Artificial Neurons using Superconductors
Physics and Astronomy
AbstractOur research focuses on the design, fabrication and testing of integrated circuits which can simulate neuron spiking dynamics on very fast timescales. These circuits are based on a low-temperature, superconducting electronics technology (Josephson junctions) that has already been successful in creating ultra-sensitive magnetometers, high-performance radiation detectors, high-speed digital processors, and the primary voltage standard in the U.S. The short spiking times in these artificial neurons combined with analog scaling properties give this approach a potentially unprecedented ability to investigate long term dynamics of large networks. In addition, these artificial neurons dissipate almost no power, making them a candidate for a low-power, neuromorphic computing technology.
By performing a series of low-temperature electronic measurements and comparing with numerical simulations, the dynamics of several Superconducting networks will be studied. We will study the analog modeling of a neural network made from Josephson junctions, focusing on the behavior of large networks. It will address questions in nonlinear dynamics, pushing toward larger, coupled systems with greater complexity. Finally, it will cross boundaries into the field of computational neuroscience and show how the computational “power” of Josephson junctions can be used to construct a fast, analog model of a neural network. Undergraduate students will be involved in cryogenic, data collection and data analysis. In addition, some students will specialize in either computational studies, electronics instrumentation, machine shop work or noise studies when we are not taking data.
Student QualificationsStudents should have taken Physics 232, 233, 334 and 336 to work on these projects, and should be interested in a career in solid state physics, superconductivity, nonlinear dynamics, electrical engineering, experimental physics or computational studies. All students will be involved in cryogenic, data collection and data analysis. In addition, some students will specialize in either computational studies, electronics instrumentation, machine shop work or noise studies when we are not taking data. The sum will be a nice broad exposure to low-temperature physics, applied physics, nonlinear dynamics and computational neuroscience.
Number of Student Researchers4 students
Project Length8 weeks
Applications open on 01/03/2022 and close on 02/04/2022