The bad and rough rotation is Poissonian

TL;DR

This paper proves that sequences derived from rough numbers multiplied by badly approximable irrationals exhibit Poissonian correlations and gaps, providing the first explicit example of such behavior and disproving a related conjecture.

Contribution

It introduces explicit sequences from rough numbers with Poissonian correlations, advancing understanding of fine-scale distribution properties in number theory.

Findings

Sequences from rough numbers have Poissonian correlations for badly approximable irrationals.

The result does not hold for almost every real number, disproving a conjecture.

Method involves equidistribution in diophantine Bohr sets, which may be of independent interest.

Abstract

Motivated by the Berry-Tabor Conjecture and the seminal work of Rudnick-Sarnak, the fine-scale properties of sequences $(a_n\alpha)_{n \in \mathbb{N}} \mod 1$ with $(a_n)_{n \in \mathbb{N}} \subseteq \mathbb{N} $ and $\alpha$ irrational have been extensively studied in the last decades. In this article, we prove that for $(a_n)_{n \in \mathbb{N}}$ arising from the set of rough numbers with explicit roughness parameters and any badly approximable $\alpha$, $(a_n\alpha)_{n \in \mathbb{N}} \mod 1$ has Poissonian correlations of all orders, and consequently, Poissonian gaps. This is the first known explicit sequence $(a_n\alpha)_{n \in \mathbb{N}} \mod 1$ with these properties. Further, we show that this result is false for Lebesgue almost every $\alpha$, thereby disproving a conjecture of Larcher and Stockinger [Math. Proc. Camb. Phil. Soc. 2020]. The method of proof makes use of an equidistribution result mod $d$ in diophantine Bohr sets which might be of independent interest.