Spectral Galerkin methods for a weakly singular Volterra integral equation of the second kind

Abstract

In this paper, two methods (the standard spectral Galerkin polynomial method and a variant where the trial space is augmented by suitably chosen functions) are applied to solve a weakly singular Volterra integral equation. The Sloan iterations of both these methods are also considered. Convergence rates for all four of these methods are proved, while taking into account the effect of quadrature. In particular, optimal rates in |$L^2$| and |$L^\infty$| are obtained for the two basic methods. Our augmented method is ill conditioned; to circumvent this drawback, a new analytical–numerical approach is introduced, which enjoys the higher-order accuracy of the augmented method without any associated ill-conditioning. It is noteworthy that this analytical–numerical method could be combined with any numerical method for this class of integral equations to yield an enhanced order of accuracy.