Watch how CPLEX can change your infeasible optimization problem into a feasible one. In this video, Dr. Meysam Cheramin, the Operations Research Specialist of Cresco International, will teach you how to use the Conflicts and Relaxations technologies of CPLEX to resolve infeasibility in your optimization problems. Dr. Meysam Cheramin has a Ph.D. in Systems and Industrial Engineering from the University of Arizona. He is the author of several academic papers on optimization and presented his work at several international conferences including the Institute for Operations Research and Management Sciences (INFORMS) and the Institute of Industrial and Systems Engineers (IISE). He also has a few years of experience in teaching optimization, simulation, and statistics at the university level.
Infeasibility in an Optimization model can be caused by:
• The existence of at least two constraints in conflict – i.e. all possible solutions to one are excluded from being a solution to the other
• Unbounded constraints or an unbounded model that render the search for an optimal solution impossible
OPL provides two techniques in the CPLEX® optimizer engine to help you resolve such problems:
• Conflict refinement finds constraints that are in conflict.
• Relaxation suggests minimal changes in constraints that will render the model feasible, by relaxing one or more of the bounds defined in the constraints.
A conflict is a set of mutually contradictory constraints and/or bounds within a model. In other words, it is impossible for all the constraints to be true. A conflict is said to be a minimal conflict if it becomes feasible when any one constraint or bound is removed from the set. This minimal conflict usually concerns a subset of the constraints in the full model, thus making it easier to analyze the source of infeasibilities. OPL’s conflict refinement technology finds a minimal conflict for you. In OPL, when a conflict occurs, it is displayed in the Conflicts tab. The information in this tab expresses the necessary change to make the model feasible: you must remove or modify at least one of the conflicting constraints.