混合互补问题牛顿型算法的二阶收敛性

在凸规划理论中，通过ＫＴ条件，往往将约束最优化问题归结为一个混合互补问题来求解。该文就正则解和一般解两种情形分别给出了求解混合互补问题牛顿型算法的二阶收敛性的充分性条件，并在一定条件下证明了非精确牛顿法和离散牛顿法所具有的二阶收敛性。

Quadratic Convergence of Newton-type Algorithms for Mix-complementarity Problems

In convex programming theory, a constrained optimization problem, by KT conditions, is usually converted into a mixed nonlinear complementarity problem. Accor-ding to regular solution and general solution, we in the paper describe and establish a sufficient condition under which the Newton-type algorithm possesses quadratic convergence property when it is applied to solving mix-complementarity problems. In addition, we also show that, when the step-size is suitably chosen, the inexact Newton's method and the discrete Newton's method converge quadratically.