Feature/qualibration graph 2 q gates cr rand cz

[KevinAVR]

This is the initial point of the Pull Request of CR and CZ (ZZ-driven) nodes into the main branch

[soonqm]

I don't think 31c is needed, the cancellation phase is known analytically. If there is any reason why it deviates, we should instead do a 2D scan of cancellation amplitude and phase, since changing only one does not lead to minimization.

$\phi_0$ and $\phi_1$ can be obtained from 31b

[soonqm]

Currently, the protocol seems fine, some analysis is there but no fitting and state update. This needs to be added, preferably from testing on actual device.

^related to the calibration protocol: Since we are defining different CR scheme, i.e. "direct", "direct+cancel", "direct+echo", "direct+cancel+echo", we require slightly modify protocol for each. Either we dump the option or somehow have a conditional calibration protocol. A quick example for the above is that for "direct" and "direct+cancel" scheme, the CR implements a pi-pulse, while the "echo" is only pi/2 which meant that we might need an additional duration routine.

[HiroQM]

I don't think 31c is needed, the cancellation phase is known analytically. If there is any reason why it deviates, we should instead do a 2D scan of cancellation amplitude and phase, since changing only one does not lead to minimization.

ϕ 0 and ϕ 1 can be obtained from 31b

@soonqm
In principle 31c indeed seems somewhat redundant. However, the actual experimental data can deviate from the procedures/theory described in the paper. Thus, there's an exhaustive list of scans including 31c. As you suggested, it might be a good idea to add some of 2D scan as protocols 32s. Besides what's already there, in fine tune-ups, we would need to add more protocol such as error amp.

[HiroQM]

31g is empty, will proceed to delete for now? I think this might be related to error amplification with respect to amplitude based on https://quantum-machines.atlassian.net/wiki/spaces/CST/pages/3746693177/Developing+Calibration+Graph+for+Cross-Resonance+CR. However, note the CR amplitude will change the calibration of cancellation pulse, so I think this will require a bit more consideration anyhow.

@soonqm
Indeed. Let's remove 31g for this PR. We'll implement this more properly later.

[HiroQM]

Currently, the protocol seems fine, some analysis is there but no fitting and state update. This needs to be added, preferably from testing on actual device.

^related to the calibration protocol: Since we are defining different CR scheme, i.e. "direct", "direct+cancel", "direct+echo", "direct+cancel+echo", we require slightly modify protocol for each. Either we dump the option or somehow have a conditional calibration protocol. A quick example for the above is that for "direct" and "direct+cancel" scheme, the CR implements a pi-pulse, while the "echo" is only pi/2 which meant that we might need an additional duration routine.

@soonqm
Yes, developing the analysis tools to pick the optimal parameters is the TODO. This could be done with rather straightforward min/max-search type of algorithm or possibly require more involved methods such as machine learning.