Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Proposition 10: n+1-variable Werner–Holevo fidelity LP

Open
QAlgorithms.Papers.Independent.Support50.P11.werner_fidelity_linear_program

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

prove2quantum

Fix integers d≥2d\ge2d≥2, n,K≥1n,K\ge1n,K≥1, α∈[0,1]\alpha\in[0,1]α∈[0,1], and independent flags s,ps,ps,p selecting NS and PPTp. Let gj,Mijg_j,M_{ij}gj​,Mij​ be the explicit coefficients of (A3) and the partial-trace formula. The operational optimum is the attained linear-program maximum

FΩs,p((W(d,α))⊗n,K)=max⁡xdn∑j=0n(nj)(1−α)n−jαjxj,F^{\Omega_{s,p}}((\mathcal W^{(d,\alpha)})^{\otimes n},K)=\max_x d^n\sum_{j=0}^n\binom nj(1-\alpha)^{n-j}\alpha^jx_j,FΩs,p​((W(d,α))⊗n,K)=xmax​dnj=0∑n​(jn​)(1−α)n−jαjxj​,

subject to 0≤xi≤d−n0\le x_i\le d^{-n}0≤xi​≤d−n for every i=0,…,ni=0,\ldots,ni=0,…,n, and

s⇒∑jgjxj=K−2,p⇒−d−nK≤∑jMijxj≤d−nK(0≤i≤n).s\Rightarrow\sum_jg_jx_j=K^{-2},\qquad p\Rightarrow-\frac{d^{-n}}K\le\sum_jM_{ij}x_j\le\frac{d^{-n}}K\quad(0\le i\le n).s⇒j∑​gj​xj​=K−2,p⇒−Kd−n​≤j∑​Mij​xj​≤Kd−n​(0≤i≤n).

A feasible maximizing vector exists and dominates every other feasible objective. Both flags impose both constraints; neither flag gives the unrestricted forward-assisted class. Integer powers retain 00=10^0=100=1 at the two endpoints of α\alphaα.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Support50_P11
Formal statement
noncomputable section

open scoped Matrix ComplexConjugate ComplexOrder ENNReal

namespace QAlgorithms.Papers.Independent.Support50.P11

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

open QAlgorithms.Papers.Independent.Information100.P21 (pt)

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

open QAlgorithms.Papers.Independent.Information100.P76 (swapMatrix)

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

open QAlgorithms.Papers.Independent.Gap20.P19 (wernerIsometry)

attribute [local instance] Classical.propDecidable

attribute [local instance] Classical.decEq

/-- PDF p.14, Proposition 10, (A4)--(A8): attained n+1-variable LP equals the
operational optimum, with NS/PPTp flags independent and neither flag giving FA. -/
theorem werner_fidelity_linear_program
    (d n K : ℕ) (hd : 2 ≤ d) (hn : 0 < n) (hK : 0 < K)
    (alpha : ℝ) (ha0 : 0 ≤ alpha) (ha1 : alpha ≤ 1) (s p : Bool) :
    ∃ x : Fin (n + 1) → ℝ, WernerLPFeasible d n K s p x ∧
      optimalCodeFidelity (flagsClass s p) (piSuperop (fun _ : Fin n => wernerHolevoChannel d alpha)) K = lpObjective d n alpha x ∧
      (∀ y, WernerLPFeasible d n K s p y → lpObjective d n alpha y ≤ lpObjective d n alpha x) := by sorry

end QAlgorithms.Papers.Independent.Support50.P11

end
Source
On the power of PPT-preserving and non-signalling codes, arXiv:1406.7142v2, PDF p. 14, Proposition 10: n+1-variable Werner–Holevo fidelity LP
Read-back

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

For every d,n,K∈Nd,n,K\in\mathbb Nd,n,K∈N satisfying d≥2d\ge2d≥2, n>0n>0n>0, and K>0K>0K>0, every real α∈[0,1]\alpha\in[0,1]α∈[0,1], and arbitrary Boolean flags s,ps,ps,p, let H={0,…,d−1}nH=\{0,\ldots,d-1\}^nH={0,…,d−1}n and N=Wα⊗nN=W_\alpha^{\otimes n}N=Wα⊗n​, where

Wα(X)=1−αd+1(tr⁡(X)I+XT)+αd−1(tr⁡(X)I−XT),N(Eaa′)bb′=∏ℓ=0n−1Wα(Eaℓaℓ′)bℓbℓ′.W_\alpha(X)=\frac{1-\alpha}{d+1}(\operatorname{tr}(X)I+X^{\mathsf T}) +\frac{\alpha}{d-1}(\operatorname{tr}(X)I-X^{\mathsf T}), \quad N(E_{aa'})_{bb'}=\prod_{\ell=0}^{n-1}W_\alpha(E_{a_\ell a'_\ell})_{b_\ell b'_\ell}.Wα​(X)=d+11−α​(tr(X)I+XT)+d−1α​(tr(X)I−XT),N(Eaa′​)bb′​=ℓ=0∏n−1​Wα​(Eaℓ​aℓ′​​)bℓ​bℓ′​​.

For 0≤i,j≤n0\le i,j\le n0≤i,j≤n, put

Gj=2−n(nj)(d+1)n−j(d−1)j,Mij=2−n∑0≤k≤min⁡(i,j)j−k≤n−i(n−ij−k)(ik)(1+d)i−k(1−d)k.G_j=2^{-n}\binom nj(d+1)^{n-j}(d-1)^j,\qquad M_{ij}=2^{-n}\sum_{\substack{0\le k\le\min(i,j)\\j-k\le n-i}} \binom{n-i}{j-k}\binom ik(1+d)^{i-k}(1-d)^k.Gj​=2−n(jn​)(d+1)n−j(d−1)j,Mij​=2−n0≤k≤min(i,j)j−k≤n−i​∑​(j−kn−i​)(ki​)(1+d)i−k(1−d)k.

Call a real vector x=(x0,…,xn)x=(x_0,\ldots,x_n)x=(x0​,…,xn​) feasible when 0≤xi≤d−n0\le x_i\le d^{-n}0≤xi​≤d−n for every iii, when s=trues=\mathrm{true}s=true implies ∑jGjxj=K−2\sum_jG_jx_j=K^{-2}∑j​Gj​xj​=K−2, and when p=truep=\mathrm{true}p=true implies −d−n/K≤∑jMijxj≤d−n/K-d^{-n}/K\le\sum_jM_{ij}x_j\le d^{-n}/K−d−n/K≤∑j​Mij​xj​≤d−n/K for every iii. Define

f(x)=dn∑j=0n(nj)(1−α)n−jαjxj.f(x)=d^n\sum_{j=0}^{n}\binom nj(1-\alpha)^{n-j}\alpha^j x_j.f(x)=dnj=0∑n​(jn​)(1−α)n−jαjxj​.

There exists a feasible xxx such that f(x)f(x)f(x) equals the following supremum of code fidelities and f(y)≤f(x)f(y)\le f(x)f(y)≤f(x) for every feasible real vector yyy. The supremum ranges over maps ZZZ from matrices on {0,…,K−1}×H\{0,\ldots,K-1\}\times H{0,…,K−1}×H to matrices on H×{0,…,K−1}H\times\{0,\ldots,K-1\}H×{0,…,K−1} satisfying these requirements: ZZZ is a channel; there is a channel E0E_0E0​ such that tr⁡2Z(X)=E0(tr⁡2X)\operatorname{tr}_2Z(X)=E_0(\operatorname{tr}_2X)tr2​Z(X)=E0​(tr2​X) for every XXX; if sss is true, there is also a channel D0D_0D0​ such that tr⁡1Z(X)=D0(tr⁡1X)\operatorname{tr}_1Z(X)=D_0(\operatorname{tr}_1X)tr1​Z(X)=D0​(tr1​X) for every XXX; and if ppp is true, Γ∘Z∘Γ\Gamma\circ Z\circ\GammaΓ∘Z∘Γ is a channel, with (ΓX)(a,b),(c,e)=X(a,e),(c,b)(\Gamma X)_{(a,b),(c,e)}=X_{(a,e),(c,b)}(ΓX)(a,b),(c,e)​=X(a,e),(c,b)​. Here a channel means a finite Kraus map ∑jKjXKj†\sum_jK_jXK_j^\dagger∑j​Kj​XKj†​ with ∑jKj†Kj=I\sum_jK_j^\dagger K_j=I∑j​Kj†​Kj​=I. The fidelity assigned to ZZZ is

Re⁡tr⁡ ⁣[φK(id⁡K⊗ΛZ)(φK)],φK=1K∑a,a′∣a,a⟩⟨a′,a′∣,ΛZ(X)bc=∑a,a′,i,j,β,γXaa′N(Eij)βγZ(E(a,β),(a′,γ))(i,b),(j,c).\operatorname{Re}\operatorname{tr}\!\left[ \varphi_K(\operatorname{id}_K\otimes\Lambda_Z)(\varphi_K)\right], \quad \varphi_K=\frac1K\sum_{a,a'}|a,a\rangle\langle a',a'|, \quad \Lambda_Z(X)_{bc} =\sum_{a,a',i,j,\beta,\gamma} X_{aa'}N(E_{ij})_{\beta\gamma} Z(E_{(a,\beta),(a',\gamma)})_{(i,b),(j,c)}.Retr[φK​(idK​⊗ΛZ​)(φK​)],φK​=K1​a,a′∑​∣a,a⟩⟨a′,a′∣,ΛZ​(X)bc​=a,a′,i,j,β,γ∑​Xaa′​N(Eij​)βγ​Z(E(a,β),(a′,γ)​)(i,b),(j,c)​.

Thus a false flag contributes no additional condition. The formulas include α=0,1\alpha=0,1α=0,1 and K=1K=1K=1, with zeroth powers interpreted as 111.

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