Lemma 10.7.1: for
ProvedWildeQIT.ZhangAudenaert.half_traceDist_eqentropyinformation-theoryp2o-gran-per-chapterp2o-plan-bookp2o-v1wilde-qit
Let and be probability distributions on a finite alphabet , let be their classical trace distance, and let . Then
where and .
The set is the optimal binary test for distinguishing from , and the identity gives the classical trace distance its operational meaning as a bias in success probability.
Differences from the source: none.
Preamble
import Mathlib import Definitions.Def_WildeQIT_FinDist import Definitions.Def_WildeQIT_classicalTraceDist
Formal statement
namespace WildeQIT.ZhangAudenaert
/-- **Lemma 10.7.1.**
Wilde, *From Classical to Quantum Shannon Theory*, arXiv:1106.1445v8, Lemma 10.7.1, (10.93),
p. 307: "Let `p` and `q` be probability distributions on a finite alphabet `𝒳`. Let
`A ≡ {x : p(x) ≥ q(x)}`. Then `(1/2)‖p − q‖₁ = p(A) − q(A)`, where `p(A) ≡ ∑_{x∈A} p(x)` and
`q(A) ≡ ∑_{x∈A} q(x)`."
**Formalization Note.** `‖p − q‖₁ = classicalTraceDist p.prob q.prob = ∑_x |p(x) − q(x)|`
(Definition 10.7.1). The set `A` is the `Finset` of letters with `q(x) ≤ p(x)`. -/
theorem half_traceDist_eq {α : Type} [Fintype α] (p q : FinDist α) :
(1 / 2) * classicalTraceDist p.prob q.prob =
∑ x ∈ Finset.univ.filter (fun x => q.prob x ≤ p.prob x), p.prob x -
∑ x ∈ Finset.univ.filter (fun x => q.prob x ≤ p.prob x), q.prob x := by sorry
end WildeQIT.ZhangAudenaert
Source
Wilde, From Classical to Quantum Shannon Theory, arXiv:1106.1445v8, Lemma 10.7.1, eq. (10.93), p. 307
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.