摘要
讨论了非线性优化中Lagrange函数的鞍点与原问题和对偶问题的最优解之间的关系,并对对偶理论中的一些性质给予详细证明.对于凸规划在一定约束规格下鞍点总是存在的,可以通过求解鞍点问题来求最优解.最后给出在不等式约束条件下求鞍点的一个迭代方法.
This paper discuses the relationship between the saddle point of Lagrange function and the optimization solving of the primal optimization problem and its dual problem, with verification of some new features in duality theory. On convex optimization, the saddle point always exists in certain constraint qualification. We can solve convex optimization problem through obtaining the saddle point. At last. It presents a iterative method for saddle point problems theorem in inequality constrained condition.
出处
《沈阳师范大学学报（自然科学版）》
CAS
2008年第3期272-274,共3页
Journal of Shenyang Normal University:Natural Science Edition
基金
国家自然科学基金资助项目(10471096)
