A New Hybrid Distributed Double Guided Genetic Swarm Algorithm for Optimization and Constraint Reasoning: Case of Max-CSPs

In this paper we propose a new distributed double guided hybrid algorithm combining the particle swarm optimization (PSO) with genetic algorithms (GA) to resolve maximal constraint satisfaction problems (Max-CSPs). It consists on a multi-agent approach inspired by a centralized version of hybrid algorithm called Genetical Swarm Optimization (GSO). Our approach consists of a set of evolutionary agents dynamically created and cooperating in order to find an optimal solution. Each agent executes its own hybrid algorithm and it is able to compute its own parameters. Our approach is compared to the GSO. It demonstrates its superiority. We reached these results thanks to the distribution using multi-agent systems, diversification and intensification mechanisms.