§II unnumbered result: superposition need not regularize to capacity
ProvedQAlgorithms.Papers.Independent.Gap48.P39.superposition_not_completeConsider the explicit binary cq broadcast channel , with a singleton Bob system and noiseless binary Charlie. Let be the raw block-regularized, oriented three-inequality superposition region. Then
and every triple in the raw superposition union has . In fact, for every positive blocklength there are deterministic words and separate POVMs sending Charlie-only messages with zero joint decoding error. The capacities use bits per channel use. This realizes the paper's explicit limitation: superposition in the stated Bob/Charlie orientation need not become complete under regularization.
import Definitions.Def_QAlgorithms_Papers_Independent_Gap48_P39
noncomputable section
open scoped BigOperators Matrix ComplexConjugate
namespace QAlgorithms.Papers.Independent.Gap48.P39
open QAlgorithms.Papers.Independent.Information100.P21 (Op)
open QAlgorithms.Papers.Independent.Information100.P03 (PureEnsemble)
open QAlgorithms.Papers.Independent.Claimed50.P08 (cqBlocks holevoInfo)
open QAlgorithms.Papers.Independent.Dependency50.P19 (both epr mutualInformation marginalRA marginalRB)
open QAlgorithms.Papers.Independent.Information100.P58 (Distribution)
open QAlgorithms.Papers.Independent.Information100.P67 (squaredFidelity conditionalEntropy groupedChannelPower)
attribute [local instance] Classical.propDecidable Classical.decEq
variable {A B C X T E : Type}
variable [Fintype A] [DecidableEq A] [Fintype B] [DecidableEq B]
variable [Fintype C] [DecidableEq C] [Fintype X] [DecidableEq X]
variable [Fintype T] [DecidableEq T] [Fintype E] [DecidableEq E]
/-- §II p.4 explicit nonregularization limitation (supporting the seed target).
The oriented superposition union is strictly smaller than capacity: every member
has zero total rate, while Charlie alone receives one bit per use with zero error. -/
theorem superposition_not_complete :
rawRegularizedSuperposition binaryExample ⊂ cqBroadcastCapacity binaryExample ∧
(∀ r ∈ rawRegularizedSuperposition binaryExample, r.1 + r.2.1 + r.2.2 = 0) ∧
cqBroadcastAchievable binaryExample (0, 0, 1) ∧
(∀ n : ℕ, 0 < n → ∃ D : CQBroadcastCode Bool Unit Bool n 1 1 (rank n 1),
D.Good binaryExample 0) := by sorry
end QAlgorithms.Papers.Independent.Gap48.P39
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Let be the Boolean input channel with a one element Bob register and Boolean Charlie register whose output at is the basis density matrix at . The assertion is the conjunction of four statements: its raw regularized superposition set is a proper subset of its broadcast capacity set; every triple in that raw set satisfies ; the triple is broadcast achievable; and for every natural , there is a length broadcast code with one common label, one Bob label, and Charlie labels whose success probability is at least for every label combination. Here a broadcast code assigns an input word to each common, Bob, and Charlie label and has Bob measurement matrices and Charlie measurement matrices , each family consisting of positive semidefinite matrices summing to the identity; its success for is , where the word output has entries equal to the product of the corresponding single letter output entries. A nonnegative rate triple is broadcast achievable when, for every , all sufficiently large positive block lengths admit such codes with label counts and success at least for every label combination; the broadcast capacity set is the closure of these achievable triples. The raw set consists of nonnegative triples for which some , with , and some probability distribution on satisfy
where , , , are the marginals of , are block diagonal with blocks and , are the two marginals of , partial traces sum diagonal entries over the removed register, and for applied by continuous functional calculus. Proper inclusion requires both inclusion and inequality of the two sets; the final finite block assertion excludes .
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.