Skip to main content

Why the primal and dual simplex process is so long?

Answered

Comments

4 comments

  • Matthias Miltenberger
    Gurobi Staff Gurobi Staff

    Hi Yansong,

    You might want to try the barrier algorithm (Method=2) to solve the root node LP relaxation. Maybe this is faster than the simplex on this model.

    Sometimes, also relatively small models can have an unexpectedly long solving time - even just to solve the LP relaxation.

    Cheers,
    Matthias

    0
  • yansong bai
    Conversationalist
    Curious
    Gurobi-versary

    Thank you for this valuable advice. I will have a try.

    0
  • Michael Khachay
    First Question
    Gurobi-versary
    First Comment

    Hello, Matthias,

    I have a similar problem with my MILP model.
    At Root relaxation, the corresponding LP problem is solved fast by the barrier method + crossover. After that Gurobi proceeds with simplex and stuckles for hours, even for Method=2.

    With best regards,
    Michael

     

     

    0
  • Tobias Achterberg
    Gurobi Staff Gurobi Staff

    In very rare cases it may even help to also use barrier for the local node relaxations. There is of course the very big downside that barrier cannot benefit from the warm start information that you have when solving node LPs, such that every single LP solve needs to start again from scratch. But sometimes, simplex struggles so much that even this severe limitation of barrier can be accepted. To use barrier also for local nodes, you can set NodeMethod=2. But as I said, this is only useful in very exceptional cases.

    Regards,

    Tobias

     

    1

Please sign in to leave a comment.