The maximum coherent information of a retained switched channel
OpenQAlgorithms.Papers.Independent.Intakenext100.P70.switchedMaximumLet be finite nonempty registers with fixed orthonormal computational bases, and let be a completely positive trace-preserving map for each basis index of . Let and define the measured, retained switch . Then
All optimizations range over every density matrix on the stated register, and all maxima are attained. In particular there are and a density matrix for which attains the left-hand maximum. Coherent information uses entropy in bits.
Thus correlations and coherent inputs in the switch cannot improve the optimum beyond a definite switch setting.
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P70
namespace QAlgorithms.Papers.Independent.Intakenext100.P70
noncomputable section
open scoped Matrix ComplexConjugate ComplexOrder
attribute [local instance] Classical.propDecidable Classical.decEq
open QAlgorithms.Papers.Independent.Information100.P22
(Op coherentInformation singleUseInformation quantumCapacity mixed)
/-- Supplementary Lemma 1, PDF p.5, S4. Optimization on the left includes all coherent
and correlated switch inputs. The extra attainment clauses express the printed maxima,
rather than assuming maximizing density matrices as hypotheses. -/
theorem switchedMaximum {S A B : Type*} [Fintype S] [Fintype A] [Fintype B]
[DecidableEq S] [DecidableEq A] [DecidableEq B]
[Nonempty S] [Nonempty A] [Nonempty B]
(branches : S → Superop A B) (hbranches : ∀ i, IsChannel (branches i)) :
singleUseInformation (cubittSwitched branches) =
sSup (Set.range (fun i => singleUseInformation (branches i))) ∧
(∀ i, ∃ rho : Op A, IsDensityMatrix rho ∧
coherentInformation (branches i) rho = singleUseInformation (branches i)) ∧
∃ i : S, ∃ rho : Op A, IsDensityMatrix rho ∧
IsDensityMatrix (cubittSelectedInput i rho) ∧
coherentInformation (branches i) rho = singleUseInformation (cubittSwitched branches) ∧
coherentInformation (cubittSwitched branches) (cubittSelectedInput i rho) =
singleUseInformation (cubittSwitched branches) := by sorry
end
end QAlgorithms.Papers.Independent.Intakenext100.P70
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Let be finite nonempty types, each with decidable equality, and let be maps from complex matrices over to complex matrices over , each admitting a finite Kraus representation with . Define the switched map by
and define the selected input by when , and otherwise. For a map and density matrix , write
where , the identity extension applies separately to each matrix block with fixed reference indices, and for applied by continuous functional calculus. The theorem asserts all three of the following: ; for every , there exists a positive semidefinite trace-one on with ; and there exist and a positive semidefinite trace-one on , whose selected input is also positive semidefinite with trace one, such that
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.