Monotonicity and Gradient Algorithms
Iterative gradient-based algorithms for ensuring the consistent reduction of error norms from iteration to iteration are described. Gradients are identified with adjoint operators which take a simple form for state space systems. The monotonicity and robustness of the algorithms for MIMO, discrete, state space systems is characterized by frequency domain conditions. The applicability of the results is extended using \(\varepsilon \)-weighted norms.