Skip to content
Tutorials Center
My learning

Topics

Lesson 34 of 39 ยท Entanglement, Measurement and Decoherence

Entanglement and Bell's Inequality

Test local hidden variables against quantum mechanics with the CHSH inequality: classical bound 2, quantum maximum 2.83, computed and simulated in Python.

In 1935 Einstein, Podolsky and Rosen argued that quantum mechanics must be incomplete. Their reasoning used a pair of particles in an entangled state: measuring one particle seems to fix the result for the other, however far away. Perhaps, they thought, each particle carried hidden instructions from the start, and the apparent "spooky action at a distance" was just the revelation of pre-existing values. In 1964 John Bell showed that this idea is testable: any theory of local hidden variables obeys an inequality that quantum mechanics violates. Experiments since the 1970s (the 1982 results of Aspect, and loophole-free tests in 2015 and later) support quantum mechanics. Clauser, Aspect and Zeilinger shared the 2022 Nobel Prize in Physics for this work.

Entangled states

A two-qubit state is separable if it can be written as a mixture of product states, and entangled otherwise. A pure state is entangled if and only if it is not a product |aโŸฉ|bโŸฉ. The standard examples are the four Bell states, such as the singlet

Text
|ฯˆโปโŸฉ = (|โ†‘โ†“โŸฉ - |โ†“โ†‘โŸฉ)/โˆš2

The singlet has total spin zero. If you measure the spin of both particles along the same axis, you always find opposite results; this is true for any axis you pick. For different axes a and b at angle ฮธ_ab the results are correlated according to the correlation

Text
E(a, b) = โŸจฯƒ_a โŠ— ฯƒ_bโŸฉ = -cos(ฮธ_ab)

where ฯƒ_a is the Pauli operator for spin along direction a (outcomes ยฑ1).

The CHSH inequality

Alice measures along one of two directions, a or a'. Bob measures along b or b'. Each outcome is ยฑ1. Define the Clauser-Horne-Shimony-Holt combination

Text
S = E(a, b) - E(a, b') + E(a', b) + E(a', b')

Local hidden variables. Suppose each pair carries a hidden variable ฮป that, together with the chosen setting, determines the outcomes: A(a, ฮป) = ยฑ1 for Alice and B(b, ฮป) = ยฑ1 for Bob. Locality means Alice's outcome does not depend on Bob's setting, and vice versa. For every ฮป,

Text
A(a)B(b) - A(a)B(b') + A(a')B(b) + A(a')B(b') = A(a)[B(b) - B(b')] + A(a')[B(b) + B(b')]

Since B(b) and B(b') are each ยฑ1, one of the brackets is 0 and the other is ยฑ2, so the whole expression is ยฑ2. Averaging over ฮป:

Text
|S| โ‰ค 2        (Bell-CHSH inequality)

Quantum mechanics. With the singlet and measurement directions in a plane, E(a, b) = -cos(a - b) for angles a, b. Choose a = 0, a' = ฯ€/2, b = ฯ€/4, b' = 3ฯ€/4. Then three correlations are -1/โˆš2 and one is +1/โˆš2, giving

Text
S = -2โˆš2 โ‰ˆ -2.828,        |S| = 2โˆš2 > 2

The inequality is violated. The value 2โˆš2 is the largest that quantum mechanics permits (Tsirelson's bound); a hypothetical theory could reach up to 4 without allowing signals faster than light, but nature stops at 2โˆš2.

Python Quantum CHSH value from the singlet state

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

No local hidden-variable model reaches it

First, count all deterministic strategies. A deterministic hidden variable assigns each of the four outcomes A(a), A(a'), B(b), B(b') a value ยฑ1; there are only 2โด = 16 possibilities. Any local hidden-variable model is a probabilistic mixture of these, so its S cannot exceed the best of the 16. You can check by brute force. Then try a natural random model: each pair gets a random angle ฮป and each detector reads the sign of cos(setting - ฮป). It reproduces perfect anticorrelation at equal settings, which is the simplest feature of the singlet, but its CHSH value stops at 2.

Python Deterministic strategies and a sampled hidden-variable model

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The sampled model matches the quantum predictions at several settings (perfect anticorrelation at equal angles, zero at right angles), but at the CHSH angles it gives |S| โ‰ˆ 2 instead of 2.83. No local assignment of pre-existing outcomes can do better.

What does the violation mean?

Experiments find |S| > 2 (values near 2.4 to 2.8 in good experiments). The assumptions behind the inequality are:

  1. Realism. Outcomes are determined by properties existing before the measurement (the hidden variable ฮป).
  2. Locality. An outcome at Alice's side does not depend on Bob's setting.
  3. Freedom of choice. The settings can be chosen independently of ฮป.

Quantum predictions violate the inequality, so at least one of these must be given up. Which one is a matter of interpretation (the next lesson discusses this honestly). Two points are agreed by everyone:

  • No signaling. Entanglement cannot be used to send information faster than light. Each side alone sees random outcomes; the correlations only show up when the two records are compared through an ordinary channel.
  • The correlations are real. "Local realism" in the sense defined above is experimentally excluded, once the experimental loopholes (detection efficiency, communication between the stations, the choice of settings) are closed, as they have been in several experiments since 2015.

Entanglement is also a resource: quantum teleportation, superdense coding and quantum key distribution protocols use it. Bell-test violation is used to certify the security of some device-independent cryptographic schemes.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What is the largest value of |S| allowed for local hidden-variable theories?

    For every hidden-variable value ฮป the combination A(a)[B(b) - B(b')] + A(a')[B(b) + B(b')] equals ยฑ2, so the average cannot exceed 2.
  2. 2. For the singlet state measured along angles a and b in a plane, the correlation E(a, b) is:

    The spins are perfectly anticorrelated for equal axes (E = -1), and E = -cos of the angle between the axes.
  3. 3. Which statement about entanglement and signaling is correct?

    Bob's marginal statistics do not depend on Alice's setting. The correlations appear only when the results are compared through an ordinary channel.
  4. 4. A Bell test with |S| = 2.7 shows that:

    |S| > 2 violates the CHSH bound derived from locality, realism and freedom of choice, so at least one assumption fails. The value 2.7 is below Tsirelson's bound, consistent with quantum theory.

Summary

  • Entangled states cannot be written as mixtures of product states; the singlet has E(a, b) = -cos(a - b).
  • Local hidden-variable theories satisfy |S| โ‰ค 2 for the CHSH combination.
  • Quantum mechanics reaches |S| = 2โˆš2 โ‰ˆ 2.83 (Tsirelson's bound) at angles 0, ฯ€/2, ฯ€/4, 3ฯ€/4.
  • Experiments, including loophole-free ones, observe violations, so local realism is untenable.
  • Entanglement does not allow faster-than-light signaling.

Next: The Measurement Problem and Interpretations