Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Conditional single-letter and activated weak locking capacity formulas

Open
QAlgorithms.Papers.Independent.Intakenext100.P63.conditional_singleLetter_and_activation

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

prove2quantum

Assume either the global constrained minimum-output Arimoto Renyi entropy additivity conjecture (16), for every pair of finite POVMs and normalized constraint states at every order α>1\alpha>1α>1, or the uniform IID smooth conditional min-entropy relation (17): for every finite normalized ensemble, 0<ε<10<\varepsilon<10<ε<1, and slack s>0s>0s>0, eventually every collective POVM satisfies Hmin⁡ε(In∣J)≥n(Sacc(I∣E)−s)H_{\min}^{\varepsilon}(I^n|J)\ge n(S_{\rm acc}(I|E)-s)Hminε​(In∣J)≥n(Sacc​(I∣E)−s). All smoothing uses normalized states and half trace-norm distance.

For a channel N\mathcal NN with a cq complementary channel in a fixed input basis,

LW(N)=max⁡pSacc(I∣E)=LW(u)(N),L_W(\mathcal N)=\max_pS_{\rm acc}(I|E)=L_W^{(u)}(\mathcal N),LW​(N)=pmax​Sacc​(I∣E)=LW(u)​(N),

where LW(u)(N)=max⁡{px,ρx}[χ({px,N(ρx)})−Iacc({px,Nc(ρx)})]L_W^{(u)}(\mathcal N)=\max_{\{p_x,\rho_x\}}[\chi(\{p_x,\mathcal N(\rho_x)\})-I_{\rm acc}(\{p_x,\mathcal N^c(\rho_x)\})]LW(u)​(N)=max{px​,ρx​}​[χ({px​,N(ρx​)})−Iacc​({px​,Nc(ρx​)})] optimizes every finite normalized mixed ensemble. Furthermore, for any finite-dimensional channel with its fixed normalized Stinespring dilation,

L‾W(N):=sup⁡k≥1{LW(N⊗idk)−log⁡2k}=sup⁡n≥11nLW(u)(N⊗n)≥LW(N).\overline L_W(\mathcal N):=\sup_{k\ge1}\{L_W(\mathcal N\otimes\mathrm{id}_k)-\log_2k\}=\sup_{n\ge1}\frac1nL_W^{(u)}(\mathcal N^{\otimes n})\ge L_W(\mathcal N).LW​(N):=k≥1sup​{LW​(N⊗idk​)−log2​k}=n≥1sup​n1​LW(u)​(N⊗n)≥LW​(N).

All spaces are nonzero, rates are in bits, and ordinary locking capacity permits only a sublinear preshared independent uniform secret. Both conclusions retain the conjectural premise. The finite-ensemble maximum defining the upper information is attained. The activated formula charges for the noiseless factor and does not equate ordinary and activated capacities in general.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P63
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Intakenext100.P63

open QAlgorithms Filter

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

open QAlgorithms.Papers.Independent.Information100.P04 (Measurement)

open QAlgorithms.Papers.Independent.Information100.P27 (shannon labelInformation)

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

open QAlgorithms.Papers.Independent.Bridge50.P47 (SymmetricIndex universalSymmetricDilation)

open QAlgorithms.Learning (IsDistribution)

open scoped Matrix ComplexConjugate ComplexOrder

attribute [local instance] Classical.propDecidable Classical.decEq

/-- PDF pp.14–15, Theorem 15: BOTH clauses retain the global conjectural premise.
Activated keyedCapacity is distinct from ordinary weak locking keyedCapacity. -/
theorem conditional_singleLetter_and_activation
    (hconj : ConstrainedRenyiAdditivity ∨ IIDUncertainty) :
    (∀ (a b e : ℕ), 0 < a → 0 < b → 0 < e →
      ∀ V : Matrix (Fin b × Fin e) (Fin a) ℂ, Vᴴ * V = 1 →
      ∀ rho : Fin a → Mat (Fin e), CqComplement (environment V) rho →
        keyedCapacity (dilationMap V) (environment V) true = optimizedEquivocation rho ∧
        optimizedEquivocation rho = upperInformation (dilationMap V) (environment V)) ∧
    (∀ (a b e : ℕ), 0 < a → 0 < b → 0 < e →
      ∀ V : Matrix (Fin b × Fin e) (Fin a) ℂ, Vᴴ * V = 1 →
        activatedCapacity (dilationMap V) (environment V) = regularizedUpper (dilationMap V) (environment V) ∧
        keyedCapacity (dilationMap V) (environment V) true ≤ activatedCapacity (dilationMap V) (environment V) ∧
        (∃ (m : ℕ) (p : Fin m → ℝ) (sigma : Fin m → Mat (Fin a)),
          IsDistribution p ∧ (∀ i, IsDensityMatrix (sigma i)) ∧
          upperInformation (dilationMap V) (environment V) =
            labelInformation p (fun i => dilationMap V (sigma i)) - shannon p +
              accessibleEquivocation p (fun i => environment V (sigma i)))) := by sorry

end QAlgorithms.Papers.Independent.Intakenext100.P63

end
Source
Weak locking capacity of quantum channels can be much larger than private capacity, arXiv:1403.6361v3, PDF p. 15, Theorem 15: conditional single-letter and activated capacity formulas
Read-back

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

Assume at least one of the following two propositions. The first states that for all positive integers a,ba,ba,b, all natural numbers i,ji,ji,j, measurements M=(Mx)x<iM=(M_x)_{x<i}M=(Mx​)x<i​ on Ca\mathbb C^aCa and R=(Ry)y<jR=(R_y)_{y<j}R=(Ry​)y<j​ consisting of positive semidefinite effects summing to the identity, density matrices σ,τ\sigma,\tauσ,τ on these respective spaces, and every α>1\alpha>1α>1, one has Cα(M⊗R,σ⊗τ)=Cα(M,σ)+Cα(R,τ)C_\alpha(M\otimes R,\sigma\otimes\tau)=C_\alpha(M,\sigma)+C_\alpha(R,\tau)Cα​(M⊗R,σ⊗τ)=Cα​(M,σ)+Cα​(R,τ), where

Cα(M,σ)=inf⁡r∈N, qz≥0, ∑zqz=1ψz⪰0, tr⁡ψz=1, ∑zqzψz=σ−αα−1log⁡2[∑zqz(∑x[Re⁡tr⁡(Mxψz)]α)1/α].C_\alpha(M,\sigma)= \inf_{\substack{r\in\mathbb N,\ q_z\ge0,\ \sum_zq_z=1\\ \psi_z\succeq0,\ \operatorname{tr}\psi_z=1,\ \sum_zq_z\psi_z=\sigma}} -\frac{\alpha}{\alpha-1}\log_2 \left[\sum_zq_z \left(\sum_x[\operatorname{Re}\operatorname{tr}(M_x\psi_z)]^\alpha\right)^{1/\alpha}\right].Cα​(M,σ)=r∈N, qz​≥0, ∑z​qz​=1ψz​⪰0, trψz​=1, ∑z​qz​ψz​=σ​inf​−α−1α​log2​​z∑​qz​(x∑​[Retr(Mx​ψz​)]α)1/α​.

The second proposition states that for all positive integers a,ea,ea,e, every probability vector ppp on {0,…,a−1}\{0,\ldots,a-1\}{0,…,a−1}, every family of density matrices ρi\rho_iρi​ on Ce\mathbb C^eCe, every 0<ε<10<\varepsilon<10<ε<1, and every s>0s>0s>0, there exists n0∈Nn_0\in\mathbb Nn0​∈N such that for all n≥n0n\ge n_0n≥n0​ and all measurements QQQ on (Ce)⊗n(\mathbb C^e)^{\otimes n}(Ce)⊗n with a positive finite number of outcomes,

n(A(p,ρ)−s)≤Hmin⁡ε(diag⁡PQ),PQ(i,j)=(∏t<npit)Re⁡tr⁡(Qj⨂t<nρit).n\bigl(A(p,\rho)-s\bigr)\le H_{\min}^{\varepsilon}\bigl(\operatorname{diag}P_Q\bigr), \qquad P_Q(i,j)=\left(\prod_{t<n}p_{i_t}\right) \operatorname{Re}\operatorname{tr}\left(Q_j\bigotimes_{t<n}\rho_{i_t}\right).n(A(p,ρ)−s)≤Hminε​(diagPQ​),PQ​(i,j)=(t<n∏​pit​​)Retr(Qj​t<n⨂​ρit​​).

Here, for any finite ensemble, A(p,ρ)A(p,\rho)A(p,ρ) is the infimum over measurements QQQ with a positive finite number of outcomes of

−∑j∑iPijlog⁡2 ⁣(Pij∑hPhj),Pij=piRe⁡tr⁡(Qjρi),-\sum_j\sum_i P_{ij}\log_2\!\left(\frac{P_{ij}}{\sum_hP_{hj}}\right), \qquad P_{ij}=p_i\operatorname{Re}\operatorname{tr}(Q_j\rho_i),−j∑​i∑​Pij​log2​(∑h​Phj​Pij​​),Pij​=pi​Retr(Qj​ρi​),

and Hmin⁡ε(X)H_{\min}^{\varepsilon}(X)Hminε​(X) is the supremum over density matrices ω\omegaω satisfying 12∥X−ω∥tr≤ε\frac12\|X-\omega\|_{\mathrm{tr}}\le\varepsilon21​∥X−ω∥tr​≤ε of the real conversion of

sup⁡{λ∈R:∃ density matrix ζ,2−λIlabel⊗ζ−ω⪰0},\sup\left\{\lambda\in\mathbb R: \exists\text{ density matrix }\zeta,\quad 2^{-\lambda}I_{\mathrm{label}}\otimes\zeta-\omega\succeq0\right\},sup{λ∈R:∃ density matrix ζ,2−λIlabel​⊗ζ−ω⪰0},

with this inner supremum taken in the extended real line and either infinity converted to 000; ∥X∥tr=Re⁡tr⁡X†X\|X\|_{\mathrm{tr}}=\operatorname{Re}\operatorname{tr}\sqrt{X^\dagger X}∥X∥tr​=RetrX†X​. To state the conclusions, define E(ρ)=sup⁡pA(p,ρ)\mathcal E(\rho)=\sup_p A(p,\rho)E(ρ)=supp​A(p,ρ), with the supremum over probability priors, H(p)=−∑ipilog⁡2piH(p)=-\sum_i p_i\log_2p_iH(p)=−∑i​pi​log2​pi​, S(X)=−Re⁡tr⁡f(X)S(X)=-\operatorname{Re}\operatorname{tr}f(X)S(X)=−Retrf(X) for the continuous functional calculus function f(x)=xlog⁡2xf(x)=x\log_2xf(x)=xlog2​x, and χ(p,ω)=S(∑ipiωi)−∑ipiS(ωi)\chi(p,\omega)=S(\sum_i p_i\omega_i)-\sum_i p_iS(\omega_i)χ(p,ω)=S(∑i​pi​ωi​)−∑i​pi​S(ωi​). For maps N,NcN,N^cN,Nc, put

U(N,Nc)=sup⁡m∈N, pi≥0, ∑ipi=1ωi⪰0, tr⁡ωi=1[χ(p,(N(ωi))i)−H(p)+A(p,(Nc(ωi))i)].U(N,N^c)= \sup_{\substack{m\in\mathbb N,\ p_i\ge0,\ \sum_i p_i=1\\ \omega_i\succeq0,\ \operatorname{tr}\omega_i=1}} \left[\chi(p,(N(\omega_i))_i)-H(p)+A(p,(N^c(\omega_i))_i)\right].U(N,Nc)=m∈N, pi​≥0, ∑i​pi​=1ωi​⪰0, trωi​=1​sup​[χ(p,(N(ωi​))i​)−H(p)+A(p,(Nc(ωi​))i​)].

Define Cw(N,Nc)C_{\mathrm w}(N,N^c)Cw​(N,Nc) as the real supremum of nonnegative rates attained by sequences of keyed codes of lengths ℓ=n+1\ell=n+1ℓ=n+1, each with positive message and key counts mℓ,kℓm_\ell,k_\ellmℓ​,kℓ​, density matrices ωk,m\omega_{k,m}ωk,m​, and for every key a decoding measurement Dk,mD_{k,m}Dk,m​. Such sequences must satisfy log⁡2kℓ/ℓ→0\log_2k_\ell/\ell\to0log2​kℓ​/ℓ→0, average error (kℓmℓ)−1∑k,m[1−Re⁡tr⁡(Dk,mN⊗ℓ(ωk,m))]→0(k_\ell m_\ell)^{-1}\sum_{k,m}[1-\operatorname{Re}\operatorname{tr}(D_{k,m}N^{\otimes\ell}(\omega_{k,m}))]\to0(kℓ​mℓ​)−1∑k,m​[1−Retr(Dk,m​N⊗ℓ(ωk,m​))]→0, and, for every s>0s>0s>0, eventually log⁡2mℓ/ℓ≥R−s\log_2m_\ell/\ell\ge R-slog2​mℓ​/ℓ≥R−s. They must also satisfy, for every u>0u>0u>0, eventually that every finite measurement QQQ with a positive number of outcomes on the environment output admits a probability vector www with

1mℓ∑m∑j∣Re⁡tr⁡(Qj(Nc)⊗ℓ(ωˉm))−wj∣≤u,ωˉm=1kℓ∑kωk,m.\frac1{m_\ell}\sum_m\sum_j \left|\operatorname{Re}\operatorname{tr} \bigl(Q_j(N^c)^{\otimes\ell}(\bar\omega_m)\bigr)-w_j\right|\le u, \qquad \bar\omega_m=\frac1{k_\ell}\sum_k\omega_{k,m}.mℓ​1​m∑​j∑​​Retr(Qj​(Nc)⊗ℓ(ωˉm​))−wj​​≤u,ωˉm​=kℓ​1​k∑​ωk,m​.

For k>0k>0k>0, define NkN_kNk​ on the input space enlarged by Ck\mathbb C^kCk through

[Nk(X)](b,r),(b′,r′)=[N((X(a,r),(a′,r′))a,a′)]b,b′,Nkc(X)=Nc(tr⁡CkX),[N_k(X)]_{(b,r),(b',r')}= \left[N\bigl((X_{(a,r),(a',r')})_{a,a'}\bigr)\right]_{b,b'}, \qquad N_k^c(X)=N^c(\operatorname{tr}_{\mathbb C^k}X),[Nk​(X)](b,r),(b′,r′)​=[N((X(a,r),(a′,r′)​)a,a′​)]b,b′​,Nkc​(X)=Nc(trCk​X),

and set Cact(N,Nc)=sup⁡k>0[Cw(Nk,Nkc)−log⁡2k]C_{\mathrm{act}}(N,N^c)=\sup_{k>0}[C_{\mathrm w}(N_k,N_k^c)-\log_2k]Cact​(N,Nc)=supk>0​[Cw​(Nk​,Nkc​)−log2​k] and Ureg(N,Nc)=sup⁡n>0U(N⊗n,(Nc)⊗n)/nU_{\mathrm{reg}}(N,N^c)=\sup_{n>0}U(N^{\otimes n},(N^c)^{\otimes n})/nUreg​(N,Nc)=supn>0​U(N⊗n,(Nc)⊗n)/n. Under the stated disjunctive assumption, for every positive a,b,ea,b,ea,b,e and every matrix V:Ca→Cb⊗CeV:\mathbb C^a\to\mathbb C^b\otimes\mathbb C^eV:Ca→Cb⊗Ce satisfying V†V=IV^\dagger V=IV†V=I, put N(X)=tr⁡E(VXV†)N(X)=\operatorname{tr}_E(VXV^\dagger)N(X)=trE​(VXV†) and Nc(X)=tr⁡B(VXV†)N^c(X)=\operatorname{tr}_B(VXV^\dagger)Nc(X)=trB​(VXV†). For every density-matrix family ρi\rho_iρi​ satisfying Nc(∣i⟩⟨j∣)=ρiN^c(|i\rangle\langle j|)=\rho_iNc(∣i⟩⟨j∣)=ρi​ when i=ji=ji=j and 000 otherwise, the conclusions are

Cw(N,Nc)=E(ρ),E(ρ)=U(N,Nc).C_{\mathrm w}(N,N^c)=\mathcal E(\rho), \qquad \mathcal E(\rho)=U(N,N^c).Cw​(N,Nc)=E(ρ),E(ρ)=U(N,Nc).

For every such isometry VVV, without an additional condition on its complementary map, the conclusions are

Cact(N,Nc)=Ureg(N,Nc),Cw(N,Nc)≤Cact(N,Nc),C_{\mathrm{act}}(N,N^c)=U_{\mathrm{reg}}(N,N^c), \qquad C_{\mathrm w}(N,N^c)\le C_{\mathrm{act}}(N,N^c),Cact​(N,Nc)=Ureg​(N,Nc),Cw​(N,Nc)≤Cact​(N,Nc),

and there exist m∈Nm\in\mathbb Nm∈N, a probability vector ppp on {0,…,m−1}\{0,\ldots,m-1\}{0,…,m−1}, and density matrices σi\sigma_iσi​ on Ca\mathbb C^aCa such that

U(N,Nc)=χ(p,(N(σi))i)−H(p)+A(p,(Nc(σi))i).U(N,N^c)= \chi(p,(N(\sigma_i))_i)-H(p)+A(p,(N^c(\sigma_i))_i).U(N,Nc)=χ(p,(N(σi​))i​)−H(p)+A(p,(Nc(σi​))i​).

Every density matrix in these definitions is positive semidefinite with trace 111; probability weights may be zero. The natural-number indices m,rm,rm,r are not explicitly required to be positive, although their probability-vector requirements exclude zero. Logarithms and quotients are total real operations, including log⁡20=0\log_2 0=0log2​0=0 and division by zero equal to zero.

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