IBM Qiskit v2.X Developer – Associate Practice Exam

A Qiskit developer wants to compute an expectation value of a Pauli observable for a parameterized circuit.

Which Qiskit primitive fits?

1. The Sampler primitive, which returns quasi-probability distributions over basis states
2. The Estimator primitive, which takes a circuit, parameters, and observables and returns expectation values
3. Neither — parameters cannot be used with primitives
4. A custom simulator outside Qiskit

A parameterized circuit must run with many different parameter values efficiently.

Which Qiskit pattern fits?

1. Hard-code the parameter values and recompile each time
2. Rebuild the circuit from scratch for every parameter value
3. Build the circuit once with Parameter objects, then submit a parameter-value array to the primitive for efficient batched execution
4. Avoid parameterization entirely

A Sampler run needs to configure shot count and result-level options.

Which approach configures shots and options on a Sampler run?

1. Set global environment variables for each option
2. Pass Sampler-specific options (shots, resilience, and similar) when invoking the primitive, per the Qiskit Runtime options API
3. Modify Qiskit source code to change defaults
4. Ignore the options and accept whatever defaults run

A developer needs to represent a two-qubit pure state for simulation.

Which Qiskit class fits?

1. A plain Python list with no Qiskit structure
2. DensityMatrix, which represents mixed states
3. Operator, which represents unitary or general operators rather than states
4. Statevector, which represents a pure quantum state as a complex amplitude vector in the computational basis

A circuit must act as a specific two-qubit unitary whose matrix form is known.

Which Qiskit technique fits?

1. Use unitary decomposition (for example UnitaryGate or the two-qubit KAK decomposition) so Qiskit translates the matrix into gates supported by the target backend
2. Apply the matrix directly to hardware without decomposition
3. Round the matrix to the nearest identity
4. Skip the gate and hope results are similar

A Runtime workflow needs to run dozens of closely related circuits while keeping backend overhead low.

Which Runtime execution mode fits?

1. Use Batch mode for unrelated single-shot jobs
2. Submit each circuit as a completely independent job
3. Avoid Qiskit Runtime and run locally only
4. Use a Session so multiple primitive calls share the same backend context and benefit from lower per-call overhead

A Runtime job must have its results retrieved after the client disconnected mid-run.

Which capability fits?

1. Wait on the command line without disconnecting, indefinitely
2. Re-submit the whole workload because there is no persistence
3. Use the job ID to fetch results via the Runtime service after reconnecting; Runtime persists jobs server-side
4. Skip the job entirely

A circuit runs poorly on real hardware because the default layout does not align with the backend’s coupling map.

Which transpilation technique fits?

1. Use the transpiler’s layout and routing passes (with an appropriate optimization_level) to map logical qubits to well-connected physical qubits
2. Disable transpilation and submit the logical circuit directly
3. Pick random physical qubits without the coupling map
4. Ignore the coupling map and hope the hardware compensates

A deep circuit accumulates many redundant gates after composition.

Which transpiler feature helps?

1. Enable gate-cancellation and commutation passes in the transpiler’s staged pipeline to reduce gate count
2. Rewrite the circuit from scratch by hand
3. Accept the redundancy and run as-is
4. Turn off the transpiler entirely

A developer sees variance in expectation-value estimates that masks the signal of interest.

Which error-mitigation technique fits?

1. Pretend the noise is not there
2. Apply Zero-Noise Extrapolation (ZNE) by running the circuit at multiple noise scales and extrapolating to the zero-noise limit
3. Discard noisy results at random
4. Run the circuit only once and accept whatever you get