Avoid using calibration parameters as input in obscure ways

[RoyStegeman]

In spectroscopy experiments the pulse sequence is constructed using the MZ gate duration and amplitude parameters. While a square pulse is fine, taking the duration and amplitude from the parameters of the MZ obscures what pulse is actually being sent. In the case of resonator punchout attenuation the amplitude is for example the one from MZ while it might make sense to set it to 1.0 or expose it as an input parameter. The duration is taken from this gate as well, which instead should probably an input parameter with sensible default as well or even just e.g. fix it to a microsecond.

I open this issue instead of a PR since I don't know if this experiment is the only such case and I think it would be good to go through all experiments to ensure there are no other situations like this.

[alecandido]

I agree: this is certainly a problem, and we generally need an extensive review of how the content of Parameters and Calibration (through CalibrationPlatform) is affecting all the experiments.

Tip

Just as an example, following up on your comment about spectroscopies, using the duration from the MZ is completely unsuitable. Spectroscopies are frequency-domain experiments, and they are best executed with continuous waves.
Instead, the MZ is the calibrated pulse used for the readout, which we may want to shorten as much as possible to implement a fast reading process (e.g. for mid-circuit measurements).
In this case, there are two different problems:

• the MZ duration is inadequate for this experiment
• in the usual workflow, it could also be just absent, since the frequency-domain measurements usually anticipate the pulsed ones

However, let me rephrase the general principle

Important

We should expose all parameters related to the specification of an experiment, to make it fully flexible for the user.

But it makes sense to avoid making all of them mandatory, using sensible defaults, possibly matching previous measurements.

E.g. for the resonator spectroscopy it makes sense to use the following defaults:

• if using a pulse (instead of a continuous wave), make it as long as possible (ideally this is driver dependent)
• if the resonator frequency is already known, center the scan around that frequency
  • since the frequency may depend on the power level, select it base on the specified power level, if the regions have already been calibrated identifying the power shift

Instead, no default power should be used, since there is no frequent choice for that (it depends on whether the user is looking for the high or low power regime, and that is fully up to the user itself).

[jevillegasd]

As a user, I prefer if the routine defaults to the parameters stored in the gate specification, and with the additional option to override either of their settings. In general I do often re-run spectroscopies often to "see" how the MZ pulse looks like before trying to optimize readouts.