Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Code inclusion and exponent order (8)–(9)

Open
QAlgorithms.Papers.Independent.Intakenext100.P20.blind_visible_order

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

prove2quantum

Let Ξ\XiΞ be an arbitrary discrete label set and let H\mathcal HH be a nonzero finite-dimensional complex Hilbert space. The source consists of normalized pure states ρi\rho_iρi​ and probabilities pi≥0p_i\ge0pi​≥0 with ∑ipi=1\sum_i p_i=1∑i​pi​=1; write ρp=∑ipiρi\rho_p=\sum_i p_i\rho_iρp​=∑i​pi​ρi​. Each memory space is finite and nonzero. Blind encoders and decoders are CPTP; a visible encoder maps labels to normalized density operators. A blind encoder is fixed before its input label.

For positive block length nnn, the word probabilities and states are the independent products of the source probabilities and states. Admissible code sequences satisfy lim sup⁡nn−1log⁡dim⁡Kn≤R\limsup_n n^{-1}\log\dim\mathcal K_n\le Rlimsupn​n−1logdimKn​≤R. An error exponent is the supremum over admissible sequences of lim inf⁡n−n−1log⁡ϵn\liminf_n -n^{-1}\log\epsilon_nliminfn​−n−1logϵn​; a success exponent is the infimum of lim inf⁡n−n−1log⁡(1−ϵn)\liminf_n -n^{-1}\log(1-\epsilon_n)liminfn​−n−1log(1−ϵn​). Logarithms are natural, and −log⁡0=+∞-\log0=+\infty−log0=+∞.

For every blind code (K,E,D)(\mathcal K,E,D)(K,E,D), the visible encoder F(i)=E(ρi)F(i)=E(\rho_i)F(i)=E(ρi​) uses the same memory and decoder and has exactly the same average overlap error. Consequently,

re,B(R)≤re,V(R),re,B∗(R)≥re,V∗(R).r_{e,B}(R)\le r_{e,V}(R),\qquad r^*_{e,B}(R)\ge r^*_{e,V}(R).re,B​(R)≤re,V​(R),re,B∗​(R)≥re,V∗​(R).

These comparisons express inclusion of blind codes in the visible model.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P20
Formal statement
noncomputable section

open scoped Matrix ENNReal ComplexOrder Topology

namespace QAlgorithms.Papers.Independent.Intakenext100.P20

open QAlgorithms.Papers.Independent.Information100.P08 (IsPure)

open QAlgorithms.Papers.Independent.Information100.P47 (IsProbability)

open QAlgorithms.Papers.Independent.Information100.P53 (SpectralData)

open QAlgorithms.Papers.Independent.Gap48.P19 (shannonNats)

open QAlgorithms.Papers.Independent.Dependency50.P28 (classicalD)

attribute [local instance] Classical.propDecidable

/-- PDF pp.2-3, code inclusion and (8)-(9): every fixed blind encoder induces
an equally sized visible encoder with identical error, hence the two exponent orders. -/
theorem blind_visible_order {L : Type*} {d : ℕ} (S : DiscretePureSource L d) (R : ℝ) :
    (∀ C : BlindCode (Fin d),
      ∃ F : L → Matrix (Fin C.memory) (Fin C.memory) ℂ,
        (∀ i, IsDensityMatrix (F i)) ∧ F = (fun i => C.encode (S.state i)) ∧
        averageLoss S.weight S.state (fun i => C.decode (F i)) false =
          averageLoss S.weight S.state (fun i => C.decode (C.encode (S.state i))) false) ∧
    blindExponent S R false false ≤ visibleExponent S R false false ∧
    visibleExponent S R false true ≤ blindExponent S R false true := by sorry

end QAlgorithms.Papers.Independent.Intakenext100.P20

end
Source
Exponents of quantum fixed-length pure state source coding, arXiv:quant-ph/0202002v6, PDF p. 2, Code inclusion and exponent order (8)–(9)
Read-back

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

Let LLL be any type, without a finiteness or countability assumption, let ddd be a natural number, and let SSS consist of a proof that d>0d>0d>0, nonnegative summable weights wiw_iwi​ satisfying ∑i′wi=1\sum'_i w_i=1∑i′​wi​=1, and matrices ρi∈Cd×d\rho_i\in\mathbb C^{d\times d}ρi​∈Cd×d, each positive semidefinite with trace 111 and expressible as vivi†v_iv_i^\daggervi​vi†​ for a unit vector viv_ivi​. Let R∈RR\in\mathbb RR∈R be arbitrary. A channel here means a map X↦∑j=1rKjXKj†X\mapsto\sum_{j=1}^{r}K_jXK_j^\daggerX↦∑j=1r​Kj​XKj†​ for some finite family satisfying ∑jKj†Kj=I\sum_jK_j^\dagger K_j=I∑j​Kj†​Kj​=I. For every blind code with positive integer memory dimension mmm, channel encoder E:Cd×d→Cm×mE:\mathbb C^{d\times d}\to\mathbb C^{m\times m}E:Cd×d→Cm×m, and channel decoder D:Cm×m→Cd×dD:\mathbb C^{m\times m}\to\mathbb C^{d\times d}D:Cm×m→Cd×d, there exists a family FiF_iFi​ of density matrices on the memory with Fi=E(ρi)F_i=E(\rho_i)Fi​=E(ρi​), and

∑i′wi ⁣[1−Re⁡Tr⁡(D(Fi)ρi)]=∑i′wi ⁣[1−Re⁡Tr⁡(D(E(ρi))ρi)].\sum'_i w_i\!\left[1-\operatorname{Re}\operatorname{Tr}(D(F_i)\rho_i)\right] = \sum'_i w_i\!\left[1-\operatorname{Re}\operatorname{Tr}(D(E(\rho_i))\rho_i)\right].i∑′​wi​[1−ReTr(D(Fi​)ρi​)]=i∑′​wi​[1−ReTr(D(E(ρi​))ρi​)].

Furthermore, define length-nnn source words, for every n≥1n\ge1n≥1, by weights wu=∏j=0n−1wujw_u=\prod_{j=0}^{n-1}w_{u_j}wu​=∏j=0n−1​wuj​​ and states (ρu)x,y=∏j=0n−1(ρuj)xj,yj(\rho_u)_{x,y}=\prod_{j=0}^{n-1}(\rho_{u_j})_{x_j,y_j}(ρu​)x,y​=∏j=0n−1​(ρuj​​)xj​,yj​​. A visible code assigns each word uuu an arbitrary memory density matrix and uses a channel decoder; a blind code uses channel encoding and decoding. For either kind of code sequence, with positive memory dimensions mnm_nmn​, restrict to sequences satisfying lim sup⁡n→∞log⁡(mn)/n≤R\limsup_{n\to\infty}\log(m_n)/n\le Rlimsupn→∞​log(mn​)/n≤R, and put en=∑u′wu[1−Re⁡Tr⁡(ρ^uρu)]e_n=\sum'_u w_u[1-\operatorname{Re}\operatorname{Tr}(\widehat\rho_u\rho_u)]en​=∑u′​wu​[1−ReTr(ρ​u​ρu​)]. Define ℓn(q)=+∞\ell_n(q)=+\inftyℓn​(q)=+∞ when q=0q=0q=0, and ℓn(q)=−log⁡(q)/n\ell_n(q)=-\log(q)/nℓn​(q)=−log(q)/n otherwise. Let EviserrE_{\mathrm{vis}}^{\mathrm{err}}Eviserr​ and EblinderrE_{\mathrm{blind}}^{\mathrm{err}}Eblinderr​ be the suprema over their respective sequences of lim inf⁡nℓn(en)\liminf_n\ell_n(e_n)liminfn​ℓn​(en​), and let EvissuccE_{\mathrm{vis}}^{\mathrm{succ}}Evissucc​ and EblindsuccE_{\mathrm{blind}}^{\mathrm{succ}}Eblindsucc​ be the infima of lim inf⁡nℓn(1−en)\liminf_n\ell_n(1-e_n)liminfn​ℓn​(1−en​). The asserted inequalities are

Eblinderr≤Eviserr,Evissucc≤Eblindsucc.E_{\mathrm{blind}}^{\mathrm{err}}\le E_{\mathrm{vis}}^{\mathrm{err}}, \qquad E_{\mathrm{vis}}^{\mathrm{succ}}\le E_{\mathrm{blind}}^{\mathrm{succ}}.Eblinderr​≤Eviserr​,Evissucc​≤Eblindsucc​.

These are extended real extrema; an empty supremum is −∞-\infty−∞ and an empty infimum is +∞+\infty+∞, so the statement also includes rates for which no code sequence satisfies the rate bound.

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