Trace operators for modulation, \(\alpha\)-modulation and Besov spaces (Q617708): Difference between revisions

The authors investigate the following problem: Given a Banach function space \(X(\mathbb R^n)\) on \(\mathbb R^n\), what are the properties of restricting \(f\in X(\mathbb R^n)\) to the hyperplane \(\{x\in\mathbb R^n:x=(x_1,\dots,x_{n-1},0)\}\)? To this end, the notion of trace of \(f\) is introduced by \(\mathbb T:f(x)\mapsto f(x_1,\dots,x_{n-1},0)\), which allows them to express this question in terms of properties of \(\mathbb T\). The paper deals with the case that \(X(\mathbb R^n)\) is an \(\alpha\)-modulation space \(M^{s,\alpha}_{p,q}(\mathbb R^n)\), where \(\alpha\in[0,1]\), \(p,q\in(0,\infty]\) and \(s\geq 0\). The endpoints \(\alpha=0\) and \(\alpha=1\) correspond to the modulation spaces and Besov spaces, respectively. \(\alpha\)-modulation spaces are examples of decomposition spaces introduced in [\textit{H. G. Feichtinger} and \textit{P. Gröbner}, Math. Nachr. 123, 97--120 (1985; Zbl 0586.46030)]. Trace theorems for modulation spaces were first considered by \textit{H. G. Feichtinger} in his technical report [Modulation spaces on locally compact Abelian groups. University of Vienna (1983), \url{http://univie.ac.at/nuhag-php/bibtex/open_files/120_fe03-1_fei\%2001\%2083.pdf}], which was later published in [Proc. Internat. Conf. on Wavelets and Applications, Chennai, January 2002. New Delhi Allied Publishers. 1--56 (2003), \url{http://www.univie.ac.at/nuhag-php/bibtex/open_files/fe031_modspa03.pdf}], and by \textit{H. Triebel} [Modulation spaces on the Euclidean \(n\)-space. Forschungsergeb., Friedrich-Schiller-Univ. Jena N/83/66 (1983; Zbl 0524.46019)]. Later, many authors considered trace theorems for Besov spaces, the most recent contribution is the paper [\textit{C. Schneider}, Math. Nachr. 284, No. 5--6, 572--586 (2011; Zbl 1219.46036)]. The paper under review contains a novel description of \(\alpha\)-modulation spaces via a new \(\alpha\)-covering of \(\mathbb R^n\). The results involve anisotropic variants of \(\alpha\)-modulation spaces. Due to the generality of the setting, the paper is very technical and a detailed description of the main results is not possible. For \(a,b\) real numbers, we denote by \(a\wedge b\) their minimum. Under some conditions on \(p,q\) and \(\alpha\), it is shown that \(\mathbb T\) is a retract from \(M^{s+\alpha/p,\alpha}_{p,p\wedge q\wedge 1}(\mathbb R^n)\) to \(M^{s,\alpha}_{p,q}(\mathbb R^{n-1})\). The authors remark that for \(\alpha=0\) and \(p,q>1\), their statement is weaker than the one known for modulation spaces and similarly for the Besov space case \(\alpha=1\). Therefore, the presented theorems for \(\alpha\)-modulation spaces are not optimal. It seems that several typos have escaped the attention of the authors and reviewers, potentially causing some confusion. Among others, the words ``retraction'' or ``reaction'' appear, when in fact the concept of ``retract'' is considered.