Modeling and Solving a Resource Allocation Problem with Soft Constraint Techniques

We studya resourceallocationproblem,which is a centralpieceof a real-world crew schedulingproblem.We first formulatetheproblemasa hybrid soft constraintsatisfaction and optimizationproblemand show that its worstcasecomplexity is NP-complete. Wethenproposeandstudyasetof decisionand optimizationmodelingschemesfor theproblem.Weconsidertheexpressi veness of thesemodelingmethods.Specifically, weexperimentallyinvestigatehow these modelingschemes interplay with the bestexisting systematicsearchand local searchmethods.Ourexperimentalresultsshow thatsoftconstrainttechniquescan beeffective on largeresourceallocationprobleminstances, andanoptimization approach is more efficient than a model checkingapproach basedon decision models.