Error correction on an array of superconducting qubits with defective components

Research Approach

This paper investigates how to maintain high-performance quantum error correction (QEC) on superconducting processors that contain imperfect, underperforming components. As quantum devices scale, fabrication variability makes it impossible to guarantee that every qubit and coupler will meet ideal performance standards. The authors compare three strategies for managing these defects on a 120-qubit superconducting processor: ignoring the defects, using a noise-informed decoder to account for them, and actively excluding them by modifying the syndrome extraction circuits.

Active Defect Exclusion

The researchers implement active defect exclusion by reconfiguring the surface code layout. When a component is identified as defective, they use 'ancilla repurposing' and 'super-stabilizer' constructions to bypass the faulty hardware. This involves merging neighboring stabilizers into higher-weight checks that allow the system to continue detecting errors without relying on the compromised components. By modifying the circuit geometry, the authors effectively quarantine the defects from the operational code.

The study demonstrates that active defect exclusion provides a substantial performance boost. In memory experiments, excluding defective couplers reduced the logical error per round from 4.49% (using an uninformed decoder) to 1.62%. In contrast, simply informing the decoder about the defects yielded only marginal improvements. Furthermore, the authors show that defect exclusion is critical for measurement-based logic gates; while standard approaches failed to suppress errors in stability experiments, the exclusion strategy successfully restored performance. When combined with leakage post-selection, the distance-5 code was able to outperform the best distance-3 code in specific bases, proving that active management of hardware defects is a viable path toward scalable fault-tolerant quantum computing.

As quantum processors grow in size, the probability of having 'perfect' hardware across the entire array drops significantly. This research provides a practical, proven framework for operating large-scale quantum computers despite inevitable fabrication imperfections. By showing that we can 'work around' faulty components rather than requiring perfect yield, this work lowers the barrier for building reliable, large-scale quantum systems.