Benchmark Generator for the IEEE WCCI-2014 Competition on Evolutionary Computation for Dynamic Optimization Problems

The field of dynamic optimization is related to the applications of nature-inspired algorithms [1]. The area is rapidly growing on strategies to enhance the performance of algorithms, but still there is limited theoretical work, due to the complexity of natureinspired algorithms and the difficulty to analyze them in the dynamic domain. Therefore, the development of BGs to evaluate the algorithms in dynamic optimization problems (DOPs) is appreciated by the evolutionary computation community. Such tools are not only useful to evaluate algorithms but also essential for the development of new algorithms. The exclusive-or (XOR) DOP generator [5] is the only general benchmark for the combinatorial space that constructs a dynamic environment from any static binaryencoded function f(x(t)), where x(t) ∈ {0, 1}, by a bitwise XOR operator. XOR DOP shifts the population of individuals into a different location in the fitness landscape. Hence, the global optimum is known during the environmental changes. In the case of permutation-encoded problems, e.g., the travelling salesman problem (TSP) where x(t) is a set of numbers that represent a position in a sequence, the BGs used change the fitness landscape. For example, the dynamic TSP (DTSP) with exchangeable cities [2]