Infeasible model in Matlab according to Gurobi
AnsweredHello,
I have a model of a real-time railway traffic management problem with a Matlab implementation. When I use Gurobi to solve the optimization problem, it does not provide a feasible solution. However, the built-in Matlab solver (intlinprog) can solve it. What can be the problem?
Thank you for your answer in advance!
-
Hi László,
There can be a few reason for the difference. One reason might be different tolerances used by Gurobi and the built-in Matlab solver.
Another reason might be that the two models you solve are slightly different. You could write your model to a human-readable LP file via the gurobi_write() function. You can open the file with any standard text editor and analyze whether you model looks as it should.
You can also let Gurobi compute an IIS, cf. How do I determine why my model is infeasible?. For this you would have to use the gurobi_iss() function and write the resullt to a file via the gurobi_write() function.
Best regards,
Jaromił0
Please sign in to leave a comment.
Comments
1 comment