Traditional Techniques of Genetic Algorithms Applied to Floating-Point Chromosome Representations
Leo Budin, Marin Golub, Andrea Budin-Posavec · 1996
The choice of chromosome representation in genetic algorithms depends on the variables of the optimization problem being solved. If the variables are realvalued, the chromosomes can be represented as fixed-point integer values which enables the use of classical genetic operators defined for binary strings. Another possible chromosome representation is using floating-point numbers directly. In this case, genetic operators have to be defined additionaly. This paper presents a different approach: floating-point chromosome representation with traditional genetic operators. In order to achieve fine local tuning of the solutions, a mapping is defined which dynamically changes the operating scope of genetic operators.