Skip to content

Documentary state: DRAFT_UNACCEPTED. This draft has NOT been editorially admitted. Mathematical claims require independent checking.

OAI-26 family 048: a counterexample to Lipman–Zariski in characteristic zero

Editorial status: DRAFT_UNACCEPTED. NO_CERTIFICATION. This is documentary coverage of a publisher-reported construction, not an independently validated counterexample.

The question in plain language

The tangent directions to a smooth geometric space vary together in a well-behaved vector bundle. On a singular space, the analogous object—the tangent sheaf—can fail to have this simple local structure. The Lipman–Zariski conjecture predicted that if a complex algebraic variety has a locally free tangent sheaf, then the variety must be smooth.

The pinned OAI-26 overview claims a counterexample: a normal affine complex surface that is singular but nevertheless has a free rank-two tangent sheaf. Freeness is stronger than local freeness, so such an example would directly contradict the unrestricted conjecture in characteristic zero.

A small example

At a smooth point of a complex surface there are two independent infinitesimal tangent directions. A rank-two free tangent sheaf means that, across the whole surface, tangent vector fields admit two global algebraic generators with no module relations beyond those of a free module.

The intuition behind Lipman–Zariski is that having such consistently organized tangent directions ought to rule out singularities. But a singular point is defined by the failure of the local algebraic geometry to be regular; an apparently well-behaved tangent module does not automatically prove local regularity.

The distinction between normal, smooth, locally free, and globally free is central. A normal surface can have isolated singularities, and freeness of its tangent sheaf is a precise module-theoretic property rather than a statement that the underlying space looks smooth.

What is known

The pinned catalogue identifies a single Family 048 manuscript, A singular normal affine surface with free tangent sheaf (September 23, 2026).

The publisher overview says it constructs a singular normal affine complex surface with free rank-two tangent sheaf, giving a characteristic-zero counterexample to the Lipman–Zariski conjecture.

This would be an existence counterexample, not a classification of all normal surface singularities with free tangent sheaves. It also should not be translated into an assertion about every algebraic category or every formulation of the conjecture without checking the source's exact definitions.

The defining equations, singular-locus calculation and proof of tangent-module freeness must be verified directly from the pinned PDF. The indexed synopsis alone does not establish the construction.

GCL verification state

The catalogue pins this family to openai/math@1581d4cbdd69a0faf34b2089614fe2a814316a6a. This article reports the family identity, source path and publisher's headline assertion.

This pass has not independently reconstructed the affine coordinate ring, checked normality or singularity, or demonstrated that the derivation module is free of rank two. No Family 048-specific Lean formalization, Comparator result, semantic-equivalence audit or MATHCERT certificate is asserted.

What remains

A source critic should read the pinned manuscript and reproduce the surface's defining equations. Check that the coordinate ring is a two-dimensional integral normal complex algebra and that the surface is singular, including an explicit local regularity or Jacobian calculation.

Then compute the tangent sheaf as the module of derivations and verify freeness of rank two, not merely reflexivity or generic rank. Confirm the theorem's precise scope and examine whether an implicit restriction in another formulation of Lipman–Zariski would exclude the example.

After source-specific criticism, validate mathematical Markdown and rendered HTML, record editorial disposition and merge through protected governance. Documentary admission remains different from proof verification and certification.

Sources and claim boundaries

Boundary: NO_CERTIFICATION. This explains the publisher's claimed counterexample rather than independently confirming the construction.


Indexed primary manuscript paths: