{"id":"9102ca38-a0a3-49a5-9826-e557ed525ffd","arxiv_id":"2411.19318","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For abelian Galois groups Γ, the p-class group splits into a ramification-forced part with infinite average rank and a conjecturally Cohen-Lenstra distributed part; a weighted moment version is proved over F_q(t).","lead":"This paper proposes a Cohen-Lenstra type description of the p-primary class group of abelian extensions of Q and F_q(t), separating a ramification-forced part from a conjecturally random part. It proves a weighted moment form of the conjecture over function fields, including cases where p divides the Galois group order.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The proof removes the admissibility hypothesis from [LWZB24, Lemma 12.10] in Lemma 10.2 without a proof; since this underlies Proposition 10.3 and Theorem 1.2(2), the weighted moment theorem is not fully supported if the removal is invalid.","rationale":"The reader's weakest-assumption analysis identifies the removal of the admissibility hypothesis in Lemma 10.2 as the primary load-bearing gap. My reading of the manuscript confirms the dependency chain: Theorem 1.2(2) is proved via Proposition 10.3, which uses Lemma 10.2 to compare Hurwitz-space counts and conclude the ratio (10.12) equals |H2|/|H1|. Lemma 10.2(2) explicitly relies on [LWZB24, Lemma 12.10] with the admissibility condition dropped, and the footnote itself concedes that the resulting groups are not admissible when Γ_p is nontrivial. No independent proof of the removal is supplied. The p-power kernel property in the Schur-covering diagram is exactly the kind of statement that can depend on structural hypotheses about the action, so the gap is real and not merely cosmetic. I have not found a second concern of comparable weight: the other substantive steps (the I-closure identity in Proposition 9.3, the Hurwitz-space counting in (10.13), and the inclusion-exclusion leading to (1.3)) appear internally coherent, and the unweighted distribution is explicitly conjectural. The paper's own footnote makes the gap self-identified, which strengthens the need for a conditional verdict. I therefore agree with the reader's CONDITIONAL assessment and recommend no change; the concern is load-bearing but not demonstrated to be fatal. The proposed concrete test—checking whether admissibility is truly unused in the proof of [LWZB24, Lemma 12.10] in the relevant non-admissible semidirect-product cases—would settle the matter.","tokens_in":61367,"tokens_out":21333,"duration_ms":176195,"concrete_test":"Independently verify the removal of admissibility in Lemma 10.2 by reading the proof of [LWZB24, Lemma 12.10] and checking every place the admissibility hypothesis is invoked, especially in proving that ker(f|ker(S_i→G_i)) in diagram (10.8) has p-power order. Re-run the argument for G=(H_i×Γ_p)⋊Γ′ with H_i a finite p-group; if the p-power kernel property cannot be derived from the exact sequence 0→H2(Γ′,Z)→H2(G,Z)→(H_i×Γ_p)_{Γ′} (or an equivalent argument), then (10.10) and Lemma 10.2(2) are unsupported and Theorem 1.2(2) is in doubt.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim Theorem 1.2(2) rests on Proposition 10.3, whose proof invokes Lemma 10.2(2) to equate Hurwitz-space main terms b(G1,c1,q,n) and b(G2,c2,q,n). In the proof of Lemma 10.2(2), the paper explicitly removes the admissibility hypothesis from [LWZB24, Lemma 12.10], asserting 'this condition can be removed because it is not used in the proof.' This assertion is load-bearing and unproved. The key step in Lemma 10.2(2) is the diagram (10.8) and the conclusion that the kernel of f|ker(S_i→G_i) has p-power order, which feeds directly into (10.10) and the equality of b-values. If admissibility is actually needed for this conclusion in the non-admissible cases G_i=(H_i×Γ_p)⋊Γ′ with Γ_p nontrivial, then the equality of b-values is unsupported, the ratio in (10.12) may not equal |H2|/|H1|, and Theorem 1.2(2) fails. The paper provides no argument beyond the footnote, so a verification is required.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the distribution of the p-part of class groups Cl(K) as K varies over totally real Galois extensions of Q or F_q(t) with a fixed finite abelian Galois group Γ. For each primitive idempotent e of Q_p[Γ] the author constructs a discrete valuation ring eZ_p[Γ] and an ideal I_e, and proves a lower bound (an analogue of genus theory) for the rank of the bad part eCl(K)/(I_e eCl(K)). The main results are: Theorem 1.2(1), that the average I_e-rank is infinite; Theorem 1.2(2), a weighted moment statement in the function field case asserting that the weighted average of #Sur_Γ(I_e eCl(K), M) is 1/|M|; Theorem 1.3, a refinement for Γ=Z/2Z and p=2; and Theorem 1.4 on the kernel of Cl(K) → ⊕_e eCl(K). The paper also states Conjecture 12.2 giving the unweighted moment and probability distribution for I_e eCl(K), and shows that the conjectures specialize to the Cohen-Lenstra-Martinet and Gerth heuristics when p ∤ |Γ| and when Γ=Z/pZ respectively. The proofs combine cohomological methods for presentations of Galois groups with restricted ramification and Hurwitz-space point counting in the function field setting.","tokens_in":1669,"tokens_out":2782,"duration_ms":365427,"significance":"If the results are correct, this is a substantial generalization of the Cohen-Lenstra-Gerth heuristic framework to all finite abelian Galois groups, including the difficult bad prime cases where p divides |Γ|. The weighted-moment technique introduced here is designed to handle infinite moments coming from Hurwitz-space counts and is likely to be of independent use. The paper gives detailed proofs of the main theorems, and the agreement of the conjectures with the known Cohen-Lenstra-Martinet and Gerth predictions provides strong internal consistency. The unweighted Conjecture 12.2 is honestly labeled as a conjecture, which is appropriate since the proof is not attempted. The main concern is a load-bearing point in the proof of the weighted moment theorem (Theorem 1.2(2)) where a hypothesis from an external lemma is removed without proof; this is discussed in the major comments.","major_comments":[{"comment":"The proof of Lemma 10.2(2) invokes [LWZB24, Lemma 12.10] to obtain Schur coverings S_i → G_i and S_{Γ'} → Γ' with the property that the kernel of f|ker(S_i→G_i) has p-power order; this property is then used in (10.10) to equate nr_{q-1} values and conclude b(G1,c1,q,n) = b(G2,c2,q,n). For the groups G_i = (H_i × Γ_p) ⋊ Γ' that arise in Proposition 10.3, the admissibility hypothesis of [LWZB24, Lemma 12.10] is not satisfied when Γ_p is nontrivial. The footnote asserts that the admissibility condition can be removed because it is not used in the proof, but no argument or reference is supplied. This is load-bearing: if the p-power-order property fails for these non-admissible groups, the equality of b-values in Lemma 10.2(2) is unsupported, and with it Proposition 10.3 and Theorem 1.2(2) collapse. The author should either give a complete proof of the removal, or modify the argument to avoid relying on the admissibility hypothesis in the non-admissible cases.","section":"§10.2, proof of Lemma 10.2(2), footnote 3"}],"minor_comments":[{"comment":"The proof of Lemma 4.4 iteratively adds primes p not in S ∪ T'(Q) to make B vanish, but it does not ensure that the added primes avoid ∪_{ℓ|p|Γ|} S_ℓ(Q). Condition (3) counts only primes outside that union. The argument appears to establish only the upper bound #(S minus (∪_{ℓ|p|Γ|} S_ℓ(Q) ∪ S ∪ T)) ≤ dim B / dim End, which is all that is later used in §6.3 and §7, but the stated equality is not justified. The lemma statement should either be weakened to an upper bound or the proof should be completed.","section":"§4.4, Lemma 4.4(3)"},{"comment":"The assertion 'Since (Hi)Γ is an abelian p-group and p∤q−1, we have m>1' is not literally correct for elements g whose image γ ∈ Γ′ has order dividing q−1: for such γ one has γ^q = γ and m = 1. The conclusion of Claim 2 still holds in that case because (1−γ) is a unit on eZ_p[Γ] so the H-coinvariant part is trivial, but the proof should be rephrased to treat the cases γ ∈ Γ_p and γ ∈ Γ′ separately.","section":"§10.2, proof of Lemma 10.2(2), Claim 2"},{"comment":"In the proof of Lemma 2.6, the phrase 'for e1 ≠ e2 in Idem(Fp)' should read 'in Idem(A)' to match the notation of the lemma.","section":"§2, Lemma 2.6"},{"comment":"The iterated-limit notation in §1.4 defines lim_{x→∞} lim_{y→∞} for a two-variable function, but Theorem 1.2(2) and Proposition 10.3 impose additional restrictions on q (p∤q(q−1), gcd(q,|Γ|)=1). It would be helpful to state explicitly that the inner limit is taken over q satisfying these restrictions.","section":"§1.4 and Theorem 1.2(2)"},{"comment":"In Definition 2.9, 'the maximal interger r' contains a typo; it should be 'integer'.","section":"§2, Definition 2.9"}],"recommendation":"major_revision","confidential_remarks":"The admissibility issue flagged in Major Comment 1 is the single genuine obstacle to the central theorem. The rest of the paper appears carefully written and the auxiliary results are plausible, but the weighted moment theorem Theorem 1.2(2) cannot be considered proved until the removal of the admissibility hypothesis in Lemma 10.2(2) is justified. If the author supplies a valid proof of that removal, the paper would constitute a significant advance. There is no indication of circularity or duplicate publication, but the cited [LWZB24, Lemma 12.10] is by the same group of authors, so a public clarification would be appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the headline: this is a substantial paper, and the weighted-moment theorem (Theorem 1.2(2)) is the first real distributional handle on class groups of abelian extensions when p divides |Γ|. The construction of the discrete valuation rings eZ_p[Γ] and the ideal I_e is new, and the paper's conjectures interpolate neatly between Cohen-Lenstra-Martinet and Gerth. The Z/2Z theorem (Theorem 1.3) is a concrete payoff, and it agrees with Smith's condition on q modulo 4.\n\nWhat it does well: the genus-theory lower bound (Theorem 1.1), the kernel result (Theorem 1.4), and the careful separation of the 'bad' part eCl(K)/I_e·eCl(K) from the 'good' part I_e·eCl(K). The proofs are detailed and the paper is honest about what is conjectural; the unweighted distribution is explicitly left as Conjecture 12.2.\n\nThe soft spot is exactly where the report flags: Lemma 10.2(2) removes the admissibility hypothesis from [LWZB24, Lemma 12.10] with only the footnote 'this condition can be removed because it is not used in the proof.' That lemma is load-bearing—it produces the Schur coverings whose kernels have p-power order, which feeds directly into (10.10) and the equality of b-values. I could not verify the removal from the text. If the admissibility condition is actually needed for G_i = (H_i × Γ_p) ⋊ Γ′, then the weighted moment theorem is unsupported. The author needs to supply a proof or a detailed reference, not a footnote.\n\nThe citation pattern is fine; the dependence on prior work by the author and Wood/Zureick-Brown is appropriate. No invented entities or free parameters. Significance if true is high.\n\nThis paper is for arithmetic statisticians and number theorists working on Cohen-Lenstra heuristics. It deserves a serious referee. I would send it to review, with the explicit instruction that the referee check Lemma 10.2 against [LWZB24] and either verify the removal or ask for a proof. If that lemma is fixed, this is a significant contribution.","headline":"Substantial new results on class group distributions for abelian extensions, with a narrow but load-bearing gap in Lemma 10.2 that needs verification before the weighted moment theorem is accepted.","tokens_in":62171,"tokens_out":4289,"would_cite":true,"duration_ms":38050,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11R29","11R32","11R45","11R58","14H30"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that, over function fields, the weighted moments of the good part of the p-class group in abelian extensions equal $1/|M|$, matching the random finite module model.","keywords":["class groups","abelian extensions","p-primary class group","genus theory","function fields","tame-cover moduli spaces","moment conjectures","ramification types"],"falsifier":"Compute the left-hand side of (1.3) for a small explicit case, such as $\\Gamma=\\mathbb{Z}/2\\mathbb{Z}$, $p=2$, and a finite module $M$, for $q\\equiv 3\\bmod 4$ and $q\\equiv 1\\bmod 4$; Theorem 1.3 predicts respective limits $1/|M|$ and $|(\\wedge^2M)[2^{v-1}]|/|M|$, so a single deviation for one module or base field would falsify the moment claim.","tokens_in":61167,"feed_emoji":"🎲","tokens_out":12862,"duration_ms":101659,"temperature":0.7,"pith_summary":"For a finite abelian group $\\Gamma$, the paper asks how the $p$-primary part of the class group of a $\\Gamma$-extension of $\\mathbb{Q}$ or $\\mathbb{F}_q(t)$ behaves as the extension varies. Its answer is a split: for each primitive idempotent $e$ of $\\mathbb{Q}_p[\\Gamma]$, the module $e\\mathrm{Cl}(K)=e\\mathbb{Z}_p[\\Gamma]\\otimes_{\\mathbb{Z}_p[\\Gamma]}\\mathrm{Cl}(K)[p^\\infty]$ contains a forced 'genus' quotient $e\\mathrm{Cl}(K)/(I_e\\cdot e\\mathrm{Cl}(K))$ whose size grows with the number of suitably ramified primes, and a complementary 'good' module $I_e\\cdot e\\mathrm{Cl}(K)$ that should be random. The paper proves the moment side of this randomness in the function-field case: with a weight function coming from the bad quotient, the average of $\\#\\mathrm{Sur}_\\Gamma(I_e\\cdot e\\mathrm{Cl}(K),M)$ over totally real $\\Gamma$-extensions of $\\mathbb{F}_q(t)$ with $\\mathrm{rDisc}(K)=q^n$, for $0\\le n\\le N$, tends to $1/|M|$ in the iterated limit $q\\to\\infty$ then $N\\to\\infty$. This recovers the classical class-group heuristic in the tame case and the cyclic $p$-extension distribution in the case $\\Gamma=\\mathbb{Z}/p\\mathbb{Z}$, and it identifies the ideal $I_e$ explicitly.","feed_headline":"Weighted class-group moments equal 1/|M| over function fields","feed_subtitle":"For abelian extensions, the forced part grows while the good part obeys the predicted 1/|M| moments.","key_machinery":"The load-bearing object is the discrete valuation ring $e\\mathbb{Z}_p[\\Gamma]$ attached to a primitive idempotent $e$ of $\\mathbb{Q}_p[\\Gamma]$, together with the ideal $I_e$ defined as the intersection, over nontrivial $\\gamma\\in\\Gamma$, of the images in $e\\mathbb{Z}_p[\\Gamma]$ of the ideals $(1-\\gamma,\\,1+\\gamma+\\cdots+\\gamma^{|\\gamma|-1})$. This ring classifies finite $e\\mathbb{Z}_p[\\Gamma]$-modules by their $I$-ranks and separates each $e\\mathrm{Cl}(K)$ into the bad quotient $e\\mathrm{Cl}(K)/(I_e\\cdot e\\mathrm{Cl}(K))$ and the good module $I_e\\cdot e\\mathrm{Cl}(K)$. The weighted-moment proof counts points on tame Galois-cover moduli spaces: extensions with a prescribed module $H$ correspond to points of these moduli spaces, an identity in Proposition 9.3 converts surjections onto $H$ into surjections onto $I_eH$ times a weight factor, and a comparison lemma for point counts identifies the dominant contribution as $1/|M|$.","core_discovery":"The central claim is that the $p$-primary class group of an abelian $\\Gamma$-extension, as a $\\mathbb{Z}_p[\\Gamma]$-module, decomposes into a statistically infinite part forced by ramification and a statistically finite part obeying the predicted moments. Concretely, each primitive idempotent $e$ of $\\mathbb{Q}_p[\\Gamma]$ gives a discrete valuation ring $e\\mathbb{Z}_p[\\Gamma]$, and there is a canonical ideal $I_e$ such that the quotient $e\\mathrm{Cl}(K)/(I_e\\cdot e\\mathrm{Cl}(K))$ has $I_e$-rank bounded below by the number of primes of certain ramification types, so its average is infinite. The complementary module $I_e\\cdot e\\mathrm{Cl}(K)$ is conjectured to be equidistributed by the unique probability measure with $M$-moment $1/|M|$; Theorem 1.2(2) proves the weighted version of this moment statement over function fields for every nontrivial $e$. The weight in the average is exactly the homomorphism count from the bad quotient to a fixed module, so the result says that the bad part, despite being infinite on average, does not change the moments of the good part.","pith_inferences":["The same weighted-moment strategy is pointed toward other bad-prime settings, such as base fields containing the relevant roots of unity, where moduli-space moments diverge; carrying it out would require tracking central-extension corrections of the kind isolated in the comparison lemma.","The bound $|I_{e_0}(e_0\\mathrm{Cl}(K))|\\le |\\wedge^2\\Gamma_p|$ suggests a general pattern: in wild abelian extensions, the trivial-representation component is controlled by the wedge square of the maximal $p$-power subgroup of $\\Gamma$, so it vanishes when that subgroup is cyclic.","A direct numerical check over $\\mathbb{F}_q(t)$ for small $q$ and $N$ could test whether the weight function and the good part are asymptotically independent; if not, the unweighted conjecture would fail in a detectable way.","The infinite average kernel when $p^2\\mid |\\Gamma|$ indicates that the idempotent decomposition misses a statistically large piece exactly in the wild case, so full distribution results may need to parametrize this kernel as well."],"forward_implications":["For every nontrivial primitive idempotent $e$, the weighted moment of the good part equals the moment of the unique random finite module measure, so the good part is fully determined by these moments.","The bad part $e\\mathrm{Cl}(K)/(I_e\\cdot e\\mathrm{Cl}(K))$ has infinite average rank, so any unweighted moment statement must fail; the weighting in (1.3) is essential in the function-field count.","When $p\\nmid |\\Gamma|$, the weight function is constant and the theorem recovers the moment version of the standard class-group heuristic; when $\\Gamma=\\mathbb{Z}/p\\mathbb{Z}$, it gives the predicted distribution of $(1-\\gamma)\\mathrm{Cl}(K)[p^\\infty]$.","The trivial idempotent component is bounded by $|\\wedge^2\\Gamma_p|$, so the full class group's growth is carried by the nontrivial $e$-components and by the kernel of the map to $\\bigoplus_e e\\mathrm{Cl}(K)$, whose average rank is infinite when $p^2\\mid |\\Gamma|$."],"supporting_citations":[{"why":"supplies the tame-cover moduli-space counting machinery and the lemma whose admissibility hypothesis Lemma 10.2 removes.","marker":"[LWZB24]"},{"why":"established the function-field moment proof by counting points on tame-cover moduli spaces, the method adapted here.","marker":"[EVW16]"},{"why":"used in Proposition 12.1 to show the moments $1/|M|$ determine a unique probability measure.","marker":"[SW22]"},{"why":"previous presentation of Galois groups with restricted ramification that the genus-type lower bound generalizes.","marker":"[Liu24]"},{"why":"gives the asymptotic count of abelian extensions with local conditions used in the appendix.","marker":"[Woo10]"},{"why":"provides the moment formulation of the class-group heuristic checked in the tame case.","marker":"[WW21]"},{"why":"the classical class-group heuristic whose moment prediction this paper extends.","marker":"[CL84]"}],"fun_headline_variants":["p-class groups split: forced part infinite, good part moments 1/|M|","Weighted moment conjecture proved for abelian function fields","Abelian extension p-class groups: infinite forced part, weighted moments for good part"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The weighted-moment theorem rests on the claim that a technical admissibility condition in the cited moduli-space count can be dropped; if that condition is genuinely needed, the point counts can differ and the moment equality could fail.","fun_headline_variants_meta":{"raw":{"variants":["p-class groups split: forced part infinite, good part moments 1/|M|","Weighted moment conjecture proved for abelian function fields","Abelian extension p-class groups: infinite forced part, weighted moments for good part"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000403,"raw_usage":{"total_tokens":2251,"prompt_tokens":1250,"completion_tokens":1001,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":866,"completion_tokens_details":{"reasoning_tokens":937}},"tokens_in":866,"tokens_out":1001,"duration_ms":9288,"temperature":1.0,"reasoning_tokens":937,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T10:18:26.112945+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the left-hand side of (1.3) for a small explicit case, such as $\\Gamma=\\mathbb{Z}/2\\mathbb{Z}$, $p=2$, and a finite module $M$, for $q\\equiv 3\\bmod 4$ and $q\\equiv 1\\bmod 4$; Theorem 1.3 predicts respective limits $1/|M|$ and $|(\\wedge^2M)[2^{v-1}]|/|M|$, so a single deviation for one module or base field would falsify the moment claim.","supporting_citations":[],"review_version":1}