A frequency domain approach is adopted in this paper to tackle the problem of validating uncertainty models described by linear fractional transforms. It is shown that the validation problem reduces to one of Nevanlinna-Pick boundary interpolation, and it can be solved by solving independently a sequence of convex programs defined with respect to each sampling frequency. In comparison to previously available algorithms hased on time domain information, the main advantage of these tests is that they have a considerably lower level of computational complexity.
Read full abstract