How to use ObjectiveCutOff and ObjectiveI​mprovement​Threshold from optimoptions in solver-based optimization problems

1 ビュー (過去 30 日間)
Rogier Doodeman
Rogier Doodeman 2021 年 10 月 12 日
コメント済み: Alan Weiss 2021 年 10 月 15 日
To speed things up I want to set the ObjectiveCutOff and ObjectiveImprovementThreshold. I know the objective solution is in the order of 10^7 so everything above 10^8 can be discarded. However, this does not do anything. Even when I set the ObjectiveCutOff to 10^4, instead of giving an error because this is inveasible the solution stays the same, in the order of 10^7. The same problem occurs with ObjectiveImprovementThreshold.
options = optimoptions('intlinprog','AbsoluteGapTolerance',1,'RelativeGapTolerance',1,...
[sol_4_months,fval_4_months] = solve(energyprob,'Options',options);
Does someone know hw to properly formulate these optimoptions?


Alan Weiss
Alan Weiss 2021 年 10 月 12 日
I think that you have a mistaken view of what these options do. Consult the options to see what they really do. I don't think that they will help you speed the solution.
It sounds to me as if the thing that will most likely speed intlinprog for you is to give a feasible initial point with an objective function value that is of the order of the solution. That will cause the branch-and-bound process to do what you want. But it is often difficult to come up with a feasible initial point.
For other potential ways of speeding intlinprog, see Tuning Integer Linear Programming. Unfortunately, none of the techniques are guaranteed to help.
Good luck,
Alan Weiss
MATLAB mathematical toolbox documentation
  2 件のコメント
Alan Weiss
Alan Weiss 2021 年 10 月 15 日
You can try it. I do not know whether it will help speed things.
Alan Weiss
MATLAB mathematical toolbox documentation


その他の回答 (0 件)

Community Treasure Hunt

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

Start Hunting!

Translated by