GEP Book

  Home
  News
  Author
  Q&A
  Tutorials
  Downloads
  GEP Biblio
  Contacts

  Visit Gepsoft

 

C. FERREIRA In H. J. Caulfield, S.-H. Chen, H.-D. Cheng, R. Duro, V. Honavar, E. E. Kerre, M. Lu, M. G. Romay, T. K. Shih, D. Ventura, P. P. Wang, Y. Yang, eds., Proceedings of the 6th Joint Conference on Information Sciences, 4th International Workshop on Frontiers in Evolutionary Algorithms, pages 614-617, Research Triangle Park, North Carolina, USA, 2002.

Mutation, Transposition, and Recombination: An Analysis of the Evolutionary Dynamics

Recombination
 

The performance of the three kinds of GEP recombination is shown in Figure 1. It is worth emphasizing that two-point recombination is the most disruptive of the recombination operators and, as shown in Figure 1, it is also the most efficient kind of recombination. Not surprisingly, the most conservative gene recombination is also the less efficient.

The dynamics characteristic of recombination (Figure 4) also show an extremely important feature of recombination, that is, the homogenizing effect of all kinds of recombination, from the most conservative to the most disruptive. For obvious reasons, these recombination-specific dynamics are called Homogenizing dynamics. Note that, in all cases, after a certain time the plot for average fitness overlaps the plot for best fitness. This indicates that the populations lost all the diversity and all the individuals have the same genetic makeup. Obviously, if populations converge to this stage before finding a good solution, they become irrevocably stuck in that point if no other, non-recombinatorial operators are available. As the small success rates obtained for recombination emphasize (Figure 1), when populations evolve by recombination only, most of the times, they converge before finding a good solution. This tells us that recombination should never be used as the only source of genetic variation.

Figure 4. A gallery of evolutionary dynamics found in populations evolving by two-point recombination (plot a), one-point recombination (plot b), and gene recombination (plot c). The success rate above each plot was determined in the experiment shown in Figure 1. In all the plots the evolutionary dynamics are of the kind Homogenizing. a) p2r = 1.0. b) p1r = 1.0. c) pgr = 1.0.


Furthermore, it is worth noticing that GEP recombination, even the most conservative gene recombination, is more disruptive than the homologous recombination that occurs in nature during sexual reproduction as the exchanged genes rarely are homologous. One of the unsolved questions of biology is the role of sex in evolution and, most of the times, biological sex in its overwhelming diversity is confounded with the homologous recombination that occurs during sexual reproduction. Consequently, many erroneously assume that homologous recombination creates great diversity. The comparison of GEP operators, especially GEP recombination operators, suggests that a more conservative recombination like homologous recombination would only be useful to maintain the status quo in periods of stasis.

Home | Contents | Previous | Next