Uniform positivity on bounded scaled windows and a fixed inner margin
ProvedBranchNearRepair.uniform_window_and_corecomplex-analysisdesenvolvimento-rhpositivityuniform-estimates
Let kₙ≥1 be integers, let hₙ>0 tend to zero, and fix A≥1. Put xₙ=kₙ/hₙ². For real ε and γ define
For all sufficiently large n, uniformly over |ε|≤1/2 and 0≤γ with γhₙ≤A, one has Pₙ(ε,γ)>0. On the inner window γhₙ≤1, the stronger lower bound Pₙ(ε,γ)≥e⁻¹/2 holds. The threshold can depend on A but not on ε or γ. This analytic statement requires no assumption that ε=0 and asserts no conclusion about RH.
Preamble
import Mathlib
Formal statement
theorem BranchNearRepair.uniform_window_and_core (k : ℕ → ℕ) (hk : ∀ n, 1≤k n)
(h : ℕ → ℝ) (hh : ∀ n, 0<h n) (ht : Filter.Tendsto h Filter.atTop (nhds 0))
(A : ℝ) (hA : 1≤A) :
∀ᶠ n in Filter.atTop, ∀ e g : ℝ, |e|≤1/2 → 0≤g → g*h n≤A →
0 < (((((k n : ℝ)/(h n)^2 : ℝ) : ℂ)/
(((k n : ℝ)/(h n)^2+g^2-e^2 : ℝ)-(2*e*g : ℝ)*Complex.I))^(k n)).re ∧
(g*h n≤1 → Real.exp (-1)/2 ≤ (((((k n : ℝ)/(h n)^2 : ℝ) : ℂ)/
(((k n : ℝ)/(h n)^2+g^2-e^2 : ℝ)-(2*e*g : ℝ)*Complex.I))^(k n)).re) := by sorrySource
Desenvolvimento RH — Ricardo Marangoni Brandão Bueno. Two-cutoff repair for the small-scale escape argument, 2026-11-16. Complete elementary Lean proof; no claim of mathematical novelty.