Gentle but rigorous introduction to genetic algorithms or what makes them tick?

Marek W. Gutowski 

Polish Academy of Sciences, Institute of Physics, al. Lotników 32/46, Warszawa 02-668, Poland

Abstract
The idea of genetic improvements of trial solutions seems quite natural and obvious. Yet, when it comes to translation of the original (optimization) problem into "genetic language", i.e. to practical realization - then the majority of potential users get stuck. In this introductory talk I am going to present not only the general idea of those algorithms but also some very simple (!) mathematics behind them.The purpose of my talk is twofold:
  • showing that overwhelming majority of practical problems may be expressed in a unified way,
  • pointing that the so called "tuning parameters", like population size, mutation ratio, and mating probability among others, need not to be problem-specific.

More precisely, only some of the tuning parameters are really important and their values should satisfy pretty simple necessary conditions (inequalities) in order to work correctly and efficiently.  These conditions are completely unrelated to the specific problem under consideration.

Small-world view of the complete space where 5-bit long chromosomes may live. 

The figure accompanies the talk "Gentle but rigorous introduction to genetic algorithms" by Marek W. Gutowski.

 

Presentation: Invited at E-MRS Fall Meeting 2007, Genetic algorithms for beginners, by Marek W. Gutowski
See On-line Journal of E-MRS Fall Meeting 2007

Submitted: 2007-07-02 13:49
Revised:   2009-06-07 00:44