Classical control, conditional inputs, and the Weyl extension
DefinitionNonadditivity_ChannelEntropyholevo-additivity-gap-v100quantum-information
For finite classical labels and a finite Kraus channel family, the input density matrix is split into diagonal label blocks. Their traces are nonnegative weights summing to one; the conditional input is the normalized block when its weight is positive and a fixed density matrix at zero weight. The controlled output is the corresponding mixture. For positive output dimension , the Weyl extension adds exactly labels and satisfies , with entropy and information in natural logarithm units here.
Definition code
import Definitions.Def_Nonadditivity_AdjointPurity
import Definitions.Def_Nonadditivity_ChannelExtensions
import Definitions.Def_Nonadditivity_Channels
import Definitions.Def_Nonadditivity_ConditionalStates
import Definitions.Def_Nonadditivity_Entropy
import Definitions.Def_Nonadditivity_EntropyMixtures
import Definitions.Def_Nonadditivity_EntropyProducts
import Definitions.Def_Nonadditivity_QuantumHolevo
import Definitions.Def_Nonadditivity_StateEnsembles
import Definitions.Def_Nonadditivity_Weyl
import Mathlib.Analysis.CStarAlgebra.ContinuousFunctionalCalculus.Order
import Mathlib.Analysis.CStarAlgebra.Matrix
import Mathlib.Analysis.Convex.Jensen
import Mathlib.Analysis.Matrix.Order
import Mathlib.Analysis.Matrix.PosDef
import Mathlib.Analysis.SpecialFunctions.Complex.CircleAddChar
import Mathlib.Analysis.SpecialFunctions.Log.Basic
import Mathlib.Analysis.SpecialFunctions.Log.NegMulLog
import Mathlib.Data.Matrix.Basic
import Mathlib.Data.Real.Sqrt
import Mathlib.LinearAlgebra.Matrix.Kronecker
import Mathlib.LinearAlgebra.Matrix.Permutation
import Mathlib.LinearAlgebra.Matrix.Trace
import Mathlib.LinearAlgebra.UnitaryGroup
import Mathlib.Order.ConditionallyCompleteLattice.Basic
import Mathlib.Tactic.FieldSimp
import Mathlib.Tactic.Linarith
import Mathlib.Tactic.NormNum
import Mathlib.Tactic.Ring
/-
Copyright (c) 2026 the Nonadditivity project contributors.
All rights reserved. See COPYRIGHT.md for licensing and attribution.
-/
/-! # Entropy and Holevo information of measured-register extensions
The conditional states and their probabilities are constructed from the
actual diagonal blocks of the input. Entropy concavity is a proved theorem.
-/
noncomputable section
namespace Nonadditivity.Channels
open Entropy
open scoped BigOperators ComplexOrder Matrix
variable {ι ο κ ζ : Type*}
variable [Fintype ι] [DecidableEq ι] [Fintype ο] [DecidableEq ο]
variable [Fintype κ] [Fintype ζ] [DecidableEq ζ]
def inputBlock (ρ : DensityMatrix (ζ × ι)) (z : ζ) : Matrix ι ι ℂ :=
selector z * ρ.matrix * (selector z).conjTranspose
theorem inputBlock_positive (ρ : DensityMatrix (ζ × ι)) (z : ζ) :
(inputBlock ρ z).PosSemidef := ρ.positive.mul_mul_conjTranspose_same _
def inputWeight (ρ : DensityMatrix (ζ × ι)) (z : ζ) : ℝ := (inputBlock ρ z).trace.re
theorem inputWeight_nonneg (ρ : DensityMatrix (ζ × ι)) (z : ζ) : 0 ≤ inputWeight ρ z :=
positive_trace_re_nonneg _ (inputBlock_positive ρ z)
theorem inputWeight_sum (ρ : DensityMatrix (ζ × ι)) : ∑ z, inputWeight ρ z = 1 := by
have ht : ∑ z, (inputBlock ρ z).trace = 1 := by
calc
(∑ z, (inputBlock ρ z).trace) =
∑ z, (ρ.matrix * ((selector z).conjTranspose * selector z)).trace := by
apply Finset.sum_congr rfl
intro z _
unfold inputBlock
rw [Matrix.trace_mul_comm, ← Matrix.mul_assoc, Matrix.trace_mul_comm]
_ = (ρ.matrix * (∑ z, (selector z).conjTranspose * selector z)).trace := by
rw [Matrix.mul_sum, Matrix.trace_sum]
_ = 1 := by rw [selector_complete, Matrix.mul_one, ρ.normalized]
have h := congrArg Complex.re ht
simpa [inputWeight] using h
def conditionalInput [Nonempty ι] (ρ : DensityMatrix (ζ × ι)) (z : ζ) : DensityMatrix ι :=
normalizePositive (inputBlock ρ z) (inputBlock_positive ρ z)
theorem inputWeight_smul_conditionalInput [Nonempty ι]
(ρ : DensityMatrix (ζ × ι)) (z : ζ) :
(inputWeight ρ z : ℂ) • (conditionalInput ρ z).matrix = inputBlock ρ z :=
trace_smul_normalizePositive _ _
namespace KrausChannel
/-- Every controlled-channel output is the explicitly normalized mixture of
its branch outputs, including branches of probability zero. -/
theorem controlled_output_mixture [Nonempty ι]
(T : ζ → KrausChannel ι ο κ) (ρ : DensityMatrix (ζ × ι)) :
(controlled T).output ρ = DensityMatrix.mixture (inputWeight ρ)
(inputWeight_nonneg ρ) (inputWeight_sum ρ)
(fun z => (T z).output (conditionalInput ρ z)) := by
apply DensityMatrix.ext
simp only [output_matrix, DensityMatrix.mixture_matrix, controlled_map]
apply Finset.sum_congr rfl
intro z _
rw [← map_smul, inputWeight_smul_conditionalInput]
rfl
/-- A uniform lower output entropy bound survives measuring and discarding
the branch register. -/
theorem controlled_output_entropy_lower [Nonempty ι]
(T : ζ → KrausChannel ι ο κ) {s : ℝ}
(hentropy : ∀ z (ρ : DensityMatrix ι), s ≤ ((T z).output ρ).vonNeumann)
(ρ : DensityMatrix (ζ × ι)) : s ≤ ((controlled T).output ρ).vonNeumann := by
rw [controlled_output_mixture]
have hconc := EntropyMixtures.densityMatrix_mixture_entropy_concave
(inputWeight ρ) (inputWeight_nonneg ρ) (inputWeight_sum ρ)
(fun z => (T z).output (conditionalInput ρ z))
have hsum : s ≤ ∑ z, inputWeight ρ z * ((T z).output (conditionalInput ρ z)).vonNeumann := by
calc
s = ∑ z, inputWeight ρ z * s := by rw [← Finset.sum_mul, inputWeight_sum, one_mul]
_ ≤ _ := Finset.sum_le_sum (fun z _ =>
mul_le_mul_of_nonneg_left (hentropy z _) (inputWeight_nonneg ρ z))
exact hsum.trans hconc
theorem outputUnitary_output (T : KrausChannel ι ο κ)
(U : unitary (Matrix ο ο ℂ)) (ρ : DensityMatrix ι) :
(T.outputUnitary U).output ρ = (T.output ρ).unitaryConjugate U := by
apply DensityMatrix.ext
exact T.outputUnitary_map U ρ.matrix
theorem outputUnitary_entropy (T : KrausChannel ι ο κ)
(U : unitary (Matrix ο ο ℂ)) (ρ : DensityMatrix ι) :
((T.outputUnitary U).output ρ).vonNeumann = (T.output ρ).vonNeumann := by
rw [outputUnitary_output, DensityMatrix.unitaryConjugate_entropy]
theorem covariantExtension_entropy_lower [Nonempty ι]
(T : KrausChannel ι ο κ) (U : ζ → unitary (Matrix ο ο ℂ)) {s : ℝ}
(hentropy : ∀ ρ : DensityMatrix ι, s ≤ (T.output ρ).vonNeumann)
(ρ : DensityMatrix (ζ × ι)) :
s ≤ ((T.covariantExtension U).output ρ).vonNeumann := by
apply controlled_output_entropy_lower
intro z σ
rw [outputUnitary_entropy]
exact hentropy σ
/-- A concrete Weyl extension uses precisely `d²` classical labels. -/
def weylExtension {d : ℕ} [NeZero d] (T : KrausChannel ι (ZMod d) κ) :
KrausChannel ((ZMod d × ZMod d) × ι) (ZMod d) ((ZMod d × ZMod d) × κ) :=
T.covariantExtension Weyl.family
theorem weylExtension_holevo_lower {d : ℕ} [NeZero d] [Nonempty ι]
(T : KrausChannel ι (ZMod d) κ) (ρ : DensityMatrix ι) :
Real.log d - (T.output ρ).vonNeumann ≤ T.weylExtension.holevo := by
apply Weyl.orbit_information_lower_bound T.weylExtension.outputs (T.output ρ)
intro q
refine ⟨labelledState q ρ, ?_⟩
change ((controlled (fun q => T.outputUnitary (Weyl.family q))).output
(labelledState q ρ)) = _
rw [controlled_output_labelled, outputUnitary_output]
/-- The complete single-use Shor/Weyl equality for a concrete Kraus channel.
No minimum-attaining input or entropy-concavity premise is assumed. -/
theorem weylExtension_holevo {d : ℕ} [NeZero d] [Nonempty ι]
(T : KrausChannel ι (ZMod d) κ) :
T.weylExtension.holevo = Real.log d - T.minimumEntropy := by
apply le_antisymm
· have h := T.weylExtension.holevo_le_of_output_entropy
(T.covariantExtension_entropy_lower Weyl.family
(fun ρ => StateEnsembles.minimumEntropy_le (outputs := T.outputs) ⟨ρ, rfl⟩))
simpa only [ZMod.card] using h
· have hglb : Real.log d - T.weylExtension.holevo ≤ T.minimumEntropy := by
apply le_csInf (T.outputs_nonempty.image _)
rintro _ ⟨σ, ⟨ρ, rfl⟩, rfl⟩
have h := T.weylExtension_holevo_lower ρ
linarith
linarith
end KrausChannel
end Nonadditivity.Channels
Source
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.