Proof of Theorem 3.3, p. 40 — KS1 ≥ S2, strictly when the P_k are not all equal
ProvedRegAggregation.Posterior.KS1_ge_S2cauchy-schwarzforecast-aggregationinequalityp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Let , and . Let
Then . If, moreover, for some , then .
In the proof of Theorem 3.3 (iii), this is what makes the scale of the inverse gamma integral in strictly positive when the forecasters disagree.
Formalization Note The page writes the upper limit of the sum in as ; it is the number of forecasters.
Preamble
import Mathlib import Definitions.Def_RegAggregation_Posterior_Model
Formal statement
namespace RegAggregation.Posterior
theorem KS1_ge_S2 (K : ℕ) (μ0 : ℝ) (P : Fin K → ℝ) :
S2 μ0 P ≤ (K : ℝ) * S1 μ0 P ∧
((∃ i k, i ≠ k ∧ P i ≠ P k) → S2 μ0 P < (K : ℝ) * S1 μ0 P) := by sorry
end RegAggregation.Posterior
Source
Satopää, Regularized Aggregation of One-off Probability Predictions, SSRN working paper 3769945 (version of 2021-08-10; published in Operations Research 2022), p. 40, proof of Theorem 3.3 (iii)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.