Lemma 4.1 (S-lemma) — if N has a positive eigenvalue, xᵀNx ⩾ 0 ⇒ xᵀMx ⩾ 0 iff M − αN ⩾ 0 for some α ⩾ 0
ProvedQuadMatIneq.Stabilization.lemma_4_1linear-matrix-inequalityp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1quadratic-formss-lemma
Let be symmetric matrices and suppose that has at least one positive eigenvalue. Then
This is Yakubovich's S-lemma, recalled in the paper as the vector-valued base case from which the matrix S-lemma (Theorem 4.7) is derived. The positive-eigenvalue hypothesis is the Slater condition: some has .
Formalization Note The index set is an arbitrary finite type; is PosSemidef. The positive-eigenvalue hypothesis uses Mathlib's eigenvalues of the symmetric matrix .
Preamble
import Mathlib import Definitions.Def_QuadMatIneq_Stabilization_QMI open Matrix
Formal statement
namespace QuadMatIneq.Stabilization
/-- Lemma 4.1 (S-lemma), p. 9. -/
theorem lemma_4_1 {k : Type*} [Fintype k] [DecidableEq k]
(M N : Matrix k k ℝ) (hM : M.IsHermitian) (hN : N.IsHermitian)
(hpos : ∃ i, 0 < hN.eigenvalues i) :
(∀ x : k → ℝ, 0 ≤ x ⬝ᵥ (N *ᵥ x) → 0 ≤ x ⬝ᵥ (M *ᵥ x)) ↔
∃ α : ℝ, 0 ≤ α ∧ (M - α • N).PosSemidef := by sorry
end QuadMatIneq.Stabilization
Source
van Waarde–Camlibel–Eising–Trentelman, arXiv:2203.12959v3, Lemma 4.1, p. 9
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.