Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Definitions for Monotonicity of quantum relative entropy revisited

Definition
QAlgorithms_Papers_Independent_Intakethird100_P84

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

prove2quantum

The objects the formalized statements of Monotonicity of quantum relative entropy revisited (arXiv:quant-ph/0209053) are stated in. Each was checked against the paper by a blind read-back and a moderator.

  • IsTwoPositive: PDF p.2, (2): positivity of the 2-by-2 block amplification; no higher amplifications or unitality are required.
  • embed12: PDF p.11, implicit tensor convention: A12 tensor I3, reassociated to I × (J × K). There is no dimension normalization.
  • petzExampleDensity: PDF p.7, Example 1: the first density; its domain is 0 < mu < 1/2.
  • exampleReference: PDF p.7, Example 1: the maximally mixed reference density in dimension 3.
  • exampleEffects: PDF p.7, Example 1: the three explicitly printed effects in source order. The off-diagonal entries in the last effect are -z and -conj(z).
  • measurementDensity: PDF pp.6–7: a finite measurement law viewed as its diagonal density. On a normalized density and POVM the entries are the source's Tr(D E_j), by cyclicity of trace.
Definition code
import Mathlib
import Definitions.Def_QAlgorithms_Defs_SDP
import Definitions.Def_QAlgorithms_Defs_Ensemble
import Definitions.Def_QAlgorithms_Defs_MatrixMapPositivity
import Definitions.Def_QAlgorithms_Papers_Independent_Dependency50_P21
import Definitions.Def_QAlgorithms_Papers_Independent_Renew46_P35

/-!
Petz, Monotonicity of quantum relative entropy revisited,
arXiv:quant-ph/0209053v1, PDF pp.2,5–12.

All matrix spaces are finite complex matrix algebras with ordinary trace.
Full-domain inequalities use the accepted extended `relativeEntropy`, with
+infinity on support failure. The finite real expression is restricted as follows.
The equality statements use the frozen `qRelEntropy` only on
invertible normalized densities. Its correction -rho+sigma has trace zero on
this domain, so its value is precisely the source's natural-log relative entropy.
The SSA equalities use the frozen entropy in bits: multiplying each side by
log 2 > 0 gives exactly the source's natural-log equality, in both directions.
For singular second arguments only the extended entropy is used.
Cited background statements are included at the complete source strength.
-/

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)

/-- PDF p.2, (2): positivity of the 2-by-2 block amplification; no higher
amplifications or unitality are required. -/
def IsTwoPositive {I J : Type*} [Fintype I] [Fintype J]
    (T : Matrix I I ℂ →ₗ[ℂ] Matrix J J ℂ) : Prop :=
  ∀ X : Matrix (Fin 2 × I) (Fin 2 × I) ℂ,
    X.PosSemidef → (idTensorSuperop T X).PosSemidef

/-- PDF p.11, implicit tensor convention: A12 tensor I3, reassociated to
I × (J × K). There is no dimension normalization. -/
def embed12 {I J K : Type*} (A : Matrix (I × J) (I × J) ℂ) :
    Matrix (I × (J × K)) (I × (J × K)) ℂ :=
  fun a b => if a.2.2 = b.2.2 then A (a.1, a.2.1) (b.1, b.2.1) else 0

/-- PDF p.7, Example 1: the first density; its domain is 0 < mu < 1/2. -/
def petzExampleDensity (mu : ℝ) : Matrix (Fin 3) (Fin 3) ℂ :=
  Matrix.diagonal ![((1 - 2 * mu : ℝ) : ℂ), (mu : ℂ), (mu : ℂ)]

/-- PDF p.7, Example 1: the maximally mixed reference density in dimension 3. -/
def exampleReference : Matrix (Fin 3) (Fin 3) ℂ :=
  Matrix.diagonal ![(1 / 3 : ℂ), (1 / 3 : ℂ), (1 / 3 : ℂ)]

/-- PDF p.7, Example 1: the three explicitly printed effects in source order.
The off-diagonal entries in the last effect are -z and -conj(z). -/
def exampleEffects (x : ℝ) (z : ℂ) : Fin 3 → Matrix (Fin 3) (Fin 3) ℂ :=
  ![Matrix.diagonal ![1, 0, 0],
    !![0, 0, 0; 0, (x : ℂ), z; 0, conj z, ((1 - x : ℝ) : ℂ)],
    !![0, 0, 0; 0, ((1 - x : ℝ) : ℂ), -z; 0, -(conj z), (x : ℂ)]]

/-- PDF pp.6–7: a finite measurement law viewed as its diagonal density.
On a normalized density and POVM the entries are the source's Tr(D E_j),
by cyclicity of trace. -/
def measurementDensity {I O : Type*} [Fintype I] [DecidableEq O]
    (E : O → Matrix I I ℂ) (D : Matrix I I ℂ) : Matrix O O ℂ :=
  Matrix.diagonal fun j => (effectProb (E j) D : ℂ)

/-- PDF p.2: the identity map satisfies the exact two-positive block condition. -/
example {I : Type*} [Fintype I] :
    IsTwoPositive (LinearMap.id : Matrix I I ℂ →ₗ[ℂ] Matrix I I ℂ) := by
  intro X hX
  exact hX

/-- PDF p.11: identity embedding retains a diagonal third-register block. -/
example {I J K : Type*} (A : Matrix (I × J) (I × J) ℂ)
    (a b : I × J) (k : K) :
    embed12 A (a.1, (a.2, k)) (b.1, (b.2, k)) = A a b := by
  simp [embed12]

/-- PDF p.7: the first weight is exactly 1-2 mu. -/
example (mu : ℝ) : petzExampleDensity mu 0 0 = ((1 - 2 * mu : ℝ) : ℂ) := by
  simp [petzExampleDensity]

/-- PDF p.7: the reference has ordinary trace one. -/
example : exampleReference.trace = 1 := by
  norm_num [exampleReference, Matrix.trace, Fin.sum_univ_three] <;>
    (change (2 / 3 : ℂ) + 1 / 3 = 1; norm_num)

/-- PDF p.7: the two off-diagonal effect entries cancel. -/
example (x : ℝ) (z : ℂ) :
    exampleEffects x z 1 1 2 + exampleEffects x z 2 1 2 = 0 := by
  simp [exampleEffects]

/-- PDF pp.6–7: measurement probabilities are the actual diagonal entries. -/
example {I O : Type*} [Fintype I] [DecidableEq O]
    (E : O → Matrix I I ℂ) (D : Matrix I I ℂ) (j : O) :
    measurementDensity E D j j = (effectProb (E j) D : ℂ) := by
  simp [measurementDensity]

end QAlgorithms.Papers.Independent.Intakethird100.P84
Source
Monotonicity of quantum relative entropy revisited, arXiv:quant-ph/0209053v1, https://arxiv.org/abs/quant-ph/0209053v1
Read-back

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

QAlgorithms.Papers.Independent.Intakethird100.P84.IsTwoPositive

For finite types I,JI,JI,J, possibly empty, and a complex linear map T:MI(C)→MJ(C)T:M_I(\mathbb C)\to M_J(\mathbb C)T:MI​(C)→MJ​(C), this defines two-positivity to mean that every positive semidefinite matrix XXX indexed by {0,1}×I\{0,1\}\times I{0,1}×I is sent to a positive semidefinite matrix indexed by {0,1}×J\{0,1\}\times J{0,1}×J by applying TTT to each block:

[(id2⊗T)(X)](r,j),(s,k)=T((X(r,a),(s,b))a,b∈I)j,k.\bigl[(\mathrm{id}_2\otimes T)(X)\bigr]_{(r,j),(s,k)} =T\bigl((X_{(r,a),(s,b)})_{a,b\in I}\bigr)_{j,k}.[(id2​⊗T)(X)](r,j),(s,k)​=T((X(r,a),(s,b)​)a,b∈I​)j,k​.

There are no nonemptiness assumptions on either index type.

QAlgorithms.Papers.Independent.Intakethird100.P84.embed12

For arbitrary types I,J,KI,J,KI,J,K and any complex matrix AAA indexed by I×JI\times JI×J, this defines a matrix indexed by I×(J×K)I\times(J\times K)I×(J×K) by

embed⁡12(A)(i,(j,k)),(i′,(j′,k′))={A(i,j),(i′,j′),k=k′,0,k≠k′.\operatorname{embed}_{12}(A)_{(i,(j,k)),(i',(j',k'))} = \begin{cases} A_{(i,j),(i',j')},&k=k',\\ 0,&k\ne k'. \end{cases}embed12​(A)(i,(j,k)),(i′,(j′,k′))​={A(i,j),(i′,j′)​,0,​k=k′,k=k′.​

No finiteness, nonemptiness, positivity, or normalization assumptions are imposed.

QAlgorithms.Papers.Independent.Intakethird100.P84.exampleEffects

For every real xxx and complex zzz, this defines a family of three complex 3×33\times33×3 matrices, indexed by 0,1,20,1,20,1,2:

E0=(100000000),E1=(0000xz0z‾1−x),E2=(00001−x−z0−z‾x).E_0=\begin{pmatrix}1&0&0\\0&0&0\\0&0&0\end{pmatrix},\qquad E_1=\begin{pmatrix}0&0&0\\0&x&z\\0&\overline z&1-x\end{pmatrix},\qquad E_2=\begin{pmatrix}0&0&0\\0&1-x&-z\\0&-\overline z&x\end{pmatrix}.E0​=​100​000​000​​,E1​=​000​0xz​0z1−x​​,E2​=​000​01−x−z​0−zx​​.

No bounds on xxx or zzz, and no positivity hypotheses, are part of the definition.

QAlgorithms.Papers.Independent.Intakethird100.P84.exampleReference

This defines the complex 3×33\times33×3 matrix

Dref=diag⁡ ⁣(13,13,13).D_{\mathrm{ref}}=\operatorname{diag}\!\left(\tfrac13,\tfrac13,\tfrac13\right).Dref​=diag(31​,31​,31​).

QAlgorithms.Papers.Independent.Intakethird100.P84.measurementDensity

For a finite type III, a type OOO with decidable equality, an arbitrary family Ej∈MI(C)E_j\in M_I(\mathbb C)Ej​∈MI​(C) indexed by j∈Oj\in Oj∈O, and an arbitrary matrix D∈MI(C)D\in M_I(\mathbb C)D∈MI​(C), this defines the matrix indexed by OOO whose off-diagonal entries are zero and whose jjj-th diagonal entry is the real number Re⁡tr⁡(EjD)\operatorname{Re}\operatorname{tr}(E_jD)Retr(Ej​D), regarded as complex:

ME(D)j,k={Re⁡tr⁡(EjD),j=k,0,j≠k.\mathcal M_E(D)_{j,k} = \begin{cases} \operatorname{Re}\operatorname{tr}(E_jD),&j=k,\\ 0,&j\ne k. \end{cases}ME​(D)j,k​={Retr(Ej​D),0,​j=k,j=k.​

Neither positivity nor trace normalization is assumed, OOO need not be finite, and an empty III makes every entry zero.

QAlgorithms.Papers.Independent.Intakethird100.P84.petzExampleDensity

For every real number μ\muμ, this defines the complex 3×33\times33×3 matrix

Dμ=diag⁡(1−2μ,μ,μ).D_\mu=\operatorname{diag}(1-2\mu,\mu,\mu).Dμ​=diag(1−2μ,μ,μ).

The parameter is unrestricted; in particular, this definition does not impose positivity, and the matrix is positive semidefinite precisely when 0≤μ≤120\le\mu\le\tfrac120≤μ≤21​.

QAlgorithms.Papers.Independent.Intakethird100.P84.statePreservingMap

Let I,J,KI,J,KI,J,K be nonempty finite types and let ρ\rhoρ, indexed by I×(J×K)I\times(J\times K)I×(J×K), be an invertible positive semidefinite complex matrix of trace one. Define

(D12)(i,j),(i′,j′)=∑kρ(i,(j,k)),(i′,(j′,k)),(D23)(j,k),(j′,k′)=∑iρ(i,(j,k)),(i,(j′,k′)),(D2)j,j′=∑k(D23)(j,k),(j′,k),(D_{12})_{(i,j),(i',j')}=\sum_k\rho_{(i,(j,k)),(i',(j',k))},\qquad (D_{23})_{(j,k),(j',k')}=\sum_i\rho_{(i,(j,k)),(i,(j',k'))},\qquad (D_2)_{j,j'}=\sum_k(D_{23})_{(j,k),(j',k)},(D12​)(i,j),(i′,j′)​=k∑​ρ(i,(j,k)),(i′,(j′,k))​,(D23​)(j,k),(j′,k′)​=i∑​ρ(i,(j,k)),(i,(j′,k′))​,(D2​)j,j′​=k∑​(D23​)(j,k),(j′,k)​,

and S(A)=−Re⁡tr⁡(u↦ulog⁡2u)(A)S(A)=-\operatorname{Re}\operatorname{tr}(u\mapsto u\log_2u)(A)S(A)=−Retr(u↦ulog2​u)(A), using spectral functional calculus. Then S(ρ)+S(D2)=S(D12)+S(D23)S(\rho)+S(D_2)=S(D_{12})+S(D_{23})S(ρ)+S(D2​)=S(D12​)+S(D23​) if and only if there exists a complex linear map γ:MI×(J×K)(C)→MJ×K(C)\gamma:M_{I\times(J\times K)}(\mathbb C)\to M_{J\times K}(\mathbb C)γ:MI×(J×K)​(C)→MJ×K​(C) satisfying all of the following: γ\gammaγ is completely positive, meaning that for every natural number rrr, including r=0r=0r=0, its blockwise action on every positive semidefinite matrix with r×rr\times rr×r blocks produces a positive semidefinite matrix; γ(1)=1\gamma(1)=1γ(1)=1; and

∀X,tr⁡(D23γ(X))=tr⁡(ρX),∀B∈MJ(C),γ(II⊗(B⊗IK))=B⊗IK.\forall X,\quad \operatorname{tr}(D_{23}\gamma(X))=\operatorname{tr}(\rho X), \qquad \forall B\in M_J(\mathbb C),\quad \gamma\bigl(I_I\otimes(B\otimes I_K)\bigr)=B\otimes I_K.∀X,tr(D23​γ(X))=tr(ρX),∀B∈MJ​(C),γ(II​⊗(B⊗IK​))=B⊗IK​.

Here XXX ranges over all complex matrices indexed by I×(J×K)I\times(J\times K)I×(J×K), without a positivity or Hermiticity restriction.

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