Update BaseEstimator import to support qiskit 2.0 by ElePT · Pull Request #2327 · Qiskit/qiskit-aer

ElePT · 2025-03-05T14:25:50Z

Summary

Qiskit/qiskit#13877 removes the non-versioned primitive V1 base class aliases. It looks like the AerSampler was already using the versioned base class, but AerEstimator wasn't. This PR updates the BaseEstimator import to BaseEstimatorV1.

Details and comments

The PR has been extended to handle more incompatibilities with 2.0.

…versioned alias

kt474

LGTM, this is also causing issues in the qiskit-ibm-runtime unit tests

wshanks · 2025-03-10T18:39:22Z

In my testing with Aer and Qiskit 2.0, this line is also problematic. Do you want to address it here as well?

qiskit-aer/qiskit_aer/primitives/sampler.py

Line 27 in 0f2ff71

from qiskit.primitives.utils import final_measurement_mapping, init_circuit

The problem addressed by the fix in this PR and the line I pointed out are both problems with trying to do import qiskit_aer.primitives with Qiskit 2.0.

wshanks · 2025-03-11T13:37:31Z

Ah, there is a helpful test run that now picks up the Qiskit rc and encounters further issues:

   File "/opt/hostedtoolcache/Python/3.10.16/x64/lib/python3.10/site-packages/qiskit_aer/primitives/sampler.py", line 24, in <module>
    from qiskit.circuit.bit import Bit
ModuleNotFoundError: No module named 'qiskit.circuit.bit'

…n 2.0

…te-primitive-imports

… but remove use in tests

…epted as a gate definition

…s to c_if and added a conditional skip depending on qiskit version

ElePT · 2025-03-12T13:07:53Z

A couple of unit tests with fractional gates were failing because of Qiskit/qiskit#14002. Given that this PR is blocking other efforts, I added a conditional skip for these tests until the issue is addresed.

ElePT · 2025-03-12T13:59:29Z

The estimator unit tests are currently failing because of another bug in qiskit: Qiskit/qiskit#14003. A fix is on the way. In the meantime, I have chosen to skip transpilation in that specific test, as it's the part of the code that triggers the qiskit bug.

…ng up the circuit to an arbitrary coupling map, which is something independent of the enable_truncation flag. The test now skips transpilation and exclusively tests the truncation (with enable_truncation=True, the test returns 2 instead of 4 qubits)

AlbertJP · 2025-03-12T15:36:14Z

        qc2.measure_all()

-        estimator = Estimator(approximation=True)
+        estimator = Estimator(approximation=True, skip_transpilation=True)


Try run_options = {'shots': None} to get exact values for the circuits that result in 0.

Alternative: change qc2 to give -1 instead of 0, then you won't have shot noise either. This can be done by changing np.pi / 2 to np.pi in the circuit, or [1] to [2] in the parameter.
(Estimator has no shot noise when the expectation value is exactly -1 or 1)

Thanks! I haven't used the aer estimator in a while and forgot how to set these.

This option is documented at

qiskit-aer/qiskit_aer/primitives/estimator.py

Lines 68 to 75 in 076fe00

When combined with the approximation option, we get the expectation values as follows:

* shots is None and approximation=False: Return an expectation value with sampling-noise w/

warning.

* shots is int and approximation=False: Return an expectation value with sampling-noise.

* shots is None and approximation=True: Return an exact expectation value.

* shots is int and approximation=True: Return expectation value with sampling-noise using a

normal distribution approximation.

. Honestly, I don't understand why it did not work. I suspect a bug in the Estimator class in combination with skip_transpilation=True.

It's likely. Something to look at in a follow-up.

gadial

Thank you for this substantial work. I left a few small comments which may be addressed but I believe we can merge without handling them as well.

gadial · 2025-03-13T06:58:03Z

-        self.assertEqual(result.results[0].header.metadata, metadata)
+        try:
+            out_metadata = result.results[0].header.metadata
+        except AttributeError:


Perhaps using hasattr is preferable to failing, although it's not that important.

gadial · 2025-03-13T07:00:40Z

        result = backend.run(circuit, shots=100).result()
        metadata = result.results[0].metadata
-        self.assertEqual(metadata["num_qubits"], 10)
+        self.assertEqual(metadata["num_qubits"], 4)


This seems to subtly change the test. Previously, there was a difference between the num_qubits value (which seems to have been determined by the now-extinct coupling map) and the number of active qubits - I believe this difference results from how truncation works in the CPP level. However, I did not manage to recreate this desired behavior yet when working with qiskit 2.0 so we might want to keep this fix as it is for now.

I actually think there is no difference in the way truncation works, the only actual difference I have seen is in the way the transpiler works between 1.4 and 2.0, but that is intentional because we added extra consistency checks between the inputs. The original test provided a backend with 5 qubits and a coupling map with 10, and this is something that transpile now processes differently. I tested transpiling only with the coupling map and the output metadata matches the original test. So maybe we could do that.

gadial · 2025-03-13T07:03:11Z

    # ---------------------------------------------------------------------
    # Test conditional
    # ---------------------------------------------------------------------
+    @unittest.skipUnless(


Can these tests be fixed to work with the new if mechanism in qiskit 2.0?

To some extent, yes. From what I have seen, the new if mechanism is only supported by some simulator methods (for example, statevector doesn't support it). I think that moving forward the tests would have to be refactored and only test conditionals with the supported simulator methods, but this is a bit tricky at the moment because we still want to support the "old" path, and there are already some tests for the new if, so I figured that the best intermediate solution would be the conditional skip. I can open an issue to keep track of this.

An issue seems like the correct way to proceed, thank you.

gadial

LGTM

Use versioned base class for BaseEstimatorV1 instead of deprecated un…

e3d9383

…versioned alias

This was referenced Mar 5, 2025

Remove v1 primitive implementations Qiskit/qiskit#13877

Merged

Unskip tests once downstream packages update to qiskit 2.0 Qiskit/qiskit-neko#54

Open

kt474 approved these changes Mar 7, 2025

View reviewed changes

kt474 mentioned this pull request Mar 7, 2025

Make sure all tests are passing with Qiskit main branch (2.0 release) Qiskit/qiskit-ibm-runtime#2169

Closed

ElePT added 2 commits March 11, 2025 14:00

Vendor removed V1 utils

cf942c3

Fix black

d51aee6

ElePT added 2 commits March 11, 2025 15:07

Replace bit with qubit

e877fc0

Remove assemble test, as assemble was deprecated in 1.4 and removed i…

e2d93d5

…n 2.0

ElePT force-pushed the update-primitive-imports branch from 45d1bbe to e2d93d5 Compare March 11, 2025 15:09

ElePT added 9 commits March 11, 2025 16:33

Merge branch 'main' of https://github.com/Qiskit/qiskit-aer into upda…

1695710

…te-primitive-imports

Treat header as dict

9fc42d1

Handle header conditionally

6da91ff

c_if is removed in 2.0, keep functionality for compatibility with 1.4…

2f4d880

… but remove use in tests

Fix access to header in tests

69c8825

Remove use of instruction_durations in transpile

e36e160

Attempt at fixing custom gate mapping. Not sure if gate = None is acc…

18344cb

…epted as a gate definition

IfElseOp is not supported by statevector simulator. Reverted uni test…

f6546e8

…s to c_if and added a conditional skip depending on qiskit version

Remove wrong import

2d0b1bb

kt474 mentioned this pull request Mar 11, 2025

Qiskit 2.0 Compatibility Qiskit/qiskit-ibm-runtime#2116

Merged

ElePT added 8 commits March 12, 2025 10:19

Fix failed target construction

983bafb

Only run tests with c_if with qiskit<2

c5865df

Address leftover custom basis gates in transpiler

17f9c34

Attempt at addressing .duration removal

fabca5e

Add unittest import missing in previous commit

f7617b8

Use new target arg in tests for relaxation noise pass

9cf7191

Fix lint

07b652f

Don't use target with backend props

edc89aa

Attempt at fixing relaxation pass

bb1869a

ElePT force-pushed the update-primitive-imports branch from a085662 to bb1869a Compare March 12, 2025 11:02

AlbertJP mentioned this pull request Mar 12, 2025

Fix multi-controlled gates in name mapping #2326

Open

use hashable type

a643439

ElePT added 2 commits March 12, 2025 14:30

Fix issue with pickle/dill (temporary skip of fractional gates)

0d57cfd

Add missing c_if skip

4d9cdd8

ElePT added 3 commits March 12, 2025 15:18

Skip transpilation to avoid qiskit bug in test

e38f0d0

Skipping transpilation affects results

41ac5a2

AlbertJP reviewed Mar 12, 2025

View reviewed changes

ElePT added 2 commits March 12, 2025 16:37

Apply suggestion to use shots None

8f78f29

Get rid of shot noise

5375346

ElePT requested a review from gadial March 12, 2025 17:20

gadial previously approved these changes Mar 13, 2025

View reviewed changes

Improve truncation test

7d554c3

ElePT dismissed gadial’s stale review via 7d554c3 March 13, 2025 08:49

Assert number of active qubits

4fe916c

gadial approved these changes Mar 13, 2025

View reviewed changes

ElePT added the automerge This PR will automatically merge once its CI has passed label Mar 13, 2025

ElePT mentioned this pull request Mar 13, 2025

Update conditional tests once c_if support is dropped in Aer #2331

Open

ElePT merged commit f77dbd3 into Qiskit:main Mar 13, 2025

	When combined with the approximation option, we get the expectation values as follows:

	* shots is None and approximation=False: Return an expectation value with sampling-noise w/
	warning.
	* shots is int and approximation=False: Return an expectation value with sampling-noise.
	* shots is None and approximation=True: Return an exact expectation value.
	* shots is int and approximation=True: Return expectation value with sampling-noise using a
	normal distribution approximation.

Conversation

ElePT commented Mar 5, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Details and comments

Uh oh!

kt474 left a comment

Choose a reason for hiding this comment

Uh oh!

wshanks commented Mar 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

wshanks commented Mar 11, 2025

Uh oh!

ElePT commented Mar 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ElePT commented Mar 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

AlbertJP Mar 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

gadial left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ElePT Mar 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

gadial left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

ElePT commented Mar 5, 2025 •

edited

Loading

wshanks commented Mar 10, 2025 •

edited

Loading

ElePT commented Mar 12, 2025 •

edited

Loading

ElePT commented Mar 12, 2025 •

edited

Loading

AlbertJP Mar 12, 2025 •

edited

Loading

ElePT Mar 13, 2025 •

edited

Loading