AI 中文总结
本文提出基于1-覆盖路径的算法，将含CNOT门序列的电路转换为适配任意量子比特连通性图的形式，应用于量子有限自动机时可降低16%-17%的CNOT电路成本，有望用于量子编译。
AI 中文摘要
量子计算广泛应用的障碍之一是高效电路合成问题。当前量子硬件的量子比特间连接有限，每个量子比特仅与少数其他量子比特相连，这意味着电路必须经过变换以适配该限制。本文提出一种算法，将包含CNOT门序列的电路转换为适用于任意量子计算机架构的形式。尽管仅在量子指纹（quantum fingerprinting）的背景下演示该算法，但类似门序列在量子算法中普遍存在，例如教材中的量子傅里叶变换。本文针对受限于两量子比特门应用的量子设备，实现了量子哈希算法（量子指纹算法）的量子电路，该限制由量子比特连通性图表示。作为该技术的应用示例，将其应用于识别一元MODₚ={a^ℓ: ℓ mod p=0}语言和EQₚ={a^ℓb^r: ℓ≡r mod p}语言的量子有限自动机。鉴于该算法带来的性能提升，例如在某一案例中实现了CNOT电路成本降低16%-17%，认为其也可在更广泛的量子编译场景中发挥作用。
英文摘要
One of the obstacles to the widespread adoption of quantum computing is the problem of efficient circuit synthesis. Current quantum hardware has limited connections between qubits, with each qubit connected to only a few others. This means that the circuit has to be transformed to accommodate this. In this paper, we present an algorithm that converts a circuit containing a sequence of CNOT gates into a form that is suitable for arbitrary quantum computer architectures. Although we demonstrate the algorithm only in the context of quantum fingerprinting, similar gate sequences are prevalent in quantum algorithms; for instance, they are present in the textbook quantum Fourier transform. We present a quantum circuit implementation of the quantum hashing algorithm (quantum fingerprinting algorithm) for a quantum device with restrictions on the application of two-qubit gates that are expressed as a qubit connectivity graph. As an example of usage of the technique, we apply it to quantum finite automata recognizing the unary $MOD_p=\{a^\ell: \ell \bmod p=0\}$ language, and the $EQ_p=\{a^\ell b^r: \ell \equiv r \pmod p\}$ language. Given the enhancements that our algorithm provides~-- for instance, in one case it achieves a 16\%--17\% decrease in CNOT circuit cost~-- we believe it could also be useful in a broader quantum compilation context.
Commentsaccpted to QUANCOM 2026
