Information-based complexity and nonparamteric worst-case system identification (Q1330145)

The paper is a survey of nonparametric approaches to identification of linear dynamic systems under nonprobabilistic assumptions on uncertains. The authors, first, introduce definitions and basic results of the information-based complexity (IBC) theory, used in the paper. Nonparametric approaches developed in the last few years can be classified in two main categories, according to the norm in which the estimation errors are measured: \(H_ \infty\) and \(l_ 1\) methods. The \(H_ \infty\) setting assumes that the true system is linear time- invariant and the available information is represented by samples of frequence response of the system, corrupted by an \(l_ \infty\)-norm bound noise. The aim is to estimate a proper, stable finite-dimensional model. In the \(l_ 1\) setting, the aim is to identify the samples of the impulse of an unknown linear time-invariant system. They provide recent results on \(H_ \infty\), \(l_ 1\), and \(l_ \infty\) identification, with particular attention to (a) evaluation of the diameter of information, (b) convergence to zero of the diameter of information, and (c) computation of optimal and almost-optimal algorithms.