在 IBM 量子计算机上通过对易 Pauli 字符串进行量子程序测试
Quantum Program Testing Through Commuting Pauli Strings on IBM's Quantum Computers
AI总结：
针对工业量子程序缺乏完整规约且难与错误缓解兼容的问题，本文提出基于对易 Pauli 字符串的 QOPS 测试方法，并在近19.5万个程序和三台 IBM 真机上验证其完美分类效果。
摘要：
量子计算最有前景的应用集中于求解搜索与优化任务，尤其是在物理模拟、量子化学和金融等领域。然而，当前量子软件测试方法在工业环境中应用时面临实际限制：（i）它们不适用于与工业最相关的量子程序；（ii）它们需要完整的程序规约，而这些程序通常无法获得完整规约；（iii）它们与 IBM 等主要工业参与者当前采用的错误缓解方法不兼容。为应对这些挑战，我们提出 QOPS，一种新颖的量子软件测试方法。QOPS 引入了基于 Pauli 字符串的测试用例新定义，以提升与不同量子程序的兼容性。QOPS 还引入了一种新的测试预言机，可直接与 IBM 的 Estimator API 等工业 API 集成，并能够利用错误缓解方法在真实的含噪声量子计算机上进行测试。我们还利用 Pauli 字符串的对易性质来放宽拥有完整程序规约的要求，使 QOPS 可实际用于工业场景中的复杂量子程序测试。我们在 194,982 个真实量子程序上对 QOPS 进行实证评估，结果表明其在测试评估方面相比现有最优方法具有有效性能，F1 分数、精确率和召回率均达到完美值。此外，我们通过在 IBM 的三台真实量子计算机上评估 QOPS 的性能，并结合工业级和开源错误缓解方法，验证了其工业适用性。
英文摘要：
The most promising applications of quantum computing are centered around solving search and optimization tasks, particularly in fields such as physics simulations, quantum chemistry, and finance. However, the current quantum software testing methods face practical limitations when applied in industrial contexts: (i) they do not apply to quantum programs most relevant to the industry, (ii) they require a full program specification, which is usually not available for these programs, and (iii) they are incompatible with error mitigation methods currently adopted by main industry actors like IBM. To address these challenges, we present QOPS, a novel quantum software testing approach. QOPS introduces a new definition of test cases based on Pauli strings to improve compatibility with different quantum programs. QOPS also introduces a new test oracle that can be directly integrated with industrial APIs such as IBM's Estimator API and can utilize error mitigation methods for testing on real noisy quantum computers. We also leverage the commuting property of Pauli strings to relax the requirement of having complete program specifications, making QOPS practical for testing complex quantum programs in industrial settings. We empirically evaluate QOPS on 194,982 real quantum programs, demonstrating effective performance in test assessment compared to the state-of-the-art with a perfect F1-score, precision, and recall. Furthermore, we validate the industrial applicability of QOPS by assessing its performance on IBM's three real quantum computers, incorporating both industrial and open-source error mitigation methods.
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。
