Galapagos Ran over 186 times, and still running , Only 8 genomes

Hi, I modeled a room with one Window and a daylightshelf that I've been testing with Diva for daylight analysis. I just tried to use Galapagos , for the Genome I used two sliders, one slider has 2 Diva materials, and the other slider has 4 different heights for a lightshelf. I started Galapagos last night and when I woke up in the morning I found out that Galapagos ran 186 times and it is still running, I thought it should only run 8 times ( 2 materials x 4 lightshelf heights = 8). Am I missing something?

Replies to This Discussion

Permalink Reply by Ninja on April 17, 2014 at 4:25pm

I set the results to be exported using Excel Exporter from TT Toolbox plug-in. I export Genomes and the fitness number. I looked at the Excell file so far and I found out that the same genomes produce different fitness numbers in different iterations. For example in the picture below the first column represents the function number , the second column records the numbers on the sliders(Genomes), I highlighted two rows with the same genomes, however the fitness number is completely different? what is going on?

Permalink Reply by Vicente Soler on April 17, 2014 at 5:45pm

Galapagos will continue unless you set a runtime limit or threshold. If you have only 8 possible combinations, it doesn't make sense to use Galapagos. Change the definition so it solves the 8 combinations and then select the best one.

Permalink Reply by Ninja on April 17, 2014 at 6:46pm

Vicente, thanks for your response. I have 8 possibilities for now just to test Galapagos, but I'm working on a school project that will have about 64 possibilities. Any ideas why the same genomes produce different fitness number though ?

What do I actually put for the threshold ? I'm new to Galapagos .

Permalink Reply by Vicente Soler on April 18, 2014 at 3:19am

The values seem close enough. Maybe DIVA/Radiance uses a non deterministic Monte Carlo approach to calculate indirect radiation so the results vary slightly on every run. You might be able to increase accuracy at the cost of speed by increasing sample count. But I'm just guessing.

If you have 64 different combinations, still do it manually rather than using Galapagos. For Galapagos to make sense to use you need lots and lots or infinite number of possible combinations.

When the fitness value is within the range of the threshold value to the target fitness it will stop.

Permalink Reply by Ninja on April 17, 2014 at 7:25pm

Another question, what is the square-shaped yellow button in Galapagos, next to the word "Display" and next to the four round buttons. When I hover over it , it says " add mutations to the population" I clicked it and nothing happened, what does it do ? and how do I use it? . Also, I'm not sure what the yellow graph on the top display show?

Permalink Reply by Vicente Soler on April 18, 2014 at 3:25am

As it says, it mutates the genes (generates values that go further away from the target fitness) so you get more variation. It's useful if you think you are stuck in a local optima. You can click it many times in a row to mutate them more.

Permalink Reply by Ninja on April 18, 2014 at 5:39pm

Awesome, Thanks.

RSS