The probabilistic behavior of the set and reset thresholds in Knowm's SDC memristors: Characterization and Simulation

This paper presents a proposal for the characterization of the set and reset thresholds for Knowms SDC memristors. The purpose is to incorporate the variability of the hysteresis cycles within the Generalized Mean Metastable Switch (GMMS) memristor model and, in this way, be able to perform simulations that reproduce these phenomena in a meaningful and computationally efficient way. We depart from the assumption that their probabilistic behavior can be well represented by using -stable random variables. The main advantage of using -stable variables is that they can capture both skewness and high variability (i.e., heavy tails), which can be exhibited by the observed phenomenon. At the same time, they also include the Gaussian random variable as a particular case, thus increasing the modeling flexibility.