AI 中文总结
本研究开发开源双同位素镱量子计算噪声模拟工具，证实双同位素阵列原位测量架构可降低表面码逻辑错误率，明确里德伯态衰减是FTQC性能的主要限制。
AI 中文摘要
中性原子量子计算机是实现容错量子计算（FTQC）的有潜力平台，但逻辑性能取决于纠错码提取过程中系统性的现实噪声因素。在双同位素镱（Yb）阵列中，数据量子比特与辅助量子比特的角色通过频谱分离，使辅助量子比特可进行原位测量，无需额外的传输或 shelving（暂存）操作。本研究量化了双同位素Yb架构在表面码存储器中的优势：基于广义保罗旋转，开发了符合实验需求、兼容Clifford门的双同位素171Yb-174Yb系统噪声模型，并将其实现为Stim的封装工具DualYbSim，打包为开源Python库。对旋转表面码和XZZX表面码的模拟显示，采用原位测量的双同位素架构在所考虑的架构中实现了最低的逻辑错误率，优于基于暂存或分区测量的单同位素方案。本研究的错误预算分析还确定里德伯态衰减是主要限制因素，其贡献占逻辑错误率标度的74%-80%，为提升FTQC性能指明了具体实验目标。
英文摘要
Neutral-atom quantum computers are a promising platform for fault-tolerant quantum computation, but logical performance depends on systemic realistic noise factors during syndrome extraction. In dual-isotope Yb arrays, the roles of data and ancilla qubits are separated spectrally, allowing ancilla qubits to be measured in place without additional transport or shelving operations. Here we quantify the advantage of a dual-isotope Yb architecture for surface code memories. We develop an experimentally motivated Clifford-compatible noise model for dual-isotope 171Yb-174Yb systems using generalised Pauli twirling and implement it as a wrapper for Stim called DualYbSim, which has been packaged as an open source Python library. Simulations of rotated and XZZX surface codes show that a dual-isotope architecture with in-place measurement achieves the lowest logical error rates among the architectures considered, outperforming single-isotope schemes based on shelving or zoned measurement. Our error-budget analysis also identifies Rydberg-state decay as the dominant limitation, contributing to 74-80% of the logical error rate scaling, highlighting concrete experimental targets for improving FTQC performance.
Comments31 pages, 9 figures, comments welcome
