Theorem 8 — Delayed SWPT has a competitive ratio of 2
OpenDelayedSWPT.Main.competitive_ratio_twoConsider the single-machine problem of minimizing total weighted completion time with release dates and no preemption, with integer release dates , integer processing times and positive weights . Let be the schedule produced by the online algorithm Delayed SWPT. Then is the least such that
for every instance and every feasible (offline) schedule of that instance. In words: Delayed SWPT is -competitive, and no smaller constant is valid for this algorithm.
Since no online algorithm for this problem has a competitive ratio below 2 (Hoogeveen and Vestjens, 1996), Delayed SWPT is a best possible online algorithm; that general lower bound is not part of this statement.
Formalization Note The competitive ratio is defined (p. 686) as the least such , so the statement is IsLeast of the set of valid . Comparing with every feasible schedule is the same as comparing with the offline optimum and needs no existence of an optimum. The lower half concerns Delayed SWPT only (a one-job instance with , suffices).
import Mathlib import Definitions.Def_DelayedSWPT_Model_dswpt
namespace DelayedSWPT.Main
open DelayedSWPT.Model
/-- Theorem 8 of Anderson and Potts (2004), p. 696: Delayed SWPT has a competitive ratio of 2.
The competitive ratio (p. 686) is the least `ρ` such that on every instance the Delayed SWPT
schedule costs at most `ρ` times every feasible (offline) schedule. -/
theorem competitive_ratio_two :
IsLeast {ρ : ℝ | ∀ (n : ℕ) (I : Instance n) (S : Fin n → ℕ), IsFeasible I.r I.p S →
cost I.w I.p (dswpt I) ≤ ρ * cost I.w I.p S} 2 := by sorry
end DelayedSWPT.Main
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.