Revision 254 · 2026-09-30 05:09 UTC atlas import import · claude-opus-5-5 “Written for the atlas by Claude Opus 5.5 at the direction of Danielle Fong (content/wiki/Statement__gaussian-moments.md).”
Revision 254 · 2026-09-30 05:09 UTC atlas import import · claude-opus-5-5 “Written for the atlas by Claude Opus 5.5 at the direction of Danielle Fong (content/wiki/Statement__gaussian-moments.md).”
Source +0 −0
… 28 unchanged lines
Revision 254 as rendered
The Gaussian moments conjecture is the probabilistic form of the Jacobian conjecture. It asks whether the kernel of Gaussian expectation is a Mathieu–Zhao subspace of the polynomial ring. Derksen, van den Essen and Zhao proved that the Gaussian moments conjecture in every dimension implies the Jacobian conjecture, and that a non-invertible cubic-homogeneous Keller map in r variables makes GMC(2r) false.[1] The operators ∂i−xi commute, have order one and constant leading coefficients, and by Gaussian integration by parts their images span kerE. So GMC is the image conjecture for this one commuting family.[1]
Open: JC₂. Refuted with an explicit witness: GMCₙ for n ≥ 3; JCₙ for n ≥ 3. Refuted by implication, no witness exhibited: none. Claimed, not yet checked: GMC₂ (a proof). Every cell not declared on its page is derived by the atlas from the relations drawn; hover a cell for its reasons. proved refuted, witnessed refuted by implication open claimed (V4) · a ring marks V0 · arrows point from the stronger statement to what it implies
Status
Three variables and up are false, with small explicit witnesses found independently of the Keller map. In three standard Gaussians, write Z,W=(X1±iX2)/2 and T=X3. Then P=W+WZ−T2−23ZT2−21Z2T2 and Q=Z satisfy E[Pm]=0 and E[QPm]=m! for every m≥1. Long also gives a six-term cubic in four variables.[2] His repository reports Lean 4 proofs not_GMC_three and not_GMC_of_three_le against an in-repository definition of GMC. They are pinned to Lean 4.34.0 and a fixed Mathlib commit, with an axiom audit.[3] The atlas has not rebuilt the project or audited the faithfulness of that definition. The tier V0 is the author's report of a kernel check.
One variable is true (Derksen–van den Essen–Zhao, Proposition 4.2).
Two variables are claimed true. Wilson's face-isolation proof is an unrefereed preprint, so the cell is drawn as a claim.[4] Long wrote of this case: “We do not settle this case.”[2]
If Wilson's proof holds, the threshold is sharp at three variables. The Jacobian conjecture's own threshold is also at three variables.