Glossary / Quantum computing
Decoding
The calculation that takes the measurements from error correction and works out which qubits suffered which errors.
The helper qubits in error correction only report clues that something is wrong, not where the error is. Decoding turns the pattern of clues into the most likely set of errors. The method Dennis and colleagues set out in 2002 — treating it as a problem of pairing up clues along the shortest paths — is still the basic approach[1].
The hard part is speed. In superconducting machines a new round of clues arrives in less than a microsecond, and if decoding falls behind the backlog grows until the computation fails. That is why dedicated chips such as FPGAs are used to get decoding down to a few hundred nanoseconds.
Related terms
- Quantum error correctionSpreading the information of fragile qubits across many of them, so that errors can be detected and fixed.
- Quantum LDPC codeA family of error-correcting codes that protect many logical qubits with few physical ones; each check looks at only a handful of qubits.
- Fault-tolerant quantum computingRunning quantum computations so that, even though components fail at some rate, layered error correction lets arbitrarily long calculations finish correctly.