The monograph describes a computationally-based methodology for controller
design that can handle typical time and frequency domain specifications and
provide a characterization of the limits of performance for a given system.
The methodology is based on optimization, where one performance objective
is optimized and the other specifications are the constraints. Many practical
performance objectives can be represented in terms of convex constraints on
the closed loop system response. Such objectives include standard norm constraints,
bandwidth constraints, time templates on the closed loop response, as
well as stability robustness conditions for unstructured uncertainty. With this,
the design problem is turned into an infinite dimensional convex optimization
problem. The infinite dimension arises from the fact that the set of feasible
closed loop maps (corresponding to a stabilizing controller) is infinite.