向量合成加速单纯形算法

An Acceleration Simplex Method by Vector Synthesis

提出了一种实现加速单纯形算法的方法。N-M单纯形算法是从最差点出发,以单纯形其他顶点的中心为寻优方向,提出的向量合成加速单纯形算法是以最好点出发,参照梯度算法思想,采用向量合成公式寻得一最好点,此最好点和单纯形除最差点外所有顶点构成新的单纯形,再进行搜索。实验证明,该方法较N-M单纯形算法、拟梯度算法速度更快,效果更好。

An accelerated method of simplex algorithm was studied. N-M simplex algorithm departures from the worst point to the center of the other vertex of the simplex as the optimization direction. Vector synthesis is a method to speed up the simplex algorithm. Referring to the gradient algorithm, the best point is found by virtue of vector synthesis formula. The best vertex and other vertex except the worst one consti-tute a new simplex for searching. Experimental results show that the method is faster and better than NM simplex algorithm and gradient algorithm.