Identifying More Equivalent Mutants via Code Similarity

Equivalent mutants are one of the major costs of mutation testing. The undecidable nature of this problem makes a fully automated solution unattainable and necessitates the manual analysis of live mutants. This paper introduces the concept of mirrored mutants, ones that affect similar code fragments. It is argued that mirrored mutants exhibit analogous behavior with respect to their equivalence. Thus, if one of them is equivalent, then the other mirrored mutants should be too. An empirical study, conducted on real world programs, investigates this argument, focusing on both intra-method and inter-method mirrored mutants. The obtained results suggest that mirrored mutants indeed exhibit this kind of behavior and thus can be utilized to ameliorate the adverse effects of the equivalent mutant problem.