Skip to main content

Obtaining 0-1 Integer Solutions with Linear Programming solvers

Ongoing

Comments

2 comments

  • Maliheh Aramon
    Gurobi Staff Gurobi Staff

    Hi Prabhu, 

    When solving an LP, Gurobi terminates as soon as finding one proven optimal solution. Running the Gurobi Optimizer using different parameters such as Seed and/or Method might result in different optimal solutions in case there are multiple optimal solutions. However, there is no guarantee on which solution will be found. If you would like for some of the decision variables to take integral values, there is no way to enforce this on the LP relaxation of the model. You need to define them as integer (binary) variables. Gurobi will then solve a MIP instead of an LP. This would increase the runtime but will ensure the integrality of variables you would like them to be integral.

     Best regards,

    Maliheh

     

    0
  • Prabhu Manyem
    First Comment
    First Question

    Maliheh,  Thank you... However, my goal (or hope) is to obtain optimal Integer solutions using Linear Programming... Or by solving at most a polynomial number of Linear Programs... for example, by adding additional constraints (Valid Inequalities) to the model.

    Is it possible to get an optimal Integer solution by some kind of post-processing?

    Thanks.

    0

Please sign in to leave a comment.