Convex optimization for robust feedback controller design

This paper deals with the Boyd-Barratt paradigm for feedback controller design. This approach combines the Youla parameterization with convex optimization. In this paper, this paradigm is accepted in its entirety, but a completely different numerical approach is adopted. We propose replacing the descent methods of Boyd and Barratt by an algorithm due to Akilov and Rubinov. This alternative approach avoids the need to compute complicated derivatives or subgradients. Instead, certain linear functionals must be computed, and this is a straightforward task. As a result, an attractive feature of the proposed approach is that the code is much shorter and more elegant. Further advantages of the approach are described, and an example is included.