This is a cute example, but the problem could probably be solved much more efficiently with standard optimization methods, like Nelder-Mead. See Numerical Recipes...
The problem with curve fitting examples for GAs is that on one side there is often a more efficient way to solve the problem and, more importantly, that it assumes you know what the equation looks like.<p>Having used (and still) genetics algorithms to build a good approximation function for a natural phenomenon, I found that I needed to spend a lot of time crafting my chromosomes structure so that it would get a better chance of eventually solving the problem. I also had to put correlation in the fitness function because it would otherwise get stuck in a local maxima where it completely eliminated most of the inputs but the resulting curve looked nothing like the targeted one.<p>However I am still not completely satisfied with the approximations I get and I would be VERY interested in recommendation for more advanced reading on the subject. Anyone?