Parallel branch-and-price algorithms for the single machine total weighted tardiness scheduling problem with sequence-dependent setup times

Scheduling problems occur in a broad range of real-world application fields and have attracted a huge set of research articles. However, there is only little research on exact algorithms for scheduling problems, many of which are NP-hard in the strong sense. We investigate the problem on a single ma...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Computers & operations research Ročník 173; s. 106804
Hlavní autori: Speckenmeyer, Philipp, Hilmer, Constanze, Rauchecker, Gerhard, Schryen, Guido
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Elsevier Ltd 01.01.2025
Predmet:
Branch-and-price algorithm
Dynamic programming
Sequence-dependent setup times
Shared-memory parallelization
Single machine scheduling
Weighted tardiness
Branch-and-price algorithm
Weighted tardiness
Sequence-dependent setup times
Dynamic programming
Single machine scheduling
Shared-memory parallelization
ISSN:0305-0548
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:Scheduling problems occur in a broad range of real-world application fields and have attracted a huge set of research articles. However, there is only little research on exact algorithms for scheduling problems, many of which are NP-hard in the strong sense. We investigate the problem on a single machine with a total weighted tardiness objective function and sequence-dependent setup times. First, we adopt a serial branch-and-price algorithm from the literature and present a modified branching strategy and a primal heuristic. Second, we use the potential of parallel computing architectures by presenting two parallel versions of the branch-and-price algorithm. Third, we conduct extensive computational experiments to show that our parallelization approaches provide substantial parallel speedups on well-known benchmark instances from the literature. We further observe that the parallel speedups achieved by our parallel algorithms are very robust among all tested instances. •We address a single machine scheduling problem with sequence-dependent setup times.•We suggest and implement two exact parallel branch and price algorithms.•The parallelization is applied to dynamic programming to solve the pricing problem.•Computational experiments are performed on an HPC cluster.•We observe an almost linear speedup for a certain range of thread counts.
ISSN:0305-0548
DOI:10.1016/j.cor.2024.106804