Low-power-consumption quantum analog Ising machine based on overdamped bistability with stochastic resonance effect

Gain-dissipative Ising machine (GIMs) are dedicated devices that utilizing the interplay of linear gain dynamics with a symmetric bistability to rapidly solve combinatorial optimization problems. However, to prevent the spin state from being switched randomly by thermal noise, the saturated fixed-point amplitude of traditional GIMs should be significantly larger than the environmental noise intensity, which results in a relatively high-power consumption. To break through the existing limitation on power consumption, this work proposes an overdamped bistable stochastic resonance unit (OBSRU) as the nonlinearity for constructing the GIM. The numerical simulation shows that OBSRU can correctly simulate the symmetric bistability of isolated spins. In addition, domain clustering dynamics benchmark shows that the stochastic resonance characteristics of the OBSRU enable GIM to effectively suppress noise-induced random spin state switching; therefore, it can work normally with small amplitude in the environment with relatively high noise level. Some prevalent MAXCUT problems, such as Moebius ladder and G-set graphs, are also used as benchmarks to evaluate the performance of OBSRU-based GIM in a noisy environment. The results show that compared with the traditional GIM, the OBSRU-based GIM can achieve the comparable accuracy with significantly lower power consumption. The results reveal that the proposed OBSRU-based GIM provides a promising architecture for low-power-consumption GIM.