The IIS of infeasible model is feasible when I run it separately

Answered

Eunseok Kim

March 20, 2023 23:15

• 
• 
• 

Hello,

I have a question about infeasible models.

Currently, Gurobi claims my model is infeasible.

So after m.feasRelaxS(0, False, False, True), m.optimize() presents that the best objective is 0. As far as I know, this means that the sum of all artificial variables is 0, and Gurobi has found a feasible solution from the Phase 1 problem.

When I ran the IIS model (.ilp file) separately, Gurobi found an optimal solution which means that the model is feasible in fact.

In this case, what can I do to resolve this issue?

Best regards

Related to

• Infeasible
• Wrong Result

0

• 

  Simranjit Kaur

  • Gurobi Staff

  March 21, 2023 03:41

  Hi,

  The behaviour you are observing occurs mostly when a model has numerical issues. Numerical issues can sometimes result in the simplex algorithm declaring the model infeasible while the IIS considers it feasible.

  Please check if your model has numerical issues. To begin with, please check the coefficient statistics of your model to see if it has large coefficient ranges. The coefficient statistics of the model are printed at the start of the Gurobi log.

  You may find the article on determining if a model has numerical issues and the Guidelines for Numerical Issues helpful. 

  Regards,

  Simran

  0

Please sign in to leave a comment.