§5.1 Bound 2: improved accessible-information bound
OpenQAlgorithms.Papers.Independent.Support50.P27.accessible_bound2Use normalized finite conditional QKD output data with public length distribution and maximum supported length . For each , use the uniform diagonal ensemble and its mean . Let retain these conditional diagonal Eve states and let replace them by , in both cases using equal uniform keys with length weight . Choose normalized representatives for every null conditional outcome. If and , with information in bits, then
Consequently the Eq. (15) privacy defect obeys
This reduces the length overhead in Bound 1. The theorem is the finite inequality; decay for growing families requires an appropriate rate restriction. The information premise concerns , rather than the generally different actual key-pair prior.
import Definitions.Def_QAlgorithms_Papers_Independent_Support50_P27
noncomputable section
namespace QAlgorithms.Papers.Independent.Support50.P27
open scoped BigOperators Matrix ComplexConjugate ComplexOrder
open QAlgorithms.Papers.Independent.Dependency50.P07
(CQData ensembleMean ensembleInformation accessibleInformation)
open QAlgorithms.Papers.Independent.Dependency50.P24 (Key)
open QAlgorithms.Papers.Independent.Dependency50.P47 (cqMatrix)
attribute [local instance] Classical.decEq Classical.propDecidable
/-- PDF pp.12--13, Section 5.1, Bound 2: full hybrid trace norm bounded by
2^(max(m)/2+1) sqrt(mu2), written as 2 sqrt(2^max(m)) sqrt(mu2).
The source's false printed Lemma 2 is NOT a hypothesis. The finite bound
survives via label-versus-mean Pinsker; no rate-uniform decay or conversion
from the actual-prior privacy condition is asserted. -/
theorem accessible_bound2 {L : ℕ} {E : Type} [Fintype E] [Nonempty E]
(D : KeyData L E) (mu2 : ℝ) (hmu2 : 0 ≤ mu2)
(hprivacy : diagonalAccessible D ≤ mu2) :
traceNorm (firstHybrid D - secondHybrid D) ≤
(∑ m, D.lengthProb m * (2 * Real.sqrt ((2 : ℝ) ^ m.val)) *
Real.sqrt (accessibleInformation (diagonalEnsemble D m))) ∧
(∑ m, D.lengthProb m * (2 * Real.sqrt ((2 : ℝ) ^ m.val)) *
Real.sqrt (accessibleInformation (diagonalEnsemble D m))) ≤
2 * Real.sqrt ((2 : ℝ) ^ maxKeyLength D) * Real.sqrt mu2 ∧
composablePrivacy D ≤
Real.sqrt ((2 : ℝ) ^ maxKeyLength D) * Real.sqrt mu2 := by sorry
end QAlgorithms.Papers.Independent.Support50.P27
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
For every natural number , finite nonempty type , data comprising nonnegative weights summing to , probability laws on , and positive semidefinite, trace-one matrices on , put , , and . Define as the real supremum over all finite families with at least one outcome and , of , where , , , and with zero summands interpreted as . Define block diagonal matrices , indexed by , whose blocks at are and , respectively, when , and otherwise. For every real satisfying , all three inequalities hold:
Here , using continuous functional calculus. Length , zero weights, and are included; the diagonal ensembles use uniform probabilities independently of .
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.