Originally posted by: rezir

I have a two-stage stochastic model. I want to use L-shaped method (Bender's Decomposition) to solve that. As you know in this method the optimality cut and feasibility cut should be added to sub-problem based on the solution of master problem. Then they should be solved iteratively. I have some experience to use OPLScript to solve similar problem. I am wondering if using C++ concert technology has some advantages over using OPLScript  in this kind of the problem (e.g., computationally of flexibility,...).

I really appreciate your help.

Hi,

just to be sure do you know that Benders decomposition is in CPLEX 12.7 : https://www.ibm.com/support/knowledgecenter/en/SSSA5P_12.7.0/ilog.odms.cplex.help/CPLEX/ReleaseNotes/topics/releasenotes127/newBenders.html

And annotation in OPL in 12.7.1

regards

I would go with C++ Concert in this case. In C++ Concert it is a lot easier (and potentially faster) to modify a problem. You also have callbacks there that make things a lot simpler, for example if you only want to solve the sub-problems whenever an integer feasible to the master is found.

Note that even in C++ Concert you can load your model from OPL files, see the oplrunsample.cpp example in the opl/examples/opl_interfaces/cpp/src directory of your distribution.