Time-frequency partitions and characterizations of modulation spaces with localization operators (Q630792): Difference between revisions

A time--frequency representation transforms a function \(f\) in \(\mathbb{R}^d\) into a function on the time--frequency space \(\mathbb{R}^d \times \mathbb{R}^d\). The goal is to obtain a description of \(f\) that is local both in time and in frequency. However, the pointwise interpretation of such time--frequency representation encounters difficulties because of the Heisenberg Uncertainty Principle. Therefore the question arises in which sense the short--time Fourier transform describes the local properties of a function and its Fourier transform. Following \textit{I. Daubechies} [IEEE Trans. Inf. Theory 34, No.4, 605--612 (1988; Zbl 0672.42007)] the authors use time--frequency operators to give meaning to the time--frequency content. By investigating a whole family of such operators, the authors obtain a new characterization of modulation spaces. This characterization relates the size of the localization operators to the global time--frequency distribution. As a by--product, they obtain a new proof of the existence of multi--window Gabor frames and extend the structure theory of Gabor frames,