针对无约束最优化问题,提出了一种能减少迭代次数的单纯形方法.基于NM(Nelder—Mead)单纯形方法和MDS(Multi-directional Search)单纯形方法的各自优点,来构造新的搜索方向.对于大多数的测试问题,该算法可以减少迭代次数并减少代入函数值次数.
In this paper, an improved simplex method based on Nelder-Mead's simplex method ( N-M ) and Multi-directional Search method (MDS) is presented for unconstrained function optimization. The effectiveness is demonstrated in that the number of function evaluations are reduced significantly for most tested problems.