Self-Assessment Quiz: Quantum Computing Systems

Twenty questions on cryogenics, control electronics, calibration, the classical stack, and what it takes to actually operate a quantum computer. Aim for 16+.

Question 1

A dilution refrigerator reaches millikelvin temperatures using:

A) A ³He/⁴He mixture exploiting the enthalpy of dilution B) Liquid nitrogen C) Peltier coolers D) Adiabatic expansion of air

Question 2

Cooling power at the mixing chamber (10 mK) is on the order of:

A) Watts B) Milliwatts C) Microwatts D) Kilowatts

Question 3

Cooling power at the 4 K stage is roughly:

A) 1–2 W B) 20 μW C) 1 kW D) 1 mW

Question 4

Control electronics generate:

A) Shaped microwave pulses with precise amplitude, phase, and timing B) DC voltages only C) Optical pulses only D) Classical logic signals only

Question 5

Timing precision required for gate pulses is on the order of:

A) Milliseconds B) Microseconds C) Nanoseconds or better D) Seconds

Question 6

Readout signals are amplified first by:

A) A near-quantum-limited amplifier (e.g. a JPA or TWPA) at the cold stage B) A room-temperature amplifier only C) An optical amplifier D) No amplification

Question 7

Calibration must be repeated periodically because:

A) Qubit frequencies and gate parameters drift over hours to days B) Software resets C) Users request it D) The refrigerator warms

Question 8

A typical full-device calibration takes:

A) Seconds B) Minutes to hours C) Weeks D) No time

Question 9

The classical control stack must decode syndromes in fault-tolerant operation within:

A) Microseconds — faster than syndrome extraction rounds B) Hours C) Seconds D) No time constraint

Question 10

If the decoder falls behind syndrome extraction:

A) A backlog accumulates and grows exponentially, breaking fault tolerance B) Nothing happens C) The computation speeds up D) Errors self-correct

Question 11

Quantum computers are usually accessed by users via:

A) Cloud queues with job scheduling B) Direct physical access C) Local installation D) Email

Question 12

Job queue times matter because:

A) Variational algorithms need many sequential quantum-classical round trips B) Queues are free C) Jobs are instant D) Queues improve fidelity

Question 13

A "session" or reserved-execution mode:

A) Holds device allocation across an iterative algorithm, avoiding requeuing each iteration B) Increases fidelity C) Reduces qubit count D) Is unnecessary

Question 14

Physical footprint of a current quantum computer is roughly:

A) A large room, dominated by the refrigerator, racks, and compressors B) A desktop C) A shipping container fleet D) A microchip

Question 15

Total electrical power draw of a typical superconducting system is dominated by:

A) Pulse-tube compressors and control electronics — tens of kilowatts B) The qubits themselves C) Lighting D) The decoder

Question 16

True or false: Quantum computers will replace classical data centres.

Question 17

True or false: The classical control system is a minor component of a quantum computer.

Question 18

True or false: Quantum computers must be recalibrated more often as they grow.

Question 19

Short answer. Explain why real-time decoding is an unsolved engineering problem at scale.

Question 20

Short answer. Describe the full stack between a user's Python script and a physical gate.

Answer Key

Q Ans Note
1 A Circulating ³He across the phase boundary provides continuous cooling below 100 mK.
2 C ~20 μW — the constraint that caps wiring (Ch. 26).
3 A Four to five orders of magnitude more than at base, which is why cryo-CMOS targets this stage.
4 A Arbitrary waveform generators with per-qubit calibrated pulse shapes.
5 C Sub-nanosecond timing and phase coherence across channels.
6 A Josephson parametric or travelling-wave amplifiers add minimal noise, which is essential for single-shot readout.
7 A TLS drift, flux drift, and thermal cycling all move parameters.
8 B Full calibration of a large device can take hours; incremental recalibration is more frequent.
9 A Decoding must keep pace with ~1 μs rounds.
10 A The backlog problem: unprocessed syndromes accumulate faster than they are cleared, and corrections arrive too late to be valid.
11 A Cloud access with queues is the dominant model.
12 A Each iteration is a separate job; queue time can dominate wall-clock by orders of magnitude.
13 A Sessions keep the allocation, which is why variational work is impractical without them.
14 A The dilution refrigerator, compressors, and electronics racks fill a room.
15 A Pulse-tube compressors alone draw ~10 kW; the quantum processor itself consumes essentially nothing.
16 False They are accelerators for specific problems, and every one is operated alongside substantial classical infrastructure.
17 False Control electronics, calibration software, compilers, and decoders represent the majority of the engineering effort and cost in a modern system.
18 True More qubits means more parameters, more frequency collisions, and more pairwise interactions to characterize — calibration time grows superlinearly, and automated calibration is itself a research area.
19 — Fault tolerance requires syndromes decoded within the time of a few extraction rounds — microseconds. The decoding problem is a 3D matching over a space-time volume growing as $O(d^3)$ per logical qubit, and a machine with thousands of logical qubits generates syndrome data at terabits per second. Meeting that requires dedicated hardware (FPGAs or ASICs) physically near the cryostat, since even the speed of light over a few metres consumes part of the budget. Falling behind is not graceful degradation: the backlog grows without bound and corrections become invalid.
20 — (1) User writes a circuit in Python. (2) The transpiler maps it to native gates and the coupling map, inserting SWAPs. (3) The scheduler converts gates into timed pulse sequences using per-qubit calibration data. (4) A job is queued and allocated to a device. (5) Arbitrary waveform generators synthesize microwave pulses at room temperature. (6) Signals travel down attenuated coaxial lines through 4 K and 100 mK stages to ~10 mK. (7) Pulses drive the qubits. (8) Readout tones probe resonators; returning signals are amplified by a quantum-limited amplifier at the cold stage, then further at 4 K and 300 K. (9) Digitizers discriminate states into bits. (10) Results are aggregated into counts and returned. Steps 2, 3, 5, and 8–9 are where most of the engineering lives.