/- Copyright 2026 The Formal Conjectures Authors. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at https://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. -/ module public import FormalConjecturesUtil

Bugeaud Collection of Conjectures and Open Questions: Pisot orbits on the Cantor set

Problem 10.61, proposed by Mendès France [MF67, Problème 1, p. 41]. For a Pisot number $\alpha > 2$ put $$C(\alpha) = \left{ (\alpha - 1) \sum_{k \ge 1} \varepsilon_k \alpha^{-k} : \varepsilon_k \in {0, 1} \right}.$$ The problem asks to prove that $(\xi \alpha^n)_{n \ge 1}$ is not uniformly distributed modulo one for any $\xi \in C(\alpha)$. In the language of [MF67] this reads $C(\alpha) \cap B(\alpha) = \emptyset$, where $B(\alpha)$ is the set of $\xi$ for which $(\xi \alpha^n)$ is uniformly distributed modulo one.

The problem is open. It is easy for an integer base and settled for two quadratic $\alpha$; the general Pisot case is what remains. [Ste26] reduces it to a statement about shift-invariant measures of the full two-shift, and proves that every true instance admits a finite certificate, so the difficulty is uniformity in $\alpha$ rather than any single $\alpha$. The variants below state the criterion and the instances of [Ste26] that concern Problem 10.61 directly; the reduction itself is not stated here, since it needs the symbolic model.

Deleting the term $n = 0$ does not change uniform distribution modulo one, so the statements indexed from $n = 1$ agree with the ones of [Ste26], which are indexed from $n = 0$. Where an orbit is confined away from an interval the stronger form, for every $n \ge 0$, is stated.

References:

    [Bug12] Bugeaud, Yann. "Distribution modulo one and Diophantine approximation." Vol. 193. Cambridge University Press, 2012. Chapter 10, Problem 10.61, p. 222.

    [MF67] Mendès France, Michel. "Nombres normaux. Applications aux fonctions pseudo-aléatoires." Journal d'Analyse Mathématique 20 (1967): 1-56.

    [Kok35] Koksma, Jurjen F. "Ein mengentheoretischer Satz über die Gleichverteilung modulo Eins." Compositio Mathematica 2 (1935): 250-258.

    [Ste26] Stephan, Ralf. "Criteria for the non-equidistribution of $(\xi \alpha^n)$ on the Cantor set $C(\alpha)$." Preprint, 2026. https://doi.org/10.13140/RG.2.2.13923.52001

@[expose] public sectionnamespace Bugeaud61

The point of $C(\alpha)$ with digit sequence $\varepsilon$, that is $(\alpha - 1) \sum_{k \ge 1} \varepsilon_k \alpha^{-k}$.

noncomputable def cantorPoint (α : ℝ) (ε : ℕ → Bool) : ℝ := (α - 1) * ∑' k : ℕ, (if ε k then (1 : ℝ) else 0) * α⁻¹ ^ (k + 1)

The set $C(\alpha)$ of Problem 10.61. For $\alpha > 2$ it is a Cantor set of Hausdorff dimension $\log 2 / \log \alpha < 1$, normalised so that $0$ is its least and $1$ its greatest element.

noncomputable def pisotCantorSet (α : ℝ) : Set ℝ := Set.range (cantorPoint α)

The Route A exponent $A(\alpha) = \log 2 / \log \alpha + \log 2 / \log(1 / \rho)$ of [Ste26], where $\rho$ is the largest modulus of a conjugate of $\alpha$ other than $\alpha$ itself. It is the sum of the box dimensions of $C(\alpha)$ and of the window in which the conjugate contributions live.

noncomputable def routeAExponent (α ρ : ℝ) : ℝ := Real.log 2 / Real.log α + Real.log 2 / Real.log ρ⁻¹

Problem 10.61. Let $\alpha > 2$ be a Pisot number. For every $\xi \in C(\alpha)$ the sequence $(\xi \alpha^n)_{n \ge 1}$ is not uniformly distributed modulo one.

@[category research open, AMS 11 37] theorem problem_10_61 (α : ℝ) (hα : IsPisot α) (hα2 : 2 < α) : ∀ ξ ∈ pisotCantorSet α, ¬ IsEquidistributedModuloOne fun n : ℕ => ξ * α ^ (n + 1) := α:ℝhα:IsPisot αhα2:2 < α⊢ ∀ ξ ∈ pisotCantorSet α, ¬IsEquidistributedModuloOne fun n ↦ ξ * α ^ (n + 1) All goals completed! 🐙

The hypothesis $\xi \in C(\alpha)$ cannot be dropped: by Koksma's metric theorem [Kok35], for a fixed $\alpha > 1$ the sequence $(\xi \alpha^n)_{n \ge 1}$ is uniformly distributed modulo one for Lebesgue-almost every $\xi$. Problem 10.61 asks a null set of starting points to defeat an almost-everywhere law.

@[category research solved, AMS 11 37] theorem problem_10_61.variants.almost_every (α : ℝ) (hα : 1 < α) : ∀ᵐ ξ ∂(MeasureTheory.volume : MeasureTheory.Measure ℝ), IsEquidistributedModuloOne fun n : ℕ => ξ * α ^ (n + 1) := α:ℝhα:1 < α⊢ ∀ᵐ (ξ : ℝ), IsEquidistributedModuloOne fun n ↦ ξ * α ^ (n + 1) All goals completed! 🐙

The covering criterion, [Ste26, Theorem C(i)], at a quadratic setup: $\alpha > 1$ is a root of $X^2 - aX - b$ whose conjugate $\beta = a - \alpha$ is non-zero of modulus less than one. If the Route A exponent $A(\alpha)$ formed from $\alpha$ and $\rho = |\beta|$ is below $1$, then Problem 10.61 holds at $\alpha$.

[Ste26] states the criterion for a Pisot number of any degree $d \ge 2$, but proves it only in this degree-two case.

@[category research solved, AMS 11 37, formal_proof using lean4 at "https://github.com/rwst/Pisot-Cantor-61/blob/a464f1cd3c8d14229a2f1d7881773987446d0df0/BB61/Criterion.lean#L230"] theorem problem_10_61.variants.covering_criterion {a b : ℤ} {α : ℝ} (hroot : α ^ 2 = a * α + b) (hα : 1 < α) (hβ : (a : ℝ) - α ≠ 0) (hconj : |(a : ℝ) - α| < 1) (hA : routeAExponent α |(a : ℝ) - α| < 1) : ∀ ξ ∈ pisotCantorSet α, ¬ IsEquidistributedModuloOne fun n : ℕ => ξ * α ^ (n + 1) := a:ℤb:ℤα:ℝhroot:α ^ 2 = ↑a * α + ↑bhα:1 < αhβ:↑a - α ≠ 0hconj:|↑a - α| < 1hA:routeAExponent α |↑a - α| < 1⊢ ∀ ξ ∈ pisotCantorSet α, ¬IsEquidistributedModuloOne fun n ↦ ξ * α ^ (n + 1) All goals completed! 🐙

The exact range of the covering criterion, [Ste26, Theorem C(ii)]: at a quadratic setup of norm $-b$ one has $A(\alpha) < 1$ if and only if $(\log_2 \alpha - 1)(\log_2(\alpha / |b|) - 1) > 1$, which for units reads $\alpha > 4$. The criterion therefore never reaches the slice $2 < \alpha \le 4$, and says nothing about a quadratic $\alpha$ with $A(\alpha) \ge 1$.

@[category research solved, AMS 11 37, formal_proof using lean4 at "https://github.com/rwst/Pisot-Cantor-61/blob/a464f1cd3c8d14229a2f1d7881773987446d0df0/BB61/RouteANormalForm.lean#L283"] theorem problem_10_61.variants.covering_criterion_range {a b : ℤ} {α : ℝ} (hroot : α ^ 2 = a * α + b) (hα : 1 < α) (hconj : |(a : ℝ) - α| < 1) (hb : b ≠ 0) : routeAExponent α |(a : ℝ) - α| < 1 ↔ 1 < (Real.logb 2 α - 1) * (Real.logb 2 α - Real.logb 2 |(b : ℝ)| - 1) := a:ℤb:ℤα:ℝhroot:α ^ 2 = ↑a * α + ↑bhα:1 < αhconj:|↑a - α| < 1hb:b ≠ 0⊢ routeAExponent α |↑a - α| < 1 ↔ 1 < (Real.logb 2 α - 1) * (Real.logb 2 α - Real.logb 2 |↑b| - 1) All goals completed! 🐙

[Ste26, Theorem D] at $\alpha = 2 + \sqrt 5$, the smallest $\alpha$ the covering criterion reaches: Problem 10.61 holds there in the strong form. There is one open interval $J \subseteq (0, 1)$, the same for every $\xi \in C(\alpha)$ and every $n \ge 0$, that ${\xi \alpha^n}$ misses.

@[category research solved, AMS 11 37, formal_proof using lean4 at "https://github.com/rwst/Pisot-Cantor-61/blob/a464f1cd3c8d14229a2f1d7881773987446d0df0/BB61/RouteA.lean#L163"] theorem problem_10_61.variants.two_add_sqrt_five : ∃ x r : ℝ, 0 < r ∧ Set.Ioo (x - r) (x + r) ⊆ Set.Ioo 0 1 ∧ ∀ ξ ∈ pisotCantorSet (2 + Real.sqrt 5), ∀ n : ℕ, Int.fract (ξ * (2 + Real.sqrt 5) ^ n) ∉ Set.Ioo (x - r) (x + r) := ⊢ ∃ x r, 0 < r ∧ Set.Ioo (x - r) (x + r) ⊆ Set.Ioo 0 1 ∧ ∀ ξ ∈ pisotCantorSet (2 + √5), ∀ (n : ℕ), Int.fract (ξ * (2 + √5) ^ n) ∉ Set.Ioo (x - r) (x + r) All goals completed! 🐙

[Ste26, Theorem D] at $\alpha = 2 + \sqrt 3$: Problem 10.61 holds there, by a confinement-gap certificate in exact $\mathbb{Z}[\sqrt 3]$ arithmetic. This $\alpha$ lies in the slice $2 < \alpha \le 4$, which the covering criterion does not reach.

@[category research solved, AMS 11 37, formal_proof using lean4 at "https://github.com/rwst/Pisot-Cantor-61/blob/a464f1cd3c8d14229a2f1d7881773987446d0df0/BB61/GapSqrtThree.lean#L298"] theorem problem_10_61.variants.two_add_sqrt_three : ∀ ξ ∈ pisotCantorSet (2 + Real.sqrt 3), ¬ IsEquidistributedModuloOne fun n : ℕ => ξ * (2 + Real.sqrt 3) ^ (n + 1) := ⊢ ∀ ξ ∈ pisotCantorSet (2 + √3), ¬IsEquidistributedModuloOne fun n ↦ ξ * (2 + √3) ^ (n + 1) All goals completed! 🐙

Sanity check: $0 \in C(\alpha)$, the point with all digits zero.

@[category test, AMS 11 37] theorem zero_mem_pisotCantorSet (α : ℝ) : 0 ∈ pisotCantorSet α := ⟨fun _ => false, α:ℝ⊢ (cantorPoint α fun x ↦ false) = 0 All goals completed! 🐙⟩

Sanity check: $1 \in C(\alpha)$ for $\alpha > 1$, the point with all digits one, since $\sum_{k \ge 1} \alpha^{-k} = 1 / (\alpha - 1)$.

α:ℝhα:1 < αhα0:0 < αhinv:α⁻¹ < 1hsum:∑' (k : ℕ), (if (fun x ↦ true) k = true then 1 else 0) * α⁻¹ ^ (k + 1) = α⁻¹ * (1 - α⁻¹)⁻¹⊢ (α - 1) * (α⁻¹ * (1 - α⁻¹)⁻¹) = 1 α:ℝhα:1 < αhα0:0 < αhinv:α⁻¹ < 1hsum:∑' (k : ℕ), (if (fun x ↦ true) k = true then 1 else 0) * α⁻¹ ^ (k + 1) = α⁻¹ * (1 - α⁻¹)⁻¹h1:α - 1 ≠ 0⊢ (α - 1) * (α⁻¹ * (1 - α⁻¹)⁻¹) = 1 All goals completed! 🐙end Bugeaud61