Dev plotting

[michael-hirsch-quelements]

No description provided.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
1 out of 5 committers have signed the CLA.

✅ omrieoqm
❌ Huo Chen
❌ ibarel
❌ michael-hirsch-quelements
❌ YaroslavBorysko

---

Huo Chen seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[TheoLaudatQM]

How is the data corresponding to the residuals retrieved? Is it based on the name or a specific xarray attribute?

[michael-hirsch-quelements]

This was one of the broken tests. Fix and documentation have been updated.

Residuals = data - fit_curve. Fit curve is defined in a FitOverlay, an example and notes have been added to the Readme.

[TheoLaudatQM]

It might be obvious, but it would be worth adding a note that residual will work only if only 1 overlay is provided and if the dataset doesn't have hue, right?

[TheoLaudatQM]

Nice, can I use this framework to define a RefLine whose value depends on a specific qubit quam attribute?
For instance:

def overlay_func(qubit_name, qubit_data, state):
    # Generate overlays based on qubit data
    return [RefLine(x=q.xy.operations["x180"].amplitude) for q in state.qubits]

fig = qplot.QualibrationFigure.plot(
    data,
    x='frequency',
    data_var='amplitude',
    overlays=overlay_func
)

[TheoLaudatQM]

@michael-hirsch-quelements this is what I mentioned in the meeting regarding adding the quam state to the overlay callback function signature.

[TheoLaudatQM]

Currently, the color bars are overlapping when the figure contains multiple plots and the color scale is kind of messed up.

[TheoLaudatQM]

As discussed in the meeting, 2 options should be implemented:

1. No colorbar is displayed (in case the range of each subplot is very different)
2. A unique colorbar is displayed with the min and max from all the subplots (in case the range of the subplots is the same, typically 0-1, -1-+1 or 0-2)

[TheoLaudatQM]

For backward compatibility reasons, and consistency with other processes, I don't think that the definition of QubitGrid should be changed. As shown in the plotting.py files, the current definition of QubitGrid should allow the implementation of a custom layout by doing:

    grid = QubitGrid(ds, [q.grid_location for q in qubits])
    for ax, qubit in grid_iter(grid):
        plot_individual_data_with_fit(ax, ds, qubit, fits.sel(qubit=qubit["qubit"]))

[YaroslavBorysko]

Agreed, QubitGrid stays unchanged. Added two optional helpers (grid_iter, make_qubit_grid_from_locations) for custom layouts and README examples, preserving backward compatibility. Will push them today

[TheoLaudatQM]

Great, thanks!
I tested the latest pushed version which runs without errors, however it seems that

• Matplotlib figures are also created
• The layout doesn't correspond to the initial figure

We can discuss it during the meeting so that I can show you.

[michael-hirsch-quelements]

For the Matplotlib figure: Per Huo, this is to keep backward compatibility since the original QubitGrid object will create a matplotlib figure at initialization and we are using the QubitGrid also for plotly.

And I believe you figured out the layout issue but correct me if I'm wrong.

[TheoLaudatQM]

Gotcha, thanks Michael I'll fix it internally.

[michael-hirsch-quelements]

@TheoLaudatQM I think Huo's latest PR has a fix that moves this functionality into griditer so heads up that we might need to review this together.