Abelian sixfolds: external review English version
User-supplied external review ·Reviewer not identified
Conditional Review of the Twenty-One Cases
The rational Hodge conjecture for abelian sixfolds
Manuscript: sixfold_hodge_revised, research draft dated October 8, 2026.
Typeset October 11, 2026; review date not supplied. This document presents the user-supplied review and does not constitute an
additional review.
Central finding: assuming all OpenAI external assertions explicitly stated in the manuscript are correct,
this review found no major mathematical gap that would invalidate the main conclusions in the twenty-one
cases.
The key steps of the main proofs currently fit together. No gap was found that would require an additional unknown
algebraicity theorem or replacement of a principal construction. This is not formal certification of every formula or
auxiliary script, but neither is it acceptance based merely on the manuscript’s final “Proved” labels.
1. Review assumptions: four external assertions
The review assumes the following assertions hold with the strength stated in the manuscript:
1. Algebraicity of split eightfold Weil classes, as stated in P .9.
2. The rational Hodge conjecture for CM abelian varieties in arbitrary dimension.
3. Algebraicity of the specified complete even-Clifford Kuga–Satake correspondence, stated on manuscript page 51.
4. Algebraicity of the Aux degree-four spin tensor, stated on manuscript pages 44–45.
The last assertion is a separate input; general algebraicity of Kuga–Satake correspondences alone cannot replace it. The
manuscript explicitly distinguishes these inputs. The review examines the author’s subsequent deductions under these
assumptions, rather than rechecking the external assertions themselves.
2. The five formerly most difficult cases
Case Decisive step Review finding
S7 Start with a Weil plane for an imaginary quadratic subfield; use quaternionic
endomorphisms and sum maps to generate the seven-dimensional exceptional
space and volume invariants on powers
No major gap found
S8 Determine the full or special unitary group; use P .11 in the balanced case for
every discriminant
No major gap found
S10 P .22 determines the derived group and determinant torus; P .11 handles the ex-
ceptional Weil plane and 9.3 passes to powers
No major gap found
N2 Aux gives a nonzero mixed cycle involving the fourfold and an auxiliary CM
factor; a CM correspondence transfers it to the prescribed surface
No major gap found ; Aux is es-
sential
N3 Construct a K3-type structure, extract matrix components of the specified KS
correspondence in the complete Clifford module, and contract to a nonzero
mixed cycle
No major gap found; requires the
specified complete KS correspon-
dence
S7, S8 and S10 share P .11. N2 and N3 construct their mixed cycles separately; their arguments cannot simply be
replaced by the conclusions for the first three cases. The principal checks follow.
1

Abelian sixfolds: external review English version
2.1 P.11: recovering an arbitrary sixfold Weil plane from a split eightfold
Let A be a sixfold of Weil type with Hermitian determinant d < 0. Add two CM elliptic curves of opposite k-types and
give their product P the Hermitian form
hP = diag(1, d).
Then
X = A × P, sig(hX ) = (4 , 4), det(hX ) = d2 ∈ Nk/Q(k×).
P .10 establishes hyperbolicity, so the hypotheses of the external eightfold theorem are met.
After obtaining algebraicity of Wk(X), the manuscript uses an actual algebraic correspondence:
Cβ(γ) = ( prA)∗
(
γ ∪ pr∗P β), 0 ̸= β ∈ Wk(P ) ⊂ NS(P )Q.
The contractions on both conjugate components are nonzero. Hence
Cβ : Wk(X)
∼
− →Wk(A).
The cycle codimension is correctly 4+1 −dim P = 3. There is no missing algebraic correspondence hidden behind
an abstract Hodge isomorphism at this step. The quaternionic generation argument in S7 and the determinant-
invariant arguments in S8 and S10 can therefore proceed.
2.2 N2: the Aux hypotheses, nonvanishing and CM transfer
The auxiliary quadratic space in Theorem 13.19 satisfies
F = K + is real quadratic, dimF V ′ = 13 ≡ 5 ( mod 8),
and sig(V ′σ) = (2 , 11) at every real embedding σ. Local isotropy and Hasse–Minkowski split a hyperbolic plane. The
complement still has real signature (1, 10), so a second hyperbolic plane splits. Thus the Witt-index requirement in
Aux is genuinely verified, not merely its dimension requirement.
The reviewer independently recomputed the exterior-algebra operation in N2.4:
CU (e13 ⊗ e23) = e3 ⊗ e123 − e123 ⊗ e3.
Antisymmetrizing the input still gives a nonzero output. Consequently, the required nonvanishing
0 ̸= z ∈ TB ⊗ H 2(D, Q)
is not an unchecked assertion. The manuscript also explains why Clifford operators in the other directions cannot cancel
this projection, and how nonvanishing for complex selectors implies the existence of rational selectors.
The CM theorem is applied to D × P , not incorrectly to B × P , which contains a non-CM factor. It realizes the selected
Hodge map
φ : H 2(D, Q) − →H 2(P, Q)
by an algebraic correspondence and transfers the previously obtained cycle to B × P . Since φ is an isomorphism on
the relevant simple substructure, (1 ⊗ φ)z ̸= 0 . No problem was found with either nonvanishing or the codimension
calculation in this transfer.
2.3 N3: extracting complete KS components and a nonzero contraction
The constructed K3-type structure satisfies
dimQ W = 8, h (W ) = (1 , 6, 1), sig(q) = (2 , 6).
The absence of rational (1, 1) classes prevents an additional Picard-number increase when W is realized as K3 tran-
scendental cohomology. The lattice inequality
ℓ(AΛW ) + 2 ≤ 10 < 14 = 22 − 8
2

Abelian sixfolds: external review English version
meets the cited Nikulin criterion. No omitted lattice-embedding condition was found.
Next, det L = u−2 and t = u−1 provide the Spin lift of the actual joint Hodge group. The complete Clifford regular
module is used, and rational abelian factors are obtained through scalar extension of rational Hom spaces. The argument
does not simply treat an individual complex half-spin representation as a rational abelian factor of the same dimension.
The reviewer checked the contraction in N3.4–N3.6:
α1(e1) = b1u−1, α 1(e2) = b2u−1,
α2(f1) = b2v−1, α 2(f2) = −b1v−1.
Contracting against q−1 and pulling back along the diagonal of B yields a nonzero multiple of
(b1 ∧ b2) ⊗ u−1 ⊗ v−1.
Since det V1 = uv, this lies in the target mixed invariant space. The algebraic transcendental projector on the K3
surface implements the contraction, and bilinearity gives descent to a rational cycle. Each of these steps is supported
by an argument. N3 does not currently lack the crucial step converting the KS correspondence into the required
target cycle.
3. The original sixteen cases
For S1–S6, S9, S11, N1 and N4–N10, this review likewise found no major mathematical gap in the main
proofs.
S1–S6. The main checks concern completeness of the low-dimensional representation candidates, correct exclusion of
diagonal couplings, and descent from quadratic invariant generation to rational divisor classes. No apparent mismatch
was found between P .2–P .4 and their applications.
S9 and S11. The polarization adjustments do more than alter real signatures. P .17 calculates the resulting rational norm
class; P .18 explicitly constructs a three-dimensional totally isotropic k-subspace. The splitting hypothesis required by
Markman’s split-sixfold theorem is therefore actually proved.
The S11 rank bound. The reviewer checked Lemma 11.6’s exclusion of a four-dimensional orbit space and indepen-
dently recomputed the three finite models in Remark 11.8. Their stabilizer orders are 96, 32, 96, and their orbit-span
dimensions are 3, 4, 1, respectively, agreeing with the manuscript. This supports exclusion of rank four. Algebraicity
of the Weil classes still comes from P .18, not from the rank computation alone.
N4, the 3 + 3 case. Lemma 15.2 does not merely compare abstract isogeny classes of tori: it tests whether the actual
Hodge cocharacters satisfy a proposed common-quotient relation, in particular excluding the potential two-dimensional
CM coupling. The remaining one-dimensional determinant relation produces precisely the stated mixed Weil plane,
handled by odd-rank polarization adjustment.
N1, N5 and N6. Mixed classes are treated by odd-rank adjustment or pullbacks of fourfold Weil classes. N7–N9 use
the theorem for products of low-dimensional factors; N10 uses quadratic invariant generation for elliptic products. The
argument does not infer the Hodge conjecture for arbitrary products merely from the conjecture for each individual
factor.
4. Two necessary boundaries
First: the sixfold itself versus all self-powers. The global theorem concerns the rational Hodge conjecture for every
abelian sixfold itself. Assertions for powers must remain restricted to the scope explicitly proved in the individual sec-
tions. The global theorem does not additionally establish the conjecture for all An for every sixfold A. The manuscript’s
conclusion already makes this distinction.
Second: no major gap found versus certification of all details. The finding is “no major gap found” , not “all
details certified.” The reviewer did not reprove the external theorems or run every verification script mentioned in the
manuscript.
3

Abelian sixfolds: external review English version
Final assessment: under the specified external assumptions, this version can be regarded as a complete
conditional proof draft covering the twenty-one cases. This review found no remaining major mathematical
gap in the author’s main proofs. In particular, withdrawal of the external results alone is not a reason to
classify S7, S8, S10, N2 or N3 as lacking an internal conditional proof.
Source pointers and provenance
• Reviewed manuscript: the user-supplied Hodge classes on abelian sixfolds: constructions and a conditional proof
(sixfold_hodge_revised). Proposition numbers and page references follow the supplied review.
• K3 lattices and periods: Daniel Huybrechts, public K3 lecture notes .
• Markman’s cited results: arXiv:2502.03415.
• Products of low-dimensional factors: Ramón Marí, arXiv:math/0505357.
Editorial note: this is an English translation of the user-supplied external review. Repeated file-reference placeholders have been
removed and identifiable links collected above. Red highlighting marks the central findings together with their external-assumption
and review-scope qualifications.
4