A Modified SQP Method for Inequality Constrained Optimization

In this paper, we proposed a modified SQR method, contrast to traditional sequential quadratic programming method, it has following merits. Firstly, at every trial point, the quadratic sub problem is feasible whether is a feasible point or not. Secondly, the method does not require all the iteration points in feasible region. Thirdly, it has no demand on penalty function, whose penalty parameter is always difficult to obtain. Under some reasonable conditions, the global convergence result of our algorithm is presented.