Gigli’s characterization of Alexandrov curvature

Claims Gigli's conjecture: Alexandrov curvature bounds match RCD spaces with a distributional sectional curvature bound.

2 papers September 24, 2026 Explained by AI

There are two ways to say a rough space curves at least as much as a model surface. One compares the shapes of triangles, and the other uses calculus, through a weak version of the curvature tensor defined by Gigli. Gigli conjectured that the two agree on a well-behaved class of rough spaces called RCD spaces. The paper claims this in every dimension of at least two. A Lean proof covers only a supporting result from the companion paper.

First step

2019

Nicola Gigli, who defined the distributional curvature tensor and posed the conjecture

Significance

Solid

It claims to connect the triangle and calculus views of sectional curvature on non-smooth spaces, as Gigli conjectured.

Trust

Formal proof listed

Field

Differential geometry

2 papers

The explanation

Start with a triangle on a unit sphere. Take the north pole and two points on the equator a quarter-turn apart. Each side has length π/2, about 1.571, and each angle is 90 degrees. Now draw a flat triangle with the same three side lengths. Its angles are 60 degrees. Measure from one corner to the middle of the opposite side. On the sphere this distance is π/2, about 1.571. In the flat triangle it is (√3/2) × (π/2), about 1.360. The sphere's triangle is fatter. That is what curvature at least 0 means in triangle terms.

In an Alexandrov space with curvature at least κ, small triangles are at least as fat as on a model surface of curvature κSource 1. This needs only distances, not smoothness. The other view uses calculus. An RCD space is a rough space with a lower bound on Ricci curvature, an averaged form of curvature, defined by optimal transportSource 2. Optimal transport studies the cheapest way to move mass from one place to another. Gigli defined a distributional sectional curvature on such spaces, tested against smooth-like functions, and conjectured the two views agreeSource 3. His construction did not by itself link a lower bound on the tensor to triangle comparisonSource 4.

The paper claims the equivalence in every dimension n of at least 2Source 5. With Hausdorff measure, the Alexandrov bound κ holds exactly when the space is RCD with distributional sectional curvature at least κSource 5. Hausdorff measure is the natural n-dimensional volume. The choice of that exact measure matters, since it rules out collapsed spacesSource 6. The test classes are Gigli's original global onesSource 7.

The hard direction goes from calculus to triangles. A weak Hessian bound is tested against volume, but a single geodesic, a shortest path, can lie in a set of zero volumeSource 8. The companion paper bridges this gap. It proves that a weak upper bound on the Hessian, the table of second derivatives, holds along every shortest pathSource 9. The main paper also needs frames along paths at every time, not only almost every timeSource 10.

How sure is this? The Lean scope note says the formalization carries a weak Hessian upper bound to every shortest path in an RCD space of finite dimension. That is the companion paper's theorem. The main equivalence is not in that note. No specialist review has been published.

What it doesn't show. This does not cover spaces whose reference measure is not exactly Hausdorff measure, or reduced versions of the RCD condition. The main equivalence has no Lean check.

Where it came from

This result comes from an unreleased OpenAI model, published in the openai/math release on 6 October 2026.

Hype check

Solid if it holds. Lean covers only the companion paper's Hessian-along-geodesics theorem, not the main equivalence. No specialist review is published.

Words used

Alexandrov space: A metric space where small triangles are at least as fat as matching triangles on a model surface of constant curvature.
RCD space: A possibly rough space with a lower bound on Ricci curvature, defined by optimal transport, and with a Hilbert-space calculus.
Ricci curvature: An averaged form of curvature that controls how volume spreads out.
Sectional curvature: How much a space bends along one flat two-dimensional slice.
Distributional: Defined by testing against nice functions instead of point by point.
Hausdorff measure: A natural way to measure n-dimensional volume using only distances.
Geodesic: A shortest path between two points.
Hessian: The table of second derivatives of a function.

The original papers (2)

Gigli’s distributional curvature characterization of Alexandrov spaces (PDF, opens in a new tab)
September 24, 2026
Weak Hessian bounds along every geodesic in RCD spaces (PDF, opens in a new tab)
September 24, 2026 Formal proof listed

Lean notes

All quoted lines (10)

An Alexandrov lower curvature bound $\CBB(\kappa)$ means that geodesic triangles satisfy lower comparison with the simply connected surface of constant curvature $\kappa$ , at the usual model scale when $\kappa>0$ .
Gigli’s distributional curvature characterization of Alexandrov spaces Back to text
Lott–Villani and Sturm independently developed synthetic lower Ricci curvature bounds through convexity properties of entropy along Wasserstein geodesics.
Gigli’s distributional curvature characterization of Alexandrov spaces Back to text
Gigli subsequently defined (…) and conjectured the characterization proved here in dimensions at least two.
Gigli’s distributional curvature characterization of Alexandrov spaces Back to text
His construction gives tensor symmetries and tensoriality, but does not by itself relate a lower tensor bound to triangle comparison.
Gigli’s distributional curvature characterization of Alexandrov spaces Back to text
For every integer $n\ge2$ and $\kappa\in\mathbb R$ , we prove that a complete separable metric space is an $n$ -dimensional Alexandrov space of curvature at least $\kappa$ if and only if, with reference measure $\mathcal H^n$ , it is a full-support $\RCD((n-1)\kappa,n)$ space whose distributional sectional curvature is at least $\kappa$ in Gigli's original global test classes.
Gigli’s distributional curvature characterization of Alexandrov spaces Back to text
The exact choice of reference measure is important: the assumption $m=\mathcal H^n$ supplies noncollapsed structure, including $n$ -dimensional Euclidean tangents almost everywhere.
Gigli’s distributional curvature characterization of Alexandrov spaces Back to text
The $\RCD$ condition is unreduced, and the test classes and curvature sign are exactly those defined above.
Gigli’s distributional curvature characterization of Alexandrov spaces Back to text
A weak inequality is tested against the reference measure, whereas a prescribed geodesic can lie in a set of measure zero.
Weak Hessian bounds along every geodesic in RCD spaces Back to text
On a full-support $\RCD(K,N)$ space with $1<N<\infty$ , we prove that a bounded globally Lipschitz function whose distributional Hessian is bounded above by a bounded continuous function satisfies the corresponding second-derivative inequality along every minimizing geodesic.
Weak Hessian bounds along every geodesic in RCD spaces Back to text
Almost-everywhere spatial regularity alone does not give a frame at every time on a path.
Gigli’s distributional curvature characterization of Alexandrov spaces Back to text

Other sources

openai/math, Lean scope note for family 356 The formalization transfers a weak Hessian upper bound to every prescribed minimizing geodesic
openai/math release notes (2026): model provenance and review status manuscripts and supporting proof artifacts produced by an internal OpenAI model

358A three-manifold without conjugate points or nonpositive curvatureDifferential geometry
359Negative Kähler curvature without bounded holomorphic coordinatesDifferential geometry
352A finite-time singularity of Calabi flowDifferential geometry
360Weak MTW curvature gives convexity and regular optimal transportDifferential geometry
351Scalar curvature and finite-time Ricci-flow singularitiesDifferential geometry

Browse Differential geometry Curvature and geometric flows Open since the 2010s Lean-checked results Solid results