12 June 2009
Eduardo Sontag
Department of Mathematics and BioMAPS, Rutgers
(on leave at MIT for AY 2008/2009)
For systems made up of interconnected components, it is usually desirable to be able to deduce global behaviors through a bottom-up analysis, based on the knowledge of the behaviors of the individual components as well as the interconnection structure. Two potential sources of difficulty in such a modular approach are (a) impedance or "dynamic retroactivity" effects due to resource sharing, and (b) feedback loops that expose modes of behavior that were "hidden" when individual subsystems had been studied in isolation. I'll make some remarks on the subject, but rather than a seminar talk describing results, I would like to use this chance to generate an informal discussion about the topic.