Theorem A.14: Compatible observables Xobs,Yobs valued in compatible Xalg,Yalg satisfy I(Xobs:Yobs)≤I(Xalg:Yobs)≤I(Xalg:Yalg).
OpenQAlgorithms.Papers.Independent.Renew46.P43.observableInformationFix one normalized ambient state. Let countable observables take values in the compatible finite intrinsic subalgebras . Classical information is defined by relative entropy of the actual joint outcome law, and hybrid information by mean posterior-state divergence; thus no difference of two infinite entropies is used.
This is the source result with its stated operational or information-theoretic domain.
import Definitions.Def_QAlgorithms_Papers_Independent_Renew46_P43
noncomputable section
namespace QAlgorithms.Papers.Independent.Renew46.P43
open scoped BigOperators Matrix ComplexConjugate ComplexOrder ENNReal Matrix.Norms.L2Operator
open Filter MeasureTheory
open QAlgorithms.Papers.Independent.Information100.P08 (IsPure)
open QAlgorithms.Papers.Independent.Information100.P06 (Below)
open QAlgorithms.Papers.Independent.Information100.P60 (copies relativeBits)
open QAlgorithms.Papers.Independent.Dependency50.P19 (count both)
open QAlgorithms.Papers.Independent.Information100.P22 (coherentInformation)
attribute [local instance] Classical.propDecidable Classical.decEq
open QAlgorithms.Papers.Independent.Claimed50.P40 (InBlocks)
attribute [local instance] QAlgorithms.Papers.Independent.Renew46.P43.matrixMeasurableSpace
/-- Theorem A.14, PDF p.72: both inequalities, using countable outcome laws and
divergence-defined hybrid information rather than infinity-minus-infinity. -/
theorem observableInformation {X Y A : IntrinsicSystem} {B D : Type*} [Countable B] [Countable D]
(E : Embedding X A) (F : Embedding Y A) (hcompat : CommutingImages E.map F.map)
(U : B → Matrix (Fin X.dim) (Fin X.dim) ℂ) (V : D → Matrix (Fin Y.dim) (Fin Y.dim) ℂ)
(hU : IsCountablePOVM U) (hV : IsCountablePOVM V)
(hUX : ∀ b, InBlocks X.block (U b)) (hVY : ∀ b, InBlocks Y.block (V b))
(rho : Matrix (Fin A.dim) (Fin A.dim) ℂ) (hrho : A.IsState rho) :
measuredInformation (fun b => E.map (U b)) (fun b => F.map (V b)) rho ≤
hybridInformation X E (fun b => F.map (V b)) rho ∧
hybridInformation X E (fun b => F.map (V b)) rho ≤
ENNReal.ofReal (intrinsicMutualInformation X Y E.map F.map rho) := by sorry
end QAlgorithms.Papers.Independent.Renew46.P43
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Let be systems with positive integer dimensions and coordinate block labels ; matrices respecting a system's blocks have zero entries between differently labelled indices. Let and be unital complex-linear embeddings which preserve the block algebras, preserve products and adjoints and are injective on those algebras, and preserve positivity under every finite amplification restricted to matrices supported within equal source-block labels. Assume their images of block-respecting matrices commute. Let be arbitrary countable types, and let and be families of positive semidefinite matrices on and , respectively, each respecting its system's blocks, such that the families have sums and , respectively. These are countable sums in the finite-dimensional matrix topology; finiteness of the outcome types is not required. Let be a positive semidefinite, trace-, block-respecting matrix on . Put , , and define the restriction map
with defined analogously, and define
For real numbers , define the extended nonnegative quantity
Also define if either matrix fails to be positive semidefinite of trace ; if both are such matrices, define it to be when , and otherwise define
Write , and set and , where the square root is given by continuous functional calculus. Then the following inequalities hold in :
The sums are the extended nonnegative sums of the displayed terms. The inverse is total, so when , and multiplication in uses .
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.