Advanced object-oriented methods and concepts for simulations of multi-body systems

A common way to simulate the motion and behaviour of a complex system is to approximate it with a multi-body system based on particles, and to evaluate the particle interactions over time. Normally, the calculations require such a large amount of computer power that methods have been developed to decrease computation time by taking advantage of new computer architectures. These methods have been in use since the early 1980s, but there is still a lackof concepts supporting the application of these methods, to build highly efficient simulation systems with reusable entities. In this paper we describe a new approach for the development of flexible, efficient methods and techniques to control multi-body systems. The major goal of our concept is the integration of advanced modern simulation techniques such as particle-based modelling and simulated annealing, and afficient data structures such as octrees and the Greengard algorithm. We introduce high-level concepts represented by agents, env ironments, and controllers based on and supportes by the object-oriented paradigm. We demonstrate the flexibility of our new simulation system for application related to environmental research and air quality control.