Galerkin schemes and the sinc-Galerkin method for singular Sturm- Liouville problems (Q915396): Difference between revisions

The computation of the eigenvalues of the Sturm-Liouville problem \(Lu(x)\equiv -u''(x)+q(x)u(x)=\lambda \rho (x)u(x),\quad a<x<b,\) \(u(a)=u(b)=0\) by Galerkin methods and by the sinc-Galerkin method is discussed and compared. The methods considered include spectral, finite element and collocation schemes, both regular and singular Sturm- Liouville problems are involved. The properties (error estimate, mode of implementation relative to varying q and \(\rho\) or the domain (a,b) and the rate of convergence) of the sinc function method [cf. the first author, Approximation of eigenvalues of Sturm-Liouville differential equations by the sinc-collocation method. Dissertation, Montana State University (1987)], the commonly used maps for the sinc methods are described and illustrated. Then the methods are shown on 5 examples: The Fourier equation (as a representative of the regular problem), the Bessel equation (on a finite interval and q(x) with a singularity at \(x=0)\), the radial Schrödinger equation with Woods-Saxon and harmonic oscillator type potentials, and the Hermite equation. The Rayleigh-Ritz (finite element) method with cubic spline functions, the spectral method with Chebyshev basis functions and other mappings in combination with spectral or collocation methods are used for comparison.