Statistical distribution of the Stern sequence (Q2421725)

In this long paper the authors study some statistical properties of the Stern sequence. The Stern sequence (or Stern's diatomic sequence) \((s_{n})_{n\in\mathbb{N}}\), is the recurrence sequence defined in the following way \[ s_{0}=0, \;s_{1}=1,\;\text{and, for}\;n\geq 1, s_{2n}=s_{n},\;s_{2n+1}=s_{n}+s_{n+1}. \] One among many striking properties of the Stern sequence is the following: the sequence \((s_{n}/s_{n+1})_{n\in\mathbb{N}}\) enumerates the set of non-negative rational numbers. The main result of the paper is the central limit theorem for the random variable \(\log S_{N}\), where \(S_{N}=s_{n}\) and \(n\) is taken uniformly from the set \(J_{N}=\mathbb{Z}\cap [2^{N},2^{N+1})\). More precisely, for some constants \(\alpha, \sigma>0\), as \(N\) tends to infinity, the values \(\log s_{n}\) are asymptotically distributed according to a Gaussian law with mean \(\alpha N\) and variance \(\sigma^{2}N\): for \(t\in\mathbb{R}\) satisfying \(t=O(N^{1/6})\) the following equality holds \[ \mathbb{P}_{N}\left[\frac{\log S_{N}-\alpha N}{\sigma\sqrt{N}}\right]=\int_{-\infty}^{t}\frac{e^{-v^{2}/2}}{\sqrt{2\pi}}dv+O\left(\frac{(1+t^2)e^{-t^{2}/2}}{\sqrt{N}}\right). \] This result answer the question from the paper of \textit{J. Lansing} [J. Integer Seq. 17, No. 7, Article 14.7.5, 18 p. (2014; Zbl 1317.11024)].