Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Corollary 2.3

Proved
QAlgorithms.Papers.Independent.Support50.P12.corollary_2_3

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.

For spectral projections PaP_aPa​ of AAA and QbQ_bQb​ of BBB, all three formulas hold:

Sf(A∥B)=Qf(A,B)+ω(f)Tr⁡A(I−B0)=f(0)Tr⁡B+Qf−f(0)(A,B)+ω(f)Tr⁡A(I−B0)=∑a[∑be0bf(a/b)Tr⁡PaQb+aω(f)Tr⁡PaQ0].\begin{aligned}S_f(A\Vert B)&=Q_f(A,B)+\omega(f)\operatorname{Tr}A(I-B^0)\\&=f(0)\operatorname{Tr}B+Q_{f-f(0)}(A,B)+\omega(f)\operatorname{Tr}A(I-B^0)\\&=\sum_a\left[\sum_{b e0}bf(a/b)\operatorname{Tr}P_aQ_b+a\omega(f)\operatorname{Tr}P_aQ_0\right].\end{aligned}Sf​(A∥B)​=Qf​(A,B)+ω(f)TrA(I−B0)=f(0)TrB+Qf−f(0)​(A,B)+ω(f)TrA(I−B0)=a∑​[be0∑​bf(a/b)TrPa​Qb​+aω(f)TrPa​Q0​].​

Moreover, Sf(A∥B)=Qf(A,B)S_f(A\Vert B)=Q_f(A,B)Sf​(A∥B)=Qf​(A,B) exactly when A0≤B0A^0\le B^0A0≤B0 or ω(f)=0\omega(f)=0ω(f)=0. Products 0(±∞)0(\pm\infty)0(±∞) mean zero. These formulas handle singular inputs without normalizing them.

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]

/-- Corollary 2.3, PDF p.6: all three expansions and the exact vanishing criterion. -/
theorem corollary_2_3 [Nonempty I] (f : ℝ → ℝ) (hf : DivergenceGenerator f)
    (A B : Mat I) (hA : A.PosSemidef) (hB : B.PosSemidef) :
    div f A B = (quasi f A B : EReal) + recessionSlope f * (tr (A * (1 - supp B)) : EReal) ∧
    div f A B = ((f 0 * tr B + quasi (fun x => f x - f 0) A B : ℝ) : EReal) +
      recessionSlope f * (tr (A * (1 - supp B)) : EReal) ∧
    div f A B = ∑ a ∈ spec A,
      (((∑ b ∈ (spec B).filter (fun b => b ≠ 0),
        b * f (a / b) * tr (spectralProjector A a * spectralProjector B b) : ℝ) : EReal) +
        (a : EReal) * recessionSlope f * (tr (spectralProjector A a * spectralProjector B 0) : EReal)) ∧
    (div f A B = (quasi f A B : EReal) ↔ Supports A B ∨ recessionSlope f = 0) := by sorry

end QAlgorithms.Papers.Independent.Support50.P12

end
Source
Quantum f-divergences and error correction, arXiv:1008.2529v6, PDF p. 6, Corollary 2.3
Read-back

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

Let III be a nonempty finite type with decidable equality, let f:R→Rf:\mathbb R\to\mathbb Rf:R→R be continuous on (0,∞)(0,\infty)(0,∞), and assume that f(x)/xf(x)/xf(x)/x converges in the extended reals as x→+∞x\to+\inftyx→+∞ to its selected limit LfL_fLf​. For positive semidefinite complex I×II\times II×I matrices A,BA,BA,B, write tr⁡X=Re⁡Tr⁡X\operatorname{tr}X=\operatorname{Re}\operatorname{Tr}XtrX=ReTrX, let sp⁡X\operatorname{sp}XspX denote distinct eigenvalues, let PaXP^X_aPaX​ be the eigenspace projection, and let SBS_BSB​ equal 111 on the nonzero eigenspaces of BBB and 000 on its kernel. Define

qg(X,Y)=Re⁡Tr⁡ ⁣[Y∑a∈sp⁡X∑b∈sp⁡Yg(a/b)PaXYPbY],q_g(X,Y)=\operatorname{Re}\operatorname{Tr}\!\left[ \sqrt Y\sum_{a\in\operatorname{sp}X}\sum_{b\in\operatorname{sp}Y} g(a/b)P^X_a\sqrt YP^Y_b \right],qg​(X,Y)=ReTr​Y​a∈spX∑​b∈spY∑​g(a/b)PaX​Y​PbY​​,

using total real division, and define Df(X,Y)D_f(X,Y)Df​(X,Y) as qf(X,Y)q_f(X,Y)qf​(X,Y) in the extended reals when ker⁡Y⊆ker⁡X\ker Y\subseteq\ker XkerY⊆kerX, and otherwise as the total selected right limit of qf(X,Y+ϵ1)q_f(X,Y+\epsilon1)qf​(X,Y+ϵ1) at zero. The following three equalities and equivalence hold:

Df(A,B)=qf(A,B)+Lf tr⁡(A(1−SB)),Df(A,B)=f(0)tr⁡B+q f−f(0)(A,B)+Lf tr⁡(A(1−SB)),Df(A,B)=∑a∈sp⁡A[∑b∈sp⁡Bb≠0bf(a/b)tr⁡(PaAPbB)+aLf tr⁡(PaAP0B)],Df(A,B)=qf(A,B)  ⟺  ker⁡B⊆ker⁡A or Lf=0.\begin{aligned} D_f(A,B) &=q_f(A,B)+L_f\,\operatorname{tr}(A(1-S_B)),\\ D_f(A,B) &=f(0)\operatorname{tr}B+q_{\,f-f(0)}(A,B) +L_f\,\operatorname{tr}(A(1-S_B)),\\ D_f(A,B) &=\sum_{a\in\operatorname{sp}A}\left[ \sum_{\substack{b\in\operatorname{sp}B\\b\ne0}} b f(a/b)\operatorname{tr}(P^A_aP^B_b) +aL_f\,\operatorname{tr}(P^A_aP^B_0) \right],\\ D_f(A,B)=q_f(A,B) &\iff \ker B\subseteq\ker A\ \text{or}\ L_f=0. \end{aligned}Df​(A,B)Df​(A,B)Df​(A,B)Df​(A,B)=qf​(A,B)​=qf​(A,B)+Lf​tr(A(1−SB​)),=f(0)trB+qf−f(0)​(A,B)+Lf​tr(A(1−SB​)),=a∈spA∑​​b∈spBb=0​∑​bf(a/b)tr(PaA​PbB​)+aLf​tr(PaA​P0B​)​,⟺kerB⊆kerA or Lf​=0.​

Here f−f(0)f-f(0)f−f(0) means x↦f(x)−f(0)x\mapsto f(x)-f(0)x↦f(x)−f(0); finite real quantities are embedded in the extended reals, and P0BP^B_0P0B​ is the projection onto ker⁡B\ker BkerB, including when 0∉sp⁡B0\notin\operatorname{sp}B0∈/spB. All extended-real arithmetic is as defined, including products involving infinite LfL_fLf​. Zero matrices are allowed, and no trace normalization is assumed.

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