Research And Implementation Of Parallel Genetic Algorithm

The paper proposes a distributed parallel genetic algorithm, and the development algorithms program under Beowulf cluster system be achieved by Java language. Paper sets up a cheap Beowulf cluster system as hardware environment of the subject. Parallel genetic algorithm uses Java RMI and multithreading to deal with the communication between each computing node of the cluster and the population pool in order to complete the "migration" strategy amang the population. Especially to design the data structure of populaton pool is the core of the subject. The population pool data structure uses multiple Hashtable and Java Vector object, which is specialized used for storage for the outstanding individual generated by all computing nodes. Because of the design of the population pool , the complex communication problems among the computing nodes has become simple. Finally, paper solves an instances problem successfully and through comparing with the results from the use of Matlab displayed advanced performance of parallel genetic algorithm which is designed by the subject.