It depends on the overall goal of you optimization approach and the concrete fitness function for this approach ...

For GA we can generally distinguish:

1) Computable fitness(function) - There is a computable fitness-function to compute the fitness of each individual with regards to the optimization goal. (e.g. fastest, shortest, most accurate, ...)

2) Interactive fitness(function) - The fitness of each individual is determined subjectively by human assessement. (e.g. by looking at it and judge what suites you most)

It depends on the overall goal of you optimization approach and the concrete fitness function for this approach ...

For GA we can generally distinguish:

1) Computable fitness(function) - There is a computable fitness-function to compute the fitness of each individual with regards to the optimization goal. (e.g. fastest, shortest, most accurate, ...)

2) Interactive fitness(function) - The fitness of each individual is determined subjectively by human assessement. (e.g. by looking at it and judge what suites you most)

I agree with Robert Fischer. As he said, it depends on the goal of the optimization. I suggest you should first specify the objective of your code. After, you should try to compare the current results with the previous one. Then, you will have some notion about the fitness error. Notice that this is one of the most simple way and perhaps it is not sufficiently necessary to your problem.

A brief but well to understand summary of GA can be found here:

http://www.ro.feri.uni-mb.si/predmeti/int_reg/Predavanja/Eng/3.Genetic%20algorithm/_01.html

Thank you all !! Your answers are really helpful.