Part II · Two Qubits — 6a

Bell’s Inequality

Entanglement looks spooky — but how do we know the qubits weren’t just carrying secret pre-agreed answers all along? Bell turned that question into a game with a scoreboard. Any “they agreed in advance” story is capped at a score of 2. Quantum mechanics blows past it.

↩ before you start · keep these handy
·From Ch. 6: a Bell pair (|00⟩+|11⟩)/√2 is entangled — each half is a fair coin, yet the pair always agrees.
·From Ch. 3: you may choose the direction you measure along, and each measurement returns a ±1 outcome.
·From 0.2: cos of an angle slides smoothly between −1 and +1 — that's the range a correlation lives in.
🔑 symbol decoder · every new mark, in plain words
Sthe CHSH score — one number that sums up four correlation measurements. E(a, b)the correlation when Alice uses setting a and Bob uses b: +1 always agree, −1 always disagree, 0 unrelated. a, a′Alice's two possible measurement directions. b, b′ are Bob's two. 2the classical ceiling — the most any "answers fixed in advance" plan can score. 2√2≈ 2.83 — Tsirelson's bound, the absolute quantum maximum. local realismthe everyday assumption that answers were set before measuring and one side can't affect the other.
feel

A game you can’t rig in advance

Alice and Bob are taken to separate rooms — no contact. Each round a referee flips a private coin for each of them and hands over a question. They each answer, instantly, with no way to know what the other was asked. Sometimes the rules reward agreeing, sometimes disagreeing.

Here’s the rub. They can scheme all they like before being separated — agree on any plan, any cheat sheet, any shared dice. Bell proved that every such plan, no matter how clever, wins at most a fixed share of rounds. Cash it out as a single score S, and no pre-arranged strategy can push S above 2. That ceiling is the inequality.

🎓 everyday picture

Two students are sent to separate sealed exam rooms. Beforehand they could memorize any shared cheat-sheet, but once split they can't talk. For a quiz whose questions are flipped at random, there's a hard limit on how often their answers can line up with the marking scheme — no cheat-sheet, however clever, beats it. Bell pinned down that exact limit. Entangled particles, astonishingly, line up more often than any cheat-sheet allows.

recapAny pre-agreed strategy is capped at a score of S = 2 — that ceiling is Bell's inequality.
play

Push the needle past 2

Give Alice & Bob a shared Bell pair instead of a cheat sheet. Alice measures along one of two fixed directions (teal); Bob picks between his two (amber). Drag the dial to rotate Bob’s pair and watch the score S climb — right through the classical ceiling of 2, up to the quantum maximum 2√2.

▸ CHSH dialS = E₁ − E₂ + E₃ + E₄
a a' b b'
{{ t.lbl }}{{ t.val }}
S ={{ sVal }}
2 · classical 2√2
{{ verdict }}
recapA shared Bell pair pushes S right past the classical 2, up to the quantum maximum 2√2.
math

Assembling the score

For each pair of settings, the correlation E runs from −1 (always disagree) to +1 (always agree). The CHSH score stitches four of them together — three added, one subtracted:

S = E(a,b) − E(a,b') + E(a',b) + E(a',b')
local realism (pre-agreed answers):  |S| ≤ 2
quantum, with a Bell pair:  |S| up to 2√2 ≈ 2.83

With Alice fixed at 0° and 90° and Bob’s pair rotated by θ, all four quantum correlations are cosines and the sum collapses to S = 2(sinθ + cosθ). That peaks at θ = 45°, giving exactly 2√2 — the dial’s top reading. No arrangement of cheat sheets can match it.

✎ worked example · the score at the best angle
1.quantum result: S = 2(sinθ + cosθ); set θ = 45°
2.sin 45° = cos 45° = 0.707
3.S = 2(0.707 + 0.707) = 2 × 1.414 = 2.83 = 2√2
4.that's 0.83 over the classical ceiling of 2 — impossible for any cheat sheet.
recapS stitches four correlations together; the quantum cosines peak at S = 2√2 ≈ 2.83.
⚠ common misconception

“Beating 2 means the qubits signal each other faster than light.” No. Alice’s answers, looked at alone, are a perfect 50/50 coin whatever Bob does — there’s no message hidden in them. The violation only shows up when the two distant logbooks are compared afterwards, over an ordinary channel.

What Bell actually kills is subtler and deeper: the comfortable idea that each qubit already carried its answer before being measured. If they had, S could never exceed 2. It does — in the lab — so they didn’t. The correlation is real, and it isn’t a stash of pre-written answers.

✓ you can now
build the CHSH score S out of four correlation measurements
say why every "answers fixed in advance" theory is stuck at |S| ≤ 2
explain why beating 2 rules out local realism yet still sends no signal
← 06 Entanglement next · 6b Mixed States