Galapagos Day 02

Another test run with the Galapagos Evolutionary Solver in Grasshopper (something I'm writing in my free time). This time I've picked an obvious problem (Volume of Box 1 minus Volume of Box 2). Thus, we expect the first box to become as large as possible and the second box to become as small as possible.

With a population of 30 Genomes, a solution very close to the ideal solution is found within 10 generations.

7 members like this

Get Embed Code

Tags:

You need to be a member of Grasshopper to add comments!

Comment by Amir Mosavi on December 18, 2010 at 4:34am: Dear David,

Is it possible to run multiple Galapagos Evolutionary Solvers at the same time in order to include more than fitness function?

Thanks,

Amir

Comment by Amir Mosavi on December 16, 2010 at 3:55pm: Hi David,

Could you please let me know the material of the Objectives in Galapagos? Do these Objectives arise just from the geometry? I mean would it be able to consider optimization objectives such as stress, fatigue, pressure? How about the optimization constraints?, Could you give me more info in this subject? How about increasing Objectives and variables and constraints?

What do you think if we try to integrate other already developed optimization packages to GH e,g. IOSO, Modefrontier or NIMBUS? or even utilizing the implemented advanced GA-based optimization cods such as NSGAII?

How do you see the future of Galapagos? are you going to integrate it with CAE packages including FEA and CFD?

Thanks,

Amir

Comment by William Lopez Campo on November 18, 2010 at 5:25am: Hi David,
Cracking idea!!!
I'm playing with it in GH 0.8 and finding it very interesting.
One thing I wanted to ask, are you considering multi-goal optimization (more than one fitness variable)?

I've posted a link in my blog to this article (http://undernda.blogspot.com/2010/11/lot-of-exciting-news.html) I hope you don't mind, please let me know if you want me to remove it.

Keep up the good work, this is mind-blowing and ground-breaking!
William.

Comment by Claudio on January 22, 2010 at 11:16am: sweeeet

Comment by tim williams on January 21, 2010 at 9:55pm: This would be incredible. I cannot wait.

Comment by Morphocode on January 19, 2010 at 3:35am: just great!.....as usual ;)

Comment by Tuan N. Tran on January 18, 2010 at 9:31pm: schweeeeeet!

Comment by Moritz Fleischmann on January 18, 2010 at 12:22pm: I have to agree with Daniel on this one, David.
Very interesting.

Comment by Daniel Piker on January 18, 2010 at 9:16am: Awesome!

Comment by David Rutten on January 18, 2010 at 2:44am: Hi Svetlin,

it's only just been running for a few days. I still need to write more advanced mating and genome-avoidance algorithms, as well as a lot of interface code. Also the Galapagos Object needs to be made a lot smarter and user friendly.

I doubt I'll release this publicly prior to Grasshopper 0.7, which I'm working on now. That is going to take 2~3 months at the very least. Also, since I'm doing Galapagos in my free time (there's not a lot of that), it's very hard to predict how long it's going to take. I'm guessing about a week of development is still needed to make it one-point-oh-worthy, but that week will be scattered over several months.

Galapagos is completely separate from Grasshopper and Rhino however, it's just a loose dll that can be used by any DotNET process, so the core library release is not tied to Grasshopper release cycles.