Title :
On constructing a shortest linear recurrence relation
Author :
Kuijper, Margreet ; Willems, Jan C.
Author_Institution :
Dept. of Electr. & Electron. Eng., Melbourne Univ., Parkville, Vic., Australia
Abstract :
It has been shown in the literature that a formulation of the minimal partial realization problem in terms of exact modeling of a behavior lends itself to an iterative polynomial solution. For the scalar case, we explicitly present such a solution in full detail. Unlike classical solution methods based on Hankel matrices, the algorithm is constructive. It iteratively constructs a partial realization of minimal McMillian degree. The algorithm is known in information theory as the Berlekamp-Massey algorithm and is used for constructing a shortest linear recurrence relation for a finite sequence of numbers.
Keywords :
information theory; iterative methods; linear systems; modelling; optimisation; Berlekamp-Massey algorithm; McMillian degree; behavior modeling; data reduction; information theory; iterative polynomial; linear systems; minimal partial realization; shortest linear recurrence relation;
Journal_Title :
Automatic Control, IEEE Transactions on