Genetic algorithms are gaining an increasing amount of interest in many domains. Even though good results are often achieved, the theoretical framework is still young. Theoretical research today is using a Markov chain as a model for genetic algorithms. The main drawback with this model is that it is only able to model very small problems.
Our research concerns the Markov chain model of the Simple Genetic Algorithm, where we aim at both simplifying the model so that it is useful for larger problems, and using it to find expressive features and correlate them to design choices. The design is today made by trial and error.
Towards Computing the Parameters of the Simple Genetic Algorithm (Jun 2001) Roger Jonsson Conferance on Evolutionary Algorithms