ABSTRACT
Folding of ribonucleic acid (RNA) molecules and evolution of their structures according to predetermined rules result in highly rugged value landscapes which originate from the complexity of genotype-phenotype relations. Extensive computer simulations were performed in order to test the predictions of the model. In addition they provide insight into the mechanism of escape from traps of the evolutionary optimization process. The cost function of evolutionary optimization is therefore often characterized as fitness landscape. An important common feature of all correlations between genotypes and phenotypic properties is the bizarre structure of value landscapes. In order to complement optimization by genetic algorithms, the value landscape was investigated by means of simulated annealing. Optimization on a highly rugged value landscape derived from RNA secondary structures— modeled by means of binary sequences— results in different scenarios depending on population size, error rate, and local details of the landscape.
