VibeMathedMath problems solved with AI

A complex structure on S2×S4S^2\times S^4

On page 408 of Calabi's 1958 paper "Construction and Properties of Some 6-Dimensional Almost Complex Manifolds" we find:

"...it is still unknown whether the sphere S6S^6 or any product manifold V2×S4V^2 \times S^4 (V2=V^2 = any closed, orientable surface) admit any complex analytic structure."

Result
Proved(see note)
Status
Resolved
AI contribution
AI-discovered
Method
Construction
Field
Differential geometry
Posed by
Calabi (1958)
Year posed
1958
Years open
68y
Solved
2026-08-30
Model
ChatGPT 5.6 Sol
Vendor
OpenAI
Collaborators
Ruiming Liang, Chenhan Liu, Yang Zhang
Verification
Unreviewed
Publication
Preprint
Significance
Disclosed cost
Wikipedia
No dedicated article

What was actually shown

The authors give an explicit construction of a complex manifold (called X+X_+) which is diffeomorphic to S2×S4S^2 \times S^4.

What the AI did

The human authors write the following:

"Statement on the use of AI. At first, GPT 5.6 Sol used the fibration method similar to [7] to construct a complex manifold XX fibreed over P1P^1 and claimed that it is homeomorphic to S2×S4S^2 \times S^4. However, the proof provided by GPT 5.6 Sol is cumbersome, incomplete, and misses an essential lemma. Then by comparing the singular fibre of XX with the singular fibres of XsphX_{sph}, we discovered that that XX is likely to be obtained by the blowup-quotient-flop process introduced in this paper, which led to the main theorem 1.2 of this article. Wall’s classification result is also introduced to the authors by GPT 5.6 Sol.

The proofs of lemma 2.3, lemma 3.1, proposition 4.6 are first generated by GPT 5.6 Sol, but it is checked and rewrote by the authors.

This article is also polished by GPT 5.6 Sol on grammars, language, notations, conventions, and some graphs."

Source

Submitted by Saul Schleimer on

Changelog3 changes

Discussion1

VibeGene07 Sep 2026, 10:59 UTC

Significance seems missing.

0