Steps to an Ecology of Actor Simulation

From an early definition of the actor concept to whole communities of concurrent processes behaving in their interactions, strategies, and competition for resources like whole ecologies of reflective agents, object-oriented concurrent programming has gone through several conceptual steps, giving rise to a structured methodology. To apply concurrent object-oriented techniques to discrete-event and behavioral simulation problems, the same kind of incremental methodology has to be defined. The first step provides minimal extensions of an actor language for discrete-event simulation. For an improvement of efficiency, the second step introduces concurrency issues in the form of a message-driven time-warp system. The third step introduces reflection as a solution to the rollback problem. It is shown how reflective techniques can be used to dynamically restore an actor's state. The fourth step defines agents as reflective actors aware of their environment, described as an ecosystem. This awareness makes agents able to make local decisions based on imperfect knowledge about the whole system or inconsistent and delayed information. In this case, reflection gives an agent a proper view of the ecosystem where it belongs. The agent becomes more adaptable to evolving circumstances: it can reason about its environment, change it, or modify itself. This constructivist approach of actor simulation techniques makes modeling easier, while it provides a flexible and powerful toolkit to solve several kinds of discrete-event or behavioral simulation problems.