Lemma 9, p. 28 — the level-k Lasserre value is a relaxation: νk(H) ≥ λmax(H)
ProvedQMCProduct.HalfApprox.lemma_9Let be a Hermitian operator on qubits and . Let be an optimal solution of level of the quantum Lasserre hierarchy, i.e. a -positive pseudo-density maximizing , so that . Then for every density matrix (positive semidefinite, trace one)
Since , this is : the hierarchy is a relaxation of the largest eigenvalue.
This is the inequality in equation (1) and the second inequality in the proof of Proposition 1.
Formalization Note is expressed through comparison with every density matrix, and through an optimal pseudo-density (equation (1) writes as a maximum). The paper proves the lemma in the equivalent vector-program form (Definition 17); the pseudo-density form is equation (1). Real parts of traces are taken; they are the full traces for Hermitian arguments.
import Definitions.Def_QMCProduct_HalfApprox_Basic set_option autoImplicit false
namespace QMCProduct.HalfApprox
/-- Lemma 9, p. 28: the level-k value `νk(H) = Tr[H ρt]` of an optimal k-positive
pseudo-density is at least `Tr[H σ]` for every density matrix `σ`, i.e. `νk(H) ≥ λmax(H)`. -/
theorem lemma_9 {n : ℕ} (k : ℕ) (H : Op n) (hH : H.IsHermitian)
(ρt : Op n) (hopt : IsOptimalLk k H ρt) :
∀ σ : Op n, QAlgorithms.IsDensityMatrix σ →
(H * σ).trace.re ≤ (H * ρt).trace.re := by sorry
end QMCProduct.HalfApprox
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.