Concerning the stability of exponential systems and Fourier matrices

Abstract

Fourier matrices naturally appear in many applications and their stability is closely tied to performance guarantees of algorithms. The starting point of this article is a result that characterizes properties of an exponential system on a union of cubes in $\mathbb{R}^d$ in terms of a general class of Fourier matrices and their extreme singular values. This relationship is flexible in the sense that it holds for any dimension $d$, for many types of exponential systems (Riesz bases, Riesz sequences, or frames) and for Fourier matrices with an arbitrary number of rows and columns. From there, we prove new stability results for Fourier matrices by exploiting this connection and using powerful stability theorems for exponential systems. This paper provides a systematic exploration of this connection and suggests some natural open questions.