摘要
提出了一个处理等式约束优化问题新的SQP算法,该算法通过求解一个增广Lagrange函数的拟Newton方法推导出一个等式约束二次规划子问题,从而获得下降方向.罚因子具有自动调节性,并能避免趋于无穷.为克服Maratos效应采用增广Lagrange函数作为效益函数并结合二阶步校正方法.在适当的条件下,证明算法是全局收敛的,并且具有超线性收敛速度.
In this paper, a new SQP method is presented to solve equality constrained optimization. It obtains descent direction by solving a equality constrained optimization subproblem which is deduced by solving the augmented Lagrangian in quasi-Newton method. The penalty parameter is adjusted automatically and avoided tending to infinity. In order to conquer Maratos effect, it takes augmented Lagrangian as a merit function and combines two-step revised method. Under some suitable assumptions, we prove that the algorithm is global convergence as well as superlinear convergence.
出处
《纯粹数学与应用数学》
CSCD
2009年第2期276-283,共8页
Pure and Applied Mathematics
基金
国家自然科学基金(10501009)
广西自然科学基金(0728206)
关键词
等式约束优化
SQP算法
等式约束二次规划
全局收敛
超线性收敛
equality constrained optimization, SQP algorithm, equality constrained quadratic programming, global convergence, superlinear convergence
