Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Definitions for Conditions for the local manipulation of Gaussian states

Definition
QAlgorithms_Papers_Independent_Gap48_P47

by Goku · Oct 22, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

The objects the formalized statements of Conditions for the local manipulation of Gaussian states (arXiv:quant-ph/0109126) are stated in. Each was checked against the paper by a blind read-back and a moderator.

  • TwoModeGaussianDensity: PDF pp.1--2 and note [12], p.4: actual normalized two-oscillator Gaussian state, specified finite covariance, arbitrary first moments and finite energy. The accepted characteristic uses W(v)=exp(i v^T O); the source's D(z) is W(Sigma^T z). The covariance normalization is unchanged by this reparametrization.
  • NormalVector: PDF p.2: the four real coordinates of the local symplectic normal form.
  • associatedVector: PDF p.2, below (1): compute the unique associated vector from the four determinants. For physical covariances the marginal determinants are positive and the discriminants are nonnegative. The larger correlation singular value is nonnegative, and the sign of the other is fixed by det B. At c=0 both correlations vanish, so the explicit zero branch avoids division by zero. No assertion about nonphysical matrices is used in any theorem.
  • logCriterionPolynomial: PDF p.2: logCriterionPolynomial, transcribed from its explicit definition, not intermediate (8).
  • f1: PDF p.2: f1 for the source and target covariances.
  • f2: PDF p.2: f2. In the stated criteria w2>0 enforces x*y != 0; ordinary division is therefore used only on its nonzero domain.
  • w1: PDF p.3, (9): upper admissible absolute correlation product.
  • w2: PDF p.3, (9): lower admissible absolute correlation product.
  • joinOccupation: PDF pp.1--2: a single retained oscillator followed by m ancillary oscillators. This is a basis relabeling, not a finite occupation cutoff.
  • firstOccupation: PDF pp.1--2: retained occupation coordinate in a local dilation.
  • restOccupation: PDF pp.1--2: all ancillary occupation coordinates, including m=0.
  • appendProduct: PDF pp.1--2 and Fig.1: independent input and ancillary density product, transported to the completed (1+m)-mode space.
  • localDilationOutput: PDF pp.1--2, Fig.1: actual bounded-unitary conjugation followed by the ancillary partial trace. The unitary and ancilla are fixed for every input.
  • IsLocalGaussianDilation: PDF pp.1--2, Fig.1 and p.4 note [10]: one local LOG, with arbitrary finite ancilla count, actual normalized Gaussian environment, and a Gaussian unitary including translations. Its action on EVERY trace-class input is fixed; complete positivity and trace preservation are explicit.
  • localProduct: PDF pp.1--2: tensor product of the two independent local maps, retaining off-diagonal blocks. The inner map acts on Bob and the outer map on Alice.
  • LOGReachable: PDF pp.1--2: actual state convertibility by LOG. The operations may be chosen for this known input, but each is a lawful channel on all inputs. The two ancillas are independent; no shared randomness is introduced.
  • LOGInSystemOne: Proposition 1, PDF pp.2--3: noisy LOG on Alice with reversible Gaussian unitary/translation on Bob, as the proof expressly permits. Independent normal-form changes are included by composition of the local operations.
  • LocalCP: PDF p.2, (3): the EXACT source plus-i local CP constraint for the passive matrix M in Gamma -> M^T Gamma M+G. No invertibility of M or G is required.
  • IsAffineLocalGaussian: PDF p.2, (2)--(3), and note [14], p.4: the affine Gaussian CPTP map, expressed on all actual density operators by its Weyl characteristic. For W(v)=exp(i v^T O), the source passive covariance convention pulls v back to M v. The noise multiplier has /4 because vacuum covariance is I. Summability excludes total-tsum defaults, including on non-Gaussian inputs.
  • localSymplecticMatrix: PDF p.2 and p.4 note [13]: block diagonal local passive matrix S1 direct-sum S2.
  • localCongruence: PDF p.2 and p.4 note [13]: local symplectic action in the source convention.
Definition code
import Mathlib
import Definitions.Def_QAlgorithms_Papers_Independent_Claimed50_P19
import Definitions.Def_QAlgorithms_Papers_Independent_Renew46_P39
import Definitions.Def_QAlgorithms_Papers_Independent_Dependency50_P10
import Definitions.Def_QAlgorithms_Papers_Independent_Frontier30_P14

/-!
Eisert--Plenio, arXiv:quant-ph/0109126v2, PDF pp.1--4.
Interleaved (X1,P1,X2,P2), hbar=1, vacuum covariance I. All density
operators act on completed oscillator spaces. New definitions are paper-local.
The source's S acts by S^T Gamma S; the accepted Gaussian unitary stores
an active matrix S^T. No measurement, communication, or correlated ancilla
is part of LOG. Six defective source claims are documented in HARD.md.
Compilation is pending the required batch handoff.
-/

noncomputable section

open scoped BigOperators ComplexOrder

namespace QAlgorithms.Papers.Independent.Gap48.P47

open QAlgorithms.Papers.Independent.Dependency50.P33
  (State IsTraceClass IsTraceClassChannel)
open QAlgorithms.Papers.Independent.Claimed50.P19
  (Cov FockIndex DensityKernel GaussianWithCovariance CharacteristicSummable
   multimodeWeylTrace covarianceCharacteristicValue modeCovarianceBlock
   standardFormCovariance)
open QAlgorithms.Papers.Independent.Renew46.P39 (AffineMultimodeGaussianUnitary)
open QAlgorithms.Papers.Independent.Frontier30.P08 (interleavedSymplecticForm)
open QAlgorithms.Papers.Independent.Dependency50.P10 (complexify)
open QAlgorithms.Papers.Independent.Frontier30.P14 (boundedUnitaryConjugation)

attribute [local instance] Classical.propDecidable Classical.decEq

/-- PDF pp.1--2 and note [12], p.4: actual normalized two-oscillator Gaussian
state, specified finite covariance, arbitrary first moments and finite energy.
The accepted characteristic uses W(v)=exp(i v^T O); the source's D(z) is
W(Sigma^T z). The covariance normalization is unchanged by this reparametrization. -/
structure TwoModeGaussianDensity where
  density : State (FockIndex 2)
  covariance : Cov 2
  gaussian : GaussianWithCovariance density.coefficients covariance

/-- PDF p.2: the four real coordinates of the local symplectic normal form. -/
structure NormalVector where
  one : ℝ
  two : ℝ
  three : ℝ
  four : ℝ

/-- PDF p.2, below (1): compute the unique associated vector from the four
determinants. For physical covariances the marginal determinants are positive
and the discriminants are nonnegative. The larger correlation singular value
is nonnegative, and the sign of the other is fixed by det B. At c=0 both
correlations vanish, so the explicit zero branch avoids division by zero.
No assertion about nonphysical matrices is used in any theorem. -/
def associatedVector (Gamma : Cov 2) : NormalVector :=
  let a := (modeCovarianceBlock Gamma 0 0).det
  let b := (modeCovarianceBlock Gamma 1 1).det
  let p := (modeCovarianceBlock Gamma 0 1).det
  let q := (p ^ 2 - Gamma.det + a * b) / Real.sqrt (a * b)
  let c := Real.sqrt ((q + Real.sqrt (q ^ 2 - 4 * p ^ 2)) / 2)
  ⟨Real.sqrt a, Real.sqrt b, c, if c = 0 then 0 else p / c⟩

/-- PDF p.2: logCriterionPolynomial, transcribed from its explicit definition, not intermediate (8). -/
def logCriterionPolynomial (a b c d : ℝ) : ℝ :=
  (a ^ 2 - 1) + (b ^ 2 - 1) * c ^ 2 * d ^ 2 + 2 * c * d -
    a * b * (c ^ 2 + d ^ 2)

/-- PDF p.2: f1 for the source and target covariances. -/
def f1 (Gamma target : Cov 2) (x y : ℝ) : ℝ :=
  let s := associatedVector Gamma
  let t := associatedVector target
  logCriterionPolynomial t.one s.one (x / s.three) (y / s.four)

/-- PDF p.2: f2. In the stated criteria w2>0 enforces x*y != 0;
ordinary division is therefore used only on its nonzero domain. -/
def f2 (Gamma target : Cov 2) (x y : ℝ) : ℝ :=
  let s := associatedVector Gamma
  let t := associatedVector target
  logCriterionPolynomial t.two s.two (t.three / x) (t.four / y)

/-- PDF p.3, (9): upper admissible absolute correlation product. -/
def w1 (Gamma target : Cov 2) : ℝ :=
  let s := associatedVector Gamma
  let t := associatedVector target
  |s.three * s.four| * t.one / s.one

/-- PDF p.3, (9): lower admissible absolute correlation product. -/
def w2 (Gamma target : Cov 2) : ℝ :=
  let s := associatedVector Gamma
  let t := associatedVector target
  |t.three * t.four| * s.two / t.two

/-- PDF pp.1--2: a single retained oscillator followed by m ancillary oscillators.
This is a basis relabeling, not a finite occupation cutoff. -/
def joinOccupation {m : ℕ} (u : FockIndex 1) (e : FockIndex m) :
    FockIndex (1 + m) :=
  fun j => if h : j.val = 0 then u 0 else e ⟨j.val - 1, by omega⟩

/-- PDF pp.1--2: retained occupation coordinate in a local dilation. -/
def firstOccupation {m : ℕ} (u : FockIndex (1 + m)) : FockIndex 1 :=
  fun _ => u ⟨0, by omega⟩

/-- PDF pp.1--2: all ancillary occupation coordinates, including m=0. -/
def restOccupation {m : ℕ} (u : FockIndex (1 + m)) : FockIndex m :=
  fun j => u ⟨j.val + 1, by omega⟩

/-- PDF pp.1--2 and Fig.1: independent input and ancillary density product,
transported to the completed (1+m)-mode space. -/
def appendProduct {m : ℕ} (X : DensityKernel 1) (sigma : DensityKernel m) :
    DensityKernel (1 + m) :=
  fun u v => X (firstOccupation u) (firstOccupation v) *
    sigma (restOccupation u) (restOccupation v)

/-- PDF pp.1--2, Fig.1: actual bounded-unitary conjugation followed by the
ancillary partial trace. The unitary and ancilla are fixed for every input. -/
def localDilationOutput {m : ℕ} (U : AffineMultimodeGaussianUnitary (1 + m))
    (sigma : DensityKernel m) (X : DensityKernel 1) : DensityKernel 1 :=
  fun u v => ∑' e : FockIndex m,
    boundedUnitaryConjugation U.unitary (appendProduct X sigma)
      (joinOccupation u e) (joinOccupation v e)

/-- PDF pp.1--2, Fig.1 and p.4 note [10]: one local LOG, with arbitrary finite
ancilla count, actual normalized Gaussian environment, and a Gaussian unitary
including translations. Its action on EVERY trace-class input is fixed;
complete positivity and trace preservation are explicit. -/
def IsLocalGaussianDilation (Phi : DensityKernel 1 → DensityKernel 1) : Prop :=
  IsTraceClassChannel Phi ∧
  ∃ (m : ℕ) (sigma : State (FockIndex m)) (gamma : Cov m)
    (U : AffineMultimodeGaussianUnitary (1 + m)),
    GaussianWithCovariance sigma.coefficients gamma ∧
    ∀ X, IsTraceClass X → Phi X = localDilationOutput U sigma.coefficients X

/-- PDF pp.1--2: tensor product of the two independent local maps, retaining
off-diagonal blocks. The inner map acts on Bob and the outer map on Alice. -/
def localProduct (Phi Psi : DensityKernel 1 → DensityKernel 1)
    (X : DensityKernel 2) : DensityKernel 2 :=
  fun u v =>
    Phi (fun a b =>
      Psi (fun c d => X (joinOccupation a c) (joinOccupation b d))
        (restOccupation (m := 1) u) (restOccupation (m := 1) v))
      (firstOccupation (m := 1) u) (firstOccupation (m := 1) v)

/-- PDF pp.1--2: actual state convertibility by LOG. The operations may be
chosen for this known input, but each is a lawful channel on all inputs.
The two ancillas are independent; no shared randomness is introduced. -/
def LOGReachable (rho target : TwoModeGaussianDensity) : Prop :=
  ∃ Phi Psi, IsLocalGaussianDilation Phi ∧ IsLocalGaussianDilation Psi ∧
    localProduct Phi Psi rho.density.coefficients = target.density.coefficients

/-- Proposition 1, PDF pp.2--3: noisy LOG on Alice with reversible Gaussian
unitary/translation on Bob, as the proof expressly permits. Independent
normal-form changes are included by composition of the local operations. -/
def LOGInSystemOne (rho target : TwoModeGaussianDensity) : Prop :=
  ∃ (Phi : DensityKernel 1 → DensityKernel 1)
    (U : AffineMultimodeGaussianUnitary 1),
    IsLocalGaussianDilation Phi ∧
    localProduct Phi (boundedUnitaryConjugation U.unitary)
      rho.density.coefficients = target.density.coefficients

/-- PDF p.2, (3): the EXACT source plus-i local CP constraint for the passive
matrix M in Gamma -> M^T Gamma M+G. No invertibility of M or G is required. -/
def LocalCP (M G : Cov 1) : Prop :=
  G.transpose = G ∧
    (complexify G + Complex.I • complexify (interleavedSymplecticForm 1) -
      Complex.I • complexify
        (M.transpose * interleavedSymplecticForm 1 * M)).PosSemidef

/-- PDF p.2, (2)--(3), and note [14], p.4: the affine Gaussian CPTP map,
expressed on all actual density operators by its Weyl characteristic.
For W(v)=exp(i v^T O), the source passive covariance convention pulls v back
to M v. The noise multiplier has /4 because vacuum covariance is I.
Summability excludes total-tsum defaults, including on non-Gaussian inputs. -/
def IsAffineLocalGaussian (Phi : DensityKernel 1 → DensityKernel 1)
    (M G : Cov 1) : Prop :=
  IsTraceClassChannel Phi ∧ LocalCP M G ∧
  ∃ d : (Fin 1 × Fin 2) → ℝ, ∀ (rho : State (FockIndex 1)) v,
    CharacteristicSummable rho.coefficients (M.mulVec v) ∧
    CharacteristicSummable (Phi rho.coefficients) v ∧
    multimodeWeylTrace (Phi rho.coefficients) v =
      multimodeWeylTrace rho.coefficients (M.mulVec v) *
        covarianceCharacteristicValue G d v

/-- PDF p.2 and p.4 note [13]: block diagonal local passive matrix S1 direct-sum S2. -/
def localSymplecticMatrix (S1 S2 : Cov 1) : Cov 2 :=
  fun i j => if i.1 = j.1 then
    (if i.1 = 0 then S1 (0, i.2) (0, j.2) else S2 (0, i.2) (0, j.2)) else 0

/-- PDF p.2 and p.4 note [13]: local symplectic action in the source convention. -/
def localCongruence (Gamma : Cov 2) (S1 S2 : Cov 1) : Cov 2 :=
  let S := localSymplecticMatrix S1 S2
  S.transpose * Gamma * S

/-! Sanity examples pin the identity-channel polynomial, actual basis ordering,
empty ancilla, uncertainty sign, and block placement. Source proofs are omitted. -/

example (a : ℝ) : logCriterionPolynomial a a 1 1 = 0 := by
  unfold logCriterionPolynomial
  ring

example (Gamma : Cov 2)
    (hthree : (associatedVector Gamma).three ≠ 0)
    (hfour : (associatedVector Gamma).four ≠ 0) :
    f1 Gamma Gamma (associatedVector Gamma).three
      (associatedVector Gamma).four = 0 := by
  dsimp [f1]
  rw [div_self hthree, div_self hfour]
  unfold logCriterionPolynomial
  ring

example {m : ℕ} (u : FockIndex 1) (e : FockIndex m) :
    firstOccupation (joinOccupation u e) = u := by
  funext i
  have hi : i = 0 := Subsingleton.elim _ _
  subst i
  simp [firstOccupation, joinOccupation]

example (u : FockIndex 1) (e : FockIndex 0) :
    joinOccupation u e 0 = u 0 := by
  simp [joinOccupation]

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

example (S1 S2 : Cov 1) :
    localSymplecticMatrix S1 S2 (1, 0) (1, 1) = S2 (0, 0) (0, 1) := by
  simp [localSymplecticMatrix]

end QAlgorithms.Papers.Independent.Gap48.P47
Source
Conditions for the local manipulation of Gaussian states, arXiv:quant-ph/0109126v2, https://arxiv.org/abs/quant-ph/0109126v2
Read-back

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

QAlgorithms.Papers.Independent.Gap48.P47.IsAffineLocalGaussian

For a map Φ\PhiΦ on one-mode complex kernels and arbitrary real 2×22\times22×2 matrices M,GM,GM,G, this predicate requires a channel condition, the conditions GT=GG^{\mathsf T}=GGT=G and G+iJ−iMTJM≥0G+iJ-iM^{\mathsf T}JM\ge0G+iJ−iMTJM≥0, and the existence of a single real two-component vector ddd such that, for every one-mode density kernel RRR and every real two-component vector vvv,

χΦR(v)=χR(Mv)exp⁡ ⁣(−vTGv4+i d⋅v).\chi_{\Phi R}(v) = \chi_R(Mv)\exp\!\left(-\frac{v^{\mathsf T}Gv}{4}+i\,d\cdot v\right).χΦR​(v)=χR​(Mv)exp(−4vTGv​+id⋅v).

Here J=\begin{psmallmatrix}0&1\\-1&0\end{psmallmatrix}, real matrices are embedded entrywise in the complex positive-semidefinite condition, and

χK(v)=∑r=0∞∑s=0∞KrsD ⁣(−v1+iv02)sr,D(z)rs=e−∣z∣2/2r!s!∑j=0min⁡(r,s)zr−j(−z‾)s−jj!(r−j)!(s−j)!.\chi_K(v)=\sum_{r=0}^{\infty}\sum_{s=0}^{\infty} K_{rs}D\!\left(\frac{-v_1+iv_0}{\sqrt2}\right)_{sr}, \qquad D(z)_{rs}=e^{-|z|^2/2}\sqrt{r!s!} \sum_{j=0}^{\min(r,s)} \frac{z^{r-j}(-\overline z)^{s-j}}{j!(r-j)!(s-j)!}.χK​(v)=r=0∑∞​s=0∑∞​Krs​D(2​−v1​+iv0​​)sr​,D(z)rs​=e−∣z∣2/2r!s!​j=0∑min(r,s)​j!(r−j)!(s−j)!zr−j(−z)s−j​.

For both χR(Mv)\chi_R(Mv)χR​(Mv) and χΦR(v)\chi_{\Phi R}(v)χΦR​(v), every inner series and the series of their sums must be summable. A kernel on an index set AAA is nuclear here if it has entries ∑t′ut(a)wt(b)‾\sum_t'u_t(a)\overline{w_t(b)}∑t′​ut​(a)wt​(b)​ for sequences ut,wt∈ℓ2(A;C)u_t,w_t\in\ell^2(A;\mathbb C)ut​,wt​∈ℓ2(A;C) with summable ∥ut∥∥wt∥\|u_t\|\|w_t\|∥ut​∥∥wt​∥; it is a density kernel if it is nuclear, every finite quadratic form is real and nonnegative, and its real diagonal is summable with sum 111. The channel condition says that Φ\PhiΦ preserves nuclear kernels, is additive on nuclear inputs and complex homogeneous on nuclear inputs, and, for every q∈Nq\in\mathbb Nq∈N, sends every density kernel KKK on N×{0,…,q−1}\mathbb N\times\{0,\ldots,q-1\}N×{0,…,q−1} to a density kernel under the amplification

Φ~(K)(r,a),(s,b)=Φ ⁣((K(j,a),(k,b))j,k∈N)rs.\widetilde\Phi(K)_{(r,a),(s,b)} = \Phi\!\left((K_{(j,a),(k,b)})_{j,k\in\mathbb N}\right)_{rs}.Φ(K)(r,a),(s,b)​=Φ((K(j,a),(k,b)​)j,k∈N​)rs​.

The test includes q=0q=0q=0, when no density kernel exists on the empty index set. The vector ddd is independent of RRR and vvv; no positive-definiteness condition on GGG is separately imposed.

QAlgorithms.Papers.Independent.Gap48.P47.IsLocalGaussianDilation

For a map Φ\PhiΦ from one-mode complex coefficient kernels to one-mode complex coefficient kernels, this predicate requires the following channel condition and dilation representation. For any occupation index set AAA, call a kernel RRR nuclear if there are sequences ur,vr∈ℓ2(A;C)u_r,v_r\in\ell^2(A;\mathbb C)ur​,vr​∈ℓ2(A;C) such that ∑r∥ur∥∥vr∥\sum_r\|u_r\|\|v_r\|∑r​∥ur​∥∥vr​∥ is summable and Rij=∑r′ur(i)vr(j)‾R_{ij}=\sum_r'u_r(i)\overline{v_r(j)}Rij​=∑r′​ur​(i)vr​(j)​; call it a density kernel if it is nuclear, every finite quadratic form ∑i,j∈Ezi‾Rijzj\sum_{i,j\in E}\overline{z_i}R_{ij}z_j∑i,j∈E​zi​​Rij​zj​ is real and nonnegative, and ∑i′Re⁡Rii=1\sum_i' \operatorname{Re}R_{ii}=1∑i′​ReRii​=1 with summable real diagonal. The channel condition says that Φ\PhiΦ preserves nuclear kernels, is additive on pairs of nuclear kernels and homogeneous over C\mathbb CC on nuclear kernels, and, for every q∈Nq\in\mathbb Nq∈N and every density kernel RRR indexed by F1×{0,…,q−1}F_1\times\{0,\ldots,q-1\}F1​×{0,…,q−1}, the kernel

Φ~(R)(j,a),(k,b)=Φ ⁣((R(i,a),(i′,b))i,i′)jk\widetilde\Phi(R)_{(j,a),(k,b)} = \Phi\!\left((R_{(i,a),(i',b)})_{i,i'}\right)_{jk}Φ(R)(j,a),(k,b)​=Φ((R(i,a),(i′,b)​)i,i′​)jk​

is a density kernel. Here Fn={0,…,n−1}→NF_n=\{0,\ldots,n-1\}\to\mathbb NFn​={0,…,n−1}→N. The dilation requirement is that there exist m∈Nm\in\mathbb Nm∈N, an environment density kernel EEE on FmF_mFm​, a real covariance matrix γ\gammaγ, and an affine Gaussian unitary UUU on ℓ2(F1+m;C)\ell^2(F_{1+m};\mathbb C)ℓ2(F1+m​;C), such that EEE has Gaussian covariance γ\gammaγ and, for every nuclear one-mode kernel XXX,

(ΦX)uv=∑e∈Fm′∑a∈F1+m′U(u,e),a(∑b∈F1+m′Xa0b0EarestbrestU(v,e),b‾).(\Phi X)_{uv} = \sum_{e\in F_m}'\sum_{a\in F_{1+m}}' U_{(u,e),a} \left(\sum_{b\in F_{1+m}}' X_{a_0b_0}E_{a_{\mathrm{rest}}b_{\mathrm{rest}}} \overline{U_{(v,e),b}}\right).(ΦX)uv​=e∈Fm​∑′​a∈F1+m​∑′​U(u,e),a​​b∈F1+m​∑′​Xa0​b0​​Earest​brest​​U(v,e),b​​​.

Gaussian covariance here means that γ\gammaγ is positive definite, γ+iΩm\gamma+i\Omega_mγ+iΩm​ is positive semidefinite, the series ∑α∈Fm(∑kαk)Re⁡Eαα\sum_{\alpha\in F_m}(\sum_k\alpha_k)\operatorname{Re}E_{\alpha\alpha}∑α∈Fm​​(∑k​αk​)ReEαα​ is summable, and there is a real mean vector hhh such that, for every real quadrature vector vvv, each inner series and the outer series in

∑α′∑β′EαβWm(v)βα=e−vTγv/4+ih⋅v\sum_\alpha'\sum_\beta' E_{\alpha\beta}W_m(v)_{\beta\alpha} = e^{-v^{\mathsf T}\gamma v/4+i h\cdot v}α∑′​β∑′​Eαβ​Wm​(v)βα​=e−vTγv/4+ih⋅v

are summable. The quadrature order is (k,0),(k,1)(k,0),(k,1)(k,0),(k,1), Ωm\Omega_mΩm​ has blocks \begin{psmallmatrix}0&1\\-1&0\end{psmallmatrix}, and Wm(v)αβ=∏kD((−vk,1+ivk,0)/2)αkβkW_m(v)_{\alpha\beta}=\prod_kD((-v_{k,1}+iv_{k,0})/\sqrt2)_{\alpha_k\beta_k}Wm​(v)αβ​=∏k​D((−vk,1​+ivk,0​)/2​)αk​βk​​, where

D(z)rs=e−∣z∣2/2r!s!∑j=0min⁡(r,s)zr−j(−z‾)s−jj!(r−j)!(s−j)!.D(z)_{rs}=e^{-|z|^2/2}\sqrt{r!s!} \sum_{j=0}^{\min(r,s)} \frac{z^{r-j}(-\overline z)^{s-j}}{j!(r-j)!(s-j)!}.D(z)rs​=e−∣z∣2/2r!s!​j=0∑min(r,s)​j!(r−j)!(s−j)!zr−j(−z)s−j​.

The affine Gaussian unitary comprises a real SSS, real displacement bbb, and unitary UUU, with SΩ1+mST=Ω1+mS\Omega_{1+m}S^{\mathsf T}=\Omega_{1+m}SΩ1+m​ST=Ω1+m​ and coefficientwise U−1W1+m(v)U=eib⋅vW1+m(STv)U^{-1}W_{1+m}(v)U=e^{i b\cdot v}W_{1+m}(S^{\mathsf T}v)U−1W1+m​(v)U=eib⋅vW1+m​(STv); coefficientwise conjugation means (UKU∗)ab=∑i′Uai(∑j′KijUbj‾)(UKU^*)_{ab}=\sum_i'U_{ai}(\sum_j'K_{ij}\overline{U_{bj}})(UKU∗)ab​=∑i′​Uai​(∑j′​Kij​Ubj​​). Both m=0m=0m=0 and the channel test q=0q=0q=0 are included; F0F_0F0​ is a singleton, whereas there is no density kernel on an empty index set. The representation is required only for nuclear XXX.

QAlgorithms.Papers.Independent.Gap48.P47.LOGInSystemOne

For two supplied two-mode Gaussian density objects ρ\rhoρ and τ\tauτ, this predicate says that there exist a one-mode kernel map Φ\PhiΦ admitting a local Gaussian dilation and a one-mode affine Gaussian unitary UUU such that applying unitary conjugation by UUU to each second-mode slice and then Φ\PhiΦ to the resulting first-mode kernel gives exactly τ\tauτ. In occupation coordinates this means

τ(u0,u1),(v0,v1)=[Φ ⁣((∑c′Uu1c(∑d′ρ(a,c),(b,d)Uv1d‾))a,b)]u0v0.\tau_{(u_0,u_1),(v_0,v_1)} = \left[ \Phi\!\left( \left( \sum_c'U_{u_1c} \left(\sum_d'\rho_{(a,c),(b,d)}\overline{U_{v_1d}}\right) \right)_{a,b} \right) \right]_{u_0v_0}.τ(u0​,u1​),(v0​,v1​)​=​Φ​(c∑′​Uu1​c​(d∑′​ρ(a,c),(b,d)​Uv1​d​​))a,b​​​u0​v0​​.

All displayed occupation coordinates range over N\mathbb NN. The Gaussian density and dilation requirements are as follows: for Fn={0,…,n−1}→NF_n=\{0,\ldots,n-1\}\to\mathbb NFn​={0,…,n−1}→N and Hn=ℓ2(Fn;C)H_n=\ell^2(F_n;\mathbb C)Hn​=ℓ2(Fn​;C), a nuclear kernel admits sequences ur,vr∈Hnu_r,v_r\in H_nur​,vr​∈Hn​ with summable ∥ur∥∥vr∥\|u_r\|\|v_r\|∥ur​∥∥vr​∥ and entries ∑r′ur(α)vr(β)‾\sum_r'u_r(\alpha)\overline{v_r(\beta)}∑r′​ur​(α)vr​(β)​; a density kernel additionally has real nonnegative quadratic forms on every finite set and summable real diagonal of sum 111. A Gaussian density object pairs a density kernel RRR with a real positive definite covariance VVV satisfying V+iΩn≥0V+i\Omega_n\ge0V+iΩn​≥0, requires summability of (∑kαk)Re⁡Rαα(\sum_k\alpha_k)\operatorname{Re}R_{\alpha\alpha}(∑k​αk​)ReRαα​, and has some real mean hhh such that, for all real quadrature vectors vvv,

∑α′∑β′RαβWn(v)βα=e−vTVv/4+ih⋅v,\sum_\alpha'\sum_\beta'R_{\alpha\beta}W_n(v)_{\beta\alpha} =e^{-v^{\mathsf T}Vv/4+i h\cdot v},α∑′​β∑′​Rαβ​Wn​(v)βα​=e−vTVv/4+ih⋅v,

with each inner series and the outer series summable. Here Ωn\Omega_nΩn​ has blocks \begin{psmallmatrix}0&1\\-1&0\end{psmallmatrix},

Wn(v)αβ=∏kD ⁣(−vk,1+ivk,02)αkβk,D(z)rs=e−∣z∣2/2r!s!∑j=0min⁡(r,s)zr−j(−z‾)s−jj!(r−j)!(s−j)!.W_n(v)_{\alpha\beta} =\prod_kD\!\left(\frac{-v_{k,1}+iv_{k,0}}{\sqrt2}\right)_{\alpha_k\beta_k}, \qquad D(z)_{rs}=e^{-|z|^2/2}\sqrt{r!s!} \sum_{j=0}^{\min(r,s)} \frac{z^{r-j}(-\overline z)^{s-j}}{j!(r-j)!(s-j)!}.Wn​(v)αβ​=k∏​D(2​−vk,1​+ivk,0​​)αk​βk​​,D(z)rs​=e−∣z∣2/2r!s!​j=0∑min(r,s)​j!(r−j)!(s−j)!zr−j(−z)s−j​.

An affine Gaussian unitary on nnn modes comprises a unitary V:Hn→HnV:H_n\to H_nV:Hn​→Hn​, a real symplectic matrix SSS satisfying SΩnST=ΩnS\Omega_nS^{\mathsf T}=\Omega_nSΩn​ST=Ωn​, and real displacement bbb, with coefficientwise V−1Wn(v)V=eib⋅vWn(STv)V^{-1}W_n(v)V=e^{i b\cdot v}W_n(S^{\mathsf T}v)V−1Wn​(v)V=eib⋅vWn​(STv). Coefficients mean Vαβ=(Veβ)αV_{\alpha\beta}=(Ve_\beta)_\alphaVαβ​=(Veβ​)α​, and conjugation means (VKV∗)αβ=∑a′Vαa(∑b′KabVβb‾)(VKV^*)_{\alpha\beta}=\sum_a'V_{\alpha a}(\sum_b'K_{ab}\overline{V_{\beta b}})(VKV∗)αβ​=∑a′​Vαa​(∑b′​Kab​Vβb​​). The local dilation condition on Φ\PhiΦ requires preservation of nuclear kernels, complex linearity on nuclear kernels, and preservation of density kernels under the slice amplification to F1×{0,…,q−1}F_1\times\{0,\ldots,q-1\}F1​×{0,…,q−1} for every q∈Nq\in\mathbb Nq∈N; it further requires an m∈Nm\in\mathbb Nm∈N, an environment Gaussian density kernel EEE, and an affine Gaussian unitary VVV on 1+m1+m1+m modes such that, for every nuclear XXX,

(ΦX)uv=∑e∈Fm′∑a∈F1+m′V(u,e),a(∑b∈F1+m′Xa0b0EarestbrestV(v,e),b‾).(\Phi X)_{uv} =\sum_{e\in F_m}'\sum_{a\in F_{1+m}}' V_{(u,e),a} \left(\sum_{b\in F_{1+m}}' X_{a_0b_0}E_{a_{\mathrm{rest}}b_{\mathrm{rest}}} \overline{V_{(v,e),b}}\right).(ΦX)uv​=e∈Fm​∑′​a∈F1+m​∑′​V(u,e),a​​b∈F1+m​∑′​Xa0​b0​​Earest​brest​​V(v,e),b​​​.

The case m=0m=0m=0, with singleton F0F_0F0​, is allowed; the q=0q=0q=0 density test is vacuous.

QAlgorithms.Papers.Independent.Gap48.P47.LOGReachable

For two supplied two-mode Gaussian density objects ρ\rhoρ and τ\tauτ, this predicate says that there exist maps Φ,Ψ\Phi,\PsiΦ,Ψ on one-mode complex kernels, each admitting the local Gaussian dilation described below, such that their successive slice action sends the coefficient kernel of ρ\rhoρ exactly to that of τ\tauτ:

τ(u0,u1),(v0,v1)=[Φ ⁣(([Ψ ⁣((ρ(a,c),(b,d))c,d)]u1v1)a,b)]u0v0.\tau_{(u_0,u_1),(v_0,v_1)} = \left[ \Phi\!\left( \left( \left[\Psi\!\left((\rho_{(a,c),(b,d)})_{c,d}\right)\right]_{u_1v_1} \right)_{a,b} \right) \right]_{u_0v_0}.τ(u0​,u1​),(v0​,v1​)​=[Φ(([Ψ((ρ(a,c),(b,d)​)c,d​)]u1​v1​​)a,b​)]u0​v0​​.

All occupation coordinates range over N\mathbb NN. To specify the objects and dilation condition, put Fn={0,…,n−1}→NF_n=\{0,\ldots,n-1\}\to\mathbb NFn​={0,…,n−1}→N, Hn=ℓ2(Fn;C)H_n=\ell^2(F_n;\mathbb C)Hn​=ℓ2(Fn​;C), and let Ωn\Omega_nΩn​ have the blocks \begin{psmallmatrix}0&1\\-1&0\end{psmallmatrix} in mode order. A nuclear kernel RRR admits ur,vr∈Hnu_r,v_r\in H_nur​,vr​∈Hn​ with summable ∥ur∥∥vr∥\|u_r\|\|v_r\|∥ur​∥∥vr​∥ and Rαβ=∑r′ur(α)vr(β)‾R_{\alpha\beta}=\sum_r'u_r(\alpha)\overline{v_r(\beta)}Rαβ​=∑r′​ur​(α)vr​(β)​. A density kernel is nuclear, has real nonnegative quadratic forms on every finite set of indices, and has summable Re⁡Rαα\operatorname{Re}R_{\alpha\alpha}ReRαα​ with sum 111. Each supplied Gaussian density object consists of such a density kernel RRR, a real covariance VVV with VVV positive definite and V+iΩnV+i\Omega_nV+iΩn​ positive semidefinite, summable (∑kαk)Re⁡Rαα(\sum_k\alpha_k)\operatorname{Re}R_{\alpha\alpha}(∑k​αk​)ReRαα​, and some real mean hhh for which

∑α′∑β′RαβWn(v)βα=e−vTVv/4+ih⋅v\sum_\alpha'\sum_\beta'R_{\alpha\beta}W_n(v)_{\beta\alpha} =e^{-v^{\mathsf T}Vv/4+i h\cdot v}α∑′​β∑′​Rαβ​Wn​(v)βα​=e−vTVv/4+ih⋅v

for every real vvv, with each inner series and the outer series summable. Here

Wn(v)αβ=∏kD ⁣(−vk,1+ivk,02)αkβk,D(z)rs=e−∣z∣2/2r!s!∑j=0min⁡(r,s)zr−j(−z‾)s−jj!(r−j)!(s−j)!.W_n(v)_{\alpha\beta} =\prod_kD\!\left(\frac{-v_{k,1}+iv_{k,0}}{\sqrt2}\right)_{\alpha_k\beta_k}, \qquad D(z)_{rs}=e^{-|z|^2/2}\sqrt{r!s!} \sum_{j=0}^{\min(r,s)} \frac{z^{r-j}(-\overline z)^{s-j}}{j!(r-j)!(s-j)!}.Wn​(v)αβ​=k∏​D(2​−vk,1​+ivk,0​​)αk​βk​​,D(z)rs​=e−∣z∣2/2r!s!​j=0∑min(r,s)​j!(r−j)!(s−j)!zr−j(−z)s−j​.

A map TTT admitting a local Gaussian dilation must preserve nuclear kernels, be complex linear on them, and send every density kernel on F1×{0,…,q−1}F_1\times\{0,\ldots,q-1\}F1​×{0,…,q−1}, for every q∈Nq\in\mathbb Nq∈N, to a density kernel when applied to its F1F_1F1​-indexed slices. It must also admit some m∈Nm\in\mathbb Nm∈N, an environment density kernel EEE with Gaussian covariance in the preceding sense, and a unitary U:H1+m→H1+mU:H_{1+m}\to H_{1+m}U:H1+m​→H1+m​, together with real S,bS,bS,b satisfying SΩ1+mST=Ω1+mS\Omega_{1+m}S^{\mathsf T}=\Omega_{1+m}SΩ1+m​ST=Ω1+m​ and U−1W1+m(v)U=eib⋅vW1+m(STv)U^{-1}W_{1+m}(v)U=e^{i b\cdot v}W_{1+m}(S^{\mathsf T}v)U−1W1+m​(v)U=eib⋅vW1+m​(STv), such that for every nuclear XXX,

(TX)uv=∑e∈Fm′∑a∈F1+m′U(u,e),a(∑b′∈F1+m′Xa0b0′Earestbrest′U(v,e),b′‾).(TX)_{uv} =\sum_{e\in F_m}'\sum_{a\in F_{1+m}}' U_{(u,e),a} \left(\sum_{b'\in F_{1+m}}' X_{a_0b'_0}E_{a_{\mathrm{rest}}b'_{\mathrm{rest}}} \overline{U_{(v,e),b'}}\right).(TX)uv​=e∈Fm​∑′​a∈F1+m​∑′​U(u,e),a​​b′∈F1+m​∑′​Xa0​b0′​​Earest​brest′​​U(v,e),b′​​​.

Unitary coefficients are Uαβ=(Ueβ)αU_{\alpha\beta}=(Ue_\beta)_\alphaUαβ​=(Ueβ​)α​, and unitary conjugations use the displayed nested-sum convention. The two maps may have different environments and unitaries. Zero environment modes are allowed, with F0F_0F0​ a singleton; the q=0q=0q=0 channel test has no density inputs.

QAlgorithms.Papers.Independent.Gap48.P47.LocalCP

For arbitrary real 2×22\times22×2 matrices MMM and GGG, indexed in the one-mode quadrature order (0,0),(0,1)(0,0),(0,1)(0,0),(0,1), this predicate means precisely

GT=GandG+iJ−iMTJM is positive semidefinite over C,J=(01−10).G^{\mathsf T}=G \quad\text{and}\quad G+iJ-iM^{\mathsf T}JM\ \text{is positive semidefinite over }\mathbb C, \qquad J=\begin{pmatrix}0&1\\-1&0\end{pmatrix}.GT=GandG+iJ−iMTJM is positive semidefinite over C,J=(0−1​10​).

Each real matrix in the complex expression is embedded entrywise into C\mathbb CC. There is no separate invertibility or symplecticity hypothesis on MMM.

QAlgorithms.Papers.Independent.Gap48.P47.NormalVector

A normal vector is an ordered collection of four arbitrary real numbers, with fields named one\mathrm{one}one, two\mathrm{two}two, three\mathrm{three}three, and four\mathrm{four}four; no equations, inequalities, or other restrictions are imposed on them.

QAlgorithms.Papers.Independent.Gap48.P47.TwoModeGaussianDensity

A two-mode Gaussian density consists of a complex kernel ρ=(ρij)\rho=(\rho_{ij})ρ=(ρij​) indexed by occupation functions i,j:{0,1}→Ni,j:\{0,1\}\to\mathbb Ni,j:{0,1}→N, a real matrix VVV indexed by {0,1}×{0,1}\{0,1\}\times\{0,1\}{0,1}×{0,1}, and the following properties. 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) such that ∑r∥ur∥∥vr∥\sum_r\|u_r\|\|v_r\|∑r​∥ur​∥∥vr​∥ converges and ρij=∑rur(i)vr(j)‾\rho_{ij}=\sum_r u_r(i)\overline{v_r(j)}ρij​=∑r​ur​(i)vr​(j)​; for every finite set SSS of occupation functions and every complex function zzz on occupation functions, ∑i,j∈Szi‾ρijzj\sum_{i,j\in S}\overline{z_i}\rho_{ij}z_j∑i,j∈S​zi​​ρij​zj​ is real and nonnegative; the family (Re⁡ρii)i(\operatorname{Re}\rho_{ii})_i(Reρii​)i​ is summable with sum 111; and the family ((i(0)+i(1))Re⁡ρii)i((i(0)+i(1))\operatorname{Re}\rho_{ii})_i((i(0)+i(1))Reρii​)i​ is summable. Moreover, VVV is positive definite and V+iΩV+i\OmegaV+iΩ is positive semidefinite, where Ω\OmegaΩ has one block (01−10)\begin{pmatrix}0&1\\-1&0\end{pmatrix}(0−1​10​) for each mode and zero entries between modes. There exists a real vector mmm, indexed by the same four quadratures as VVV, such that for every real quadrature vector vvv, putting zk=(−v(k,1)+iv(k,0))/2z_k=(-v_{(k,1)}+i v_{(k,0)})/\sqrt2zk​=(−v(k,1)​+iv(k,0)​)/2​, each inner series below is summable, the series of their sums is summable, and

∑i∑jρij∏k=01D(zk)j(k),i(k)=exp⁡ ⁣(−14∑α,βvαVαβvβ+i∑αmαvα).\sum_i\sum_j \rho_{ij}\prod_{k=0}^{1}D(z_k)_{j(k),i(k)} =\exp\!\left(-\frac14\sum_{\alpha,\beta}v_\alpha V_{\alpha\beta}v_\beta +i\sum_\alpha m_\alpha v_\alpha\right).i∑​j∑​ρij​k=0∏1​D(zk​)j(k),i(k)​=exp​−41​α,β∑​vα​Vαβ​vβ​+iα∑​mα​vα​​.

Here, for all z∈Cz\in\mathbb Cz∈C and a,b∈Na,b\in\mathbb Na,b∈N,

D(z)ab=e−∣z∣2/2a! b!∑ℓ=0min⁡(a,b)za−ℓ(−z‾)b−ℓℓ! (a−ℓ)! (b−ℓ)!.D(z)_{ab} =e^{-|z|^2/2}\sqrt{a!\,b!} \sum_{\ell=0}^{\min(a,b)} \frac{z^{a-\ell}(-\overline z)^{b-\ell}} {\ell!\,(a-\ell)!\,(b-\ell)!}.D(z)ab​=e−∣z∣2/2a!b!​ℓ=0∑min(a,b)​ℓ!(a−ℓ)!(b−ℓ)!za−ℓ(−z)b−ℓ​.

QAlgorithms.Papers.Independent.Gap48.P47.appendProduct

For every natural number mmm, every complex kernel XXX indexed by pairs of functions {0}→N\{0\}\to\mathbb N{0}→N, and every complex kernel σ\sigmaσ indexed by pairs of functions {0,…,m−1}→N\{0,\ldots,m-1\}\to\mathbb N{0,…,m−1}→N, their appended product is the kernel indexed by occupation functions u,v:{0,…,m}→Nu,v:\{0,\ldots,m\}\to\mathbb Nu,v:{0,…,m}→N with entries

(Xappendσ)uv=Xufirst,vfirst σurest,vrest,ufirst(0)=u(0),urest(k)=u(k+1),(X\mathbin{\mathrm{append}}\sigma)_{uv} =X_{u_{\mathrm{first}},v_{\mathrm{first}}}\, \sigma_{u_{\mathrm{rest}},v_{\mathrm{rest}}}, \qquad u_{\mathrm{first}}(0)=u(0),\quad u_{\mathrm{rest}}(k)=u(k+1),(Xappendσ)uv​=Xufirst​,vfirst​​σurest​,vrest​​,ufirst​(0)=u(0),urest​(k)=u(k+1),

with the corresponding definitions for vvv. No density, positivity, normalization, or summability conditions are imposed on either kernel. For m=0m=0m=0, both remaining occupation functions are the unique empty function ∅\varnothing∅, so the result has entries Xuvσ∅,∅X_{uv}\sigma_{\varnothing,\varnothing}Xuv​σ∅,∅​.

QAlgorithms.Papers.Independent.Gap48.P47.associatedVector

For any real matrix Γ\GammaΓ indexed by mode–quadrature pairs (k,α)∈{0,1}×{0,1}(k,\alpha)\in\{0,1\}\times\{0,1\}(k,α)∈{0,1}×{0,1}, let Γkl\Gamma_{kl}Γkl​ denote its 2×22\times22×2 block with entries (Γkl)αβ=Γ(k,α),(l,β)(\Gamma_{kl})_{\alpha\beta}=\Gamma_{(k,\alpha),(l,\beta)}(Γkl​)αβ​=Γ(k,α),(l,β)​, and set

a=det⁡Γ00,b=det⁡Γ11,p=det⁡Γ01,q=p2−det⁡Γ+abab,c=q+q2−4p22.a=\det\Gamma_{00},\qquad b=\det\Gamma_{11},\qquad p=\det\Gamma_{01},\qquad q=\frac{p^2-\det\Gamma+ab}{\sqrt{ab}},\qquad c=\sqrt{\frac{q+\sqrt{q^2-4p^2}}2}.a=detΓ00​,b=detΓ11​,p=detΓ01​,q=ab​p2−detΓ+ab​,c=2q+q2−4p2​​​.

The associated vector has four real components

(a, b, c, {0,c=0,p/c,c≠0.).\left(\sqrt a,\ \sqrt b,\ c,\ \begin{cases}0,&c=0,\\ p/c,&c\ne0.\end{cases}\right).(a​, b​, c, {0,p/c,​c=0,c=0.​).

There are no symmetry, positivity, or physicality assumptions on Γ\GammaΓ. Every square root is the total real square-root function, which is 000 on negative arguments, and division by 000 has value 000.

QAlgorithms.Papers.Independent.Gap48.P47.f1

For arbitrary real 4×44\times44×4 matrices Γ,T\Gamma,TΓ,T, indexed by (k,α)∈{0,1}×{0,1}(k,\alpha)\in\{0,1\}\times\{0,1\}(k,α)∈{0,1}×{0,1}, and arbitrary x,y∈Rx,y\in\mathbb Rx,y∈R, write HklH_{kl}Hkl​ for the 2×22\times22×2 block (H(k,α),(l,β))α,β(H_{(k,\alpha),(l,\beta)})_{\alpha,\beta}(H(k,α),(l,β)​)α,β​ of either matrix HHH. Set

A=det⁡Γ00,B=det⁡Γ11,P=det⁡Γ01,Q=P2−det⁡Γ+ABAB,C=Q+Q2−4P22,D={0,C=0,P/C,C≠0,A=\det\Gamma_{00},\quad B=\det\Gamma_{11},\quad P=\det\Gamma_{01},\quad Q=\frac{P^2-\det\Gamma+AB}{\sqrt{AB}},\quad C=\sqrt{\frac{Q+\sqrt{Q^2-4P^2}}2},\quad D=\begin{cases}0,&C=0,\\P/C,&C\ne0,\end{cases}A=detΓ00​,B=detΓ11​,P=detΓ01​,Q=AB​P2−detΓ+AB​,C=2Q+Q2−4P2​​​,D={0,P/C,​C=0,C=0,​

and put a=det⁡T00a=\sqrt{\det T_{00}}a=detT00​​, b=Ab=\sqrt Ab=A​, u=x/Cu=x/Cu=x/C, and v=y/Dv=y/Dv=y/D. Then

f1(Γ,T;x,y)=(a2−1)+(b2−1)u2v2+2uv−ab(u2+v2).f_1(\Gamma,T;x,y)=(a^2-1)+(b^2-1)u^2v^2+2uv-ab(u^2+v^2).f1​(Γ,T;x,y)=(a2−1)+(b2−1)u2v2+2uv−ab(u2+v2).

Neither matrix is assumed symmetric, positive, or physical; real square roots of negative arguments and quotients with zero denominator have value 000, so the definition includes C=0C=0C=0 and D=0D=0D=0.

QAlgorithms.Papers.Independent.Gap48.P47.f2

For arbitrary real 4×44\times44×4 matrices Γ,T\Gamma,TΓ,T, indexed by (k,α)∈{0,1}×{0,1}(k,\alpha)\in\{0,1\}\times\{0,1\}(k,α)∈{0,1}×{0,1}, and arbitrary x,y∈Rx,y\in\mathbb Rx,y∈R, write HklH_{kl}Hkl​ for the 2×22\times22×2 block (H(k,α),(l,β))α,β(H_{(k,\alpha),(l,\beta)})_{\alpha,\beta}(H(k,α),(l,β)​)α,β​. Define

A=det⁡T00,B=det⁡T11,P=det⁡T01,Q=P2−det⁡T+ABAB,C=Q+Q2−4P22,D={0,C=0,P/C,C≠0.A=\det T_{00},\quad B=\det T_{11},\quad P=\det T_{01},\quad Q=\frac{P^2-\det T+AB}{\sqrt{AB}},\quad C=\sqrt{\frac{Q+\sqrt{Q^2-4P^2}}2},\quad D=\begin{cases}0,&C=0,\\P/C,&C\ne0.\end{cases}A=detT00​,B=detT11​,P=detT01​,Q=AB​P2−detT+AB​,C=2Q+Q2−4P2​​​,D={0,P/C,​C=0,C=0.​

With a=Ba=\sqrt Ba=B​, b=det⁡Γ11b=\sqrt{\det\Gamma_{11}}b=detΓ11​​, u=C/xu=C/xu=C/x, and v=D/yv=D/yv=D/y, the function is

f2(Γ,T;x,y)=(a2−1)+(b2−1)u2v2+2uv−ab(u2+v2).f_2(\Gamma,T;x,y)=(a^2-1)+(b^2-1)u^2v^2+2uv-ab(u^2+v^2).f2​(Γ,T;x,y)=(a2−1)+(b2−1)u2v2+2uv−ab(u2+v2).

No symmetry, positivity, physicality, or nonzero-denominator assumptions are imposed. Real square roots of negative arguments are 000, and division by 000 is 000; in particular, x=0x=0x=0 and y=0y=0y=0 are included.

QAlgorithms.Papers.Independent.Gap48.P47.firstOccupation

For every natural number mmm and every occupation function u:{0,…,m}→Nu:\{0,\ldots,m\}\to\mathbb Nu:{0,…,m}→N, the first occupation is the function on the singleton {0}\{0\}{0} whose value is u(0)u(0)u(0). This also applies when m=0m=0m=0.

QAlgorithms.Papers.Independent.Gap48.P47.joinOccupation

For every natural number mmm, every occupation function u:{0}→Nu:\{0\}\to\mathbb Nu:{0}→N, and every occupation function e:{0,…,m−1}→Ne:\{0,\ldots,m-1\}\to\mathbb Ne:{0,…,m−1}→N, their joined occupation function j:{0,…,m}→Nj:\{0,\ldots,m\}\to\mathbb Nj:{0,…,m}→N is defined by j(0)=u(0)j(0)=u(0)j(0)=u(0) and j(k)=e(k−1)j(k)=e(k-1)j(k)=e(k−1) for 1≤k≤m1\le k\le m1≤k≤m. The case m=0m=0m=0 is included: eee is the unique function on the empty set, and the joined function has only the value u(0)u(0)u(0).

QAlgorithms.Papers.Independent.Gap48.P47.localCongruence

For any real 4×44\times44×4 matrix Γ\GammaΓ indexed by two modes and their two quadratures, and any real 2×22\times22×2 one-mode matrices S1,S2S_1,S_2S1​,S2​, this definition returns

STΓS,S=(S100S2),S^{\mathsf T}\Gamma S, \qquad S=\begin{pmatrix}S_1&0\\0&S_2\end{pmatrix},STΓS,S=(S1​0​0S2​​),

using the 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). No symmetry, positivity, or physical covariance condition on Γ\GammaΓ, and no symplecticity condition on S1,S2S_1,S_2S1​,S2​, is required.

QAlgorithms.Papers.Independent.Gap48.P47.localDilationOutput

For every natural number mmm, let Fn={0,…,n−1}→NF_n=\{0,\ldots,n-1\}\to\mathbb NFn​={0,…,n−1}→N be the set of nnn-mode occupation indices and Hn=ℓ2(Fn;C)H_n=\ell^2(F_n;\mathbb C)Hn​=ℓ2(Fn​;C). The inputs are an affine Gaussian unitary UUU on 1+m1+m1+m modes and arbitrary complex kernels XXX on F1×F1F_1\times F_1F1​×F1​ and σ\sigmaσ on Fm×FmF_m\times F_mFm​×Fm​; neither kernel is assumed to be a density kernel. An affine Gaussian unitary on nnn modes comprises a real matrix SSS, a real vector bbb, and a complex linear isometric equivalence U:Hn→HnU:H_n\to H_nU:Hn​→Hn​, satisfying SΩnST=ΩnS\Omega_nS^{\mathsf T}=\Omega_nSΩn​ST=Ωn​ and U−1Wn(v)U=e ib⋅vWn(STv)U^{-1}W_n(v)U=e^{\,i b\cdot v}W_n(S^{\mathsf T}v)U−1Wn​(v)U=eib⋅vWn​(STv), where conjugation is interpreted coefficientwise by nested infinite sums, Ωn\Omega_nΩn​ has the blocks \begin{psmallmatrix}0&1\\-1&0\end{psmallmatrix} in mode order, and

Wn(v)αβ=∏k=0n−1D ⁣(−vk,1+ivk,02)αkβk,D(z)rs=e−∣z∣2/2r!s!∑j=0min⁡(r,s)zr−j(−z‾)s−jj!(r−j)!(s−j)!.W_n(v)_{\alpha\beta}=\prod_{k=0}^{n-1}D\!\left(\frac{-v_{k,1}+iv_{k,0}}{\sqrt2}\right)_{\alpha_k\beta_k}, \qquad D(z)_{rs}=e^{-|z|^2/2}\sqrt{r!s!} \sum_{j=0}^{\min(r,s)} \frac{z^{r-j}(-\overline z)^{s-j}}{j!(r-j)!(s-j)!}.Wn​(v)αβ​=k=0∏n−1​D(2​−vk,1​+ivk,0​​)αk​βk​​,D(z)rs​=e−∣z∣2/2r!s!​j=0∑min(r,s)​j!(r−j)!(s−j)!zr−j(−z)s−j​.

Writing Uαβ=(Ueβ)αU_{\alpha\beta}=(Ue_\beta)_\alphaUαβ​=(Ueβ​)α​, with eβe_\betaeβ​ the occupation basis vector, and writing a=(a0,arest)∈F1+ma=(a_0,a_{\mathrm{rest}})\in F_{1+m}a=(a0​,arest​)∈F1+m​, the defined output is the one-mode kernel

(localDilationOutput⁡(U,σ,X))uv=∑e∈Fm′ ∑a∈F1+m′U(u,e),a(∑b∈F1+m′Xa0b0 σarestbrest U(v,e),b‾).\bigl(\operatorname{localDilationOutput}(U,\sigma,X)\bigr)_{uv} = \sum_{e\in F_m}'\ \sum_{a\in F_{1+m}}' U_{(u,e),a} \left( \sum_{b\in F_{1+m}}' X_{a_0b_0}\, \sigma_{a_{\mathrm{rest}}b_{\mathrm{rest}}}\, \overline{U_{(v,e),b}} \right).(localDilationOutput(U,σ,X))uv​=e∈Fm​∑′​ a∈F1+m​∑′​U(u,e),a​​b∈F1+m​∑′​Xa0​b0​​σarest​brest​​U(v,e),b​​​.

Here (u,e)(u,e)(u,e) places uuu's occupation first and then the occupations of eee. The sums have exactly the displayed nesting; no summability condition is imposed, and a nonsummable complex series has the total infinite-sum value 000. The case m=0m=0m=0 is included: F0F_0F0​ is a singleton, so the outer environment sum has one term.

QAlgorithms.Papers.Independent.Gap48.P47.localProduct

For arbitrary maps Φ,Ψ\Phi,\PsiΦ,Ψ from one-mode complex kernels to one-mode complex kernels and an arbitrary two-mode complex kernel XXX, the defined two-mode kernel is obtained by applying Ψ\PsiΨ to each second-mode slice, then applying Φ\PhiΦ to the resulting first-mode kernel. Identifying one-mode occupation indices with N\mathbb NN and two-mode indices with pairs of natural numbers, its entries are exactly

(localProduct⁡(Φ,Ψ,X))(u0,u1),(v0,v1)=[Φ ⁣(([Ψ ⁣((X(a,c),(b,d))c,d∈N)]u1v1)a,b∈N)]u0v0.\bigl(\operatorname{localProduct}(\Phi,\Psi,X)\bigr)_{(u_0,u_1),(v_0,v_1)} = \left[ \Phi\!\left( \left( \left[\Psi\!\left((X_{(a,c),(b,d)})_{c,d\in\mathbb N}\right)\right]_{u_1v_1} \right)_{a,b\in\mathbb N} \right) \right]_{u_0v_0}.(localProduct(Φ,Ψ,X))(u0​,u1​),(v0​,v1​)​=[Φ(([Ψ((X(a,c),(b,d)​)c,d∈N​)]u1​v1​​)a,b∈N​)]u0​v0​​.

The joining operation places the first occupation before the second; the first and remaining occupation operations extract those respective coordinates. No linearity, positivity, trace preservation, density, Gaussianity, or summability hypothesis is imposed on these inputs.

QAlgorithms.Papers.Independent.Gap48.P47.localSymplecticMatrix

For any two real 2×22\times22×2 matrices S1,S2S_1,S_2S1​,S2​, indexed by the one-mode quadratures (0,0),(0,1)(0,0),(0,1)(0,0),(0,1), this definition produces the real 4×44\times44×4 matrix

S=(S100S2)S=\begin{pmatrix}S_1&0\\0&S_2\end{pmatrix}S=(S1​0​0S2​​)

in the two-mode 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): entries between different modes are 000, the mode-000 block is S1S_1S1​, and the mode-111 block is S2S_2S2​. No symplecticity condition on either input is assumed.

QAlgorithms.Papers.Independent.Gap48.P47.logCriterionPolynomial

For every four real numbers a,b,c,da,b,c,da,b,c,d, the value of this real-valued function is

(a2−1)+(b2−1)c2d2+2cd−ab(c2+d2).(a^2-1)+(b^2-1)c^2d^2+2cd-ab(c^2+d^2).(a2−1)+(b2−1)c2d2+2cd−ab(c2+d2).

No restrictions are imposed on the four arguments.

QAlgorithms.Papers.Independent.Gap48.P47.restOccupation

For every natural number mmm and every occupation function u:{0,…,m}→Nu:\{0,\ldots,m\}\to\mathbb Nu:{0,…,m}→N, the remaining occupation function has domain {0,…,m−1}\{0,\ldots,m-1\}{0,…,m−1} and value u(k+1)u(k+1)u(k+1) at kkk. When m=0m=0m=0, the result is the unique function on the empty set.

QAlgorithms.Papers.Independent.Gap48.P47.w1

For arbitrary real 4×44\times44×4 matrices Γ,T\Gamma,TΓ,T, indexed by (k,α)∈{0,1}×{0,1}(k,\alpha)\in\{0,1\}\times\{0,1\}(k,α)∈{0,1}×{0,1}, write HklH_{kl}Hkl​ for the 2×22\times22×2 block (H(k,α),(l,β))α,β(H_{(k,\alpha),(l,\beta)})_{\alpha,\beta}(H(k,α),(l,β)​)α,β​. Set

A=det⁡Γ00,B=det⁡Γ11,P=det⁡Γ01,Q=P2−det⁡Γ+ABAB,C=Q+Q2−4P22,D={0,C=0,P/C,C≠0.A=\det\Gamma_{00},\quad B=\det\Gamma_{11},\quad P=\det\Gamma_{01},\quad Q=\frac{P^2-\det\Gamma+AB}{\sqrt{AB}},\quad C=\sqrt{\frac{Q+\sqrt{Q^2-4P^2}}2},\quad D=\begin{cases}0,&C=0,\\P/C,&C\ne0.\end{cases}A=detΓ00​,B=detΓ11​,P=detΓ01​,Q=AB​P2−detΓ+AB​,C=2Q+Q2−4P2​​​,D={0,P/C,​C=0,C=0.​

Then

w1(Γ,T)=∣CD∣det⁡T00A.w_1(\Gamma,T)=\frac{|CD|\sqrt{\det T_{00}}}{\sqrt A}.w1​(Γ,T)=A​∣CD∣detT00​​​.

There are no symmetry, positivity, or physicality assumptions on the matrices. Negative arguments of real square roots give 000, and division by 000 gives 000, including when the final denominator vanishes.

QAlgorithms.Papers.Independent.Gap48.P47.w2

For arbitrary real 4×44\times44×4 matrices Γ,T\Gamma,TΓ,T, indexed by (k,α)∈{0,1}×{0,1}(k,\alpha)\in\{0,1\}\times\{0,1\}(k,α)∈{0,1}×{0,1}, write HklH_{kl}Hkl​ for the 2×22\times22×2 block (H(k,α),(l,β))α,β(H_{(k,\alpha),(l,\beta)})_{\alpha,\beta}(H(k,α),(l,β)​)α,β​. Set

A=det⁡T00,B=det⁡T11,P=det⁡T01,Q=P2−det⁡T+ABAB,C=Q+Q2−4P22,D={0,C=0,P/C,C≠0.A=\det T_{00},\quad B=\det T_{11},\quad P=\det T_{01},\quad Q=\frac{P^2-\det T+AB}{\sqrt{AB}},\quad C=\sqrt{\frac{Q+\sqrt{Q^2-4P^2}}2},\quad D=\begin{cases}0,&C=0,\\P/C,&C\ne0.\end{cases}A=detT00​,B=detT11​,P=detT01​,Q=AB​P2−detT+AB​,C=2Q+Q2−4P2​​​,D={0,P/C,​C=0,C=0.​

Then

w2(Γ,T)=∣CD∣det⁡Γ11B.w_2(\Gamma,T)=\frac{|CD|\sqrt{\det\Gamma_{11}}}{\sqrt B}.w2​(Γ,T)=B​∣CD∣detΓ11​​​.

Neither matrix is required to be symmetric, positive, or physical. Real square roots of negative arguments and quotients with zero denominator have value 000, so a vanishing final denominator is included.

Human review
  • Endorsed by Shuze Chen · Oct 22, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Goku · Oct 22, 2026

    Confirmed by the mission captain (proposal self-audit).

  • 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