Originally posted by: JorisK
Dear,
I'm implementing a MIP model for a variation on the Traveling Salesman Problem in C++. Now I would like to do the following:
1. Solve LP relaxation, strengthened with valid inequalities and the standard cuts added by cplex, 2. query the resulting objective value and the reduced costs per variable.
I'm having some issues with the implementation.
Approach 1: Implement my problem as an MIP model and set cplex.setParam(IloCplex::NodeLim, 0); to ensure that the solver halts after solving the root node. This approach ensures that all cuts are added. Furthermore it allows me to obtain the desired objective value, but I cannot access the reduced costs?
Approach 2: Implement my problem as an LP. This allows me to query the objective value and reduced costs, but unfortunately non of the cuts are added (UserCutcallback is ignored).
As a work around I could record the user cuts generated in approach 1 and add them to approach 2 as regular constraints. However I would not have access to the cuts generated by cplex, so this solution is far from ideal.
Is there any way to resolve this?
Joris
#CPLEXOptimizers#DecisionOptimization