A hybrid genetic algorithm for multiway graph partitioning

Abstract

A hybrid genetic algorithm for multiway graph partitioning is proposed. The algorithm includes an e cient local optimization heuristic. Starting at an initial solution, the heuristic iteratively improves the solution using cyclic movements of vertices. The suggested heuristic performed well in itself and as a local optimization engine in the hybrid genetic algorithm. Combined with the framework of hybrid genetic algorithms, it showed signi cant performance improvement. 1