The Bhatia–Davis inequality
ProvedFamousTheorems.bhatia_davis_inequalityThe Bhatia–Davis inequality. Let be a random variable with almost surely and mean . Then
Bhatia and Davis proved this in 2000. Since , it sharpens Popoviciu's inequality. Equality holds exactly when takes only the values and . It is used for bounds on variances of bounded observables in statistics and in quantum mechanics.
Formalization note. Mathlib's ProbabilityTheory.variance_le_sub_mul_sub. The measure is a probability measure, is a.e. measurable, and the mean is ∫ ω, X ω ∂μ.
Preamble
import Mathlib
Formal statement
namespace FamousTheorems
open MeasureTheory
theorem bhatia_davis_inequality {Ω : Type*} {mΩ : MeasurableSpace Ω} {μ : Measure Ω} [IsProbabilityMeasure μ] {a b : ℝ} {X : Ω → ℝ}
(h : ∀ᵐ ω ∂μ, X ω ∈ Set.Icc a b) (hX : AEMeasurable X μ) :
ProbabilityTheory.variance X μ ≤ (b - ∫ ω, X ω ∂μ) * (∫ ω, X ω ∂μ - a) := by sorry
end FamousTheoremsSource
Listed in Mathlib's curated theorem manifests (docs/1000.yaml); formalized in Mathlib as `ProbabilityTheory.variance_le_sub_mul_sub`. Proof here reduces to that Mathlib result.