Quadratic assignment [chapter]

Volker Nissen
Handbook of Evolutionary Computation  
The quadratic assignment problem (QAP) is of practical relevance in different fields of application, such as hospital layout planning, machine scheduling, and component placement on printed circuit cards. We discuss the QAP as a base model for facility layout problems and present an evolutionary solution technique derived from evolution strategy (ES), one of the mainstream forms of evolutionary algorithms (EAs). It is empirically evaluated on a test suite of QAPs. The implemented ES for
more » ... orial problems (CES) is deliberately not hybridized with other solution techniques. This gives an idea of the potential of ES on this standard problem in the light of recent debates about the competitiveness of EAs for solving combinatorial problems. CES is a good heuristic for solving QAPs and does not require tuning of strategy parameters on individual problem instances. Even though CES is superior over the classical 2-Opt procedure and an evolutionary approach proposed by Tate and Smith (1995) , it is not fully competitive with Threshold Accepting, one of the most efficient heuristics currently available for QAPs. Implications of these results are discussed. 100 std dev./mean
doi:10.1887/0750308958/b386c125 fatcat:klgmwravsjf7ngtiqkqx5xg3yq