Proof of Observation 1, p. 8779 (with (5), p. 8771) — the a-neutral probability of a uniform source takes every value in [0, 1]
ProvedSourceTheory.InsensMatch.source_prob_rangeAssume structural Assumption 1. Let be a source (a subalgebra of events on which is convex ranged) that is uniform with a-neutral probability and source function , so that on . Then takes every value in the unit interval:
The paper proves this range property for risk in the proof of Observation 1 ("Convex rangedness implies that w's domain is the entire [0,1]") and states that the proof for a source function (5) is identical. It is what lets every triple of probabilities summing to 1 be realised by a partition of the source, which turns preference conditions on events into conditions on functions of probabilities.
Formalization Note. is only finitely additive; the paper calls the countably additive case elementary and defers the general case to its Online Appendix A.
import Mathlib import Definitions.Def_SourceTheory_Uniform_Model
namespace SourceTheory.InsensMatch
open Set
/-- Proof of Observation 1 (p. 8779), applied to a uniform source by (5) (p. 8771): convex
rangedness forces the a-neutral probability `P_𝒜` of a uniform source to take every value in
`[0, 1]`. -/
theorem source_prob_range {S Γ : Type*} (𝓔 : Set (Set S)) (pref : (S → Γ) → (S → Γ) → Prop)
(U : Γ → ℝ) (W : Set S → ℝ) (h1 : SourceTheory.Uniform.Assumption1 𝓔 pref U W)
(𝒜 : Set (Set S)) (h𝒜 : SourceTheory.Uniform.IsSource 𝓔 W 𝒜) (PA : Set S → ℝ) (wA : ℝ → ℝ)
(hPA : SourceTheory.Uniform.IsUniformWith pref U 𝒜 PA) (hwA : SourceTheory.Uniform.IsSourceFunction W 𝒜 PA wA) :
∀ q ∈ Icc (0 : ℝ) 1, ∃ A ∈ 𝒜, PA A = q := by sorry
end SourceTheory.InsensMatch
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.