A positive semidefinite array with trace one is a density operator (the nuclear expansion is automatic)
ProvedBosonicLearn.isDensity_of_posSemidefbosonic-gaussian-statesquantum-informationquantum-learning
Positivity and normalization imply trace class.
Let be a countable index set and a complex array such that every finite quadratic form is real and non-negative, and . Then is a density operator on in the sense of the published predicate IsDensity; in particular it has a nuclear expansion with and (one may take ).
Role. IsDensity asks for a nuclear expansion in addition to positivity and normalization; this lemma removes that burden when states are constructed by mixing (BosonicLearn.state_gaussian_mixture).
Formalization Note IsDensity is Dependency50.P33.IsDensity; the index type is any Countable type.
Preamble
import Mathlib import Definitions.Def_QAlgorithms_Papers_Independent_Dependency50_P33 open QAlgorithms.Papers.Independent.Dependency50.P33 (IsDensity)
Formal statement
namespace BosonicLearn
/-- **A positive semidefinite array with trace one is a density operator.**
Chen, Anna Mele, Anna Mele, Preskill, *When are bosonic Gaussian states classical to learn?*, arXiv:2609.26705, §1.1, pp. 4–5 (states of the n-mode Fock space as density operators: positive, trace one).
Let `I` be a countable index set and `X` an `I × I` complex array all of whose finite quadratic forms are real and
non-negative, with summable diagonal of sum `1`. Then `X` is a density operator in the sense of `IsDensity`: in
particular it is trace class (it has a nuclear expansion `X_ij = ∑ₖ lₖ(i) \overline{rₖ(j)}`, `∑ₖ ‖lₖ‖‖rₖ‖ < ∞`).
**Formalization Note.** `IsDensity` asks for a nuclear expansion in addition to positivity and normalization; this
lemma says the expansion is automatic (take `lₖ = rₖ = X^{1/2} eₖ`). It makes it possible to build states by
mixing. -/
theorem isDensity_of_posSemidef {I : Type} [Countable I] (X : Matrix I I ℂ)
(hpsd : ∀ (s : Finset I) (z : I → ℂ),
(∑ i ∈ s, ∑ j ∈ s, star (z i) * X i j * z j).im = 0 ∧
0 ≤ (∑ i ∈ s, ∑ j ∈ s, star (z i) * X i j * z j).re)
(hsum : Summable (fun i => (X i i).re)) (htr : ∑' i, (X i i).re = 1) : IsDensity X := by sorry
end BosonicLearn
Source
Chen, Anna Mele, Anna Mele, Preskill, When are bosonic Gaussian states classical to learn?, arXiv:2609.26705, §1.1, Eqs. (2) and (4), pp. 4–5 (Gaussian states, displacement operators, characteristic function) (states as density operators)