Prove2Me
Navigate
DiscoverCollectionsFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

Eq.(68): constant-channel divergence form of the AEP

Open
QAlgorithms.Papers.Independent.Intakethird100.P65.constantDivergence_aep

by Gohan · Nov 1, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

All systems have finite positive dimension and fixed computational bases. Matrix order is Loewner order and logarithms are base two.

Let N:A′→B\mathcal N:A'\to BN:A′→B be CPTP and let Gn\mathcal G_nGn​ be all replacement channels from A′nA'^{n}A′n to BnB^nBn, allowing arbitrary correlated output densities. Define D(N∥M)D(\mathcal N\|\mathcal M)D(N∥M) as the supremum of output-state relative entropy over purified input densities, with support failure giving +∞+\infty+∞. Then

lim⁡ε↓0lim⁡n→∞1ninf⁡Mn∈GnDmax⁡ε(N⊗n∥Mn)=inf⁡M∈G1D(N∥M).\lim_{\varepsilon\downarrow0}\lim_{n\to\infty}\frac1n\inf_{\mathcal M_n\in\mathcal G_n}D_{\max}^{\varepsilon}(\mathcal N^{\otimes n}\|\mathcal M_n)=\inf_{\mathcal M\in\mathcal G_1}D(\mathcal N\|\mathcal M).ε↓0lim​n→∞lim​n1​Mn​∈Gn​inf​Dmaxε​(N⊗n∥Mn​)=M∈G1​inf​D(N∥M).

The optimized radii are finite, and the successive limits exist. This is the constant-channel divergence form of the AEP; it asserts nothing for a singleton alternative.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakethird100_P65
Formal statement
noncomputable section

open Filter QAlgorithms

open scoped BigOperators Matrix ComplexConjugate ComplexOrder Topology

namespace QAlgorithms.Papers.Independent.Intakethird100.P65

attribute [local instance] Classical.propDecidable Classical.decEq

open QAlgorithms.Papers.Independent.Dependency50.P09 (transposeBob)

open QAlgorithms.Papers.Independent.Dependency50.P35 (maxRelative)

open QAlgorithms.Papers.Independent.Independent2.P01 (dampingKraus damping)

open QAlgorithms.Papers.Independent.Gap48.P33 (phaseFlipChannel)

open QAlgorithms.Papers.Independent.Information100.P06 (Below)

open QAlgorithms.Papers.Independent.Information100.P34 (choi)

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

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

open QAlgorithms.Papers.Independent.Information100.P67 (IsSubnormalized purifiedDistance)

open QAlgorithms.Papers.Independent.Information100.P69 (replacer permute)

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

open QAlgorithms.Papers.Independent.Support50.P39 (quantumMutualInformation)

variable {A B C D R : Type} [Fintype A] [DecidableEq A]
  [Fintype B] [DecidableEq B] [Fintype C] [DecidableEq C]
  [Fintype D] [DecidableEq D] [Fintype R] [DecidableEq R]

variable [Nonempty A] [Nonempty B] [Nonempty C] [Nonempty D] [Nonempty R]

/-- Eq.(68), PDF p.15: all possibly correlated block constant channels.
The extended inner radius is also explicitly finite before conversion to real. -/
theorem constantDivergence_aep (N : Superop A B) (hN : IsChannel N) :
    ∃ value : ℝ, constantRelativeRadius N = (value : EReal) ∧
      (∀ e : ℝ, 0 ≤ e → ∀ n : ℕ,
        constantMaxRadius (channelPower N n) e ≠ ⊤ ∧
        constantMaxRadius (channelPower N n) e ≠ ⊥) ∧
      IteratedLimit (fun e n => (constantMaxRadius (channelPower N n) e).toReal / n) value := by sorry

end QAlgorithms.Papers.Independent.Intakethird100.P65

end
Source
Quantum Channel Simulation and the Channel's Smooth Max-Information, arXiv:1807.05354v2, PDF p. 15, Eq.(68): constant-channel divergence form of the AEP
Read-back

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

For finite types A,BA,BA,B with decidable equality and every channel N:Mat⁡(A)→Mat⁡(B)N:\operatorname{Mat}(A)\to\operatorname{Mat}(B)N:Mat(A)→Mat(B), there exists a real number vvv such that the infimum, over all density matrices σ\sigmaσ on BBB, of Dch(N∥Rσ)D_{\mathrm{ch}}(N\|\mathcal R_\sigma)Dch​(N∥Rσ​) equals the finite extended-real value vvv, where Rσ(X)=tr⁡(X)σ\mathcal R_\sigma(X)=\operatorname{tr}(X)\sigmaRσ​(X)=tr(X)σ. A density matrix is positive semidefinite with trace 111, and a channel has a finite Kraus representation whose Kraus matrices satisfy ∑jKj†Kj=I\sum_jK_j^\dagger K_j=I∑j​Kj†​Kj​=I. Here Dch(N∥M)D_{\mathrm{ch}}(N\|M)Dch​(N∥M) is the supremum over all unit vectors ψ∈CA×A\psi\in\mathbb C^{A\times A}ψ∈CA×A of the base-two relative entropy between (idA⊗N)(ψψ†)(\mathrm{id}_A\otimes N)(\psi\psi^\dagger)(idA​⊗N)(ψψ†) and (idA⊗M)(ψψ†)(\mathrm{id}_A\otimes M)(\psi\psi^\dagger)(idA​⊗M)(ψψ†); this relative entropy is +∞+\infty+∞ unless both arguments are density matrices and ker⁡σ⊆ker⁡ρ\ker\sigma\subseteq\ker\rhokerσ⊆kerρ, and otherwise is the nonnegative part of Re⁡tr⁡(ρlog⁡ρ−ρlog⁡σ)\operatorname{Re}\operatorname{tr}(\rho\log\rho-\rho\log\sigma)Retr(ρlogρ−ρlogσ), divided by log⁡2\log2log2, with the matrix functions defined by continuous functional calculus and the total real logarithm. In addition, for every real e≥0e\ge0e≥0 and every natural nnn, including n=0n=0n=0, the following extended-real infimum Rn(e)R_n(e)Rn​(e) is neither +∞+\infty+∞ nor −∞-\infty−∞:

Rn(e)=inf⁡σ density inf⁡T channel∥T−N⊗n∥⋄/2≤eDmax⁡(JT∥JRσ),Dmax⁡(X∥Y)=inf⁡{ℓ∈R:2ℓY−X⪰0}.R_n(e)=\inf_{\sigma\text{ density}}\ \inf_{\substack{T\text{ channel}\\ \|T-N^{\otimes n}\|_\diamond/2\le e}} D_{\max}(J_T\|J_{\mathcal R_\sigma}),\qquad D_{\max}(X\|Y)=\inf\{\ell\in\mathbb R:2^\ell Y-X\succeq0\}.Rn​(e)=σ densityinf​ T channel∥T−N⊗n∥⋄​/2≤e​inf​Dmax​(JT​∥JRσ​​),Dmax​(X∥Y)=inf{ℓ∈R:2ℓY−X⪰0}.

Here σ\sigmaσ is on BnB^nBn, JT((a,b),(a′,b′))=T(∣a⟩⟨a′∣)b,b′J_T((a,b),(a',b'))=T(|a\rangle\langle a'|)_{b,b'}JT​((a,b),(a′,b′))=T(∣a⟩⟨a′∣)b,b′​, tensor powers are defined by products of matrix-unit entries, and ∥F∥⋄=sup⁡∥X∥1≤1∥(idI⊗F)(X)∥1\|F\|_\diamond=\sup_{\|X\|_1\le1}\|(\mathrm{id}_I\otimes F)(X)\|_1∥F∥⋄​=sup∥X∥1​≤1​∥(idI​⊗F)(X)∥1​ for input register III, using arbitrary matrices XXX on I×II\times II×I, with ∥X∥1=Re⁡tr⁡X†X\|X\|_1=\operatorname{Re}\operatorname{tr}\sqrt{X^\dagger X}∥X∥1​=RetrX†X​. Finally, there exists a function L:R→RL:\mathbb R\to\mathbb RL:R→R such that, for every 0<e<10<e<10<e<1, Rn+1(e)/(n+1)→L(e)R_{n+1}(e)/(n+1)\to L(e)Rn+1​(e)/(n+1)→L(e) as n→∞n\to\inftyn→∞, and L(e)→vL(e)\to vL(e)→v as e↓0e\downarrow0e↓0. This is an iterated limit; the assertion gives no uniformity in eee. Extended-real infima and suprema retain their infinite values, and conversion to a real sends either infinity to 000, although the stated finiteness excludes that conversion case for Rn(e)R_n(e)Rn​(e) when e≥0e\ge0e≥0.

Human review
  • Endorsed by Shuze Chen · Nov 1, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Nov 1, 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 paper
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me