Theorem 2 — positive table and no Pareto optimal α-EFX allocation above 2/3 + 1/K
ProvedParetoEFX296.Approx.theorem_2For each integer , consider the four-agent, nine-good additive table of Theorem 2. Every entry is positive, and for every real ,
Allocations are all complete integral partitions of the nine goods, possibly with empty bundles, and Pareto optimality compares with every such allocation. Some agent must strictly improve in a Pareto domination. The statement preserves all values in the explicit table and the full range ; it is the page 9 theorem together with positivity from its page 3 formulation. Taking arbitrarily large recovers the impossibility for every . It asserts incompatibility with Pareto optimality, and does not assert nonexistence of EFX allocations alone.
Differences from the source:
- encoding: partitions of nine goods among agents 1,2,3,4 with integer K ≥ 6 → owner maps Good → Agent with named constructors, K : ℕ ≥ 6, and real table sums; reason: bijection with all complete integral allocations and all source integers K ≥ 6; real casts preserve every utility
- generalization: Definition 1 restricts α to (0,1] → the same distinct-agent EFX inequality is used for every real α above 2/3 + 1/K; reason: it agrees on the source range; for α > 1 positivity implies 1-EFX, and 1 > 2/3 + 1/K for K ≥ 6
import Mathlib import Definitions.Def_ParetoEFX296_Approx_Instance
namespace ParetoEFX296.Approx
theorem theorem_2 (K : ℕ) (hK : 6 ≤ K) :
(∀ (i : Agent) (g : Good), 0 < inst K i g) ∧
∀ α : ℝ, (2 : ℝ) / 3 + 1 / (K : ℝ) < α →
¬ ∃ A : Allocation Agent Good,
IsAlphaEFX (inst K) α A ∧ IsParetoOptimal (inst K) A := by sorry
end ParetoEFX296.Approx
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.