Abstract: With a large amount of computer resources becoming available in the network, grid can been used to handle a single problem simultaneously. Selecting more than one outstanding resource from numerous resources is an NP problem. In this paper, a new adaptive parallel genetic algorithm (NAPGA) is proposed, with which parallel model simulation of the grid resources selection strategy on C+MPI platform is made. The result indicates that the algorithm can effectively solve the problem of premature convergence, and produce results that are better than a new improved parallel genetic algorithm (NIPGA). Searching and converging disciplinarian of genetic algorithm are discussed.

Key words: genetic algorithm, MPI, grid