Factorization of the matrix polynomial \(A(\lambda)=A_ 0\lambda ^ t+A_ 1\lambda ^{t-1}+\dots +A_{t-1}\lambda +A_ t\) (Q1123953)

For a matrix polynomial \(A(\lambda)=\sum^{t}_{j=0}A_{t-j}\lambda^ j\), where \(A_ i\) are \(n\times n\) matrices over a (commutative) field F, define its companion pencil \[ C(\lambda) = \left[ \begin{matrix} I & 0 & \ldots & 0 & 0 \\ 0 & I & \ldots & 0 & 0 \\ \vdots & \vdots && \vdots & \vdots \\ \vdots & \vdots & & I & 0 \\ 0 & 0 & \ldots & 0 & A_0 \end{matrix} \right] \lambda - \left[ \begin{matrix} 0 & I & 0 & \ldots & 0 \\ 0 & 0 & I & \ldots & 0 \\ \vdots & \vdots & & & \vdots \\ \vdots & \vdots & & & -I \\ -A_ t & -A_{t-1} & \ldots & \ldots & -A_1 \end{matrix} \right] \] Main result: A(\(\lambda)\) admits complete factorization of the form \(A(\lambda)=(A_ 0\lambda -L_ t)(I_ n\lambda -L_{t-1})...(I_ n\lambda -L_ 1)\) if and only if C(\(\lambda)\) is strictly equivalent to the bidiagonal pencil \[ L(\lambda)=\left[ \begin{matrix} I\lambda-L_1 & -I & 0 & \ldots & 0 \\ 0 & I\lambda-L_2 & -I & \ldots & 0 \\ \vdots & \vdots & \vdots && \vdots \\ \vdots & \vdots & \vdots && -I \\ 0 & 0 & 0 & \ldots & A_0\lambda-L_ t \end{matrix} \right] \] and, moreover, the constant matrices \(P\) and \(Q\) satisfying \(PC(\lambda)Q=L(\lambda)\) can be chosen block lower triangular with \(I\)'s on the main diagonal.