Electro-Photo-Sensitive Memristor for Neuromorphic and Arithmetic Computing

Memristors ``remember'' how much current has flowed through them, and are a promising means to combine information processing and nonvolatile storage on the same physical platform. They are also sought for $n{{0}{0ex}}e{{0}{0ex}}u{{0}{0ex}}r{{0}{0ex}}o{{0}{0ex}}m{{0}{0ex}}o{{0}{0ex}}r{{0}{0ex}}p{{0}{0ex}}h{{0}{0ex}}i{{0}{0ex}}c$ hardware, in which neuronal responses to input are mimicked, with the goal of attaining some of the brain's computing strengths. A memristor based on $I{{0}{0ex}}I{{0}{0ex}}I-V$ semiconductor quantum dots shows spike-timing-dependent plasticity, the key characteristic for learning and memory in neural networks. Furthermore, applying low-power light pulses allows the system to perform arithmetic operations, under optoelectronic control.