Stochastic Constraint Optimisation with Applications in Network Analysis

Stochastic Constraint (optimisation) Problems (SCPs) are problems that combine weighted model counting (WMC) with constraint satisfaction and optimisation. We present an extensive study of methods for exactly solving SCPs in network analysis, where the underlying probability distributions have a monotonic property. These methods use knowledge compilation to address the model counting problem; subsequently, either a constraint programming (CP) solver or mixed integer programming (MIP) solver is used to solve the overall SCP. To configure the space of parameters of these approaches, we propose to use the framework of programming by optimisation. The result shows that a CP-based pipeline obtains the best performance.