New Technique Thoroughly Tests Quantum Programs Without Detailed Blueprints

Quantum software testing represents a critical hurdle in the development of dependable quantum computing applications. Researchers Asmar Muqeet from Simula Research Laboratory and University of Oslo, Shaukat Ali from Simula Research Laboratory and Oslo Metropolitan University, and Paolo Arcaini from National Institute of Informatics Tokyo, working in collaboration across Norway and Japan, present a novel approach to address limitations in current testing methodologies. Their work introduces SB-QOPS, a search-based quantum program testing technique utilising commuting Pauli strings, which moves beyond reliance on simplistic test inputs and exhaustive program specifications. This innovative method redefines test cases and employs a measurement-centric oracle, substantially improving test budget efficiency and the potential to identify subtle program faults. Through extensive evaluation on circuits containing up to 29 qubits, utilising real quantum computers from IBM and Quantinuum, alongside simulations, the team demonstrates that SB-QOPS significantly surpasses existing techniques, achieving 100% fault detection and showcasing portability across diverse quantum platforms.

Scientists have developed SB-QOPS, a direct extension to a previously proposed QOPS approach, redefining test cases in terms of Pauli strings. The method introduces a measurement-centric oracle that exploits their commutation properties, enabling effective testing of quantum programs while reducing the need for full program specifications. SB-QOPS systematically explores the test space using an expectation-value-based fitness function, which guides the search towards test cases likely to reveal faults within the quantum programs. The Genetic Algorithm consistently yielded the highest quality solutions in terms of fitness optimisation, indicating its superior ability to navigate the complex search landscape. While the (1+1) Evolutionary Algorithm identified failing test cases at a faster rate, the Genetic Algorithm demonstrated a more robust and comprehensive search capability. Hill Climbing, although less effective than the other two, still contributed to the overall understanding of search strategy performance within the context of quantum program testing. Achieving a fault-detection score of 100%, SB-QOPS successfully identified errors in quantum circuits containing up to 29 qubits, representing a substantial advancement over IBM, IQM, and Quantinuum, confirming the adaptability of the approach to diverse quantum architectures and broadening its potential applicability. The effectiveness of SB-QOPS on real quantum hardware was shown to be critically dependent on the implementation of robust error mitigation techniques, such as those provided by IBM’s Estimator API, highlighting the importance of mitigating noise in practical quantum software testing. Quantum Computing (QC) has introduced new programming paradigms and computational capabilities, giving rise to the need for reliable quantum software engineering practices. As a result, Quantum Software Testing has emerged as a critical research area, facing fundamental challenges including the impossibility of directly observing quantum states without inducing state collapse, the inherent nondeterminism of quantum program outputs, and the limited availability and reliability of current quantum hardware. These characteristics render classical testing techniques largely unsuitable and necessitate the development of novel, quantum-aware testing strategies. Prior work has proposed a variety of approaches for testing quantum programs, including combinatorial testing, fuzzing, property-based testing, metamorphic testing, and mutation testing. Despite this progress, several key challenges remain unresolved, including the reliance on simplistic input states, the absence of robust and scalable test oracles, the difficulty of handling quantum noise, and the requirement for complete program specifications. Most existing testing methods restrict test inputs to computational basis states |0⟩ and |1⟩, which are incompatible for testing programs that handle input states in superposition, such as quantum search and optimisation algorithms. QOPS was proposed as a novel testing approach for quantum programs, representing test cases using Pauli strings and shifting the focus from input initialisation to measurement operations. This design makes QOPS particularly suitable for testing programs that inherently rely on superposition. Furthermore, QOPS introduces a new oracle that exploits the commuting properties of Pauli strings, which significantly reduces the need for exhaustive program specifications and is compatible with modern error mitigation techniques, including those provided by IBM’s Estimator AP.
In this work, researchers extend the original QOPS approach into SB-QOPS by introducing a search-based test generation strategy that systematically explores the search space. Instead of relying on unguided random exploration, SB-QOPS employs an expectation-value-based fitness function that effectively steers the search toward failing test cases. The empirical evaluation was improved by scaling the analysis to larger quantum circuits of up to 29 qubits and investigating the applicability of SB-QOPS across multiple quantum computing architectures by conducting experiments on IBM, IQM, and Quantinuum platforms. They also explicitly analysed the impact of quantum noise and studied the role of error mitigation techniques in enabling reliable quantum program testing on real quantum computers. Genetic Algorithm (GA), Hill Climbing (HC), and the (1+1) Evolutionary Algorithm ((1+1) EA); an extensive assessment of SB-QOPS under noisy conditions, covering both simulated noise and executions on real quantum computers, with and without error mitigation; and a cross-platform evaluation demonstrating the applicability of SB-QOPS on diverse quantum computing architectures, including IBM, IQM, and Quantinuum. Experimental results show that, compared to the original QOPS approach, SB-QOPS is substantially more effective at uncovering subtle faults. Among the evaluated search strategies, GA consistently produces higher-quality solutions in terms of fitness optimisation, whereas the (1+1) EA identifies failing test cases more quickly. Both GA and (1+1) EA detect faults in larger quantum circuits with an average fault detection score of 100%. Moreover, SB-QOPS is architecturally portable and can be executed across different quantum computing platforms, but its effectiveness on real quantum hardware critically depends on the availability of strong error mitigation techniques, such as IBM’s zero-noise extrapolation enhanced with probabilistic error amplification, to ensure reliable test assessment. Qubits constitute the fundamental unit of information in quantum computing, existing in a superposition of the computational basis states |0⟩ and |1⟩, described by complex probability amplitudes (α). The state of a qubit is expressed as |ψ⟩= α0|0⟩+ α1|1⟩, where the squared modulus of each amplitude determines the probability of obtaining the corresponding measurement outcome, constrained by the normalization condition |α0|2 + |α1|2 = 1. Quantum circuits and gates are used to program gate-based quantum computers, consisting of sequences of quantum gates that manipulate the quantum states of qubits. Quantum gates are unitary operators that change a qubit’s state based on a unitary matr.

The relentless pursuit of reliable quantum software has long been hampered by a fundamental tension between thorough testing demands and the resources available from near-term quantum devices. Existing methods, reliant on painstakingly crafted test cases or exhaustive program specifications, struggle to scale beyond trivial examples or offer meaningful validation on actual hardware. This new work offers a compelling step forward through a more intelligent approach to test case generation and fault detection, leveraging the mathematical properties of Pauli strings to reduce the need for detailed program knowledge while maintaining diagnostic power. What distinguishes this advance is its portability and demonstrated performance across multiple quantum computing platforms, signalling a move towards vendor-agnostic testing tools. While promising, the true test will be sustained performance as quantum circuits grow in complexity and the impact of noise becomes more pronounced. Looking ahead, this line of inquiry could converge with emerging techniques in quantum error mitigation and verification, combining intelligent test case generation with methods for characterising and correcting errors. Furthermore, the principles underpinning this approach may extend beyond software testing, informing the development of more robust and verifiable quantum algorithms. The challenge now is to move beyond demonstrating fault detection to actively diagnosing and correcting the root causes of errors in increasingly sophisticated quantum programs.