Shifted skew-symmetric iteration methods for nonsymmetric linear complementarity problems (Q5962123)

The author considers the linear complementarity problem consisting in finding element \(z\in\mathbb{R}^n\) such that \[ Mz+ q\geq 0,\quad z\geq 0,\;z^T(Mz+ q)= 0, \] where \(M\) is a given, large, sparse \((n\times n)\)-matrix, and \(q\in\mathbb{R}^n\) is a given vector. Matrix \(M\) can be splitted into a sum of two matrices \(H\), \(S\) such that \(H\) is symmetric and S is skew-symmetric. Using this splitting so called shifted symmetric and shifted skew-symmetric splitting of matrix Al is introduced. By making use of the shifted splitting of \(M\) two iteration methods for solving the linear complementarity problem are proposed: 1) Shifted skew-symmetric iteration method, 2) Inexact skew-symmetric iteration method. Conditions of convergence of the proposed methods are presented and the corresponding convergence theorems are proved. Numerical experience with the method shows that both methods are superior to the well-known Lemke's method.