OctoInformBeta

OpenAI preprint, September 2026 · §7. Flatness for every finite exponent

The flatness model: The polynomial Pε(z)P_\varepsilon(z) of a real coefficient vector at any complex zz, and the signings, whose entries are all ±1\pm 1

Real signings and their Littlewood polynomials.

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

Take a Littlewood polynomial with NN coefficients and look at its modulus on the unit circle, parametrized by z=e(t)z = e(t) with e(t)=e2πite(t) = e^{2\pi i t} and t∈[0,1]t \in [0,1]. Its root mean square there is N\sqrt{N}, and the flatness problem asks how close to N\sqrt{N} the modulus can be kept at every point of the circle at once.

The library’s result on flatness for every finite exponent is built on two notions: the polynomial of a real coefficient vector, and the signing.

The Definition

Definition 1. Let N≥0N \ge 0 be an integer. For a coefficient vector ε=(ε0,…,εN−1)\varepsilon = (\varepsilon_0, \dots, \varepsilon_{N-1}) with real entries, the polynomial PεP_\varepsilon of ε\varepsilon is the function on the complex plane whose value at a complex number zz is Pε(z)=∑k=0N−1εkzk.P_\varepsilon(z) = \sum_{k=0}^{N-1} \varepsilon_k z^k .(1)✦ A signing of length NN is a coefficient vector ε=(ε0,…,εN−1)\varepsilon = (\varepsilon_0, \dots, \varepsilon_{N-1}) with real entries such that εk∈{−1,1}for every k∈{0,…,N−1}.\varepsilon_k \in \{-1, 1\} \qquad \text{for every } k \in \{0, \dots, N-1\}.(2)✦ The polynomial PεP_\varepsilon of a signing ε\varepsilon is its Littlewood polynomial.

What Is Checked

The library states the polynomial (1) over a real coefficient vector of any length NN and a point zz of the complex plane, with complex values, and the signing condition (2) over the same real coefficient vectors.