0
$\begingroup$

I need some advice, any hints will be appreciated. I have a MILP and I am using Gurobi to solve it. I have a feasible integer solution. When I use mipstart, after 30 min of execution the gap is extremely high (100%). It advances a little.

However, when I do not provide the known integer feasible solution, after 30 min an integer solution is found.

So, should I use the feasible solution? If not, are there any parameters in Gurobi that I can adjust and make the lower bound better?

$\endgroup$
3
  • $\begingroup$ Without seeing the model (or at least a log), there is no way to tell. It could be a bad model formulation, bad out of the box solver performance, or some other reason. You might want to try the tuning tool first to find better parameters: gurobi.com/documentation/9.0/refman/… $\endgroup$ – Walter Sebastian Gisler Jun 11 at 6:08
  • $\begingroup$ It's unclear what you're asking. In general, providing a starting solution is beneficial and helps the solver find the optimal solution. $\endgroup$ – mattmilten Jun 12 at 13:56
  • $\begingroup$ You said it's a feasible integer solution. When you pass that to mipstart, do you know whether gurobi uses it? For example, you don't see a message like "User MIP start did not produce a new incumbent solution" in gurobi's log? You can also try to fix all the variables in your model to the values of your feasible solution. If you get an infeasible solution, you know that what you thought was a feasible solution, was in fact infeasible (this is very common) $\endgroup$ – EhsanK Jun 13 at 19:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.