Equation 2.2.5 — Additivity of matrix variance
ProvedTroppMatrixConcentration.variance_additivityLet be independent measurable Hermitian random complex matrices on a probability space, with and finite second moments. Write . Then
No zero-mean assumption is imposed. The identity is matrix-valued and keeps the summation inside subsequent spectral norms.
import Definitions.Def_TroppMatrixConcentration_probability import Definitions.Def_TroppMatrixConcentration_dilation import Mathlib.Analysis.Convex.Function open MeasureTheory ProbabilityTheory open scoped Matrix.Norms.L2Operator
namespace TroppMatrixConcentration
theorem variance_additivity {Ω : Type*} [MeasurableSpace Ω]
(μ : Measure Ω) [IsProbabilityMeasure μ] {d N : ℕ} [NeZero d]
(X : Fin N → Ω → Matrix (Fin d) (Fin d) ℂ)
(hMeas : ∀ k, Measurable (X k))
(hHerm : ∀ k, ∀ᵐ ω ∂μ, (X k ω).IsHermitian)
(hL2 : ∀ k, MemLp (X k) 2 μ) (hIndep : iIndepFun X μ) :
(∫ ω, ((∑ k, X k ω) - ∫ ω', ∑ k, X k ω' ∂μ) ^ 2 ∂μ) =
∑ k, ∫ ω, (X k ω - ∫ ω', X k ω' ∂μ) ^ 2 ∂μ := by sorry
end TroppMatrixConcentration
Read-back
What the Lean code literally says, in plain math · gpt-6
Let be a measurable space, let be a probability measure on it, let be a positive integer, and let be any nonnegative integer. For each , let be measurable, Hermitian -almost everywhere (that is, almost everywhere), and square-integrable with respect to the operator norm induced by the Euclidean norm, so that ; assume that the family is jointly independent under . Define , , and , where all matrix-valued integrals are Bochner integrals. Then as an equality of complex matrices, with each square denoting ordinary matrix multiplication of the centered matrix by itself. The means need not vanish. The case is included: the hypotheses on the empty family are vacuous and both sides are the zero matrix; the case is excluded.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.