Incorrect parameters in the RB runcard example

[biankawolo]

Dear Qibo!
Fot standard RB for one qubit "if a dictionary is given (in actions.yml) the list will be built" - not true :) There seems to be a problem with yml file. I put in my actions:

depths:
     start: 1
     stop: 50
      step: 25

and it does not generate any report.
Thanks for help!

[alecandido]

@biankawolo just a general comment: to provide help for a bug, it is important to be able to reproduce (otherwise there will be a lot more guessing-in-the-wild involved).

So, please, whenever you file a bug report (starting from this one), provide:

1. a full minimal runcard to reproduce it
   • do not include parts that are not necessary
   • just the minimum that is needed (to the best of your understanding) to generate the undesired behavior
2. the versions of the sotware used, in particular
   • Qibocal version/branch
   • (Qibolab, if manually installed - otherwise it should be inferred from Qibocal)
   • which platform you're using (the name is in the runcard, so just point to the branch of qibolab_platforms_qrc)

[biankawolo]

@alecandido
Yest You are fully right! But this was just a formal trace of a problem because I spoke about it with Edoardo and he will have a look at it. Thanx!

[alecandido]

Yest You are fully right! But this was just a formal trace of a problem because I spoke about it with Edoardo and he will have a look at it. Thanx!

That's great, I'm confident @Edoardo-Pedicillo will be able to help you :)

However, the purpose of issues is also to track possible problems even as a reference to the other users (which may encounter the same ones), and to the other developers (which may help to solve it, or point it to other users).
So, making it reproducible may (in general) help you and rest of the Qibocal crowd :D

[biankawolo]

@alecandido good point! I have not thought about that:) thankx!

[biankawolo]

So Edoardo told me to put it all, because it might be a bigger issue.
I'm using qibolab: OPX1000 branch,
qibolab_platforms_qrc ---qw5q-platinum
qibocal---- main
My actions :

  # - id: standard rb
  #   priority: 0
  #   operation: standard_rb
  #   parameters:
  #     depths:
  #       start: 1
  #       stop: 102
  #       step: 25
  #     niter: 3
  #     nshots: 500
  #     uncertainties: std
  #     #n_bootstrap: 10
  #     # unrolling: True

The routine tries to execute itself and then at the end it crashes
And the error log looks like this:

Traceback (most recent call last):
  File "/nfs/users/bianka.woloncewicz/.envs/october/bin/qq", line 8, in <module>
    sys.exit(command())
  File "/nfs/users/bianka.woloncewicz/.envs/october/lib/python3.10/site-packages/click/core.py", line 1157, in __call__
    return self.main(*args, **kwargs)
  File "/nfs/users/bianka.woloncewicz/.envs/october/lib/python3.10/site-packages/click/core.py", line 1078, in main
    rv = self.invoke(ctx)
  File "/nfs/users/bianka.woloncewicz/.envs/october/lib/python3.10/site-packages/click/core.py", line 1688, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
  File "/nfs/users/bianka.woloncewicz/.envs/october/lib/python3.10/site-packages/click/core.py", line 1434, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/nfs/users/bianka.woloncewicz/.envs/october/lib/python3.10/site-packages/click/core.py", line 783, in invoke
    return __callback(*args, **kwargs)
  File "/nfs/users/bianka.woloncewicz/qibocal/src/qibocal/cli/_base.py", line 75, in auto
    autocalibrate(runcard, folder, force, update)
  File "/nfs/users/bianka.woloncewicz/qibocal/src/qibocal/cli/autocalibration.py", line 42, in autocalibrate
    history = runcard.run(
  File "/nfs/users/bianka.woloncewicz/qibocal/src/qibocal/auto/runcard.py", line 63, in run
    instance.run_protocol(
  File "/nfs/users/bianka.woloncewicz/qibocal/src/qibocal/auto/execute.py", line 131, in run_protocol
    completed = task.run(platform=self.platform, targets=self.targets, mode=mode)
  File "/nfs/users/bianka.woloncewicz/qibocal/src/qibocal/auto/task.py", line 160, in run
    completed.results, completed.results_time = operation.fit(completed.data)
  File "/nfs/users/bianka.woloncewicz/qibocal/src/qibocal/auto/operation.py", line 42, in wrapper
    out = func(*args, **kwds)
  File "/nfs/users/bianka.woloncewicz/qibocal/src/qibocal/protocols/randomized_benchmarking/standard_rb.py", line 104, in _fit
    return fit(data.qubits, data)
  File "/nfs/users/bianka.woloncewicz/qibocal/src/qibocal/protocols/randomized_benchmarking/utils.py", line 710, in fit
    error_bars = data_uncertainties(
  File "/nfs/users/bianka.woloncewicz/qibocal/src/qibocal/protocols/randomized_benchmarking/utils.py", line 211, in data_uncertainties
    (100 - method) / 2,
TypeError: unsupported operand type(s) for -: 'int' and 'str'

If you need anything else please let me know!

[alecandido]

I assume you're using the Qibolab opx1000 branch, qiboteam/qibolab#1045, and the qibolab_platforms_qrc's qw5q-qm branch, qiboteam/qibolab_platforms_qrc#179.

So, the problem apparently is that method is a str, and it doesn't make much sense to subtract 100 by a string of text...

Investigation

Not sure in the first place why it is called method, but it is assigned to data.uncertainties in the preceding call:

qibocal/src/qibocal/protocols/randomized_benchmarking/utils.py

Line 712 in 731c9b1

method=data.uncertainties,

The data are just the RBData, as expected:

qibocal/src/qibocal/protocols/randomized_benchmarking/standard_rb.py

Lines 94 to 104 in 731c9b1

	def _fit(data: RBData) -> StandardRBResult:
	"""Takes a data frame, extracts the depths and the signal and fits it with
	an exponential function y = Ap^x+B.
	Args:
	data (RBData): Data from the data acquisition stage.
	Returns:
	StandardRBResult: Aggregated and processed data.
	"""
	return fit(data.qubits, data)

generated by the rb_acquisition()

qibocal/src/qibocal/protocols/randomized_benchmarking/standard_rb.py

Line 91 in 731c9b1

return rb_acquisition(params, platform, targets)

(everything linear here).

The rb_acquisition() fills data after them being initialized in the setup() function

qibocal/src/qibocal/protocols/randomized_benchmarking/utils.py

Line 506 in 731c9b1

data, noise_model, backend = setup(params, platform, single_qubit=True)

qibocal/src/qibocal/protocols/randomized_benchmarking/utils.py

Lines 520 to 528 in 731c9b1

	data.register_qubit(
	RBType,
	(qubit_id, depth),
	dict(
	samples=samples[index[0] : index[1]][:, nqubit],
	),
	)
	data.circuits = indexes
	data.npulses_per_clifford = npulses_per_clifford

which is setting the .uncertainties attribute after the params one

qibocal/src/qibocal/protocols/randomized_benchmarking/utils.py

Line 377 in 731c9b1

uncertainties=params.uncertainties,

which in this case are the StandardRBParameters

qibocal/src/qibocal/protocols/randomized_benchmarking/standard_rb.py

Lines 26 to 36 in 731c9b1

	class StandardRBParameters(Parameters):
	"""Standard Randomized Benchmarking runcard inputs."""
	depths: Union[list, Depthsdict]
	"""A list of depths/sequence lengths.
	If a dictionary is given the list will be build.
	"""
	niter: int
	"""Sets how many iterations over the same depth value."""
	uncertainties: Optional[float] = None

So, the question is: why the uncertainties parameter in your runcard is a str instead of a float?

I.e. the error is all generated by this:

     uncertainties: std

[alecandido]

I guess you just used the runcard from

qibocal/runcards/actions_qq.yml

Lines 370 to 379 in 731c9b1

	# - id: standard rb inhomogeneous
	# operation: standard_rb
	# targets: [0, 1, 3]
	# parameters:
	# depths: [1, 3, 3, 5]
	# niter: 5
	# nshots: 50
	# uncertainties: std
	# noise_model: PauliErrorOnX
	# noise_params: [0.01, 0.01, 0.01]

that's the explanation.

Which is now conflicting with the documentation of that field

qibocal/src/qibocal/protocols/randomized_benchmarking/standard_rb.py

Lines 36 to 42 in 731c9b1

	uncertainties: Optional[float] = None
	"""Method of computing the error bars of the signal and uncertainties of
	the fit.
	If ``None``,
	it computes the standard deviation. Otherwise it computes the corresponding confidence interval. Defaults `None`.
	"""

That was changed in 8fde113 in #718.
So, the example runcard is simply outdated (which should be fixed @Edoardo-Pedicillo), but you should put a numerical value in that field. Or just omit the field (delete that line from your runcard), and rely on the default value.

[biankawolo]

Thanx I ll try that!

[alecandido]

Thanx I ll try that!

Whenever you tried, remember to report the outcome!

[biankawolo]

It works!

[alecandido]

Thanks for the confirmation!