Comparing search techniques for finding subtle higher order mutants

Subtle Higher Order Mutants (HOMs) are those HOMs that cannot be killed by existing test suites that kill all First Order Mutants (FOMs) for the program under test. Subtle HOMs simulate complex, real faults, whose behavior cannot be simulated using FOMs. However, due to the coupling effect, subtle HOMs are rare in the exponentially large space of candidate HOMs and they can be costly to find even for small programs. In this paper we propose new search techniques for finding subtle HOMs and extend our prior work with new heuristics and search strategies. We compare the effectiveness of six search techniques applied to Java and AspectJ programs. Our study shows that more subtle HOMs were found when the new heuristics and search strategies were used. The programming language (Java or AspectJ) did not affect the effectiveness of any search technique.