(21), App. B, p. 2325 — 0 ≤ F_β(w) − max_j(w_j − y_j) ≤ β⁻¹ log p = φ⁻¹ for β = φ log p
ProvedHighDimCLT.Hyperrect.eq_21high-dimensional-cltlog-sum-expp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1smooth-max
Let , and , and put . Let be the smooth maximum. Then for every ,
The smooth maximum approximates the maximum coordinate of within , which is what lets the proof of Lemma 5.1 replace the indicator of by a smooth function of .
Formalization Note The maximum is the finite supremum over ; with the logarithm is positive, so .
Preamble
import Mathlib import Definitions.Def_HighDimCLT_Hyperrect_Setting open MeasureTheory ProbabilityTheory
Formal statement
namespace HighDimCLT.Hyperrect
theorem eq_21 (p : ℕ) (hp : 3 ≤ p) (φ : ℝ) (hφ : 1 ≤ φ) (y w : E p) :
0 ≤ Fβ (φ * Real.log p) y w - (⨆ j : Fin p, (w j - y j)) ∧
Fβ (φ * Real.log p) y w - (⨆ j : Fin p, (w j - y j)) ≤ (φ * Real.log p)⁻¹ * Real.log p ∧
(φ * Real.log p)⁻¹ * Real.log p = φ⁻¹ := by sorry
end HighDimCLT.Hyperrect
Source
Chernozhukov, Chetverikov and Kato, Central limit theorems and bootstrap in high dimensions, Ann. Probab. 45 (2017), p. 2325, App. B, display (21)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.