An interior-point algorithm for P∗(κ)-linear complementarity problem based on a new trigonometric kernel function

In this paper, an interior-point algorithm for P∗(κ)-Linear Complementarity Problem (LCP) based on a new parametric trigonometric kernel function is proposed. By applying strictly feasible starting point condition and using some simple analysis tools, we prove that our algorithm has O((1κ)nlog⁡nlog⁡nϵ) iteration bound for large-update methods, which coincides with the best known complexity bound. Moreover, numerical results confirm that our new proposed kernel function is doing well in practice in comparison with some existing kernel functions in the literature.