Probabilities in the TorX test derivation algorithm
Loe M. G. Feijs, Nicolae Goga, Sjouke Mauw · 2000
We propose to extend the TorX algorithm for automatic test derivation with explicit probabilities. Using these probabilities, the generated test suite can be tuned and optimised with respect to the chances of nding errors in the implementation. The main result of this paper is a theorem that shows that the optimal balance between giving stimuli and checking responses is determined by the ration of inputs and outputs along a typical test trace. A simulation experiment demonstrates that this gives rise to an improved error detection capability. Keywords: testing, test generation, probabilities, tools. 1 Introduction The part of the software development process where the application of formal methods is expected to have considerable impact in the near future is the phase of testing. Manual derivation and execution of test cases leads to an expensive, time consuming and suboptimal testing process. We think that problem areas such as regression testing and conformance testing will benet...