Neural Network Approach for Nonlinear Complementarity Problem and Quadratic Programming with Second-Order Cone Constraints

No Thumbnail Available



Journal Title

Journal ISSN

Volume Title



This dissertation focuses on two types of optimization problems, nonlinear complementarity problem (NCP for short) and quadratic programming with second-order cone constraints (SOCQP for short). Based on NCP-function and SOC-complementarity function, we propose suitable neural networks for each of them, respectively. For the NCP-function, we propose new one which is the generalization of natural residual function for NCP. It is a discrete generalization of natural residual function phinr, denoted as phinrp. Besides being a NCP-function, we also show its twice dierentiability and present the geometric view. In addition, we utilize neural network approach to solving nonlinear complementarity problems and quadratic programming problems with second-order cone constraints. By building neural networks based on dierent families of smooth NCP or SOCCP-functions. Our goal is to study the stability of the equilibrium with respect to dierent neural network models. Asymptotical stability are built in most neural network models. Under suitable conditions, we show the equilibrium being exponentially stable. Finally, the simulation results are reported to demonstrate the effectiveness of the proposed neural network.



Nonlinear Complementarity Problem, Second-Order Cone, Neural Network, Nonlinear Complementarity Problem, Second-Order Cone, Neural Network