Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Two separability certificates along the Gaussian covariance iteration

Open
QAlgorithms.Papers.Independent.Intakenext100.P68.separabilityCriterion

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

prove2quantum

Let n,m≥1n,m\ge1n,m≥1 be finite local mode counts, with interleaved quadratures (X1,P1,…)(X_1,P_1,\ldots)(X1​,P1​,…), [Xj,Pj]=i[X_j,P_j]=i[Xj​,Pj​]=i, and doubled covariance γab=2Re⁡⟨(Ra−da)(Rb−db)⟩\gamma_{ab}=2\operatorname{Re}\langle(R_a-d_a)(R_b-d_b)\rangleγab​=2Re⟨(Ra​−da​)(Rb​−db​)⟩. Thus vacuum covariance is III. Put Jr=⨁j=1r(0−110)J_r=\bigoplus_{j=1}^r\begin{pmatrix}0&-1\\1&0\end{pmatrix}Jr​=⨁j=1r​(01​−10​). A covariance matrix (CM) is real symmetric and satisfies γ−iJ⪰0\gamma-iJ\succeq0γ−iJ⪰0. All inequalities below are Loewner inequalities.

Let γ0=(A0C0C0TB0)\gamma_0=\begin{pmatrix}A_0&C_0\\C_0^T&B_0\end{pmatrix}γ0​=(A0​C0T​​C0​B0​​) be a physical n∣mn|mn∣m CM. Construct the sequence as follows: a nonphysical matrix is followed by zero; otherwise put Xj=Cj(Bj−iJ)+CjTX_j=C_j(B_j-iJ)^+C_j^TXj​=Cj​(Bj​−iJ)+CjT​, Aj+1=Bj+1=Aj−Re⁡XjA_{j+1}=B_{j+1}=A_j-\operatorname{Re}X_jAj+1​=Bj+1​=Aj​−ReXj​, and Cj+1=−Im⁡XjC_{j+1}=-\operatorname{Im}X_jCj+1​=−ImXj​. All inverses are spectral Moore–Penrose inverses. Every positive-index matrix has split n∣nn|nn∣n. Set Lj=Aj−∥Cj∥opIL_j=A_j-\|C_j\|_{\mathrm{op}}ILj​=Aj​−∥Cj​∥op​I for j≥1j\ge1j≥1, where the norm is Euclidean induced. Separability always means the trace-norm closed convex hull of full oscillator product densities, independently of covariance witness inequalities.

Let ρ\rhoρ be any actual normalized Gaussian density of covariance γ0\gamma_0γ0​ and arbitrary first moments. Then both implications hold:

(∃N≥1: AN⪰̸iJ)⟹ρ is not separable,(∃N≥1: LN⪰iJ)⟹ρ is separable.(\exists N\ge1:\ A_N\not\succeq iJ)\Longrightarrow\rho\text{ is not separable},\qquad (\exists N\ge1:\ L_N\succeq iJ)\Longrightarrow\rho\text{ is separable}.(∃N≥1: AN​⪰iJ)⟹ρ is not separable,(∃N≥1: LN​⪰iJ)⟹ρ is separable.

These are sound rejection and acceptance certificates. They do not assert finite termination for every input.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P68
Formal statement
noncomputable section

open MeasureTheory Filter

open scoped BigOperators Topology ComplexOrder

namespace QAlgorithms.Papers.Independent.Intakenext100.P68

attribute [local instance] Classical.propDecidable Classical.decEq

open QAlgorithms.Papers.Independent.Dependency50.P33
  (State PureState IsSeparable IsBarycenter productCoeff)

open QAlgorithms.Papers.Independent.Dependency50.P10 (complexify)

open QAlgorithms.Papers.Independent.Frontier30.P08 (QuadratureIndex)

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

open QAlgorithms.Papers.Independent.Claimed50.P47 (Occupation)

/-- PDF p.2, Theorem 1, BOTH certificates, for an actual normalized Gaussian
input with arbitrary first moments. Tail k means N=k+1 >= 1. No termination,
invertibility, local-witness-existence or separability assumption is added. -/
theorem separabilityCriterion {n m : ℕ} (hn : 0 < n) (hm : 0 < m)
    (g : CovarianceBlocks n m) (rho : State (Occupation (Fin n) × Occupation (Fin m)))
    (d : QuadratureIndex n ⊕ QuadratureIndex m → ℝ) (hG : HasGaussianCovariance rho g d) :
    ((∃ k : ℕ, ¬ Uncertainty (tail g k).A) → ¬ IsSeparable rho.coefficients) ∧
    ((∃ k : ℕ, Uncertainty (lowerWitness (tail g k))) →
      IsSeparable rho.coefficients) := by sorry

end QAlgorithms.Papers.Independent.Intakenext100.P68

end
Source
Separability Criterion for all bipartite Gaussian States, arXiv:quant-ph/0104050v1, PDF p. 2, Theorem 1 (Separability criterion)
Read-back

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

For every pair of positive natural numbers n,mn,mn,m, put Qr=Fin⁡r×{0,1}Q_r=\operatorname{Fin}r\times\{0,1\}Qr​=Finr×{0,1}, Sr=NFin⁡rS_r=\mathbb N^{\operatorname{Fin}r}Sr​=NFinr, and S=Sn×SmS=S_n\times S_mS=Sn​×Sm​. Take arbitrary real blocks g=(A,B,C)g=(A,B,C)g=(A,B,C) of sizes Qn,QmQ_n,Q_mQn​,Qm​, a state ρ\rhoρ with kernel RRR on SSS, and a real vector ddd on Qn⊔QmQ_n\sqcup Q_mQn​⊔Qm​. A state kernel KKK on any label set LLL here admits an expansion Kuv=∑rar(u)br(v)‾K_{uv}=\sum_r a_r(u)\overline{b_r(v)}Kuv​=∑r​ar​(u)br​(v)​ with ar,br∈ℓ2(L;C)a_r,b_r\in\ell^2(L;\mathbb C)ar​,br​∈ℓ2(L;C) and ∑r∥ar∥∥br∥<∞\sum_r\|a_r\|\|b_r\|<\infty∑r​∥ar​∥∥br​∥<∞, has real nonnegative quadratic forms on every finite subset, and has summable real diagonal of sum one. Define JrJ_rJr​ by same-mode entries J(a,0),(a,1)=−1J_{(a,0),(a,1)}=-1J(a,0),(a,1)​=−1, J(a,1),(a,0)=1J_{(a,1),(a,0)}=1J(a,1),(a,0)​=1, with other entries zero. Assume G=\begin{psmallmatrix}A&C\\C^{\mathsf T}&B\end{psmallmatrix} is symmetric and G−i(Jn⊕Jm)⪰0G-i(J_n\oplus J_m)\succeq0G−i(Jn​⊕Jm​)⪰0. Reorder G,dG,dG,d into Γ,d′\Gamma,d'Γ,d′ by putting all coordinate-000 entries before all coordinate-111 entries, and set \Omega=\begin{psmallmatrix}0&I\\-I&0\end{psmallmatrix}. The remaining Gaussian hypothesis is that for every real vector ξ\xiξ in this grouped order, with za=−(ξfirst,a+iξsecond,a)/2z_a=-(\xi_{\mathrm{first},a}+i\xi_{\mathrm{second},a})/\sqrt2za​=−(ξfirst,a​+iξsecond,a​)/2​, the inner and outer series below are summable and satisfy

∑u∑vRuvWξ(v,u)=exp⁡ ⁣(−14ξTΩTΓΩξ+iξTΩd′),\sum_u\sum_v R_{uv}W_\xi(v,u) =\exp\!\left(-\frac14\xi^{\mathsf T}\Omega^{\mathsf T}\Gamma\Omega\xi +i\xi^{\mathsf T}\Omega d'\right),u∑​v∑​Ruv​Wξ​(v,u)=exp(−41​ξTΩTΓΩξ+iξTΩd′),

where Wξ(u,v)W_\xi(u,v)Wξ​(u,v) is the product over both subsystems’ modes of

D(z)pq=e−∣z∣2/2p! q!∑j=0min⁡(p,q)zp−j(−z‾)q−jj!(p−j)!(q−j)!D(z)_{pq} =e^{-|z|^2/2}\sqrt{p!\,q!} \sum_{j=0}^{\min(p,q)} \frac{z^{p-j}(-\overline z)^{q-j}}{j!(p-j)!(q-j)!}D(z)pq​=e−∣z∣2/2p!q!​j=0∑min(p,q)​j!(p−j)!(q−j)!zp−j(−z)q−j​

at the corresponding occupation indices. To define the iteration, call a block triple h=(Ah,Bh,Ch)h=(A_h,B_h,C_h)h=(Ah​,Bh​,Ch​) physical when its full block matrix is symmetric and its complexification minus iii times the corresponding JJJ-block sum is positive semidefinite. Let H(M)=0H(M)=0H(M)=0 for non-Hermitian MMM, and otherwise obtain H(M)H(M)H(M) by taking reciprocal eigenvalues with 0−1=00^{-1}=00−1=0. Set F(h)=0F(h)=0F(h)=0 as a triple when hhh is not physical; otherwise set

Zh=ChH(Bh−iJsecond)ChT,F(h)=(Ah−Re⁡Zh, Ah−Re⁡Zh, −Im⁡Zh).Z_h=C_hH(B_h-iJ_{\text{second}})C_h^{\mathsf T}, \qquad F(h)=(A_h-\operatorname{Re}Z_h,\ A_h-\operatorname{Re}Z_h,\ -\operatorname{Im}Z_h).Zh​=Ch​H(Bh​−iJsecond​)ChT​,F(h)=(Ah​−ReZh​, Ah​−ReZh​, −ImZh​).

Put T0=F(g)T_0=F(g)T0​=F(g), Tk+1=F(Tk)T_{k+1}=F(T_k)Tk+1​=F(Tk​), and write Tk=(Ak,Bk,Ck)T_k=(A_k,B_k,C_k)Tk​=(Ak​,Bk​,Ck​), whose blocks are all on QnQ_nQn​. The theorem asserts both implications

(∃k∈N: Ak−iJn⪰̸0)⟹R is not separable,(∃k∈N: Ak−∥Ck∥opI−iJn⪰0)⟹R is separable.\left(\exists k\in\mathbb N:\ A_k-iJ_n\not\succeq0\right) \Longrightarrow R\text{ is not separable}, \qquad \left(\exists k\in\mathbb N:\ A_k-\|C_k\|_{\mathrm{op}}I-iJ_n\succeq0\right) \Longrightarrow R\text{ is separable}.(∃k∈N: Ak​−iJn​⪰0)⟹R is not separable,(∃k∈N: Ak​−∥Ck​∥op​I−iJn​⪰0)⟹R is separable.

Here ∥Ck∥op\|C_k\|_{\mathrm{op}}∥Ck​∥op​ is its Euclidean operator norm, and “separable” means that for every ε>0\varepsilon>0ε>0, some finite convex combination YYY of product kernels K⊗LK\otimes LK⊗L, with K,LK,LK,L state kernels on Sn,SmS_n,S_mSn​,Sm​, satisfies τ(R−Y)<ε\tau(R-Y)<\varepsilonτ(R−Y)<ε; τ(T)\tau(T)τ(T) is the infimum in [0,∞][0,\infty][0,∞] of ∑r∥ar∥∥br∥\sum_r\|a_r\|\|b_r\|∑r​∥ar​∥∥br​∥ over the summable expansions of TTT described above. The existential iteration index includes k=0k=0k=0, which refers to the first updated triple F(g)F(g)F(g).

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