针对带约束的非线性规划问题,提出一个修正共轭梯度投影算法,并且用不严格互补条件证明了算法具备全局收敛性和局部超线性收敛性;另一方面,算法的每步迭代只计算一次共轭投影矩阵,避免了求解二次规划或求两个投影矩阵,因而算法在计算量上有所改进.
In this paper, a revised conjugate projection gradient method for nonlinear inequality constrained optimization problem is proposed. The author proves that the algorithm is super-linearly without strict complementary condition under some mild assumption. On the other hand, the amount of computation is fewer since the gradient matrix only needs to be computed once in each iteration.