Theorem 4.1.1 — Matrix Gaussian and Rademacher series
ProvedTroppMatrixConcentration.ch4_matrix_seriesLet be a finite family of fixed complex matrices, with . On an arbitrary probability space let be independent measurable real random variables, either all standard normal or all Rademacher. Put and define the actual second-moment statistic . Then
All norms are Euclidean operator norms. At the tail right side is for and zero for . The finite family may be empty. Both scalar distributions are included in the same theorem via a disjunction on the whole family's laws.
import Definitions.Def_TroppMatrixConcentration_ch4_scalar_laws import Definitions.Def_TroppMatrixConcentration_dilation open MeasureTheory ProbabilityTheory open scoped Matrix.Norms.L2Operator
namespace TroppMatrixConcentration
theorem ch4_matrix_series {Ω : Type*} [MeasurableSpace Ω]
(μ : Measure Ω) [IsProbabilityMeasure μ] {m n N : ℕ} [NeZero m] [NeZero n]
(B : Fin N → Matrix (Fin m) (Fin n) ℂ)
(g : Fin N → Ω → ℝ) (hMeas : ∀ k, Measurable (g k))
(hIndep : iIndepFun g μ)
(hLaw : (∀ k, standardGaussianLaw μ (g k)) ∨ (∀ k, rademacherLaw μ (g k))) :
let Z := fun ω => ∑ k, g k ω • B k
let v := rectSecondMoment μ Z
v = max (spectralNorm (∑ k, B k * (B k).conjTranspose))
(spectralNorm (∑ k, (B k).conjTranspose * B k)) ∧
(∫ ω, spectralNorm (Z ω) ∂μ) ≤ Real.sqrt (2 * v * Real.log (m + n)) ∧
∀ t : ℝ, 0 ≤ t → (μ {ω | t ≤ spectralNorm (Z ω)}).toReal ≤
gaussianSeriesTail (m + n) v t := by sorry
end TroppMatrixConcentration
Read-back
What the Lean code literally says, in plain math · gpt-6
For every measurable space , every probability measure on it, all positive natural numbers , every natural number (including ), every family of complex matrices , and every family of measurable real-valued functions that are jointly independent with respect to , suppose either that every pushforward measure is the real Gaussian distribution with mean and variance , or that every equals ; this is a disjunction of two uniform choices for the whole family. Write for the conjugate transpose of a matrix , and for its operator norm between the complex Euclidean spaces of its column and row dimensions. Define , with multiplication by the real scalar acting on the complex matrix, and define , where these are Bochner integrals (using the convention that such an integral is zero when its integrand is not integrable). Then all three conclusions hold: ; , with the natural logarithm and nonnegative real square root; and, for every real , the real value of is at most if , at most if and , and at most if and . The real value of the event measure is its extended-nonnegative-real-to-real conversion, which is the ordinary finite value here because is a probability measure. The tail event uses the non-strict inequality, so at it is all of and the bound is in either variance branch. If , all family hypotheses are vacuous, all sums are empty, and , the expectation bound is , and the stated zero-variance tail bounds apply; zero variance is also included for nonempty families. The assumptions exclude and .
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.