Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§5.3, difference-of-entropies identity

Open
QAlgorithms.Papers.Independent.Intakenext100.P45.conditionalEntropyDifference

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

prove2quantum

For densities ρ,σ\rho,\sigmaρ,σ with ψ0=Tr⁡(ρsupp⁡σ)>0\psi_0=\operatorname{Tr}(\rho\operatorname{supp}\sigma)>0ψ0​=Tr(ρsuppσ)>0, set R=ρ/ψ0R=\rho/\psi_0R=ρ/ψ0​ and X=Rsupp⁡σX=R\operatorname{supp}\sigmaX=Rsuppσ. The operator XXX has trace one and is diagonalizable with nonnegative spectrum. With the spectral logarithm set to zero on its zero eigenspace,

Sσ(ρ∥σ)=Sgen(X∥σ)−Sgen(X∥R).S_\sigma(\rho\|\sigma)=S_{\rm gen}(X\|\sigma)-S_{\rm gen}(X\|R).Sσ​(ρ∥σ)=Sgen​(X∥σ)−Sgen​(X∥R).

Both traces defining the generalized relative entropies are real. This trace extension permits non-self-adjoint XXX and unnormalized RRR; it is not the normalized-state Umegaki function.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P45
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Intakenext100.P45

open QAlgorithms Filter

open scoped Matrix ComplexConjugate ComplexOrder ENNReal BigOperators Topology

open QAlgorithms.Papers.Independent.Information100.P07
  (support positivePart positiveProjection IsProjection)

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

open QAlgorithms.Papers.Independent.Information100.P60 (Mat copies)

open QAlgorithms.Papers.Independent.Dependency50.P28 (logNonnegativeExtended relativeEntropy)

open QAlgorithms.Papers.Independent.Claimed50.P15 (positiveMatrixRpow)

attribute [local instance] Classical.propDecidable Classical.decEq

variable {I J : Type} [Fintype I] [DecidableEq I] [Fintype J] [DecidableEq J]

/-- arXiv:0708.4282v1, PDF p.26: §5.3, difference-of-entropies identity. -/
theorem conditionalEntropyDifference (rho sigma : Mat I)
    (hrho : IsDensityMatrix rho) (hsigma : IsDensityMatrix sigma)
    (hpsi : 0 < (rho*support sigma).trace.re) :
    let R := (((rho*support sigma).trace.re)⁻¹ : ℂ) • rho
    let X := R*support sigma
    (∃ d, DiagonalizesNonnegative X d) ∧ X.trace=1 ∧
    supportConditionalEntropy rho sigma = generalizedRelative X sigma-generalizedRelative X R ∧
    (X*(nonnegativeSpectrumLog X-cfc Real.log sigma)).trace.im=0 ∧
    (X*(nonnegativeSpectrumLog X-cfc Real.log R)).trace.im=0 := by sorry

end QAlgorithms.Papers.Independent.Intakenext100.P45

end
Source
Asymptotic Error Rates in Quantum Hypothesis Testing, arXiv:0708.4282v1, PDF p. 26, §5.3, difference-of-entropies identity
Read-back

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

Let III be a finite type with decidable equality, and let ρ,σ\rho,\sigmaρ,σ be positive semidefinite complex I×II\times II×I matrices with trace one. Let SσS_\sigmaSσ​ be the orthogonal projection onto the operator range of σ\sigmaσ, assume ψ=Re⁡tr⁡(ρSσ)>0\psi=\operatorname{Re}\operatorname{tr}(\rho S_\sigma)>0ψ=Retr(ρSσ​)>0, and put R=ψ−1ρR=\psi^{-1}\rhoR=ψ−1ρ and X=RSσX=RS_\sigmaX=RSσ​. Define L(Y)\mathcal L(Y)L(Y) as follows: if there exist a nonnegative real family did_idi​ and matrices B,CB,CB,C with BC=CB=1BC=CB=1BC=CB=1 and Y=Bdiag⁡(d)CY=B\operatorname{diag}(d)CY=Bdiag(d)C, choose one such triple and set L(Y)=Bdiag⁡(log⁡d)C\mathcal L(Y)=B\operatorname{diag}(\log d)CL(Y)=Bdiag(logd)C; if no such triple exists, set L(Y)=0\mathcal L(Y)=0L(Y)=0. Here the total real logarithm has log⁡0=0\log0=0log0=0. Define G(Y,Z)=Re⁡tr⁡[Y(L(Y)−log⁡Z)]G(Y,Z)=\operatorname{Re}\operatorname{tr}[Y(\mathcal L(Y)-\log Z)]G(Y,Z)=Retr[Y(L(Y)−logZ)], where log⁡Z\log ZlogZ denotes continuous functional calculus for that logarithm, and define Hc(ρ,σ)=Re⁡tr⁡[R(log⁡R−log⁡σ)Sσ]H_c(\rho,\sigma)=\operatorname{Re}\operatorname{tr}[R(\log R-\log\sigma)S_\sigma]Hc​(ρ,σ)=Retr[R(logR−logσ)Sσ​]. The declaration asserts that XXX admits a diagonalization X=Bdiag⁡(d)CX=B\operatorname{diag}(d)CX=Bdiag(d)C with di≥0d_i\ge0di​≥0 and BC=CB=1BC=CB=1BC=CB=1, that tr⁡X=1\operatorname{tr}X=1trX=1 as a complex equality, and that

Hc(ρ,σ)=G(X,σ)−G(X,R),Im⁡tr⁡[X(L(X)−log⁡σ)]=0,Im⁡tr⁡[X(L(X)−log⁡R)]=0.H_c(\rho,\sigma)=G(X,\sigma)-G(X,R),\qquad \operatorname{Im}\operatorname{tr}[X(\mathcal L(X)-\log\sigma)]=0,\qquad \operatorname{Im}\operatorname{tr}[X(\mathcal L(X)-\log R)]=0.Hc​(ρ,σ)=G(X,σ)−G(X,R),Imtr[X(L(X)−logσ)]=0,Imtr[X(L(X)−logR)]=0.

The diagonalizing matrices are required to be mutual inverses; no unitary condition is imposed on them.

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