Population Distribution in a cEA

The population of a cEA is composed by a set of individuals that interact among them along the different generations. These individuals represent potential solutions to the problem and a search path towards the optimal solution. So as bigger the population be easier will be finding a solution, although more system resources will be needed during the execution.

Typical population sizes in evolutionary computation range from 1 (for certain evolution strategies) to many thousands (for genetic programming).

In the case of cEAs, the population is usually represented as a 2-D toroidal grid in which nodes are placed the individuals. In Figure 1 we can see how a torus can be obtained from a 2-D grid.

Figure 1. Construction of a toroidal shape from a plane

In cEA implementations, a toroidal shape of the population is easily obtained by linking the individuals placed on the extremes of a 2D grid like shown in Fig. 2.

Figure 2. Toroidal grid structure used for holding individuals in most cEAs

back to home