摘要
针对一般的光滑约束最优化问题,提出一种原始对偶不可行内点算法,该算法运用3个值函数使算法能收敛到局部极小点而非其他一阶最优性点,并通过将等式约束的罚项和松弛变量的障碍项添加到目标函数中转化原问题.计算结果证明了算法的可行性和有效性.
Aiming at general smooth constrained optimization problems, we proposed a primal-dual infeasible interior-point method. We added the penalty term of the sum of equality constraint and the barrier term of slack variables to the object function so as to convert the original problem. In the method, there are three merit functions to make the algorithm convergence to the local minimum point rather than other first order optimization point. We gave computational results, showing that the algorithm can solve nonlinear programming problems in an efficient way.
出处
《吉林大学学报（理学版）》
CAS
CSCD
北大核心
2009年第4期677-682,共6页
Journal of Jilin University:Science Edition
基金
国家自然科学基金(批准号:10771020)
关键词
内点法
原始对偶
非线性规划
interior-point method
primal-dual
nonlinear programming
