An algebraic computational approach to the identifiability of Fourier models

Computer algebra and in particular Grobner bases are powerful tools in experimental design (Pistone and Wynn, 1996, Biometrika 83, 653-666). This paper applies this algebraic methodology to the identifiability of Fourier models. The choice of the class of trigonometric models forces one to deal with complex entities and algebraic irrational numbers. By means of standard techniques we have implemented a version of the Buchberger algorithm that computes Grobner bases over the complex rational numbers and other simple algebraic extensions of the rational numbers. Some examples are fully carried out. (C) 1998 Academic Press.