Distinguished Lecture Series

4:00 pm on Wednesday, March 19, 2014
5:00 pm on Wednesday, March 19, 2014
Contact Name:
Fed Chavez
The College of Engineering Distinguished Lecture Series celebrates high impact research in engineering and annually honors one of our faculty engaged in outstanding research. This showcasing event will be a public forum allowing all members of the University community to meet a distinguished scholar selected from College of Engineering faculty discussing a topic of recognized excellence. This year's Distinguished Scholar and current award recipient is Professor Christos G. Cassandras. TITLE: "Complexity Made Simple* *at a Small Price" ABSTRACT: " There are some fundamental complexity limits that provide a starting point for any effort to solve complex problems. When it comes to the design, control, and optimization of complex dynamic systems, the goal of this lecture is to show that there are ways to solve many hard problems by exploiting their specific structure, by asking the “right” questions, and by challenging some conventional engineering approaches. First, trial-and-error techniques are often used to systematically learn and predict the behavior of a complex system. These are invariably slow, inefficient, and intrusive (since one has to disrupt the system with a “trial” in order to learn its effect…) I will describe how this learning can sometimes be accomplished at a fraction of the usual brute-force trial-and-error process through simple “thought experiments” constructed at a “small price.” I will include some glimpses at the foundations of the theory on which this approach is built and illustrate it with specific applications. Second, it is known that decomposition and abstraction methods can sometimes provide accurate solutions or significantly simplify a hard problem at the “small price” of some loss of accuracy. I will discuss how these methods can be used in large distributed systems and whether this “small price” can be quantified. Finally, when it comes to today’s wireless, networked world, conventional time-driven methods for sampling, control, and communication should be challenged. For example, communication actions dictated solely by a “clock” drain precious battery power, exacerbate security risks, and are often unnecessary; they may be instead event-driven at the “small price” of identifying the proper events triggering these actions. In this vein, an interesting question is “what is the least amount of communication required within a team of cooperating agents to achieve a given goal?” I will address this question for a class of problems where the goal is provably met while saving energy and enhancing security."