Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§7.5: alphabet cardinality sufficiency

Open
QAlgorithms.Papers.Independent.Support50.P02.alphabet_sufficient

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

prove2quantum

Let N:A′B′→C\mathcal N:A'B'\to CN:A′B′→C be a completely positive, trace-preserving channel between finite, positive-dimensional complex quantum registers. Alice and Bob control separate inputs. Codes use independent sender preparations, local encoders, one joint memoryless channel and one joint decoder, with external references untouched and no free shared resources. Fidelity is squared Uhlmann fidelity, trace distance is the full trace norm, and entropies are in bits. Write CQ(N)CQ(\mathcal N)CQ(N) and Q(N)Q(\mathcal N)Q(N) for the closed operational classical–quantum and quantum–quantum capacity regions, using joint output fidelity and nonnegative rates.

For a pure ensemble {px,ϕxA′}\{p_x,\phi_x^{A'}\}{px​,ϕxA′​} and an independent pure state ΨBB′\Psi^{BB'}ΨBB′, let σXBC=∑xpx∣x⟩⟨x∣⊗(id⁡B⊗N)(ϕx⊗Ψ)\sigma^{XBC}=\sum_xp_x|x\rangle\langle x|\otimes(\operatorname{id}_B\otimes\mathcal N)(\phi_x\otimes\Psi)σXBC=∑x​px​∣x⟩⟨x∣⊗(idB​⊗N)(ϕx​⊗Ψ). Write HHH for von Neumann entropy in bits, I(U;V)=H(U)+H(V)−H(UV)I(U;V)=H(U)+H(V)-H(UV)I(U;V)=H(U)+H(V)−H(UV), and Ic(U⟩V)=H(V)−H(UV)I_c(U\rangle V)=H(V)-H(UV)Ic​(U⟩V)=H(V)−H(UV).

Put L=min⁡{∣A′∣,∣C∣}2+1L=\min\{|A'|,|C|\}^2+1L=min{∣A′∣,∣C∣}2+1. Every such ensemble can be replaced by a pure ensemble on at most LLL letters, keeping Bob's pure pair fixed and preserving both I(X;C)I(X;C)I(X;C) and Ic(B⟩CX)I_c(B\rangle CX)Ic​(B⟩CX). If ∣A′∣≤∣C∣|A'|\le|C|∣A′∣≤∣C∣, the average Alice input can also be preserved; otherwise the average Charlie output can be preserved. Consequently,

CQ(1)(N;all finite alphabets)=CQ(1)(N;∣X∣≤L).CQ^{(1)}(\mathcal N;\text{all finite alphabets})=CQ^{(1)}(\mathcal N;|X|\le L).CQ(1)(N;all finite alphabets)=CQ(1)(N;∣X∣≤L).

This makes the alphabet optimization finite-dimensional while retaining the same one-use region.

Formalization Note The replacement may use new pure states. The corrected connected domain is the image of all pure inputs, rather than a fixed finite family.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Support50_P02
Formal statement
noncomputable section

open scoped BigOperators Matrix ComplexConjugate

namespace QAlgorithms.Papers.Independent.Support50.P02

open QAlgorithms.Papers.Independent.Information100.P67
  (squaredFidelity receiverChannel complementaryChannel IsDegradable)

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

open QAlgorithms.Papers.Independent.Information100.P03 (PureEnsemble)

open QAlgorithms.Papers.Independent.Dependency50.P19 (epr)

attribute [local instance] Classical.propDecidable Classical.decEq

variable {A B C E R S I J : Type*}

variable [Fintype A] [DecidableEq A] [Fintype B] [DecidableEq B]

variable [Fintype C] [DecidableEq C] [Fintype E] [DecidableEq E]

variable [Fintype R] [DecidableEq R] [Fintype S] [DecidableEq S]

variable [Fintype I] [DecidableEq I] [Fintype J] [DecidableEq J]

/-- Alphabet sufficiency, §7.5, PDF p.36, also Theorem 1 p.8.
Corrected domain: replacement points may be NEW pure states from the continuous
image of ALL pure inputs. The printed finite-family simplex argument does not
justify same-support sparsification. The smaller-input branch preserves average
INPUT; the smaller-output branch preserves average output. Bob's pure pair stays fixed. -/
theorem alphabet_sufficient [Nonempty A] [Nonempty B] [Nonempty C]
    (N : Superop (A × B) C) (hN : IsChannel N) :
    cqOne N none = cqOne N (some (min (Fintype.card A) (Fintype.card C) ^ 2 + 1)) ∧
    (∀ D : CQMacData A B, ∃ P : PureEnsemble A,
      P.count ≤ min (Fintype.card A) (Fintype.card C) ^ 2 + 1 ∧
      let Dnew : CQMacData A B := { D with ensemble := P }
      Dnew.classical N = D.classical N ∧
      Dnew.conditionalCoherent N = D.conditionalCoherent N ∧
      (if Fintype.card A ≤ Fintype.card C then pureEnsembleAverage P = pureEnsembleAverage D.ensemble
       else traceLeft (Dnew.averageOutput N) = traceLeft (D.averageOutput N))) := by sorry

end QAlgorithms.Papers.Independent.Support50.P02

end
Source
Capacity Theorems for Quantum Multiple Access Channels: Classical-Quantum and Quantum-Quantum Capacity Regions, arXiv:quant-ph/0501045v3, PDF p. 36, §7.5: alphabet cardinality sufficiency
Read-back

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

For arbitrary nonempty finite types A,B,CA,B,CA,B,C, each with decidable equality, and every map NNN on complex matrices from A×BA\times BA×B to CCC of the form N(X)=∑jKjXKj†N(X)=\sum_jK_jXK_j^\daggerN(X)=∑j​Kj​XKj†​ for finitely many matrices satisfying ∑jKj†Kj=I\sum_jK_j^\dagger K_j=I∑j​Kj†​Kj​=I, put L=(min⁡{∣A∣,∣C∣})2+1L=(\min\{|A|,|C|\})^2+1L=(min{∣A∣,∣C∣})2+1. Consider data DDD consisting of a finite ensemble (wx,αx)(w_x,\alpha_x)(wx​,αx​) of unit vectors on AAA, with nonnegative real weights summing to one, a positive integer kkk, and a unit vector β\betaβ on R×BR\times BR×B, where R={0,…,k−1}R=\{0,\ldots,k-1\}R={0,…,k−1}. Define

ωx=(idR⊗N)(∣αx⟩⟨αx∣⊗∣β⟩⟨β∣),ωˉ=∑xwxωx,χ(D)=S(tr⁡Rωˉ)−∑xwxS(tr⁡Rωx),J(D)=∑xwx(S(tr⁡Rωx)−S(ωx)),\omega_x=(\mathrm{id}_R\otimes N) \bigl(|\alpha_x\rangle\langle\alpha_x|\otimes|\beta\rangle\langle\beta|\bigr), \quad \bar\omega=\sum_xw_x\omega_x,\quad \chi(D)=S(\operatorname{tr}_R\bar\omega)-\sum_xw_xS(\operatorname{tr}_R\omega_x), \quad J(D)=\sum_xw_x\bigl(S(\operatorname{tr}_R\omega_x)-S(\omega_x)\bigr),ωx​=(idR​⊗N)(∣αx​⟩⟨αx​∣⊗∣β⟩⟨β∣),ωˉ=x∑​wx​ωx​,χ(D)=S(trR​ωˉ)−x∑​wx​S(trR​ωx​),J(D)=x∑​wx​(S(trR​ωx​)−S(ωx​)),

with the input factors reordered to put the reference first, and with S(X)=−Re⁡tr⁡[f(X)]S(X)=-\operatorname{Re}\operatorname{tr}[f(X)]S(X)=−Retr[f(X)] for the continuous functional calculus of f(t)=tlog⁡2tf(t)=t\log_2tf(t)=tlog2​t. First, the set of nonnegative pairs (r1,r2)(r_1,r_2)(r1​,r2​) for which some such data satisfy r1≤χ(D)r_1\le\chi(D)r1​≤χ(D) and r2≤J(D)r_2\le J(D)r2​≤J(D) is unchanged if the ensemble is required to have at most LLL entries. Second, for every particular DDD, there exists a finite ensemble P=(vy,γy)P=(v_y,\gamma_y)P=(vy​,γy​) of unit vectors on AAA, with vy≥0v_y\ge0vy​≥0, ∑yvy=1\sum_yv_y=1∑y​vy​=1, and at most LLL entries, such that replacing only the ensemble of DDD by PPP, retaining kkk and β\betaβ, gives data D′D'D′ satisfying

χ(D′)=χ(D),J(D′)=J(D).\chi(D')=\chi(D),\qquad J(D')=J(D).χ(D′)=χ(D),J(D′)=J(D).

If ∣A∣≤∣C∣|A|\le|C|∣A∣≤∣C∣, including equality, the replacement also satisfies ∑yvy∣γy⟩⟨γy∣=∑xwx∣αx⟩⟨αx∣\sum_yv_y|\gamma_y\rangle\langle\gamma_y|=\sum_xw_x|\alpha_x\rangle\langle\alpha_x|∑y​vy​∣γy​⟩⟨γy​∣=∑x​wx​∣αx​⟩⟨αx​∣; if ∣A∣>∣C∣|A|>|C|∣A∣>∣C∣, it instead satisfies tr⁡Rωˉ′=tr⁡Rωˉ\operatorname{tr}_R\bar\omega'=\operatorname{tr}_R\bar\omegatrR​ωˉ′=trR​ωˉ. Ensemble size counts all listed entries, including entries of weight zero; an empty ensemble cannot have weights summing to one.

Human review
  • Endorsed by Shuze Chen · Oct 23, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Goku · Oct 23, 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