On the Local Equivalence of 2D Color Codes and Surface Codes with Applications

Arun B. Aloshious, Arjun Nitin Bhagoji, Pradeep Kiran SarvepalliarXiv 2018 · DOI · arXiv

Abstract

This study establishes that any 2D topological color code, even those lacking translational symmetry, is locally equivalent to two copies of a related surface code. Using a linear algebraic framework, we demonstrate how to project color code syndromes onto surface codes for efficient decoding and provide ancilla-free circuits for code transformation. Performance benchmarks on square-octagonal lattices show a 5.3% bit-flip threshold and a 30.8% erasure threshold, highlighting the practical utility of this equivalence.

Introduction

Topological Stabilizer Codes are pivotal for fault-tolerant quantum computing. Surface codes are renowned for high error thresholds, while color codes offer advantages in performing complex logical gates. Previous research into their relationship often assumed translation invariance—a condition where the lattice structure repeats perfectly. This paper relaxes that assumption, proving the equivalence for arbitrary 2D color codes and exploring applications in decoding and code-switching.

Methodology

The authors employ a linear algebraic and coding-theoretic framework to bridge the two code families.

Lattice Contraction

A fundamental component of the proof is the definition of a map $\pi$ that "contracts" specific colored faces of a 2D color code lattice. This process generates a new graph $\tau_c(\Gamma)$ which serves as the foundation for a surface code.

Lattice contraction mapping a 2D color code lattice to a surface code lattice.

Syndrome Projection and Decoding

The equivalence allows for "Syndrome Projection," where error signals from a color code are mapped onto two independent surface codes. This enables the use of established surface code algorithms, such as Minimum Weight Perfect Matching (MWPM), to decode the more complex color code.

Mapping of color code operators to surface code operators for syndrome projection.

Key Insights

  • The Power of Two: Every 2D color code is mathematically equivalent to exactly two copies of a surface code.
  • Local Equivalence: The mapping is a local property, meaning it does not require a perfectly repeating lattice. It holds for any valid 2-colex structure.
  • In-Place Transformation: The transformation between codes can be achieved using local Clifford circuits (CNOT, Hadamard, SWAP) without the need for additional ancilla qubits.

Results

Simulations on square-octagonal (4.8.8) lattices provided the following benchmarks:

  • Bit-Flip Threshold: A threshold of ~5.3% was achieved using hard-decision decoding via the projection method.
  • Erasure Threshold: On the quantum erasure channel, the mapping achieved a threshold of 30.8%.
  • Linear Complexity: The transformation circuits scale linearly ($O(n)$) with the number of qubits, ensuring low overhead for code switching.

Conclusion

This work provides a rigorous foundation for the relationship between 2D color codes and surface codes. By removing the requirement for translation invariance, it opens the door for more flexible hardware implementations. Future work will extend these results to planar architectures with boundaries and explore joint decoding strategies to further improve error thresholds.

Summary updated

Abstract

In recent years, there have been many studies on local stabilizer codes. Under the assumption of translation and scale invariance Yoshida classified such codes. His result implies that translation invariant 2D color codes are equivalent to copies of toric codes. Independently, Bombin, Duclos-Cianci, and Poulin showed that a local translation invariant 2D topological stabilizer code is locally equivalent to a finite number of copies of Kitaev's toric code. In this paper we focus on 2D topological color codes and relax the assumption of translation invariance. Using a linear algebraic framework we show that any 2D color code (without boundaries) is locally equivalent to two copies of a related surface code. This surface code is induced by color code. We study the application of this equivalence to the decoding of 2D color codes over the bit flip channel as well as the quantum erasure channel. We report the performance of the color code on the square octagonal lattice over the quantum erasure channel. Further, we provide explicit circuits that perform the transformation between 2D color codes and surface codes. Our circuits do not require any additional ancilla qubits.
BibTeX
@misc{aloshious2018local,
  author = {Arun B. Aloshious and Arjun Nitin Bhagoji and Pradeep Kiran Sarvepalli},
  title = {{On the Local Equivalence of 2D Color Codes and Surface Codes with Applications}},
  year = {2018},
  eprint = {1804.00866},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1804.00866},
}
Figure