Quantum hardware makers must improve physical error rates and scale their systems to accommodate larger numbers of qubits—tasks that pose significant challenges. By using soft information, we have a powerful tool to ultimately help them reach desirable logical error rates more effectively.
A new arXiv paper, "Reducing Quantum Error Correction Overhead Using Soft Information," demonstrates that soft information decoding enables low logical error rates while utilising fewer qubits than previously required.
The findings indicate that soft information significantly improves error suppression for both superconducting qubits and neutral atom qubits in various Quantum Error Correction (QEC) codes, including surface codes and bivariate bicycle (BB) codes.
Surface codes are like a grid designed to efficiently manage localised errors, while bivariate bicycle codes use a more complex network structure to tackle different error scenarios. Both aim to improve the reliability of quantum computing but do so in distinct ways.
Soft information, essentially, gives the decoder a better real-time picture of the errors that may have happened. This allows the decoder to make more accurate predictions of the system’s state, enhancing its ability to suppress errors.
In our research, we simulated noisy QEC circuits for surface codes and BB codes, employing a realistic physics-based measurement model. By implementing a range of decoders, including our Local Clustering Decoder (LCD) and variants of belief propagation, we assessed the error suppression capabilities.
The improvements were notable: soft information increased error suppression rates by 10% for superconducting qubits and 20% for neutral atom qubits. Moreover, when applied to BB codes, soft information yielded an order-of-magnitude reduction in logical error probability.
What's particularly exciting is that our work marks the first demonstration of soft information's effectiveness on neutral atoms and its benefits for decoding BB codes.
This technique not only enhances QEC now but also shows promise for future scalability. By leveraging soft information—a method grounded in the physical properties of qubits—we unlock greater potential for efficient quantum computing.
For more insights, check out the full paper here.