A Case Study of Metamorphic Test for Monte Carlo Numerical Calculation Programs

The Monte Carlo numerical calculation program is based on non-deterministic theory, which encounters the ‘test oracle’ problem. Metamorphic Test (MT) is an effective method to alleviate this problem. This paper applies MT to test the Monte Carlo numerical calculation programs and proposes a new method for deriving the Metamorphic Relations (MRs) in MT. Firstly, MRs are analyzed through the random sampling model of under-test programs. Secondly, MT is performed. Finally, the results and the effectiveness of MT are analysed. The experiments show that MRs can effectively detect errors in the program, and the proposed method can alleviate the ‘test oracle’ problem in Monte Carlo numerical calculation programs.