[Jump to code hierarchy]

Bivariate bicycle (BB) code [1]

Description

Two-block CSS code on \(n=2\ell m\) qubits with check matrices \(H_X=[A|B]\) and \(H_Z=[B^T|A^T]\) [1; Sec. 4]. Here, \(A\) and \(B\) are polynomials in two commuting cyclic shifts \(x\) and \(y\) of orders \(\ell\) and \(m\). As in the toric code, all checks are shifts of one \(X\)-check and one \(Z\)-check on a two-dimensional periodic grid [1; Sec. 3].

The qubit connectivity graph is not quite a 2D grid and is instead decomposable into two planar subgraphs of degree three; there exists an optimized layout minimizing Euclidean communication distance for check operators [2]. There are \(n/2\) \(X\)-type and \(n/2\) \(Z\)-type check operators, each of weight \(|A|+|B|\), the number of terms in \(A\) and \(B\) [1; Sec. 4]. Codes can be classified by the weight of their checks, e.g., by BB\(w\) where \(w\) is the check weight. The weight-six BB codes of Ref. [1] are on par with the surface code in terms of threshold and admit a much higher ancilla-added encoding rate. The price is check operators that are not geometrically local in a two-layer nearest-neighbor layout [1].

Multivariate (e.g., trivariate) bicycle codes generalize BB codes to Abelian 2BGA codes with polynomials in more than two variables [3]; other variants and generalizations exist [4]. Versions with open boundary conditions are tile codes [5], obtained by pruning [6] or by condensing boundary anyons [7,8]. There exist qudit BB codes that achieve \(kd^2 / n = 20\) [9].

BB codes as lattice codes on twisted tori

A BB code with polynomials \(A\) and \(B\) is the quotient by a torus of a 2D translation-invariant CSS code with two qubits per unit cell [10]. In the polynomial representation, the stabilizer generators of that code are \((A,B|0,0)\) and \((0,0|\overline{B},\overline{A})\) and their translations [10]. The bar denotes the antipode map \(x^iy^j\mapsto x^{-i}y^{-j}\). The torus is specified by two lattice vectors \((0,\alpha)\) and \((\beta,\gamma)\), which impose \(y^{\alpha}=1\) and \(x^{\beta}y^{\gamma}=1\) on \(n=2\alpha\beta\) qubits [10]. Every torus can be brought to this form by a unimodular change of basis, and the case \(\gamma\neq 0\) is called a twisted torus [10; Sec. II.B]. The polynomials of the gross code on the twisted torus with lattice vectors \((0,30)\) and \((6,6)\) yield a \([[360,12,d\leq 24]]\) code whose checks have range three [10; Sec. III.B and Table IV].

A BB code satisfies the topological order condition iff \(A\) and \(B\) are coprime as Laurent polynomials [10,11]. Its \(m\)-type anyons are then the elements of the quotient ring \(\mathbb{F}_2[x^{\pm1},y^{\pm1}]/\langle A,B\rangle\) [10]. The number of logical qubits on any torus is at most \(k_{\max}=2\dim\mathbb{F}_2[x^{\pm1},y^{\pm1}]/\langle A,B\rangle\), twice the number of independent \(m\)-type anyons [10; Thm. 1]. On the torus with lattice vectors \((0,\alpha)\) and \((\beta,\gamma)\), the number of logical qubits is [10; Thm. 3] \begin{align} k=2\dim\frac{\mathbb{F}_2[x^{\pm1},y^{\pm1}]}{\langle A,B,y^{\alpha}-1,x^{\beta}y^{\gamma}-1\rangle}~.\tag*{$(1)$} \end{align}The maximum \(k=k_{\max}\) is attained iff \(y^{\alpha}-1\) and \(x^{\beta}y^{\gamma}-1\) lie in the ideal \(\langle A,B\rangle\) [10; Corr. 2.2]. Otherwise, \(k<k_{\max}\) [10,12].

Weight-six BB codes whose polynomials \(A=1+x+x^ay^b\) and \(B=1+y+x^cy^d\) extend the toric-code pair \(1+x\), \(1+y\) by one monomial each include the checks of the 6.6.6 color code and of the gross code [10; Exams. 2 and 3].

Protection

A BB code with \(A=B\) and \(k>0\) has distance two regardless of check weight [13; Thm. 1].

Rate

When ancilla qubit overhead is included, the encoding rate of the weight-six codes of Ref. [1] surpasses that of the surface code. A general \([[n,k,d]]\) bivariate bicycle code requires \(n\) ancilla qubits for encoding, meaning that its ancilla-added encoding rate is \(k/2n\). BB codes have zero design rate, so all of their logical qubits come from dependent stabilizer generators [14].

Transversal and Permutation-Based Gates

Logical Pauli operators and fold-transversal gates studied in Ref. [15,16].

Gates

Certain bivariate bicycle codes admit a cup product structure and can thus have logical gates in the Clifford hierarchy implemented by constant-depth Clifford circuits [17].

Decoding

BP-OSD decoder [18] has been extended [1] to account for measurement errors (i.e., the circuit-based noise model [19]).Decoding under circuit-level noise has been studied for the BP, BP+OSD, and AutDEC decoders [20].Transformer-based neural-network decoder [21].Matching decoder [22].Small-size and intermediate-size matching decoders adapt the cell-matching decoder for 2D lattice stabilizer codes to BB codes [23; Sec. VI].

Fault Tolerance

Fault-tolerant state initialization using lattice surgery techniques [24,25] and an ancillary surface code [1].Syndrome extraction circuit requires seven layers of CNOT gates regardless of code length [1].The depth-7 syndrome extraction schedules studied in Ref. [1] are not generally distance-preserving; for the \([[144,12,12]]\) code, all \(936\) depth-7 variants obtained by reordering the CNOT layers satisfy \(d_{\mathrm{circ}}\leq 10 < d\).Some long-range check operators can be measured less frequently than others [26].Syndrome extraction circuits called morphing circuits [27], generalizing circuits for the color code [28].

Code Capacity Threshold

Fully postselected failure rates of weight-six BB codes under bit-flip noise cross near \(p=1/(2+\sqrt{2})\approx 0.2929\) [29]. This value is the self-dual critical point of zero-rate PSD code families with a unique threshold [29]. Without postselection, the \(k=12\) weight-six family of Ref. [30] has a threshold of \(8.38(6)\%\) under BP-OSD decoding [29].

Threshold

Admits a \(0.8\%\) pseudo-threshold for circuit-level noise under BP-OSD decoder [1] (cf. [19]).

Notes

A database of bivariate bicycle codes is available in QECDB [31].A catalogue of BB codes found by LLM-guided search is available at qcode-discovery [13].See Ref. [10; Tables I-IV] for weight-six BB codes of this form on twisted tori with \(n\leq 400\), and Ref. [11; Tables I and II] for weight-eight self-dual BB codes with \(n\leq 200\).See Ref. [32] for further self-dual BB codes.

Cousins

  • Honeycomb (6.6.6) color code— Certain bivariate bicycle codes are equivalent to a family of 6.6.6 color codes [6; Exam. 1]. The 6.6.6 color code on a torus is the self-dual BB code with \(A=1+x+y\) [11; Exam. 2].
  • Self-dual CSS code— BB codes with \(B=\overline{A}\), where the bar denotes the antipode map \(x^iy^j\mapsto x^{-i}y^{-j}\), have \(H_X=H_Z\) and are self-dual CSS codes [11]. Weight-six ones, \(A=1+x^ay^b+x^cy^d\), have \(k_{\max}=4|ad-bc|\) and reduce to \(|ad-bc|\) decoupled copies of the 6.6.6 color code [11; Thm. 1 and Exam. 2]. Weight-eight ones are doubly even, and \(A=1+x+y+y^{-1}\) yields the \([[16,4,4]]\) and \([[64,8,8]]\) codes [11; Table I].
  • Abelian topological code— BB codes have been investigated in terms of their anyons and topological order [10,12]. A BB code satisfies the topological order condition iff its polynomials are coprime [10]. Its \(m\)-type anyons are then the elements of \(\mathbb{F}_2[x^{\pm1},y^{\pm1}]/\langle A,B\rangle\), and the number of logical qubits on any torus is at most twice the dimension of this ring [10; Thm. 1].
  • Generalized bicycle (GB) code— GB codes (BB codes) are 2BGA codes over the cyclic group \(\mathbb{Z}_{\ell}\) (Abelian group \(\mathbb{Z}_{r} \times \mathbb{Z}_{s}\)). The two codes are the same when \(r\) and \(s\) are relatively prime due to the isomorphism \(\mathbb{Z}_{r} \times \mathbb{Z}_{s} \cong \mathbb{Z}_{\ell = rs}\). BB codes on twisted tori with lattice vectors \((0,\ell)\) and \((1,\gamma)\) reduce to GB codes upon substituting \(x=y^{-\gamma}\) [10; Sec. III.C].
  • \([[27,3,3]]\) bivariate tricycle code— The \([[18,2,3]]\) BB code with \(a=x^2y+x^2y^2\) and \(b=1+xy^2\) is the 2D component code of the \([[27,3,3]]\) bivariate tricycle code [33].
  • \([[81,3,5]]\) bivariate tricycle code— The \([[54,2,6]]\) BB code with \(a=xy^3+x^2y\) and \(b=1+xy^8\) is the 2D component code of the \([[81,3,5]]\) bivariate tricycle code [33].
  • Tile quantum code— Tile codes are open-boundary (planar) analogues of bivariate bicycle (BB) codes, which arise by tiling a torus with stabilizer tiles [5]. Tile codes instead have open boundaries and thus retain true \(O(1)\)-locality on a planar 2D lattice [5]. Sharing the same bulk stabilizer tiles does not yield the same code. The planar \([[288,8,12]]\) tile code has the same bulk stabilizers as the \((-2,2)\)-BB code. That BB code requires a \(217\times 217\) torus to attain its maximal logical dimension \(k=16\) and yields a \([[98,6,12]]\) code on a \(7\times 7\) torus [8]. The logical dimension of a BB code is determined by the period of the same cellular automaton that generates tile-code logical operators [34].

Primary Hierarchy

Parents
Bivariate bicycle codes are Abelian 2BGA (equivalently, two-variable multivariate bicycle) codes over groups of the form \(\mathbb{Z}_{r} \times \mathbb{Z}_{s}\).
Bivariate bicycle codes are permutationally self-dual. Their standard \(XZ\)-duality exchanges the two blocks of qubits while sending \(x\mapsto x^{-1}\) and \(y\mapsto y^{-1}\). This involution yields a Hadamard-type fold-transversal gate [15; Secs. 3 and 4.3]. BB codes on \(\mathbb{Z}_{\ell}\times\mathbb{Z}_{\ell}\) whose two polynomials are exchanged by swapping \(x\) and \(y\) admit a second \(XZ\)-duality. It sends \(x\mapsto y^{-1}\) and \(y\mapsto x^{-1}\) within each block and yields a phase-type fold-transversal gate [15; Sec. 4.3].
BB codes are mirror codes up to qubit permutations and Hadamard gates [35; Prop. 3.21].
A BB code with polynomials \(A\) and \(B\) is the quotient by a torus \(\mathbb{Z}^2/\Lambda\) of a 2D translation-invariant CSS code with two qubits per unit cell [10,11]. The generators of that code are \((A,B|0,0)\) and \((0,0|\overline{B},\overline{A})\), with range given by the degrees of the Laurent representatives of \(A\) and \(B\) [10]. Holding the polynomials fixed while enlarging the torus, including twisted tori, yields a 2D lattice stabilizer code family with checks of bounded range [7,10].
Bivariate bicycle (BB) code
Children
The \([[16,4,4]]\) code is the weight-eight self-dual bivariate bicycle (BB) code with \(A=1+x+y+y^{-1}\) and \(B=\overline{A}\) on the twisted torus with lattice vectors \((0,4)\) and \((2,2)\) [11; Table I].
The \([[18,4,4]]\) color code is simultaneously the bivariate bicycle code with \(\ell=m=3\), \(A=x+1+y^2\) and \(B=y+1+x^2\) [36], i.e., an Abelian 2BGA code over \(\mathbb{Z}_3\times\mathbb{Z}_3\). It is equivalently the \(\ell=m=3\) member of the bivariate bicycle family \(A=1+x+xy\), \(B=1+y+xy\), which yields 6.6.6 color codes whenever \(\ell\) and \(m\) are multiples of three [6; Exam. 1]. It is also the weight-six self-dual BB code with \(A=1+x+y\) on that torus [11; Exam. 2].
The \([[6,4,2]]\) error-detecting code is the weight-six self-dual BB code with the 6.6.6 color-code polynomial \(A=1+x+y\) on the twisted torus with lattice vectors \((3,0)\) and \((1,1)\) [11; Exam. 2].
The three-quarters-gross code is the BB code with \(A=1+x+x^{-1}y^3\) and \(B=1+y+x^3y^{-1}\) on the torus with lattice vectors \((0,9)\) and \((6,0)\) [10; Table I].
The half-gross code places the checks of the gross code on the \(6\times 6\) torus [1; Table 3].
Directional codes form a subfamily of BB codes. Writing the support of an \(X\)-type stabilizer generator on horizontal and vertical edges as bivariate polynomials \(f\) and \(g\), the parity-check matrices are \(H_X=[B(f)\mid B(g)]\) and \(H_Z=[B(\overline{g})\mid B(\overline{f})]\), where \(B\) denotes the binary representation of the quotient ring and the bar denotes the reciprocal polynomial [37]. Every stabilizer generator then has weight \(w\), and every data qubit participates in exactly \(w\) checks, so the Tanner graph is \((w,w)\)-regular [37]. Directional codes are 2D lattice stabilizer codes whose generators are supported on translates of a fixed bounded string of the square lattice [38].
The gross code is the BB code with \(A=1+x+x^{-1}y^3\) and \(B=1+y+x^3y^{-1}\) on the torus with lattice vectors \((0,12)\) and \((6,0)\) [10; Exam. 3 and Table II].
The toric code on any torus \(\mathbb{Z}^2/\Lambda\), including twisted tori, is the BB code with \(A=1+x\) and \(B=1+y\) [10; Exam. 1 and Appx. B].

References

  1. [1]

    S. Bravyi, A. W. Cross, J. M. Gambetta, D. Maslov, P. Rall, and T. J. Yoder, “High-threshold and low-overhead fault-tolerant quantum memory”, Nature 627, 778 (2024) arXiv:2308.07915 DOI

  2. [2]

    C. Poole, T. M. Graham, M. A. Perlin, M. Otten, and M. Saffman, “Architecture for fast implementation of quantum low-density parity-check codes with optimized Rydberg gates”, Physical Review A 111, (2025) arXiv:2404.18809 DOI

  3. [3]

    L. Voss, S. J. Xian, T. Haug, and K. Bharti, “Multivariate Bicycle Codes”, (2025) arXiv:2406.19151

  4. [4]

    M. Wang and F. Mueller, “Coprime Bivariate Bicycle Codes and Their Layouts on Cold Atoms”, Quantum 10, 2009 (2026) arXiv:2408.10001 DOI

  5. [5]

    V. Steffan, S. H. Choe, N. P. Breuckmann, F. R. F. Pereira, and J. N. Eberhardt, “Tile Codes: High-Efficiency Quantum Codes on a Lattice with Boundary”, (2025) arXiv:2504.09171

  6. [6]

    J. N. Eberhardt, F. R. F. Pereira, and V. Steffan, “Pruning qLDPC codes: Towards bivariate bicycle codes with open boundary conditions”, (2024) arXiv:2412.04181

  7. [7]

    Z. Liang, B. Yang, J. T. Iosue, and Y.-A. Chen, “Operator algebra and algorithmic construction of boundaries and defects in (2+1)D topological Pauli stabilizer codes”, (2026) arXiv:2410.11942

  8. [8]

    Z. Liang, J. N. Eberhardt, and Y.-A. Chen, “Planar quantum low-density parity-check codes with open boundaries”, (2025) arXiv:2504.08887

  9. [9]

    Z. Liang and Y.-A. Chen, “Generalized \(\mathbb{Z}_p\) toric codes as qudit low-density parity-check codes”, (2026) arXiv:2602.20158

  10. [10]

    Z. Liang, K. Liu, H. Song, and Y.-A. Chen, “Generalized Toric Codes on Twisted Tori for Quantum Error Correction”, PRX Quantum 6, (2025) arXiv:2503.03827 DOI

  11. [11]

    Z. Liang and Y.-A. Chen, “Self-dual bivariate bicycle codes with transversal Clifford gates”, (2026) arXiv:2510.05211

  12. [12]

    K. Chen, Y. Liu, Y. Zhang, Z. Liang, Y.-A. Chen, K. Liu, and H. Song, “Anyon Theory and Topological Frustration of High-Efficiency Quantum Low-Density Parity-Check Codes”, Physical Review Letters 135, (2025) arXiv:2503.04699 DOI

  13. [13]

    J. Cruz-Benito, A. W. Cross, D. Kremer, and I. Faro, “Evolutionary Discovery of Bivariate Bicycle Codes with LLM-Guided Search”, (2026) arXiv:2606.02418

  14. [14]

    J. Y. Lee, K. Okada, N. Maskara, K. Kasai, and H. Zhou, “Design Principles for Ultra-High-Rate Quantum Codes”, (2026) arXiv:2609.30069

  15. [15]

    J. N. Eberhardt and V. Steffan, “Logical Operators and Fold-Transversal Gates of Bivariate Bicycle Codes”, (2024) arXiv:2407.03973

  16. [16]

    H. Sayginel, S. Koutsioumpas, M. Webster, A. Rajput, and Dan E Browne, “Fault-Tolerant Logical Clifford Gates from Code Automorphisms”, (2025) arXiv:2409.18175

  17. [17]

    N. P. Breuckmann, M. Davydova, J. N. Eberhardt, and N. Tantivasadakarn, “Cups and Gates I: Cohomology Invariants and Logical Quantum Operations”, Communications in Mathematical Physics 407, (2026) arXiv:2410.16250 DOI

  18. [18]

    P. Panteleev and G. Kalachev, “Degenerate Quantum LDPC Codes With Good Finite Length Performance”, Quantum 5, 585 (2021) arXiv:1904.02703 DOI

  19. [19]

    A. G. Fowler, A. M. Stephens, and P. Groszkowski, “High-threshold universal quantum computation on the surface code”, Physical Review A 80, (2009) arXiv:0803.0272 DOI

  20. [20]

    S. Koutsioumpas, H. Sayginel, M. Webster, and Dan E Browne, “Automorphism Ensemble Decoding of Quantum LDPC Codes”, (2025) arXiv:2503.01738

  21. [21]

    J. Blue, H. Avlani, Z. He, L. Ziyin, and I. L. Chuang, “Machine Learning Decoding of Circuit-Level Noise for Bivariate Bicycle Codes”, Quantum 10, 2149 (2026) arXiv:2504.13043 DOI

  22. [22]

    K. Sahay, D. J. Williamson, and B. J. Brown, “A matching decoder for bivariate bicycle codes”, (2026) arXiv:2602.22770

  23. [23]

    S. J. S. Tan, I. Gill, E. Huang, P. Liu, C. Zhao, H. Dehghani, A. Kubica, H. Zhou, and A. Dua, “Generalized matching decoders for 2D topological translationally-invariant codes”, (2026) arXiv:2603.05402

  24. [24]

    L. Z. Cohen, I. H. Kim, S. D. Bartlett, and B. J. Brown, “Low-overhead fault-tolerant quantum computing using long-range connectivity”, Science Advances 8, (2022) arXiv:2110.10794 DOI

  25. [25]

    Q. Xu, J. P. B. Ataides, C. A. Pattison, N. Raveendran, D. Bluvstein, J. Wurtz, B. Vasic, M. D. Lukin, L. Jiang, and H. Zhou, “Constant-Overhead Fault-Tolerant Quantum Computation with Reconfigurable Atom Arrays”, (2023) arXiv:2308.08648

  26. [26]

    N. Berthusen, D. Devulapalli, E. Schoute, A. M. Childs, M. J. Gullans, A. V. Gorshkov, and D. Gottesman, “Toward a 2D Local Implementation of Quantum Low-Density Parity-Check Codes”, PRX Quantum 6, (2025) arXiv:2404.17676 DOI

  27. [27]

    M. H. Shaw and B. M. Terhal, “Lowering Connectivity Requirements for Bivariate Bicycle Codes Using Morphing Circuits”, Physical Review Letters 134, (2025) arXiv:2407.16336 DOI

  28. [28]

    C. Gidney and C. Jones, “New circuits and an open source decoder for the color code”, (2023) arXiv:2312.08813

  29. [29]

    L. H. English, H. Luo, Y. Wang, B. Srivastava, S. D. Bartlett, and D. J. Williamson, “Duality constrains optimal thresholds in quantum error correction”, (2026) arXiv:2607.21160

  30. [30]

    B. C. B. Symons, A. Rajput, and D. E. Browne, “Sequences of Bivariate Bicycle Codes from Covering Graphs”, (2026) arXiv:2511.13560

  31. [31]

    S. Burton, “qecdb.org: Quantum Error Correction Database”, URL

  32. [32]

    Q. Xu, H. Zhou, D. Bluvstein, M. Cain, M. Kalinowski, J. Preskill, M. D. Lukin, and N. Maskara, “Batched high-rate logical operations for quantum LDPC codes”, (2025) arXiv:2510.06159

  33. [33]

    C. Li, J. Preskill, and Q. Xu, “Transversal dimension jump for product qLDPC codes”, (2026) arXiv:2510.07269

  34. [34]

    N. P. Breuckmann, S. H. Choe, J. N. Eberhardt, F. R. F. Pereira, and V. Steffan, “Logical Operators and Derived Automorphisms of Tile Codes”, (2025) arXiv:2511.14589

  35. [35]

    A. B. Khesin and J. Z. Lu, “Mirror codes: High-threshold quantum LDPC codes beyond the CSS regime”, (2026) arXiv:2603.05496

  36. [36]

    K. Wang et al., “Demonstration of low-overhead quantum error correction codes”, Nature Physics (2026) arXiv:2505.09684 DOI

  37. [37]

    B. Gu, T. Noszko, V. Steffan, J. N. Eberhardt, J. Roffe, J. Eisert, and S. Koutsioumpas, “Nearest-neighbour gates are all you need: High-rate quantum low-density parity-check codes on a planar grid”, (2026) arXiv:2606.19482

  38. [38]

    G. P. Gehér, D. Byfield, and A. Ruban, “Directional Codes: a new family of quantum LDPC codes on hexagonal- and square-grid connectivity hardware”, (2026) arXiv:2507.19430

Page edit log

Your contribution is welcome!

on github.com (edit & pull request)

— see instructions

Zoo Code ID: qcga

Cite as:
“Bivariate bicycle (BB) code”, The Error Correction Zoo (V. V. Albert & P. Faist, eds.), 2026. https://errorcorrectionzoo.org/c/qcga, arXiv:2606.11484
BibTeX:
@incollection{eczoo_qcga,
title={Bivariate bicycle (BB) code},
booktitle={The Error Correction Zoo},
year={2026},
editor={Albert, Victor V. and Faist, Philippe},
eprint={2606.11484},
doi={10.48550/arXiv.2606.11484},
url={https://errorcorrectionzoo.org/c/qcga}
}
Share via:
Twitter | Mastodon |  | E-mail
Permanent link:
https://errorcorrectionzoo.org/c/qcga

Cite as:

“Bivariate bicycle (BB) code”, The Error Correction Zoo (V. V. Albert & P. Faist, eds.), 2026. https://errorcorrectionzoo.org/c/qcga, arXiv:2606.11484

Github: https://github.com/errorcorrectionzoo/eczoo_data/edit/main/codes/quantum/qubits/stabilizer/qldpc/balanced_product/lp/bb/qcga.yml.