Theoretical and practical issues in single-machine scheduling with two job release and delivery times

We study a single-machine scheduling problem with two allowable job release and delivery times. This special case of a strongly NP-hard single-machine scheduling problem with an arbitrary number of job release and delivery times and with the objective to minimize the maximum job completion time rema...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Journal of scheduling Ročník 24; číslo 6; s. 615 - 647
Hlavní autoři: Reynoso, Alejandro, Vakhania, Nodari
Médium: Journal Article
Jazyk:angličtina
Vydáno: New York Springer US 01.12.2021
Springer Nature B.V
Témata:
Algorithms
Artificial Intelligence
Business and Management
Calculus of Variations and Optimal Control; Optimization
Completion time
Dynamic programming
Heuristic
Heuristic methods
Job shops
Operations Research/Decision Theory
Optimality criteria
Optimization
Polynomials
Scheduling
Supply Chain Management
Single-machine scheduling
Release time
Delivery time
Polynomial and pseudo-polynomial time algorithms
SUBSET SUM
Time complexity
ISSN:1094-6136, 1099-1425
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:We study a single-machine scheduling problem with two allowable job release and delivery times. This special case of a strongly NP-hard single-machine scheduling problem with an arbitrary number of job release and delivery times and with the objective to minimize the maximum job completion time remains NP-hard. Nevertheless, it is more transparent and accessible than the general version. Hence, it permitted us to establish some nice properties that yielded simple and efficient solution methods. On the one hand, the restricted setting is useful by its own right since it fits some real-life applications. On the other hand, the presented case study is helpful in the solution of a more general setting with a constant number of job release and delivery times. In particular, the optimality conditions and the heuristics methods that we propose here for the restricted setting might be generalized to the extended setting. The established optimality criteria are explicit conditions under which the restricted problem can be solved optimally in time O ( n log n ) . The extensive computational study showed that at least one of these conditions is satisfied for practically all the randomly generated 50 million problem instances while applying our heuristics to these instances. We report a favorable practical performance of our polynomial-time subroutine that invokes our five heuristics and verifies the proposed optimality conditions consecutively. These conditions were verified for the solutions delivered by the five heuristics individually and in a combined fashion as four pairs of heuristics and as all the five heuristics together. Fifty million problem instances were randomly generated using uniform distribution for each of these ten combinations. For the 50 million instances generated for the last (overall) combination, the schedule created by at least one of the heuristics has satisfied at least one of our optimality conditions (so all these problem instances were solved optimally). We also addressed the (theoretical) possibility that none of our conditions is satisfied, and showed how a known dynamic programming algorithm for SUBSET SUM problem can be used for the solution of the scheduling problem in pseudo-polynomial time. For a considerable part of the tested instances, one of the five scheduling heuristics has solved optimally also SUBSET SUM problem.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1094-6136
1099-1425
DOI:10.1007/s10951-021-00708-4