Documentary state: EDITORIALLY_ADMITTED. The manuscript has a protected editorial admission, not a theorem certification.
OAI-26 family 050: a counterexample to Griffiths' positivity conjecture
Editorial status: SOURCE_CRITIC_COMPLETE; protected acceptance is determined by the admissions ledger. Mathematical status: NO_CERTIFICATION. This article reports the source claim, not independent verification of the construction.
The question in plain language
A vector bundle attaches a vector space to every point of a geometric space. An ample vector bundle satisfies an algebraic positivity condition. Griffiths' positivity conjecture proposed that such positivity should be expressible through a smooth Hermitian metric whose curvature is strictly positive in the Griffiths sense.
The pinned Family 050 overview claims counterexamples: ample rank-two holomorphic vector bundles on the quadric surface that admit no smooth Hermitian metric with strictly Griffiths-positive curvature. The assertion separates algebraic positivity from this particular differential-geometric notion of positivity.
A small example
For a complex vector bundle, a Hermitian metric measures lengths and angles in each fiber. Its curvature describes how these measurements vary over the base. Griffiths positivity requires a strict positive curvature inequality when evaluated against nonzero tangent directions and nonzero bundle vectors.
Ample line bundles admit positive-curvature metrics in the familiar projective setting. A rank-two bundle has more directions to compare, so extending intuition directly from line bundles can be misleading.
The base in the indexed counterexample is
the product of two complex projective lines, also called the quadric surface. The claim is therefore not confined to a high-dimensional or nonprojective base.
What is known
The protected Family 050 catalogue lists one manuscript, Ample rank-two bundles on the quadric surface without Griffiths-positive metrics (September 24, 2026).
The pinned primary PDF, Theorem 1.1, constructs one rank-two algebraic bundle \(G\) on \(S=\mathbb{P}^1\times\mathbb{P}^1\). For every positive integer \(m\), the power-map pullback \(E_m=f_m^*G\otimes\mathcal{O}(1,1)\) is ample; there exists a threshold \(m_0\) such that for every \(m\geq m_0\), \(E_m\) admits no smooth Hermitian metric with strictly Griffiths-positive curvature. The source explicitly notes that \(m_0\) is ineffective. This produces counterexamples in rank two on the quadric surface.
This is an existence statement: it does not say all ample rank-two bundles fail Griffiths positivity. Nor does the absence of a strictly Griffiths-positive metric automatically rule out every weaker metric positivity condition.
The exact bundle construction, proof of ampleness, and obstruction to positive curvature require direct verification from the pinned manuscript rather than inference from the overview.
GCL verification state
The source identity is openai/math@1581d4cbdd69a0faf34b2089614fe2a814316a6a. The 18-page primary PDF was downloaded at this exact revision (Git blob 3df7e220f480a17f408b5a1c99b0f56483a2384e, 389,997 bytes; SHA-256 af1c052828ec09138c76a64edeed05edd429e4634bf61a4fecdda9d86fd1eb9b). The abstract, Theorem 1.1, the quantitative ampleness statement of Proposition 3.1, and introductory proof outline were read.
The bounded source critic has checked the exact quantifier split between ampleness for every positive integer \(m\) and metric nonexistence for all sufficiently large \(m\), but has not independently checked the algebraic bundle data or analytic obstruction. No Family 050-specific Lean proof, Comparator replay, semantic-equivalence audit or MATHCERT certificate is established.
What remains
A source critic has checked the manuscript's theorem wording, the precise power-map family, and the distinction between all positive powers and sufficiently large powers. Mathematical follow-up must verify the three-graph elementary transformation, the very-ampleness certificate for the symmetric powers, and the invariant-measure obstruction to every smooth strictly Griffiths-positive Hermitian metric in the large-power range.
The source critic and exact article hash must be registered, the Markdown/compendium tests and HTML output checked, and the result admitted through protected merge/readback. None of these editorial checks independently validates Griffiths' conjecture or its purported counterexamples.
Sources and claim boundaries
Boundary: NO_CERTIFICATION. The article describes a publisher-reported counterexample, not a verified construction or certified proof.
Indexed primary manuscript paths:
- Ample rank-two bundles on the quadric surface without Griffiths-positive metrics —
preprints/ample-rank-two-bundles-on-the-quadric-surface-without-griffiths-positive-metrics-September-24-2026/paper.pdf