Lemma 5.4.1 — Matrix Chernoff mgf and cgf bounds
ProvedTroppMatrixConcentration.ch5_chernoff_mgf_cgfLet be a measurable random complex matrix on a probability space, , Hermitian almost surely. Assume and almost surely, for a common . For every real , put when , and when . Then
All expectations are Bochner integrals; boundedness supplies integrability. Both signs of , as well as , are included. At the matrix vanishes almost surely and the displayed relations have their deterministic interpretation.
import Definitions.Def_TroppMatrixConcentration_ch5_chernoff_functions open MeasureTheory ProbabilityTheory open scoped Matrix.Norms.L2Operator
namespace TroppMatrixConcentration
theorem ch5_chernoff_mgf_cgf {Ω : Type*} [MeasurableSpace Ω]
(μ : Measure Ω) [IsProbabilityMeasure μ] {d : ℕ} [NeZero d]
(X : Ω → Matrix (Fin d) (Fin d) ℂ) (L : ℝ) (hL : 0 ≤ L)
(hMeas : Measurable X) (hHerm : ∀ᵐ ω ∂μ, (X ω).IsHermitian)
(hBound : ∀ᵐ ω ∂μ, 0 ≤ lambdaMin (X ω) ∧ lambdaMax (X ω) ≤ L)
(θ : ℝ) :
loewnerLE (∫ ω, matrixExp (θ • X ω) ∂μ)
(matrixExp (chernoffCgfCoefficient L θ • (∫ ω, X ω ∂μ))) ∧
loewnerLE (matrixLog (∫ ω, matrixExp (θ • X ω) ∂μ))
(chernoffCgfCoefficient L θ • (∫ ω, X ω ∂μ)) := 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, every positive integer (so is excluded), every Borel-measurable function with rows and columns indexed by , and every real , suppose that is Hermitian for -almost every and that and for -almost every , where denotes the spectrum of as an element of the real matrix algebra (on Hermitian matrices these extrema are the smallest and largest eigenvalues). Then, for every real , including zero and negative values, set , , and if , and otherwise; both and are positive semidefinite, meaning that each is Hermitian and has nonnegative quadratic form on every vector in . Here the integrals are matrix-valued Bochner integrals, is the normed-algebra exponential, scalar multiplication is by real scalars, and is the continuous functional calculus applied to the real logarithm and (in particular, on positive definite Hermitian matrices it applies the usual logarithm to each eigenvalue). The assumptions impose Hermitianity and spectral bounds only almost everywhere, permit , and impose no sign restriction on ; when they force almost everywhere, so the two positive semidefinite differences are zero, and when the differences are likewise zero. The probability-measure assumption requires , excluding an empty underlying space.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.