{"id":"10d99d23-e078-47a1-b434-77be19017a7c","arxiv_id":"2507.09303","paper_version":3,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For fixed k, the asymptotic Mahler measure of Gaussian periods of conductor kn+1 is n times the Mahler measure of the cyclovariety x0+F_k(x)=0, which is asymptotically (1/2)log k.","lead":"The paper shows that for primes p=kn+1 with k fixed, the logarithmic Mahler measure of the Gaussian period of degree n grows like n times the Mahler measure of an associated 'cyclovariety' defined from k-th roots of unity. It then computes how those limiting Mahler measures grow with k, connecting a classical Lehmer-style height problem to Calabi-Yau geometry, random walks, and algebraic dynamics.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem B's explicit rate in (1.5) rests on the unverified normalization and constant C_Z=1 in imported Wasserstein bound [61]; the qualitative limit is independently safe, but the advertised error term needs a direct check.","rationale":"The reader's weakest_assumption already identifies this point: the quantitative part of Theorem B is imported from the preprint [61] and is not re-proven in the paper. I agree with that diagnosis. The qualitative asymptotic m(alpha_n) ~ m(C_k) n is independently secured by Myerson's equidistribution theorem and the Hlawka-Koksma argument, so the core identification of the growth rate is not at risk. What is at risk is the explicit bound (1.5), which is stated as a theorem and is the strongest form of the claim. Because the proof of (1.5) rests on three non-reproduced identifications -- the normalization of nu_q, the equality mu_g = F_k^* mu_{T^d}, and the value C_Z = 1 -- I would make acceptance conditional on a direct verification of these items. The paper is otherwise careful, self-aware, and computationally supported; there is no sign of circularity or fitted parameters. If the check in concrete_test succeeds, the reader's ACCEPT stands without revision. If it fails, Theorem B should be weakened to the qualitative statement plus whatever corrected bound is obtained.","tokens_in":62997,"tokens_out":21252,"duration_ms":255949,"concrete_test":"Write out [61, Prop. 3.4 / Cor. 3.6 / Rem. 3.7(2)] for Z=mu_k and check: (a) nu_q is normalized over all of F_p, including the a=0 contribution, so that (3.4) holds; (b) the construction in [60, Sec. 3] identifies mu_g with F_k^* mu_{T^d}; and (c) [61, Lem. 3.3] yields C_Z=1 for Z=mu_k. Recompute the Wasserstein bound under these identifications; if the constant differs from 4 sqrt(3k(k+1)) by any factor, revise (1.5) accordingly. A numerical cross-check for k=2 or k=4 with one large p in N_k can detect gross normalization errors but cannot by itself locate the source.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The quantitative part of Theorem B is the last inequality in (3.2): W(nu_1, nu_2) <= 4 sqrt(3k(k+1)) / p^{1/phi(k)}, obtained by citing [61, Cor. 3.6 and Remark 3.7(2)] with Z the set of k-th roots of unity. The proof asserts without re-derivation that (i) the measure nu_q in [61] is exactly nu_1 = (1/p) sum_{a in F_p} delta_{tr_H(zeta_p^a)}, including the a=0 atom; (ii) the limit measure mu_g in [61] equals F_k^* mu_{T^d}; and (iii) the constant C_Z from [61, Lem. 3.3] is 1 for Z = mu_k. Each of these identifications feeds directly into the right-hand side of (1.5): a different normalization would shift the log(k)/p term, a different limit measure would change m(C_k), and a C_Z different from 1 would rescale the constant 4 sqrt(3k(k+1)). The qualitative statement (1.4) does not depend on this: Myerson's equidistribution (Remark 3.3.4) plus the Hlawka-Koksma bound (3.9) with finite variation from Lemma 3.3.7 give the limit independently. Thus the fragile point is exactly the explicit, k-uniform error term advertised in the theorem.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies the asymptotic Mahler measure of the Gaussian periods α_n = tr_H(ζ_p), where p = κ(n)n + 1 is prime and κ(n) is the least k for which kn+1 is prime. For fixed k and n ∈ N_k, it proves that m(α_n) ∼ m(C_k)n, where C_k is the Laurent hypersurface x_0 + F_k(x) = 0 whose Newton polytope is the cyclopolytope, and it gives an explicit k-uniform error bound. The proof compares the empirical measure of the Gaussian periods with the push-forward of Haar measure under F_k. The paper also proves geometric properties of C_k (reflexive polytopes, log Calabi–Yau compactifications), interprets C_k as an algebraic dynamical system with entropy m(C_k), proves asymptotic expansions of m(C_k) for k = 2q^r, k = 2^r and k = q using Bessel functions, develops an algorithm for high-precision computation of m(C_k), proves by computer search that m_q = m(α_q) for q = 3,5,7, and finally derives a conditional framework under which m(α_n) ≍ n log log n for almost all odd n.","tokens_in":63326,"tokens_out":10900,"duration_ms":132377,"significance":"The paper's central achievement is a parameter-free asymptotic identification between the height of a natural family of cyclotomic integers and the multivariate Mahler measure of a family of log Calabi–Yau varieties. The qualitative limit is independently supported by Myerson's equidistribution theorem, and the paper is careful to label its conjectural input. The geometric and dynamical interpretations of the cyclovariety are valuable, and the computer-assisted minimality results in Section 8 are documented with publicly available code. The main caveat is that the explicit quantitative rate in Theorem B, and the constant terms in Theorem E, depend on imported results whose precise hypotheses and normalizations are not verified in the text.","major_comments":[{"comment":"The advertised explicit bound (1.5) is obtained by citing [61, Cor. 3.6 and Remark 3.7(2)] with Z equal to the set of k-th roots of unity. The proof asserts, without re-derivation, that the measure ν_q of [61] is exactly ν_1 = (1/p)∑_{a∈F_p} δ_{tr_H(ζ_p^a)}, that the limit measure μ_g equals F_k^* μ_{T^d}, and that the constant C_Z from [61, Lem. 3.3] equals 1 for this Z. Each of these identifications feeds directly into the right-hand side of (1.5): a different normalization would shift the (log k)/p term, a different limit measure would change m(C_k), and a C_Z different from 1 would rescale the constant 4√(3k(k+1)). Since Theorem B advertises an explicit k-uniform error term, the authors should either prove these identifications in the present paper, state the exact form of the imported result with all hypotheses verified, or weaken the statement to the qualitative limit (1.4), which is independently secured by Myerson's equidistribution and the Hlawka–Koksma argument in §3.3.","section":"§3.2, Eqs. (1.5) and (3.2)"},{"comment":"The proof of the asymptotic expansion (1.7) reduces to the evaluation ∫_0^∞ log(x) √(2/(πk)) e^{-x^2/(2k)} dx = (1/2)log k − (1/2)log 2 − (1/2)γ, which is stated in the text as 'computed the last integral using Mathematica'. This integral determines the constant terms −γ/2 and −(log 2)/2 in Theorem E, so it should be proved directly (for example, by differentiating the Gamma integral) or a precise reference should be supplied. Similarly, the bounds from [69] used in Lemmas 6.5.1 and 6.5.2 should be quoted with their exact constants and hypotheses, since they control the claimed O(k^{-3/2}) uniformity in x.","section":"§6.3, proof of Theorem E"}],"minor_comments":[{"comment":"In the display following (2.1), the inequality '#N_k(X) ⩽ N'_k(X)' should read '#N_k(X) ⩽ #N'_k(X)'.","section":"§2.1"},{"comment":"The sentence 'Since P_k is projective smooth toric by Lemma 4.2.1(i)' is inaccurate for k = 2q^r with r > 1; smoothness in that case follows from Lemma 4.2.1(ii), so the reference should cover both parts.","section":"§4.3"},{"comment":"References [19] and [20] appear to be the same article by Boyd; one of them should be replaced by the intended reference or merged.","section":"References"},{"comment":"The term 'außerwesentliche Diskriminantenteiler' in Lemma 8.4.1 is not standard in English-language papers; a parenthetical translation or definition would improve readability.","section":"§8.4"},{"comment":"In the proof of Lemma 8.5.1, the sentence 'q^q divides the coefficients of h(qx)' is a shorthand for a coefficient-by-coefficient divisibility; stating the resulting valuations v_q(b_j) ≥ q − j directly would be clearer.","section":"§8.5"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern about [61] lands precisely on Eq. (3.2) and the explicit form of (1.5). I do not regard this as a fatal flaw, because the qualitative limit is independently established and the missing piece is a verification of hypotheses or a reformulation of the quantitative statement. An appendix verifying the normalization and C_Z = 1, or a conditional formulation of the rate, would resolve the issue. The paper is otherwise strong and within the journal's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe takeaway: this paper delivers the first exact asymptotic for the Mahler measure of Gaussian periods and ties it to a natural family of multivariate Mahler measures. Theorem B's qualitative limit is solid, and Theorem E's Bessel-function asymptotic for m(C_k) is genuinely new and convincing. The computational proof of Theorem F is extensive, code is posted, and the authors are unusually clear about what is proven versus conjectured.\n\nWhat is actually new: the identification m(alpha_n)/n -> m(C_k) for fixed k, the first asymptotic expansion m(C_k) = (log k)/2 - gamma/2 - (log 2)/2 + ..., and the proof that m_q = m(alpha_q) for q = 7. The paper also connects these quantities to random walks, toric geometry, and algebraic dynamics in a useful way.\n\nThe soft spots are real but narrow. The explicit rate in (1.5) rests on the imported Wasserstein bound from [61] without re-deriving the normalization of nu_q, the identification of the limit measure, or the constant C_Z = 1. If that preprint has hidden hypotheses, the claimed k-uniform error term would need adjustment. The qualitative limit (1.4) does not depend on this—it follows from Myerson plus the Hlawka–Koksma argument in Section 3.3—so the fragility affects the advertised bound, not the central convergence statement. Two minor points: one integral in Theorem E is evaluated with Mathematica, and the Bessel-function bounds are imported; both are plausible but not machine-checked.\n\nThe paper's own conjectures (9.2.1, 9.2.2, G) are clearly labeled, and the dependence on the prime tuplets conjecture is stated. No circularity or fitting detected.\n\nWho this is for: anyone working on Mahler measure, heights of algebraic integers, or arithmetic dynamics will get real value. It deserves a serious referee. I would send it out, asking the referee to check the identifications in the proof of Theorem B, ideally by having the authors include a short verification of C_Z = 1 or a precise reference with lemma and page.\n\nRecommended: accept for peer review with targeted requests for clarification on the imported bound.","headline":"Strong paper with a novel asymptotic identification; the advertised error term leans on an imported constant that isn't re-proved, but the core result holds up.","tokens_in":63894,"tokens_out":3129,"would_cite":true,"duration_ms":33190,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11R06","11R18","11K38","14J33","52B20","60G50"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that the Mahler measure of Gaussian periods grows asymptotically as n times the multivariate Mahler measure of a cyclovariety, with an explicit error bound.","keywords":["Mahler measure","Gaussian periods","cyclotomic integers","Weil height","log Calabi-Yau varieties","algebraic dynamical systems","quantitative equidistribution","random walks"],"falsifier":"Fix $k=4$ and compute $m(\\alpha_n)/n$ with high precision for all $n\\in N_4$ up to $p\\approx10^6$; the claimed inequality predicts that $|m(\\alpha_n)/n - m(C_4)| \\le 4\\sqrt{3\\cdot4\\cdot5}/p^{1/2} + 2\\log4/p$. Any persistent violation above that envelope would disprove the quantitative Theorem B. A more direct test is to recompute the constant $C_Z$ for $Z=\\mu_k$ in the cited preprint's normalization in the specific case $k=2$ or $k=4$; if it is not 1, the explicit bound (1.5) needs correction.","tokens_in":62778,"feed_emoji":"🔢","tokens_out":9265,"duration_ms":97899,"temperature":0.7,"pith_summary":"Gaussian periods are trace-of-root-of-unity integers that generate cyclic extensions of the rationals; the paper's main theorem states that, for each fixed $k\\ge2$, as the conductor grows through values with $\\kappa(n)=k$, the logarithmic Mahler measure $m(\\alpha_n)$ of the period is asymptotic to $n\\,m(C_k)$, where $C_k$ is a specific Laurent hypersurface built from the $k$-th roots of unity. This gives an explicit rate $|m(\\alpha_n)/n - m(C_k)|\\le 4\\sqrt{3k(k+1)}/p^{1/\\phi(k)} + 2\\log(k)/p$ with $p=kn+1$. The identification turns the old question of minimal height in cyclic Galois extensions into a comparison between a discrete average over residues and an integral against Haar measure on a torus.","feed_headline":"Gaussian period heights track a cyclovariety's Mahler measure","feed_subtitle":"For each fixed k, these cyclotomic integers' Mahler measure grows as n times a multivariate Mahler measure, with explicit error.","key_machinery":"The central object is the cyclovariety $C_k$, the hypersurface $x_0+F_k(x_1,\\dots,x_d)=0$ whose Laurent polynomial $F_k$ has exponent vectors given by the coordinates of all $k$-th roots of unity in a basis of $\\mathbb{Z}[\\zeta_k]$ ($d=\\phi(k)$). The Gaussian period $\\alpha_n=\\sum_{\\sigma\\in H}\\zeta_p^\\sigma$ is the trace over the unique subgroup $H\\subset(\\mathbb{Z}/p)^\\times$ of order $k$, so its conjugates are the values of $F_k$ at an equidistributed and structured set of points. The argument carries through the equality of the relevant measures and the transfer of a quantitative equidistribution bound from the finite set of $k$-th roots of unity to those sample points.","core_discovery":"On the paper's own terms: the logarithmic Mahler measure (equivalently, the Weil height) of the Gaussian period $\\alpha_n$ is, up to smaller-order error, exactly $n$ times the multivariate Mahler measure of the cyclovariety $C_k$ when $n$ runs through the set $N_k$ on which the least prime in $kn+1$ is exactly $k$. The proof shows $m(\\alpha_n)/n$ is a Riemann sum or a push-forward delta measure of the same one-variable function $\\log^+|F_k|$, while $m(C_k)$ is the corresponding integral against the normalized Haar measure; quantitative equidistribution—via Wasserstein distance and a Weyl-type theorem for the sample points—controls the difference.","pith_inferences":["If the paper's dictionary is correct, a comparable construction for other prescribed Galois groups should yield hypersurfaces whose Mahler measure, times the degree, is the asymptotic height of an extremal generator; checking this for dihedral or elementary abelian groups is a direct test of the paper's general thesis.","The explicit error terms in Theorem B are only as trustworthy as the imported equidistribution constant that the authors assume equals 1; a reader who wants the qualitative result can rely on the classical equidistribution proof, but the claimed uniformity in $k$ and $p$ should be checked against the Hlawka–Koksma bound that the paper derives separately.","The random-walk formulation suggests that the asymptotic of $m(C_k)$ for all $k$ follows from a local central-limit statement for the distance of a correlated walk on the cyclopolytope; the Bessel-function representation in Section 6 is a natural starting point for that estimate."],"forward_implications":["For each fixed $k$, the height of $\\alpha_n$ is asymptotic to $n\\,m(C_k)$, so the minimal non-cyclotomic height in cyclic degree-$n$ extensions grows linearly in $n$ along every fixed $k$.","Combined with the proven asymptotic for $m(C_k)$ when $k=2q^r$ or $k=q$, the leading growth is $(n/2)\\log\\kappa(n)$ along those subsequences, pinning down lower-order terms as well.","Conjecture G asserts that for almost all odd $n$ the true growth is $m(\\alpha_n)\\asymp n\\log\\log n$, which would make the linear lower bound in (1.10) tight up to a $\\log\\log n$ factor for typical odd $n$.","The cyclovariety supports a $\\mathbb{Z}^{d+1}$-algebraic dynamical system of positive entropy $m(C_k)$, which is Bernoulli and (for a large family of $k$) non-expansive, so the asymptotic constant has a dynamical reading.","For $k=2$, the paper's Euler–Maclaurin argument gives the subleading constant $(\\log 2)/2$ in the expansion of $m(\\alpha_n)$."],"supporting_citations":[{"why":"Supplies the Wasserstein-distance bound that gives the explicit power-saving error term in Theorem B.","marker":"[61]"},{"why":"Proves the Weyl-type equidistribution of the finite-field sample points, yielding the qualitative convergence (1.4).","marker":"[82]"},{"why":"Identifies the measure of the relevant push-forward with the Haar measure on the torus, which is needed to match the two Mahler measures.","marker":"[60]"},{"why":"Provides an alternative route to the qualitative convergence for k-admissible integers.","marker":"[106]"},{"why":"Supplies the known closed values of $m(C_2)$ and $m(C_3)$ used as benchmarks for the cyclovariety's Mahler measure.","marker":"[100]"},{"why":"Proves the conjectured closed form for $m(C_4)$, confirming the pattern for the smallest even $k$.","marker":"[24]"},{"why":"Establishes that the topological entropy of the associated algebraic dynamical system equals the Mahler measure of the defining polynomial.","marker":"[73]"},{"why":"Provides the dictionary between algebraic properties of the variety and dynamical properties (ergodicity, mixing, non-expansiveness) used in Theorem D.","marker":"[94]"},{"why":"Gives the linear lower bound for $m_n$ that Conjecture G would sharpen to $n\\log\\log n$ for almost all odd $n$.","marker":"[93]"}],"fun_headline_variants":["Gaussian period Mahler measure ~ n × cyclovariety's","For Gaussian periods, Mahler measure ≈ n × cyclovariety's","Gaussian periods' Mahler measure: n times cyclovariety's","Period Mahler measure asymptotically n times cyclovariety's","Gaussian period heights track n-fold cyclovariety Mahler measure"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The quantitative error bound in the main theorem rests on an imported equidistribution estimate for the set of $k$-th roots of unity, including the claim that a certain constant equals 1; if that constant differs, or if the measure identification fails, the explicit decay in $p^{-1/\\phi(k)}$ would not hold as stated, though the qualitative limit would survive by the classical equidistribution theorem.","fun_headline_variants_meta":{"raw":{"variants":["Gaussian period Mahler measure ~ n × cyclovariety's","For Gaussian periods, Mahler measure ≈ n × cyclovariety's","Gaussian periods' Mahler measure: n times cyclovariety's","Period Mahler measure asymptotically n times cyclovariety's","Gaussian period heights track n-fold cyclovariety Mahler measure"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.003281,"raw_usage":{"total_tokens":12344,"prompt_tokens":878,"completion_tokens":11466,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":494,"completion_tokens_details":{"reasoning_tokens":11369}},"tokens_in":494,"tokens_out":11466,"duration_ms":93670,"temperature":1.0,"reasoning_tokens":11369,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T18:00:01.869132+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fix $k=4$ and compute $m(\\alpha_n)/n$ with high precision for all $n\\in N_4$ up to $p\\approx10^6$; the claimed inequality predicts that $|m(\\alpha_n)/n - m(C_4)| \\le 4\\sqrt{3\\cdot4\\cdot5}/p^{1/2} + 2\\log4/p$. Any persistent violation above that envelope would disprove the quantitative Theorem B. A more direct test is to recompute the constant $C_Z$ for $Z=\\mu_k$ in the cited preprint's normalization in the specific case $k=2$ or $k=4$; if it is not 1, the explicit bound (1.5) needs correction.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves the Weyl-type equidistribution of the finite-field sample points, yielding the qualitative convergence (1.4)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Identifies the measure of the relevant push-forward with the Haar measure on the torus, which is needed to match the two Mahler measures."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides an alternative route to the qualitative convergence for k-admissible integers."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the known closed values of $m(C_2)$ and $m(C_3)$ used as benchmarks for the cyclovariety's Mahler measure."},{"cited_title":"On the Mahler measure of $(1+x)(1+y)+z$","cited_arxiv_id":"2305.02992","evidence_quote":"Proves the conjectured closed form for $m(C_4)$, confirming the pattern for the smallest even $k$."},{"cited_title":"Math.101(1990), no","cited_arxiv_id":null,"evidence_quote":"Establishes that the topological entropy of the associated algebraic dynamical system equals the Mahler measure of the defining polynomial."},{"cited_title":"128, Birkhäuser Ver- lag, Basel, 1995","cited_arxiv_id":null,"evidence_quote":"Provides the dictionary between algebraic properties of the variety and dynamical properties (ergodicity, mixing, non-expansiveness) used in Theorem D."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the linear lower bound for $m_n$ that Conjecture G would sharpen to $n\\log\\log n$ for almost all odd $n$."}],"review_version":1}