Originally posted by: gokhan_ceyhan

Hi,

When I solve my mip problem with Interactive solver, it says that mip - integer infeasible and gap is infinite. On the other hand, when I write conflict comment, it says there is no conflict. What is the reason of this problem and what should I do to see the conflicts?

Gokhan,

This could be due to numerical problems. Can you attach your model here? If not, what is the output of 'disp prob stats'?

Originally posted by: gokhan_ceyhan

Here is the "disp prob stats" info:

Objective sense      : Maximize
Variables            :    1901  [Fix: 48,  Box: 877,  General Integer: 976,
                                 Qobj: 853]
Objective nonzeros   :     974
Objective Q nonzeros :     853
Linear constraints   :    2138  [Less: 1969,  Greater: 97,  Equal: 72]
  Nonzeros           :    7793
  RHS nonzeros       :     295

Variables            : Min LB: 0.0000000        Max UB: 199.7100       
Objective nonzeros   : Min   : 0.06798640       Max   : 2.079934e+007  
Objective Q nonzeros : Min   : 2.948491e-008    Max   : 1267.732       
Linear constraints   :
  Nonzeros           : Min   : 1.000000e-005    Max   : 1.449771e+007  
  RHS nonzeros       : Min   : 1.000000e-005    Max   : 1.271616e+007  

OK, the range of your coefficients is huge. With that it could well be that you are facing numerical troubles.

You should at least enable the numerical emphasis parameter. Any chance to reduce the range of the coefficients?

Originally posted by: gokhan_ceyhan

Activating numerical emphasis is not working also. Is there any common way to reduce range of the coefficients?

Moreover, although there is a limitation on the elapsed time, CPLEX could not stop within the given time limit (It starts branching, but stuck in a point). Is the reason of this situation also related with the numerical issues?

There is no general guide for reducing the range of coefficients since that depends on the real-world problem you are trying to model. Scaling (parts of) the data sometimes can help.

CPLEX should respect a time limit even if there are numerical issues. If CPLEX violates a time limit then this smells like a bug. Can you share your model and exact parameter settings so that we can try to reproduce here?