Binary Varibales get negative -0.0 values

回答済み

Mert Eroglu

2022年12月27日 14:29

• 
• 
• 

Hello, 

How is it possible to get -0.0 when they are binary variables?

Thanks.

関連

• Numerics

0

• 正式なコメント

  

  Simranjit Kaur

  • Gurobi Staff

  2026年11月13日 01:51

  This post is more than two 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 try Gurobot, our chatbot interface offering instant, expert-level support.
• 

  Gwyneth Butera

  • Gurobi Staff

  2022年12月27日 15:25

  Have you had a chance to look at our Knowledge Base article Why does Gurobi sometimes return non-integral values for integer variables? 

  0
• 

  Mert Eroglu

  2022年12月27日 15:27

  • 
  • 
  • 

  Yeah,

  I set IntFeasTol to 1e-9 and IntegralityFocus to 1, but it didn't changed the outcome.

  Thanks for the fast reply.

  0
• 

  Mario Ruthmair

  • Gurobi Staff

  2022年12月27日 18:44

  Yes, things like this still can happen due to limited precision arithmetics.
  We usually do a very rough test for binary variables in a solution: if x > 0.5 then x=1 else x=0.

  0
• 

  Mert Eroglu

  2022年12月28日 06:40

  • 
  • 
  • 

  for j in range(n):
      for i in range(m):
          model.addConstr(y_2022[j,i] >= 0.5 + eps - M*(1-B))
          model.addConstr(y_2022[j,i] <= 0.5 + M*(B))
          model.addConstr((B==1) >> (y_2022[j,i] == 1))
          model.addConstr((B==0) >> (y_2022[j,i] == 0))
  
  I added something like this, but this time i didn't have any outcome, are there
  something wrong with this?
  
  Thanks

  0
• 

  Mario Ruthmair

  • Gurobi Staff

  2022年12月28日 07:25

  I probably was not clear with my last comment. The idea is not to add further constraints to ensure that binary variables get exactly value 0 or 1. This is simply not possible on all our machines with limited precision arithmetics.

  The test I proposed is meant for analyzing the solution after the optimization process has already finished. If the value for a binary variable is larger than 0.5, then it is actually 1 (or True) in the solution, 0 otherwise.

  0
• 

  Mert Eroglu

  2022年12月28日 13:07

  • 
  • 
  • 

  Oh OK, i thought adding such constraint would delete the - sign on the 0.0's but it's the same thing.

  Thanks for your support.

  0

投稿コメントは受け付けていません。