Course 4 studio · QR-4
These are the words the month requires you to define explicitly. A passing week-4 card uses this level of precision — not circular slogans.
The complex number in front of a basis state; probability is the squared magnitude.
If a state is Σ α_i |i⟩, then |α_i|² is the probability of outcome i in the computational basis (Born rule). Phases (the arguments of the α_i) do not show up in that basis’s probabilities, but they drive interference in other bases and later in the circuit.
Why it matters. Confusing amplitude with probability is the most common week-2 error.
Related: Superposition · Interference · Measurement
The parameterized circuit shape you choose to variationally optimize (VQE/QAOA).
Hardware-efficient ansatze, problem-inspired ansatze, and QAOA layers are different bets. A bad ansatz cannot be saved by a better classical optimizer. Document it like you would document a neural-net architecture.
Why it matters. “We used VQE” is incomplete without the ansatz name and depth.
Related: VQE · QAOA
A defined comparison: same problem, stated metrics, stated conditions, reproducible enough to argue about.
Good benchmarks fix the instance, the metric (e.g. Sharpe, energy, PR-AUC, feasibility), the budget (time, shots, iterations), and the uncertainty. Bad benchmarks change the instance when the method changes. Hardware-vs-simulator must be labeled.
Why it matters. Week 4 capstones are graded on the table, not the adjective.
Related: Classical baseline · Metric
A classical unit of information that is either 0 or 1.
A bit is a definite binary value. Classical computers store, copy, and error-correct bits cheaply. Quantum algorithms still read out bits at the end; the quantum part is the state evolution before measurement.
Why it matters. Every quantum workflow still ends in bits. That is why classical baselines remain the comparison point.
Related: Qubit · Measurement
A sequence of gates and measurements on a register of qubits.
A quantum circuit is the standard programming model for gate-model machines. Depth, two-qubit gate count, and connectivity all affect whether a circuit is runnable on a given chip. Transpilation rewrites the circuit to the device’s native gates.
Why it matters. “We ran it on a quantum computer” always means some compiled circuit, shots, and a backend — not a metaphor.
Related: Quantum gate · Transpilation · Shot
The best honest classical method, metric, and setup you will compare against before claiming anything quantum.
A baseline names the problem instance, data, classical algorithm (or family), hyperparameters, hardware, and metric. “We beat a random guess” is not a baseline. “We beat a tuned classical solver on the same QUBO” might be. Hybrid work still needs the classical-only number.
Why it matters. This is Quantum Global Group’s non-negotiable. The Portfolio Lab exists to make the habit concrete.
Related: Benchmark · Quantum advantage · QUBO
A two-qubit gate: flip the target if the control is |1⟩. The standard way to create entanglement from a product state.
CNOT|c⟩|t⟩ → |c⟩|t ⊕ c⟩ in the computational basis. Starting from H on the control and |0⟩ on the target yields a Bell pair. In algorithms it is the workhorse “if” of reversible computing.
Why it matters. Entanglement is not a feeling; it is usually this gate plus superposition.
Related: Entanglement · Circuit
A correlation between quantum systems that cannot be explained by assigning each system its own independent state.
An entangled state of two or more qubits cannot be written as a product of single-qubit states. Measurement statistics can violate classical bounds (Bell). Entanglement is a resource for algorithms and communication; it is not faster-than-light signaling.
Why it matters. CNOT-after-Hadamard is the first circuit that makes this real. If you skip it, “entanglement” stays a slogan.
Related: CNOT · Bell pair · No-signaling
Computing with encoded logical qubits so that errors can be corrected faster than they accumulate.
Logical qubits are built from many physical qubits plus error-correcting codes. Resource estimates for useful fault-tolerant algorithms are still large. “We will have fault tolerance soon” is a research statement, not a procurement spec, unless a vendor gives a logical-qubit experiment with stated assumptions.
Why it matters. Separates roadmap slides from runnable work. Learners must know which bucket a claim is in.
Related: NISQ · Logical qubit
A score for how close two quantum states or how accurate a gate/process is; 1 is perfect.
State fidelity is a standard overlap-based measure between an ideal state and an experimental one. Gate and process fidelity describe operations. Reported “99.9% two-qubit fidelity” is about that operation, not about your full algorithm succeeding.
Why it matters. Vendor headlines quote fidelities. Practitioners ask: of what, on which gate, and what does that do to my circuit depth?
Related: Noise · NISQ
The gate that turns |0⟩ into an equal superposition of |0⟩ and |1⟩ (and vice versa, up to phase conventions).
H|0⟩ = (|0⟩+|1⟩)/√2. It is the usual way to open superposition in textbook circuits. Applying H to every qubit of |00…0⟩ prepares the uniform superposition used in many algorithms.
Why it matters. It is the most-used single-qubit gate in introductory circuits and in Qolour’s teaching sequence.
Related: Superposition · Quantum gate
A loop that uses both classical and quantum (or quantum-inspired) pieces, each doing what it is good at.
Almost all near-term application work is hybrid: classical data prep, encoding, parameter updates, and post-processing around a quantum or quantum-inspired kernel. “Hybrid” is not a dodge; it is the architecture.
Why it matters. IBM Business Foundations and this lab agree: quantum will not replace classical computers.
Related: VQE · QAOA · Classical baseline
Amplitudes add and cancel, so some measurement outcomes become more likely and others less.
Because amplitudes are complex, paths through a circuit can constructively or destructively interfere. Grover, QAOA phase separators, and many textbook algorithms are interference engineering. Noise wrecks interference, which is why NISQ results need careful baselines.
Why it matters. This is the mechanism behind most “why would this be faster?” stories that are not just analog simulation of chemistry.
Related: Amplitude · Superposition · QAOA
The operation that produces a classical outcome from a quantum state and typically collapses the state.
Measurement is defined relative to a basis. Repeating an experiment builds a distribution. You do not get the amplitudes themselves from a single shot. Shot count, readout error, and basis choice all belong in a results paragraph.
Why it matters. People under-count shots and over-claim certainty. Measurement is where statistics re-enter the story.
Related: Superposition · Shot · Fidelity
Noisy Intermediate-Scale Quantum: today’s devices, too noisy for full error correction, large enough to run interesting circuits.
Coined to name the era of 50–1000+ noisy qubits. Variational algorithms (VQE, QAOA) and analog / annealing approaches are NISQ-era bets. Fault-tolerant algorithms (Shor at cryptographically relevant scale, large Grover) are not NISQ deliverables.
Why it matters. Business Foundations week is partly about not selling fault-tolerant timelines as this quarter’s IT project.
Related: Fault tolerance · VQE · QAOA
Unwanted processes that make the real device differ from the ideal circuit.
Decoherence, gate error, crosstalk, and readout error all fall under noise. Simulators can be noiseless, noisy, or fake-backends. Ignoring noise is how toy demos become false advantage claims.
Why it matters. Week 4 comparisons must say whether the quantum path was noiseless simulation, noisy simulation, or hardware.
Related: Fidelity · NISQ · Shot
Classical cryptographic algorithms designed to resist attacks by large quantum computers (especially Shor’s algorithm).
PQC is not quantum communication and not QKD. NIST has standardized algorithms (e.g. ML-KEM, ML-DSA). Migration is a multi-year crypto-agility program: inventory, vendors, certificates, protocols. It is a parallel workforce track to quantum computing.
Why it matters. Security-pathway learners must not confuse “we need PQC” with “we need a quantum computer.”
Related: Shor’s algorithm · Quantum readiness
Quantum Approximate Optimization Algorithm: a variational circuit for combinatorial problems, usually from an Ising / QUBO cost.
QAOA alternates cost and mixer unitaries, with parameters tuned classically. It is a hybrid algorithm. Depth p, parameter training, and the quality of the QUBO encoding dominate results. It is not a magic replacement for Gurobi.
Why it matters. Appears in the Portfolio Lab and in IBM-adjacent application stories. Learners should know it is variational and hybrid.
Related: QUBO · VQE · Hybrid algorithm
A quantum (or hybrid) method that is better than the best relevant classical method on a stated metric for a stated problem.
Advantage is not a vibe. It requires a problem, a classical baseline, a quantum method, and a comparison (time, quality, energy, memory, cost). Quantum supremacy / quantum computational advantage historically referred to a contrived sampling task, not a business KPI.
Why it matters. This is the word executives misuse most. Week 4 exists so the cohort will not.
Related: Classical baseline · Benchmark · Quantum supremacy
A reversible operation on one or more qubits; the analog of a logic gate, but unitary.
Gates are unitary matrices (or physically, calibrated pulses that approximate them). Common one-qubit gates: H, X, Y, Z, T. The workhorse two-qubit gate is CNOT. A universal set can approximate any unitary to arbitrary accuracy.
Why it matters. If you cannot say what H and CNOT are for, you cannot read a circuit diagram in a paper or a vendor demo.
Related: Circuit · Hadamard · CNOT
A moving measure of whether a person or organization can use quantum technology and respond to quantum-era risk — not a certificate you hang once.
IBM describes organizational quantum readiness across strategy, technology, and operations: talent, use-case skill, hybrid architecture, governance, and protection against quantum threats. For a person, this month treats readiness as: know your fit, share the vocabulary, hold qubit intuition, and refuse results without a baseline.
Why it matters. It is the name of the program. It is not a synonym for “we bought a time on a QPU.”
Related: Use case · Post-quantum cryptography
A historical term for a quantum device performing a task believed to be infeasible for classical computers, often not a useful task.
Google’s 2019 sampling experiment popularized the phrase. The community has largely moved to “quantum computational advantage” and, for applications, to domain-specific advantage. Do not use supremacy when you mean “our optimizer looked good in a slide.”
Why it matters. Vocabulary hygiene. Using the wrong word destroys trust with technical reviewers.
Related: Quantum advantage
The basic unit of quantum information; unlike a bit, it has a state that can be a combination of 0 and 1 until measured.
A qubit is a two-level quantum system whose state is a normalized vector in a two-dimensional complex space, often written α|0⟩ + β|1⟩. Measurement in a chosen basis yields a classical outcome and generally destroys the prior superposition. Hardware implementations vary (superconducting circuits, trapped ions, photonics, spins); the information model is shared.
Why it matters. If you cannot say what a qubit is without saying “it’s like a bit but magic,” you are not ready to evaluate a vendor or a paper.
Related: Superposition · Measurement · Bit
Quadratic Unconstrained Binary Optimization: a standard way to write discrete problems for annealers and many quantum-inspired solvers.
A QUBO minimizes xᵀ Q x for binary x. Many portfolio, routing, and selection problems can be encoded this way, often with penalty terms for constraints. QAOA and quantum annealing families consume QUBO / Ising forms. Encoding quality often matters more than the solver brand.
Why it matters. This is the bridge from “I have a business constraint” to “I have something a quantum-inspired method can eat.”
Related: QAOA · Ising · Classical baseline
One execution of a circuit ending in measurement; statistics need many shots.
Estimating an expectation or a probability to useful precision can take thousands of shots. Hardware queues, cost, and noise all scale with shots. Always report shot count next to a quantum number.
Why it matters. Missing shot counts are a grading fail in week 4.
Related: Measurement · Benchmark
A qubit (or register) can occupy a combination of basis states, with complex amplitudes, until it is measured.
Superposition is not “being in two places” as a metaphor for indecision. It is a linear combination of basis states. Algorithms use superposition so that a circuit can act on many basis states at once; interference then shapes which outcomes become likely.
Why it matters. Without superposition there is no quantum parallelism to interfere. It is necessary, not sufficient, for advantage.
Related: Interference · Measurement · Amplitude
Rewriting a circuit to the gates and connectivity a specific backend actually supports.
Logical CNOT on non-adjacent qubits becomes a chain of swaps. That inflates depth and error. Comparing two “same” circuits on different backends without transpilation notes is not a benchmark.
Why it matters. Software-pathway people live here. Everyone else must know it exists.
Related: Circuit · Noise
A specific problem, owner, data, and success metric — not an industry name on a slide.
IBM’s course distinguishes good use cases (strategic, scalable, explorably small, hybrid-near-term) from poor ones (mission-critical bets on unproven theory, or problems that do not scale). A use case that cannot name a classical baseline is not ready for a quantum pilot.
Why it matters. Week 3 memo. If you cannot write it, you cannot fund it.
Related: Quantum readiness · Classical baseline
Variational Quantum Eigensolver: a hybrid method that uses a parameterized circuit and a classical optimizer to estimate an energy / cost.
VQE was designed for chemistry Hamiltonians and is also used as a variational cost minimizer in other encodings. The ansatz, optimizer, and shot noise determine whether you are doing science or fitting a random landscape.
Why it matters. IBM Runtime paths in this repo use VQE-shaped workflows. The cohort should not confuse “we ran VQE” with “we beat classical.”
Related: Hybrid algorithm · Ansatz · QAOA