OctoInformBeta

OpenAI preprint, September 2026 · §1. Introduction

Littlewood polynomials and the normalized minimum maximum: The polynomial ∑k<Nεkzk\sum_{k<N}\varepsilon_k z^k of a real vector ε\varepsilon at a complex point zz, and the signings, vectors with every entry ±1\pm 1

Littlewood polynomials built from real coefficient vectors and signings.

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

Fix a length NN and signs ε0,…,εN−1\varepsilon_0, \dots, \varepsilon_{N-1}, each equal to −1-1 or 11. On the unit circle, z=e(t)z = e(t) with e(x)=exp⁡(2πix)e(x) = \exp(2\pi i x) and tt in the circle 𝕋=ℝ/ℤ\mathbb{T} = \mathbb{R}/\mathbb{Z}, the polynomial with these coefficients is the trigonometric polynomial ∑k<Nεke(kt)\sum_{k<N} \varepsilon_k e(kt), whose mean square over 𝕋\mathbb{T} is NN. The flatness problem asks how close to N\sqrt{N} its maximum modulus can be brought by the choice of the signs.

The Definition

Definition 1. Let NN be a natural number. The polynomial PεP_\varepsilon of a coefficient vector ε=(ε0,…,εN−1)\varepsilon = (\varepsilon_0, \dots, \varepsilon_{N-1}) of length NN with real entries is the function on the complex plane given by Pε(z)=∑k<Nεkzk,z∈ℂ.P_\varepsilon(z) = \sum_{k<N} \varepsilon_k z^k, \qquad z \in \mathbb{C}.(1)✦ A signing of length NN is a coefficient vector ε\varepsilon of length NN with real entries such that εk∈{−1,1}for 0≤k<N.\varepsilon_k \in \{-1, 1\} \qquad \text{for } 0 \le k < N.(2)✦ The Littlewood polynomial of a signing ε\varepsilon of length NN is the polynomial PεP_\varepsilon of ε\varepsilon.

In the paper, the result reads: a Littlewood polynomial of length NN is P(z)=∑k<NεkzkP(z) = \sum_{k<N} \varepsilon_k z^k with every εk∈{−1,1}\varepsilon_k \in \{-1, 1\}, the page’s PεP_\varepsilon for a signing ε\varepsilon; its maximum modulus is ∥P∥∞=max⁡t∈𝕋|P(e(t))|\|P\|_\infty = \max_{t \in \mathbb{T}} |P(e(t))|, and mNm_N is the minimum of ∥P∥∞/N\|P\|_\infty / \sqrt{N} over all choices of the signs.

What Is Checked

The definitions (1) and (2) are stated over a coefficient vector ε\varepsilon of length NN with real entries, for every natural number NN, and the polynomial (1) over every complex number zz.