Good Quantum LDPC Codes
Good quantum LDPC codes achieve constant rate and distance proportional to block length at once, the asymptotically optimal scaling long thought hard to reach.
What good means
A family of codes is asymptotically good if, as the block length n grows, both the rate k/n and the relative distance d/n stay bounded below by positive constants. Classical good LDPC codes have existed for decades. Whether quantum good LDPC codes existed was open for years, because the CSS commutation constraint makes it hard to keep distance large while keeping checks sparse.
The resolution
A sequence of constructions in the early 2020s settled the question affirmatively. Building on balanced and lifted products of classical codes with expansion properties, these families provably achieve constant rate and linear distance with bounded-weight stabilizers. The proofs use expander-graph arguments to lower-bound the distance.
- Constant rate: the number of logical qubits is a fixed fraction of physical qubits.
- Linear distance: d grows proportionally to n, not merely as its square root.
- Bounded check weight: every stabilizer stays low weight at all sizes.
- Existence is proven; efficient decoders and hardware layouts are still being developed.
The significance is that, in principle, the overhead of fault tolerance can be made a constant multiplier rather than a growing one. This is a sharp contrast with the surface code, whose overhead per logical qubit grows with the target logical error rate.
Two gaps separate theory from machines. First, decoding these codes efficiently and accurately is harder than matching, so belief propagation with added heuristics is the leading approach. Second, the required qubit connectivity is non-local, which suits reconfigurable-atom and modular architectures more than a fixed planar grid. Closing both gaps is where much current effort is focused.