QCRMut:量子电路随机变异体生成工具
QCRMut: Quantum Circuit Random Mutant generator tool
AI总结:
提出QCRMut工具,通过不可变位置保留电路结构并生成随机变异体,在稳定Qiskit版本上实现高效、代表性的变异测试,与穷举变异相比得分相当且更稳定。
摘要:
随着量子计算走向实际部署,确保量子软件的可靠性变得越来越重要。变异测试在此背景下是一种有前景的技术;然而,现有的穷举变异生成器主要针对旧版Qiskit(0.x)开发,限制了其对当前稳定版本的适用性。本文提出QCRMut,一种兼容稳定Qiskit版本的量子电路变异测试工具,它通过不可变位置支持电路结构的有控保留,并作为穷举变异的替代方案,实现高效、有代表性的随机变异体生成。我们根据四个设计原则开发QCRMut:唯一性、相似性、代表性和覆盖性。我们通过在一组量子电路基准套件上比较随机采样变异子集与穷举变异产生的变异分数,对工具进行实证评估。我们进一步分析了对随机种子的敏感性,并应用统计检验评估观察到的差异的稳健性。最后,我们将QCRMut与现有变异测试工具进行比较。在整个基准套件中,QCRMut生成的随机变异子集获得的变异分数与穷举变异相当。结果在不同随机种子下稳定,并突显了现有方法中由我们的方法解决的实际局限性。总体而言,QCRMut通过结合与稳定Qiskit版本的兼容性、有控结构保留、统计上合理的评估和高效的变异体生成,为量子电路的变异测试提供了实用且可扩展的解决方案。此外,它还能对先前工具无法处理的电路进行变异和分析。
英文摘要:
As quantum computing moves towards practical deployment, ensuring the reliability of quantum software becomes increasingly important. Mutation testing is a promising technique in this context; however, existing exhaustive mutation generators have primarily been developed for legacy versions of Qiskit (0.x), limiting their applicability to current stable releases. This work presents QCRMut, a mutation testing tool for quantum circuits compatible with stable Qiskit versions, which supports controlled preservation of circuit structure through immutable positions and enables efficient, representative random mutant generation as an alternative to exhaustive mutation. We develop QCRMut according to four design principles: unicity, similarity, representativity, and coverability. We evaluate the tool empirically by comparing mutation scores obtained from randomly sampled mutant subsets against those produced by exhaustive mutation over a benchmark suite of quantum circuits. We further analyse sensitivity to random seeds and apply statistical tests to assess the robustness of the observed differences. Finally, we compare QCRMut with existing mutation testing tools. Across the benchmark suite, randomly generated mutant subsets produced by QCRMut achieve mutation scores that are comparable to those obtained via exhaustive mutation. The results are stable across different random seeds and highlight practical limitations in existing approaches that are addressed by our method. Overall, QCRMut provides a practical and extensible solution for mutation testing of quantum circuits by combining compatibility with stable Qiskit releases, controlled structure preservation, statistically sound evaluation, and efficient mutant generation. In addition, it enables the mutation and analysis of circuits that cannot be handled by previous tools.
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。