Stabilizers and Syndromes

A stabilizer is a parity check. It is a measurement that asks "are these qubits the same or different?" It never asks what value either one holds. The answers from all the checks form a bit pattern called the syndrome. Each syndrome points to exactly one fix. That is how a code can find an error without collapsing the superposition it protects.

How can you check a qubit without reading it?

Error correction seems to demand the impossible. To fix an error you must look for it. But measuring a qubit collapses its superposition — the very thing the code is protecting. A superposition is a state that is a mix of |0⟩ and |1⟩, with weights called amplitudes. Reading the qubit forces it to one value and wipes out the weights.

The way out is to measure only how qubits relate to each other. A parity check on two qubits asks one question: "are you the same or different?" That is the XOR of the two values, and nothing more.

An everyday example: two friends each flip a coin, and you ask only "did you get the same side?" A "yes" tells you nothing about whether it was heads or tails. The coin picture breaks in one way. Coins already landed; you just didn't see them. Qubits in superposition haven't "picked" a value at all, and the check keeps it that way.

In quantum terms, this is measuring the operator Z⊗Z. The symbol ⊗ just means the two single-qubit Zs are bundled into one joint question asked of the pair. The answer comes back as +1 or −1, not as either qubit's own value. It answers +1 for |00⟩ and |11⟩ ("same"). It answers −1 for |01⟩ and |10⟩ ("different").

Here is why that is safe. Take the entangled state α|00⟩ + β|11⟩ (α and β are the amplitudes). Both branches, |00⟩ and |11⟩, give the same answer: "same." Since every branch agrees, the measurement learns nothing about α and β. The superposition survives untouched. The check would only disturb a state whose branches disagree about the answer. And that is exactly what an error creates. Errors get caught. Data passes through unread.