Revealed overlap detects a full overlap deviation
ProvedZeroFreeRRG.Hardcore.reveal_overlap_miss_concentrationFor and , if the full empirical pair overlap differs from the product marginal by more than in some coordinate, the independent Bernoulli() reveal set misses the paper’s revealed-deviation event with probability at most . This is the first Bernstein bound in the proof of Lemma 5.6.
Preamble
import Mathlib import Definitions.Def_ZeroFreeRRG_Hardcore_RevealOverlap set_option autoImplicit false open Set Filter MeasureTheory AFIsingKS ZeroFreeRRG.Hardcore open scoped Classical BigOperators
Formal statement
theorem ZeroFreeRRG.Hardcore.reveal_overlap_miss_concentration (m : Model) (hm : m.Valid)
(t s η : ℝ) (ht : m.PosParam t) (hs : 0 < s) (hs1 : s ≤ 1)
(hη : 0 < η) (hη1 : η ≤ 1) (n : ℕ) (σ σ' : Fin n → m.Spin)
(hdev : fullOverlapDeviation m t η σ σ') :
(∑ U : Finset (Fin n), revealProb s U *
(if ¬ revealedOverlapH m t s η U (fun v => σ v.1) (fun v => σ' v.1) then 1 else 0))
≤ 2*Real.exp (-(s*η^2/12)*n) := by sorrySource
Chen–Wang–Zhang, Zero-Freeness for Random Regular Graphs up to the Replica Symmetry Breaking Threshold, https://github.com/sigact-stoc/stoc2027/releases/download/papers-1/1461.pdf, p. 39, two Bernstein inequalities in the proof of Lemma 5.6.