MIP model takes a large time to find the first feasible solution. what are possible solutions to decrease this time?

回答済み

leili Soltanisehat

2021年04月13日 23:14

• 
• 

I have a linear MIP model and it takes a very long time to find the first feasible solution despite the small size of the problem (Variable types: 0 continuous, 2162 integer (320 binary) )

I read through a lot of pages in Gorubi web pages but didn't find the answer. I have two main questions:

1- What are the possible options that can reduce the time in which the MIP can be solved?

2- I also don't know how callback may help to reduce the run time. I am wondering if you can help me to understand this,

Later on, I need to run my model for larger real data, and when I try my MIP model to this large data, the model never ends and gives the results. even it doesn't give any optimality gap for termination. 

 

 

here is the log file for small example (Variable types: 0 continuous, 2162 integer (320 binary))

 

....

 

0

• 正式なコメント

  

  Simranjit Kaur

  • Gurobi Staff

  2025年06月23日 02:56 編集済み

  This post is more than three years old. Some information may not be up to date. For current information, please check the Gurobi Documentation or Knowledge Base. If you need more help, please create a new post in the community forum. Or why not try our AI Gurobot?.
• 

  Jaromił Najman

  • Gurobi Staff

  2021年04月14日 09:30

  Hi Ieili,

  What are the possible options that can reduce the time in which the MIP can be solved?

  The documentation on Most important parameters should hold the information you are looking for. You should update to the latest version of Gurobi. Then, you could start with setting the NoRelHeurTime parameter to, e.g., 600 seconds, which activates the no relaxation heuristic.

  I also don't know how callback may help to reduce the run time. I am wondering if you can help me to understand this

  Callbacks may reduce the runtime if you have a heuristic or an algorithm which can construct good feasible solutions out of either the relaxation solution point or out of a given feasible point. Other than than, callbacks can be used to implement lazy constraint and/or user cuts.

  Other than that, you could try to improve the coefficient ranges of your matrix and right hand sides. Our Guidelines for Numerical Issues provide a good starting point for improving the numerics of a given problem.

  Best regards,
  Jaromił

  0

投稿コメントは受け付けていません。