Claimed AI-assisted

Counterexample to the smooth Carathéodory conjecture on umbilic points

An explicit support function gives a smoothly embedded two-sphere bounding a convex body with exactly one umbilic point, so the C-infinity form of Carathéodory's 1922 conjecture is false.

Model
Claude; Codex
Field
Mathematics
Date
2026-08-19
Human collaborators
Levent Alpöge, John-Paul Smith
Problem posed
1922 · open 104 yrs

What was found

Carathéodory's conjecture, Problem 8.1 of Ghomi's list and traceable to 1922, asks whether every closed convex surface in R^3 carries at least two umbilic points. Hamburger settled the real-analytic case in 1940-41 and that theorem is untouched. The counterexample is explicitly a C-infinity object: an explicit support function produces a smoothly embedded two-sphere bounding a convex body with exactly one umbilic point. The same family disproves the smooth Loewner conjecture, whose member at k = 1 has an isolated trace-free Hessian zero of the wrong index. The gap between the smooth and the real-analytic case is the whole content of the result.

Novelty check

Carathéodory's conjecture is a named problem with a long literature and a standing partial theorem, Hamburger's real-analytic case from 1940-41, and it appears as Problem 8.1 in Ghomi's published problem list. The claim is a counterexample in the smooth category only, which is consistent with rather than contradicting Hamburger. No prior smooth counterexample appears in the problem list or in the vibemathed record. The construction is new work, not a retrieval of something already in the literature.

Caveats and known objections

Announced in an X post rather than a preprint, and unrefereed. The Lean formalization is sorry-free at a commit but not merged, and vibemathed records its statement as read rather than compiled and audited, so nothing here is machine-checked end to end for this registry's purposes; the grade stays at claimed. Autonomy is graded ai-assisted rather than higher on a deliberately strict reading of two separate model roles, neither of which is the discovery: Alpöge's announcement credits John-Paul Smith and Claude with checking the construction, which makes the model a verifier of a human construction, and the Lean development was, in its author's words, developed with Codex and parallel proof-review agents, which is formalization of a human result. Only the smooth case falls; the classical real-analytic form of the conjecture remains true.

Also recorded at

vibemathedmathbb-c-infty-caratheodory-conjecture

Nothing mechanically. It is a parallel listing of the same result, carrying its own verification label rather than an independent check.

Nobody outside the lab has checked this yet.

Reading the primary source closely enough to say whether it supports the claim counts as a check, and you are credited on the entry.

Or on GitHub: submit a check challenge the grade send a correction or send a pull request

Entry history (1 event)
  1. AddedEntered the registry graded Claimed and AI-assisted.

Entries are never deleted. A grade that does not hold up is downgraded on the record, with the reason beside it.

Graded claimed for verification and ai-assisted for autonomy. What these mean.

Cite this entry

Plain text
whataifound.org. (2026). Counterexample to the smooth Carathéodory conjecture on umbilic points. whataifound.org: A Registry of AI Scientific and Mathematical Discoveries. https://whataifound.org/finding/2026-08-19-caratheodory-umbilic
BibTeX
@misc{whataifound-independent-2026-umbilic,
  title        = {Counterexample to the smooth Carathéodory conjecture on umbilic points},
  author       = {{whataifound.org}},
  year         = {2026},
  howpublished = {whataifound.org: A Registry of AI Scientific and Mathematical Discoveries},
  note         = {Result by Independent. Verification: Claimed. Autonomy: AI-assisted.},
  url          = {https://whataifound.org/finding/2026-08-19-caratheodory-umbilic}
}

Related findings

← All mathematics findings in the registry