Information Technology Reference
In-Depth Information
200
Best fitness
Avg fitness
150
100
50
0
0 0 0 0 0 0 0 0 0 0 0
Generations
Figure 12.6.
Evolutionary dynamics characteristic of non-homogenizing systems.
This population evolved under a mutation rate of 0.05. Note the oscillatory pattern
on average fitness and the wide gap between best and average fitness.
of the algorithm increased a little (see Table 12.2). Note that the success rate
increases slightly comparatively to the individual performances obtained for
the recombinational operators working separately. Thus, for this study, we
will choose the general settings shown in the fifth column of Table 12.2.
Notwithstanding, these populations exhibit the same kind of homogenizing
dynamics characteristic of populations undergoing only one type of recom-
bination at a time (Figure 12.7). This further reinforces the hypothesis that
recombination is conservative and, therefore, plays a major role at maintain-
ing the status quo. Note that, in this particular case, by generation 54 the
plots for average and best fitness merged and all individuals became geneti-
cally identical. This might be seen as a good thing, especially if all the indi-
viduals became equal and perfect. Recall, however, that in complex real-
200
Best fitness
Avg fitness
150
100
50
0
0 0 0 0 0 0 0 0 0 0 0
Generations
Figure 12.7.
Homogenizing populations undergoing extensive recombination.
This dynamics was obtained for a successful run of the experiment summarized in
the fifth column of Table 12.2.
