No label defined (Q4037006)

[For the entire collection see Zbl 0755.00019.] Two square matrices \(A\) and \(B\) are called algebraically shift equivalent if there exist rectangular matrices \(X\) and \(Y\) such that \(XA=BX\), \(AY=YB\), \(YX=A^ \ell\), \(XY=B^ \ell\) for some integer \(\ell\). The following theorem is proved: Let \(S\) be a principal ideal domain all of whose proper factor rings are finite. Let \(\alpha\) run over some index set. Let \(B\) and \(B_ \alpha\) be square matrices with characteristic polynomial \(f\); let \(C_ \alpha\) and \(C_ \alpha'\) be square matrices with characteristic polynomial \(g\), where \(f\) and \(g\) have no common divisors; let \(X_ \alpha\) and \(X_ \alpha'\) be rectangular matrices, all of these subject to the constraint \[ \left[ \begin{matrix} C_ \alpha & X_ \alpha \\ 0 & B\end{matrix}\right] \sim \left[ \begin{matrix} B_ \alpha & X_ \alpha'\\ 0 & C_ \alpha'\end{matrix} \right], \] where \(\sim\) denotes either algebraic shift or conjugacy with respect to \(S\). Then only finitely many equivalence classes appear among the \(B_ \alpha\) satisfying these constraints.