Analysis of Skyrmion Shuffling Chamber Stochasticity for Neuromorphic Computing Applications

Abstract

In this study, micromagnetic simulations of a magnetic skyrmion reshuffling chamber for probabilistic computing applications are performed. The skyrmion shuffling chamber is modeled with a custom current density masking technique to capture current density variation, grain boundary variations, and anisotropy changes. The results show that the skyrmion oscillatory dynamics contribute to the system's stochasticity, allowing uncorrelated signals to be achieved with a single chamber. Our findings indicate that uncorrelated signals are generally achieved at all temperatures simulated, with the skyrmion diameter playing a role in the resulting stochasticity. Furthermore, we find that local temperature control has the benefit of not affecting the overall skyrmion diameter, while still perturbing the skyrmion trajectory. The results from varying chamber size, global temperature, and local temperature are analyzed using Pearson correlation coefficient (PCC) and p-value. This research contributes to the development of tunable probabilistic computing devices and artificial synapses using magnetic skyrmions.