摘要
本文首先给出由线性等式和不等式以及部分变量非负组成的约束集的一个新的转轴运算。它是以往转轴运算的推广。然后,以此为基础,建立该约束条件下的非线性规划的一个拓广的既约梯度法,它是既约梯度法的广泛推广和改进。算法不需增加任何松驰变量,以致提高问题的维数,扩大问题的规模;方法直接对原问题进行求解。本文算法对一般线性约束规划具有广泛的实用性,其处理技巧带有普遍意义。在非退化假设下,本文算法具有全局收敛性。
In this paper, a new pivoting method for the constained set with linear equality and inequalty, as well as a part ot nonneqative varibles is qiven at first . Then, base on this, an extension of reduced gradient for nonlinear programming with such constrained conditions is presented. It is a wide extension and modification of reduced gradient method. This method solves the original problem directly, and no slack vector is introduced such the dimension and the scale of the problem enlarging. This method possesses wide practical uses for general linearly constrained programming, and its technique has universal sense. Our method Possesses global convergence under nondegenerate assumption.
出处
《经济数学》
1993年第1期70-78,共9页
Journal of Quantitative Economics
