This page describes the computational technique called Genetic Algorithms (GA). It is a method for solving general optimisation problems. In fact, GAs can solve a larger class of problems, but are usually used for optimisation.

Optimisation problems can be thought of as finding the minimum of some complicated n-dimensional surface, where n is the number of parameters to change.

In a Genetic Algorithm, converagnce on the global minimum is encouraged by allowing a collection of potential solutions to complete for a finite resource. Each of the potential solutions contains a value for each parameter. The value of a particular parameter within a trial solution can be thought of as a "gene."

Pairs of randomly chosen parent solutions breed to form offspring. The children have some genes copied from one parent and the remainder from the other. After breeding is completed, the distinction between parents and children is removed, and all the solutions vie with each other for inclusion in the next generation and thereafter the chance to be parents.

In addition to breeding, GAs incorperate a mutation step. After all required children have been bread, some of the child grenes are changed to a random number. Mutating increases diversity and facilitates in escaping from local minima. These steps of a basic GA are shown in the following diagram.

Webmaster Paul Millar email paulm@astro.gla.ac.uk