Systems Seminar

A New Approach to Computational Complexity Problems Associated with Robustness

Prof. B. R. Barmish
UW ECE Dept.

Abstract

Many robustness problems can be formulated in terms of a a design vector x, a vector of uncertain parameters q with bounding rectangle Q and a specification described by an nonlinear inequality F(x,q) < 0. The objective of robust design is to find a feasible vector x such that F(x,q) < 0 is satisfied for all q in Q.

The motivation for this seminar is derived from the fact that large classes of such problems are known to have high computational complexity --- they frequently become impractical to solve as the dimension of q increases. This seminar describes a new approach to this computational complexity problem. To this end, a new concept called "approximate feasibility" is seen to play a central role.

Time and Place: Wed., Oct. 27, 3:30-4:30 pm in 4610 Engr. Hall.

SYSTEMS SEMINAR WEB PAGE: http://www.cae.wisc.edu/~gubner/seminar/