TY - GEN
T1 - Generating Failing Test Suites for Quantum Programs With Search
AU - Wang, Xinyi
AU - Arcaini, Paolo
AU - Yue, Tao
AU - Ali, Shaukat
N1 - Publisher Copyright:
© 2021, Springer Nature Switzerland AG.
PY - 2021
Y1 - 2021
N2 - Testing quantum programs requires systematic, automated, and intelligent methods due to their inherent complexity, such as their superposition and entanglement. To this end, we present a search-based approach, called Quantum Search-Based Testing (QuSBT), for automatically generating test suites of a given size depending on available testing budget, with the aim of maximizing the number of failing test cases in the test suite. QuSBT consists of definitions of the problem encoding, failure types, test assessment with statistical tests, fitness function, and test case generation with a Genetic Algorithm (GA). To empirically evaluate QuSBT, we compared it with Random Search (RS) by testing six quantum programs. We assessed the effectiveness of QuSBT and RS with 30 carefully designed faulty versions of the six quantum programs. Results show that QuSBT provides a viable solution for testing quantum programs, and achieved a significant improvement over RS in 87% of the faulty programs, and no significant difference in the rest of 13% of the faulty programs.
AB - Testing quantum programs requires systematic, automated, and intelligent methods due to their inherent complexity, such as their superposition and entanglement. To this end, we present a search-based approach, called Quantum Search-Based Testing (QuSBT), for automatically generating test suites of a given size depending on available testing budget, with the aim of maximizing the number of failing test cases in the test suite. QuSBT consists of definitions of the problem encoding, failure types, test assessment with statistical tests, fitness function, and test case generation with a Genetic Algorithm (GA). To empirically evaluate QuSBT, we compared it with Random Search (RS) by testing six quantum programs. We assessed the effectiveness of QuSBT and RS with 30 carefully designed faulty versions of the six quantum programs. Results show that QuSBT provides a viable solution for testing quantum programs, and achieved a significant improvement over RS in 87% of the faulty programs, and no significant difference in the rest of 13% of the faulty programs.
KW - Genetic algorithms
KW - Quantum programs
KW - Software testing
UR - https://www.scopus.com/pages/publications/85117086479
U2 - 10.1007/978-3-030-88106-1_2
DO - 10.1007/978-3-030-88106-1_2
M3 - 会议稿件
AN - SCOPUS:85117086479
SN - 9783030881054
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 9
EP - 25
BT - Search-Based Software Engineering - 13th International Symposium, SSBSE 2021, Proceedings
A2 - O’Reilly, Una-May
A2 - Devroey, Xavier
PB - Springer Science and Business Media Deutschland GmbH
T2 - 13th International Symposium on Search-Based Software Engineering, SSBSE 2021
Y2 - 11 October 2021 through 12 October 2021
ER -