Answers are grounded in this paper and cite their sources.
Existing hardware implementations of Gaussian random number generators (RNGs) rely on large arrays of binary devices, resulting in substantial hardware footprints or sampling-time overheads. Here, we present a racetrack memory-based random number generator (RTM-RNG) that directly generates analog Gaussian random number streams from a single domain-wall-based nanoscale racetrack. A geometrically engineered racetrack tip creates a localized current-density gradient, enabling field-free stochastic domain-wall nucleation driven by nanosecond current pulses. The domain-wall nucleation probability is continuously tunable via an electrically generated Oersted field or an external magnetic field, providing direct control over the mean of the output distribution. We demonstrate more than 256 distinguishable states from a single nanoscopic RTM-RNG device, thus making a Gaussian RNG with more than 8-bit resolution possible. Compared with conventional binary random bitstreams, the RTM-RNG achieves high-quality Gaussian distributions up to 83 times faster.
No takes yet. Share an insight, caveat, or question.
Arnold et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: