Solving Quadratic Assignment Problem in parallel using local search with simulated annealing elements

The Quadratic Assignment Problem (QAP) is one of the hardest among the combinatorial optimization problems with wide practical applications. It is impossible to solve it by exact methods, since it is a NP-hard problem. This paper presents a parallel algorithm for solving the QAP based on well-known meta-heuristics, which are combined and improved, to reach the best possible results. There is detailed description of whole algorithm with explanation of impact of individual steps. Outcomes of the algorithm are compared with several other algorithms on some instances of QAPLIB. As it is obvious from summary, proposed algorithm achieves better results than other algorithms.