IBM C1000-179 — Fundamentals of Quantum Computing Using Qiskit v2.X Developer Passing target: 24/34. All code outputs and the circuit drawing were verified by execution with qiskit 2.5.1 / qiskit-ibm-runtime 0.48.0 / qiskit-aer.

Answer Key

Q	Ans	Q	Ans	Q	Ans	Q	Ans
1	C	10	C	19	C	28	D
2	A	11	A	20	A, C	29	A
3	B	12	D	21	B	30	C
4	C	13	B	22	A	31	B
5	B	14	A	23	B	32	D
6	B	15	C	24	C	33	B
7	A	16	B	25	B	34	C
8	D	17	D	26	B
9	A	18	A	27	C

Explanations

Q1 — C (['a', 'b']). qc.parameters is a view sorted alphabetically by name, not by insertion order. Even though b was used first (rx(b, 0)), the view lists a before b. Reusing a qubit for two parameterized gates is perfectly legal. Verified by execution.

Q2 — A (2 1). compose() returns a new circuit by default (inplace=False), so qc3 has h + cx = 2 instructions while qc1 is untouched at size 1. Verified by execution.

Q3 — B. H on q0 creates the superposition, then the CX chain (0→1, then 1→2) spreads the correlation to all three qubits: (|000⟩ + |111⟩)/√2. A gives the product state |+++⟩; C's second CX undoes the first, leaving a Bell pair on q0/q1 and q2 in |0⟩; D produces the deterministic |111⟩ (no superposition).

Q4 — C ({'11': 1000}). x(0) guarantees the mid-circuit measurement yields 1, so the if_test body fires every shot and x(1) runs. Both clbits read 1 → bitstring '11' for all 1000 shots. Verified by execution with AerSimulator.

Q5 — B. Routing inserts SWAPs to satisfy the coupling map. Layout only chooses the initial virtual→physical mapping; translation rewrites gates into the basis set; scheduling deals with timing.

Q6 — B (['cx', 'h']). inverse() reverses the instruction order and inverts each gate. H and CX are both self-inverse, so the names are unchanged — only the order flips. There are no hdg/cxdg gates. Verified by execution.

Q7 — A (-1.0). Label '01' means q1=0, q0=1 (little-endian). Pauli('IZ') puts Z on qubit 0 (rightmost character). Z on |1⟩ gives eigenvalue −1. Verified by execution.

Q8 — D (['ZIXI']). from_sparse_list places Z on qubit 3 and X on qubit 1. Reading the string left→right is q3 q2 q1 q0: Z I X I. Verified by execution.

Q9 — A (True). This is the conjugation identity HXH = Z. Operator.equiv compares up to global phase, and a Pauli converts cleanly to an Operator. Verified by execution.

Q10 — C. In a Pauli string the rightmost character is qubit 0, so "Z on q1 ⊗ Z on q0" is "ZZ" and "X on q1" is "XI". Coefficients pair positionally in from_list, so [("ZZ", 2.0), ("XI", 1.0)] is correct. A puts X on qubit 0; D swaps the coefficients.

Q11 — A (qubit 0). Little-endian: the rightmost character of 'XYZ' is qubit 0, so Z→q0, Y→q1, X→q2.

Q12 — D. After H+CX the state is (|00⟩+|11⟩)/√2; x(1) flips qubit 1 (the left bit of each label): (|10⟩+|01⟩)/√2 → {'01': 0.5, '10': 0.5}. Verified by execution.

Q13 — B. Sequentially dependent, iterative jobs are the textbook Session use case: the dedicated window avoids re-queuing between iterations. Batch is for independent jobs; job mode queues each request separately.

Q14 — A. The v2.x pattern is Session(backend=...) as a context manager and SamplerV2(mode=session). There is no session=True kwarg and no session.sampler() factory in the current API.

Q15 — C. V2 primitives on IBM hardware accept only ISA circuits (already expressed in the backend's basis gates and respecting its connectivity). The job is rejected — the service does not auto-transpile.

Q16 — B ({'10': 200}). x(1) sets qubit 1; in the little-endian bitstring q1 is the left bit → '10', deterministically for all 200 shots requested via the PUB's third element. Verified by execution.

Q17 — D. service.least_busy(operational=True, simulator=False) is the QiskitRuntimeService method. C is the old provider-era helper function.

Q18 — A. measure_all() creates the register named 'meas' with 3 bits; 100 shots were requested. x(0) sets the rightmost bit → '001'. So num_shots=100, num_bits=3, counts {'001': 100}. Verified by execution.

Q19 — C. Sampler PUB parameter values for k parameter sets over p parameters have shape (k, p): a list of three inner lists of two values each. A single flat list is one parameter set (and here it has the wrong length).

Q20 — A and C. Sampler options include dynamical_decoupling (suppression) and default_shots. resilience_level and ZNE (resilience.zne_mitigation) are Estimator-only — the Sampler returns raw samples and has no mitigation levels.

Q21 — B. With multiple classical registers, each appears as its own BitArray field named after the register. alpha recorded q0 (flipped to 1), beta recorded q1 (still 0): {'1': 50} {'0': 50}. Verified by execution.

Q22 — A ([1. 1. 0.]). For the Bell state: ⟨ZZ⟩ = 1 (perfect correlation), ⟨XX⟩ = 1, and ⟨IZ⟩ = 0 (each qubit alone is maximally mixed). One PUB happily carries an array of observables. Verified by execution.

Q23 — B ([ 1. -1.]). The PUB broadcasts one observable across two parameter sets. ry(0)|0⟩ = |0⟩ → ⟨Z⟩ = 1; ry(π)|0⟩ = |1⟩ → ⟨Z⟩ = −1. evs has shape (2,). Verified by execution.

Q24 — C. After transpilation the logical qubits live on arbitrary physical qubits, so the observable must be remapped with obs.apply_layout(isa_qc.layout). apply_layout is a method of the observable, not the circuit (D).

Q25 — B. Level 1 (default) applies measurement-error mitigation (TREX) and twirling; level 2 adds ZNE. PEC is not enabled by a resilience level preset — it is opt-in via options.resilience.pec_mitigation.

Q26 — B. Gates render left→right in application order (H, then Z, then CX) with q0 on the top wire; the CX control dot sits on q0 and the ⊕ target on q1. Drawing verified by execution.

Q27 — C. Each qubit of a Bell pair is maximally mixed, so its reduced Bloch vector has zero length — the arrows vanish at the sphere centers. This is the visual signature of entanglement in plot_bloch_multivector.

Q28 — D. On the qsphere, blob size encodes amplitude magnitude and blob color encodes the amplitude's phase.

Q29 — A. number_to_keep=k shows the k most frequent outcomes and lumps the remainder into a single "rest" bar — nothing is silently dropped.

Q30 — C. Jobs are retrieved through the service: QiskitRuntimeService().job(job_id), then .result(). Primitives have no load classmethod; sessions do not retrieve jobs.

Q31 — B (0.2). ⟨Z⟩ = (n₀ − n₁)/shots = (600 − 400)/1000 = 0.2. Option C flips the sign convention (Z has eigenvalue +1 for |0⟩).

Q32 — D (16384). Standard error scales as 1/√shots. Halving the error requires 4× the shots: 4096 × 4 = 16384.

Q33 — B. qasm3.loads parses a QASM 3 string; qasm3.load reads from a file. from_qasm_str and qasm2.loads are QASM 2-only and will choke on OPENQASM 3.0;.

Q34 — C. QASM 3 declares bit[2] c; and qubit[2] q; and writes measurements as assignments: c[0] = measure q[0];. The qreg/creg/measure q -> c forms are QASM 2 syntax. Output string verified by execution.

Distractor Analysis — 10 Trickiest

Q1 (param order). Distractor A ['b', 'a'] is the insertion-order assumption — the single most common assign_parameters list-binding bug. Qiskit sorts the parameter view alphabetically, which is exactly why positional binding can silently scramble values.
Q4 (if_test). A {'01': 1000} catches people who read the bitstring big-endian (clbit 0 on the left). D tempts those who think mid-circuit measurement of |1⟩ is probabilistic. The state is deterministic end to end.
Q7 (Pauli IZ). B 1.0 is the big-endian trap: reading 'IZ' as "Z on qubit 1" and the label '01' as "q0=0". Two endian errors cancel for palindromes but not here — the exam loves this.
Q8 (from_sparse_list). B ['IXIZ'] is the mirrored (big-endian) string. A ['ZXII'] comes from treating the index list [3, 1] as left-to-right string positions instead of qubit indices.
Q12 (bell + x). A keeps the original Bell distribution (forgetting the X entirely); B assumes the X collapses things deterministically. The X flips the left bit of each branch label — position is the trap.
Q16 (sampler counts). A {'01': 200} is the endian flip. C {'1': 200} tempts those who think only measured-and-flipped qubits appear — but the register has 2 bits, so bitstrings have 2 characters.
Q19 (PUB shape). A is a flat list, which the Sampler interprets as a single parameter set — with 3 values against 2 parameters it fails; even shaped right it would not mean "3 sets". The (k, p) 2-D shape is the rule to burn in.
Q20 (sampler options). B resilience_level is the flagship trap: it exists on Estimator options only. D moves a real Estimator ZNE path onto the Sampler. Both look plausible because the two primitives share the DD/twirling suppression options.
Q23 (broadcasting). D ("two PUBs required") reflects V1 thinking where each run took one binding. C ("only the first set is used") preys on uncertainty about silent truncation — V2 PUBs never truncate; they broadcast or raise.
Q26 (drawing). C is the same circuit drawn with the wires swapped — plausible if you believe q0 renders at the bottom (it renders at the top unless reverse_bits=True). A reverses control and target on the CX; D swaps the H/Z order.