{"id":"0e684691-8d1c-460c-94a7-a9a2caeddd23","arxiv_id":"2607.27364","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For every global function field K and finite abelian p-group G, the multivariate generating function that counts sub-G-extensions by successive higher-ramification heights is rational.","lead":"Global function fields are curves over finite fields, and this paper counts their abelian p-group extensions by several ramification heights at once. It proves the multivariate counting generating function is rational, giving exact formulas and asymptotics for all base curves, including non-rational ones.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The formula's local counting engine imports the filtration lemma [Gun26, Lemma 4.1] without reproof; a hidden failure there would invalidate Theorem 6.3, and no independent check is supplied.","rationale":"I read the proof in good faith and found no internal inconsistency. The exact sequence, Selmer bounds, character orthogonality, and the coefficient recursion all check out; the p-Selmer argument for non-rational curves is a genuine improvement, and the correction to [Lag15] is credible. The single place where the central claim is not self-contained is the filtration lemma imported from [Gun26]. That lemma is prior work of the author and is standard in nature, so I do not regard the dependency as a reason to reject; the reader's weakest_assumption points to the same spot. A failure there would change the answer, but no concrete failure was found. Hence the verdict is unchanged.","tokens_in":18561,"tokens_out":58021,"duration_ms":502382,"concrete_test":"Independently re-derive [Gun26, Lemma 4.1] for a non-prime residue field, e.g. p=2 and K_P=F_4((π)): construct the group isomorphism α explicitly and verify α(Λ^{k+1})=U^{k+1}_P for the first several k by computing the filtrations on a finite quotient U^1_P/U^N_P. Then recompute Lemma 4.10 directly for the same field and G=C_2 with a handful of k∈J_1, using the definition of exjump, and compare against the c_i,d_i formula. If both match, the dependency is sound; if α(Λ^{k+1}) misses U^{k+1}_P for some k, the local Euler factors in Theorem 6.2 are wrong.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 6.3 expresses F_K as a finite product/sum of Euler factors whose exponents c_i,d_i come from Lemma 4.10. Lemma 4.10 counts Hom(R^0_P/R^k_P,G), and Lemma 4.6 identifies the set of local characters with exjump_{P,j}≤k_j with this Hom-group. Both rely on Lemma 4.4, which asserts that the coordinate subgroup R^k_P = ∏ p^{λ_i(k)}Z_p^d is carried by the isomorphism α:R^0_P≅U^1_P to the subgroup generated by (U^{k_j+1}_P)^{p^j}. Lemma 4.4 is not proved independently; it is imported verbatim from [Gun26, Lemma 4.1]. If that filtration statement were false for some place type, say for residue degree d>1 where the Z_p^d coordinates are less obvious, then the local generating functions computed in Theorem 6.2 and hence the rational expression in Theorem 6.3 would not be the correct multivariate count. The paper gives no argument that [Gun26, Lemma 4.1] covers arbitrary completions of K, and the delicate part of Lemma 4.4's proof is exactly the coordinate-wise identity (4.1), which is checked only assuming the imported lemma. This is a load-bearing external dependency rather than a demonstrated error.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the multivariate counting of wild abelian extensions of a global function field K of characteristic p. For a finite abelian p-group G of exponent p^e, it defines e height functions deg(exjump_i(φ) - p·exjump_{i+1}(φ)) and forms the generating function F_K(X_0,...,X_{e-1}) = (1/|G|) ∑_{φ∈Hom(Γ_K,G)} ∏_i X_i^{deg(exjump_i(φ)-p·exjump_{i+1}(φ))}. The main result, Theorem 1.1 / Theorem 6.3, asserts that this multivariate generating function is rational and gives an explicit formula as the product of the trivial-character Euler factor ∏_i H_i(X_i) plus a finite sum over nontrivial characters of Ext^1_Z(Pic^0_K,G), each term involving a partial product of the H_i and a reciprocal Hasse-Weil zeta factor times a polynomial B_χ. The proof proceeds by class field theory, reducing global counts to a boundary map δ into Ext^1_Z(Pic^0_K,G), then to local unit-group filtrations. The local counting engine combines a filtration of U^1_P (Lemma 4.4, building on [Gun26, Lemma 4.1]), a count of homomorphisms R^0_P/R^k_P→G (Lemma 4.10), and a Selmer-group comparison of the filtrations H^k_P and I_r (Section 5). The paper also derives single-variable rationality and pole/asymptotic statements for the Artin-Schreier conductor in Theorem 7.5.","tokens_in":122,"tokens_out":28561,"duration_ms":737533,"significance":"If correct, this is a substantial advance: it gives exact, not merely asymptotic, multivariate counting formulas valid for every global function field, removing the rationality restriction of earlier work such as [Gun26] and introducing a genuinely multivariate height statistic. The explicit form in Theorem 6.3, with the Hasse-Weil zeta function and finitely many correction polynomials B_χ, is concrete enough to produce exact formulas for fixed K and G (see Example 6.4) and to yield uniform asymptotics and a negative-residue pole statement in Theorem 7.5. The paper also identifies and corrects a mistake in [Lag15] (Remark 5.7). The proof is detailed and internally coherent: the local generating function recursion in Theorem 6.2 is intricate but the key steps, including the Selmer obstruction and the boundary-map comparison, fit together. I checked the formulas in the small case G=C_p with e=1 against the elliptic-curve example and the local count of Lemma 4.10, and they agree.","major_comments":[],"minor_comments":[{"comment":"The letter d is used both for the residue degree of a place in Definition 4.2 and for the sequence d_i in Definition 4.7. Consider using d_P or r_P for the residue degree to avoid a notational collision that could confuse readers of Lemmas 4.4 and 4.10.","section":"Definition 4.2 and Definition 4.7"},{"comment":"The existence of the isomorphism α is imported from [Gun26, Lemma 4.1], and the paper proves only the coordinate identity (4.1). Since all of the subsequent local counting depends on this lemma, please state explicitly in the text that [Gun26, Lemma 4.1] applies to completions of an arbitrary global function field of characteristic p, including all residue degrees d≥1; if that lemma was originally stated only for d=1, a proof for general d should be supplied.","section":"Lemma 4.4 proof"},{"comment":"The tuple notation \"(k_0+p^g,...,k_{g-1}+p,k_g,k_{g+1},...)\" is not well-defined for g=0. The accompanying sentence indicates that the g=0 map is the identity, but the display should be adjusted, for example by writing the map for g≥1 and treating g=0 separately, to prevent a misreading of the induction.","section":"Theorem 6.2, Equation (6.3)"},{"comment":"The condition \"k_0≥...≥k_{r-1}>0 = k_r = k_{r+1}=...\" is terse and could be misread. Rephrase as \"k_0≥...≥k_{r-1}≥1 and k_j=0 for all j≥r\" for clarity.","section":"Corollary 5.5"},{"comment":"The positivity of C_i is asserted to follow from Remark 4.9(b)(c) together with the properties of Z_K. Please state explicitly that Z_K has no zeros in the closed disc |X|≤q^{-1}, since this standard fact is used to justify the nonzero denominators and the sign computation.","section":"Lemma 7.1(a)"}],"recommendation":"minor_revision","confidential_remarks":"This is a strong paper whose central result appears sound. The only substantive concern raised in review is the dependence of Lemma 4.4 on [Gun26, Lemma 4.1]; I do not regard this as a fatal or circular dependency, but the authors should state the exact scope of that cited lemma. The paper is well within the scope of a number-theory journal and represents a clear advance over the rational-field results in [Gun26]. I recommend minor revision rather than acceptance as-is only because the local-filtration dependency and a few notational points deserve explicit clarification."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Fabian Gundlach proves that the multivariate generating function counting sub-G-extensions of an arbitrary global function field K (any genus, any constant field) by e ramification heights is rational, and gives an explicit formula in terms of the Hasse-Weil zeta function. This is a genuine upgrade over [Gun26], which covered only rational K and a single height. The theorem is new and, as far as I can tell, correct.\n\nThe proof is careful and mostly self-contained. The use of p-Selmer groups to control the boundary map into Ext^1(Pic^0_K,G) is well executed, and the correction of Lagemann's mistake (Remark 5.7) is a real service. The local recursion in Theorem 6.2 is intricate but the final closed form is credible; I checked small cases like G=C_p, e=1 and the elliptic curve example, and they come out right. The asymptotics in Theorem 7.3 and the single-variable corollary are clean and useful.\n\nThe one soft spot is the dependence on Lemma 4.4, imported from [Gun26, Lemma 4.1]. The paper proves the combinatorial identity (4.1) that makes the filtration explicit, but the existence of the isomorphism α with the right subgroups is cited, not reproved. If that lemma were only proved for rational K, or failed for residue degree d>1, the local counts and hence Theorem 6.3 would fall. I think the stress-test note overstates the risk: the lemma is published and local, and there's no indication it is restricted to the rational case. But I would ask the author to state explicitly that it holds for every place of every global function field, or give the one-paragraph proof. This is a clarification, not a fix.\n\nThe only other minor issue is readability: the recursion in Theorem 6.2 is dense, and the 'small trick' in equation (6.3) deserves a sentence or two more. A serious referee will follow it, but it pushed me.\n\nBottom line: this is a solid, significant paper. The main theorem is a real result with explicit content, not just an abstract rationality claim. It deserves a careful referee and publication. I'd recommend accepting it into the review process; if I were refereeing, I'd ask for the clarification above and otherwise recommend acceptance.","headline":"Solid, significant extension of Gun26 to arbitrary base curves and multivariate ramification heights; the main theorem holds and the only real caveat is an imported local lemma that deserves an explicit generality statement.","tokens_in":19346,"tokens_out":4131,"would_cite":true,"duration_ms":35857,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11R45","11R37","11S40","30B10","11S15"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that for any finite abelian p-group G and any global function field K of characteristic p, the multivariate generating function counting sub-G-extensions by all successive higher-ramification height functions is rational…","keywords":["wild abelian extensions","function fields","higher ramification groups","rational generating functions","Artin-Schreier conductor","Hasse-Weil zeta function","p-Selmer group","abelian p-groups"],"falsifier":"Take $K=\\mathbb{F}_q(t)$, $G=C_p$, and compare the first coefficients of $Z_K(qX)Z_K(q^{p-1}X^p)/(Z_K(q^p X^p)Z_K(X))$ with a direct enumeration of Artin-Schreier extensions $y^p-y=f$ over $\\mathbb{F}_q(t)$ ordered by their conductor; any disagreement in the coefficient of $X^n$ for a small $n$ would show that Theorem 6.3 computes the wrong generating function.","tokens_in":18325,"feed_emoji":"🧮","tokens_out":11153,"duration_ms":93252,"temperature":0.7,"pith_summary":"Fix a finite abelian p-group G of exponent $p^e$ and a global function field $K$ of characteristic $p$. The paper establishes that the multivariate generating function counting sub-$G$-extensions $\\varphi$ of $K$ by the $e$ degrees $\\deg(\\mathrm{exjump}_i(\\varphi)-p\\,\\mathrm{exjump}_{i+1}(\\varphi))$ is rational, and writes it explicitly as a finite sum of products of translates of the Hasse-Weil zeta function of the base curve. This turns a counting problem with infinitely many local contributions at every place into a closed-form object. The result gives exact, not merely asymptotic, formulas for the number of extensions with prescribed successive ramification heights for every base curve, together with a product asymptotic when all heights grow.","feed_headline":"Wild abelian extensions counted by one rational formula","feed_subtitle":"For every function field in characteristic p, exact extension counts by ramification heights become a finite zeta-function formula.","key_machinery":"The load-bearing object is the explicit filtration of the local pro-$p$ unit group $U_P^1 = 1+\\pi_P O_P$. For $k=(k_0,k_1,\\ldots)$ in the cone $J_e=\\{k_i\\ge pk_{i+1},\\ k_i=0\\text{ for }i\\ge e\\}$, define $\\lambda_i(k)=\\min\\{\\lambda\\ge 0 : i>k_\\lambda\\}$ and $R_P^k=\\prod_{i\\ge 1,\\,p\\nmid i} p^{\\lambda_i(k)} \\mathbb{Z}_p^{d}$ with $d=\\deg(P)$. Lemma 4.4 supplies an isomorphism $\\alpha:R_P^0\\xrightarrow{\\sim} U_P^1$ that sends $R_P^k$ onto the subgroup generated by $(U_P^{k_j+1})^{p^j}$ for all $j$, converting the ramification-jump conditions $\\mathrm{exjump}_{P,j}(\\varphi)\\le k_j$ into the vanishing of $\\varphi$ on $R_P^k$. Lemma 4.10 then gives the local count $|\\mathrm{Hom}(R_P^0/R_P^k,G)| = Q_P^{\\sum (c_i v_i + d_{i+1} w_i)}$ when $k_i-pk_{i+1}=v_i+pw_i$. The $p$-Selmer group of $K$ controls the boundary map $\\delta$ to $\\mathrm{Ext}^1_{\\mathbb{Z}}(\\mathrm{Pic}^0_K,G)$, ensuring that the character-sum over the local obstructions differs from its naive Euler product only by finitely many places, which explains the finite correction polynomials $B_\\chi$.","core_discovery":"The central claim is Theorem 6.3: for every finite abelian group $G$ of exponent $p^e$ and every global field $K$ of characteristic $p$, $$F_K(X_0,\\ldots,X_{e-1}) = \\frac{1}{|G|} \\sum_{\\varphi\\in\\mathrm{Hom}(\\Gamma_K,G)} \\prod_{i=0}^{e-1} $X_i^{{\\deg(\\mathrm{exjump}}$_i(\\varphi)-p\\,\\mathrm{exjump}_{i+1}(\\varphi))}$$ is rational. More precisely, $F_K$ equals $\\prod_{i=0}^{e-1} H_i(X_i)$ plus a finite sum over the nontrivial characters $\\chi$ of $\\mathrm{Ext}^1_{\\mathbb{Z}}(\\mathrm{Pic}^0_K,G)$ of $\\big(\\prod_{i=0}^{r_\\chi-1} H_i(X_i)\\big)\\,\\frac{1}{Z_K(q^{d_{r_\\chi}}X_{r_\\chi})}\\,B_\\chi(X_{r_\\chi},\\ldots,X_{e-1})$, where $H_i(X_i)=\\frac{Z_K(q^{c_i}X_i)Z_K(q^{d_{i+1}}X_i^p)}{Z_K(q^{pc_i}X_i^p)Z_K(q^{d_i}X_i)}$ and $B_\\chi$ is a polynomial with integer coefficients and constant term 1, which is identically 1 when the genus of $K$ is at most 1. Thus rationality survives for non-rational base fields and the count tracks all $e$ successive drops in the higher-ramification filtration at once.","pith_inferences":["Substituting $X_i=Y^{a_i}$ for any nonnegative weights would give exact rational generating functions for any linear combination of the $e$ height functions, so the same closed-form phenomenon should hold for a whole family of coarser inertial invariants; the paper only carries out the substitution $X_i=X^{p^i}$ for the Artin-Schreier conductor.","A natural stress test is the non-abelian analogue: fix a non-abelian $p$-group $N$ and count surjective maps $\\Gamma_K\\to N$ by the same height functions. The paper's structure suggests the obstruction is whether a filtration like $R_P^k$ exists for the relevant pro-$p$ quotients, and if it does, rationality of the corresponding generating function would be a plausible extension.","Because the formula is built from the Hasse-Weil zeta function, the Riemann hypothesis for function fields implies that the corrections to the main asymptotic are oscillatory and controlled by the zeros of $Z_K$; this analytic refinement is not spelled out in the paper."],"forward_implications":["For any prescribed tuple $(n_0,\\ldots,n_{e-1})$, the exact number of sub-$G$-extensions with heights $n_i$ is the coefficient of $\\prod X_i^{n_i}$ in a known finite sum of Hasse-Weil factors, so counting is a finite algebraic operation once the curve's zeta function is known.","When all $n_i$ tend to infinity the counted number is asymptotic to $(\\prod C_i) \\prod q^{(1+c_i)n_i}$ and is bounded by a constant times the same product for every tuple, identifying the dominant pole $X_i=q^{-1-c_i}$ as the source of all exponential growth.","The single-variable Artin-Schreier-conductor generating function $F_K^{\\mathrm{asc}}(X)=F_K(X,X^p,\\ldots,X^{p^{e-1}})$ is rational and has a unique innermost simple pole at $X=q^{-a}$ with $a=1+\\dim_{\\mathbb{F}_p}G[p]$, giving the asymptotic $Cq^{an}$ for extensions counted by conductor degree.","For a base curve of genus at most one, all correction polynomials $B_\\chi$ are equal to 1, so the rational formula reduces to the product of zeta-function translates with no additional finite factor."],"supporting_citations":[{"why":"Supplies the local filtration lemma (Lemma 4.1) identifying $R_P^k$ with the subgroup generated by $(U_P^{k_j+1})^{p^j}$, the engine of all local counts; also the prior rationality result being generalized.","marker":"[Gun26]"},{"why":"Provides the logarithmic-derivative argument for Lemma 2.3 relating $p$-Selmer valuations to differentials, which controls the boundary map $\\delta$.","marker":"[Lag12]"},{"why":"Earlier Artin-Schreier-Witt counting whose Proposition 4.2 the paper corrects in Remark 5.7; motivates the Selmer-group method for groups of exponent larger than $p$.","marker":"[Lag15]"},{"why":"Source for Lemma 2.1 and Proposition 6.14 on representatives in $O_P^\\times/O_P^{\\times p}$ and Selmer valuations used in the local analysis.","marker":"[Pot26]"},{"why":"Classical class-field-theoretic counting of abelian extensions by discriminant that this proof adapts through the idele-class character sum.","marker":"[Wri89]"},{"why":"Models the idea of counting by several inertial invariants simultaneously, cited by the paper for the multivariate height-function setup.","marker":"[EV05]"},{"why":"Provides the analytic-combinatorics transfer theorem used to pass from the innermost pole of the rational generating function to coefficient asymptotics.","marker":"[FS09]"}],"fun_headline_variants":["Rational series counts wild abelian extensions by ramification","Wild abelian extensions: exact counts via finite rational formula","One rational series for wild abelian extension counts","Multivariate counting of wild abelian extensions made rational","Wild p-extensions: rational generating function for all jumps"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof rests on the imported lemma that the subgroup generated by $(U_P^{k_j+1})^{p^j}$ inside the local pro-$p$ unit group at $P$ is exactly the product $R_P^k$; if this filtration identification fails for some place type, the explicit rational formula would not be the true generating function.","fun_headline_variants_meta":{"raw":{"variants":["Rational series counts wild abelian extensions by ramification","Wild abelian extensions: exact counts via finite rational formula","One rational series for wild abelian extension counts","Multivariate counting of wild abelian extensions made rational","Wild p-extensions: rational generating function for all jumps"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000251,"raw_usage":{"total_tokens":1550,"prompt_tokens":934,"completion_tokens":616,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":550,"completion_tokens_details":{"reasoning_tokens":539}},"tokens_in":550,"tokens_out":616,"duration_ms":5853,"temperature":1.0,"reasoning_tokens":539,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T15:26:42.506739+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take $K=\\mathbb{F}_q(t)$, $G=C_p$, and compare the first coefficients of $Z_K(qX)Z_K(q^{p-1}X^p)/(Z_K(q^p X^p)Z_K(X))$ with a direct enumeration of Artin-Schreier extensions $y^p-y=f$ over $\\mathbb{F}_q(t)$ ordered by their conductor; any disagreement in the coefficient of $X^n$ for a small $n$ would show that Theorem 6.3 computes the wrong generating function.","supporting_citations":[],"review_version":2}