Documentation

LeanPool.ThreeGap.EuclideanNN

The Euclidean isometry reduction: gapVal is the Euclidean nearest-neighbour distance #

The combinatorial bound gap_count_euclidean (g₂ ≤ 6) counts distinct values of gapVal (deltaE α) N q. This file identifies that quantity with the genuine Euclidean torus nearest-neighbour distance, completing the Euclidean five-distance bound g₂ ≤ 6 as a statement about the actual orbit {0, α, …, Nα} on 𝕋².

The isometry reduction (TorusReduction.gapVal_eq_nnDist) used only the symmetry of the cost, so it is stated here generically for any symmetric c : ℤ → ℝ and instantiated at the Euclidean defect c = deltaN (euclNorm 2) α (symmetric by deltaN_neg). Axiom-clean.

noncomputable def ThreeGap.EuclideanRecords.nnDistC (c : ) (N q : ) :

The nearest-neighbour distance for a symmetric cost c, abstractly: the minimum of c (q − j) over the other points j ∈ {0,…,N} ∖ {q}.

Equations
Instances For
    theorem ThreeGap.EuclideanRecords.gapVal_eq_nnDistC (c : ) (hsymm : ∀ (t : ), c (-t) = c t) {N : } (hN : 2 N) {q : } (hq : q N) :
    Chevallier.gapVal (fun (n : ) => c n) N q = nnDistC c N q

    Generic isometry reduction. For a symmetric cost c, the gap value equals the nearest-neighbour distance, via the value-preserving offset bijection j ↦ |q − j|.

    noncomputable def ThreeGap.EuclideanRecords.nnDistE (α : Fin 2) (N q : ) :

    The Euclidean torus nearest-neighbour distance of among {0, α, …, Nα}, via d_𝕋(iα, jα) = deltaN (euclNorm 2) α (i − j).

    Equations
    Instances For
      theorem ThreeGap.EuclideanRecords.nnDistE_count_le (α : Fin 2) {k₀ : Fin 2} (hirr : Irrational (α k₀)) {N : } (hN : 2 N) :

      g₂ ≤ 6 for the actual Euclidean nearest-neighbour distances on 𝕋² (unconditional). For any α : Fin 2 → ℝ with an irrational coordinate and N ≥ 2, the orbit {0, α, …, Nα} has at most six distinct nearest-neighbour distances in the Euclidean metric. (The sharp ≤ 5 needs Romanov's K = 4.)