors show that for \(n>1\) and \(m=n-1\) (implying the same result for \(1\leq m\leq n-1\)) there is a null set \(N\subset \mathbb{R}^n\) containing a point of differentiability for every Lipschitz map \(f: \mathbb{R}^n\to \mathbb{R}^m\). One important ingredient in the proof is the porosity of the set of the so-called irregular points; roughly speaking these are the points where we lack uniformity when we speak about differentiability. The crucial property of porous sets (roughly put, sets with large holes on arbitrarily small scales) that is used here is that they have measure zero with respect to certain measures. The proof is inspired by techniques stemming from the setting of (infinite-dimensional) Banach spaces detailed in [\textit{J. Lindenstrauss} et al., Fréchet differentiability of Lipschitz functions and porous sets in Banach spaces. Princeton, NJ: Princeton University Press (2012; Zbl 1241.26001)]. The methods are used for finding a point of differentiability nearby regular points.