Model is "Either infeasible or unbonded" reported by Gurobi, however feasible and bounded after providing with partial initial guess

Answered

XINWEI SHEN

• March 03, 2020 08:46

I'm calling Gurobi by Matlab to solve a power distribution system planning problem. And in one case, the model is "Either infeasible or unbonded" reported by Gurobi, however become feasible and bounded (a optimal solution is produced) after providing Gurobi with partial initial guess.

So what would possiblely make things wrong?

0

• 

  XINWEI SHEN

  • March 03, 2020 08:52
  • Edited

  

  

  

  Some additional information:

  1. This is a MISOCP model;

  2. The optimization status code is 12, which means that "Optimization was terminated due to unrecoverable numerical difficulties."

  So what can I do to find the problem of the solving process and make the model solvable?

  0
• 

  Silke Horn

  Gurobi Staff

  • March 03, 2020 12:10

  As you have seen yourself from the status code, the most likely explanation is that the model has numerical issues -- e.g. due to huge coefficient ranges. What are your model's coefficient statistics? (You can find them at the beginning of the log file.)

  Our Numerics Guidelines may also be helpful.

  0

Please sign in to leave a comment.