Numerical and speed issues related to objective coeficients in multiobjective model
回答済みHello, I am usually solving multiobjective model with 3 obejctives where objective coefficients range is about 1e11. Gurobi warns me about numerical issues but mailny matrix numerical issues are discussed in documentation and my matrix range is ok.
I have noticed that when coefficients within one objective are in smaller range (I tried to reduce this to maybe 1e7 from about 1e9) gurobi is able to solve it faster.
Should reducing coefficient range provide some speedup or is this just a coincidence?
Also should I try to minimalize this range completely or maybe use a greater number of multiobjectives?
-
正式なコメント
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?. -
Hi Daniel,
This page of the documentation also discusses objective ranges. It is recommended for the coefficients of the objective function to be not larger than 10^4, so reducing the size of the coefficients can definitely help.
Concerning the use of a greater number of multi-objectives, this is also a good approach, typically when you have objective coefficients that appear in groups of similar ranges, because if you have only large coefficients, then the objective function can be scaled, while scaling becomes much more difficult if you have a mix of large and small coefficients.
Best regards,
Elisabeth
0
投稿コメントは受け付けていません。
コメント
2件のコメント