OctoInform Beta

OpenAI preprint, September 2026 · §1. Introduction · §1

Theorem 1.1 [Main theorem]: On any probability space, and sequentially mixing of every order along spaced sequences of time tuples

Invertible mixing transformations are mixing of all higher orders.

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

Let TT be an invertible measure-preserving transformation of a probability space (Ω,μ)(\Omega,\mu), and suppose that TT is mixing. Is TT then mixing of every order? In order three the question is whether, for measurable sets AA, BB and CC, the sets AA, T−mBT^{-m}B and T−m−nCT^{-m-n}C become independent in the limit as mm and nn both tend to infinity. This is Rokhlin’s multiple-mixing problem for a single transformation.

The library states a positive answer in every order k≥3k\ge 3, on an arbitrary measurable space Ω\Omega.

Mixing of Higher Orders

Throughout, Ω\Omega is a measurable space. A bimeasurable bijection of Ω\Omega is a bijection T:Ω→ΩT\colon\Omega\to\Omega that is measurable with measurable inverse; its integer powers are TnT^n, n∈ℤn\in\mathbb Z, and T−nAT^{-n}A is the preimage of a set AA under TnT^n. We write ℕ={0,1,2,…}\mathbb N=\{0,1,2,\dots\}.

A bimeasurable bijection TT of Ω\Omega is mixing for a measure μ\mu on Ω\Omega if, for all measurable sets A,B⊆ΩA,B\subseteq\Omega, lim|n|→∞μ(A∩T−nB)=μ(A)μ(B),n∈ℤ.\lim_{|n|\to\infty}\mu\bigl(A\cap T^{-n}B\bigr)=\mu(A)\,\mu(B),\qquad n\in\mathbb Z.(1)✦

The layout times t0,…,tk−1t_0,\dots,t_{k-1} of a tuple of gaps g=(g1,…,gk−1)g=(g_1,\dots,g_{k-1}), positive integers, are the partial sums t0=0t_0=0 and ti=g1+…+git_i=g_1+\dots+g_i for 1≤i≤k−11\le i\le k-1.✦

A bimeasurable bijection TT of Ω\Omega is mixing of order kk for a measure μ\mu on Ω\Omega if, for all measurable sets A0,…,Ak−1⊆ΩA_0,\dots,A_{k-1}\subseteq\Omega, limminjgj→∞μ(⋂i=0k−1T−tiAi)=∏i=0k−1μ(Ai),\lim_{\min_j g_j\to\infty}\mu\Bigl(\bigcap_{i=0}^{k-1}T^{-t_i}A_i\Bigr) =\prod_{i=0}^{k-1}\mu(A_i),(2)✦ where t0,…,tk−1t_0,\dots,t_{k-1} are the layout times of the gaps g=(g1,…,gk−1)g=(g_1,\dots,g_{k-1}).

Mixing Along Spaced Times

A sequence (dr)r∈ℕ(d_r)_{r\in\mathbb N} of kk-tuples dr=(dr(0),…,dr(k−1))d_r=(d_r(0),\dots,d_r(k-1)) of natural numbers is spaced if dr(i)+r<dr(j)for all r∈ℕ and 0≤i<j≤k−1.d_r(i)+r<d_r(j)\qquad\text{for all } r\in\mathbb N \text{ and } 0\le i<j\le k-1.(3)✦

A bimeasurable bijection TT of Ω\Omega is sequentially mixing of order kk for a measure μ\mu on Ω\Omega if, for every spaced sequence (dr)(d_r) of kk-tuples and all measurable sets A0,…,Ak−1⊆ΩA_0,\dots,A_{k-1}\subseteq\Omega, limr→∞μ(⋂i=0k−1T−dr(i)Ai)=∏i=0k−1μ(Ai).\lim_{r\to\infty}\mu\Bigl(\bigcap_{i=0}^{k-1}T^{-d_r(i)}A_i\Bigr) =\prod_{i=0}^{k-1}\mu(A_i).(4)✦

The Claim

Theorem 1. Let μ\mu be a probability measure on a measurable space Ω\Omega, and let TT be a bimeasurable bijection of Ω\Omega that preserves μ\mu and is mixing for μ\mu. Then, for every integer k≥3k\ge 3, Tis mixing of order k for μ.T\ \text{is mixing of order } k \text{ for } \mu.(5)✦

What Is Checked

The claim is stated over a probability measure μ\mu on an arbitrary measurable space Ω\Omega, a bimeasurable bijection TT of Ω\Omega that preserves μ\mu, its integer powers, mixing (1), and mixing of order kk (2) at the layout times of positive gaps.

For every measure μ\mu on Ω\Omega and every k∈ℕk\in\mathbb N, if TT is sequentially mixing of order kk for μ\mu, then Tis mixing of order k for μ.T\ \text{is mixing of order } k \text{ for } \mu.(6)✦

If TT is mixing for a measure μ\mu on Ω\Omega, then, for all measurable sets A,B⊆ΩA,B\subseteq\Omega, limn→∞μ(A∩T−nB)=μ(A)μ(B),n∈ℕ.\lim_{n\to\infty}\mu\bigl(A\cap T^{-n}B\bigr)=\mu(A)\,\mu(B),\qquad n\in\mathbb N.(7)✦

Let μ\mu be a probability measure on Ω\Omega preserved by TT, and suppose that (7) holds for all measurable sets A,B⊆ΩA,B\subseteq\Omega. Then, for every k∈ℕk\in\mathbb N, Tis sequentially mixing of order k for μ.T\ \text{is sequentially mixing of order } k \text{ for } \mu.(8)✦

Let μ\mu be a probability measure on Ω\Omega preserved by TT such that (7) holds for all measurable sets A,B⊆ΩA,B\subseteq\Omega, and let k≥3k\ge 3 be an integer. If TT is sequentially mixing for μ\mu of every order less than kk, then Tis sequentially mixing of order k for μ.T\ \text{is sequentially mixing of order } k \text{ for } \mu.(9)✦