§III/§VI: smooth entropy and mutual-information specializations
ProvedQAlgorithms.Papers.Independent.Support50.P21.smooth_specializationsAll systems are nonzero finite-dimensional complex Hilbert spaces, and logarithms have base two.
Let , let and be density operators, and let be a density operator. The smooth entropy substitutions are
For the original marginals , define . Only the first argument is smoothed; the comparison operator is held fixed. These are definitions with the paper's trace-norm smoothing and rank-based max-entropy convention.
import Definitions.Def_QAlgorithms_Papers_Independent_Support50_P21
noncomputable section
namespace QAlgorithms.Papers.Independent.Support50.P21
open QAlgorithms QAlgorithms.Entanglement Filter
open scoped Matrix ComplexConjugate ComplexOrder
attribute [local instance] Classical.propDecidable Classical.decEq
-- Use the accepted declarations directly; no paper-local aliases hide dimension binders.
open QAlgorithms.Papers.Independent.Claimed50.P33
(FiniteState IsAverageLocalMonotone IsAdaptiveLOCCMonotone)
open QAlgorithms.Papers.Independent.Information100.P63 (Mat)
open QAlgorithms.Papers.Independent.Dependency50.P35 (tr elog maxRelative)
open QAlgorithms.Papers.Independent.Information100.P67 (IsSubnormalized)
open QAlgorithms.Papers.Independent.Information100.P07 (support)
open QAlgorithms.Papers.Independent.Information100.P71 (supportPower SupportIncluded)
variable {I J : Type} [Fintype I] [DecidableEq I] [Fintype J] [DecidableEq J]
section Statements
variable [Nonempty I] [Nonempty J]
/-- PDF pp.4,7: literal smooth substitutions, retaining the original marginals
and the paper's rank-based max convention. These are definitional identities. -/
theorem smooth_specializations (eps : ℝ) (heps : 0 ≤ eps)
(rho : Mat I) (hrho : IsDensityMatrix rho)
(rhoAB : Mat (I × J)) (hAB : IsDensityMatrix rhoAB)
(sigmaB : Mat J) (hB : IsDensityMatrix sigmaB) :
smoothHMin eps rho = -traceSmoothMax eps rho 1 ∧
smoothHMax eps rho = -traceSmoothMin eps rho 1 ∧
smoothConditionalMin eps rhoAB sigmaB =
-traceSmoothMax eps rhoAB (Matrix.kronecker (1 : Mat I) sigmaB) ∧
smoothConditionalMax eps rhoAB sigmaB =
-traceSmoothMin eps rhoAB (Matrix.kronecker (1 : Mat I) sigmaB) ∧
smoothMutualMin eps rhoAB =
traceSmoothMin eps rhoAB (Matrix.kronecker (traceRight rhoAB) (traceLeft rhoAB)) ∧
smoothMutualMax eps rhoAB =
traceSmoothMax eps rhoAB (Matrix.kronecker (traceRight rhoAB) (traceLeft rhoAB)) := by sorry
end Statements
end QAlgorithms.Papers.Independent.Support50.P21
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
Let be finite index types with decidable equality, let , and let be positive semidefinite complex matrices of trace on , , and , respectively. For matrices on the same index type, define and , where is the orthogonal projection onto the range of , and for and otherwise. Define and as, respectively, the extended-real infimum of and supremum of over positive semidefinite satisfying and , with . The six asserted identities give the named smooth quantities by these expressions:
where and . The six symbols on the left are defined by the corresponding expressions on the right; there is no optimization over . The allowed include and values at least , and the smoothing matrices may have reduced trace or be zero.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.