Parallel simulation methods for dynamical systems: issues and challenges

Abstract

In this paper we present and discuss a number of important issues related to the fast simulation of complex dynamical systems. Such simulations are required frequently in modern engineering applications and they are also amongst the most expensive of all computational tasks. In many industries it is of vital importance to improve the speed of simulation of a given process whilst using moderate-priced hardware. This can be achieved by using multiprocessor systems together with parallel processing techniques. The aim of this paper is to highlight the possibilities of enhancing parallelism in the simulation of dynamical systems described by large sets of differential equations. The emphasis is put onto decomposition and hierarchical approaches leading to computational tasks of ‘large granularity’ but in order to make this presentation self-contained and as complete as possible, the other relevant aspects are addressed as well. It is also aimed in this paper to show that the complexity of the issues being faced with leads to many challenging problems related to the methodology of parallel simulation as well as to the particular methods and their implementations using multiprocessor networks.