Adaptive parallelization of strategies in agent based systems

C. Geiger ([email protected]), B. Kaltho ([email protected]) Heinz Nixdorf Institut, Fuerstenallee 11, 33102 Paderborn, Germany Abstract In this paper we present an approach for adaptive partitioning of strategies in agent oriented systems based on algorithmic skeletons [2]. This provides the user with parallel programming templates for modeling agent strategies and their e cient parallelization. Information present in these skeletons supports the decision for an e cient partition. First, we mention agent based modeling and our validation example taken from the eld of manufacturing. Types of strategies are described and the idea of skeletons is introduced. Then we present the concept of PARTITIONER, a meta agent which computes an adaptive parallelization based on the available knowledge provided by the used skeletons. Finally, rst implementation results are discussed and future research in this area is motivated.