Swarm is a service to the community of researchers building computers simulations. Computer programs have become an important aspect of experimental science and yet few general purpose tools exist to help people write models. The goal of Swarm is to provide consistent experimental tools in the form of libraries of carefully designed, implemented and tested code. Such libraries can also be developed and exchanged within user communities that organize and maintain the model-building components applicable to their shared needs.
Swarm defines a structure for simulations, a framework within which models are built. The core commitment is to a discrete-event simulation of multiple agents using an object-oriented representation. To these basic choices Swarm adds the concept of the ``swarm,'' a collection of agents with a schedule of activity. The swarm is the basic structural element of a model: it is the basic collection that allows scaling within large models, and also supports the modeling of complex, multi-level dynamics including agents that model their own world. In all these choices, the goal of Swarm is to enable a higher level of representation for simulations, thereby making it easier to understand, implement, repeat, and communicate computer models.
Since Swarm's first limited beta release in October 1995, some thirty groups of users have installed Swarm and are actively writing models with it. There is already one finished paper that presents results accomplished with Swarm [5]; in the coming months we expect a series of further models developed by Swarm users, with published results following close behind.
In the next few months we will have a finished, version 1.0 release of Swarm. Our goal is to create a distribution that is simple, complete, well documented, and bug-free. Our longer term goal is to foster a community of modelers, researchers who use computer software for experimental science. Swarm is already helping to provide focal point for discussion of simulation techniques and methodology. Swarm can also facilitate the sharing of modeling components and libraries within particular research communities, fostering an important form of intellectual exchange. Finally, a formalized framework for model definition establishes a necessary standard of specification for computer programs used as tools in experimental science.
