§4.1.2, p. 44 — tail integration at a positive split point
ProvedIntroMatrixConc.GaussianSeries.tail_integrationmatrix-concentrationp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-bookp2o-v1random-matrices
Let be a measurable complex random matrix on a probability space, with positive variance statistic . Suppose that for every ,
For every split point , the squared norm is integrable and
This is the quantitative tail-integration step used for the upper half of (4.1.7).
Formalization Note The page applies the step to a Gaussian series. The statement isolates the step for any measurable matrix satisfying exactly its stated tail bound; positive variance makes the displayed denominator meaningful.
Preamble
import Mathlib import Definitions.Def_TroppMatrixConcentration_ch4_scalar_laws open MeasureTheory ProbabilityTheory TroppMatrixConcentration open scoped Matrix.Norms.L2Operator noncomputable section
Formal statement
namespace IntroMatrixConc.GaussianSeries
/-- Tropp §4.1.2, p. 44: integrate the displayed tail bound at a split point. -/
theorem tail_integration {Ω : Type*} [MeasurableSpace Ω]
(μ : Measure Ω) [IsProbabilityMeasure μ]
{d₁ d₂ : ℕ} [NeZero d₁] [NeZero d₂]
(Z : Ω → Matrix (Fin d₁) (Fin d₂) ℂ)
(hMeas : Measurable Z)
(hv : 0 < rectSecondMoment μ Z)
(hTail : ∀ t : ℝ, 0 ≤ t →
(μ {ω | t ≤ spectralNorm (Z ω)}).toReal ≤
(d₁ + d₂ : ℕ) * Real.exp (-(t ^ 2) / (2 * rectSecondMoment μ Z)))
(a : ℝ) (ha : 0 < a) :
Integrable (fun ω => spectralNorm (Z ω) ^ 2) μ ∧
(∫ ω, spectralNorm (Z ω) ^ 2 ∂μ) ≤
a ^ 2 + 2 * rectSecondMoment μ Z * (d₁ + d₂ : ℕ) *
Real.exp (-(a ^ 2) / (2 * rectSecondMoment μ Z)) := by sorry
end IntroMatrixConc.GaussianSeries
Source
Tropp, An Introduction to Matrix Concentration Inequalities, arXiv:1501.01571v1, §4.1.2, p. 44, tail-integration display below (4.1.7)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.