Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Quantum fixed-error strong converse with corrected logarithm

Open
QAlgorithms.Papers.Independent.Support50.P38.theorem21

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

prove2quantum

Let XXX be finite nonempty, PPP a probability law, and ρx\rho_xρx​ density matrices on a fixed finite nonzero-dimensional Hilbert space. Let M(n,λ)M(n,\lambda)M(n,λ) be the minimum positive code dimension over all visible maps into normalized code states followed by a linear completely positive trace-preserving decoder, with no shared randomness and PnP^nPn-average half trace norm error at most λ\lambdaλ against ρx1⊗⋯⊗ρxn\rho_{x_1}\otimes\cdots\otimes\rho_{x_n}ρx1​​⊗⋯⊗ρxn​​. Write

χ=S(∑xP(x)ρx)−∑xP(x)S(ρx),\chi=S\left(\sum_xP(x)\rho_x\right)-\sum_xP(x)S(\rho_x),χ=S(x∑​P(x)ρx​)−x∑​P(x)S(ρx​),

where SSS is von Neumann entropy in bits. For every fixed 0<λ<10<\lambda<10<λ<1,

lim inf⁡n→∞log⁡2M(n,λ)n≥χ.\liminf_{n\to\infty}\frac{\log_2M(n,\lambda)}n\ge\chi.n→∞liminf​nlog2​M(n,λ)​≥χ.

This is the quantum strong converse lower bound.

Source correction. The printed theorem omits the logarithm. Its definition of MMM as a dimension, the exponential bound in its proof, and Question 24 require log⁡2M\log_2Mlog2​M.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Support50_P38
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Support50.P38

open scoped BigOperators Matrix ComplexConjugate

open Filter

open QAlgorithms.Papers.Independent.Information100.P56 (IsLaw)

open QAlgorithms.Papers.Independent.Frontier30.P12 (IsStochastic)

open QAlgorithms.Papers.Independent.Information100.P27 (shannon labelInformation)

attribute [local instance] Classical.propDecidable

/-- Theorem 21, PDF pp.10–11: quantum fixed-error strong converse.
SOURCE CORRECTION: insert log_2 before M. The printed display omits log, although p.10
defines M as dim C, p.11 proves an exponential dimension bound, and Question 24 includes log.
The p.11 success-prefactor inversion is also a proof typo and is not assumed here. -/
theorem theorem21 {A B : Type} [Fintype A] [Fintype B] [DecidableEq B]
    [Nonempty A] [Nonempty B] (p : A → ℝ) (rho : A → Matrix B B ℂ)
    (hp : IsLaw p) (hrho : ∀ a, QAlgorithms.IsDensityMatrix (rho a))
    (error : ℝ) (he : 0 < error) (he1 : error < 1) :
    labelInformation p rho ≤ Filter.liminf (quantumSizeRate p rho error) atTop := by sorry

end QAlgorithms.Papers.Independent.Support50.P38

end
Source
Compression of sources of probability distributions and density operators, arXiv:quant-ph/0208131v1, PDF p. 10, Theorem 21
Read-back

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

Let A,BA,BA,B be finite nonempty types, with decidable equality on BBB. Let p:A→Rp:A\to\mathbb Rp:A→R be nonnegative with ∑ap(a)=1\sum_a p(a)=1∑a​p(a)=1, and let every complex B×BB\times BB×B matrix ρa\rho_aρa​ be positive semidefinite with trace 111. For every real ε\varepsilonε with 0<ε<10<\varepsilon<10<ε<1,

S ⁣(∑ap(a)ρa)−∑ap(a)S(ρa)≤lim inf⁡k→∞log⁡2Mk+1(ε)k+1.S\!\left(\sum_a p(a)\rho_a\right)-\sum_a p(a)S(\rho_a) \le \liminf_{k\to\infty}\frac{\log_2 M_{k+1}(\varepsilon)}{k+1}.S(a∑​p(a)ρa​)−a∑​p(a)S(ρa​)≤k→∞liminf​k+1log2​Mk+1​(ε)​.

Here S(X)=−Re⁡tr⁡(f(X))S(X)=-\operatorname{Re}\operatorname{tr}(f(X))S(X)=−Retr(f(X)), where f(t)=tlog⁡2tf(t)=t\log_2 tf(t)=tlog2​t is applied by functional calculus using the total logarithm with log⁡20=0\log_2 0=0log2​0=0. For each block length NNN, MN(ε)M_N(\varepsilon)MN​(ε) is the infimum in N\mathbb NN of the positive integers MMM for which one can assign a positive semidefinite, trace-one M×MM\times MM×M matrix σx\sigma_xσx​ to every word x∈ANx\in A^Nx∈AN, and choose a map

D(X)=∑j=0r−1KjXKj†,∑j=0r−1Kj†Kj=IMD(X)=\sum_{j=0}^{r-1}K_jXK_j^\dagger, \qquad \sum_{j=0}^{r-1}K_j^\dagger K_j=I_MD(X)=j=0∑r−1​Kj​XKj†​,j=0∑r−1​Kj†​Kj​=IM​

for some natural number rrr and complex matrices KjK_jKj​ with rows indexed by BNB^NBN and columns indexed by {0,…,M−1}\{0,\ldots,M-1\}{0,…,M−1}, such that

∑x∈AN(∏i=0N−1p(xi))12∥ρx(N)−D(σx)∥1≤ε,(ρx(N))y,z=∏i=0N−1(ρxi)yi,zi.\sum_{x\in A^N}\left(\prod_{i=0}^{N-1}p(x_i)\right) \frac12\left\|\rho_x^{(N)}-D(\sigma_x)\right\|_1\le\varepsilon, \qquad \bigl(\rho_x^{(N)}\bigr)_{y,z} =\prod_{i=0}^{N-1}(\rho_{x_i})_{y_i,z_i}.x∈AN∑​(i=0∏N−1​p(xi​))21​​ρx(N)​−D(σx​)​1​≤ε,(ρx(N)​)y,z​=i=0∏N−1​(ρxi​​)yi​,zi​​.

The trace norm is ∥Z∥1=Re⁡tr⁡Z†Z\|Z\|_1=\operatorname{Re}\operatorname{tr}\sqrt{Z^\dagger Z}∥Z∥1​=RetrZ†Z​. The encoder must assign a density matrix even to words of zero source probability. The natural-number infimum is 000 if its defining set is empty; the rate sequence uses only positive block lengths N=k+1N=k+1N=k+1.

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