Lemma 3, p. 36 — for π ∈ Π_AP and k > q: λ_H^{k−q} π_q ≤ π_k ≤ λ^{k−q} π_q; if λ_H > 0, every π_k ∈ (0, 1)
ProvedNeedPersuasion.Threshold.lemma_3detailed-balancep2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1queueing
Work in the model of Anunrojwong, Iyer and Manshadi under its standing assumptions, and let be an admission policy, so that is a probability distribution on with for every .
- For all in ,
- If , then for every .
The two-sided geometric sandwich lets one compare the tail of any admission policy with the tails of the extreme policies "everyone joins" and "only high-need users join"; it is the comparison used in the proof of Theorem 1 to show that truncating a threshold policy at is stochastically dominated by the original.
Preamble
import Mathlib import Definitions.Def_NeedPersuasion_Threshold_Model
Formal statement
namespace NeedPersuasion.Threshold
theorem lemma_3 (M : Model) (hM : M.Standing) (π : ℕ → ℝ) (hπ : π ∈ PiAP M) :
(∀ k q : ℕ, q < k → M.lamH ^ (k - q) * π q ≤ π k ∧ π k ≤ M.lam ^ (k - q) * π q) ∧
(0 < M.lamH → ∀ k : ℕ, 0 < π k ∧ π k < 1) := by sorry
end NeedPersuasion.Threshold
Source
arXiv:2005.07253v3, Lemma 3, p. 36
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.