Proposition 1, p. 607 — under (1)–(3),
ProvedStoneRegression.Criterion.proposition_1lp-convergencenonparametric-regressionp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1weight-functions
Let be i.i.d. -valued with law and let be a sequence of Borel weights. Suppose that conditions (1)–(3) hold: there is with for every nonnegative Borel and ; there is with for ; and in probability for every .
Let and let be a Borel function on with . Then
This is the basic approximation step of the paper: weights concentrated near reproduce in a weighted sense. It feeds Propositions 2 and 5 and thereby the sufficiency half of Theorem 1.
Formalization Note. The expectation is a lower Lebesgue integral in over the product law of ; is the corresponding lintegral against . Joint Borel measurability of the weights is the series' standing assumption.
Preamble
import Mathlib import Definitions.Def_StoneRegression_Criterion_Setting open MeasureTheory ProbabilityTheory Filter Topology open scoped ENNReal NNReal BoundedContinuousFunction
Formal statement
namespace StoneRegression.Criterion
/-- Proposition 1 (Stone 1977, p. 607). Under (1)–(3), for `r ≥ 1` and a Borel `f` with `E|f(X)|ʳ < ∞`,
`E ∑ᵢ |W_{ni}(X)| |f(Xᵢ) − f(X)|ʳ → 0`. Sample indices are 0-based. -/
theorem proposition_1 {d : ℕ} (μ : Measure (EuclideanSpace ℝ (Fin d))) [IsProbabilityMeasure μ]
(W : WeightSeq d) (hW : MeasurableWeights W)
(h1 : ∃ C : ℝ≥0, 1 ≤ C ∧ Cond1 μ W C) (h2 : ∃ D : ℝ, 1 ≤ D ∧ Cond2 μ W D) (h3 : Cond3 μ W)
(r : ℝ) (hr : 1 ≤ r) (f : EuclideanSpace ℝ (Fin d) → ℝ) (hf : Measurable f)
(hfr : ∫⁻ x, ‖f x‖ₑ ^ r ∂μ < ∞) :
Tendsto (fun n => ∫⁻ ω, ∑ i : Fin n,
‖wAt W n ω i‖ₑ * ‖f (ω (i.val + 1)) - f (ω 0)‖ₑ ^ r ∂(xLaw μ)) atTop (𝓝 0) := by sorry
end StoneRegression.Criterion
Source
Stone (1977), Ann. Statist. 5, Proposition 1, p. 607; proof pp. 607–608
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.