Numerical computation of minimal polynomial bases: a generalized resultant approach (Q2484243)

A new resultant based method for the computation of minimal polynomial bases of a polynomial matrix is proposed. The algorithm utilizes the left null space structure of successive generalized Sylvester resultants of a polynomial matrix to obtain the coefficients of the minimal polynomial bases of the left kernel of the given polynomial matrix. The entire procedure can be implemented using only orthogonal decompositions and the coefficients of the resulting minimal polynomial basis have the appealing property of being orthonomal.