Originally posted by: GGR
Hi
I do think the CPO solver does not succeed to prove the optimality after finding an incumbent at the known optimal cost. That is not very surprising as proof of optimality are generally very hard to achieve using Constraint Programming.
For the problem of OPL process failing, it could be, a possible hypothesis is a memory usage issue. There is a log that is regularly printed from engine information.
For trying to speed up the proof, you could
Stronger propagation: create if needed a new settings file and set the propagation controls parameters of constraint programming items in function of the global constraint you are using.
If your model has different flavors of variable, some being more important to decide first, create a search phase from this variables.
You can also break symmetries (using constraints), finding redundant constraints or adding constraints that tells conditions that are only true at optimality.
For the erroneous behavior of OPL, it could be a memory issue. A way to check it, is to follow the information on memory usage in the log of the engine View <<Engine Log>>. In any case, if you're allowed to send us your OPL project that we can investigate.
Hope that helps
#DecisionOptimization#OPLusingCPOptimizer