研究了单调线性互补问题的一种内点法,将牛顿方向和中心路径方向相结合,通过求解一个线性方程组得到搜索方向;在每次迭代中,寻找使得新的迭代点满足可行性要求且同时使得势函数值下降的步长参数,进而建立了求解单调线性互补问题的一种势下降内点算法,并证明该算法经过多项式次迭代之后收敛到原问题的一个最优解,数值实验表明此方法是有效的。
Combined with Newton direction and centering direction,search direction was obtained by solving a linear system.In every iteration,step parameter which made new points feasible and values of potential function decreased was acquired.Then potential-reduction interior point algorithm for monotone linear complementarity problem was established and its global convergence was proved.Numerical results were reported which demonstrates very good computational performance on monotone linear complementarity problem.