Multi-agent integer programming

Abstract

Many real-life optimization problems such as planning and scheduling require finding the best allocation of scarce resources among competing activities. These problems may be modeled and solved by means of mathematical programming. This paper explores a distributed multi-agent approach to mathematical programming, and demonstrates the approach in the case of integer programming. The important characteristics of the multi-agent approach consist in that the behavior-based computation performed by the agents is parallel and goal-driven in nature, and has low time complexity.