Spaces of variable smoothness and integrability: characterizations by local means and ball means of differences (Q692620)

From MaRDI portal
Revision as of 17:07, 18 April 2024 by Importer (talk | contribs) (‎Changed an Item)
scientific article
Language Label Description Also known as
English
Spaces of variable smoothness and integrability: characterizations by local means and ball means of differences
scientific article

    Statements

    Spaces of variable smoothness and integrability: characterizations by local means and ball means of differences (English)
    0 references
    0 references
    0 references
    6 December 2012
    0 references
    The well-known scales of Besov and Triebel-Lizorkin spaces \(B^s_{p,q}\) and \(F^s_{p,q}\) on \(\mathbb{R}^n\), with \(s\in\mathbb{R}\) and \(p,q\in(0,\infty]\) (\(p<\infty\) for the \(F\)-spaces), can be defined through the quasinorms \[ \|f\|_{B^s_{p,q}} = \left\| \left( 2^{\nu s}(\varphi_\nu\, \hat{f})^\vee \right)_{\nu\in\mathbb{N}_0}\right\|_{\ell_q(L_p)} \quad \text{and} \quad \|f\|_{F^s_{p,q}} = \left\| \left( 2^{\nu s}(\varphi_\nu\, \hat{f})^\vee \right)_{\nu\in\mathbb{N}_0}\right\|_{L_p(\ell_q)}, \] where \(\{\varphi_\nu\}_{\nu\in\mathbb{N}_0}\) is a dyadic resolution of unity and the symbols \(\,\hat{}\,\) and \(\,^\vee\,\) denote the Fourier transform and its inverse, respectively. As is known, different such systems \(\{\varphi_\nu\}_{\nu\in\mathbb{N}_0}\) produce the same spaces up to equivalent quasinorms. In the present paper, the authors consider \(2\)-microlocal spaces \(B^{\mathbf{w}}_{p(\cdot),q(\cdot)}\) and \(F^{\mathbf{w}}_{p(\cdot),q(\cdot)}\) of variable integrability, which are defined as above by replacing the power sequence \((2^{\nu s})_{\nu\in\mathbb{N}_0}\) by a more general admissible weight sequence \(\mathbf{w}=(w_\nu)_{\nu\in\mathbb{N}_0}\) (which may depend also on the space variable) and the sequence spaces \(\ell_q(L_p)\) and \(L_p(\ell_q)\) by the mixed spaces \(\ell_{q(\cdot)}(L_{p(\cdot)})\) and \(L_{p(\cdot)}(\ell_{q(\cdot)})\), respectively. Even for constant exponents \(p\) and \(q\), the \(2\)-microlocal spaces can be seen as a slight generalization of variable smoothness spaces and spaces of generalized smoothness. The mixed sequence spaces \(\ell_{q(\cdot)}(L_{p(\cdot)})\) and \(L_{p(\cdot)}(\ell_{q(\cdot)})\) were recently introduced by \textit{A. Almeida} and \textit{P. Hästö} [J. Funct. Anal. 258, No. 5, 1628--1655 (2010; Zbl 1194.46045)] and \textit{L. Diening}, \textit{P. Hästö} and \textit{S. Roudenko} [J. Funct. Anal. 256, No. 6, 1731--1768 (2009; Zbl 1179.46028)] to define Besov and Triebel-Lizorkin scales \(B^{s(\cdot)}_{p(\cdot),q(\cdot)}\) and \(F^{s(\cdot)}_{p(\cdot),q(\cdot)}\) with variable smoothness and integrability. Admitting all the three indices variable provides the natural framework to study traces and Sobolev embeddings since an interaction between the parameters is required in this situation. The main aim of this paper is to characterize the general spaces \(B^{\mathbf{w}}_{p(\cdot),q(\cdot)}\) and \(F^{\mathbf{w}}_{p(\cdot),q(\cdot)}\). The first main result is a local means characterization using Peetre maximal functions. The second main result provides a characterization of these spaces in terms of ball means of differences. Both statements are given under the usual regularity conditions on the integrability exponents, typically the \(\log\)-Hölder continuity.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    Besov spaces
    0 references
    Triebel-Lizorkin spaces
    0 references
    2-microlocal spaces
    0 references
    variable exponents
    0 references
    variable smoothness
    0 references
    variable integrability
    0 references
    local means
    0 references
    Peetre maximal functions
    0 references
    differences
    0 references
    0 references
    0 references