§3.2, p. 12 — the QA pool is well-defined: exactly one p* ∈ Δⁿ has Σᵢ wᵢ g(pᵢ) as a subgradient of G
ProvedQAPooling.MaxMin.qaPool_existsUniqueconvex-optimizationp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1proper-scoring-rulequasi-arithmetic-pooling
Let be a regular proper scoring rule whose expected reward is continuous on , and let be an exposure function of . Let be forecasts with non-negative weights adding to . Then there is exactly one such that
i.e. exactly one quasi-arithmetic pool .
This is the well-definedness claim following Definition 3.6: it turns "the unique " of the definition into a theorem, and it is what makes the QA-pool hypothesis of Theorem 4.1 satisfiable.
Formalization Note Continuity of is the paper's standing assumption of p. 10. Subgradients are relative to (footnote 10).
Preamble
import Mathlib import Definitions.Def_QAPooling_MaxMin_Model
Formal statement
namespace QAPooling.MaxMin
/-- The QA pool is well-defined (§3.2, p. 12): for forecasts in `D` with non-negative weights adding
to 1, there is exactly one `p* ∈ Δⁿ` at which `Σ_i w_i g(p_i)` is a subgradient of `G`. -/
theorem qaPool_existsUnique {n m : ℕ} (s : ScoringRule n) (g : (Fin n → ℝ) → Fin n → ℝ)
(hreg : IsRegular s) (hprop : IsProper s)
(hGcont : ContinuousOn (expectedReward s) (stdSimplex ℝ (Fin n)))
(hg : IsExposure s g)
(p : Fin m → Fin n → ℝ) (hp : ∀ i, p i ∈ forecastDomain s)
(w : Fin m → ℝ) (hw0 : ∀ i, 0 ≤ w i) (hw1 : ∑ i, w i = 1) :
∃! pstar, IsQAPool s g p w pstar := by sorry
end QAPooling.MaxMin
Source
Neyman & Roughgarden, From Proper Scoring Rules to Max-Min Optimal Forecast Aggregation, arXiv:2102.07081v2, p. 12, §3.2, paragraph after Remark 3.7 ('The QA-pool is well-defined, i.e. exists and is unique'), Definition 3.6
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.