Execution & results

Live preview versus an explicit run

The live preview updates as you edit. It helps inspect the current circuit; it is separate from submitting a saved experiment.

An explicit Run records the submitted circuit and settings through the saved job flow. Sign in to access owner-scoped records in My Jobs. If you edit afterward, the saved result still belongs to the earlier submitted circuit; check the stale-circuit notice and run details.

Choose a backend in Execute

1. Open Execute and choose the simulator or hardware path.
2. Review the available framework, backend and target options. Availability and compatibility depend on the current backend catalog.
3. Check width and operation support, especially for mid-circuit measurement, reset and feed-forward.
4. Set shots and noise options where supported. Hardware providers may require your own credentials and an available target.
5. Run the circuit and follow its status. Open the saved job for the recorded circuit, execution settings and result.

A disabled or unavailable backend is not a promise of a working execution. Follow the displayed reason; a simulator with a similar name may have different supported operations or limits.

Interpret probabilities and counts

Exact results give computed probabilities for the submitted configuration. Sampled results give counts or frequencies from a finite number of shots. An outcome with a small nonzero probability may be absent from a small sample.

Choose the register that matches the algorithm’s output. Classical bits describes measurement records; a qubit-state probability chart may include work wires that are not part of the answer.

For the ideal Bell example, 00 and 11 each have probability 1/2. Counts of 49 and 51 out of 100 shots are consistent with that prediction. Unexpected 01 or 10 outcomes call for checking gates, measurement mapping and any selected noise.

Reopen and compare saved runs

My Jobs lists your saved Composer executions. Search or filter to find an experiment, then open its detail page to inspect the submitted snapshot, backend, status and recorded outcome.

Select two completed saved runs for comparison. Compare the circuit and settings first. Distribution distance is meaningful only when measured registers are compatible; the comparison page checks that compatibility.

An algorithm label records where the circuit originated. It does not certify that an edited circuit still implements the original algorithm or that a result demonstrates quantum advantage.

When something looks wrong

For an empty preview, check validation messages and whether the local simulation engine loaded. For an execution failure, inspect the job’s error and selected backend’s requirements.

For an unexpected histogram, check gate order, controls, angles, classical mapping, measurement basis, shots and noise. For a mismatch with a saved result, verify that you are comparing the same submitted circuit revision.

A run can fail because a provider target or server is unavailable even when the circuit is valid. Preserve the error message and job identifier when reporting the problem.