Explicit Petz mutual information and its minimizing state
OpenQAlgorithms.Papers.Independent.Intakenext100.P15.petz_mutual_information_formulaAll Hilbert spaces are finite-dimensional and nonzero. A density operator is positive semidefinite with trace one; singular states are included. All logarithms are base two. For real , denotes the sandwiched Rényi divergence and the traditional Petz divergence, with value on support failure and negative spectral powers taken on the support.
For a bipartite density , let and put , , and . Partial-trace cyclicity gives .
The quantity has the finite explicit value
Here and is a density attaining that infimum. Thus the source minimum includes singular inputs. The formula expresses state mutual information through a single partial trace and spectral root.
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P15
noncomputable section
open scoped Matrix ComplexOrder ComplexConjugate Topology
open Filter
namespace QAlgorithms.Papers.Independent.Intakenext100.P15
attribute [local instance] Classical.propDecidable
attribute [local instance] Classical.decEq
-- These are pinned accepted definitions, not other chunks from the current run.
open QAlgorithms.Papers.Independent.Bridge50.P43 (NormalizedDensity payoff)
open QAlgorithms.Papers.Independent.Dependency50.P09 (Divergence logExt sandwichedD petzD)
open QAlgorithms.Papers.Independent.Information100.P34 (choi)
open QAlgorithms.Papers.Independent.Information100.P63 (schatten)
open QAlgorithms.Papers.Independent.Information100.P71
(supportPower SupportIncluded stateRelative channelPower diamond)
/-- PDF p.15, Corollary 8, (6.20)--(6.24). The minimum is attained by the
explicit normalized Sibson state, including rank-deficient inputs. -/
theorem petz_mutual_information_formula
{A B : Type} [Fintype A] [Fintype B] [DecidableEq A] [DecidableEq B]
[Nonempty A] [Nonempty B] (rho : NormalizedDensity (A × B)) (alpha : ℝ) (ha : 1 < alpha) :
0 < sibsonZ alpha rho.1 ∧ IsDensityMatrix (sibsonState alpha rho.1) ∧
stateInfo (petzD alpha) rho.1 =
((alpha / (alpha - 1) * Real.logb 2 (sibsonZ alpha rho.1) : ℝ) : EReal) ∧
petzD alpha rho.1 (Matrix.kronecker (traceRight rho.1) (sibsonState alpha rho.1)) =
stateInfo (petzD alpha) rho.1 := by sorry
end QAlgorithms.Papers.Independent.Intakenext100.P15
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Let be finite nonempty index sets with decidable equality, let be a positive semidefinite trace-one matrix on , and let . Set , with partial traces defined by summing the corresponding diagonal index. For a positive semidefinite matrix , let denote spectral functional calculus with value at zero eigenvalues and value at nonzero eigenvalues, including when . Put
For this , define when , and otherwise define it as , where for and for , using extended-real arithmetic. Define
where ranges over matrices on . The assertion is that , that is positive semidefinite with trace , and that
The real expression in the first equality is regarded as an extended real number, and singular density matrices are included.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.