Ergodicity of triangular billiards with an irrational angle
Lean is a language for mathematical statements and proofs that a computer can verify. This page shows one statement file from OpenAI’s release, possible paper matches and independent reports. The code-like names below identify the definitions or theorems selected by the proof-verification settings.
Theorem selected for verification:OAI.TriangularBilliards.irrational_triangle_billiard
A Lean statement states the theorem and leaves its proof as sorry on purpose. OpenAI’s proof is in a separate solution module, and a program called Comparator checks that proof against this statement.
Definitions needed to read it
Local definitions the statement names, followed through the definitions they name in turn. The match is by name in the source text, so a name reached only through notation can be missed; the full file is below.
structureFlightChain (Q : Triangle) (z : Phase) where
Show the remaining 18 lines
time : ℤ → ℝ point : ℤ → ℂ direction : ℤ → Circle wall : ℤ → Fin 3 increasing : StrictMono time unbounded_below : ∀ t : ℝ, ∃ n : ℤ, time n < t unbounded_above : ∀ t : ℝ, ∃ n : ℤ, t < time n zero_before : time 0 < 0 zero_after : 0 < time 1 on_side : ∀ n : ℤ, point n ∈ Q.side (wall n) flight : ∀ n : ℤ, point (n + 1) = point n + (time (n + 1) - time n) • (direction n : ℂ) inside : ∀ (n : ℤ) (t : ℝ), time n < t → t < time (n + 1) → point n + (t - time n) • (direction n : ℂ) ∈ Q.table start_point : z.1 = point 0 + (-time 0) • (direction 0 : ℂ) start_direction : z.2 = direction 0 specular : ∀ n : ℤ, (direction n : ℂ) = reflect (Q.tangent (wall n)) (direction (n - 1) : ℂ)
defFlightChain.at {Q : Triangle} {z : Phase} (c : FlightChain Q z) (t : ℝ) : Phase := by classical exact if h : ∃ n : ℤ, c.time n ≤ t ∧ t < c.time (n + 1) thenlet n := Classical.choose h (c.point n + (t - c.time n) • (c.direction n : ℂ), c.direction n)else z
defbilliardFlow (Q : Triangle) (t : ℝ) (z : Phase) : Phase := by classical exact if h : Nonempty (FlightChain Q z) then (Classical.choice h).at t else z
defMainConclusion (Q : Triangle) : Prop := phaseMeasure Q univ = 1 ∧ (∀ᵐ z ∂phaseMeasure Q, Nonempty (FlightChain Q z)) ∧ (∀ (z : Phase) (c d : FlightChain Q z), c = d) ∧ (∀ t : ℝ, MeasurePreserving (billiardFlow Q t) (phaseMeasure Q) (phaseMeasure Q)) ∧ (∀ s t : ℝ, billiardFlow Q (s + t) =ᵐ[phaseMeasure Q] (billiardFlow Q s ∘ billiardFlow Q t)) ∧ (∀ A : Set Phase, MeasurableSet A → (∀ t : ℝ, phaseMeasure Q (((billiardFlow Q t) ⁻¹' A) ∆ A) = 0) → phaseMeasure Q A = 0 ∨ phaseMeasure Q A = 1)
Possible paper match, with no recorded review
Atlas selected this possible match by comparing words in the Lean statement with every statement in the family. No review of whether the statements mean the same thing is recorded.
Words of the Lean statement, its names split into words, that also appear in the theorem: time, trajectories, vertex, specular, reflection, probability, direction, interior and twelve more. Found in the paper’s title, which its main theorem states: ergodicity. Shared only with the release’s name for the Lean statement or OpenAI’s note on it: ergodic, speed, least, one and unit.
Statements in Ergodicity of triangular billiards with an irrational angle come first because the release lists only that paper for this Lean statement.
Theorem 1 of Ergodicity of triangular billiards with an irrational angle
Theorem
Let Q be the interior of a nondegenerate Euclidean triangle with angles \alpha,\beta,\gamma. Suppose that at least one of
\alpha/\pi,\qquad \beta/\pi,\qquad \gamma/\pi
is irrational. The unit-speed billiard flow \Phi_t, with specular reflection at the open sides, is ergodic on Q\times S^1 for the probability measure
Here trajectories hitting a vertex in either time direction are discarded. In particular, if a measurable set A satisfies \mu(\Phi_t^{-1}A\mathbin\triangle A)=0 for every t\in\R, then \mu(A)\in\{0,1\}.
Let \(Q\) be the interior of a nondegenerate Euclidean triangle with angles
\(\alpha,\beta,\gamma\). Suppose that at least one of
\[
\alpha/\pi,\qquad \beta/\pi,\qquad \gamma/\pi
\]
is irrational.
The unit-speed billiard flow \(\Phi_t\), with specular reflection at the
open sides, is ergodic on \(Q\times S^1\) for the probability measure
\[
d\mu=\frac{dA\,d\theta}{2\pi\operatorname{Area}(Q)}.
\]
Here trajectories hitting a vertex in either time direction are discarded.
In particular, if a measurable set \(A\) satisfies
\(\mu(\Phi_t^{-1}A\mathbin\triangle A)=0\) for every \(t\in\R\), then
\(\mu(A)\in\{0,1\}\).
Other possible paper matches, closest wording first
Corollary 7 of Ergodicity of triangular billiards with an irrational angle. Words of the Lean statement, its names split into words, that also appear in the theorem: using, real, conclusions, outward, three, zero and 16 more. Shared only with the release’s name for the Lean statement or OpenAI’s note on it: unit and one.
Theorem 1.1 of Weak mixing of triangular billiards with an irrational angle. Words of the Lean statement, its names split into words, that also appear in the theorem: billiard, irrational, flow, normalized, measure, angle and four more. Shared only with the release’s name for the Lean statement or OpenAI’s note on it: relative, speed, least, one and unit.
Proposition 5.1 of Weak mixing of triangular billiards with an irrational angle. Words of the Lean statement, its names split into words, that also appear in the theorem: nonzero, irrational, times, angle, triangle, pi and two more. Shared only with the release’s name for the Lean statement or OpenAI’s note on it: relative, almost and everywhere.
Proposition 2.1 of Weak mixing of triangular billiards with an irrational angle. Words of the Lean statement, its names split into words, that also appear in the theorem: flow, regular, normalized, measure, triangle, nondegenerate and one more.
Proposition 5 of Ergodicity of triangular billiards with an irrational angle. Words of the Lean statement, its names split into words, that also appear in the theorem: regular. Shared only with the release’s name for the Lean statement or OpenAI’s note on it: almost and everywhere.
How Atlas selects possible matches: it compares words in the Lean statement with every statement in the family. It also splits Lean names into words. The release’s label and note receive less weight; a main theorem wins a close match. If the Lean wording yields no clear match, Atlas uses the catalogue description. This selects passages to compare; it does not establish that they mean the same thing.
Reported Comparator runs
three independent groups report that Comparator accepted runs for this Lean statement file: Carlos Toledo’s cert-machine; MathVet; Neal McBurnett’s openai-math-lean-tooling. Each group reports its own Comparator run for this statement file. Atlas has not repeated the runs.
For a configured theorem, Comparator tests whether OpenAI’s proof proves the exact Lean statement. Some files instead configure definitions for comparison by name and type. Acceptance does not establish that the statement means the same thing as the paper’s theorem.
Every Comparator challenge re-run on Linux with nanoda switched on, decided by a preregistered rule; every family’s challenge statements read against OpenAI’s headline, clause by clause. Decider: Comparator (leanprover/comparator d03acab1 = tag v4.34.0, built with the release’s toolchain v4.34.1), lean4export 076e8e57, landrun 811cfff5, nanoda_lib 3a240721 — on GitHub’s Linux runners (.github/workflows/openai-math-kernel.yml), one j…
AcceptedTheir term: “CERTIFIED”.
Re-run through Comparator with the independent nanoda kernel switched on: Lean’s kernel and nanoda both accept OpenAI’s solution.
An independent, model-assisted, pre-referee reading of each family’s challenge statements against its headline claim under a published rubric (PROTOCOL.md), with its own Comparator runs per challenge. Its protocol describes a planned error-rate study; this import does not establish that the study has been completed.
AcceptedTheir term: “comparator-sandboxed-pass”.
Re-run through Comparator in a sandboxed virtual machine: accepted.
Each challenge’s solution is built and checked by Comparator at the pinned release, on Lean leanprover/lean4:v4.34.1, recording the verdict, the check time and the peak memory; the run was still in progress at this snapshot (192 accepted, 224 pending).
AcceptedTheir term: “accepted”.
Re-run through Comparator with Lean’s kernel: accepted (the check took 185 s; peak memory 6.5 GiB).
If Comparator is run under this configuration and accepts, the solution proves the configured theorem using only the allowed axioms, and the Lean kernel accepts every step.
This configuration lists no definition holes, so every definition the statement uses is compared as it is written in this file.
Acceptance leaves open whether the Lean statement means the same thing as the paper’s theorem, whether the paper’s other claims are correct, and whether the result is new. Those questions require comparing the definitions and statements, reviewing the paper’s arguments, and reading the earlier literature.
Do family 150’s Lean statements express the advertised result?
These assessments compare all the family’s Lean statements with the whole result OpenAI describes in its catalogue. They do not assess this statement alone. Each group uses its own published rules.
Neither group finds that the Lean statements express or imply the whole advertised result. They describe what the statements express in different terms.
Finds part of it stated, or a narrower or weaker statementTheir term: “weaker-statement”.
The challenge statements do not imply the headline claim: they state a nontrivially weaker statement. Their record
Their note: MainConclusion states ergodicity (a.e.-invariant measurable sets have measure 0 or 1), plus a.e. existence and uniqueness of flight chains, measure preservation and the a.e. flow law; weak mixing, the summary's headline, is strictly stronger and is not stated (docs also only say 'ergodic').
Definitions the group asks readers to compare with the paper: billiardFlow Q t z := if h : Nonempty (FlightChain Q z) then (Classical.choice h).at t else z (flow defined by choice on chains; sound because the same statement proves chain uniqueness ∀ z c d : FlightChain Q z, c = d; singular states are frozen)
OpenAI’s own note on what this Lean statement covers
“The formalization proves that the unit-speed billiard flow in every nondegenerate Euclidean triangle with at least one angle irrational relative to $\pi$ is ergodic for normalized area times uniform angular measure. It constructs the flow outside the null set of exceptional trajectories, proves uniqueness of the flight chain there, and establishes measure preservation and the flow law almost everywhere. No genericity or Diophantine condition on the irrational angle is assumed.”
An AI agent drafted an interpretation of that note, classifying this statement as “states a paper’s theorem as the paper does”. No human review of the draft is recorded.
Supporting material
The complete Lean statement file
126 lines, copied from release version fd4aeeb. A name defined in this file links to its line; pointing at the name shows its definition.
Declared in this fileOpenAI’s library at fd4aeebOther linked names open the current documentation for Mathlib, Lean’s mathematical library.
Links in the file: 39 to its own declarations, 48 to Mathlib’s documentation. 17 declared names also link to the same declaration in OpenAI’s library. OpenAI uses Mathlib version d13f23b. The linked documentation describes the current version, so names that have since changed may be missing there.
import Mathlibnamespace OAI/-!The phase space is represented in the ambient plane × circle, with measuresupported on the interior of the triangle. Values on side-collision statesat a fixed time are immaterial to the modulo-null-set conclusion. Thetrajectories themselves include the collisions and impose specular reflection.-/open MeasureTheory Setopenscoped ENNReal symmDiffnoncomputablesectionnamespace TriangularBilliards/-- Three affinely independent vertices in the Euclidean plane. -/structureTrianglewherevertex : Fin3 → ℂnondegenerate : AffineIndependentℝ vertexnamespace Triangle/-- The open table; the complex plane carries its ordinary real Euclidean metric. -/deftable (Q : Triangle) : Setℂ := interior (convexHullℝ (range Q.vertex))/-- The open side from vertex `i` to the next vertex, indexed cyclically. -/defside (Q : Triangle) (i : Fin3) : Setℂ :=openSegmentℝ (Q.vertex i) (Q.vertex (i + 1))/-- A tangent vector to side `i`, with no choice of inward or outward normal. -/deftangent (Q : Triangle) (i : Fin3) : ℂ := Q.vertex (i + 1) - Q.vertex i/-- Interior angle at vertex `i`. -/defangle (Q : Triangle) (i : Fin3) : ℝ :=InnerProductGeometry.angle (Q.vertex (i + 1) - Q.vertex i) (Q.vertex (i + 2) - Q.vertex i)defHasIrrationalAngle (Q : Triangle) : Prop := ∃ i : Fin3, Irrational (Q.angle i / Real.pi)end Triangle/-- Reflection in the real line spanned by the nonzero tangent `e`.In coordinates with `e` real this is complex conjugation, preserving thetangential component and reversing the normal component. -/defreflect (e v : ℂ) : ℂ := (e / star e) * star vinstancecircleMeasurableSpace : MeasurableSpaceCircle := borelCircleinstancecircleBorelSpace : BorelSpaceCircle := ⟨rfl⟩abbrevPhase := ℂ × Circle/-- A complete nonsingular billiard trajectory, indexed by its collisions.Flight zero contains the initial time strictly. Both ends of the collisiontime sequence are unbounded; collisions occur only in open sides, neververtices. These are geometric conditions, not ergodicity/analytic assumptions. -/structureFlightChain (Q : Triangle) (z : Phase) wheretime : ℤ → ℝpoint : ℤ → ℂdirection : ℤ → Circlewall : ℤ → Fin3increasing : StrictMono timeunbounded_below : ∀ t : ℝ, ∃ n : ℤ, time n < tunbounded_above : ∀ t : ℝ, ∃ n : ℤ, t < time nzero_before : time 0 < 0zero_after : 0 < time 1on_side : ∀ n : ℤ, point n ∈ Q.side (wall n)flight : ∀ n : ℤ, point (n + 1) = point n + (time (n + 1) - time n) • (direction n : ℂ)inside : ∀ (n : ℤ) (t : ℝ), time n < t → t < time (n + 1) → point n + (t - time n) • (direction n : ℂ) ∈ Q.tablestart_point : z.1 = point 0 + (-time 0) • (direction 0 : ℂ)start_direction : z.2 = direction 0specular : ∀ n : ℤ, (direction n : ℂ) = reflect (Q.tangent (wall n)) (direction (n - 1) : ℂ)/-- The geometric state on a chain, using outgoing velocity at a collision. -/defFlightChain.at {Q : Triangle} {z : Phase} (c : FlightChain Q z) (t : ℝ) : Phase := byclassicalexactif h : ∃ n : ℤ, c.time n ≤ t ∧ t < c.time (n + 1) thenlet n := Classical.choose h (c.point n + (t - c.time n) • (c.direction n : ℂ), c.direction n)else z/-- The actual specular flow on complete regular chains. Singular initialstates are assigned their initial state; the theorem must prove these forma null set. No nonemptiness of the regular set is postulated. -/defbilliardFlow (Q : Triangle) (t : ℝ) (z : Phase) : Phase := byclassicalexactif h : Nonempty (FlightChain Q z) then (Classical.choice h).at t else z/-- Uniform angular probability, parameterized by [0,2π). -/defangularMeasure : MeasureCircle :=Measure.mapCircle.exp ((ENNReal.ofReal (2 * Real.pi))⁻¹ • volume.restrict (Ico0 (2 * Real.pi)))/-- Exactly normalized area in the open table times uniform angular measure. -/defphaseMeasure (Q : Triangle) : MeasurePhase := ((volume Q.table)⁻¹ • volume.restrict Q.table).prod angularMeasure/-- The main conclusion, together with the geometric/nonvacuity guaranteesneeded by the concrete definition of the actual two-sided billiard. -/defMainConclusion (Q : Triangle) : Prop :=phaseMeasure Q univ = 1 ∧ (∀ᵐ z ∂phaseMeasure Q, Nonempty (FlightChain Q z)) ∧ (∀ (z : Phase) (c d : FlightChain Q z), c = d) ∧ (∀ t : ℝ, MeasurePreserving (billiardFlow Q t) (phaseMeasure Q) (phaseMeasure Q)) ∧ (∀ s t : ℝ, billiardFlow Q (s + t) =ᵐ[phaseMeasure Q] (billiardFlow Q s ∘ billiardFlow Q t)) ∧ (∀ A : SetPhase, MeasurableSet A → (∀ t : ℝ, phaseMeasure Q (((billiardFlow Q t) ⁻¹' A) ∆ A) = 0) →phaseMeasure Q A = 0 ∨ phaseMeasure Q A = 1)theoremirrational_triangle_billiard (Q : Triangle) (hQ : Q.HasIrrationalAngle) :MainConclusion Q := bysorryend TriangularBilliardsendend OAI
Every extracted statement of the accompanying papers
The statements the atlas extracted automatically from each accompanying paper’s TeX source, with links to the full sources.
Here trajectories hitting a vertex in either time direction are discarded. In particular, if a measurable set A satisfies \mu(\Phi_t^{-1}A\mathbin\triangle A)=0 for every t\in\R, then \mu(A)\in\{0,1\}.
Let \(Q\) be the interior of a nondegenerate Euclidean triangle with angles
\(\alpha,\beta,\gamma\). Suppose that at least one of
\[
\alpha/\pi,\qquad \beta/\pi,\qquad \gamma/\pi
\]
is irrational.
The unit-speed billiard flow \(\Phi_t\), with specular reflection at the
open sides, is ergodic on \(Q\times S^1\) for the probability measure
\[
d\mu=\frac{dA\,d\theta}{2\pi\operatorname{Area}(Q)}.
\]
Here trajectories hitting a vertex in either time direction are discarded.
In particular, if a measurable set \(A\) satisfies
\(\mu(\Phi_t^{-1}A\mathbin\triangle A)=0\) for every \(t\in\R\), then
\(\mu(A)\in\{0,1\}\).
Here trajectories hitting a vertex in either time direction are discarded. In particular, if a measurable set A satisfies \mu(\Phi_t^{-1}A\mathbin\triangle A)=0 for every t\in\R, then \mu(A)\in\{0,1\}.
Ergodicity of triangular billiards with an irrational angle
TeX label thm:main, build/source/main.tex, line 88
A Lean statement states the theorem and leaves its proof as sorry on purpose. OpenAI’s proof is in a separate solution module, and a program called Comparator checks that proof against this statement.
Recorded comparisons with the paper
Atlas records no contributed comparison of this statement with the paper.
Contributed Comparator run reports
Atlas records no contributed report of running Comparator for this statement file.
Questions to resolve when comparing this statement with the paper
Both independent readers find this family's challenges state less than its headline, while the atlas's AI draft classes this file as stating a paper's theorem as the paper does; check which part of the headline it covers.