The objective of our present experiments is to check that CPTEST is able to localize fault after its detection. We fed CPTEST with faulty CPUTs of four well-known CP problems, namely n-queens, social golfer, golomb rulers and car sequencing problems. We take small instances of these problems where it is more interesting to detect and localize fault on small instances then hard instances. To evaluate our framework, we use a standard process of fault injection [1].

All our experiments were performed on Intel Core2 Duo CPU 2.40Ghz machine with 2.00 GB of RAM.

[1] Jeffrey A. Clark and Dhiraj K. Pradhan. Fault injection. Computer, 28(6):47–56, 1995.