(relative Frobenius bound)
ProvedBosonicLearn.loewner_trace_boundbosonic-gaussian-statesquantum-informationquantum-learning
A Loewner sandwich bounds the relative Frobenius norm.
Let be a real positive definite matrix, , and a real symmetric matrix with . Then
Role. In the proof of Theorem 8, , , and ; this is Eq. (215), the bound on the quantity of Theorem 7.
Formalization Note The squared Frobenius norm is written as the trace , the form in which BosonicLearn.theorem_7 states . Loewner order is Matrix.PosSemidef of the difference; is Fintype.card ι. Pure linear algebra.
Preamble
import Mathlib open scoped Matrix
Formal statement
namespace BosonicLearn
/-- **Relative Frobenius bound from a Loewner sandwich.**
Chen, Anna Mele, Anna Mele, Preskill, *When are bosonic Gaussian states classical to learn?*, arXiv:2609.26705, §7.2, proof of Theorem 8, Eq. (215), p. 62: "0 ⪯ Σ^{−1/2}(Σ̃ − Σ)Σ^{−1/2} ⪯ (2α_ν ζ/(1 − ζ)) I. … Since for a
positive semidefinite matrix ‖R‖_F² is the sum of the squares of its eigenvalues, we obtain
‖Σ^{−1/2}(Σ̃ − Σ)Σ^{−1/2}‖_F ≤ (2α_ν ζ/(1 − ζ)) √(2n)."
Let `S` be a real positive definite `d × d` matrix and `D` a real symmetric matrix with `0 ⪯ D ⪯ c S`, `c ≥ 0`.
Then `‖S^{−1/2} D S^{−1/2}‖_F² = Tr(S⁻¹ D S⁻¹ D) ≤ c² d`.
**Formalization Note.** The squared Frobenius norm is written as `Tr(S⁻¹ D S⁻¹ D)`, the form in which
`theorem_7` states `Δ_rel²`. Loewner order is positive semidefiniteness of the difference. -/
theorem loewner_trace_bound {ι : Type} [Fintype ι] [DecidableEq ι] (S D : Matrix ι ι ℝ) (hS : S.PosDef)
(hD : D.PosSemidef) (c : ℝ) (hc : 0 ≤ c) (hDS : (c • S - D).PosSemidef) :
(S⁻¹ * D * S⁻¹ * D).trace ≤ c ^ 2 * Fintype.card ι := by sorry
end BosonicLearn
Source
Chen, Anna Mele, Anna Mele, Preskill, When are bosonic Gaussian states classical to learn?, arXiv:2609.26705, Eq. (215), p. 62 (proof of Theorem 8)