qiskit_noise_learning.analysis.FlipPostSelect

class qiskit_noise_learning.analysis.FlipPostSelect(creg_identifier: Callable[[list[str]], Iterator[tuple[str, ...]]] | None = None, mode: Literal['node', 'edge'] = 'edge', apply_measurement_flips: bool = False)

Bases: AnalysisStage

Apply a mask to raw data based on bit-flip failures across classical registers.

This post-selection stage identifies groups of one or two cregs, marks the failed bits of each group, and masks shots based on the structure of the failures:

• Two cregs (base, ps): bit j failed if its outcome is the same in both, i.e. it did not flip between the two measurements.
• One creg (base,): bit j failed if its outcome is True. This is the natural rule when a creg is expected to read out all-zeros, and coincides with the two-creg rule for a ps register of all ones.

Given the failed bits, the mode determines which shots are discarded:

• "node": Shots are discarded if at least one bit failed.
• "edge": Shots are discarded if there exists a pair of neighbouring qubits in the coupling map for which both bits failed.

Parameters

• creg_identifier – A callable that, given a list of present creg names, returns an iterator over tuples of creg names to post-select on. Each tuple holds either one or two names, selecting the corresponding rule above. Defaults to pairing each creg named "*_ps" with "*" when the latter is present, and treating it on its own otherwise.
• mode – Post-selection mode; either "node" or "edge".
• apply_measurement_flips – Whether or not to include measurement flips in the comparison. Defaults to False, i.e. comparing the physical measurement outcomes.

Raises

ValueError – If mode is not "node" or "edge".

__init__

Methods

Attributes

input_level

output_level