A note on the Ehrhard inequality

Abstract

We prove that for λ ∈ [0,1] and A, B two Borel sets in $ℝ^n$ with A convex, $Φ^{-1}(γ_n(λA + (1-λ)B)) ≥ λΦ^{-1}(γ_n(A)) + (1-λ)Φ^{-1}(γ_n(B))$, where $γ_n$ is the canonical gaussian measure in $ℝ^n$ and $Φ^{-1}$ is the inverse of the gaussian distribution

Locations

• Studia Mathematica - View - PDF