Lemma 3: Holevo information of direct-sum mixtures
ProvedQAlgorithms.Papers.Independent.Gap48.P46.direct_sum_informationAll quantum systems are nonzero finite-dimensional complex Hilbert spaces. States are positive semidefinite trace-one matrices, and channels are CPTP maps on the full matrix algebra. All entropies and logarithms are in bits. An ensemble is a finite probability distribution of states, with zero weights permitted. For a channel , write for the maximum ensemble Holevo information at mean , for the minimum average output entropy at that mean, and for the maximum of over a nonempty closed mean constraint .
Let be a nonempty finite family of channels with a common input and possibly different output spaces, and let with . Put . For every fixed input ensemble ,
Here each ensemble chi is its output-mean entropy minus its average output entropy. This is an ensemble identity, allowing , rather than an identity between independently optimized ensembles.
import Definitions.Def_QAlgorithms_Papers_Independent_Gap48_P46
noncomputable section
namespace QAlgorithms.Papers.Independent.Gap48.P46
open scoped Matrix ComplexConjugate ENNReal ComplexOrder
open Filter
attribute [local instance] Classical.propDecidable
attribute [local instance] Classical.decEq
open QAlgorithms.Papers.Independent.Information100.P62 (tensorChannel)
open QAlgorithms.Papers.Independent.Information100.P54 (relativeEntropyBits)
open QAlgorithms.Papers.Independent.Dependency50.P23 (IsEB)
open QAlgorithms.Papers.Independent.Information100.P08
(Ensemble ensembleAverage outputHolevo mixedRoof)
variable {I O J K : Type} [Fintype I] [Fintype O] [Fintype J] [Fintype K]
variable [DecidableEq I] [DecidableEq O] [DecidableEq J] [DecidableEq K]
/-- PDF p.8, Lemma 3: identity for a fixed ensemble, without separate branch
optimization. Branch output spaces may differ and weights may vanish. -/
theorem direct_sum_information [Nonempty I] {T : Type} [Fintype T] [DecidableEq T]
[Nonempty T] {B : T → Type} [∀ t, Fintype (B t)] [∀ t, DecidableEq (B t)]
[∀ t, Nonempty (B t)] (q : T → ℝ) (hq : ∀ t, 0 ≤ q t) (hsum : ∑ t, q t = 1)
(Phi : (t : T) → Superop I (B t)) (hPhi : ∀ t, IsChannel (Phi t)) (e : Ensemble I) :
outputHolevo (flaggedChannelSum q Phi) e = ∑ t, q t * outputHolevo (Phi t) e := by sorry
end QAlgorithms.Papers.Independent.Gap48.P46
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Let be finite nonempty types with decidable equality, and for each , let be a finite nonempty type with decidable equality. Let be real numbers with , and let be channels from matrices on to matrices on , each admitting a finite Kraus representation with Kraus adjoint-products summing to the identity. Define a map to matrices on the disjoint union by
For every finite ensemble of density matrices on , with , , and , one has
Here a density matrix is positive semidefinite with trace , and , with zero terms equal to . Both the ensemble weights and the branch weights may vanish.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.