Complexity in simplicity: Flexible agent-based state space exploration

In this paper, we describe a new flexible framework for state space exploration based on cooperating agents. The idea is to let various agents with different search patterns explore the state space individually and communicate information about fruitful subpaths of the search tree to each other. That way very complex global search behavior is achieved with very simple local behavior. As an example agent behavior, we propose a novel anytime randomized search strategy called frustration search. The effectiveness of the framework is illustrated in the setting of priced timed automata on a number of case studies.