Wavelets, tiling, and spectral sets (Q1847940): Difference between revisions

A function \(\phi\) is a wavelet if there exist a set of \(d\times d\) real matrices \(\mathcal D\) and a set \({\mathcal T} \subset {\mathbb R}^d\) such that \(\{|\det (D)|^{1/2} \phi(Dx - \lambda): D \in {\mathcal D}, \lambda \in {\mathcal T}\}\) forms an orthogonal basis for \(L^2({\mathbb R}^d)\). \(\mathcal D\) is the dilation set and \(\mathcal T\) is the translation set. If \(\widehat \phi = \chi_\Omega\) for some \(\Omega \subset {\mathbb R}^d\), then we refer to \(\Omega\) as a wavelet set. A set \(\Omega\) with nonzero and finite measure is called a spectral set if there exists a set \(\mathcal S\) such that \(\{e^{2\pi i s \xi}: s \in {\mathcal S}\}\) is an orthogonal basis for \(L^2(\Omega)\). Then \((\Omega, {\mathcal S})\) is called a spectral pair. The paper under review establishes a tie between wavelet sets and spectral sets. Namely, if \((\Omega, {\mathcal T})\) is a spectral pair and if \(\{D^T(\Omega): D \in {\mathcal D}\}\) tiles \({\mathbb R}^d\), then \(\check \chi_\Omega\) is a wavelet with respect to \(\mathcal D\) and \(\mathcal T\). Conversly, if \(\check \chi_\Omega\) is a wavelet with respect to \(\mathcal D\) and \(\mathcal T\), and if \(0 \in {\mathcal T}\), then \((\Omega, {\mathcal T})\) is a spectral pair and \(\{D^T(\Omega): D \in {\mathcal D}\}\) tiles \({\mathbb R}^d\). Further, the structure of dilation sets of MSF wavelets is studied, and the author provides examples of wavelet sets associated with irregular sets of dilations and translations. In particular, such more general dilation sets (as compared to the typical \({\mathcal D} = \{A^j: j \in {\mathbb Z}\}\)) allow the existence of simple, nonfractal-like, wavelet sets.