The EPR Paradox — Physica

Equivalent forms

\langle (\sigma_A\cdot\hat a)(\sigma_B\cdot\hat b)\rangle = -\hat a\cdot\hat b

P(\text{same}) = \sin^2\!\tfrac{\theta}{2}

EPR meant to expose a flaw and instead handed physics its sharpest tool: entanglement is not a bug but the resource behind quantum computing and cryptography.

Symbol	Name	SI unit	Dimension	Range
$\|\Psi^-\rangle$	Singlet (Bell) state Maximally entangled, total spin 0	dimensionless	1	—
$\theta$	Relative measurement angle Angle between Alice's and Bob's detector axes	rad	1	0 – 3.14159
$E(\hat a,\hat b)$	Spin correlationoutput Expectation of the product of the two outcomes	dimensionless	1	—

Where it holds

Predictions hold for ideal entangled pairs with spacelike-separated, freely-chosen measurements. Real tests must close detection and locality loopholes. No information travels faster than light — the marginal outcome at each side is perfectly random; only correlations (revealed by later classical comparison) show the effect.

Discovery

Einstein, Podolsky & Rosen · John Bell (resolution) · 1935

In a 1935 paper provocatively titled 'Can Quantum-Mechanical Description of Physical Reality Be Considered Complete?', Einstein, Podolsky and Rosen built a thought experiment they believed forced the answer 'no.' Bohr fired back within months, defending the theory. The debate looked philosophical until 1964, when John Bell — working alone at CERN — derived an inequality that any local-hidden-variable theory must obey but quantum mechanics violates. Alain Aspect (1982) and later loophole-free experiments (2015) measured the violation; the 2022 Nobel Prize honored that work.

Try this

Can measuring a particle here instantly fix the state of its twin light-years away?

Einstein, Podolsky and Rosen argued quantum mechanics must be incomplete because it seemed to permit 'spooky action at a distance.' The maths of entanglement turns their paradox into a precise, testable prediction.

Real-world applications

Quantum key distribution (Ekert protocol) — eavesdropping breaks the correlations
Quantum teleportation and entanglement swapping
Device-independent randomness certification
Entanglement as the fuel of quantum-computing speedups

Common misconceptions

No faster-than-light signaling occurs — each side alone sees pure randomness; the correlation appears only when results are compared classically
The particles do not have predetermined opposite spins along all axes; assuming so is exactly what Bell's theorem rules out
Entanglement is not a 'connection' you can push a signal through — measuring A does not let you control B's outcome

Experimental verification

Aspect's 1982 photon-polarization experiments with fast-switched analyzers violated Bell's inequality. Loophole-free tests (Hensen 2015, NIST, Vienna) closed locality and detection loopholes simultaneously, confirming the quantum prediction to many standard deviations.

Derivation

Take the singlet |

\Psi ^{-}

⟩ = (|↑↓⟩

-

|↓↑⟩

)/\sqrt{2}

.

Measuring A along â and B along b̂, the quantum correlation is E(â,b̂

) =

⟨

\Psi ^{-}|(\sigma \cdot

â

)(\sigma \cdot b

̂

)|\Psi ^{-}

⟩

= -

â

\cdot b

̂

= -\cos \theta

.

The probability the two spins come out opposite-aligned along axes separated

by \theta is \cos ^{2}(\theta /2)

; same-aligned is

\sin ^{2}(\theta /2)

.

A local hidden-variable model, by contrast, is bounded by Bell's inequality |E(a,

b) - E

(a,

b')| + |E

(a'

,

b) + E(a'

,

b')| \le 2

, whereas quantum mechanics reaches

2\sqrt{2}

(Tsirelson's bound) at the optimal angles — a clear, measurable gap.

What if…

What if hidden variables were real (Einstein's view)?

Correlations would obey Bell's inequality $(S \le 2)$ . Experiments measure $S \approx 2.8$ , so local hidden variables are ruled out.

What if you could read B before comparing notes with A?

B alone is 50/50 random — no message arrives. The 'spooky' part only shows up in the joint statistics, preserving no-signaling.

What if the detectors are perpendicular

(\theta = 90^{\circ})

?

$E = -\cos 90^{\circ} = 0$ — the outcomes become completely uncorrelated.

1

Correlation at parallel detectors

Given ·

theta:: 0

Find · E(â, b̂) and probability of opposite outcomes

Steps

E(â,b̂ $) = -\cos \theta = -\cos 0 = -1$
P(opposite $) = \cos ^{2}(\theta /2) = \cos ^{2}0 = 1$ (always opposite)

Result ·

E = -\cos 0 = -1

; P(opposite

) = \cos ^{2}0 = 1

— perfectly anti-correlated, as expected for a singlet.

2

Maximal Bell violation angle

Given · Optimal CHSH angles

(0^{\circ}

,

45^{\circ}

,

22.5^{\circ}

,

67.5^{\circ})

Find · CHSH value S

Steps

Each correlation term $= \cos 45^{\circ} = 1/\sqrt{2}$
$S = |E$ (a, $b) - E$ (a, $b')| + |E$ $(a'$ , $b) + E(a'$ , $b')| = 4\cdot (1/\sqrt{2}) = 2\sqrt{2}$
$2\sqrt{2} \approx 2.83$ exceeds the classical bound of 2

Result ·

S = 2\sqrt{2} \approx 2.83

> 2 — quantum mechanics beats every local hidden-variable theory.

Bell–CHSH Inequality→Spin-1/2 Operators→Born Rule→Density Matrix→Stern–Gerlach Deflection→