LMI Representation of Convex Sets

Friday, January 19, 2007 - 2:30pm - 3:20pm
EE/CS 3-180
Victor Vinnikov (Ben Gurion University of the Negev)
I will discuss the characterization of convex sets in m
which can be represented by Linear Matrix Inequalities, i.e., as feasible
sets of semidefinite programmes. There is a simple necessary condition,
called rigid convexity, which has been shown to be sufficient for sets in
the plane and is conjectured to be sufficient (in a somewhat weakened
sense) for any m.

This should be contrasted with the situation for matrix convex sets that
will feature in the talk of Scott McCullough, where all the available
evidence suggests that any matrix convex set with noncommutative algebraic
boundary admits an LMI representation.

This is a joint work with Bill Helton.
MSC Code: