*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Title: Computing with Dynamical Systems: Bridging Device Technologies with Algorithms
Committee:
Dr. Raychowdhury, Advisor
Dr. Yalamanchili, Chair
Dr. Romberg
Abstract:
The objective of the proposed research is to create alternative computing models and architectures, other than traditional Turing machine or Von Neumann style models, which utilize the continuous time dynamics, deterministic or stochastic, of novel beyond-CMOS devices to solve hard computing problems like optimizations. These new beyond-CMOS devices have physical dynamics which can be used to emulate much higher level functions than merely switches and logic gates. Another aspect of computing power obtained through such devices is noise induced stochasticity. More importantly, the dynamics of a network of such interconnected and coupled devices can be much more complex, even more than arithmetic operations that form the basis of Turing machine models. In these cases, the computing models need to be modified or recreated to utilize such computing abilities of devices and their networks. Dynamical systems that result from networks of interacting nodes, like oscillators and neurons, can have emergent properties with similarities and analogies to various non-heuristic algorithms for hard optimization problems. As such, a constructive way towards analyzing and creating such computing models is to study those connections between algorithms and dynamical systems that are possible by designing circuits and architectures using devices. Better device-architecture-algorithm co-design can be enabled by understanding and generalizing these connections, which in-turn will provide better research targets for new devices to be used in unconventional platforms that run physics-based algorithms.
