Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Positive Petz payoff uniformly near order one

Open
QAlgorithms.Papers.Independent.Intakenext100.P15.petz_positive_payoff_near_one

by Gohan · Oct 31, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

All 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 α>1\alpha>1α>1, D~α\widetilde D_\alphaDα​ denotes the sandwiched Rényi divergence and DαD_\alphaDα​ the traditional Petz divergence, with value +∞+\infty+∞ on support failure and negative spectral powers taken on the support.

Write ΓABN=(id⁡A⊗N)(∣Γ⟩⟨Γ∣)\Gamma^{\mathcal N}_{AB}=(\operatorname{id}_A\otimes\mathcal N)(|\Gamma\rangle\langle\Gamma|)ΓABN​=(idA​⊗N)(∣Γ⟩⟨Γ∣), where ∣Γ⟩=∑i∣i⟩∣i⟩|\Gamma\rangle=\sum_i|i\rangle|i\rangle∣Γ⟩=∑i​∣i⟩∣i⟩ is unnormalized, and ΩN(ρ)=(ρ1/2⊗IB)ΓABN(ρ1/2⊗IB)\Omega^{\mathcal N}(\rho)=(\rho^{1/2}\otimes I_B)\Gamma^{\mathcal N}_{AB}(\rho^{1/2}\otimes I_B)ΩN(ρ)=(ρ1/2⊗IB​)ΓABN​(ρ1/2⊗IB​). Its reference marginal is ρ\rhoρ and its channel-input marginal is ρT\rho^{\mathsf T}ρT. Define I~α(N)=sup⁡ρinf⁡σD~α(ΩN(ρ)∥ρ⊗σ)\widetilde I_\alpha(\mathcal N)=\sup_\rho\inf_\sigma\widetilde D_\alpha(\Omega^{\mathcal N}(\rho)\Vert\rho\otimes\sigma)Iα​(N)=supρ​infσ​Dα​(ΩN(ρ)∥ρ⊗σ) over all normalized densities. The finite attained extrema are retained as conclusions in the minimax statement.

Define I(N)=sup⁡ρinf⁡σD(ΩN(ρ)∥ρ⊗σ)I(\mathcal N)=\sup_\rho\inf_\sigma D(\Omega^{\mathcal N}(\rho)\Vert\rho\otimes\sigma)I(N)=supρ​infσ​D(ΩN(ρ)∥ρ⊗σ) with Umegaki divergence, and Iα(N)=sup⁡ρinf⁡σDα(ΩN(ρ)∥ρ⊗σ)I_\alpha(\mathcal N)=\sup_\rho\inf_\sigma D_\alpha(\Omega^{\mathcal N}(\rho)\Vert\rho\otimes\sigma)Iα​(N)=supρ​infσ​Dα​(ΩN(ρ)∥ρ⊗σ). For CPTP N\mathcal NN and real R>I(N)R>I(\mathcal N)R>I(N),

∃β∈(1,2),∀α∈(1,β),α−1α(R−Iα(N))>0.\exists\beta\in(1,2),\quad\forall\alpha\in(1,\beta),\qquad \frac{\alpha-1}{\alpha}(R-I_\alpha(\mathcal N))>0.∃β∈(1,2),∀α∈(1,β),αα−1​(R−Iα​(N))>0.

For every order in this neighborhood the Petz channel supremum is finite and attained by an input density. The neighborhood depends on N,R\mathcal N,RN,R and is uniform over all input states. Restricting it within order two preserves the near-one claim and avoids the unrestricted maximum defect of (6.6).

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P15
Formal statement
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.17, Lemma 10, (6.43). The Petz channel information is only used in
a right neighborhood of one, chosen within alpha<2. Its finite maximum is attained
in that neighborhood; no all-order Petz maximum is asserted. -/
theorem petz_positive_payoff_near_one
    {A B : Type} [Fintype A] [Fintype B] [DecidableEq A] [DecidableEq B]
    [Nonempty A] [Nonempty B] (N : Superop A B) (hN : IsChannel N)
    (R : ℝ) (hR : channelInfo stateRelative N < (R : EReal)) :
    ∃ beta : ℝ, 1 < beta ∧ beta < 2 ∧ ∀ alpha : ℝ, 1 < alpha → alpha < beta →
      0 < payoff alpha R (channelInfo (petzD alpha) N) ∧
      ∃ (v : ℝ) (rho : NormalizedDensity A),
        channelInfo (petzD alpha) N = (v : EReal) ∧ inputInfo (petzD alpha) N rho = (v : EReal) := by sorry

end QAlgorithms.Papers.Independent.Intakenext100.P15

end
Source
Multiplicativity of completely bounded ppp-norms implies a strong converse for entanglement-assisted capacity, arXiv:1310.7028v3, PDF p. 17, Lemma 10: a positive Petz payoff near one, (6.43)
Read-back

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

Let A,BA,BA,B be finite nonempty index sets with decidable equality, and let NNN be a map from complex matrices on AAA to complex matrices on BBB admitting a finite Kraus representation N(X)=∑jEjXEj†N(X)=\sum_jE_jXE_j^\daggerN(X)=∑j​Ej​XEj†​ with ∑jEj†Ej=IA\sum_jE_j^\dagger E_j=I_A∑j​Ej†​Ej​=IA​. A density matrix means a positive semidefinite matrix of trace 111. For positive semidefinite XXX, write X[t]X^{[t]}X[t] for spectral functional calculus with value 000 at zero eigenvalues and xtx^txt at nonzero eigenvalues. Let JNJ_NJN​ be the matrix on A×BA\times BA×B with (JN)(a,b),(a′,b′)=N(Ea,a′)b,b′(J_N)_{(a,b),(a',b')}=N(E_{a,a'})_{b,b'}(JN​)(a,b),(a′,b′)​=N(Ea,a′​)b,b′​, where Ea,a′E_{a,a'}Ea,a′​ is the matrix unit, and set ΩN(ρ)=(ρ[1/2]⊗IB)JN(ρ[1/2]⊗IB)\Omega_N(\rho)=(\rho^{[1/2]}\otimes I_B)J_N(\rho^{[1/2]}\otimes I_B)ΩN​(ρ)=(ρ[1/2]⊗IB​)JN​(ρ[1/2]⊗IB​). For a divergence DDD, define

ID(N,ρ)=inf⁡σ a density matrix on BD(ΩN(ρ)∥ρ⊗σ),ID(N)=sup⁡ρ a density matrix on AID(N,ρ).I_D(N,\rho)=\inf_{\sigma\text{ a density matrix on }B} D(\Omega_N(\rho)\|\rho\otimes\sigma), \qquad I_D(N)=\sup_{\rho\text{ a density matrix on }A}I_D(N,\rho).ID​(N,ρ)=σ a density matrix on Binf​D(ΩN​(ρ)∥ρ⊗σ),ID​(N)=ρ a density matrix on Asup​ID​(N,ρ).

The divergence DU(X∥Y)D_U(X\|Y)DU​(X∥Y) used in the hypothesis is +∞+\infty+∞ unless both matrices are density matrices and ker⁡Y⊆ker⁡X\ker Y\subseteq\ker XkerY⊆kerX; otherwise it equals

max⁡ ⁣{Re⁡Tr⁡[g(X)−Xlog⁡Y], 0}log⁡2,\frac{\max\!\left\{ \operatorname{Re}\operatorname{Tr}[g(X)-X\log Y],\,0\right\}}{\log2},log2max{ReTr[g(X)−XlogY],0}​,

where g(x)=xlog⁡xg(x)=x\log xg(x)=xlogx, the logarithm and ggg are applied by functional calculus, and log⁡0=g(0)=0\log0=g(0)=0log0=g(0)=0. For α>1\alpha>1α>1, Dα(X∥Y)D_\alpha(X\|Y)Dα​(X∥Y) is +∞+\infty+∞ if ker⁡Y⊈ker⁡X\ker Y\not\subseteq\ker XkerY⊆kerX, and otherwise equals (α−1)−1L(Re⁡Tr⁡[X[α]Y[1−α]])(\alpha-1)^{-1}\mathcal L(\operatorname{Re}\operatorname{Tr}[X^{[\alpha]}Y^{[1-\alpha]}])(α−1)−1L(ReTr[X[α]Y[1−α]]), with L(u)=log⁡2u\mathcal L(u)=\log_2uL(u)=log2​u for u>0u>0u>0 and −∞-\infty−∞ for u≤0u\le0u≤0. For every real RRR satisfying IDU(N)<RI_{D_U}(N)<RIDU​​(N)<R, the assertion is that there exists a real β\betaβ with 1<β<21<\beta<21<β<2 such that, for every real α\alphaα with 1<α<β1<\alpha<\beta1<α<β,

0<α−1α(R−IDα(N)),0< \frac{\alpha-1}{\alpha}\bigl(R-I_{D_\alpha}(N)\bigr),0<αα−1​(R−IDα​​(N)),

and there exist a real vvv and an input density matrix ρ\rhoρ such that IDα(N)=v=IDα(N,ρ)I_{D_\alpha}(N)=v=I_{D_\alpha}(N,\rho)IDα​​(N)=v=IDα​​(N,ρ). All divergence extrema and the payoff use extended-real arithmetic; the real value and maximizing input may depend on α\alphaα.

Human review
  • Endorsed by Shuze Chen · Oct 31, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Oct 31, 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