Implement flux crosstalk matrix

[stavros11]

I added the possibility to investigate crosstalk in the fluxlines in the existing qubit_flux routine, because it was needed to explain some of the mysteries in the calibration of qw25qB. It can be used by passing the new flux_qubits parameter, for example:

platform: qw25qB

qubits: ["B1", "B2"]

actions:
  
  - id: qubit flux crosstalk
    priority: 0
    # main: t1
    operation: qubit_flux
    parameters:
      freq_width: 20_000_000
      freq_step: 400_000
      bias_width: 0.8
      bias_step: 0.0075
      drive_amplitude: 0.01
      flux_qubits: ["B1", "B2", "B3", "B4", "B5"]
      nshots: 1000
      relaxation_time: 10000

will perform spectroscopy on qubits B1 and B2 (multiplexed) while sweeping the bias of B1, B2, ..., B5 one at a time. Note that this will result to five (in general len(flux_qubits)) executions, but I think the biases need to be swept one at a time in order to extract meaningful results.

If flux_qubits is not given the behavior of the routine will be the same as before. The only difference is the color of the fit (because I changed the order of the plots), but we could force the old color if we need. Also, the colorbar in the crosstalk case is a bit unreadable but I did not try to fix it yet.

The same routine can be used to create avoided crossings:

In this plot we are sweeping the bias of qubit B2 (high frequency) while performing spectroscopy and reading B1 (low frequency).

Checklist:

• Reviewers confirm new code works as expected.
• Tests are passing.
• Coverage does not decrease.
• Documentation is updated.
• Compatibility with Qibo modules (Please edit this section if the current pull request is not compatible with the following branches).
  • Qibo: master
  • Qibolab: main
  • Qibolab_platforms_qrc: main

[codecov]

Codecov Report

Merging #462 (030198b) into main (5fb3700) will increase coverage by 0.06%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #462      +/-   ##
==========================================
+ Coverage   95.95%   96.01%   +0.06%     
==========================================
  Files          61       61              
  Lines        4421     4489      +68     
==========================================
+ Hits         4242     4310      +68     
  Misses        179      179              

Flag	Coverage Δ
unittests	96.01% <100.00%> (+0.06%)	⬆️

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

Files Changed	Coverage
...erization/flux_dependence/qubit_flux_dependence.py	100.00%
...ation/flux_dependence/resonator_flux_dependence.py	100.00%
...rotocols/characterization/flux_dependence/utils.py	100.00%

[andrea-pasquale]

Thanks @stavros11, generally it looks good to me.
You can find below a few comments.
Just a tiny remark. The colormaps tend to overlap when you use more than one flux qubit.

I think that it should be possible to separate the color maps.

[stavros11]

Thanks for the review @andrea-pasquale. I fixed your comments and removed the phase from the plots in order to have the matrix structure, as discussed in a previous meetings. I also implemented the crosstalk matrix for the resonator spectroscopy routine since this was also a TODO. Note that I disabled the colorbar from the matrix plots because I was unable to properly align it in the general case (because the columns of the matrix depends on len(flux_qubits)).

Other than that, this should be ready. I am not sure why the tests/coverage are failing in the CI but doesn't seem related to this PR.

[andrea-pasquale]

Thanks @stavros11.
After relaunching the jobs the CI seems to work.
I have a few small suggestions.
I guess that the first step will be to perform also a post-processing in order to compute the cross-talk matrix. We can do it in a separate. I like already the possibility of visualizing the cross-talk in this way.

[stavros11]

I guess that the first step will be to perform also a post-processing in order to compute the cross-talk matrix. We can do it in a separate.

Yes, it would be useful to add fitting to this routine. I didn't do it yet because it is not very trivial: if {qubit} - {flux_qubit} are not interacting then we should get wide parabolas (ideally straight lines if no crosstalk), which I guess can be used to calculate the crosstalk matrix. However, if {qubit} - {flux_qubit} interact, we may get avoided crossings which should be handled differently by the fitting.

[andrea-pasquale]

Yes, it would be useful to add fitting to this routine. I didn't do it yet because it is not very trivial: if {qubit} - {flux_qubit} are not interacting then we should get wide parabolas (ideally straight lines if no crosstalk), which I guess can be used to calculate the crosstalk matrix. However, if {qubit} - {flux_qubit} interact, we may get avoided crossings which should be handled differently by the fitting.

Right... and also the fitting for the standard protocol (one qubit - one flux protocol) is very unstable at the moment, we should also try to improve the fitting before trying to go for something like this.
Thanks for all the updates!