next up previous

4.1.3 Mutation

To simulate the introduction of new mutations, we simply add a random integer to the new child. Thus to create a new child named x from the ith member of the current generation, the code is:

    x = g[i] + poisson(u);
where poisson() is a function that implements a Poisson random number generator and u is the expected number of mutations.

Note that the variable u corresponds to the parameter of the mathematical model. For small values of the expected value the Poisson procedure described in Numerical Recipes [3] works fine. The average number of mutations per individual measured empirically over a wide variety of species is about 1.0, and the procedure in Numerical Recipes works for , so it should be sufficient.