Coupled map lattices as spatio-temporal fitness functions: Landscape measures and evolutionary optimization

Coupled Map Lattices (CML) can be interpreted as spatio-temporal fitness landscapes which may pose a dynamic optimization problem. In this paper, we analyze such dynamic fitness landscapes in terms of the landscape measures modality, ruggedness, information content and epistasis. These measures account for different aspects of problem hardness. We use an evolutionary algorithm to solve the dynamic optimization problem and study the relationship between performance criteria of the algorithm and the landscape measures. In this way we relate problem hardness to expectable performance.