On the asymptotic behavior of the number variance of random points on the unit torus

Abstract

The study of the distribution of real sequences in the unit torus has a long and rich history. There are many notions which describe the ``pseudo-random'' behavior of a given deterministic sequence (often of arithmetic origin), such as equidistribution (which describes the distribution on a ``global scale'') or the pair correlation (which is a ``local-scale'' statistics). Another popular and important statistics, which can be applied to all ranges from global via intermediate to local, is the number variance $V_N(s)$, which describes the mean-square fluctuation of the number of elements of an initial segment of the sequence in an interval of length $s$, when moving this interval around the torus. In the terminology of quantum chaology, where such statistics are used to describe the distribution of the energy levels of a system, pseudo-random behavior is called Poissonian behavior. Since the number variance is an important indicator to qualify pseudo-random behavior, it is quite surprising that the literature does not seem to contain a precise description of the distribution and of the almost sure asymptotic behavior of the number variance in the truly random case (i.i.d.\ random points), even in the comparatively simple setup of uniformly distributed points on the unit torus. In the present paper we calculate the distribution of the number variance with high precision throughout a wide range of the parameter $s$, and determine the precise threshold (in terms of $s$) where the ``Poissonian'' asymptotics $V_N(s) \sim N s$ breaks for a typical realization of a random sequence. Key technical input comes from the Komlós--Major--Tusnády theorem and the Karhunen-Loève decomposition.