Theorem 2.5 — the LP schedule's mean busy times are optimal for (R)
ProvedSingleMachineSched.LPRelax.theorem_2_5linear-programminglp-relaxationmean-busy-timep2o-batch-p200bp2o-gran-per-chapterp2o-plan-paperp2o-v1scheduling
Let jobs have integral processing times , integral release dates and weights , indexed so that
Let be the mean busy time of job in the LP schedule. Then is an optimal solution to the relaxation (R):
- satisfies every constraint , ;
- for every feasible ,
Consequently .
Formalization Note Jobs are Fin n, 0-based, and the sortedness is the hypothesis j ≤ k → w_k/p_k ≤ w_j/p_j. Optimality is stated for the objective of (R), ; it differs from by the constant , so the two forms are equivalent. The statement is made for (the paper's standing assumption is ).
Preamble
import Mathlib import Definitions.Def_SingleMachineSched_LPRelax_LPSchedule import Definitions.Def_SingleMachineSched_Shared_RelaxationR
Formal statement
namespace SingleMachineSched.LPRelax
/-- Theorem 2.5: with the jobs indexed by nonincreasing `w_j / p_j`, the mean busy time vector
`M^LP` of the LP schedule is an optimal solution to (R). -/
theorem theorem_2_5 {n : ℕ} (p r : Fin n → ℕ) (w : Fin n → ℝ)
(hp : ∀ j, 0 < p j) (hw : ∀ j, 0 ≤ w j)
(hsort : ∀ j k : Fin n, j ≤ k → w k / p k ≤ w j / p j) :
Shared.FeasibleR p r (mLP p r) ∧
∀ M : Fin n → ℝ, Shared.FeasibleR p r M → Shared.objR p w (mLP p r) ≤ Shared.objR p w M := by sorry
end SingleMachineSched.LPRelax
Source
Goemans, Queyranne, Schulz, Skutella & Wang, Single Machine Scheduling with Release Dates, SIAM J. Discrete Math. 15(2) (2002), p. 173, Theorem 2.5
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.