p. 159 — a solution satisfying selection, INV, SYM, IIA and PAR picks the maximizer of the Nash product
ProvedNashBargainingProblem.Axiomatic.nash_axioms_force_nash_product_maxA bargaining problem is a pair with compact and convex, the disagreement point, and some with and ; let be the set of them. Let assign to each a point and assume:
- (selection) ;
- (INV) for , and : if then ;
- (SYM, Nash's assumption 8) if and , then ;
- (IIA, Nash's assumption 7) if , both , and , then ;
- (PAR, Nash's assumption 6) if with and , then .
Then for every , writing : , , and
That is, the solution is the point of the set in the first quadrant (relative to ) where the product of the utility gains is maximized. This is the assertion Nash proves on p. 159, the necessity half of his characterization.
Formalization Note The five hypotheses are copied verbatim from the left-hand side of the Open platform theorem NashBargaining.nash_bargaining_solution_unique, and the conclusion is its last conjunct with in place of , so this milestone is the "only if" direction of that goal. The paper fixes at the origin by the normalization of p. 158 ("assign the number zero to this anticipation"); the translation by , i.e. INV with , is that normalization. The paper's INV is implicit: utilities are determined only up to positive scale (p. 158), and its assumption 8 allows symmetry after a rescaling, which with INV is the literal SYM above.
import Mathlib
namespace NashBargainingProblem.Axiomatic
theorem nash_axioms_force_nash_product_max
(B : Set (Set (ℝ × ℝ) × (ℝ × ℝ)))
(hB : B = {P | IsCompact P.1 ∧ Convex ℝ P.1 ∧ P.2 ∈ P.1 ∧
∃ s ∈ P.1, P.2.1 < s.1 ∧ P.2.2 < s.2})
(g : B → ℝ × ℝ)
-- g is a bargaining solution: it selects a point of S
(h_sel : ∀ (S : Set (ℝ × ℝ)) (d : ℝ × ℝ) (h : (S, d) ∈ B), g ⟨(S, d), h⟩ ∈ S)
-- INV: invariance under positive affine rescalings of the two utilities
(h_inv : ∀ (S : Set (ℝ × ℝ)) (d : ℝ × ℝ) (h : (S, d) ∈ B)
(α₁ α₂ β₁ β₂ : ℝ) (L : ℝ × ℝ → ℝ × ℝ), 0 < α₁ → 0 < α₂ →
(∀ s : ℝ × ℝ, L s = (α₁ * s.1 + β₁, α₂ * s.2 + β₂)) →
∀ h' : (L '' S, L d) ∈ B, g ⟨(L '' S, L d), h'⟩ = L (g ⟨(S, d), h⟩))
-- SYM: symmetric problems get symmetric outcomes
(h_sym : ∀ (S : Set (ℝ × ℝ)) (d : ℝ × ℝ) (h : (S, d) ∈ B), d.1 = d.2 →
(∀ s₁ s₂ : ℝ, (s₁, s₂) ∈ S ↔ (s₂, s₁) ∈ S) →
(g ⟨(S, d), h⟩).1 = (g ⟨(S, d), h⟩).2)
-- IIA: independence of irrelevant alternatives
(h_iia : ∀ (S T : Set (ℝ × ℝ)) (d : ℝ × ℝ) (hS : (S, d) ∈ B) (hT : (T, d) ∈ B),
S ⊆ T → g ⟨(T, d), hT⟩ ∈ S → g ⟨(S, d), hS⟩ = g ⟨(T, d), hT⟩)
-- PAR: Pareto efficiency (a point of S strictly improved upon by some t ∈ S is never chosen)
(h_par : ∀ (S : Set (ℝ × ℝ)) (d : ℝ × ℝ) (h : (S, d) ∈ B), ∀ s ∈ S, ∀ t ∈ S,
s.1 < t.1 → s.2 < t.2 → g ⟨(S, d), h⟩ ≠ s) :
∀ (S : Set (ℝ × ℝ)) (d : ℝ × ℝ) (h : (S, d) ∈ B),
d.1 ≤ (g ⟨(S, d), h⟩).1 ∧ d.2 ≤ (g ⟨(S, d), h⟩).2 ∧
∀ s ∈ S, d.1 ≤ s.1 → d.2 ≤ s.2 → s ≠ g ⟨(S, d), h⟩ →
(s.1 - d.1) * (s.2 - d.2) <
((g ⟨(S, d), h⟩).1 - d.1) * ((g ⟨(S, d), h⟩).2 - d.2) := by sorry
end NashBargainingProblem.Axiomatic
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.