Theorem 6.1, proof — joining an edge used by one other player raises by times the new share
ProvedPriceOfStability.WeightedPotential.join_shared_edgeLet be a weighted cost-sharing game with weights and edge costs in which each edge lies in the strategy spaces of at most two players. Let be a profile and an edge used in by exactly one player , and let a player switch to a feasible strategy containing , giving the profile . Then
where is the weight on in ; that is, the change in the edge potential equals times the cost incurs on after joining.
This is the computation the paper displays in the proof of Theorem 6.1, the basic case of the weighted potential identity.
Formalization Note "The edge already supported another player " is encoded as: the set of users of in is exactly . The conclusion states both the closed form and the form times 's new payment for .
import Mathlib import Definitions.Def_PriceOfStability_WeightedPotential_Model
namespace PriceOfStability.WeightedPotential
variable {ι E : Type*} [Fintype ι] [DecidableEq ι] [Fintype E] [DecidableEq E]
/-- Anshelevich et al., SIAM J. Comput. 38 (2008), Theorem 6.1, proof, p. 1620 (PDF p. 19),
displayed computation: "Consider a player i and an edge e that player i joins. If the edge already
supported another player j, then i's cost for using e is c_e w_i/(w_i+w_j), while the change in
Φ_e(S) is c_e(w_i − w_i w_j/(w_i + w_j)) = c_e w_i^2/(w_i + w_j). Thus the change in potential when
i joins e equals the cost i incurs, scaled up by a factor of w_i."
In a standard weighted game in which every edge lies in the strategy spaces of at most two players,
let `S` be a profile in which edge `e` is used by exactly one player `j ≠ i` (so `i` does not use
it), and let `i` switch to a feasible strategy `T ∋ e`. Then the edge potential of `e` rises by
`c_e wᵢ²/(wᵢ + wⱼ)`, which is `wᵢ` times `i`'s payment `(wᵢ/W_e) c_e` for `e` after the switch.
**Formalization Note.** "The edge already supported another player j" and "player i joins e" are
`users S e = {j}` together with `e ∈ T`; the hypotheses of Theorem 6.1 are kept as binders. -/
theorem join_shared_edge (G : WeightedGame ι E) (hG : G.IsStandard)
(hspace : ∀ e, (Finset.univ.filter (fun i => e ∈ strategySpace G i)).card ≤ 2)
(S : ι → Finset E) (hS : IsProfile G S) (i j : ι) (hij : i ≠ j)
(T : Finset E) (hT : T ∈ G.strategies i) (e : E)
(hjoin : users S e = {j}) (heT : e ∈ T) :
edgePotential G (Function.update S i T) e - edgePotential G S e
= G.edgeCost e * G.weight i ^ 2 / (G.weight i + G.weight j) ∧
edgePotential G (Function.update S i T) e - edgePotential G S e
= G.weight i * (G.weight i / edgeWeight G (Function.update S i T) e * G.edgeCost e) := by sorry
end PriceOfStability.WeightedPotential
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.