AI 中文总结
针对量子计算机资源不足问题，MosaiQC框架采用混合热启动与细化优化，结合线路和门切割及快速近似二次分配求解器，提升局部保真度，降低运行时和采样开销成本，在速度和质量间实现卓越权衡，减少电路切割放置计算开销。
AI 中文摘要
当前量子计算机的量子比特资源尚无法满足大多数实际量子算法的需求。为克服这一限制，人们探索了通过量子电路切割将算法分成若干部分的方法。许多此类工作要么呈现指数级扩展，要么远离最优解。本文提出了MosaiQC这一新颖框架以改进现有电路切割框架。它采用混合热启动与细化优化来寻找切割解决方案，允许同时进行线路和门切割，还能实现混合尺寸的硬件分区。细化阶段纳入快速近似二次分配求解器以更好地放置硬件分区，相较于基线算法，平均局部保真度提高了19.56%±6.17%。在运行时和采样开销成本方面，分别有2.88倍的改进以及平均16.84%的切割减少（导致平均5.83·10¹¹倍的开销减少）。MosaiQC在运行速度和解决方案质量之间展现出卓越的权衡，同时增添了多数竞争对手所没有的基本特性，证明了可扩展启发式优化能大幅降低日益大型的量子电路切割放置的计算开销。
英文摘要
Current quantum computers do not yet have the required qubit resources to meet the demands of most practical quantum algorithms. To circumvent this constraint, the practice of dividing these algorithms into parts through quantum circuit cutting has been explored. Many of these works either show exponential scaling or are far from optimal solutions. In this paper, MosaiQC is presented as a novel framework to improve upon existing circuit cutting frameworks. A hybrid warmstart with refinement optimization is used to find cutting solutions, allowing the combination of both wire and gate cuts. Additionally, MosaiQC enables hardware partitions of mixed sizes. Furthermore, the refinement stage incorporates a fast approximate quadratic assignment solver to better place hardware partitions, demonstrating a mean local fidelity improvement of $19.56 \% \pm 6.17\%$ over the baseline algorithm. In runtime and sampling overhead costs, improvements of $2.88 \times$ and an average of $16.84\%$ cut reduction (resulting in an average $5.83 \cdot 10^{11} \times$ overhead reduction) are observed. MosaiQC demonstrates a superior trade-off for run speed and solution quality, while adding fundamental features excluded by most competitors. With this, MosaiQC demonstrates that scalable heuristic optimization can substantially reduce the computational overhead of circuit-cut placement for increasingly large quantum circuits.
Comments32 pages, 20 figures
