Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Concavity, tagged measurement affinity, and minimax

Open
QAlgorithms.Papers.Independent.Intakenext100.P63.accessible_concavity_affinity_minimax

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

prove2quantum

Fix normalized states (ρi)i∈I(\rho_i)_{i\in I}(ρi​)i∈I​ on a nonzero finite-dimensional space EEE, with a finite nonempty label set III. For a prior ppp and finite POVM QQQ, let η(p,Q)=H(I∣J)\eta(p,Q)=H(I|J)η(p,Q)=H(I∣J) for the law piTr⁡(ρiQj)p_i\operatorname{Tr}(\rho_iQ_j)pi​Tr(ρi​Qj​). Let p(0),p(1)p^{(0)},p^{(1)}p(0),p(1) be probability vectors, let 0≤λ≤10\le\lambda\le10≤λ≤1, and let Q(0),Q(1)Q^{(0)},Q^{(1)}Q(0),Q(1) be POVMs. Their tagged mixture has disjoint outcome sets and effects λQj(0)\lambda Q_j^{(0)}λQj(0)​ and (1−λ)Qk(1)(1-\lambda)Q_k^{(1)}(1−λ)Qk(1)​. Then

η(λp(0)+(1−λ)p(1),Q(0))≥λη(p(0),Q(0))+(1−λ)η(p(1),Q(0)),\eta(\lambda p^{(0)}+(1-\lambda)p^{(1)},Q^{(0)})\ge\lambda\eta(p^{(0)},Q^{(0)})+(1-\lambda)\eta(p^{(1)},Q^{(0)}),η(λp(0)+(1−λ)p(1),Q(0))≥λη(p(0),Q(0))+(1−λ)η(p(1),Q(0)), η(p(0),λQ(0)⊕(1−λ)Q(1))=λη(p(0),Q(0))+(1−λ)η(p(0),Q(1)),\eta(p^{(0)},\lambda Q^{(0)}\oplus(1-\lambda)Q^{(1)})=\lambda\eta(p^{(0)},Q^{(0)})+(1-\lambda)\eta(p^{(0)},Q^{(1)}),η(p(0),λQ(0)⊕(1−λ)Q(1))=λη(p(0),Q(0))+(1−λ)η(p(0),Q(1)), max⁡pSacc(p;ρ)=max⁡pmin⁡Qη(p,Q)=min⁡Qmax⁡pη(p,Q).\max_p S_{\rm acc}(p;\rho)=\max_p\min_Q\eta(p,Q)=\min_Q\max_p\eta(p,Q).pmax​Sacc​(p;ρ)=pmax​Qmin​η(p,Q)=Qmin​pmax​η(p,Q).

The extrema are attained, including the inner minimum for each prior. Entropies are in bits, with zero-probability terms contributing zero. Retaining the random branch tag is essential to the affinity assertion.

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 p.8, Lemma 5: concavity in priors, tagged-measurement affinity, and minimax.
The displayed max/min are attained; they are not replaced by unattained bounds. -/
theorem accessible_concavity_affinity_minimax {I E J K : Type}
    [Fintype I] [Nonempty I] [Fintype E] [DecidableEq E] [Nonempty E]
    [Fintype J] [Fintype K]
    (rho : I → Mat E) (hrho : ∀ i, IsDensityMatrix (rho i))
    (p0 p1 : I → ℝ) (hp0 : IsDistribution p0) (hp1 : IsDistribution p1)
    (Q : J → Mat E) (R : K → Mat E) (hQ : IsPOVM Q) (hR : IsPOVM R)
    (lambda : ℝ) (hl0 : 0 ≤ lambda) (hl1 : lambda ≤ 1) :
    (lambda * ensembleConditionalBits p0 rho Q + (1 - lambda) * ensembleConditionalBits p1 rho Q ≤
      ensembleConditionalBits (fun i => lambda * p0 i + (1 - lambda) * p1 i) rho Q) ∧
    (ensembleConditionalBits p0 rho (taggedEffects lambda Q R) =
      lambda * ensembleConditionalBits p0 rho Q + (1 - lambda) * ensembleConditionalBits p0 rho R) ∧
    (optimizedEquivocation rho = sInf {h | ∃ M : Measurement E, h = measurementWorstPrior rho M}) ∧
    (∃ p : I → ℝ, IsDistribution p ∧ accessibleEquivocation p rho = optimizedEquivocation rho) ∧
    (∀ p : I → ℝ, IsDistribution p → ∃ M : Measurement E,
      ensembleConditionalBits p rho M.effect = accessibleEquivocation p rho) ∧
    (∃ M : Measurement E, measurementWorstPrior rho M = optimizedEquivocation rho) := 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. 8, Lemma 5: concavity, tagged-measurement affinity, and minimax
Read-back

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

Let I,EI,EI,E be finite nonempty types, with decidable equality on EEE, and let J,KJ,KJ,K be finite types. Let ρi\rho_iρi​ be positive semidefinite complex matrices on EEE with trace 111, let p0,p1:I→Rp_0,p_1:I\to\mathbb Rp0​,p1​:I→R be nonnegative functions summing to 111, let Q=(Qj)j∈JQ=(Q_j)_{j\in J}Q=(Qj​)j∈J​ and R=(Rk)k∈KR=(R_k)_{k\in K}R=(Rk​)k∈K​ consist of positive semidefinite matrices summing to the identity, and let 0≤λ≤10\le\lambda\le10≤λ≤1. Write

H(p,Q)=−∑j∑iPijlog⁡2 ⁣(Pij∑hPhj),Pij=piRe⁡tr⁡(Qjρi).H(p,Q)=-\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).H(p,Q)=−j∑​i∑​Pij​log2​(∑h​Phj​Pij​​),Pij​=pi​Retr(Qj​ρi​).

Define A(p)=inf⁡MH(p,M)A(p)=\inf_M H(p,M)A(p)=infM​H(p,M), where MMM ranges over measurements with any positive finite number of outcomes, define E=sup⁡pA(p)\mathcal E=\sup_p A(p)E=supp​A(p), and define W(M)=sup⁡pH(p,M)W(M)=\sup_p H(p,M)W(M)=supp​H(p,M), with both suprema over all nonnegative priors on III summing to 111. The tagged measurement Q⊔λRQ\sqcup_\lambda RQ⊔λ​R, indexed by the disjoint union J⊔KJ\sqcup KJ⊔K, has effects λQj\lambda Q_jλQj​ and (1−λ)Rk(1-\lambda)R_k(1−λ)Rk​. The declaration asserts all of

λH(p0,Q)+(1−λ)H(p1,Q)≤H(λp0+(1−λ)p1,Q),H(p0,Q⊔λR)=λH(p0,Q)+(1−λ)H(p0,R),E=inf⁡MW(M),\lambda H(p_0,Q)+(1-\lambda)H(p_1,Q) \le H(\lambda p_0+(1-\lambda)p_1,Q), \qquad H(p_0,Q\sqcup_\lambda R) =\lambda H(p_0,Q)+(1-\lambda)H(p_0,R), \qquad \mathcal E=\inf_M W(M),λH(p0​,Q)+(1−λ)H(p1​,Q)≤H(λp0​+(1−λ)p1​,Q),H(p0​,Q⊔λ​R)=λH(p0​,Q)+(1−λ)H(p0​,R),E=Minf​W(M),

together with the existence of a probability prior ppp satisfying A(p)=EA(p)=\mathcal EA(p)=E, the existence for every probability prior ppp of a measurement MMM satisfying H(p,M)=A(p)H(p,M)=A(p)H(p,M)=A(p), and the existence of a measurement MMM satisfying W(M)=EW(M)=\mathcal EW(M)=E. All logarithms and quotients in HHH are total real operations, so zero-probability terms use log⁡20=0\log_2 0=0log2​0=0 and division by zero equal to zero. The endpoints λ=0,1\lambda=0,1λ=0,1 are included, retaining zero effects on the discarded tagged branch; although J,KJ,KJ,K may syntactically be empty, their measurement hypotheses cannot hold on nonempty EEE in that case.

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