feat: ionq minimal integration into cloud provider platform by vprusso · Pull Request #725 · unitaryfoundation/metriq-gym

vprusso · 2026-03-13T23:21:31Z

Closes: #724

Example

Dispatch a small benchmark to IonQ simulator

  mgym job dispatch metriq_gym/schemas/examples/mirror_circuits.example.json -p ionq -d simulator

Then poll the result (replace with your job ID)

  mgym job poll <job-id>

Standalone script:

  from qbraid.runtime import IonQProvider
  from metriq_gym.ionq.device import patch_ionq_device
  from qiskit import QuantumCircuit

  provider = IonQProvider()
  device = provider.get_device("simulator")
  patch_ionq_device(device)

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

  job = device.run(qc, shots=100)
  result = job.result()
  print(result.data.get_counts())  # {'00': 50, '11': 50}

Using the IonQ Quantum Cloud platform, we can see that launching this job on the (free) simulator, that the result we get back is successful:

github-actions · 2026-03-13T23:21:59Z

PR Preview Action v1.8.1
🚀 View preview at https://unitaryfoundation.github.io/metriq-gym/pr-preview/pr-725/
Built to branch `gh-pages` at 2026-03-22 17:46 UTC. Preview will be ready when the GitHub Pages deployment is complete.

- Add IonQDevice dispatchers for version, connectivity_graph (all-to-all) in qplatform/device.py - Add IonQJob dispatchers for execution_time and job_status in qplatform/job.py - Add ionq/device.py workaround for qiskit_ionq bug that sets qubit count to device total instead of circuit size - Pass shots from params when loading IonQ jobs during polling (IonQ API omits shots in job details) - Add unit tests for all new IonQ qplatform functions and device patch

Copilot

Pull request overview

Adds minimal IonQ support paths needed for running metriq-gym benchmarks through the qBraid provider layer, including device metadata utilities, job timing/status extraction, and a runtime workaround for an IonQ/qiskit_ionq conversion issue.

Changes:

Add IonQ implementations for version(), connectivity_graph(), execution_time(), and job_status() in the qplatform utilities.
Patch IonQ device setup to bypass qiskit_ionq’s buggy conversion path by converting Qiskit circuits to QASM2 before submission.
Add unit tests covering IonQ device metadata, connectivity, and job timing/status behavior.

Reviewed changes

Copilot reviewed 7 out of 9 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
`metriq_gym/qplatform/job.py`	Adds IonQJob handling for execution time and status extraction.
`metriq_gym/qplatform/device.py`	Adds IonQDevice handling for version and all-to-all connectivity graph.
`metriq_gym/run.py`	Patches IonQ devices at setup time; injects `shots` into `load_job` kwargs for IonQ result reconstruction.
`metriq_gym/ionq/device.py`	Introduces IonQDevice.run patch to force QASM2 submission when qiskit_ionq is installed.
`metriq_gym/ionq/__init__.py`	Adds IonQ package module.
`tests/unit/qplatform/test_job.py`	Adds IonQ job tests for execution time and status.
`tests/unit/qplatform/test_device.py`	Adds IonQ device tests for version/connectivity/metadata.
`tests/unit/ionq/test_device.py`	Adds tests for QASM2 conversion helper and device patching behavior.
`tests/unit/ionq/__init__.py`	Adds IonQ test package module.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

metriq_gym/qplatform/job.py

+    exec_time_ms = meta.get("execution_time") or meta.get("predicted_execution_time")
+    if exec_time_ms is None:


metriq_gym/ionq/device.py

+which correctly preserves the circuit's qubit count.
+"""
+
+import importlib


tests/unit/ionq/test_device.py

+    def test_patched_device_converts_qiskit_to_qasm(self):
+        """After patching, Qiskit circuits should be converted to QASM2 strings."""
+        from qbraid.runtime import IonQDevice
+
+        class FakeDevice:
+            pass
+
+        FakeDevice.run = IonQDevice.run
+        device = FakeDevice()
+
+        patch_ionq_device(device)
+
+        # The patched run should convert QC -> QASM before calling original
+        qc = QuantumCircuit(2, 2)
+        qc.h(0)
+        qc.measure([0, 1], [0, 1])
+
+        # Directly test the conversion function instead of the full patch chain
+        result = _convert_qiskit_to_qasm2(qc)
+        assert isinstance(result, str)
+        assert "OPENQASM 2.0" in result
+


cosenal · 2026-03-16T16:30:30Z

metriq_gym/ionq/device.py

+When qiskit_ionq is installed, qBraid's IonQDevice.run() uses qiskit_ionq's
+transpilation path which sets the circuit qubit count to the device's total
+qubit count (e.g. 29) rather than the circuit's actual qubit count. This
+causes IonQ's simulator to reject circuits that would otherwise run fine.


Is this something we need to handle in metriq-gym, or raise it with qBraid?

Ideally, we should raise an issue about this on the qBraid issues board.

cosenal

I tested it with

mgym job dispatch config.json -p ionq -d simulator

and config.json being

{
    "benchmark_name": "WIT",
    "num_qubits": 7,
    "shots": 8192
}

I got 0.0 as a result for the expectation value, so something is wrong.

vprusso · 2026-03-17T11:42:56Z

I tested it with
mgym job dispatch config.json -p ionq -d simulator
and config.json being
{
    "benchmark_name": "WIT",
    "num_qubits": 7,
    "shots": 8192
}
I got 0.0 as a result for the expectation value, so something is wrong.

Two separate qBraid issues were causing it:

IonQJob._get_counts() uses normalize_data() which drops leading zeros from bitstrings when only
one computational state is observed (e.g. returns {"0": 8192} instead of {"0000000": 8192}). Benchmarks that do exact key lookups (mirror circuits, QML kernel) get 0 matches.
IonQ measures all qubits regardless of measurement gates (qBraid's QASM -> IonQ conversion explicitly discards them). So WIT's 7-qubit circuit with measure q[5] -> c[0] returns 7-bit strings, but binary_expectation_value looks for "1" (a 1-bit key).

Both are now fixed with a centralized patch to qBraid's IonQJob class, but ideally, we would want another MR that actually patches qBraid for this. Thoughts, @cosenal ?

Resolve uv.lock and pyproject.toml conflicts. Remove ionq extra from qbraid dependency to drop qiskit-ionq per qBraid/qBraid#1141 workaround.

vprusso force-pushed the 724-ionq-integration branch from 3f30ab0 to dd6e062 Compare March 13, 2026 23:31

vprusso added 2 commits March 13, 2026 19:35

fix: ruff

2caf6c5

chore: minor nits

78e09a3

vprusso requested a review from cosenal March 14, 2026 11:55

vprusso marked this pull request as ready for review March 14, 2026 11:55

Copilot AI review requested due to automatic review settings March 14, 2026 11:55

Copilot started reviewing on behalf of vprusso March 14, 2026 11:56 View session

Copilot AI reviewed Mar 14, 2026

View reviewed changes

vprusso added 2 commits March 14, 2026 09:39

chore: simplify overly cautious guard checks

6b8cc07

chore: co-pilot comments

33ca4ab

cosenal reviewed Mar 16, 2026

View reviewed changes

cosenal requested changes Mar 16, 2026

View reviewed changes

vprusso added 3 commits March 16, 2026 23:09

chore: patch ionq job

1d8c980

chore: patch ionq job

8e95eba

chore: patch ionq job

cf01691

vprusso requested a review from cosenal March 17, 2026 11:38

vprusso added 3 commits March 17, 2026 07:56

chore: uv.lock conflict fix

c01465b

fix: ionq/qbraid bitstring big-endian issue

5aed5c3

chore: merge main, resolve conflicts, drop qiskit-ionq

c51294b

Resolve uv.lock and pyproject.toml conflicts. Remove ionq extra from qbraid dependency to drop qiskit-ionq per qBraid/qBraid#1141 workaround.

This was referenced Mar 25, 2026

Upstream custom provider integrations to qBraid #730

Open

Fix IonQ qiskit_ionq qubit count bug upstream in qBraid #734

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: ionq minimal integration into cloud provider platform#725

feat: ionq minimal integration into cloud provider platform#725
vprusso wants to merge 11 commits intomainfrom
724-ionq-integration

vprusso commented Mar 13, 2026 •

edited

Loading

Uh oh!

github-actions bot commented Mar 13, 2026 •

edited

Loading

Built to branch `gh-pages` at 2026-03-22 17:46 UTC.
Preview will be ready when the GitHub Pages deployment is complete.

Uh oh!

Copilot AI left a comment

Uh oh!

cosenal Mar 16, 2026

Uh oh!

vprusso Mar 17, 2026

Uh oh!

cosenal left a comment

Uh oh!

vprusso commented Mar 17, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

		exec_time_ms = meta.get("execution_time") or meta.get("predicted_execution_time")
		if exec_time_ms is None:

Conversation

vprusso commented Mar 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Example

Uh oh!

github-actions bot commented Mar 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Built to branch gh-pages at 2026-03-22 17:46 UTC. Preview will be ready when the GitHub Pages deployment is complete.

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

cosenal Mar 16, 2026

Choose a reason for hiding this comment

Uh oh!

vprusso Mar 17, 2026

Choose a reason for hiding this comment

Uh oh!

cosenal left a comment

Choose a reason for hiding this comment

Uh oh!

vprusso commented Mar 17, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

vprusso commented Mar 13, 2026 •

edited

Loading

github-actions bot commented Mar 13, 2026 •

edited

Loading

Built to branch `gh-pages` at 2026-03-22 17:46 UTC.
Preview will be ready when the GitHub Pages deployment is complete.