Super-resolution by means of Beurling minimal extrapolation

Abstract

Let $M(\mathbb{T}^d)$ be the space of complex bounded Radon measures defined on the $d$-dimensional torus group $(\mathbb{R}/\mathbb{Z})^d=\mathbb{T}^d$, equipped with the total variation norm $\|\cdot\|$; and let $\hat\mu$ denote the Fourier transform of $\mu\in M(\mathbb{T}^d)$. We address the super-resolution problem: For given spectral (Fourier transform) data defined on a finite set $\Lambda\subset\mathbb{Z}^d$, determine if there is a unique $\mu\in M(\mathbb{T}^d)$ of minimal norm for which $\hat\mu$ equals this data on $\Lambda$. Without additional assumptions on $\mu$ and $\Lambda$, our main theorem shows that the solutions to the super-resolution problem, which we call minimal extrapolations, depend crucially on the set $\Gamma\subset\Lambda$, defined in terms of $\mu$ and $\Lambda$. For example, when $\Gamma=0$, the minimal extrapolations are singular measures supported in the zero set of an analytic function, and when $\Gamma\geq 2$, the minimal extrapolations are singular measures supported in the intersection of $\Gamma\choose 2$ hyperplanes. By theory and example, we show that the case $\Gamma=1$ is different from other cases and is deeply connected with the existence of positive minimal extrapolations. This theorem has implications to the possibility and impossibility of uniquely recovering $\mu$ from $\Lambda$. We illustrate how to apply our theory to both directions, by computing pertinent analytical examples. These examples are of interest in both super-resolution and deterministic compressed sensing. Our concept of an admissibility range fundamentally connects Beurling's theory of minimal extrapolation with Candes and Fernandez-Granda's work on super-resolution. This connection is exploited to address situations where current algorithms fail to compute a numerical solution to the super-resolution problem.