Peter Kassakian
Convex Approximation and Optimization with Applications in Magnitude Filter Design and Radiation Pattern Synthesis

Peter writes: In this talk I will present two recent results from my thesis. The first is a
method for designing audio equalizers using convex optimization. I will
present the method emphasizing its flexibility and the convenience gained by
using convex optimization in general. The second result is more abstract, and
is related to solving nonconvex problems by approximating them as convex.
Though such techniques can yield high-quality solutions, they often do not
find the global optimum. I will present a novel result that exploits
complex-value structure in a nonconvex quadratic problem to prove that the
solution accuracy is better than that of a real-valued counterpart. The proof
is simple and self-contained, providing insight into the geometry of these
problems.

Add to iCal
Find on Google Maps
Share

Monday, May 8, 2006, 8:00pm to 10:00pm