ABSTRACT
This chapter introduces and discusses variants of the genetic algorithm both for trees and two-dimensional bitmaps. It is well known that the functionality of genetic algorithms (GA) highly depends on finding a suitable encoding for the data occurring in the optimization problem to be solved. GA for trees have mainly been used in the context of genetic programming (GP), as in Reference for example, a relatively technique, proven to be a rather versatile tool for automatic program generation. The programs and the corresponding trees are constructed from a set of functions and a set of terminal symbols, defined by the user. The algorithm starts with crude, randomly generated items, using a model grammar with generic design rules, and applies genetic operators to let these models evolve. The model is based on the following components: Grammar based, structured object modelling, Parametrization of numerical model features and constraints, and Performance feedback.
