Every passive Gaussian state is for a Hermitian
OpenBosonicLearn.passive_gaussian_fugacityThe surjectivity half of the bijection of Lemma 2 of the paper.
Let be a passive Gaussian state on modes: a Gaussian state whose density operator is block diagonal in the total photon number sectors. Then there is a positive semidefinite matrix with positive definite such that
Status of the other clauses of Lemma 2. That is a passive Gaussian state (with covariance in real form) is BosonicLearn.fugacity_isGaussianState; the formula , hence uniqueness of , is BosonicLearn.gammaFock_one_photon; the mean photon number is BosonicLearn.fugacity_photon_number. This statement is the only part of BosonicLearn.lemma_2 that is not yet proved.
Possible route. Block diagonality makes the characteristic function invariant under , which forces mean and a covariance with Hermitian; put ; positivity and normalization of give ; conclude with the fact that a state is determined by its characteristic function (or by its Husimi function, BosonicLearn.husimi_fourier).
Formalization Note IsPassive ρ is the mission predicate (Gaussian for some mean and covariance, and vanishing Fock matrix elements between different total photon numbers); the conclusion is equality of coefficient arrays with fugacityCoeff R.
import Mathlib import Definitions.Def_BosonicLearn_fugacityCoeff import Definitions.Def_BosonicLearn_IsPassive open QAlgorithms.Papers.Independent.Dependency50.P33 (State) open QAlgorithms.Papers.Independent.Claimed50.P47 (Occupation) open scoped Matrix ComplexOrder
namespace BosonicLearn
/-- **Every passive Gaussian state has a fugacity matrix.**
Chen, Anna Mele, Anna Mele, Preskill, *When are bosonic Gaussian states classical to learn?*, arXiv:2609.26705, Lemma 2, p. 20: "There is a bijection between n-mode passive Gaussian states ρ and n-by-n Hermitian matrices 0 ≤ R < I.
Given R, which we call the fugacity matrix, the corresponding passive Gaussian state is ρ = det(I − R)Γ(R)."
If `ρ` is a passive Gaussian state (Gaussian, and block diagonal in the total photon number), then there is a
positive semidefinite `R` with `1 − R` positive definite such that `ρ = det(1 − R)Γ(R)`.
**Formalization Note.** This is the surjectivity half of the bijection of `lemma_2`; the other clauses of Lemma 2
(`det(1 − R)Γ(R)` is a passive Gaussian state, the formula `R = P_1ρP_1/Tr(P_0ρ)` and hence uniqueness of `R`, and
the mean photon number) are proved in `fugacity_isGaussianState`, `gammaFock_one_photon` and
`fugacity_photon_number`. -/
theorem passive_gaussian_fugacity {n : ℕ} (ρ : State (Occupation (Fin n))) (hρ : IsPassive ρ) :
∃ R : Matrix (Fin n) (Fin n) ℂ, R.PosSemidef ∧ (1 - R).PosDef ∧
ρ.coefficients = fugacityCoeff R := by sorry
end BosonicLearn