OctoInformBeta

OpenAI preprint, September 2026 ยท Section 2. Finite-field square constraints

Section 2 (setting): strict squares and normalized kernels: Strict-square indicators and products of constraints, with uniform finite averages

Directed square constraints between two lists.

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

Directed square constraints on two lists over ๐”ฝp\mathbb F_p, with pp an odd prime, specify allowed differences between selected pairs of entries. A strict square zz in a field FF is a nonzero element of FF of the form u2u^2 with uโˆˆFu\in F.

For a nonempty finite set ฮฉ\Omega and a function f:ฮฉโ†’โ„f\colon\Omega\to\mathbb R, the uniform average ๐”ผฯ‰โˆˆฮฉf(ฯ‰)\mathbb E_{\omega\in\Omega} f(\omega) is the real number defined by (1): ๐”ผฯ‰โˆˆฮฉf(ฯ‰)=1|ฮฉ|โˆ‘ฯ‰โˆˆฮฉf(ฯ‰).\mathbb E_{\omega\in\Omega} f(\omega) = \frac{1}{|\Omega|}\sum_{\omega\in\Omega} f(\omega).(1) For finite index sets II and JJ, averages over pairs of lists use this convention on ฮฉ=๐”ฝpIร—๐”ฝpJ\Omega=\mathbb F_p^I\times\mathbb F_p^J.

The Definition

Definition 1. For a field FF, the strict-square indicator sF:Fโ†’โ„s_F\colon F\to\mathbb R is the function defined in (2): sF(z)={1,zโ‰ 0 and z=u2 for some uโˆˆF,0,otherwise.s_F(z)= \begin{cases} 1,& z\ne0\text{ and }z=u^2\text{ for some }u\in F,\\ 0,& \text{otherwise}. \end{cases}(2)โœฆ

Let pp be prime, let II and JJ be index sets, and let EโІIร—JE\subseteq I\times J be finite. Choose a direction map ฯƒ:Iร—Jโ†’{+,โˆ’}\sigma\colon I\times J\to\{+,-\}. The square constraint kernel KE,ฯƒ:๐”ฝpIร—๐”ฝpJโ†’โ„K_{E,\sigma}\colon\mathbb F_p^I\times\mathbb F_p^J \to\mathbb R is the function defined in (3), for lists x=(xi)iโˆˆIx=(x_i)_{i\in I} and y=(yj)jโˆˆJy=(y_j)_{j\in J}: KE,ฯƒ(x,y)=โˆ(i,j)โˆˆE{s๐”ฝp(yjโˆ’xi),ฯƒ(i,j)=+,s๐”ฝp(xiโˆ’yj),ฯƒ(i,j)=โˆ’.K_{E,\sigma}(x,y) =\prod_{(i,j)\in E} \begin{cases} s_{\mathbb F_p}(y_j-x_i),&\sigma(i,j)=+,\\ s_{\mathbb F_p}(x_i-y_j),&\sigma(i,j)=-. \end{cases}(3)โœฆ

What Is Checked

The indicator in (2) is defined over any field, and the kernel in (3) is defined over every prime field, with arbitrary index sets and a finite set of directed constraints.