Parametric Local Studies

I’m hoping to run some parametric studies for both comfort and daylight running on my local machine.

As I understand, we currently cannot setup more than one run per study when running locally.

As a workaround I have used the fly component to generate a single study for each parametric run instead. When I run the simulation, each is saved to a subfolder with the name of the id of that run.

However I run into a problem when trying to fetch the results. The check study status component can only get one study at a time. I planned to use fly again to cycle through the subfolders, however this throws up an error as some of the requisite files (inputs.json, job.json) are located either in the parent folder or the subfolder.

What would be the best way to fetch all of the data from my parametric runs? Is there any plans to support multiple runs when running locally? This would be useful in allowing you to bring in the user data etc.

1 Like

I can second @davidjbrownstein on this. I recall when we wanted to report both energy and daylight outputs for a parametric geometry, we realized that multiple recipes can not be integrated into the same run. Even tough we split the study runs into two parts, I remember having a little bit hard time when extracting the results for the identical parametric geometry iterations.

Thanks,

1 Like

Hi @davidjbrownstein and @hlkocalioglu,

Sorry that it took me several days to get back to you. Every time I started to write my answer, something came up. :smiley: It looks like your questions are related, but are not the same.

This shouldn’t be the case. You can have as many runs as needed as part of the study. They will just run one-at-a-time.

Let me try it on my end, and see if I can recreate the issue. As I mentioned, this should already be supported. What happens when you try to run a study with several runs?

In this case, you should run them as two separate studies, load the results back separately, save the results, and merge the generated CSVs using the IDs. This is not an ideal workflow but it should work without any issues.

Hi all, just jumping in because I am also really interested in the answers to these questions.

Regarding the question about running a study with more than one run locally, I always get the following error message when I try:

@davidjbrownstein, have you had the same experience?

@mostapha, I may not be on the most up to date version of Pollination.

Cheers,
Anthony

1 Like

Hi @anthonyschneider, you will get proper answer from the team but from my understanding, when you are running a parametric study locally, you need to do two things:

  • Enable pollinate input (right click on the Pollinate component)
  • Enable fly input (right click on the fly component)

By doing this, you should be able to run multiple studies locally without any issue.

Hope this will be helpful,
Levent

2 Likes

Thanks all for the responses. I get exactly the same result as shown in @anthonyschneider screenshot. When using data recorder to plug in a number of runs, it shows this as not being possible.

I could do what @hlkocalioglu suggests, which will create separate studies for each iteration (rather than runs), but then I run into the issue of fetching the results as they are all separate studies. Whereas I can only get the results of one study at a time.

1 Like

Thank you all! I had to reset my memory to remember how the local runs are set up. You’re correct that you can only schedule

Supporting more than one run in study takes some reworking of the code that schedules and runs the studies. I haven’t checked the code for some time and I can’t really help with giving an estimate on how much it would be or when we’re going to get around it. That said, you should be able to run the studies, and load the results from several studies, and compile them back together.

I tried putting an example together and I was able to set up everything correctly for the daylight simulation but I can recreate the error for job.json for energy simulation. Let me check this with @mingbo and get back to you.

P.S.

I’m well aware that our efforts for developing better solutions for running simulations using Pollination hasn’t been the best in the past year or so. The main reason is that we have shifted our main focus on the geometry preparation, and we have suggested everyone to use Ladybug Tools for running the studies. Sorry about that! It is something that we had to do. It will take us some time but we are working our way back into integrating the simulation workflows on top of all the geometry work that we have done. We will start from adding them to the Model Editor, and slowly but surely make our way towards Rhino/Grasshopper plugins. Thank you for being patient with us as we go through this process.

2 Likes

A quick update: @mingbo and I decided to make some improvements to the local workflow before posting it here. Namely, the List Studies component will return the list of local studies, so you don’t have to create the folder names. We’re also fixing the issue with the missing job.json file for failed studies.

2 Likes