{"id":"799cbac0-b5df-4d65-bb0b-44ce72ce22f0","arxiv_id":"2411.12971","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"For Weil-Petersson random hyperbolic surfaces of large genus, the normalized log-determinant of the Laplacian converges in L^1 with polynomial rate, and its beta-moments converge to E^beta for beta below 2 but diverge for beta at least 2.","lead":"The paper proves a quantitative concentration statement for the log-determinant of the Laplacian on random hyperbolic surfaces of large genus. This gives a polynomial decay rate and moment convergence, sharpening earlier work of Naud.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the proof is internally consistent, and its main external dependency (the Wu–Xue filling-count estimates used in §4.3.3) appears to be applied correctly.","rationale":"I read the paper as a chain of deterministic inequalities with imported probabilistic inputs. The only place where the final polynomial rate could fail is Section 4.3.3, because term (III) after division by (g-1) must decay; the long and simple parts are safe. The non-simple part relies on a filling-geodesic count with exponential suppression in the boundary length. Tracing (37)–(44), the grouping by Euler characteristic is exhaustive, the use of Theorem 4 is consistent, and the endgame choice R=a log g, a<8/9, yields a positive δ. I also checked the moment arguments in Theorem 2 and Proposition 7; the threshold β=2 is exactly where E[sys^{-(β+2)/2}] fails, and Lemma 9 converts systole divergence into log-det divergence. The paper explicitly states its dependence on Naud's theorem (2) and on [22,32], which is a limitation but not a hidden one. Minor issues: ε1 range in (38) should be (0,1/2) to match Theorem 4, and the abstract's β∈[1,2) versus theorem's (0,2) is harmless. Overall, I find no concrete failure mode that would change the ACCEPT verdict.","tokens_in":15200,"tokens_out":36658,"duration_ms":394316,"concrete_test":"Re-derive equations (38)–(44) using the verbatim statements of [31, Theorem 18], [32, Theorem 4], and [32, Props 33 and 35]. Specifically, verify that Theorem 4 has exactly the factor ((1-ε)/2) on ℓ(∂Y) and that the averaged bound (39) carries the same weight e^{-(1-ε1)/2ℓ(∂Y)} with rate L^{66}e^{L/2+ε1L}/g. If these match, substitute into (45) with R=a log g to recover the g^{9a/4-2+o(1)} term and the condition a<8/9; if the boundary-length suppression is weaker or has a different coefficient, recompute the allowed range of a and check whether δ remains positive.","verdict_should_be":"UNCHANGED","load_bearing_attack":"No significant objection identified. The central estimate that could break the polynomial rate in Theorem 1 is the non-simple geodesic contribution to term (III), controlled by Theorem 4 and the averaged bounds (39), (42). I checked the chain (37)–(44) against the stated assumptions: the boundary-length suppression has the correct sign, the topology splitting into |χ|≤16 and 17≤|χ|≤[4L_g/2π] covers all possible subsurfaces by Gauss–Bonnet, and the exponent R/4+(1/2+ε1)L_g with L_g≈4R gives the announced g^{9a/4-2} term. The endgame therefore closes for any a<8/9. The only genuine risk is that the quoted constants/exponents from [31,32] are not exactly as stated; this is an external verification issue, not a defect in the paper's argument. A minor wording issue is ε1∈(0,1) in (38) versus ε1∈(0,1/2) in Theorem 4; since ε1 can be chosen small, this is harmless.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","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.","tokens_in":15411,"tokens_out":31688,"duration_ms":284145,"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.","major_comments":[],"minor_comments":[{"comment":"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":"Section 2.1, Lemma 9"},{"comment":"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":"Section 4.3.3, equations (38) and (45)"},{"comment":"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.","section":"Section 2.2, Lemma 3 and Section 4.3.3"},{"comment":"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.","section":"Abstract and Theorem 2"}],"recommendation":"minor_revision","confidential_remarks":"The paper relies heavily on two papers by the second author (Wu–Xue), but those are published independent results and are used as black boxes; I see no circularity or novelty-disclosure concern. The main mathematical argument appears sound, and the issues I found are local or typographical."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","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.\n\nThe 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.\n\nSoft 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.\n\nIf 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.\n\nThis 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.","headline":"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.","tokens_in":15983,"tokens_out":1901,"would_cite":true,"duration_ms":18272,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["32G15","58J52","30F60"],"pacs":[],"model":"deepseek-v4-flash","headline":"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…","keywords":["regularized determinant of the Laplacian","Selberg zeta function","Weil-Petersson random surfaces","large genus asymptotics","moduli space of hyperbolic surfaces","closed geodesic counting","Weil-Petersson volumes"],"falsifier":"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.","tokens_in":14962,"feed_emoji":"📐","tokens_out":11414,"duration_ms":104409,"temperature":0.7,"pith_summary":"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.","feed_headline":"As genus grows, Laplacian log-determinant deviation decays like g^-δ","feed_subtitle":"It upgrades known convergence in probability to a quantitative rate and locates the moment threshold at β=2.","key_machinery":"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.","core_discovery":"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.","pith_inferences":["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."],"forward_implications":["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."],"supporting_citations":[{"why":"Supplies the convergence-in-probability baseline and the moment bounds on |log det(Δ_X)| that are used to control the complement of the good set and the moment growth.","marker":"[21]"},{"why":"Supplies the filling closed geodesic counting theorem and the spectral gap decay estimate; these are the main estimates that make the good set have measure 1 - O(g^{-ε0}).","marker":"[32]"},{"why":"Provides an alternate formulation of the filling geodesic bound and the spectral gap estimate used in the proof.","marker":"[31]"},{"why":"Provides the integration formula converting simple closed geodesic sums into Weil-Petersson volume integrals, used for the simple-geodesic term and for the systole moment estimate.","marker":"[17]"},{"why":"Establishes the determinant identity linking det(Δ_X) to Z'_0(1) and the universal constant E.","marker":"[5]"},{"why":"Gives the same determinant identity and the value of E, the target of concentration.","marker":"[25]"},{"why":"Supplies the elementary bound on non-iterated closed geodesics and the Collar Lemma used to count short geodesics.","marker":"[4]"},{"why":"Provides the Weil-Petersson volume estimates used in the integration over moduli space and in the systole moment computation.","marker":"[22]"}],"fun_headline_variants":["Laplacian determinant deviation decays like g^-δ on moduli space","Universal decay rate for Laplacian determinant averages on moduli space","Mean absolute deviation of log-determinant decays as genus grows","Moment threshold at β=2 for normalized Laplacian determinant","Log-determinant deviation on moduli space shrinks like g^-δ"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"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.","fun_headline_variants_meta":{"raw":{"variants":["Laplacian determinant deviation decays like g^-δ on moduli space","Universal decay rate for Laplacian determinant averages on moduli space","Mean absolute deviation of log-determinant decays as genus grows","Moment threshold at β=2 for normalized Laplacian determinant","Log-determinant deviation on moduli space shrinks like g^-δ"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001298,"raw_usage":{"total_tokens":5276,"prompt_tokens":904,"completion_tokens":4372,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":520,"completion_tokens_details":{"reasoning_tokens":4279}},"tokens_in":520,"tokens_out":4372,"duration_ms":31691,"temperature":1.0,"reasoning_tokens":4279,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T17:00:33.485064+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"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.","supporting_citations":[{"cited_title":"Determinants of laplacians on random hyperbolic surfaces","cited_arxiv_id":null,"evidence_quote":"Supplies the convergence-in-probability baseline and the moment bounds on |log det(Δ_X)| that are used to control the complement of the good set and the moment growth."},{"cited_title":"Random hyperbolic surfaces of l arge genus have first eigenvalues greater than 3 16 − ǫ","cited_arxiv_id":null,"evidence_quote":"Supplies the filling closed geodesic counting theorem and the spectral gap decay estimate; these are the main estimates that make the good set have measure 1 - O(g^{-ε0})."},{"cited_title":"Prime geodesic theorem and clos ed geodesics for large genus","cited_arxiv_id":null,"evidence_quote":"Provides an alternate formulation of the filling geodesic bound and the spectral gap estimate used in the proof."},{"cited_title":"Growth of weil-petersson volumes a nd random hyperbolic sur- face of large genus","cited_arxiv_id":null,"evidence_quote":"Provides the integration formula converting simple closed geodesic sums into Weil-Petersson volume integrals, used for the simple-geodesic term and for the systole moment estimate."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the determinant identity linking det(Δ_X) to Z'_0(1) and the universal constant E."},{"cited_title":"Determinants of laplacians","cited_arxiv_id":null,"evidence_quote":"Gives the same determinant identity and the value of E, the target of concentration."},{"cited_title":"Large genus asymptoti cs for lengths of separat- ing closed geodesics on random surfaces","cited_arxiv_id":null,"evidence_quote":"Provides the Weil-Petersson volume estimates used in the integration over moduli space and in the systole moment computation."}],"review_version":1}