OpenAI claim summary for Math-418
mathematicsTop Open Problemscontributionliterature_reportClaim: literature report
Summary of a claim published in openai/math; not a claim by this platform.
Codimension-one varifold singular bound: A singular set of Hausdorff dimension at most m minus one has zero m-dimensional Hausdorff measure, for every positive dimension and codimension. The varifold stationarity, integrality and open smooth-neighborhood regularity conventions agree with our definition.
Almost-everywhere varifold regularity: The main theorem is exactly zero m-dimensional Hausdorff measure of the singular support of an arbitrary stationary integral m-varifold, with all positive m and codimensions and no additional stability, minimization or multiplicity restriction.
The claimed result implies the stated question and also supplies the additional strength described above.
Limitations: The linked openai/math material has not been examined or verified by the platform.
Manuscript
Contents
Measure-zero singular set of stationary integral varifolds
OpenAI's claim
Codimension-one varifold singular bound: For a stationary integral varifold of dimension $m$, the refinement claims a singular-set Hausdorff bound of $m-1$; the Euclidean ambient dimension can be any larger dimension, and the assertion begins at $m=1$.Almost-everywhere varifold regularity: For Euclidean stationary integral varifolds of dimension $m$, the initial paper claims that singular points carry no $m$-dimensional Hausdorff measure; off this exceptional set there are smooth minimal neighborhoods with one constant integer multiplicity; a separate argument handles round spheres.
Mathematical statement of the claimed result
Mathematical claim 1.
Let $V$ be a stationary integral $m$-varifold in open $U\subset\mathbb R^{m+n}$, with $m,n\ge1$. Its weight has the form $\mu=\theta\mathcal H^m|_E$ with integer multiplicity, and stationarity requires $\int\operatorname{tr}(P_{T_xV}DX)\,d\mu=0$ for every compactly supported $C^1$ ambient vector field. Define singular points as support points without a neighborhood equal to one smooth embedded minimal submanifold with constant positive integer multiplicity. The claimed bound is \[\dim_H\operatorname{Sing}V\le m-1.\] Stability, minimization and a codimension-one assumption are not required.Mathematical claim 2.
For arbitrary integers $m,n\ge1$, take an integral varifold $V$ of dimension $m$ on some open $U\subset\mathbb R^{m+n}$. If its first variation vanishes, it is claimed to satisfy \[\mathcal H^m(\operatorname{Sing}V)=0.\] Regularity here requires the local support to be a single smooth embedded minimal manifold with constant integer multiplicity. No stability, minimizing property, orientability, density upper bound or codimension restriction is imposed.Outline of the argument
Codimension-one varifold singular bound: Smooth reference graphs are fitted at prescribed flat scales, with controlled frequency intervals covering transitions between scales. Centered heights are normalized so that their second moments remain nonzero while their first moments vanish. The limiting height and small mean-curvature errors provide the dimension-reduction estimate needed to sharpen the earlier almost-everywhere result.Almost-everywhere varifold regularity: Integer multiplicity, height estimates, and heat-contact comparisons are combined in a signed-excess induction. Packing of witness scales closes the estimates when sheets merge. A center chosen from minimal fits and multiplicity defects gives inward doubling, forcing the residual height to vanish and the exceptional support to have zero relevant Hausdorff measure.
Formal verification
Codimension-one varifold singular bound: The pinned catalogue lists no Lean formalization for this result.Almost-everywhere varifold regularity: The pinned catalogue lists no Lean formalization for this result.
This site has not run the Lean development. These links describe OpenAI’s selected formal statements and should be read with their stated scope.
Status caveat
OpenAI reports that an internal model produced the manuscripts. Their arguments have not been independently verified as proofs by this site. Any portion outside the documented formal scope may contain mistakes, and the catalogue label reports the announced mathematical claim.Sources
- OpenAI, A Codimension-One Bound for the Singular Set of a Stationary Integral Varifold, 2026-10-05; citation key
OAI:A-Codimension-One-Bound-for-the-Singular-Set-of-a-Stationary-Integral-Varifold-October-5-2026. PDF (GitHub); PDF (direct); Manuscript page and citation; Latest version; Statement source: thm:main. - OpenAI, Almost-everywhere regularity of stationary integral varifolds, 2026-09-23; citation key
OAI:Almost-everywhere-regularity-of-stationary-integral-varifolds-September-23-2026. PDF (GitHub); PDF (direct); Manuscript page and citation; Latest version; Statement source: thm:main. - OpenAI catalogue entry.