Deep linear network: four-point recursion (eq. 3.20)
ProvedDeepLearningTheory.linearNet_four_point_recursionFor a zero-bias deep linear network at initialization and a single input , for every layer and neurons of layer ,
Formalization Note The identity is stated for all ; for the right side involves the deterministic input and reduces to eq. (3.18).
import Mathlib import Definitions.Def_DLT_DeepLinearNetwork open MeasureTheory ProbabilityTheory open scoped NNReal
namespace DeepLearningTheory
theorem linearNet_four_point_recursion {Ω : Type*} [MeasurableSpace Ω] (P : Measure Ω)
[IsProbabilityMeasure P] (n : ℕ → ℕ) (CW : ℝ≥0) (W : Ω → ℕ → ℕ → ℕ → ℝ)
(hW : IsLinearNetInit P n CW W) (x : ℕ → ℝ) (ℓ i₁ i₂ i₃ i₄ : ℕ)
(hi₁ : i₁ < n (ℓ + 1)) (hi₂ : i₂ < n (ℓ + 1)) (hi₃ : i₃ < n (ℓ + 1))
(hi₄ : i₄ < n (ℓ + 1)) :
∫ ω, linearPreact n (W ω) x (ℓ + 1) i₁ * linearPreact n (W ω) x (ℓ + 1) i₂ *
linearPreact n (W ω) x (ℓ + 1) i₃ * linearPreact n (W ω) x (ℓ + 1) i₄ ∂P
= (CW : ℝ) ^ 2 * wickDelta4 i₁ i₂ i₃ i₄ * (1 / (n ℓ : ℝ) ^ 2) *
∑ j ∈ Finset.range (n ℓ), ∑ k ∈ Finset.range (n ℓ),
∫ ω, linearPreact n (W ω) x ℓ j * linearPreact n (W ω) x ℓ j *
linearPreact n (W ω) x ℓ k * linearPreact n (W ω) x ℓ k ∂P := 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.
Let be a probability space, widths, and satisfying IsLinearNetInit. For every input , every and all :
where is if , is the three-term Kronecker-delta tensor, and (so for the inner integrals are of constants).
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.