Investigating F# as a development tool for distributed multi-agent systems

In this paper we explore the use of F#, a functional language, as a means of developing distributed multi-agent systems. In order to achieve this goal, we started with the implementation of the MapReduce algorithm. We also implemented the algorithm in JADE, hence allowing for an initial comparison of the language-platform combination from an usability point of view with respect to the development of distributed agent systems. The results include an analysis of their underlying threading models as well as of the overall system performance. We show how the algorithmic specification can be easily mapped onto the constructs offered by the language. In the final discussion we identify some interesting features of F# that make it suitable for implementing distributed multi-agent systems.