Skip to main content

Get variables by name

Ongoing

Comments

6 comments

  • Maliheh Aramon
    Gurobi Staff Gurobi Staff

    Hi Wenbo, 

    We can use list comprehension as below:

    [var for var in model.getVars() if "gamma" in var.VarName]
    The above will iterate over all variables. To do it more efficiently such that only the variables \(\texttt{gamma}\) are retrieved, we can do it like below:
    names_to_retrieve = (f"gamma[{i},{j}]" for i in range(20) for j in range(2))
    [model.getVarByName(name) for name in names_to_retrieve]

    Best regards,

    Maliheh

     

    0
  • Maedeh Sharbaf
    Gurobi-versary
    First Comment
    First Question

    Dear Maliheh,

    I'm using Gurobi in Python and I'm trying to solve a model that have multiple optimal solution, with one binary and one continuous variable.

    y=m.addVars(SC_id,vtype=GRB.BINARY,name='facility')
    x=m.addVars(PP_id,SC_id,lb=0,ub=1,vtype=GRB.CONTINUOUS,name='assignment')

    Now, I want to retrieve decision variables for every optimal solution. I wrote a piece of code as following:

    count=0
    for aa in range(m.SolCount):
    count+=1
    m.Params.SolutionNumber = aa
    print("solution: ", count)
    for var in m.getVars():
    if var.xn > 0:
    if "assignment" in var.VarName:
    print ('%s %g' %(var.varName, var.xn))

    and this is the results of my code:

    solution: 1
    assignment[11,1] 1
    assignment[12,1] 1
    assignment[23,4] 0.6
    .
    .

    Now I want to save the [i,j] inside the brackets along with the value of assignment. Because right now I have them in terms of string. Since I want to compute Risk of each solution, I need to have value of decision variables for each solution. If I had one optimal solution, below code would work (R[i,j] was computed before). But now that I have multiple optimal solution, I do not know how to access the decision variables for each solution number.

    A=0
    for i in PP_id:
    for j in SC_id:
    if x[i,j].x>0:
    A+=x[i,j].x*R[i,j]

    Thanks for your time,

    I really appreciate your help

     

     

     

     

     

    0
  • Maliheh Aramon
    Gurobi Staff Gurobi Staff

    Hi Maedeh, 

    The simplest approach would be to manipulate the variable string names to retrieve the indices \(i\) and \(j\). You can modify the assignment variable names, \(\texttt{assignment[i,j]}\) like below:

    i, _, j = (
    var.VarName.replace("assignment", "")
    .replace("[", "")
    .replace("]", "")
    .partition(",")
    )

    A cleaner approach would be to use regular expressions like below, but the first approach is easier to understand.

    import re
    pattern = re.compile
    ("assignment\[((?P<i>\d+),(?P<j>\d+))\]")
    i, j = pattern.match(var.VarName).groupdict().values()
     
    I hope this answers your question.
     
    Best regards,
    Maliheh
    0
  • Sophie Wuyckens
    First Question
    Gurobi-versary
    Conversationalist

    Hi Maliheh,

    Quick question, I wanted to know the value of some of my variables during the optimization using a callback function (where=MIPSOL). However, I get an error during the callback when I want to access the .X attribute of my variables. Is there any reason for that ? Any workaround ? Thank you.

    0
  • Maliheh Aramon
    Gurobi Staff Gurobi Staff

    Hi Sophie, 

    You need to use the method Model.cbGetSolution(vars) to retrieve values from a new MIP solution. There is a clear example of how to do this at the bottom of the documentation page linked above that you can check.

    Best regards,

    Maliheh 

    0
  • Sophie Wuyckens
    First Question
    Gurobi-versary
    Conversationalist

    Hi Maliheh,

    I saw this solution but the problem is I want to only access specific variables by name and not by the variable object.I tried the following but I still get an error:

    names_to_retrieve = (f"{name}[{i}]" for i in range(M))
    vars_obj = [model.getVarByName(name) for name in names_to_retrieve]
    vars_value = [model.cbGeSolution(var) for var in vars_ob]
    print("Name = {}, Objective value = {}".format(name, sum(vars_value))

    Thank you in advance for your help.

    Sophie

    0

Please sign in to leave a comment.