Local behavior of traces of Besov functions: prevalent results (Q1937458)

The paper deals with regularity properties of traces of functions belonging to some Besov or Sobolev spaces. The authors show that in the particular genericity meaning of prevalence, the traces can be more regular than what standard theorems of traces give. They also compute the singularity spectrum of the traces. To obtain their results, they use wavelet techniques. One of the main problems the authors had to handle was to get the universal measurability of their ``key sets'' involved within their proofs. The main tool to get prevalence was to construct a Borel measure based on a suitable probe. Let us give a short description of the structure of the paper. The main results are Theorem 1.1 and Theorem 1.6 (see Section 1): (1.1) Let \(D\in \mathbb{N}^*\), \(0<p,s<\infty\), with \(s-D/p >0\), and let \(q\in (0,\infty]\). If \(f\in B^s_{p,\infty}(\mathbb{R}^D)\), then, for Lebesgue-almost all \(a\in \mathbb{R}^{D-d},\;f_a\in\bigcup_{s'<s}B^{s'}_{p,\infty}(\mathbb{R}^d)\). Moreover, if \(f\in B^s_{p,q}(\mathbb{R}^D)\) and \(q<p\) (resp., \(q=p\)), almost every trace \(f_a\) belongs to \(B^s_{p,qp/(p-q)}(\mathbb{R}^d)\) (resp., \(B^s_{p,\infty}(\mathbb{R}^d))\). (1.6) Let \(0<p<\infty\), \(0<q\leq \infty\) and \(0<s-d/p<\infty\). For almost all \(f\) in \(B^s_{p,q}(\mathbb{R}^D)\), for Lebesgue-almost all \(a\in \mathbb{R}^{D-d}\), the following holds: (i) The spectrum of singularities of \(f_a\) is \[ d_{f_a}(h)=\left\{\begin{matrix} d+p(h-s)\;\text{ if }\;h\in[s-d/p,s],\\ -\infty\;\text{otherwise}.\end{matrix}\right. \] (ii) For every open set \(\Omega\subset \mathbb{R}^d\), the set \(\left\{x_0\in\Omega\;:\;h_{f_a}(x_0)=s\right\}\) has full Lebesgue measure in \(\Omega\). Section 5 is devoted to the proof of Theorem 1.1. Section 2 (Preliminaries) provides definitions, tools and results needed to get Theorem 1.6, in fact, its ``localisation version'' Theorem 2.2. In Section 3, the main theorem (1.6) (in fact, its ``localised'' version 2.2) is obtained, using an ancillary result (Theorem 3.2, proved in Section 4). Section 6 is devoted to the extension of the results to submanifolds of \(\mathbb{R}^D\) instead of subspaces.