There is a suite of benchmark functions (Shi & Eberhart 1998) which are commonly used to test the performance of numeric optimization algorithms. These functions are chosen because of their particularities, which render their optimization difficult. These comprise
A numeric optimization algorithm can be believed to sport a good performance a large number of problems if it can accurately solve this benchmark. Even though it represents a subset of the possible optimization problems, it is our belief that it represents the subset of such problems that we would like to solve.
The criterion given in the definiation of the functions is a value for the fitness function below which a solution is located in the region of the global optimum.