This paper presents a global error bound for the projected gradient by using the value function,which is appeared in sequential quadratic programming(SQP) method.Further,by the error bound,we study the convergence of a feasible solution sequence.%文章利用序列二次规划(SQP)方法中的价值函数为约束最优化问题的投影梯度提供了一个全局误差界,并利用这个全局误差界给出了可行解点列具有收敛性的充分与必要条件.
展开▼