Conclusion Part II.

            Currently, our results seem inferior to the improvement
 obtained with other means in table 1–3 . In the experiments executed thus far
         in this project, the distributed genetic algorithm manipulates
                the evaluation-coefficients within the GNUchess
                program in a smaller number (6 or 12) of groups
and uses a small population for optimization. As points out, the limited success
                             of our past experiments
                  may have its cause in the limited number of
             parameters which we optimize, or simply the fact that
            the original GNU-chess program has a family of builtin
               evaluation-coefficients which are close to optimal
                             within the search space.8

無題 8

  • 1.
    Conclusion Part II. Currently, our results seem inferior to the improvement obtained with other means in table 1–3 . In the experiments executed thus far in this project, the distributed genetic algorithm manipulates the evaluation-coefficients within the GNUchess program in a smaller number (6 or 12) of groups and uses a small population for optimization. As points out, the limited success of our past experiments may have its cause in the limited number of parameters which we optimize, or simply the fact that the original GNU-chess program has a family of builtin evaluation-coefficients which are close to optimal within the search space.8