Positive bidiagonal factorization of tetradiagonal Hessenberg matrices (Q6051136): Difference between revisions

The authors study the tetradiagonal Hessenberg matrix of the form \[ T = \begin{bmatrix} c_0 & 1 & 0 & \dots & \dots & \dots\\ b_1 & c_1 & 1& \ddots &&\\ a_2 & b_2 & c_2 & 1 & \ddots & &\\ 0 & a_3 & b_3 & c_3 & 1 &\ddots&\\ \vdots & \ddots & \ddots & \ddots & \ddots & \ddots\\ \vdots && \ddots & \ddots & \ddots & \ddots \end{bmatrix}, \] where \(a_n > 0\). They study the existence of a positive bidiagonal factorization (PBF) \(T = L_1 L_2 U\), where \(L_1\), \(L_2\), and \(U\) are bidiagonal matrices. They first summarize the theory pertaining to the tridiagonal case, emphasizing that a suitably shifted Jacobi matrix, denoted as \(\mathbf{J} + s \mathbf{I}\), exhibits oscillatory behavior. It is established that \(\mathbf{J}\) is oscillatory if and only if it possesses PBF properties. Then the authors use truncations and continued fractions to obtain their main results. They prove the existence of a PBF for the tetradiagonal matrix in the finite-dimensional case. For the infinite-dimensional case, the authors prescribe a boundary condition in the presence of PBF, extrapolating the finite PBF to the semi-infinite case. This extrapolation shows the existence of a PBF when a specific nonnegative infinite continued fraction is unequivocally positive. Furthermore, they carry out an analysis of the oscillatory Toeplitz matrices, establishing their capability to admit PBF. Finally, they prove that oscillatory banded Hessenberg matrices are organized in rays, where the origin of the ray does not have a PBF, but all the interior points of the ray do possess a PBF.