OctoInformBeta

OpenAI preprint, September 2026 · §1. Introduction

§1: Gaussian partitions and their first moments: The squared-moment value and an explicit propeller family in Euclidean space

Gaussian partitions and their first moments.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

The Gaussian partition problem concerns how a finite division of Euclidean space distributes first moments among its cells. The three-sector propeller is the explicit family distinguished by two coordinates.

The Definition

Definition 1. Let dd and kk be nonnegative integers. The ambient Euclidean space ℝd\mathbb R^d is the real coordinate space in (1), with its usual inner product: ℝd:=ℝ{0,…,d−1}.\mathbb R^d:=\mathbb R^{\{0,\ldots,d-1\}}.(1)✦ The measure γd\gamma_d on ℝd\mathbb R^d is the standard Gaussian measure in (2), where IdI_d is the identity operator: γd:=𝒩(0,Id).\gamma_d:=\mathcal N(0,I_d).(2)✦

A Gaussian partition A=(A1,…,Ak)A=(A_1,\ldots,A_k) of ℝd\mathbb R^d is a labelled family of sets satisfying (3), where ℬ(ℝd)\mathcal B(\mathbb R^d) is the Borel sigma algebra: Ai∈ℬ(ℝd)(1≤i≤k),γd(ℝd\⋃i=1kAi)=0,γd(Ai∩Aj)=0(1≤i<j≤k).\begin{aligned} &A_i\in\mathcal B(\mathbb R^d) &&(1\le i\le k),\\ &\gamma_d\!\left(\mathbb R^d\setminus\bigcup_{i=1}^k A_i\right)=0,\\ &\gamma_d(A_i\cap A_j)=0 &&(1\le i<j\le k). \end{aligned}(3)✦

For a set A⊆ℝdA\subseteq\mathbb R^d, its unnormalized Gaussian centroid z(A)∈ℝdz(A)\in\mathbb R^d is the first moment in (4): z(A):=∫Axdγd(x).z(A):=\int_A x\,d\gamma_d(x).(4)✦ The squared-moment value F(A)F(A) of a labelled family A=(A1,…,Ak)A=(A_1,\ldots,A_k) of subsets of ℝd\mathbb R^d is the real number in (5): F(A):=∑i=1k∥z(Ai)∥2.F(A):=\sum_{i=1}^k\|z(A_i)\|^2.(5)✦

For a nonnegative integer jj and x∈ℝdx\in\mathbb R^d, the zero-extended coordinate xj∈ℝx_j\in\mathbb R is the real number in (6): xj:={x(j),j<d,0,j≥d.x_j:= \begin{cases} x(j),&j<d,\\ 0,&j\ge d. \end{cases}(6) The propeller family 𝒫d,k=(P1,…,Pk)\mathcal P^{d,k}=(P_1,\ldots,P_k) in ℝd\mathbb R^d is the labelled family of sets in (7): Pi:={{x∈ℝd:|x1|≤3x0},i=1,{x∈ℝd:0≤x1,3x0≤x1},i=2,{x∈ℝd:x1≤0,3x0≤−x1},i=3,⌀,i>3,1≤i≤k.P_i:= \begin{cases} \{x\in\mathbb R^d:|x_1|\le\sqrt3\,x_0\},&i=1,\\ \{x\in\mathbb R^d:0\le x_1,\ \sqrt3\,x_0\le x_1\},&i=2,\\ \{x\in\mathbb R^d:x_1\le0,\ \sqrt3\,x_0\le-x_1\},&i=3,\\ \varnothing,&i>3, \end{cases} \qquad 1\le i\le k.(7)✦

In the paper, the result reads: for positive integers dd and kk, let γd\gamma_d be standard Gaussian measure on ℝd\mathbb R^d; a measurable labelled partition (A1,…,Ak)(A_1,\ldots,A_k) is understood up to Gaussian null sets, allows empty cells and arbitrary probabilities, and has unnormalized centroids z(Ai)=∫Aixdγd(x)z(A_i)=\int_{A_i}x\,d\gamma_d(x) and value F(A)=∑i=1k∥z(Ai)∥2F(A)=\sum_{i=1}^k\|z(A_i)\|^2.

What Is Checked

The definitions are stated over Euclidean spaces of arbitrary nonnegative dimension and finite labelled families of sets, with Borel measurability and Gaussian almost-everywhere uniqueness imposed by (3).