Artificial Intelligence, Automated Reasoning, and Symbolic Computation

Volume 2385 of the series Lecture Notes in Computer Science pp 102-116


Indefinite Integration as a Testbed for Developments in Multi-agent Systems

  • J. A. CampbellAffiliated withDepartment of Computer Science, University College London

* Final gross prices may vary according to local VAT.

Get Access


Coordination of multiple autonomous agents to solve problems that require each of them to contribute their limited expertise in the construction of a solution is often ensured by the use of numerical methods such as vote-counting, payoff functions, game theory and economic criteria. In areas where there are no obvious numerical methods for agents to use in assessing other agents’ contributions, many questions still remain open for research. The paper reports a study of one such area: heuristic indefinite integration in terms of agents with different single heuristic abilities which must cooperate in finding indefinite integrals. It examines the reasons for successes and lack of success in performance, and draws some general conclusions about the usefulness of indefinite integration as a field for realistic tests of methods for multi-agent systems where the usefulness of “economic” criteria is limited. In this connection, the role of numerical taxonomy is emphasised.