Between theory and practice: guidelines for an optimization scheme with genetic algorithms - Part I: single-objective continuous global optimization

TL;DR

This paper provides practical guidelines for applying genetic algorithms to single-objective continuous global optimization problems, emphasizing their versatility and robustness for practitioners without proposing new methods.

Contribution

It offers a collection of practical tips and insights for non-experts on how to effectively use genetic algorithms in real-world optimization tasks.

Findings

Genetic algorithms are versatile and robust for optimization.

Practical tips improve the application of genetic algorithms.

Genetic algorithms remain central despite many available techniques.

Abstract

The rapid advances in the field of optimization methods in many pure and applied science pose the difficulty of keeping track of the developments as well as selecting an appropriate technique that best suits the problem in-hand. From a practitioner point of view is rightful to wander "which optimization method is the best for my problem?". Looking at the optimization process as a "system" of intercon- nected parts, in this paper are collected some ideas about how to tackle an optimization problem using a class of tools from evolutionary computations called Genetic Algorithms. Despite the number of optimization techniques available nowadays the author of this paper thinks that Genetic Algorithms still play a central role for their versatility, robustness, theoretical framework and simplicity of use. The paper can be considered a "collection of tips" (from literature and personal experience) for the non-computer-scientist that has to deal with optimization problems both in the science and engineering practice. No original methods or algorithms are proposed.