Add BackendSamplerV2

[t-imamichi]

Summary

This PR adds BackendSamplerV2 as a BaseSamplerV2 implementation using a backend.
This requires BackendV2 object that supports memory option.

I made tests based on those for StatevectorSampler.

Notes

• I intentionally specified only BackendV2 as backend option though it still supports BackendV1 because I think we need to support only the latest version of backends. If requested, I can add BackendV1.
  • I added BackendV1 as well as BackendV2 support.
• I didn't add transpile feature because Runtime Primitives V2 do not support transpile during execution. Users need to transform input circuits into ISA format beforehand.

Details and comments

[qiskit-bot]

One or more of the the following people are requested to review this:

• @Qiskit/terra-core
• @ajavadia
• @ikkoham
• @levbishop
• @t-imamichi

[coveralls]

Pull Request Test Coverage Report for Build 8297352993

Details

• 98 of 99 (98.99%) changed or added relevant lines in 2 files are covered.
• 4 unchanged lines in 2 files lost coverage.
• Overall coverage increased (+0.04%) to 89.316%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
qiskit/primitives/backend_sampler_v2.py	97	98	98.98%

Files with Coverage Reduction	New Missed Lines	%
crates/qasm2/src/expr.rs	1	93.81%
crates/qasm2/src/lex.rs	3	92.19%

Totals
Change from base Build 8297051089:	0.04%
Covered Lines:	59820
Relevant Lines:	66976

---

💛 - Coveralls

[ikkoham]

Basically LGTM, but we need more reviews.

[ikkoham]

Do we need to check for measure instruction? Should we rather check measure instruction?

[t-imamichi]

I'm not sure whether we need more validations. This is basically same as the check of StatevectorSampler.

qiskit/qiskit/primitives/statevector_sampler.py

Lines 161 to 166 in ff2bbfe

	if any(len(pub.circuit.cregs) == 0 for pub in coerced_pubs):
	warnings.warn(
	"One of your circuits has no output classical registers and so the result "
	"will be empty. Did you mean to add measurement instructions?",
	UserWarning,
	)

[1ucian0]

I tested this in the following way:

with Aer:

from qiskit import QuantumCircuit, transpile
from qiskit_aer import Aer
from qiskit.primitives import BackendSamplerV2

qasm_simulator = Aer.get_backend('qasm_simulator')
qasm_simulator_sampler = BackendSamplerV2(backend=qasm_simulator)

circuit = QuantumCircuit(2)
circuit.h(0)
circuit.cx(0, 1)
circuit.measure_all()

jobs = qasm_simulator_sampler.run([transpile(circuit, qasm_simulator)])
jobs.result()[0].data.meas.get_counts()

with fake backend (V1 and V2):

from qiskit import QuantumCircuit, transpile
from qiskit_ibm_runtime.fake_provider import FakeAlmadenV2 
from qiskit.primitives import BackendSamplerV2

fake_almaden = FakeAlmadenV2()
fake_almaden_sampler = BackendSamplerV2(backend=fake_almaden)

circuit = QuantumCircuit(2)
circuit.h(0)
circuit.cx(0, 1)
circuit.measure_all()

jobs = fake_almaden_sampler.run([transpile(circuit, fake_almaden)])
print(jobs.result()[0].data.meas.get_counts())

It works as expected.

I was surprised with the following errors:

• access databin items with subscripts seems like it is not working. That is print(jobs.result()[0].data['meas'].get_counts()) is failing with 'DataBin' object is not subscriptable. Probably unrelated with this PR.
• When accidentally calling the primitive with a circuit instead of an array with a circuit jobs = fake_almaden_sampler.run(transpile(circuit, fake_almaden)), the error AttributeError: 'U2Gate' object has no attribute 'parameters' which seems suspicious as it is an access error. It might be because something deeper is wrong.
• I was unable to seed the simulated running. The reference implementation supports sampler.run(qc, shots=shots, seed=42) but fake_almaden_sampler.run([transpile(circuit, fake_almaden)], seed=42) is not supported, it seems.

[1ucian0]

I find these tests hard to read. At the same time, they are based on test/python/primitives/test_backend_sampler.py, which I also find hard to read. So probably for a follow up.

[1ucian0]

Maybe an example with Fake7QPulseV1 similar to #11928 (comment) ?

[t-imamichi]

Thank you for your review, Luciano. I add some comments to your surprises.

1. access databin items with subscripts are not implemented since the first PR Add V2 primitive container classes #11524. Do we need to add it in a separate PR, @ihincks?
2. SamplerV2 accepts only an iterable of pubs or publikes.

   qiskit/qiskit/primitives/base/base_sampler.py

   Line 167 in f4b50fb

   self, pubs: Iterable[SamplerPubLike], *, shots: int | None = None

   So, jobs = fake_almaden_sampler.run(transpile(circuit, fake_almaden)) is interpreted as individual instruction of the transpiled circuit as an input publike. It should be jobs = fake_almaden_sampler.run([transpile(circuit, fake_almaden)]) (put a circuit in a list).
3. It does not support any run_options currently. I suppose a backend can set the seed in the constructor, i.e., BackendSamplerV2(backend=BackendSomething(seed=seed)). If it is preferable to accept run options in run method, I will add it.

[ihincks]

Thank you @t-imamichi I only have a few minor comments. Other than those, this implimentation LGTM.

[t-imamichi]

I added BackendV1 support and tests with Fake7QPulseV1.

[t-imamichi]

I added options to set seed_simulator.

[1ucian0]

Do you mind adding an example like #11928 (comment) ?

[1ucian0]

with seeding, to you still need this _assert_allclose?

[t-imamichi]

Yes. It is necessary because unit tests use both an exact simulator and noise simulators.

[1ucian0]

Some non-blocking comments. It they can be addressed in this PR, great. Otherwise, in a follow up

[t-imamichi]

I updated the code based on #11931 as follows.

• updated the docstrings
• moved default_shots to options
• added a sample code

[1ucian0]

Let's merge it! Thanks!

[t-imamichi]

Fixed a conflict