摘要
针对通常精确罚函数方法在可行域边界不可微的缺点,构造了一类可微的精确罚函数方法,使得它能采用无约束优化方法中许多有效的解析方法。作者提出了精确罚函数的构造,讨论了它的性质,证明了算法的收敛性,并给出了数值计算实例。
In general, the exact penalty function is not differentialble in the adge of feasible region. A differentiable exact penalty function is given in this paper. So that the analytical unconstrained methods can be used. This paper gives the form and some properties of the exact penalty function, convergence theorem of the algorithm and numerical examples.
基金
中科院管理
决策
信息管理系统开放实验室资助课题
关键词
精确罚函数
约束优化问题
无约束优化方法
exact penalty function
constrained optimization problem
unconstrained optimization method
