On the spectra of a Cantor measure (Q1016263)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | On the spectra of a Cantor measure |
scientific article |
Statements
On the spectra of a Cantor measure (English)
0 references
5 May 2009
0 references
The paper deals with spectral measures \(\mu\), i.e., special probability measures on \(\mathbb{R}^d\), for which a set \(\Lambda\subset\mathbb{R}^d\) -- the spectrum -- exists such that \(\{e^{2\pi i \lambda x}\}_{\lambda\in\Lambda}\) is an orthonormal basis for \(L^2(\mu)\). More precisely, the paper essentially concentrates on the measure \(\mu_4\) which is connected with the modified Cantor set \[ X_4 = \left\{ \sum_{k=1}^\infty a_k 4^{-k}: a_k \in \{0,2\}\right\} \] which can also be obtained as the attractor of an appropriate iterated function system. In [\textit{P. E. T. Jørgensen} and \textit{S. Pedersen}, J. Anal. Math. 75, 185--228 (1998; Zbl 0959.28008)], the spectrum \(\Lambda=\{\sum_{k=0}^n d_k 4^k: d_k \in \{0,1\}\), \(n\geq 0\}\) for \(\mu_4\) was already shown. In contrast to the usual (middle-third) Cantor set the analysis of \(\mu_4\) and \(X_4\) is less complete, regarding, for instance, the problem to characterize all spectra or the maximal families of orthogonal exponentials, as well as the question whether such a maximal family must be a basis in \(L^2(\mu)\). This paper aims to answer these questions. The first main result, Theorem 3.3, contains a criterion for a set of mutually orthogonal exponentials to be maximal, involving some invariance of a special spectral labeling of the associate binary tree. A similar condition is sufficient to imply the existence of a spectrum for \(\mu_4\) (Theorem 3.10). In a separate section the authors discuss the situation when the binary tree is labeled with other digits than \(\{0,1,2,3\}\). This leads to counterparts of the above results (Theorem 4.4). The last section is devoted to basic properties of the spectra of \(\mu_4\), illustrated with various examples. In particular, the authors return to their sufficient condition obtained in Section~3 and construct a special labeling which does not satisfy their assumptions, but admits a spectrum.
0 references
Fourier series
0 references
affine fractals
0 references
spectrum
0 references
spectral measure
0 references
Hilbert spaces
0 references
attractor
0 references