Grafted List as Galapagos Fitness ?

Hi Guys,

I plugged a grafted list of values into the fitness input of Galapagos (please see below) and would like to reach 40 for each of the values.

Although Galapagos is running fine, I don't think it is considering the multiple values in the list, it might be flattening the whole list...? Is there a way of making it work for all the values ?

Many thanks,

Arthur

Replies to This Discussion

Permalink Reply by Manuel on September 21, 2011 at 5:34pm

if you want that those values to be 40, then make a substraction between 40 and that list you have (flatten first). Then use the mass addition on the substraction result and set galapagos to "Minimize".

When all those value reach 40, then the substraction will be zero for all values and the mass addition will be zero (galapalos will try to get that).

Permalink Reply by David Rutten on September 22, 2011 at 3:17am

When you feed multiple values into Galapagos, it will average them first, then minimize/maximize/optimize the average. You won't be able to make all values be 40 this way, as {40,40,40,40} would be equally fit as {30,50,25,55}.

David Rutten

david@mcneel.com

Poprad, Slovakia

Permalink Reply by Arthur Mamou-Mani on September 22, 2011 at 3:43am

Thanks a lot David and Manuel,

Will Galapagos average the values for both a flattened or a Grafted list?

The way I solved this for now is counting the number of items which are more or less equal to 40 through dispatch and list length (below) and maximize the Galapagos result.

Permalink Reply by David Rutten on September 22, 2011 at 6:15am

All supplied values will be averaged. Galapagos can only operate on a single number, so it has to either ignore values or take them all into account.

David Rutten

david@mcneel.com

Poprad, Slovakia

Permalink Reply by Arthur Mamou-Mani on September 22, 2011 at 11:41am

Thanks David, finding a function to introduce several Fitnesses is fascinating...

Permalink Reply by Manuel on September 22, 2011 at 10:45am

yo can use booleans as number (false=0, true=1), then use a mass addition and set it as fitness (if you want the true values then set galapagos to maximize)

Permalink Reply by Arthur Mamou-Mani on September 22, 2011 at 11:40am

Thanks Manuel, that's very good, runs much faster.

Permalink Reply by Andrew Wallace on November 22, 2011 at 6:35pm

I mass add the difference and then solve for the minimum.

Permalink Reply by David Rutten on November 23, 2011 at 8:57am

The latest version of Galapagos can actually handle multiple fitness values. It will simply compute the average of all supplied values. This is not exactly the same as mass-addition, but the difference is only important when the number of values changes between genomes.

David Rutten

david@mcneel.com

Poprad, Slovakia

RSS