Comparative Study between the Improved Implementation of 3 Classic Mutation Operators for Genetic Algorithms

Mutation is the most important Genetic Algorithms operator, allowing them to thoroughly explore the design space of an optimization problem. If designed correctly it also allows for the exploitation of promising solutions, task usually attributed to crossover. This study compares the performance of three classic mutation operators: uniform, polynomial and Gaussian. The tool used is the OOGA framework which implements an improved and unified variant of the mutation operators. GA performance is evaluated on a benchmark structural optimization problem using three criteria: accuracy, reliability and efficiency. The optimum configuration of each operator is also explored by varying mutation parameters over a range of possible values. Overall the study is aimed at the optimization practitioners, offering them the means to make informed decisions about the right mutation operator and its setting for particular problems.