Multiagent Coordination Using Graph Structured Mathematical Optimization

We address the problem of solving mathematical programs defined over a graph where nodes represent agents and edges represent interaction among agents. We focus on the class of graph structured linear and quadratic programs (LPs/QPs) which can model important multiagent coordination frameworks such as distributed constraint optimization (DCOP). For DCOPs, our framework provides a key benefit of modelling functional constraints among agents (e.g. resource, network flow constraints) in a much more tractable fashion. Our framework is also more general than previous work on solving graph-based LPs/QPs as it can model a richer class of objective function and constraints than previous work. Our iterative approach has several desirable properties—it is guaranteed to converge to the optimal solution for LPs, it works for general cyclic graphs, it is memory efficient making it suitable for resource limited agents, and has anytime property. Empirically, our approach provides solid empirical results on several standard benchmark problems when compared against previous approaches.