Global asymptotic dynamics of a class of nonlinearly coupled neural networks with delays

Abstract

This work presents an effective approach to the study of the global asymptotic dynamics of general coupled systems. Under the developed framework, the problem of establishing global synchronization or global convergence reduces to solving a corresponding system of linear equations. We illustrate this approach with a class of neural networks that consist of a pair of sub-networks under various types of nonlinear and delayed couplings. We study both the synchronization and the asymptotic synchronous phases of the dynamics, including global convergence to zero, global convergence to multiple synchronous equilibria, and global synchronization with nontrivial synchronous periodic solutions. Our investigation also provides theoretical support to some numerical findings, and improves or extend some results in the literature.