Programmable Fading Memory in Atomic Switch Systems for Error Checking Applications

Disruptive technology in computational devices is required as the universal computing machines approach quantum mechanical limits. Integration of state-of-the-art memristive devices provides optimal scaling of current technologies beyond this limit through the adoption of neuromorphic models. Universal computing machines pioneered by Alan Turing are strictly based on top-down intelligent design. Neuromorphic models instead engage in bottom-up programmability by emulating mammalian brain design and characteristics. Here we show the design, characterization, and implementation of a massively parallel memristor neuromorphic network based on metal chalcogenide atomic switch network (ASN) systems with key characteristics such as short- and long-term potentiation, power-law dynamics, and scale-free topology.