One-shot achievability for separable-output pure-state cq channels
OpenQAlgorithms.Papers.Independent.Intakenext100.P76.oneShotCodingLet have normalized pure outputs on a separable complex Hilbert space. Let be any probability law on its measurable input alphabet, with the pure-state projector map measurable in the trace-norm Borel structure. Compute . For , define . Then a finite deterministic code with equiprobable messages and one joint POVM satisfies
The same lower bound holds for the extended supremum over all such finite codes. This is a one-shot achievability statement without an output-dimension or input-support cutoff.
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P76
noncomputable section
namespace QAlgorithms.Papers.Independent.Intakenext100.P76
open MeasureTheory Filter
open QAlgorithms.Papers.Independent.Information100.P63 (logTwo normalCDF normalQuantile)
open QAlgorithms.Papers.Independent.Information100.P54 (thermalEntropy)
open QAlgorithms.Papers.Independent.Dependency50.P39 (coherentCoefficient)
open QAlgorithms.Papers.Independent.Next46.P32 (PositiveKernel)
open scoped BigOperators ENNReal Topology
attribute [local instance] Classical.propDecidable
attribute [local instance] Classical.decEq
/-- Theorem 3, PDF p.6, (4.1)–(4.2), proof pp.6–8: one-shot coding on any
separable-output pure-state cq channel and any measurable input law. The spectral
law is a decomposition of the computed mean output, not an unrelated density.
Integer cardinalities are actual finite code witnesses; retain M-1 interference
terms when rounding. The decoder uses average error as in (3.15). -/
theorem oneShotCoding {X : Type*} {I : Type} [MeasurableSpace X] [Countable I]
(W : X → I → ℂ) (mu : Measure X) [IsProbabilityMeasure mu]
(hnorm : ∀ x, IsUnitCoefficientVector (W x))
(hmeas : ∀ a b, Measurable (fun x => Matrix.vecMulVec (W x) (star (W x)) a b))
(p : ℕ → ℝ) (hspec : HasSpectralLaw (meanOutput W mu) p)
(epsilon slack : ℝ) (he : 0 < epsilon) (he1 : epsilon < 1)
(hs : 0 < slack) (hse : slack < epsilon) :
let R := spectralInfEntropy p (epsilon - slack) - logTwo (4 * epsilon / slack ^ 2)
HasFiniteCode W epsilon R ∧ (R : EReal) ≤ logMaxCodeSize W epsilon := by sorry
end QAlgorithms.Papers.Independent.Intakenext100.P76
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Let be any measurable type, let be any countable type, let , and let be a probability measure on . Assume that for every , the family is summable with sum , and that every function is measurable. Define the matrix by . Let be nonnegative, summable, and have sum , and assume there are coefficient vectors such that each with has summable squared moduli totaling , distinct vectors with positive weights satisfy , and for all ; vectors with zero weight have no normalization or orthogonality requirement. For real numbers and , put
where , and the displayed supremum is the real supremum used by the definition. Then there exist a positive integer , an encoding , and complex matrices indexed by , such that every finite quadratic form , for finite and arbitrary , is real and nonnegative, for every , and
Moreover, , embedded in the extended real line, is at most the extended-real supremum of over all such positive finite codes whose displayed average error is at most . The encoding need not be injective.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.