基于正基算法的收敛性研究

Research for Convergence of Algorithm Based on Positive Basis

对于线性约束的非线性规划问题,利用正基给出了一种算法,在每次迭代时无需求解线性规划子问题;并证明在目标函数连续时,算法在有限步终止于一个K-T点,或产生一个无穷点列其聚点是一个K-T点。

A algorithm is gived by using positive basis, which are relevant to nonlinear constrained programming. The new algorithm does not need linear programming problem to be soved in every iterative step and it is easy to implement. The global convergence is also proved.