Ernst Messerschmid
Parallel and Distributed Development and Simulation of Atmospheric Models
V. Mastrangelo, I. Mehilli, F. Schmidt, M.Weigele, J.Kaltenbach, A.Grohmann, and R.Kopetzky
Parallel and distributed computing is a way to meet the increasing demand for engineering and computational power in order to perform scientific and technical simulations. To make use of these new paradigms we are developing an agent based (simulation) environment which is operating on various computers at different locations in Europe. In order to set up a multi-agent system and to integrate modules, objects and services it is necessary to
- Modularize the overall problem
- Develop a strategy of distribution
- Parallelize and optimise the computationally expensive modules
- Encapsulate modules as independent processes
- Develop a strategy for the integration of (competing) services into a simulation system Solutions to most of the problems are proposed.
Communication between agents takes place on the basis of CORBA mechanisms. A special Service Agent Layer (SAL) provides methods to make agents from modules, objects or services. Experiences with this architecture and the implementation of parts of it will be reported. Further possibilities and difficulties of this approach are outlined by referring to the example of the dispersion of air carried particles. This example consists of four different parts, of which the simulation of the wind field and the simulation of air carried particle transport form the computationally expensive core of the application. They are run on a CRAY-T3E of IDRIS-CNRS in Paris.
Table of Contents | Author Index | CUG Home Page | Home (Title Page)
|