{"id":"50c4287f-46c1-404a-b342-8f088139009f","arxiv_id":"2607.18341","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The Hodge–Weil classes of a general abelian fourfold of discriminant one are algebraic because they are pulled back from the second Chern class of a projective six-dimensional hyperkähler manifold containing its Kummer fourfold.","lead":"This paper gives another proof that all Hodge classes on a general abelian fourfold of Weil type with trivial discriminant are algebraic. The proof embeds the fourfold's Kummer variety into a six-dimensional hyperkähler manifold and pulls back a natural algebraic class.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'complete family' step in §7.4 lacks a dominance argument: the 4D family of Kummer fourfolds may only cover a proper subfamily, so Theorem 0.2's 'very general' quantifier is unsupported.","rationale":"The paper's main theorem (Thm 0.2) is the load-bearing existence result: it must produce, for every very general Weil fourfold of discriminant one, a K3[3] manifold containing its Kummer fourfold. The construction in §5–§7 produces a 4-dimensional family of such pairs by deforming a special pair (K(A×A),M) over the Hodge locus of the anti-invariant lattice BW. The reader's weakest_assumption correctly points to the delicate geometric identification in Prop 6.2 and the deformation in Prop 7.2. My independent reading identifies a different, arguably more central gap: even if Prop 6.2 holds, the proof in §7.4 does not establish that this 4-dimensional family sweeps out the complete moduli of Weil fourfolds. The sentence 'we thus obtain a complete family' is an assertion without a dominance or Torelli argument. The alternative proof's 'conversely' assumes the very identification between periods of the K3-type sub-Hodge structure T_1 and isogeny classes of Weil fourfolds that must be proven. This is a concrete, localizable missing proof, and it directly threatens the 'very general' quantifier of Thm 0.2 and hence the pull-back argument in Thm 0.3. I do not think the paper is wrong; the gap may be fillable by a standard dominance argument or a reference to known results on Weil fourfolds. Since the reader already flagged the proof as conditional, I see no reason to change the verdict, but the specific concern should be made explicit. The proposed differential-rank check is a feasible way to settle whether the family is complete.","tokens_in":43952,"tokens_out":45384,"duration_ms":387958,"concrete_test":"Let P: U_{λ⊕BW} → M_W be the period map sending t to the isogeny class of the abelian fourfold B_t obtained from the Kummer fourfold Z_t⊂M_t by the double-cover/Nakano contraction of Lemma 7.3. At the central point t=0 (the Kummer fourfold of A×A for A the Jacobian of a genus-two curve), compute the differential dP: T_0 U_{λ⊕BW} → T_{[B_0]} M_W. Verify that it has rank 4, i.e., that the four independent period directions in λ⊥ induce four independent deformations of B_0 inside the 4-dimensional moduli of Weil fourfolds of discriminant one. If rank=4, the image is Euclidean open and, by algebraicity/constructibility of the relative Hilbert scheme used in §7.4, contains a Zariski open subset, completing the 'complete family' step. If rank<4, the theorem only covers a proper subfamily and the 'very general' quantifier fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central assertion that Theorem 0.2 covers every very general Weil fourfold B′ with discriminant one rests on the step in §7.4: 'From the four dimensional family of Hodge structures on λ⊥ ... we thus obtain a complete family of Weil type abelian fourfolds...'. What has actually been established is that, for each d and each open U⊂Ω_Λ, Ratner/density gives some t∈U with transcendental lattice λ⊥ and with K(B_t)⊂M_t. It is not shown that the period map from U_{λ⊕BW} (equivalently, from the period domain of λ⊥) to the moduli of Weil fourfolds of field Q(√−d) and discriminant one is dominant or generically finite. Both spaces have dimension 4, but without a rank-4 differential or an algebraic dominance argument the image could be a proper subfamily; a very general B′ could then lie outside the image, and the pull-back of c2(TX) in §1.14 would not be available for it. The alternative proof in §7.4 invokes algebraicity of Y_{λ⊕BW} and the relative Hilbert scheme, but still moves from 'every general K3-type Hodge structure occurs' to 'every general Weil fourfold occurs' via a 'conversely' that assumes the identification between the period of T_1 and the isogeny class of B′. That identification is exactly what needs proof. This is a genuine missing support, not a mere terseness: without it, the construction may only yield a lower-dimensional subfamily, and the Hodge-theoretic conclusion does not follow for the stated 'very general' member.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper gives a new proof of the Hodge conjecture for very general abelian fourfolds of Weil type with imaginary quadratic field and trivial discriminant. The strategy is to show that, up to isogeny, the Kummer fourfold K(B) of such a fourfold B embeds as a submanifold of a projective hyperkähler sixfold X of K3^[3] type whose transcendental Hodge structure has dimension six (Theorem 0.2). Since c_2(T_X) is algebraic and is proportional to the BBF polarization class, the pullback of c_2(T_X) to B is an algebraic Hodge–Weil class; the K-action then forces all Hodge–Weil classes to be algebraic (Theorem 0.3). Sections 1 sets up the Hodge theory; Sections 2–4 study the Barnes–Wall lattice, MBM classes, and birational involutions on K3^[3] manifolds; Sections 5–6 construct a specific birational involution on S^[3] whose fixed locus contains a Kummer fourfold, prove an isomorphism with the Kummer fourfold, and Section 7 deforms the pair to cover very general Weil fourfolds.","tokens_in":44379,"tokens_out":8263,"duration_ms":74547,"significance":"The main theorem was already proved by Markman and by Floccari–Fu, so the paper does not break new ground on the Hodge conjecture itself. Its interest lies in the explicit geometric route: replacing singular OG6 varieties by better-known K3^[3] type manifolds, identifying the fixed locus of a birational involution with a Kummer fourfold, and relating the second Chern class to Hodge–Weil classes. If the construction is correct, it gives a concrete cycle representative and a conceptually different proof of a known result. The paper is long and imports many deep theorems (Global Torelli, MBM classification, Bakker–Lehn, SYZ theorem); its originality is in the combination. However, the current version contains load-bearing gaps, especially in the 'complete family' step and in a computational lattice classification, and therefore cannot be accepted as written.","major_comments":[{"comment":"The step 'From the four dimensional family of Hodge structures on λ⊥ ... we thus obtain a complete family of Weil type abelian fourfolds' is not justified. For fixed d, the construction yields some t in each open U with transcendental lattice λ⊥ and with K(B_t) ⊂ M_t. But the period map from the 4-dimensional domain P(λ⊥)∩Ω_Λ to the 4-dimensional moduli of Weil fourfolds with field Q(√−d) and discriminant one is not shown to be dominant or generically finite. Without a rank-4 differential or an algebraic dominance argument, the image could be a proper subfamily, so a very general Weil fourfold B′ need not be covered. The 'conversely' in the alternative proof of §7.5 assumes precisely the identification between the constructed family and the complete family of Weil fourfolds; this is circular. This gap is load-bearing for Theorem 0.2 and hence for Theorem 0.3.","section":"Section 7.4 and 7.5"},{"comment":"The classification of vectors of square −12 in the Barnes–Wall lattice relies on an unreported Magma computation: 'With Magma we found that there are three orbits ... characterized by its cardinality 61440.' This computation is used to obtain exactly 256 ppMBM classes, which underpins the existence of the 256 disjoint P^3's and the contraction in Proposition 4.4. Similarly, the proof of Corollary 5.6 asserts 'one checks, with a computer' that no extra classes lie in the anti-invariant lattice. No code, output, or reproducible verification is provided. Since these finite lattice statements are checkable, the gap is fixable, but as it stands a central part of the geometric construction depends on unverified computation.","section":"Proposition 3.9 and Corollary 5.6"},{"comment":"The assertion that V cannot be contained in the indeterminacy locus of the birational map S^[3] ⇢ M because it would then be covered by rational curves is not justified as stated. Being contained in the union of rational curves contracted by a sequence of flops does not imply that the subvariety V itself is covered by rational curves: for example, a K3 surface inside P^3 is contained in a uniruled variety but is not covered by rational curves. The indeterminacy locus of a composition of flops can have components of dimension 4, and one would need to prove that any 4-dimensional subvariety contained in such a component is uniruled. The argument that V has Kodaira dimension zero therefore does not rule out the possibility that V lies in the indeterminacy locus. This point is load-bearing for the identification of the fixed component with the Kummer fourfold.","section":"Proposition 6.2, Step 1"}],"minor_comments":[{"comment":"The statement that the pullback of a holomorphic 2-form from X to B is non-zero 'for dimension reasons' would be clearer if the argument were spelled out: a trivial pullback would give a 4-dimensional isotropic tangent subspace of a hyperkähler sixfold, which is impossible.","section":"Section 1.14"},{"comment":"The notation b_X(D,D') should presumably be B_X(D,D'), the BBF form. Also, the sudden introduction of the 16 subsets S_i and the tropes of the Kummer quartic would benefit from a short explanatory paragraph before the enumerative assertion.","section":"Proposition 3.9"},{"comment":"The claim that the anti-invariant lattice is exactly BW is proved by a computer check of the discriminant group. Even if one accepts the computer check, the surrounding text should say which classes are tested and why the check is exhaustive.","section":"Section 5.6"},{"comment":"In the proof of Lemma 7.6, the projection p_2 is said to be flat and hence open. The incidence variety of tangent hyperplanes is not obviously flat over the target projective space; this point needs a reference or a direct argument if the lemma is used.","section":"Section 7.5, Lemma 7.6"},{"comment":"The paper relies on many imported results with differing notational conventions; a table of lattices and their names (BW, Λ, λ⊥, v⊥, etc.) would substantially improve readability.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The dominance gap in §7.4 is the main obstacle: without it, Theorem 0.2 is not established for very general Weil fourfolds, and Theorem 0.3 does not follow. The Magma dependence is also a reproducibility concern. In contrast, the Hodge-theoretic part (Section 1) is mostly sound and clearly written. If the authors can supply a genuine dominance argument or prove that the constructed family identifies with the vG2/Markman complete family, the result is likely correct. I recommend major revision, not rejection, because the gap appears fixable within the paper's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a new proof of Markman's theorem, not a new theorem, and the authors say so plainly. The genuinely new part is Theorem 0.2—an explicit embedding of the Kummer fourfold of a general Weil fourfold into a projective K3[3] sixfold—and the pullback of c2(T_X) to get a Hodge–Weil class. That mechanism is clean and, as far as I can tell, correct: c2 is algebraic on any projective manifold, so once you have the embedding, the Hodge conjecture for these fourfolds follows.\n\nCredit where due: the paper is honest, carefully organized, and uses the heavy machinery (global Torelli, MBM classification, Bakker–Lehn, SYZ) in a way that I could not find a contradiction in. The Barnes–Wall lattice analysis and the geometric description of the fixed locus as a Kummer fourfold are substantive. The self-citations to [vG1] and [vG2] are appropriate; the proof of Lemma 1.6 is a useful shortcut avoiding Kuga–Satake.\n\nSoft spots. Proposition 3.9 relies on a Magma orbit count that is not shipped; the authors add a geometric description of the 256 classes, which helps, but a certificate or code would be better. That is minor. The more serious concern is the step in §7.4, and I think the stress-test note lands. The paper moves from a four-dimensional family of Hodge structures on λ⊥ to a 'complete family' of Weil fourfolds with discriminant one, and then to the conclusion that the very general Weil fourfold appears. That requires dominance of the period map from the moduli of Weil fourfolds to the period domain of the K3-type sub-Hodge structure, or at least an explicit argument that the image is not a proper subfamily. As written, it is a missing step, and it is load-bearing for the 'very general' quantifier in Theorem 0.2. I believe it is repairable: the period map for Weil fourfolds is classical (van Geemen's spinor map, Lombardo), and Remark 7.8 shows the authors know that family. A referee should ask for the step to be written out or for a citation that the period map is dominant. This is a gap in exposition/argument, not a false statement as far as I can tell.\n\nWho this is for: complex geometers working on hyperkähler manifolds, abelian varieties, and the Hodge conjecture. It is a serious paper and deserves a serious referee. My recommendation: send it to peer review, with a request to fix §7.4 and to provide the Magma verification.","headline":"New proof of a known theorem via K3[3] manifolds; the explicit c2 mechanism is nice, but the 'complete family' step in §7.4 needs a dominance argument or explicit citation.","tokens_in":44875,"tokens_out":9440,"would_cite":false,"duration_ms":81124,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14C30","14J42","14K05","14J28"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves the Hodge conjecture for very general abelian fourfolds of Weil type with discriminant one by realizing their Kummer variety inside a projective K3[3] hyperkähler sixfold and pulling back the second Chern class.","keywords":["Hodge conjecture","abelian fourfolds of Weil type","Hodge-Weil classes","hyperkähler sixfolds","K3[3] type","Kummer varieties","second Chern class","Barnes-Wall lattice"],"falsifier":"Compute the analytic germ of the fixed locus at one of the 256 singular points of the contracted variety cM in Proposition 6.2; if it is not isomorphic to the quotient singularity C^4/{±1}, the isomorphism between the fixed component and the singular Kummer fourfold fails, and Theorem 0.2 cannot be true. A more arithmetic falsifier: find a very general Weil fourfold of discriminant one whose Kummer variety does not deform to have a K3-type transcendental lattice of dimension six in H^2; then the embedding of Theorem 0.2 does not exist.","tokens_in":43851,"feed_emoji":"","tokens_out":5606,"duration_ms":43340,"temperature":0.7,"pith_summary":"The paper establishes the Hodge conjecture for the very general abelian fourfold of Weil type with trivial discriminant, providing a new proof via K3[3] type hyperkähler sixfolds. The key is Theorem 0.2: every such fourfold is isogenous to one whose Kummer variety embeds into a projective hyperkähler sixfold X of K3[3] type. Because X is projective, its second Chern class c2(T_X) is algebraic; pulling it back to the abelian fourfold produces a Hodge-Weil class. Since the imaginary quadratic field K acts on the two-dimensional space of Hodge-Weil classes, one algebraic class forces all of them to be algebraic, proving the conjecture.","feed_headline":"A Chern class settles the Hodge conjecture for Weil fourfolds","feed_subtitle":"Pulling back c2 from a K3[3] sixfold makes every Hodge-Weil class algebraic.","key_machinery":"The load-bearing object is the BBF form q_X on H^2(X) of a K3[3] manifold and its associated class q∨_X ∈ Sym^2 H^2(X), which is proportional to the second Chern class c2(T_X). The construction of the embedding uses a birational symplectic involution on a Hilbert scheme S^[3] of a Kummer K3 surface whose anti-invariant lattice is the Barnes-Wall lattice BW; a 4-dimensional component of the fixed locus is the Kummer fourfold K(A×A), and deformation theory over the Hodge locus of BW produces the required family of pairs (K(B), X) with transcendental dimension six.","core_discovery":"The central discovery is that the second Chern class of a K3[3] hyperkähler sixfold is universal for Hodge-Weil classes: for a very general abelian fourfold B' of Weil type with discriminant one, there is an embedding of the Kummer variety K(B) of an isogenous B into a projective K3[3] manifold X, and the pull-back of c2(T_X) is a non-zero Hodge-Weil class in HW(B',K). The proof uses the Beauville-Bogomolov-Fujiki form: its associated class q∨_X in H^4(X) is proportional to c2(T_X), and its restriction defines a polarization class on a K3-type sub-Hodge structure T_1 ⊂ H^2(B'), which Proposition 1.11 identifies as a Hodge-Weil class. Since the K-action spans HW(B',K) from any non-trivial ele","pith_inferences":["The paper hints at a concrete way to construct explicit surfaces in B whose classes are not complete intersections: take a global section of N_0 ⊗ L for sufficiently ample L; the zero locus gives a 2-cycle representing a multiple of the Hodge-Weil class.","The proof likely extends to any family of hyperkähler manifolds where the fixed locus of a symplectic birational involution contains a Kummer fourfold with the right Hodge numbers; the key numerical input is the proportionality q∨_X = λ c2(X), which holds for all known hyperkähler types.","The use of ergodic-theoretic density results is shown to be replaceable by a purely lattice-theoretic argument, which may make the approach more accessible to algebraic geometers and amenable to effective versions over number fields.","The discriminant-one hypothesis is exactly where T splits as T_1 ⊕ xT_1; for non-trivial discriminant the quaternion algebra is a division algebra and the argument collapses, suggesting the Hodge conjecture for non-trivial discriminant needs a genuinely different mechanism."],"forward_implications":["If correct, the Hodge conjecture holds for all very general abelian fourfolds of Weil type with trivial discriminant.","The algebraic cycles representing Hodge-Weil classes can be taken as pullbacks of c2(T_X), giving an explicit, if indirect, description.","The same proof shows the pulled-back c2 lies in the subspace of H^4(B) generated by algebraic classes and is not an intersection of divisor classes.","The method fails for non-trivial discriminant: the Hodge structure T has no K3-type sub-Hodge structure, so no such embedding can exist.","The approach suggests that any algebraic class in the image of the cup product Sym^2 T_1 → H^4(B) is Hodge-Weil, and the K-action then propagates algebraicity."],"fun_headline_variants":["Chern class c2 proves Hodge for Weil fourfolds","One K3[3] Chern class settles Hodge-Weil","Pullback of c2 makes Hodge classes algebraic","Hodge conjecture for Weil fourfolds via c2","K3 sixfold's c2 closes Hodge-Weil case"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The proof requires that the 4-dimensional component of the fixed locus of the involution is isomorphic, not just birational, to the Kummer fourfold of A×A; if that identification failed away from an open set, the inclusion of the transcendental lattice of X into H^2(B) and the pull-back argument would break.","fun_headline_variants_meta":{"raw":{"variants":["Chern class c2 proves Hodge for Weil fourfolds","One K3[3] Chern class settles Hodge-Weil","Pullback of c2 makes Hodge classes algebraic","Hodge conjecture for Weil fourfolds via c2","K3 sixfold's c2 closes Hodge-Weil case"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000209,"raw_usage":{"total_tokens":1225,"prompt_tokens":706,"completion_tokens":519,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":450,"completion_tokens_details":{"reasoning_tokens":433}},"tokens_in":450,"tokens_out":519,"duration_ms":4540,"temperature":1.0,"reasoning_tokens":433,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T17:34:43.812610+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the analytic germ of the fixed locus at one of the 256 singular points of the contracted variety cM in Proposition 6.2; if it is not isomorphic to the quotient singularity C^4/{±1}, the isomorphism between the fixed component and the singular Kummer fourfold fails, and Theorem 0.2 cannot be true. A more arithmetic falsifier: find a very general Weil fourfold of discriminant one whose Kummer variety does not deform to have a K3-type transcendental lattice of dimension six in H^2; then the embedding of Theorem 0.2 does not exist.","supporting_citations":[],"review_version":1}