Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Four properties of sandwiched quasi-relative entropy

Open
QAlgorithms.Papers.Independent.Support50.P19.quasiEntropy_properties

by Goku · Oct 23, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

For positive operators A,BA,BA,B on a nonzero finite complex space and α>1\alpha>1α>1, let Q~α(A∥B)=Tr⁡(B(1−α)/(2α)AB(1−α)/(2α))α\widetilde Q_\alpha(A\Vert B)=\operatorname{Tr}(B^{(1-\alpha)/(2\alpha)}AB^{(1-\alpha)/(2\alpha)})^\alphaQ​α​(A∥B)=Tr(B(1−α)/(2α)AB(1−α)/(2α))α, with generalized inverses on the support and value +∞+\infty+∞ on support failure. For every unitary UUU, positive C,DC,DC,D on another nonzero finite space, and density operator τ\tauτ on that space,

Q~α(UAU†∥UBU†)=Q~α(A∥B),Q~α(A⊗C∥B⊗D)=Q~α(A∥B)Q~α(C∥D),Q~α(A⊗τ∥B⊗τ)=Q~α(A∥B).\begin{aligned}\widetilde Q_\alpha(UAU^\dagger\Vert UBU^\dagger)&=\widetilde Q_\alpha(A\Vert B),\\\widetilde Q_\alpha(A\otimes C\Vert B\otimes D)&=\widetilde Q_\alpha(A\Vert B)\widetilde Q_\alpha(C\Vert D),\\\widetilde Q_\alpha(A\otimes\tau\Vert B\otimes\tau)&=\widetilde Q_\alpha(A\Vert B).\end{aligned}Q​α​(UAU†∥UBU†)Q​α​(A⊗C∥B⊗D)Q​α​(A⊗τ∥B⊗τ)​=Q​α​(A∥B),=Q​α​(A∥B)Q​α​(C∥D),=Q​α​(A∥B).​

In addition, if 1<α≤21<\alpha\le21<α≤2, every finite probability distribution pxp_xpx​ and positive families Ax,BxA_x,B_xAx​,Bx​ satisfy

Q~α ⁣(∑xpxAx∥∑xpxBx)≤∑xpxQ~α(Ax∥Bx).\widetilde Q_\alpha\!\left(\sum_xp_xA_x\middle\Vert\sum_xp_xB_x\right)\le\sum_xp_x\widetilde Q_\alpha(A_x\Vert B_x).Q​α​(x∑​px​Ax​​x∑​px​Bx​)≤x∑​px​Q​α​(Ax​∥Bx​).

Products and sums use extended nonnegative arithmetic, including 0⋅∞=00\cdot\infty=00⋅∞=0. These four properties underpin data processing.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Support50_P19
Formal statement
noncomputable section

open scoped Matrix ComplexOrder ComplexConjugate ENNReal

open Filter Topology

namespace QAlgorithms.Papers.Independent.Support50.P19

attribute [local instance] Classical.propDecidable Classical.decEq

open QAlgorithms.Papers.Independent.Dependency50.P35
  (tr Dominates spectralPower elog elogNN rawTrace cq)

open QAlgorithms.Papers.Independent.Dependency50.P09 (sandwichedD petzD relativeD)

open QAlgorithms.Papers.Independent.Information100.P56 (sandwichedTrace)

open QAlgorithms.Papers.Independent.Information100.P27 (average)

open QAlgorithms.Papers.Independent.Information100.P67
  (IsIsometry receiverChannel complementaryChannel)

open QAlgorithms.Papers.Independent.Claimed50.P48 (FiniteKrausCP)

open QAlgorithms.Papers.Independent.Dependency50.P23
  (UACode uaCodeSuccess ordinaryClassicalCapacity IsEB IsHadamard)

open QAlgorithms.Papers.Independent.Information100.P08 (Ensemble)

open QAlgorithms.Papers.Independent.Information100.P71 (channelPower)

open QAlgorithms.Papers.Independent.Information100.P62 (tensorChannel)

open QAlgorithms.Papers.Independent.Information100.P63 (tensorPower)

open QAlgorithms.Papers.Independent.Claimed50.P26
  (CovariantIrreducible minOutput UnassistedExponentialConverse)

variable {I J K L : Type} [Fintype I] [Fintype J] [Fintype K] [Fintype L]

variable [DecidableEq I] [DecidableEq J] [DecidableEq K] [DecidableEq L]

/-- PDF pp.12–13, Theorem 5: all four properties. Only joint convexity needs alpha≤2;
the tensor factor tau is a normalized density operator. -/
theorem quasiEntropy_properties [Nonempty I] [Nonempty J] (a : ℝ) (ha : 1 < a)
    (A B : Matrix I I ℂ) (hA : A.PosSemidef) (hB : B.PosSemidef) :
    (∀ U : Matrix.unitaryGroup I ℂ,
      sandwichedTrace a ((U : Matrix I I ℂ) * A * (U : Matrix I I ℂ)ᴴ)
        ((U : Matrix I I ℂ) * B * (U : Matrix I I ℂ)ᴴ) = sandwichedTrace a A B) ∧
    (∀ C D : Matrix J J ℂ, C.PosSemidef → D.PosSemidef →
      sandwichedTrace a (Matrix.kronecker A C) (Matrix.kronecker B D) =
        sandwichedTrace a A B * sandwichedTrace a C D) ∧
    (∀ tau : {X : Matrix J J ℂ // IsDensityMatrix X}, sandwichedTrace a (Matrix.kronecker A tau.1) (Matrix.kronecker B tau.1) =
      sandwichedTrace a A B) ∧
    (a ≤ 2 → ∀ m : ℕ, ∀ p : Fin m → ℝ, (∀ x, 0 ≤ p x) → (∑ x, p x) = 1 →
      ∀ X Y : Fin m → Matrix I I ℂ, (∀ x, (X x).PosSemidef) → (∀ x, (Y x).PosSemidef) →
        sandwichedTrace a (average p X) (average p Y) ≤
          ∑ x, ENNReal.ofReal (p x) * sandwichedTrace a (X x) (Y x)) := by sorry

end QAlgorithms.Papers.Independent.Support50.P19

end
Source
Strong converse for the classical capacity of entanglement-breaking and Hadamard channels via a sandwiched Renyi relative entropy, arXiv:1306.1586v4, PDF p. 12, theorem5
Read-back

What the Lean code literally says, in plain math · gpt-6.1-sol

Let I,JI,JI,J be finite nonempty index sets with decidable equality, let a>1a>1a>1 be real, and let A,BA,BA,B be positive semidefinite complex matrices indexed by III, without normalization or nonzero assumptions. For positive semidefinite XXX, let X[t]X^{[t]}X[t] act on positive eigenvalues by x↦xtx\mapsto x^tx↦xt and on zero eigenvalues by zero. Define the extended nonnegative quantity

Qa(X,Y)={+∞,ker⁡Y⊈ker⁡X,max⁡ ⁣{0,Re⁡tr⁡[(Y[(1−a)/(2a)]XY[(1−a)/(2a)])[a]]},ker⁡Y⊆ker⁡X.Q_a(X,Y)= \begin{cases} +\infty,&\ker Y\not\subseteq\ker X,\\ \displaystyle\max\!\left\{0,\operatorname{Re}\operatorname{tr} \left[(Y^{[(1-a)/(2a)]}XY^{[(1-a)/(2a)]})^{[a]}\right]\right\}, &\ker Y\subseteq\ker X. \end{cases}Qa​(X,Y)={+∞,max{0,Retr[(Y[(1−a)/(2a)]XY[(1−a)/(2a)])[a]]},​kerY⊆kerX,kerY⊆kerX.​

The following all hold: for every unitary UUU indexed by III, Qa(UAU†,UBU†)=Qa(A,B)Q_a(UAU^\dagger,UBU^\dagger)=Q_a(A,B)Qa​(UAU†,UBU†)=Qa​(A,B); for every positive semidefinite pair C,DC,DC,D indexed by JJJ, Qa(A⊗C,B⊗D)=Qa(A,B)Qa(C,D)Q_a(A\otimes C,B\otimes D)=Q_a(A,B)Q_a(C,D)Qa​(A⊗C,B⊗D)=Qa​(A,B)Qa​(C,D); and for every positive semidefinite τ\tauτ indexed by JJJ with tr⁡τ=1\operatorname{tr}\tau=1trτ=1, Qa(A⊗τ,B⊗τ)=Qa(A,B)Q_a(A\otimes\tau,B\otimes\tau)=Q_a(A,B)Qa​(A⊗τ,B⊗τ)=Qa​(A,B). Furthermore, if a≤2a\le2a≤2, then for every natural number mmm, every family of real weights p0,…,pm−1≥0p_0,\ldots,p_{m-1}\ge0p0​,…,pm−1​≥0 with ∑xpx=1\sum_xp_x=1∑x​px​=1, and every pair of positive semidefinite matrix families Xx,YxX_x,Y_xXx​,Yx​ indexed by III,

Qa ⁣(∑xpxXx,∑xpxYx)≤∑xpxQa(Xx,Yx).Q_a\!\left(\sum_xp_xX_x,\sum_xp_xY_x\right) \le\sum_xp_xQ_a(X_x,Y_x).Qa​(x∑​px​Xx​,x∑​px​Yx​)≤x∑​px​Qa​(Xx​,Yx​).

Products and sums here are taken in [0,+∞][0,+\infty][0,+∞], where 0⋅(+∞)=00\cdot(+\infty)=00⋅(+∞)=0. Zero matrices and zero weights are included. Although m=0m=0m=0 is quantified over, its weight-sum hypothesis cannot hold.

Human review
  • Endorsed by Shuze Chen · Oct 24, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Goku · Oct 24, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me