SCO Seminar - Robin Bulsink, University of Groningen

Title: Applications of conic programming duality to systems and control

Abstract:

Conic programming is the optimization of an inner product subject to feasibility constraints, in terms of inclusion in a cone. Inspired by the rich existing duality theory in semidefinite programming (SDP), we develop a theory of duality in the conic programming context. We find that this methodology both generalizes SDP and features simpler, more elegant proofs than those in SDP. Finally, it is shown how this duality theory can be applied to study notable problems in systems and control theory; specifically solutions Lyapunov inequalities and the (infinite-horizon) linear quadratic regulator (LQR) problem.