QMutBench: A Dataset of Quantum Circuit Mutants
Eñaut Mendiluze Usandizaga, Paolo Arcaini, Thomas Laurent, Shaukat Ali · 2026
Read QMutBench: A Dataset of Quantum Circuit Mutants by Eñaut Mendiluze Usandizaga, Paolo Arcaini, Thomas Laurent, Shaukat Ali (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-15870.txt.
Introduction
Mendiluze Usandizaga, Laurent, Arcaini, and Ali present QMutBench, a dataset of over 700,000 quantum circuit mutants, faulty variants used to judge how well test cases catch bugs in quantum software. An online interface lets users select mutants by original circuit, desired survival rate, and fault type, such as the faulty gate involved. The collection aims to give developers a shared yardstick for comparing quantum testing techniques.
Catalog record
| CREATOR | Eñaut Mendiluze Usandizaga, Paolo Arcaini, Thomas Laurent, Shaukat Ali |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 21246 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.15870
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
QMutBench: A Dataset of Quantum Circuit Mutants Thanks: E. Mendiluze Usandizaga is supported by Simula’s internal strategic project on quantum software engineering. P. Arcaini is supported by the ASPIRE grant No. JPMJAP2301, JST. S. Ali is supported by the Qu-Test project (Project #299827) funded by the Research Council of Norway and Oslo Metropolitan University’s Quantum Hub. Eñaut Mendiluze Usandizaga, Paolo Arcaini, Thomas Laurent, Shaukat Ali Abstract Quantum software testing has attracted interest in recent years, prompting the development of various techniques to automate the testing of quantum software. These techniques generate test cases that must be assessed for their effectiveness in detecting faults. Such an assessment requires benchmarks of faulty programs. However, there is a lack of benchmarks containing faults. In this data showcase, we propose QMutBench , a dataset that contains over 700,000 quantum circuit mutants representing different faults. The dataset is accessible via an online interface with selection criteria, such as the original quantum circuit(s) from which mutants are generated, the desired survival rate of the selected mutants, and other mutation characteristics (e.g., the type of faulty quantum gate). QMutBench provides quantum software developers and testers with an accessible online dataset to obtain benchmarks of mutants necessary to assess either the quality of the test cases generated by their testing technique or to compare different testing techniques. It also enables the development of new mutation-guided quantum software testing techniques. Online Interface: https://enautmendi.github.io/QMutBench/ Code & Data: https://github.com/EnautMendi/QMutBench Video: https://youtu.be/6rjmiR16hdc I Introduction Quantum software promises to solve complex problems faster than classical software by utilising quantum phenomena such as superposition and entanglement. Such phenomena present new challenges for quantum software testing [ 1 ] . Thus, several quantum software testing techniques have emerged [ 2 , 3 , 4 ] . As the field of quantum software testing grows, it is essential to have systematic methods to assess the effectiveness of tests generated by these techniques. Typically, tests are assessed using benchmarks of faulty programs to measure their fault detection capability. However, in this new field, benchmarks of real faulty quantum programs remain scarce [ 5 , 6 ] . To address this limitation, mutation analysis, a technique that introduces artificial faults to programs to create their faulty versions, is often used to assess test effectiveness. In the context of quantum software testing, specialised tools have been developed for quantum-specific mutation analysis [ 7 , 8 ] . Given a set of quantum programs (specified as quantum circuits ), these tools can automatically generate a large number of quantum circuit mutants for each original quantum circuit. However, these tools lack an automated and systematic method, supported by strong empirical evidence, to select a representative subset of mutants from all those generated. In this data showcase, we address the challenge mentioned above, and we introduce QMutBench , a dataset of quantum circuits accessible via a selection interface. We provide a large set of quantum circuit mutants and allow users to apply various selection criteria to obtain a benchmark of mutants that meet their specific requirements. QMutBench is built based on the result of an extensive empirical evaluation on quantum circuit mutants [ 9 ] . The empirical evaluation was conducted on over 300 quantum circuits and over 700,000 quantum circuit mutants. The evaluation examined how these mutants interact with quantum circuit characteristics and their effects on survival rate , a metric quantifying the likelihood of survival (i.e., not being detected) for a mutant with specific characteristics. While that empirical evaluation offers valuable insights and a dataset of mutants for future research, its complexity limits the dataset’s usability. QMutBench addresses this issue by offering users filtering options to select a subset of mutants matching their criteria. We refer to the subset of mutants obtained from the full dataset as a benchmark , as it is the set of mutants used to assess test effectiveness. QMutBench is implemented as an online interface that users can easily access to select a subset of mutants that meet their selection criteria (e.g., type of quantum algorithm, desired survival rate , or specific quantum gates used in the mutants). The selected subset of mutants together with the original quantum circuit are provided as a set of open quantum assembly language ( QASM ) files [ 10 ] . This format is supported by most quantum computing platforms and tools, allowing for straightforward conversion to their respective programming languages. Researchers could use the dataset to evaluate and improve new techniques for testing quantum software. Three main use cases that we identify are: (i) First, the benchmarks obtained from QMutBench can be used to assess the quality of tests generated by a specific quantum software testing technique. (ii) Second, the benchmarks acquired from QMutBench can assist in developing new mutation-guided quantum testing techniques. (iii) Third, the benchmarks provided by QMutBench can be used to evaluate the effectiveness of quantum debugging techniques in locating and repairing faults. II Background This section introduces some concepts of quantum computing and quantum mutation analysis techniques. II-A Quantum Computing Quantum computing uses distinct characteristics, such as superposition and entanglement , to perform computations that can potentially perform certain tasks faster than classical computing. Quantum superposition is a key feature that enables computational speed-up as it allows a quantum bit ( qubit ) to exist in multiple states at the same time, in contrast to a classical bit, which can only exist in one state at a time (i.e., either 0 or 1) [ 11 ] . Quantum entanglement allows qubits to be correlated such that the state of one qubit is linked to the state of another qubit and a change in one will instantly affect the other [ 11 ] . Quantum computers are programmed using quantum circuits , which consist of several qubits and quantum gates that perform operations on those qubits. Fig. 1 shows a visual representation of a quantum circuit drawn in the IBM Quantum Composer [ 12 ] . Fig. 1: An example of a quantum circuit alongside examples of the applications of mutation operators. The circuit has four qubits ( q[0] to q[3] ) and a group of four classical bits (denoted collectively as c_{4} ). The circuit also shows examples of single-qubit gates (e.g., the Hadamard gate) and multi-qubit gates (e.g., the controlled-NOT gate). Mutations introduced by quantum mutation operators are highlighted in orange, along with the positions of the gates in the circuit. For instance, the Remove mutation operator removes the controlled-NOT gate to create a mutant of the original circuit. The circuit consists of four qubits, q[0] to q[3] , and four classical bits, collectively represented as c_{4} . A sequence of quantum gates is applied to the qubits. For example, a single-qubit Hadamard gate (denoted as H) is applied to q[0] , placing it into an equal superposition of the states 0 and 1. A multi-qubit gate (the controlled-NOT (CNOT)) is applied between q[0] and q[1] , where the value of q[0] determines whether the state of q[1] is to be flipped. The measurement operations in Fig. 1 measure the state of the qubits and store the measured values into classical bits. These operations collapse the quantum superpositioned states into definite classical states. II-B