Chapter 39 Exercises: Bell Tests, Entanglement, and Reality

Part A: Conceptual Questions (one star)

These questions test your understanding of the core ideas. No calculations required.

A.1 State the two assumptions of local realism precisely. For each, give a concrete physical example of what it means and a concrete example of what violating it would look like.

A.2 A friend says: "Bell's theorem proves that quantum mechanics is nonlocal — entangled particles communicate faster than light." Is this correct? Explain carefully what Bell's theorem does and does not prove, and distinguish between nonlocal correlations and nonlocal signaling.

A.3 Explain why the detection loophole is a genuine logical concern and not just an experimental inconvenience. Construct a verbal argument for how a local hidden variable model could exploit imperfect detectors to fake a Bell violation.

A.4 In the E91 protocol, why does a high CHSH value guarantee security? Explain the physical reasoning connecting Bell violation to the absence of eavesdropper information.

A.5 What is the freedom-of-choice loophole? Why is it qualitatively different from the detection and locality loopholes? Can it ever be completely closed?

A.6 Explain the conceptual difference between BB84 and E91. Which protocol's security relies on the no-cloning theorem, and which relies on Bell's theorem? Can you describe a scenario where one protocol is secure but the other is not?

A.7 What does "device-independent" mean in the context of quantum key distribution? Why is this property desirable? Under what (extreme) threat model does DI-QKD provide security that standard QKD does not?

A.8 The Tsirelson bound states that $|S| \leq 2\sqrt{2}$ for quantum mechanics. The algebraic maximum is $|S| = 4$. Explain in physical terms why quantum mechanics does not saturate the algebraic maximum. What physical principle prevents $|S| = 4$?

Part B: Applied Problems (two stars)

These problems require direct application of the chapter's key equations and concepts.

B.1: CHSH for Non-Optimal Settings

Consider a Bell test with the singlet state $|\Psi^-\rangle$ where Alice measures at angles $\alpha = 0°$ and $\alpha' = 90°$ from the $z$-axis, and Bob at $\beta = 30°$ and $\beta' = -30°$.

(a) Compute the four quantum correlations $E_{\text{QM}}(\hat{a}, \hat{b})$, $E_{\text{QM}}(\hat{a}, \hat{b}')$, $E_{\text{QM}}(\hat{a}', \hat{b})$, $E_{\text{QM}}(\hat{a}', \hat{b}')$.

(b) Compute the CHSH parameter $S$. Does it violate the inequality?

(c) How does this compare with the optimal value of $2\sqrt{2}$? What fraction of the maximum violation is achieved?

(d) Optimize Alice's settings (keeping Bob's fixed at $30°$ and $-30°$) to maximize $|S|$. What are the optimal angles for Alice?

B.2: Statistical Power

You are designing a Bell test with the singlet state and optimal CHSH settings.

(a) Derive the approximate number of trials $N$ needed to detect a Bell violation at $3\sigma$ significance.

(b) Repeat for $5\sigma$ significance.

(c) If your source produces $1{,}000$ entangled pairs per second and your detector efficiency is $\eta = 85\%$, how many seconds of data collection are needed for a $5\sigma$ result?

(d) Now suppose your visibility is $V = 80\%$ instead of $V = 1$. How does this change the required data collection time? (Hint: the expected CHSH value decreases, so you need more data to distinguish it from $2$.)

B.3: Werner States and Bell Violations

A Werner state is $\hat{\rho} = V|\Psi^-\rangle\langle\Psi^-| + \frac{1-V}{4}\hat{I}_4$, where $V$ is the visibility.

(a) Show that the CHSH parameter for a Werner state with optimal settings is $S = 2\sqrt{2}V$.

(b) Find the minimum visibility $V_{\min}$ for a CHSH violation.

(c) Show that for $V < 1/\sqrt{2}$, the Werner state admits a local hidden variable model (you may cite Bell's result rather than constructing the model explicitly).

(d) For $1/3 < V < 1/\sqrt{2}$, the Werner state is entangled (it violates the Peres-Horodecki criterion) but does not violate the CHSH inequality. What does this tell you about the relationship between entanglement and Bell violation?

B.4: CHSH with the $|\Phi^+\rangle$ State

Consider the Bell state $|\Phi^+\rangle = \frac{1}{\sqrt{2}}(|00\rangle + |11\rangle)$ instead of the singlet.

(a) Compute the quantum correlation $E_{\text{QM}}(\hat{a}, \hat{b})$ for this state, where $\hat{a}$ and $\hat{b}$ are measurement axes in the $xz$-plane at angles $\alpha$ and $\beta$ from the $z$-axis. (Hint: the singlet gives $-\cos\theta$; what does $|\Phi^+\rangle$ give?)

(b) Find the optimal CHSH settings for $|\Phi^+\rangle$.

(c) Compute the maximum CHSH value. Is it the same as for the singlet?

(d) Explain why all maximally entangled two-qubit states give the same maximum CHSH value.

B.5: Detection Loophole Threshold

In a Bell test with imperfect detectors (efficiency $\eta$), undetected events are assigned outcome $+1$ (the "no-detection = $+1$" strategy).

(a) For the singlet state with optimal CHSH settings, show that the effective CHSH parameter becomes $S_{\text{eff}} = \eta^2 \cdot 2\sqrt{2} + (1 - \eta^2) \cdot S_{\text{no-click}}$, where $S_{\text{no-click}}$ depends on the assignment strategy.

(b) For the strategy "assign $+1$ to all no-detection events," compute $S_{\text{no-click}}$.

(c) Find the minimum efficiency $\eta_{\min}$ for which $|S_{\text{eff}}| > 2$.

(d) Eberhard's inequality has a lower detection efficiency threshold than CHSH. Why might an experimentalist prefer Eberhard's inequality even though it is harder to violate maximally?

B.6: The Tsirelson Bound

(a) Define the CHSH operator $\hat{\mathcal{S}} = \hat{A} \otimes (\hat{B} - \hat{B}') + \hat{A}' \otimes (\hat{B} + \hat{B}')$, where $\hat{A}, \hat{A}', \hat{B}, \hat{B}'$ are Hermitian operators with eigenvalues $\pm 1$.

(b) Show that $\hat{\mathcal{S}}^2 = 4\hat{I} - [\hat{A}, \hat{A}'] \otimes [\hat{B}, \hat{B}']$.

(c) Show that $\|[\hat{A}, \hat{A}']\| \leq 2$ for operators with eigenvalues $\pm 1$, and similarly for $[\hat{B}, \hat{B}']$.

(d) Conclude that $\|\hat{\mathcal{S}}\| \leq 2\sqrt{2}$.

(e) Show that equality holds when $\hat{A} = \hat{\sigma}_z$, $\hat{A}' = \hat{\sigma}_x$, $\hat{B} = (\hat{\sigma}_z + \hat{\sigma}_x)/\sqrt{2}$, $\hat{B}' = (\hat{\sigma}_z - \hat{\sigma}_x)/\sqrt{2}$.

Part C: Advanced and Synthesis Problems (three stars)

These problems require deeper analysis, synthesis of multiple concepts, or computational work.

C.1: Building a Local Hidden Variable Model

Construct an explicit local hidden variable model for the following scenario:

The hidden variable $\lambda$ is a unit vector uniformly distributed on the unit circle (2D, not 3D).

(a) Define Alice's outcome as $A(\hat{a}, \lambda) = \text{sign}(\hat{a} \cdot \lambda)$ and similarly $B(\hat{b}, \lambda) = -\text{sign}(\hat{b} \cdot \lambda)$. Compute $E(\hat{a}, \hat{b})$ as a function of $\theta_{ab}$ (the angle between $\hat{a}$ and $\hat{b}$).

(b) Compute the CHSH parameter for this model with the optimal quantum settings. Verify that $|S| = 2$.

(c) Now try to beat this model. Can you find any local hidden variable model (any $A$, $B$, $\rho(\lambda)$) that gives $|S| > 2$? Explain why or why not.

(d) Modify the model so that Alice's outcome is a noisy version: $A(\hat{a}, \lambda) = \text{sign}(\hat{a} \cdot \lambda)$ with probability $p$ and $-\text{sign}(\hat{a} \cdot \lambda)$ with probability $1-p$. What is the maximum CHSH value as a function of $p$? For what range of $p$ is there a violation?

C.2: Bell Test with Qutrit States

Extend the analysis to three-dimensional systems (qutrits).

(a) The maximally entangled qutrit state is $|\Psi\rangle = \frac{1}{\sqrt{3}}(|00\rangle + |11\rangle + |22\rangle)$. Write out this state explicitly in the $9$-dimensional product basis.

(b) The Collins-Gisin-Linden-Massar-Popescu (CGLMP) inequality is the natural generalization of CHSH to qutrits. It has a classical bound of $2$ and a quantum maximum of approximately $2.9149$ (compared to $2\sqrt{2} \approx 2.8284$ for CHSH). What does the larger violation ratio tell us about qutrit entanglement versus qubit entanglement?

(c) Discuss the implications for device-independent QKD: would qutrit-based protocols offer higher key rates than qubit-based protocols?

C.3: Simulating Loopholes (Computational)

Using the code from code/example-01-bell-complete.py as a starting point:

(a) Implement the detection loophole. Add a parameter $\eta$ (detection efficiency) and simulate a Bell test where each particle is independently detected with probability $\eta$. Only pairs where both are detected contribute to the CHSH calculation. Plot $S$ vs. $\eta$ for $\eta$ from $0.5$ to $1.0$ and verify the threshold at $\eta \approx 0.828$.

(b) Implement the locality loophole. Suppose Bob's detector has access to Alice's setting with probability $\epsilon$ (modeling imperfect spacelike separation). Design a local hidden variable strategy that exploits this information and plot $S$ vs. $\epsilon$.

(c) Combine both loopholes. Find the region in $(\eta, \epsilon)$ space where a local hidden variable model can fake a Bell violation.

C.4: Quantum Key Distribution Simulation

(a) Implement the BB84 protocol. Simulate Alice preparing qubits, Eve intercepting (intercept-resend attack), and Bob measuring. Plot the quantum bit error rate (QBER) as a function of Eve's interception probability.

(b) Implement the E91 protocol. Simulate Alice and Bob measuring entangled pairs, computing CHSH for the test settings and extracting the key from the matched-basis settings. Verify that without eavesdropping, $S \approx 2\sqrt{2}$ and the key bits are anticorrelated.

(c) Add an eavesdropper to the E91 protocol who performs an optimal individual attack. Show that the CHSH value decreases and the QBER increases. Plot the relationship between $S$ and QBER.

(d) Compute the secure key rate as a function of QBER for both protocols.

C.5: The Fine-Grained Uncertainty Relation and Bell Violations

The connection between Bell violations and uncertainty relations is deep. Consider the following:

(a) Show that for a single qubit, the outcomes of measurements along $\hat{a}$ and $\hat{a}'$ satisfy an uncertainty relation: $\Delta A \cdot \Delta A' \geq |\langle [\hat{A}, \hat{A}']\rangle / 2|$.

(b) Compute this for $\hat{A} = \hat{\sigma}_z$ and $\hat{A}' = \hat{\sigma}_x$ in the state $|+z\rangle$.

(c) Argue (qualitatively) that the Tsirelson bound $2\sqrt{2}$ is a consequence of the uncertainty principle: the more strongly Alice's measurements are incompatible, the stronger the Bell violation — but the uncertainty principle limits how incompatible they can be.

(d) In a hypothetical "superquantum" theory where the uncertainty principle is relaxed, could $|S| > 2\sqrt{2}$? Relate this to the Popescu-Rohrlich (PR) box, which achieves $|S| = 4$.

C.6: Historical Analysis

(a) Read Bell's original 1964 paper "On the Einstein Podolsky Rosen Paradox" (Physics 1, 195-200). It is only six pages. Summarize the key argument in your own words. What assumptions does Bell make that are often overlooked in textbook treatments?

(b) Bell's paper considers the original EPR scenario (position-momentum entanglement), not the CHSH version. How does the CHSH inequality improve upon Bell's original inequality for experimental purposes?

(c) The original Freedman-Clauser (1972) experiment tested a different form of Bell inequality than CHSH. What was it, and why did Clauser, Horne, Shimony, and Holt (1969) develop the CHSH form?

C.7: Self-Testing the Singlet

The following is an outline of the self-testing argument. Fill in the details.

(a) Suppose Alice and Bob observe $S = 2\sqrt{2}$ in a CHSH experiment. Show that this implies $\langle \hat{\mathcal{S}} \rangle = 2\sqrt{2}$, which means $|\psi\rangle$ is an eigenstate of $\hat{\mathcal{S}}$ with eigenvalue $2\sqrt{2}$.

(b) Show that $\hat{\mathcal{S}}|\psi\rangle = 2\sqrt{2}|\psi\rangle$ implies the following relations: - $\hat{A}|\psi\rangle = \frac{1}{\sqrt{2}}(\hat{B} + \hat{B}')|\psi\rangle$ - $\hat{A}'|\psi\rangle = \frac{1}{\sqrt{2}}(\hat{B} - \hat{B}')|\psi\rangle$

(c) From these relations, derive that $\hat{A}$ and $\hat{A}'$ anticommute on $|\psi\rangle$, and similarly $\hat{B}$ and $\hat{B}'$.

(d) Argue that the anticommutation relations, combined with the eigenvalue-$\pm 1$ constraint, imply that each subsystem is effectively a qubit and the state is (up to local isometry) the singlet state.

C.8: Bell Violations and Relativity

(a) In special relativity, Alice and Bob's measurements in a Bell test can be spacelike separated — neither event is in the light cone of the other. Explain why this means there is no frame of reference in which Alice's measurement "causes" Bob's outcome or vice versa.

(b) If there is no causal connection, how do the correlations arise? Discuss this from the perspectives of (i) Copenhagen, (ii) Many-Worlds, and (iii) Bohmian mechanics.

(c) In Bohmian mechanics, the nonlocal influence is frame-dependent — different Lorentz frames disagree on which measurement happened "first." Is this a problem for the theory? Does it conflict with special relativity?

(d) Some physicists argue that Bell correlations are best understood as "retrocausal" — the future measurement settings influence the past preparation of the state. Evaluate this proposal. What are its advantages and disadvantages?

C.9: The Kochen-Specker Theorem and Contextuality

While Bell's theorem rules out local hidden variables, the Kochen-Specker (KS) theorem rules out non-contextual hidden variables.

(a) State the Kochen-Specker theorem. What is the difference between "local" and "non-contextual" hidden variables?

(b) The simplest KS proof uses 18 vectors in a four-dimensional Hilbert space. Explain the basic structure of the proof without working through all 18 vectors.

(c) Contextuality is now understood as a resource for quantum computation (analogous to how entanglement is a resource for Bell violations). Briefly describe the connection.

Part D: Computational Projects (four stars)

These are open-ended projects that require significant programming and analysis.

D.1: Full Bell Test Simulator

Using the code from this chapter as a starting point, build a comprehensive Bell test simulator that:

(a) Allows the user to specify: source state (any two-qubit state), measurement settings (arbitrary axes for Alice and Bob), number of trials, detection efficiency, visibility, dark count rate, and timing jitter.

(b) Generates synthetic data for both quantum and LHV models.

(c) Performs a complete statistical analysis: CHSH computation, error bars, p-value, confidence interval, power analysis.

(d) Produces publication-quality plots: correlation functions vs. angle, CHSH parameter vs. visibility, p-value vs. number of trials, comparison of quantum vs. LHV predictions.

(e) Includes a "loophole dashboard" that reports which loopholes are closed for given experimental parameters.

D.2: QKD Protocol Comparison

Implement BB84, E91, and a simplified version of device-independent QKD. For each:

(a) Simulate the full protocol: state preparation, measurement, sifting, error estimation, key extraction.

(b) Add an eavesdropper with several attack strategies: intercept-resend, optimal individual, and optimal collective.

(c) Compute the secure key rate as a function of channel loss (distance).

(d) Compare the three protocols in terms of: key rate, maximum distance, tolerance to noise, assumptions required, and security guarantees.

(e) Present your results in a "QKD protocol comparison report" with plots and analysis.