Orientation Guides

Multi-Qubit Circuits, Entanglement, and No-Hype Protocols

Add a second qubit and the state space stops being a collection of independent parts: gates become conditional, correlations appear that no local description can reproduce, and the word entanglement earns its keep. This guide builds the smallest entangling circuit by hand and strips the hype from the three protocols everyone name-drops.

Listen to this chapter

Entanglement is the most abused word in quantum computing, and the abuse is unnecessary. The concept has a one-line mathematical test, and everything the field actually uses — protocols, algorithms, error correction — follows from that test, not from any spooky story told around it.

By the end you will be able to derive a Bell state with pencil and paper, decide for yourself whether a given two-qubit state is entangled, and explain superdense coding, teleportation, and quantum key distribution in terms of resources spent and results delivered.

Core concepts: tensor product, CNOT, Bell state, separability, no-signaling, no-cloning.

Notice what the meters cannot show you: each wire alone is a coin flip, yet the two flips always agree. The correlation lives in the pair, not in either qubit — and that is all entanglement means.

What changes when a second qubit arrives

Multi-qubit computing is where the state space grows exponentially, gates become conditional, and entanglement appears. It is also where bad explanations become tempting, so this guide keeps four disciplines in view throughout:

Combine systems with tensor products, never with wishful notation.
Use multi-qubit gates to create correlations that no collection of independent one-qubit states can represent.
Use measurement to extract classical data — and expect it to destroy the rest.
Respect no-signaling: entanglement never sends usable information faster than light.

Basis ordering

For two qubits, this book uses a fixed ordering, and the fixedness matters:

∣00 ⟩ ∣01 ⟩ ∣10 ⟩ ∣11 ⟩ = 1000 = 0100 = 0010 = 0001

A general two-qubit pure state is $∣ ψ ⟩ = a_{00} ∣00 ⟩ + a_{01} ∣01 ⟩ + a_{10} ∣10 ⟩ + a_{11} ∣11 ⟩$ , with the squared magnitudes of the four amplitudes summing to one. Most early mistakes in multi-qubit work are ordering mistakes; decide the convention once and never improvise it again.

Tensor products

Combining two one-qubit states multiplies every component of the first vector by every component of the second:

(a b) \otimes (c d) = a c a d b c b d

Worked through on basis states:

∣0 ⟩ \otimes ∣1 ⟩ = (10) \otimes (01) = 1 \cdot 0 1 \cdot 1 0 \cdot 0 0 \cdot 1 = 0100 = ∣01 ⟩

Two qubits give a four-dimensional state space; ten give a thousand; three hundred give more amplitudes than there are atoms in the observable universe. That growth is the source of both the excitement and the simulation cost.

The CNOT gate

CNOT uses the first qubit as control and the second as target, flipping the target exactly when the control is 1:

∣00 ⟩ ∣01 ⟩ ∣10 ⟩ ∣11 ⟩ \mapsto ∣00 ⟩ \mapsto ∣01 ⟩ \mapsto ∣11 ⟩ \mapsto ∣10 ⟩

As a matrix in the fixed basis ordering:

CNOT = 1000010000010010

It is the quantum version of an if-statement, and because it is linear it acts on superpositions branch by branch — which is exactly how the next example manufactures entanglement.

Worked example: create a Bell state

Start with $∣00 ⟩$ and apply $H$ to the first qubit:

(H \otimes I) ∣00 ⟩ = \frac{∣0 ⟩ + ∣1 ⟩}{2} \otimes ∣0 ⟩ = \frac{∣00 ⟩ + ∣10 ⟩}{2}

Now apply CNOT with the first qubit as control. It leaves $∣00 ⟩$ alone and flips the second branch:

CNOT \frac{∣00 ⟩ + ∣10 ⟩}{2} = \frac{CNOT ∣00 ⟩ + CNOT ∣10 ⟩}{2} = \frac{∣00 ⟩ + ∣11 ⟩}{2}

The result is the Bell state $∣ Φ^{+} ⟩ = (∣00 ⟩ + ∣11 ⟩) / 2$ . Measure both qubits in the computational basis and you get 00 half the time, 11 half the time, and never 01 or 10. The outcomes are perfectly correlated: whatever the first qubit shows, the second shows the same.

Why this is entangled

A two-qubit product state factors into independent one-qubit states, $(a ∣0 ⟩ + b ∣1 ⟩) \otimes (c ∣0 ⟩ + d ∣1 ⟩)$ , which expands to $a c ∣00 ⟩ + a d ∣01 ⟩ + b c ∣10 ⟩ + b d ∣11 ⟩$ . Compare the coefficients and you get a test: a two-qubit pure state is separable only if its amplitudes satisfy

a_{00} a_{11} = a_{01} a_{10}

For $Φ^{+}$ , the left side is $\frac{1}{2} \cdot \frac{1}{2} = \frac{1}{2}$ and the right side is $0 \cdot 0 = 0$ . The test fails, so no choice of local one-qubit states can reproduce the joint state. That is the whole definition: entangled means not factorable. No distance, no drama, just algebra.

Entanglement is not ordinary correlation

A classical random process can also produce correlated bits — 00 half the time, 11 half the time — and in one measurement basis that looks exactly like the Bell state. The difference shows up when you measure in other bases: Bell states produce structured correlations there too, and no classical local hidden-variable model can reproduce the full set of statistics observed in Bell-type experiments.

For computation the practical point is simpler still: entanglement means the joint state cannot be represented as independent local states. It is a state-space and correlation-structure resource — not a communication loophole, and not evidence that the universe is whispering between particles.

No faster-than-light signaling

The correlations are real, but they carry no controllable message. Three facts close every loophole:

Each local measurement outcome is individually random.
The correlation becomes visible only when the two parties compare classical records.
Comparing records requires ordinary, slower-than-light classical communication.

Teleportation respects the same rule. It transfers an unknown quantum state using a shared entangled pair plus two classical bits, and the receiver cannot reconstruct anything until the classical bits arrive. Relativity survives quantum mechanics intact.

No-cloning in one page

Suppose a universal cloning operation $U$ existed, with $U ∣ ψ ⟩ ∣0 ⟩ = ∣ ψ ⟩ ∣ ψ ⟩$ for any state $∣ ψ ⟩$ . Applied to two different inputs it would give $U ∣ a ⟩ ∣0 ⟩ = ∣ a ⟩ ∣ a ⟩$ and $U ∣ b ⟩ ∣0 ⟩ = ∣ b ⟩ ∣ b ⟩$ .

Unitary operations preserve inner products. Before cloning, the inner product of the two input states is $⟨ a ∣ b ⟩$ , since the blank state contributes a factor of one. After cloning it is $(⟨ a ∣ b ⟩)^{2}$ . Preservation demands $⟨ a ∣ b ⟩ = (⟨ a ∣ b ⟩)^{2}$ , which holds only when the states are identical or perfectly distinguishable — never for arbitrary unknown states.

So universal cloning is impossible, and the proof is three lines of linear algebra. This is not an engineering limitation awaiting a clever fix; it is a theorem, and Part III shows how the field turns it from a constraint into a security resource.

Protocols without hype

Superdense coding

Resource: one shared Bell pair, plus one qubit sent from Alice to Bob.

Outcome: Bob recovers two classical bits from Alice's single transmitted qubit.

What it does not mean:

It does not transmit information without a physical qubit making the trip.
It does not beat relativity.
It spends pre-shared entanglement; the pair is consumed.

Teleportation

Resource: one shared Bell pair, plus two classical bits sent from Alice to Bob.

Outcome: Bob reconstructs Alice's unknown qubit state.

What it does not mean:

It does not move an object; it moves the state of a qubit.
It does not copy the state — Alice's original is destroyed, as no-cloning requires.
It does not complete before the classical bits arrive.

BB84 and quantum key distribution

Resource: quantum states prepared in non-commuting bases, plus an authenticated classical channel.

Outcome: a shared key with eavesdropping detection, under clearly stated assumptions about the devices and the channel.

What it does not replace: the migration to post-quantum cryptography across ordinary internet infrastructure, which solves a different problem for a fraction of the operational cost. Part VIII weighs both honestly.

What this means on hardware

Entangling gates are usually the hard part of a machine. One-qubit rotations are comparatively easy; making two qubits interact strongly enough to entangle, while keeping cross-talk, leakage, calibration drift, and measurement error under control, is where modalities differ. The questions that matter:

What is the native entangling gate?
What is its measured fidelity?
How does that fidelity change as the system grows?
How connected are the qubits?
Does routing force long chains of swap gates?
Can error-correction cycles run fast enough to keep up?

The application connection

Entanglement is necessary for many quantum advantages, but it is never sufficient by itself. The weak version of an application claim runs: "this uses entanglement, so it will beat classical computing." The strong version names the problem structure being exploited, the algorithm that exploits it, the hardware that can run the circuit within error limits, and the classical baseline that loses under stated assumptions. Anything between those two poles is marketing.

What this buys you in diligence

When a company shows you an entanglement demo, the serious questions are:

Can the system create entanglement at scale, not just between two qubits?
Can it preserve entanglement long enough to compute with it?
Can it measure some qubits without destroying information still needed later?
Can it run repeated error-correction cycles?
Do logical operations improve as code distance or encoding size increases?

Entanglement demos are table stakes. Company value lives in reliable, repeatable, scalable operation.

Exercises

Compute $∣1 ⟩ \otimes ∣0 ⟩$ in the fixed basis ordering.
Write CNOT's action on all four two-qubit basis states from memory.
Derive the Bell state from $∣00 ⟩$ using $H$ on the first qubit and then CNOT.
Test whether $(∣00 ⟩ + ∣01 ⟩) / 2$ is entangled.
Test whether $(∣00 ⟩ + ∣11 ⟩) / 2$ is entangled.
Explain why no-cloning does not prevent copying classical bits.
Explain teleportation to a technical audience without science-fiction language.
Pick a hardware company and identify its main entangling operation or error-correction route.

Check your understanding

You have this guide when you can, without notes:

Use tensor products with a fixed basis ordering.
Derive a Bell state by hand.
Distinguish entanglement from ordinary correlation, and say what experiment separates them.
Explain no-cloning through inner-product preservation.
Explain why teleportation and entanglement allow no faster-than-light message.