p. 47 — Σ_j G_I(x)_ij = 1, so G_I(x) lies in Δ
ProvedFixpNash.DivFree.G_isMixedProfileConsider a finite game in which every player has a nonempty set of pure strategies, and let be any real vector indexed by player–strategy pairs. Then , defined by , is a mixed strategy profile: for every player ,
In the paper this is the remark that, by the choice of , the map sends its domain into itself, so that is a self-map of a compact convex set and its fixed points are meaningful.
Formalization Note. The paper states the sum identity and concludes ; the Lean conclusion is the membership (AGT.IsMixedProfile), which contains the sum identity and the nonnegativity that is immediate from the . It is stated for every real , a slight generalization of . Every strategy set is assumed nonempty, the paper's tacit assumption.
import Mathlib import Definitions.Def_agt_games import Definitions.Def_DGPNash_NashMap_nashMap import Definitions.Def_FixpNash_DivFree_Map
namespace FixpNash.DivFree
/-- p. 47: `∑_{j ∈ Sᵢ} G_I(x)ᵢⱼ = 1` for every player `i`, so `G_I(x)` lies in `Δ`
(its entries are nonnegative by construction). Holds for every real vector `x`. -/
theorem G_isMixedProfile {ι : Type*} [Fintype ι] [DecidableEq ι]
{S : ι → Type*} [∀ i, Fintype (S i)] [∀ i, DecidableEq (S i)]
(u : ι → (∀ i, S i) → ℝ)
(hS : ∀ i, Nonempty (S i)) (x : ∀ i, S i → ℝ) :
AGT.IsMixedProfile (G u x) := by sorry
end FixpNash.DivFree
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.