Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Consequence 23

Open
QAlgorithms.Papers.Independent.Intakenext100.P56.discord_recovery_minimum

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

prove2quantum

All systems are finite-dimensional complex Hilbert spaces with nonempty finite bases. States are positive semidefinite operators of trace one; pure kets have norm one. All marginal states below are partial traces of the stated joint state. Logarithms are natural, and spectral powers vanish on the kernel, including at exponent zero.

Assume the sandwiched conditional-information parameter conjecture (5.41), including its value at one. For any state ρAB\rho_{AB}ρAB​,

D(Aˉ;B)ρ≥min⁡∑x∣φx⟩⟨φx∣=IA−log⁡F(ρAB,(EρA,φ⊗idB)(ρAB)).D(\bar A;B)_\rho\ge\min_{\sum_x|\varphi_x\rangle\langle\varphi_x|=I_A}-\log F\bigl(\rho_{AB},(\mathcal E_{\rho_A,\varphi}\otimes\mathrm{id}_B)(\rho_{AB})\bigr).D(Aˉ;B)ρ​≥∑x​∣φx​⟩⟨φx​∣=IA​min​−logF(ρAB​,(EρA​,φ​⊗idB​)(ρAB​)).

Here FFF is squared Uhlmann fidelity. With qx=⟨φx∣ρA∣φx⟩q_x=\langle\varphi_x|\rho_A|\varphi_x\rangleqx​=⟨φx​∣ρA​∣φx​⟩, the map prepares ρA1/2∣φx⟩⟨φx∣ρA1/2/qx\rho_A^{1/2}|\varphi_x\rangle\langle\varphi_x|\rho_A^{1/2}/q_xρA1/2​∣φx​⟩⟨φx​∣ρA1/2​/qx​ after outcome xxx when qx>0q_x>0qx​>0, and prepares ρA\rho_AρA​ otherwise. It is a single entanglement-breaking map. The minimum is attained; −log⁡0=+∞-\log0=+\infty−log0=+∞, and full rank is not required.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P56
Formal statement
noncomputable section

open Filter Topology

open scoped Matrix ComplexConjugate ComplexOrder

namespace QAlgorithms.Papers.Independent.Intakenext100.P56

attribute [local instance] Classical.propDecidable Classical.decEq

open QAlgorithms

open QAlgorithms.Papers.Independent.Information100.P56 (IsLaw entropy average holevo)

open QAlgorithms.Papers.Independent.Information100.P71 (supportPower)

open QAlgorithms.Papers.Independent.Dependency50.P21 (channelAdjoint petz extendedLog marginalAB)

open QAlgorithms.Papers.Independent.Information100.P21 (Op FiniteLOCC)

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

open QAlgorithms.Papers.Independent.Dependency50.P07 (bipartiteProduct)

open QAlgorithms.Papers.Independent.Information100.P67 (squaredFidelity)

open QAlgorithms.Papers.Independent.Advance50.P25 (renyiEntropyNats)

variable {A B C D E X : Type}

variable [Fintype A] [Fintype B] [Fintype C] [Fintype D] [Fintype E] [Fintype X]

variable [Nonempty A] [Nonempty B] [Nonempty C] [Nonempty D] [Nonempty E] [Nonempty X]

open scoped Matrix.Norms.L2Operator

/-- arXiv:1410.1443v3, PDF pp.22–23, Consequence 23, (5.47).
The minimum is attained; rho_A need not be invertible. -/
theorem discord_recovery_minimum (hconjecture : SandwichedParameterConjecture)
    (rho : Op (A × B)) (hrho : IsDensityMatrix rho) :
    ∃ m : ℕ, 0 < m ∧ ∃ phi : Fin m → EuclideanSpace ℂ A,
      KetPOVM phi ∧ discordRecoveryCost rho phi ≤ renyiDiscord 1 rho ∧
      ∀ n : ℕ, 0 < n → ∀ other : Fin n → EuclideanSpace ℂ A,
        KetPOVM other → discordRecoveryCost rho phi ≤ discordRecoveryCost rho other := by sorry

end QAlgorithms.Papers.Independent.Intakenext100.P56

end
Source
Rényi squashed entanglement, discord, and relative entropy differences, arXiv:1410.1443v3, PDF p. 22, Consequence 23
Read-back

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

Assume the following global parameter-monotonicity statement: for every triple of finite nonempty register types H,K,LH,K,LH,K,L, every positive semidefinite trace-one matrix ω\omegaω on H×(K×L)H\times(K\times L)H×(K×L), and all real 0<s≤t0<s\le t0<s≤t, one has I~s(ω)≤I~t(ω)\widetilde{\mathcal I}_s(\omega)\le\widetilde{\mathcal I}_t(\omega)Is​(ω)≤It​(ω). Here Pu(X)P_u(X)Pu​(X) is spectral functional calculus with scalar value 000 at 000 and xux^uxu otherwise; writing H ⁣e(X)=−Re⁡Tr⁡(Xlog⁡X)H_{\!e}(X)=-\operatorname{Re}\operatorname{Tr}(X\log X)He​(X)=−ReTr(XlogX), the value at s=1s=1s=1 is H ⁣e(ωHL)+H ⁣e(ωKL)−H ⁣e(ωL)−H ⁣e(ω)H_{\!e}(\omega_{HL})+H_{\!e}(\omega_{KL})-H_{\!e}(\omega_L)-H_{\!e}(\omega)He​(ωHL​)+He​(ωKL​)−He​(ωL​)−He​(ω), while for s≠1s\ne1s=1,

I~s(ω)=log⁡Re⁡Tr⁡Ps ⁣(L0R[IH⊗(STS)]RL0)s−1,L0=P1/2(ω),R=P(1−s)/(2s)(ωHL)⊗IK,S=IK⊗P(s−1)/(2s)(ωL),T=P(1−s)/s(ωKL),\widetilde{\mathcal I}_s(\omega)= \frac{\log\operatorname{Re}\operatorname{Tr} P_s\!\left(L_0R[I_H\otimes(STS)]RL_0\right)}{s-1}, \quad L_0=P_{1/2}(\omega),\quad R=P_{(1-s)/(2s)}(\omega_{HL})\otimes I_K,\quad S=I_K\otimes P_{(s-1)/(2s)}(\omega_L),\quad T=P_{(1-s)/s}(\omega_{KL}),Is​(ω)=s−1logReTrPs​(L0​R[IH​⊗(STS)]RL0​)​,L0​=P1/2​(ω),R=P(1−s)/(2s)​(ωHL​)⊗IK​,S=IK​⊗P(s−1)/(2s)​(ωL​),T=P(1−s)/s​(ωKL​),

with RRR acting on H,LH,LH,L. For every positive semidefinite trace-one matrix ρ\rhoρ on finite registers A×BA\times BA×B, there then exist a positive integer mmm and vectors ϕx∈CA\phi_x\in\mathbb C^Aϕx​∈CA, x<mx<mx<m, satisfying ∑x∣ϕx⟩⟨ϕx∣=IA\sum_x|\phi_x\rangle\langle\phi_x|=I_A∑x​∣ϕx​⟩⟨ϕx​∣=IA​, for which the cost Cρ(ϕ)C_\rho(\phi)Cρ​(ϕ) is at most D1(ρ)\mathcal D_1(\rho)D1​(ρ) and is no larger than Cρ(χ)C_\rho(\chi)Cρ​(χ) for every positive integer nnn and every other family χy∈CA\chi_y\in\mathbb C^Aχy​∈CA, y<ny<ny<n, satisfying the same completeness equation. To define this cost, set ρA=Tr⁡Bρ\rho_A=\operatorname{Tr}_B\rhoρA​=TrB​ρ, Px=∣ϕx⟩⟨ϕx∣P_x=|\phi_x\rangle\langle\phi_x|Px​=∣ϕx​⟩⟨ϕx​∣, qx=Re⁡Tr⁡(PxρA)q_x=\operatorname{Re}\operatorname{Tr}(P_x\rho_A)qx​=ReTr(Px​ρA​), and

Rϕ(Z)=∑xTr⁡(PxZ){ρA1/2PxρA1/2/qx,qx>0,ρA,qx≤0.Cρ(ϕ)=−ℓ ⁣(F(ρ,(Rϕ⊗idB)(ρ))),\mathcal R_\phi(Z)=\sum_x\operatorname{Tr}(P_xZ) \begin{cases} \rho_A^{1/2}P_x\rho_A^{1/2}/q_x,&q_x>0,\\ \rho_A,&q_x\le0. \end{cases} \qquad C_\rho(\phi)=-\ell\!\left(F(\rho,(\mathcal R_\phi\otimes\mathrm{id}_B)(\rho))\right),Rϕ​(Z)=x∑​Tr(Px​Z){ρA1/2​Px​ρA1/2​/qx​,ρA​,​qx​>0,qx​≤0.​Cρ​(ϕ)=−ℓ(F(ρ,(Rϕ​⊗idB​)(ρ))),

where F(X,Y)=(Re⁡Tr⁡X YX)2F(X,Y)=(\operatorname{Re}\operatorname{Tr}\sqrt{\sqrt X\,Y\sqrt X})^2F(X,Y)=(ReTrX​YX​​)2, and ℓ(u)=−∞\ell(u)=-\inftyℓ(u)=−∞ for u≤0u\le0u≤0, otherwise ℓ(u)=log⁡u\ell(u)=\log uℓ(u)=logu in the extended reals. The quantity D1(ρ)\mathcal D_1(\rho)D1​(ρ) is the extended-real infimum, over all positive finite POVMs MxM_xMx​ on AAA, of H ⁣e((ΩM)HL)+H ⁣e((ΩM)KL)−H ⁣e((ΩM)L)−H ⁣e(ΩM)H_{\!e}((\Omega_M)_{HL})+H_{\!e}((\Omega_M)_{KL})-H_{\!e}((\Omega_M)_L)-H_{\!e}(\Omega_M)He​((ΩM​)HL​)+He​((ΩM​)KL​)−He​((ΩM​)L​)−He​(ΩM​), where H=(x,A)H=(x,A)H=(x,A), K=BK=BK=B, L=yL=yL=y, and ΩM((x,i),(b,y);(x′,i′),(b′,y′))\Omega_M((x,i),(b,y);(x',i'),(b',y'))ΩM​((x,i),(b,y);(x′,i′),(b′,y′)) equals ∑k,l(Mx)ikρ(k,b),(l,b′)(Mx′)i′l‾\sum_{k,l}(\sqrt{M_x})_{ik}\rho_{(k,b),(l,b')}\overline{(\sqrt{M_{x'}})_{i'l}}∑k,l​(Mx​​)ik​ρ(k,b),(l,b′)​(Mx′​​)i′l​​ when x=y,x′=y′x=y,x'=y'x=y,x′=y′, and 000 otherwise. The minimizing vectors may be zero or unnormalized, and a zero fidelity gives recovery cost +∞+\infty+∞.

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