Averages of determinants of Laplacians over moduli spaces for large genus

Yunhui Wu; Yuxin He

Averages of determinants of Laplacians over moduli spaces for large genus

T0 review · 0 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper proves that on a Weil-Petersson random hyperbolic surface of large genus the normalized log determinant of the Laplacian concentrates around a universal constant E: the expected absolute deviation is O(g^{-δ}), the β-moments…

desk verdict A solid quantitative upgrade to Naud's convergence-in-probability result: the L^1 decay rate and the sharp moment threshold beta=2 are new, and the proof chain checks out. read the letter →

arxiv 2411.12971 v2 pith:TSK4MFYI submitted 2024-11-20 math.GT math.DGmath.SP

Yuxin He , Yunhui Wu This is my paper · ORCID

classification math.GTmath.DGmath.SP MSC 32G1558J5230F60

keywords regularizeddeterminantoftheLaplacianSelbergzetafunctionWeil-Peterssonrandomsurfaceslargegenusasymptoticsmodulispacehyperbolicclosedgeodesiccountingvolumes

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

On a random hyperbolic surface sampled from the Weil-Petersson measure, the regularized determinant of the Laplacian is studied as a function of the surface. The paper proves that the normalized quantity log det(Δ_X)/(4π(g−1)) concentrates around a universal constant E≈0.0538: the expected absolute deviation from E decays as a power of the genus. It further proves that the β-th moments converge to E^β for every β in (0,2), while for β≥2 the unnormalized moments diverge. These are quantitative upgrades of a previously known convergence-in-probability result, and they identify the threshold β=2 as the point where moment finiteness breaks down.

What carries the argument

The argument is carried by the Selberg trace formula identity log det Δ_X = 4π(g−1)E + γ0 − ∫$_0^{1}$ S_X(t)/t dt − ∫_1^∞ (S_X(t)−1)/t dt, where S_X(t) is a sum over closed geodesics weighted by length and length-squared Gaussian factors. The proof partitions S_X(t) into long, simple, and non-simple closed geodesic contributions. Long and simple contributions are controlled through a spectral gap bound and the integration formula for simple closed geodesic sums in terms of Weil-Petersson volumes; the non-simple contribution is controlled by a counting estimate for filling closed geodesics, those that cut the surface into simply connected or boundary-homotopic pieces, whose bound carries an exponential penalty for long subsurface boundaries. A good set of surfaces with a uniform spectral gap and few short geodesics is shown to contain almost all Weil-Petersson mass, so the three contributions can be integrated and then optimized against a cutoff R(g) chosen to grow logarithmically in g.

What would settle it

Compute, for growing genus, the Weil-Petersson average of the weighted count of those cutting geodesics that the proof needs to be polynomially small; if the average grows faster than any power of the genus, the main decay estimate would fail.

Watch

Extended reading notes

Core claim

The central discovery is a decay estimate for the average of the Selberg zeta derivative at s=1: there is a universal 0<δ<1 with E_WP[|log Z'_0(1)|/(4π(g−1))] = O($g^{{-δ}}$). Since log det(Δ_X)=4π(g−1)E + log Z'_0(1), this is equivalent to a statement about the mean absolute deviation of the normalized determinant from E. The same machinery gives the moment statement: for β∈(0,2), E_WP[|log det(Δ_X)/(4π(g−1))|^β] → E^β, while for β≥2 the integral of |log det(Δ_X)|^β over M_g is infinite. The threshold β=2 matches the moment behaviour of the reciprocal systole on the same probability space.

Load-bearing premise

The proof leans on the estimate that closed geodesics which cut a surface into very simple pieces, and which stay shorter than L, become exponentially rarer when the cut-out piece has a long boundary; if that exponential rarity failed, the dominant error term would not be polynomially small.

Editorial extensions

If this is right

Theorem 1 upgrades the earlier convergence-in-probability of the normalized Selberg zeta derivative to a rate that is uniform in genus.
For every β in (0,2), the normalized determinant has all β-th moments tending to E^β, so the distribution of log det(Δ_X)/(4π(g−1)) clusters around E without Gaussian-scale fluctuations.
The divergence of moments for β≥2 is sharp and comes from surfaces with a very short systole, matching the reciprocal-systole moment threshold.
The rate is produced by spectral gap and geodesic counting inputs, so any improvement in those inputs would directly improve the exponent δ.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A natural next question, not addressed here, is the optimal value of δ; the proof bounds δ by the spectral-gap exponent, so sharper spectral gap estimates would automatically sharpen the rate.
The same long/simple/non-simple decomposition of the geodesic heat sum should apply to other models of random hyperbolic surfaces, such as random covers, where analogous spectral and counting bounds are known; one would expect the same moment threshold β=2.
The divergence at β=2 hints at a heavy-tailed limiting law for the normalized determinant, with the second moment diverging at a definite rate such as log g; the paper neither confirms nor rules out such a rate.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Let me cut to it: this is a solid quantitative upgrade to Naud's convergence-in-probability theorem for the Laplacian determinant on Weil-Petersson random surfaces. The genuinely new content is Theorem 1, a polynomial L^1 decay rate for the normalized log Z'_0(1), and Theorem 2, which pins the moment threshold at beta=2: convergence for beta in (0,2) and divergence for beta>=2. That threshold, tied to systole fluctuations, is a nice new ingredient and not in Naud's earlier work.

The proof is coherent. The decomposition in (25) and the bounds (35), (36), (44), (45) fit together, and the endgame in Section 4.4 genuinely produces a positive delta for any a<8/9. I checked the non-simple geodesic contribution, the place I'd worry most, and the sign and size of the boundary-length suppression work; the splitting by |chi| covers all cases. Proposition 7 on the systole moment is clean and does what it needs to.

Soft spots are minor. First, the load-bearing imported estimates are the Wu-Xue filling count (Theorem 4) and the spectral gap bound (22). If any of those constants were wrong the exponent delta would shift, but that is an external verification matter, not a defect in this paper's argument. Second, Theorem 2 is stated and proved assuming Naud's convergence-in-probability result, with Theorem 1 later implying it; that ordering is slightly awkward but not circular since Theorem 2's proof never invokes Theorem 1. Third, the abstract says beta in [1,2) while Theorem 2 actually covers (0,2); a harmless wording difference.

If I had to assign one real risk, it is that the final delta is non-explicit because it depends on eta, alpha, and epsilons, but that is not a flaw, just a feature.

This paper is for people working on spectral theory of random hyperbolic surfaces. It deserves a serious referee; I'd send it out. It is not a framework-shifting paper, but it closes a real gap and the proof looks right.

Referee Report

0 major / 4 minor

Summary. The paper proves quantitative large-genus asymptotics for the Weil-Petersson average of the regularized Laplacian determinant. Theorem 1 states that the mean absolute deviation of log Z'_0(1)/(4π(g-1)), equivalently of log det(Δ_X)/(4π(g-1)) from the universal constant E, decays like O(g^{-δ}) for some uniform δ∈(0,1). Theorem 2 states that the L^β mean of log det(Δ_X)/(4π(g-1)) converges to E for every β∈(0,2), and that the unnormalized integral of |log det(Δ_X)|^β diverges for β≥2. The proofs combine the Selberg trace formula, Naud's concentration estimates, Wu–Xue spectral gap and filling-geodesic counting theorems, Mirzakhani's integration formula, and Weil–Petersson volume bounds.

Significance. If correct, Theorem 1 provides the first quantitative rate of convergence in Naud's concentration theorem for determinants on Weil–Petersson random surfaces, and Theorem 2 identifies the exact L^β threshold. The technical core—controlling non-simple closed geodesics via Wu–Xue filling counts—is applied with care, and the endgame transparently balances the various polynomial decays. The paper is clearly structured and makes its external dependencies explicit; the main estimates are imported from published work and are used in a way that is internally consistent. The result will be of interest to the random hyperbolic geometry and spectral theory communities.

minor comments (4)

[Section 2.1, Lemma 9] The definition of \tilde G(u) contains the factor e^{-1/(4t)}; however, the subsequent inequality e^{1/4}\tilde G(u) \ge \int_0^1 t^{-3/2} e^{-u^2/(4t)} dt is only correct if the factor is e^{-t/4}. Please correct the displayed definition of \tilde G(u); with e^{-t/4} the inequality follows from e^{-t/4} \ge e^{-1/4} on [0,1].
[Section 4.3.3, equations (38) and (45)] Theorem 4 is stated for 0<ε<1/2, but equation (38) fixes ε1∈(0,1) and equation (45) says "for any ε1>0". Since the argument only needs ε1 arbitrarily small, the inconsistency is harmless, but the stated ranges should be aligned.
[Section 2.2, Lemma 3 and Section 4.3.3] Lemma 3 is stated for closed hyperbolic surfaces, but in Section 4.3.3 it is applied to subsurfaces Y⊂X with geodesic boundary. The application is valid because every closed geodesic in Y is a closed geodesic in X, but this should be said explicitly to avoid confusion.
[Abstract and Theorem 2] The abstract restricts the L^β convergence claim to β∈[1,2), while Theorem 2 proves it for every β∈(0,2). Please make the abstract match the theorem, or explicitly note that the case β∈(0,1) is already due to Naud.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the new decay estimate is obtained by applying independent published counting and spectral-gap theorems to the Selberg zeta representation.

full rationale

The paper's central identity (1), imported from Sarnak and D'Hoker–Phong, expresses log det(Δ_X) as log Z'_0(1) plus a fixed universal constant E·4π(g−1); therefore the averaged deviation studied in Theorem 1 is exactly the averaged size of log Z'_0(1)/(4π(g−1)) up to that constant, and no quantity is fitted then renamed as a prediction. The proof of Theorem 1 in Section 4 uses the Selberg trace formula identity (20), the Wu–Xue spectral gap estimate (22), the Wu–Xue filling-geodesic counting theorem (Theorem 4), and Mirzakhani's integration formula; each is an independent published theorem whose assumptions do not include the determinant statement being proved. The coauthorship overlap on [31,32] is a self-citation, but the cited results concern spectral gaps and closed-geodesic counting, not Laplacian determinants, and they are externally established rather than assumed for the present target. Theorem 2 is proved conditionally on Naud's independent concentration theorem (2), not on Theorem 1, and the moment bound in Lemma 8 rests on independent spectral and systole estimates. The choice R(g)=a log g with a<8/9 is an analytic parameter choice, not a fit to determinant data. The only delicate point is whether the constants and exponents in [31,32] are exactly as quoted; that is an external verification matter, not circularity.

Assumptions & free parameters 2 free parameters · 6 assumptions · 0 invented entities

The central claims rest on no fitted empirical constants. The constant E is imported from the known determinant identity (1) of D'Hoker-Phong and Sarnak. The proof parameters listed above are arbitrary choices with constraints, not data fits. The main mathematical inputs are prior theorems: the Selberg trace formula, Mirzakhani's integration formula, Weil-Petersson volume asymptotics, and the Wu-Xue and Naud estimates. They are treated as axioms for this paper.

free parameters (2)

R(g)=a log g exponent a = unspecified; constrained to 0<a<8/9
Chosen in Section 4.4 to balance exponential terms in bound (48); no numerical value is needed because the theorem is existential in delta.
eta and alpha in good event A(g) = unspecified; eta in (0,3/16), alpha in (0,1)
Chosen in (21) to define the high-probability event; the exponent epsilon_0 depends on them, but the final claim is uniform in the resulting delta.

assumptions (6)

standard math Selberg trace formula and determinant identity (3), (14)
Used to express log det and log Z'_0 in terms of S_X(t); established in [2,4,5,21,25].
standard math Mirzakhani integration formula (Theorem 6)
Core tool for Weil-Petersson expectations; cited from [17].
standard math Weil-Petersson volume asymptotics (5), (6)
Imported from [17,18,22] and used in Proposition 7 and Section 4.3.2.
domain assumption Wu-Xue filling geodesic counting theorem (Theorem 4) and spectral gap bound (22)
Imported from [31,32]; the counting theorem is the load-bearing estimate for non-simple geodesics in Section 4.3.3.
domain assumption Naud's convergence-in-probability theorem (2) and moment estimates (10), (11)
Assumed in Section 3 for Theorem 2; Theorem 1 later implies (2), so the dependency is removable in the final theorem set.
standard math Collar lemma and the bound N_0^X(1) <= 3g-3
Used in Lemma 8 to control the geodesic counting term; standard facts from [4,11].

how reviews work

0 comments

Cite this review

Pith. "Pith review of Averages of determinants of Laplacians over moduli spaces for large genus." pith.science (2026). https://pith.science/paper/TSK4MFYI

@misc{pith2026241112971,
  author       = {Pith},
  title        = {Pith review of: Averages of determinants of Laplacians over moduli spaces for large genus},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TSK4MFYI}},
  note         = {Machine review of arXiv:2411.12971}
}

abstract

Let $\mathcal{M}_g$ be the moduli space of hyperbolic surfaces of genus $g$ endowed with the Weil-Petersson metric. We view the regularized determinant $\log \det(\Delta_{X})$ of Laplacian as a function on $\mathcal{M}_g$ and show that there exists a universal constant $E>0$ such that as $g\to \infty$, (1) the expected value of $\left|\frac{\log \det(\Delta_{X})}{4\pi(g-1)}-E \right|$ over $\mathcal{M}_g$ has rate of decay $g^{-\delta}$ for some uniform constant $\delta \in (0,1)$; (2) the expected value of $\left|\frac{\log \det(\Delta_{X})}{4\pi(g-1)}\right|^\beta$ over $\mathcal{M}_g$ approaches to $E^\beta$ whenever $\beta \in [1,2)$.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 29 canonical work pages

[1]

Friedman-Ramanuja n functions in ran- dom hyperbolic geometry and application to spectral gaps

Nalini Anantharaman and Laura Monk. Friedman-Ramanuja n functions in ran- dom hyperbolic geometry and application to spectral gaps. arXiv e-prints , page arXiv:2304.02678, April 2023

arXiv 2023
[2]

The spectrum of hyperbolic surfaces

Nicolas Bergeron. The spectrum of hyperbolic surfaces . Springer, 2016

work page 2016
[3]

Random construction of R iemann surfaces

Robert Brooks and Eran Makover. Random construction of R iemann surfaces. J. Diﬀerential Geom. , 68(1):121–157, 2004

work page 2004
[4]

Geometry and spectra of compact Riemann surfaces

Peter Buser. Geometry and spectra of compact Riemann surfaces . Springer Science & Business Media, 2010

2010
[5]

Eric D’Hoker and D. H. Phong. On determinants of Laplacia ns on Riemann surfaces. Comm. Math. Phys. , 104(4):537–545, 1986

work page 1986
[6]

Short ge- odesic loops and Lp norms of eigenfunctions on large genus random surfaces

Cliﬀord Gilmore, Etienne Le Masson, Tuomas Sahlsten, an d Joe Thomas. Short ge- odesic loops and Lp norms of eigenfunctions on large genus random surfaces. Geom. Funct. Anal., 31(1):62–110, 2021

work page 2021
[7]

Spectral distribution of twisted Laplacian on typical hyperbolic surfaces of high genus

Yulin Gong. Spectral distribution of twisted Laplacian on typical hyperbolic surfaces of high genus. Comm. Math. Phys. , 405(7):Paper No. 158, 41, 2024

work page 2024
[8]

The Selberg Trace Formula for PSL(2, R): Volume 2 , volume 1001

Dennis A Hejhal. The Selberg Trace Formula for PSL(2, R): Volume 2 , volume 1001. Springer, 2006

work page 2006

Show all 32 references

[9]

Spectral gap for weil–petersson random surfa ces with cusps

Will Hide. Spectral gap for weil–petersson random surfa ces with cusps. Int. Math. Res. Not. IMRN , 2023(20):17411–17460, 2023

2023
[10]

Short geodesics and small eige nvalues on random hyper- bolic punctured spheres

Will Hide and Joe Thomas. Short geodesics and small eige nvalues on random hyper- bolic punctured spheres. Comment. Math. Helv. , page to appear, 2022

2022
[11]

Collars on riemann surfaces

Linda Keen. Collars on riemann surfaces. In Discontinuous groups and Riemann sur- faces (Proc. Conf., Univ. Maryland, College Park, Md., 1973 ), pages 263–268, 1974

1973
[12]

Quantum ergodic ity for eisenstein series on hyperbolic surfaces of large genus

Etienne Le Masson and Tuomas Sahlsten. Quantum ergodic ity for eisenstein series on hyperbolic surfaces of large genus. Math. Ann. , 389(1):845–898, 2024

2024
[13]

Towards optimal spe ctral gaps in large genus

Michael Lipnowski and Alex Wright. Towards optimal spe ctral gaps in large genus. Ann. Probab., 52(2):545–575, 2024

2024
[14]

Explicit spectral gaps for random covers of Rie- mann surfaces

Michael Magee and Fr´ ed´ eric Naud. Explicit spectral gaps for random covers of Rie- mann surfaces. Publ. Math. Inst. Hautes ´Etudes Sci. , 132:137–179, 2020

2020
[15]

A rand om cover of a compact hyperbolic surface has relative spectral gap 3 16 − ε

Michael Magee, Fr´ ed´ eric Naud, and Doron Puder. A rand om cover of a compact hyperbolic surface has relative spectral gap 3 16 − ε. Geom. Funct. Anal. , 32(3):595– 661, 2022

2022
[16]

Simple geodesics and Weil-Peterss on volumes of moduli spaces of bordered Riemann surfaces

Maryam Mirzakhani. Simple geodesics and Weil-Peterss on volumes of moduli spaces of bordered Riemann surfaces. Invent. Math. , 167(1):179–222, 2007

2007
[17]

Growth of weil-petersson volumes a nd random hyperbolic sur- face of large genus

Maryam Mirzakhani. Growth of weil-petersson volumes a nd random hyperbolic sur- face of large genus. Journal of Diﬀerential Geometry , 94(2):267–300, 2013

2013
[18]

Lengths of closed geo desics on random surfaces of large genus

Maryam Mirzakhani and Bram Petri. Lengths of closed geo desics on random surfaces of large genus. Comment. Math. Helv. , 94(4):869–889, 2019

2019
[19]

Benjamini-Schramm convergence and spectr a of random hyperbolic surfaces of high genus

Laura Monk. Benjamini-Schramm convergence and spectr a of random hyperbolic surfaces of high genus. Anal. PDE , 15(3):727–752, 2022. 20 YUXIN HE AND YUNHUI WU

2022
[20]

Spectral convergence of the D irac operator on typical hyperbolic surfaces of high genus

Laura Monk and Rares Stan. Spectral convergence of the D irac operator on typical hyperbolic surfaces of high genus. Annales Henri Poincare, to appear, July 2023

2023
[21]

Determinants of laplacians on random hyperbolic surfaces

Fr´ ed´ eric Naud. Determinants of laplacians on random hyperbolic surfaces. Journal d’Analyse Math´ ematique, 151:265–291, 2023

2023
[22]

Large genus asymptoti cs for lengths of separat- ing closed geodesics on random surfaces

Xin Nie, Yunhui Wu, and Yuhao Xue. Large genus asymptoti cs for lengths of separat- ing closed geodesics on random surfaces. Journal of Topology , 16(1):106–175, 2023

2023
[23]

GOE statistics on the moduli space of sur faces of large genus

Ze´ ev Rudnick. GOE statistics on the moduli space of sur faces of large genus. Geom. Funct. Anal., 33(6):1581–1607, 2023

2023
[24]

On the central limit theo rem for linear eigenvalue statistics on random surfaces of large genus

Ze´ ev Rudnick and Igor Wigman. On the central limit theo rem for linear eigenvalue statistics on random surfaces of large genus. J. Anal. Math. , 151(1):293–302, 2023

2023
[25]

Determinants of laplacians

Peter Sarnak. Determinants of laplacians. Communications in mathematical physics , 110(1):113–120, 1987

1987
[26]

Harmonic analysis and discontinuous gro ups in weakly symmetric spaces with applications to dirichlet series

Atle Selberg. Harmonic analysis and discontinuous gro ups in weakly symmetric spaces with applications to dirichlet series. J. Indian Math. Soc. , 20:47–87, 1956

1956
[27]

Arbitrarily small spectral gap s for random hyperbolic surfaces with many cusps

Yang Shen and Yunhui Wu. Arbitrarily small spectral gap s for random hyperbolic surfaces with many cusps. arXiv e-prints , page arXiv:2203.15681, March 2022

2022 arXiv
[28]

The Fenchel-Nielsen deformation

Scott Wolpert. The Fenchel-Nielsen deformation. Ann. of Math. (2) , 115(3):501–528, 1982

1982
[29]

Scott A. Wolpert. Asymptotics of the spectrum and the Se lberg zeta function on the space of Riemann surfaces. Comm. Math. Phys. , 112(2):283–315, 1987

1987
[30]

Asymptotics of the spectrum and the sel berg zeta function on the space of riemann surfaces

Scott A Wolpert. Asymptotics of the spectrum and the sel berg zeta function on the space of riemann surfaces. Communications in mathematical physics , 112(2):283–315, 1987

1987
[31]

Prime geodesic theorem and clos ed geodesics for large genus

Yunhui Wu and Yuhao Xue. Prime geodesic theorem and clos ed geodesics for large genus. J. Eur. Math. Soc. (JEMS), to appear, 2022

2022
[32]

Random hyperbolic surfaces of l arge genus have ﬁrst eigenvalues greater than 3 16 − ǫ

Yunhui Wu and Yuhao Xue. Random hyperbolic surfaces of l arge genus have ﬁrst eigenvalues greater than 3 16 − ǫ. Geometric and Functional Analysis , 32(2):340–410, 2022. Yau Mathematical Sciences Center and Department of Mathema tical Sci- ences, Tsinghua University, Beijing, ...

2022

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

Friedman-Ramanuja n functions in ran- dom hyperbolic geometry and application to spectral gaps

Nalini Anantharaman and Laura Monk. Friedman-Ramanuja n functions in ran- dom hyperbolic geometry and application to spectral gaps. arXiv e-prints , page arXiv:2304.02678, April 2023

arXiv 2023

[2] [2]

The spectrum of hyperbolic surfaces

Nicolas Bergeron. The spectrum of hyperbolic surfaces . Springer, 2016

work page 2016

[3] [3]

Random construction of R iemann surfaces

Robert Brooks and Eran Makover. Random construction of R iemann surfaces. J. Diﬀerential Geom. , 68(1):121–157, 2004

work page 2004

[4] [4]

Geometry and spectra of compact Riemann surfaces

Peter Buser. Geometry and spectra of compact Riemann surfaces . Springer Science & Business Media, 2010

2010

[5] [5]

Eric D’Hoker and D. H. Phong. On determinants of Laplacia ns on Riemann surfaces. Comm. Math. Phys. , 104(4):537–545, 1986

work page 1986

[6] [6]

Short ge- odesic loops and Lp norms of eigenfunctions on large genus random surfaces

Cliﬀord Gilmore, Etienne Le Masson, Tuomas Sahlsten, an d Joe Thomas. Short ge- odesic loops and Lp norms of eigenfunctions on large genus random surfaces. Geom. Funct. Anal., 31(1):62–110, 2021

work page 2021

[7] [7]

Spectral distribution of twisted Laplacian on typical hyperbolic surfaces of high genus

Yulin Gong. Spectral distribution of twisted Laplacian on typical hyperbolic surfaces of high genus. Comm. Math. Phys. , 405(7):Paper No. 158, 41, 2024

work page 2024

[8] [8]

The Selberg Trace Formula for PSL(2, R): Volume 2 , volume 1001

Dennis A Hejhal. The Selberg Trace Formula for PSL(2, R): Volume 2 , volume 1001. Springer, 2006

work page 2006