How to set stopping criteria in mixed integer programming?

Hi guys~
Recently, I am using mixed integer programming to solve a problem.
Actually, I want to let it stop when Relative gap is less than 2%.
options = optimoptions(@intlinprog,'RelativeGapTolerance',0.2,'Display','iter')
[x,fval]=intlinprog(f,iint,A,b,[],[],lb,ub,[],options);
Yet, it seems like this setting doesn't work. It have been running for 30min...
Any suggetions? Thank you very much!

1 Comment

I assume 'RelativeGapTolerance',0.02, for the output underneath the setting of options.

Sign in to comment.

 Accepted Answer

Hello Frank,
The stopping tolerance is automatically modified to
tolerance = min(1/(1+|L|), RelativeGapTolerance)
I estimate the lower bound, L, for your problem at around 96, so intlinprog tolerance for stopping on relative gap would be around 0.01, i.e. 1%. It looks like intlinprog cannot reduce the gap from 1.54% in the following 30 minutes.
You have couple of options:
  1. Increase AbsoluteGapTolerance, e.g. to 2.
  2. Divide the objective function by a factor 2 or more, perhaps 10.
  3. Set CutGeneration and/or Heuristics to 'intermediate' and see if this reduces the time to solution.
Kind Regards,
Derya

1 Comment

Btw, how could I find the running time complexity of the solver
[x,fval]=intlinprog(f,iint,A,b,[],[],lb,ub,[],options);
Thank you very much!

Sign in to comment.

More Answers (0)

Products

Release

R2021a

Asked:

on 18 May 2021

Commented:

on 25 Nov 2021

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!