Deterministic Multi-step Crossover Fusion: A Handy Crossover Composition for GAs

Multi-step crossover fusion (MSXF) is a promising crossover method using only the neighborhood structure and the distance measure, when heuristic crossovers are hardly introduced. However, MSXF works unsteadily according to the temperature parameter, like as Simulated Annealing. In this paper, we introduce deterministic multi-step crossover fusion (dMSXF) to take this parameter away. Instead of the probabilistic acceptance of MSXF, neighbors are restricted to be closer to the goal solution, the best candidate of them is selected definitely as the next step solution. The performance of dMSXF is tested on 1max problem and Traveling Salesman Problem, and its superiority to conventional methods, e.g. uniform crossover, is shown.