Efficient Near-Optimal Decoding of the Surface Code through Ensembling

Area of Science:

• Quantum Information Science
• Computer Science

Background:

• Accurate decoding is crucial for fault-tolerant quantum computation.
• Existing decoders often struggle with high logical error rates on complex quantum error-correcting codes.
• Ensembling methods offer potential improvements but can be computationally expensive.

Purpose of the Study:

• To introduce and evaluate a novel ensembling method called harmonization for improving decoding accuracy.
• To assess the effectiveness of harmonization on repetition and surface codes.
• To develop a confidence measure for decoding based on ensemble consensus.

Main Methods:

• Harmonization combines multiple "noisy" decoders to produce more accurate predictions.
• Ensembles of minimum weight perfect matching-based decoders were benchmarked.
• A layered decoding scheme was proposed, utilizing ensemble consensus as a confidence measure.

Main Results:

• Harmonized ensembles achieved lower logical error rates compared to individual decoders on repetition and surface codes.
• Accuracy approached maximum-likelihood levels with increasing ensemble sizes.
• The layered scheme demonstrated significant accuracy gains with manageable computational overhead.

Conclusions:

• Harmonization is a viable ensembling technique for enhancing decoder accuracy in quantum error correction.
• The method provides a practical approach to achieving high-fidelity, real-time decoding.
• Ensemble consensus serves as an effective confidence metric for adaptive decoding strategies.