{"id":"a75f6fba-221e-41c5-9a3b-fb8c7c07982b","arxiv_id":"2501.09515","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper conjectures that the ideal factorization of L(E,χ,1) equals the product of minimal polynomials of Galois generators on the eigenspaces of the Tate-Shafarevich group, and provides numerical evidence via visualization and 11-descent.","lead":"This paper states a precise conjecture connecting the prime factorization of a twisted L-value of an elliptic curve to the Galois-module structure of its Tate-Shafarevich group. It then tests the conjecture on nine CM elliptic curves using a newly developed 11-descent algorithm over the quintic field Q(ζ11)+.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.6 silently assumes BSD: the claimed verification uses ConjecturalSha() to fix #X(E/Kχ)[11]=11^2, so the Galois-module conclusion is not established from the stated GRH-only hypotheses.","rationale":"The reader identified the hidden BSD dependence through ConjecturalSha() as the load-bearing assumption of the numerical verification, and that is the same concern that lands most directly when reading §3.5 against Theorem 1.6. The theorem states only GRH as a hypothesis, yet the computation uses the analytic Sha order to know #X(E/Kχ)[11]=11^2, to conclude X(E/Kχ)[11∞]=X(E/Kχ)[11], and to decide which of the candidate eigenspaces is actually present. This is not circular with respect to Conjecture 1.3, but it means the numerical evidence is conditional on the full BSD formula as well as GRH. The p-descent algorithm itself is a real contribution: the norm-relations decomposition in Proposition 3.8, the eigenvalue restriction in Proposition 3.10, and the Cassels–Tate symmetry in Proposition 3.14 are concrete and reusable. My recommended verdict therefore remains CONDITIONAL, matching the reader's verdict; no verdict change is needed, but the paper should restate Theorem 1.6 or explicitly include BSD among its assumptions, and should correct the label mismatches before publication.","tokens_in":1043,"tokens_out":1105,"duration_ms":56059,"concrete_test":"For at least 7056.bg1, rerun the 11-descent from §3.5 using the authors' Magma code, modified so that no quantity from ConjecturalSha() or any analytic Sha order is used. Compute all elements of Sel_11(E/Kχ) from the exact R(F_i,S_i;11) data, i.e., fully enumerate the unit-group and class-group contributions for i=1,...,4 (or prove which are empty) and check local conditions directly. If the resulting F_11[Gal(K/Q)]-module has dimension 2 with eigenspace pair {x−3,x−4} (matching (L(E,χ))_11 = p1p̄1), then Theorem 1.6 survives with only GRH; if dimension is not 2 or the pair differs, the theorem as stated is not proved. A cheaper preliminary check: grep the published code [Shu24] for ConjecturalSha and rerun with that input replaced by an independent bound from the descent computation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central numerical claim (Theorem 1.6) is that Conjecture 1.3 holds for nine CM curves with p=11 and all primitive characters of Kχ=Q(ζ11)+, assuming GRH. The proof in §3.5 uses more than GRH. For 7056.bg1, the text states: “|Xan(E/Q)| = 1 and Xan(E/Kχ) ≃ F_11^2 ⊕ F_31^2 (using ConjecturalSha())” and then “Since |X(E/K)[11]| = 11^2, Proposition 3.14 also shows that only one of u1 or u2 represents an element Sel_11(E/Kχ)[11].” ConjecturalSha() computes the analytic order of Sha, which is only known to equal the true order if the full BSD formula (Conjecture 1.1(2)) holds. That size is load-bearing in two places: it fixes the dimension of the Selmer module, and it is used with Proposition 3.14 to discard one of the two candidate eigenspaces. Without an independent proof that #X(E/Kχ)[11]=11^2, the computation only proves that certain elements lie in the Selmer group; it does not prove that the full Galois module has the predicted eigenvalue set, nor that the other candidate is absent. Thus Theorem 1.6, as stated with only GRH, is stronger than the computation establishes. This is a scope/rigor gap, not an internal inconsistency of Conjecture 1.3. The label discrepancies (6400.a1 vs 6400.r1; 57600.ch1 vs 57600.ch2) are secondary but should be fixed.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes Conjecture 1.3, which predicts the ideal factorization of the normalized twisted L-value L(E,χ) in terms of the minimal polynomials of the Galois eigenspaces of X(E/Kχ)[p]. The authors prove Theorem 2.2, a congruence-based result showing that, under p-congruence and Greenberg–Vatsal hypotheses, the ideal (p) divides the twisted L-value; Corollary 2.3 connects this to the product of eigen-polynomials. They then develop an 11-descent algorithm over C5-fields for CM elliptic curves, using norm relations to reduce the computation to degree-20 fields, and apply it to nine LMFDB curves, claiming in Theorem 1.6 that Conjecture 1.3 is verified for p=11 and all primitive characters through Q(ζ11)+, assuming GRH.","tokens_in":18167,"tokens_out":14766,"duration_ms":148153,"significance":"If Conjecture 1.3 is correct, it gives a very concrete Galois-module refinement of BSD-type formulas for twisted L-values, going beyond the norm-level predictions in earlier work. Theorem 2.2 is a clean and valuable consequence of the Greenberg–Vatsal congruences. The descent method in Section 3 is a substantial computational contribution: it makes 11-descent over a C5-field practical for CM curves, and the code is publicly available. However, the numerical verification is not as unconditional as Theorem 1.6 suggests, because the computation uses BSD-type predictions in two load-bearing places.","major_comments":[{"comment":"Theorem 1.6 is stated under GRH alone, but the verification in §3.5 uses additional conjectural input. For 7056.bg1 the text says '|Xan(E/Q)| = 1 and Xan(E/Kχ) ≃ F_11^2 ⊕ F_31^2 (using ConjecturalSha())' and then 'Since |X(E/K)[11]| = 11^2, Proposition 3.14 also shows that only one of u1 or u2 represents an element Sel_11(E/Kχ)[11].' ConjecturalSha() computes the order predicted by the BSD formula, so the equality between the analytic and algebraic Tate–Shafarevich groups is exactly Conjecture 1.1(2), not a consequence of GRH. This equality is load-bearing: without it the descent computation gives only a lower bound on Sel_11(E/Kχ), and the absence of the conjugate eigenspace is not established. In addition, the norm computation 'By Theorem 1.7' uses the BSD-type formula of Theorem 1.7, which also assumes the Stevens Manin constant conjecture and BSD for E over Q and Kχ. Remark 3.17 already says 'if we are ready to assume the BSD-conjecture', so the theorem statement should either include these hypotheses or be phrased as a conditional verification.","section":"§3.5 and Theorem 1.6"},{"comment":"The list of curves in Theorem 1.6 does not match the final table: the theorem names '6400.a1' and '57600.ch1', whereas the table lists '6400.r1' and '57600.ch2'. The prose in §3.5 also refers to 6400.a1. Since the numerical claim is attached to specific LMFDB labels, this discrepancy must be resolved before the theorem can be reproduced from its statement.","section":"Theorem 1.6 and final table in §3.5"}],"minor_comments":[{"comment":"The hypothesis list contains the condition 'E(K)[p∞] = E(Q)[p∞]' twice; the duplicate should be removed.","section":"Conjecture 1.3"},{"comment":"The expression 'Lp(E2, χ, 1) = 0' is not defined in the proof; it should presumably be 'L(E2, χ, 1) = 0' or 'LΣ(E2, χ, 1) = 0'.","section":"Proof of Theorem 2.2"},{"comment":"The notation in Remark 3.19 is confusing: if X(E/Kχ)[11] ≃ (Z/11^2 Z)^2, then its 11-torsion has order 11^2, while the surrounding text discusses an order of 11^4; the intended group and notation should be clarified.","section":"§3.5 and Remark 3.19"},{"comment":"The text says the pair (E1,E2) satisfies the conditions of Theorem 2.1, but Theorem 2.1 assumes Ei(K)/pEi(K) = 0, while E2(Kχ) has rank 4 and hence E2(Kχ)/11E2(Kχ) is nonzero; the index in the hypothesis of Theorem 2.1 needs to be stated correctly.","section":"§2.2 and Theorem 2.1"},{"comment":"The verification for 'all primitive Dirichlet characters' is only described for one embedding and one displayed matching of hθ to p1 or p̄1; a short per-character table would make the claim transparent.","section":"§3.5 numerical verification"}],"recommendation":"major_revision","confidential_remarks":"The central mathematical idea and the theoretical argument in Theorem 2.2 appear sound. The main problem is that Theorem 1.6 overstates what the numerical computation establishes, since ConjecturalSha() and the use of Theorem 1.7 silently introduce BSD and the Manin constant conjecture in addition to GRH. This is fixable by restating the theorem with the correct hypotheses and by correcting the LMFDB label mismatches. I do not see a reason to reject the paper, but the numerical theorem's statement must be made accurate before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is worth reading for two reasons. First, Conjecture 1.3 is a real refinement of the BSD philosophy: instead of predicting only the size of X(E/K), it predicts the Galois module structure through the p-part of the factorization of L(E, chi, 1). That is new and goes beyond DEW21 Remark 43. Second, the p-descent procedure in Section 3 is a genuine technical advance: by using CM and norm relations from Biasse-Fieker-Hofmann-Page, the authors reduce degree 600 class group and unit computations to degree 20 fields and make 11-descent over a C5 field practical. The code is on GitHub, and the numerical evidence for nine curves is substantial. Theorem 2.2 is a clean congruence argument using Greenberg-Vatsal, and Corollary 2.3 is honest about what is needed to get from divisibility by p to the eigenspace factorization. The main weakness is in the statement of Theorem 1.6. The theorem says assuming GRH, but the verification in Section 3.5 uses more than GRH. For 7056.bg1 the text says Xan(E/K_chi) is isomorphic to F_11^2 plus F_31^2 using ConjecturalSha(), and then uses |X(E/K)[11]| = 11^2 to decide which candidate eigenspace represents a Selmer element. ConjecturalSha() computes the BSD-predicted order, so the completeness of the descent search is conditional on BSD, not just GRH. If BSD fails for these curves, the Galois module structure could be different, and the stated verification does not establish Conjecture 1.3. This is a scope/rigor gap, not an internal contradiction; the conjecture itself and the descent method are independent of this issue. Still, the hypotheses of Theorem 1.6 need to be corrected to include BSD, or the authors should provide an unconditional bound on the Selmer group that does not rely on the analytic order. There are also small presentation issues: the labels in Theorem 1.6 (6400.a1, 57600.ch1) differ from the final table (6400.r1, 57600.ch2), and Conjecture 1.3 repeats the condition E(K)[p^infty]=E(Q)[p^infty] twice. Minor, but should be fixed. Who this is for: arithmetic geometers working on BSD, Galois module structure, or explicit descent; also people who need to actually compute Selmer groups over number fields. I would send this to a serious referee. The numerical evidence needs tightening, but the conjecture and the computational method deserve attention.","headline":"New Galois-module refinement of BSD plus a genuinely practical p-descent method; Theorem 1.6 overstates what is verified by leaning on BSD.","tokens_in":18863,"tokens_out":3086,"would_cite":true,"duration_ms":27925,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11G05","11Y99","11G40"],"pacs":[],"model":"deepseek-v4-flash","headline":"For an elliptic curve and a primitive Dirichlet character of squarefree order, the p-part of the normalized twisted L-value is conjectured to equal a product of minimal polynomials attached to the Galois eigenspaces of the p-torsion of…","keywords":["Tate-Shafarevich group","twisted L-values","Galois module structure","11-descent","complex multiplication","Dirichlet characters","C5-number fields"],"falsifier":"For the curve $y^2 = x^3 - 262395x + 51731946$, run Algorithm 2 over $\\mathbb{Q}(\\zeta_{11})^+$ and enumerate the full 11-Selmer group from the computed degree-20 unit and class groups, without imposing the analytic order $11^2$ from the BSD formula. Conjecture 1.3 predicts exactly the eigenspace polynomials $x-3$ and $x-4$; an additional Selmer element with an eigenvalue outside $\\{3,4,3^{-1},4^{-1}\\}$, or a total order different from $11^2$, would falsify the conjecture and expose the hidden BSD input.","tokens_in":17629,"feed_emoji":"🔢","tokens_out":12180,"duration_ms":111449,"temperature":0.7,"pith_summary":"This paper proposes a precise bridge between the analytic and algebraic sides of the Birch–Swinnerton-Dyer story: for an elliptic curve $E/\\mathbb{Q}$ and a primitive Dirichlet character $\\chi$ of squarefree order $d$, the prime-$p$ part of the ideal generated by the normalized twisted L-value $L(E,\\chi,1)$ should equal, as an ideal in $\\mathbb{Z}[\\zeta_d]$, a product of minimal polynomials $h_\\theta$ evaluated at the character value, one factor for each irreducible Galois eigenspace of the $p$-torsion of the Tate–Shafarevich group $\\Sha(E/K_\\chi)$. If true, the ideal factorization of twisted L-values becomes a computable Galois-module invariant rather than a transcendental computation. The paper verifies the conjecture in the first nontrivial case — order-5 characters and 11-torsion — for nine complex-multiplication curves over the real cyclotomic field $\\mathbb{Q}(\\zeta_{11})^+$, assuming GRH, and for one further curve by visualization. To do so it develops a practical algorithm for 11-descent over $C_5$ number fields, reducing the needed class-group and unit-group computations from degree 600 to degree-20 subfields.","feed_headline":"Galois eigenspaces predict twisted L-value factorizations","feed_subtitle":"Under GRH, an 11-descent over a C5 field verifies the prediction for nine CM curves, plus one by visualization","key_machinery":"The carrying object is the 11-descent algorithm over a $C_5$ field. It uses the Schaefer–Stoll embedding of the Selmer group into $L^\\times/(L^\\times)^p$ for the étale algebra $L$ attached to the 11-torsion, reduces to the CM case so that $[L:\\mathbb{Q}] = 2(p-1) = 20$, and then uses norm relations to compute the S-unit and S-class groups of $q$ subfields $F_i$ of degree 20 instead of the degree-100 compositum. Proposition 3.10 locates each Galois eigenspace in a specific subfield $F_i$, and Proposition 3.14 uses the Cassels–Tate pairing to force eigenvalues to appear in reciprocal pairs $\\alpha, \\alpha^{-1}$ with equal multiplicity, halving the search space. The output is the list of minimal polynomials $h_\\theta = t - \\alpha_\\theta$ together with the eigenspace dimensions, to be compared with the ideal factorization of $L(E,\\chi)$.","core_discovery":"The central assertion is Conjecture 1.3: for an odd prime $p$, a primitive Dirichlet character $\\chi$ of squarefree order $d$ with conductor coprime to the conductor of $E$, and an embedding $\\iota$ of $\\mathbb{Q}(\\zeta_d)$ into $\\mathbb{C}$, the $p$-part of the principal ideal generated by the normalized value $\\iota^{-1}(L(E,\\chi))$ is predicted to be $\\prod_\\theta (h_\\theta(\\chi(\\tau)^{d/d_\\theta}), p)$, where $\\tau$ generates $\\mathrm{Gal}(K_\\chi/\\mathbb{Q})$, $d_\\theta$ is the order of the roots of $h_\\theta$, and $h_\\theta$ is the minimal polynomial of the matrix giving the action of $\\tau$ on the $\\theta$-isotypic component of $\\Sha(E/K_\\chi)[p]$. The paper's supporting results include Theorem 2.2, showing that $p$ divides the normalized L-value of one of two $p$-congruent curves when the other's twisted L-value vanishes, and Corollary 2.3, transferring this to the eigenspace product under BSD and a visualization hypothesis; this verifies Conjecture 1.3 for the curve 9450du1. For the CM curves, Theorem 1.6 records the GRH-conditional verification for nine curves with $p=11$ and all primitive Dirichlet characters factoring through $\\mathbb{Q}(\\zeta_{11})^+$. In the worked example with the curve $y^2 = x^3 - 262395x + 51731946$, the algorithm outputs eigenspace polynomials $x-3$ and $x-4$, matching the computed factorization $\\mathfrak{p}_1\\overline{\\mathfrak{p}}_1$ of the L-value ideal.","pith_inferences":["A reader should note that the GRH-conditional verification also relies on the Birch–Swinnerton-Dyer conjecture: the worked examples take the analytic order of $\\Sha(E/K_\\chi)[11]$ to be $11^2$, and if BSD failed for one of the nine curves the descent search could be incomplete.","Because all tested characters factor through $\\mathbb{Q}(\\zeta_{11})^+$, every eigenvalue is a fifth root of unity in $\\mathbb{F}_{11}$; the genuinely higher-degree case $d_\\theta > 1$ of Conjecture 1.3 remains untested, and the pairing argument would need the field-extension version sketched in Remark 3.16.","The norm-relations reduction to degree-20 subfields is a general mechanism: the same strategy could make $p$-descent practical for other cyclic extensions and for Jacobians of cyclic covers, not only CM elliptic curves, which would give a larger family of test cases.","If the conjecture is true, the correspondence gives a new way to predict Selmer ranks: for rank-zero curves the factor $(h_\\theta(\\chi(\\tau)^{d/d_\\theta}),p)$ records both the eigenvalues and multiplicities, so the order $|\\Sha(E/K_\\chi)[p]|$ is the product of the degrees of these factors."],"forward_implications":["Conjecture 1.3, if proved, turns the $p$-part of a transcendental L-value into a purely algebraic object: one can compute it by determining the Galois eigenspace decomposition of $\\Sha(E/K_\\chi)[p]$ via descent.","The descent procedure is not tied to $p=11$: it works in principle for any prime $p$ splitting in the CM order with $q \\mid p-1$, and the authors expect 31-descent over a $C_5$ field to become feasible with more computational power.","The visualization theorem constrains what the L-value can look like: when Sha torsion is fully visible, Remark 2.5 shows the ideal must be a power of a single prime $(p)^n$, so examples with richer factorization require descent rather than visualization.","The Cassels–Tate pairing forces the eigenspace decomposition to be symmetric under $\\alpha \\leftrightarrow \\alpha^{-1}$; Conjecture 1.3 therefore automatically respects complex conjugation of the L-value ideal.","For the nine CM curves, the matching of $(h_\\theta(\\chi(\\tau)^{d/d_\\theta}), p)$ with the factorized L-value gives new numerical evidence that BSD-type formulas for Artin twists see the full Galois module structure of Sha, not just its order."],"supporting_citations":[{"why":"Defines the normalized twisted L-value $L(E,\\rho)$, supplies the BSD-type formula for Artin twists, and its Remark 43 is the origin of the conjecture.","marker":"[DEW21]"},{"why":"Provides the p-descent machinery: the embedding of $H^1(G_K,E[p])$ into $L^\\times/(L^\\times)^p$ and the CM version of the construction used in Algorithm 2.","marker":"[SS03]"},{"why":"Gives the exact sequence that computes $R(L,S;p)$ from S-class groups and S-units, which is the computational heart of the descent.","marker":"[PS97]"},{"why":"Supplies the p-adic L-function congruence used in Theorem 2.2 to transfer divisibility between p-congruent curves.","marker":"[GV00]"},{"why":"Supplies the visualization theorem identifying visible elements of the Tate–Shafarevich group in a congruent curve's cokernel, used in the order-5 verification.","marker":"[Fis16]"},{"why":"Supplies the norm-relations method that splits S-unit and S-class computations down to degree-20 subfields, making the descent practical.","marker":"[BFHP22]"},{"why":"Supplies the Cassels–Tate pairing properties used to prove that eigenvalues come in reciprocal pairs with equal multiplicity.","marker":"[PS99]"},{"why":"Supplies integrality of twisted L-values, so the ideal factorization in the BSD-type formula does not depend on BSD.","marker":"[WW22]"}],"fun_headline_variants":["11-descent links Tate-Shafarevich to twisted L-values","Galois eigenspaces dictate L-value factorizations","CM curves verify 11-descent L-value prediction","Twisted L-values read off Galois eigenspaces","11-descent over C5 fields cracks L-value code"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The numerical evidence assumes, beyond the stated GRH hypothesis, that the analytic order of the 11-torsion of the Tate–Shafarevich group supplied by the BSD formula is the true order $11^2$; if BSD fails for any of the tested curves, the descent search could be incomplete and the match with the L-value factorization would not be established.","fun_headline_variants_meta":{"raw":{"variants":["11-descent links Tate-Shafarevich to twisted L-values","Galois eigenspaces dictate L-value factorizations","CM curves verify 11-descent L-value prediction","Twisted L-values read off Galois eigenspaces","11-descent over C5 fields cracks L-value code"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000678,"raw_usage":{"total_tokens":3153,"prompt_tokens":1087,"completion_tokens":2066,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":703,"completion_tokens_details":{"reasoning_tokens":1985}},"tokens_in":703,"tokens_out":2066,"duration_ms":17193,"temperature":1.0,"reasoning_tokens":1985,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T19:56:31.588519+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For the curve $y^2 = x^3 - 262395x + 51731946$, run Algorithm 2 over $\\mathbb{Q}(\\zeta_{11})^+$ and enumerate the full 11-Selmer group from the computed degree-20 unit and class groups, without imposing the analytic order $11^2$ from the BSD formula. Conjecture 1.3 predicts exactly the eigenspace polynomials $x-3$ and $x-4$; an additional Selmer element with an eigenvalue outside $\\{3,4,3^{-1},4^{-1}\\}$, or a total order different from $11^2$, would falsify the conjecture and expose the hidden BSD input.","supporting_citations":[],"review_version":1}