Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Example 2.7

Proved
QAlgorithms.Papers.Independent.Support50.P12.example_2_7

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

prove2quantum

Let A,BA,BA,B be positive semidefinite operators on a nonzero finite-dimensional complex Hilbert space. Let f:[0,∞)→Rf:[0,\infty)\to\mathbb Rf:[0,∞)→R be continuous on (0,∞)(0,\infty)(0,∞) with signed extended recession slope ω(f)=lim⁡x→∞f(x)/x\omega(f)=\lim_{x\to\infty}f(x)/xω(f)=limx→∞​f(x)/x. Write SfS_fSf​ for the paper's quantum divergence, Qf(A,B)=⟨B1/2,f(LARB−1)B1/2⟩HSQ_f(A,B)=\langle B^{1/2},f(L_AR_{B^{-1}})B^{1/2}\rangle_{\rm HS}Qf​(A,B)=⟨B1/2,f(LA​RB−1​)B1/2⟩HS​, and X0X^0X0 for the support projection. All powers vanish on the kernel, including exponent zero. Traces are representation traces.

Put fα(x)=xαf_\alpha(x)=x^\alphafα​(x)=xα for α>0\alpha>0α>0, and f0(x)=1f_0(x)=1f0​(x)=1 for x>0x>0x>0, f0(0)=0f_0(0)=0f0​(0)=0. Then

Sfα(A∥B)=Tr⁡AαB1−α+ω(fα)Tr⁡A(I−B0),α≥0.S_{f_\alpha}(A\Vert B)=\operatorname{Tr}A^\alpha B^{1-\alpha}+\omega(f_\alpha)\operatorname{Tr}A(I-B^0),\qquad \alpha\ge0.Sfα​​(A∥B)=TrAαB1−α+ω(fα​)TrA(I−B0),α≥0. Sfα(A∥B)={Tr⁡AαB1−α,0≤α<1 or A0≤B0,+∞,α>1, A0ot≤B0.S_{f_\alpha}(A\Vert B)=\begin{cases}\operatorname{Tr}A^\alpha B^{1-\alpha},&0\le\alpha<1\text{ or }A^0\le B^0,\\+\infty,&\alpha>1,\ A^0 ot\le B^0.\end{cases}Sfα​​(A∥B)={TrAαB1−α,+∞,​0≤α<1 or A0≤B0,α>1, A0ot≤B0.​

For α≥0\alpha\ge0α≥0, αe1\alpha e1αe1, the Rényi entropy is (α−1)−1log⁡Sfα(A∥B)(\alpha-1)^{-1}\log S_{f_\alpha}(A\Vert B)(α−1)−1logSfα​​(A∥B) with the paper's extended logarithm. Also Sxlog⁡x(A∥B)S_{x\log x}(A\Vert B)Sxlogx​(A∥B) is Tr⁡A(log⁡∗A−log⁡∗B)\operatorname{Tr}A(\log^*A-\log^*B)TrA(log∗A−log∗B) on supported pairs and +∞+\infty+∞ otherwise. These are Petz, rather than sandwiched, formulas, without trace normalization.

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]

/-- Example 2.7, PDF pp.6–7: support powers, unnormalized Petz Rényi entropy,
and x log x. The alpha=1 identity is retained. -/
theorem example_2_7 [Nonempty I] (A B : Mat I) (hA : A.PosSemidef) (hB : B.PosSemidef) :
    (∀ a : ℝ, 0 ≤ a →
      div (supportScalarPower a) A B = (tr (supportPower A a * supportPower B (1 - a)) : EReal) +
        recessionSlope (supportScalarPower a) * (tr (A * (1 - supp B)) : EReal)) ∧
    (∀ a ∈ Ico (0 : ℝ) 1,
      div (supportScalarPower a) A B = (tr (supportPower A a * supportPower B (1 - a)) : EReal)) ∧
    (∀ a : ℝ, 1 < a → div (supportScalarPower a) A B =
      if Supports A B then (tr (supportPower A a * supportPower B (1 - a)) : EReal) else ⊤) ∧
    (∀ a : ℝ, 0 ≤ a → a ≠ 1 →
      petzRenyi a A B = (((a - 1)⁻¹ : ℝ) : EReal) *
        (if div (supportScalarPower a) A B = ⊤ then ⊤ else logNonnegativeExtended (div (supportScalarPower a) A B).toReal)) ∧
    div (fun x => x * Real.log x) A B = relativeEntropy A B := by sorry

end QAlgorithms.Papers.Independent.Support50.P12

end
Source
Quantum f-divergences and error correction, arXiv:1008.2529v6, PDF p. 6, Example 2.7
Read-back

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

Let III be a finite, nonempty index type with decidable equality, and let A,BA,BA,B be positive semidefinite complex matrices indexed by III. Write tr⁡X=Re⁡Tr⁡X\operatorname{tr}X=\operatorname{Re}\operatorname{Tr}XtrX=ReTrX, X[s]=cfc⁡(x↦0 if x=0, otherwise xs)(X)X^{[s]}=\operatorname{cfc}(x\mapsto 0\text{ if }x=0,\text{ otherwise }x^s)(X)X[s]=cfc(x↦0 if x=0, otherwise xs)(X), and PX=X[0]P_X=X^{[0]}PX​=X[0]; thus zero eigenvalues contribute zero even when s=0s=0s=0 or s<0s<0s<0. Write A≼suppBA\preccurlyeq_{\mathrm{supp}}BA≼supp​B for ker⁡B⊆ker⁡A\ker B\subseteq\ker AkerB⊆kerA. For positive semidefinite U,VU,VU,V, let

qf(U,V)=∑λ∈spec⁡U∑μ∈spec⁡Vμf(λ/μ)Re⁡Tr⁡(Pλ(U)Pμ(V)),Df(U,V)={qf(U,V),ker⁡V⊆ker⁡U,lim⁡ε↓0qf(U,V+εI),otherwise,q_f(U,V)=\sum_{\lambda\in\operatorname{spec}U}\sum_{\mu\in\operatorname{spec}V} \mu f(\lambda/\mu)\operatorname{Re}\operatorname{Tr}(P_\lambda(U)P_\mu(V)), \qquad D_f(U,V)= \begin{cases} q_f(U,V),&\ker V\subseteq\ker U,\\ \displaystyle\lim_{\varepsilon\downarrow0}q_f(U,V+\varepsilon I),&\text{otherwise}, \end{cases}qf​(U,V)=λ∈specU∑​μ∈specV∑​μf(λ/μ)ReTr(Pλ​(U)Pμ​(V)),Df​(U,V)=⎩⎨⎧​qf​(U,V),ε↓0lim​qf​(U,V+εI),​kerV⊆kerU,otherwise,​

where the spectral sums use distinct eigenvalues and their orthogonal eigenspace projections, division by zero is defined as zero, and the second branch uses the formal total limit-selection operation in the extended real line. Put fa(x)=0f_a(x)=0fa​(x)=0 when x=0x=0x=0 and fa(x)=xaf_a(x)=x^afa​(x)=xa otherwise, and let waw_awa​ be the extended-real limit-selection value of fa(x)/xf_a(x)/xfa​(x)/x as x→+∞x\to+\inftyx→+∞. Define L(t)=log⁡tL(t)=\log tL(t)=logt for t>0t>0t>0 and L(t)=−∞L(t)=-\inftyL(t)=−∞ otherwise, and

Ra(U,V)={1a−1L ⁣(tr⁡(U[a]V[1−a])),ker⁡V⊆ker⁡U or a<1,+∞,otherwise.R_a(U,V)= \begin{cases} \dfrac{1}{a-1}L\!\left(\operatorname{tr}(U^{[a]}V^{[1-a]})\right), &\ker V\subseteq\ker U\ \text{or }a<1,\\ +\infty,&\text{otherwise}. \end{cases}Ra​(U,V)=⎩⎨⎧​a−11​L(tr(U[a]V[1−a])),+∞,​kerV⊆kerU or a<1,otherwise.​

Then all of the following hold:

Dfa(A,B)=tr⁡(A[a]B[1−a])+wa tr⁡(A(I−PB))(a≥0),Dfa(A,B)=tr⁡(A[a]B[1−a])(0≤a<1),Dfa(A,B)={tr⁡(A[a]B[1−a]),ker⁡B⊆ker⁡A,+∞,otherwise(a>1),Ra(A,B)=1a−1{+∞,Dfa(A,B)=+∞,L ⁣(Dfa(A,B).toReal),otherwise(a≥0, a≠1).\begin{aligned} D_{f_a}(A,B) &=\operatorname{tr}(A^{[a]}B^{[1-a]}) +w_a\,\operatorname{tr}(A(I-P_B)) &&(a\ge0),\\ D_{f_a}(A,B) &=\operatorname{tr}(A^{[a]}B^{[1-a]}) &&(0\le a<1),\\ D_{f_a}(A,B) &=\begin{cases} \operatorname{tr}(A^{[a]}B^{[1-a]}),&\ker B\subseteq\ker A,\\ +\infty,&\text{otherwise} \end{cases} &&(a>1),\\ R_a(A,B) &=\dfrac{1}{a-1} \begin{cases} +\infty,&D_{f_a}(A,B)=+\infty,\\ L\!\left(D_{f_a}(A,B).\mathrm{toReal}\right),&\text{otherwise} \end{cases} &&(a\ge0,\ a\ne1). \end{aligned}Dfa​​(A,B)Dfa​​(A,B)Dfa​​(A,B)Ra​(A,B)​=tr(A[a]B[1−a])+wa​tr(A(I−PB​))=tr(A[a]B[1−a])={tr(A[a]B[1−a]),+∞,​kerB⊆kerA,otherwise​=a−11​{+∞,L(Dfa​​(A,B).toReal),​Dfa​​(A,B)=+∞,otherwise​​​(a≥0),(0≤a<1),(a>1),(a≥0, a=1).​

Here extended-real values are used throughout these equalities, and toReal\mathrm{toReal}toReal sends either infinity to zero. Finally,

Dx↦xlog⁡x(A,B)={Re⁡Tr⁡ ⁣(cfc⁡(x↦xlog⁡x)(A)−A cfc⁡(log⁡)(B)),ker⁡B⊆ker⁡A,+∞,otherwise.D_{x\mapsto x\log x}(A,B)= \begin{cases} \operatorname{Re}\operatorname{Tr}\!\left(\operatorname{cfc}(x\mapsto x\log x)(A) -A\,\operatorname{cfc}(\log)(B)\right),&\ker B\subseteq\ker A,\\ +\infty,&\text{otherwise}. \end{cases}Dx↦xlogx​(A,B)={ReTr(cfc(x↦xlogx)(A)−Acfc(log)(B)),+∞,​kerB⊆kerA,otherwise.​

No trace normalization or invertibility is assumed; either matrix may be zero.

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