Can Gurobi not be able to solve a MILP problem if the objective function is formulate differently?

Ongoing

Marina Santarelli

• December 15, 2020 11:09

I have to solve a MILP program and for certain values associate to the variables in objective function I get a feasible solution. However. when I try to change the cost value associated to some variables,  Gurobi does not return a feasible solution. Why? 

0

• 

  Jaromił Najman

  Gurobi Staff

  • December 15, 2020 11:54

  Hi Marina,

  Could you share LOG files showing the issue? In best case you could also share the two model instances where you observe the behavior you described. Please have a look at our Posting to the Community Forum article for further information on sharing models.

  Best regards,
  Jaromił

  0
• 

  Marina Santarelli

  • December 15, 2020 14:44

  

  

  

  This is my objective function:

  f = [zeros(nv_ev + nv_column,1); f_bill; ; 0.56*20; 0.56*40; 0.56*80; 0.56*120 ];
  
  

  where f_bill is  a vector of values between 0.1205 and 0.125.

  If I change f_bill such that my new objective function is:

  f = [zeros(nv_ev + nv_column,1); f_bill*0.5; ; 0.56*20; 0.56*40; 0.56*80; 0.56*120 ];

  Gurobi returns no feasible solution. 

   

  0
• 

  Jaromił Najman

  Gurobi Staff

  • December 15, 2020 15:28

  We will continue this issue in an internal ticket.

  0

Please sign in to leave a comment.