Characterizing \(W^{2,p}\) submanifolds by \(p\)-integrability of global curvatures (Q2376328)

The authors show sufficient and necessary (geometric) conditions for a \(C^1\) compact \(m\)-manifold immersed in a Euclidean space \(\Sigma\subset\mathbb{R}^n\), to be an embedded \(C^{1,\tau}\cap W^{2,p}\)-submanifold, where \(p>m\) and \(\tau=1-m/p\). In fact, these conditions apply to a more general family of topological spaces, called \(m\)-fine compact sets. Thus, specialists in geometric measure theory and potential theory should be interested in this paper. The whole work relies on a background of deep results. The first section presents definitions and states sufficient and necessary conditions. First of all, an \(m\)-fine compact set is a set satisfying two conditions: (i) Ahlfors regularity, i.e., the \(m\)-Hausdorff measure of \(\Sigma\) in balls of small radius \(r\) is bounded from below by \(C r^m\), where \(C\) is a constant which only depends on \(\Sigma\); (ii) ``hole-control'' in small scales, i.e., that the bilateral beta numbers \(\theta_\Sigma(x,r)\) are upper controlled by the beta numbers \(\beta_\Sigma(x,r)\) -- these numbers give a notion of what is going on geometrically in small scales \(r>0\) around \(x\in\Sigma\). Let \(\operatorname{conv}(x_0,\dotsc,x_m)\) be the convex hull of \(x_0,\dotsc,x_m\in\mathbb{R}^n\), i.e., the affine simplex with these vertices. For an \(m\)-fine set \(\Sigma\), the global Menger curvature at a point \(x_0\in\Sigma\) is defined as \[ \mathcal{K}_G(x_0)=\sup_{x_1,\dotsc,x_m+1\in\Sigma}\frac{\mathcal{H}^{m+1}(\operatorname{conv}(x_0,x_1,\dotsc,x_{m+1}))}{\max |x_i-x_j|}, \] where \(\mathcal{H}^{m+1}\) is the \((m+1)\)-Hausdorff measure in \(\mathbb{R}^n\). Let \(R_{\mathrm{tp}}:\Sigma\times\Sigma\times G(n,m)\to[0,\infty]\) be the tangent-point radius defined as \[ R_{\mathrm{tp}}(x,y;H)=\frac{|x-y|}{2\operatorname{dist}(y,x + H)}. \] Assuming that \(H:\Sigma\to G(n,m)\) is given, the global tangent-point curvature is defined as \[ \mathcal{K}_{\mathrm{tp}}(x,H)=\sup_{y\in\Sigma}\frac{1}{R_{\mathrm{tp}}(x,y;H(x))}. \] In the case of a \(C^1\)-manifold the obvious choice for \(H\) is the tangent space at the point \(x\). The energy of these curvatures, being finite, leads to the desired conditions. For an \(m\)-fine compact set \(\Sigma\subset \mathbb{R}^n\) and \(p>m\), the following conditions are equivalent: (i) \(\Sigma\) is an embedded \(C^{1,\tau}\cap W^{2,p}\)-manifold without boundary; (ii) \(\mathcal{K}_G\in L^p(\Sigma,\mathcal{H}^m)\); (iii) there is a map \(H:\Sigma\to G(n,m)\) such that \(\mathcal{K}_{tp}(\cdot, H)\in L^p(\Sigma,\mathcal{H}^m)\). The second section is devoted to some geometrical aspects relative to the Grassmanian space \(G(n,m)\), this is by far the simplest section of the paper. This section ends with two key theorems. One of them, due to \textit{P. Hajłasz} [Potential Anal. 5, No. 4, 403--415 (1996; Zbl 0859.46022)] shows that \(f\in L^p(\Omega)\) belongs to \(W^{1,p}\) if and only if there exists \(g\in L^p(\Omega)\) such that \(|f(x)-f(y)|\leq|x-y|(g(x)+g(y))\). This is the tool to improve the class of the Sobolev space. The other one, due to \textit{G. David} et al. [Commun. Pure Appl. Math. 54, No. 4, 385--449 (2001; Zbl 1031.28004)] gives a condition on \(m\)-fine compact sets to be \(C^{1,\tau}\) submanifolds without boundary. This condition will be satisfied when conditions (ii) or (iii) in the main theorem are provided. The third section is the core of the paper where the implications (ii) \(\Rightarrow\) (i) and (iii) \(\Rightarrow\) (i) are shown. The work in this section is highly non-trivial. A lot of estimates are needed in order to have control on all the defined quantities by constants depending only on \(m\) and \(p\), in fact, a deep analysis of Morrey estimates for the energies of Menger and tangent-point curvatures is done. The final section gives the easier reverse implications. Here, the weight of the arguments relay in the expression of beta numbers for \(W^{2,p}\)-graphs. Notice, that the definition of \(W^{2,p}\)-manifolds is given for the first time at this stage (\(\Sigma\) will be \(W^{2,p}\) if it locally is the graph of a \(W^{2,p}\)-function). Remark that \(W^{2,p}\) implies \(C^1\) by the Sobolev embedding theorem. A significative number of estimates come from previous works (see, e.g., [the second and the third author, Math. Z. 257, 107--130 (2007; Zbl 1354.49028); \textit{S. Blatt} and the first author, Adv. Math. 230, No. 3, 839--852 (2012; Zbl 1246.53005)]). An exhaustive list of references can be found in the paper.