Re: Evovlers and fitness functions.

Hi Cristian

On Sep 11, 3:01 pm, Cristian210273 <cristian210...@xxxxxxxx> wrote:

In this case, the max fitness a warrior could obtain should
be:

(30*3)*100+(30*3)*100=18000

Using this fitness, I obtain warriors that gain about the 52%
of the total score.

Equivalent to a hill score of 156, sounds fine. Your opponents
will get < 48% of the total score on average.

Running more battles would reduce the score noise.

Do you think that the fitness I use is wrong? Or should I search
for the problem elsewhere (in the evolution algorithm or in the
benchmark warriors I use)? If the problem is in the fitness
function, can you suggest me a more suitable algorithm?

Have you tried submitting anything to the nano hill yet?

Cheers,

sf

--
http://sfghoul.blogspot.com / http://corewar.atspace.com

.