Author(s)

J.J. Grannell

Abstract

Invited Paper Large scale problems - efficiency, accuracy and adapt ivity J.J. Grannell Department of Mathematical Physics, University College Cork, Cork, Ireland INTRODUCTION The feasibility of applying a boundary element method (BEM) to obtain sufficiently accurate solutions of large scale problems depends primarily on the efficiency and stability of the numerical algorithms and also on available computer capacity. It is essential to design BEM algorithms which have minimal computational complexity to bring large scale problems into the feasible range. A number of fast techniques has emerged in recent years primarily for high frequency scattering problems in fluid mechanics and electrodynamics. For brevity, the discussion will be confined mainly to this class of ideas. One approach to cost reduction is to design BEM methods having fast (exponential) convergence. Such rates have been theoretically predicted for static problems (cf. Postell and Stephan [19] and the refe

Keywords