Factorizations for \(n\)D polynomial matrices (Q5957407): Difference between revisions

The paper presents a constructive scheme for extracting a nontrivial factor from a given \( n\)D polynomial matrix whose maximal order minors satisfy certain conditions. Let \( F(z)\) be an \(m \times l\) matrix (\(m \geq l\)) with entries in \(C[z_1,\dots,z_n]\), shortly \(C[z]\), \(a_i(z)\), for \(i=1,\dots, b\), \(l \times l\) minors and \(d_0(z)\) a common divisor of \(a_1(z),\dots,a_{\beta}(z)\), i.e. \(a_i(z)= d_0(z) e_i(z)\). \( F(z)\) is said to admit a general matrix factorization if \(F(z)=F_0(z)G_0(z)\), with \( F_0 \in C^{m \times l}[z]\), \(G_0 \in C^{l \times l}[z]\) and \(\text{det} G_0=d_0(z)\). Two special cases of the general matrix factorization are the minor prime factorization (\(m > l\) and \(d_0(z)=\text{gcd}(a_1(z),\dots,a_{\beta}(z))\) and the determinantal factorization (\(m=l\) and \(\beta=1\)). It is shown that the general matrix factorization exists for three classes of \(n\)D polynomial matrices for which \(d_0(z)\) is of a special type (that can be practically tested) and \(d_0(z)\), \(e_1(z),\dots e_n(z)\) have no common zeros. The proposed factorization scheme can be performed in a finite number of steps. These results provide sufficient conditions that are, in general, not necessary and a partial solution. The results can be theoretically generalized to the case of arbitrary \(n\)D polynomial matrices but the practical computation of the factors is still an open problem. Three examples are worked out in detail to illustrate the proposed scheme.