Proof of Theorem 1 — product strategy earns zero
ProvedHartSchmeidler.Finite.product_strategy_payoff_zerocorrelated-equilibriumfinite-gamesp2o-batch-pfp2bp2o-gran-per-chapterp2o-plan-paperp2o-v1product-strategies
Let be a lottery over all triples in the auxiliary game. Suppose that, for each player , is a probability vector on satisfying equation (2) for the nonnegative slice and every . Set . Then
This is the product-strategy calculation closing the nonnegative-response condition in the proof of Theorem 1.
Formalization Note Each is required to be a lottery, and the are slices of one global lottery. The calculation uses the exact auxiliary payment appearing in the correspondence milestone.
Preamble
import Definitions.Def_HartSchmeidler_Finite_Game
Formal statement
namespace HartSchmeidler.Finite
open Finset
/-- Proof of Theorem 1, p. 20: the product of the playerwise lotteries is
a mixed strategy of player I, and equation (2) makes its payoff zero. -/
theorem product_strategy_payoff_zero {ι : Type*} [Fintype ι] [DecidableEq ι]
{S : ι → Type*} [∀ j, Fintype (S j)] [∀ j, DecidableEq (S j)]
[∀ j, Nonempty (S j)] (h : ι → (∀ j, S j) → ℝ)
(y : Deviation S → ℝ) (hy : AGT.IsLottery y)
(x : ∀ i, S i → ℝ) (hx : ∀ i, AGT.IsLottery (x i))
(hbalance : ∀ (i : ι) (s : ∀ j, S j),
∑ r : S i, x i r * ∑ t : S i, y ⟨i, (r, t)⟩ *
(h i (Function.update s i r) - h i (Function.update s i t)) = 0) :
AGT.IsLottery (fun r : ∀ j, S j => ∏ j, x j (r j)) ∧
(∑ r, ∑ c, (∏ j, x j (r j)) * y c * auxPayoff h r c) = 0 := by sorry
end HartSchmeidler.Finite
Source
Hart and Schmeidler, Existence of Correlated Equilibria, Math. Oper. Res. 14 (1989), p. 20, proof of Theorem 1 (‘Define x(r)’ through ‘which equals zero by (2)’); https://doi.org/10.1287/moor.14.1.18
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.