一种用于架构感知 ZX 多项式综合与优化的递归分区方法
A recursively partitioned approach to architecture-aware ZX Polynomial synthesis and optimization
AI总结:
本文针对 PauliOpt 架构感知综合提出递归分区的启发式搜索与分治方法,从 ZX 多项式综合优化量子电路,并在随机电路上取得显著优势。
摘要:
在 ZX-calculus 中,从 phase gadgets(相位小工具)综合量子电路有助于量子电路优化。我们的工作为 PauliOpt 的架构感知综合算法提供了另一种表述:用基于启发式的搜索取代 PauliOpt 的随机方法,并采用分治法从 ZX 多项式综合出优化电路。我们将该算法与 PauliOpt 及其他最先进的优化库进行了比较。虽然我们注意到其在高度结构化电路上性能较差,例如用于 Max-Cut 的 QAOA 表述,但我们证明了其在随机电路上的显著优势,这凸显了采用架构感知方法的优势。
英文摘要:
The synthesis of quantum circuits from phase gadgets in the ZX-calculus facilitates quantum circuit optimization. Our work provides an alternative formulation for the architecture-aware synthesis algorithm of PauliOpt by replacing the stochastic approach of PauliOpt with a heuristic based search and utilizes a divide and conquer method to synthesize an optimized circuit from a ZX polynomial. We provide a comparison of our algorithm with PauliOpt and other state-of-the-art optimization libraries. While we note poorer performance for highly structured circuits, as in the QAOA formulation for Max-Cut, we demonstrate a significant advantage for randomized circuits, which highlights the advantages of utilizing an architecture-aware methodology.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。