Research On Secomd Order Mutants Reduction Method Based On Multi-objective Differential Evolution Algorithm

The second-order mutation test artificially injects two defects into the original program.Compared with the first-order mutation test,it can better simulate the real error,and has been paid more and more attention and applied.However,the high cost of second-order mutation testing has hindered its application.How to reduce the number of mutants is an important research topic to reduce the cost of mutation testing.In the second-order mutation test,a large number of redundant mutants that are not helpful to improve the quality of the test set and are easy to be killed by the test set will be generated;At the same time,it will also generate mutants similar to the original program and difficult to kill.These mutants can not only be used to improve the quality of the test set,but also replace the first-order mutants that make up it to reduce the test cost.However,when traditional methods solve this problem,on the one hand,it is difficult to accurately find such high-quality second-order mutants,on the other hand,they lack effective treatment strategies for potential equivalent mutants,and have the problems of slow convergence and easy to fall into local optimal solutions.In this thesis,the multiobjective evolutionary algorithm based on decomposition is applied to the reduction of second-order mutants.Firstly,the method proposed in this thesis includes modifying the source code of Mujava project to support second-order mutation,and recording the necessary features of mutants.Secondly,the improved multiobjective evolutionary algorithm based on decomposition is selected as the core algorithm to transform the second-order mutant reduction problem into a multiobjective optimization problem.Finally,in order to verify the feasibility of the method,This thesis uses three typical laboratory procedures in Git Hub to evaluate the proposed method,and compares it with three typical multiobjective optimization algorithms.Experimental results suggest that the proposed method performs better in the dominance and diversity of solution set than other algorithms.It can not only greatly reduce the number of second-order mutants,but also find more strongly subsuming second-order mutants.