QCQI – Chapter 9 Detailed Summary: Quantum Error Correction, Stabilizers, and Noise

Course Notes (MSU)

Learning Targets¶

• Model noise with Pauli channels and Kraus operators; use the Pauli transfer picture.

• State and apply Knill–Laflamme (K–L) error-correction conditions.

• Build and analyze stabilizer codes: generators, syndromes, degeneracy, and decoding.

• Implement 3-qubit, phase-flip, and Shor 9-qubit ideas; understand CSS/Steane intuition.

• Glimpse fault tolerance: transversal gates, cat-state extraction, and the threshold idea.

Noise Models & Error Bases¶

A single-qubit CPTP map can be expanded in the Pauli basis: $\mathcal{E}(\rho)=\sum_{P\in\{I,X,Y,Z\}} p_P\,P\rho P$ for a Pauli channel, or via Kraus operators $\{K_\ell\}$ as $\mathcal{E}(\rho)=\sum_\ell K_\ell \rho K_\ell^\dagger$.

Eureka! Because $\{I,X,Y,Z\}$ form an operator basis, correcting all single-qubit Paulis implies correction of arbitrary small errors (by linearity).

Examples¶

Depolarizing: $(1-p)\rho + \tfrac{p}{3}(X\rho X + Y\rho Y + Z\rho Z)$. Phase Damping: dephases off-diagonals in $Z$-basis. Amplitude Damping: $E_0=\begin{pmatrix}1&0\\0&\sqrt{1-\gamma}\end{pmatrix}$, $E_1=\begin{pmatrix}0&\sqrt{\gamma}\\0&0\end{pmatrix}$.

Knill–Laflamme Conditions¶

A code subspace $\mathcal{C}=\mathrm{span}\{|\bar{i}\rangle\}$ corrects $\{E_a\}$ iff

Intuition: error pairs act like a scalar on the code space—no leakage about the logical basis labels.

Eureka! “All damage $\to$ syndrome”: the environment learns the error type but not the logical state.

Distance, Detection, and Degeneracy¶

The distance $d$ is the minimum weight of a Pauli that maps one codeword to another (up to stabilizers). A code detects up to $d-1$ errors and corrects up to $t=\lfloor(d-1)/2\rfloor$ errors. A degenerate code has distinct errors sharing the same syndrome (coset leaders differ by stabilizers).

Stabilizer Formalism¶

Let $S\subset\mathcal{P}_n$ be an abelian subgroup with $-I\notin S$. The code space is the +1 eigenspace of $S$. If $|S|=2^{n-k}$ then we have an $[[n,k]]$ code; measuring the $r=n-k$ generators gives a syndrome $s\in\{0,1\}^r$.

Eureka! Commutation $\Rightarrow$ simultaneous measurability: we can extract $r$ classical bits without collapsing the logical superposition.

Syndrome Extraction (Concept)¶

Canonical Small Codes¶

Repetition (bit-flip) $[[3,1,1]]$¶

$|\bar{0}\rangle=|000\rangle$, $|\bar{1}\rangle=|111\rangle$, stabilizers $Z_1Z_2$, $Z_2Z_3$. A single $X$ error flips one parity; majority vote recovers the bit.

Phase-Flip via Hadamards¶

Wrap the above with $H^{\otimes 3}$ to protect against $Z$ errors.

Shor 9-Qubit $[[9,1,3]]$ (Idea)¶

Concatenate phase-protection within triples and bit-flip protection across triples to correct any single-qubit $X,Y,Z$ error.

Eureka! Protect $Z$ and $X$ errors separately in complementary bases; $Y$ follows since $Y=iXZ$.

CSS and Steane Intuition¶

Given $C_2\subset C_1$ classical linear codes, CSS builds an $[[n,k]]$ code with $X$-checks from $C_2^\perp$ and $Z$-checks from $C_1$. The Steane $[[7,1,3]]$ uses Hamming $(7,4,3)$ for both parts.

Fault-Tolerance (Glimpse)¶

Transversal gates prevent a single fault from fanning out. Cat-state ancillas limit hook errors during extraction. Threshold theorem: below a hardware-dependent error rate, scalable computation is possible with polylog overhead.

Pseudo-code Aids¶

Pseudo-code: 3-qubit bit-flip: encode $\to$ syndrome $\to$ correct

Encode: CNOT($q_0\to q_1$); CNOT($q_0\to q_2$) Syndrome: measure $Z_1Z_2$ and $Z_2Z_3$ via ancillas Decode: map syndrome $(s_{12},s_{23})\in\{10,01,11\}$ to $X$ on $\{q_3,q_1,q_2\}$ respectively Recover: majority vote or inverse of encode

Pseudo-code: General stabilizer check measurement

Prepare ancilla in $|0\rangle$ (for $Z$-type) or $|+\rangle$ (for $X$-type) For each $Z$ on data: CNOT(data$\to$anc) For each $X$ on data: CNOT(anc$\to$data) or basis-change $H$ tricks Measure ancilla in $Z$; the bit is the generator eigenvalue

Pseudo-code: Knill–Laflamme numeric test

Compute projector $P$ onto code space; for errors $\{E_a\}$ compute $G_{ab}=P E_a^\dagger E_b P$ Verify $G_{ab}\propto I$ in the code basis

Hands-on Notebook (Multi-Backend)¶

Run: L13_QCQI_Ch09_Quantum_Error_Correction_and_Noise

Set backend in the first cell to cirq, pennylane, braket, pyquil, or qiskit.

Mini-Labs¶

1. Monte Carlo logical error vs. physical error $p$ for the 3-qubit code; compare to $3p^2-2p^3$.

2. Convert to phase-flip protection and test against $Z$-noise.

3. (Stretch) Sketch a Shor encoder/decoder; correct a single arbitrary Pauli on one qubit using an ideal simulator.