Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§3, Eqs. (27),(28): displacement does not lower GRE

Open
QAlgorithms.Papers.Independent.Intakenext100.P46.displacement_nonDecrease

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

prove2quantum

Let ρ\rhoρ and σ\sigmaσ be centered Gaussian density operators on the same positive finite number of modes, represented by any two registers (either may be empty), and suppose σ\sigmaσ has a finite positive exponential matrix MMM. For real grouped phase vector zzz, let σz=W(z)σW(z)∗\sigma_z=W(z)\sigma W(z)^*σz​=W(z)σW(z)∗, where W(z)=exp⁡(iFTz)W(z)=\exp(iF^Tz)W(z)=exp(iFTz). Its mean is −Δz-\Delta z−Δz and its covariance equals that of σ\sigmaσ. Then

D(ρ∥σz)=D(ρ∥σ)+12(Δz)TM(Δz)≥D(ρ∥σ).D(\rho\Vert\sigma_z)=D(\rho\Vert\sigma)+\tfrac12(\Delta z)^TM(\Delta z)\ge D(\rho\Vert\sigma).D(ρ∥σz​)=D(ρ∥σ)+21​(Δz)TM(Δz)≥D(ρ∥σ).

A common displacement first centers a noncentered input. The correction is positive, not an assertion that all displaced competitors have equal divergence. The displaced density is constructed by the bounded Weyl kernel on the completed oscillator space; its existence, Gaussian characteristic function, and preservation of separability are conclusions.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P46
Formal statement
noncomputable section

namespace QAlgorithms.Papers.Independent.Intakenext100.P46

open scoped BigOperators ENNReal ComplexOrder Matrix

attribute [local instance] Classical.propDecidable Classical.decEq

open QAlgorithms.RealRegisters (Hilbert)

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

open QAlgorithms.Papers.Independent.Dependency50.P13
  (finiteRelativeEntropy compression entropyWeight)

open QAlgorithms.Papers.Independent.Claimed50.P47
  (Occupation bipartiteCharacteristic bipartiteWeyl gaussianStateCharacteristic)

-- These accepted definitions have the same completed Hilbert-space meaning.

open QAlgorithms.Papers.Independent.Next46.P13 (coordinateBasis)

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

open QAlgorithms.Papers.Independent.Claimed50.P41 (IsDiagonal)

/-- Chen, PDF p.6, (27)–(28): the constructed Weyl displacement is an actual
Gaussian density, preserves separability, and has the positive quadratic cost.
A common local displacement first centers nonzero input means. A single mode
is included by allowing one empty register, with a nonempty total mode set. -/
theorem displacement_nonDecrease {A B : Type} [Fintype A] [Fintype B]
    [Nonempty (A ⊕ B)]
    (rho sigma : State (Occupation A × Occupation B))
    (alpha beta M : ModeMatrix (A ⊕ B)) (z : GroupedPhase (A ⊕ B) → ℝ)
    (hrho : HasGaussian rho 0 alpha) (hsigma : HasGaussian sigma 0 beta)
    (hfinite : CMEM beta M) :
    ∃ sigmaz : State (Occupation A × Occupation B),
      sigmaz.coefficients = weylDisplacedDensity sigma z ∧
      HasGaussian sigmaz (-(groupedCCR (A ⊕ B)).mulVec z) beta ∧
      (IsSeparable sigmaz.coefficients ↔ IsSeparable sigma.coefficients) ∧
      0 ≤ displacementPenalty M z ∧
      operatorRelativeEntropy rho sigmaz = operatorRelativeEntropy rho sigma + ENNReal.ofReal (displacementPenalty M z) ∧
      operatorRelativeEntropy rho sigma ≤ operatorRelativeEntropy rho sigmaz := by sorry

end QAlgorithms.Papers.Independent.Intakenext100.P46

end
Source
Gaussian relative entropy of entanglement, arXiv:quant-ph/0402109v3, PDF p. 6, §3, Eqs. (27),(28): displacement does not lower GRE
Read-back

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

Let A,BA,BA,B be finite types with I=A⊔BI=A\sqcup BI=A⊔B nonempty, allowing either individual type to be empty, and write OE=NE\mathcal O_E=\mathbb N^EOE​=NE, J=(01−10)J=\begin{pmatrix}0&1\\-1&0\end{pmatrix}J=(0−1​10​) on I⊔II\sqcup II⊔I, and D(h)=diag⁡(diag⁡h,diag⁡h)D(h)=\operatorname{diag}(\operatorname{diag}h,\operatorname{diag}h)D(h)=diag(diagh,diagh). Let ρ,σ\rho,\sigmaρ,σ be states indexed by OA×OB\mathcal O_A\times\mathcal O_BOA​×OB​: their coefficient matrices admit expansions Rij=∑rur(i)vr(j)‾R_{ij}=\sum_r u_r(i)\overline{v_r(j)}Rij​=∑r​ur​(i)vr​(j)​ with ur,vr∈ℓ2u_r,v_r\in\ell^2ur​,vr​∈ℓ2 and ∑r∥ur∥∥vr∥<∞\sum_r\|u_r\|\|v_r\|<\infty∑r​∥ur​∥∥vr​∥<∞, every finite quadratic form is real and nonnegative, and their real diagonals are summable with sum 111. Let α,β,M\alpha,\beta,Mα,β,M be real matrices indexed by I⊔II\sqcup II⊔I, and let z∈RI⊔Iz\in\mathbb R^{I\sqcup I}z∈RI⊔I. Assume ρ,σ\rho,\sigmaρ,σ have centered Gaussian characteristics with covariances α,β\alpha,\betaα,β, respectively. Explicitly, each covariance aaa is symmetric and a−i2Ja-\tfrac i2Ja−2i​J is positive semidefinite, and for every ξ\xiξ, every inner coefficient sum and the outer coefficient sum in

χR(ξ)=∑u∑vRuvW(ξ)vu=exp⁡ ⁣(−12ξTJTaJξ)\chi_R(\xi)=\sum_u\sum_vR_{uv}W(\xi)_{vu} =\exp\!\left(-\tfrac12\xi^{\mathsf T}J^{\mathsf T}aJ\xi\right)χR​(ξ)=u∑​v∑​Ruv​W(ξ)vu​=exp(−21​ξTJTaJξ)

are summable. Here W(ξ)uv=∏j∈Idujvj(−(ξL,j+iξR,j)/2)W(\xi)_{uv}=\prod_{j\in I}d_{u_jv_j}(-(\xi_{L,j}+i\xi_{R,j})/\sqrt2)W(ξ)uv​=∏j∈I​duj​vj​​(−(ξL,j​+iξR,j​)/2​), interpreting occupations across the two parts by restriction, and

dmn(w)=e−∣w∣2/2m! n!∑r=0min⁡(m,n)wm−r(−wˉ)n−rr!(m−r)!(n−r)!.d_{mn}(w)=e^{-|w|^2/2}\sqrt{m!\,n!} \sum_{r=0}^{\min(m,n)} \frac{w^{m-r}(-\bar w)^{n-r}}{r!(m-r)!(n-r)!}.dmn​(w)=e−∣w∣2/2m!n!​r=0∑min(m,n)​r!(m−r)!(n−r)!wm−r(−wˉ)n−r​.

Assume also that there exist an invertible real matrix SSS with SJST=JSJS^{\mathsf T}=JSJST=J and numbers gj>12g_j>\tfrac12gj​>21​ such that

β=SD(g)ST,M=S−TD ⁣(j↦log⁡2gj+12gj−1)S−1.\beta=SD(g)S^{\mathsf T},\qquad M=S^{-\mathsf T}D\!\left(j\mapsto\log\frac{2g_j+1}{2g_j-1}\right)S^{-1}.β=SD(g)ST,M=S−TD(j↦log2gj​−12gj​+1​)S−1.

For states R,QR,QR,Q, define D(R∥Q)D(R\Vert Q)D(R∥Q) as follows: replace each coefficient matrix by a chosen bounded operator TR,TQT_R,T_QTR​,TQ​ with that matrix, using zero if no such operator exists; take the supremum over all finite orthonormal families, including the empty family, of F(X,Y)F(X,Y)F(X,Y) for the compressions Xij=⟨vi,TRvj⟩X_{ij}=\langle v_i,T_Rv_j\rangleXij​=⟨vi​,TR​vj​⟩, Yij=⟨vi,TQvj⟩Y_{ij}=\langle v_i,T_Qv_j\rangleYij​=⟨vi​,TQ​vj​⟩, where

F(X,Y)={[Re⁡tr⁡((xlog⁡x)(X)−X(log⁡)(Y)−X+Y)]+,ker⁡Y⊆ker⁡X,+∞,otherwise.F(X,Y)= \begin{cases} \left[\operatorname{Re}\operatorname{tr} \bigl((x\log x)(X)-X(\log)(Y)-X+Y\bigr)\right]_+, &\ker Y\subseteq\ker X,\\ +\infty,&\text{otherwise}. \end{cases}F(X,Y)={[Retr((xlogx)(X)−X(log)(Y)−X+Y)]+​,+∞,​kerY⊆kerX,otherwise.​

Functions of matrices here use continuous functional calculus, and [a]+=max⁡(a,0)[a]_+=\max(a,0)[a]+​=max(a,0) is embedded into the extended nonnegative reals. Then there exists a state σz\sigma_zσz​ whose coefficients are exactly

(σz)ij=∑aW(Jz)ia(∑bσabW(Jz)jb‾).(\sigma_z)_{ij} =\sum_a W(Jz)_{ia}\left(\sum_b\sigma_{ab}\overline{W(Jz)_{jb}}\right).(σz​)ij​=a∑​W(Jz)ia​(b∑​σab​W(Jz)jb​​).

This state has covariance β\betaβ and mean d=−Jzd=-Jzd=−Jz: it satisfies the same physical-covariance and coefficient-summability conditions above, with characteristic

χσz(ξ)=exp⁡ ⁣(−12ξTJTβJξ+iξTJ(−Jz)).\chi_{\sigma_z}(\xi) =\exp\!\left(-\tfrac12\xi^{\mathsf T}J^{\mathsf T}\beta J\xi +i\xi^{\mathsf T}J(-Jz)\right).χσz​​(ξ)=exp(−21​ξTJTβJξ+iξTJ(−Jz)).

Furthermore, σz\sigma_zσz​ is separable if and only if σ\sigmaσ is separable. Here separability means belonging to the following closure of the real convex hull of coefficient products Raa′Qbb′R_{aa'}Q_{bb'}Raa′​Qbb′​ of states on OA,OB\mathcal O_A,\mathcal O_BOA​,OB​: a matrix XXX belongs to that closure precisely when for every ε>0\varepsilon>0ε>0 there is a matrix YYY in that convex hull for which ∥X−Y∥nuc<ε\|X-Y\|_{\mathrm{nuc}}<\varepsilon∥X−Y∥nuc​<ε, where

∥K∥nuc=inf⁡Kij=∑rur(i)vr(j)‾∑r∥ur∥∥vr∥<∞[∑r∥ur∥∥vr∥]+,\|K\|_{\mathrm{nuc}} =\inf_{\substack{K_{ij}=\sum_r u_r(i)\overline{v_r(j)}\\ \sum_r\|u_r\|\|v_r\|<\infty}} \left[\sum_r\|u_r\|\|v_r\|\right]_+,∥K∥nuc​=Kij​=∑r​ur​(i)vr​(j)​∑r​∥ur​∥∥vr​∥<∞​inf​[r∑​∥ur​∥∥vr​∥]+​,

with value +∞+\infty+∞ if there is no such expansion. Finally, with

pM(z)=12∑i,j(Jz)iMij(Jz)j,p_M(z)=\tfrac12\sum_{i,j}(Jz)_iM_{ij}(Jz)_j,pM​(z)=21​i,j∑​(Jz)i​Mij​(Jz)j​,

one has

pM(z)≥0,D(ρ∥σz)=D(ρ∥σ)+[pM(z)]+,D(ρ∥σ)≤D(ρ∥σz).p_M(z)\ge0,\qquad D(\rho\Vert\sigma_z)=D(\rho\Vert\sigma)+[p_M(z)]_+,\qquad D(\rho\Vert\sigma)\le D(\rho\Vert\sigma_z).pM​(z)≥0,D(ρ∥σz​)=D(ρ∥σ)+[pM​(z)]+​,D(ρ∥σ)≤D(ρ∥σz​).

The vector zzz is unrestricted, including z=0z=0z=0; the equality and inequality are in the extended nonnegative reals and therefore also include infinite relative-entropy values.

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