Remark 4.8
OpenQAlgorithms.Papers.Independent.Support50.P12.remark_4_8Let and be concrete unital algebras on nonzero finite-dimensional complex Hilbert spaces, and let be complex linear. The adjoint is for . A substochastic map has a Schwarz-contracting adjoint; stochastic additionally means trace preserving. Let . Powers are zero on kernels and traces are not normalized.
For operator convex , let contain the positive maps monotone for on all positive pairs. This class is convex and closed under compatible composition. A co-Schwarz bound is , ; it is equivalent, with the same , to a Schwarz bound after transposing the input algebra. Co-substochastic means that the adjoint satisfies the co-Schwarz bound with . Under trace preservation of and the alternative or ,
also holds for co-substochastic maps. Every convex combination of a stochastic and a co-stochastic map belongs to . This uses the paper's specific decomposability, without replacing it by the entire CP plus co-CP cone.
import Definitions.Def_QAlgorithms_Papers_Independent_Support50_P12
noncomputable section
namespace QAlgorithms.Papers.Independent.Support50.P12
open Filter MeasureTheory Set
open QAlgorithms.Papers.Independent.Information100.P60 (Mat copies)
open QAlgorithms.Papers.Independent.Information100.P30 (MatrixMap)
open QAlgorithms.Papers.Independent.Dependency50.P28 (tr LE logNonnegativeExtended pinch)
open QAlgorithms.Papers.Independent.Dependency50.P21 (channelAdjoint spectralFunction petz relativeEntropy)
open QAlgorithms.Papers.Independent.Frontier18.P17 (hs)
open QAlgorithms.Papers.Independent.Next46.P23 (Supports)
open QAlgorithms.Papers.Independent.Support21.P03 (ExtendedConvex)
open QAlgorithms (spectralProjector)
open QAlgorithms.Learning (IsDistribution)
open QAlgorithms.Papers.Independent.Information100.P04 (absOp)
open scoped BigOperators Matrix ComplexConjugate ComplexOrder ENNReal Topology Kronecker
attribute [local instance] Classical.propDecidable Classical.decEq
open QAlgorithms.Papers.Independent.Information100.P71 (supportPower)
variable {I J K : Type} [Fintype I] [DecidableEq I] [Fintype J] [DecidableEq J]
[Fintype K] [DecidableEq K]
/-- Remark 4.8, PDF pp.20–21: convex/compositional closure, transpose reduction,
co-substochastic monotonicity, and the stated stochastic/co-stochastic decomposability. -/
theorem remark_4_8 [Nonempty I] [Nonempty J] [Nonempty K]
(A : Alg I) (B : Alg J) (C : Alg K) (f : ℝ → ℝ)
(hf : DivergenceGenerator f) (hconv : SemidefiniteOperatorConvex f) :
(∀ (Phi Psi : MatrixMap I J), AlgebraExtension A B Phi → AlgebraExtension A B Psi →
MonotoneMap A f Phi → MonotoneMap A f Psi → ∀ t ∈ Icc (0 : ℝ) 1,
MonotoneMap A f (((1 - t : ℝ) : ℂ) • Phi + (t : ℂ) • Psi)) ∧
(∀ (Phi : MatrixMap I J) (Psi : MatrixMap J K), AlgebraExtension A B Phi → AlgebraExtension B C Psi →
MonotoneMap A f Phi → MonotoneMap B f Psi → MonotoneMap A f (Psi.comp Phi)) ∧
(∀ Phi : MatrixMap I J, AlgebraExtension A B Phi → ∀ c : ℝ,
CoBound A Phi c ↔ ∃ Psi : MatrixMap I J, (∀ X ∈ A, Phi X = Psi (Xᵀ)) ∧
0 ≤ c ∧ (∀ X : Mat I, Xᵀ ∈ A → X.PosSemidef → (Psi X).PosSemidef) ∧
∀ X : Mat I, Xᵀ ∈ A → LE ((Psi X)ᴴ * Psi X) ((c : ℂ) • Psi (Xᴴ * X))) ∧
(∀ (Phi : MatrixMap I J), CoSubstochastic A B Phi → ∀ X ∈ A, ∀ Y ∈ A,
X.PosSemidef → Y.PosSemidef → tr (Phi Y) = tr Y →
(tr (Phi X) = tr X ∨ 0 ≤ recessionSlope f) → div f (Phi X) (Phi Y) ≤ div f X Y) ∧
(∀ (Phi Psi : MatrixMap I J), AlgebraStochastic A B Phi → CoStochastic A B Psi →
∀ t ∈ Icc (0 : ℝ) 1, MonotoneMap A f (((1 - t : ℝ) : ℂ) • Phi + (t : ℂ) • Psi)) := by sorry
end QAlgorithms.Papers.Independent.Support50.P12
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Let be nonempty finite index types with decidable equality, and let be unital complex star subalgebras of their respective matrix algebras. All maps quantified below are complex linear. Write and . Say that has the extension property from to if its entire ambient range lies in and the entire range of lies in . For a function , write for the conjunction of , positivity of on positive elements of , and for every ; write for the same requirements with in place of . A map is substochastic from to when it has the extension property and ; it is co-substochastic when it has the extension property and ; the corresponding stochastic and co-stochastic conditions additionally require for every . Let be continuous on , with in the extended real line at infinity, and operator convex on positive semidefinite matrices in every size. Define
Here powers vanish at zero, is the distinct Hermitian spectrum and is empty otherwise, is the orthogonal eigenspace projection, division by zero is zero, and the selected limit is the total extended real limit operation. Call a function monotone on for when it is positive on positive elements of and for every positive semidefinite . The declaration asserts that convex combinations , , are monotone whenever both maps have the extension property and are monotone; that is monotone on whenever the maps have extension properties and and are respectively monotone on and ; that, for every having extension property and every real , is equivalent to the existence of a complex-linear such that for every , , whenever and , and whenever ; that every co-substochastic satisfies for positive semidefinite with and either or ; and that every convex combination of a stochastic and a co-stochastic is monotone on . These assertions include zero and unnormalized positive matrices.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.