galapagos: genome weight

hey guys!

is there a way to give the genomes different weights in evolution? i tried to multiply or divide them. but i was wondering if there is a way to rank them.

for example: #1 minimal coordinate diffenence between a given and a searched point

#2 shortest way to connect the searched point to another point

thanks

Replies to This Discussion

Permalink Reply by madmax on October 10, 2011 at 2:48pm

please help!

Permalink Reply by Joshua Seskis on October 11, 2011 at 3:18am

Sounds like the travelling salesman is in town.

Create a F(X) with 2 variables. Set the first variable as your distance apart, set your second variable as the shortest way.

The function f(x) = variable1 + variable2/100

Make the output of that module your fitness. This way when you get your results in Galapagos they should better reflect the distances apart rather than the shortest path.

It will help if you upload the file to look at. David made a great guide to understanding genetic algorithms and they can be found here.

Permalink Reply by madmax on October 11, 2011 at 6:17am

thank you!

but the variables in your funktion have the same weight,

maybe this way:

The function f(x) = variable1 + (variable2*2)/100

the variable2 has twice the weight of variable1

or:

The function f(x) = variable1 + (variable2*2)/3

what do you mean?

PS: what do you mean with: "Sounds like the travelling salesman is in town."?

cheers max

Permalink Reply by Joshua Seskis on October 11, 2011 at 6:33am

'variable 1' = a

'variable 2' = b

fitness = a + (b/100)

Thats what I was suggesting. The travelling salesman is a problem used to test optimisation procedures, it sounded like you were trying to set one up. http://en.wikipedia.org/wiki/Travelling_salesman_problem

Permalink Reply by madmax on October 11, 2011 at 6:51am

this was jutst a randaom example...

the the salesman problem sounds interesting.

thank you!

RSS