非线性约束条件下的SQP可行方法

SQP feasible algorithm on solving the nonlinear programming

非线性规划求解问题,一直是人们关心的热点问题。Zhu和Zhang利用对具有不等式约束的非线性规划构造出新的超线性收敛的SQP算法,每次迭代只需解一个二次规划子问题,还可自动修正可行方向以避免Marotos效应,并在较弱条件下保持算法的整体收敛性。研究将Zhu和Zhang工作,推广到更一般具有等式约束和具有不等式约束的非线性规划。

The nonlinear programming problem is a hotspot problem that people care about all the time.Zhu and Zhang make use of the nonlinear programming with inequality constraint to construct a new sequence quadratic programming method. We need only solving one quadratic sub-problem at each iteration and modify the feasible direction automatically in order to avoid the Marotos effect. The method could retain the global convergence under the weak condition. We extend the work of Zhu and Zhang to the nonlinear programming with equality and inequality constraint and obtain a better sequence quadratic programming method.