Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Linear quantum entropy power inequality, Eq. (4) = Eq. (18)

Open
QAlgorithms.Papers.Independent.Intakenext100.P62.entropyBeamSplitterClosed

by Gohan · Oct 31, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

Let n≥1n\ge1n≥1 and let ρX,ρY\rho_X,\rho_YρX​,ρY​ be arbitrary normalized states on the full nnn-mode bosonic Fock spaces. Their joint input is the product ρX⊗ρY\rho_X\otimes\rho_YρX​⊗ρY​; entanglement within either input is unrestricted. Let BλB_\lambdaBλ​ retain the port with quadratures λRX+1−λRY\sqrt\lambda R_X+\sqrt{1-\lambda}R_Yλ​RX​+1−λ​RY​. Write SSS for extended nonnegative von Neumann entropy in nats.

For every 0≤λ≤10\le\lambda\le10≤λ≤1,

S(Bλ(ρX⊗ρY))≥λS(ρX)+(1−λ)S(ρY).S(B_\lambda(\rho_X\otimes\rho_Y))\ge\lambda S(\rho_X)+(1-\lambda)S(\rho_Y).S(Bλ​(ρX​⊗ρY​))≥λS(ρX​)+(1−λ)S(ρY​).

No energy, entropy-finiteness, Gaussianity or smoothness assumption is imposed. Infinite entropy is retained, with 0⋅∞=00\cdot\infty=00⋅∞=0. This controls entropy production under physical mixing, including the endpoints.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P62
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Intakenext100.P62

open scoped ENNReal BigOperators

open QAlgorithms.Papers.Independent.Dependency50.P33 (State productCoeff)

open QAlgorithms.Papers.Independent.Frontier18.P07 (extendedEntropyNats)

open QAlgorithms.Papers.Independent.Claimed50.P19 (FockIndex)

open QAlgorithms.Papers.Independent.Claimed50.P39 (g displacementNoiseChannel)

open QAlgorithms.Papers.Independent.Frontier18.P18 (operatorCoeff boundedOperator)

open QAlgorithms.Papers.Independent.Gap48.P26 (weightedPhotonExpectation)

open QAlgorithms.Papers.Independent.Bridge50.P28
  (classicalPhotonCapacityBits thermalNoiseChannelKernel coherentSignalRateBits)

open QAlgorithms.Papers.Independent.Next46.P17
  (beamSplitter bosonicEntropyPower diffusionChannel TotalFisherRegular totalFisher)

/-- PDF pp.2,7, (4)/(18): linear quantum EPI for arbitrary independent
normalized full n-mode states. No Gaussianity, finite entropy, finite energy or
smoothness hypothesis is imposed. ENNReal uses 0*infinity=0 at lambda=0,1. -/
theorem entropyBeamSplitterClosed {n : ℕ} (hn : 0 < n)
    (lam : ℝ) (hlam0 : 0 ≤ lam) (hlam1 : lam ≤ 1)
    (rhoX rhoY : State (FockIndex n)) :
    ENNReal.ofReal lam * extendedEntropyNats rhoX.coefficients +
      ENNReal.ofReal (1 - lam) * extendedEntropyNats rhoY.coefficients ≤
    extendedEntropyNats (beamSplitter lam
      (productCoeff rhoX.coefficients rhoY.coefficients)) := by sorry

end QAlgorithms.Papers.Independent.Intakenext100.P62

end
Source
Limits on classical communication from quantum entropy power inequalities, arXiv:1205.3407v2, PDF p. 2, Linear quantum entropy power inequality, Eq. (4) = Eq. (18)
Read-back

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

For every integer n>0n>0n>0, every real 0≤λ≤10\le\lambda\le10≤λ≤1, and every two states X,YX,YX,Y on Fn={0,…,n−1}→NF_n=\{0,\ldots,n-1\}\to\mathbb NFn​={0,…,n−1}→N, the theorem asserts

[λ]+H(X)+[1−λ]+H(Y)≤H(Bλ(X,Y))in [0,∞].[\lambda]_+H(X)+[1-\lambda]_+H(Y)\le H(B_\lambda(X,Y)) \quad\text{in }[0,\infty].[λ]+​H(X)+[1−λ]+​H(Y)≤H(Bλ​(X,Y))in [0,∞].

Here [x]+=max⁡(x,0)[x]_+=\max(x,0)[x]+​=max(x,0), embedded in [0,∞][0,\infty][0,∞], and the entropy of any complex matrix AAA on a label set LLL is

H(A)=sup⁡s⊆Ls finite[(Re⁡tr⁡As)log⁡(Re⁡tr⁡As)−Re⁡tr⁡ ⁣(Aslog⁡fcAs)]+,H(A)=\sup_{\substack{s\subseteq L\\s\text{ finite}}} \left[ (\operatorname{Re}\operatorname{tr}A_s)\log(\operatorname{Re}\operatorname{tr}A_s) -\operatorname{Re}\operatorname{tr}\!\left(A_s\log_{\mathrm{fc}}A_s\right) \right]_+,H(A)=s⊆Ls finite​sup​[(RetrAs​)log(RetrAs​)−Retr(As​logfc​As​)]+​,

where AsA_sAs​ is the principal submatrix and log⁡fc\log_{\mathrm{fc}}logfc​ is continuous functional calculus using the total real logarithm. The empty submatrix is included. The output matrix is defined by

Bλ(X,Y)ab=∑k∈Fn′∑u∈Fn2′∑v∈Fn2′βλ((a,k),u)Xu1v1Yu2v2βλ((b,k),v),βλ(o,p)=∏j=0n−1hλ(o1(j),o2(j);p1(j),p2(j)),B_\lambda(X,Y)_{ab} =\sum_{k\in F_n}'\sum_{u\in F_n^2}'\sum_{v\in F_n^2}' \beta_\lambda((a,k),u)X_{u_1v_1}Y_{u_2v_2} \beta_\lambda((b,k),v), \qquad \beta_\lambda(o,p)=\prod_{j=0}^{n-1} h_\lambda(o_1(j),o_2(j);p_1(j),p_2(j)),Bλ​(X,Y)ab​=k∈Fn​∑′​u∈Fn2​∑′​v∈Fn2​∑′​βλ​((a,k),u)Xu1​v1​​Yu2​v2​​βλ​((b,k),v),βλ​(o,p)=j=0∏n−1​hλ​(o1​(j),o2​(j);p1​(j),p2​(j)),

with

hλ(r,s;p,q)={r! s!p! q!∑i=0p∑j=0q1{i+j=r}(pi)(qj)(λ)i(1−λ)p−i(1−λ)j(−λ)q−j,r+s=p+q,0,otherwise.h_\lambda(r,s;p,q)= \begin{cases} \displaystyle \sqrt{\frac{r!\,s!}{p!\,q!}} \sum_{i=0}^{p}\sum_{j=0}^{q} \mathbf1_{\{i+j=r\}}\binom pi\binom qj (\sqrt\lambda)^i(\sqrt{1-\lambda})^{p-i} (\sqrt{1-\lambda})^j(-\sqrt\lambda)^{q-j}, &r+s=p+q,\\ 0,&\text{otherwise}. \end{cases}hλ​(r,s;p,q)=⎩⎨⎧​p!q!r!s!​​i=0∑p​j=0∑q​1{i+j=r}​(ip​)(jq​)(λ​)i(1−λ​)p−i(1−λ​)j(−λ​)q−j,0,​r+s=p+q,otherwise.​

Each state AAA admits a countable expansion Aab=∑r′ur(a)vr(b)‾A_{ab}=\sum_r'u_r(a)\overline{v_r(b)}Aab​=∑r′​ur​(a)vr​(b)​ with ur,vr∈ℓ2(Fn;C)u_r,v_r\in\ell^2(F_n;\mathbb C)ur​,vr​∈ℓ2(Fn​;C) and ∑r∥ur∥∥vr∥<∞\sum_r\|u_r\|\|v_r\|<\infty∑r​∥ur​∥∥vr​∥<∞, has real nonnegative finite quadratic forms, and has a summable real diagonal with sum 111. No finite-energy or finite-entropy assumption is made. Both endpoints λ=0,1\lambda=0,1λ=0,1 are included, and 0⋅∞=00\cdot\infty=00⋅∞=0; n=0n=0n=0 is excluded. No additional summability hypothesis is imposed on the beam-splitter sums, which use the total complex infinite-sum convention.

Human review
  • Endorsed by Shuze Chen · Oct 31, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Gohan · Oct 31, 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