SamplerV2

menos.qiskit.MENOSSampler implements Qiskit's BaseSamplerV2. Given measured circuits it returns samples, one BitArray per classical register, in the layout of Qiskit's StatevectorSampler.

MENOSSampler(*, default_shots: int = 1024, noise_model: NoiseModel | None = None,
             seed: int | None = None, readout_mitigation: bool = False)
MENOSSampler.run(pubs, *, shots=None) -> PrimitiveJob

from qiskit import QuantumCircuit

from menos.qiskit import MENOSSampler

qc = QuantumCircuit(2, 2)
qc.h(0)
qc.cx(0, 1)
qc.measure([0, 1], [0, 1])

result = MENOSSampler(seed=7).run([qc], shots=1000).result()[0]
counts = result.data.c.get_counts()
print(counts)
assert set(counts) == {"00", "11"} and sum(counts.values()) == 1000
assert result.metadata["seed"] == 7

Behavior

• Seeds. seed=None draws a fresh seed for each run() call. The seed used is recorded in every result's metadata["seed"], so any run can be reproduced. Pub j runs under the j-th seed derived from it, the rule run_batch uses.
• Shot order. MENOS produces counts, not a per-shot memory, so a BitArray lists its shots in ascending outcome order, not in sampling order. Counts, bitstrings as a multiset and expectation values are unaffected.
• Noise. noise_model is a qiskit_aer.noise.NoiseModel, applied as by MENOSBackend, including readout errors.
• Parameters. Parameterized pubs are refused; bind parameters first.
• Mitigation. readout_mitigation=True adds a mitigated distribution to each result's metadata. The samples themselves are never changed. See Readout Mitigation.