value the variable

Answered

Niloofar Shirvani

• January 31, 2022 11:56

Hello All!

I have a MIP model in gurobi and I want to value the variables.

in other words I have a binary variable called x(r,i,j). it is 1 when vehicle r goes from i to j.

I want to value this variable like this in loop and compare the objectives:

for j in N:

x[(1,0,j)].ub=1

x[(1,0,j)].lb=1

it is important to me to know what is difference when vehicle 1 after node 0 go to node 1 or 2 or 3 or ...

0

• 

  Jaromił Najman

  Gurobi Staff

  • January 31, 2022 14:40

  Hi Niloofar,

  You can access the value of an optimization variable after at least one feasible solution has been found via the X attribute. Example mip1.py shows how to print variable values after a successful optimization.

  Best regards, 
  Jaromił

  0
• 

  Niloofar Shirvani

  • January 31, 2022 16:04

  

  

  

  Actually I do not want to print variable value, I want to give them value by myself and then compare each solution.

  0
• 

  Jaromił Najman

  Gurobi Staff

  • January 31, 2022 16:43
  • Edited

  If you want to fix a variable to a particular value you can either add an equality constraint fixing it to a constant value or setting a variable's lower and upper bound to the same value.

  # fix via equality constraint
  x = model.addVar(vtype=GRB.BINARY)
  model.addConstr(x==1)
  # fix via bounds
  x.lb = 1
  x.ub = 1
  # [...]
  # optimize model with fixed values to get new solution and objective value
  model.optimize()

  Best regards, 
  Jaromił

  0
• 

  Niloofar Shirvani

  • January 31, 2022 17:50

  

  

  

  Dear,

  Actually I want to give value 1 to all of x variable but in order.

  for example: at first x(1,0,2)=1 and get model objective, then x(1,0,3)=1 and get model objective by this criteria again. 

  this would be loop for value this variable.

  please give me hand to get the solution.

  0
• 

  Jaromił Najman

  Gurobi Staff

  • February 01, 2022 08:29

  You can use the code from my previous comment and put it in a \(\texttt{for}\)-loop to achieve what you need.

  for j in J:
    # fix variable to desired value
    x[(1,0,j)].lb = 1
    x[(1,0,j)].ub = 1
    # optimize model
    model.optimize()
    # if at least one solution point has been found, 
    # set obj to objective value and infinity otherwise
    if (model.SolCount > 0)
      obj = model.ObjVal
    else
      obj = GRB.INFINITY
    # do something with the objective value for this run
    # [...]
    # reset bound of the variable
    x[(1,0,j)].lb = 0
    x[(1,0,j)].ub = 1

  You can find more information about model and variable attributes in the attributes documentation.

  Best regards, 
  Jaromił

  0
• 

  Niloofar Shirvani

  • February 01, 2022 08:39

  

  

  

  Thank a lot. It works.

  0
• 

  Niloofar Shirvani

  • February 01, 2022 10:02

  

  

  

  Sorry, I have another question.

  how can I bring the objective value in a dictionary or table with their variable?

  0
• 

  Jaromił Najman

  Gurobi Staff

  • February 01, 2022 11:35
  • Edited

  Note that this is no longer Gurobi related. You can use a standard Python dictionary for this

  d = {}
  for j in J:
    #[..]
    d[(1,0,j)] = obj
  
  # access objective value for fixed variable (1,0,0) 
  print(d[(1,0,0)])

  0
• 

  Niloofar Shirvani

  • February 01, 2022 13:36

  

  

  

  Thanks again.

  0

Please sign in to leave a comment.