Uniform distribution of saddle connection lengths (with an appendix by Daniel El-Baz and Bingrong Huang) (Q2026745)

A saddle connection on a translation surface is a geodesic segment joining cone points (with no cone points in the interior of the segment). There are countably infinitely many saddle connections on any translation surface, and finitely many of length less than any finite number. The main result of this paper is that, for almost every surface, when saddle connection lengths are ordered in an increasing list, the lengths are uniformly distributed modulo 1. It is open whether the ``almost every'' assumption can be relaxed. This problem fits into the general framework of ``counting saddle connections'', which has produced an extensive literature. Advances on such problems are an important marker of progress in Teichmüller dynamics. The question that the authors address has a somewhat different flavor than the usual questions studied, which tend to be amenable to renormalization techniques via the \(\mathrm{SL}_2(\mathbb{R})\) action on the space of translation surfaces. The authors do use the \(\mathrm{SL}_2(\mathbb{R})\) action, but in a ``perturbative'' fashion, rather than rescaling lengths to become \(O(1)\), which would not help much for understanding lengths modulo \(1\). However, their proof relies on an effective counting result due to [\textit{A. Nevo} et al., Adv. Math. 360, Article ID 106890, 29 p. (2020; Zbl 1451.37049)], whose proof does use renormalization techniques. The authors approach the issue of equidistribution via the Weyl Criterion. Part of their proof resembles a proof of the Strong Law of Large Numbers via moments. To get the required approximate independence, they show that that if two vectors are not too close in length, then they behave somewhat independently when perturbed by small elements of \(\mathrm{SL}_2(\mathbb{R})\) (see Lemma 12). In the context of saddle connections, one gets this length separation using the results of \textit{A. Nevo} et al..