Abstract: "We propose a new parallel simulated annealing algorithm. Each processor maintains one solution and performs the annealing process concurrently at a constant temperature that differs from processor to processor, and the solutions obtained by the processors are exchanged occasionally i...

Ausführliche Beschreibung

Bibliographische Detailangaben

Hauptverfasser:	Kimura, Kouichi (VerfasserIn), Taki, Kazuo (VerfasserIn)
Format:	Buch
Sprache:	English
Veröffentlicht:	Tokyo, Japan 1991
Schriftenreihe:	Shin-Sedai-Konpyūta-Gijutsu-Kaihatsu-Kikō <Tōkyō>: ICOT technical report 673
Schlagworte:	Algorithms Combinatorial optimization
Zusammenfassung:	Abstract: "We propose a new parallel simulated annealing algorithm. Each processor maintains one solution and performs the annealing process concurrently at a constant temperature that differs from processor to processor, and the solutions obtained by the processors are exchanged occasionally in some probabilistic way. An appropriate cooling schedule is automatically constructed from a given set of temperatures that are assigned to the processors. Thus we can avoid the task of carefully reducing the temperature over the course of time, which is essential for the performance of the conventional sequential algorithm. In this paper we propose a scheme for the probabilistic exchange of solutions and justify it from the viewpoint of probability theory We have applied our algorithm to a graph-partitioning problem. Results of experiments, and comparison with those by the sequential annealing algorithm and the Kernighan-Lin algorithm, are discussed.
Beschreibung:	9 S.