Gaussian normalization (eq. 1.30)
ProvedDeepLearningTheory.gaussian_integral_normalizationLet be a symmetric positive-definite real matrix. Then
This fixes the normalization of the multivariable Gaussian distribution (1.31), so that its density integrates to one.
import Mathlib import Definitions.Def_DLT_GaussianIntegrals open MeasureTheory
namespace DeepLearningTheory
theorem gaussian_integral_normalization {N : ℕ} (K : Matrix (Fin N) (Fin N) ℝ)
(hK : K.PosDef) :
∫ z : Fin N → ℝ, Real.exp (-(1 / 2) * gaussQuadForm K z)
= Real.sqrt ((2 * Real.pi) ^ N * K.det) := by sorry
end DeepLearningTheoryRead-back
What the Lean code literally says, in plain math · Aristotle (Harmonic) - non-blind, same agent as the drafter
Non-blind read-back. This read-back was written by the same agent that drafted the Lean statement, with full knowledge of the source text and of the intended meaning. It is not independent testimony and must not be mistaken for a blind audit; reviewers should compare the Lean code against the source themselves.
For every and every real matrix that is positive definite (in Mathlib's sense: symmetric and for all nonzero ), the Lebesgue integral over of equals . The integral is a Bochner integral (it would be if the integrand were not integrable). The case is included: the integral over the one-point space is and the right side is .
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.