Researchers at BTQ Technologies, the University of Sheffield, and Macquarie University have developed a quantum error recovery method tailored for permutation-invariant (PI) codes, offering a potential advantage in noisy environments where traditional stabilizer codes struggle. The team reports demonstrating a recovery algorithm specifically for “collective and local symmetric correlated amplitude-damping (AD) noise,” a challenging type of error where stabilizer codes often require additional overhead. Their work includes a newly proposed “CAD9” code, supported on nine qubits, which outperforms the 7-qubit AAB code by more than one order of magnitude in correcting these global symmetric AD errors. CAD codes correct √N-1 errors. When correcting for global symmetric AD errors at an AD strength of p = 1 × 10-3, the CAD9 code outperforms the 7-qubit AAB code by more than one order of magnitude, and by more than two orders of magnitude at p = 1 × 10-4. For the CAD4 code, which perfectly corrects global symmetric AD error, the compiled recovery circuit consists of 10 system and system-ancilla gates realizable from linear geometric phase gates, providing a direct path from optimized recovery maps to experimentally implementable, low-overhead protocols.
This lessened need for individual qubit control is particularly beneficial in densely packed quantum systems where precise addressing introduces operational noise. Unlike traditional approaches, the new method leverages the unique properties of PI codes, allowing for simpler recovery operation circuits. The CAD codes, including the CAD4 and CAD9 examples, are designed specifically for global symmetric AD errors, with the CAD9 code outperforming the seven-qubit AAB code by more than one order of magnitude at AD strengths of 1 x 10-3 and by more than two orders of magnitude at 1 x 10-4.
Stabilizer codes, the foundation of much current quantum error correction research, are increasingly challenged by amplitude-damping (AD) noise, a spontaneous energy decay where qubits transition from |1⟩ to |0⟩. While effective against Pauli errors, these codes struggle with AD’s non-Pauli nature, often requiring significant overhead to maintain fidelity in environments dominated by this type of decay. Researchers are now focusing on bespoke, hardware-adapted codes to circumvent these limitations, seeking strategies that minimize resource demands. Researchers at BTQ Technologies and the University of Sheffield, along with colleagues at Macquarie University, have introduced a new family of PI codes, termed CAD codes, with examples on 4 and 9 qubits designed for global symmetric AD errors. The CAD9 code outperforms the 7-qubit AAB code by more than one order of magnitude. CAD codes correct √N-1 errors. When correcting for global symmetric AD errors, specifically at an AD strength of p = 1 × 10-3 and exceeding that advantage at p = 1 × 10-4.
This performance boost stems from the ability of PI codes to bypass the need for individual addressability, relying instead on uniform, collective control over the entire qubit register. Researchers at Macquarie University, the University of Sheffield, and BTQ Technologies demonstrated that optimized quantum error recovery (QER) can suppress entanglement infidelity in the logical codespace by several orders of magnitude. The CAD9 code outperforms the 7-qubit AAB code by more than one order of magnitude. CAD codes can correct √N−1 errors, while gnu codes correct ⌊√N−1⌋ errors. When correcting for global symmetric AD errors, specifically at an AD strength of p = 1 × 10-3, the CAD9 code shows an advantage that increases at p = 1 × 10-4. The team has devised a method to translate optimized recovery maps into practical quantum circuits using ancilla-assisted coherent control.
Their work focuses on permutation-invariant (PI) codes, which offer a potential advantage by simplifying control requirements compared to traditional stabilizer codes. They utilize a method to translate optimized recovery maps into practical quantum circuits using ancilla-assisted coherent control, and note that CAD codes on N qubits can correct √N – 1 collective AD errors, while gnu codes correct ⌊√N – 1⌋ errors.
This approach is particularly valuable as physical quantum systems mature and demand hardware-efficient strategies. They’ve developed a four-qubit CAD4 code capable of perfectly correcting global symmetric AD errors, with a recovery circuit compiled from just ten system and system-ancilla gates. These gates can be realized using linear geometric phase gates, simplifying potential hardware implementation.
Researchers are increasingly focused on tailoring quantum error correction to specific hardware limitations, moving beyond generic approaches to bespoke codes and decoding algorithms. A key challenge lies in mitigating amplitude-damping (AD) noise, where qubits decay from |1⟩ to |0⟩, which poses difficulties for traditional stabilizer codes requiring increased resource overhead. Researchers at BTQ Technologies, Macquarie University, and the University of Sheffield benchmarked several short-length PI codes, including a newly proposed family termed CAD codes, against both global and local symmetric AD noise. Their findings reveal that optimized QER can substantially reduce entanglement infidelity within the logical codespace. The CAD9 code outperforms the 7-qubit AAB code by more than one order of magnitude at an AD strength of p = 1 × 10-3 and by more than two orders of magnitude at p = 1 × 10-4.
This improvement is particularly notable when contrasted with the “gnu” code, a competing PI code. While gnu codes with parameters g=n=√N and u=1 on N qubits correct ⌊√N−1⌋ errors, CAD codes can correct √N−1 errors. This work suggests PI codes, particularly the CAD family, could serve as a valuable, efficient layer within larger fault-tolerant quantum computing architectures.
This approach moves beyond simply correcting errors; it focuses on implementing optimized recovery maps directly into experimentally viable protocols. This allows for a streamlined recovery process, scaling linearly with the Kraus rank when considering dominant primitives. Demonstrating the practicality of this method, researchers compiled a recovery circuit for the CAD4 PI code, designed to counteract first-order global symmetric AD noise.
Source: https://arxiv.org/abs/2607.02346
