Lemma 7.5.1 — Intrinsic dimension trace inequality
ProvedTroppMatrixConcentration.ch7_intrinsic_dimensionFor every real function convex on with , and every positive semidefinite complex matrix , ,
The function of the matrix is defined by spectral functional calculus. Zero matrices are included using the totalized convention , so both sides are zero. No nonnegativity assumption on is imposed.
import Definitions.Def_TroppMatrixConcentration_ch7_intrinsic open MeasureTheory ProbabilityTheory open scoped Matrix.Norms.L2Operator ComplexOrder
namespace TroppMatrixConcentration
theorem ch7_intrinsic_dimension {d : ℕ} [NeZero d]
(φ : ℝ → ℝ) (hConvex : ConvexOn ℝ (Set.Ici 0) φ) (hZero : φ 0 = 0)
(A : Matrix (Fin d) (Fin d) ℂ) (hPSD : A.PosSemidef) :
traceFunction φ A ≤ intrinsicDimension A * φ (spectralNorm A) := by sorry
end TroppMatrixConcentration
Read-back
What the Lean code literally says, in plain math · gpt-6
For every natural number , every function that is convex on and satisfies , and every complex matrix , with rows and columns indexed by , assume that is positive semidefinite, meaning that and is real and nonnegative for every . Then , where is the operator norm induced by the Euclidean norm and is the real continuous functional calculus of : if is a unitary diagonalization, then . Equivalently, the asserted inequality is , with eigenvalues counted with multiplicity. Division is total, with division by zero defined to be zero; in particular, is allowed and the inequality then reads . The case is excluded, but singular nonzero matrices are allowed; no nonnegativity of or restriction on its values at negative arguments is assumed.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.