Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Definitions for Peres-Horodecki separability criterion for continuous variable systems

Definition
QAlgorithms_Papers_Independent_Intakenext100_P92

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

prove2quantum

The objects the formalized statements of Peres-Horodecki separability criterion for continuous variable systems (arXiv:quant-ph/9909044) are stated in. Each was checked against the paper by a blind read-back and a moderator.

  • LocalMatrix: PDF p.3, (15): one-mode canonical matrices; both quadratures are retained.
  • quadratureVariance: PDF pp.2--3, (6),(8): variance of the linear observable x^T xi.
  • TwoModeMoments: PDF p.2, (6): actual finite moments of this normalized oscillator density. Finite energy supplies the quadratic-form domain. Directional Weyl derivatives fix all first and second moments; explicit symmetry fixes the antisymmetric part that directional derivatives alone cannot detect. No uncertainty or PPT condition is assumed. The characteristic uses exp(i x^T xi).
  • bobTranspose: PDF pp.2,4: Bob transpose in the real Hermite/Fock basis, not the adjoint.
  • reflectionMatrix: PDF p.2, (5): Lambda=diag(1,1,1,-1).
  • reflectedCovariance: PDF p.3, (9): reflected covariance; physicality of this matrix is a test.
  • ReflectedUncertainty: PDF p.3, (10): V+i Lambda Omega Lambda/2 is positive semidefinite.
  • simonInvariant: PDF pp.3--4: I1,I2,I3,I4 in Simon's order. The fourth is the trace contraction, whereas the accepted discord invariant with index 3 is det V.
  • SimonCriterion: PDF p.4, (19): the invariant second-moment inequality, independent of separability and of any operator positivity assertion about partial transpose.
  • localArea: PDF p.3, (11): the two independent local symplectic products.
  • SeparableVarianceBound: PDF p.3, (11): uncertainty sum, quantified over every real pair.
  • LocalSymplectic: PDF pp.3--4, (15): active local canonical transformation.
  • localCovarianceAction: PDF pp.3--4, (14)--(15): active, rather than passive, covariance congruence.
  • localReflection: PDF p.4: local sigma3=diag(1,-1), in one-mode coordinate labels.
  • exampleX: PDF p.3 example: X=q1+p1+q2+p2.
  • exampleY: PDF p.3 example: Y=q1-p1-q2+p2.
  • lowerAction: PDF pp.1--3: annihilation on the algebraic number-state core. This formula is used on finite-support vectors; no bounded annihilation operator is claimed.
  • raiseAction: PDF pp.1--3: creation on the same core; the vacuum boundary is explicit.
  • quadratureAction: PDF pp.1--3: q=(a+a^dagger)/sqrt(2), p=(a-a^dagger)/(i sqrt(2)).
  • linearAction: PDF p.2, (8): actual linear canonical observable on the finite-particle core.
  • coherentProductKernel: PDF pp.4--5: coherent product projector, using normalized full-space Fock coefficients. No choice of an unrealized vector or photon cutoff occurs.
  • IsClassicalTwo: PDF pp.4--5: positive generalized P representation as a probability measure. All coefficient integrals are explicitly integrable. Dirac and other singular measures are allowed; this is not a smooth-P-density restriction.
  • localUnitaryCoefficients: PDF pp.4--5: actual tensor-product coefficients of local oscillator unitaries. Fin 1 -> Nat is explicitly relabeled by its sole occupation coordinate.
  • localStateAction: PDF pp.4--5: actual (U1 tensor U2) rho (U1 tensor U2)^dagger.
Definition code
import Mathlib
import Definitions.Def_QAlgorithms_Papers_Independent_Advance50_P48
import Definitions.Def_QAlgorithms_Papers_Independent_Gap48_P47

/-!
R. Simon, arXiv:quant-ph/9909044v1, PDF pp.1--5.
Two complete oscillator modes, Alice | Bob; hbar=1; interleaved q1,p1,q2,p2;
half-anticommutator covariance, vacuum I/2. Accepted paper modules are used
only as pinned definition providers under the independent-paper override.
No theorem of another paper is used as a substitute for a source result.
The all-input L2 Wigner interface and two source ambiguities are deferred.
Compilation and independent moderation are pending the batch handoff.
-/

noncomputable section

namespace QAlgorithms.Papers.Independent.Intakenext100.P92

open MeasureTheory
open scoped BigOperators ENNReal ComplexOrder

open QAlgorithms.Papers.Independent.Dependency50.P33
  (IsDensity IsSeparable IsBarycenter)
open QAlgorithms.Papers.Independent.Dependency50.P39 (coherentCoefficient)
open QAlgorithms.Papers.Independent.Advance50.P48
  (TwoKernel GaussianTwo IsDiscordCovariance characteristicTwo
   covarianceModeBlock totalPhotonEnergy)
open QAlgorithms.Papers.Independent.Frontier30.P04 (modePhase Covariance twoModeStandard)
open QAlgorithms.Papers.Independent.Claimed50.P30 (canonicalOmega)
open QAlgorithms.Papers.Independent.Gap48.P47 (localSymplecticMatrix)
open QAlgorithms.Papers.Independent.Renew46.P39 (AffineMultimodeGaussianUnitary)
open QAlgorithms.Papers.Independent.Frontier30.P14 (conjugateKernel unitaryCoeff)

attribute [local instance] Classical.propDecidable Classical.decEq

/-- PDF p.3, (15): one-mode canonical matrices; both quadratures are retained. -/
abbrev LocalMatrix := Matrix (modePhase 1) (modePhase 1) ℝ

/-- PDF pp.2--3, (6),(8): variance of the linear observable x^T xi. -/
def quadratureVariance (V : Covariance 2) (x : (modePhase 2) → ℝ) : ℝ :=
  ∑ i, ∑ j, x i * V i j * x j

/-- PDF p.2, (6): actual finite moments of this normalized oscillator density.
Finite energy supplies the quadratic-form domain. Directional Weyl derivatives
fix all first and second moments; explicit symmetry fixes the antisymmetric
part that directional derivatives alone cannot detect. No uncertainty or PPT
condition is assumed. The characteristic uses exp(i x^T xi). -/
structure TwoModeMoments (rho : TwoKernel) (d : (modePhase 2) → ℝ) (V : Covariance 2) : Prop where
  density : IsDensity rho
  symmetric : V.IsSymm
  finiteEnergy : totalPhotonEnergy rho < ⊤
  firstDerivative : ∀ x : (modePhase 2) → ℝ,
    HasDerivAt (fun t : ℝ => characteristicTwo rho (t • x))
      (Complex.I * ((∑ i, d i * x i : ℝ) : ℂ)) 0
  secondDerivative : ∀ x : (modePhase 2) → ℝ,
    HasDerivAt (deriv (fun t : ℝ => characteristicTwo rho (t • x)))
      (((-quadratureVariance V x - (∑ i, d i * x i) ^ 2 : ℝ) : ℂ)) 0

/-- PDF pp.2,4: Bob transpose in the real Hermite/Fock basis, not the adjoint. -/
def bobTranspose (rho : TwoKernel) : TwoKernel :=
  fun m n => rho (m.1, n.2) (n.1, m.2)

/-- PDF p.2, (5): Lambda=diag(1,1,1,-1). -/
def reflectionMatrix : Covariance 2 :=
  Matrix.diagonal (fun i => if i.1 = 1 ∧ i.2 = 1 then -1 else 1)

/-- PDF p.3, (9): reflected covariance; physicality of this matrix is a test. -/
def reflectedCovariance (V : Covariance 2) : Covariance 2 :=
  reflectionMatrix * V * reflectionMatrix

/-- PDF p.3, (10): V+i Lambda Omega Lambda/2 is positive semidefinite. -/
def ReflectedUncertainty (V : Covariance 2) : Prop :=
  Matrix.PosSemidef (fun i j => (V i j : ℂ) + Complex.I *
    (((reflectionMatrix * canonicalOmega 2 * reflectionMatrix) i j / 2 : ℝ) : ℂ))

/-- PDF pp.3--4: I1,I2,I3,I4 in Simon's order. The fourth is the trace
contraction, whereas the accepted discord invariant with index 3 is det V. -/
def simonInvariant (V : Covariance 2) (k : Fin 4) : ℝ :=
  let A := covarianceModeBlock V 0 0
  let B := covarianceModeBlock V 1 1
  let C := covarianceModeBlock V 0 1
  let J := canonicalOmega 1
  if k = 0 then A.det else if k = 1 then B.det else if k = 2 then C.det
  else Matrix.trace (A * J * C * J * B * J * C.transpose * J)

/-- PDF p.4, (19): the invariant second-moment inequality, independent of
separability and of any operator positivity assertion about partial transpose. -/
def SimonCriterion (V : Covariance 2) : Prop :=
  (simonInvariant V 0 + simonInvariant V 1) / 4 ≤
    simonInvariant V 0 * simonInvariant V 1 +
      (1 / 4 - |simonInvariant V 2|) ^ 2 - simonInvariant V 3

/-- PDF p.3, (11): the two independent local symplectic products. -/
def localArea (x y : (modePhase 2) → ℝ) (mode : Fin 2) : ℝ :=
  x (mode,0) * y (mode,1) - x (mode,1) * y (mode,0)

/-- PDF p.3, (11): uncertainty sum, quantified over every real pair. -/
def SeparableVarianceBound (V : Covariance 2) : Prop :=
  ∀ x y : (modePhase 2) → ℝ,
    |localArea x y 0| + |localArea x y 1| ≤ quadratureVariance V x + quadratureVariance V y

/-- PDF pp.3--4, (15): active local canonical transformation. -/
def LocalSymplectic (S : LocalMatrix) : Prop :=
  S * canonicalOmega 1 * S.transpose = canonicalOmega 1

/-- PDF pp.3--4, (14)--(15): active, rather than passive, covariance congruence. -/
def localCovarianceAction (V : Covariance 2) (S1 S2 : LocalMatrix) : Covariance 2 :=
  let S := localSymplecticMatrix S1 S2
  S * V * S.transpose

/-- PDF p.4: local sigma3=diag(1,-1), in one-mode coordinate labels. -/
def localReflection : LocalMatrix :=
  Matrix.diagonal (fun i => if i.2 = 0 then 1 else -1)

/-- PDF p.3 example: X=q1+p1+q2+p2. -/
def exampleX : (modePhase 2) → ℝ := fun _ => 1

/-- PDF p.3 example: Y=q1-p1-q2+p2. -/
def exampleY : (modePhase 2) → ℝ := fun i => if i.1 = i.2 then 1 else -1

/-- PDF pp.1--3: annihilation on the algebraic number-state core. This formula
is used on finite-support vectors; no bounded annihilation operator is claimed. -/
def lowerAction (mode : Fin 2) (psi : (ℕ × ℕ) → ℂ) (m : ℕ × ℕ) : ℂ :=
  if mode = 0 then (Real.sqrt (m.1 + 1 : ℝ) : ℂ) * psi (m.1 + 1,m.2)
  else (Real.sqrt (m.2 + 1 : ℝ) : ℂ) * psi (m.1,m.2 + 1)

/-- PDF pp.1--3: creation on the same core; the vacuum boundary is explicit. -/
def raiseAction (mode : Fin 2) (psi : (ℕ × ℕ) → ℂ) (m : ℕ × ℕ) : ℂ :=
  if mode = 0 then
    if m.1 = 0 then 0 else (Real.sqrt (m.1 : ℝ) : ℂ) * psi (m.1 - 1,m.2)
  else if m.2 = 0 then 0 else (Real.sqrt (m.2 : ℝ) : ℂ) * psi (m.1,m.2 - 1)

/-- PDF pp.1--3: q=(a+a^dagger)/sqrt(2), p=(a-a^dagger)/(i sqrt(2)). -/
def quadratureAction (i : (modePhase 2)) (psi : (ℕ × ℕ) → ℂ) : (ℕ × ℕ) → ℂ :=
  fun m => if i.2 = 0 then
    (lowerAction i.1 psi m + raiseAction i.1 psi m) / (Real.sqrt 2 : ℂ)
  else (lowerAction i.1 psi m - raiseAction i.1 psi m) /
    (Complex.I * (Real.sqrt 2 : ℂ))

/-- PDF p.2, (8): actual linear canonical observable on the finite-particle core. -/
def linearAction (x : (modePhase 2) → ℝ) (psi : (ℕ × ℕ) → ℂ) : (ℕ × ℕ) → ℂ :=
  fun m => ∑ i, (x i : ℂ) * quadratureAction i psi m

/-- PDF pp.4--5: coherent product projector, using normalized full-space
Fock coefficients. No choice of an unrealized vector or photon cutoff occurs. -/
def coherentProductKernel (z : ℂ × ℂ) : TwoKernel := fun m n =>
  coherentCoefficient z.1 m.1 * star (coherentCoefficient z.1 n.1) *
    (coherentCoefficient z.2 m.2 * star (coherentCoefficient z.2 n.2))

/-- PDF pp.4--5: positive generalized P representation as a probability measure.
All coefficient integrals are explicitly integrable. Dirac and other singular
measures are allowed; this is not a smooth-P-density restriction. -/
def IsClassicalTwo (rho : TwoKernel) : Prop :=
  IsDensity rho ∧ ∃ mu : Measure (ℂ × ℂ),
    IsProbabilityMeasure mu ∧ IsBarycenter mu coherentProductKernel rho

/-- PDF pp.4--5: actual tensor-product coefficients of local oscillator unitaries.
Fin 1 -> Nat is explicitly relabeled by its sole occupation coordinate. -/
def localUnitaryCoefficients (U1 U2 : AffineMultimodeGaussianUnitary 1) : TwoKernel :=
  fun m n => unitaryCoeff U1.unitary (fun _ => m.1) (fun _ => n.1) *
    unitaryCoeff U2.unitary (fun _ => m.2) (fun _ => n.2)

/-- PDF pp.4--5: actual (U1 tensor U2) rho (U1 tensor U2)^dagger. -/
def localStateAction (U1 U2 : AffineMultimodeGaussianUnitary 1)
    (rho : TwoKernel) : TwoKernel :=
  conjugateKernel (localUnitaryCoefficients U1 U2) rho

/-! Sanity checks of local definitions; these do not prove source results. -/

example (V : Covariance 2) : quadratureVariance V 0 = 0 := by simp [quadratureVariance]

example (rho : TwoKernel) (d : (modePhase 2) → ℝ) (V : Covariance 2)
    (h : TwoModeMoments rho d V) : V.IsSymm := h.symmetric

example (rho : TwoKernel) : bobTranspose (bobTranspose rho) = rho := rfl

example : reflectionMatrix (1,1) (1,1) = -1 := by simp [reflectionMatrix]

example (V : Covariance 2) : reflectedCovariance V =
    reflectionMatrix * V * reflectionMatrix := rfl

example (V : Covariance 2) : ReflectedUncertainty V ↔
    Matrix.PosSemidef (fun i j => (V i j : ℂ) + Complex.I *
      (((reflectionMatrix * canonicalOmega 2 * reflectionMatrix) i j / 2 : ℝ) : ℂ)) := Iff.rfl

example (V : Covariance 2) : simonInvariant V 2 = (covarianceModeBlock V 0 1).det := by
  simp [simonInvariant]

example (V : Covariance 2) : SimonCriterion V ↔
    (simonInvariant V 0 + simonInvariant V 1) / 4 ≤
      simonInvariant V 0 * simonInvariant V 1 +
        (1 / 4 - |simonInvariant V 2|) ^ 2 - simonInvariant V 3 := Iff.rfl

example : localArea exampleX exampleY 0 = -2 := by norm_num [localArea, exampleX, exampleY]

example : localArea exampleX exampleY 1 = 2 := by norm_num [localArea, exampleX, exampleY]

example (V : Covariance 2) (h : SeparableVarianceBound V) :
    |localArea exampleX exampleY 0| + |localArea exampleX exampleY 1| ≤
      quadratureVariance V exampleX + quadratureVariance V exampleY := h _ _

example : LocalSymplectic (1 : LocalMatrix) := by simp [LocalSymplectic]

example (V : Covariance 2) (S1 S2 : LocalMatrix) : localCovarianceAction V S1 S2 =
    localSymplecticMatrix S1 S2 * V * (localSymplecticMatrix S1 S2).transpose := rfl

example : localReflection (0,1) (0,1) = -1 := by simp [localReflection]

example (psi : (ℕ × ℕ) → ℂ) : lowerAction 0 psi (0,0) = psi (1,0) := by
  simp [lowerAction]

example (psi : (ℕ × ℕ) → ℂ) : raiseAction 0 psi (0,0) = 0 := by simp [raiseAction]

example (psi : (ℕ × ℕ) → ℂ) : quadratureAction (0,0) psi (0,0) =
    (lowerAction 0 psi (0,0) + raiseAction 0 psi (0,0)) / (Real.sqrt 2 : ℂ) := rfl

example (psi : (ℕ × ℕ) → ℂ) : linearAction 0 psi = 0 := by
  funext m
  simp [linearAction]

example : coherentProductKernel (0,0) (0,0) (0,0) = 1 := by
  simp [coherentProductKernel, coherentCoefficient]

example (rho : TwoKernel) (h : IsClassicalTwo rho) : IsDensity rho := h.1

example (U1 U2 : AffineMultimodeGaussianUnitary 1) (m n : ℕ × ℕ) :
    localUnitaryCoefficients U1 U2 m n =
      unitaryCoeff U1.unitary (fun _ => m.1) (fun _ => n.1) *
        unitaryCoeff U2.unitary (fun _ => m.2) (fun _ => n.2) := rfl

example (U1 U2 : AffineMultimodeGaussianUnitary 1) (rho : TwoKernel) :
    localStateAction U1 U2 rho = conjugateKernel (localUnitaryCoefficients U1 U2) rho := rfl

/-! Statements, one per retained node. -/

end QAlgorithms.Papers.Independent.Intakenext100.P92
Source
Peres-Horodecki separability criterion for continuous variable systems, arXiv:quant-ph/9909044v1, https://arxiv.org/abs/quant-ph/9909044v1
Read-back

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

QAlgorithms.Papers.Independent.Intakenext100.P92.IsClassicalTwo

For any complex matrix ρ\rhoρ indexed by N2\mathbb N^2N2, this predicate requires all of the following: there exist sequences uj,vj∈ℓ2(N2;C)u_j,v_j\in\ell^2(\mathbb N^2;\mathbb C)uj​,vj​∈ℓ2(N2;C) such that ∑j∥uj∥∥vj∥\sum_j\|u_j\|\|v_j\|∑j​∥uj​∥∥vj​∥ converges and ρab=∑juj(a)vj(b)‾\rho_{ab}=\sum_j u_j(a)\overline{v_j(b)}ρab​=∑j​uj​(a)vj​(b)​ for every a,ba,ba,b; for every finite F⊆N2F\subseteq\mathbb N^2F⊆N2 and every function w:N2→Cw:\mathbb N^2\to\mathbb Cw:N2→C, the number ∑a,b∈Fw(a)‾ρabw(b)\sum_{a,b\in F}\overline{w(a)}\rho_{ab}w(b)∑a,b∈F​w(a)​ρab​w(b) has zero imaginary part and nonnegative real part; the real diagonal entries are summable with ∑aRe⁡ρaa=1\sum_a\operatorname{Re}\rho_{aa}=1∑a​Reρaa​=1; and there exists a probability measure μ\muμ on C2\mathbb C^2C2 such that, for every m,n∈N2m,n\in\mathbb N^2m,n∈N2, the function z↦Kz(m,n)z\mapsto K_z(m,n)z↦Kz​(m,n) is integrable and

ρmn=∫C2Kz(m,n) dμ(z),Kz(m,n)=∏a=12e−∣za∣2zamazana‾ma! na!.\rho_{mn}=\int_{\mathbb C^2}K_z(m,n)\,d\mu(z), \qquad K_z(m,n)= \prod_{a=1}^{2} \frac{e^{-|z_a|^2}z_a^{m_a}\overline{z_a^{n_a}}} {\sqrt{m_a!\,n_a!}}.ρmn​=∫C2​Kz​(m,n)dμ(z),Kz​(m,n)=a=1∏2​ma​!na​!​e−∣za​∣2zama​​zana​​​​.

No discrete-support requirement is imposed on μ\muμ; the diagonal normalization excludes the zero matrix.

QAlgorithms.Papers.Independent.Intakenext100.P92.LocalMatrix

A local matrix is a real matrix whose rows and columns are indexed by {0}×{0,1}\{0\}\times\{0,1\}{0}×{0,1}; thus it is a 2×22\times22×2 real matrix, with its coordinates labelled (0,0)(0,0)(0,0) and (0,1)(0,1)(0,1). No symmetry, positivity, or other condition is imposed.

QAlgorithms.Papers.Independent.Intakenext100.P92.LocalSymplectic

For any real 2×22\times22×2 matrix SSS with coordinates labelled (0,0),(0,1)(0,0),(0,1)(0,0),(0,1), being locally symplectic means precisely

SJST=J,J=(01−10).SJS^{\mathsf T}=J, \qquad J=\begin{pmatrix}0&1\\-1&0\end{pmatrix}.SJST=J,J=(0−1​10​).

QAlgorithms.Papers.Independent.Intakenext100.P92.ReflectedUncertainty

For any real matrix VVV indexed by Q={0,1}×{0,1}Q=\{0,1\}\times\{0,1\}Q={0,1}×{0,1}, reflected uncertainty means that the complex matrix

V+i2RΩRV+\frac{\mathrm i}{2}R\Omega RV+2i​RΩR

is positive semidefinite, where R=diag⁡(1,1,1,−1)R=\operatorname{diag}(1,1,1,-1)R=diag(1,1,1,−1) and Ω=diag⁡(J,J)\Omega=\operatorname{diag}(J,J)Ω=diag(J,J), with J=(01−10)J=\begin{pmatrix}0&1\\-1&0\end{pmatrix}J=(0−1​10​), in the coordinate order (0,0),(0,1),(1,0),(1,1)(0,0),(0,1),(1,0),(1,1)(0,0),(0,1),(1,0),(1,1). Thus the required matrix is V+i2diag⁡(J,−J)V+\frac{\mathrm i}{2}\operatorname{diag}(J,-J)V+2i​diag(J,−J). Positive semidefiniteness requires this matrix to be Hermitian and every complex quadratic form z∗(V+i2RΩR)zz^*(V+\frac{\mathrm i}{2}R\Omega R)zz∗(V+2i​RΩR)z to be nonnegative.

QAlgorithms.Papers.Independent.Intakenext100.P92.SeparableVarianceBound

For any real matrix VVV indexed by Q={0,1}×{0,1}Q=\{0,1\}\times\{0,1\}Q={0,1}×{0,1}, the separable variance bound means that every pair of real vectors x,yx,yx,y indexed by QQQ, including zero vectors, satisfies

∣x(0,0)y(0,1)−x(0,1)y(0,0)∣+∣x(1,0)y(1,1)−x(1,1)y(1,0)∣≤∑i,j∈QxiVijxj+∑i,j∈QyiVijyj.\left|x_{(0,0)}y_{(0,1)}-x_{(0,1)}y_{(0,0)}\right| + \left|x_{(1,0)}y_{(1,1)}-x_{(1,1)}y_{(1,0)}\right| \le \sum_{i,j\in Q}x_iV_{ij}x_j+ \sum_{i,j\in Q}y_iV_{ij}y_j.​x(0,0)​y(0,1)​−x(0,1)​y(0,0)​​+​x(1,0)​y(1,1)​−x(1,1)​y(1,0)​​≤i,j∈Q∑​xi​Vij​xj​+i,j∈Q∑​yi​Vij​yj​.

There is no separate symmetry, positivity, or separability hypothesis in this definition.

QAlgorithms.Papers.Independent.Intakenext100.P92.SimonCriterion

For any real matrix VVV indexed by {0,1}×{0,1}\{0,1\}\times\{0,1\}{0,1}×{0,1}, let Aab=V(0,a),(0,b)A_{ab}=V_{(0,a),(0,b)}Aab​=V(0,a),(0,b)​, Bab=V(1,a),(1,b)B_{ab}=V_{(1,a),(1,b)}Bab​=V(1,a),(1,b)​, and Cab=V(0,a),(1,b)C_{ab}=V_{(0,a),(1,b)}Cab​=V(0,a),(1,b)​ for a,b∈{0,1}a,b\in\{0,1\}a,b∈{0,1}, and let J=(01−10)J=\begin{pmatrix}0&1\\-1&0\end{pmatrix}J=(0−1​10​). The Simon criterion is precisely the inequality

det⁡A+det⁡B4≤det⁡A det⁡B+(14−∣det⁡C∣)2−tr⁡(AJCJBJCTJ).\frac{\det A+\det B}{4} \le \det A\,\det B+ \left(\frac14-|\det C|\right)^2 -\operatorname{tr}(AJCJBJC^{\mathsf T}J).4detA+detB​≤detAdetB+(41​−∣detC∣)2−tr(AJCJBJCTJ).

The definition imposes no additional symmetry, positivity, or density condition.

QAlgorithms.Papers.Independent.Intakenext100.P92.TwoModeMoments

For a complex kernel ρ=(ρab)a,b∈N2\rho=(\rho_{ab})_{a,b\in\mathbb N^2}ρ=(ρab​)a,b∈N2​, a real vector ddd indexed by Q={0,1}×{0,1}Q=\{0,1\}\times\{0,1\}Q={0,1}×{0,1}, and a real matrix VVV indexed by QQQ, this proposition requires all the following conditions: there are sequences ur,vr∈ℓ2(N2;C)u_r,v_r\in\ell^2(\mathbb N^2;\mathbb C)ur​,vr​∈ℓ2(N2;C) with ∑r∥ur∥∥vr∥<∞\sum_r\|u_r\|\|v_r\|<\infty∑r​∥ur​∥∥vr​∥<∞ and ρab=∑rur(a)vr(b)‾\rho_{ab}=\sum_r u_r(a)\overline{v_r(b)}ρab​=∑r​ur​(a)vr​(b)​ for every a,ba,ba,b; for every finite F⊆N2F\subseteq\mathbb N^2F⊆N2 and every z:N2→Cz:\mathbb N^2\to\mathbb Cz:N2→C, the number ∑a,b∈Fza‾ρabzb\sum_{a,b\in F}\overline{z_a}\rho_{ab}z_b∑a,b∈F​za​​ρab​zb​ has imaginary part zero and nonnegative real part; the family (Re⁡ρaa)a(\operatorname{Re}\rho_{aa})_a(Reρaa​)a​ is summable with sum 111; Vij=VjiV_{ij}=V_{ji}Vij​=Vji​ for all i,ji,ji,j; and

∑a∈N2max⁡ ⁣((a1+a2)Re⁡ρaa,0)<∞.\sum_{a\in\mathbb N^2}\max\!\left((a_1+a_2)\operatorname{Re}\rho_{aa},0\right)<\infty.a∈N2∑​max((a1​+a2​)Reρaa​,0)<∞.

In addition, define qV(x)=∑i,j∈QxiVijxjq_V(x)=\sum_{i,j\in Q}x_iV_{ij}x_jqV​(x)=∑i,j∈Q​xi​Vij​xj​ and define the complex characteristic function by

χρ(x)=∑a∈N2∑b∈N2ρab D ⁣(−x(0,1)+ix(0,0)2)b1a1D ⁣(−x(1,1)+ix(1,0)2)b2a2,\chi_\rho(x)= \sum_{a\in\mathbb N^2}\sum_{b\in\mathbb N^2} \rho_{ab}\, D\!\left(\frac{-x_{(0,1)}+\mathrm i x_{(0,0)}}{\sqrt2}\right)_{b_1a_1} D\!\left(\frac{-x_{(1,1)}+\mathrm i x_{(1,0)}}{\sqrt2}\right)_{b_2a_2},χρ​(x)=a∈N2∑​b∈N2∑​ρab​D(2​−x(0,1)​+ix(0,0)​​)b1​a1​​D(2​−x(1,1)​+ix(1,0)​​)b2​a2​​,

where, for z∈Cz\in\mathbb Cz∈C and m,n∈Nm,n\in\mathbb Nm,n∈N,

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 every real vector xxx indexed by QQQ, including x=0x=0x=0, the real-variable function fx(t)=χρ(tx)f_x(t)=\chi_\rho(tx)fx​(t)=χρ​(tx) must be differentiable at 000, with

fx′(0)=i∑i∈Qdixi,f_x'(0)=\mathrm i\sum_{i\in Q}d_ix_i,fx′​(0)=ii∈Q∑​di​xi​,

and its derivative function must itself be differentiable at 000, with

(fx′)′(0)=−qV(x)−(∑i∈Qdixi)2,(f_x')'(0)=-q_V(x)-\left(\sum_{i\in Q}d_ix_i\right)^2,(fx′​)′(0)=−qV​(x)−​i∈Q∑​di​xi​​2,

the last real number being viewed as complex. The characteristic function uses the displayed nested infinite sums without separate summability hypotheses; a nonsummable sum is assigned zero. Likewise, the derivative function used in the second condition is assigned zero wherever the original function is not differentiable.

QAlgorithms.Papers.Independent.Intakenext100.P92.bobTranspose

For any complex kernel ρ\rhoρ indexed by N2×N2\mathbb N^2\times\mathbb N^2N2×N2, its defined transpose on Bob’s indices is the kernel

(ρB)(m1,m2),(n1,n2)=ρ(m1,n2),(n1,m2)(m1,m2,n1,n2∈N).(\rho^{\mathrm B})_{(m_1,m_2),(n_1,n_2)} =\rho_{(m_1,n_2),(n_1,m_2)} \qquad(m_1,m_2,n_1,n_2\in\mathbb N).(ρB)(m1​,m2​),(n1​,n2​)​=ρ(m1​,n2​),(n1​,m2​)​(m1​,m2​,n1​,n2​∈N).

This exchanges the second components of the row and column indices and does not conjugate entries. No density or positivity assumption is required.

QAlgorithms.Papers.Independent.Intakenext100.P92.coherentProductKernel

For every pair z=(z1,z2)∈C2z=(z_1,z_2)\in\mathbb C^2z=(z1​,z2​)∈C2, this defines a complex matrix indexed by m,n∈N2m,n\in\mathbb N^2m,n∈N2 with entries

Kz(m,n)=cz1(m1)cz1(n1)‾cz2(m2)cz2(n2)‾,cα(j)=e−∣α∣2/2αjj!.K_z(m,n) = c_{z_1}(m_1)\overline{c_{z_1}(n_1)} c_{z_2}(m_2)\overline{c_{z_2}(n_2)}, \qquad c_\alpha(j)= \frac{e^{-|\alpha|^2/2}\alpha^j}{\sqrt{j!}}.Kz​(m,n)=cz1​​(m1​)cz1​​(n1​)​cz2​​(m2​)cz2​​(n2​)​,cα​(j)=j!​e−∣α∣2/2αj​.

All nonnegative occupation indices, including zero, are included; in particular α0=1\alpha^0=1α0=1, also when α=0\alpha=0α=0.

QAlgorithms.Papers.Independent.Intakenext100.P92.exampleX

This defines the real-valued function on the four coordinates (a,r)∈{0,1}×{0,1}(a,r)\in\{0,1\}\times\{0,1\}(a,r)∈{0,1}×{0,1} that equals 111 at every coordinate.

QAlgorithms.Papers.Independent.Intakenext100.P92.exampleY

This defines the real-valued function on {0,1}×{0,1}\{0,1\}\times\{0,1\}{0,1}×{0,1} that equals 111 when its two coordinate indices agree and −1-1−1 otherwise:

Y(0,0)=Y(1,1)=1,Y(0,1)=Y(1,0)=−1.Y_{(0,0)}=Y_{(1,1)}=1,\qquad Y_{(0,1)}=Y_{(1,0)}=-1.Y(0,0)​=Y(1,1)​=1,Y(0,1)​=Y(1,0)​=−1.

QAlgorithms.Papers.Independent.Intakenext100.P92.linearAction

For any real coefficient function x:{0,1}×{0,1}→Rx:\{0,1\}\times\{0,1\}\to\mathbb Rx:{0,1}×{0,1}→R and any function ψ:N2→C\psi:\mathbb N^2\to\mathbb Cψ:N2→C, this defines the function

m⟼∑a=01[x(a,0)2(Laψ(m)+Raψ(m))+x(a,1)i2(Laψ(m)−Raψ(m))].m\longmapsto \sum_{a=0}^{1} \left[ \frac{x_{(a,0)}}{\sqrt2}\bigl(L_a\psi(m)+R_a\psi(m)\bigr) + \frac{x_{(a,1)}}{\mathrm i\sqrt2}\bigl(L_a\psi(m)-R_a\psi(m)\bigr) \right].m⟼a=0∑1​[2​x(a,0)​​(La​ψ(m)+Ra​ψ(m))+i2​x(a,1)​​(La​ψ(m)−Ra​ψ(m))].

Here L0ψ(m)=m1+1ψ(m1+1,m2)L_0\psi(m)=\sqrt{m_1+1}\psi(m_1+1,m_2)L0​ψ(m)=m1​+1​ψ(m1​+1,m2​), L1ψ(m)=m2+1ψ(m1,m2+1)L_1\psi(m)=\sqrt{m_2+1}\psi(m_1,m_2+1)L1​ψ(m)=m2​+1​ψ(m1​,m2​+1), and Raψ(m)R_a\psi(m)Ra​ψ(m) lowers the corresponding argument with factor ma+1\sqrt{m_{a+1}}ma+1​​, returning zero when that argument is zero. The sum has four coordinate contributions, real coefficients are embedded into C\mathbb CC, and no normalization or summability is required.

QAlgorithms.Papers.Independent.Intakenext100.P92.localArea

For any real vectors x,yx,yx,y indexed by {0,1}×{0,1}\{0,1\}\times\{0,1\}{0,1}×{0,1} and any mode m∈{0,1}m\in\{0,1\}m∈{0,1}, their local area in mode mmm is defined by

am(x,y)=x(m,0)y(m,1)−x(m,1)y(m,0).a_m(x,y)=x_{(m,0)}y_{(m,1)}-x_{(m,1)}y_{(m,0)}.am​(x,y)=x(m,0)​y(m,1)​−x(m,1)​y(m,0)​.

The vectors are unrestricted, including zero vectors; this definition retains the sign of the displayed expression.

QAlgorithms.Papers.Independent.Intakenext100.P92.localCovarianceAction

For any real matrix VVV indexed by {0,1}×{0,1}\{0,1\}\times\{0,1\}{0,1}×{0,1}, and any two real matrices S1,S2S_1,S_2S1​,S2​ indexed by the single-mode coordinates {0}×{0,1}\{0\}\times\{0,1\}{0}×{0,1}, this defines the real matrix SVSTSVS^{\mathsf T}SVST, where SSS has S1S_1S1​ and S2S_2S2​ as its two diagonal mode blocks and zeros between modes:

S(a,r),(b,s)={(S1)(0,r),(0,s),a=b=0,(S2)(0,r),(0,s),a=b=1,0,a≠b.S_{(a,r),(b,s)} = \begin{cases} (S_1)_{(0,r),(0,s)},&a=b=0,\\ (S_2)_{(0,r),(0,s)},&a=b=1,\\ 0,&a\ne b. \end{cases}S(a,r),(b,s)​=⎩⎨⎧​(S1​)(0,r),(0,s)​,(S2​)(0,r),(0,s)​,0,​a=b=0,a=b=1,a=b.​

No symmetry, positivity, or symplectic condition is imposed on these arguments.

QAlgorithms.Papers.Independent.Intakenext100.P92.localReflection

This defines the real matrix on the single-mode coordinates {0}×{0,1}\{0\}\times\{0,1\}{0}×{0,1} whose diagonal entries are 111 at (0,0)(0,0)(0,0) and −1-1−1 at (0,1)(0,1)(0,1), and whose off-diagonal entries are zero; in that coordinate order it is diag⁡(1,−1)\operatorname{diag}(1,-1)diag(1,−1).

QAlgorithms.Papers.Independent.Intakenext100.P92.localStateAction

For any complex matrix ρ\rhoρ indexed by N2\mathbb N^2N2 and any two supplied single-mode affine Gaussian unitary structures, this defines the complex matrix

ρmn′=∑a∈N2Kma(∑b∈N2ρabKnb‾),Kma=(U1ea1)(m1)(U2ea2)(m2).\rho'_{mn} = \sum_{a\in\mathbb N^2} K_{ma}\left(\sum_{b\in\mathbb N^2}\rho_{ab}\overline{K_{nb}}\right), \qquad K_{ma}=(U_1e_{a_1})(m_1)(U_2e_{a_2})(m_2).ρmn′​=a∈N2∑​Kma​(b∈N2∑​ρab​Knb​​),Kma​=(U1​ea1​​)(m1​)(U2​ea2​​)(m2​).

The one-mode Hilbert space is ℓ2(N;C)\ell^2(\mathbb N;\mathbb C)ℓ2(N;C), using the identification of a one-coordinate occupation string with its value, and eje_jej​ is its coordinate vector. Each supplied structure consists of a complex-linear bijective isometry UℓU_\ellUℓ​, a real matrix SℓS_\ellSℓ​, and a real vector dℓd_\elldℓ​, satisfying SℓJSℓT=JS_\ell JS_\ell^{\mathsf T}=JSℓ​JSℓT​=J for J=(01−10)J=\begin{pmatrix}0&1\\-1&0\end{pmatrix}J=(0−1​10​), together with the coefficient identity

∑aHma(∑bD(v)abHnb‾)=eidℓ⋅vD(SℓTv)mn,Hma=(Uℓ−1ea)(m),\sum_a H_{ma}\left(\sum_b D(v)_{ab}\overline{H_{nb}}\right) = e^{\mathrm i d_\ell\cdot v}D(S_\ell^{\mathsf T}v)_{mn}, \qquad H_{ma}=(U_\ell^{-1}e_a)(m),a∑​Hma​(b∑​D(v)ab​Hnb​​)=eidℓ​⋅vD(SℓT​v)mn​,Hma​=(Uℓ−1​ea​)(m),

for every v∈R2v\in\mathbb R^2v∈R2, where

D(v)mn=e−∣z∣2/2m! n!∑j=0min⁡(m,n)zm−j(−z‾)n−jj! (m−j)! (n−j)!,z=−v1+iv02.D(v)_{mn} = e^{-|z|^2/2}\sqrt{m!\,n!} \sum_{j=0}^{\min(m,n)} \frac{z^{m-j}(-\overline z)^{n-j}}{j!\,(m-j)!\,(n-j)!}, \qquad z=\frac{-v_1+\mathrm i v_0}{\sqrt2}.D(v)mn​=e−∣z∣2/2m!n!​j=0∑min(m,n)​j!(m−j)!(n−j)!zm−j(−z)n−j​,z=2​−v1​+iv0​​.

There is no density, positivity, normalization, or summability hypothesis on ρ\rhoρ. The defining sums are the stated nested infinite sums, with a nonsummable complex sum assigned zero by the total infinite-sum convention.

QAlgorithms.Papers.Independent.Intakenext100.P92.localUnitaryCoefficients

For any two supplied single-mode affine Gaussian unitary structures, this defines a complex matrix on N2\mathbb N^2N2 by

Kmn=(U1en1)(m1) (U2en2)(m2).K_{mn} = (U_1e_{n_1})(m_1)\,(U_2e_{n_2})(m_2).Kmn​=(U1​en1​​)(m1​)(U2​en2​​)(m2​).

Here the one-mode Hilbert space is the space of square-summable complex functions on the singleton-indexed occupation strings, identified with ℓ2(N;C)\ell^2(\mathbb N;\mathbb C)ℓ2(N;C), and eje_jej​ is its coordinate vector. Each supplied structure consists of a real 2×22\times22×2 matrix SaS_aSa​, a real vector dad_ada​, and a complex-linear bijective isometry UaU_aUa​, with SaJSaT=JS_aJS_a^{\mathsf T}=JSa​JSaT​=J, where J=(01−10)J=\begin{pmatrix}0&1\\-1&0\end{pmatrix}J=(0−1​10​), and with the following coefficient identity for every v∈R2v\in\mathbb R^2v∈R2: conjugating the matrix D(v)D(v)D(v) by Ua−1U_a^{-1}Ua−1​ gives eida⋅vD(SaTv)e^{\mathrm i d_a\cdot v}D(S_a^{\mathsf T}v)eida​⋅vD(SaT​v). Explicitly, writing Hmb=(Ua−1eb)(m)H_{mb}=(U_a^{-1}e_b)(m)Hmb​=(Ua−1​eb​)(m), this identity means ∑bHmb(∑cD(v)bcHnc‾)=eida⋅vD(SaTv)mn\sum_b H_{mb}(\sum_c D(v)_{bc}\overline{H_{nc}})=e^{\mathrm i d_a\cdot v}D(S_a^{\mathsf T}v)_{mn}∑b​Hmb​(∑c​D(v)bc​Hnc​​)=eida​⋅vD(SaT​v)mn​, where D(v)mn=Dmn((−v1+iv0)/2)D(v)_{mn}=\mathcal D_{mn}((-v_1+\mathrm i v_0)/\sqrt2)D(v)mn​=Dmn​((−v1​+iv0​)/2​) and

Dmn(z)=e−∣z∣2/2m! n!∑j=0min⁡(m,n)zm−j(−z‾)n−jj! (m−j)! (n−j)!.\mathcal D_{mn}(z)= e^{-|z|^2/2}\sqrt{m!\,n!} \sum_{j=0}^{\min(m,n)} \frac{z^{m-j}(-\overline z)^{n-j}} {j!\,(m-j)!\,(n-j)!}.Dmn​(z)=e−∣z∣2/2m!n!​j=0∑min(m,n)​j!(m−j)!(n−j)!zm−j(−z)n−j​.

The defined matrix KKK uses only the two isometries’ coefficients.

QAlgorithms.Papers.Independent.Intakenext100.P92.lowerAction

For either mode a∈{0,1}a\in\{0,1\}a∈{0,1}, any function ψ:N2→C\psi:\mathbb N^2\to\mathbb Cψ:N2→C, and any m=(m1,m2)∈N2m=(m_1,m_2)\in\mathbb N^2m=(m1​,m2​)∈N2, this defines

Laψ(m)={m1+1 ψ(m1+1,m2),a=0,m2+1 ψ(m1,m2+1),a=1.L_a\psi(m)= \begin{cases} \sqrt{m_1+1}\,\psi(m_1+1,m_2),&a=0,\\ \sqrt{m_2+1}\,\psi(m_1,m_2+1),&a=1. \end{cases}La​ψ(m)={m1​+1​ψ(m1​+1,m2​),m2​+1​ψ(m1​,m2​+1),​a=0,a=1.​

The real square roots are regarded as complex scalars. There is no normalization or summability requirement on ψ\psiψ, and zero occupation numbers are included.

QAlgorithms.Papers.Independent.Intakenext100.P92.quadratureAction

For a coordinate i=(a,r)∈{0,1}×{0,1}i=(a,r)\in\{0,1\}\times\{0,1\}i=(a,r)∈{0,1}×{0,1} and any function ψ:N2→C\psi:\mathbb N^2\to\mathbb Cψ:N2→C, this defines another function on N2\mathbb N^2N2 by

Qa,rψ(m)={(Laψ(m)+Raψ(m))/2,r=0,(Laψ(m)−Raψ(m))/(i2),r=1.Q_{a,r}\psi(m)= \begin{cases} \bigl(L_a\psi(m)+R_a\psi(m)\bigr)/\sqrt2,&r=0,\\ \bigl(L_a\psi(m)-R_a\psi(m)\bigr)/(\mathrm i\sqrt2),&r=1. \end{cases}Qa,r​ψ(m)={(La​ψ(m)+Ra​ψ(m))/2​,(La​ψ(m)−Ra​ψ(m))/(i2​),​r=0,r=1.​

Here L0ψ(m)=m1+1ψ(m1+1,m2)L_0\psi(m)=\sqrt{m_1+1}\psi(m_1+1,m_2)L0​ψ(m)=m1​+1​ψ(m1​+1,m2​) and L1ψ(m)=m2+1ψ(m1,m2+1)L_1\psi(m)=\sqrt{m_2+1}\psi(m_1,m_2+1)L1​ψ(m)=m2​+1​ψ(m1​,m2​+1); R0ψ(m)R_0\psi(m)R0​ψ(m) is zero when m1=0m_1=0m1​=0 and otherwise equals m1ψ(m1−1,m2)\sqrt{m_1}\psi(m_1-1,m_2)m1​​ψ(m1​−1,m2​), while R1ψ(m)R_1\psi(m)R1​ψ(m) is zero when m2=0m_2=0m2​=0 and otherwise equals m2ψ(m1,m2−1)\sqrt{m_2}\psi(m_1,m_2-1)m2​​ψ(m1​,m2​−1). No normalization or summability assumption is imposed on ψ\psiψ.

QAlgorithms.Papers.Independent.Intakenext100.P92.quadratureVariance

For any real matrix VVV indexed by Q={0,1}×{0,1}Q=\{0,1\}\times\{0,1\}Q={0,1}×{0,1} and any real vector x=(xi)i∈Qx=(x_i)_{i\in Q}x=(xi​)i∈Q​, the defined quadrature variance is

qV(x)=∑i∈Q∑j∈QxiVijxj.q_V(x)=\sum_{i\in Q}\sum_{j\in Q}x_iV_{ij}x_j.qV​(x)=i∈Q∑​j∈Q∑​xi​Vij​xj​.

Neither symmetry nor positivity of VVV is assumed, and xxx may be zero.

QAlgorithms.Papers.Independent.Intakenext100.P92.raiseAction

For either mode a∈{0,1}a\in\{0,1\}a∈{0,1}, any function ψ:N2→C\psi:\mathbb N^2\to\mathbb Cψ:N2→C, and any m=(m1,m2)∈N2m=(m_1,m_2)\in\mathbb N^2m=(m1​,m2​)∈N2, this defines

Raψ(m)={0,a=0, m1=0,m1 ψ(m1−1,m2),a=0, m1>0,0,a=1, m2=0,m2 ψ(m1,m2−1),a=1, m2>0.R_a\psi(m)= \begin{cases} 0,&a=0,\ m_1=0,\\ \sqrt{m_1}\,\psi(m_1-1,m_2),&a=0,\ m_1>0,\\ 0,&a=1,\ m_2=0,\\ \sqrt{m_2}\,\psi(m_1,m_2-1),&a=1,\ m_2>0. \end{cases}Ra​ψ(m)=⎩⎨⎧​0,m1​​ψ(m1​−1,m2​),0,m2​​ψ(m1​,m2​−1),​a=0, m1​=0,a=0, m1​>0,a=1, m2​=0,a=1, m2​>0.​

The real square roots are regarded as complex scalars. No normalization or summability is required, and the boundary branches explicitly return zero.

QAlgorithms.Papers.Independent.Intakenext100.P92.reflectedCovariance

For any real matrix VVV indexed by Q={0,1}×{0,1}Q=\{0,1\}\times\{0,1\}Q={0,1}×{0,1}, its reflected covariance is defined as RVRRVRRVR, where R=diag⁡(1,1,1,−1)R=\operatorname{diag}(1,1,1,-1)R=diag(1,1,1,−1) in the coordinate order (0,0),(0,1),(1,0),(1,1)(0,0),(0,1),(1,0),(1,1)(0,0),(0,1),(1,0),(1,1). Equivalently, its (i,j)(i,j)(i,j) entry is riVijrjr_iV_{ij}r_jri​Vij​rj​, where r(1,1)=−1r_{(1,1)}=-1r(1,1)​=−1 and every other ri=1r_i=1ri​=1. No symmetry or positivity assumption on VVV is imposed.

QAlgorithms.Papers.Independent.Intakenext100.P92.reflectionMatrix

The reflection matrix is the real diagonal matrix indexed by Q={0,1}×{0,1}Q=\{0,1\}\times\{0,1\}Q={0,1}×{0,1} whose diagonal entry at (1,1)(1,1)(1,1) is −1-1−1 and whose other diagonal entries are 111. In the coordinate order (0,0),(0,1),(1,0),(1,1)(0,0),(0,1),(1,0),(1,1)(0,0),(0,1),(1,0),(1,1), it is

R=diag⁡(1,1,1,−1).R=\operatorname{diag}(1,1,1,-1).R=diag(1,1,1,−1).

QAlgorithms.Papers.Independent.Intakenext100.P92.simonInvariant

For any real matrix VVV indexed by Q={0,1}×{0,1}Q=\{0,1\}\times\{0,1\}Q={0,1}×{0,1} and any k∈{0,1,2,3}k\in\{0,1,2,3\}k∈{0,1,2,3}, define the 2×22\times22×2 blocks Aab=V(0,a),(0,b)A_{ab}=V_{(0,a),(0,b)}Aab​=V(0,a),(0,b)​, Bab=V(1,a),(1,b)B_{ab}=V_{(1,a),(1,b)}Bab​=V(1,a),(1,b)​, and Cab=V(0,a),(1,b)C_{ab}=V_{(0,a),(1,b)}Cab​=V(0,a),(1,b)​ for a,b∈{0,1}a,b\in\{0,1\}a,b∈{0,1}, and let J=(01−10)J=\begin{pmatrix}0&1\\-1&0\end{pmatrix}J=(0−1​10​). The defined invariant is

Ik(V)={det⁡A,k=0,det⁡B,k=1,det⁡C,k=2,tr⁡(AJCJBJCTJ),k=3.I_k(V)= \begin{cases} \det A,&k=0,\\ \det B,&k=1,\\ \det C,&k=2,\\ \operatorname{tr}(AJCJBJC^{\mathsf T}J),&k=3. \end{cases}Ik​(V)=⎩⎨⎧​detA,detB,detC,tr(AJCJBJCTJ),​k=0,k=1,k=2,k=3.​

No symmetry or positivity assumption on VVV is imposed.

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