A Novel Locally Active Memristor and Its Application in Neuron Coupling Synchronization

As a nonvolatile and nonlinear circuit element, memristors can be used to simulate neural synapses. Now that the local activity is an origin of complexity, a locally active memristor can establish complex artificial neural networks to explore brain-like computing. In this study, a proposed memristor has nonvolatile and local activity confirmed by a Power-Off-Plot (POP) and a DC v–i plot, with bistable hysteresis loops. It is used to simulate neural synapses to couple two different classical neuronal models. This small neural network induced by the memristor exhibits complex dynamic behaviors such as coexistence bifurcations, anti-periodic bifurcations, free parameter switching, coexistence firing, chaos, periodicity, and so on. Further, it is observed that two neurons achieve a firing synchronous and verified by phase difference and synchronization error. In addition, the neuronal burst mode gradually disappears as the parameter increases.