Is the branch and bound tree saved for future re-optimization?

Answered

Hauk Bjørnland

April 08, 2021 10:02

• 
• 

Hi, 

I have a MIP model which I solve through Gurobi's BB-tree, adding some lazy constraints along the way. After the model finds the optimal solution I sometimes want to add a new constraint and continue the process (optimize the model again). I know that when I optimize the model again it will start from the previous solution, but is the branch and bound tree also "saved" after the Gurboi finds the optimal solution in the first iteration? The reason I am asking is that I want to avoid spending time on solutions already visited in the previous iterations of my program. 

Related to

• MIPStart
• Modify

0

• Official comment

  

  Simranjit Kaur

  • Gurobi Staff

  June 23, 2025 02:57 Edited

  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

  April 21, 2021 09:20 Edited

  Hi Hauk,

  In general, when you optimize a model with Gurobi and you interrupt it for some reason and then continue the optimization without altering the model, all previous information, i.e., also the B&B tree, will be preserved. However, when you add a constraint in between, this changes the problem structure. Gurobi will try to warm-start the second optimization run, but cannot re-use all previous information, i.e., the B&B tree is not re-used in this case.

  Best regards,
  Jaromił

  0
• 

  Kenny Girón

  March 13, 2023 08:58

  • 
  • 

  Hi,

  I am optimizing a MILP problem using Gurobi through Yalmip. I execute an optimization problem, then optimize some other different MILP problems as a verification. Depending on the information given by the second stage verification problem, I alter a parameter in the constraints on the first stage/tier optimization problem. 

  Would it be wise to study if Gurobi can still use the information from the previous optimization as a map for further optimizations?

  I also want to avoid spending time on solutions that will also be included in the new optimization problem.

  0
• 

  Jaromił Najman

  • Gurobi Staff

  March 13, 2023 10:52

  Hi Kenny,

  Would it be wise to study if Gurobi can still use the information from the previous optimization as a map for further optimizations?

  It might be worth trying to re-use the optimal solution point found in the first optimization run as a MIPStart for your verification run, cf. How do I use MIP starts?

  This might save some solution time in your verification optimization run.

  Best regards, 
  Jaromił

  0

Post is closed for comments.