Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§5, Eqs. (37),(38): type (i)/(ii) separability criteria and three-parameter border families

Open
QAlgorithms.Papers.Independent.Intakenext100.P46.type_i_ii_borders

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

prove2quantum

Let γA,γB≥1/2\gamma_A,\gamma_B\ge1/2γA​,γB​≥1/2, D=diag⁡(γA,γB,γA,γB)D=\operatorname{diag}(\gamma_A,\gamma_B,\gamma_A,\gamma_B)D=diag(γA​,γB​,γA​,γB​) and x>0x>0x>0. Construct type-I covariances by TIDTITT_IDT_I^TTI​DTIT​, where TI=(R(−r)⊕R(r))X(x−1)T_I=(R(-r)\oplus R(r))X(x^{-1})TI​=(R(−r)⊕R(r))X(x−1), and type-II covariances by TIIDTIITT_{II}DT_{II}^TTII​DTIIT​, where TII=(Θ(−θ)⊕Θ(−θ))X(x−1)T_{II}=(\Theta(-\theta)\oplus\Theta(-\theta))X(x^{-1})TII​=(Θ(−θ)⊕Θ(−θ))X(x−1). Here X(x)=diag⁡(x,x−1/2,x−1/2,x)X(x)=\operatorname{diag}(\sqrt{x},x^{-1/2},x^{-1/2},\sqrt{x})X(x)=diag(x​,x−1/2,x−1/2,x​), R(r)=(cosh⁡rsinh⁡rsinh⁡rcosh⁡r)R(r)=\begin{pmatrix}\cosh r&\sinh r\\\sinh r&\cosh r\end{pmatrix}R(r)=(coshrsinhr​sinhrcoshr​), and Θ(t)=(cos⁡tsin⁡t−sin⁡tcos⁡t)\Theta(t)=\begin{pmatrix}\cos t&\sin t\\-\sin t&\cos t\end{pmatrix}Θ(t)=(cost−sint​sintcost​). For type II require x>max⁡{γA/γB,γB/γA}x>\max\{\sqrt{\gamma_A/\gamma_B},\sqrt{\gamma_B/\gamma_A}\}x>max{γA​/γB​​,γB​/γA​​} or 0<x<min⁡{γA/γB,γB/γA}0<x<\min\{\sqrt{\gamma_A/\gamma_B},\sqrt{\gamma_B/\gamma_A}\}0<x<min{γA​/γB​​,γB​/γA​​}. Both constructed covariances are physical. Write L=(2γA2−1/2)(2γB2−1/2)L=(2\gamma_A^2-1/2)(2\gamma_B^2-1/2)L=(2γA2​−1/2)(2γB2​−1/2). Separability is equivalent, respectively, to

L≥sinh⁡2(2r)[(x2+x−2)γAγB+γA2+γB2],L\ge\sinh^2(2r)[(x^2+x^{-2})\gamma_A\gamma_B+\gamma_A^2+\gamma_B^2],L≥sinh2(2r)[(x2+x−2)γA​γB​+γA2​+γB2​], L≥sin⁡2(2θ)[(x2+x−2)γAγB−γA2−γB2].L\ge\sin^2(2\theta)[(x^2+x^{-2})\gamma_A\gamma_B-\gamma_A^2-\gamma_B^2].L≥sin2(2θ)[(x2+x−2)γA​γB​−γA2​−γB2​].

When both eigenvalues exceed 1/21/21/2, equality is equivalent to the actual Gaussian covariance border, and the finite EM is T−Tdiag⁡(m,m)T−1T^{-T}\operatorname{diag}(m,m)T^{-1}T−Tdiag(m,m)T−1 with mj=ln⁡((2γj+1)/(2γj−1))m_j=\ln((2\gamma_j+1)/(2\gamma_j-1))mj​=ln((2γj​+1)/(2γj​−1)). On this border the relevant sine or hyperbolic sine is nonzero. Put KI=[L/sinh⁡2(2r)−γA2−γB2]/(γAγB)K_I=[L/\sinh^2(2r)-\gamma_A^2-\gamma_B^2]/(\gamma_A\gamma_B)KI​=[L/sinh2(2r)−γA2​−γB2​]/(γA​γB​) or KII=[L/sin⁡2(2θ)+γA2+γB2]/(γAγB)K_{II}=[L/\sin^2(2\theta)+\gamma_A^2+\gamma_B^2]/(\gamma_A\gamma_B)KII​=[L/sin2(2θ)+γA2​+γB2​]/(γA​γB​). Then K≥2K\ge2K≥2 and

x=K+K2−42orx=K−K2−42.x=\sqrt{\frac{K+\sqrt{K^2-4}}2}\quad\text{or}\quad x=\sqrt{\frac{K-\sqrt{K^2-4}}2}.x=2K+K2−4​​​orx=2K−K2−4​​​.

Thus the finite border search has three independent parameters. The pure-mode separability criteria remain included; scalar equality alone is not used to classify their border. The finite exponential matrix refers to the actual normalized Gibbs density: after a bounded unitary implementing the stated symplectic transformation, its number-basis weights are ∏j(1−e−mj)e−mjnj\prod_j(1-e^{-m_j})e^{-m_j n_j}∏j​(1−e−mj​)e−mj​nj​. This operator meaning is part of the conclusion, alongside the covariance normal form.

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 pp.8–9, (37),(38): separability criteria for constructed
families, finite-chart border equivalences, EM construction, and elimination of
x by the other three parameters. Pure covariances remain in the separability
clauses; they are not declared to be borders from a scalar equality alone. -/
theorem type_i_ii_borders (gA gB : ℝ) (hA : 1 / 2 ≤ gA) (hB : 1 / 2 ≤ gB) :
    (∀ r x : ℝ, 0 < x →
      IsPhysicalCovariance (typeICM gA gB r x) ∧
      (CovSeparable (typeICM gA gB r x) ↔ borderRightI gA gB r x ≤ borderLeft gA gB) ∧
      (1 / 2 < gA → 1 / 2 < gB →
        GaussianEM (typeICM gA gB r x)
          (emFrom (squeezeBoth (-r) * xSqueeze x⁻¹) (pairValues gA gB)) ∧
        (CovBorder (typeICM gA gB r x) ↔ borderRightI gA gB r x = borderLeft gA gB) ∧
        (CovBorder (typeICM gA gB r x) →
          Real.sinh (2 * r) ≠ 0 ∧ PositiveRoots
            ((borderLeft gA gB / Real.sinh (2 * r) ^ 2 - (gA ^ 2 + gB ^ 2)) / (gA * gB)) x))) ∧
    (∀ t x : ℝ, TypeIIDomain gA gB x →
      IsPhysicalCovariance (typeIICM gA gB t x) ∧
      (CovSeparable (typeIICM gA gB t x) ↔ borderRightII gA gB t x ≤ borderLeft gA gB) ∧
      (1 / 2 < gA → 1 / 2 < gB →
        GaussianEM (typeIICM gA gB t x)
          (emFrom (rotateBoth (-t) * xSqueeze x⁻¹) (pairValues gA gB)) ∧
        (CovBorder (typeIICM gA gB t x) ↔ borderRightII gA gB t x = borderLeft gA gB) ∧
        (CovBorder (typeIICM gA gB t x) →
          Real.sin (2 * t) ≠ 0 ∧ PositiveRoots
            ((borderLeft gA gB / Real.sin (2 * t) ^ 2 + (gA ^ 2 + gB ^ 2)) / (gA * gB)) x))) := by sorry

end QAlgorithms.Papers.Independent.Intakenext100.P46

end
Source
Gaussian relative entropy of entanglement, arXiv:quant-ph/0402109v3, PDF p. 8, §5, Eqs. (37),(38): type (i)/(ii) separability criteria and three-parameter border families
Read-back

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

For every pair of real numbers gA,gB≥12g_A,g_B\ge\frac12gA​,gB​≥21​, use the order A,BA,BA,B for the two modes and the order qA,qB,pA,pBq_A,q_B,p_A,p_BqA​,qB​,pA​,pB​ for their quadratures, and put

J=(0I2−I20),N(gA,gB)=diag⁡(gA,gB,gA,gB),H(r)=(cosh⁡rsinh⁡rsinh⁡rcosh⁡r),R(t)=(cos⁡tsin⁡t−sin⁡tcos⁡t),Dx=diag⁡(x−1/2,x1/2,x1/2,x−1/2)(x>0),SI(r,x)=diag⁡(H(−r),H(r))Dx,SII(t,x)=diag⁡(R(−t),R(−t))Dx,αI=SIN(gA,gB)SIT,αII=SIIN(gA,gB)SIIT,M(S)=(ST)−1N ⁣(log⁡2gA+12gA−1,log⁡2gB+12gB−1)S−1,L=(2gA2−12)(2gB2−12),BI(r,x)=sinh⁡2(2r)((x2+x−2)gAgB+gA2+gB2),BII(t,x)=sin⁡2(2t)((x2+x−2)gAgB−gA2−gB2).\begin{gathered} J=\begin{pmatrix}0&I_2\\-I_2&0\end{pmatrix},\qquad N(g_A,g_B)=\operatorname{diag}(g_A,g_B,g_A,g_B),\\ H(r)=\begin{pmatrix}\cosh r&\sinh r\\\sinh r&\cosh r\end{pmatrix}, \qquad R(t)=\begin{pmatrix}\cos t&\sin t\\-\sin t&\cos t\end{pmatrix},\\ D_x=\operatorname{diag}(x^{-1/2},x^{1/2},x^{1/2},x^{-1/2})\quad(x>0),\\ S_I(r,x)=\operatorname{diag}(H(-r),H(r))D_x,\qquad S_{II}(t,x)=\operatorname{diag}(R(-t),R(-t))D_x,\\ \alpha_I=S_I N(g_A,g_B)S_I^{\mathsf T},\qquad \alpha_{II}=S_{II}N(g_A,g_B)S_{II}^{\mathsf T},\\ M(S)=(S^{\mathsf T})^{-1} N\!\left(\log\frac{2g_A+1}{2g_A-1}, \log\frac{2g_B+1}{2g_B-1}\right)S^{-1},\\ L=(2g_A^2-\tfrac12)(2g_B^2-\tfrac12),\\ B_I(r,x)=\sinh^2(2r)\bigl((x^2+x^{-2})g_Ag_B+g_A^2+g_B^2\bigr),\\ B_{II}(t,x)=\sin^2(2t)\bigl((x^2+x^{-2})g_Ag_B-g_A^2-g_B^2\bigr). \end{gathered}J=(0−I2​​I2​0​),N(gA​,gB​)=diag(gA​,gB​,gA​,gB​),H(r)=(coshrsinhr​sinhrcoshr​),R(t)=(cost−sint​sintcost​),Dx​=diag(x−1/2,x1/2,x1/2,x−1/2)(x>0),SI​(r,x)=diag(H(−r),H(r))Dx​,SII​(t,x)=diag(R(−t),R(−t))Dx​,αI​=SI​N(gA​,gB​)SIT​,αII​=SII​N(gA​,gB​)SIIT​,M(S)=(ST)−1N(log2gA​−12gA​+1​,log2gB​−12gB​+1​)S−1,L=(2gA2​−21​)(2gB2​−21​),BI​(r,x)=sinh2(2r)((x2+x−2)gA​gB​+gA2​+gB2​),BII​(t,x)=sin2(2t)((x2+x−2)gA​gB​−gA2​−gB2​).​

The following terminology specifies the predicates in the assertion. A physical covariance α\alphaα means αT=α\alpha^{\mathsf T}=\alphaαT=α and α−i2J\alpha-\frac i2Jα−2i​J is positive semidefinite. A state on a countable label set Λ\LambdaΛ is a complex kernel ρ\rhoρ admitting sequences un,vn∈ℓ2(Λ)u_n,v_n\in\ell^2(\Lambda)un​,vn​∈ℓ2(Λ) with ∑n∥un∥∥vn∥<∞\sum_n\|u_n\|\|v_n\|<\infty∑n​∥un​∥∥vn​∥<∞ and ρab=∑nun(a)vn(b)‾\rho_{ab}=\sum_nu_n(a)\overline{v_n(b)}ρab​=∑n​un​(a)vn​(b)​, whose every finite quadratic form is real and nonnegative, and whose real diagonal is summable with sum 111. Its trace size, and that of any kernel, is the infimum of these nuclear-expansion costs in [0,∞][0,\infty][0,∞], with an empty infimum equal to ∞\infty∞. A bipartite state is separable when its kernel belongs to the trace-size closure of the real convex hull of product state kernels; this closure means that, for every ε>0\varepsilon>0ε>0, a kernel in that convex hull lies at trace size less than ε\varepsilonε from it. For a finite mode set FFF, let WF(ξ)uv=∏j∈FD(zj)ujvjW_F(\xi)_{uv}=\prod_{j\in F}D(z_j)_{u_jv_j}WF​(ξ)uv​=∏j∈F​D(zj​)uj​vj​​, where zj=−(ξqj+iξpj)/2z_j=-(\xi_{q_j}+i\xi_{p_j})/\sqrt2zj​=−(ξqj​​+iξpj​​)/2​ and

D(z)mn=e−∣z∣2/2m! n!∑h=0min⁡(m,n)zm−h(−z‾)n−hh! (m−h)! (n−h)!.D(z)_{mn} =e^{-|z|^2/2}\sqrt{m!\,n!} \sum_{h=0}^{\min(m,n)} \frac{z^{m-h}(-\overline z)^{n-h}}{h!\,(m-h)!\,(n-h)!}.D(z)mn​=e−∣z∣2/2m!n!​h=0∑min(m,n)​h!(m−h)!(n−h)!zm−h(−z)n−h​.

For bipartite occupation labels, use the product of the Weyl kernels on the two parties. Write H(ρ,d,α)\mathcal H(\rho,d,\alpha)H(ρ,d,α) when α\alphaα is physical and, for every real quadrature vector ξ\xiξ, each inner series and the outer series in χρ(ξ)=∑u∑vρuvW(ξ)vu\chi_\rho(\xi)=\sum_u\sum_v\rho_{uv}W(\xi)_{vu}χρ​(ξ)=∑u​∑v​ρuv​W(ξ)vu​ are summable and

χρ(ξ)=exp⁡ ⁣(−12ξTJTαJξ+iξTJd).\chi_\rho(\xi) =\exp\!\left(-\tfrac12\xi^{\mathsf T}J^{\mathsf T}\alpha J\xi +i\xi^{\mathsf T}Jd\right).χρ​(ξ)=exp(−21​ξTJTαJξ+iξTJd).

Write S(α)\mathcal S(\alpha)S(α) when there exists a separable bipartite state ρ\rhoρ satisfying H(ρ,0,α)\mathcal H(\rho,0,\alpha)H(ρ,0,α). Write B(α)\mathcal B(\alpha)B(α) when there exists such a centered Gaussian state ρ\rhoρ that is separable and, additionally, there exist d0,α0d_0,\alpha_0d0​,α0​ satisfying H(ρ,d0,α0)\mathcal H(\rho,d_0,\alpha_0)H(ρ,d0​,α0​) such that, for every ε>0\varepsilon>0ε>0, there are a nonseparable state τ\tauτ and a covariance β\betaβ satisfying H(τ,d0,β)\mathcal H(\tau,d_0,\beta)H(τ,d0​,β) and ∣βij−(α0)ij∣<ε|\beta_{ij}-(\alpha_0)_{ij}|<\varepsilon∣βij​−(α0​)ij​∣<ε for every entry. Finally, E(α,M)\mathcal E(\alpha,M)E(α,M) means both of the following: there exist an invertible real matrix TTT with TJTT=JTJT^{\mathsf T}=JTJTT=J and mode values hj>12h_j>\frac12hj​>21​ such that α=TN(h)TT\alpha=T N(h)T^{\mathsf T}α=TN(h)TT and M=(TT)−1N(log⁡((2h+1)/(2h−1)))T−1M=(T^{\mathsf T})^{-1}N(\log((2h+1)/(2h-1)))T^{-1}M=(TT)−1N(log((2h+1)/(2h−1)))T−1; and every centered Gaussian density kernel with covariance α\alphaα, now indexed by joint occupation functions, has a representation CU(Gm)C_U(G_m)CU​(Gm​), where mj>0m_j>0mj​>0, Gm(u,v)=1u=v∏j(1−e−mj)e−mjujG_m(u,v)=\mathbf1_{u=v}\prod_j(1-e^{-m_j})e^{-m_ju_j}Gm​(u,v)=1u=v​∏j​(1−e−mj​)e−mj​uj​, UUU is a unitary on the occupation ℓ2\ell^2ℓ2-space, and there is an invertible symplectic T0T_0T0​ such that M=(T0T)−1N(m)T0−1M=(T_0^{\mathsf T})^{-1}N(m)T_0^{-1}M=(T0T​)−1N(m)T0−1​ and

CU∗(WF(ξ))=WF(−JT0TJξ).C_{U^*}(W_F(\xi))=W_F(-JT_0^{\mathsf T}J\xi).CU∗​(WF​(ξ))=WF​(−JT0T​Jξ).

Here N(h)N(h)N(h) repeats the mode diagonal in the two quadrature blocks, and unitary conjugation is defined by the nested coefficient sums CU(K)ij=∑aUia(∑bKabUjb‾)C_U(K)_{ij}=\sum_aU_{ia}\bigl(\sum_bK_{ab}\overline{U_{jb}}\bigr)CU​(K)ij​=∑a​Uia​(∑b​Kab​Ujb​​). With these meanings, the theorem asserts

for every r∈R, x>0:αI is physical,S(αI)  ⟺  BI(r,x)≤L;if gA>12 and gB>12, thenE(αI,M(SI)),B(αI)  ⟺  BI(r,x)=L,B(αI) ⟹ sinh⁡(2r)≠0,KI≥2,x=KI+KI2−42 or x=KI−KI2−42,KI=L/sinh⁡2(2r)−(gA2+gB2)gAgB;for every t,x∈R with x>0 andx>max⁡ ⁣(gA/gB,gB/gA)orx<min⁡ ⁣(gA/gB,gB/gA):αII is physical,S(αII)  ⟺  BII(t,x)≤L;if gA>12 and gB>12, thenE(αII,M(SII)),B(αII)  ⟺  BII(t,x)=L,B(αII) ⟹ sin⁡(2t)≠0,KII≥2,x=KII+KII2−42 or x=KII−KII2−42,KII=L/sin⁡2(2t)+(gA2+gB2)gAgB.\begin{aligned} &\text{for every }r\in\mathbb R,\ x>0:\quad \alpha_I\text{ is physical},\qquad \mathcal S(\alpha_I)\iff B_I(r,x)\le L;\\ &\quad\text{if }g_A>\tfrac12\text{ and }g_B>\tfrac12,\text{ then}\\ &\qquad \mathcal E(\alpha_I,M(S_I)),\qquad \mathcal B(\alpha_I)\iff B_I(r,x)=L,\\ &\qquad \mathcal B(\alpha_I)\ \Longrightarrow\ \sinh(2r)\ne0,\quad K_I\ge2,\quad x=\sqrt{\frac{K_I+\sqrt{K_I^2-4}}2} \ \text{or}\ x=\sqrt{\frac{K_I-\sqrt{K_I^2-4}}2},\\ &\qquad K_I=\frac{L/\sinh^2(2r)-(g_A^2+g_B^2)}{g_Ag_B};\\[2mm] &\text{for every }t,x\in\mathbb R\text{ with }x>0\text{ and}\\ &\qquad x>\max\!\left(\sqrt{g_A/g_B},\sqrt{g_B/g_A}\right) \quad\text{or}\quad x<\min\!\left(\sqrt{g_A/g_B},\sqrt{g_B/g_A}\right):\\ &\quad \alpha_{II}\text{ is physical},\qquad \mathcal S(\alpha_{II})\iff B_{II}(t,x)\le L;\\ &\quad\text{if }g_A>\tfrac12\text{ and }g_B>\tfrac12,\text{ then}\\ &\qquad \mathcal E(\alpha_{II},M(S_{II})),\qquad \mathcal B(\alpha_{II})\iff B_{II}(t,x)=L,\\ &\qquad \mathcal B(\alpha_{II})\ \Longrightarrow\ \sin(2t)\ne0,\quad K_{II}\ge2,\quad x=\sqrt{\frac{K_{II}+\sqrt{K_{II}^2-4}}2} \ \text{or}\ x=\sqrt{\frac{K_{II}-\sqrt{K_{II}^2-4}}2},\\ &\qquad K_{II}=\frac{L/\sin^2(2t)+(g_A^2+g_B^2)}{g_Ag_B}. \end{aligned}​for every r∈R, x>0:αI​ is physical,S(αI​)⟺BI​(r,x)≤L;if gA​>21​ and gB​>21​, thenE(αI​,M(SI​)),B(αI​)⟺BI​(r,x)=L,B(αI​) ⟹ sinh(2r)=0,KI​≥2,x=2KI​+KI2​−4​​​ or x=2KI​−KI2​−4​​​,KI​=gA​gB​L/sinh2(2r)−(gA2​+gB2​)​;for every t,x∈R with x>0 andx>max(gA​/gB​​,gB​/gA​​)orx<min(gA​/gB​​,gB​/gA​​):αII​ is physical,S(αII​)⟺BII​(t,x)≤L;if gA​>21​ and gB​>21​, thenE(αII​,M(SII​)),B(αII​)⟺BII​(t,x)=L,B(αII​) ⟹ sin(2t)=0,KII​≥2,x=2KII​+KII2​−4​​​ or x=2KII​−KII2​−4​​​,KII​=gA​gB​L/sin2(2t)+(gA2​+gB2​)​.​

In the second quantified clause, x>0x>0x>0 is required together with the displayed disjunction. The initial assumptions allow either gAg_AgA​ or gBg_BgB​ to equal 12\frac1221​; the exponential-matrix and border conclusions are conditional on both inequalities being strict. The parameters r,tr,tr,t otherwise range over all real numbers, including values where their displayed hyperbolic sine or sine vanishes.

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