Theorem 2: equality in monotonicity (seed paper result)
OpenQAlgorithms.Papers.Independent.Intakethird100.P84.equalityInMonotonicityLet and be nonzero finite-dimensional complex Hilbert spaces, and let be complex-linear, trace preserving and 2-positive. Let be invertible density matrices whose images are also invertible. Density matrices are positive semidefinite with ordinary trace one. Write , using natural logarithms, and let be the computed trace adjoint. Then each of the following is separately equivalent to :
This characterizes exactly when coarse graining preserves the relative entropy of a faithful pair. No commutativity, complete positivity or unitality of is assumed.
Formalization Note The frozen relative entropy includes the correction , whose trace vanishes for trace-one densities. The adjoint is calculated from matrix units; powers are .
import Definitions.Def_QAlgorithms_Papers_Independent_Intakethird100_P84
noncomputable section
namespace QAlgorithms.Papers.Independent.Intakethird100.P84
open scoped Matrix ComplexConjugate ComplexOrder
attribute [local instance] Classical.propDecidable Classical.decEq
-- Pinned accepted definitions, inspected at their recorded source locations.
open QAlgorithms.Papers.Independent.Renew46.P35 (traceAdjoint)
open QAlgorithms.Papers.Independent.Dependency50.P21 (imaginaryPower marginalAB)
/-- Petz, arXiv:quant-ph/0209053v1, PDF p.5, Theorem 2.
Both printed characterizations are individually equivalent to entropy equality.
All four densities are invertible; T is only two-positive and trace preserving. -/
theorem equalityInMonotonicity {I J : Type*}
[Fintype I] [Fintype J] [Nonempty I] [Nonempty J]
(T : Matrix I I ℂ →ₗ[ℂ] Matrix J J ℂ)
(hT : IsTwoPositive T) (htrace : ∀ B, (T B).trace = B.trace)
(D1 D2 : Matrix I I ℂ)
(hD1 : IsDensityMatrix D1) (hD2 : IsDensityMatrix D2)
(hTD1 : IsDensityMatrix (T D1)) (hTD2 : IsDensityMatrix (T D2))
(hInvD1 : IsUnit D1) (hInvD2 : IsUnit D2)
(hInvTD1 : IsUnit (T D1)) (hInvTD2 : IsUnit (T D2)) :
((qRelEntropy D1 D2 = qRelEntropy (T D1) (T D2)) ↔
∀ t : ℝ, traceAdjoint T (imaginaryPower (T D1) t * imaginaryPower (T D2) (-t)) =
imaginaryPower D1 t * imaginaryPower D2 (-t)) ∧
((qRelEntropy D1 D2 = qRelEntropy (T D1) (T D2)) ↔
traceAdjoint T (cfc Real.log (T D1) - cfc Real.log (T D2)) =
cfc Real.log D1 - cfc Real.log D2) := by sorry
end QAlgorithms.Papers.Independent.Intakethird100.P84
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Let be nonempty finite types, and let be complex linear, trace preserving on every matrix, and two-positive, meaning that applying to each block of any positive semidefinite matrix indexed by produces a positive semidefinite matrix. Let and both be invertible positive semidefinite matrices of trace one. Write , interpret scalar functions of these matrices by spectral functional calculus, set to have eigenvalues on the eigenspaces of , and define
where is the matrix unit. The declaration asserts both equivalences
and
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.