Probabilistic Choice Operators as Global Constraints: Application to Statistical Software Testing

Probabilistic Concurrent Constraint Programming (PCCP) [3] is an extension of Concurrent Constraint Programming (CCP) [5] where probabilistic choice operators are introduced to represent the randomness or uncertain behaviour of processes. A probabilistic choice between two processes can be though of as flipping a coin : head the first process is triggered, tail it is the second. Based on this theoretical framework, it seems possible to extend the classical CCP over finite domains framework [4] with probabilistic choice operators. Our aim is to define probabilistic choice operators as global constraints of the CCP over finite domains paradigm [4] and to apply this framework to deal with a specific Software Testing problem [1]. Global constraints are a good way for giving global semantics to complex constraints. Furthermore, such operators appear to the user like single constraints and so can be awaked and treated efficiently by the constraint propagation algorithm. A part of our work is to establish the relationships between probabilistic choice operators, global constraints and the PCCP semantic framework. Gupta et al. pioneered the inclusion of probabilistic choice operators in CCP to address several applications areas, such as stochastic processes [3].