Description
Quantum error-correcting codes allow for fault-tolerant quantum computation with noisy hardware, but they incur an overhead: multiple noisy qubits are required to encode and manipulate quantum information robustly. One way to reduce this overhead is to use biased-noise qubits, with structured noise channels dominated by a single error type, such as dephasing or erasure. These errors are much easier to correct, which lowers the number of qubits required to build error-correcting codes with algorithmically-relevant error rates. However, biased-noise qubits also come with additional hardware requirements: a qubit biased toward dephasing benefits from a control scheme that preserves this bias, while an erasure qubit benefits from mid-circuit erasure detection. Both are experimentally challenging to implement. Here, we analyse the overhead of biased-noise error-correcting codes in practical settings where the full suite of bias-preserving tools may not be available, and find that appreciable overhead reduction persists even when bias-preserving gates or mid-circuit erasure detection is unavailable.
| I am the presenting author | Yes |
|---|