{"id":"26a562db-c73c-41ac-87ae-521f7e11fd36","arxiv_id":"2607.13536","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"The order of the rational cuspidal class group of X_1(N) is given by an explicit product over even Dirichlet characters involving generalized Bernoulli numbers B_{2,χ}, valid for all N ≥ 5.","lead":"This paper proves an explicit formula for the order of the rational cuspidal class group of the modular curve X_1(N) for every N ≥ 5. The formula is a product of generalized Bernoulli numbers and simple arithmetic factors, and the proof connects a classical torsion problem to a determinant of Bernoulli polynomials.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Main risk is Proposition 5.6's dependence on a precise reading of Streng's [30, Thm 1.2]; a misstated congruence lattice would change (div(S):div(U)) and hence Theorem 1.1.","rationale":"The reader's weakest assumption correctly identifies the use of Streng's theorem as the most load-bearing external dependency. I reviewed the main internal steps: the Bernoulli-transform determinant computation in Sections 2–4, the divisor-matrix/index computation in Proposition 5.4, the use of Streng in Proposition 5.6, and the sign-positivity argument in the proof of Theorem 1.1. I spot-checked N=5 through both the determinant formula and the final cuspidal-class-group formula, and the constants, Bernoulli numbers, Euler factors, and sign all match. The only point where the argument could silently fail is if the quoted generator/congruence statement from Streng is not exactly as used; that would propagate directly to Theorem 1.1. Since this is a verifiable transcription risk rather than a detected flaw, and since Streng's result is published and peer-reviewed, the reader's ACCEPT verdict should stand unless the proposed check reveals a discrepancy.","tokens_in":16422,"tokens_out":35982,"duration_ms":319269,"concrete_test":"Locate [30, Theorem 1.2] and compare its exact statement with the version used in Proposition 5.6: (a) the list of generators H_1,...,H_⌊N/2⌋; (b) the two congruences, especially whether the second modulus is gcd(2,N)N or just N; (c) the claim that the representation u=c·H_1^{e_1}...H_r^{e_r} is unique modulo constants. Then recompute (S:U)=|im ψ| for N=4,5,6,8 and confirm it equals 12 gcd(2,N)N; if the modulus or generator set differs, recompute Proposition 5.6 and the final formula.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 1.1 is assembled as (Div^0(C):div(S))·(div(S):div(U)) in Eq. (15). The second factor is fixed in Proposition 5.6 as 12 gcd(2,N)N by invoking Streng's [30, Theorem 1.2]: every modular unit on the Γ_1(N)-model is uniquely c·H_1^{e_1}...H_r^{e_r} with exponents satisfying ∑e_a ≡ 0 mod 12 and ∑a^2 e_a ≡ 0 mod gcd(2,N)N. This is the only non-elementary external input in the proof of that index; if the second modulus, the generator set, or the stated 'unique product plus congruences' description is not accurately transcribed, then U has a different index in S and the final formula changes. The same proof also relies on the author's preprint [6] for Lemmas 5.2 and 5.3, but those lemmas are stated and sketched in the paper, while the Streng theorem is quoted without proof. I checked the Bernoulli-determinant computation (Theorem 4.1) and the N=5 arithmetic and found no internal inconsistency; the residual risk is concentrated in the faithful transcription of an external theorem.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper gives a closed formula for the order of the rational cuspidal class group C^Q_1(N) of X_1(N) for every N ≥ 5. The proof proceeds in two main steps: first, an independent evaluation of the Bernoulli determinant D_{k,N} (Theorem 4.1), obtained by decomposing the Bernoulli transform into Dirichlet-character eigenspaces and factoring the resulting matrix over prime powers; second, a geometric computation of the two indices in Eq. (15): the index of the Siegel-unit lattice in the rational cusp-divisor group (Proposition 5.4) and the index of the full modular-unit lattice in that Siegel-unit lattice (Proposition 5.6), the latter using Streng's theorem on generators of modular units on X_1(N). The paper also defines a speculative higher-weight analogue of the cuspidal class group and conjecturally relates its order to D_{k,N}.","tokens_in":16764,"tokens_out":38011,"duration_ms":348964,"significance":"If the proof is accepted, Theorem 1.1 is the first uniform formula for the rational cuspidal class group of X_1(N) for composite N, covering and unifying earlier prime-level and twice-prime-level results. The determinant evaluation in Theorem 4.1 is a useful result in its own right, and the paper connects it both to known formulas for cuspidal subgroups and to open questions involving Eisenstein symbols. The argument is largely explicit and checkable: the determinant computation is written out, the N=5 case gives |C^Q_1(5)|=1 as required by genus 0, and the theorem specializes correctly to the known formula for X_1(p). These cross-checks substantially increase confidence in the central claim.","major_comments":[{"comment":"The factor (div(S):div(U)) = 12 gcd(2,N)N is load-bearing for Theorem 1.1, and its proof rests entirely on the exact form of Streng's theorem [30, Theorem 1.2], including the generator set H_1,...,H_r and the two congruences modulo 12 and gcd(2,N)N. The theorem is not stated in the paper, and the promised direct check for N=4,5 is omitted. Since a mis-transcription of the second modulus or of the generator set would change this index and hence the final formula, please state Streng's theorem explicitly in the form used, and include the N=4,5 check for surjectivity of the map ψ.","section":"§5.6, Eq. (15)"},{"comment":"Proposition 5.4 depends on the parametrization of Galois orbits of cusps (Lemma 5.2) and on the divisor formula for the Siegel units (Lemma 5.3). Both proofs are largely delegated to the author's preprint [6] (Lemmas 5.9 and 5.11), and the exceptional case N=4 in Lemma 5.3 is asserted without calculation. Since these lemmas feed directly into the determinant matrix and the degree computation, an error in any of these statements would propagate to Theorem 1.1. Please either include full proofs or state these as self-contained lemmas with enough detail to verify the exceptional cases.","section":"§5, Lemmas 5.2–5.3 and Prop. 5.4"}],"minor_comments":[{"comment":"The proof says the basis is given by [i]+[-i] for 0≤i≤⌊N/2⌋. For i=0 (and for i=N/2 when N is even) this vector is twice the characteristic function of the corresponding orbit, which affects the coefficient of the row in the displayed matrix. The determinant statement is correct, but the row-scaling discussion should be worded more carefully to avoid confusion.","section":"§2, Lemma 2.1"},{"comment":"The displayed constant in Eq. (16) is difficult to read in the typeset version (the factors involving 2, 3, 5 and the powers of gcd(2,N) are hard to disentangle). Please re-typeset all powers explicitly.","section":"Eq. (16)"},{"comment":"The higher-weight cuspidal class group is introduced as an image in a higher Chow group, but finiteness is only conjectural and the relation to D_{k,N} is explicitly speculative. The section would be clearer if the statement were labelled as a conjecture rather than presented as an expected theorem.","section":"§6, Definition 6.1"}],"recommendation":"major_revision","confidential_remarks":"The central formula appears sound, and I found no internal inconsistency in the determinant computation or in the numerical cross-checks. My recommendation of major revision, rather than acceptance, is driven by the fact that the proof of Theorem 1.1 hinges on Proposition 5.6, whose exact congruence lattice is quoted from an external theorem without being stated. This is readily fixable: reproducing Streng's theorem and the N=4,5 checks would remove the main residual risk. I do not see grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe main thing you should know: this paper proves a fully explicit formula for |C^Q_1(N)|, the rational cuspidal class group of X_1(N), for every N ≥ 5. That's new for arbitrary N; earlier work covered primes, prime powers, and N=2p. The formula is a product of Bernoulli numbers and Euler factors, and it matches known special cases. I spot-checked N=5 and it works.\n\nWhat's genuinely good: the Bernoulli determinant computation (Theorem 4.1) is elegant and self-contained. The paper views D_{k,N} as a group determinant, decomposes by Dirichlet characters, and reduces to small matrices at each prime. That part is carefully done. The application to the cuspidal class group is a clever use of Streng's generators of modular units. The paper is honest about overlap: Liu [21] computed the same determinant family by different methods, and for prime N the final formula is equivalent to known results.\n\nSoft spots, in proportion: the main risk is Proposition 5.6, where the index (div(S):div(U)) is taken from Streng's [30, Theorem 1.2]. The theorem is quoted as: every modular unit is uniquely c·H_1^{e_1}...H_r^{e_r} with exponents satisfying two congruences mod 12 and mod gcd(2,N)N. If those moduli or the generator set are misstated, the factor 12 gcd(2,N)N changes and Theorem 1.1 collapses. I haven't checked Streng's paper line by line, but the exposition suggests the author read it carefully. Still, that is the one non-elementary external input, and it is load-bearing. Also, two auxiliary lemmas (5.2 and 5.3) come from the author's own preprint [6]; they are stated with sketches, so less concerning.\n\nMinor caveats: determinant formulas have ± signs, but the sign is resolved. Surjectivity in Prop 5.6 for N=4,5 is asserted without details; that's trivial to verify. The higher-weight speculations in Section 6 are explicitly speculative; they don't affect the main theorem.\n\nOverall: the central argument holds up. This deserves a serious referee. I'd send it to review, with particular attention to the transcription of Streng's theorem and the two lemmas from [6]. The paper is a solid contribution to the study of cuspidal subgroups, and the determinant formula may be useful beyond this context.\n\nRecommendation: engage with it. Have a referee check the index computation carefully.","headline":"New explicit formula for the rational cuspidal class group of X_1(N) for all N; proof is sound but hinges on a precise reading of Streng's theorem.","tokens_in":17289,"tokens_out":1927,"would_cite":true,"duration_ms":18821,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11G18","11G16","14G35","11B68"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves an explicit formula for the order of the rational cuspidal class group of the modular curve X_1(N) for every integer N ≥ 5, built from a determinant of second-Bernoulli-polynomial values.","keywords":["modular curves","cuspidal subgroup","rational cuspidal class group","modular units","Bernoulli polynomials","Bernoulli determinant","Dirichlet characters","Siegel functions"],"falsifier":"Compute the right-hand side of Theorem 1.1 for a concrete composite N (e.g., N = 12 or N = 15) and compare it with a direct computation of the order of C^Q_1(N) obtained by writing down the divisor matrix of the Siegel units, reducing modulo principal divisors, and taking Galois invariants; a single mismatch would refute the theorem. A cheaper check is to verify Theorem 4.1 numerically for D_{2,11} or D_{3,11} with a computer algebra system, since the cuspidal-group proof depends on that determinant evaluation.","tokens_in":16313,"feed_emoji":"🧮","tokens_out":8934,"duration_ms":77377,"temperature":0.7,"pith_summary":"The paper proves a closed formula for the order of the rational cuspidal class group of the modular curve X_1(N) — the subgroup of the Jacobian generated by Galois-invariant degree-zero cuspidal divisors — for every integer N ≥ 5. The formula displays a product over even primitive Dirichlet characters of generalized Bernoulli numbers B_{2,χ}, multiplied by local Euler factors at primes dividing N and a prefactor built from divisors of N. The proof reduces the group-order computation to evaluating a determinant D_{2,N} whose entries are values of the second Bernoulli polynomial at fractional parts, and the same technique yields an explicit non-vanishing formula for the whole family D_{k,N} for every k ≥ 2. The result matters because it replaces a patchwork of known special cases (prime powers, twice a prime) with one uniform expression, and it gives arithmetic meaning to a family of classical determinants. The paper also defines a higher-weight analogue of the cuspidal class group and conjectures its order is governed by the analogous determinant D_{k,N}.","feed_headline":"Explicit formula gives size of cuspidal group for every N","feed_subtitle":"A closed product of Bernoulli numbers and local prime factors now gives the order for every N ≥ 5.","key_machinery":"The engine of the paper is the Bernoulli transform B_{k,N}: f ↦ Σ_y B_k({xy/N}) f(y). Its determinant on the even (resp. odd) subspace of functions on Z/NZ is 2^{⌊(N−1)/2⌋} D_{k,N}, so evaluating the determinants D_{k,N} reduces to diagonalizing the transform. The transform commutes with the action of (Z/NZ)^×, hence splits into Dirichlet-character eigenspaces V_{N,χ}; on each eigenspace the matrix becomes a product of local matrices over the primes dividing N, whose determinants are computed by induction. On the arithmetic side, the index computation rests on Siegel functions: the divisors of g_{0,a} are read off from B_2({ak/N}), and Streng's theorem describing the full group of modular un","core_discovery":"The central claim, Theorem 1.1, is that for every integer N ≥ 5, |C^Q_1(N)| equals gcd(2,N)^2 · 9/(2^{N−5} N) · ∏_{d|N, d≥3} d^{φ(d)/2} · ∏_{p|N} p^2/(p^2−1) times the product over even primitive Dirichlet characters χ of conductor dividing N of B_{2,χ}^{σ_0(N/Nχ)} (N/Nχ)^{(1/2)σ_0(N/Nχ)} · ∏_{p|N, p∤Nχ} (1−χ(p)p^{−2})^{v_p(N)σ_0(N^{(p)}/Nχ)}. The proof goes by computing the cokernel of the divisor map from the group of modular units to the degree-zero cuspidal divisors: it first expresses the index of the subgroup generated by Siegel units as |D_{2,N}| up to explicit simple factors, and then uses Streng's theorem to compute the index of the full modular-unit group inside that subgroup as 12","pith_inferences":["The same determinant-eigenspace method could be applied to the full cuspidal subgroup C_1(N) (not just its Galois-invariant part); a formula there would be a major step toward the conjecture that C^Q_1(N) equals the rational torsion of J_1(N), which the paper leaves open.","The higher-weight cuspidal group introduced in Section 6 can be tested numerically for small N: for an elliptic curve with a rational N-torsion point, the residue map on the tame K_2 group should be controlled by D_{3,N}, so a computer calculation for N = 7 could confirm or refute the speculated link.","The factorization over primes suggests that the cuspidal class group itself (not just its order) may decompose according to the prime factors of N; the paper does not establish this, but the formula for the determinant is the natural first step toward such a structural theorem.","Because D_{k,N} is a 'paratrophic' determinant in the sense of Frobenius, the explicit formula likely extends to other congruence subgroups by replacing the character set P^{(-1)^k}_N with the corresponding set; this would give analogous cuspidal formulas for X_0(N) or X(N)."],"forward_implications":["For every N ≥ 5, the order of C^Q_1(N) is computable in closed form from Dirichlet L-values and elementary factors; no case-by-case modular computation is needed.","Theorem 4.1 proves all determinants D_{k,N} are non-zero, giving a uniform proof of independence for the Siegel functions and for the K_2 elements whose non-vanishing was previously checked only for N = 7, 8, 10.","When N is prime, the general formula reduces to the known expression |C^Q_1(p)| = p^{(p−1)/2} / 2^{p−3} ∏_{χ ≠ 1 even} B_{2,χ}, recovering previous results in one stroke.","The factorization of the determinant over the primes dividing N yields an Euler-product-style expression for the cuspidal group order, with each prime contributing a local factor.","The non-vanishing of D_{2,N} combined with Proposition 5.4 gives an independent proof that the Siegel functions g_{0,a} with 1 ≤ a ≤ ⌊N/2⌋ form a Q-basis of U_Q."],"fun_headline_variants":["Formula for rational cuspidal class group order","Bernoulli determinants fix cuspidal group size","Exact order of cuspidal group via Bernoulli numbers","New formula: cuspidal class group order for all N","Bernoulli trick computes cuspidal group order"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The formula's one external load-bearing input is Streng's theorem describing every modular unit on X_1(N) as a unique product of specified Siegel functions whose exponents satisfy two congruences; if that description is inaccurate, the index (div(S):div(U)) = 12 gcd(2,N)N would change and the final formula would fail.","fun_headline_variants_meta":{"raw":{"variants":["Formula for rational cuspidal class group order","Bernoulli determinants fix cuspidal group size","Exact order of cuspidal group via Bernoulli numbers","New formula: cuspidal class group order for all N","Bernoulli trick computes cuspidal group order"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00017,"raw_usage":{"total_tokens":1084,"prompt_tokens":700,"completion_tokens":384,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":444,"completion_tokens_details":{"reasoning_tokens":303}},"tokens_in":444,"tokens_out":384,"duration_ms":3757,"temperature":1.0,"reasoning_tokens":303,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T04:55:43.027116+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the right-hand side of Theorem 1.1 for a concrete composite N (e.g., N = 12 or N = 15) and compare it with a direct computation of the order of C^Q_1(N) obtained by writing down the divisor matrix of the Siegel units, reducing modulo principal divisors, and taking Galois invariants; a single mismatch would refute the theorem. A cheaper check is to verify Theorem 4.1 numerically for D_{2,11} or D_{3,11} with a computer algebra system, since the cuspidal-group proof depends on that determinant evaluation.","supporting_citations":[],"review_version":1}