The champernowne constant is not Poissonian (Q2422604)

Let \(\Vert\cdot\Vert\) denote the distance to the nearest integer. A sequence \((x_{n})_{n\in N}\) of real numbers in the unit interval \([0,1[\) has Poissonian pair correlations if \[ \lim(1/N)\mathrm{Card}\{1\leq k\neq m\leq N:\Vert x_{k}-x_{m}\Vert\leq(s/N)\}=2s,\text{ for every }s\geq 0.\tag{\(\ast\)} \] The concept of Poissonian pair correlations has its origin in quantum mechanics, where the spacings of energy levels of integrable systems were studied (see [\textit{I. Aichinger} et al., in: Contemporary computational mathematics -- a celebration of the 80th birthday of Ian Sloan. In 2 volumes. Cham: Springer. 1--16 (2018; Zbl 1405.81033)] and the references therein for detailed information on that topic). Over the years the topic has attracted wide attention [\textit{D. R. Heath-Brown}, Math. Proc. Camb. Philos. Soc. 148, No. 3, 385--407 (2010; Zbl 1239.11081); \textit{R. Nair} and \textit{M. Pollicott}, Isr. J. Math. 157, 219--238 (2007; Zbl 1124.11033); \textit{Z. Rudnick} and \textit{A. Zaharescu}, Acta Arith. 89, No. 3, 283--293 (1999; Zbl 0961.11028); \textit{Z. Rudnick} et al., Invent. Math. 145, No. 1, 37--57 (2001; Zbl 1006.11041); \textit{Z. Rudnick} and \textit{P. Sarnak}, Commun. Math. Phys. 194, No. 1, 61--70 (1998; Zbl 0919.11052); \textit{S. Steinerberger}, Indag. Math., New Ser. 29, No. 5, 1167--1178 (2018; Zbl 1425.11141); J. Number Theory 208, 47--58 (2020; Zbl 1445.11077)] and recently, \textit{C. Aistleitner} et al. [Isr. J. Math. 222, No. 1, 463--485 (2017; Zbl 1388.11043)] could give a strong link between the concept of Poissonian pair correlations and the additive energy of a finite set of integers, a notion that plays an important role in many mathematical fields. In this work, the authors study the pair correlation statistics for the sequence of shifts of \(\alpha\), \(x_{n}=\{2^n\alpha\}\); \(n=1, 2, 3,\dots\), where \(\alpha\) is chosen as the Champernowne constant in base 2, i.e., \(\alpha=0.1101110010111011\dots_2\) and \(\{\cdot\}\) denotes the fractional part of a real number. It is a well known fact that the Champernowne constant in base 2 is normal to base 2. Moreover, we know that the sequence \((\{2^n\alpha\})_{n\in N}\) is uniformly distributed modulo 1 if and only if \(\alpha\) is normal. It is well known that \((x_{n})_{n\in N}\) has Poissonian pair correlations for almost all normal numbers (in the sense of Lebesgue), but the authors show that it does not have this property for all normal numbers, as it fails to be Poissonian for the Champernowne constant. Therefore, they investigate the distribution of the spacings between the sequence elements of \((\{2^n\alpha\})_{n\in N}\), the only reasonable choice for \(\alpha\) is a normal number. They obtain the following result: Theorem: The sequence \((\{2^n\alpha\})_{n\in N}\) where \(\alpha\) is the Champernowne constant in base 2, i.e., \(\alpha=0.1101110010111011\dots_2\) does not have Poissonian pair correlations. The proof of this theorem is given by the authors. In the last section, they state an open problem, which involves the notion of weak pair correlations, introduced by Steinerberger in [\textit{S. Steinerberger}, Acta Arith. 180, No. 2, 183--199 (2017; Zbl 1381.11059)], a concept that relaxes the requirements of \((\ast)\).