We propose a DC (Difference of Convex fonctions) programming approach for solving asymmetric eigenvalue complementarity problem (EiCP). This problem is equivalent to a Nonlinear Program (NLP) that minimizing a nonconvex polynomial merit function on a convex set defined by linear constraints. EiCP has a solution if and only if NLP has zero global optimal value. We will reformulate the NLP as a DC Program (DCP) and then solve the later one by an efficient DC Algorithm (DCA). Some preliminary numerical results are also reported.
展开▼
