arXiv:2611.00985
Normal Factor Graph Representations of Stabilizer Quantum Codes with Applications
July X. Li, Pascal O. Vontobel
Abstract
We develop normal factor graph (NFG) representations of stabilizer quantum error-correcting codes and entanglement-assisted quantum error-correcting codes. We introduce two complementary representations. Quantum-code NFGs describe stabilizer and normalizer label codes and provide a graphical framework for establishing self-orthogonality, constructing codes from component codes, and identifying encoding and decoding circuits. Quantum-probability NFGs represent the quantum-mechanical probabilities associated with encoding, noise, and measurement. We also introduce a doubleedge formulation that makes the relationship between these two representations explicit. We apply the framework to convolutional and tail-biting convolutional stabilizer codes, as well as to two classes of low-density parity-check stabilizer codes. Finally, we extend the quantum-probability representation to entanglementassisted codes.
AI Summary
The paper asks how normal factor graphs (NFGs) can represent stabilizer codes while exposing commutation constraints, circuit structure, and quantum-mechanical probabilities. It develops two complementary descriptions: quantum-code NFGs for stabilizer and normalizer label codes, and quantum-probability NFGs for encoding, noise, recovery, and measurement. A double-edge formulation groups paired quantum variables and relates the graphical structures. For quantum-code NFGs, local constraints are additive or linear codes. A graph-duality theorem, including a proof of the reverse inclusion missing from an earlier treatment, identifies the normalizer representation by dualizing local codes. Self-orthogonal local codes consequently produce a self-orthogonal global code. Opening and closing graph components connects code composition with Clifford-circuit decomposition. The paper also gives a decoding-circuit normal form with six gate-type stages, C–P–C–P–H–C, or five using controlled-$Z$; these stages are not bounds on physical circuit depth. The main constructive application is to convolutional and tail-biting stabilizer codes. Explicit algorithms generate self-orthogonal elementary constraints and encoding circuits using $\nu=\sum_i\nu_i$ memory qubits for additive label codes. For $\mathrm{GF}(4)$-linear label codes, the construction uses between $\nu/2$ and $\nu$ memory qubits. Examples reduce memory from three qubits to one or two relative to cited shift-register constructions. Further applications graphically re-establish commutation for hypergraph-product and toric codes and extend probability representations to entanglement-assisted codes. The memory results concern the supplied constructions rather than a general optimality theorem. Developing decoding algorithms that exploit these representations while explicitly accounting for degeneracy is left for future work.
Error found: The Section IV-C Hermitian modification of the convolutional-code state-generation algorithm fails on the paper's own Section IV-F example, producing elementary-code rows with nonzero Hermitian inner product that its diagonal corrections cannot repair. The toric-code proof incorrectly infers S=C_Z from S-perp=C_Z; its intended argument requires S-perp=C_Z-perp.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.