(2/2) Combine Pauli measurements and postselection

[ddddddanni]

This commit implements the postselection symmetry functionality in CircuitToPauliStringsParameters. Circuits provided with postselection_symmetries will now successfully filter measurement results based on the expected eigenvalues rather than defaulting to the confusion matrix method.

[ddddddanni]

I tested using https://colab.sandbox.google.com/drive/1xY5N6NBm2kSL_kChUr566hLNlwiNApY6?resourcekey=0-D_sl5TxF7eo9A_azwwb8nQ#scrollTo=w8fWdb-2Ki6c, results seem to be fine

[eliottrosenberg]

LGTM! Thanks, @ddddddanni!

Some testing in this colab.

[eliottrosenberg]

Actually, maybe the docstrings could make it a bit clearer that it doesn't perform readout benchmarking if postselection symmetries are included and that the parameters readout_repetitions and num_random_bitstrings are ignored in this case.

[ddddddanni]

Actually, maybe the docstrings could make it a bit clearer that it doesn't perform readout benchmarking if postselection symmetries are included and that the parameters readout_repetitions and num_random_bitstrings are ignored in this case.

Done! I added this: https://screenshot.googleplex.com/6B7SV7HRxZWyH9C

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.63%. Comparing base (b3366ab) to head (e16729b).

Additional details and impacted files

@@           Coverage Diff            @@
##             main    #7970    +/-   ##
========================================
  Coverage   99.63%   99.63%            
========================================
  Files        1110     1110            
  Lines       99738    99855   +117     
========================================
+ Hits        99376    99493   +117     
  Misses        362      362            

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[pavoljuhas]

Please check if we can avoid some code repetition (see inline).

We should also check if measure_pauli_strings can have a simpler interface.

@eliottrosenberg - is it required for measure_pauli_strings to handle an argument with a mixture of post-selection-symmetry and confusion-matrix cases?

[pavoljuhas]

This is a bit over-complicated interface. Both readout_repetitions and num_random_bitstrings arguments are mandatory, but unused for the post-selection symmetry case.

Would it work to include readout_repetitions and num_random_bitstrings in CircuitToPauliStringsParameters where they would be mutually exclusive with postselection_symmetries and the split this interface to 2 overloads:

1 legacy interface (if feasible deprecate):

@overload
def measure_pauli_strings(
    circuits_to_pauli: (
        Mapping[circuits.FrozenCircuit, Sequence[ops.PauliString]
        | Sequence[Sequence[ops.PauliString]]]
    ),
    sampler: work.Sampler,
    pauli_repetitions: int,
    readout_repetitions: int,
    num_random_bitstrings: int,
    ...
)

2 preferred interface (readout_repetitions, num_random_bitstrings in CircuitToPauliStringsParameters)

@overload
def measure_pauli_strings(
    circuits_to_pauli: list[CircuitToPauliStringsParameters],
    sampler: work.Sampler,
    pauli_repetitions: int,
    ...
)

[ddddddanni]

I'm not sure if it's better to put readout_repetitions and num_random_bitstrings under CircuitToPauliStringsParameters, because these two parameters are more like global parameters that could take effects to all circuits, while CircuitToPauliStringsParameters is only for one circuit. wdyt?