Looping the whole generative design study

Not sure what you are after, but perhaps running more generations would work?

Can you give me some more insight as to why?

I’m basically trying to solve the same problem I had in a different post but in a different and more efficient way. I want the GD study to run all possible options for a specific Number Slider. I thought about creating a loop and running the generative design study for every possible option in this Number Slider. On a side note, I tried increasing the population size as well as the # of Generations, I still don’t get all possible options.

If you are using optimize than won’t ever see ALL possible generations, as only the best performing make it into the hall of fame (what the displayed options are referred to as).

Try setting the building index selector to be a static input of say 0 and vary all the rest, then do the same study but set the building index selector to 1, and again for 2, 3 and so on.

Yes that would help, but I find it to be inefficient solution. Again, is it possible to loop the whole GD study? I’m not sure if that’s even a doable thing.

I mean like to integrate it in an external code…

Any idea Jacob? I value your opinion and I look forward to hearing from you! Thank you for your continuous help!

I have one thought but will have to spend some time hacking apart the system to see if it is even feasible. Hope to have the time to do so this weekend. :slight_smile:

1 Like

That sounds great! I’m so excited for it :slight_smile: Thanks again!

Depending on how many options you have for that one variable things could get reaaaaal messy, but you could always remove the slider and have your graph run all of them by default. You’d have a list of outputs to isolate in this case, but then you’d always be looking at every variation of that option with each solution.

Yes I had a list of the options, until I realized that the output of the GD study can’t be a list!

Correct. You’d have to isolate each item in the list as a singular output. That’s what I meant when I said it could get messy fast. I don’t know how many options you currently have on your number slider, but you’d have to duplicate at least that many outputs.

I currently have around 20 options (but expecting to have more than 200 later on!!).

Yup. Ignore my idea.

However I really think you need to rethink what you’re after here. If you’re going to have 200 options on a single variable and you’re going to want to have outcomes with all 200 variations for that one variable while still randomizing other variables… that’s a lot of iterations. I believe we talked about this before and it seems like you’re trying to run the same optimization on multiple objects at the same time. Which just isn’t efficient.

1 Like

Doable, although it may be messy. Basically multiply the number of outputs on this graph by 50 or so:

1 Like

Imagine how inefficient this process is when manually done! That’s why I thought about Generative Design hoping to get all options done at the same time!

You meant to multiply the number of outputs by the number of options that I have, right?

Yes - and run all options for each study at once.

Yes, but this will make my graph x200 bigger!

No- just adds 199 new watch nodes and a codeblock to the graph for each output.

Overall compute time on that will likely be better as well.

1 Like

will try that! Thank you