Genetic Algorithms: Selection and Crossover

ALife Home

Introduction

Automata

Genetic Algorithms

Neural Networks

Experiments

People

Timeline

References

The following two Flash(c) animations illustrate the processes of Selection and Crossover.

Selection

This is where the a number of individuals of size probability P(s) are chosen in proportion to their fitness in relation to the fitness of the whole population.

Crossover

A number of these selected individuals of size probability P(c) are taken two at a time, a random point is chosen on their genotype sequence and they exchange genotype bits from that point. Two new individuals are therefore produced.

The individuals not selected for crossover have in effect undergone asexual reproduction from their parent and are just copied to the new generation along with the newly created ones; the parents of all the individuals are not copied.

Fitness Objective

The objective of the fitness function used on the GA below, is to maximize the number of 1's in the genotype of the individual.


Below we can see that after crossover has completed we have increased the total fitness of the population from 6 to 7, and our best individual has an increased fitness of 3 instead of 2.


michael@obrienm.com
Last Updated: Ottawa, Canada

Genetic Algorithms Genetic Algorithms Genetic Algorithms: Optimization Strategies Genetic Algorithms: Optimization Strategies Artificial Life Home Artificial Life Home