“…With the feature of inherent variability, RRAM shows great potential to be used as low-cost and energy-efficient stochastic number generator enabling stochastic computing, which emulates the generation of neural spikes processed by the human brain in the form of long sequences of noisy voltage spikes (Gaba et al, 2013;Suri et al, 2013;Knag et al, 2014;Moons and Verhelst, 2014;Ielmini and Wong, 2018;Wang et al, 2018;Carboni and Ielmini, 2019;Zhao et al, 2019). For the design and optimization of these brain-inspired computing systems, related physics-based models and simulation platforms have been developed to bridge the link between device, circuit, and system, which aims to meet the requirement for the device-circuitsystem co-design (Gao et al, 2011;Guan et al, 2012;Huang et al, 2013Huang et al, , 2017Huang et al, , 2018Chen et al, 2017;Larcher et al, 2017;Pedretti et al, 2017;Zhao et al, 2019;Cai et al, 2020;Liao et al, 2020).…”