{"id":"f9b92a3a-acfb-4e26-adc1-d472f9c612cf","arxiv_id":"2506.13729","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For an etale abelian cover of curves, the Weil Hodge classes on the associated Prym variety are represented by algebraic cycles.","lead":"This paper proves that certain Hodge classes on Prym varieties attached to abelian covers of curves are algebraic. It reframes Chad Schoen's cycle construction through geometric class field theory, which generalizes his result from cyclic covers to all abelian covers.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 4.2's Schur-lemma argument is invalid: the G0-action is misdefined and lower Kunneth summands share nontrivial characters with U, so the identification of U with U_Weil is unsupported.","rationale":"The reader's weakest_assumption is exactly the step I would stress-test. I agree with the CONDITIONAL verdict. My reading finds the objection is stronger than 'under-justified': the proof of Lemma 4.2 contains a group-action error and a misapplication of Schur's lemma. The subgroup G0 as defined lies in N subset H, hence acts trivially on H^h(W,Q), contradicting the claim that it induces the standard G-action on U. Even repairing G0 to the diagonal subgroup, the lower Macdonald summands can contain nontrivial characters of G (for g(C') = 3, h = 4, G = Z/3, the map delta -> delta^2 is bijective on characters), so a non-faithful action does not imply the absence of common constituents with U. Since Lemma 4.2 is what places phi*U inside the top Kunneth factor and hence identifies U with U_Weil through Abel-Jacobi and Lefschetz operators, Theorem 1.2 is not established by the present argument. The rest of the paper, including the spectral sequence computation in Theorem 1.1 and the explicit construction of algebraic cycles, appears sound, and the claimed theorem may well be true. This is therefore a conditional-accept situation rather than a rejection; a concrete small-case computation would settle whether the lemma's conclusion itself holds.","tokens_in":14266,"tokens_out":38709,"duration_ms":402497,"concrete_test":"Specialize to C' of genus 3 with G = Z/3 (so h = 4), where the group-theoretic obstruction is clearest. Compute explicitly the class of the algebraic cycle phi^{-1}(Q_0) in H^4(Sym^4 C,Q), or the difference phi*[Q_0] - phi*[Q_1], using the known tautological/intersection ring of Sym^4 C, and decompose it via Macdonald's formula H^4(Sym^4 C) = /\\^4 H^1(C) (+) /\\^2 H^1(C)(-1) (+) Q(-2). If the /\\^2 H^1(C)(-1) component is nonzero, Lemma 4.2 is false and Theorem 1.2's identification breaks; if it vanishes, the lemma's conclusion survives and should be reproved by a componentwise calculation that avoids Schur's lemma and the incorrect G0-action.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma 4.2 is the load-bearing step: it asserts phi*U lies in the top Kunneth factor /\\^h H^1(C,Q), which is what lets the authors identify U with U_Weil via Abel-Jacobi and Lefschetz operators (Theorem 1.2 and diagram 4.3.1). The proof is not merely under-justified; it is incorrect as written. First, the subgroup G0 = {(g,-g,0,...,0)} is contained in N = {sum t_k = 0}, hence in H, so it acts trivially on H^h(W,Q) = H^h(C^h,Q)^H. It therefore cannot induce the standard G-action on U, contradicting the text. Second, even if G0 is replaced by the diagonal subgroup, Schur's lemma does not give the claimed vanishing: U is reducible, and a non-faithful action can still contain nontrivial characters. For example, when g(C') = 3, h = 4 and G = Z/3, the Macdonald summand /\\^2 H^1(C,Q)(-1) has nontrivial G0-character constituents (e.g. character 27 + 24lambda + 40lambda^2 under the stated G0), and under the diagonal action the map delta -> delta^{h-2i} is an isomorphism of the character group. Thus the lower factors share characters with U, and 'not faithful' does not imply absence of common irreducible constituents. Consequently the projection of phi*U to lower Kunneth factors is not shown to vanish, and Theorem 1.2's identification of U with U_Weil is unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies algebraic cycles on Prym varieties associated with etale abelian covers C -> C1 of smooth projective curves over C. The authors use geometric class field theory to construct an abelian cover W of the symmetric product Sym^h(C1), with h = 2g(C1) - 2, and prove (Theorem 1.1) a description of the cohomology of W: it agrees with that of Sym^h(C1) except in middle degree, where an additional |G|-1 dimensional Hodge substructure U generated by algebraic cycles appears. They then claim (Theorem 1.2) that the Weil-type Hodge classes U_Weil = Λ^h_{Q[G]^nt} H^1(B,Q) inside H^h(B,Q) of the associated Prym variety B are algebraic, by identifying U with U_Weil through Abel-Jacobi maps and Lefschetz operators. If correct, this extends Schoen's theorem for cyclic covers and primitive characters to arbitrary abelian covers and all nontrivial characters.","tokens_in":14612,"tokens_out":21124,"duration_ms":202073,"significance":"The potential significance is substantial: Theorem 1.2 would establish algebraicity of the full non-trivial-character part of the cohomology of Prym varieties attached to abelian covers, generalizing Schoen's cyclic primitive-character result. The paper also offers a conceptual reinterpretation via unramified geometric class field theory, and the proof of Theorem 1.1 via the Leray spectral sequence with explicit cycle classes Q_t is concrete and appears sound. The main result, however, rests on Lemma 4.2, whose proof contains a false group-action assertion and an invalid Schur's lemma argument. Since the identification of U with U_Weil is not established, the central algebraicity theorem is currently unsupported.","major_comments":[{"comment":"The subgroup G0 = {(g,-g,0,...,0)} is contained in N = {Σ t_k = 0}, hence in H = N S_h, so it acts trivially on H^h(W,Q) = H^h(C^h,Q)^H. The sentence claiming that the composition G0 -> G^h S_h -> \\tilde H/H ≅ G is an isomorphism and that the induced G0-action on U is the standard G-action is therefore false: the image of G0 in \\tilde H/H is zero. This invalidates the equivariance framework used in the proof of Lemma 4.2.","section":"§4.2 (diagram 4.1.2 and following paragraph)"},{"comment":"Even if G0 is replaced by a subgroup that does induce the G-action, the Schur's lemma step is invalid. U is not an irreducible G-representation in general, and a non-faithful action on a Kunneth summand does not preclude common irreducible constituents with U. For instance, when G = Z/3 and g(C1) = 3 (so h = 4), the summand Λ^2 H^1(C,Q)(-1) of H^4(Sym^4 C,Q) contains a nontrivial character: for v in the trivial part H^1(C1,Q) and w in the χ-isotypic part of H^1(C,Q), the class v∧w transforms by χ under the diagonal action and by χ^{-1} under the stated G0-action, and such characters also occur in U. Hence the projection of φ^*U to a lower Kunneth factor is not shown to vanish, and the containment φ^*U ⊆ Λ^h H^1(C,Q) is not proved.","section":"§4.3, Lemma 4.2"},{"comment":"The identification of φ^*U with Φ^*U_h and the final conclusion of Theorem 1.2 depend on Lemma 4.2. Since Lemma 4.2 is not established, the chain of isomorphisms in diagram (4.3.1) does not prove that the Weil Hodge classes U_Weil are algebraic. A corrected proof of Lemma 4.2, or a different argument for containment of φ^*U in the top Kunneth factor, is required.","section":"§4.4, Theorem 4.3 and diagram (4.3.1)"}],"minor_comments":[{"comment":"The notation 'Q' for the special fiber P^{g-1} is confusing because Q is also the coefficient field; please use a different symbol such as P.","section":"§3.2, Proposition 3.2"},{"comment":"The section heading contains a typo: 'Prym V ariety' should be 'Prym Variety'.","section":"§4 heading"},{"comment":"The notation \\bar σ is not defined; please specify the complex conjugation action on Hom_Q(F_i,C).","section":"§2.3, Lemma 2.6"},{"comment":"The construction of the morphism φ via the universal property of the fiber product is only sketched in one sentence; a slightly more detailed explanation would improve readability.","section":"§4.1, Lemma 4.1"}],"recommendation":"major_revision","confidential_remarks":"The paper has a promising idea and a clean proof of Theorem 1.1, but the proof of the main theorem is currently not valid. The false statement about G0 is not a mere typo: it is the foundation of the equivariance argument in Lemma 4.2. I recommend a major revision focused on repairing that lemma. If the authors cannot supply a correct proof of the containment φ^*U ⊆ Λ^h H^1(C,Q), the manuscript should not be accepted."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [colleague],\n\nRead the Patel–Zhang paper. It extends Schoen's theorem from cyclic primitive-character covers to all finite abelian covers and all nontrivial characters. That is a real mathematical step, and the geometric class field theory framing is a genuinely useful way to see why the cycles exist. Theorem 1.1's Leray spectral sequence computation looks coherent, and the construction of the algebraic cycles Q_t is explicit and checkable.\n\nThe soft spot is Lemma 4.2, and it is load-bearing. The subgroup G0 = {(g,-g,0,...,0)} lies in N, hence in H, so it acts trivially on H^h(W,Q). It cannot induce the standard G-action on U. That looks like a typo—possibly they meant (g,0,...,0)—but the Schur's lemma argument is not right even after that correction. U is reducible, and a non-faithful action on a lower Kunneth summand does not imply the summand has no characters in common with U. For G = Z/3 and h = 4, the ^2 H^1(C) summand contains nontrivial characters, so the projection of phi*U to that summand is not shown to vanish. The containment phi*U ⊂ ^h H^1(C) is therefore not established.\n\nThis is not a cosmetic complaint. The identification of U with U_Weil in Theorem 1.2 passes through that containment via diagram (4.3.1). If Lemma 4.2 fails, the chain of maps does not identify the two Hodge structures. I don't think the final theorem is false—Schoen's result and the overall framework make it plausible—but the proof as written has a real gap at exactly the point where the main theorem needs support.\n\nThe rest of the paper is solid. The representation theory setup is standard, the comparison with Schoen's setting in Section 5 is helpful, and the references look appropriate. The paper is worth serious referee time. A referee should ask for a corrected Lemma 4.2, likely via a direct character computation on the Macdonald summands, and then re-check the isomorphism in (4.3.1).\n\nRecommendation: send to peer review, with the lemma as the main issue.","headline":"A serious generalization of Schoen's theorem with a genuine new framing, but Lemma 4.2's proof is wrong as written and it is the load-bearing step.","tokens_in":15148,"tokens_out":12530,"would_cite":false,"duration_ms":112641,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14C30","14K22","14H40"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that all top-wedge Hodge classes on Prym varieties attached to abelian covers of curves are algebraic cycles.","keywords":["Hodge conjecture","algebraic cycles","Prym varieties","abelian covers","geometric class field theory","Weil classes","symmetric products","Abel-Jacobi maps"],"falsifier":"Compute, for a small abelian cover such as a $\\mathbb{Z}/3$-cover of a genus-3 curve, the decomposition of $H^4(\\mathrm{Sym}^4 C,\\mathbb{Q})$ by the diagonal subgroup $G_0$; if the projection of $\\phi^*U$ to any summand $\\wedge^{4-2i}H^1(C,\\mathbb{Q})$ with $i>0$ is nonzero, Lemma 4.2 and the identification with $U_{\\mathrm{Weil}}$ fail. A direct dimension count of $H^4(W,\\mathbb{Q})$ for the same cover would also test the predicted decomposition of Theorem 1.1.","tokens_in":14053,"feed_emoji":"","tokens_out":11544,"duration_ms":103817,"temperature":0.7,"pith_summary":"This paper proves that certain Hodge classes on Prym varieties are algebraic, meaning they are the cohomology classes of actual algebraic cycles. Starting from an étale cover $C \\to C'$ of smooth projective curves with finite abelian Galois group $G$, the Prym variety $B$ is the part of the Jacobian of $C$ on which $G$ acts through non-trivial characters. The paper shows that the full top exterior power of $H^1(B,\\mathbb{Q})$ over the non-trivial part of the group ring, a Hodge substructure of dimension $|G|-1$, is generated by algebraic cycles. This removes the cyclic and primitive-character restrictions from an earlier result in the literature. If correct, it supplies algebraic representatives for a natural family of Hodge classes that are generally not divisor classes.","feed_headline":"Prym Hodge classes from abelian covers are algebraic","feed_subtitle":"Top exterior powers of the non-trivial character parts are shown to be algebraic cycles.","key_machinery":"The machine that drives the argument is the base-change diagram from unramified geometric class field theory: because $C \\to C'$ is abelian, it is pulled back from an isogeny $A \\to J(C')$, so there is a commutative square in which the cover $W \\to \\mathrm{Sym}^h(C')$ sits over $A \\to J(C')$. The distinguished object is the special fiber $\\mathbb{P}^{g(C')-1} = |K_{C'}|$ of $AJ_h$ over the canonical point of $J(C')$; in $W$ its inverse image is $|G|$ disjoint projective spaces indexed by the group. Their classes, after quotienting by the diagonal class, form the summand $U$ in the computed decomposition $H^*(W,\\mathbb{Q}) \\cong H^*(\\mathrm{Sym}^h(C'),\\mathbb{Q}) \\oplus U(-h/2)$. Lemma 4.2 then uses the diagonal subgroup $G_0 \\subset G^h$ acting on $C^h$, together with Schur's lemma, to show that $\\phi^*U$ sits inside the top Künneth factor $\\wedge^h H^1(C,\\mathbb{Q})$ of $H^h(\\mathrm{Sym}^h C,\\mathbb{Q})$; Macdonald's symmetric-product formula, the Abel-Jacobi map, and the Lefschetz operator complete the transfer of algebraicity from $U$ to $U_{\\mathrm{Weil}}$.","core_discovery":"The central claim is Theorem 1.2: for every étale abelian cover $C \\to C'$ with $g(C') \\ge 2$ and $h = 2g(C') - 2$, the subspace $U_{\\mathrm{Weil}} = \\Lambda^h_{\\mathbb{Q}[G]^{\\mathrm{nt}}} H^1(B,\\mathbb{Q}) \\subset H^h(B,\\mathbb{Q})$ consists of Hodge classes and is represented by algebraic cycles. Over $\\mathbb{C}$ this subspace is spanned, one vector per non-trivial character $\\chi$, by the wedge of the $h$ copies of the $\\chi$-isotypic component; it is a Hodge substructure by a pairing of conjugate Hodge types, and it has dimension $|G|-1$. The proof constructs the cycles on an auxiliary space: the Abel-Jacobi map from $\\mathrm{Sym}^h(C')$ to $J(C')$ has a special fiber $\\mathbb{P}^{g(C')-1}$ over the canonical class, and its preimage in the abelian cover $W$ of $\\mathrm{Sym}^h(C')$ is a union of $|G|$ projective spaces whose classes, modulo their $G$-invariant sum, span exactly the summand $U$ appearing in Theorem 1.1. Pulling $U$ back to $\\mathrm{Sym}^h(C)$, pushing forward by the Abel-Jacobi map to $J(C)$, and applying powers of the Lefschetz operator identifies $U$ with $U_{\\mathrm{Weil}}$ on the Prym variety; the algebraicity of the Lefschetz inverse on abelian varieties then makes the classes algebraic.","pith_inferences":["The paper does not spell this out, but the same class-field-theory base change should carry the construction into the ramified setting with Jacobians with modulus, where the special fiber would be replaced by a corresponding Brill-Noether-like locus; this would give algebraic representatives for Prym-type classes attached to ramified abelian covers.","A direct consequence of the argument is that algebraicity of these classes is a formal consequence of the base-change diagram plus the Lefschetz standard conjecture on abelian varieties; one would not expect it to depend on special properties of cyclotomic fields, unlike the cyclic examples that motivated it.","One testable extension is to realize the same cycles in $\\ell$-adic cohomology and verify that they are Tate classes under a finite-field analogue of the abelian cover; the paper notes the proofs adapt to étale cohomology, so such a check would be a natural next step."],"forward_implications":["For every étale abelian cover with base genus at least 2, the full nontrivial-character Weil class space $U_{\\mathrm{Weil}}$ is algebraic, so the Hodge conjecture holds on the Prym variety for this entire subspace.","The result covers all nontrivial characters at once; in cyclic covers whose order is not prime these include characters that the earlier primitive-character theorem did not reach.","The same proof, as the paper notes for an arbitrary rational representation $V$ of $G$, produces algebraic cycles for the corresponding top-wedge classes on each isotypic abelian subvariety $J(C)_V$.","Because the Lefschetz standard conjecture for abelian varieties is already known, the algebraicity conclusion transfers through the Lefschetz operator without needing the full Hodge conjecture."],"supporting_citations":[{"why":"Supplies the cyclic primitive-character theorem and the method of constructing algebraic cycles that this paper extends to all abelian covers.","marker":"[11]"},{"why":"Source for the geometric class field theory result behind the base-change diagram (Theorem 2.1).","marker":"[14]"},{"why":"Macdonald's decomposition of the cohomology of symmetric products is used to locate the pulled-back classes.","marker":"[7]"},{"why":"Kleiman's proof of the Lefschetz standard conjecture for abelian varieties lets the paper invert powers of the Lefschetz operator algebraically.","marker":"[4]"},{"why":"Used in Lemma 2.6 to show the top exterior power over the group ring consists of Hodge classes.","marker":"[10]"},{"why":"Used in Theorem 2.1 for the statement that Abel-Jacobi maps induce isomorphisms on abelianized fundamental groups.","marker":"[3]"},{"why":"Gives the projective-bundle description of Abel-Jacobi maps at degree $2g-2$, including the special fiber over the canonical class.","marker":"[13]"},{"why":"Provides the construction of abelian subvarieties from rational representations, used to define the Prym variety and its isotypic factors.","marker":"[6]"}],"fun_headline_variants":["Prym Hodge classes proven algebraic via abelian covers","Geometric class field theory yields algebraic Prym Hodge classes","Abelian covers make Prym Hodge classes algebraic","Algebraicity of Hodge classes on generalized Pryms shown","Class field theory reveals algebraicity of Prym Hodge classes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that pulling the algebraic classes $U$ back to the symmetric product of the covering curve places them in the very top cohomology slice and nowhere else; the proof infers this from how the deck-transformation group acts on the lower slices, and if any lower slice shared the same group characters the identification with the Prym's Weil classes would break.","fun_headline_variants_meta":{"raw":{"variants":["Prym Hodge classes proven algebraic via abelian covers","Geometric class field theory yields algebraic Prym Hodge classes","Abelian covers make Prym Hodge classes algebraic","Algebraicity of Hodge classes on generalized Pryms shown","Class field theory reveals algebraicity of Prym Hodge classes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000353,"raw_usage":{"total_tokens":1911,"prompt_tokens":924,"completion_tokens":987,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":540,"completion_tokens_details":{"reasoning_tokens":904}},"tokens_in":540,"tokens_out":987,"duration_ms":8664,"temperature":1.0,"reasoning_tokens":904,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:00:04.989285+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for a small abelian cover such as a $\\mathbb{Z}/3$-cover of a genus-3 curve, the decomposition of $H^4(\\mathrm{Sym}^4 C,\\mathbb{Q})$ by the diagonal subgroup $G_0$; if the projection of $\\phi^*U$ to any summand $\\wedge^{4-2i}H^1(C,\\mathbb{Q})$ with $i>0$ is nonzero, Lemma 4.2 and the identification with $U_{\\mathrm{Weil}}$ fail. A direct dimension count of $H^4(W,\\mathbb{Q})$ for the same cover would also test the predicted decomposition of Theorem 1.1.","supporting_citations":[{"cited_title":"Hodge classes on self-products of a variety with an automorphism.Compositio Math., 65(1):3–32, 1988","cited_arxiv_id":null,"evidence_quote":"Supplies the cyclic primitive-character theorem and the method of constructing algebraic cycles that this paper extends to all abelian covers."},{"cited_title":"Springer-Verlag, New York, 1988","cited_arxiv_id":null,"evidence_quote":"Source for the geometric class field theory result behind the base-change diagram (Theorem 2.1)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Macdonald's decomposition of the cohomology of symmetric products is used to locate the pulled-back classes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Kleiman's proof of the Lefschetz standard conjecture for abelian varieties lets the paper invert powers of the Lefschetz operator algebraically."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Used in Lemma 2.6 to show the top exterior power over the group ring consists of Hodge classes."},{"cited_title":"The geometry and fundamental group of permutation products and fat diagonals","cited_arxiv_id":null,"evidence_quote":"Used in Theorem 2.1 for the statement that Abel-Jacobi maps induce isomorphisms on abelianized fundamental groups."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the projective-bundle description of Abel-Jacobi maps at degree $2g-2$, including the special fiber over the canonical class."},{"cited_title":"Rodr´ ıguez.Decomposition of Jacobians by Prym varieties, volume 2310 ofLect","cited_arxiv_id":null,"evidence_quote":"Provides the construction of abelian subvarieties from rational representations, used to define the Prym variety and its isotypic factors."}],"review_version":2}