Global stability for discrete Cohen-Grossberg neural networks with finite and infinite delays (Q714573): Difference between revisions

The authors consider a general discrete Cohen-Grossberg neural network (DCGNN) with finite and infinite delays: \[ \begin{multlined} u^{(i)}_{n+1}=u^{(i)}_{n}-a_i\left(u^{(i)}_{n}\right)\\ \times\left[b_i\left(u^{(i)}_{n}\right)-\sum_{j=1}^lc_{ij}f_j\left(u^{(j)}_{n}\right)- \sum_{j=1}^ld_{ij}f_j\left(u^{(j)}_{n-\nu_{ij}}\right)\right.\\ \left. -\sum_{j=1}^le_{ij}f_j\left(\sum_{s=1}^{\infty}K_{ij}(s)u^{(j)}_{n-s}\right)+J_i\right], \;i\in\{ 1, 2,\dots,I\}, \end{multlined} \] where \(l\geq2\) is the number of neurons in the network; \(u_n^{(i)}\) denotes the state associated with the \(i\)-th neuron; \(a_i(u)\) represents the amplification function; \(b_i(u)\) is called an appropriately behaved function; \(c_{ij}\), \(d_{ij}\), \(e_{ij}\) denote the synaptic connection weights of unit \(j\) to unit \(i\); \(f_i(\cdot)\) is a measure of response or activation to its incoming potential; \(\nu_{ij}\) means the transmission delay along the axon of unit \(j\) to unit \(i\); \(K_{ij}(x)\) corresponds to the delay kernel and \(J_i\) is the constant external input. Sufficient conditions on the global stability for the DCGNN are obtained by applying Lyapunov' direct method and a new method on graph theory, introduced by \textit{M.~Y.~Li} and \textit{Z.~Shuai} [J. Differ. Equations, 248, No. 1, 1--20 (2010; Zbl 1190.34063)]. To show the effectiveness of the proved result an illustrative example is given.