Why is the solution for binary variables 0.999999?

回答済み

Tianyu Mao

2023年07月09日 09:20

• 
• 
• 

Hello

In my program, I have set a set of binary decision variables, with code 

Zeta = md.addVars(V, vtype=GRB.BINARY, name='Zeta[i]')

During the process of adjusting the constraints, I found that the solution for this set of decision variables can be different from 0 or 1. If the solution of this decision variable is 0 or 1, then this decision variable is normally effective. If it is a different value, then this decision variable is not functioning properly. I would like to know what may be causing this problem. Thank you for your reply.

（1）Invalid

Zeta:
10: 0.999999 11: 0.9999979999979998 12: 0.999998999999

（2）Effective

Zeta:
12: 1.0

 

0

• 

  Simranjit Kaur

  • Gurobi Staff

  2023年07月09日 23:08

  Hi Tianyu,

  As explained in the article Why does Gurobi sometimes return non-integral values for integer variables?, the solution value of a binary variable can deviate from 0 or 1. The tolerance for such deviation is defined by the IntFeasTol parameter (default value 1e-5).  

  If the problem with binary variables having a value like 0.999.. only occurs when you try to read them after optimization, you can simply round the values once the optimization process is done. Otherwise, you can try setting the IntegralityFocus parameter to 1, which will prompt Gurobi to take extra steps to remove the integrality violation while maintaining feasibility.

  Best regards,

  Simran

   

   

  1
• 

  Tianyu Mao

  2023年07月10日 06:43

  • 
  • 
  • 

  Thank you for your reply! I found your explanation very helpful.

  0

サインインしてコメントを残してください。