Computing the eigenvalues of a class of nonlocal Sturm-Liouville problems (Q969869)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Computing the eigenvalues of a class of nonlocal Sturm-Liouville problems
scientific article

    Statements

    Computing the eigenvalues of a class of nonlocal Sturm-Liouville problems (English)
    0 references
    0 references
    8 May 2010
    0 references
    The author proposes a methodology to compute eigenvalues for Sturm-Liouville problems with integral and nonlocal conditions: \[ \begin{cases} -y'' + q(x) y = \lambda y, & x\in [0,1], \\ {\mathcal X}_0(y)=0, \quad {\mathcal X}_1(y)=0. \end{cases} \] with, \[ {\mathcal X}_i(y)=\int_0^1 [y(t)d\psi_{i,1}(t)+y'(t)d\psi_{i,2}(t)], \quad i=0,1, \] where integrations are understood in the Rieman-Stieljes sense, and with suitable hypothesis on the functions that appear in the formulation of the problem. The methodology is based on the regularized sampling method introduced by the same author. The result is constructed considering the solutions (\(y_c(x,\mu)\) and \(y_s(x,\mu)\)) of two base problems (essentially the Sturm-Liouville equation with given \(\lambda=\mu^2\) and boundary conditions \(y(0)=1\), \(y'(0)=0\) for \(y_c\) and \(y(0)=1\), \(y'(0)=1\) for \(y_s\)). Then it is known that the Sturm-Liouville operator has a point spectrum that coincides with the set of the square zeros \(\mu_k\), (\(k=0,1,2,\ldots\)) of the entire function \(B(\mu)\) defined as \[ B(\mu)= \left| \begin{matrix} {\mathcal X}_0(y_c) & {\mathcal X}_0(y_s) \\ {\mathcal X}_1(y_c) & {\mathcal X}_1(y_s) \end{matrix} \right|. \] The paper develops some lemmas and theorems to bound the growth of \(y_c\), \(y_s\), their derivatives and other functions in the Paley-Wiener space related with these. In the analysis, a few Whittaker-Shannon-Kotel'nikov series appear that need to be truncated for practical applications. The truncation error is also bounded, and the result is that \(B(\mu)\) is approximated by a truncated series \(B_N(\mu)\) within a known accuracy. Finally, taking into account the Rouche theorem, the author bounds \(|\mu_N - \mu^*|\) where \(\mu^*\) is a zero of \(B(\mu)\) of a given multiplicity and \(\mu_N\) is the corresponding approximation. In the last section of the paper, the results of the methodology are illustrated with two examples.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    second order Sturm-Liouville problems
    0 references
    eigenvalue problems
    0 references
    Whittaker-Shannon-Kotel'nikov theorem
    0 references
    nonlocal condition
    0 references
    integral boundary condition
    0 references
    error bounds
    0 references
    numerical examples
    0 references
    regularized sampling method
    0 references
    point spectrum
    0 references
    0 references