your idead are pointing to the right direction. You could use integers as
alleles encoding the cities as you described. This would be what I would
have done. With such encoding you are on the safe side. There are no
performance issues here. Of course the number of chromosomes is an important
factor to consider. Also the number of genetic operators and natural
selectors is important here. But when using the DefaultConfiguration you
should be doing right.
On the JGAP homepage (http://jgap.sourceforge.net) there are quite some
documents helping getting started. If your questions are not answered there,
please don't hesitate sending them to the JGAp list as concrete questions to
be added to the documentation and we will manage to do this.
> -----Original Message-----
> I would be very grateful if someone could provide some
> pointers/help on designing a Chromosome? I have looked at the
> examples and seen how the Chromosomes have been set up there
> and this been a good help however I am trying to represent my
> own types of Chromosome/solutions and I am not sure how this
> should be set up.
> Lets say for example my Chromosome/solution is the ideal
> house for a first time buyer I want the solution to include:
> Number of Rooms
> Number of Gardens
> If it has a garage
> This type of data can be stored as Strings, ints and booleans
> is this how I should proceed with my design? Can it include
> different types? Should I even be using strings if I know
> what the possible solutions can be e.g.
> Address = North City or South City - then should I just
> encode this as an int allele e.g. North City = 0; South City
> = 1. What is best practice?
> Also are there performance overheads I should consider?
> Is there any documents/tutorials on this that I could take a look at?
> Thanks for any help,