Proof of Theorem 2, pp. 23–24 — equidistance from iff for some
ProvedSeqLoss.Equidistance.equidistant_iff_prefixcost-sharingeuclidean-distancep2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Let have total and let . Then is at equal Euclidean distance from all ideal allocations if and only if there is with
This turns the geometric condition of Theorem 2 into a linear system in the prefix sums of ; the paper's intermediate form is .
Formalization Note The hypothesis is added: the paper divides by in passing between the two forms. For every is the zero vector, every is equidistant, and the prefix condition fails in general, so the equivalence needs it. The statement holds for every real , not only allocations. Agents are 0-based, so is the sum over indices .
Preamble
import Mathlib import Definitions.Def_SeqLoss_FixedFraction_Setting import Definitions.Def_SeqLoss_Equidistance_IdealAllocations
Formal statement
namespace SeqLoss.Equidistance
theorem equidistant_iff_prefix {n : ℕ} (ℓ x : Fin n → ℝ) (hL : SeqLoss.FixedFraction.total ℓ ≠ 0) :
IsEquidistantFromIdeals ℓ x ↔
∃ α : ℝ, ∀ k : ℕ, 1 ≤ k → k ≤ n →
(∑ i ∈ Finset.univ.filter (fun i : Fin n => i.val < k), x i) - SeqLoss.FixedFraction.total ℓ / 2 = α * (k : ℝ) := by sorry
end SeqLoss.Equidistance
Source
Gudmundsson, Hougaard & Ko, Sharing Sequentially Triggered Losses, accepted manuscript (2023), pp. 23–24, Appendix A, proof of Theorem 2, second and third displays
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.