An adaptive method for Volterra-Fredholm integral equations on the half line (Q1019657)

Motivated by problems arising from real life applications the authors focus on the numerical solution of Volterra-Fredholm integral equations (VFIEs) on unbounded spatial domains, with kernels that present variable peaks along the space. The numerical method presented in this paper, based on a trapezoidal direct quadrature formula along space and time, exploits the characteristics of the kernel and just covers the peaks of the kernel, \(k\). Section 2 details the difficulties encountered in the numerical treatment of the VFIEs under discussion. Conditions on the kernel, \(k\), and the forcing term, \(f\), are stated and details of the new method are presented. In section 3 the global error of convergence is analysed at the mesh points. Conditions on \(k\) and \(f\) under which convergence is achieved are stated. A theorem relating to convergence is presented and proved. Bounds are established which ensure order 2 convergence. Results of numerical experiements are included in section 4 to illustrate convergence of the new method and to demonstate the potential for a high computational cost. Performance of the method is compared with those of a classical implementation of an \(m\)-nodes Gauss-Laguerre formula on the half-line. The authors briefly discuss the use of parallel computation to reduce the implementation time. They also refer to the need for a stability analysis of the new method and indicate that this will be the subject of a later paper.