Bell's Theorem, Teleportation, and Quantum Cryptography
The Bell states and their correlations lead to the CHSH inequality, proved in full with its local bound 2 and its quantum maximum of two root two, and to the three experiments of 2015 that closed the loopholes at once. The unit then turns entanglement into a resource: it runs the teleportation protocol step by step, shows why nothing travels faster than light, and builds the BB84 and E91 protocols for a secret key, ending with the entangled photons that the Micius satellite sent to two stations 1,203 km apart.
Unit 8 closed the Foundation masterclass with a Bell inequality proved by counting and with the experiments that broke it. This unit returns to entanglement with the tools of Units 9 to 13 and treats it as a resource, which is what it has become since the 1990s. The four-setting inequality that every modern experiment uses is derived in full, its quantum maximum is computed from the spin states of Unit 11, and the three experiments of 2015 that closed the loopholes at once are examined closely enough to see what “loophole-free” means. The second half shows what an entangled pair can do that no classical channel can: carry a quantum state from Alice to Bob with two ordinary bits, and produce a secret key whose security rests on a Bell test instead of on the difficulty of a mathematical problem.
1 The Bell states and their correlations
Unit 13 built the four Bell states from a Hadamard gate and a CNOT gate. In the computational basis they read \begin{equation} \ket{\Phi^\pm} = \frac{1}{\sqrt{2}}\big(\ket{00} \pm \ket{11}\big) , \qquad \ket{\Psi^\pm} = \frac{1}{\sqrt{2}}\big(\ket{01} \pm \ket{10}\big) , \label{eq:bellstates} \end{equation} and they are orthogonal to one another, so together they form a basis of the two-qubit space, the Bell basis. Any one of them turns into any other under a single gate on one qubit alone. For example, an X gate on the second qubit of \(\ket{\Phi^+}\) gives \(\ket{\Psi^+}\), a Z gate gives \(\ket{\Phi^-}\), and X followed by Z, which is the Y gate up to a phase, gives \(\ket{\Psi^-}\). Section 4 uses this to repair a teleported state with at most two gates.
For Bell tests the most convenient of the four is \(\ket{\Psi^-}\). When the qubit is a spin-1/2 particle with \(\ket{0} = \ket{+z}\) and \(\ket{1} = \ket{-z}\), this state is the singlet of Unit 11, the state of two spins with total angular momentum zero. Unit 11 showed that the two spins of a singlet are found opposite when both are measured along \(z\); the derivation box below shows that the same holds along every axis, and works out what happens when the two axes differ.
In a Bell test Alice measures the spin of particle 1 along a direction \(\vec{a}\) and Bob measures particle 2 along \(\vec{b}\). Each records \(+1\) for spin up along the chosen axis and \(-1\) for spin down, so that Alice’s outcome \(A\) and Bob’s outcome \(B\) are numbers whose product is \(+1\) when the two agree and \(-1\) when they differ. All directions in this unit lie in the \(xz\)-plane, at the angles \(\theta_a\) and \(\theta_b\) from the \(z\) axis, which loses nothing because the settings that give the largest violation can always be chosen in one plane. The spin states along such a direction are those of Unit 11 with the azimuth angle \(\phi\) set to zero, \begin{equation} \ket{+a} = \cos\tfrac{\theta_a}{2}\,\ket{0} + \sin\tfrac{\theta_a}{2}\,\ket{1} , \qquad \ket{-a} = -\sin\tfrac{\theta_a}{2}\,\ket{0} + \cos\tfrac{\theta_a}{2}\,\ket{1} , \label{eq:spinstates} \end{equation} where the half angles are the reason that spins differ from the photon polarizations of Unit 8, as the end of this section explains.
The quantity every Bell experiment reports is the average of the product \(AB\) over many pairs, called the correlation \begin{equation} E(a,b) = P(+,+) + P(-,-) - P(+,-) - P(-,+) = \ev{AB} , \label{eq:corr} \end{equation} where \(P(+,-)\) is the probability that Alice finds \(+1\) and Bob finds \(-1\), and so on. The correlation is \(+1\) when the two outcomes always agree, \(-1\) when they are always opposite, and \(0\) when they are independent, and it is the only quantity the CHSH inequality of the next section needs.