Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

General SDP and covariant dispersion converse

Open
QAlgorithms.Papers.Independent.Support50.P04.general_outer_bounds

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

prove2quantum

Let N:A→B\mathcal N:A\to BN:A→B be a CPTP channel on finite positive-dimensional complex Hilbert spaces with fixed orthonormal bases. Write R^(n;ε)\widehat R(n;\varepsilon)R(n;ε) and R^cpp(n;ε)\widehat R^{\mathrm{cpp}}(n;\varepsilon)Rcpp(n;ε) for the maximum unassisted and classical-post-processing assisted entanglement-transmission rates. A code has a positive integer message dimension MMM, rate log⁡2M/n\log_2 M/nlog2​M/n, and squared fidelity to the normalized maximally entangled target at least 1−ε1-\varepsilon1−ε. A cpp encoder may retain sender memory; its decoder is LOCC across that memory and the receiver outputs, with its final quantum output at the receiver and no preshared entanglement. Let JN=(id⊗N)(∣A∣ΦA)J_{\mathcal N}=(\mathrm{id}\otimes\mathcal N)(|A|\Phi_A)JN​=(id⊗N)(∣A∣ΦA​) be the unnormalized Choi matrix. Define Γ(ρ)={Λ⪰0:Λ⪯ρT⊗IB, tr⁡ΛJN≥1−ε}\Gamma(\rho)=\{\Lambda\succeq0:\Lambda\preceq\rho^T\otimes I_B,\ \operatorname{tr}\Lambda J_{\mathcal N}\ge1-\varepsilon\}Γ(ρ)={Λ⪰0:Λ⪯ρT⊗IB​, trΛJN​≥1−ε}. Set f(N,ε)=inf⁡ρ∈S(A)inf⁡Λ∈Γ(ρ)sup⁡Jtr⁡(ΛJ)f(\mathcal N,\varepsilon)=\inf_{\rho\in\mathcal S(A)}\inf_{\Lambda\in\Gamma(\rho)}\sup_J\operatorname{tr}(\Lambda J)f(N,ε)=infρ∈S(A)​infΛ∈Γ(ρ)​supJ​tr(ΛJ), where the comparison Choi operators satisfy J⪰0J\succeq0J⪰0, JTA⪰0J^{T_A}\succeq0JTA​⪰0, and tr⁡BJ=IA\operatorname{tr}_B J=I_AtrB​J=IA​. Fix 0<ε<10<\varepsilon<10<ε<1. All logarithms are binary and Φ−1\Phi^{-1}Φ−1 is the standard Gaussian quantile. Remainder constants may depend on the fixed channel and error, and are chosen before the block length: there are C>0C>0C>0 and n0≥2n_0\ge2n0​≥2 such that the displayed bound holds for every integer n≥n0n\ge n_0n≥n0​. For every integer n≥1n\ge1n≥1, f(N⊗n,ε)>0f(\mathcal N^{\otimes n},\varepsilon)>0f(N⊗n,ε)>0 and

R^cpp(n;ε)≤−1nlog⁡2f(N⊗n,ε).\widehat R^{\mathrm{cpp}}(n;\varepsilon)\le-\frac1n\log_2 f(\mathcal N^{\otimes n},\varepsilon).Rcpp(n;ε)≤−n1​log2​f(N⊗n,ε).

Let ω=(id⊗N)(ΦA)\omega=(\mathrm{id}\otimes\mathcal N)(\Phi_A)ω=(id⊗N)(ΦA​). The Rains set is PPT′={σ⪰0:∥σTB∥1≤1}\mathrm{PPT}^{\prime}=\{\sigma\succeq0:\|\sigma^{T_B}\|_1\le1\}PPT′={σ⪰0:∥σTB​∥1​≤1}. Put IR(N)=min⁡σ∈PPT′D(ω∥σ)I_R(\mathcal N)=\min_{\sigma\in\mathrm{PPT}^{\prime}}D(\omega\|\sigma)IR​(N)=minσ∈PPT′​D(ω∥σ), where DDD is Umegaki relative entropy in bits, infinite on support failure. Let ΠR\Pi_RΠR​ be its full minimizer set and let VRεV_R^{\varepsilon}VRε​ be the maximum of the relative-information variance V(ω∥σ)V(\omega\|\sigma)V(ω∥σ) over ΠR\Pi_RΠR​ if ε<1/2\varepsilon<1/2ε<1/2, and the minimum otherwise. If N\mathcal NN is covariant for the full input unitary group with an output unitary representation, then IRI_RIR​ is finite and

R^cpp(n;ε)≤IR(N)+VRε(N)/n Φ−1(ε)+Clog⁡2n/n.\widehat R^{\mathrm{cpp}}(n;\varepsilon)\le I_R(\mathcal N)+\sqrt{V_R^\varepsilon(\mathcal N)/n}\,\Phi^{-1}(\varepsilon)+C\log_2n/n.Rcpp(n;ε)≤IR​(N)+VRε​(N)/n​Φ−1(ε)+Clog2​n/n.

Both clauses are converse bounds.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Support50_P04
Formal statement
noncomputable section

open scoped Matrix ComplexConjugate ComplexOrder

open QAlgorithms

namespace QAlgorithms.Papers.Independent.Support50.P04

variable {A B E : Type} [Fintype A] [DecidableEq A]

variable [Fintype B] [DecidableEq B] [Fintype E] [DecidableEq E]

/-- arXiv:1504.04617v2, PDF pp.7,11–13, Result 4, Eqs.(13)–(14).
Printed (13) omits 1/n; changed to per-use normalization as Eq.(1) and Eq.(39)
require. Covariance is the full-unitary convention explicitly stated on p.12. -/
theorem general_outer_bounds [Nonempty A] [Nonempty B]
    (N : Superop A B) (hN : IsChannel N) (epsilon : ℝ)
    (he0 : 0 < epsilon) (he1 : epsilon < 1) :
    (∀ n : ℕ, 0 < n → 0 < metaValue (piSuperop (fun _ : Fin n => N)) epsilon ∧
      cppBoundary N n epsilon ≤ ((-QAlgorithms.Papers.Independent.Information100.P80.log2 (metaValue (piSuperop (fun _ : Fin n => N)) epsilon) / n : ℝ) : EReal)) ∧
    (FullUnitaryCovariant N →
      ∃ a : ℝ, rainsInformation N = (a : EReal) ∧
        dispersionUpper (fun n => cppBoundary N n epsilon) a (rainsVariance N epsilon) epsilon) := by sorry

end QAlgorithms.Papers.Independent.Support50.P04

end
Source
Quantum Coding with Finite Resources, arXiv:1504.04617v2, PDF p. 7, Result 4, Eqs. (13)–(14)
Read-back

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

Let A,BA,BA,B be arbitrary finite nonempty types with decidable equality, let N\mathcal NN be a map from complex matrices on AAA to complex matrices on BBB admitting a finite Kraus representation N(X)=∑jEjXEj†\mathcal N(X)=\sum_jE_jXE_j^\daggerN(X)=∑j​Ej​XEj†​, ∑jEj†Ej=IA\sum_jE_j^\dagger E_j=I_A∑j​Ej†​Ej​=IA​, and let 0<ϵ<10<\epsilon<10<ϵ<1. Throughout, a channel means such a finite normalized Kraus sum, ∥X∥1=Re⁡tr⁡X†X\|X\|_1=\operatorname{Re}\operatorname{tr}\sqrt{X^\dagger X}∥X∥1​=RetrX†X​, log⁡2x=ln⁡x/ln⁡2\log_2x=\ln x/\ln2log2​x=lnx/ln2, and a density matrix is positive semidefinite with trace 111. Define BnB_nBn​ to be the extended-real supremum of real RRR for which there are positive integers M,qM,qM,q, a channel A:M→(An×q)\mathcal A:M\to(A^n\times q)A:M→(An×q), and a channel D:(q×Bn)→M\mathcal D:(q\times B^n)\to MD:(q×Bn)→M with R≤log⁡2M/nR\le\log_2M/nR≤log2​M/n and fidelity at least 1−ϵ1-\epsilon1−ϵ. The fidelity is Re⁡tr⁡[ΦM(id⊗T)(ΦM)]\operatorname{Re}\operatorname{tr}[\Phi_M(\mathrm{id}\otimes\mathcal T)(\Phi_M)]Retr[ΦM​(id⊗T)(ΦM​)], where ΦM=M−1∑i,j∣i,i⟩⟨j,j∣\Phi_M=M^{-1}\sum_{i,j}|i,i\rangle\langle j,j|ΦM​=M−1∑i,j​∣i,i⟩⟨j,j∣ and T\mathcal TT applies A\mathcal AA, reorders its output to put qqq first, applies idq⊗N⊗n\mathrm{id}_q\otimes\mathcal N^{\otimes n}idq​⊗N⊗n, and then applies D\mathcal DD. The decoder is additionally required to have the form F∘tr⁡a∘L∘Ad⁡W⊗V\mathcal F\circ\operatorname{tr}_a\circ\mathcal L\circ\operatorname{Ad}_{W\otimes V}F∘tra​∘L∘AdW⊗V​, for positive integers a,ba,ba,b, isometries W:q→aW:q\to aW:q→a, V:Bn→bV:B^n\to bV:Bn→b, and a channel F:b→M\mathcal F:b\to MF:b→M. Here L\mathcal LL, on a×ba\times ba×b, must be the limit of single-outcome instruments AjA_jAj​ coarse-grained from compatible finite local-history instruments KjK_jKj​: an instrument has completely positive complex-linear branches, summable outputs on every matrix, and trace-preserving sum; complete positivity means positivity under every finite identity amplification. Each local round is a branchwise tensor product in which one party performs a countable-outcome instrument and the other performs an outcome-dependent channel. Each KjK_jKj​ has a positive finite number of such rounds, Kj+1K_{j+1}Kj+1​ extends each history of KjK_jKj​ by one local round using the fixed natural-number pairing, coarse-graining sums branches over fibers of an outcome-label function, and convergence means sup⁡ω density∥(id⊗(Aj−L))(ω)∥1→0\sup_{\omega\text{ density}}\|(\mathrm{id}\otimes(A_j-\mathcal L))(\omega)\|_1\to0supω density​∥(id⊗(Aj​−L))(ω)∥1​→0, with identity-register dimension ababab. For any channel Q:I→O\mathcal Q:I\to OQ:I→O, define its unnormalized Choi matrix by JQ((i,o),(j,o′))=Q(∣i⟩⟨j∣)o,o′J_{\mathcal Q}((i,o),(j,o'))=\mathcal Q(|i\rangle\langle j|)_{o,o'}JQ​((i,o),(j,o′))=Q(∣i⟩⟨j∣)o,o′​, let Γ\GammaΓ transpose its first tensor factor, and define

μϵ(Q)=inf⁡ρ density on I, 0≤Λ≤ρT⊗IORe⁡tr⁡(ΛJQ)≥1−ϵ sup⁡J≥0, ΓJ≥0tr⁡OJ=IIRe⁡tr⁡(ΛJ).\mu_\epsilon(\mathcal Q)= \inf_{\substack{\rho\text{ density on }I,\ 0\le\Lambda\le\rho^{\mathsf T}\otimes I_O\\ \operatorname{Re}\operatorname{tr}(\Lambda J_{\mathcal Q})\ge1-\epsilon}} \ \sup_{\substack{J\ge0,\ \Gamma J\ge0\\\operatorname{tr}_O J=I_I}} \operatorname{Re}\operatorname{tr}(\Lambda J).μϵ​(Q)=ρ density on I, 0≤Λ≤ρT⊗IO​Retr(ΛJQ​)≥1−ϵ​inf​ J≥0, ΓJ≥0trO​J=II​​sup​Retr(ΛJ).

The theorem first asserts, for every natural n>0n>0n>0,

0<μϵ(N⊗n),Bn≤−log⁡2μϵ(N⊗n)n.0<\mu_\epsilon(\mathcal N^{\otimes n}), \qquad B_n\le-\frac{\log_2\mu_\epsilon(\mathcal N^{\otimes n})}{n}.0<μϵ​(N⊗n),Bn​≤−nlog2​μϵ​(N⊗n)​.

For its second assertion, put ω=JN/∣A∣\omega=J_{\mathcal N}/|A|ω=JN​/∣A∣, let TBT_BTB​ denote partial transpose on BBB, and set

P={σ≥0:∥σTB∥1≤1},Df(X∥σ)=Re⁡tr⁡(f(X)−Xln⁡σ)ln⁡2,f(x)=xln⁡x.\mathcal P=\{\sigma\ge0:\|\sigma^{T_B}\|_1\le1\},\quad D_f(X\|\sigma)= \frac{\operatorname{Re}\operatorname{tr}(f(X)-X\ln\sigma)}{\ln2}, \quad f(x)=x\ln x.P={σ≥0:∥σTB​∥1​≤1},Df​(X∥σ)=ln2Retr(f(X)−Xlnσ)​,f(x)=xlnx.

Functional calculus is used in these expressions; define the extended-real D(X∥σ)D(X\|\sigma)D(X∥σ) to equal Df(X∥σ)D_f(X\|\sigma)Df​(X∥σ) when ker⁡σ⊆ker⁡X\ker\sigma\subseteq\ker Xkerσ⊆kerX, and +∞+\infty+∞ otherwise. Let a∗=inf⁡σ∈PD(ω∥σ)a_*=\inf_{\sigma\in\mathcal P}D(\omega\|\sigma)a∗​=infσ∈P​D(ω∥σ), let O={σ∈P:D(ω∥σ)=a∗}\mathcal O=\{\sigma\in\mathcal P:D(\omega\|\sigma)=a_*\}O={σ∈P:D(ω∥σ)=a∗​}, and define

V(ω∥σ)=Re⁡tr⁡ ⁣[ω(log⁡2ω−log⁡2σ)2]−Df(ω∥σ)2,vϵ={sup⁡σ∈OV(ω∥σ),ϵ<12,inf⁡σ∈OV(ω∥σ),ϵ≥12.V(\omega\|\sigma)= \operatorname{Re}\operatorname{tr}\!\left[\omega(\log_2\omega-\log_2\sigma)^2\right] -D_f(\omega\|\sigma)^2,\qquad v_\epsilon= \begin{cases} \sup_{\sigma\in\mathcal O}V(\omega\|\sigma),&\epsilon<\tfrac12,\\ \inf_{\sigma\in\mathcal O}V(\omega\|\sigma),&\epsilon\ge\tfrac12. \end{cases}V(ω∥σ)=Retr[ω(log2​ω−log2​σ)2]−Df​(ω∥σ)2,vϵ​={supσ∈O​V(ω∥σ),infσ∈O​V(ω∥σ),​ϵ<21​,ϵ≥21​.​

If there exists a group homomorphism U↦VUU\mapsto V_UU↦VU​ from the unitary group on AAA to that on BBB such that N(UXU†)=VUN(X)VU†\mathcal N(UXU^\dagger)=V_U\mathcal N(X)V_U^\daggerN(UXU†)=VU​N(X)VU†​ for every unitary UUU and every matrix XXX, then there is a real aaa with a∗=aa_*=aa∗​=a, and there exist C>0C>0C>0, n0≥2n_0\ge2n0​≥2 such that, for every n≥n0n\ge n_0n≥n0​,

Bn≤a+vϵn zϵ+Clog⁡2nn.B_n\le a+\sqrt{\frac{v_\epsilon}{n}}\,z_\epsilon+ C\frac{\log_2n}{n}.Bn​≤a+nvϵ​​​zϵ​+Cnlog2​n​.

Here zϵ=inf⁡{z:∫−∞ze−x2/2/2π dx≥ϵ}z_\epsilon=\inf\{z:\int_{-\infty}^z e^{-x^2/2}/\sqrt{2\pi}\,dx\ge\epsilon\}zϵ​=inf{z:∫−∞z​e−x2/2/2π​dx≥ϵ}. The variance selection uses the displayed real infimum or supremum, without an additional attainment hypothesis; the comparison-set definition allows σ=0\sigma=0σ=0, for which the support condition fails for a nonzero density matrix. All finite-block conclusions explicitly require n>0n>0n>0.

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