The horocycle flow at prime times (Q488026)

In the article under review, the authors prove that the orbit of a non-periodic point at prime values of the horocycle flow in the modular surface is dense in a set of positive measure. Assuming the Ramanujan/Selberg conjectures, they prove in some special cases that set of positive measure can be replaced by the entire modular surface. In addition, they give an effective version of \textit{S. G. Dani}'s theorem for the discrete horocycle flow [Ergodic Theory Dyn. Syst. 2, 139--158 (1982; Zbl 0504.22006)]. Let \((X,T)\) be a dynamical system. For \(x \in X,\) let \(\theta_x = \{T^n x : n \geq 1 \}\) be the orbit of \(x,\) and define \(P_x = \{T^p x : p~\text{prime} \}.\) Let \(G\) be a connected Lie group, \(\Gamma\) a lattice in \(G,\) \(u \in G\) an \(\text{Ad}_G\) unipotent element, and \(X = \Gamma \setminus G.\) Let \(T:X \rightarrow X\) be defined by \(T(\Gamma g) = \Gamma g u.\) Set \(x = \Gamma g.\) Then \textit{M. Ratner}'s theorem [Ann. Math. (2) 118, 277--313 (1983; Zbl 0556.28020)] says that the closure \(\overline{\theta_x}\) is homogenous and the orbit \(xu^n,\) \(n=1,2,\dots\) is equidistributed in \(\overline{\theta_x}.\) In the paper under review, the authors consider the case of \text {\(X = \text{SL}(2,\mathbb{Z})\setminus\text{SL}(2,\mathbb{R}).\)} Here \(\overline{\theta_x}\) is either finite, a closed horocycle of length \(l, 0 < l < \infty,\) or is all of \(X\) [\textit{G. A. Hedlund}, Duke Math. J. 2, 530--542 (1936; Zbl 0015.10201)]. In the last case, \(x\) is said to be \textit{generic.} For \(N \geq 1, x \in X, \) define the probability measure \(\pi_{x,N}\) by \[ \pi_{x,N} = \frac{1}{\pi(N)} \sum_{p < N} \delta_{x u^p}. \] Here for \(\zeta \in X,\) \(\delta_\zeta\) is the delta mass at \(\zeta\) and \(\pi(N) \) is the number of primes less than \(X.\) Let \(d\mu_G\) be the normalized volume measure on \(X.\) The first main result is {Theorem 1.1} (\textit{Non-concentration at primes}). Let \(x\) be generic and \(v_x\) a weak limit of \(\pi_{x,N},\) then \[ dv_x \leq 10 d\mu_G. \] Here, as \(N \rightarrow \infty, \pi_{x,N}\) converges weakly in the sense of integrating against continuous functions on the one-point compactification of \(X.\) A corollary (Corollary 1.2) of the theorem is that \(\text{Vol}(\overline{P_x}) \geq \frac{1}{10}.\) Next consider the special case of Hecke orbits. Let \[ g = H_N := \begin{pmatrix} N^{-\frac{1}{2}} & 0 \\ 0 & N^{\frac{1}{2}} \\ \end{pmatrix}, ~ u = \begin{pmatrix} 1 & 1 \\ 0 & 1 \\ \end{pmatrix}. \] The other main result is {Theorem 1.3} (\textit{Prime Hecke orbits are dense}). Let \(v\) be a weak limit of the measures \(\pi_{H_N,N}.\) Assuming the Ramanujan/Selberg Conjectures concerning the automorphic spectrum of \(\text{GL}_2/\mathbb{Q}\) ([the first author, J. Funct. Anal. 184, No. 2, 419--453 (2001; Zbl 1006.11022)]), we have \[ \frac{1}{5}d\mu_G \leq dv \leq \frac{9}{5} d\mu_G. \]