Solving Permutation Problems with the Ordering Messy Genetic Algorithm

Abstract

Although successful at first, messy genetic algorithms had minimum attention within the evolutionary computation community for the past few years. This chapter presents an ordering messy genetic algorithm (OmeGA) that is able to solve difficult permutation problems efficiently. Starting with a brief introduction to the fast messy genetic algorithm (fmGA), the chapter continues by proposing a robust representation model--the random keys--that proved to work successfully for representing permutations. The design of OmeGA is described and ordering deceptive problems are discussed in detail. Thereafter, experimental results that show the random key-based simple genetic algorithm (RKGA) being outperformed by its messy competitor in 32-length ordering deceptive problems are presented. The OmeGA is completely independent from the underlying chromosome's coding scheme and finds the global optimal solution in problems with both tightly and loosely coded building blocks. The chapter finally demonstrates the OmeGA's scale-up behavior.