Cutoff and optimal solution

Answered

Alexia Payan

• January 28, 2021 18:39

Hello All,

I get the following output from my model solve:

What does the current node "cutoff" objective mean? I still get an optimal solution printout after this happens with a gap of 11.1% corresponding to the Heuristics solution. Does it mean that the search along the tree was stopped because no better solution than the Heuristics one could be found?

Sometimes, I also get an "infeasible" objective value for the current node but I also get a solution printout corresponding to the latest Heuristics solution found. Is that a similar explanation as for the "cutoff" case?

0

• Official comment

  

  Juan Orozco

  Gurobi Staff

  • January 28, 2021 22:56
  • Edited

  As explained in the section MIP Logging of our reference manual, The sub-section Current Node of the Branch-and-Cut (B&C) Tree Search provides information on the specific node that was explored at that point in the B&C tree. Specifically, the tag cutoff indicates that the associated node was fathomed, since the objective value of its relaxation is worse than the objective value of the incumbent solution. Similarly, the tag infeasible indicates that the associated node was deemed as infeasible (note that this does not imply that the whole model is infeasible).
• 

  Alexia Payan

  • January 28, 2021 23:32

  

  

  

  Hello Juan,

  Thank you for your reply. Regarding both the "cutoff" case and the "infeasible" case, can I still trust the "optimal" solution that is being provided by the optimizer?

  Also, does this mean that the entire B&C tree has been explored when either case occur?

  Thanks.

  0
• 

  Juan Orozco

  Gurobi Staff

  • January 29, 2021 14:35

  As long as  you model doesn't have numerical issues, you can definitely trust the optimal solution reported by our Solver

  0

Please sign in to leave a comment.