An Overview of Approximation Methods for Control of PDEs

Tuesday, March 15, 2016 - 2:00pm - 3:00pm
Keller 3-180
With all the algorithmic advancements in forward PDE approximations, development of proper approximations for control and optimization of PDES remains quite challenging. These approximations should preserve system-theoretic properties of the systems such as detectability, stablizability, identifiability, etc., which add to the convergence requirements of the approximation schemes which are used for forward simulation. If there is no early investment on the choice of the right approximation, then there will more time spent later for the controller design of a model which may not capture all the salient features/dynamics of the original problem. The perils and pitfalls of choice of wrong approximation methods for the controlled problem will be illustrated by examples.