Example 6.6
ProvedQAlgorithms.Papers.Independent.Support50.P12.example_6_6Write for the unnormalized Petz Rényi entropy, , , and . Logarithms are natural and . The Petz map is .
In the diagonal algebra on , set and , and let merge the first two diagonal entries into the first. This map is CPTP, , , and for every real the input and output trace overlaps are and . Therefore
but . No stochastic map recovers both inputs. This is the explicit endpoint-preservation counterexample.
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]
/-- Example 6.6, PDF p.31: explicit CPTP Chernoff-preserving but irreversible pair.
The printed '-log psi' typo is not copied, since psi is already logarithmic. -/
theorem example_6_6 :
AlgebraCompletelyPositive (diagonalAlgebra (Fin 3)) merge ∧ AlgebraTracePreserving (diagonalAlgebra (Fin 3)) merge ∧
AlgebraDensity (diagonalAlgebra (Fin 3)) exampleRho ∧ AlgebraDensity (diagonalAlgebra (Fin 3)) exampleSigma ∧
merge exampleRho = Matrix.diagonal ![(1 : ℂ), 0, 0] ∧
merge exampleSigma = Matrix.diagonal ![(1 / 2 : ℂ), 0, 1 / 2] ∧
(∀ a : ℝ, tr (supportPower exampleRho a * supportPower exampleSigma (1 - a)) =
(2 + Real.rpow 4 a) / 6) ∧
(∀ a : ℝ, tr (supportPower (merge exampleRho) a * supportPower (merge exampleSigma) (1 - a)) =
Real.rpow 2 (a - 1)) ∧
chernoff (merge exampleRho) (merge exampleSigma) = (Real.log 2 : EReal) ∧
chernoff exampleRho exampleSigma = (Real.log 2 : EReal) ∧
petzRenyi 0 (merge exampleRho) (merge exampleSigma) = (Real.log 2 : EReal) ∧
petzRenyi 0 exampleRho exampleSigma = (Real.log 2 : EReal) ∧
petz exampleSigma merge (merge exampleRho) = Matrix.diagonal ![(1 / 3 : ℂ), 2 / 3, 0] ∧
petz exampleSigma merge (merge exampleRho) ≠ exampleRho ∧
¬ Reversible (diagonalAlgebra (Fin 3)) (diagonalAlgebra (Fin 3)) merge exampleRho exampleSigma := 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 the algebra of diagonal complex matrices, and set
For every natural , including , applying to each block of a positive semidefinite matrix indexed by gives a positive semidefinite matrix whenever all its blocks lie in ; preserves the complex trace on every element of ; and are positive semidefinite with real trace one. Moreover,
Writing for spectral powers that take value at zero for every exponent, one has, for every real ,
For positive semidefinite , define , , and , where for and otherwise. Then
The map
where is the Hilbert–Schmidt adjoint, satisfies . Finally, there is no complex-linear that recovers both and from their images, has both its range and its adjoint’s range contained in , preserves trace on , and whose adjoint is positive on and satisfies for every .
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.