The SDK currently has a README and CONTRIBUTING.md but no worked examples or API reference. This limits discoverability and onboarding, especially for unitaryHACK contributors.
What's needed
Example notebooks (examples/)
Jupyter notebooks covering common use cases:
- Bell state + local execution — the simplest possible end-to-end example
- VQE with gradient-free optimiser — COBYLA optimisation loop using
@task/@workflow to parallelise Pauli measurements
- Multi-backend benchmark — run the same circuit on Local, Braket SV1, and IBM, compare results
- Circuit conversion roundtrip — show
Circuit → Qiskit → Circuit, Circuit → Braket → Circuit, etc.
- Parameter sweep — use
@workflow to fan out over a range of rotation angles in parallel
API reference
Auto-generated reference docs (e.g. via Sphinx or pdoc) covering:
marqov.circuits.Circuit — all gates, converters, importers
marqov.executors — each executor, config classes, ExecutionResult
marqov.workflows — @task, @workflow, WorkflowDispatch
marqov.simulation.noise — NoiseModel, noise channels
Notes
- Notebooks should run without cloud credentials (use
LocalExecutor or mock cloud calls where needed)
- API reference should be auto-generated from existing docstrings — the inline docs are already thorough
The SDK currently has a README and CONTRIBUTING.md but no worked examples or API reference. This limits discoverability and onboarding, especially for unitaryHACK contributors.
What's needed
Example notebooks (
examples/)Jupyter notebooks covering common use cases:
@task/@workflowto parallelise Pauli measurementsCircuit → Qiskit → Circuit,Circuit → Braket → Circuit, etc.@workflowto fan out over a range of rotation angles in parallelAPI reference
Auto-generated reference docs (e.g. via Sphinx or pdoc) covering:
marqov.circuits.Circuit— all gates, converters, importersmarqov.executors— each executor, config classes,ExecutionResultmarqov.workflows—@task,@workflow,WorkflowDispatchmarqov.simulation.noise—NoiseModel, noise channelsNotes
LocalExecutoror mock cloud calls where needed)