Lemma-11
OpenQAlgorithms.Papers.Independent.Gap20.P04.lemma11Let be a CPTP channel on nonzero finite-dimensional registers, with receiver and environment obtained by tracing the same Stinespring isometry. Optimize over all finite classical labels , probability distributions, and normalized pure conditional states on a finite reference and input . For the resulting state, write , , , and , with entropy in bits. Rates are nonnegative, and denotes consumed ebits. The capacity regions are the closures of positive-blocklength scaled unions in Section III; their one-use unions have no added closure.
Choose one point on each supporting face maximizing , , of the full CQ region, and one point maximizing , , of the full CE region. Include the resource endpoints and limits. Then
Here allows lowering both nonnegative rates; allows lowering and increasing consumption . A single selection can omit points on constant-slope faces, which are recovered by convex combinations; no unique or smooth parametrization is asserted.
import Definitions.Def_QAlgorithms_Papers_Independent_Gap20_P04
noncomputable section
namespace QAlgorithms.Papers.Independent.Gap20.P04
open QAlgorithms
open QAlgorithms.Papers.Independent.Claimed50.P24
(Dilation PhysicalDilation ChannelEB Hadamard EqualMax AdditiveMax
dephasingDilation dephasingGamma cloningDilation)
open QAlgorithms.Papers.Independent.Information100.P67
(receiverChannel complementaryChannel dilationPower)
open QAlgorithms.Papers.Independent.Information100.P32 (diagonal)
open QAlgorithms.Papers.Independent.Dependency50.P24 (ensembleHolevoBits)
open QAlgorithms.Papers.Independent.Gap48.P44 (cqTimeSharing ceTimeSharing)
open scoped BigOperators Matrix ComplexConjugate
attribute [local instance] Classical.propDecidable Classical.decEq
variable {I B E J D F : Type}
variable [Fintype I] [Fintype B] [Fintype E]
variable [Fintype J] [Fintype D] [Fintype F]
variable [DecidableEq I] [DecidableEq B] [DecidableEq E]
variable [DecidableEq J] [DecidableEq D] [DecidableEq F]
variable [Nonempty I] [Nonempty B] [Nonempty E]
variable [Nonempty J] [Nonempty D] [Nonempty F]
/-- PDF pp.11,21, Lemma 11: one support optimizer per weight recovers the entire
resource region by convex combinations and endpoint limits. On the boundary,
nonselected points recovered by nontrivial combinations lie on constant-slope faces.
The selection hypotheses refer to full regularized regions, not one-use objectives
for an arbitrary channel, and make no uniqueness or differentiability assertion. -/
theorem lemma11 (U : Dilation I B E) (hU : PhysicalDilation U)
(cqSelect ceSelect : ℝ → ℝ × ℝ)
(hCQ : ∀ lam : ℝ, 1 ≤ lam → CQSupport (capacityCQ U) lam (cqSelect lam))
(hCE : ∀ lam ∈ Set.Icc (0 : ℝ) 1,
CESupport (capacityCE U) lam (ceSelect lam)) :
capacityCQ U = selectedCQHull (capacityCQ U) cqSelect ∧
capacityCE U = selectedCEHull (capacityCE U) ceSelect := by sorry
end QAlgorithms.Papers.Independent.Gap20.P04
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 let satisfy ; these nonemptiness conditions and the isometry equation are what “physical dilation” requires. Define and . Here a density matrix is positive semidefinite with trace , and , where is applied by continuous functional calculus. For any dilation , consider every family consisting of positive integers , probabilities with , and vectors of norm for every , including labels of probability zero. Set , , and, using the receiver and environment maps of , define
Let contain exactly the pairs with for which some such family satisfies and , and let contain exactly those with for which some such family satisfies and . For each positive integer , define , and set
Thus these capacity definitions use only positive block lengths. Let be arbitrary functions satisfying, for every real , and for all , and, for every , and for all . The endpoint parameters are included, and the functions’ values outside these parameter ranges are unrestricted. Then both equalities hold:
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.