It's about time.  Galapagos is wonderful but this looks like it will be a lot more useful.  Nice work!

It looks very interesting! Let me give a try :)

This may sound a bit premature but, Robert will there be some example files in this group related to Octopus usage in next few days?

Thank you for sharing this Rob! Here is my first try...

Hi Mostapha,
Can you tell me more about your testing?
What did you want to achieve?
Thank you.

Awesome! Thanks Robert! Here's my try.

I find it quite difficult to navigate through the results at this stage :) If I may throw in a few ideas, like redefining the axis variables, logarithmic scales, display thresholds, better marking management - or at least add contrast!

Hi Djordje,

I tried to find the optimum form for a dumb building based on 3 parameters (building orientation, and roof angle in both directions), and two objectives (minimize incident solar radiation during the cooling period, and maximize incident solar radiation during the heating period). The graph on the right side of the image shows that Octopus was able to find the Pareto frontier.

I captured a video that shows how I set up this problem for Octopus. You should be able to set-up your problem in a similar way...

Thank you for such a detailed reply Mostapha.
Are those energy components you are using, the ones from the future Ladybug plugin?

To be honest, I haven't tried Octopus, as I do not have Rhino 4, and it currently only works on v5.
What I am confused with is the difference between the Galapagos and Octopus. Why couldn't you use Galapagos instead? Fitness would be sum of -(objective 1) and objective 2, and the Fitness should be Minimized.
?

Octopus is very interesting,and is a nice job.

BUT I can not make the Fitness working as the way I do with Galaogas.

I was wondering if someone can share some example files.

Thank you!

Best,

Lucas.

Hi Djordje,

Yes! I did used Ladybug for radiation studies.

I'm not the best person on the list to tell the differences between Galapagos and Octopus, but generally speaking single-objective optimization is more restricted than Multi-Objective one.

Design is a multi-objective problem by nature and it is not always that easy to simplify all the objectives as one single objective. For example what if I had area of the building as the 3rd objective and view to a point of interest as the forth? How you want to sum them up? (there are of course some way to do that but it's not that easy/straight forward)

There is a great value in having a multiple-objective evolutionary algorithm in Grasshopper that let you explore the result space in multiple dimensions!

Thanks to Robert again!

Thank you for clearing that up Mostapha,
I guess I will just have to wait until the Rhino4 Octopus version comes out, and check it by myself.

Rhino 5 is available as an evaluation version limited to 90 days of full usage.. an upgrade is fairly cheap, so I would go for it since the advantages are quite remarkable.

Mostapha stated it right in principle - weighting the different factors of a single-objective optimization is quite tricky and practically impossible for many problems. Further multi-objective optimization gives you a number of optimized solutions (that are differently weighted trade-offs between the objectives) that you can choose from - so the final decision is up to the designer. Octopus also tries (so far rather simple) to incorporate user-interaction by letting the search be densified near solutions that have been marked 'preferred' when the optimization was paused.

i might make a version for rhino 4 over christmas, before the actual work on octopus starts over.

Thank you for the advice Robert.

Pd: just a little bug, in the context menu for each solution sphere, when the sphere is on the base, the context menu is partly outside of the main window so i cant select some of the options for those spheres

i hope i had explained well............ 

Great work on the solver!  I just downloaded Octopus and enjoyed the improved interface.  We worked on an interface in our office as well, here's the blog post if you're interested.

Thanks Erick! Octopus just uses the interface-parts of Galapagos - it adds a pareto-based fitness assignment and elitism strategy to its GA. So, besides the GA-concept and the canvas-interfacing, its very different from Galapagos. On multi-objective optimization also see http://en.wikipedia.org/wiki/Multi-objective_optimization 

Nice job on using GH and Galapagos cross-platform! also like the processing interfaces, they appear just amazingly clean and fast..!

scopes for future work (going to happen) on octopus are not precisely defined yet. But modular, cloud-based thinking is healthy I guess!

This looks like an interesting alternative to Galapagos. To be clear, one of the main differences here is that Octopus allows/uses multiple fitness functions instead of just the one?

If so, how is this operationally different from me just creating my own weighting of factors and combining them into one fitness equation, like we do with Galapagos now? Is there some additional added strength to separating them out?

Seems really interesting!

Sometimes you can't make a viable fitness function. In my example, I would have to combine very different values in different units (or without unit), different expected range (some may vary by 10%, others may double or be reduced by half), and dependant one on another.

Plus there may be a hierarchy, for instance my first two objectives are absolutely primordial, third one is important but I don't have a clue what its value must be, and then, when I'm happy with those I can focus on the last two.

So the good thing with octopus is that you can pause the evolution at some point an play God, i.e. looking individuals that respect my main objectives, and resuming evolution with those marked as "preferred".

IS there any way to pass a panel instead of slider to Octopus?

You mean to plug in the Panel instead of number slider in "Genome" plug of Octopus? No you can't do that. You need either Gene pools or number sliders for the "Genome" input, as only these tho are able of changing it's values using slider. Panel could not do that.

Thanks for the info. Two more questions:

1) there is no way that I pass the range other than slider?

2) WHAT is the slider in the pareto frontier? Is it narrowing down the pareto frontier?

Hi Nick,
For now you will have to define the range within the slider. So yes, you are right.
Not sure I understood you about the second question. You will get the pareto frontier as a result of your octopus "analysis".

why i´m getting this error when i try to open some of the example file...
octopus component is in the grasshopper components folder, like the other plugins...
thanks

Are you using Rhino4?
Octopus works only on Rhino5.

R5 & Gh 0.9.0014

If your Octopus file is already in appropriate folder (File -> Special Folders -> Components folder), then try unblock it (Right-click on it -> Properties > unblock it).
Will that help?

Djordje, i did it, but i still have the same error...

i restart rhino, and now it works, thanks!

I'm having trouble getting Octopus 3.0 or 3.1 to run in GH 0.9.0061. I believe I have it set up correctly, but double-clicking on the component seems to have no effect whatsoever. The file isn't blocked... Any thoughts anyone? Does Octopus run in 0.9.0061?

Thanks.

Ok, solved that issue. The .gha and .dll were in: 

C:\Program Files (x86)\Grasshopper for Rhino 5\Components

but they need to be in:

C:\Users\Rob\AppData\Roaming\McNeel\Rhinoceros\5.0\Plug-ins\Grasshopper {A4.....CF}\0.9.61.0\Components

Octopus is running perfectly now. What a cool plugin. Thanks!

Hi,

I downloaded version 3.1 and I can't get it to work. I have put files HelixToolkit.Wpf.dll and Octopus.gha in C:\Users\"username"\AppData\Roaming\McNeel\Rhinoceros\5.0\Plug-ins\Grasshopper {B45A29B1-4343-4035-989E-044E8580D9CF}\0.9.64.0\Components\

Octopus appears in the ribbon, and the example files load correctly, but when I click start (calculation), it does nothing. 

Thanks

You need to unblock the .dll file.

-A.

Hello Abraham, 

the files are unblocked. I can use the Octopus component and upon opening the example files I can see the results of the default calculation. But when I click Start again the log file 01_singleObjective_Diversification_OctopusErrorLog_20131119.txt  says:

Error, reduction of environmental selection did not yield the correct archive size

I have previously used Octopus 0.1 flawlessly.

OK, it seems to be working. Only thing is on each start I only get one generation, instead of many. I guess I have to mess around with the options. 

Also the text is very difficult to read. I didn't realize until now that there were numbers on the fields. 

Is there any version of Octopus that works with Grasshopper 0.9.0064? I can load the component, but it crashes whenever I delete it from the canvas.

Ha I'm getting an "Awesome" tab when I load the Octopus.e gha. What is this thing?!

Oh this is a mistake in compiling the plug-in, this component was purely for development, and not supposed to be in the official release.

Sorry for the confusion,

Best

Robert

Haha no prob. You should keep the Awesome tab for the next release. Thanks for all your work in this area.

Can Octopus pick the best solution itself? Where?

How does Octopus deal with multicriteria weighting and multicriteria scores?

What is the algorithm to calculate hypervolume metric? Is it HSO?

Does the latest version of Octopus work with Grasshopper 0.9.0076.... download page says it is compatible with 0.9.0075.

If only works with 0.9.0075 any tips on where to find thta version of Grasshopper? 

Thanks.

works with the latest version, too!

Thanks Robert!

Hello Robert, Hope everything is great with you, I have a little problem that the Convergence Graphs and Hypervolume Graph  don't appear in the optimization process ... I designed an office unit with a shading element. The parameters are the geometrical properties of the shading element. The targeted performance is DAYLIGHT AUTONOMY, PMV AND ENERGY CONSUMPTION , I used DIVA for that. So What do you think might be the problem.
P.S. when I run the GH files came with the OCTOPUS . it works just fine and both graphs appear. Thanks in advance 

Israa,

you are in generation 0. The graphs will be shown in the next ones.

Best

Ok (Y) , Thanks a lot :) 

I have a question concerning EVALPOOLSIZE ... what does determine its values.. i made population size 4 and max number of generation is 3 ... does the values of these parameters affect it !.. I really don't understand it clearly 

Thanks in advance

I have another conern my simulation does not go far than ZERO generations .. Despite i made it 3 and population size 4 .. just as a test ... Could you think what might be the problem !

Hello Robert Vier, hope you are doing great 

I have a little serious problem, my optimization collapsed at the11th generation, where it should continue till 15 generation. The octopus saved my last 10 generations "no problem about that" but corrupted my 11th, any way I can start from generation 10 and re continue my optimization process , THANKS IN A MILLION