scientific article; zbMATH DE number 1455218

Authors Garth I. Gaudry, Brian Jefferies, Werner J. Ricker

Publication date 8 August 2001

Brenner's theorem Calderon-Zygmund theory Coste's theorem Fourier-Stieltjes transforms multipliers spectral measures vector-valued convolution operators vector-valued singular integral operators

Mathematics Subject Classification ID

Multipliers for harmonic analysis in several variables (42B15) Measures on groups and semigroups, etc. (43A05) (L^p)-spaces and other function spaces on groups, semigroups, etc. (43A15) Summability and bases; functional analytic aspects of frames in Banach and Hilbert spaces (46B15) Vector-valued measures and integration (46G10)

Review text

Several of the basic results of Calderon-Zygmund theory have been worked out in the setting of vector-valued singular integral operators but many basic vector-valued analogues remain open problems, in part, because some basic notions have not been worked out satisfactorily in the vector-valued setting. The current memoir represents an attempt to lay down the foundations of vector-valued convolution operators in a manner consistent with the goal of extending basic results of operator theory from the setting of real-valued functions to the setting of functions taking values in a separable Hilbert space \(H\). To put the appropriate emphasis on convolution the authors work in the setting of a locally compact Abelian group \(G\) (though as noted, some results extend to non-Abelian cases). The main results include characterizations of operator-valued Fourier-Stieltjes transforms and of strong \(L^{1}\)-multipliers, and a characterization in terms of absolutely continuous measures of those convolution operators that can be represented by integration against an operator-valued function (Coste's theorem). Of particular interest is a collection of negative results that illustrate precisely how \(L^{p}\) multiplier results can break down in the vector-valued case when \(p\neq 2\). First, because the operator-valued measure in question is not, generally, self-adjoint, it is not guaranteed that a Fourier \(p\)-multiplier is also a Fourier \(q\)-multiplier, \(\frac{1}{p}+\frac{1}{q}=1\). The authors also show that convolution with a spectral measure is rarely bounded on \(L^{p}\) in the vector-valued case when \(p\neq 2\). As a consequence one can show that the group of shifts of an infinite group gives rise to a Fourier \(p\)-multiplier, where \(H=L^{2}(G)\), if and only if \(p=2\). In a different direction, Brenner's theorem states that given an \(n\)-tuple \( A=(A_{1},\dots,A_{n})\) of self-adjoint operators on a finite dimensional Hilbert space \(H\), \(\Phi :\) \(x\mapsto e^{-ix\cdot A}\) is an \(\mathcal{L}(H)\)-valued \(p\)-multiplier for some and hence all \(p\neq 2\) if and only if the operators \(A_{i}\) pairwise commute. The authors show that, in the case \(H\) is infinite dimensional, that commutativity of the \(A_{i}\) is insufficient to guarantee that \(\Phi \) defines a \(p\)-multiplier for some \(p\neq 2\). The first part of this memoir sets the stage by sorting out just what are the appropriate notions of measurability, variation and regularity of a measure, almost everywhere existence and so forth in the case of operator-valued measures. For \(1\leq p\leq \infty \) one then calls a linear operator \(T\) from \(L^{p}(G,H)\) to itself a Fourier \(p\)-multiplier operator if there is a measurable operator-valued function \(\Phi :\Gamma \rightarrow L(H)\) that is essentially bounded in operator norm and such that, for each \( f\in L^{p}\cap L^{2}(G,H)\) one has (Tf)\((\gamma)=\Phi (\gamma) \widehat{f}(\gamma)\) a.e. on \(\Gamma \), the dual group of \(G\).

Recommendations

Cited in

(8)

This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4484935)