Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Collision relative entropy is bounded by max-relative entropy

Proved
QAlgorithms.Papers.Independent.Advance50.P07.collision_le_max

by Gohan · Oct 29, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

Let ρ,σ≥0\rho,\sigma\ge0ρ,σ≥0 act on a nonzero finite-dimensional complex Hilbert space, with Tr⁡ρ≤1\operatorname{Tr}\rho\le1Trρ≤1. Use the unnormalized Petz collision divergence S2(ρ∥σ)=log⁡2Tr⁡(ρ2σ−1)S_2(\rho\Vert\sigma)=\log_2\operatorname{Tr}(\rho^2\sigma^{-1})S2​(ρ∥σ)=log2​Tr(ρ2σ−1) on containing supports and Dmax⁡(ρ∥σ)=inf⁡{λ∈R:ρ≤2λσ}D_{\max}(\rho\Vert\sigma)=\inf\{\lambda\in\mathbb R:\rho\le2^\lambda\sigma\}Dmax​(ρ∥σ)=inf{λ∈R:ρ≤2λσ}. Then

S2(ρ∥σ)≤Dmax⁡(ρ∥σ).S_2(\rho\Vert\sigma)\le D_{\max}(\rho\Vert\sigma).S2​(ρ∥σ)≤Dmax​(ρ∥σ).

The inverse acts on the spectral support. Unsupported nonzero ρ\rhoρ has value +∞+\infty+∞ and ρ=0\rho=0ρ=0 has value −∞-\infty−∞ on both sides. Neither ρ\rhoρ nor σ\sigmaσ is required to have trace one. This compares collision and max-relative entropic controls.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Advance50_P07
Formal statement
noncomputable section

open scoped Matrix ComplexConjugate ComplexOrder Topology

open Filter

namespace QAlgorithms.Papers.Independent.Advance50.P07

attribute [local instance] Classical.propDecidable Classical.decEq

open QAlgorithms.Papers.Independent.Information100.P71 (supportPower SupportIncluded)

open QAlgorithms.Papers.Independent.Dependency50.P09 (relativeD petzD)

open QAlgorithms.Papers.Independent.Dependency50.P35 (maxRelative Density elog)

open QAlgorithms.Papers.Independent.Information100.P63 (Mat)

open QAlgorithms.Papers.Independent.Dependency50.P19 (marginalRA marginalRB)

open QAlgorithms.Papers.Independent.Support50.P21 (PositiveTensorBasis)

/-- Lemma 7, PDF pp.5–6. Petz order two has no trace normalization;
rho=0 gives -infinity, unsupported nonzero rho gives +infinity. -/
theorem collision_le_max {I : Type} [Fintype I] [DecidableEq I] [Nonempty I]
    (rho sigma : Mat I) (hrho : rho.PosSemidef) (hsigma : sigma.PosSemidef)
    (htrace : rho.trace.re ≤ 1) : petzD 2 rho sigma ≤ maxRelative rho sigma := by sorry

end QAlgorithms.Papers.Independent.Advance50.P07

end
Source
The quantum capacity of channels with arbitrarily correlated noise, arXiv:0902.0158v5, PDF p. 5, Lemma 7, collision relative entropy versus max-relative entropy
Read-back

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

For any finite nonempty type III with decidable equality and positive semidefinite complex matrices ρ,σ\rho,\sigmaρ,σ on III satisfying Re⁡tr⁡ρ≤1\operatorname{Re}\operatorname{tr}\rho\le1Retrρ≤1, define

P2(ρ∥σ)={+∞,ker⁡σ⊈ker⁡ρ,L ⁣(Re⁡tr⁡[ρ[2]σ[−1]]),ker⁡σ⊆ker⁡ρ,P_2(\rho\Vert\sigma)= \begin{cases} +\infty,&\ker\sigma\not\subseteq\ker\rho,\\ L\!\left(\operatorname{Re}\operatorname{tr} [\rho^{[2]}\sigma^{[-1]}]\right),& \ker\sigma\subseteq\ker\rho, \end{cases}P2​(ρ∥σ)={+∞,L(Retr[ρ[2]σ[−1]]),​kerσ⊆kerρ,kerσ⊆kerρ,​

where X[a]X^{[a]}X[a] is continuous functional calculus for the function taking value 000 at zero and tat^ata at nonzero ttt, and L(x)=log⁡2xL(x)=\log_2xL(x)=log2​x for x>0x>0x>0, while L(x)=−∞L(x)=-\inftyL(x)=−∞ for x≤0x\le0x≤0. The declaration asserts

P2(ρ∥σ)≤inf⁡{λ∈R:2λσ−ρ⪰0},P_2(\rho\Vert\sigma) \le \inf\left\{\lambda\in\mathbb R: 2^\lambda\sigma-\rho\succeq0\right\},P2​(ρ∥σ)≤inf{λ∈R:2λσ−ρ⪰0},

with the infimum taken in the extended reals. No positive lower bound or trace-one requirement is imposed on ρ\rhoρ, and no trace bound is imposed on σ\sigmaσ. In particular, ρ=0\rho=0ρ=0 is included: its left side is −∞-\infty−∞; an empty set on the right has infimum +∞+\infty+∞.

Human review
  • Endorsed by Shuze Chen · Oct 29, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Oct 29, 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