A restricted shift completeness problem (Q715618)

From MaRDI portal
Revision as of 10:03, 30 January 2024 by Import240129110113 (talk | contribs) (Added link to MaRDI item.)
scientific article
Language Label Description Also known as
English
A restricted shift completeness problem
scientific article

    Statements

    A restricted shift completeness problem (English)
    0 references
    0 references
    0 references
    0 references
    31 October 2012
    0 references
    The authors identify the orthogonal complement of a set of functions obtained from shifts of a single generator \(f\) in \(L^2([0,1])\) with support contained in \([0,a]\), \(0<a<1\). This result solves a problem posed by Carlsson and Sundberg which puts additional restrictions on \(f\) and specialized to \(a=1/2\) [\textit{M. Carlsson} and \textit{C. Sundberg}, ``On translations of a function and its Fourier transoform; orthogonal decompositions of \(L^2([0,1])\)'' (work in progress)]. The main result is as follows: Let \(0<a<1\) and \(f \in L^2([0,1])\) be such that the convex hull of the support of \(f\) is \([0,a]\). Consider the roots \(L=\{\lambda_k\}\) of the analytic continuation of the Fourier transform of \(f\), and let \(n_k\) be the multiplicity of the root \(\lambda_k\), then the orthogonal complement of \(\{\tau_t f\}_{0 \leq t \leq 1 -a}\), where \(\tau_t f(x) = f(x-t)\), is the closed linear span of the set \[ \{ x \mapsto x^s e^{- i \overline{\lambda_k} x }: \lambda_k \in L,~ 0 \leq s <n_k \} \, . \] It is a standard exercise in the theory of reproducing kernels to verify that the elements in this set are indeed orthogonal to the shifts of \(f\), because the inner product with one of these kernel functions yields the point evaluation of the Fourier transform, or of a derivative of the Fourier transform at a (multiple) root. What is shown here is that the two closed subspaces spanned by the shifts of \(f\) and by the kernel functions indexed by the roots of the Fourier transform of \(f\) and by their multiplicity form a direct sum decomposition of the whole space \(L^2([0,1])\). The proof uses inner-outer factorization, the Hilbert transform, Beurling density theorems, and a recent result by \textit{M. Mitkovski} and \textit{A. Poltoratski} who characterized Pólya sequences in terms of long sequences of intervals [Adv. Math. 224, No. 3, 1057--1070 (2010; Zbl 1204.30018)].
    0 references
    exponentials systems
    0 references
    completeness
    0 references
    Fourier transform
    0 references
    Pólya sets
    0 references

    Identifiers