Abstract: In this talk, I will introduce a computing paradigm that exploits the natural tendency of a network of coupled oscillators to settle to the ground state. Our first-of-its kind test chip with almost 2,000 coupled oscillators implemented in a standard 65nm technology shows that solutions to NP-hard problems such as max-cut can be found with higher efficiency than traditional digital computation. Extensive testing shows that coupled oscillators can probabilistically explore the energy landscape to find a good minima point. This has implications on a wide range of intractable problems that can be mapped to a quadratic unconstrained binary cost function.
Bio: Chris H. Kim is the Louis John Schnell professor in Electrical and Computer Engineering at the University of Minnesota. He is the recipient of the Distinguished Mcknight University Professorship, UMN Taylor Award for Distinguished Research, SRC Technical Excellence Award, NSF CAREER Award, and Mcknight Foundation Land-Grant Professorship. His group has expertise in digital, mixed-signal, and memory IC design, with special emphasis on quantum-inspired solvers, circuit reliability, hardware security, memory circuits, radiation effects, time-based circuits, beyond-CMOS technologies, and machine learning hardware. He is an IEEE fellow.