Proof of Proposition 5, p. 26 — for losses in [0, 1], each 𝒦ᵢ(R) is nondecreasing in R on R ≥ 0
ProvedRiskControl.WSR.K_monotonebetting-confidence-boundp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1risk-controlupper-confidence-bound
Let , , and let be a sample with every . Then for every index the capital of Proposition 5 (see the definition RiskControl.WSR.Process) is a nondecreasing function of on :
This monotonicity is what turns the bound at a single value of (the true mean) into a statement about the infimum .
Formalization Note The paper says "increasing in almost surely"; the statement is the nondecreasing version (for , or when , is constant in ), and "almost surely" is the condition that the observed losses lie in . The range is the range of the infimum defining ; there every factor is nonnegative because and .
Preamble
import Mathlib import Definitions.Def_RiskControl_WSR_Process
Formal statement
namespace RiskControl.WSR
/-- Proof of Proposition 5, p. 26 (arXiv:2101.02703v3): for a sample of losses in `[0, 1]`,
every `𝒦_i(R; λ)` is nondecreasing in `R` on `R ≥ 0`. -/
theorem K_monotone (n : ℕ) (δ : ℝ) (ω : Fin n → ℝ) (hω : ∀ j, ω j ∈ Set.Icc (0 : ℝ) 1)
(i : ℕ) :
MonotoneOn (fun R => K n δ ω i R) (Set.Ici 0) := by sorry
end RiskControl.WSR
Source
Bates, Angelopoulos, Lei, Malik & Jordan, arXiv:2101.02703v3, Proof of Proposition 5, p. 26, sentence after the second display
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.