Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proposition 1: optimal ensemble criterion

Open
QAlgorithms.Papers.Independent.Gap48.P46.optimal_ensemble_criterion

by Goku · Oct 22, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

All quantum systems are nonzero finite-dimensional complex Hilbert spaces. States are positive semidefinite trace-one matrices, and channels are CPTP maps on the full matrix algebra. All entropies and logarithms are in bits. An ensemble is a finite probability distribution of states, with zero weights permitted. For a channel Φ\PhiΦ, write χΦ(ρ)\chi_\Phi(\rho)χΦ​(ρ) for the maximum ensemble Holevo information at mean ρ\rhoρ, H^Φ(ρ)\widehat H_\Phi(\rho)HΦ​(ρ) for the minimum average output entropy at that mean, and C‾(Φ;A)\overline C(\Phi;\mathcal A)C(Φ;A) for the maximum of χΦ\chi_\PhiχΦ​ over a nonempty closed mean constraint A\mathcal AA.

Let A\mathcal AA be nonempty, closed and convex. Let e={πi,ρi}e=\{\pi_i,\rho_i\}e={πi​,ρi​} have mean ρav∈A\rho_{\rm av}\in\mathcal Aρav​∈A. Write χΦ(e)=H(∑iπiΦ(ρi))−∑iπiH(Φ(ρi))\chi_\Phi(e)=H(\sum_i\pi_i\Phi(\rho_i))-\sum_i\pi_iH(\Phi(\rho_i))χΦ​(e)=H(∑i​πi​Φ(ρi​))−∑i​πi​H(Φ(ρi​)), and let D2D_2D2​ be quantum relative entropy, taking value +∞+\infty+∞ on support violation. Then

e is optimal⟺∑jμjD2(Φ(ωj)∥Φ(ρav))≤χΦ(e) for every ensemble {μj,ωj} whose mean lies in A.e\text{ is optimal}\quad\Longleftrightarrow\quad\sum_j\mu_jD_2(\Phi(\omega_j)\|\Phi(\rho_{\rm av}))\le\chi_\Phi(e)\ \text{for every ensemble }\{\mu_j,\omega_j\}\text{ whose mean lies in }\mathcal A.e is optimal⟺j∑​μj​D2​(Φ(ωj​)∥Φ(ρav​))≤χΦ​(e) for every ensemble {μj​,ωj​} whose mean lies in A.

Zero-weight distance terms are zero. This criterion tests feasible ensemble averages; individual component states need not lie in the constraint.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Gap48_P46
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Gap48.P46

open scoped Matrix ComplexConjugate ENNReal ComplexOrder

open Filter

attribute [local instance] Classical.propDecidable

attribute [local instance] Classical.decEq

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

open QAlgorithms.Papers.Independent.Information100.P54 (relativeEntropyBits)

open QAlgorithms.Papers.Independent.Dependency50.P23 (IsEB)

open QAlgorithms.Papers.Independent.Information100.P08
  (Ensemble ensembleAverage outputHolevo mixedRoof)

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

variable [DecidableEq I] [DecidableEq O] [DecidableEq J] [DecidableEq K]

/-- PDF pp.4–5, Proposition 1: optimality iff the extended distance bound holds for
every ensemble with feasible mean, including components outside the constraint. -/
theorem optimal_ensemble_criterion [Nonempty I] [Nonempty O]
    (Phi : Superop I O) (hPhi : IsChannel Phi) (A : Set (Matrix I I ℂ))
    (hA : Constraint A) (hconvex : Convex ℝ A) (e : Ensemble I) (he : (ensembleAverage e) ∈ A) :
    IsOptimalConstrainedEnsemble Phi A e ↔ ∀ f : Ensemble I, (ensembleAverage f) ∈ A →
      shorEnsembleDistance f Phi (ensembleAverage e) ≤ ENNReal.ofReal (outputHolevo Phi e) := by sorry

end QAlgorithms.Papers.Independent.Gap48.P46

end
Source
On Shor's channel extension and constrained channels, arXiv:quant-ph/0306196v2, PDF p. 4, Proposition 1: optimal ensemble criterion
Read-back

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

Let I,OI,OI,O be finite nonempty types with decidable equality, and let Φ\PhiΦ be a channel, so Φ(X)=∑jKjXKj†\Phi(X)=\sum_jK_jXK_j^\daggerΦ(X)=∑j​Kj​XKj†​ for all XXX for a finite family satisfying ∑jKj†Kj=1\sum_jK_j^\dagger K_j=1∑j​Kj†​Kj​=1. Let A\mathcal AA be a nonempty closed convex set of positive semidefinite trace-one matrices on III. An ensemble consists of finitely many density matrices ρi\rho_iρi​ with real weights pi≥0p_i\ge0pi​≥0 summing to 111; put ρˉe=∑ipiρi\bar\rho_e=\sum_i p_i\rho_iρˉ​e​=∑i​pi​ρi​ and χΦ(e)=H(Φ(ρˉe))−∑ipiH(Φ(ρi))\chi_\Phi(e)=H(\Phi(\bar\rho_e))-\sum_i p_iH(\Phi(\rho_i))χΦ​(e)=H(Φ(ρˉ​e​))−∑i​pi​H(Φ(ρi​)), where H(X)=−∑jλjlog⁡2λjH(X)=-\sum_j\lambda_j\log_2\lambda_jH(X)=−∑j​λj​log2​λj​, with zero terms equal to 000. Define χΦ(ρ)=sup⁡ρˉf=ρχΦ(f)\chi_\Phi(\rho)=\sup_{\bar\rho_f=\rho}\chi_\Phi(f)χΦ​(ρ)=supρˉ​f​=ρ​χΦ​(f) and CΦ(A)=sup⁡ρ∈AχΦ(ρ)C_\Phi(\mathcal A)=\sup_{\rho\in\mathcal A}\chi_\Phi(\rho)CΦ​(A)=supρ∈A​χΦ​(ρ). For density matrices τ,σ\tau,\sigmaτ,σ, define the extended nonnegative quantity

D2(τ∥σ)={[Re⁡tr⁡(τln⁡τ−τln⁡σ)]+ln⁡2,ker⁡σ⊆ker⁡τ,+∞,otherwise,D_2(\tau\Vert\sigma)= \begin{cases} \displaystyle \frac{\bigl[\operatorname{Re}\operatorname{tr}(\tau\ln\tau-\tau\ln\sigma)\bigr]_+}{\ln2}, &\ker\sigma\subseteq\ker\tau,\\ +\infty,&\text{otherwise}, \end{cases}D2​(τ∥σ)=⎩⎨⎧​ln2[Retr(τlnτ−τlnσ)]+​​,+∞,​kerσ⊆kerτ,otherwise,​

where [x]+=max⁡(x,0)[x]_+=\max(x,0)[x]+​=max(x,0), and the matrix functions use spectral functional calculus, assigning ln⁡0=0\ln0=0ln0=0 and 0ln⁡0=00\ln0=00ln0=0. Fix an ensemble eee with ρˉe∈A\bar\rho_e\in\mathcal Aρˉ​e​∈A. Then ρˉe∈A\bar\rho_e\in\mathcal Aρˉ​e​∈A and χΦ(e)=CΦ(A)\chi_\Phi(e)=C_\Phi(\mathcal A)χΦ​(e)=CΦ​(A) hold if and only if, for every ensemble f=(qi,σi)f=(q_i,\sigma_i)f=(qi​,σi​) with ∑iqiσi∈A\sum_iq_i\sigma_i\in\mathcal A∑i​qi​σi​∈A,

∑iqiD2 ⁣(Φ(σi) ∥ Φ(ρˉe))≤[χΦ(e)]+.\sum_iq_iD_2\!\left(\Phi(\sigma_i)\,\middle\Vert\,\Phi(\bar\rho_e)\right) \le [\chi_\Phi(e)]_+.i∑​qi​D2​(Φ(σi​)∥Φ(ρˉ​e​))≤[χΦ​(e)]+​.

This inequality is in [0,+∞][0,+\infty][0,+∞], with 0⋅(+∞)=00\cdot(+\infty)=00⋅(+∞)=0.

Human review
  • Endorsed by Shuze Chen · Oct 22, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Goku · Oct 22, 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