Eq. (10): first-layer covariance
ProvedLesHouchesWidth.covariance_first_layerFor a network with Gaussian initialization and , any inputs and any first-layer neuron ,
This is the base case of the NNGP kernel recursion.
Formalization Note In Lecture 1 this quantity is written ; the series uses the layer indexing of Lectures 4–5.
import Mathlib import Definitions.Def_LesHouchesWidth_GaussianMLP
namespace LesHouchesWidth
open MeasureTheory ProbabilityTheory
theorem covariance_first_layer (σb σw : ℝ) (φ : ℝ → ℝ) (n : ℕ → ℕ) (L : ℕ)
(x x' : Fin (n 0) → ℝ) (i : Fin (n 1)) :
∫ θ, mlpZ (σb ^ 2) (σw ^ 2) φ θ x 1 i * mlpZ (σb ^ 2) (σw ^ 2) φ θ x' 1 i
∂(stdGaussianParams n L) =
nngpKernel (σb ^ 2) (σw ^ 2) φ 1 x x' := by sorry
end LesHouchesWidthRead-back
What the Lean code literally says, in plain math · Aristotle (Harmonic) - non-blind, same agent as drafter
Non-blind read-back. This read-back was written by the same agent (Aristotle, by Harmonic) that drafted the Lean statement, with full knowledge of the source and of the intended meaning. It is not independent testimony and should not be treated as a blind audit; an independent read-back is still recommended before launch.
Let be arbitrary, let be arbitrary (it plays no role), and let and be arbitrary. Let and let . With parameters distributed as i.i.d. standard Gaussians, biases and first-layer weights , the theorem asserts
where . There are no hypotheses. If , the right side is , since division by zero gives .
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.