求解二阶锥互补问题的预估校正算法

Predictor-Corrector Method for Second-order Cone Complementarity Problem

运用一般内点算法求解二阶锥互补问题时算法性能易受初始点选取影响,文中基于一个新的对称扰动光滑函数,在光滑化牛顿算法的基础上引入预估校正步,给出了求解二阶锥互补问题的预估校正算法.结果表明:该算法不依赖于初始点的选取,且不需要额外运算就能使算法产生的迭代序列保持在给定邻域内.

The interior-point algorithm can be used to solve the secona-oraer cone comptementanty problem,but its performance is affected by the choice of the initial point. Based on a new perturbed symmetrically smoothing function, a predictor-corrector smoothing method is introduced to solve the second-order cone complementarity problem. The new algorithm can keep the iteration sequence in the given neighborhood without depending on the selection of the initial point and with no additional computations.