Strict mixed difference for the log-product potential
ProvedZeroFreeRRG.Hardcore.log_product_strict_mixed_differenceLet be nonnegative on a finite set, with strict inequalities at distinct indices , and let . If every other is strictly positive, then the mixed difference of between and the two-coordinate increase is strictly positive. This is the positive-probability strictness step in the hardcore Bethe comparison.
Preamble
import Mathlib open scoped BigOperators
Formal statement
namespace ZeroFreeRRG.Hardcore
theorem log_product_strict_mixed_difference {ι : Type} [Fintype ι] [DecidableEq ι] (θ : ℝ) (hθ : 0 < θ)
(x y : ι → ℝ) (hx : ∀ i, 0 ≤ x i) (hxy : ∀ i, x i ≤ y i)
(i j : ι) (hij : i ≠ j) (hi : x i < y i) (hj : x j < y j)
(hrest : ∀ a, a ≠ i → a ≠ j → 0 < x a) :
Real.log (1+θ*∏ a, (if a ∈ ({i,j} : Finset ι) then y a else x a)) +
Real.log (1+θ*∏ a, x a) >
Real.log (1+θ*∏ a, (if a = i then y a else x a)) +
Real.log (1+θ*∏ a, (if a = j then y a else x a)) := by sorry
end ZeroFreeRRG.HardcoreSource
Chen, Wang, and Zhang, Zero-Freeness for Random Regular Graphs up to the Replica Symmetry Breaking Threshold, https://github.com/sigact-stoc/stoc2027/releases/download/papers-2/1461.pdf, proof of Lemma 4.17, printed p. 30.