Theorem 3 (§VIII.B): smooth min-relative entropy gives spectral inf-divergence
ProvedQAlgorithms.Papers.Independent.Support50.P21.spectralInf_smoothMinAll systems are nonzero finite-dimensional complex Hilbert spaces, and logarithms have base two.
Fix a nonzero finite-dimensional system . Let be density operators and on for . Define and
Then, in the extended real topology,
The order of limits is as displayed, with no iid, support, trace-bound-on-, or finite-rate assumptions. This identifies the spectral inf-divergence rate with the smoothed asymptotic quantity.
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]
/-- Theorem 3, PDF pp.9–10, (67): first n→infinity, then eps→0+.
Proof (78) requires arbitrarily small slack at liminf, not a new hypothesis. -/
theorem spectralInf_smoothMin (rho sigma : (k : ℕ) → Mat (PositiveTensorBasis I k))
(hrho : ∀ k, IsDensityMatrix (rho k)) (hsigma : ∀ k, (sigma k).PosSemidef) :
Tendsto (minAsymptotic rho sigma) (nhdsWithin 0 (Set.Ioi (0 : ℝ)))
(nhds (spectralInf rho sigma)) := 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 a finite index type with decidable equality. For every integer , let be complex matrices indexed by the functions , with positive semidefinite of trace and positive semidefinite, possibly zero and without trace normalization. Let denote the orthogonal projection onto the range of , define for and otherwise, and put
where and the supremum is extended-real. For , set , let be its spectral projection onto the nonnegative eigenvalues, including zero, and set . Then, in the topology of the extended reals,
The outer limit runs through strictly positive real . The families need not consist of tensor powers or satisfy any consistency across , and already has one coordinate. The final supremum is extended-real, with value if its defining set is empty.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.