講演情報
[20a-C401-4]Noise-robust logic gate based on double-well Dikandé–Kofané potential
〇(D)Zhiqiang Liao1, Keying Huang1, Siyi Tang1, Hiroyasu Yamahara1, Munetoshi Seki1, Hitoshi Tabata1 (1.Univ. of Tokyo)
キーワード：
logic gate、noise robustness、logical stochastic resonance
Logic gates are essential for performing computations, making decisions, and executing algorithms. However, real-world logic gates often encounter noise and uncertainties that can impact their performance. To ensure accurate information processing and minimize errors, designing noise robust logic gates is important. Moreover, the majority of previous logic gates were electronic. With the advancement of neuromorphic devices, the development of logic gates based on other physical media holds significant value. To address the aforementioned research gap, in this study, we utilize the bistable lattice with Dikandé–Kofanéto (DK) potential to construct noise-robust logic gates. DK potential is mostly generated in one-dimensional system of interacting particles, such as atoms, molecules and ions. In this simulation study, we investigate the performance of DK-based logic gate (DKLG) under three conditions. Specifically, these conditions include a noise-free scenario, a scenario involving Gaussian white noise, and a scenario where the system is influenced by noise pulses. For the latter two cases, we make a comparison with quadstable logic gate (QLG). QLG is a previously studied logic gate with the state-of-the-art noise robustness. The obtained results shed light on the parameter-induced logical stochastic resonance phenomenon in DK-based systems. Notably, in the presence of Gaussian white noise, the noise margin for robust logical operations in DKLG is comparable to that of QLG. Additionally, when discontinuous noise pulses are combined with thermal noise, DKLG demonstrates superior robustness compared to QLG. Our findings underscore the potential benefits of utilizing DK potential derived from materials as noise-robust substitutes for traditional logic gate.