Re: JGAP - simple question re evolve() and the fitness function evaluate()

Esmail wrote:

In order to get a feel for the JGAP package I took a look at the
minimizing change code and wrote my own little test program.

I now have a question about evolve() and the fitness function
evaluate() I wrote.

If I have a population of 10, and call evolve(), shouldn't my fitness
function be called a constant time? I would think 10 times.

However, when I run my basic GA my evaluation function gets called
between 10 and 13 times .. I am not quite sure why this would happen.

Likewise if I set the number of evolutions to say 3, I'd expect my
fitness function evaluate() to be called 3 times as much as before, but
that doesn't happen either.


Thanks,
Esmail


anyone?

(or am I using a package that's not that popular ie used much by others
- should I be using something different?)
.

Relevant Pages

  • Re: Software simulation of hardware evolution
    ... evolve the behaviour of each gate/class in the form of a truth table, ... and then build higher-level blocks using a tree-like hierarchical ... A fitness function would have to be a series of test cases to check ...
    (comp.ai.genetic)
  • Re: GA to move tanks on grid (Advice for newbie needed).
    ... How do I represent my tank movements as a set ... How would they evolve? ... how to I represent this decision tree? ... the victory percentage as part of the fitness function. ...
    (comp.ai.games)
  • Re: GA to move tanks on grid (Advice for newbie needed).
    ... How do I represent my tank movements as a set ... How would they evolve? ... "Application of Genetic Programming to the "Snake Game"":http://www.gamedev.net/reference/articles/article1175.asp ... the victory percentage as part of the fitness function. ...
    (comp.ai.games)