ABSTRACT
First principle models of physical systems are often represented by state-space descriptions in which the various components of the state represent well-defined physical quantities. Variations, perturbations, or uncertainties in physical parameters lead to uncertainty in the model. Often, this uncertainty is reflected by variations in well-distinguished parameters or coefficients in the model, while, in addition, the nature and/or range of the uncertain parameters may be known, or partially known. Since very small parameter variations may have a major impact on the dynamics of a system, it is of evident importance to analyze parametric uncertainties of dynamical systems. Suppose that δ = (δ1, . . . , δp) is the vector that expresses the ensemble of all uncertain quantities in a given dynamical system. Then there are at least two distinct cases that are of independent interest:
a. Time-invariant parametric uncertainties: the vector δ is a fixed but unknown element of an uncertainty set δ ⊆ Rp.
