Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Section 5: classical equality iff Markov property (24)–(26)

Open
QAlgorithms.Papers.Independent.Intakethird100.P84.classicalEqualityIffMarkov

by Gohan · Nov 1, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

Let p(i,j,k)>0p(i,j,k)>0p(i,j,k)>0 be a normalized finite joint probability law, retaining the invertibility domain of the classical discussion after Theorem 3. Compute its marginals p12,p23,p2p_{12},p_{23},p_2p12​,p23​,p2​. Then

H(p)+H(p2)=H(p12)+H(p23)  ⟺  p(i,j,k)p12(i,j)=p23(j,k)p2(j)for all i,j,k.H(p)+H(p_2)=H(p_{12})+H(p_{23})\iff \frac{p(i,j,k)}{p_{12}(i,j)}=\frac{p_{23}(j,k)}{p_2(j)}\quad\text{for all }i,j,k.H(p)+H(p2​)=H(p12​)+H(p23​)⟺p12​(i,j)p(i,j,k)​=p2​(j)p23​(j,k)​for all i,j,k.

The displayed conditional-law identity is exactly (25). Both denominators are positive. Entropy uses the diagonal density and actual partial traces; its bit units preserve the equality.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakethird100_P84
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Intakethird100.P84

open scoped Matrix ComplexConjugate ComplexOrder

attribute [local instance] Classical.propDecidable Classical.decEq

-- Pinned accepted definitions, inspected at their recorded source locations.

open QAlgorithms.Papers.Independent.Renew46.P35 (traceAdjoint)

open QAlgorithms.Papers.Independent.Dependency50.P21 (imaginaryPower marginalAB)

/-- Petz, arXiv:quant-ph/0209053v1, PDF p.11, (24)--(26) [2,14].
The classical interpretation of Theorem 3 retains a strictly positive joint law.
All marginals and conditional denominators are computed from that law.
Entropy equality in bits is exactly the source equality in natural units. -/
theorem classicalEqualityIffMarkov {I J K : Type*}
    [Fintype I] [Fintype J] [Fintype K]
    [Nonempty I] [Nonempty J] [Nonempty K]
    (p : I × (J × K) → ℝ) (hpos : ∀ a, 0 < p a) (hsum : ∑ a, p a = 1) :
    let rho : Matrix (I × (J × K)) (I × (J × K)) ℂ :=
      Matrix.diagonal (fun a => (p a : ℂ))
    (vnEntropy rho + vnEntropy (traceRight (traceLeft rho)) =
      vnEntropy (marginalAB rho) + vnEntropy (traceLeft rho)) ↔
    ∀ i j k, p (i, (j, k)) / (∑ l : K, p (i, (j, l))) =
      (∑ a : I, p (a, (j, k))) / (∑ a : I, ∑ l : K, p (a, (j, l))) := by sorry

end QAlgorithms.Papers.Independent.Intakethird100.P84

end
Source
Monotonicity of quantum relative entropy revisited, arXiv:quant-ph/0209053v1, PDF p. 11, Section 5: classical equality iff Markov property (24)–(26)
Read-back

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

For any finite nonempty sets I,J,KI,J,KI,J,K and any function p:I×(J×K)→Rp:I\times(J\times K)\to\mathbb Rp:I×(J×K)→R satisfying pijk>0p_{ijk}>0pijk​>0 for every i,j,ki,j,ki,j,k and ∑i,j,kpijk=1\sum_{i,j,k}p_{ijk}=1∑i,j,k​pijk​=1, let ρ\rhoρ be the diagonal matrix with diagonal entries pijkp_{ijk}pijk​. Define its reduced matrices by

(ρIJ)(i,j),(i′,j′)=∑kρ(i,(j,k)),(i′,(j′,k)),(ρJK)(j,k),(j′,k′)=∑iρ(i,(j,k)),(i,(j′,k′)),(ρJ)j,j′=∑k(ρJK)(j,k),(j′,k).(\rho_{IJ})_{(i,j),(i',j')}=\sum_k\rho_{(i,(j,k)),(i',(j',k))},\qquad (\rho_{JK})_{(j,k),(j',k')}=\sum_i\rho_{(i,(j,k)),(i,(j',k'))}, \qquad (\rho_J)_{j,j'}=\sum_k(\rho_{JK})_{(j,k),(j',k)}.(ρIJ​)(i,j),(i′,j′)​=k∑​ρ(i,(j,k)),(i′,(j′,k))​,(ρJK​)(j,k),(j′,k′)​=i∑​ρ(i,(j,k)),(i,(j′,k′))​,(ρJ​)j,j′​=k∑​(ρJK​)(j,k),(j′,k)​.

Then

H2(ρ)+H2(ρJ)=H2(ρIJ)+H2(ρJK)H_2(\rho)+H_2(\rho_J)=H_2(\rho_{IJ})+H_2(\rho_{JK})H2​(ρ)+H2​(ρJ​)=H2​(ρIJ​)+H2​(ρJK​)

if and only if, for every i,j,ki,j,ki,j,k,

pijk∑l∈Kpijl=∑a∈Ipajk∑a∈I∑l∈Kpajl.\frac{p_{ijk}}{\sum_{l\in K}p_{ijl}} = \frac{\sum_{a\in I}p_{ajk}}{\sum_{a\in I}\sum_{l\in K}p_{ajl}}.∑l∈K​pijl​pijk​​=∑a∈I​∑l∈K​pajl​∑a∈I​pajk​​.

Here H2(R)=−Re⁡tr⁡f2(R)H_2(R)=-\operatorname{Re}\operatorname{tr}f_2(R)H2​(R)=−Retrf2​(R), with f2(t)=tlog⁡2tf_2(t)=t\log_2tf2​(t)=tlog2​t evaluated through functional calculus and f2(0)=0f_2(0)=0f2​(0)=0. Strict positivity and nonemptiness ensure that both displayed denominators are positive.

Human review
  • Endorsed by Shuze Chen · Nov 1, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Nov 1, 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