Singular Schur-complement positivity with both kernel conditions
OpenQAlgorithms.Papers.Independent.Intakenext100.P68.singularSchurComplementLet and be symmetric positive-semidefinite matrices, and let be arbitrary. Here count matrix coordinates, not bosonic modes. Put . The following are equivalent:
The inverses are spectral Moore–Penrose inverses, including zero eigenvalues. The kernel conditions are essential in the singular case. Lean records the two equivalences with as their common left-hand side.
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P68
noncomputable section
open MeasureTheory Filter
open scoped BigOperators Topology ComplexOrder
namespace QAlgorithms.Papers.Independent.Intakenext100.P68
attribute [local instance] Classical.propDecidable Classical.decEq
open QAlgorithms.Papers.Independent.Dependency50.P33
(State PureState IsSeparable IsBarycenter productCoeff)
open QAlgorithms.Papers.Independent.Dependency50.P10 (complexify)
open QAlgorithms.Papers.Independent.Frontier30.P08 (QuadratureIndex)
open QAlgorithms.Papers.Independent.Claimed50.P19 (Cov)
open QAlgorithms.Papers.Independent.Claimed50.P47 (Occupation)
/-- PDF p.4, Appendix Lemma 1: both singular Schur-complement equivalences.
Here n,m count matrix coordinates, not oscillator modes. Both kernel clauses
are retained, with no invertibility or positive-mode-count restriction. -/
theorem singularSchurComplement {n m : ℕ}
(A : Matrix (Fin n) (Fin n) ℝ) (B : Matrix (Fin m) (Fin m) ℝ)
(C : Matrix (Fin n) (Fin m) ℝ) (hA : A.PosSemidef) (hB : B.PosSemidef) :
((Matrix.fromBlocks A C C.transpose B).PosSemidef ↔
(∀ v : Fin m → ℝ, B.mulVec v = 0 → C.mulVec v = 0) ∧
(A - C * hermitianInverse B * C.transpose).PosSemidef) ∧
((Matrix.fromBlocks A C C.transpose B).PosSemidef ↔
(∀ v : Fin n → ℝ, A.mulVec v = 0 → C.transpose.mulVec v = 0) ∧
(B - C.transpose * hermitianInverse A * C).PosSemidef) := by sorry
end QAlgorithms.Papers.Independent.Intakenext100.P68
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
For any natural numbers , any real positive semidefinite matrix , any real positive semidefinite matrix , and any real matrix , both equivalences below hold. Here means the spectral functional calculus for a Hermitian matrix , taking , and means zero for a non-Hermitian matrix:
Neither nor is assumed invertible. Zero dimensions are included, with empty matrices and the corresponding zero-dimensional vector spaces.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.