From Bell Experiments to Quantum Information

Bell experiments transformed entanglement from a philosophical puzzle into an experimentally testable and operationally useful feature of quantum theory. EPR correlations raised the question of completeness and locality. Bell’s theorem made the question experimentally sharp. Later quantum information theory turned entanglement into a resource for communication, computation, cryptography, metrology, and device-independent certification.

For the original arguments and experiments, begin with EPR Argument, Bell’s Theorem as Historical Turning Point, and Bell Inequality Experiments. The state calculations use Bell States, Entangled States, and Entanglement in Quantum Information.

EPR Correlations

The EPR argument used perfect correlations to challenge the completeness of quantum mechanics. In modern two-spin language, the clean prototype is the singlet state

$$\lvert\psi^-\rangle = \frac{ \lvert01\rangle-\lvert10\rangle }{\sqrt2}.$$

If both parties measure spin along the same axis, their outcomes are perfectly anticorrelated. More generally, for measurement directions $\mathbf a$ and $\mathbf b$, quantum mechanics predicts

$$E(\mathbf a,\mathbf b) = -\mathbf a\cdot\mathbf b$$

for ideal spin-$1/2$ singlet measurements with outcomes encoded as $\pm1$.

The historical force of EPR was not that quantum mechanics fails to calculate correlations. It calculates them very well. The question was whether those correlations imply that the quantum state is incomplete, or whether nature itself violates assumptions about separability, locality, or predetermined values.

Bell Inequality Violations

Bell’s theorem showed that models with local factorization and setting independence obey inequalities that quantum mechanics can violate. In the CHSH scenario, Alice chooses between settings $a,a'$, Bob chooses between $b,b'$, and each outcome is $\pm1$. Define

$$S = E(a,b) + E(a,b') + E(a',b) - E(a',b').$$

Under the CHSH local hidden-variable assumptions,

$$\lvert S\rvert\le2.$$

Quantum mechanics can reach

$$\lvert S\rvert=2\sqrt2$$

for suitable measurements on an entangled state. The maximum quantum value is the Tsirelson bound, not the algebraic maximum $4$.

A Bell test compares correlations from separated measurement choices. Bell violations rule out local hidden-variable models satisfying the relevant assumptions, while quantum mechanics remains compatible with no-signaling.

The experimental story moved from early tests to Aspect’s time-varying analyzer experiments and then to modern loophole-free Bell tests. For a compact derivation, use CHSH Inequality, and the historical arc is Bell Inequality Experiments and Loophole-Free Bell Tests.

Entanglement as Nonclassical Correlation

Quantum information theory reframes entanglement as a structured nonclassical correlation between subsystems. A pure bipartite state is entangled when it cannot be written as a product

$$\lvert\psi\rangle_{AB} \ne \lvert\phi\rangle_A \otimes \lvert\chi\rangle_B.$$

For two qubits, the Bell states form a maximally entangled basis. The state $\lvert\psi^-\rangle$ is one example, but the broader language includes Schmidt decomposition, reduced density matrices, entanglement entropy, concurrence, witnesses, and LOCC transformations.

Bell nonlocality and entanglement are related but not identical:

• every Bell-violating state is entangled;
• not every entangled mixed state violates a given Bell inequality without further processing or choices;
• steering, entanglement, and Bell nonlocality are distinct levels of nonclassical correlation;
• Bell tests certify more than ordinary entanglement witnesses, but under different assumptions.

This distinction matters for reliable language. Entanglement is the resource concept used across quantum information. Bell violation is a stronger, scenario-dependent statement about observed correlations.

No-Signaling

Bell violations do not allow faster-than-light communication. The reason is visible in the reduced-state formalism. If Alice and Bob share $\rho_{AB}$, Alice’s local statistics are computed from

$$\rho_A = \operatorname{Tr}_B\rho_{AB}.$$

Bob’s choice of measurement basis cannot change Alice’s averaged local density operator. It can change correlations revealed when Alice and Bob later compare records, but it cannot transmit a controllable message through the marginal statistics alone.

For the ideal singlet, each local spin outcome has probability one half. More general Bell experiments can have biased marginals; no-signaling requires independence from the remote setting, not a uniform local distribution. The nonclassical structure appears in joint correlations:

$$E(a,b) = \sum_{A,B=\pm1} AB\,p(A,B|a,b).$$

No-signaling is not an optional patch added to hide nonlocality. It is built into the tensor-product and reduced-state structure of ordinary quantum mechanics. For the conceptual treatment, see Entanglement in Foundations and What These Experiments Do and Do Not Prove.

Device-Independent Protocols

Bell tests also created a new idea: certify some quantum behavior from observed input-output correlations, without fully trusting the internal description of the devices. This is the seed of device-independent quantum information.

In a device-independent setting, one treats the apparatus as a box with inputs and outputs. A Bell violation can certify that the behavior is incompatible with local hidden-variable models. With additional assumptions and finite-statistical analysis, such correlations can support:

• randomness certification;
• device-independent quantum key distribution;
• self-testing of states and measurements;
• entanglement certification without a detailed device model;
• security proofs based on observed nonlocal correlations.

Device-independent protocols still need assumptions: spacelike separation or trusted isolation as appropriate, independent setting choices, calibrated timing, finite-statistics treatment, and protection against side channels. A statistically significant violation does not automatically provide a positive secret-key rate: noise, finite data, leakage, and the protocol’s security proof determine what can be extracted.

Noise Separates Entanglement From CHSH Violation

Consider a singlet mixed with uniform two-qubit noise:

$$\rho_p=p\lvert\psi^-\rangle\langle\psi^-\rvert+(1-p)\frac{I_4}{4}, \qquad 0\le p\le1.$$

The noise contributes zero to spin-product correlations, so $E_p(\mathbf a,\mathbf b)=-p\,\mathbf a\cdot\mathbf b$. The largest CHSH magnitude with spin-component measurements is therefore $2\sqrt2p$, giving a violation only when $p>1/\sqrt2$.

At $p=0.6$, the CHSH maximum is approximately $1.70$. Yet the singlet probability is

$$\langle\psi^-\rvert\rho_p\lvert\psi^-\rangle =\frac{1+3p}{4}=0.70.$$

Any product state overlaps a maximally entangled two-qubit state with squared magnitude at most $1/2$; convex mixtures preserve that bound. Thus this state is entangled despite not violating CHSH with these measurements. This example does not claim that all other Bell scenarios or processing protocols behave the same way.

To work with the state, continue with Schmidt Decomposition and Entanglement Entropy. To understand what the apparatus must be trusted to do, study Certification of Entanglement. The broader learning route is Quantum Information.

Exercises

1. Why does a Bell violation rule out more than ordinary classical correlation?

Solution

Ordinary classical correlations can be produced by shared randomness. Bell inequalities constrain correlations that can arise from local hidden-variable models with shared randomness and local response functions. A Bell violation exceeds that bound, so it cannot be explained by shared randomness satisfying the Bell assumptions.

2. Explain why Bell violations do not allow Alice to signal to Bob.

Solution

Bob’s local statistics are determined by his reduced density operator. Alice’s choice of measurement can change correlations revealed after comparing data, but it does not change Bob’s averaged marginal distribution in a controllable way. Without the later classical comparison channel, Bob cannot infer Alice’s setting from his local outcomes alone.

3. For the noisy singlet above with $p=0.9$, calculate the maximum CHSH magnitude and the local density operator. Does the violation make either local output predictable?

Solution

The maximum is $2\sqrt2(0.9)\approx2.55>2$. Tracing either term over Bob gives $\rho_A=I_2/2$, so Alice’s spin outcomes remain equally likely on every axis. A larger joint correlation does not create a locally controllable result.

References

• A. Aspect, P. Grangier, and G. Roger, “Experimental Tests of Realistic Local Theories via Bell’s Theorem,” Physical Review Letters 47, 460-463, 1981.
• J. S. Bell, “On the Einstein Podolsky Rosen Paradox,” Physics 1, 195-200, 1964.
• D. Bohm, Quantum Theory, Prentice-Hall, 1951.
• N. Brunner, D. Cavalcanti, S. Pironio, V. Scarani, and S. Wehner, “Bell nonlocality,” Reviews of Modern Physics 86, 419-478, 2014.
• J. F. Clauser, M. A. Horne, A. Shimony, and R. A. Holt, “Proposed Experiment to Test Local Hidden-Variable Theories,” Physical Review Letters 23, 880-884, 1969.
• A. Einstein, B. Podolsky, and N. Rosen, “Can Quantum-Mechanical Description of Physical Reality Be Considered Complete?,” Physical Review 47, 777-780, 1935.
• B. Hensen et al., “Loophole-free Bell inequality violation using electron spins separated by 1.3 kilometres,” Nature 526, 682-686, 2015.
• M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information, Cambridge University Press, 2010.
• R. F. Werner, “Quantum states with Einstein-Podolsky-Rosen correlations admitting a hidden-variable model,” Physical Review A 40, 4277–4281, 1989, DOI: 10.1103/PhysRevA.40.4277.