Author(s)

M. Zerroukat

Abstract

The solution of the diffusion equation using the boundary element method has a high computational cost due to the inherent time history constraint in the integral representation. This time-history dependence becomes imprac- tical and inadequate for problems where computations are to be performed for extended times. The computation of the solution at n domain points using m boundary points at the fc-th time step requires an amount of com- puter operations of the order O (km?-\-knm). This paper presents a scheme that requires a computaional cost of the order of only O(m? + nm), where the dependence from the past fc-steps is removed. 1 Introduction Numerical computations of time dependent partial differential equa- tions using the boundary integral approach have been greatly ham- pere

Keywords