Nearly optimal spectral gaps for random Belyi surfaces
Yang Shen, Yunhui Wu · 2025
Read Nearly optimal spectral gaps for random Belyi surfaces by Yang Shen, Yunhui Wu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-02517.txt.
Introduction
Yang Shen and Yunhui Wu confirm a conjecture about the vibrations of a typical curved surface. Take a random hyperbolic surface built by the Brooks-Makover construction, gluing together idealised pieces at random. Its Laplace spectrum has a gap above zero, and the size of that gap controls how quickly heat, waves, and random walks spread and mix across the surface. The best gap any such surface could hope for is bounded by the one-quarter threshold.
The authors prove that a random surface in this model has a spectral gap exceeding one quarter minus a small term. That term shrinks like a constant divided by the logarithm of the surface's size parameter. In other words, random surfaces come arbitrarily close to the optimal gap as they grow, which is exactly the nearly optimal spectral gap conjecture for this model. The proof runs to fifty pages, with an appendix.
Catalog record
| CREATOR | Yang Shen, Yunhui Wu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 106 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-02517
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Nearly optimal spectral gaps for random Belyi surfaces Yang Shen, Yunhui Wu Abstract . In this paper, we show that for any ϵ > 0 \epsilon>0 , a random hyperbolic surface in the Brooks-Makover model has a spectral gap greater than 1 4 − ϵ \frac{1}{4}-\epsilon , confirming the nearly optimal spectral gap conjecture in this model. 1. Introduction Let X_{g} be a closed hyperbolic surface of genus g , and let \lambda_{1}(X_{g}) be its first Laplacian eigenvalue. It is known from e.g. Huber [ Hub74 ] or Cheng [ Che75 ] that \limsup\limits_{g\to\infty}\lambda_{1}(X_{g})\leq\frac{1}{4} for any sequence of closed hyperbolic surfaces \{X_{g}\}_{g\geq 2} . A long-standing conjecture of Buser [ Bus84 ] predicted the sharpness of this bound, i.e., the existence of a sequence \{X_{g}\} of closed hyperbolic surfaces with \lambda_{1}(X_{g})\to\frac{1}{4} as g\to\infty . This was first achieved by Hide-Magee [ HM23 ] , who essentially proved the nearly optimal spectral gap for random covers of a cusped hyperbolic surface. For closed surfaces, research has focused on three prominent randomness models: the Brooks-Makover model, the Weil-Petersson model, and the covering model. It is widely conjectured that a random closed hyperbolic surface in all these three models has nearly optimal spectral gap. In what follows, we briefly review the progress toward this conjecture across the three models. Brooks-Makover Model : Brooks and Makover [ BM04 ] proved that a random closed hyperbolic surface in their model has a uniform spectral gap c>0 , where the constant c is implicit. This was not only the first proof of a uniform spectral gap for random closed hyperbolic surfaces but also remains the only such result within their specific model. Weil-Petersson Model : The study of spectral gaps for random closed hyperbolic surfaces in this model began with Mirzakhani [ Mir13 ] , who proved a lower bound 0.00247 . This bound was significantly improved to \left(\frac{3}{16}-\epsilon\right) for any \epsilon>0 by Wu-Xue [ WX22 ] using a completely different method, a result independently obtained by Lipnowski-Wright [ LW24 ] . In a series of subsequent works, Anantharaman-Monk raised the lower bound further to \left(\frac{2}{9}-\epsilon\right) in [ AM23 ] and eventually to \left(\frac{1}{4}-\epsilon\right) in the remarkable work [ AM25 ] , resolving the nearly optimal spectral gap conjecture for the Weil-Petersson model. Most recently, Hide-Macera-Thomas [ HMT25 ] provided an alternative proof via the polynomial method, additionally refining the \epsilon term to a significant explicit polynomial error of order \frac{1}{g^{\delta}} for some uniform \delta>0 . Covering Model : Magee-Naud-Puder [ MNP22 ] proved that for any \epsilon>0 and any fixed closed hyperbolic surface X , the eigenvalues (with multiplicity) of a random covering surface of X within \left(0,\frac{3}{16}-\epsilon\right) coincide with those of X . In a recent remarkable work [ MPvH25 ] , Magee-Puder-van Handel used the polynomial method to improve the bound from \left(\frac{3}{16}-\epsilon\right) to \left(\frac{1}{4}-\epsilon\right) , resolving the nearly optimal spectral gap conjecture for the covering model. Recall that for all n\geq 1 , set \mathcal{F}_{n}^{\star}=\left\{(\Gamma,\mathcal{O});\ \begin{matrix}\Gamma\text{ is a $3-$regular graph with $2n$ vertices,}\\ \mathcal{O}\text{ is an orientation on }\Gamma\end{matrix}\right\}. As introduced by Brooks-Makover [ BM04 ] , there are two associated hyperbolic surfaces for each (\Gamma,\mathcal{O})\in\mathcal{F}_{n}^{\star} : a cusped hyperbolic surface S^{O}(\Gamma,\mathcal{O}) which is obtained by gluing 2n hyperbolic ideal triangles in a certain way, and a closed hyperbolic surface S^{C}(\Gamma,\mathcal{O}) which is the conformal compactification of S^{O}(\Gamma,\mathcal{O}) by filling in all the cusps. It is known that such constructions produce all the so-called Belyi surfaces that are dense in the space of all Riemann surfaces in a certain sense (see e.g. [ Bel79 , Gam06 ] ). For each n , the set \mathcal{F}_{n}^{\star} is finite, and hence there is a classical probability measure denoted by \textnormal{Prob}_{\textnormal{BM}}^{n} on \mathcal{F}_{n}^{\star} , which was introduced by Bollabás [ Bol88 ] . As mentioned previously, the Brooks-Makover model has stood as the final piece of the puzzle among the three models of random closed hyperbolic surfaces, being the only one for which the nearly optimal spectral gap conjecture was still unproven. Our work completes this picture by resolving the conjecture (as e.g. formulated in [ Pet24 , Page 71] ). Specifically, we prove Theorem 1.1 . For any \epsilon>0 , \lim\limits_{n\to\infty}\textnormal{Prob}_{\textnormal{BM}}^{n}\left((\Gamma,\mathcal{O})\in\mathcal{F}_{n}^{\star};\ \lambda_{1}\left(S^{C}(\Gamma,\mathcal{O})\right)>\frac{1}{4}-\epsilon\right)=1. It is known from [ BM04 ] that a random cusped hyperbolic surface S^{O}(\Gamma,\mathcal{O}) has the so-called large cusps property. This together with [ BM01 ] tells that as n\to\infty , for a random (\Gamma,\mathcal{O})\in\mathcal{F}_{n}^{\star} , \lambda_{1}\left(S^{C}(\Gamma,\mathcal{O})\right)\geq\min\left\{\frac{1}{4}-\epsilon,(1-o(1))\cdot\lambda_{1}\left(S^{O}(\Gamma,\mathcal{O})\right)\right\}. Therefore, to prove Theorem 1.1 , it suffices to show that a random cusped hyperbolic surface S^{O}(\Gamma,\mathcal{O}) has no eigenvalues in \left(0,\frac{1}{4}-\epsilon\right) for any \epsilon>0 . Our approach is inspired by the work [ MPvH25 ] of Magee-Puder-van Handel that builds upon the polynomial method in e.g. [ CGVTvH24 , CGVvH24 , MdlS24 ] . Meanwhile, we believe that the novel ingredients developed in this work will find further applications in studying the asymptotics of other geometric quantities in this model of random hyperbolic surfaces. For more related results of spectral gaps on random hyperbolic surfaces, one may refer to e.g. [ Hid23 , HT25 , SW25 ] for the Weil-Petersson model and to e.g. [ MN20 , HM23 , LM25 ] for the covering model. Similar results on spectral gaps also hold for random graphs. Alon [ Alo86 ] conjectured and Friedman [ Fri08 ] proved that a random d- regular (d\geq 3) graph has nearly optimal spectral gap, as the number N of vertices goes to infinity. A recent breakthrough of Huang-Mckenzie-Yau [ HMY25 ] proved that as N\to\infty , about 69\% d- regular graphs with N vertices are Ramanujan graphs, that is, have optimal spectral gaps. It is natural to ask whether an analogous result holds for all these three models of random closed hyperbolic surfaces. That is, does there exist a constant c\in(0,1) such that \lim\limits_{g\to\infty}\textnormal{Prob}\left(\lambda_{1}(X)\geq\frac{1}{4}\right)=c? where the constant c>0 may be different for these three models of random closed hyperbolic surfaces. The existence of infinite families of Ramanujan graphs was first proved in [ LPS88 , Mar88 , MSS15 ] . We remark here that it remains even open that there exists a closed hyperbolic surface X_{g} of genus g such that \lambda_{1}(X_{g})\geq\frac{1}{4} for all g\geq 2 . Strategy on the proof of Theorem 1.1 . We first give a new geometric description for each cusped hyperbolic surface S^{O}(\Gamma,\mathcal{O}) in the Brooks-Makover model. For n\geq 1 , assume that S_{6n} is the permutation group of 6n elements. Set \mathcal{E}_{n}=\left\{(\sigma,\tau);\ \begin{matrix}\sigma,\tau\in S_{6n},\text{ $\sigma$ has order $2$ and no fixed point,}\\ \text{$\tau$ has order $3$ and no fixed point}.\end{matrix}\right\}. For each (\sigma,\tau)\in\mathcal{E}_{n} , it induces a group homomorphism \Phi(\sigma,\tau):\textnormal{PSL}(2,\mathbb{Z})\to S_{6n}, which corresponds to a degree 6n covering hyperbolic surface S^{O}(\sigma,\tau) of the modular surface \mathbb{H}/\textnormal{PSL}(2,\mathbb{Z}) . For each cusped hyperbolic surface S^{O}(\Gamma,\mathcal{O}) in the Brooks-Makover model, it is isometric to certain S^{O}(\sigma,\tau) (see e.g. Theorem 2.5 ). The two major differences between the Brooks-Makover model and the covering model of cusped hyperbolic surfaces (see e.g. [ HM23 ] ) are as follows: (i) for any cusped hyperbolic surface X , its fundamental group \pi_{1}(X) is a free group, while \textnormal{PSL}(2,\mathbb{Z})\simeq\mathbb{Z}_{2}\star\mathbb{Z}_{3} is not; (ii) all degree n covering hyperbolic surfaces of a fixed cusped hyperbolic surface X correspond to all group homomorphisms \phi\in\textnormal{Hom}(\pi_{1}(X),S_{n})