首先回顾了采用最钝角行、列主元规则求解线性规画问题的原始、对偶可行解的主要过程,阐述了其与众不同的特性.然后构造了2个特殊的辅助问题,并证明了最钝角行、列主元规则的过程实际上分别等价于采用原始、对偶单纯形算法求解相应的辅助问题.此外,还对嵌套的pricing规则进行了回顾,并基于最优解的启发式特征刻画给出了该规则的一个几何解释.
First, the main procedures and the distinctive features of the most-obtuse-angle(MOA)row or column pivot rules are introduced for achieving primal or dual feasibility in linear programming. Then, two special auxiliary problems are constructed to prove that each of the rules can be actually considered as a simplex approach for solving the corresponding auxiliary problem. In addition, the nested pricing rule is also reviewed and its geometric interpretation is offered based on the heuristic characterization of an optimal solution.