An alternative to gradient-based optimization methods is to let the optimization be guided solely by fitness values. This kind of optimization has no explicit knowledge of how the fitness landscape looks, but merely considers the fitness function to be a black box that takes candidate solutions as input and produces a fitness value as output. This is known in this chapter as Derivate-free optimization, direct search, heuristic optimization, meta-heuristics, black-box optimization, and so on. We will use these terms a lot!