Code inclusion and exponent order (8)–(9)
OpenQAlgorithms.Papers.Independent.Intakenext100.P20.blind_visible_orderLet be an arbitrary discrete label set and let be a nonzero finite-dimensional complex Hilbert space. The source consists of normalized pure states and probabilities with ; write . Each memory space is finite and nonzero. Blind encoders and decoders are CPTP; a visible encoder maps labels to normalized density operators. A blind encoder is fixed before its input label.
For positive block length , the word probabilities and states are the independent products of the source probabilities and states. Admissible code sequences satisfy . An error exponent is the supremum over admissible sequences of ; a success exponent is the infimum of . Logarithms are natural, and .
For every blind code , the visible encoder uses the same memory and decoder and has exactly the same average overlap error. Consequently,
These comparisons express inclusion of blind codes in the visible model.
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P20
noncomputable section
open scoped Matrix ENNReal ComplexOrder Topology
namespace QAlgorithms.Papers.Independent.Intakenext100.P20
open QAlgorithms.Papers.Independent.Information100.P08 (IsPure)
open QAlgorithms.Papers.Independent.Information100.P47 (IsProbability)
open QAlgorithms.Papers.Independent.Information100.P53 (SpectralData)
open QAlgorithms.Papers.Independent.Gap48.P19 (shannonNats)
open QAlgorithms.Papers.Independent.Dependency50.P28 (classicalD)
attribute [local instance] Classical.propDecidable
/-- PDF pp.2-3, code inclusion and (8)-(9): every fixed blind encoder induces
an equally sized visible encoder with identical error, hence the two exponent orders. -/
theorem blind_visible_order {L : Type*} {d : ℕ} (S : DiscretePureSource L d) (R : ℝ) :
(∀ C : BlindCode (Fin d),
∃ F : L → Matrix (Fin C.memory) (Fin C.memory) ℂ,
(∀ i, IsDensityMatrix (F i)) ∧ F = (fun i => C.encode (S.state i)) ∧
averageLoss S.weight S.state (fun i => C.decode (F i)) false =
averageLoss S.weight S.state (fun i => C.decode (C.encode (S.state i))) false) ∧
blindExponent S R false false ≤ visibleExponent S R false false ∧
visibleExponent S R false true ≤ blindExponent S R false true := by sorry
end QAlgorithms.Papers.Independent.Intakenext100.P20
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Let be any type, without a finiteness or countability assumption, let be a natural number, and let consist of a proof that , nonnegative summable weights satisfying , and matrices , each positive semidefinite with trace and expressible as for a unit vector . Let be arbitrary. A channel here means a map for some finite family satisfying . For every blind code with positive integer memory dimension , channel encoder , and channel decoder , there exists a family of density matrices on the memory with , and
Furthermore, define length- source words, for every , by weights and states . A visible code assigns each word an arbitrary memory density matrix and uses a channel decoder; a blind code uses channel encoding and decoding. For either kind of code sequence, with positive memory dimensions , restrict to sequences satisfying , and put . Define when , and otherwise. Let and be the suprema over their respective sequences of , and let and be the infima of . The asserted inequalities are
These are extended real extrema; an empty supremum is and an empty infimum is , so the statement also includes rates for which no code sequence satisfies the rate bound.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.