Boundary value problems for second order linear difference equations: application to the computation of the inverse of generalized Jacobi matrices (Q2331739)

Generalized Jacobi matrices are mean square matrices of order \(n+2\) of the form \[ M=\|a_{ij}\|_{i,j=1}^{n+2}, \qquad a_{ij}=0, \quad |i-j|\ge2, \] except for \(4\) entries \(\{a_{1,n}, a_{1,n+1}, a_{n+1,1}, a_{n+1,2}\}\) which can be nonzero. The main problem under consideration is to find conditions when such matrix is invertible, and to compute the inverse. The authors show that the inverse of generalized Jacobi matrices is tightly related to a certain boundary value problem for a second-order linear difference equation associated with the Schrödinger equation. The conditions that ensure the uniqueness of the solution of such boundary value problem lead to the invertibility conditions for the matrix, whereas solutions for suitable problems provide explicitly the entries of the inverse matrix as the linear combinations of Chebyshev functions.