Almost periodic oscillation of octonion-valued neural networks with delays on time scales

Abstract

In this paper, we propose an octonion-valued neural network model governed by dynamic equations, and study the existence and stability of its almost periodic solutions by employing the fixed point method and the time scale calculus theory. The method we use is the direct one, that is, we do not split the considered systems into real-valued ones, but directly study octonion-valued systems. Even when the time scale T=R, in other words, even if we consider continuous time real-valued systems, our results are novel. Finally, we illustrate the effectiveness of our results through a numerical example and computer simulation.