A geometric characterization of a class of Poisson type distributions (Q682874)

A previous theorem by \textit{N. Lev} and \textit{A. Olevskii} [Invent. Math. 200, No. 2, 585--606 (2015; Zbl 1402.28002)] states that the measures on \(\mathbb R^n\) which have both their support and spectrum uniformly discrete are supported on a finite union af translates of a single lattice. The main theorem of the present article is a variation of this statement for tempered distributions: Theorem. Let \(u\) be a tempered distribution on \(\mathbb R^n\) with support \(\Lambda\), and let \(\Sigma\) be the support of its Fourier transform. Suppose that both sets \(\Lambda-\Lambda\) and \(\Sigma-\Sigma\) are discrete, and that at least one of them is uniformly discrete. Then \(\Lambda\) is included in a finite union af translates of a single lattice. Note that the hypothesis of the theorem can be rephrased as ``\(\Lambda\) and \(\Sigma\) are uniformly discrete, and one of them is Meyer''. The proof of this theorem consists in applying Lev and Olevskii's theorem to an autocorrelation distribution associated to the tempered distribution \(u\); it appears that this autocorrelation is a measure. A Poisson's sommation formula is also obtained.