Hi, I am trying to solve a Maximization MIP problem with gurobi. It takes ages to solve the problem. I tried different warm-up starts, and I always receive "MIP start did not produce a new incumbent solution".
Hence, I decided to give the optimal solution of my model as a MIP start to see how faster the gurobi solves my model. The strange thing is that I still receive "MIP start did not produce a new incumbent solution" and yet from the log file I can see the gurobi used my MIP start as an incumbent. Moreover, the solving time didn't improve. Below, I explain the procedures:
1: I run the model and store decision variables values of the optimal solution.
2: Again, I run the same model but this time, I give a warm-up start which is the optimal solution I got from the previous step.
Do you have any idea why I receive "MIP start did not produce a new incumbent solution" with an optimal starting solution? and why the gurobi still uses it as an incumbent (I am sure the reason is not that gurobi can find a solution as good as mine, because I checked the first model, and the incumbent the gurobi start with, has much lower objective value than mine)? and why the solving time didn't improve?
I appreciate your help.
Please sign in to leave a comment.