How can I speed up the running time?

Ongoing

Seyedeharmaghan Bahreini

• December 04, 2020 17:13

Dear Gurobi community,

I have a MILP model coded in python and solved by Gurobi. It has 216 SOS constraints, 5330 non-negative continuous variables and 0 integer (0 binary) variable. It takes almost 490s to solve the problem. 

Can you please give me any insights to speed up the running time?

 

Here is the log file: https://www.dropbox.com/s/ew5tccte0gbthov/log_linear_pipe.txt?dl=0 

0

• 

  Jaromił Najman

  Gurobi Staff

  • December 07, 2020 10:11

  This is a cross-post to Speed up Gurobi to solve an MILP.

  0
• 

  Jaromił Najman

  Gurobi Staff

  • December 07, 2020 10:16
  • Edited

  Hi,

  As mattmilten already mentioned in the post at Stackoverflow, you should not increase the feasibility tolerance that much, because it deteriorates the quality of your solution. Did you have a look at the documentation on most important parameters for MIP models?

  Best regards,
  Jaromił

  0
• 

  Seyedeharmaghan Bahreini

  • December 07, 2020 17:43

  

  

  

  Hi Jaromil,

  Yes, I almost tried different parameters for the MILP model. I could reduce the running time to 99.71 seconds by changing the NodeMethod to 0, the Method to 1 and using the maximum feasibility tolerance but still, I need a faster solution. 

  Regarding the feasibility tolerance, I understand your point. However, if I reduce the feasibility tolerance, it becomes really slow. So, I'm looking for a way to reaching a reasonable solution in a short time.

  Thank you!

  Armaghan

  0
• 

  Jaromił Najman

  Gurobi Staff

  • December 09, 2020 08:23
  • Edited

  Hi Armaghan,

  You could also try the Gurobi tuning tool.

  Best regards,
  Jaromił

  0

Please sign in to leave a comment.