Intermittent and metastable chaos in a memristive artificial neuron with inertia

Inspired by rapid experimental development of diffusive memristors, we propose a computational model of a memristive artificial neuron that takes into consideration inertia of metallic nanoparticles within the dielectric layer of the core-memristor. This model displays rich nonlinear dynamics, which has been speculated to be key for successful emulation of living biological neurons by neuromorphic devices. We found out four characteristic dynamical regimes realized in the system depending on inertness of the nanoparticles. For low-inertia particles, the artificial neuron biased by an applied DC-voltage demonstrates either steady state or regular periodic oscillations. For higher inertia, metastable and intermittent chaos can appear in the system. We analyse the transitions between these regimes and draw parallels between our model and biological neurons.