Genetic algorithms as shrinkers in property-based testing

Fang-Yi Lo, Chao-Hong Chen, Ying-Ping Chen

Abstract: This paper proposes the use of genetic algorithms as shrinkers for shrinking the counterexamples generated by QuickChick, a property-based testing framework for Coq. The present study incorporates the flexibility and versatility of evolutionary algorithms into the realm of rigorous software development, in particular, making the results of property-based testing observable and comprehensible for human. The program code for merge sort is investigated as a showcase in the study. Due to the lack of similar proposals in the literature, random sample is used to compete with the proposal for comparison. The experimental results indicate that the proposed genetic algorithm outperforms random sample. Moreover, the minimal counterexamples, through which programmers are able to pinpoint the program mistakes with ease, can be successfully obtained by using genetic algorithms as shrinkers.