Quantum Error Correction and Fault Tolerance

Summary

Error correction is the bridge from noisy physical qubits to logical qubits that can run deep algorithms. The key idea: encode information across many physical qubits so that errors can be detected and corrected without learning the encoded quantum state.

Notes

Why classical intuition fails

You can’t just copy an unknown quantum state (no-cloning theorem).
Measuring qubits directly would collapse the state.

So QEC uses syndrome measurements: measure carefully chosen observables that reveal where errors happened, not the logical state itself.

Codes, stabilizers, and syndromes

A quantum code encodes k logical qubits into n physical qubits.
Stabilizer codes describe the code space as the +1 eigenspace of a set of commuting Pauli operators (stabilizers).
Measuring stabilizers yields syndromes (patterns of ±1) that indicate which error likely occurred.

Surface code (very high level)

Qubits arranged on a 2D lattice; stabilizers correspond to “plaquettes” and “stars”.
Pros:
- High threshold error rate (≈1% region, implementation-dependent).
- Local interactions (good for planar hardware).
Cons:
- Large overhead: thousands of physical qubits per high-quality logical qubit for demanding tasks.

Logical qubits and thresholds

A logical qubit is an encoded qubit; logical gates operate on the encoded space.
If physical error rate p is below threshold, then:
- Increasing the code distance (more physical qubits) can make logical error rate p_L fall roughly like (p / p_th)^((d+1)/2).
- This is the core “scaling” property you need for FTQC.

Biased noise and bosonic codes (link to Alice & Bob)

In some devices, one error type (e.g. bit flips) is much rarer than another (e.g. phase flips) → biased noise.
Bosonic codes encode qubits into harmonic oscillators (modes) rather than discrete qubits.
Cat qubits are a type of bosonic code that:
- strongly suppress one error channel (e.g. bit flips)
- allow tailored codes that exploit this bias for lower overhead.

Alice & Bob’s approach fits here: leverage biased noise + bosonic codes + concatenated schemes to reduce the number of physical qubits needed per logical qubit.

References

06-hardware-modalities.md
08-alice-and-bob.md