# Failure of integer-degree harmonic dimension comparison

> Proves Yau's Euclidean bound on polynomial-growth harmonic functions fails under nonnegative Ricci curvature.

Canonical: https://closedform.nishachay.tech/e/openai-math-2026-361/
Last updated: 2026-10-14

**Proves Yau's Euclidean bound on polynomial-growth harmonic functions fails under nonnegative Ricci curvature.**

**Trust:** Formal proof listed · **Papers:** 2 · **First paper:** September 25, 2026 · **Field:** Differential geometry · **Entry:** 361

*Explained by AI.* Drafted by one AI model and checked sentence by sentence against the paper by a different one. Spot a mistake? Report an error.

A harmonic function is a smooth function that equals its own average around every small sphere. On flat three-dimensional space, there are exactly (k + 1)² independent ones that grow no faster than distance to the power k. Yau asked if nonnegative Ricci curvature always keeps this count at or below the flat one. The first paper builds a counterexample in dimension 16, and Lean, a computer proof checker, has a formal statement of it. A companion paper claims counterexamples on three-dimensional space for every large k.

## At a glance

- **First step:** 1989 (Peter Li and Luen-Fai Tam, the sharp bound for linear-growth harmonic functions)
- **Significance:** Major. It answers the integer-degree form of Yau's comparison question negatively, with a Lean statement in one dimension.
- **Trust:** Formal proof listed
- **Field:** Differential geometry (2 papers)

## The explanation

On flat three-dimensional space, list the harmonic polynomials of degree at most 2. A harmonic function equals its own average around every small sphere. There is 1 constant. There are 3 linear ones: x, y and z. There are 5 independent quadratic ones, such as xy and x² − y². That makes 1 + 3 + 5 = 9. In general, degree at most k gives 1 + 3 + … + (2k + 1) = (k + 1)²[1]. On flat space, a harmonic function that grows no faster than distance to the power k is such a polynomial.

Ricci curvature is a kind of average curvature. Nonnegative Ricci curvature means space does not spread out faster than flat space. The comparison problem asks whether nonnegative Ricci curvature always keeps the count at or below the flat count, for every whole-number k[2]. Li and Tam proved the sharp bound for linear growth, k = 1. Colding and Minicozzi proved the count is finite for every degree[3]. Donnelly found failures at fractional degrees between one and two. Those do not decide the whole-number case[4].

The first paper answers no. It builds a complete metric with nonnegative Ricci curvature in some even dimension n ≥ 8. This metric has more harmonic functions of growth at most k than flat space[5]. It allows the choices n = 16 and k = 50000[6]. The flat count there is about 4.68 × 10⁵⁸. The key idea is a cone whose cross-section slowly changes shape. On a cone, each angular direction carries one growth rate d, set by d(d + m − 1) = λ, where λ is an eigenvalue[7]. On a round two-sphere, λ = l(l + 1) gives d = l. A moving cross-section lets one function use several rates in turn. The obstacle is that a tiny part in a faster direction can take over after a long stretch[8].

The second paper claims the same failure on three-dimensional space. For large k it builds metrics with at least c(k + 1)² harmonic functions, for any c below 9v/4[9]. Here v, between 4/9 and 1, measures volume growth relative to flat space. With c = 2 and k = 10, that is at least 242 functions instead of 121. The metric is chosen after k and may depend on it[10].

How sure is this? Lean, a computer proof checker, has a formal statement of the first paper's result in dimension 16 at degree 50000. So that counterexample is proved, as far as the Lean statement matches the paper. The three-dimensional companion and the tangent-cone claims are not in the Lean scope. No specialist review is published.

**What it doesn't show.** This does not give one fixed metric that beats the count for every k; the metric depends on k. It does not overturn the known finite-dimensionality bounds, and the three-dimensional companion 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

Major, with a Lean statement of one counterexample. The metric depends on the degree, and the three-dimensional result is a claim without a Lean check. No specialist review is published.

## Words used

- **Harmonic function:** A smooth function equal to its own average around every small sphere.
- **Polynomial growth:** Growing no faster than a fixed power of the distance from a base point.
- **Ricci curvature:** An average of curvature over directions; nonnegative means volume grows no faster than in flat space.
- **Asymptotic volume ratio:** How the volume of huge balls compares with flat balls of the same radius.
- **Tangent cone at infinity:** The cone a space looks like when viewed from very far away.
- **Eigenvalue:** A number λ for which some nonzero function is turned into λ times itself.

## The original papers (2)

- [A counterexample to integer-degree harmonic dimension comparison](https://github.com/openai/math/blob/main/preprints/A-counterexample-to-integer-degree-harmonic-dimension-comparison-September-25-2026/paper.pdf) - September 25, 2026 · Formal proof listed
- [A Three-Dimensional Counterexample to Integer-Degree Harmonic Dimension Comparison](https://github.com/openai/math/blob/main/preprints/A-Three-Dimensional-Counterexample-to-Integer-Degree-Harmonic-Dimension-Comparison-September-26-2026/paper.pdf) - September 26, 2026

- [Lean notes](https://github.com/openai/math/blob/main/lean/docs/361.md)

## Quoted lines from the paper (10)

[1] From [A Three-Dimensional Counterexample to Integer-Degree Harmonic Dimension Comparison](https://github.com/openai/math/blob/main/preprints/A-Three-Dimensional-Counterexample-to-Integer-Degree-Harmonic-Dimension-Comparison-September-26-2026/paper.pdf):

> For an integer $k\geq0$, $\mathcal H_k(\mathbb R^3,g_{\mathrm E})$ consists of harmonic polynomials of degree at most $k$; summing their homogeneous dimensions gives \begin{equation} h_k(\mathbb R^3,g_{\mathrm E})=\sum_{l=0}^k(2l+1)=(k+1)^2. \end{equation}

[2] From [A counterexample to integer-degree harmonic dimension comparison](https://github.com/openai/math/blob/main/preprints/A-counterexample-to-integer-degree-harmonic-dimension-comparison-September-25-2026/paper.pdf):

> The integer-degree comparison problem asks whether $\operatorname{Ric}_g \geq0$ forces $h_k(M,g)\leq h_k(\mathbb R^n,g_{\mathrm E})$ for every integer $k\geq1$.

[3] From [A counterexample to integer-degree harmonic dimension comparison](https://github.com/openai/math/blob/main/preprints/A-counterexample-to-integer-degree-harmonic-dimension-comparison-September-25-2026/paper.pdf):

> Colding and Minicozzi proved finite dimensionality at every fixed degree and later obtained bounds of the optimal order $d^{n-1}$ as $d$ grows.

[4] From [A Three-Dimensional Counterexample to Integer-Degree Harmonic Dimension Comparison](https://github.com/openai/math/blob/main/preprints/A-Three-Dimensional-Counterexample-to-Integer-Degree-Harmonic-Dimension-Comparison-September-26-2026/paper.pdf):

> This subquadratic excess does not by itself establish an integer-degree violation.

[5] From [A counterexample to integer-degree harmonic dimension comparison](https://github.com/openai/math/blob/main/preprints/A-counterexample-to-integer-degree-harmonic-dimension-comparison-September-25-2026/paper.pdf):

> There are an even integer $n\geq8$, an integer $k\geq2$, and a complete smooth metric $g$ on $\mathbb R^n$ such that $\operatorname{Ric}_g\geq0$ and \[ h_k(\mathbb R^n,g)>h_k(\mathbb R^n,g_{\mathrm E}). \]

[6] From [A counterexample to integer-degree harmonic dimension comparison](https://github.com/openai/math/blob/main/preprints/A-counterexample-to-integer-degree-harmonic-dimension-comparison-September-25-2026/paper.pdf):

> The exact finite spectral selection in Appendix … permits the choices \(n=16\) and \(k=50000\).

[7] From [A counterexample to integer-degree harmonic dimension comparison](https://github.com/openai/math/blob/main/preprints/A-counterexample-to-integer-degree-harmonic-dimension-comparison-September-25-2026/paper.pdf):

> On the metric cone $dr^2+r^2h$ over an $m$-dimensional link, an angular eigenfunction of the nonnegative Laplacian with eigenvalue $\lambda$ has a growing homogeneous harmonic mode $r^d\phi$, where \[ d(d+m-1)=\lambda,\qquad d\geq0. \]

[8] From [A counterexample to integer-degree harmonic dimension comparison](https://github.com/openai/math/blob/main/preprints/A-counterexample-to-integer-degree-harmonic-dimension-comparison-September-25-2026/paper.pdf):

> The challenge is to arrange that succession exactly: an arbitrarily small component in an unwanted faster direction can dominate after a sufficiently long interval.

[9] From [A Three-Dimensional Counterexample to Integer-Degree Harmonic Dimension Comparison](https://github.com/openai/math/blob/main/preprints/A-Three-Dimensional-Counterexample-to-Integer-Degree-Harmonic-Dimension-Comparison-September-26-2026/paper.pdf):

> For every \(v\in(4/9,1)\) and \(c\in(1,9v/4)\), there is an integer \(k_0(v,c)\) such that for every integer \(k\geq k_0(v,c)\) there is a complete smooth metric \(g_k\) on \(\mathbb R^3\), Euclidean near the origin, with \(\operatorname{Ric}_{g_k}\geq0\), \(\operatorname{AVR}(g_k)=v\), and \[ h_k(\mathbb R^3,g_k)\geq c(k+1)^2>(k+1)^2 =h_k(\mathbb R^3,g_{\mathrm E}). \]

[10] From [A Three-Dimensional Counterexample to Integer-Degree Harmonic Dimension Comparison](https://github.com/openai/math/blob/main/preprints/A-Three-Dimensional-Counterexample-to-Integer-Degree-Harmonic-Dimension-Comparison-September-26-2026/paper.pdf):

> The metric is chosen after \(k\) and may depend on it.

## Other sources

- [openai/math, Lean scope note for family 361](https://github.com/openai/math/blob/main/lean/docs/361.md): “The construction uses dimension $16$ and degree $50000$.”
- [openai/math release notes (2026): model provenance and review status](https://github.com/openai/math/blob/main/README.md): “manuscripts and supporting proof artifacts produced by an internal OpenAI model”

## Related

- [A finite-time singularity of Calabi flow](https://closedform.nishachay.tech/e/openai-math-2026-352/) - Differential geometry
- [Katok's entropy rigidity conjecture](https://closedform.nishachay.tech/e/openai-math-2026-339/) - Differential geometry
- [Yau's uniformization conjecture](https://closedform.nishachay.tech/e/openai-math-2026-338/) - Differential geometry
- [Spectral scalar curvature, Urysohn width, and macroscopic dimension](https://closedform.nishachay.tech/e/openai-math-2026-336/) - Differential geometry
- [Gromov’s integral scalar-curvature bound for simplicial volume](https://closedform.nishachay.tech/e/openai-math-2026-335/) - Differential geometry
- [Differential geometry](https://closedform.nishachay.tech/science/mathematics/differential-geometry/) - Field · 29 results
- [Curvature and geometric flows](https://closedform.nishachay.tech/topic/curvature-and-geometric-flows/) - Topic · 17 results
- [Open since the 1980s](https://closedform.nishachay.tech/open-since/1980s/) - Era · 60 results
- [Lean-checked results](https://closedform.nishachay.tech/trust/lean-checked/) - Trust · 242 results
- [Report an error](https://github.com/nishachay/closed-form/issues/new?template=report-error.yml&labels=entry-error&title=%5Bentry%20361%5D%20)