(8), PDF p. 6 — the cost to agent i of link j is c_i^j = (M^j + (1 − p_i^j)w_i)/s_j
ProvedWorstCaseEq.Speeds.eq8_linkCostConsider agents with traffic on parallel links with speeds , and let be a mixed profile in which every is a probability distribution. Fix an agent and a link , and let agent place its traffic on link with certainty while every other agent keeps its mixed strategy. Then the expected delay of agent equals
where is the expected traffic on link .
This is the speed version of (2): the expected cost of a pure deviation of one agent is an explicit linear expression in the other agents' probabilities, which is what makes the equilibria computable in closed form.
Formalization Note The pure deviation is the profile in which agent 's distribution is replaced by the point mass on ; its expected cost is minus the expected payoff of the published agt_games vocabulary. No positivity of speeds or traffic is needed for the identity, so none is assumed.
import Mathlib import Definitions.Def_agt_games import Definitions.Def_WorstCaseEq_Speeds_Model
namespace WorstCaseEq.Speeds
theorem eq8_linkCost {n m : ℕ} (w : Fin n → ℝ) (s : Fin m → ℝ) (p : Fin n → Fin m → ℝ)
(hp : AGT.IsMixedProfile p) :
∀ i j, -AGT.expectedPayoff (payoff w s) (Function.update p i (WorstCaseEq.Identical.pureLottery j)) i
= linkCost w s p i j ∧
linkCost w s p i j = (WorstCaseEq.Identical.expTraffic w p j + (1 - p i j) * w i) / s j := by sorry
end WorstCaseEq.Speeds
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.