Theorem 8.1 — the diagonal formula is the unique sequence of fuzzy values; it satisfies (5) and picks the core of concave homogeneous games
ProvedFuzzyGames.Values.theorem_8_1Let be the space of continuously differentiable coalitional worth functions on fuzzy coalitions with , normed by the norm of the cube. A sequence of fuzzy values is a family of continuous linear maps satisfying Pareto optimality, symmetry and atomicity.
- There exists a sequence of fuzzy values, given by the diagonal formula
- It is unique: every sequence of fuzzy values satisfies for all and .
- For every linear operator with ,
where is the transpose of . 4. If is moreover concave and positively homogeneous on , then is the unique element of the core .
The theorem is an axiomatic characterisation of the diagonal (Aumann–Shapley) value for smooth games with fuzzy coalitions.
Formalization Note The paper writes "" in the last sentence, read as . Elements of are on all of (any function on the cube extends), and continuity of means with the seminorm of the cube; without this continuity uniqueness would fail. Parts 1–2 say that the formula (4) is, up to the irrelevant component, the only sequence of fuzzy values. Part 3 is stated for the formula (4), which by 2 is the sequence; the game is an element with and . In part 4, concavity and positive homogeneity (, , ) are required on the orthant , to which §2 (2) extends ; the core quantifies over the cube as in §2 (4)(a).
import Mathlib import Definitions.Def_FuzzyGames_Values_Basic
namespace FuzzyGames.Values
/-- Theorem 8.1 (Aubin 1981, §8, p. 10). (i) The diagonal formula (4),
`ψ_n v = ∫_0^1 Dv(tτ^N) dt`, defines a sequence of fuzzy values; (ii) every sequence of fuzzy
values coincides with it on `V^n` for every `n ≥ 1`; (iii) property (5): for every linear
`A : ℝ^m → ℝ^n` with `Aτ^M = τ^N`, `ψ_m(v ∘ A) = A* ψ_n v`, where
`(A* c)_j = ∑_i (A e_j)_i c_i`; (iv) if `v ∈ V^n` is concave and positively homogeneous on
`ℝ^n_+`, then `ψ_n v` is the unique element of the core of `v`. -/
theorem theorem_8_1 :
(∃ ψ : (n : ℕ) → Vn n →ₗ[ℝ] (Fin n → ℝ), IsSequenceOfFuzzyValues ψ ∧
∀ n : ℕ, 0 < n → ∀ v : Vn n, ψ n v = diagValue v.1) ∧
(∀ ψ : (n : ℕ) → Vn n →ₗ[ℝ] (Fin n → ℝ), IsSequenceOfFuzzyValues ψ →
∀ n : ℕ, 0 < n → ∀ v : Vn n, ψ n v = diagValue v.1) ∧
(∀ (n m : ℕ) (A : (Fin m → ℝ) →ₗ[ℝ] (Fin n → ℝ)), A 1 = 1 →
∀ (v : Vn n) (w : Vn m), (∀ σ : Fin m → ℝ, w.1 σ = v.1 (A σ)) →
∀ j : Fin m, diagValue w.1 j = ∑ i, A (Pi.single j 1) i * diagValue v.1 i) ∧
(∀ (n : ℕ) (v : Vn n), ConcaveOn ℝ (Set.Ici 0) v.1 → IsPosHomogeneous v.1 →
FuzzyGames.TUCore.fuzzyCore v.1 = {diagValue v.1}) := by sorry
end FuzzyGames.Values
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.