Figure 5

Optimizing the settings for binGA on the reversible GMAK model. The best combination of mutation and crossover operators for binGA on the reversible GMAK model is adaptive mutation and bit-simulated crossover. For this pair the influence of the mutation and crossover probabilities p _m and p _c is examined and plotted in (a). Each experiment is repeated 20 times. The combination p _m = 0.2 and p _c = 1.0 improved the average fitness most successfully. Figure 5(b) depicts the impact of the population size for this setting. The variance decreases with an increasing population size and reaches its minimum at 1,000. However, the best single result cannot be surpassed.