Eigenstructure of order-one-quasiseparable matrices. Three-term and two-term recurrence relations (Q2484224): Difference between revisions

The paper deals with the computation of eigenvalues and eigenvectors of order-one-quasiseparable matrices, which are defined by \[ r_{ij}=\left\{ \begin{matrix} p_{i}a_{ij}^{\times }q_{j}\;\;\;\;\;1\leq j<i\leq N \\ d_{i}\;\;\;\;\;\;1\leq i=j\leq N \\ g_{i}b_{ij}^{\times }h_{j}\;\; \;\;\;1\leq i<j\leq N \end{matrix}\right. \tag{*} \] where the symbol \(a_{ij}^{\times}=a_{i-1}\cdots a_{j+1}.\) The \(N^{2}\) entries of the matrix are in fact generated by \(p_{i}( i=2,\dots,N) \), \( q_{i}( j=1,\dots,N-1) ,\) \(a_{k}( k=2,\dots,N-1) ,\) \( g_{i}( i=1,\dots,N-1) ,\) \(h_{j}( j=2,\dots,N) \), \( b_{k}( k=2,\dots,N-1) ,\) and \(d_{k}( k=1,\dots,N) .\) From the structure of the matrix various, two-term and three-term, recurrence relations for the characteristic polynomials follow and these yields fast Newton type iteration algorithms. \newline Sufficient conditions for an eigenvalue to be simple and explicit formulae for the corresponding eigenvector are obtained.