Convergence dynamics and pseudo almost periodicity of a class of nonautonomous RFDEs with applications (Q2566787)

The authors study the asymptotic behaviour of the solutions of the following system of retarded differential equations with finite delay \[ \dot{x}_i(t)=-b_ix_i(t)+f_i(t,x_1(t),\ldots,x_n(t),x_1(t-\tau_{i1}(t)), \ldots,x_n(t-\tau_{in}(t))) +I_i(t), \] \(i=1,\ldots,n, \) where all functions are bounded and continuous, \(b_i(t)>0\), \(0\leq\tau_{ij}(t)\leq\tau\), \(\tau=const>0\), and \(f_i\) satisfies a global Lipschitz condition. The system incorporates the Hopfield-type neural network and the bidirectional associative memory neural network as special cases. The system is said to be globally exponentially stable, if for any two solutions \(x(t;t_0,\varphi)\) and \(y(t;t_0,\psi)\) there exist positive numbers \(M\) and \(\varepsilon\) such that \( | x(t;t_0,\varphi)-y(t;t_0,\psi)| _p\leq M\exp[-\varepsilon(t-t_0)]\| \varphi-\psi\| _p,\;\;t>t_0. \) Delay-dependent and delay-independent sufficient conditions for the global exponential stability of the system are obtained. A delay-independent criterion for existence and uniqueness of a pseudo almost-periodic solution is established. A concept of the pseudo almost-periodicity has been introduced by \textit{C.~Zhang} [J. Math. Anal. Appl. 181, 62--76 (1994; Zbl 0796.34029)]. Some examples arising from neural networks and population ecology are considered as applications of the new results. The paper ends by concluding remarks and numerical simulations concerning pseudo almost-periodic solutions to a Hopfield neural network with two neurons.