龙格库塔程序的似然蜕变关系识别方法

Identification method for likely metamorphic relations of Runge-Kutta program

为解决蜕变测试中的蜕变关系识别问题,提出一种似然蜕变关系识别方法。根据程序算法层面的数学性质,结合领域知识给出识别输入模式的指导规则;利用SPSS工具从程序输出结果中自动化地挖掘出相应的输出模式,使输出模式的挖掘过程既简单又高效;通过产生新的测试数据来检验输入、输出模式,获得一批有意义的似然蜕变关系。通过常微分方程龙格库塔算法程序实例,验证该方法能有效地识别出似然蜕变关系,为识别真正的蜕变关系提供了一种途径。

To solve the problem of identification of metamorphic relations in metamorphic testing,a method of identifying likely metamorphic relations was proposed.According to the mathematical properties of the program algorithm,a guide rule for identifying input patterns which combined with domain knowledge was provided.The SPSS tool was used to automatically mine the corresponding output patterns from the program output results,so that the output pattern mining process was both simple and efficient.New test data were generated to verify the input-output pattern,and the meaningful likely metamorphic relations were obtained.With an example of the Runge-Kutta algorithm of ordinary differential equations,it is verified that the proposed method can effectively identify likely metamorphic relations,and provides a way to identify the true metamorphic relations.