Theorem 4.1 — MinWork is a strongly truthful n-approximation mechanism
ProvedAlgMechDesign.MinWork.minwork_strongly_truthful_n_approxConsider task scheduling on unrelated machines with agents and tasks, agent needing time for task . Let be any MinWork allocation rule (each task goes to an agent with minimal declared time, ties broken arbitrarily) and let
be the MinWork payments. Then:
- the mechanism is strongly truthful: truth-telling is a dominant strategy for every agent, and every positive misreport is strictly worse than the truth for some positive declarations of the others;
- is an -approximation: for every positive type vector and every allocation , , where is the make-span.
This is the upper bound of the paper's task scheduling section; the paper's lower bounds (Theorems 4.6, 4.10, 4.12) compare against it.
Formalization Note The statement quantifies over every allocation rule satisfying the MinWork specification, so it holds for every tie-breaking rule. The hypothesis makes the second-best minimum well defined; [NeZero n] is implied by it and only makes the make-span's maximum typecheck. Running time ("polynomial time") is not modelled.
import Mathlib import Definitions.Def_AlgMechDesign_MinWork_Model import Definitions.Def_AlgMechDesign_MinWork_Mechanism
namespace AlgMechDesign.MinWork
/-- Theorem 4.1: for `n ≥ 2` agents, any number `k` of tasks and every tie-breaking rule, the
MinWork mechanism is strongly truthful and an `n`-approximation for task scheduling. -/
theorem minwork_strongly_truthful_n_approx {n k : ℕ} [NeZero n] (hn : 2 ≤ n)
(alloc : (Fin n → Fin k → ℝ) → (Fin k → Fin n)) (hmin : IsMinWorkAlloc alloc) :
IsStronglyTruthful alloc (minWorkPay hn alloc) ∧ IsApprox (n : ℝ) alloc := by sorry
end AlgMechDesign.MinWork
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.