Dear
I'm trying to compute a MUS/IIS from a CPMpy model using both CPLEX and CP Optimizer's conflict refiner. The tricky part is that a valid CPMpy constraint may correspond to multiple CPLEX/CP Optimizer constraints and hence these constraints need to be grouped in order for the CPMpy-level MUS to still be minimal. CPLEX' conflict refiner does allow for grouping of constraints and according to the documentation, so should CP Optimizer. However the actual API does not seem to allow for groups as a parameter of the refine_conflict function. This limitation could be circumvented within CPMpy itself by reifying all the constraints, and use the same indicator variable for constraints in the same group. But I would expect a native grouping to be a better approach. Since it is mentioned in the docs and is already part of CPLEX I was hoping this is something that can be added to the python API for CPO as well.
Kind regards
Orestis
Dear
I'm trying to compute a MUS/IIS from a CPMpy model using both CPLEX and CP Optimizer's conflict refiner. The tricky part is that a valid CPMpy constraint may correspond to multiple CPLEX/CP Optimizer constraints and hence these constraints need to be grouped in order for the CPMpy-level MUS to still be minimal. CPLEX' conflict refiner does allow for grouping of constraints and according to the documentation, so should CP Optimizer. However the actual API does not seem to allow for groups as a parameter of the
refine_conflictfunction. This limitation could be circumvented within CPMpy itself by reifying all the constraints, and use the same indicator variable for constraints in the same group. But I would expect a native grouping to be a better approach. Since it is mentioned in the docs and is already part of CPLEX I was hoping this is something that can be added to the python API for CPO as well.Kind regards
Orestis