The discrete collocation method for nonlinear integral equations (Q4031986)

If the collocation method is used to solve an integral equation like the Urysohn equation, \(x(s)=\int_ DK(s,t,x(t))dt\), \(s\in D\), where \(D\) is a closed bounded set in \(\mathbb{R}^ m\) \((m\geq 1)\), and where \(K(s,t,u)\) defines a completely continuous nonlinear integral operator on some open subset of \(C(D)\) into \(C(D)\), the resulting integrals must usually be evaluated numerically. In this paper the authors present a general framework for the so-called discrete collocation method; in contrast to other studies of this method, they do not treat it as a perturbation of the ``exact'' collocation method and thus obtain stronger results than appear to be possible with earlier types of analysis. The present approach also leads to an elegant derivation of superconvergence results for discrete collocation methods. Finally, the analysis is applied to a variant of the collocation method due to \textit{S. Kumar} and \textit{I. H. Sloan} [Math. Comp. 48, 585-593 (1987; Zbl 0616.65142)] for Hammerstein equations.