Y. Tanimura, T. Hiroyasu, M. Miki, and K. Aoi (Japan)
Grid Computing, Evolutionary Computing, Optimization
The computational Grid can offer users tremendous com puter resources. Many researchers are developing the Grid middleware and the typical results are Grid RPCs. However, application models are restricted to use when Grid RPCs are applied. In this paper, we proposed the ”EVOLVE/G” system for developer to construct evolution ary computation (EC) system on the computational grid. The EVOLVE/G has a tree topology of data communica tion. In the EVOLVE/G, there are Agent and some Work ers. Since the data can be transferred between Agent and Worker, any logical model of EC can be implemented by the EVOLVE/G. Furthermore, it has mechanism of clustering nodes. Therefore, the effective model can be constructed on the Grid environment. In this paper, the Grid calculation model of Parallel Simulated Annealing us ing the Genetic Crossover (PSA/GAc) is built using the EVOLVE/G and presented the experimental results in the real Grid environment. Consequently, it is shown that the examination of the calculation model using the EVOLVE/G is effective.
Important Links:
Go Back
