A Preliminary Study on Generating Well-Formed Q# Quantum Programs for Fuzz Testing

Abstract

Generative Sequence-To-Sequence models have been proposed for the task of generating well-formed programs, an important task for fuzz testing tools such as compilers. In this paper, we propose a Sequence-to-Sequence model to generate well-formed Q# Quantum programs. The ratio of syntactically valid programs among 1,000 Q# files generated by our model is 79.6%. In addition, we also contribute with a dataset of 1,723 Q# files taken from publicly available repositories on GitHub, which can be used by the growing community of Quantum Software Engineering.