A simple universal routing strategy for reducing the connectivity requirements of quantum LDPC codes
Guangqi Zhao, Fei Yan, Xiaotong Ni · 2025
Read A simple universal routing strategy for reducing the connectivity requirements of quantum LDPC codes by Guangqi Zhao, Fei Yan, Xiaotong Ni (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-00850.txt.
Introduction
Guangqi Zhao, with two co-authors, Fei Yan and Xiaotong Ni, presents a universal routing strategy to reduce connectivity needs of quantum low-density parity-check codes. These codes cut error correction overhead but demand dense, long-range connections that challenge hardware, especially superconducting processors.
Zhao and his co-authors observe that ancilla qubits, helper qubits, form short loops with data qubits in any quantum code. This lets stabilizer measurement circuits route data qubit information through ancillas when direct connections are unavailable. For bivariate bicycle codes, they remove up to fifty percent of long-range connections while roughly doubling circuit depth, with circuit-level distance largely preserved. The method also applies to surface codes, matching a known hexagonal connectivity requirement.
Catalog record
| CREATOR | Guangqi Zhao, Fei Yan, Xiaotong Ni |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-00850
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A simple universal routing strategy for reducing the connectivity requirements of quantum LDPC codes Guangqi Zhao, Fei Yan, Xiaotong Ni Abstract Quantum low-density parity-check codes reduce quantum error correction overhead but require dense, long-range connectivity that challenges hardware implementation, particularly for superconducting processors. We address this problem by demonstrating that long-range connections can be reduced at the cost of increased syndrome extraction circuit depth. Our approach is based on the observation that X and Z ancilla qubits form short loops with data qubits—a property that holds for any quantum code. This enables implementing stabilizer measurement circuits by routing data qubit information through ancilla qubits when direct connections are unavailable. For bivariate bicycle codes, we remove up to 50% of long-range connections while approximately doubling the circuit depth, with the circuit-level distance remaining largely preserved. This method can also be applied to surface codes, achieving the same hexagonal connectivity requirement as McEwen et al. McEwen et al. . Our routing approach for designing syndrome extraction circuits is applicable to diverse quantum codes, offering a practical pathway toward their implementation on hardware with connectivity constraints. I Introduction While quantum bits provide significant computational advantages over classical bits, they are inherently more susceptible to errors. Quantum error correction (QEC) is therefore essential for executing long quantum circuits and algorithms. Among various QEC schemes, the surface code Kitaev 2003 ; Bravyi and Kitaev 1998 has emerged as particularly promising due to its local connectivity requirements, with recent experiments demonstrating sub-threshold error rates on superconducting platforms Acharya et al. 2025 . However, the surface code’s very low rate imposes steep resource overheads. This has motivated interest in quantum LDPC (qLDPC) codes with higher encoding rates Freedman et al. 2002 ; Evra et al. 0 ; Kaufman and Tessler 2021 ; Hastings et al. 2021 ; Panteleev and Kalachev 2022a ; Breuckmann and Eberhardt 2021 ; Panteleev and Kalachev 2022b ; Leverrier and Zemor 2022 ; Dinur et al. 2023 . Significant effort has gone into constructing and optimizing qLDPC codes for experimental systems Tremblay et al. 2022 ; Bravyi et al. 2024 ; Xu et al. 2023 ; Liang et al. 2025 ; Steffan et al. 2025 ; Shaw and Terhal 2025 ; Gehér et al. 2025 . This will be a long-lasting but worthwhile process, as current platforms have yet to achieve low logical error rates even with the surface code. A central objective is to reduce the number of required long-range operations. In mobile platforms such as neutral atoms, shuttling time can dominate the cost of fault-tolerant computation. In solid-state platforms such as superconducting processors, each direct long-range gate requires a physical coupler Wang et al. 2025 ; Song et al. 2025 ; Magnard et al. 2020 ; Campagne-Ibarcq et al. 2018 ; Yan et al. 2022 ; Renger et al. 2025 . Such gates are typically slower, and increased connectivity degrades fidelity and amplifies crosstalk (see Appendix VII.2 ). Bivariate bicycle (BB) codes Bravyi et al. 2024 are a recently introduced family of qLDPC codes that combine strong performance with a pseudo-2D layout and commonly use weight-6 stabilizers. In the standard “toric” layout, each measurement qubit typically performs four local and two long-range gates to data qubits Bravyi et al. 2024 . To further reduce connectivity, “morphing circuits” Shaw and Terhal 2025 lower the local degree from four to three but still require two long-range couplers, leaving the main hardware bottleneck. In this work, we introduce a routing-based syndrome measurement approach that systematically reduces long-range connectivity for BB codes. Our method removes up to 50% of long-range connections while largely preserving circuit-level distance, at the cost of roughly doubling circuit depth. Concretely, the BB requirement can be reduced to four local connections and one long-range connection per qubit. The same idea transfers to the surface code, achieving degree-3 connectivity with performance comparable to mid-cycle circuits McEwen et al. . These reductions can simplify recent qLDPC demonstrations Wang et al. 2025 and enable less demanding multilayer layouts Mathews et al. 2025 . This manuscript is organized as follows. In Sec. II , we discuss background information, including the structure of surface and BB codes, as well as previous work on reducing connectivity requirements in surface codes. Then, in Sec. III , we demonstrate our routing approach for the surface code with reduced connectivity, specifically showing how to perform syndrome measurements for the surface code on a hexagonal lattice via routing. For performance evaluation, we conduct circuit-level simulations using Stim Gidney 2021 with a detector error model Gidney 2021 ; Higgott and Gidney 2025 that fully captures error mechanisms at the circuit level (See Appendix VII.3 for a short introduction to the detector error model). Results show that our circuit preserves the code distance, meaning the circuit-level distance matches the code distance. By applying this routing strategy, we demonstrate that up to 50% of long-range couplers can be eliminated while largely maintaining circuit-level distances. Furthermore, the logical error rate of our circuit is comparable to that of conventional circuits, despite the reduced connectivity requirements. The code used to conduct the simulation is available in our code repository on GitHub git . II Background II.1 Surface code Surface codes Bravyi and Kitaev 1998 ; Freedman and Meyer 2001 represent the most developed quantum error correction approach and potentially the most promising codes realizable in the near term Acharya et al. 2025 ; Acharya et al. 2023 , due to their local two-dimensional structure. For surface codes on an L\times L square lattice, characterized as [[n,k,d]]=[[L^{2},1,L]] quantum codes, they saturate the theoretical trade-off bound for two-dimensional codes where kd^{2}\leq O(n) Bravyi et al. 2010 . Fig. 1 shows the structure of a rotated surface code. In the bulk, each ancilla qubit connects to four data qubits, and each data qubit connects to four ancilla qubits, resulting in 4-fold local connectivity in a 2D layout. Conventionally, syndrome extraction circuits for surface codes are implemented by performing CNOT gates across these connections. Figure 1: 2D square lattice layout of rotated surface code. Although the conventional circuits for X and Z stabilizer measurements are well-known, for clear comparison with our new circuit discussed in the following sections, we illustrate them in Fig. 2 along with the backward transformation of the measured Z stabilizers. We can see that we are indeed measuring the code stabilizer operators when no error occurs in the circuit. This is known as stabilizer flows, as described in McEwen et al. . Through careful arrangement of operations, the circuit depth of surface codes can be optimized to just 4 layers of CNOT or CZ gates. This circuit has demonstrated impressive logical performance in experiments Krinner et al. 2022 ; Acharya et al. 2023 ; Acharya et al. 2025 ; Zhao et al. 2022 ; Bluvstein et al. 2025 , and it can be efficiently decoded through matching algorithms Dennis et al. 2002 ; Fowler 2013 ; Higgott 2022 . (a) Z stabilizer measurement circuit (b) X stabilizer measurement circuit Figure 2: Conventional syndrome extraction circuit for surface code. (a) shows the circuit for Z stabilizer measurement, while (b) shows the X -type stabilizer measurement. Blue and red lines indicate the flows of stabilizer through the CNOTs.