Conclusions

In this paper, we have proved that an evolutionary algorithm is an efficient way of solving oracle-type problemas like Mastermind. The genetic algorithm performs as well or better than globally optimal and exhaustive search algorithms, can be applied to problems of any size, and the number of combinations evaluated (and thus the time taken to reach the solution) grows linearly. The algorithm also surpasses other evolutionary algorithms like Bento's [18] and our own [17].

In fact, this algorithm has been working since May 1999 in the Genetic
Mastermind 2.0 site, at * http://geneura.ugr.es/~jmerelo/GenMM*. From this site, the full code
for GenMM can be downloaded; in fact, it is included as an example for
the EOlib (Evolving|Evolutionary objects) library, which is available
from *http://geneura.ugr.es/~jmerelo/EO *.