Cantor sets are not tangent homogeneous (Q2292997)

A subset \(A\subset\mathbb R^N\) is said to be \textit{differentiably homogeneous} if for every pair of points \(x, y\in A\) there exist neighborhoods \(O_x,O_y\) of \(x,y\), a bijection \(f :A\cap O_x\to A\cap O_y\) with \(f(x) = y\), a non-singular linear map \(d:\mathbb R^N\to\mathbb R^N\) and an infinitesimal function \(\alpha:\mathbb R^N\to\mathbb R\) such that, \(\bullet\) \(f(u)=y+d(u-x)+(u-x)\alpha(u-x)\) for each \(u\in A\cap O_x\); \(\bullet\) \(f^{-1}(v)=x+d^{-1}(v-y)+(v-y)\alpha(v-y)\) for each \(v\in A\cap O_y\). The main result (Theorem 5.3) states that no subset of \(\mathbb R^3\) homeomorphic to the Cantor set (e.g. Antoine's necklace) is differentiably homogeneous. The proof is very similar to [\textit{E. V. Shchepin} and \textit{D. Repovš}, J. Math. Sci., New York 100, No. 6, 2716--2726 (1999; Zbl 0977.53003); translation from Itogi Nauki Tekh., Ser. Sovrem. Mat. Prilozh., Temat. Obz. 61, 190--207 (1999)], which is not mentioned (also similarity to [\textit{D. Repovš} et al., Proc. Am. Math. Soc. 124, No. 4, 1219--1226 (1996; Zbl 0863.53004); \textit{A. Skopenkov}, Topology Appl. 154, No. 9, 1894--1897 (2007; Zbl 1118.57025)] is not properly described, see (2) below). So it is not clear to the reviewer what the new ideas are. See also a related paper [\textit{G. della Sala}, ``Curves homogeneous under analytic transformations'', Preprint, \url{arXiv:1602.02525}]. Reviewer's remark (1) In the second paragraph of \S4 the phrase `A cone is \(\ell\)-dense if and only if it contains a line' is, at face value, incorrect. Indeed, take a cone \(C\) to be the union of coordinate axes (e.g. in the plane). Then \(C\) contains a line but it is not 1-dense (because the line \(x_1=\ldots=x_N\) is a 1-dimensional subspace of the convex hull of \(C\) containing no line in common with \(C\).) (Also, this phrase could not be correct because it implies that \(\ell\)-density does not depend on \(\ell\) and is equivalent to a much simpler property `contains a line'.) However, this should be read to mean `A cone whose convex hull has lineability $\ell$ etc.', which is correct; the author's wording is somewhat sloppy here. (2) The difference of the main result (Theorem 5.3) and its proof from [Zbl 0863.53004, Zbl 0977.53003 and Zbl 1118.57025] is subtle but it is not properly described. The following phrase in \S1.1 is not a proper description: `we do not require \(f\) to be defined on an open set in \(\mathbb R^N\), or a local homeomorphism, or differentiable anywhere besides \(x\)'. Indeed, in [Zbl 1118.57025] (and so essentially in [Zbl 0863.53004]) differentiability anywhere besides \(x\) was not used in the proof (although it is present in the definition). It is not written which parts of the proof in [Zbl 1118.57025] (and so essentially in [Zbl 0863.53004]) use that \(f\) is defined on an open set in \(\mathbb R^N\), or is a local homeomorphism.