Distribution of Zeckendorf expressions

TheoremPowers of two are the only binary fundamental sequence#●ready to prove

Let \(H \colon \mathbb{N} \to \mathbb{N}\) be a strictly increasing sequence of positive integers. Then every natural number (including \(0\)) equals \(\sum_{k \in s} H_k\) for exactly one finite set \(s \subseteq \mathbb{N}\) if and only if \(H_k = 2^k\) for all \(k\).

public theorem_wanted bijective_finset_sum_iff_eq_two_pow {H : ℕ → ℕ} (hpos : ∀ k, 0 < H k)
    (hmono : StrictMono H) :
    Function.Bijective (fun s : Finset ℕ => ∑ k ∈ s, H k) ↔ ∀ k, H k = 2 ^ k
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 1.

TheoremThe Fibonacci numbers are the only Zeckendorf fundamental sequence#●ready to prove

Let \(H \colon \mathbb{N} \to \mathbb{N}\) be a strictly increasing sequence of positive integers, and call a finite set \(s \subseteq \mathbb{N}\) nonadjacent if \(k \in s\) implies \(k + 1 \notin s\). Then every natural number (including \(0\)) equals \(\sum_{k \in s} H_k\) for exactly one nonadjacent finite set \(s\) if and only if \(H_k = F_{k+2}\) for all \(k\), where \(F_0 = 0\), \(F_1 = 1\) and \(F_{n+2} = F_{n+1} + F_n\) are the Fibonacci numbers.

public theorem_wanted bijective_nonadjacent_finset_sum_iff_eq_fib {H : ℕ → ℕ}
    (hpos : ∀ k, 0 < H k) (hmono : StrictMono H) :
    Function.Bijective (fun s : {s : Finset ℕ // ∀ k ∈ s, k + 1 ∉ s} => ∑ k ∈ s.1, H k) ↔
      ∀ k, H k = Nat.fib (k + 2)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 1.

TheoremDuality formula for binary expansions without adjacent ones#●ready to prove

For \(x \in \mathbb{N}\) let \(z(x)\) be the number of \(n < x\) whose binary expansion has no two adjacent digits equal to \(1\). For a finite set \(A \subseteq \mathbb{N}\), put \(\bar A = A\) if there is no \(j\) with \(j, j + 1 \in A\), and otherwise \(\bar A = \{k \in A : k \ge j\}\), where \(j\) is the largest index with \(j, j + 1 \in A\). Then \(z\bigl(\sum_{k \in A} 2^k\bigr) = \sum_{k \in \bar A} F_{k+2}\), where \(F_0 = 0\), \(F_1 = 1\) and \(F_{n+2} = F_{n+1} + F_n\) are the Fibonacci numbers.

public theorem_wanted card_zeckendorfBinary_lt_sum_two_pow (A : Finset ℕ) :
    let J := A.filter (fun j => j + 1 ∈ A)
    let Abar := if h : J.Nonempty then A.filter (fun k => J.max' h ≤ k) else A
    ((Finset.range (∑ k ∈ A, 2 ^ k)).filter (fun n => n &&& (n >>> 1) = 0)).card =
      ∑ k ∈ Abar, Nat.fib (k + 2)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 2.

DefinitionAscending lexicographic order#✓fully proved

For coefficient functions \(\mu, \nu \colon \mathbb{N} \to \mathbb{N}\) with finite support, \(\mu \lt _a \nu\) means that there is an index \(k\) with \(\mu_k < \nu_k\) and \(\mu_j = \nu_j\) for all \(j > k\): at the largest index where they differ, \(\mu\) is smaller.

public def AscLexLT (μ ν : ℕ →₀ ℕ) : Prop :=
  ∃ k, μ k < ν k ∧ ∀ j, k < j → μ j = ν j
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.2).
Used by 7

TheoremExample of the ascending lexicographic order#✓fully proved

Write \((a_0, a_1, a_2, a_3, a_4)\) for the coefficient function with finite support whose entry at index \(i \le 4\) is \(a_i\) and whose other entries are \(0\). Then \((1, 2, 10, 3, 7) \lt _a (1, 3, 1, 4, 7)\) in the ascending lexicographic order: the largest index where they differ is \(3\), where \(3 < 4\).

public theorem ascLexLT_example :
    AscLexLT (Finsupp.single 0 1 + Finsupp.single 1 2 + Finsupp.single 2 10 +
        Finsupp.single 3 3 + Finsupp.single 4 7)
      (Finsupp.single 0 1 + Finsupp.single 1 3 + Finsupp.single 2 1 + Finsupp.single 3 4 +
        Finsupp.single 4 7)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.2, example).
Uses 1

DefinitionImmediate predecessor#✓fully proved

Let \(E\) be a collection of finitely supported coefficient functions \(\mathbb N \to \mathbb N\), and let \(\lt _a\) be the ascending lexicographic order. A finitely supported coefficient function \(\mu\) is the immediate predecessor of \(\nu\) in \(E\) if \(\mu \in E\), \(\mu \lt _a \nu\), and every \(\xi \in E\) with \(\xi \lt _a \nu\) satisfies \(\xi = \mu\) or \(\xi \lt _a \mu\). That is, \(\mu\) is the largest element of \(E\) that is \(\lt _a \nu\).

public def IsImmediatePredecessor (E : Set (ℕ →₀ ℕ)) (ν μ : ℕ →₀ ℕ) : Prop :=
  μ ∈ E ∧ AscLexLT μ ν ∧ ∀ ξ ∈ E, AscLexLT ξ ν → ξ = μ ∨ AscLexLT ξ μ
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.2).
Uses 1 · Used by 10

DefinitionImmediate successor#✓fully proved

Let \(E\) be a collection of finitely supported coefficient functions \(\mathbb N \to \mathbb N\), and let \(\lt _a\) be the ascending lexicographic order. A finitely supported coefficient function \(\nu\) is the immediate successor of \(\mu\) in \(E\) if \(\nu \in E\), \(\mu \lt _a \nu\), and every \(\xi \in E\) with \(\mu \lt _a \xi\) satisfies \(\xi = \nu\) or \(\nu \lt _a \xi\). That is, \(\nu\) is the smallest element of \(E\) that is \(>_a \mu\).

public def IsImmediateSuccessor (E : Set (ℕ →₀ ℕ)) (μ ν : ℕ →₀ ℕ) : Prop :=
  ν ∈ E ∧ AscLexLT μ ν ∧ ∀ ξ ∈ E, AscLexLT μ ξ → ξ = ν ∨ AscLexLT ν ξ
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.2).
Uses 1 · Used by 6

DefinitionZeckendorf collections#✓fully proved

Index coefficient functions from \(1\), let \(\beta^n\) be the basis coefficient function with entry \(1\) at index \(n\), and for a coefficient function \(\mu\) let \(\operatorname{res}_n(\mu)\) keep the entries of \(\mu\) at indices \(\le n\) and \(\operatorname{res}^n(\mu)\) keep those at indices \(\ge n\), the other entries being \(0\). A set \(E\) of finitely supported coefficient functions \(\mathbb N \to \mathbb N\) is a Zeckendorf collection when it contains \(0\) and \(\beta^n\) for every \(n \ge 1\), every \(\mu \in E\) has only finitely many elements of \(E\) below it in the ascending lexicographic order \(\lt _a\), and for every \(\mu \in E\) and every immediate successor \(\tilde\mu\) of \(\mu\) in \(E\) with \(\tilde\mu \ne \beta^1 + \mu\) there is an index \(n \ge 2\) such that \(\operatorname{res}_{n-1}(\mu)\) is the immediate predecessor of \(\beta^n\) in \(E\) and \(\tilde\mu = \beta^n + \operatorname{res}^n(\mu)\).

public def IsZeckendorfCollection (E : Set (ℕ →₀ ℕ)) : Prop :=
  0 ∈ E ∧ (∀ i, Finsupp.single i 1 ∈ E) ∧ (∀ μ ∈ E, {ν | ν ∈ E ∧ AscLexLT ν μ}.Finite) ∧
    ∀ μ ∈ E, ∀ ν, IsImmediateSuccessor E μ ν → ν ≠ Finsupp.single 0 1 + μ →
      ∃ m, 1 ≤ m ∧ IsImmediatePredecessor E (Finsupp.single m 1) (μ.filter (· < m)) ∧
        ν = Finsupp.single m 1 + μ.filter (m ≤ ·)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.2).
Uses 3 · Used by 7

TheoremZeckendorf collections contain zero#✓fully proved

Every Zeckendorf collection \(E\) of finitely supported coefficient functions \(\mathbb N \to \mathbb N\) contains the zero coefficient function.

public theorem IsZeckendorfCollection.zero_mem {E : Set (ℕ →₀ ℕ)} (h : IsZeckendorfCollection E) :
    0 ∈ E
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.2).
Uses 4

DefinitionThe maximal \(L\)-block \(\theta^{m+1}\)#✓fully proved

Let \(L = (e_1, \dots, e_N)\) be a list of natural numbers and \(m \in \mathbb N\), with indices of coefficient functions starting at \(0\). The maximal \(L\)-block \(\theta^{m+1}\) is the finitely supported coefficient function whose entry at each index \(i < m\) is \(e_j\), where \(j - 1 = (m - 1 - i) \bmod N\), and whose entries at the indices \(i \ge m\) are \(0\) (every entry is \(0\) when \(L\) is empty).

In the paper's one-based indexing, \(\theta^n = \mathrm{rev}_{n-1}(\beta^*)\), where \(\beta^*\) is the coefficient function with \(\beta^*_k = e_j\) for \(k \equiv j \pmod N\) and \(\mathrm{rev}_{n-1}(\beta^*)\) lists the first \(n - 1\) entries of \(\beta^*\) in reverse order, followed by zeros.

public noncomputable def periodicTheta (L : List ℕ) (m : ℕ) : ℕ →₀ ℕ :=
  ∑ i ∈ Finset.range m, Finsupp.single i (L.getD ((m - 1 - i) % L.length) 0)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 10 (arXiv:2308.12872v1, Section 2.2).
Used by 4

TheoremThe maximal block \(\theta^4\) in the Fibonacci case#✓fully proved

For \(L = (1, 0)\), the maximal \(L\)-block \(\theta^4\) is \((1, 0, 1)\): its entries at indices \(0\), \(1\), \(2\) are \(1\), \(0\), \(1\), and all its entries from index \(3\) on are \(0\).

public theorem periodicTheta_one_zero_three :
    periodicTheta [1, 0] 3 = Finsupp.single 0 1 + Finsupp.single 2 1
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 10 (arXiv:2308.12872v1, Section 2.2).
Uses 1

DefinitionL-blocks#✓fully proved

Let \(L\) be a list of natural numbers, and for each \(m\) let \(\theta^{m+1}\) be the maximal \(L\)-block, which vanishes from index \(m\) on (indices are zero-based). A finitely supported coefficient function \(\zeta\) is an \(L\)-block with support interval \([s, m - 1]\) if \(1 \le m\), \(s < m\), and either \(s = 0\) and \(\zeta = \theta^{m+1}\), or \(\zeta_i = \theta^{m+1}_i\) for every \(i > s\), \(\zeta_s < \theta^{m+1}_s\) and \(\zeta_i = 0\) for every \(i < s\).

In the paper's indexing this is a block at index \(n - 1\) with \(n = m + 1 \ge 2\).

public def IsLBlock (L : List ℕ) (ζ : ℕ →₀ ℕ) (lo m : ℕ) : Prop :=
  1 ≤ m ∧ lo < m ∧
    ((lo = 0 ∧ ζ = periodicTheta L m) ∨
      (ζ.filter (lo < ·) = (periodicTheta L m).filter (lo < ·) ∧ ζ lo < periodicTheta L m lo ∧
        ζ.filter (· < lo) = 0))
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 10 (arXiv:2308.12872v1, Section 2.2).
Uses 1 · Used by 1

DefinitionPeriodic Zeckendorf collection#✓fully proved

For a list \(L\) of natural numbers, the periodic Zeckendorf collection determined by \(L\) is the set of finitely supported coefficient functions \(\mu\) that can be written as a sum \(\mu = \zeta_1 + \dots + \zeta_r\) of finitely many \(L\)-blocks \(\zeta_1, \dots, \zeta_r\) (possibly none, which gives \(\mu = 0\)) whose support intervals are pairwise disjoint.

public def periodicZeckendorfCollection (L : List ℕ) : Set (ℕ →₀ ℕ) :=
  {μ | ∃ blocks : List ((ℕ →₀ ℕ) × ℕ × ℕ), (∀ b ∈ blocks, IsLBlock L b.1 b.2.1 b.2.2) ∧
    blocks.Pairwise (fun a b => a.2.2 ≤ b.2.1 ∨ b.2.2 ≤ a.2.1) ∧ μ = (blocks.map Prod.fst).sum}
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 4 and the definition following it (arXiv:2308.12872v1, Section 2.2).
Uses 1 · Used by 14

DefinitionEvaluation of a coefficient function at an integer sequence#✓fully proved

For a sequence \(H\) of natural numbers and a finitely supported coefficient function \(\varepsilon \colon \mathbb{N} \to \mathbb{N}\), the evaluation of \(\varepsilon\) at \(H\) is the natural number \(\mathrm{eval}_H(\varepsilon) = \sum_{k \ge 0} \varepsilon_k H_k\).

public def zeckendorfEval (H : ℕ → ℕ) (ε : ℕ →₀ ℕ) : ℕ :=
  ε.sum fun k a => a * H k
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 1 (arXiv:2308.12872v1, Section 1).
Used by 9

TheoremZeckendorf collections are determined by their immediate predecessors#●ready to prove

Index coefficient functions from \(0\), and for \(m \in \mathbb N\) let \(\beta^{(m)}\) be the coefficient function with entry \(1\) at index \(m\) and \(0\) elsewhere. Suppose that for each \(m \ge 1\) a finitely supported coefficient function \(\theta^{(m)}\) is given whose entry at index \(m - 1\) is nonzero and whose entries at all indices \(j \ge m\) are \(0\). Then there is a unique Zeckendorf collection \(E\) such that, for every \(m \ge 1\), \(\theta^{(m)}\) is the immediate predecessor of \(\beta^{(m)}\) in \(E\).

public theorem_wanted existsUnique_zeckendorfCollection (θ : ℕ → ℕ →₀ ℕ)
    (hθ : ∀ m, 1 ≤ m → θ m (m - 1) ≠ 0 ∧ ∀ j, m ≤ j → θ m j = 0) :
    ∃! E : Set (ℕ →₀ ℕ), IsZeckendorfCollection E ∧
      ∀ m, 1 ≤ m → IsImmediatePredecessor E (Finsupp.single m 1) (θ m)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 3 (arXiv:2308.12872v1, Section 2.2, quoting Chang 2021).
Uses 2

TheoremPeriodic collections are Zeckendorf collections#●ready to prove

Let \(L = (e_1, \dots, e_N)\) be a list of natural numbers with \(N \ge 2\) and \(e_1 \ne 0\), and let \(E\) be the periodic Zeckendorf collection determined by \(L\) (the sums of finitely many \(L\)-blocks with pairwise disjoint support intervals). Index coefficient functions from \(1\) and let \(\beta^n\) be the basis coefficient function with entry \(1\) at index \(n\). Then \(E\) is a Zeckendorf collection, and for every \(n \ge 2\) the immediate predecessor of \(\beta^n\) in \(E\) is the maximal \(L\)-block \(\theta^n\).

public theorem_wanted isZeckendorfCollection_periodic (L : List ℕ) (hL : 2 ≤ L.length)
    (he : L.getD 0 0 ≠ 0) :
    IsZeckendorfCollection (periodicZeckendorfCollection L) ∧
      ∀ m, 1 ≤ m → IsImmediatePredecessor (periodicZeckendorfCollection L)
        (Finsupp.single m 1) (periodicTheta L m)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 4 (arXiv:2308.12872v1, Section 2.2, quoting Chang 2021).
Uses 4

TheoremFundamental sequences of Zeckendorf collections#●ready to prove

Let \(E\) be a Zeckendorf collection and index coefficient functions and sequences from \(0\). Suppose that, for each \(m \ge 1\), \(\theta^{(m)}\) is the immediate predecessor in \(E\) of the coefficient function with entry \(1\) at index \(m\) and \(0\) elsewhere. For a sequence \(H\) write \(\mathrm{eval}_H(\mu) = \sum_k \mu_k H_k\). Then there is a unique strictly increasing sequence \(H\) of positive integers such that \(\mathrm{eval}_H\) is a bijection from \(E\) onto \(\mathbb N\). Moreover, every such \(H\) satisfies \(H_m = H_0 + \mathrm{eval}_H(\theta^{(m)})\) for all \(m \ge 1\), and for all \(\mu, \nu \in E\), \(\mu \lt _a \nu\) (that is, \(\mu_k < \nu_k\) at the largest index \(k\) where \(\mu\) and \(\nu\) differ) if and only if \(\mathrm{eval}_H(\mu) < \mathrm{eval}_H(\nu)\).

public theorem_wanted existsUnique_zeckendorf_fundamentalSequence {E : Set (ℕ →₀ ℕ)}
    (hE : IsZeckendorfCollection E) (θ : ℕ → ℕ →₀ ℕ)
    (hθ : ∀ m, 1 ≤ m → IsImmediatePredecessor E (Finsupp.single m 1) (θ m)) :
    (∃! H : ℕ → ℕ, StrictMono H ∧ (∀ k, 0 < H k) ∧ Set.BijOn (zeckendorfEval H) E Set.univ) ∧
      ∀ H : ℕ → ℕ, StrictMono H → (∀ k, 0 < H k) → Set.BijOn (zeckendorfEval H) E Set.univ →
        (∀ m, 1 ≤ m → H m = H 0 + zeckendorfEval H (θ m)) ∧
          ∀ μ ∈ E, ∀ ν ∈ E, (AscLexLT μ ν ↔ zeckendorfEval H μ < zeckendorfEval H ν)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 5 (arXiv:2308.12872v1, Section 2.2, quoting Chang 2021).
Uses 4

TheoremCriterion for one Zeckendorf collection to contain another#●ready to prove

Let \(E\) and \(\tilde E\) be Zeckendorf collections. For \(n \ge 2\) let \(\beta^n\) be the basis coefficient function with entry \(1\) at position \(n\) (positions counted from \(1\)) and \(0\) elsewhere, and let \(\hat\beta^n\) be the immediate predecessor of \(\beta^n\) in \(E\) (the largest element of \(E\) that is \(\lt _a \beta^n\) in the ascending lexicographic order). Then \(E \subseteq \tilde E\) if and only if \(\hat\beta^n \in \tilde E\) for all \(n \ge 2\).

public theorem_wanted zeckendorfCollection_subset_iff {E E' : Set (ℕ →₀ ℕ)}
    (hE' : IsZeckendorfCollection E') (hE : IsZeckendorfCollection E) (θ : ℕ → ℕ →₀ ℕ)
    (hθ : ∀ m, 1 ≤ m → IsImmediatePredecessor E (Finsupp.single m 1) (θ m)) :
    E ⊆ E' ↔ ∀ m, 1 ≤ m → θ m ∈ E'
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Proposition 1 (arXiv:2308.12872v1, Section 2.2).
Uses 2

DefinitionCounting function of a subcollection#✓fully proved

For a set \(E\) of finitely supported coefficient functions \(\mu : \mathbb{N} \to \mathbb{N}\), a sequence \(H : \mathbb{N} \to \mathbb{N}\) and a real number \(x\), \(z(x)\) is the number of natural numbers \(n < x\) of the form \(n = \sum_k \mu_k H_k\) for some \(\mu \in E\).

public noncomputable def subcollectionCount (E : Set (ℕ →₀ ℕ)) (H : ℕ → ℕ) (x : ℝ) : ℕ :=
  {n : ℕ | (n : ℝ) < x ∧ ∃ μ ∈ E, zeckendorfEval H μ = n}.ncard
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 1 and Section 3 (arXiv:2308.12872v1, Sections 1 and 3.2).
Uses 1 · Used by 5

TheoremThe counting function \(z\) vanishes for \(x \le 0\)#✓fully proved

Let \(E\) be a set of finitely supported coefficient functions and \(H\) a sequence of natural numbers, and for real \(x\) let \(z(x)\) be the number of natural numbers \(n < x\) of the form \(n = \sum_k \mu_k H_k\) with \(\mu \in E\). Then \(z(x) = 0\) for every real \(x \le 0\).

public theorem subcollectionCount_of_nonpos (E : Set (ℕ →₀ ℕ)) (H : ℕ → ℕ) {x : ℝ}
    (hx : x ≤ 0) : subcollectionCount E H x = 0
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 3 (arXiv:2308.12872v1, Section 3.2).
Uses 2

TheoremPower-law bounds for the subcollection count#●ready to prove

Let \(L\) and \(\tilde L\) be lists of natural numbers of length at least \(2\) with nonzero first entries, and let \(E\) and \(\tilde E\) be the periodic Zeckendorf collections they determine, with \(E\) a proper subset of \(\tilde E\). Let \(\tilde H\) be a strictly increasing sequence of positive integers such that \(\mu \mapsto \sum_k \mu_k \tilde H_k\) is a bijection from \(\tilde E\) onto \(\mathbb N\), and let \(z(x)\) be the number of natural numbers \(n < x\) of the form \(\sum_k \mu_k \tilde H_k\) with \(\mu \in E\). Then there are reals \(\gamma\), \(a\) and \(b\) with \(0 < \gamma < 1\), \(a > 0\) and \(b > 0\) such that \(a < z(x) / x^\gamma < b\) for all sufficiently large real \(x\).

public theorem_wanted exists_rpow_bounds_subcollectionCount (L L' : List ℕ) (hL : 2 ≤ L.length)
    (he : L.getD 0 0 ≠ 0) (hL' : 2 ≤ L'.length) (he' : L'.getD 0 0 ≠ 0)
    (hsub : periodicZeckendorfCollection L ⊂ periodicZeckendorfCollection L') {H' : ℕ → ℕ}
    (hmono : StrictMono H') (hpos : ∀ k, 0 < H' k)
    (hbij : Set.BijOn (zeckendorfEval H') (periodicZeckendorfCollection L') Set.univ) :
    ∃ γ a b : ℝ, 0 < γ ∧ γ < 1 ∧ 0 < a ∧ 0 < b ∧ ∀ᶠ x in Filter.atTop,
      a < (subcollectionCount (periodicZeckendorfCollection L) H' x : ℝ) / x ^ γ ∧
        (subcollectionCount (periodicZeckendorfCollection L) H' x : ℝ) / x ^ γ < b
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 1 (arXiv:2308.12872v1, Section 1, with the standing assumption of Section 3 that E is a proper subcollection).
Uses 3

TheoremLimsup of the count of binary expansions without adjacent ones#●ready to prove

For real \(x\) let \(z(x)\) be the number of natural numbers \(n < x\) (including \(0\)) whose binary expansion has no two adjacent \(1\)s, let \(\varphi\) be the golden ratio and let \(\gamma = \log_2 \varphi\). Then \(\limsup_{x \to \infty} z(x) / x^{\gamma} = 3^{\gamma} (\varphi + 2) / 5\).

public theorem_wanted limsup_zeckendorfBinaryCount_div_rpow :
    Filter.limsup (fun x : ℝ =>
      ({n : ℕ | (n : ℝ) < x ∧ n &&& (n >>> 1) = 0}.ncard : ℝ) /
        x ^ Real.logb 2 Real.goldenRatio) Filter.atTop =
      3 ^ Real.logb 2 Real.goldenRatio * (Real.goldenRatio + 2) / 5
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 1, Equation (1) (arXiv:2308.12872v1, Section 1).

TheoremLiminf of the count of binary expansions without adjacent ones#●ready to prove

For real \(x\) let \(z(x)\) be the number of natural numbers \(n < x\) (including \(0\)) whose binary expansion has no two adjacent \(1\)s, let \(\varphi\) be the golden ratio and let \(\gamma = \log_2 \varphi\). Then \(\liminf_{x \to \infty} z(x) / x^{\gamma} = (3\varphi + 1) / 5\).

public theorem_wanted liminf_zeckendorfBinaryCount_div_rpow :
    Filter.liminf (fun x : ℝ =>
      ({n : ℕ | (n : ℝ) < x ∧ n &&& (n >>> 1) = 0}.ncard : ℝ) /
        x ^ Real.logb 2 Real.goldenRatio) Filter.atTop =
      (3 * Real.goldenRatio + 1) / 5
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 1, Equation (1) (arXiv:2308.12872v1, Section 1).

TheoremAsymptotic form of the subcollection count#●ready to prove

Let \(L\) and \(\tilde L\) be lists of natural numbers of length at least \(2\) with nonzero first entries, and let \(E\) and \(\tilde E\) be the periodic Zeckendorf collections they determine, with \(E\) a proper subset of \(\tilde E\). Let \(\tilde H\) be a strictly increasing sequence of positive integers such that \(\mu \mapsto \sum_k \mu_k \tilde H_k\) is a bijection from \(\tilde E\) onto \(\mathbb N\), and let \(z(x)\) be the number of natural numbers \(n < x\) of the form \(\sum_k \mu_k \tilde H_k\) with \(\mu \in E\). Then there are reals \(\gamma\) and \(\tilde\varphi\) with \(0 < \gamma < 1\) and \(\tilde\varphi > 1\), and a continuous function \(\Phi : \mathbb R \to \mathbb R\), such that \(z(x) \sim x^\gamma \Phi(\{\log_{\tilde\varphi} x\})\) as \(x \to \infty\), where \(\{t\}\) is the fractional part of \(t\), and \(\Phi\) is differentiable at almost every point of \((0, 1)\).

public theorem_wanted exists_isEquivalent_subcollectionCount (L L' : List ℕ) (hL : 2 ≤ L.length)
    (he : L.getD 0 0 ≠ 0) (hL' : 2 ≤ L'.length) (he' : L'.getD 0 0 ≠ 0)
    (hsub : periodicZeckendorfCollection L ⊂ periodicZeckendorfCollection L') {H' : ℕ → ℕ}
    (hmono : StrictMono H') (hpos : ∀ k, 0 < H' k)
    (hbij : Set.BijOn (zeckendorfEval H') (periodicZeckendorfCollection L') Set.univ) :
    ∃ γ φ' : ℝ, 0 < γ ∧ γ < 1 ∧ 1 < φ' ∧ ∃ Φ : ℝ → ℝ, Continuous Φ ∧
      Asymptotics.IsEquivalent Filter.atTop
        (fun x : ℝ => (subcollectionCount (periodicZeckendorfCollection L) H' x : ℝ))
        (fun x : ℝ => x ^ γ * Φ (Int.fract (Real.logb φ' x))) ∧
      ∀ᵐ t ∂MeasureTheory.volume, t ∈ Set.Ioo (0 : ℝ) 1 → DifferentiableAt ℝ Φ t
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 2 (arXiv:2308.12872v1, Section 1).
Uses 3

TheoremGrowth of fundamental sequences#●ready to prove

Let \(L\) and \(\tilde L\) be lists of natural numbers of length at least \(2\) with nonzero first entries, and let \(E\) and \(\tilde E\) be the periodic Zeckendorf collections they determine, with \(E\) a proper subset of \(\tilde E\). Let \(H\) and \(\tilde H\) be strictly increasing sequences of positive integers, indexed from \(0\), such that \(\mu \mapsto \sum_k \mu_k H_k\) is a bijection from \(E\) onto \(\mathbb N\) and \(\mu \mapsto \sum_k \mu_k \tilde H_k\) is a bijection from \(\tilde E\) onto \(\mathbb N\). Then there are reals \(\alpha > 0\), \(\tilde\alpha > 0\), \(\varphi\), \(\tilde\varphi\) and \(r\) with \(1 < \varphi < \tilde\varphi\) and \(0 < r < 1\) such that, as \(n \to \infty\), \(H_n - \alpha \varphi^n = O(\varphi^{nr})\), \(\tilde H_n - \tilde\alpha \tilde\varphi^n = O(\tilde\varphi^{nr})\) and, with \(\gamma = \log_{\tilde\varphi} \varphi\), \(H_n / \tilde H_n^{\gamma} \to \alpha / \tilde\alpha^{\gamma}\).

public theorem_wanted fundamentalSequence_div_rpow_tendsto (L L' : List ℕ) (hL : 2 ≤ L.length)
    (he : L.getD 0 0 ≠ 0) (hL' : 2 ≤ L'.length) (he' : L'.getD 0 0 ≠ 0)
    (hsub : periodicZeckendorfCollection L ⊂ periodicZeckendorfCollection L') {H H' : ℕ → ℕ}
    (hmono : StrictMono H) (hpos : ∀ k, 0 < H k)
    (hbij : Set.BijOn (zeckendorfEval H) (periodicZeckendorfCollection L) Set.univ)
    (hmono' : StrictMono H') (hpos' : ∀ k, 0 < H' k)
    (hbij' : Set.BijOn (zeckendorfEval H') (periodicZeckendorfCollection L') Set.univ) :
    ∃ α α' φ φ' r : ℝ, 0 < α ∧ 0 < α' ∧ 1 < φ ∧ φ < φ' ∧ 0 < r ∧ r < 1 ∧
      (fun n : ℕ => (H n : ℝ) - α * φ ^ n) =O[Filter.atTop] (fun n : ℕ => φ ^ ((n : ℝ) * r)) ∧
      (fun n : ℕ => (H' n : ℝ) - α' * φ' ^ n) =O[Filter.atTop]
        (fun n : ℕ => φ' ^ ((n : ℝ) * r)) ∧
      Filter.Tendsto (fun n : ℕ => (H n : ℝ) / (H' n : ℝ) ^ Real.logb φ' φ) Filter.atTop
        (nhds (α / α' ^ Real.logb φ' φ))
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 7 (arXiv:2308.12872v1, Section 3.1).
Uses 2

TheoremReduction lemma for counting a subcollection#●ready to prove

Let \(E \subseteq \tilde E\) be Zeckendorf collections, and let \(\tilde H\) be a strictly increasing sequence of positive integers such that \(\varepsilon \mapsto \sum_k \varepsilon_k \tilde H_k\) is a bijection from \(\tilde E\) onto \(\mathbb{N}\). For a natural number \(x\) let \(R(x)\) be the set of natural numbers \(n < x\) of the form \(n = \sum_k \varepsilon_k \tilde H_k\) with \(\varepsilon \in E\). Let \(\mu \in \tilde E\) be nonzero, suppose that \(\mu\) has an immediate predecessor \(\check\mu\) in \(E\) (the largest element of \(E\) that is \(\lt _a \mu\) in the ascending lexicographic order), and suppose that \(\check\mu\) has an immediate successor \(\bar\mu\) in \(E\) (the smallest element of \(E\) that is \(>_a \check\mu\)). Then \(R(\sum_k \mu_k \tilde H_k) = R(\sum_k \bar\mu_k \tilde H_k)\).

public theorem_wanted subcollectionLT_eval_eq_eval_bar {E E' : Set (ℕ →₀ ℕ)}
    (hE : IsZeckendorfCollection E) (hE' : IsZeckendorfCollection E') (hsub : E ⊆ E')
    {H' : ℕ → ℕ} (hmono' : StrictMono H') (hpos' : ∀ k, 0 < H' k)
    (hbij' : Set.BijOn (zeckendorfEval H') E' Set.univ) {μ μc μb : ℕ →₀ ℕ} (hμ : μ ∈ E')
    (hμ0 : μ ≠ 0) (hc : IsImmediatePredecessor E μ μc) (hb : IsImmediateSuccessor E μc μb) :
    {n : ℕ | n < zeckendorfEval H' μ ∧ ∃ ε ∈ E, zeckendorfEval H' ε = n} =
      {n : ℕ | n < zeckendorfEval H' μb ∧ ∃ ε ∈ E, zeckendorfEval H' ε = n}
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 7 (arXiv:2308.12872v1, Section 3.2, with the paper's definition of μ̌ and μ̄ there).
Uses 4

TheoremDuality formula#●ready to prove

Let \(E \subseteq \tilde E\) be Zeckendorf collections. Let \(H\) and \(\tilde H\) be strictly increasing sequences of positive integers such that \(\mu \mapsto \sum_k \mu_k H_k\) is a bijection from \(E\) onto \(\mathbb N\) and \(\mu \mapsto \sum_k \mu_k \tilde H_k\) is a bijection from \(\tilde E\) onto \(\mathbb N\). For real \(x\), let \(z(x)\) be the number of natural numbers \(n < x\) of the form \(n = \sum_k \nu_k \tilde H_k\) with \(\nu \in E\). Let \(\mu \in \tilde E\) be nonzero, let \(\check\mu\) be the immediate predecessor of \(\mu\) in \(E\) (the largest element of \(E\) below \(\mu\) in the ascending lexicographic order), and let \(\bar\mu\) be the immediate successor of \(\check\mu\) in \(E\). Then \(z\left(\sum_k \mu_k \tilde H_k\right) = \sum_k \bar\mu_k H_k\).

public theorem_wanted subcollectionCount_eval_eq_eval_bar {E E' : Set (ℕ →₀ ℕ)}
    (hE : IsZeckendorfCollection E) (hE' : IsZeckendorfCollection E') (hsub : E ⊆ E')
    {H H' : ℕ → ℕ} (hmono : StrictMono H) (hpos : ∀ k, 0 < H k)
    (hbij : Set.BijOn (zeckendorfEval H) E Set.univ) (hmono' : StrictMono H')
    (hpos' : ∀ k, 0 < H' k) (hbij' : Set.BijOn (zeckendorfEval H') E' Set.univ)
    {μ μc μb : ℕ →₀ ℕ} (hμ : μ ∈ E') (hμ0 : μ ≠ 0) (hc : IsImmediatePredecessor E μ μc)
    (hb : IsImmediateSuccessor E μc μb) :
    subcollectionCount E H' (zeckendorfEval H' μ) = zeckendorfEval H μb
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 8 (arXiv:2308.12872v1, Section 3.2).
Uses 5

DefinitionDescending lexicographic order#✓fully proved

For coefficient functions \(\varepsilon, \varepsilon' \colon \mathbb{N} \to \mathbb{N}\), \(\varepsilon \lt _d \varepsilon'\) means that there is an index \(k\) with \(\varepsilon_k < \varepsilon'_k\) and \(\varepsilon_j = \varepsilon'_j\) for all \(j < k\): at the smallest index where they differ, \(\varepsilon\) is smaller.

public def DescLexLT (ε ε' : ℕ → ℕ) : Prop :=
  ∃ k, ε k < ε' k ∧ ∀ j, j < k → ε j = ε' j
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.3).
Used by 7

DefinitionThe maximal \(L^*\)-block#✓fully proved

The maximal \(L^*\)-block. Let \(L = (e_1, \dots, e_N)\) be a list of natural numbers and \(s\) an index, with indices starting at \(0\). The maximal \(L^*\)-block starting at \(s\) is the coefficient function whose entry at index \(i\) is \(0\) for \(i < s\), and \(e_j\) for \(i \ge s\), where \(j - 1 = (i - s) \bmod N\) (every entry is \(0\) when \(L\) is empty). It is the paper's \(\bar\beta^{s+1}\).

public def maxLStarBlock (L : List ℕ) (s i : ℕ) : ℕ :=
  if i < s then 0 else L.getD ((i - s) % L.length) 0
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.3).
Used by 19

TheoremThe maximal block \(\bar\beta^2\) for \(L = (2, 3, 0)\)#✓fully proved

For \(L = (2, 3, 0)\), the maximal \(L^*\)-block starting at index \(1\) (indices start at \(0\); the paper's \(\bar\beta^2\)) has entries \(0, 2, 3, 0, 2\) at the indices \(0, 1, 2, 3, 4\).

public theorem maxLStarBlock_example :
    (List.range 5).map (maxLStarBlock [2, 3, 0] 1) = [0, 2, 3, 0, 2]
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.3, example).
Uses 1

DefinitionProper \(L^*\)-blocks#✓fully proved

Proper \(L^*\)-blocks. Let \(L\) be a list of natural numbers, index coefficient functions from \(1\), and let \(\bar\beta^s\) be the maximal \(L^*\)-block starting at index \(s\). For \(s \ge 1\), a function \(\zeta : \mathbb N \to \mathbb N\) is a proper \(L^*\)-block with support interval \([s, k]\) when \(s \le k\), \(\zeta_i = \bar\beta^s_i\) for all \(i < k\), \(\zeta_k < \bar\beta^s_k\), and \(\zeta_i = 0\) for all \(i > k\). Since \(\bar\beta^s\) vanishes below \(s\), so does \(\zeta\).

public def IsProperLStarBlock (L : List ℕ) (ζ : ℕ → ℕ) (s k : ℕ) : Prop :=
  s ≤ k ∧ (∀ i, i < k → ζ i = maxLStarBlock L s i) ∧ ζ k < maxLStarBlock L s k ∧
    ∀ i, k < i → ζ i = 0
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.3).
Uses 1 · Used by 3

DefinitionSums of disjoint proper \(L^\ast\)-blocks#✓fully proved

Let \(L\) be a list of natural numbers, and for each \(s\) let \(\bar\beta^{s+1}\) be the maximal \(L^*\)-block starting at (zero-based) index \(s\). Let \(I\) be a finite or infinite set of pairs \((s, k)\) of natural numbers. A coefficient function \(\varepsilon : \mathbb N \to \mathbb N\) is the sum of the disjoint proper \(L^*\)-blocks with support intervals \([s, k]\), \((s, k) \in I\), if \(s \le k\) for every \((s, k) \in I\), the intervals \([s, k]\) of distinct pairs in \(I\) are pairwise disjoint, for every \((s, k) \in I\) we have \(\varepsilon_i = \bar\beta^{s+1}_i\) whenever \(s \le i < k\) and \(\varepsilon_k < \bar\beta^{s+1}_k\), and \(\varepsilon_i = 0\) for every index \(i\) that lies in none of the intervals.

public def IsLStarBlockSum (L : List ℕ) (I : Set (ℕ × ℕ)) (ε : ℕ → ℕ) : Prop :=
  (∀ p ∈ I, p.1 ≤ p.2) ∧ I.PairwiseDisjoint (fun p => Set.Icc p.1 p.2) ∧
    (∀ p ∈ I, (∀ i, p.1 ≤ i → i < p.2 → ε i = maxLStarBlock L p.1 i) ∧
      ε p.2 < maxLStarBlock L p.1 p.2) ∧
    ∀ i, (∀ p ∈ I, i ∉ Set.Icc p.1 p.2) → ε i = 0
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.3).
Uses 1 · Used by 13

DefinitionThe periodic collection \(E^*\)#✓fully proved

The periodic collection \(E^*\). For a list \(L\) of natural numbers, \(E^*\) is the set of nonzero coefficient functions \(\varepsilon : \mathbb{N} \to \mathbb{N}\) that are sums of finitely or infinitely many disjoint proper \(L^*\)-blocks, that is, the \(\varepsilon \ne 0\) that are \(L^*\)-block sums over some set of support intervals. It is the collection \(E^*\) for the interval \((0, 1)\) determined by \(L\).

public def lStarCollection (L : List ℕ) : Set (ℕ → ℕ) :=
  {ε | ε ≠ 0 ∧ ∃ I, IsLStarBlockSum L I ε}
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.3).
Uses 1 · Used by 28
Used byMonotonicity of the ratio under steep increments · Strict local maxima of \(\delta^{\ast}\) at the closure points · Where \(\delta^*\) attains its minimum · Nearby expansions share long prefixes · Real Zeckendorf expansions · Infinitely many unit reductions stay in the \(L^\ast\)-closure · Classification of the local extrema of \(\delta^\ast\) · Local minima of \(\delta^\ast\) at finitely supported points · Fundamental sequence of an \(L^\ast\)-collection · Where \(\delta^\ast\) can attain its maximum · Decomposition along \(L^*\)-blocks (Corollary 1) · The closure \(\bar E^*\) of \(E^*\) · \(\bar\beta^1\) lies in \(\bar E^*\) · Fixed points of \(\overline{\mathrm{rev}}\) in \(\tilde E^*\) · The tail of an \(L^\ast\)-block decomposition lies in \(\tilde E^\ast\) · Membership in the set \(U\) · \(\delta^*\) has no local extremum at infinite expansions in \(E^*\) · \(\varepsilon\) and \(\inf(\varepsilon)\) have the same value · Shifted maximal blocks scale geometrically · Monotonicity of the ratio under fundamental increments (Corollary 3) · Generic upper bound for \(\delta^*\) (Lemma 17) · Shifting a unit one entry down increases \(\delta^*\) (Lemma 19) · The function \(\delta_1^\ast\) · Lemma 13: behaviour of \(\overline{\mathrm{rev}}\) near a point · Lemma 10: \(\mathrm{rev}_n\) of the bar of \(\mathrm{rev}_n(\varepsilon)\) · The set \(U\) of evaluations of \(\tilde E^* \setminus \bar E^*\) · Proposition 3: \(U\) is a disjoint union of open intervals · The zero function is not in the periodic Zeckendorf collection for \((0, 1)\)

TheoremThe zero function is not in the periodic Zeckendorf collection for \((0, 1)\)#✓fully proved

For every list \(L\) of natural numbers, the zero coefficient function \(\mathbb{N} \to \mathbb{N}\) does not belong to the periodic Zeckendorf collection \(E^*\) for the interval \((0, 1)\) determined by \(L\).

public theorem zero_notMem_lStarCollection (L : List ℕ) : (0 : ℕ → ℕ) ∉ lStarCollection L
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.3).
Uses 2

TheoremReal Zeckendorf expansions#●ready to prove

Let \(L = (e_1, \dots, e_N)\) be a list of natural numbers with \(N \ge 2\) and \(e_1 \ne 0\), and let \(E^*\) be its \(L^*\)-collection; indices of coefficient functions and sequences start at \(0\). There is a unique strictly decreasing sequence \((Q_m)_{m \ge 0}\) in \((0, 1)\) such that \(\varepsilon \mapsto \sum_k \varepsilon_k Q_k\) is a bijection from \(E^*\) onto \((0, 1)\). Moreover, every such sequence \(Q\) has the following properties: for \(\varepsilon, \delta \in E^*\), \(\varepsilon \lt _d \delta\) (that is, \(\varepsilon_k < \delta_k\) at the smallest index \(k\) where \(\varepsilon\) and \(\delta\) differ) if and only if \(\sum_k \varepsilon_k Q_k < \sum_k \delta_k Q_k\); for every \(m\), \(Q_m = \sum_k \bar\beta_k Q_k\), where \(\bar\beta\) is the maximal \(L^*\)-block starting at index \(m + 1\); and if \(\omega > 0\) satisfies \(-1 + \sum_{k=1}^{N-1} e_k \omega^k + (1 + e_N) \omega^N = 0\), then \(Q_m = \omega^{m+1}\) for every \(m\).

public theorem_wanted existsUnique_lStar_fundamentalSequence (L : List ℕ) (hL : 2 ≤ L.length)
    (he : L.getD 0 0 ≠ 0) :
    (∃! Q : ℕ → ℝ, StrictAnti Q ∧ (∀ n, 0 < Q n ∧ Q n < 1) ∧
      Set.BijOn (fun ε : ℕ → ℕ => ∑' k, (ε k : ℝ) * Q k) (lStarCollection L) (Set.Ioo 0 1)) ∧
    ∀ Q : ℕ → ℝ, StrictAnti Q → (∀ n, 0 < Q n ∧ Q n < 1) →
      Set.BijOn (fun ε : ℕ → ℕ => ∑' k, (ε k : ℝ) * Q k) (lStarCollection L) (Set.Ioo 0 1) →
      (∀ ε ∈ lStarCollection L, ∀ δ ∈ lStarCollection L,
        DescLexLT ε δ ↔ ∑' k, (ε k : ℝ) * Q k < ∑' k, (δ k : ℝ) * Q k) ∧
      (∀ m, Q m = ∑' k, (maxLStarBlock L (m + 1) k : ℝ) * Q k) ∧
      ∀ ω : ℝ, 0 < ω →
        -1 + ∑ k ∈ Finset.Ico 1 L.length, (L.getD (k - 1) 0 : ℝ) * ω ^ k +
          (1 + (L.getD (L.length - 1) 0 : ℝ)) * ω ^ L.length = 0 →
        ∀ m, Q m = ω ^ (m + 1)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 6 (arXiv:2308.12872v1, Section 2.3, quoting Chang 2021).
Uses 3

DefinitionThe coefficient function \(\inf(\varepsilon)\)#✓fully proved

Let \(L\) be a list of natural numbers and \(\varepsilon : \mathbb{N} \to \mathbb{N}\) a coefficient function, with indices starting at \(0\). If \(\varepsilon\) has a last nonzero entry, at index \(c\), then \(\inf(\varepsilon)\) agrees with \(\varepsilon\) at the indices \(i < c\), equals \(\varepsilon_c - 1\) at \(c\), and agrees with the maximal \(L^*\)-block starting at index \(c + 1\) at the indices \(i > c\). Otherwise, that is, when \(\varepsilon\) has infinite support or is zero, \(\inf(\varepsilon) = \varepsilon\).

In the paper's one-based indexing the last nonzero entry sits at index \(c + 1\), and \(\inf(\varepsilon) = \operatorname{res}_{c+1}(\varepsilon) - \beta^{c+1} + \bar\beta^{c+2}\), where \(\operatorname{res}_{c+1}(\varepsilon)\) keeps the entries of \(\varepsilon\) at the indices at most \(c + 1\), \(\beta^{c+1}\) has a single entry \(1\) at index \(c + 1\) and zeros elsewhere, and \(\bar\beta^{c+2}\) is the maximal \(L^*\)-block starting at index \(c + 2\) (all in paper indices).

public noncomputable def lStarInf (L : List ℕ) (ε : ℕ → ℕ) : ℕ → ℕ := by
  classical
  exact if h : ∃ c, ε c ≠ 0 ∧ ∀ i, c < i → ε i = 0 then
    fun i => if i < h.choose then ε i else if i = h.choose then ε i - 1
      else maxLStarBlock L (h.choose + 1) i
  else ε
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.3).
Uses 1 · Used by 4

Theorem\(\inf(\varepsilon) = \varepsilon\) for infinite support#✓fully proved

Let \(L\) be a list of natural numbers and \(\varepsilon : \mathbb{N} \to \mathbb{N}\) a coefficient function with infinitely many nonzero entries (for every \(c\) there is \(i > c\) with \(\varepsilon_i \ne 0\)). Then \(\inf(\varepsilon) = \varepsilon\), where \(\inf\) is taken with respect to \(L\).

public theorem lStarInf_of_infinite_support (L : List ℕ) {ε : ℕ → ℕ}
    (hε : ∀ c, ∃ i, c < i ∧ ε i ≠ 0) : lStarInf L ε = ε
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2 (arXiv:2308.12872v1, Section 2.3).
Uses 2

Theorem\(\varepsilon\) and \(\inf(\varepsilon)\) have the same value#●ready to prove

Let \(L\) be a list of natural numbers of length at least \(2\) with nonzero first entry, let \(E^*\) be the set of nonzero coefficient functions \(\mathbb N \to \mathbb N\) that are sums of finitely or infinitely many disjoint proper \(L^*\)-blocks, and let \(Q\) be a strictly decreasing sequence in \((0, 1)\) such that \(\varepsilon \mapsto \sum_k \varepsilon_k Q_k\) is a bijection from \(E^*\) onto \((0, 1)\); indices start at \(0\). For \(\varepsilon \in E^*\) define \(\inf(\varepsilon)\) as follows: if \(\varepsilon\) has infinitely many nonzero entries, then \(\inf(\varepsilon) = \varepsilon\); otherwise, with \(c\) the largest index such that \(\varepsilon_c \ne 0\), \(\inf(\varepsilon)\) agrees with \(\varepsilon\) at the indices below \(c\), has entry \(\varepsilon_c - 1\) at index \(c\), and agrees at the indices above \(c\) with the maximal \(L^*\)-block starting at index \(c + 1\). Then every \(\varepsilon \in E^*\) satisfies \(\sum_k \varepsilon_k Q_k = \sum_k \inf(\varepsilon)_k Q_k\).

public theorem_wanted tsum_lStarInf_mul_eq (L : List ℕ) (hL : 2 ≤ L.length)
    (he : L.getD 0 0 ≠ 0) {Q : ℕ → ℝ} (hQ : StrictAnti Q) (hQ01 : ∀ n, 0 < Q n ∧ Q n < 1)
    (hbij : Set.BijOn (fun ε : ℕ → ℕ => ∑' k, (ε k : ℝ) * Q k) (lStarCollection L)
      (Set.Ioo 0 1)) {ε : ℕ → ℕ} (hε : ε ∈ lStarCollection L) :
    ∑' k, (ε k : ℝ) * Q k = ∑' k, (lStarInf L ε k : ℝ) * Q k
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 4 (arXiv:2308.12872v1, Section 2.3).
Uses 2

TheoremNearby expansions share long prefixes#●ready to prove

Let \(L\) be a list of natural numbers of length at least \(2\) with nonzero first entry, let \(E^*\) be its \(L^*\)-collection, and let \(Q\) be a strictly decreasing sequence in \((0, 1)\) such that \(\varepsilon \mapsto \sum_k \varepsilon_k Q_k\) is a bijection from \(E^*\) onto \((0, 1)\); indices start at \(0\). Let \(\varepsilon \in E^*\), put \(x = \sum_k \varepsilon_k Q_k\), and let \(n \in \mathbb N\). Define \(\inf(\varepsilon)\) as follows: if \(\varepsilon\) has infinitely many nonzero entries, then \(\inf(\varepsilon) = \varepsilon\); otherwise, with \(c\) the largest index such that \(\varepsilon_c \ne 0\), \(\inf(\varepsilon)\) agrees with \(\varepsilon\) at indices below \(c\), has entry \(\varepsilon_c - 1\) at index \(c\), and agrees at indices above \(c\) with the maximal \(L^*\)-block starting at index \(c + 1\). Then there is \(\delta > 0\) such that every \(\varepsilon' \in E^*\) with \(x < \sum_k \varepsilon'_k Q_k < x + \delta\) satisfies \(\varepsilon'_i = \varepsilon_i\) for all \(i \le n\), and every \(\varepsilon' \in E^*\) with \(x - \delta < \sum_k \varepsilon'_k Q_k < x\) satisfies \(\varepsilon'_i = \inf(\varepsilon)_i\) for all \(i \le n\).

public theorem_wanted exists_prefix_eq_of_lStar_near (L : List ℕ) (hL : 2 ≤ L.length)
    (he : L.getD 0 0 ≠ 0) {Q : ℕ → ℝ} (hQ : StrictAnti Q) (hQ01 : ∀ n, 0 < Q n ∧ Q n < 1)
    (hbij : Set.BijOn (fun ε : ℕ → ℕ => ∑' k, (ε k : ℝ) * Q k) (lStarCollection L)
      (Set.Ioo 0 1)) {ε : ℕ → ℕ} (hε : ε ∈ lStarCollection L) (n : ℕ) :
    ∃ δ > 0,
      (∀ ε' ∈ lStarCollection L, ∑' k, (ε k : ℝ) * Q k < ∑' k, (ε' k : ℝ) * Q k →
        ∑' k, (ε' k : ℝ) * Q k < ∑' k, (ε k : ℝ) * Q k + δ → ∀ i ≤ n, ε' i = ε i) ∧
      (∀ ε' ∈ lStarCollection L, ∑' k, (ε k : ℝ) * Q k - δ < ∑' k, (ε' k : ℝ) * Q k →
        ∑' k, (ε' k : ℝ) * Q k < ∑' k, (ε k : ℝ) * Q k → ∀ i ≤ n, ε' i = lStarInf L ε i)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 5 (arXiv:2308.12872v1, Section 2.3).
Uses 2

TheoremProper blocks below a maximal block#●ready to prove

Let \(L = (e_1, \dots, e_N)\) with \(N \ge 2\) and \(e_1 \ne 0\), let \(\varepsilon \colon \mathbb{N} \to \mathbb{N}\) and \(b \in \mathbb{N}\) with \(\varepsilon_i = 0\) for all \(i < b\), and let \(\bar\beta\) be the maximal \(L^*\)-block starting at index \(b\). Then \(\varepsilon \lt _d \bar\beta\) in the descending lexicographic order if and only if there is \(k \ge b\) such that the function equal to \(\varepsilon_i\) for \(i \le k\) and to \(0\) for \(i > k\) is a proper \(L^*\)-block with support interval \([b, k]\).

public theorem_wanted descLexLT_maxLStarBlock_iff (L : List ℕ) (hL : 2 ≤ L.length)
    (he : L.getD 0 0 ≠ 0) {ε : ℕ → ℕ} {b : ℕ} (hε : ∀ i, i < b → ε i = 0) :
    DescLexLT ε (maxLStarBlock L b) ↔
      ∃ k, b ≤ k ∧ IsProperLStarBlock L (fun i => if i ≤ k then ε i else 0) b k
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 8 (arXiv:2308.12872v1, Section 3.3).
Uses 3

TheoremDecomposing \(L^*\)-blocks into \(\tilde L^*\)-blocks#●ready to prove

Decomposing \(L^*\)-blocks into \(\tilde L^*\)-blocks. Let \(L\) and \(\tilde L\) be lists of natural numbers of length at least \(2\) with nonzero first entries whose periodic Zeckendorf collections satisfy \(E \subsetneq \tilde E\). If \(\zeta\) is a nonzero proper \(L^*\)-block with support interval \([i, b]\), then \(\zeta\) is a sum of disjoint proper \(\tilde L^*\)-blocks whose support intervals all lie in \([0, b]\), one of which has support interval \([a, b]\) for some \(a\).

public theorem_wanted exists_lStarBlockSum_of_properLStarBlock (L L' : List ℕ)
    (hL : 2 ≤ L.length) (he : L.getD 0 0 ≠ 0) (hL' : 2 ≤ L'.length) (he' : L'.getD 0 0 ≠ 0)
    (hsub : periodicZeckendorfCollection L ⊂ periodicZeckendorfCollection L') {ζ : ℕ → ℕ}
    {i b : ℕ} (hζ : IsProperLStarBlock L ζ i b) (hζ0 : ζ ≠ 0) :
    ∃ I, IsLStarBlockSum L' I ζ ∧ (∀ p ∈ I, p.2 ≤ b) ∧ ∃ a, (a, b) ∈ I
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 9 (arXiv:2308.12872v1, Section 3.3).
Uses 3

TheoremThe tail of an \(L^\ast\)-block decomposition lies in \(\tilde E^\ast\)#●ready to prove

Let \(L\) and \(\tilde L\) be lists of natural numbers of length at least \(2\) with nonzero first entries, such that the periodic Zeckendorf collection \(E\) of \(L\) is a proper subset of the periodic Zeckendorf collection \(\tilde E\) of \(\tilde L\), and let \(\tilde E^*\) be the set of nonzero coefficient functions \(\mathbb N \to \mathbb N\) that are sums of finitely or infinitely many disjoint proper \(\tilde L^*\)-blocks. Indices start at \(0\). Let \(\varepsilon \in \tilde E^*\) and \(b \in \mathbb N\), let \(\eta\) agree with \(\varepsilon\) at the indices \(i < b\) and vanish at the indices \(i \ge b\), and let \(\mu\) vanish at the indices \(i < b\) and agree with \(\varepsilon\) at the indices \(i \ge b\). Suppose that for some set \(I\) of pairs \((s, k)\) with \(k < b\), \(\eta\) is the sum of the disjoint proper \(L^*\)-blocks with support intervals \([s, k]\), \((s, k) \in I\), and that \(\mu = \bar\beta\) or \(\bar\beta \lt _d \mu\), where \(\bar\beta\) is the maximal \(L^*\)-block starting at index \(b\) (the paper's \(\bar\beta^{b+1}\)) and \(\lt _d\) is the descending lexicographic order. Then \(\mu \in \tilde E^*\).

public theorem_wanted mem_lStarCollection_of_lStarBar_decomposition (L L' : List ℕ)
    (hL : 2 ≤ L.length) (he : L.getD 0 0 ≠ 0) (hL' : 2 ≤ L'.length) (he' : L'.getD 0 0 ≠ 0)
    (hsub : periodicZeckendorfCollection L ⊂ periodicZeckendorfCollection L') {ε : ℕ → ℕ}
    (hε : ε ∈ lStarCollection L') {b : ℕ} {I : Set (ℕ × ℕ)}
    (hI : IsLStarBlockSum L I (fun i => if i < b then ε i else 0)) (hIb : ∀ p ∈ I, p.2 < b)
    (hμ : maxLStarBlock L b = (fun i => if i < b then 0 else ε i) ∨
      DescLexLT (maxLStarBlock L b) (fun i => if i < b then 0 else ε i)) :
    (fun i => if i < b then 0 else ε i) ∈ lStarCollection L'
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Corollary 2 (arXiv:2308.12872v1, Section 3.3).
Uses 5

TheoremShifted maximal blocks scale geometrically#●ready to prove

Let \(L = (e_1, \dots, e_N)\) and \(\tilde L\) be lists of natural numbers of length at least \(2\) with nonzero first entries, and let \(\tilde Q\) be a strictly decreasing sequence with values in \((0, 1)\) such that \(\varepsilon \mapsto \sum_k \varepsilon_k \tilde Q_k\) is a bijection from \(\tilde E^*\), the periodic Zeckendorf collection for \((0, 1)\) determined by \(\tilde L\), onto \((0, 1)\). For \(s \in \mathbb{N}\) let \(\bar\beta_s\) be the maximal \(L^*\)-block starting at index \(s\): it is \(0\) at indices \(i < s\) and \(e_j\) at indices \(i \ge s\) with \(j - 1 \equiv i - s \pmod N\). Then \(\sum_k (\bar\beta_s)_k \tilde Q_k = \tilde Q_0^{\,s} \sum_k (\bar\beta_0)_k \tilde Q_k\) for every \(s \in \mathbb{N}\).

Indices are zero-based: \(\bar\beta_s\) is the paper's \(\bar\beta^{s+1}\) and \(\tilde Q_0\) is its \(\tilde Q_1 = \tilde\omega\).

public theorem_wanted tsum_maxLStarBlock_mul_eq (L L' : List ℕ) (hL : 2 ≤ L.length)
    (he : L.getD 0 0 ≠ 0) (hL' : 2 ≤ L'.length) (he' : L'.getD 0 0 ≠ 0) {Q' : ℕ → ℝ}
    (hQ' : StrictAnti Q') (hQ'01 : ∀ n, 0 < Q' n ∧ Q' n < 1)
    (hbij' : Set.BijOn (fun ε : ℕ → ℕ => ∑' k, (ε k : ℝ) * Q' k) (lStarCollection L')
      (Set.Ioo 0 1)) (s : ℕ) :
    ∑' k, (maxLStarBlock L s k : ℝ) * Q' k = Q' 0 ^ s * ∑' k, (maxLStarBlock L 0 k : ℝ) * Q' k
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 16 (arXiv:2308.12872v1, Section 4.1).
Uses 2

TheoremNumber of proper \(L^*\)-blocks with a given support interval#●ready to prove

Number of proper \(L^*\)-blocks with a given support interval. Let \(L = (e_1, \dots, e_N)\) be a list of natural numbers with \(N \ge 2\) and \(e_1 \ne 0\), let \(s \le k\) be natural numbers, and let \(j\) be the integer with \(1 \le j \le N\) and \(k - s + 1 \equiv j \pmod N\). Then there are exactly \(e_j\) proper \(L^*\)-blocks with support interval \([s, k]\).

public theorem_wanted ncard_properLStarBlock (L : List ℕ) (hL : 2 ≤ L.length)
    (he : L.getD 0 0 ≠ 0) {s k : ℕ} (hsk : s ≤ k) :
    {ζ : ℕ → ℕ | IsProperLStarBlock L ζ s k}.ncard = L.getD ((k - s) % L.length) 0
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 15 (arXiv:2308.12872v1, Section 4.1).
Uses 1
Statement usesProper \(L^*\)-blocks

DefinitionNumber of \(L^*\)-block sums of length \(n\)#✓fully proved

Number of \(L^*\)-block sums of length \(n\). For a list \(L\) of natural numbers and \(n \ge 0\), with coefficient functions indexed from \(1\), \(C_n\) is the number of functions \(\varepsilon : \mathbb N \to \mathbb N\) that are sums of disjoint proper \(L^*\)-blocks with support intervals inside \([1, n]\). The zero function (the empty sum) is counted, so \(C_0 = 1\).

public noncomputable def lStarBlockSumCount (L : List ℕ) (n : ℕ) : ℕ :=
  {ε : ℕ → ℕ | ∃ I, IsLStarBlockSum L I ε ∧ ∀ p ∈ I, p.2 < n}.ncard
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Proposition 4 (arXiv:2308.12872v1, Section 4.1).
Uses 1 · Used by 3

TheoremThe empty block-sum count#✓fully proved

For every list \(L\) of natural numbers, \(C_0 = 1\), where \(C_n\) is the number of sums of disjoint proper \(L^*\)-blocks with support intervals inside \([1, n]\): for \(n = 0\) the zero function is the only such sum.

public theorem lStarBlockSumCount_zero (L : List ℕ) : lStarBlockSumCount L 0 = 1
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Proposition 4 (arXiv:2308.12872v1, Section 4.1, S_0 = {0}).
Uses 3

TheoremRecurrence for the number of \(L^*\)-block sums#●ready to prove

Recurrence for the number of \(L^*\)-block sums. Let \(L = (e_1, \dots, e_N)\) be a list of natural numbers with \(N \ge 2\) and \(e_1 \ne 0\), index coefficient functions from \(1\), and let \(C_n\) be the number of functions \(\mathbb N \to \mathbb N\) that are sums of disjoint proper \(L^*\)-blocks with support intervals inside \([1, n]\). Then \(C_k = \sum_{j=1}^{k} e_j C_{k-j}\) for \(1 \le k \le N\), and \(C_n = \sum_{k=1}^{N-1} e_k C_{n-k} + (1 + e_N) C_{n-N}\) for \(n \ge N + 1\).

public theorem_wanted lStarBlockSumCount_recurrence (L : List ℕ) (hL : 2 ≤ L.length)
    (he : L.getD 0 0 ≠ 0) :
    (∀ k, 1 ≤ k → k ≤ L.length → lStarBlockSumCount L k =
      ∑ j ∈ Finset.Icc 1 k, L.getD (j - 1) 0 * lStarBlockSumCount L (k - j)) ∧
    ∀ n, L.length + 1 ≤ n → lStarBlockSumCount L n =
      ∑ k ∈ Finset.Ico 1 L.length, L.getD (k - 1) 0 * lStarBlockSumCount L (n - k) +
        (1 + L.getD (L.length - 1) 0) * lStarBlockSumCount L (n - L.length)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Proposition 4 (arXiv:2308.12872v1, Section 4.1).
Uses 1

TheoremGenerating function of the \(L^\ast\)-block-sum counts#●ready to prove

Let \(L = (e_1, \dots, e_N)\) be a list of natural numbers with \(N \ge 2\) and \(e_1 \ne 0\). For each \(k\) let \(C_k\) be the number of coefficient functions that are sums of disjoint proper \(L^*\)-blocks with support intervals inside \([1, k]\) (the zero function included), and put \(f(x) = 1 - \sum_{k=1}^{N-1} e_k x^k - (1 + e_N) x^N\). Then there is a real \(r > 0\) such that for every real \(x\) with \(-r < x < r\) the series \(\sum_{k \ge 0} C_k x^k\) converges absolutely, with sum \((1 - x^N) / f(x)\).

public theorem_wanted hasSum_lStarBlockSumCount (L : List ℕ) (hL : 2 ≤ L.length)
    (he : L.getD 0 0 ≠ 0) :
    ∃ r : ℝ, 0 < r ∧ ∀ x ∈ Set.Ioo (-r) r,
      HasSum (fun k => (lStarBlockSumCount L k : ℝ) * x ^ k)
        ((1 - x ^ L.length) / (1 - ∑ k ∈ Finset.Ico 1 L.length, (L.getD (k - 1) 0 : ℝ) * x ^ k -
          (1 + (L.getD (L.length - 1) 0 : ℝ)) * x ^ L.length))
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Proposition 5 (arXiv:2308.12872v1, Section 4.1).
Uses 1

StructurePeriodic Zeckendorf pairs#✓fully proved

Two lists \(L\) and \(\tilde L\) of natural numbers form a periodic Zeckendorf pair when both have length at least \(2\) and a nonzero first entry, and the periodic Zeckendorf collection \(E\) determined by \(L\) is a proper subset of the periodic Zeckendorf collection \(\tilde E\) determined by \(\tilde L\).

This is the standing assumption of Sections 3 and 4 of the source.

public structure IsPeriodicZeckendorfPair (L L' : List ℕ) : Prop where
  two_le_length : 2 ≤ L.length
  head_ne_zero : L.getD 0 0 ≠ 0
  two_le_length' : 2 ≤ L'.length
  head_ne_zero' : L'.getD 0 0 ≠ 0
  ssubset : periodicZeckendorfCollection L ⊂ periodicZeckendorfCollection L'
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 3.1.
Uses 1 · Used by 20

TheoremPeriodic Zeckendorf pairs are nested#✓fully proved

If the lists \(L\) and \(\tilde L\) form a periodic Zeckendorf pair, then the periodic Zeckendorf collection determined by \(L\) is contained in the periodic Zeckendorf collection determined by \(\tilde L\).

public theorem IsPeriodicZeckendorfPair.subset {L L' : List ℕ} (h : IsPeriodicZeckendorfPair L L') :
    periodicZeckendorfCollection L ⊆ periodicZeckendorfCollection L'
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 3.1.
Uses 2

DefinitionEvaluation of a coefficient function at a real sequence#✓fully proved

For a real sequence \(Q\) and a coefficient function \(\varepsilon \colon \mathbb{N} \to \mathbb{N}\), not necessarily finitely supported, the evaluation \(\sum \varepsilon Q\) is the sum of the series \(\sum_{k \ge 0} \varepsilon_k Q_k\), taken to be \(0\) when the series is not absolutely summable.

public noncomputable def zeckendorfRealEval (Q : ℕ → ℝ) (ε : ℕ → ℕ) : ℝ :=
  ∑' k, (ε k : ℝ) * Q k
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2.1.
Used by 19

TheoremEvaluating the zero coefficient function#✓fully proved

For every real sequence \(Q\), the evaluation of the zero coefficient function at \(Q\) is \(0\).

public theorem zeckendorfRealEval_zero (Q : ℕ → ℝ) : zeckendorfRealEval Q 0 = 0
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Section 2.1.
Uses 1

StructureFundamental sequence of an \(L^\ast\)-collection#✓fully proved

Let \(L\) be a list of natural numbers with \(L^*\)-collection \(E^*\). A real sequence \((Q_n)_{n \ge 0}\) is a fundamental sequence of \(E^*\) if it is strictly decreasing, \(0 < Q_n < 1\) for every \(n\), and the map \(\varepsilon \mapsto \sum_k \varepsilon_k Q_k\) is a bijection from \(E^*\) onto the open interval \((0, 1)\).

public structure IsLStarFundamentalSequence (L : List ℕ) (Q : ℕ → ℝ) : Prop where
  strictAnti : StrictAnti Q
  mem_Ioo : ∀ n, Q n ∈ Set.Ioo (0 : ℝ) 1
  bijOn : Set.BijOn (zeckendorfRealEval Q) (lStarCollection L) (Set.Ioo 0 1)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 6 and Definition 18.
Uses 2 · Used by 17

TheoremFundamental sequences of \(E^*\) are positive#✓fully proved

Fundamental sequences of \(E^*\) are positive. Let \(L\) be a list of natural numbers and let \(Q\) be a fundamental sequence of the collection \(E^*\) determined by \(L\). Then \(Q_n > 0\) for every \(n\).

public theorem IsLStarFundamentalSequence.pos {L : List ℕ} {Q : ℕ → ℝ}
    (h : IsLStarFundamentalSequence L Q) (n : ℕ) : 0 < Q n
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 6.
Uses 1

DefinitionReversal of the first \(n\) entries#✓fully proved

For \(n \in \mathbb N\) and a coefficient function \(\varepsilon : \mathbb N \to \mathbb N\), \(\mathrm{rev}_n(\varepsilon)\) is the finitely supported coefficient function whose entry at index \(j\) is \(\varepsilon(n - 1 - j)\) for \(j < n\) and \(0\) for \(j \ge n\) (indices start at \(0\)). In the paper's one-based indexing this is \(\mathrm{rev}_n(\varepsilon) = (\varepsilon_n, \dots, \varepsilon_1)\).

public noncomputable def zeckendorfRev (n : ℕ) (ε : ℕ → ℕ) : ℕ →₀ ℕ :=
  ∑ j ∈ Finset.range n, Finsupp.single j (ε (n - 1 - j))
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 5.
Used by 7

TheoremReversal of the first three entries#✓fully proved

For every coefficient function \(\varepsilon : \mathbb N \to \mathbb N\), \(\mathrm{rev}_3(\varepsilon)\) is the finitely supported coefficient function with entries \(\varepsilon(2)\), \(\varepsilon(1)\), \(\varepsilon(0)\) at indices \(0\), \(1\), \(2\) and \(0\) at every other index. In the paper's one-based indexing, \(\mathrm{rev}_3(\varepsilon) = (\varepsilon_3, \varepsilon_2, \varepsilon_1)\).

public theorem zeckendorfRev_three (ε : ℕ → ℕ) :
    zeckendorfRev 3 ε =
      Finsupp.single 0 (ε 2) + Finsupp.single 1 (ε 1) + Finsupp.single 2 (ε 0)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 5.
Uses 1

DefinitionThe coefficient function \(\bar\mu\)#✓fully proved

Let \(E\) be a set of finitely supported coefficient functions \(\mathbb{N} \to \mathbb{N}\) and \(\mu\) a finitely supported coefficient function. If \(\mu\) has an immediate predecessor \(\check\mu\) in \(E\) (the largest element of \(E\) that is \(\lt _a \mu\) in the ascending lexicographic order) and \(\check\mu\) has an immediate successor in \(E\) (the smallest element of \(E\) that is \(>_a \check\mu\)), then \(\bar\mu\) is that immediate successor; otherwise \(\bar\mu = 0\).

public noncomputable def zeckendorfBar (E : Set (ℕ →₀ ℕ)) (μ : ℕ →₀ ℕ) : ℕ →₀ ℕ := by
  classical
  exact if h : ∃ μb μc, IsImmediatePredecessor E μ μc ∧ IsImmediateSuccessor E μc μb then
    h.choose else 0
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 20.
Uses 2 · Used by 5

TheoremThe bar of a coefficient function is a successor of its predecessor#✓fully proved

Let \(E\) be a set of finitely supported coefficient functions \(\mathbb{N} \to \mathbb{N}\) and \(\mu\) a finitely supported coefficient function. Call \(\nu\) an immediate predecessor of \(\mu\) in \(E\) if it is a largest element of \(E\) that is \(\lt _a \mu\) in the ascending lexicographic order, and call \(\xi\) an immediate successor of \(\nu\) in \(E\) if it is a smallest element of \(E\) that is \(>_a \nu\). Let \(\bar\mu\) be an immediate successor in \(E\) of an immediate predecessor of \(\mu\) in \(E\) if such a pair exists, and \(\bar\mu = 0\) otherwise. If some immediate predecessor of \(\mu\) in \(E\) has an immediate successor in \(E\), then \(\bar\mu\) is the immediate successor in \(E\) of some immediate predecessor of \(\mu\) in \(E\).

public theorem zeckendorfBar_spec {E : Set (ℕ →₀ ℕ)} {μ : ℕ →₀ ℕ}
    (h : ∃ μb μc, IsImmediatePredecessor E μ μc ∧ IsImmediateSuccessor E μc μb) :
    ∃ μc, IsImmediatePredecessor E μ μc ∧ IsImmediateSuccessor E μc (zeckendorfBar E μ)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 20.
Uses 3

DefinitionThe closure \(\bar E^*\) of \(E^*\)#✓fully proved

The closure \(\bar E^*\) of \(E^*\). Let \(L\) be a list of natural numbers, \(E^*\) the collection determined by \(L\), and index coefficient functions from \(1\). The collection \(\bar E^*\) consists of the elements of \(E^*\) together with every function \(\varepsilon : \mathbb N \to \mathbb N\) for which there is an index \(b \ge 0\) such that the function agreeing with \(\varepsilon\) at the indices \(\le b\) and vanishing elsewhere is a sum of disjoint proper \(L^*\)-blocks with support intervals inside \([1, b]\), and \(\varepsilon_i = \bar\beta^{b+1}_i\) for all \(i \ge b + 1\), where \(\bar\beta^{b+1}\) is the maximal \(L^*\)-block starting at index \(b + 1\). These extra functions have the form \(\sum_m \zeta^m + \bar\beta^{b+1}\) with proper \(L^*\)-blocks \(\zeta^m\); the case \(b = 0\) gives \(\bar\beta^1\).

public def lStarClosure (L : List ℕ) : Set (ℕ → ℕ) :=
  lStarCollection L ∪ {ε | ∃ b I, IsLStarBlockSum L I (fun i => if i < b then ε i else 0) ∧
    (∀ p ∈ I, p.2 < b) ∧ ∀ i, b ≤ i → ε i = maxLStarBlock L b i}
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 22.
Uses 3 · Used by 11

Theorem\(\bar\beta^1\) lies in \(\bar E^*\)#✓fully proved

\(\bar\beta^1\) lies in \(\bar E^*\). For every list \(L\) of natural numbers, the maximal \(L^*\)-block starting at index \(0\) (the paper's \(\bar\beta^1\)) belongs to the collection \(\bar E^*\) determined by \(L\). This is the case of the definition of \(\bar E^*\) with no proper blocks, in which the maximal block starts at index \(0\).

public theorem maxLStarBlock_zero_mem_lStarClosure (L : List ℕ) :
    maxLStarBlock L 0 ∈ lStarClosure L
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 22.
Uses 4

DefinitionThe map \(\overline{\mathrm{rev}}\)#✓fully proved

Fix a list \(L\), and let \(E\) be its periodic Zeckendorf collection, \(\bar\beta^1\) the maximal \(L^*\)-block starting at index \(0\), \(\lt _d\) the descending lexicographic order, \(\mathrm{rev}_n\) the reversal of the first \(n\) entries, and \(\bar\mu\) the bar of a finitely supported \(\mu\) relative to \(E\). For a coefficient function \(\varepsilon : \mathbb N \to \mathbb N\), if \(\varepsilon \lt _d \bar\beta^1\) fails then \(\overline{\mathrm{rev}}(\varepsilon) = \bar\beta^1\). If \(\varepsilon \lt _d \bar\beta^1\), then for each index \(i\) the entry of \(\overline{\mathrm{rev}}(\varepsilon)\) at \(i\) is the eventual value, as \(n \to \infty\), of the entry of \(\mathrm{rev}_n(\overline{\mathrm{rev}_n(\varepsilon)})\) at \(i\) when that entry is eventually constant, and \(0\) when it is not.

public noncomputable def zeckendorfRevBar (L : List ℕ) (ε : ℕ → ℕ) : ℕ → ℕ := by
  classical
  exact if DescLexLT ε (maxLStarBlock L 0) then
    fun i => if h : ∃ v, ∀ᶠ n in Filter.atTop,
        zeckendorfRev n (zeckendorfBar (periodicZeckendorfCollection L) (zeckendorfRev n ε)) i = v
      then h.choose else 0
  else maxLStarBlock L 0
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 23.
Uses 5 · Used by 8

Theorem\(\overline{\mathrm{rev}}\) of a function not below \(\bar\beta^1\)#✓fully proved

Let \(L\) be a list of natural numbers, let \(\bar\beta^1\) be the maximal \(L^*\)-block starting at index \(0\), let \(\lt _d\) be the descending lexicographic order, and let \(\overline{\mathrm{rev}}\) be the map determined by \(L\). For every coefficient function \(\varepsilon : \mathbb N \to \mathbb N\) such that \(\varepsilon \lt _d \bar\beta^1\) fails, \(\overline{\mathrm{rev}}(\varepsilon) = \bar\beta^1\).

public theorem zeckendorfRevBar_of_not_descLexLT (L : List ℕ) {ε : ℕ → ℕ}
    (h : ¬ DescLexLT ε (maxLStarBlock L 0)) : zeckendorfRevBar L ε = maxLStarBlock L 0
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 23.
Uses 6

DefinitionThe function \(\delta_1^\ast\)#✓fully proved

Let \(L\) and \(\tilde L\) be lists of natural numbers and \(Q\) and \(\tilde Q\) real sequences with first terms \(\omega = Q_0\) and \(\tilde\omega = \tilde Q_0\) (indices start at \(0\)), and put \(\gamma = \log_{\tilde\omega} \omega\). Let \(\tilde E^*\) be the collection \(E^*\) determined by \(\tilde L\), let \(\overline{\mathrm{rev}}\) be the map determined by \(L\), and write \(\sum \varepsilon R = \sum_k \varepsilon_k R_k\) for the evaluation of a coefficient function \(\varepsilon\) at a real sequence \(R\). For real \(y\): if some \(\varepsilon \in \tilde E^*\) satisfies \(\sum \varepsilon \tilde Q = y\), then \(\delta_1^*(y) = \sum \overline{\mathrm{rev}}(\varepsilon) Q / y^\gamma\) for a chosen such \(\varepsilon\); otherwise \(\delta_1^*(y) = 0\).

In the paper \(\delta_1^*\) is used on \([\tilde\omega, 1)\).

public noncomputable def zeckendorfDeltaStarOne (L L' : List ℕ) (Q Q' : ℕ → ℝ) (y : ℝ) : ℝ := by
  classical
  exact if h : ∃ ε ∈ lStarCollection L', zeckendorfRealEval Q' ε = y then
    zeckendorfRealEval Q (zeckendorfRevBar L h.choose) / y ^ Real.logb (Q' 0) (Q 0) else 0
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 23.
Uses 3 · Used by 2

DefinitionThe function \(\delta^*\)#✓fully proved

The function \(\delta^*\). Let \(L\) and \(\tilde L\) be lists and \(Q\) and \(\tilde Q\) real sequences, let \(\tilde\omega\) be the first term of \(\tilde Q\), and let \(\delta_1^*\) be the function \(\delta_1^*\) of \(L\), \(\tilde L\), \(Q\), \(\tilde Q\). For real \(y\): if some natural number \(m\) satisfies \(\tilde\omega^{m+1} \le y < \tilde\omega^m\), then \(\delta^*(y) = \delta_1^*(y / \tilde\omega^m)\) for a chosen such \(m\); otherwise \(\delta^*(y) = 0\).

When \(0 < \tilde\omega < 1\), the intervals \([\tilde\omega^{m+1}, \tilde\omega^m)\) partition \((0, 1)\), so \(m\) is unique and \(\delta^*\) vanishes outside \((0, 1)\). The printed Definition 23 reduces \(y\) by powers of the first term \(\omega\) of \(Q\); since \(\delta_1^*\) lives on \([\tilde\omega, 1)\), Proposition 2 and Equation (12) evaluate \(\delta^*\) on \([\tilde\omega, 1)\) as \(\delta_1^*\), and the proof of Theorem 9 glues \(\delta_1^*(\tilde\omega) = \lim_{x \to 1^-} \delta_1^*(x)\), the reduction is by powers of \(\tilde\omega\) (Notation 1).

public noncomputable def zeckendorfDeltaStar (L L' : List ℕ) (Q Q' : ℕ → ℝ) (y : ℝ) : ℝ := by
  classical
  exact if h : ∃ m : ℕ, Q' 0 ^ (m + 1) ≤ y ∧ y < Q' 0 ^ m then
    zeckendorfDeltaStarOne L L' Q Q' (y / Q' 0 ^ h.choose) else 0
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 23.
Uses 1 · Used by 13

Theorem\(\delta^*\) agrees with \(\delta_1^*\) on \((\tilde\omega, 1)\)#✓fully proved

\(\delta^*\) agrees with \(\delta_1^*\) on \([\tilde\omega, 1)\). Let \(L\) and \(\tilde L\) be lists and \(Q\) and \(\tilde Q\) real sequences such that the first term \(\tilde\omega\) of \(\tilde Q\) satisfies \(0 < \tilde\omega < 1\). Then for every real \(y\) with \(\tilde\omega \le y < 1\), the functions \(\delta^*\) and \(\delta_1^*\) of \(L\), \(\tilde L\), \(Q\), \(\tilde Q\) agree at \(y\): \(\delta^*(y) = \delta_1^*(y)\).

public theorem zeckendorfDeltaStar_of_mem_Ico {L L' : List ℕ} {Q Q' : ℕ → ℝ} (h0 : 0 < Q' 0)
    (h1 : Q' 0 < 1) {y : ℝ} (hy : Q' 0 ≤ y) (hy1 : y < 1) :
    zeckendorfDeltaStar L L' Q Q' y = zeckendorfDeltaStarOne L L' Q Q' y
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 23.
Uses 2

DefinitionThe generic exponent#✓fully proved

Let \(Q\) and \(\tilde Q\) be real sequences with first terms \(\omega\) and \(\tilde\omega\), and put \(\gamma = \log_{\tilde\omega} \omega\). The generic exponent \(p\) of \(Q\) and \(\tilde Q\) is the smallest positive integer \(p\) with \(\gamma \tilde\omega^{p-1} < \omega^p\), and \(p = 0\) if no positive integer satisfies this inequality.

In the paper, \(p\) is the exponent of the generic upper bound of the function \(\delta^*\) (Definition 25).

public noncomputable def zeckendorfGenericExponent (Q Q' : ℕ → ℝ) : ℕ :=
  sInf {p : ℕ | 0 < p ∧ Real.logb (Q' 0) (Q 0) * Q' 0 ^ (p - 1) < Q 0 ^ p}
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 25.
Used by 7

TheoremThe generic exponent satisfies its defining inequality#✓fully proved

Let \(Q\) and \(\tilde Q\) be real sequences with first terms \(\omega\) and \(\tilde\omega\), put \(\gamma = \log_{\tilde\omega} \omega\), and let \(p\) be the generic exponent of \(Q\) and \(\tilde Q\). If some positive integer \(q\) satisfies \(\gamma \tilde\omega^{q-1} < \omega^q\), then \(p\) is positive and \(\gamma \tilde\omega^{p-1} < \omega^p\).

public theorem zeckendorfGenericExponent_spec {Q Q' : ℕ → ℝ}
    (h : ∃ p : ℕ, 0 < p ∧ Real.logb (Q' 0) (Q 0) * Q' 0 ^ (p - 1) < Q 0 ^ p) :
    0 < zeckendorfGenericExponent Q Q' ∧
      Real.logb (Q' 0) (Q 0) * Q' 0 ^ (zeckendorfGenericExponent Q Q' - 1) <
        Q 0 ^ zeckendorfGenericExponent Q Q'
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Definition 25.
Uses 1
Statement usesThe generic exponent

DefinitionThe set \(U\) of evaluations of \(\tilde E^* \setminus \bar E^*\)#✓fully proved

The set \(U\) of evaluations of \(\tilde E^* \setminus \bar E^*\). For lists \(L\) and \(\tilde L\) and a real sequence \(\tilde Q\), let \(\tilde E^*\) be the \(\tilde L^*\)-collection and \(\bar E^*\) the closure collection of \(L\). The set \(U \subseteq \mathbb R\) is the image of \(\tilde E^* \setminus \bar E^*\) under the evaluation \(\varepsilon \mapsto \sum_k \varepsilon_k \tilde Q_k\).

public noncomputable def zeckendorfSmoothSet (L L' : List ℕ) (Q' : ℕ → ℝ) : Set ℝ :=
  zeckendorfRealEval Q' '' (lStarCollection L' \ lStarClosure L)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Notation 2.
Uses 3 · Used by 4

TheoremMembership in the set \(U\)#✓fully proved

Let \(L\) and \(\tilde L\) be lists of natural numbers and let \(\tilde Q = (\tilde Q_k)\) be a sequence of real numbers. Let \(\tilde E^*\) be the collection \(E^*\) determined by \(\tilde L\), let \(\bar E^*\) be the closure \(\bar E^*\) determined by \(L\), and let \(U\) be the image of \(\tilde E^* \setminus \bar E^*\) under the map \(\varepsilon \mapsto \sum_k \varepsilon_k \tilde Q_k\). A real number \(x\) lies in \(U\) if and only if \(x = \sum_k \varepsilon_k \tilde Q_k\) for some \(\varepsilon \in \tilde E^*\) with \(\varepsilon \notin \bar E^*\).

public theorem mem_zeckendorfSmoothSet_iff {L L' : List ℕ} {Q' : ℕ → ℝ} {x : ℝ} :
    x ∈ zeckendorfSmoothSet L L' Q' ↔
      ∃ ε ∈ lStarCollection L', ε ∉ lStarClosure L ∧ zeckendorfRealEval Q' ε = x
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Notation 2.
Uses 4

TheoremReversals characterize the nonzero elements of \(\bar E^\ast\)#●ready to prove

Let \(L\) be a list of natural numbers of length at least \(2\) with nonzero first entry, let \(E\) be the periodic Zeckendorf collection determined by \(L\), and let \(\bar E^*\) be the closure of the collection \(E^*\) determined by \(L\). For \(n \in \mathbb N\) and a coefficient function \(\varepsilon : \mathbb N \to \mathbb N\), let \(\mathrm{rev}_n(\varepsilon)\) be the finitely supported coefficient function whose first \(n\) entries are the first \(n\) entries of \(\varepsilon\) in reverse order and whose other entries are \(0\). Then a nonzero coefficient function \(\varepsilon : \mathbb N \to \mathbb N\) satisfies \(\mathrm{rev}_n(\varepsilon) \in E\) for every \(n \in \mathbb N\) if and only if \(\varepsilon \in \bar E^*\).

This reads the source's Lemma 3 for nonzero \(\varepsilon\) and counts \(\bar\beta^1\), the maximal \(L^*\)-block starting at the first index, as an element of \(\bar E^*\). As printed (no \(\varepsilon \ne 0\), decompositions with at least one block) the lemma fails at \(\varepsilon = 0\), since \(\mathrm{rev}_n(0) = 0 \in E\), and at \(\varepsilon = \bar\beta^1\), since \(\mathrm{rev}_n(\bar\beta^1)\) is the maximal \(L\)-block \(\theta^{n+1}\), which lies in \(E\).

public theorem_wanted forall_zeckendorfRev_mem_iff_mem_lStarClosure (L : List ℕ)
    (hL : 2 ≤ L.length) (he : L.getD 0 0 ≠ 0) {ε : ℕ → ℕ} (hε : ε ≠ 0) :
    (∀ n, zeckendorfRev n ε ∈ periodicZeckendorfCollection L) ↔ ε ∈ lStarClosure L
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 3.
Uses 3

TheoremDecomposition along \(L^*\)-blocks (Corollary 1)#●ready to prove

Decomposition along \(L^*\)-blocks (Corollary 1). Let \(L\) and \(\tilde L\) be a periodic Zeckendorf pair, let \(E^*\) and \(\tilde E^*\) be the collections determined by \(L\) and \(\tilde L\), and let \(\varepsilon \in \tilde E^*\). Index coefficient functions from \(1\). For \(b \ge 0\), let \(\eta_b\) agree with \(\varepsilon\) at the indices \(\le b\) and vanish elsewhere, and let \(\mu_b = \varepsilon - \eta_b\). Call \(b\) a decomposition index of \(\varepsilon\) if \(\eta_b\) is a sum of disjoint proper \(L^*\)-blocks with support intervals inside \([1, b]\) and \(\bar\beta^{b+1} \le_d \mu_b\), where \(\bar\beta^{b+1}\) is the maximal \(L^*\)-block starting at index \(b + 1\) and \(\le_d\) means equal or smaller in the descending lexicographic order. Then \(\varepsilon \in E^*\) if and only if \(\varepsilon\) has no decomposition index, and if \(\varepsilon \notin E^*\), then \(\varepsilon\) has exactly one decomposition index.

public theorem_wanted lStarBar_decomposition {L L' : List ℕ} (hp : IsPeriodicZeckendorfPair L L')
    {ε : ℕ → ℕ} (hε : ε ∈ lStarCollection L') :
    (ε ∈ lStarCollection L ↔ ¬ ∃ b I, IsLStarBlockSum L I (fun i => if i < b then ε i else 0) ∧
        (∀ p ∈ I, p.2 < b) ∧ (maxLStarBlock L b = (fun i => if i < b then 0 else ε i) ∨
          DescLexLT (maxLStarBlock L b) (fun i => if i < b then 0 else ε i))) ∧
    (ε ∉ lStarCollection L → ∃! b, ∃ I, IsLStarBlockSum L I (fun i => if i < b then ε i else 0) ∧
        (∀ p ∈ I, p.2 < b) ∧ (maxLStarBlock L b = (fun i => if i < b then 0 else ε i) ∨
          DescLexLT (maxLStarBlock L b) (fun i => if i < b then 0 else ε i)))
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Corollary 1.
Uses 5

TheoremLemma 10: \(\mathrm{rev}_n\) of the bar of \(\mathrm{rev}_n(\varepsilon)\)#●ready to prove

Let \((L, \tilde L)\) be a periodic Zeckendorf pair, \(E\) the periodic Zeckendorf collection of \(L\), and \(E^*\) and \(\tilde E^*\) the \(L^*\)- and \(\tilde L^*\)-collections. Indices start at \(0\). Write \(\lt _d\) for the descending lexicographic order, \(\bar\beta^{s+1}\) for the maximal \(L^*\)-block starting at index \(s\), \(\mathrm{rev}_n\) for the reversal of the first \(n\) entries, \(\bar\mu\) for the bar of a finitely supported \(\mu\) relative to \(E\), and \(\mathrm{res}_n(\varepsilon)\) for \(\varepsilon\) with every entry at an index \(\ge n\) replaced by \(0\). Let \(\varepsilon \in \tilde E^*\) with \(\varepsilon \lt _d \bar\beta^1\). If \(\varepsilon \in E^*\), then \(\mathrm{rev}_n(\overline{\mathrm{rev}_n(\varepsilon)}) = \mathrm{res}_n(\varepsilon)\) for every \(n \ge 1\). Moreover, let \(b \ge 1\) be such that \(\mathrm{res}_b(\varepsilon)\) is a sum of disjoint proper \(L^*\)-blocks whose support intervals consist of indices below \(b\), and such that the coefficient function \(\mu\) obtained from \(\varepsilon\) by replacing every entry at an index below \(b\) by \(0\) satisfies \(\mu = \bar\beta^{b+1}\) or \(\bar\beta^{b+1} \lt _d \mu\). Then for all sufficiently large \(n\), \(\mathrm{rev}_n(\overline{\mathrm{rev}_n(\varepsilon)})\) equals \(\mathrm{res}_b(\varepsilon) + \beta^b\) if \(\bar\beta^{b+1} \lt _d \mu\), where \(\beta^b\) has entry \(1\) at index \(b - 1\) and \(0\) elsewhere, and equals \(\mathrm{res}_n(\varepsilon)\) otherwise.

MetaMathlibExt.zeckendorfRev_bar_zeckendorfRev_eq · 15 lines
public theorem_wanted zeckendorfRev_bar_zeckendorfRev_eq {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {ε : ℕ → ℕ} (hε : ε ∈ lStarCollection L')
    (hlt : DescLexLT ε (maxLStarBlock L 0)) :
    (ε ∈ lStarCollection L → ∀ n, 1 ≤ n →
      ⇑(zeckendorfRev n (zeckendorfBar (periodicZeckendorfCollection L) (zeckendorfRev n ε))) =
        fun i => if i < n then ε i else 0) ∧
    ∀ b I, 1 ≤ b → IsLStarBlockSum L I (fun i => if i < b then ε i else 0) →
      (∀ p ∈ I, p.2 < b) →
      (maxLStarBlock L b = (fun i => if i < b then 0 else ε i) ∨
        DescLexLT (maxLStarBlock L b) (fun i => if i < b then 0 else ε i)) →
      ∀ᶠ n in Filter.atTop,
        ⇑(zeckendorfRev n (zeckendorfBar (periodicZeckendorfCollection L) (zeckendorfRev n ε))) =
          if DescLexLT (maxLStarBlock L b) (fun i => if i < b then 0 else ε i) then
            (fun i => if i < b then ε i else 0) + ⇑(Finsupp.single (b - 1) 1)
          else fun i => if i < n then ε i else 0
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 10.
Uses 8

TheoremFixed points of \(\overline{\mathrm{rev}}\) in \(\tilde E^*\)#●ready to prove

Fixed points of \(\overline{\mathrm{rev}}\) in \(\tilde E^*\). Let \(L\) and \(\tilde L\) be lists of natural numbers of length at least \(2\) with nonzero first entries, such that the periodic Zeckendorf collection \(E\) of \(L\) is a proper subset of the periodic Zeckendorf collection \(\tilde E\) of \(\tilde L\). Let \(\tilde E^*\) be the collection \(E^*\) determined by \(\tilde L\), and let \(\bar E^*\) and \(\overline{\mathrm{rev}}\) be taken with respect to \(L\). Then for every \(\varepsilon \in \tilde E^*\), \(\varepsilon \in \bar E^*\) if and only if \(\overline{\mathrm{rev}}(\varepsilon) = \varepsilon\).

public theorem_wanted mem_lStarClosure_iff_zeckendorfRevBar_eq {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {ε : ℕ → ℕ} (hε : ε ∈ lStarCollection L') :
    ε ∈ lStarClosure L ↔ zeckendorfRevBar L ε = ε
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 11.
Uses 4

Theorem\(\overline{\mathrm{rev}}\) of the reversal of an element of \(\tilde E\) (Lemma 12)#●ready to prove

Let \(L\) and \(\tilde L\) form a periodic Zeckendorf pair, let \(E\) and \(\tilde E\) be the periodic Zeckendorf collections determined by \(L\) and \(\tilde L\), and index coefficient functions from \(0\). Let \(\mu \in \tilde E\) and \(t \ge 1\) be such that \(\mu_{t-1} \ne 0\) and \(\mu_j = 0\) for all \(j \ge t\). Let \(\overline{\mathrm{rev}}\) be the map determined by \(L\), \(\bar\beta^1\) the maximal \(L^*\)-block starting at index \(0\), \(\mathrm{rev}_t\) the reversal of the first \(t\) entries, \(\bar\mu\) the bar of \(\mu\) relative to \(E\), \(\lt _a\) the ascending lexicographic order, and \(\theta^{t+1}\) the maximal \(L\)-block that vanishes from index \(t\) on. If \(\mu = \theta^{t+1}\) or \(\mu \lt _a \theta^{t+1}\), then \(\overline{\mathrm{rev}}(\mathrm{rev}_t(\mu)) = \mathrm{rev}_t(\bar\mu)\); if \(\theta^{t+1} \lt _a \mu\), then \(\overline{\mathrm{rev}}(\mathrm{rev}_t(\mu)) = \bar\beta^1\).

In the paper \(t\) is the order of \(\mu\), and \(\theta^{t+1}\) appears as the immediate predecessor \(\hat\beta^{t+1}\) in \(E\) of the basis function \(\beta^{t+1}\), whose only nonzero entry is a \(1\) at index \(t\).

public theorem_wanted zeckendorfRevBar_zeckendorfRev_of_order {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {μ : ℕ →₀ ℕ} (hμ : μ ∈ periodicZeckendorfCollection L')
    {t : ℕ} (ht : 1 ≤ t) (hord : μ (t - 1) ≠ 0 ∧ ∀ j, t ≤ j → μ j = 0) :
    ((μ = periodicTheta L t ∨ AscLexLT μ (periodicTheta L t)) →
      zeckendorfRevBar L (zeckendorfRev t μ) =
        ⇑(zeckendorfRev t (zeckendorfBar (periodicZeckendorfCollection L) μ))) ∧
    (AscLexLT (periodicTheta L t) μ → zeckendorfRevBar L (zeckendorfRev t μ) = maxLStarBlock L 0)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 12.
Uses 8

TheoremTransition from the subcollection count to \(\delta^\ast\)#●ready to prove

Let \(L\) and \(\tilde L\) form a periodic Zeckendorf pair, with periodic Zeckendorf collections \(E \subsetneq \tilde E\), and let \(Q\) and \(\tilde Q\) be fundamental sequences of the \(L^*\)-collection \(E^*\) and the \(\tilde L^*\)-collection \(\tilde E^*\). Put \(\omega = Q_0\), \(\tilde\omega = \tilde Q_0\) and \(\gamma = \log_{\tilde\omega} \omega\), and let \(\delta^*\) be the function determined by \(L\), \(\tilde L\), \(Q\) and \(\tilde Q\). Let \(H\) and \(\tilde H\) be strictly increasing sequences of positive integers such that \(\mu \mapsto \sum_k \mu_k H_k\) is a bijection from \(E\) onto \(\mathbb N\) and \(\mu \mapsto \sum_k \mu_k \tilde H_k\) is a bijection from \(\tilde E\) onto \(\mathbb N\), and suppose \(H_n \omega^n \to \alpha\) and \(\tilde H_n \tilde\omega^n \to \tilde\alpha\) as \(n \to \infty\). For each natural number \(x\) let \(\mu_x \in \tilde E\) satisfy \(\sum_k (\mu_x)_k \tilde H_k = x\), let \(n_x\) be one more than the largest index in the support of \(\mu_x\) (\(n_x = 0\) if \(\mu_x = 0\)), and let \(z(x)\) be the number of natural numbers \(n < x\) of the form \(\sum_k \mu_k \tilde H_k\) with \(\mu \in E\). Then, as \(x \to \infty\) through the natural numbers, \(z(x) / x^\gamma\) is asymptotically equivalent to \(\delta^*(y_x) \, \alpha / \tilde\alpha^\gamma\), where \(y_x = \sum_k \rho_k \tilde Q_k\) and \(\rho = \operatorname{rev}_{n_x}(\mu_x)\) lists the first \(n_x\) entries of \(\mu_x\) in reverse order.

Indices are zero-based.

MetaMathlibExt.isEquivalent_subcollectionCount_div_rpow_deltaStar · 16 lines
public theorem_wanted isEquivalent_subcollectionCount_div_rpow_deltaStar {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q Q' : ℕ → ℝ} (hQ : IsLStarFundamentalSequence L Q)
    (hQ' : IsLStarFundamentalSequence L' Q') {H H' : ℕ → ℕ} (hmono : StrictMono H)
    (hpos : ∀ k, 0 < H k) (hbij : Set.BijOn (zeckendorfEval H) (periodicZeckendorfCollection L) Set.univ)
    (hmono' : StrictMono H') (hpos' : ∀ k, 0 < H' k)
    (hbij' : Set.BijOn (zeckendorfEval H') (periodicZeckendorfCollection L') Set.univ) {α α' : ℝ}
    (hα : Filter.Tendsto (fun n => (H n : ℝ) * Q 0 ^ n) Filter.atTop (nhds α))
    (hα' : Filter.Tendsto (fun n => (H' n : ℝ) * Q' 0 ^ n) Filter.atTop (nhds α'))
    {μ : ℕ → ℕ →₀ ℕ}
    (hμ : ∀ x, μ x ∈ periodicZeckendorfCollection L' ∧ zeckendorfEval H' (μ x) = x) :
    Asymptotics.IsEquivalent Filter.atTop
      (fun x : ℕ => (subcollectionCount (periodicZeckendorfCollection L) H' x : ℝ) /
        (x : ℝ) ^ Real.logb (Q' 0) (Q 0))
      (fun x : ℕ => zeckendorfDeltaStar L L' Q Q'
          (zeckendorfRealEval Q' (zeckendorfRev ((μ x).support.sup (· + 1)) (μ x))) *
        (α / α' ^ Real.logb (Q' 0) (Q 0)))
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Proposition 2.
Uses 8

TheoremLemma 13: behaviour of \(\overline{\mathrm{rev}}\) near a point#●ready to prove

Let \((L, \tilde L)\) be a periodic Zeckendorf pair, let \(E^*\) and \(\tilde E^*\) be the \(L^*\)- and \(\tilde L^*\)-collections, \(\bar E^*\) the closure collection of \(L\), \(\overline{\mathrm{rev}}\) the map determined by \(L\), \(\inf\) the \(L^*\)-infimum map, and \(\tilde Q\) a fundamental sequence of \(\tilde E^*\). For \(\eta \in \tilde E^*\) write \(x(\eta) = \sum_k \eta_k \tilde Q_k\), and let \(\varepsilon \in \tilde E^*\). If \(\varepsilon \notin \bar E^*\), there is \(\delta > 0\) such that \(\overline{\mathrm{rev}}(\varepsilon') = \overline{\mathrm{rev}}(\varepsilon)\) for every \(\varepsilon' \in \tilde E^*\) with \(|x(\varepsilon') - x(\varepsilon)| < \delta\). If \(\varepsilon \in \bar E^* \setminus E^*\), there is \(\delta > 0\) such that \(\inf(\overline{\mathrm{rev}}(\varepsilon')) = \overline{\mathrm{rev}}(\varepsilon)\) for every \(\varepsilon' \in \tilde E^*\) with \(x(\varepsilon) < x(\varepsilon') < x(\varepsilon) + \delta\), and for every \(n\) there is \(\delta > 0\) such that \(\overline{\mathrm{rev}}(\varepsilon')\) and \(\overline{\mathrm{rev}}(\varepsilon)\) agree at all indices \(i < n\) for every \(\varepsilon' \in \tilde E^*\) with \(x(\varepsilon) - \delta < x(\varepsilon') < x(\varepsilon)\). If \(\varepsilon \in E^*\), then for every \(n\) there is \(\delta > 0\) such that \(\overline{\mathrm{rev}}(\varepsilon')\) and \(\overline{\mathrm{rev}}(\varepsilon)\) agree at all indices \(i < n\) for every \(\varepsilon' \in \tilde E^*\) with \(|x(\varepsilon') - x(\varepsilon)| < \delta\). Indices start at \(0\).

MetaMathlibExt.zeckendorfRevBar_eventually_near · 17 lines
public theorem_wanted zeckendorfRevBar_eventually_near {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q' : ℕ → ℝ} (hQ' : IsLStarFundamentalSequence L' Q')
    {ε : ℕ → ℕ} (hε : ε ∈ lStarCollection L') :
    (ε ∉ lStarClosure L → ∃ δ > 0, ∀ ε' ∈ lStarCollection L',
      |zeckendorfRealEval Q' ε' - zeckendorfRealEval Q' ε| < δ →
        zeckendorfRevBar L ε' = zeckendorfRevBar L ε) ∧
    (ε ∈ lStarClosure L → ε ∉ lStarCollection L →
      (∃ δ > 0, ∀ ε' ∈ lStarCollection L', zeckendorfRealEval Q' ε < zeckendorfRealEval Q' ε' →
        zeckendorfRealEval Q' ε' < zeckendorfRealEval Q' ε + δ →
          lStarInf L (zeckendorfRevBar L ε') = zeckendorfRevBar L ε) ∧
      ∀ n, ∃ δ > 0, ∀ ε' ∈ lStarCollection L',
        zeckendorfRealEval Q' ε - δ < zeckendorfRealEval Q' ε' →
          zeckendorfRealEval Q' ε' < zeckendorfRealEval Q' ε →
            ∀ i < n, zeckendorfRevBar L ε' i = zeckendorfRevBar L ε i) ∧
    (ε ∈ lStarCollection L → ∀ n, ∃ δ > 0, ∀ ε' ∈ lStarCollection L',
      |zeckendorfRealEval Q' ε' - zeckendorfRealEval Q' ε| < δ →
        ∀ i < n, zeckendorfRevBar L ε' i = zeckendorfRevBar L ε i)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 13.
Uses 7

TheoremInfinitely many unit reductions stay in the \(L^\ast\)-closure#●ready to prove

Let \(L = (e_1, \dots, e_N)\) be a list of natural numbers with \(N \ge 2\) and \(e_1 \ne 0\), and let \(E^* \subseteq \bar E^*\) be the \(L^*\)-collection and the \(L^*\)-closure determined by \(L\). Let \(\varepsilon \in \bar E^*\), and suppose it is not the case that \(\varepsilon \in E^*\) and \(\varepsilon\) has finite support. Then there are infinitely many indices \(r\) such that \(\varepsilon_r \ge 1\) and the coefficient function obtained from \(\varepsilon\) by lowering its entry at \(r\) by one lies in \(\bar E^*\).

public theorem_wanted infinite_setOf_sub_single_mem_lStarClosure (L : List ℕ)
    (hL : 2 ≤ L.length) (he : L.getD 0 0 ≠ 0) {ε : ℕ → ℕ} (hε : ε ∈ lStarClosure L)
    (hfin : ¬ (ε ∈ lStarCollection L ∧ ∃ M, ∀ i, M < i → ε i = 0)) :
    {r | 1 ≤ ε r ∧ Function.update ε r (ε r - 1) ∈ lStarClosure L}.Infinite
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 14.
Uses 2

TheoremProposition 3: \(U\) is a disjoint union of open intervals#●ready to prove

Let \((L, \tilde L)\) be a periodic Zeckendorf pair, \(E^*\) and \(\tilde E^*\) the \(L^*\)- and \(\tilde L^*\)-collections, \(\bar E^*\) the closure collection of \(L\), and \(\tilde Q\) a fundamental sequence of \(\tilde E^*\). Write \(x(\eta) = \sum_k \eta_k \tilde Q_k\), \(\tilde\omega = \tilde Q_0\), and \(\bar\beta^{b+1}\) for the maximal \(L^*\)-block starting at index \(b\) (indices start at \(0\)), and let \(U = \{x(\varepsilon) : \varepsilon \in \tilde E^* \setminus \bar E^*\}\). Call a pair \((b, \eta)\) with \(b \in \mathbb N\) and \(\eta : \mathbb N \to \mathbb N\) admissible if \(\eta\) is a sum of disjoint proper \(L^*\)-blocks whose support intervals consist of indices below \(b\) and \(\eta + \bar\beta^{b+1} \notin E^*\). Then \(U\) is the union, over all admissible pairs \((b, \eta)\), of the open intervals \((x(\eta + \bar\beta^{b+1}), x(\eta) + \tilde\omega^b)\); the intervals of two distinct admissible pairs are disjoint; and \(U\) is open and measurable.

MetaMathlibExt.zeckendorfSmoothSet_eq_iUnion · 16 lines
public theorem_wanted zeckendorfSmoothSet_eq_iUnion {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q' : ℕ → ℝ} (hQ' : IsLStarFundamentalSequence L' Q') :
    zeckendorfSmoothSet L L' Q' =
      (⋃ (b : ℕ) (η : ℕ → ℕ) (_ : (∃ I, IsLStarBlockSum L I η ∧ ∀ p ∈ I, p.2 < b) ∧
          η + maxLStarBlock L b ∉ lStarCollection L),
        Set.Ioo (zeckendorfRealEval Q' (η + maxLStarBlock L b))
          (zeckendorfRealEval Q' η + Q' 0 ^ b)) ∧
    (∀ b η b' η', (∃ I, IsLStarBlockSum L I η ∧ ∀ p ∈ I, p.2 < b) →
      η + maxLStarBlock L b ∉ lStarCollection L →
      (∃ I, IsLStarBlockSum L I η' ∧ ∀ p ∈ I, p.2 < b') →
      η' + maxLStarBlock L b' ∉ lStarCollection L → (b, η) ≠ (b', η') →
      Disjoint (Set.Ioo (zeckendorfRealEval Q' (η + maxLStarBlock L b))
          (zeckendorfRealEval Q' η + Q' 0 ^ b))
        (Set.Ioo (zeckendorfRealEval Q' (η' + maxLStarBlock L b'))
          (zeckendorfRealEval Q' η' + Q' 0 ^ b'))) ∧
    IsOpen (zeckendorfSmoothSet L L' Q') ∧ MeasurableSet (zeckendorfSmoothSet L L' Q')
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Proposition 3.
Uses 7

TheoremContinuity and local monotonicity of \(\delta^{\ast}\)#●ready to prove

Let \(L\) and \(\tilde L\) form a periodic Zeckendorf pair, let \(Q\) be a fundamental sequence of the collection \(E^*\) determined by \(L\), and let \(\tilde Q\) be a fundamental sequence of the collection \(\tilde E^*\) determined by \(\tilde L\). Then the function \(\delta^*\) built from \(L, \tilde L, Q, \tilde Q\) is continuous on \((0, 1)\), and at each point \(x\) of the smooth set \(U\) of \(L, \tilde L, \tilde Q\) it is differentiable and strictly decreasing on some neighbourhood of \(x\).

public theorem_wanted continuousOn_zeckendorfDeltaStar {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q Q' : ℕ → ℝ} (hQ : IsLStarFundamentalSequence L Q)
    (hQ' : IsLStarFundamentalSequence L' Q') :
    ContinuousOn (zeckendorfDeltaStar L L' Q Q') (Set.Ioo 0 1) ∧
      ∀ x ∈ zeckendorfSmoothSet L L' Q', DifferentiableAt ℝ (zeckendorfDeltaStar L L' Q Q') x ∧
        ∃ s ∈ nhds x, StrictAntiOn (zeckendorfDeltaStar L L' Q Q') s
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 9.
Uses 4

Theorem\(\delta^*\) is not differentiable at the evaluations of \(\bar E^*\)#●ready to prove

\(\delta^*\) is not differentiable at the evaluations of \(\bar E^*\). Let \(L\) and \(\tilde L\) be lists of natural numbers of length at least \(2\) with nonzero first entries whose periodic Zeckendorf collections \(E\) and \(\tilde E\) satisfy \(E \subsetneq \tilde E\). Let \(Q\) be a fundamental sequence of the collection \(E^*\) determined by \(L\), let \(\tilde Q\) be a fundamental sequence of the collection \(\tilde E^*\) determined by \(\tilde L\), and let \(\delta^*\) be the function determined by \(L\), \(\tilde L\), \(Q\) and \(\tilde Q\). For every \(\varepsilon\) in the closure \(\bar E^*\) determined by \(L\), \(\delta^*\) is not differentiable at \(\sum_k \varepsilon_k \tilde Q_k\).

public theorem_wanted not_differentiableAt_zeckendorfDeltaStar {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q Q' : ℕ → ℝ} (hQ : IsLStarFundamentalSequence L Q)
    (hQ' : IsLStarFundamentalSequence L' Q') {ε : ℕ → ℕ} (hε : ε ∈ lStarClosure L) :
    ¬ DifferentiableAt ℝ (zeckendorfDeltaStar L L' Q Q') (zeckendorfRealEval Q' ε)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 10.
Uses 5

TheoremThe smooth set \(U\) has Lebesgue measure \(1\)#●ready to prove

Let \(L\) and \(\tilde L\) form a periodic Zeckendorf pair, let \(\tilde E^*\) be the collection \(E^*\) determined by \(\tilde L\), let \(\tilde Q\) be a fundamental sequence of \(\tilde E^*\), and let \(\bar E^*\) be the closure determined by \(L\). Let \(U\) be the image of \(\tilde E^* \setminus \bar E^*\) under the evaluation \(\varepsilon \mapsto \sum_k \varepsilon_k \tilde Q_k\). Then \(U\) has Lebesgue measure \(1\).

public theorem_wanted volume_zeckendorfSmoothSet {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q' : ℕ → ℝ} (hQ' : IsLStarFundamentalSequence L' Q') :
    MeasureTheory.volume (zeckendorfSmoothSet L L' Q') = 1
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 11.
Uses 3

TheoremLocal minima of \(\delta^\ast\) at finitely supported points#●ready to prove

Let \(L\) and \(\tilde L\) form a periodic Zeckendorf pair, let \(Q\) and \(\tilde Q\) be fundamental sequences of the \(L^*\)-collection \(E^*\) and the \(\tilde L^*\)-collection \(\tilde E^*\), and let \(\delta^*\) be the function determined by \(L\), \(\tilde L\), \(Q\) and \(\tilde Q\). If \(\varepsilon \in E^*\) has finite support, then \(\delta^*\) has a local minimum at \(\sum_k \varepsilon_k \tilde Q_k\).

public theorem_wanted isLocalMin_zeckendorfDeltaStar_of_finite {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q Q' : ℕ → ℝ} (hQ : IsLStarFundamentalSequence L Q)
    (hQ' : IsLStarFundamentalSequence L' Q') {ε : ℕ → ℕ} (hε : ε ∈ lStarCollection L)
    (hfin : ∃ M, ∀ i, M < i → ε i = 0) :
    IsLocalMin (zeckendorfDeltaStar L L' Q Q') (zeckendorfRealEval Q' ε)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Proposition 6.
Uses 5

Theorem\(\delta^*\) has no local extremum at infinite expansions in \(E^*\)#●ready to prove

\(\delta^*\) has no local extremum at infinite expansions in \(E^*\). Let \(L\) and \(\tilde L\) be lists of natural numbers of length at least \(2\) with nonzero first entries whose periodic Zeckendorf collections satisfy \(E \subsetneq \tilde E\). Let \(Q\) be a fundamental sequence of the collection \(E^*\) determined by \(L\), let \(\tilde Q\) be a fundamental sequence of the collection \(\tilde E^*\) determined by \(\tilde L\), and let \(\delta^*\) be the function determined by \(L\), \(\tilde L\), \(Q\) and \(\tilde Q\). If \(\varepsilon \in E^*\) has infinitely many nonzero entries, then \(\delta^*\) has neither a local minimum nor a local maximum at \(\sum_k \varepsilon_k \tilde Q_k\).

public theorem_wanted not_isLocalExtr_zeckendorfDeltaStar_of_infinite {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q Q' : ℕ → ℝ} (hQ : IsLStarFundamentalSequence L Q)
    (hQ' : IsLStarFundamentalSequence L' Q') {ε : ℕ → ℕ} (hε : ε ∈ lStarCollection L)
    (hinf : ∀ M, ∃ i, M < i ∧ ε i ≠ 0) :
    ¬ IsLocalMin (zeckendorfDeltaStar L L' Q Q') (zeckendorfRealEval Q' ε) ∧
      ¬ IsLocalMax (zeckendorfDeltaStar L L' Q Q') (zeckendorfRealEval Q' ε)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Proposition 7.
Uses 5

TheoremStrict local maxima of \(\delta^{\ast}\) at the closure points#●ready to prove

Let \(L\) and \(\tilde L\) form a periodic Zeckendorf pair, let \(Q\) be a fundamental sequence of the collection \(E^*\) determined by \(L\), and let \(\tilde Q\) be a fundamental sequence of the collection \(\tilde E^*\) determined by \(\tilde L\). Let \(\varepsilon \in \bar E^* \setminus E^*\), where \(\bar E^*\) is the closure collection of \(L\), and put \(x = \sum_k \varepsilon_k \tilde Q_k\). Then, for the function \(\delta^*\) built from \(L, \tilde L, Q, \tilde Q\), we have \(\delta^*(y) < \delta^*(x)\) for all \(y \ne x\) sufficiently close to \(x\).

public theorem_wanted eventually_zeckendorfDeltaStar_lt_of_mem_lStarClosure {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q Q' : ℕ → ℝ} (hQ : IsLStarFundamentalSequence L Q)
    (hQ' : IsLStarFundamentalSequence L' Q') {ε : ℕ → ℕ} (hε : ε ∈ lStarClosure L)
    (hε' : ε ∉ lStarCollection L) :
    ∀ᶠ y in nhdsWithin (zeckendorfRealEval Q' ε) {zeckendorfRealEval Q' ε}ᶜ,
      zeckendorfDeltaStar L L' Q Q' y < zeckendorfDeltaStar L L' Q Q' (zeckendorfRealEval Q' ε)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Proposition 8.
Uses 6

TheoremClassification of the local extrema of \(\delta^\ast\)#●ready to prove

Let \(L\) and \(\tilde L\) form a periodic Zeckendorf pair, let \(Q\) and \(\tilde Q\) be fundamental sequences of the \(L^*\)-collection \(E^*\) and the \(\tilde L^*\)-collection \(\tilde E^*\), let \(\bar E^*\) be the \(L^*\)-closure, and let \(\delta^*\) be the function determined by \(L\), \(\tilde L\), \(Q\) and \(\tilde Q\). For a coefficient function \(\varepsilon\) write \(x_\varepsilon = \sum_k \varepsilon_k \tilde Q_k\). Then (i) for every \(\varepsilon \in \tilde E^*\) such that either \(\varepsilon \notin \bar E^*\), or \(\varepsilon \in E^*\) and \(\varepsilon\) has infinite support, \(\delta^*\) has neither a local minimum nor a local maximum at \(x_\varepsilon\); (ii) for every \(\varepsilon \in E^*\) with finite support, \(\delta^*\) has a local minimum at \(x_\varepsilon\); and (iii) for every \(\varepsilon \in \bar E^* \setminus E^*\), \(\delta^*\) has a local maximum at \(x_\varepsilon\).

public theorem_wanted isLocalExtr_zeckendorfDeltaStar_classification {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q Q' : ℕ → ℝ} (hQ : IsLStarFundamentalSequence L Q)
    (hQ' : IsLStarFundamentalSequence L' Q') :
    (∀ ε ∈ lStarCollection L', (ε ∉ lStarClosure L ∨
        (ε ∈ lStarCollection L ∧ ∀ M, ∃ i, M < i ∧ ε i ≠ 0)) →
      ¬ IsLocalMin (zeckendorfDeltaStar L L' Q Q') (zeckendorfRealEval Q' ε) ∧
        ¬ IsLocalMax (zeckendorfDeltaStar L L' Q Q') (zeckendorfRealEval Q' ε)) ∧
    (∀ ε ∈ lStarCollection L, (∃ M, ∀ i, M < i → ε i = 0) →
      IsLocalMin (zeckendorfDeltaStar L L' Q Q') (zeckendorfRealEval Q' ε)) ∧
    ∀ ε ∈ lStarClosure L, ε ∉ lStarCollection L →
      IsLocalMax (zeckendorfDeltaStar L L' Q Q') (zeckendorfRealEval Q' ε)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 12.
Uses 6

TheoremGeneric upper bound for \(\delta^*\) (Lemma 17)#●ready to prove

Generic upper bound for \(\delta^*\) (Lemma 17). Let \(L\) and \(\tilde L\) be lists of length at least \(2\) with nonzero first entries whose periodic Zeckendorf collections satisfy \(E \subsetneq \tilde E\), and let \(Q\) and \(\tilde Q\) be fundamental sequences of the collections \(E^*\) and \(\tilde E^*\) of \(L\) and \(\tilde L\). Write \(\omega\) and \(\tilde\omega\) for the first terms of \(Q\) and \(\tilde Q\), \(\gamma = \log_{\tilde\omega} \omega\), \(p\) for the generic exponent of \(Q\) and \(\tilde Q\), \(\delta^*\) for the function \(\delta^*\) of \(L\), \(\tilde L\), \(Q\), \(\tilde Q\), \(\overline{\mathrm{rev}}\) for the reversal map of \(L\) (Definition 23), \(\varepsilon_k\) for the \(k\)th entry of a coefficient function \(\varepsilon\) (\(k \ge 1\)), and \(\sum \varepsilon Q = \sum_{k \ge 1} \varepsilon_k Q_k\) (so that \(Q_1 = \omega\)). Then \(\delta^*(\sum \varepsilon \tilde Q) < (\omega / \tilde\omega)^p / \gamma\) for every \(\varepsilon \in \tilde E^*\), and every \(\varepsilon \in \tilde E^*\) with \(\sum \varepsilon \tilde Q \ge \tilde\omega\) satisfies \(\delta^*(\sum \varepsilon \tilde Q) < \sum \overline{\mathrm{rev}}(\varepsilon) Q / \sum \varepsilon \tilde Q < (\omega / \tilde\omega)^p / \gamma\).

Without \(\sum \varepsilon \tilde Q \ge \tilde\omega\) (the normalization \(\varepsilon_1 \ge 1\) that the proof of Lemma 17 assumes) the middle bound fails: for \(L = (1, 0)\), \(\tilde L = (1, 1)\) (so \(p = 2\)) and \(\varepsilon = \beta^4\) (a single \(1\) in the fourth entry), \(\overline{\mathrm{rev}}(\varepsilon) = \varepsilon\) and the middle quotient is \(16 \omega^4 = 2.33436\ldots\), while \((\omega / \tilde\omega)^2 / \gamma = 2.20076\ldots\).

public theorem_wanted zeckendorfDeltaStar_lt_genericBound {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q Q' : ℕ → ℝ} (hQ : IsLStarFundamentalSequence L Q)
    (hQ' : IsLStarFundamentalSequence L' Q') :
    (∀ ε ∈ lStarCollection L', zeckendorfDeltaStar L L' Q Q' (zeckendorfRealEval Q' ε) <
      (Q 0 / Q' 0) ^ zeckendorfGenericExponent Q Q' / Real.logb (Q' 0) (Q 0)) ∧
    ∀ ε ∈ lStarCollection L', Q' 0 ≤ zeckendorfRealEval Q' ε →
      zeckendorfDeltaStar L L' Q Q' (zeckendorfRealEval Q' ε) <
          zeckendorfRealEval Q (zeckendorfRevBar L ε) / zeckendorfRealEval Q' ε ∧
        zeckendorfRealEval Q (zeckendorfRevBar L ε) / zeckendorfRealEval Q' ε <
          (Q 0 / Q' 0) ^ zeckendorfGenericExponent Q Q' / Real.logb (Q' 0) (Q 0)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 17.
Uses 7

TheoremMonotonicity of the ratio under steep increments#●ready to prove

Let \(L\) and \(\tilde L\) form a periodic Zeckendorf pair, let \(Q\) be a fundamental sequence of the collection \(E^*\) determined by \(L\), and let \(\tilde Q\) be a fundamental sequence of the collection \(\tilde E^*\) determined by \(\tilde L\). Put \(\omega = Q_0\), \(\tilde\omega = \tilde Q_0\) and \(\gamma = \log_{\tilde\omega} \omega\), and let \(p\) be the generic exponent of \(Q, \tilde Q\). Let \(\varepsilon \in \tilde E^*\) with \(x = \sum_k \varepsilon_k \tilde Q_k \ge \tilde\omega\), and put \(y = \sum_k \overline{\mathrm{rev}}(\varepsilon)_k Q_k\), with \(\overline{\mathrm{rev}}\) taken with respect to \(L\). If \(\Delta x > 0\), \(\Delta y > 0\) and \((\omega / \tilde\omega)^p \le \Delta y / \Delta x\), then \(y / x^\gamma < (y + \Delta y) / (x + \Delta x)^\gamma\).

The hypothesis \(x \ge \tilde\omega\) follows the reading of Lemma 17. As printed (for all \(\varepsilon \in \tilde E^*\)) the lemma fails for \(L = (1, 0)\), \(\tilde L = (1, 1)\), \(\varepsilon\) the coefficient function with a single \(1\) at index \(3\) (\(x = 1/16\), \(y = \omega^4\), \(y / x^\gamma = 1\)), \(\Delta x = 10^{-8}\) and \(\Delta y = (\omega / \tilde\omega)^2 \Delta x\): the right side is \(0.99999999364\ldots\).

public theorem_wanted div_rpow_lt_add_div_add_rpow_of_genericBound {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q Q' : ℕ → ℝ} (hQ : IsLStarFundamentalSequence L Q)
    (hQ' : IsLStarFundamentalSequence L' Q') {ε : ℕ → ℕ} (hε : ε ∈ lStarCollection L')
    (hx : Q' 0 ≤ zeckendorfRealEval Q' ε) {Δx Δy : ℝ} (hΔx : 0 < Δx) (hΔy : 0 < Δy)
    (hΔ : (Q 0 / Q' 0) ^ zeckendorfGenericExponent Q Q' ≤ Δy / Δx) :
    zeckendorfRealEval Q (zeckendorfRevBar L ε) /
        zeckendorfRealEval Q' ε ^ Real.logb (Q' 0) (Q 0) <
      (zeckendorfRealEval Q (zeckendorfRevBar L ε) + Δy) /
        (zeckendorfRealEval Q' ε + Δx) ^ Real.logb (Q' 0) (Q 0)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 18.
Uses 6

TheoremMonotonicity of the ratio under fundamental increments (Corollary 3)#●ready to prove

Let \(L\) and \(\tilde L\) form a periodic Zeckendorf pair, let \(Q\) be a fundamental sequence of the collection \(E^*\) determined by \(L\), and let \(\tilde Q\) be a fundamental sequence of the collection \(\tilde E^*\) determined by \(\tilde L\); indices start at \(0\). Put \(\omega = Q_0\), \(\tilde\omega = \tilde Q_0\) and \(\gamma = \log_{\tilde\omega} \omega\), let \(p\) be the generic exponent of \(Q\) and \(\tilde Q\), let \(\delta^*\) be the function \(\delta^*\) of \(L\), \(\tilde L\), \(Q\), \(\tilde Q\), and let \(\overline{\mathrm{rev}}\) be the map determined by \(L\). Let \(\varepsilon \in \tilde E^*\) with \(x = \sum_k \varepsilon_k \tilde Q_k \ge \tilde\omega\), put \(y = \sum_k \overline{\mathrm{rev}}(\varepsilon)_k Q_k\), and let \(m\) be a natural number with \(p \le m + 1\). Then \(\delta^*(x) = y / x^\gamma\) and \(y / x^\gamma < (y + Q_m) / (x + \tilde Q_m)^\gamma\).

The hypothesis \(x \ge \tilde\omega\) follows the reading of Lemma 17; in the paper's one-based indexing \(Q_m\) is \(Q_n\) with \(n = m + 1 \ge p\). As printed (for all \(\varepsilon \in \tilde E^*\)) the corollary fails for \(L = (1, 0)\), \(\tilde L = (1, 1)\), \(n = p = 2\) and \(\varepsilon = \beta^{11} + \beta^{13}\) (single ones in the paper's entries \(11\) and \(13\)), for which \(\overline{\mathrm{rev}}(\varepsilon) = \varepsilon\): the left side \(1.18363\ldots\) exceeds the right side \(1.01645\ldots\).

public theorem_wanted zeckendorfDeltaStar_lt_add_fundamental {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q Q' : ℕ → ℝ} (hQ : IsLStarFundamentalSequence L Q)
    (hQ' : IsLStarFundamentalSequence L' Q') {ε : ℕ → ℕ} (hε : ε ∈ lStarCollection L')
    (hx : Q' 0 ≤ zeckendorfRealEval Q' ε) {m : ℕ} (hm : zeckendorfGenericExponent Q Q' ≤ m + 1) :
    zeckendorfDeltaStar L L' Q Q' (zeckendorfRealEval Q' ε) =
        zeckendorfRealEval Q (zeckendorfRevBar L ε) /
          zeckendorfRealEval Q' ε ^ Real.logb (Q' 0) (Q 0) ∧
      zeckendorfRealEval Q (zeckendorfRevBar L ε) /
          zeckendorfRealEval Q' ε ^ Real.logb (Q' 0) (Q 0) <
        (zeckendorfRealEval Q (zeckendorfRevBar L ε) + Q m) /
          (zeckendorfRealEval Q' ε + Q' m) ^ Real.logb (Q' 0) (Q 0)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Corollary 3.
Uses 7

TheoremWhere \(\delta^\ast\) can attain its maximum#●ready to prove

Let \(L\) and \(\tilde L\) be lists of natural numbers forming a periodic Zeckendorf pair, let \(N\) be the length of \(L\), let \(E^*\) and \(\tilde E^*\) be the collections determined by \(L\) and \(\tilde L\), let \(\bar E^*\) be the closure of \(E^*\), and let \(Q\) and \(\tilde Q\) be fundamental sequences of \(E^*\) and \(\tilde E^*\). Let \(\delta^*\) be the function \(\delta^*\) of \(L\), \(\tilde L\), \(Q\), \(\tilde Q\), and let \(p\) be the generic exponent of \(Q\) and \(\tilde Q\). Indices start at \(0\): put \(\omega = Q_0\) and \(\tilde\omega = \tilde Q_0\), write \(\sum \varepsilon \tilde Q = \sum_k \varepsilon_k \tilde Q_k\) for a coefficient function \(\varepsilon : \mathbb N \to \mathbb N\), and for \(s \in \mathbb N\) let \(\bar\beta_s\) be the maximal \(L^*\)-block starting at index \(s\). First, if \(y \in (0, 1)\) and \(\delta^*(x) \le \delta^*(y)\) for all \(x \in (0, 1)\), then \(y = \sum \varepsilon \tilde Q\) for some \(\varepsilon \in \bar E^*\) with \(\varepsilon \notin E^*\). Second, let \(\ell \ge 1\) be a natural number and let \(\eta : \mathbb N \to \mathbb N\) be a sum of disjoint proper \(L^*\)-blocks (possibly none) whose support intervals \([s, k]\) all satisfy \(k < \ell - 1\). If \(\delta^*(x) \le \delta^*\bigl(\sum (\eta + \bar\beta_{\ell - 1}) \tilde Q\bigr)\) for all \(x \in (0, 1)\), then

\[ \ell < \max\left(2,\ p + 1 + \frac{N \ln(1 / \omega) + \ln\left(1 - \sum \bar\beta_0 \tilde Q + \tilde\omega^N\right)} {\ln(1 / \tilde\omega) - \ln(1 / \omega)}\right). \]

In the source's one-based indexing \(\bar\beta_s\) is \(\bar\beta^{s+1}\), so the blocks \(\zeta^m\) of \(\eta\) lie inside \([1, \ell - 1]\), the point is \(\sum (\sum_m \zeta^m + \bar\beta^\ell) \tilde Q\), and with \(\varphi = 1 / \omega\), \(\tilde\varphi = 1 / \tilde\omega\) and \(\rho = \sum \bar\beta^1 \tilde Q\) the right side is the source's \(\max\{2, p^*\}\).

public theorem_wanted isMaxOn_zeckendorfDeltaStar_bound {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q Q' : ℕ → ℝ} (hQ : IsLStarFundamentalSequence L Q)
    (hQ' : IsLStarFundamentalSequence L' Q') :
    (∀ y ∈ Set.Ioo (0 : ℝ) 1, IsMaxOn (zeckendorfDeltaStar L L' Q Q') (Set.Ioo 0 1) y →
      ∃ ε ∈ lStarClosure L, ε ∉ lStarCollection L ∧ zeckendorfRealEval Q' ε = y) ∧
    ∀ (ℓ : ℕ) (η : ℕ → ℕ) (I : Set (ℕ × ℕ)), 1 ≤ ℓ → IsLStarBlockSum L I η →
      (∀ p ∈ I, p.2 < ℓ - 1) →
      IsMaxOn (zeckendorfDeltaStar L L' Q Q') (Set.Ioo 0 1)
        (zeckendorfRealEval Q' (η + maxLStarBlock L (ℓ - 1))) →
      (ℓ : ℝ) < max 2 (zeckendorfGenericExponent Q Q' + 1 +
        (L.length * Real.log (Q 0)⁻¹ +
          Real.log (1 - zeckendorfRealEval Q' (maxLStarBlock L 0) + Q' 0 ^ L.length)) /
        (Real.log (Q' 0)⁻¹ - Real.log (Q 0)⁻¹))
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 13.
Uses 9

TheoremWhere \(\delta^*\) attains its minimum#●ready to prove

Where \(\delta^*\) attains its minimum. Let \(L\) and \(\tilde L\) be lists of natural numbers of length at least \(2\) with nonzero first entries whose periodic Zeckendorf collections satisfy \(E \subsetneq \tilde E\). Let \(Q\) and \(\tilde Q\) be fundamental sequences of the \(L^*\)-collections \(E^*\) of \(L\) and \(\tilde E^*\) of \(\tilde L\), let \(\delta^*\) be the function on \((0, 1)\) built from \(L\), \(\tilde L\), \(Q\) and \(\tilde Q\), and let \(p\) be the generic exponent of \(Q\) and \(\tilde Q\). Indices start at \(0\). Then there is \(\varepsilon \in E^*\) with \(\varepsilon_i = 0\) for every \(i \ge \max(2, p) - 1\) such that \(\delta^*(\sum_k \varepsilon_k \tilde Q_k) \le \delta^*(y)\) for all \(y \in (0, 1)\). Moreover, if \(p \le 2\), then \(\delta^*(\tilde Q_0) \le \delta^*(y)\) for all \(y \in (0, 1)\) and \(\delta^*(\tilde Q_0) = 1\).

public theorem_wanted exists_isMinOn_zeckendorfDeltaStar {L L' : List ℕ}
    (hp : IsPeriodicZeckendorfPair L L') {Q Q' : ℕ → ℝ} (hQ : IsLStarFundamentalSequence L Q)
    (hQ' : IsLStarFundamentalSequence L' Q') :
    (∃ ε ∈ lStarCollection L, (∀ i, max 2 (zeckendorfGenericExponent Q Q') - 1 ≤ i → ε i = 0) ∧
      IsMinOn (zeckendorfDeltaStar L L' Q Q') (Set.Ioo 0 1) (zeckendorfRealEval Q' ε)) ∧
    (zeckendorfGenericExponent Q Q' ≤ 2 →
      IsMinOn (zeckendorfDeltaStar L L' Q Q') (Set.Ioo 0 1) (Q' 0) ∧
        zeckendorfDeltaStar L L' Q Q' (Q' 0) = 1)
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Theorem 14.
Uses 6

TheoremShifting a unit one entry down increases \(\delta^*\) (Lemma 19)#●ready to prove

Shifting a unit one entry down increases \(\delta^*\) (Lemma 19). Let \(N \ge 3\), let \(L = (1, \dots, 1, 0)\) (\(N - 1\) ones followed by \(0\), the \(N\)th order Zeckendorf collection) and \(\tilde L = (N - 1, \dots, N - 1)\) (\(N\) entries, the base-\(N\) collection), and let \(Q\) and \(\tilde Q\) be fundamental sequences of the collections \(E^*\) and \(\tilde E^*\) of \(L\) and \(\tilde L\). Write \(\varepsilon_k\) for the \(k\)th entry of a coefficient function (\(k \ge 1\)), \(\beta^k\) for the coefficient function with a single \(1\) in entry \(k\), \(\sum \varepsilon \tilde Q = \sum_k \varepsilon_k \tilde Q_k\), \(\delta^*\) for the function \(\delta^*\) of \(L\), \(\tilde L\), \(Q\), \(\tilde Q\), and \(p\) for the generic exponent of \(Q\) and \(\tilde Q\). Then \(p = 1\) if \(N \ge 4\). Moreover, let \(\varepsilon \in E^*\) with \(\varepsilon_1 \ge 1\) and let \(n \ge 3\) with \(\varepsilon_n = 1\). If some decomposition of \(\varepsilon\) as a sum of disjoint proper \(L^*\)-blocks has a block whose support interval ends at entry \(n - 1\), then \(\varepsilon' = \varepsilon + \beta^{n-1} - \beta^n\) lies in \(E^*\) and \(\delta^*(\sum \varepsilon \tilde Q) < \delta^*(\sum \varepsilon' \tilde Q)\).

As printed, \(p = 1\) fails for \(N = 3\): with \(\omega\) and \(\tilde\omega\) the first terms of \(Q\) and \(\tilde Q\) and \(\gamma = \log_{\tilde\omega} \omega\), \(\gamma = 0.55467\ldots\) exceeds \(\omega = 0.54368\ldots\), so \(p = 2\). The inequality \(\delta^*(\sum \varepsilon \tilde Q) < \delta^*(\sum \varepsilon' \tilde Q)\) fails without \(\varepsilon_1 \ge 1\) (the normalization used where the paper applies it): \(\varepsilon = \beta^4\) and \(\varepsilon' = \beta^3\) give the value \(1\) on both sides.

public theorem_wanted zeckendorfDeltaStar_lt_shift_baseN {N : ℕ} (hN : 3 ≤ N) {Q Q' : ℕ → ℝ}
    (hQ : IsLStarFundamentalSequence (List.replicate (N - 1) 1 ++ [0]) Q)
    (hQ' : IsLStarFundamentalSequence (List.replicate N (N - 1)) Q') :
    (4 ≤ N → zeckendorfGenericExponent Q Q' = 1) ∧
    ∀ ε ∈ lStarCollection (List.replicate (N - 1) 1 ++ [0]), 1 ≤ ε 0 →
      ∀ (n i : ℕ) (I : Set (ℕ × ℕ)), 3 ≤ n → ε (n - 1) = 1 →
        IsLStarBlockSum (List.replicate (N - 1) 1 ++ [0]) I ε → (i, n - 2) ∈ I →
        ε + Pi.single (n - 2) 1 - Pi.single (n - 1) 1 ∈
            lStarCollection (List.replicate (N - 1) 1 ++ [0]) ∧
          zeckendorfDeltaStar (List.replicate (N - 1) 1 ++ [0]) (List.replicate N (N - 1)) Q Q'
              (zeckendorfRealEval Q' ε) <
            zeckendorfDeltaStar (List.replicate (N - 1) 1 ++ [0]) (List.replicate N (N - 1)) Q Q'
              (zeckendorfRealEval Q' (ε + Pi.single (n - 2) 1 - Pi.single (n - 1) 1))
Sources
  • Sungkon Chang, Distribution of Zeckendorf Expressions, INTEGERS 25 (2025), Article A25, Lemma 19.
Uses 6
45 fully proved46 ready to prove

The dependency explorer needs JavaScript and its graph data.

Download the complete graph data.