Minimization using Galapagos converges to a larger value than the minimum of all iterated fitness

Replies to This Discussion

Permalink Reply by Robert Vier on February 9, 2014 at 10:22am

you did set the solver to minimize your fitness value, did you?!

Permalink Reply by patric guenther on February 9, 2014 at 11:22am

even for a maximization it would not be the optimized solution. how do you calculate the fitness value? maybe something wrong with your equation....? would you like to share your gh file? 

Permalink Reply by Pil Lauridsen on February 10, 2014 at 1:02am

thank you for the fast reply.

My equation is quiet heavy! I'm simulating indoor climate using MATLAB and one simulation loop runs for about 1 minute. Last night I stop the solver cause the fitness started to fluctuate in a pattern. Like a standing wave with big amplitude :(

See figure:

 

 

Before going to bed I start a new optimization where:

 

Population: 20

Initial boost: 5

Maintain: 10% (thought maintaining more of the good solution would help the solver in the right "direction")

Inbreeding: +90% (cause the best solutions will be near each other)

 

After 8 hours the fitness iteration look like this figure:

Permalink Reply by Pil Lauridsen on February 10, 2014 at 12:49am

That thought came through my mind as well. I agree it nearly look like an maximization but I just double check – the solver is set to minimize.