OctoInformBeta

OpenAI preprint, September 2026 · §1. Introduction

Theorem 1: Almost every phase state has a unique complete trajectory, and invariant sets satisfy a zero-one law

Irrational angles force ergodic billiard motion.

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

A particle moves at unit speed inside a triangle and reflects specularly at its open sides. The phase state records its position in ℂ\mathbb C and its direction in the unit circle S1S^1.✦ Ergodicity concerns the measurable sets of phase states that remain invariant under this motion up to sets of measure zero.

A nondegenerate triangle QQ is a triple of noncollinear vertices q0,q1,q2∈ℂq_0,q_1,q_2\in\mathbb C, with indices read modulo three.✦ For a nondegenerate triangle QQ with vertices q0,q1,q2q_0,q_1,q_2, the open billiard table ΩQ\Omega_Q is the interior of their convex hull, the open side LiL_i is the open segment joining qiq_i to qi+1q_{i+1}, and the side tangent vector eie_i is their difference, as in (1): ΩQ=int⁡conv⁡{q0,q1,q2},Li=(qi,qi+1),ei=qi+1−qi.\Omega_Q=\operatorname{int}\operatorname{conv}\{q_0,q_1,q_2\},\qquad L_i=(q_i,q_{i+1}),\qquad e_i=q_{i+1}-q_i.(1) Here (qi,qi+1)(q_i,q_{i+1}) denotes the open line segment. The interior angle αi\alpha_i at qiq_i is the Euclidean angle in (2); QQ has an irrational angle when the condition on the right holds: αi=∠(qi+1−qi,qi+2−qi),αiπ∉ℚfor some i∈{0,1,2}.\alpha_i=\angle(q_{i+1}-q_i,q_{i+2}-q_i),\qquad \frac{\alpha_i}{\pi}\notin\mathbb Q\quad\text{for some }i\in\{0,1,2\}.(2)

Write Phase\mathrm{Phase} for this phase space. The uniform angular probability mS1m_{S^1} is the image of normalized Lebesgue measure on [0,2π)[0,2\pi) under θ↦eiθ\theta\mapsto e^{i\theta}.✦ The normalized area-angular measure μ\mu on Phase\mathrm{Phase} is the product measure in (3), where dAdA is planar Lebesgue measure and |ΩQ||\Omega_Q| is the area of the table: mS1=(θ↦eiθ)*(12πdθ|[0,2π)),μ=(1|ΩQ|dA|ΩQ)⊗mS1.m_{S^1}=(\theta\mapsto e^{i\theta})_* \left(\frac{1}{2\pi}\,d\theta\big|_{[0,2\pi)}\right),\qquad \mu=\left(\frac{1}{|\Omega_Q|}\,dA\big|_{\Omega_Q}\right)\otimes m_{S^1}.(3)✦

Complete Trajectories

For a side tangent ee and a velocity v∈ℂv\in\mathbb C, specular reflection Re(v)R_e(v) is the reflected velocity defined by (4): Re(v)=ee¯v¯.R_e(v)=\frac{e}{\overline e}\,\overline v.(4)✦ A complete regular billiard trajectory cc in the triangle QQ through z=(x,v)∈Phasez=(x,v)\in\mathrm{Phase} is a sequence of collision times τn∈ℝ\tau_n\in\mathbb R, collision points pn∈ℂp_n\in\mathbb C, directions vn∈S1v_n\in S^1 and side indices kn∈{0,1,2}k_n\in\{0,1,2\}, indexed by n∈ℤn\in\mathbb Z, with the conditions in (5). For a phase state z∈Phasez\in\mathrm{Phase}, 𝒞Q(z)\mathcal C_Q(z) is the set of complete regular billiard trajectories in the triangle QQ through zz. τn<τn+1,limn→−∞τn=−∞,limn→+∞τn=+∞,τ0<0<τ1,pn∈Lkn,pn+1=pn+(τn+1−τn)vn,pn+(t−τn)vn∈ΩQwhenever τn<t<τn+1,x=p0−τ0v0,v=v0,vn=Rekn(vn−1)(n∈ℤ).\begin{gathered} \tau_n<\tau_{n+1},\qquad \lim_{n\to-\infty}\tau_n=-\infty,\qquad \lim_{n\to+\infty}\tau_n=+\infty,\qquad \tau_0<0<\tau_1,\\ p_n\in L_{k_n},\qquad p_{n+1}=p_n+(\tau_{n+1}-\tau_n)v_n,\\ p_n+(t-\tau_n)v_n\in\Omega_Q \quad\text{whenever }\tau_n<t<\tau_{n+1},\\ x=p_0-\tau_0v_0,\qquad v=v_0,\qquad v_n=R_{e_{k_n}}(v_{n-1})\quad(n\in\mathbb Z). \end{gathered}(5) For a complete regular trajectory c∈𝒞Q(z)c\in\mathcal C_Q(z) and real time tt, the trajectory state c(t)∈Phasec(t)\in\mathrm{Phase} consists of its position and direction at tt, with the outgoing direction used at a collision time. For real tt, the billiard flow Φt:Phase→Phase\Phi_t:\mathrm{Phase}\to\mathrm{Phase} sends zz to the state at tt along a complete regular billiard trajectory through zz whenever one exists, and fixes zz otherwise.

An invariant set modulo μ\mu is a subset A⊆PhaseA\subseteq\mathrm{Phase} satisfying (6), where △\triangle denotes symmetric difference: μ(Φt−1A△A)=0(t∈ℝ).\mu\bigl(\Phi_t^{-1}A\mathbin{\triangle}A\bigr)=0 \qquad(t\in\mathbb R).(6) A μ\mu-null-measurable set AA in Phase\mathrm{Phase} is a subset of Phase\mathrm{Phase} measurable for the completion of μ\mu.

The Claim

Theorem 1 (Irrational triangle billiards). Let QQ be a nondegenerate triangle with an irrational angle. Its billiard flow is a regular ergodic billiard flow in the sense of (7): $$\begin{gathered} \mu(\mathrm{Phase})=1,\qquad \mathcal C_Q(z)\ne\varnothing\quad\text{for }\mu\text{-almost every }z,\\ \#\mathcal C_Q(z)\leq1\quad(z\in\mathrm{Phase}),\\ \Phi_t\text{ is measurable},\qquad (\Phi_t)_*\mu=\mu \quad(t\in\mathbb R),\\ \Phi_{s+t}=\Phi_s\circ\Phi_t\quad\mu\text{-almost everywhere} \quad(s,t\in\mathbb R),\\ \text{and the measurable invariant set law \eqref{eq:claim2} holds.} \end{gathered}$$(7)✦ For a measurable invariant set AA modulo μ\mu, the invariant set law is (8): μ(A)=0orμ(A)=1.\mu(A)=0\quad\text{or}\quad\mu(A)=1.(8)✦ For a μ\mu-null-measurable invariant set AA modulo μ\mu, the completed invariant set law is (9): μ(A)=0orμ(A)=1.\mu(A)=0\quad\text{or}\quad\mu(A)=1.(9)✦

In the paper, the result reads: if ΩQ\Omega_Q is the interior of a nondegenerate Euclidean triangle and at least one of its angle ratios αi/π\alpha_i/\pi is irrational, then the unit-speed billiard flow with specular reflection at open sides, after discarding trajectories that hit a vertex in either time direction, is ergodic for dμ=dAdθ/(2π|ΩQ|)d\mu=dA\,d\theta/(2\pi|\Omega_Q|); in particular, every measurable set AA with μ(Φt−1A△A)=0\mu(\Phi_t^{-1}A\mathbin{\triangle}A)=0 for every t∈ℝt\in\mathbb R has μ(A)=0\mu(A)=0 or μ(A)=1\mu(A)=1.

What Is Checked

The claims are stated over a nondegenerate triangle QQ with an irrational angle, its complete regular trajectories and billiard flow on Phase\mathrm{Phase}, and the normalized area-angular measure μ\mu.