A predictor-corrector smoothing Newton method for symmetric cone complementarity problems (Q613219)

The authors consider the following symmetric cone complementarity problem: Find \(x\in J\) such that \(x\in K\), \(F(x)\in K\), \(\langle x,F(x)\rangle= 0\), where \(J\) is a finite-dimensional vector space over the real field endowed with inner product \(\langle.,.\rangle\), \(K\) is a symmetric cone in \(J\), \(F: J\to J\) is a continuously differentiable nonlinear operator. Such problems have wide applications e.g. in engineering and management science and provide a unified framework for various complementarity problems known from the literature. The authors show that problems having the form considered above are solvable under mild assumptions. An algorithm for solving such problems is proposed. Both local and global convergence of the proposed algorithm is proved. The effectivity of the proposed method is demonstrated on numerical examples in the concluding part of the paper.