objective function: use of info not in genome, and dependence on other genomes

Ramon Diaz-Uriarte ramon-diaz at teleline.es
Mon Nov 6 08:00:30 EST 2000


Dear All,

I am having trouble figuring out how to adapt galib to my specific problem.  In
each generation, each "individual" can fight with others a specified number
of times.  Fitness is not a function of a genome by itself, but depends on
the other genomes too (a typical game theory problem); fitness is a function of
the outcome of these encounters; the outcome is a function of the genome, plus
some other variables (and the values of these other variables are NOT
inherited, and are themselves affected by the outcome of the fights).  I am not
sure how to:

1. Code and use those other variables that affect fitness and need to be
modified within a generation. To evaluate each genome I need to take into
account both the genome itself and those other variables.

2. To assign fitness, I need to use an objective function that will be passed
the genomes plus the other extra variables of each
individual,  and then (according to a set of pre-specified rules for fights
between individuals) will return the fitness (and change the values of those
other extra variables). I do not see how to do this.

Thanks,

Ramon


-- 
Ramón Díaz-Uriarte
Triana 47
28016 Madrid
Spain

email:ramon-diaz at teleline.es
Phone: +-34-918-513-966
       +-34-657-186-407	











More information about the galib mailing list