betaFeedback

WorksMathematics

Uniform large-genus asymptotics of Weil-Petersson volumes

Abstract

We prove an expansion in 1/g1/g of the Weil–Petersson volumes Vg,nV_{g,n} that is uniform over all cusp counts, from fixed nn to arbitrarily large n/gn/g. An explicit leading term and single-factorial remainder bounds give exponentially small relative errors after truncation at a suitable order proportional to gg. A positive recursion along the string flow connects these volumes to pure ψψ-class intersection numbers, where Painlevé~I fixes the normalization; uniform coefficient extraction then recovers individual volumes. We determine Zograf's constant C=1/πC=1/\sqrtπ and resolve Mirzakhani's two questions on volume ratios for growing cusp counts and adjacent coefficient ratios uniformly over all degrees. Via Mirzakhani's integration formula, the same volume estimates control random surfaces at two distinct scales. At n/g→c∈(0,∞)n/\sqrt g\to c\in(0,\infty), the entire positive Laplace spectrum below 1/41/4 converges to an explicit Poisson point process generated by geodesics cutting off two cusps. Together with Huang's theorem, this characterizes the transition for n≥1n\ge1: the spectral gap tends in probability to 1/41/4 exactly when n/g→0n/\sqrt g\to0, and to 00 exactly when n/g→∞n/\sqrt g\to\infty. At density n/g→λ∈(0,∞)n/g\toλ\in(0,\infty), we identify the local limit around a typical point; its cusp-filling is conformally a disc. As λ→∞λ\to\infty, these limits approach the infinite Weil–Petersson sphere of Budd and Curien. We prove their parabolicity conjecture and, by a growing-radius coupling, answer their scaling-limit question: after deep cusps are trimmed, the rescaled infinite sphere converges to the Brownian plane.

Acknowledgements

We thank Hao Xu for helpful correspondence on the work of Liu and Xu, the uniformity ranges of the two deformation arguments, and the determination of Zograf’s constant. We acknowledge GeWu SciLab for providing the collaborative research environment in which this project was developed.

Kun Chen is supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDB1680102). Renqiu Xia is supported by the National Natural Science Foundation of China (Grant No. 6260070460).

2020 Mathematics Subject Classification. Primary 32G15; Secondary 11F72, 14H10, 14H81, 41A60, 58J50, 60D05, 60F05.

Key words and phrases. Weil–Petersson volume, moduli space of curves, intersection number, large genus, random hyperbolic surface, spectral gap, local limit theorem.

School of Mathematical Sciences, Peking University, Beijing 100871, China.

Department of Modern Physics, University of Science and Technology of China, Hefei, Anhui 230026, China.

CAS Key Laboratory of Theoretical Physics, Institute of Theoretical Physics, Chinese Academy of Sciences, Beijing 100190, China.

School of Artificial Intelligence, Shanghai Jiao Tong University, Shanghai, China.

Wuhu Qiheng Co., Ltd., Wuhu, China.

Institute of Theoretical Physics, Chinese Academy of Sciences, Beijing 100190, China.

Introduction

Let Mg,n(L)\mathcal{M}_{g,n}(\mathbf{L}) be the moduli space of hyperbolic surfaces of genus gg with nn labeled geodesic boundary components of lengths L=(L1,…,Ln)\mathbf{L}=(L_1,\ldots,L_n), where 2g−2+n>02g-2+n>0 and a boundary component of length zero is a cusp. Its Weil–Peterson volume, in the orbifold convention, is

Vg,n(L)=1(3g−3+n)!∫Mg,n(L)ωWP3g−3+n,Vg,n=Vg,n(0,…,0).V_{g,n}(\mathbf{L})=\frac{1}{(3g-3+n)!}\int_{\mathcal{M}_{g,n}(\mathbf{L})}\omega_{\mathrm{WP}}^{3g-3+n},\qquad V_{g,n}=V_{g,n}(0,\ldots,0).

These volumes connect intersection theory on moduli space with the geometry of random hyperbolic surfaces: their coefficients in the boundary lengths are intersection numbers, and through Mirzakhani’s integration formula their ratios give expected counts of geometric configurations [141, 142, 144]. See [57, 145, 187, 45] for surveys.

Two classical asymptotic theories cover opposite ends of the problem: Mirzakhani and Zograf obtained an expansion as g→∞g\to\infty with nn fixed [147], and Kaufmann–Manin–Zagier and Manin–Zograf obtained the large-nn asymptotics at fixed genus [117, 136]. In the first expansion the coefficient of order g−ig^{-i} is a polynomial in nn of degree 2i2i, with leading term (−1)in2i/(i!(2π2)i)(-1)^i n^{2i}/(i!(2\pi^2)^i). These leading terms alone sum to exp⁡(−n2/(2π2g))\exp(-n^2/(2\pi^2g)), so the effective parameter is n2/gn^2/g: once n≍gn\asymp\sqrt{g}, terms of every fixed order contribute, and powers of 1/g1/g no longer order the corrections asymptotically. The two endpoint theories alone therefore do not describe arbitrary simultaneous growth of genus and cusp count.

We extend the large-genus theory across the full range of cusp densities. A single real parameter YY, the cusp-density parameter, runs over a compact interval [0,y0][0,y_0] as n/gn/g runs over [0,∞][0,\infty] and locates the saddle point of the cusp generating function. Three ideas make the expansion uniform in nn. Moving along the string flow replaces the alternating sum that expresses Weil–Petersson volumes through ψ\psi intersection numbers by a positive recursion, whose kernels form a compact family over [0,y0][0,y_0]. The dilaton and KdV equations, controlled by this positivity, determine the dependence on YY; the remaining normalization is fixed at the endpoint Y=y0Y=y_0, where only pure ψ\psi intersection numbers enter and Painlevé I supplies their asymptotics. Finally, each generating function is a series with nonnegative coefficients times an explicit function raised to a power proportional to the genus; this factorization converts its asymptotics into asymptotics of individual coefficients, uniformly in nn. The same parameter connects the volume problem to the geometry of random surfaces: the scales n≍gn\asymp\sqrt{g} and n≍gn\asymp g mark changes in the low spectrum of these surfaces and in their geometry near a typical point.

Volumes (Part 1). Theorem 1.1 is a single large-genus expansion

V^g,n=V^g,n(1+A1(g,n)2g−1+A2(g,n)(2g−1)2+⋯ )\widehat{V}_{g,n}=\widehat{V}_{g,n}\left(1+\frac{A_1(g,n)}{2g-1}+\frac{A_2(g,n)}{(2g-1)^2}+\cdots\right)

valid uniformly over every cusp count, with an explicit leading term V^g,n\widehat{V}_{g,n}, coefficients bounded uniformly in nn, and single-factorial remainder bounds. In the fixed-cusp limit it identifies the constant conjectured by Zograf; at every density it gives the volume-ratio profiles, and for the one-boundary polynomial it gives the adjacent-coefficient asymptotics in all degrees. These results resolve the two questions posed by Mirzakhani [144, Section 1.4]: the asymptotics of Vg,n/Vg−1,n+2V_{g,n}/V_{g-1,n+2} when n=n(g)→∞n=n(g)\to\infty, and the ratios of adjacent coefficients of Vg,1(L)V_{g,1}(L) when the degree grows with the genus. Theorems 1.3 and 1.4 give the respective answers uniformly over all cusp counts and all adjacent degrees. Liu and Xu subsequently determined Zograf’s constant by a different deformation [131]; the two approaches are compared at the beginning of Part 1.

Spectrum (Part 2). The threshold n≍gn\asymp\sqrt{g}, at which the fixed-cusp expansion breaks down, is also a spectral phase transition. Both are governed by n2/gn^2/g: the expected number of geodesics of bounded length that cut off two cusps is of this order (Subsection 1.6). Huang showed that the spectral gap of a Weil–Petersson random surface tends to 1/41/4 when n=o(g)n=o(\sqrt{g}) [101]. At the critical scale the full positive spectral point process below 1/41/4 converges to a Poisson process of explicit model energies, one for each short two-cusp pair of pants (Theorem 15.1); the gap therefore has a nondegenerate random limit, and tends to zero above the scale. Above the critical scale the lowest eigenvalues follow rescaled Poisson laws in specified ranges, and fixed-energy windows carry deterministic limiting spectral measures. The volume theorem supplies the statistics of short curves; the spectral work is to show that the states these curves create exhaust the low spectrum.

Local geometry (Part 3). At positive density n/g→λ∈(0,∞)n/g \to\lambda\in(0,\infty) the same volume ratios, with based holonomies retained in the integration formula, determine the surface seen from a uniformly chosen point and the conformal type of its cusp-filling. As λ→∞\lambda\to\infty this family reaches the area-rooted infinite WP sphere of Budd and Curien [31]. We prove that this endpoint is parabolic, with recurrent Brownian motion and ergodic geodesic flow, as they conjectured [31, Section 7.3], whereas at every finite density the cusp-filled surface is conformally a disc. A growing-radius coupling of their tree models shows that the endpoint surface, rescaled and with its deep cusps trimmed, converges to the Brownian plane, answering another of their questions.

The common step is geometric integration: a cut expresses the law of a finite configuration through volumes of its complementary surfaces. For spectra, these configuration counts must be matched to the entire low spectral subspace; for local geometry, the relative positions of the based loops must also be retained. Figure 1 shows how these two uses of the volume theorem lead to the applications; Table 2 lists the questions from the literature that are answered here.

Flowchart showing the volume theory and its two geometric uses

Figure 1. One volume theory, two geometric uses. Both branches use volume ratios through geometric integration and then the additional arguments inside their boxes. The labelled arrow is the proved fixed-radius local limit as λ→∞\lambda\to\infty. The two arrows into the bottom box show the endpoint surface and the separate coupling argument used for its Brownian-plane scaling limit. Part 3 uses no spectral theorem from Part 2. The all-degree coefficient theorem in Part 1 also uses harmonic comparison directly.

SourceQuestionAnswered in
Mirzakhani [144, §1.4]Vg,n/Vg−1,n+2V_{g,n}/V_{g-1,n+2} when n=n(g)→∞n=n(g)\to\inftyThm. 1.3, all nn
Mirzakhani [144, §1.4]adjacent coefficients at growing degreeThm. 1.4, all degrees
Shen–Wu [172, Question 3]does Vg,n2/(Vg,n−1Vg,n+1)→1V_{g,n}^{2}/(V_{g,n-1}V_{g,n+1})\to1?yes: Thm. 1.3
Shen–Wu [172, Question 2]uniform spectral gap when n/g→∞n/\sqrt{g}\to\infty?no: Cor. 14.1a^{a}
Budd–Lions [32, Open Question 1.6]tight spectrum at other weights, fixed countsThm. 14.8, Cor. 14.9
Hide–Thomas [98, §1.1]linear count of small residual eigenvaluesThm. 14.7
Hide–Thomas [98, Remark 1.8]spectral measures with many cuspsThms. 19.2–19.4b^{b}
Budd–Curien [31, §7.3]Brownian-plane limit of the infinite sphereThm. 21.1
Budd–Curien [31, §7.3]conjecture: the infinite sphere is parabolic, recurrent, with ergodic geodesic flowyes: Thm. 20.2, Cor. 20.3
Budd–Curien [31, §7.3]hyperbolic analogue of the infinite sphereThm. 20.1c^{c}

Table 2. Questions from the literature answered or treated in this paper.

The cusp-density parameter

All our formulas are expressed through a single real parameter. We define it first and then explain heuristically why it is the natural parameter; the proofs come later.

The genus-zero relation. Let

X(y)=yJ1(2y)=y−y22+y312−⋯ ,y0=j0,124,xc=X(y0),X(y)=\sqrt{y}J_1(2\sqrt{y})=y-\frac{y^2}{2}+\frac{y^3}{12}-\cdots,\qquad y_0=\frac{j_{0,1}^2}{4},\qquad x_c=X(y_0),

where j0,1j_{0,1} is the first positive zero of J0J_0. Since X′(y)=J0(2y)X'(y)=J_0(2\sqrt{y}), the function XX increases from 00 to xcx_c on [0,y0][0,y_0]; write Y(x)Y(x) for the inverse function. Consider the normalized cusp generating functions

Φg(x)=∑n≥0Vg,n(2π2)3g−3+nxnn!.(1)\Phi_g(x)=\sum_{n\geq0}\frac{V_{g,n}}{(2\pi^2)^{3g-3+n}}\frac{x^n}{n!}. \tag*{(1)}

with the convention that terms with 2g−2+n≤02g-2+n\leq0 are omitted. In genus zero, Φ0′′(x)=Y(x)\Phi_0''(x)=Y(x); this is the classical genus-zero computation (cf. [117, 136]), and in our normalization it follows from the string equation (see Step 1 of the proof of Theorem 14.2 in Appendix § 10.1). We call xx the cusp weight: it weights each cusp in {(1.1)}. The cusp weight xx and the parameter Y=Y(x)∈[0,y0)Y=Y(x)\in[0,y_0) determine each other, and we use whichever is more convenient.

The parameter functions. For 0≤Y<y00\leq Y<y_0, the entire function

KY(z)=∫01X′(Y+z2(1−s2)) dsK_Y(z)=\int_0^1 X'\left(Y+\frac{z}{2}(1-s^2)\right)\,ds

has only positive simple zeros (§ 3.1); it is the kernel of the recursion in § 3.4. Write its first zero as 2Δ(Y)2\Delta(Y) and put T(Y)=Y+Δ(Y)T(Y)=Y+\Delta(Y). The substitution v=T−Δs2v=T-\Delta s^2 shows that T=T(Y)T=T(Y) is the unique root in (y0,π2/4)(y_0,\pi^2/4) of

∫YTX′(v)T−v dv=0.(2)\int_Y^T\frac{X'(v)}{\sqrt{T-v}}\,dv=0. \tag*{(2)}

Finally define

I(Y)=∫YT(Y)X′(v)T(Y)−v dv,λ(Y)=2X(Y)Δ(Y)I(Y).(3)I(Y)=\int_Y^{T(Y)}X'(v)\sqrt{T(Y)-v}\,dv,\qquad\lambda(Y)=\frac{2X(Y)\sqrt{\Delta(Y)}}{I(Y)}. \tag*{(3)}

The next paragraph motivates these definitions. In § 3.3 we show that I>0I>0 and that λ\lambda is a strictly increasing bijection from [0,y0)[0,y_0) onto [0,∞)[0,\infty). At Y=0Y=0 one has Δ=T=π2/4\Delta=T=\pi^2/4, I=π/4I=\pi/4 and λ=0\lambda=0; as Y↑y0Y \uparrow y_0, Δ→0\Delta\to0 and λ→∞\lambda\to\infty.

Why λ\lambda is the cusp density. Let h=2g−1h=2g-1 and x=X(Y)x=X(Y). We shall prove (Proposition 7.1) that, uniformly for 0≤Y<y00 \le Y < y_0,

Φg′(x)=Γ(h)ϖ(Y)π4gπh(2I(Y))−h(1+O(log⁡gg)).(4)\Phi'_g(x)=\frac{\Gamma(h)\varpi(Y)}{\pi4^g\sqrt{\pi h}}\left(\sqrt{2}I(Y)\right)^{-h}\left(1+O\left(\frac{\log g}{g}\right)\right). \tag*{(4)}

where ϖ\varpi is the explicit positive function in (6) below. Thus the cusp weight changes the exponential growth only through the action I(Y)I(Y).

References

No references were extracted for this paper.

Paper details

vlnc:W_dftxg499
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

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