非线性规划一般约束条件的SQP方法

A Successive Quadratic Programming Algorithm with Global and Superlinear Convergence Properties

提出了一种新的处理等式和不等式约束条件优化问题的SQP方法，计算过程中每一步迭代只需解一个二次规划。在一定条件下，证明了算法的全局和二步超线性收敛性，其优点是具有较小的计算量，避免了Maratos现象的发生。

This paper present a new variant of successive quadratic programming methods with inequality and equality constraint. In calculation process, this methods need only to solve a quadratic programming problem at each iteration. In certain conditions, the global and fast local convergence properties of this algorithm is proved, whose step size of unity is acceptable and whose Maratos effect is obviated.