Small planted likelihood mean under compact variational hypotheses
OpenZeroFreeRRG.Hardcore.overlap_likelihood_model_reductionFor the paper’s positive-parameter hardcore or Potts model, suppose Condition 1 holds, the cavity-law set is compact, and the Bethe and reveal-perturbation functionals are continuous on it. Then for every , some reveal probability makes the planted expected pinned log-likelihood divided by at most for all sufficiently large admissible . The explicit compactness and continuity hypotheses isolate analytic properties used in the proof of the exponential overlap lemma.
Preamble
import Mathlib import Definitions.Def_ZeroFreeRRG_Hardcore_BetheFunctional import Definitions.Def_ZeroFreeRRG_Hardcore_Moments set_option autoImplicit false open Set Filter MeasureTheory AFIsingKS ZeroFreeRRG.Hardcore open scoped Classical BigOperators
Formal statement
theorem ZeroFreeRRG.Hardcore.overlap_likelihood_model_reduction (m : Model) (hm : m.Valid) (d : ℕ) (hd : 3 ≤ d) (t : ℝ) (ht : m.PosParam t)
(hC1 : Condition1 m d t)
(hcompact : IsCompact (cavityLaws m t))
(hbethe : ContinuousOn (betheK m t d) (cavityLaws m t))
(hpert : ContinuousOn (fun ν : ProbabilityMeasure (m.Spin → ℝ) =>
(d:ℝ) * ∫ μ, (∑ a, m.pi t a * (Mmul m t μ a - 1)^2)
∂(ν : Measure (m.Spin → ℝ))) (cavityLaws m t)) :
∀ ε : ℝ, 0 < ε → ∃ s : ℝ, 0 < s ∧ s < 1 ∧ ∃ N : ℕ,
∀ n ≥ N, Even (n*d) →
(1:ℝ)/n * plantedPinnedE m t s d n (fun M σ U => pinnedLL m t M σ U) ≤ ε*s := by sorrySource
Conditional source-specific reduction in 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, proof of Lemma 5.6 and equation (29).