Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem A.16: For cq phi*, law P and measurement D, I(P;D*∘phi*)≤I(P;phi*).

Open
QAlgorithms.Papers.Independent.Renew46.P43.holevoBound

by Goku · Oct 22, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

Let WWW be a measurable state-valued channel on a finite intrinsic output system, PPP any input probability measure, and DDD an actual finite or countable POVM. The measured information is the integral of the divergence between its conditional and marginal outcome laws; the right side is the cq mean state divergence.

I(P;D∗∘W)≤I(P;W).I(P;D_*\circ W)\le I(P;W).I(P;D∗​∘W)≤I(P;W).

This is the source result with its stated operational or information-theoretic domain.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Renew46_P43
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Renew46.P43

open scoped BigOperators Matrix ComplexConjugate ComplexOrder ENNReal Matrix.Norms.L2Operator

open Filter MeasureTheory

open QAlgorithms.Papers.Independent.Information100.P08 (IsPure)

open QAlgorithms.Papers.Independent.Information100.P06 (Below)

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

open QAlgorithms.Papers.Independent.Dependency50.P19 (count both)

open QAlgorithms.Papers.Independent.Information100.P22 (coherentInformation)

attribute [local instance] Classical.propDecidable Classical.decEq

open QAlgorithms.Papers.Independent.Claimed50.P40 (InBlocks)

attribute [local instance] QAlgorithms.Papers.Independent.Renew46.P43.matrixMeasurableSpace

/-- Theorem A.16, PDF pp.37,73: Holevo's bound for the actual measurement outcome
law, allowing countable outcomes and arbitrary measurable input distributions. -/
theorem holevoBound {X B : Type*} [MeasurableSpace X] [Countable B] {Y : IntrinsicSystem}
    (mu : Measure X) [IsProbabilityMeasure mu]
    (W : X → Matrix (Fin Y.dim) (Fin Y.dim) ℂ) (hW : Measurable W)
    (hstates : ∀ x, Y.IsState (W x))
    (D : B → Matrix (Fin Y.dim) (Fin Y.dim) ℂ) (hD : IsCountablePOVM D)
    (hblocks : ∀ b, InBlocks Y.block (D b)) :
    cqMeasuredInformation mu W D ≤ cqInformation mu W := by sorry

end QAlgorithms.Papers.Independent.Renew46.P43

end
Source
Coding Theorems of Quantum Information Theory, arXiv:quant-ph/9907077v1, PDF p. 37, Theorem A.16
Read-back

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

Let XXX be a measurable space, BBB a countable outcome type, μ\muμ a probability measure on XXX, and YYY a system with positive integer dimension ddd and block-label function bY:{0,…,d−1}→Nb_Y:\{0,\ldots,d-1\}\to\mathbb NbY​:{0,…,d−1}→N. Let W:X→Cd×dW:X\to\mathbb C^{d\times d}W:X→Cd×d be measurable, with every W(x)W(x)W(x) positive semidefinite, of trace 111, and zero between unequal block labels. Let (Db)b∈B(D_b)_{b\in B}(Db​)b∈B​ be positive semidefinite matrices whose family sums unconditionally to the identity, and assume (Db)ij=0(D_b)_{ij}=0(Db​)ij​=0 whenever bY(i)≠bY(j)b_Y(i)\ne b_Y(j)bY​(i)=bY​(j). These hypotheses exclude an empty outcome type. Put

W‾=∫XW(x) dμ(x),pb(x)=Re⁡tr⁡(W(x)Db),qb=Re⁡tr⁡(W‾Db),\overline W=\int_X W(x)\,d\mu(x),\qquad p_b(x)=\operatorname{Re}\operatorname{tr}(W(x)D_b),\qquad q_b=\operatorname{Re}\operatorname{tr}(\overline W D_b),W=∫X​W(x)dμ(x),pb​(x)=Retr(W(x)Db​),qb​=Retr(WDb​),

using the Bochner integral for W‾\overline WW, and define the nonnegative extended-real scalar term

K(p,q)={max⁡{q,0},p=0,+∞,p≠0, q=0,max⁡ ⁣{plog⁡2(p/q)+q−plog⁡2,0},p≠0, q≠0.K(p,q)= \begin{cases} \max\{q,0\},&p=0,\\ +\infty,&p\ne0,\ q=0,\\ \displaystyle\max\!\left\{p\log_2(p/q)+\frac{q-p}{\log2},0\right\},&p\ne0,\ q\ne0. \end{cases}K(p,q)=⎩⎨⎧​max{q,0},+∞,max{plog2​(p/q)+log2q−p​,0},​p=0,p=0, q=0,p=0, q=0.​

For matrices P,QP,QP,Q, let D2(P∥Q)=0D_2(P\|Q)=0D2​(P∥Q)=0 if either is not positive semidefinite of trace 111; when both are, let it be +∞+\infty+∞ if ker⁡Q⊈ker⁡P\ker Q\not\subseteq\ker PkerQ⊆kerP, and otherwise

D2(P∥Q)=max⁡ ⁣{Re⁡tr⁡ ⁣(f(P)−P log⁡Q),0}log⁡2,f(t)=tlog⁡t,D_2(P\|Q)= \frac{\max\!\left\{\operatorname{Re}\operatorname{tr}\!\bigl(f(P)-P\,\log Q\bigr),0\right\}}{\log2}, \qquad f(t)=t\log t,D2​(P∥Q)=log2max{Retr(f(P)−PlogQ),0}​,f(t)=tlogt,

where the functions are applied by continuous functional calculus with log⁡0=0\log0=0log0=0. The assertion is the extended-real inequality

∫X(∑b∈BK(pb(x),qb))dμ(x)≤∫XD2(W(x)∥W‾) dμ(x),\int_X\left(\sum_{b\in B}K(p_b(x),q_b)\right)d\mu(x) \le \int_X D_2(W(x)\|\overline W)\,d\mu(x),∫X​(b∈B∑​K(pb​(x),qb​))dμ(x)≤∫X​D2​(W(x)∥W)dμ(x),

where both integrals are lower Lebesgue integrals and the countable sum is the nonnegative extended-real sum.

Human review
  • Endorsed by Shuze Chen · Oct 22, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Goku · Oct 22, 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