本文研究了线性互补问题内点算法.利用全牛顿步长求解迭代方向,获得了算法迭代复杂性为O(nlogn/ε),推广了Roos等关于线性规划问题不可行内点算法,其复杂性与目前最好的不可行内点算法复杂性一致.
This paper proposes an infeasible interior-point algorithm with full-Newton step for linear complementarity problem, which is an extension of Roos' results on linear optimization. At last, we prove that the algorithm has O(n log n/ε) polynomial complexity, which coincides with the best known one for the infeasible interior-point Mgorithm at present.