Stabilizer Slicing: Coherent Error Cancellations in LDPC Codes

TL;DR

This paper introduces stabilizer slicing, a technique that suppresses coherent errors in LDPC stabilizer codes by destructive interference, significantly reducing logical error rates and enhancing fault-tolerance in quantum computing.

Contribution

The paper presents a novel stabilizer slicing method that leverages destructive interference of overrotations to suppress coherent errors in LDPC codes, demonstrating substantial error rate improvements.

Findings

Complete elimination of coherent overrotation errors with 3-body Hamiltonian gates.

135-fold reduction in logical error rate for Surface-17 with pure coherent errors.

89-fold improvement in Bacon-Shor-13 with conventional ion trap gates.

Abstract

Coherent errors are a dominant noise process in many quantum computing architectures. Unlike stochastic errors, these errors can combine constructively and grow into highly detrimental overrotations. To combat this, we introduce a simple technique for suppressing systematic coherent errors in low-density parity-check (LDPC) stabilizer codes, which we call stabilizer slicing. The essential idea is to slice low-weight stabilizers into two equally-weighted Pauli operators and then apply them by rotating in opposite directions, causing their overrotations to interfere destructively on the logical subspace. With access to native gates generated by 3-body Hamiltonians, we can completely eliminate purely coherent overrotation errors, and for overrotation noise of 0.99 unitarity we achieve a 135-fold improvement in the logical error rate of Surface-17. For more conventional 2-body ion trap gates, we observe an 89-fold improvement for Bacon-Shor-13 with purely coherent errors which should be testable in near-term fault-tolerance experiments. This second scheme takes advantage of the prepared gauge degrees of freedom, and to our knowledge is the first example in which the state of the gauge directly affects the robustness of a code's memory. This work demonstrates that coherent noise is preferable to stochastic noise within certain code and gate implementations when the coherence is utilized effectively.