implies every eigenvalue of has (symplectic eigenvalues )
ProvedBosonicLearn.symplectic_spectrum_lowerbosonic-gaussian-statesquantum-informationquantum-learning
Ordinary-to-symplectic eigenvalue comparison.
Let be a real symmetric matrix with for some , and . Then every complex eigenvalue of satisfies .
For a covariance matrix the spectrum of is with the symplectic eigenvalues, so implies that all symplectic eigenvalues are at least .
Role. This verifies the hypothesis of Theorem 7 (symplectic eigenvalues of at least ) for states of the promise class in the proof of Theorem 8.
Formalization Note The conclusion has the form of the hypothesis hsympl of BosonicLearn.theorem_7: for every z ∈ spectrum ℂ ((Ω * S).map (↑)), c ≤ ‖z‖, with Ω = Matrix.fromBlocks 0 1 (-1) 0 in the grouped quadrature order. The hypothesis is Matrix.PosSemidef (S − c • 1). Pure linear algebra.
Preamble
import Mathlib open scoped Matrix
Formal statement
namespace BosonicLearn
/-- **Ordinary-to-symplectic eigenvalue comparison.**
Chen, Anna Mele, Anna Mele, Preskill, *When are bosonic Gaussian states classical to learn?*, arXiv:2609.26705, §7.2, proof of Theorem 8, p. 62: "The same promise also implies that every symplectic eigenvalue of Σ is at
least 1/2 + ν, by the ordinary-to-symplectic eigenvalue comparison proved above."
Let `S` be a real symmetric `2n × 2n` matrix with `S ⪰ c I`, `c ≥ 0`, and `Ω = [[0, I], [−I, 0]]`. Then every
complex eigenvalue `z` of `ΩS` satisfies `|z| ≥ c`. For a covariance matrix the spectrum of `ΩΣ` is
`{± i ν_j}` with `ν_j` the symplectic eigenvalues, so all symplectic eigenvalues are at least `c`.
**Formalization Note.** This is the hypothesis `hsympl` of `theorem_7` (with `νm = c`), in the same form:
`z ∈ spectrum ℂ ((Ω * S).map (↑))`. -/
theorem symplectic_spectrum_lower {n : ℕ} (S : Matrix (Fin n ⊕ Fin n) (Fin n ⊕ Fin n) ℝ) (c : ℝ)
(hc : 0 ≤ c) (hS : (S - c • (1 : Matrix (Fin n ⊕ Fin n) (Fin n ⊕ Fin n) ℝ)).PosSemidef) :
∀ z ∈ spectrum ℂ ((Matrix.fromBlocks 0 1 (-1) 0 * S).map (fun x : ℝ => (x : ℂ))),
c ≤ ‖z‖ := by sorry
end BosonicLearn
Source
Chen, Anna Mele, Anna Mele, Preskill, When are bosonic Gaussian states classical to learn?, arXiv:2609.26705, p. 62 ("the ordinary-to-symplectic eigenvalue comparison", proof of Theorem 8); Theorem 7, p. 55