Fix bugs related to plotting

[andrea-pasquale]

Closes #266.

Checklist:

• Reviewers confirm new code works as expected.
• Tests are passing.
• Coverage does not decrease.
• Documentation is updated.

[codecov]

Codecov Report

Merging #269 (b55bac8) into main (b48a3f5) will decrease coverage by 0.32%.
The diff coverage is 52.63%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #269      +/-   ##
==========================================
- Coverage   36.26%   35.95%   -0.32%     
==========================================
  Files          19       19              
  Lines        1456     1452       -4     
==========================================
- Hits          528      522       -6     
- Misses        928      930       +2     

Flag	Coverage Δ
unittests	35.95% <52.63%> (-0.32%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/qibocal/plots/allXY.py	10.98% <ø> (ø)
src/qibocal/plots/flipping.py	15.90% <ø> (ø)
src/qibocal/plots/rabi.py	5.97% <ø> (ø)
src/qibocal/plots/ramsey.py	15.90% <ø> (ø)
src/qibocal/plots/spectroscopies.py	4.71% <0.00%> (-0.04%)	⬇️
src/qibocal/plots/spin_echo.py	15.90% <ø> (ø)
src/qibocal/plots/t1.py	15.90% <ø> (ø)
src/qibocal/fitting/methods.py	71.42% <90.90%> (-0.59%)	⬇️

[stavros11]

Thanks for fixing these issues. I tested after merging this to qmsweepers branch and spectroscopies, Rabi amplitude and length and allXY work for me.

For Ramsey I am still getting an error, now in builders.py update_platform_runcard line 282, that int cannot convert 1D array. It may be related to the fact that fitting failed and fits.csv does not have data for all qubits. It may be good to make this robust to fitting failures, because Ramsey fit fails quite often but we would still like to get the report.

I am now trying T1.

EDIT: More specifically what I have in mind is that update_platform_runcard checks if fitting data exist for the qubit. If yes it proceeds as now, otherwise it could print a warning like "Fitting data for qubit {qubit} not found. Runcard was not updated." and proceed with the code without failing.

EDIT2: T1 seems to work too. So far the issue remains only for Ramsey.

[andrea-pasquale]

For Ramsey I am still getting an error, now in builders.py update_platform_runcard line 282, that int cannot convert 1D array. It may be related to the fact that fitting failed and fits.csv does not have data for all qubits. It may be good to make this robust to fitting failures, because Ramsey fit fails quite often but we would still like to get the report.

@stavros11 you mean ramsey_frequency_detuned, right?
The routine is pretty complicated and not optimized for multiplex I think.

[stavros11]

@stavros11 you mean ramsey_frequency_detuned, right? The routine is pretty complicated and not optimized for multiplex I think.

Sorry, I just saw this comment, I replied to the issue first. No, I am only using the normal ramsey, I have never used the detuned one. Even from before I remember that the fit for ramsey rarely works and even when it does the fitted curve is completely off the real one.

[andrea-pasquale]

This was the bug. For some reasons, when the fitting fails we were setting all the keys to 0 (including the qubit key). This can be very dangerous since you could have two columns with the same qubit.

[andrea-pasquale]

Thanks for the clarification.
I was able to spot the bug (see my comment).
Let me know if it works now @stavros11.

[stavros11]

Thanks for the update, it appears to work now. Ramsey fit still fails for most qubits but the report is still generated.

[andrea-pasquale]

Coverage is not passing because we are missing some tests for fitting methods.
Given that this is out of the scope of this PR I suggest to merge this now.
In the meantime I will open an issue for the missing tests.