betaFeedback

WorksMathematics

On the kissing number of hyperbolic surfaces with a small systole

Abstract

Parlier proved that there exists a constant U>0U>0 such that for any hyperbolic surface SS of genus gg with systole ℓ\ell the following holds: Kiss(S)≤Ueℓℓg Kiss(S)\leq U \frac {e^\ell}\ell g where Kiss(S)Kiss(S) is the maximum number of homotopically distinct systoles in SS (Kissing number). We show that there exists a constant V>0V>0 and a sequence SnS_n of hyperbolic homogeneous surfaces of genus gn→∞g_n\to \infty with a bounded systole such that Kiss(Sn)≥Vgn Kiss(S_n)\geq Vg_n for all n≥1n\geq 1. This shows that Parlier's theorem is sharp for a fixed systole length. Our proof uses Bruhat–Tits buildings of type A~2\tilde A_2 and Hamiltonian surfaces.

Introduction

In his paper [6], Hugo Parlier proves that the number of homotopically distinct systoles (kissing number) in a closed hyperbolic surface SS of genus gg and systoles ℓ\ell is bounded above by

Kiss⁡(S)≤Ueℓℓg\operatorname{Kiss}(S) \le U \frac{e^\ell}{\ell}g

where U>0U > 0 is a constant. This implies that the number of systoles in any hyperbolic surface of genus gg is bounded above by

g2log⁡g.\frac{g^2}{\log g}.

The search for lower bounds has a long history (see [6]). Paul Schmutz Schaller [8, 9] has shown that there exists a family of arithmetic surfaces with kissing number bounded below by

Kiss⁡(S)≥g4/3−ε∀ε>0.\operatorname{Kiss}(S) \ge g^{4/3-\varepsilon} \quad\forall\varepsilon> 0.

This is the best known construction to date and therefore there is a gap between Schmutz Schaller’s lower bound g4/3−εg^{4/3-\varepsilon} and Parlier’s upper bound g2log⁡g\frac{g^2}{\log g}. In [9], Schmutz Schaller conjectures that the bound g4/3g^{4/3} is optimal.

Our main theorem shows that Parlier upper bound is sharp in the case of a surface with a bounded systole.

Theorem 1.1. There exists a constant V>0V > 0 and a sequence SnS_n of hyperbolic homogeneous surfaces of genus gn→∞g_n \to\infty with a bounded systole such that

Kiss⁡(Sn)≥Vgn\operatorname{Kiss}(S_n) \ge Vg_n

for all n≥1n \ge1.

The systole in these surfaces is in fact constant in nn equal to ℓ≈8.69\ell\approx8.69. The proof of the theorem relies on the construction of a certain type of regular surfaces that are associated with a lattice in PGL⁡3(F2((T)))\operatorname{PGL}_3(\mathbb{F}_2((T))). These surfaces are “Hamiltonian surfaces”, in the sense of [2]. They will visit every vertex and every edge of a large quotient of the Bruhat–Tits building associated with PGL⁡3(F2((T)))\operatorname{PGL}_3(\mathbb{F}_2((T))), exactly once.

The notion of Hamiltonian surfaces derives from a new construction of 2-complexes using an operation of “systolic filling”, in a finite covering space associated with a given surface of infinite genus. This approach was first studied in [1] for systolic fillings associated with a pinched torus, and later in [2], for systolic fillings of hyperbolic surfaces of infinite genus, naturally associated with the group Aut⁡(F2)\operatorname{Aut}(F_2).

On AI usage

No use of AI was made in the conception or the writing of this paper.

Construction of the Hamiltonian surface

By Hamiltonian surface in a 2-complex XX, we mean a surface which visits every vertex and every edge precisely once. A Hamiltonian surface is said to be periodic if it endowed with a free and cocompact action of a countable group.

Theorem 2.1. The Bruhat–Tits building XX of type A~2\widetilde{A}_2 over the field F2((T))\mathbb{F}_2((T)) contains three distinct periodic Hamiltonian surfaces, which are cyclically permuted by a group GG of automorphisms of XX that acts simply transitively by type-rotating automorphisms.

The group GG in this theorem was introduced in [3]. It will be described explicitly in the proof.

Proof. We first review the construction of an explicit embedding into PGL⁡3(F2((T)))\operatorname{PGL}_3(\mathbb{F}_2((T))) of an amalgamated group (of Tits type) following the work of [10, 5, 7, 3, 4]. Namely, we consider the Frobenius group of order 21 defined by:

F=⟨p,s∣p7=s3=1, sps−1=p2⟩≃⟨p,s∣p7=s3=1, sps−1=p4⟩F = \langle p,s \mid p^7=s^3=1,\ sps^{-1}=p^2\rangle\simeq\langle p,s \mid p^7=s^3=1,\ sps^{-1}=p^4\rangle

in which Z/3Z\mathbb{Z}/3\mathbb{Z} acts on Z/7Z\mathbb{Z}/7\mathbb{Z} by conjugacy. It is well-known [7] that two distinct copies of Z/3Z\mathbb{Z}/3\mathbb{Z} in FF generate a chamber system for the building of SL⁡3(2)\operatorname{SL}_3(2).

Furthermore, Tits has shown in [10] (see also [7 p. 48]) that there are three ways to amalgamate three copies of FF together along two distinct copies of Z/3Z\mathbb{Z}/3\mathbb{Z} in each group, in order to define a group, which does not collapse, and acts by automorphisms of the Bruhat–Tits building of PGL⁡3(F2((T)))\operatorname{PGL}_3(\mathbb{F}_2((T))).

In [5], Kohler, Meixner, and Wester have shown that one of these three amalgams embeds in PGL⁡3(F2((T)))\operatorname{PGL}_3(\mathbb{F}_2((T))). Furthermore, they use this embedding to construct exponentially many pairwise non-isomorphic finite transitive chamber systems of type A~2\widetilde{A}_2. It will be convenient to use a closely related embedding of the following group GG, which was introduced by Cartwright, Mantero, Steger, and Zappa in [3, 4]. Thus, we consider the group GG on seven generators aia_i and seven relations:

aiai+1ai+3a_i a_{i+1} a_{i+3}

where i∈Z/7Zi \in\mathbb{Z}/7\mathbb{Z}. This presentation is an A~2\widetilde{A}_2-triangle presentation in the sense of [3 p. 149]. The automorphism group of this triangle presentation is FF and the extension G~\widetilde{G} of GG by FF is generated by the aia_i, pp and ss with commutation relations:

paip−1=ap(i)sais−1=as(i)p a_i p^{-1}=a_{p(i)} \qquad s a_i s^{-1}=a_{s(i)}

where p(i)=i+1p(i)=i+1 and s(i)=2is(i)=2i.

The following explicit embedding theorem is established in [3 §4, Example] and described in [4 p. 177].

Theorem 2.2. The map

a0↦(T000T1+T01+T1),p↦(101100010),s↦(100011010)a_0 \mapsto\begin{pmatrix} T & 0 & 0 \\ 0 & T & 1+T \\ 0 & 1+T & 1 \end{pmatrix}, \quad p \mapsto\begin{pmatrix} 1 & 0 & 1 \\ 1 & 0 & 0 \\ 0 & 1 & 0 \end{pmatrix}, \quad s \mapsto\begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 1 \\ 0 & 1 & 0 \end{pmatrix}

defines to an embedding of G~\widetilde{G} into PGL⁡3(F2((T)))\operatorname{PGL}_{3}(\mathbb{F}_{2}((T))).

The relation with the embedding of Kohler, Meixner, and Wester is explained in the last part of §4 in [3].

It is shown in [3] that the group GG acts simply transitively on the vertices of the building XX of PGL⁡3(F2((T)))\operatorname{PGL}_{3}(\mathbb{F}_{2}((T))). Furthermore, the action is by type-rotating automorphisms. In fact, the type-rotating assumption essentially yields a presentation of GG of the form given above, namely the A~2\widetilde{A}_{2}-triangle presentation of GG. For all this, we refer [3 Theorem 3.1].

The type-rotating assumption implies that the group GG admits a subgroup G3G_{3} of index 3, which act by type-preserving automorphisms. Furthermore, the presentation of G3G_{3} is easily computed, following [3]. In the particular case above, we find a group on 21 generators ai,ja_{i,j} and 21 relations, as follows:

ai,jai+1,j+1ai+3,j+2a_{i,j}a_{i+1,j+1}a_{i+3,j+2}

where i∈Z/7Zi\in\mathbb{Z}/7\mathbb{Z} and j∈Z/3Zj\in\mathbb{Z}/3\mathbb{Z}. We let X3X_{3} denote the corresponding finite chamber system.

Our main observation is:

Lemma 2.3. The set Σp\Sigma_{p} of 14 chambers ai,jai+1,j+1ai+3,j+2a_{i,j}a_{i+1,j+1}a_{i+3,j+2} for all i∈Z/7Zi\in\mathbb{Z}/7\mathbb{Z} and all jj omitting the value pp in Z/3Z\mathbb{Z}/3\mathbb{Z} defines a Hamiltonian surface in X3X_{3}.

Proof. Since one value of jj is omitted, it is clear that the generator ai,ja_{i,j} appears only twice in the given set of chambers, for all i∈Z/7Zi\in\mathbb{Z}/7\mathbb{Z} and j∈Z/3Zj\in\mathbb{Z}/3\mathbb{Z}. To prove that XpX_{p} is a hamiltonian surface, it suffices to prove that its links form a Hamiltonian cycle in the links of X3X_{3}.

The generic chamber CijC_{ij} in Σp\Sigma_{p} is given by:

Figure 1. Chamber CijC_{ij}

By definition, the link of Σp\Sigma_{p} is the graph with edges:

(ai,j−,ai+1,j+1+),(ai+1,j+1−,ai+3,j+2+),(ai+3,j+2−,ai,j+)\left(a_{i,j}^{-},a_{i+1,j+1}^{+}\right),\left(a_{i+1,j+1}^{-},a_{i+3,j+2}^{+}\right),\left(a_{i+3,j+2}^{-},a_{i,j}^{+}\right)

where i∈Z/7Zi\in\mathbb{Z}/7\mathbb{Z} and j∈Z/3Z∖pj\in\mathbb{Z}/3\mathbb{Z}\setminus p. It is straightforward to verify that this graph is a union of three 14-gons. □

To conclude the proof of the theorem, it suffices to prove that GG permutes cyclically the three lifts of the surfaces Σp⊂X3\Sigma_p \subset X_3 to the universal cover XX. This follows because the action is type-rotating. It is clear that each lift is a periodic Hamiltonian surface in XX on which G3G_3 act by automorphisms. □

The following result was stated without proof in [2], we can now provide a proof.

Corollary 2.4. The Bruhat–Tits building of type A~2\widetilde{A}_2 over F2((T))\mathbb{F}_2((T)) contains uncountably many Hamiltonian surfaces.

Proof. If KK denotes a vertex stabilizer in PGL⁡3(F2((T)))\operatorname{PGL}_3(\mathbb{F}_2((T))) and Σ\Sigma is one of the three Hamiltonian surfaces in XX, then the orbit KΣK\Sigma is uncountable, by the orbit–stabilizer theorem. □

Proof of Theorem 1.1

We establish the following theorem which is our main result.

Theorem 3.1. There exists a sequence SnS_n of simplicially hyperbolic homogeneous surfaces of genus gn→∞g_n \to\infty with bounded systole ℓ=3\ell= 3 such that

Kiss⁡(Sn)=72(gn−1)\operatorname{Kiss}(S_n) = \frac{7}{2}(g_n - 1)

for all $n \ge 1.

We note that Theorem 1.1 follows immediately from this result by replacing every equilateral triangle in SnS_n with the corresponding hyperbolic triangle of constant curvature and vertex angle π7\frac{\pi}{7}. Each systole of length 33 gives rise to a systole of length

ℓ=3×2arccosh⁡(12sin⁡(π/14))≃8.69\ell= 3 \times2\operatorname{arccosh}\left(\frac{1}{2\sin(\pi/14)}\right) \simeq8.69

in the corresponding smooth surface.

Proof. Let XX denote the Bruhat–Tits building of PGL⁡3(F2((T)))\operatorname{PGL}_3(\mathbb{F}_2((T))). Consider one of the three periodic hamiltonian surface SS in XX defined in Theorem 2.1. This surface is G3G_3-periodic, where G3G_3 is of index 33 in GG. We will need to use here the group GG defined in [3] and reviewed in the previous paragraph.

Since G3G_3 is linear, it is residually finite. Therefore, there exists a finite index subgroup K0K_0 of G3G_3 for which the injectivity radius of the cover space X↠X/K0X \twoheadrightarrow X/K_0 is at least 22. This implies that the systole in S/K0S/K_0 is at least 33.

Let (Kn)n≥0(K_n)_{n\geq0} be a decreasing sequence of normal finite index normal subgroups of G3G_3, and let Sn:=S/KnS_n := S/K_n. The 22-complex Xn:=X/KnX_n := X/K_n is obtained from SnS_n by systolic filling in the sense of [2]. The systoles of SnS_n have length 33, and XnX_n is obtained canonically from SnS_n by adjoining a triangle for every systole. This is in fact easy to see directly from the construction of SS in Theorem 2.1. Namely, in Lemma 2.3, the surface Σp\Sigma_p is obtained by omitting the chambers corresponding to a single index p∈Z/3Zp \in\mathbb{Z}/3\mathbb{Z}. The systolic filling of SnS_n is obtained by adjoining all chambers which are lifts of the chambers corresponding to this index. In this construction, every systole is a boundary of such a chamber. It has length 33, and meets every edge precisely once. The residual finiteness of GG and the presence of XX is crucial to find the correct covering space of S/GS/G.

It is clear that the surfaces SnS_n are simplicially hyperbolic with (by construction) 14 equilateral triangles meeting at every vertex. Furthermore, they are homogeneous surfaces acted upon by the group Hn:=G/KnH_n := G/K_n.

It remains to compute the number of systoles and the genus of SnS_n. Since every systole meets an edge exactly once, the number of systoles is en/3e_n/3 where ene_n is the number of edges of SnS_n. Since SnS_n is Hamiltonian, the number of edges in SnS_n equals the number of edges in XnX_n.

Since GG acts freely transitively by type rotating automorphisms on XX, and G3G_3 acts freely by type preserving automorphisms of XX, the action of G3G_3 on X3X_3 is free with 3 orbits of vertices, 21 orbits of edges, and 21 orbits of faces. Furthermore, the action HnH_n on XnX_n is free with quotient complex X/G3X/G_3. In particular, the surface SnS_n contains precisely en=21∣Hn∣e_n = 21|H_n| edges and therefore 7∣Hn∣7|H_n| systoles. Furthermore, the Euler characteristic of SnS_n is

2−2gn=χ(Sn)=(v0−e0+f0)∣Hn∣=−4∣Hn∣.2 - 2g_n = \chi(S_n) = (v_0 - e_0 + f_0)|H_n| = -4|H_n|.

This shows that

Kiss⁡(Sn)=7∣Hn∣=72(gn−1)\operatorname{Kiss}(S_n) = 7|H_n| = \frac{7}{2}(g_n - 1)

for all n≥1n \ge1. □

References

  1. [1]Barré, Pichot. Surgery on Aut(F_2), Bull. Belg. Math. Soc. Simon Stevin 30(1): 31-50 (july 2023). DOI: 10.36045/j.bmss.210809.DOI
  2. [2]Barré, Pichot. The Hamiltonian surface of Aut(F_2), preprint (2025).arxiv.org/abs/2507.00213
  3. [3]Cartwright, D. I., Mantero, A. M., Steger, T., Zappa, A. (1993). Groups acting simply transitively on the vertices of a building of type à 2, I. Geometriae Dedicata, 47(2), 143-166.DOI
  4. [4]Cartwright, D. I., Mantero, A. M., Steger, T., Zappa, A. (1993). Groups acting simply transitively on the vertices of a building of type à 2, II: The cases q= 2 and q= 3. Geometriae Dedicata, 47(2), 167-223.
  5. [5]Kohler, P., Meixner, T. and Wester, M., ’The affine building of type A2 over a local field of characteristic 2’, Archiv Math. 42 (1984), 400-407.
  6. [6]Parlier, H. Kissing numbers for surfaces. Journal of Topology, 6(3), 777-791 (2013).
  7. [7]Ronan, M., Lectures on Buildings, Academic Press, New York, 1989.
  8. [8]Schmutz Shaller, P. Extremal Riemann surfaces with a large number of systoles. In Extremal Riemann surfaces (San Francisco, CA, 1995), vol. 201 of Contemp. Math. Amer. Math. Soc., Providence, RI, 1997, pp. 9–19.
  9. [9]Schmutz Shaller, P. Geometry of Riemann surfaces based on closed geodesics. Bull. Amer. Math. Soc. (N.S.) 35, 3 (1998), 193–214.
  10. [10]Tits, J., Buildings and Group Amalgamations, London Maths. Soc. Lecture Notes, 121 (Proceedings of Groups - St. Andrews 1985), Cambridge Univ. Press, 1986, 110-127.DOI

Paper details

vlnc:W_t4uqf77d
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents