{"id":"730cacad-e8c0-4499-9aca-41fbaf9a12a6","arxiv_id":"2501.12529","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Multiple Dirichlet series heuristics reproduce, and for elliptic curve twists correct, the standard moment predictions for four families of L-functions, with each recipe term matched to a residue.","lead":"This paper develops a heuristic based on multiple Dirichlet series to guess the average sizes of special values of four families of L-functions, including Dirichlet characters, quadratic characters, modular forms, and elliptic curves. It shows that a standard recipe fails for elliptic curve twists, proposes a modified version, and matches every term of the recipe with a residue of the Dirichlet series.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Sign error in §5: replacing χ_d(−N) by χ_d(N) creates the new odd-swap terms; with the correct sign they vanish and the elliptic-curve modification collapses.","rationale":"The reader's conditional verdict focuses on the unproved meromorphic continuation conjectures. Those are legitimate caveats, but they are explicitly heuristic assumptions and do not invalidate the formal residue-to-term correspondence. The more damaging problem is internal to the elliptic-curve section: the new modified recipe is derived with χ_d(N) in place of the χ_d(−N) that appears in the paper's own functional equation (5.1). The missing factor χ_d(−1) is nontrivial on positive fundamental discriminants, and restoring it makes all odd-|J| residue terms vanish, because the relevant arithmetic condition M(−N)^{|J|}n_1⋯n_k=□ is impossible. Consequently Conjecture 5.2 collapses to Conjecture 5.1, and the appendix's purported verification of a two-term first moment is not a verification of the modified recipe as stated. Sections 2–4, where the same residue-to-recipe correspondence is worked out, may well be correct; but the paper's headline claim that the CFKRS recipe requires modification for elliptic-curve twists is unsupported and appears false for the family summed over all positive fundamental discriminants. The statement could be salvaged by restricting to a single root-number class, where the root number is constant and the standard CFKRS recipe already gives both swap terms; but that is not what the paper proves, and its summed-over-d formulation must be withdrawn. Because the central novelty fails, the verdict should move from CONDITIONAL to REJECT.","tokens_in":23489,"tokens_out":25454,"duration_ms":253497,"concrete_test":"Recompute the functional-equation chain (5.2.10) using χ_d(−N) and evaluate the residue at σ_J(w)=1 for |J|=1: it is H_{E,−MN}(1−s), which by the defining sum (5.1.6) is the empty sum and equals 0. As an independent check, take a concrete non-square conductor, e.g. N=32 for y²=x³−x, set M=1 and k=1, and compute the smooth first moment with α=1/log X at X=10^5 using the approximate functional equation; compare the coefficient of X^{1−2α} with the two-term prediction (6.2.6). The honest character average of χ_{8d}(−ℓN) over d has no pole, so this coefficient should be consistent with zero up to the expected error term.","verdict_should_be":"REJECT","load_bearing_attack":"The central novelty is Case (4). The functional equation (5.1) has the factor χ_d(−N), and (5.2) is written correctly. But (5.1.4) and the modified-recipe derivation use χ_d(N) instead of χ_d(−N). For even |J| this is harmless, but for odd |J| the missing χ_d(−1)=(−1/d) is not constant over positive fundamental discriminants. Repeating the functional-equation step (5.2.10) with the correct character gives A_{E,M}(s;w)=∏_{j∈J}X_E(s_j)A_{E,M(−N)^{|J|}}(s^J;w+2∑_{j∈J}s_j−|J|). For |J| odd the first argument is negative, and the residue at w=1 given by Proposition 5.4 is H_{E,M(−N)^{|J|}}(S∖S_J∪S^−_J)=0 by definition (5.1.6), since M(−N)^{|J|}n_1⋯n_k cannot be a square. Thus the odd-swap residues advertised after (5.2.10) do not exist, and Conjecture 5.2 reduces to Conjecture 5.1. The same sign omission appears in the appendix: M^−_E should contain χ_{8d}(−ℓN), whose average over d has no pole because −ℓN is not a square; the second term in (6.2.6) is therefore not the honest Shen-style asymptotic. The claimed disagreement between CFKRS and the multiple Dirichlet series heuristic is an artifact of the sign, not a genuine correlation effect.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops multiple Dirichlet series (MDS) heuristics for shifted moments of four families of L-functions: even primitive Dirichlet characters of conductor up to Q, quadratic Dirichlet L-functions, quadratic twists of a level-one Hecke eigencuspform, and quadratic twists of a fixed elliptic curve with non-square conductor. For the first three families, the authors compute the residues of the associated multiple Dirichlet series and show a term-by-term match with the CFKRS recipe prediction. For the elliptic-curve family, they claim that the unmodified CFKRS recipe misses terms with an odd number of functional-equation applications, because the root number ε(E)χ_d(−N) is correlated with the Dirichlet coefficients; they propose a modified recipe and state that the appendix verifies the first moment by adapting a result of Shen. The asymptotic predictions are explicitly conditional on strong meromorphic-continuation conjectures, and the rigorous content is mainly the residue computations in Propositions 2.4, 3.2, 4.2, and 5.4.","tokens_in":23883,"tokens_out":33369,"duration_ms":328796,"significance":"If the claims are correct, the paper provides a useful dictionary between residues of multiple Dirichlet series and terms of the CFKRS recipe, and it identifies a genuine subtlety in the recipe for families with sign-coefficient correlation. The residue computations are standard and appear correct, and the conjectured analytic properties are stated precisely enough to be falsifiable. The main weakness is that the appendix's verification of the elliptic-curve first moment is only sketched; since the paper's central novelty is the failure of the unmodified recipe, this verification is load-bearing and needs to be either fully supplied or explicitly downgraded to a conjecture.","major_comments":[{"comment":"The appendix claims that a 'straightforward modification' of Shen's proof yields the asymptotic formula (6.2.6) with a power-saving error term, but the required modifications are not actually given. The function R_E(α, ℓ) is never defined, the treatment of bad Euler factors at primes dividing N is not explained, and the passage from Shen's χ_{8d}-family to the family of all positive fundamental discriminants coprime to N is not justified. Since the assertion that Conjecture 5.1 is false for the first moment rests entirely on (6.2.6), this gap is load-bearing and must be addressed before the paper's central claim can be accepted.","section":"§6.2, Eq. (6.2.6)"},{"comment":"Equations (5.1.4), (5.2.10), and the appendix use χ_d(N) where the functional equation (5.1) has χ_d(−N). The equality χ_d(−N)=χ_d(N) does hold because d runs over positive fundamental discriminants, for which χ_d(−1)=1, but this fact is never stated. The odd-|J| terms in Conjecture 5.2 depend on this point, and the current notation invites a spurious sign-error objection; please add an explicit remark and use a single convention throughout.","section":"§5.1–5.2, (5.1.4), (5.2.10)"},{"comment":"The paper states that Conjecture 5.1 is false by comparison with (6.2.6). However, (6.2.6) is itself derived only under an unproved extension of Shen's theorem and under the meromorphic-continuation Conjecture 5.5. The falsity claim should be framed as conditional on these two unproved inputs unless the appendix is made fully rigorous.","section":"§5.1, Conjecture 5.1"}],"minor_comments":[{"comment":"The displayed formula uses a product over J, but it should be a sum over J; the surrounding text and the subsequent equations clearly intend a sum.","section":"Eq. (5.1.4)"},{"comment":"The subset notation ⊂ should be ⊆; otherwise the poles corresponding to applying the functional equation to all L-factors are not removed by the product.","section":"Conjectures 2.6 and 4.3"},{"comment":"The approximate functional equation is written for L(1/2+α, E_d), but in the definition of M_E(α, ℓ) the argument is E_{8d}; please correct this to E_{8d} throughout the appendix.","section":"Appendix, after (6.2)"},{"comment":"Reference [31] contains a formatting error in the authors' name ('Radziwi/suppress l/suppress l'); this should be fixed.","section":"References"},{"comment":"The functions R_E(α, ℓ) and R_E(−α, ℓN) are used without definition; please define them explicitly and relate them to H_{E,M} from (5.1.6).","section":"Appendix"}],"recommendation":"major_revision","confidential_remarks":"The stress-test's sign-error concern does not land on the mathematics as written: for positive fundamental discriminants d, χ_d(−1)=1, so χ_d(−N)=χ_d(N). The real weakness is the appendix's unproven 'straightforward modification' of Shen's result, which currently carries the paper's central conclusion that the unmodified CFKRS recipe fails for the first moment. The novelty of Case (4) is significant, but it needs to be supported either by a detailed proof or by a more modest framing as a conditional heuristic."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this for the framework, not for the case-4 punchline. The main new claim—that the multiple Dirichlet series heuristic forces a modification of the CFKRS recipe for quadratic twists of elliptic curves—does not survive contact with the functional equation's minus sign. In (5.1) the authors correctly write the twist root number as ε(E)χ_d(−N). In (5.1.4) and (5.2.10) they then use χ_d(N). For |J| even this makes no difference; for |J| odd the missing χ_d(−1) is not constant on positive fundamental discriminants, and its average over the family kills the term. Concretely, the corrected functional equation yields A_{E,M}(s;w) = ... A_{E,M(−N)^{|J|}}; for odd |J| the first argument is negative, and the residue at w=1 is H_{E,M(−N)^{|J|}} = 0 because the square condition in (5.1.6) is empty. So the odd-swap residues advertised after (5.2.10) don't exist, Conjecture 5.2 collapses into Conjecture 5.1, and the root-number correlation story disappears. The appendix does not rescue the full-family claim: it works with χ_{8d}, where χ_{8d}(−1)=1, so that minus sign is genuinely harmless there; but the main body's family includes odd discriminants. (The stress-test's appendix objection is thus wrong, while its §5 objection is right.)\n\nThe rest of the paper is worth reading. Section 2, the conductor-averaged unitary family, is a genuine new application of the MDS heuristic, and the residue computations in Propositions 2.4, 3.2, 4.2 and 5.4 look correct. The one-to-one correspondence between residue terms and recipe terms is a useful organizing principle, and it is made explicit without overclaiming. The conjectures about meromorphic continuation are honestly labelled, and the comparison to prior DGH and CFKRS work is careful. That said, the predictions for the first three families reproduce known CFKRS formulas; the novelty was supposed to be case 4, and that is the part that fails.\n\nBottom line: as a heuristic paper it deserves a serious referee, because the framework is reusable and the error is fixable in principle—but the current version's headline result is an artifact, and the authors should be asked to redo §5 with the sign straightened out before any claim about CFKRS failure appears.","headline":"The claimed elliptic-curve failure of CFKRS is a sign error away from disappearing; the rest of the paper is a solid MDS heuristic framework.","tokens_in":24329,"tokens_out":22776,"would_cite":false,"duration_ms":215508,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11M06"],"pacs":[],"model":"deepseek-v4-flash","headline":"Multiple Dirichlet series predicts L-function moments and fixes an elliptic-curve failure","keywords":["multiple Dirichlet series","moments of L-functions","CFKRS recipe","quadratic twists of elliptic curves","root number","symplectic family","unitary family","shifted moments"],"falsifier":"Evaluate the weighted first moment of $L(1/2+\\alpha,E_d)$ for a fixed elliptic curve of non-square conductor. The unmodified recipe predicts only the diagonal term, while the corrected prediction adds the swap term $X^{2-2s} X_E(s)\\tilde g(2-2s)H_{E,MN}(1-s)$ with $s=1/2+\\alpha$; an asymptotic or high-precision numerical evaluation would reveal which term is present, and the paper's appendix says the extra term appears.","tokens_in":1872,"feed_emoji":"🔢","tokens_out":5663,"duration_ms":161192,"temperature":0.7,"pith_summary":"Using multiple Dirichlet series, the paper predicts asymptotic formulas for shifted moments of four families of L-functions: even primitive Dirichlet characters (unitary), quadratic Dirichlet L-functions (symplectic), quadratic twists of a fixed modular form (orthogonal), and quadratic twists of a fixed elliptic curve. For the first three families, the prediction agrees term by term with the CFKRS recipe. For the elliptic-curve family, the recipe as usually stated is wrong, because averaging root numbers separately from the Dirichlet coefficients drops a correlation; the paper gives a corrected version and verifies the correction on the first moment. The main structural finding is a one-to-one correspondence between the residues of the multiple Dirichlet series and the terms of the recipe prediction. If the paper's continuation conjectures hold, each family's moment formula follows with a power-saving error term.","feed_headline":"L-function moment recipe fails for elliptic-curve twists","feed_subtitle":"A multiple Dirichlet series analysis shows why root numbers and coefficients must be averaged together.","key_machinery":"The central object is the multiple Dirichlet series built from moments of the family's L-functions, for example $A_{M,N}(s_1,\\ldots,s_k;z_1,\\ldots,z_k;w)=\\sum_{q\\ge 1}\\sum_{\\chi\\bmod q}^{\\flat}\\chi(M)\\chi(N)L(s_1,\\chi)\\cdots L(z_k,\\chi)/q^w$, with analogous sums for quadratic characters and quadratic twists. Its pole at $w=2$ or $w=1$ has residue equal to the diagonal main term of the recipe, and the functional equations of the L-functions shift that pole to locations $\\sigma_{J,H}(w)=2$ or $\\sigma_J(w)=1$ whose residues match the remaining recipe terms. The load-bearing conjecture is that the series continues meromorphically to a tube domain (a multidimensional analogue of a vertical strip) containing the central point, with only those shifted poles and with polynomial growth; the paper verifies convergence and the pole-residue statements only in smaller regions.","core_discovery":"The central claim is that the multiple Dirichlet series heuristic and the CFKRS recipe produce the same shifted-moment asymptotics, term by term, for the unitary, symplectic and orthogonal families, and that the elliptic-curve family is the one place where an unmodified recipe fails. The mechanism is a pole at $w=2$ (unitary) or $w=1$ (the other families) whose residue is the diagonal term, plus additional poles created by applying the functional equations of the individual L-functions; the residues of those shifted poles are exactly the swap terms of the recipe. In the elliptic-curve case, the functional equations introduce a factor $\\chi_d(N)^{|J|}$, so the shifted series is the same multiple Dirichlet series with $M$ replaced by $M N^{|J|}$; the usual recipe step of averaging root numbers first is what loses this factor. The paper states the precise continuation assumptions (Conjectures 2.6, 3.3, 4.3 and 5.5) under which these residue computations become asymptotics, and the appendix shows for the first moment that the corrected prediction holds while the unmodified one does not.","pith_inferences":["If the continuation conjectures hold beyond the leading pole, the multiple Dirichlet series should also encode secondary lower-order terms of the type already found for the third moment of quadratic Dirichlet L-functions, terms that the recipe does not predict.","The root-number and coefficient correlation exposed here is likely present in every family with genuinely varying root numbers, such as quadratic twists of higher-level newforms or twists restricted to arithmetic progressions, where the corrected averaging prescription would also be needed.","The residue-to-term dictionary suggests a practical route for generating new moment predictions: compute shifted poles of the multiple Dirichlet series rather than executing the full recipe.","A second-moment computation for elliptic-curve twists would sharply discriminate between the two versions of the recipe, since at that order the two predictions differ by several terms, not just one."],"forward_implications":["Conditional on the continuation conjectures, Propositions 2.8, 3.4, 4.4 and 5.6 turn the residue computation into asymptotic formulas with power-saving error terms, matching the CFKRS predictions for the first three families.","For the elliptic-curve family, the unmodified Conjecture 5.1 fails at the first moment, while the modified Conjecture 5.2 reproduces the asymptotic formula obtained in the appendix.","Every admissible subset of L-functions flipped by the approximate functional equation corresponds to one pole of the multiple Dirichlet series, so the recipe's swap terms have a uniform analytic origin.","The same machinery is remarked to extend to other families and to related statistics such as Ratios conjectures and one-level density of low-lying zeros."],"supporting_citations":[{"why":"The recipe whose shifted-moment predictions the paper reproduces in three families and corrects in the elliptic-curve family.","marker":"[9]"},{"why":"Origin of the multiple Dirichlet series heuristic for moments that this paper extends to four families.","marker":"[16]"},{"why":"Provides the first-moment asymptotic method adapted in the appendix to verify the corrected elliptic-curve prediction.","marker":"[32]"},{"why":"Supplies the analytic properties of L_D(w,chi) and a ratios-conjecture treatment used in the symplectic and orthogonal residue computations.","marker":"[7]"},{"why":"Determines explicit regions of meromorphic continuation for the real-character multiple Dirichlet series, supporting the continuation conjectures.","marker":"[8]"},{"why":"Gives the functional equation and root-number factor for quadratic twists of elliptic curve L-functions used in Section 5.","marker":"[31]"},{"why":"Establishes modularity of elliptic curves over Q, letting the level-one first-moment argument apply to elliptic-curve twists.","marker":"[2]"}],"fun_headline_variants":["Elliptic twists expose gap in L-function moment recipe","Root-number link forces elliptic L-moment correction","Dirichlet series agree with CFKRS but elliptic curves don't","L-function moments: elliptic case needs adjusted recipe","Shifted L-moments: multiple Dirichlet series vs. CFKRS"],"cache_read_input_tokens":26368,"weakest_assumption_plain":"The predictions hinge on the unproved assumption that every multiple Dirichlet series under study extends meromorphically into a large enough complex region around the critical point, with only the expected shifted poles and polynomial growth; the rigorous results in the paper reach that region only partially.","fun_headline_variants_meta":{"raw":{"variants":["Elliptic twists expose gap in L-function moment recipe","Root-number link forces elliptic L-moment correction","Dirichlet series agree with CFKRS but elliptic curves don't","L-function moments: elliptic case needs adjusted recipe","Shifted L-moments: multiple Dirichlet series vs. CFKRS"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000684,"raw_usage":{"total_tokens":3121,"prompt_tokens":977,"completion_tokens":2144,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":593,"completion_tokens_details":{"reasoning_tokens":2061}},"tokens_in":593,"tokens_out":2144,"duration_ms":15232,"temperature":1.0,"reasoning_tokens":2061,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T17:06:25.220061+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate the weighted first moment of $L(1/2+\\alpha,E_d)$ for a fixed elliptic curve of non-square conductor. The unmodified recipe predicts only the diagonal term, while the corrected prediction adds the swap term $X^{2-2s} X_E(s)\\tilde g(2-2s)H_{E,MN}(1-s)$ with $s=1/2+\\alpha$; an asymptotic or high-precision numerical evaluation would reveal which term is present, and the paper's appendix says the extra term appears.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The recipe whose shifted-moment predictions the paper reproduces in three families and corrects in the elliptic-curve family."},{"cited_title":"Diaconu, D","cited_arxiv_id":null,"evidence_quote":"Origin of the multiple Dirichlet series heuristic for moments that this paper extends to four families."},{"cited_title":"The first moment of quadratic twists of modular L-functions","cited_arxiv_id":null,"evidence_quote":"Provides the first-moment asymptotic method adapted in the appendix to verify the corrected elliptic-curve prediction."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the analytic properties of L_D(w,chi) and a ratios-conjecture treatment used in the symplectic and orthogonal residue computations."},{"cited_title":"Moments of real Dirichlet $L$-functions and multiple Dirichlet series","cited_arxiv_id":"2402.07473","evidence_quote":"Determines explicit regions of meromorphic continuation for the real-character multiple Dirichlet series, supporting the continuation conjectures."},{"cited_title":"Radziwi/suppress l/suppress l and K","cited_arxiv_id":null,"evidence_quote":"Gives the functional equation and root-number factor for quadratic twists of elliptic curve L-functions used in Section 5."},{"cited_title":"Breuil, B","cited_arxiv_id":null,"evidence_quote":"Establishes modularity of elliptic curves over Q, letting the level-one first-moment argument apply to elliptic-curve twists."}],"review_version":1}