摘要
对具有线性等式和不等式约束的线性规划问题给出了一种内点法,利用寻优方向选择参加投影矩阵计算的约束,使少部分约束参加运算,从而减少了问题的求解规模,有效地提高了求解速度,同时也节省了存贮量。
An interior point method for solving linear programming with hybridlinear constraints is presented. The method can uniformly treat equality and inequality constraints, and can also choose a few candidate constraints to take partin the projective computation using its descending direction. In this way, the pr-oblem scale can highly be reduced, especially for large scale problem. Numericaltests have snown that our algorithm can effectively improve the convergence rateThe convergence proof is presented too.
出处
《中山大学学报（自然科学版）》
CAS
CSCD
1992年第4期19-25,共7页
Acta Scientiarum Naturalium Universitatis Sunyatseni
基金
国家自然科学基金
中山大学高等学术研究中心基金
关键词
线性规划
内点法
投影矩阵
约束
linear programming
interior point method
search direction
projective matrix
