Error in the objective functionsyn

Answered

Bushido Pora

April 10, 2023 11:53

• 
• 
• 

Im trying to add a quadratic expression to my objective function but fail.

Here's the code

x = model.addMVar((a, b), vtype=GRB.BINARY, name = "x_{i,j}")

qexpr = gp.QuadExpr(0)

for s in range(s):

    for i in(a):

        for j in range(b):

            for k in range(b):

                if j != k:

                    qexpr.add(x[i,j]*x[i,k]*data[k,j])

But I get the following error 

"File src\gurobipy\quadexpr.pxi:176, in gurobipy.QuadExpr.add() GurobiError: Invalid argument to QuadExpr addition"

I tried with a linear expression lexpr.add(x[i,j]) but it led to

"GurobiError: Unsupported type (<class 'gurobipy.mquadexpr'="">) for LinExpr addition argument"

I read this may be due to the fact that x is not a Var but an Mvar which are not handled by expression but I'm not sure thus I'm writing this post.

Hope you can help me

Bushido

Related to

• Matrix API
• Syntax Error

0

• 

  Mario Ruthmair

  • Gurobi Staff

  April 13, 2023 09:52

  Hi Bushido,

  QuadExpr does not support MVars. In particular, if x is an MVar, then x[i,j] is a 0-dimensional MVar, not a Var object. You could fix this in several ways:

  • You keep the MVar but use the item() method to get a Var object, i.e.:

  qexpr.add(x[i,j].item() * x[i,k].item(), data[k,j])

  • You use classical Vars which makes sense if you do not exploit matrix operations on variables as in your code piece, i.e.

  x = model.addVars(a, b, vtype=GRB.BINARY, name = "x")

  Independent from that, note that QuadExpr.add() sets the coefficient in the second parameter.

  Best regards,
  Mario

  0
• 

  Bushido Pora

  April 16, 2023 18:30

  • 
  • 
  • 

  Hello, 

  Thank you for your answer.

  Indeed using Vars directly was a solution however I now have a scalar expressed as a vector Matrix vector product in the objective and cant manage to make it work.

  a  = model.addMVar((3,5), vtype=GRB.BINARY)
  b = np.ones((5,5))
  qexpr = gp.QuadExpr(0)
  qexpr.add(a[0,:] @ b @ a[1,:].transpose())

  leads to :

  GurobiError: Invalid argument to QuadExpr addition
  
  

   

  0
• 

  Mario Ruthmair

  • Gurobi Staff

  April 18, 2023 08:42

  This is similar to before: You are working with MVars but want to store an expression in a QuadExpr. You need to use either an MQuadExpr or work with Vars instead of MVars.

  MQuadExpr are temporary objects, there is no constructor. But you can create an expression by assignment:

  qexpr = a[0, :] @ b @ a[1, :].transpose()

  and then extend it via operators "+=".

  Depending on what you want to express in your objective function, there might be a one-liner based on matrix notation instead of summing up terms.

  0
• 

  Bushido Pora

  April 20, 2023 11:08

  • 
  • 
  • 

  Thanks a lot.

  0

Please sign in to leave a comment.