Display plots from different URIs

[mfiaudrin-epsyl]

This PR aims to use interpolation in order to allow users to display several plots from diffent URIs in a same graph.

[read-the-docs-community]

Documentation build overview

📚 imas-ibex | 🛠️ Build #32654506 | 📁 Comparing 260df2f against latest (0c56cfe)

  🔍 Preview build  

48 files changed · ± 48 modified

± Modified

• index.html
• users_manual/features.html
• users_manual/how_to_launch.html
• developers_manual/backend_development/adding_new_data_source.html
• developers_manual/backend_development/adding_new_downsampling_method.html
• developers_manual/backend_development/api-hidden.html
• developers_manual/backend_development/api.html
• developers_manual/backend_development/backend_development.html
• developers_manual/backend_development/backend_development_introduction.html
• developers_manual/backend_development/benchmarking.html
• and 38 more...

[mfiaudrin-epsyl]

Hi @jwasikpsnc,

I am implementing the feature to display several plots from different URIs using your interpolation functionality.

I have encountered a bug with the interpolation calling data/plot_data (Code 500). There are my params:

• uri: imas:hdf5?path=/home/ITER/tribolp/public/imasdb/ITER/3/135011/98#pf_active:0/coil[:]/current/data
• interpolate_over: imas:hdf5?path=/work/imas/shared/imasdb/ITER/3/135011/7/#pf_active:0/coil[:]/current/data

And there is the full route:
http://127.0.0.1:52879/data/plot_data?uri=imas%3Ahdf5%3Fpath%3D%2Fhome%2FITER%2Ftribolp%2Fpublic%2Fimasdb%2FITER%2F3%2F135011%2F98%23pf_active%3A0%2Fcoil%5B%3A%5D%2Fcurrent%2Fdata&interpolate_over=imas%3Ahdf5%3Fpath%3D%2Fwork%2Fimas%2Fshared%2Fimasdb%2FITER%2F3%2F135011%2F7%2F%23pf_active%3A0%2Fcoil%5B%3A%5D%2Fcurrent%2Fdata

I share you a screenshot of the error, it looks that string type is not handle by the interpolation:

Can you please do some investigations and fix it?
Thanks !

[mfiaudrin-epsyl]

Hi @jwasikpsnc,

Is it possible to also have the interpolate_over parameter in the data/field_value route ?

[jwasikpsnc]

Not at the moment. What do you need it for?

[mfiaudrin-epsyl]

Ok, at the moment I can use data/plot_data since it's not blocking.

I would like to call a first time the interpolation with data/plot_data to have the new selected plot to include in the graph with the new common coordinates.

I will then have to update the other plot(s) of the graph to also interpolate them. I would like to do it by calling data/field_value because I just need to have the data of the plot since I already have the coordinates.

I think it would perform a little better since less data will be send. But we can discuss of that later if it's a lot of works for an imperceptible improves.

[jwasikpsnc]

data/field_value handles only data (without coordinates) extracted from IMAS database.
In order to interpolate data i need original coordinates that I already have in data/plot_data endpoint logic.
Getting access to coordinates in data/field_value would need extensive development and would make this endpoint too similar to data/plot_data, thus it would be better to use plot_data in this case.

After all, AFAIR you use data/field_falue to extract error bands.
Before, we didn't had to worry about coordinates, but now we have to.
In this case I think it would be reasonable to use plot_data to get error bands using logic simillar to this:

if not interpolate:
    error_bands = backend.field_value(...)
else:
    error_bands = backend.plot_data(interpolate_over=...)

[mfiaudrin-epsyl]

Hi @jwasikpsnc,
The interpolation isn't working as expected when I try to get error bands.

Here is an example where I want to get an ip_error_upper data with an interpolation based on A/ip, B/ip calling /data/plot_data:

• uri : imas:hdf5?user=public;pulse=100002;run=1;database=iterdb;version=3#equilibrium:0/time_slice[:]/global_quantities/ip_error_upper
• interpolate_over:
  • imas:hdf5?user=public;pulse=100002;run=1;database=iterdb;version=3#equilibrium:0/time_slice[:]/global_quantities/ip (this one is probably useless because A/ip_error_upper and A/ip should have same coordinates)
  • imas:hdf5?path=/home/ITER/tribolp/public/imasdb/ITER/3/135011/98#equilibrium:0/time_slice[:]/global_quantities/ip

The same problem occurs for ip_error_lower.

The error received is (466 Code): Error: IDS name and node path should be the same for source and target URI when interpolating data.

It looks that the interpolation is only working for same paths so error_upper and error_lower aren't interpolables.

Can you do some investigations and fix it ?

[jwasikpsnc]

Hi @mfiaudrin-epsyl
This is intended behavior.
You should put URI1/time_slice[:]/global_quantities/ip_error_upper in uri argument
and URI2/time_slice[:]/global_quantities/ip_error_upper in interpolate_over.
Both URIs must point to the same path.
Do not repeat the same URI in both arguments.
Please let me know if this answer finds you well.

[mfiaudrin-epsyl]

@jwasikpsnc Thank you for your feedback, it should works when all URIs have error band data.
Unfortunately in this case there is no error band data in URI B so I can't interpolate the error bands from URI A.

Do you think we can Add a BE rule to allow to interpolate with a different path only when the difference is at the end with "_error_upper" or "_error_lower" since coordinates should be the same between the plotted data and his error_upper / error_lower ?

[jwasikpsnc]

Ok, now i understand
I will open new issue

[mfiaudrin-epsyl]

Hello @olivhoenen,
I have implemented the E2E test to ensure that coordinates between 2 plots from different URIs have a common one.
I set the PR to ready so that you can test it and approves it.

To remind, this feature allow the user to plot several plots from different URIs in a same graph and compare them in a scientific way thanks to the union of coordinates to have a common ones.
Feel free to test with different data. There is the ones I used for this implementation:

• URI 1 : https://zenodo.org/records/17062700/files/iter_scenario_53298_seq1_DD3.nc
• URI 2 : https://zenodo.org/records/17062700/files/iter_disruption_113112_1.nc
• Path : #equilibrium/time_slice[0]/profiles_1d/psi

NB: This feature include interpolation in plus of combination of coordinates. It has been asked by ITER to remove it. It should be done from BE in an other branch. No changes are expected from FE.