Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Lemma 3.5

Open
QAlgorithms.Papers.Independent.Support50.P12.lemma_3_5

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

prove2quantum

Let A1⊆B(H1)\mathcal A_1\subseteq B(H_1)A1​⊆B(H1​) and A2⊆B(H2)\mathcal A_2\subseteq B(H_2)A2​⊆B(H2​) be concrete unital algebras on nonzero finite-dimensional complex Hilbert spaces, and let Φ:A1→A2\Phi:\mathcal A_1\to\mathcal A_2Φ:A1​→A2​ be complex linear. The adjoint is for ⟨X,Y⟩HS=Tr⁡X∗Y\langle X,Y\rangle_{\rm HS}=\operatorname{Tr}X^*Y⟨X,Y⟩HS​=TrX∗Y. A substochastic map has a Schwarz-contracting adjoint; stochastic additionally means trace preserving. Let A,B∈A1,+A,B\in\mathcal A_{1,+}A,B∈A1,+​. Powers are zero on kernels and traces are not normalized.

For every positive Φ\PhiΦ,

Φ(B0AB0)Φ(B)−1Φ(B0AB0)≤Φ(B0AB−1AB0).\Phi(B^0AB^0)\Phi(B)^{-1}\Phi(B^0AB^0)\le\Phi(B^0AB^{-1}AB^0).Φ(B0AB0)Φ(B)−1Φ(B0AB0)≤Φ(B0AB−1AB0).

If A0≤B0A^0\le B^0A0≤B0, also Φ(A)Φ(B)−1Φ(A)≤Φ(AB−1A)\Phi(A)\Phi(B)^{-1}\Phi(A)\le\Phi(AB^{-1}A)Φ(A)Φ(B)−1Φ(A)≤Φ(AB−1A). If, in addition, Φ\PhiΦ is trace non-increasing, then

Sx2(Φ(A)∥Φ(B))=Tr⁡Φ(A)2Φ(B)−1≤Tr⁡A2B−1=Sx2(A∥B).S_{x^2}(\Phi(A)\Vert\Phi(B))=\operatorname{Tr}\Phi(A)^2\Phi(B)^{-1}\le\operatorname{Tr}A^2B^{-1}=S_{x^2}(A\Vert B).Sx2​(Φ(A)∥Φ(B))=TrΦ(A)2Φ(B)−1≤TrA2B−1=Sx2​(A∥B).

The support hypothesis applies only to the latter two assertions. These comparisons underpin general monotonicity.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Support50_P12
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Support50.P12

open Filter MeasureTheory Set

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

open QAlgorithms.Papers.Independent.Information100.P30 (MatrixMap)

open QAlgorithms.Papers.Independent.Dependency50.P28 (tr LE logNonnegativeExtended pinch)

open QAlgorithms.Papers.Independent.Dependency50.P21 (channelAdjoint spectralFunction petz relativeEntropy)

open QAlgorithms.Papers.Independent.Frontier18.P17 (hs)

open QAlgorithms.Papers.Independent.Next46.P23 (Supports)

open QAlgorithms.Papers.Independent.Support21.P03 (ExtendedConvex)

open QAlgorithms (spectralProjector)

open QAlgorithms.Learning (IsDistribution)

open QAlgorithms.Papers.Independent.Information100.P04 (absOp)

open scoped BigOperators Matrix ComplexConjugate ComplexOrder ENNReal Topology Kronecker

attribute [local instance] Classical.propDecidable Classical.decEq

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

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

/-- Lemma 3.5, PDF p.11: unconditional compression inequality; (3.4) and (3.5)
inherit support inclusion, and only the final clause requires trace non-increase. -/
theorem lemma_3_5 [Nonempty I] [Nonempty J] (A : Alg I) (B : Alg J)
    (Phi : MatrixMap I J) (hc : AlgebraExtension A B Phi) (hp : PositiveOnConcreteAlgebra A Phi)
    (X Y : Mat I) (hX : X ∈ A) (hY : Y ∈ A) (hXp : X.PosSemidef) (hYp : Y.PosSemidef) :
    LE (Phi (supp Y * X * supp Y) * supportPower (Phi Y) (-1) * Phi (supp Y * X * supp Y))
      (Phi (supp Y * X * supportPower Y (-1) * X * supp Y)) ∧
    (Supports X Y → LE (Phi X * supportPower (Phi Y) (-1) * Phi X)
      (Phi (X * supportPower Y (-1) * X))) ∧
    (Supports X Y → AlgebraTraceNonincreasing A Phi →
      div (supportScalarPower 2) (Phi X) (Phi Y) = (tr (Phi X ^ 2 * supportPower (Phi Y) (-1)) : EReal) ∧
      tr (Phi X ^ 2 * supportPower (Phi Y) (-1)) ≤ tr (X ^ 2 * supportPower Y (-1)) ∧
      (tr (X ^ 2 * supportPower Y (-1)) : EReal) = div (supportScalarPower 2) X Y) := by sorry

end QAlgorithms.Papers.Independent.Support50.P12

end
Source
Quantum f-divergences and error correction, arXiv:1008.2529v6, PDF p. 11, Lemma 3.5
Read-back

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

Let I,JI,JI,J be finite, nonempty index types with decidable equality, let A,B\mathcal A,\mathcal BA,B be unital complex star subalgebras, and let Φ:Mat⁡(I)→Mat⁡(J)\Phi:\operatorname{Mat}(I)\to\operatorname{Mat}(J)Φ:Mat(I)→Mat(J) be complex linear. Define [Φ†(Z)]ij=∑k,l[Φ(Eij)]kl‾Zkl[\Phi^\dagger(Z)]_{ij}=\sum_{k,l}\overline{[\Phi(E_{ij})]_{kl}}Z_{kl}[Φ†(Z)]ij​=∑k,l​[Φ(Eij​)]kl​​Zkl​. Assume that Φ\PhiΦ maps every ambient matrix into B\mathcal BB, that Φ†\Phi^\daggerΦ† maps every ambient matrix into A\mathcal AA, and that Φ\PhiΦ maps positive semidefinite members of A\mathcal AA to positive semidefinite matrices. For positive semidefinite X,Y∈AX,Y\in\mathcal AX,Y∈A, write Z[s]Z^{[s]}Z[s] for functional calculus assigning zero to eigenvalue zero and xsx^sxs elsewhere, put PY=Y[0]P_Y=Y^{[0]}PY​=Y[0], and interpret U≤VU\le VU≤V as positive semidefiniteness of V−UV-UV−U. Then

Φ(PYXPY)(ΦY)[−1]Φ(PYXPY)≤Φ(PYXY[−1]XPY).\Phi(P_YXP_Y)(\Phi Y)^{[-1]}\Phi(P_YXP_Y) \le \Phi(P_YXY^{[-1]}XP_Y).Φ(PY​XPY​)(ΦY)[−1]Φ(PY​XPY​)≤Φ(PY​XY[−1]XPY​).

If ker⁡Y⊆ker⁡X\ker Y\subseteq\ker XkerY⊆kerX, then additionally

Φ(X)(ΦY)[−1]Φ(X)≤Φ(XY[−1]X).\Phi(X)(\Phi Y)^{[-1]}\Phi(X)\le\Phi(XY^{[-1]}X).Φ(X)(ΦY)[−1]Φ(X)≤Φ(XY[−1]X).

If this support inclusion holds and Φ\PhiΦ also satisfies tr⁡(ΦZ)≤tr⁡Z\operatorname{tr}(\Phi Z)\le\operatorname{tr}Ztr(ΦZ)≤trZ for every positive semidefinite Z∈AZ\in\mathcal AZ∈A, where tr⁡=Re⁡Tr⁡\operatorname{tr}=\operatorname{Re}\operatorname{Tr}tr=ReTr, then

D2(ΦX,ΦY)=tr⁡ ⁣((ΦX)2(ΦY)[−1])≤tr⁡(X2Y[−1])=D2(X,Y).D_2(\Phi X,\Phi Y) =\operatorname{tr}\!\left((\Phi X)^2(\Phi Y)^{[-1]}\right) \le\operatorname{tr}(X^2Y^{[-1]}) =D_2(X,Y).D2​(ΦX,ΦY)=tr((ΦX)2(ΦY)[−1])≤tr(X2Y[−1])=D2​(X,Y).

Here D2(U,V)D_2(U,V)D2​(U,V) is the extended-real quantity obtained from f(x)=0f(x)=0f(x)=0 at x=0x=0x=0 and f(x)=x2f(x)=x^2f(x)=x2 otherwise: it equals ∑λ,μμf(λ/μ)Re⁡Tr⁡(Pλ(U)Pμ(V))\sum_{\lambda,\mu}\mu f(\lambda/\mu)\operatorname{Re}\operatorname{Tr}(P_\lambda(U)P_\mu(V))∑λ,μ​μf(λ/μ)ReTr(Pλ​(U)Pμ​(V)) when ker⁡V⊆ker⁡U\ker V\subseteq\ker UkerV⊆kerU, and otherwise uses the formal total extended-real limit-selection value of that expression with V+εIV+\varepsilon IV+εI as ε↓0\varepsilon\downarrow0ε↓0; spectra use distinct eigenvalues and division by zero is zero. The traces in the final chain are embedded into the extended real line where required. No invertibility or trace normalization is assumed, and zero matrices are allowed.

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