Grasshopper

algorithmic modeling for Rhino

Hi,

I have a real problem with reopening saved simulations and extracting phenotypes because the gene values changed.

I ran a simulation for 29 hours and checked the genes right after the simulation finished and they matched the range of the slider values, since I ran into this problem before but thought I fixed it.

Today I reopen the saved file and the genes have changed. The problem seems to only affect sliders with floating point numbers, e.g. a slider/gene pool from 0.00 to 1.00 gets changed to numbers from 1 to ~100.

The consequence is that I'm not able to export phenotypes, since it uses wrong genes, the exported solutions look totally wrong and they infringe my intersection constraint.

My guess is that the number stands for a floating point number inside the range of the slider values, so 35 in {0;0;0} of the first screenshot means 0.35, so I would be able to recreate the right genes.

The only workaround I can think of to generate the real solutions is to use Wallacei Analytics + Selection to find the the solutions I want and recreate them using their Branch + recreated slider values.

Since the file exceeds the limit in the forum I uploaded it here: https://www.dropbox.com/s/f5my1wt2708m17i/190808_Wallacei%2BGalpago...

Greetings from Weimar,

Daniel

screenshots:

Views: 40

Replies to This Discussion

Hi Daniel,

interesting problem. I downloaded your file and inspected it.. indeed something doesnt look right. I tried to recreate the problem on my machine but was unsuccessful. I copied your definition to a new GH file and ran a small simulation using your definition (generation count 4;generation size 4), at the end of the simulation i checked the genomes and everything was fine (which presumably was the same for you), i saved the file, restarted rhino and opened the file again, exported a solution from wallacei and the genomes still matched.

can you give me some more info on exactly the steps you took (i.e. did you modify the file in anyway after running wallacei? did you reopen the file another machine?).. essentially any information that would help me recreate the problem... dont worry about being very specific.. actually the more specific the better.

in the meantime however! and to make sure you dont have to run the simulation for another 29 hours, i edited your definition so that you can recreate your phenotypes by extracting a solution's genome. I edited the genomes so that the decimal numbers are back to what they should be... essentially all you need to do is select the generation/solution number you want using the sliders i highlighted in red (at the beginning of the definition) and the correct phenotype should be recreated. This should help you continue your research while we attempt to recreate the problem and resolve the issue. the edited definition can be downloaded through the link below:

https://www.dropbox.com/s/kwaq8ywvyqrklp5/190808_Wallacei%2BGalpago...

Hi Mohammed,

thank you for the fast answer :)

I came across this post Decode genome, not recognizing decimals and think that could be the problem, since I'm located in Germany and use a german version of Windows, that should also be the reason why you can't recreate the problem. I changed my decimal seperator now and will test it after lunch and post an update.

I didn't modify the file in anyway after running wallacei, but I copied it to another machine. After realizing there is a problem I opened the original file on the original used machine and had the same results.

Thank you for the file, I did it in a simlar way and was able to recreate the solutions but noticed that the intersection constraint is ignored a lot (eg. Gen 99, Solution 30). I think I will replace it with a test for colisson instead and hope that works better, but I will have to start a new run again sometime tomorrow.

great.. looking forward to your update.

I tried it first with "." as seperator (4 gen count, 4 gen size), saved and opened again and the numbers matched.

To be sure I then went back to "," did the same cyle and after opening the saved file the numbers didn't match.

So I guess it's save to say that this is the solution :)

Great! We actually addressed this issue in an earlier release of Wallacei (where region (ie decimal separator) doesn’t impact the genomes) but it must have slipped through the cracks in later versions. We will permanently fix it in the next update (the next update is scheduled for the coming few weeks). thanks again for raising this !

Glad I could help :) looking forward to the next version!

RSS

About

Translate

Search

Photos

  • Add Photos
  • View All

© 2019   Created by Scott Davidson.   Powered by

Badges  |  Report an Issue  |  Terms of Service