DOcplexcloud now supports two CPLEX features to help you locate and resolve infeasibilities in your model. You can use these features, conflict refiner and FeasOpt, when you know or suspect that certain constraints or bounds in your model are incompatible. These features are available for models in
The conflict refiner examines elements within an infeasible model that can be removed from the conflict in order to arrive at a minimal conflict. This minimal conflict can make it easier to identify the source of infeasibilities in the original model.
To invoke the conflict refiner, you can submit an infeasible CPLEX model along with a
.feasibility file containing the line:
DOcplexcloud runs the conflict refiner and returns an
.json file containing details of the minimal conflict. This minimal conflict helps you to identify which constraints or bounds need to be reconsidered in order to produce a feasible model.
The FeasOpt feature suggests how to repair an infeasible model according to preferences you set. FeasOpt selectively relaxes bounds and constraints in a way that minimizes a user-defined weighted penalty function. FeasOpt tries to suggest the least change to your model that would achieve feasibility. It does not modify the model itself.
To invoke the FeasOpt feature, you can submit a
.feasibility file containing the constraints that feature in a minimal conflict. You add a preference value to each constraint to identify how willing you are to relax that constraint.
DOcplexcloud runs the FeasOpt algorithm and returns an
.json file containing a solution to the relaxed model along with a set of values for infeasible constraints and/or bounds.
You can download a sample illustrating the use of the conflict refiner and FeasOpt from the DropSolve samples. You can read more about these features in the online CPLEX documentation: