A General Purpose Distributed Implementation of Simulated Annealing

In this paper, we present a problem independent general purpose parallel implementation of simulated annealing on distributed message-passing multiproces-sor systems. The sequential algorithm is studied and we give a classiication of combinatorial optimization problems together with their neighborhood structures. Several parallelization approaches are examined c on-sidering their suitability for problems of the various classes. For typical representatives of the diierent classes good p arallel simulated annealing implementations are p r esented. A new parallel SA-algorithm is presented, that works simultaneously on several markov chains and decreases the number of chains dynamically. With a p arallel self-adapting cooling schedule this method yields good r esults. All algorithms are implemented in OCCAM-2 on a free c onngurable transputer system. Measurements on various numbers of processors up to 128 transputers are p r esented.