Introducing Bar Systems: A Class of Swarm Intelligence Optimization Algorithms

We present Bar Systems: a family of very simple al- gorithms for different classes of complex optimization problems in static and dynamic environments by means of reactive multi agent systems. Bar Systems are in the same line as other Swarm Intelli- gence algorithms; they are loosely inspired in the behavior a staff of bartenders can show while serving drinks to a crowd of customers in a bar or pub. We will see how Bar Systems can be applied to CONTS, a NP-hard scheduling problem, and how they achieve much better re- sults than other greedy algorithms in the "nearest neighbor" style. We will also prove this framework to be general enough to be applied to other interesting optimization problems like generalized versions of flexible Open-shop, Job-shop and Flow-shop problems.